Invasion and metastasis of human tumors are multistep processes requiring both cell-cell and cell-matrix interactions within the host tissue. The result of such interactions is the production, release and activation of a variety of cytokines and growth factors, thus generating signals to directly or indirectly promote tumor growth and survival.
One important feature of tumor cells is their ability to invade the extracellular matrix of the host tissue utilizing a variety of different proteases. Increased expression and activity of matrix metalloproteinases (MMPs) have been shown in many tumor tissues to facilitate the breakdown and invasion of the extracellular matrix, as well as to release active growth factors and cytokines.1 In addition, several other proteases, including members of the cysteine, aspartate and serine protease families, have been shown to be involved in degradation of extracellular matrix components. However, the involvement of ADAM (A Disintegrin And Metalloproteinase) proteases in this process has been only reported in a few studies.2, 3, 4, 5, 6
ADAMs are a family of multidomain glycoproteins highly homologous to the class P-III snake venom metalloproteinase-disintegrins.7 All ADAM family members consist of a regulatory prodomain and a metalloproteinase, a disintegrin-like domain and a cysteine-rich domain. Furthermore, ADAMs are characterized by an epidermal growth factor (EGF)-like domain, a transmembrane domain and a short cytoplasmic domain. Half of the ADAM proteins are predicted to be active metalloproteinases, although for most of them the identification of specific substrates is still lacking. In the case of Fertilin α (ADAM-1) as well as for ADAM-9,8 ADAM-10,4 ADAM-12,9 ADAM-1710, 11 and ADAM-28,12 proteolytic activity toward matrix macromolecules has been demonstrated in vitro.
ADAMs have been shown to be involved in the release of extracellular domains of transmembrane proteins, such as interleukin (IL)-6 receptors, FAS-Ligand, TGF-α, TNF-α, HB-EGF, and L-selectin, a process known as ectodomain shedding.13 The release of soluble variants of these proteins may result in autocrine and distal paracrine effects, which do not occur in the cell-surface-bound forms.
Apart from their proteolytic activity, ADAMs have been reported to be involved in cell adhesion through their disintegrin domains. ADAMs can interact with integrins on adjacent cells through their RGD sequence, as observed for ADAM-15, which interacts with ανβ3 and α5β1 integrins on hematopoietic cells.14 ADAM-9 was found to adhere to α6β1 integrin and to enhance cell migration on the basement membrane component laminin-5.15 Furthermore, ADAM-9 was also shown to interact with ανβ5 integrin through their disintegrin sequence ECD.15, 16, 17
The in vivo functions of most ADAM family members are not well understood. ADAM-9 is widely expressed and highly conserved between mouse and human.18 It exists in 2 forms, secreted and cell-surface-bound, generated by alternative splicing.19
Apart from the integrin-binding ability, ADAM-9 has also been shown to process amyloid precursor protein and fibronectin in vitro.5, 20 However, ADAM-9 null mice failed to show an obvious phenotype during development and adulthood.20
Recently, the expression of several ADAM family members was reported in cell lines derived from different human tumors.21, 22 Overexpression of ADAM-9, ADAM-10 and ADAM-12 was shown in vivo and in vitro in several human carcinomas,23, 24, 25 suggesting that these ADAMs may play a role in cancer progression.
In the present study we have examined the expression and regulation of ADAM-9 in human melanoma tissue in vivo and in cultured human melanoma cell lines. ADAM-9 was detected in melanoma cells in vivo and in vitro and in peritumoral fibroblasts. We also extended our study to the analysis of ADAM-9 mRNA expression and modulation by matrix components, as well as cytokines and growth factors in melanoma cell lines with different metastatic potential.
Material and methods
Melanoma cell lines and cell culture
Four classes of melanoma cell lines were used in this study. The cell lines MV3, BLM and MeWo have been demonstrated to be highly metastatic with early and frequent formation of metastasis after subcutaneous inoculation in nude mice. SK mel 23, SK mel 28 and WM 115 are low invasive melanoma cells.26, 27, 28 The melanoma cell lines IF6, 530 and WM 164 also have only low invasive potential and show no organ metastasis unless injected intravenously in mice.27, 29, 30 In addition, the melanoma cell lines WM 75, WM 793, WM 1366, 1205 Lu and VMM5, which have not been characterized in vivo, were included. All cell lines were cultured in RPMI-1640 supplemented with 10% FCS, 2 mM glutamine and 100 U/ml of each penicillin and streptomycin, and were passaged by trypsinization.
Growth and proliferation experiments
For growth inhibition, MV3 melanoma cells were cultured as monolayers on plastic until subconfluency. The cells were washed twice with PBS and starved in serum-free medium for 24 hr. Afterward, the melanoma cells were grown in medium supplemented with 10% FCS for different time periods, as indicated in Figure 3. Alternatively, growth inhibition was induced by treatment of the cells with increasing concentrations of the transcription inhibitor Actinomycin D (Sigma, Taufkirchen, Germany) for 24 hr, followed by RNA isolation.
Growth on/in extracellular matrix proteins
Three-dimensional collagen type I gel cultures were prepared as described previously.31 Briefly, cells were seeded into collagen gels (3.5 × 105 cells/ml) containing 1 mg/ml porcine skin type I collagen (DGF Stoess AG, Eberbach, Germany). Attached or free-floating collagen lattices were prepared in tissue culture or bacterial dishes, respectively. In addition, dishes were coated with 20 μg/ml fibronectin (Roche, Mannheim, Germany), monomeric collagen type I (DGF Stoess AG), laminin-1 (Sigma), gelatin (Sigma), collagen type IV (Sigma) or matrigel (BD Pharmingen, Heidelberg, Germany) at 4°C overnight before seeding the cells (1.2 × 105 cells/cm2) for 48 hr.
Growth factor and cytokine stimulation
MV3 melanoma cells were plated at subconfluent density in medium containing 10% FCS for 24 hr. After washing twice with PBS, the medium was replaced by medium containing 1% FCS and various growth factors and cytokines: TNF-α (20 ng/ml; Roche), EGF (30 ng/ml; Peprotech), bFGF (10 ng/ml; Peprotech, Frankfurt, Germany), TGF-β (5 ng/ml; R&D), PDGF-BB (10 ng/ml; Peprotech), IL-6 (100 U/ml; Roche), IL-1α (10 U/ml; R&D), and Activin A (5 ng/ml; R&D Systems, Wiesbaden, Germany). Cells were cultured for an additional 24 hr before RNA preparation.
Northern blot analysis
Total RNA was isolated from melanoma cells cultivated as described by direct lysis into guanidine thiocyanate followed by phenol-chloroform extraction.31, 32 A total of 10 μg of total RNA was resolved by formaldehyde-agarose gel electrophoresis, blotted onto Hybond-N+ membranes (Amersham Biosciences, Braunschweig, Germany). Filters were hybridized according to published protocols with random-primed 32P-labeled cDNA probes for ADAM-9 (EST 186431; GenBank Acc: AA314599) and histone-3.33 Membranes were stripped for 5 min at 95°C in 0.1% SDS, 0.1% SSC. As a control for equal loading and integrity of the RNA the filters were hybridized with a 32P-labeled 18S rRNA oligonucleotide.34 All experiments were quantified by determining the signal intensities using ImageQuant software supplied to “Personal Densitometer” from Molecular Dynamics (Sunnyvale, CA) and normalized to the signal intensities of the 18S loading control.
Cryosections (8 μm) of human melanoma specimens from 20 patients were fixed in acetone at 4°C for 10 min, followed by blocking of nonspecific binding sites with 20% FCS in PBS (pH 7.4). Primary antibodies were dissolved in PBS containing 10% FCS and applied onto sections overnight at 4°C. After repetitive washing, bound antibodies were detected with an alkaline phosphatase (AP)-labeled anti-mouse/anti-rabbit polymer (Dako Envision®; Dako, Hamburg, Germany) and neofuchsin as substrate, as described previously.35 The following antibodies were used: polyclonal rabbit anti-ADAM-9 raised against the disintegrin domain of the protein (1 μg/ml; Chemicon, Hofheim, Germany) and anti-S100 rabbit polyclonal antibody (ready to use solution; Dako). Negative controls were performed by omission of primary antibodies.
Protein extracts were prepared by lysis of cells in RIPA buffer (150 mM NaCl; 1% NP-40; 0,5% sodium deoxycholate; 0,1% SDS; 50 mM Tris pH 8.0) on ice for 1 hr. Lysates were centrifuged at 15,000g for 20 min at 4°C. Protein concentration was determined using a commercial assay (Bio-Rad, Munich, Germany). Then, 20 μg of the protein samples were resolved on a 10% SDS-polyacrylamide gel and transferred to nitrocellulose membranes Hybond-C Super™ (Pharmacia-Biotech). After blocking of nonspecific binding sites with 5% skimmed milk in PBS containing 0.5% Tween (v/v), the blots were incubated with the primary antibody overnight at 4°C. The goat anti-human ADAM-9 antibody was used at a dilution of 1:500 (R&D). The mouse anti-human actin was used at a final dilution of 1:5,000 (MPXXBiomedicals, Irvine, CA). Bound primary antibodies were detected using a horseradish peroxidase-conjugated secondary antibody (1:2,000; Dako) and visualized with the ECL system (ECL™; Pharmacia Biotech, Freiburg, Germany).
ADAM-9 expression in melanoma cells in vivo and in vitro
ADAM-9 protein expression was analyzed by immunohistochemistry (Fig. 1) in nevocytic nevi (Fig. 1a and b) was compared to the expression in primary human melanoma (Fig. 1c–f, i and k) and metastatic melanoma (Fig. 1g and h). No staining was detectable in the nevus cell nests composed mainly of aggregates of normal melanocytes (Fig. 1b). In contrast, in a primary nodular melanoma (Fig. 1d), ADAM-9 protein is strongly expressed at the periphery of the tumor, in melanoma cells that stained positive for the melanoma marker S100 (Fig. 1f). In addition, in areas of tumor-stroma interactions (Fig. 1k), ADAM-9 protein was detected both in the tumor cells (Fig. 1i) as well as in stromal cells close to the tumor cell nests. In metastatic melanoma only very faint and diffuse staining was observed (Fig. 1h).
A large collection of human melanoma cell lines, characterized by different invasive capacities in nude mice was examined for ADAM-9 expression (Fig. 2). Northern blot analysis revealed a considerable variation of ADAM-9 expression levels. Quantification of the signals showed that ADAM-9 mRNA levels did not correlate with the invasive capacities of the different cell lines. For example, ADAM-9 transcript levels were comparable in the high invasive cell line VMM5 and the low invasive cell line 530.
RT-PCR analysis performed according to Hotoda et al.19 failed to show the splice variant of ADAM-9, suggesting that the detected transcript encodes the full-length form (data not shown).
Regulation of ADAM-9 expression by proliferation and cell-cell contacts
The ADAM-9 expression pattern observed in vivo in melanoma tissue sections and in vitro in melanoma cell lines indicates that ADAM-9 expression could be regulated by processes involved in cell growth and/or cell activation.
MV3 melanoma cells were synchronized by serum starvation for 24 hr, and cell growth was stimulated by addition of 10 % FCS (Fig. 3a). Under these conditions, ADAM-9 mRNA expression was maximally induced at 6 hr of serum stimulation, whereas histone-3 mRNA levels, which are a measure of cell proliferation,33 could not be detected before 24 hr of cultivation in serum-containing medium. In the same culture conditions we have analyzed ADAM-9 protein expression in total lysates by Western blot, as shown in Figure 3b. ADAM-9 protein was always detected as 2 immunoreactive bands migrating at ∼110 and 80 kDa, corresponding to inactive and active forms of the enzyme, respectively.18 Both protein forms were detectable under all treatments with the highest level of pro- and mature ADAM9 after 6 hr of FCS stimulation and a reduction to control levels after 48 hr.
Cell growth was blocked by different concentrations of the transcription/proliferation inhibitor Actinomycin D.36 ADAM-9 and histone-3 mRNA levels were analyzed after 24 hr of incubation with different concentrations of Actinomycin D by Northern blot (Fig. 3c). Histone-3 transcript levels were dose-dependently reduced and completely abolished at the highest inhibitor concentration of 1 μg/ml Actinomycin D. In contrast, ADAM-9 mRNA expression was only slightly reduced by increasing inhibitor concentrations.
In order to investigate whether cell-cell contacts have an impact on ADAM-9 synthesis, MV3 melanoma cells were plated at high and low cell densities. RNA analysis of these cultures did not reveal changes in ADAM-9 mRNA levels, suggesting that homotypic cell-cell contacts are not involved in regulation of ADAM-9 expression (data not shown).
Regulation of ADAM-9 expression by cell-matrix interactions
Degradation of extracellular matrix components has been shown to play a key role in tumor invasion and metastasis. Increased proteolysis of matrix is achieved by different types of proteases, of which many have been reported to be induced and activated by cell-matrix interactions in melanoma cells.37, 38, 39, 40 Therefore, we investigated whether interaction of melanoma cells with various extracellular matrix components could modulate ADAM-9 expression. As shown in Figure 4a, none of the matrix proteins used as planar substrates significantly altered ADAM-9 transcript levels. In contrast, contact of MV3 cells with a 3-dimensional environment composed of fibrillar collagen type I, either under tension (bound lattice, bl) or in free retracting lattices (fl) resulted in a marked reduction of ADAM-9 mRNA levels (Fig. 4b). This reduction was comparable in fully-contracted and noncontracted lattices but absent on planar collagen substrates, underlining the importance of a 3-dimensional collagenous environment in the regulation of ADAM-9 expression.
To further explore whether this downregulation is due to the specific interaction with fibrillar collagen, melanoma cells were cultured within alginate gels, providing a 3-dimensional environment void of specific cell-matrix interactions. In the absence of an anchoring matrix, in these gels the melanoma cells appeared as rounded single cells (data not shown). ADAM-9 transcript levels were unchanged in alginate gel cultures when compared with the levels observed under monolayer conditions (Fig 4b; lane 4). Thus cell contact with fibrillar collagen type I appear to specifically downregulate ADAM-9 mRNA expression.
Regulation of ADAM-9 in melanoma cells by cytokines and growth factors
To examine the regulation of ADAM-9 by inflammatory cell-derived factors, human melanoma cells in culture were treated with different cytokines and growth factors, which play a role in cutaneous wound repair and tumor-stroma interaction (Fig. 5). For this purpose, cells were plated for 24 hr in medium containing 10% FCS, which was then replaced by medium containing 1% FCS and the various stimuli for additional 24 hr. Treatment of cells with IL-1α for 24 hr reduced the expression of ADAM-9 mRNA 2.7-fold as detected by Northern blot hybridization. Densitometric analysis and normalization to 18S rRNA levels of 3 independent experiments displayed a significant decrease of ADAM-9 transcripts only when melanoma cells were treated with IL-1α but not with the other cytokines and growth factors.
Key characteristics of human cancer are unrestrained proliferation and altered organization influenced by the tissue microenvironment.41 Multiple signals from the surrounding environment, such as growth factors, cytokines, and homotypic and heterotypic cell-cell and cell-matrix interactions, have been shown to induce and activate a variety of proteases in tumor cells and fibroblasts,37, 42, 43, 44 which are thought to be essential for local degradation of connective tissue during tumor invasion and metastasis.40 Degradation of extracellular matrix components is accomplished through the combined action of different proteases including members of the serine, cysteine, aspartate and MMP families. Beside these proteolytic enzymes, the family of membrane-anchored ADAM proteins is especially well suited to fulfil many functions facilitating tumor invasion. ADAMs can function as integrin ligands as well as metalloproteinases that degrade extracellular matrix components and shed ectodomains of proteins. Such sheddase activity could result in the release of soluble factors contributing to tumor progression.
Apart from preliminary studies on cell lines and small numbers of human tumors, knowledge regarding the role of ADAMs in malignancy is limited.21, 22, 23, 45, 46 Our studies provided the first analysis of ADAM-9 expression in skin melanoma and in melanoma cell lines. We demonstrated that in melanoma tissue, ADAM-9 protein is strongly expressed in tumor cells invading the surrounding dermis, as well as in stromal cells adjacent to the tumor front. In contrast, in nevi, ADAM-9 expression was absent in both nevus cells and in the stromal cells close to nevus cell nests. ADAM-9 mRNA expression could not be detected in normal skin,23 although ADAM-9 protein seemed to be expressed at low levels in basal keratinocytes.6 Thus our observations suggest that induction of ADAM-9 expression in melanoma and stromal cells may be directly controlled by cellular changes associated with tumor progression.
To gain insight into the potential role of ADAM-9 in melanoma cell invasion, we also analyzed ADAM-9 expression in a large collection of human melanoma cell lines with different invasive and metastatic capacities. All melanoma cell lines showed ADAM-9 mRNA expression in varying amounts. However, comparison of the RNA levels with the invasiveness of the cell lines did not indicate any significant correlation.
A hallmark of tumor cells and tumor cell lines is enhanced and unrestrained growth. Interestingly, Nelson et al.47 showed an association of mitotic arrest deficient 2 β (MAD2β) with the cytoplasmic domain of ADAM-9. MAD2β is structurally related to MAD2, which is a component of the mitotic checkpoint mechanism, thus indicating a link between ADAM-9 function and cell cycle progression. To determine whether ADAM-9 mRNA expression is influenced by processes involved in tumor cell proliferation, we analyzed ADAM-9 mRNA levels under different growth conditions in MV3 melanoma cells. Induction of proliferation by serum stimulation of quiescent melanoma cells resulted in an increase of ADAM-9 mRNA after only 6 hr. At this early time point no changes in histone-3 mRNA levels, which is a measure of cell proliferation, were detected, thereby indicating that ADAM-9 expression is upregulated prior to initiation of proliferation. To corroborate this, MV3 melanoma cells were growth-arrested by the transcription inhibitor Actinomycin D.36 ADAM-9 mRNA levels were not significantly altered by this treatment, whereas histone-3 mRNA levels showed a steady decrease, being completely abolished at the highest inhibitor concentration used. These data suggest that ADAM-9 expression in melanoma cells is not directly linked to cell cycle progression. From our data we cannot exclude that differences in mRNA stability for the 2 analyzed transcripts might account for the observed differences. However, to date no data are available about ADAM-9 mRNA stability.
Several lines of investigation demonstrate alterations of the expression of various cell-cell and cell-matrix receptor proteins during the transformation of benign melanocytes to melanoma cells.48 From our results we can conclude that homotypic cell-cell interactions are not involved in ADAM-9 regulation. This does not exclude that heterotypic cell-cell interactions, such as interaction between tumor cells and stromal fibroblasts, inflammatory cells or endothelial cells may play an important role in ADAM-9 expression. Interactions between tumor cells and the surrounding matrix have been reported to provide tumor cells with the ability to invade tissues and overcome microenvironmental control. On tumor cells various integrin molecules, such as α2β1, α5β1 and αvβ3, which mediate binding to dermal connective tissue constituents, were found to be strongly induced during melanoma progression.49, 50 To test whether ADAM-9 expression is regulated by cell-matrix interactions, as suggested by the in vivo observations, MV3 cells were cultured on different matrix components. When MV3 melanoma cells were cultured on plates coated with matrigel (a basement membrane like composite), fibronectin, gelatin and collagen type I monomers, no or only little alteration of ADAM-9 mRNA levels was observed. However, when the cells were cultured inside 3-dimensional fibrillar collagen type I lattices either under mechanical tension, i.e. “attached”, or devoid of tension, i.e. “relaxed”, a strong downregulation of ADAM-9 transcript levels was detected. This suggests that the 3-dimensional environment, but not the forces generated within the gels, are responsible for the observed changes. In addition, culture of melanoma cells within 3-dimensional gels composed of alginate did not reveal alterations of ADAM-9 transcripts levels, thereby identifying fibrillar collagen type I being associated with ADAM-9 downregulation.
It is well accepted that tumor cells can modify the surrounding connective tissue and alter the metabolism of resident fibroblasts such that they synthesize a stroma conducive for tumor cell migration. These interactions are mediated in part by cytokines and growth factors released by tumor cells, fibroblasts or/and inflammatory cells.51 The strong expression of ADAM-9 in the invading front of the melanoma and in stromal cells in close vicinity to the tumor suggests that tumor-stroma interactions are contributing to ADAM-9 induction in vivo. We therefore tested different growth factors and cytokines, which are known to be involved in melanoma progression, for their influence on ADAM-9 expression.52, 53 In our study IL-1α significantly reduced ADAM-9 mRNA expression in MV3 melanoma cells. The latter finding corroborated a previous report of ADAM-9 downregulation by IL-1α in chondrocytes.54 Interestingly, these cytokines have been shown to induce expression of other proteases such as the MMPs.55
In summary, we have shown that ADAM-9 is upregulated in melanoma in vivo at the border of the tumor cell nests, where contacts with the surrounding stroma occur, thus implicating a role for ADAM-9 in melanoma progression. In vitro fibrillar collagen type I was identified as a negative regulator for ADAM-9 expression, indicating that melanoma-type I collagen interactions might not be involved in ADAM-9 induction observed in vivo. Additionally, the restricted localization of ADAM-9 protein at the tumor front might indicate a link between ADAM-9 expression and cell-cell interactions involved in locomotion of rapidly moving cells. This was suggested by Nath et al.15 showing that cell-cell interactions between integrin receptors and ADAM-9 resulted in a marked induction of fibroblast motility. Further studies will be needed to elucidate the role of ADAM-9 in migratory processes during tumor invasion and metastasis and to dissect the role of the metalloproteinase and the disintegrin domains for melanoma progression.
This work was supported by Wilhelm Sander-Stiftung (1999.093.2; to C.M. and R.N.) and by the Koeln Fortune Program/Faculty of Medicine, University of Cologne (Nr. 10/2004, to P.Z.). J.W.F. was supported by the University of Virginia Cancer Center, Polifarma, Ltd. and NIH award U54 GM64346.