Stimulation and Inhibition of Uveal Melanoma Invasion by HGF, GRO, IL-1α and TGF-β

Abstract

purpose. To investigate potential factors involved in uveal melanoma migration and invasion in vitro.

methods. Using a microchemotaxis chamber, the effects were studied of a range of stimulators and inhibitors on a series of 10 primary uveal melanomas and 2 uveal melanoma cell lines, by assessing invasion through an 8-μm pore membrane, precoated with an extracellular matrix solution. In addition, invasion in response to the effect of cells and conditioned media derived from the liver and other tissues was studied for one uveal melanoma culture, by using double-chambered wells, and invasion was assessed through an 8-μm pore membrane, precoated with synthetic extracellular matrix. In all instances, invading cells were counted under ×400 magnification on the lower surface of the membrane. Levels of invasion were correlated with histopathologic markers of prognosis.

results. Conditioned media and cells derived from other tissues, including the liver, increased cellular invasion of the uveal melanoma cell line studied. For specific regulators, maximum stimulation of invasion was induced by hepatic growth factor (HGF), growth-related oncogene (GRO), and macrophage inflammatory protein (MIP)-1β, whereas significant inhibition was induced by IL-1α, TGF-β1, and TGF-β2.

conclusions. The primary site of metastasis in patients with uveal melanoma is the liver. For the degree of site specificity commonly seen, regulators involved in the process may be expressed at the secondary sites, promoting adhesion, migration, invasion, and proliferation of tumor cells. HGF, GRO, MIP-1β, IL-1α, TGF-β1, and TGF-β2 may play a significant role in regulating invasion of uveal melanoma cells.

The development of metastatic disease is highly complex. Ocular tumor cells detach from the primary tumor mass; migrate; invade through the extracellular matrix, basement membrane, and endothelial cell wall; and circulate in the vascular system, before establishing themselves in the target organ, where they undergo the reverse process before proliferating into new foci. ¹ Abrogation of any of the processes prevents formation of secondary disease, and, accordingly, only a small percentage of tumor cells establish themselves as metastatic foci. ² Posterior uveal melanoma is the most common intraocular malignancy in adults, ³ and unlike cutaneous melanoma, uveal melanoma disseminates mainly through the hematogenous system and predominantly to the liver. ⁴ Metastatic liver disease is the leading cause of death and can develop after a long disease-free interval. ⁵ To date, little progress has been made in the detection and treatment of metastatic disease, and after detection of liver metastases, median survival is still less than 1 year. ^{6 7}  

Little is known about the factors that regulate growth and cellular invasion in uveal melanoma. Inflammatory responses and malignant progression are now thought to have much in common. ⁸ Many inflammatory mediators regulate directional migration and invasion of both leukocytes and tumor cells, and recent evidence has suggested a direct role for chemokines in guiding tumor cells to a specific secondary site. ⁹ Because metastatic uveal melanoma cells appear to target the liver in most cases, further investigation of factors that promote and inhibit migration of uveal melanoma cells is essential to our understanding of metastasis in uveal melanoma. 

Inflammatory cytokines and chemokines, produced by tumor and host cells, have been implicated in the metastatic progression of cutaneous melanoma. ^{10 11 12 13} The neutrophil-activating lymphokine, interleukin (IL)-8 can induce migration of cutaneous melanoma in vitro, ¹⁴ and expression of high levels of IL-8 by metastatic melanoma associates with upregulation of the transcriptional activity of matrix metalloproteinase (MMP)-2 ¹⁵ and increased cellular invasion through reconstituted basement membranes. ^{14 15} Other factors, including hepatocyte growth factor (HGF), produced by the liver, are known to be motogenic factors for several tumor types, ¹⁶ and both HGF and its receptor c-Met, have been linked with progression of cutaneous melanoma. 

Very few investigators have sought to determine the specific role of the factors that potentially regulate migration of uveal melanoma cells. TGF-β is thought to contribute to the immunosuppressive environment of the orbit ¹⁷ and is known to suppress the growth of melanocytes. This effect is lost in both cutaneous and ocular melanoma cells. ^{18 19} Evidence also suggests that interferon (IFN)-α and -γ are inhibitory to proliferation of uveal melanoma cells, by increasing HLA expression in uveal melanoma cell lines and effecting a host immune response, but immunotherapy with IFN-γ has had limited success in patients with uveal melanoma. ²⁰ Growth factors have also been shown to regulate cellular invasion in uveal melanoma—in particular, HGF and epidermal growth factor (EGF) have been shown to increase invasion and are possibly involved in the targeting of the liver. ^{21 22} More recently, expression of the EGF receptor (EGFR) has been correlated with death from metastatic disease. ²³  

To determine why uveal melanomas in particular target the liver, we undertook to investigate which factors may regulate migration and invasion of uveal melanoma cells. In addition to the effects of growth factors, cytokines, and chemokines, we also explored the roles that tissues themselves may play in the selective colonization of the liver, by studying the effect that cells derived from liver and other tissues may exert. 

Materials and Methods

Clinical Material

A series of 10 primary posterior uveal melanomas were cultured to produce short-term cultures (STCs), which were serially passaged and used within seven passages. Histopathologic details are recorded in Table 1 . Because of the constraints involved in culturing solid tumors, the study was restricted to 10 melanomas, from which we were able to obtain a sufficient number of cells for all aspects of the study. Two uveal melanoma cell lines, previously established in this laboratory, were also included in the study (Sheffield Ocular Melanoma [SOM]-157d and -196B). These cell lines were used to determine control levels of cellular invasion. SOM-196B is a relatively invasive cell line, whereas SOM-157d is virtually noninvasive. Fresh samples were obtained from primary posterior uveal melanomas at enucleation and processed as reported previously. ²⁴ Ethical approval for all samples was sought before collection, and protocols adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all patients before collection of tumor samples. STCs and cell lines were maintained by serial passage in RPMI-1640, supplemented with penicillin (100 U/mL), streptomycin (100 μg/mL), amphotericin B (5 μg/mL), EGF (0.2 μg/mL), fetal calf serum (20%), and glucose (0.2%) at 37°C in an atmosphere of 5% carbon dioxide and 95% air. Melanoma status was confirmed by immunohistochemistry for three melanoma-associated markers (Melan-A, S-100, and HMB45). 

Chemotaxis and Chemoinvasion Assays

Migration and invasion were assessed by methods previously described by Falk et al. ²⁵ and Albini et al. ²⁶  

Differential Tissue Stimulation of Uveal Melanoma Cellular Invasion Determined Using Chemoinvasion Assays in Double-Chambered Wells.

All assays used the control invasive cell line SOM-196B, because of an unlimited source of material. Chemoinvasion assays were performed in a modified Boyden chamber system. Cell types used as a source of viable cells and conditioned media (CM) included hepatoma cells (HepG2; European Collection of Cell Cultures [ECACC]), primary human liver endothelial cells (HULECs; the gift of Lance J. Burns, Department of Surgery and Anaesthetic Sciences, University of Sheffield, UK), lung carcinoma cells (A549; ECACC), primary human dermal fibroblasts (JFF; the gift of Declan Donovan, Department of Pathology, University of Sheffield, UK), and primary normal adult keratinocytes (NAK 3.2l; the gift of Stephen St. John-Smith, Department of Dermatology, University of Sheffield, UK). 

Briefly, to investigate the effects of different cell types on SOM-196B invasion, cells were grown to confluence (5 × 10⁵ cells) in appropriate media in a 24-well plate and were then used in assays in double-chambered wells (Transwell; Costar UK, Ltd., High Wycombe, UK). Before the assay, media were replaced with RPMI with 0.1% BSA. Filters with an 8-μm pore polycarbonate membrane (Costar UK, Ltd.), precoated with an artificial basement membrane (1.5 μg/mL, Matrigel; Becton Dickinson Labware, Bedford, MA), were placed in the 24-well plates, thus including the different cell types in the lower chamber as potential chemoattractants. SOM-196B cultures were dissociated from tissue culture flasks and resuspended in RPMI-1640 medium with 0.1% bovine serum albumin (BSA) and were added to the upper chamber (1 × 10⁵/well), and RPMI-1640 medium with 0.1% (BSA) was added to the lower chamber. Alternatively, to assess the effect of CM on SOM-196B invasion, CM was collected after incubating each cell type to be assessed with RPMI with 0.1% BSA for 24 hours. Before use, CM was filter sterilized (0.2-μm-pore filter) and added to the lower chambers of the test wells. Negative control experiments were conducted in which neither cells nor CM was added to the lower chamber. Assays were incubated at 37°C for 24 hours. Noninvading cells were removed from the upper chamber by gently wiping the upper surface of the membrane with a cotton swab. Membranes were fixed in ethanol and stained with Gill hematoxylin. Levels of invasion were assessed by counting the number of cells present in 10 fields on the lower surface of the membrane, under light microscope (×400). Each experiment was performed in triplicate and repeated three times. 

The Effects of Growth Factors, Cytokines, and Chemokines Determined in Boyden Chamber Chemotaxis and Chemoinvasion Assays.

In this aspect of the study, we used cells from primary uveal melanomas, which were established as STCs, in addition to the two cell lines (SOM-157d and -196B). Because cell number was restricted in STCs, a 48-well microchemotaxis chamber (Neuroprobe; The Laboratory Sales Co., Ltd., Cambs, UK) was used to maximize the number of factors tested with the limited source of primary material available. 

Standardization of Chemotaxis

To select the regulators that would have the most overt effect on uveal melanomas, initial studies of migration were performed on the uveal melanoma cell lines (SOM-196B and -157d) and two STCs (SOM-267 and -269) from which sufficient cells were available. The use of these cell lines and STCs allowed a range of concentrations to be tested to establish parameters. Proposed stimulatory and inhibitory factors included IL-1α, IL-1β, TNFα, IL-8, growth-related oncogene (GRO), macrophage inflammatory protein (MIP)-1α, MIP-1β, TGF-β1, TGF-β2, EGF, regulated on activation normal T-cell expressed and secreted (RANTES), and HGF (Table 2) . These were diluted to a range of concentrations with RPMI-1640 with 0.1% BSA, according to the manufacturer’s instructions, and were added to the lower chamber for stimulating migration before the fibronectin-coated (6.5 μg/mL) membrane was applied, or included with the cells in the upper chamber, for inhibition studies. ²⁷ Control wells were set up to obtain baseline migration, and each assay was performed in triplicate. For stimulation assays, wells with only RPMI-1640 with 0.1% BSA in the lower wells, were included as the negative control. Conversely, in inhibition experiments, wells in which the inhibitor was absent in the upper chambers acted as the positive control. After incubation, the membrane was detached from the chamber and nonmigrating cells on the upper surface of the membrane were gently wiped away. The membrane was fixed in ethanol, stained with rapid stain (Hema Gurr; BDH/Merck, Poole, UK) and mounted on a glass slide for analysis. Levels of migration were assessed by counting the number of cells present in five fields on the lower surface of the membrane, under light microscope (×400). Each assay was performed in triplicate. The concentrations of regulators that elicited the maximum response varied in each culture. The ranges are noted in Table 2 . For the most effective regulators, chemoinvasion assays were subsequently performed on ECM-coated (15 μg/mL) membranes on a series of 10 STCs. 

Chemoinvasion Assays

To study invasion, a combined extracellular matrix (ECM) solution (15 μg/mL), consisting of equal quantities of fibronectin, type IV collagen, and laminin in phosphate-buffered saline was used to coat the 8-μm-pore polycarbonate membranes. Membranes were air-dried at room temperature before use. Briefly, the most effective regulators were diluted in RPMI-1640 with 0.1% BSA and added to the appropriate wells (stimulators were added to the lower wells and inhibitors were added to the upper wells). Tumor cells (3 × 10⁵) suspended in RPMI-1640 with 0.1% BSA were seeded into the upper wells, and the chamber was incubated in a humidified chamber for 5 hours at 37°C. To assess baseline cellular invasion, appropriate controls were included in the assays, as described earlier, and each assay was performed in triplicate. After incubation, invasion was assessed as detailed earlier. 

To distinguish chemotaxis from chemokinesis in stimulation experiments, for SOM-196B, a checkerboard assay was used. Varying concentrations of regulators were added to the upper well only, the lower well only, or both wells, and the patterns of migration were observed. If cell movement occurred only in the presence of a concentration gradient, the migration was considered to be chemotaxis. To confirm stimulation by HGF, for SOM-196B, the HGF receptor (c-Met) was functionally blocked with a c-Met-blocking antibody (Chemicon, Temecula, CA), diluted to the recommended concentrations in RPMI-1640 with 0.1% BSA (0.5–2 μg/mL) and added to the cells 15 minutes before the assay. 

Results of invasion and migration assays using the chemotaxis chamber were expressed as a percentage of the appropriate control levels of invasion. For stimulators the negative control was used, whereas for inhibitors the positive control was used. 

Data Analysis

To assess the effect of different cells and conditioned media on SOM-196B invasion, significance was calculated by analysis of variance (ANOVA). A Student’s t-test was used to determine significance for the effects of growth factors, cytokines, and chemokines on uveal melanoma migration and invasion. These tests were used because there was no evidence to suggest that the data would not fit the underlying assumptions of the tests. In all cases P < 0.05 was taken as significant, and data at that level were used to establish that stimulation or inhibition of migration or invasion was significantly increased or decreased, respectively, compared with appropriate control invasion. Because a large number of tests were performed, there is only a small possibility that the significant differences observed would have occurred by chance. 

Results

Because uveal melanomas primarily target the liver, it is possible that factors specifically associated with hepatic cells contribute to this selection. To explore this possibility, the effect of the liver-derived cells HepG2 and HULECs and their CM on the invasion of one uveal melanoma culture (SOM-196B) was investigated. Both cells and CM from HepG2 and HULECs were shown to significantly increase invasion of SOM-196B (P < 0.05). These results were further compared with the effects of cells derived from other tissues, from sites not commonly colonized by uveal melanoma. With the exception of A549 cells, all other cell types and the appropriate CM was shown to increase invasion of SOM-196B significantly (P < 0.05). Results of these invasion studies are presented in Figure 1 . 

Because the metastatic process involves both directed invasion (chemoinvasion) and migration (chemotaxis), to identify specific factors that may regulate liver colonization, migration was initially assessed in two cell lines (SOM-157d and -196B) and two STCs (SOM-267 and -269) through a polycarbonate membrane coated with fibronectin (6.5 μg/mL), which provided a minimal barrier to migration and promoting attachment. ²⁷ Established cell lines were used for preliminary migration assays, because sufficient cells were available to optimize assay conditions and to prescreen a panel of 12 regulators for their stimulatory or inhibitory effects on uveal melanoma cells. The data were supported with the results from two STCs (SOM-267 and -269). Results from these experiments are presented in Figure 2 . In summary, maximum stimulation of cultures was shown by MIP-1α, GRO, and HGF, inducing significant migration (P < 0.05) in all four cultures. MIP-1β was also shown to increase migration (P < 0.05) significantly in three cultures (SOM-157d, -196B, and -269). With the exception of HGF, checkerboard analysis of stimulatory regulators confirmed chemotaxis in all cases. For HGF, a slight chemokinetic response was observed (data not shown). In invasive cultures (SOM-196B and -267), inhibition of migration by IL-1α, TGF-β1, and TGF-β2 was significant (P < 0.01). Because of unavailability of cells when working with STCs at low passage, only the most effective cytokines were used for further studies of chemoinvasion. IL-1β, TNF-α, EGF, RANTES, and IL-8 produced a limited or mixed response in the preliminary experiments and were thus not included in further studies. 

To assess directed invasion, chemoinvasion, through membranes coated with an ECM solution (15 μg/mL) was assessed in 10 STCs and two cell lines for HGF, MIP-1α, MIP-1β, IL-1α, TGF-β1, and TGF-β2. Confirmation of melanoma status was ascertained, because all uveal melanoma cultures expressed at least one marker associated with their nature as melanomas (data not shown). Basal invasion levels without any stimulation or inhibition varied between cultures. Because invasion involves active degradation of ECM components, the use of an ECM solution to study invasion, was possibly more appropriate. The results of these invasion assays are summarized in Figures 3 and 4 . Tumors were classified as noninvasive, weakly invasive, moderately invasive, invasive, and highly invasive (Table 1) relating to levels of cellular invasion in vitro. For each regulator, the effect of a range of concentrations was assessed, according to the manufacturer’s instructions (Table 2) . Results presented in Figures 3 and 4 represent the concentrations at which maximum effect was recorded for each culture. 

Briefly, invasion was significantly increased (P < 0.05) by MIP-1β in 9 of 12 tumors, whereas only 6 of 12 tumors significantly responded to MIP-1α (P < 0.05). GRO produced a significant invasive response in 9 of 12 tumors (P < 0.05). The largest invasive response was produced by HGF, and all cultures responded significantly (P < 0.05). It was particularly interesting that cells with a low invasive ability (SOM-157d, -281, and -286) demonstrated the greatest response to HGF (Figs. 5a 5b) . Two tumors (SOM-157d and -286) could not be used for inhibition studies, because basal levels of invasion were insufficient for analysis of inhibition. IL-1α was the most effective inhibitor of invasion in all cultures tested. However, there was no correlation between the invasive ability of any tumor and inhibition by IL-1α. The effects of TGF-β1 and -β2 were diverse and showed variation between cultures. In two cultures, invasion was stimulated by both isoforms (SOM-280 and -282) whereas in others, invasion was inhibited. TGF-β1 significantly inhibited invasion (P < 0.05) in 4 of the 10 tumors (SOM-196B, -267, -277, and -288), whereas TGF-β2 significantly inhibited invasion in two cultures (SOM-196B and -277). Conversely, invasion of SOM-280 and -282 was stimulated by both isoforms, but the effect was only significant for SOM-282 invasion (P < 0.05). Both isoforms had no effect in 4 of the 10 tumors (SOM-279, -281, -289, and -290). In addition SOM-288 did not respond to TGF-β2. 

Previous immunohistochemical staining of uveal melanoma and melanocyte cultures showed that uveal melanoma cells stained strongly for the HGF receptor c-Met, and melanocytes stained weakly (data not shown). Preincubating SOM-196B cells for 15 minutes with a functional blocking antibody to c-Met caused a decrease in HGF-stimulated invasion by 50.7% (Figs. 5c 5d 5e) . However, in the absence of HGF, after preincubation with the c-Met functional blocking antibody, invasion was decreased by 11.2%. 

Discussion

The primary site of metastases in patients with uveal melanoma is the liver. Other less-common sites of spread include the subcutaneous and pulmonary tissue. ⁷ For the degree of metastatic site specificity common in melanoma, regulators involved in the process may be expressed at the secondary sites, promoting adhesion, migration, invasion, and proliferation of tumor cells. The importance of host and tumor-derived growth factors, cytokines, and chemokines in tumor progression is becoming increasingly evident, affecting various stages, including tumor cell migration, leukocyte infiltration, cell adhesion molecules, and angiogenesis. ⁸ Moreover, to effect the required response, the tumor cells involved must express the necessary receptors. ^{28 29} Recent evidence has shown that breast cancer and malignant melanoma cells express distinct and nonrandom patterns of chemokine receptors that may play a critical role in determining the metastatic destination of these tumor cells. ³⁰ To investigate potential factors involved in liver colonization by uveal melanoma cells, the effect of cells derived from the liver and other tissues on the invasion of one uveal melanoma culture (SOM-196B) was studied. To extend this further, the effects of specific potential stimulatory and inhibitory factors involved in uveal melanoma cellular migration and invasion on a series of 10 primary uveal melanoma cultures and two cell lines were investigated. 

CM, and in most cases the cells themselves, of all tissue types investigated significantly increased invasion of SOM-196B, with the exception of A549 cells (Fig. 1) . Because of the constraints involved in culturing human hepatocytes, HepG2 cells were used to provide an approximation and as such were found to secrete factors inducing invasion of SOM-196B. Because metastatic tumor cells arrest in the microvasculature, endothelial cells derived from the liver were also studied as a possible source of soluble chemotactic factors for uveal melanoma cells. Both HULECs and the respective CM were found to increase invasion of SOM-196B significantly, an observation in agreement with previous work in which CM from different types of endothelial cells was shown to be chemotactic for a variety of tumor cells, including a malignant melanoma cell line ³¹ ; but the effect of endothelial cells themselves was not specifically investigated. 

However, cells and CM from primary keratinocytes and fibroblasts were shown to have a similar effect on SOM-196B invasion, and although A549 cells themselves did not significantly stimulate invasion, the corresponding CM did. It was interesting that the cells HepG2 cells, SOM-196B cells, and HULECs themselves were found to be comparatively more simulative than the appropriate CM, yet this was not the case with NAK 3.2, A459, and JFF cells. Because metastatic spread of uveal melanoma cells to the skin and lung occurs only occasionally, ⁷ it is possible that these results reflected this pattern, suggesting that the liver itself is an active participant in colonization by uveal melanoma. In addition, the observation that SOM-196B cells and CM also significantly induced invasion of themselves, infers that an autocrine motility response exists, but because checkerboard analysis was not performed in this aspect of our study, it was not possible to distinguish between intrinsic motility and chemotactic responses to autocrine and paracrine factors. Similar observations have been made previously, using metastatic variants of RAW117 large-cell lymphoma cells. ³² In this published study, CM from lung microvessel endothelial cells, and lung fibroblasts, together with its own CM, stimulated migration of highly metastatic lung variants. 

To further investigate the regulation of uveal melanoma migration and invasion in vitro, a panel of four cultures only (SOM-196B, -157d, -267, and -269) was initially selected to assess the effects of a wide selection of regulators on migration (Table 2) . Preliminary studies on cell migration through fibronectin-coated membranes identified significant stimulation of migration by MIP-1α, MIP-1β, HGF, and GRO, whereas IL-1α, TGF-β1, and TGF-β2 significantly inhibited migration (Fig. 2) . IL-1β, TNF-α, EGF, RANTES, and IL-8 produced a limited or mixed response in the preliminary experiments and were therefore not included in further studies. It is perhaps surprising that EGF did not produce a consistent significant response in this study, because previous evidence has suggested that expression of EGFR by uveal melanoma cell lines correlates with localization to the liver. ²² However, because primary cultures in this present study differed in their responses, our results may have been biased by the specific tumors used. 

The regulators with the most overt response (MIP-1α, MIP-1β, HGF, GRO, IL-1α, TGF-β1, and TGF-β2) were shown to effect cellular invasion in a series of 10 STCs differentially, probably reflecting innate variations in invasive abilities of the primary cultures. IL-1α and HGF were shown to significantly inhibit and stimulate invasion, respectively, in all cultures studied, whereas the effects of other regulators such as TGF-β1 and -β2, were less consistent between tumors and thus may have a more specific implication for individual tumors (Figs. 3 4) . However, it must be taken into consideration that, because the tumors (both STCs and cell lines) were all surgically resected, all could be classified as large tumors and thus associated with a poor prognosis. Consequently, this may have had an affect on the responses that we observed. 

This investigation has confirmed previous studies of uveal melanoma that suggested that these cells are responsive to HGF. ²¹ Although some movement in response to HGF was thought to be due to random motility, the observation that all tumors studied had a significant response to HGF, which was relatively high in comparison with the response to other factors tested, suggests specific involvement of HGF in uveal melanoma metastasis (Fig. 3) . It is also of particular interest that cultures classified as noninvasive and weakly invasive, showed the greatest responses to HGF and that such increased responses were not seen with the other stimulatory factors. There are several possible explanations for this effect. First, in this situation, it is possible that, these invasive melanomas (SOM-196B, -267, -277, -279, -280, -282, -288, -289, and -290) are already stimulated by autocrine HGF, additional stimulation by inclusion of this growth factor in the assay would only produce a limited response in comparison to weakly invasive melanomas (SOM-157d, -281, and -286) where autocrine stimulation may be less apparent. Evidence suggests that invasive uveal melanoma cells are capable of expressing HGF, with micrometastases in the liver having been found to stain positively for HGF. ²¹ Second, the difference may in part arise from the involvement of the HGF receptor, c-Met. All uveal melanoma cultures tested were found to express c-Met (data not shown), and blocking of c-Met in SOM-196B with a functional blocking antibody completely abrogated the invasive response to HGF (Figs. 5c 5d 5e) , thus confirming that HGF directly stimulates even invasive uveal melanoma cells. However, blocking c-Met in the absence of stimulation by HGF also decreased invasion of SOM-196B, suggesting that either innate production of HGF by the tumor itself was blocked, or c-Met itself contributes to the invasive response. Certainly, much evidence exists suggesting the role of c-Met in the invasion of other tumors, ^{33 34} and there is growing evidence implicating both c-Met and HGF in a number of autocrine and paracrine responses in a variety of cell types, promoting, among other responses, junctional breakdown, directed migration and invasion, and cell survival. The possibility therefore exists that other pathways, perhaps used by the more invasive melanomas, are also responsible for stimulating c-Met in uveal melanomas, and recently it has been reported that autocrine TGFα, binding to its receptor (EGFR), causes phosphorylation and activation of c-Met, in the absence of HGF. ³⁵ Therefore, by blocking c-Met, alternative pathways such as the TGFα/EGFR pathway may also be abrogated and the subsequent response prevented, potentially explaining the decrease in invasion noted after c-Met was blocked, in the absence of HGF. 

In the eye, HGF has been found in both the aqueous and vitreous humor and is expressed by a number of cell types, including retinal endothelial cells. ^{36 37} Concentrations of HGF in the vitreous humor were also shown to increase significantly migration of retinal endothelial cells. ³⁷ Expression of and the response to HGF by retinal endothelial cells and the presence in the vitreous humor may therefore have a significant part in neovascularization and subsequent development of uveal melanoma and invasion of the vitreous compartment. In connection with metastasis to the liver, production of HGF is associated with Kupffer cells and sinusoidal endothelial cells. ³⁸ The ubiquitous expression of c-Met by uveal melanoma cells has been shown in this and other studies, indicating that HGF potentially plays a pivotal role in uveal melanoma invasion and growth at both the primary and secondary sites. ²¹  

In contrast to HGF, IL-1α showed significant inhibition of invasion through ECM components in all cultures tested (Fig. 4) , including cultures classified as only weakly invasive. No correlation could be made between the inhibitory responses and the invasion levels in vitro. The roles of both IL-1α and -1β in metastasis are complex and are often contradictory, affecting a wide range of activities. As yet, no evidence exists supporting the significance of IL-1 in uveal melanoma metastasis. In contrast, several studies have considered the potential role of IL-1 in progression of cutaneous melanoma. For example, IL-1α and IL-1β may inhibit growth of early-stage cutaneous melanoma. As the tumor progresses, resistance to the effects of both isoforms may be acquired. ^{39 40} IL-1β and TNFα have also been associated with enhanced cell adhesion and promotion of liver metastases of cutaneous melanomas through the upregulation of vascular cell adhesion molecule (VCAM)-1 on hepatic sinusoidal cells. ^{41 42} When considering the effect on migration and invasion, in contradiction of the results of the present study, IL-1α and IL-1β have been commonly associated with promotion of migration for many tumor types. ^{43 44} During inflammatory responses in the eye, IL-1 is produced by monocytes, macrophages, and resident corneal cells, affecting a number of processes, including promotion of neovascularization and chemotaxis. ⁴⁵ Moreover, this cytokine has been implicated in the abrogation of the immune privileged nature of the ocular environment. ⁴⁶ Thus, an explanation of the inhibitory response of IL-1α on cellular invasion in uveal melanoma remains unclear. 

In preliminary investigations of migration through fibronectin, both TGF-β1 and -β2 significantly inhibited migration (Fig. 2) . However, in accordance with previous studies of other tumors, in subsequent experiments on invasion of the 10 primary posterior uveal melanomas, the inhibitory effects of the two isoforms were less pronounced, and showed intertumor variation ²⁷ (Fig. 4) . Thus the effects of both isoforms may have a specific effect on the individual tumor. Significant inhibition by both isoforms was seen in two cases (SOM-196B and -277), whereas stimulation of invasion was observed in two tumors (SOM-280 and -282). Both isoforms had no effect in 4 of the 10 cultures studied (SOM-279, -281, -289, and -290). 

The effects of TGF-β on uveal melanoma are not well known, but loss of TGF-β2 receptor expression has been observed in uveal melanoma, and the effect was cell line dependent. ⁴⁷ The eye is known to be an immune-privileged site in which both adaptive and immune responses are downregulated. ⁴⁸ TGF-β, present in the aqueous humor, has been shown to affect the local cytokine milieu preferentially and to induce apoptosis in antigen-presenting cells (APCs), influencing this state of immune deviation. ⁴⁹ Uveal melanomas rarely develop in the iris, and it is possible that TGF-β significantly inhibits tumor growth. Moreover, iris melanomas, developing in the anterior chamber, surrounded by aqueous humor rarely metastasize. ⁵⁰ Therefore the inhibitory effects of TGF-β may have a significant role in development of primary uveal melanoma in the orbit. TGF-β may have multiple effects, as in the early stages of tumorigenesis, because it inhibits cell growth by inducing an arrest in the cell cycle. However, advanced malignant cells often acquire resistance to growth inhibition by TGF-β. ^{51 52} This biphasic effect has also been seen in cutaneous malignant melanoma, ⁵³ where melanocytic and nonmetastatic melanoma cells have shown little response to TGF-β, whereas highly metastatic melanoma cells have shown increased migration induced by TGF-β. ^{54 55} It is therefore of note that invasion of SOM-280 was stimulated by both TGF-β isoforms. Clinically, the patient had three foci of deep scleral invasion, inferring stimulation of invasion in vivo. The number of tumors investigated in this study was too small to draw any reliable conclusions, but it is possible that both uveal and cutaneous melanomas are comparable in possessing a biphasic response to TGFβ isoforms. 

Although chemokines are known to be involved in the migration of lymphocytes during the immune response, ^{56 57 58} studies reporting the effects of MIP-1α, MIP-1β, and GRO on tumor cell migration are limited. In our study, most cultures exhibited a significant invasive response to MIP-1β, whereas only half of the cultures studied responded to MIP-1α, and the response was generally of a lower magnitude when compared with the response to MIP-1β (Fig. 3) . Of note, SOM-282 had a negligible response to MIP-1α and -1β, and was also one of the two cultures stimulated by TGF-β. Most tumors in this study demonstrated a significant invasive response to GRO. There is limited information on the effects of GRO on the tumorigenesis of melanoma cells, but it is known to act as an autocrine growth factor and has been shown to induce migration and proliferation of the cutaneous form. ^{59 60} Forced overexpression of GRO in immortalized mouse melanocytes also enables cells to form tumors when subcutaneously injected into nude mice, ⁶⁰ and this regulator may consistently stimulate tumors of melanocytic origin. 

We have reported herein the in vitro effects of a series of significant different cell types, growth factors, cytokines, and chemokines on the migration and invasion of uveal melanoma cells. Of particular interest have been the responses shown to HGF, GRO, MIP-1β, IL-1α, and TGF-β1 and -β2. It is therefore possible that these factors play an important role in the regulation of uveal melanoma migration and invasion. IL-1α and TGF-β may play a role in inhibiting tumor growth in situ, whereas HGF and GRO may be particularly involved in promoting invasion and tumorigencity in vivo. All tumors used to establish STCs had been resected within the past 16 months, and consequently it was not possible to correlate responses in vitro with long-term prognoses. 

²  Contributed equally to the work.

 Supported by Yorkshire Cancer Research, National Eye Research Council (Bristol, United Kingdom) and The Humane Research Trust.

 Submitted for publication January 22, 2002; revised April 22, 2002; accepted May 20, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Histopathological Details of Patients Treated for Posterior Uveal Melanoma by Enucleation

SOM Cell Line/Tumor	Age	Sex	Location	Volume (mm3)	Cell Type	Metastatic Disease	Extrascleral Spread	Invasive Behavior in Culture*	Status
157d	73	M	Choroid	3647	Epithelioid	Yes	No	Non-invasive	Dead
196B	80	M	Choroid	885	Mixed	No	No	Invasive	Alive at 48 months
267	51	M	Choroid	1395	Spindle A	No	No	Moderately invasive	Alive at 19 months
269	72	M	Ciliary Body	2273	Spindle B	No	No	Non-invasive	Alive at 19 months
277	74	M	Choroid	1476	—	No	Yes	Highly invasive	Alive at 16 months
279	72	M	Choroid	1263	Spindle B	No	No	Invasive	Alive at 16 months
280	85	M	Choroid	2621	Mixed	No	3 Foci of deep scleral invasion	Moderately invasive	Alive at 16 months
281	54	M	Choroid	1273	Spindle B	No	No	Weakly invasive	Alive at 16 months
282	64	M	Ciliary Body	1141	Mixed	No	No	Invasive	Alive at 13 months
286	79	F	Choroid/Ciliary Body	677	Spindle B (occasional epithelioid cells)	No	No	Non-invasive	Dead (cause unknown)
288	59	M	Choroid/Ciliary Body	1158	Mixed	No	No	Invasive	Alive at 11 months
289	88	F	Choroid	1477	Mixed	No	No	Invasive	Alive at 11 months
290	53	M	Choroid	754	Mixed	No	—	Moderately invasive	Alive at 11 months

*  Invasive behaviour in culture was classified as follows: <15 cells/field mean migration, noninvasive; 15–25 cells/field mean migration, weakly invasive; 25–45 cells/field mean migration, moderately invasive; 45–75 cells/field mean migration, invasive; >75 cells/field mean migration, highly invasive.

Table 2.

 

Sources of Cytokines and Chemokines Used in the Study

Cytokine or Chemokine (Recombinant Human)	Optimum Concentration Range (ng/ml)	Reference Code	Source
IL-1α	0.1–10	IL001	Chemicon, (IemeculaR CA)
IL-1β	0.1–10	201-LB	R&D Systems Ltd., (Oxon, UK)
TNFα	0.1–1.0	T6674	Sigma Chemical Co. Ltd. (Dorset, UK)
IL-8	150–400	208-IL	R&D Systems
GRO	10–40	GF005	Chemicon
MIP-1α	1.0–60	GF010	Chemicon
MIP-1β	10–60	271-BME	R&D Systems
TGF-β1	0.01–1.0	T7039	Sigma Chemical Co.
TGF-β2	0.01–0.1	T7039	Sigma Chemical Co.
EGF	0.01–0.1	E-4127	Sigma Chemical Co.
RANTES	0.1–20	GF020	Chemicon
HGF	20–80	294-HG005	R&D Systems Ltd.,

Figure 1.

 

Mean (bars, SE) invasion of SOM-196B cells, when stimulated by cells and CM from different cellular sources. *P < 0.05, when compared with a negative control (tumor cells without stimulation by cells or CM).

Figure 2.

 

Simulation and inhibition of migration of four uveal melanoma cell cultures, expressed as a percentage of the control levels of migration.

Figure 3.

 

Stimulation of invasion of uveal melanoma cells by MIP-1α, MIP-1β, GRO, and HGF, expressed as a percentage of the control levels of invasion. *P < 0.05 and P < 0.01 when compared with a negative control (tumor cells without stimulation).

Figure 4.

 

Inhibition of invasion of uveal melanoma cells by IL-1α, TGF-β1, and TGF-β2, expressed as a percentage of the control levels of invasion. *P < 0.05 and P < 0.01, when compared with a positive control (tumor cells without inhibition). X, cultures with levels of basal invasion insufficient for inhibition (SOM-157d and -286).

Figure 5.

 

Invasion responses of SOM-157d and -196B to HGF and blocking of HGF-stimulated invasion by c-Met blocking antibody. All images show invading cells on the underside of the Boyden chamber membrane (×400 magnification). Nuclei are stained with Gill hematoxylin. Invasion response of SOM-157d cells to (a) serum-free media (0.1% BSA; negative control) and (b) HGF (60 ng/mL). Invasion response of SOM-196B cells to (c) serum-free media and (d) HGF (20 ng/mL). (e) Effect of blocking c-Met on the invasion of SOM-196B cells after stimulation with HGF (20 ng/mL).

The authors thank Robin Farr for assistance with photographic work and Irene Canton for help with counting cells.