Colipase stabilizes the lid domain of pancreatic triglyceride lipase.

Pancreatic lipase is characterized by increased activity against water-insoluble substrates and by dependence on another protein, colipase, for binding to the substrate interface. In most models of pancreatic lipase activity, colipase functions to anchor lipase on the substrate interface. Recent studies of the x-ray crystal structure of the complex between colipase and lipase suggest another function for colipase in maintaining the active conformation of lipase. We tested this hypothesis by introducing mutations into colipase at position 15, a residue that contacts the lid domain lipase in the open conformation. Multiple mutant colipases were expressed and shown to have decreased activity. To further investigate the function of the interaction between Glu15 of colipase and lipase, we examined one mutant, E15R, in detail. This mutant had 175-fold less activity compared with wild-type colipase. Although E15R had decreased activity, it was as effective as wild-type lipase in anchoring lipase to mixed emulsions of bile salt and tributyrin. The importance of the interaction with the lid domain was tested by determining the activity of E15R with lid deletion mutants of lipase. E15R was as active as wild-type colipase with these mutant lipases. These results indicate that Glu15 is critical for activity of the colipase-lipase complex at an interface and that colipase has a function in lipolysis in addition to anchoring lipase to an interface. We propose that this function is to stabilize the lid domain of lipase in the open conformation, thereby facilitating lipolysis.

For a number of years, lipases and phospholipases have been models for the interaction of proteins with lipid-water interfaces like those found in cell membranes, in lipoprotein particles, or in emulsions of dietary fats. Interest in the function of lipolytic enzymes has increased in recent years because of the role that phospholipases play in intracellular signaling (1)(2)(3). The kinetic mechanism of lipolytic enzymes is distinguished from that of enzymes hydrolyzing water-soluble substrates in a homogenous milieu by an additional, distinct step in the catalytic mechanism, adsorption of the water-soluble lipase from the aqueous phase onto the water-insoluble substrate interface prior to hydrolysis (4). Most lipases accomplish the adsorption step readily, but there are several exceptions including lipoprotein lipase and the archetype of the lipase gene family, pancreatic triglyceride lipase (PTL). 1 Under physiological conditions, PTL is prevented from adsorbing to its substrate, dietary triglycerides, by bile salts and phospholipids (5). Another pancreatic exocrine protein, colipase, facilitates the adsorption of PTL to the substrate lipid-water interface.
Until recently, the only or predominant role of colipase in the lipolytic mechanism of PTL was believed to be anchoring of PTL to the substrate interface (6). An additional role was suggested by two papers reporting the three-dimensional structures of porcine colipase with human PTL obtained in the absence or in the presence of mixed micelles (7,8). Comparison of these two structures revealed a marked conformational change in a surface loop of PTL, termed the lid domain. In the absence of mixed micelles, the lid domain occupied a position that covered the catalytic site residues and prevented substrate diffusion into the active site. No interactions between the lid domain and colipase were present in this conformation. In the presence of mixed micelles, the lid domain moved into a markedly different position that opened the active site. Furthermore, this movement created new interactions between residues in the lid and colipase residues Glu 15 and Arg 38 (9). Both of these residues are conserved in colipase isolated from many species including dogfish, chicken, and various mammals, implicating a role for these residues in colipase function (10,11). The formation of these new interactions between colipase and the PTL lid domain indicated that colipase may function to stabilize the lid domain in the open, active conformation, thereby facilitating PTL activity. We directly tested the hypothesis that the interaction between colipase and the lid domain of PTL is functionally important by creating, expressing, and characterizing colipase mutants with amino acid substitutions at position 15.

MATERIALS AND METHODS
Construction of the Mutants-All manipulations of DNA were done by standard methods (12). Mutations were introduced into the cDNA encoding human procolipase by the polymerase chain reaction (PCR) overlap extension method (13). Primers containing a specific mutation were used to produce the E15R and the E15Q mutants. The other mutants were produced by using oligonucleotide primers that were a mixture of all possible combinations of nucleotides at positions 1 and 2 of the codon encoding the residues at position 15. In this last instance multiple subclones of the PCR product were sequenced to identify the other mutants. Each mutant was fully sequenced by the dideoxynucloetide method with Sequenase following the manufacturer's instructions. The mutant cDNAs were subcloned into pSVL for transfection of COS-1 cells and into pVL1392 for expression in Sf9 cells.
Expression of Recombinant Proteins-All of the mutant colipases were expressed in COS-1 cells transfected by the DEAE method as described previously for pancreatic lipase mutants (13). The production of recombinant baculovirus and expression in Sf9 cells was done as described for wild-type colipase (14). Purification was done on an immunoaffinity column as described (14).
Analysis of Recombinant Protein-The medium from COS-1 cells was * This work was supported by National Institutes of Health Grant DK42120 and American Heart Association Grant 96009490 and was done during the tenure of an Established Investigatorship from the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. analyzed by SDS-PAGE followed by immunoblot with a polyclonal antibody to human colipase. Purity of the purified, recombinant colipase was assessed by SDS-PAGE. The CD spectrum was recorded an a Jasco J-600 spectropolarimeter in the far ultraviolet region of the CD spectrum, 190 -250 nm. Colipase was dissolved in water at a concentration of 0.01 mM.
Colipase Assays-The activity of colipase and the mutants was done in a Radiometer VIT 90 pH-STAT or in a radioactive assay as described for lipase (14). Either recombinant or native human pancreatic triglyceride lipase was added at the concentrations given in the figure legends.
Binding to Tributyrin-Binding assays were done in 0.1 M Tris-Cl, pH 8.0, 0.1 M NaCl, 2 mM CaCl 2 , and the stated concentration of taurodeoxycholate. Tributyrin was added to the binding buffer in a ratio of 0.5 ml of tributyrin to 14.5 ml of buffer, and the mixture was emulsified by homogenization and sonication. Triplicate assays were done in 1.7-ml microfuge tubes. 1 ml of buffer and the indicated amounts of colipase and lipase were added. The tubes were shook at room temperature for 1 min, and the phases were separated by centrifugation in a microfuge at high speed for 10 min. 500 l of the upper, aqueous phase was assayed by the titrametric method. The 100% value was determined in an identical tube containing buffer without tributyrin that was handled in parallel with the other samples.

RESULTS AND DISCUSSION
The functional importance of Glu 15 to the activity of PTL was tested by introducing mutations into this position. Initially, six mutants, E15G, E15D, E15Q, E15R, E15T, and E15W, were constructed and expressed in COS-1 cells (13). The medium of cells transfected with each mutant was examined by SDS-PAGE and immunoblot with rabbit polyclonal anti-human colipase antibody to determine if the mutant proteins were successfully synthesized and secreted (Fig. 1A). Each mutant was present in the medium, showing that mutations at Glu 15 did not grossly disrupt protein folding. The activity of each mutant was measured with both radioactive and titrametric assays (15). No activity was detected in any of the mutants, whereas activity in an equivalent amount of wild-type colipase activity was easily detected (data not shown). This initial screening suggested that preservation of glutamic acid at position 15 was critical for maintaining the activity of colipase.
Because kinetic and biophysical measurements of the Glu 15 mutants required larger amounts of protein than could be conveniently expressed in COS-1 cells, one mutant, E15R, was expressed in a baculovirus system to produce larger amounts of recombinant protein (14). This mutation introduced both a change in charge and side-chain length at position 15, which would maximally stress any interactions that may occur between this residues and the lid domain of PTL. Additionally, preliminary experiments revealed that this mutant was highly expressed in the baculovirus system. The expressed mutant was purified to homogeneity by immunoaffinity chromatography and run as a single band on SDS-PAGE (Fig. 1B). The secretion of E15R by Sf9 cells provided the initial evidence that the mutant was not misfolded. That the mutant had folded correctly was confirmed by determining the CD spectrum of the mutant (Fig. 2). The spectrum of E15R colipase was indistinguishable from the spectrum of wild-type colipase consistent with the proper folding of the mutant colipase.
To determine if E15R possessed activity, the mutant was tested over a broad range of concentrations (Fig. 3). At high concentrations, E15R reactivated PTL fully. But the concentration of the mutant to give half-maximal activity was 175-fold greater than for wild-type colipase, 1.7 ϫ 10 Ϫ9 M versus 3.0 ϫ 10 Ϫ7 M. The molar ratio of wild-type colipase to PTL at this concentration was 0.47, whereas the ratio of E15R to PTL was 83. Longer incubations, 60 min, at low concentrations of E15R did not give any activity, demonstrating that the low activity was not due to a long lag time. These findings suggested that the mutation of Glu 15 to Arg 15 decreased the affinity of the mutant colipase for PTL and implied that Glu 15 is critical for the interaction of colipase with PTL. The decreased interaction could affect PTL function through two possible mechanisms. First, the mutant colipase may interfere with opening of the lid domain. Second, the mutant colipase may not anchor PTL to mixed micelles.
The ability of E15R to anchor PTL to mixed micelles was tested directly. PTL was incubated with an emulsion of tributyrin and taurodeoxycholate in the absence or the presence of varying amounts of colipase, wild type, or E15R (15). The organic phase was separated from the aqueous phase by centrifugation, and the amount of PTL remaining in the aqueous phase was determined by activity measurements. When PTL was incubated with tributyrin alone, 10% or less remained in the aqueous phase. If taurodeoxycholate was also present, greater than 80% of the PTL was found in the aqueous phase (Fig. 4). If either colipase or E15R was included, the inhibition of binding by taurodeoxycholate was overcome in a concentrationdependent manner. The ability of E15R to anchor PTL to mixed micelles was slightly less than that of wild-type colipase, but the difference could not explain the large difference in activity between the two. This result suggested that E15R could interact with PTL at an interface and separated, for the first time, the effect of colipase on PTL activity from its effect on anchoring PTL.
Because Glu 15 was found to interact with a lid domain residue in the crystal structure of the colipase-PTL complex, we tested the activity of E15R with two lid domain mutants we had previously created and characterized (15). These mutants contained deletions in the lid domain. One deleted the ␣-helix at the end of the surface loop, residues 248 -257, retaining Asn 241 , which bonds to Glu 15 , and the other deleted the entire lid domain, including Asn 241 , from residue 240 to residue 260. Both of these mutants were active and required colipase for full activity, and importantly, both mutants had good activity  (13). An aliquot of the medium was analyzed by SDS-PAGE and immunoblot and was assayed for colipase activity. The immunoblot is shown, and the mutation is indicated over each lane. B, E15R colipase and wild-type colipase were expressed in a baculovirus system and purified by immunoaffinity chromatography. 5 g of purified protein was separated by SDS-PAGE and stained with Coomassie Blue. WT, wild type. FIG. 2. CD spectrum of wild-type and E15R colipase. Both wildtype and E15R colipase showed a positive band centered at 225 nm and a negative band at 192 nm. The spectra were similar to that reported for porcine colipase (16). WL, wavelength.
Colipase and the Lid of Pancreatic Lipase 10 against monomeric substrates (15). That is, they did not require an interface for activity, demonstrating that the catalytic site was available to substrate even in the absence of an interface that would trigger conformational changes in PTL. If E15R interfered with the opening of the lid domain preventing substrate access to the active site, then E15R should have relatively better activity with the deletion mutants that already have an accessible active site.
When we compared E15R to wild-type colipase with these mutants, E15R activity was about 2% of wild-type colipase with PTL compared with 20% with the short lid deletion, 248 -257, and 170% with the complete lid deletion, 240 -260 (Fig. 5A). We confirmed the increased activity of E15R with the 240 -260 deletion by measuring activity at various ratios of colipase to lipase (Fig. 5B). At each concentration, E15R was more active than wild-type colipase. The explanation for the increased activity of E15R with the lid domain mutant was not determined by these experiments and may be fortuitous. These results suggest that the decreased activity of E15R results from a detrimental interaction with PTL that hinders opening of the lid domain. Deletion of the lid domain residues removes that hindrance and permits a functional interaction of E15R with PTL that is consistent with the hypothesis that Glu 15 interacts with the lid domain residues and stabilizes the lid in the open, active conformation.
Pancreatic colipase functions to restore activity to bile saltinhibited PTL. Abundant experimental evidence has demonstrated that one function of colipase is to anchor PTL to the surface of the substrate (6). Recent structural data suggested an additional role for colipase in the lipolytic mechanism, stabilizing the open, active conformation of PTL (7,8). Our findings that E15R has decreased affinity for PTL, yet it has preserved the ability to effectively anchor PTL to the substrate, provide direct evidence for this hypothesis, and support a model of lipolysis that includes separate roles for colipase in PTL binding and activity. In this model, colipase-mediated binding of PTL does not require interactions with the lid domain, but for activity, PTL must move into the open conformation, presumably triggered by an interface, and form a stable, high affinity complex of colipase and lipase on the substrate interface. Once a complex is formed, colipase stabilizes the open conformation and both increases the affinity of PTL for the interface and allows catalysis to proceed. If the open conformation does not form, then PTL disassociates from colipase and the substrate. It is in the formation of a high affinity complex of colipase and lipase in the open conformation on an interface that is disrupted by mutations in Glu 15 .
FIG. 3. Activity of colipase and E15R colipase. Activity of colipase was determined by a pH-STAT assay with 2.5 g of recombinant human PTL in each assay (14). Closed circles, wild-type colipase; closed squares, E15R colipase.
FIG. 4. Colipase-mediated binding of PTL to tributyrin. PTL was incubated with emulsions of tributyrin and taurodeoxycholate and either wild-type or E15R colipase. After a 1-min incubation, the organic phase was separated from the aqueous phase by centrifugation (15). The amount of lipase remaining in the aqueous phase was determined by the pH-STAT method and expressed as a percentage of the total lipase activity added. The amount of E15R colipase that completely bound lipase did not inhibit the activity of native colipase with PTL in test reactions. Open bars, wild-type colipase; hatched bars, E15R colipase. The errors bars show standard deviation from three separate experiments. PL, pancreatic lipase.

FIG. 5. E15R activation of PTL lid domain mutants.
The lid domain mutants were described previously (15). A, activity was determined by the pH-STAT method with 2.5 g of the indicated lipase and a 2-molar excess of wild-type or E15R colipase. The results were expressed as a percentage of the activity obtained with each lipase and wild-type colipase. Open bars, wild-type colipase; closed bars, E15R colipase. The lipase assayed is given below each pair of bars. Standard deviation for four separate experiments is shown. B, the 240 -260 lid domain deletion mutant was assayed with varying amounts of wildtype or E15R colipase by the pH-STAT method. Closed circles, E15R colipase; closed squares, wild-type colipase. The error bars give standard deviation for three separate experiments. hPL, human pancreatic lipase.
Colipase and the Lid of Pancreatic Lipase 11