Dietary free and esterified cholesterol absorption in cholesterol esterase (bile salt-stimulated lipase) gene-targeted mice.

The involvement of pancreatic cholesterol esterase (bile salt-stimulated lipase) in cholesterol absorption through the intestine has been controversial. We have addressed this issue by using homologous recombination in embryonic stem cells to produce mice lacking a functional cholesterol esterase gene. Cholesterol esterase knockout mice and their wild type counterparts were fed a bolus dose of [3H]cholesterol and a trace amount of [beta-14C]sitosterol by gavage. The ratio of the two radiolabels excreted in the feces over a 24-h period was found to be similar in the control and cholesterol esterase-null mice. Similar results were observed when the radiolabeled sterols were supplied in an emulsion with phospholipid and triolein or in lipid vesicles with phosphatidylcholine. Cholesterol absorption results were similar between the control and cholesterol esterase-null mice regardless of whether the animals were fed a low fat diet or a high fat/high cholesterol diet. The rate of [3H]cholesterol appearance in the serum of the gene-targeted mice paralleled that observed in control animals. In contrast to these results, when experiments were performed with [3H]cholesteryl oleate instead of [3H]cholesterol, a higher amount of the 3H radiolabel was found excreted in feces and dramatically less of the radiolabel was detected in the serum of the cholesterol esterase-null mice in comparison with that detected in control animals. Serum cholesterol levels were not significantly different between control and cholesterol esterase-null mice fed either control or an atherogenic diet. These results indicate that cholesterol esterase is responsible for mediating intestinal absorption of cholesteryl esters but does not play a primary role in free cholesterol absorption.

Cholesterol esterase, also called bile salt-stimulated lipase or carboxyl ester lipase (abbreviated as CEL), 1 is a lipolytic enzyme capable of hydrolyzing triacylglycerol, phospholipid, lysophospholipid, and cholesteryl esters. The enzyme is synthesized in the acinar cells of the pancreas and is stored in zymogen granules. Cholesterol esterase is released into the intestinal lumen upon food ingestion and constitutes 1-2% of total protein in pancreatic juice (1).
While the high concentration of CEL in pancreatic juice suggests that it may play a role in mediating nutrient absorption, the precise physiologic function of the enzyme remains controversial. Early studies with isolated intestinal cells suggested a role for CEL in dietary cholesterol absorption (2). However, subsequent studies yielded contradictory results. For example, using pancreatic diverted rats, Watt and Simmonds (3) showed normal absorption and esterification of cholesterol. In contrast, using the same experimental system, Gallo et al. (4) showed an 80% reduction in cholesterol absorption, which could be restored by infusion of pancreatic juice containing CEL but not by juice depleted of the enzyme.
Cholesterol absorption has also been studied using a variety of inhibitors. Bennett Clark and Tercyak (5) demonstrated a reduction in cholesterol transmucosal transport in rats with inhibited acyl CoA:cholesterol acyltransferase and normal pancreatic function, which suggested that acyl CoA:cholesterol acyltransferase, and not CEL, was responsible for this process. However, using similar inhibitors, Gallo et al. (6) showed no inhibition of cholesterol absorption, which again suggested the involvement of CEL. In later studies, CEL inhibitors, such as the phenoxyphenyl carbamates WAY-121,751 and WAY-121,898, were shown to be effective inhibitors of cholesterol absorption in normal and cholesterol-fed rats and dogs (7). Thus, whole animal studies have not consistently shown the importance of CEL in cholesterol absorption.
The possible role of CEL in mediating intestinal absorption of cholesterol has also been investigated in vitro without resolution. Bhat and Brockman (8) showed that incubation of rat intestinal sacs with cholesterol-containing micelles in the presence of CEL resulted in a 3-5-fold enhancement of intracellular cholesterol and cholesteryl ester accumulation compared with intestinal sacs incubated in the absence of the enzyme. More recently, Lange and colleagues, using Caco-2 cells as a model for intestinal epithelium, showed that CEL addition was necessary for the transfer of exogenous cholesterol to a "physiologically important pool" that could be esterified and assembled into lipoproteins (9). In contrast to these results, our laboratory could not demonstrate CEL-mediated uptake of unesterified cholesterol by Caco-2 cells (10). Our in vitro data were confirmed and extended in a recent publication by Fisher and colleagues (11). Both laboratories reported that the enzyme was only effective in facilitating cellular uptake of esterified cholesterol.
In an attempt to resolve this controversy, we have used the approach of gene targeting in embryonic stem (ES) cells to produce mice lacking in CEL. The CEL(Ϫ/Ϫ) mice provide a unique in vivo model to assess the physiological function of the bile salt-stimulated cholesterol esterase.

EXPERIMENTAL PROCEDURES
Cloning of the Mouse Cholesterol Esterase Gene and Production of the Targeting Vector-A strain 129 mouse genomic library made in -DASH phage vector was obtained from Dr. Thomas Doetschman at our institution and used to isolate the mouse CEL gene. The 2-kb full-length rat cholesterol esterase cDNA (12) was used as the probe for screening the library. A positive clone that also hybridized with probes corresponding to both the 5Ј-flanking region and the 3Ј-flanking region of the rat CEL gene was selected for further characterization. Restriction mapping, Southern hybridization with various cholesterol esterase cDNA fragments, and partial nucleotide sequencing were performed to determine the intron and exon locations of the mouse CEL gene.
A 4.7-kb SacI DNA fragment, encoding sequences from 540 bp upstream of exon 1 to intron 7 of the mouse cholesterol esterase gene, was subcloned into a similarly-digested PTZ18U plasmid. A 1.75-kb fragment containing a thymidine kinase promoter-driven neomycin resistance gene (neo R ) was isolated from SspI/HincII-digested pMC1Neo (Stratagene) and subcloned into the unique BalI site in exon 4 of the 4.7-kb SacI clone (Fig. 1). A plasmid containing neo R inserted in the same orientation as the CEL gene was selected for the gene-targeting experiment. After CsCl purification, the targeting vector was digested with SacI, and the 6.5-kb DNA fragment containing the disrupted CEL gene sequence was purified by agarose gel electrophoresis.
Targeted Disruption of the Cholesterol Esterase Gene in ES Cells-Gene targeting experiments were performed using the R1 ES cell line derived from the 129 mouse strain by Nagy et al. (13). Cells were grown and passaged as described (14). When ES cells were grown at high densities, the medium was supplemented with leukemia inhibitory factor (LIF) at ϳ500 units/ml (15)(16)(17). On the day of the experiment, 5.5 ϫ 10 7 ES cells in 0.5 ml of culture medium were electroporated in the presence of 3 pmol of the targeting DNA, using an IBI GeneZapper 450 set to deliver 200 microfarads at 800 V/cm. Surviving cells (ϳ50%) were cultured on G418-resistant feeders (a kind gift from Dr. Tom Doetschman of this institute) in selection medium containing G418 at 300 g/ml (added 24 h after electroporation) for 3 days and 250 g/ml thereafter. After 7 days, colonies resistant to G418 selection were picked and expanded individually in 24-well dishes. After 3 or 4 days of growth, approximately half of the cells in each colony were used to isolate DNA while the remaining cells were maintained for colony expansion and freezing.
Southern Blot Analysis-Colonies were screened for homologous recombination by Southern blot analysis. Genomic DNA was prepared from each colony (18), digested with various restriction enzymes according to the manufacturer's conditions, fractionated on 0.7-0.8% agarose gels, and transferred (19) to Nytran Plus membranes (Schleicher & Schuell, Inc.). Blots were prehybridized, hybridized, and washed according to the method of Church and Gilbert (20). Hybridization probes were purified from low melting agarose gels and labeled to a specific activity of ϳ1 ϫ 10 9 dpm/g using a random primer kit from Ambion, Inc.
Generation of Chimeric Mice-Mice with targeted disruption of the CEL gene were produced according to standard procedures using C57BL/6 blastocysts and (C57BL/6 ϫ C3H/HeN) F1 pseudopregnant females as foster mothers (21). Chimeric mice were identified by the degree of agouti coat color on the black background. Male chimeric mice were mated with female Black Swiss mice (Taconic Farms, Germantown, NY) to test for germ line transmission of the ES cell genome. Agouti pups carrying the modified allele were identified by polymerase chain reaction analysis of ear punch DNA. Heterozygotes from different parents were mated to obtain mice with homozygous disruption of the cholesterol esterase gene. Mice were maintained in a temperature and humidity-controlled room with a 12-h light/dark cycle and were allowed food (Teklad LM485 rodent diet) and water ad libitum. All animals of the F 2 generation were fed this same diet supplemented with retinol and tocopherol (Sigma) at a dose designed to deliver 10 units/day/mouse of retinol and 0.2 units/day/mouse of tocopherol. These unesterified forms of vitamins A and E were provided to compensate for any deficiencies in vitamin ester absorption resulting from the loss of CEL activity. Some animals were fed an atherogenic diet (TD 88051, Harlan Teklad), which was based on Purina Mouse Chow 5015 and also contained 7.5% cocoa butter (15.75% final fat content), 1.25% cholesterol, and 0.5% sodium cholate. This diet was also supplemented with unesterified vitamins. Animals were fed the atherogenic diet for at least 6 weeks before analysis.
Mouse Genotyping by Polymerase Chain Reaction-Ear punch tissue was incubated at 55°C in 25 l of digestion buffer (18) and 400 g/ml proteinase K for at least 4 h. After digestion, samples were briefly vortexed, the hair was allowed to settle, and 2 l of the suspension was diluted into 50 l of H 2 O, boiled for 5 min, and immediately placed on ice. One l of the boiled sample was used within 4 h for polymerase chain reaction analysis using the procedure described by Kim and Smithies (22). Primer concentrations were each 2.5 M, nucleotide triphosphates were each 200 M, and one unit of Taq polymerase (Life Technologies) was used per reaction. Reactions (30 l) were prepared on ice, preheated to 90°C for 1 min, and subjected to 33 cycles of amplification consisting of 30 s at 94°C and 2 min at 65°C. Samples were analyzed on 1.5% agarose gels.
The presence of neo R was determined using the primers described by Kim and Smithies (22), which amplify a 555-bp portion of this gene. The CEL gene was analyzed with these primers and an additional set of primers that amplifies exon 4. The upstream CEL primer, 5Ј-CCCTT-TCAGTGTCCCACAACCT-3Ј, and the downstream CEL primer, 5Ј-TCACTATTCCCGCTCTTACAGTC-3Ј, amplify a 244-bp fragment from the wild type exon 4 but do not amplify the targeted allele because of the insertion of neo R between their cognate sequences. Identical conditions were used for both primer sets but in separate reactions. A positive result with the exon 4 primers only was scored as wild type. Positive results with both sets indicated a heterozygote, and a positive result with the neo R primers only was scored as a homozygous knockout.
Cholesterol Ester Lipase Determination-Levels of CEL protein in the mouse pancreas were determined by immunoblot assay. The mice were euthanized, and the abdominal cavities were opened. The pancreas was removed and homogenized on ice in a solution containing 10 mM sodium phosphate, pH 6.2, 0.1 M NaCl, 1 mM EDTA, 0.02% sodium azide, 1.5% glycerol, and 0.02% soybean trypsin inhibitor (Sigma). The homogenate was centrifuged at 4°C for 1 h at 100,000 ϫ g to precipitate particulate fractions. Twenty-five g of the 100,000 ϫ g supernatant fraction from each sample was analyzed by electrophoresis on a 10% SDS-polyacrylamide gel (23). The electrophoresed samples were either stained with Coomassie Blue or transferred to nitrocellulose (24) for immunoblotting with affinity-purified rabbit anti-rat cholesterol esterase and 125 I-labeled anti-rabbit IgG (Amersham Corp.) as described (25). CEL activity was determined as described (26)  Cholesterol Absorption Studies-The single-dose, dual-isotope feeding method, originally described by Zilversmit and Hughes (27) and validated for measuring cholesterol absorption in mouse by Dueland et al. (28), was used to determine cholesterol absorption efficiency in control and CEL gene-targeted mice. Test meal was prepared either as a lipid emulsion or as sonicated vesicles. For absorption studies, mice were housed in metabolic cages where they had free access to food and water. Animals were allowed to adjust to the cages for at least 24 h before beginning the test. On the day of the experiment, mice were administered 50 l of the test meal by gavage approximately 3-4 h before the beginning of their dark cycle. Feces were collected for the following 24 h. The samples were homogenized in water and then extracted with an equal volume of chloroform/methanol (2:1, v/v). The aqueous phase was re-extracted once with chloroform. The organic phases from each sample were combined, their volumes were measured, and an aliquot was used for scintillation counting. Counting efficiency was calculated using the external standard, channel ratio method. Cholesterol absorption efficiency, determined as percentage of administered dose absorbed, was calculated based on the formula described by Grundy et al. (29) as follows: {1 -(( 3 H-dpm/ 14 Cdpm) in feces/( 3 H-dpm/ 14 C-dpm) administered)} ϫ 100. Total recovery of the [␤-14 C]sitosterol over the 24-h period ranged from 66 to 97%. Differences in cholesterol absorption between groups were evaluated for statistical significance by Mann-Whitney rank sum and Student's t tests using SigmaStat software from Jandel Corporation.
Serum Cholesterol Analysis-Blood was collected from euthanized animals by severing the inferior vena cava. Samples were allowed to clot on ice, and serum was collected after centrifugation. Total cholesterol levels were determined using commercially available enzymatic kits from Sigma.

Targeting of the Cholesterol Esterase Gene in Mouse Embryonic Stem Cells-Restriction mapping and
Southern blot analysis revealed that the mouse cholesterol esterase gene is highly homologous to the rat CEL gene and contains 11 exons interrupted by 10 introns (30). A 4.7-kb SacI fragment that contains exon 1 to intron 7 was used to construct the targeting DNA as described under "Experimental Procedures" and schematically outlined in Fig. 1B. The insertion of neo R into exon 4 of the CEL gene disrupts the CEL coding sequence immediately preceding the active site domains of cholesterol esterase (25,30). Additionally, any truncated polypeptide derived from the disrupted CEL gene could not be secreted by the pancreas due to the absence of the exon 11 domain, which is important for this function (31). Thus, homologous recombination of the endogenous CEL gene with this targeting construct will provide an animal with a CEL-null phenotype.
The electroporation of 5.5 ϫ 10 7 mouse ES cells with the targeting DNA resulted in approximately 4,800 G418-resistant colonies. One-fourth of the colonies were picked and expanded individually in 24-well dishes. A total of 268 colonies were selected for Southern blot analysis to screen for homologous recombination between the targeting DNA and the resident CEL gene. For the initial screening, ES colony DNA was digested with EcoRI and hybridized with an ϳ1100-bp PstI/SacI DNA fragment corresponding to genomic sequence 5Ј from the targeting DNA (Fig. 1C). As shown in Fig. 2A, the wild-type allele gives rise to a fragment Ͼ30 kb in length, while a correctly targeted gene yields a fragment 6.5 kb in length due to the insertion of two EcoRI sites present in the thymidine kinase promoter of neo R (Fig. 1B). To confirm that the putative targeting events had taken place as planned, additional aliquots of the ES colony DNA were digested with XbaI and hybridized with an ϳ1100-bp BglII/SalI DNA fragment corresponding to sequences 3Ј from the targeting DNA (Fig. 1D). Fig.  2B shows that the wild type allele yields a 7.2-kb fragment, while the correctly targeted allele yields a 9.0-kb fragment due to the insertion of 1.75 kb of DNA corresponding to the selectable marker cassette. Of the 268 colonies screened, 11 were positive in both tests. The overall targeting efficiency was 4.4%.
Site-specific integration of the targeting DNA at the CEL locus was confirmed by additional Southern blot analysis with both the 5Ј-and 3Ј-flanking probes. The addition of the 1.75-kb neo R cassette to the endogenous CEL gene resulted in an 8.3-kb HindIII fragment that hybridized with the 5Ј probe in addition to the 6.5-kb band observed for the controls (Fig. 2A). Using the 3Ј probe, a 7.5-kb SphI band resulted from the insertion of an SphI site in neo R in targeted clones in addition to the 18-kb SphI band observed for the wild type allele (Fig. 2B). These hybridization patterns were consistent with those predicted for the site-specific insertion of the neo R cassette into exon 4 of the endogenous CEL gene (Fig. 1). A total of eight enzymes, informative with either the 5Ј or 3Ј probe, were used to confirm that the gene targeting had occurred as planned. In addition, a neo R -specific probe was used to confirm that the targeting DNA had inserted in only one site in the genome (data not shown).
Two of the 11 cell lines with proper CEL gene targeting were used to generate chimeric mice. One cell line yielded only one chimeric mouse, which was female and had only ϳ5% agouti coat color. However, the second cell line produced 22 chimeric mice (from 119 injected and reimplanted blastocysts), all with extensive agouti coat color. Nineteen of these were male, and 15 of the 19 were able to transmit the modified gene to their offspring. Progeny from these test matings (chimerics ϫ Black Swiss), which carried the modified allele, were bred to generate homozygous knockout animals. Southern blot analysis of the genomic DNA from representative wild type, heterozygous, and homozygous CEL-targeted mice is shown in Fig. 3. Because of an apparent restriction fragment length polymorphism between the ES cells and the outbred mice used in the initial breeding, XbaI was not informative, and NcoI was used as a diagnostic enzyme for the 3Ј end of the recombination. This enzyme yields a 9.5-kb fragment from the wild type allele and a 6.5-kb fragment from the targeted allele due to the insertion of the NcoI site in the neo R gene (see Fig. 1

for details).
General Characteristics of CEL Gene-targeted Mice-The targeted allele was transmitted by the chimeric males at approximately the expected frequency (40%, 39 CELϩ/Ϫ mice of 97 agouti pups), indicating no obvious disadvantage to embryos or

FIG. 2. Southern blot analysis of wild type and CEL genetargeted ES colonies.
A, representative colonies digested with EcoRI or HindIII and hybridized with the 5Ј probe. The wild type allele gives rise to a Ͼ30-kb fragment with EcoRI, while the targeted allele yields a fragment of 6.7 kb due to EcoRI sites present in neo R . When digested with HindIII, the wild type allele is 6.7 kb, while the targeted allele is 8.5 kb due to the presence of neo R . B, representative colonies digested with SphI and XbaI and hybridized with the 3Ј probe. Wild type DNA yields an 18-kb fragment with SphI, while targeted DNA yields a 7.5-kb fragment due to an SphI site present in neo R . When digested with XbaI, the wild type DNA yields a 7.3-kb fragment, and the targeted DNA yields a 9.1-kb fragment due to the insertion of neo R . W, wild type; T, targeted.
To verify that the gene targeting abolished expression of CEL protein, pancreatic homogenates from control, heterozygous, and homozygous CEL gene-targeted mice were examined for CEL expression using immunoblotting techniques (Fig. 4). The levels of CEL protein in pancreatic extracts of the heterozygous animals were approximately half those of the wild type mice. No CEL protein was detected in homogenates of the CEL(Ϫ/Ϫ) mice. Furthermore, no CEL-immunoreactive polypeptides of any size were detected, indicating that no fusion protein or truncated protein was being produced as a result of the modified gene. These extracts were also assayed for cholesteryl ester hydrolytic activity. Table I shows that cholesteryl oleate hydrolysis is reduced 98% in the pancreatic extracts of CEL(Ϫ/Ϫ) animals.
Cholesterol Absorption Studies-The role of CEL in mediating absorption of dietary cholesterol and cholesteryl esters was assessed by comparing the amount of radiolabeled sterol excreted in the feces after its infusion into the stomach of normal and homozygous CEL gene-targeted mice. Since a role for CEL in the absorption of cholesteryl esters is not in dispute, the initial study used  In contrast to the cholesteryl ester result, when unesterified [ 3 H]cholesterol was included in the emulsion instead of the cholesteryl oleate, no significant difference was found between wild type and CEL-null mice in absorption of the radiolabeled sterol (Table III). Interestingly, males were found to absorb significantly less (59.6 Ϯ 3.05%) cholesterol than females (72.0 Ϯ 1.61%), but this difference was independent of their CEL genotype. For clarity, results from male and female mice are combined in Table III.
The presence of cholesteryl ester in the core of an emulsion particle has been shown to increase the partition of free cholesterol from the surface to the core (32). Our results show that CEL is necessary for digestion of this core cholesteryl ester. In the absence of CEL, the free cholesterol may remain sequestered and unabsorbed. To test this possibility, animals were fed [ 3 H]cholesterol in an emulsion that contained unlabeled cholesteryl ester along with phospholipid and triglyceride. Table III also shows that the presence of the cholesteryl ester in the core had no effect on the ability of CEL(Ϫ/Ϫ) mice to absorb the free cholesterol.
Published literature indicates that dietary cholesterol and biliary cholesterol may be absorbed from the intestine by different mechanisms (33). Experiments were undertaken to determine the ability of wild type and CEL gene-targeted mice to absorb unesterified cholesterol presented in a vesicular complex with phospholipids, similar to that present in the biliary tract. [␤-14 C]Sitosterol was used as a marker of recovery as described above. The results showed that, regardless of the ratio of cholesterol to phospholipid used to prepare the lipid vesicles, there was no significant difference between the ratios of [ 3 H]cholesterol to [␤-14 C]sitosterol recovered in the feces of wild type versus CEL gene-targeted mice (Table III).
The report that phenoxyphenyl carbamate inhibitors of CEL result in delayed absorption of cholesterol (7) prompted additional experiments to compare the rate at which the radiolabel from cholesterol and cholesteryl esters appears in the serum of control and CEL gene-targeted mice. In these experiments, the amount of 3 H in 15 l of serum was determined at various times after gastric infusion of emulsified radiolabeled sterol. The infusion of the unesterified [ 3 H]cholesterol resulted in the progressive appearance of the radiolabel in the serum of both wild type and CEL gene-targeted mice with a maximum at ϳ10 h (Fig. 5) and a slow decline thereafter. No significant difference in the rate of radiolabeled cholesterol appearance in the serum was observed between the two groups of animals in this case. In contrast, when the radiolabel was supplied as emulsified [ 3 H]cholesteryl oleate, the serum level of the radiolabel after 12 h was ϳ8-fold higher in the CEL(ϩ/ϩ) mice than in the CEL-null mice. In fact, very little radiolabel was detected in the serum of the gene-targeted mice (Fig. 5).
To examine the possibility that CEL plays a role in cholesterol absorption when mice are fed a high fat, high cholesterol, atherogenic diet, we studied the absorption of free and esterified cholesterol in wild type and CEL-null mice fed this diet for 6 weeks. As shown in Table IV, cholesteryl oleate absorption was reduced in the CEL-null mice, while free cholesterol was absorbed similarly by the wild type and CEL gene-targeted mice. The percentage of cholesterol absorbed was decreased relative to normal diet in both the free cholesterol and cholesteryl ester experiments due to the high level of cholesterol in the atherogenic diet. In fact the total mass of absorbed cholesterol is increased. DISCUSSION The results of the current study show that disruption of the CEL gene has no significant effect on the ability of mice to absorb unesterified cholesterol from the gastrointestinal tract. Similar results were observed regardless of the physical characteristics of the substrate or the dietary conditions of the animals. Furthermore, similar results were observed when cholesterol absorption was determined based on the amount of nonabsorbed cholesterol present in the feces or on the appearance of the radiolabeled cholesterol in the serum. These observations demonstrate that CEL is not necessary for cholesterol flux across the intestinal epithelium. In contrast to its effect on unesterified cholesterol absorption this study shows that CEL Although CEL does not appear to be essential for intestinal absorption of unesterified cholesterol, results of this study demonstrate unequivocally that CEL plays a primary role in absorption of cholesteryl esters. These results are consistent with in vitro studies from this laboratory and others that showed a role of CEL in facilitating the uptake of esterified cholesterol but not unesterified cholesterol by intestinal cells (10,11). Curiously, our fecal sterol experiments indicate a base-line level of cholesteryl ester absorption that is independent of CEL (Table II). However, pancreatic extracts from CEL(Ϫ/Ϫ) mice lack significant esterolytic activity (Table I). Also, the amount of radiolabel appearing in the serum of cholesteryl ester-fed, CEL-null mice was not consistent with this base line (Fig. 5). One explanation for this discrepancy is that a minor pathway for absorption of these nutrients exists and that our serum assay was insufficiently sensitive. Alternatively, some cholesteryl ester may remain associated with cell membranes or lipid vesicles of the intestinal epithelium due to hydrophobic interactions.
The current study, showing normal absorption of unesterified cholesterol in CEL-null mice, resolves the discrepancy in previously published data. Our results are consistent with those of Watt and Simmonds (3), which showed that unesterified cholesterol absorption was independent of pancreatic proteins. Gallo et al. (4) showed that CEL-depleted pancreatic juice could not restore cholesterol absorption in pancreatic-diverted rats; however, intestinal lymph flow in the CEL-depleted group was severely compromised in those experiments (11). Our current results are also in agreement with in vitro data that showed no CEL requirement for unesterified cholesterol uptake by Caco-2 cells (10,11). The CEL-stimulated uptake in Caco-2 cells reported by Lopez-Candales et al. (9), while intriguing, did not reflect physiologically relevant levels of cholesterol in the gastrointestinal tract.
Until the availability of an animal model lacking in CEL, such as the gene-targeted mice described here, the most physiologically relevant experiments regarding the role of this enzyme in cholesterol absorption were performed by feeding animals CEL inhibitors. Two classes of such inhibitors, phenoxyphenyl carbamates (7) and the lipstatins (35), were reported to reduce cholesterol absorption in normal and cholesterol-fed animals. However, the lipstatins were shown to also inhibit pancreatic lipase (35). Thus, their inhibitory effect on cholesterol absorption may be related to the inhibition of lipid emulsion hydrolysis and the release of unesterified cholesterol for diffusion through the mucosa (36). In support of this possibility was the observation that tetrahydrolipstatin had no effect on intestinal uptake of unesterified cholesterol from phospholipid-bile salt mixed micelles (35). Although the carbamate inhibitors were reported to be more specific for CEL and did not inhibit the activity of pancreatic lipase or acyl CoA: cholesterol acyltransferase in vitro, other possible side effects that can modulate cholesterol absorption in vivo may exist in the carbamate-treated animals.
Although our data show that CEL does not participate in free cholesterol absorption, the wide range of absorption values (37-87%) as well as additional studies with inbred strains of mice support the hypothesis that cholesterol absorption is regulated by at least one gene. 2 The recent report by Kirk et al. (37) on the responsiveness of different strains of mice to dietary fat and cholesterol with respect to cholesterol absorption and serum lipid parameters also supports this hypothesis. Candidate genes potentially involved in this process include the cholesterol transfer protein (38), which may be the same as, or closely related to, sterol carrier protein-2 (39,40), acyl CoA: cholesterol acyltransferase (5,41), pancreatic lipase (36), and liver fatty acid binding protein (42). Additional experiments are necessary to investigate the physiological role of these and other proteins in cholesterol absorption.
Although CEL does not play a primary role in free cholesterol absorption, its abundance in the intestinal lumen suggests that it may play a different role in the absorption of lipid-based nutrients. This enzyme may complement other lipolytic enzymes to increase the efficiency of dietary fat absorption by the small intestine. For example, CEL has been shown to be more efficient than pancreatic lipase in the hydrolysis of long chain  polyenoic fatty acids (43,44). CEL is also capable of hydrolyzing phospholipids and lysophospholipids (45) and thus may play a role in the assimilation of dietary phospholipids. More importantly, a pancreas-derived, vitamin-ester hydrolytic activity has been ascribed to CEL (46,47), suggesting its importance in the absorption of fat-soluble vitamins, which are primarily esterified in dietary sources. The presence of CEL in the milk of many mammalian species has led to the proposal that the milk CEL is critically important for digestion of milk triglycerides, the major source of energy in infants, before the maturation of the pancreas (48). Finally, in addition to its presence in the gastrointestinal tract, CEL is also found to be synthesized by the liver and is present in serum (12, 26, 49 -51), where its level is correlated to that of serum cholesterol and LDL (52). Thus, CEL may play a role in the modulation of lipoprotein structure and metabolism. The knockout mice described in this report will provide a useful tool to address the role of CEL in various aspects of lipid absorption and metabolism.