Langmuir 2006, 22, 10145-10151
10145
Surface-Supported Bilayers with Transmembrane Proteins: Role of the Polymer Cushion Revisited Mikhail Merzlyakov,† Edwin Li,† Ivan Gitsov,‡ and Kalina Hristova*,† Department of Materials Science and Engineering, Johns Hopkins UniVersity, Baltimore, Maryland 21218, and The Michael M. Szwarc Polymer Research Institute and Department of Chemistry, College of EnVironmental Science and Forestry, State UniVersity of New York, Syracuse, New York 13210 ReceiVed July 7, 2006. In Final Form: September 6, 2006 Protein lateral mobility in surface-supported bilayers is often much lower than the mobility of the lipids. In the present study we explore whether the incorporation of a PEG cushion between the bilayer and the substrate increases the lateral mobility of transmembrane proteins in bilayers produced via directed assembly, a method based on LangmuirBlodgett deposition techniques. In our experiments, the PEG cushions were incorporated by adding PEG lipids to the protein/lipid monolayer at the air/water interface, at the first step of bilayer assembly. The protein and lipid mobilities in 160 different bilayers, with various PEG molecular weights and PEG lipid concentrations, were measured and compared. We found that the measured diffusion coefficients do not depend on the PEG molecular weight or the PEG lipid concentration and are very similar to the values measured in the absence of PEG. Therefore, contrary to our expectations, we found that a PEG cushion does not necessarily increase protein mobility, suggesting that the low protein mobility is not a consequence of protein-substrate interactions. Furthermore, we showed that the low protein mobility is not due to protein aggregation. The major determinant of protein mobility in surface-supported bilayer systems appears to be the method of bilayer assembly. While proteins were always mobile if the bilayers were prepared using the directed assembly method, in the presence and absence of a PEG cushion, other bilayer assembly protocols resulted in complete lack of protein mobility.
Introduction Surface-supported lipid bilayer platforms are being used to probe protein-protein1 and protein-lipid interactions, as well as domain formation,2,3 in membranes. A general problem with these platforms is the limited protein mobility, which is believed to be due to protein-substrate interactions. On the basis of this assumption, it has been hypothesized that the addition of a PEG “cushion” in platform design, by incorporating PEG lipids into the monolayer facing the support, will “lift” the bilayer away from the substrate, decrease the protein-substrate interactions and, ultimately, increase protein mobility.4-9 This “PEG cushion” hypothesis, however, has not been fully verified. In one study, the incorporation of PEG lipids in the bilayer decreased, but did not eliminate, the immobile protein fraction.5 Other studies have demonstrated that the incorporated PEG lipids can obstruct lipid and protein diffusion.6 Furthermore, the optimum PEG lipid concentration that ensures the highest protein lateral mobility has not yet been identified. Systematic investigations should explore the system complexity arising due to variations in the configurations of the PEG chains as a function of the PEG lipid concentration: at a particular * To whom correspondence should be addressed. E-mail:
[email protected]. Phone: (410) 516-8939. Fax: (410) 516-5293. † Johns Hopkins University. ‡ State University of New York. (1) Li, E.; Hristova, K. Langmuir 2004, 20, 9053-9060. (2) Stottrup, B. L.; Veatch, S. L.; Keller, S. L. Biophys. J. 2004, 86, 29422950. (3) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; Gratton, E. Biophys. J. 2001, 80, 1417-1428. (4) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58-64. (5) Wagner, M. L.; Tamm, L. K. Biophys. J. 2000, 79, 1400-1414. (6) Deverall, M. A.; Gindl, E.; Sinner, E. K.; Besir, H.; Ruehe, J.; Saxton, M. J.; Naumann, C. A. Biophys. J. 2005, 88, 1875-1886. (7) Munro, J. C.; Frank, C. W. Langmuir 2004, 20, 10567-10575. (8) Heibel, C.; Maus, S.; Knoll, W.; Ruhe, J. Org. Thin Films 1998, 695, 104-118. (9) Shen, W. W.; Boxer, S. G.; Knoll, W.; Frank, C. W. Biomacromolecules 2001, 2, 70-79.
concentration, known as the crossover concentration, ξ, a transition occurs between spaced randomly coiled polymers (the mushroom regime) to laterally interacting, pushed away from the surface, elongated PEG chains (the brush regime).10 A question remains of whether the configurational regime of the PEG chains affects mobility. Tamm and colleagues observed a sharp ∼30% decrease in the mobile fraction at the crossover concentration.5 However, the diffusion coefficients of the proteins in the mobile fraction did not depend on the PEG lipid concentration. Naumann and colleagues, however, observed a gradual decrease in lipid diffusion coefficients upon increasing the PEG lipid concentrations from 5% to 30%.11 Yet another question pertaining to the polymer cushion requirements, addressed so far in a single study by Tamm and colleagues,5 is whether the nature of the PEG attachment to the substrate (physical or chemical) affects mobility. These authors have compared the effect of chemical tethering and physical adsorption of the PEG chains to the glass substrate and have observed that chemical tethering does not have a significant effect on lipid and protein mobility. In our laboratory, we have found that protein mobility in surface-supported bilayers is greatly influenced by the method of supported bilayer assembly. In one approach that we have explored, vesicles, containing proteins, were fused onto solid substrates to form bilayers.1 We have shown that the lipids in these bilayers are mobile and the protein-protein interactions, as assessed by Fo¨rster resonance energy transfer (FRET), are identical to the protein-protein interactions within suspended liposomes. However, the proteins were always immobile, on clean and on modified glass. In a second approach, we have used a novel method, termed “directed assembly”, to produce surface-supported bilayers with (10) Kenworthy, A. K.; Hristova, K.; Needham, D.; McIntosh, T. J. Biophys. J. 1995, 68, 1921-1936. (11) Naumann, C. A.; Prucker, O.; Lehmann, T.; Ruhe, J.; Knoll, W.; Frank, C. W. Biomacromolecules 2002, 3, 27-35.
10.1021/la061976d CCC: $33.50 © 2006 American Chemical Society Published on Web 10/20/2006
10146 Langmuir, Vol. 22, No. 24, 2006
MerzlyakoV et al.
Figure 1. Methods of surface-supported bilayer assembly. (A) Directed assembly method.12 A monolayer of lipids, PEG lipids, and peptides is compressed and deposited on a glass coverslip using the Langmuir-Blodgett method (I). The second monolayer is deposited using LS deposition (IIa) or VF (IIb). Both LS deposition and VF result in the formation of a surface-supported bilayer with transmembrane peptides (III), as previously demonstrated using oriented circular dichroism.12 (B) Method of Tamm and colleagues.5 An LB lipid/PEG lipid monolayer is deposited (I) and incubated with vesicles containing the peptides (II). This results in the formation of a supported bilayer with embedded peptides (III). (C) Vesicle fusion on a solid support.1 A clean substrate is incubated with vesicles containing the peptides (I). The vesicles break and fuse to the surface to produce a continuous bilayer on a polymer cushion (II).
transmembrane (TM) peptides.12 The directed assembly method is based on Langmuir-Blodgett (LB) deposition techniques and produces surface-supported bilayers in two steps (see Figure 1A): (1) a monolayer, containing proteins and lipids, is deposited on a clean glass slide by the LB technique; (2) a second monolayer, containing only lipids, is deposited to “complete” the bilayer. The novelty in the approach is the incorporation of the proteins into the monolayer at the air/water interface, at the first step of bilayer assembly. Using this procedure, we have obtained bilayers with 100% mobile peptides with transmembrane orientation.12 The peptide mobility was low, however, approximately 3 orders of magnitude lower than the lipid mobility.12 In the present study we explore whether the incorporation of a PEG cushion between the bilayer and the substrate increases protein mobility in bilayers produced via directed assembly. In our experiments, the PEG cushions were incorporated by adding PEG lipids to the protein/lipid monolayer at the air/water interface, at the first step of bilayer assembly. The mobilities of lipids and peptides in 160 different bilayers, with various PEG lengths and PEG lipid concentrations, with PEG chemically tethered to the surface and physically adsorbed, were measured and compared. Materials and Methods Materials. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N[methoxy(poly(ethylene glycol))-5000] (mPEG5000-PE, PEG5K), mPEG3000-PE (PEG3K), mPEG2000-PE (PEG2K), and mPEG1000PE (PEG1K) were obtained from Avanti Polar Lipids (Alabaster, AL). N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)-1,2-dihexadecanoyl-snglycero-3-phosphoethanolamine, triethylammonium salt (NBD-PE) was purchased from Molecular Probes (Eugene, OR). 1,2-Dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[propylmethyldimethoxysilyl(poly(ethylene glycol))-1000] (DPS1K) and DPS3K were produced following a synthetic strategy depicted in Schemes 1 and 2. (12) Merzlyakov, M.; Li, E.; Hristova, K. Langmuir 2006, 22, 1247-1253.
Scheme 1
The first stage shown in Scheme 1 is a typical Williamson ether synthesis with poly(ethylene glycol) of nominal molecular weight 1000 (n ) 23) or 3000 (n ) 68) (Polymer Standards Service) and (3-chloropropyl)methyldimethoxysilane (Gelest, Inc.), taken in equimolar amounts. The reaction is performed in dry tetrahydrofuran with NaH as the base catalyst. The monosubstituted derivative (product A) is isolated as a white waxy solid in 38% yield by flash chromatography using silica gel (60 Å, 63-200 µm, Sorbent Technologies) and dichloromethane as the eluent. NMR analysis was performed on a Bruker Avance 600 MHz spectrophotometer in CDCl3 at room temperature with the solvent signal as the internal standard. 1H NMR (600 MHz, CDCl3): δ 0.06 (s, SiCH3, 3H) 0.74 (t, CH2Si, 2H), 1,72 (q, CH2CH2CH2, 2H), 2.53 (t, OCH2CH2, 2H), 3.56 (s, CH3OSi, 6H), 3.68 (s, CH2CH2O, 92H or 272H). The second stage depicted in Scheme 2 involves the interaction of 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine (Avanti Polar lipids, Inc.) and the succinimidyl carbonate derivative of the propylmethyldimethoxysilyl-PEG (prepared from product A by a known procedure13) in chloroform and triethylamine as the catalyst.13 (13) Zalipsky, S. Bioconjugate Chem. 1993, 4, 296-299.
Surface-Supported Bilayers with TM Proteins
Langmuir, Vol. 22, No. 24, 2006 10147 Scheme 2
Product C is isolated by preparative aqueous column chromatography over Sephadex G-50 (20-80 µm, Sigma-Aldrich) as a white solid in 58% yield. 1H NMR (600 MHz, CDCl3): δ 0.06 (s, SiCH3, 3H) 0.74 (t, CH2Si, 2H), 0.86 (t, CH2CH3, 6H), 1.2-1.3 (m, CH2, 48H), 1,7 (q, CH2CH2CH2, 2H), 2.26 (t, CH2CO, 4H), 2.52 (t, OCH2CH2, 2H), 3.35 (m, OCH2CH2NH, 2H), 3.43 (m, OCH2CH2NH, 2H), 3.56 (s, CH3OSi, 6H), 3.68 (s, CH2CH2O, 92H or 272H), 3.97 (br m, CH2OP, 4H), 4.12-4.20 (br m, CH2OCONH, 2H), 4.36 (dd, CH2OCOCH2, 2H), 5.22 (m, CHOCOCH2, 1H). The FGFR3 TM domain, TMFGFR3, RRAGSVYAGILSYGVGFFLFILVVAAVTLCRLR, and its variant, TMFGFR3E, RRAGSVYAGILSYGVGFFLFILVVEAVTLCRLR, were synthesized and purified as described.14 The peptides were labeled with the fluorophores Cy3, fluorescein, or rhodamine, by coupling the dyes to the single cysteine in the sequence, close to the C-terminus.15 The labeled proteins were eluted in 100% acetonitrile and further purified by HPLC. Samples collected from the HPLC column (Varian) were lyophilized and redissolved in a solution of HFIP/MeOH (1:2). The solvents were removed, and the peptides were redissolved in 2,2,2trifluoroethanol (TFE). The helicity of the peptides was confirmed using circular dichroism (CD) measurements using a J-710 spectropolarimeter (Jasco, Easton, MD). Stock solutions of lipids in chloroform and peptides in TFE were mixed together in the required proportions (see Table 2 below), and the mixtures were stored in the freezer. Substrate Preparation. Coverslips (24 × 40 × 0.15 mm, catalog no. 12-544-12, Fisher Scientific) were cleaned by sonication for 10 min in 2-propanol (IPA), then in acetone, and again in IPA. After the second sonication in IPA, the slides were rinsed in deionized (DI) water and soaked in 30% hydrogen peroxide, 70% sulfuric acid for >30 min. The slides were extensively rinsed with DI water and were stored in water until use. Bilayer Deposition: Directed Assembly Method (Figure 1A). A lipid monolayer containing the peptides and PEG lipids was deposited on a glass coverslip by the LB method12 as follows: First, two clean wet coverslips were stacked together and were immersed vertically into the clean subphase of a Langmuir-Blodgett trough (model 611, Nima Technologies, Coventry, England). Next, a solution of lipids, peptides, and PEG lipids in TFE/chloroform was spread (14) Iwamoto, T.; You, M.; Li, E.; Spangler, J.; Tomich, J. M.; Hristova, K. Biochim. Biophys. Acta 2005, 1668, 240-247. (15) Li, E.; You, M.; Hristova, K. Biochemistry 2005, 44, 352-360.
dropwise at the air/water interface of the open trough (600 cm2). Thus, the difference between this protocol and a previously published one12 is the addition of PEG lipids to the monolayer at the air/water interface. After spreading, the solvents were allowed to evaporate for 30 min, and the monolayers were cycled once at a 100 cm2/min compression/expansion rate up to 32 mN/m surface pressure and were then compressed to 32 mN/m. The monolayer was transferred on the outer surfaces of the coverslips during withdrawal from the subphase at a rate of 15 mm/ min at a surface pressure of 32 mN/m, as described in detail elsewhere.12 The second lipid monolayer was deposited by either the Langmuir-Schaefer (LS) method (Figure 1A, IIa) or vesicle fusion (VF) (Figure 1A, IIb).12 While the behaviors of bilayers produced via the LB/LS and the LB/VF routes of assembly were identical, the LB/LS route was preferred for systematic studies and comparisons because it allows better control over the lipid density in the second leaflet, which is determined by the applied surface pressure. For LS deposition, about ∼30 nmol of POPC lipids was deposited on the surface of the open trough; the solvent was allowed to evaporate for 30 min, and the monolayer was compressed at 100 cm2/min up to 32 mN/m surface pressure. The LB monolayer-coated coverslip, held horizontally with the coated face down, was lowered gently until it touched the surface. A microscope slide with two adhesive tape spacers was used to “scoop out” the coverslip, by positioning the edges of the coverslip on the adhesive tape spacers. The gap between the coverslip and the microscope slide was filled with water or buffer, and the perimeter of the coverslip/microscope slide sandwich was sealed with silicon grease. LS deposition was usually performed within 40 min after the LB deposition of the first monolayer. Bilayers, formed using this protocol, were stable for at least 1 day, if properly sealed to prevent dehydration. To achieve chemical attachment of the PEG chains to the substrate, silane-terminated DPS lipids were used instead of the commercially available nonreactive PEG lipids. The DPS lipids were synthesized as described in the Materials. After LB deposition, DPS lipids could not be removed with organic solvent, thus suggesting successful chemical attachment at room temperature. Choice of PEG Lipid Concentrations in the LB Monolayer. All LB monolayers (monolayers facing the support) contained POPC as the host component, 1 mol % NBD-PE, 0.1 mol % peptide, and PEG lipids. The concentrations of the PEG lipids in the LB monolayer
10148 Langmuir, Vol. 22, No. 24, 2006
MerzlyakoV et al.
Table 1. Calculation of the Crossover Concentration ξ as a Function of the PEG Chain Length PEG lipid PEG1K PEG2K PEG3K PEG5K
no. of EG units, N 22 45 67 113
mole fraction (mol %) of PEG lipids at crossover, ξ ) A/RF2 ) (70 Å2)/RF2
Flory radius (Å), RF ) (3.5 Å)N3/5 22.4 34.4 43.6 59.7
14.0 5.9 3.7 2.0
Table 2. Mole Fraction of PEG Lipids in the LB Monolayer for the Various PEG Lipids Used in the Study PEG lipid
0.25ξ
0.5ξ
ξ
2ξ
4ξ
PEG1K DPS1K PEG2K PEG3K DPS3K PEG5K
3.5 3.5 1.5 0.9 0.9 0.5
7.0 7.0 3.0 1.9 1.9 1.0
14.0 14.0 5.9 3.7 3.7 2.0
28.0 28.0 11.8 7.4 7.4 4.0
56.0 56.0 23.6 14.8 14.8 8.0
were chosen as multiples of the crossover concentration ξ, which is a function of the molecular weight of the PEG chain. The concentrations used were 0.25ξ, 0.5ξ, ξ, 2ξ, and 4ξ. The crossover concentration ξ marks the transition from the “mushroom” to the “brush” regime10 and describes the case when the PEG random coils are barely touching each other. The size of the random coil is given by the Flory radius, and at the crossover concentration ξ, the average spacing between the PEG lipids in the LB monolayer is equal to the Flory radius. Table 1 shows the calculated Flory radius, RF, and the mole fraction of PEG lipids at the crossover concentration ξ as a function of the PEG molecular weight. For the calculation, we have assumed that the POPC area in the LB monolayer is A ) 70 Å2. Table 2 shows the mole fraction of PEG lipids in the LB monolayer facing the support, used in the experiments. Experiments were set up for the 6 × 5 ) 30 different compositions shown in Table 2, thus exploring the effect of various PEG molecular weights and PEG lipid concentrations, as well as means of attachment to the substrate (chemical for DPS lipids or physical). At least four different bilayers were prepared for each composition. In addition, approximately 40 bilayers were prepared that explored conditions not shown in Table 2: no PEG lipids, different TM domain variants, different fluorescent dyes (fluorescein or rhodamine instead of Cy3), and various time intervals between the LB and LS monolayer depositions. Other Bilayer Assembly Methods. Two other methods, in addition to the directed assembly method, were used to prepare surface-supported protein/lipid bilayers on polymer cushions. Method of Tamm and Colleagues (Figure 1B).5 An LB monolayer containing POPC, NBD-PE, and PEG lipids (but no proteins) was deposited on a coverslip using the LB method. Next, lipid vesicles with incorporated proteins were fused to complete the bilayer. The difference between this method and the directed assembly method is the incorporation of the proteins in the vesicles, not in the LB monolayer. Vesicle Fusion on a Solid Support (Figure 1C).1 Vesicles were prepared from POPC, NBD-PE, PEG lipids, and peptides. Clean glass substrates were incubated with these vesicles. It has been shown that such liposomes break and fuse on the surface to produce a continuous bilayer on a polymer cushion.16 The measured diffusion coefficient of the lipids in these bilayers, ∼2 µm2/s, suggested that a continuous bilayer was indeed produced. Measurements of Diffusion Coefficients. The mobility of lipids and peptides in supported bilayers was measured using fluorescence recovery after photobleaching (FRAP) with an Eclipse E600 fluorescence microscope (Nikon) equipped with a mercury lamp and a SPOT RT camera (Diagnostic Instruments). Diffusion coefficients were determined using the boundary profile evolution (BPE) method described elsewhere.12 Briefly, a relatively large area (16) Albertorio, F.; Diaz, A. J.; Yang, T. L.; Chapa, V. A.; Kataoka, S.; Castellana, E. T.; Cremer, P. S. Langmuir 2005, 21, 7476-7482.
Table 3. Diffusion Coefficients (µm2/s) of NBD-PE in Supported Bilayers of Different LB Monolayer Compositions, Prepared via the Directed Assembly Method (LB/LS Deposition)a PEG lipid
0.25ξ
0.5ξ
ξ
2ξ
4ξ
PEG1K DPS1K PEG2K PEG3K DPS3K PEG5K
2.41 ( 0.66 2.94 ( 0.12 2.34 ( 0.33 3.13 ( 0.24 2.80 ( 0.18 2.45 ( 0.12
2.95 ( 0.19 2.29 ( 0.18 3.01 ( 0.27 2.65 (0.23 2.31 ( 0.24 2.73 ( 0.34
2.28 ( 0.25 2.5 ( 0.2 2.68 ( 0.09 2.39 ( 0.21 2.38 ( 0.15 2.23 ( 0.23
2.26 ( 0.11 2.64 ( 0.51 2.25 ( 0.31 2.70 ( 0.22 2.43 ( 0.08 3.61 ( 0.41
0.69 ( 0.06 2.43 ( 0.30 1.95 ( 0.18 2.00 ( 0.47 1.80 ( 0.29 2.05 ( 0.15
a The diffusion coefficient of NBD-PE in supported bilayers containing no PEG was 2.5 ( 0.2 µm2/s (for both LB/LS and LB/VF12 deposition) and 2.5 ( 0.2 µm2/s (for LB/VF deposition12).
(∼1/4 of the whole field size) was bleached, and the time evolution of the intensity profile of the boundary region was monitored during recovery. The boundary profile was fitted to a Gaussian error function, yielding the diffusion depth. The square of the diffusion depth versus recovery time is a straight line with a slope equal to the diffusion coefficient. Bilayer Quality Assessment. The roughness of the fluorescence intensity profile, r, was used as an estimate of bilayer quality. The roughness r is defined as the ratio r ) σ/F0, where F0 is the average fluorescence intensity and σ is the standard deviation of the intensity profile around this average value. A bilayer of good quality with no visible defects has a smooth intensity profile and therefore low roughness, while a bilayer of poor quality, showing visible defects, exhibits higher intensity profile roughness. Alternatively, the amount of visible defects in bilayers can be quantified from histograms of pixel intensities of the captured fluorescence images. There is a strong correlation between histogram width and roughness, both being the lowest for bilayers of highest visual quality (not shown). FRET Measurements. FRET analysis was used to assess the degree of peptide oligomerization in supported bilayers, as described previously.15 Fluorescein-labeled peptides acted as FRET donors, and rhodamine-labeled peptides acted as FRET acceptors. Bilayers were prepared with different donor-to-acceptor ratios, but the same total peptide concentration. Fluorescence spectra were recorded from single supported bilayers in a fluorometer, as described in detail elsewhere.17 FRET efficiencies were calculated from the decrease in donor intensity in the presence of the acceptor.18 The functional dependence of the measured FRET efficiency on the acceptor mole fraction gives information about the aggregation state of the protein in the bilayer.19
Results Effect of PEG Cushions on Peptide Mobility. The diffusion coefficients of bilayers with TM peptides, produced via the method of directed assembly (Figure 1A), were measured using FRAP and the BPE method, as described previously.12 The calculated diffusion coefficients, as a function of the PEG chain length and concentration, are shown in Table 3 for lipids and in Table 4 for peptides. The results are also plotted in Figure 2. Each value, reported in Tables 3 and 4, is the average value determined for at least four different bilayers prepared under identical conditions. As seen in Tables 3 and 4 and in Figure 2, no systematic changes in diffusion coefficients are observed upon variations in the PEG length and concentration. It appears that PEG2K gives a slightly higher diffusion coefficient, particularly for concentrations above the crossover. The measured values are very similar to the values, measured in the absence of PEG, for both LB/LS bilayers (2.5 ( 0.2 µm2/s for lipids, (6.6 ( 3.3) × 10-3 µm2/s for peptides, measured here) and LB/VF bilayers (2.5 ( 0.2 µm2/s for lipids, (17) Merzlyakov, M.; Li, E.; Casas, R.; Hristova, K. Langmuir 2006, 22, 6986-6992. (18) You, M.; Li, E.; Wimley, W. C.; Hristova, K. Anal. Biochem. 2005, 340, 154-164. (19) Adair, B. D.; Engelman, D. M. Biochemistry 1994, 33, 5539-5544.
Surface-Supported Bilayers with TM Proteins
Langmuir, Vol. 22, No. 24, 2006 10149
Table 4. Diffusion Coefficients (µm2/s × 103) of Cy3-TMFGFR3 in Supported Bilayers of Different Compositions, Prepared via the Directed Assembly Method (LB/LS Deposition)a PEG lipid
0.25ξ
0.5ξ
ξ
2ξ
4ξ
PEG1K 7.9 ( 4.8 9.0 ( 1.6 9.2 ( 1.4 9.4 ( 0.6 7.5 ( 1.6 DPS1K 13 ( 2.9 14.6 ( 0.6 8.2 ( 1.4 8.0 ( 1.9 5.6 ( 3.6 PEG2K 14.2 ( 2 13.3 ( 1.8 14.6 ( 1.6 19.4 ( 2.6 19.8 ( 3.7 PEG3K 7.6 ( 0.8 12.7 ( 1 8.0 ( 1.2 9.4 ( 1 7.7 ( 1.9 DPS3K 7.3 ( 1.3 12.2 ( 2.4 8.7 ( 2.5 8.2 ( 0.7 5.3 ( 1.1 PEG5K 7.4 ( 1.6 10.8 (3.5 8.9 ( 0.8 7.0 ( 2.4 4.9 ( 0.4 a
The diffusion coefficient of Cy3-TMFGFR3 in supported bilayers containing no PEGs was (6.6 ( 3.3) × 10-3 µm2/s (for LB/LS deposition) and (6.5 ( 2.0) × 10-3 µm2/s (for LB/VF deposition12).
Table 5. Roughness r of NBD-PE Intensity Profiles in Supported Bilayers of Different LB Monolayer Compositionsa PEG lipid
0.25ξ
0.5ξ
ξ
2ξ
4ξ
PEG1K DPS1K PEG2K PEG3K DPS3K PEG5K
0.014 0.018 0.004 0.014 0.011 0.012
0.031 0.012 0.006 0.005 0.005 0.064
0.036 0.038 0.006 0.012 0.009 0.015
0.013 0.019 0.008 0.006 0.010 0.043
0.008 0.014 0.005 0.007 0.021 0.023
a The roughness r is defined as the ratio r ) σ/F0, where F0 is the average fluorescence intensity and σ is the standard deviation of the intensity profile around this average value. The reported roughness values are the averages from four different experiments for each composition. Bilayers were prepared using the directed assembly method (LB/LS deposition).
Table 6. Roughness r of Cy3-TMFGFR3 Intensity Profiles in Supported Bilayers of Different LB Monolayer Compositionsa PEG lipid
0.25ξ
0.5ξ
ξ
2ξ
4ξ
PEG1K DPS1K PEG2K PEG3K DPS3K PEG5K
0.020 0.029 0.018 0.028 0.029 0.023
0.022 0.021 0.020 0.019 0.019 0.030
0.024 0.027 0.020 0.022 0.021 0.019
0.021 0.025 0.018 0.021 0.025 0.033
0.016 0.031 0.018 0.028 0.045 0.029
a
The reported values are the averages from four different experiments for each composition. Bilayers were prepared using the directed assembly method (LB/LS deposition).
Figure 2. Diffusion coefficients of NBD-PE (A) and Cy3-TMFGFR3 (B) as a function of (1) the type of PEG lipids used for cushioning and (2) the PEG lipid concentration. Values are from Tables 3 and 4.
(6.5 ( 2.0) × 10-3 µm2/s for peptides) reported previously.12 We therefore conclude that the presence of PEG does not have a significant effect on the diffusion coefficients. The exact method of deposition of the second monolayer (LS or VF, Figure 1A, IIa or IIb) does not have an effect on protein and lipid mobility
either. The mobility of TMFGFR3E is very similar to the mobility of TMFGFR3 (not shown). Furthermore, the mobility is not affected by the particular fluorophore usedsCy3, fluorescein, or rhodamine (not shown). Our findings demonstrate that the PEG cushion is neither a necessary nor a required component in the design and fabrication of surface-supported bilayers with mobile transmembrane peptides. Contrary to our expectations, the incorporation of a PEG cushion did not increase the low peptide mobility in the surfacesupported bilayers produced via directed assembly. These results strongly suggest that the low peptide mobility is not due to peptide-substrate interactions, but is intrinsic to the bilayer itself. Effect of PEG Chemical Tethering on Peptide Mobility. The diffusion coefficients of lipids and peptides in the presence of DPS, a silane-terminated lipid, are reported in Tables 3 and 4 for two different PEG molecular weights, 1000 and 3000. Comparison of the values in Tables 3 and 4 suggests that the chemical attachment does not affect the mobility of peptides and lipids, consistent with a previous report.5 Therefore, the use of silane-terminated PEG lipids, which need to be customsynthesized, does not offer an advantage over the use of the commercially available nonreactive PEG lipids (Avanti) in terms of protein mobility. Quality of Bilayers, Assessed by Fluorescence Microscopy. Next we investigated whether the incorporation of a PEG cushion affects the quality of the bilayers. To do so, we calculated the fluorescence intensity profile roughness r of bilayers containing 1 mol % NBD-PE and 0.1 mol % Cy3-TMFGFR3, as described in Materials and Methods. Results for various PEG chain lengths and PEG lipid concentrations, shown in Table 5 (for NBD-PE) and Table 6 (for Cy3-TMFGFR3), reveal no general trends. We see that the intensity profile roughness is generally low when PEG2K is used. We conclude that the quality of the bilayers is the highest for PEG2K, but the effect is small. Diffusion Coefficients and Bilayer Quality. A question arises of whether the quality of the bilayers affects lipid and peptide
10150 Langmuir, Vol. 22, No. 24, 2006
MerzlyakoV et al.
Figure 4. Measured FRET efficiency for TMFGFR3E as a function of the acceptor fraction. Bilayers, prepared via the directed assembly method (LB/VF deposition), contained fluorescein-TMFGFR3E (donor) and rhodamine-TMFGFR3E (acceptor). The total TMFGFR3E concentration in the bilayers was 0.3 mol %, while the donor-to-acceptor ratio was varied. The FRET efficiency was measured from the decrease in the donor fluorescence in the presence of the acceptor, as described in detail elsewhere.20 Error bars represent the standard deviations based on four different experiments. The dependence of the FRET efficiency on the acceptor mole fraction is linear, indicating that only monomers and dimers, but no large peptide aggregates, exist in the supported bilayer (see the text for details).
Figure 3. Diffusion coefficients of NBD-PE (A) and Cy3-TMFGFR3 (B) in supported bilayers prepared via the directed assembly method (LB/LS deposition) as a function of the intensity profile roughness. The profile roughness r is defined as the ratio r ) σ/F0, where F0 is the average fluorescence intensity and σ is the standard deviation of the intensity profile around this average value (see Materials and Methods). (A) No correlation between NBD-PE (i.e., lipid) diffusion coefficients and NBD intensity profile roughness. (B) A weak general trend for peptides: the diffusion coefficient increases as the roughness decreases (bilayer quality improves).
mobility. In Figure 3, we show the diffusion coefficients as a function of the intensity profile roughness for lipids and peptides. In Figure 3A, we see no correlation between lipid diffusion and intensity profile roughness. Figure 3B shows a weak general trend for peptides: the diffusion coefficient increases as the quality of the bilayer improves. Peptide Oligomerization Probed by FRET. Next, we investigated whether the peptides in the supported bilayer are clustering in oligomers which are large enough to explain the low peptide mobility, but small enough to be below the optical resolution. To answer this question, we measured FRET efficiencies as a function of the acceptor fraction at constant total peptide concentration, as previously described.1,20,21 If highorder oligomers were to form, the FRET efficiency would not depend on the acceptor mole fraction, since any donor would have a high probability of oligomerizing with at least one acceptor and of being quenched. If, however, only dimers and no higher (20) Li, E.; You, M.; Hristova, K. J. Mol. Biol. 2006, 356, 600-612. (21) You, M.; Li, E.; Hristova, K. Biochemistry 2006, 45, 5551-5556.
order oligomers were to form, the probability of any donor to dimerize with an acceptor would be proportional to the acceptor fraction, such that the FRET efficiency would increase linearly with the acceptor fraction (the derivation of linearity can be found in ref 19). The investigated TM peptides, which correspond to the TM domain of human FGFR3, have a natural propensity to form sequence-specific dimers in the plane of the membrane, such that the monomers and dimers are in equilibrium.15,18 We have previously observed that the FRET efficiency increases linearly with the acceptor mole fraction in free-standing liposomes.15,20 Would such a linear relationship hold in the supported bilayers? To answer this question, supported bilayers containing peptides labeled with fluorescein (FRET donor) and rhodamine (FRET acceptor) were prepared via the directed assembly method. FRET spectra were recorded in a fluorometer as described in detail in ref 17. The total peptide concentration was fixed at 0.3 mol %, while the donor-to-acceptor ratio was varied. The measured FRET efficiencies, calculated from the decrease in donor fluorescence as described,18 are shown in Figure 4. We see that the FRET efficiency is linear with the acceptor mole fraction, indicating that the peptides in the surface-supported bilayer exist in a monomer-dimer equilibrium and do not form large aggregates. Effect of Dehydration. As discussed in Materials and Methods, bilayers were completed within 40 min after the deposition of the LB monolayer. A 40 min time period is required for removing the first monolayer from the air/water interface, spreading a new monolayer, chloroform evaporation, and monolayer compression. During that time, the coverslips coated with the LB monolayers were set aside in the dust-free cabinet of the trough. To investigate the sensitivity of the LB monolayers to dehydration, LS monolayers were deposited 40 min, 2 h, and 4 h after the deposition of an LB monolayer containing PEG2K lipids at twice the crossover concentration, 2ξ. Diffusion coefficients were measured using FRAP and the BPE method, as described in Materials and Methods. When the interval between LB and LS deposition was increased from 40 min to 2 h, the
Surface-Supported Bilayers with TM Proteins
diffusion coefficient of the lipids decreased from 2.25 ( 0.31 to 0.67 ( 0.33 µm2/s, while the diffusion coefficient of the peptides decreased from (19.4 ( 2.6) × 10-3 to (8.4 ( 2.5) × 10-3 µm2/s. Hydration of the bilayers for several hours did not improve the mobility. If LS deposition was performed 4 h after LB deposition, the overall mobility of lipids and peptides was further reduced. Furthermore, there were areas with immobile peptides and lipids, with no distinct borders between regions containing mobile and immobile molecules. It therefore appears that LB dehydration prior to LS deposition significantly reduces protein and lipid mobility. We also formed bilayers with two identical leaflets by depositing an LS monolayer immediately after LB deposition, such that both the LB and the LS monolayers originated from the same monolayer at the air/water interface. The peptide and lipid diffusion coefficients did not increase with respect to the values measured when there was a 40 min interval between LB and LS deposition. Therefore, it appears that the dehydration of the LB monolayer in the course of the first 40 min after deposition is not significant or its effect is reversible. Effect of PEG Peptide Modification on Peptide Mobility. Next, we investigated whether the attachment of a PEG chain to the peptide has an effect on peptide mobility. To answer this question, PEG chains of molecular weight 333 or 509 (Quanta Bio Design) were coupled to the N-terminus using succinimidyl chemistry. Bilayers containing PEG peptides were prepared using the directed assembly method. In some cases, both PEG lipids and PEG peptides were added to the LB monolayer. We found that the lateral mobility of the peptides is not affected by their PEG modification: the diffusion coefficients in bilayers containing PEG peptides, or both PEG peptides and PEG lipids, were very similar to the values reported in Table 4. This finding further supports the idea that the observed low peptide mobility in the bilayer is not due to peptide-substrate interactions, but is intrinsic to the bilayer itself. Effect of the Assembly Method on Peptide Mobility. We have previously shown that the directed assembly method is the only method that produces bilayers with mobile FGFR3 TM peptides in the absence of a PEG cushion.12 Here we explored whether bilayers with mobile proteins can be produced using the method of Tamm and colleagues5 (Figure 1B) or vesicle fusion on a solid support (Figure 1C) when a PEG cushion is incorporated between the support and the bilayer. In the method of Tamm and colleagues, the PEG cushion was incorporated by adding PEG lipids to the LB monolayer, as shown in Figure 1B. After that, lipid vesicles with incorporated proteins were fused to complete the bilayer. FRAP experiments on such supported bilayers showed complete lack of protein mobility, despite the presence of the PEG cushion and despite full lipid mobility. For the vesicle fusion studies, we incorporated PEG lipids into vesicles containing the TM peptides; the vesicles were then incubated with clean glass substrates. It has been shown that such PEG vesicles break and fuse on the surface to produce a continuous bilayer on a polymer cushion.16 The measured lipid diffusion coefficient in these bilayers, ∼2 µm2/s, suggested that a continuous bilayer was indeed produced. However, the peptides exhibited a complete lack of lateral mobility. Thus, the presence of a PEG cushion in bilayers assembled following the procedure of Tamm and colleagues or vesicle fusion did not ensure protein mobility. On the other hand, proteins were always mobile if the bilayers were prepared using the directed assembly method, in the presence and absence of a PEG cushion. Thus, the major determinant of protein mobility in supported bilayers appears to be the method of bilayer assembly.
Langmuir, Vol. 22, No. 24, 2006 10151
Discussion We have previously demonstrated that the mobility of the FGFR3 TM peptide in surface-supported bilayers is low, about 3 orders of magnitude lower than the lipid mobility.12 It is generally believed that the low protein mobility is due to interactions between the protein and the substrate. Therefore, in this study, we investigated whether the protein mobility can be increased by incorporating a polymer cushion between the bilayer and the substrate. Our results demonstrate that the addition of a PEG cushion to the supported bilayer does not enhance the lateral mobility of the particular protein we study, the TM domain of FGFR3. Furthermore, the mobility does not depend on the molecular weight or the concentration of the incorporated PEG lipids. For the highest PEG molecular weight and the highest PEG lipid concentration, the distance between the substrate and the bilayer should exceed 100 Å. Therefore, there should be no physical contact or even long-range interactions between the substrate and the proteins in the bilayer. Thus, it appears that the low protein mobility is not due to protein-substrate interactions. A second possibility to explain the low protein diffusion coefficient is the formation of large protein aggregates within the supported bilayer. Aggregates of hundreds of proteins could move very slowly. However, the linear dependence of the measured FRET efficiency on the acceptor mole ratio in Figure 4 demonstrates that only dimers form due to sequence-specific protein-protein interactions, as shown previously.15 Furthermore, we have recently shown that the free energy of dimerization of the FGFR3 TM domain in surface-supported bilayers is similar to that in liposomes, indicating that protein interactions in surfacesupported bilayers and in free-standing vesicles are similar;17 this is a second indication that large protein aggregates do not form. Thus, the observed low protein mobility is not due to uncontrolled protein aggregation. We therefore propose that the mobility of the FGFR3 TM peptide is intrinsically low. The diffusion of model TM peptides, as well as naturally occurring TM domains such as TMFGFR3 (corresponding to the TM domain of a human protein), within lipid bilayers should be studied further to uncover the general and sequence-specific factors that govern protein mobility.
Conclusions In this paper, we have revisited the effect of the PEG cushion on protein mobility in surface-supported bilayers. We have investigated whether the incorporation of PEG lipids into bilayers produced via the directed assembly method increases the mobility of one transmembrane protein, the TM domain of FGFR3. We have found that the measured diffusion coefficients do not depend on the PEG molecular weight and the PEG lipid concentration and are very similar to the values measured in the absence of PEG. Therefore, contrary to our expectations, we have found that PEG cushions do not necessarily increase protein mobility, suggesting that the low mobility is not due to protein-substrate interaction. Furthermore, we have shown that the low mobility is not a consequence of protein aggregation. The major determinant of protein mobility in supported bilayers appears to be the method of bilayer assembly. Furthermore, our work confirmed previous findings that chemical attachment of the PEG chain to the substrate does not offer advantages over physical adsorption. Therefore, robust surface-supported bilayers with transmembrane proteins can be assembled without customsynthesized components. Acknowledgment. This research was supported by a grant from The National Science Foundation (NSF MCB 0315663). LA061976D
