Enhancement of Proton Transfer in Ion Channels by Membrane

Apr 15, 2009 - Department of Cell and Molecular Physiology, Loyola UniVersity Medical Center, 2160 South First AVenue,. Maywood, Illinois 60153, and ...
2 downloads 0 Views 2MB Size
Download PDF
J. Phys. Chem. B 2009, 113, 6725–6731

6725

Enhancement of Proton Transfer in Ion Channels by Membrane Phosphate Headgroups Debra L. Wyatt,† Carlos Marcelo G. de Godoy,‡ and Samuel Cukierman*,† Department of Cell and Molecular Physiology, Loyola UniVersity Medical Center, 2160 South First AVenue, Maywood, Illinois 60153, and Nu´cleo de Pesquisas Tecnolo´gicas, UniVersidade de Mogi das Cruzes, Mogi das Cruzes, Sa˜o Paulo 08087-911, Brazil ReceiVed: January 5, 2009; ReVised Manuscript ReceiVed: March 18, 2009

The transfer of protons (H+) in gramicidin (gA) channels is markedly distinct in monoglyceride and phospholipid membranes. In this study, the molecular groups that account for those differences were investigated using a new methodology. The rates of H+ transfer were measured in single gA channels reconstituted in membranes made of plain ceramides or sphingomyelins and compared to those in monoglyceride and phospholipid bilayers. Single-channel conductances to protons (gH) were significantly larger in sphingomyelin than in ceramide membranes. A novel and unsuspected finding was that H+ transfer was heavily attenuated or completely blocked in ceramide (but not in sphingomyelin) membranes in low-ionic-strength solutions. It is reasoned that H-bond dynamics at low ionic strengths between membrane ceramides and gA makes channels dysfunctional. The rate of H+ transfer in gA channels in ceramide membranes is significantly higher than that in monoglyceride bilayers. This suggests that solvation of the hydrophobic surface of gA channels by two acyl chains in ceramides stabilizes the gA channels and the water wire inside the pore, leading to an enhancement of H+ transfer in relation to that occurring in monoglyceride membranes. gH values in gA channels are similar in ceramide and monoglyceride bilayers and in sphingomyelin and phospholipid membranes. It is concluded that phospho headgroups in membranes have significant effects on the rate of H+ transfer at the membrane gA channel/solution interfaces, enhancing the entry and exit rates of protons in channels. Introduction Gramicidin A (gA) is a highly hydrophobic pentadecapeptide secreted by Bacillus breVis. In lipid bilayers, the primary structure of gA [an alternating sequence of D and L hydrophobic amino acids: HCO-L-Val1-Gly2-L-Ala3-D-Leu4-L-Ala5-DVal6-L-Val7-D-Val8-L-Trp9-D-Leu10-L-Trp11-D-Leu12-LTrp13-D-Leu14-L-Trp15-NH-(CH2)2-OH] defines a righthanded β6.3 helix in which the amino acid side chain residues are in contact with the core of the lipid membrane and the carbonyl and amide groups line the pore of the protein.1-3 The association via six intermolecular H-bonds between the amino termini of two gA peptides, each located in a distinct monolayer of a lipid bilayer, forms a water-filled ion channel that is selective for monovalent cations.4 Disruption of the intermolecular H-bonds between the two gA monomers leads to its dissociation and loss of channel function. Two gA peptides can be covalently linked to various molecules.3,5-10 Because covalently linked gA channels do not dissociate, their average open time far exceeds that in native gA channels. There are several experimental advantages in studying covalently linked gA channels, one of them being that experimental manipulations that affect ionic permeation in those channels can be analyzed independently of the intrinsic gating (association and dissociation of gA monomers) of gA channels. Even though the molecular details are not known, the structure and function of membrane proteins are generally affected by the composition of the lipid membranes. The simplicity of gA channels and the ease with which they can be reconstituted in * Corresponding author. E-mail: [email protected]. Phone: (708)2169471. Fax: (708)216-6308. † Loyola University Medical Center. ‡ Universidade de Mogi das Cruzes.

various lipid membranes offers an opportunity to unravel the molecular groups in the membrane that affect channel function. The function of gA channels is significantly dependent on the lipid composition of the membrane.4,8,11 gA channels reconstituted in monoglyceride versus phospholipid membranes display meaningful differences in their permeabilities to monovalent cations. The relative simplicity of gA channels and their distinctive behavior in different membrane environments offer a precious opportunity to unravel the molecular basis of modulation of membrane protein function by membrane lipids. This opportunity cannot be found in complex cellular systems. In particular, under nonsaturating proton concentrations ([H+]) in bulk solutions, the single-channel conductance to H+ (gH ) iH/Vm, where iH is the measured H+ current through a singlechannel molecule and Vm is the applied transmembrane voltage) in native and covalently linked gA channels in phospholipid membranes is ∼10-fold larger than that in monoglycerides.11-13 Moreover, in phospholipid membranes, a single adsorption isotherm accurately represents the relationship between gH and [H+]. By contrast, in monoglycerides, this relationship is considerably complex and not yet understood.13-17 The structural differences between monoglycerides and phospholipids are the presence of a second acyl chain and a phosphate headgroup in the latter. The major question addressed by this study was: Does the presence of an extra acyl chain and/or phosphate headgroup in phospholipid membranes account for the major quantitative differences in the rate of H+ transfer in gA channels in relation to monoglyceride membranes? This question cannot be addressed experimentally by adding either a phosphate headgroup or a second acyl chain to monoglycerides and comparing the rate of H+ transfer with gA channels reconstituted in phospholipid membranes. The addition of a phosphate headgroup or a second acyl chain to a monoglyceride

10.1021/jp900087g CCC: $40.75  2009 American Chemical Society Published on Web 04/15/2009

6726

J. Phys. Chem. B, Vol. 113, No. 19, 2009

forms detergents or diacylglycerides, respectively, that do not meet the structural requirements to form lipid bilayers and are, in fact, membrane destabilizers.33 Nevertheless, the question seems essential for the understanding of the molecular features in membranes that modulate the function of membrane proteins. To address the proposed question, a novel experimental approach was developed in this study. It was reasoned that, if stable planar lipid bilayers can be assembled from plain ceramides or sphingolipids, then insights could be gathered from the effects of the specific molecular structures responsible for distinct gA functions in monoglyceride and phospholipid membranes. In this study, the effects of the major structural differences among monoglycerides, ceramides, phospholipids, and sphingomyelins on H+ transfer in gA channels were addressed. The effects of plain ceramide or sphingomyelin bilayers on the rate of H+ transfer in gA channels were studied. The results shed light on the distinctive effects of specific molecular groups in membrane lipids that affect the function of membrane proteins. These results could be of significance not only for a better understanding of signal transduction pathways and membrane protein function in cell biology but also for the development or improvement of technological processes that use H+ currents. Materials and Methods 1. Planar Lipid Membranes. Lipids were purchased from Avanti Polar Lipids Inc. (Alabaster, AL). Lipid membranes were formed from decane solutions (∼60 mM) of (a) N-oleoyl-D-erythrosphingosine or ceramide C18:1 (NODS) and (b) (2S,3R,4E)-2acylaminooctadec-4-ene-3-hydroxy-1-phosphocholine (sphingomyelin, SPM). These lipids were obtained in chloroform. A small aliquot was transferred to an assay tube, and the chloroform was evaporated with a gentle flux of N2. Dry sphingolipids were immediately resuspended in decane. Fresh solutions of lipids in decane (∼20 µL) were prepared and used for approximately 2-3 h of experiments. At room temperature, these solutions become gels. Immediately before forming a planar membrane, the lipid solution was warmed to ∼37 °C until it became liquid. Bilayers were then formed by spreading the decane solution onto a ∼0.1-mm-diameter hole in a polystyrene partition that separated two distinct aqueous compartments. The formation of lipid bilayers was assessed by capacitance measurements and visual inspection. Bilayers formed using this methodology with either NODS or SPM were extremely stable and had very high leak resistances (at least 10 GΩ in HCl concentrations higher than 2 M). The stability and leakiness of NODS and SPM membranes were comparable to those of membranes formed with the more conventional 1-palmitoyl-2-oleoyl phosphatidylcholine (POPC) and glycerylmonooleate (GMO). The thickness of a lipid bilayer has a strong influence on the ionic permeation and gating in various gA channels.16 Consequently, the thicknesses of the new bilayers used in this study were calculated following measurements of the area and the capacitance of lipid membranes.31,32 The thicknesses of distinct decane-containing bilayers were similar (median ( sem [standard error of the mean], n): GMO/decane (45.9 ( 1.0 Å, 16), POPC/decane (47.4 ( 0.7 Å, 12), SPM/decane (46.7 ( 2.2 Å, 8), NODS/decane (47.9 ( 1.3 Å, 7).31,32 Consequently, the distinct experimental results obtained in various bilayers in this study cannot be attributed to variations in bilayer thickness. The main structural differences among GMO, NODS, SPM, and POPC are illustrated in Figure 1. NODS has a sphingosine backbone (note the arrow pointing at the double bond between

Wyatt et al.

Figure 1. Molecular structures of membrane components used in this study. From left to right: GMO, NODS, SPM, and POPC. GMO and NODS lack the phosphate headgroup present in SPM and POPC. GMO and NODS differ by the presence of a second acyl chain in the latter. Notice the presence of a cis C4sC5 double bond (indicated by an arrow) and a peptide bond in NODS. These features are essential for the formation of stable membranes, which does not occur with diacylglycerides. The difference between NODS and SPM is the presence of a phosphocholine group. As with NODS, SPM has an amide group and a cis C4sC5 double bond (arrow) that can donate and receive H-bonds, respectively. In the SPM molecule, a H-bond between the NH group and an O is indicated by a dashed line. The cis C4sC5 double bond is essential for bilayer formation with NODS. For the sake of clarity, most hydrogens in the structures in this figure were omitted. C, O, H, N, and P are represented in gray, red, white, blue, and orange, respectively.

C4 and C5 in the sphingosine moiety). An oleate group is linked to the C2 of the glycerol molecule via an amino group as in a peptide bond. These two features constrain the flexibility of the acyl groups, maximizing the intermolecular van der Waals interactions between adjacent acyl chains. Thus, stable membranes can be formed, as will be shown below. H-bonds between adjacent NODS molecules in a lipid membrane (see below) also enhance the membrane stability. The cis C4sC5 double bond in NODS that would accept an intermolecular H-bond from an adjacent amide H is also essential for the formation of membranes. Reduction of this double bond results in C18:1 dihydrosphingosine, which did not form lipid bilayers in our study, thus strengthening previous argument. The amide group in NODS also constrains the flexibility of the acyl chain via coplanarity of C, O, N, and H as in peptide bonds18 and donates H-bonds. Replacement of C1-hydroxyl in NODS by a phosphocholine group results in SPM. Another structural feature of interest that aids in understanding of our experimental results is the likely intramolecular H-bond between the phosphate oxygen and the amide hydrogen in the SPM molecule (Figure 1, dotted line in SPM). This has been suggested in crystallographic18 and in molecular dynamics studies.19-22 This structural feature is used in this study to rationalize the dysfunction or inactivation of gA channels in SPM membranes at low ionic strengths. 2. Experimental Setup and Analysis. Symmetrical solutions of HCl were used in both compartments across the membrane. Because POPC and SPM membranes are positively charged under our experimental conditions (but not GMO and NODS membranes), H+ concentrations at the interface between the membrane protein and the solution ([H+]x)0) were calculated using Gouy-Chapman-Stern models.12,13 Relationships be-

H+ Transfer and Membrane Phosphate Groups

J. Phys. Chem. B, Vol. 113, No. 19, 2009 6727

tween single-channel conductances were plotted against [H+]bulk for GMO and NODS membranes and against [H+]x)0 for POPC and SPM membranes. Only single-channel experiments were analyzed. To this end, gA channels were added from a methanol stock solution to only one bilayer compartment at a concentration of ∼10-9 M for native gA and ∼10-11 M for the dioxolane-linked gA channels. Single-channel conductances of hundreds of channels were measured under each set of experimental conditions and plotted in histograms. These histograms were fitted with simple Gaussian distributions. The experimental points in the graphs correspond to the median ( standard error of the mean (sem) of the distributions. Unpaired Student’s t tests were applied for single-channel conductances of gA channels in various lipid membranes. These results are shown in the figure legends. It was previously reported23,24 that short- and long-chain ceramides form ion channels in phospholipid/cholesterol membranes. The methodology used in the earlier works is different from ours. Nevertheless, it is important to state that, in our experiments, channel activity was never detected in plain sphingolipid membranes in the absence of gA peptides. Results and Discussion 1. H+ Transfer in gA Channels Reconstituted in Plain NODS or GMO Membranes: Channels in NODS Membranes Become Dysfunctional at Low Ionic Strength. Figure 2 shows log-log plots of gH versus [H+]bulk in NODS (open and solid squares) in native gA (panel A) and in the singlestranded- (SS-) dioxolane-linked gA channels (panel B). For GMO and ceramide membranes [H+]x)0 ) [H+]bulk. Because POPC and SPM membranes are positively charged under the experimental conditions of Figure 2, [H+]x)0 is considerably smaller than [H+]bulk. The proton concentration at the membrane/ solution interface ([H+]x)0) was calculated using GouyChapman-Stern models (see refs 12, 13, and 33 for details and discussion). In the concentration range of 0.2 M < [H+] < 2 M, the open squares (gA channels in NODS bilayers) were well approximated by a straight line in both panels. However, at [H+] < 0.2 M, gH became increasingly attenuated, and at [H+] < 0.025 M, the transfer of H+ (as well as monovalent alkalines, results not shown) in the channel ceased. This unexpected finding contrasts with H+ transfer measured in gA channels in other membranes at much lower H+ concentrations. In particular, single-channel activity in the submilimolar range of [H+]7,8,12-17 has been frequently observed in this (see below) as well as in our previous studies. The short black lines (starting at 25 mM H+ concentration) drawn in both panels represent the linear dependence of gH on H+ concentration assuming that gH is entirely determined by a diffusion-limited process in that range of H+ concentration. Notice that gH-[H+] relationships (open squares in NODS bilayers) at low H+ concentrations are considerably steeper than that for a diffusion-limited process. This suggests that significant qualitative changes in H+ permeation through the gA channels are occurring. Because gA peptides transfer H+ at considerably lower H+ concentration when reconstituted in other membranes7,8,12-17 (see Figure 2, POPC curve and solid triangles), it is likely that gA channels at low H+ concentrations in NODS membranes become dysfunctional or inactivated. In an attempt to determine the factors that caused the loss of proton transfer at [H+] < 0.025 M in distinct gA channels in NODS membranes, several experimental variables were modified in trying to restore or enhance gH at low H+ concentration. The ionic strength of solutions accounted for the loss of gA

Figure 2. log-log plots of gH versus [H+]x)0 in (A) native gA and (B) the SS-dioxolane-linked gA channels in various lipid membranes. Experimental measurements are plotted as mean ( sem. The error bars are considerably smaller than the size of symbols in these graphs. In both panels, the black continuous lines (identified as POPC and GMO) connect experimental measurements of gH (symbols not shown) of gA channels in GMO and POPC membranes. The triangles and squares (both open and solid) represent measurements of gH in SPM and NODS membranes, respectively. The open squares represent measurements of gH in solutions containing HCl only. The solid squares represent gH measurements in HCl solutions with an additional 200 mM NaCl. The dashed straight lines connecting the solid squares in both panels are best power-function (linear on a log-log scale) fits to gH values in (A) native gA and (B) SS-dioxolane-linked channels in NODS bilayers. Notice that these straight lines connecting the solid squares have slopes of (A) 0.82 and (B) 0.85, indicating that the rate-limiting step of proton conduction in gA channels does not occur in bulk solution.14 These slopes are similar to that for the GMO membrane (0.75). The thick straight lines (starting at [H+] ) 25 mM, open squares) at the bottom of the graphs show the dependence of gH on the H+ concentration (slope ) 1) under the assumption that gH is entirely limited by the diffusion of protons in bulk HCl solutions. There are significant differences between various sets of gH data (unpaired Student’s t test): gH values in both native and covalently linked gA channels in POPC are significantly larger than those in SPM membranes (p < 0.0001). gH values in POPC and SPM bilayers are significantly larger than those in GMO or NODS membranes (p < 2 × 10-7). In the concentration range of 0.03 M < [H+] < 1 M, gH values in native gA channels (but not in SS-dioxolane-linked gA channels) in NODS membranes (solid squares) are significantly larger than those in GMO bilayers (continuous curve in panel A, p < 0.002).

function in NODS bilayers at low HCl concentrations. This is demonstrated in Figure 2, in which the solid squares represent gH measurements in solutions containing 0.2 M NaCl in addition to HCl. In both panels A and B, at high H+ concentration, there is no major difference between the open and solid squares. However, at low ionic strengths, differences between gH values in those two solutions (HCl and HCl + 0.2 M NaCl) were

6728

J. Phys. Chem. B, Vol. 113, No. 19, 2009

Wyatt et al.

Figure 3. (Left) Single-channel recordings of native gA channels in (A) GMO and (B,C) NODS bilayers. The ionic concentration in each recording is indicated at the top of the panel. Notice the significant attenuation of proton currents in panel B (compared to panel A) and their recovery in panel C following the addition of 200 mM NaCl. (Right) Corresponding histogram distributions of gH values under the various experimental conditions indicated on the left. The dotted line in each panel of the left column represents the zero current level (all channels are closed). Channel openings and closures are seen as upward and downward deflections, respectively. Single-channel currents were digitized at 2 kHz and low-pass-Besselfiltered at 200 Hz. The applied transmembrane voltage was 50 mV.

immediately noticeable. gH was markedly enhanced when NaCl was added to low H+ concentration. An additional figure in the Supporting Information describes the effects of various NaCl concentrations on gH in 10 and 25 mM H+. The increase in ionic strength enhances gH continuously. This suggests that the functional state of gA channels at low ionic strengths can be modulated by subtle intermolecular interactions at the membrane/ channel/solution interface (discussed below) and not by major mesoscopic changes in the membrane phase. It must be remarked that the possible contribution of Na+ ions to the overall single-channel conductance of gA channels in solutions containing NaCl and HCl is negligible. Even in symmetrical 200 mM NaCl solutions, the single-channel conductances to Na+ are of only a few picosiemens.34 Furthermore, increasing the ionic strength by adding 30-50 mM Ca2+ (at a much lower concentration than NaCl), which does not permeate gA channels, also prevented the lack of gA channel function in NODS bilayers at low ionic strength (results not shown). Figure 3 shows typical recordings of single native gA channels in GMO (top panel) and NODS (middle and bottom panels) membranes at an applied transmembrane voltage of 50 mV. These recording were obtained in 20 and 25 mM H+ (top and middle panels, respectively) and at 25 mM H+ with an additional 200 mM NaCl (bottom panel). The corresponding histograms of gH values in various experiments are shown next to the recordings. In the middle panel, single H+ currents were extremely attenuated in relation to the top panel, as also depicted in the graphs of Figure 2. The median distribution of gH declined

from approximately 40 to 10 pS (see the corresponding histograms). Upon addition of NaCl, a major enhancement of the single-channel currents occurred (bottom panels). This enhancement also occurred with KCl and with smaller concentrations of CaCl2 (40 mM), which suggests that the ionic strength of solution is the cause of the dysfunctional gA channels at low salt concentrations. As mentioned in the Introduction, native gA peptides form ion channels via dimerization of peptides across lipid membranes. The use of covalently linked SS dioxolane gA dimers in experiments similar to those shown in Figure 3 serves one purpose of eliminating the possibility of an effect of low ionic strength on the dimerization process of native gA channels reconstituted in ceramide membranes. In Figure 4, singlechannel recordings and corresponding histograms of the SSdioxolane-linked gA channels in plain ceramide bilayers in 50 mM HCl (panel A) and with an additional 200 mM NaCl (panel B) are also shown. In this figure, sequential incorporations of channels (notice the significantly longer open times as compared to native gA channels) are illustrated. The experimental results in Figures 3 and 4 demonstrate that, at relatively low ionic strength, various gA channels reconstituted in ceramide membranes become dysfunctional. Other experimental variables (solution tonicity, applied transmembrane voltages, and temperature) were not effective in restoring channel activity in NODS at low ionic concentrations. Another possible interpretation of ion channel dysfunction at low ionic strength is that the incorporation of gA channels in

H+ Transfer and Membrane Phosphate Groups

J. Phys. Chem. B, Vol. 113, No. 19, 2009 6729

Figure 4. Single-channel recordings of SS-dioxolane-linked gA channels in NODS membranes and corresponding gH histograms in (A) 25 mM HCl and (B) 25 mM HCl + 200 mM NaCl. See the Figure 3 caption for an explanation of the notation. Notice in this figure the considerably longer open times in the covalently linked gA channels in comparison with those in the native gA recordings (Figure 3).

NODS did not occur. To that end, experiments were performed in which ion channels were added to low-ionic-strength solutions, and no functional ion channel was observed after 15 min. However, when the solution ionic strength was increased by adding NaCl or CaCl2, gA channel activity immediately became evident, suggesting that channels were already incorporated into the membrane at low ionic strength but were not functional. It is proposed that, at low ionic strength, gA peptides are already present in the NODS membrane but are not properly folded as functional ion channels. This suggestion is based on structural features of the NODS molecule, namely, the cis C4sC5 double bond and the amide group in the sphingosine moiety (see Figure 1). These features determine patterns of intraand intermolecular H-bonds18 that have recently been studied and contrasted with those in POPC membranes.19-22 In particular, the electron-dense cis C4sC5 double bond and the amide group at the solution/membrane interfaces can receive and donate intermolecular H-bonds, respectively. At high ionic strength, the partial charges of these groups are screened. However, at low ionic strength, the electrostatic potential of these groups could form stronger H-bonds with the carbonyls and amide groups in the outer loops of the gA peptide. These H-bonds could have the effect of distorting the local secondary structure of the peptide, leading to dysfunctional channels. A distinct and not exclusive possibility is that, at low ionic strength, intermolecular H-bonds between NODS molecules would allow a closer packing of the lipid membrane25 that would prevent appropriate solvation of gA in the membrane core peptide, leading to its unfolding and channel inactivation. Notice that the loss of channel function did not occur with channels reconstituted in SPM membranes (see triangles in Figure 2 and discussion below). In SPM bilayers, the amide group frequently donates an intramolecular H-bond to a phosphate oxygen,9-22 and this could prevent the unfolding of the peptide leading to channel dysfunction as in the mechanisms discussed above. To further inquire into the molecular origin of channel inactivation in NODS membranes, attempts were made to form

bilayers using C18:1 dihydrosphingosine. The difference between this molecule and NODS is the lack of the double bond present in the sphingosine moiety of the former, eliminating the capacity to accept H-bonds. However, membranes cannot be formed with C18:1 dihydrosphingosines. Although our hypothesis could not have been further corroborated, these experiments demonstrate unequivocally that the cis C4sC5 double bond is essential for the formation of plain NODS bilayers as discussed in the Materials and Methods section. In the H+concentration range of 0.03-1.0 M, gH values in native gA channels in NODS are larger than those in GMO membranes (Figure 2A). For the SS-dioxolane-linked gA channels, however, the gH values (Figure 2B) were quite similar in both GMO and NODS membranes, and they were wellrepresented by straight lines (with slopes of ∼0.75; see ref 14) in log(gH)-log([H+]) plots (Figure 1B). The linearization of the log(gH)-log([H+]) plots has been previously demonstrated.14-16 The double acyl chains in NODS might provide a better solvation of native gA channels in the middle of the channel than the single acyl chain of GMO. This could optimize the intermolecular H-bonds between gA peptides in the bilayer core and provide more stability for the native gA channel, thus enhancing the rate of H+ transfer in the middle of the channel.15,16 In the SS-dioxolane-linked gA channels (Figure 1B), the interaction between the two gA peptides is already optimized by the constrained and continuous transition caused by the SS-dioxolane linker, and the gH values are similar in both NODS and GMO bilayers. In summary, the enhancement of H+ transfer in native gA channels by NODS in relation to GMO bilayers could be a consequence of more favorable and stable interactions between the two gA peptides in the core of the bilayer due to a better solvation of gA peptides by double acyl chains. This would explain the linearization of log-log plots for H+ transfer in gA channels. However, the significant conclusion regarding the rates of proton transfer in gA channels in monoglyceride versus phospholipid membranes, which was the original motivation for

6730

J. Phys. Chem. B, Vol. 113, No. 19, 2009

this study, is that adding a second acyl chain to the membrane lipid does not enhance the rate of H+ transfer in gA channels to the level seen in a phospholipid membrane. 2. Ion Channel Function in SPM and POPC Bilayers. Membrane Phosphate Headgroups Enhance H+ Transfer. Figure 2 also shows that gH values in gA channels in SPM membranes (triangles) are considerably larger (3-6 times) than those in NODS membranes (squares). The similarities in the gH-[H+] relationships between GMO and NODS bilayers and between POPC and SPM membranes (continuous curves in Figure 213,14) suggest that the differences between gH values in SPM (or POPC) and NODS (or GMO) membranes are caused by the presence of phospho headgroups in the former bilayers. Differences between the gH values in POPC and SPM membranes could be a consequence of different protonation association constants (pKa values) of the phospholipid headgroups, which are used to calculate [H+]x)0 (see Materials and Methods). In these plots, the same pKa value of 1 used for POPC26 was assumed for SPM membranes. Possible mechanisms by which phospholipid membranes enhance the rate of proton transfer in various gA channels compared to monoglyceride membranes have been discussed before.7,8,12-16 It is of interest to compare and better understand the experimental results in this work with respect to results recently obtained from molecular dynamics calculations of H+ transfer in gA channels imbedded in GMO or phosphatidylcholine membranes.30 The free energy barrier for H+ transfer in gA channels imbedded in GMO membranes was reported to be significantly larger than that in gA channels in phospholipid membranes.30 In particular, those calculations found the following: (1) Water molecules at the interface between the membrane channel and the solution are structureless in a GMO membrane and are considerably more organized at the interfaces of a phospholipid membrane. Those structures create a negligible free energy barrier for H+ entry/exit at the mouths of gA channels in phospholipid membranes that contrasts with the significant barrier present in GMO interfaces. (2) A gA channel in a phospholipid membrane fluctuates considerably less than one in a GMO bilayer. This would stabilize the water wire inside the channel by lowering the entropic contribution to the activation energy of H+ transfer14,15 along the water wire inside the channel’s pore. Concluding Remarks The main question addressed in this study concerns the molecular basis of the higher rate of H+ transfer in gA channels reconstituted in phospholipid compared to monoglyceride membranes. Experiments were performed in various gA channels in plain ceramide (NODS) or sphingolipid (SPM) membranes. The structural difference between these molecules is the lack of a phosphocholine group in the former. The rate of H+ transfer in gA channels is significantly larger (and comparable to that in phospholipid membranes) in SPM membranes than in NODS membranes. This could relate to a significantly more organized structure of water molecules at the membrane-channel/ solution interfaces that increases the entrance and exit rates of proton transfer in and out of the channel.14,16,30 On the other hand, the presence of a second acyl chain among various lipid components in membranes enhances gH in a significant manner. This might result from a better solvation of the side chain residues, resulting in stabilization of the gA channel and its water wire inside the pore.30 In the course of this experimental work, an original and unexpected finding was that, in low-ionic-strength solutions, gA

Wyatt et al. channels became dysfunctional or completely inactivated in NODS membranes only. It would be of interest to study the interactions between NODS and gA channels using computational methods. Such an approach could unravel basic phenomena related to the modulation of membrane protein function by membrane lipids and test our working hypothesis. The experimental results in this study can be of relevance to cell biology. The conversion of SPM into NODS in biological membranes via phospholipases is thought to be an important signal transduction pathway in biology.27-29 Our findings provide direct evidence that the function of a membrane protein surrounded by SPM molecules in cell membranes can be modulated indirectly by phospholipases that modify SPM molecules into NODS. Another possible implication of our experimental results pertains to various technological applications that use H+ currents. Proton transfer is part of various technological developments including proton electromotive fuel cells and artificial photosynthesis. The knowledge that phosphate headgroups on the membrane surface appreciably enhance the rate of H+ transfer across molecules or channels could be of significance in developing new or more efficient technology. Acknowledgment. Supported by grants from the U.S. NIH (GM05976) and Fundac¸a˜o de Amparo a` Pesquisa do Estado de Sa˜o Paulo (FAPESP, Brazil). Supporting Information Available: Figure showing the effects of various concentrations of NaCl (distinct ionic strengths) on gH values in the SS-dioxolane-linked gA channel reconstituted in plain ceramide membranes at two H+ concentrations of 10 and 25 mM. This information is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Arseniev, A. S.; Barsukov, I. L.; Bystrov, V. F.; Lomize, A. L.; Ovchinikov, Y. A. FEBS Lett. 1985, 186, 168–174. (2) Ketchem, R. R.; Roux, B.; Cross, T. A. Structure 1997, 5, 1655– 1669. (3) Urry, D. W. Proc. Natl. Acad. Sci. U.S.A. 1971, 68, 672–676. (4) Hladky, S. B.; Haydon, D. A. Biochem. Biophys. Acta 1972, 274, 294–312. (5) Armstrong, K. M.; Quigley, E. P.; Quigley, P.; Crumrine, D. S.; Cukierman, S. Biophys. J. 2001, 80, 1810–1818. (6) Bamberg, E.; Janko, K. Biochem. Biophys. Acta 1977, 465, 486– 499. (7) Cukierman, S.; Quigley, E. P.; Crumrine, D. S. Biophys. J. 1997, 73, 2489–2502. (8) Quigley, E. P.; Quigley, P.; Crumrine, D. S.; Cukierman, S. Biophys. J. 1999, 77, 2479–2491. (9) Stankovic, C. J.; Heinemann, S. H.; Delfino, J. M.; Sigworth, F. J.; Schreiber, S. L. Science 1989, 244, 813–817. (10) Rudnev, V. S.; Ermishkin, L. N.; Fonina, L. A.; Rovin, Y. G. Biochem. Biophys. Acta 1981, 642, 196–202. (11) Phillips, L. R.; Cole, C. D.; Hendershoot, R. J.; Cotton, M.; Cross, T. A.; Busath, D. D. Biophys. J. 1999, 77, 2492–2501. (12) Godoy, C. M. G.; Cukierman, S. Biophys. J. 2001, 81, 1430–1438. (13) Chernyshev, A.; Cukierman, S. Biophys. J. 2006, 91, 580–587. (14) Cukierman, S. Biophys. J. 2000, 78, 1825–1834. (15) Chernyshev, A.; Cukierman, S. Biophys. J. 2002, 82, 182–192. (16) Chernyshev, A.; Armstrong, K. M.; Cukierman, S. Biophys. J. 2003, 84, 238–250. (17) Gowen, J. A.; Markham, J. C.; Morrison, S. E.; Cross, T. A.; Busath, D. D.; Mapes, E. J.; Schumaker, M. F. Biophys. J. 2002, 83, 880–898. (18) Pascher, I. Biochim. Biophys. Acta 1976, 455, 433–451. (19) Chiu, S. W.; Vasudevan, S.; Jakobsson, E.; Mashl, R. J.; Scott, H. L. Biophys. J. 2003, 85, 3624–3635. (20) Hyvo¨nen, M. T.; Kovanen, P. T. J. Phys. Chem. 2003, 107, 9102– 9108. (21) Mombelli, E.; Morris, R.; Taylor, W.; Fraternali, F. Biophys. J. 2003, 84, 1507–1517. (22) Niemel¨; a, P.; Hyvo¨nen, M. T.; Vattulainen, I. Biophys. J. 2004, 87, 2976–2989.

H+ Transfer and Membrane Phosphate Groups (23) Siskind, L. J.; Colombini, M. J. Biol. Chem. 2004, 275, 38640– 38644. (24) Siskind, L. J.; Davoodi, A.; Lewin, N.; Marshall, S.; Colombini, M. Biophys. J. 2003, 85, 1560–1575. (25) Brockman, H. L.; Mosen, M. M.; Brown, R. E.; He, L.; Chun, J.; Byun, H.; Bittman, R. Biophys. J. 2004, 87, 1722–1731. (26) Marsh, D. CRC Handbook of Lipid Bilayers; CRC Press: Boca Raton, FL, 2004; pp 81-85. (27) Blitterswijk, W. J. V.; Luit, A. H. V.; Veldman, R. J.; Verheij, M.; Borst, J. Biochem. J. 2003, 369, 199–211. (28) Cremesti, A. E.; Goni, F. M.; Kolesnick, R. FEBS Lett. 2002, 531, 47–53. (29) Futerman, A. H.; Hannun, Y. A. EMBO Rep. 2004, 5, 777–782.

J. Phys. Chem. B, Vol. 113, No. 19, 2009 6731 (30) Qin, Z.; Tepper, H. L.; Voth, G. A. J. Phys. Chem. B 2007, 111, 9931–9939. (31) Benz, R.; Frohlich, O.; Lauger, P.; Montal, M. Biochem. Biophys. Acta 1975, 394, 323–334. (32) Lundbæk, J. A.; Andersen, O. S. J. Gen. Physiol. 1994, 104, 645– 673. (33) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 1992. (34) Quigley, E. P.; Crumrine, D. S.; Cukierman, S. J. Membr. Biol. 2000, 174, 207–212.

JP900087G