Uveal Melanoma: No Expression of HLA-G

Abstract

purpose. To investigate whether uveal melanoma cells express HLA-G, a nonclassical HLA class I molecule that has been shown to be a critical mediator in the inhibition of natural killer (NK) cell–mediated cytolysis.

methods. Eleven human uveal melanoma cell lines were analyzed for the expression of HLA-G by flow cytometry, immunocytochemistry, Western blot analysis, and RT-PCR followed by Southern blot analysis. Two HLA-G–specific monoclonal antibodies were used, 87G and MEM-G/1. In addition, HLA-G expression was determined on frozen tissue sections of 17 primary uveal melanomas.

results. With all HLA-G detection methods, no evidence for HLA-G expression by uveal melanoma cells was found. In contrast, the trophoblast cell line JEG-3 clearly expressed HLA-G transcripts and protein in all cases. Furthermore, interferon-γ did not induce HLA-G expression in the uveal melanoma cell lines. Notably, all cell lines expressed HLA-E, and this expression was significantly enhanced by interferon-γ.

conclusions. Because none of the uveal melanoma cell lines nor any of the primary uveal melanomas displayed expression of HLA-G, it is unlikely that HLA-G plays a role, direct or indirect, in the modulation of cellular immunity against uveal melanoma tumors.

HLA-G is a nonclassical HLA class I molecule, preferentially expressed in fetal placental tissues. In contrast to the classical HLA-A, -B, and -C class I molecules, HLA-G is characterized by tissue-restricted distribution and limited polymorphism. ¹ An important function of HLA-G in immunosurveillance is the inhibition of natural killer (NK) cell–mediated cytolysis. This inhibition can be achieved either by direct interaction of HLA-G with inhibitory receptors expressed by NK cells such as ILT-2, p49, and KIR2DL4, ^{2 3 4} or indirectly through HLA-E. It has been shown that a peptide derived from the leader sequence of HLA-G, similar to leader sequence peptides of other HLA molecules, associates with the peptide-binding groove of HLA-E. This allows the HLA-E molecules to reach the cell surface, and, as such, they are involved in the inhibition of NK cell–mediated cytolysis by interacting with the c-type lectin inhibitory receptor CD94/NKG2A. ^{5 6} In addition to its NK cell inhibitory capacities, HLA-G has been shown to inhibit T-cell functions, at the level of proliferation, as well as cytolysis. ^{7 8} Because of these immunosuppressive properties, HLA-G expression may be involved in tumor immunosurveillance, especially because tumors frequently display altered HLA class I phenotypes. 

Uveal melanoma is the most common primary malignant intraocular tumor in adults, and carries a high mortality rate, due to liver metastases. Concerning the expression of HLA, uveal melanoma substantially differs from other tumor types: global HLA class I downregulation is rare ⁹ and a high expression of HLA-A and HLA-B on the primary tumor is correlated with a worse patient survival. ¹⁰ There is evidence that metastatic tumor cells that leave the eye and enter the bloodstream to settle down in the liver display increased HLA class I expression. ^{11 12} These data suggest that NK cells may play a protective role in the development of metastatic disease by killing tumor cells with low HLA class I expression. Murine studies support this hypothesis. ¹³  

As for the expression of HLA-G and -E in uveal melanoma, no data are yet published. Therefore, the purpose of the present study was to investigate whether HLA-G could play a role in immunosurveillance of uveal melanoma. To firmly establish the presence of HLA-G in uveal melanoma cells, we used several detection methods and analyzed uveal melanoma cell lines and tissue sections of primary uveal melanomas. Because HLA-G can function through HLA-E, the tumor cell lines were also tested for HLA-E expression. 

Materials and Methods

Uveal Melanoma Cell Lines and Tumor Specimens

Eleven human uveal melanoma cell lines were tested for HLA-G expression. Eight cell lines (92-1; Mel-202, -270, -285, and -290; and OCM-1, -3, and -8) were obtained from primary uveal melanomas. Cell line OMM-1 was derived from a uveal melanoma skin metastasis, and cell lines OMM-1.3 and -1.5 were obtained from a liver metastasis in the same patient from which cell line Mel-270 was derived. Cell line 92-1 was established in our laboratory, ¹⁴ cell lines Mel-202, -270, -285, and -290 and OMM-1.3 and -1.5 were the generous gift of Bruce R. Ksander (Schepens Eye Institute, Harvard Medical School, Boston, MA). Cell lines OCM-1, -3, and -8 were kindly provided by June Kan-Mitchell (University of California, San Diego, CA) and cell line OMM-1 by Gregorius P. M. Luyten (Rotterdam University Hospital). The cell lines were cultured in RPMI 1640 or DMEM (Gibco, Paisley, UK) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 3 mM l-glutamine (Gibco) and 2% penicillin-streptomycin (Gibco). Cell lines were passaged weekly, with a maximum passage number of 30. 

Frozen sections were obtained from 17 primary uveal melanomas. The research protocol followed the tenets of the Declaration of Helsinki. Two tumors were in the ciliary body, and the other tumors were in the choroid. Five tumors consisted of only spindle cells, whereas in the other tumors, both spindle and epithelial cells were present. Tumor diameter and prominence varied from 6 to 24 mm (mean, 11.9 mm) and from 2 to 10 mm (mean, 6.0 mm), respectively. 

HLA-G Antibodies

In our study, we used the 87G antibody, kindly provided by Myriam Onno, (University of Rennes, France). This murine IgG2a antibody recognizes membrane-bound and soluble HLA-G molecules and showed no apparent cross-reactivity with other class I molecules when tested on diverse HLA class I–transfected cells. ¹⁵  

In addition, we used the new murine HLA-G–specific antibody MEM-G/1, raised against denatured recombinant HLA-G protein. Western blot analyses were performed to test the specificity of this MEM-G/1 antibody. Several recombinant human proteins including HLA-A2, -B8, -Cw3, -Cw4, -Cw6, -Cw7, -E, and -G were subjected to electrophoresis and subsequent Western blot analysis. The MEM-G/1 antibody detected exclusively recombinant human HLA-G protein. In addition, we tested cell lysate of MHC class-I–negative 721.221 cell lines transfected to express each of these MHC class-I molecules individually. Immunoreactivity was found exclusively in case of HLA-G–transfected 721.221 cells. 

Flow Cytometry

Single-cell suspensions of 11 human uveal melanoma cell lines were subjected to flow cytometry. Cells were first incubated with the HLA-G–specific 87G antibody (diluted 1:50 in 1% PBS-BSA) or with the W6/32 antibody (1:40) for monomorphic HLA class I staining. After washing, cells were incubated with FITC-conjugated rabbit anti-mouse immunoglobulin (1:50; Dako, Glostrup, Denmark). Both incubations were performed for 1 hour on ice. Measurements and subsequent evaluation of the data were performed using a flow cytometer (FACScan; BD Biosciences, Mountain View, CA). The JEG-3 trophoblast cell line was used as a positive control. 

Western Blot Analysis

For protein extraction of the uveal melanoma lines, cells were freshly harvested from culture, washed with PBS, and lysed for 30 minutes on ice, using 20 mM Tris-HCl (pH 7.4) 50 mM NaCl, and 1% Nonidet-P40, with 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml leupeptin (Roche Molecular Biochemicals, Indianapolis, IN) and 100 U/ml aprotinin (Calbiochem, La Jolla, CA). Debris was removed by centrifugation at 13,000g for 10 minutes at 4°C. Protein concentrations were determined using a detergent-compatible protein assay (Bio-Rad, Hercules, CA). Soluble protein (10 μg) was separated by electrophoresis on 15% polyacrylamide gels containing SDS and blotted onto nitrocellulose. After blocking (SuperBlock solution; Pierce, Rockford, IL) for 2 hours at room temperature (RT), the filters were incubated with HLA-G–specific MEM-G/1 antibody for 1 hour at RT, washed, and incubated with horseradish peroxidase–conjugated secondary goat anti-mouse antibody for 45 minutes at RT. The membranes were then washed again and incubated with substrate (SuperSignal; Pierce) for 10 minutes at RT in the dark. Then films (Biomax Light; Eastman Kodak, Rochester, NY) were exposed to the filters for up to 5 minutes at RT and subjected to automated development. The JEG-3 trophoblast cell line and HLA-G–transfected 721.221 cells were used as a positive control. 

RT-PCR and Southern Blot Analysis

Total RNA was isolated from 11 subconfluent uveal melanoma cell cultures using an RNA extraction method (RNAzol; Cinna/Biotecx Laboratories, Houston, TX). RNA samples (5 μg) were transcribed into cDNA using avian myoblastosis virus (AMV) reverse transcriptase (Promega, Madison, WI). The cDNA was subjected to a semiquantitative PCR using the following HLA-G1–specific oligonucleotide primers: forward primer 5′-AGACGCCAAGGATGGTGGTCA-3′ and reverse primer 5′-AGGAAAGGTGATTGGGGAAGG-3′. PCR amplifications (2 μl cDNA in a total reaction volume of 100 μl) were run at 94° for 1 minute, at 60° for 1 minute and at 72° for 1 minute for 35 cycles. The amplification of reduced glyceraldehyde-phosphate dehydrogenase (GAPDH) was performed in the same manner to check RNA quality. The JEG-3 trophoblast cell line was used as a positive control. Similarly, HLA-E mRNA levels were determined, using the forward 5′-TCCGAGCAAAAGTCAAAT-3′ and the reverse 5′-AGATCCAAGGAGAACCAG-3′ primers. 

PCR amplification products were separated by electrophoresis on a 2% agarose gel and then, in the case of HLA-G, additionally blotted onto a membrane (Hybond N+; Amersham, Aylesbury, UK). The blot was hybridized with a ³²P-labeled HLA-G probe and analyzed by phosphorescent imaging (PhosphorImager, Molecular Dynamics, Sunnyvale, CA). 

Immunocytochemistry and Histochemistry

Frozen sections of 17 primary uveal melanomas and cytospin preparations of 11 uveal melanoma cell lines (fixed in acetone at 4°C for 10 minutes) were washed in PBS and incubated for 1 hour with the HLA-G–specific antibodies 87G or MEM-G/1, or with 1% BSA-PBS as a negative control. After washing, the slides were then incubated with biotinylated anti-mouse immunoglobulins (LSAB-2 kit; Dako) for 30 minutes, washed again and incubated with horseradish peroxidase–conjugated streptavidin (LSAB-2 kit) for 30 minutes. The peroxidase reaction was developed using 5% 3-amino-9-ethyl-carbazole (Sigma, Diesenhofen, Germany ) in 0.1 M sodium acetate buffer (pH 5) containing 0.05% H₂O₂. The sections were counterstained with Mayer’s hematoxylin and finally embedded in Kaiser’s glycerin. Cytospins of JEG-3 cells were used as a positive control. 

Results

The purpose of the study was to investigate whether uveal melanoma cells express HLA-G. Therefore we examined 11 human uveal melanoma cell lines, significantly varying in HLA class I expression ¹⁶ and metastatic potential. ¹³ Flow cytometry analyses to determine HLA-G surface expression, using the 87G antibody, showed that none of the uveal melanoma cell lines displayed HLA-G expression, as exemplified in cell lines Mel-270 and OMM-1.5 (Fig. 1) . The JEG-3 trophoblast cell line, which served as the positive control, clearly showed expression of HLA-G. Western blot analysis using the MEM-G/1 antibody, showed that in all uveal melanoma cell lines HLA-G protein expression was absent, comparable to the HLA-G–negative controls, Jar and 721.221 cells. In contrast, both in JEG-3 and HLA-G–transfected 721.221 cells, HLA-G protein was easily detectable. The absence of HLA-G expression as determined with the MEM-G/1 antibody, was confirmed by immunocytochemistry, using the 87G antibody. 

To determine whether the absence of HLA-G protein expression correlated with the absence of HLA-G transcription, RT-PCR was used. None of the uveal melanoma cell lines expressed HLA-G transcripts. In contrast, varying levels of HLA-E mRNA transcripts were detected in all cell lines. Culturing the cells in the presence of IFN-γ (200 U/ml, 48 hours) did not result in the induction of HLA-G mRNA in the uveal melanoma cell lines (Fig. 2) . To increase the sensitivity, a Southern blot analysis was performed with the PCR products. With the exception of the control JEG-3 cell line, no HLA-G was detected. Notably, HLA-E expression was increased after IFNγ treatment in all cell lines investigated (Fig. 2) . 

In addition to the uveal melanoma cell lines, 17 primary uveal melanomas, showing a large heterogeneity in clinical and histopathologic tumor characteristics, were examined for HLA-G expression by immunohistochemistry. Comparable to the cell lines, no evidence for HLA-G expression by uveal melanoma cells was found. Cytospins of JEG-3 cells were reactive with the HLA-G antibodies. Summarizing, from cell surface expression to the transcription level, not one of the uveal melanoma cell lines and not one of the primary uveal melanomas expressed HLA-G. 

Discussion

In the present study, HLA-G1 expression in uveal melanoma cells was studied. In our extensive analysis of a heterogeneous group of uveal melanoma cell lines and primary tumors, uveal melanoma cells did not express HLA-G. In contrast, in all cell lines HLA-E transcripts were detected. This suggests that in uveal melanoma HLA-G does not play a role, direct or indirect through HLA-E, in the tumor’s escape from cellular immunity. 

In cutaneous melanoma, Cabestre et al. ¹⁷ and Paul et al. ^{18 19} have provided data that support HLA-G expression in cells lines and primary tumors of cutaneous melanoma. ^{17 18 19} Similar studies by other investigators, however, did not find evidence for HLA-G expression by skin melanoma cells, either at the transcription level or at the protein level. ^{20 21 22} Studies of expression of HLA-G by different tumor types report heterogeneous results, ranging from the presence of HLA-G mRNA transcripts in the majority of tumors to a complete absence of HLA-G transcription in all tumors analyzed. ^{20 21 23} HLA-G protein expression, however, is only rarely found. ^{18 20 21 23} From this standpoint, it is not surprising that we did not find HLA-G protein expression in our present study, but the absence of HLA-G transcription may imply a new difference in HLA expression characteristics between uveal and cutaneous melanoma. 

In general, tumor cells have all kinds of strategies to escape immunosurveillance, such as downregulation of HLA class I expression, generation of an immunosuppressive microenvironment, and possibly the expression of HLA-G. In contrast to most other tumor types, uveal melanomas arise in an immune-privileged site. Therefore, their adjustments to survive may be quite different from that of other tumors. In uveal melanoma, a complete downregulation of HLA class I or transporters associated with antigen processing (TAP) is rare. ^{9 24} Because surface expression of HLA-E is dependent on an adequate level of HLA class I expression and an intact TAP mechanism, ²⁵ uveal melanoma cells have the capacity to express HLA-E on the cell surface. This is supported in our present study in which we demonstrated HLA-E expression in uveal melanoma cell lines at the transcription level. This suggests that uveal melanoma cells can escape lysis by CD94/NKG2A-positive NK cells through HLA-E and may not need to express HLA-G. 

Furthermore, besides an MHC class I–mediated inhibition of NK cell function, uveal melanoma cells may escape NK cell immunosurveillance by the production of macrophage migration-inhibitory factor (MIF). MIF, produced by numerous cells in the eye and present in the aqueous humor, has been shown to protect intraocular tumor cells against NK cell–mediated lysis. ^{26 27} Recently, Repp et al. ²⁸ demonstrated that uveal melanoma cells themselves, especially metastatic cells, also can produce bioactive MIF. By this mechanism, as well as by the expression of HLA class I, uveal melanoma cells can gain protection against NK cells outside the immune-privileged environment of the eye. Of note is that in cutaneous melanoma, MIF expression by tumor cells has been correlated with tumor progression. ²⁹  

In conclusion, in this study no evidence was found for HLA-G expression by uveal melanoma cells. This suggests that, in uveal melanoma, there is no role, either direct or indirect through HLA-E, for HLA-G in the inhibition of NK cell–mediated lysis. 

 Supported by Algemene Nederlandse Vereniging ter Voorkoming van Blindheid, The Netherlands; the Jan Dekker Foundation and Dr. Ludgardine Foundation, The Netherlands; and the J. A. Cohen Institute for Radiopathology and Radiation Protection (IRS), The Netherlands.

 Submitted for publication July 6, 2001; revised August 21, 2001; accepted September 5, 2001.

 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.

 

Figure 1.

 

Flow cytometry of monomorphic HLA class I and HLA-G surface expression in uveal melanoma cell lines Mel-270 (primary) and OMM-1.5 (metastatic) and in trophoblast cell line JEG-3 (positive control). Cells were labeled with W6/32 antibody (anti-HLA class I) or 87G antibody (anti-HLA-G), followed by FITC-conjugated rabbit anti-mouse immunoglobulin (filled traces). Open traces: isotype antibody control.

Figure 2.

 

RT-PCR analyses of HLA-G and HLA-E transcripts in 11 uveal melanoma cell lines (lanes 1–22), B cells (lane 23), and JEG-3 trophoblast cells (lane 24, positive control). Uveal melanoma cells were either untreated (odd numbers) or cultured in the presence of IFNγ (equal numbers). The following uveal cell lines were tested: 92-1 (lanes 1, 2), Mel-202 (lanes 3, 4), Mel-270 (lanes 5, 6), Mel-285 (lanes 7, 8), Mel-290 (lanes 9, 10), OCM-1 (lanes 11, 12), OCM-3 (lanes 13, 14), OCM-8 (lanes 15, 16), OMM-1 (lanes 17, 18), OMM-1.3 (lanes 19, 20), and OMM-1.5 (lanes 21, 22).