The lipid-protein interface in biological membranes.

A significant fraction of the lipid in many biological membranes is at the lipid-protein interface. The ESR and NMR data are in basic agreement that there is a dynamic equilibrium between lipid at the interface and the bulk bilayer. The lipid contact with the hydrophobic surfaces of the protein is spatially disordered compared to the bilayer lipids. The spatial disordering on the protein surface leads to the prediction that cooperative chain melting would not occur between lipid tails directly contacting the protein. This is in agreement with most, but not all of the DSC data. While there are some disagreements in the ESR studies, most of the quantitative data support the conclusion that the protein-associated lipid is motionally restricted under physiologically relevant conditions. In general, the NMR data are in agreement that exchange between boundary and bilayer regions is rapid on the NMR time scale at physiological temperatures, although there are some differences in interpretation of the lipid dynamics. From the available data, several kinds of lipid binding sites may be involved. Most of these sites are probably nonspecific, but with some additional sites exhibiting specificity for the chemical properties of the polar head group. The relative binding constants can vary within the boundary layer with several exchange rates applying. Although most of the exchange rates are rapid, perhaps more rapid than specific mechanistic steps in the enzyme reaction, there is a characterizable set of thermodynamic parameters for the boundary and bilayer equilibrium. Although many of the lipid binding sites may have very low relative binding constants, they must be higher than the binding constants for nonspecific protein-protein contacts. One probable function of the boundary is to act as a molecular spacer, preventing indiscriminate protein-protein aggregation in the two-dimensional lipid solvent. Other roles are suggested by the higher relative binding constants of some specific lipids.

Evidence that the zymogen of phospholipase A2 binds to a negatively charged lipid-water interface.

Evidence is presented that the zymogen of porcine pancreatic phospholipase A2 (prophospholipase A2) interacts with a lipid-water interface provided that the interface has a net negative surface charge. Fluorescence spectroscopy and non-equilibrium gel filtration indicate that binding of prophospholipase A2 (proPLA) to mixed detergent micelles is dependent on the presence of an anionic detergent. Prophospholipase binding is accompanied by a change in the environment of the single tryptophan residue qualitatively similar to that observed when the active enzyme, phospholipase A2 (PLA), binds to micelles. In addition, the rate of tryptic activation of prophospholipase is significantly reduced in the presence of negatively-charged mixed micelles, whereas no change in rate occurs when neutral micelles are present. These observations suggest that the lack of catalytic activity of the zymogen toward organized substrates carrying a negative surface charge cannot be explained by a failure to bind at the lipid-water interface.

Synthesis of charged amphipathic nitroxide lipid spin labels and an example of their application in membrane studies.

The synthesis of a series of amphipathic nitroxide lipid spin labels is reported. Thus, 12-proxylhexadecanol has been converted into the versatile fatty acid spin label 14-proxylstearic acid. This substance was used to prepare 14-proxylstearyltrimethylammonium methanesulfonate, a positively charged label, and 14-proxylstearylmethyl phosphate sodium salt, a negatively charged label. Also prepared in the doxyl series were quaternary ammonium salts derived from 16-doxyl- and 7-doxylstearic acid. The positively charged and negatively charged proxyl labels were used in a preliminary experiment to investigate the role of charge in their interaction with reconstituted cytochrome oxidase. The average binding affinity of the negatively charged label is approximately 2-fold higher than that of the positively charged label at pH 7.4. At pH 5.5 the average relative affinity for negatively charged label is about 3.5-fold higher than that of positively charged label, suggesting that the ionizable group(s) on the protein can interact with the lipid headgroup.

Lipid-lipid and lipid-protein interactions in membranes.

Over the past decade spectroscopic methods (fluorescence, ESR, and NMR) have been used to provide new information about the molecular dynamics of lipid-lipid and lipid-protein interactions in membranes. The various methods of characterizing isotropic and anisotropic motion are described. Lipid bilayers are highly dynamic, with rapid acyl chain motion and rapid lateral diffusion in the plane of the membrane. In membranes where proteins penetrate through the bilayer, a large hydrophobic surface area exists in contact with the bilayer lipids. Lipids at the protein interface are in dynamic equilibrium with the remaining pools of bilayer. The protein has been shown spectroscopically to have some influence on the dynamics of the nearest neighbor lipids, leaving the rest of the bilayer relatively unperturbed. Evidence is summarized that, in some cases, the lipid composition in the interfacial region is influenced by the protein.

The potential role of photoelectron microscopy in the analysis of biological surfaces.

The photoelectric effect provides the basis for an imaging technique useful for the study of biological surfaces. The photoelectron microscope (PEM) employs a UV lamp to photoeject electrons from the specimen surface. The electrons are then accelerated and imaged using electron optics. Photoelectron micrographs often resemble scanning electron micrographs, but the origin of contrast is different and these two techniques are complementary. Scanning Electron Microscopy (SEM) is unsurpassed in applications where specimens have pronounced relief or where elemental analysis is required. The advantages of PEM are a new origin of contrast, high sensitivity to fine topographical detail, short depth of information, and low specimen conductivity requirements. Photoelectron images of model systems, cell surfaces and cytoskeletal elements have been obtained.

Lipid binding to the amphipathic membrane protein cytochrome b5.

The lipid binding properties of the membrane protein cytochrome b(5) (detergent-extracted from calf liver microsomal preparations) were characterized by studying the interaction of spin-labeled lipids (5-, 12-, and 16-doxylstearic acid and 5- and 16-doxylphosphatidyl-choline, where doxyl refers to the nitroxide moiety) with cytochrome b(5), using electron spin resonance spectroscopy. The intact cytochrome b(5) molecule immobilizes all of the lipid spin labels, while the segment of cytochrome b(5) released by trypsin does not affect lipid mobility. The immobilization of lipid spin labels on the hydrophobic surface of intact cytochrome b(5) is not appreciably altered by associating the protein with liposomes. Differences in polarity of the lipid binding sites between cytochrome b(5) and phospholipid vesicles were also observed. The lipid binding sites on cytochrome b(5) are hydrophobic by conventional criteria, but are more polar than the interior of fluid phospholipid bilayers.

Phosphatidylcholine exchange between the boundary lipid and bilayer domains in cytochrome oxidase containing membranes.

A phospholipid spin label, 16-doxylphosphatidylcholine, is employed in a study of lipid--protein interactions in cytochrome oxidase containing membranes. Two methods are used to label the membranous cytochrome oxidase: dispersion in cholate with subsequent detergent removal, and fusion with vesicles of the pure phospholipid label in the absence of detergent. A fraction of the label is immobilized, which is calculated to fall in the range of 0.17--0.21 mg of phospholipid/mg of protein (0.15--0.19 after correction for lipids not extracted by chloroform--methanol). This narrow range of values is independent of methods of labeling, protein isolation, and lipid depletion within experimental error. When labeling by fusion is utilized, the patches of pure phosphatidylcholine spin label diffuse in the plane of the bilayer, become diluted, and demonstrate exchange with bound phospholipid. These observations are evidence that boundary lipid, as reflected by the partitioning of the phosphatidylcholine label, is in equilibrium with adjacent bilayer regions and that it consists of a relatively constant amount of phospholipid associated with the hydrophobic portion of the protein.

Lipid-protein interactions in cytochrome c oxidase. A comparison of covalently attached phospholipid photo-spin-label with label free to diffuse in the bilayer.

The aim of this study was to clarify the possible origins of the motion-restricted electron spin resonance (ESR) spectral component observed in membranes. For this purpose, a phospholipid photo-spin-label was synthesized, characterized, and used to study lipid-protein interactions in beef heart cytochrome c oxidase. The probe was designed with a nitroaryl azide incorporated in the phospholipid head group, and a spin-label on the sn-2 side chain, and was radiolabeled. The resulting molecule, 1-palmitoyl-2-(14-proxyl [2-3H]stearoyl)-sn-glycero-3-phospho-N-(4-azido-3-nitrophenyl)ethanolami ne, was stable under subdued light and during the procedures required to reconstitute cytochrome c oxidase in phospholipid bilayers. Upon photolysis, the photo-spin-label reacted with the protein in high yields (50% attached). There was no detectable destruction of the spin-label. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of cytochrome c oxidase after reaction with the photo-spin-label showed highest levels of attachment to bands I, III, and VII, with some labeling of other bands. The labeling pattern demonstrated a distribution of attachment sites, which was needed for the spin-labeling studies. ESR spectra of the attached labels at 25 degrees C indicated a constant fraction of motion-restricted lipid chains, independent of the lipid to protein ratio. In contrast, a spin-labeled phosphatidylcholine and the prephotolyzed photo-spin-label, both free to diffuse in the bilayer, exhibited behavior in agreement with the multiple equilibria binding model. These results, as well as data obtained with membranes frozen at -196 degrees C, show how several situations that lead to a motion-restricted ESR line shape can be distinguished. This study provides additional evidence that the fraction of lipids normally in contact with protein, and not aggregation artifacts, accounts for the observed motion-restricted component of ESR spectra of reconstituted cytochrome c oxidase in phospholipid bilayers.

Activation of porcine pancreatic phospholipase A2 by the presence of negative charges at the lipid-water interface.

The effect of surface charge on the porcine pancreatic phospholipase A2 catalyzed hydrolysis of organized substrates was examined through initial rate enzyme kinetic measurements. Two long-chain phospholipid substrates, phosphatidylglycerol (PG) and phosphatidylcholine (PC), were solubilized in seven detergents differing in polar head-group charge. The neutral or zwitterionic detergents selected were Triton X-100, Zwittergent 314, lauryl maltoside, hexadecylphosphocholine (C16PN), and 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate. The negatively and positively charged detergents used were cholate and CTAB, respectively. In general, the negatively charged phospholipid PG was hydrolyzed much more rapidly than the neutral (zwitterionic) phospholipid PC. The rate of hydrolysis of PG was rapid when solubilized in all the neutral detergents and in cholate but was essentially zero in the positively charged CTAB. Conversely, hydrolysis of PC was negligible when solubilized in neutral detergents, except C16PN, and was maximal in the negatively charged detergent, cholate. The rate of hydrolysis of PC solubilized in a neutral detergent became significant only when a negative surface charge was introduced by addition of SDS. Taken together, these kinetic measurements indicate that the surface charge on the lipid aggregates is an important factor in the rate of hydrolysis of phospholipid substrates and the highest activity is observed when the net surface charge is negative. Fluorescence and electron spin resonance (ESR) spectroscopic data provide additional support for this conclusion. The fluorescence emission spectrum of the single tryptophan of phospholipase A2 is a sensitive monitor of interfacial complex formation and shows that interaction of the protein with detergent micelles is strongly dependent on the presence of a negatively charged amphiphile.(ABSTRACT TRUNCATED AT 250 WORDS)

Detergent inactivation of sodium- and potassium-activated adenosinetriphosphatase of the electric eel.

The stability of the sodium- and potassium-activated adenosinetriphosphatase (Na,K-ATPase) of the electric eel, Electrophorus electricus, was studied in five detergents in an effort to establish conditions for reconstitution of this membrane protein into defined phospholipids. The Na,K-ATPase activity of purified electric organ membranes as well as the ATPase is stable for at least 1 month of storage at 0 degrees C in the absence of detergents. At low concentrations of detergents, the enzyme is also stable for several days, but irreversible inactivation occurs rapidly as the detergent concentration is further increased. This inactivation begins at well-defined threshold concentrations for each detergent, and these concentrations generally occur in the order of the detergent critical micelle concentrations. Increasing the concentration of the electric organ membranes causes a linear increase in the inactivation threshold concentrations of Lubrol WX, deoxycholate, and cholate. The onset of inactivation evidently occurs when the mole fraction of detergent associated with the membrane lipids reaches a critical value in the narrow range of 0.2-0.4, in contrast to the large differences in the bulk concentrations of these detergents. The eel Na,K-ATPase is more sensitive to detergents than the sheep kidney enzyme.

Orientation of lipid spin labels in lecithin multilayers.

Nitroxide-labeled stearic acid and cholestane have been incorporated into lecithin multilayers. The pronounced anisotropy of the preparations reflects a nonrandom structure of the lecithin film and demonstrates that these lipid probes can report structural information from lipid regions resembling those thought to occur in membranes. These two probes differ with respect to the orientation of the NO bond to the long axis, but both orient with their long axes perpendicular to the phospholipid film. Simulated spectra, calculated on the basis of a Gaussian distribution of orientations, are in agreement with the experimental results.

Evidence for boundary lipid in membranes.

Cytochrome oxidase (EC 1.9.3.1) isolated from beef-heart mitochondria with an appropriate phospholipid content forms vesicular structures. Lipid-protein interactions in this model membrane system were studied with the lipid spin label, 16-doxylstearic acid. As the phospholipid/protein ratio is varied, two spectral components are observed. At low phospholipid/protein ratios (

Competition between cholesterol and phosphatidylcholine for the hydrophobic surface of sarcoplasmic reticulum Ca2+-ATPase.

A multiple equilibrium binding model is used to examine phospholipid and cholesterol binding with the transmembranous protein Ca2+-ATPase (calcium pump). The protein was reconstituted in egg phosphatidylcholine bilayers by lipid substitution of rabbit muscle sarcoplasmic reticulum. Electron spin resonance spectra of a phosphatidylcholine spin-label and a recently developed cholesterol spin-label show two major spectral contributions, a motionally restricted component consistent with interactions between the label and the protein surface and another component characteristic of motion of the label in a fluid lipid bilayer. The number of lipid binding (or contact) sites at the hydrophobic surface of the protein is calculated to be N = 22 +/- 2. Experiments with intact sarcoplasmic reticulum membranes give approximately the same value for N. The relative binding constants are Kav approximately 1 for the phosphatidylcholine label and Kav approximately 0.65 for the cholesterol spin-label. Thus, cholesterol does contact the surface of the protein, but with a somewhat lower probability than phosphatidylcholine. This is confirmed by competition experiments where unlabeled cholesterol and the phospholipid spin-label are both present in the bilayer. Evidently the flexible acyl chains of the phospholipid molecules accommodate more readily to the irregular surface of the protein than does the rigid steroid structure of cholesterol.

Lipid environments in the yolk lipoprotein system. A spin-labeling study of the lipovitellin/phosvitin complex from Xenopus laevis.

Lipid/protein and lipid/lipid interactions in the yolk lipoprotein complex from Xenopus laevis were examined by introducing a series of lipid spin-labels into the complex and observing the electron spin resonance spectra as a function of the position of the label along the lipid chains, temperature, pH, and charge on the lipid polar head group. Analyses of the spectra show that, in addition to the expected component arising from lipid associated with protein, a second component with increased segmental flexibility and the greater temperature dependence characteristic of lipid/lipid interactions is observed. These spin-labeling data and supporting compositional data indicate that much of the lipid is organized into a lipid-rich region or pool, consistent with the earlier model derived from electron microscopy and diffraction data and with companion 31P and 2H nuclear magnetic resonance data reported in the preceding paper [Banaszak, L. J., & Seelig, J. (1982) Biochemistry (preceding paper in this issue)]. The bilayer-like component exhibits a greater restriction of motion compared to vesicles of the isolated lipids at the same temperature, as would be expected for a relatively small lipid pool. Phospholipids exchange between the two motionally distinguishable environments. The equilibrium binding undergoes a shift between these two environments as a function both of pH and of the charge on the phospholipid polar head group. This shift in average binding affinity is opposite in direction to that reported for membrane proteins and implicates negatively charged groups on the protein that repel negatively charged phospholipids. This effect is greatly reduced by alkaline phosphatase treatment, suggesting that some of the lipid binding sites are in close proximity to phosphorylated residues on the protein.

Charge selectivity at the lipid-protein interface of membranous Na,K-ATPase.

Lipid interactions with the integral membrane protein Na,K-ATPase (ATP phosphohydrolase, EC 3.6.1.3) purified from the electric organ of Electrophorus electricus were studied by spin labeling. A protein-associated component (boundary layer) in equilibrium with the fluid bilayer is clearly evident in the electron spin resonance spectra. The influence of charge on this equilibrium was determined by varying the head group of the lipid while maintaining the chain length and the position of the label constant. The lipid spin labels were 14-proxylstearylmethyl phosphate and the corresponding dimethylphosphate, alcohol, and quaternary amine. By using a pairwise spectral analysis, as well as a conventional spectral analysis, the binding affinity was found to decrease in the order of negative greater than neutral greater than positive charges. The fraction bound decreased from about 0.57 for the negatively charged phosphate to 0.25 for the positively charged quaternary amine. The amount of each bound lipid was nearly constant over the temperature range investigated (5-35 degrees C). High salt concentrations reversibly abolished the selectivity between the labels, confirming the role of charge in the binding equilibria.

Lipid--protein multiple binding equilibria in membranes.

Phospholipids at the lipid--protein interface of membrane proteins are in dynamic equilibrium with fluid bilayer. In order to express the number of binding sites (N) and the relative binding constants (K) in terms of measurable quantities, the equilibrium is formulated as an exchange reaction between lipid molecules competing for hydrophobic sites on the protein surface. Experimental data are reported on two integral membrane proteins, cytochrome oxidase and (Na,-K)-ATPase, reconstituted into defined phospholipids. Electron spin resonance measurements on reconstituted preparations of beef heart cytochrome oxidase in 1,2-dioleoyl-sn-3-phosphatidylcholine containing small quantities of the spin-labeled phospholipid 1-palmitoyl-2-(14-proxylstearoyl)-sn-3-phosphatidylcholine (PC*) gave a linear plot of bilayer/bound PC* vs. the lipid/protein ratio as predicted by the theory, with K congruent to 1 and N = 40 (normalized to heme aa3). This demonstrates that the spin-label moiety attached to the hydrocarbon chain does not significantly perturb the binding equilibria. In the second experimental system, (Na,K)-ATPase purified from rectal glands of Squalus acanthias was reconstituted with defined phosphatidylcholines as the lipid solvent and spin-labeled phospholipids with choline or serine head groups (PC*, PS*) as the solute. The (Na,K)-ATPase has a larger number of lipid binding or contact sites (N = 60-65 per alpha 2 beta 2 dimer) and exhibits a detectably larger average binding constant for the negatively charged phosphatidylserine than for the corresponding phosphatidylcholine. These results show that a multiple equilibria, noninteracting site binding treatment can account for the behavior of lipids exchanging between the protein surface and the lipid bilayer. Selective sites among a background of nonselective sites are experimentally detectable as a change in the measured relative binding constant.

Multiple equilibria binding treatment of lipid and detergent interactions with membrane proteins. Application to cytochrome c oxidase solubilized in cholate.

A modified multiple binding equilibria treatment is presented that allows determination of thermodynamic parameters of the interaction of phospholipids with integral membrane proteins solubilized in excess detergent. Lipid binding is modeled as a series of exchange reactions between lipid molecules and detergent molecules at the hydrophobic protein surface. A general equation is derived which expresses a relative association constant (K) and the total number of contact sites at the lipid-protein interface (N) in terms of experimentally measurable variables. A useful simplification of the general equation occurs when the amount of detergent is high relative to the total number of lipid binding sites in the sample. Computer simulations show that in cases we have examined there appears to be an experimentally accessible range of detergent to protein molar ratios where the approximation at high detergent is useful for analyzing experimental data. This model is used to examine the competition between cholate and spin-labeled phospholipids for the hydrophobic surfaces of bovine heart cytochrome c oxidase. We find, for example, that K = 12 +/- 2 for phosphatidylcholine relative to cholate (i.e., the cholate molecules are relatively easily displaced by membrane lipids). This helps to explain the experimental observation that cholate is an effective detergent both for solubilizing cytochrome c oxidase and for reconstituting this protein into a defined lipid bilayer environment. An excess of cholate readily displaces almost all of the native phospholipids, and the protein is dispersed in cholate micelles. However, when phospholipids are added back, the cholate molecules at the protein surface are replaced because of the higher relative binding of the phospholipids. Observed differences between the behavior of phosphatidylcholine and phosphatidylglycerol suggest that reconstitution in cholate is a selective process in which detergent molecules in localized areas on the protein surface are more readily displaced by certain phospholipids.

Identifying regions of membrane proteins in contact with phospholipid head groups: covalent attachment of a new class of aldehyde lipid labels to cytochrome c oxidase.

A series of amine-specific reagents based on the benzaldehyde reactive group have been synthesized, characterized, and used to study beef heart cytochrome c oxidase reconstituted in phospholipid bilayers. The series contained three classes of reagents: lipid-soluble phosphodiesters having a single hydrocarbon chain, phospholipid analogues, and a water-soluble benzaldehyde. All reagents were either radiolabeled or spin-labeled or both. The Schiff bases formed by these benzaldehydes with amines were found to be reversible until the addition of the reducing agent sodium cyanoborohydride, whereas attachment of lipid-derived aliphatic aldehydes was not readily reversible in the absence of the reducing agent. The benzaldehyde group provides a convenient method of controlling and delaying permanent attachment to integral membrane proteins until after the reconstitution steps. This ensures that the lipid analogues are located properly to identify amine groups at the lipid-protein interface rather than reacting indiscriminately with amines of the hydrophilic domains of the protein. The benzaldehyde lipid labels attach to cytochrome c oxidase with high efficiency. Typically, 20% of the amount of lipid label present was covalently attached to the protein, and the number of moles of label incorporated per mole of protein ranged from 1 to 6, depending on the molar ratios of label, lipid, and protein. The efficiency of labeling by the water-soluble benzaldehyde was much less than that observed for any of the lipid labels because of dilution effects, but equivalent levels of incorporation were achieved by increasing the label concentration. Electron spin resonance spectra of a nitroxide-containing phospholipid analogue covalently attached to reconstituted cytochrome c oxidase exhibited a large motion-restricted component, which is characteristic of spin-labeled lipids in contact with the hydrophobic surfaces of membrane proteins. The line shape and splittings were similar for covalently attached label and label free to diffuse and contact the protein molecules in the bilayer, providing independent evidence that the coupling occurs at the protein-lipid interface. The distribution of the benzaldehyde reagents attached to the polypeptide components of cytochrome c oxidase was examined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The labeling pattern observed for the lipid analogues was not affected by the presence of the nitroxide moiety on the acyl chains but was dependent on the molar ratio of labeling reagent to protein.(ABSTRACT TRUNCATED AT 400 WORDS)

Rotational dynamics of the single tryptophan of porcine pancreatic phospholipase A2, its zymogen, and an enzyme/micelle complex. A steady-state and time-resolved anisotropy study.

The rotational dynamics of the single tryptophan of porcine pancreatic phospholipase A2 and its zymogen (prophospholipase A2) have been studied by polarized fluorescence using steady-state and time-resolved single-photon counting techniques. The motion of Trp-3 in phospholipase A2 consists of a rapid subnanosecond wobble of the indole ring with an amplitude of about +/- 20 degrees accompanied by slower isotropic rotation of the entire protein. The rotational correlation times for overall particle rotational diffusion are consistent with conventional hydrodynamic theory. When phospholipase A2 binds to micelles of n-hexadecylphosphocholine, the amplitude of the fast ring rotation decreases. The whole particle rotational correlation time of the enzyme/micelle complex is smaller than the minimum value calculated from hydrodynamic theory. A similar result is obtained for the micelle itself by using the lipophilic probe transparinaric acid. These low values for the particle correlation times can be understood by postulating that an isotropic motion of the fluorophore in the small detergent particles contributes to the angular reorientation of the fluorophore. The internal reorientational motion of the tryptophan in the zymogen, prophospholipase A2, is of larger amplitude than that observed for the enzyme; specifically, the proenzyme exhibits a motion with a significant amplitude on the nanosecond time scale. This additional freedom of motion is attributed to segmental mobility of the N-terminal residues of prophospholipase A2. This demonstrates that this region of the protein is flexible in the zymogen but not in the processed enzyme. The implications of these findings for the mechanism of surface activation of phospholipase A2 are discussed by analogy with a trypsinogen-trypsin activation model.