Apoprotein B Structure and Receptor Recognition of Triglyceride-rich Low Density Lipoprotein (LDL) Is Modified in Small LDL but Not in Triglyceride-rich LDL of Normal Size*

We compared the effect of lipid composition and par- ticle size of triglyceride-rich low density lipoprotein (LDL) upon apoprotein B conformation and binding to the LDL receptor. Three groups of triglyceride-rich LDL were studied: (a) LDL isolated from chronic hy- pertriglyceridemic individuals (HTG-LDL); (b) normal LDL made triglyceride-rich by in vitro incubation with triglyceride emulsion and the neutral lipid transfer protein (R-LDL); and (c) LDL from normolipidemic individuals made acutely hypertriglyceridemic by in- travenous infusion of 10% Intralipid (IV-LDL). HTG-LDL was small and dense, whereas R-LDL g/ml) (Beckman Co.) 45,000 rpm 20 h. The g/ml was washed by spinning at the same density. The LDL was then collected and dialyzed in 0.19 M NaCl, 0.25 mM EDTA, pH 7.4, and then filtered using a 45-pm filter (Gelman Sciences, Ann Arbor, MI). All procedures on LDL and storage (under argon gas) were at 4 "C. LDL from Chronic Hypertriglyceridemic Subjects (HTG-LDL)-- Ten chronic hypertriglyceridemic individuals with diverse clinical conditions (type I hyperchylomicronemia (n = 2), diabetes mellitus (n = 2), glycogen storage disease type 1 (n = 4), and type IV phenotype hyperlipidemia (n = 2)) participated in this study. The mean levels of plasma triglycerides f S.E. were 1,740 f 339 mg/dl (range 755-3980) and of cholesterol, 380 f 65 mg/dl (range 148-726). Their plasma was used to isolate HTG-LDL by sequential ultracentrifugation (as described for the separation of N-LDL) or by nonequilibrium rate zonal ultracentrifugation using a Ti-14 zonal rotor (Beckman) and a discontinuous NaBr gradient of 1.0-1.3 g/ml (23, 29). Zonal ultracentrifugation runs lasted exactly 170 min. The effluent was monitored continuously at absorbance of 280 nm, and the fraction containing LDL was collected and concentrated under vacuum at 4 "C. The LDL was then dialyzed, filtered, and stored as explained above. ultracentrifugation, free cholesterol/protein, esterified cholesterol/protein, and phospholipid/protein.

We compared the effect of lipid composition and particle size of triglyceride-rich low density lipoprotein (LDL) upon apoprotein B conformation and binding to the LDL receptor. Three groups of triglyceride-rich LDL were studied: (a) LDL isolated from chronic hypertriglyceridemic individuals (HTG-LDL); ( b ) normal LDL made triglyceride-rich by in vitro incubation with triglyceride emulsion and the neutral lipid transfer protein (R-LDL); and ( c ) LDL from normolipidemic individuals made acutely hypertriglyceridemic by intravenous infusion of 10% Intralipid (IV-LDL). HTG-LDL was small and dense, whereas R-LDL and IV-LDL had normal size. HTG-LDL, but not R-LDL or IV-LDL, exhibited decreased binding to the LDL receptor on human skin fibroblasts in studies at 4 "C and reduced degradation at 37 "C. Apoprotein B conformation was assessed by circular dichroism and by analyzing the immunoreactivity of different monoclonal antibodies. HTG-LDL but not R-LDL or IV-LDL showed a change in the CD spectra and a consistent decrease in the immunoreactivity of monoclonal antibody 3F5 (2.5-fold) which recognizes an epitope adjacent to the receptor binding domain of apoprotein B. These findings suggest that in triglyceride-rich LDL, the relative content of neutral lipid in the core of LDL in the absence of changes in the size of the particle does not significantly affect apoprotein B conformation or its affinity for the LDL receptor.
High levels of low density lipoprotein (LDL),' the major $3 To whom correspondence should be addressed Dept. of Pediatrics, Columbia University, 630 W. 168th St., New York, NY 10032.
The abbreviations used are: LDL, plasma low density lipoprotein; apo, apoprotein; N-LDL; LDL isolated from normolipidemic control subjects; HTG-LDL, LDL isolated from subjects with chronic hypertriglyceridemia; R-LDL, LDL enriched in triglyceride after remodeling by in vitro incubations with triglyceride emulsions; IV-LDL, LDL isolated from normal subjects after acute hypertriglyceridemia induced by intravenous infusion of triglyceride emulsions; VLDL, very low density lipoprotein; Mab, monoclonal antibody; PBS, phosphatebuffered saline. carrier of cholesterol in plasma, are associated with increased risk of atherosclerosis (1). LDL exhibits a wide heterogeneity in lipid composition, immunoreactivity, size and density in normal (2,3) and dyslipidemic individuals (4,5). It has been suggested that these differences may affect the interaction of LDL with its receptor (6) and consequently, the metabolism of cholesterol.
Apoprotein B-100 (apoB) is the major apoprotein of LDL and its ligand for the LDL receptor. ApoB is a very hydrophobic molecule containing 4,536 amino acid residues and a high number of repeating subunits (7,8). Although the size and insolubility of apoB have hindered progress in its structural analysis, the use of monoclonal antibodies (Mabs) against different epitopes of apoB has been useful in determining the general structure, the binding region to the LDL receptor, and more recently in characterizing genetic variants of the molecule (9)(10)(11)(12)(13)(14). The putative binding region of apoB to the LDL receptor consists of a cluster of positively charged amino acids located between amino acids 2835 and 4189 and appears to be placed in the aqueous face of the molecule (8). The size of the apoB-containing lipoproteins (very low, intermediate, and low density lipoproteins) has been suggested to be an important determinant of the conformation of the apoB molecule and its binding to the LDL receptor as VLDL is converted to LDL during lipolysis (6,8,15,16). Still, mechanisms regulating structure and conformation of apoB in LDL itself are not well understood. Of interest, LDL lipid composition, particularly the triglyceride content, has been proposed as an important modulator of apoB interaction with cell receptors (17)(18)(19)(20)(21).
The purpose of this study was to understand better how the size and the lipid composition of LDL might affect apoB conformation and its binding to the LDL receptor. We chose to study triglyceride-rich LDL because LDL triglyceride and cholesteryl ester contents are relatively simple to modify both in human dyslipidemias and by in vitro incubations of LDL (5, [23][24][25][26]. LDL particles were isolated from normolipidemic volunteers, from a variety of genetically distinct chronic hypertriglyceridemic individuals, or after in vitro or in vivo modification of the LDL lipid content using lipid emulsions. Our results demonstrate: ( a ) that changes in LDL lipid composition, especially in the triglyceride content, in the absence of change in size, do not affect either the apoB conformation or the binding to the LDL receptor, and ( b ) that in small LDL, apoB has alterations in configuration, which is associated with a lower affinity to the LDL receptor.

Cells
Human fibroblasts from foreskin of normal newborns were plated from frozen cells (6th-12th passages) at densities of 5 X lo', 2.5 X lo', or 1 X lo' cells/well in plates of 6, 12, or 24 wells, respectively.

Lipoproteins
Normal LDL (N-LDLI-Fasting plasma from normolipidemic subjects was obtained, and an antiproteolytic mixture (25 pl/ml) was added to reach a final concentration of 1.2 g/liter EDTA, 0.1 g/liter NaN3, and 100,000 kallikrein inhibitory units/liter aprotinin. The density was adjusted with NaBr, and LDL was isolated by sequential ultracentrifugation (1.025 < d < 1.050 g/ml) using a 50.3 rotor (Beckman Co.) at 45,000 rpm for 20 h. The top fraction of the d = 1.050 g/ml run was washed by spinning at the same density. The LDL was then collected and dialyzed in 0.19 M NaCl, 0.25 mM EDTA, pH 7.4, and then filtered using a 45-pm filter (Gelman Sciences, Ann Arbor, MI). All procedures on LDL and storage (under argon gas) were at 4 "C.
LDL from Chronic Hypertriglyceridemic Subjects (HTG-LDL)--Ten chronic hypertriglyceridemic individuals with diverse clinical conditions (type I hyperchylomicronemia (n = 2), diabetes mellitus (n = 2), glycogen storage disease type 1 (n = 4), and type IV phenotype hyperlipidemia ( n = 2)) participated in this study. The mean levels of plasma triglycerides f S.E. were 1,740 f 339 mg/dl (range 755-3980) and of cholesterol, 380 f 65 mg/dl (range 148-726). Their plasma was used to isolate HTG-LDL by sequential ultracentrifugation (as described for the separation of N-LDL) or by nonequilibrium rate zonal ultracentrifugation using a Ti-14 zonal rotor (Beckman) and a discontinuous NaBr gradient of 1.0-1.3 g/ml (23,29). Zonal ultracentrifugation runs lasted exactly 170 min. The effluent was monitored continuously at absorbance of 280 nm, and the fraction containing LDL was collected and concentrated under vacuum at 4 "C. The LDL was then dialyzed, filtered, and stored as explained above. Most LDL were separated by nonequilibrium rate zonal ultracentrifugation, and no differences in composition or other properties, e.g. cell binding, were observed between LDL separated by either method.
Remodeled in Vitro LDL (R-LDL)-N-LDL was incubated in the presence of a triglyceride-phospholipid emulsion (20% Intralipid) and lipoprotein-deficient plasma (d > 1.21 g/ml, previously heated at 56 "C for 30 min to inactivate lecithin cholesterol acyltransferase) as a source for cholesteryl ester transfer protein (25). To rid Intralipid of excess phospholipids, Intralipid was washed twice by centrifugation in an SW 40 rotor (Beckman) at 25,000 rpm for 15 min, with a buffer containing 0.19 M NaCl, 1 mM EDTA, pH 8.5, at 4 "C. The top of every wash was resuspended in the same buffer into a volume equal to the original volume of Intralipid (25). In a typical experiment, 6 mg of LDL esterified cholesterol was incubated with 60 mg of emulsion triglyceride and 300-400 mg protein of d > 1.21 g/ml plasma in a solution of 0.19 M NaCl, 1 mM EDTA, 100 kallikrein inhibitory units/ml aprotinin, 10 pg/ml leupeptin, 1 pg/ml pepstatin, pH 8.25. The total volume of the mixture was 5 ml. The tubes were flushed with argon and put in a slowly shaking water bath at 37 "C for variable periods of time (0, 8, or 24 h). After incubation, the density was adjusted to 1.019 g/ml with NaBr, and the emulsion fraction was separated by ultracentrifugation in a 50.3 rotor at 45,000 rpm for 20 h. The density of the infranate was then adjusted to 1.063 g/ml and the LDL isolated and washed as described above. This method allows for separation of LDL, free of contamination from the emulsion fraction (23,25).
In Vivo Modified LDL (IV-LDL)-Seven normolipidemic volunteers were made acutely hypertriglyceridemic by a 6-h infusion of 10% Intralipid at a rate of 0.3 g of triglyceride/kg/h (30). After 6 h of Intralipid infusion the mean f S.E. of plasma triglyceride and cholesterol levels were 1,657 f 209 and 185 f 12 mg/dl, respectively. LDL before and after infusion was isolated by nonequilibrium rate zonal ultracentrifugation.
'=I-LDL-N-LDL was iodinated with lz5I using the monochloride method of McFarlane (31) modified by Bilheimer et al. (32). The specific activity ranged between 200 and 600 cpm/ng LDL protein, and more than 98% of the total radioactivity precipitated with trichloracetic acid (27).
SizelDensity of LDL-To assess LDL size and relative densities, nonequilibrium rate zonal ultracentrifugation was performed in d > 1.006 g/ml plasma from chronically hypertriglyceridemic subjects, from normolipidemic individuals after 6 h of intravenous infusion of 10% Intralipid, as well as on samples of R-LDL (d > 1.020 g/ml).
Our previous work has shown a close inverse correlation between the size of LDL and plasma triglyceride levels, as well as the effluent volume of LDL after zonal centrifugation (5). Similarly, the zonal effluent volume very closely correlates ( r = 0.99) with LDL flotation constants and molecular weight (26). R-LDL size was also evaluated by using 2-16% nondenaturing polyacrylamide gels (33). Typically 5 pg of LDL was combined with a solution of 40% sucrose and 0.1% bromphenol blue (3:1, v/v) and run on a buffer of 0.09 M Tris, 0.08 M boric acid, 0.0025 M EDTA at pH 8.4. The gels were preequilibrated at 4 "C and 125 volts for 20 min. The samples were initially run at 70 volts for 30 min and then at 125 volts for 24 h. After electrophoresis the gels were stained with Sudan Black B as described (33). Electron microscopy of several experimental LDL was also performed as reported previously (5). Lipoproteins were negatively stained with 2% sodium phosphotungstate, pH 7.4, on collodian carbon-coated grids. Electron micrographs were obtained with a JEM 1200 EX electron microscope (JEOL) at an instrument magnification of 80,000 and LDL concentration of 0.05-0.5 mg/ml. Results were expressed as the average of two measurements of the diameter of 100 particles. Cell Binding-To assess the affinity of different LDL for the LDL receptor, in most of experiments, competition studies were done in duplicate at 4 "C on monolayers of human skin fibroblasts; the medium (pH 7.4) contained Dulbecco's modified Eagle's medium, 20 mM Hepes, 5% lipoprotein-deficient serum, 5 pg/ml Iz6I-LDL, and increasing amounts of experimental LDL (27,28). After 2.5 h of incubation, the cells were washed three times with PBS, 0.2% albumin, pH 7.4, followed by two incubations of 10 min each with the same buffer and two rapid washes with PBS alone. At the end, the monolayers were harvested by dissolution in 0.1 N NaOH, and the amount of lz5I-LDL radioactivity was counted. Protein determination was then done on aliquots of the solution using the Lowry method (34). The amount of LDL bound per mg of total cell protein was calculated and the results expressed as percent of displacement of lZ5I-LDL in the absence of competitors. The results represent the average of duplicates which varied less than 10%. LDL degradation by human skin fibroblasts was evaluated by incubation of 20 pg/ml lZ5I-LDL in the presence of diverse amounts of competitive experimental LDL at 37 "C for 5 h. At the end of the incubation the medium was collected, and the proteolytic degradation of Iz6I-LDL was measured as described previously (27). Saturation binding assays at 4 "C were done in duplicate using increasing amounts of 'Z51-experimental LDL in the presence and absence of a 50-fold excess of unlabeled LDL, and the total, nonspecific, and specific binding was calculated (27,28). , MRW is the mean residue weight of the amino acids, 1 is the path length of the cell in cm, and c is the concentration of protein in g/ml. Spectra are analyzed for a-helix, p-sheet and turn and random coil at 1-nm intervals between 240 and 200 nm, as described (35).

Immunoreactivity of LDL ApoB with Monoclonal Antibodies
The radioimmunoassay with different anti-apoB monoclonal antibodies was slightly modified from that described previously (6). Immulon I1 Removawells (Dynatech Laboratories, Chantilly, VA) were coated by an overnight incubation with 200 pl of reference LDL (30 pg/ml in 5 mM glycine, pH 9.2) and subsequently saturated by incubation for 1 h with 250 p1 of 1% bovine serum albumin-PBS, pH 7.4. Serial dilutions (150 pl) of test and control LDL were prepared in microtiter plates. 150 pl of Mab, appropriately diluted in 1% bovine serum albumin-PBS, was added to the diluted LDL and allowed to incubate for 4 h at room temperature. Aliquots (200 pl) of the LDL-Mab mixture were transferred to the LDL-coated Removawells that had been washed with a solution of 0.15 M NaCl containing 0.025% Tween 20. The wells were incubated overnight and again washed with the Tween-saline solution as above. Two hundred pl of '"I-rabbit anti-mouse IgG (10) (1-2 X lo4 cpm/ng) diluted to 83 ng/ml was added to each well and incubated overnight. The wells were washed with the Tween-saline solution as above and counted for bound radioactivity. Five Mabs mapping different epitopes of the apoB molecule were used 1D1, 2D8, 3F5, 4G3, and 3A10, which react between residues 474 and 539, 1438 and 1480, 2835 and 2922, 2980 and 3084, and 3441 and 3569, respectively (11,12).

Other Analyses
The protein content of LDL was measured by Lowry's method (34) and in many instances by parallel determination of apoB by radioimmunoassay (36). The correlation between these two methods was very close with r = 0.97 ( n = 51). Triglyceride and total and unesterified cholesterol were analyzed by enzymatic kits (Boehringer Mannheim 877557, 236691, and 310378, respectively) and phospholipid by the Barlett method (37). The weight concentration of esterified cholesterol was estimated as that of the sterol moiety plus fatty acid as (total cholesterolfree cholesterol) X 1.67. The apoprotein composition of LDL was analyzed by one-dimensional SDS-polyacrylamide gel electrophoresis (23). Generally 5-10 pg of protein was applied to each lane. The electrophoresis was performed using a constant 150 voltage. The gels were stained with 0.1% Coomassie Brilliant Blue R and destained with a solution of 25% methanol, 10% acetic acid-distilled water (v/v). Lipid oxidation of LDL was assessed by the determination of thiobarbituric acid reacting substances (38) using 1,1,3,3 tetramethoxipropane (T-1642 Sigma) as standard (39). The total volume of the assay was 1.5 ml, and 50-100 pg of LDL protein was analyzed.
Statistical Analysis-Differences in the relative weight composition of different LDLs were evaluated for statistical significance using the two-tailed paired and nonpaired Student's t test. The EC, (concentration of ligand causing 50% binding inhibition) value of competitive studies on fibroblasts was obtained by linear transformation of the data according to the Hill equation (40). The EC, value for the Mabs reacting with apoB was determined by using the ALLFIT program (41). The ratio between EC, of experimental LDL and EC, of the appropriate control in each case was calculated, and the statistical significance was established by Student's t test. Linear transformation of LDL saturation binding data and & values were calculated by Scatchard plot analysis (40).

RESULTS
LDL Composition-Compared with N-LDL, LDL from subjects with chronic hypertriglyceridemia had relative enrichment in protein (1.4-fold) and triglyceride (%fold) along with a relative decrease in phospholipid and free and esterified cholesterol (20,30, and 25%, respectively) ( Table I). Since the amount of apoB protein per particle remains constant the changes in lipid are best expressed as LDL lipid to protein ratio. HTG-LDL compared with N-LDL exhibited a 2-fold enrichment in triglyceride, a 45% reduction of phospholipid, and a decrease in free and esterified cholesterol of 60 and 45%, respectively.   (Table 11). These compositional changes induced in vitro were similar to those observed in HTG-LDL. Compared with the native N-LDL used for the in vitro experiments, R-LDL after 8 h of incubation did not show changes in the relative weight composition of protein or phospholipid, but the triglyceride increased %fold with a reduction in both free and esterified cholesterol of about 20%. Prolonging the incubation to 24 h resulted in a minimal increase in the relative weight composition of protein, but the phospholipid content remained unchanged. However, the triglyceride increased &fold, and the free and esterified cholesterol decreased by 40 and 45%, respectively. The protein/phospholipid ratios remained similar after an 8or 24-h incubation. R-LDL that was incubated in the presence of triglyceride emulsion but in the absence of lipoprotein deficient d > 1.21 g/ml plasma only showed a decrease in the relative weight of free cholesterol by 40% after a 24-h incubation. N-LDL that was incubated in the presence of plasma d > 1.21 g/ml but in the absence of triglyceride emulsion (control R-LDL) did not have any significant compositional changes.
Compared with the composition of LDL from normal volunteers, at (time Oh ), LDL after 6 h of intravenous lipid infusion (IV-LDL) had an increase in triglyceride (1.5-fold) and phospholipid (1.4-f01d), whereas the content of free and esterified cholesterol was reduced by about 10% (Table 111).
No degradation of apoB or other changes in the apoprotein composition of N-LDL, HTG-LDL, R-LDL, or IV-LDL were detected by SDS-polyacrylamide gel electrophoresis. No lipid peroxidation of LDL was found by thiobarbituric acid reacting substance analysis (data not shown).
LDL Size-To evaluate whether the modifications in lipid composition of LDL were associated with changes in size and relative densities of the particles, the elution profiles of LDL isolated by nonequilibrium rate zonal ultracentrifugation were compared (Fig. 1). The main peak of HTG-LDL consistently appeared later in the rotor effluent than did N-LDL (mean f S.E. 228 f 6.2 ml and 180 f 4.2 ml, respectively, p < 0.05).
In contrast, R-LDL eluted slightly earlier (mean f S.E. 173 f 5.4 ml) than the N-LDL used for the in vitro remodeling of LDL (193 f 6.3 ml, p < 0.05). IV-LDL showed the same elution pattern as did preinfusion control LDL (187 f 6.4 uerszu 189 f 7.0 ml, respectively). The difference in elution volume suggests that triglyceride-rich HTG-LDL was smaller than N-LDL, whereas R-LDL with its increased triglyceride relative to cholesteryl ester was a more buoyant particle than

Relative weight compositions of R-LDL
Normal LDL (6 mg of LDL esterified cholesterol) was incubated at 37 'C with 20% Intralipid (60 mg of triglyceride) and 300-400 mg of protein of d > 1.21 g/ml plasma as a source of cholesteryl ester transfer protein, in a solution of 0.19 M NaC1, 1 mM EDTA, 100 kallikrein inhibitory units/ml aprotinin, 10 pg/ml leupeptin, 1 pg/ml pepstatin, pH 8.5, for 8

Relative weight composition of IV-LDL
LDL from seven normolipidemic subjects was isolated before (0 h) and immediately after 6 h of intravenous infusion of 10% Intralipid (time 6 h) as described under "Experimental Procedures." Infusion time Weight ratio of triglyceride/protein, free cholesterol/protein, esterified cholesterol/protein, and phospholipid/protein.
native LDL. IV-LDL did not show significant deviations in size or density when compared with N-LDL. This was confirmed by size analysis of different LDL particles using nondenaturing gradient gels ( d a t a not shown) and electron microscopy (Fig. 2). Only HTG-LDL showed a significant decrease in size, whereas R-LDL and IV-LDL were similar in size to N-LDL.
LDL Binding to the LDL Receptor-To determine whether differences in composition, size, or immunoreactivity of LDL correlate with the ability of LDL to bind to the LDL receptor, the displacement of normal '261-LDL by unlabeled experimental LDL and the saturation binding of LDL were evaluated on cultured human skin fibroblasts. In competition cell binding studies performed at 4 "C, HTG-LDL but not R-LDL or IV-LDL showed a significant decrease, about 4-fold lower, in binding to the LDL receptor, compared with control N-LDL (Fig. 3). The ratio of the ECso for experimental LDL to normal control LDL (in the same experiment) was 4.1 & 0.9 for HTG-LDL (n = 10, p < 0.05 versus N-LDL) but not different for R-LDL or IV-LDL (1.3 & 0.2 (n = 11) and 1.2 k 0.3 ( n = 3), respectively). Scatchard analysis (Fig. 4) also demonstrated that, compared with N-LDL, HTG-LDL but not R-LDL had about (-fold lower affinity for the LDL receptor (Kd) and decreased maximal binding capacity ( Bmm).
To examine whether small amounts of apoE not detected by 2-16% SDS-polyacrylamide gel electrophoresis might be affecting binding of HTG-LDL or N-LDL, the cell binding of 5 wg/ml '261-LDL and '261-HTG-LDL were measured at 4 "C, in the presence and absence of 15 pg/ml Mab ID7, which selectively inhibits the apoE binding to the LDL receptor (16). The presence of Mab ID7 did not change the differences in binding affinities between N-LDL and HTG-LDL (data not shown).
To determine if changes in binding affinity of HTG-LDL were also reflected in cellular LDL uptake and degradation, the degradation of 12'I-LDL was evaluated in competition studies at 37 "C. Fig. 5 shows that compared with control N-LDL, HTG-LDL but not R-LDL had decreased ability to compete for LDL degradation.
To investigate if there is a graded effect of HTG-LDL size upon HTG-LDL receptor affinity, different fractions of small and dense LDL were isolated from each of two hypertriglyceridemic donors by nonequilibrium zonal centrifugation, and then the ability of each fraction to displace normal '261-LDL from the LDL receptor on human skin fibroblasts was assessed at 4 "C. Compared with N-LDL, both fractions showed a decreased affinity for the LDL receptor, but the fractions containing the smallest and most dense LDL had the lowest binding affinities (Fig. 6). This is a consistent finding in any given subject. Note also that both fractions of LDL from subject A depicted in Fig. 6 had only slight triglyceride enrichment.
Circular Dichroism of LDL-To analyze if the changes in cell binding of HTG-LDL were associated with modifications in secondary structure of apoB, CD of the LDL particles was performed. As shown in Fig. 7A, compared with N-LDL, HTG-LDL had a distinct spectra and decreased percent contribution of a-helix. Seven of nine HTG-LDL fractions isolated from six hypertriglyceridemic individuals showed similar findings. The secondary structure of apoB in R-LDL remained unchanged despite the major changes in core lipid composition (Fig. 7B). ApoB Immunoreactivity with Monoclonal Antibodies-To ascertain if changes in size or composition of LDL were associated with alterations of apoB conformation at specific sites, the immunoreactivity of diverse epitopes across the apoB molecule was analyzed using a panel of monoclonal 515 EFFLUENT VOLUME(ml1 FIG. 1. Nonequilibrium rate zonal ultracentrifugation elution profiles of LDL. LDL was isolated from plasma d > 1.006 g/ ml, using a discontinuous NaBr gradient of 1.0-1.3 g/ml as described under "Experimental Procedures." Elution is from left to right with larger and lighter particles eluting earlier in the rotor effluent. In panels A-C the dotted line represents N-LDL, N-LDL before in vitro incubation with 20% Intralipid and plasma d > 1.21 g/ml, and from a normal subject before intravenous infusion of 10% Intralipid, respectively. The continuous line shows the elution of small/dense LDL from an HTG-LDL subject (panel A ) , an R-LDL subject after 24 h of incubation (panel B ) , and LDL isolated after 6 h of intravenous infusion of Intralipid (IV-LDL) in a normolipidemic volunteer (panel C ) . In panel C the small peak eluting before the main LDL peak in IV-LDL represents liposomal-like particles.
antibodies. Fig. 8A shows that compared with N-LDL, HTG-LDL from three hypertriglyceridemic subjects had a major decrease in the immunoreactivity of Mab 3F5, which identifies a region immediately adjacent to the LDL binding region of apoB (residues 2835-2922). Seven of eight HTG-LDL showed similar decreases in immunoreactivity at this epitope, and the ratio of the ECso (concentration of LDL causing 50% inhibition of antibody binding) of HTG-LDL to N-LDL was 2.5 f 0.5, p < 0.05. In two cases HTG-LDL also had increased immunoreactivity with Mab 3A10 (an epitope in the LDL receptor binding domain of apoB), but no significant differ- ences were found with other Mabs. In contrast, only minor or no changes were found when the immunoreactivity of R-LDL (n = 7-10) or IV-LDL (n = 6) was assessed with the same panel of apoB Mabs (e.g. Fig. 8, B and C ) . Thus, an increase in triglyceride content or a decrease in unesterified or esterified cholesterol had no significant effect on apoB conformation at the sites examined as judged by apoB immunoreactivity, unless such compositional changes were accompanied by a reduction in LDL size. Therefore, since enrichment of LDL with triglyceride alone is not sufficient to account for alterations in apoB structure, binding, and cell uptake via the LDL receptor, other factors, such as size, must be considered.

DISCUSSION
We analyzed the effects of both lipid composition and size upon apoB conformation and function in triglyceride-rich LDL. The results of our investigation provide evidence that the triglyceride content alone is not a major determinant of the overall structure of apoB and suggest that the size of LDL plays an important role in determining the conformation of the apoB molecule near its receptor recognition site and its affinity for the LDL receptor. In agreement with previous reports (5, 42)) we found that in contrast to N-LDL, HTG-LDL is predominantly smaller, denser, relatively enriched in triglyceride, and depleted in phospholipid and both free and esterified cholesterol. We also confirmed that HTG-LDL had a decreased affinity for the LDL receptor (17)(18)(19). This was accompanied by a decrease in total cell binding and maximal specific binding (Bmax) suggesting that a decrease in the HTG-LDL affinity for the LDL receptor is associated with partial occupancy of the cell receptors. Moreover, different fractions of small LDL also showed decreased cell binding, and the smallest fractions had even poorer LDL affinity for the LDL receptor. Small HTG-LDL consistently showed decreased immunoreactivity of Mab 3F5, which recognizes a boundary epitope (residues 2835-2922) to the binding region of apoB (11-12) and inhibits, by about 65%, the binding of LDL to the LDL receptor (10). In two subjects HTG-LDL also exhibited changes in affinity for Mab 3A10 which recognizes a distinct epitope of the putative binding region of apoB (10-12).
The changes in CD spectra of small HTG-LDL but not R-LDL demonstrate that these changes in the apoB receptor in conformation of apoB in small HTG-LDL might be masked.
Despite major changes in lipid composition, neither R-LDL nor IV-LDL showed significant or consistent modifications in size, apoB secondary structure, immunoreactivities, or binding to the LDL receptor. These results suggest first, that the relative content of neutral lipid in the core of LDL, in the absence of changes in the size of the particle, does not significantly modify apoB conformation or its affinity for the LDL receptor; second, that in the case of small LDL, there is a direct effect of LDL size upon its affinity for the LDL receptor; and third, that a decrease in free cholesterol or an increase in phospholipid in the LDL surface does not have important consequences on apoB configuration in its receptor binding domain or binding to the LDL receptor.
Notice that the spectrum of N-LDL was very similar to the spectrum of a different N-LDL shown in panel A and that the spectrum of the R-LDL was very similar (a-helix 41%, j3 sheet 23%, j3 turn 8%, random coil 28%) to the spectrum before in uitro incubation. The HTG-LDL, and 3 R-LDL.
results are representative of the CD spectra found in 10 N-LDL, 7 triglyceride lowering treatment which decreases LDL triglyceride content and also normalizes LDL size (43). It has also been described that triglyceride-rich LDL produced in vitro has reduced binding to the LDL receptor (17, 19, 21, 22). Differences in our in vitro results from those of others are likely because we used a triglyceride-phospholipid emulsion instead of VLDL as the triglyceride donor and esterified cholesterol acceptor particle. In our experience and that of others, incubating LDL in the presence of VLDL always results in enrichment of LDL with C apoproteins (17,23,26), which can decrease its cellular uptake (44,45).
Still, the LDL core composition likely exerts effects independent of LDL size on at least some regions of apoB.
Kunitake et d. (46) found that the immunoreactivity of apoB in triglyceride-rich LDL from subjects with Tangier disease differed from N-LDL. They described a decrease in immunoreactivity at an epitope between residues 1031 and 1084, an area of apoB not probed in our studies. However, no changes in conformation were detected in the LDL receptor recognition domain, and the cell binding of LDL was not reported. Similarly, decreased immunoreactivity of Mab MB3, which binds near residue 1022 of apoB, was reported in another study in triglyceride-rich LDL (22).
Although our findings suggest that LDL size may be more important than surface or core lipid composition in modifying apoB conformation and binding to the LDL receptor, other possibilities, such as the following, must also be considered.
1. Fatty acyl composition of LDL lipids in hypertriglyceridemic subjects may be different from that in normal or emulsion-modified LDL. This seems unlikely since other groups have not found changes in LDL receptor binding which can be related to modifications in lipid fatty acyl composition Under conditions very different from our experiments, maximal hydrolysis of LDL phospholipid to lysophosphatidylcholine produced by phospholipase A2 resulted in decreased binding to the LDL receptor and changes in apoB conformation (51), whereas LDL phospholipolysis with phospholipase D induced increased cell uptake by macrophages (52).
3. Since the group of hypertriglyceridemic individuals studied was genetically heterogeneous it is unlikely that a single genetic defect affecting both apoB secondary and tertiary structures can account for our findings. Nevertheless, genetic heterogeneity may certainly explain some of the variability we observed among the HTG-LDL, e.g. a decrease of LDL immunoreactivity with Mab 3A10 in some donors.
Evidence from other groups also supports the hypothesis that size, more than the core triglyceride/cholesteryl ester ratio, is a major factor determining apoB structure and function. Krieger et al. (53,54) showed that replacing the lipid core with either cholesteryl linoleate, trilinolein, or methylinoleate did not affect LDL precipitation by antibodies or its cell uptake. Small, dense fractions of cholesterol ester-poor LDL have been reported to have altered immunoreactivity to monoclonal antibodies recognizing epitopes near or in the LDL receptor binding domain (6,lO-12,55), and in one report this was associated with decreased binding and degradation of LDL in cultured cells (55). Nigon et al. (56) described that in LDL of normal composition, lighter (d 1.024-1.029 g/ml) and denser (d 1.035-1.043 g/ml) fractions of LDL had lower LDL receptor binding affinities than N-LDL of intermediate density (1.029-1035 g/ml). Recently, decreased binding of large LDL has been attributed to steric hindrance produced by the crowding of LDL particles on receptor lattices (57,58).
Some hypertriglyceridemic subjects also have large LDL as well as small LDL populations (5). We have found, in a preliminary study, that abnormally large triglyceride or esterified cholesterol-rich LDL particles also have decreased affinity for the LDL receptor (59). Thus, LDL may have an optimal size for maximum receptor affinity, LDL with a size that deviates substantially from this optimum will have decreased receptor affinity.
ApoB is a highly hydrophobic molecule, and the binding region to the LDL receptor presumably lies exposed on the surface-water interface of the LDL particle. Further, lysineand arginine-rich domains, such as the receptor binding region, appear to be particularly mobile (60). We hypothesize that the decreasing ratio of curvature in small LDL not only affects apoB structure but also forces changes in conformation upon specific plastic domains of the apoB molecule, e.g. the binding region to the LDL receptor, resulting in lower receptor affinity of small LDL.
It has been shown that small LDL may have an increased susceptibility to oxidation (49,50), a high affinity for the arterial proteoglycans (61), increased catabolism by nonreceptor-mediated pathway (62,63) and is associated with increased risk of myocardial infarction (64). Thus, LDL size may be important in modifying its metabolism by changing its affinity for the LDL receptor in certain tissues such as the liver and directing LDL toward other catabolic pathways, such as scavenger pathways in the arterial wall.