High Density Lipoprotein Interconversions in Rat and Man as Assessed with a Novel Nontransferable Apolipopeptide*

A nontransferable peptide analog of a plasma apoli- poprotein diacyl lipid-associating peptide (diLAP) in-corporates into model reassembled high density lipo- proteins (R-HDL). In whole plasma in vitro, diLAP irreversibly transfers to native rat HDLz and human HDLs, but not to rat HDLl or human HDLz. The rate of transfer is dependent on the physical state of the lipid in the R-HDL. Exogenous cholesterol promotes the formation of larger HDL. When diLAP-labeled R-HDL were injected into rats, the diLAP that initially associated with HDLz transferred to HDLl over a period of 48 h. The rate of clearance of diLAP-labeled HDL was slower than that of apoA-I. The liver was the preferred site for diLAP-labeled HDLl uptake. In con- trast, diLAP-labeled HDL, were associated with liver, ovaries, and adrenal glands, with the adrenal grands exhibiting the highest specific association. DiLAP was not found in the kidneys. These data show that 1) rat HDL is cleared more slowly than rat apoA-I; 2) HDL is removed from the plasma compartment as a particle; 3) there are tissue-specific differences in the removal of rat HDLl and HDL2; 4) HDLz is a precursor to HDLI;

Plasma lipoproteins, which are composed of a variety of lipids and proteins, are remodeled by spontaneous and protein-mediated processes to produce mature particles that transport lipids among various extravascular compartments. Lipoprotein interconversions are an important but poorly defined aspect of the remodeling process. One of the best understood examples of lipoprotein interconversion is the lipolytic cascade that begins with very low density lipoproteins and terminates with the formation of low density lipoproteins. This process has been relatively easy to study because the apoB-100 component of very low density lipoproteins, which is nontransferable, can serve as a reliable marker of the fate of the host particle through the entire cascade. High density lipoproteins (HDL)' also undergo interconversions in the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ll To whom correspondence should be addressed Mail Station A601, the Methodist Hospital, 6565 Fannin St., Houston, T X 77030.
The abbreviations used are: HDL, high density lipoprotein(s); R-HDL, model reassembled high density lipoprotein(s); LCAT, lecithin:cholesterol acyltransferase; DMPC, 1,2-dimyristoyl-sn-glycero-3-phosphorylcholine; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphorylcholine; diLAP, diacyl lipid-associating peptide. plasma compartment, but have been difficult to characterize because there are no reliable markers for the processing of HDL, even though it is clear that substantial remodeling of this relatively long-lived particle occurs. Indirect evidence of the interconversion of HDL subfractions has been derived from studying the changes associated with lipolysis. These studies showed that postprandial lipemia leads to triglyceriderich HDL2 that are converted to HDLB (Patsch et al., 1984(Patsch et al., , 1987. Moreover, the in vitro studies showed that postprandial HDL, from hyperlipemic subjects can be converted to HDL3 (Patsch et al., 1987), a process requiring apoA-I transfer among different particles.
HDL exhibit different rates of degradation according to the exchangeability of the marker employed to monitor plasma decay kinetics. Cholesterol and phospholipids exchange too rapidly to be used as a marker for HDL (Stein and Stein, 1966;Phillips et al., 1987); cholesteryl esters, triglycerides, and phosphatidylcholines are transferred between lipoprotein subclasses by specific protein factors (Jarnagin et al., 1987;Tall, 1986;Barter et al., 1987). Cells in culture also selectively remove cholesteryl esters from HDL (Pittman et al., 1987). Nevertheless, when apoA-I is used as a marker for HDL, the lifetimes obtained are on the order of 4-5 days in normal human subjects (Shepherd et al., 1978;Schaefer et al., 1986). These values should be considered a lower limit for the lifetime of HDL because studies in rat clearly show that the liver is the major site for the removal of lipids and apoproteins, whereas the kidney is an important site for apoprotein catabolism (Glass et al., 1983;Ponsin et al., 1986). This and other evidence based upon the very rapid transfer of apoC proteins and acylated apolipopeptide analogs (McKeone et al. 1988; Hickson-Bick et al., 1988) suggest that the soluble apoproteins are poor markers for the identity and metabolic fate of a lipoprotein particle.
In vivo studies of synthetic apolipopeptides have shown that the fraction of plasma peptide that is degraded in the kidneys is a predictable function of its affinity for HDL (Ponsin et al., 1984(Ponsin et al., , 1986. The results suggested the possibility of synthesizing an apolipopeptide that was sufficiently hydrophobic to prevent its transfer between lipoproteins within the timeframe of lipoprotein remodeling. This paper describes the synthesis, characterization, and in vitro and in vivo testing of a nontransferable apolipopeptide (diLAP) (Fig. l), which is an analog of a previously described peptide that binds phospholipids and activates LCAT (Pownall et al., 1980;Ponsin et al., 1984Ponsin et al., , 1986. EXPERIMENTAL PROCEDURES Materials-ApoA-I and apoE were isolated from normal human plasma as previously described (Pownall et al., 1978;Rall et al., 1986). 1,2-Dimyristoyl-sn-glycero-3-phosphorylcholine (DMPC) and l-palmitoyl-2-oleoyl-sn-glycero-3-phosphorylcholine (POPC) were purchased from Avanti Polar Lipids, Inc. (Birmingham, AL). Fatty acidfree bovine serum albumin and cholesterol were obtained from Sigma.
["'I]Iodide was from Amersham Corp. Female  were used for the in vivo experiments.
DiLAP was prepared by a modification of the procedure of Ponsin e t al. (1984). The 15-residue peptide was synthesized by conventional solid-phase methods on an Applied Biosystems Model 430 peptide synthesizer. While attached to the resin, 2-hexadecyloctadecanoic acid was coupled to the amino-terminal end of the peptide chain in the presence of dicyclohexylcarhodiimide and N,N-dimethylaminopyridine; 2-hexadecyloctadecanoic acid was synthesized via malonic ester synthesis followed by decarboxylation. The peptide was cleaved and deprotected with HF and purified by high performance liquid chromatography.
Methods-ApoA-I and albumin were labeled by the chloramine-T method (Greenwood et al., 1963). Because of its low solubility in water, the peptide was first dissolved in 2-propanol, and then the solution was added to the labeling mixture; peptide or protein was used within 2 days of labeling. The protein or peptide was separated from the label by gel filtration. R-HDL were prepared by the cholate removal technique of Matz and Jonas (1982). Typically, the phospholipids and apoA-I or apoE were combined in respective molar ratios of 501. This mixture was added to the labeled peptide, to which sufficient sodium cholate was added to discharge the turbidity of the lipid dispersion. Exhaustive dialysis against a Tris buffer (10 mM, pH 7.4) containing 150 mM sodium chloride, 10 mM EDTA, and 10 mM sodium azide gave an optically clear solution of R-HDL into which the labeled peptide was incorporated as a tracer. The Tris buffer was used in all in vitro experiments, but was replaced by a phosphate buffer (30 mM, pH 7.4) for the in vivo studies.
Serum Clearance Kinetics-The kinetics of turnover of labeled R-HDL were performed as previously described (Ponsin et al., 1986).
T o obviate significant formation of '251-labeled thyroid hormones during serum decay measurements, the rats were injected intraperitoneally with 1 ml of potassium iodide (1 mg/ml) 1 day before and at the time of injection. Catheters were introduced into the left and right femoral veins of rats under ether anesthesia. One catheter was used for injection of R-HDL, and the other was used for sampling. Between samplings, a solution of 0.09% NaCl was infused through the catheter at the rate of 1 ml/h. The injected samples consisted of "'1-diLAP as R-HDL. The sera were sampled (0.1-0.2 ml) periodically and counted for ' *' I radioactivity. The serum volume was estimated at 4% of body weight. I n Viuo Association ~f~~~I -L a b e l e d Peptide or Protein with Organs-Rats under ether anesthesia were injected via the femoral vein with the radiolabeled peptide or protein as a component of R-HDL. After 2 or 48 h, the animals were killed and their plasma (1 ml) was reinjected into a second set of rats, which were exsanguinated via the abdominal aorta within 7 min of injection. The major organs were immediately removed; and after careful cleaning, each was weighed and counted. For the liver, several samples were removed and counted, and the results were averaged. The results were expressed in terms of a volume of serum having the same radioactivity; the distribution volumes thus defined were expressed in microliters/gram of organ. T h e nonspecific association of serum radioactivity with organs was estimated by measuring the organ distribution volumes of '251-labeled albumin in some of the rats (Ponsin et al., 1986).
Gradient Gel Electrophoresis-HDL subfractions were characterized by gradient gel electrophoresis as described by Blanche et al. (1981), whose electrophoretic criteria were used to distinguish HDL1, HDL,., HDL2b, and HDL3. Following in vitro or in vivo incubations of R-HDL with plasma, small samples were applied to commercial polyacrylamide gradient gels (PAA 2/16 or 4/30, Pharmacia LKB Biotechnology Inc.). Samples were electrophoresed a t 2 "C for at least 24 h in a buffer of 90 mM Tris, 80 mM boric acid, 3 mM EDTA, and 3 mM azide. Samples were applied in -15 p1 of buffer containing 40% sucrose. Following electrophoresis, the gels were cut into 32 strips and counted, and the radioactivity was plotted versus gel slice number.
All experiments with rats were conducted with at least three sets of three animals. I n Vitro Transfer Kinetics-The kinetics of desorption of "'I-diLAP from HDL, were determined according to Retzinger et al. (1985) and McKeone et al. (1988). Rat HDL, were labeled by adding '2'I-diLAP-labeled R-HDL to rat plasma followed by flotation at d 1.125 g/ml. At 37 "C, the labeled HDL, were mixed with latex beads coated with egg phosphatidlycholine. The mixture of beads and HDL2 was gently stirred; and over a period of 48 h, samples were removed, and the beads were pelleted, washed a t 4 "C, and counted. Similar studies of peptide transfer from latex heads to rat HDL were also performed. A trace of '"I-diLAP was added to the POPC before it was used to coat the latex beads. The beads were mixed with a 10fold excess of rat HDL and stirred at 37 "C for 48 h, during which aliquots were collected, and the radioactivity that remained in the pellet after washing at 4 "C was counted. The half-times for transfer were estimated from an exponential plot of the fraction of counts remaining with the latex beads as a function of time.
I n Vitro Transformation of R-HDL-One ml of rat or human plasma was mixed with 20 pl of '2511-diLAP-labeled R-HDL (-10 pg of protein) a t various temperatures; in some experiments, 10 p1 of an ethanol solution of free cholesterol (150 pg) was injected into the plasma -10 s before the addition of the R-HDL. At various times, a sample was removed and cooled in an ice bath until all succeeding samples were collected. Each was applied to a lane of polyacrylamide gel for electrophoresis a t 2-4 "C, and the resulting electrophoretic pattern was analyzed. For rat plasma, the rate of transfer was calculated from the rate at which radioactivity appeared in the part of the gel corresponding to HDL2. Human plasma samples were subjected to ultracentrifugal flotation in sodium bromide a t d 1.125 g/ml at 2 "C for 5 h in a Beckman Model T L 100 ultracentrifuge. The floating and sedimenting fractions, HDL2 and HDL,, respectively, were then analyzed by polyacrylamide gel electrophoresis. Preliminary experiments established that there was no transformation at 4 "C.

RESULTS
Clearance Kinetics-The measurements of the kinetics of transfer of "'1-diLAP from HDL, to latex beads at 37 "C (data not shown) were fitted to an exponential curve from which a transfer half-time of 4 days was calculated. Analysis of these data showed that over a period of 48 h, 28% of the 'T-diLAP transferred from the HDLz to the beads. When donor and acceptor identities were reversed, 8% of the peptide transferred within 48 h, and a half-time of >2 weeks was obtained. These data indicate that lZ5I-diLAP transfers between hydrophobic surfaces. However, within the timeframe of our in vivo studies that last up to 48 h, this peptide can be used as a nontransferable marker of HDL.
To determine how fast the label disappeared from the plasma compartment and what transformations of the HDL were associated with the labeled HDL, the behavior of the "'I-diLAP injected into rats was followed in two ways. The '2'II-diLAP-labeled R-HDL was injected into rats, and samples were collected and split into two parts. One part was counted to calculate the kinetics of peptide turnover, and the other was applied to gradient gels to follow the distribution of the labeled peptide among HDL subfractions. Although the turnover kinetics of the peptide were nonexponential (Fig. M ) , we determined a half-time of -12 h, a value that is larger than that observed previously when apoA-I and other exchangeable peptides were used (Glass et al., 1983;Ponsin et al., 1986). Fig. 3 is a plot of the distribution of I2'I-diLAP among HDL subfractions as a function of time. Before injection, a single peak of radioactivity was found near the migration distance of HDL, (Fig. 3, upper left panel). Within 5 min of the time of injection, all of the radioactivity was found in the range of HDL,. Between 2 and 5 h, a shoulder in the radioactivity appeared and after 20 h, a distinct peak corresponding to the migration distance of HDL, emerged. After 48 h, nearly all of the remaining radioactivity was associated with the fraction that migrates as HDL,. Knowing the distri- bution of lZ5I-diLAP between HDL subfractions (Fig. 3) and the serum clearance time (Fig. 2), it is possible to calculate the rate of clearance of HDL, and HDL, and plot the kinetics of conversion of the HDL in several ways. Fig. 2B shows that the total peptide radioactivity shifted from HDL, to HDLl over the 48-h interval. At the same time, the fraction of counts appearing as HDL, declined and that associated with HDLl initially grew and then also declined (Fig. 2C). Finally, Fig. 2 0 shows that the ratio of T -d i L A P associated with HDL, and HDL, increased from nil to >4 over the same time interval.
Organ Distribution of R-HDL Containing ApoA-Z and lZ5Z-DiLAP-Studies of 12sI-apoA-I-labeled HDLz directly injected into the animals showed that nearly all of the radioactivity was associated with liver, kidney, adrenal glands, and ovaries (Fig. 4). These results were similar to earlier reports (Glass et al., 1983;Ponsin et al., 1986). To differentiate the organ distribution of HDLz and HDLl, the difference in the kinetics of clearance of the two lZ5I-diLAP-labeled subfractions cited above was used. Rats injected with '251-diLAP-labeled R-HDL were exsanguinated after 2 or 48 h. The plasma, which contained essentially 'T-diLAP-labeled HDLz or HDLl, respectively, was reinjected into another set of rats, and the association with various tissues was determined. In both cases, virtually no '"I-diLAP was found in renal tissue. The highest specific association of '251-diLAP-labeled HDL2 was found with adrenal glands followed by ovaries and liver, which exhibited about the same specific association. Association of lZ5I-diLAP-labeled HDLl with adrenal glands and ovaries was much lower than that of 12sI-diLAP-labeled HDL, and was comparable to that with liver.
In Vitro Transformations of Plasma HDL-Additional experiments were conducted in vitro to determine the mechanism by which the lZ5I-diLAP-labeled HDLz was converted to HDL,. '251-diLAP-labeled R-HDL was incubated with rat plasma in vitro at 37 "C (Fig. 5, A-D), and the electrophoretic migration of the lZ5I-diLAP was followed as a function of incubation time in both the presence and absence of exogenous cholesterol, which was added to the mixture at t = 0 by injection as an ethanol solution. At the earliest time point ( 5 min), there was already a bimodal distribution of the lZ5I-diLAP, with most of it co-migrating with HDLz. Between 3 and 30 h, the distribution of lz5I-diLAP radioactivity with HDLz was virtually unchanged. This behavior differed from in uiuo behavior, where there was a time-dependent transfer of the lZ51-diLAP-labeled R-HDL to a species that migrated as HDLZ and then HDL1. These data are also consistent with the behavior of a nontransferable peptide since none of the label transferred to the HDLl fraction over this time interval. The in vitro transformation was greatly modified by exogenous free cholesterol (Fig. 5, E-H). After 3 h, lZ5I-diLAP was already shifting from HDLz to larger particles. After 22 and 30 h, the lZ5I-diLAP label co-migrated with lipoprotein particles having molecular weights much greater than that of HDLz and with dimensions that increased with time. However, unlike the in uivo experiments, at t = 30 h, the 'T-diLAP was associated with particles having dimensions that were considerably larger than those of any normal HDL subspecies.
When human rather than rat plasma was used, the lZ5I-diLAP was first recovered in the HDL, subfraction, but was gradually transferred to HDLz as incubation time was increased. Fig. 6 contains representative electrophoretic patterns after incubation for 1 and 24 h. These data show that the peptide is initially associated with HDL3; but after 24 h, most of the label is found in the larger HDL, particles. No difference in the kinetics of label transfer from HDLB to HDLp was observed when apoE was substituted for the apoA-I in the R-HDL.
Initial Association of lZ5I-DiLAP with Plasma HDL-Since we observed a rapid association of lZ5I-diLAP-labeled R-HDL with plasma HDL, we attempted to study this process at lower temperatures to better define its energetics and mechanism. These data are shown in Fig. 7, which contains plots of the rates of transfer of '251-diLAP from labeled R-HDL to HDL. With R-HDL composed of either apoA-I or apoE with POPC or DMPC, the rate increased with increasing temperature. The differences are more apparent in Arrhenius plots (Fig.  8). With R-HDL that contains POPC and apoA-I or apoE, these plots were linear, with activation energies and rate constants that are given in Table I  was replaced with DMPC, nonlinear plots having two line segments that intersect at the gel-to-liquid crystalline transition temperature (T,) of DMPC were observed; this behavior was found with R-HDL composed of either apoA-I or apoE.

DISCUSSION
In humans, HDL are thought to be critical in reverse cholesterol transport from peripheral tissues to the liver ( Phil-lips et al., 1987;Schwartz et al., 1982;Koo et al., 1985). Therefore, a better understanding of HDL remodeling within the plasma compartment is important in developing quantitative models of in vioo cholesterol metabolism. HDL are highly polymorphic (Patsch and Patsch, 1986); and the metabolic relationships between HDL subfractions, although poorly defined, remain an important component of cholesterol transport. The major difficulty in determining whether a given subfraction of HDL is a product or precursor of other fractions is that all of the components of HDL exchange freely via either spontaneous or protein-mediated pathways (Tall, 1986;Phillips et al., 1987;McKeone et al., 1988;Shepherd et al., 1978). Using a series of synthetic acylated apolipopeptides, we showed that their hydrophobicity was the major determinant of their affinity for HDL (Ponsin et al., 1984(Ponsin et al., , 1986Hickson-Bick et al., 1988;McKeone et al., 1988). According to two criteria, association with HDL and activation of plasma LCAT, these peptides retained many of the properties of apoA-I, but contained hydrophobic prosthetic groups that could be changed for a given experiment according to the required affinity for a lipoprotein surface. On the basis of these observations and a model  for peptide association with HDL, we synthesized a new diacyl peptide that is very hydrophobic and is predicted to be practically nontransferable among lipoprotein subclasses. The hydrophobicity of diLAP is sufficiently large to obviate any significant transfer during a 2-day study.
Transformation of HDL Particles in Human Plasma-When '"I-diLAP-labeled R-HDL were incubated with human plasma at 37 "C, the radioactivity corresponding to lZ5I-diLAP was transferred into the HDL3 subfraction faster than the discs and HDL could be separated. This was followed by a time-dependent transfer of the radioactivity into HDL,. Gradient gel electrophoresis indicated that this process proceeded in two distinct steps characterized by HDL having different physical properties. In the first step, the particles containing '"'I-diLAP exhibited a density corresponding to that of HDL,, but were not fully remodeled to the size of mature HDL2. During the second step, the dimensions of the particle containing lZ5I-diLAP continued to increase in size by acquiring additional lipid. On the basis of one widely used model ( Edelstein et al., 1979;Shen et al., 1977), this was likely to be mostly core or neutral lipid. After 24 h, most of the lz5I-diLAP appeared in a density and size range corresponding to that of mature HDL,. The two differently sized particles in the density range of HDL, have been characterized in great detail by Blanche et al. (1981), who designated the small and large fractions as HDL,,,, and HDLPbgge, respectively. Based on this designation, our data suggest that HDL, are converted to HDLPbr with HDL,, being an intermediate.
Transformation of HDL Particles in Rat Plasma-In contrast to human plasma, rat plasma contains little or no detectable HDL, (Oschry and Eisenberg, 1982). The majority of rat HDL appears in the HDL, density range, while a minor part resides in a larger particle (HDL,) that is relatively enriched in apoE. After rats were injected with 'T-diLAPlabeled R-HDL, the radioactivity was immediately recovered in particles having the dimensions of HDL,. AS time passed, a n increasing fraction of the label appeared in the fractions corresponding to HDL1; and at the end of 48 h, nearly all of the label in the plasma was found in HDL1, suggesting that HDLl is not a precursor to HDL,. The calculated plasma decay curve of HDL, corresponded to a half-life of -7 h.
Plasma HDL, may diminish over a period of time through two or more routes; these include 1) the conversion of HDL, t o HDL, within the plasma compartment and 2) the direct clearance of HDL, via transport to extravascular compartments. These two metabolic pathways are characterized by the rate constants kl and kp, respectively; the sum of these two rate constants is the apparent rate constant that may be calculated from kapp = kl + k, = 0.693/t1,, = 0.693/7 h = 0.1 h-'.
Previous studies in the rat using either labeled apoA-I or HDL containing lZ5I-labeled synthetic apolipopeptides have shown that the plasma clearance of HDL is associated with a half-time of -9 h (Glass et al., 1983;Ponsin et al., 1986), which corresponds to a clearance rate constant of 0.08 h-'. Both the apoA-I and the synthetic peptides used in previous studies readily exchanged between HDL particles. Since nearly all of the labeled apoA-I or synthetic peptide is associated with HDL,, the clearance rate constant of apoA-I is similar (if not identical) to kz. Thus, we can estimate that the rate constant ( k l ) for the conversion of HDL2 to HDL, is 0.02 h-'. From this, we infer that the transformation of HDL, into HDL, in normal rats represents a minor pathway in the catabolism of HDL and that the major pathway is the direct removal of HDL,.
The total half-time for the disappearance of lZ5I-diLAP is about one-third longer than that of '251-apoA-I. We attribute the lower lifetime to the absence of renal filtration of diLAP. A major pathway for the removal of apoA-I from the plasma compartment is through uptake and degradation in the kidneys. This pathway is not used for the uptake of cholesteryl esters, which are not soluble enough to form free monomers in the aqueous phase (Glass et al., 1983). Similarly, the diLAP does not transfer to the aqueous phase. Therefore, this pathway is not expected to contribute to the disappearance of 1251-diLAP in HDL, and lz5I-diLAP is removed only as a component of an HDL particle. In the absence of this additional catabolic route, the plasma half-time of lz5I-diLAP and the HDL particles with which it is associated is 12 h in rat.
Our data are the first unambiguous evidence that HDL is removed from plasma as a particle. The difference in the rates of removal of the various components of HDL from plasma is characteristic of a dynamic particle containing many chemical species that readily transfer to or exchange with other lipoproteins and cells that are in contact with the vascular compartment. Only measurements with a nonexchangeable species would accurately reflect the clearance kinetics for the entire particle. This route is presumed to occur via a specific receptor (Graham and Oram, 1987), and our data suggest that this occurs primarily in the liver, adrenal glands, and ovaries. The fractions of HDL, and HDLz that are targeted to these three tissue sites are different, and this implies that there are some tissue-specific factors that modulate the behavior of the HDL receptor. When labeled model HDL discs were incubated in normal plasma in vitro, only HDLp were formed; as previously reported by Gavish et al. (1987), no HDLl were found. However, when excess cholesterol was added to the incubation medium, HDL, were converted into much larger particles, suggesting that the formation of HDL1 may depend upon a source of additional cholesterol. Thus, factors that regulate the amount of cholesterol available for the formation of the larger particles could also determine which tissue sites will remove HDL from plasma.
I n Vivo Organ Distribution Volumes of HDL Particles in Rat-Our results permit us to distinguish the tissue sites where HDL, is localized from those of HDL,. As expected from previous studies (Ponsin et al., 1986;Glass et al., 1983), HDLz distributed into the steroidogenic tissues, liver, adrenal glands, and ovaries. HDLl accumulation was comparable to that of HDL, in the liver, but was reduced by 65-75% in adrenal glands and ovaries. Although it is well known that HDL are the main source of cholesterol for steroidogenic tissues in rats, our data show that this property is restricted, for the most part, to the HDLz subfraction. Of importance, no lZ5I-diLAP was found in the kidneys. This observation has important implications for HDL metabolism. First, it confirms that the transport of HDL apoproteins into the kidney is largely due to the fraction that is lipid-free and in the aqueous part of the plasma compartment. Second, it supports our conclusion that this peptide is nontransferable and remains in solution only when bound to lipid surfaces.
Initial Transformation of HDL Discs into HDL-The R-HDL resembled the nascent HDL discs that were secreted by rat liver (Hamilton et al., 1976). When these labeled HDLlike discs were incubated at 37 "C in human or rat plasma or were injected into rats, the labeled peptide always appeared in the HDL for too short a time to permit isolation of any of the discs from the initial mixture. Instead, the labels were found in the major HDL subfractions of rat and man, HDLz and HDL3, respectively. This was unexpected in view of the current concept that nascent HDL are transformed into spherical HDL by the successive effects of sequestering free cholesterol from other lipoprotein or tissue pools and action of the enzyme LCAT. Although this concept is based upon the high LCAT reactivity of nascent HDL, it does not require a priori that the observed morphological changes from discoidal to spherical HDL are necessarily driven only by LCAT activity; moreover, there is no evidence that this is the predominant mechanism in plasma in uiuo. In view of the rapidity of the initial transformation relative to the LCAT reaction, it is unlikely that the in uiuo HDL conversion process requires LCAT activity.
To distinguish spontaneous remodeling processes from those that are LCAT-driven, model HDL discs composed of DMPC or POPC were incubated with rat plasma as a function of temperature. When POPC was the phospholipid source, the transformation velocity increased with temperature and gave linear Arrhenius plots. In contrast, when DMPC, which exhibits a gel-to-liquid crystalline transition temperature at 23.7 "C (Hinz and Sturtevant, 1972), was substituted for POPC in the same experiment, the Arrhenius plots were composed of two straight lines that intersected at the transition temperature of the lipid. These data support the concept that the initial transformation of the R-HDL (and perhaps nascent HDL) into spherical HDL is purely a physicochemical process. One possible mechanism is that fragments of nascent or model HDL fuse with the surface of the smallest preexisting HDL particles, namely HDLz and HDL3 in rat and human plasma, respectively. This would permit the particle surface to increase in the metabolic step that precedes the action of LCAT, which forms the cholesteryl ester-rich core.
On the basis of these studies, a model of HDL interconversions can be proposed. The first step is the physicochemical association of R-HDL with HDLz, a process that is rapid and dependent upon the physical state of the lipid in the R-HDL. This process might also emulate the association of transient HDL species, which derive their cholesterol from peripheral tissue, with a preformed HDLz (Francone et al., 1989) and lead to the formation of a larger HDL species. Concomitantly, there may be spontaneous monomeric transfer of cholesterol from peripheral tissue to the HDLz. A large fraction of the rat plasma HDL is removed as HDL2, with the highest specific tissue association being found with ovaries and adrenal glands. However, the highest total association of HDLz with an organ was with liver. The association of the HDLz with liver cells may be mediated by an HDL-specific receptor (Graham and Oram, 1987), whereas uptake of HDL, is thought to involve B, E-receptors (Gordon et al., 1983; Eisenberg et al., 1984). Additional HDL remodeling is achieved by LCAT, which forms the end product of a cholesteryl esterrich HDL1. Future studies with diLAP should help determine how these processes occur in other mammalian species and what impact various drugs and diets have on them.