Bile Salt-stimulated Carboxyl Ester Lipase Influences Lipoprotein Assembly and Secretion in Intestine

Bile salt-stimulated carboxyl ester lipase (CEL), also called cholesterol esterase, is one of the major proteins secreted by the pancreas. The physiological role of CEL was originally thought to be its mediation of dietary cholesterol absorption. However, recent studies showed no difference between wild type and CEL knockout mice in the total amount of cholesterol absorbed in a single meal. The current study tests the hypothesis that CEL in the intestinal lumen may influence the type of lipoproteins produced. A lipid emulsion containing 4 mm phospholipid, 13.33 mm[3H]triolein, and 2.6 mm[14C]cholesterol in 19 mm taurocholate was infused into the duodenum of lymph fistula CEL(+/+) and CEL(−/−) mice at a rate of 0.3 ml/h. Results showed no difference between CEL(+/+) and CEL(−/−) mice in the rate of cholesterol and triglyceride transport from the intestinal lumen to the lymph. However, CEL(−/−) mice produced predominantly smaller lipoproteins, whereas the CEL(+/+) mice produced primarily large chylomicrons and very low density lipoprotein. The proximal intestine of CEL(−/−) mice was also found to possess significantly less ceramide hydrolytic activity than that present in CEL(+/+) mice. By using Caco2 cells grown on Transwell membranes as a model, sphingomyelinase treatment inhibited the secretion of larger chylomicron-like lipoproteins without affecting total cholesterol secretion. In contrast, the addition of CEL to the apical medium increased the amount of large lipoproteins produced and alleviated the inhibition induced by sphingomyelinase. Taken together, this study identified a novel and physiologically significant role for CEL, namely the promotion of large chylomicron production in the intestine. The mechanism appears to be mediated through CEL hydrolysis of ceramide generated during the lipid absorption process.

sized in the acinar cells of the pancreas and is stored in zymogen granules. Upon food ingestion, CEL is released into the intestinal lumen where it constitutes 1-5% of total protein in pancreatic juice (3,4). The same enzyme is also present as a major protein in milk and as a minor constituent in liver, activated macrophages, and endothelial cells (5)(6)(7)(8)(9).
The abundance of CEL in pancreatic juice and in milk of a number of mammals led to the early speculation of its role in dietary lipid absorption (10 -12). However, the precise role of CEL in dietary lipid absorption remains controversial despite over 20 years of investigations. For example, one study showed that infusion of pancreatic juice containing CEL, but not juice devoid of CEL by immunoprecipitation, restored normal cholesterol absorption in pancreatectomized rats (10,13). On the other hand, another study showed that cholesterol absorption was not affected by pancreatic diversion (14). In vitro tissue culture studies also failed to resolve this issue, with experiments showing that the inclusion of CEL can either facilitate or have no effect on cholesterol transport by the enterocyte-like Caco-2 cells (12,15,16). Although recent studies with CEL gene-targeted mice demonstrated that CEL deficiency did not alter the total amount of cholesterol absorbed from a bolus meal over a 24-h period (17,18), whether CEL has any influence on the rate of intestinal cholesterol absorption and/or the type of lipoproteins produced in the intestine remains unknown.
One difficulty in assigning specific role(s) for each protein in lipid absorption is the complexity of the process. After lipid digestion and solubilization with bile salt micelles in the intestinal lumen (19 -22), lipid nutrients are absorbed by enterocytes, resynthesized, and assembled into lipoproteins (a process that entails shuttling through several intracellular compartments) prior to their secretion into the lymph. The first step of this process is microsomal triglyceride transfer proteinmediated lipidation of apoB in the lumen of the smooth endoplasmic reticulum (23,24). The lipid-poor apoB particles fuse with apoB-free triacylglycerol-rich particles in the rough endoplasmic reticulum (23,24), and these partially matured lipoproteins are then delivered to the Golgi for final processing prior to secretion into the lymphatics (25).
Increasing evidence shows that the transport of pre-chylomicrons from the endoplasmic reticulum to the Golgi is a vesicular process mediated by pre-chylomicron transport vesicles (26). Recent studies (27,28) have shown that naturally occurring long chain ceramides, such as those generated from sphingomyelin hydrolysis, are promoters of Golgi disassembly and are capable of disrupting protein trafficking through intracellular secretory pathways. In view of observations that CEL possesses lipoamidase activity (29) and is capable of hydrolyzing ceramide (30), this study was undertaken to test the possibility that CEL may influence intestinal lipoprotein assembly and transport. Bacterial sphingomyelinase (SMase) and C 6 -ceramide were from Matreya, Inc. (Pleasant Gap, PA). Human CEL was purified from human milk (generous gift from Dr. Ron Jandecek, Procter & Gamble, Inc., Cincinnati, OH) using the same procedure as described previously for the purification of rat pancreatic CEL (7). Formvar-coated grids used for electron microscopy were obtained from Electron Microscopy Sciences (Fort Washington, PA). Free and total cholesterol analysis kits were purchased from Wako Chemicals (Richmond, VA). The human colonic adenocarcinoma cell line Caco2 cells was obtained from the American Type Culture Collection (Manassas, VA; ATCC HTB 37). Costar Transwells (6-well; 3-m pore) were obtained from Fisher. Dulbecco's modified Eagle's medium with 4.5 g/liter glucose and fetal bovine serum were purchased from Invitrogen.

Materials
Lymphatic Lipid Transport-Male C57BL/6 mice were purchased from Jackson Laboratories (Bar Harbor, ME). The CEL(Ϫ/Ϫ) mice were produced and bred in our institutional animal facility. The characteristics of these mice have been reported previously (17). Under halothane anesthesia, the major intestinal lymph duct of CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice was cannulated superior to the superior mesenteric artery as described (31,32). The lymph cannula was primed with a heparin sodium solution (1,000 units/ml) to prevent clotting. A silicone tube was passed through the fundus of the stomach and extended into the duodenum. The fundal incision was closed using a purse-string suture. Postoperatively, the animals were infused with a 5% dextrose saline solution (145 mM NaCl, 4 mM KCl, 0.28 mM dextrose) at a constant rate of 0.3 ml/h. The animals were maintained overnight at 30°C before lipid infusion. On the day of experiments, mice were infused with an emulsion that consisted of 4 mM egg phosphatidylcholine, 13.33 mM [ 3 H]triolein, and 2.6 mM [ 14 C]cholesterol in 19 mM taurocholate at a rate of 0.3 ml/h. Lymph was collected for 1 h prior to lipid infusion and served as the fasting lymph. After beginning lipid infusion, lymph was collected hourly for 6 h. Appearance of radiolabel in lymph was determined by scintillation counting.
Lipoprotein Size Determination-Lipoproteins from mesenteric lymph were adsorbed onto 300-mesh Formvar-coated grids, air-dried, and stained either with 1% phosphotungstic acid, pH 6.9, for 30 s or dual stained with 4% osmium tetroxide and 1% phosphotungstic acid in 0.1% sucrose as described (33). Electron micrographs were taken of random areas of several grids at a magnification of ϫ40,000 using a Hitachi H-600 transmission electron microscope (Hitachi Ltd., Tokyo, Japan). The diameters of 400 particles were measured from representative photographs to determine particle sizes.
Ceramide Hydrolytic Activity-Substrate for measuring ceramide hydrolytic activity was prepared by dissolving 0.5 mol of unlabeled C-6 ceramide and 1 Ci of [ 3 H]C 6 -ceramide in ethanol and then dried under nitrogen. The sample was resuspended in 10 ml of buffer containing 20 mM Tris-HCl, pH 8.5, 4 mM taurocholate and then sonicated for 10 min at 4°C. Ceramidase assay was commenced by adding 0.1 ml of purified CEL solution in phosphate-buffered saline to 0.5 ml of the ceramide substrate. Incubation was continued at 37°C for 1 h. Reaction was terminated by addition of 1 ml of 50 mM sodium borate, 50 mM sodium carbonate, pH 10.0. Fatty acids were extracted by addition of 3 ml of methanol/chloroform/heptane (1.41:1.25:1.0, v/v/v) and centrifugation for 30 min at 5,000 ϫ g. A 0.75-ml aliquot of the top aqueous phase was added to 15 ml of scintillation fluid for radioactivity determination in a scintillation counter. Extraction efficiency was determined empirically to be 80%, which was corrected in the report of results.
For the determination of ceramide hydrolytic activity in small intestine, the tissue was removed from CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice immediately after their sacrifice. The intestine was flushed with cold saline and then sectioned into four equal fractions proximal to distal. The mucosa was scraped using glass microscope slides and suspended in 5 ml of buffer containing 300 mM mannitol, 1 mM EGTA, 2.4 mM Tris-HCl, pH 7.1. Particulate fraction containing brush border membranes was prepared by a modification of Kessler's divalent cation precipitation method (34). Briefly, 20 ml of H 2 O was added to the mucosal scrapings for homogenization with a Dounce homogenizer. A 0.25-ml aliquot of 1 M MgCl 2 was added to each homogenate, and the sample was incubated on ice for 15 min prior to centrifugation at 3000 ϫ g for 15 min. The resulting supernatant was centrifuged at 27,000 ϫ g for 30 min. Pellet containing cellular membrane proteins, including brush border membranes, was resuspended in 60 mM mannitol, 5 mM EGTA, 12 mM Tris-HCl, pH 7.1. Protein content was determined by the Lowry procedure (35), and 50 g of membrane proteins were used for ceramidase activity determinations.
Lipid substrates containing 50 M cholesterol, 30 M monoolein, 1.6 mM oleic acid, and 1 mM taurocholate were prepared by drying contents under N 2 and dissolving in serum-free Dulbecco's modified Eagle's medium by sonication. Trace amounts of [ 14 C]cholesterol were added to the micelles to monitor cholesterol transport. SMase and CEL, in phosphate-buffered saline, were added to micelle preparations at final concentrations of 100 milliunits/ml and 10 g/ml, respectively. On the day of the experiments, the Caco2 cells were washed twice with phosphatebuffered saline. 2 ml of fresh serum-free Dulbecco's modified Eagle's medium were added to the basolateral surface, and 2 ml of the micellar preparations were added to the apical surface. Incubations were continued at 37°C for 24 h. At the end of the incubation period, the basolateral medium was collected. An aliquot of the medium was used for radioactivity determinations to measure cholesterol secretion. Another aliquot of the basolateral medium was subjected to density gradient ultracentrifugation for lipoprotein separation.
Differential Gradient Ultracentrifugation-Basolateral medium was collected and brought to 4 ml with serum-free Dulbecco's modified Eagle's medium. Solid KBr was added to adjust the density to 1.10 g/ml. The sample was overlaid with 3 ml each of 1.063 and 1.019 g/ml, and 2 ml of 1.006 g/ml KBr density solutions and then centrifuged for 33 min at 40,000 rpm in a Beckman SW41Ti rotor at 15°C. Large chylomicrons (with flotation rate of S f Ͼ 400) were obtained from the top 1-ml fraction after centrifugation.

Lipid Absorption Rate and Lymph Lipoprotein Particle Size of CEL(ϩ/ϩ) and CEL(Ϫ /Ϫ) Mice-
The role of CEL in intestinal lipid transport was explored by comparing the rate of lymphatic absorption of radiolabeled lipids that were infused into the duodenum in lymph fistula CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice. An emulsion containing 4 mM phospholipid, 13.33 mM [ 3 H]triolein, and 2.6 mM [ 14 C]cholesterol in 19 mM taurocholate was infused into each animal at a rate of 0.3 ml/h. Lymph was collected hourly to monitor radiolabeled lipid output by the intestine. No difference in the absorption rate of infused cholesterol from the intestinal lumen to the mesenteric lymph was observed between CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice (Fig. 1A). Radiolabeled fatty acids derived from [ 3 H]triolein were also secreted into the lymph at a similar rate for both CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice (Fig. 1B). However, when lymph from these animals was analyzed by negative staining electron microscopy, a significant difference in the size of lipoprotein particles was observed (Fig. 2). The CEL(ϩ/ϩ) mice produced lipoproteins with sizes ranging from 40 to 260 nm (Fig. 2, A and  C). Approximately 55% of the lymph lipoproteins displayed sizes Ͼ80 nm indicating a predominance of chylomicron production in the wild type mice. In contrast, lymph lipoproteins in CEL(Ϫ/Ϫ) mice were much smaller, ranging in size from 20 to 140 nm (Fig. 2, B and D). Less than 25% of the lymph lipoproteins in CEL(Ϫ/Ϫ) mice were the size of large chylomicrons (Ͼ80 nm). Most of the lymph lipoproteins (75%) in CEL(Ϫ/Ϫ) mice were 30 -80 nm in diameter, suggesting that the intestines of CEL(Ϫ/Ϫ) mice secrete mostly VLDL sized lipoproteins. Analysis of the cholesterol content in the lymph of these animals revealed a higher percentage of the cholesterol was esterified in CEL(Ϫ/Ϫ) mice in comparison with that observed in CEL(ϩ/ϩ) mice (66 versus 45%, respectively). Thus, the difference in lipoprotein particle size observed in CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice cannot be attributed to the cholesteryl ester hydrolytic activity of CEL.
Ceramidase Activity of CEL-Previous studies (36, 37) with cultured intestinal cells suggested that intestinal lipoprotein secretion may be controlled by ceramide release after membrane sphingomyelin hydrolysis. In view of our previous observation (29) that CEL also displays lipoamidase activity, experiments were performed to determine whether CEL influences ceramide metabolism in intestine. An initial study (30) was performed to confirm the ceramide hydrolytic activity of CEL. Incubation of C 6 -ceramide with CEL resulted in a CEL concentration-dependent release of free fatty acids from ceramide (Fig. 3). We then compared ceramide hydrolytic activity in the intestine of CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice. Membrane proteins prepared from the proximal intestinal mucosa of CEL(ϩ/ϩ) mice were found to contain approximately twice the ceramide hydrolytic activity as that present in similar preparations from CEL(Ϫ/Ϫ) mice (Fig. 4). In contrast, ceramide hydrolytic activity in membrane preparations from the mid-to distal intestinal fractions was similar between CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice (Fig. 4). Because intestinal lipid absorption occurs primarily in the proximal intestine (38), these results support the hypothesis that the CEL influence on intestinal lipoprotein assembly and transport may be related to the ceramidase activity of this protein. the presence or absence of bacterial SMase with or without CEL. Transport was monitored by the appearance of radiolabel in the basolateral media. Secreted lipoproteins were then fractionated by differential gradient ultracenrifugation to separate chylomicrons from higher density particles. Consistent with results reported by Luchoomun and Hussain (39), the incubation of differentiated Caco-2 cells with oleic acid-supplemented medium resulted in the secretion of large chylomicron-sized lipoproteins. Incubation of Caco-2 cells with the micellar substrate in the presence or absence of CEL had no effect on the total amount of cholesterol secreted into the basolateral media (Fig. 5A). However, the amount of micellar derived [ 14 C]cholesterol secreted into the basolateral media as the low density large chylomicron fractions was found to be increased in the incubations containing CEL (Fig. 5B). In contrast, generation of endogenous ceramide in Caco2 cells by incubation with bacterial SMase decreased the amount of micelle-derived [ 14 C]cho-lesterol found in the larger chylomicron-like particles (Fig.  5B) without significant effect on the total amount of [ 14 C]cholesterol secreted into the basolateral media (Fig. 5A). Interestingly, the addition of CEL not only alleviated the SMase inhibition of large chylomicron secretion, but the amount of [ 14 C]cholesterol in the chylomicron fraction was also higher in incubations containing both CEL and SMase than in control cells incubated without these proteins or with only CEL (Fig. 5B). DISCUSSION Lipid absorption occurs predominantly in the proximal part of the intestine (38) and continues into the distal intestine after a high fat load (40). Interestingly, CEL secreted by the pancreas is also present in absorptive cells throughout the small intestine and is especially prominent in the proximal gut (41). Although this CEL localization suggests its possible involvement in dietary lipid absorption and transport, its precise role in the lipid absorption process has not been defined. This study oleic acid, 1 mM taurocholate) were added to the apical compartment, and the incubation was continued for 24 h at 37°C. When included, SMase and CEL in phosphate-buffered saline were added at a concentration of 100 milliunits/ml and 10 g/ml, respectively. An equivalent volume of buffer without enzyme was added to the control wells. At the end of the incubation period, an aliquot of the basolateral media was used for radioactivity measurement to determine the total amount of [ 3 H]cholesterol secreted by the Caco-2 cells (A). The remaining portion of the basolateral media was subjected to differential gradient ultracentrifugation to isolate chylomicron particles (S f Ͼ 400) (B). The reported data represent the mean Ϯ S.D. from three different experiments. Bars with different letters are significantly different from each other at p Ͻ 0.05. identified a novel physiologic role of CEL in dictating the type of lipoproteins produced by the intestine. We used CEL-null mice to demonstrate that the lack of a functional CEL gene resulted in diminished production of large sized chylomicrons and the concomitant increase in the amount of smaller VLDL sized intestinal lipoproteins after lipid infusion in vivo. The lack of CEL in the CEL(Ϫ/Ϫ) mice may either directly result in defective lipid processing in the intestine or alternatively delaying lipid absorption to the distal portion of the intestine where the smaller sized VLDL may be produced. Although we cannot distinguish these possibilities in the current study, previous studies indicated that lipoproteins secreted by distal intestine were either larger (42) or the same size (43) as chylomicrons secreted by the proximal intestine. Therefore, it seems likely that CEL directly influences lipoprotein assembly and secretion in the proximal intestine and dictates the type of lipoproteins being produced. The in vitro cell culture experiments showing that addition of CEL increases the production of large sized chylomicron-like particles by Caco-2 cells are supportive of this hypothesis.

Effect of Ceramide Accumulation and CEL on Lipid Transport in Caco
Our current study also showed reduced ceramide hydrolytic activity in the proximal intestine of CEL(Ϫ/Ϫ) mice. This observation suggests that CEL may promote chylomicron production through ceramide hydrolysis. The importance of sphingomyelin and its metabolic product ceramide in controlling lipid trafficking in mammalian cells is well documented in the literature. Previously, Field and colleagues (44,45) showed that cholesterol uptake by enterocytes is a two-step process in which the first step requires micellar cholesterol from the apical side embedding into the plasma membrane of the enterocytes. The membrane-bound cholesterol is then transported to endoplasmic reticulum where lipoprotein assembly occurs (44,45). The amount of cholesterol that can be accommodated in plasma membrane is strongly correlated with its sphingomyelin content, and influx of membrane cholesterol into the cell interior requires hydrolysis of the membrane sphingomyelin (46,47). However, cell culture studies showed that the rate of exogenous cholesterol uptake by intestinal cells decreased when SMase was added to the culture media (37). Although SMase also decreased the amount of triacylglycerol secreted to the basolateral media, the amount of cholesterol and phospholipid secreted into the basolateral media was not affected by SMase treatment (36,37). Analysis of the lipoprotein composition data reported in these studies revealed that SMase may affect the secretion of larger sized chylomicrons with minimal effect on the secretion of smaller size VLDL. The SMase-induced alteration in intestinal cholesterol transport was attributed to the ability of the digestion product, ceramide, to inhibit basolateral secretion of intestinal lipoproteins (36). Although the exact mechanism by which ceramide inhibits chylomicron production is unknown at the present time, it is likely that the excess ceramide generated during cholesterol transport promotes Golgi disassembly and interrupts the transport of pre-chylomicron transport vesicles to this organelle.
The previous in vitro cell culture experiments cited above were conducted in the absence of other pancreatic enzymes. In a physiological setting, pancreatic enzymes including CEL are secreted into the pancreatic juice and are present in the intestinal lumen. The CEL can interact with heparin-like molecules on the surface of enterocytes (12) and be taken up into the cell interior (15, 48 -50). The presence of CEL may alleviate ceramide inhibition of intracellular lipid trafficking and promote chylomicron production. Thus, SMase hydrolysis of membrane sphingomyelin, along with CEL hydrolysis of ceramide, are two important processes in the intracellular lipid transport process that are required for the production of large chylomicrons in the intestine. The observed synergism between SMase and CEL in promoting large chylomicron secretion by Caco-2 cells in culture is supportive of this hypothesis. A schematic diagram showing the participation of both SMase and CEL in intestinal lipid transport and lipoprotein assembly is presented in Fig. 6. We hypothesize that ceramide generated as a result of sphingomyelin hydrolysis, by either luminal or cytoplasmic SMase (37, 51), can be further hydrolyzed by CEL in the lumen or endocytosed into the cell interior (15, 48 -50). The hydrolysis of this SMase-generated metabolic product is necessary for FIG. 6. Schematic diagram depicting the proposed participation of SMase and CEL in cholesterol transport in enterocytes. This diagram shows that the initial step of cholesterol absorption is the intercalation of the micellar cholesterol from the lumen to the apical membrane of enterocytes. The second step of cholesterol transport from the plasma membrane to the cell interior requires the hydrolysis of the sphingomyelin by SMase present in the intestinal lumen or in the cell interior. Ceramide generated as a result of sphingomyelin hydrolysis can then be hydrolyzed by CEL present in either the lumen (or membrane-bound) or CEL endocytosed into the cells. Lack of ceramide hydrolysis will result in blockage of large lipoprotein assembly and secretion by the intestinal cells.
proper lipid trafficking through the Golgi and the assembly and secretion of large sized chylomicrons.
It is important to note that ceramide hydrolytic activity in CEL(Ϫ/Ϫ) mice decreased by ϳ50% only in the proximal intestine, with little or no change in the distal gut. These results, which are consistent with results reported by others (52), indicated the presence of additional enzyme(s) in the intestine capable of ceramide hydrolysis. However, the intestinal ceramidase is apparently unable to compensate for the lack of CEL in promoting chylomicron assembly and secretion in the CEL(Ϫ/Ϫ) mice. It is possible that differences in anatomic location of the two enzymes in the intestine (52) may account for the inability of the ceramidase to replace CEL in mediating large chylomicron production. It is also possible that the remaining ceramidase activity in the proximal intestine of CEL(Ϫ/Ϫ) mice is insufficient to support the assembly and secretion of large chylomicrons after lipid infusion. The differentiation of these two possibilities will require additional experimentation with the generation of tissue-specific ceramidase overexpression transgenic mice.
The participation of CEL in chylomicron production may be of clinical importance in determining the relationship between dietary lipid transport and risk of atherosclerosis and obesity. Previous studies using in vitro generated lipoproteins with sizes comparable with large chylomicrons (S f Ͼ400) and the smaller VLDL (S f ϭ 20 ϭ 400) showed that the chylomicrons are cleared from circulation by the liver more rapidly than the smaller sized VLDL (53)(54)(55)(56). The delayed hepatic clearance of smaller postprandial lipoproteins may promote obesity due to increase transport of the dietary fat to adipose tissues. The smaller sized VLDL are also more likely to penetrate the arterial wall. The VLDL trapped within the subendothelial space may promote atherogenesis (57). Thus, CEL in the gastrointestinal tract may protect against these adverse effects by promoting the formation of large chylomicrons in response to fat feeding. Additional studies comparing diet-induced obesity and atherosclerosis in CEL(ϩ/ϩ) and CEL(Ϫ/Ϫ) mice are warranted.