Site-directed Mutagenesis of the Basic N-terminal Cluster of Pancreatic Bile Salt-dependent Lipase

Previous studies have postulated the presence of a heparin-binding site on the bile salt-dependent lipase (BSDL), whereas two bile salt-binding sites regulate the enzyme activity. One of these sites may overlap with the tentative heparin-binding site at the level of an N-terminal basic cluster consisting of positive residues Lys32, Lys56, Lys61, Lys62, and Arg63. The present study uses specific site-directed mutagenesis to determine the functional significance of this basic cluster. Mutations in this sequence resulted in recombinant enzymes that were able to bind to immobilized and to cell-associated heparin before moving throughout intestinal cells. Recombinant BSDL was fully active on soluble substrate, but mutants were less active on micellar cholesteryl oleate in comparison with the wild-type enzyme. Activation studies by primary (sodium taurocholate) and by secondary (sodium taurodeoxycholate) bile salts revealed that the activation of BSDL by sodium taurocholate at concentrations below the critical micellar concentration, and not that evoked by micellar bile salts, was affected by substitutions, suggesting that this N-terminal basic cluster likely represents the specific bile salt-binding site of BSDL. Substitutions also affected the activation of the enzyme promoted by anionic phospholipids, extending the function of this site to that of a cationic regulatory site susceptible to accommodate anionic ligands.

Previous studies have postulated the presence of a heparin-binding site on the bile salt-dependent lipase (BSDL), whereas two bile salt-binding sites regulate the enzyme activity. One of these sites may overlap with the tentative heparin-binding site at the level of an N-terminal basic cluster consisting of positive residues Lys 32 , Lys 56 , Lys 61 , Lys 62 , and Arg 63 . The present study uses specific site-directed mutagenesis to determine the functional significance of this basic cluster. Mutations in this sequence resulted in recombinant enzymes that were able to bind to immobilized and to cell-associated heparin before moving throughout intestinal cells. Recombinant BSDL was fully active on soluble substrate, but mutants were less active on micellar cholesteryl oleate in comparison with the wild-type enzyme. Activation studies by primary (sodium taurocholate) and by secondary (sodium taurodeoxycholate) bile salts revealed that the activation of BSDL by sodium taurocholate at concentrations below the critical micellar concentration, and not that evoked by micellar bile salts, was affected by substitutions, suggesting that this Nterminal basic cluster likely represents the specific bile salt-binding site of BSDL. Substitutions also affected the activation of the enzyme promoted by anionic phospholipids, extending the function of this site to that of a cationic regulatory site susceptible to accommodate anionic ligands.
Bile salt-dependent lipase (BSDL, EC 3.1.1.13) 1 is a lipolytic enzyme secreted by the acinar pancreatic cell into the duodenum, where it plays a significant role in dietary lipid digestion. BSDL has been shown to display wide substrate reactivities ranging from the hydrolysis of both long-chain and short-chain fatty acid esters of glycerol, as well as phospholipids, lysophospholipids, and esters of cholesterol and of the fat-soluble vitamin A, E, and D (1). BSDL, unlike other lipases, is characterized by a unique activation mechanism requiring the binding of bile salts. Early studies have proposed that bile salts interact with two sites on the protein (2). One site, specific for primary bile salts, is associated with enzyme dimerization and activation, whereas the second is less specific, is able to bind indistinctly primary and secondary bile salts, and is involved in the enzyme binding to micellar or aggregated substrates (3,4). More recently, the presence of these two bile salt-binding sites has been detected on the human milk counterpart enzyme referred to as bile salt-stimulated lipase (5). These two bile salt-binding sites have been tentatively localized on each BSDL molecule forming the dimeric BSDL-taurocholate complex crystal (6,7). The first one, proximal to the catalytic site, could be identified as the specific binding site (8). The second one is located in a depression region on the back side of the catalytic domain of the enzyme. Binding of a monomeric primary bile salt to the specific site leads to the opening of a loop comprised of residues His 115 to Tyr 125 of the bovine BSDL, a loop that otherwise is in a closed conformation which might hinder substrate binding (6,7). Using chemical modification approaches, tyrosine and arginine have been identified as key residues for BSDL interaction with bile salts (3,4). Furthermore, a recent study demonstrated that Arg 63 is essential for the enzyme activity on substrates such as cholesteryl oleate solubilized by sodium taurocholate (9).
It has been shown that BSDL binds to heparin-like molecules lining the intestinal wall and that this binding is reversed by the addition of soluble heparin (10). Furthermore, BSDL located on the epithelial cell surface may facilitate the uptake of hydrolyzed dietary lipids and cholesterol (11). However, the association of BSDL with membrane-associated heparin of intestinal cells may be a prerequisite for the uptake of BSDL by these cells and the consecutive transcytotic motion of the enzyme throughout the enterocyte (12). Several groups have proposed that a basic cluster in the N-terminal domain of BSDL may be involved in the binding to heparin (6,8). This Nterminal cluster of positively charged residues forms a cationic protrusion at the surface of the protein (6,8). The side chains of Lys 31 , Lys 56 , and Lys 58 lie on one side of a groove, whereas those of Lys 61 , Lys 62 , and Arg 63 lie on the other side (6). These latter cationic residues are common to all BSDL (13). Another basic cluster consisting of seven positively charged residues (Lys 336 , Arg 423 , Lys 429 , Arg 454 , Arg 458 , Lys 462 , and Lys 503 ) has been localized on the BSDL surface (6). This cluster could facilitate the binding of BSDL (possibly dimers) to micelles (6).
On the surface of the BSDL dimer, these residues form two rows of positively charged residues parallel to each other that could also represent another putative heparin-binding site. From those data, it looks possible that heparin-binding and bile salt-binding surfaces may overlap to some degree.
The close localization of the sequence Lys-Lys-Arg 63 , forming the putative heparin site with the disulfide bridge formed by Cys 64 and Cys 80 , leads us to also postulate that this site may be involved in the sensing of the BSDL folding and the secretion of the protein (1). Furthermore, cationic residues of this cluster may not only accommodate the sulfate group of taurine-conjugated bile salts and those of heparin but also the anionic polar head group of phospholipids such as phosphatidylserine, phosphatidylinositol, and phosphatidic acid that have been shown to regulate neutral cholesterol ester hydrolase (14). The aim of the present study is to use a site-directed specific mutagenesis approach to determine the functional significance of the Nterminal basic cluster of BSDL.

EXPERIMENTAL PROCEDURES
Materials-Unless otherwise stated, all A grade chemicals were purchased from Sigma. Culture medium Ham's F12 was from Invitrogen. Taq polymerase was purchased from CLONTECH (Palo Alto, CA) and was a part of the GC-rich PCR kit. Polyclonal antibodies (pAbL10) against a peptide sequence (Val 590 to Gln 605 ) of the rat BSDL were raised in rabbit and purified on protein A-Sepharose. These antibodies only recognized rat pancreatic BSDL.
Site-directed Mutagenesis-The pECE-1 transcript, initially containing wild-type cDNA encoding rat BSDL (15), was subcloned in Escherichia coli cells and mutagenized. For this purpose, oligonucleotide primers designed to cover the sequence encoding each residue to be mutagenized were used in PCR reactions in combination with the overlap extension method (16). The sequence of primers allows the mutagenesis of the sequence Lys-Lys-Arg 63 into Ile-Ile-Ala 63 , that of Lys 32 and Lys 56 into Ile 32 and Ile 56 , respectively, and can be released upon request. The oligonucleotide primers, each complementary to opposite strands of the cDNA sequence of BSDL in pECE-1 vector, were extended using the Advantage GC-rich PCR kit from CLONTECH. The amount of the initial substrate was maintained below 100 ng. Independently of the pair of primers used, the amplification was performed on a PerkinElmer Life Sciences 2400 GeneAmp PCR system using a prerun cycle (94°C, 100 s), and 12 reaction cycles were programmed as follows: denaturation (94°C, 20 s), annealing (60°C, 30 s), and extension (68°C, 4 min). The reaction was terminated by an incubation at 68°C for 4 min. A treatment with DpnI endonuclease was used to eliminate the methylated parental DNA template. About 10 l of the digested material were used to transform competent Escherichia coli cells (Top10 FЈ strain), which were spread on agar plates containing ampicillin and incubated at 37°C overnight. Colonies were picked up and cultured in Luria-Bertani medium supplemented with 50 g/ml ampicillin. Cells were pelleted and plasmid cDNA isolated using a miniprep kit method (NucleoSpin Plus, Macherey-Nagel). The first mutagenized plasmid, including the sequence Ile-Ile-Ala 63 , was then used to mutagenize Lys 32 into Ile 32 . This latter plasmid was finally used as template to substitute Lys 56 by Ile 56 . In short, two plasmids were obtained, the first one bearing mutations on the sequence Lys-Lys-Arg 63 representing the putative heparin-binding site and the second one including mutation of the latter sequence and that of Lys 32 and Lys 56 (representing the basic N-terminal cluster of BSDL). These two plasmids were referred to as 3M (for three mutations, K61I/K62I/R63A) and 5M (for five mutations, K32I/K56I/K61I/K62I/R63A), respectively. Plasmids were completely sequenced using specific primers for BSDL sequence (13) to detect any misincorporation of bases by the polymerase within the BSDL cDNA. Mutagenized cDNA of BSDL was then digested using EcoRI restriction enzyme and ligated into pECE-1 vector from which the wild-type cDNA encoding BSDL has been excised previously following the same restriction procedure. This eliminates undesired substitution in the vector. Plasmids pECE-1-3M and pECE-1-5M, bearing the desired mutation, were amplified and sequenced as above, and plasmids with the right orientation were transfected into CHO-K1 cells.
Co-transfection-pECE-1-3M and pECE-1-5M plasmids encoding mutagenized BSDL were independently co-transfected with pMAM-neo vector (ratio of pMAM-neo/pECE-1, 1/25), an expression vector that carries the resistance to Geneticin (G418). This material was transfected into CHO-K1 cell line using LipofectAMINE according to manufacturer's procedure (Invitrogen). Transfected cells were stabilized in Ham's F12 medium supplemented with G418 (1 mg/ml). The different clones were then isolated by the end-dilution procedure and maintained under G418 selection for at least 3 weeks. Control cells were transfected, according to the same protocol, with the empty pMAM-neo vector, and corresponding positive clones (CHO-control) were selected as above indicated. The CHO-K1 cell line expressing the wild-type BSDL referred to as clone 3B has been already described (17).
Protein Preparation-Transfected CHO cells were grown to about 80% confluence, and they were then washed twice with incomplete PBS buffer (10 mM sodium phosphate buffer at pH 7.4 with 0.15 M NaCl and without Mg 2ϩ and Ca 2ϩ ions) and scraped with a rubber policeman. Cells obtained from a 100-mm diameter dish culture were sedimented and homogenized in 0.5 ml of lysis buffer (5 mM Mes, 0.1 M NaCl, 1 mM EDTA, 1 mM benzamidine, and 0.5% Nonidet P40, pH 6.0) by sonication (15 s, 4 watts, 4°C). Homogenates were quickly cleared by centrifugation at 14,000 ϫ g for 30 min at 4°C, and the supernatants were immediately used for enzymatic assays.
Recombinant BSDL variants were produced by transfected CHO-K1 cells cultured in Ham's F12 medium (100 units/ml penicillin and 100 g/ml streptomycin) for 16 h in the absence of FCS. Cell culture media were then used as recombinant enzyme source.
Enzyme Activity and Protein Determinations-The esterolytic activity of BSDL was recorded using 4-nitrophenyl hexanoate (4-NPC) as substrate (18) in the absence or in the presence of sodium taurocholate (NaTC, 4 mM). When required, sodium taurodeoxycholate (NaTDC) or phospholipids (see below) were added in the cuvette. The concentration of 4-NPC in each assay cuvette was determined after the complete hydrolysis of the substrate provoked by adding 100 l of 5 N NaOH and using a molar absorption coefficient of 5150 M Ϫ1 cm Ϫ1 at the isosbestic point (348 nm) of 4-nitrophenol and its anion (19). The critical micellar concentration (CMC) of bile salts in the assay condition of BSDL was recorded by the absorbance of rhodamine 6G (30 M) at 540 nm (3).
The cholesterol esterase activity of BSDL was determined on cholesteryl [1-14 C]oleate (PerkinElmer Life Sciences). Briefly, to a mixture made with 125 l of Tris-HCl buffer (100 mM, pH 7.5) with or without 14.5 mM sodium taurocholate, 100 l of recombinant enzyme and 25 l of water were added to 10 l of an ethanolic solution of cholesteryl [ 14 C]oleate. After incubation for 30 min at 37°C under agitation, the reaction was stopped with 0.83 ml of Dole mixture (20) containing 100 M non-radiolabeled oleic acid used as vector. The released [ 14 C]oleic acid was extracted from the medium by a liquid-liquid system partition as described (20), and the radioactivity was estimated by counting (PCS mixture, Amersham Biosciences) with an LKB scintillation counter. Protein concentration was determined with the bicinchoninic acid test from Pierce using BSA as standard.
Preparation of Phospholipid Suspension-A stock solution of phospholipids in chloroform was evaporated under nitrogen and resuspended at a concentration of 2 mg/ml in 20 mM, pH 7.4, Tris-HCl buffer. Suspension was obtained by sonication of lipid (1 min, 10 watts). The phospholipid suspension was then diluted in the BSDL assay to the required final concentration.
SDS-PAGE and Immunoblotting-SDS-PAGE was performed in 7.5% polyacrylamide and 0.1% SDS as described by Laemmli (21) using a Bio-Rad Mini Protean II apparatus. After electrophoretic migration, proteins were electrotransferred onto nitrocellulose membranes at 4 mA/cm 2 for 18 h. The efficiency of the electrotransfer was checked by staining the nitrocellulose membrane with 2% Ponceau S solution. Nitrocellulose membranes were air-dried, and transferred proteins were detected by immunodetection using polyclonal antibodies pAbL10 (1 g/ml) specific for the rat pancreatic BSDL. Membranes were developed for 10 min with a mixture of nitro blue tetrazolium and 5-bromo-4-chloro-3-indoyl phosphate (0.5 mM each) in 0.1 M Tris/HCl buffer (pH 9.5), 100 mM NaCl, and 1 mM MgCl 2 and then analyzed by densitometric scanning and quantified using the Image program (National Institutes of Health, Bethesda, MD) Northern Blotting-Total RNAs were extracted from transfected CHO cells using Trizol reagent (Invitrogen). After migration on 1% agarose gel, RNAs (about 20 g) were transferred onto a nitrocellulose membrane (22). cDNA probes specific for pancreatic BSDL (500 bp) and for ␤-actin (300 bp) were obtained and used as described previously (13). Probes were radiolabeled by using [␣-32 P]dCTP (PerkinElmer Life Sciences) and the random primed DNA labeling kit (Roche Molecular Biochemicals). Conditions used for dot-blot dilution and quantification have been already described (22).
Binding of Wild-type and Recombinant Mutants of BSDL to Heparin-Sepharose-Heparin-Sepharose (Sigma) column (3-ml wet gel) was equilibrated in a Tris-HCl buffer (10 mM, pH 7.0, 50 mM NaCl, loading buffer). Wild-type or recombinant mutants of BSDL expressed during 16 h in the conditioned medium (without FCS) of transfected CHO-K1 cells were concentrated by centrifugation on Amicon filters (Bedford, MA) up to 3.8 -4.0 ml. This medium containing 0.3-0.4 BSDL units as determined on 4-NPC was loaded onto the heparin-Sepharose column and incubated for 16 h at 4°C under rotation. Unbound material was then eluted (1 ml/min) with the loading buffer, and bound material was further eluted with a linear NaCl concentration gradient from 50 to 500 mM in 10 mM, pH 7.0, Tris-HCl buffer. Two washings were then performed successively with 1 and 2.5 M NaCl in the Tris-HCl buffer. The elution profile was monitored by recording BSDL activity on 4-NPC.

Expression of Recombinant Mutagenized BSDL-
The expression of BSDL bearing the mutagenized putative heparinbinding site K61I, K62I, and R63A and that bearing mutation of the basic N-terminal cluster K32I, K56I, K61I, K62I, and R63A was examined in all clones selected for G418 resistance (i.e. six clones with three mutations and three clones with five mutations) and compared with that of the wild-type 3B clone (17) and with that of the control clone (i.e. transfected with the pMAM-neo plasmid only). For this purpose, all positive clones were cultured to subconfluence, their respective conditioned media were withdrawn for further analyses, and cells were washed with PBS, harvested, and lysed. Alternatively, before lysis, cells expressing recombinant BSDL were washed with PBS containing 0.25 M NaCl (a salt concentration that should liberate BSDL associated with heparanoids of the outer CHO cell surface, see below). Under these conditions, the amount of BSDL released from membranes never exceeded 5% of that found in the cell lysate (as assessed from the enzyme activity). The cell lysates were cleared and analyzed on SDS-PAGE and with Western blotting. Their activity on 4-NPC in the presence of 4 mM NaTC was also recorded. As shown in Fig. 1, all selected clones (excepted control clones) expressed BSDL albeit at various levels. The expression level of BSDL protein correlated with the activity on 4-NPC recorded in each lysate. When the BSDL activity monitored in each lysate was reported to the corresponding amount of protein determined from Western blotting quantification, similar ratios were obtained (230 Ϯ 56, 222 Ϯ 58 and 180 Ϯ 50 for the wild-type enzyme (clone 3B), K61I/K62I/R63A mutants (clones 3M1-3M7), and K32I/K56I/ K61I/K62I/R63A mutants (clones 5M1-5M5), respectively). This means that substitutions do not affect significantly the enzyme activity on 4-NPC. The esterolytic activity recorded in control clone lysates (clones C1-C3) could be due to endogenous esterases (23). The presence of BSDL was also examined in the conditioned medium of selected clones. Western blotting (Fig.  2) showed that BSDL is present in all culture media except those of control clones. Once again, the activity on 4-NPC recorded in these culture media paralleled the amount of BSDL as quantified by Western blotting.
These data indicated that, independently of the number of mutagenized residues of the N-terminal basic cluster, BSDL is normally expressed and secreted by transfected CHO cells. The amount of BSDL secreted by each clone corroborates the amount of enzyme that is expressed by the clone in question. Overall, in the presence of 4 mM NaTC, K61I/K62I/R63A and K32I/K56I/K61I/K62I/R63A mutagenized BSDL had comparable activity on synthetic water-soluble ester substrate such as 4-NPC than the wild-type enzyme.
From this point and besides clone 3B expressing the wildtype BSDL, two clones expressing K61I/K62I/R63A and K32I/ K56I/K61I/K62I/R63A mutagenized BSDL (i.e. clones 3M2 and clone 5M5, respectively) expressing an amount of enzyme comparable with that of clone 3B were selected and further analyzed. Dot-blot and Northern blot analyses were used to assess the mRNA abundance and size in selected clones (Fig. 3). Dot-blot quantification indicated that the BSDL mRNA was in similar amounts in stably transfected CHO cell clones expressing wild-type (clone 3B), K61I/K62I/R63A (clone 3M2), and K32I/K56I/K61I/K62I/R63A BSDL (clone 5M5), whereas no mRNA can be detected in control clone C1. The mRNA (2.0 kb) encoding BSDL detected in clone 3B expressing the wild-type enzyme and that detected in clone 3M2 and 5M5 expressing K61I/K62I/R63A and K32I/K56I/K61I/K62I/R63A mutagenized variants of BSDL, respectively, was of the expected size (15). Also, the cDNA probe for ␤-actin hybridized with a transcript of the right size (13), present in all selected clones. The amount of mRNA encoding BSDL, equivalent in all selected clones, correlated quite well with the amount of BSDL expressed by the corresponding clone. Furthermore, and except for the control clone C1 that does not secrete BSDL, the three selected clones expressing wild-type, K61I/K62I/R63A, and K32I/K56I/K61I/ K62I/R63A BSDL (i.e. 3B, 3M2 and 5M5 clones) secreted the enzyme at the same rate (19.4 Ϯ 2.1 10 Ϫ3 units/mg of cell proteins/h).
Binding of Recombinant Wild-type and Mutant BSDL to Immobilized Heparin-Wild-type BSDL (clone 3B), K61I/K62I/ R63A (clone 3M2), and K32I/K56I/K61I/K62I/R63A (clone 5M5) mutagenized recombinant variants of BSDL were characterized based on heparin binding. Heparin-Sepharose chromatography was used to measure the relative affinities of mutants for heparin, and the position of the peak in the heparin-Sepharose chromatogram reflects the affinity of protein for immobilized heparin. Therefore the position of the peak is expected to shift to a lower salt concentration upon mutation of any residue that forms a heparin-binding site in the wild-type enzyme. Accordingly, the same amount of recombinant BSDL was chromatographed on a heparin-Sepharose column, and after elution of unbound material, bound enzyme was eluted with a linear gradient in NaCl. The results showed that mutations of either the putative heparin-binding sequence Lys-Lys-Arg 63 or of this sequence associated with the substitutions of Lys 32 and Lys 56 did not affect the binding of BSDL to Sepharose-immobilized heparin. Wild-type and mutagenized variants of BSDL were eluted at a similar volume of the NaCl gradient (Fig. 4). Such results strongly suggest that amino acids of the N-terminal cationic cluster of BSDL, i.e. Lys-Lys-Arg 63 , Lys 32 , and Lys 56 likely do not participate in binding to immobilized heparin.
Transcytosis of Recombinant Wild-type and Mutants BSDL throughout Intestinal Cells-Heparin-like molecules lining intestinal microvilosities have been implicated in the binding of BSDL to the intestinal wall (10). This binding should precede the transcytotic motion of the enzyme throughout enterocytes (17). Therefore, if mutagenized BSDL is still capable of binding to immobilized heparin molecules, it should a priori also be taken up by and move throughout intestinal cells. We attempted to demonstrate this specific point by examining the transcytosis of BSDL through Int407 intestinal cells cultured in Transwell inserts to form a tight epithelium (12). As shown in Fig. 5A, the wild-type (clone 3B), K61I/K62I/R63A (clone 3M2), and K32I/K56I/K61I/K62I/R63A (clone 5M5) recombinant variants of BSDL were allowed to move throughout Int407 cells at the same rate, from the apical reservoir to the lower reservoir of the Transwell insert. No activity on 4-NPC in the presence of 4 mM NaTC can be recorded with time in the lower reservoir when Int407 cells were incubated with the conditioned medium of control clone. Overall, independently of mutations, the transcytosis of recombinant BSDL after 3 h of incubation was inhibited by ϳ50% when heparin was co-incubated with the enzymes in the apical reservoir (Fig. 5B). At the end of the incubation of Int407 cells with recombinant variants of BSDL, the cell epithelium was exhaustively washed (12), and then the cells were scraped and lysed, and BSDL present in cell lysate was finally determined by recording the enzyme activity. As shown in Fig. 5C, the activity of BSDL was not different in cell lysate of Int407 cells incubated with recombinant variants of the enzyme expressed by clones 3B, 3M2, and 5M5 and was decreased by heparin. This result indicated that the amount of enzyme taken up by Int407 cells is independent upon mutagenized residues and inhibited by heparin.
Catalytic Activity of the Recombinant Variants of BSDL-
The activity of the wild-type (clone 3B), K61I/K62I/R63A (clone 3M2), and K32I/K56I/K61I/K62I/R63A (clone 5M5) recombinant variants of BSDL was recorded in the presence of 4 mM NaTC as a function of the 4-NPC concentration. The results showed that mutations of either Lys-Lys-Arg 63 or of Lys-Lys-Arg 63 , Lys 32 , and Lys 56 did not alter the hydrolytic activity of recombinant enzymes on this water-soluble substrate (Fig. 6). The double-reciprocal plot indicated that the affinity constant and the maximal velocity for 4-NPC hydrolysis was of the same order of magnitude for each recombinant variant.
The bile salt-dependent hydrolysis of cholesterol [ 14 C]oleate was also recorded, and analysis of the enzyme kinetic data (Fig.  7) revealed that the major difference between the mutants and the wild-type enzyme resided in the maximal velocity, which significantly decreased with the number of mutations, the K32I/K56I/K61I/K62I/R63A (clone 5M5) mutant being much less active than the K61I/K62I/R63A (clone 3M2) mutant, which is itself much less active than the wild-type enzyme (clone 3B). The affinity constant for cholesteryl [ 14 C]oleate showed no significant difference between the wild-type and mutagenized enzymes (approx. 10 M).
Activation of the Esterolytic Activity of Mutagenized and Wild-type BSDL by Bile Salts-We next examined the effect of primary and secondary bile salts on the esterolytic activity of mutagenized and wild-type BSDL. As shown in Fig. 8A, increasing concentrations of the primary bile salt NaTC also enhanced the activity of recombinant variants of BSDL. Although activation kinetics of the wild-type BSDL and of the K61I/K62I/R63A mutant of BSDL were parallel, that of the K32I/K56I/K61I/K62I/R63A mutant differed a little and appeared biphasic. At NaTC concentrations below 150 M (Fig.  8A, inset), the K32I/K56I/K61I/K62I/R63A BSDL mutant activity on 4-NPC was low. Increasing the NaTC concentration led to a higher activity, which reached that of the wild-type BSDL and K61I/K62I/R63A BSDL mutant for NaTC concentrations close to 500 M. Then the activity of all recombinant variants of BSDL remained similar with NaTC concentrations up to 4 mM (Fig. 8A). When examining the effects of a secondary bile salt such as NaTDC, no difference in activation kinetics can be recorded between recombinant variants of BSDL (Fig. 8B) even at concentrations below 150 M (Fig. 8B, inset). Clearly the maximal velocity of BSDL is reached above the CMC (CMC ϭ 0.5 mM) of NaTDC (Fig. 8B), whereas this maximal activity is obtained below the CMC (CMC ϭ 1.4 mM) of NaTC (Fig. 8A).
These data indicated that the interaction of NaTC with the K32I/K56I/K61I/K62I/R63A BSDL mutant differed from that of K61I/K62I/R63A mutant and wild-type BSDL and suggested that mutagenized residues are implicated in the binding of NaTC to the enzyme.
Activation of the Esterolytic Activity of Mutagenized and Wild-type BSDL by Acid Phospholipids-In the light of the cationic nature of the mutagenized residues, we wondered whether the N-terminal basic cluster may not be involved in the binding of acidic lipids to BSDL. We therefore investigated the effect of phosphatidylserine, phosphatidylinositol, and phosphatidic acid, all acid phospholipids, and phosphatidylcholine (a neutral-charged phospholipid) and compared this effect with that of NaTC. For these experiments, the activity of recombinant BSDL was recorded on 4-NPC in the presence of each phospholipid (125 M), NaTC or NaTDC. As depicted in Fig. 9, clearly and as already shown, NaTC (4 mM) activated both wild-type and mutagenized recombinant BSDL to the FIG. 3. Northern blot analyses. RNA extracted from CHO cells expressing the wild-type enzyme (clone 3B), expressing the K61I/K62I/R63A BSDL mutant (clone 3M2), and expressing the K32I/K56I/ K61I/K62I/R63A BSDL mutant (clone 5M5) and from control clone C1 were separated on 1% agarose gel and transferred onto a nitrocellulose membrane. The membrane was then hybridized with specific probes for BSDL (left) and ␤-actin (right). Alternatively, RNA was dotted onto a nitrocellulose membrane in decreasing rank amounts from 5 to 0.2 g/ well using a Bio-Rad device. The membranes were then probed as above. The relative abundance of mRNA encoding BSDL and ␤-actin was estimated after the quantification of the dark intensity of each spot obtained on the autoradiogram using the National Institutes of Health program (Bethesda, MD) by comparing the slope of the regression line of the dark intensity (in arbitrary units) versus the amount (in g) of RNA. E, RNA extracted from clone 3B; ‚, RNA extracted from clone 3M2; ▫, RNA extracted from clone 5M5; X, RNA extracted from control clone C1.

FIG. 4. Heparin-Sepharose affinity chromatography of recombinant variants of BSDL.
Serum-free conditioned medium of CHO cells expressing the wildtype enzyme (clone 3B), expressing the K61I/K62I/R63A BSDL mutant (clone 3M2), and expressing the K32I/K56I/K61I/ K62I/R63A BSDL mutant (clone 5M5) was applied to a heparin-Sepharose column (3-ml wet gel, approx. 300 -400 ϫ 10 Ϫ3 units) and agitated for 16 h at 4°C. The gel was then allowed to settle, and the column was washed with 10 mM Tris-HCl buffer, pH 7.0, containing 50 mM NaCl. Proteins bound to the affinity column were eluted with a NaCl gradient of 50 -500 mM in Tris-HCl buffer, pH 7.0. The elution profile was monitored by recording the BSDL activity on 4-NPC in the presence of 4 mM NaTC. E, BSDL expressed by clone 3B; ‚, BSDL expressed by clone 3M2; ▫, BSDL expressed by clone 5M5. same extent, whereas at 125 M, this bile salt poorly activated the K32I/K56I/K61I/K62I/R63A BSDL mutant (compare with the activity in the absence of bile salt). Independently of its concentration, NaTDC activated BSDL mutants and the wildtype enzyme to the same extent. When examining the effect of phospholipids on the esterolytic activity of BSDL, anionic phos- At the end of the incubation of Int407 cells with BSDL, cells were exhaustively washed, harvested, and lysed, and the amount of BSDL taken up by Int407 cells was determined by recording the enzyme activity (C). In cell lysate, the BSDL activity was defined as the difference of activity on 4-NPC recorded in the presence and in the absence of NaTC (23). Results are averages of two independent determinations. pholipids phosphatidylserine, phosphatidylinositol, and phosphatidic acid (125 M) significantly enhanced the wild-type BSDL activity on 4-NPC to a significant value approximately half that promoted by NaTC at 4 mM, whereas the effect of zwitterionic phosphatidylcholine (125 M) is much less significant and close to the effect of NaTDC at the same concentration. The activating effect of acid phospholipids on mutagenized variants of BSDL is less pronounced, and interestingly, the more basic charges were substituted, the less the BSDL activity is increased by anionic phospholipids. The effect of phosphatidylcholine on 4-NPC hydrolysis was not different between recombinant mutants and wild-type BSDL.

DISCUSSION
Two site-directed mutants of BSDL were constructed to define the functionality of the N-terminal basic cluster or putative heparin-binding site consisting of residues Lys 32 , Lys 56 , Lys 61 , Lys 62 , and Arg 63 . Site-directed mutagenesis was used to alter some amino acids involved in binding properties, and a limitation of this method is that there exists a possibility that the amino acid substitution causes a loss of function by altering the structure of the protein. Informative elements such as preservation of enzymatic activity suggest that the overall structure has not been perturbed, but one cannot state with certainty that this is the case. Despite this limitation and in the absence of information such as x-ray crystallography, site-di-rected mutagenesis is a useful tool for identifying potential structural features of proteins.
In this study, five clones bearing three substitutions (K61I, K62I, and R63A) and three clones with five mutations (K32I, K56I, K61I, K62I, and R63A) were obtained after selection with G418 and compared with the clone 3B expressing the wild-type BSDL (17). The activity of recombinant wild-type and mutant BSDL always paralleled the amount of enzyme detected by Western blotting in corresponding clones, demonstrating that the overall structure of the enzyme was not significantly altered by substitutions. Each selected clone expressing BSDL also secreted BSDL, and among these, clone 3B expressing the wild-type enzyme, clone 3M2 expressing the K61I/K62I/R63A BSDL mutant, and clone 5M5 expressing the K32I/K56I/K61I/ K62I/R63A BSDL mutant that displayed similar amounts of BSDL mRNA, of BSDL protein, and of BSDL activity also secreted the enzyme at the same rate. These results demonstrated that BSDL is expressed and secreted by either clone and bring evidence that the substitutions in part of the sequence Lys-Lys-Arg 63 , which is close to the Cys 64 residue that forms a bridge with Cys 80 (6), do not interfere with the folding and the degradation of BSDL (1, 24).
The putative heparin-binding domain of BSDL was predicted from analyses of the primary amino acid sequence of the enzyme based on comparison with the consensus sequence of other heparin-binding proteins (25). However, once subjected to site-directed mutagenesis, no modification in heparin binding as determined on Sepharose-immobilized heparin can be recorded with the K61I/K62I/R63A BSDL mutant and with the K32I/K56I/K61I/K62I/R63A BSDL mutant as compared with the wild-type recombinant enzyme. This result agrees with that of Liang et al. (9) on the R63A mutant of BSDL. Furthermore, recombinant BSDL mutants were still capable of transcytosis throughout Int407 intestinal cells. This motion likely necessitates a first interaction of the enzyme with heparanoids lining the cell brush border, and heparin, which inhibited this interaction (10), also decreased the uptake and transcytosis of recombinant mutants of BSDL, as shown here. Consequently, the N-terminal cationic cluster of BSDL, including Lys 32 , Lys 56 , Lys 61 , Lys 62 , and Arg 63 , might not be the major BSDL domain implicated in heparin binding. On this enzyme and beside the N-terminal cationic cluster located near the enzyme catalytic site (7,8), another positive area was detected in the C-terminal primary sequence of BSDL. In the human enzyme, this cluster consists of six positively charged residues, five of which are also conserved in all known BSDL (13). This positive area could facilitate the binding of the BSDL dimer to anionic micelles (6). Another possibility is that this cluster helps BSDL (possibly a dimer) to anchor to heparin molecules lining the brush-border membrane. The localization of this cluster, remote from the catalytic site, may leave the enzyme totally functional and agree with the results of Spilburg et al. (26), showing that the interaction of BSDL with membrane-associated heparin enhanced the enzyme activity on micelles of NaTC-cholesteryl oleate. Therefore it could be that the C-terminal positive cluster of BSDL represents the major functional site for binding to immobilized or membrane-associated heparin. Multiple cationic sites susceptible to bind (or not) to heparin are also present in lipoprotein lipase and hepatic lipase sequences (27)(28)(29).
From this point, the functional significance of the N-terminal basic cluster of BSDL remains an open question. The position of the N-terminal basic cluster, proximal to the active site (7), and its possible overlap with a site for bile salt binding lead us to investigate the catalytic properties and activation of mutagenized BSDL. The activity of the two BSDL mutants on the water-soluble 4-NPC substrate in the presence of 4 mM NaTC is not altered as compared with the wild-type BSDL. However, the maximal velocity of the enzyme on cholesteryl oleate in bile salt micelles decreased with the number of mutations without affecting the affinity constant. This latter result confirms recent data showing that the substitution of Arg 63 (R63A mutant) affected the enzyme activity on micellar cholesteryl esters (9). Recording the effect of bile salt on the esterolytic activity of BSDL showed that NaTC at concentrations below the CMC (125 M), which do bind to the specific site (2, 3), did not activate the K32I/K56I/K61I/K62I/R63A BSDL mutant, whereas micellar NaTC (at concentrations above the CMC, i.e. Ͼ 1. 4 mM) activated K61I/K62I/R63A and K32I/K56I/K61I/ K62I/R63A mutants to the same extent as the wild-type enzyme. Further, NaTDC, a secondary bile salt, which likely did not interact with the specific bile salt-binding site of BSDL (3, 4), equally activated both mutants and wild-type BSDL, reach- ing the maximal velocity above the CMC of NaTDC (i.e. Ͼ 0.5 mM). These results demonstrated that the interaction of NaTC, and not that of micellar bile salts, with BSDL is affected by amino acid substitutions in the N-terminal basic cluster. Consequently, this cluster likely represents the specific bile saltbinding site of BSDL susceptible to bind NaTC at concentrations below the CMC (3).
BSDL was also detected in blood (30), where it is associated with atherogenic light density lipoproteins (31). BSDL is synthesized to a limited extent by human-monocytes-macrophages (32), endothelial cells (33,34), eosinophils (35), and the liver (22,36,37). BSDL has also been localized in necrotic areas of the pancreas, consecutive to an acute pancreatitis (38). The presence of BSDL in such a wide range of tissues and organs, normal or pathological, suggests a broad physiological function in the body. The underlying question about the physiological function of BSDL outside the duodenum concerns the activation of the enzyme, which in the absence of bile salt cannot hydrolyze lipid substrates (2,39). We, therefore, have extended the effect of anionic bile salts to acidic phospholipids that have been shown to regulate neutral cholesterol esterase of alveolar macrophages (14). As shown here, anionic phospholipids such as phosphatidylserine, phosphatidylinositol, and phosphatidic acid, contrary to the zwitterionic phosphatidylcholine, are able to enhance the activity of the wild-type BSDL to a value higher than that of NaTC at the same concentration. Furthermore, the two mutants of BSDL used here were less and less activable with an increasing number of mutagenized amino acids. These data suggest that the N-terminal cationic cluster of BSDL could be, in fact, not only a specific bile salt-binding site but more generally a cationic regulatory site capable of accommodating anionic ligands. Occupancy of this site by soluble heparin may explain the inhibition promoted by this ligand on the human BSDL activity on NaTC-cholesteryl oleate micelles (26). The presence of this regulatory site on BSDL could be physiologically relevant as it may be involved in regulating the enzyme activity once, for example, in the atherosclerotic lesions of the vascular endothelium, where both BSDL 2 and acid phospholipids (40) were detected.