Regulated Vectorial Secretion of Cholesteryl Ester Transfer Protein (LTP-I) by the CaCo-2 Model of Human Enterocyte Epithelium*

We have investigated the human CaCo-2 enterocyte model for secretion of the plasma cholesteryl ester transfer protein, LTP-I. CaCo-2 cells secrete a cholesteryl ester transfer protein which possesses molecular identity with plasma LTP-I, demonstrated by anti- LTP-I immunoblot analysis and immunoinhibition of all cell-secreted cholesteryl ester transfer activity. When CaCo-2 are cultured on permeable membranes, cholesteryl ester transfer activity is detected only in the lower culture compartment. Thus, CaCo-2 vecto- rially sort and secrete LTP-I, as well as the intestinal apolipoproteins, from the basolateral cellular domain. Over a 24-h period, CaCo-2 secrete cholesteryl ester transfer activity in a time-dependent manner, at approximately twice the rate of HepGZ. Furthermore, CaCo-2 enterocytes, but not HepGZ hepatocytes, regulate LTP-I secretion in response to fatty acid concen- tration in the culture medium. Based on these obser-vations, we speculate that the intestine may be the principal regulated source of human plasma LTP-I. We have reported that the human plasma cholesteryl ester (CE)’ transfer protein, LTP-I (l), mediates the transfer of neutral lipids, such as triglycerides, and negative lipids, such as phosphatidylcholine, in

macrophages (7) and to HepGZ hepatocytes (8). It is intriguing that both the human monocyte-derived macrophage (9) and the HepGZ hepatocyte (10) synthesize and secrete LTP-I. However, the principal cell and tissue sources of LTP-I in human plasma remain a matter of speculation.
It has recently been reported that human intestine contains the mRNA for LTP-I (11). However, neither the specific cell source of this mRNA, nor the nature of the secreted protein or its regulation are known. We have therefore adopted the CaCo-2 membrane model of enterocyte epithelium (12) as a paradigm for studying intestinal synthesis and secretion of The following experiments demonstrate for the first time: 1) CaCo-2 enterocytes secrete functional LTP-I; 2) as with the intestinal apolipoproteins, postconfluent CaCo-2 cells cultured on permeable membranes secrete LTP-I in a polar fashion, from the basolateral (serosal) cellular domain; 3) LTP-I secretion by the CaCo-2 epithelium cultured on membranes is regulated by "luminal" fatty acids.

Materials
The human CaCo-2 enterocyte and HepG2 hepatocyte lines were purchased from ATCC (American Type Culture Collection, Rockville, MD). All culture media, supplements, and reagents were from Gibco Laboratories and Sigma Chemical Co. T-75 culture flasks were from Corning (Ithaca, NY); "Transwell" plates were from Costar (Van Nuys, CA). Histochemical stains and reagents were purchased from Sigma. All radioisotopes were from Du Pont-New England Nuclear. Supplies and reagents for electron microscopy were from Ted Pella, Inc. (Tustin, CA). All centrifugation equipment and supplies were from Beckman Instruments.

Methods
Cell Culture"HepG2 hepatocarcinoma cells were cultured and maintained as described (10). Cell stocks of CaCo-2 colonic adenocarcinoma cells were cultured in Dulbecco's modified minimal essential medium (DMEM) with 15% fetal calf serum and maintained as described (13). For experiments described here, cells were seeded on permeable culture membrane (PCM) cups essentially as described by Traber et al. (12). Fetal calf serum-DMEM growth medium in both upper and lower chambers was replaced daily from the time of plating to initiation of experiments at approximately 14 days postconfluency. Confluent cell monolayers at approximately 14 days postconfluency which had developed functional tight junctions (see below) were washed twice with Earle's saline buffer supplemented with Ca*+ and M$+ to maintain tight junction integrity (14), followed by a single DMEM wash, and finally replaced in serum-free (SF) DMEM to condition medium for analysis of secreted lipid transfer activity (SF-DMEM consisted of DMEM supplemented with 4 mg/ml lactalbumin hydrolysate). This final SF-DMEM wash was frozen for later use as a negative control in the radioassay of CE transfer activity. Following experiments, HepG2 and CaCo-2 cell viability was estimated by trypan blue exclusion, and cell monolayers were extracted with 0.1 N NaOH for determination of monolayer protein. All CaCo-2 cells used in these experiments were from passages 20-25; HepG2 were from passages 75-80.
Electron Microscorn-Transmission electron microscopy of CaCo-2 cultured on membranes, to confirm polar cell morphology, was performed by standard methods as described (13).
Assessment of CaCo-2 Monolayer Integrity-Functional integrity of CaCo-2 epithelial monolayers was assessed essentially as described (15). Briefly, permeability of intercellular tight junctions was tested by monitoring bidirectional diffusion of horseradish peroxidase across the cellular monolayer, at both 4 and 37 "C. Horseradish peroxidase diffusion was tested across every CaCo-2 monolayer, and only those manifesting a permeability barrier to macromolecular diffusion were used for further experiments.
Fatty Acid Challenge-CaCo-2 cells were grown on PCM to approximately 14 days postconfluency. CaCo-2 monolayers which were determined to have impermeable intercellular junctions were washed once with Earle's saline, once with SF-DMEM, and finally SF-DMEM added to both upper and lower culture chambers. SF-DMEM in the upper PCM chamber was supplemented with sodium oleate to 0, 62.5, 125, 250, or 500 pM with a sodium o1eate:albumin emulsion, prepared according to the method of Patsch et al. (16). HepG2 cells were cultured to approximately 90% confluency and washed in preparation for incubations as described (10). HepG2 were then incubated with SF-DMEM in the presence of these same concentrations of sodium oleate. Following a 24-h incubation at 37 "C, cell-conditioned media were collected and centrifuged to remove cell debris. For CaCo-2 cells, medium was collected from both upper and lower culture chambers. Conditioned medium samples from CaCo-2 and HepG2 were then assayed for CE transfer activity.
Radioussay of Cholesteryl Ester TransferActiuity-["C]Cholesteryl ester (CE)-HDL donor lipoprotein was prepared as described previously (1). Approximate specific activity of this (CE)-HDL donor lipoprotein was approximately 3500 dpm/pg of CE (assuming a molecular weight of 650 for CE).
Radioassays of CE transfer activity in cell-conditioned media were performed essentially as described (10). Transfer activity was defined as the percent of labeled lipid substrate transferred from donor lipoprotein to acceptor lipoprotein mediated by a specific protein or protein complex present in cell culture media or in human plasma.
Maintaining assay conditions such that activity was assayed in the linear range, transfer activity was expressed as units/volume/mg of cell protein, where U/V= 100 kt/Vand t = 12 h (17). All calculations took into account the small amount of spontaneous transfer by subtractingendogenous transfer activity in matched negative controls from total transfer observed in experimental samples.
Western Blot Annlysis of CaCo-2 and HepG2 Lipid Transfer Protein-Goat anti-human LTP-I (anti-LTP, developed against homogenous human LTP-I exhibiting amino acid sequence identity with cholesteryl ester transfer protein (11)). purified by immunoaffnity as described (lo), was used to probe for LTP-I in CaCo-2-and HepG2conditioned media, by Western blot analysis according to the method of Towbin et al. (18).

CaCo-2 and HepG2 Secretion of CE Transfer Activity-
CaCo-2 cells cultured on PCM to approximately 14 days postconfluency secreted cholesteryl ester transfer activity, with 8.7 f 2.2 total units/h/mg of cell protein CE transfer activity secreted over a 24-h period during serum-free incubation (mean +-S.D., N = 10, range = 5.8 -12.5 units/h/ mg). This is approximately twice the rate we observed for secretion of CE transfer activity by HepG2 hepatocytes (10). As with HepG2 secretion of CE transfer activity, CaCo-2 cells secreted CE transfer activity in a linear, time-dependent manner for at least 24 h (Fig. 1). In cell-conditioned medium from CaCo-2 cultured on membranes, all CE transfer activity was detected in medium of the basolateral compartment of the PCM apparatus. Thus, CaCo-2 transfer CE transfer activity was vectorially sorted to, and secreted from, the baso-  lateral cellular domain, as are intestinal apolipoproteins secreted by this cell (12). Molecular Identity of CaCo-2 CE Transfer Protein-That LTP-I protein was responsible for CE transfer activity detected in cell-conditioned medium was suggested by immunoinhibition of 100% of CaCo-2-and HepG2-secreted CE transfer activity by incubation with specific anti-LTP-I (Fig.  2); immunoinhibition followed a dose-dependent pattern. Thus, the only source of CE transfer activity detected in these samples was LTP-I. LTP-I immunoreactive protein was not detected in CaCo-2 apical conditioned medium, providing further evidence that CaCo-2 cells vectorially sort and secrete LTP-I from the basolateral cellular domain. Furthermore, immunoblot analysis of secreted proteins revealed a single band of LTP-I immunoreactive material in both CaCo-2-and HepG2-conditioned media, with an apparent M, = 63,000 (Fig. 2, inset). This is consistent with the size of LTP-I purified to homogeneity from human plasma (l), and with that reported for LTP-I visualized by monoclonal antibody immunoblot (20). The apparent molecular weight and broad nature of the CaCo-2 and HepG2 LTP-I bands suggests that LTP-I is secreted from these cells in its glycosylated form (21), as LTP-I cDNA encodes a M, = 54,000 protein which contains four potential sites for N-linked glycosylation (11). Preliminary data from solution-hybridization of CaCo-2 RNA using a portion of the LTP-I cDNA (11) indicates that CaCo-2 do indeed transcribe mRNA for LTP-I, corroborating immunoinhibition and Western blot data that the molecular identity of CaCo-2 lipid transfer protein is LTP-I (not shown).
Regulation of CaCo-2 LTP-I Secretion-CaCo-2 cells cultured on permeable membranes upregulated secretion of CE transfer activity in response to increasing concentrations of oleate in the apical PCM culture compartment (Fig. 3); LTP-I activity was detected only in the basolateral medium compartment, The effect appeared to saturate at approximately 250 p t~ sodium oleate under our culture conditions. Secretion of cholesteryl ester transfer activity by HepG2 hepatocytes was unaltered by similar concentrations of fatty acid (Fig. 3). Thus, although HepG2 hepatocytes were confirmed to constitutively synthesize and secrete LTP-I in uitro, HepG2 cells did not regulate LTP-I secretion in response to this lipid challenge.
Spontaneous (background) transfer of labeled lipid substrates under our assay conditions was 4 % of the experimental activity in all cases. In addition to the usual controls, DMEM were incubated with donor-acceptor lipoprotein substrate in the presence or absence of a known source of LTP-I (chromatographed from human plasma, eluted from phenyl-Sepharose with water as described (l), purified approximately 250-fold). This was necessary to determine the effect of these concentrations of sodium oleate on apparent transfer activity of a constant source of LTP-I, since it has been suggested that increasing fatty acid concentration may increase the apparent CE transfer activity of LTP-I (22). However, we observed no effect of sodium oleate on CE transfer activity manifest by a constant source of LTP-I under our experimental conditions.

DISCUSSION
The principal rationale for investigating the CaCo-2 enterocyte model for synthesis and secretion of the plasma cholesteryl ester transfer protein was the Northern blot demonstration of LTP-I mRNA in polyadenylated RNA extracted from human intestine (11). The present data suggest that the enterocyte is responsible for at least some of this intestine LTP-I mRNA. We have demonstrated CaCo-2 secretion, and confirmed HepG2 secretion, of the LTP-I protein by: 1) secreted cholesteryl ester transfer activity, 2) anti-LTP-I immunoinhibition of 100% of secreted CE transfer activity, and 3) immunoblot technique revealing a single reactive protein, confirming the molecular identity of the protein responsible for all CE transfer activity secreted by CaCo-2 and HepG2 cells. We have previously demonstrated immunoinhibition of 100% of HepG2-secreted Ce transfer activity (10). Curiously, this is in contrast to a subsequent report of 20% residual CE transfer activity following immunoprecipitation of HepG2-secreted LTP-I, which suggests secretion of more than one cholesteryl transfer protein by HepG2 (21). However, our present findings are consistent with our earlier demonstration that a single protein is responsible for the CE transfer activity detected in human plasma (1) and for CE transfer activity secreted by HepG2 cells (10).
The principal cellular and tissue sources of human plasma LTP-I remain a matter of speculation. Cholesteryl ester transfer activity has been detected in concentrated rabbit liver perfusate, suggesting that liver is a source of plasma LTP-I (23). Furthermore, synthesis and secretion of LTP-I by the human HepG2 hepatocyte model (10) suggests that the liver parenchymal cell may be responsible for cholesteryl ester transfer activity detected in liver perfusate, although the hepatic Kupffer macrophage may contribute as well (9). Enterocyte synthesis of LTP-I is suggested by the presence of LTP-I mRNA in intestinal extract and confirmed by the present data. However, the relative hepatic and intestinal contributions to the total plasma LTP-I pool is unknown.
Data on the physiological correlates of CE transfer activity in vivo suggest that LTP-I protein mass is upregulated in response to both acute (24) and chronic (25) lipemia. As the immediate site of lipid absorption, the intestinal enterocyte is suspect as a source and regulatory site for proteins involved in lipid metabolism (26), including LTP-I. In uiuo, dietary triglycerides are lipolyzed to free fatty acids and monoglycerides in the intestinal lumen. These readily enter the enterocyte epithelium, where they are used to resynthesize triglycerides to be packaged and secreted with intestinal apolipoproteins. The present data are consistent with tissue-specific upregulation of LTP-I secretion by intestinal epithelium in response to fatty acid flux. Thus, dietary fat intake may acutely upregulate LTP-I levels in order to meet increased demands for lipid transport in postprandial plasma. In this context, it is noteworthy that LTP-I, the cholesteryl ester transfer protein, also mediates the net mass transfer of triglycerides and phospholipids between the plasma lipoproteins. Thus, plasma and tissue lipid homeostasis during alimentary lipemia may be maintained by upregulating LTP-I, as demonstrated here, and perhaps coordinately by upregulating 1ecithin:cholesterol acyltransferase activity, as suggested elsewhere (27).
It has been difficult to assess in vivo the question of whether the correlation of LTP-I-mediated CE transfer activity with plasma lipemia is causal or consequential. The data presented here suggest that acute lipid challenge causes upregulation of LTP-I secretion by enterocytes but not hepatocytes. Based on these observations, we anticipate in vivo data to corroborate the intestine as a principal regulated source of plasma LTP-I in humans.
The specific mechanism by which free fatty acids may upregulate protein secretion is unknown. However, sterols have been shown to control levels of protein mass and mRNA for other proteins which regulate cell cholesterol levels (28-30). A similar mechanism may be operating in the present case, perhaps indirectly through an intracellular sterol carrier or fatty acid binding protein (31). The present data do not speak to the question of whether the observed regulation occurs at transcriptional, translational, or secretory levels. We will need to collate solution-hybridization quantitation of specific LTP-I mRNA with quantitation of secreted LTP-I lipid transfer activity and mass in order to address this question.
The precise physiological roles and clinical significance of LTP-I remain uncertain. The hypothesized role of LTP-I in reverse cholesterol transport (3), and reports of LTP-I-mediated CE transfer from peripheral cells (6, 7) and to liver cells (8), support a preventive role of LTP-I in arterial lipid deposition. Alternatively, transfer of additional CE to the atherogenic apo-B-containing lipoproteins could potentially exacerbate an atherogenic condition. Moreover, growing evidence of synthesis and secretion by many cell types suggests that LTP-I may play a more pervasive role in lipid metabolism than previously suspected. The advent of our present, and previous (9, lo), models of cellular LTP-I production should help us understand the regulation of LTP-I synthesis and secretion, as well as its role in lipid metabolism. Finally, the present data lend further validity to CaCo-2 cells cultured on membranes as a model of enterocyte epithelium.