A Synthetic S6 Segment Derived from KvAP Channel Self-assembles, Permeabilizes Lipid Vesicles, and Exhibits Ion Channel Activity in Bilayer Lipid Membrane*

KvAP is a voltage-gated tetrameric K+ channel with six transmembrane (S1–S6) segments in each monomer from the archaeon Aeropyrum pernix. The objective of the present investigation was to understand the plausible role of the S6 segment, which has been proposed to form the inner lining of the pore, in the membrane assembly and functional properties of KvAP channel. For this purpose, a 22-residue peptide, corresponding to the S6 transmembrane segment of KvAP (amino acids 218–239), and a scrambled peptide (S6-SCR) with rearrangement of only hydrophobic amino acids but without changing its composition were synthesized and characterized structurally and functionally. Although both peptides bound to the negatively charged phosphatidylcholine/phosphatidylglycerol model membrane with comparable affinity, significant differences were observed between these peptides in their localization, self-assembly, and aggregation properties onto this membrane. S6-SCR also exhibited reduced helical structures in SDS micelles and phosphatidylcholine/phosphatidylglycerol lipid vesicles as compared with the S6 peptide. Furthermore, the S6 peptide showed significant membrane-permeabilizing capability as evidenced by the release of calcein from the calcein-entrapped lipid vesicles, whereas S6-SCR showed much weaker efficacy. Interestingly, although the S6 peptide showed ion channel activity in the bilayer lipid membrane, despite having the same amino acid composition, S6-SCR was significantly inactive. The results demonstrated sequence-specific structural and functional properties of the S6 wild type peptide. The selected S6 segment is probably an important structural element that could play an important role in the membrane interaction, membrane assembly, and functional property of the KvAP channel.

Ion channels belong to a large family of proteins that catalyze the diffusion of inorganic ions down their electrochemical gra-dients across the cell membranes (1)(2)(3). It is known that electric fields exist or arise in living cells and tissues affecting conductance of ion channels (1)(2)(3). These ion channel proteins are involved in versatile physiological activities and are responsible for all electrical signaling in living creatures (4 -6). Ion channels have been named according to the name of ions that they allow to diffuse through the pores they form. For example, K ϩ channels are responsible for the diffusion of K ϩ ions and constitute an important class of ion channels (7)(8)(9). K ϩ channels modulate the resting potential and action potential duration of neurons, myocytes, and endocrine cells and stabilize the membrane potential of excitable and non-excitable cells (10 -13).
K ϩ channels that are activated by membrane depolarization are termed voltage-gated K ϩ (Kv) 4 channels. These Kv channels are tetramers containing six transmembrane segments and one pore region in each monomer. The first four transmembrane segments, S1-S4, form a module that somehow controls the opening and closing of the pore. The fifth (S5) and sixth (S6) transmembrane segments along with the H5 segment form the pore region (14). The S5 and S6 segments form the pore helix of the K ϩ channel, delineating the outer and inner linings of the pore, respectively (7,(15)(16)(17)(18)(19). The role of the pore region of K ϩ channels in the selectivity of K ϩ ions has been shown by several point mutations. For example, a single point mutation in the S6 segment (A463C) of the Shaker K ϩ channel decreased the internal K ϩ affinity by a factor of ϳ1,000 in the millimolar range of K ϩ ions (20). Hackos et al. (21), by using a scanning mutagenesis approach, identified a PVP motif in the Shaker K ϩ channel S6 segment and suggested that its intracellular portion could form an activation gate.
Membrane interaction and assembly are the key steps associated with the functional activity of K ϩ channels. It is believed that both protein-protein (e.g. interaction of a channel protein with itself or other membrane proteins) and protein-lipid interactions are involved in the appropriate assembly of any ion channel including K ϩ channels in the membrane (7,15,18,19). Despite a huge amount of work done on K ϩ channels or ion channels in general, very little is known about the structural biology of membrane interaction of these channel proteins. This is due to the inherent complications in purification of the membrane proteins and their crystallization in the presence of lipids. Synthetic peptides are being used to dissect structuralfunctional relationships in ion channels (22)(23)(24)(25)(26)(27). However, how the different conserved segments of a K ϩ channel participate in the membrane interaction and contribute to the assembly and function of the whole protein is still not very well studied. Toward this end, we synthesized a 22-residue segment comprising the amino acid (AA) region 218 -239, which belongs to the S6 transmembrane region of the KvAP channel. In addition, a scrambled peptide (S6-SCR), which is the same size and has the same amino acid composition of the wild type peptide but with some reorganization in the sequence of hydrophobic amino acids, was also synthesized to investigate the importance of the specific amino acid sequence in the structure and activity of the synthetic S6 segment. The phospholipid membrane interaction and membrane assembly of the synthetic S6 segment and its analog were studied after labeling the peptides with the fluorescent probes 7-nitrobenz-2-oxa-1,3-diazole (NBD) and rhodamine (Fig. 1B). In addition, bilayer lipid membrane (BLM) electrophysiological studies were performed to explore ion channel activity of these peptides.
Peptide Synthesis, Fluorescent Labeling, and Purification-The peptides were synthesized manually on solid phase. Stepwise solid phase synthesis was carried out on rink amide MBHA resin (0.15 mmol) utilizing the standard Fmoc chemistry using a di-isopropylcarbodiimide/1-hydroxybenzotriazole or 1,1,3,3tetramethyluronium tetrafluoroborate/1-hydroxybenzotriazole/N,NЈ-diisopropylethylamine coupling procedure (28 -31). Deprotection of the ␣-amino group and the coupling of amino acids were checked by Kaiser test (32) for primary amines. After the synthesis was over, each peptide was cleaved from the resin with simultaneous deprotection of side chains by treatment with a mixture of TFA/phenol/thioanisole/1,2-ethanedithiol/ water (82.5:5:5:2.5:5, v/v) for 6 -7 h. Labeling at the N terminus of the peptides was achieved by a standard procedure reported earlier (33)(34)(35). In brief, 15-20 mg of a resin-bound peptide was treated with 25% piperidine (in N,NЈ-dimethylformamide) to remove the Fmoc group from the N-terminal amino group. The resin was washed and dried. Then Fmoc-deprotected resinbound peptides were incubated with tetramethylrhodamine succinimidyl ester (2-3 eq) in dimethylformamide in the presence of 5% diisopropylethylamine for 48 -72 h, which ultimately resulted in the formation of N ␣ -Rho-peptides. Similarly, resin-bound peptides were treated with 4-fluoro-7-nitrobenz-2-oxa-1,3-diazole (2-3 eq) to obtain N ␣ -NBD-peptides. After sufficient labeling, the resins were washed with N,NЈ-dimethylformamide and dichloromethane to remove the unreacted probe. The peptides were cleaved from the resin as above and precipitated with dry ether. All the peptides were purified by reverse phase HPLC on an analytical Vydac C 4 column using a linear gradient of 0 -80% acetonitrile in 45 min with a flow rate of 0.6 ml/min. Both acetonitrile and water contained 0.05% TFA. The purified peptides were ϳ90% homogeneous as shown by HPLC. The detected molecular masses of the peptides by electrospray MS analysis were close to their desired values.
Preparation of Large Unilamellar Vesicles-Large unilamellar vesicles were prepared by a standard procedure (33,35,36) as follows. Dry lipids containing the required amounts of PC/PG (1:1, w/w) were dissolved in CHCl 3 /CH 3 OH (2:1, v/v) in a small glass vial. Solvents were evaporated under a stream of nitrogen, resulting in the formation of a thin film on the wall of the glass vessel. Films were also dried overnight under vacuum to remove traces of solvents. The thin film was resuspended in buffer at a concentration of 8.2 mg/ml by vortex mixing. The lipid dispersions were then sonicated in a bath-type sonicator (Laboratory Supplies Co.) for 10 -20 min until they became transparent. The lipid concentration was determined by phosphorus estimation (37).
Circular Dichroism (CD) Experiments-The CD spectra of peptides were recorded in phosphate-buffered saline (PBS, pH 7.4), 1% SDS, and ϳ400 M PC/PG lipid vesicles by utilizing a Jasco J-710 spectropolarimeter, averaged over three scans, and base line-corrected. Noise reduction was performed by using the manufacturer's software. The spectropolarimeter was calibrated routinely with 10-camphorsulfonic acid. The samples were scanned at room temperature (ϳ30°C) with the help of a capped quartz cuvette of 0.2-cm path length in the wavelength range of 250 -195 nm. An average of four to six scans was taken for each sample with a scan speed of 20 nm/min and a data interval of 0.5 nm for a peptide concentration of 10 -20 M. The fractional helicities were calculated using the formula ( 100 and 0% helix contents were considered to have mean residue ellipticity values of Ϫ32,000 and Ϫ2,000 respectively at 222 nm (40).
Membrane Binding Experiments-The affinity of the peptides for phospholipid vesicles was determined by binding experiments as reported earlier (32,(41)(42)(43). In brief, large unilamellar lipid vesicles were added gradually to a freshly dissolved NBD-labeled peptide of ϳ0.15 M concentration at room temperature. Fluorescence intensities of NBD-S6 alone and after each addition of lipid vesicles were recorded on a PerkinElmer Life Sciences spectrofluorometer (model LS-50B) with the excitation and emission wavelengths of 467 and 525 nm, respectively. Because the emission maximum of NBD-S6-SCR was at a longer wavelength, the fluorescence intensity for this peptide was recorded at 530 nm with the excitation wavelength at 467 nm. The excitation and emission slits were fixed at 8 and 6 nm, respectively, for both the peptides. The contributions of lipid to any of the recorded signals were measured by titrating the unlabeled peptide (at the concentration of the NBD-labeled peptide) with the same amount of lipid vesicles and subtracted from the original fluorescence signal. The binding isotherms were analyzed using the equation where X b * is defined as the molar ratio of bound peptide per 60% of the total lipid, assuming that the peptides were initially partitioned only over the outer leaflet of the large unilamellar vesicles as suggested by Beschiaschvili and Seelig (43). K p * represents the partition coefficient, and C f indicates the concentration of the free peptide at equilibrium. X b can be calculated by extrapolating the fluorescence signal F infinity (fluorescence signal when all the peptide molecules are bound to lipid) from a double reciprocal plot of F (peptide fluorescence in the presence of lipid) versus C L (lipid concentration). The fraction of peptide bound (f b ) was determined by the equation where F is the fluorescence of the peptide when it is bound to lipid, and F 0 is the fluorescence of the peptide in its unbound state. When f b is known, C f can easily be calculated for each concentration of the lipid. K p * can easily be determined from the slope of the plot of X b * and C f . The partition coefficient of each of the peptides was determined as the average of the values obtained from two to three independent experiments as described previously (35,40). However, note that there are some limitations in determining the binding isotherm of a peptide by this equation (43). Because X b * is a function of concentration of free peptide, it is desirable to have the X b * values with respect to free peptide concentrations over a large concentration range of a peptide. Due to aggregation of peptide in the aqueous phase, low solubility of the peptide, or peptide-induced damage of membrane organization, sometimes it is difficult to get meaningful X b * values over a broad range of peptide concentration. Therefore, the binding experiments are often limited to a low concentration of peptides. Furthermore, the concentration of free peptide (C f ) in solution is calculated indirectly from its bound fraction.
Enzymatic Cleavage Experiments-To detect the location of the NBD-labeled peptides in their membrane-bound state, enzymatic cleavage experiments were performed as reported earlier (32,44). In brief, PC/PG lipid vesicles were first added to an NBD-labeled peptide. When the major portion of the peptide was bound to the lipid vesicles as detected by the saturation of the fluorescence level, proteinase K (final concentration, 10.0 g/ml) was added. In this experiment, the fluorescence of NBD-S6 was recorded at 525 nm with respect to time (in s) and with the excitation wavelength set at 467 nm. Because NBD-S6-SCR exhibited an emission maximum at longer wavelength, fluorescence of this peptide was recorded at 530 nm with the same excitation wavelength as for the S6 peptide. In the control experiment, proteinase K was first added to NBD-labeled peptides, and then lipid vesicles were added.
Fluorescence Resonance Energy Transfer Experiment-Fluorescence energy transfer experiments were performed with the excitation wavelength set at 467 nm and an emission range of 500 -600 nm. ϳ0.15 M NBD-labeled peptide was taken in a fluorometer cuvette. Then ϳ400 M concentration of the phospholipid vesicles was added to the NBD-labeled peptide to ensure that the peptides were bound to the membrane. Now Rho-labeled acceptor peptide was added to the donor peptidelipid complex. Energy transfer from the donor to acceptor was determined by subtracting the acceptor fluorescence in the presence of lipid and unlabeled peptide from the fluorescence signal obtained in the presence of donor, acceptor, and lipid vesicles.
The efficiency of energy transfer (E) was determined by the decrease in the fluorescence of the donor in the presence of the acceptor as reported earlier (32,35). The percentage of energy transfer was calculated by the equation where I D0 and I DA are the fluorescence intensities of the NBDlabeled donor peptide in the absence and presence of the Rholabeled acceptor peptide, respectively, at the emission maxima of the donor after correcting the light scattering of the lipid vesicles and emission of the acceptor.
Calcein Release from Calcein-entrapped Lipid Vesicles-Calcein-entrapped lipid vesicles were prepared with a self-quenching concentration (60 mM) of the dye in 10 mM HEPES at pH 7.4 as reported earlier (35,36,40,45). Briefly, a thin film of lipid (PC/PG) was resuspended in calcein solution, vortexed for 1-2 min, and then sonicated in a bath-type sonicator. The nonencapsulated calcein was removed from the liposome suspension by gel filtration using a Sephadex G-50 column. Usually lipid vesicles are diluted to ϳ10-fold after passing through a G-50 column. The eluted calcein-entrapped lipid vesicles were diluted further in the same buffer to a final lipid concentration of ϳ3.0 M for the experiment. Peptide-induced release of calcein from the lipid vesicles was monitored by the increase in fluorescence due to the dilution of the dye from its selfquenched concentration. Fluorescence was monitored at room temperature with excitation and emission wavelengths fixed at 490 and 520 nm, respectively. The peptide-induced release of calcein from the calcein-entrapped lipid vesicles was measured in terms of percentage of fluorescence recovery (F t ) as defined by the equation (33) where I t is the observed fluorescence after the addition of a peptide at time t, I 0 is the initial fluorescence of calcein-entrapped vesicles, and I f is the increase in fluorescence observed after the addition of Triton X-100 (0.1% final concentration) to the dye-entrapped vesicle suspension.
Reconstitution of S6 in Planar Lipid Bilayers and Electrophysiological Recording-Peptides were reconstituted into the planar lipid bilayers according to the method of Roos et al. (42) as was also reported earlier (46,47). Briefly, the apparatus consisted of a polystyrene cuvette (Warner Instruments) with a thin wall separating two aqueous compartments containing 500 mM KCl, 5 mM MgCl 2 , and 10 mM HEPES (pH 7.4). The polystyrene divider had a circular aperture with a diameter of 150 m. Aqueous compartments were connected to an integrating patch amplifier (Axopatch 200B, Axon Instruments) through a matched pair of Ag/AgCl electrodes. The cis chamber was connected to the head stage (CV-203BU) of the amplifier, and the trans chamber was held at virtual ground. A solution of diphytanoyl phosphatidylcholine and cholesterol (6:1) in n-decane (10 l) was painted over the aperture to form the membrane. The stock solutions of the peptides were made in water except S6-SCR, which was dissolved in 10% DMSO and later diluted with double distilled water to get the desired concentration. Reconstitution of these peptides in BLM was initiated by adding 50 pg of each of these peptides in BLM buffer. The buffer was homogenized using a magnetic stirrer. Channel current was recorded using Digidata (1440A, Axon Instruments) and the acquisition software CLAMPEX (PCLAMP 10.2., Axon Instruments). Single channel recording of the peptides was per-formed in a symmetric bath solution (1 ml in both cis and trans chambers) at an applied membrane potential in the range of Ϫ200 to ϩ200mV and s sampling frequency 1 kHz using a low pass filter (frequency, 200 Hz).
Analysis of Electrophysiological Data-Steady state conductance (current/voltage) of these peptides was calculated from the single channel current data using the software CLAMPFIT (PCLAMP 10.2, Axon Instruments) and data were analyzed in Origin 5.0 (Originlab Corp.).

RESULTS
Peptide Design-The sequence alignment of the S6 segment comprising the AA region 218 -239 is shown in Fig. 1A. It is clear that appreciable sequence homology in this segment exists among the KvAP and other K ϩ channel proteins. Furthermore, Schiffer-Edmundson wheel projection of the S6 peptide shows the segregation of hydrophobic and hydrophilic amino acids in opposite sides, indicating a considerable amphipathic character of the selected S6 segment (Fig. 1B). Therefore, this segment containing the above amino acid region was selected. To investigate the importance of the specific amino acid sequence in the structure and activity of the selected S6 segment, a scrambled peptide, S6-SCR, of the same size and composition was designed. Because the selected segment contains mostly hydrophobic amino acids, positions of some of these hydrophobic amino acids were interchanged with adjacent less hydrophobic amino acids. In S6-SCR, positions of isoleucine residues at 4 and 11 were interchanged with positions of alanine at 6 and glycine at 12, respectively. The Schiffer and Edmundson presentation of S6-SCR shows that it possesses a significant amphipathic character compared with the wild type peptide, indicating its potential to bind to phospholipid membrane. To determine the assembly of the peptides by energy transfer experiments, these peptides were labeled by energy donor and acceptor fluorescent probes, namely NBD and rhodamine, respectively, at their N termini. AA sequences of the S6 and S6-SCR are shown in Fig. 1C.
S6-derived Peptides Bind to Phospholipid Vesicles-The S6 segment is an important component of the pore formed by a voltage-gated K ϩ channel in the membrane. Therefore, it was of interest to evaluate the phospholipid membrane interaction of the selected S6 segment of KvAP channel along with its analog and study their assembly therein. To detect the ability of the peptides to bind to phospholipid vesicles, these peptides were covalently labeled at the N terminus by a fluorescent probe, NBD. Fluorescence of NBD is sensitive to the dielectric constant of the medium, and this probe has been used extensively to detect the membrane interaction of peptides (34, 35, 41, 48 -50). Fluorescence of NBD-labeled peptides was recorded in PBS (pH 7.4) as well as in the presence of negatively charged PC/PG lipid vesicles. As shown in Fig. 2, NBD-labeled S6 and S6-SCR exhibited broad emission maxima at ϳ545 nm in PBS. This emission maximum indicated the location of the probe in the hydrophilic environment (51). However, in the presence of lipid vesicles, the fluorescence of wild type S6 peptide enhanced appreciably concomitant with a shift of its emission maximum to a shorter wavelength (513-514 nm). This shift of the emission maximum to a shorter wavelength along with the enhancement of fluorescence of NBD-S6 indicated the relocation of the NBD probe in the hydrophobic environment resulting from the binding of the S6 peptide onto the lipid vesicles. Furthermore, the emission maximum of 513-514 nm exhibited by NBD-labeled S6 indicated that the N terminus of the peptide was inserted into the lipid bilayer (52). However, NBD-S6-SCR exhibited a significantly different emission spectrum in the presence of negatively charged phospholipid vesicles as compared with that of the wild type S6 peptide. The emission maximum of NBD-S6-SCR was located at ϳ530-nm wavelength, which was much longer as compared with that exhibited by NBD-S6 peptide (Fig. 2). The results indicated that the N ter-minus of S6-SCR was probably localized closer to the surface of the membrane (52). Perhaps the interchange of positions among these hydrophobic amino acids in S6 peptide altered the localization of its N terminus.
S6-derived Peptides Bind to Phospholipid Vesicles with Appreciable Affinity and Form Aggregates-The affinity of S6 and S6-SCR toward the PC/PG phospholipid vesicles was determined by binding experiments utilizing their NBD-labeled versions. Fluorescence signals of NBD-labeled peptide in the presence of an increasing amount of PC/PG vesicles were plotted with respect to lipid/peptide molar ratio.   NBD-labeled peptides to the lipid vesicles. Conventional binding isotherms were generated by plotting X b * with respect to C f as described under "Experimental Procedures" (Fig. 3B). The low C f value corresponds to the high fraction of bound peptide, indicating a saturated bound state for peptides after which the addition of lipids leads to no significant increase in further binding of peptides to lipid vesicles. Therefore, the low C f values are considered to be appropriate to calculate the partition coefficient (53)(54)(55). Furthermore, at low C f values, which correspond to low peptide to lipid ratios, the amount of membranebound peptide varies linearly with the concentration of free peptide in solution (55). Therefore, partition coefficients of NBD-labeled peptides for the phospholipid vesicles were estimated from the initial slopes of the curves extrapolating to zero C f values (Fig. 3C). The calculated partition coefficient for NBD-S6 peptide was 1.61 ϫ 10 4 M Ϫ1 , which is an appreciable value and similar to the other surface-active peptides derived from antimicrobial peptides, bacterial toxins, and viral fusion proteins (32,41,56). NBD-S6-SCR showed a significant affinity for PC/PG lipid vesicles as indicated by its respective partition coefficient of 4.8 ϫ 10 4 M Ϫ1 (Fig. 3C). The value of the partition coefficient changes with the equilibrium concentration of free peptide in solution as demonstrated by Fig. 3B. The nature of the binding curve also provides an idea about the assembly of a peptide in its membrane-bound state. The binding isotherm of NBD-S6 showed distinct features in terms of changes in X b * values with respect to C f values. At lower C f values, the binding isotherm of NBD-S6 contained an initial "lag" phase, and the curve was flat. However, after the lag phase at a certain threshold value of C f , the X b * value increased sharply as reflected by the sharp upward rise of the binding isotherm curve of NBD-S6. The peptides that show this kind of binding isotherm manifest the cooperative binding process where peptides are first incorporated by a simple partition process into the membrane and then aggregate to form a pore (35,41,42,55,(57)(58)(59). Thus, the binding isotherm of S6 suggests that it formed large aggregates in the presence of negatively charged lipid vesicles, although the size of the aggregate cannot be estimated from this experiment. The self-assembly of S6 supports the proposal that S6 may serve as a structural unit that could assist in the tetramerization of the whole KvAP channel protein. Interestingly, the binding isotherm of NBD-S6-SCR at low C f values showed an initial "lag phase" with a flat curve as observed for NBD-S6 that rose upward at a threshold value of C f . This nature of the binding isotherm indicated that NBD-S6-SCR also aggregated in its membrane-bound state.
However, the nature of the binding isotherm for NBD-S6-SCR in PC/PG lipid vesicles was significantly different from that of NBD-S6. A careful look at these binding isotherms indicated that the upward rise of the binding isotherm of NBD-S6-SCR at a threshold higher C f value was less sharp than that of NBD-S6 (Fig. 3B). This difference in the shape of the binding isotherms revealed that this scrambled S6 analog bound to lipid vesicles with weaker cooperativity as compared with the S6 peptide. In other words, NBD-S6-SCR formed weaker aggregates in its membrane-bound state than the wild type NBD-S6 peptide. Similar inferences have been made from the comparable binding isotherms for other peptides (58, 60). Thus, the results suggest that although NBD-S6-SCR bound to phospholipid vesicles with significant affinity it formed weaker aggregates in comparison with the NBD-S6 peptide.
Detection of Localization of S6 Peptide and Its Analog into Lipid Bilayer-To detect the localization of the S6-derived peptides onto the membrane, proteolytic cleavage experiments were performed with their NBD-labeled versions in their membrane-bound states (32,36,44). The basis of the experiment is that NBD-labeled peptides, bound onto the membrane surface, are easily cleaved by a proteolytic enzyme like proteinase K, which can be monitored by the prompt decrease in NBD fluorescence from the characteristic membrane-bound level. On the other hand, a membrane-inserted peptide will not be accessible to proteinase K, and therefore its NBD fluorescence will not decrease easily. The profiles for proteolytic cleavage experiments with NBD-labeled S6 peptide in the presence of negatively charged PC/PG lipid vesicles are shown in Fig. 4. At time point 1, NBD-labeled peptides were added to the buffer followed by the addition of PC/PG lipid vesicles at time point 2. Addition of lipid vesicles resulted in a sharp increase in NBD fluorescence due to binding of the peptide to the lipid vesicles. When NBD fluorescence reached a plateau, indicating a saturation of binding of the NBD-labeled peptide to the lipid vesicles, proteinase K was added at time point 3. The addition of proteinase K resulted in a slow decrease in NBD fluorescence (Fig. 4, profile a, black scale), indicating that NBD-S6 protected itself to some extent from the digestion by proteinase K. The 50% decrease in NBD fluorescence, resulting from the cleavage of NBD-S6 in its membrane-bound state, took place at ϳ500 s (profile a, black scale), whereas the same for NBD-S6-SCR took ϳ300 s (profile a, gray scale), suggesting that the N terminus of wild type S6 peptide was to some extent more protected from proteolytic cleavage when bound to the membrane as compared with its analog. However, in control experiments when proteinase K was directly added to these NBD-labeled peptides, they were easily cleaved because after the addition of lipid vesicles no significant enhancement in NBD fluorescence, which is and S6-SCR peptides (gray scale), and then proteinase K was added to each of the membrane-bound peptides. 1, 2, and 3 indicate the addition of peptides, lipid vesicles, and proteinase K, respectively. In control experiments (profiles b), proteinase K was first added to the NBD-labeled S6 (black scale) and S6-SCR (gray scale) peptides, and then lipid vesicles were added. In profiles b, 1, 4, and 5 stand for the addition of peptide, proteinase K, and lipid vesicles, respectively. The final concentration of proteinase K was 10 g/ml for profiles a and b. a.u., arbitrary units. characteristic of binding of these NBD-labeled peptides to lipid vesicles, was observed (Fig. 4, profiles b, black and gray scales).
S6 Peptide Adopted Significant Helical Structure in Membrane-mimetic Environments-Circular dichroism experiments were performed with S6 and S6-SCR in PBS, in SDS micelles, and in the presence of PC/PG lipid vesicles. In PBS, none of the peptides showed any appreciable helical structure (hence not shown). However, in the presence of SDS micelles (1% (w/v) in water), the wild type S6 peptide adopted a significant helical structure as evidenced by the characteristic CD spectrum (Fig. 5A). The mean residue ellipticity values of S6 at 222 nm in different environments and corresponding percentage of helicity are shown in Table 1. Also, in the presence of PC/PG vesicles, the wild type S6 peptide showed appreciable helical structure (Table 1). In contrast to S6 peptide, S6-SCR showed appreciably reduced mean residue ellipticity values in the presence of either SDS micelles or PC/PG lipid vesicles (Fig.  5B). The data clearly indicated that substitution of the hydrophobic amino acids in the S6 peptide significantly altered its secondary structure in phospholipid membrane and membrane-mimetic environments.
Difference between S6 Peptide and Its Analog in Self-assembly onto Phospholipid Vesicles-To investigate whether the selected S6 segment could contribute in the assembly of the KvAP channel, self-association properties of S6 and S6-SCR peptides were examined in the presence of phospholipid vesicles by fluorescence energy transfer experiments. To perform the energy transfer experiments, the peptides were labeled by a fluorescence energy donor, NBD, and an energy acceptor, rhodamine, respectively, as reported earlier (32,35,36). Fig. 6, A and B, show the fluorescence spectra of energy transfer experiments between NBD-S6 and Rho-S6 and between NBD-S6-SCR and Rho-S6-SCR, respectively, in the presence of PC/PG lipid vesicles. Fluorescence spectra of NBD-labeled peptides in PC/PG lipid vesicles were recorded in the presence of increasing amounts of corresponding Rho-labeled peptides. As evident from Fig. 6A, the addition of energy acceptor Rho-S6 to the membrane-bound donor NBD-S6 resulted in an appreciable decrease in NBD fluorescence concomitant with the increase in rhodamine fluorescence. This decrease in NBD fluorescence resulted due to energy transfer from the NBD-labeled S6 to Rho-labeled S6 and suggested that S6 peptide molecules selfassembled in the negatively charged PC/PG lipid vesicles. The extent of the decrease in NBD fluorescence for NBD-S6-SCR after the addition of corresponding Rho-labeled peptide was to some extent lower as compared with that observed for NBD-S6 and Rho-S6, suggesting a weaker self-association of S6-SCR peptide onto the lipid vesicles (Fig. 6B). To confirm whether the observed energy transfer was due to the assembly of the peptides, the percentages of energy transfer observed for these two pairs were compared with that of the randomly distributed energy donor and acceptor as described earlier (42,61). Fig. 6C shows such plots, which clearly indicate that the energy transfer efficiency between NBD-S6 and Rho-S6 is much above the random distribution level, whereas the energy transfer efficiency between NBD-S6-SCR and Rho-S6-SCR is marginally above the random distribution. Altogether, the results suggested that the interchange of positions of the selected hydrophobic amino acids in the 22-residue S6 peptide altered its self-association property onto the negatively charged lipid vesicles.
Only S6 Wild Type Peptide Permeabilized Phospholipid Vesicles Appreciably-To determine whether or not the 22-residue peptide comprising the amino acid region 218 -239 can permeabilize the lipid vesicles, release of calcein from calcein-entrapped lipid vesicles in the presence of S6 peptide was measured. Permeabilization of lipid vesicles in the presence of S6 scrambled peptide was also examined. Fig. 7, A and B, show the experimental fluorescence profiles of calcein release induced by S6 and S6-SCR peptides, respectively. A sharp increase in fluorescence (Fig. 7A) after the addition of S6 peptide clearly indicated its significant efficacy in inducing release of calcein from the calcein-entrapped lipid vesicles, which further suggested the ability of the S6 peptide to induce permeation in the negatively charged PC/PG vesicles. In contrast to the S6 peptide, S6-SCR induced a significantly lesser release of calcein from calcein-entrapped PC/PG lipid vesicles (Fig. 7B). Fig. 7C depicts the plots of percentage of fluorescence recovery induced by the  peptides with respect to their concentrations, which is a measure of the ability of a peptide to permeabilize the lipid vesicles. The results (Fig. 7C) clearly indicated a significant efficacy of the selected S6 segment to induce leakage in phospholipid ves-icles; however, S6-SCR with rearranged positions of some of the hydrophobic amino acids but the same amino acid composition as the S6 peptide was not appreciably active in permeabilizing the same kind of lipid vesicles. The data probably reflect the importance of the specific amino acid sequence in maintaining the ability of S6 peptide to permeabilize the phospholipid membrane. The efficacy of this S6 segment to permeabilize the lipid vesicles is comparable with the pore-forming peptides or pore-

Synthetic S6 Segment from KvAP Channel
forming segment of a toxin or ion channel protein (35,(62)(63)(64)(65). Furthermore, fluorescently labeled S6 peptide also induced appreciable calcein release (ϳ90% as compared with its unlabeled version) from PC/PG lipid vesicles, suggesting that labeling of the peptide did not alter its activity.

S6 Peptide Showed Ion Channel Activity in Bilayer Lipid
Membrane-When incorporated in BLM, the selected synthetic S6 segment showed remarkable openings at both positive and negative potentials (ϩ200 mV). Fig. 8, A and C, show the single channel recordings (current versus time) of S6 incorporated into the diphytanoyl phosphatidylcholine bilayer membrane at applied membrane potentials of ϩ100 and Ϫ100 mV along with the corresponding current histograms (Fig. 8, B and  D). The corresponding current versus voltage (I-V) plot, shown in Fig. 9, indicates that although the opening behavior of the S6 single channel is similar in both the positive and negative applied potentials it is not ideally symmetrical. Fig. 9 shows that the I-V relation is nonlinear. The fact that I varied with V nonlinearly showed that S6 peptide formed a weak voltage-dependent ion channel (discussed in the next section). Fig. 10 shows the opening probability as a function of applied membrane potential for the S6 peptide in the range of Ϫ200 to ϩ200 mV. Single channel currents were recorded at membrane potentials in the range of Ϫ200 to ϩ200 mV. f shows the mean current with S.E. for three independent experiments; error bars indicate S.E. Experimental conditions are the same as in Fig. 8. The pattern appears to be bell-shaped, and the opening probability was maximal between Ϫ75 and ϩ75 mV. Fig. 11 shows the variation of single channel conductance of S6 with the bilayer membrane potential. As shown in Fig. 11, the single channel conductance of S6 peptide lies in the range of 200 to 260 pS in the positive membrane potential, whereas the same lies in the range of 100 to 160 pS in the negative applied potential. We optimized the peptide concentration such that we were able to get clear-cut openings that were analyzable. However, when the ion channel activity of S6-SCR was examined like S6 peptide in bilayer lipid membrane at several voltages, significantly different results were observed. S6-SCR showed ion channel activity only at high voltages (Ϯ175 mV) that was much weaker in comparison with wild type S6 peptide (supplemental Fig. 1). The I-V plot for S6-SCR shows that current approaches 0 as the voltage falls below 100 mV (supplemental Fig. 2). The channel opening probability of S6-SCR was low in the range of ϩ125 to ϩ200 as compared with S6 and did not show any significant opening in the ϩ100-mV range (supplemental Fig. 2). The conductance of the channel formed by S6-SCR varied between 30 and 60 pS in both the positive and negative voltage range (supplemental Table 1). These results indicated that the interchange of positions of the hydrophobic amino acids in S6 peptide significantly impaired its ion channel activity. The data further indicated a probable role of the specific sequence of these hydrophobic amino acids in the ion channel activity of S6 peptide.

DISCUSSION
The results presented here show the structural and functional characterization of a 22-residue S6 wild type peptide and its scrambled analog derived from the amino acid region 218 -239 of KvAP channel. Experiments with an NBD-labeled version of the peptide indicated that the S6 peptide as well as its analog bound to negatively charged PC/PG lipid vesicles (Fig.  2). Considering the overall negative charge of bacterial membrane, PC/PG (1:1, w/w) lipid vesicles were chosen as the model membrane as also used by others (35, 66 -68). However, these peptides can also bind to zwitterionic PC/cholesterol (8:1, w/w) lipid vesicles (data not presented). The emission spectrum of NBD-S6 (Fig. 2) exhibited a maximum at a wavelength of 513-514 nm, indicating that the location of the N terminus of the peptide was appreciably inside the lipid bilayer. This was further supported by the evidence that NBD-S6 in its membranebound state was not cleaved easily by a proteolytic enzyme, proteinase K (Fig. 4). However, the scrambled analog exhibited an emission maximum in the range of ϳ530 nm (Fig. 2) in the presence of the same kind of lipid vesicles, suggesting that the interchange of positions of some of the hydrophobic amino acids in S6 peptide significantly abrogated the localization of its N terminus onto the lipid vesicles. Besides, proteinase K cleaved this scrambled NBD-labeled peptide in its membranebound state to some extent faster than the parent peptide. Titration with lipid vesicles indicated that NBD-S6 peptide bound to PC/PG lipid vesicles with appreciable affinity (Fig. 3,  A, B, and C). Moreover, the binding isotherm of NBD-S6 (Fig.  3B) showed a sharp upward rise at a higher threshold C f value, indicating cooperative binding of the peptide onto the lipid vesicles where the peptide molecules are first incorporated into the membrane by simple adsorption and then aggregate therein. This further suggested that the S6 peptide formed large aggregates onto the lipid vesicles. Similar kinds of binding isotherms for lipid vesicles have been reported for peptides derived from channel-forming peptides such as pardaxin and alamethicin, pore-forming toxins, and the antimicrobial peptide magainin (35,41,56,59,69). The significant affinity of S6 toward the lipid vesicles, its localization appreciably inside the lipid bilayer, and its ability to self-assemble (Fig. 6) in the presence of lipid vesicles match with the fact that S6 is the porelining region of the KvAP channel (7,18,19). Interestingly, FIGURE 10. Single channel opening probability distribution curve of S6 peptide. The pattern shows Gaussian distribution at an applied membrane potential between Ϫ200 and ϩ200 mV. f shows the mean probability with S.E. for three independent experiments. Experimental conditions are the same as in Fig. 8. NBD-S6-SCR showed very significant affinity toward the negatively charged lipid vesicles that was to some extent higher than NBD-S6 peptide, indicating that the interchange of positions of some of the hydrophobic amino acids did not disturb the affinity of S6 peptide for lipid vesicles (Fig. 3, B and C). Nevertheless, the molecular basis of the higher affinity of NBD-S6-SCR than its parent S6 peptide is not understood at present. However, the comparative sharpness of the upward rise of binding isotherms at threshold C f values indicates a weaker cooperativity of NBD-S6-SCR than NBD-S6 peptide in binding to the negatively charged lipid vesicles (Fig. 3B). The selected S6 segment also possesses the AA sequence that exhibited significant helical structure in a phospholipid membrane or membrane-mimetic environment ( Fig. 5A and Table 1). However, the secondary structure of S6-SCR ( Fig. 5B and Table 1) clearly indicated a significant effect of the alteration in the amino acid sequence of S6 peptide in its helical structure either in phospholipid membrane or in membrane-mimetic SDS micelles.
One of the important issues is the assembly of the channel protein in membrane and how different segments of the protein participate in it. Fluorescence energy transfer experiments between the NBD-and Rho-labeled S6 peptides (Fig. 6, A and C) clearly indicated the potential of the S6 segment to contribute in the assembly of the whole KvAP protein in membrane. A similar FRET experiment demonstrated a weak self-assembly for S6-SCR in PC/PG lipid vesicles, suggesting the requirement of a specific amino acid sequence in the self-assembly of S6 peptide (Fig. 6, B and C).
Considering the fact that S6 is an integral part of the pore region of KvAP channel, its ability to cause leakage in the lipid vesicles was examined. Indeed, the wild type S6 peptide but not its analog S6-SCR induced the release of calcein from the calcein-entrapped lipid vesicles at an appreciably low peptide concentration, indicating a sequence-specific and significant activity of the synthetic S6 segment in permeabilizing the lipid vesicles (Fig. 7). Having obtained good experimental support in favor of the membrane-permeabilizing ability of the S6 peptide, ion channel activity of this synthetic segment was examined in BLM by a standard procedure as described under "Experimental Procedures" and reported earlier (47,56). As shown in Figs. 8,9,10, and 11, S6 peptide showed weakly voltage-sensitive ion channel activity as evident from the I-V plot. There are similar examples in the literature. For example, KcsA channel does not contain any voltage sensor but shows weak voltage dependence (70,71). It has been shown that gating can be voltage-dependent without voltage sensors (72). The I-V curve of S6 is asymmetric as in the case of KcsA channel (70). In light of the above discussions, we may conclude that S6 peptide has weakly voltage-dependent channel activity similar to KcsA channel. A similar kind of I-V plot not passing through zero has also been reported for other channels (73). Therefore, it may be inferred that the channel formed by S6 peptide is a "non-ohmic," weakly voltage-dependent ion channel. Furthermore, the opening probability of S6 channel showed distinct features; it is bellshaped and showed a maximum between Ϫ75 and ϩ75 mV. A similar bell-shaped opening probability of voltage-gated ion channels has been reported for the Maxi-chloride channel from the apical membrane of the syncytiotrophoblast of human pla-centa (74) and voltage-dependent anion channel in the plasma membrane of neurons (75).
Different models have been suggested for voltage sensor movement and opening of the intracellular S6 gate (76). A twogate theory has been proposed; the first gate is movement of the voltage sensor domain, and the second is the intracellular S6 gate (77). It has been postulated that depolarization moves the voltage sensor that opens the gate so that it does not hinder S6. Further depolarization opens the S6 gate so that it conducts the ions (77). When we compared the KvAP channel with S6, we found that the open probability curve of S6 is bell-shaped, whereas that of the whole KvAP channel is more sigmoidal (78). This could be because of the fact that the S6 segment lacks the voltage sensor domain. A typical KvAP channel shows complex gating kinetics as it exhibits different pre-open states before the channel opens (78). Moreover, it undergoes inactivation upon prolonged exposure to the voltage (78). The open probability of S6 is maximal between ϩ75 and Ϫ75 mV, which shows that the S6 gate remains open under resting membrane potentials and also under depolarizing membrane potential. We speculate that the high open probability of S6 peptide could be due to the lack of the voltage sensor domain. If the voltage sensor domain were present we probably would not have gotten this kind of curve. Thus, our results support the role of the voltage sensor segment in shielding/suppressing the S6 gate, which is important for controlling the flow of ions through KvAP.
It was reported (79) that the conductance of KvAP channel in planar lipid bilayers of 1-palmitoyl-2-oleoyl-sn-glycero-3phosphoglycerol and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine was 170 pS, whereas Morris and co-workers (80) reported that the unitary conductance of KvAP channel was 150 Ϯ 26 pS. We observed that the conductance of S6 varied from 100 to 160 pS at negative applied potentials and from 200 to 260 pS at positive applied potentials (Fig. 11). Considering the fact that S6 is only a small part of the whole KvAP protein and that the conductance of an ion channel in BLM depends on the chemical and mechanical properties of the lipids used to form the membrane (78), the conductance observed for the synthetic S6 segment matches fairly well with KvAP channel. Interestingly, the activity of S6-SCR in bilayer lipid membrane under different applied voltages was significantly weaker than the S6 peptide (supplemental Table 1 and Figs. 1  and 2). Unlike the S6 peptide, S6-SCR showed poor activity only at higher voltages; the opening probability of this ion channel was very low when voltage was ϳ100 mV (supplemental Fig. 2). Altogether, these BLM studies indicated a significant effect of rearrangement of the hydrophobic amino acids of S6 on its ion channel activity, suggesting the importance of a specific amino acid sequence in S6. We also designed another scrambled analog (S6-SCR1) in which some of the hydrophobic amino acids were substituted by polar amino acid residues, although its total amino acid composition was maintained the same as the S6 peptide; in this analog, isoleucines at positions 4 and 11 and valine at position 14 were replaced by serine at position 5, threonine at position 13, and serine at position 15, respectively (supplemental Fig. 3A). This scrambled analog (S6-SCR1) exhibited a membrane permeabilization property even lower than that of S6-SCR (supplemental Fig. 3B), and it did not exhibit any ion channel activity at the voltages tested for S6 and S6-SCR. S6-SCR1 exhibited significantly reduced helical structures in the presence of PC/PG lipid vesicles and SDS micelles (supplemental Fig. 3C) as compared with the other two peptides, and detailed membrane binding studies of this peptide was not performed. These results further strengthened the role of a specific amino acid sequence of S6 in maintaining its structural and functional properties.
It is a well known fact that S6 forms the activation gate of KvAP channel (76). Within the S6 lies a glycine residue, which is conserved in bacterial potassium channels (81). The gate opens through the glycine of S6 in KvAP channel (76,79,81,82). On the other hand, in a Shaker-like eukaryotic channel, the S6 activation gate opens due to a bend at the Pro-Val-Pro motif (77). In Shaker/Kv1.2 channels, which are eukaryotic channels, Pro-Val-Pro lies in place of Leu-Ile-Gly in KvAP. When Ile was interchanged with Gly, S6-SCR showed some activity but not at all the voltages. In S6-SCR1, more drastic amino acid substitutions were made, e.g. Ile of "Leu-Ile-Gly" was replaced by Thr (hydrophobic by hydrophilic), and therefore we probably saw the maximum abrogation in its ion chancel activity. This led us to speculate that the LIG stretch could be important in the formation of the S6 gate.
To the best of our knowledge, the present study is the first report on the ion channel activity of a small peptide from the S6 segment of a potassium channel. A synthetic 22-amino acid residue peptide designed from the putative pore region of a voltage-sensitive sodium channel formed ion channels in the lipid bilayer but did not show any voltage dependence (27). However, voltage sensitivity in artificial porin-like peptide has been reported (83). There are reports on the ion channel activity of peptides derived from receptors, ion channels, and toxins (24, 25, 84 -91) as well as designer peptides (83,88,92). It has been reported previously that mutations in the S6 segment alter the gating property of the whole channel (14,21,(93)(94)(95)(96). The voltage-gated channels are among those membrane proteins for which direct structural information is insufficient; thus, peptide models are used to explain their function. The ability of S6 peptide to form a weakly voltage-gated ion channel suggests that this particular amino acid sequence could be involved in the formation of the pore region of the native KvAP channel. In other studies, peptides derived from the H5 segment of Shaker potassium channel were found to give rise to single channel conductance (97,98).
S6-SCR showed higher affinity (Fig. 3C) than the parent S6 peptide, although it exhibited weaker aggregation (Fig. 3) and self-association (Fig. 6) properties than its parent peptide, S6. However, more noticeably, S6-SCR was significantly less active than the wild type S6 peptide either to permeabilize the lipid vesicles or to exhibit ion channel activity in BLM. We speculate that even after replacement of more hydrophobic isoleucine residues with less hydrophobic glycine and alanine residues, because the peptide still possess appreciable amphipathic properties, it maintains its self-assembly property to some extent onto the lipid vesicles. However, this interchange of positions of hydrophobic amino acids significantly impaired the secondary structure of S6 in the presence of phospholipid membrane and in membrane-mimetic environments. Moreover, its localization onto the lipid vesicles was disturbed in PC/PG lipid vesicles as a result of these interchanges of positions of hydrophobic amino acids. We anticipate that due to the abrogation of secondary structure S6-SCR probably did not form strong aggregates, and it failed to penetrate inside the bilayer of the lipid vesicles and localized mostly onto the surface of the membrane as evidenced by the blue shift and proteolytic cleavage experiments with NBDlabeled peptide (Figs. 2 and 4). Therefore, S6-SCR probably shows much weaker membrane-permeabilizing and ion channel activities. Perhaps at very high voltages, significant alteration in S6-SCR aggregates occurs that may result in better ion channel activity of this peptide. More drastic interchanges of amino acid sequences were made in S6-SCR1 that severely disturbed its secondary structures. Probably, therefore, S6-SCR1 exhibited the least membrane-permeabilizing property and the weakest ion channel activity among the three peptides.
Taken together, the present observations suggest that the 22-residue peptide derived from the S6 segment of KvAP channel possesses the primary amino acid sequence to bind to phospholipid vesicles with appreciable affinity and to self-assemble and form pores therein. More interestingly, the S6 peptide also formed weakly voltage-dependent ion channels in BLM, which could be related to the potential contribution of the individual helices in folding of the corresponding full-length integral membrane protein as was suggested by others as well (99 -102). Significant effects of alterations of amino acid sequence of S6 on its secondary structure, localization, and membrane-permeabilizing and ion channel activity were observed. Characterization of two scrambled S6 analogs with the same amino acid composition indicated the importance of a specific amino acid sequence in this synthetic segment. Altogether, these data are consistent with the fact that this S6 peptide was derived from the pore-lining region of a voltage-gated ion channel. The results obtained from studies with this synthetic peptide suggest a probable crucial role of the selected 22-residue stretch of the S6 segment in KvAP channel and are in agreement with earlier studies on the function of the S6 segment in KvAP channel (20,21,94,(103)(104)(105)(106). The present study is important because it addresses the question how S6 behaves independently of the voltage sensor segment. Although the synthetic S6 segment showed a weakly voltage-sensitive ion channel property in BLM, its difference from the whole KvAP channel is prominent in terms of voltage sensitivity. Thus, the present results show the characterization of the ion channel property of the synthetic S6 segment and the importance of the voltage sensor domain in maintaining the voltage sensitivity of a voltage-gated ion channel. How the mutations of S6 hydrophobic amino acids, the positions of which have been rearranged in the present study, can influence the properties of whole KvAP channel protein could be a subject of future investigation. Nevertheless, the results obtained from the present study may be useful in investigating the structural and functional properties of the corresponding segment in homologous channel proteins.