Formation of non-beta 6.3-helical gramicidin channels between sequence-substituted gramicidin analogues.

Using the linear gramicidins as an example, we have previously shown how the statistical properties of heterodimeric (hybrid) channels (formed between the parent [Val1]gramicidin A (gA) and a sequence-altered analogue) can be used to assess whether the analogue forms channels that are structurally equivalent to the parent channels (Durkin, J. T., R. E. Koeppe II, and O. S. Andersen. 1990. J. Mol. Biol. 211:221-234). Generally, the gramicidins are tolerant of amino acid sequence alterations. We report here an exception. The optically reversed analogue, gramicidin M- (gM-) (Heitz, F., G. Spach, and Y. Trudelle. 1982. Biophys. J. 40:87-89), forms channels that are the mirror-image of [Val1]gA channels; gM- should thus form no hybrid channels with analogues having the same helix sense as [Val1]gA. Surprisingly, however, gM- forms hybrid channels with the shortened analogues des-Val1-[Ala2]gA and des-Val1-gC, but these channels differ fundamentally from the parent channels: (a) the appearance rate of these heterodimers is only approximately 1/10 of that predicted from the random assortment of monomers into conducting dimers, indicating the existence of an energy barrier to their formation (e.g., monomer refolding into a new channel-forming conformation); and (b), once formed, the hybrid channels are stabilized approximately 1,000-fold relative to the parent channels. The increased stability suggests a structure that is joined by many hydrogen bonds, such as one of the double-stranded helical dimers shown to be adopted by gramicidins in organic solvents (Veatch, W. R., E. T. Fossel, and E. R. Blout. 1974. Biochemistry. 13:5249-5256).     

Induction of conductance heterogeneity in gramicidin channels

In previous work from our laboratory, 5-10% of the channels formed by [Val1]gramicidin A have conductances that fall outside the narrow range that conventionally has defined the standard gramicidin channel [e.g., see Russell et al. (1986) Biophys. J. 49, 673]. Reports from other laboratories, however, show that up to 50% of [Val1]gramicidin channels have conductances that fall outside the range for standard channels [e.g., see Prasad et al. (1986) Biochemistry 25, 456]. This laboratory-to-laboratory variation in the distribution of gramicidin single-channel conductances suggests that the conductance variants are induced by some environmental factor(s) [Busath et al. (1987) Biophys. J. 51, 79]. In order to test whether extrinsic agents can induce such conductance heterogeneity, we examined the effects of nonionic or zwitterionic detergents upon gramicidin channel behavior. In phospholipid bilayers, detergent addition induces many changes in gramicidin channel behavior: all detergents tested increase the channel appearance rate and average duration; most detergents decrease the conductance of the standard channel; and all but one of the detergents increase the conductance heterogeneity. These results show that the conductance heterogeneity can result from environmental perturbations, thus providing a possible explanation for the laboratory-to-laboratory variation in the heterogeneity of gramicidin channels. In addition, the differential detergent effects suggest possible mechanisms by which detergents can induce the conformational perturbations that result in gramicidin single-channel conductance variations.     

On the conductance heterogeneity in membrane channels formed by gramicidin A. A cooperative study.

The relative frequency of low-conductance variants of gramicidin A channels in lipid bilayers was determined in parallel experiments in two different laboratories. A common gramicidin stock solution was tested in both labs and, initially, gave rise to significantly different proportions (9% v. 23%) of "mini" channels in the two labs. The lipid and gramicidin solutions were exchanged to identify the source of the difference: When using solutions prepared in lab A (Andersen), lab B (Busath) observed 9% minis, consistent with the original findings in lab A; when using the gramicidin solution prepared in lab B, lab A observed 18% minis, consistent with the original findings in lab B. The experimental apparatus and analysis techniques are therefore not the source of the discrepancy; rather, the difference appears to stem from some factor(s) related to the gramicidin, lipid, and electrolyte solutions. It appears that the mini frequency cannot reflect intrinsic characteristics of the channel-forming peptide, but rather must, at least in part, reflect environmental modulations of channel properties. This has implications for the interpretation of multi-channel experiments on gramicidin A.     

Characterization of micellar-packaged gramicidin A channels.

The effects of heating, on an aqueous gramicidin A lysolecithin system, were examined by carbon-13 nuclear magnetic resonance (¹³C-NMR), circular dichroism (CD), and sodium-23 nuclear magnetic resonance (²³Na-NMR), and the results are collectively interpreted to indicate micellar-packaging of gramicidin channels and cation occupancy in the channel. ¹³C-NMR of the gramicidin-lysolecithin system demonstrates a decrease in mobility of the micellar lipid on heating which is indicative of incorporation of gramicidin into the hydrophobic core of the micelle. A unique and reproducible CD spectrum is obtained for the heat incorporated state. Sodium-23 spin-lattice relaxation times (T₁) demonstrated sodium interaction to be dependent on heat incorporation. The T₁ identified interaction is blocked by silver ion which is known to block sodium transport through the channel in lipid bilayer studies. The temperature dependence of the sodium-23 line width defines an exchange process with an activation energy of 6.8 kcal/mole which is essentially the same as the activation energy reported for transport through the channel in lecithin bilayer studies, and the sodium exchange process is blocked by thallium ion which is also known to block sodium transport through the channel.     

Bioactivity of topologically confined gramicidin A dimers

The d-/l-peptide gramicidin A (gA) is well known as a pivotal ion channel model and shows a broad spectrum of bioactivities such as antibiosis, antimalarial activity, as well as hemolysis. We applied inter-chain disulfide bonds to constrain the conformational freedom of gA into parallel and antiparallel dimeric topologies. Albeit the constructs were not found to be monoconformational, CD- and IR-spectroscopic studies suggested that this strategy indeed restricted the conformational space of the d-/l-peptide construct, and that β-helical secondary structures prevail. Correlative testing of gA dimers in antimicrobial, antimalarial, and ion conduction assays suggested that the tail-to-tail antiparallel single stranded β^6.3 helix dominantly mediates the bioactivity of gA. Other conformers are unlikely to contribute to these activities. From these investigations, only weakly ion conducting gA dimers were identified that retained nM antimalarial activity.     

Comparison of gramicidin A and gramicidin M channel conductance dispersities

To explore the possible role of Trp side chains in gramicidin channel conductance dispersity, we studied the dispersity of gramicidin M (gM), a gramicidin variant in which all four tryptophan residues are replaced with phenylalanine residues, and its enantiomer, gramicidin M(-) (gM(-)), and compared them to that of gramicidin A (gA). The conductances of highly purified gM and gM(-) were studied in alkali metal solutions at a variety of concentrations and voltages, in seven different types of lipid, and in the presence of detergent. Like gA channels, the most common gM channel conductance forms a narrow band. However, unlike gA channels, where the remaining 5-30% of channel conductances are broadly distributed below (and slightly above) the main band, in gM there is a narrow secondary band with <50% of the main peak conductance. This secondary peak was prominent in NaCl and KCl, but significantly diminished in CsCl and RbCl. Under some conditions, minor components can be observed with conductances yet lower than the secondary peak. Interconversions between the primary conductance state and these yet lower conductance states were observed. The current-voltage relations for both primary and secondary gM channel types have about the same curvature. The mean lifetime of the secondary channel type is below one third that of the primary type. The variants represent state deviations in the peptide or adjacent lipid structure.     

Distinction between dipolar and inductive effects in modulating the conductance of gramicidin channels.

The ion permeability of transmembrane channels formed by the linear gramicidins is altered by amino acid sequence substitutions. We have previously shown that the polarity of the side chain at position one is important in modulating a channel's conductance and ion selectivity [Russel et al. (1986) Biophys. J. 49, 673-686]. Changes in polarity could alter ion permeability by (through-space) ion-dipole interactions or by (through-bond) inductive electron shifts. We have addressed this question by investigating the permeability characteristics of channels formed by gramicidins where the NH2-terminal amino acid is either phenylalanine or one of a series of substituted phenylalanines: p-hydroxy-, p-methoxy-, o-fluoro-, m-fluoro-, or p-fluorophenylalanine. The electron-donating or -withdrawing nature, as quantified by the Hammett constant, ranges from -0.37 to +0.34 for these side chains. Channels formed by these gramicidins show a more than 2.5-fold variation in their Na+ conductance, but the conductance variations do not rank in the order of the Hammett constants of the side chains. Inductive effects cannot therefore be of primary importance in the modulation of the gramicidin single-channel conductance by these side chains. The results support previous suggestions that electrostatic interactions between side chain dipoles and permeating ions can modify the energy profile for ion movement through the gramicidin channel and thus alter the conductance.     

Membrane surface-charge titration probed by gramicidin A channel conductance.

We manipulate lipid bilayer surface charge and gauge its influence on gramicidin A channel conductance by two strategies: titration of the lipid charge through bulk solution pH and dilution of a charged lipid by neutral. Using diphytanoyl phosphatidylserine (PS) bilayers with CsCl aqueous solutions, we show that the effects of lipid charge titration on channel conductance are masked 1) by conductance saturation with Cs+ ions in the neutral pH range and 2) by increased proton concentration when the bathing solution pH is less than 3. A smeared charge model permits us to separate different contributions to the channel conductance and to introduce a new method for "bilayer pKa" determination. We use the Gouy-Chapman expression for the charged surface potential to obtain equilibria of protons and cations with lipid charges. To calculate cation concentration at the channel mouth, we compare different models for the ion distribution, exact and linearized forms of the planar Poisson-Boltzmann equation, as well as the construction of a "Gibbs dividing surface" between salt bath and charged membrane. All approximations yield the intrinsic pKain of PS lipid in 0.1 M CsCl to be in the range 2.5-3.0. By diluting PS surface charge at a fixed pH with admixed neutral diphytanoyl phosphatidylcholine (PC), we obtain a conductance decrease in magnitude greater than expected from the electrostatic model. This observation is in accord with the different conductance saturation values for PS and PC lipids reported earlier (, Biochim. Biophys. Acta. 552:369-378) and verified in the present work for solvent-free membranes. In addition to electrostatic effects of surface charge, gramicidin A channel conductance is also influenced by lipid-dependent structural factors.     

Low conductance sodium channels in canine cardiac Purkinje cells.

Low conductance sodium (Na) channels have been observed in nerve, skeletal muscle, and cardiac cells. In cardiac tissues the higher amplitude, more commonly observed Na channel was first investigated in detail by Cachelin et al. (Cachelin, A.B., J.E. de Peyer, S. Kokubun, and H. Reuter, 1983, J. Physiol. (Lond.), 340:389-402). They also reported low amplitude Na channel events. We have studied this low conductance Na channel in single canine cardiac Purkinje cells using cell-attached patches. Patch pipette solutions contained either 140 or 280 mM NaCl, and cells were bathed in a solution of 150 mM KCl to bring their resting potential close to zero. In 140 mM Na+, during steps to -50 mV, the lower and higher openings had amplitudes of 0.57 +/- 0.2 and 1.2 +/- 0.2 pA (means +/- SD of Gaussian fits). In 280 mM Na+ at -50 mV, amplitudes were 0.72 +/- 0.2 and 1.55 +/- 0.2 pA. Over a substantial voltage range, the lower events had amplitudes of about one-third that of the higher events. The frequency of the low conductance openings varied in different patches from zero to 22% of total openings. Histograms of open durations and latencies at several voltages suggested no difference in kinetics between the two channel events. The behavior of the low conductance channels was more consistent with a second population of channels rather than a second open state.     

Ion channels formed by chemical analogs of gramicidin A.

Channel-forming peptides such as gramicidin A offer the opportunity to study the relationship between chemical structure and transport properties of an ion channel. This article summarizes a number of recent experiments with chemical analogs and derivatives of gramicidin A using artificial lipid bilayer membranes. The introduction of negative charges near the channel mouth leads to an increase in the cation transport rate. Hybrid channels consisting of a neutral and a negatively charged or of a positively and a negatively charged half-channel may be formed. The current-voltage characteristic of these hybrid channels exhibits a pronounced asymmetry.Experiments with charged derivatives of gramicidin A have been used in order to distinguish between different structural models of the dimeric channel; these studies strongly support Urry's model of a single-stranded, head-to-head associated helical dimer. In a further set of experiments gramicidin analogs with modified amino acid sequence were studied. It was found that a single substitution (tryptophan replaced by phenylalanine) leads to marked changes in the conductance of the channel. Analogs with a simplified amino acid sequence such as (L-Trp-D-Leu)7-L-Trp or L-Trp-Gly-(L-Trp-D-Leu)6-L-Trp are able to form cation permeable channels with similar properties as gramicidin A.     

On the helix sense of gramicidin A single channels.

In order to resolve whether gramicidin A channels are formed by right- or left-handed beta-helices, we synthesized an optically reversed (or mirror image) analogue of gramicidin A, called gramicidin A-, to test whether it forms channels that have the same handedness as channels formed by gramicidin M- (F. Heitz et al., Biophys. J. 40:87-89, 1982). In gramicidin M- the four tryptophan residues have been replaced with phenylalanine, and the circular dichroism (CD) spectrum therefore reflects almost exclusively contributions from the polypeptide backbone. The CD spectrum of gramicidin M- in dimyristoylphosphatidylcholine vesicles is consistent with a left-handed helical backbone folding motif (F. Heitz et al., Biophys. Chem. 24:149-160, 1986), and the CD spectra of gramicidins A and A- are essentially mirror images of each other. Based on hybrid channel experiments, gramicidin A- and M- channels are structurally equivalent, while gramicidin A and A- channels are nonequivalent, being of opposite helix sense. Gramicidin A- channels are therefore left-handed, and natural gramicidin A channels in phospholipid bilayers are right-handed beta 6.3-helical dimers.     

A molecular dynamics study of gating in dioxolane-linked gramicidin A channels.

The gating transition of the RR and SS dioxolane ring-linked gramicidin A channels were studied with molecular dynamics simulations using a detailed atomic model. It was found that the probable reaction path, describing the transition of the ring from the exterior to the interior of the channel where it blocked the permeation pathway, involved several steps including the isomerization of the transpeptide plane dihedral angle of Val1. Reaction coordinates along this pathway were defined, and the transition rates between the stable conformers were calculated. It was found, in good accord with experimental observations, that the calculated blocking rate for the RR-linked channel was 280/s with a mean blocking time of 0.04 ms, whereas such blocking did not occur in the case of the SS-linked channel. An important observation is that the resulting lifetime for the blocked state of the RR-linked channel was in good accord with the experimental observations only when the calculations were performed in the presence of a potassium ion inside the channel.     

The channel properties of possible gramicidin dimers

Extending previous work (Sung & Jordan, 1987 a, Biophys. J. 51, 661-672; 1988, Biophys. J.54, 519-526), we describe channel properties of five possible gramicidin dimers by studying dimerization energies and axial electrical potentials. Unlike the head-to-head dimer (the predominant channel former), both tail-to-tail and head-to-tail dimers with the same beta-helical monomer structure as the head-to-head dimer only form four intermonomer hydrogen bonds and are much less stable. Were channels formed from these dimers to be observed, their electrical potential profiles suggest that they should be cation selective, probably conduct less than the head-to-head dimer, have a central cation binding site, bind cations preferentially if crystallizable, and in the case of the head-to-tail dimer, rectify. Like the antiparallel double stranded helical dimer (a possible minor conducting pathway) the parallel double stranded helical dimer has 28 interstrand hydrogen bonds, but its hydrogen bond network is quite distorted and it is much less stable. If it formed, its electrical potential profile suggests that it would be cation selective, bind anions preferentially if crystallizable, rectify, and at high enough voltages, might exhibit a conductance greater than that of the antiparallel form.     

Properties of gramicidin A channels in erythrocyte membranes

Studies for the cation permeability properties of the gramicidin A channel in erythrocyte membranes are presented. It is shown that gramicidin A interacts with the membrane in a cooperative manner, creating aggregates of the antibiotic molecules in the lipid lattice of the membrane. Cationic channels exist in these aggregates with the following order of selectivity: Rb+ greater than Cs+ greater K+ greater than Na+. The cation permeability of the channels depends on the media surrounding the membrane. This finding has been explained on the basis of Hodgkin-Keynes theory for single-file ion diffusion through extra-narrow pores.     

Photolysis of gramicidin A channels in lipid bilayers

We have tested the hypothesis that peptide tryptophan groups can control the ionic conductance of transmembrane channels. We report here that single gramicidin A channels change conductance state when the peptide tryptophans are flash photolyzed with ultraviolet light. The current flow through planar lipid bilayers containing multiple gramicidin A channels decreases irreversibly when exposed to ultraviolet light. The current-loss action spectrum peaks sharply at the 280 nm absorption maximum of the gramicidin A tryptophans. Gramicidin channel sensitivity to ultraviolet light is found to be about 20-fold higher than that of frog node sodium channels which is even more than expected based on the high tryptophan content of gramicidin. Channels which survive an ultraviolet light exposure exist in a wide variety of different low-conductance forms. The broad distribution of the single channel conductance of these partially photolyzed channels is attributable to the loss of different combinations of the dimer's normal complement of eight tryptophans per channel. Flash photolysis of single channels results in discrete conductance state changes. Partially photolyzed single channels manifest a further conductance cascade when exposed to a second flash of ultraviolet light. Analysis of the photolysis conductance turn-off process indicates that gramicidin A is a multistate electrochemical unit where the peptide tryptophan groups can modulate the flow of ions through the transmembrane channel.     

Energy-minimized conformation of gramicidin-like channels. I. Infinitely long poly-(L,D)-alanine beta 6.3-helix.

The energy-minimized conformation of an infinitely long poly-(L,D)-alanine in single-stranded beta 6.3-helix was calculated by the molecular mechanics method. When energy minimization was started from a wide range of initial geometries, six optimized conformations were obtained and identified as the right- and left-handed counterparts of the beta 4.5-, beta 6.3-, and beta 8.2-helices. It was found that their conformation energies increase in this order, the beta 4.5-helix having the lowest energy. The backbone dihedral angles of the energy-minimized beta 6.3-helix were: phi L = -116 degrees (or -131 degrees), psi L = 122 degrees (or 111 degrees), phi D = 131 degrees (or 116 degrees), psi D = -111 degrees (or -122 degrees), omega L = 173 degrees (or 173 degrees), and omega D = -173 degrees (or -173 degrees) for the right-handed (or left-handed) helix. This helix was composed of 6.30 residues/turn with a pitch of 4.97 A. All the alpha-carbons of L- and D-configurations appeared on one common circular helix. Interestingly, small deviations (approximately 7 degrees) of the peptide bonds from the planar structure caused a considerable lowering of the conformation energy, and, at the same time, they produced more favorable fitting of the hydrogen bonds; the carbonyl oxygens and the nearest-neighbor alpha-hydrogens also took more favorable relative positions.