Extent of cross-linking of amino-phospholipid neighbors in the erythrocyte membrane as influenced by the concentration of difluorodinitrobenzene

The degree of cross-linking of phospholipids to phospholipids and phospholipids to proteins in the erythrocyte membrane is dependent on the concentration of diflurodinitrobenzene. With ghosts isolated from human erythrocytes, the optimal extent of cross-linking of neighboring phosphatidylethanolamine molecules occurs at 50 μM difluorodinitrobenzene, the optimal extent of cross-linking of neighboring phosphatidylserine molecules occurs at 125 μM diflurodinitrobenzene and the optimal cross-linking of phosphatidylethanolamine to phosphatidylserine occurs at 75 μM difluorodinitrobenzene. Up to 37% of the total amino-phospholipids are cross-linked to membrane protein, the major part occurring with phosphatidylserine. Under optimal conditions of difluorodinitrobenzene concentration, 60% of the total phosphatidylethanolamine is cross-linked to phosphatidylethanolamine, 27% of the total phosphatidylethanolamine is cross-linked to phosphatidylserine, 24% of the total phosphatidylserine is cross-linked to phosphatidylserine and 44% of the total phosphatidylserine is cross-linked to phosphatidylethanolamine.     

Chemical synthesis of dinitrodiphenylsulfone derivatives of ethanolamine and serine and its application to the study of neighbor analysis of amino-phospholipids in the erythrocyte membrane

Summary The dinitrodiphenylsulfone derivatives of serine and ethanolamine have been prepared and their chromatographic and spectral properties are described. This cross-linking agent was used to determine the neighbor analysis of amino-phospholipids in the erythrocyte membrane. The results with erythrocyte ghosts show that at 50 m probe 31–50% of the total phosphatidylethanolamine is cross-linked to itself and 10–12% of the phosphatidylethanolamine is cross-linked to phosphatidylserine. Approximately 10–12% of the phosphatidylserine is cross-linked to itself and 16–20% of phosphatidylserine is cross-linked to phosphatidylethanolamine. The cross-linking of amino-phospholipids of ghosts with difluorodinitrodiphenylsulfone (9 Å span) is compared with cross-linking of these phospholipids by difluorodinitrobenzene (5 Å span). It is important to use the same sample of ghosts for this type of study since biological variability was seen in ghosts prepared from different batches of stored blood.     

Effect of cross-linking agents on insulin associated responses in adipocytes

The effect of 10 bifunctional cross-linking agents and four monofunctional analogues was studied on isolated adipocytes. [125I]Insulin binding and degradation, basal and insulin-stimulated glucose oxidation, and 3-O-methyl glucose uptake were measured. Two cross-linkers, which possess succinimide ester residues (disuccinimidyl suberate and dithiobis(succinimidyl propionate)) and react selectively with amino groups, appeared to react relatively specifically with the insulin receptor. Both produced a slight stimulation of basal glucose transport and metabolism, a marked inhibition of insulin-stimulated glucose transport and metabolism, and a marked decrease in insulin binding. Pretreatment of cells with unlabelled insulin partially blocked the effect of disuccinimidyl suberate, and as has been previously shown, disuccinimidyl suberate cross-linked insulin to its receptor. A monofunctional analogue of these compounds was 100-fold less active in altering cellular metabolic activity. Bisimidates, such as dimethyl suberimidate, dimethyl adipimidate, and dimethyl dithiobispropionimidate, also react with free amino groups but are more hydrophilic. These agents produced similar effects on glucose oxidation as the succinimide esters, but had little or no effect on insulin binding. The effects of these agents are not blocked by insulin and they do not cross-link insulin to its receptor. Mixed bifunctional reagents containing either a succinimide ester or an imidate and a group which reacts with thiols produced effects similar to the cross-linkers containing two succinimide groups or bisimidates, respectively. The bifunctional arylating agents difluorodinitrobenzene and bis(fluoronitrophenyl)sulfone produce marked effects on insulin binding and glucose oxidation at micromolar concentrations, but the monofunctional analogue fluorodinitrobenzene is almost equally active suggesting that with these compounds chemical modifications and not cross-linking was important. With neither the mixed bifunctional reagents, nor the arylating agents, did insulin pretreatment alter the effect of cross-linker and none of these agents cross-linked [125I]insulin to its receptor. These data suggest that the insulin receptor possesses a free amino group in a hydrophobic environment in its active site. A reactive amino group in a hydrophilic environment as well as other reactive groups are also present in some component of the insulin receptor-effector complex. Chemical modification or cross-linking of these functional groups results in an inhibition or mimicking of insulin action. Further study will be required to identify the exact locus of these sites.     

The interaction of 1-fluoro-2,4-dinitrobenzene with amino-phospholipids in membranes of intact erythrocytes,modified erythrocytes,and erythrocytes ghosts

Summary 1-Fluoro-2,4-dinitrobenzene (FDNB) has been used to study the availability of amino-containing phospholipids in erythrocyte membranes and ghosts in an aqueous, isotonic medium. It was found that the addition of bovine serum albumin (BSA) to the medium protects the cells from cation leak and protects some of the amino-phospholipids from reacting with the probe. In isotonic medium without BSA, 46% of the phosphatidylethanolamine and 12% of the phosphatidylserine of erythrocytes and 73% and 21% of these respective lipids of ghosts react with the probe. In the presence of 70 m BSA, 31% of phosphatidylethanolamine and 6.5% of phosphatidylserine of erythrocytes and 59% and 16% of these respective lipids of ghosts react with the probe. The labeling of these lipids does not change under conditions of varying tonicity, or after treatment of erythrocytes with pronase or lysolecithin. The data suggest that 46% of phosphatidylethanolamine and 12% of phosphatidylserine of the erythrocyte membrane are free in a lipid bilayer; 27% and 9% of these respective lipids are loosely bound to proteins which are lost during the preparation of ghosts and 27% of the phosphatidylethanolamine and 79% of the phosphatidylserine are tightly bound to core proteins which remain part of the erythrocyte membrane even after hemolysis.     

Properties of the strongly immobilized signal observed in spin-labeled erythrocytes

A strongly immobilized signal from fatty acid spin labels was observed in human erythrocytes treated with oxidizing agents such as glutaraldehyde, hydrogen peroxide, phenylhydrazine and copper-ortho-phenanthroline. This signal was also observed in freshly prepared ghosts treated with potassium superoxide and in old erythrocyte ghosts. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of these samples demonstrated the diffuse, nondiscrete bands of high molecular weight due to the cross-linking of membrane proteins. The temperature and pH dependences of the outer hyperfine splitting of this signal were very similar to those of bovine serum albumin. We propose that the strongly immobilized signal reflects the interaction of the lipids with the cross-linked products of membrane proteins.     

Properties of the strongly immobilized signal observed in spin-labeled erythrocytes

A strongly immobilized signal from fatty acid spin labels was observed in human erythrocytes treated with oxidizing agents such as glutaraldehyde, hydrogen peroxide, phenylhydrazine and copper-ortho-phenanthroline. This signal was also observed in freshly prepared ghosts treated with potassium superoxide and in old erythrocyte ghosts. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of these samples demonstrated the diffuse, nondiscrete bands of high molecular weight due to the cross-linking of membrane proteins. The temperature and pH dependences of the outer hyperfine splitting of this signal were very similar to those of bovine serum albumin. We propose that the strongly immobilized signal reflects the interaction of the lipids with the cross-linked products of membrane proteins.     

Investigation of the properties of bovine heart creatine kinase cross-linked with dimethyl suberimidate

Dimeric bovine heart creatine kinase (EC 2.7.3.2, ATP: creatine N-phosphotransferase) has been cross-linked with the bifunctional reagent dimethyl suberimidate at several concentrations to yield modified enzyme with enhanced stability towards heat denaturation. The degree of thermal stability is dependent on the degree of cross-linking with optimal stabilization occurring when approx. half of all the available amino groups are covalently attached to dimethyl suberimidate. Accelerated storage studies were performed and the results used to predict the storage time of the native and modified enzyme at lower temperatures. The cross-linked derivative was predicted to have a longer shelf-life at 4 degrees C than the native enzyme. Modification caused a reduction in the specific activity of the enzyme. The pH profile was altered following cross-linking, but the Michaelis constants were not changed. The modified enzyme exhibited a marked resistance to the action of some denaturing agents.     

The asymetric arrangement of phospholipids in the human erythrocyte membrane

In erythrocytes treated with 2,4,6-trinitrobenzenesulfonate (a non-penetrating probe) for 24 hours, a maximum of 33% of the phosphatidylethanolamine and none of the phosphatidylserine reacts with this reagent. In erythrocyte ghosts, however, 95% of the phosphatidylethanolamine and over 50% of the phosphatidylserine reacts in 90 minutes under the same conditions. When extracted erythrocyte lipids are treated with 2,4,6-trinitrobenzenesulfonate in either a chloroform-methanolbicarbonate or a sonicated aqueous bicarbonate system, both phosphatidylethanolamine and phosphatidylserine react essentially to completion within minutes. We interpret these results to indicate the localization of nearly all of the phosphatidylserine on the interior surface of the membrane thus demonstrating an asymmetric distribution of phospholipids in the erythrocyte membrane.     

Selective chemical modification of plasma membrane ectoenzymes

As part of an investigation of the organization of cell surface macromolecular assemblies, we have treated intact central nervous system cells with chemical probes which react convalently with proteins and aminophospholipids. Selective alterations of the enzymatic activities of ecto-ATPases, ecto-5'-nucleotidases and cholinesterases were obtained under appropriate reaction conditions. The cross-linking reagent, 1,5-difluoro-2,4-dinitrobenzene, was a potent inactivator of ecto-ATPase of C6 glioblastoma, IMR-32 neuroblastoma and of a primary rat astroblast cell line (RB). Ecto-5'-nucleotidase and acetylcholinesterase were less sensitive to difluorodinitrobenzene. 1-Fluoro-2,4-dinitrobenzene at concentrations which inactivated ecto-ATPase had little effect on ecto-5'-nucleotidase. Conversely, 2,4,6-trinitrobenzenesulfonic acid was a potent inactivator of ecto-5'-nucleotidase but had no effect on ecto-ATPase. The difluorodinitrobenzene inactivation of ecto-ATPase and of ecto-5'-nucleotidase as well as the fluorodinitrobenzene inactivation of ecto-ATPase could be prevented by the presence of the appropriate substrates in the reaction medium. In the presence of protecting nucleotide substrates, a decrease in reactivity with proteins and lipids was observed when the isotopic probe fluorodinitro[3H]-benzene was used.     

Arrangement of phosphatidylserine and phosphatidylethanolamine in the erythrocyte membrane

Cross-linking of phosphatidylethanolamine and phosphatidylserine in the erythrocyte membrane with the reagent difluorodinitrobenzene was studied as a function of temperature, time and concentration of difluorodinitrobenzene. The optimal extent of cross-linking of phosphatidylethanolamine to phosphatidylethanolamine, phosphatidylethanolamine to phosphatidylserine and phosphatidylserine to phosphatidylserine was expressed as molar ratios of these three different cross-linked species. The experimental results were compared to different models of a phospholipid monolayer containing phosphatidylethanolamine and phosphatidylserine in which phosphatidylserine was arranged primarily as singles (having 6 phosphatidylethanolamine neighbors) as clusters of dimers, trimer and tetramers or as large clusters. In the various model monolayers each lipid component has 6 neighbors. The models which are consistent with the experimental results are those in which phosphatidylserine and phosphatidylethanolamine occur as small clusters in a non-random array.     

Identification of suberimidate cross-linking sites of four histone sequences in H1-depleted chromatin. Histone arrangement in nucleosome core

The arrangement of 8 histones in the nucleosome core has been investigated by identifying the sites of 4 histone sequences cross-linked with a bifunctional amino-group reagent, dimethyl suberimidate, selected from among 4 diimidoesters of various linker lengths examined. H1-depleted calf thymus chromatin was allowed to react with 14C-labeled suberimidate at pH 8.5 and 0 degrees C. The cross-linked chromatin was then digested exhaustively with trypsin. Almost all the histone fragments were released from the chromatin with 0.25 M HCl and chromatographed on several columns and on paper. Cross-linked peptides were detected by analyzing the content of radioactive suberimidoylbislysine after acid hydrolysis. The chromatographic procedure developed here showed that the whole histone fragments contained 29 mol% of the total linked reagent as suberimidoylbisylsine. The 5 finally purified cross-linked peptides were identified from the total and N-terminal amino acids of each pair of peptides separated by two-dimensional cellulose thin layer chromatography after cutting the linker by ammonolysis. Thus, intramolecular cross-linking was found between Lys-5 and Lys-9 of H2A, and Lys-34 and Lys-85 of H2B, while intermolecular cross-linking was found between Lys-24 (or 27) of H2B and Lys-74 of H2A, Lys-85 of H2B and Lys-91 of H4, and Lys-120 of H2B and Lys-115 of H3 and/or Lys-77 of H4. Most of these lysine residues are located in the DNA-binding segments of the 4 histone sequences identified previously [Kato, Y. & Iwai, K, (1977) J. Biochem. 81, 621--630]. All the 5 or 6 cross-links can be located in a heterotypic tetramer consisting of one molecule each of H2A, H2B, H3, and H4, and a model of the histone arrangement in the tetramer is proposed. Two such tetramers may compose to the histone octamer in the nucleosome core.     

Cell-free synthesis of myosin by cardiac myofibrillar ribosomes

Human erythrocyte ghosts were treated with a bifunctional cross-linking reagent, dimethyl adipimidate dihydrochloride. On SDS-polyacrylamide electrophoresis of the cross-linked membrane proteins after solubilization, sialoglycoproteins and the proteins disappeared from the original band positions and appeared in a new band of aggregates.     

Identification of retrovirus matrix proteins by lipid-protein cross-linking.

Dimethyl suberimidate and its analogs are symmetrical bifunctional reagents that form amidine linkages with primary amino groups. These reagents have been used previously to study nearest neighbor relationships of proteins in viruses and in complex structures such as ribosomes. Dimethyl suberimidate also reacts with phosphatidylethanolamine, which can be radioactively labeled specifically with [¹⁴C]ethanolamine. When enveloped viruses containing radioactive phosphatidylethanolamine are exposed to the diimido ester, a fraction of the radioactivity becomes linked to viral structural proteins. Upon separation by sodium dodecyl sulfate/polyacrylamide gel electrophoresis, the lipid-protein complexes can be visualized on fluorograms of the gels. The cross-linking of lipids to proteins is specific, since it requires the viral structure to be intact, and since only certain proteins become chemically linked to phosphatidylethanolamine even though all the proteins react with dimethyl suberimidate. In vesicular stomatitis virus, the structure of which has been well characterized, only the glycoprotein and the matrix protein become linked to lipid. This is consistent with their known locations protruding outwards and inwards from the virus membrane, respectively. Thus we infer that the cross-linking technique can be used to identify proteins in close proximity to the lipid bilayer. In the avian leukemia and sarcoma viruses the protein designated p 19, and in the murine leukemia viruses p 15 become linked to radioactive lipid. Since avian p19 and murine p15 are internal structural proteins, we infer that they are equivalent to the matrix protein defined for other other enveloped viruses.     

Cross-linking of the enzymes in the glycosome of Trypanosoma brucei

Glycosomes, the microbody-like organelles containing mainly glycolytic enzymes, were purified from the long slender bloodstream form of Trypanosoma brucei EATRO 110 monomorphic strain by an improved method in which the protozoa were frozen and thawed in 15% glycerol to free, from the plasma membrane, much of the variant surface glycoprotein which used to constitute the major contaminant of our purified glycosomes. The purified glycosomes have 11 major proteins, 6 of which, tentatively identified as phosphofructose kinase, hexokinase, 3-phosphoglycerate kinase, aldolase, glyceraldehyde-3-phosphate dehydrogenase, and alpha-glycerophosphate dehydrogenase, constitute 87% of the total glycosomal protein. The bifunctional cross-linking reagents dimethyl suberimidate and dimethyl-3,3'-dithiobispropionimidate can penetrate the glycosomal membrane and cause extensive cross-linking of all the major glycosomal proteins. The cross-linked complex, insoluble in 0.1% Triton X-100 plus 0.15 M NaCl, contains all the glycosomal enzyme activities with only partial inactivations. All the enzymes are probably cross-linked into one large complex since they all sediment rapidly to the bottom of a 5-20% (v/v) sucrose density gradient. This successful cross-linking with reagents of span lengths of 11-12 A suggests close proximities among the glycosomal enzymes which may explain the extraordinarily high rate of glycolysis in T. brucei. Whether such a close association represents specific spatial arrangement required for genuine substrate channeling among the enzymes will be verified by future kinetic studies of the cross-linked enzyme complex.     

Effects of membrane lipids and -proteins and cytoskeletal proteins on the kinetics of cholesterol exchange between high density lipoprotein and human red blood cells, ghosts and microvesicles.

To better understand the effects of plasma membrane lipids and proteins and the cytoskeleton on the kinetics of cellular cholesterol efflux, the effects of (1), selectively depleting either sphingomyelin (SM) or phosphatidylcholine (PC); (2), cross-linking the cytoskeleton, and (3), removing certain cytoskeletal and integral membrane proteins on radiolabelled cholesterol efflux from red blood cells (RBC) have been studied. When RBC were treated with either phospholipase A2 or sphingomyelinase C to hydrolyze either 30-40% of the PC or 40-50% of the SM, respectively, the halftimes (t1/2) for cholesterol efflux to excess HDL3 were not significantly altered, with the values being 4.4 +/- 0.8 h or 3.7 +/- 0.4 h, respectively, compared to 4.6 +/- 0.6 h for control RBC. To investigate the effects of the cytoskeleton on the rate of free cholesterol (FC) desorption from the plasma membrane, the cytoskeletal proteins were cross-linked by either heat-treatment or exposure to diamide and cholesterol efflux from ghosts of these cells was measured. Cross-linking the cytoskeletal proteins by diamide treatment resulted in no significant change in t1/2 for treated (3.6 +/- 0.6 h) compared to control (4.2 +/- 0.4 h) ghosts: this suggests that the cytoskeleton does not play a large role in modulating cholesterol efflux. To investigate the effects of membrane proteins on cholesterol efflux, RBC microvesicles, containing mainly band 3 and 4 proteins and little of the cytoskeletal proteins, such as spectrin (bands 1,2) or actin (band 5), were obtained by incubation with the ionophore A23187. With excess HDL3 present, microvesicles exhibited a t1/2 of 4.2 +/- 1.9 h (compared to the t1/2 of 4.2 +/- 0.4 h for control ghosts). The results described in this paper suggest that neither changing the SM/PC ratio in the membrane nor cross-linking the cytoskeletal proteins nor removing the cytoskeleton changes the t1/2 for cholesterol efflux to excess HDL3. Presumably, the cholesterol-phospholipid interactions are insensitive to these perturbations in membrane structure.     

Cross-linking studies on the conformation and dimerization of myelin basic protein in solution.

Myelin basic protein was isolated from both cat and bovine central nervous system. Cat and bovine myelin basic protein, which are shown to be similar by tryptic mapping, exhibit identical behavior when cross-linked with the bifunctional reagent difluorodinitrobenzene. Myelin basic protein is cross-linked into only a dimer under certain conditions in the presence of sodium dodecyl sulfate. In contrast, many oligomers are formed when myelin basic protein is cross-linked in the absence of detergent. The formation of cross-linked dimers in the absence of other oligomer formation suggests that the protein is at least partly dimeric in the presence of sodium dodecyl sulfate. The conformation of them myelin basic protein monomer in sodium dodecyl sulfate was also studied. N-Bromosuccinimide and cyanogen bromide cleavage reactions were used to demonstrate that difluorodinitrobenzene had introduced intramolecular cross-links between the two peptides resulting from each of the cleavage ractions. However, these types of intramolecular cross-links cannot be detected under conditions in which only dimers have formed. Some of the lysine residues which are modified by difluorodinitrobenzene were identified by tryptic mapping. In several respects, the conformation of myelin basic protein in a sodium dodecyl sulfate solution appears to be similar to the conformation of the protein in the membrane.     

Determination of the native subunit pattern of gizzard tropomyosin using antibodies specific to each subunit

The gizzard tropomyosin molecule is composed of two subunits at 1:1 molar ratio. Possible composites of the tropomyosin molecule are two kinds of homodimer (one for each subunit), a heterodimer of two subunits, or a mixture of heterodimer and homodimer(s). We tried to evaluate the native subunit composition of gizzard tropomyosin by cross-linking experiments and immunological methods using specific antibodies to each subunit. For the cross-linking experiment we used dimethyl suberimidate, an amino group-specific cross-linker, in the presence of dithiothreitol to avoid artificial oxidative intersubunit cross-linking. When gizzard tropomyosin was cross-linked, it generated several products which might correspond to dimers formed by intersubunit cross-linkage. When the reaction was carried out for a long time, non-cross-linked subunits completely disappeared and two or three major cross-linked products arose. All of these cross-linked products were recognized by both of the specific antibodies to each subunit. These results indicated that the predominant part, if not all, of gizzard tropomyosin is present as heterodimer.     

Evaluation of Integrin Molecules Involved in Substrate Adhesion

Integrins were cross-linked to their extracellular matrix ligands using non-penetrating chemical cross-linkers. This procedure did not disturb the distribution of integrin in the adhesion structure and adhesion plaque integrin staining remained even when the cultures were extracted with ionic detergents. 80-90% of the pi integrin in the cross-linked culture was extracted with RIPA buffer and the remaining 10-20% was recovered following reversal of the cross-linking. This separated two distinct integrin pools, one which can be cross-linked to substrate bound extracellular matrix and one which is not. The specificity of this procedure for cross-linking of integrins involved in substrate adhesion was demonstrated using NIH 3T3 cells which express both α5β1 and α5β1 integrins. α6 was cross-linked only in cells plated on laminin whereas α5 was cross-linked when fibronectin was present. Using antisera directed to the cytoplasmic domains of either α5 or β1 integrin, it was demonstrated that these domains can be blocked in the intact cell but the blocking can be removed using ionic detergent extraction after chemical cross-linking. The extracellular matrix associated with the substrate surface but not that associated with the media exposed surface is both cross-linked and retained on the plastic dish following cross-linking.     

Cross-linking experiments with the adenosine triphosphatase of sarcoplasmic reticulum.

The proteins of sarcoplasmic reticulum were cross-linked by rapid oxidation of thiol groups with I2. About two-thirds of the thiols were oxidized without any significant cross-linking, implying an extensive formation of intramolecular disulphide bonds. When the thiols were completely oxidized at room temperature a series of oligomers containing up to five molecules were observed, as well as large aggregates which were excluded from the gels. Complete oxidation at -10 degrees C left most of the ATPase (adenosine triphosphatase) as monomer. Similar results were obtained when copper-phenanthroline complexes or dimethyl suberimidate were used as cross-linking reagents. We conclude that most of the cross-linked species arise by linking of randomly colliding ATPase molecules which are present in the membrane at very high concentration.     

Effect of pH on the cross-bridge arrangement in synthetic myosin filaments.

Synthetic thick filaments were cross-linked with dimethyl suberimidate at various pH values over the range pH 6.8---8.3. The rate of cross-linking myosin heads to the thick filament surface decreases significantly over a narrow pH range (7.4--8.0) despite the fact that the rate of the chemical reaction (amidination of lysine side chains) shows a positive pH dependence. The fall in rate cannot be ascribed to dissociation of the filament during the cross-linking reaction since the sedimentation boundary of the cross-linked filament (pH 8.3) remains unaltered in the presence of high salt (0.5 M). The decreased rate of cross-linking is also not caused by a shift in reactivity of a small number of highly reactive lysine groups, since the time course of cross-linking (pH 7.2) is unaffected by preincubation with a monofunctional imidate ester. Our results suggest that the heads of the myosin molecules move away from the thick filament surface at alkaline pH but are held close to the surface at neutral pH.