Interaction of Membrane Lipids With Amphiphilic Molecules and Transmembrane Proteins

A. Lipid Order Parameter in the Presence of Amphiphilic Molecules
The outer lipid membrane surface of eukaryotic cells is generally uncharged. Amphiphilic, water-soluble molecules such as local anesthetics, viral or antibiotic peptides, or peptide toxins therefore partition into the bilayer interface because of their hydrophobicity. All these compounds are found to decrease the order of lipid membranes. This is illustrated in Fig. 8 which shows the effect of incorporating the cationic peptide fragment 828–848 from the carboxy-terminus of the envelope glycoprotein gp41 of HIV-1 (P828) into bilayers composed of 1-stearoyl(d35)- 2-oleoyl-sn- glycero-3-phosphoserine. A modest reduction of the lipid chain order near the glycerol backbone and a significant reduction towards the bilayer center are observed, indicating a decrease in the lateral packing density of the membrane and a corresponding increase of the cross-sectional area of the fatty acyl chains. An area expansion upon membrane penetration of amphiphilic compounds was also shown with molecular dynamics simulation for local anesthetics and peptides. The observation of an area increase upon insertion of local anesthetics is consistent with the phenomenon of pressure reversal of local anesthesia, which may be due to the anisotropic compression of lipid membranes under hydrostatic pressure and the consequent release of anesthetic molecules.>

B. Order and Fluidity in the Presence of Transmembrane Proteins
Hydrophobic transmembrane peptides aggregate in aqueous solution and therefore do not enter a lipid membrane spontaneously. In model membranes, peptide insertion is achieved by cosolubilization of peptide and lipid in an organic solvent (detergent solution) and subsequent evaporation of the solvent (equilibrium dialysis against detergent-free buffer). Reconstitution studies show that transmembrane peptides and proteins barely perturb the lipid bilayer order, suggesting a fluid-like match between the lipid acyl chains and the outer protein surface. The investigation of hydrophobic transmembrane peptides of different lengths has led to the conclusion that the average thickness of the lipid bilayer is significantly perturbed only in cases of a large mismatch between peptide length and membrane thickness. When the hydrophobic part of the peptide was larger (smaller) than that of the pure bilayer, the membrane thickness was increased (decreased).

Figure 8 Influence of peptide P828S on the hydrocarbon chain order of 1- stearoyld 35-2-oleoyl-sn- glycero-3- phosphoserine at 32°C. The smoothed order parameter profile derived from dePaked nuclear magnetic resonance powder patterns has lost the information characteristic for the beginning of the fatty acyl chains seen in Fig. 5. (A) > 2H NMR order parameter profiles of SOPS-d35 in the absence of P828s () and at lipid/peptide molar ratios of 20:1 () and 10:1 (), respectively. (B) The peptideinduced difference in order parameters along the chain at molar lipid/peptide ratios of 20:1 () and 10:1 (). Peptide-induced order changes are largest in the bilayer center, suggesting that the peptide acts as a spacer that is located in the membrane’s interface region. [From Smondyrev and Berkowitz (2000). Biophys. J. 78, 1672.]

Larger intrinsic membrane proteins may span the membrane with several helices and perform functional tasks that can be quantified by biochemical assays.Two different approaches have been employed to study the lipid–protein interaction. One is to purify and delipidate transmembrane proteins and to reconstitute them with selectively deuterated lipids; the other is to incorporate deuterated fatty acids or other deuterated substrates into biological membranes by means of the biosynthetic pathway. In the latter case, the intact biological membrane is compared with aqueous bilayer dispersions formed from the extracted lipids. In the following we will discuss examples for the two types of assays.

Cytochrome C oxidase catalyzes the transfer of electrons from cytochrome C to molecular oxygen and is one of the best investigated intrinsic membrane proteins. The beef-heart enzyme can be purified in an almost lipid-free form and can be functionally reconstituted by incorporation into different lipid systems since the natural lipid composition is usually not required for reconstitution of an active enzyme (see Fig. 9).

The interaction of cytochrome C oxidase with lipid membranes has been investigated by means of spin-label electron paramagnetic resonance (epr) and by ²H−, ¹⁴N−, and ³¹P−NMR experiments. The spin label method showed two motionally distinct lipid populations, with the slower component being attributed to the lipids interacting directly with the protein (“boundary lipids”). In contrast, NMR measurements of cytochrome C oxidase functionally reconstituted with headgroup and chain deuterated lipids revealed only one homogeneous population of lipids. The anisotropy of the segmental movements characterized by means of the residual ²H and ¹⁴N quadrupole splittings and the ³¹P chemical shielding anisotropy as well as the segmental fluctuations, determined by measuring the ²H− and ³¹P spin-lattice (T₁) relaxation times (Fig. 10), closely resemble those of pure lipid bilayers. Taken together, the anisotropy parameters as well as the T₁ relaxation times provide no evidence for any strong polar or hydrophobic interaction between the lipid and the protein, neither in terms of a conformational change of the headgroup nor in terms of a significant immobilization of individual segments. The only noticeable difference between the NMR spectra of reconstituted membranes and pure lipid bilayers was a line broadening in the presence of protein, which probably arises from slower motions.

Similar results were obtained in reconstitution experiments with lipophilin and proteolipid apoprotein-lecithin systems, sarcoplasmic reticulum Ca²⁺, Mg²⁺-ATPase, rhodopsin, and glycophorin. In all these cases deuterium NMR revealed only one lipid population while the epr spectra (as far as available) showed two components. The results further showthat proteins either disorder or have little effect on hydrocarbon chain order in membranes above the gel-to-liquid crystal phase transition temperature, T_c, of the pure lipids.

The question as to how an intrinsic protein affects the lipid environment was also investigated in systems containing a relatively low amount of lipid such as in partially delipidated cytochromeCoxidase surrounded by only 130 lipid molecules or in the crystalline lipovitellin/phosvitin complex containing about 100 phospholipid molecules in an interior cavity. In both systems the lipids remain in a fluid phase. Only when the lipid pool of cytochrome C oxidase was reduced to 6 to 18 molecules was a distinct broadening of the ²H-NMR linewidth observed, indicating a lipid motion which was no longer axially symmetric. But even under these conditions, the total width of the spectrum was still considerably narrower than that observed for immobilized phospholipids in solid crystals.

A second, much-debated question is whether or not cardiolipins form a long-lived complex with cytochrome C oxidase. To answer this question, the remaining lipids in partially (130 lipids per protein) and highly delipidated (“lipid-depleted”; 6 to 18 lipids per protein) cytochrome C oxidase were analyzed. In the partially delipidated preparation, approximately 11 cardiolipins, 54 phosphatidylethanolamines, and 64 phosphatidylcholines were found; in the “lipid-depleted” state, the corresponding numbers are 1 or 2 cardiolipins, 3 to 8 phosphatidylethanolamines, and 2 to 8 phosphatidylcholines. This result supports a fast exchange (>10⁴ s⁻¹) and is in contrast to earlier contentions that cardiolipin is the only remaining lipid in “lipid-depleted” cytochrome C oxidase. However, recent X-ray results showthat the residual lipids in cytochrome C oxidase crystals are also heterogeneous and may not even contain cardiolipin. The random distribution of the remaining lipids is in accordance with a fast exchange between lipids on and off the protein surface and suggests that cardiolipin (which may have a potential role in electron transfer reactions) is at best interacting transiently rather than permanently with cytochrome C oxidase.

The results obtained in reconstitution studies were confirmed with natural membranes. The natural systems investigated are, for example, Acholesplasma laidlawii (grown on a medium supplemented with specifically deuterated or perdeuterated fatty acids), cardiolipin- or glycerol-auxotroph Escherichia coli (grown in tissueculture medium containing selectively deuterated fatty acids or phosphatidyl glycerol), and mouse fibroblast L-M cells (grown in tissue-culture medium containing selectively deuterated choline or ethanolamine). The membranes of these systems showed very similar fatty acid and headgroup motion, ordering, and orientation as the membranes formed from the extracted lipids without protein. No long-lived lipid–protein complexes were observed for neutral or negatively charged