Evidence for the separation of albumin- and apo A-I-dependent mechanisms of cholesterol efflux from cultured fibroblasts into human plasma.

The role of albumin has been studied in the plasma-mediated efflux of cholesterol from cultured fibroblasts. Immunoaffinity chromatography of plasma on immobilized anti-albumin antibody decreased by 25-50% total efflux catalyzed by plasma. The remainder of the efflux-promoting capacity of plasma was deleted by immunoaffinity chromatography on antibody to apolipoprotein A-I, the major apoprotein of high density lipoprotein. Both components of efflux were saturable with half-saturation at 0.5-1.0% (v/v) plasma. However, the net transport of sterol from cells to medium catalyzed by lecithin:cholesterol acyltransferase, was not reduced by the deletion of the albumin-catalyzed component of efflux. This finding was confirmed with congenitally analbuminemic plasma. These results indicate that efflux to albumin and to high density lipoprotein in plasma represent independent mechanisms; only the latter is coupled to net transport.

The role of albumin has been studied in the plasmamediated efflux of cholesterol from cultured fibroblasts. Immunoaffinity chromatography of plasma on immobilized anti-albumin antibody decreased by 25-50% total efflux catalyzed by plasma. The remainder of the efflux-promoting capacity of plasma was deleted by immunoaffinity chromatography on antibody to apolipoprotein A-I, the major apoprotein of high density lipoprotein. Both components of efflux were saturable with half-saturation at 0.5-1.0% (v/v) plasma. However, the net transport of sterol from cells to medium, catalyzed by lecithin:cholesterol acyltransferase, was not reduced by the deletion of the albumin-catalyzed component of efflux. This finding was confirmed with congenitally analbuminemic plasma. These results indicate that efflux to albumin and to high density lipoprotein in plasma represent independent mechanisms; only the latter is coupled to net transport.
Cholesterol efflux from cultured cells is catalyzed by plasma (1)(2)(3), as well as by a number of components isolated from plasma, such as high and low density lipoproteins (1,2,4,5) and albumin (1,6). Efflux promoted by plasma or isolated lipoproteins furthermore is saturable, and on such grounds it has been suggested (7,8) that efflux might involve binding of sterol acceptors to cell surface sites. Binding sites for lipoprotein apoproteins have indeed been demonstrated on the surfaces of many cells (7)(8)(9)(10)(11)(12). On the other hand, biophysical evidence (13,14) has suggested rather that efflux was limited by diffusion across an unstirred water layer, without any necessity of cell-acceptor interaction.
In studies with whole plasma, it was recently shown that at least the bulk of efflux was closely coupled to the activities of lecithin:cholesterol acyltransferase and the cholesteryl ester transfer factor (3,15,16) leading to sterol net transport in normal plasma from cells to medium. In whole plasma, the largest part of efflux was dependent upon unassociated apo A-I'; however, a smaller rate of efflux was catalyzed by recrystallized human serum albumin (containing no apo A-I) at its physiological concentration (3). However, albumin has been shown to form no association with cholesterol in purified form (17). Several major unanswered questions therefore remain about the potential role of albumin in the promotion of * This research was supported by National Institutes of Health Grant HL-23738 and through Arteriosclerosis SCOR HL-14237. 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.
LThe abbreviations used are: apo A-I, apolipoprotein A-I; PBS, phosphate-buffered saline. cholesterol efflux and net transport. Does albumin in whole plasma have the same activity in the promotion of efflux as does the isolated protein? Does either apo A-I-or albumindependent efflux involve cell-surface binding? And if there is in plasma an albumin-dependent component of efflux, is it, as efflux-linked to apo A-I, coupled to lecithin:cholesterol acyltransferase activity and steryl ester transfer? The present experiments were designed to determine, in plasma, the significance of albumin-dependent efflux of cholesterol.

EXPERIMENTAL PROCEDURES
Cell Culture-Infant preputial skin fibroblasts were maintained at 37 C in the presence of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (3). For individual experiments, cells were plated at a density of 1-10 x 104 cells/6-cm Falcon dish. Three days before reaching the required density, the dishes were changed into medium containing calf serum labeled with [ 3 H]cholesterol (3,15). After incubation (72 h), the labeled medium was removed; the cells were washed three times with PBS containing recrystallized human serum albumin (Sigma Chemical Co.) (4 mg/ml; pH 7.4) and then three times with PBS alone. The cells were then incubated (15-60 min, 37 C) with PBS containing fractions of human plasma prepared by immunoaffinity chromatography (3,15).
Preparation of Plasma Fractions-Blood from normocholesterolemic laboratory donors was collected into one-twentieth volume of 0.2 M sodium citrate maintained in ice, and, also at ice temperatures, plasma was obtained by centrifugation (2000 x g, 30 min). Coagulation of plasma during later incubation was inhibited by removal of fibrinogen by chromatography on columns of antibody (raised in rabbits) to human fibrinogen, coupled to Sepharose 2B. These columns removed no detectable amount of cholesterol from plasma, plasma apoproteins, or albumin, as determined by the corresponding quantitative radial immunoassays (3,15,16).
Plasma apo A-I, apo B, apo E, and albumin were selectively removed from plasma by immunoaffinity chromatography with the corresponding immobilized antibodies. Recrystallized human serum albumin had been first purified from any apo A-I by passing it through immobilized anti-apo A-I before injection into rabbits of the albumin in complete Freund's adjuvant.
The complete removal of antigen, after specific immunoaffinity column chromatography, was confirmed in each experiment by immunoassay, if necessary, after concentration by reverse dialysis, under conditions such that a 2% contaminant of the original plasma antigen concentration would have been detected.
Assay of Sterol Efflux-Efflux was determined as the rate of appearance of [3H]cholesterol radioactivity in the medium from cells preincubated with the labeled sterol. As previously described (3), the specific activity of effluxed sterol was not significantly different from that of cell-free or ester sterol. The dishes of cells were washed three times with phosphate-buffered saline containing 0.05% albumin (pH 7.4) at ice temperature, then three times with PBS alone, to remove fetal calf serum. Three ml of medium containing plasma (free of fibrinogen) at the indicated dilution were then added, 1 ml was withdrawn for determination of initial radioactivity and free and ester sterol mass, and the plate containing the remaining 2 ml of medium was then incubated for 60 min at 37 C. To determine sterol net transport from cells to medium, empty dishes containing the same plasma medium were incubated simultaneously (3,15). Dishes containing cells and empty dishes were prepared and incubated in pentuplicate. At the end of the incubation period, a further 1-ml sample was taken for determination of radioactivity and free and ester sterol mass, and the remaining medium was washed from the dishes with PBS-albumin and PBS (three changes each). The cells were solubilized with 0.1 N NaOH, and cell-specific radioactivity was determined. The rate of appearance of radioactivity was linear under the conditions described in these studies. Efflux into medium containing no plasma was 3-6% of plasma values under the same conditions.
Assay of Sterol Net Transport-Pentuplicate empty dishes, and the same number of dishes containing fibroblasts, were incubated with medium containing plasma, as described for efflux. A 1-ml initial sample was taken from each dish for determination of free and ester cholesterol mass (18). After a 60-min incubation at 37 °C, a second 1ml sample was taken from each dish for a further free and ester sterol determination. Sterol net transport was determined as the difference in the decrease of medium-free sterol (during incubation at 37 C) between empty dishes and dishes containing cultured fibroblasts (3).
As previously demonstrated (3), the increase in lecithin:cholesterol acyltransferase-derived steryl ester is the same in the presence or absence of fibroblasts and, hence, total sterol demand for esterification is also equivalent. Hence, the difference in medium-free cholesterol decrease in the presence and absence of cells represents net sterol mass entering the medium during the incubation period. The concept has been further validated by cell-medium balance studies under different incubation conditions, as detailed below.
Preparation of Immobilized Lecithin-Cholesterol Liposomes-Liposomes were prepared from egg lecithin, dioleyl phosphatidylethanolamine (Sigma) (5%, w/w), and cholesterol (40%, w/w) in distilled water in the French pressure cell (19). The single walled vesicles so obtained (20) were then complexed to CNBr-activated Sepharose via the free amino group of the basic phospholipid. Any unreacted Sepharose active sites were deactivated with ethanolamine, and the prepared gel was finally washed with distilled water and ethylene glycol bis(fi-aminoethyl ether)-N,N,N',N'-tetraacetic acid before use in the incubation experiments.
Analysis of Antibody-Associated Lipids-Columns of immobilized antibody, through which plasma had been passed, were then washed with 0.9% NaCl-EDTA (1 mM) for 40 column volumes. Under these conditions, no further protein was detectable in the eluate. Lipids and proteins bound to the antibody gel were then released with 10 volumes of 3 M NaCNS (pH 7.0). The protein content of the eluate was determined after dialysis, and its composition by quantitative radial immunoassays. To extract lipids, the thiocyanate solution was mixed with an equal volume of CHO30H and CHCI3. Internal standards of 1,2-[H]cholesterol, 1-[ 14 C]oleic acid, 1-[ 14 C]lysolecithin, and ['H]lecithin (all from New England Nuclear) were added. The chloroform phase was washed (three times) with twice its volume of 0.15 M NaCl (pH 5.5). Sterol was determined enzymatically (18) in portions of the chloroform phase; unesterified fatty acids were determined by titration (21). Lecithin and lysolecithin were separated by thin layer chromatography on silica gel layers on glass plates developed in chloroform/methanol/water (65:35:3, v/v) and extracted with chloroform/methanol (1:2, v/v), and lipid phosphorus was determined (22). Recovery through the purification procedure was determined from the radioactivity of the internal lipid standards. Protein was assayed with the Folin phenol reagent (23).

Albumin-and Apo A-I-dependent
Efflux-When albumin was removed from plasma by immunoaffinity chromatography on immobilized anti-albumin antibody, there was a significant decrease in the ability of the plasma to promote cholesterol efflux from labeled fibroblasts. This decrease was evident at all concentrations of plasma ( Fig. 1). An analogous decrease in efflux was obtained when lipoproteins containing apo A-I were removed from the plasma with the corresponding immunoaffinity column. In eight experiments, the sum of the efflux-promoting activities removed by antibodies to apo A-I and albumin was 95 +± 9% of the activity of unfractionated plasma under the same conditions (range, 92-102%). The efflux components associated with the presence of apo A-I and of albumin therefore represented essentially the whole of efflux into plasma. As further shown in Fig. 1, both compo-nents of efflux showed a considerable degree of saturation when plasma was present at increasing concentrations. The proportion of apo A-I-dependent efflux was reproducible (+±10%) with different plasma samples from the same source, but varied between 40 and 75% of total effiux in plasma samples from six different normocholesterolemic donors.
A possible relationship between the two components of plasma efflux was further investigated by sequential affinity chromatography on antibodies to apo A-I and albumin. If the pathways in fact represented independent mechanisms of efflux, the effect of the removal of albumin should be independent of the previous removal of apo A-I. As shown in Fig.  2, this was found to be the case. In the experiment shown, the  reduction in efflux caused by the removal of albumin from whole plasma was 0.58 ng g-of cell sterol min -', while the further reduction by the same antibody, in plasma from which lipoproteins containing apo A-I had already been removed, was 0.52 ng pg-1 of cell sterol min -. This finding indicates that not only are the apo A-I-dependent and albumin-dependent components of efflux additive, but also that they are independent.
Effect of Cell Density on Efflux Rates-As shown in Fig. 3, the efflux rates promoted by unfractionated plasma were strongly dependent upon cell density. At low density (<1 #Lg of cell sterol/6-cm culture dish), total efflux was 10-fold greater than in confluent cells. Minimal efflux rates were reached when the dishes were -50% confluent. There was no difference in the sterol/protein ratio of the cells between the highest and lowest densities used (mean, 36.8 + 0.9 g of sterol/mg of protein; range, 36.0-37.4 g of sterol/mg of protein). A similar effect of cell density has also been observed for cultured vascular smooth muscle and microvessel endothelial cells. 2 There was no difference in efflux rates at any cell density when normal fibroblasts were compared to fibroblasts lacking lipoprotein high affinity receptors. This confirms and extends the earlier finding of Wu and Bailey (4). The property of the density dependence of efflux was used to determine how the proportions of albumin-dependent and apo A-I-dependent efflux would depend on absolute efflux rates. As shown in Table I, there was no significant difference in the proportion of each component of efflux when efflux rates varied over a 12-fold range. Apolipoprotein E, a component of cholesterol-rich lipoproteins in plasma (24), has been implicated in sterol transport from cells. There was no effect of the removal of apo E from plasma on sterol efflux from fibroblasts at any cell density. These findings indicated that the rates of apo A-I-and albumin-dependent components of efflux are proportionate rather than absolute.  (5,7,12). Such binding has been suggested as a necessary intermediate step in sterol efflux (7). To determine whether the functions of either albumin or apo A-I in efflux require cell membranes as cholesterol donor, the labeled fibroblasts in the incubation experiments were replaced by labeled cholesterol-lecithin liposomes containing 40% (w/w) cholesterol, similar in proportion to the cell membranes (25). As shown in Table II, the characteristics of efflux into plasma from these liposomes were almost identical with those from cultured fibroblasts. The proportions of total efflux dependent upon the presence of apo A-I and albumin were also similar to those found in the cell experiments. This finding indicates that liposomes with a similar cholesterol-lecithin ratio to fibroblast membranes are as active as the membranes themselves in efflux to cholesterol acceptors in plasma. Cholesterol net transport from liposomes to plasma was also at rates similar to those found with fibroblasts, indicating the coupling of efflux to esterification in plasma (Table II). As with fibroblasts, the rate of cholesterol net transport was equivalent to the rate of apo A-I-dependent efflux.

Effects of Metabolic Inhibitors and Temperature on Efflux
Rates-Further investigation of the mechanism and regulation of efflux was made using a range of metabolic and transport inhibitors. As shown in Table III, there was little effect on total efflux by any of these agents except colchicine, which mediated a modest but significant increase in total efflux. However, the proportions of albumin-and apo A-I- Plasma, unfractionated or after removal of individual antigens by affinity chromatography, was incubated at a dilution of 1.2% (v/v) with 6-cm dishes of fibroblasts containing 0.46, 3.67, or 14.3 g of sterol/dish, and labeled with [ 3 H]cholesterol to specific activities of 6.0 x 104, 5.3 x 104, and 3.6 x 104 cpm//pg of sterol, respectively. Each experimental point represents the mean ± S.D. of quadruplicate dishes of cells. 3 mrl of plasma or plasma fractions in PBS were added to each dish, an initial 1-ml sample was taken for analysis, and a second 1-ml sample was taken after incubation for 60 min at 37 C.  Efflux and net transport from lecithin-cholesterol liposomes to plasma Lecithin-cholesterol liposomes, prepared as described under "Experimental Procedures" to contain 5% (w/w) phosphatidylethanolamine (Ref. 19) were complexed with CNBr-Sepharose 6B. Assay tubes were prepared to contain 10.0 pLg of cholesterol (specific activity, 9.4 x 10o cpm/jxg) in liposome complex form, and either plasma, or plasma from which individual antigens had been removed, at a concentration of 1.2% (v/v) in PBS, in an initial volume of 3 ml. After removal of an initial 1.0-ml medium sample, assays were incubated for 2 h at 37 °C, after which a second sample was taken for analysis. The gel was settled by centrifugation (5 main, 1000 x g at 4 °C) before collection of samples for determination of final medium cholesterol mass and radioactivity. Tubes containing no gel were incubated simultaneously to permit determination of cholesterol net transport, as described under "Experimental Procedures." dependent efflux were unchanged, and net transport increased 15-20%. The effect of temperature was determined over the range of 0-37 C. Activation energies of efflux were calculated from the Arrhenius plots obtained with unfractionated or albuminor apo A-I-depleted plasma (Fig. 4). The activation energy of total efflux was 10.7 ± 1.0 kcal/mol (three determinations, range of 9.9-11.4 kcal/mol). Although slightly different plots were obtained for fractionated plasma, such differences did not reach significance and the energy of efflux was in large part independent of the nature of the sterol acceptor in plasma.

Sterol Efflux and Sterol Net Transport in Plasma-As
reported earlier (3), efflux may or may not result in sterol net transport, depending in large part on whether efflux is coupled to sterol esterification catalyzed by the lecithin:cholesterol acyltransferase reaction. When efflux was assayed isotopically, a significant portion of total efflux was dependent upon the presence of albumin but when albumin was removed from plasma by immunoaffinity chromatography, there was no reduction in either lecithin:cholesterol acyltransferase activity

TABLE III Effects of metabolic inhibitors on sterol efflux
Dishes of cells containing 10 pg of sterol labeled with [ 3 Hlcholesterol were washed with PBS and preincubated, with the same buffer alone or with the antimetabolites indicated below, for 60 min. This medium was then replaced with 1.2% plasma (v/v)-PBS (3 ml), either alone or with the same antimetabolite. After removal of an initial 1ml medium sample, the cells were incubated for 60 mmin at 37 °C, after which a second 1-ml medium sample was taken for analysis of radioactivity. Experimental values represent means + S.D. of triplicate determinations. EGTA, ethylene glycol bis(8i-aminoethyl ether) -N,N,N',N'- or the rate of sterol net transport from the fibroblasts to the culture medium (Fig. 5). This finding indicates that only that efflux mediated by apo A-I is significant in promoting net transport of sterol from cells to plasma-containing medium. This result was further validated by balance studies in which both cell and medium cholesterol mass was measured during net transport. In a representative study (of three), initial cell cholesterol mass was 11.8 ± 0.1 pg/dish and final cell cholesterol mass was 11.1 ± 0.1 ,g/dish, a difference of 0.7 pg. Using the difference assay described under "Experimental Procedures," cholesterol net transport, in terms of medium cholesterol mass, was 0.65 pg/dish under the same conditions. When apo A-I was removed from the plasma medium by immunoaffinity chromatography, cholesterol net transport was reduced to 0.04 pg, while after removal of albumin it was 0.65 #g, compared to the (0.65 -0.04)) or 0.61 pg expected by difference. The specific activity of the cells in this experiment was 9875 cpm/pg. Initial cell radioactivity was 1.05 + 0.03 x 10 5 cpm and final radioactivity was 0.94 ± 0.01 x 105 cpm, for a total efflux of 0.11 x 105 cpm or a calculated total efflux of 1.1 pg of cholesterol. This was reduced by anti-apo A-I affinity chromatography to 0.44 x 104 cpm or a calculated 0.67 pg of sterol mass. This can be compared with the reduction of 0.61 pg of sterol determined by direct mass measurement. Finally, the efflux in the absence of albumin in the same experiment was 0.58 X 10 4 cpm or a decrease of 5200 cpm, or 0.53 pg. The sum of apo A-I-and albumin-dependent efflux in this case is therefore 1.14 pg of sterol versus a measured total efflux of 1.10 pg. The efflux to albumin (0.53 g) can be compared with the difference between total efflux and net transport (1.1 -0.65 = 0.45 pg of cholesterol).
These results therefore support the concept that the indirect measures of efflux and net transport from medium values reflect the expected changes within the cells, and that only efflux coupled to apo A-I mediates sterol mass transport. In the three studies, differences calculated and found between initial and final cell cholesterol levels were 6-16%.
To confirm that fractionation by affinity chromatography was not a factor in this finding, human analbuminemic plasma was used in place of normal plasma in the cell incubation studies. As shown in Table IV, in this plasma, essentially the whole of efflux is dependent upon apo A-I, and sterol net transport was as high as in normal plasma. This finding  5. Effects of removal of plasma albumin on sterol net transport. Cultured fibroblasts in 6-cm dishes, or empty dishes, were incubated in pentuplicate with unfractionated plasma or plasma from which albumin (ALB) had been removed by affinity chromatography. Incubation was for 60 min at 37 C. Initial and final 1-ml samples were taken for analysis of free and ester sterol (18). Esterification (EST) was determined from the decrease in medium-free sterol during incubation in the absence of cells. Sterol net transport (TRANS) was determined from the difference between the decrease in medium-free sterol in the presence and absence of fibroblasts, as described under "Experimental Procedures." 1 i -supports the concept that albumin plays no essential role in sterol transport and esterification in plasma.

Lipids Associated with Albumin in Plasma-The results
reported above indicate that albumin promotes sterol efflux but that this efflux represents exchange and is not coupled to net transport. This requires the presence on albumin of an exchangeable pool of free sterol. No interaction could earlier be detected between highly purified albumin and cholesterol in terms of binding constants (17). This should not exclude, however, interaction between cholesterol and albumin in the presence of other lipids. The possible existence of an albuminassociated sterol pool was investigated by immunoaffinity chromatography of plasma from which lipoproteins had been previously removed with the appropriate immobilized antibodies. As shown in Fig. 6, elution of the bound albumin after extensive washing indicated the presence of free sterol together with phospholipid and large amounts of unesterified fatty acid. The molar ratio between free sterol and phospho- b Total sterol esterified in the medium in the same period (micrograms of sterol esterified/dish).
cSterol contributed from cells to medium in the same period (micrograms/dish). Apo A-I levels in control and analbuminemic plasma were 1.32 and 1.71 pg/ml, respectively. The absence of albumin from the plasma of the analbuminemic donor has been documented elsewhere (29). lipid was -0.5 and total albumin-associated cholesterol represented about 1% of plasma free sterol. This finding indicates that, in human plasma, a small proportion of total free sterol is bound to albumin and may function there as a pool exchangeable with sterol from cell membranes. DISCUSSION There has been considerable disagreement in the recent literature on the nature of the physiological acceptors of cell sterol efflux. Studies using radiolabeled cells have found that a wide range of plasma fractions in purified form car. promote sterol efflux (1-6). These fractions include each of the major plasma lipoprotein classes, lipoprotein-deficient plasma, and albumin either alone or in the presence of phospholipids. On the other hand, in studies where mass transport of sterol between cells and their medium was measured, either directly (3) or indirectly via the effects of cell sterol on a receptor assay (26), efflux was almost wholly dependent upon the presence of high density lipoproteins (27) and, in particular, apoprotein A-I in a lipoprotein form unassociated with other apolipoproteins. The solution to this paradox is shown in the present study to lie in the finding that plasma contains two distinct and independent pathways, both of which promote the efflux of sterol from cultured cells. However, only one of these pathways, that dependent upon apo A-I, mediates sterol net transport from cells. The other, dependent upon the presence of albumin, takes part in an exchange reaction that is not coupled to net transport. While the function of this latter pathway is presently not evident, it clearly represents a considerable potential artifact in studies of sterol transport, since as much as half of total efflux from labeled cells may be uninvolved in the regulation of cell cholesterol content. It was previously shown that, in unfractionated plasma, the apo A-Idependent component of efflux is coupled to the activity of lecithin:cholesterol acyltransferase. In the present study, this reaction was not decreased in the absence of plasma albumin, whether in fractionated normal or in analbuminemic plasma.
The contributions of the two pathways to efflux were independent of total efflux when the latter was varied by temperature or cell density. This finding suggests that these acceptors act at a point distal to the regulatory step of efflux. This conclusion is reinforced by the observation that the contributions of the albumin-and apo A-I-dependent routes of efflux were essentially the same whether cultured cells or cholesterol-lecithin vesicles immobilized on Sepharose provided the labeled substrate, when the concentration of sterol was the same. Furthermore, the energy of activation for efflux of cholesterol from cells to plasma was also highly similar to that found for cholesterol efflux from vesicles to a variety of media (13,14,28). Two types of theory are currently held on the mechanism of sterol efflux. One involves binding of acceptors to the cell surface and subsequent dissociation of the carrier, enriched with cholesterol, into the medium. The second envisages a rate-limiting diffusion step from the cell surface across an unstirred water layer, after which cholesterol would be incorporated with plasma acceptors. The results of the present study appear most in accord with the second concept, since the kinetics of efflux from cell surfaces and from liposomes were very similar, although an interaction with the membrane surface that is not rate limiting would not be ruled out. The present study, in agreement with an earlier report (4), found no difference in efflux rates from fibroblasts homozygous for deficiency of low density lipoprotein receptors (FH cells) and control cells. However, leukocytes heterozygous for the same genetic factor have been reported to show an increased efflux rate (30). Further studies with a variety of cultured cell types from this disorder will be required to determine the basis of this difference.
It should be emphasized that only a small proportion of the total apo A-I or albumin in plasma appears to be involved in the sterol efflux reaction. As previously described (3), only about 5% of total apo A-I is in unassociated form and it is this fraction which is coupled to esterification via the lecithin:cholesterol acyltransferase reaction. While the recovery of cholesterol with albumin using affinity chromatography is probably incomplete, only a small proportion of the albumin in plasma (0.6 umol/ml) could contain bound cholesterol; even the binding of sterol to a single site per molecule would involve the major part of plasma free sterol, a finding incompatible with the affinity chromatography of plasma lipoproteins (3), as well as extensive ultracentrifugal data. Whether this portion is distinguished by physical or chemical properties (for example, association with specific phospholipids) is presently unknown. Taken together, the findings of this study indicate that in unfractionated plasma there is considerable heterogeneity in the associations of free sterol molecules derived from cells with plasma factors. These associations have significance because they determine, at least in part, the subsequent metabolic fate of the sterol.