CHAPTER 1
Biophysics of the membrane interface and its involvement in protein targeting and translocation
A. Watts and T.J.T. Pinheiro
Department of Biochemistry, South Parks Road, University of Oxford, Oxford OX| 3QU, U.K.
Introduction
The initial site of association for any component that may partition into and then, as one possibility, traverse a membrane is the polar/apolar interface of the membrane. Whether or not a protein or lipid acts as the target site, such associations are driven initially, and possibly subsequently, by electrostatic forces. These forces are important not only in ionic interactions and conductance effects, but also in determining the structure and activity of membrane proteins, including protein insertion and translocation. Here, the relevant thermodynamic and electrostatic aspects of membrane protein association and insertion will be reviewed. As specific examples of such associations, the mode of interaction and kinetics of association of several peptides and proteins will be presented using a range of biophysical approaches, although it should be stated that the area is highly complex, and no simple explanations exist for any systems, and much information is piecemeal and incomplete. This short resume cannot hope to include every aspect of the topic, but some indication of contemporary methods and results will be given.
Thermodynamic and practical consideration of protein–membrane associations
When considering membrane protein insertion and translocation, the thermodynamics of the interactions are important, driven, as they are initially, by electrical forces. The free energy of peptide binding to a membrane can be approximated from:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where ΔGel denotes the electrostatic effects (see below), ΔGimm is the positive binding energy caused by peptide immobilization, ΔGfob is the energy gained from the hydrophobic effects, ΔGpol is the energy contribution from backbone and side-chain hydrogen bonding, and ΔGlip describes the lipid perturbation effects. Any configurational energy associated with peptide-membrane binding is included within the ΔGpol term.
Biophysical methods are able to allow some estimates of these various terms, although too little is known at present to permit complete descriptions of the association mechanisms. The value of ΔGimm for the peptide on the membrane surface, when compared with isotropically moving peptide, is not really known. However, when compared with the more favourable disordered unfolded helixforming peptide, the folded and thus more ordered helix has an unfavourable energy of ΔGimm ~5.23 kj · mol-1 (~1.25 kcal · mol-1) per peptide bond, giving approx. 84–126 kj · mol-1 (20-30 kcal · mol-1) for a 20-amino-acid helix. Against this, a similar transmembrane helix would form 16 hydrogen bonds in the non-aqueous environment of the bilayer core, giving a ΔGpol of approx.—402 kj · mol-1 (-96 kcal · mol-1), with only small contributions from van der Waals interactions between lipids and side chains. Estimates of ΔGimm + ΔGfob of at least 84 kj · mol-1 (20 kcal · mol-1) have been made for the desolvation/hydration of signal peptides. However, it has been pointed out that the 'macroscopic' and 'microscopic' hydrophobic effects in membranes and protein binding sites are very different in magnitude and that highly curved surfaces can produce an anomalous, high, hydrophobic energy of binding. Furthermore, for peptide insertion and folding, and again on energetic grounds, the insertion of an unfolded chain is extremely unfavourable [ΔGimm +176 kj · mol-1 (+42 kcal · mol-1)] when compared with insertion of a folded chain [ΔGimm -126 kj · mol-1 (-30 kcal · mol-1)] (20–22 residues). From similar arguments, assembled ß-structures are also unfavourably inserted into a bilayer. Based on such considerations, it has been argued that a polypeptide coil cannot be inserted into a bilayer and then fold, but rather a secondary structure must be formed either in the aqueous phase or, at the latest, at the membrane/water interface before insertion.
The maximum energy from inserting the hydrophobic residues of the LamB wild-type peptide into the bilayer (ΔGfob) is approx. -381 kj · mol-1 (-91 kcal · mol-1), which is very substantial and favourable. Also, the free energy required for transfer of unbonded polar groups is much higher [OH, +16.7 kj · mol-1 (+4.0 kcal · mol-1); -NH2, +20.9 kj · mol-1 (+ 5.0 kcal · mol-1); COOH, +20.1 kj · mol-1 (+4.8 kcal · mol-1); C = O, +8.4 kj · mol-1 (+2.0 kcal · mol-1)] than for bonded -NHO = C pairs [+2.3 kj · mol-1 (+0.55 kcal · mol-1)].
Lipid perturbation effects play an insignificant role in the energetics of peptide insertion and translocation. Ordering of lipids around the peptide may be significant, but this is offset by the favourable interaction of acyl chains with hydrophobic peptide residues. Partially embedded peptides, or those on the bilayer surface, cannot be assessed so readily and lipid phase separation, cavity formation and lipid head-group perturbations are all relatively poorly described. However, as a general contribution to the entropic changes upon peptide binding, the restriction of a peptide on the bilayer surface, when compared with the isotropically moving peptide, may be countered by the disorder of the induced lipid perturbation effects, resulting in a balance (no gain or loss) of these two contributions to the overall energetics of the interaction.
Other factors of importance in the energetics of transmembrane helix assembly are the stability of helix–helix interactions which overcome the entropies of helix separation [in the range 4.2–41.9 kj · mol-1 (1-10 kcal · mol-1)], especially when helix packing is better accommodated in the bilayer than lipid–helix packing lipids.
The major (~70%) contribution to the binding energy of signal peptides, from the studies reported to date, is therefore the hydrophobic (ΔGfob) interaction, although theory and experiment are still not within range of each other, and too little is known at present about the mechanism of peptide and protein interactions with biomernbranes to be able to describe the energetics of the process in detail.
Electrostatic contributions at the membrane interface
The electrostatic interactions between a peptide or protein and a membrane interface are likely to be more important in the initial events of peptide associations than in the later stabilization of the peptide–membrane complex. Such interactions are known to involve positively charged amino acid side chains and negative charges in lipid head-groups at the membrane interface. However, the dynamics and contributions of these multi-phase and multi-step interactions are not well understood. In particular, the way in which electrostatic interactions can induce gross conformational distortions of the peptides or proteins during contact with the lipid interface is only now being explored and recognized. These electrostatic contributions are significant and are the result of many complex factors, not least of these being the localization of very different chemical groups (carbonyls, methyls, methylene, phosphates, cholines, carboxvlic acids, primary amino groups, etc.) of the lipids within the structured bilayer (Figure 1).
The electrical profile of a lipid bilayer is a complicated sum of multiple potentials (Figure 1). Although the transmembrane potential (ΔΨ) and surface potentials (Ψso; Ψsi) can be reasonably well described, the internal membrane potentials (Ψo; Ψi) can be quite varied, being made up predominantly of the membrane dipole potential and adsorption potentials. Typical values for the transmembrane potential are 10-100 mV, and in biological membranes the cytoplasmic side is negative relative to the outside of the membrane. The transmembrane potential can readily be calculated for equilibrium and steady-state conditions (Nernst and Goldman–Hodgkin–Katz equations) and is, itself, responsible for driving translocation of peptides across membranes.
The surface potentials (Ψso; Ψsi) are the electrostatic potential at the membrane/aqueous interface relative to that in the corresponding bulk phase. In all biological systems, this potential is negative due to the existence of anionic phospholipids such as phosphatidylserine, but also cardiolipin and phosphatidylglycerol (PG), with a very small amount (<1%, w/w) of phosphatidic acid as a turnover product. Biomembranes typically have 10–20% of their area covered by negatively charged lipids, which is 1 electronic charge per 100 nm2; this gives, together with protein charges, surface potentials (Ψso; Ψsi) of between -8 and -30 mV. Direct measurements of the changes in surface potential can be made using the highly sensitive, non-perturbing deuterium NMR approach, as done for peptides, proteins and ions.
Internal potentials (Ψi; Ψo) arise from either dipoles or charges in the low dielectric interior of the membrane. There are at least two contributions that give rise to these potentials. First, the polar molecules at the membrane/water interface give rise to a substantial (several hundred millivolts) membrane dipole surface potential (ΔΨiP = Ψso - Ψsi) due to their organization, with positive inside, and hence make a significant contribution to the total transmembrane potential. Secondly, any transfer of charge into the bilayer, such as ions, peptides or channel-gating charges, also contribute to this internal adsorption potential (ΔΨip = Ψs - Ψi), because their effects are not screened by changes in the ionic strength of the bulk medium. Thus, in peptide and protein adsorption and translocation, it is the intramembrane potential (ΔΨip) (Figure 1) that is modified substantially by such events, and not simply the total transmembrane potential, in which the internal potential is then directly reflected. Importantly, gross inhomogeneous structural changes in bilayer lipids and associated proteins, such as phase separation and concomitant boundary mismatches between phase-separated regions, can occur when proteins and peptides interact with a membrane, and it is not easy to see how these can be accounted for in generalized, thermodynamic models of biomembrane electrostatics and energetics.
The free energy of attraction of counter-ions to a charged surface on which the charges are smeared uniformly, and the statistical tendency of the counter-ion to diffuse away from the region of high concentration, is given by the Gouy–Chapman relationship, which holds in many homogeneous situations. However, if the charges are fixed (discrete charges), or the charges are far apart (further than the Debye length, ~1 nm), the Gouy–Chapman theory breaks down, as it does for multivalent charged species such as peptides and proteins. In particular, when charges are buried in the low dielectric interior of the membrane, even though their density may be low, they produce a large contribution to the intramembrane potential (Figure 1). Thus, for proteins and peptides that associate with a membrane surface and possibly translocate across the bilayer, discrete charge effects need to be considered, as does the possibility that the lipids may or may not be uniformly smeared out across the whole membrane; without detailed knowledge of lateral phase separation of lipids induced by polyvalent species, this cannot be comprehensively addressed.
No large protein can be considered as a point charge, and only charges that are within the Debye length (~1 nm) of the membrane interface will experience any significant potential. Thus how membrane interfacial electrostatic forces attract peptides or proteins, and affect the orientation of an approaching peptide or protein with respect to a membrane surface, is not known, although the high fields (105 V · cm-1) near the surface of a biomembrane could clearly induce structural changes in the protein or peptide. It is normally assumed, and has been confirmed experimentally, that one charged lipid associates with one charged protein residue. More recent evidence from studies on polybasic peptides also confirms this observation.
Creation of mismatch zones in membranes as potential translocation or insertion sites
Peripherally bound peptides and proteins are known to induce lateral phase separation in model membranes, a feature that is not accommodated in the classical Singer–Nicholson model of the biomembrane, Such a phenomenon is much more difficult to demonstrate in biological membranes due to their heterogeneity, but it is generally agreed that it does occur, as suggested and schematically represented (Figure 2) by Vaz.
To investigate the effect of the electrostatically induced lipid lateral phase separation of a charged peptide, we have used a non-perturbing method to probe the surface properties of anionic lipid bilayers. Polymyxin-B is a cyclic peptide antibiotic specific for Escherichia coli, and it is thought that instability induced between laterally phase-separated regions of the bilayer causes cell lysis. Deuterons, placed in the lipid head-groups of the bilayers [on either PG or phosphatidylcholine (PC) lipids, in separate experiments], show that, when the peptide binds, PG molecules are extracted from the bilayer, in which PG/PC molecules are ideally mixed, into new peptide-PG complexes. The lifetimes of these complexes are longer than 10-3 s, since the deuterium NMR spectra are not averaged by rapid diffusion of the lipids into and out of the phase-separated regions, although this does, of course, happen at the molecular level. No estimate of the size of the laterally separated areas could be made, and they may contain just one peptide, or an aggregate of peptides. The regions away from the complexes are essentially free of PG molecules, with charges being titrated between the peptide and PG molecules.
Interaction of basic proteins with anionic lipids
To assess the mechanisms of protein–lipid interactions at the bilayer surface, the interaction of the peripheral spinal cord myelin basic protein (18400 Da) has been studied, again using solid-state deuterium NMR, Here, the number of accessible charges (32, mainly lysines) on the protein were titrated stoichiometrically with PG lipids. No interaction was observed with PC bilayers, implying that the hydrophobic contribution to the interaction was relatively small. Each charged residue was satisfied with one PG head-group, and the exchange of lipids into the protein–lipid complex was fast, implying that the lifetime of the complex was less than 10-3 s.
