A 70-kDa Apolipoprotein Designated ApoJ Is a Marker for Subclasses of Human Plasma High Density Lipoproteins*

p1 range

can predict premature artery disease (6) since these conditions are associated with low HDL-cholesterol. The major physiological role postulated for HDL, consistent with its inverse association with disease, is to mediate reverse cholesterol transport (7)(8)(9). In this role, HDL collect unesterified cholesterol from tissues, facilitate its esterification by lecithin: cholesterol acyltransferase (LCAT) which utilizes apoA1 as an activating cofactor (lo), serve as donors of cholesteryl esters (CE) to other lipoproteins in reactions catalyzed by the cholesteryl ester transfer protein (CETP) (1 l), and deliver CE to the liver for excretion. HDL can also supply functional apolipoproteins to other lipoproteins, specifically apoCI1 to chylomicrons and very low density lipoproteins in the plasma for activation of lipoprotein lipase (12) and apoE to chylomicrons in the intestinal lymphatics (13). Our understanding of HDL metabolism is surprisingly limited. Depending on the method of isolation, at least eight discrete HDL subclasses have been identified (14,15). The metabolic interrelationship between the subclasses is not well established nor have specific functions been assigned to the different subclasses. It may be that different HDL subclasses are targeted to different tissues. The irreversible uptake of HDL by tissues has been studied in animal models, particularly the rat (16). The important sites of HDL delivery are the liver and steroidogenic tissues. HDL-cholesterol appears to enter cells by receptor-mediated (17, 18) and receptorindependent pathways (19)(20)(21). Even the origin of HDL particles is uncertain. Nascent HDL are synthesized both in the liver and intestine (22)(23)(24). In addition, surface constituents released during the catabolism of triglyceride-rich lipoproteins can contribute to one or all of the HDL subclasses (25-27). Ellsworth et al. (28) and Hopkins et al. (29) presented evidence that CETP activity is necessary for the speciation of HDL and particularly that which occurs as a consequence of the hydrolysis of very low density lipoprotein-triglycerides catalyzed by lipoprotein lipase (28). The normal distribution of HDL subclasses is markedly altered in patients with complete deficiency of CE transfer activity (30). A role for CETP in HDL speciation is consistent with its capacity to bind to HDL (31). The HDL with which CETP is associated may be the preferred substrates from which CE is transferred to other lipoproteins.
The rate-limiting step in the lipid transfer reaction is the formation of a ternary complex comprised of CETP and the donor and acceptor lipoproteins (32). If CETP is already complexed with HDL, the transfer process is favored (32). Cheung and colleagues (33) reported that, in uiuo, CETP is associated with specific subclasses of HDL which contain apoA1 but not apoAI1. Francone et al. (34) determined that the HDL containing apoA1 and a substantial portion of the CETP activity involve only about 2% of the total plasma apoA1 and can also contain apoD and LCAT activity. These findings lend credence to the hypothesis that the CETP-HDL interaction is physiologically relevant. The apparent association of CETP with specific populations of HDL raises the possibility that CETP activity is regulated by lipoprotein constituents.
In fact, the accumulation of lipolytic products (35) and cholesterol (36) at the lipoprotein surface can, respectively, enhance and retard CETP-catalyzed lipid transfer. Apolipoproteins may also regulate CETP activity (37, 3.8). We assume that the activity of CETP is controlled by the lipid and apolipoprotein constituents with which it associates in uiuo. To obtain a more complete understanding of CETP and to identify apolipoproteins which are associated with it, we used an immunochemical approach.
Monoclonal antibodies were raised against partially purified CETP. These antibodies were screened to select those which recognized CETP and those which recognized CETP-associated proteins. Much to our surprise, one of the antibodies which recognized a CETP-associated protein identified a new apolipoprotein of HDL which we have designated apolipoprotein J (apoJ). This report describes the isolation of apoJ and the characterization of lipoproteins containing apoJ.

EXPERIMENTAL PROCEDURES
Materials-Polyacrylamide gel reagents, nitrocellulose paper, and secondary antibodies for monoclonal antibody characterization and electroimmunoblotting were obtained from Bio-Rad or from Iles-Yeda. The MonoQ anion-exchange column. eradient gels. high molecular weight protein standards, and amph&es wereGb&i&d from Pharmacia LKB Biotechnology Inc. Prestained protein standards were purchased from Bethesda Research Laboratories. Trifluoroacetic acid was purchased from Pierce Chemical Co. and HPLC grade acetonitrile from Fisher. Cholesterol and triglyceride quantitation kits were purchased from Sigma. General chemicals were purchased from Sigma or Fisher. Tissue culture reagents were obtained from M. A. Bioproducts or GIBCO. P3X63Ag8 cells, purchased from the American Type Culture Collection, were grown in ascites of BALBc female mice. Polyclonal rabbit antiapoA1 antibody was purchased from Behring Diagnostics. Monoclonal apoA1 antibody was provided by Dr.

Purification of Cholesteryl Ester
Transfer Protein-CETP was isolated from human plasma by affinity chromatography as described by Busch et al. (39). Typically, this one-step procedure resulted in a 3,500-fold purification of CETP. CETP was also purified 15,000-fold by column chromatography, using the procedure described by Jarnagin et al. (40) through the CM-cellulose chromatographic step. One fig of purified CETP (15,000-fold) catalyzed the transfer of 18.8f 1.7% of the TG and 24.8+0.8% of the CE in radiolabeled-LDL to HDL during a 3-h incubation at 37 "C.
Neutral Lipid Transfer Assay-In the typical assay, radiolabeled-LDL (12.5 pg of cholesterol) was the donor and unlabeled HDL (25 pg of cholesterol) was the acceptor. Radiolabeled lipoprotein substrates were prepared by incubating phosphatidylcholine emulsions containing 20% by weight neutral lipid (300 ,.&i of [3H]TG and ["Cl CE) overnight at 37 "C with 100 ml of plasma and 5,5'-dithiobis-2nitrobenzoic acid to inhibit cholesterol esterification (41): radiolabeled lipoproteins were isolated by ultracentrifugal f&&on (42). Radiolabeled-LDL and unlabeled HDL were incubated for 3-15 h at 37 'C in the absence and presence of catalyst. The amount of radiolabeled lipid in HDL was determined following precipitation of LDL by heparin/Mn*+ (43).
The effect of polyclonal antiCETP (39) and monoclonal antibody (mAb) 11 on lipid transfer was determined by direct addition of antibody and by-indirect immunoprecipitation, respectively. Purified CETP (40) was incubated overnight at 4 "C with varving concentrations of purified goat antiCETP-IgG (1 mg/ml) or purified mAbl1 (0.7 mg/ml) in 50 mM Tris-HCl, pH 7.4. Aliquots containing poly-clonal antibodies were centrifuged, and the supernatants were assayed directly; mAbl1 and its bound antigen were precipitated by the addition of a high affinity goat antimouse antibody, and the supernatants were assayed for lipid transfer activity. Preimmune goat immunoglobulin and a purified nonrelated IgG,, antibody from the secreting myeloma P3X63Ag8 served as controls.

Preparation of Morwclonul
Antibodies-CBGFl/J mice were injected intraperitoneally with 50 pg of partially purified CETP (3500fold) (39) in an equal volume of Freund's complete adjuvant. After 4 and 6 weeks, the mice were boosted with 50 pg of antigen in Freund's incomplete adjuvant. Fusion was conducted 3 days after the second boost by a modification of the method of Kennett et al. (44). Mouse P3X63Ag8.653 myeloma cells and erythrocyte-free spleen cells from immunized mice were mixed in a 1:lO ratio in the presence of 40% polyethylene glycol (PEG-4000), diluted, and centrifuged at 400 X g for 5 min. The cells were washed gently with serum-free Dulbecco's modified Eagle's medium, resuspended (4 X 10' cells/ml) in Dulbecco's modified Eagle's medium containing 20% fetal bovine serum, 10% minimum essential Alpha medium, and HAT (100 PM hypoxanthine with 0.8 PM aminopterin and 16 PM thymidine), and cloned at limiting dilution.
Supernatants from wells containing single clones were assayed for production of antibody to partially purified CETP by an enzymelinked immunoassay (EIA). Partially purified CETP (10 pg) was added to each well of a 96-well microtiter plate and incubated overnight at 4 "C. Immediately prior to use, the plate was washed three times with plasma density buffer (PDB, 10 mM Tris-HCl, pH 7.4, plus 150 mM NaCl, and 1 mM EDTA) containing 0.5% (v/v) Tween 20 (PDB-Tween). Unreacted sites were blocked for 1 h at 37 "C with 200 pl of PDB-Tween containing 3% bovine serum albumin, and 100 ~1 of culture supernatant were added. Specific reactivity was detected with alkaline phosphatase-labeled rabbit antimouse IgG (1:5000 dilution). Positive wells were quantitated using an Artek plate reader.

Zmmurwaffinity
Purification of Apolipoprotein J-MAbll was covalently attached to Affi-Gel, and the resulting column was employed to purify apoJ from plasma. Fast protein liquid chromatograpbypurified mAbl1 (5 mg/ml) was dialyzed into 0.1 M HEPES, pH 7.0, and cross-linked to Affi-Gel 10 (6 ml, Bio-Rad), according to the manufacturer's instructions. Unreacted sites were blocked with 0.6 ml of 1 M ethanolamine-HCl, pH 8.1. The affinity matrix had 17 mg/ ml of bound mAb. Fresh plasma (25 ml) was diluted 1:3 with phosphate-buffered saline containing 1 mM phenylmethylsulfenyl fluoride, 1 mM EDTA, 1 mM benzamidine, 25 kallikrein units/ml of aprotinin, and 0.02% NaN, and chromatographed at 4 "C on mAbll-Affi-Gel or on a control column of Affi-Gel alone blocked with ethanolamine-HCl. Unbound protein was eluted with 300 bed volumes of PDB and 30 bed volumes of 3 x PDB. Bound protein was eluted with 8 volumes of 3 M KSCN or 50 mM triethylamine (TEA) or 1 M acetic acid (HOAc), as indicated. The eluted samples (mAbl1 eluate) were dialyzed immediately into 10 mM NH,HCOa and concentrated by ultrafiltration to 500 pg/ml or lyophilized.

A Monoclonal
Antibody Raised against Partially Purified CETP Recognizes a 70-kDa Plasma Protein-A preparation of CETP purified 3500.fold (39) was used to raise monoclonal antibodies to characterize CETP and associated proteins. The antigen was comprised of a major protein of 67-70 kDa, a protein of 58 kDa and minor proteins, as determined by SDS-PAGE. The 67-70-kDa protein was distinct from albumin and had CETP activity when electroeluted from the gel (54). Since CETP was the catalytic protein of interest in any complexes containing it, the isolation of CETP mAb was the highest priority. Antibodies which reacted with partially purified CETP, identified by EIA, were initially chosen. Positive clones were subsequently screened by electroimmunoblotting of plasma and partially purified CETP, and the few mAb which recognized the 67-70.kDa protein were selected with the expectation that these were CETP mAb. A highly reactive positive clone, designated clone 11 and producing an IgG,, mAbll, was expanded and subcloned at limiting dilution. Antibody from all subclones reacted identically to the parent clone, as assessed by EIA and electroimmunoblotting, and the clonality of the antibody secreted by the parent clone was confirmed by two-dimensional gel electrophoresis. The reactivity of mAbl1 toward proteins in human plasma and in partially purified CETP, determined by electroimmunoblot analysis, is illustrated in Fig. 1. Under nonreducing (NR) conditions, mAbl1 recognized a 'IO-kDa protein in plasma (panel A) and in partially purified CETP (panel I?).
In the presence of 20 mM DTT (reduced, R), the protein recognized by the antibody migrated as a heterogeneous band at -35 kDa. Based on electroimmunoblot analysis, the 70(NR)/35(R)-kDa protein was also present in CETP preparations purified 15,000-fold (not shown). Human serum albumin was not recognized by mAbl1. CETP activity has been reported to be associated both with 70- (40)  Purified CETP (15,000-fold, 1 rg) was incubated overnight at 4 "C with varying amounts of purified goat IgG raised against partially purified CETP (1 mg/ml) (panel A) or purified mAbl1 (0.7 mg/ml) (panel R) in 50 mM Tris-HCl, pH 7.4. Antigen-antibody complexes were removed by centrifugation as described under "Experimental Procedures," and the supernatants were assayed (3 h at 37 "C) for CE (U) and TG (A-A) transfer, using radiolabeled-LDL and unlabeled HDL. Preimmune goat immunoglobulin and a purified nonrelated IgG, from the secreting myeloma P3X63Ag8 had no effect on the transfer of TG or CE. The data are represented as percent of untreated controls.

CETP-The
ability of mAbl1 to remove CETP activity from plasma supported the possibility that mAbl1 is an antibody specific for CETP. To test t.his hypothesis further, we asked whether mAbl1 could inhibit CETP activity. Increasing concentrations of mAbl1 were added to a transfer reaction consisting of radiolabeled LDL, unlabeled HDL, and CETP, purified l&000-fold.
To establish that CETP could be inhibited by antibodies, we showed that the polyclonal antibody raised against partially purified CETP (39) removed transfer activity in a concentration-dependent manner (Fig. 3, panel   A). Addition of similar concentrations of preimmune goat immunoglobulin had no effect. In contrast, direct addition of mAbl1 had no effect on the catalyzed transfer of CE or TG (not shown). Moreover, mAbl1 failed to remove CE or TG transfer activity (panel B) when a high affinity goat antimouse IgG secondary antibody was included with mAbl1 to achieve immunoprecipitation.
The mAb from the secreting myeloma P3X63Ag8 used as a control in the indirect immunoprecipitation experiments had no effect on CE or TG activity. To confirm that the 70(NR)/35(R)-kDa protein was removed from the CETP preparation by mAbl1, the precipitates were collected by centrifugation and the 70(NR)/35(R)-kDa protein was detected by electroimmunoblotting. Although mAbl1 removed the 70(NR)/35(R)-kDa protein without removing transfer activity, there was a consistent and reproducible increase in TG transfer which occurred at low anti-body concentrations.
Removal of the 70(NR)/35(R)-kDa protein by indirect immunoprecipitation without concomitant removal of transfer activity indicates that the 70(NR)/35(R)-kDa protein is not CETP or another protein with CE and TG lipid transfer capability. Moreover, the data suggest that it is not an inhibitor of neutral lipid transfer.

The mAb1 I Eluate Contains the 70(NR)/35(R)-kDa
Protein and upoAZ--The finding that the antigen recognized by mAbl1 was not CETP raised two questions. Has the 70(NR)/ 35(R)-kDa protein been described previously and is it an apolipoprotein?
If it is an apolipoprotein, the mAbl1 eluate should consist of lipid and perhaps other apolipoproteins.
To address these questions, mAbl1 eluate was obtained for structural studies. Fresh human plasma, diluted in buffer containing protease inhibitors, was chromatographed on mAbll-Affi-Gel and on a control Affi-Gel column to which no antibody was attached. The resulting eluates were analyzed by SDS-PAGE (Fig. 4). Panel A illustrates the CBB-stained gel of the eluate from control-Affi-Gel (lanes 1 and 2) and from mAbll-Affi-Gel (lanes 3 and 4). The 70.kDa protein and its reduced forms at 35 kDa reacted with mAbl1 (panel B, lanes 7 and 8, respectively).
The mAbl1 immunoreactivity which remained at 70 kDa in the presence of reductant was not affected by alkylation, suggesting that a fraction of the 70-kDa protein was resistant to reduction. Since the 70-kDa protein was not visible in the CBB-stained gel under reducing conditions, its contribution to the total 70-kDa species was small. The 2%kDa protein which was present in the mAbl1 eluate but not in the control eluate was apoA1, identified by electroimmunoblotting with mAbB7 (panel C, lanes Control and mAbl1 eluates were prepared by chromatographing 25 ml of human plasma, diluted 1:3 in phosphatk-buffered &line containing protease inhibitors. on Affi-Gel (control) or mAbll-Affi-Gel as specified under "Experimental Procedures." Bound protein was elutedin 50 mM TEA and dialyzed immediately into 10 mM NH,HCO,. Equal volumes (1 ml for stained gel analysis and 100 ~1 for electroimmunoblot analysis) of each eluate were lyophilyzed, solubilized, and electrophoresed for 5.5 h at 30 mA. The proteins were stained with CBB (panel A) or transferred to nitrocellulose and incubated with apoJ mAbl1 (panel B) or apoA1 mAbB7 (panel C). The absence or presence of 20 mM DTT is denoted as nonreduced (NR) or reduced (R), respectively. sequence analysis of electroeluted protein (not shown). Chromatography of plasma on mAbll-Affi-Gel removed a small amount of the total apoA1 in plasma, 2.2% (n = 3).
In sum, chromatography of plasma over mAbll-Affi-Gel removed a 70-kDa disulfide-linked dimer of 35-kDa subunits, and apoA1, a marker of HDL. In other studies, mAbl1 eluate was also found by electroimmunoblot analysis to contain very minor amounts of apoI1, apoAIV, apoE, and apoD; however, no apoB or LCAT was detected. The 70(NR)/35(R)-kDa protein recognized by mAbl1 was designated apolipoprotein J. The apolipoprotein designation was justified by the fact that mAbl1 eluate also contained lipid. MAb eluate was 10.6 -t 1.8% lipid as: 6.3 5 1.4% phospholipid, 3.3 2 0.3% cholesterol, and 1.0 + 0.3% triglyceride (n = 3). Negligible apoJ or apoA1 bound to control-Affi-Gel which indicates that apoJ and apoA1 were specifically removed from plasma by mAbl1. The method of elution of the columns, using TEA, HOAc, or KSCN, did not influence the results. The amount of mAbl1 eluate protein typically obtained from 25 ml of plasma, under conditions where >90% (n = 3) of the mAbl1 immunoreactivity was removed, was 1.7 mg, about 80% of which was apoJ based on relative CBB staining intensity. Therefore, we estimate that the plasma apoJ concentration is -9 mg/dl. the 35-kDa species, after reduction which was not detected by CBB (Fig. 5, left panel). The p1 of apoA1, the 28-kDa protein in mAbl1 eluate, was 5.6, as reported (57). MAbll did not recognize two other additional proteins, a 90.kDa protein with a p1 of 6.4-6.9 and a 45-kDa protein with a p1 of 5.6. These proteins were not characterized further in this study.

ApoJ Consists of Two
MAbll Eluute Contains Distinct Lipoprotein Species-The presence of both apoA1 and lipid in the mAbl1 eluate suggested the existence of macromolecular complexes. To assess this possibility, the eluate was analyzed by nondenaturing GGE (Fig. 6). Panel A of Fig. 6 shows a CBB-stained gel of mAbl1 eluate compared with HDL (d 1.063-1.21 g/ml), isolated by sequential ultracentrifugation, for reference. HDL consisted of lipoproteins which ranged in size from 100 to 400 For electroimmunoblots @nnels C and I)), mAbl1 eluate (17 and 27 peg for apoJ and apoA1 analysis, respectively) or HDL (135 and 15 rg for apoJ and apoA1 analysis, respectively) resolved by GGE and transferred to nitrocellulose, were incubated with apoJ-specific mAbl1 (r~nnel C) and apoAIspecific mAbB7 (panel I)).
kDa, consistent with the sizes of ultracentrifugally isolated HDL reported by Blanche et al. (58). MAbll eluate consisted of four major species of 80,160,240, and 340 kDa and a minor component of 520 kDa. The relative distribution of these particles, determined by densitometric scanning, was 80 kDa, 35%; 160 kDa, 8%; 240 kDa, 17%; 340 kDa, 27%; 520 kDa, 13%. The sizes and the distribution of particles in mAbl1 eluate are consistent with the size distribution of immunoaffinity purified HDL analyzed by GGE (59). An identical gel stained with Sudan Black (panel B) revealed that the 160-, 240-, and 340-kDa complexes contained lipid. If lipid was present in the 80-and 520-kDa complexes, it was below the level detectable by this assay. All sizes of complexes contained apoJ, as determined by electroimmunoblotting (panel C). Using mAbB7 (panel D), apoA1 was revealed in the 160-, 240-, and 340-kDa complexes; low levels of apoA1 immunoreactivity were also detected in the 80-and 520-kDa com-plex%.
For Fig. 6, mAbl1 eluate was obtained with TEA. The results were the same when HOAc was used to elute bound protein from the column. Furthermore, a comparison of fresh plasma and mAbl1 eluate by nondenaturing GGE coupled with antiapoJ immunostaining revealed that the sizes of apoJlipoproteins in plasma and in mAbl1 eluate were the same. This result indicates that elution of apoJ-lipoproteins from the immunoaffinity column did not alter their sizes. The proteins of the mAbl1 eluate complexes resolved by nondenaturing GGE were also evaluated by SDS-PAGE to confirm the presence of apoJ and apoA1. The complexes, designated l-5 (Fig. 6, panel A), were electroeluted and subjected to SDS-PAGE in the presence of DTT followed by electroimmunoblot analysis. The samples shown in the left panel of Fig. 7 were stained with mAbI1, and the results establish that complexes l-5 contained apoJ of the expected reduced size of -35 kDa. Electroimmunoblot analysis with apoA1 mAbB7 (Fig. 7, right panel) confirmed that apoA1, the 28-kDa protein, was also present in each complex. HDL (d 1.063-1.21 g/ml) were also evaluated for the presence of apoJ and apoA1 by electroimmunoblotting. ApoJ was not detected in these HDL which had been isolated by sequential ultracentrifugal flotation.
As expected, apoA1 was detected (Fig. 6 The coml)lexes in mAhl1 resolved 1)~ GGE and designated l-5 in Fig. 6 (80 rg/ml), obtained in 50 mu TEA, were stained with I)hosphotunpstic acid as sl)ecified under "Experimental Procedures." phosphotungstic acid and visualized by electron microscopy (Fig. 8). The size of the particles in each sample was estimated by measuring the diameters of 100 particles in a representative field of each micrograph. MAbll eluate (Fig. 8, COP) consisted of spherical particles which ranged in diameter from 5 to 16 nm. Of these, approximately 50% were 5-9 nm, 30% 9-12 nm, and 20% 12-16 nm. Of the particles, 70% were <lo nm and 30% were >lO nm in diameter. The size distribution of particles in mAbl1 eluate agrees well with the range of 5-16 nm reported for HDL isolated by immunoaffinity chromatography (60). HDL isolated by immunoaffinity chromatography have a broader size range than those isolated by ultracentrifugal flotation (60 L'ersus 61). The HDL which we isolated by ultracentrifugal flotation ranged from 5 to 12 nm in diameter (Fig. 8, bottom): approximately 45% were 5-9 nm, 45% 9-12 nm and 10% 1 '2-16 nm. bomparison of the Distribution of apoAI and apoJ in Human Plasma-Plasma from two donors (female and male) was rapidly fractionated in a discontinuous salt gradient (50) to minimize disruption of lipoprotein particles relative to that which can result from sequential ultracentrifugal flotation in which long centrifugation times are used. The gradient, se- Peaks of apoJ occurred in both the HDLL and HDL.1 region, fractions which also contained apoA1. In the plasma from both subjects, apoJ was present in the HDLL density range Cd 1.06S1.125 g/ml), in the tailing edge of HDL, Cd 1.125-1.21 g/ml) and in the density range containing very high density lipoproteins (VHDL, d 1.21-1.25 g/ml). No apoJ was detected in fractions of cl < 1.063 g/ml, but apoJ was also present in fractions of d > 1.25 g/ml. Within the density range of 1.063-1.25 g/ml, there were fractions which contained apoA1 but no detectable apoJ, supporting the GGE results which suggest that apoJ is associated with discrete subclasses of HDL.

DISCUSSION
This report establishes the existence of a unique protein, designated apoJ, which is associated with plasma lipoproteins. ApoJ is a 70-kDa disulfide-linked dimer of -35kDa subunits, both of which have pI values between 4.9 and 5.4. The lipoproteins which contain apoJ are HDL. This conclusion is supported by two lines of evidence. First, apoJ and apoA1, a marker of HDL, are associated with the same sized particles in mAbl1 eluate. Electrophoresis of mAbl1 eluate on gradient gels under nondenaturing conditions resolves four major and one minor discrete subclasses of lipoproteins. ApoJ and apoA1 exist within all five subclasses. The molecular masses of the major subclass are within the range reported for HDL, SO-340 kDa. The minor subclass of 520 kDa is larger than typical HDL. Second, electron microscopic evidence indicates that mAbl1 eluate contains spherical particles with diameters in the HDL range, 5-16 nm. The majority of the apoJ-containing HDL are small and dense, as determined by GGE and by discontinuous density gradient ultracentrifugation, suggesting that they can be classified as HDL, and VWDL. Taken together, the data support a bimodal density distribution of apoJ-HDL with the greatest percentage of apoJ-HDL in the HDLs-VHDL classes.
Weighed against the evidence that apoJ defines distinct subclasses of HDL is the puzzling result that no apoJ was detected in HDL by electroimmunoblot analysis. The amount of apoJ in plasma is in the range of that of apoE and the apoCs. The abundance of apd, predicted on the basis of its yield from immunoaffinity chromatography of plasma samples isolated from three donors, is about 9 mg/dl. The range of apoJ determined in preliminary experiments for more than 41 donors, using a quantitative EIA, is 7-19 mg/d12 If most of the apoJ in plasma were in HDL, apoJ should have been detected in HDL by electroimmunoblotting, but was not. Since the HDL used in this study were isolated by sequential ultracentrifugal flotation, it is possible that apoJ (like apoA1 (62,63), apoE (64), and, in particular, apoAIV (65) and apoD (66)) dissociates from lipoproteins during prolonged centrifugation. Using a short centrifugation time in a vertical rotor, apoJ is detected in HDL. Dissociation of apoJ from lipoproteins will make it difficult to determine the structure of the physiolo~cal complexes of this apolipoprotein, as has been the situation for apoAIV (67) and apoD (66). Isolation of apoJ by immunoaffinity chromatography, as we have done in this study, offers the least disruptive alternative to evaluating apoJ-HDL. Proteins with properties similar to those of the apoJ subunits have been previously reported to exist in immunoaffinity isolated HDL (60).
The use of selected immunosorption to isolate specific subpopulations of lipoproteins (for example, immunoaffinity matrices specific for apoA1, apoAI1, apoB and apoE) has provided convincing support for the use of this strategy to study plasma lipoprotein structure and interrelationships (68-70). One advantage is that specific lipoproteins, containing the apolipoprotein of interest, are obtained without appreciable structural disruption. Another is the oppo~unity to obtain minor subclasses of HDL which contain apolipoproteins which are present in low abundance in plasma. Immunoaffinity chromatography has been particularly useful in studying subpopulations of HDL. Using antiapoA1 and antiapoAI1 immunoaffinity matrices, lipoproteins which contain both apoA1 and apoAI1 have been separated from those which contain apoA1 without apoAI1 (33). CETP and LCAT are associated with the apoAI-HDL not the apoAI/apoAII-HDL (33, 34). We have shown that CETP activity can also be associated with HDL subclasses comprised of apoJ and apoA1; however no LCAT was detected in these subclasses by immunochemical analysis.
The presence of CETP activity in mAbl1 eluate implies an association between apoJ-HDL and CETP. ApoJ can also copurify with CETP during sequential column chromatography. CETP, a 'IO-kDa glycoprotein with a p1 of 5.2, has been purified to homogeneity (40,71), and its primary structure deduced from its cDNA (72). We established here that CETP and apoJ are different proteins. MAbll depleted apoJ from a preparation of CETP but failed to remove neutral lipid transfer activity. We have now confirmed the separate identities of apoJ and CETP by sequence analysis (73). Documentation of an association between CETP activity and apoJ does not prove a physiological role for apoJ in the structure-function of CETP. A role is suggested, however, by the fact that mAbl1 consistently enhances the transfer of TG facilitated by CETP. The basis of this enhancement is not known. It is not clear at the present time how the lipoproteins containing apoAI, apoD, and CETP activity (34) and those containing apoA1, apoJ, and CETP activity are related, if at all. The most significant consequence of this investigation is the identification of apd, an apolipoprotein which defines discrete subclasses of apoAI-confining HDL. This finding emphasizes the heterogeneity of HDL. In addition to heterogeneity in size and density, HDL are heterogeneic in protein constituents. Apolipoprotein heterogeneity is a consequence of the minor proteins in HDL such as apoJ. The HDL subclasses which contain apoJ may account for only 2% of the HDL, based on the amount of apoA1 associated with them. Although low in abundance, these HDL subclasses may have an important physiologic function. Among apolipoproteins in HDL, apoJ is unique in its molecular weight, subunit structure, and isoelectric point. The identification of apoJ in HDL presents exciting opportunities. According to current concepts, small HDL mature into large HDL as a result of cholesterol efflux and esterification processes. Thus, the apoJ-HDL in the HDL* density range may be products of the small dense apoJ-HDL. The apoJ-HDL in the HDL~/VHDL density range have properties app~pria~ for cholesterol acceptors and LCAT substrates. They are small, poor in lipid, and contain the LCAT activator apoA1. Castro and Fielding (74) recently reported that small HDL, present in low abundance in plasma, preferentially accept effluxed cholesterol from cultured fibroblasts. Small HDL are also the preferred sites of LCAT action (75, 76). In fact, within the HDLB density range, the maximum velocity of the LCAT reaction correlates inversely with the size of the HDLS (76). The bimodal density distribution of apoJ-HDL suggests that an understanding of apoJ and apoJ-HDL subclasses may provide insight into the process of reverse cholesterol transport.