According to the two-phase model of dietary fat digestion, the concerted action of preduodenal and duodenal lipases digests dietary fats into sn-2 monoacylglycerols and free fatty acids (Hofmann and Borgstrom 1962 and 1964). These products are solubilized with bile salts to form mixed micelles that facilitate the absorption of the fatty acids across the enterocyte membrane. In humans, digestion begins in the stomach where gastric lipase, an acid lipase synthesized by gastric chief cells, releases about 15% of the fatty acids (Carrière et al. 1993). Lipases, secreted by pancreatic acinar cells, complete fat digestion in the proximal small intestine.

Three of the lipases expressed by the pancreas have highly homologous structures. One of these, colipase-dependent triglyceride lipase is clearly essential for efficient dietary fat digestion as evidenced by the steatorrhea present in patients with congenital pancreatic triglyceride lipase deficiency (Figarella et al. 1980, Ghishan et al. 1984). The exact proportion of acyl chains released by pancreatic triglyceride lipase is not clearly defined, but data in patients with congenital pancreatic triglyceride lipase deficiency suggest that pancreatic triglyceride lipase digests at least 50% of dietary triglycerides. The other two homologous proteins were recently described and have been called pancreatic lipase-related proteins 1 and 2 because of their resemblance to pancreatic triglyceride lipase (Giller et al. 1992, Grusby et al. 1990, Payne et al. 1994, Wishart et al. 1993). The roles of pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 in fat digestion are not known, but their description and characterization along with the determination of the tertiary structure of pancreatic triglyceride lipase and the characterization of recombinant pancreatic triglyceride lipase mutants have provided insights into the biochemical mechanism of lipolysis and the physiology of dietary fat digestion.

LIPASE GENE FAMILY

The subsequent characterization of the genes encoding these lipases increased the likelihood that the genes have a common ancestral gene. Alignment of the exon-intron structure for the human genes encoding pancreatic triglyceride lipase, hepatic lipase and lipoprotein lipase showed conservation of the exon-intron divisions, which indicated that the genes are related (Kirchgessner et al. 1989, Sims et al. 1993). Furthermore, an Alu element, a repetitive DNA sequence, was present in the corresponding introns of pancreatic triglyceride lipase and lipoprotein lipase. Classification of the Alu sequences suggested that the genes diverged within the last 40-50 million years (Chuat et al. 1992, Sims et al. 1993).

The description of the human pancreatic triglyceride lipase-related proteins added two new members to the lipase gene family (Giller et al. 1992). Pancreatic triglyceride lipase-related protein 1 was 68% identical to pancreatic triglyceride lipase and the related protein 2 was 65% identical to pancreatic triglyceride lipase (see below). Additionally, the gene encoding dog pancreatic triglyceride lipase-related protein 1 has an exon-intron organization identical to that of pancreatic triglyceride lipase (Mickel et al. 1989). The presence of these genes, so closely related to the gene encoding pancreatic triglyceride lipase, clearly established the presence of a lipase gene family.

PHYSIOLOGY

The presence of mRNA does not ensure that the mRNA is translated and the protein is expressed. The protein must be detected by other means. Both pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 were detected in the medium of AR42J cells, a cell line with acinar cell properties, suggesting that these homologues may also be present in pancreatic secretions (Fig. 1) (Kuhlman et al. 1996). Pancreatic secretions have not been examined for the presence of pancreatic triglyceride lipase-related protein 1 protein by any methods; thus, it is not known if pancreatic triglyceride lipase-related protein 1 is secreted by the pancreas. Pancreatic triglyceride lipase-related protein 2 protein was detected by immunoblot in secretions obtained from a cannulated rat pancreatic duct (Wagner et al. 1994). Even though the presence of pancreatic triglyceride lipase-related protein 2 in pancreatic secretions suggests that it may participate in fat digestion, the physiological function is not known for either pancreatic triglyceride lipase-related protein 1 or pancreatic triglyceride lipase-related protein 2. 

A clue to the function of the pancreatic triglyceride lipase-related proteins was suggested by examining the temporal pattern of expression for all three genes during development in the rat (Payne et al. 1994). In the adult human pancreas, the mRNA encoding pancreatic triglyceride lipase-related proteins 1 and 2 were 4- and 24-fold lower, respectively, than the mRNA for pancreatic triglyceride lipase (Giller et al. 1992). Similar distributions were observed in the adult rat pancreas, but the pattern was quite different in the newborn rat (Payne et al. 1994). In newborns, mRNA encoding pancreatic lipase-related protein 1 predominated. mRNA encoding pancreatic triglyceride lipase-related protein 2 was present at lower levels, and mRNA encoding pancreatic triglyceride lipase was not detected. This initial observation suggested that the three mRNA species were discordantly regulated during development.

To expand this observation, the presence of mRNA encoding each species was determined by dot-blot analysis of pancreas RNA harvested at various postnatal and prenatal ages. The mRNA encoding pancreatic triglyceride lipase, pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 were low prior to birth (Fig. 2). Within hours after birth, the mRNA encoding pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 rose to their maximum levels, whereas the mRNA for pancreatic triglyceride lipase remained low. During the suckling period, the relative levels of mRNA encoding pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 fell to the lower levels found in the adult pancreas. In contrast, mRNA encoding pancreatic triglyceride lipase remained low until the suckling-weanling transition when the levels reached adult values. Similar temporal regulation of the mRNA encoding pancreatic triglyceride lipase may also occur in human infants because they have a relative deficiency of pancreatic triglyceride lipase activity in the first year of life (Fredikzon and Olivecrona 1978). The temporal expression pattern of pancreatic triglyceride lipase, pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 suggests that the pancreatic triglyceride lipase-related proteins may help to compensate for the low pancreatic triglyceride lipase activity in suckling animals to ensure efficient fat digestion in newborns who consume more fat per kilogram than at any other age.

PROTEIN STRUCTURE

The primary sequences of pancreatic triglyceride lipase and the pancreatic triglyceride lipase-related proteins are remarkably conserved. Human and rat pancreatic triglyceride lipase are synthesized as 465 amino acid proteins with a short, 16 amino acid signal peptide. The mature protein is 449 amino acids (Lowe et al. 1989, Payne et al. 1994). Similarly, pancreatic triglyceride lipase-related protein 1 from both species has a short signal peptide of 17 amino acids, but the predicted sequence for the mature rat pancreatic triglyceride lipase-related protein 1, originally reported as pancreatic triglyceride lipase but later shown to be pancreatic triglyceride lipase-related protein 1 by sequence and enzymatic criteria, is longer than that of human pancreatic triglyceride lipase-related protein 1, 456 residues versus 451 (Giller et al. 1992, Wicker-Planquart and Puigserver 1992). The difference is explained by additional amino acids at the C-terminus of rat pancreatic triglyceride lipase-related protein 1. Human and rat pancreatic triglyceride lipase-related protein 2 may also differ, but in the length of the signal peptide not in the length of the mature protein (Giller et al. 1992, Payne et al. 1994). The reported human cDNA predicted a 17 amino acid signal peptide, whereas the rat cDNA had an open reading frame that predicted a 30 amino acid signal peptide; however, there was a second methionine, 16 amino acids from the N-terminus of the mature protein, that could function as the translation start site. Thus the rat protein may also have a short signal peptide. The predicted size of the mature protein was 452 amino acids for both species. In addition to the human and rat proteins, the sequences for porcine, rabbit, horse and guinea pig pancreatic triglyceride lipase, canine pancreatic triglyceride lipase-related protein 1 (originally reported as pancreatic triglyceride lipase), and mouse and coypu pancreatic triglyceride lipase-related protein 2 have been reported (Aleman-Gomez et al. 1992, Carrière et al. 1994, Hjorth et al. 1993, Kerfelec et al. 1986, Lowe et al. 1989, Payne et al. 1994, Thirstrup et al. 1994, Wicker-Planquart and Puigserver 1992, Winkler et al. 1990, Wishart et al. 1993). Comparisons of the available predicted amino acid sequences of pancreatic triglyceride lipase and the pancreatic triglyceride lipase-related proteins reveal that the primary sequences are 63-68% identical and 77-81% homologous (Fig. 3).

The primary sequence data have been supplemented by the tertiary structure of human pancreatic triglyceride lipase determined by radiograph crystallography at 2.8 nm (Winkler et al. 1990). This analysis shows that pancreatic triglyceride lipase is divided into two domains, a globular N-terminal domain formed by a central -sheet core extending from amino acids 1-335 and a C-terminal domain consisting of a -sheet sandwich structure (Fig. 4). The domains are separated by a short, unstructured stretch of amino acids and the domains are stabilized by seven disulfide bonds. The observed folding pattern of pancreatic triglyceride lipase is conserved in a number of hydrolases including several fungal lipases. This finding and the amino acid sequence homology of pancreatic triglyceride lipase to hepatic and lipoprotein lipase support the concept of a lipase gene family that evolved from a common ancestral hydrolase.

Several other important observations came from the crystal structure of pancreatic triglyceride lipase. First, examination of the tertiary structure supported earlier suspicions that the enzymatic mechanism of pancreatic triglyceride lipase resembles that of serine proteases and involves a serine-histidine-aspartic acid catalytic triad. A serine residue at position 153 had been previously identified as the potential nucleophilic residue by chemical modification and inactivation of porcine pancreatic triglyceride lipase (Verger 1984). In the crystal structure, Ser153 is located in the N-terminal domain with its side chain hydrogen bonded to His264, which in turn is hydrogen bonded to Asp177 in a topology that mimics the catalytic triad of trypsin.

Another observation about the crystal structure of pancreatic triglyceride lipase was the finding that the proposed catalytic triad is covered by several surface loops. The position of these loops would sterically hinder access of substrate to the catalytic site. A disulfide bridge between Cys238 and Cys262 defines the largest of these loops, termed the lid domain (Fig. 4). Two shorter loops, formed by residues 76-80, the -5 fold and by residues 213-217, also block the catalytic site. It is clear that these loops would have to change conformation and open the active site to substrate before lipolysis could proceed.

Although there are no crystal structures of the pancreatic triglyceride lipase-related proteins, analysis of the primary structures suggests that they should have conformations that are similar to pancreatic triglyceride lipase. Most of the differing amino acids are in unstructured surface loops, whereas the conserved amino acids are in regions that form the major determinants of secondary and tertiary structure. Notably, the disulfides and the regions of central -sheets and -sheet sandwich structure are conserved as are the components of the active site, the -5 fold, the nucleophile elbow (residues 175-183) and the catalytic triad. The lid domains are conserved in length and have nearly identical amino acid sequences. Given these similarities, it is highly probable that the tertiary structures of pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 resemble that of pancreatic triglyceride lipase.

LIPOLYSIS

SUBSTRATE SPECIFICITY

The substrate specificities of pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase-related protein 2 have been less well characterized, but differences have been noted between these proteins and pancreatic triglyceride lipase. Pancreatic triglyceride lipase-related protein 1 did not hydrolyze triglycerides, phospholipids or galactolipids, and a substrate has not been identified for this pancreatic triglyceride lipase homologue (Andersson et al. 1996, Giller et al. 1992, Payne et al. 1994). The reason for this lack of activity is not known and several potential explanations exist. Pancreatic triglyceride lipase-related protein 1 may require a different substrate, di- or monoglycerides, for example, or it may require a cofactor other than colipase, or pancreatic triglyceride lipase-related protein 1 may be an inactive homologue of pancreatic triglyceride lipase. This last explanation seems unlikely given the mutational rate of pancreatic triglyceride lipase-related protein 1 and the location of the mutations. Nonfunctional genes mutate at higher rates than their functional counterparts, and this has not occurred with pancreatic triglyceride lipase-related protein 1. Pancreatic triglyceride lipase-related protein 1 is as conserved among various species as are pancreatic triglyceride lipase and pancreatic triglyceride lipase-related protein 2. Furthermore, the functional residues of pancreatic lipase-related protein 1, such as the catalytic triad, and the major determinants of structure have been preserved, leaving the majority of divergent residues concentrated in other regions of the protein. This nonrandom pattern of divergent residues is in distinct contrast to the expected results of random mutations, which would distribute mutations throughout the protein. Certainly, mutations in biologically important regions should have occurred in a nonfunctional gene that is thought to have diverged along with pancreatic triglyceride lipase from a common ancestral gene over 90 million years ago (Giller et al. 1992, Sims et al. 1993). The conservation of functional residues and the apparent lack of random divergence between pancreatic triglyceride lipase-related protein 1 and pancreatic triglyceride lipase argue strongly for a biological role of pancreatic triglyceride lipase-related protein 1. Additional substrates must be tested and other roles considered before a definitive statement about the physiological function of pancreatic triglyceride lipase-related protein 1 can be made.

Pancreatic triglyceride lipase-related protein 2, on the other hand, has easily detectable activity against a variety of substrates (Table 1). It has the highest specific activity against triglycerides and can hydrolyze short-, medium- and long-chain triglycerides (Andersson et al. 1996, Jennens and Lowe 1995b, Thirstrup et al. 1994). It has lower, but significant activity against phospholipids and galactolipids. The activity against phospholipids may be advantageous if pancreatic triglyceride lipase-related protein 2 is important in digesting phospholipid-coated breast milk fat globules. An alternative reason for the phospholipase activity is suggested by the observation that pancreatic triglyceride lipase-related protein 2 is tightly associated with the zymogen granule membrane (Wishart et al. 1993). In this location, pancreatic triglyceride lipase-related protein 2 may participate in the release of the zymogen granule contents into the pancreatic ducts by altering membrane properties through hydrolysis of membrane phospholipids.

Table 1. Properties of pancreatic triglyceride lipase (PTL) and pancreatic triglyceride lipase-related protein 2 (PLRP2) [View Table]

The amino acids that determine the differences in substrate specificity between pancreatic triglyceride lipase and pancreatic triglyceride lipase-related protein 2 have not been identified. Simple comparison of the sequences corresponding to the catalytic site in the pancreatic triglyceride lipase crystal structure reveals remarkable conservation between pancreatic triglyceride lipase and pancreatic triglyceride lipase-related protein 2 with no obvious changes that should greatly affect the properties of the catalytic site. The most striking differences are in the lid domain, but these residues would be expected to move away from the catalytic site by analogy to pancreatic triglyceride lipase and should not affect substrate specificity. It is possible that the conformation of the catalytic site is affected by other regions of the protein and differs significantly from that of pancreatic triglyceride lipase, but such changes cannot be predicted by simple inspection of the primary sequences. The crystal structure of pancreatic triglyceride lipase-related protein 2 is necessary to aid in resolving the issue of substrate specificity.

BILE SALTS AND COLIPASE

Even though the function of converting procolipase to colipase is unsettled, the function of colipase in restoring activity to bile salt-inhibited pancreatic triglyceride lipase is well established. What is less clear are the molecular details of the mechanism involved in the interaction between pancreatic triglyceride lipase and colipase. There are kinetic and physical data that colipase and pancreatic triglyceride lipase form a complex at the substrate interface. The most dramatic evidence for this structure was provided by X-ray crystallography of a complex between porcine procolipase and human pancreatic triglyceride lipase (van Tilbeurgh et al. 1992). In this structure, procolipase was bound to the pancreatic triglyceride lipase C-terminal domain through two salt bridges involving Lys400 and Asp390 of pancreatic triglyceride lipase (Fig. 5). Procolipase formed three major loops stabilized by disulfide bridges. The tips of the loops contained primarily hydrophobic groups and extended away from pancreatic triglyceride lipase, suggesting that these loops may mediate interactions with the substrate interface (Fig. 6). Procolipase did not form any interactions with the N-terminal domain and did not alter the conformation of pancreatic triglyceride lipase from that reported for the crystal structure of pancreatic triglyceride lipase alone.

The importance of the interaction with the C-terminal domain was brought into question by a report of two pancreatic triglyceride lipase mutants lacking the C-terminal domain (Jennens and Lowe 1995a). The mutants, truncated after Phe336 or Tyr341, had a specific activity that was only 20% of that of wild-type pancreatic triglyceride lipase, but the activity was dependent on colipase. In this same study, pancreatic triglyceride lipase mutants with alterations in Lys400 and Asp390 were reported to be fully active, indicating that the salt bridges formed with colipase are not required for the efficient reactivation of pancreatic triglyceride lipase by colipase. These results demonstrated that the C-terminal domain is not essential for the interaction between colipase and pancreatic triglyceride lipase to occur, although the decreased activity of the deletion mutants could be attributed entirely or in part to the lack of that interaction. Importantly, these findings indicated that colipase could interact with the N-terminal domain of pancreatic triglyceride lipase.

An interaction of colipase with the N-terminal domain of pancreatic triglyceride lipase was also demonstrated by a second crystal structure of procolipase complexed with pancreatic triglyceride lipase (van Tilbeurgh et al. 1993). In this study, the crystallization media contained mixed micelles of phospholipid and detergent, which formed an interface. Colipase was associated with the C-terminal domain of pancreatic triglyceride lipase, but it also formed an interaction with the N-terminal domain of pancreatic triglyceride lipase (Fig. 7). Several conformational changes in pancreatic triglyceride lipase allowed these interactions to form. The most dramatic was the rearrangement of the lid domain, resulting in movement of these residues toward colipase and in the uncovering of the catalytic site. A hinge movement of the C-terminal domain facilitated the formation of interactions between Glu15 of colipase and Asn241 in the lid and between Arg38 of colipase and Ser243 and Val246 of the lid. The importance of these interactions in the mechanism of lipolysis has not been ascertained, but it is reasonable to speculate that the bonds between colipase and the lid domain stabilize the lid in the open conformation and permit substrate to enter the active site.

Further insights into the molecular details of the interaction of colipase with pancreatic triglyceride lipase have been provided by comparisons with pancreatic triglyceride lipase-related protein 2 (Jennens and Lowe 1995b). This homologue showed much different behavior in the presence of bile salts and with colipase than did pancreatic triglyceride lipase. Rat pancreatic triglyceride lipase-related protein 2 was not inhibited by even supraphysiological concentrations of bile salts. In this respect, rat pancreatic triglyceride lipase-related protein 2 differed from coypu pancreatic triglyceride lipase-related protein 2, which was completely inhibited by bile salts (Thirstrup et al. 1994). Colipase increased the activity of rat pancreatic triglyceride lipase-related protein 2, but the increase was only 1.5- to 4-fold, much less than colipase increases pancreatic triglyceride lipase activity. Coypu pancreatic triglyceride lipase-related protein 2 did not show any response to the addition of colipase. These observations demonstrated that pancreatic triglyceride lipase-related protein 2 can overcome the presence of bile salts without colipase and that it contains residues that function like colipase and permit activity in the presence of bile salts. These functional differences between pancreatic triglyceride lipase-related protein 2 and pancreatic triglyceride lipase provide a unique opportunity to identify the residues responsible for important interactions with colipase and with mixed micelles of bile salts and triglycerides.

INTERFACIAL ACTIVATION

The crystal structure of the colipase-pancreatic triglyceride lipase complex obtained in the presence of mixed micelles clearly showed a conformational change in pancreatic triglyceride lipase (Fig. 7). Not only did the lid domain move, but its movement induced changes in another loop near the catalytic site, the -5 loop. Together these movements configured the catalytic site to accept substrate. Concurrent studies on fungal lipases have also shown similar conformational changes in a surface loop homologous to the pancreatic triglyceride lipase lid domain. The authors of these studies proposed that the conformational rearrangement of pancreatic triglyceride lipase is the physical correlate of interfacial activation (Brzozowski et al. 1991, Grochulski et al. 1993, van Tilbeurgh et al. 1993).

This hypothesis is supported by two additional studies. First, a pancreatic lipase that is quite homologous to the pancreatic triglyceride lipase-related protein 2 family with one exception was isolated from guinea pig (Hjorth et al. 1993). It is missing sequence corresponding to the lid domain and was predicted to have an open catalytic site (Fig. 8). This hypothesis was confirmed experimentally by demonstrating that the lipase possesses high activity against monomeric substrates and is not activated by interfaces. Apparently, the guinea pig lipase is a naturally occurring mutant with a deletion of the lid domain. The properties of this lipase support the enzyme model of interfacial activation.

In the second study, deletion mutants of pancreatic triglyceride lipase were created, expressed and characterized (Jennens and Lowe 1994). One mutation removed all of the lid domain and the second mutation removed a portion of the lid domain, the -helix covering the catalytic site (Fig. 9). Both mutants had decreased, but easily detectable activity against emulsions of triglycerides and taurodeoxycholate compared with the wild-type lipase. Like the guinea pig lipase, the mutants were not activated by interfaces and could hydrolyze water-soluble substrates at a high rate (Fig. 10). This finding demonstrated that the lid domain is required for interfacial activation. A role for the lid domain in interfacial binding was also demonstrated in this study by showing that the mutants have greatly decreased binding to tributyrin and taurodeoxycholate emulsions. These results further supported the enzyme model of interfacial activation and implicated the lid domain in the mechanism of interfacial activation and lipid binding.

Despite the attractiveness of this model, which centers on the conformational changes seen in the lid domain, the presence of a lid domain is not sufficient to predict interfacial activation. This complexity arose after the description of other pancreatic triglyceride lipase-related protein 2 family members. Both rat and coypu pancreatic triglyceride lipase-related protein 2 have sequence homologous to the lid domain, but neither possess interfacial activation (Payne et al. 1994, Thirstrup et al. 1994). These results mean either that the pancreatic triglyceride lipase-related protein 2 sequence corresponding to the pancreatic triglyceride lipase lid does not occupy the same position in the tertiary structure or that the lid domain in the pancreatic triglyceride lipase-related protein 2 lipases is more dynamic and spends a large portion of time in the open position.

The corollary of the latter hypothesis is that the lid domain of pancreatic triglyceride lipase is less mobile and stabilized in the closed position, presumably through interactions between lid domain and main chain residues. Inspection of the lid domain sequence of pancreatic triglyceride lipase and pancreatic triglyceride lipase-related protein 2 suggests a potential explanation for the presumed differences in mobility or stability of the lid domains (Fig. 8). Two residues, Arg257 and Asp258, in the pancreatic triglyceride lipase lid are conserved across species and form salt bridges and hydrogen bonds with residues in the core of the protein. These contacts change with the position of the lid and act to stabilize the lid in the open position. In the pancreatic triglyceride lipase-related protein 2 lipases, these two residues are not conserved and even vary among members of the pancreatic triglyceride lipase-related protein 2 family. These differences may result in fewer bonds stabilizing the conformation of the pancreatic triglyceride lipase-related protein 2 lid, allowing more mobility, with a significant portion of time spent with the lid in the open position. As a result, monomeric substrate is able to enter the catalytic site without a conformational change in the lid domain triggered by an interface as happens with pancreatic triglyceride lipase.

CONCLUSION