Cytokines play a crucial role in tumor initiation and progression. Here, we demonstrate that interleukin (IL)-6 is a key factor by driving tumor progression from benign to malignant, invasive tumors in the HaCaT-model of human skin carcinoma. IL-6 activates STAT3 and directly stimulates proliferation and migration of the benign noninvasive HaCaT-ras A-5 cells in vitro. Furthermore, IL-6 induces a complex, reciprocally regulated cytokine network in the tumor cells that includes inflammatory and angiogenic factors such as IL-8, GM-CSF, VEGF and MCP-1. These IL-6 effects lead to tumor cell invasion in organotypic cultures in vitro and to the formation of malignant and invasive s.c. tumors in vivo. Tumor invasion is supported by the IL-6 induced overexpression of MMP-1 in vitro and in vivo. These data demonstrate a key function of IL-6 in the progression of skin SCCs by regulating a complex cytokine and protease network and suggest new therapeutic approaches to target this central player in skin carcinogenesis.
More than a century ago, Virchow postulated a link between inflammation and cancer, based on the presence of leukocytes in neoplastic tissue. In the meantime, this concept has been confirmed by a number of studies. In fact, infection and chronic inflammation contribute to about 25% of cancer cases.1 Immune cells are involved in the regulation of most stages in cancer development2 and elevated serum levels of proinflammatory cytokines like interleukin (IL)-1, IL-6, IL-8 or TNF-α are often associated with tumor progression and a bad clinical prognosis.3–6
IL-6, a pleiotropic cytokine with a wide range of biological functions,7 is secreted by different cell types, e.g., fibroblasts, macrophages, T- and B lymphocytes, endothelial cells and activated keratinocytes.8
IL-6 binds to a specific IL-6R (α-chain; CD126, gp 80). This complex associates with two molecules of the ubiquitously expressed gp130 receptor (β-chain, CD130) that functions as signal transducing receptor subunit for IL-6 and other members of the IL-6 family (e.g., IL-11, LIF and oncostatin M).9, 10 Receptor activation triggers three main signaling pathways: the JAK/STAT, the MAP kinase and the PI3-kinase/Akt pathway.11, 12
IL-6 is upregulated in inflammatory processes like infection, trauma, rheumatoid arthritis, systemic lupus erythematosus, psoriasis and cancer.7, 9, 11, 13 In cancer, high serum levels of IL-6 correlate with disease severity and a bad clinical outcome.9, 14, 15 However, IL-6 can also exert antitumorigenic effects, e.g., by promoting antitumor activities of macrophages and by contributing to the activation of killer cells.9 One example for controversial IL-6 effects is breast cancer. Whereas IL-6 inhibits tumor cell growth in vitro,16 it also confers multi drug resistance to breast cancer cells8 and promotes metastasis in advanced tumor stages.17 Additionally, IL-6 can induce angiogenesis either directly in ovarian cancer18 or indirectly by stimulating the release of VEGF from cervical cancer and glioblastoma cells.19 In support of this, IL-6 deficient mice exhibit reduced angiogenesis and delayed wound healing.20
In our study, we demonstrate that IL-6 is a key factor for mediating tumor progression from a benign to a malignant, invasive tumor phenotype in the HaCaT skin carcinogenesis model. IL-6 activates STAT3-mediated signal transduction and strongly stimulates proliferation and migration of the benign HaCaT-ras A-5 cells. Transfection of these cells with IL-6 to induce overexpression even promotes tumor cell invasion in organotypic cultures (OTCs) in vitro. Importantly, IL-6 induces a complex, reciprocally regulated cytokine network in the tumor cells that includes inflammatory cytokines like IL-8, GM-CSF and MCP-1 as well as the angiogenic factor VEGF and leads to malignant and invasive tumor growth in vivo. These IL-6 induced cytokines again promote tumor cell proliferation and migration, while the IL-6-mediated overexpression of MMP-1 can facilitate tumor cell invasion. These data demonstrate a key function of IL-6 in the progression of skin tumors by regulating a complex cytokine network and by inducing the expression of invasion-supporting proteases. Thus, IL-6 and its cytokine network act in an autocrine manner on the tumor cells themselves and are able to activate a tumor supporting inflammatory and angiogenic stroma. This suggests IL-6 as a central player in skin carcinogenesis that might be a prominent target for new therapeutic approaches.
Material and Methods
Cell lines used in our study are the malignant cell lines HaCaT-ras A-5RT1 and A-5RT321 and the benign cell line HaCaT-ras A-5.21, 22 Benign HaCaT-ras A-5 cells were transfected with the pZeoSV2(−) expression vector containing the coding sequences for IL-6 using the Gene Jammer kit (Stratagene, La Jolla, CA). The cell lines HaCaT-ras A-5IL-6D (moderate IL-6 expression: 120 pg/ml × 106 cells), A-5IL-6E (high IL-6 expression: 470 pg/ml × 106 cells) and A-5IL-6F (high IL-6 expression: 460 pg/ml × 106 cells) were generated. Control transfectants A-5C3 (IL-6 expression: 45 pg/ml × 106 cells) contain the vector alone.
HaCaT-ras A-5, A-5RT1 and A-5RT3 cells were cultivated in 4× MEM (modified Eagle's medium) + 10% FCS + neomycin (200 μg/ml, PAA Laboratories, Coelbe, Germany), the transfectants were cultivated in 4× MEM + 10% FCS + 200 μg/ml Neomycin + 200 μg/ml Zeozin, as described.23 Cells were passaged at a split ratio of 1:6 to 1:10, routinely tested for mycoplasma contamination24 and always found to be negative.
A 1,120 bp EcoRI/HindIII cDNA fragment of IL-6 was isolated from the expression plasmid pGEM-225 and ligated into the multiple cloning site of pZeoSV2(−) (Invitrogen, Karlsruhe, Germany). The resulting IL-6 expression plasmid was verified by restriction digest and sequence analysis.
Stimulation assay and ELISA
A total of 1 × 105 HaCaT-ras A-5, A-5IL-6E or A-5C3 cells were seeded per 6-cm culture dish in medium containing 10% FCS. After 24 hr, cells were shifted to medium with 0% FCS (control) or with 0% FCS and different concentrations of IL-6, GM-CSF, VEGF, MCP-1 and IL-8 (all R&D Systems, Wiesbaden, Germany). Alternatively, an IL-6 antibody (MAB2061, 1 μg/ml, R&D Systems) or an irrelevant mouse IgG antibody (M7894, Sigma-Aldrich, Taufkirchen, Germany) were used. Conditioned medium was harvested 3, 5 and 7 days poststimulation, centrifuged for 5 min at 10,000g and stored in aliquots at −80°C. Cells were trypsinized and counted. Secretion of G-CSF, GM-CSF, PDGF-BB, HGF, IL-8, MCP-1, VEGF, IL-6 and MMP-1 were measured by ELISA using Quantikine Immunoassay kits (R&D Systems). Samples were tested in duplicate and experiments were repeated twice. Data shown are mean values ± SD.
Cell proliferation (SYBRR Green Assay)
Cells were seeded in 24-well plates at a density of 2 × 104 cells per well (HaCaT-ras A-5, A-5RT3, A-5IL-6F) or of 1 × 104 cells per well (HaCaT-ras A-5, A-5C3, A-5IL-6E) in medium containing 10% FCS. Twenty-four hours later, medium was shifted to 0% FCS with or without GM-CSF, IL-6 (both 50 ng/ml) and/or neutralizing antibodies against either factor (AF-215-N against GM-CSF, MAB2061 against IL-6, both 1 μg/ml; all R&D Systems) or an irrelevant IgG as control (M7894, Sigma-Aldrich) and incubated for 72 hr. After washing with PBS, the 24-well plates were stored at −20°C. For the assay plates were thawed and incubated with lysis buffer (PBS with 0.1% Triton-X 100 and 1 μg/ml SYBRR Green I nucleic acid stain (Molecular Probes, Leiden, Netherlands) at RT in the dark for 1 hr. Cell proliferation reflected by the quantity of cellular DNA was measured by determining fluorescence intensity in a fluorescent reader (Fluoroscan, Labsystems, Niantic, CT). Experiments were done in triplicate and repeated twice. Data shown are mean values ± SD.
A total of 1 × 106 cells per well were seeded into 6-well plates and grown to confluence. Twenty-four hours later, the cell monolayer was disrupted using a cell scraper of 1-cm width and the borders of the cell-free area were marked. The medium was replaced by serum-free medium containing 50 or 75 ng/ml IL-6. Cell migration was documented by microscopic photos taken after 0, 24 and 48 hr at 200× magnification. Migration area was determined using the analySIS software (Soft Imaging Systems, Münster, Germany). Experiments were done in duplicate and repeated twice. Data shown are mean values ± SD.
Epithelial growth in 3D OTCs in vitro
Dermal equivalents either containing 2.5 × 105 human dermal fibroblasts or without fibroblasts were prepared with native type I rat collagen (final concentration of 3 mg/ml) as described.23, 26 Tumor cells (1 × 106 of HaCaT-ras A-5C3, A-5IL-6D, A-5IL-6E and A-5IL-6F, respectively) were plated on top. After 24 hr, cultures were raised to the air-medium interface by lowering the medium level. For 3 weeks, two cultures per week were harvested and processed for histology and cryostat sectioning. Data are representative of two independent experiments with two replicas each.
Tumorigenicity assays in vivo
Tumor growth was analyzed by subcutaneous injection of 5 × 106 cells (HaCaT-rasA-5, A-5C3 and A-5IL-6E) in a volume of 100 μl into the interscapular region of 4- to 6-week-old athymic nude mice (Swiss nu/nu). Tumor size was measured weekly, and tumor volume was calculated as follows (l × w2) × 0.52, where w represents the smallest diameter of the tumor. Tumors were taken out after 90 days and further processed for histology and cryostat sectioning.
RNA isolation and reverse transcription
Total RNA was isolated with the RNeasy Mini kit (Qiagen, Hilden, Germany). RT-PCR was carried out with the Omniscript RT Kit (Qiagen). Reverse transcription was done in a volume of 20 μl, using 2 μg total RNA; 10× RT-buffer; 5 mM each of dATP, dGTP, dCTP and dTTP (Sigma-Aldrich, Taufkirchen, Germany) and 20 U/μl RNAse-inhibitor (Roche Diagnostics); 50 U/μl Omniscript Reverse Transcriptase; 1 μM Random-Hexamer Primer and 1 μM Oligo dT16 Primer.
Oligonucleotide primers and PCR
Twenty-microliter PCR reaction contained 2 μl of the RT-reaction and 5 U Taq DNA polymerase (Qiagen). MgCl2 concentration and annealing temperature were optimized for each primer set. Sense and antisense primers were synthesized according to the sequence extracted from the nucleotide database. The primers used were as follows: Gp80 (IL-6R α subunit): 72°C; 5′-primer, GTGTCCATGTGCGTCGCCAGTAG; 3′-primer, TCACCTCGCTCACCTCGGGC; Gp130 (IL-6R subunit): 58°C; 5′-primer, ATTCGGACAGCTTGAACAGA; 3′-primer, GGTTAGGCGGTGTATTAAAT; GAPDH: 60°C; 5′-primer, GGTGAAGG TCGGAGTCAACGGA; 3′-primer, GAGGGATCTCGCTCC TGGAAGA; IL-6: 66.5°C; 5′-primer, ATGAACTCCTTCTCC ACAAGCGC; 3′-primer, GAAGAGCCCTCAGGCTGGAC TG. All primers pairs spanned intron–exon splice sites to ensure that PCR products did not derive from any DNA present in the RNA preparations.
Cryosections were mounted on slides, fixed for 5 min in 80% methanol at 4°C, followed by 2 min in acetone at −20°C and re-hydrated in PBS as described.23 Stained sections were examined and photographed with the Leica DM RBE microscope (Leica GmbH, Bensheim, Germany) fitted with epifluorescence.
The following primary antibodies were used: Anti-pan-cytokeratin, guinea pig polyclonal, (GP14, Progen, Heidelberg Germany); anti-Ki67, mouse monoclonal (#Dia505, Dianova, Hamburg Germany); anti-murine CD31, rat monoclonal (#01951D, BD PharMingen, Heidelberg, Germany); anti-human collagen IV, rabbit polyclonal (#20451, Novotec, Lyon, France), anti-murine neutrophils, rat monoclonal (#MCA771G, Serotec, Duesseldorf Germany); anti-tissue macrophages, rat monoclonal (ER-MP23; Acris, Bad Nauheim, Germany).
The following secondary antibodies were used: Anti-guinea pig, from donkey, DTAF (#706-015-148, Dianova); anti-guinea pig, from donkey, AMCA (#706-156-148, Dianova); anti-mouse from goat Texas red (#115-076-062, Dianova); anti-rat from donkey Texas red (#712-076-153, Dianova) and anti-rabbit from goat Texas Red (#111-076-045, Dianova).
In situ hybridization
In situ hybridization was performed nonradioactively as described.27 DIG signals were detected by anti-DIG alkaline phosphatase (DIG-AP) (Roche Diagnostics) and NBT/BCIP substrate reaction (Promega, Mannheim, Germany). The plasmid for hMMP-1 was kindly provided by P. Angel DKFZ, Heidelberg, Germany. Sections were counterstained against pan-keratin and collagen type IV and examined as described above. For better visualization of DIG signals, colors were reassigned with the analySIS software (Soft Imaging Systems).
Immunoprecipitation and Western blotting
After 24-h serum starvation HaCaT-ras A-5 cells were cultured with or without (control) 100 ng/ml IL-6 (R&D Systems). Cells were lysed after 0, 5, 15, 30 and 45 min in ice-cold NP-40 lysis buffer (150 mM NaCl, 20 mM Tris pH 7,4, 1 mM EDTA, 1 mM ZnCl2, 1 mM MgCl2, 1 mM Na3VO4, 10% glycerin, 1% NP-40, protease inhibitor cocktail (Roche Diagnostics). Lysates were centrifuged for 10 min at 13,000 rpm to separate cytoplasmic (supernatant) and nuclear (pellet) fraction. Immunoprecipitation was performed from the cytoplasmic fraction overnight at 4°C with anti-STAT3 antibody (#9132, Cell Signaling Technology, Inc, MA) and 5% A/G Plus agarose beads (sc-2003, Santa Cruz Biotechnology, Inc Heidelberg, Germany). Recombinant GST-STAT3 (30 ng, kindly provided by U. Klingmueller, DKFZ Heidelberg, Germany) was added as internal control to the immunoprecipitation. Beads were washed with lysis buffer and precipitated proteins were eluted in Laemmli sample buffer for 5 min at 95°C. Subsequently, proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Phosphorylated STAT3 protein was detected with a primary phospho-STAT3 antibody (Tyr705; #9131, Cell Signaling), a horseradish peroxidase coupled secondary antibody (Protein A-HRP, GE Healthcare, München, Germany) and subsequent ECL reaction according to the manufacturers description (GE Healthcare).
A two-tailed Student's t test was used for data analysis, with *p < 0.05 considered as statistically significant and **p < 0.001 as highly significant.
3D: three-dimensional; IL: interleukin; JAK: Janus-activated kinase; MMP: matrix metalloproteinases; OTC: organotypic culture; STAT: signal transducers and activators of transcription Socs; TNF: tumor necrosis factor
IL-6 expression is increased with tumor progression in human skin SCC
We previously demonstrated a critical contribution of the hematopoietic factors G-CSF and GM-CSF to tumor growth and progression in the HaCaT skin carcinogenesis model23 and verified their tumor promoting role for human gliomas, meningiomas and head and neck squamous cell carcinomas.26, 28, 29 Further comparison of the growth factor expression profile of benign, low grade and high grade malignant HaCaT-ras cell lines showed an increase in IL-6 mRNA expression corresponding to tumor progression (Fig. 1a, upper lane). High grade malignant HaCaT-ras A-5RT3 cells also secreted considerably more IL-6 protein (50–170 pg/ml × 106 cells) than benign HaCaT-ras A-5 cells (6–14 pg/ml × 106 cells). All HaCaT-ras cell lines expressed the mRNA for the signal transducing IL-6R gp130 and the ligand binding subunit IL-6R α (gp80) (Fig. 1a, lower lanes). Normal keratinocytes express IL-6 only in a highly regulated nonconstitutive fashion. Thus, the enhanced and constitutive expression of IL-6 together with the respective IL-6R in high grade malignant HaCaT-ras cells suggested an autocrine stimulatory loop.
IL-6 dependent autocrine stimulation of tumor cell proliferation and migration and activation of STAT3
Potential autocrine effects of IL-6 on HaCaT-ras tumor cells were analyzed by cultivating benign HaCaT-ras A-5 cells in the presence of increasing amounts of IL-6. IL-6 promoted the proliferation (Fig. 1b) and migration (Fig. 1c) of benign HaCaT-ras A-5 cells in a dose-dependent manner. To confirm the activation of IL-6 associated signal transduction in the tumor cells, we analyzed the IL-6 induced phosphorylation of STAT3, the main mediator of IL-6 signaling30 and a signal transduction molecule that is inappropriately activated in a wide range of human cancers.31 Benign HaCaT-ras A-5 cells were stimulated with 100 ng/ml IL-6 and STAT3 phosphorylation was analyzed by immunoprecipitation and Western blot. The amount of phosphorylated STAT3 increased immediately after IL-6 stimulation and remained elevated for 45 min (Fig. 1d), suggesting its role in promoting proliferation and migration of the keratinocyte tumor cells.
IL-6 promotes epithelial growth in OTCs and invasion in fibroblast containing OTCs in vitro
The proliferation and migration promoting effect of IL-6 on tumor cells further suggested a tumor progression promoting role. To address this question, we constitutively overexpressed IL-6 in the low expressing benign HaCaT-ras A-5 cells (Fig. 1a). Two HaCaT-ras transfectant clones expressing high levels of IL-6 (A-5IL-6E: 470 pg/ml × 106 cells, A-5IL-6F: 460 pg/ml × 106 cells) and one clone with moderate IL-6 expression (A-5IL-6D: 120 pg/ml × 106 cells) were selected. HaCaT-ras transfectants with vector alone (A-5C3: 45 pg/ml IL-6 × 106 cells) and/or parental HaCaT-ras A-5 cells (6–14 pg/ml IL-6 × 106 cells) were used as controls in all subsequent experiments. Despite the coexpression of IL-6 and its receptors, IL-6 expression did not significantly enhance tumor cell proliferation in monolayer culture in vitro (data not shown).
In contrast, high IL-6 expression resulted in clearly enhanced epithelial growth in 3D OTCs on a collagen type I gel. All HaCaT-ras transfectants formed a multilayered epithelium on top of the collagen gel (Fig. 2a, shown for HaCaT-ras A-5C3 and A-5IL-6E), yet throughout the whole observation period epithelia of the IL-6 transfectants A-5IL-6E were clearly thicker compared to the control-transfectants A-5C3 (Fig. 2a, data shown for week 1). This was in part mediated by an increased proliferation of the IL-6 transfectants HaCaT-ras A-5IL-6E (Fig. 2b, Ki67 staining, shown for week 1).
In OTCs with normal human dermal fibroblasts embedded in the collagen gel, resembling more closely the in vivo situation, a similar proliferation-stimulating effect of IL-6 was observed (data not shown). Additionally, the HaCaT-ras transfectants with high IL-6 expression showed invasive tumor cell growth in these cultures starting from the second week. In contrast, transfectants with low and moderate IL-6 expression did not grow invasively into the collagen gel (Fig. 2c, shown for weeks 2 and 3, arrows mark the invasive areas). These results clearly demonstrated that IL-6 can promote tumor cell invasion in an in vivo-like OTC system.
IL-6 induces the expression of additional cytokines in a cytokine network
The IL-6-mediated tumor cell invasion raised the question whether invasion was solely mediated by IL-6 or additionally promoted by IL-6 induced growth factors. The IL-6-mediated STAT3 activation and the known involvement of STAT3 signaling in the induction of tumor promoting cytokines and proteases32 suggested that IL-6 might activate additional growth factors. Therefore, we analyzed the influence of IL-6 on the secretion of factors that were previously shown to promote tumor progression in HaCaT skin tumors such as G-CSF, GM-CSF, HGF, IL-8, MCP-1, PDGF-BB and VEGF-A. Benign HaCaT-ras A-5 cells were cultivated for 3, 5 and 7 days in the presence of IL-6 (IL-6, 50 ng/ml), either with or without an IL-6 neutralizing antibody (Ab, 1 μg/ml) and cytokine secretion was measured by ELISA. Unstimulated HaCaT-ras A-5 cells were used as control (Co). To confirm the IL-6 specific effects on cytokine expression, the HaCaT-ras transfectants with high IL-6 expression (A-5IL-6E) and the control transfectants (A-5C3) were cultivated for the same time period in the presence and absence of IL-6 neutralizing antibodies and conditioned media were analyzed for cytokine secretion.
IL-6 stimulation markedly enhanced the expression of GM-CSF, IL-8, MCP-1 and VEGF in benign HaCaT-ras A-5 cells (Figs. 3a–3d). In contrast, IL-6 did not affect the expression of G-CSF, HGF and PDGF-BB (data not shown). In agreement with these findings, we also observed an upregulation of GM-CSF, IL-8, MCP-1 and VEGF in the IL-6 transfectants HaCaT-ras A-5IL-6E that could be inhibited by an IL-6 neutralizing antibody (Figs. 3e–3f, shown for GM-CSF and IL-8 at day 5). The antibody had no effect on the control transfectants A-5C3.
Reciprocal induction of factors from the cytokine network
IL-6 and the IL-6 induced cytokines (i.e., GM-CSF, IL-8, MCP-1 and VEGF) have complex pleiotropic functions, suggesting a cytokine network with reciprocal regulatory interactions. To address this assumption, benign HaCaT-ras A-5 cells were stimulated with the IL-6 induced cytokines and secretion of different cytokines was measured in conditioned media by ELISA. Unstimulated HaCaT-ras A-5 cells were used as controls (Co). Indeed, the IL-6 induced cytokines reciprocally regulate each others expression in benign HaCaT-ras A-5 cells. In addition to the IL-6 dependent upregulation of GM-CSF (see Figs. 3a and 3e), GM-CSF secretion was also enhanced by VEGF and IL-8 (Fig. 4a, left panel). IL-8 in turn was upregulated by VEGF and GM-CSF (Fig. 4a, right panel) and MCP-1 secretion was markedly stimulated by VEGF and IL-8 (Fig. 4b). Finally, secretion of IL-6 itself was slightly enhanced in response to VEGF and GM-CSF (Fig. 4c). Thus our data demonstrate a reciprocal regulation of IL-6, GM-CSF and VEGF and establish an IL-6 induced cytokine network with feed back regulation loops that may promote growth and progression of human skin SCCs.
IL-6 induced proliferation is GM-CSF dependent and vice versa
The reciprocal regulation of GM-CSF and IL-6 extended beyond the level of protein secretion and included their stimulatory effect on tumor cell proliferation. IL-6 and GM-SCF potently stimulated the proliferation of benign HaCaT-ras A-5 cells in comparison to unstimulated HaCaT-ras A-5 cells (control) (Fig. 4d, and compare Fig. 1b). Interestingly, GM-CSF neutralizing antibodies (Ab) completely abrogated the IL-6 induced cell proliferation in A-5 cells (Fig. 4d, left panel) and vice versa, IL-6 neutralizing antibodies (Ab) inhibited the GM-CSF-mediated increase in cell proliferation (Fig. 4d, right panel).
IL-6 promotes invasive tumor growth in vivo
To finally confirm the tumor growth and invasion promoting effects of IL-6 and its growth factor network in vivo, we subcutaneously injected the IL-6 overexpressing transfectants HaCaT-ras A-5IL-6E, the control transfectants HaCaT-ras A-5C3 and the parental HaCaT-ras A-5 cells into athymic nude mice. Control transfectants (A-5C3) and parental A-5 cells formed small tumors for a period of about 30 days and completely regressed thereafter (Fig. 5a, left panel, data shown for A-5C3). In contrast, the IL-6 transfectants A-5IL-6E formed progressively enlarging tumors (Fig. 5a, right panel). Histological analysis revealed an invasive and vascularized tumor tissue with an abundant accumulation of macrophages and neutrophils in the tumor stroma and neutrophils infiltrating the tumor tissue (Fig. 5b). Additionally, we observed a strong expression of MMP-1 mRNA in the tumor cells at the tumor stroma border of the HaCaT-ras A-5IL-6E tumors (Fig. 5c) that was completely lacking in control tumors (data not shown). This suggested an IL-6-mediated upregulation of MMP-1 that possibly facilitates tumor vascularization and invasion. The direct effect of IL-6 on MMP-1 expression was confirmed in vitro where we could demonstrate a strong dose-dependent increase in MMP-1 protein secretion in IL-6 stimulated benign HaCaT-ras A-5 cells that could be blocked by IL-6 neutralizing antibodies (Fig. 5d). Thus, our data clearly demonstrate that IL-6 promotes the progression of a benign skin tumor to an invasive, well vascularized SCC.
IL-6 is a pleiotropic cytokine that plays a role in the physiology of virtually every organ system and whose deregulated expression is associated with chronic inflammatory processes like rheumatoid arthritis, autoimmune disease and cancer.7 In a variety of tumor types, high serum levels of IL-6 correlate with increased malignancy and a poor clinical outcome.9, 14, 15 In agreement with these clinical data, IL-6 expression is strongly upregulated upon progression from benign HaCaT-ras A-5 tumors to highly malignant HaCaT-ras A-5RT3 skin SCCs. In our study, we analyzed potential progression promoting functions of IL-6 in our HaCaT skin carcinogenesis model in vitro and in vivo by stimulating benign HaCaT-ras A-5 cells with IL-6 and by generating IL-6 overexpressing HaCaT-ras A-5 transfectants. We identified IL-6 as a key mediator of a complex, tightly regulated cytokine network that promotes the progression to malignant, invasive SCCs.
IL-6 stimulation directly promoted the proliferation and migration of benign HaCaT-ras A-5 cells in monolayer culture. This effect was also observed in 3D OTCs of the HaCaT-ras transfectants A-5IL-6E with high IL-6 expression that showed a clearly enhanced epithelial thickness and tumor cell proliferation compared to control transfectants. These results demonstrated that IL-6 is an autocrine growth factor that maintains its effect in a tissue context over an extended period of 3 weeks. Direct tumor promoting effects of IL-6 became evident in OTCs containing fibroblasts as cells of the stroma. In these cultures, the HaCaT-ras A-5 transfectants with high IL-6 expression (A-5IL-6E and A-5IL-6F) showed invasive growth into the fibroblast containing collagen gel, whereas the transfectants with low and moderate IL-6 expression did not grow invasively. This suggested a stroma-dependent invasion stimulatory effect of IL-6 that is subject of further studies in our lab and highlights the central role of the tissue context for studying tumor progression promoting mechanisms in vitro.
Direct autocrine tumor promoting effects of IL-6 have been also demonstrated in multiple myeloma by increasing proliferation33 and preventing apoptosis.34 Similarly, IL-6 has been identified as an autocrine stimulator of melanoma progression.35 In melanoma like in many other cancer types, STAT3 is the main intracellular mediator of IL-6 signaling,32 leading to enhanced cell proliferation, migration and survival.36–38 In agreement with this, STAT3 phosphorylation was strongly increased in IL-6 stimulated HaCaT-ras A-5 cells. Besides these direct effects on cell proliferation and migration, STAT3 was demonstrated to indirectly stimulate angiogenesis, invasion and metastasis of melanomas by upregulating bFGF, VEGF and MMP-232, 39 and by promoting an MMP-1-mediated invasion of colon cancer cells.40 This indicated that IL-6 signaling exerts pleiotropic effects by the induction of additional growth factors and prompted us to analyze the influence of IL-6 on the expression of well known tumor progression promoting cytokines in HaCaT-ras A-5 cells.
Indeed, IL-6 stimulated the expression of GM-CSF, IL-8, MCP-1 and VEGF in the benign HaCaT-ras A-5 cells, an effect that could be blocked by an IL-6 neutralizing antibody and was also observed for the IL-6 overexpressing HaCaT-ras transfectants A-5IL-6E. These data extended the stimulatory effects of IL-6 to the role of a central regulator of additional tumor promoting factors.15
The IL-6 stimulated expression of different pleiotropic cytokines41–43 suggested the presence of a complex cytokine network rather than a unidirectional cytokine regulation. Indeed, further analyses revealed several reciprocal feed back loops within the cytokine network of the tumor cells. VEGF markedly stimulated the expression of GM-CSF, IL-8 and MCP-1 and slightly affected the IL-6 expression of HaCaT-ras A5 cells. Additionally, IL-8 slightly increased the GM-CSF expression of HaCaT-ras A-5 cells and strongly enhanced their MCP-1 expression. Finally, also GM-CSF slightly upregulated IL-6 and IL-8 expression by the HaCaT-ras A-5 cells (Fig. 6). These data establish a novel IL-6 induced cytokine network with complex reciprocal feed back loops that includes a number of factors known for their progression promoting function.
Strikingly, the regulatory interaction of IL-6 and GM-CSF extended beyond the level of a mere stimulation of protein secretion, but rather included an interactive effect on tumor cell proliferation. Accordingly, the IL-6 dependent proliferation was efficiently blocked by a GM-CSF neutralizing antibody and vice versa, the GM-CSF stimulated proliferation was abrogated by an IL-6 neutralizing antibody. This observation suggests an overlap in the signaling pathways and target genes of IL-6 and GM-CSF and is subject of further studies. Thus we provide the first experimental evidence for IL-6 as regulator of a complex, reciprocally regulated cytokine network controlling proliferation, migration and invasion of tumor cells.
In support of the in vitro-data suggesting a central role of IL-6 as a tumor promoting factor with pleiotropic functions, IL-6 overexpression promoted the progression of benign skin tumors to malignant, invasive SCCs in vivo. In contrast to the small tumor nodules that transiently formed after s.c. injection of the parental HaCaT-ras A-5 cells and the control transfectants, the IL-6 overexpressing transfectants HaCaT-ras A-5IL-6E induced persistent tumor growth with tumor sizes up to 300 mm3 after 90 days. The transient tumor formation of the control transfectants and the parental cells during the first 30 days might be explained by an activation of the innate immune system in a wound like reaction and a subsequent eradication of the tumor cells. In contrast, IL-6 induces the re-programming of the tumor stroma and potentially mediates the transition from an innate immune response to a more sustained, adaptive response.44 IL-6 polarizes macrophages towards a tumor supporting phenotype (M2)45 and accordingly plays a crucial role in tumorigenesis associated with chronic inflammation.44 In agreement with this, the A-5IL-6E transfectant tumors were invasive, well vascularized and infiltrated by inflammatory cells. In addition, a strong MMP-1 expression was detected in tumor cells at the tumor stroma border zone in vivo and was confirmed in vitro by the IL-6 induced overexpression of MMP-1 in benign HaCaT-ras A-5 cells. This MMP-1 overexpression might be mediated by STAT3 signaling as described for colon cancer40 and seems to play an important role in tumor progression of the HaCaT-ras tumor cells, possibly by facilitating tumor cell invasion. In support of this hypothesis, MMP-1 expression is associated with the progression of dysplasia to carcinoma in oral SCCs46 and seems to play a role in early steps of tumor development in skin SCCs.46, 47
However, in addition to the overexpression of IL-6 by the HaCaT-ras cells, our studies revealed the important role of stromal cells in supporting invasive tumor growth. This became already obvious in 3D OTCs in vitro, where the HaCaT-ras transfectants with high IL-6 expression exhibited invasive growth into the collagen gel in vitro when normal human dermal fibroblasts were embedded in the collagen gel, thus demonstrating that invasive tumor growth is dependent on the stromal counterpart. This was further substantiated by the prominent inflammatory cell recruitment and the high degree of vascularization in the s.c. tumors of IL-6 overexpressing tumor cells. Thus we conclude that in addition to autocrine effects, IL-6 also exerts paracrine effects on the tumor environment that contribute to invasive tumor growth. In addition to IL-6 itself, the IL-6 induced upregulation of GM-CSF, IL-8 and VEGF in the tumor cells are likely to be involved in the activation of a tumor supporting stromal environment, since these factors are known as crucial mediators of inflammatory cell and endothelial cell activation.41–43 In agreement with this, various findings from the literature describe strong paracrine effects of IL-6 in tumorigenesis, e.g., in prostate cancer, including stimulation of angiogenesis.18, 19, 48–51 Thus, our results clearly demonstrate that the direct IL-6-mediated activation of tumor and stromal cells together with the cell activation and recruitment that is mediated by the IL-6 regulated cytokine network play a crucial role for the establishment of an invasive tumor phenotype.
Taken together, our study provides new insights into the mechanisms of an IL-6-mediated tumor progression that might explain the direct correlation of elevated IL-6 serum levels and cancer severity and poor prognosis. We demonstrate that IL-6 is a key regulator of tumor progression by inducing a complex, reciprocally regulated cytokine and protease network that includes GM-CSF, IL-8, MCP-1, VEGF and MMP-1 and stimulates malignant progression by autocrine and paracrine mechanisms. These findings highlight IL-6 as a promising target for efficient cancer therapy that is directed against both the activated tumor cells as well as the tumor supporting stroma.
We thank A. Pálfi and H. Steinbauer for excellent technical assistance.