Int. J. Cancer: 52,645-652 (1992) C) 1992 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I Union lnternationale Contre le Cancer

CYSTATIN C AND CATHEPSIN B IN HUMAN COLON CARCINOMA: EXPRESSION BY CELL LINES AND MATRIX DEGRADATION Olivier CORIICCHIATO',Jean-Franqois CAJOT',Magnus ABRAHAMSON~, Shu Jin CHAN-?, Daniel KEPPLER' and Bernard SORDAT',4 'Swis.s Institute ,forEsperirnentul Cancer Research (ISREC), 1066 Epnlinge.7, Switzerlund; 2Depnrtnieritof Clinical Chemistry, Univrrsity o f l u r i d , University Hospital, S-22185 Liind, Sweden; iind 3Depurtrnent of Biochemistry and Moleciilar Biology, Unir,ersi/y of Chicago, Chicago, IL 60637, USA. Expression of the cysteine proteinase cathepsin B and i t s physiological inhibitor cystatin C was analyzed in vitro in I human fibrosarcoma and 4 human colon carcinoma cell lines. Cystatin C antigen as well as cathepsin B activity were detected in the conditioned media of the 5 cell lines. The corresponding cell extracts expressed high levels of cathepsin B activity, whereas only trace amounts of cystatin C antigen could be found. Northern-blot analysis revealedthe presence in the 5 cell lines of a 0.8-kb cystatin C mRNA transcript and 2 cathepsin B transcripts of 2.3 and 4.3 kb. Pepsin treatment of tumor-cellreleased cathepsin B induced an average 7.3-fold increase in activity, indicating that the enzyme was mainly present as a latent form in conditioned medium. The pepsin-activatedcathepsin B from one colon carcinoma cell line was further characterized using the cysteine proteinase inhibitors E-64, recombinant cystatin C, a cystatin-C-derived peptidyl inhibitor (Z-LVGCHN,), and cathepsin-B-specificdiazomethyl ketone inhibitors (Z-FT(OBzl)-CHN2, Z-FS(OBzl)-CHN,). This activity was totally neutralized by recombinant cystatin C, suggesting a potential for interaction between released extracellular cathepsin B and cystatin C. In vitro assays of degradation of extracellular matrix showed that cyrteine proteinase inhibitors could decrease matrix degradation induced by pepsin-activated conditioned media. With colon cells, this inhibition was not observed, indicating a requirement for an extracellular activation of latent cathepsin B. Our data provide evidence that cystatin C and latent cathepsin B are both released extracellularly by colon carcinoma cells in vitro. They suggest that cystatin C and cathepsin B interactions may participate, in an as yet unelucidated way, in the modulation of the invasive phenotype of human colonic tumors. CJ 1992 Wilqy-Liss,Inc.

Tumor invasion and metastasis is a multi-step process that involves, at various stages, the penetration of host ECM by cancer cclls. Matrix degradation has been shown to involve a potent proteolytic cascade in which tumor-cell-associated urokinase-type plasminogen activator and collagenase(s) play an important role (Mignatti et a/., 1986). A further refinement in the control of these complex proteolytic events is the recent identification of specific proteinase inhibitors expressed by tumor cclls such as plasminogen activator inhibitors and tissue inhibitors of metalloproteinases. These proteinase inhibitors may represent important regulatory components of the proteolytic potential of tumor cells. Cathepsin B is a lysosomal enzyme involved in physiological events such as intracellular protein catabolism and prohormone activation. The following lines of evidence suggest that tumor-cell-associated cathepsin B may play a role in malignancy and have extracellular functions (reviewed by Sloanc et a/., 1990). ( 1 ) Cathepsin B activity and mRNA levels have been found to be increased in extracts of tumor vs. adjacent tissues (Keppleret a/., 1988; Murnane et al., 1991). ( 2 ) Active cathepsin B and cathepsin-B precursors are secreted into the medium of various cell types in culture, including non-small-cell lung cancer, rabbit V2 carcinoma (Baici and Kndpfel, 1986), and human colon carcinoma cells (Keppler et a/., 1988; Maciewicz et a/., 1989), and have been reported to be localized on the cell surface or associated with cell membranes (Weiss et ul., 1990). ( 3 ) The tumor-cell-associated cathepsin B

remains active near neutral pH, in contrast to the lysosomal enzyme, and to the enzyme produced by normal fibroblasts (Baici and Knopfel, 1986). (4) Cathepsin B is a broadspectrum proteolytic enzyme capable of degrading various constituents of the ECM (Lah et ul.. 1989a), directly or by activation of other proteolytic pathways such as activation of pro-collagenases (Eeckhout and Vaes. 1977). ( 5 ) Intracellular cathepsin-B activity has been correlated with the metastatic potential of tumor cells in various experimental systems, including lung-colony assay (Sloanc et a/., 1990), and with early stages of tumor development. The activity of cysteine proteinases can be regulated by the cystatins, a superfamily of evolutionarily related proteins. Inhibitors of cysteine proteinases have also been found to be associated with malignancy in several studies: (1) The intracellular family 1 cystatins, cystatin A and cystatin B, are negatively regulated in malignant squamous epithelia (Jarvinen er ul., 1987). (2) Cystatin A from human sarcoma has a lowered inhibitory potential compared to cystatin A from human spleen (Lah et ul., 19896). (3) Keren et a/. (1989) reported that butanol extraction of mouse melanoma cells increases cystcine proteinase activity from the membrane fractions, and concomitantly causes release of inhibitors. (4) Using mouse melanoma cell variants, Rozhin et ul. (1990) reported that increased cathepsin-B activity was partly due to a decrease in global endogenous cysteine proteinase inhibitors, and was correlated to the invasive phenotype in vivo. Few reports have addressed the role of the extracellular family 2 cystatins in malignancy, however, which should be of interest for the suggested extracellular function of cathepsin B in such states. Cystatin C is a member of cystatin family 2 which is expressed in a variety of human tissues and which displays a very high affinity for cathepsin B and other human cysteine proteinases. Therefore, it is a potentially important physiological inhibitor of cysteine proteinase activity in human extracelM a r fluids (Abrahamson et nl., 1Y86). To our knowledge, no evidence of secretion of cystatin C by colon carcinoma cells has so far been reported, and no JTo whom correspondence and requests for reprints should be addressed. Ahhrevin/ions: rCystC. recombinant human cystatin C; BSA, bovine serum albumin; DMEM, Dulbecco's modified Eagle's medium; FCS, fetal calf serum, E-64. L-trans-epoxysuccinyl-leucylamido(4-guanidino)butane; EGTA, ethylene glycol-tetraacetate: 2-Arg-Arg-NHMec, N-benzyloxycarbonyl- L-arginyl- L-arginine- 4-methyl -7 coumarylamide; 2-LVG-CHN2, N-benzyloxycarbonyl-L-leucyl-L-valyl-L-glycine diazomethyl ketone, Z-FT(OBz1)CHN.. N-benzyloxycarbonyl-Lphenylalanyl-L-0-benzyl-threonine-dIazomethylketone; Z-FS(0Bzl)CHN>, N-benzyloxycarbonyl-L-phenylalanyl-L-O-benzyl-serinediazomethyl ketone; Pro-uPA. pro-urokinase type plasminogen activator; DTT, d,l-dithiothreitol: ECM. extracellular matrix; NHzMec, 4-methyl-7-coumarylamine; MAb. monoclonal antibody.

Received: April 16, 1992 and in revised form July 15, 1992.

646

CORTICCHIATO ET AL.

demonstration of a possible extracellular regulation of secreted, tumor-associated, cysteine proteinases by cystatin C has yet been established. The present study illustrates the concomitant expression and secretion of cystatin C and cathepsin B by human colorectal cancer cell lines and analyzes their possible interactions in the context of an enzyme-inhibitor pair for regulated extracellular proteolysis. MATERIAL AND METHODS

with good linearity up to 6 hr of reaction. All assays were performed in parallel with serial dilutions of purified bovine cathepsin B showing a linear release of NH2Mec (r = 0.99). Increasing dilutions of NH2Mec were used to establish a calibration curve. Sample activity is expressed as cathepsin B milliunits/lOh cells (one unit of cathepsin B is defined as the amount of enzyme that will release 1 p M of NHzMecimin). Latent cathepsin B was activated by treatment with pepsin (Keppler et a!., 1988). Samples were acidified to pH 3.0 with 10 M HCI before addition of 100 pg/ml pepsin and incubated for 1 hr at 37°C. Cathepsin-B activity was then assayed after adjusting the pH of the samples to 6.3 with 1 M NaOH. The “total cathepsin-B activity” was defined as the activity measured in conditioned medium after the pepsin activation. Inhibition of cell-secreted cathepsin-B activity was analyzed using pepsin-treated conditioned medium from Col15 cell culture. Before addition of the fluorogenic substrate, conditioned medium or DMEM alone (blank) was incubated for 15 min at 37°C with cysteine proteinase inhibitors: rCystC, E-64, Z-LVG-CHN2. Z-FS(0Bzl)-CHN,, and Z-FT(0Bzl)-CHN2, at concentrations ranging between and 10-I” M. Respective blanks were subtracted.

Material Z-Arg-Arg-NHMec was purchased from Bachem (Dubendorf, Switzerland), and bovine cathepsin B, DTT, E-64, ortho-phenanthroline, pepstatin, EDTA, E-ACA, aprotinin, pronase, pepsin (grade 1:10,000) and purified laminin (from EHS-sarcoma) from Sigma (Buchs, Switzerland). Leupeptin was from Boehringer Mannheim (Rotkreuz, Bern, Switzerland), NaIz5I from Amersham (Zurich, Switzerland), iodogen from Pierce (Rockford, IL), RNA molecular size markers from Gibco-BRL (Basel, Switzerland) and 10-kDa cut-off filtration membrane from Amicon (Diaflo ultrafiltration membrane, Lexington, MA). rCystC was produced in E. coli and the cystatin-C-derived inhibitor Z-LVG-CHN? synthesized as ear- Assay of cystatin C antigen Samples were assayed for cystatin C antigen in a double lier (Abrahamson et al., 1991). Z-FS(OBzl)CHN2 and Z-FT(OBzl)CHN2 (Shaw et al., 1983) were kindly donated by Dr. E. sandwich enzyme immunoassay, as described in detail (Olafsson et al., 1988): monospecific antibodies from a rabbit Shaw (Basel, Switzerland). antiserum raised against cystatin C from human urine were used to coat microtiter plates. Dilutions of samples to be Cell culture assayed were added, followed by a murine MAb against human The cell lines used in this study include 4 human colon cystatin C. The MAb used, HCC3, recognizes human cystatin carcinomas (HT29, SW620, Col12, Co115) and the HT1080 C both in free form and in complex with cysteine proteinases, human fibrosarcoma. All cells were obtained from the ATCC and does not cross-react with human cystatin A. B, S, or SU. (Rockville, MD), except Co112 and Co115 which were previ- Cystatin-C-bound MAbs were detected with horseradish peroxously established in our laboratory. Tumor-cell-conditioned idase-labelled rabbit antibodies against mouse immunoglobumedia were obtained by growing cells to confluence in 75-cm2 lins (DakoPatts, Copenhagen, Denmark), and enzyme activity flasks with DMEM containing 10% FCS. The medium was was measured using hydrogen peroxide/ABTS as substrate then removed and the cells washed twice with DMEM and (hO5). Dilutions of a solution of isolated rCystC were used to cultured for 30 hr under serum-free conditions in 30 ml construct a standard curve. DMEM -0.1% BSA. Conditioned media were harvested, centrifuged for 5 min at 750 g and stored at -80°C until use. Cell extracts were prepared by washing the culture flasks with Noithenz -blot a tia lysis Poly(A)+-RNA was isolated, subjected to 0.9% agarose gel 20 ml of 0.9% NaCl followed by scraping into 5 mlO.9% NaCI. This material was washed twice with 10 ml 0.9% NaCl and electrophoresis (4 pg RNA per lane) and transferred to a centrifuged for 5 min at 750g then the cells were counted and Gene Screen (NEN, Boston, MA) membrane. R N A molecular analyzed for viability by the Trypan-blue exclusion test ( > 90% size markers were included in the gel. Isolation, transfer and viability). The cell pellet was resuspended at 2 x loh cclls/ml hybridization conditions, as well as processing of filters, were in lysis buffer (10 mM sodium-phosphate buffer, pH 6.3, 0.2% as described by Qian et al. (1989). The filters were preTriton X-100). Cell lysis was achieved by 20 successive aspira- hybridized for 1 hr at 42°C and subsequently hybridized for 18 tions through a Pasteur pipette followed by freezing and hr at 42°C in the presence of 108 cpm of random-primerthawing. The cell extracts were then centrifuged for 20 rnin at labeled cDNA probes. The probes used in this study were the PstI-EcoRI (1.1 kb) fragment of plasmid phCB79 (= human 10.000g and the supernatants stored at -80°C. cathepsin B cDNA clone) and the EcoRI (0.78 kb) fragment of pUC18iC6a (= human cystatin C cDNA clone) (Chan et al., Assay of cathepsin-B activity 1986; Abrahamson et al., 1987). Filters were then washed Cathcpsin-B activity was determined fluorometrically using conditions of high stringency (3 X 10 rnin at 65°C in 300 the synthetic substrate Z-Arg-Arg-NHMec (Keppler et al., under m l 2 x SSC, 0.1% SDS and 3 X 20 min at 20°C in 0.1 X SSC, 1988). Briefly, samples (conditioned media or extracts) were 0.1% SDS) and exposed to Kodak type XAR-5 X-ray film. mixed with an equal volume of reaction buffer (0.2 M phosphate buffer, p H 6.3,2 mM EDTA, 0.2 mM DTT, 0.4% Triton X-100, 5 p M Z-Arg-Arg-NHMec) and incubated for 5 hr at Extracellular matrix degradation assay Preparation of ’H-proline-radiolabeled ECM (R22-ECM) 37°C. The reaction was stopped by adding 1/10 vol of 0.5 M iodoacetic acid and the release of NHIMec was monitored with was performed essentially as described by Jones and De CIerck a spectrofluorometer (Kontron, Basel, Switzerland, type (1982). Matrix degradation was initiated by plating (at day 1) SFM25.340 nm excitation, 433 nm emission wavelengths). The 50,000 cells per 13-mm-diameter well in 1 ml DMEM containspecificity of the substrate for cysteine proteinases was as- ing 5% FCS; 24 hr later, medium was aspirated, the wells were sessed using the E-64 inhibitor. To determine the linearity of washed twice with serum-free DMEM, and respective inhibithe assay, kinetic studies were carried out by removing at tors (aprotinin, rCystC, lIYbM) diluted in serum-free DMEM various times aliquots of the reaction mixtures and measuring were added (or serum-free DMEM alone in control wells). At their fluorescence after stopping the reaction. Preliminary days 3 and 4, supernatants were collected and counted for results showed a time-dependent increase in activity over 24 hr released radioactivity in a scintillation counter. Fresh serum-

CYSTATIN C AND CATHEPSIN B

647

free DMEM and inhibitors wcre added daily. To ensure that each well contained an equivalent amount of radioactivity, total degradation was induced at day 4 by adding pronase to the wells (0.1 mg/ml, 48 hr incubation). Assays were performed in quadruplicate. For cell-free experiments, conditioned media, previously activated by pepsin (and adjusted to pH 6.3) or not, were incubated in the wells for 48 or 96 hr in the prcsence of DTT (1.5 mM). DMEM treated under the same conditions was incubated in control wells, and these respective blanks were subtracted from the samples. Results are expressed as relative percentage of degradation compared to the degradation induced by non-activated conditioned medium or by cells '1 I one. Degradation assay using i2rI-lamitiin-coated plates Laminin was iodinated with Na1251by the iodogen method. '?51-laminin-coated plates were prepared as follows: glutaraldehyde was allowed to bind covalently to the polyvinyl surface of 24-well tissue-culture plates (2.5% in 0.1 M sodium bicarbonate buffer, pH 9.5) for 2 hr at room temperature; 0.22~-filtered '"1-laminin (150,000 cpm per well) mixed with 1 pg per well of non-labeled laminin in 0.2 ml of PBS buffer was incubated overnight at room temperature. Plates were then washed twice with HzO, incubated overnight with DMEM containing 1 nigiml BSA to saturate residual free radicals of glutaraldehyde, and stored at 4°C. For the cell-free assays, pepsinactivated conditioned medium (adjusted to pH 6.3) from Col15 cells was concentrated (9.5 X ) by membrane filtration (cut-off 10 kDa) and added to the plates (500 pl per well) in the presence or absence of 1.5 mM DTT and inhibitors (aprotinin, 1W5 M; rCystC lo-" M). As a control, DMEM was treated under the same conditions (pepsin treatment, DTT, inhibitors), and incubated in coated wells. After 48 hr incubation. cpm released were counted in a gamma counter, and the respective blanks were subtracted. For cell-induced degradation, 100,000 Co115 cells were seeded per coated well in DMEM-SC; FCS at day 1. At day 2, medium was changed for serum-free DMEM after 2 washings and inhibitors were added M; Z-LVG-CHN2, M; (E-64, 1W5 M; rCystC, 7 x Z-FS(OBzl)CHN2, M; aprotinin, M; pepstatin, M; orthophenanthroline. M; EDTA, M; leupeptin, M). Medium was collected at day 3 and counted for radioactivity. then fresh serum-free medium and inhibitors wcre added. At day 4 the medium was counted. RESULTS

Expression arid release of cathepsin B by colon carcinoma arid fihrosarconia cells Transcription of cathepsin B was analyzed by Northern blots in 4 colon carcinoma cell lines (HT29, SW620, Co115, Co112) and 1 fibrosarcoma cell line (HT1080), to determine qualitatively the size of the cathepsin B and cystatin C transcripts expressed. Hybridization of the filters with a human cathepsin B cDNA probe revealed a major transcript of 2.3 kb as well as a larger 4.3-kb transcript (Fig. l a ) for all the cell lines. Cathepsin-B activity was then investigated in conditioned media from cultures of the cells. Figure 2a shows that cathepsin-B activity could be directly detected, after minimal dilution (2-fold), in crude conditioned media derived from all cell lines investigated as well as after pepsin treatment, which should activate latent forms of the enzyme present in the media. Conditions for activation by pepsin pre-treatment were optimized and found to be optimal for 1 hr incubation and 0.1 mgiml pepsin at p H 3.0 (data not shown). The presence of a secreted latent form of cathepsin B could thus be identified in all conditioned media analyzed. The relative increase in activity shown in Figure 2a was variable among cell lines. i.e., activity increased 2.8-, 5.1-, 7.3-, 8.2- and 13.3-fold for Co112, Co115, SW620, HT29 and HT1080 cells, respectively. The

FIGURE 1 - Northern-blot analysis of cathepsin B ( a ) and cystatin C (b) mRNA from human colon carcinoma and fibrosarcoma cells. The same filter was used for the successive hybridization to human cDNA probes specific to cathepsin B and cystatin C as described in the text. The poly(A)+-RNA samples that were blotted to the filter after electrophoresis were from: lane 1, HT29 colon carcinoma cells; 2, SW620 cells; 3, Co115 cells; 4, C0112 cells; 5 , HT1080 fibrosarcoma cells.

extent of cell death, as provoked by overgrowth and medium acidification, did not significantly influence the levels of secretion or of spontaneous activation (data not shown). The corresponding cell extracts also expressed cathcpsin-Bspecific activity (from 0.026 to 0.332 mU/lOfI cells). No correlation between extra- and intra-cellular levels could be established. Cystatin-C expression and secretion The conditioned media used for cathepsin-B analyses were also used to determine the levels of released cystatin-C antigen. All 5 cell lines tested released cystatin C into the culture medium, ranging between 8 ng/lOh cells and 57 ng/106

648

CORTICCHIATO ET AL. 0.00

0.10

0.20

HT29

B

HT29

HT29

SW620 SW620

SW620

COl15

COl15

co112

c0112

Coil2

HT1080

HT1080

HT1080

I

C0115

I

0.00

1.00

2.00

Calhepsin B activity (mU/106cells)

0

10 20 30 40 50 60 Cystatin C antigen (ng/106cells)

70

0

1

2 3 4 5 6 Relative ratio Cathepsin B I Cystatin C

FIGURE 2 - Cathepsin-B activity and cystatin-C antigen released by cultured cancer cells. Results are shown for HT29, Co115 and Col12 colon carcinoma cell lines and HT1080 fibrosarcoma cells. Conditioned media were collected after 30 hr incubation at 37°C from subconfluent flasks. The fluorogenic substrate assay of cathepsin-B activity and the immunoassay of cystatin C are described in the text. Results were normalized to a constant amount of cells (ie., rnU cathepsin-B activity and ng cystatin-C/106 cells) and represent the average 2 SD of 2 separate experiments performed i n triplicate. ( a )Cathepsin-B activity was measured before and after incubation with pepsin as described in the text. (b) Conditioned media were analyzed for the presence of cystatin-C antigen. (c) Relative ratio of cathepsin-B activity over cystatin-C antigen released in conditioned media. Results are expressed in arbitrary units, individual ratios being compared to that of Co112 cells to which a relative value of 1 was given.

cells (Fig. 2b). In contrast, trace amounts of cystatin-C antigen only were detected in cell extracts (below 1 ng/106 cells, data not shown). Northern-blot analysis of corresponding mRNAs confirmed the concomitant expression of cystatin C and cathepsin B: the same filter as was used for cathepsin B analysis, rehybridizcd with a human cystatin-C cDNA probe, revealed that a single 0.8-kb mRNA transcript was expressed by all 5 cell lines investigated (Fig. lb).

Ratio between secreted cathepsin B and cystatin C When assessing the relative ratio of total (ie., pepsin generated) cathepsin B activity over cystatin C antigen concentrations in conditioned cell media, it was observed that HT1080 and Col IS cells exhibited a higher ratio cornpared to the other cell lines (Fig. 2C). Both cell lines have been shown to express a proteinase-mediated prominent capacity to degrade ECM (Cajot et al., 1990). Inhibition of secreted cathepsin-B activity by ~ w i o u proteinase s inhibitors In the context of inhibition of cell-associated cystcine proteinase activities, a variety of inhibitors were tested for their ability to modulate total cathepsin-B activity expressed extracellularly by the tumor cells. A dose-dependent inhibition of CollS-cell-secreted cathepsin B was obtained with the broad-spectrum inhibitor of cysteine proteinases, E-64, as well as for rCystC, the cystatin-C-based peptidyl diazomethyl ketone inhibitor Z-LVG-CHN2(rate constant for human cathepsin B inactivation at p H 6.0: 10,600 M-l s-l; Abrahamson et a/., 1991) and with the diazomethyl ketone inhibitors designed to be specific to cathepsin B, Z-FS(OBzl)-CHN2 and Z-FT( 0Bzl)CHN2 (rate constants for bovine cathepsin-B inactivation at pH 5.4: 2,330 M-I s-l and 64,000 M-' s-I M, respectively; Shaw et al., 1983) (Fig. 3 ) . Complete inhibition was obtained with 0.1 FM of E-64 or rCystC, with 1.0 pM of Z-LVG-CHN2, or Z-FT(0Bzl)-CHN2 and with 10 KM of Z-FS(OBzl)-CHN2. A similar pattern of inhibition was also obtained when purified bovine cathepsin B was uscd as a control (data not shown). confirming that the activity detected after activation in thc conditioned media was of cathcpsin-B type.

lnhlbltor concentratlon ( M )

FIGURE3 - Inhibition of tumor-cell-secreted cathepsin B by various cysteine protease inhibitors. The latent cathepsin B present in conditioned medium derived from Co115 coloncarcinoma cells was activated by pepsin treatment before incubation with increasing concentrations of inhibitors or with buffer alone as described in the text. Residual activity was measured and expressed as a percentage of the activity obtained in the absence of inhibitors. Molar concentrations of inhibitors are indicated on the horizontal axis.

Extracellular matrix degradation by conditioned culture media and cells Biosynthetically radiolabeled ECM produced by smoothmuscle cells (R22) was used to test the capacity of cells or conditioned media to lyse matrix constituents. Previous studies had demonstrated that the R22 ECM contains collagens, glycoproteins, elastin and proteoglycans (Jones and DeClerck, 1982). The inset in Figure 4a shows that cysteine proteinases (here purified bovine cathepsin B) required reducing conditions (presence of DTT) to significantly degrade the ECM. The lysis induced by conditioned media was also enhanced in the presence of DTT (Fig. 4a): media activated by pepsin were able to induce a more than '-fold increase in R22-ECM

649

CYSTATIN C AND CATHEPSIN B

Actlvated Cond. Med. + DlT

*I

-1

Actlvated Cond. Med. Cells + rCyst C Cond. Med. t DlT

a n d . Med.

17

m+

Relatlve 56 degradation so loo

O' 0

200

100

300

0

Relatlve X degradallon

25

50

75

100

I 125

Relatlve X degradation

81

Activated Cond. Med +DlTtrCystC Actlvated Cond. Med. t DlT

1 0

I

1000

2000

3000

Relatlve cpm released

FIGURE4 - Degradation of R22-ECM and '2SI-Laminin by conditioned culture medium from Co115 cells. (a) Conditioned media (Cond Med) from Co115 cells in culture were incubated for 96 hr on R22 plates, and the influence of pepsin activation and DTT sensitivity is shown. Respective blanks were subtracted and results are given as relative percentages of degradation compared to the degradation induced by non-treated conditioned medium. Inset: Two dilutions (SmU: Sigma milliunits, units as defined by the manufacturer) of purified bovine cathepsin B were incubated for 48 hr on R22-cell-derived ECM as described in the text. DTT was added to show the dependence of cathepsin B on oxidoreductive conditions. Results are given as relative percentages of degradation compared to that for 100 SmU of bovine cathepsin B under reducing conditions. (6) 12sI-laminincoated plates were incubated (48 hr, 37°C) with conditioned media (Cond. Med.) from Col15 cells. These media had previously been activated by pepsin and concentrated 9.5 fold by membrane filtration (cut-off 10 kDa). DTT (1.5 mM) and inhibitors (rCyst C: M. aprotinin (Aprot): lo-' M) were added prior to incubation. Aliquots of 400 FI were then counted for radioactivity and respective blanks (DMEM + respective treatments) were subtracted. degradation in the presence of DTT, compared to conditioned mcdia not treated. This degradation was further characterized on 1251-laminincoated plates. Concentrated (9.5 X ) conditioned media generated a radioactive release which was inhibited by aprotinin (a serine proteinase inhibitor), but also by rCystC (Fig. 4 h ) . These data confirm the involvement of cysteine proteinases in laminin degradation in vitro. Under the conditions of the experiments, rCystC inhibited 76% and aprotinin 60% of the degradation. Taken together, these results illustrate the protcolytic potential of tumor-cell-secreted latent forms of cysteine proteinases, in particular that of cathepsin B.

FIGURE5 - Degradation of a 3H-proline-ECM by Co115 cells. Co115 cells (50 x lo3)per well were seeded onto 3H-proline-ECM produced by R22 cells (R22-ECM) in the presence of 5% FCS. At day 2, medium was changed for serum-free DMEM and inhibitors were added (aprotinin and rCyst C. both at 10-"M). The release of radiolabeled fragments was measured at day 3 (as described in the text). Degradation in the presence of inhibitors is expressed as percentages of the degradation induced by cells alone. Additional experiments using Col15 cells plated on R22ECM were carried out to monitor the degradative potential of the cells in a direct cell-matrix interaction. Inhibitors were added to determine which class(es) of proteolytic enzymes were implicated in the degradation. Inhibitors used in concentrations similar to those used in previous experiments showed little or no cytotoxic or cytostatic activity (data not shown). Cells were seeded in the presence of 5% FCS for 24 hr. and then cultured in the absence of FCS. No exogenous plasminogen was added to the culture medium. The pattern of degradation was the same during the first and the second 24 hr of FCS starvation. This could indicate that seric inhibitors from the FCS, which could be retained by the ECM, did not interfere with the assay. However, similar results were obtained when cells were cultured in DMEM-2% FCS. Co115 cells, as well as the other 4 cell lines tested, were able to release a soluble radioactivity which was dependent on cell number and time corresponding to labelled proteolytic matrix fragments (data not shown). Figure 5 shows that aprotinin was able to prevent the cell-mediated degradation, whereas rCystC had no significant effect. Similar results were obtained when Col15 cells were seeded on lzSI-labelledlaminin-coated wells and cultured under serumfree conditions: neither E-64, rCystC. pepstatin, leupeptin, ortho-phenanthroline, Z-FS(OBzl)CHN2, Z-LVG-CHN2, nor EDTA were able to inhibit the degradation at the concentrations indicated in the "Results". In contrast, aprotinin at M lowered the degradation to about 50% of the initial level (data not shown). DISCUSSION

Previous studies have suggested a relationship between expression of cathepsin-B activity and the malignant phenotype (reviewed by Sloane et al., 1990). However, the involvement of cathepsin B in malignancy has remained circumstantial. In this regard, some conflicting results have been reported (Ostrowski et al., 1986; Persky ef al., 1986), and no direct evidence for an extracellular function of cathepsin B, or for a regulation of extracellular cathepsin B activity by cystatin(s), has been demonstrated. The actual events which directly or indirectly control the cathepsin-B-related proteolytic potential of a given tumor-cell type are not yet understood.

650

CORTICCHIATO E T A L .

Cuthepsiti B Our data show that 4 human colon carcinoma lines and one fibrosarcoma line all express cathepsin-B mRNA and secrete cathepsin B. Two transcripts of cathepsin-B mRNA were found. Both contain the full protein-encoding region and their difference in size can be explained by alternative splicing in the 3’ untranslated region of the precursor mRNA by the use of diverse sites of polyadenylation (Qian et a/., 1989). Expression of a largesized transcript appears not to be tumor-specific. Nevertheless, malignant tissues (vs. adjacent tissues) were found to express a higher level of this transcript, which has been correlated with early stages of progression in colonic tumors (Murnane et al., 1991). We have shown that the tumor-cell-secreted cathepsin-B activity was reproducibly increased by pepsin treatment, indicating that the enzyme is mainly secreted in a latent form. Although secretion of cathepsin B by colon carcinoma cells in culture had been previously reported. the immunological technique used did not differentiate between the activities expressed by the mature enzyme after pepsin activation and the inactive proform (Maciewicz et al., 1989), or the studies did not address the possible concomitant secretion of an inhibitor (Keppler et a/., 1988). Secretion of latent cathcpsin B, activatable by pepsin treatment, has been described in various models. The latent form is now considered as a true proenzyme, although interactions of cathepsin B with inhibitors yielding high-MW inactive complexes offer an additional explanation of latency, which has been poorly investigated. The question of the abnormal mechanism(s) of secretion, which misroute procathepsin B in many tumor cell types, also remains unanswered. Other cysteine proteinases are expressed by tumor cells: cathepsin L is another lysosomal cysteine proteinase, which requires acidic pH to be functional. In contrast to cathepsin B, no tumor-associated form of cathepsin L, expressing pH optimum close to neutrality, has been found. In the present study, cathepsin B and cathepsin L were differentiated by using specific substrates, and by an inhibitory pattern toward specific inhibitors. Cjlstatiti C A significant aspect of our study is the demonstration that cystatin C is expressed and secreted by one fibrosarcoma and 4 colon carcinoma cell lines. It has often been suggested that interaction(s) with cysteine proteinase inhibitors would be a crucial point in the regulation of cysteine proteinase activities. Nevertheless, no endogenous secreted inhibitor has, to our knowledge, been earlier identified in cancer cells. Our inhibition studies of activated conditioned media demonstrated that the cysteine proteinase activity expressed by colon cancer cells is sensitive not only to synthetic inhibitors but also to rCystC. These results confirm that the released enzyme activity measured indeed has cathcpsin B specificity since lysosomal bovine cathepsin B gave the same pattcrn of inhibition. In this context, a comparison between the total amount of extraccllular cathepsin B activity and secreted cystatin C is of interest. We showed that 2 of the cell lines, Co115 and HT1080, expressed a higher ratio than the other cell lines. In plasma membrane fractions of mouse melanoma cell subpopulations, Rozhin et al. (1989) calculated a similar ratio for cathepsin L and cysteine proteinase inhibitors and this ratio correlated with “high” metastatic potential as tested in a lung colony assay. The cathepsin-Bicystatin-C ratio could indicate, for these 2 particular cell lines, an imbalance in favor of the proteolytic enzyme.

As discussed above, it is possible that cystatin-C antigen detected in the samples of conditioned media represents an inhibitor-enzyme complex form. Total inhibition of Col15secreted cathepsin-B activity was obtained with 100 nM of recombinant cystatin C. However, only about 1.2 nM of endogenous cystatin C was detected in conditioned medium of these cells (Fig. 2, based on lo6 cells/ml medium). According to the inhibition experiments using rCystC (Fig. 3), a 1.2-nM concentration is able to inhibit 50% of the total cathepsin B present in the medium. These data indicate that the amount of cystatin C released by tumor cells may only be sufficient to partially inhibit the coexpressed cathepsin B. The Ki value of 0.25 nM for cystatin C (Barrett et al., 1984) is compatible with a total cathepsin-B inhibition at 100 nM and a partial inhibition at 1.2 nM. However, these in vitro observations may differ significantly from the in vivo situation where the enzyme/ inhibitor balance may vary to a large extent in response to hormones, growth factors, and cell-matrix interactions. The rather high concentration of cystatin C in plasma (75 nM) (Abrahamson et al., 1986) may also contribute to in vivo regulation of cysteine proteinase activity by tumor cells. Alternatively, as previously reported for cystatin A in human sarcoma cells (Lah et ul., 19896), tumor cystatin C could be altered and could display a reduced inhibitory capacity towards cysteine proteinases, in particular cathepsin B. Under non-malignant conditions, an allelic variant of cystatin C has been identified in the vessel wall of patients with amyloid angiopathy (Ghiso et al., 1986). Cathepsiti B itivolvement in ECM proteolysis The co-expression and concomitant release by human colon carcinoma cells of a cathepsin-B-like cysteine proteinase and of cystatin C underline some of the complex mechanismswhich may participate in pericellular proteolysis. However, no direct experimental evidence in support of cathepsin-B-mediated degradation of ECM components, either by secreted or by membrane-associated enzyme, has to our knowledge yet been provided. Sloane et al. (1990) found a correlation between intracellular cathepsin-B level and lung colonization in vivo. Inhibition of ECM degradation by cysteine proteinase inhibitors has only been shown in an amnion invasion assay with cathepsin-L inhibitors (Yagel et al., 1989). In contrast, neither Mignatti et al. (1986), Persky et al. (1986), nor Keren et ul. (1989) wcrc ablc to dcmonstratc a cystcinc proteinase inhibitor sensitivity in their models. We confirm that, under cellculture conditions, the secreted cathepsin B appears essentially unable to induce significant laminin and E C M degradation. This contrasts to the results of Weiss et al. (1990), showing that plasma-membrane-associated cathepsin B from bladder carcinoma cells appears able to degrade laminin at pH 6.2 without prior activation. Our results indicate that, under the conditions of the assay, a serine proteinase-dependent pathway is mainly responsible for proteolysis of ECM components. This can be interpreted as the involvement of plasmin in ECM proteolysis, and furthermore indicates that no detectable cell-induced activation of latent cysteine proteinases occurs in culture on ECM. In contrast to the results obtained with cell cultures, we show that induced (i.e., pepsin) activation of the latent form of cathepsin B present in conditioned media can unmask a cysteine proteinase activity now able to lyse ECM components at pH 6.3. This was demonstrated by the dependence on oxido-reductive conditions, and inhibition by rCystC. Taken together, our data corroborate activity measurements showing that cathepsin B is mainly released in a latent form, suggest that no activating mechanism occurs under culture conditions, and demonstrate the proteolytic potential of latent secreted cathepsin B.

65 1

CYSTATIN C AND CATHEPSIN B

Thesc apparcntly conflicting results indicate that particular in V i l a mechanisms may trigger a cysteine-proteinase-related

protcolytic pathway involving multiple regulatory steps. In this regard, intratumoral environment and stromal influences could be critical in regulation of the cathepsin-B-related activity at 3 levels at least: oxido-reductive conditions potentiating cysteine proteinase activities, presence of other proteinases able to activate proforms, and presence (and distribution) of cysteine proteinase inhibitors such as cystatin C. We have previously shown that colon carcinoma cell lines express significant levels of plasminogen activators, which convert plasminogen into active plasmin, a potent ECM proteolytic enzyme. and plasminogen activator inhibitors (Cajot ct ul., 1990). We now report that colon carcinoma cells in addition to expressing serine proteinase activity, also release cathepsin B and its inhibitor cystatin C. In breast tumors, Sloane et al. (1990) have proposed that cathepsin-B precursors may be activated in vitv by cathepsin D (a pathway cstablished by Nishimura eta/., 1988). This activation could also be induced, in other tumor types, by metalloproteinases (Eeckhout and Vaes, 1977). Kobayashi et a/. (1991) have shown that pro-uPA, either soluble or bound to a cell receptor, can be activated in vitro not only by traces of plasmin or kallikrcin but also by purified cathepsin B. Our results, although not directly supportive, do not contradict this hypo-

thetical cooperative pathway which could establish a link between the extracellular expression of latent cysteine proteinase activities and the serine-proteinase-dependent degradation of ECMs. Every step of these proteolytic cascades of enzyme activation may in turn be regulated at different levels: pH dependence for cathepsin D, PA-inhibitors for serine proteinases (Cajot et al., 1990) and, as suggested in this study, cystatins (e.g., cystatin C) for cysteine proteinases. Further investigations into the actual mechanism of cathepsin B activation in vivo, the nature of the enzyme-tumor-cellsurface association, and the potential interactions with natural inhibitor(s) should contribute to a better understanding of the exact role of this tumor-associated cysteine proteinase in malignancy. ACKNOWLEDGEMENTS

We are grateful to Mr. D. Bachmann for technical assistance, and to Dr. Trin-Thang Chi&n (ISREC, Epalinges) for helpful comments during the preparation of the manuscript. This work was supported by the Swiss Science Foundation (grants 3.401-0.86 and 31.26642.89), by the “Ligue NeuchBteloise contre le Cancer” and by the Swedish Medical Research Council (projects 09291 and 09915).

REFERENCES ABRAHAMSON, M., BARRETT,A.J., SALVESEN.G. and GRUBB,A,, Isolation of six cysteine proteinase inhibitors from human urine. Their physicochemical and enzyme kinetic properties and concentrations in biological fluids. J. bid. Chem.. 261, 11282-1 1289 (1986). ABRAHAMSON, M., GRUBB,A,, OLAFSSON,I. and LUNDWALL, A,, Molecular clonine and seauence analvsis of cDNA coding for the precursor of the himan cydeine proteinase inhibitor cystaticc. FEBS Lett.. 216,229-233 (1987). ABRAHAMSON. M., MASON,R.W., HANSSON. H., BUTTLE,D.J., GRUBB, K., Human cystatin C: role of the N-terminal A. and OHLSSON. segment in the inhibition of human cysteine proteinases and in its inactivation by leucocyte elastase. Biochem. J., 273,621-626 (1991). BAICI. A. and KNOPFtl., M., Cysteine proteinases produced by cultured rabbit V2 carcinoma cells and skin fibroblasts. Inf. J. Cancer, 38, 753-761 (1986). B A R R E n , A.J., DAvits. M.E. and GRLIBB, A,, The place of human y-trace (cystatin C) amongst the cysteine proteinase inhibitors. Biod7em. hiopliys. RPS.Cornm., 120. 631-636 (1984). CAJOT, J.F., BAMAT.J., BERGONZELLI, G., KRUITHOF, E.. MEDCALF, R.. TESTUZ,J. and SORDAT,B., Plasminogen-activator inhibitor type 1 is a Dotent natural inhibitor of extracellular matrix deeradation bv tibroiarcoma m d colon carcinoma cells. Proc. nut. Atad. %I. (Wask.i 87,6339-634Y (1990) CHAN,S.J., SkGUNDO, B.S., MCCORMICK, M.B. and STEINER, D.F.. Nucleotide and predicted amino acid sequences of cloned human and mouse preprocathepsin B cDNAs. Proc. nut. Acad. Sci. (Wash.), 83, 7721-7725 (1986). ELCKHOUT, Y . and VAES, G., Further studies on the activation of procollagenase, the latent precursor of bone collagenase. Effects of lysosomal cathepsin B. plasmin and kallikrein, and spontaneous activation. Biochem. J.. 166,21-31 (1977). GHISO.J., JENSSON. 0. and FRANGIONE, B., Amyloid fibrils in hereditaw cerebral hemorrhage with amvloidosis o f Icelandic tvne is a vahant o f gamma-trace