Contact of high-invasive, but not low-invasive, melanoma cells to native collagen I induces the release of mature cathepsin B

Authors


Abstract

Metastasis of malignant tumor cells involves cell–cell and cell–matrix interactions, which regulate the expression and localization of proteolytic enzymes. In the present study, we investigated the expression and localization of the lysosomal cysteine proteinase cathepsin B and its natural inhibitors cystatin A, B and C in high- (MV3), intermediate- (SKmel28) and low-invasive (SKmel23, WM164) human melanoma cell lines grown on plastic or in contact with monomeric or fibrillar collagen type I. Neither the transcript levels of cathepsin B nor those of the natural inhibitors, cystatin B and C, were altered by the interaction of melanoma cells with collagen type I. However, protein expression and cellular localization of cathepsin B and its inhibitors were markedly affected. In contrast to low-invasive cells, high-invasive cells constitutively released procathepsin B when cultured on plastic. In addition, contact of invasive cells with fibrillar collagen type I resulted in the release of both mature forms of the protease. Perturbation studies using inhibitory antibodies against the β1 subunit of the integrin receptor indicated a role for the β1 integrin receptor family in the regulation of cathepsin B release. Cystatin B protein expression was much lower in high-invasive cells in both culture conditions, when compared to low-invasive cells. Cystatin C expression was comparable in all cells, but cell contact to fibrillar collagen type I induced its expression. These results strongly implicate a pivotal role of cell–matrix interactions for the regulation of cathepsin B localization and activity in melanoma cells. © 2005 Wiley-Liss, Inc.

Tumor invasion and metastasis are multistep processes, which include tumor–host interactions mediated by cell–cell and cell–matrix contacts.1 To invade adjacent tissue and metastasize, tumor cells have to cross various extracellular barriers like extracellular matrix (ECM), basement membranes and interstitial stroma. Degradation of ECM components and remodeling of connective tissue and basement membranes requires the concerted action of various classes of intra- and/or extracellular proteases.2 These include serine proteases (plasmin, tissue-type plasminogen activator, urokinase-type plasminogen activator and Cat G), matrix metalloproteases (MMP-2, -9, -1, -7, -3 and MT1-MMP), aspartic proteases (Cat D) and cysteine proteases (Cat B, H and L).3, 4, 5, 6, 7, 8

Cat B (EC 3.4.22.16) is a lysosomal cysteine protease of the papain family of proteases,9 which is also involved in a number of disease processes, such as osteoarthritis10, 11 and rheumatoid arthritis.12 This enzyme has also been shown to be involved in tumor invasion and metastasis. Increase in Cat B activity or upregulation of transcripts/protein expression was found in various types of carcinomas, like breast,13 lung,14 gastrointestinal15 and urogenital16, 17 carcinomas. In addition, Cat B has also been shown to be overexpressed in chondrosarcomas,18 gliomas,19 osteoclastomas20 and melanomas.21, 22, 23 In human esophageal tumors, amplification of the cathepsin B gene was first identified.24 In pathological conditions, altered expression, redistribution, secretion and increased activity was reported to be caused by elevated transcription, by the use of alternative promoters and by alternative splicing.25, 26

Cat B is synthesized on the rough endoplasmic reticulum as a preproenzyme containing a signal sequence, which is cleaved off to reveal the inactive 46 kDa proenzyme that is transported from the Golgi to late endosomes where it is converted to an active 31 kDa single-chain (sc) form. Further processing of the enzyme into an active double-chain (dc) form of 25/26 and 5 kDa occurs in lysosomes, the final localization of the enzyme.27 Its major function inside lysosomes is the degradation of proteins that have been taken up by the cell or originate from other cellular compartments.28, 29 However, besides the classical Cat B trafficking pathway, 2 alternative pathways are proposed to take place in tumor cells: (i) proCat B is secreted from the cells via the secretory pathway where it may exist extracellularly as soluble or membrane-bound proCat B; (ii) active Cat B is trafficked to secretory lysosomes, which either undergo exocytosis or fuse with the plasma membrane, resulting in soluble or membrane-bound active Cat B, respectively.27 The functional significance of tumor-associated Cat B is ascribed to its participation in ECM degradation, either by direct proteolysis of matrix components or by activating other proteases, thereby initiating proteolytic cascades.27 Active Cat B can extracellularly activate proMMP-330 and prouPA.31 Mai et al.32 observed an interaction of ProCat B with the annexin II light chain (p11) on the surface of human breast carcinoma and glioma cells, which might lead to activation of proCat B. Annexin II as well as prouPA, uPA and uPAR were reported to be localized in caveolae.33, 34 Colorectal carcinoma cells (HCT 116) contain a mutant K-ras gene, which mediates the localization of Cat B to caveolae.35 Furthermore, using caveolin-1-antisense transfection of HCT 116 cells, it could be demonstrated that caveolin-1 regulates the expression and localization of Cat B, prouPA, uPAR and p11.35, 36 By using quenched-fluorescent collagen type IV as a substrate to analyze degradation patterns of 2 breast cancer cell lines, Sameni et al.37 showed that, in BT20 cells, Cat B was involved in the extracellular degradation of the substrate, whereas in BT549 cells, degradation by Cat B occurred intracellularly and was dependent on endocytic uptake of the substrate. Szpaderska and Frankfater38 identified a role of intracellular Cat B in matrigel invasion and demonstrated decreased invasiveness of A375M melanoma cells after transfection of these cells with a Cat B-antisense expressing vector.

Abbreviations:

CA-074, (L-3-trans-propylcarbamoyloxirane-2-carbonyl)-L-isoleucyl-L-proline; Cat, cathepsin; 3D/2D/1D, three/two/one-dimensional; DTT, dithiothreitol; E-64, trans-epoxysuccinyl-L-leucylamino-(4-guanidino)-butane; ECM, extracellular matrix; MMP, matrix metalloprotease; MT, membrane-type; proCat B, procathepsin B; uPA, urokinase-type plasminogen activator; uPAR, urokinase-type plasminogen activator receptor.

The activity of cysteine proteases is regulated by various endogenous inhibitors of the cystatin superfamily39, 40 like cystatins (family I and II) and kininogens (family III). Cystatin A and B (family I) are the main intracellular inhibitors of cysteine proteases, whereas cystatin C (family II) acts primarily extracellularly being a secreted inhibitor.41 Several groups have suggested that alterations in the expression of cysteine proteases and cystatins at mRNA and protein levels, as well as in the activity and trafficking of enzyme and inhibitor molecules, correlate with the malignancy of various tumors.6

In a recent study, we could demonstrate marked extracellular cysteine protease activity generated by high-invasive melanoma cells invading dead deepidermized human dermis. Inhibition of these activities led to a significant reduction of invading tumor cells. Extracellular Cat B may, therefore, play a major role for the invasive process of melanoma cells.42

Interaction of cells with ECM components plays a significant role in regulating the expression of matrix degrading proteases. For instance, Langholz et al.43 could show the important role of α2β1 integrin receptors in the regulation of MMP-1 expression in human skin fibroblasts. By using a similar approach, Zigrino et al.44 demonstrated that also MT1-MMP expression is regulated by the activation of α2β1 integrin receptors. Koblinski et al.45 suggested an important role of α1β1 and α2β1 integrins in the induction of proCat B secretion in human breast fibroblasts cultured on collagen type I. In a recent study, Podgorski et al.46 analyzed prostate cancer cell–stroma interactions in the context of bone microenvironment in vitro and in vivo. The authors proved that these interactions modulate the expression and activity of Cat B.

The purpose of the present study was to analyze the role of cell–matrix interactions on the expression and cellular localization of Cat B in human melanoma cell lines of different invasive abilities. Since Cat B is involved in intra- and also extracellular degradation of protein substrates, we also investigated the expression and localization patterns of the cysteine protease inhibitors, cystatin A, B and C.

Our results clearly indicate that Cat B expression correlates with the invasiveness of the cells, being the strongest in invasive and the lowest in noninvasive melanoma cells. On the contrary, the expression of its endogenous inhibitor cystatin B was markedly reduced in high-invasive cells. Analysis of the Cat B distribution revealed that only high-invasive melanoma cells constitutively release proCat B and the occurrence of extracellular active sc or dc forms of the protease were exclusively observed after interaction of the cells with native collagen type I fibrils.

Material and methods

Antibodies

Monoclonal mouse antibodies directed against human cathepsin B (clone CB 59-4B11; final concentration 2 μg/ml) and human cystatin B (clone RJMW 2E7; final concentration 100 ng/ml) were kindly provided by Dr. E. Weber, Institute of Physiological Chemistry, Martin Luther University Halle-Wittenberg, Germany. The polyclonal goat antihuman cystatin C (clone P-14; final concentration 4 μg/ml) antibody was purchased from Santa Cruz Biotechnology (Heidelberg, Germany). Rabbit antihuman actin (clone C11; final concentration 6 μg/ml) polyclonal antibody was from Sigma–Aldrich Chemie GmbH (Taufkirchen, Germany). Horseradish peroxidase-labeled secondary antibodies, rabbit antimouse IgG (650 ng/ml), swine antirabbit IgG (140 ng/ml) and rabbit antigoat IgG (250 ng/ml) were purchased from DAKO (Hamburg, Germany). The monoclonal mouse antibody 4B4 directed against the human β1-integrin chain was obtained from Coulter (Hialeah, FL); the rat antihuman monoclonal antibody mAb13, directed against the human β1-integrin chain, as well as the purified rat IgG2aκ controls were from BD Pharmingen (Heidelberg, Germany). The monoclonal mouse antibody P1E6 (antihuman α2 integrin chain) as well as the purified mouse IgG1 was purchased from Chemicon International (Hofheim, Germany). The monoclonal antibodies 6F1 and 12F1, both directed against the α2-integrin chain, were kindly provided by Prof Barry Coller (Rockefeller University, New York) and Prof. VL Woods Jr. (University of California, San Diego, USA), respectively.

Cell culture

HaCaT and Jurkat cells were cultured in either DMEM (Invitrogen GmbH, Karlsruhe, Germany) or RPMI-1640 (PAA Laboratories GmbH, Pasching, Austria) both enriched with 10% fetal calf serum (PAA Laboratories GmbH, Pasching, Austria), 2 mM glutamine (Biochrom AG, Berlin, Germany), 0.01% (w/v) L-ascorbic acid (Sigma–Aldrich Chemie GmbH, Taufkirchen, Germany) and 100 U/ml each of streptomycin and penicillin (Biochrom AG, Berlin, Germany). Cells were cultured either as monolayers (HaCaT) or in suspension (Jurkat) and routinely passaged by trypsinization.

Four human melanoma cell lines were used for the following experiments: the high metastatic cell line MV347 and the low metastatic cell lines SKmel28,48, 49 SKmel2348, 50 and WM164.51 All cell lines were characterized in an in vitro invasion model described by Dennhöfer et al.42 We have used a modified system to characterize melanoma cell invasion through basement membrane. The cell lines were classified according to their ability to penetrate the basement membrane and designated as high- (MV3), intermediate- (SKmel28) or low-invasive (SKmel23, WM164).

All melanoma cell lines were maintained in RPMI-1640 supplemented with 10% fetal calf serum, 2 mM glutamine, 100 mM each of nonessential amino acids (Biochrom AG, Berlin, Germany) and 100 U/ml each of streptomycin and penicillin. Cells were routinely passaged by trypsinization.

Culture in/on collagen type I matrices

Monomeric type I collagen coatings (1D; IBFB, Leipzig, Germany) were prepared by overnight incubation of culture dishes with 100 μg/ml collagen solution in 0.1% (v/v) acetic acid at 4°C. Fibrillar collagen coating (2D) and 3D collagen lattices were prepared as previously described.52 Briefly, monomeric collagen type I and 10× RPMI-1640 were combined in a 10:1 ratio and neutralized by the addition of 0.1 N NaOH, thus allowing fibril formation (final collagen concentration, 1 mg/ml). 3.5 × 104 cells/cm2 were seeded onto uncoated, 1D-, 2D-coated plates or into 3D lattices (3.5 × 104 cells/cm3). After 48 h, coated plates and lattices were washed 3 times with PBS and incubated in serum-free medium for additional 24 h. Control cultures of melanoma cells were performed on uncoated (monolayers) tissue-culture plates.

For the inhibition studies within 3D collagen lattices, protease inhibitors were added to the collagen–cell suspensions prior to polymerization. The following inhibitors were used: E-64 (10 μM, Serva, Heidelberg, Germany), CA-074 (10 μM, Merck Biosciences, Schwalbach/Ts., Germany), 1,10-phenanthroline (5 mM, Sigma–Aldrich Chemie GmbH, Taufkirchen, Germany), Pefabloc®SC (0.1 mM and 0.25 mM, Serva, Heidelberg, Germany) and Pepstatin A (1–20 μM, Serva, Heidelberg, Germany). Controls were set up by adding methanol, the solvent of Pepstatin A and 1,10-phenanthroline (0.68 and 0.5%, respectively). After 48 h, lattices were washed 3 times with PBS and incubated in serum-free medium containing newly added inhibitors for additional 24 h. Cell viability after inhibitor treatment was assessed by trypan blue viability test using 0.2% (v/v) trypan blue solution (Sigma–Aldrich Chemie GmbH, Taufkirchen, Germany).

For blocking experiments with integrin antibodies, cells were harvested from confluent monolayer cultures by trypsinization, centrifuged (5 min, 272g) and resuspended in complete media. Cells were preincubated in the presence of antibodies (4.5 μg/1.5 × 105 Zellen) for 30 min at 37°C and incorporated into 3D collagen lattices.

Preparation of conditioned media, cell lysates and crude plasma membranes

Serum-free conditioned media from monolayers, 1D-, 2D- and 3D- cultures were centrifuged (5 min, 272g) to eliminate debris and suspended cells and analyzed. Equal amounts of conditioned media, normalized to the cell number (amount of conditioned media according to 1.5 × 105 cells), were precipitated with trichloroacetic acid (10% (w/v) final concentration)/Triton X-100 (0.1% (v/v) final concentration) and protein pellets resuspended in Laemmli sample buffer (50 mM Tris, pH 6.8, 100 mM DTT, 2% (w/v) SDS, 0.1% (w/v) bromphenol blue, 10% (v/v) glycerol). Cell numbers were obtained by isolating cells either by trypsinization (monolayer cultures) or by extraction from collagen type I (1D, 2D and 3D) with 400 U/ml collagenase type I (Cell Systems®, St. Katharinen, Germany) at 37°C for 10 min. After washing with PBS for 2 times (10 min, 420g), cells were resuspended in PBS enriched with a protease inhibitor mixture (1 mM EDTA, 2 mM AEBSF, 130 μM bestatin, 14 μM E-64, 1 μM leupeptin, 0.3 μM aprotinin); Sigma–Aldrich Chemie GmbH (Taufkirchen, Germany). Cell numbers were determined after trypan blue exclusion test with 0.2% (v/v) trypan blue solution (Sigma–Aldrich Chemie GmbH, Taufkirchen, Germany) to asses the total number of viable cells. Lysates (amounts normalized according to 2 × 104 cells) were prepared by sonication on ice (3 × 10 s, cycle 0.5, 60% amplitude [126 μm, 276 W/cm2]) followed by centrifugation (5 min, 1,660g). The preparation of crude plasma membrane was performed as described earlier.44

SDS-PAGE and immunoblot analysis

Cell lysates, crude plasma membranes or concentrated conditioned media, normalized to the cell number and dissolved in Laemmli sample buffer, were subjected to SDS-PAGE using 15% (w/v) acrylamide gels. Proteins were transferred to Hybond-C Extra™ (Amersham Biosciences, Braunschweig, Germany) in a semidry blotting chamber. Following blockage of nonspecific binding sites with 5% (w/v) milk in Tris-buffered saline (TBS) containing 0.1% (v/v) Tween 20 (TBS-T), blots were incubated for 16 h at 4°C with primary antibodies. After incubation with horseradish peroxidase-labeled secondary antibodies in 5% milk-TBS-T, reactive proteins were detected using enhanced chemiluminescence (ECL™ Western Blotting Detection Reagents, Amersham Biosciences, Braunschweig, Germany). Purified human liver cathepsin B (250 ng/lane; Merck Biosciences, Schwalbach/Ts., Germany) was used as a positive control.

RNA isolation, reverse transcription and polymerase chain reaction

Total RNA was isolated from HaCaT cells, Jurkat cells or melanoma cells grown as monolayers, 1D-, 2D- or 3D- cultures, using peqGOLD TriFast™ according to the manufacturer's instructions (peQLab, Erlangen, Germany). Collagen lattices were disrupted by mechanical shearing through 20-G syringes for 10 times prior to RNA extraction. Two microgram of total RNA was reverse transcribed according to the manufacturer's instructions (SuperScript™ II RNAseH-reverse transcriptase, Invitrogen GmbH, Karlsruhe, Germany) using oligo d/T (Invitrogen GmbH, Karlsruhe, Germany) as primers in a total volume of 20 μl. 2-μl cDNA were subjected to PCR amplification with specific primers for cathepsin B (NCBI accession number NM_001908) forward 5′-CTGGTCAACTATGTCAACAAACGG-3′ and reverse 5′-GAAGTCCGAATACACAGAGAAAGC-3′ (amplified using RNA from WM164/monolayer), cystatin B (accession number NM_000100) forward 5′-ACCCCAGCGCCTACTTGG-3′ and reverse 5′-GGAAGAGAAATGCAAAAGCAGC-3′ (amplified using RNA from MV3/monolayer), cystatin C (accession number BC013083) forward 5′-AGCCAGCAACGACATGTACC-3′ and reverse 5′-GCTACTATTTTATTGCAGGAGGTGG-3′ (amplified using RNA from HaCaT/monolayer), cystatin A (accession number NM_005213) forward 5′-GAAATCCAGGAGATTGTTGATAAGG-3′ and reverse 5′-CATGACTCAGTAGCCAGTTGAAGG-3′ (amplified using RNA from Jurkat/monolayer) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH, accession number BC023632) forward 5′-TCATGACCACAGTCCATGCCATCAC-3′ and reverse 5′-GCCAAATTCGTTGTCATACCAGGAAATGA-3′. Amplification of GAPDH cDNA was used as a positive control of each PCR reaction. Negative controls were performed for each primer pair by omitting the corresponding cDNA template. Amplification of cDNA was achieved within 30 cycles of denaturation (94°C, 30 s), annealing (54°C, cathepsin B; 59°C, cystatin B; 62°C, cystatin C; 58°C, cystatin A and GAPDH, 30 s) and extension (72°C, 40 s, cathepsin B; 30 s, cystatin B, cystatin C and GAPDH; 20 s, cystatin A). PCR products were analyzed on 1.2% (w/v) agarose gels in Tris acetate EDTA buffer (TAE) containing 0.5 μg/ml ethidium bro'mide. Amplified fragments were identified by size and by sequencing after purification with E.Z.N.A. Cycle-Pure (peQLab Biotechnologie GmbH, Erlangen, Germany), according to the manufacturer's instructions. Characterized amplification products of cathepsin B, cystatin A, B and C were employed as cDNA probes for Northern blot analysis as described below.

Northern blot analysis

Total RNA isolated from cells, 5–10 μg, were resolved by 1% formaldehyde agarose gel electrophoresis and transferred overnight onto positively charged nylon membranes (peQLab, Erlangen, Germany). Membranes were UV-crosslinked and hybridized (42°C, overnight) to random primed 32P-labeled cDNA probes. The 32P-labeling for all cDNA probes was carried out using Ladderman™ Labeling Kit (TaKaRa Europe S.A., Gennevilliers, France), according to the manufacturer's instructions. After hybridization, membranes were washed 2 times at RT in 2× SSC, 0.1% (w/v) SDS, and hybridized probes were visualized on Hyperfilm™ ECL (Amersham Biosciences, Braunschweig, Germany). All experiments were quantified by determining the signal intensities using ImageJ 1.32j software (Wayne Rasband, National Institutes of Health) and normalized to the signal intensities of the 18S loading control.

Results

Expression and distribution of intra- and extracellular Cat B in melanoma cell lines cultured as monolayers or in 3D collagen type I lattices

Human melanoma cell lines, characterized by a different invasive capacity in vitro and in nude mice, were compared with respect to Cat B expression by Northern blot analysis. Densitometric evaluation and normalization to 18S rRNA levels revealed up to 5-fold higher levels of the 4 and 2.2 kb Cat B transcripts in high-invasive MV3 and intermediate-invasive SKmel28 cells grown on plastic, when compared to the low-invasive SKmel23 and WM164 cells (Fig. 1). Interaction of cells with ECM components has been reported to be responsible for the modulation of various proteases including Cat B.43, 44, 45, 53, 54, 55 We, therefore, analyzed whether interaction of melanoma cells with a relevant component of the ECM, such as collagen type I, may regulate the expression and cellular localization of Cat B. Compared to the corresponding monolayer controls, no or little alterations of transcript levels were observed upon the interaction of the melanoma cell lines with native collagen type I (Fig. 1).

Figure 1.

Northern blot analysis of Cat B mRNA in MV3, SKmel28, WM164 and SKmel23 melanoma cell lines. Cells were either cultured as monolayers on plastic culture dishes (ML) or in 3D collagen type I lattices (3D). Five microgram of total RNA were resolved by formaldehyde agarose gel electrophoresis and blotted onto positively charged nylon membranes. Both major transcripts of Cat B (2.2 and 4 kb) were detected after hybridization with a specific 32P-labeled cDNA probe, generated as described in “Materials and Methods”. As a control for loading 18S rRNA, hybridizations of the membranes are shown. The quantification of Cat B mRNA transcripts was performed densitometrically and normalized according to 18S rRNA signals.

High levels of intracellular Cat B protein were detected in the high-invasive MV3 and the intermediate-invasive SKmel28 cells, when compared to the low-invasive SKMel23 and WM164 cells. No differences in the intracellular levels of Cat B were observed when melanoma cell lines were either grown as monolayers on plastic or seeded into 3D collagen type I lattices (Fig. 2). Mature single-chain (sc) and double-chain (dc) forms of Cat B were detected in all cell lysates analyzed. The dc form of the enzyme consists of a heavy- (dc-h) and a light-chain (dc-l), connected via a disulfide bond.56

Figure 2.

Comparison of intracellular Cat B protein of MV3, SKmel28, WM164 and SKmel23 cells. Cells were either cultured as monolayers on plastic culture dishes (ML) or in 3D collagen type I lattices. Cell lysates were normalized according to the cell number corresponding to 2 × 104 cells and analyzed by immunoblotting using mouse antihuman Cat B primary antibody (clone CB 59-4B11; final concentration 2 μg/ml). Purified human liver Cat B (250 ng) served as a positive control (Lane 1). sc, single-chain; dc-h, heavy-chain of the double-chain form.

Interestingly, analysis of conditioned media from cells grown as monolayers showed that only MV3 cells released proCat B, whereas SKmel28 cells released only the sc form (Fig. 3a). On the contrary, low-invasive SKmel23 and WM164 cells released no Cat B under these culture conditions (Fig. 3a).

Figure 3.

Extracellular Cat B protein in conditioned media of MV3, SKmel28, WM164 and SKmel23 melanoma cell lines. Cells were either cultured as monolayers on plastic culture dishes (ML) or in 3D collagen type I lattices. Conditioned media were normalized according to the cell number corresponding to 1.5 × 105 cells, precipitated and analyzed by immunoblotting using mouse antihuman Cat B primary antibody (clone CB 59-4B11; final concentration 2 μg/ml). Purified human liver Cat B (250 ng/lane) served as a positive control. Level of extracellularly released Cat B in melanoma cell lines cultured as monolayers (a) and in 3D collagen type I lattices (b). The additional 30–40 kDa form of Cat B is detected, exclusively after contact of MV3 cells to collagen I (see asterisk). sc, single-chain; dc-h, heavy-chain of the double-chain form.

Also, after contact to 3D fibrillar collagen type I (3D), no Cat B protein was found in the conditioned media of the low-invasive SKmel23 and WM164 cells (Fig. 3b), whereas in conditioned media of high-invasive MV3 cells, in addition to proCat B, both mature forms of the protease (sc and dc-h forms) were present (Fig. 3b). Furthermore, an additional band corresponding to Cat B was detected at a molecular weight between 30 and 40 kDa (Fig. 3b, see asterisk), which might hypothetically correspond to glycosylated single-chain Cat B as an intermediate form occurring during proCat B processing. Conditioned media of SKmel28 cells showed, in addition to single-chain Cat B, also the fully processed dc form of the protease (Fig. 3b, sc and dc-h), when seeded into collagen type I lattices.

Analysis of crude plasma membrane fractions of melanoma cell lines either cultured on plastic or in collagen lattices failed to show any membrane-bound form of the enzyme (data not shown).

Expression and distribution of cystatin A, B and C

Regulation of protease activity does not only occur through the regulation of synthesis and secretion, but also through the inhibitory activity exerted by endogenous inhibitors. To gain more insight into the putative regulatory mechanisms implicated in the regulation of Cat B activity, we have analyzed the expression and distribution of 3 cysteine protease inhibitors in high- and low-invasive melanoma cell lines (MV3 and WM164, respectively). No remarkable differences in cystatin B and C transcript levels in MV3 compared to WM164 cells were detected, when the cells were cultured as monolayers on plastic or after contact of cells to native collagen type I, demonstrated by densitometric analysis of the inhibitors transcripts versus the corresponding 18S signals (Fig. 4a, left panel and Fig. 4b, left panel). Cystatin A transcripts were not detectable in both cell lines and in both culture conditions (data not shown). Furthermore, conditioned media and lysates were analyzed with respect to inhibitor expression and distribution via immunoblotting. The amount of cystatin B protein varied among the cell lines being almost undetectable in MV3 cells and low in SKmel28 as compared to the SKmel23 and WM164 cells lines (Fig. 4a, middle). Interaction of the cells with native collagen type I resulted in a considerable increase of cystatin B protein in WM164 (Fig. 4a, right panel), whereas only a slight increase was detected in MV3 cells (Fig. 4a, right panel), when compared to monolayer controls. Extracellular cystatin C protein was detected in conditioned media of MV3 and WM164 cells only after growth into 3D collagen type I lattices, but not in conditioned media from monolayer cultures (Fig. 4b, right panel). Cystatin C was detected only as a dimeric form of 26 kDa molecular weight, whereas 300 ng of the purified human cystatin C clearly appeared in the monomeric form with a molecular weight of 13 kDa. SDS-PAGE and immunoblot analysis of 500 ng purified human cystatin C revealed the dimerization of the inhibitor due to the high protein concentration and elevated temperatures (Fig. 4b) (Abrahamson, M.; personal communication).

Figure 4.

Expression and localization of cystatin B (A) and cystatin C (B) in melanoma cells. Cells were either cultured as monolayers on plastic culture dishes (ML) or in 3D collagen type I lattices. Ten microgram of total RNA were analyzed by Northern blot. Membranes were hybridized with 32P-labeled cDNA probes for cystatin B and C, generated as described in “Materials and Methods” and detecting specific mRNA species, respectively. Differences in loading are shown by 18S rRNA hybridization of the membranes. The quantification of cystatin B and C mRNA transcripts was performed densitometrically and normalized according to 18S rRNA signals. mRNA levels versus 18S are presented as arbitrary units below each corresponding lane. Cell lysates corresponding to 2 × 104 cells were analyzed by immunoblotting using mouse antihuman cystatin B primary antibody (clone RJMW 2E7; final concentration 100 ng/ml). Equal loading of protein samples is monitored by immunoblotting using polyclonal rabbit antihuman actin antibody (clone C11; final concentration 6 μg/ml). Cell conditioned media were normalized according to the cell number corresponding to 1.5 × 105 cells, precipitated and analyzed by immunoblotting using polyclonal goat antihuman cystatin C antibody (clone P-14; final concentration 4 μg/ml). Purified human cystatin C (300 or 500 ng) served as a positive control (lane 5 and 6, respectively).

Influence of collagen type I conformation (monomeric or fibrillar) for the release of Cat B by melanoma cell lines

To determine whether the appearance of extracellular mature sc or dc forms of Cat B depends on the interaction of melanoma cells with a complex network of native collagen type I fibrils (3D), we cultured high-invasive MV3 cells on culture dishes coated with either monomeric (1D) or a thin layer of fibrillar collagen type I (2D). Conditioned media of cells cultured on dishes coated with monomeric collagen type I, revealed exclusively proCat B (Fig. 5), whereas conditioned media of cells grown on 2D coated dishes revealed the same results as cells seeded into 3D collagen lattices. Upon contact to fibrillar collagen type I, also sc and dc forms of Cat B were detected in conditioned media of MV3 cells (Fig. 5, see also Fig. 3b).

Figure 5.

The role of cell-collagen fibril interactions in extracellular mature Cat B. MV3 cells were cultured on dishes coated with either monomeric or fibrillar collagen type I. After 72 h of culture, the conditioned media were normalized to the cell number (1.5 × 105 cells), precipitated and analyzed by immunoblotting using mouse antihuman Cat B primary antibody (clone CB 59-4B11; final concentration 2 μg/ml). Purified human liver Cat B (250 ng, lane 3) served as a positive control. sc, single-chain; dc-h, heavy-chain of the double-chain form.

Processing of extracellularly released proCat B

MV3 cells grown as monolayer cultures secrete prevalently proCat B, however, after interaction of MV3 cells with collagen type I fibrils, in addition to secretion of proCat B, we also detected both mature forms of the enzyme (Fig. 3b). We, therefore, analyzed, whether the extracellularly released proCat B is activated by autocatalysis or is processed by other proteases present in the extracellular environment. For this purpose, MV3 cells were seeded into 3D collagen type I lattices and incubated for 72 h in the presence of specific protease inhibitors in concentrations nontoxic for the cells as assessed by trypan blue exclusion test. Neither the addition of E-64 nor CA-074 led to a conspicuous increase of proCat B (Fig. 6a), when compared to the control conditioned media (cells incubated without inhibitor). After incubation with 1,10-phenanthroline or with Pefabloc®SC, total extracellular Cat B was remarkably decreased (Fig. 6b). In these conditions, collagen gel contraction was completely inhibited. Adding the same inhibitors to cells cultured on 2D coated dishes revealed also a decrease of total extracellular Cat B (data not shown). No influence of aspartic proteases on Cat B activation was observed, as pretreatment of cells with increasing concentrations of the specific inhibitor Pepstatin A, even at the highest concentrations used (20 μM), revealed no differences in the amount of extracellular Cat B, as compared to the control conditioned media of cells cultured without inhibitor or of cells incubated with 0.68% (v/v) methanol the solvent of Pepstatin A (Fig. 6c).

Figure 6.

The role of extracellular proteases in proCat B processing. MV3 cells were cultured in 3D collagen type I lattices in the absence (control) or presence of protease inhibitors (as indicated) or their corresponding solvent MeOH. (a) 10 μM E-64, 10 μM CA-074; (b) 5 mM 1,10-phenanthroline (final concentration of solvent 0.5% MeOH), 0.1 mM and 0.25 mM Pefabloc; (c) 1–20 μM Pepstatin A (20 μM of the inhibitor correspond to a final concentration of 0.68% MeOH). Conditioned media were normalized to the cell number according to 1.5 × 105 cells, precipitated and analyzed via immunoblotting using mouse antihuman Cat B primary antibody (clone CB 59-4B11; final concentration 2 μg/ml). Purified human liver Cat B served as a positive control (250 ng/lane). sc, single-chain; dc-h, heavy-chain of the double-chain form.

Involvement of collagen binding integrin receptors in cathepsin B release and processing

The mature forms of Cat B were exclusively found in high- and intermediate-invasive melanoma cell lines after interaction of the cells with native collagen type I (Fig. 3b). Binding of cells to collagen type I and contraction of the lattices are known to occur through specific integrin receptor molecules e.g., α2β1 integrin, which has already been shown to be involved in the regulation of several proteases43, 44, 57 and also in the induction of proCat B secretion in human breast fibroblasts.45 To analyze whether the collagen binding integrin α2β1 plays a role in mobilization of mature Cat B from an internal pool or in processing of extracellular proCat B, inhibition studies using several monoclonal antibodies raised against the α2 and β1 chains of the integrin receptor were performed. MV3 cells were preincubated with inhibitory antibodies to the integrin chain α2 (P1E6, 6F1, 12F1) or β1 (4B4, mAb13), or by a combination of both and cultured within 3D collagen lattices. Compared to control lattices, which were incubated with cells in the absence of inhibitory antibodies, no reduction in collagen fibril contraction was observed when cells were treated with P1E6. Also, after preincubation with 6F1 or 12F1, only a slight contraction of collagen fibrils could be detected. Furthermore, neither preincubation of MV3 cells with P1E6, 6F1 nor 12F1 (Fig. 7a) significantly altered the total amount of extracellular Cat B, when compared to the control conditioned media of cells incubated with purified IgG1. Both inhibitory antibodies raised against the β1 integrin chain, 4B4 and mAb13 blocked the ability of MV3 cells to contract collagen fibrils according to previously published data.58, 59 Preincubation with mAb13 led to a decrease in extracellular proCat B, as compared to the control conditioned media of cells incubated with purified IgG2aκ (Fig. 7a). Only when MV3 cells were treated with 4B4, a marked reduction of both proCat B and mature forms of the protease was detected (Fig. 7a). Similar reduction in extracellular Cat B levels was also observed after preincubation with a combination of both anti-α2 and anti-β1 inhibitory antibodies. Extracellular Cat B levels in these conditions were similar to those observed after preincubation with solely the β1 integrin blocking antibody mAb13 (Figs. 7a and 7b).

Figure 7.

Involvement of collagen binding β1 integrin receptors in Cat B release. (a) MV3 cells were preincubated with either inhibitory anti-α2 (P1E6, 6F1, 12F1) or anti-β1 (4B4, mAb13) antibodies or (b) with a combination of both (P1E6 / mAb13, 6F1 / mAb13, 12F1 / mAb13) and cultured in 3D collagen type I lattices. Controls were set up using purified IgG1 and IgG2aκ. After 72 h of culture, the conditioned media were normalized to the cell number according to 1.5 × 105 cells, precipitated and analyzed via immunoblotting using mouse antihuman Cat B primary antibody (clone CB 59-4B11; final concentration 2 μg/ml). Additional bands of the IgG heavy or light chains are caused by the recognition of mouse IgG1 isotype by the mouse antihuman Cat B antibody. sc, single-chain; dc-h, heavy-chain of the double-chain form.

Discussion

Previous studies by Fröhlich et al.22 observed an upregulation of Cat B mRNA, protein expression and increased activity levels in melanoma specimens, when compared to normal skin tissues. These authors also suggested a positive correlation between increased Cat B transcripts expression and the malignancy of melanoma cells.

To gather more insight in the molecular mechanisms underlying tumor-associated Cat B regulation, we now compared Cat B transcription and protein expression levels of high-, intermediate- and low-invasive human melanoma cell lines and observed a clear correlation of increased intracellular Cat B protein and mRNA transcripts with the invasive capacity of cells. Transcripts as well as levels of intracellular Cat B protein were elevated in high-invasive MV3 and intermediate-invasive SKmel28 cell, in comparison to low-invasive melanoma cell lines. In vivo, regulation of protease activities involves interactions of tumor cells with components of the ECM as well as interactions with cellular components of the peritumoral stroma. In vitro, several studies have explored the role of cell–matrix interactions in the regulation of proteolytic activities. Various authors have shown that culture of human dermal fibroblasts in contact with fibrillar type I collagen regulates the expression and activity of several MMPs such as MMP-1, -3, -13 and MT1-MMP.43, 44, 53, 54 A similar modulatory role of fibrillar collagen on the expression and activation of MMP-2 has also been shown for melanoma cells.55 We have herein addressed whether native fibrillar collagen type I also exerts a similar regulatory role toward synthesis, cellular localization and activity of Cat B. However, no changes in either Cat B transcripts or intracellular protein levels were observed in the various melanoma cell lines when cultured in contact with collagen.

As a number of studies have described the redistribution of Cat B to the plasma membrane and also the secretion of the protease,15, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70 we also characterized the extracellular occurrence of Cat B. Here, we observed that, in contrast to low-invasive cells, high-invasive MV3 cells constitutively released proCat B. In addition to proCat B, the extracellular occurrence of both mature forms of the protease was dependent on cell–collagen I interactions and solely observed in high-invasive MV3 and intermediate-invasive SKmel28 cells. The increase of total Cat B protein, after cell–collagen I interaction, was not paralleled by enhanced mRNA transcription thus suggesting that it is released from intracellular pools (e.g., lysosomes). In agreement with this last hypothesis, studies on 12T human breast fibroblasts and MCF-10A epithelial cell lines revealed increases in Cat B protein in the absence of transcriptional regulation.45, 71 These authors hypothesized that enhanced Cat B expression may be regulated at the posttranscriptional level by either stabilization of Cat B protein or increasing translation due to cell–collagen type I interactions. Elevated translation might also be a result of alternative splicing and of the use of alternative promoters.26

Cavallo-Medved and Sloane27 proposed recently that proCat B is secreted by cancer cells via the secretory pathway and exists extracellularly as soluble proCat B or as a membrane-associated component. Several proteases, including Cat D72 and G, tPA, uPA and elastases,73, 74 have been shown to activate proCat B. In vitro, we could not observe any inhibition of proCat B processing upon treatment with cysteine and aspartic protease inhibitors (E-64 and Pepstatin A). Inhibition of Cat B activity by CA-074 failed to show a significant increase of extracellularly proCat B thus excluding autocatalytic processing. In contrast, no extracellular released Cat B was detected when MV3 cells were treated with either MMPs or serine proteases inhibitors, in these conditions the cells also lost their ability to contract collagen I fibrils. Several proteases have been shown to be involved in cell-migration processes, namely serine proteases and MMPs such as MMP-1, -2, -3, -12 and -13.75, 76, 77 It has also been shown, that contraction of collagen lattices is decreased in MMP-3 deficient fibroblasts.78, 79 We conclude that the occurrence of extracellular mature Cat B results from cell–collagen type I interactions, which are dependent also on MMP activity.

Conventional lysosomes can, in many cell types, respond to rises in the intracellular free Ca2+ concentration by fusing with the plasma membrane.80, 81 In addition, Linebaugh et al.82 reported that secretion of active Cat B in tumor cells is independent of proCat B secretion. We found high concentrations of sc and dc-h Cat B in conditioned media of high- and intermediate-invasive cell lines after interaction with native collagen I, probably as a consequence of stimulated lysosomal exocytosis. In MV3 cells, small amounts of a differentially glycosylated form of Cat B (30–40 kDa) were detected extracellularly. This additional form most probably originates from glycosylated proCat B, which is secreted from the Golgi apparatus or transGolgi network, instead of being transported to lysosomes. Like proCat B, this form was never traceable intracellularly or in conditioned media of intermediate-invasive SKmel28 cells. Because occurrence of extracellular mature Cat B was not observed in cells cultured on monomeric collagen, the cell contact with native collagen I fibrils was essential for triggering this process. Increased expression of MMP-1 and -13 in fibroblasts was also found to be dependent on integrin binding to fibrillar collagen type I.53, 54 Both, α1β1 and α2β1 integrins are able to bind native collagen type I.83, 84 Recently, Koblinski et al.45 could show that stimulation of proCat B secretion from breast fibroblasts can be induced through redistribution of activated β1 integrin subunits upon contact with collagen type I. By perturbation studies, with antibodies raised against the single chains of the α2β1 integrin receptor, we could observe inhibition of Cat B release only upon β1 integrin subunit ligation/inhibition. In addition, collagen gel contraction of MV3 cells was abolished. From our data, we can conclude that collagen type I binding, through a β1-containing integrin, leads to processing of secreted proCat B and, in addition, lysosomal exocytosis of mature Cat B. Inhibition of the β1 integrin chain leads to the inhibition of various integrin heterodimers besides to the collagen I binding integrins α1β1 and α2β1. Recently, other collagen binding integrins have been described, α10β1 and α11β1;85, 86 however, it appears that α10β1 is exclusively expressed in collagen II expressing tissues, and the α11β1 expression is limited to mesenchymal cells.87 However, up to date no data are available about the expression of these integrin receptors in melanoma cells.

The activity of lysosomal cysteine proteases is regulated in a number of ways, the most important being zymogen activation and inhibition by endogenous inhibitors.88 Sloane et al.89 proposed that an imbalance between Cats and cysteine protease inhibitors, associated with metastatic tumor cell phenotype, might facilitate tumor cell invasion and metastasis. In all melanoma cell lines tested, cystatin A mRNA was not detectable. No differences in mRNA transcript levels of cystatin B and C were observed comparing high- and low-invasive cells cultured on plastic or in 3D collagen I lattices. Interestingly, both inhibitors seemed to be regulated on the posttranscriptional level, since marked differences in inhibitor protein expression were observed and not depicted by the analysis of transcript levels. Cystatin B expression correlated with the invasiveness of the analyzed cell lines, being the lowest in MV3 and the highest in low-invasive cells. Differences in the levels of cystatin B protein might result from repression of translation or incomplete mRNA processing even though it is not clear to us by which mechanism. Thus, suggesting that in addition to an increased level of intra- and extracellular Cat B, the decreased synthesis of cystatin B further contributes to the invasiveness of high-invasive melanoma cells. Cystatin B protein is synthesized without signal peptide and localized mainly intracellularly, where it may protect cells from uncontrolled action of endogenous cysteine proteases.90 In contrast, cystatin C is synthesized containing a signal peptide, which is responsible for extracellular targeting of this inhibitor.39 Interestingly, Kos et al.70 found significantly elevated levels of Cat B and cystatin C in sera of metastatic melanoma patients. Our results clearly show that cystatin C protein expression is induced by cell–collagen I interactions, leading to the detachment of hypothetically mRNA bound repressor proteins and resulting in similar amounts of extracellularly located protein in high- and low-invasive melanoma cells. Nevertheless, highest levels of extracellular Cat B are only found in high-invasive cells, thus indicating an imbalance between the protease and cystatin C in these cells. Furthermore, the reduced cystatin B expression of high-invasive cells might point out the necessity of both types of inhibitors to effectively impair extra- and intracellular Cat B activity.

In conclusion, increased intracellular levels of Cat B, as well as constitutive secretion of proCat B and low cystatin B expression correlate with the high-invasive phenotype of melanoma cells. In addition, only by contact to fibrillar collagen type I the release of mature Cat B is induced in high-invasive melanoma cells, but not in low-invasive ones. The results presented herein underline the importance of tumor cell-matrix interactions in the regulation of proteolytic activities leading to the progression of malignant melanoma.

Acknowledgements

This work was supported by the Köln Fortune Programm (133/2003 to N. H. and 10/2004 to P. Z.) of the Faculty of Medicine, University of Cologne and the Center for Molecular Medicine (CMMC; BMFT/IDZ 10, Grant 01 GB 950/04) to C. M.

Ancillary