|Home | About | Journals | Submit | Contact Us | Français|
The S4 transmembrane segment in voltage-gated ion channels, a highly basic α helix, responds to changes in membrane potential and induces channel opening. Earlier work by others indicates that the S4 segment interacts with lipids in plasma membrane, but its mechanism is unclear. Working with synthetic tryptophan-labeled S4 peptides, we characterized binding of autonomous S4 to lipid membranes. The binding free energy (5.2 ± 0.2 kcal/mol) of the peptide-lipid interaction was estimated from the apparent dissociation constants, determined from the changes in anisotropy of tryptophan fluorescence induced by addition of lipid vesicles with 30 mol% phosphatidylglycerol. The results are in good agreement with the prediction based on the Wimley-White hydrophobicity scale for interfacial (IF) binding of an alpha-helical peptide to the lipid bilayer (6.98 kcal/mol). High salt inhibited the interaction, thus indicating that the peptide/membrane interaction has both electrostatic and non-electrostatic components. Furthermore, the synthetic S4 corresponding to the Shaker potassium channel was found to spontaneously penetrate into the negatively charged lipid membrane to a depth of about 9 Å. Our results revealed important biophysical parameters that influence the interaction of S4 with the membrane: they include fluidity, surface charge, and surface pressure of the membrane, and the α helicity and regular spacing of basic amino-acid residues in the S4 sequence.
It has been traditionally thought that integral proteins form transmembrane domains with hydrophobic residues in contact with the phospholipid acyl chains, and that the membrane localization of polar sequences is thermodynamically unfavorable . Recent evidence from apparently divergent areas of study, such as voltage-gated ion channels and cell-penetrating peptides, seems to question this popular belief [2–4].
Hodgkin and Huxley conceived voltage-dependent ionic channels in order to account for the propagation of action potential in the squid giant axon . The ideas of the inevitable presence of charged gating moieties in these channels to—accommodate for the ionic selectivity—have been around for four decades. Most of these channels respond to membrane potential by a domain, consisting of four trans-membrane helices, called voltage-sensing domain (VSD), which regulates the conformational changes that allow for the passage of ions through the pore domain (S5-pore-S6) . The S4 region of the voltage sensor domain has been the key to this function . Data from x-ray crystallography have revealed finer information at atomic resolution . While valuable insights have been gained from these experiments, the exact mechanism involved, including the actual movement of the charged S4 domain within the channel or in the membrane, is yet to be fully understood.
Crystal structure of the bacterial voltage-gated channel KvAP was elucidated on the protein obtained by detergent solubilization and in complex with Fab fragments [8, 9]. Questions have been raised regarding the apparent loose organization of the voltage sensor domain—whether it was due to the absence of lateral pressure of lipid bilayer or due to unit cell crystal packing between the tetramers. The possibility of non-native conformation of the resolved structure has been considered [10, 11]. The more recent structure of the mammalian Shaker family potassium channel Kv1.2 seems to be in a better agreement with the available biochemical and functional data . Long et al. came to the conclusion that the voltage sensors are essentially independent domains inside the membrane and that (in the open state) two of the four Arg residues on S4 face the lipid, while the other two are buried in the sensor .
By modifying the biophysical hydrophobicity scale of amino acids, as devised by Wimley and White , and taking into account specific positions of charged amino acids along an α helix, Hessa et al. showed that the free-energy cost for insertion of charged α helices like S4 is not as large as previously thought [14, 15]. In our preliminary sequence analyses, we noticed that S4 segment exhibits homologous pattern of positive charge distribution (83%, as opposed to just 25% identity in amino acids) with the cell-penetrating peptide (CPP) penetratin (Table 1), which has been shown to translocate through cell membrane [2, 16]. While S4 and other CPPs share the property of being positively charged, there are several differences: 1) basic residues in S4 are more regularly distributed, i.e., every third residue; 2) S4 assumes a specific α-helical conformation, and 3) S4 stably partitions in the membrane, while CPPs often perturb the membrane integrity. In the present work we quantitatively characterized the interaction of a synthetic voltage sensor motif with the model lipid membrane, identified the key residues on the peptide that are necessary for this interaction, and specified membrane properties that favor the insertion and/or movement of the voltage sensor domain. We have chosen to employ an in vitro system—model membranes made of chemically defined lipids—which allows for better control of variables, but we believe that our observations are relevant to native plasma membrane in intact cells. It is hoped that our results will contribute to better understanding of the interaction between the native S4 segment and lipid membranes in voltage gating, and perhaps also to emergence of a unifying mechanism for membrane partitioning of polycationic macromolecules.
Synthetic peptides used in our experiments are listed in Table 1. The fluorescent analog of S4 and the mutant peptides with >95% purity were obtained from either Abgent (San Diego, CA) or Peptide2.0, Inc. (Chantilly, VA). Lipids egg PC (L-α-phosphatidylcholine), egg PG (L-α-phosphatidylglycerol), DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine), DMPG (1,2-dimyristoyl-sn-glycero-3-phospho-(1′-rac)-glycerol), brominated lipids (6,7)Br2-PC (1-palmitoyl-2-stearoyl-6,7-dibromo-sn-glycero-3-phosphocholine), (9,10)Br2-PC (1-palmitoyl-2-stearoyl-9,10-dibromo-sn-glycero-3-phosphocholine) and (11–12)Br2-PC (1-palmitoyl-2-stearoyl-11,12-dibromo-sn-glycero-3-phosphocholine) were obtained from Avanti Polar Lipids (Alabaster, AL). Fluorescence probes DPH (1,6-diphenyl-1,3,5-hexatriene) and calcein were obtained from Invitrogen-Molecular Probes (Carlsbad, CA). Various salts and buffers were obtained either from Sigma-Aldrich or Invitrogen at highest purity.
Small unilamellar vesicles (SUV) were prepared by sonication . Desired amounts of lipids were dissolved in chloroform and dried under a stream of nitrogen. The resulting dry thin film was hydrated in 0.5 ml buffer, so that the stock concentration of lipid was 13 mM. This lipid suspension was sonicated for 30 min (PC vesicles) or 10 min (PG-containing vesicles) at 4 °C with Branson Sonifier S-150D (Branson Ultrasonics Corporation, Danbury, CT). Large unilamellar vesicles (LUV) were obtained by extrusion, using LiposoFast extruder. The lipid suspension was vortexed, frozen and thawed five times and extruded 21 times through stacks of two polycarbonate membranes (Whatman Nucleopore, Clifton, NJ) with a pore diameter of 100 nm. For calcein loaded vesicles, calcein was dissolved in distilled water and pH adjusted to 7.4, such that the nominal calcein concentration in solution to hydrate the lipid film was 80 mM. The lipid suspension was sonicated as described above and the un-entrapped calcein was removed by passing 0.1 mL of the SUV suspension through Sephadex G-25 column (1 cm × 30 cm) eluted with HEPES buffer. The vesicles thus obtained were immediately used for experiments or stored at 4 °C for at most 48 hrs.
Fluorescence measurements were performed with an ISS K2 fluorometer (Champaign, IL) equipped with a xenon lamp, variable slits, and a microprocessor-controlled photomultiplier in 1 cm × 1 cm or 0.3 cm × 0.3 cm quartz cuvettes. The excitation and emission wavelengths were 280 nm and 360 nm, respectively, for tryptophan; 355 nm and 430 nm for DPH; and 490 nm and 515 nm for calcein. When excitation was in the UV region, a 305 nm high-pass filter placed on the emission side was used to reduce light scattering from the vesicles. Titrations were carried out at room temperature and repeated 5 times. The data were corrected for dilution, and analyzed using Microcal Origin 7.0 (Microcal Software, Northampton, MA).
Fluorophore concentrations were determined with Beckman Coulter DU 800 UV/Vis Spectrophotometer (Beckman Coulter, Inc., Fullerton, CA), using the following molar extinction coefficients at the corresponding absorption maxima: 5,600 cm−1M−1 for tryptophan; 81,200 cm−1M−1 for calcein; and 88,000 cm−M−1 for DPH. The same spectrophotometer was used to record absorption spectra of fluorescently labeled lipid vesicles used in FRET measurements.
Fluorescence anisotropy was measured with the same fluorometer in the L format with the Glan-Thomson prism polarizers placed in the excitation and emission paths. Data were collected in 5 individual determinations, each with 50 iterations and fitted with a single hyperbola (Eq. 1):
where r is the fluorescence anisotropy, x is the concentration of lipid, Δrmax is the maximum change in fluorescence anisotropy at saturating concentration of lipid, Kd is the apparent dissociation constant and r0 is the initial fluorescence anisotropy. The dissociation constant values were used to calculate the free energy change (ΔG) of the peptide/lipid interaction using Equation 2 :
where R is the gas constant and T is the absolute temperature.
Tryptophan-labeled peptides in the NaCl-HEPES buffer were mixed with or without 100 μM of PC/PG (7:3) SUV and fluorescence intensity was measured at increasing concentration of acrylamide (up to 0.2 M). The data were analyzed using the Stern-Volmer equation, Eq. 3, with the intrinsic protein fluorescence multiplied by the factor 10ε[Q]/2 to correct for the acrylamide inner filter effect, using an extinction coefficient ε of 4.3 cm−1M−1 for acrylamide at 280 nm :
where, F/F0 is the ratio of quenched and unquenched fluorescence intensities, [Q] is the molar concentration of the quencher, and KSV is the Stern-Volmer quenching constant.
For quenching with brominated lipid, the lipid vesicles (SUVs) were prepared with 70% dibromo-PC lipids, with bromine at various positions of the acyl chain. The peptide to lipid ratio was maintained at 1:400. The lipid-induced decrease in the tryptophan fluorescence intensity was used for data analysis.
RET was measured with the same fluorometer. Tryptophan-labeled peptides were titrated with SUV labeled with 0.5% DPH. Efficiency E of energy transfer between the two fluorophores was calculated using the Eq. 4 :
where AA(λD) is the absorbance of the acceptor at the donor excitation wavelength, AD(λD) is the absorbance of the donor at its excitation wavelegth, IAD(λD) is fluorescence intensity of the acceptor excited at the donor wavelength in the presence of the donor, and IA(λD) is fluorescence intensity of the acceptor excited at the donor wavelength in the absence of the donor.
Data are expressed as mean ± SE. Statistical differences between means were analyzed using a paired or unpaired Student’s t test. A value of P less than 0.05 was considered significant. All data analysis was performed using Origin 6 software (Origin Labs, Northampton, MA).
Earlier work by others indicated that S4 accomplishes voltage sensing by relocating its position relative to the membrane in response to voltage change. S4 is cationic at neutral pH, and therefore its movement within the lipid membrane poses a significant thermodynamic challenge . Previously, Hessa et al. showed that the S4 segment of a truncated Shaker K+ channel inserted into the endoplasmic reticulum membrane via translocon-mediated integration . We designed a series of in vitro experiments using synthetic peptides and model lipid membranes, which bypasses the necessity for in vitro translation and microsome preparations, and reexamined the inherent ability of S4 to partition in the membrane.
Solvent accessibility studies have suggested a direct contact of S4 with the hydrophobic core of the lipid bilayer . We examined this with fluorescence spectroscopy, using a tryptophan-labeled synthetic S4 peptide from shaker K+ channel S4. The use of the polarity-sensitive fluorophore tryptophan (W) obviated introducing non-amino acid moiety into the peptide.
Various peptides were designed (Table 1) in order to probe the significance of the specific arrangement of amino-acid residues in the S4 transmembrane segment. To test for the possibly different propensities of the two termini to interact with the membrane we used peptides with tryptophan at C terminus (S4-W), N terminus (W-S4), and in the center (S4(F9W, W18F)). To test whether the regular spacing of cationic amino-acid residues at every 3rd position throughout the peptide affects the membrane interaction, one or more arginine residues were shifted in the sequence by one (S4-W (R10 −1)) or two positions (S4-W (R10 −2), (S4-W (R7,10 −2), S4-W (R4,7,10 −2), and S4-W (R4,7,10,13 −2)). Finally, to determine the role of helicity in the peptide-membrane interaction we disrupted the helix by placing proline in the center of the sequence (S4-W (F9P)).
Lipid-induced changes in the anisotropy of the tryptophan fluorescence were used to measure binding of the peptides to the membrane. Increasing concentrations of SUV were sequentially added to a fixed concentration of the peptide (3 μM), as shown in Fig. 1. Binding of fluorescent peptides to lipid vesicles decreases the fluorophore’s diffusion and thus increases its fluorescence anisotropy. Values of the apparent dissociation constant (Kd) and the maximum change in anisotropy (Δrmax) were extracted from data using Eq. 1, and values of apparent dissociation constant (Kd) thus obtained were used to estimate the apparent Gibbs free energy change (ΔG) of the S4W interaction with the PC/PG membrane according to Eq. 2. As shown in Fig. 1, altering the membrane composition had a significant effect on the interaction of the peptide with the lipid membrane. Inclusion of 30 mol% of the acidic phospholipid phosphatidylglycerol (PG) in the membrane (i) increased the peptide’s affinity (the value of ΔG changed from −4.0 ± 0.9 kcal/mol to −5.2 ± 0.2 kcal/mol (the latter is an average of the two peptides with tryptophan on the two opposite termini, see Table 1); and (ii) greatly increased the observed change in the peptide’s fluorescence anisotropy, indicating that S4 is more tightly immobilized in the negatively charged membrane.
The obtained values of ΔG were compared with the theoretical ones, calculated using the Wimley-White interfacial (IF) hydrophobicity scale  (http://blanco.biomol.uci.edu/mpex/). Theoretical ΔG values for transfer of S4-W from water to bilayer in the extended and α-helical conformations were +0.22 kcal/mol and −6.98 kcal/mol, respectively. Thus, the ΔG values obtained from our data are in good agreement with the Wimley-White prediction for the peptide in the α-helix form.
To investigate possible effects of the spatial arrangement of the charged residues along the peptide sequence on the membrane interaction, fluorescence anisotropy changes were quantified for each peptide from Table 1. Because of the logarithmic relationship between ΔG and Kd, variations in ΔG (Table 2), were not conspicuous, but they were statistically significant nevertheless. The extent of anisotropy change (Δrmax), while following the same trend as for ΔG, varied more conspicuously (Fig. 1b). The latter variations indicate that in addition to the overall affinity of the peptide helices towards the membrane (binding energetics, ΔG), also the location, orientation and/or mobility of the bound peptides with respect to the membrane (Δrmax) depend on the peptides’ sequences. Switching the tryptophan from the C terminus (S4-W) to the N terminus (W-S4) or to the center (S4(F9W, W18F)) did not result in significant changes (0.14 ± 0.02, 0.14± 0.02, and 0.16 ± 0.02, respectively). However, a change in the spatial arrangement of the cationic residues (S4-W(R10−1), S4-W(R10−2), S4-W(R7,10−2), S4-W(R4,7,10−2), S4-W(R4,7,10,13−2) and S4-W(Rregain−2)) has brought about a decrease in values of Δrmax (0.09 ± 0.01, 0.05 ± 0.02, 0.09 ± 0.02, 0.11 ± 0.01, 0.12 ± 0.02, and 0.14 ± 0.01, respectively). The most significant changes were observed in peptides where only Arg10 (S4-W(R10−1), S4-W(R10−2)) or Arg7 and Arg10 (S4-W(R7,10−2)) were shifted. When three or four arginines were shifted by two positions, changes ceased to be statistically significant. Reverting to the original amino-acid sequence brought fluorescence anisotropy back to the original value. Disruption of the peptide’s helical conformation (S4-W (F9P)) also resulted in a significant decrease in the anisotropy change (0.05 ± 0.02). These results suggest that both the spatial arrangement of cationic amino-acid residues and the helical confirmation are important for the S4-membrane interaction.
None of the peptides breached membrane integrity even at high peptide/lipid ratios (1:30, data not shown), as evidenced by no significant dye release from the calcein-loaded SUV in a calcein-release assay .
Sensitivity of tryptophan fluorescence to the polarity of its environment was used to gain information on peptides’ localization in the membrane. In less polar (or more hydrophobic) environment, tryptophan emission spectrum exhibits blue shift, i.e., a shift of the spectrum to shorter wavelengths, compared to tryptophan in aqueous solvent . Fluorescence spectra of S4 peptides were recorded at increasing concentrations of PC/PG SUV (Fig. 2a) and magnitudes of the blue shift were evaluated with respect to the spectrum in the absence of the lipid. At saturation, the native peptide S4W exhibited a 14 ± 2 nm blue shift upon binding to the membrane. Replacing egg PC with sphingomyelin gave similar results (data not shown), which suggests that the chemical nature of the neutral lipid is not relevant to the interaction. Location of tryptophan at the N terminus or C terminus also appeared to be irrelevant: peptides S4-W and W-S4 showed similar blue shifts (14 ± 2 nm and 13 ± 2 nm, respectively). However, the peptide S4(F9W, W18F), with tryptophan in the center of the helix, showed a significantly greater blue shift (21 ± 3 nm). This suggests that the helix may be bent, with the center buried deeper in the lipid membrane than the ends. The mutated peptides S4-W(R10−1), S4-W(R10−2), S4-W(R7,10−2), S4-W(R4,7,10−2), S4-W(R4,7,10,13−2) and S4-W(Rregain−2) showed the following values of blue shift: 10 ± 1 nm, 4 ± 2 nm, 12 ± 2 nm, 12 ± 1 nm, 11 ± 2 nm, and 14 ± 1 nm, respectively. Only shifting of Arg10 produced a significantly smaller blue shift, indicating that the regular pattern of basic amino acid residues in the center of the peptide is more important than that at the termini. Also, disruption of the peptide’s helicity by placing a proline in the center (S4-W(F9P)) resulted in a significantly smaller blue shift, 3 ± 1 nm.
The above data show that the synthetic S4 peptide spontaneously binds to the lipid membrane. Furthermore, the blue shift indicates that at least a portion of the peptide is in contact with the non-polar lipid core. To determine the depth of the peptide’s penetration into the membrane we employed several experimental approaches. First, we measured exposure of the membrane-bound peptide to water by quenching its fluorescence with the water-soluble quencher acrylamide . Fig. 3 shows that the Stern-Volmer quenching constants (KSV, Eq. 4) for S4-W in the absence and presence of lipid were 27 ± 3 M−1 and 10 ± 1 M−1, respectively. The lipid partially shielded the C-terminal tryptophan against acrylamide quenching. To compare aqueous accessibility of various peptides, we calculated the ratio of quenching constants in the presence and absence of lipid (KSV(lipid)/KSV(no lipid)) for each peptide. The ratios for S4-W, W-S4, and S4(F9W, W18F), were found to be 0.37 ± 0.02, 0.41 ± 0.03 and 0.28 ± 0.02, respectively (Fig. 3b). The two termini are shielded from water to the same extent, but the center of the peptide is shielded more. This confirms the conclusion based on the blue-shift data. These results are not consistent with the transmembrane orientation of the peptide, where the two termini would not be quenched to the same extent by a quencher applied from one side of the membrane. Even if the peptides did span the membrane in a random orientation, there would always be a fraction of fluorophores that would be unquenchable (on the other side from the side with the quencher), which would result in curved Stern-Volmer plots in Fig. 3A. Our results thus suggest that S4 lies in the membrane more or less parallel with the membrane surface.
The KSV(lipid)/KSV(no lipid) ratios for the peptides with arginine shifts, S4-W(R10−1), S4-W(R10−2), S4-W(R7,10−2), S4-W(R4,7,10−2), S4-W(R4,7,10,13−2) and S4-W(Rregain−2), were found to be 0.54 ± 0.03, 0.78 ± 0.09, 0.49 ± 0.04, 0.41 ± 0.04, 0.44 ± 0.05, and 0.39 ± 0.03, respectively. The values indicate that peptides in which only Arg10 and Arg 7 are shifted retain more exposure to water in the presence of lipid than the peptides in which the flanking arginines (4 and 13) are shifted as well. And again, the peptide with disrupted helicity (S4-W(F9P)) showed very high KSV(lipid)/KSV(no lipid) ratio (0.75 ± 0.10). These result indicate that two peptides, S4-W(R10−2) and S4-W(F9P), remain most exposed to water upon binding to the membrane; they probably lie on the membrane peripheral. The other peptides, with the quenching ratios of less than 0.5, are more protected from aqueous quenching they may be shielded or buried below the lipid head groups, at the level of phosphate or perhaps deeper.
To confirm this, we determined the relative location of different residues in the synthetic S4 peptide by RET. The membrane was labeled with the hydrophobic fluorophore DPH, which is a good acceptor of energy from tryptophan. Assuming that most of the DPH molecules lie in the bilayer core, energy transfer efficiency E is inversely proportional to the depth of the peptide’s penetration in the membrane. The native peptide S4W showed high efficiency, 88 ± 6% (Fig. 5). A comparison between RET efficiencies of all peptides is shown in Fig. 4b. The observed pattern was the same as in the previous experiments described above: the peptides that exhibited the lowest RET efficiency were again those with shifted Arg10 and disrupted helicity, (S4-W(R10−2) and S4W(F9P), with efficiencies of 38 ± 14% and 34 ± 12%, respectively. This indicates that tryptophan in these peptides lies the farthest from the center of the lipid bilayer.
Since both donor and acceptor molecules diffuse in the membrane, we did not attempt to translate efficiency values into distances. Rather, we confirmed these qualitative conclusions by quenching with lipidic quenchers that have quenching groups at defined depths in the hydrophobic core, such as brominated lipids (70% dibromo-PC and 30% PG) . Bromine is a known quencher of tryptophan fluorescence and the approximate distance of bromines from the membrane surface is 9 Å for 6,7-dibromo-PC, 12 Å for 9,10-dibromo-PC and 14 Å for 11,12-dibromo-PC. For the native peptide S4W the extents of quenching with 6,7-dibromo-PC, 9,10-dibromo-PC and 11,12-dibromo-PC were 74 ± 4%, 52 ± 4%, and 26 ± 7%, respectively (Fig. 5a). The highest quenching with the lipid that is brominated at carbon positions 6 and 7 indicates that the peptide’s C terminus (where the tryptophan is located) lies just below the polar head groups, no more than 9 Å from the bilayer surface. Values of quenching by 6,7-dibromo-PC of all peptides are in Fig. 5b. The familiar pattern emerged again, confirming the previous conclusions: the peptides with shifted Arg10 and with disrupted helicity showed the smallest degree of quenching, indicating that tryptophan in these peptides lies closer to the surface of the bilayer than tryptophan in the other peptides, where all, or almost all basic residues were shifted in phase. The latter peptides penetrate deeper below the phosphate, into the hydrophobic core, similarly to the native peptide S4-W.
The fluorescence anisotropy data have shown significant differences in interactions of the peptide with membranes of different lipid composition. We noticed a difference in S4-W interaction with addition of 30% of anionic lipids in the membrane (Fig. 1a). To maintain the physiologic relevance we limited the amount of anionic charge in the membrane to 30%. Enhanced immobilization of the cationic peptide with increased anionic charge on the membrane, led us to investigate the role of electrostatic forces in the peptide-lipid interaction. High ionic strength can be used to attenuate electrostatic interactions . We used 2 M sodium chloride to shield the electrostatic peptide-lipid interactions and thus to probe the involvement of electrostatic interactions in the overall binding of S4 to membranes, using fluorescence spectroscopy, in particular spectral shift and anisotropy. Fig. 6 shows the effect of increased ionic strength on the peptide-lipid interaction: the addition of salt, either before or after adding the lipid, reversed the blue shift to the original value in the absence of lipid. Subsequent quantitative analysis (Fig. 6c) has shown that electrostatics accounted for 54% and non-electrostatics accounted for 46% of the overall change. These results, along with those from varying lipid composition, indicate that in addition to the electrostatics other forces, such as hydrophobic and van der Waals, participate in the peptide binding. This gives support to the hypothesis that the anionic phosphate head groups of the lipid membrane attract a cationic peptide randomly structured in the aqueous phase. Close to the membrane surface, the peptide undergoes a conformational change to alpha helix, which is, more or less reversibly, stabilized in the lipid bilayer. Atomic structure of Kv1.2 revealed that S4 indeed exists in an alpha helical conformation . Many amphipathic peptides also assume α-helical structure upon interaction with lipid membranes [28, 29].
Properties of the membrane, such as, e.g., fluidity (gel state vs. liquid-crystal state), have been shown to affect membrane interactions . In the following experiments we studied the effect of membrane fluidity on S4 binding. We used chemically defined lipids DMPC and DMPG, which have a clear phase transition at around 23 °C [30, 31]. Vesicles prepared from these lipids are in the gel state at temperatures below 23 °C, and in the liquid-crystal state above 23 °C. Fluorescence anisotropy changes (Fig. 7a) in S4-W were significantly higher when the peptide bound to the fluid membrane than when it interacted with the membrane in the gel state (0.12 ± 0.02 and 0.03 ± 0.01, respectively). Apparent binding free energy ΔG paralleled the trend: in the gel state it became much less negative, −2.2 ± 0.4 kcal/mol.
To ascertain that this is solely an effect of the physical state of the membrane, we modulated the latter by means other than temperature, namely, by doping the membrane with cholesterol. Addition of 33% cholesterol, the main constituent of lipid rafts , decreases the fluidity of the membrane and broadens or abolishes the phase transition. As shown in Fig. 7b, binding of S4-W to cholesterol-containing membranes brought about a significantly lower change in anisotropy (0.029 ± 0.004) than binding to cholesterol-free membranes (0.14 ± 0.02). In addition to fluorescence anisotropy, all the previously used parameters were determined and compared for interactions of S4-W with cholesterol-free and cholesterol-containing membranes (Fig. 7c): blue shift (14 ± 2 vs. 3.1 ± 1.2), acrylamide quenching (the KSV(lipid)/KSV(no lipid) ratio of 0.37 ± 0.03 vs. 0.84 ± 0.05), efficiency of RET to DPH (88 ± 7 % vs. 24 ± 12 %), quenching with 6,7-dibromo-PC (74 ± 6% vs. 18 ± 8 %), and ΔG (−5.2 ± 0.2 kcal/mol vs. −2.5 ± 0.3 kcal/mol). Since Kd and ΔG are measures of binding affinity or “strength”, our results indicate that liquid-crystal state of the membrane is necessary for efficient binding and membrane partitioning of the S4 peptide. The other measured parameters, such as the maximum fluorescence anisotropy change and the magnitude of blue shift, RET and quenching, do not report on thermodynamics of binding, but rather on different “modes” of binding, as for instance, the depth of membrane penetration and the microenvironment of the tryptophan as reporting moiety. We note that while all the studied peptides bind to the PC/PG membrane, they do so with various affinities and “modes”, depending on the amino acid sequence.
The charged segment S4 is thought to sense membrane potential by moving its charges in the electric field in the membrane. Evidence for the S4 movement, albeit quite indirect, is based on residue accessibility studies , gating current measurements in the presence of pore-blocking toxins [33, 34], and distance determinations using FRET and LRET . More recently, S4 segment in a mutant variety of shaker has been shown to have proton channel and omega-current activity . These results were consistent with both of the competing models of S4 movement in the membrane, namely: the voltage-sensing paddle model and the helical screw model . While the paddle model predicts, or indeed requires, the direct contact between the charged S4 segment and the hydrophobic tails of the lipid molecules, the helical screw model predicts that the S4 segment moves perpendicularly to the plane of the membrane within a proteinaceous sheath of the channel wall. Despite extensive efforts, a comprehensive picture of the voltage sensor movement has not yet emerged. A major setback in elucidating the mechanism was the fact that channels with mutated arginines in the middle of S4 segment were either non-functional or did not even express on the cell membrane [38, 39]. Our work provides experimental evidence that the autonomous voltage-sensing peptide S4 of the potassium channel spontaneously binds and partitions in the lipid membrane, without any assistance from the rest of the channel protein. S4 was found to spontaneously penetrate into the negatively charged lipid membrane, with the termini about 9 Å below the membrane surface. The peptide’s high positive charge at the neutral pH does not seem to pose an obstacle, as long as the periodicity of 3 is maintained in the distribution of the basic amino acids along the S4 sequence. This periodicity distributes the positive charges unequally on the two sides of the alpha helix.
While our work was in progress, Doherty et al.  published their results of solid-state NMR determination of KvAP S4 orientation in the lipid membrane. According to their model, both termini of the peptide are close to the opposing membrane surface, such that S4 spans the thinned lipid bilayer in a tilted orientation identical to that observed in the crystal structure of the whole voltage-sensing domain. Our data are consistent with this model and further support the notion that the orientation of S4 is determined by peptide-lipid interaction, with minimal, if any, influence of the rest of the channel protein.
To summarize, in the present work we characterized binding of autonomous S4 to lipid membranes. The apparent binding free energy (5.2 ± 0.2 kcal/mol for PC/PG SUV) estimated from the apparent dissociation constant, which, in turn, was determined from lipid-induced changes in anisotropy of tryptophan fluorescence, is not much different from the value calculated using the Wimley-White hydrophobicity scale for binding of an α-helical peptide to the lipid (6.98 kcal/mol). The peptide/membrane interactions have both electrostatic and non-electrostatic components. The peptide lies more or less parallel to the membrane surface, in the depth of about 9 Å. A cartoon model of the S4-membrane interaction consistent with our data is in Fig. 8. Important characteristics of the two ligands (i.e., the membrane and the peptide) that influence binding and penetration of S4 into the membrane include: the fluidity, surface charge, and surface pressure of the membrane, and the α-helicity and the regular spacing of the basic aminoacid residues along the S4 sequence.
An important question that our work has left unanswered is the movement of S4 in the membrane in response to changes in electric potential across the membrane. We note that the in vitro system employed in this study is suitable for addressing this issue .
We thank members of the Li laboratory for valuable discussions and comments over the course of this investigation. This work was supported by grants GM078579 and MH084691 from the National Institutes of Health (NIH).
Role of the funding source
NIH did not play a role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.