The invasive potential of human melanoma cell lines correlates with their ability to alter fibroblast gene expression in vitro and the stromal microenvironment in vivo



The tumor microenvironment is thought to play an important role in invasion and metastasis. Previously, we have shown that signaling from melanoma cells can alter the gene expression profiles of fibroblasts in vitro and in vivo. To investigate whether the capacity to signal fibroblasts and alter host gene expression profiles is correlated to the invasive potential of specific human melanoma cell lines, we assayed changes in gene expression of fibroblasts when cocultured with the human melanoma cell lines BLM, MV3, A2058, SK-mel28 and WM164. Results indicated that the gene expression of key chemokines and cytokines, such as IL-1B, IL-8, IL-6 and CCL2/MCP1, was significantly upregulated in fibroblasts cocultured with the invasive melanoma lines BLM and MV3 compared to fibroblasts cocultured with noninvasive WM164 cells. The results were verified by quantitative RT-PCR as well as by protein assay and supported by immunohistochemistry of human invasive melanoma. Furthermore, a role for fibroblast-secreted IL-1B in the invasion of melanoma was demonstrated in vitro, where siRNA silencing of IL-1B in melanoma-stimulated fibroblasts resulted in a diminution of melanoma invasion. Although CCL2/MCP1, a chemoattractant for macrophages, was shown to be upregulated in fibroblasts cocultured with metastatic melanoma cell lines, immunohistochemical analysis of human melanoma also indicated CCL2/MCP1 production associated with the melanoma. In summary, these experiments indicate that the invasiveness of melanoma can partly be correlated to its ability to stimulate host stromal fibroblasts to give rise to the secretion of chemokines that generate a microenvironment that is conductive for melanoma invasion and metastasis. © 2009 UICC

The capacity for invasion and metastasis is one of the hallmarks of tumor malignancy. The role of host–tumor communication and the tumor microenvironment in tumor cell invasion and metastasis has been recognized as being critical in this process.1 The complex interaction between invasive tumor cells, stromal cells, lymphocytes and extracellular matrix as represented by the invasive tumor microenvironment is only beginning to be understood. Further investigation in tumor microenvironments will generate new possibilities in terms of identifying key mechanisms of tumor cell invasion and metastasis and potentially therapeutic nodes for attenuating primary tumor metastasis.2, 3

Fibroblasts comprise one of the principal cell types of the stromalcompartment, and recent studies have demonstrated that stromal fibroblasts are important in tumor cell migration through the stroma.4, 5 Stimulation of resident stromal fibroblasts by the presence of invasive tumor cells gives rise to alterations in the gene and protein expression patterns of these cells, which have been interpreted as proinvasive and prometastatic.6 Upregulation of key chemokine and cytokine expression in stromal fibroblasts in the presence of migrating human melanoma appears to promote conditions for infiltration of macrophages and neutrophils, which may support the metastatic process as well as provide conditions favorable for tumor cell chemotaxis and invasion.1, 7, 8

A variety of melanoma cell lines have been characterized as having different potentials for invasion and metastasis in animal models of melanoma.9–14 The molecular characteristics of some of these lines have been investigated with the aim of shedding light of critical components in the metastatic process. Using gene expression analysis, de Wit et al. (2005) found a subset of genes that were differentially expressed in metastatic and nonmetastatic melanoma cell lines. Similarly, differences in the gene expression patterns of a metastatic melanoma cell line NMCL-1 and a nonmetastatic cell line 530 were observed using DNA microarrays.15 This approach has recently been extended by investigating differential gene expression in nevocellular nevus and melanoma metastasis lesions from human melanoma patients with results indicating specific differences in gene expression patterns in these tissue samples.16 Taken together, there appears to be a growing body of evidence that suggests the gene expression patterns of nonmetastatic and metastatic melanoma are different; however, how this relates to stromal communication and its outcome is less well understood.

In this investigation, we have extended this view by exploring whether the invasive/metastatic potential of melanoma cell lines is correlated with their ability to communicate with fibroblasts and subsequently alter their gene and protein expression patterns in a proinvasive, prometastatic manner. Using a limited number of cell lines, we observed a correlation of the melanoma cell lines' invasive and metastatic potential with their ability to alter fibroblast gene and protein expression patterns. Many of the altered genes and proteins are documented to be associated with tumor invasion and metastases. This study underscores the role of host–tumor communication in the microenvironment and identifies some of components potentially involved in tumor cell invasion and metastasis.

Material and methods

Cell lines and culture conditions

A human foreskin-derived fibroblast cell line, HS 68, was used to study the mechanism of melanoma invasion. The HS 68 fibroblast cell line and the human melanoma cell lines A2058, WM164 and SK-Mel-28 were purchased from ATCC (Manassas, VA). The BLM and MV3 melanoma cell lines were a gift from the Department of Dermatology, University of Cologne, Germany. Fibroblasts cells were maintained in DMEM (Invitrogen, Carlsbad, CA), containing 10% fetal bovine serum (FBS) and supplemented with 4 mM L-glutamine, 1.5 g/l sodium pyruvate, 4.5 g/l glucose, essential and nonessential amino acids; all melanoma lines were maintained in DMEM supplemented with 10% FBS. Cells were incubated at 37°C with 5% CO2.

Fibroblast–melanoma coculture

To model and assess the communication between melanomas and fibroblast cell cultures, we used a coculture setting performed in transwell chamber. Fibroblasts were plated at a density of 3 × 104 cells per lower well in complete media in 24-well plate. After 24 hr of incubation, the media in the wells were changed with serum-free DMEM. Inserts (Becton Dickinson Labware, Bedford, MA) coated with rehydrated matrigel (200 μg/filter, 25-mm, Collaborative Biomedical Products, Bedford, MA) were placed in each well, and 3 × 104 melanoma cells in 0.2 ml serum-free DMEM were added in each insert. To prevent direct tumor cell migration, we used a filter with 1-μm pore size, thus allowing only soluble factors to pass through the filter. After 24 hr of incubation, fibroblasts in the bottom compartment and melanoma cells in upper compartment were harvested and used for gene expression profiling.

Gene expression analysis

To identify differentially expressed genes in the human melanoma cell lines BLM, MV3, A2058, SK-mel28 and WM164 and human HS68 fibroblasts as a result of coculture, gene expression profiling was performed using the Affymetrix Human Genome U133 plus 2 GeneChips containing ∼54,600 full-length genes and ESTs. Two GeneChips were used for each experimental condition with experimental replicates. The procedures including total RNA isolation, cDNA synthesis, cRNA labeling, microarray hybridization and image acquisition and analysis were performed as previously described.1

To examine the effect of coculture of the melanoma cells and fibroblasts on their gene expression, gene expression profiling was performed on both the cocultured melanoma cell lines and fibroblasts using the conditions described in the previous section and in the section describing the melanoma-fibroblast coculture conditions.

Gene expression data analysis

After microarray images were obtained, dChip software17 was used to analyze the profiling data. Briefly, the image file (.CEL file) generated by Affymetrix Microarray Suite 5.0. was converted into DCP files using dChip as described by Li and Wong.18 The DCP files were normalized and raw gene expression data were generated using model-based analysis. To filter out the resulting gene list, following criteria were set for the fibroblasts GeneChips: 0.55< SD/mean <1,000 (SD/mean is coefficient of variation, a parameter representative of the variation across samples), gene present call in 20% samples, and expression level >50 in 30% of samples. For melanoma cell lines, genes which met the arbitrary criteria were selected: 0.90< SD/mean <1,000: gene present call in 20% samples, and expression level >100 in 40% of samples. Hierarchical clustering was performed using the filtered gene lists. Functional grouping was performed using expression analysis systematic explorer.19 To identify the genes overexpressed in fibroblasts following coculture with highly metastatic melanoma BLM and MV3, compared to the nonmetastatic melanoma WM164 cell line, we used the following criteria: fold change ≥2 in both fibroblasts cocultured with BLM and MV3 and fold change <0.5 in fibroblasts cocultured with WM164.

Validation of genechip expression data

To confirm expression data for selected genes identified in the chip analyses, real-time quantitative RT-PCR was carried out using the TaqMan® PCR reagent kit and ABI PRISM™ 7900 Sequence Detector as previously described1 for CXCL1, CXCL2, IL-8, CCL2/MCP1 and IL-1B. All PCRs reactions were carried out in triplicates using equal amounts of each cDNA with sample equivalent to 50 ng of starting total RNA. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was used as internal control. Primer sequences: (forward) CATGCCAGCCACTGTGATAGA, (reverse) ATTCCCCTGCCTTCACAATG for CXCL1; (forward) GGTTTGCAGATATTCTCTAGTCATTTGT, (reverse) GGATTCCTCAGCCTCTATCACAGT for CXCL2; (forward) TGTTGA ATTACGGAATAATGAGTTAGAAC, (reverse) CAAGTTTCAACCAGCAAGAAATTACT for IL-8; (forward) AACCTGAACCTTCCAAAGATGG, (reverse) TCTGGCTTGTTCCTCACTACT for IL-6; (forward) CAA-GCA-GAA-GTG-GGT-TCA-GGA-T, (reverse) TTA-GCT-GCA-GAT-TCT-TGG-GTT-GT for CCL2/MCP1; (forward) CTCGCCAGTGAAATGATGGCT, (reverse) GTCGGAGATTCGTAGCTGGCT for IL-1B. (Forward) GAAGATGGTGATGGGATTTCCA, (reverse) GATTCCACCCATGGCAAATT for human GAPDH.

In vitro melanoma invasion assays

The effect of soluble factors released from fibroblasts as a result of coculture with melanoma cells on the invasive behavior of melanoma cells was measured by quantifying the number of cells penetrating through matrigel-coated transwell inserts (Becton Dickinson Labware, Bedford, MA). Bottom wells were seeded with 3 × 104 HS 68 fibroblasts in 24-well plate a day before the experiment. Polycarbonate filters (6.5-mm diameter, 8-μm pore size, Becton Dickinson Labware, Bedford, MA) precoated with a layer of matrigel were seeded with 3 × 104 of the melanoma cell lines A2058, BLM, MV3, SK-mel28 or WM-164 onto the filters and placed into the wells over the fibroblasts. After 24 hr of incubation, melanoma cells were removed from upper surface of the filter. The filters were then rinsed in PBS and stained with diff-quick kit (Baxter, FL). The cells that had migrated to the opposite side were counted microscopically by examining 3 randomly selected fields on each filter. For controls, invasion of melanoma cell lines was assessed using serum-free media in the bottom wells.

Preparation of melanoma-stimulated fibroblast conditioned media

Melanoma cell lines were grown to ∼70–80% confluence, and then the medium was replaced with serum-free DMEM and incubated for another 72 hr followed by conditioned media collection and centrifugation. Fibroblast cultures at 80% confluence were incubated for 4 hr in serum-free DMEM, followed by 5 hr of incubation in 10% of the conditioned melanoma media without serum. The melanoma conditioned media-stimulated fibroblast media were then collected, centrifuged, sterilized by filtration and stored at −80°C before use in the THP-1 cell migration assays. For the preparation of unstimulated fibroblast control media, the cells were incubated in serum-free DMEM for 9 hr, and then the conditioned media were collected, centrifuged and filtered.

Protein microarray analysis of cytokine concentrations

Cytokine and chemokine concentrations in: (i) conditioned media of fibroblasts stimulated with 10% melanoma-condition medium, (ii) conditioned media of melanoma cell lines and (iii) conditioned media of fibroblasts were measured by the protein array FASTQuant Human II Microspot Assay (Whatman Schleicher&Schuell, Boston, MA). This assay allows for the detection of the 10 chemokines/cytokines: IL-1B, IL-2, IL-6, GM-CSF, IL-4, CCL2/MCP1, RANTES, IL-8, IL-10 and IL 12/p70. Using standard curves, the minimal chemokines/cytokine concentration detected by this method was of 3pg/ml for IL-1B, IL-2, IL-6 and GM-CSF, 4pg/ml for IL-4, 5 pg/ml for CCL2/MCP1 and RANTES, 10 pg/ml for IL-8 and 30 pg/ml for IL-10 and IL-12p70.

Fibroblast gene silencing and in vitro invasion assays

To assess the role of IL-1B produced by melanoma-stimulated fibroblasts in melanoma invasion, small interfering RNA (siRNA) for IL-1B expression was obtained from Invitrogen (Carlsbad, CA). An irrelevant siRNA (Stealth RNAi Negative Control, Invitrogen) was used for transfection as a nonspecific control. Fibroblastswere plated at 50–60% confluence in complete culture medium in 24-well plates and incubated overnight. Cells were then transfected with siRNA (50 pmol/set/well) for 24 hr using Lipofectamine 2000 reagent (Invitrogen, CA) according to instructions from the manufacture. IL-1B gene expression levels after silencing were assayed by RT-PCR. The effect on BLM melanoma invasion cocultured with the IL-1B knockdown fibroblasts compared to control fibroblasts using the transwell invasion assay was performed as described earlier.


Cryosections of 8 μm thickness were fixed with cold acetone for 5 min and rinsed for 10 min in phosphate-buffered saline (PBS). Sections were blocked for 1 hr with 10% fetal calf serum (FCS) in PBS before applying the primary antibodies diluted in PBS–FCS for 16 hr at 4°C. After 3, 15 min washes, bound antibodies were detected with alkaline phosphatase-labeled anti-mouse/anti-rabbit polymer (DAKO Envision, Hamburg, Germany) and neofuchsin as substrate. Nuclei were counterstained with hematoxyline solution for 1 min (Shandon, Pittsburg, PA).

The following antibodies were used: monoclonal antibody directed against the melanoma marker HMB45 was purchased from DAKO (Hamburg, Germany); monoclonal antibody to human Interleukin 1(1:100) was purchased from BioSource (Laboserv, Giessen, Germany). Mouse anti-human CCL2/MCP-1 (1:50) and CD 68 (1:50) were from BD Bioscience (Heidelberg, Germany).


Gene expression profiles of melanoma cell lines

The gene expression profiling of the highly invasive and metastatic BLM and MV3 melanoma cell lines,9, 11, 12 the low/moderate invasive A2058 and SK-mel28 cell lines13, 20 and the nonmetastatic WM164 cell line21, 22 followed by hierarchal cluster analysis showed the highly invasive and metastatic cell lines BLM and MV3 cell lines cluster together. Whereas, the melanoma cells lines A2058, SK-mel28, which display moderate to low invasive phenotypes, and WM164, which is not invasive, clustered together (Fig. 1a). However, from this cluster, the nonmetastatic melanoma line WM164 and poorly metastatic melanoma line SK-mel28 gene expression patterns were the least similar to the other melanoma cell lines (Fig. 1a). Interestingly, the gene expression profiles of the various melanoma cell lines mirrored the invasive/metastatic phenotype described for those cell lines.9, 11–13, 20–22

Figure 1.

Gene expression profiles of melanoma cells with differing invasive/metastatic potentials. (a) Hierarchal cluster of the GeneChip expression data of melanoma cell lines BLM, MV3, A2058, SK-mel28 and WM164. A total of 1,558 genes were filtered out as differentially expressed among the cell lines based on the following criteria: CV > 90%, p call > 20%, expression level > 100 in >=40% sample. (b) Hierarchal cluster of GeneChip expression data of human fibroblasts cocultured with melanoma cell lines. A total of 1,183 genes whose expression level satisfied the following criteria were included: CV > 55%, p call > 20%, expression level > 50 in >=30% sample. The expression level of each gene is standardized to have mean 0 and standard deviation 1. Red color represents expression levels above the mean expression level and blue color represents an expression level lower than the mean.

Ontological characterization of gene expression profiles of the invasive melanoma cell lines BLM and MV3 compared to noninvasive WM164 line

To gain insight into differences between invasive/metastatic melanoma and noninvasive melanoma, we examined the ontology of differentially expressed genes between these 2 categories of melanoma as represented by BLM and MV3 lines and the WM164 line. Genes whose expression increased (fold change ≥ 2) or decreased (fold change < 0.5) in both BLM and MV3 cell lines when compared to the same genes in WM164 cell line were selected and used for ontological characterization. The top 10 ontological categories are shown in Table I. Interestingly, the ontological categories of extracellular matrix, cell communication, collagen, cell adhesion were identified from the upregulated gene lists, whereas the ontological categories of pigmentation, melanin biosynthesis, tyrosine metabolism were identified from the list of downregulated genes of the high-invasive BLM and MV3 lines compared to the noninvasive WM164 cell line (Table II). Not surprisingly, these results suggest an important role for extracellular matrix, cell adhesion and migration in microenvironment in the invasive, metastatic process associated with melanoma.

Table I. Top Ontology Categories of 693 Genes Upregulated in BLM and MV3 Compared to WM164 (Fold Change ≥ 2)
GO groupGene categoryList hitsList totalPopulation hitsPopulation totalEASE score
Cellular componentExtracellular913301,30212,3823.40E-18
Cellular componentExtracellular matrix3533031812,3823.81E-12
Cellular componentExtracellular space4033041812,3826.62E-12
Cellular componentCollagen103303412,3821.73E-07
Biological processCell communication1363313,06912,6021.36E-11
Biological processDevelopment933311,85312,6022.83E-10
Biological processMorphogenesis613311,08112,6021.57E-08
Biological processOrganogenesis5533196312,6026.71E-08
Biological processCell adhesion3933159712,6023.48E-07
Molecular functionExtracellular matrix structural constituent153379712,9091.99E-07
Table II. Top Ontology Categories of 704 Genes Downregulated in BLM and MV3 Compared to WM164 (Fold Change < 2)
GO groupGene categoryList hitsList totalPopulation hitsPopulation totalEASE score
Biological processDevelopment863261,85312,6023.55E-08
Biological processPigmentation83262512,6022.29E-06
Biological processNeurogenesis3032644112,6023.66E-06
Biological processMelanin biosynthesis5326612,6026.25E-06
Biological processMelanin biosynthesis from tyrosine5326612,6026.25E-06
Biological processMelanin metabolism5326712,6021.43E-05
Biological processMorphogenesis523261,08112,6021.63E-05
Biological processPigment metabolism73262412,6022.55E-05
Biological processOrganogenesis4632696312,6026.89E-05
Biological processTyrosine metabolism53261112,6020.000124

Melanoma cell line gene expression profiles following coculture with fibroblasts

Analysis of the gene expression patterns of the 5 melanoma cell lines following coculture with fibroblasts showed little change compared to the lines cultured alone (data not shown) suggesting that fibroblasts had little effect on melanoma gene expression regardless of their invasive and metastatic potentials.

Fibroblast gene expression profiles cocultured with melanoma cell lines

Hierarchical cluster analysis (Fig. 1b) of the gene expression profiles of the melanoma-stimulated fibroblasts as a result of coculture with the melanoma cell lines showed that the global gene expression patterns of fibroblasts cocultured with high-invasive BLM and MV3 were more similar to one another and clustered distinctly from fibroblasts cocultured with less-invasive A2058, SK-mel28 and WM164. Overall, the clustering of the melanoma-treated fibroblast gene expression profiles was similar to that observed for the melanoma cell line profiles (Fig. 1a). Thus, the potential for melanoma cells to stimulate fibroblast gene expression during coculture melanoma cell line appears to reflect their invasive/metastatic potential and suggests that the alteration of gene expression in those fibroblasts may provide insight into the role of fibroblasts in the microenvironment to promote melanoma invasion and metastasis.

Ontological characterization of gene expression profiles of fibroblasts cocultured with melanoma cell lines

Genes whose expression increased (fold change ≥ 2) in both the fibroblasts cocultured with BLM or MV3 and decreased (fold change <0.5) in fibroblasts cocultured with WM164 cell line were selected and used for ontological characterization. A total of 510 genes met these criteria (Table III) For the downregulated gene list, genes with fold change <0.5 in fibroblasts cocultured with BLM and MV3 and fold change ≥ 2 in fibroblasts cocultured with WM164 were selected (total 79 genes). The 10 highest scoring gene ontology categories are shown in Table IV. From the categories seen in Table III, genes related to cytokine and chemokine activity and inflammatory response were upregulated in fibroblasts cocultured with BLM and MV3. Genes associated with phosphoric diester hydrolase activity and neurofilament were downregulated.

Table III. Top Ontology Categories of 510 Genes Significantly Increased in Fibroblasts Cocultured With BLM and MV3 (Fold Change ≥ 2), Which Are Not Changed or Decreased in Fibroblasts Cocultured With WM164 (Fold Change < 2)
GO groupGene categoryList hitsList totalPopulation hitsPopulation totalEASE score
Molecular functionChemokine activity103214712,9091.88E-06
Molecular functionChemokine receptor binding103214712,9091.88E-06
Molecular functionTranscription regulator activity563211,16912,9092.36E-06
Molecular functionChemoattractant activity103214912,9092.71E-06
Molecular functionG-protein-coupled receptor binding103214912,9092.71E-06
Molecular functionCytokine activity1832120712,9091.57E-05
Cellular componentExtracellular573081,30212,3822.6E-05
Biological processInflammatory response1631218012,6023.96E-05
Cellular componentExtracellular space2630841812,3824.37E-05
Biological processCellular process1933126,39912,6025.34E-05
Table IV. Top Ontology Categories of 79 Genes Significantly Decreased in Fibroblasts Cocultured With BLM and MV3 (Fold Change < 2), Which Are Increased in Fibroblasts Cocultured With WM164 (Fold Change ≥ 2)
GO groupGene categoryList hitsList totalPopulation hitsPopulation totalEASE score
Molecular functionPhosphoric diester hydrolase activity3477312,9090.03
Cellular componentNeurofilament245812,3820.03
Biological processCellular process29456,39912,6020.06
Biological processCentral nervous system development34512112,6020.07
Molecular function3′\,5′-cyclic-nucleotide phosphodiesterase activity2472512,9090.09
Molecular functionTransferase activity\, transferring groups other than amino-acyl groups34713812,9090.09
Molecular functionCyclic-nucleotide phosphodiesterase activity2472912,9090.10
Molecular functionTransferase activity\, transferring acyl groups34714912,9090.10
Biological processResponse to external stimulus9451,37012,6020.10
Biological processImmune response64572512,6020.11

In Table V, we show the nonredundant list of the genes that populate the ontology categories of chemokine activity, chemokine receptor binding, chemoattractant activity, cytokine activity, inflammatory response and their fold changes in the fibroblasts cocultured with the various melanoma cell lines. Examination of these data indicates that invasive/metastatic melanoma gives rise to characteristic alterations of fibroblast gene expression in a manner which suggests that stimulation of stromal fibroblasts by invasive/metastatic melanoma may play an integral role in the process.

Table V. Upregulated Cytokine, Chemokine and Related Genes in Fibroblasts Cocultured With BLM and MV3 Cell Lines Compared to WM164 Cell Line
Gene descriptionGene SymbolFb (BLM)Fb (MV3)Fb (A2058)Fb (SK-mel 28)Fb (WM164)
Chemokine (C-X-C motif) ligand 3CXCL3196.8519.
Chemokine (C-X-C motif) ligand 2CXCL2126.618.128.670.620.81
Chemokine (C-X-C motif) ligand 6CXCL674.526.557.360.80.87
Chemokine (C-X-C motif) ligand 1CXCL154.2222.8511.310.340.39
Interleukin 1, betaIL-1B50.7399.312.021.9
Chemokine (C-C motif) ligand 11CCL1147.8813.743.21.011.57
Interleukin 8IL-835.5912.3216.460.380.56
Chemokine (C-X-C motif) ligand 5CXCL521.383.032.4811.14
Complement component 3C319.759.922.790.91.25
Interleukin 6IL-618.773.061.520.220.2
Nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 4NFATC49.6612.081.251.581.88
Chemokine (C-C motif) ligand 13CCL139.063.623.141.751.34
Leukemia inhibitory factorLIF8.872.843.610.850.9
Interleukin-1 receptor-associated kinase 2IRAK27.342.545.251.281.39
Chemokine (C-X-C motif) ligand 14CXCL146.975.590.691.251.8
Molecule possessing ankyrin repeats induced by lipopolysaccharideMAIL6.523.353.040.490.37
Bradykinin receptor B1BDKRB15.652.553.141.351.17
Fms-related tyrosine kinase 3 ligandFLT3LG5.575.172.081.831.83
Chemokine (C-C motif) ligand 2CCL24.112.892.851.361.25
Vascular endothelial growth factor BVEGFB4.085.080.891.061.27
Transforming growth factor, beta 1TGFB13.794.631.11.141.09
Bone morphogenetic protein 1BMP13.642.941.151.251.13
Tyrosyl-tRNA synthetaseYARS2.162.120.780.70.74

Validation of select genes upregulated in fibroblasts cocultured with the BLM and MV3 melanoma cell lines

Before further investigation, we selected CXCL1, CXCL2, IL-8, IL-6, IL-1B and CCL2/MCP1 for expression validation quantitative RT-PCR in fibroblasts cocultured with either the invasive/metastatic melanoma BLM line or the noninvasive/metastatic WM 164 line based on their role in tumor metastasis and inflammation (Table VI). Compared to fibroblasts that were not in coculture, the mRNA of these genes was significantly increased in fibroblasts cocultured with BLM with fold changes ranging from 4 to 126 compared to only 0.4- to 2-fold changes observed in fibroblasts cocultured with WM164. The results essentially validated the expression results observed by GeneChip analysis.

Table VI. Quantitative RT-PCR Verification of Genechip Data
Gene SymbolFb(BLM)Fb(MV3)Fb(A2058)Fb(SK-mel 28)Fb(WM164)
CXCL2126.6042.36 ± 0.2718.123.15 ± 0.018.671.34 ± 0.040.620.69 ± 0.020.810.27 ± 0.03
CXCL154.2236.47 ± 0.5322.852.33 ± 0.1011.310.85 ± 0.010.340.16 ± 0.010.390.12 ± 0.01
IL-8139.05224.76 ± 1.6613.079.18 ± 0.0923.973.29 ± 0.020.920.97 ± ± 0.01
IL-618.7747.89 ± ± 0.011.521.54 ± ± ± 0.01
IL-1B38.1013.26 ± ± 0.018.992.02 ± 0.031.841.32 ± 0.011.960.94 ± 0.01
CCL2/MCP14.1115.12 ± 0.012.894.25 ± 0.012.852.64 ± 0.011.361.73 ± ± 0.01

Protein levels of various cytokines and chemokines in fibroblast media following stimulation with BLM or MV 3 conditioned media

The production of IL-8, IL-1B and CCL2/MCP1 from fibroblasts in response to stimulation by BLM or MV3 condition media was significantly greater than observed in response to WM164 condition media (Table VII). In the case of RANTES, however, the protein was seen to be present in the melanoma conditioned media as well. No significant difference of IL-4, IL-6, IL-2, GM-CSF and IL 12p70 concentrations was observed in the melanoma media-stimulated fibroblast media compared to controls.

Table VII. Protein Measurement in Conditioned Media of Fibroblasts Stimulated With Melanoma and Melanoma Media
 Fb (BLM)Fb (MV3)Fb (A2058)Fb (SK-mel 28)Fb (WM164)BLMMV3A2058SK-mel 28WM164Fb
  1. ND, not detected. Value represents pg/ml.


Localization of IL1-B, CCL2/MCP-1 expression and macrophage infiltration in human primary nodular melanoma

Analysis of IL-1B expression in human melanoma specimens showed expression of this chemokine primarily to be associated with stromal fibroblasts (Fig. 2), whereas CCL2/MCP-1 expression was observed in both melanoma cell nests and the surrounding stromal tracks (Fig. 3b). Staining for the presence of macrophages indicated that they were generally associated with the regions of the stromal tracks (Fig. 3c) perhaps indicative of a functional significance of the stroma supporting macrophage migration.

Figure 2.

Immunohistochemical staining of IL-1B in specimens of a human primary nodular melanoma. (a) Staining of melanoma with melanoma antigen HMB 45. Arrow head indicates melanoma nests; (b) staining of IL-1B in melanoma specimen. Arrow points at stromal branches expressing IL-1B. The bar represents 50 μm.

Figure 3.

Immunohistochemical staining of CCL2/MCP-1 and macrophages in specimens of a human primary nodular melanoma. (a) Staining of melanoma with melanoma antigen HMB 45. Arrow head indicates melanoma nests; Arrow indicates stromal branches around the melanoma. (b) Staining of CCL2/MCP-1 in melanoma specimen. (c): Staining of macrophage with CD 68 in melanoma specimen. Arrow heads indicate macrophage infiltrated in the stromal track close to the melanoma nests. The bar represents 100 μm.

Role of melanoma coculture with fibroblasts on invasion in vitro

As seen in Figure 4a, all of the melanoma cell lines used in this investigation were capable of invading matrigel with somewhat better invasion observed with the BLM, MV3 and A2058 lines. Upon coculture with fibroblasts, the invasive potential of all the lines was enhanced with the greatest number of invaded cells observed with the BLM and MV3 cell lines cocultured with fibroblast. siRNA silencing of IL-1B in fibroblasts was observed to decrease the transcripts of IL-1B as assayed by quantitative RT PCR, to ∼15% of the untreated fibroblasts (Fig. 4b). BLM invasion when cocultured with the IL-1B silenced fibroblasts decreased by ∼45% (Fig. 4c) suggesting fibroblast IL-1B plays an important, although perhaps not exclusive, role in promoting melanoma invasion.

Figure 4.

Melanoma-stimulated fibroblast promotion of melanoma invasion. (a) The in vitro invasive ability of melanoma cell lines BLM, MV3, A2058, SK-mel28 and WM164 cocultured with fibroblasts compared to conditioned fibroblast media. Fb represents melanoma invasion while in coculture with fibroblasts. SFM represents melanoma invasion in the presence of fibroblast conditioned media. (b) IL-1B levels in siRNA silenced fibroblasts cocultured with BLM compared to fibroblasts. Mock and nonspecific transfection denotes the use of transfection reagent or a nonspecific siRNA. (c) Decreased melanoma invasion in the presence of IL-1B silenced fibroblasts.


Metastasis is a critical step in the progression of malignancy and continues to be a formidable problem in the treatment of human cancers.23 Recently, there has been an increased awareness of the importance of tumor microenvironment in metastasis. The interaction between the stromal compartment and tumor cells is bidirectional in terms of an impact on the metastatic process.24 One scenario is that tumor cells may modulate gene expression in host cells residing in the microenvironment to generate a permissive/conducive environment for tumor invasion. Likewise, host cells in microenvironment may induce genetic changes and clonal selection in tumor cells that generates a subpopulation of tumor cells with the capability to invade into the surrounding tissue.25, 26 Many recent studies support the first hypothesis; for example, Stearman et al.27 reported that macrophages in lung tumor environment displayed a unique gene expression signature suggesting a role in promoting tumor metastasis. Richardson et al.28 conducted experiments to compare expression profiles in the 4 major components of microenvironment: tumor epithelium, tumor associated stroma, normal epithelium and normal stroma using human prostate cancer specimens. In these studies, they determined that tumor-associated stroma showed a predominant upregulation of transcripts compared to normal stroma. Previous studies by our laboratory studies corroborated their observations.1 In those experiments, we explored the effects of melanoma and stroma fibroblast cross-talk via gene expression profiling, and observed that human fibroblasts only moderately affected the gene expression pattern of the cocultured A2058 human melanoma cells but were themselves significantly affected by the melanoma cells. In our investigation, we have explored whether the ability of melanoma to alter fibroblast gene expression is correlated with their invasive/metastatic capability by performing gene expression profiling on fibroblasts cocultured with 5 representative melanoma cell lines BLM, MV3, A2058, SK-mel28 and WM164. As observed in animal models of melanoma metastasis, these melanoma cell lines demonstrate differences in their ability to invade the stroma and form metastases. BLM and MV3 are highly invasive and metastatic cell lines as evaluated by subcutaneous inoculation into nude mice,9, 11, 12 whereas A2058 and SK-mel28, derived from melanoma metastases, are aggressive tumor cell lines but are somewhat less metastatic in animal models.13, 20 WM164 is not metastatic in nude mice models of melanoma metastasis.21, 22 Gene expression of profiling of these lines mirror these biological attributes in that highly invasive BLM and MV3 cell line have similar gene expression patterns, which are distinct to those of less-invasive melanoma cell lines.

Interestingly, the gene expression profiles of melanoma-cocultured fibroblasts clustered together with a pattern that mirrored the invasive/metastatic potential of the melanoma cell lines with which they were cocultured. This reflection of melanoma cell line invasiveness/metastatic potential in their ability to stimulate cocultured fibroblasts suggests a potential role of invasive/metastatic melanoma in orchestrating invasion and metastasis via the host stromal fibroblasts. Many mechanisms including angiogenesis,29 extracellular matrix remodeling30 and adhesion/invasion have been suggested to aid in the metastasis of neoplastic cells. In our profiling data, the genes upregulated in fibroblasts stimulated by coculture with BLM or MV3 cells fall into a number of functional categories associated with chemokine upregulation, clearly indicating that one of the primary responses of fibroblasts to coculture with invasive/metastatic melanoma lines is an inflammatory response of sorts predominated by expression of chemokines and cytokines. Inflammatory cells and mediators are extensively involved in the invasion and metastasis of malignant cells. For example, coinjection of bone marrow-derived mesenchymal stem cells enhanced the metastatic potency of MDA-MD-231 human breast carcinoma cells growing at subcutaneous sites via upregulating CCL5-CCR5 signaling.31 Overexpression of CXCL14 and CXCL12 chemokines in breast cancer-associated myoepithelial cells and myofibroblasts promote tumor cell migration, invasion and metastasis.32, 33 On the basis of the data in our study, we observed that cytokine and chemokine CXC and CC family members (CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, IL-6, IL-1B, IL-8, CCL2/MCP1) and growth factors (VEGF, FGF) were consistently upregulated in fibroblasts cocultured with invasive/metastatic melanoma BLM and MV3 cell lines compared to fibroblasts cocultured with nonmetastatic melanoma WM164. Among these genes, of particular relevance to the malignant process is IL-1B. IL-1B, a secreted form of IL-1, has been suggested to function as a master cytokine in regulation of tumor invasiveness.34–36 Vidal-Vanaclocha et al. observed that stroma-derived IL-1B promoted melanoma liver metastasis following intrasplenic injection of recombinant IL-1B or LPS, a strong IL-1 inducer, whereas neutralizing IL-1B along with IL-1 antibody reduced the metastasis and increased survival rates.37, 38 Another study demonstrated that in IL-1B knockout mice, local tumor or lung metastases of B16 melanoma cells were not observed compared to wild-type mice.39 IL-1B promotes tumor invasion mainly via inducing the expression of diverse proinflammatory molecules, such as CXC chemokines, IL-8, IL-6 and CCL2/MCP-1.34–36 The CXC chemokine family has been found to be associated with tumorigenesis, angiogenesis and metastasis.40, 41 High nuclear CXCL1 in breast cancer has been associated with decreased survival and lymph node involvement.42 Protein expression of CXCL1 regulated the invasive ability of bladder cancer cells via MMPs.43 CXCL2 has been shown to promote outgrowth of colorectal liver metastasis.44 IL-6 is a cytokine with a wide variety of biological functions. In carcinoma of the cervix and melanoma, high gene expression of IL-6 with tumor invasiveness is correlated with poor prognosis.45, 46 IL-8 is defined as one of the most important angiogenic chemokines47, 48 in that it exhibits potent angiogenic activities both in vivo and in vitro.47–49 The tumor-derived IL-8 has been reported to increase tumor progression and migration in a variety of human cancers including melanoma, gastric cancer and lung cancer.50–52 Furthermore, IL-8 production by melanoma cells directly correlated with their metastatic potential in nude mice.53 CCL2/MCP1 is one of the chemotactic factors predominantly associated with macrophage chemoattraction, which was implicated to be involved in the progression in breast, ovarian, bladder and lung cancer.54–56 In our expression data, IL-1B was observed to be significantly increased in fibroblasts cocultured with high-invasive melanoma cells along with a concomitant upregulation of a group of other proinflammatory genes known to be associated with IL-1B expression including CXCL 1,2,3,5,6,14, IL-8, CCL2/MCP-1 and VEGF. Immunohistochemical staining for CCL2/MCP-1, a key chemoattractant for macrophages, was observed in both melanoma nests and surrounding stromal tracks in human nodular melanoma. Taken together, these results underscore IL-1B's importance in the regulation of a chemokine inflammation pathway elicited by coculture with fibroblasts.

As noted earlier, we demonstrated IL-1B expression to be associated with stromal tracks in human melanoma. To further investigate the effect of melanoma-stimulated fibroblast-derived IL-1B on melanoma invasion, we conducted a silencing experiment via siRNA on fibroblasts cocultured with BLM melanoma, the cell line which demonstrated the greatest ability to stimulate a proinflammatory response in cocultured fibroblasts. The IL-1B silenced fibroblasts cocultured with BLM showed a marked decrease in promoting melanoma invasion compared to that from control fibroblasts. This is consistent with a previous report that high IL-1B expression is important in promoting melanoma tumor invasion via a generalize, widespread inflammatory reaction.57–59

Recent advances in the understanding of the complex biology of the microenvironment that underlie tumor invasion and migration indicate that the enhancement of cancer progression and metastasis might be regulated by cellular interactions within the microenvironment, whereby a multitude of cytokines and chemokines provide functional cues60 orchestrating the outcome. Our data support this concept in view of the significant alteration of gene expression and protein expression of fibroblasts; we observed as a result of their communication with invasive/metastatic melanoma in the microenvironment. The molecular signals and their mechanism of transduction between host stromal fibroblasts and invasive/metastatic melanoma are currently under investigation in our laboratory.


This work was supported by the Center for Molecular Medicine, University of Cologne (CMMC, B1 to C.M.) and the University of Virginia Cancer Center (J.W.F.).