Triglyceride Transfer Is Required for Net Cholesteryl Ester Transfer between Lipoproteins in Plasma by Lipid Transfer Protein EVIDENCE FOR A HETERO-EXCHANGE TRANSFER MECHANISM DEMONSTRATED BY USING NOVEL MONOCLONAL ANTIBODIES*

In order to investigate the role of lipid transfer pro- tein (LTP) in plasma lipoprotein metabolism, monoclonal antibodies (mAbs) have been raised against LTP isolated from rabbit plasma. mAbs 2-86 and 3-9F inhibited both [SHlcholesteryl ester (CE) and [‘Hltriglyceride (TG) transfer from low density lipoprotein (LDL) to high density lipoprotein (HDL) mediated by LTP. Although 3-9F cross-reacted with human LTP, 2-8G was species-specific for rabbit LTP. mAb 14-8H inhibited only tSH]TG but not [‘HICE transfer and was cross-reactive with hu- man LTP. mAbs 2-86 and 3-9F interfered with association of LTP with lipid microemulsions, again 2-86 with species specificity, whereas 14-8H did not affect LTP-microemulsion binding. Thus, mAbs 3-9F and 14-8H were used for further study in human plasma. By total inhibition of CE and TG transfer by 3-9F, LTP was shown to be responsible for net mass transfer of neutral lipids between lipoprotein classes in plasma, namely of CE from HDL to very low density lipoprotein (VLDL) and TG from VLDL to LDL and HDL. Selective inhibition of TG transfer by mAb 14-8H

In order to investigate the role of lipid transfer protein (LTP) in plasma lipoprotein metabolism, monoclonal antibodies (mAbs) have been raised against LTP isolated from rabbit plasma. mAbs 2-86 and 3-9F inhibited both [SHlcholesteryl ester (CE) and ['Hltriglyceride (TG) transfer from low density lipoprotein (LDL) to high density lipoprotein (HDL) mediated by LTP. Although 3-9F cross-reacted with human LTP, 2-8G was speciesspecific for rabbit LTP. mAb 14-8H inhibited only tSH]TG but not ['HICE transfer and was cross-reactive with human LTP. mAbs 2-86 and 3-9F interfered with association of LTP with lipid microemulsions, again 2-86 with species specificity, whereas 14-8H did not affect LTPmicroemulsion binding. Thus, mAbs 3-9F and 14-8H were used for further study in human plasma. By total inhibition of CE and TG transfer by 3-9F, LTP was shown to be responsible for net mass transfer of neutral lipids between lipoprotein classes in plasma, namely of CE from HDL to very low density lipoprotein (VLDL) and TG from VLDL to LDL and HDL. Selective inhibition of TG transfer by mAb 14-8H was also able to inhibit such net neutral lipid transfer. Such effect of these antibodies was demonstrated more remarkably in the presence of cholesterol esterification. Thus, TG transfer activity of LTP was shown to be required for net CE transfer, suggesting that net neutral lipid transfer in plasma between lipoproteins occurred mainly by a hetero-exchange mechanism. Inhibition of net neutral lipid transfer in plasma did not affect cholesterol esterification occurring predominantly on HDL. Consequently, mAb inhibition of TG transfer in plasma leads to CE accumulation in HDL. It is possible that hyperalphalipoproteinemia may be induced by a mutation in LTP that causes a selective defect in TG transfer activity.
In the plasma of certain vertebrates exists lipid transfer activity which can mediate the transfer of cholesteryl ester (CE)' between the cores of the various classes of lipoproteins ' The abbreviations used are: CE, cholesteryl ester; LTP, lipid trans--~~ ~ ~ (Ha and Barter, 1982). Immunochemically, it has been demonstrated that a single protein, lipid transfer protein (LTP), also known as cholesteryl ester transfer protein, of M, = 74,000 is responsible for the transfer activity of neutral lipid, mostly CE and triglyceride (TG) in human plasma (Hesler et al., 1988). The physiological role of LTP in lipoprotein metabolism and its relation to atherosclerosis have been understood further by the availability of neutralizing monoclonal antibodies (mAbs) to LTP and in vivo studies using it (Whitlock et al., 1989), evaluation of human subjects genetically deficient in LTP, and development of transgenic mice expressing LTP (reviewed by . These studies have demonstrated that LTP plays a role in transferring of CE out of high density lipoproteins (HDL) to apolipoprotein (apo) E-containing lipoproteins. Most plasma CE arises on HDL due to 1ecithin:cholesterol acyltransferase (LCAT) activity (Glomset, 19681, and transfer to apoB-containing lipoproteins allows for more efficient clearance from the circulation by receptor-mediated processes at the liver. These facts seem to be consistent with the hypothesis that, in the reverse cholesterol transport pathway whereby peripheral tissue cholesterol is removed by lipoproteins and returned to the liver for catabolism, LTP performs an important step as the mediator of CE transfer from HDL to apoB-containing lipoproteins. However, as a consequence of the reactions mentioned above, LTP lowers HDL and may raise low density lipoprotein (LDL) to give an undesirable impact on atherosclerosis based on the known positive and negative correlations of coronary heart disease risk with LDL and HDL cholesterol levels, respectively (Gordon and Rifkind, 1989). Thus, it is still unclear what the overall role of LTP is in plasma lipoprotein metabolism. Further understanding of the mechanism and role of the plasma LTP reaction is required especially in regard to the research of cardiovascular disease from the viewpoints of regulating LDL and HDL cholesterol levels and of atheromatous plaque prevention and regression.
It is proposed that LTP catalyzes a net CE transfer from HDL to other lipoproteins mainly by hetero-exchange with TG either by a direct equimolar exchange or by mediating transfer of both substrate lipids in each direction (Morton and Zilversmit, 1983;. I t is therefore important to study the relationship between CE and TG transfer by LTP in order to understand the mechanism of net neutral lipid transfer. Much information about structure-function relationships of LTP has been derived fer protein; TG, triglyceride; mAb, monoclonal antibody; HDL, high transferase; FCS, fetal calf serum; HAT, hypoxanthine-aminopterin-density lipoprotein; apo, apolipoprotein; LCAT, 1ecithin:cholesterol acylthymidine; LPDP, lipoprotein-deficient plasma; LDL, low density lipoprotein; PC, phosphatidylcholine; PAGE, polyacrylamide gel electrophoresis; DTNB, 5,5'-dithiobis(2-nitrobenzoic acid); VLDL, very low density lipoprotein. from studies of one mAb, TP2, originally shown to inhibit both human LTP-mediated CE and TG transfer (Hesler et al., 1988). The TP2 epitope was mapped to the COOH-terminal26 amino acids of LTP and binding of TP2 Fab fragment to LTP affected neutral lipid binding and transfer to similar degrees (Swenson et al., 1989). Point mutagenesis studies suggest that hydrophobic residues in this region are directly involved in neutral lipid binding (Wang et al., 19931, and a basic model of how this binding site is formed has been proposed . In addition, it was shown that the transfer of CE and TG by LTP is strictly competitive, although it seems to prefer CE (Ohnishi et al., 1994a). These results strongly indicate that LTP has a single or at least common binding sites for both CE and TG, which may have different affinity for the two substrates. Interestingly, Fukasawa et al. (1992) have reported establishment of mAbs against human LTP, one of which is inhibitory toward both CE and TG transfer and the other novel clone, LT-J1, dissociates CE and TG transfer. However, these mAbs were not further characterized for effects on other parameters of lipid transfer such as neutral lipid binding or lipoprotein binding to LTP.
Rabbit LTP is highly similar to the human counterpart in protein sequence, but there are differences, such as a 19-residue extension near the COOH-terminal in the rabbit protein (Nagashima et al., 1988) and perhaps functionally some difference in transfer specificity (Morton and Zilversmit, 1983) and distribution in plasma (Tall et al., 1987). We have purified rabbit LTP and shown that human and rabbit LTP are similar in specific activity of in vitro CE transfer (KO et al., 1993) but slightly different in their physicochemical properties (KO et al., 1993;Ohnishi et al., 1994b). Further effort was made to raise mAbs against rabbit LTP to study the mechanism of the LTP. Through use of different screening approaches, transfer assays, and a lipid particle binding assay, we hoped to obtain novel mAbs which might be usefkl probes for investigating the mechanism and physiological role of the LTP reaction further.
In this paper, we report the generation and partial characterization of novel inhibitory mAbs against rabbit LTP and the use of them to assess the role of LTP in net lipid transfer among lipoproteins in incubated human plasma. One mAb was capable of dissociating the two neutral lipid transfer functions of LTP, CE, and TG transfer by specifically inhibiting TG transfer. Using these mAbs, it was shown that the net transfer of CE between lipoprotein classes by LTP in relation to the LCAT reaction is likely to regulate plasma HDL cholesterol level. More interestingly, we tested, using the TG transfer-specific inhibitory mAb, the hypothesis that exclusive CE transfer by LTP does not result in the net transfer of CE from HDL to other lipoproteins and demonstrated that TG transfer is indeed required for such net transfer of CE, providing direct evidence for a "hetero-exchange" mechanism for net mass transfer of both lipids.
EXPERIMENTAL PROCEDURES Immunizations-Female 6-week-old BALB/c mice (University of Alberta Laboratory Animal Services) were immunized with active native rabbit LTP (approximately 3 pg) isolated as described (KO et al., 1993), combined with monophosphoryl lipid A plus trehalose dimycolate adjuvant (RIB1 Immunochem) by intraperitoneal injection. At week 3, a secondary injection (3 pg of LTP) was given and after detection in tailbleed serum (control serum was negative) at week 4 of positive inhibitory activity toward rabbit LTP mediated QH]CE transfer (below), mice were boosted again at week 6. At week 12, final boosting intraperitoneally and intravenously (3 pg of LTP with and without adjuvant, respectively) was performed 3 days before fusion.
Fusion and d b Production-Mouse myeloma cell line P3x63 Ag8 were cultured in myeloma medium (Dulbecco's modified Eagle's medium (low glucose), with 8% fetal calf serum (FCS) and 50 pg/ml gentamycin) at 37 "C in a CO, (7%) incubator, splitting at 1:40. On fie day of the fusion, 10' cells were collected (1,500 rpm for 10 min, benchtop centrifuge) in rinse medium (RPMI 1640 with 5% FCS, 50 p~ 2-mercaptoethanol (Sigma), and 50 pglml gentamycin). The mouse was sacrificed by cervical dislocation, and the spleen was excised aseptically, placed into a Petri dish, and washed with prewarmed rinse medium. The spleen was then homogenized in 5 ml of rinse medium using two sterilized frosted microscope slides, and splenocytes were collected and resuspended in 10 ml of rinse medium. The splenocytes and myeloma cells were combined in a sterile centrifuge tube (Corning), and the volume was adjusted to 50 ml with RPMI 1640 medium alone. The cells were collected and washed once more with RPMI medium before drying the pellet carefully. Then 1 ml of prewarmed 50% polyethylene glycol (Sigma) was added with gentle shaking, followed by 1-min standing. Then 500 pl of RPMI 1640 was added over 45 s, followed by 500 pl over 15 s, 5 ml over 5 min, and 15 ml over 5 min. One ml of FCS and 25 ml of hypoxanthine-aminopterin-thymidine (HAT) medium (WMI 1640 with 10% FCS, 50 p~ 2-mercaptoethanol, 50 pg/ml gentamycin, 1 x HAT (Sigma)) were then added, and cell density was adjusted to 2 x lo6 fused celldml with further HAT medium before aliquoting 100 pl/well into sterile 96 well plates (Corning). Next day, 150 pL of HAT medium was added; subsequently, media changes of 170 pl were performed twice a week for 2 weeks with HAT medium, followed by subculturing and expansion in HT medium (HAT medium, substituting 1 x hypoxanthine-Thymidine (HT, Sigma) for HAT) for 4 weeks, during which preliminary screening by inhibition assay or dot blotting (below) was performed. Positive wells were cloned by limiting dilution at least three times with a splenocyte feeder cell method. Large amounts of individual mAbs were produced by intraperitoneal injection of clonal hybridomas in phosphate-buffered saline (3 x lo6 cells) into pristane (Sigma) primed BALB/c mice and harvesting of serum and ascites fluid 1-2 weeks later. Antibodies (IgG) were purified by 25-50% ammonium sulfate precipitation and protein A-Sepharose 4 (Pharmacia Biotech Inc.) chromatography under the high salt condition (Harlow and Lane, 1988).
Inhibition Assay Screening and Immunotitration Studies-Inhibition assay screening and immunotitration studies evaluated the ability of mAbs to inhibit I3H1CE or L3HlTG transfer mediated in vitro by rabbit or human lipoprotein depleted plasma (LPDP) as an LTP source. Rabbit and human LPDP were prepared by d = 1.25 NaBr centrifugation of fresh human or previously frozen rabbit (Pelfreez, Rogers, A R ) plasma in a Ti 70 rotor (Beckman) at 184,000 x g for 60 h followed by removal of floated lipoproteins and dialysis of the infranatant with LTP buffed1 m~ ethylenediaminetetraacetic acid (EDTA) (KO et al., 1993). For screening of tailbleed serum and hybridoma supernatants, 25 pl of rabbit LPDP was pretreated with diluted sera or supernatant for 2 h at 4 "C before performing our standard assay to measure r3H]CE transfer from human LDL to HDL (KO et al., 1993) on the mixture. Wells secreting antibodies inhibitory toward 13HlCE transfer were cloned by limiting dilution, yielding clones 2-8G and 3-9F.
For the immunotitration studies, various amounts of IgG fraction were preincubated with 25 pl of rabbit or 50 pl of human LPDP for 2 h at 4 "C before performance of L3HlCE (above) or L3HITG transfer assay on the mixture (final volume 250 pl). For the L3HlTG transfer assay, isotopically labeled LDL donor was prepared as described for L3HJCE donor particles (Nishikawa, et al., 1986;KO et al., 1993), except substituting vesicles prepared with 0.5 mCi of [3H]glycerol trioleate (Amersham Corp.) and 4.5 mg of phosphatidylcholine (PC) (Avanti Polar Lipids) in the method. Final TG concentration and specific radioactivity of radiolabeled LDL thus obtained was 1.27 m~ and 26,600 dpdnmol, respectively, and L3HlTG assays contained 10 pg of TG in LDL and the same amount of HDL (40 pg of total cholesterol, 3.7 pg of TG) as for the L3HlCE transfer assays in a final volume of 250 pl.
Dot Blot Screening-Hybridoma culture supernatants (2 pl) were spotted onto nitrocellulose membranes (Bio-Rad) which had been preabsorbed with pure native rabbit LTP (by soaking in 1-5 pg of LTP/cm2 for 1 h) and then blocked with 5% milk in this buffered saline (KO et al., 1993). Adherent antibodies were detected by treatment of the membrane with horseradish peroxidase-linked sheep anti-mouse I& (Amersham) and fluorography (KO et al., 1993). Wells secreting antibodies recognizing LTP were cloned by limiting dilution using the dot blot screen, yielding clone 14-8H.
Pyrene-Lipid Dansfer Assays-'hnsfer of pyrene-CE and -TG were measured between artificially prepared and apoh-I-activated PCPTG microemulsions of about 25 nm, approximating the size of LDL, from donor to a 9-fold excess of acceptor microemulsions Ohnishi et al., 1994a). When mAbs were tested, they were preincubated with isolated rabbit (KO et al., 1993) and human (Ohnishi et al., 1990) LTP for at least 30 min at 4 "C before performing the assay.

Net Lipid Dansfer
Gel Electrophoresis and Zmmunoblotting-Polyacrylamide gel electrophoresis (PAGE) in the presence of sodium dodecyl sulfate (SDS) was performed as described previously (KO et al., 1993). Nondenaturing PAGE was performed in the absence of SDS using the same mini-gel apparatus and a 4% acrylamide stacking, 8% separating gel run in 89 m~ Tris-HC1, 89 m M boric acid, and 2 m M EDTA (pH 8.4) at 120 V constant voltage for 1 h. The sample buffer was 2 x concentrated buffer above containing 20% glycerol and 0.02% bromphenol blue, and samples were mixed 1:l (v:v) before loading directly to the gel. Electrophoretic transfer to nitrocellulose membranes and immunostaining/ fluorography after either procedure were performed as described previously (KO et al., 1993), substituting various mAbs as the primary antibody.
Microemulsion Binding Analysis-The effects of mAb treatment on LTP binding to small PCmG microemulsions (Tajima et al., 1983;KO et al., 1993) was assessed by preincubating 2 pg of rabbit or human LTP with 20 pg of various IgG fractions for 30 min at 4 "C before addition of LTP, 0.5% ovalbumin (Sigma) with or without microemulsion (100 pg in TG and 70 pg in PC) in a total volume of 200 pl and incubation at 37 "C for 15 min. The samples were then centrifuged and separated as described previously (KO et al., 1993) before lop1 aliquots were analyzed by SDS-PAGE immunoblotting as described above, using mAb 14-8H (recognizes native and denatured rabbit and human LTP) as the primary antibody (30 pg of IgG/ml) to immunostain LTP in all fractions. Radioisotopic Assays of Neutral Lipid Tkansfer in Plasma-Human plasma mediated isotopic neutral lipid transfer from LDL to HDL was measured employing our in vitro assay for [3H]CE or [3H]TG transfer activity mentioned above using 50 pl of plasma. The transfer of isotopic neutral lipids from HDL within human plasma was determined using [3H]CE and c3H1TG-labeled HDL isolated as the flow-through material from the dextran sulfate-cellulose column during preparation of the respectively labeled LDL (Francis et al., 1991) after ultracentrifugal flotation (d < 1.21 g/ml fraction) and dialysis. Either radiolabeled HDL (30 pg total cholesterol) was added to 0.5 ml of human plasma, thus increasing HDL cholesterol of the sample by about 10%. At the indicated times, 150-pl aliquots were removed and placed on ice before total and HDL radioactivity were measured. HDL radioactivity was determined in the supernatant following treatment of plasma with heparin (250 unitdml) and MnCl, (0.09 M) (Tall et al., 19871, and apoB containing lipoprotein associated radioactivity was the difference between total and HDL counts.
Plasma Incubation Experiments-Fasting (12 h) plasma was obtained from 6 healthy subjects (three male, three female) by collection of blood into 0.1% EDTA and 4 "C centrifugation. Care was taken to cool the blood as rapidly as possible, including immersion of the catheter tubing in ice water during blood collection. Plasma was then incubated at 37 "C for 24 h in the absence or presence of mAb. An aliquot was kept at 4 "C, a temperature at which LTP and LCAT are inactive, serving as the control. Some plasma was treated with 2 m M 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) before storing 2 ml at 4 "C. Three other DTNBtreated portions were then either untreated or pretreated with mAb (64 pg of IgG/ml) before 37 "C incubation. The same storage and treatments were also performed on plasma samples without chemical modification by DTNB. Following incubations, all samples were transferred to ice, and very low density lipoprotein (VLDL) (0.25 ml) was isolated by ultracentrifugation at d = 1.006 g/ml in a TL-100 rotor (Beckman) at 100,000 rpm for 2 h at 4 "C followed by suction of the d > 1.006 g/ml fraction (1.75 ml) from the tube bottom. HDL was isolated from the d > 1.006 g/ml fraction as the supernatant following heparidMnC1, treatment (above To measure cholesterol esterification, intact plasma was untreated or pretreated with mAb before incubation at 37 "C. At the indicated time

Mechanism by LTP
intervals, aliquots were removed onto ice, followed by heparin/MnCI, treatment (above) to separate HDL from apoB-containing lipoproteins.
CE content in total plasma and HDL was determined (above), and values for apoB-containing lipoproteins were obtained by difference. Analytical Methods-Protein was determined by Lowry assay (Lowry et al., 19511, with bovine serum albumin (Pierce Chemical Co.) as a standard, except for purified LTP, which was determined by BCA assay (Pierce) with the same standard. TG was measured with an enzymatic kit (Wako), and IgG subclassing was performed with an mAb isotyping kit (Pierce) using their antigen-independent screening protocol.
Other Materials-For cell culture, Dulbecco's modified Eagle's medium, RPMI 1640, FCS, and gentamycin were from Life Technologies, Inc. Other reagents were obtained as specified or generally from Sigma, Bio-Rad, or Fisher. P3x63 Ag8 cells were kindly supplied by Dr. Tim Mosmann (Department of Immunology, University ofAlberta), and TP2 mAb was a generous gift from Dr. Yves Marcel (University of Ottawa Heart Institute) (Hesler et al., 1988).

Raising of Inhibitory Monoclonal Antibodies against Rabbit
LTP-Hybridoma culture supernatants were initially screened by two methods in order to detect different antibodies against LTP. First, supernatants were screened for ability to inhibit in vitro [3HlCE transfer between lipoproteins mediated by rabbit LTP in order to obtain antibodies affecting LTP function, yielding clones 2-8G and 3-9F. Second, supernatants were screened by dot blotting, initially in hope of obtaining antibodies which would recognize but might not inhibit LTP, yielding clone 14-8H. The 14-8H was negative in the inhibition assay screen. Surprisingly, the supernatants from the inhibitory clones bound only very weakly in the dot blotting screen (not shown). Isotyping of the supernatants revealed that all three hybridoma clones secreted mAbs of the IgG, subclass with K light chains. Subsequently, the clones were used for production of mouse ascites fluid, and the IgG fraction was purified and used for the rest of the studies.
Characterization of the Monoclonal Antibodies-Zn vitro immunotitration studies were carried out to examine the effect of all three mAbs on neutral lipid transfer. r3H]CE transfer from LDL to HDL mediated by rabbit LTP was inhibited by 2-8G and 3-9F, two mAbs originally obtained by inhibition screening, in a dose-dependent manner, whereas 14-8H, the mAb obtained by dot blotting, had little effect (Fig. 1). Although 2-8G inhibited rabbit LTP, it was unable to inhibit r3H]CE transfer mediated by human LTP (Fig. 1). When r3H]TG transfer from LDL to HDL was analyzed, in addition to 2-8G and 3-9F, 14-8H was also able to inhibit rabbit LTP in a dose-dependent manner (Fig. 1). Once again, 2-8G had no effect on human LTP, whereas 3-9F and 14-8H did cross-react (Fig. 1). An alternative assay system measuring the transfer of pyrene-containing neutral lipid analogs between artificially prepared microemulsions was then used to further characterize the effects of these mAbs on the LTP reaction. In this system, LTP is active only in the presence of helical apolipoproteins, i e .
apoA-I . The results are listed in Table I as the apparent first order rate constant k l L , determined from the initial slopes of log transformation of the data as described previously (Milner et al., 1991;Ohnishi and Yokoyama, 1993). In this assay, all three mAbs inhibited both CE and TG transfer mediated by LTP with 2-8G again being species specific for rabbit LTP. Curious results were obtained for 14-8H, which did not inhibit isotopic CE transfer between lipoproteins (Fig. 11, but did inhibit pyrene-CE transfer between artificial lipid microemulsions. It was noted earlier that 2-8G and 3-9F gave a weak signal during dot blotting to immobilized rabbit LTP, whereas 14-8H reacted strongly. SDS-PAGE immunoblotting with each mAb against purified rabbit and human LTP also revealed strong binding by 14-8H to denatured LTP, whereas binding of 2-8G  (50 111) LPDP were treated with the indicated amounts of IgG fraction containing mAb 2-8G (squares), 3-9F (diamonds), or 14-8H (circles) before performing in vitro L3H1CE or L3H1TG transfer assay between lipoproteins on the mixtures. In the absence of mAb, control rabbit LTP CE transfer activity was 2403 d p d 2 5 pl/h and control human LTP CE transfer activity was 2352 d p d 5 0 pVh, and control rabbit LTP TG transfer activity was 1736 d p d 2 5 pl/h, and control human LTP TG transfer activity was 1223 d p d 5 0 pVh. The results are representative of three separate experiments.

TABLE I Apparent first order rate constants for the transfer of pyrene-lipids by human and rabbit LTP between lipid microemulsions in the presence
of mAbs Pure rabbit and human LTP (0.5 pg for pyrene-CE transfer and 1 pg for pyrene-TG transfer, respectively) was treated with 5 pg of IgG fraction containing mAb 2-8G, 3-9F, or 14-8H before performing pyrene-CE or -TG transfer assay in the presence of apoA-I between microemulsions. Details of the experimental procedure and the calculation of the rate constant are available in the reference by Ohnishi et al. (1994a and 3-9F were hardly detected (Fig. 2). Apparent also are the slight differences in M, between rabbit (74,000) and human LTP (66,00049,000) (KO Ohnishi et al., 1990).
Recognition of denatured rabbit and human LTP by TP2, an mAb raised against human LTP (Hesler et al., 1988) also validates that 14-8H recognizes LTP authentically (Fig. 2, lanes 11  and 12). In contrast, nondenaturing PAGE immunoblotting was able to demonstrate interaction of all three clones to rabbit LTP, as well as the lack of recognition of the human LTP by 2-8G (Fig. 3). The interaction of 3-9F was slightly weaker toward human LTP than rabbit LTP. In nondenaturing PAGE, both LTPs migrated slower than bovine serum albumin (66 kDa) but faster than transferrin (81 m a ) , behavior consistent with LTP being a monomer under these conditions in agreement with our sedimentation equilibrium result (Ohnishi et al., 1994b). Overall, the PAGE results showed that 14-8H binds to  lanes 1-4) or immunostaining of the LTP following electrophoretic transfer to nitrocellulose membrane (lanes 5-12 ). Lunes 1 and 4 contain Bio-Rad SDS-PAGE molecular weight standards (low range) and an epitope resistant to denaturation by SDS. The lack of binding of 2-8G and 3-9F to denatured antigen suggests that their interaction with LTP may be toward more complex epitopes formed by LTP in its native protein conformation.
We previously showed that both rabbit and human LTP bind to lipid microemulsions strongly ( KO et al., 1993;Ohnishi et al., 1994b). Therefore, the effects of our mAbs on the association of LTP with microemulsion was studied in order to examine the modes of their inhibitory action. Microemulsion binding of LTP was estimated by SDS-PAGE immunoblotting (Fig. 4). Consistent with the previous results, a major portion of rabbit or human LTP partitions to the bottom half fraction when centrifuged in the absence of lipid (Fig. 4, leftmost columns). In the presence of the microemulsion, LTP becomes almost entirely recovered in the top half fraction upon centrifugation (Fig. 4, second from left columns), also consistent with previous results (KO Ohnishi et al., 1994b). When rabbit LTP is pretreated with 2-8G, appreciable LTP was recovered in the bottom half fraction in the presence of lipid, indicating that 2-8G, the rabbit-specific combined CEmG transfer inhibitory mAb, interfered with lipid association of rabbit LTP (Fig. 4 A , central column). On the other hand, with human LTP, 2-8G had no effect on the lipid association of LTP (Fig. 4B, central column), consistent with its species specificity toward rabbit LTP demonstrated in lipid transfer inhibition assays and nondenaturing PAGE immunoblotting. mAb 3-9F completely blocked lipid association of both rabbit or human LTP with microemulsion (Fig. 4, second from right columns). On the other hand, 14-8H, the CE/TG transfer dissociating antibody, seemed to have no effect on LTP lipid association (Fig. 4, rightmost (Fig. 51, consistent with the results of titration (Fig. 1) (Fig. 6). In this experimental condition, VLDL in plasma provided an additional exchangeable pool of TG for CE in HDL, approximately 35 molar percent of the CE in apoB-containing lipoproteins, expanding capacity of the acceptors for radiolabeled CE in this assay. Eliminating this additional pool by 14-8H may have lead to less appearance of [3H]CE in the acceptor fraction from HDL. Nevertheless, 14-8H allowed CE transfer to occur between lipoproteins while completely inhibiting TG transfer, whereas 3-9F completely blocked both transfer of CE and TG. From the rate of ["HICE transfer, it can be estimated that much more than 90% of CE molecules in each lipoprotein fraction should undergo the LTPmediated transfer during an incubation for 24 h (below), regardless of the presence of 14-8H Jones, 1979, 1980;Kurasawa et al., 1985).
Effect of Monoclonal Antibodies on Net Neutral Lipid Dunsfer in Human Plasma-Human plasma, with LCAT inactivated by DTNB treatment, was incubated at 37 "C for 24 h to demonstrate net CE and TG transfer among the various lipoprotein classes in absence of new CE synthesis, compared with 4 "C controls (Table 11). Following the incubation period, although HDL-CE decreased, this was more than compensated for by increases in VLDL and LDL-CE, both of which may be overestimated as suggested by the larger errors associated with these measurements. The higher error in LDL CE values in all the conditions is unexplained and occurred in a similar study (Yen et al., 1989) and would account largely for the apparent increase in total CE values of all the incubated groups. The lack of change in freeltotal cholesterol ratios among all the groups would suggest that LCAT was effectively inhibited in this experiment. A decrease in VLDL TG, on the other hand, was well compensated for by increases in LDL and HDL TG levels. As well, overall net increase of core neutral lipid content was observed in LDL, but not in VLDL or HDL, where an increase in one neutral lipid was compensated by an equimolar decrease in the other, following plasma incubation. Pretreatment of DTNB treated human plasma with either mAb 3-9F or 14-8H practically abolished the neutral lipid transfer among the lipoprotein classes normally seen upon incubation, including the changes in LDL total core lipid content. Thus, LTP activity in the absence of CE generation by LCAT leads to net transfer of CE from HDL to VLDL and perhaps LDL, with TG transfer from VLDL to LDL and HDL. mAb 3-9F, which inhibits both radioisotopic CE and TG transfer mediated by LTP, was able to abolish these changes. mAb 14-8H, which permits transfer of CE molecules and not TG among different lipoprotein classes in plasma, also abolishes such net transfer of CE and TG among the lipoproteins.
Effect of Monoclonal Antibodies on Net Neutral Lipid Dunsfer in Plasma with Active LCAT- Table I11 shows the results of a similar experiment as described in Table 11, performed in the absence of DTNB. Following 37 "C incubation for 24 h of intact plasma, compared with the 4 "C control, VLDL-TG dropped markedly, accompanied by increases in LDL and HDL-TG levels. CE levels increased in all the lipoprotein classes, although the change in LDL was not significant. The increase in HDL-CE reflects the net result of synthesis of CE by LCAT and removal of CE by LTP. This role of LTP in mediation of net CE within incubated human plasma ( n = 3). Squares, untreated plasma; diamonds, mAb 3-SF-treated plasma; circles, mAb 14-8Htreated plasma. mass transfer out of HDL is further illustrated when lipid transfer is blocked by either mAb 3-9F that inhibits both CE and TG transfer or 14-8H that selectively inhibits TG transfer. Both antibodies practically abolished any net neutral lipid changes in VLDL and LDL as well as changes in HDL-TG levels. In the presence of either antibody, the increase of CE was almost exclusively in HDL resulting in its marked build up in HDL, showing that CE transfer out of HDL is prevented. Thus, the experiments in the presence of cholesterol esterification demonstrated in a more prominent manner than its absence that permitting LTP to transfer only CE but not TG between lipoproteins prevents net removal of CE from HDL. Again, core neutral lipid content was increased in LDL upon plasma incubation with LTP active, and both mAbs inhibited this change.
Cholesterol esterification was observed upon incubation of intact human plasma at 37 "C for 24 h ( Table 111). The results also indicated that no cholesterol esterification activity on VLDL or LDL, since the increase of CE was only in HDL when LTP was completely inhibited, in contrast to a previous report by Yen et al. (1989). However, the presence of either mAb did not affect the total cholesterol esterification in plasma, indicating that LTP-mediated neutral lipid transfer, including CE transfer out of HDL, does not regulate the LCAT reaction. This was further demonstrated by an initial time course of choles-  Table I11 are consistent with the notion that DTNB treatment in Table I1 was effective in specifically inhibiting LCAT without affecting LTP function, which may not be the case for all LCAT inhibitors, such as p-chloromercuriphenylsulfonate (Hopkins and Barter, 1980;Morton and Zilversmit, 1983).

DISCUSSION
Three mAbs were generated against rabbit LTP: 2-8G, 3-9F, and 14-8H. Characterization of their effects on neutral lipid transfer between lipoproteins by LTP revealed that 3-9F inhibited both isotopic CE and TG transfer, whereas 14-8H inhibited only TG transfer, both cross-reacting with human LTP. These two antibodies were therefore used to study total inhibition and dissociation of CE and TG transfer by LTP in human plasma. The conclusions from the present study are: 1) LTP is responsible for net transfer of CE from HDL to VLDL and of TG from VLDL to LDL and HDL; 2) inhibition of CE and TG transfer completely blocked net transfer of these neutral lipids; 3) inhibition of only TG transfer but not CE transfer also resulted in almost complete blockade of all net transfer of the neutral lipids among lipoproteins; and 4) LCAT esterifies cholesterol only in HDL, and neither types of LTP inhibition influenced the esterification reaction. Thus, it was demonstrated that TG transfer activity of LTP is essentially required for LTP to mediate net CE transfer from HDL.
Our combined CEEG transfer inhibitory mAbs 2-8G and 3-9F appear to be novel in that they recognize a native epitope, and their binding to LTP interferes with its association with lipid microemulsions. mAb 2-8G is species-specific for rabbit LTP in this reaction as well. In contrast, the well studied mAb TP2 which also inhibits both CE and TG transfer mediated by LTP recognizes an epitope stable to denaturation and binding of mAb increases LTP binding to lipoproteins . More information about the epitope location of any of our mAbs is unavailable at this moment, but the difference in effects of our mAbs on binding of LTP to lipid particle surfaces compared with TP2 suggest that they may bind in a different location than TP2. Thus, our mAbs may not recognize the COOH-terminal region of LTP identified as the TP2 epitope, but perhaps some other region crucial for lipid transfer function. One candidate is a putative "interfacial recognition region" away from the TP2 epitope (Swenson et al., 1989) involved in lipoprotein Effects of incubation of human plasma with LCAT inhibited, and with or without anti-LTP mAb 3-9F or 14-8H, on lipoprotein lipid composition Plasma samples (n = 6) were pretreated with DTNB and antibody, then incubated and treated as described under 'Materials and Methods." All values are in milligramddl as mean (& S.E.) and statistically significant differences between 4 "C control values are shown.

TABLE I11
Effects of incubation of human plasma containing active L C G and with or without anti-LTP mAb 3-9F or 14-8H, on lipoprotein lipid composition Plasma samples (n = 6) were pretreated with antibody, then incubated and treated as described under "Materials and Methods." All values are in milligramddl as mean (2 S.E.) and statistically significant differences between 4 "C control values are shown. binding (Pattnaik and Zilversmit, 1979;Morton, 19851, which is perhaps a hydrophilic helix containing a lysine triplet (Wang et al., 1991). mAbs 2-8H and 3-9F may be binding to or near such a region to disrupt LTP lipoprotein binding as well as total neutral lipid transfer, isotopic or pyrene labeled. In any case, the major conclusions of this study do not depend on identification of the epitopes where the mAbs bind. There is good evidence that a single neutral lipid binding site involved in lipid transfer is associated with the COOH-terminal region of LTP . The fact that CE and TG compete for transfer (Morton and Zilversmit, 1983;Ohnishi et al., 1994a) is consistent with such a hypothesis. Dissociation of CE and TG transfer by LTP was somehow demonstrated by treatments with mercurial compounds Barter, 1980, 1982;Morton and Zilversmit, 1982) and by TP2 Fab fragments (Swenson et al., 1989). In light of the information about the role of the LTP COOH-terminal, a likely explanation is that transfer of the larger substrate, TG, is more susceptible to inhibition by certain treatments for steric reasons. This may be supported by the finding that the CE molecule has higher apparent affinity for LTP than TG (Ohnishi et al., 1994a). Thus, our 14-8H and the LT-Jl of others (Fukasawa et al., 1992) may act like TP2 Fab fragment in a more extreme manner. Pyrene-CE transfer was inhibitable by 14-8H, likely due to the increased bulk of the pyrene group limiting its access to the neutral lipid binding site. mAb 14-8H did not interrupt association of LTP with lipid microemulsions, a condition possibly necessary (but not sufficient) for LTP-mediated lipid transfer within the pyrene assay system. Presumably, 14-8H does not decrease binding of lipoprotein by LTP and allows the transfer mechanism to proceed, except that only transfer of CE can occur.
The use of these novel inhibitory mAbs raised against rabbit plasma LTP permitted us to investigate the role of LTP in human plasma lipoprotein metabolism and the basic mechanism of neutral lipid transfer mediated by LTP. The two mAbs employed in this study were 3-9F and 14-8H, which both recognize and inhibit human as well as rabbit LTP. In normolipidemic fasting plasma incubated at 37 "C, there was net transfer of CE from HDL to VLDL (and perhaps LDL) and of TG from VLDL to LDL and HDL. These changes indicate that the neutral lipid pools distributed among the various lipoprotein classes in plasma are subject to net re-distribution by some specific (protein-mediated) or nonspecific (i.e. collision or diffusion) processes when isolated plasma is incubated. The directions of the neutral lipid mass changes are consistent with the reverse cholesterol transport pathway, as well as with an equilibration of the cores of the various lipoprotein classes according to their neutral lipid (i.e. TG/CE) compositional ratios (Table 11). Treatment of plasma with 3-9F, the mAb which to-taIly inhibits isotopic CE and TG transfer, blocked the net transfer of neutral lipid between the lipoprotein classes normally seen upon plasma incubation, thus implicating plasma LTP entirely in this process. mAb 14-8H, which strongly inhibited isotopic TG transfer, also blocked the net transfer of both CE and TG (Table 11). The fact that selective inhibition of TG transfer is sufficiently able to abolish net mass transfer of CE (and TG) is consistent with a mechanism whereby CE and TG transfer are both required for net neutral lipid transfer as the result of reciprocal heteroexchange (Morton and Zilversmit, 1983). Thus, when one transfer function, TG transfer, is inhibited, and assuming that the mAb treatment did not affect the mechanism of LTP, LTP is converted to being a transfer protein only able to mediate CE homo-exchange and thus no longer CEPTG hetero-exchange, resulting in no net neutral lipid transfer by LTP. In this condition isotopic CE transfer strictly represents exchange transfer of CE molecules that can occur without net CE mass transfer.
The precise mechanism of LTP remains controversial, with kinetic evidence for either carrier (Barter and Jones, 1980) or ternary complex mediated processes (Ihm et al., 1982). mAb 14-8H blocked pyrene-lipid transfer between lipid microemulsions but did not affect LTP-microemulsion interaction. It has also been shown that the presence of activator (apoA-I) of the transfer reaction does not affect LTP binding to lipid (Ohnishi et al., 1994b). Therefore, the apparent stable association of LTP with lipid, which otherwise may support a ternary complex mechanism, is not enough for the transfer of lipid.
The results of this study also provided strong evidence for inter-relationship of CE and TG transfer mediated by LTP, which was first indicated by using reconstituted lipoproteins (Morton and Zilversmit, 1983). It has been clearly demonstrated in this work that, by selectively eliminating TG transfer activity of LTP, net transfer of CE was almost completely inhibited between lipoproteins in human plasma, especially between HDL and VLDL. In other words, the net CE transfer function of LTP within plasma is closely linked to its TG transfer function, since loss of TG transfer activity (and not of CE transfer activity) is accompanied by loss of net neutral lipid transfer capability under the conditions of this study. Presumably, specifically inhibiting CE transfer would also have the same result, although this has not been experimentally demonstrable using chemical modification reagents or mAbs to date.
Somewhat inconsistent results were seen for the LDL fraction, in which net expansion of core neutral lipid content, mainly due to an increase in TG, seemed to occur as a result of LTP action during plasma incubation (Tables I1 and 111). For this to occur, LTP must be able to mediate direct mass transfer under some conditions, whereby it is able to deliver TG to LDL without picking up another lipid. Nevertheless, this process was also inhibitable by either mAb 3-9F or 14-8H.
Our mAbs also allowed us to investigate the effects of inhibition of net neutral lipid transfer on plasma LCAT activity. It was demonstrated that inhibition of net neutral lipid transfer between lipoproteins did not influence plasma LCAT activity, which was shown to reside predominantly on HDL. During incubation of plasma with active LCAT, increase of CE was found in all lipoprotein fractions but relatively more in HDL, indicating that the lipid transfer does not completely overcome the esterification reaction. However, a large portion of the CE increment generated by LCAT is transferred to apoB-containing lipoproteins and would thus increase clearance of HDL-CE from plasma in uiuo. The fact that TG transfer-specific inhibition of LTP causes CE accumulation in HDL suggests that screening of hyperalphalipoproteinemia should be done through both CE and TG transfer assays, because there is a possibility for a genetic defect of LTP selectively in TG transfer causing such a lipoprotein abnormality.