Analysis of the membrane-interacting domains of myelin basic protein by hydrophobic photolabeling.

Myelin basic protein is a water soluble membrane protein which interacts with acidic lipids through some type of hydrophobic interaction in addition to electrostatic interactions. Here we show that it can be labeled from within the lipid bilayer when bound to acidic lipids with the hydrophobic photolabel 3-(trifluoromethyl)-3-(m-[125I]iodophenyl)diazirine (TID) and by two lipid photolabels. The latter included one with the reactive group near the apolar/polar interface and one with the reactive group linked to an acyl chain to position it deeper in the bilayer. The regions of the protein which interact hydrophobically with lipid to the greatest extent were determined by cleaving the TID-labeled myelin basic protein (MBP) with cathepsin D into peptides 1-43, 44-89, and 90-170. All three peptides from lipid-bound protein were labeled much more than peptides from the protein labeled in solution. However, the peptide labeling pattern was similar for both environments. The two peptides in the N-terminal half were labeled similarly and about twice as much as the C-terminal peptide indicating that the N-terminal half interacts hydrophobically with lipid more than the C-terminal half. MBP can be modified post-translationally in vivo, including by deamidation, which may alter its interactions with lipid. However, deamidation had no effect on the TID labeling of MBP or on the labeling pattern of the cathepsin D peptides. The site of deamidation has been reported to be in the C-terminal half, and its lack of effect on hydrophobic interactions of MBP with lipid are consistent with the conclusion that the N-terminal half interacts hydrophobically more than the C-terminal half. Since other studies of the interaction of isolated N-terminal and C-terminal peptides with lipid also indicate that the N-terminal half interacts hydrophobically with lipid more than the C-terminal half, these results from photolabeling of the intact protein suggest that the N-terminal half of the intact protein interacts with lipid in a similar way as the isolated peptide. The similar behavior of the intact protein to that of its isolated peptides suggests that when the purified protein binds to acidic lipids, it is in a conformation which allows both halves of the protein to interact independently with the lipid bilayer. That is, it does not form a hydrophobic domain made up from different parts of the protein.

An effect of colchicine on galactosyl- and sialyltransferases of rat liver Golgi membranes.

Colchicine inhibited the activity of the galactosyl- and sialyltransferases of rat liver Golgi membranes. The sialyltransferase was more sensitive to the drug than galactosyltransferase since it was inhibited to a greater extent and at lower concentrations of colchicine than the galactosyltransferase. Two soluble enzymes, i.e. that from rat serum and that isolated from bovine milk, were not inhibited by colchicine. Even with very high concentrations of colchicine a marked stimulation of activity was observed. The data suggest that the inhibition observed in the Golgi membranes is in some way related to the arrangement of the enzymes in the lipid bilayer. In support of this hypothesis, the milk galactosyltransferase became very sensitive to colchicine after incorporation of the enzyme into lipid vesicles. The incorporation of colchicine into Golgi membranes was shown to decrease the order parameter as determined by electron spin resonance which reflects an increased fluidity of the Golgi membranes. A change in fluidity may be responsible for the inhibition of enzyme activity at least in part.

Adhesion of acidic lipid vesicles by 21.5 kDa (recombinant) and 18.5 kDa isoforms of myelin basic protein.

Myelin basic protein (MBP) is thought to be responsible for adhesion of the intracellular surfaces of compact myelin to give the major dense line. The 17 and 21.5 kDa isoforms containing exon II have been reported by others to localize to the cytoplasm and nucleus of murine oligodendrocytes and HeLa cells while the 14 and 18.5 kDa isoforms lacking exon II are confined to the plasma membrane. However, we show that the exon II(-) 18.5 kDa form and a recombinant exon II(+) 21.5 kDa isoform both caused similar aggregation of acidic lipid vesicles, indicating that they should have similar abilities to bind to the intracellular lipid surface of the plasma membrane and to cause adhesion of those surfaces to each other. The circular dichroism spectra of the two isoforms indicated that both had a similar secondary structure. Thus, both isoforms should be able to bind to and cause adhesion of the cytosolic surfaces of compact myelin. The fact that they do not could be due to differences in post-translational modification in vivo, trafficking through the cell and/or subcellular location of synthesis, but it is not due to differences in their lipid binding.

Similar effect of proteolipid apoproteins from human myelin (lipophilin) and bovine white matter on the lipid phase transition.

The proteolipid apoprotein from bovine white matter has been reported to increase the phase transition temperature of dimyristoyl phosphatidylcholine, in contrast to a proteolipid apoprotein fraction from human myelin, called lipophilin, which decreases the enthalpy without altering the phase transition temperature. Since these results lead to different conclusions concerning the structure and amount of boundary lipid surrounding these proteins, the effects of the two proteins on the phase transition of dimyristoyl phosphatidylcholine were compared. Neither protein has any effect on the phase transition temperature, regardless of the method of delipidation of the protein, the amount of residual lipid, the method of incorporation into vesicles, or heating rates used for differential scanning calorimetry. However, a higher melting component was observed when decomposition of the lipid to lysophosphatidylcholine had occurred. Addition of as little as 6% of the decomposition products of dimyristoyl phosphatidylcholine, lysodimyristoyl phosphatidylcholine and myristic acid, is enough to produce a higher-temperature peak. The intensity of this peak increases with increasing protein concentration similar to the reported result on the bovine white matter proteolipid. The question as to whether the protein-induced decrease in enthalpy is due to boundary lipid or entrapment of lipid in protein aggregates was also addressed by studying the appearance of the intramembranous protein particles by freeze-fracture electron microscopy at temperatures above and below the phase transition and between the premelt and main transitions. The protein is randomly dispersed above the phase transition. At low concentrations, below the phase transition, it clusters, forming particle-free and particle-rich areas, but does not aggregate. At higher concentrations it is randomly dispersed below the premelt and main transition but is clustered between the premelt and main transition. Since the protein is more randomly dispersed above the transition than below, the reduction in enthalpy of the freezing transition was compared to that of the melting transition and was found to be identical, suggesting that the withdrawal of lipid from the phase transition is probably not due to lipid entrapment but due to the formation of a boundary lipid interface between the protein and the bulk lipid.

Preparation and properties of vesicles of a purified myelin hydrophobic protein and phospholipid. A spin label study.

Lipophilin, a hydrophobic protein purified from the proteolipid of normal hupid and protein in 2-chloro-ethanol followed by dialysis against buffer. This method resulted in homogeneous incorporation of the protein into lipid vesicles as judged by sedimentation on a sucrose gradient and freeze fracture electreter and the freeze fracture faces contained intramembrane particles. The effect of lipophilin on the organization of the lipid was studied by use of spin label probes. Two distinct components were present in the spectrum of fatty acid spin labels in the lipid-protein vesicles. One was immobilized presumably due to the presence of boundary lipid around the protein and the second component waicles and probably due to a lamellar phase but with a slightly greater order parameter. Lipophilin was found to increase the order parameter linearly with increasing concentration of protein incorporated into the vesicles. However, the phase transition temperature as measured from the 2,2,6,6-tetramethyl piperidine-1-oxyl (TEMPO) solubility parameter was unchanged.

Characterization of anti-cerebroside sulfate antisera using a theoretical model to analyse liposome immune lysis data.

Antisera to the acidic glycolipid cerebroside sulfate (sulfogalactosyl ceramide) were raised in rabbits by several different methods. Reactivity with cerebroside sulfate was detected from complement-mediated lysis of liposomes composed of phosphatidylcholine/cholesterol/cerebroside sulfate and containing the spin label tempocholine chloride as a marker substance. Both cholesterol rich particles and lipid bilayer liposomes containing phosphatidylcholine and cholesterol were effective carriers for cerebroside sulfate, in combination with methylated bovine serum albumin for intravenous immunization, and with Freunds complete adjuvant for subcutaneous immunization. The antisera raised by the different methods were characterized with respect to their cross reactivity with other lipids and the relative concn of specific antibodies and their affinities for cerebroside sulfate using a theoretical model developed earlier [Vistnes A. I. (1984) J. Immun. Meth. 68, 251] for analysis of data from immune lysis of liposomes. Differences in these properties, both of which can affect antibody titer, could be detected for antisera raised by different methods and obtained at different times after immunization. Some of the antisera also reacted non-specifically to varying degrees with other anionic lipids indicating that the anti-cerebroside sulfate antibodies could bind non-specifically to anionic lipids by electrostatic interactions. This suggested that basic amino acid residues may be an important part of the antibody receptor binding site for the glycolipid head group. An important implication of this result is that antibodies raised against anionic glycolipids should be tested for non-specific binding to anionic phospholipids.

Suppression of experimental allergic encephalomyelitis by the encephalitogenic peptide, in solution or bound to liposomes.

We have investigated the ability of liposome-bound encephalitogenic peptide to suppress experimental allergic encephalomyelitis (EAE) in the guinea pig. EAE was induced by challenge with the encephalitogenic peptide, residues 113-122 of human myelin basic protein (MBP) in complete Freund's adjuvant. The peptide was acylated with stearic acid in order to anchor it to the lipid bilayer. The liposomal-bound peptide effectively suppressed clinical signs of EAE at relatively low doses, when given subcutaneously or intraperitoneally without incomplete Freund's adjuvant, several days after challenge. In vitro proliferation of lymphocytes from treated, protected animals in response to the peptide was greatly decreased but that to the purified protein derivative of tuberculin antigen was not, indicating an antigen-specific effect. However, histological signs of EAE were not reduced. The free peptide in solution was somewhat less effective when given intraperitoneally but was as or nearly as effective as liposome-bound peptide when given subcutaneously. Binding to liposomes may decrease the rate of clearance or degradation of the peptide when given intraperitoneally.

Antigen-targeted liposome-encapsulated methotrexate specifically kills lymphocytes sensitized to the nonapeptide of myelin basic protein.

Antigen targeting of liposome-encapsulated cytotoxic drugs to specific lymphocytes may be a useful approach for antigen-specific immunosuppressive treatment of autoimmune diseases in which a specific antigen is involved. The feasibility of utilizing this approach was investigated using experimental allergic encephalomyelitis as an animal model for an autoimmune response. The encephalitogenic determinant of myelin basic protein for the guinea pig is contained in residues 114-122, the so-called nonapeptide. We have acylated the nonapeptide at its N-terminal to anchor it in the lipid bilayer of liposomes containing the cytotoxic drug methotrexate. The nonapeptide on the surface of the liposomes then allows targeting of the liposomal methotrexate in vitro to anti-nonapeptide T lymphocytes obtained from guinea pigs with experimental allergic encephalomyelitis. Treatment with the nonapeptide-targeted liposomal methotrexate inhibited proliferation of anti-nonapeptide lymphocytes significantly more than that of control lymphocytes. These included non-sensitized lymphocytes, stimulated with phytohemagglutinin, or lymphocytes sensitized to different, unrelated proteins, the purified protein derivative of tuberculin and keyhole limpet hemocyanin, and stimulated with their specific antigens. Furthermore, nonapeptide-targeted liposomes had a greater cytotoxic effect on anti-nonapeptide T cells than untargeted liposomes. The results indicated that specific targeting to and killing of anti-nonapeptide cells was achieved, although improvements of the treatment are necessary before its use can be attempted in vivo.

Myelin basic protein: a multifunctional protein.

Myelin basic protein (MBP), the second most abundant protein in central nervous system myelin, is responsible for adhesion of the cytosolic surfaces of multilayered compact myelin. A member of the 'intrinsically disordered' or conformationally adaptable protein family, it also appears to have several other functions. It can interact with a number of polyanionic proteins including actin, tubulin, Ca(2+)-calmodulin, and clathrin, and negatively charged lipids, and acquires structure on binding to them. It may act as a membrane actin-binding protein, which might allow it to participate in transmission of extracellular signals to the cytoskeleton in oligodendrocytes and tight junctions in myelin. Some size isoforms of MBP are transported into the nucleus and thus they may also bind polynucleotides. Extracellular signals received by myelin or cultured oligodendrocytes cause changes in phosphorylation of MBP, suggesting that MBP is also involved in signaling. Further study of this very abundant protein will reveal how it is utilized by the oligodendrocyte and myelin for different purposes.

Exposure of galactosylceramide to galactose oxidase in liposomes: dependence on lipid environment and ceramide composition.

Factors which influence the accessibility, or exposure, of the carbohydrate head group of the glycolipid galactosylceramide (GalCer) at the membrane surface have been examined in lipid model membranes using the technique of galactose oxidase-tritiated sodium borohydride labeling. Both the ceramide composition of GalCer and the lipid composition of its membrane environment were varied. We have shown that GalCer is oxidized in a membrane environment, by purification of the labeled galactosyl moiety of the glycolipid by high-performance anion exchange chromatography. Using semisynthetic molecular species of GalCer with acyl chain lengths ranging from 16 to 26 carbons, incorporated into liposome membranes of egg phosphatidylcholine (PC), and reverse-phase HPLC separation of mixtures of the molecular species, we have shown that increasing the fatty acid chain length of GalCer increases its oxidation by galactose oxidase. In addition, the degree of oxidation is reduced when the fatty acid chain of GalCer is hydroxylated. GalCer incorporated into liposomes containing synthetic species of PC with different fatty acid chain lengths (together with cholesterol) was oxidized less as the PC acyl chain length, and hence the bilayer thickness, was increased. The oxidation of GalCer in liposomes composed of sphingomyelin/cholesterol was reduced compared to its oxidation in PC liposomes. Furthermore, changes in the fatty acid chain length of GalCer had no effect on its oxidation in sphingomyelin liposomes. These findings indicate that the ceramide composition and lipid membrane environment can influence the exposure of the lipid carbohydrate, and hence, they could modulate the receptor activity of glycolipids at the membrane surface.

Serum decreases permeability of some compounds from liposomes.

Although serum is generally regarded to increase the permeability of liposomes containing entrapped substances, we found that a low concentration of serum (10%) significantly reduced the permeability of liposomes to the spin label tempocholine chloride and the polar drug methotrexate, although it increased the permeability of the lipid-soluble drug actinomycin D. Liposomes containing sphingomyelin and cholesterol were considerably less permeable than liposomes containing phosphatidylcholine and cholesterol. Although a higher concentration of serum (88%) increased the permeability of liposomes containing either lipid, the amount of tempocholine which had leaked from sphingomyelin-containing liposomes in 88% serum after 50 h at 37 degrees C was only 25%, three times less than that from phosphatidylcholine-containing liposomes. Thus the effect of serum on liposome permeability depends on the compound entrapped as well as the type of lipid used.

A carbohydrate-carbohydrate interaction between galactosylceramide-containing liposomes and cerebroside sulfate-containing liposomes: dependence on the glycolipid ceramide composition.

Galactosylceramide (GalCer) and cerebroside sulfate (CBS) are the major glycolipids found in myelin. They occur in greater concentrations in this membrane than any other. Recently, it was reported that these two glycolipids can participate in a heterotypic carbohydrate-carbohydrate interaction [Hakomori et al. (1991) Glycoconjugate J. 8, 178]. In the present study, the effect of changes in the ceramide composition of both GalCer and CBS on this interaction has been examined. The interaction was monitored by measuring the aggregation of small unilamellar phosphatidylcholine/cholesterol liposomes containing GalCer with similar liposomes containing CBS, through the increase in optical density at 450 nm. Aggregation depends on the addition of a divalent cation and varies inversely with the ionic radius of the cation. Aggregation occurred at millimolar concentrations of divalent cation and was inhibited and reversed by the addition of EDTA. A lesser degree of homotypic self-aggregation of GalCer and of CBS liposomes also occurred in the presence of divalent cations, but the sum of this self-aggregation was significantly less than the heterotypic interaction between the two types of liposomes. Changes in the ceramide composition of GalCer and CBS significantly affected the extent of their interaction with each other. Increasing the fatty acid chain length of either GalCer or CBS resulted in increased aggregation. Hydroxylation of the fatty acid also increased the degree of aggregation of GalCer and CBS liposomes. These findings indicate that a divalent cation-mediated GalCer-CBS interaction could play a role in cell recognition and membrane adhesion phenomena such as the formation of compact multilamellar myelin.(ABSTRACT TRUNCATED AT 250 WORDS)

Aggregation of acidic lipid vesicles by myelin basic protein: dependence on potassium concentration.

In the compacted multilayered myelin sheath of the central nervous system, myelin basic protein (MBP) is thought to be responsible for adhesion of the intracellular surfaces by electrostatic interactions with acidic lipids. Noncompacted regions of myelin containing cytosol exist and can take up potassium released into the extracellular fluid after the axonal action potential. Therefore, the effect of K+ concentration on the ability of MBP to aggregate large unilamellar vesicles (LUVs) containing phosphatidylcholine (PC) and 10-20% acidic lipid was investigated. At MBP to lipid ratios where there was an excess of acidic lipid, physiological increases in K+ concentration up to about 100 mM greatly increased MBP-mediated aggregation of the LUVs by shielding the negative charge on the vesicle surface. Thus, changes in K+ concentration during the axonal action potential could regulate MBP-mediated adhesion of the intracellular myelin surfaces of noncompacted regions of myelin such as the paranodal loops. It could thus regulate the volume of these cytosolic regions, allowing MBP to have a dynamic function in myelin. Concentrations of K+ above 150 mM caused dissociation of MBP from LUVs containing PC and a single acidic lipid. LUVs containing the lipid composition estimated to be characteristic of the cytoplasmic leaflet of myelin (Cyt.-LUVs) were found to interact uniquely with MBP, resulting in greater aggregation, greater sensitivity to K+ concentration, and resistance to dissociation at high K+ concentrations. The latter suggested that electrostatic interactions were not the only force involved in binding of MBP to the Cyt.-LUVs. Hydrogen bonding of the protein to the lipid head groups and hydrophobic interactions due to penetration of hydrophobic amino acid side chains into the bilayer could also occur. The greater involvement of hydrophobic interactions of MBP with Cyt.-LUVs compared to PC/acidic lipid LUVs was confirmed from greater labeling of MBP bound to Cyt.-LUVs by the hydrophobic photolabeled TID. Cholesterol and phosphatidylethanolamine together were found to be responsible for the greater MBP-mediated aggregation of Cyt.-LUVs and the greater TID labeling of MBP bound to Cyt.-LUVs compared to PC/acidic lipid LUVs. Thus, the lipid composition of the intracellular surface of myelin is well suited to allow MBP to mediate adhesion of apposing intracellular membranes and to respond in a dynamic way in some regions of myelin, such as the paranodal loops, to changes in K+ concentration resulting from nerve conduction.

Effect of basic protein from human central nervous system myelin on lipid bilayer structure.

The effect of myelin basic protein from normal human central nervous system on lipid organization has been investigated by studying model membranes containing the protein by differential scanning calorimetry or electron spin resonance spectroscopy. Basic protein was found to decrease the phase transition temperature of dipalmitoyl phosphatidylglycerol, phosphatidic acid, and phosphatidylserine. The protein had a greater effect on the freezing temperature, measured from the cooling scan, than on the melting temperature, measured from the heating scan. These results are consistent with partial penetration of parts of the protein into the hydrocarbon region of the bilayer in the liquid crystalline state and partial freezing out when the lipid has been cooled below its phase transition temperature. The effect of the protein on fatty acid chain packing was investigated by using a series of fatty acid spin labels with the nitroxide group located at different positions along the chain. If the protein has not yet penetrated, it increases the order throughout the bilayer in the gel phase, probably by decreasing the repulsion between the lipid polar head groups. Above the phase transition temperature, when parts of it are able to pentrate, it decreases the motion of the lipid fatty acid chains greatly near the polar head group region, but has little or no effect near the interior of the bilayer. Upon cooling again the protein still decreases the motion near the polar head group region but increases it greatly in the interior. Thus, the protein penetrates partway into the bilayer, distorts the packing of the lipid fatty acid chains, and prevents recrystallization, thus decreasing the phase transition temperature. The magnitude of the effect varied with the lipid and was greatest for phosphatidic acid and phosphatidylglycerol. It could be reversed upon cooling for phosphatidylglycerol but not phosphatidic acid. The protein was only observed to decrease the phase transition temperature of phosphatidylserine upon cooling. It had only a small effect on phosphatidylethanolamine and no effect on phosphatidylcholine. Thus, the protein may penetrate to a different extent into different lipids even if it binds to the polar head group region by electrostatic interactions.

Pressure effect on the membrane action of a nerve-blocking spin label.

A reversible nerve-blocking spin label, 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) was used to study the nature of anesthetic-binding sites in nerve membranes as a function of pressure. The nerve-blocking effect of TEMPO is enhanced under pressure. At atmospheric pressure, TEMPO blocks nerve conduction by solubilizing in the apolar region of the nerve membrane. However, the nerve-conduction-block by TEMPO at 150 atm of helium was related to the binding of TEMPO to a pressure-induced high-affinity polar site in the nerve membrane. The new TEMPO-binding site could not be detected in lipid model membranes and, thus, the involvement of membrane protein in the new site was inferred. Pressure may induce a nerve membrane conformation change in the presence of TEMPO. The observation that under different pressure, a single anesthetic, i.e., TEMPO, was capable of blocking nerve conduction by binding to two different sites within the nerve membrane, supports the view that there are multiple anesthetic receptor sites, which differ in chemical composition and location within the nerve membrane. These sites, when occupied by different classes of anesthetics, produce the general phenomenon of nerve-conduction block. The enhancement of nerve-conduction block by pressure may be due to the increased concentration of TEMPO in the new site in the nerve membrane under pressure.

Specific targeting of phototoxic haptenated liposomes to a hapten-specific B cell lymphoma.

A method is reported to eliminate B lymphocytes specific for a haptenated lipid by using the lipid hapten to target a photosensitive drug to them. The photosensitizer eosin was coupled to a phospholipid and incorporated into trinitrophenol (TNP)-bearing small unilamellar vesicles of egg phosphatidylcholine (PC) and cholesterol in order to target the photosensitizer to B lymphoma cells (A20-HL) with TNP-specific membrane IgM receptors in vitro. Exposure of the treated cells to visible light led to an antigen-specific toxic effect indicated by inhibition of cell proliferation. A significantly higher concentration of liposomal eosin was required to inhibit control B cells. These were genetically identical B lymphoma cells (A20-2J) which lack only the DNA for the surface antigen receptor. Furthermore, pretreatment with TNP-conjugated keyhole limpet hemocyanin or anti-IgM antibody abolished the antigen-specific toxic effect, confirming that the TNP-targeted liposomal eosin mediates its effect by binding to the Ig antigen receptors on TNP-specific B cells. Incubation of cells with the TNP-bearing phototoxic liposomes at 4 degrees C instead of 37 degrees C did not alter the antigen-specific targeting effect, suggesting that uptake of the liposomal drug into the cells is not necessary for its toxic effect. Replacement of the liposomal phospholipid (egg PC) with saturated species of PC having higher phase transition temperatures or with sphingomyelin caused a decrease of the antigen-specific effect. These results demonstrate the potential use of antigen-bearing liposomal phototoxic drugs for the purpose of targeting and eliminating B cells with antigen-specific surface Ig receptors.

Dependence of the surface expression of the glycolipid cerebroside sulfate on its lipid environment: comparison of sphingomyelin and phosphatidylcholine.

The influence of the membrane lipid environment on the reactivity with antibody of the acidic glycolipid cerebroside sulfate (CBS) was examined by using a spin membrane immunoassay. Fewer antibodies in a polyclonal anti-CBS antiserum recognized the antigen in a bovine brain sphingomyelin/cholesterol (SM/CHOL) environment than in dipalmitoylphosphatidylcholine/cholesterol (DPPC/CHOL). Changes in the CBS ceramide group appeared to have less influence on antibody recognition of CBS in SM/CHOL than in DPPC/CHOL [Crook et al. (1986) Biochemistry 25, 7488-7494]. Although the fatty acid chain length of phosphatidylcholine strongly influences CBS recognition, the fatty acid chain length of sphingomyelin had only a moderate effect on CBS recognition and did not account for the decreased recognition in SM compared to in DPPC. Inhibition studies revealed that the antibodies which recognize CBS in SM/CHOL (S antibodies) form a population distinct from those which recognize CBS in DPPC/CHOL (P antibodies). The specificity of the P and S antibodies was examined further by comparing the efficacy of various substances, which share chemical features with the components of CBS in a SM/CHOL or DPPC/CHOL environment, to inhibit lysis of liposomes containing CBS. Intact CBS, cholesterol, and a phosphocholine lipid, at certain antigen densities, were required for optimal recognition of the antigen, especially by the P antibodies, suggesting that a complex of all three lipids in a multivalent array may be recognized by these antibodies. The S antibodies may recognize a smaller complex or monomers of CBS.(ABSTRACT TRUNCATED AT 250 WORDS)

Dependence of boundary lipid on fatty acid chain length in phosphatidylcholine vesicles containing a hydrophobic protein from myelin proteolipid.

Lipophilin, a hydrophobic protein fraction, purified and delipidated from the proteolipid of human myelin, possesses a layer of boundary lipid surrounding it when incorporated into lipid vesicles. The protein reduces the energy absorbed during the lipid phase transition, indicating that the boundary lipid does not go through the phase transition. The amount of boundary lipid was estimated by plotting the enthalpy of the transition against the protein to lipid mole ratio and extrapolating to deltaH = 0 for a number of synthetic phosphatidylcholines, to determine the ability of fatty acid chains of varying length to interact with the protein. The amount of boundary lipid was found to be similar, 21-25 molecules per molecule of lipophilin, for fatty acid chains of length 14-18 carbons but somewhat less, 16 molecules of lipid per molecule of protein, for a fatty acid chain length of 12 or for one with a trans double bond (18:1tr). No preferential interaction was observed with a lipid containing a particular fatty acid chain length when the protein was incorporated into a mixture of these lipids. These results suggest that the binding of lipids to the boundary layer of other membrane proteins and enzymes may not depend significantly on lipid fatty acid chain length.