Chymotrypsin C (Caldecrin) Stimulates Autoactivation of Human Cationic Trypsinogen*

Trypsin-mediated trypsinogen activation (autoactivation) facilitates digestive zymogen activation in the duodenum but may precipitate pancreatitis if it occurs prematurely in the pancreas. Autoactivation of human cationic trypsinogen is inhibited by a repulsive electrostatic interaction between the unique Asp218 on the surface of cationic trypsin and the conserved tetra-aspartate (Asp19-22) motif in the trypsinogen activation peptide (Nemoda, Z., and Sahin-Tóth, M. (2005) J. Biol. Chem. 280, 29645-29652). Here we describe that this interaction is regulated by chymotrypsin C (caldecrin), which can specifically cleave the Phe18-Asp19 peptide bond in the trypsinogen activation peptide and remove the N-terminal tripeptide. In contrast, chymotrypsin B, elastase 2A, or elastase 3A (proteinase E) are ineffective. Autoactivation of N-terminally truncated cationic trypsinogen is stimulated ∼3-fold, and this effect is dependent on the presence of Asp218. Because chymotrypsinogen C is activated by trypsin, and chymotrypsin C stimulates trypsinogen activation, these reactions establish a positive feedback mechanism in the digestive enzyme cascade of humans. Furthermore, inappropriate activation of chymotrypsinogen C in the pancreas may contribute to the development of pancreatitis. Consistent with this notion, the pancreatitis-associated mutation A16V in cationic trypsinogen increases the rate of chymotrypsin C-mediated processing of the activation peptide 4-fold and causes accelerated trypsinogen activation in vitro.

The digestive enzyme cascade is a tightly regulated activation process of pancreatic zymogens in the duodenum. First, trypsinogen is activated to trypsin by enteropeptidase (enterokinase), and in turn, trypsin activates all other protease zymogens. Trypsin also activates trypsinogen, in a proteolytic reaction termed autoactivation, which is thought to have a physiological role in facilitating zymogen activation in the duodenum. However, the unique ability of trypsinogen to autoactivate renders this zymogen a potentially harmful disease-causing agent, since inappropriate autoactivation within the pancreas might initiate an autodigestive process and result in pancreatitis. The human pancreas produces three isoforms of trypsinogen encoded by separate genes, the PRSS1 (protease, serine 1), PRSS2, and PRSS3 genes (for a recent review, see Ref. 1 and references therein). On the basis of their relative isoelectric points and electrophoretic mobility, the iso(pro)enzymes are commonly referred to as cationic trypsinogen (PRSS1), anionic trypsinogen (PRSS2), and mesotrypsinogen (PRSS3). PRSS1 (ϳ60 -70%) and PRSS2 (ϳ30 -40%) account for the majority of trypsinogens in the pancreatic juice, whereas PRSS3 is secreted in relatively low amounts. Further-more, this minor isoform exhibits defective inhibitor binding, and it cannot autoactivate or activate other protease zymogens (2).
The causative role of trypsinogen in pancreatitis is supported by the identification of mutations in the PRSS1 gene of patients with hereditary pancreatitis (3)(4)(5)(6)(7). In contrast, genetic variants of PRSS2 or PRSS3 have not been described in association with chronic pancreatitis (8). Three PRSS1 mutations, namely R122H (ϳ70%), N29I (ϳ25%), and A16V (ϳ4%), have been found with relatively high frequency in multiple families, whereas 18 additional genetic variants have been identified only in very few patients (for an up-to-date list, see the pancreatitis mutation data base on the World Wide Web at www.uni-leipzig.de/pancreasmutation). Biochemical analyses of the frequently found R122H and N29I mutations as well as a subset of rare mutations (D19A, D22G, K23R, N29T) indicated that the common phenotypic change in the pancreatitis-associated mutants is an increased propensity for autoactivation (9 -13). However, the third most frequently detected mutation, A16V, which alters the N-terminal amino acid of trypsinogen, has appeared to be an exception so far, because in a recent study, the A16V mutant failed to exhibit increased autoactivation (14).
Recently, we demonstrated that the conserved tetra-aspartate sequence (Asp 19 -22 ) in the trypsinogen activation peptide plays an essential role in suppressing autoactivation of human cationic trypsinogen (15). Two inhibitory interactions were identified that employ the activation peptide: one between Asp 22 and the conserved hydrophobic S2 subsite on trypsin and another between Asp 21 and the unique Asp 218 , which forms part of the S3 subsite on trypsin. Together, these interactions can suppress the rate of autoactivation by more than 2 orders of magnitude. Interestingly, the inhibitory mechanism of autoactivation due to the Asp 21 -Asp 218 interaction appears to be specific for human cationic trypsinogen, since Asp 218 is rarely found in vertebrate trypsinogens, which mostly contain a Tyr residue at the corresponding position. Here we report that not only is this inhibitory interaction unique to humans, but it is also regulated by a novel mechanism, which involves proteolytic processing of the activation peptide by chymotrypsin C. Autoactivation of the N-terminally truncated cationic trypsinogen increases significantly, which can facilitate physiological zymogen activation in the duodenum, and may play a role in the precipitation of pancreatitis in carriers of the A16V cationic trypsinogen mutation.

EXPERIMENTAL PROCEDURES
Nomenclature-The genetic abbreviations PRSS1 (protease, serine 1) and PRSS2 (protease, serine 2) are used to denote human cationic trypsinogen and anionic trypsinogen, respectively. Similarly, chymotrypsinogens B and C are abbreviated as CTRB and CTRC, and proelastase 2A and proelastase 3A are indicated as ELA2A and ELA3A. Note that chymotrypsin C is also known in the literature as caldecrin. Amino acid residues in the trypsinogen sequences are numbered according to their position in the native preproenzyme, starting with Met 1 . The first amino acid of the mature cationic trypsinogen is Ala 16 . The term "autoactiva-tion" is used to describe trypsin-mediated trypsinogen activation. The intrinsic catalytic activity of trypsinogen is ϳ10 8 -fold lower than the activity of trypsin; therefore, its impact on trypsinogen activation is negligible in our experiments (16).
Human Pancreatic Juice Samples-Deidentified samples of pure human pancreatic juice were received in lyophilized form, as a kind gift from Dr. Niels Teich (University of Leipzig, Germany). The juice was collected from patients who underwent pancreas transplantation with bladder drainage of exocrine pancreatic secretions. To prevent zymogen activation, Trasylol (aprotinin) was added at the time of collection. See Ref. 17 for details.
Recombinant Trypsinogen Preparations-Because the routinely used recombinant trypsinogen preparations expressed in Escherichia coli Rosetta(DE3) cells contain abnormal N termini (see below), in this study, recombinant trypsinogens were also made in the E. coli LG-3 strain and human embryonic kidney (HEK) 2 293T cells, which produce trypsinogens with homogenous, authentic N-terminal sequences. In the experiments, which analyze rates of chymotrypsin C-mediated N-terminal processing of cationic trypsinogen, recombinant preparations from LG-3 or HEK cells were used exclusively. On the other hand, we observed that N-terminal processing of trypsinogen preparations from E. coli Rosetta(DE3) cells is markedly accelerated. Therefore, these trypsinogen preparations proved to be valuable in experiments, which examined the effect of N-terminal processing. The source of the recombinant preparations used in the different experiments is clearly indicated in the figure legends.
Purification of Native Protease Zymogens from Pancreatic Juice-Fifty mg of lyophilized pancreatic juice powder was solubilized with 1 ml of 10 mM HCl, and insoluble material was removed by centrifugation. The supernatant was diluted with 3 ml of water and loaded onto a MonoQ anion exchange column (Amersham Biosciences) equilibrated with 20 mM Tris-HCl (pH 8.0). Proteins were eluted with a 0 -0.5 M NaCl gradient at a 1 ml/min flow rate, and 1 ml fractions were collected. Aliquots of fractions were subjected to trypsin and chymotrypsin activity assays before and after activation with enteropeptidase or trypsin, respectively. Briefly, a 10-l fraction was mixed with 40 l of assay buffer (0.1 M Tris-HCl (pH 8.0), 1 mM CaCl 2 ) and 150 l of N-benzyloxycarbonyl-Gly-Pro-Arg-p-nitroanilide trypsin substrate or N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide chymotrypsin substrate, dissolved in assay buffer, was added to a final concentration of 0.14 and 0.15 mM, respectively. After recording the absorbance change at 405 nm for 1 min in a microplate reader, 0.032 nM recombinant human enteropeptidase (R & D Systems, Minneapolis, MN) or 10 nM cationic trypsin was added (final concentrations). After a 1-min (enteropeptidase) or 5-min (trypsin) incubation, the rate of p-nitroaniline release was measured again for 1 min. Zymogens were further purified from high activity fractions by ecotin affinity chromatography, as reported previously (18).
Expression of Recombinant Protease Zymogens in the E. coli Rosetta (DE3) Strain-The pTrap-T7 expression plasmids harboring the PRSS1, PRSS2, and ELA2A genes were described earlier (9,10,19,20). Note that in this expression plasmid, the sequence encoding the secretory signal peptide is deleted, and a codon for an initiator methionine is placed before the mature zymogen sequence. The PRSS1 D218Y and D218S mutants and the PRSS2 Y218D mutant were constructed previously (15). The cDNA for CTRC was PCR-amplified from IMAGE clone 5221216 (GenBank TM accession number BI832476; this corresponds to the reported HC1 variant with Arg 80 instead of Trp 80 ) with the sense primer 5Ј-GCC TGT GCC ATG GCT TGT GGG GTG CCC AGC TTC CCG CCC AAC-3Ј and antisense primer 5Ј-CCC AGC GTC GAC TCA CAG CTG CAT TTT CTC GTT GAT CC-3Ј. The sense primer introduces a Met-Ala sequence in place of the secretory signal peptide of CTRC. The cDNA for proelastase 3A (ELA3A) was PCRamplified from IMAGE clone 3950453 (GenBank TM accession number BC007028), using the sense primer 5Ј-GCC GTT GCC ATG GCT GGC TAT GGC CCA CCT TCC TCT CAC TCT TCC-3Ј and the antisense primer 5Ј-TTG GTT GTC GAC TTA GTG GCT TGC TAT GGT CTC CTC AAT CC-3Ј. The sense primer replaces the secretory signal peptide of ELA3A with a Met-Ala-Gly sequence, and the antisense primer changes the amber stop codon (TAG) to ochre (TAA). The CTRC and ELA3A PCR products were digested with NcoI and SalI restriction enzymes and cloned under the control of the T7 promoter in the pTrap-T7 expression plasmid.
Recombinant protease zymogens were expressed in the E. coli Rosetta(DE3) strain as cytoplasmic inclusion bodies. Small scale expression and in vitro refolding of the zymogens was carried out as reported previously (9, 10), except that 6 M guanidine-HCl was used to solubilize the proelastases and chymotrypsinogen C from inclusion bodies.
Expression of Recombinant Human Cationic Trypsinogen with an Authentic N Terminus in the E. coli LG-3 Strain-Trypsinogens expressed from the pTrapT7 plasmid in E. coli Rosetta(DE3) contain a mixture of abnormal N termini, which consists of ϳ70% Met-Ala 16 -Pro 17 -Phe 18 -and ϳ30% Pro 17 -Phe 18 -sequences. Recently, we developed an E. coli expression system that was designed to produce recombinant trypsinogens with a homogenous, intact N-terminal sequence of Ala 16 -Pro 17 -Phe 18 -. The method involves the use of a self-splicing miniintein fused in frame with the N terminus of human cationic trypsinogen and a newly engineered E. coli strain deficient in aminopeptidase P, designated E. coli LG-3. Details of the construction of the intein-trypsinogen fusion construct and the LG-3 strain are described elsewhere (14). Wild-type cationic trypsinogen and mutants A16V, N29I, and R122H were expressed as intein fusions in the LG-3 strain. After C-terminal self-cleavage of the intein, trypsinogens with the native N terminus (Ala 16 -Pro 17 -Phe 18 -) accumulated in inclusion bodies. In vitro refolding and purification of trypsinogens was then carried out according to the protocol described previously (9,10,18).
Expression of Human Cationic Trypsinogen in Human Embryonic Kidney (HEK) 293T Cells-The PRSS1 cDNA was PCR-amplified from IMAGE clone 6217518 (GenBank TM accession CA778152) and cloned into the pcDNA3.1(Ϫ) plasmid using the XhoI and BamHI restriction sites. Mutation A16V was generated by site-directed mutagenesis. HEK 293T cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum, 1% penicillin, and 4 mM L-glutamine. Cells were grown in the same medium but without penicillin before transfection in 75-cm 2 flasks to 95% confluence. Transfections were carried out in Opti-MEM I reduced serum medium with 2 mM L-glutamine (Invitrogen) using 32 g of plasmid DNA and 80 l of Lipofectamine 2000 TM reagent (Invitrogen). After 5 h, the medium was supplemented with Dulbecco's modified Eagle's medium and fetal bovine serum to a 10% final concentration. After 24 h, cells were washed with Opti-MEM containing 2 mM L-glutamine and 1 mM benzamidine and covered with 20 ml of the same medium. Conditioned medium was then harvested and replenished with fresh medium every other day for 10 -12 days. Collected media were pooled (100 -120 ml), and trypsinogens were purified by ecotin affinity chromatography (18).
Autoactivation assays and calculation of initial rates using progress curve analysis with KINSIM and FITSIM computer programs (21,22) were described in Ref. 15. Autoactivation reactions contained 2 M trypsinogen and were initiated by the addition of 10 nM trypsin (final concentrations). The initiating trypsin was always prepared from the corresponding trypsinogen preparation.
The rate of N-terminal processing of trypsinogens by CTRC, CTRB, ELA3A, and ELA2A was measured using SDS-PAGE. Zymogens were activated in 0.1 M Tris-HCl (pH 8.0) and 10 mM CaCl 2 with 10 nM trypsin (final concentration). Trypsinogens (2 M concentration) were incubated with the indicated concentrations of active proteases in 0.1 M Tris-HCl (pH 8.0), 1 mM CaCl 2 , and 1 mM benzamidine. Aliquots (100 l) were withdrawn from the activation mixtures, and trypsinogen was precipitated with trichloroacetic acid (10% final concentration). The precipitate was recovered by centrifugation, dissolved in Laemmli sample buffer, and heat-denatured at 95°C for 5 min in the absence of a reducing agent. Electrophoretic separation was performed on 13% SDS-PAGE minigels in standard Tris-glycine buffer, and gels were stained with Brilliant Blue R.

An N-terminally Processed Cationic Trypsinogen Variant in Pancre-
atic Juice Is Associated with Increased Autoactivation-We have isolated cationic and anionic trypsinogens from three deidentified human pancreatic juice samples (PS7, PS13, and PS19). Our aim was to compare properties of autoactivation of native human trypsinogens to those of recombinant trypsinogens produced in E. coli. Unexpectedly, autoactivation of cationic trypsinogen isolated from PS19 was significantly increased relative to cationic trypsinogen from PS7, PS13, or from E. coli, which were comparable. Fig. 1A demonstrates time courses of autoactivation at pH 8.0, and Fig. 1B shows the pH dependence of the rate of autoactivation between pH 4.0 and 8.0. The increased autoactivation of cationic trypsinogen from PS19 was detectable between pH 6.0 and 8.0 and showed a maximum at pH 7.0. In contrast, anionic trypsinogen isolated from PS19 autoactivated slower than anionic trypsinogen from PS7, PS13, or E. coli; however, the difference was not significant (see Fig. 1C for time courses at pH 8.0 and Fig. 1D for pH dependence). SDS-PAGE analysis of native and recombinant trypsinogens under reducing and nonreducing conditions revealed that native trypsinogens purified from PS19 are heterogeneous and exhibit two bands (Fig. 2). The doublet was best appreciated under nonreducing conditions in the 13% polyacrylamide gel used. The two cationic trypsinogen bands were transferred to polyvinylidene difluoride membrane and subjected to Edman degradation, which yielded two distinct N-terminal sequences (Fig. 2). One of the sequences corresponded to the N terminus of intact, mature cationic trypsinogen starting with Ala 16 , whereas the other sequence showed a truncated N terminus, with the Ala 16 -Pro 17 -Phe 18 tripeptide missing. N-terminal sequencing of the anionic trypsinogen doublet revealed the same proteolytic modification. Interestingly, the shortened N-terminal sequence was also observed by Guy et al. (23), when the N-terminal sequences of native human trypsinogens were first determined (23). Clearly, a so far unidentified proteolytic activity in pancreatic juice cleaves off the N-terminal tripeptide of human trypsinogens. Remarkably, N-terminal processing stimulates autoactivation of cationic trypsinogen, whereas anionic trypsinogen remains largely unaffected.
Chymotrypsin C Activity Is Responsible for Stimulation of Autoactivation of Human Cationic Trypsinogen-The cleavage after Phe 18 suggested that a chymotrypsin-like enzyme was responsible for the observed  N-terminal processing of human trypsinogens. Therefore, we set out to identify this enzyme in human pancreatic juice. We based our experimental approach on the reasonable assumption that the unknown enzyme would cleave the synthetic chromogenic chymotrypsin substrate N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide, which is analogous to the N terminus of human trypsinogens in the P1-P2-P3 positions. Furthermore, it seemed likely that the unknown enzyme was present or activated in PS19, but not in PS7 and PS13, and thus comparative analysis of the three juices would be informative. The juice samples were loaded onto a MonoQ anion exchange column and eluted with a NaCl gradient (0 -0.5 M). The flow-through and eluted fractions were collected and analyzed for the presence of chymotrypsin activity before and after incubation with low concentrations of trypsin, which was added to activate proenzymes. Trypsin activity was also measured in each fraction before and after activation of trypsinogens with enteropeptidase. Fig. 3 demonstrates the elution (UV absorbance at 280 nm) and enzyme activity profiles for all three juice samples. Before activation with trypsin, chymotrypsin activity was negligible in the PS7 and PS13 fractions but was clearly detectable in the flow-through and fractions 18 and 19 from PS19 (see orange trace in Fig. 3A). After activation with trypsin, three distinct chymotrypsin activity peaks emerged in all three juice samples, corresponding to the flow-through (peak I) and to fractions 14 and 15 (peak II) and fractions 18 and 19 (peak III). No spontaneous trypsin activity was observed in any of the chromatography fractions from the three juice samples. After activation with enteropeptidase, two trypsin activity peaks appeared, which represented cationic trypsinogen (fractions 24 -26) and anionic trypsinogen (fractions [27][28][29]. The presence of chymotrypsin activity in peak III of PS19 even without activation with trypsin, suggested that the chymotrypsin-like enzyme in peak III was responsible for the N-terminal processing of human trypsinogens. To investigate this notion, ecotin affinity chromatography was used to purify the putative chymotrypsinogens from peaks I-III. The purified proteins were electrophoresed on 13% minigels, transferred to polyvinylidene difluoride membrane, and subjected to N-terminal protein sequencing. Peak II contained a single prominent band, which was identified as chymotrypsinogen B. Peak III exhibited three bands: an intense upper band corresponding to chymotrypsinogen C, a faint middle band that contained proelastase 3A, and a relatively weak lower band, which was chymotrypsinogen B. Finally, peak I purified from the MonoQ flow-through also contained all three bands, with chymotrypsinogen B as the predominant component (Fig. 4A). The chymotrypsinogen C band in peak III exhibited a fuzzy appearance on gels, suggesting that the protein might be glycosylated. Indeed, treatment with peptide:N-glycosidase F (New England Biolabs) resulted in a mobility shift on gels, whereas the mobility of chymotrypsinogen B was unaffected (not shown). The addition of trypsin also caused small but detectable mobility shifts for all proteins, indicating that the zymogens were activated by trypsin (not shown). Activity assays confirmed that trypsin-mediated activation resulted in the development of chymotrypsin activity in all three ecotin-affinity purified MonoQ peaks.
To test the effect of the purified chymotrypsins on trypsinogen function, autoactivation of recombinant cationic trypsinogen was measured in the presence of proteins from peaks I-III. Clearly, peak III stimulated autoactivation to a significant extent, whereas peak I was only marginally active, and peak II was devoid of stimulatory activity (Fig. 4B).
Efforts to purify chymotrypsinogen C completely free from the two small contaminating proteins (i.e. chymotrypsinogen B and proelastase 3A) present in peak III were unsuccessful; therefore, we expressed chymotrypsinogen C and proelastase 3A recombinantly in E. coli. To rule out that human elastase 2A, which may also cleave phenylalanyl peptide bonds, can process the N terminus of trypsinogens, recombinant proelastase 2A was also made. Fig. 4C demonstrates that recombinant human chymotrypsin C markedly stimulated autoactivation of cationic trypsinogen, whereas ELA3A and ELA2A had no stimulatory activity. Maximal chymotrypsin C-induced stimulation of the rate of autoactivation was ϳ3-fold at pH 8.0.
Taken together, the results clearly establish that human chymotrypsin C specifically stimulates autoactivation of human cationic trypsinogen, whereas chymotrypsin B, elastase 3A, and elastase 2A are ineffective.
Chymotrypsin C Excises the N-terminal Tripeptide of Human Trypsinogens-To confirm that stimulation of autoactivation by chymotrypsin C is mediated through N-terminal processing of cationic trypsinogen, the effect of purified peaks II and III as well as recombinant proteases was analyzed on SDS-polyacrylamide gels under nonreducing conditions, which clearly resolved the trypsinogen, N-terminally truncated trypsinogen and trypsin bands. For these experiments, three different cationic trypsinogen preparations were used: native trypsinogen from pancreatic juice and recombinant trypsinogens expressed in the aminopeptidase P-deficient E. coli LG-3 strain or in HEK 293T cells (see "Experimental Procedures"). Trypsinogen produced in E. coli LG-3 or in After injection of the 4-ml sample volume, the column was developed with a linear gradient of 0 -0.5 M NaCl in 20 mM Tris-HCl (pH 8.0) at a flow rate of 1 ml/min. The elution profile of proteins was followed by UV absorbance at 280 nm, indicated by the black dashed line. Chymotrypsin activity was determined using the synthetic chromogenic substrate N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide before (solid orange line) and after (solid red line) activation with trypsin, as described under "Experimental Procedures." The three peaks with chymotrypsin-like activity were designated I, II, and III. There was no measurable chymotrypsin activity in the PS7 and PS13 fractions before activation with trypsin. Trypsin activity was assayed by N-benzyloxycarbonyl-Gly-Pro-Arg-p-nitroanilide hydrolysis after activation by enteropeptidase (solid blue line). There was no detectable trypsin activity before enteropeptidase activation. The enzyme activities are indicated as absorbance change at 405 nm in 1 min (mOD/min).
HEK cells accurately mimics the intact N terminus of the native proenzyme (Ala 16 -Pro 17 -Phe 18 -). As shown in Fig. 5A, native chymotrypsin C from peak III proteolyzed the N terminus of cationic trypsinogen, and the rate of processing was comparable for native and recombinant cationic trypsinogens from HEK cells or E. coli LG-3. Similarly, recombinant chymotrypsin C also efficiently removed the N-terminal tripeptide from cationic trypsinogen, confirming the identity of the active component in peak III (Fig. 5C). In sharp contrast, chymotrypsin B isolated from peak II (Fig. 5B) or recombinant elastase 3A and elastase 2A (Fig.  5C) were completely devoid of processing activity.
Although not shown, we also observed that cationic trypsinogen produced in E. coli Rosetta(DE3) was processed by chymotrypsin C at a markedly increased rate (ϳ20-fold). The enhanced processing is probably due to the extra N-terminal methionine residue (Met-Ala 16 -Pro 17 -Phe 18 -) present in the bulk of trypsinogen expressed in E. coli Rosetta(DE3). This unexpected observation offered the unique opportunity to use this type of recombinant trypsinogen preparation in experiments when rapid and complete processing was desirable (e.g. to examine the functional consequences of N-terminal processing).
N-terminal Processing of Cationic Trypsinogen by Chymotrypsin C Relieves the Inhibitory Effect of Asp 218 on Autoactivation-The findings described above establish a novel regulatory mechanism through which autoactivation of cationic trypsinogen is controlled by chymotrypsin C. Furthermore, the experiments presented thus far raise the exciting question of why N-terminal trimming of the trypsinogen activation peptide FIGURE 4. Effect of native and recombinant pancreatic proteases with chymotrypsin or elastase activity on the autoactivation of cationic trypsinogen. Ecotin affinity chromatography was carried out to purify the protease zymogens from the three MonoQ peaks with chymotrypsin activity (see Fig. 3). Peak I was from the flow-through fractions 2-6, peak II contained fractions 13-16, and peak III was pooled from fractions 18 -20. In the experiments presented here, zymogens purified from PS13 were used. Identical results were obtained with zymogens purified from juice sample PS7 (not shown). Recombinant cationic trypsinogen, chymotrypsinogen C, proelastase 2A and proelastase 3A were expressed in E. coli Rosetta(DE3) and purified with ecotin affinity chromatography. A, SDS-PAGE analysis of the ecotin affinity-purified peaks I, II, and III from pancreatic juice. Results of N-terminal sequencing are indicated. The N-terminal sequences correspond to chymotrypsinogen C (upper band), proelastase 3A (middle band), and chymotrypsinogen B (lower band). The Cys residues in the sequences were inferred from their expected positions. B, trypsin-mediated trypsinogen activation was carried out as described in the legend to Fig. 1A in the absence or presence of ϳ20 nM purified peak fractions (final concentrations). Protein concentrations were estimated based on the UV absorbance at 280 nm, using the theoretical extinction coefficient of the predominant zymogen, as described under "Experimental Procedures." Because activation reactions contained 10 nM initial trypsin concentration, peak fractions were added directly as zymogens. Identical results were obtained when fractions were first preactivated with 10 nM trypsin (not shown). The rates of autoactivation calculated from progress curve analysis were as follows: control (open circles), 1.7 nM/min; peak I (solid diamonds), 2 nM/min; peak II (solid triangles), 1.7 nM/min; peak III (inverted solid triangles), 3.4 nM/min. C, the effect of purified recombinant pancreatic proteases on autoactivation of cationic trypsinogen. Initial concentrations of reactants were 2 M cationic trypsinogen, 10 nM cationic trypsin, and 40 nM CTRC, ELA3A, or ELA2A, added as purified zymogens. Autoactivation was measured as described in the legend to  Benzamidine was included in the reaction to prevent autoactivation of trypsinogens. Nonreducing SDS-PAGE was used to separate the two trypsinogen forms on 13% polyacrylamide gels, which adequately resolved the intact trypsinogen, the N-terminally processed trypsinogen, and the trypsin bands, as indicated. A, N-terminal processing by chymotrypsin C from peak III. Trypsinogens purified from pancreatic juice (native) or expressed recombinantly in human embryonic kidney 293T cells (HEK) or in E. coli LG-3 (LG) were incubated with chymotrypsin C from peak III at ϳ40 nM final concentration. B, lack of N-terminal processing of cationic trypsinogen (from E. coli LG-3) by chymotrypsin B from peak II (40 nM final concentration). C, N-terminal processing of cationic trypsinogen by recombinant pancreatic proteases. Recombinant trypsinogen expressed in E. coli LG-3 (LG) was used at 2 M concentration. CTRC was at 30 nM concentration, ELA3A was at 100 nM, and ELA2A was at 100 nM. See "Experimental Procedures" for details. Chymotrypsinogen B and C from peaks II and III, respectively, were purified from juice sample PS13 for the experiments presented here. Identical results were obtained when zymogens purified from juice sample PS7 were used (not shown).
increases autoactivation in a manner that is specific for cationic trypsinogen. In this context, we recently identified a unique inhibitory interaction in human cationic trypsinogen, which involves an electrostatic repulsion between Asp 218 on trypsin and Asp 21 within the tetra-aspartate motif (Asp 19 -22 ) of the trypsinogen activation peptide (15). Asp 218 is not present in the large majority of vertebrate trypsins, and typically a Tyr residue occupies the same position. Human anionic trypsinogen also contains a Tyr at position 218, and, consequently, the electrostatic inhibitory mechanism is not operational in this isoform. The differential effect of the chymotrypsin C-mediated N-terminal processing on the two human trypsinogens suggested that disruption of the Asp 218 -Asp 21 interaction might be the mechanism underlying autoactivation stimulation. To test this notion, cationic trypsinogen mutant D218Y was treated with chymotrypsin C, and autoactivation was measured. Fig. 6, A and B, show that mutation of Asp 218 to Tyr (D218Y) abolished the stimulatory effect of chymotrypsin C. Although not shown, we also tested the D218S mutant, which contains a different uncharged residue (Ser) at position 218, and, again, no stimulation was observed. In the converse experiment, Asp 218 was introduced into human anionic trypsinogen (mutant Y218D). Incubation of wild-type anionic trypsinogen with chymotrypsin C had no effect on autoactivation (Fig. 6C), whereas autoactivation of the Y218D mutant was notably stimulated (Fig.  6D). In conclusion, the observations provide compelling evidence that chymotrypsin C stimulates autoactivation of human cationic trypsinogen through alleviation of the Asp 218 -mediated inhibition.
The Pancreatitis-associated Mutation A16V Increases N-terminal Processing of Cationic Trypsinogen-The observations presented above raise the possibility that chymotrypsin C-mediated stimulation of autoactivation might play a role in the pathogenesis of human pancreatitis, where premature trypsinogen activation is believed to be an early, initi-ating event. To obtain support for this theory, we have tested the effect of the three cationic trypsinogen mutations that are most frequently found in human hereditary pancreatitis. Previous studies have demonstrated that mutations N29I and R122H increase autoactivation of human cationic trypsinogen (9 -11); however, mutation A16V has no such effect (14). In the experiments presented in Fig. 7, chymotrypsin C-mediated N-terminal processing of wild-type cationic trypsinogen and mutant A16V was followed on SDS-PAGE under nonreducing conditions. Remarkably, chymotrypsin C processed the A16V mutant 4-fold more rapidly than wild-type cationic trypsinogen. Recombinant trypsinogens expressed either in E. coli LG-3 or HEK cells were tested, with identical results. Analysis of mutation N29I revealed a small but reproducible (1.2-1.4-fold) stimulation in N-terminal processing, whereas mutation R122H had no effect (not shown). The results strongly suggest that stimulation of autoactivation of cationic trypsinogen by chymotrypsin C plays a role in chronic pancreatitis associated with the A16V mutation.

DISCUSSION
In this study, we demonstrated that proteolytic removal of the N-terminal tripeptide from human cationic trypsinogen by chymotrypsin C results in marked stimulation of autoactivation. The chymotrypsin C-processed cationic trypsinogen becomes a better substrate for trypsin due to the disruption of the inhibitory interaction between Asp 218 on cationic trypsin and Asp 21 within the conserved tetra-Asp motif of the trypsinogen activation peptide. Presumably, N-terminal truncation of cationic trypsinogen by chymotrypsin C results in a conformational change within the remainder of the activation peptide, which repositions Asp 21 and thereby mitigates the Asp 21 -Asp 218 electrostatic repulsion. Previous work demonstrated that complete abolition of the Asp 21 -Asp 218 interaction by replacing Asp 218 with Tyr resulted in an ϳ11-fold increase in autoactivation (15). Therefore, the 3-fold increase in autoactivation upon chymotrypsin C treatment indi-catesthatthelossoftheN-terminaltripeptideonlypartlyrelievestheAsp 218dependent inhibition. Chymotrypsin C also cleaves off the N-terminal tripeptide from human anionic trypsinogen, but the proteolytic processing has  no significant effect on the autoactivation of this trypsinogen isoform, which contains a Tyr in place of Asp 218 . However, if an Asp residue is artificially introduced into anionic trypsinogen in place of Tyr 218 , autoactivation of the resulting mutant becomes stimulated by chymotrypsin C.
Chymotrypsin C was first isolated from pig pancreas as an anionic chymotrypsin that exhibited protease activity distinct from the cationic bovine or porcine chymotrypsin A or the anionic bovine chymotrypsin B (24). Thus, chymotrypsin C displayed a broad specificity toward tyrosyl, phenylalanyl, methionyl, tryptophanyl, leucyl, glutaminyl, and asparaginyl peptide bonds but, characteristically, hydrolyzed leucyl peptide bonds in synthetic and natural substrates with significantly higher activity than chymotrypsin A or B (24,25). In the bovine pancreas, chymotrypsinogen C was found in a binary complex with procarboxypeptidase A or in ternary complex with procarboxypeptidase A and proproteinase E (Refs. 26 and 27 and references therein). The crystal structure of the ternary complex has been determined (28,29). Chymotrypsin C is almost certainly the same protein as caldecrin, a serum calcium-decreasing protein isolated from porcine and rat pancreas and later cloned from rat and human pancreas (30 -34). Porcine, rat, and human caldecrins have chymotrypsin-like activity after activation with trypsin and exhibit a very high sequence identity to bovine chymotrypsin C (75-81%), suggesting that these proteins represent orthologs of the same digestive enzyme. The protease activity and the effect on serum calcium appear to be distinct and unrelated functions, although both require activation of the zymogen by trypsin. The gene for human chymotrypsin C (the CTRC gene) is located on chromosome 1, in the vicinity of the proelastase 2A and 2B (ELA2A and ELA2B) genes. Indeed, the extent of homology between human chymotrypsin C and elastase 2A is higher (64%) than between chymotrypsin C and chymotrypsin B (43%).
The discovery that chymotrypsin C stimulates autoactivation of human cationic trypsinogen establishes a novel positive feedback mechanism, which facilitates activation of human trypsinogens in the digestive enzyme cascade of humans. Activation of secreted pancreatic trypsinogens in the duodenum is initially catalyzed by enteropeptidase. The active trypsin can in turn activate trypsinogen (autoactivation) and other pancreatic protease zymogens, including chymotrypsinogen C. Active chymotrypsin C can process the N terminus of still unactivated trypsinogens and thereby enhance autoactivation of the most abundant cationic isoform. As a result, overall conversion of trypsinogen to trypsin is accelerated, and the digestive protease capacity in the chyme may be increased. Chymotrypsin C appears to be universally expressed in the pancreas of various species; however, the stimulatory effect described here is probably human-specific due to the conspicuous absence of Asp 218 in vertebrate trypsinogens. In addition, the N-terminal Ala 16 -Pro 17 -Phe 18 -sequence is also unique to human trypsinogens, and other N termini may not be good substrates for chymotrypsin C.
The observation that the pancreatitis-associated mutation A16V markedly increases N-terminal processing of cationic trypsinogen by chymotrypsin C provides compelling evidence that the chymotrypsin C-mediated stimulation of autoactivation plays a role in the pathogenesis of human pancreatitis. Inappropriate intrapancreatic trypsinogen activation is an early event in pancreatitis, and cationic trypsinogen mutations that in vitro accelerate autoactivation are frequently associated with hereditary pancreatitis. The A16V mutation was first reported in 1999, and almost 10% of pediatric patients with chronic pancreatitis were shown to be heterozygous carriers (35). In follow-up studies, the mutation was also identified in adult patients with chronic pancreatitis, and to date ϳ25 patients (pediatric and adult) have been documented worldwide (6, 36 -38). Interestingly, in contrast to the more frequent N29I and R122H mutations, which show autosomal dominant inheritance, the A16V mutation was more often found in sporadic cases with no family history. This suggests that this mutation might have a somewhat different mechanism of action than N29I or R122H. Since the A16V mutation affects the first amino acid of the activation peptide, which forms part of the signal peptide cleavage site, it was proposed that the mutation might cause defective intracellular transport of cationic trypsinogen. However, such a mechanism seems unlikely, because the signal peptide cleavage reaction is typically not influenced by the amino acid distal to the cleavage site, and valine is frequently found in this position in other secretory proteins, including human mesotrypsinogen. Furthermore, we measured the rate of trypsinogen secretion from transiently transfected HEK 293T cells and found no difference between wild-type and A16V mutant cationic trypsinogens (not shown), indicating that signal peptide cleavage is not affected by the mutation. Recently, we have characterized the effect of mutation A16V on autoactivation of cationic trypsinogen using recombinant trypsinogens produced in the aminopeptidase P-deficient E. coli LG-3 strain as self-splicing inteintrypsinogen fusions. In contrast to the previously characterized N29I and R122H variants, mutation A16V had no stimulatory effect on autoactivation, and even a slight inhibition was observed (14). The present study offers a plausible mechanistic explanation for the pathogenic action of the A16V mutation, which is fundamentally similar to the effect of the other mutations and yet differs substantially enough to account for the different clinical phenotype. Thus, mutation A16V stimulates autoactivation of cationic trypsinogen but does it indirectly by means of chymotrypsin C. Consequently, some degree of premature trypsinogen activation followed by activation of chymotrypsinogen C must occur before the detrimental effect the A16V mutation is manifested. This stands in contrast to N29I or R122H and a handful of other mutations, which directly increase autoactivation of cationic trypsinogen and lead to the development of pancreatitis with a higher probability.
From the three juice samples studied here, only one (PS19; see Fig. 3) contained spontaneous chymotrypsin C activity and N-terminally processed trypsinogens. It remains unclear why only in this particular sample was chymotrypsinogen C prematurely activated. No clinical information was available with the deidentified juice samples obtained for this study. To address this question, characterization of chymotrypsin C activity and trypsinogen processing in juice samples from patients with well defined pancreatic pathologies will be necessary.