Interaction of D-amino acid incorporated analogues of pardaxin with membranes.

The influence of specific L- to D-amino acid substitutions on the interaction of pardaxin, a shark repellent neurotoxin polypeptide, with phospholipid vesicles and human erythrocytes is described. Twelve modified, truncated, or fluorescently labeled [with the fluorophore 7-nitrobenz-2-oxa-1,3-diazole-4-yl (NBD) at their N-terminal amino acid] analogues of pardaxin were synthesized by a solid-phase method. Fluorescence measurements were used to monitor the interaction of the analogues with membranes [Rapaport, D., & Shai, Y. (1991) J. Biol. Chem. 266, 23769-23775]. Upon titration of solutions containing the NBD-labeled peptides with small unilamellar vesicles, the fluorescent emission spectra of all NBD-labeled peptides displayed similar blue-shifts, in addition to enhanced intensities, upon relocation of the probe to the more apolar environment. Binding isotherms were constructed from which surface partition constants, in the range of 10(4) M-1, were derived. The existence of an aggregation process, suggested by the shape of the binding isotherms, could be associated only with those analogues in which the N-helix (residues 1-9) was not perturbed. The alpha-helical content of the analogues was estimated by circular dichroism (CD) spectroscopy, both before and after binding to vesicles at neutral pH. The ability of the peptides to dissipate a diffusion potential and to cause calcein release, as well as to lyse human erythrocytes, served to functionally characterize the peptides. The results support a two alpha-helix model, with a bend at position 13, as best describing pardaxin in its membrane-bound state.(ABSTRACT TRUNCATED AT 250 WORDS)     

Membrane interactions of the sodium channel S4 segment and its fluorescently-labeled analogues.

A 24-amino acid peptide corresponding to the S4 segment of the sodium channel was synthesized. In order to perform fluorescence energy transfer measurements and to monitor the interaction of the peptide with lipid vesicles, the peptide was selectively labeled with fluorescence probes at either its N- or C-terminal amino acids. The fluorescent emission spectra of 7-nitrobenz-2-oxa-1,3-diazol-4- yl-(NBD-)labeled analogues displayed blue shifts upon binding to small unilamellar vesicles (SUV), reflecting the relocation of the fluorescent probe to an environment of increased apolarity. The results revealed that both the N- and C-terminus of the S4 segment are located within the lipid bilayer. Titration of solutions containing NBD-labeled peptides with SUV was used to generate binding isotherms, from which surface partition constants, in the range of 10(4) M-1, were derived. The shape of the binding isotherms as well as fluorescence energy transfer measurements suggest that aggregation of peptide monomers within the membrane readily occurs in acidic but not in zwitterionic vesicles. Furthermore, the results provide good correlation between the incidence of aggregation in PC/PS vesicles and the ability of the peptides to permeate the vesicle's membrane. However, a transmembrane diffusion potential had no detectable effect on the location of the peptide within the lipid bilayer or on its aggregation state.(ABSTRACT TRUNCATED AT 250 WORDS)     

Interaction of antimicrobial dermaseptin and its fluorescently labeled analogues with phospholipid membranes.

Dermaseptin, a 34 amino-acid residue antimicrobial polypeptide [Mor, A., Nguyen, V. H., Delfour, A., Migliore-Samour, D., & Nicolas, P. (1991) Biochemistry 30, 8824-8830] was synthesized and selectively labeled at its N-terminal amino acid with either 7-nitrobenz-2-oxa-1,3-diazole-4-yl (NBD), rhodamine, or fluorescein. The fluorescent emission spectra of the NBD-labeled dermaseptin displayed a blue-shift upon binding to small unilamellar vesicles (SUV), reflecting the relocation of the fluorescent probe to an environment of increased apolarity. Titrations of solutions containing NBD-labeled dermaseptin with SUV composed of zwitterionic or acidic phospholipids were used to generate binding isotherms, from which were derived surface partition constants of (0.66 +/- 0.06) x 10(4) M-1 and (2.8 +/- 0.3) x 10(4) M-1, respectively. The shape of the binding isotherms, as well as fluorescence energy transfer measurements, suggests that some aggregation of membrane-bound peptide monomers occurs in acidic but not in zwitterionic vesicles. The preferential susceptibility of the peptide to proteolysis when bound to zwitterionic but not to acidic SUV suggests that these aggregates might then penetrate a relatively short distance into the hydrophobic region of the acidic membrane. Furthermore, the results provide good correlation between the peptide's strong binding and its ability to permeate membranes composed of acidic phospholipids, as revealed by a dissipation of diffusion potential and a release of entrapped calcein from SUV.     

pH-dependent pore formation properties of pardaxin analogues

The interaction of pardaxin, a shark-repellent neurotoxin, and its charge-modified analogues with vesicles and human erythrocytes is described. The following six analogues and derivatives were synthesized by a solid phase method: [Glu8, Glu16]pardaxin, [N1-succinamido,Glu8,Glu16]pardaxin, [N1,Lys8,Lys16-triacetyl]pardaxin, des-[1----9]pardaxin (Shai, Y., Bach, D., and Yanovsky, A. (1990) J. Biol. Chem. 265, 20202-20209), and des-[1----9] [Glu16]pardaxin. The relative hydrophobic characteristics of the analogues were examined using reverse-phase high performance liquid chromatography. The pH-dependent spectroscopic and functional characteristics of the analogues were also investigated at either neutral or acidic pH. Spectroscopic characterization was achieved by measuring circular dichroism both before and after binding to vesicles, at either neutral or acidic pH. The ability of the peptides to dissipate a diffusion potential, to cause calcein release or the pH-dependent release of 8-aminonaphthalene-1,3,6-trisulfonic acid disodium salt/p-xylene-bis[pyridinium bromide] from sonicated unilamellar liposomes, as well as measurements of cytolytic activity on human erythrocytes, served to functionally characterize the peptides. We show a direct correlation between alpha-helical content, the analogues' hydrophobicity, and their pore-forming properties at the different pH values tested. We also demonstrate that the charge of the N terminus and of the peptide backbone, but not of the C terminus, affects the secondary structure as well as the activities of the analogues. Finally, we show that the cytolytic activity of pardaxin at neutral pH is not retained by any of the analogues.     

Channel formation properties of synthetic pardaxin and analogues.

Six analogues of teh 33-residue shark repellent neurotoxin pardaxin were synthesized by the solid phase method: [Ala13]pardaxin, [Gly14,Gly15]pardaxin, des[1----9]pardaxin, [N1-succinamido]pardaxin, C33-dihydroxyethylamido]pardaxin, and C33-[diaminoethylamido]pardaxin. The spectroscopic and functional characterizations of the analogues are described. The peptides were characterized spectroscopically by circular dichroism (CD) before and after binding to soybean vesicles. They were characterized functionally by measuring their potential to evoke the dissipation of diffusion potential and calcein release from sonicated unilamellar soybean liposomes, by determining their ability to create single channels in planar bilayers, and by measuring their cytolytic activity on human erythrocytes. The behavior of the analogues modified at the C terminus is similar to that of pardaxin. [N'-succinamido]Pardaxin, however, reveals an increase in alpha-helicity both alone and in the presence of liposomes. It has the same potency as pardaxin to dissipate diffusion potential, to evoke calcein release and to produce single channels in lipid bilayers, but at a slower rate than that of pardaxin. It has more than 70-fold less cytolytic activity than pardaxin. [Ala13] Pardaxin has twice the alpha-helical content than pardaxin, both alone and in the presence of vesicles, yet it has less effect on the diffusion potential and calcein release, and it does not have cytolytic activity on human erythrocytes. Both [Gly14,Gly15]pardaxin and des[1----9]pardaxin are much less potent than pardaxin in all effects. However des[1----9]pardaxin exhibits a slight change in alpha-helicity upon binding to vesicles, whereas [Gly14,Gly15]pardaxin does not. The results support a model in which pardaxin is composed of two putative alpha-helices separated by proline. The N-terminal alpha-helix is important for the insertion of the peptide to the lipid bilayer, and the C-terminal amphiphilic alpha-helix is the ion channel lining segment of pardaxin.     

Application of a fluorogenic reagent, ammonium 7-fluorobenzo-2-oxa-1,3-diazole-4-sulfonate for detection of cystine-containing peptides

A simple, sensitive and reliable method for the detection of cystine-containing peptides has been developed. A peptide bridged with a disulfide bond was reduced and cleaved with tributylphosphine, and then coupled with a thiol specific reagent, ammonium 7-fluorobenzo-2-oxa-1,3-diazole-4-sulfonate, under alkaline conditions. After incubation at 60 degrees C for 1 h, the fluorescent derivative formed was measured with excitation at 385 nm and emission at 515 nm. The intensity of fluorescence labeled to the peptide was very stable and the peptide containing disulfide was quantitatively determined in the range of 100 pmol to 10 nmol, when oxidized glutathione was used as a standard. This method was very useful for specific detection of cystine-containing peptides in the column effluents on reverse phase high performance liquid chromatography.     

Reversible surface aggregation in pore formation by pardaxin.

The mechanism of leakage induced by surface active peptides is not yet fully understood. To gain insight into the molecular events underlying this process, the leakage induced by the peptide pardaxin from phosphatidylcholine/ phosphatidylserine/cholesterol large unilamellar vesicles was studied by monitoring the rate and extent of dye release and by theoretical modeling. The leakage occurred by an all-or-none mechanism: vesicles either leaked or retained all of their contents. We further developed a mathematical model that includes the assumption that certain peptides become incorporated into the vesicle bilayer and aggregate to form a pore. The current experimental results can be explained by the model only if the surface aggregation of the peptide is reversible. Considering this reversibility, the model can explain the final extents of calcein leakage for lipid/peptide ratios of > 2000:1 to 25:1 by assuming that only a fraction of the bound peptide forms pores consisting of M = 6 +/- 3 peptides. Interestingly, less leakage occurred at 43 degrees C, than at 30 degrees C, although peptide partitioning into the bilayer was enhanced upon elevation of the temperature. We deduced that the increased leakage at 30 degrees C was due to an increase in the extent of reversible surface aggregation at the lower temperature. Experiments employing fluorescein-labeled pardaxin demonstrated reversible aggregation of the peptide in suspension and within the membrane, and exchange of the peptide between liposomes. In summary, our experimental and theoretical results support reversible surface aggregation as the mechanism of pore formation by pardaxin.     

Study of factors affecting the determination of total plasma 7-fluorobenzo-2-oxa-1,3-diazole-4-sulfonate (SBD)–thiol derivatives by liquid chromatography

A detailed investigation of the factors affecting the determination of total plasma 7-fluorobenzo-2-oxa-1,3-diazole-4-sulfonate (SBD)–thiol derivatives (i.e. cysteine, homocysteine and cysteinylglycine) is described. Essentially, this assay entails extracting specific thiols by plasma disulphide bond reduction, protein precipitation, sulphydryl compound derivatization with the thiol-specific fluorogenic reagent ammonium 7-fluorobenzo-2-oxa-1,3-diazole-4-sulphonate (SBD-F), and subsequent separation with isocratic reversed-phase high-performance liquid chromatography. By improving the reliability of several analytical parameters (composition of the mobile phase, pretreatment of the sample using different reducing and protein precipitation agents, and optimization of the derivatization of thiols with SBD-F), a number of critical issues can be identified and solved.     

New Fluorescent Bichromophoric Probes for the Donor–Donor Energy Migration (DDEM) Studies

Three fluorescent probes were synthesized for studying the excitation energy migration between two identical fluorophores. Each probe has two identical fluorescent groups (dansyl, 7-nitrobenzo-2-oxa-1,3-diazole-4-yl, or fluoresceinyl) linked by the rigid bis-(8-aminooctyl)amide of 4,4"-biphenyldicarbonic acid or flexible dotriacontanedioic acid spacer, which enables the intramolecular energy migration through the distance of 3.2–3.5 nm.     

Pardaxin induces aggregation but not fusion of phosphatidylserine vesicles

The effects on membranes of pardaxin, an amphipathic polypeptide, purified from the gland secretion of the Red Sea Moses sole flatfish Pardachirus marmoratus are dose-dependent and range from formation of voltage-gated, cation-selective pores to lysis. We have now investigated the interactions of pardaxin with small unilamellar liposomes. Light scattering showed that pardaxin (10⁻⁷–10⁻⁹M) mediated the aggregation of liposomes composed of phosphatidylserine but not of phosphatidylcholine. Aggregation of phosphatidylserine vesicles was impaired by vesicle depolarization. Furthermore, pardaxin-mediated aggregation between fluorescent-labeled PS vesicles was accompanied by leakage of the vesicle contents, and not by fusogenic process within the aggregates. We suggest that pardaxin is a unique polypeptide, that induces vesicle aggregation and membrane destabilization, but not membrane fusion; the mechanism of the aggregation activity of pardaxin is related to its amphipathic properties.     

New fluorescent bichromophoric probes for studying mechanisms of donor-donor energy migration (DDEM)

Three fluorescent probes were synthesized for studying the excitation energy migration between two identical fluorophores. Each probe has two identical fluorescent groups (dansyl, 7-nitrobenz-2-oxa-1,3-diazole-4-yl, or fluoresceinyl) linked by the rigid bis-(8-aminooctyl)amide of 4,4'-biphenyldicarbonic acid or flexible dotriacontanedioic acid spacer, which enables the intramolecular energy migration through the distance of 3.2-3.5 nm.     

Helix-helix association of a lipid-bound amphipathic alpha-helix derived from apolipoprotein C-II.

The interaction of a peptide derived from the sequence of apolipoprotein C-II (apoC-II) with a model lipid surface has been investigated by fluorescence spectroscopy. ApoC-II19-39, labeled at the N-terminus with 7-nitrobenz-2-oxa-1,3-diazole (NBD), bound to small unilamellar vesicles of phosphatidylcholine with a dissociation constant of 6 microM. The lipid-bound NBD-labeled peptide exhibited a red-edge excitation shift in its emission maximum and anisotropy, consistent with insertion of the probe into the motionally restricted, polar environment provided by the bilayer interface. The small Stokes shift of the NBD fluorophore permits electronic energy homotransfer between peptides on the lipid surface and results in depolarization of the NBD emission. At high surface densities of lipid-bound peptide, the anisotropy of the NBD probe was 33% lower than in corresponding samples in which electronic energy homotransfer was prevented by the addition of an unlabeled peptide. The efficiency of energy transfer between probes was not consistent with a random distribution of peptides on the lipid surface, indicating instead the self-association of lipid-bound apoC-II19-39. We propose that the role of this sequence in apoC-II is not only to mediate binding of protein to a lipid surface, but also to stabilize the lipoprotein complexes by associating with other amphipathic helices within apoC-II and with other apolipoproteins.     

Large-scale preparation of asymmetrically labeled fluorescent lipid vesicles

A method for producing lipid vesicles containing fluorescent phospholipid analogues localized to the inner leaflet of their membrane was developed. Incubation of a 450-fold molar excess of serum albumin with lipid vesicles symmetrically labeled with 1 mol % 1-palmitoyl-2-(N-4-nitrobenzo-2-oxa-1,3-diazolyl)amino-caproyl phosphatidylcholine resulted in the removal of 99% of the fluorescent lipid from the outer leaflet. Asymmetrically labeled vesicles were separated from albumin/lipid complexes by gel filtration chromatography. Vesicles prepared in this manner were unable to transfer fluorescent lipid to cells during liposome-cell incubations. Liposomes asymmetrically labeled with other 4-nitrobenzo-2-oxa-1,3-diazole (NBD)-phospholipid analogues were also prepared. Removal of amino-dodecanoyl-NBD-labeled lipids from the outer leaflet of liposomes required three times more bovine serum albumin, and 48 h of incubation. This method can be used to produce large amounts of asymmetrically labeled liposomes suitable for use in investigating a variety of membrane phenomena.     

Transport of exogenous fluorescent phosphatidylserine analogue to the Golgi apparatus in cultured fibroblasts

We have examined intracellular transport and metabolism of the fluorescent analogue of phosphatidylserine, 1-palmitoyl-2-(N-[12[(7-nitrobenz-2-oxa-1,3-diazole-4-yl)amino] dodecanoyl])-phosphatidylserine ([palmitoyl-C12-NBD]-PS) in cultured fibroblasts. When monolayer cultures were incubated with liposomes containing (palmitoyl-C12-NBD)-PS at 37 degrees C, fluorescent PS was transported to the Golgi apparatus. NBD-containing analogues of phosphatidylcholine, phosphatidylethanolamine (PE), or phosphatidic acid did not accumulate in the Golgi apparatus under the same experimental conditions. We suggest that the transport is not due to endocytosis, but is the result of incorporation and trans-bilayer movement of the (palmitoyl-C12-NBD)-PS at the plasma membrane followed by translocation of the lipid from plasma membrane to the Golgi apparatus via nonvesicular mechanisms. Uptake of fluorescent PS was inhibited by depletion of cellular ATP and was blocked by structural analogues of the lipid or by pretreatment of cells with glutaraldehyde or N-ethylmaleimide. After incorporation into the cell, fluorescent PS was metabolized to fluorescent PE. The intracellular distribution of fluorescence changed during the conversion. In addition to the Golgi apparatus, mitochondria also became labeled.     

Interfacial activation of porcine pancreatic phospholipase A(2) studied with 7-nitrobenz-2-oxa-1,3-diazol-4-yl-labeled lipids

The interfacial activation of porcine pancreatic phospholipase A(2) (PLA(2)) during the hydrolysis of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine liposomes at different temperatures has been monitored by fluorescence changes of the 7-nitrobenz-2-oxa-1,3-diazol-4-yl (NBD) lipid derivatives 1-palmitoyl-2-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]dodecanoyl]-sn-glycero-3-phosphocholine (C(12)-NBD-PC) and 12-[(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)]dodecanoic acid (C(12)-NBD-FA) inserted in the substrate vesicles. These long-chain monitors, in contrast to the previously used C(6)-NBD-PC, detect latency times of PLA(2) action, similar to those measured by the classic titrimetric, pH-stat method. Interestingly, hydrolysis of the host vesicles results in a decrease in fluorescence not only of C(12)-NBD-PC, a substrate analog, but also of product derivative C(12)-NBD-FA. Ultrafiltration experiments show that C(12)-NBD-FA does not migrate to the aqueous phase upon hydrolysis of the host liposomes. Besides, in a simulated hydrolysis experiment in which increasing proportions of palmitic acid and 1-palmitoyl-sn-glycero-3-phosphocholine were cosonicated with 1,2-dipalmitoyl-sn-glycero-3-phosphocholine, C(12)-NBD-PC fluorescence was insensitive to products, whereas C(12)-NBD-FA did show a decreased emission intensity as in the actual hydrolysis experiments. The phenomenon is triggered above a critical concentration of products (10 mol%) suggesting that cosegregation of NBD-FA (either added as such or generated by hydrolysis of C(12)-NBD-PC) and products may be related to the decrease in fluorescence. Phase separation should create microdomains of increased C(12)-NBD-FA surface density and cause concentration quenching. In addition, and taking into account that the NBD group may be located near the interfacial region, it is possible that in segregating with products, the fluorescent moiety of C(12)-NBD-FA becomes exposed to microenvironments of higher surface polarity, which further decreases its quantum yield.     

Insertion and pore formation driven by adsorption of proteins onto lipid bilayer membrane-water interfaces

We describe the binding of proteins to lipid bilayers in the case for which binding can occur either by adsorption to the lipid bilayer membrane-water interface or by direct insertion into the bilayer itself. We examine in particular the case when the insertion and pore formation are driven by the adsorption process using scaled particle theory. The adsorbed proteins form a two-dimensional "surface gas" at the lipid bilayer membrane-water interface that exerts a lateral pressure on the lipid bilayer membrane. Under conditions of strong intrinsic binding and a high degree of interfacial converge, this pressure can become high enough to overcome the energy barrier for protein insertion. Under these conditions, a subtle equilibrium exists between the adsorbed and inserted proteins. We propose that this provides a control mechanism for reversible insertion and pore formation of proteins such as melittin and magainin. Next, we discuss experimental data for the binding isotherms of cytochrome c to charged lipid membranes in the light of our theory and predict that cytochrome c inserts into charged lipid bilayers at low ionic strength. This prediction is supported by titration calorimetry results that are reported here. We were furthermore able to describe the observed binding isotherms of the pore-forming peptides endotoxin (alpha 5-helix) and of pardaxin to zwitterionic vesicles from our theory by assuming adsorption/insertion equilibrium.     

Solution conformations of peptides representing the sequence of the toxin pardaxin and analogues in trifluoroethanol-water mixtures: Analysis of CD spectra

The cytolytic activities and conformational properties of pardaxin (GFFALIPKIISSPLFKTLLSAVGSALSSSGEQE), a 33-residue linear peptide that exhibits unusual shark repellent and cytolytic activities, and its analogues have been examined in aqueous environment and trifluoroethanol (TFE) using CD spectroscopy. A peptide corresponding to the 1–26 segment and an analogue where P7 has been changed to A show greater hemolytic activity than pardaxin. While the peptide corresponding to the N-terminal 18-residue segment does not exhibit hemolytic activity, its analogue where P7 is replaced by A is hemolytic. The secondary structural propensities of the peptides were inferred by deconvolution of the experimental spectra into pure components. Pardaxin, its variant where proline at position 7 was replaced by alanine, and shorter peptides corresponding to N-terminal segments exist in multiple conformations in aqueous medium that are comprised of β-turn, β-sheet, and distorted helical structures. With increasing proportions of TFE, while helical conformation predominates in all the peptides, both distorted and the regular α-helices appear to be populated. Analysis of CD spectra by deconvolution methods appears to be a powerful tool for delineating multiple conformations in peptides, especially membrane-active peptides that encounter media of different polarity ranging from aqueous environment to one of low dielectric constant in the hydrophobic interior of membranes. Our study provides further insights into the structural requirements for the biological activity of pardaxin and related peptides. © 1997 John Wiley & Sons, Inc. Biopoly 41: 635–645, 1997     

Dehydroergosterol structural organization in aqueous medium and in a model system of membranes.

The aggregation of delta 5,7,9(11),22-ergostatetraen-3 beta-ol (dehydroergosterol or DHE), a fluorescent analog of cholesterol, was studied by photophysical techniques. It was concluded that the aqueous dispersions of DHE consist of strongly fluorescent microcrystals, and no evidence for self-quenching in micellar-type aggregates was found. The organization of DHE in model systems of membranes (phospholipid vesicles) is strongly dependent on the vesicle type. In small unilamellar vesicles, no evidence for aggregation is obtained, and the fluorescence anisotropy is rationalized on the basis of a random distribution of fluorophores. On the contrary, in large unilamellar vesicles (LUVs), a steeper concentration depolarization was observed. To explain this, a model that takes into account transbilayer dimer formation was derived. This was further confirmed from observation of excitonic absorption bands of 22-(N-7-nitrobenz-2-oxa-1,3-diazol-4-yl-amino)-23,24-bisnor- 5-cholen-3 beta-ol (NBD-cholesterol) in LUV, which disappear upon sonication. It is concluded that, in agreement with recent works, sterol aggregation is a very efficient process in large vesicles (and probably in natural membranes), even at very low concentrations (approximately 5 mol%).     

The structure and organization of synthetic putative membranous segments of ROMK1 channel in phospholipid membranes.

The hydropathy plot of ROMK1, an inwardly rectifying K+ channel, suggests that the channel contains two transmembrane domains (M1 and M2) and a linker between them with significant homology to the H5 pore region of voltage-gated K+ channels. To gain structural information on the pore region of the ROMK1 channel, we used a spectrofluorimetric approach and characterized the structure, the organization state, and the ability of the putative membranous domains of the ROMK1 channel to self-assemble and coassemble within lipid membranes. Circular dichroism (CD) spectroscopy revealed that M1 and M2 adopt high alpha-helical structures in egg phosphatidylcholine small unilamellar vesicles and 40% trifluoroethanol (TFE)/water, whereas H5 is not alpha-helical in either egg phosphatidylcholine small unilamellar vesicles or 40% TFE/water. Binding experiments with 4-fluoro-7-nitrobenz-2-oxa-1,3-diazole (NBD)-labeled peptide demonstrated that all of the peptides bind to zwitterionic phospholipid membranes with partition coefficients on the order of 10(5) M-1. Tryptophan quenching experiments using brominated phospholipids revealed that M1 is dipped into the hydrophobic core of the membrane. Resonance energy transfer (RET) measurements between fluorescently labeled pairs of donor (NBD)/acceptor (rhodamine) peptides revealed that H5 and M2 can self-associate in their membrane-bound state, but M1 cannot. Moreover, the membrane-associated nonhelical H5 serving as a donor can coassemble with the alpha-helical M2 but not with M1, and M1 can coassemble with M2. No coassembly was observed between any of the segments and a membrane-embedded alpha-helical control peptide, pardaxin. The results are discussed in terms of their relevance to the proposed topology of the ROMK1 channel, and to general aspects of molecular recognition between membrane-bound polypeptides.