Presence of Phospholipid-Neutral Lipid Complex Structures in Atherosclerotic Lesions as Detected by a Novel Monoclonal Antibody*

A novel monoclonal antibody (ASH1a/256C) that recognizes atherosclerotic lesions in human and Watanabe heritable hyperlipidemic (WHHL) rabbit aortae is described. When 123I-labeled ASH1a/256C antibody is injected intravenously into WHHL rabbits, it associates specifically with fatty streaks on the aorta. The antigen recognized by the antibody is lipid, based on extraction with chloroform and methanol from WHHL rabbit tissues. The antigen, purified by high performance liquid chromatography, was shown to be phosphatidylcholine (PC), which contains unsaturated fatty acyl groups based on analyses utilizing1H and 13C nuclear magnetic resonance, Fourier transfer-infrared spectrum, and mass spectrometry. The antibody did not react with other classes of phospholipids or neutral lipids when tested using an enzyme-linked immunosorbent assay. When PC was mixed with either cholesterol, cholesteryl ester, or triacylglycerol, however, the reactivity of the antibody to PC increased up to 8-fold. Homogenates of aorta tissue obtained from normal and WHHL rabbits were fractionated using sucrose density gradient ultracentrifugation in which neutral lipid droplets, cellular membranes, and proteins are separated. The phospholipid content in cellular membrane fractions from WHHL rabbits was twice as high as that of normal rabbits, and there was an enormous difference in the antigenic activity in these fractions. The content of cholesterol in the cellular membrane fraction of WHHL rabbits was approximately 50 times higher than that of normal rabbits. Addition of neutral lipids to the cellular membrane fraction of normal rabbit markedly increased the antigenic activity. Atheromatous lesions in thickened WHHL rabbit aortic intima that were rich in lipid droplets were stained positively with ASH1a/256C immunohistochemically. These results strongly suggest that PC-neutral lipid complex domains are formed in atherosclerotic lesions.

Intracellular and extracellular accumulation of neutral lipids in the arterial intima is a typical feature of atherosclerotic lesions. In the early stages of atherosclerosis, foam cells that accumulate cholesteryl ester (CE) 1 droplets in their cytosol are formed from macrophages and smooth muscle cells (1)(2)(3). Several types of scavenger receptors, which are capable of binding and taking up modified low density lipoproteins (LDL), have been shown to play crucial roles in foam cell formation (4,5). In advanced lesions, neutral lipids are also accumulated in the extracellular space, and cholesterol crystals can form (1, 3, 6 -8). Neutral lipids may be deposited in the extracellular spaces when foam cells eventually die either by necrosis or apoptosis (9). However, little is known of the mechanisms of extracellular deposition of neutral lipids or the fate of foam cells. Furthermore, it is not known whether lipid accumulation affects cellular responses in the lesions.
Multiple factors are closely involved in the formation of these lesions, including lipoprotein metabolism, smooth muscle cell proliferation, endothelial cell malfunction, formation of modified LDLs, and accumulation of foam cells (4, 5, 10 -12). To establish useful tools for the investigation of the mechanisms of atherogenesis, a series of monoclonal antibodies using homogenates of human atheroma as immunogen has been raised. Through characterization of these anti-atheroma antibodies, the presence of vitronectin (13,14), oxidized phosphatidylcholine (PC) (15,16), and cross-linked proteins (17) in human and rabbit atherosclerotic lesions have been demonstrated.
In this study, a monoclonal antibody was selected that bound to fatty streaks using an in vitro artery wall binding assay. Strips of aorta from Watanabe heritable hyperlipidemic (WHHL) rabbits were incubated with hybridoma culture media followed by a 125 I-labeled second antibody. The antibodies that bound to the surface of fatty streak but not to the normal endothelium were selected. The monoclonal antibody ASH1a/ 256C (a murine monoclonal antibody against surface of human atheroma), which bound atherosclerotic lesions in vivo and immunohistochemically, recognized PC containing polyunsaturated fatty acyl groups (PUFA). The content of PC in atherosclerotic lesions was at most twice that of normals, although the antigenicity of the lesion homogenates was more than eight times higher than that of the normal aortae. The reactivity of this antibody to PC was greatly increased in the presence of neutral lipids, suggesting that certain complex structures of PC and neutral lipids are present in atherosclerotic lesions.

MATERIALS AND METHODS
Preparation of Monoclonal Antibody-Atherosclerotic areas of human abdominal aorta were cut into pieces and homogenized with a Polytron® homogenizer in SVE solution (0.25 M sucrose, 1 mM EDTA, 1% ethanol, pH 7.4). After centrifugation at 220 ϫ g for 10 min at 4°C, the supernatant was recovered and used as immunogen. BALB/c mice (8 weeks old, female) were immunized three times with the homogenate of human atheroma over a period of 3 months (18). Spleens were removed from the immunized mice 3 days after the final injection. The spleen cells were fused with the murine myeloma cell line P3/U1 using polyethylene glycol-4000 and cultured in HY medium (DMEM: NCTC109 medium ϭ 8:1 containing 1 mM sodium pyruvate, 5 g/ml insulin, 0.16 mg/ml oxaloacetate, 7% fetal calf serum) containing hypoxanthine, aminopterin, and thymidine (19). Antibody titers in the culture medium of hybridomas were tested by enzyme-linked immnosorbent assay (ELISA) and an in vitro binding assay to WHHL aorta. Hybridomas showing anti-atheroma reactivity were cloned by limiting dilution procedure twice.
To select anti-atheroma antibodies, homogenates of human atheroma, WHHL rabbit atheroma, and normal aorta obtained from control rabbits as well as human and rabbit sera were used as antigens for ELISA. For those clones that were positive to homogenates of human and rabbit atheroma and negative to the other antigens, immunohistochemical staining of frozen sections (4 -6 m) of WHHL rabbit aorta and human atheroma were performed. Then the in vitro binding assay to WHHL aorta was performed for the selected clones that stained atherosclerotic lesions immunohistochemically. Strips of WHHL rabbit aorta (4 ϫ 15 mm) were incubated with the culture medium of the selected hybridoma clones followed by 125 I-labeled goat anti-mouse Ig(GϩM) (New England Nuclear Co.). After rinsing the strips with phosphate-buffered saline (PBS) five times, autoradiography was performed. The ascites obtained from mice bearing P3 U1 myeloma, which was not hybridized with any cells, was used as control. One of the clones that showed positive spots corresponding to areas of fatty streaks was isolated and was named ASH1a/256C. The antibody produced by this clone was partially purified from ascites of mice bearing the hybridoma using ammonium sulfate precipitation. Its immunoglobulin class was IgM. During investigating this antibody, the hybridoma clone has been recloned five times so far, and no change in the reactivity of the antibody has been observed.
Purification of Antigen Recognized by ASH1a/256C-Aorta or kidney from WHHL rabbits were cut into pieces and homogenized using Polytron® homogenizer as described previously (14). After removed of cellular debris by centrifugation at 220 ϫ g for 10 min, the supernatant was collected. Lipids were extracted from the homogenate using the method of Bligh and Dyer (21). The lipid extracts were dried under an argon gas stream and then applied onto a silica gel column to separate phospholipids from neutral lipids. After washing the column with chloroform followed by chloroform:methanol (9:1) to remove neutral lipids, polar lipids including those with antigenic activity were eluted with chloroform:methanol:water (6:4:1). The eluate was then fractionated using straight phase high performance liquid chromatography (HPLC) (column: LiChrosorb Si60, 4 ϫ 250 mm, Merck, Germany) by gradient elution with hexane:2-propanol:water (44:55:1 to 33:55:12). The flow rate was 0.5 ml/min. The antigenic activity, which was eluted at 45 min, was separated completely from neutral lipids and glycolipids by this purification step. The antigen recovered from the HPLC was rechromatographed on the same column with another solvent system chloroform:methanol:water (4:5:1) at a flow rate of 0.2 ml/min. The antigenic activity was eluted as a single peak at 23 min.
Two-dimensional Thin Layer Chromatography-The purified antigen and PC standard (4 g each) were spotted onto a silica gel TLC plate. The plates were developed with hexane:diethylether (1:1) followed by chloroform:methanol:water (6:4:1) in the same direction. The plate was then developed in the direction perpendicular to the first run with chloroform:methanol:acetic acid:acetone:water (6:2:4:2:1). The samples were visualized by spraying molybdophospholic acid onto the plate (23).
Structural Analyses-Proton NMR spectra of the purified antigen (2.7 mg) and sn-1-palmitoyl-2-linoleoyl PC (2 mg) dissolved in (CD 3 ) 2 SO were obtained using a GSX-400 spectrometer (Jeol) with 512-pulse scanning at 400 MHz (24). Two-dimensional cross-relaxation spectra (NMR-COSY) were obtained using 256-pulse scanning at 400 MHz. Proton chemical shifts were indicated in ppm downfield from tetramethylsilane. 13 C NMR spectrum was obtained using 61,440-pulse scanning at 100 MHz using the same spectrometer. Carbon chemical shifts were indicated in ppm with reference to the internal solvent (CD 3 ) 2 SO. Fourier transfer infrared spectra of the antigen were obtained using a Fourier transfer infrared spectrum 8000 spectrometer (Jasco, Japan). Fast atom bombardment mass spectrometry of the antigen were obtained using JMS-SX102 A (Jeol), triethanolamine as the matrix. Liquid chromatography-linked mass spectra of the antigen were examined using JMS-LX2000 spectrometer (Jeol) with a Hiber LiChroCART RP-18 column (4 ϫ 250 mm; 7 m; Merck) under the same conditions as described above.
Measurement of Antigenic Activity-Reactivity of ASH1a/256C to various materials was determined by ELISA. Aquaous samples, such as homogenates of atheroma, were coated onto 96-well microtiter plates (Falcon number 3912) that had been pretreated with 2% glutaraldehyde for 2 h. After incubating the plates at 37°C for 1 h, the surfaces of the microtiter wells were blocked by incubating with Tris-buffered saline containing 2% skimmed milk. The plates were incubated with ASH1a/ 256C antibody diluted with Tris-buffered saline containing 2% skimmed milk followed by alkaline phosphatase-conjugated goat antimurine Ig(GϩM) antibody (Tago Inc., AMI3705). After washing extensively with Tris-buffered saline containing 0.05% Tween 20, the plates were incubated with p-nitrophenylphosphate (1 mg/ml) dissolved in 1 M diethanolamine-HCl buffer, pH 9.8 at 37°C for the appropriate time periods. The absorbance at 405 nm was measured photometrically using an ELISA plate reader (Bio-Rad).
When the antigenic activities of the lipids were tested, their methanol solutions were placed into microtiter wells without the pretreatment with glutaraldehyde. The plates were incubated at 37°C for 5-10 min to remove the methanol, after which the surfaces of the microtiter wells were blocked with 0.3 M sucrose.
Density Gradient Ultracentrifugation of Atheromatous Lipids-Homogenates of atherosclerotic lesions from WHHL and normal rabbit aorta (6 mg of protein) were fractionated using a sucrose density gradient ultracentrifugation according to the method described previously (20). Briefly, a linear gradient of SVE solutions containing 53 to 0% sucrose was layered on top of the homogenates containing 64% sucrose. After centrifugation at 89,000 ϫ g for 75 min at 4°C using RPS-27 rotor (Hitachi), samples were collected from each ml from the bottom to the top of the gradient.
Histochemical Study of WHHL Rabbit Aorta-Frozen sections (4 -6 m) of WHHL rabbit fatty streaks were obtained and fixed with 10% neutral formalin immediately after autopsy. The sections were incubated with ASH1a/256C (ascites) followed by fluorescein isothiocyanate-conjugated goat anti-mouse Ig(GϩM) (Organin Teknika Corp., Durham, NC). The adjacent WHHL section was stained with 0.1% oil-red O in 60% 2-propanol for 10 min, and the section was counter stained with Mayer's hematoxylin for 5 min after washing off any excess oil-red O with 2-propanol.
Other Analytical Methods-The amounts of total cholesterol were measured by a cholesterol oxidase method using the Cholestase-V kit (Nissui, Co.) (25,26). Levels of phospholipids were determined by measuring phosphorus in organic extracts using malachite green according to the method of Zhou and Arther (27). Protein concentrations were measured by the Bradford method using the Bio-Rad protein assay kit with BSA as the standard (28).

A New Antibody That Binds Specifically to Atheromatous
Lesions-In an attempt to obtain monoclonal antibodies against atherosclerotic lesions, hybridoma clones from mice immunized with homogenates of fatty streak lesions of human atheroma were prepared. Anti-atheroma clones were selected by ELISA using homogenates of atheroma from humans and WHHL rabbits for initial screening, followed by immunohistochemical staining using frozen sections of WHHL rabbit aorta.
Then candidate clones were further tested using a binding assay to WHHL rabbit aorta strips. Clones reactive to materials in human and rabbit sera proteins were omitted. A clone was finally established after these selections and was named ASH1a/256C (atheroma, surface, human). 123 I-Labeled ASH1a/256C antibody was injected intravenously into normal and WHHL rabbits. These rabbits were sacrificed 48 h after injection, and the distribution of the labeled antibody in isolated aortas was visualized by autoradiography (Fig. 1). Fatty streaks were observed in the WHHL rabbit aorta but not in the normal rabbit aorta. Lesion formation was prominent in the aortic arch and at the points of vessel branching. The radioactivity was co-localized with the atherosclerotic plaques in the WHHL rabbit aorta. In contrast, the area that was free of visible lesions in the WHHL aorta and the aorta from normal rabbit were negative. The antibody reacted strongly to atheromatous homogenates from human and WHHL rabbits but did not react to homogenates from normal rabbits (Fig. 2). Furthermore, this antibody also bound to atheromatous lesions in WHHL rabbit aortae as shown by an in vitro binding assay (see "Materials and Methods"). These results show that this monoclonal antibody recognizes atherosclerotic lesions both in vivo and in vitro.
Antigen Purification-The antigen of ASH1a/256C was effectively extracted with chloroform and methanol from homogenates of rabbit aorta with the residual fractions having no antigenicity, suggesting that the antigen is likely to be lipid (Fig. 2). The reactivity of the antibody to the lipid extracts from the WHHL rabbit aorta was 8-fold greater than that from the same amount (10 g of protein) of normal rabbit aorta homogenate. When the same amount of phospholipid extracted from either the WHHL aorta or normal rabbit aorta was used as antigen, the antigenic activity in WHHL extract was 3.9-fold higher than the extract from normal rabbit by phospholipid basis (data not shown).
When the antigenicity of homogenates of several tissues to ASH1a/256C was examined by ELISA, kidney and xanthoma as well as aorta from WHHL rabbits showed strong activities (data not shown). The lipid extracts obtained from atheroma and kidney of WHHL rabbits were fractionated by silica gel column chromatography followed by HPLC. The identity of the antigens obtained from atheroma and kidney was investigated by the following experiments. First, when the partially purified antigens were analyzed by TLC immunostaining using ASH1a/ 256C, both samples showed single bands with the same Rf values (Fig. 3). Second, the antigens from these tissues were eluted from the HPLC column with the same retention times were coated onto microtiter plates. Then ASH1a/256C was added to the plates to carry out the ELISA assay as described under "Materials and Methods" to measure their antigenicity.
(data not shown). Finally, the same molecular mass numbers were obtained for these antigens by liquid chromatographymass spectrometry analysis (data not shown). Therfore, the antigens in atheroma and kidney could be identical.
The antigen was purified from both the aorta and kidney of WHHL rabbits. The antigen purified from the kidney was used to perform structural analyses (see below), because the quantity of the antigen purified from rabbit aorta was very limited. Fig. 4 shows data from the antigen purification from kidney, and the profiles were almost the same as those of aorta. Initially, a step-wise elution from a silica gel column was performed to remove large amount of neutral lipids (Fig. 4A). The antigen was eluted in fraction III (chloroform:methanol:water, 6:4:1), whereas fractions I and II had no reactivity. Triacylglycerol and CE were mostly recovered in fraction I (data not shown).
Fraction III was then applied to a straight phase HPLC with a gradient elution using hexane:2-propanol:water (44:55:1 to 33:55:12). The antigenic activity was eluted at 44 min as a single peak (Fig. 4B). This fraction was further purified on the same HPLC column using a different solvent system (Fig. 4C). The purified antigen, which was eluted at 23 min as a single peak by the second HPLC, showed a single spot on two-dimensional TLC (data not shown).
Structural Analyses of the Antigen-The antigen purified from WHHL rabbit kidney underwent a number of structural analyses. No signal corresponding to either ketone, aldehyde, acid anhydride, or free carboxylic acid was observed by Fourier transfer infrared spectrum of the antigen; however, the spectrum did suggest the presence of ester bonds (CϭO; 1735 cm Ϫ1 ) (data not shown). The presence of two ester bonds (CϭO; 172 ppm) was confirmed by a 13 C NMR spectrum (Fig. 5A). Signals corresponding to two CϭC double bonds (127 and 129 ppm) were also observed in the 13 C NMR spectrum.
Furthermore, one-dimensional and two-dimensional NMR analysis (NMR-COSY) of the antigen was performed to identify its molecular structure. The signals marked in alphabets in the 1 H NMR spectrum of the antigen were identified as described in the legend of Fig. 5. The spectrum of the antigen was found to be very similar to that of sn-1-palmitoyl-2-linoleoyl PC (Fig.  5, C and D). One particular signal (d ϭ 3.1 ppm; 9H, marked with asterisks in Fig. 5 (C and D), corresponds to signal e) did not interact with any other signal, suggesting that there is no proton in close proximity to the nine hydrogen atoms in the FIG. 3. Immunological identity of the antigens from atherosclerotic aorta and kidney of WHHL rabbits. Partially purified antigens were prepared as described under "Materials and Methods" and the legend of Fig. 4. The antigens from atherosclerotic aorta (lanes 1 and 2) and kidney (lanes 3 and 4) of WHHL rabbits were developed on a TLC plate with hexane:diethylether (1:1) followed by chloroform: methanol:water (6:4:1). Then the antigen was detected with ASH1a/ 256C as described under "Materials and Methods. "   FIG. 4. Purification of the ASH1a/256C antigen. A, the lipid extract from WHHL rabbit kidney was applied to a silica gel column (bed volume, 15 ml), which was equilibrated with chloroform. The sample was eluted with chloroform (fraction I), then with chloroform:methanol (9:1) (fraction II), and finally with chloroform:methanol:water (6:4:1) (fraction III). Eluate was collected from each 10 ml. Neutral lipids were mostly eluted in fraction I. Antigenic activity (closed circles) was measured by ELISA. B, fraction III recovered in A was applied onto a straight phase HPLC column (first separation). The chromatography was carried out as a gradient elution with the following solvent system: hexane:2-propanol:water (44:55:1 to 33:55:12). The flow rate was 0.5 ml/min. The eluate was collected each minute. C, The partially purified antigen recovered in B was then applied to the same silica gel HPLC column as B but eluted isocratically with the solvent system chloroform: methanol:water (4:5:1). The flow rate was 0.2 ml/min. The eluate was collected each minute.
FIG . 5. NMR analyses of the purified antigen. A, 13 C NMR spectrum of the purified antigen in (CD 3 ) 2 SO. B, 1 H NMR spectrum of the purified antigen in (CD 3 ) 2 SO. C and D, two-dimensional cross-relaxation spectra (NMR-COSY) of the purified antigen (2.7 mg) (C) and authentic sn-1-palmitoyl-2-linoleoyl PC (D) in (CD 3 ) 2 SO are shown. The signals at 3.1 ppm (asterisk) that do not interact any other peak were identified as antigen, as is the case with the N-trimethylamino group of the authentic sn-1-palmitoyl-2-linoleoyl PC. These results strongly suggest that the antigen is PC.
Analyses of the antigen by fast atom bombardment mass spectrometry showed several peaks ranging from m/z ϭ 756 -808. One of the peaks (m/z ϭ 758) corresponds to palmitoyllinoleoyl PC. The molecular species of the antigenic PC were separated by reverse phase HPLC (Fig. 6). Several antigenic peaks appeared, and a major antigenic peak at 21 min and a large peak at 29 min were identified by liquid chromatographymass spectrometry as palmitoyl-linoleoyl PC and stearoyl-linoleoyl PC, respectively. It appears that the antigenic PC consists of several molecular species with different combinations of fatty acids. The possibility that certain compounds other than PC are present in the purified antigen is very unlikely for two reasons: first, the antigen was purified to homogeneity by twodimensional TLC by which most of the phospholipid classes were separated, and, second, all of the signals (apart from one corresponding to the N-trimethylamino group in the NMR-COSY analysis) interacted with other signals. Therefore all of the signals were related to one structure. These results confirm that the monoclonal antibody ASH1a/256C recognizes PC molecules containing PUFA.
Specificity of the Antigen Recognition-To investigate specificity of ASH1a/256C to recognize PC, reactivity of the antibody to various phospholipids, neutral lipids, and PC-related com-pounds was examined by ELISA (Table I, experiment 1). The antibody did not react to phosphatidylethanolamine, monomethyl phosphatidylethanolamine, or dimethyl phosphatidylethanolamine, indicating that the binding was specific for the choline-containing head group. Because these three phospholipids were prepared from egg PC by a head exchange reaction, their fatty acid compositions are essentially the same (palmitic acid, 50%; oleic acid, 25%, palmitoleic acid; linoleic acid, 16%; stearic acid, 8%). Other phospholipids such as phosphatidylserine and phosphatidylinositol had no reactivity with the antibody. All the neutral lipids tested were also negative. Platelet-activating factor and sphingomyelin were not antigenic, although they share the choline head group. It seems that not only the choline head group but certain combinations of acyl groups are necessary for antigen recognition.
Reactivity of the antibody to various molecular species of PC was examined using chemically synthesized PCs (Table I,  recognition, but the number of PUFAs is not the determinant of the specificity of the antibody. When PC containing PUFA are incubated with metal ion, various peroxidation products including 9-CHO PC and 5-CHO PC are formed (16). ASH1a/256C failed to bind with the aldehyde-contianing oxidized products of PC (Table I, experiment 3). Preincubation of the antibody solution with 9-CHO PC, 5-CHO PC, egg lysoPC, or platelet-activating factor did not decrease the reactivity of the antibody to bind with 1-stearoyl-2-linleoyl PC (data not shown). sn-1-Stearoyl-2-linoleoyl PC was incubated with ferrous ion and ascorbate, and the change of the antigenicity of the PC during the peroxidation reaction was determined (Table II). The antigenicity for FOH1a/DLH3, which recognizes oxidized PC, appeared strongly after 3 h of oxidation, whereas the reactivity of ASH1a/256C to the PC decreased. These results indicate that the antibody does not bind with oxidized products of PC.
The Effect of Neutral Lipids on the Antigenicity of PC-As mentioned above, the antigenic activity was effectively extracted with chloroform and methanol from homogenates from rabbit tissues. Recovery of the antigenic activity was, however, reduced significantly during the purification of the antigen. The final yield of antigen activity was approximately 6.6%. It is noteworthy that the specific activity of the antigen was normalized by the amount of phosphorus decreased during the purification. A possibility to be considered is that there may be activators of antigen-antibody interaction in the homogenates. One of the major characteristics of atherosclerotic lesions is accumulation of neutral lipids; to see whether neutral lipids enhance the antigenicity of PC, the reactivity of ASH1a/256C to PC in the presence of neutral lipids was measured using ELISA. Addition of cholesterol, CE, or triacylglycerol markedly enhanced its reactivity to sn-1-stearoyl-2-linoleoyl PC (Fig. 7), whereas the neutral lipids themselves were not reactive to the antibody (Table I). These results show that neutral lipids are capable of increasing the binding of the antibody to PC. LDL, a huge particle containing phospholipids, neutral lipids, and apolipoprotein B, was not found to be a good antigen. When human LDL, copper-oxidized LDL, or high density lipoprotein were coated onto microtiter plates, no reactivity was observed with the antibody ASH1a/256C (data not shown).
The lipid droplets in aorta homogenates were separated from cellular membranes and proteins by sucrose density gradient ultracentrifugation. As shown in Fig. 8B, the antigenic activity in WHHL rabbit atheroma separated into two peaks, the top fractions and the middle fractions. These fractions correspond to lipid droplets and cellular membranes, respectively. The distribution of the antigenic activity corresponds to the  ELISA was performed on antigens (6.5 nmol each/well) as described under "Materials and Methods." The results are expressed as relative reactivity to egg PC (experiment 1) or to sn-1-linoleoyl-2-stearoyl PC (experiment 2). The absorbance obtained for egg PC (experiment 1) and sn-1-linoleoyl-2-stearoyl PC (experiments 2 and 3) were 0.96, 1.08, and 1.08, respectively.   II Effect of peroxidation of PC on its antigenicity to ASH1a/256C sn-1-Stearoyl-2-linoleoyl PC (0.4 mM) was incubated with ferrous sulfate (40 M) and ascorbate (0.4 mM) in PBS at 37°C for indicated periods. Oxidized lipids extracted from the reaction mixture were mixed with BSA and then placed onto microtiter wells as antigen (6.5 nmol PC/well). ELISA was performed using ASH1a/256C and FOH1a/DLH3 as described under "Materials and Methods." The results are expressed as relative reactivity to the highest values obtained by these antibodies. The absorbance obtained with ASH1a/256C (0 min) and FOH1a/DLH3 (3 h) were 0.685 and 0.579, respectively. Note that the reaction with ASH1a/256C was less effective under this experimental condition than the data in Table I, because the antigen suspended in PBS as lipid-BSA mixture was coated onto micrototier wells without glutaraldehydepretreatment. The decrease in the ELISA reaction of ASH1a/256C during the oxidation of PC was equally observed under the other experimental conditions. amounts of both phospholipid and cholesterol. In the case of normal rabbit aorta, there was no antigenic activity, although phospholipids were localized in fraction 7. The cholesterol content in fraction 7 was about 1 ⁄100 of that of the corresponding fraction of WHHL rabbit (Fig. 8A). These results suggest that antigenicity in rabbit aorta is greatly affected by cholesterol accumulation in the tissue.
To confirm the effect of cholesterol on the reactivity of PC in atherosclerotic lesions, an aliquot of cholesterol was added to each fraction obtained from normal rabbit aorta by sucrose density gradient centrifugation (Fig. 9). The ASH1a/256C antibody strongly reacted to the top and middle fractions following the addition of cholesterol, especially to fraction 6, which contained cellular membrane phospholipids. Similar enhanced antigenicity was also observed by addition of either cholesteryl oleate or triolein (data not shown). These results show that addition of neutral lipids to normal vessel wall increases the antigenicity of PC as observed in WHHL rabbit atheroma.
Immunohistochemical Analysis-Serial sections of WHHL rabbit atherosclerotic aorta were stained with ASH1a/256C and oil-red O to study the localization of antigenic PCs and lipid deposits. Large intracellular lipid droplets related to foam cells and small lipid droplets in extracellular matrix were observed when stained with oil-red O (Fig. 10A). ASH1a/256C stained the area where small lipid droplets were profusely deposited (Fig. 10B), whereas the antibody did not stain the endothelium and the media. These immunohistochemical observations, together with the other results, strongly suggest that the ASH1a/256C antibody does not recognize normal cellular membranes but rather that certain structures of PCs complexed with neutral lipids formed in atherosclerotic lesions. DISCUSSION This paper describes the preparation of a novel monoclonal antibody that recognizes fatty streaks in human atherosclerotic FIG. 8. Separation of the antigenic materials in atherosclerotic aorta by sucrose density gradient centrifugation. Homogenates of aortas from WHHL and normal rabbits were fractionated using sucrose density gradient centrifugation. The antigenic activity in each fraction was measured by ELISA (horizontal bars). Amounts of phospholipids (open circles) and total cholesterol (closed circles) were measured following lipid extraction with chloroform and methanol. aorta. This antibody was selected by reactivity to homogenates of atheroma using ELISA and to atheromatous plaques in aortic strips using an in vitro binding assay. The antibody also recognized atherosclerotic lesions of WHHL aorta in vivo.
The antigen is a lipophilic compound, based on the effective extraction from WHHL rabbit aortae by use of organic solvents. The antigen was purified by repetitive HPLC to a single spot on two-dimensional TLC. From extensive spectrometric analyses the purified antigen was identified as PC. Other phospholipids and neutral lipids were inactive. By reverse phase HPLC, the purified antigen was shown to contain several antigenic molecular species of PC. One major antigenic species was confirmed to be sn-1-palmitoyl-2-linoleoyl PC, by comparison with authentic PC and by use of liquid chromatography-mass spectrometry. Judging from the reactivity of the antibody to various molecular species of PC and PC analogs, it was concluded that the choline head group is necessary for antigen recognition and that at least one PUFA is also required.
It is intriguing that the monoclonal antibody that recognizes PC binds to atherosclerotic lesions in in vivo and in vitro binding assays, despite PC, a major component of cellular membranes, having a ubiquitous distribution in whole animal tissues. It is possible that the microenvironments of PC molecules in normal aorta and atherosclerotic lesions are different. The current data indicate that PC mixed with neutral lipids such as cholesterol was highly reactive with the antibody, although the neutral lipids themselves were not antigenic. Fractionation of aortic homogenates by density gradient centrifugation showed that fractions rich in both phospholipids and neutral lipids were antigenic, and, furthermore, addition of neutral lipids to the PC-rich fraction from normal aorta markedly increased its antigenicity. From these observations, it is proposed that the monoclonal antibody ASH1a/256C is likely to recognize particular conformations or packing structures of PC molecules that are formed in the presence of high concentrations of neutral lipids.
In atherosclerotic lesions there are a number of foam cells that accumulate neutral lipids as cytoplasmic and lysosomal droplets (1)(2)(3). Immunohistochemical studies showed that the ASH1a/256C antigen present in atherosclerotic lesions of WHHL aorta preferentially found in areas rich in small lipid droplets but not in areas rich in oil-red O-positive foam cells, suggesting that the lipid droplets in foam cells are not putative antigenic PC-neutral lipid complexes. It is known that lipid droplets in the extracellular space are smaller in size than those in foam cells (8,29,30). Smaller lipid droplets contain mainly free cholesterol rather than CE (31)(32)(33)(34)(35), whereas intracellular lipid droplets consist mainly of cholesteryl oleate, which forms liquid crystal structures (6). It is possible that the lipids accumulated in the extracellular space may form certain types of phospholipid-neutral lipid mixed structures. Chao et al. (30) reported that in rabbit atherosclerotic lesions the lipid droplets deposited in the extracellular space were enriched with cholesterol and sphingomyelin. The small lipid droplets accumulated extracellularly may be liposome-like mutilamellar vesicles consisting of phospholipids and unesterified cholesterol (31,32).
It has been shown that cell death either by necrosis or by apoptosis is frequently seen in atherosclerotic lesions (9,33,34). Lysosomal hydrolysis of CE in foam cells during the development of atherosclerosis increases the intracellular free cholesterol:phospholipid ratio, which causes damage to the cells (35)(36)(37) When lipid-laden foam cells die during necrosis, the cytosolic lipid droplets are released into extracellular spaces. Lipid droplets may interact with phospholipids derived from fragmented membranes to form a new complex structure. In the extracellular space, the molar ratio between free cholesterol and phospholipids changes during the development of atherosclerosis (37). It is thought that free cholesterol-derived cell death may produce extracellular deposits of lipid droplets that are rich in free cholesterol. When foam cells derived from J774 murine macrophages in culture were maintained for a week, the cells that eventually died left traces of cellular materials such as fragmented membranes, attached focal adhesions, and small lipid droplets. The scenario suggested above is also supported by our recent experiments showing that ASH1a/ 256C reacts to the small lipid droplets left after the foam cells die in culture. 2 The possibility that antigenic-PC neutral lipids form complexes without being accumulated in macrophages and smooth muscle cells cannot be ruled out. From a series of extensive electromicroscopic studies Guyton and co-workers (2,3,7,8,29) proposed that free cholesterol-rich particles in the extracellular space could be formed without prior accumulation of lipids in foam cells. This group has shown that extracellular lipid vesicles accumulate in early lesions prior to the appearance of lipid-laden foam cells.
A number of monoclonal antibodies that recognize atherosclerotic materials have been prepared by many investigators; however, few of them have succeeded in identifying their antigenic materials. An anti-oxidized LDL monoclonal antibody, FOH1a/DLH3, that recognizes foam cells has previously obtained (15). Its antigen was identified as oxidized products of PC including 9-CHO PC (16). The specificity of ASH1a/256C is clearly different from that of FOH1a/DLH3. The former does not bind to OxPC or oxidized LDL, and the latter does not recognize native PC species. Another monoclonal antibody recognizing atherosclerotic lesions prepared in a previous study, EMR1a/212D, specifically stained extracellular regions of atheroscletic intima from WHHL rabbits in immunohistochemical studies (18). The antibody was shown to recognize rabbit vitronectin (13), and, using this antibody, accumulation of subtypes of vitronectin with small molecular masses was demonstrated (14).
The antibody reported in the present study is unique in that it represents unusual structures of common lipid complexes. Further study is needed to understand the physical properties of the putative antigenic PC-neutral lipid complex in the lesions. Finally, this antibody can bind to atherosclerotic lesions in vivo, thus applications for immuno-diagnosis and drug delivery systems may be possible in the future.