Computational studies of colicin insertion into membranes: the closed state

Colicins are water-soluble toxins that, upon interaction with membranes, undergo a conformational change, insert, and form pores in them. Pore formation activity is localized in a bundle of 10 α-helices named the pore-forming domain (PFD). There is evidence that colicins attach to the membrane via a hydrophobic hairpin embedded in the core of the PFD. Two main models have been suggested for the membrane-bound state: penknife and umbrella, differing in regard to the orientation of the hydrophobic hairpin with respect to the membrane. The arrangement of the amphipathic helices has been described as either a compact three-dimensional structure or a two-dimensional array of loosely interacting helices on the membrane surface. Using molecular dynamics simulations with an implicit membrane model, we studied the structure and stability of the conformations proposed earlier for four colicins. We find that colicins are initially driven towards the membrane by electrostatic interactions between basic residues and the negatively charged membrane surface. They do not have a unique binding orientation, but in the predominant orientations the central hydrophobic hairpin is parallel to the membrane. In the inserted state, the estimated free energy tends to be lower for the compact arrangements of the amphipathic helix, but the more expanded ones are in better agreement with experimental distance distributions. The difference in energy between penknife and umbrella conformations is small enough for equilibrium to exist between them. Elongation of the hydrophobic hairpin helices and membrane thinning were found unable to produce stabilization of the transmembrane configuration of the hydrophobic hairpin.     

The helix bundle: A reversible lipid binding motif

Apolipoproteins are the protein components of lipoproteins that have the innate ability to inter convert between a lipid-free and a lipid-bound form in a facile manner, a remarkable property conferred by the helix bundle motif. Composed of a series of four or five amphipathic α-helices that fold to form a helix bundle, this motif allows the en face orientation of the hydrophobic faces of the α-helices in the protein interior in the lipid-free state. A conformational switch then permits helix–helix interactions to be substituted by helix–lipid interactions upon lipid binding interaction. This review compares the apolipoprotein high-resolution structures and the factors that trigger this switch in insect apolipophorin III and the mammalian apolipoproteins, apolipoprotein E and apolipoprotein A-I, pointing out the commonalities and key differences in the mode of lipid interaction. Further insights into the lipid-bound conformation of apolipoproteins are required to fully understand their functional role under physiological conditions.     

A novel amphipathic cell-penetrating peptide based on the N-terminal glycosaminoglycan binding region of human apolipoprotein E

In the direct cell membrane penetration, arginine-rich cell-penetrating peptides are thought to penetrate into cells across the hydrophobic lipid membranes. To investigate the effect of the amphipathic property of arginine-rich peptide on the cell-penetrating ability, we designed a novel amphipathic cell-penetrating peptide, A2-17, and its derivative, A2-17KR, in which all lysine residues are substituted with arginine residues, based on the glycosaminoglycan binding region in the N-terminal α-helix bundle of human apolipoprotein E. Isothermal titration calorimetry showed that A2-17 variants have a strong ability to bind to heparin with high affinity. Circular dichroism and tryptophan fluorescence measurements demonstrated that A2-17 variants bind to lipid vesicles with a structural change from random coil to amphipathic α-helix, being inserted into the hydrophobic membrane interiors. Flow cytometric analysis and confocal laser scanning microscopy demonstrated the great cell penetration efficiency of A2-17 variants into CHO-K1 cells when incubated at low peptide concentrations (2 μM or less), suggesting that the increased amphipathicity with α-helix formation enhances the cell membrane penetration ability of arginine-rich peptides. Interestingly, A2-17KR exhibited lower efficiency of cell membrane penetration compared to A2-17 despite of their similar binding affinity to lipid membranes. Since high peptide concentrations (typically >10 μM) are usually prerequisite for efficient cell penetration of arginine-rich peptides, A2-17 is a unique amphipathic cell-penetrating peptide that exhibits an efficient cell penetration ability even at low peptide concentrations.     

Configurations of the N-terminal amphipathic domain of the membrane-bound M13 major coat protein

The M13 major coat protein has been extensively studied in detergent-based and phospholipid model systems to elucidate its structure. This resulted in an L-shaped model structure of the protein in membranes. An amphipathic alpha-helical N-terminal arm, which is parallel to the surface of the membrane, is connected via a flexible linker to an alpha-helical transmembrane domain. In the present study, a fluorescence polarity probe or ESR spin probe is attached to the SH group of a series of N-terminal single cysteine mutants, which were reconstituted into DOPC model membranes. With ESR spectroscopy, we measured the local mobility of N-terminal positions of the protein in the membrane. This is supplemented with relative depth measurements at these positions by fluorescence spectroscopy via the wavelength of maximum emission and fluorescence quenching. Results show the existence of at least two possible configurations of the M13 amphipathic N-terminal arm on the ESR time scale. The arm is bound either to the membrane surface or in the water phase. The removal or addition of a hydrophobic membrane-anchor by site-specific mutagenesis changes the ratio between the membrane-bound and the water phase fraction.     

Membrane binding of MARCKS-related protein studied by tryptophan fluorescence spectroscopy

MARCKS-related protein (MRP) is a peripheral membrane protein whose binding to membranes is mediated by the N-terminal myristoyl moiety and a central, highly basic effector domain. MRP mediates cross-talk between protein kinase C and calmodulin and is thought to link the actin cytoskeleton to the plasma membrane. Since MRP contains no tryptophan residues, we mutated a phenylalanine in the effector domain to tryptophan (MRP F93W) and used fluorescence spectroscopy to monitor binding of the protein to phospholipid vesicles. We report in detail the evaluation procedure necessary to extract quantitative information from the raw data. The spectra of MRP F93W obtained in the presence of increasing amounts of lipid crossed at an isosbestic point, indicating a simple transition between two states: free and membrane-bound protein. The change in fluorescence toward values typical of a more hydrophobic environment was used to quantify membrane binding. The partition coefficient agreed well with values obtained previously by other methods. To study the interaction of the N-terminus of MRP with membranes, a tryptophan residue was also introduced at position 4 (MRP S4W). Our data suggest that only the myristoylated N-terminus interacted with liposomes. These results demonstrate the versatility of site-directed incorporation of tryptophan residues to study protein-membrane interactions.     

Förster resonance energy transfer measurements are consistent with a helical bundle model for lipid-free apolipoprotein A-I

Apolipoprotein (apo) A-I mutants were constructed for FRET studies to distinguish between two possible lipid-free conformers, a globular helix bundle and an elongated helical hairpin. Mutants containing a single Trp at position 50 were prepared by replacing Trps at positions 8, 72, and 108 with Phe (W@50). Two mutants were constructed from W@50 by incorporating Cys at Arg83 (W@50R83C) or Arg173 (W@50R173C) for attachment of the fluorescent probe AEDANS. Secondary structure of the mutants is very similar to wild type (wt) apo A-I, and fluorescence emission indicates that W50 is protected from solvent. Thermal stabilities of the AEDANS-labeled mutants are also similar to wt. These results indicate that no discernible changes occur in structure or stability as a result of mutations or labeling. The FRET data from W@50 to AEDANS are well-represented by a single distance distribution function with a distance of approximately 22 A for W@50R83C and approximately 19 A for W@50R173C. These distances are consistent with theoretical values calculated from a helical bundle model but not from a helical hairpin. A probability distance distribution function yields significantly small half-width values of 5.6 and 3.7 A, respectively, suggesting low conformational dynamics in both mutants. Differential scanning calorimetry (DSC) was performed on wt and a C-terminal deletion mutant, Delta(187-243), to obtain information on domain architecture. Contrary to expectations, both proteins unfold cooperatively. The results are consistent with the presence of a single folded domain within residues 1-186. These results support the presence of a discrete globular bundle conformation for lipid-free apo A-I.     

Lipid binding specificity of bovine α-lactalbumin: A multidimensional approach

Many soluble proteins are known to interact with membranes in partially disordered states, and the mechanism and relevance of such interactions in cellular processes are beginning to be understood. Bovine α-lactalbumin (BLA) represents an excellent prototype for monitoring membrane interaction due to its conformational plasticity. In this work, we comprehensively monitored the interaction of apo-BLA with zwitterionic and negatively charged membranes utilizing a variety of approaches. We show that BLA preferentially binds to negatively charged membranes at acidic pH with higher binding affinity. This is supported by spectral changes observed with a potential-sensitive membrane probe and fluorescence anisotropy measurements of a hydrophobic probe. Our results show that BLA exhibits a molten globule conformation when bound to negatively charged membranes. We further show, using the parallax approach, that BLA penetrates the interior of negatively charged membranes, and tryptophan residues are localized at the membrane interface. Red edge excitation shift (REES) measurements reveal that the immediate environment of tryptophans in membrane-bound BLA is restricted, and the restriction is dependent on membrane lipid composition. We envision that understanding the mechanism of BLA–membrane interaction would help in bioengineering of α-lactalbumin, and to address the mechanism of tumoricidal and antimicrobial activities of BLA–oleic acid complex.     

Structural features of split and unsplit βαβ-units

In this study, 1064 nonhomologous “unsplit”, “one-strand split” and “two-strand split” right-handed βαβ-units having standard α-helices and loops up to seven residues in length have been analyzed. It was found that the α-helices in these kinds of βαβ-units have different distributions of the hydrophobic and hydrophilic amino acid residues along the chain. In the unsplit βαβ-units, most α-helices have hydrophobic residues in positions N4-N7-N8-N11 or N6-N7-N10, where N1 is the first N-terminal residue. In the one-strand split βαβ-units, most α-helices have hydrophobic residues in positions N4-N7-N8-N11 and those in two-strand split βαβ-units in positions N4-N5-N8-N12. On the other hand, in all kinds of βαβ-units, there are commonly occurring hydrophobic stripes of type C4-C7-C8 at the C-terminal parts of the α-helices. As a rule, the C- and N-terminal hydrophobic stripes overlap and the extent of their overlapping determine the length of α-helices.     

Solution structure and lipid membrane partitioning of VSTx1, an inhibitor of the KvAP potassium channel

VSTx1 is a voltage sensor toxin from the spider Grammostola spatulata that inhibits KvAP, an archeabacterial voltage-activated K(+) channel whose X-ray structure has been reported. Although the receptor for VSTx1 and the mechanism of inhibition are unknown, the sequence of the toxin is related to hanatoxin (HaTx) and SGTx, two toxins that inhibit eukaryotic voltage-activated K(+) channels by binding to voltage sensors. VSTx1 has been recently shown to interact equally well with lipid membranes that contain zwitterionic or acidic phospholipids, and it has been proposed that the toxin receptor is located within a region of the channel that is submerged in the membrane. As a first step toward understanding the inhibitory mechanism of VSTx1, we determined the three-dimensional solution structure of the toxin using NMR. Although the structure of VSTx1 is similar to HaTx and SGTx in terms of molecular fold and amphipathic character, the detailed positions of hydrophobic and surrounding charged residues in VSTx1 are very different than what is seen in the other toxins. The amphipathic character of VSTx1, notably the close apposition of basic and hydrophobic residues on one face of the toxin, raises the possibility that the toxin interacts with interfacial regions of the membrane. We reinvestigated the partitioning of VSTx1 into lipid membranes and find that VSTx1 partitioning requires negatively charged phospholipids. Intrinsic tryptophan fluorescence and acrylamide quenching experiments suggest that tryptophan residues on the hydrophobic surface of VSTx1 have a diminished exposure to water when the toxin interacts with membranes. The present results suggest that if membrane partitioning is involved in the mechanism by which VSTx1 inhibits voltage-activated K(+) channels, then binding of the toxin to the channel would likely occur at the interface between the polar headgroups and the hydrophobic phase of the membrane.     

Lipid-protein interactions

Intrinsic membrane proteins are solvated by a shell of lipid molecules interacting with the membrane-penetrating surface of the protein; these lipid molecules are referred to as annular lipids. Lipid molecules are also found bound between transmembrane α-helices; these are referred to as non-annular lipids. Annular lipid binding constants depend on fatty acyl chain length, but the dependence is less than expected from models based on distortion of the lipid bilayer alone. This suggests that hydrophobic matching between a membrane protein and the surrounding lipid bilayer involves some distortion of the transmembrane α-helical bundle found in most membrane proteins, explaining the importance of bilayer thickness for membrane protein function. Annular lipid binding constants also depend on the structure of the polar headgroup region of the lipid, and hotspots for binding anionic lipids have been detected on some membrane proteins; binding of anionic lipid molecules to these hotspots can be functionally important. Binding of anionic lipids to non-annular sites on membrane proteins such as the potassium channel KcsA can also be important for function. It is argued that the packing preferences of the membrane-spanning α-helices in a membrane protein result in a structure that matches nicely with that of the surrounding lipid bilayer, so that lipid and protein can meet without either having to change very much.     

Dissecting quasi-equivalence in nonenveloped viruses: membrane disruption is promoted by lytic peptides released from subunit pentamers, not hexamers

Nonenveloped viruses often invade membranes by exposing hydrophobic or amphipathic peptides generated by a proteolytic maturation step that leaves a lytic peptide noncovalently associated with the viral capsid. Since multiple copies of the same protein form many nonenveloped virus capsids, it is unclear if lytic peptides derived from subunits occupying different positions in a quasi-equivalent icosahedral capsid play different roles in host infection. We addressed this question with Nudaurelia capensis omega virus (NωV), an insect RNA virus with an icosahedral capsid formed by protein α, which undergoes autocleavage during maturation, producing the lytic γ peptide. NωV is a unique model because autocatalysis can be precisely initiated in vitro and is sufficiently slow to correlate lytic activity with γ peptide production. Using liposome-based assays, we observed that autocatalysis is essential for the potent membrane disruption caused by NωV. We observed that lytic activity is acquired rapidly during the maturation program, reaching 100% activity with less than 50% of the subunits cleaved. Previous time-resolved structural studies of partially mature NωV particles showed that, during this time frame, γ peptides derived from the pentamer subunits are produced and are organized in a vertical helical bundle that is projected toward the particle surface, while identical polypeptides in quasi-equivalent subunits are produced later or are in positions inappropriate for release. Our functional data provide experimental support for the hypothesis that pentamers containing a central helical bundle, observed in different nonenveloped virus families, are a specialized lytic motif.     

Membrane interactions of antimicrobial peptides from Australian frogs

The membrane interactions of four antimicrobial peptides, aurein 1.2, citropin 1.1, maculatin 1.1 and caerin 1.1, isolated from Australian tree frogs, are reviewed. All four peptides are amphipathic α-helices with a net positive charge and range in length from 13 to 25 residues. Despite several similar sequence characteristics, these peptides compromise the integrity of model membrane bilayers via different mechanisms; the shorter peptides exhibit a surface interaction mechanism while the longer peptides may form pores in membranes.     

The determinants of hydrophobic mismatch response for transmembrane helices

Hydrophobic mismatch arises from a difference in the hydrophobic thickness of a lipid membrane and a transmembrane protein segment, and is thought to play an important role in the folding, stability and function of membrane proteins. We have investigated the possible adaptations that lipid bilayers and transmembrane α-helices undergo in response to mismatch, using fully-atomistic molecular dynamics simulations totaling 1.4 μs. We have created 25 different tryptophan-alanine-leucine transmembrane α-helical peptide systems, each composed of a hydrophobic alanine–leucine stretch, flanked by 1–4 tryptophan side chains, as well as the β-helical peptide dimer, gramicidin A. Membrane responses to mismatch include changes in local bilayer thickness and lipid order, varying systematically with peptide length. Adding more flanking tryptophan side chains led to an increase in bilayer thinning for negatively mismatched peptides, though it was also associated with a spreading of the bilayer interface. Peptide tilting, bending and stretching were systematic, with tilting dominating the responses, with values of up to ~ 45° for the most positively mismatched peptides. Peptide responses were modulated by the number of tryptophan side chains due to their anchoring roles and distributions around the helices. Potential of mean force calculations for local membrane thickness changes, helix tilting, bending and stretching revealed that membrane deformation is the least energetically costly of all mismatch responses, except for positively mismatched peptides where helix tilting also contributes substantially. This comparison of energetic driving forces of mismatch responses allows for deeper insight into protein stability and conformational changes in lipid membranes.     

Membrane binding mechanism of an RNA virus-capping enzyme

The RNA replication complex of Semliki Forest virus is bound to cytoplasmic membranes via the mRNA-capping enzyme Nsp1. Here we have studied the structure and liposome interactions of a synthetic peptide (245)GSTLYTESRKLLRSWHLPSV(264) corresponding to the membrane binding domain of Nsp1. The peptide interacted with liposomes only if negatively charged lipids were present that induced a structural change in the peptide from a random coil to a partially alpha-helical conformation. NMR structure shows that the alpha-helix is amphipathic, the hydrophobic surface consisting of several leucines, a valine, and a tryptophan moiety (Trp-259). Fluorescence studies revealed that this tryptophan intercalates in the bilayer to the depth of the ninth and tenth carbons of lipid acyl chains. Mutation W259A altered the mode of bilayer association of the peptide and abolished its ability to compete for membrane association of intact Nsp1, demonstrating its crucial role in the membrane association and function of Nsp1.     

Packing interactions of aib-containing helices: Molecular modeling of parallel dimers of simple hydrophobic helices and of alamethicin

α-Aminoisobutyric acid (Aib) is a helicogenic α,α-dimethyl amino acid found in channel-forming peptaibols such as alamethicin. Possible effects of Aib on helix–helix packing are analyzed. Simulated annealing via restrained molecular dynamics is used to generate ensembles of approximately parallel helix dimers. Analysis of variations in geometrical and energetic parameters within ensembles defines how tightly a pair of helices interact. Simple hydrophobic helix dimers are compared: Ala₂₀, Leu₂₀, Aib₂₀, and P20, the latter a simple channel-forming peptide [G. Menestrina, K. P. Voges, G, Jung, and G. Boheim (1986) Journal of Membrane Biology, Vol. 93, pp. 111–132]. Ala₂₀ and Leu₂₀ dimers exhibit well-defined ridges-in-grooves packing with helix crossing angles (Ω) of the order of +20°. Aib₂₀ α-helix dimers are much more loosely packed, as evidenced by a wide range of Ω values and small helix-helix interaction energies. However, when in a 3₁₀ conformation Aib₂₀ helices pack in three well-defined parallel modes, with Ω ca. ?15°, +5°, and 10°. Comparison of helix–helix interaction energies suggests that dimerization may favor the 3₁₀ conformation. P20, with 8 Aib residues, also shows looser packing of α-helices. The results of these studies of hydrophobic helix dimers are analyzed in the context of the ridges-in-grooves packing model. Simulations are extended to dimers of alamethicin, and of an alamethicin derivative in which all Aib residues are replaced by Leu. This substitution has little effect on helix–helix packing. Rather, such interactions appear to be sensitive to interactions between polar side chains. Overall, the results suggest that Aib may modulate the packing of simple hydrophobic helices, in favor of looser interactions. For more complex amphipathic helices, interactions between polar side chains may be more critical. © 1995 John Wiley & Sons, Inc.     

Conformational changes of colicin Ia channel-forming domain upon membrane binding: a solid-state NMR study

Channel-forming colicins are bactericidal proteins that spontaneously insert into hydrophobic lipid bilayers. We have used magic-angle spinning solid-state nuclear magnetic resonance spectroscopy to examine the conformational differences between the water-soluble and the membrane-bound states of colicin Ia channel domain, and to study the effect of bound colicin on lipid bilayer structure and dynamics. We detected ¹³C and ¹⁵N isotropic chemical shift differences between the two forms of the protein, which indicate structural changes of the protein due to membrane binding. The Val Cα signal, unambiguously assigned by double-quantum experiments, gave a 0.6 ppm downfield shift in the isotropic position and a 4 ppm reduction in the anisotropic chemical shift span after membrane binding. These suggest that the α-helices in the membrane-bound colicin adopt more ideal helical torsion angles as they spread onto the membrane. Colicin binding significantly reduced the lipid chain order, as manifested by ²H quadrupolar couplings. These results are consistent with the model that colicin Ia channel domain forms an extended helical array at the membrane-water interface upon membrane binding.     

On the stoichiometry of Deinococcus radiodurans Dps-1 binding to duplex DNA

DNA protection during starvation (Dps) proteins, dodecameric assemblies of four-helix bundle subunits, contribute to protection against reactive oxygen species. Deinococcus radiodurans, which is characterized by resistance to DNA damaging agents, encodes two Dps homologs, of which Dps-1 binds DNA with high affinity. DNA binding requires N-terminal extensions preceding the four-helix bundle core. Composed of six Dps-1 dimers, each capable of DNA binding by N-terminal extensions interacting in consecutive DNA major grooves, dodecameric Dps-1 would be predicted to feature six DNA binding sites. Using electrophoretic mobility shift assays and intrinsic tryptophan fluorescence, we show that dodecameric Dps-1 binds 22-bp DNA with a stoichiometry of 1:6, consistent with the existence of six DNA binding sites. The stoichiometry of Dps-1 binding to 26-bp DNA is 1:4, suggesting that two Dps-1 dodecamers can simultaneously occupy opposite faces of this DNA. Mutagenesis of an arginine (Arg132) on the surface of Dps-1 leads to a reduction in DNA binding. Altogether, our data suggest that duplex DNA lies along the dimer interface, interacting with Arg132 and the N-terminal α-helices, and they extend the hexagonal packing model for Dps-DNA assemblies by specifying the basis for occupancy of available DNA binding sites.     

Fluorescence energy transfer distance measurements using site-directed single cysteine mutants. The membrane insertion of colicin A

The ion-channel-forming C-terminal fragment of colicin A binds to negatively charged lipid vesicles and provides an example of insertion of a soluble protein into a lipid bilayer. The soluble structure is known from X-ray crystallography and consists of a ten-helix bundle containing a hydrophobic helical hairpin. In this work fluorescence spectroscopy was used to study the membrane-bound structure. An extrinsic probe, N'-(iodoacetyl)-N'-(5-sulfol-naphthyl)ethylenediamine (IAEDANS) was attached to mutant proteins each of which bears a unique cysteine residue. Three mutants K39C (helix 2), T127C (between helices 6 and 7) and S16Crpt (helix 1, which bears a decapeptide repeat before the mutation) gave useful derivatives. In the soluble protein they showed emission wavelengths decreasing in the order K39C greater than T127C greater than S16Crpt and although all showed blue shifts on addition of dimyristoylphosphatidylglycerol (DMPG) this order was maintained in the membrane-bound state. These shifts were not indicative of deep membrane insertion. Polarization of IAEDANS revealed differences in mobility between mutants. The three tryptophan residues were used as a compound donor to IAEDANS in resonance energy transfer distance determinations. The values obtained for the soluble form were 1.2 A to 3.2 A longer than in the crystal structure. On addition of lipids the indicated distances increased: S16Crpt-I(AEDANS) 6.45 A (22%), K39C-I 5.45 A (18%) and T127C-I 2.4 A (14%). N-bromosuccinimide (NBS) completely abolishes the tryptophan emission from the thermolytic fragment. When lipids were added to a mixture containing ten NBS-treated channel-forming fragments to one IAEDANS labelled fragment the indicated distances increased rather more: S16Crpt-I 9.7 A (38%), K39C-I 8.1 A (36%) and T127C-I 2.5 A (16%). This showed that intermolecular transfer reduces the distance estimated in samples containing only labelled protein. The ensemble of results shows that the amphipathic helices of the C-terminal fragment open out on the surface of the lipid bilayer during the initial phase of membrane insertion.