RNA was prepared from 8 primary UM cultures grown under hypoxic or normoxic conditions for 24 hours and subjected to quantitative polymerase chain reaction analysis. The ordinate shows fold difference in transcript level in hypoxic culture relative to the normoxic culture reference (reference = 1). Asterisks indicate P < .05 derived from paired t tests of ΔCT values comparing hypoxic and normoxic cultures. Error bars indicate SEM.
Hypoxia stimulates vascular endothelial growth factor (VEGF) production (A) but not CCL2 (B) and IL6 (C). Under normoxic conditions, CCL2 and IL6 (P = .11) expression seems higher in M3 tumors, whereas VEGF (P = .88) showed no difference (Mann-Whitney test). IL6 was significantly increased in M3 tumors under hypoxic conditions (P = .03). Asterisk indicates P < .05.
Monocytes were added in an insert, and TSN was added underneath the insert. After 16 hours, the number of migrated cells was counted with flow cytometry. A mixture of CCL2 and lipopolysaccharide (LPS) served as a positive control. The TSN from all 5 cell lines had a migration-inducing capacity, which was independent of normoxic or hypoxic conditions.
The presence of CCL2 was determined in tumor supernatants (TSNs) from the uveal melanoma cell lines OCM8 (A) and OMM2.5 (B) grown under hypoxic and normoxic conditions for 24, 48, 72, and 96 hours. The CCL2 concentrations in the TSN of cell line 92.1 and Mel270 were less than 5 pg/mL. The TSN of cell line Mel290 only showed a low level of CCL2 after 96 hours (not shown). C, Primary uveal melanoma cell cultures 10-019 and 12-009 and the lung adenocarcinoma cell culture 10-015 express CCL2.
Peripheral blood monocytes were cultured for 6 days with granulocyte-macrophage colony-stimulating factor and interleukin 4 in 20% control medium or in culture medium with 20% tumor supernatant (TSN) of the indicated cultures or cell lines. Cells were analyzed for the expression of the monocyte-derived dendritic cell (mo-DC) marker CD1, macrophages and DC marker CD206, the macrophage marker CD14, and M2 marker CD163. The TSN of CC8 and of the lung adenocarcinoma cell line 10.015 were able to skew mo-DCs toward CD14+ macrophages with a high CD163 expression, whereas the TSN from the primary UM cell cultures (10-019 and 12-009) or the UM cell line OCM8 (under either normoxic or hypoxic conditions) inhibited DC maturation (low CD1a) but did not affect macrophage differentiation.
Bronkhorst IHG, Jehs TML, Dijkgraaf EM, Luyten GPM, van der Velden PA, van der Burg SH, Jager MJ. Effect of Hypoxic Stress on Migration and Characteristics of Monocytes in Uveal Melanoma. JAMA Ophthalmol. 2014;132(5):614-621. doi:10.1001/jamaophthalmol.2014.43
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Among the characteristics of uveal melanoma that are associated with a poor prognosis are a large tumor size and the presence of increased numbers of lymphocytes and macrophages. In rapidly growing tumors, reduction in oxygen tension may occur with increased distance from blood vessels, which we hypothesize may lead to an inflammatory microenvironment, further stimulating tumor growth.
To analyze whether hypoxia induces uveal melanoma cells to express proinflammatory cytokines and whether tumor supernatant (TSN) affects monocyte migration and differentiation.
Design and Setting
The expression of proinflammatory genes in freshly cultured uveal melanoma samples was studied in an in vitro 24-hour hypoxic culture system using quantitative polymerase chain reaction. In addition, cell lines cultured under normoxic and hypoxic conditions were used. The effect of TSN on monocyte chemotaxis was tested using a transwell migration system and by analyzing monocyte differentiation. The levels of the cytokines CCL2, IL6, and PGE2 in TSN were determined by enzyme-linked immunosorbent assay.
Five cell lines (OCM8, 92.1, Mel270,Mel290 and OMM2.5) and 11 primary short-term cultures.
Exposure of freshly cultured uveal melanoma cells to hypoxia led to an increased expression of the proinflammatory cytokines PLGF (OMIM 601121), TGFβ (OMIM 190180), END1 (OMIM +131240), and ICAM1 (OMIM 147840) and a lower expression of AIMP1 (OMIM 603605) (EMAP2), CCL2 (MCP-1) (OMIM +158105), and IL1b (OMIM *147720). The TSN from cultured melanoma cell lines induced chemotaxis of monocytes, but this was independent of the normoxic or hypoxic state. The TSN of 1 cell line and 2 primary uveal melanoma cultures inhibited the dendritic cell maturation and did not induce M2 macrophage polarization in vitro.
Conclusions and Relevance
Our results indicate that under hypoxic conditions, immune response genes are differentially expressed in cultured primary uveal melanoma cells. The TSN from uveal melanoma cell lines is capable of affecting the chemotactic response and maturation of monocytes in vitro, but this is irrespective of hypoxia.
Under physiologic conditions, tissue is characterized by normoxia. Tissue oxygenation is severely disturbed during pathologic conditions, such as cancer, which is associated with a local decrease in Po2 (ie, hypoxia).1 A hypoxic environment influences the formation of metastases (eg, by creating selective pressure and by promoting outgrowth of cells that can circumvent the oxygenation restrictions). Hypoxia induces stabilization of hypoxia-inducible factor (HIF) 1α, which upregulates cell matrix adhesion, invasion, and tumor angiogenesis via a variety of mechanisms. In addition, hypoxic changes of the tumor cells themselves may induce expression of specific patterns of genes, which confer a survival advantage on cancer cells.2 Consequently, tumor hypoxia functions as a key mediator of tumor progression by promoting the expansion of cells with a more aggressive phenotype; it may also influence therapy efficiency.
Zimmerman et al3 hypothesized that all melanomas have a relatively uniform slow growth rate before enucleation. Subsequently, this hypothesis was discussed as each melanoma appeared to grow at a constant rate that varied widely among different melanomas.4 Interestingly, most large tumors and epithelioid and mixed-cell melanomas demonstrated fast growth rates. Burnier et al5 were interested in tumor vascularization by analyzing the relationship of proliferating cells to blood vessels in retinoblastoma; however, they did not explore this in uveal melanoma (UM) because this field has only recently emerged. HIF-1 has been reported to play a critical role in UM progression by increasing the expression of a number of target genes involved in invasion.6 Moreover, HIF-1 has been identified as one of the most critical biomarkers that can predict UM metastasis.7 In UM, the genetic progression from 1A-1B-2A to 2B probably represents an adaptation to multiple selective pressures, including hypoxia,7,8 immune responses9,10 and other factors.11 When the HIF-1α protein is stabilized and translocated to the nucleus, it induces transcription of target genes involved in oxygen delivery and energy metabolism. HIF-1α was found to be constitutively stabilized and active in at least 50% of human UM cases.12 This can happen in hypoxic conditions but also under influence of other environmental stress, such as inflammation.
Inflammation is an important characteristic of malignancy, which includes the presence of tumor-associated macrophages (TAMs) and lymphocytes.13 Tumor cells produce chemotactic cytokines and growth factors that recruit peripheral blood monocytes into the tissue, where they subsequently differentiate into tissue macrophages. Chemoattractants that accomplish this recruitment include colony-stimulating factor 1 (CSF-1, also known as macrophage colony-stimulating factor), monocyte chemotactic protein 1 (MCP-1, also known as CCL2),14 CCL5 (also known as RANTES ), CCL7, CCL8, CXCL12 (also known as SDF-1), platelet-derived growth factor, and vascular endothelial growth factor (VEGF).15 Macrophages can differentiate into cells with a proinflammatory (M1) phenotype or into the M2 phenotype that is oriented toward tumor growth, angiogenesis, and immunosuppression. M2 macrophages produce anti-inflammatory cytokines, including interleukin (IL) 10 and TGF-β (OMIM 190180),15 as well as molecules associated with increased angiogenesis and metastasis, such as VEGF.16 In UM, an increased density of TAMs correlates with tumor size, microvessel density, and a poor prognosis.17 Moreover, this study found that almost all macrophages in UM belong to the M2 subtype.18
Monocytes tend to migrate to hypoxic areas because hypoxia stimulates the production of cytokines, such as VEGF, that attract macrophages.19 In UM, larger tumor size is associated with a higher risk of metastases,20 and we hypothesize that the exponential growth of UM will induce ischemia and cytokine production and lead to the influx of inflammatory cells. It is already known that when UM cells are cultured in a hypoxic environment, production of VEGF is increased through induction of the HIF-1α pathway.21 We determined whether hypoxia influenced the expression of macrophage-specific chemokines in primary UM and UM cell lines and whether hypoxia-derived cytokines help macrophages to acquire a polarized M2 phenotype.
Fresh tissue from 11 tumors, obtained immediately after enucleation, was placed in AmnioChrome Pro medium (Lonza Group Ltd) to develop a primary cell culture. Primary tumor cells were cultured for a maximum of 4 passes, providing a pure tumor cell population. Characteristics of the primary tumor are given in the Table. Tumor cells were cultured under normoxic or hypoxic conditions for 24 hours, after which time the expression of selected inflammation- and hypoxia-related genes were analyzed. Institutional review board approval was waived by the Medical Ethics Committee of the Leiden University Medical Center.
Chromosome 3 status was determined by standard cytogenetic testing (ie, karyotyping and fluorescence in situ hybridization on cultured cells). When either test on cultured cells showed monosomy 3, the tumor was categorized as having monosomy 3.
Using an RNeasy Mini Kit (Qiagen), we extracted RNA from primary cultures. Complementary DNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad). Quantitative polymerase chain reaction (qPCR) was performed in duplicate on selected genes that are associated with inflammation. GAPDH (OMIM 138400), β-actin (OMIM *102630), RPL13 (OMIM 113703), and RPS11 (OMIM *180471) were initially included for selecting suitable reference genes. Primers were designed with Beacon Designer (Biosoft). The qPCR was performed according to our standard laboratory protocol, as described previously.22 By using CFX-384, it was possible to perform qPCR synchronically on several genes. The PCR reaction settings were 95°C for 3 minutes, 40 cycles containing 96°C for 10 seconds, and 60°C for 30 seconds. The results were subsequently validated with the iQ5 Bio-Rad system (Bio-Rad). Calculation of the gene expression was as follows: the CT value of each sample obtained from qPCR was normalized to the reference genes (genes that are stably expressed in the tissue). Because 2 genes were stably expressed (BACT and RPS11, as determined with the geNorm software23), the gene of interest was corrected for the geometric mean of these reference genes according to the method used by Vandesompele et al.24 The calculated values were the normalized values of each sample.
Tumor supernatant (TSN) of the UM cell lines (OCM8, 92.1, Mel 270, Mel 290, and OMM2.5), primary UM cell cultures (10.019 and 12.009), and a lung adenocarcinoma culture (metastasis in the eye; 10.015) was collected. Tumor cell lines were grown in RPMI 1640 supplemented with 10% fetal calf serum, penicillin-streptomycin, and l-glutamine. The UM cell lines were grown in flasks at 80% to 90% confluency and harvested with trypsin and EDTA. The primary cell cultures were grown in T-flasks. For the migration assay, 1 × 106 cells were plated in T75 flasks in 6 mL of medium. Supernatant was collected after 24, 48, 72, and 96 hours and was stored at −20°C. For the differentiation assay, 100 000 cells of OCM8, 150 000 cells of 92.1, and 250 000 cells of Mel 270 were plated in 2 mL per well of a 6-well culture plate and cultured for 4 days. Supernatant was stored at −80°C.
The protocol for the migration assay was provided by A. M. van der Does, PhD (Departments of Physiology and Pharmacology and Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm, Sweden). Monocyte migration was measured with a modification of the Boyden method using a 96-well microchemotaxis Transwell system with a membrane of 8-µm pore size (Corning HTS Transwell 96-well permeable support). CD14 purified monocytes were added in the insert, and in the lower chamber we added the TSN of interest. CD14 purified monocytes were resuspended in medium and placed in the upper compartment (1 × 105 cells in 100 µL). The TSN, medium (negative control), and medium with MCP-1 and lipopolysaccharide (positive control) (100 μL; 100 ng/mL of CCL2 and 200 ng/mL of lipopolysaccharide) were added to the lower compartments and subsequently incubated for 16 hours at 37°C. The next day, we removed the 96-well filter rack and pipetted the liquid from the lower wells into small fluorescence-activated cell sorting tubes. The wells were rinsed with 100 μL of 5 mM EDTA, which was added to the tubes. Migratory responses were quantified by counting monocytes obtained from the wells with flow cytometry (30 seconds).
The influence of tumor-produced cytokines on macrophage differentiation was determined as described previously.25 With this in vitro test, it is possible to show a differentiation into M2 macrophages. In short, to establish a monocyte-derived dendritic cell (mo-DC) culture, peripheral blood mononuclear cells were obtained from buffy coats of healthy donors. CD14+ monocytes (95% purity) were isolated using magnetic-activated cell sorted cell separation and stored in liquid nitrogen until further use. Monocytes were thawed and cultured in 24-well plates in a density of 0.5 million cells per well in the presence of IL-4 (500 U/mL), granulocyte-macrophage colony-stimulating factor (GM-CSF) (800 U/mL) (mo-DC), and 20% TSN. The 20% TSN of the cervical cancer cell line CC-8, which is known to induce M2 macrophages, was used as a positive control. After 3 days, fresh medium with cytokines and TSN was added. At day 6, the cells were stained and analyzed for differentiation by flow cytometry.
Mouse mAbs to human CD14 (PE), CD1a (APC), CD206 (FITC), and CD163 (APC) were used. Cells were recorded (20 000 per live gate) using a BD FACSCalibur with Cell Quest software (BD Biosciences) and analyzed by FlowJo software (TreeStar Inc).
CCL2, IL6, and PGE2 production in the TSN were measured with a commercially available enzyme-linked immunosorbent assay kit (eBioscience).
We determined the effect of hypoxia on 8 primary UM cultures using qPCR and observed a similar gene response in all 8 cultures, with upregulation of VEGF, the positive control GLUT1, and the other proinflammatory genes PLGF (OMIM 601121),TGFβ, END1 (OMIM +131240), and ICAM1 (OMIM 147840). In contrast, AIMP1 (OMIM 603605) (endothelial monocyte-activating polypeptide 2), CCL2 (MCP-1) (OMIM +158105), and IL1b (OMIM *147720) were significantly downregulated (paired t tests) (Figure 1).
When comparing the inflammatory gene expression of the cultures of 4 disomy and 4 monosomy 3 tumors, CCL2 and IL6 (both P = .11) expression tended to be higher in monosomy 3 tumors, whereas all others, such as VEGF (P = .88), showed no difference (Mann-Whitney test) (Figure 2). Of the 2 other primary UM cultures for which TSN was used in further experiments, 10-019 was the disomy tumor and 12-009 the monosomy 3 tumor.
Because we hypothesized that hypoxia would stimulate the production of monocyte-chemotactic factors, we determined the effect of TSN obtained from normoxic and hypoxic cell lines. The TSN derived from the UM cell lines and primary cultures stimulated the in vitro migration of peripheral blood–derived monocytes. However, we did not observe a difference in the effect of TSN from cells grown under either normoxic or hypoxic conditions (Figure 3).
We specifically analyzed CCL2 (MCP-1) because this is one of the best recruiters of monocytes. The cultured UM cell lines OCM8, OMM2.5, and Mel 290 expressed CCL2 in a time-dependent manner with a maximum peak after 96 hours in culture, but 92.1 and Mel270 did not. The concentration of CCL2 was lower when the cells were grown under hypoxia than normoxia (Figure 4A and B). The primary cultures expressed CCL2 (Figure 4C).
We analyzed whether TSN from UM cells was able to influence the in vitro differentiation of mo-DCs into M2 macrophages. The mo-DCs were obtained by culturing peripheral blood mononuclear cells from healthy donors with GM-CSF and IL-4 and are characterized by a high expression of CD1a and a low expression of CD14. We used the TSNs of cell line CC-8 and of lung adenocarcinoma cell culture 10.015 as positive controls because these inhibit DC differentiation (inhibition of CD1) and stimulate M2 formation (characterized by high CD14 and high CD163 expression). The TSN from the 2 primary cultures and cell line OCM8 inhibited the further differentiation of DCs (very low expression of CD1), whereas none of the primary cultures or UM cell lines (under either hypoxic or normoxic conditions) were able to stimulate differentiation of monocytes into CD14+ macrophages (Figure 5).
Our tested cell lines did not produce IL6 or PGE2. Primary UM culture 10.019 did not express IL6 or PGE2, but 12.009 and the lung adenocarcinoma 10.015 did. This finding was verified on messenger RNA levels: IL6 had extremely low expression in all cell lines tested (data not shown) but was detectable in primary cultures (Figure 2).
Part of the inflammatory response seen in malignant UM involves the massive migration of immune cells from the circulation to the site of the tumor, which may be due to local hypoxia. We hypothesize that, under hypoxic conditions, UM acquires an inflammatory phenotype and activates immune gene expression by its tumor cells. Cytokine production by tumor cells may subsequently affect the differentiation of peripheral blood monocytes inside the tumor into immunosuppressive and proangiogenic M2 macrophages. If this would indeed be the case, it will be interesting to see whether targeting hypoxia-dependent signaling pathways26 will enable a clinically significant reduction in the protumoral inflammatory response in UM.
Eltzschig and Carmeliet27 discuss the link between hypoxia and inflammation and its implications in cancer. The link between hypoxia and inflammation is regulated by HIF, which might interact with the nfκB (nuclear factor κB transcription factor) that regulates inflammation. We hypothesized that hypoxia is essential for infiltration and activation in vivo. The master transcription factor HIF-1α binds to hypoxia response elements in the promoter of its target genes to activate their expression.28 One of these targets is VEGF, which is increased under the influence of hypoxia in vitro21 and overexpressed in 20 of 60 cases (33%) of primary UM.12 Besides affecting VEGF, hypoxia-induced changes in gene expression in UM cells may help to increase the production of inflammatory signaling molecules. Hypoxia led to upregulation and downregulation of inflammatory genes in primary UM cultures, with VEGF as the most differentially expressed molecule. Because the VHL gene (an important regulator of VEGF) is located on chromosome 3, we compared changes in expression with the tumor’s chromosome 3 status: IL6 and CCL2 had a differential expression.
We preferred to use primary UM cultures because they probably represent the situation better than do cell lines. Besides chromosome 3 status, size of original tumor, presence of extracellular matrix patterns (vascular loops and networks), and cell type might have an influence on tumor behavior because they may influence hypoxia. However, because the tissues came from enucleated eyes, tumor size was large overall. Usually, the cell type was mixed, and vascular loops and networks were seen in 4 of 5 monosomy 3 tumors (Table).
Because ischemia affected a variety of cytokines and chemokines, we wondered what the overall effect of TSN on monocyte migration would be. Using a monocyte migration assay, we found that TSN from primary cultures and cell lines stimulated monocyte migration. To our surprise, we did not find a difference between TSN from cells cultured under hypoxic or normoxic conditions.
Macrophage phenotype can vary in different areas of a tumor. In a mammary adenocarcinoma model, TAMs with high expression of major histocompatibility complex class II can localize to normoxic tumor tissues and express M1 markers and antiangiogenic chemokines, whereas TAMs with low expression of major histocompatibility complex class II were found in hypoxic tumor tissues, preferentially expressed M2 markers, and had greater proangiogenic functions.29 Most macrophages in UM are of the M2 type. It was found that TSN of cervical cancer cell lines can skew mo-DCs into an M2 phenotype and that this process is dependent on IL6 and PGE2.25 With use of the differentiation assay, none of the TSN obtained from the UM cell lines grown under normoxic and hypoxic conditions, or the primary cell cultures, could differentiate mo-DCs into macrophages. Consequently, there must be other soluble factors that play a role in UMs rather than in cervical cancer cell lines and lung adenocarcinomas.30 Another study31 found that UM cell lines did not express IL6 or PGE2, cytokines that are considered essential in the induction of M2 macrophages, and this correlates with our observation that we could not observe any M2 differentiation. We observed IL6 expression by primary cultures, especially in the monosomy 3 group. However, production of interferon γ may have hampered the differentiation of monocytes into M2 macrophages because it is known to induce M1 macrophages.32 Therefore, it could be that, in our TSN, the presence of interferon γ overruled the effect of IL6 and PGE2.
Although we have linked the macrophage phenotype to the cytokine milieu in which the macrophages reside, we have to keep in mind that the type of receptor interaction on the macrophage may also play a role in the polarization of these cells into M1 or M2 macrophages. Further research has to be performed to analyze the behavior and interactions of macrophages with tumor cells in vivo that can lead to further understanding of direct tumor cell–macrophage interaction.
Culturing primary UM under hypoxia resulted in an altered balance in cytokine expression. CD163 M2 macrophages are detected in situ, and although TSN, whether from normoxic or hypoxic cell lines, did not differentiate monocytes toward M2-like macrophages in vitro, TSN from one cell line and from both tested primary UM cultures was able to inhibit DC maturation. Furthermore, all TSN stimulated macrophage migration. It seems that the generation of an inflammatory environment around the tumor, which corresponds to a worse prognosis, is not oxygenation dependent but a primary characteristic of UM cells.
Submitted for Publication: September 4, 2013; final revision received December 11, 2013; accepted December 23, 2013.
Corresponding Author: Martine J. Jager, MD, PhD, Department of Ophthalmology, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, the Netherlands (M.J.Jager@lumc.nl).
Published Online: March 13, 2014. doi:10.1001/jamaophthalmol.2014.43.
Author Contributions: Dr Bronkhorst had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bronkhorst, Jehs, Dijkgraaf, van der Burg, Jager.
Acquisition, analysis, or interpretation of data: Bronkhorst, Jehs, Dijkgraaf, Luyten, van der Velden, Jager.
Drafting of the manuscript: Bronkhorst, Dijkgraaf, van der Burg, Jager.
Critical revision of the manuscript for important intellectual content: Bronkhorst, Jehs, van der Velden, van der Burg, Jager.
Statistical analysis: Bronkhorst.
Obtained funding: Jager, Luyten.
Administrative, technical, or material support: Dijkgraaf, Luyten, Jager.
Study supervision: van der Velden, van der Burg, Jager.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was financially supported by the Board of Directors of the Leiden University Medical Centre (PhD position recipient: Dr Bronkhorst), Stichting Nederlands Oogheelkundig Onderzoek, Rotterdamse Stichting Blindenbelangen, and Landelijke Stichting voor Blinden en Slechtzienden.
Role of the Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.