Inflammation-Induced Chemokine Expression in Uveal Melanoma Cell Lines Stimulates Monocyte Chemotaxis

Abstract

Purpose.: Uvealmelanoma (UM) is the most common primary intraocular tumor in adults and the presence of infiltrating leucocytes is associated with a poor prognosis. Little is known how infiltrating leucocytes influence the tumor cells. The purpose of this study was to investigate the effect of activated T cells on the expression of chemotactic cytokines in UM cells. Furthermore, we examined the ability of stimulated UM cells to attract monocytes.

Methods.: We used an in vitro coculture system in which UM cell lines and T cells were cultured together, but separated by a membrane. Uveal melanoma gene expression was quantified using a microarray. Protein expression in the supernatant was quantified with ELISA or cytometric bead array. For the monocyte migration assay, a transwell plate was used.

Results.: Gene-expression analysis of UM cell lines showed that coculture with activated T cells resulted in an upregulation of chemokines such as CXCL8, CXCL9, CXCL10, CXCL11, CCL2, CCL5, VEGF, intracellular adhesion molecule 1 (ICAM1), and granulocyte-macrophage colony-stimulating factor (GM-CSF). The upregulation of these molecules was confirmed at the protein level. This increase of chemokines coincided with an increased chemotactic capacity of the supernatant toward monocytes.

Conclusions.: Cytokines derived from activated T cells shifted the UM cell transcriptome toward a more inflammatory state, including upregulation of several chemokines, which led to an increased migration of monocytes. Therefore, UM cells might actively participate in generating a tumor-promoting inflammatory microenvironment.

Introduction

Uveal melanoma (UM) is a relatively rare tumor. However, it is the most frequently-occurring primary intraocular tumor in Caucasian adults with an incidence rate of 6.3 cases per million per year. Treatment options are limited and once metastases have developed, which happens in up to 50% of all cases, survival is poor. ¹  

Several prognostic parameters of UM that are associated with an unfavorable outcome have been identified, such as the presence of epithelioid cells, a large tumor size, loss of a copy of chromosome 3, a specific gene-expression profile, ² and expression of several immunologic markers. ³ For instance, it has been shown that the expression of the chemokine receptor CCR7 on UM is correlated with poor survival. ⁴ Several studies have shown a high density of CD3⁺ T cells, ⁵ CD8⁺, and CD4⁺ T lymphocytes and regulatory T (Treg) cells in UM. ^6–9 In contrast with other tumors, the presence of tumor-associated lymphocytes (TAL) in UM is associated with a bad prognosis. ^8–10  

Besides various TAL, a high density of tumor-associated macrophages (TAM) are found in some UM, which also correlates to a worse prognosis. ¹¹ Chemoattractants such as CCL2, CCL5, CXCL8, and VEGF recruit TAMs from the blood stream into the tissue, where the TAMs can differentiate into either M1 or M2 macrophages. M1 macrophages are considered to be antitumorigenic, whereas the M2 macrophages promote tumor growth through enhanced angiogenesis, tissue repair, and immune suppression within the tumo. ¹² UM contain especially M2 macrophages. ^13,14  

Cytokines, which are part of the inflammatory phenotype of UM, have been measured in the aqueous ^15–17 or vitreous fluid. ^15,18 In eyes containing a UM, expression of CXCL8, CXCL10, CCL2, CCL5, VEGF, IL-6, TNF-α, and IFN-γ correlated with tumor size. ¹⁸  

Interferon-γ is not only found to be high in UM-containing eyes ^16,18 but is also found in the peripheral blood of the patients, and is associated with increased metastasis formation. ¹⁹ Interferon-γ exerts a direct effect on UM cell lines: stimulating UM cells with IFN-γ resulted in an increased expression of the immunosuppressive molecules indoleamine 2,3-dioxygenase, ²⁰ and PD-L1, ²¹ the adhesion molecule intracellular adhesion molecule (ICAM) 1, ²² and the inflammation-related molecules of the HLA classes I and II. ^23–25 In contrast with many other tumors, in UM, an increased expression of HLA classes I and II is correlated to a worse prognosis. ^26,27  

It is clear that inflammation along with tumor-infiltrating leucocytes (TIL) is an important characteristic of UM, ^5–8,12,13 but the pathophysiological background of this leucocyte infiltration is not yet known. 

Bronkhorst et al. ²⁸ have previously shown that hypoxia leads to the upregulation of several inflammatory genes such as VEGF, ICAM1, and CXCL8 in primary UM cultures, but hypoxia by itself did not induce increased monocyte migration. Thus, there must be other conditions that are involved in monocyte migration. One of these might be the interaction with TIL, which modulate the UM cells. 

This study aimed to investigate the effect of soluble factors derived from activated T cells on gene and protein expression of different chemokines of UM cell lines and their ability to attract monocytes. We found that coculture of UM cell lines with activated T cells in a membrane insert resulted in an increased secretion of chemokines and an increased monocyte migration. 

Materials and Methods

Ethics Statement

Ethical approval was waived by the local ethics committee “Den Videnskabsetiske Komité Københavns og Frederiksberg Kommuner,” which deemed that no results obtained were related to the donors and no samples were stored. Samples were identified by date of blood sampling only, and analyzed anonymously. Verbal informed consent of blood sampling was obtained. The study adhered to the tenets of the Declaration of Helsinki. 

Cell Culture

The following three primary human UM cell lines were used. Mel 290 ²⁹ and Mel 270, ³⁰ which were kindly provided by Hans E. Grossniklaus, MD (Emory Eye Center, Atlanta, GA, USA) and 92.1, which came from one of our laboratories (Leiden University Medical Center, Leiden, The Netherlands). ³¹ These cell lines were obtained from primary UM and are well characterized. ^29–31 All cell lines were maintained in Roswell Park Memorial Institute medium 1640 (RPMI) supplemented with 2 mg/mL sodium hydrogen carbonate (Substrate Department at the Panum Institute, Copenhagen, Denmark), with 10% fetal calf serum (FCS; Seralab, West Sussex, UK), 300 μg/mL L-glutamine and 200 μg/mL penicillin/streptomycin (Gibco, Paisley, UK). There were 100,000 cells of Mel 270 and Mel 290, and 80,000 cells of 92.1 plated into 6-well plates (Becton Dickinson, Le Pont De Claix, France). When Mel 290 was confluent and Mel 270 and 92.1 were subconfluent, medium was changed for all the experiments to X-VIVO 15 serum-free medium (Lonza BioWhittaker, Verviers, Belgium) containing 300 μg/mL L-glutamine, 100 μg/mL penicillin/streptomycin, and 2.5 μg/mL amphotericin B (Gibco, Paisley, UK), which did not have an effect on the further growth of the UM cells as they grew until confluency. X-VIVO 15 serum-free medium was used to allow the downstream use of the supernatants from the UM cell lines in the monocyte migration assay. Supernatants were harvested for analysis from the UM cell lines after they were cultured in X-VIVO 15 serum-free medium for 64 hours at 37°C and 5% CO₂. 

T-Cell Purification and UM–T-Cell Coculture

T cells were purified from fresh whole blood from healthy young volunteers, as previously described.³² The purity of T cells was checked with a FITC-labeled CD3 antibody (Dako, Glostrup, Denmark) and found to be above 91% (Supplementary Material S1). Uveal melanoma cell lines were changed to X-VIVO 15 serum-free medium and T cells were cultured in 0.4 μm polyethylene terephthalate membrane inserts (BD Biosciences, Franklin Lakes, NJ, USA) over UM cell lines. A fixed number of T cells was added to each culture (1.5 × 10⁶ T cells/well in a total volume of 3 mL), to ensure comparable levels of T-cell–derived soluble factors in the medium. T cells had been activated using Dynabeads CD3/CD28 T cell Expander (Invitrogen, Oslo, Norway) and cells were cocultured with UM cells for 64 hours at 37°C and 5% CO₂. The ratio of T cells to beads was 1:1. The distribution of CD4/CD8 T cells was checked with a PerCP-Cy5.5-labeled CD4 antibody (BioLegend, San Diego, CA, USA) and an APC-Cy7-labeled CD8 antibody (BD, Biosciences) before and after stimulation. Before stimulation, the ratio of CD4/CD8 was found to be 78% (SD = 3%)/14% (SD = 3%) and after stimulation 57% (SD = 4%)/27% (SD = 5%; Supplementary Material S2). After coculture, supernatant was collected and RNA was purified from the UM cell lines using the NucleoSpin RNA II Kit (Macherey-Nagel, Düren, Germany) in accordance with the manufacturer's instructions. 

Enzyme-Linked Immunosorbent Assay

Interferon-γ in supernatant was quantified by sandwich ELISA as previously described. ³³ The assay was performed with duplicates on four independent replicates. 

Microarrays

Ribonucleic acid was labeled and hybridized to a genome-wide microarray Human Gene 1.0 ST (Affymetrix, Santa Clara, CA, USA) at the Copenhagen University Hospital Microarray Center according to Affymetrix protocols. The microarray data was used as screening tool and performed once for each UM cell line. All microarray data are Minimum Information about a Microarray Experiments (MIAME) compliant, and raw data have been submitted to the Gene Expression Omnibus (GEO, accession number GSE55983). 

Multiplex Protein Determination

Protein concentration of CXCL8, CXCL9, CXCL10, CCL2, CCL5, VEGF, GM-CSF, and ICAM1 was quantified using the BD Cytometric Bead Array (CBA; BD Biosciences) according to manufacturer's instructions. Briefly, we first performed a preliminary experiment to determine the optimal dilutions for each sample type. Supernatants from untreated UM cells were assayed at 4×, activated T cells at 4× and 20× dilutions, and coculture supernatants at 10×, 30×, 100×, 150×, and 200× dilutions. For each protein, the dilution that yielded a concentration within the standard curve was used. The data was analyzed using CBA software supplied by BD Bioscience. The assay was performed with duplicates on three to four independent replicates. 

Monocyte Purification and Migration Assay

Monocytes were purified from peripheral blood mononuclear cells from healthy young volunteers by CD14+ selection using MACS cell separation (Miltenyi Biotec, Bergisch Gladbach, Germany). Monocyte purity was checked with a FITC-labeled CD14 antibody (BioLegend) and found to be above 95% (Supplementary Material S3). For the transwell migration assay 100,000 cells were placed in a 5.0-μm pore size polycarbonate transwell insert (Nalgene Nunc, Naperville, IL, USA). For studies with the neutralizing antibody, supernatants were preincubated for 3.5 hour at room temperature with 12 μg/mL mouse IgG2a antihuman IFN-γ antibodies (R&D System, Minneapolis, MN, USA) before adding to the lower chamber. We found no significant effect of the isotype control (ExBio Praha a.s., Vestec, Czech Republic) on monocyte migration. The plates were incubated for 2.5 hours at 37°C. The filters were removed and the migrated monocytes were detached using PBS containing 20 mM EDTA and 2% BSA (Sigma-Aldrich Corp., St. Louis, MO, USA). The number of migrated cells was counted on a LSR II flow cytometer (BD Biosciences) and analyzed with FlowJo version 7.6.5 for Windows (Tree Star, Ashland, OR, USA). For the positive control we used 100% FCS. In order to assess the effect of IFN-γ on cell migration we added 50 ng/mL recombinant IFN-γ (R&D Systems, Minneapolis, MN, USA) to X-VIVO 15 culture medium with 10% FCS. The assay was performed in duplicates on four independent cocultures and with two to three different monocyte donors. 

Statistics

Statistical analysis was performed using the graphical and statistical software (GraphPad Prism version 4.03; GraphPad Software, La Jolla, CA, USA), and P values less than 0.05 were considered significant. The migration assay and the multiplex protein determination were analyzed with one-way ANOVA and Turkey's multiple comparison test. The singleplex ELISA was analyzed with a t-test. 

Results

Activated T Cells Express IFN-γ

Activated T cells express an array of cytokines. In order to verify that the T cells were fully stimulated following incubation with CD3/CD28 beads, we quantified the inflammatory cytokine IFN-γ in the supernatants of the UM cell lines grown alone, cocultured with activated T cells and T cells grown alone. The result showed that activated T cells express IFN-γ and that coculture of UM cell lines with T cells led to an increased expression of IFN-γ by T cells. None of the supernatants of the three unstimulated UM cell lines contained IFN-γ constitutively (Fig. 1). 

Figure 1

 

Interferon-γ production following co-incubation of UM cell lines with activated T cells. Interferon-γ protein concentration was determined in supernatants from activated T cells grown alone or in coculture with UM cell lines. Uveal melanoma cell lines (92.1, Mel 270, and Mel 290) and activated T cells (T+) were cultured separately or in coculture with activated T cells in a membrane insert. *P < 0.05 and **P < 0.01.

UM Cell Lines Upregulate Chemokine Gene Expression After Coculture With Activated T Cells

We wondered whether the presence of TAM might be due to the presence of cytokine-producing TIL in UM, and tested this in a coculture model. In order to determine whether the UM cell lines 92.1, Mel 270, and Mel 290 change their gene expression in response to activated T cells, we cultured the UM cell lines alone or with activated T cells in a membrane insert. After 64 hours, gene expression was analyzed with a whole-transcriptome microarray. Using a greater than 2-fold criterion and an expression level above the median, more than 350 genes were found to be upregulated and more than 200 genes were found to be downregulated in each UM cell line in response to coculture with activated T cells. Based on the purpose of this study, we focused on genes encoding chemokines and other proteins involved in leucocyte migration such as ICAM1, VEGF-A, and GM-CSF. 

Nine chemotactic genes were upregulated by 92.1 after coculture with activated T cells, of which the chemokines CXCL9, CXCL10, and CXCL11 were most inducible and found to be more than 150-fold upregulated (Fig. 2A). 

Figure 2

 

Quantitative analysis of gene expression of chemotactic factors by UM cell lines. Uveal melanoma cell lines were grown alone (white bars) or cocultured with activated T cells (black bars) in a membrane insert. After 64 hours, RNA was harvested and analyzed with a Human Gene 1.0 ST Array. (A) 92.1 upregulated nine genes, (B) Mel 270 upregulated 11 genes, and (C) Mel 290 upregulated 14 genes. Results are shown as expression values. (D) Schematic representation of upregulated genes in UM cell lines 92.1, Mel 270, and Mel 290. n = 1. Arbitrary unit (a.u.)

The UM cell line Mel 270 upregulated 11 chemotactic genes after coculture with activated T cells, of which the chemokines CCL5, CXCL10, and CXCL11 were most inducible and found to be more than 80-fold upregulated (Fig. 2B). 

In Mel 290 cells, coculture resulted in an upregulation of 14 chemotactic genes, of which the chemokines CCL5, CXCL10, and CXCL11 were most inducible and found to be more than 100-fold upregulated (Fig. 2C). 

Comparing the three different UM cell lines with each other, seven genes were commonly upregulated namely CCL2, CCL5, CXCL8, CXCL9, CXCL10, CXCL11, and ICAM1 (Fig. 2D). 

Increased Protein Expression of Chemokines After Coculture With Activated T Cells

To validate that the gene upregulation translated into secreted proteins, we used a cytometric bead array to quantify the protein concentration of CXCL8, CXCL9, CXCL10, CXCL11, CCL2, CCL5, VEGF, ICAM1, and GM-CSF in supernatants (Fig. 3). All three UM cell lines expressed VEGF and ICAM1 constitutively, while activated T cells alone secreted CCL5, GM-CSF, and ICAM1. 

Figure 3

 

Chemokine production following coculture of UM cell lines with activated T cells. A quantitative protein expression analysis was performed by cytometric bead array for CXCL8, CXCL9, CXCL10, CXCL11, CCL2, CCL5, VEGF, GM-CSF, and ICAM1. Uveal melanoma cell lines were grown alone (white bars), cocultured with activated T cells in a membrane insert (black bars), and activated T cells were grown alone (gray bars). After 64 hours, supernatants were collected and analyzed. The cell lines only expressed VEGF and ICAM1 constitutively. Coculture with activated T cells in a membrane insert resulted in an upregulation of all the measured proteins, except for 92.1, in which VEGF was slightly downregulated. Activated T cells expressed CCL5, GM-CSF, and ICAM1. n = 4 for 92.1 and Mel 270 and n = 3 for Mel 290 and T+. Error bars are SEM. **P < 0.01 and ***P < 0.001.

In the UM cell line 92.1, the chemokines CXCL9, CXCL10, CXCL11, CCL5, GM-CSF, and the adhesion molecule ICAM1 were significantly upregulated after coculture with activated T cells. CXCL8 and CCL2 were also upregulated, however this increase did not reach significance. Vascular endothelial growth factor was slightly downregulated after coculture. 

In Mel 270, CXCL8, CXCL10, CCL2, CCL5, VEGF, and GM-CSF were significantly upregulated, and in Mel 290, CXCL8, CCL2, CCL5, VEGF, and GM-CSF were significantly upregulated. 

Increased Chemokine Expression Leads to Increased Monocyte Migration

Because we could measure an increased secretion of chemokines involved in monocyte migration, such as CCL2, CCL5, and CXCL8 after coculture with activated T cells, we tested the different supernatants in a transwell migration assay. Figure 4 shows that supernatants from all three unstimulated UM cell lines attracted monocytes and the migration increased significantly in response to coculture with activated T cells. The UM cell line Mel 270 showed the highest increase of monocyte migration after coculture with activated T cells (3.7-fold), followed by 92.1 with an induction of 2.4-fold, and Mel 290 with a 1.5-fold induction. Because it is reported that IFN-γ can inhibit monocyte migration ³⁴ and activated T cells express IFN-γ, we neutralized this inflammatory cytokine in the supernatants of UM cell lines cocultured with activated T cells. Although Mel 270 cells have a GNAQ mutation ³⁵ and express the melanoma marker MelanA, ³⁶ whereas Mel 290 cells neither have a GNAQ nor a GNA11 ³⁵ mutation and do not express MelanA, ³⁶ they secreted a similar concentration of the above measured proteins when stimulated with activated T cells. Therefore, we chose to neutralize IFN-γ for Mel 290 and 92.1. We found that monocyte migration was significantly increased (1.5-fold for 92.1 and 2.1-fold for Mel 290) when neutralizing IFN-γ in supernatants after coculture with activated T cells (Fig. 4). 

Figure 4

 

Influence of blocking IFN-γ on monocyte migration. Supernatants from UM cell lines cocultured with activated T cells attracted more monocytes compared to UM cell lines alone. Interferon-γ exerted an inhibitory effect on monocyte migration. Blocking IFN-γ resulted in increased monocyte migration. Results are shown as percentage of migrated monocytes compared to 100% FCS (positive control). The experiment was performed in four independent UM cultures, three independent monocyte donors; for studies with anti–IFN-γ two independent monocyte donors were used. Error bars are SEM. *P < 0.05, **P < 0.01, and ***P < 0.001. X-VIVO 15 (serum-free culture medium).

Discussion

The purpose of this study was to investigate whether activated T cells were able to modify UM cells so that they would produce cytokines that attract monocytes. Therefore, we studied whether cytokines derived from CD3/CD28-activated T cells have an effect on UM cell lines through a membrane insert. Further, we tested the effect of UM supernatants on monocyte migration. 

The data from the microarray analysis showed that the UM cell lines 92.1, Mel 270, and Mel 290 upregulated over 350 genes in response to inflammatory stress. Based on several reports where an increased infiltration of T cells ^6,7 and M2 macrophages ^13,14 are observed in UM and associated with a worse prognosis, we chose to focus in this study on soluble factors, which play a role in leucocyte attraction. We found that coculture with activated T cells stimulate UM cells to secrete CXCL8, CXCL9, CXCL10, CXCL11, CCL2, CCL5, VEGF, GM-CSF, and ICAM1, of which only VEGF and ICAM1 are being produced by the cell lines without stimulation (Fig. 3). 

Our results suggest that the tumor tissue itself contributes to the synthesis of inflammation-related chemokines such as CCL2, CCL5, CXCL8, CXCL10, and VEGF, which are found in the vitreous fluid of eyes with uveal melanoma. ¹⁸ Additionally, we found that supernatants from UM cell lines cocultured with activated T cells are able to attract more monocytes compared with supernatants from unstimulated UM cell lines (Fig. 4). This shows that UM cells are actively involved in recruiting monocytes. We also showed that the UM cell lines secrete the adhesion molecule ICAM1, which facilitates the influx of monocyte into the tissue and it has been shown that ICAM1 is expressed in the aqueous fluid of UM eyes. ¹⁷  

Our findings support the assumption that infiltrating T cells might modulate the in vivo production of several chemokines such as CXCL8, CXCL10, CXCL11, and CCL2 by UM, through the production of T cell–derived cytokines. This inflammatory microenvironment created by the T cells and UM, can contribute to the attraction of monocytes. Since the chemokine system is a pleiotropic system, the secreted chemokines by UM might not only attract monocytes but also in addition more T cells, which in turn creates a vicious cycle leading to a tumor-promoting inflammatory microenvironment. It seems likely that the ability of the UM cell lines to attract monocytes is specific to UM, since supernatant from RPE cells that have been stimulated with activated T cells and also secrete a number of chemokines ³³ does not lead to an increased monocyte attraction (unpublished data). 

Once the monocytes have migrated into the eye, IFN-γ might trap the macrophages at the site of inflammation since this inflammatory cytokine is able to inhibit monocyte migration through the blocking of actin polymerization and polarization. ³⁴ Indeed, blocking of IFN-γ in our in vitro system further enhanced the migration of monocytes toward the supernatants (Fig. 4). 

The majority of macrophages found in UM belong to the pro-angiogenic and immune suppressive M2 type, ^13,14 and therefore favor tumor growth. ³⁷ The chemokines CCL2 and CXCL10 can make monocytes differentiate into an M2 phenotype, ³⁸ while the T cell–derived cytokine GM-CSF not only increased the migration of monocytes, ^39,40 but in combination with IFN-γ can also stimulate macrophages toward an M1 phenotype, which has an antitumor effect. ⁴¹ Several studies have described the interaction between melanoma cells and monocytes in high detail. Wang et al. ⁴² investigated the crosstalk between skin melanoma cells and monocytes where they found that tumor supernatant from cell cultures is able to differentiate monocytes into a mixed population of M1/M2 macrophages. Furthermore, it has been shown that macrophage colony-stimulating factor (M-CSF) expressed by melanoma cells induce expression of VEGF-A in monocytes and conditioned medium from monocyte-derived macrophages can also enhance the secretion of VEGF-A by melanoma cells. ⁴³ A recent study ²⁷ has shown that UM cell lines cultured under hypoxic conditions did not differentiate monocytes into M2 macrophages. Therefore, the interaction between monocytes and UM cells needs to be further elucidated and how the inflamed UM tumor microenvironment might favor a differentiation of monocytes to M2 macrophages. 

Several chemokines are associated with cancer biology, and play a role in cancer cell survival, regulate angiogenesis, and influence metastasis. ⁴⁴ For example, CXCL8 has been shown to promote angiogenesis, whereas the chemokines CXCL9, CXCL10, and CXCL11 show angiostatic activities in a variety of tumors and promote tumor inhibition. ⁴⁵ CXCL10 can have a proliferative but also an antiproliferative effect on tumors. ⁴⁶ This illustrates the complex network of chemokines and their pleiotropic effects in cancer. 

From our results, we conclude that under inflammatory conditions UM cells shift their transcriptome toward a more inflammatory state, which results in an increased secretion of chemokines, leading to an increased attraction of monocytes. Therefore, UM cells exposed to inflammatory stress might actively participate in maintaining a tumor-promoting inflammatory microenvironment, known as the inflammatory phenotype. ⁴⁷ This phenotype also includes an increased level of HLA-A and -B antigen expression, which may be related to the production of IFN-γ by TIL. As IFN-γ at the same time stimulates the expression of PD-L1 and makes UM cells resistant to cytotoxic T cell–mediated lysis, the presence of an inflammatory infiltrate may be a sign that there may be a role for immunomodulation in the treatment of UM prior to trying immunotherapy. ²³ Further research needs to be done to understand the complex network of the tumor microenvironment and how all the different cell types play together before immunotherapy may be a successful treatment option. 

Supplementary Materials

pdf S1 

pdf S2 

pdf S3 

Acknowledgments

The authors thank Luise Bruen Hemmingsen for excellent technical assistance. 

Presented at the annual meeting of the Association for Research in Vision and Ophthalmology (ARVO), Seattle, Washington, United States, 2013. The data was awarded with the ARVO-NED ARVO travel award. 

Supported by grants from Værn om Synet and the Graduate School of Health and Medical Sciences, University of Copenhagen (Copenhagen, Denmark) and a grant of the Board of Directors of the Leiden University Medical Center (IHGB; Leiden, The Netherlands). 

Disclosure: T. Jehs, None; C. Faber, None; H.B. Juel, None; I.H.G. Bronkhorst, None; M.J. Jager, None; M.H. Nissen, None