The Mechanism of the Remodeling of High Density Lipoproteins by Phospholipid Transfer Protein*

Phospholipid transfer protein (PLTP) remodels high density lipoproteins (HDL) into large and small particles. It also mediates the dissociation of lipid-poor or lipid-free apolipoprotein A-I (apoA-I) from HDL. Remodeling is enhanced markedly in triglyceride (TG)-en-riched HDL (Rye, K.-A., Jauhiainen, M., Barter, P. J., and Ehnholm. C. (1998) J. Lipid. Res. 39, 613–622). This study defines the mechanism of the remodeling of HDL by PLTP and determines why it is enhanced in TG-en-riched HDL. Homogeneous populations of spherical reconstituted HDL (rHDL) containing apoA-I and either cholesteryl esters only (CE-rHDL; diameter 9.3 nm) or CE and TG in their core (TG-rHDL; diameter 9.5 nm) were used. After 24 h of incubation with tography a Q Fast Flow Biotech) attached to an FPLC system (17). This gave homogeneous preparations of apoA-I, which appeared as a single band after electrophoresis on a 20% SDS-polyacrylamide gel (Phast System; Amersham Pharmacia Biotech) and staining with Coo- massie Blue. Isolation of Lecithin:Cholesterol Acyltransferase— Lecithin:choles- terol acyltransferase was isolated from samples of pooled, human plasma as described (18, 19). Activity was assessed using 1-palmitoyl- 2-oleoyl phosphatidylcholine (POPC)/unesterified cholesterol (UC)/ apoA-I discoidal rHDL labeled with [1 a ,2 a - 3 H]cholesterol ([ 3 H]UC) (Amersham Pharmacia Biotech) as the substrate (20). The assay was linear when less than 30% of the [ 3 H]UC was esterified. The prepara- tion used in this study generated 2.36 m mol of CE/ml lecithin:choles-terol acyltransferase/h. Isolation of CETP— CETP was prepared as described (21). Transfer activity was assessed as the transfer of [ 3 H]CE from [ 3 H]CE-HDL 3 to low density lipoproteins (22, 23). The assay was linear when less than 30% of the [ 3 H]CE transferred from HDL 3 to low density lipoproteins. Activity is expressed in units/ml, with 1 unit being the transfer activity of 1 ml of a preparation of pooled, human lipoprotein-deficient plasma. The CETP used in this study had 22 units of activity/ml. Isolation of PLTP— PLTP was isolated from samples of pooled hu- man plasma by ammonium sulfate precipitation, ultracentrifugation, Macroprep Hydrophobic Interaction chromatography (Bio-Rad), DEAE- Sepharose Fast Flow chromatography (Amersham Pharmacia cross-linking rHDL suberate the rHDL were isolated by ultracentrifugation and subjected to nonde- naturing polyacrylamide gradient gel electrophoresis. The profiles represent scans of Coomassie-stained gels. Particle diameters were calcu- lated by reference to known high molecular weight standards. i , PLTP mediates the fusion of two rHDL particles, each of which contain three molecules of apoA-I, to give an unstable particle with six apoA-I molecules. The fusion product either ( ii ) rearranges into three small particles with two molecules of apoA-I/particle or ( iii ) is converted into a large particle with four molecules of apoA-I in a process that is accompanied by the dissociation of two molecules of apoA-I.

tography on a column of Q Sepharose Fast Flow (Amersham Pharmacia Biotech) attached to an FPLC system (Amersham Pharmacia Biotech) (17). This gave homogeneous preparations of apoA-I, which appeared as a single band after electrophoresis on a 20% SDS-polyacrylamide gel (Phast System; Amersham Pharmacia Biotech) and staining with Coomassie Blue.
Isolation of CETP-CETP was prepared as described (21). Transfer activity was assessed as the transfer of [ 3 H]CE from [ 3 H]CE-HDL 3 to low density lipoproteins (22,23). The assay was linear when less than 30% of the [ 3 H]CE transferred from HDL 3 to low density lipoproteins. Activity is expressed in units/ml, with 1 unit being the transfer activity of 1 ml of a preparation of pooled, human lipoprotein-deficient plasma. The CETP used in this study had 22 units of activity/ml.
Isolation of PLTP-PLTP was isolated from samples of pooled human plasma by ammonium sulfate precipitation, ultracentrifugation, Macroprep Hydrophobic Interaction chromatography (Bio-Rad), DEAE-Sepharose Fast Flow chromatography (Amersham Pharmacia Biotech), and heparin-Sepharose Fast Flow chromatography (Amersham Pharmacia Biotech) (9). These procedures were carried out at room temperature on an FPLC system. PLTP activity was determined as the transfer of 1,2-di-[1-14 C]palmitoyl L-3-phosphatidylcholine (Amersham Pharmacia Biotech) from 1,2-di-[1-14 C]palmitoyl L-3-phosphatidylcholine-labeled small unilamellar egg PC vesicles to HDL (24). The activities of the PLTP preparations used in this study ranged from 1.7 to 3.1 mol of phospholipid transferred/ml of PLTP/h. Activities in individual incubations are presented in the figure legends. The PLTP preparations were free of CETP and lecithin:cholesterol acyltransferase activities and appeared as single bands after electrophoresis on a 20% SDS Phast gel (Amersham Pharmacia Biotech) and silver staining.
Preparation of rHDL-Discoidal rHDL containing POPC, UC, and apoA-I were prepared by the cholate dialysis method (25) and converted into spherical CE-rHDL by incubation at 37°C for 24 h with low density lipoproteins and lecithin:cholesterol acyltransferase (26). The spherical CE-rHDL were isolated by ultracentrifugation at 4°C in the 1.07 Ͻ d Ͻ 1.21 g/ml density range, with two 24-h spins at the lower density and a single 16-h spin at the upper density (15). The 1.07 g/ml spins were carried out at 55,000 rpm using a Beckman Ti 55.2 rotor in a Beckman L8-M ultracentrifuge. The 1.21 g/ml spin was carried out at 100,000 rpm using a Beckman TLA 100.4 rotor in a Beckman TL-100 Tabletop ultracentrifuge. The CE-rHDL were dialyzed extensively against Trisbuffered saline (10 mM Tris-HCl, 150 mM NaCl, pH 7.4) containing 1 mM EDTA-Na 2 and 0.01% (w/v) NaN 3 (TBS) before use.
Spherical CE-rHDL were enriched with TG by incubation at 37°C for either 2 or 20 min with the phospholipid/triglyceride emulsion, Intralipid (KabiVitrum AB, Stockholm, Sweden) and CETP (9). Under these conditions, 12 and 32%, respectively, of the rHDL CE were replaced by TG. The TG-rHDL were isolated by ultracentrifugation as described for CE-rHDL and dialyzed against TBS before use.
Remodeling of rHDL by PLTP-All incubations were carried out in stoppered plastic tubes in a shaking water bath maintained at 37°C. Details of the individual incubations are presented in the figure legends. When the incubations were complete, the rHDL were isolated by ultracentrifugation in the 1.063 Ͻ d Ͻ 1.25 g/ml density range with a single spin at the upper and lower densities. The spins were conducted at 100,000 rpm for 16 h at 4°C using a TLA 100.4 rotor in a Beckman TL-100 Tabletop ultracentrifuge. The rHDL were concentrated in a CF25 membrane cone (Amicon®) and then applied to a prepacked HR 10/30 Superose-6 column (Amersham Pharmacia Biotech) attached to an FPLC system. The rHDL were eluted from the column at a flow rate of 0.3 ml/min. Fractions were collected at 1-min intervals.
Time Course of the Remodeling of rHDL by PLTP-CE-rHDL and TG-rHDL were either maintained at 4°C, incubated at 37°C for 24 h in the absence of PLTP, or incubated at 37°C for 1, 3, 6, 12, and 24 h in the presence of PLTP. Aliquots of the unprocessed incubation mixtures were subjected to nondenaturing 3-40% polyacrylamide gradient gel electrophoresis, transferred electrophoretically to nitrocellulose membranes, and immunoblotted with polyclonal sheep anti-human apoA-I antiserum (Roche Molecular Biochemicals). The transferred bands were detected by ECL (Amersham Pharmacia Biotech) (27).
Phospholipid Transfer Studies-To study phospholipid transfers from vesicles to rHDL, [ 14 C]POPC-labeled small unilamellar POPC vesicles were mixed with unlabeled CE-rHDL or unlabeled TG-rHDL and incubated at 37°C for 1, 3, 5, 10, and 20 min in the presence of PLTP (24). The vesicles were precipitated with a MnCl 2 /heparin solution (24,28), and the [ 14 C]POPC content of the rHDL in the supernatant was determined by liquid scintillation counting (Beckman LS 6000 TA Liquid Scintillation Systems, Beckman Instruments, Inc., Fullerton, CA). Precipitation of the vesicles with MnCl 2 /heparin was quantitative, while more than 95% of the rHDL remained in solution.
Transfer rates were calculated as the slope of the initial, linear section of plots of the percentage of phospholipid transferred between the rHDL and vesicles as a function of time.
Interaction of PLTP with rHDL-The binding of PLTP to CE-rHDL and TG-rHDL was studied using surface plasmon resonance analysis on a BIAcore 2000 system. Rabbit anti-mouse Fc (RAMFc) was covalently attached to CM5 research grade sensor chips as described (29). Purified monoclonal antibodies were injected individually at 10 l/min and captured at 25°C by the RAMFc. The human apoA-I-specific antibodies AI-1.2 (120 g/ml), AI-11 (5 g/ml), and AI-137.1 (13.5 g/ml) were used individually to capture CE-rHDL and TG-rHDL. The human PLTPspecific antibody, mAb 66 (100 g/ml) was used to capture the PLTP. The analyte PLTP or rHDL was exposed to the immobilized rHDL or PLTP, respectively, at 25°C with a buffer flow rate of 10 l/min.
For the apoA-I-specific mAb studies, a typical experiment consisted of the generation of a single chip of immobilized RAMFc followed sequentially by exposure of each flow cell to the same concentration of antibody and rHDL (25 nM). Then, 40 l of PLTP analyte (125-1000 nM) was injected, with each flow cell receiving a different analyte concentration.
For the PLTP-specific mAb studies, the RAMFc chip was exposed sequentially to 100 g/ml of mAb 66 and 1 M PLTP followed by CE-rHDL or TG-rHDL at 100 -6000 nM.
Data were analyzed with the BIAevaluation version 3.0 software, and the Langmuir model for 1:1 binding was used to calculate k a and k d rate constants.
Epitope Mapping of Anti-PLTP mAb 66 -Epitope mapping was performed using a Peptide-Spot filter containing 140 peptides, each 13 amino acid residues long, with 11 amino acid overlaps (Jerini Bio Tools, GmbH, Berlin, Germany). The filter was first soaked in methanol for 10 min at room temperature and further treated with monoclonal antibody mAb 66 as described (30). The bound antibodies were detected by ECL (30). According to this analysis, mAb 66 reacts with the region of amino acids 225-235 of PLTP, ASTSNLDMDFR. The molecular model of PLTP (31) predicts that this peptide region, especially the sequence STSN, is in a turn on the surface of PLTP and well exposed for antigenic reactions. 2 Structural Studies-Phospholipid acyl chain and head group packing order was determined by labeling rHDL with 1,6-diphenyl-1,3,5hexatriene (DPH) and 1-(4-trimethylammoniumphenyl)-6-phenyl-1,3,5hexatriene p-toluenesulfonate (TMA-DPH) (32,33), respectively. Steady state fluorescence polarization of the DPH-and TMA-DPHlabeled rHDL was measured at 5°C intervals from 5 to 50°C using an excitation wavelength of 366 nm. The molar ratio of rHDL phospholipid/ probe was 500:1, and the final phospholipid concentration was 0.5 mM.
Lipid-water interfacial hydration was assessed by labeling the rHDL with 6-propionyl-2-(dimethylamino)-naphthalene (PRODAN) (34). Uncorrected 390 -600-nm emission spectra of the PRODAN-labeled rHDL were recorded using an excitation wavelength of 366 nm and excitation and emission band passes of 5 and 6 nm, respectively. The molar ratio of rHDL phospholipid/probe was 500:1, and the final phospholipid con-centration was 0.5 mM. The ratio of the emission intensities at 440 and 490 nm was measured at 5°C intervals from 5 to 50°C.
The unfolding of apoA-I in CE-rHDL and TG-rHDL was assessed by incubation with 0 -8 M guanidine hydrochloride (GdnHCl) (21,35). The rHDL were added to aliquots of 50 mM Tris-HCl, pH 8.0, containing varying concentrations of GdnHCl. The final apoA-I concentration was 20 g/ml. Wavelengths of maximum fluorescence were determined from 300 -380-nm emission scans using an excitation wavelength of 295 nm. The respective excitation and emission band passes were 10 and 5 nm. Initial readings (t ϭ 0 h) were made at 25°C, immediately after the rHDL were added to the GdnHCl. Subsequent measurements were made after 2, 5, 8, and 24 h of incubation at 25°C.
Other Methods-Nondenaturing 3-40% polyacrylamide gradient gel electrophoresis was used to quantitate rHDL size (36). The gels were stained with Coomassie Blue and scanned with a Sharp JX-610 scanner (Sharp, Japan). Imagemaster software was used to quantitate particle size by reference to high molecular weight standards of known diameter (Amersham Pharmacia Biotech). Agarose gel electrophoresis was used to determine rHDL surface charge (37).
A Cobas Fara centrifugal analyzer (Roche Diagnostics, Zurich, Switzerland) was used for the compositional analyses. ApoA-I concentrations were determined either by the method of Lowry et al. (38), using bovine serum albumin as a standard, or by an immunoturbidometric assay (39). Enzymatic kits (Roche Molecular Biochemicals) were used to measure phospholipid, UC, and total cholesterol concentrations. CE concentrations were calculated as the difference between the total cholesterol and UC concentrations. TG was quantitated enzymatically (40).
The number of apoA-I molecules/particle was determined by covalently cross-linking the rHDL with bis(sulfosuccinimidyl) suberate (41). The cross-linked rHDL and a control sample of cross-linked, lipidfree apoA-I were electrophoresed on nondenaturing polyacrylamide gradient gels. The number of apoA-I molecules/particle was determined by reference to the cross linked lipid-free apoA-I.
Statistical Analysis-Analysis of variance, two-factor with repeated measures, was used to assess differences between data sets. Significance was determined as p Ͻ 0.05. The data analysis package in Microsoft Excel 98 was used for these analyses. (Tables I  and II, Figs. 1-3)-Spherical CE-rHDL, prepared as described under "Experimental Procedures," were enriched with TG by incubation for 20 min with Intralipid and CETP. The respective diameters of the CE-rHDL and TG-rHDL were 9.3 and 9.5 nm, and their PL/UC/CE/TG/apoA-I molar ratios were 95.3/6.8/ 68.1/1.8/3.0 and 103.2/6.2/54.2/22.9/3.0 ( Table I). The small amount of TG in the CE-rHDL probably represents traces of Intralipid co-isolating with the rHDL rather than spontaneous transfer of TG from Intralipid to the rHDL (21). Since crosslinking with bis(sulfosuccinimidyl) suberate showed that the CE-rHDL and TG-rHDL both contained three molecules of apoA-I/particle, the stoichiometries are expressed relative to three molecules of apoA-I. As judged by agarose gel electrophoresis, TG enrichment did not affect rHDL surface charge, with the CE-rHDL and TG-rHDL both having electrophoretic mobilities of Ϫ0.49 m⅐s Ϫ1 /V⅐cm Ϫ1 .

Remodeling of CE-rHDL and TG-rHDL by PLTP
CE-rHDL and TG-rHDL size was not affected by incubation for 24 h in the absence of PLTP (Fig. 1). After 24 h of incubation in the presence of PLTP, ϳ76% of the original CE-rHDL were converted into large (11.3-nm) and small (7.7-nm) particles. The remaining CE-rHDL were unchanged in size. When the TG-rHDL were incubated for 24 h with PLTP, they were completely converted into large (11.3-nm) and small (7.7-nm) particles, thus confirming the earlier report that TG-enrichment enhances the remodeling of HDL by PLTP (9).
The large and small conversion products were resolved by gel permeation chromatography (Fig. 2). The lipid and protein constituents in the rHDL that were either maintained at 4°C or incubated at 37°C in the absence of PLTP eluted as a single peak between fractions 12 and 27. This reflects the monodispersity of the rHDL preparations. The large particles generated by incubation in the presence of PLTP were enriched in CE, while the small conversion products were enriched in apoA-I. Selected fractions were pooled as indicated (7) and subjected to nondenaturing gradient gel electrophoresis (Fig.  3). Profiles A and B show, respectively, the CE-rHDL and TG-rHDL, that were either maintained at 4°C or incubated at 37°C for 24 h in the absence of PLTP. Profiles C and D represent the large (11.3-nm) and small (7.7-nm) conversion products, respectively. The number of apoA-I molecules/particle in the pooled samples was determined by cross-linking (Table II). Whereas the original CE-rHDL and TG-rHDL contained three molecules of apoA-I/particle, the large and small conversion products contained four and two molecules of apoA-I/particle, respectively. The composition of the pooled samples and the recoveries of the individual rHDL constituents are shown in Table II. The large particles contained approximately twice as many phospholipid and core lipid molecules as the original rHDL, indicating that they are derived from a fusion product. Following incubation with PLTP, the recovery of rHDL core lipids exceeded that of the surface constituents. This indicates  1. PLTP-mediated changes in rHDL size. CE-rHDL (final apoA-I concentration 92.5 g/ml) and TG-rHDL (final apoA-I concentration 92.5 g/ml; 28.0% TG) were either maintained at 4°C or incubated at 37°C for 24 h in the absence or presence of PLTP (final activity 2.2 mol of PL transferred/ml of PLTP/h). The final volume of the incubation mixtures was 4.79 ml. When the incubations were complete, the rHDL were isolated by ultracentrifugation and subjected to nondenaturing polyacrylamide gradient gel electrophoresis. The profiles represent scans of Coomassie-stained gels. Particle diameters were calculated by reference to known high molecular weight standards. that phospholipids and UC as well as apoA-I dissociated from the CE-rHDL and TG-rHDL during the incubation.
Time Course of the Remodeling of CE-rHDL and TG-rHDL by PLTP (Fig. 4)-The time dependence of the remodeling of CE-rHDL and TG-rHDL by PLTP was also examined (Fig. 4). CE-rHDL and TG-rHDL were incubated for 0 -24 h with PLTP. Aliquots of the incubation mixtures were subjected to nondenaturing polyacrylamide gradient gel electrophoresis and immunoblotted with an anti-apoA-I polyclonal antibody. Tracks A and B show rHDL that were either maintained at 4°C or incubated at 37°C for 24 h in the absence of PLTP. Tracks C, D, E, F, and G represent rHDL that were incubated in the presence of PLTP for 1, 3, 6, 12, and 24 h. Lipid-free apoA-I is shown in track H. A proportion of the CE-rHDL were converted into small particles during the first 6 h of incubation with PLTP. Large CE-rHDL conversion products and dissociated apoA-I were apparent after 12 h of incubation. In the case of TG-rHDL, small particles were apparent at 1 h, while the large conversion products and dissociated apoA-I appeared after 3 h of incubation with PLTP.
The finding that the small conversion products were formed before the large conversion products provides convincing evidence that these particles are generated by two independent processes. Furthermore, as both the large and small TG-rHDL conversion products appeared more rapidly than the CE-rHDL conversion products, it follows that TG enrichment enhances both processes. The additional finding that the dissociation of apoA-I coincided with the appearance of the large conversion products suggests that their formation is related.
Influence of TG Enrichment on Phospholipid Transfers (Fig.  5)-To determine whether transfers of phospholipids from small unilamellar vesicles to TG-HDL were also enhanced relative to CE-rHDL, the rHDL were incubated with To determine whether the different rates of transfer could be explained by modification of the vesicles by rHDL surface constituents, transfers of phospholipids from rHDL to the vesicles were also examined. [ 14 C]POPC-labeled CE-rHDL and [ 14 C]POPC-labeled TG-rHDL were incubated for 0 -20 min with PLTP and unlabeled POPC vesicles. There was no detectable transfer of POPC from the CE-rHDL to the vesicles (open diamonds). Less than 5% of the TG-rHDL phospholipids transferred to the vesicles (closed diamonds). This indicates that the different rates of transfer of phospholipids from the vesicles to the rHDL cannot be explained by rHDL phospholipids altering the vesicle structure.
Binding of CE-rHDL and TG-rHDL to PLTP (Table III)-There was no difference between the interaction of soluble PLTP with CE-rHDL and TG-rHDL that were immobilized on the biosensor surface using either of three separate apoA-Ispecific monoclonal antibodies (AI-1.2, AI-11, or AI-137.1). Likewise, no difference was observed between the interaction of CE-rHDL and TG-rHDL with PLTP when the PLTP was immobilized on the biosensor surface with the PLTP-specific antibody mAb 66 (Table III).
The Influence of TG Enrichment on rHDL Structure (Table  IV, Fig. 6)-Since the enhanced remodeling of TG-rHDL compared with CE-rHDL could not be explained in terms of differences in the binding to PLTP, the possibility that it was due to structural differences between the particles was investigated.
Spherical CE-rHDL were enriched with increasing amounts of TG by incubation with Intralipid and CETP for 2 or 20 min.

FIG. 2. Elution profiles of CE-rHDL and TG-rHDL. CE-rHDL and TG-rHDL
were either maintained at 4°C or incubated at 37°C for 24 h in the absence or presence of PLTP as described in the legend to Fig. 1. When the incubations were complete, the rHDL were isolated by ultracentrifugation, concentrated by ultrafiltration, and applied to a HR 10/30 Superose-6 column. The concentrations of the individual rHDL constituents in each fraction are shown. The values represent the mean of triplicate determinations that varied by less than 10%. Selected fractions were pooled as indicated (7). Control incubations containing CE-rHDL and Intralipid but no CETP were also carried out. The rHDL that were incubated with CETP contained TG as 12.2 and 32.4% of the total core lipids (Table IV). TG enrichment increased the size of the rHDL slightly from 9.2 to 9.5 nm.
Phospholipid acyl chain and head group packing order was determined from the steady state fluorescence polarization of rHDL labeled with DPH and TMA-DPH (Fig. 6). The DPHlabeled CE-rHDL that were either maintained at 4°C (data not shown) or incubated in the absence of CETP (closed squares) had comparable polarization values. As the TG content of the rHDL increased, phospholipid acyl chain packing order decreased as evidenced by the progressive reduction in the polarization values for the rHDL with The polarization values for the TMA-DPH-labeled CE-rHDL (closed squares) and the TG-rHDL with 12.2% TG (open trian-gles) and 32.4% TG (open circles) were not significantly different, indicating that TG enrichment did not affect rHDL phospholipid head group packing order.
The rHDL lipid-water interfacial hydration was assessed by labeling with PRODAN and comparing the ratio of the intensities of the fluorescence emission spectra at 440 and 490 nm as a function of temperature. No differences were observed between the CE-rHDL and TG-rHDL (results not shown). This indicates that TG enrichment has no effect on rHDL lipidwater interfacial hydration.
The Influence of TG Enrichment on the Unfolding of apoA-I in rHDL (Table IV, Figs. 7 and 8)-The unfolding of the apoA-I in spherical CE-rHDL and rHDL containing either 12.2 or 32.4% TG is shown in Fig. 7. The rHDL were incubated with 0 -8 M GdnHCl for 0 (closed diamonds), 5 (open squares), and TABLE II Physical properties of the pooled rHDL fractions CE-rHDL and TG-rHDL were incubated with PLTP as described in the legend to Fig. 1. Selected fractions were pooled as indicated (7) in Fig.  2. The composition, size, and number of apoA-I molecules/particle were determined as described under "Experimental Procedures." The stoichiometries represent the mean of triplicate determinations, which varied by less than 10%.  24 h (closed triangles). The results for the CE-rHDL are shown in Fig. 7A. Panels B and C represent rHDL with 12.2 and 32.4% TG, respectively. The concentration of GdnHCl required to achieve 50% unfolding of apoA-I is shown in Table IV. These values were either calculated directly from Fig. 7, or from a plot of the concentration of GdnHCl versus the free energy of unfolding of apoA-I (35). The results show that the concentration of GdnHCl required to unfold apoA-I decreases with increasing rHDL TG content.
The kinetics of the unfolding of apoA-I in CE-rHDL (closed diamonds) and rHDL with 12.2% (open squares) and 32.4% (closed triangles) TG are shown in Fig. 8. These experiments were carried out in the presence of 4.0 M GdnHCl The apoA-I in the rHDL with 32.4% TG unfolded more rapidly than the apoA-I in the rHDL with 12.2% TG or the apoA-I in CE-rHDL. When these results are considered with the data in Table IV, it follows that enrichment with TG destabilizes the apoA-I in rHDL.

DISCUSSION
The ability of PLTP to remodel HDL into large and small particles, mediate the dissociation of apoA-I from HDL, and transfer phospholipids between HDL and other lipoproteins is well documented (1)(2)(3)(4)(5)(6)(7)(8). Earlier work from this laboratory has also established that TG enrichment enhances the remodeling of HDL by PLTP (9). While the PLTP-mediated remodeling of HDL and the dissociation of apoA-I are both dependent on efficient phospholipid transfers (42), the mechanism by which these processes occur is not understood. The present study was carried out in order to elucidate the mechanism of the remodeling of HDL by PLTP and reasons why it is enhanced in TG-enriched particles.
These goals were achieved by (i) characterizing the large and small particles that were formed when CE-rHDL and TG-rHDL were incubated with PLTP, (ii) determining the time sequence for the formation of the large and small particles and the dissociation of apoA-I, and (iii) defining the structural changes that occur when rHDL are enriched with TG.
The relationship between the time sequence of the remodeling of rHDL into large and small particles and the dissociation of apoA-I was studied by incubating CE-rHDL and TG-rHDL

TABLE III Interaction of PLTP with CE-rHDL and TG-rHDL
The association rate constants (k a ) and dissociation rate constants (k d ) for PLTP and CE-rHDL or TG-rHDL were measured by surface plasmon resonance analysis as described under "Experimental Procedures." Multiple interactions were studied including the interaction of soluble PLTP with rHDL that had been immobilized on the biosensor chip with three unique apoA-I-specific monoclonal antibodies. The interaction of soluble rHDL with PLTP immobilized with the PLTP-specific monoclonal antibody 66 was also studied. Monoclonal antibodies AI-11, AI-137.1, and AI-1.2 bind apoA-I epitopes containing amino acid residues 96 -111, 137-147, and 1-19, respectively.  with PLTP and monitoring rHDL size changes as well as the dissociation of apoA-I (Fig. 4). These results showed that the small conversion products were formed before the large conver-sion products and that the appearance of the large conversion products coincided with the dissociation of apoA-I. Furthermore, both the large and small TG-rHDL conversion products were generated more rapidly than the CE-rHDL conversion products.
In the case of CE-rHDL, small particles were formed during the first 6 h of incubation with PLTP, while the large conversion products and dissociated apoA-I appeared after 12 h of incubation. These time differences indicate that the large and small conversion products are formed by independent processes. Furthermore, the finding that the formation of the large particles coincided with the dissociation of apoA-I suggested that both of these species were generated via a common pathway.
The data in Table II show that the large CE-rHDL and TG-rHDL conversion products contain more surface and core lipid constituents/particle than the original rHDL, indicating that they must have been formed by particle fusion. Moreover, the data in Table III, showing that TG enrichment does not affect the binding of PLTP to rHDL, suggests that the rate at which the rHDL are remodeled is determined by processes that occur after PLTP has mediated particle fusion. This finding, together with the data in Table II and Fig. 4, indicate that the PLTP-mediated remodeling of rHDL involves the following events (Fig. 9). Interaction of PLTP with two rHDL particles gives a large, unstable fusion product with six molecules of apoA-I (Fig. 9i). The fusion product has one of the following two fates: it rearranges into three small particles, each of which has two molecules of apoA-I/particle, in a process that is not accompanied by the dissociation of apoA-I (Fig. 9ii), or two molecules of apoA-I dissociate from the fusion product, forming a more stable, large conversion product with four molecules of apoA-I (Fig. 9iii).
The possibility that the small conversion products are further remodeled into large conversion products was also considered. The data in Table II indicate that this pathway requires interactions between three small conversion products and the concomitant dissociation of two molecules of apoA-I. Although the results in Fig. 4 indicate that formation of large conversion products by this pathway is feasible, it is likely that trimolecular collisions of this type are not energetically favorable. As  9. Proposed mechanism for the remodeling of rHDL by PLTP. i, PLTP mediates the fusion of two rHDL particles, each of which contain three molecules of apoA-I, to give an unstable particle with six apoA-I molecules. The fusion product either (ii) rearranges into three small particles with two molecules of apoA-I/particle or (iii) is converted into a large particle with four molecules of apoA-I in a process that is accompanied by the dissociation of two molecules of apoA-I. such, this pathway is unlikely to be a major source of the large conversion products.
The results of the spectroscopic studies give an insight as to why the TG-rHDL are remodeled by PLTP more rapidly than CE-rHDL. The data in Fig. 8 show that the apoA-I in TG-rHDL unfolds more readily than the apoA-I in CE-rHDL. This is in agreement with what has been reported by other investigators (43) and is most likely caused by TG partitioning from the core into the particle surface and preventing apoA-I ␣-helices from intercalating between the rHDL phospholipid acyl chains (44). This may be why apoA-I dissociates more rapidly from TG-rHDL than from CE-rHDL and why the PLTP-mediated remodeling of HDL is enhanced by TG enrichment.
The apoA-I that dissociated from the CE-rHDL and TG-rHDL during the incubations with PLTP appeared as three bands when subjected to nondenaturing gradient gel electrophoresis and immunoblotting (Fig. 4). One of the bands was comparable in size with lipid-free apoA-I. The smallest band is possibly the 23-kDa fragment of apoA-I that is generated when either lipid-free or lipid-associated apoA-I are incubated with PLTP (45). The largest of the three bands may represent apoA-I complexed with small amounts of rHDL lipids. This is consistent with the reduced recovery of surface constituents relative to core lipids in the large and small conversion products (Table  II). It is possible that the apoA-I that is associated with small amounts of lipid may be comparable with the pre-␤ 1 -migrating HDL in human plasma that have been identified as the initial acceptors of cellular cholesterol from peripheral tissues (12,46). Fig. 5 shows that PLTP-mediated phospholipid transfers from small unilamellar vesicles to TG-rHDL are enhanced in TG-rHDL relative to CE-rHDL. The additional finding that there is minimal transfer of phospholipids in the reverse direction, from the rHDL to the vesicles, excludes the possibility that this result can be explained by rHDL phospholipids altering the vesicle surface. The increased transfer of phospholipids from the vesicles to the TG-rHDL can be explained by an enhanced ability of the TG-rHDL to accommodate additional phospholipids. This is consistent with the results in Fig. 6, which show that TG enrichment decreases the packing order of rHDL phospholipid acyl chains. This is probably a result of TG partitioning from the core into the rHDL surface and generating packing defects that can be occupied by additional phospholipid molecules. These findings are also in agreement with the observation that the TG content of HDL correlates positively with the rate of PLTP-mediated phospholipid transfers in plasma (47).
In conclusion, this study provides the first insight into the mechanism by which PLTP mediates the remodeling of HDL. The results show that PLTP acts as a fusogen when it interacts with rHDL and that the fusion product is subsequently remodeled into large and small particles. Evidence that the large and small conversion products are formed by two independent processes is also presented. In addition, we show that the apoA-I in TG-rHDL is destabilized compared with the apoA-I in CE-rHDL and that this destabilization may be responsible for the enhanced PLTP-mediated remodeling of TG-rHDL.