Cathepsin B Mediates Tumor Necrosis Factor-induced Arachidonic Acid Release in Tumor Cells*

Arachidonic acid (AA) generated by cytosolic phospholipase A2 (cPLA2) has been suggested to function as a second messenger in tumor necrosis factor (TNF)-induced death signaling. Here, we show that cathepsin B-like proteases are required for the TNF-induced AA release in transformed cells. Pharmaceutical inhibitors of cathepsin B blocked TNF-induced AA release in human breast (MCF-7S1) and cervix (ME-180as) carcinoma as well as murine fibrosarcoma (WEHI-S) cells. Furthermore, TNF-induced AA release was significantly reduced in cathepsin B-deficient immortalized murine embryonic fibroblasts. Employing cPLA2-deficient MCF-7S1 cells expressing ectopic cPLA2 or cPLA2-deficient immortalized murine embryonic fibroblasts, we showed that cPLA2 is dispensable for TNF-induced AA release and death in these cells. Furthermore, TNF-induced cathepsin B-dependent AA release could be dissociated from the cathepsin B-independent cell death in MCF-7S1 cells, whereas both events required cathepsin B activity in other cell lines tested. These data suggest that cathepsin B inhibitors may prove useful not only in the direct control of cell death but also in limiting the damage-associated inflammation.

Phospholipase A 2 (PLA 2 ) 1 family of esterases plays a significant role in inflammation because PLA 2 s can cleave arachidonic acid (AA) from the sn-2 position of membrane phospholipids and provide it to cyclo-and lipooxygenases, which catalyze its conversion into prostanoids and leukotrienes, respectively (1,2). In addition to generating AA for synthesis of inflammatory mediators, PLA 2 s have been suggested to mediate cell death induced by death and antigen receptors, hydrogen peroxide, or lack of focal adhesions and to modulate drugand ischemia/reperfusion-induced tissue injury (3)(4)(5)(6)(7)(8)(9)(10). Contrary to the rapid activation of PLA 2 s during the inflammatory response, death-associated AA release is a late event occurring downstream of mitochondrial changes and effector caspase activation (7,(11)(12)(13).
The PLA 2 activity is found in several enzymes that can be classified based on their Ca 2ϩ dependence into secretory PLA 2 s, which require Ca 2ϩ at mM levels, cytosolic PLA 2 s (cPLA 2 s), which require Ca 2ϩ at M levels and display specificity for phospholipids containing AA in the sn-2 position, Ca 2ϩ -independent PLA 2 s (iPLA 2 s), and platelet-activating factor acetyl hydrolases. The definition of the role of individual PLA 2 enzymes in cell death has been based mainly on the ability of pharmacological inhibitors to confer protection. The obtained data suggesting that iPLA 2 and cPLA 2 mediate cell death induced by anti-Fas and tumor necrosis factor (TNF) or oxidative stress, respectively (5)(6)(7)14), need, however, to be re-evaluated, because it has become clear that the commonly used PLA 2 inhibitors are not as specific as originally assumed (1). Instead, they cross-react with several PLA 2 s and possibly even with other enzymes like cyclo-oxygenase. Genetic evidence support, however, a role for cPLA 2 in cell death and tissue injury. Mice deficient for cPLA 2 ␣ show reduced postischemic tissue injury and resistance to brain injury induced by 1-methyl-4-phenyl-1,2,3,6,-tetrahyropyride (8); a TNF-resistant cPLA 2 -deficient clone of L929 fibrosarcoma cells can be rendered TNF-sensitive upon ectopic expression of cPLA 2 (3); and depletion of melanoma cells of cPLA 2 by antisense oligonucleotides protects them against TNF-induced cytotoxicity (4).
TNF is a multifunctional cytokine that can elicit several biological responses like apoptosis, inflammation, and stress response (15,16). TNF-induced apoptosis is triggered by the activation of caspase-8 at the receptor complex. Caspase-8 can either directly or via a mitochondrial amplification pathway activate downstream effector caspases, including caspase-3, which execute the apoptotic cell death. Interestingly, TNFinduced AA release in tumor cells depends on caspase activation (7), and both cPLA 2 and iPLA 2 are caspase substrates (9,17). It has recently become evident that in addition to the classic caspase-mediated apoptosis, TNF can activate mechanistically different caspase-independent death pathways resulting in apoptosis-or necrosis-like morphology (18 -23). Effector molecules implicated in these alternative pathways include AA, reactive oxygen intermediates, receptor-interacting protein RIP, Rac GTP-binding protein, and cathepsin B-like proteases. Cathepsin B-like cysteine proteases participate also in the final execution of TNF-induced caspase-dependent cell death in various human tumor cell lines (22) and murine hepatocytes (24). In all cathepsin-mediated cell death models, cathepsins translocate from the lysosomes into the cytosol before cathepsin-mediated morphological changes (25).
This study was designed to investigate the role of cathepsin B in TNF-induced AA release in cells dying either in caspasedependent or -independent manner and to study the role of individual PLA 2 enzymes and their cleavage in TNF-induced cell death. For this purpose we analyzed the effect of various well characterized protease inhibitors on TNF-induced AA release, cytotoxicity, and cleavage of PLA 2 s in three TNF-sensitive tumor cell lines. The results obtained were confirmed genetically by employing immortalized/transformed murine embryonic fibroblasts (iMEFs) from wild type, cathepsin B-, cPLA 2 -, or caspase-3-deficient mice or by expressing ectopic cPLA 2 or its noncleavable mutants in MCF-7S1 cells lacking the endogenous enzyme. The data obtained suggest a novel role for cathepsin B-like enzymes in the regulation of AA metabolism and inflammation.
Plasmids-The cDNA encoding human cPLA 2 (kindly provided by D. Sharp, Eli Lilly) was subcloned into pBSII (Stratagene) and pEBS7 (11) vectors. The substitution of the C-terminal aspartic acid (D) to a glutamic acid (E) in the caspase cleavage site (DELD 522 ) of cPLA 2 was performed by altering the codon GCT to GAG using the QuikChange TM site-directed mutagenesis kit (Stratagene) in pBSII-cPLA 2 . The deletion of the cDNA encoding either the C-terminal 227 amino acids (cPLA 2 -N) or N-terminal 522 amino acids (cPLA2-C) of human cPLA 2 was performed by PCR with pBSII-cPLA 2 as a template. Subsequently, the obtained cDNAs with or without N-terminal FLAG-tag (GAC TAC AAG GAC GAC GAT GAC AAG) were subcloned between HindIII and NotI sites of multicloning sites of pN1 (pEGFP-N1 from Clontech without EGFP) or pEBS7 vectors by PCR. The expected sequences were verified by DNA sequencing. To create pCEP-vimentin, cDNA encoding vimentin (kindly provided by Nobuhiro Morishima, Saitama, Japan) was subcloned into the multicloning site of pCEP4-puro vector (the pCEP4 vector from Invitrogen with a replacement of the hygromycin resistance gene with the puromycin resistance gene from pBABE-puro). pcDNA3-neo was from Invitrogen, and pcDNA-casp3 was kindly provided by Vishva Dixit (Ann Arbor, MI).
Transfections-The transfections of MCF-7S1 cells with the indicated plasmids were performed by electroporation as described previously (11). Two days later successfully transfected cells were selected by adding 150 g/ml hygromycin, 200 g/ml G418, and/or 5 g/ml puromycin. Experiments were performed with cell pools that had been in culture for 1-3 weeks. Transient transfections of WEHI-S cells were performed using LipofectAMINE plus reagents (Invitrogen).
Immunoblotting and Antibodies-Immunodetection of proteins separated by SDS-PAGE and transferred to nitrocellulose was performed with enhanced chemiluminescence Western blotting reagents (Amersham Biosciences). To obtain cPLA 2 -specific antisera, rabbits were immunized with ovalbumin-coupled peptides (CPDPYVELFISTTPDSRK sequence, corresponding to amino acids 42-58 or CSVSLSNVEAR-RFFNKEFLSKPKA, corresponding to amino acids 726 -749 of human cPLA 2 sequence) at Neosystem. The obtained polyclonal antisera recognizing the N-terminal or the C-terminal part of cPLA 2 are referred to as anti-cPLA 2 -N and anti-cPLA 2 -C, respectively. Other primary antibodies used included polyclonal rabbit antiserum, recognizing a Cterminal sequence of iPLA 2 (Cayman Chemical), goat polyclonal anticaspase-3 antiserum (Santa Cruz Biotechnology), and murine monoclonal antibodies against cathepsin B (Oncogene Research Products), FLAG (anti-FLAG®M2, Sigma), and Hsc70 (clone N69, kindly provided by Boris A. Margulis, St. Petersburg, Russia). Peroxidaseconjugated secondary antibodies were from DAKO A/S.
Analysis of TNF-induced Cleavage of Ectopic cPLA 2 Protein in WEHI-S Cells-Two days after the transfections of 6 ϫ 10 6 WEHI-S cells with the indicated plasmids, cells were stimulated with TNF and/or zVAD-fmk for 8 h, and the ectopic FLAG-cPLA 2 was immunoprecipitated with anti-FLAG antibody. Briefly, cells were harvested by scraping, washed once in PBS, resuspended in 1 ml of Tris lysis buffer (20 mM Tris-HCl, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 0.5% Nonidet P-40, pH 7.7) containing protease inhibitors, and snap-frozen in liquid nitrogen. After centrifugation of the thawed lysate (14.000 ϫ g for 10 min), the supernatant was precleared by incubation with 50 l of a 50% solution of protein G-Sepharose TM beads (Amersham Biosciences) in phosphate-buffered saline for 2 h at 4°C with rotation. Beads were removed by centrifugation, 0.95 l of anti-FLAG-antibody was added to the supernatant, and the mixture was rotated for 18 h at 4°C. 60 l of Sephadex protein G beads (50% in phosphate-buffered saline) was added, and the rotation was continued for an additional 2 h. The beads were washed 5 times with Tris lysis buffer, dissolved in 70 l of 1ϫ Laemmli sample buffer, boiled for 5 min, and analyzed by Western blot analysis.
Assay for Phospholipase A 2 -like Activity in Cells and in Vitro-The analysis of AA release was done essentially as described previously employing cells prelabeled with [5,6,8,9,11,12,14, H]AA (0.25 Ci/ ml; PerkinElmer Life Sciences) (28). The radioactivity released from cells treated with TNF and protease inhibitors was compared with that from cells incubated with protease inhibitors alone. The spontaneous AA release was always less than 15% of the total. 3 H released from cells killed by freezing and thawing did not differ significantly from the spontaneous release, indicating that all the radioactive AA was incorporated into the membranes and did not get released by unspecific lysis of the cells.
To measure the PLA 2 activity in the lysates of transfected cells, cells (5 ϫ 10 6 cells/ml) were resuspended in a lysis buffer (80 mM KCl, 10 mM Hepes pH 7.4, 1 mM EDTA, 1 mM EGTA, 40 g/ml leupeptin, 25 g/ml pepstatin A, 1 mM sodium fluoride, 0.2 mM ammonium(meta)vanadate, and 4 mM DTT), and the cytosolic fraction was isolated by sonication and centrifugation. The reaction was performed at 37°C for 30 min in a total volume of 525 l containing 100 l of cytosolic cell lysate and 50 pmol of 1-stearoyl-2-[1-14 C]arachidonoylphosphatidylcholine (55 mCi/ mmol, Amersham Biosciences) in a reaction buffer (80 mM KCl, 10 mM Hepes, pH 7.4, 1 mM EDTA, 4.7 mM Ca 2ϩ , 5 mM DTT, and 200 g bovine serum albumin) and terminated by the addition of 750 l of stop solution (chlorophorm/methanol/10 M hydrochloric acid (volume 2:1: 0.01), 0.05% butylated hydroxytoluene, and 0.1 mg/ml of mouse liver lipid (kindly provided by G. Hiller). After vortexing, the lipid phase was added to a silica acid (Merck) column equilibrated with 1 ml of chloroform, and phospholipids were eluted with 2.1 ml of methanol. Eluates of both free fatty acids and phospholipids were air-dried and mixed with 3 ml of Ecosint A scintillation fluid and counted in a scintillation counter.

TNF-induced AA Release in MCF-7S1 and ME-180as Cells Is Dependent on Caspases and Cathepsin B-like Proteases-In
TNF-sensitive human tumor cells, TNF-induced PLA 2 activa-tion as well as the release of cysteine cathepsins from lysosomes into the cytosol are dependent on caspase activity (7,22). To test whether cysteine cathepsins and PLA 2 are in the same caspase-dependent signaling pathway, we tested the effect of various protease inhibitors on TNF-induced PLA 2 activity. Treatment of MCF-7S1 cells with 20 ng/ml TNF for 18 h triggered a 2.5-fold increase in AA release as compared with untreated cells. As expected, inhibition of caspases by zVAD-fmk, DEVD-CHO, or IETD-CHO almost completely inhibited TNFinduced AA release in MCF-7S1 breast cancer cells (Fig. 1A). In addition to caspase inhibitors, also a broad spectrum cysteine cathepsin inhibitor (zFA-fmk) and a more specific inhibitor of cathepsin B (CA-074-Me) partially inhibited TNF-induced AA release. Similarly, zVAD-fmk, DEVD-CHO, and CA-074-Me effectively inhibited TNF-induced 2.1-fold increase in AA release in ME-180as human cervix carcinoma cells (Fig. 1B). Except for CA-074-Me-treated MCF-7S1 cells dying as effectively as untreated cells despite the significantly reduced AA release, the ability of all other protease inhibitors to inhibit PLA 2 -like activity correlated well with their ability to inhibit TNF-induced cell death (Fig. 1, A and B). These data indicate that caspases and cathepsin B are upstream of both PLA 2 activation and death in ME-180as cells, whereas in MCF-7S1 cells cathepsin B activity is required for TNF-induced PLA 2 activation but not death. Other cysteine cathepsins than cathepsin B appear, however, to contribute to TNF-induced execution of MCF-7S1 cells because a broad-spectrum cysteine cathepsin inhibitor, zFA-fmk, was partially protective (Fig.  1A). We could unfortunately not investigate the role of PLA 2like activity in TNF-induced death of human tumor cells, because PLA 2 inhibitors were highly toxic to these cells. The ability of CA-074-Me to inhibit TNF-induced PLA 2 activation without inhibiting death indicated, however, that AA release is not required for TNF-induced death in MCF-7S1 cells.
Cathepsin B, but Not Caspases, Is Required for TNF-induced PLA 2 -like Activity in WEHI-S Cells-Careful titration of protease inhibitor concentrations to obtain specific inhibition of caspases has revealed that contrary to MCF-7S1 and ME-180as cells, TNF-induced death of WEHI-S murine fibrosarcoma cells as well as that of several other fibroblast-derived murine cells is enhanced by caspase-8 inhibition (19,22). These data prompted us to reevaluate our earlier data suggesting that the TNF-induced AA release in WEHI-S cells is caspase-dependent (7). To avoid the inhibition of other cysteine proteases than the caspases, we used zVAD-fmk at concentrations from 1 to 5 M that were sufficient to inhibit the TNF-induced caspase-8 activation (demonstrated by the inhibition of the autocatalytic processing of caspase-8 into an ϳ43-kDa intermediate fragment) without having a significant effect on cysteine cathepsins or calpains (22). As demonstrated earlier, the inhibition of caspases by 1-5 M zVAD-fmk sensitized WEHI-S cells to cytotoxicity induced by 100 pg/ml TNF (Fig. 1C). This sensitization was accompanied by a significant increase in AA release. Higher zVAD-fmk concentrations (Ն50 M) that in addition to caspases effectively inhibit cathepsin B and other cysteine cathepsins completely inhibited both TNF-induced AA release and cell death. The inhibition of cathepsin B activity by CA-074-Me also effectively attenuated TNF-induced cell death and AA release in WEHI-S cells. Furthermore, the enhanced death and PLA 2 -like activity induced by the combination of TNF and zVAD-fmk were effectively inhibited by CA-074-Me. In addition to cysteine cathepsin inhibitors, an antioxidant butylated hydroxyanisole blocked both death and AA release induced by TNF in WEHI-S cells (Fig. 1C). As demonstrated earlier in slightly different experimental conditions (7), the inhibition of PLA 2 -like activity by 20 M AACOCF 3 also conferred nearly complete protection to WEHI-S cells against cytotoxicity induced by a 12-h treatment with 100 pg/ml TNF (not shown). These data suggest that a cathepsin B-dependent but caspaseindependent PLA 2 activity is required for TNF-induced cytotoxicity to occur in WEHI-S cells.
TNF-induced AA Release Is Defective in Fibroblasts Lacking Cathepsin B-To test genetically whether cathepsin B really is required for the TNF-induced activation of PLA 2 -like enzymes, we compared the ability of TNF to induce AA release in iMEFs originating from wild type and cathepsin B-deficient mice (26). TNF induced a dose-dependent release of AA in wild type iMEFs at concentrations ranging from 10 pg/ml to 1 ng/ml ( Fig. 2A). Contrary to wild type cells, no AA release was detected after stimulation of cathepsin B-deficient iMEFs with 10 or 100 pg/ml TNF. At a higher concentration (1 ng/ml) TNF induced some AA release also in cathepsin B-deficient iMEFs, but this release was significantly less than in wild type iMEFs. The differences in TNF-induced AA release between these cell lines reflected differences in their sensitivity to TNF-induced cytotoxicity (Fig. 2B).

PLA 2 Isoforms Are Differentially Expressed in Tissues and
Cell Lines-Both iPLA 2 and cPLA 2 can trigger the release of AA, and both enzymes have been implicated in death receptorinduced cell death (1). Therefore, we studied the expression of cPLA 2 and iPLA 2 proteins in cell lines used in this study. WEHI-S cells and iMEFs expressed high levels of cPLA 2 but no detectable iPLA 2 (Fig. 3A). A similar expression pattern was observed in murine brain, whereas spleen contained both cPLA 2 and iPLA 2 , and thymus, testis, and lung tissues expressed mainly iPLA 2 . MCF-7S1 cells expressed iPLA 2 , but no cPLA 2 , and ME-180as cells expressed both isoforms (Fig. 3B).
Both cPLA 2 and iPLA 2 have been reported to be cleaved by caspases during apoptosis (9). Ectopic and endogenous cPLA 2 was effectively cleaved in TNF-treated MCF-cPLA 2 and ME-180as cells, respectively, into an ϳ64 kDa N-terminal fragment and a 32-kDa C-terminal fragment (Fig. 3B, not shown). Even though the amount of full-length iPLA 2 was slightly reduced in TNF-treated ME-180as and MCF-cPLA 2 cells, we failed to detect any cleavage product in these samples.
TNF-induced AA Release Peaks before the Cleavage of cPLA 2 in WEHI-S Cells-Because caspase-3-mediated cleavage of cPLA 2 at Asp 522 inactivates the enzyme, we speculated that zVAD-fmk-mediated enhancement of TNF-induced cytotoxicity and AA release could be due to the inhibition of cPLA 2 cleav-age. However, the increase in TNF-triggered AA release and death induced by 1 M zVAD-fmk was surprisingly accompanied by an increase in cPLA 2 cleavage (Fig. 4, A and B). Furthermore, CA-074-Me, which inhibited TNF-induced AA release and death, also inhibited the cleavage of cPLA 2 in WEHI-S cells. Thus, the cleavage of cPLA 2 may rather represent a post-activation event and function as a negative feedback mechanism. In line with this, the comparison of the kinetics of TNF-induced AA release and cPLA 2 cleavage revealed that the cleavage occurred later than the AA release and that cells still contained plenty of full-length cPLA 2 at the time when AA release reached its peak (Fig. 4, A and B).
To test whether the cleavage of cPLA 2 in WEHI-S cells took place at the DELD 522 caspase-3 cleavage site, we transiently transfected WEHI-S cells with vectors either expressing FLAGtagged wild type cPLA 2 or cPLA 2 mutated in the caspase-3 cleavage site (DELD 522 ) by the substitution of Asp 522 with Glu 522 (cPLA 2 -DELE). Despite several attempts, we could not express ectopic cPLA 2 in WEHI-S cells at levels detectable by immunoblotting. To be able to detect and analyze ectopic proteins, we immunoprecipitated them with the anti-FLAG-antibody from a large number of transiently transfected cells. Akin to the endogenous cPLA 2 , ectopic FLAG-cPLA 2 was cleaved into a 65-kDa cleavage fragment (the slightly larger size due to the FLAG sequence) whether zVAD-fmk was present or not (Fig. 4C). A similar 65-kDa fragment was seen in the lysates of TNF-treated MCF-FLAG-cPLA 2 cells dying in a caspase-dependent manner. The mutant FLAG-cPLA 2 -DELE remained intact upon WEHI-S cell stimulation with TNF or TNF plus zVAD-fmk. These data suggest that zVAD-fmk enhances the TNF-induced activity of a protease capable of cleaving cPLA 2 at or around Asp 522 .
Next, we investigated the dependence of the TNF-induced cleavage of cPLA 2 on cathepsin B and caspase-3 in iMEFs. TNF treatment of wild type iMEFs resulted in the appearance of the 64-kDa N-terminal cleavage fragment of cPLA 2 (Fig. 5). The amount of full-length cPLA 2 was only slightly reduced, and the 64-kDa N-terminal cleavage fragment was not detected in TNF-treated cathepsin B-deficient iMEFs, whereas caspase-3 deficiency had no effect on the TNF-induced cleavage pattern of cPLA 2 .

Cysteine Cathepsins Cleave cPLA 2 into a 50-kDa Fragment in
Vitro-To test whether cysteine cathepsins that normally function at acidic lysosomal pH could be responsible for the TNFinduced cleavage of cPLA 2 , we tested their ability to cleave cPLA 2 at neutral pH in vitro. At pH 7.5, both cathepsin B and L cleaved cPLA 2 into an ϳ50-kDa fragment clearly distinct from the 64-kDa cleavage fragment produced by caspase-3 in vitro or detected in lysates of TNF-treated cells (Fig. 6).
TNF Fails to Activate cPLA 2 in MCF-7S1 Cells-Because MCF-7S1 cells express no cPLA 2 , we asked whether ectopic expression of cPLA 2 could enhance the TNF-induced PLA 2 activation and sensitize MCF-7S1 cells to TNF-induced apoptosis. Contrary to WEHI-S cells, we could easily create stable MCF-7S1 cells expressing high levels of ectopic cPLA 2 (Fig.  7A). MCF-7S1 cells expressing wild type cPLA 2 displayed similar sensitivity to TNF as vector-transfected control cells (Fig.  7, A and B). Ectopic expression of cPLA 2 failed to enhance the ability of MCF-7S1 cells to release AA in response to TNF (Fig.  7C). To test whether the ability of cPLA 2 to sensitize MCF-7S1 cells to TNF-induced cytotoxicity was regulated by the TNFinduced cleavage of cPLA 2 , we next expressed the noncleavable cPLA 2 -DELE mutant or N-terminal (amino acids 1-522) and C-terminal (amino acids 523-749) cPLA 2 fragments mimicking the fragments created by the cleavage at Asp 522 in MCF-7S1 cells. All three proteins could be expressed in MCF-7S1 cells, but their expression had no effect on cell viability or TNFinduced AA release (Fig. 7, A and B). To ensure that the ectopically expressed cPLA 2 was functional, we next analyzed the cytosols from vector-and cPLA 2 -transfected MCF-7S1 cells for PLA 2 activity in an in vitro PLA 2 enzyme assay. The cytosols from MCF-cPLA 2 cells had almost 20 times higher PLA 2 activity than the cytosols from vector-transfected control cells, indicating that the ectopic enzyme was functional (Fig. 7D).
Because enzymes inhibited by DEVD-CHO (caspase-3-like enzymes) are necessary for TNF-induced AA release (Fig. 1A) and MCF-7S1 cells are deficient in caspase-3 (29), we next asked whether co-expression of cPLA 2 and caspase-3 would result in enhanced TNF-induced cytotoxicity and AA release. Whereas the ectopic expression of caspase-3 clearly sensitized MCF-7S1 cells to TNF-induced cytotoxicity, the sensitivity of MCF-7S1 cells transfected with both cPLA 2 and caspase-3 was indistinguishable from that of cells transfected with only caspase-3 (Fig. 8). Furthermore, the ectopic expression of caspase-3, cPLA 2 , or their combination had no effect on TNFinduced AA release. Interestingly, ectopic caspase-3 did not increase the cleavage of cPLA 2 to the "caspase-specific" 64-kDa fragment but triggered an appearance of an ϳ50-kDa cleavage fragment resembling that produced by cathepsins in vitro ( Fig.  6 and 8A).
Finally, we asked whether the lack of vimentin, which has been reported to bind cPLA 2 and augment its activation (30), could explain the inability of TNF to activate cPLA 2 in MCF-7S1 cells. Ectopic expression of vimentin or a combination of cPLA 2 and vimentin had, however, no effect on TNF-induced cytotoxicity or AA release in MCF-7S1 cells (data not shown). All together, these data suggest that cPLA 2 is dispensable for TNF-induced AA release and cytotoxicity in MCF-7S1 cells. cPLA 2 Is Also Dispensable for TNF-induced AA Release and Cytotoxicity in iMEFs-To further assess the role of cPLA 2 in TNF-induced death and AA release, we compared TNF-induced death and AA release in cPLA 2 -deficient and wild type iMEFs. Surprisingly, TNF-induced AA release in cPLA 2 -deficient iMEFs was not significantly different from that in cPLA 2expressing wild type cells (Fig. 9A). Furthermore, cPLA 2 -deficient cells displayed slightly increased sensitivity to TNF-induced death (Fig. 9B). DISCUSSION In this report we identified a new role for cathepsin B. In addition to its functions in disposal of cellular proteins in lysosomes, enhancement of tumor cell invasion by proteolysis of extracellular matrix (31), onset of acute pancreatitis (32) and execution of TNF-induced cell death (22), our new data indicate that cathepsin B is an important regulator of TNF-induced AA release. Using pharmacological protease inhibitors, we showed that cathepsin B-like activity mediates TNF-induced release of AA in cancer cell lines of different origins. Furthermore, the significantly reduced ability of TNF to induce AA release in cathepsin B-deficient iMEFs confirmed that cathepsin B protein is a part of this signaling pathway.
Whereas the data obtained from all cell models used in this study was consistent regarding the necessity of cathepsin B for  ). B, the indicated cells were treated with 1-100 ng/ml rhTNF for 48 h and analyzed for survival by MTT assay. C, the AA release from MCF-pEBS7 and MCF-cPLA 2 cells treated in parallel with 1-100 ng/ml rhTNF for 18 was measured as described in the legend for Fig. 1. The release from untreated control cells is set to 100%, and the relative PLA 2 activity in the treated cells is shown. B and C, the values represent the means of triplicate determinations ϮS.D. The experiments were repeated twice with similar results. D, the catalytic PLA 2 activity in cytosolic fractions of MCF-pEBS7 and MCF-cPLA 2 cells was determined by an in vitro PLA 2 enzyme assay. The cPLA 2 activity is expressed as the percentage of hydrolyzed free fatty acids of total fatty acids. The data represent the mean values from two independent experiments ϮS.D.
FIG. 8. Forced expression of caspase-3 sensitizes MCF7-S1 cells to TNF-induced cell death but is not associated with an additional increase in AA release. MCF-7 cells were transfected with indicated combinations of pcDNA-neo (neo), pEBS7-hygro (hygro), pcDNA-casp-3 (casp-3), and pEBS7-cPLA 2 (cPLA 2 ). A, the expression levels of transfected proteins in pool-transfected cells that were left untreated (Ϫ) or treated with 20 ng/ml rhTNF (ϩ) for 24 h were analyzed by Western blot analysis using anti-cPLA 2 -N, anti-caspase-3 or anti-Hsc70. The migration of molecular weight markers is indicated on the right. The survival of cells transfected as indicated above was analyzed by a 48-h MTT assay (B), and the AA release was measured in a parallel experiment after 18 h of stimulation with TNF as described in Fig. 1 (C). The release from untreated control cells is set to 100%, and the relative PLA 2 activity in the treated cells is shown. The values represent the means of triplicate determinations ϮS.D. The experiment was repeated twice with similar results. *, p value Ͻ 0.01; **, p value Ͻ 0.05; data are as compared with similarly treated neo ϩ hygro or neo ϩ cPLA 2 cells, respectively. the TNF-induced AA release, the requirement of AA release for the final execution of the cells was not absolute. The ability of a cPLA 2 /iPLA 2 inhibitor AACOCF 3 to protect WEHI-S cells from TNF-induced death and the nearly perfect correlation between AA release and cell death upon combined treatments with TNF and various protease inhibitors strongly suggest that AA participates in the execution of WEHI-S cells. On the contrary, the inhibition of AA release by CA-074-Me had no effect on the sensitivity of MCF-7S1 cells to TNF, indicating that AA release is dispensable for TNF-induced cell death in these cells. Supporting this hypothesis, ectopic expression of caspase-3 increased the death rate of MCF-7S1 cells upon TNF treatment without affecting TNF-induced AA release. It has been shown recently that the inhibition of PLA 2 activity suppresses TNFand anti-Fas-induced phosphatidylserine exposure in U937 leukemia cells (9). Thus, PLA 2 activity may contribute to cell death by two very different means, i.e. by directly participating in the execution of the cell death and/or by triggering the exposure of phosphatidylserine, which in turn leads to the phagocytosis of the cell in vivo. Furthermore, AA is a ratelimiting precursor for the synthesis of inflammatory prostanoids and leukotrienes. Thus, even in cells where cathepsin B-mediated AA release does not play a direct role in cell death in vitro, it may affect the phagocytosis of cells as well as the overall tissue damage induced by the associated inflammation in vivo.
The dissociation of AA release and cell death in MCF-7S1 cells clearly indicates that TNF-induced AA release is a result of a specific cathepsin B-dependent signaling pathway rather than an unspecific bi-product released from dying cells. Earlier literature suggests that AA release associated with TNF-induced cytotoxicity is mediated by cPLA 2 (3,4). The ability of TNF to induce AA release in cPLA 2 -deficient iMEFs and MCF-7S1 cells challenges this notion. Furthermore, TNF failed to activate ectopically expressed cPLA 2 in MCF-7S1 cells even when the caspase cleavage site of cPLA 2 was mutated or cPLA 2 was co-expressed with its putative co-activators, caspase-3 or vimentin. Whereas iPLA 2 may be the enzyme responsible for TNF-induced AA release in MCF-7S1 cells, enzymes distinct from cPLA 2 and iPLA 2 must mediate the AA release in iMEFs. iMEFs did not express iPLA 2 at levels detectable by Western blot analysis, and their AA release in response to TNF was rather enhanced than inhibited by cPLA 2 deficiency. AACOCF 3 inhibits cPLA 2 and iPLA 2 but not secretory PLA 2 s (1). Thus, the data showing that WEHI-S cells that are protected against TNF-induced cytotoxicity by AACOCF 3 do not express detectable levels of iPLA 2 suggest that cPLA 2 could be the responsible enzyme in these cells. It should, however, be noted that in addition to PLA 2 s, several non-PLA 2 enzymes such as acyltransferases or arachidonyl-CoA synthetase, can trigger increased AA release (1), and the effect of AACOCF 3 on these enzymes is not known. To address the question more directly, we therefore attempted to establish stable subclones of WEHI-S cells with either increased or decreased cPLA 2 expression. Despite great difficulties in altering cPLA 2 protein expression to a degree detectable by Western blot analysis, our data obtained by statistical analysis of 9, 17, or 10 single cell clones transfected with empty vector or cPLA 2 cDNA in sense or antisense orientation, respectively, strongly suggest, that this enzyme is required for TNF-induced cell death in WEHI-S cells. 2 These data are in agreement with earlier results showing that cPLA 2 expression is crucial for the cytotoxic action of TNF in another murine fibrosarcoma cell line, L929 (3). Contrary to MEFs and MCF-7S1 and ME-180as cells, TNF-induced cell death in L929 and WEHI-S cells is caspase-independent and can be inhibited by antioxidants like butylated hydroxyanisole (18,22,33). Thus, cPLA 2 -mediated AA release may be specifically involved in the execution of alternative TNF-induced death pathways that are caspase-independent but reactive oxygen intermediate-dependent in murine fibrosarcoma cells. In line with this, reactive oxygen intermediates have been reported to activate cPLA 2 enzyme via an increase in intracellular calcium and/or lipo-oxygenation (6,34). Butylated hydroxyanisole inhibited TNF-induced AA release in WEHI-S cells, suggesting that reactive oxygen intermediates indeed act upstream of AA release in TNF-treated WEHI-S cells. AA produced by cPLA 2 has in turn been suggested to contribute to TNF-induced cytotoxicity via induction of the mitochondrial permeability transition and cytochrome c release (35). Alternatively, the PLA 2 activity may be involved in the apoptosisassociated accumulation of phosphatidylserine in the outer leaflet of the plasma membrane (9).
The N-terminal cleavage fragment of cPLA 2 may act as a dominant negative molecule by preventing the access of fulllength cPLA 2 to its substrates (9). Our data show that the cleavage of cPLA 2 in WEHI-S cells is a late event occurring clearly after the initiation of the AA release. Moreover, the more pronounced cleavage of cPLA 2 in the presence of caspase inhibitors was associated with increased AA release. The late and incomplete cleavage of cPLA 2 suggests that the full-length cPLA 2 contributes to the cell death in WEHI-S cells, and the limited cleavage may represent a negative feedback response that takes place in cells that are already committed to die.
Whereas our data clearly indicate that cathepsin B activity is required for TNF-induced AA release in tumor cells, the mechanism by which this occurs remains obscure. We have earlier shown that TNF induces the release of cysteine cathepsins from the lysosomal compartment into the cytosol in sensitive target cells (22). This release occurs before cell death and AA release, suggesting that cytosolic rather than lysosomal ca-2 D. Wissing, unpublished data. thepsin B triggers AA release. The ability of pharmacological inhibitors of cathepsin B activity to block TNF-induced AA release indicates that the protease activity of cathepsin B is required for this pathway. Thus, the question arises of whether cathepsin B activity is sufficiently stable at the neutral cytosolic pH. Our in vitro experiments show that cathepsins can cleave cPLA 2 at pH 7.5, giving rise to a specific 50-kDa cleavage fragment, whereas their excessive exopeptidase activity observed at acidic pH is inhibited at neutral pH. Similarly, cysteine cathepsins have been shown to cleave pro-inflammatory caspases (36,37), Bid (38), poly(ADP-ribose) polymerase (39), and protein kinase B/Akt 3 at neutral pH. We could detect a cleavage fragment of cPLA 2 resembling that produced by cathepsins (ϳ50 kDa) only in TNF-treated MCF-casp3-cPLA 2 cells. The requirement of caspase-3 overexpression for this cleavage could be explained by data showing that caspase-3 expression greatly enhances the cytosolic cysteine cathepsin activity in TNF-treated MCF-7S1 cells. 4 TNF-induced cleavage of cPLA 2 into a 64-kDa fragment was inhibited in CA-074-Me-treated WEHI-S cells and cathepsin B-deficient iMEFs, suggesting that cathepsin B activates a cPLA 2 -processing protease in the cytosol of these cells. Even though this unknown protease requires an intact caspase-3 cleavage site, it is unlikely that it is caspase-3 because treatment of cells with 1-10 M zVAD-fmk or 200 M DEVD-fmk enhanced this cleavage. 5 Our data indicate that cathepsins maintain their endopeptidase activity at neutral intracellular pH, and thus, they may also cleave and activate other enzymes that can trigger AA release. Interestingly, cathepsin B-mediated cleavage activates pro-inflammatory proteases including pro-caspases-1 and -11 (36,37). These proteases can in turn activate pro-inflammatory cytokines. Whether caspases-1 and -11 act downstream of cathepsin B in the TNF-induced signaling pathway leading to AA release remains, however, to be studied.
Ischemia induces apoptotic and/or necrotic cell death associated with cytokine production and strong inflammatory response resulting in increased tissue damage. Interestingly, cathepsin B (40 -42) and cPLA 2 (6,8) have been implicated as effector molecules in ischemic tissue damage and cell death. Thus, it is tempting to speculate that cathepsin B could trigger cPLA 2 activation and AA release in such ischemic conditions and thereby control ischemia-associated cell death, inflammation, and phagocytosis. If this proves to be true, cathepsin B inhibitors may turn out to be more effective than expected in the combat against ischemia-induced tissue damage and other conditions associated with cathepsin B-mediated cell death and/or AA release.