Hormone-sensitive lipase is a cholesterol esterase of the intestinal mucosa.

The identity of the enzymes responsible for lipase and cholesterol esterase activities in the small intestinal mucosa is not known. Because hormone-sensitive lipase (HSL) catalyzes the hydrolysis of acylglycerols and cholesteryl esters, we sought to determine whether HSL could be involved. HSL mRNA and protein were detected in all segments of the small intestine by Northern and Western blot analyses, respectively. Immunocytochemistry experiments revealed that HSL was expressed in the differentiated enterocytes of the villi and was absent in the undifferentiated cells of the crypt. Diacylglycerol lipase and cholesterol esterase activities were found in the different segments. Analysis of gut from HSL-null mice showed that diacylglycerol lipase activity was unchanged in the duodenum and reduced in jejunum. Neutral cholesterol esterase activity was totally abolished in duodenum, jejunum, and ileum of HSL-null mice. Analysis of HSL mRNA structure showed two types of transcripts expressed in equal amounts with alternative 5'-ends transcribed from two exons. This work demonstrates that HSL is expressed in the mucosa of the small intestine. The results also reveal that the enzyme participates in acylglycerol hydrolysis in jejunal enterocytes and cholesteryl ester hydrolysis throughout the small intestine.

sons revealed that HSL belongs to a family of esterases which is mainly represented by prokaryotic enzymes (2,3). From a structural point of view, HSL is the most complex protein of the family. Sequence alignments together with biochemical experiments suggest that adipocyte HSL is composed of two structural domains (4,5). The first 315 amino acids make up the N-terminal domain, which shows very little sequence similarities to other known proteins. The region responsible for the interaction with adipocyte lipid-binding protein (ALBP) was mapped to this domain (6,7). In adipose tissue, ALBP could increase the hydrolytic activity of HSL through its ability to bind and sequester fatty acids and through specific proteinprotein interactions. The C-terminal domain is divided in two functional parts, a catalytic core and a regulatory module. The latter is composed of 150 amino acids, including all of the known phosphorylation sites of HSL. Unlike other known mammalian triacylglycerol lipases, the activity of HSL is regulated by phosphorylation. The phosphorylation sites of protein kinase A, extracellular signal-regulated kinase, and AMP-dependent protein kinase have been mapped (8 -10). The catalytic core is the region that shows homology with the other members of the family. Modeling of the part revealed that it adopts an ␣/␤-hydrolase fold that harbors the catalytic triad constituted by Ser 423 , Asp 703 , and His 733 (4,11).
Several forms of HSL transcripts and the exon-intron organization of the HSL gene have been characterized in humans. The 88-kDa adipocyte HSL is translated from a 2.8-kb mRNA and encoded by 9 exons (12,13). The transcription start site was mapped in a short noncoding exon called exon B. In the adenocarcinoma cell line HT29, two mRNA species are found, the adipocyte HSL mRNA and a mRNA with a different 5Ј-end transcribed from exon A. Two testicular forms of HSL have been characterized. The 3.9-kb mRNA encodes a 120-kDa protein that contains a unique N-terminal region encoded by exon T1, a region that presumably forms a third structural domain in this isoform (14). The 3.3-kb mRNA encodes a protein that is identical to the adipocyte HSL form (15). However, the mRNA species differ in their 5Ј-ends. Exon usage is mutually exclusive, exon T2 being only transcribed in testis and exon B being transcribed in adipose tissue.
The nature and role of lipases and esterases participating in the digestion of dietary lipids in the lumen of the gastrointestinal tract are well established. In addition to the enzymes in the lumen, there is evidence of lipase activity in enterocytes (16). The presence of cholesterol esterase activity is more elusive. The exact identity of the enzymes responsible for the hydrolysis of intracellular acylglycerols and cholesteryl esters is still unclear. Pancreatic triacylglycerol lipase, microsomal triacylglycerol hydrolase, and pancreatic cholesterol esterase have all been suggested to be responsible for the hydrolytic activities (17)(18)(19). Pancreatic triacylglycerol lipase may be synthesized by the small intestine and accounts for the alkaline lipase activity of the enterocytes (19). Microsomal triacylglycerol hydrolase could also be involved (18). However, it has been shown that most of the lipase activity is cytosolic (20). Pancreatic cholesterol esterase, also called bile salt-stimulated lipase, is able to hydrolyze cholesteryl esters and triacylglycerols. In the absence of bile salts, the contribution of this enzyme is presumably minor (21). Because HSL is a cytosolic enzyme with a wide range of hydrolytic activities, the purpose of the present paper was to determine whether HSL was expressed in the intestine and could contribute to the hydrolysis of intracellular lipids.

EXPERIMENTAL PROCEDURES
Preparation of Intestinal Mucosa-French guidelines for the use and care of laboratory animals were followed. Male Swiss mice were fed ad libitum a standard chow (UAR A04, Usine d'Alimentation Rationnelle).
To study the expression of HSL along the gastro-colic axis of the gut, the small intestine from the pylorus to the ileocaecal valvula was removed, flushed with 0.9% NaCl at 4°C, and divided into five equal segments. The mucosa was scraped off at 4°C with a spatula. The first segment is considered to be the duodenum; segments 2-4, the jejunum; and segment 5, the ileum.
Northern Blot Analysis-Total RNA was extracted from adipose tissue and intestinal mucosa by the method of Chomczynski and Sacchi (22). RNA was denatured, subjected to electrophoresis on a 1% (w/v) agarose gel, and transferred to GeneScreen membranes (PerkinElmer Life Sciences). Rat intestinal fatty acid-binding protein (I-FABP, a gift from Dr. J. I. Gordon, Washington University, St. Louis, MO), mouse ALBP (a gift from Dr. P. Grimaldi, INSERM U 470, Université de Nice Sophia-Antipolis, Nice, France), and mouse HSL cDNA were used as probes. They were labeled with [␣-32 P]dCTP (3,000 Ci/mmol; Amersham Biosciences) using the Megaprime kit (Amersham Biosciences). A 24-residue oligonucleotide specific for rat 18 S rRNA was used as probe to ensure that equivalent quantities of RNA were loaded and transferred. This oligonucleotide was 5Ј-end labeled with T 4 polynucleotide kinase and [␥-32 P]ATP (3,000 Ci/mmol; Amersham Biosciences).
Analysis of 5Ј-cDNA Ends and Real Time Quantitative PCR of Intestinal HSL mRNAs-Total RNA was isolated using RNASTAT-60 (AMS Biotechnology). Total RNA (1 g) was treated with DNase I (DNase I amplification grade, Invitrogen), then retrotranscribed using random hexamers (Amersham Biosciences) and Thermoscript reverse transcriptase (Invitrogen) according to the manufacturer's recommendations. Combinations of the different primers and amplicon sizes are shown in Table I. The PCRs were performed on a Biometra apparatus with 94°C for 2 min followed by 35 amplification cycles (94°C for 20 s, 58 or 60°C for 30 s, 72°C for 30 s). This was followed by an additional elongation step of 7 min at 70°C. The PCR products were electrophoresed on agarose gels. Real time quantitative PCR was performed on a GeneAmp 7000 Sequence Detection System using SYBR green chemistry (Applied Biosystems). 18 S rRNA was used as control to normalize gene expression using the Ribosomal RNA Control Taqman Assay kit (Applied Biosystems).
Preparation of Whole Cell and Cytosolic Homogenates from Intestine and Adipose Tissue-Intestinal mucosa were homogenized in 4 volumes of homogenization buffer (0.25 M sucrose, 1 mM EDTA, pH 7.0, 1 mM dithioerythritol, 20 g/ml leupeptin, 20 g/ml antipain) using syringe and needle (19 and 25 gauge) to obtain whole cell homogenates. Homogenates were centrifuged at 110,000 ϫ g at 4°C for 45 min to prepare fat-free cytosolic supernatants. Cytosolic homogenates were prepared from adipose tissue samples as described for the intestine except that a glass/Teflon potter was used for homogenization. Protein concentrations were determined with a Bio-Rad protein assay using bovine serum albumin as standard.
Western Blot Analysis of HSL-Samples of 50 g of proteins from intestine supernatants and 15 g of proteins from adipose tissue supernatants were subjected to 10% SDS-PAGE, transferred onto nitrocellulose membrane (Hybond ECL, Amersham Biosciences), and probed with specific polyclonal anti-rat HSL antibody. Immunoreactive protein was determined by enhanced chemiluminescence reagent (Amersham Biosciences) and visualized by exposure to Fujifilm.
Immunocytochemistry-The jejunum and ileum were rapidly filled in and rinsed with 10% formalin, pH 7.0. Pieces were removed and immersed in the same fixative for 6 h at 4°C. After rinsing overnight with sodium phosphate, pH 7.4, 20% sucrose, specimens were frozen (Ϫ40°C in isopentane). Sections (10 m thick) were sliced on a cryostat HM 500 (Microm) at Ϫ25°C and mounted on glass slides coated with poly(Llysine). The sections were hydrated with NaCl/P i for 10 min, then incubated with 5% goat serum in 0.2% Triton X-100, NaCl/P i for 20 min. After being washed with 0.2% Triton X-100, NaCl/P i , sections were incubated with polyclonal anti-rat HSL antibodies generated in rabbits (1:500) in 5% goat serum, 0.2% Triton X-100, NaCl/P i , in a humid chamber for 6 h at room temperature. After several washes with 0.2% Triton, NaCl/P i , sections were incubated with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (Sigma) at a dilution of 1:160 for 60 min. After washing with NaCl/P i and addition of mounting media (Sigma) and coverslips, slides were examined with an ultraviolet visible confocal microscope (Leica TCS 4D).
Enzyme Activity Assays-In vitro enzymatic activities were performed on whole cell and cytosolic fractions. Diacylglycerol lipase and cholesteryl ester hydrolase activities were measured using phospholipid-stabilized emulsions of 1(3)-monooleoyl-2-O-monooleylglycerol or cholesterol oleate, respectively (21). Total esterase activity was determined using p-nitrophenyl butyrate as substrate. One unit of hydrolase activity is equivalent to 1 mol of fatty acid released/min at 37°C.
Generation of HSL-null Mice-HSL-null mice were generated by targeted disruption of the HSL gene in SV129-derived embryonic stem cells by standard procedures (23). In brief, the cDNA encoding the Aequorea victoria green fluorescent protein was inserted in-frame into exon 5 of the HSL gene, followed by a neomycin resistance gene, thereby disrupting the catalytic domain. The herpes simplex thymidine kinase gene was inserted at the 3Ј-end of the construct. Detailed description of the targeting construct will be published elsewhere. 2 After electroporation of embryonic stem cells, 96 colonies resistant to both G418 and ganciclovir were isolated, 10 of which showed homologous recombination as determined by Southern blot analysis. Two of these colonies were used for generation of two independent HSL-null mouse lines. In the present work animals from both of these lines were used with almost identical results, thus results from only one of the strains are presented. The studies were approved by the Animal Ethics Committee at Lund University. Statistical Analysis-Data are presented as the means Ϯ S.E. Values from HSL-null and wild-type mice were compared using the Mann-Whitney nonparametric test (Stat View software, Abacus concepts).

RESULTS
Northern Blot Analysis of HSL mRNA-The distribution of HSL mRNA along the gastro-colic axis was analyzed by Northern blot (Fig. 1A). HSL mRNA is present in the five segments of the small intestine (Fig. 1B). No expression of HSL mRNA was detected in colon (data not shown). Using quantitative RT-PCR, the HSL mRNA level was 8.8 Ϯ 0.7-fold higher in adipose tissue than in intestinal mucosa (n ϭ 5).
To determine that the specificity of the detected HSL mRNA signal did not derive from contaminating visceral adipose tissue, a series of hybridization was performed with different probes that were tested on intestinal and adipose tissue samples. As shown in Fig. 1C, ALBP mRNA that encodes an adipocyte-specific fatty acid-binding protein was only detected in the adipose tissue sample. There was no hybridization of the probe in the intestinal mucosa sample. Considering the quantities of total RNA loaded (15 g for adipose tissue and 30 g for intestine) and the intensity of the ALBP mRNA signal obtained in adipose tissue, a weak contamination of intestinal sample by visceral adipose tissue would thus have been detected. On the contrary, I-FABP, which encodes the intestinal fatty acid-binding protein, was only detected in the intestinal mucosa sample.
Western Blot Analysis of HSL-Western blot analysis with supernatants from the five parts of small intestinal mucosa was performed and developed with specific antibodies against rat HSL. A protein with similar apparent molecular mass (82 kDa) as the adipocyte murine HSL was detected in the different intestinal parts (Fig. 2). A lower molecular band that may correspond to a proteolytic fragment was present in intestinal samples.
Immunocytochemistry Analysis of HSL-Immunocytochemistry experiments were performed on jejunum (Fig. 3, A and B) and ileum (Fig. 3, C and D) sections. HSL protein was detected in the differentiated cells of the villi and was absent in the undifferentiated cells of the crypt.
Enzyme Activity Assays-Cholesterol esterase activity was measured with cholesterol oleate. Using a diacylglycerol analog in which only the first ester bond can be hydrolyzed, diacylglycerol lipase activity can be determined without simultaneously measuring monoacylglycerol lipase activity. Enzymatic activity determination on intestinal musosa cytosolic fractions indicated the presence of a cholesterol esterase and a lipase along the small intestine (Fig. 4). Cholesterol esterase and diacylglycerol lipase activities were 9-and 4-fold lower in jejunum than in white adipose tissue, respectively. Diacylglycerol lipase activity was inhibited using 100 M diethyl-p-nitrophenyl phosphate by 75 Ϯ 3% (range from 68.2 to 82.5) in all intestinal segments and by 94 Ϯ 1% in white adipose tissue.
Analysis of HSL-null Mice Intestine-In vitro enzymatic assays were realized in HSL-null mice intestine and compared with those from wild-type littermates. Cholesterol esterase activity was totally abolished in the cytosolic fractions of duodenum, jejunum, and ileum of HSL-null mice (Fig. 5A). Although diacylglycerol lipase activity was unchanged in the duodenum of HSL-null mice, it was significantly reduced in jejunum (Fig.  5B). Enzymatic assays were also performed in whole cell homogenates. There was almost no cholesterol esterase activity in the various parts of HSL-null mouse small intestine (Fig. 6A). Diacylglycerol lipase activity was decreased in jejunum (Fig. 6B). Total esterase activity was not modified by the lack of HSL (Fig. 6C). As shown in Fig. 7, Western blot analysis performed on HSL-null mouse intestine showed complete disappearance of the 82-kDa protein. A higher molecular mass band was detected in wild-type mice. It may correspond to the 89 kDa band observed in some rat tissues expressing HSL (24).  1. Northern blot analysis of HSL mRNA. A, the small intestine was divided into five segments. The first segment is considered to be the duodenum (D); segments 2-4, the jejunum (J); and segment 5, the ileum (I). Total RNA from intestinal mucosa (30 g) was resolved on a 1% agarose gel containing 2.2 M formaldehyde, transferred, and fixed to a nylon membrane. B, the bar graph represents HSL mRNA data normalized to 18 S rRNA for difference in total RNA loading. Values are the means Ϯ S.E. from three independent determinations. C, RNA from adipose tissue (15 g) or from jejunal mucosa (30 g) was resolved on a 1% agarose gel containing 2.2 M formaldehyde, transferred, and fixed to a nylon membrane. Hybridizations were performed with HSL, ALBP, I-FABP, and 18 S rRNA probes.

FIG. 2. Western blot analysis of intestinal HSL.
The small intestine was divided into five parts from duodenum to ileum, and the mucosa was scraped off. Cytosolic fractions were prepared from the five intestinal segments (numbered 1-5) and visceral white adipose tissue (WAT). Lanes were loaded with 50 g of protein for intestine and 15 g of protein for adipose tissue, subjected to SDS-PAGE, and Western blotted with anti-rat HSL antibody. The arrow shows the murine HSL protein (82 kDa). The lower molecular mass band in intestinal samples may correspond to a proteolytic fragment. Analysis of the 5Ј-Ends of Intestinal HSL mRNA-Two forms of HSL transcripts have been characterized in the adenocarcinoma cell line HT29 (13). The 5Ј-ends of the two forms are transcribed either from exon B or from exon A. In an attempt to characterize intestinal HSL transcripts, different primers were used in RT-PCR (Table I) with mRNA from intestinal mucosa. As expected, the use of primers in exon 1 led to an amplification of a 206-bp PCR product in adipose tissue and intestine (Fig. 8).
Using different antisense primers in exon 1 with sense primers designed either in exon A or in exon B, we could detect the different PCR products with the expected size in intestine. These results suggest that two HSL mRNA with mutually exclusive 5Ј-ends coexist in enterocytes. The relative abundance of exon Aand exon B-containing transcripts was determined using quantitative RT-PCR on adipose tissue and intestinal mucosa total RNA (n ϭ 4). The ratio of exon B to exon A transcripts was 4.4 Ϯ 0.2 in the adipose tissue and 1.1 Ϯ 0.1 in the intestine. The data reveal that exons A and B are used equally in the enterocytes.  7. Western blot of intestinal mucosa extracts from wildtype and HSL-null mice. Cytosolic fractions were prepared from jejunal and ileal segments. Lanes were loaded with 50 g of protein, subjected to SDS-PAGE, and Western blotted with anti-rat HSL antibody. The arrow shows the murine HSL protein (82 kDa). The higher molecular mass band in wild-type mice may correspond to a previously described HSL form in rat tissues (24).

DISCUSSION
Here we have found that HSL contributes to lipase activity and is the major cholesterol esterase in the intestinal mucosa. HSL mRNA and protein were readily detected along the small intestine. Immunohistochemistry data revealed that the enzyme is expressed preferentially in differentiated cells of the villi. Data from HSL-null mice showed that HSL does not account for a significant part of total esterase activity. However, the enzyme is responsible for all neutral cholesteryl ester hydrolase activity both in the cytosolic and particulate fractions. It also contributes between one-third and one-half of diacylglycerol lipase activity in the jejunum.
The size of intestinal HSL mRNA and protein corresponds to those of mouse adipocyte HSL, i.e. a 2.6 -2.8-kb mRNA and an 82-kDa protein. The 5Ј-ends of the mRNA species are transcribed from two exons that correspond to human exons A and B. Exon A is located in the mouse gene Ϸ 7 kb upstream of exon B (25). Use of the two exons is mutually exclusive. Interestingly, we have shown previously that HSL is expressed in the human adenocarcinoma cell line of intestinal origin HT29 (26). The two mRNA 5Ј-ends were found in HT29 HSL mRNA (13). In human and mouse adipose tissue, the main transcription start site is located in exon B. Exon A-containing transcripts are found at very low levels in humans (13) and are much less abundant than mRNA with exon B in mice ((25) and present work). To date, there is little information on the mechanisms controlling tissue-specific expression of HSL. We have shown that the first 95 bp of human exon T1 5Ј-flanking region conferred expression of a reporter gene exclusively in testis of transgenic mice (27). Exon B 5Ј-flanking region contains an active promoter with an E-box and two GC-boxes as functional cis-acting elements (28). In adipocytes, the E-box mediates the glucose-mediated induction of HSL gene expression. However, the sequences responsible for the adipose tissue-specific expression of HSL are not present in this region because the pattern of promoter activity up to 2.4 kb was similar in adipocytes and in HeLa cells that do not express HSL (13). No data are yet available for the exon A 5Ј-flanking region. In the enterocytes, exons A and B are represented in equal amounts, suggesting that two alternative promoters control HSL gene expression providing the possibility of distinct transcriptional regulation.
Lipid processing through the intestine is a complex pathway with multiple control steps. The intestine is unable to transport neutral lipids into the lymph at the rate with which they are absorbed, especially at high input rates. Nearly half of the triacylglycerol mass infused into rat intestine does not appear in the lymph (29). It is unlikely that the lipids are oxidized because ␤-oxidation of lipid entering the mucosa from the lumen is limited. There is no evidence that triacylglycerols are transported via the portal vein (30). These studies suggest that some triacylglycerols in the enterocyte are undergoing hydrolysis. In support of this concept, a mucosal triacylglycerol pool distinct from the chylomicron triacylglycerol precursor pool has been characterized (31). Lipolysis of the mucosal pool has been shown both in vitro and in vivo. Both acidic and alkaline lipase activities have been described in the mucosa (20,32). Because most of the lipolytic activity was found at neutral or basic pH, the physiological importance of the acidic lipase is unclear.
Here, we confirm that significant neutral lipase activity is found in the enterocyte. This activity was inhibited by diethylp-nitrophenyl phosphate as shown previously for mucosal lipolysis in triolein-infused rats (31). Lipase activity was found in the different parts of the small intestine. Data from HSL-null mice show that HSL contributes to lipase activity in the distal section but not in the first part of the small intestine. Recently, Mansbach and colleagues (19) showed that pancreatic lipase was expressed in the intestine with most of the enzyme detected in the first quarter (19). Altogether, the data suggest that the hydrolysis of mucosal triacylglycerols is caused by pancreatic lipase in the proximal part of the small intestine and HSL in the more distal parts.
The nature of the enzyme responsible for the hydrolysis of cholesteryl esters in the intestine has remained unclear. Pancreatic cholesterol esterase (bile salt-stimulated lipase) is internalized upon binding to the surface of enterocytes (33). The esterase could hydrolyze intracellular cholesteryl esters or conversely participate, at acid pH, in the esterification of cholesterol (17,34). However, the intracellular esterase activity in the absence of cofactors such as bile salts may be very low. Contribution to cholesterol esterification is also unlikely because studies on knockout mice revealed that the enzyme is responsible for mediating intestinal absorption of cholesteryl esters but does not influence free cholesterol absorption (35). In contrast, acyl-CoA:cholesterol acyltransferase 2-deficient mice are resistant to diet-induced hypercholesterolemia (36). Localization of pancreatic cholesterol esterase in intestinal epithelium may therefore not be related to intracellular metabolism. There is evidence that the enzyme, via an apical-to-basolateral transcytotic pathway, is released at the basolateral membrane level and may contribute to serum pancreatic cholesterol esterase activity (37). Our data reveal that HSL and not pancreatic cholesterol esterase accounts for neutral cholesterol esterase activity in the small intestine.
The expression of HSL in the enterocytes may open new paths in our understanding of cholesterol intestinal absorption and metabolism. HSL-mediated hydrolysis of the intracellular pool of cholesteryl esters may contribute together with the esterification process mediated by acyl-CoA:cholesterol acyltransferase-2 and cholesterol transport mediated by ATP-binding cassette (ABC) transporters to the control of cholesterol homeostasis. Several transporters are expressed in the intestinal epithelium. ABCA1 is expressed in the small intestine and may modulate cholesterol absorption. However, data from ABCA1-deficient mice are conflicting (38,39). Studies in patients with sitosterolemia (40, 41) and in transgenic mice overexpressing ABCG5 and ABCG8 (42) suggest that the halftransporters participate in cholesterol efflux. Hydrolysis of FIG. 8. RT-PCR amplification of HSL mRNA 5-ends. Total RNA was extracted from intestinal mucosa (Int.) and white adipose tissue (WAT). RT-PCR was performed using primers derived from various exons (Ex.) with (ϩ) and without (Ϫ) reverse transcriptase. Information on the primers is provided in Table I. cholesteryl esters by HSL may produce free cholesterol for export through ABC transporters into the lumen. Because of the unique properties of HSL, the present work paves the way for future studies on lipid metabolism in the enterocyte.