Hereditary Pancreatitis-associated Mutation Asn21 → Ile Stabilizes Rat Trypsinogen in Vitro *

Mutations Arg117 → His and Asn21 → Ile in human trypsinogen-I have been recently associated with hereditary pancreatitis (HP). The Arg117→ His substitution is believed to cause pancreatitis by stabilizing trypsin against autolytic degradation, while the mechanism of action of Asn21 → Ile has been unknown. In an effort to understand the effect(s) of this mutation, Thr21 in the highly homologous rat trypsinogen-II was replaced with Asn or Ile, and the recombinant zymogens and their active trypsin forms were studied. Kinetic parameters of all three trypsins were comparable, and the active enzymes suffered autolysis at similar rates, indicating that neither catalytic properties nor proteolytic stability of trypsin are influenced by mutations at position 21. When incubated at pH 8.0, 37 °C, pure zymogens underwent autoactivation with concomitant trypsinolytic degradation in a Ca2+-dependent fashion. Thus, in the presence of 5 mm Ca2+, autoactivation and digestion of the zymogens after Arg117and Lys188 were observed, while in the presence of 1 mm EDTA autoactivation and cleavage at Lys188were reduced, and zymogenolysis at the Arg117 site was enhanced. Overall rates of zymogen degradation in [Asn21]- and [Ile21]trypsinogens were higher in Ca2+ than in EDTA, while [Thr21]trypsinogen demonstrated inverse characteristics. Remarkably, both in the presence and absence of Ca2+, [Ile21]trypsinogen exhibited significantly higher stability against autoactivation and proteolysis than zymogens with Asn21 or Thr21. The observations suggest that autocatalytic trypsinogen degradation may be an important defense mechanism against excessive trypsin generation in the pancreas, and trypsinogen stabilization by the Asn21 → Ile mutation plays a role in the pathogenesis of HP.

Hereditary pancreatitis (HP) 1 is a rare autosomal dominant disorder with recurrent attacks of acute pancreatitis and frequent progression to chronic pancreatitis (1). Recently, mutations Arg 117 3 His and Asn 21 3 Ile in the human cationic trypsinogen (TG-I) have been associated with HP (2)(3)(4)(5). It was proposed that the Arg 117 3 His mutation eliminates a trypsinsensitive cleavage site on one of the surface loops of trypsin, a so-called autolysis loop, and stabilizes trypsin against autocatalytic digestion (2). Degradation of trypsin by autolysis or a trypsin-like enzymes in the pancreas is thought to serve as a protective mechanism against excessive trypsin liberation in pathological states (2)(3)(4)(5). On the basis of in vitro studies it has been long known that Arg 117 is one of the autocatalytic cleavage sites in bovine trypsin (6). More recently, Vá rallyai et al. (7) demonstrated that in rat trypsin cleavage at Arg 117 or Lys 61 destabilizes the intervening loop with subsequent loss of activity due to rapid and extensive proteolysis at multiple cleavage sites. Importantly, mutation of Arg 117 to Asn resulted in dramatic stabilization of trypsin, thereby providing the first direct indication that mutations at this position lead to autolysisresistant trypsins (7). In contrast to the relative abundance of evidence supporting a clear pathogenic scheme for the Arg 117 3 His mutation, practically nothing is known about the mechanism of action of the Asn 21 3 Ile mutation. At least two speculative models have been offered (3)(4)(5)8): (i) the mutation may decrease the accessibility of Arg 117 and cause increased autolytic stability, possibly due to a salt bridge formation with Glu 24 , or (ii) the mutation may increase the rate of autoactivation by altering the secondary structure of the activation peptide region. In both cases the end result would be excessive trypsin accumulation leading to autodigestion of the pancreatic tissue. To test these models, in the present study we replaced Thr 21 in the rat TG-II with Asn or Ile, and the purified recombinant zymogens and their active trypsin forms were characterized. The results clearly indicate that the Asn 21 3 Ile mutation significantly decreases autoactivation and zymogen degradation without affecting trypsin stability or activity and suggest that unwanted zymogen stabilization may play an important role in the pathogenesis of HP.
Construction of TG Mutants-Mutations were introduced into the rat anionic TG gene (TG-II) by oligonucleotide-directed site-specific mutagenesis, using the "overlap extension" PCR method (12). The PCR fragments were digested with restriction enzymes BamHI-EcoRI (mutants Thr 21 3 Asn and Thr 21 3 Ile), EcoRI-SacI (mutant Arg 117 3 His), or XhoI-SacI (mutant Lys 188 3 Asn) and ligated into the similarly treated TG gene in the expression vector pTrap (13). The DNA sequence of the mutations as well as the entire subcloned PCR fragments were verified by dideoxy sequencing.
Expression and Purification of Recombinant TGs-Wild-type and mutant TGs were expressed to the periplasm in E. coli SM138 as described previously (13,14). In a typical experiment, 2-liter cultures of SM138/pTrap in Luria-Bertani medium with 100 g/ml ampicillin were grown to saturation overnight, and periplasmic fractions were isolated by osmotic shock. Tris-HCl (pH 8.0) and NaCl were added to a final concentration of 20 mM and 0.2 M, respectively, and the approximately 200-ml periplasm was applied directly to a 2-ml ecotin affinity column (11). The column was washed with 20 mM Tris-HCl (pH 8.0), 0.2 M NaCl, and zymogens were eluted with 50 mM HCl. The pH of the eluate was * 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.
re-adjusted to 8.0 with Tris-HCl, and the approximately 4-ml zymogen preparation was re-chromatographed on the same ecotin column. The second eluate was essentially homogenous trypsinogen, no other contaminating bands were detectable on Coomassie Blue or silver-stained gels. Concentrations of zymogen solutions were estimated from their ultraviolet absorbance using an extinction coefficient of 38,000 M Ϫ1 cm Ϫ1 at 280 nm (14). In addition, to ensure that identical amounts of wild-type and mutant zymogens were used in the experiments presented below, serial dilutions of the different mutants were electrophoresed on 12% SDS-PAGE gels, and the intensities of the Coomassie Blue-stained bands were compared. Typical trypsinogen yields were 100 -200 g of purified zymogen/liter of culture.
Assay of Trypsin Activity-Trypsin activity was determined using the synthetic chromogenic substrate N-CBZ-Gly-Pro-Arg-p-nitroanilide. Kinetics of the chromophor release were followed at 410 nm in 0.1 M Tris-HCl (pH 8.0), 1 mM CaCl 2 , at 37°C.
Autolysis of Trypsin-Autolytic degradation of trypsin was followed by residual activity measurements (7). Trypsinogens were activated in a volume of 110 l with 60 ng of enterokinase for 30 min at 22°C in the presence of 0.1 M Tris-HCl (pH 8.0), 5 mM CaCl 2 . Trypsin solutions were then incubated at 37°C without any further additions (i.e. in the presence of 5 mM Ca 2ϩ ) or after addition of K-EDTA (pH 8.0) to a final concentration of 15 mM. At the indicated times 2-5-l samples were withdrawn for trypsin activity determination.
Autoactivation of Trypsinogens-Trypsinogens (150 pmol) were incubated at 37°C in 0.1 M Tris-HCl (pH 8.0) in the presence of 5 mM CaCl 2 or 1 mM K-EDTA in a volume of 110 l. At the indicated times a sample of 5-10 l was removed for trypsin activity assay. Alternatively, reactions were terminated by trichloroacetic acid precipitation, proteins were separated on a 12% reducing SDS-PAGE, and bands were visualized by Coomassie Blue staining.

RESULTS
Construction of Mutants-Rat anionic TG (TG-II) is probably the best characterized and most widely used recombinant TG/ trypsin model system. Homology with the human cationic TG (TG-I) is extensive (80% identity), and the N-terminal regions are particularly well conserved between the two proteins (15). Interestingly, one of the very few differences is found at position 21, where the rat species contains a Thr. In an attempt to explore the significance of the amino acid side chain at position 21, and to understand the effect(s) of the Asn 21 3 Ile change observed in the TG-I of HP patients, we replaced Thr 21 in rat TG-II with Asn or Ile, and the recombinant zymogens and their active trypsin forms were studied. PCR mutagenesis was carried out as described under "Experimental Procedures," and mutant TG genes were cloned into the expression vector pTrap (13) behind the alkaline phosphatase promoter and signal sequence. E. coli SM138 was transformed with verified clones, and TG was expressed into the periplasmic space in a constitutive fashion. Zymogens were purified to homogeneity from periplasmic extracts with a one-step affinity procedure using immobilized ecotin (11) as detailed under "Experimental Procedures." Due to the relatively low levels of expression, the rapid purification technique and the acidic conditions used for elution, remarkably pure zymogen preparations without any detectable trypsin activity were obtained.
Enterokinase Activation of Zymogens-Time courses of enterokinase-catalyzed conversion of TG to trypsin were followed by SDS-PAGE and Coomassie Blue staining. As shown in Fig.  1A, in the presence of 5 mM Ca 2ϩ at 22°C, enterokinase activated zymogens with Thr 21 or Ile 21 at comparable rates, while activation of [Asn 21 ]TG was at least 2-fold accelerated. Although not shown, no other degradation products were observed on the stained gels. In a different experiment, enterokinase activation of zymogens was followed by continuously monitoring trypsin activity 37°C, in the presence of 1 mM Ca 2ϩ . Under these conditions the initial appearance of trypsin activity reflects predominantly enterokinase action, without significant interference from autocatalytic activation. Fig. 1B  Catalytic Properties and Autolysis of Mutant Trypsins-Zymogens were activated by enterokinase and catalytic parameters of the active enzymes were determined using the chromogenic substrate N-CBZ-Gly-Pro-Arg-p-nitroanilide (Table I). No significant differences were detected in the K m , k cat , or k cat /K m values of trypsins with Thr, Asn, or Ile at position 21, indicating that catalytic properties of trypsin are not influenced by mutations at this position.
It has been proposed that the Asn 21 3 Ile mutation may decrease the accessibility of the autolytic site at Arg 117 and cause increased autolytic stability, in a manner that is similar to the effects of the Arg 117 3 His mutation (3)(4)(5). To verify this suggestion, autolytic degradation of trypsins with Thr 21 , Asn 21 , or Ile 21 were characterized together with the Asn 21 /Arg 117 3 His double mutant. When solutions of trypsin were incubated at 37°C, enzyme activity was gradually lost as a function of time (Fig. 2). Surprisingly, rates of autolytic degradation showed no appreciable difference when trypsins with different substitutions at position 21 were compared, while stability of the Asn 21 /Arg 117 3 His mutant was significantly increased. As described previously (7), 5 mM Ca 2ϩ afforded significant protection against autolysis (compare Fig. 2, A and B); however, it had no effect on the relative degradation patterns of the trypsin mutants. The results indicate that autolytic stability is not altered by mutations at position 21 and suggest essential mechanistic differences between the effects of the Asn 21 3 Ile and Arg 117 3 His mutations in HP.
Autoactivation of Mutant Zymogens in the Presence of Ca 2ϩ -One of the prevailing theories to explain the effects of the Asn 21 3 Ile mutation suggests that altered accessibility of the activation site may lead to excessive trypsin formation through increased autoactivation (3)(4)(5). To examine this possibility, in a series of experiments autoactivation of zymogens was characterized by activity assays and SDS-PAGE. When incubated at 37°C in the presence of 5 mM Ca 2ϩ , pure zymogens underwent relatively rapid autoactivation and proteolytic degradation (Fig. 3) ]TG was converted to trypsin and proteolytic products, and at 1.5 h no more intact zymogen was detectable on gels (Fig. 3A). In contrast, [Thr 21 ]TG and [Ile 21 ]TG were almost completely stable at least up to 1 and 1.5 h, respectively. Rapid autoactivation of the Asn 21 -zymogen was also evident from the early appearance of trypsin activity. Differences were most pronounced at 1 h, when more than 20-fold higher trypsin activity was detectable in [Asn 21 ]TG samples relative to [Ile 21 ]trypsin activity. It is important to note that maximal activity achieved during autoactivation was only 40 -50% of the potential maximal value, as determined after enterokinase activation of zymogens. This apparent loss of activity is presumably due to the significant proteolytic degradation also observed on gels (Fig. 3A). On the routinely used 12% gels four major stable degradation products were identified by Coomassie Blue staining (bands I-IV in Fig. 3A). Appearance of bands I and III in all three zymogen variants coincided with the  B, aliquots of 5-10 l were withdrawn from reaction mixtures at indicated times and trypsin activity was determined with the synthetic substrate N-CBZ-Gly-Pro-Arg-p-nitroanilide. Activity was expressed as percentage of the potential total activity, as determined on similar zymogen samples activated with enterokinase. Each point represents the average of two measurements. emergence of the trypsin band, indicating that proteolysis of zymogens is due to the trypsin generated by autoactivation. No differences were observed in the overall proteolytic pattern among the three zymogens studied. Rat TG-II migrates at an anomalously high molecular weight in SDS-PAGE (11), therefore reliable estimation of fragment sizes by comparing to molecular mass standards was not feasible. However, changes in the intensities of bands I-IV during the investigated time course suggested that band I may get converted to band II, and cleavage of band III may give rise to band IV (see Fig. 3A). In addition, molecular mass differences between bands I-II and III-IV, respectively, appeared to be similar to the difference between trypsinogen and trypsin, suggesting that bands I and III have an intact trypsinogen N terminus which gets "activated" by trypsin and results in bands II and IV. This assumption was further supported by the observation that a 1 min "flashdigestion" by enterokinase almost quantitatively converted bands I and III to II and IV, respectively (Fig. 4A). Subsequently, bands I-IV were electroblotted to polyvinylidene difluoride membrane and subjected to N-terminal protein sequencing. The results confirmed the previous assumptions and demonstrated that bands I and III have an intact zymogen N terminus (13), and bands II and IV start with the N-terminal sequence of trypsin (Fig. 4A). The internal cleavage sites responsible for the generation of these fragments were first tentatively identified by comparing the relative mobility differences of the fragments on gels to possible trypsinolytic peptide bonds in the protein. The calculations indicated that bands I and II were generated by digestion of the Lys 188 -Asp 189 peptide bond, while bands III and IV were the result of cleavage of the Arg 117 -Val 118 bond. Mutagenesis of Arg 117 and Lys 188 confirmed that these two cleavage sites were responsible for fragments I-IV. Thus, degradation patterns of the Asn 21 /Arg 117 3 His mutant contained only bands I and II (Fig. 4B), while degradation of Asn 21 /Lys 188 3 Asn yielded only bands III and IV (Fig. 4C). Interestingly, in the latter mutant a new stable proteolytic fragment became apparent, too. Note that autoactivation kinetics of the Asn 21 /Arg 117 3 His 117 zymogen were similar to those of [Asn 21 ]TG (Fig. 3A), while the Asn 21 /Lys 188 3 Asn mutant exhibited a slower rate of autoactivation (Fig.  4C), in all likelihood due to its somewhat impaired catalytic activity.
Autoactivation of Mutant Zymogens in the Absence of Ca 2ϩ -Comparison of zymogen behavior at 37°C in the presence or absence of Ca 2ϩ revealed mutant-specific changes in autoactivation and zymogen degradation patterns (Fig. 5). Thus, in the presence of 1 mM EDTA the rate of trypsin formation by autoactivation was significantly decreased in both [Asn 21 ]TG and [Ile 21 ]TG, in agreement with previous observations that Ca 2ϩ is required for optimal digestion at the Lys 15 activation site. Proteolysis at Arg 117 was remarkably enhanced, and only one major proteolytic fragment (band III) was apparent on gels. This observation also supports previous findings, that Arg 117 and the autolysis loop are stabilized by Ca 2ϩ (7). Digestion at Lys 188 was minimal, indicating that this site is probably not Ca 2ϩ -sensitive, and cleavage efficiency here is only dependent on the amount of trypsin generated during autoactivation. Note that conversion of band III to band IV was also abolished, which is also explained by limited trypsin generation and the Ca 2ϩ dependence of this process. Importantly, the stabilizing effect of Ile 21 was even more pronounced under these conditions, and only insignificant zymogen degradation was observed up to 3 h, whereas the majority of [Asn 21 ]trypsinogen was degraded by 1.5 h. [Thr 21 ]TG exhibited unexpected properties under Ca 2ϩ -free conditions. Autoactivation was decreased; however, this inhibition was far less pronounced than in [Asn 21 ]TG or [Ile 21 ]TG. In turn, the relatively higher amounts of trypsin resulted in massive digestion at Arg 117 (band III). The overall effect was a surprisingly rapid zymogen degradation with little trypsin liberation (Fig. 5A). Monitoring trypsin activity during autoactivation in 1 mM EDTA revealed a pattern consistent with those observed on gels (Fig. 5B). Only [Thr 21 ]TG exhibited significant activity; however, even at its maximum it was below 20% of the total potential activity, as determined on enterokinase activated zymogen samples. Trypsin liberated from [Asn 21 ]TG was minimal, while [Ile 21 ]TG samples did not yield any measurable signal above background over the 3-h time period. DISCUSSION The present study is the first attempt of an in-depth biochemical investigation into the effects of the HP-associated Asn 21 3 Ile mutation in human cationic TG. Since high yield recombinant expression and purification protocols for the human trypsinogens are not readily available yet, we used the homologous rat TG model system and replaced Thr 21 with Asn or Ile, and properties of the three zymogens and their active trypsin forms were compared. More specifically, we tested two theories providing speculative explanations while this mutation leads to the HP phenotype (3)(4)(5)8). (i) In analogy to the effects of the Arg 117 3 His mutation, it was suggested that the Asn 21 3 Ile mutation may indirectly decrease autolytic cleavage at Arg 117 and causes increased autolytic stability of trypsin. (ii) Alternatively, the relative proximity of the Asn 21 3 Ile mutation to the activation peptide region prompted the hypothesis, that enhanced autoactivation may increase trypsin liberation. A common feature of both models that HP-associated mutations are supposed to lead to uncontrolled, excessive trypsin activity and tissue autodigestion. Surprisingly, our observations demonstrate that mutations at position 21 have no effect whatsoever on autolytic stability of trypsin and autoactivation is not increased in [Ile 21 ]TG. In sharp contrast, the primary effect of the Asn 21 3 Ile mutation appears to be stabilization of the activation peptide region, resulting in significantly decreased autoactivation rates (Figs. 3 and 5). While the exact mechanism of this stabilization is not clear yet, structural rearrangements in the activation peptide (5) are likely to be responsible for the altered proteolytic accessibility and/or cleavage efficiency of the Lys 15 site. This notion is also supported by the reduced rates of activation by enterokinase in [Ile 21 ]TG relative to [Asn 21 ]TG (see Fig. 1).
One important consequence of the decreased autoactivation rates in [Ile 21 ]TG is the delayed onset of autocatalytic zymogen degradation (see Figs. 3 and 5). Proteolytic inactivation of TG by trypsin generated during autoactivation appears to be highly efficient, and it is intriguing to speculate that this might function as a "failsafe" mechanism of controlling excessive trypsin liberation in the pancreas. In the presence of Ca 2ϩ , only about 50% of the total potential trypsin activity can be produced from [Asn 21 ]TG by autoactivation, while in the absence of Ca 2ϩ this number is well below 10%. We found that cleavage of TG occurs first at Arg 117 and Lys 188 . Digestion of the Lys 188 -Asp 189 bond in bovine trypsin has been shown to significantly impair activity (16), while cleavage of the Arg 117 -Val 118 bond per se does not inactivate trypsin (6); however, it destabilizes the autolysis loop and leads to widespread trypsinolysis in this region (7). Cleavage at the Lys 15 and Arg 117 sites are inversely controlled by Ca 2ϩ , which enhances proteolysis at Lys 15 and inhibits at Arg 117 . Remarkably, as best demonstrated by the behavior of [Thr 21 ]TG, the combination of these two opposing actions of Ca 2ϩ appears to determine the overall reaction kinetics of zymogen degradation. Thus, in the absence of Ca 2ϩ , inhibition of proteolysis at Lys 15 is less pronounced in [Thr 21 ]TG than in [Asn 21 ]TG or [Ile 21 ]TG, and relatively higher levels of trypsin are liberated, which in turn can more efficiently attack the sensitized Arg 117 site. From these observations it is also apparent, that the amino acid side chain at position 21 is an important determinant of the effect of Ca 2ϩ on the activation site. Further studies are required to clarify whether amino acid 21 influences the Ca 2ϩ binding affinity of the activation peptide or modulates the effects of the bound Ca 2ϩ .
It is generally believed that in pancreatitis pathologic trypsin generation is initiated inside the acinar cells, presumably in a low Ca 2ϩ environment (17). We hypothesize that rapid and efficient reduction of the "potentially hazardous" zymogen storage pool under these conditions can be critical in preventing widespread autoactivation. Overall rates of zymogen degradation in different TG species are primarily determined by amino acid 21. With the exception of human TG-I and -II, mammalian trypsinogens carry a Thr residue at this position, which appears to allow for rapid zymogenolysis in a low Ca 2ϩ environment. The presence of Asn and Ile in human TG-I and -II, respectively, are unique and so far not observed in any other species. While the evolutionary rationale behind these changes is not yet understood, both substitutions lead to slower zymogen degradation in the absence of Ca 2ϩ , which may weaken an important defense mechanism against uncontrolled trypsin generation and autodigestion. In HP caused by the Asn 21 3 Ile mutation of TG-I, the highly stable Ile 21 -zymogen is overpro-duced, causing a significant increase in the risk of pancreatitis. The increased stability against autoactivation of human TG-II relative to TG-I has been described previously (18), and based on our results this may be attributed solely to the Asn versus Ile difference at position 21.
The proposed pathomechanism of HP described above provides an attractive working model, which should be also readily testable. However, given the complexity of the pancreatic zymogen synthesis, storage, and secretion, we cannot rule out that the Asn 21 3 Ile mutation affects other processes along this pathway, too. One such possibility is that the mutation may decrease affinity for trypsin inhibitors found in the pancreas, e.g. the pancreatic secretory trypsin inhibitor. Although we have not tested this idea in detail in the present study, inhibition experiments with bovine pancreatic trypsin inhibitor (aprotinin) showed no significant differences between the inactivation profiles of trypsins with Asn 21 , Thr 21 , or Ile 21 (not shown).
Finally, we need to remember that an obvious caveat to the conclusions of the present study is the use of the rat anionic TG/trypsin model system rather than the human cationic TG. TR, trypsin, I, III, IV, major proteolytic fragments described in the legends to Figs. 3 and 4. B, aliquots of 5-10 l were withdrawn from reaction mixtures at indicated times, and trypsin activity was determined with the synthetic substrate N-CBZ-Gly-Pro-Arg-p-nitroanilide. Activity was expressed as percentage of the potential total activity, as determined on similar zymogen samples activated with enterokinase. Each point represents the average of two measurements.
Although the rodent enzyme is 80% identical to the human isoform, this does not guarantee that the Asn 21 3 Ile mutation will have the exact same effect in the two species. Therefore, further experiments to confirm the present findings will be mandatory as human recombinant enzymes become available.