Cholesteryl Ester Transfer Protein Is Secreted by Hep G 2 Cells and Contains Asparagine-linked Carbohydrate and Sialic Acid*

A cholesteryl ester transfer protein (CETP) of ap- parent M, 74,000 has recently been purified from human plasma. Cholesteryl ester transfer activity was found to accumulate in the medium of cultured Hep G2 cells. The transfer activity was removed by immuno- precipitation with specific antibodies to the plasma CETP. Sodium dodecyl sulfate gel electrophoresis of immunoprecipitates prepared from the medium of cells pulsed with [S6S]methionine revealed a broad specific band of protein of M, 72,000 to 76,000; by contrast, immunoprecipitates of cellular homogenates showed a sharp specific band of M, 58,000. The M, 72,000 to 76,000 band disappears, concomitant with the appear- ance of lower M, products, upon neuraminidase or glycopeptidase F treatment of medium immunoprecipitates or of purified CETP. The results indicate that liver cells have the capacity to synthesize and secrete CETP. The CETP peptide acquires asparagine-linked carbohydrate and sialic acid during intracellular proc- essing. One or more plasma proteins facilitate the transfer and exchange of cholesteryl esters (CE),’ triglycerides, and phos-pholipids among the lipoproteins (1-5). Recently, a plasma cholesteryl ester transfer protein (CETP) has been purified

model liver cell, the Hep G2 cell, and to examine the molecular form of the newly secreted CETP.

MATERIALS AND METHODS
Cholesteryl Ester Transfer Activity in Hep G2 Cell Medium-Hep G2 cells were grown on 100-mm collagen-coated plastic Petri dishes in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 volume % fetal bovine serum, 2 mM glutamine, 100 units of penicillin/ml and 0.1 mg of streptomycin/ml. Cells (3 X lo5) were seeded on 35-mm collagen-coated Petri plates and grown for 2 days until 75% confluent. Experimental plates and cell-free controls were washed three times with DMEM and reincubated with fresh full medium (DMEM plus fetal bovine serum, glutamine, penicillin, and streptomycin). Three plates per time point (24,48,72, and 96 h) were used, two with cells and one cell-free control. CE transfer activity was measured by the transfer of radiolabeled 3H-cholesteryl esters from HDL to LDL, as previously described (1). To further increase the sensitivity of the assay, final assay volumes were reduced to 200 pl and samples were incubated for 24 h in a 37 "C shaking water bath. Following incubation, lipoproteins were separated by ultracentrifugation a t d = 1.063 g/ml or by precipitation of LDL with heparin/ MnC12, with similar results. Facilitated cholesteryl ester transfer (Fig. 1) was measured by subtracting the blank value (transfer of radiolabeled cholesteryl ester from HDL to LDL in media from cell-free plates) from the experimental value (transfer in cell-conditioned media). Generally, the blank value was about 10% of the experimental value.
Immunoprecipitation of CETP Activity in Hep G2 Cell Medium-Hep G2 cells were grown on 100-mm Petri dishes as described above, washed twice with phosphate-buffered saline (PBS), washed once with DMEM, and then incubated with 10 ml of medium containing one of the following: (a) anti-CETP rabbit IgG ( 8 4 pg/ml) prepared by ammonium sulfate precipitation and DEAE-cellulose chromatography as previously described (l), (b) nonimmune rabbit IgG (84 pg/ ml), (c) goat anti-rabbit IgG, or ( d ) no IgG. After 48 h, the medium was harvested and a 30-fold excess of goat anti-rabbit IgG or an equal volume of 50 mM Tris-HC1, pH 7.5, 150 mM NaC1, 2 mM EDTA was added. This solution was reincubated for 24 h at 4 "C. Immunoprecipitates were pelleted in a microfuge at top speed and 0.5 ml of the supernatant was assayed for CE transfer activity.
[35S]Methionine Labeling of Newly Synthesized Protein-Hep G2 cells were grown until 65% confluent, washed three times with PBS and incubated for 8 h in 10 ml of methionine-free DMEM. Cells were rewashed with PBS three times and reincubated with 5 ml of fresh methionine-free DMEM containing 1 mCi of [35S]methionine (Du Pont-New England Nuclear, 1086 Ci/mmol) for 20 h. Seventy percent of the radiolabel was incorporated into secreted protein at this time. Labeled protein was immunoprecipitated, as described (6, 7). The medium was cleared using inactivated Staphylococcus aureus. Either 100 pg of anti-CETP IgG or 100 pg of nonimmune IgG was added per 0.5 ml of media for 24 h at 4 "C. Immunoprecipitated protein was dissociated from immune complexes by solubilization in 0.125 M Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, and 0.5% bromphenol blue and heating at 100 "C for 5 min. Samples were analyzed by electrophoresis on a 6-16% polyacrylamide Laemmli SDS slab gel and subjected to autoradiography. To prepare Hep G2 cell homogenates, cells from a confluent 100-mm dish that had been pulsed with [35S]methionine were washed three times with PBS and harvested by scraping into 2 ml of 50 mM Tris-HC1, pH 7.4, 150 mM NaC1, 2 mM EDTA, containing 2 mM phenylmethylsulfonyl fluoride and 2 mM benzamidine. Five hundred-pl aliquots were treated as described above to immunoprecipitate newly synthesized protein.
Digestion of Purified CETP with Glycohydrolases-Two-pg samples of purified CETP (1) were digested with glycopeptidase F or neuraminidase, P-galactosidase, or 0-glycanase (Genzyme). The glycopeptidase F incubation mixture consisted of 1 unit of glycopeptidase F in 100 mM sodium phosphate, pH 8. Neuraminidase digestions were performed with 60 or 180 milliunits of neuraminidase in 50 mM sodium acetate, pH 5.5, 150 mM NaC1, 4 mM CaC1,. P-Galactosidase incubation consisted of 40 units of @galactosidase in 60 mM sodium maleate, pH 6.0. All three incubation mixtures contained 1% Triton X-100 (w/v), 0.02% SDS, and 3 mM dithiothreitol, 2 mM diethyl pnitrophenyl phosphate (E 600) in a final volume of 100 pl. Samples of purified CETP (2 pg) were also treated with 0-glycanase (endo-0-N-acetylgalactosaminidase) by incubating with 60 milliunits of neuraminidase and 1 or 4 milliunits of 0-glycanase in 20 mM Tris maleate, pH 5.5,2 mM calcium acetate, 2 mM diethyl p-nitrophenyl phosphate in a final volume of 100 pl. T o measure the activity of purified CETP following removal of carbohydrate, these digestions were performed in the same buffers without SDS or Triton X-100. The digestions were incubated for 16 h a t 37 "C under argon. Samples were prepared for electrophoresis by precipitating once with 1 ml of cold acetone. Protein pellets were solubilized as described above for SDS-PAGE except that the samples were heated a t 60 "C for 20 min.

RESULTS
There was a time-dependent accumulation of cholesteryl ester transfer activity in the medium of cultured Hep G2 cells, as shown for a representative experiment in Fig. 1. A similar secretion of CE transfer activity by Hep G2 cells was found in five separate experiments. The transfer activity was demonstrated in an assay which measures the transfer of radiolabeled cholesteryl esters from HDL to LDL. The results of this assay can be influenced by several nonspecific variables. T o determine if the cholesteryl ester transfer activity was in fact due to the previously described CETP (l), the medium was subjected to immunoprecipitation, using specific IgG prepared by immunization of a rabbit with purified CETP (1). This resulted in removal of more than 80% of the CE transfer activity from the medium (Fig. 2). By contrast, nonimmune I g G or second antibody alone had no significant effect on cholesteryl ester transfer activity in the medium. Thus, there is specific immunoprecipitation of cholesteryl ester transfer activity from the medium of Hep G2 cells by CETP-specific antibodies.
In order to visualize CETP in the immunoprecipitates, Hep G2 cells were pulsed with ["SJmethionine to label newly synthesized proteins, the medium was precipitated with CETP antibodies, and the immunoprecipitates were analyzed by SDS-PAGE and autoradiography. There was specific immunoprecipitation of a protein of M , approximately 72,000-76,000 (Fig. 3, lane D), identical in electrophoretic mobility to the previously purified CETP (1). Although both nonimmune and immune I g G caused precipitation of several other lesser bands, only the band of M , 72,000-76,000 was specifically immunoprecipitated (cf. Fig. 3, lanes C and D). In contrast to the results obtained with the cellular media, immunoprecipitates of cellular homogenates showed a sharp specific band of average M , 58,000 (Fig. 3, lane B The broadness of the band of CETP in SDS gels, as well as its high M , (72,000-76,000), relative to a probable intracellular precursor ( M , 58,000), suggested that the mobility of CETP in SDS gels might be influenced by the presence of carbohydrate. Accordingly, the purified CETP was treated with a variety of enzymes, which remove carbohydrate, and then analyzed by SDS gel electrophoresis and immunoblotting with specific CETP antibodies. Treatment of CETP with glyco-peptidase F (to remove asparagine ("linked sugars) resulted in the formation of three sharp bands of M , 64,000, 62,000, and 60,000 (Fig. 4, lane B). When the amount of glycopeptidase F was increased (&fold), the lowest of the bands became most pronounced, but there was no further decrease in M , (not shown). This result suggests that the three bands resulted from partial digestion of CETP, with a limiting M , of about 60,000 resulting from complete removal of N-linked sugars. Digestion of purified CETP with neuraminidase resulted in a smaller change in M , to about 67,000 (Fig. 4, lane C). Although the neuraminidase treatment consistently resulted in a 4-to 6-kDa decrease in M , ( n = 7), the apparent M , was slightly different depending on the electrophoresis conditions (cf. Fig.   4, lanes A and C, with Fig. 4, lanes I and G). Treatment of CETP with higher doses of neuraminidase resulted in formation of two additional products of much lower M , (53,000 and 45,000, Fig. 4, lane M ) . However, this pattern was noted to be similar to that produced by a number of proteases acting on CETP, including trypsin, chymotrypsin, and an endogenous protease copurifying with the CETP, with the latter producing initial degradation fragments of M , 56,000 and 48,000, which can be seen in the heavily loaded lane of control CETP in Fig. 4, lane J. When an inhibitor of serine proteases (2 mM E 600) was included, the major product of neuraminidase digestion still had M , 67,000 but the lower M , degradation fragments were much reduced in amount (Fig. 4, lane L). Thus, removal of sialic acid produced a protein of apparent M , 67,000, and the lower M , fragments resulted from a protease contaminating the neuraminidase preparation. By contrast, the product of glycopeptidase F digestion was not influenced by the presence of protease inhibitors. Sequential treatments with neuraminidase and then glycopeptidase F or glycopeptidase F and then neuraminidase produced no further change in M , compared to glycopeptidase F alone, implying that the sialic acid of CETP is attached to the N-linked sugar. Also, treatment of purified CETP with neuraminidase and then 0-glycanase (to remove core disaccharides of structure Gal@(1,3)GalNAc linked to either serine or threonine) resulted in no change in mobility (Fig. 4, lanes E and F) compared to neuraminidase alone (Fig. 4, lane G). Treatment of purified CETP with P-galactosidase (to remove terminal galactose) resulted in no change in mobility; the CETP band was diffuse after this treatment because of the high salt content in the buffer (Fig. 4, lane K ) but a similar effect was produced by adding the buffer alone to control CETP. Similar results to Fig. 4 were obtained when immunoprecipitates of newly synthesized CETP were treated with glycopeptidase F ( M , of product 59,000), neuraminidase (68,000) and P-galactosidase ( M , 74,000). To see if N-linked sugars were necessary for secretion of CETP, cells were treated with tunicamycin (2 pg/ml). This resulted in a pronounced (>go%) reduction in the amount of the 74-kDa band present in the medium; no lower M , forms of CETP were immunoprecipitated from the medium of tunicamycin-treated cells (not shown).
In further experiments the activity of purified CETP was measured following removal of carbohydrate. The removal of carbohydrate resulted in relatively little change in cholesteryl ester transfer activity which was 94% of control (neuraminidase) or 84% of control (glycopeptidase F at maximum dose).
In this set of experiments, complete digestion of the parent CETP molecule was verified by SDS-PAGE.

DISCUSSION
The present study demonstrates the secretion of cholesteryl ester transfer activity by cultured Hep G2 cells. The removal of activity by precipitation with specific CETP antibodies, along with the appearance of a single specific band corresponding to CETP in immunoprecipitates, shows that the transfer activity was due to secretion of CETP by Hep G2 cells. The coincidence in M , of the secreted form of CETP with that of CETP purified from plasma (1) indicates that the latter is not itself a breakdown fragment formed during the purification procedure. Faust and Albers (8) have also recently reported the secretion of cholesteryl ester transfer activity by Hep G2 cells; however, this report did not provide molecular identification of the CETP.
The broadness of the band of CETP precipitated from medium ( Fig. 3) or isolat.ed from plasma (1) suggested the possibility of protein microheterogeneity. Neuraminidase or glycopeptidase F treatment of the CETP resulted in an increase in mobility and sharpening of the CETP band in SDSpolyacrylamide gels, forming a peptide of similar M , to a CETP precursor ( M , 58,000) specifically immunoprecipitated from cellular homogenates.' These results suggest that the CETP peptide ( M , < 58,000) acquires N-linked carbohydrate and sialic acid during intracellular processing. Consistent with this suggestion, a cDNA to the CETP, obtained from a human liver library, encodes an M , 53,000 protein which displays four potential asparagine-linked glycosylation sites (9). Since glycopeptidase F cleaves at the N-glycosidic linkage (lo), the presence of three distinct bands in partial digests with glycopeptidase F (Fig. 4, lane B ) suggests that at least three of the four sites are glycosylated. Despite treatments of CETP with neuraminidase, glycopeptidase F, /3-galactosidase, and O-gly-* In an attempt to demonstrate a higher M, form of CETP within cells, cells were pulsed for 1 h with ["SS]methionine and then chased for 1 h with cold methionine. However, no specific bands between 53 and 74 kDa were seen, suggesting rapid secretion of CETP following the addition of sialic acid in the Golgi apparatus. canase, in increasing doses or in combinations, we were not secrete a lipid transfer activity with some characteristics of able to produce a peptide of M , 53,000. Several possibilities CETP (13), and the mRNA for CETP is present in spleen could explain the residual difference of about 6 kDa between and other peripheral tissues (9). Thus, it is likely that CETP the glycopeptidase F-digested protein and the cDNA-encoded synthesis is not confined to the liver and intestine, like most peptide (8). For example, there could be 0-linked sugar, not of the apoproteins, but rather that it shows a more widespread susceptible to digestion by neuraminidase and 0-glycanase, tissue distribution, and perhaps, like apoE (14), participates or additional post-translational modifications such as phos-in the localized redistribution of cholesterol within tissues. phorylation o r addition of fatty acids.
Although the addition of N-linked sugars seemed to be necessary for the normal secretion of CETP (as shown by lack of secretion of CETP by tunicamycin-treated cells), the removal of carbohydrate from purified CETP resulted in only minor reductions of cholesteryl transfer activity. Thus, despite the unusually high content of hydrophobic amino acid residues in CETP (l), the N-linked carbohydrate and sialic acid are not essential for activity of purified CETP. By contrast, small amounts of phospholipid bound to purified CETP are needed to maintain cholesteryl ester transfer activ-it^.^ Previously, cholesteryl ester transfer activity has been shown to accumulate in the perfusate of rabbit liver, suggesting hepatic synthesis and secretion of a cholesteryl ester transfer protein (11). The present investigation indicates that hepatocytes have the capacity to synthesize and secrete CETP. The CETP can increase the transfer of CE from HDL into cultured Hep G2 cells (la), raising the possibility that CETP secreted by the liver cell might enhance the cellular uptake of CE in the same tissue microenvironment. Cultured monocyte macrophages have been shown to synthesize and