INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The transfer of protons in proteins and across biological membranes is an essential phenomenon in biology (Deamer and Nichols, 1989; DeCoursey and Cherny, 1994, 1999; Oliver and Deamer, 1994). ATP production is triggered by proton transfer in membrane proteins. Chains of water molecules and proton shuttle by amino acids in energy transducing enzymes have been demonstrated in the photosynthetic reaction center (Baciou and Michel, 1995), and in cytochrome c oxidase (Riistma et al., 1997), respectively. Despite its central role in life, the essential mechanisms underlying protein-assisted proton flow are not completely understood. Because gramicidin A (gA) is a water-filled ion channel pore with a proton conductance that can easily be measured (Akeson and Deamer, 1991; Cukierman et al., 1997; Eisenman et al., 1980; Heinemann, 1990; Hladky and Haydon, 1972; Levitt and Decker, 1988; Myers and Haydon, 1972), this ion channel has been used in both experimental and theoretical fronts as a model to promote our understanding of the basic rules of proton transport in proteins (Akeson and Deamer, 1991; Cukierman et al., 1997; Phillips et al., 1999; Pomès and Roux, 1996).

gA is a pentadecapeptide secreted by Bacillus brevis. Its unusual alternating sequence of D and L amino acids determines a ^6.3 helix structure in different molecular environments (Arseniev et al., 1985; Ketchem et al., 1993). In lipid bilayers, six intermolecular H-bonds are established between the amino termini of two gA's, resulting in the formation of an ion channel that is selective for monovalent cations only. Disruption of these H-bonds destabilizes the ion channel, leading to its disappearance. In 1989, Stankovic et al. developed an ingenious strategy for covalently linking two gA monomers. Using a dioxolane ring that allows a continuous and constrained transition between the two  helices of gA's, those authors demonstrated that the covalently linked gA dimer formed ion channels in lipid bilayers. As anticipated, covalently linked gA channels had lifetimes considerably longer than that of the monomer-monomer association of gA's via H-bonds. Because tartaric acid is the starting compound in the synthesis of the dioxolane linker, two stereochemically different dioxolane-linked gA dimers were synthesized in the dimerization process of gA's: the SS and the RR dimers (Stankovic et al., 1989). There are several advantages in using the dioxolane-linked gA's to study proton currents in proteins. One of them is the possibility of introducing discrete changes in the polarity or stereochemistry of the dioxolane linker (Stankovic et al., 1989, 1990; Cukierman et al., 1997). Because these molecular changes do not hamper the formation of ion channels in planar lipid bilayers, the structure-function relationships in a relatively simple protein can then be studied in relation to proton flow.

Previous work from our laboratory has determined the single-channel proton conduction and gating (opening and closing of the channel) in dioxolane-linked gA's in different lipid environments (Cukierman et al., 1997; Quigley et al., 1998). A racemic mixture of tartaric acid was used in our initial chemical synthesis, leading to the formation of both the SS and the RR dioxolane-linked gA dimers. It was not possible to separate these different dimers by conventional purification procedures. Consequently, our previous single-channel experiments in lipid bilayers with the product of the dimerization reaction revealed the presence of different channels with open state durations considerably longer than those of conventional gA channels. Unfortunately, it was not possible to unequivocally correlate the structure of each dimer (RR or SS) with its functional characteristics in lipid bilayers. Parenthetically, this illustrates the significant theoretical difficulties confronting the elucidation of the precise relationship between structure and function in ion channels. Even when the structure of a "simple" ion channel is relatively well known, as is the case with gA channels, it is very difficult, if not impossible at present, to predict the function of an ion channel from its structure and vice versa (Koeppe and Andersen, 1996).

Our interest in the basic mechanisms of proton conduction in proteins prompted the synthesis of each of the dioxolane-linked gA stereoisomers individually. To this end, the chemical synthesis of linked gA's started with the SS or RR tartaric acids (Stankovic et al., 1989, 1990). In this study, the resulting SS or RR gA dimers were studied individually in different lipid bilayers in HCl solutions. Significant differences in single-channel gating and proton conductances were found between these two different dioxolane-linked dimers. We have now concluded that the previously characterized D₁ dimer (Cukierman et al., 1997; Quigley et al., 1998) is indeed the SS stereoisomer of the dioxolane-linked gramicidin A channel. Moreover, our experimental results suggest that differences in single-channel proton currents between the SS and RR dimers must be caused by differences in water dynamics inside the pores of these channels (Cukierman et al., 1997; Pomès and Roux, 1996; Quigley et al., 1998). In turn, this must reflect appreciable changes in H-bond energetics between water molecules inside the channel and carbonyl groups lining the pore of the SS or RR gA dimers.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Synthesis and purification of different gA dimers

Dioxolane-linked gA dimers were prepared in three consecutive steps. First, the SS or RR stereospecific dioxolane dicarboxylic acids were synthesized from D-diethyl ester tartrate and L-diethyl ester tartrate, respectively. The diethyl ester protecting groups were removed, and the diacids were purified for further reactions. Second, natural gA was desformylated and purified. Third, the dioxolanes were coupled to two deprotected gA's and purified for biophysical studies.

The SS dioxolane linking reagent was prepared from the corresponding D-diethyl tartrate ester (Stankovic et al., 1989, 1990) .The diethyl tartrate was refluxed neat with diethoxymethane and catalytic p-toluenesulfonic acid under argon for 24-48 h at 95°C. The intermediate bis ethoxymethyl diester was converted to the closed ring under slow distillation at 105-115°C until no diethoxymethane or ethanol was collected. The remaining dioxolane diethyl ester was separated from p-TsOH and unreacted diethyl tartrate by silica gel chromatography. Thin-layer chromatography and H-NMR were utilized to determine fractions containing dioxolane-linked diester without unreacted diethyl tartrate. The dioxolane diester was saponified with 1 M NaOH for 90 min. Neutralization, acidification, and extraction with diethyl ether followed. The aqueous fraction was concentrated and triturated with ethyl acetate to yield the SS dioxolane dicarboxylic acid.

In parallel to the production of the respective dioxolane dicarboxylic acid stereoisomers, desformylated gA was prepared and purified. In contrast to our previous studies (Cukierman et al., 1997; Quigley et al., 1998), in which gA was purified by large-scale flash chromatography from the natural mixture of gramicidins A, B, and C, the current syntheses utilized purified gA from Fluka (Milwaukee, WI). This gA was desformylated and used in both the SS and RR synthesis. gA was desformylated by anhydrous hydrogen chloride (acetyl chloride in methanol, 2.9 M), with stirring for 60 min at room temperature. The reaction mixture was concentrated, dissolved in glacial acetic acid, and lyophilized for 24 h. Desformyl gA was separated from unreacted gA on a water-jacketed Ag11A8 column (1 × 50 cm), eluted with MeOH and 2 N NH₄OH in MeOH. Desformyl gA purity was further enhanced by preparative high-performance liquid chromatography (HPLC) on a reverse-phase C18 column, using methanol/water (Waters Co., Milford, MA).

The respective SS or RR linking reagent was covalently attached to two deprotected gA's via a diphenylphosphorylazide-mediated coupling in dimethyl formamide at 20°C for 48-96 h with catalytic triethylamine (15%). The reaction was quenched with MeOH and lyophilized. This crude product was purified by preparative HPLC on reverse-phase C18 with isocratic 95/5 MeOH/H₂O at 1.5 ml/min. The dimer fractions from multiple runs were pooled, concentrated, and repurified by HPLC to minimize gA contamination. Between purifications of either stereoisomer of the dimers, the injector, tubing, column, and detector were flushed extensively.

It is important to discuss the possibility of different products from the dimerization reaction. 1) Commercially available gA (Fluka, Milwaukee, WI) was used for the synthesis of the RR or SS dimers. The contamination of gramicidin C (gC) or B (gB) in that sample was very low. In our previous studies (Cukierman et al., 1997; Quigley et al., 1998), gA was purified from gB and gC by flash chromatography. The purity of these different gA samples was assessed by thin-layer chromatography and reverse-phase HPLC. Even if small amounts of gC or gB had remained throughout the synthesis reaction, the possibility of heterodimer (gA-gB, gA-gC) or homodimer (gB-gB, gC-gC) formation would certainly not have occurred in significant yields. Moreover, heterodimers have different retention times, and were not detected by the photodiode array of the HPLC. 2) The stability of gA was high under different phases of synthesis purification. Tryptophan degradation did not occur over the time course of dimer synthesis and purification. 3) The respective SS and RR dioxolane diethyl ester was separated by column chromatography from unreacted diethyl tartrate. The RR and SS linking reagents were assessed before and after removal of the ester protecting group for the continued presence of the closed dioxolane ring. Thus it is not likely that the ring hydrolyzed during the dimerization reaction. 4) It could be argued that a desformyl gA and a half-reacted SS or RR dioxolane-gA monomer form ion channels in lipid bilayers. However, desformyl gA was removed by ion exchange chromatography and further purified by HPLC. Desformyl gA and gA eluted before gA dimers. Therefore, it is not likely that the two components of these potential asymmetrical heterodimers would have been present in the bilayer set-up in significant amounts. In conclusion, our bilayer results with channels with very long open times, in relation to natural gA channels, suggest that those are indeed the SS and RR dioxolane-linked dimers.

NMR characterization

Proton NMR (Varian VXR400 or Varian VXR300) was used to characterize the individual linking reagents and the gA dimers. Two singlets appeared for the dioxolane link at 5.25 (s, 2H) and 4.75 (s, 2H), corresponding to the methylene protons of the dioxolane ring and the proton on the chiral carbons.

Molecular modeling of gA dimers

Molecular models of the SS and RR dioxolane-linked gramicidin dimers were developed using Insight and CHARMm 22 and manipulated using WebLab Viewer Pro (these three different software packages were from Molecular Simulations, San Diego, CA). Gramicidin A coordinates were retrieved from the Brookhaven Protein Data Bank (access code 1grm). The two stereospecific linking reagents were produced. Gramicidin A was digitally desformylated, and the respective linking reagent was docked into the space created by deletion of the two N-terminal valines. Models with the dioxolane group facing into the channel were constructed from the previous models by nearly symmetrical distortion of the amino acids near the linker, rotation of the dioxolane ring, and then attempting to restore helicity and continuity of the lumen by allowed rotations (peptide bond torsions or isomerizations were not attempted). For the purposes of these models, the dioxolane ring was treated as a rigid entity.

Characterization of ion channels in lipid bilayers

Experimental procedures were the same as before (Cukierman et al., 1997; Quigley et al., 1998). Briefly, membranes were formed onto a 0.1-mm-diameter hole in a polystyrene cup (cis side) nested inside a plastic chamber that formed the trans side. Membranes had the following compositions: 1) PEPC 4:1 (60 mM in decane), 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (PE), and 1-palmitoyl-2-oleoyl-phosphatidylcholine (PC); 2) glycerylmonooleate (GMO) (60 mM in decane). Phospholipids were obtained from Avanti Lipids (Alabaster, AL), and GMO was from NuCheck (Elysian, MN). Experiments were performed at room temperature (21-23°C).

Both sides of the membrane were connected to an Axopatch 1D amplifier (Axon Instruments, Foster City, CA) via Ag/AgCl wires immersed in solutions. Two different voltage-clamp protocols were applied across the lipid bilayers: 1) a steady DC voltage (range: 0-400 mV) was used, and 2) in PEPC bilayers, voltage clamp ramps from 0 to 380 mV were generated in ~7.5 s. Because GMO bilayers are far less stable at high voltages than PEPC bilayers, voltage ramps from 0 to ~200-250 mV were used in the former. The single-channel recordings shown in this study are representative of typical experiments. The SS and RR dimers each have consistent and reproducible electrophysiological characteristics that are easily identifiable during a bilayer experiment by their relatively long open durations, shapes of the single-channel current-voltage relationships, and their single-channel conductances. In this paper, n is the number of different channels recorded in different lipid bilayers of a given composition.

In the experimental conditions of this study (low pH), PE and PC are protonated. Consequently, PEPC bilayers are positively charged. In contrast, GMO membranes are neutral. To compare proton concentrations in different GMO or PEPC bilayers as seen by the channel openings, the concentration of protons at the PEPC-membrane/solution interface ([H⁺]_x = 0) was calculated using a model based on the Gouy-Chapman-Stern model (Cukierman, 1991; Cukierman et al., 1997). This calculation rests on several assumptions that may not be entirely correct. One of these assumptions relates to the pK of phospholipid protonation in the lipid bilayer. Not only is this value unknown, but it is expected that pK shifts as phospholipids become protonated. Therefore, calculations of [H⁺]_x = 0 must be understood as a first rough approximation for the proton concentration at the membrane/solution interface.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In this section, some structural differences between the SS and RR dimers that are relevant for proton conduction in gA dimer channels will be presented (Figs. 1 and 2, and Table 1). A description of the functional differences between proton conduction and gating in the SS versus RR dimers in different experimental conditions will follow (Figs. 3-8, and Table 2).

View larger version (45K): [in this window] [in a new window]   FIGURE 1   Close-up at the dioxolane linker in the SS (left) and RR (right) dimers as viewed from inside the lumen of the channel. H-bonds are indicated in yellow dashed lines. See text and Table 1 for a detailed discussion of the figure. Calibration bars at the bottom of the panels represent 2.5 Å.

Fig. 1 shows an inside-the-pore view of the dioxolane group (identified by carbons in yellow) connecting two gA molecules. Only the three central loops from the -helix are shown in this figure: the central loop, and those immediately above (Val₇, upper left), and below (Val'₇the prime identifies a residue in a different gA monomerlower left) that central loop. The left and right panels represent the SS and RR dimers, respectively. H-bonds in this figure are represented by yellow dashed lines. In natural gA, three symmetrical pairs of H-bonds stabilize the formation of an ion channel in the bilayer (Val₁  Ala'₅; Ala'₅  Val₁; Ala₃  Ala'₃). This H-bonding motif continues along the peptide, leading to a ^6.3 helix. Deletion of the formyl groups and covalent attachment of the SS dioxolane linker do not significantly distort the dipeptide (Stankovic et al., 1989). This is consistent with the previous functional finding that proton conduction in the SS is similar to natural gA in lipid bilayers (Cukierman et al., 1997). The two carbonyls of the dicarboxylic acid replace the formyl groups, allowing the same intrastrand hydrogen bonds with Val⁷. In both panels of Fig. 1 the dioxolane ring is facing away from the lumen of the pore.

Significant structural differences were noted between the SS and RR linked dimers. The change in chirality from SS to RR creates distortion between the two gA monomers, disrupting interstrand hydrogen bonding. Note the significant difference in the pitch of the dioxolane ring between the SS and RR dimers in Fig. 1. The C₂-C₃ bonds of the SS and RR dioxolane rings differ by ~95° relative to each other. Whereas in the SS dimer the dioxolane ring is parallel to the plane of the bilayer, in the RR it is nearly perpendicular to the plane of the bilayer. Because of differences in chirality at the linking reagent, distortions of the peptide backbone were noticed in the RR in relation to the SS dimer.

To better understand the alterations of secondary structure in the two stereospecific dimers, the geometries of H-bonds were examined as markers of carbonyl and amino orientation. Table 1 lists the unique inter- and intrastrand hydrogen bonds from the dioxolane linker to the C terminus. The parameters given are the H O distance, the N-H  O donor angle, the CO  H acceptor angle, and the torsion between donor and acceptor. Column 1 lists the hydrogen bonds in order of amino donor to carbonyl acceptor, starting from the center of the channel and radiating outward for a single monomer only. Several differences in H-bonding between the SS and RR linked gramicidin dimers are noteworthy. There was some variability in the H  O distances as also documented by Crouzy et al. (1994), and one H-bond (Val'¹  Ala⁵) was not formed in the RR dimer because of the pronounced tilt about the dioxolane link. This tilt increased the distance between the donor H (in Val'¹) to the acceptor O (in Ala⁵) from 1.95 Å in the SS to 3.56 Å in the RR, eliminating the H-bond in the latter. Some changes were also noticed in the donor and acceptor angles as well as torsion between the SS and RR. Because of some asymmetry in the distortion, only one of the pairs of Val'¹ to Ala⁵ was lost. The planes of some of the peptide bonds, which are generally perpendicular to the plane of the bilayer in the SS dimer, become tilted in the RR linked gA. This results in an increased pitch and tilt of the amino and carbonyl groups. This initial survey does not take into account peptide bond isomerization or carbonyl polarity. But it does demonstrate that even a subtle chirality change can cause significant alterations in protein secondary structural motifs (Crouzy et al., 1994). These structural differences between the SS and RR dimers are likely to be relevant for proton conduction through the pore. Because water molecules inside the channel interact electrostatically with the wall of the channel (Pomès and Roux, 1996), a change in the geometries of the peptide carbonyls can cause differences in water-peptide interactions. In turn, this may have a significant impact on proton transfer along the pore (see Discussion).

Fig. 2 shows cross-sectional views of the pore of the SS (upper row) and RR (lower row) dimers. Molecular dynamic simulations by Crouzy et al. (1994), in which dimer closures were caused by rotation of the dioxolane linker inside the channel lumen (Stankovic et al., 1989, 1990), prompted the models of the SS and RR dimers with the dioxolane linker protruding inside the pore. The right panels in Fig. 2 show the SS and RR dimers with the lumen partially obstructed by the dioxolane linker. In Fig. 2, O is represented by red, N by blue, hydrogens are omitted, and the dioxolane carbons are yellow. The white bar in each panel is 3 Å. The panels to the left show the same structures as seen in Fig. 1 (with the dioxolane ring pointing away from the pore). In the upper left panel, the orientation of the SS dioxolane is parallel to the plane of the bilayer and perpendicular to the barrel of peptide and hydrogen bonds. In the lower left panel, note that the pitch of the RR dioxolane ring is nearly perpendicular to that of the SS, whereas the overall pore diameter is essentially unchanged. However, and as discussed in relation to Fig. 1 and Table 1, the orientations of carbonyls are different between the SS-exo and RR-exo. This observation is significant because proton transfer in these dimers is markedly different (see below). Evidently, the distribution and orientation of carbonyls have a far more significant impact on proton transfer than the diameter of the pore itself.

View larger version (34K): [in this window] [in a new window]   FIGURE 2   Cross-sectional views of the pore of the SS and RR dimers. The upper row shows the SS dimer with the dioxolane linker outside (SS-exo, left) and inside (SS-endo, right) the pore. The lower row shows the RR-exo (left) and RR-endo (right). The carbons of the dioxolane linker are identified in yellow. The calibration mark in each panel represents 3 Å. See text for discussion.

The two right panels in Fig. 2 demonstrate the peptide distortion caused by rotating the respective SS and RR dioxolane rings into the lumen of the channel. This has been suggested to be the mechanism for channel closures in dioxolane-linked dimers (Stankovic et al., 1989). By rotating multiple - angles and leaving peptide bonds planar, as described in Materials and Methods, it is possible for the linker to protrude inside the pore lumen. In this process, the RR dimer required less peptide reorganization than in the SS. In the RR, much of the distortion was at the interface, with carbonyls directed into the lumen. In the SS, considerably more rotation was required to reestablish some measure of helicity in the protein. Note that the carbonyls in the upper right panel that are approximately orthogonal to the axis of the channel project into the lipid bilayer.

Fig. 3 shows single-channel I-V plots of the SS and RR dimers reconstituted in PEPC bilayers in 1 M [H⁺]_bulk. The RR has a significantly lower g_H than the SS dimer (Table 2). Linear portions of the I-V plot in Fig. 4 have a g_H of 516 and 228 pS for the SS and RR dimers, respectively. In the SS dimer, the I-V plot becomes sublinear around 100 mV at that proton concentration (Cukierman et al., 1997; Quigley et al., 1998). In contrast, the I-V relationship in the RR dimer is clearly supralinear in that same voltage range, as shown in the bottom panel of Fig. 3.

View larger version (23K): [in this window] [in a new window]   FIGURE 3   Single-channel I-V plots in response to voltage ramps applied to SS and RR dimers in PEPC (1 M HCl). Dashed lines represent the linear part of the gH. Recordings were low-pass filtered at 2 kHz. Note single-channel closures in both recordings.

View larger version (30K): [in this window] [in a new window]   FIGURE 4   Continuous recording of a single-channel SS dimer in a PEPC bilayer at a membrane potential of 60 mV (1 M HCl). The recording was low-pass filtered at 1 kHz and sampled at 4 kHz.

In Fig. 4, a continuous recording of a single SS dimer in PEPC at a transmembrane potential of 60 mV is shown. The typical lifetime of a SS dimer in lipid bilayers is determined essentially either by the duration of the experiment or by the lifetime of the bilayer (Cukierman et al., 1997; Quigley et al., 1998). In the specific case of Fig. 4, this single SS dimer lasted for the entire duration of the experiment (~70 min). We have conducted experiments in which the SS dimer remained stable in the bilayer for over 2 h. The SS dimer gates with very rapid, not completely resolved closures, even at recording frequencies of 10 kHz (Cukierman et al., 1997).

Fig. 5 shows a typical continuous recording of an RR dimer in a PEPC bilayer at a transmembrane potential of 50 mV. In this recording, the RR channel was incorporated into the bilayer a few seconds (time necessary to activate the videotape recorder) before the start of this recording. In contrast to what was mentioned before in relation to Fig. 4 (SS dimer), the RR dimer showed relatively long duration closures, and typically, the channel "disappeared" from the bilayer, as shown at the end of recording in Fig. 5. The disappearance (no channel opening observed within 1 min) of the RR channel from the bilayer will be referred to as inactivation. Sometimes the channel "reappeared" several minutes later in the bilayer. Evidently, it is not possible to ascertain whether it is the same or a different RR channel that was incorporated into the bilayer. Because of its limited lifetime in the bilayer, dwell-time distributions of fast closed states of the RR channel could not have been determined (small number of closing events).

View larger version (24K): [in this window] [in a new window]   FIGURE 5   Continuous recording of single channel proton currents in the RR dimer in a PEPC bilayer at a transmembrane potential of 50 mV (1 M HCl). The recording was low-pass filtered at 1 kHz and sampled at 4 kHz. The end of the recording in the lower panel identifies the "inactivation" of the channel.

Because 1) gA dimers have different single-channel properties in different lipid bilayers (Cukierman et al., 1997) and 2) the RR and SS dimers were previously studied in HCl solutions in GMO bilayers (Stankovic et al., 1989, 1990), it was of interest to perform experiments in GMO bilayers. In Fig. 6, I-V plots of the SS and RR dimers are shown in two different [H⁺]_bulk. As with PEPC bilayers, g_H in the RR is significantly smaller than in the SS dimer. In 122 mM [H⁺]_bulk, the single-channel conductances were 172 and 54 pS for the SS and RR dimers, respectively. In 1 M [H⁺]_bulk, g_H's were 828 and 328 pS (Table 2). Contrary to what is shown for PEPC bilayers, there were no major differences in the shape of I-V plots between the SS and RR dimers in GMO bilayers.

View larger version (23K): [in this window] [in a new window]   FIGURE 6   Single-channel I-V plots in response to voltage ramps applied to SS and RR dimers in GMO under different HCl concentrations. Dashed lines represent single-channel conductances of 828 and 328 pS (upper recordings), and 172 and two channels of 54 pS each (lower recordings). Single-channel recordings were low-pass filtered at 1 kHz (upper recording) and 300 Hz (lower recording).

In 1 M [H⁺]_bulk, the proton concentration at the PEPC/solution interface is 122 mM (see discussion of the underlying assumptions in Materials and Methods). g_H's (for both the SS and RR dimers) in GMO in 122 mM [H⁺]_bulk are considerably smaller than in PEPC with 1 M [H⁺]_bulk (Table 2).

Figs. 7 and 8 show continuous recordings in GMO bilayers of the SS and RR dimers, respectively. In Fig. 7, a single SS dimer was recorded in 2 M HCl at a transmembrane potential of 100 mV. The stable recording of the single-channel activity is characterized by fast flickers to the closed state as shown in Fig. 4. In Fig. 8, a single RR dimer was recorded in a 1 M HCl solution at 50 mV. Short-duration closures were also observed as with the SS dimer. However, RR channels were inactivated, as discussed before in relation to Fig. 5. The RR channel shown in Fig. 8 lasted ~90 s in the bilayer. The gatings of the SS and RR channels were qualitatively similar in the GMO and PEPC bilayers.

View larger version (28K): [in this window] [in a new window]   FIGURE 7   Continuous recording of single-channel proton currents in the SS dimer in a GMO bilayer at a transmembrane potential of 100 mV (2 M HCl). The recording was low-pass filtered at 4 kHz, sampled at 4 kHz, and decimated.

View larger version (19K): [in this window] [in a new window]   FIGURE 8   Continuous recording of single-channel proton currents in the RR dimer in a GMO bilayer at a transmembrane potential of 50 mV (1 M HCl). The recording was low-pass filtered at 1 kHz, sampled at 2 kHz, and decimated.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The novel findings reported in this study are: 1) g_H in the SS dimer is significantly larger than in the RR dimer; 2) shapes of I-V plots in PEPC (but not in GMO) bilayers are different between the SS and RR dimers; 3) the open state of the SS channel is stable, whereas the RR dimer eventually inactivates into a long-lasting closed state. Based on single-channel properties in different experimental conditions, it is concluded that our previously studied D₁ dimer (Cukierman et al., 1997; Quigley et al., 1998) is indeed the SS dioxolane linked dimer. In addition, differences between the H-bonding pattern in the RR and SS dimers modeled in this study could account for the observed differences in g_H between these channels.

Single-channel proton conductance in different gA dimers and bilayers

Proton transfer in water

Proton transfer in gA channels

Modulation of proton conduction in gA channels: implications for proton transfer in proteins

Gating of the SS and RR dimers

The pioneering work of Stankovic et al. (1989, 1990) revealed that the SS dimer remained essentially in the open configuration, whereas the RR dimer showed fast closing events. These results were obtained in different solutions (KCl or HCl). Molecular dynamics simulations (Crouzy et al., 1994) showed that in both the SS and RR dimers, the lowest energy minimum is attained when the dioxolane link is outside the pore (Fig. 2, left panels). In the RR dimer there is an energetically favorable path for the dioxolane to move from outside of to inside the pore in a sequence of different chemical steps (Fig. 2, bottom row). In contrast, this movement in the SS dimer is energetically more costly but clearly not impossible (see Crouzy et al., 1994). The H-bonded chain of water molecules inside the pore of the dimers is interrupted when the dioxolane link is inside the pore. This means that H⁺ transfer can no longer occur along the water wire. Consequently, the fast flickers that were only observed in the RR by Stankovic et al. (1989, 1990), and in our experiments with both the SS and RR dimers, could be explained by the movement of the dioxolane linker into the pore. Although Crouzy et al. (1994) favored that said movement would occur in the RR dimer only, those authors acknowledged the possibility that different kinetics could occur in different lipid environments, and with different (qualitative and/or quantitative) ions inside the pore of dioxolane-linked dimers. Because in our experimental conditions, the open state of natural gA channels lacks the fast closures seen with gA dimers (Cukierman et al., 1997), fast closing flickers in both the RR and SS dimers must be directly or indirectly (see below) related to the presence of the dioxolane linker in these molecules.

Fast closures were also noticed in gA channels without dioxolane linkers. For example, carbon suboxide-linked gA channels have fast closing events during the long openings of the channel (Bamberg and Janko, 1977). Fast closures in natural gA channels in lipid bilayers have also been demonstrated (Ring, 1986). Following the discussion above, it is possible that the reorganization of H-bonds in the peptide backbone affects the dynamics of water molecules inside the pore, causing interruptions in H⁺ transfer (seen as channel closures). An important factor to consider is that the dynamic flexibility of the helical chain of gA affects the mobility of waters inside the pore (Chiu et al., 1991). This could also modulate proton transfer and gating in gA channels.

Experimental results of this study clearly demonstrated that both the SS and RR dimers have fast closing events in either GMO or PEPC bilayers. Moreover, it was noticed that the RR dimer "inactivates." Thus, whereas the SS dimer gating is consistent with a simple O (open)  C (closed) kinetic scheme (Cukierman et al., 1997), gating in the RR is more complex, with an additional inactivated (I) state (I  C  O  I). It is not clear what is causing inactivation in the RR dimer. One possibility is that there is a very stable conformation of the dioxolane linker inside the RR dimer, interrupting the water chain for prolonged times. Another possibility could be a more dramatic conformational change of the entire protein structure of the RR dimer, resulting in the disappearance of channel activity.

Comparison with previous experimental results

It has been reported (Stankovic et al., 1990) that g_H in different dioxolane linked dimers was nearly the same in 40 mM HCl. It was also reported that the SS dimer remains in the open state essentially 100% of the time, whereas the RR dimer showed fast closing events. The results described in this paper are not in agreement with those experimental findings. The only methodological difference between our studies is that we have used the planar bilayer system, whereas Stankovic and colleagues formed bilayers at the tip of a glass pipette. Bilayers at the pipette tip can be under tension (Heinemann, 1990). The extent to which the behavior of dioxolane-linked gA dimers is affected by tension (as with natural gA channels; see Lundbæk et al., 1997) is an interesting topic for future development.

Previous work from our own laboratory (Cukierman et al., 1997; Quigley et al., 1998) have used the product of dimerization reaction from an initial racemic mixture of D- and L-tartaric acids. Both RR and SS dimers, unequivocally individualized in the present study by their long open times, were seen in previous experiments. We can now conclude that 1) The previously designated D₁ (or gA~D₁~gA) channel is the SS dimer. 2) A channel designated as gA~D₂~gA (or D₂) was identified in our lipid bilayers. This channel has a gating behavior completely different from that of the RR or SS channels. D₂ channels are in the closed state most of the time and have brief openings and a short lifetime in lipid bilayers. D₂ channels were also infrequently seen (as before) in experiments with solutions containing the pure SS or RR dimer. Is it possible that D₂ channels represent a different conformation of dioxolane-linked dimers?