Evidence for the presence of lipid-free monomolecular apolipoprotein A-1 in plasma.

The first step in reverse cholesterol transport is a process by which lipid-free or lipid-poor apoA-1 removes cholesterol from cells through the action of ATP binding cassette transporter A1 at the plasma membrane. However the structure and composition of lipid-free or -poor apoA-1 in plasma remains obscure. We previously obtained a monoclonal antibody (MAb) that specifically recognizes apoA-1 in preβ1-HDL, the smallest apoA-1-containing particle in plasma, which we used to establish a preβ1-HDL ELISA. Here, we purified preβ1-HDL from fresh normal plasma using said antibody, and analyzed the composition and structure. ApoA-1 was detected, but neither phospholipid nor cholesterol were detected in the purified preβ1-HDL. Only globular, not discoidal, particles were observed by electron microscopy. In nondenaturing PAGE, no difference in the mobility was observed between the purified preβ1-HDL and original plasma preβ1-HDL, or between the preβ1-HDL and lipid-free apoA-1 prepared by delipidating HDL. In sandwich ELISA using two anti-preβ1-HDL MAbs, reactivity with intact plasma preβ1-HDL was observed in ELISA using two MAbs with distinct epitopes but no reactivity was observed in ELISA using a single MAb, and the same phenomenon was observed with monomolecular lipid-free apoA-1. These results suggest that plasma preβ1-HDL is lipid-free monomolecular apoA-1.

peptide was confi rmed to be the same as the theoretical molecular mass by MALDI-MS, and the purity was confi rmed to be more than 95% by HPLC. Cyanogen bromide (CNBr) fragments of apoA-1 were prepared using the procedure reported by Morrison et al. ( 28 ) as follows. Lipid-free apoA-1 (0.1 mg) dialyzed with purifi ed water was lyophilized and dissolved in 0.2 ml of 70% trifl uoroacetic acid (TFA). CNBr dissolved in 0.2 ml of 70% TFA to a concentration of 1 mg/ml was added to the apoA-1 solution; the bottle was suffused with nitrogen gas and sealed, followed by incubation at room temperature for 24 h in the dark. After incubation, 3 ml of purifi ed water was added and lyophilized. Each fragment was identifi ed by molecular mass using SDS-PAGE and N-terminal amino acid analysis. MAb55201 was obtained by our group as described previously ( 21 ). The mouse anti-infl uenza A MAb obtained by our group was used as the control IgG. MAb55201 or the control IgG-bound sepharose column was prepared by using CNBr-activated Sepharose 4B (GE Healthcare) in accordance with the procedure recommended by the manufacturer. Polyacrylamide slab gels used for nondenaturing PAGE or SDS-PAGE, and the protein transfer kits used for Western blotting were purchased from Cosmo Bio Co., Ltd. (Tokyo). DTNB used for the LCAT-dependent conversion experiment was purchased from Sigma-Aldrich, and dissolved with the phosphate buffer [0.1 mol/l KH 2 PO 4 -Na 2 HPO 4 (pH 7.4)] just prior to use, followed by addition to plasma or pre ␤ 1-HDL-depleted plasma to a fi nal concentration of 2 mmol/l. Goat anti-apoA-1 polyclonal antibody (PAb) and goat anti-apoA-2 PAb were prepared as described previously ( 21 ). Biotin-labeled antibodies were prepared with EZ-Link Sulfo-NHS-LC-Biotin (Thermo Fisher Scientifi c, Inc.).

Purifi cation of plasma pre ␤ 1-HDL
Plasma (40 ml) was passed through the control IgG-bound Sepharose column (5 ml) to remove fractions nonspecifi cally binding to mouse IgG or the Sepharose beads. The eluent (40 ml) was then passed through the MAb55201-bound Sepharose column (2 ml) to bind pre ␤ 1-HDL. The unbound fraction was used as the pre ␤ 1-HDL-depleted plasma for the experiments described later. After washing the MAb55201-bound Sepharose column with 30 ml of PBS, the adsorbed fraction was eluted with 10 ml of 0.1 mol/l glycine-HCl buffer (pH 2.0) containing 0.15 mol/l NaCl, and collected in 1 ml fractions. The absorbance at 280 nm of each fraction was measured, and the fractions (Fr. 4-6) having absorbance of more than 0.4 were immediately pooled and applied to a Superdex200 16/60 column (GE Healthcare) for further separation by gel fi ltration chromatography. The chromatography was run with 0.15 mol/l NaCl solution at a fl ow rate of 1.0 ml/min and the eluent was collected in 1.0 ml fractions by a fraction collector. The pre ␤ 1-HDL concentration of each fraction was determined by ELISA using MAb55201 (pre ␤ 1-HDL-ELISA) ( 21 ). Those fractions containing a high concentration of pre ␤ 1-HDL were then collected. The purifi cation process was performed four times, once each for the plasma samples obtained from the four healthy subjects, and all operations were performed at 4°C. The purity of pre ␤ 1-HDL in the purifi ed sample was confi rmed by nondenaturing 2D electrophoresis, and the concentration of pre ␤ 1-HDL was determined by pre ␤ 1-HDL-ELISA. In order to examine the possibility that MAb55201 bound to Sepharose beads may strip apoA-1 from HDL, we carried out affi nity chromatography under the same conditions at 1/10 scale of the above using HDL (1.063 g/ml < d < 1.21 g/ml) prepared by ultracentrifugation instead of plasma. Two milliliters of HDL (140 g/ml of cholesterol) was applied to a MAb55201-bound Sepharose column (0.2 ml). After washing the column with 3 ml of PBS, the adsorbed fraction was eluted with 1 ml of glycine-HCl for analysis cannot be easily purifi ed from plasma. The concentration of plasma pre ␤ 1-HDL is much lower than that of other HDL subfractions, and pre ␤ 1-HDL is very unstable in plasma ( 20 ). We previously obtained a monoclonal antibody (MAb) 55201 (MAb55201) that specifically recognizes apoA-1 in pre ␤ 1-HDL and developed an ELISA for easily measuring plasma pre ␤ 1-HDL concentrations using the MAb ( 21 ). Recently a variety of research results on pre ␤ 1-HDL using the ELISA were reported, in particular the relationships between plasma pre ␤ 1-HDL concentrations and coronary artery disease and medications have been noted (22)(23)(24)(25)(26).
The aim of the present study is to clarify the composition and structure of plasma pre ␤ 1-HDL. The MAb55201 we obtained reacts with apoA-1 in plasma pre ␤ 1-HDL and delipidated apoA-1 (lipid-free apoA-1), and does not react with apoA-1 in other HDL sub-fractions ( 21 ). We fi rst purifi ed pre ␤ 1-HDL from fresh plasma of healthy subjects with affi nity chromatography using MAb55201, and with gel fi ltration chromatography to analyze the protein and lipid composition and to examine its shape under electron microscopy. We then analyzed intact plasma pre ␤ 1-HDL and lipid-free apoA-1 by nondenaturing PAGE, nondenaturing 2D electrophoresis, and size-exclusion chromatography, and examined whether the concentration of lipid-free apoA-1 added to pre ␤ 1-HDL-depleted plasma decreases in an LCAT-dependent conversion into ␣ -HDL, which is a phenomenon unique to plasma pre ␤ 1-HDL, for comparison of intact plasma pre ␤ 1-HDL and lipid-free apoA-1. In addition, we obtained another anti-pre ␤ 1-HDL MAb (MAb55205), which recognizes a different epitope on apoA-1 from that of MAb55201, and investigated whether the apoA-1 in pre ␤ 1-HDL is a monomer or dimer by an immunochemical procedure using the anti-pre ␤ 1-HDL antibodies.

Materials
Blood was obtained on the day experiments were performed from healthy volunteers (employees of Sekisui Medical Co., Ltd.) having a normal plasma lipid profi le (total cholesterol, <220 mg/ dl; triglyceride, <150 mg/dl; and HDL-cholesterol, >40 mg/dl). Written informed consent was obtained from all volunteers. The subjects fasted overnight, and venous blood was drawn into vacuum blood collection tubes containing EDTA-2Na (Terumo Corporation, Tokyo, Japan). The blood samples were immediately chilled in ice water for 30 min and centrifuged at 2°C, 3,000 rpm for 30 min to separate the plasma. Lipid-free apoA-1 was prepared by delipidating the HDL fraction (1.063 g/ml < d < 1.21 g/ml) and subsequent purifi cation as described previously ( 21 ). Lipidfree apoA-1 is known to self-associate to form dimers and larger multimers, depending on various conditions that include apoA-1 concentration and ionic strength in the solution ( 27 ). The purifi ed lipid-free apoA-1 was dialyzed against PBS [0.15 mol/l NaCl, 20 mmol/l phosphate buffer (pH 7.2)] to a concentration of 0.1 mg/ml or lower, and stored at Ϫ 80°C to avoid self-association before use. Twenty-fi ve peptides of apoA-1 were purchased from Protein Purity Ltd. (Gunma, Japan). The molecular mass of each where and S represent standard deviations of blank absorbance (n = 10) and slopes of calibration curves, respectively.

Nondenaturing PAGE
Plasma and pre ␤ 1-HDL-depleted plasma were diluted 10-fold with sample buffer containing 31% sucrose, 0.06% EDTA, and 0.01% Bromophenol blue; and purifi ed pre ␤ 1-HDL and lipidfree apoA-1 were diluted with the sample buffer to a concentration of 10 g/ml. The unbound fraction and the eluted fraction in affi nity chromatography using HDL were each diluted 2-fold with the sample buffer . Each sample was applied to a 15-25% polyacrylamide gradient slab gel with 5 l/well and electrophoresed at 75 V for 24 h at 4°C. Then Western blot analysis was performed in the same manner as described above for nondenaturing 2D electrophoresis. A high molecular mass calibration kit for native electrophoresis (GE Healthcare) was used as the molecular size marker.

Size-exclusion chromatography
Plasma, pre ␤ 1-HDL-depleted plasma, and lipid-free apoA-1 added to pre ␤ 1-HDL-depleted plasma were individually separated using size-exclusion chromatography in accordance with the methods of Nanjee and Brinton ( 31 ). Fifty microliters of each buffer (pH 2.0). The eluted fraction was neutralized by adding 0.05 ml of 2 mol/l Tris-HCl buffer (pH 8.0), followed by analysis by nondenaturing PAGE.

Nondenaturing 2D electrophoresis
Purifi ed pre ␤ 1-HDL (4 l) or plasma (2 l) was added to a 0.75% agarose gel in 50 mmol/l Barbital buffer (pH 8.6) prepared on gel-bond fi lm with a thickness of 1 mm, and run at 540 V for 60 min at 4°C. Strips of agarose gel were cut out, set on 10-20% polyacrylamide gradient slab gels, and run at 75 V for 24 h at 4°C in 90 mmol/l Tris, 80 mmol/l boric acid, and 3 mmol/l EDTA (pH 8.3). The separated HDL subfractions in the polyacrylamide gel were transferred onto poly vinylidene difl uoride (PVDF) membranes using a semidry electroblotter, and then analyzed by Western blot using goat anti-apoA-1 PAb, HRP-conjugated rabbit anti-goat IgG antibody (DAKO), and diaminobenzidine.

Electron microscopy
Purifi ed pre ␤ 1-HDL solution was dropped on a 400 mesh grid coated with carbon fi lm for dispersion. The adsorbed specimen was soaked with 2% uranyl acetate at 4°C for 10 s for negative staining and examined in a transmission electron microscope, JEM-1200EX (JEOL Ltd., Tokyo, Japan).

Protein analysis
The protein content was determined with a Micro BCA kit (Thermo). The protein composition was investigated by SDS-PAGE and mass spectrometric analysis as follows. The purifi ed fraction (0.22 g) diluted in a sample buffer containing SDS and 2-mercaptoethanol was boiled and electrophoresed on a 15% polyacrylamide gel. Mass spectrometric analysis and the pretreatment were performed as described previously ( 29 ). Briefl y, protein in the gel was detected by MS-compatible silver staining. The stained band was cut out and put into a 1.5 ml tube. Cysteine disulfi de bonds were reduced with DTT and alkylated with iodoacetamide. In-gel digestion with trypsin was performed at 37°C overnight, followed by extracting the peptides with 50% acetonitrile containing 5% TFA. The extracts were dried and redissolved in 0.1% formic acid, followed by LC/MS/MS analysis. LC/MS/MS analysis was carried out in a LCQ Deca XP ion trap mass spectrometer (ThermoFinnigan, USA) equipped with a nano-LC electrospray ionization source (AMR, Japan), interfaced on-line with a capillary HPLC system (Paradigm MS4; Michrom BioResources, USA). MS/MS data were analyzed using SEQUEST, a computer program that allows the correlation of experimental data with theoretical spectra generated from known protein sequences, to be compared against the latest version of the public nonredundant protein database of the National Center for Biotechnology Information.

Lipid analysis
The amounts of choline-containing phospholipids, cholesterol, and triglyceride were determined by enzymatic methods using commercial reagents (Wako Pure Chemical Industries, Ltd., Osaka, Japan) as follows. Each standard, diluted with saline (0.15 mol/l NaCl) and purifi ed pre ␤ 1-HDL (0.2 ml), were added to the appropriate enzyme solution (0.2 ml) in glass tubes and incubated at 37°C for 5 min. Absorbance at 600 nm was measured by a spectrophotometer, UV-2400PC (Shimadzu Corporation, Kyoto, Japan). In order to validate each lipid assay, limit of detection (LOD) and limit of quantifi cation (LOQ) were calculated using the following equations, in accordance with the appropriate ICH guideline ( 30 ). diluted 2,121-fold as the antigen, and goat anti-apoA-1 or anti-apoA-2 PAb as the biotin-labeled antibody in each plate. Reactivity to plasma gel fi ltration fractions was investigated by sandwich ELISA using HRP-labeled goat anti-apoA-1 PAb with plasma fractions separated on a fast protein liquid chromatography system (GE Healthcare) composed of two TSK gel G3000SW columns (7.5 mm × 60 cm), a TSK gel G3000SW column (7.5 mm × 30 cm) (Tosoh, Japan), and a Superdex 200HR1030 column (10 mm × 30 cm).

Epitope of anti-apoA-1 MAbs
To determine the epitope of each antibody, fi rst, reactivity to 25 kinds of apo-A1 peptide was investigated in the following peptide-competitive ELISA. Lipid-free apoA-1 (50 l/well), diluted to 0.5 g/ml with PBS, was added to an ELISA plate, followed by incubation at 4°C overnight. After washing and blocking the plate, various concentrations of each peptide were added (25 l/ well). Then each MAb (25 l/well), diluted to 200 ng/ml, was added, followed by incubation at room temperature for 1 h. After washing, HRP-labeled goat anti-mouse IgG antibody (50 l/well) was added, followed by incubation at room temperature for 1 h. After washing, the substrate solution (50 l/well) containing o -phenylenediamine and H 2 O 2 was added, followed by incubation at room temperature for 10 min. After stopping the enzymatic reaction by adding 1.5N H 2 SO 4 (50 l/well), the absorbance at 492 nm was measured by a plate reader. Next, reactivity of each MAb to CNBr fragments of apoA-1 was investigated by Western blot analysis. After separating apoA-1 CNBr fragments (5 g) by SDS-PAGE, the separated fragments in the polyacrylamide gel were transferred onto a PVDF membrane. The fragments on the membrane were fi rst reacted with each MAb and then reacted with HRP-conjugated goat anti-mouse IgG antibody, followed by detection of bands by diaminobenzidine. We then carried out an antibody-competitive ELISA to examine whether the epitope of MAb55205 is different from that of MAb55201. Various concentrations of each MAb (25 l/well) were added (25 l/well) to a plate coated with lipid-free apoA-1, as described above in the peptide-competitive ELISA test. Then biotin-labeled MAb55201 (1 g/ml) was added (25 l/well), followed by incubation at room temperature for 1 h. After washing, HRP-labeled streptavidin was added (50 l/well), followed by incubation at room temperature for 0.5 h. After a further washing, the enzymatic reaction was carried out as in the peptide-competitive ELISA.

Sandwich ELISA using anti-apoA-1 MAbs
To investigate the number of apoA-1 molecules per particle of plasma pre ␤ 1-HDL, sandwich ELISA using MAb55201 and MAb55205 was performed, with sandwich ELISA using only MAb14208 serving as control. Each MAb (50 l/well), diluted to sample was applied to a Superdex 200 HR 10/30 column connected in series to a Superdex 75 HR 10/30 column and separated in a 50 mmol/l Tris-HCl buffer (pH 7.4), containing 150 mmol/ l NaCl, 1 g/l sodium EDTA, and 1 g/l NaN 3 at a flow rate of 0.25 ml/min. After discarding 12.5 ml of the eluent, 40 fractions of 0.5 ml each were collected. The concentrations of pre ␤ 1-HDL and lipid-free apoA-1 in all fractions were determined by pre ␤ 1-HDL-ELISA ( 21 ). The apoA-1 concentrations of fractions separated from plasma were determined by sandwich ELISA using goat anti-apoA-1 PAb as described previously ( 21 ).

LCAT-dependent conversion experiment
Plasma and lipid-free apoA-1 added to pre ␤ 1-HDL-depleted plasma (0.5 ml) were individually incubated at 37°C with or without DTNB. A portion of each sample was taken at intervals and diluted 11-fold with a stabilization buffer ( 20 ) containing 50% sucrose. The diluted samples were immediately stored at Ϫ 80°C, and later the concentrations of pre ␤ 1-HDL and lipid-free apoA-1 were determined by pre ␤ 1-HDL-ELISA ( 21 ).

Preparation of new anti-apoA-1 MAbs
A hybridoma cell line producing MAb55205 was developed in the same manner as described previously for MAb55201 ( 21 ). Briefl y, after immunization of a Balb/c mouse with lipid-free apoA-1, the mouse was euthanized, and the spleen cells were fused with murine myeloma cells according to the method of Kohler and Milstein ( 32 ). Screening of hybridoma cells was performed using two kinds of ELISA with biotin-labeled goat anti-apoA-1 or apoA-2 PAb. Hybridomas were selected that produced antibodies showing reactivity in ELISA using anti-apoA-1 PAb, but not in ELISA using anti-apoA-2 PAb. A hybridoma cell line producing anti-apoA-1 MAb 14208 (MAb14208) was prepared as follows. After immunization and cell fusion were performed as described above, a hybridoma cell producing an antibody that reacted strongly with apoA-1 in apoA-1-immobilized ELISA was selected. The selected hybridomas were cloned by limited dilution and then injected into the peritoneal region of mice to harvest ascites, followed by purifi cation of IgG from the ascites using protein A-Sepharose.

Specifi city of anti-apoA-1 MAbs
The specifi city of the newly developed antibodies (MAb55205 and MAb14208) was assessed based on their reactivity to plasma proteins, LpA-1 (lipoprotein containing apoA-1 but not apoA-2), LpA-1:A-2 (lipoprotein containing both apoA-1 and apoA-2), and plasma gel fi ltration fractions, as described previously ( 21 ). Reactivity to plasma proteins was investigated by Western blotting using plasma proteins separated by SDS-PAGE. Reactivity to LpA-1 and LpA-1:A-2 was investigated by sandwich ELISA using plasma The absorbance of each sample was 0.0031, 0.0033, 0.0043, and 0.0033, respectively (mean of duplicates). All values were less than LOD (0.0899 g/ml).
c The absorbance of each sample was 0.0022, 0.0020, 0.0024, and 0.0023, respectively (mean of duplicates). All values were less than LOD (0.0694 g/ml). d The absorbance of each sample was 0.0023, 0.0044, 0.0022, and 0.0036, respectively (mean of duplicates). All values were less than LOD (0.339 g/ml). e Values are calculated on the basis of the following molecular weights: apoA-1, 28,331 Da; phospholipid, 774 Da; cholesterol, 387 Da; and triglyceride, 886 Da . 5 g/ml with PBS, was added to an ELISA plate, followed by incubation at 4°C overnight. After washing the plate with PBS, blocking solution was added, followed by incubation at room temperature for 1 h. This was then removed and various concentrations of lipid-free apoA-1 or plasma (50 l/well) were added to the plate, followed by incubation at room temperature for 1 h. After washing, each biotin-labeled MAb (50 l/well), diluted to 1 g/ml, was added and incubated for 1 h. After washing, HRPlabeled streptavidin (50 l/well) was added and incubated for 30 min. After washing, the HRP reaction was performed as described above in the peptide-competitive ELISA test. The blocking and dilution steps in the sandwich ELISA were performed using 1% BSA-PBS containing no detergent in order to preserve the three-dimensional structure of the lipoprotein, and the washing steps after the antigen reaction were performed using 0.1% BSA-PBS.

Purifi cation and analysis of plasma pre ␤ 1-HDL
We purifi ed pre ␤ 1-HDL from each plasma sample of the four healthy volunteers. The concentration and the amount of pre ␤ 1-HDL in the purifi ed fractions were 19.0 ± 1.7 g/ml (mean ± SD, n = 4) and 152 ± 13.4 g (mean ± SD, n = 4), respectively. In nondenaturing 2D electrophoresis, only pre ␤ 1-HDL spots were detected in the purifi ed fraction ( Fig. 1A ). In electron microscopy, only globular, not discoidal, particles were observed ( Fig. 1B ). In SDS-PAGE, only a band with a molecular mass of 28 kDa was detected ( Fig. 1C ). In LC/MS/MS analysis of the band, apoA-1 was detected, but other proteins, including apoM, were not detected (supplementary Table I). The protein concentration in the purifi ed fraction was 22.5 ± 3.5 g/ml (mean ± SD, n = 4), while any lipids, including phospholipids, cholesterol, and triglyceride, were not detected at all ( Table 1 ). The LOD of the lipid assays for phospholipids, cholesterol, and triglyceride were 89.9 ng/ml, 69.4 ng/ml, and 339 ng/ml, respectively ( Fig. 2 ).

Sandwich ELISA using anti-apoA-1 MAbs
Reactivity was not observed for either lipid-free apoA-1 or plasma pre ␤ 1-HDL in sandwich ELISA using only MAb55201 or only MAb55205 as the immobilized and labeled antibody, although it was observed for both lipidfree apoA-1 and plasma pre ␤ 1-HDL in sandwich ELISA using a combination of MAb55201 and MAb55205 ( Fig. 6A,  B, D, E ). On the other hand, in sandwich ELISA using MAb14208 as both the immobilized and labeled antibody, reactivity was not observed for lipid-free apoA-1, but was seen for plasma HDL ( Fig. 6C, F ). We performed this experiment two more times using plasma obtained from two other healthy volunteers and obtained the same results (data not shown). These results suggest that there is only one epitope recognized by each MAb in a particle of plasma pre ␤ 1-HDL, i.e., plasma pre ␤ 1-HDL is composed of one molecule of apoA-1, as is lipid-free apoA-1.

DISCUSSION
In the present study, we investigated the composition and structure of plasma pre ␤ 1-HDL. With respect to the composition of pre ␤ 1-HDL, several different results have been reported ( 8,18,19,31 ). Castro and Fielding ( 8 ) analyzed a small amount of pre ␤ 1-HDL extracted from polyacrylamide gel after separating plasma by nondenaturing 2D electrophoresis, and reported that pre ␤ 1-HDL was composed of apoA-1, phospholipids, and unesterifi ed cholesterol. Their report is the basis of the view that pre ␤ 1-HDL is composed of apoA-1, phospholipids, and unesterifi ed cholesterol. Nanjee and Brinton ( 31 ) separated a very small apoA-1-containing particle having a size of 5.8-6.3 nm (diameter) from plasma by using two kinds of analytical columns for size-exclusion chromatography. The particle is speculated to be the same as pre ␤ 1-HDL because its content in plasma is 2-15% of total apoA-1, its particle size is the smallest among plasma apoA-1-containing particles, it has pre ␤ mobility in agarose gel electrophoresis, and its concentration markedly falls when plasma is incubated at 37°C for 2 h. It has been shown that the particle is composed only of apoA-1 and phospholipids in a molar ratio of one to two. In the present study, we tried to obtain a large amount of pure pre ␤ 1-HDL, suffi cient to analyze its composition by affi nity chromatography using a MAb55201bound column and subsequent preparative gel fi ltration lipid-free apoA-1, added to pre ␤ 1-HDL-depleted plasma, decreased in a time-dependent manner when incubated at 37°C, while this decrease was not observed in the presence of DTNB. Moreover, the shift in the lipid-free apoA-1 concentration approximately corresponded to that in the pre ␤ 1-HDL concentration when the plasma was incubated ( Fig. 3D ).

Specifi city of anti-apoA-1 MAbs
In Western blotting analysis, MAb55205 and MAb14208 specifi cally reacted with apoA-1 having a molecular mass of 28 kDa (data not shown). In sandwich ELISA using biotin-labeled goat anti-apoA-1 or anti-apoA-2 PAb, MAb55205, like MAb55201, showed reactivity in ELISA using anti-apoA-1 PAb but not anti-apoA-2 PAb; whereas MAb14208 showed reactivity in ELISA using either anti-apoA-1 PAb or anti-apoA-2 PAb ( Fig. 4A ). With regard to the plasma gel fi ltration fractions, MAb55205, like MAb55201, showed reactivity only to those fractions with a molecular mass of less than 67 kDa containing pre ␤ 1-HDL; whereas MAb14208 showed reactivity not only for that fraction but also for other fractions containing HDL of higher molecular mass ( Fig. 4B ). These fi ndings indicate that MAb55205, like MAb55201, specifi cally recognizes apoA-1 associated with pre ␤ 1-HDL; whereas MAb14208 reacts with apoA-1 associated with all plasma HDL fractions.

Epitope of anti-apoA-1 MAbs
In peptide-competitive ELISA, MAb14208 reacted with only a peptide of aa97-116 in 25 peptides of apoA-1, that no phospholipids are present at all in the purifi ed pre ␤ 1-HDL, we can affi rm that at least phosphatidyl choline and sphingomyelin, previously reported ( 1 ) to be constituents of pre ␤ 1-HDL, are not present. We compared the purifi ed pre ␤ 1-HDL with intact plasma pre ␤ 1-HDL by nondenaturing PAGE, to examine the possibility that lipids were dissociated from apoA-1 in pre ␤ 1-HDL during the purifi cation. Our results showed no difference in mobility in nondenaturing PAGE between the purifi ed pre ␤ 1-HDL and intact plasma pre ␤ 1-HDL ( Fig. 3A ). It has been reported that the particle size in nondenaturing PAGE remarkably increases when a small amount of lipid binds to lipid-free apoA-1 ( 37,38 ). Conversely, the possibility that lipids were dissociated from apoA-1 in pre ␤ 1-HDL during the purifi cation process is extremely low because the particle size of pre ␤ 1-HDL remained unchanged. We also performed the affi nity chromatography under the same conditions using HDL prepared by ultracentrifugation, which does not contain lipid-free apoA-1 and/or pre ␤ 1-HDL, and confi rmed that neither lipid-free apoA-1 nor pre ␤ 1-HDL were stripped from HDL by the action of MAb55201 bound to Sepharose beads during the chromatography process ( Fig. 3A ). These fi ndings suggest that plasma pre ␤ 1-HDL does not contain lipids at all or, if it does, the amount is extremely small.
With respect to the structure, pre ␤ 1-HDL is postulated to be discoidal in many articles ( 2,3,39,40 ). However there is no report to directly prove that plasma pre ␤ 1-HDL has a discoidal shape ( 6 ). Discoidal HDL, having a pre ␤ electrophoretic mobility, is known to be synthesized in vitro as a model particle of plasma pre ␤ -HDL. The particle forms a structure which is called the double belt model chromatography using 40 ml of fresh normal plasma. As a result, we succeeded in purifying 139-169 g of pre ␤ 1-HDL from each plasma sample. We fi rst investigated the protein composition of the purifi ed pre ␤ 1-HDL, and found that only apoA-1 was detected. Wolfrum, Poy, and Stoffel ( 33 ) have reported that apoM is required for pre ␤ -HDL formation and is a component of pre ␤ -HDL. However, apoM was not detected in the purifi ed pre ␤ 1-HDL in the present study (supplementary Table I). The apoMcontaining pre ␤ -HDL previously reported might have included the larger pre ␤ -HDLs, pre ␤ 2-HDL, or pre ␤ 3-HDL, which could account for the different results in the two studies. In line with our results, low or undetectable levels of apoM in pre ␤ -HDL have been reported in other studies ( 34,35 ). We next investigated the lipid composition of the purifi ed pre ␤ 1-HDL, and found that no lipid was detected ( Table 1 ). We determined the LOD of each lipid assay to examine the possibility that sensitivity may have been insuffi cient. The results showed that the LOD values are, relative to the amount of protein, 0.40, 0.31, and 1.5% for phospholipids, cholesterol, and triglyceride, respectively. In molar ratios, these are 0.11 mol of phospholipids, 0.23 mol of cholesterol, and 0.48 mol of triglyceride to 1 mol of apoA-1. In other words, even if lipid is contained in the purifi ed pre ␤ 1-HDL, the amount is less than one molecule to one molecule of apoA-1. The enzymatic method we used to measure phospholipids cannot detect cholinefree phospholipids such as phosphatidyl ethanolamine, although it can detect choline-containing phospholipids including phosphatidyl choline, sphingomyelin, and lysophosphatidyl choline, comprising about 95% of plasma phospholipids ( 36 ). Therefore, although we cannot assert IC50, Concentration of competitive peptide required to cause 50% inhibition; >100, inhibition was not detected even at peptide concentrations of 100 mol/l . than 1.21 g/ml ( 11 ). This fact suggests that the discoidal HDL in lymph is different from plasma pre ␤ 1-HDL. Regarding the number of apoA-1 molecules per particle of pre ␤ 1-HDL, there are two hypotheses, one suggesting that there are two molecules and the other proposing one molecule ( 7 ). The hypothesis of two molecules is speculation based on the molecular mass determined by nondenaturing PAGE and size-exclusion chromatography ( 5,31 ). In the present study, the molecular masses of plasma pre ␤ 1-HDL determined by nondenaturing PAGE and size-exclusion chromatography are 60 kDa ( Fig. 3A ) and 50 kDa ( Fig. 3C ), respectively. It is well known that molecular mass determined by nondenaturing PAGE and size-exclusion chromatography is affected by the electrical charge and shape of the particle. Formisano, Brewer, and Osborne ( 27 ) reported the molecular mass of the apoA-1 monomer, determined to be 28 kDa by sedimentation equilibrium, to be 50 kDa in size-exclusion chromatography. Because the frictional ratio of apoA-1 has been reported to be 1.4 ( 44 ), it may be diffi cult to determine the exact molecular mass of the apoA-1 monomer or dimer by size exclusion chromatography or nondenaturing PAGE. On the other hand, the hypothesis of one molecule is based on the results of a cross-linking study ( 37,45,46 ). Nakamura et al. ( 45 ) reported that only the apoA-1 monomer was detected when plasma pre ␤ 1-HDL separated by nondenaturing PAGE was cross-linked in polyacrylamide gel, followed by SDS-PAGE. Our result, based on immunochemical analysis, is consistent with their report ( Fig. 6 ).
We compared intact plasma pre ␤ 1-HDL with monomolecular lipid-free apoA-1 obtained by delipidating plasma HDL and subsequent purifi cation, due to the fact that plasma pre ␤ 1-HDL appeared to be monomolecular lipidfree apoA-1. We demonstrated that no difference is to be found in their mobility in nondenaturing PAGE and nondenaturing 2D electrophoresis ( Fig. 3A, B ) and in their elution volume in size-exclusion chromatography ( Fig. 3C ). We also compared lipid-free apoA-1 with plasma pre ␤ 1-HDL in a LCAT-dependent conversion experiment. It is known that when plasma is incubated at 37°C, the concentration of plasma pre ␤ 1-HDL decreases and this decrease is blocked by the presence of DTNB ( 10,21,45,47 ). Although the mechanism has not yet been clarifi ed, the phenomenon seems to be a LCAT-dependent effect because it has been reported that the same effect is shown when anti-LCAT antibody is used instead of DTNB ( 47 ), and that plasma LCAT activity is positively correlated with the decreased level of plasma pre ␤ 1-HDL in the experiment ( 25 ). In the present study, lipid-free apoA-1 added to pre ␤ 1-HDL-depleted plasma showed the same variation in concentration as plasma pre ␤ 1-HDL in a LCAT-dependent manner ( Fig. 3-D ). These results suggest that plasma pre ␤ 1-HDL is monomolecular lipid-free apoA-1.
Plasma pre ␤ 1-HDL is known to play an important role in cholesterol effl ux, but it is still controversial whether pre ␤ 1-HDL is the substrate or the fi rst product of the ABCA1-mediated cholesterol effl ux ( 1, 2, 5-7 ). Castro and Fielding ( 8 ) reported that cell-derived cholesterol is fi rst detected at the position of pre ␤ 1-HDL in nondenaturing that characteristically contains two molecules of apoA-1 per particle ( 39,41 ). In the present study, we did not fi nd any discoidal particle, but found only globular particles in the purifi ed pre ␤ 1-HDL through electron microscopy ( Fig. 1B ). In addition, we found that apoA-1 in pre ␤ 1-HDL is one molecule per particle through an immunochemical procedure using anti-pre ␤ 1-HDL MAbs ( Fig. 6 ). The view that pre ␤ 1-HDL has a discoidal shape may be based on the fact that pre ␤ 1-HDL exists in lymph at a relatively high concentration ( 42 ) and some HDL in lymph has a discoidal shape ( 43 ). However the density of discoidal HDL in lymph is reported to be 1.063-1.21 g/ml ( 43 ), whereas the density of plasma pre ␤ 1-HDL is reported to be higher 2D electrophoresis when fi broblasts loaded with 3 H-labeled cholesterol are incubated with plasma. The result seems to demonstrate that pre ␤ 1-HDL is the fi rst product of cholesterol effl ux by apoA-1. In recent years it has been reported that "pre ␤ 1-HDL-like" lipid-poor apoA-1 or nascent HDL is formed when lipid-free apoA-1 retrieves phospholipids and cholesterol from cells mediated by ABCA1 in experiments using various culture cells ( 38,46,48,49 ). However, the particle sizes of those pre ␤ 1-HDLlike particles determined by nondenaturing PAGE are all larger than albumin with a diameter of 7.1 nm, and are clearly different from plasma pre ␤ 1-HDL, which is a particle smaller than albumin. On the other hand, it is known that the cholesterol effl ux from fi broblasts to plasma markedly declines when pre ␤ 1-HDL is depleted from plasma ( 10 ), or when antibodies specifi cally recognizing pre ␤ 1-HDL are added to plasma ( 9 ). The results seem to demonstrate pre ␤ 1-HDL itself retrieves cholesterol, i.e., pre ␤ 1-HDL is the substrate of the cholesterol effl ux. It is also reported that pre ␤ 1-HDL-specifi c cholesterol effl ux remarkably declines when fi broblasts are pretreated with protease, while it is not infl uenced by protease when blood Fig. 6. Sandwich ELISA using anti-apoA-1 MAbs. A-C: Reactivity with lipid-free apoA-1 in each sandwich ELISA using biotin-labeled MAb55201 (A), biotin-labeled MAb55205 (B), and biotin-labeled MAb14208 (C). D, E: Reactivity with plasma pre ␤ 1-HDL in each sandwich ELISA using biotin-labeled MAb55201 (D) and biotin-labeled MAb55205 (E). Plasma with a pre ␤ 1-HDL concentration of 26.9 g/ml was used after diluting 441-to 3,580-fold. The horizontal axis indicates the plasma pre ␤ 1-HDL concentration. F: Reactivity with plasma HDL in sandwich ELISA using biotin-labeled MAb14208. Plasma with an apoA-1 concentration of 1.3 mg/ml was used after diluting 14,112-to 451,584-fold. The horizontal axis indicates the plasma apoA-1 concentration. Values are presented as the mean value ± SD of triplicate measurements.