MET Oncogene Inhibition as a Potential Target of Therapy for Uveal Melanomas

Abstract

Purpose.: The purposes of this study were to investigate the frequency of MET activation in uveal melanomas (UMs), to study the potential molecular mechanism for its activation, and to assess the utility of MET inhibition as a potential therapy for UM.

Methods.: The frequency of MET activation in UMs was studied by using immunohistochemistry and Western blot analysis in 46 primary UMs and six UM cell lines. Sequencing was used for detection of activating mutations in the MET gene, and the effect of selective MET inhibition was assessed by cell proliferation and migration assays.

Results.: The results showed that the majority (82.5%) of the 46 UMs expressed activated MET protein. Three of the UM cell lines, C918, 92.1, and MEL202, showed strong MET and pMET expression, whereas the other three showed weaker expression. Sequence analysis identified no activating mutations in MET in any of the 22 tumors or in the six UM cell lines. Selective MET blocking showed inhibition of tumor cell proliferation at an IC₅₀ ranging from 2.5 to 5.2 μM. A significant inhibition of UM cell migration was also observed starting at 1.25 μM.

Conclusions.: The results indicate that MET is activated in a significant number of UMs and also that MET activation in UMs is most likely through indirect gene activation rather than copy number alteration or mutation involving the MET gene. MET inhibition could be a target of therapy for UM.

Uveal melanoma (UM), including choroidal and iris melanomas, are the most common primary intraocular tumor in adults. The tumor spreads through a hematogenous route, mostly to the liver, which is affected in approximately 90% of patients with metastatic disease. ¹ Once metastatic, UMs are highly fatal, and patients rarely survive more than a year from the time of diagnosis. ² Currently, there is no available systemic therapy for these tumors, and identifying novel therapies therefore is crucial for improving survival. ³  

Activated MET oncoprotein, also known as cMET, appears to play an important role in tumorigenesis of a wide variety of tumors. ^4–6 In UM, MET is overexpressed in 60% to 86% of solid tumors, and, at least in univariate analysis, it is associated with tumor aggressiveness. ^7,8 It has been suggested that MET plays a role in the preferential spread of UM to the liver, ⁹ a notion that is supported by experimental evidence from a murine model of selective liver metastasis. ¹⁰ Therapies targeting the MET/HGF pathway ^11–14 may also be interesting prospects for treatment of UM. 

MET serves as the high-affinity receptor for hepatocyte growth factor/scatter factor (HGF/SF), a disulfide-linked heterodimeric molecule produced predominantly by mesenchymal cells and acting primarily on MET-expressing epithelial cells in an endocrine and/or paracrine fashion. ^15,16 The cellular responses to MET stimulation by HGF/SF are important in mediating a wide range of biological activities, including embryologic development, wound healing, tissue regeneration, angiogenesis, growth, invasion, and morphogenic differentiation. ⁴  

The mechanism of MET activation in UMs is still unknown. The chromosomal location of MET on 7q31 is not commonly amplified in UMs. The frequency of MET-activating mutations in UMs is still unknown. The role of indirect MET regulation through loss of a negative regulator such as VHL, located on chromosome 3, which is commonly altered in UM, is also not clear. In this study we investigated the frequency of MET activation in UM, studied the molecular mechanism of its activation, and assessed the potential utility of selective MET inhibition as a potential therapy for UM. 

Materials and Methods

Patient Samples

A total of 46 primary UMs from patients treated with enucleation were studied. In 22 tumors, frozen tissue was available for DNA sequencing and RNA expression studies. Control tissues for RNA expression and Western blot were obtained from the choroid of three nontumorous eyes obtained from the Ohio Lions Eye Bank and collected within less than 24 hours from the time of death. Formalin fixed, paraffin-embedded tissue from five nontumorous eyes were used as controls for the immunohistochemical staining for pMET and MET in the retina and choroid. DNA obtained from 50 healthy individuals was used for assessment of the frequency of unreported single-nucleotide polymorphisms (SNPs) detected in tumor samples by MET gene sequencing. All specimens (uveal melanomas and normal controls) were collected as per Institutional Ethical Review Board–approved protocols (2003C0057 and 2006C0045) and in accordance with the Declaration of Helsinki. 

Cell Lines

Uveal melanoma cell lines C918 ¹⁷ and MUM2C (or OCM1, originally established by June Kan-Mitchell, University of Texas at El Paso) were kindly provided by Mary Hendrix (Northwestern University Feinberg School of Medicine, Chicago, IL). The uveal melanoma cell lines MEL202 (originally established by Bruce R. Ksander, Schepens Eye Research Institute, Boston, MA), OCM3 and OCM8 (originally established by June Kan-Mitchell), and 92.1 (originally established by Martine J. Jager, Leiden University, The Netherlands) were obtained from the European Searchable Tumor Line Database and Cell Bank (http://www.ebi.ac.uk/ipd/estdab/ provided in the public domain by the European Bioinformatics Institute, Hinxton, UK). All cell lines were reported to be from primary uveal melanomas. Authentication of the cell lines was achieved by short tandem repeat (STR) profiling with the markers D8S1179, D16S539, D7S820, AMEL, CSF1PO, D3S1358, D5S818, D13S317, D18S51, D21S72, and TPOX and was compared to published profiles of uveal melanoma cell lines. ¹⁸ Authentication confirmed that the two cell lines OCM3 and OCM8 have the same genetic origin. The cell lines were grown in RPMI medium (Invitrogen Gibco, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin. Normal retinal pigmented epithelial cells (ARPE-19) and normal fibroblasts (MRC-5) were obtained from American Type Culture Collection (ATCC, Manassas, VA) and were grown according to the provider's protocol. 

Quantitative Reverse Transcription-PCR Assay

5′ Nuclease-quantitative (real-time) PCR (qRT-PCR) assays (TaqMan; Applied Biosystems, Inc., [ABI], Foster City, CA) were performed with predeveloped assays, according to the manufacturer's protocol (model 9700 real-time PCR system; ABI). The reactions were performed in triplicate for each sample in 25 μL, with a final dilution of 1× each of PCR universal master mix and 1× of the predeveloped probes. Only probes amplifying the cDNA and not the genomic DNA were selected. In addition to MET, an endogenous control GUSB was tested in separate reactions. The PCR reaction settings were 95°C for 3 minutes, then 40 to 50 cycles of 95°C for 15 seconds, and 60°C for 1 minute. The relative expression levels were assessed by the comparative CT method, according to our previously published protocol. ¹⁹  

Immunohistochemistry

Tissue microarray (TMA) preparation was performed at the histology core facility, Department of Pathology, The Ohio State University. TMAs were prepared from paraffin-embedded tissue, as previously reported. ¹⁹ Representative cores of at least 3-mm from each tumor were used. For tumors with mixed morphology up to six cores representing different morphologic regions were used. Ten external control tissues were also studied (liver, nerve, heart, lymph node, thyroid, prostate, testis, pancreas, placenta, and kidney) to ensure consistency between the experiments. 

Polyclonal rabbit anti-MET antibodies for total (MET) ²⁰ and phosphorylated protein (pY1230/pY1234/pY1235, pMET) ²¹ were obtained from Invitrogen. A rabbit polyclonal antibody for the α chain of HGF was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Deparaffinized tissue sections were heat pretreated with citrate buffer (pH 6; total MET and HGF) or 0.01 mM EDTA (pH 8; pMET). Immunostaining without the primary antibody was used as the negative control. Antibody dilutions were 1:200 for total MET, 1:50 for HGF, and 1:100 for pMET, with overnight incubation at 4°C. For detection of the immunostaining, a chromogen (NovaRED; Vector Laboratory, Burlingame, CA) was used to produce a dark red stain to minimize interference from the melanin pigment. After the tissues were counterstained with hematoxylin and mounted, the slides were evaluated (Eclipse 400E microscope; Nikon Tokyo, Japan). The positive control for MET was a TMA prepared from multiple primary liver tumors, whereas the positive control for pMET protein was the HTB-30 cell line stimulated for 30 minutes with HGF 100 ng/mL and processed with routine formalin fixation and paraffin embedding as cell pellets. 

Immunohistochemistry was performed twice for each antibody, and the average score of the two experiments was assessed. The immunohistochemical staining signals were scored on a 0 to 2 scale based on the intensity of staining and the percentage of stained cells in each core. The average staining of multiple tissue cores for each tumor was assessed with slight modification of our published protocol. ²² A final score of 1 to 200 was obtained by multiplying the percentage of stained cells by the intensity of staining (0–2). Final scores less than 10 were deemed negative, whereas scores from 10 to 100 indicated weak to moderate staining. Strong staining was considered for scores above 100. Samples with staining limited only to the periphery of the tissue cores were excluded from our scoring assessment. Also, areas of tumors with intense melanin pigmentation were excluded from our analysis. 

Western Blot Analysis

For total protein extraction, the cells were lysed by incubation for 10 minutes with ice-cold 1× cell lysis buffer (Cell Signaling Technology, Boston, MA), spiked with 1 mM PMSF and 1× phosphatase inhibitor cocktail 2 (Sigma-Aldrich, St. Louis, MO) immediately before use. Nuclear and cytoplasmic proteins were extracted with NE-PER nuclear and cytoplasmic extraction reagents (Pierce Biotechnology, Rockford, IL), spiked with protease inhibitors according to the manufacturer's protocol. Protein concentrations were determined with a BCA protein assay kit (Pierce Biotechnology). For Western blot analysis, 20 to 50 μg of protein extracts per sample were loaded on 10% to 12% Tris-HCl gel (Ready-Gel; Bio-Rad Laboratories, Hercules, CA) and transferred to nitrocellulose filters (Bio-Rad Laboratories). 

For MET and pMET, the filters were probed with the antibodies used for immunohistochemistry, at 1:500 dilution for primary and 1:2000 for secondary antibodies. For HGF, we used the same antibody as was used for immunostaining at 1:200 dilution for the primary antibody and 1:1000 dilution for the secondary antibody. To assess the efficiency of nuclear fractionation, we used a monoclonal antibody to lamin B1 (Invitrogen) at 1:500 dilution. The equality of loading was assessed by blotting with a monoclonal antibody for β-actin (Sigma-Aldrich, St. Louis, MO), as well as Ponceau S staining (0.1% wt/vol in 5% acetic acid; Sigma, St. Louis, MO). Signals were developed with chemiluminescent substrate (SuperSignal West Pico; Pierce Biotechnology). Western blot analysis was repeated at least twice. 

DNA Extraction and Sequencing

DNA was extracted from fresh-frozen tumor tissues (DNeasy; Qiagen, Valencia, CA). The DNA integrity was assessed by minigel electrophoresis. Extracted DNAs were stored at 4°C until used in the experiments. Mutational screening of all the reported activating mutations in the MET gene ²³ were performed by direct sequencing of fragments obtained by polymerase chain reaction (PCR) on a DNA sequencer (model 3730; ABI). The sequence result was read by aligning with the reference sequence provided in GenBank, accession number M35073 (http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). 

Cell Proliferation Assay

The selective MET tyrosine kinase inhibitor SU11274 [(3Z)-N-(3-chlorophenyl)-3-([3,5-dimethyl-4-[(4-methylpiperazin-1-yl)carbonyl]-1H-pyrrol-2-yl]methylene)-N-methyl-2-oxo-2,3-dihydro-1H-indole-5-sulfonamide] ²⁴ was obtained from Sigma-Aldrich. The cell proliferation assay was performed in triplicate on 1 to 3 × 10³ cells/well. The cells were plated with 10 serial dilutions of the drug, ranging from 100 to 0.1 μM and incubated at 37°C for approximately 72 hours. Cell viability was then assessed (CellTiter 96 AQueous One assay; Promega, Madison, WI), according to the manufacturer's protocol. The absorbance was obtained at a wave length of 490 nm by spectrophotometry (Epoch Microplate Spectrophotometer; Biotech, Winooski, VT), and the half-maximum inhibitory concentration (IC₅₀) was assessed (Gen5 data analysis software, ver. 1.09; Biotech). For the IC₅₀ determination, this program uses the dose–response equation Y = {(A − D)/[1 + (X/C) ^B ]} + D, where X is the drug concentration, Y is absorbance at 490 nm, A is the upper asymptote, B is the slope, C is IC₅₀, and D is the lower asymptote. 

Cell Migration Assay

The uveal melanoma cell lines C918 and OCM-3 were seeded in T-75 flasks and grown to 80% confluence. The cells were starved overnight through incubation in serum-free medium, trypsinized, washed, and seeded on cell migration plates (Transwell Chambers; Corning, Inc., Lowell, MA) with 8.0-μm pore polycarbonate membrane inserts (Corning, Lowell, MA) in serum-free medium at a concentration of 150 × 10³ cells/well. The cells in each insert were treated with a serial dilution of SU11274 (0–10 μM). The inserts were placed in wells containing 600 μL of growth medium with 10% fetal bovine serum and incubated overnight. After fixation and staining with thiazin blue stain (Wescor Inc., Logan, UT), the number of migrating cells in the bottom chamber were counted in five high-power fields by two independent investigators. The experiment was performed in duplicate, and the average of the two experiments was evaluated. To assess the cytotoxic effect of the SU11274 at the concentrations used for testing cell migration inhibition, we performed an independent experiment by counting the cells incubated overnight with the same concentrations of the drug as that used in the cell migration inhibition experiment. Statistical analysis was performed with single-factor ANOVA (Office 2003 Excel software; Microsoft, Redmond, CA). 

Results

Quantitative Reverse Transcription-PCR Assay

qRT-PCR was performed on 18 samples with available high-quality RNA. Figure 1 shows the average expression level in these samples compared with the expression level in the choroid of three nontumorous eyes. Only two (11.1%) samples showed expression levels equal to that in the control cells (<1.5 times normal), six (33.3%) samples showed a mildly elevated expression (1.5–3.0 times normal), and 10 samples (55.6%) showed more than 3.0 times normal expression. 

Figure 1.

 

MET gene expression using quantitative RT-PCR in UM tissues. Relative expression of MET in 18 UM tissues compared with expression in the choroid of three nontumorous eyes. The expression of GUSB was used as an internal control. Six (33.3%) samples showed a mild elevated expression (1.5- to 3.0-fold of the normal) and 10 samples showed overexpression (>3.0-fold of the normal). Chr 3 (chromosome 3) status: M3, monosomy 3; P3, partial chromosome 3 alteration not involving the 3p25 region; P3*, partial chromosome 3 loss involving the 3p25 region; ID3, acquired isodisomy of chromosome 3; and D3, disomy of chromosome 3.

In the 18 samples, the status of chromosome 3 loss by genotyping and/or molecular cytogenetics was available (Fig. 1). VHL is a negative regulator of MET, and it is located on chromosome 3. Loss of the VHL chromosomal region (3p25.3) was observed in 11 of 18 tumors. In addition one tumor showed acquired isodisomy of chromosome 3, whereas the remaining six showed either disomy of chromosome 3 or partial chromosome 3 alterations that did not involve the VHL gene chromosomal region (Fig. 1). No association between the MET gene expression and loss of the VHL gene region was observed. 

Immunohistochemistry

Total MET staining was mostly cytoplasmic, but weak nuclear staining was also observed. Both nuclear and cytoplasmic staining were observed for the pMET antibody. The nuclear localization of pMET protein was confirmed by Western blot analysis of nuclear/cytoplasmic fractionated proteins (Fig. 2A). Total MET was positive in almost all samples (Figs. 2B, 2C) with strong staining in 26 (56.5%) of 46 and weak to moderate staining in 18 (39.1%) of 46 of the tumor tissues. Only two tumors were negative for MET staining. Both nuclear and cytoplasmic staining were observed for the pMET protein (Figs. 2E, 2F). Nuclear staining was strong in 15 (32.5%) of 46 of the tumors, weak to moderate in 23 (50%), and negative in 8 (17.4%). Cytoplasmic staining was strong in 13 (28.3%) of 46 tumors, weak to moderate in 25 (54.3%), and negative in 7 (15.2%; Fig. 3B). 

Figure 2.

 

Nuclear and cytoplasmic expression of MET and pMET in UMs. (A) Western blot showing MET expression in both the cytoplasm (C) and nucleus (N) of the C918 and OCM3 cell lines. The expression of pMET was primarily localized to the nucleus, as shown by the staining for lamin B1. (B) Representative UM case with mostly nuclear MET expression. (C) Representative UM case with strong cytoplasmic and nuclear expression of MET. (D) pMET expression in both tumor and nontumor tissue. (E, F) Representative cases of pMET expression in UM showing prominent nuclear staining with areas of cytoplasmic staining. Scale bars: (B, C, E, F) 50 μm; (D) 100 μm.

Figure 3.

 

Summary of MET, pMET, and HGF expressions in uveal melanoma cell lines and primary tumors. (A) Western blot of MET, pMET, and HGF expression in UM cell lines. Antibody to the cytoplasmic domain of MET detects the preprocessed polypeptide (170 kDa), the full-length protein (145 kDa), and three related polypeptides (110, 60, and 40 kDa) formed by proteolytic cleavage. HGF was expressed in all cell lines. (B) HGF, MET, and pMET expression in 36, 46, and 46 primary tumors, respectively, as detected by immunohistochemistry. pMET-N and pMET-C represent nuclear and cytoplasmic expression of phospho-MET, respectively. (C) The expression of pMET and MET in primary tumors detected by immunostaining in correlation with HGF expression. −ve, negative; +, weak to moderate; and ++, strong expression.

In tumor-containing eyes the nontumorous tissues including the retina, retinal pigmented epithelial (RPE) cells, and choroidal melanocytes showed weak to moderate staining for MET and pMET (Fig. 2D). The staining in the retina of the tumor samples was mostly in the inner nuclear and outer plexiform layers. We included five nontumorous eyes as control specimens, and all showed minimal pMET cytoplasmic staining in the retina but not in the choroid or RPE. This result was confirmed by Western blot analysis in three additional control eyes (Supplementary Fig. S1). 

Immunostaining for HGF was successful in 36 tumors. Negative staining was observed in 2 (5.6%) of 36, weak to moderate staining in 10 (27.8%) of 36 and strong staining in 24 (66.7%) of 40 of the uveal melanomas (Fig. 3C). 

DNA Sequencing

There was no activating mutation identified in the MET gene in either the 22 primary tumors or the six UM cell lines included. However, three single nucleotide polymorphisms (SNPs), 4099C>T, 4071G>A, and 4146G>A, were observed in exons 20 and 21 in some of the UM samples. All three SNPs had no associated amino acid changes. The frequency of such SNPs was not significantly different between tumors and the 50 normal control eyes (data not shown) suggesting that such SNPs are not pathogenic. 

Correlation between MET Protein Expression and Molecular Alterations in MET and VHL Genes

For 17 samples, data on the protein expression as well as the copy number alteration in both the 7q31 region (the location of the MET gene) and the 3p25 region (the location of the VHL gene) were available (Table 1). Of the 17 tumors, 11 showed loss of the 3p25 region, mostly in the form of monosomy of chromosome 3, and two showed polysomy of the 7q31 region. There was no significant correlation between the expression of MET and pMET and the copy number alteration in the VHL gene. 

Table 1.

 

MET, HGF, and pMET Protein Expression Compared with Genomic Alterations in the 7q31 and 3p25 Chromosomal Regions

Tumor ID	MET	HGF	pMET-N	pMET-C	7q31 (MET)	3p25 (VHL)
UM0002	++	+	++	++	Gain	Acquired isodisomy
UM0004	++	NA	−ve	+	No change	Loss
UM0006	+	+	++	+	No change	No change
UM0011	+	+	+	+	No change	Loss
UM0017	+	+	+	+	Gain	No change
UM0021	++	++			No change	Loss
UM0030	++	+	+	+	ND	Loss
UM4033	++	++	++	+	No change	Loss
UM4035	++	++	−ve	−ve	No change	No change
UM4036	++	++	++	+	No change	Loss
UM5069	++	++	+	+	No change	No change
UM5070	++	+	++	+	No change	Loss
UM5071	−	++	−ve	−ve	No change	No change
UM5072	+	++	−ve	+	No change	Loss
UM5074	++	++	+	+	No change	Loss
UM6002	++	+	++	−ve	No change	Loss
UM5068	++	++	+	+	No change	Loss
UM5073	++	++	+	−ve	No change	Loss
UM5048	++	+	+	+	ND	Loss
UM5047	++	++	+	+	ND	Loss
UM5040	++	++	++	++	ND	Loss

 pMET-N, nuclear phospho MET; p-MET-C, cytoplasmic phospho MET; −ve, negative; +, weak to moderate; ++, strong expression.

Western Blot Analysis

Western blot analysis showed strong expression of the full-length MET protein (145 kDa) in the C918, MEL202, and 92.1 cell lines and moderate expression in OCM1, OCM3, and OCM8. The levels of truncated MET protein (60 kDa) were higher in the C918, OCM1, OCM3, and OCM8 cell lines. The levels of the 40-kDa proapoptotic fragment were higher in the C918 and MEL202 cell lines (Fig. 3A). Cytoplasmic/nuclear fractionation was performed for C918 and OCM3 cell lines and showed mostly cytoplasmic expression of MET, whereas pMET expression was mostly nuclear with weak cytoplasmic expression in C918 and undetectable cytoplasmic expression in OCM3 (Fig. 2A). The MET antibody identified mostly nuclear localization of the full-length protein. We observed inhibition of MET phosphorylation in UM cell lines by SU11274 starting at a concentration of 2.5 μM, which is close to its IC₅₀ (Fig. 4B). 

Cell Proliferation Assay

SU11274 showed inhibition of cellular proliferation at IC₅₀ values ranging from 2.5 ± 0.98 μM (92.1) to 4.7 ± 0.91 μM (OCM1) (Fig. 4A, Table 2). The results represent the average of at least three independent experiments, each performed in triplicate. The IC₅₀ of SU11274 in the UM cell lines was significantly lower than its IC₅₀ in the normal control RPE and MRC5 cells; however, there was no significant difference between the IC₅₀ in different UM cell lines. Cytologic features of cell injury in the form of cytoplasmic vacuoles were seen in UM cell lines starting from 2.5 μM (ranging from 2.5 to 6.25 μM) of SU11274 (Fig. 4E) and were more prominent in OCM1, OCM3, and OCM8 cell lines. Cytologic features of cell death including cellular shrinkage, pyknosis, karyorrhexis, and karyolysis were observed in all cell lines at higher concentrations (5–10 μM). 

Figure 4.

 

Cell proliferation inhibition of uveal melanoma cell lines by the selective MET inhibitor SU11274. (A) The effect of serial dilutions (25–0.1 μM) of SU11274 on UM cell lines compared with its effect on the nontumor control cells RPE and MRC5. (B) SU11274 blocked the serum-induced activation of MET protein starting at the 2.5-μM concentration. (C) Untreated OCM3 cells and (D) OCM3 cells after overnight incubation with 2.5 μM SU11274 in growth medium. The latter showed prominent cytotoxic effects in the form of cellular vacuoles.

Table 2.

 

Half Maximum Inhibition (IC₅₀) of the Selective MET Inhibitor SU11274 in Uveal Melanoma and Control Cell Lines

Cell Line	IC50 μM (SD)*
C918	3.8 (1.04)
OCM3	2.96 (0.58)
OCM8	5.2 (1.71)
MUM2C (OCM1)	4.7 (0.91)
MEL202	2.96 (0.94)
92.1	2.5 (0.98)
RPE (ARPE-19)	>50
MRC-5	>50

 Results represent the mean (SD) of at least three different experiments each performed in triplicate at different times.

Cell Migration Assays

Cell migration assays were performed on OCM3 and C918 cell lines with a low and a high full-length MET protein expression, respectively. Both showed a significant (P < 0.0001, ANOVA) inhibition of cellular migration in both cell lines, starting at a concentration of 1.25 μM of SU11274_. The migration inhibition was slightly more prominent in the C918 cell line. However, the difference in migration inhibition was not significantly different between the two cell lines at the 1.25-μM concentration, and the migration inhibition was similar at higher concentrations (Fig. 5). 

Figure 5.

 

Inhibition of UM cell line migration by the selective MET inhibitor SU11274. Summary of the effect of increasing concentration of SU11274 on the migration of the UM cell lines. (A) Untreated cells. (B–D) Inhibition of cell migration to the bottom chamber with overnight incubation with increasing concentrations (1.25, 2.5, and 5 μM, respectively) of the selective MET inhibitor SU11274. (E) Relative migration inhibition of OCM3 and C918 cells after overnight incubation with increasing concentrations of SU11274. (F) Relative cell proliferation inhibition of the same drug concentrations as used in (E).

Discussion

Drug treatment for UM is currently limited to metastatic tumors, and the effectiveness of such therapy is highly debated. ^2,3 A recent meta-analysis of the published peer-reviewed articles indicates that there is no compelling scientific evidence of any survival benefit of any method of treatment for any subgroup of patients with metastatic UM. ³ The lack of effectiveness and high toxicity of currently used therapeutic regimens have hindered the utilization of systemic therapy for early management of UM. Identifying new therapies for UM could improve the overall prognosis of this malignancy. Our study results are in agreement with findings in earlier reports indicating the high frequency of overexpression of MET protein in UMs ^7,8 and further show that, in most UMs, MET is in the activated, phosphorylated form. Increased transcriptional activation was also evident from MET RNA assessment in the tumors. The high frequency of MET activation in UMs supports its role in the pathogenesis of these tumors, as well as its potential utility as a therapeutic target. 

We investigated the potential mechanism of MET activation in UM and studied several commonly reported mechanisms for activation of the MET gene observed in other cancers, including activating mutations in the kinase and/or the transmembrane domains, amplification of the MET gene, and indirect activation through overexpression of HGF or loss of negative regulation by the VHL gene. ^5,6,12–14 The VHL gene was selected since monosomy 3, the chromosomal location of VHL gene, is the most common cytogenetic alteration in UM. We identified no activating mutations in the MET gene in any of the primary tumor samples or the cell lines studied, suggesting that in contrast to that in cutaneous melanomas, ²⁵ activating mutations of MET play no role in MET activation in UM. We also identified a low frequency of gains in chromosome 7 in the UMs that was consistent with findings in published reports, ^26–28 suggesting that copy number alteration in MET is not the major molecular mechanism for its activation in UM. The lack of direct gene activation suggests that the MET activation in UM is mostly through an indirect mechanism. Our finding of expression of MET and pMET in the normal tissues of the tumorous eyes but not the nontumorous ones suggest a paracrine and/or autocrine mechanism for its activation in UM. In support of an autocrine mechanism for MET activation in UM, serum starving the UM cell lines did not result in any significant decrease in either MET or pMET expression. In addition, we identified coexpression of HGF in all UM cell lines studied as well as in most of the primary UMs, suggesting its importance as a potential autocrine factor for MET activation. However, the absence of a clear correlation between the degree of HGF expression and MET activation suggests the contribution of other potential autocrine/paracrine factors that should be further investigated. For example, a recent report on cutaneous melanoma suggests the contribution of melanoma chondroitin sulfate proteoglycan in enhancement of the expression of both HGF and MET, and such a role must be further studied in UM. ²⁹ Our results suggest no significant association between VHL gene haploinsufficiency and expression of MET or pMET. Such findings must be further validated because of our relatively small sample size. 

Our next task was to investigate the utility of MET inhibition as a potential therapy for UM. Using a small molecule, SU11274, that selectively inhibits the MET receptor, ²⁴ we identified significant inhibition of cell proliferation of the UM cell lines included in the study. In addition, we observed a prominent inhibition of cell migration at concentrations lower than the IC₅₀ of SU11274 on UM cell lines. The drug acted selectively against tumor cells but not normal RPE or fibroblast cells, with IC₅₀ ranging from 2.5 to 5.2 μM (Fig. 4, Table 1). The IC₅₀ values are similar to those previously reported in other tumor cell lines responsive to MET inhibition. ^21,25,30 They are also similar to the IC₅₀ of SU11274 in transfected tumor cells with activating mutant variants of the MET oncogene. ³¹  

We identified a consistent nuclear and cytoplasmic expression of pMET in most of the UMs included in our study. We performed nuclear/cytoplasmic fractionation of several cell lines that confirmed the nuclear localization of the activated pMET (Fig. 2). The nuclear localization of pMET expression is in line with the findings of other investigators who identified nuclear translocation of MET as a crucial step for downstream activation through calcium signaling. ³² Several other reports also support the nuclear localization of MET protein ^33–35 (reviewed by Hanaa et al. ³⁶ ), In our study, nuclear MET was mostly the full size protein, similar to the findings of Gomes et al. ³² in liver cancer cells. However, other investigators identified the translocation of only the 60-kDa truncated protein to the nucleus in breast and non–small cell lung cancers. ^35,37 Such differences should be investigated further. 

Several reports have suggested the existence of a proapoptotic property of MET through proteolytic cleavage by caspases to generate the 40-kDa fragment p40. ^38,39 This fragment contains the catalytic tyrosine kinase domain of MET and is not further anchored to the membrane. ⁴⁰ It is worth noting that we observed a prominent 40-kDa fragment of MET in at least two of six cell lines (C918 and MEL202; Fig. 3A). Recognizing the dual pro- and antiapoptotic function of MET is crucial when an anti-MET therapy is considered. In our study, one of the interesting findings was that the degree of cell proliferation inhibition was not directly associated with the level of total MET/pMET expression. This finding may suggest that in tumors with high MET expression, inhibition of MET may also affect its proapoptotic function, which could diminish the effectiveness of an MET inhibitor. Further studies of the contribution of the proapoptotic MET function in inducing resistance to MET inhibition as well as the utilization of combined targeted therapy are currently in progress. 

In conclusion, our study indicates that MET is activated in a large number of UMs, suggesting that it is important in the pathogenesis of UM. The results also suggest that selective MET inhibition is a potentially useful therapeutic target for treatment of UMs. Finally, our findings indicate that MET is probably activated in UM through an indirect mechanism that does not involve copy number alteration or activating mutation in the MET gene. 

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Footnotes

 Supported by the Patti Blow Research fund in Ophthalmology and a Selective Investment grant from the College of Medicine, The Ohio State University.

Footnotes

 Disclosure: M.H. Abdel-Rahman, None; G. Boru, None; J. Massengill, None; M.M. Salem, None; F.H. Davidorf, None

The authors thank Colleen Cebulla, MD, PhD, for a thorough review of the manuscript and thoughtful comments.