GASTROENTEROLOGY 2010;138:726–737

Cathepsin L Inactivates Human , Whereas -Deletion Reduces the Severity of in Mice

THOMAS WARTMANN,* JULIA MAYERLE,‡ THILO KÄHNE,§ MIKLÓS SAHIN–TÓTH,ʈ MANUEL RUTHENBÜRGER,‡ RAINER MATTHIAS,* ANNE KRUSE,‡ THOMAS REINHECKEL,¶ CHRISTOPH PETERS,¶ F. ULRICH WEISS,‡ MATTHIAS SENDLER,‡ HANS LIPPERT,* HANS–ULRICH SCHULZ,* ALI AGHDASSI,‡ ANNEGRET DUMMER,‡ STEFFEN TELLER,‡ WALTER HALANGK,* and MARKUS M. LERCH‡

*Division of Experimental Surgery, Department of Surgery, Otto-von-Guericke-Universität Magdeburg, Magdeburg; ‡Department of Medicine A, Ernst-Moritz-Arndt- Universität Greifswald, Greifswald; §Institute of Experimental Internal Medicine, Department of Internal Medicine, Otto-von-Guericke-Universität Magdeburg, Magdeburg, Germany; ʈDepartment of Molecular and Cell Biology, Boston University, Goldman School of Dental Medicine, Boston, Massachusetts; and ¶Institut für Molekulare Medizin und Zellforschung, Albert-Ludwigs-Universität Freiburg, Freiburg, Germany

BACKGROUND & AIMS: is charac- notable example is trypsinogen, which reaches the small terized by an activation cascade of digestive in the intestine as precursor zymogen, is activated by brush- pancreas. The first of these, trypsinogen, can be converted to border enterokinase, and then acts as the master active by the peptidase (CTSB). We in activating other digestive . A pivotal role of investigated whether cathepsin L (CTSL) can also process trypsin in the initiating events of pancreatitis can be trypsinogen to active trypsin and has a role in pancreatitis. assumed because trypsin mutations are the most com- METHODS: In CTSL-deficient (CtslϪ/Ϫ) mice, pancreatitis mon autosomal-dominant changes associated with pan- was induced by injection of cerulein or infusion of tauro- creatitis.1–3 In the absence of enterokinase in the pan- cholate into the pancreatic duct. Human tissue, pancreatic creas, other mechanisms must be operative through juice, mouse pancreatitis specimens, and recombinant en- which a premature and intrapancreatic activation of zymes were studied by enzyme assay, immunoblot, N-ter- trypsinogen can be initiated. One of the best documented minal sequencing, immunocytochemistry, and electron mi- of these mechanisms is trypsinogen activation by cathep- Ϫ Ϫ Ϫ Ϫ croscopy analyses. Isolated acini from Ctsl / and Ctsb / sin B (CTSB), a lysosomal . Biochemically, CTSB mice were studied. RESULTS: CTSL was expressed in hu- has long been shown to be a trypsinogen activator,4 and, man and mouse pancreas, colocalized with trypsinogen in in experimental studies, it was demonstrated that inhi- secretory vesicles and , and secreted into pancre- bition of CTSB or genetic deletion of the ctsb gene Ϫ Ϫ atic juice. Severity of pancreatitis was reduced in Ctsl / protects not only against premature trypsinogen activa- ACES AND PANCREAS,

IIR TRACT BILIARY mice, whereas apoptosis and intrapancreatic trypsin 5

BASIC–LIVER, tion but also against pancreatitis. Although the role of activity were increased. CTSL-induced cleavage of CTSB in trypsinogen activation and pancreatitis is now trypsinogen occurred 3 amino acids toward the C-termi- firmly established,6–8 little is known about a potential nus from the CTSB activation site and resulted in a role of other lysosomal enzymes. truncated, inactive form of trypsin and an elongated Cathepsin L (CTSL) is another member of the propeptide (trypsinogen activation peptide [TAP]). This family of cysteine proteinases with similar enzymatic elongated TAP was not detected by enzyme-linked im- properties. CTSL exhibits a much stronger endoproteo- munosorbent assay (ELISA) but was effectively converted lytic activity than CTSB9 and could therefore, if it were to an immunoreactive form by CTSB. Levels of TAP thus expressed in the pancreas, be an even more important generated by CTSB were not associated with disease se- regulator of protease activation in pancreatitis. Recently verity, although this is what the TAP-ELISA is used to established functions of CTSL include distinct steps in determine in the clinic. CONCLUSIONS: CTSL inacti- major histocompatibility complex processing, the matu- vates trypsinogen and counteracts the ability of CTSB ration of enkephalin in neuroendocrine cells, and the to form active trypsin. In mouse models of pancreati- degradation and recycling of growth factor receptors in tis, absence of CTSL induces apoptosis and reduces keratinocytes.10–12 When released into the cytosol, CTSL disease severity. has been suggested to regulate apoptosis and to process nuclear transcription factors.13 cute pancreatitis is thought to be caused by autodi- gestion of the pancreas by its own digestive enzymes. Abbreviations used in this paper: CTSB, cathepsin B; CTSL, L; TAP, trypsinogen activation peptide. Physiologically, most digestive proteases are secreted as © 2010 by the AGA Institute precursor zymogens and only acquire activity after cleav- 0016-5085/10/$36.00 age and activation by proteolytic processing. The most doi:10.1053/j.gastro.2009.10.048 February 2010 CATHEPSIN L IN PANCREATITIS 727

In the present study using human and mouse material, we found that CTSL is abundantly expressed in the exocrine pancreas, sorted into the lysosomal as well as the secretory pathway of acinar cells, and secreted into pan- creatic juice. In mice, in which the ctsl gene was deleted, experimental pancreatitis was significantly less severe and involved a dramatic shift to cell injury via apoptosis. Surprisingly, and in complete contrast to CTSB, CTSL was found to very effectively inactivate trypsinogen and trypsin in vivo, in isolated acini, and in vitro.

Materials and Methods Please refer to Supplementary Materials and Methods for more detailed descriptions.

Induction of Experimental Pancreatitis in Cathepsin L-Deficient Mice CTSL-deficient (CtslϪ/Ϫ) mice were generated by gene targeting in embryonic day 14 mouse embryonic stem cells as described by Roth et al.14 The CtslϪ/Ϫ mice lack CTSL activity in all organs but do not show pheno- typic alterations of the pancreas (Figure 1). Cerulein pancreatitis was induced by 7 hourly injections of supra- maximal (50 ␮g/kg/body weight/intraperitoneal) cer- ulein injections and taurocholate-induced pancreatitis by infusion of 50 ␮L 2% taurocholate into the pancreatic 5,15 duct as previously described. Figure 1. CTSL in the human and mouse pancreas and the CtslϪ/Ϫ pancreas. In panel A, by Western blots with CTSL antibody pro-CTSL, Preparation of Serum and Tissue Samples single and heavy chain CTSL in human pancreas and pancreatic juice were detected. Panel B indicates expression and subcellular localiza- Mice were killed at intervals between 1 and 24 tion of CTSL in normal human pancreas (red fluorescence) and panel C hours after the first injection of cerulein and 6 hours colocalization of trypsinogen (green) with CTSL (red) in normal mouse after intraductal taurocholate application. Blood and tis- pancreas. The pancreas of CtslϪ/Ϫ mice appears completely normal on sue were harvested as previously reported.5 Pancreatic H&E-stained sections (D, with an islet and duct at the top) and electron microscopy (E, with normal arrangement of zymogen granules and acini were prepared using purified (Serva, mitochondria). Bars indicate 50 ␮m and asterisks the acinar lumen. Heidelberg, Germany).5,15 BASIC–LIVER, BILIARY TRACT PANCREAS, AND Biochemical Assays Trypsin and trypsinogen after enterokinase ac- tivation were measured at 37°C using 64 ␮mol/L Human and Animal Material for Morphology BOC-Gln-Ala-Arg-7-amido-4-methylcoumarin as a sub- and Morphometry strate.5,15 Serum amylase and lipase activities were deter- Human material from donor pancreas as well as mined enzymatically by commercially available assays pancreatic juice from controls and pancreatitis patients (Roche-Hitachi, Basel, Switzerland). Trypsinogen activa- was obtained as previously described16 under an ethics tion peptide (TAP) was assayed using an enzyme immu- committee-approved protocol and with informed patient noassay (Biotrin, Dublin, Ireland), and protein concen- consent. For animal experiments, we collected tissue sam- trations were determined according to Bradford. Serum ples at selected time intervals of pancreatitis as previously cytokines were measured by fluorescence-activated cell reported.5,8 For experiments involving the detection of sorter analysis using a bead-based mouse immunoassay apoptotic cells, free 3’OH-DNA termini were labeled (Becton-Dickinson, Franklin Lakes, NJ). -3 activ- using the terminal deoxynucleotidyl-transferase (TdT) ity was measured fluorometrically using Rodamin110- method with fluorescein-labeled digoxigenin nucleo- DEVD as substrate (Invitrogen, Eugene, OR). Lung my- tides.5,8 Methods for electron microscopy and the quan- eloperoxidase was measured using O-dianisidine and tification of areas of necrosis by morphometry are de- 5 8 H2O2 as previously reported. scribed in detail elsewhere. 728 WARTMANN ET AL GASTROENTEROLOGY Vol. 138, No. 2

Proteolytic Processing of Trypsin and as well as heavy chain CTSL (Figure 1A). Detection of Trypsinogen by CTSB and CTSL CTSL in pancreatic juice is direct evidence that some

The human cationic trypsin (PRSSI1) expression CTSL is sorted into the secretory pathway of the pancreas plasmid was constructed as previously reported and and actively secreted into the ducts. Immunofluorescence expressed in Escherichia coli Rosetta (DE3).17 Bovine labeling of human pancreatic tissue from an organ donor trypsinogen, enterokinase, and CTSB were from Sigma- localized CTSL to an intracellular vesicle compartment in Aldrich and bovine trypsin and CTSL from Calbiochem acinar cells consistent with zymogen granules and lyso- (Bad Soden, Germany). For CTSL detection, we used somes (Figure 1B). Double labeling with trypsinogen in antibodies 33/2 and 3G10 (mouse monoclonal) kindly untreated mice showed a distinct staining of CTSL pre- provided by H. Kirschke/E. Weber (Halle, Germany). For dominately in the lysosmal compartment (Cy3 red) and trypsin(ogen) detection, we used trypsin antibody from of trypsinogen (FITC green) in zymogen granules (Figure Chemicon (clone AB1832A). Trypsin activity, trypsinogen 1C) with some colocalization of both enzymes (yellow content, and TAP (Biotrin) were determined as previously fluorescence). That subcellular redistribution increases rap- reported.5 Isolated pancreatic acini were prepared as de- idly during experimental pancreatitis (see below for Fig- scribed in detail elsewhere15 with Ile-Pro-Arg-rhodamin- ure 2). Untreated CtslϪ/Ϫ mice have a completely normal 110 as trypsin substrate, PhiPhiLux (Calbiochem) as appearing exocrine pancreas on either light or electron caspase-3 substrate, and propidium iodine exclusion microscopy (Figure 1D and E) (Roth, Karlsruhe, Germany) to quantitate necrosis. Localization of Cathepsin L and Pancreatic Enzyme Activity Measurements and Cleavage Injury Product Detection Induction of pancreatitis lead to an increased co- CTSL activity was determined using the fluoro- ␮ localization of trypsin (green, Figure 2A) and CTSL (red, genic substrate Z-Phe-Arg-AMC (64 mol/L final concen- Figure 2A) at the apical portion of acinar cells and in 8 tration). CTSL-treated trypsin and trypsinogen samples cytoplasmic vesicles. An additional presence of both en- or poly(ADP-ribose) polymorase cleavage during apopto- zymes in the cytosol could also not be excluded (lower sis was investigated by sodium dodecyl sulfate-polyacryl- tissue margin in Figure 2A). In subcellular fractionation amide (SDS-PAGE) and cleavage studies, a shift of CTSL from a -enriched to a products identified by Western blotting and N-terminal zymogen -enriched fraction was found (not sequencing.18 Mass spectrometry was used as previously shown). Both observations indicate that the colocaliza- described.18 CTLS in pancreatic juice was detected with tion of CTSL and trypsin increases during the early monoclonal anti-cathepsin L antibody (3G10, 1:1000 di- course of pancreatitis. On electron microscopy (Figure lution).8 2B) and light microscopy, wild-type and CtslϪ/Ϫ animals ACES AND PANCREAS,

IIR TRACT BILIARY developed morphologic signs of acute pancreatitis in-

BASIC–LIVER, Data Presentation and Statistical Analysis cluding acinar cell vacuolization, formation of auto- Data in graphs are expressed as means Ϯ standard phagic vesicles, and overt necrosis. When quantitated by error of mean from 5 or more experiments per group. Ϫ Ϫ morphometry, CTSL-deleted animals had less extensive Statistical comparison between the Ctsl / and the ϩ ϩ necrosis than wild-type animals (Figure 2C). When the Ctsl / groups at various time intervals was done by Student t test for independent samples using SPSS (SPSS, pancreas was stained for apoptotic cells by terminal Inc, Chicago, IL) for Windows (Microsoft Corp, Red- deoxynucleotidyl transferase-mediated deoxyuridine mond, WA). Differences were considered significant at a triphosphate nick-end labeling assay (Figure 2D), the level of P Ͻ .05. Data presentation was performed with result was reversed. CTSL-deleted animals developed a Origin 7.0 for Windows (OriginLab, Northampton, MA). much greater extent of acinar cell apoptosis (Figure 2F) than their wild-type littermates, and this difference was already significant in untreated control animals. We in- Results vestigated this further by quantitating the cleavage of Subcellular Localization and Sorting of CTSL PARP and the generation of caspase-3 activity in either Into the Secretory Pathway pancreatic homogenates from pancreatitis animals or in A prerequisite for a biologic relevance of CTSL in isolated acini following supramaximal cerulein stimula- pancreatitis would be its expression in the exocrine pan- tion (Figure 2E and F). In all experiments, apoptosis- creas and its colocalization with trypsinogen. All of these dependent mechanisms were up-regulated in the absence have previously been established for CTSB5,8 and could of CTSL. This indicates that CTSL is involved in the also be operative for CTSL because of great similarities cell-death pathways of pancreatic acinar cells including, between the 2 lysosomal proteases. Blotting CTSL in the when CTSL is deleted, a prominent shift from necrosis- pancreatic juice of mice and humans resulted in a strong dominant acinar cell injury to apoptosis during pancre- signal for pro-CTSL (31 kilodaltons), single chain CTSL, atitis. February 2010 CATHEPSIN L IN PANCREATITIS 729 BASIC–LIVER, BILIARY TRACT PANCREAS, AND

Figure 2. Cell injury and apoptisis in cerulein-induced pancreatitis. Panel A: 3 hours of pancreatitis in a wild-type animal labeled with antibody against trypsinogen (green) and CTSL (red fluorescence). (B) Electron microscopy of 3 hours pancreatitis in the CtslϪ/Ϫ mouse. Note whirl-like arrangement of the ER in lower right corner, large autophagic vacuoles, of which 1 contains a nucleus, and a necrotic cell in the top center. (C) Morphometry of acinar tissue necrosis in wild type and CtslϪ/Ϫ animals over 24 hours of pancreatitis. (D) 3-OH-nick-end labelling (TUNEL) of apoptotic nuclei in CtslϪ/Ϫ animals over 24 hours of pancreatitis. This difference is already present in untreated animals. (E) PARP cleavage and caspase-3 activity in pancreatic tissue after 8 hours of pancreatitis. (F) Caspase-3 activity in living pancreatic acini after 60 minutes of supramaximal cerulein stimulation. Bars indicate 1 ␮m. Values denote means Ϯ standard error of mean (SEM) for 5 or more measurements.

Serum Pancreatic Enzymes and Trypsinogen correspond to acinar cell injury.5 As previously reported Activation for CtsbϪ/Ϫ mice,5 the CTSL-deficient animals had a Induction of cerulein pancreatitis was followed by much milder disease course, ie, 33% lower amylase and a rapid and biphasic increase in serum activities of amy- 25% lower lipase activities. This corresponded not only to lase (Figure 3A) and lipase (Figure 3B), which is known to the lower extent of necrosis in exocrine tissue (Figure 2C) 730 WARTMANN ET AL GASTROENTEROLOGY Vol. 138, No. 2 ACES AND PANCREAS, IIR TRACT BILIARY BASIC–LIVER,

Figure 3. Disease severity and trypsinogen activation. Deletion of CTSL greatly decreased pancreatitis-associated hyperamylasemia (A) and hyperlipasemia (B) as well as serum levels of IL-6, MCP-1, and TNF-␣ (C) over 24 hours. (D) Trypsin activity and (E) TAP generation in pancreatic tissue homogenates over 24 hours of pancreatitis. In strong contrast to serum pancreatic enzyme levels and the extent of tissue necrosis (Figure 2), intrapancreatic trypsin activity is greatly increased in CtslϪ/Ϫ animals, whereas TAP levels are similar in both groups. Values denote means Ϯ SEM for 5 or more measurements. February 2010 CATHEPSIN L IN PANCREATITIS 731

but also to a greatly reduced inflammatory response of the animals as indicated by the reduction of serum in- terleukin-6, monocyte chemoattractant protein-1, and tumor necrosis factor ␣ (Figure 3C). Under most clinical and experimental conditions, the degree of disease severity in pancreatitis is paralleled by the extent of intrapancreatic trypsinogen activation. It therefore came as a complete surprise to find greatly increased free trypsin activities in the pancreas of CtslϪ/Ϫ mice during experimental pancreatitis (Figure 3D). The different trypsin activities in the pancreas were not due to different expression levels of trypsinogen because CtsbϪ/Ϫ, CtslϪ/Ϫ, and wild-type mice had comparable trypsinogen levels in the pancreas under resting condi- tions (not shown). The observation was made even more puzzling by the fact that recovery of TAP—the trypsino- gen activation peptide that is generated during activation of trypsinogen to trypsin—did not much differ between wild-type and CtslϪ/Ϫ animals (Figure 3E). This suggests that trypsinogen activation is not altered in CtslϪ/Ϫ mice, whereas the degradation of trypsinogen and trypsin is highly dependent on the presence of CTSL. It should be noted that the total tissue content of TAP in molar terms exceeded that of measurable trypsin activity by one order of magnitude. This excess of TAP over trypsin activity can indicate an inactivation of newly formed trypsin by CTSL Figure 4. Taurocholate-induced pancreatitis. CtslϪ/Ϫ and Ctslϩ/ϩ an- and other proteases15 or an inhibition of trypsin activity imals were killed after 6 hours of taurocholate-induced pancreatitis. On by endogenous trypsin inhibitors. On the other hand, morphology (A and B), by serum pancreatic enzyme activities (C and D), TAP may also be formed by sequential cleavage via CTSL and according to parameters of systemic inflammation (E: IL-10; F: lung ), the disease severity is reduced in the absence of and CTSB, a possibility we tested in the experiments CTSL. Values denote means Ϯ SEM for 5 or more measurements. below. These findings also indicate that CTSL deletion has completely opposite effects on the severity of pancre- atitis and on the intrapancreatic activity of trypsin. The concentrations of CTSB for proteolytic activation of trypsinogen were chosen to reflect the conditions in ex- Taurocholate-Induced Pancreatitis perimental pancreatitis in which only 0.1% to 1% of total To test whether the absence of CTSL has a similar trypsinogen is known to be activated to trypsin. Under severity-reducing effect on other models of pancreatitis these conditions, CTSB produced only a very weak tryp- BASIC–LIVER, that are largely independent of trypsinogen activation, we sin band (Figure 5A). At the same concentration, CTSL BILIARY TRACT PANCREAS, AND induced taurocholate-induced pancreatitis in mice. Here produced a much more effective cleavage of trypsinogen, again, the morphologic damage after 6 hours (Figure 4A resulting in a protein-product in the molecular weight and B), or the increase in serum pancreatic enzymes range of active trypsin (compared with enterokinase). (Figure 4C and D) or the markers of inflammation in Ten- to 100-fold higher concentrations of CTSB were serum (Figure 4E) and lung tissue (Figure 4F) were found needed to produce similarly prominent trypsinogen Ϫ Ϫ to be reduced in Ctsl / animals. cleavage (inset below CTSB in Figure 5A). In contrast to the incubation with CTSB, in which the Proteolytic Cleavage of Trypsinogen by CTSL cleavage resulted in the generation of trypsin activity To characterize further the effect of CTSL on (Figure 5B) and the formation of TAP (Figure 5C), the trypsinogen and trypsin, we used purified or recombinant trypsinogen processing by CTSL resulted neither in active enzymes. The proteolytic processing of trypsinogen was trypsin nor in the generation of TAP. To determine its followed by Western blotting, by measuring trypsin ac- structure, the trypsin fragment generated by CTSL cleav- tivity, and by quantification of TAP. In Figure 5A, immu- age from human and bovine trypsinogen was N-termi- noblots of incubations with enterokinase under optimal nally sequenced (Figure 6A). CTSL cleaved trypsinogen at catalytic conditions show a rapid and total conversion of the --IVG2GYN-site (G26-G27 in bovine-1 and human bovine trypsinogen to trypsin. Control incubations at pH cationic trypsinogen). Hereby, a truncated trypsin with- 5.5 in the presence of 10 mmol/L Caϩϩ indicated that out the first 3 amino acids IVG is formed, and this bovine trypsinogen did not autoactivate over 3 hours. protein completely lacks trypsin activity. On the other 732 WARTMANN ET AL GASTROENTEROLOGY Vol. 138, No. 2

Figure 5. Proteolytic processing of trypsinogen. Cleavage of trypsinogen by CTSB and CTSL was studied in vitro. (A) CTSL cleaves trypsinogen to a lighter protein that corresponds to active trypsin generated by enterokinase (EK) but is processed much more rapidly than by CTSB. Unlike

ACES AND PANCREAS, CTSB-generated trypsin, however, this protein has no trypsin activity (B) nor does CTSL-cleavage generate immunoreactive TAP (C). CTSB and IIR TRACT BILIARY BASIC–LIVER, CTSL were used in equimolar amounts. To reach comparable cleavage efficiency, a 100-fold CTSB concentration had to be used (inset below CTSB in A).

hand, the cleaved trypsinogen activation peptide that is Caϩϩ rise in acinar cells.21 Because CTSL activity is pH elongated by 3 amino acids (IVG) is not immunoreactive dependent and Caϩϩ stabilizes the trypsin(ogen) struc- against the TAP antibody. These experiments show that, ture,22 the processing of trypsin(ogen) by CTSL may whereas CTSB cleavage of trypsinogen results in active depend on the ionic environment within that subcellular trypsin and immunoreactive TAP, the cleavage by CTSL 3 compartment. We therefore investigated the effect of pH amino acids towards the C-terminus from the CTSB on CTSL activity measured as degradation of human processing site degrades trypsinogen to an inactive deg- recombinant trypsin (reduction in trypsin activity) and radation product and a nonimmunoreactive TAP-IVG cleavage of a CTSL peptide substrate Z-Phe-Arg-AMC peptide. This peptide generated by CTSL is rapidly pro- (Figure 6B and C). The rate of trypsin inactivation was cessed further by CTSB as shown in Figure 3 of the strongly pH dependent in the range between 4.2 and 4.8 Supplementary Material. in the presence of Caϩϩ and in the range between 5.2 and The Role of pH and Ca؉؉ on Proteolytic 5.8 in the absence of Caϩϩ (Figure 6B). Activity of CTSL Cleavage increased from pH 3.6 to 6.2 and was independent of the Previous investigations of the initial intracellular calcium concentration (Figure 6C). The rate of trypsin events during pancreatitis have shown that trypsinogen inactivation by CTSL is thus approximately 10 times activation begins in vesicular organelles19 and that they faster at pH 4.0 than at pH 5.5. It can therefore be represent an acidic compartment.20 Furthermore, it has concluded that the rate of trypsin cleavage by CTSL is been demonstrated that trypsinogen activation after su- mainly determined by pH-dependent changes in the sur- pramaximal cerulein stimulation depends on a sustained face charge of the trypsin molecule, which results in a February 2010 CATHEPSIN L IN PANCREATITIS 733 BASIC–LIVER, BILIARY TRACT PANCREAS, AND

Figure 6. Trypsin cleavage by CTSL and effect of pH and Caϩϩ. Human cationic trypsinogen or active trypsin were incubated for 3 hours with CTSL or enterokinase (EK) and submitted to SDS-PAGE followed by N-terminal sequencing. Note that the inactive CTSL-generated protein is 3 amino acids shorter (IVG) than active trypsin generated by enterokinase (A). Inactivation of human trypsin was measured as residual trypsin activity after incubation with CTSL at the pH indicated (B). Proteolytic cleavage of trypsin by CTSL exhibited a completely different dependence on pH and Caϩϩ compared with CTSL cleavage of the peptide substrate (C).

higher substrate affinity to CTSL. CTSL activity is, in trypsinogen and trypsin degradation by CTSL and, more itself, not calcium dependent. prominently so, at an acidic pH. We could further identify We then studied the inactivation and protein processing an additional CTSL-induced cleavage site in active trypsin at via CTSL for up to 3 hours (Supplementary Figure 1A). We position E82–G83. The detailed results of these experiments found that EDTA removal of calcium greatly increased are found in Supplementary Figures 1 and 2. 734 WARTMANN ET AL GASTROENTEROLOGY Vol. 138, No. 2

Processing of TAP by CTSB When we synthetized the activation peptide cleav- age product of CTSL (TAP-IVG or APFDDDDKIVG), which is not immunoreactive in the TAP-enzyme-linked immunosorbent assay, and exposed it to CTSB, it was rapidly converted to TAP (6-fold faster compared with the TAP generation from trypsinogen by CTSB). This indicates that sequential cleavage of trypsinogen by first CTSL and subsequently CTSB generates very large amounts of TAP (50% of equimolar enterokinase). TAP generated under these conditions no longer reflects active trypsin or disease severity.23 The details and data of this experiment are found in the Supplementary Materials.

Trypsin Activity and TAP in Isolated Pancreatic Acini While CTSB is a trypsinogen-activating enzyme,5 our data show that CTSL degrades trypsinogen and tryp- sin. To study whether this has a direct effect on acinar cells, we performed a series of in vitro experiments using freshly isolated acini. In these, we inhibited CTSL or CTSB before measuring trypsinogen activation in re- sponse to supramaximal cerulein. Figure 7A shows the effect of a specific CTSL inhibitor (1-Naphthalenylsulfo- nyl-Ile-Trp-aldehyde) and the CTSB inhibitor CA-074Me. CTSB inhibition reduced trypsinogen activation by more than 70%, whereas the inhibition of CTSL led to an increased level of active trypsin (ϩ50%). To confirm this, we also compared acinar cell preparations from CtslϪ/Ϫ, CtsbϪ/Ϫ, and wild-type mice (Ctsb/Ctslϩ/ϩ) in response to supramaximal cerulein. In the absence of CTSB, trypsino- gen activation was greatly decreased as indicated by sig-

ACES AND PANCREAS, nificantly reduced levels of active trypsin and TAP (Figure IIR TRACT BILIARY BASIC–LIVER, 7B and C). In the absence of CTSL, on the other hand, trypsin activity was increased to 170% compared with wild-type controls, but this was paralleled by a reduced Figure 7. Inhibition or deletion of CTSB and CTSL in pancreatic acini. rate of TAP formation. Lower TAP formation in the In panel A, pancreatic acini were preincubated for 30 minutes with 10 absence of CTSL may therefore indicate that a significant nmol/L CTSL inhibitor (1-Naphthalenylsulfonyl-Ile-Trp-aldehyde) or 1 ␮mol/L CTSB inhibitor CA-074Me then stimulated with cerulein. Trypsin portion of TAP in wild-type animals is generated via the activity was compared over 120 minutes. Inhibition of CTSL increased pathway identified above, in which CTSL rapidly gener- and the inhibition of CTSB decreased intra-acinar cell trypsin activity. In ates TAP-IVG and this is subsequently converted to TAP B and C, acini were prepared from wild-type, CTSL-deleted, and by CTSB. The alternative explanation of an extended CTSB-deleted animals. Trypsin activity and TAP were determined after half-life of trypsin (rather than higher activity) is less cerulein stimulation for 1 hour. The absence of CTSL increased, whereas absence of CTSB decreased intra-acinar cell trypsin activity. likely because, under CTSL inhibition (Figure 7A) in acini The generation of TAP was reduced in both knockout strains. Data Ϫ Ϫ or in acini of Ctsl / animals (not shown), the time represent means of 5 or more experiments Ϯ SEM. Significant differ- interval when trypsin activity returns to prestimulation ences are indicated with asterisks when P was Ͻ.05. levels is the same as in controls. It should be noted that the amount of TAP greatly exceeded the trypsin activities in molar terms in acini as well as in vivo (Figure 3F and pancreas by its own digestive proteases. Under physio- G). CTSB is thus physiologically and critically involved in logic conditions, the pancreas is protected by a variety of both pathways of TAP generation. mechanisms that include storage and processing of di- gestive enzymes in membrane-confined vesicles, trans- port of proteases to the lumen as inactive precursor Discussion zymogens, presence of protease inhibitors, and absence of The underlying mechanism of acute pancreatitis the physiologic activator enterokinase from the pancreas. has long been thought to involve autodigestion of the Several studies have shown that these protective mecha- February 2010 CATHEPSIN L IN PANCREATITIS 735

nisms are apparently overwhelmed in the early phase of located 3 amino acids towards the C-terminus from the pancreatitis and that protease activation, specifically the physiologic (ie, for enterokinase and CTSB) cleavage site premature and intrapancreatic activation of trypsinogen, Lys-Ile and removes the N-terminus of mature trypsin, is an inherent characteristic of human and several exper- thus impairing the catalytic center of trypsin.25 Incuba- imental models of pancreatitis.15,20,23 tion of mature trypsin with CTSL, on the other hand, One well-documented mechanism that permits the resulted in a negligible loss of the (Ile-Val-)-N-terminus. protease activation cascade to begin within acinar cells is This indicates a conformational change of trypsinogen the activation of trypsinogen by CTSB. Several studies upon physiologic cleavage of TAP by either CTSB or have shown with purified or recombinant enzymes,4,8 autoactivation, which prevents further processing of the with isolated preparations of pancreatic acini, or with N-terminus by CTSL. animal models of pancreatitis5,15 that CTSB is a potent We also identified an additional cleavage site for CTSL activator of trypsinogen and that its deletion or inhibi- at position E82–G83. This cleavage site is located in the tion greatly reduces intrapancreatic protease activation calcium-binding loop and affects trypsinogen as well as and the severity of pancreatitis. A prerequisite for the trypsin. Under conditions at which trypsin(ogen) binds activation of trypsinogen by CTSB is supposed to be a Caϩϩ, eg, pH 5.5, the cleavage by CTSL is strongly sup- redistribution of lysosomal enzymes into the zymogen- pressed. On the other hand, EDTA removal of Caϩϩ or containing secretory compartment and a colocalization decreasing pH to 4.0 strongly enhanced trypsin(ogen) of CTSB with trypsinogen. Both conditions are met when degradation. Consistent with these findings, the initial pancreatitis begins in either the human or animal pan- inactivation mechanisms for trypsinogen or trypsin by creas.5 Whether other lysosomal proteases can have sim- CTSL follow different pathways: (1) trypsinogen is pri- ilar functions in the pancreas or in pancreatitis was marily cleaved in the N-terminal region in a Caϩϩ-inde- previously unknown. pendent manner (G26-G27), and (2) the primary cleavage In the present study, we found that CTSL, the second of trypsin occurs predominantly at the Caϩϩ-binding site most common lysosomal cysteine proteinase besides (E82–G83). Which of these 2 prevails in vivo is impossible CTSB, is abundantly present in human and mouse pan- to predict, but, during experimental pancreatitis, the ab- creatic acinar cells. We further found that CTSL and sence of CTSL results in a manifold increase in intrapan- CTSB are both localized in the lysosomal as well as the creatic trypsin activity. pancreatic secretory compartment and that their colocal- The proteolytic processing of trypsinogen by entero- ization with zymogens further increases during pancre- kinase or CTSB produces active trypsin and the cleaved atitis and may even spread to the cytosol. Deletion of the propeptide in equal stoichiometric amounts. The ctsl gene, which does not affect the pancreas under phys- amount of TAP therefore reflects the extent of trypsino- iologic conditions, has 2 effects in experimental pancre- gen activation much more accurately than activity mea- atitis: (1) it greatly increases intrapancreatic trypsin ac- surements, eg, when trypsin is rapidly inactivated by tivity because CTSL is a trypsin(ogen) inactivator and thus autodegradation,23 endogenous inhibitors, or proteolytic an antagonist of CTSB; and (2) it greatly reduces the degradation. Because of its relative stability and the avail- severity of pancreatitis, possibly by shifting the cellular ability of specific antibodies, TAP is increasingly recog- effects of pancreatitis towards apoptosis. nized as a stand-alone parameter for pancreatitis severity

23 BASIC–LIVER, CTSB and CTSL are widely expressed members of the and has found its way into clinical practice. Our results BILIARY TRACT PANCREAS, AND papain family of cysteine proteinases.9 Recent experimen- now indicate a second pathway in which the generation tal results suggest that they not only catalyze bulk pro- of trypsin activity and TAP does not develop in parallel. teolysis but also take part in proteolytic processing of We found that the primary cleavage of trypsinogen by protein substrates.9 For CTSB, we and others have shown CTSL creates a TAP-IVG peptide that escapes detection that it acts as a trypsinogen-activating enzyme in vivo by the TAP antibody. However, its subsequent conver- and in vitro4,5 and that this process involves the cleavage sion by CTSB produces immunoreactive TAP to a of a Lys-Ile bond releasing active trypsin and the propep- much greater extent than via direct CTSB activation of tide TAP. Because of its optimum at a pH Ͻ6, CTSB trypsinogen. Obviously, this pathway combines the exerts its catalytic activity in an acidic intracellular com- highly effective endoproteolytic activity of CTSL with the partment, where zymogen activation has also been shown exoproteolytic activity of CTSB. In the pathologic situa- to take place.20 We found that CTSL, the second most tion of acute pancreatitis, these large amounts of TAP abundant cysteine proteinase, shares the same intracellu- may no longer reflect trypsin activity. lar compartment and confirmed24 that it possesses a Our experiments, using isolated pancreatic acini with much higher endoproteolytic activity than CTSB. either specific enzyme inhibitors or from CTSL and We further found that CTSL very effectively cleaves CTSB knockout animals, confirmed this fundamental trypsinogen at position G26-G27 resulting in its inacti- difference between the 2 lysosomal hydolases. CTSB was vation. Human cationic trypsinogen was found to be confirmed as an activator of trypsinogen and of the cleaved at the exact same position. This cleavage site is intracellular protease cascade, whereas CTSL was identi- 736 WARTMANN ET AL GASTROENTEROLOGY Vol. 138, No. 2

fied as its antagonist and a potent inactivator of trypsino- Gastroenterology at www.gastrojournal.org, and at doi: gen and trypsin. Taken together, these data suggest that 10.1053/j.gastro.2009.10.048. alterations in the structure or function of CTSL could represent an important mechanism in the pathogenesis References of human pancreatitis. 1. Whitcomb DC, Gorry MC, Preston RA, et al. Hereditary pancreati- The effect of CTSL on disease severity seems to be tis is caused by a mutation in the cationic trypsinogen gene. Nat completely uncoupled from its effect on trypsinogen ac- Genet 1996;14:141–145. tivation. It not only affects pancreatic injury directly but 2. Simon P, Weiss FU, Sahin-Tóth M, et al. Hereditary pancreatitis also translates into less systemic inflammation, even in a caused by a novel PRSS1 mutation (Arg-122ϾCys) that alters autoactivation and autodegradation of cationic trypsinogen. J Biol model such as taurocholate-induced pancreatitis in Chem 2002;277:5404–5410. which trypsinogen activation might not play such a cru- 3. Simon P, Weiss FU, Zimmer KP, et al. Spontaneous and sporadic cial role in determining disease severity. It is also paral- trypsinogen mutations in idiopathic pancreatitis. JAMA 2002; leled by increased caspase-3 activation, PARP cleavage, 288:2122. and apoptosis in the pancreas of CtslϪ/Ϫ animals. That 4. Greenbaum LM, Hirshkowitz A, Shoichet I. The activation of trypsinogen by cathepsin B. J Biol Chem 1959;234:2885–2890. such a shift to apoptosis-dominant cell injury improves 5. Halangk W, Lerch MM, Brandt-Nedelev B, et al. Role of cathepsin the outcome of pancreatitis has previously been re- B in intracellular trypsinogen activation and the onset of acute ported.26 Cathepsins, on the other hand, have long been pancreatitis. J Clin Invest 2000;106:773–781. considered to be guardians of the cellular homeostasis 6. Mahurkar S, Idris MM, Reddy DN, et al. Association of cathepsin and, when released into the cytoplasm, to influence ap- B gene polymorphisms with tropical calcific pancreatitis. Gut 2006;55:1270–1275. optotic pathways through a number of different mecha- 7. Weiss FU, Behn CO, Simon P, et al. Cathepsin B gene polymor- 27 nisms. A release of cathepsins into the cytoplasm has phism Val26 is not associated with idiopathic chronic pancreati- recently been shown to also occur in pancreatitis.28 tis in European patients. Gut 2007;56:1322–1323. Crossbreeding of CtslϪ/Ϫ with CtsbϪ/Ϫ mice results in a 8. Kukor Z, Mayerle J, Kruger B, et al. Presence of cathepsin B in the lethal phenotype around postnatal day 12 and is caused human pancreatic secretory pathway and its role in trypsinogen activation during hereditary pancreatitis. J Biol Chem 2002;277: by significant neuronal cell apoptosis, whereas neither 21389–21396. the CTSL nor the CTSB knockout alone has such an 9. Kirschke H, Barrett AJ. Cathepsin L—a lysosomal cysteine pro- effect. This suggests that some in vivo function of CTSB teinase. Prog Clin Biol Res 1985;180:61–69. can be compensated for by CTSL and vice versa.29 Other 10. Yasothornsrikul S, Greenbaum D, Medzihradszky KF, et al. Ca- experimental observations would also be in line with a thepsin L in secretory vesicles functions as a prohormone-pro- 30,31 cessing enzyme for production of the enkephalin peptide neuro- predominantly antiapoptotic role of CTSL. In the transmitter. Proc Natl Acad SciUSA2003;100:9590–9595. setting of pancreatitis, however, this antiapoptotic func- 11. Nakagawa T, Roth W, Wong P, et al. Cathepsin L: critical role in tion of CTSL appears to contribute to disease severity, Ii degradation and CD4 T cell selection in the thymus. Science and the absence or inhibition of CTSL would have a 1998;280:450–453.

ACES AND PANCREAS, 12. Reinheckel T, Hagemann S, Dollwet-Mack S, et al. The lysosomal IIR TRACT BILIARY

BASIC–LIVER, beneficial therapeutic effect. cathepsin L regulates keratinocyte prolifera- In conclusion, the present study establishes CTSL as a tion by control of growth factor recycling. J Cell Sci 2005;118: potent trypsin(ogen)-inactivating factor in vivo and in 3387–3395. vitro and thus as an antagonist of CTSB in the digestive 13. Cirman T, Oresic K, Mazovec GD, et al. Selective disruption of protease cascade that triggers pancreatitis. The antiapop- lysosomes in HeLa cells triggers apoptosis mediated by cleavage totic function of CTSL, on the other hand, affects disease of Bid by multiple papain-like lysosomal cathepsins. J Biol Chem severity in a manner that is unrelated to the initiating 2004;279:3578–3587. 14. Roth W, Deussing J, Botchkarev VA, et al. Cathepsin L deficiency protease cascade. As far as the understanding of acute as molecular defect of furless: hyperproliferation of keratinocytes pancreatitis is concerned, these data indicate further that and pertubation of hair follicle cycling. FASEB J 2000;14:2075– (1) greater intrapancreatic trypsin activity does not nec- 2086. essarily correlate with greater acinar cell injury and that 15. Halangk W, Kruger B, Ruthenburger M, et al. Trypsin activity is not (2) higher TAP levels do not always reflect higher levels of involved in premature, intrapancreatic trypsinogen activation. Am J Physiol Gastrointest Liver Physiol 2002;282:G367–G374. active trypsin or disease severity. The sequence in which 16. Kukor Z, Toth M, Pal G, et al. Human cationic trypsinogen. lysosomal proteases find their substrates and the subcel- Arg(117) is the reactive site of an inhibitory surface loop that lular compartment in which they either constitutively or controls spontaneous zymogen activation. J Biol Chem 2002; pathophysiologically colocalize with digestive enzymes as 277:6111–6117. well as the biophysical properties of that compartment 17. Sahin-Tóth M. Human cationic trypsinogen. Role of Asn-21 in zymogen activation and implications in hereditary pancreatitis. ultimately determine whether their action is disease pro- J Biol Chem 2000;275:22750–22755. voking or protective. 18. Mayerle J, Schnekenburger J, Kruger B, et al. Extracellular cleav- age of E-cadherin by leukocyte during acute experimen- Supplementary Materials tal pancreatitis in rats. Gastroenterology 2005;129:1251–1267. 19. Lerch MM, Saluja AK, Runzi M, et al. Luminal endocytosis and Note: To access the supplementary material intracellular targeting by acinar cells during early biliary pancre- accompanying this article visit the online version of atitis in the opossum. J Clin Invest 1995;95:2222–2231. February 2010 CATHEPSIN L IN PANCREATITIS 737

20. Lerch MM, Saluja AK, Dawra R, et al. The effect of chloroquine 30. Zheng X, Chu F, Mirkin BL, et al. Role of the proteolytic hierarchy administration on 2 experimental models of acute pancreatitis. between cathepsin L, and caspase-3 in regulation of Gastroenterology 1993;104:1768–1779. cellular susceptibility to apoptosis and autophagy. Biochim Bio- 21. Kruger B, Albrecht E, Lerch MM. The role of intracellular calcium phys Acta 2008;1783:2294–2300. signaling in premature protease activation and the onset of 31. Zheng X, Chu F, Chou PM, et al. Cathepsin L inhibition sup- pancreatitis. Am J Pathol 2000;157:43–50. presses drug resistance in vitro and in vivo: a putative mecha- 22. Sahin-Toth M, Toth M. High-affinity Caϩϩ binding inhibits autoac- nism. Am J Physiol Cell Physiol 2009;296:C65–C74. tivation of rat trypsinogen. Biochem Biophys Res Commun 2000; 275:668–671. 23. Neoptolemos JP, Kemppainen EA, Mayer JM, et al. Early predic- Received April 17, 2009. Accepted October 9, 2009. tion of severity in acute pancreatitis by urinary trypsinogen acti- vation peptide: a multicentre study. Lancet 2000;355:1955– Reprint requests 1960. Address requests for reprints to: Markus M. Lerch, MD, 24. Rocken C, Menard R, Buhling F, et al. of serum Department of Medicine A, Ernst-Moritz-Arndt-University, Greifswald amyloid A and AA amyloid proteins by cysteine proteases: cathep- Friedrich-Loeffler-Strasse 23A, D-17475 Greifswald, Germany. sin B generates AA amyloid proteins and cathepsin L may prevent e-mail: [email protected]; fax: (49) 3834-867234. their formation. Ann Rheum Dis 2005;64:808–815. Acknowledgments 25. Bode W, Fehlhammer H, Huber R. Crystal structure of bovine The authors thank Dr B. Schmidt, Zentrum für Biochemie und trypsinogen at 1-8 A resolution. I. Data collection, application of Molekulare Zellbiologie, Göttingen, for sequencing assistance and V. Patterson search techniques and preliminary structural interpre- Krause, C. Jechorek, K. Mülling, and N. Loth for technical assistance tation. J Mol Biol 1976;106:325–335. and D. Schwenn for writing assistance. 26. Mareninova OA, Sung KF, Hong P, et al. Cell death in pancreatitis: T.W. and J.M. contributed equally as first authors. protect from necrotizing pancreatitis. J Biol Chem W.H. and M.M.L. contributed equally as senior authors. 2006;281:3370–3381. 27. Chwieralski CE, Welte T, Buhling F. Cathepsin-regulated apopto- Conflicts of interest sis. Apoptosis 2006;11:143–149. The authors disclose no conflicts. 28. Dawra R, Talukdar R, Dudeja V, et al Release of cathepsin B into cytosol induces apoptosis in pancreatic acinar cells through Funding mitochondrial pathways during pancreatitis. Gastroenterology Supported by grants from the Deutsche Forschungsgemeinschaft 2009;136(Suppl 1):A372. DFG HA2080/7-1, LE 625/8-1, LE 625/9-1, MA4115/1-2, DFG GRK 29. Felbor U, Kessler B, Mothes W, et al. Neuronal loss and brain 840 E3 and E4; Mildred Scheel Stiftung 10-2031-Le I and 10-6977- atrophy in mice lacking cathepsins B and L. Proc Natl Acad Sci Re, BMBF-NBL3 01 ZZ 0403; and Alfried-Krupp Foundation U S A 2002;99:7883–7888. (Graduiertenkolleg Tumorbiologie) and NIH DK058088 to MST. BASIC–LIVER, BILIARY TRACT PANCREAS, AND