A Surface Loop Covering the Active Site of Human Pancreatic Lipase Influences Interfacial Activation and Lipid Binding"

The distinguishing feature of lipases is their increased activity at an oil-water interface, termed interfacial activation. X-ray crystallography of lipases suggested a mechanism for interfacial activation by revealing conformational changes in several surface loops that cover the active site. In one conformation, these loops pre- vented substrate from entering the active site, and, in the other conformation, movement of the loops opened the active site. We tested the role of the major surface loop, the lid domain, in human pancreatic lipase (hPL) function by creating deletions in this region and ex- pressing the mutant proteins in baculovirus-infected insect cells. The mutants were tested for activity against tributyrin and triolein, colipase interaction, interfacial activation, and binding to tributyrin. The purified mutants had decreased activity against both tributyrin and triolein compared to wild-type hPL and did not show a preference for either substrate. Although colipase was required for maximum activity in the presence of bile salts, the mutants had significant activity against tribu- tyrin, but not triolein, in the absence of colipase. Both mutants were active against monomers of tributyrin demonstrating that they did not require an interface for activity. Finally, both mutants had decreased binding to tributyrin particles. These results suggest that the lid domain in hPL mediates interfacial activation and influ- ences interfacial binding. infeded Beckman Purification of Recombinant Lipases-Both mutants and hPL were purified from the harvested insect cell medium. Wild-type hPL was purified by immunoafinity chromatography.' Preliminary experiments that the two mutants bound and eluted from the immu- noafinity column, but were inactivated by the elution buffer. Thus, an alternative purification method was developed. Benzamidine, 2 mM, and phenylmethylsulfonyl fluoride, 0.25 mM, were added to the clarified medium, and the medium was dialyzed overnight against 15 liters of 10 mM Tris-HCI, pH 8.0, 2 mM benzamidine, and 0.25 mnl phenylmethyl- sulfonyl fluoride. The was applied to a DEAE-Blue Sepharose column (200-ml bed volume) and washed through with 10 mM Tris-HCI, pH 8.0. Both mutants were in the pass-through. The pH of the pass-through was adjusted to 6.2 with 0.5 M succinate and loaded onto a CM-Sepharose column equilibrated in 30 mM Tris succinate, pH 6.2. The column was washed with equilibration buffer and eluted with a salt gradient from 0 to 300 mM NaCl in the equilibration buffer. Lipase was located by activity against triolein. The active fractions were pooled and concentrated by ultrafiltration over an Amicon YM30 membrane. spectrophotometrically an E,, 10.0 nm Bradford with bovine serum albumin as

The absorption and assimilation of dietary fats depends on the actions of multiple lipolytic enzymes (1,2). Many of these lipases have homologous protein and gene structures suggesting that lipases form a gene family evolved from a common ancestral hydrolase (3,4). The archetype of this family is pancreatic triglyceride lipase (PL),' and information about PL has been extrapolated to other lipase family members. For instance, experiments based on earlier PL studies produced new knowledge about hepatic and lipoprotein lipases (5)(6)(7)(8).
Human pancreatic triglyceride lipase (hPL) has the characteristic and intriguing property of preferring water-insoluble substrates that form oil-water interfaces over water-soluble substrates (9). In dilute solutions of short-chain triglycerides where monomers predominate, hPL displays little activity. Once the concentration of triglyceride exceeds its solubility, particles form and the reaction velocity increases dramatically, a property termed interfacial activation. *This work was funded by National Institutes of Health Grant DK42120 and was done during the tenure of an Established Investigatorship from the American Heart Association (to M. E, L.). The costs of charges. This article must therefore be hereby marked "advertisement" publication of this article were defrayed in part by the payment of page in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Three-dimensional structures of lipases have demonstrated that the active sites of hPL and of several fungal lipases are covered by peptide loops that would prevent diffusion of substrate into the active site unless these loops move (10, 11). Movement of the loops was observed in hPL co-crystallized with procolipase and mixed micelles of phospholipid and detergent (12). Similar movements were seen in a fungal lipase crystallized with an irreversible inhibitor and were fortuitously observed in the crystal structure of another fungal lipase (13,14). These groups proposed that the movements produced the phenomenon of interfacial activation (12, 14, 15).
One of the loops in hPL, a surface helix of 23 amino acids bounded by CysZ3' and CysZ6', the lid domain, adopted a markedly different conformation when substrate or inhibitor was present in the crystals (12,16). The new position of the lid opened the active site, created a hydrophobic surface, allowed movement of a second loop, the p5 loop, which interacted with the lid in the closed conformation, and permitted new contacts between the open lid and colipase. The description of a guinea pig lipase with a 4-amino acid lid domain suggested that the lid domain did prevent diffusion substrate into the active site (17). The enzyme was highly active against monomeric substrates and did not exhibit interfacial activation consistent with the hypothesis that movement of the lid domain constitutes interfacial activation.
The participation of the lid domain in interfacial activation has not been tested directly by creating mutations in the lid domain and determining the effects on interfacial activation. Mutations have been introduced into the putative lid domain of lipoprotein lipase (5,6). The authors demonstrated effects of the mutations on activity and substrate specificity, but did not measure interfacial activation.
We tested the contribution of the hPL lid domain to interfacial activation and lipid binding by introducing deletions into the cDNA encoding hPL and expressing the mutant lipases in COS-1 cells or baculovirus-infected insect cells. The mutants were tested for activity against tributyrin and triolein. Their sensitivity to bile salts and interaction with colipase was determined, and interfacial activation was measured. EXPERIMENTAL PROCEDURES Construction of Mutants-The previously described hPL cDNA was the template for mutagenesis (18). The amino acid numbering is based on the hPL amino acid sequence. Two deletions were introduced into the region between Cys238 and CysZ6' by the polymerase chain reaction overlap extension method (19,20). Amino acids LysZ4' to Alaz6" were deleted in one mutant, and residues Aspz4' to ArgS7 were deleted in the other mutant. The polymerase chain reaction products were subcloned into pGEM3Z, and the sequence was confirmed by the dideoxynucleotide chain termination method according to the Sequenase protocol. Each mutant was then subcloned into pVL1392 for preparing recombinant baculovirus for infecting insect cells. Plasmid DNA was prepared with Qiagen columns according to the manufacturer's instructions.
In&tion of Insect Cells with Recombinant Baculouirus-Fkcombinant baculovirus was prepared as described previously, and Sf9 or Hi-5 insect cells grown in spinner flasks with serum-free medium (Ex-Cell 400) were infeded with the virus (21). The medium was harvested 3-4 days postinfection by centrifugation at 5000 rpm for 10 min in a Beckman 52-21 centrifuge with a JA-20 rotor.
Purification of Recombinant Lipases-Both mutants and hPL were purified from the harvested insect cell medium. Wild-type hPL was purified by immunoafinity chromatography.' Preliminary experiments demonstrated that the two mutants bound and eluted from the immunoafinity column, but were inactivated by the elution buffer. Thus, an alternative purification method was developed. Benzamidine, 2 mM, and phenylmethylsulfonyl fluoride, 0.25 mM, were added to the clarified medium, and the medium was dialyzed overnight against 15 liters of 10 mM Tris-HCI, pH 8.0, 2 mM benzamidine, and 0.25 mnl phenylmethylsulfonyl fluoride. The medium was applied to a DEAE-Blue Sepharose column (200-ml bed volume) and washed through with 10 mM Tris-HCI, pH 8.0. Both mutants were in the pass-through. The pH of the passthrough was adjusted to 6.2 with 0.5 M succinate and loaded onto a CM-Sepharose column equilibrated in 30 mM Tris succinate, pH 6.2. The column was washed with equilibration buffer and eluted with a salt gradient from 0 to 300 mM NaCl in the equilibration buffer. Lipase was located by activity against triolein. The active fractions were pooled and concentrated by ultrafiltration over an Amicon YM30 membrane.
Analysis of Lipases-SDS-PAGE and immunoblot were accomplished as described previously (22). Lipase activity was measured against ['Hltriolein as described or by a pH-STAT technique. For potentiometric measurements, the sample was added to 1 mM Tris-HCI, pH 8.0,0.15 M NaCI, 2 mM CaCI,, and 4 mM taurodeoxycholate. The assay substrate was prepared by adding 0.5 ml of tributyrin to 15  Lipase concentration in the medium was determined by immunoblot as described (22). The concentration of the lipases purified from insect cell medium was determined spectrophotometrically using an E,, = 10.0 at 280 nm and by Bradford assay with bovine serum albumin as a standard.

RESULTS
Expression of hPL and the Lid Domain Mutants-Two mutants were created in the region of the lid domain. One, the 240-260 deletion mutant, removed virtually all of the lid domain to test the importance of the domain in lipase function. The second mutation, the 247-258 deletion mutant, was designed to remove the a-helix covering the active site and to minimize disruption of the remaining lid domain residues.
Both lid domain mutants and hPL were expressed in recombinant baculovirus-infected insect cells. Earlier, we and others demonstrated that milligram quantities of wild-type PL could be expressed in this system (21,23). After infecting insect cells with the recombinant baculovirus, the presence of the lid domain mutants in the medium was demonstrated by SDS-PAGE and immunoblot (data not shown.) Larger scale infections were done, and 3 to 5 mg of each protein was isolated from the medium. The isolated proteins were homogeneous by SDS-PAGE analysis and protein staining (Fig. 1). These purified recombinant proteins were used for further characterization of the role of the lid domain in lipolysis.
Activity of the Mutant Lipases-The lipase activity of the recombinant lipases was determined in the presence of colipase and taurodeoxycholate by the pH-STAT technique with tributyrin, a short-chain triglyceride, as substrate (Fig. 2 A ) . Lipolytic activity was readily detectable and linear over a t least 5 min demonstrating that the insect cells made functional lipases. With longer incubation times, the reaction rate for the 248-257 deletion decreased (inset, Fig. 2 B ) . Because only 2% of the available fatty acids were released, substrate depletion is not an adequate explanation for the decreasing activity. Alternatively, decreased stability of the lid domain mutants or product inhibition could explanation the fall off in activity.
To determine if the lid domain mutants had decreased activity against long-chain triglycerides, we tested the lid domain M. L. Jennens and M. E. Lowe, submitted for publication. mutant activity against ["Hltriolein. A previous study demonstrated that deleting the lid in lipoprotein lipase increased activity against tributyrin and greatly decreased activity against triolein (6). The authors speculated that the lid domain influenced substrate binding and specificity. hPL and both mutants had activity against triolein, in contrast to the result with lipoprotein lipase (Fig. 3). Interestingly, the lid deletion mutants, but not wild-type hPL, had a long lag phase with triolein that was not seen with tributyrin. The assay system did not influence the lag phase because the mutants also had an exaggerated lag phase with the pH-STAT technique and triolein (data not shown).
The specific activities of hPL and the two mutants for tributyrin and triolein were determined by the pH-STAT method to detect marked differences in activity against either substrate. Regardless of the substrate, the deletion mutants had lower specific activities than wild-type hPL when colipase was in the assay ( Table I). The decrease in specific activity was about &fold for the 248-257 deletion mutant for both tributyrin and triolein. Likewise, the specific activity of the 240-260 deletion mutant decreased about 17-fold for both substrates. Neither mutant had a significant preference for either substrate indicating that the lid domain may not contribute to or influence the substrate binding pocket.
Although maximum activity of the wild-type hPL and the lid deletion mutants required colipase for both substrates, the mutants had significant activity against tributyrin in the absence of colipase (Table I). In contrast, detectable hydrolysis of triolein required colipase. When tributyrin was the substrate, colipase increased lipase activity 22-fold for wild-type hPL compared to just 1.5-fold for the lid deletion mutants demonstrating a marked difference in bile salt inhibition of tributyrin hydrolysis between the mutants and wild-type hPL.
Interaction with Colipase-One potential explanation for the decreased activity of the lid domain mutants is decreased binding of the mutant lipases to colipase. To test this possibility, we initially attempted to measure colipase binding to the lid domain mutants by several previously published methods and obtained inconsistent results, presumably because of the weak binding of colipase and hPL in free solution. Consequently, we employed a competition assay to determine if the 240-260 lid deletion mutant could inhibit wild-type hPL activity against triolein. If wild-type hPL and the 240-260 deletion were added to the same reaction, triolein hydrolysis during short incubations would be a result of the wild-type hPL activity and not the mutant because of the mutant's lag phase. Activity would only be inhibited by molar excesses of the mutant if the mutant bound colipase with the same affinity as the wild-type hPL.  Activity would be normal or minimally decreased if the mutant had significantly decreased affinity for colipase. The assay conditions included a large excess of substrate, suboptimal colipase concentrations, and a short incubation period. hPL had about 30% of maximal activity under these conditions. Addition of the 240-260 deletion mutant inhibited lipase activity in a dose-dependent fashion (Fig. 4). Adding a 2-fold molar excess of colipase overcame the inhibition suggesting that the 240-260 deletion mutant inhibited wild-type hPL by competing for colipase binding and that the interaction of the mutant with colipase was not vastly different from the interaction with the wild-type hPL (Fig. 4). Furthermore, the interaction must occur quickly and could not account for the lag phase seen with the lid deletion mutants.

Interfacial Activation of Human Pancreatic Lipase
Binding of the Lid Domain Mutants to Substrate-Another possible explanation for decreased activity of the mutant lipases is decreased binding affinity of the mutants for the substrate interface. Tributyrin, which can be dispersed in aqueous solutions more readily than longer-chain triglycerides and easily separated from the aqueous phase by centrifugation, presents an interface that binds native porcine PL (24). This system was used to directly measure binding of the lid deletion mutants to a substrate interface (Table 11). Short incubations with tributyrin dispersed in buffer followed by centrifugation removed virtually all of wild-type hPL from the aqueous phase, but most of the lid deletion mutants remained in the aqueous phase. Binding of hPL was inhibited by the presence of taurodeoxycholate and was restored if colipase was added. A larger percentage of the 248-257 deletion mutant bound to the lipid phase if colipase was present, but the extent of binding was still less than for wild-type. These results demonstrated decreased binding of the lid domain mutants to interfaces and suggested that the lid domain forms all or part of the binding site.
Interfacial Actiuation-Interfacial activation was determined by measuring activity of the mutants over a range of tributyrin concentrations. The activity of wild-type hPL increased sharply when tributyrin reached its saturation point demonstrating interfacial activation (Fig. 5). Both mutants reached maximal activity at tributyrin concentrations below the saturation point indicating that they hydrolyzed monomeric substrate and did not require an interface for activation (Fig. 5 ) . These findings are consistent with participation of the lid domain in interfacial activation.
Bile Salt Inhibition of the Lid Domain Mutants-To characterize the bile salt inhibition of the lid domain mutants, we measured the activity of the lid mutants at various taurodeoxycholate concentrations with tributyrin as the substrate (Fig.  6). The mutant's activity against monomers should have obscured an effect of bile salts as reported for the guinea pig lipase, but bile salts had clear effects on the mutants activity. Low concentrations of taurodeoxycholate stimulated the mutants and the wild-type hPL (17). Higher concentrations of bile salts inhibited the lid deletion mutants to a much greater extent than the inhibition of wild-type hPL. The lid deletion mutants had almost 5-fold higher activity at 0.5 m M taurodeoxycholate than at the standard assay concentration of 4 mM taurodeoxycholate. The increased sensitivity of the lid domain mutants to bile salt inhibition suggests that the mutants are more susceptible to denaturation by bile salts or to displacement from substrate by bile salts. DISCUSSION We constructed and expressed two mutant lipases with deletions in a surface loop between Cys238 and Cys2'j2, the lid domain, that covers the active site and blocks access of substrate t o the binding pocket. Although infected insect cells secreted both mutants in amounts comparable to the wild-type hPL and both mutants had activity against triglyceride substrates, the lid domain mutants differed from the wild-type hPL in several important properties. The mutants showed decreased activity, decreased binding to an interface, increased sensitivity to bile salt inhibition, and, importantly, did not demonstrate interfacial activation.
The lid domain of lipases has been implicated by other studies in the mechanism of interfacial activation and in lipid binding of lipases (10-17). Our data provide direct evidence that the lid domain participates in the mechanism of interfacial activation. This observation is consistent with the hypothesis that the lid domain blocks access of substrate to the active site and that movement of the lid is required for activity. At least two mechanisms for inducing interfacial activation by lid domain movement are plausible. First, binding to an interface could trigger The three-dimensional structure of the procolipase-hPL complex in the open form revealed a continuous hydrophobic plateau formed by procolipase and the lid domain (12). The authors proposed that this hydrophobic surface is the lipid binding site of the complex. The formation of the surface by regions from both procolipase and hPL provides a potential explanation for the role of colipase in potentiating the binding of hPL to lipid interfaces. The decreased binding of the deletion mutants to tributyrin is consistent with the notion that the lid domain forms part of the lipid binding site. The preserved activity against triolein argues that the deletion mutants can bind to interfaces and that other regions of the procolipase-hPL complex participate in binding to interfaces.

Interfacial Activation of
Although the decreased binding of the lid domain mutants to a substrate interface is sufficient to explain the decreased activity, increased bile salt sensitivity, and long lag phase of the mutants, other explanations are also possible. The crystal structure of the open and closed forms of the procolipase-hPL complex demonstrated conformational changes in other regions of hPL, notably in the p5 loop (12). These changes constrain Phe7* in position to form the oxyanion hole and help stabilize the oxyanion intermediate. Because lid domain residues interact with residues in the p5 loop, deletion of lid domain residues may affect the position of the p5 loop and destabilize the oxyanion intermediate. Conformational changes in the p5 loop or in other regions of the active site could negatively affect substrate binding affinity or slow substrate turnover. Finally, the interaction with colipase may be inhibited by deletions in the lid domain.
Analysis of the three-dimensional structure of the procolipase-lipase complex co-crystallized with mixed micelles did reveal interactions of procolipase with lid domain residues, Val247, Ser"?44, and AsnZ4l (12). All 3 residues were preserved in the 248-257 deletion mutant and were absent in the 240-260 mutant. Yet, both mutants required colipase for activity against triolein and the 240-260 deletion mutant could effectively compete with hPL for colipase. Although neither of these experiments would detect small differences in the binding of colipase to the mutants, the binding was not critically affected by the deletions. In contrast, the inability of colipase to restore binding of the 248-257 deletion mutant to tributyrin suggested that the interaction of colipase with the mutant differed from the interaction of colipase with wild-type hPL. Thus, altered interactions between colipase and the mutants remains a possible explanation for the decreased activity and long lag phase of the lid domain mutants.
The results presented in this paper demonstrate the role of the lid domain, first postulated from the three-dimensional structures of lipases, in interfacial binding and interfacial activation. Movement of the lid domain away from the active site may be the only structural correlate of interfacial activation, but other structures may play a role. The x-ray data suggested that other loops also moved, perhaps, as a direct result of the lid domain movement. The effects of deleting portions of the lid domain on other regions of hPL is not clear from the crystal structures or the present studies. Removing the lid domain residues may induce the other loops to assume the positions Human Pancreatic Lipase they normally occupy in the open conformation or to assume positions that do not normally occur. The molecular basis for the changes induced in the lid deletion mutants requires additional detailed knowledge of the mutants' structure that can be provided by their crystal structure.