Binding of high density lipoprotein to cultured fibroblasts after chemical alteration of apoprotein amino acid residues.

Cultured extrahepatic cells possess a specific high affinity binding site (receptor) for high density lipoprotein (HDL) that is induced by cholesterol delivery to cells. To characterize the binding recognition site(s) on HDL, the ability of HDL to interact with cultured human fibroblasts was assayed after chemical alteration of specific apoprotein amino acid residues. Reduction and alkylation, acetylation, and cyclohexanedione treatment of HDL3 had little or no effect on its cellular binding. Treatment of HDL3 with tetranitromethane (TNM), however, caused a large dose-dependent decrease in binding, with maximum inhibition at 3 mM. Amino acid analysis of the TNM-treated particles showed specific alteration of tyrosine residues, but sodium dodecyl sulfate-gel electrophoresis demonstrated apoprotein cross-linking coincident with decreased binding. These results suggest that modification of HDL tyrosine residues and/or cross-linking of HDL apoproteins alters the ligand site recognized by the HDL receptor. Gradient gel electrophoresis, molecular sieve chromatography, and electron microscopy showed only minor changes in size distribution and shape of HDL3 particles after treatment with 3 mM TNM, but at higher TNM concentrations, coalescence and aggregation of particles was evident. Treatment of HDL3 with 3 mM TNM affected neither its promotion of the low affinity (receptor-independent) cholesterol efflux from cells nor its ability to accept cholesterol from an albumin suspension, yet promotion of high affinity (receptor-dependent) cholesterol efflux from cells was abolished. The finding that TNM treatment of HDL3 decreases both its receptor binding and its promotion of cholesterol efflux from cells without substantial alteration of its physical properties supports the hypothesis that the HDL receptor functions to facilitate cholesterol transport from cells.

Binding of High Density Lipoprotein to Cultured Fibroblasts after Chemical Alteration of Apoprotein Amino Acid Residues* (Received for publication, June 5, 1985) Eliot A. Brinton Cultured extrahepat.ic cells possess a specific high affinity binding site (receptor) for high density lipoprotein (HDL) that is induced by cholesterol delivery to cells. To characterize the binding recognition site(s) on HDL, the ability of HDL to interact with cultured human fibroblasts was assayed after chemical alteration of specific apoprotein amino acid residues. Reduction and alkylation, acetylation, and cyclohexanedione treatment of HDL3 had little or no effect on its cellular binding. Treatment of HDLs with tetranitromethane (TNM), however, caused a large dose-dependent decrease in binding, with maximum inhibition at 3 mM. Amino acid analysis of the TNM-treated particles showed specific alteration of tyrosine residues, but sodium dodecyl sulfate-gel electrophoresis demonstrated apoprotein cross-linking coincident with decreased binding. These results suggest that modification of HDL tyrosine residues and/or cross-linking of HDL apoproteins alters the ligand site recognized by the HDL receptor. Gradient gel electrophoresis, molecular sieve chromatography, and electron microscopy showed only minor changes in size distribution and shape of HDL3 particles after treatment with 3 m M TNM, but at higher TNM concentrations, coalescence and aggregation of particles was evident. Treatment of HDL3 with 3 mM TNM affected neither its promotion of the low affinity (receptor-independent) cholesterol efflux from cells nor its ability to accept cholesterol from an albumin suspension, yet promotion of high affinity (receptor-dependent) cholesterol efflux from cells was abolished. The finding that TNM treatment of HDL3 decreases both its receptor binding and its promotion of cholesterol efflux from cells without substantial alteration of its physical properties supports the hypothesis that the HDL receptor functions to facilitate cholesterol transport from cells.
Population studies have shown an inverse correlation between plasma HDL' levels and the incidence and prevalence . The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: HDL, high density lipoprotein; LDL, low density lipoprotein; apo, apoprotein; TNM, tetranitromethane; PBS, phosphate-buffered saline; SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis; 1, liter; Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid. of atherosclerosis (1-3), suggesting that HDL may protect against atherogenesis. HDL has been shown to remove cholesterol from cells in vitro (4-9) and has been reported to mediate cholestrol removal from the periphery i n vivo (10,11), facilitating its delivery to the liver (12) where it either can be catabolized or excreted from the body. This "reverse cholesterol transport" by HDL (13) may therefore contribute to the protective effect of HDL reported in population studies. Recent studies from our laboratory have provided evidence that removal of cholesterol from cells may be mediated by the interaction of HDL with a single class of high affinity binding site on the cell surface (14,15). Since the number of these sites appears to increase with increasing cellular cholesterol content, we postulated that they may represent receptors for HDL that act to facilitate transport of cholesterol from cells to HDL particles.
The possible interaction of HDL with a cell-surface receptor on cultured cells implies that HDL possesses specific recognition sites for this receptor within one or more of its components. Recent studies have provided conflicting evidence to the nature of the components involved in the interaction of HDL with cultured cells. Using recombinant proteoliposomes containing HDL apoproteins and phospholipids, results from several laboratories including our own have suggested that the recognition sites lie within the HDL apoproteins, particularly apo-A-I but possibly apo-A-I1 and others (16)(17)(18). Recently, Tabas and Tall (19) concluded that surface lipids, not proteins, are the major components that interact with cultured cells. The purpose of the present study was to test the effects of chemical modification of specific apoprotein amino acid residues on the ability of HDLB to bind to cultured fibroblasts and promote efflux of cellular cholesterol. Results show that modification of HDL apoproteins by treatment with tetranitromethane decreases both cellular binding and HDL-mediated cholesterol efflux, providing evidence that both processes are closely coupled and involve the interaction of HDL apoproteins with the cell surface.

EXPERIMENTAL PROCEDURES
Cells-Normal human skin fibroblasts were grown from explants of punch biopsies of skin from the inner thighs of normal volunteers in plastic tissue culture flasks containing Dulbecco's minimum essential medium plus 10% fetal bovine serum (growth medium) at 37 "C in humidified incubators equilibrated with 5% COZ, 95% air. Cells were trypsinized from stock flasks (2-14 passages), seeded in 35-mm plastic Petri dishes using 2 ml of medium containing 5 X lo4 cells, and grown in growth medium until just before reaching confluency, usually 6-10 days after plating. Cells were then washed twice with a PBS-albumin medium (0.2 g/l of KCl, 0.2 g/l of KHzPO,, 8.0 g/1 of NaCl, 2.16 g/1 of NazHP04 7Hz0, 2.0 g/l of bovine serum albumin, at pH 7.4) and incubated for 48 h with serum-free culture medium containing albumin, 2.0 g/l, plus the indicated concentration of cholesterol. Cholesterol was dissolved in 95% ethanol (10 mg/ml) before addition to the albumin medium. The final ethanol concentra-495 tion in the incubation medium was 0.5%, equivalent to a final concentration of cholesterol of 50 pg/ml. After the preincubation periods, the indicated parameters of cell cholesterol metabolism were assayed. Unless indicated otherwise, each data point for the figures and tables represents the mean of determinations on duplicate dishes.
Liposomes-Phospholipid proteoliposomes were made by cholate dialysis (22) using egg lecithin (Applied Science Laboratory) and apo-A-I in an approximate molar ratio of 300:l. This method previously has been reported to produce uniform round bilamellar vesicles which, when prepared with cholesterol, are biologically active substrates for the enzyme lecithin-cholesterol acyltransferase (22). When indicated, vesicle phospholipids were radiolabeled by the addition of L-W [methyl-3H]dipalmitoylphosphatidylcholine to the egg lecithin.
Chemical Modification of Lipoproteins and Liposomes-Acetylation was performed according to the method of Fraenkel-Conrat (23) as modified by Basu et al. (24). An HDL3 solution containing 2-5 mg of protein/ml was diluted 1:l (v:v) with saturated sodium acetate and, while being continuously stirred for 90 min in an ice bath, acetic anhydride was added in 2-8 equal aliquots to a total volume equivalent to 1.5 pl/mg of HDL, protein. Alkylation and reduction of HDL, was carried out according to the mercaptoethanol-iodoacetamide method of Weisgraber et al. (25) using a concentration of HDL, similar to that used in acetylation. Cyclohexanedione (Aldrich) treatment was performed following the method of Patthy and Smith (26) as modified by Mahley et al. (27) using the same HDL, concentrations as described for acetylation. Effective modification was confirmed by agarose gel electrophoresis and amino acid analysis.
For each TNM (Aldrich) modification, a fresh stock solution was prepared by dilution of TNM at room temperature in 95% ethanol to a concentration 100 times that of the desired reaction concentration (e.g. 72 pl of TNM plus 128 p1 of ethanol to make 3 M stock for a 30 mM reaction concentration). The stock was then added to solutions containing HDL or proteoliposomes at protein concentrations of 0.5-1.5 mg/ml in 0.15 M NaCl and 5 mM EDTA, pH 7.4. The mixture was allowed to stand at room temperature for 1 h, after which it was dialyzed extensively against saline/EDTA at 4 "C. This procedure was based on previously described methods (28,29) and on methods personally communicated by Dr. Y.-D. I. Chen, Stanford University.
HDL Binding-To determine cell binding of '251-HDL3 at 37 "C, cells were rapidly washed three times with PBS-albumin medium (wash medium) and incubated at 37 "C with serum-free culture medium containing 1.0-2.0 mg/ml albumin and lZ51-HDL3 at a protein concentration of 2 pg/ml. After 1 h, dishes were chilled on ice, washed rapidly three times with ice-cold wash medium, incubated twice for 10 min with the same medium, and washed twice again rapidly with cold wash medium containing no albumin. Cells were then digested in 0.1 N NaOH and an aliquot was assayed for radioactivity and another aliquot was assayed for protein content. When the binding assay was performed at 4 "C, cells were washed twice at room temperature and chilled to 4 "C while being bathed with the third wash medium. Cells were then incubated at 4 "C with serum-free medium containing 1.0-2.0 mg/ml albumin, 10 mM Hepes (pH 7.4), and ' ' ' I-HDL,. After 2 h, cells were washed seven times by the same procedure used for the 37 "C binding assay and digested with NaOH. Previous studies (14) demonstrated that the major components in the ' "I-HDL3 fraction that bound to cells were the apo-A-I-and A-IIcontaining particles. Competitive binding studies performed at both 37 "C and 4 "C produced similar results. fH]Cholesterol Efflux, Cholesterol Ester Formation, and Sterol Mass-Cellular lipids were extracted by the hexane-isopropyl alcohol method described by Brown et al. (5). Esterified and unesterified cholesterol were separated by thin layer chromatography, saponified in ethanolic KOH, and assayed for radioactivity and mass by methods previously described (15). Cholesterol mass was measured by the cholesterol oxidase method of Heider and Boyett (30). To assay the rate of efflux of cholesterol from cells, fibroblasts prelabeled with [3H]cholesterol were subsequently incubated with nonradioactive medium and the amount of 3H radioactivity and cholesterol mass appearing in the medium was measured. The rate of cholesteryl ester formation in cells was determined as described previously (15). Briefly, after the indicated incubation, cells were washed rapidly at room temperature with PBS and serum-free culture medium containing [''C]oleic acid (20-30 p~, 2 pCi/ml) bound to albumin (0.05-0.10%) was added to the dishes. After 1 h at 37 "C, cells were chilled on ice and washed twice with cold PBS and cholesteryl esters were extracted and isolated for measurement of 14C radioactivity.
Electrophoresis and Gel Filtration Chromatography-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out by the method of Laemmli (31) using a gradient of 10%-15% SDS and a current of 5 mA overnight and then 30 mA for 1 h. Samples were boiled in the presence of dithiothreitol for 3 min in SDS solubilization medium prior to use. Gradient gel electrophoresis was performed on a Pharmacia Electrophoresis Apparatus (GE-4) using slab gradient gel PAA 4/30 (Pharmacia) at a constant voltage of 125 V for 24 h. Gels were stained with Coomassie Blue. Agarose gel electrophoresis was performed using the Paragon system for lipoprotein electrophoresis (Beckman) at pH 8.6. The gels were stained with Fat Red 78.
Gel filtration chromatography was performed by two methods. For method A, samples were applied to a Bio-Gel A-5m column (1.5 X 90 cm, 200-400 mesh, 1.5 X 90 cm) (Bio-Rad) in buffer of 100 mM NaCl, 10 mM EDTA, and 100 mM Tris-HC1 (pH 7.5) at a protein concentration of 1 mg/ml. The column samples were eluted at 6-8 ml/h and 2-ml fractions were assayed for protein content by absorbance at 280 nm. For method B, high performance gel filtration was performed as described by Chen and Albers (32) using a Waters high performance liquid chromatography system (Milford, MA) equipped in series with a 7.5 X 40 mm pre-column (Waters Associates), a 7.5 X 600 mm Bio-Si1 TSK-250 column (Bio-Rad Laboratories), and a 7.5 X 600 mm TSK-4000 column (Beckman). The columns were equilibrated with 10 mM Tris buffer containing 140 mM NaCl and 1 mM EDTA, pH 7.4, and were eluted with the same buffer.
Radial Immunodiffusion and Immunoprecipitation-Immunoprecipitability of HDL3 was assessed using sera from rabbits immunized to purified human apo-A-I (33, 34). Single HDL, and TNM-HDL, were immunoprecipitated by the addition of 200 pl of rabbit anti[human apo-A-Ilantisera (33) to 25 pg of HDL protein in 500 p1 of culture medium. Samples were incubated for 1 h at 37 "C, followed by 24 h at 4 'C, and then were centrifuged at 1000 X g for 20 min. Lipids were extracted from the washed immunoprecipitates in hexane:isopropyl alcohol (3:2, v:v).
Amino Acid Analysis-For amino acid analysis, samples were hydrolyzed in constant boiling 6 N HCl in a sealed nitrogen atmosphere for 24 h. Analysis was performed using a Durrum D-500 amino acid analysis (36). With this hydrolysis procedure, the effects of TNM treatment and reduction/alkylation were determined directly from the disappearance of tyrosine and cysteine residues, respectively. Results demonstrated that modification of cysteine residues by sulfide-acetamide bond formation was irreversible under these hydrolysis conditions. For cyclohexanedione-treated samples, hydrolysis was performed in the presence of mercaptoacetic acid (2%, v:v) to stabilize the cyclohexanedione derivative of arginine (26,271. It was estimated that cyclohexanedione treatment modified 30 to 40% of the arginine residues in HDL3. To estimate the degree of acetylation of lysine residues, the indirect method of Wofiy and Singer (37) as modified by Wiesgraber et al. (25) was used. Briefly, native and acetylated HDL3 were treated with dinitrofluorobenzene to convert nonacetylated lysines to acid-stable dinitrophenylated derivatives prior to hydrolysis. Results indicated that more than 80% of the lysine residues were acetylated by acetic anhydride treatment.
Fatty Acid Analysis-Lipids were extracted from HDL in chloroform:methanol(2:1, v:v) and lipid classes were separated by thin layer chromatography by previously described methods (15). The phospholipid silica spot and the combined triglyceride plus cholesteryl ester (neutral lipid) spots were suspended in ethanolic (95%) KOH (0.1 M) and lipids were saponified for 1 h at 80 "C. The liberated free fatty acids were extracted in hexane, the solvent was evaporated under Nz, the lipid was redissolved in petroleum ether, and an equal volume of 1% H2SO4 in methanol was added. After heating at 80 "C for 2 h, the methyl esters were extracted in hexane and the solvent was dried under NZ. Samples were resuspended in 0.1 ml of hexane and analyzed in a Hewlett-Packard 5790A gas chromatographer with flame ionization detector and HP3390A integrator. Fatty acids were separated on a 30 m X 0.25 mm inside diameter, 0.2-p-thick capillary column.
The area under the peaks for each fatty acid was integrated and expressed as per cent of the total area for all peaks.
Other Methods-Electron microscopy of HDL, negatively stained with 1% sodium phosphotungstate at pH 7.4 was performed by previously described methods (38). Lipoprotein triglyceride and phospholipid mass were determined by previously described methods (39,40). Protein content of lipoproteins and cells was determined by the method of Lowry et al. (41). RESULTS To identify the domain(s) on HDL apoproteins that are recognized by the HDL binding site, we employed methods of protein alteration specific for certain amino acid residues. Acetylation, reduction and alkylation, or cyclohexanedione treatment were performed on HDL3 to alter lysine, cysteine, and arginine residues, respectively. These modifications caused little or no change in the ability of HDLB to compete for binding of '251-HDL3 to fibroblasts (Fig. la), suggesting no significant involvement of these amino acid residues in recognition of HDL3 by its cellular binding site.
Recent reports (42, 54) suggested that alteration of HDL by TNM, presumably specific for tyrosine residues, abolished its binding to hepatocyte membranes. To determine if tyrosine-containing segments of human HDL apoproteins constitute the ligand site recognized by the HDL binding site of cultured fibroblasts, the binding of TNM-treated HDL, was compared with that of control HDL, by means of competition studies. When HDL3 had been treated with TNM at a concentration of 3 mM, its ability to compete for Iz5I-HDL3 binding was almost totally lost (Fig. lb). Even at a 10-fold excess, TNM-treated HDL3 competed poorly with the Iz5Ilabeled ligand. This is in contrast to the effective competition by native HDL3 seen even at low concentrations. To examine for possible dose dependency of this TNM effect, HDLB was treated wtih TNM at concentrations of mM to 30 mM and tested for its ability to compete for lZ5I-HDL3 binding (Fig. 2). There was no significant effect on binding competition up to a TNM concentration of 0.1 mM. Above 0.1 mM TNM there was a precipitous decrease in competition and then a plateau beginning at 3 mM. At the higher concentrations of TNM, less than 15% of lZ5I-HDL3 binding was displaced, even at a 10-fold excess of unlabeled ligand (Fig. lb).
Previous studies demonstrated that both major HDL subclasses, HDL, and HDL,, interact with the HDL binding site on cultured human skin fibroblasts (14). On a protein basis, HDL, and HDL3 compete for Iz5I-HDL3 binding to similar extents (Fig. 3). Competitive binding of both HDLz and HDL3 was nearly abolished when the lipoproteins were treated with TNM.
Treated HDL3 samples across a range of TNM concentrations were subjected to agarose electrophoresis along with untreated HDL3 (Fig. 4). Increased electrophoretic mobility was noticeable at a TNM concentration of 0.3 mM, corresponding to the TNM concentration at which HDL, binding competition was decreased (Fig. 2).
TNM alteration is relatively specific for tyrosine residues, especially at a pH of 8 or less (29,43). Amino acid analysis of altered and unaltered HDL, showed that TNM treatment caused a shift of the entire tyrosine peak into the phenylalanine peak (Fig. 5), without substantial changes in the other major amino acid peaks (Table I)

Effects of TNM treatment on amino acid composition of HDIn
The area under the chromatographic peak representing each amino acid residue was measured for native and 10 mM TNM-treated HDL, in separate chromatography runs of equal amounts of protein. The values for the TNM-treated HDLs were divided hy those for the native particle. Only the major peaks are shown. A similar degree of tyrosine modification was observed for HDL proteins treated with 3 mM TNM (data not shown). Amino  hydrolysis used for amino acid analysis destroys tryptophan, it is not known if this amino acid is also affected by T N M treatment.
TNM-altered HDL, was compared to native HDLa by 10-15% gradient SDS-PAGE (Fig. 6) from the gel. The oligomer formation and eventual failure of entry into the gel strongly suggested that cross-linking between HDL apoproteins was caused by TNM treatment (44).
Gradient gel electrophoresis demonstrated that marked aggregation of HDLa particles did not occur until the HDLn was treated with TNM in excess of 3 mM (Fig. 7). Although increases in particle size became evident at 3 mM TNM, nearly all of the particles had the same distribution as untreated HDLa. Similar results were observed when particles were analyzed by molecular sieve chromatography (Fig. 8). At a TNM concentration of 3 mM, the peak was shifted slightly to the left, indicatinga slight increase in average Stokes radius of the particles. This suggested a minor change in particle size distribution that could have occurred with apoprotein cross-linking. In contrast, at 10 mM TNM, there was a significant shift and broadening of the peak, suggesting a spectrum of increase in particle size. Electron microscopy also revealed that treatment of HDL, with 3 mM TNM had little effect on particle morphology (Fig. 9). At 3 mM TNM. the particles were similar in size and shape to native HDLn and  there was no evidence of interparticle aggregation. In contrast, at TNM concentrations of 10-30 mM, the particles appeared to aggregate in clumps and chains and much larger particles were formed, suggesting that treatment of HDLs with the higher T N M concentrations led to coalescence of two or more HDL particles. Measurements of the lipid mass in HDLR demonstrated that TNM treatment caused modest changes in the lipid to protein ratio over the TNM concentration range that induced loss of HDL binding (Fig. 10). With the exception of unesterified cholesterol, the ratio of lipids to protein in HDLn progressively decreased with increasing concentrations of TNM. The unesterified cholesterol to protein ratio increased to a maximum value at the lowest concentration of T N M tested. This increase in unesterified cholesterol content in association with decreases in esterified lipids suggested that TNM may have promoted ester hydrolysis. However, the relative composition of fatty acids in both phospholipids and neutral lipid esters (triglyceride plus cholesteryl ester) appeared to be  (Table   II), indicating that, unless all ester bonds were affected equally, hydrolysis of HDL lipid esters could not account for the decrease in lipid to protein ratio. The fatty acid composition data also showed that TNM did not have a marked effect on the degree of unsaturation of acyl groups, indicating that nitration of lipid double bonds did not occur to any significant extent.
T o further investigate the possibility that TNM blocked HDL binding by altering the interaction between HDI, apoproteins and lipids, studies were conducted using recombinant phospholipid-apoprotein vesicles. Preliminary studies (16) demonstrated that proteoliposomes prepared with egg lecithin and apo-A-I, the major apoprotein in HDL, were as effective as native HDLa particles in competing with ""I-HDLn binding. When apo-A-I proteoliposomes were treated with 3 mM TNM, there was a broadening of the apo-A-I hand on SDS polyacrylamide gels and a slight retardation in mobility, similar to that observed with HDLn (compare Fig. 11, inwt, to   6). TNM treatment of the proteoliposome almost completely abolished its ability to compete for "'II-HDLs binding (Fig. 11). When proteoliposomes were prepared with ["HI phosphatidylcholine and characterized by high performance liquid chromatography, the apo-A-I and radiolabeled phospholipid eluted together as a single peak with an apparent size similar to HDL (Fig. 12A). Except for a slightly broader peak, the elution profile for TNM-treated proteoliposomes was the same as that of the untreated vesicle, with both protein and phospholipid eluting together in a single population of particles (Fig. 12R). These results indicate that TNM treatment blocks competitive binding of apo-A-I proteoliposomes without causing marked alteration in the size distribution and lipid composition of particles.
Since HDLa treat,ed with 3 mM T N M showed greatly decreased binding competition and yet had minimal particle aggregation or distortion of particle size, this preparation was selected for studies of HDL-mediated cholesterol efflux from cultured cells. T o determine any direct effect of TNM on the ability of HDL to accept cholesterol, exchange of ["Hlcholesterol from an aqueous albumin suspension was measured in the presence of native or TNM-treated HDL, (Table 111). Both the treated and untreated HDL particles were isolated from the cholesterol-albumin suspension by immunoprecipitation using antisera containing anti-apo-A-I demonstrated by radial immunodiffusion to possess near-complete reactivity toward TNM (3 mM)-treated HDL:I. The ability of HDI,, to accept cholesterol from the albumin suspension was unaltered by TNM treatment. Despite normal ability to accept cholesterol in a cell-free system, TNM-treated HDLZI was much less effective than HDL, in promoting cholesterol transport from cultured fibroblasts, whether measured by the appearance of [3HH]cholesterol or unesterified cholesterol mass in the medium or by the decrease in ["H]cholesterol or cholesterol mass in the cell (Table IV). When ["Hjcholesterol efflux from cells was measured as a function of the concentration of native HDLs, both a high and a low affinity component were evident (Fig. 13). TNM treatment greatly reduced the high affinity component of efflux without affecting the low affinity component.
T o examine the dose dependency of the TNM effect on HDL-mediated cholesterol transport from cells, HDL, particles treated with different concentrations of TNM were compared for their ability to remove cholesterol from cells over short term incubations (4 h). As an assay of net cholesterol transport, the relative rate of cholestewl ester formation by cells was measured after exposure to HDL:l. Previous studies (5,6,53) have demonstrated that the cholesterol esterification rate is a function of the amount of excess unesterified cholesterol in cells and thus can be used as a sensitive biochemical assay for changes in net flux of cholesterol into or out of cells.
With an increasing dose of TNM, there was a reduction in the ability of HDLs to remove cholesterol from cells as evidenced by a diminished ability to suppress the rate of cholesteryl ester formation (Fig. 14). The TNM-mediated reduction in cholesterol transport was dose-dependent over a concentration range below 3 mM, similar to that observed for its effect on cellular binding of HDL,. As with hinding, the major TAME 111 Effect of TNM treotment on thP nhility of HDI, lo orrrpt rho/f7strro~ from o cholesterol-olhumin susprnsion ['HH]Cholesterol in ethanol (10 pg/ml, 1 pCi/ml) was added t o medium containing 2 mg/ml alhumin as described under "Experimental Procedures" and incuhated for 25 h at 37 "C. For one set of tuhes (25 h), either native or 3 mM TNM-treated HDIn was added to the medium at zero time and rahhit antiserum was added after 24 h of incuhation. For the other set of tuhes ( 1 h). lipoprotein and antiserum were added simultaneously after 24 h. Both sets were then chilled on ice after an additional I-h incuhation at 3'7 "C (with antisera) and incuhatetl for 24 h at 4 "C. Tubes were then centrifuged, the pellet was washed with PRS-alhumin at 4 * C , and lipids were extracted with hexane-isopropyl alcohol and assayed for 'H. Values are the mean rt S.E. of three incuhations after suhtrartion of values for tubes incuhated with control sera.   /ml), washed five times with PBS-albumin, and then incubated with serum-free medium containing albumin plus the indicated lipoprotein (20 pg/ml). After 48 h, the cells were chilled to 0 "C, the medium was collected and assayed for [3H]cholesterol and cholesterol mass, and the washed cells were extracted with hexane/isopropyl alcohol for unesterified cholesterol mass and radioactivity determinations. Net change in medium cholesterol mass was calculated as the difference in unesterified cholesterol in medium from dishes with and without cells. Cell-free values for medium containing no HDL3, native HDL,, or TNM-HDL, were 0.12, 0.52, and 0.59 pg of cholesterol/ml, respectively. Net change in cellular cholesterol mass and radioactivity was calculated from the difference between zero time on cultured fibroblasts. This implies a lack of involvement of lysine and cysteine residues in recognition of HDL by its binding site. Cyclohexanedione treatment caused only a slight decrease in competitive binding of HDL3, suggesting that arginine residues are also not involved in the interaction of HDL with cells. However, since this treatment modified less than half the arginine residues in HDL3, the possibility of arginine involvement cannot be completely eliminated. The degree of inhibition of LDL binding to its receptor by cyclohexanedione treatment was found to be a function of the number of arginine residues modified (27).
Chen and colleagues originally reported (42) that TNM modification of rat HDL altered its interaction with isolated hepatocyte membranes and perfused rat liver and they suggested that tyrosine was present in the HDL binding recognition site. To test for possible tyrosine involvement in the active ligand site of human HDL for its receptor on extrahepatic cells, native and TNM-modified HDL were compared in their ability to compete with HDL for its binding to human fibroblasts. In agreement with the earlier report (42), TNM treatment of HDL profoundly decreased its competitive binding to fibroblasts. Amino acid analysis of TNM-HDL in the present study showed that tyrosine was specifically altered, suggesting the presence of tyrosine at the active ligand site. Furthermore, the increased mobility of TNM-HDL on agarose gel electrophoresis was consistent with the concept that formation of 3-nitrotyrosine had occurred, resulting in an increased negative charge of the particle. However, it is impossible to draw firm conclusion regarding tyrosine involvement at the active ligand site in light of the results of SDS-PAGE analysis which clearly showed protein cross-linking due to TNM treatment. Significantly, the TNM dose-response curve for binding competition paralleled the degree of cross-linking evident by SDS-PAGE. Cross-linking as a side reaction to T N M was first described in 1968 by Doyle and colleagues (46) in the case of collagen, y-globulin, and carboxypeptidase A. TNM has been reported to cross-link certain other proteins (47, 48), although many additional proteins do not appear to undergo this side reaction (46, 49, 50). Even though TNMmediated cross-linking is believed to specifically involve tyrosine residues (47,51), the cross-linking reaction itself alters the secondary structure of an affected protein. Thus, TNM cross-linking is 1ikely.to affect accessibility of many amino acid residues other than tyrosine at some distance from the cross-linking site.
Although involvement of specific amino acids was not demonstrated conclusively, the current study provides additional evidence that binding of HDL, to cholesterol-treated cells is mediated by HDL apoproteins rather than lipid. SDS-PAGE analysis revealed dose-dependent TNM cross-linking of HDL apoproteins which closely paralleled the dose-dependent effect of TNM on binding. As evidenced by gradient gel filtration, electrophoresis, and electron microscopy, the size and shape of HDL, particles was not altered substantially by treatment with 3 mM TNM, a concentration that led to maximum suppression of HDL, binding to cells. At concentrations below 3 mM, TNM had only a modest effect on the lipid to protein ratio of HDL,. On an HDL, protein basis, TNM treatment led to the same degree of loss (-25% at 3 mM TNM) of all the major lipid classes except unesterified cholesterol. Analysis of HDL lipids by gas chromatography indicated that the loss in phospholipid, cholesteryl ester, and triglyceride was not the result of extensive hydrolysis of ester bonds or nitration of acyl group double bonds. It is possible that TNM treatment led to a modification of the physical properties of a proportion of HDL, lipids so that they were incompletely recovered during extraction. This conclusion is supported by results from a recent study by Chacko (54), who demonstrated that TNM treatment caused an increase in the proportion of phospholipid and cholesteryl esters that remained associated with protein during lipid extraction, suggesting that the apparent lipid loss was due to cross-linking of some of the HDL lipids to apoprotein. Regardless of the mechanism involved, it is unlikely that the relatively minor changes in lipid composition alone could account for the striking suppression of competitive binding that was observed when HDL3 was treated with 3 mM TNM.
Additional evidence that protein is involved in the interaction of HDL with its cell-surface binding site was provided by studies comparing competitive binding of native and recombinant HDL particles that have different protein and lipid compositions. On a protein basis, HDL3 and HDLZ compete for lZ5I-HDL3 to the same degree, even though HDLz is a larger particle containing substantially more lipid. Treatment with 3 mM TNM suppressed competitive binding of both HDL subclasses to the same extent. While protein-free egg-lecithin liposomes are poor competitors, proteoliposomes containing apo-A-I are as effective as HDLB in their ability to compete for HDL binding (17). Treatment of apo-A-I vesicles with 3 mM TNM markedly suppressed their competitive binding with only minor changes in size distribution of the particles. This suppression in binding occurred even though there was no decrease in the phospholipid to apoprotein ratio of these particles, providing additional support for the conclusion that the inhibitory effect of TNM was not due to changes in particle lipid composition.
In a recent report (19), Tabas and Tall concluded that association of HDL, with cultured cells involves the interaction of surface lipids of HDL, with the cell membranes. Their conclusions were based in part on results showing that trypsin digestion of HDL, proteins did not eliminate the ability of HDL, to interact with cells. It is possible, however, that the binding domains on HDL are located within the hydrophobic regions of the apoprotein and may be protected against trypsin digestion. Since TNM is also hydrophobic, these regions may be particularly sensitive to cross-linking by TNM. When purified lipid-depleted apo-A-I was treated with 3 mM TNM in aqueous solution, very little cross-linking was evident and tyrosine residues were unmodified (data not shown), suggesting that the lipid environment was an important factor in the modification process. That the receptor binding domains on apoproteins can be located in areas protected against proteolysis is supported by results showing that treatment of LDL with trypsin had little effect on its ability to interact with its receptor (52).
Several recent studies have demonstrated that TNM treatment of HDL abolishes its interaction with high affinity binding sites on human liver (55), rat liver (54), and rat ovarian (56) membranes. Although high affinity binding was reduced, TNM-treated HDL was still capable of stimulating steroidogenesis by rat ovarian cells (56), suggesting that delivery of HDL sterol to this cell type is not mediated by binding of HDL to its high affinity site. In apparent contrast to delivery of HDL cholesterol to ovarian cells, our previous studies (14, 15) suggested that removal of cholesterol from cultured fibroblasts is facilitated by binding of HDL to high affinity sites on the cell surface. Since treatment of HDL3 with 3 mM TNM suppresses its ability to interact with its high affinity binding site without causing major changes in lipid composition or particle morphology, we tested the feasibility of using TNM-treated HDL, as a negative control for studies evaluating the role of HDL binding in removal of cholesterol from cells.
The current and previous studies (14, 15) demonstrated that, as with cellular binding of HDL, a typical saturation curve for promotion of cholesterol efflux from fibroblasts contains both a high affinity, saturable component and a low affinity, nonsaturable component. We postulated that the high affinity component may represent transport of cholesterol from cells that is mediated by HDL binding to its high affinity site (14, 15,53). This concept is supported by present results showing that treatment of HDL, with TNM decreased its ability to promote cholesterol transport from fibroblasts by the high affinity process. As with HDL binding, the reduction in HDL-mediated cholesterol transport by TNM treatment was dose-dependent over a TNM concentration range of 1-3 mM. The ability of HDL, to promote cholesterol efflux by the low affinity process appeared to be unaffected by TNM treatment, suggesting that the capacity of the particle to accept cellular cholesterol was not abolished by TNM treatment. Moreover, TNM treatment failed to alter the ability of HDL, to sequester cholesterol when HDL3 was added to a cholesterol-albumin suspension. Although cellular binding of HDL, does not appear to be essential for transport of cholesterol from extrahepatic cells to HDL3 (7), results from the current study support the hypothesis that HDL binding may facilitate removal of cholesterol from cells when they become acutely overloaded with cholesterol. This facilitative process may be mediated by a specific cell-surface receptor for HDL that is induced in response to cholesterol loading (15).