Expression of Classic and Nonclassic HLA Class I Antigens in Uveal Melanoma

Abstract

purpose. Because the expression of classic and nonclassic HLA antigens is crucial for the recognition and elimination of tumor cells by cytotoxic T and/or NK cells, we analyzed the HLA-A, -B, -C, and -G expression in uveal melanoma specimens from 18 patients.

methods. Tumor specimens and EDTA plasma samples from 18 patients treated by primary enucleation or tumor resection for primary uveal melanoma in the University Eye Clinic, Essen, Germany, were collected immediate after surgery. After solubilization of tumor tissue and specific immunoprecipitation of classic HLA-A, -B, and -C and nonclassic HLA-G antigens the tumor samples were analyzed by one-dimensional isoelectric focusing (1D-IEF) and Western blot analysis. In parallel, patients were typed for HLA-A, -B, and -C class I antigens by PCR with sequence-specific primers (PCR-SSP). In addition, HLA-A2 and -G expression was investigated by immunohistochemistry in paraffin-embedded tumor sections from these patients.

results. In 9 (50%) of 18 specimens, a full HLA-A and -B antigen expression pattern was detected by 1D-IEF. In six (33.3%) tumor specimens, an HLA class I allotype was missing (HLA-A2, -A28, -A29, -B18, -B35, and -B55), in two cases a haplotypic loss (HLA-A2, -B44 and HLA-A2, -B13) and in another case an allotype-specific loss combined with a haplotypic loss (HLA-A26, -A32, -B41) were observed. HLA-C and -G antigens were not detectable in any of the tumor samples by biochemical methods used.

conclusions. A considerable portion of the uveal melanomas tested showed a loss of classic HLA class I antigens, which may enable them to escape from the immunosurveillance of cytotoxic T cells. HLA-C and -G antigens were not found in uveal melanoma tissue implying a susceptibility for NK lysis.

Therapy of uveal melanoma within the eye is successful in most patients. However, in 30% to 50% of the patients, metastasis occurs, and the outcome is fatal in less than 1 year after manifestation of dissemination of the tumor, most commonly in the liver. ^{1 2} To improve the survival rate of patients with uveal melanoma, an adjuvant therapy is needed. 

Immunotherapy is a promising approach, because it is based on the “natural” way to detect and destroy disseminated tumor cells by the host immune system. Cytotoxic lymphocytes, which include CD4⁺ cytotoxic T lymphocytes (CTLs), CD8⁺ CTLs, natural killer (NK) cells, and lymphokine-activated killer (LAK) cells, can kill tumor cells by using one of the following two pathways: secretion of cytotoxic cytokines or calcium-dependent or calcium-independent contact-dependent cytotoxicity. ³ During the past two decades, numerous trials of immunotherapy have been conducted in patients with malignancies. Both promising and disappointing results have been reported. However, in most tumors, mechanisms develop that enable them to escape host immune surveillance, which results in primary failure of immunotherapy or even secondary failure after an initially successful response. Many pathways of immune escape are known, including signaling defects in T cells, contact-induced anergy of T cells, and downregulation of HLA expression in tumor tissue. ⁴  

The latter is probably the best-established pathway of tumor escape, as shown in many different types of malignancy. ^{5 6} Especially in uveal melanoma there is evidence that the downregulation of HLA-A and -B antigens correlates with a favorable patient outcome, ^{7 8} which is completely different from the situation known to exist with other tumors. These data suggest a protective role of NK cells in the development of metastasis. ^{7 8} A recent study demonstrated further that uveal melanomas do not express the HLA-G antigen, which is a nonclassic HLA class I molecule known to inhibit NK cell–mediated cytotoxicity. ⁹ Although the body of evidence is increasing that influencing the NK cell pathway of cytotoxicity would be the most promising immunotherapy in patients with uveal melanoma, no established approach to perform this in a clinical trial exists. Therefore, immunotherapy based on HLA class I restricted CTLs remains an alternative when planning an adjuvant therapy in these patients. 

Crucial for an immune-response–based immunotherapy (e.g., vaccine) is the expression pattern of HLA-class I molecules in the tumor tissue. For example, most of the vaccines used are HLA-A2 restricted. For this reason, it is very important to know the HLA-A2 expression pattern in the targeted tumor. Very little is known about the allotypic HLA expression in uveal melanomas. Some information was obtained by two previous immunohistochemical studies in tumor specimens and cell lines, using a small panel of antibodies (specific for A-locus A2, A3; B-locus Bw4, Bw6). ^{10 11} Recently another study of cell lines from uveal melanomas in which a broader panel of antibodies and flow cytometry were used demonstrated a high frequency of allotype-specific downregulation of HLA class I molecules. ¹²  

In the present study we screened 18 primary uveal melanomas for the expression of HLA-A, -B, -C, and -G antigens using another approach, one-dimensional isoelectric focusing (1D-IEF) and SDS-PAGE, followed by immunoblot analysis allowing an allotypic expression analysis. 

Materials and Methods

Human Subjects and Tissue Samples

Frozen tissue from 18 patients was obtained immediately after surgery (enucleation, n = 14; tumor resection, n = 4) for uveal melanoma performed in the University Eye Clinic at Essen. The remaining part of the tumor (and the eye globe) was embedded in paraffin. Blood samples were collected after surgery. The research protocol adhered to the tenets of the Declaration of Helsinki. Histologic examination showed that 12 tumors involved the ciliary body and 6 were localized in the choroid. The mean of the largest tumor diameter was 13 mm. Twelve of the tumors consisted of spindle cells, five were of mixed cell type, and one consisted predominantly of epithelioid cells. In two cases, extraocular tumor growth was manifested. 

Monoclonal and Polyclonal Antibodies and Cell Lines

For immunoprecipitation of HLA-A, -B, and -C antigens, mAb W6/32, which recognizes a conformational epitope of HLA class I antigens in connection with β2-microglobulin (β2-m), was used. To immunoprecipitate HLA-G antigens specifically, mAb BFL1.1 (Coulter-Immunotech Diagnostic, Hamburg, Germany) and mAb MEM/G9 (Biozol, Munich, Germany) were used. After Western blot analysis, HLA-A and -B, antigens were visualized by a rabbit antiserum toward denatured HLA class I heavy chain (RaHC) ¹³ and an alkaline phosphatase-conjugated goat anti-rabbit IgG (Dianova, Hamburg, Germany). For the detection of HLA-C and -G antigens in Western blot analysis, two rabbit antisera (RaHLA-C and RaHLA-G) were generated against peptides with amino acid sequences DRETQKYKRQAQ and EEETRNTKAHAQTDRM specific for the HLA-C α1 domain on residue 85-96 and for HLA-G α1 domain on residue 63-81, respectively (Eurogentec, Seraing, Belgium). The specificity of RaHLA-C was validated by Western blot after mAb W6/32 immunoprecipitation and 1D-IEF, by 51 B-lymphoblastoid cell lines (B-LCLs) with known HLA-A, -B, and -C phenotypes. ¹⁴ The HLA-C typing of the cell lines covered HLA-C allotypes from HLA-C*01 to HLA-C*17. The results showed that RaHLA-C recognized all HLA-C allotypes, except HLA-C*07 and -C*15 but cross-reacted with HLA-B46. Antiserum RaHLA-G recognized all HLA-G isoforms without cross-reaction with classic HLA-A, -B, and -C. 

For the detection of HLA-A2 in paraffin-embedded tissue sections, the anti-HLA-A2 mAbs 0397HA and 0791HA (One Lambda, Canoga Park, CA) were used. Pilot studies revealed that these mAbs cross-react with HLA-A9 (A23 and A24) in immunoprecipitation experiments (data not shown). HLA-G antigens were detected in paraffin-embedded tissue sections by mAb BFL1.1 (Coulter-Immunotech Diagnostic). 

Two human B-lymphoblastoid cell lines with known HLA class I types (HLA-A2, -A24, -B8, -B61 and HLA-A2, -A29, -B7, -B44) served as reference markers (M1 and M2) in 1D-IEF. The choriocarcinoma cell line JEG3, which expresses HLA-G antigens served as the control in HLA-G–specific immunoblots. The B-LCL AMAI (HLA-A28,-; B53,-; Cw*04,-) was used as a positive control for the HLA-C–specific immunoblots. ¹⁴ All cell lines were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, 20 mM l-glutamine, penicillin (20,000 U/L), and streptomycin (20,000 U/L) at 37°C in a moist atmosphere containing 5% CO₂. 

Biochemical Methods

After mechanical homogenization of frozen tissue specimens (approximately 100 mg) membrane-anchored antigens were solubilized in 1 mL lysis buffer (1% TX114, 50 mM Tris, 5 mM MgCl₂, 2 mM phenylmethylsulfonyl fluoride [pH 7.5]) for 30 minutes at 4°C. Cell fragments and DNA were pelleted by centrifugation at 10,000g at 4°C, and the supernatant was transferred to a fresh tube. The supernatant containing solubilized antigens was incubated for 5 minutes at 37°C, followed by centrifugation at 300g for 10 minutes. The upper aqueous phase was discarded and the detergent phase, enriched with cell-surface–anchored molecules, was filled up to 1 mL with phosphate-buffered saline (PBS; pH 7.2). The protein concentration of each tissue lysate was determined with the bicinchoninic acid protein assay (BCA; Pierce, Rockford, IL). All tissue lysates were diluted with PBS to a fixed protein concentration of 0.1 mg/mL. The reference cell lines M1 and M2 were treated as the tissue specimens. 

To precipitate solubilized HLA class I molecules, 1 mL of detergent lysate was incubated with 100 μL of immunomagnetic beads (Dynabeads TM M280; Dynal, Hamburg, Germany) precoated with mAb W6/32 (5 μg/mL), mAb BFL1.1 (10 μg/mL), or mAb MEM/G9 (10 μg/mL) overnight at 4°C under rotation. For 1D-IEF analysis the beads were treated with 0.2 U neuraminidase type VIII (Sigma-Aldrich Chemie GmbH, Deisenhofen, Germany) in 50 mM sodium acetate (pH 5.5) and 1 mM CaCl₂ for 5 hour at 37°C. Then 50 μL of 1D-IEF sample buffer (9.5 M urea, 2% Ampholyte [Pharmacia Biotech AB, Uppsala, Sweden; pH 3.5–10], 2% Triton X-100) was added. For SDS-PAGE analysis the beads were incubated for 5 minutes at 95°C with 1% SDS and 150 mM β-mercaptoethanol. 

1D-IEF, SDS-PAGE, and Western blot analysis were performed as described previously. ^{15 16} Bound HLA-A, -B, and -G antigens were detected by either an overnight incubation with antiserum RaHC or RaHLA-C diluted to 1:10.000 in PBS-Tween (0.05%) and with alkaline phosphatase-conjugated goat anti-rabbit IgG diluted 1:1000 in PBS-Tween (0.05%) for 1 hour. Bound antibodies were visualized by nitroblue tetrazolium (0.1 mg/mL), 5-bromo-4-chloro-3-indolyl-phosphate (0.05 mg/mL), and MgCl₂ (1 mM) in Tris-HCl (0.1 M, pH 9.6). 

HLA Class I Typing

HLA class I typing of the patients was performed with DNA samples isolated from frozen EDTA blood specimens by PCR with sequence-specific primers (PCR-SSP), as described previously. ¹⁷  

Immunohistochemical Methods

Immunolocalization of expression of HLA-G and -A2 antigens was performed by an immunohistochemical technique specifically adapted to detect classic MHC class I molecules in formalin-fixed and paraffin-embedded tissue sections. ¹⁷ Briefly, 2-μm tissue sections were placed on triethoxysilylpropylamin-coated glass slides (2%; Merck, Darmstadt, Germany), dried at 37°C overnight, and subsequently dewaxed in xylene. After rehydration through a graded series of ethanol, tissue sections were incubated in antigen-retrieval solution (1 mM EDTA in dH₂O; pH 8.0), microwave heated up to 100°C for 5 minutes at 300 W, and rinsed twice in dH₂O and 50 mM Tris-buffered saline (TBS). Endogenous peroxidase was blocked by treatment with 0.05% H₂O₂ for 10 minutes at room temperature, followed by washing in TBS. Slides were subsequently incubated with mAbs 0397HA and 0791HA (both, 1:100) and BFL1.1.2 (1:75) followed by biotinylated rabbit anti-mouse Ig Ab (at 1:300, Dako, Glostrup, Denmark) and streptavidin-biotin-complex/horseradish peroxidase (Dako). After sections were washed in TBS, signals were amplified with biotinylated tyramide (diluted 1:500, Dako) activated by H₂O₂ (at 1:1000) and visualized by subsequent incubation with streptavidin-biotin-complex/alkaline phosphatase (Dako) and the chromogen neofuchsin, according to the manufacturer’s instructions. Finally, sections were counterstained with hemalum. Negative controls included omission of the first, second, and/or third antibody, and replacement of the primary antibody with rabbit preimmune or nonimmune IgG or monoclonal mouse IgG of irrelevant specificity. 

Results

Expression Analysis of Classic HLA-A, -B, and -C Antigens

HLA expression in uveal melanoma tissue samples was studied after immunoprecipitation with mAb W6/32 by 1D-IEF and Western blot. As summarized in Figure 1 and Table 1 9 (50%) of 18 tissue specimens presented the full HLA-A and -B expression pattern. A selective HLA allotype-specific loss was found in six (33.3%) tissue specimens, which three affected the HLA-A locus (HLA-A2, A28, and A29) and three the HLA-B locus (HLA-B18, B35, B55). In two tumor specimens (2 and 8) a haplotype (HLA-A2, B44 and HLA-A2, B13, respectively) and, in another sample (11), a complete HLA-A locus (HLA-A26, -A32) with an HLA-B allotype (HLA-B41) expression were missing. 

The immunohistochemical examination of HLA-A2 in six tissue samples revealed a homogenous expression (>75% of the tumor cells) in three samples, a heterogeneous expression (25%–75% of the tumor cells) in two, and negative expression in one. The latter tumor (2) also showed an absence of HLA-A2 expression in biochemistry assays, whereas tumor 18, also negative in biochemistry, stained very weakly (<25% of the tumor cells) with anti-HLA-A2 (Table 1) . 

Because HLA-C molecules interact preferentially with NK cell receptors, all tumor specimens were also studied for HLA-C expression using the specific RaHLA-C antiserum as detection reagent after Western blot analysis. However, in none of the tumor tissues was HLA-C surface expression detectable (data not shown). 

Expression Analysis of Nonclassic HLA-G Antigens

The expression analysis by HLA-G specific immunoprecipitation and Western blot analysis revealed a positive signal in none of uveal melanoma tissue specimens. The immunohistochemical examination of 17 tumors showed that only two tumors (3 and 4) had heterogeneous expression (25%–75% of the tumor cells) of HLA-G, whereas the remaining tumors were negative (data not shown). 

Discussion

In the present study only 50% of the uveal melanoma tissues showed a complete HLA-A and -B expression pattern, whereas the other half of the tumors represented losses of HLA-A and/or -B allotypes. Change in HLA expression is a well known phenomenon in tumors and, even in uveal melanomas, many groups have reported a downregulation of HLA class I antigens which, surprisingly, has been associated with a better prognosis. ^{7 18} The downregulation of HLA class I antigens is a well known mechanism of immune escape. Thus, when planning immunotherapy it is important to know the HLA allotype expressed in the individual tumors. ⁴ The present study underlines once again the specific difficulties for T-cell–based immunotherapy in uveal melanoma, because every second tumor showed a loss of HLA-A and/or -B antigens. 

In the past, attempts to determine the allele-specific HLA expression in uveal melanoma have been based on immunohistochemical methods ^{8 10 11 19} and flow cytometry. ¹² These studies showed that uveal melanomas had a differential HLA expression. However, due to methodological limitations (e.g., availability of specific antibodies) only a small panel of allotypes could be studied. In the present study we were able to screen the allele expression of HLA-A and -B allotypes by a biochemical approach, resulting possibly in more reliable expression data compared with allotypic monoclonal HLA antibodies that often cross-react with various HLA antigens. 

The role of NK cells in the biology of uveal melanoma has been the subject of an intriguing long-lasting discussion during the past years. ²⁰ In clinicopathological studies ^{7 18} and using experimental models, ^{21 22 23} investigators have argued that uveal melanoma cells are possibly susceptible to NK-cell–mediated lysis. Altered HLA expression is a possible mechanism to influence NK-cell–mediated lysis. For example, HLA-C or -G gene products are able to inhibit NK cell function. Therefore, we looked in addition for the expression of these HLA antigens in our tumor specimens. HLA-C antigens were not detected in any of the tumors studied, and thus it is unlikely that these antigens play a role in immunology of this tumor. To our knowledge this is the first report on analysis of HLA-C expression in uveal melanomas. HLA-G was not detected by biochemistry or immunohistochemistry; only in 2 of 17 tumors did a portion of tumor cells stain positive. Recently, Hurks et al. ⁹ found no HLA-G in uveal melanoma tissue of 17 tumors and in 11 tumor-derived cell lines. The authors also analyzed the HLA-E expression, which may also inhibit NK cells through HLA-G, and they found a broad expression of that antigen. Although we detected some HLA-G–positive tumor cells by immunohistochemistry, in most of the specimens HLA-G was absent, thus confirming the report of Hurks et al. ⁹ The absence of HLA-C and -G expression in uveal melanomas does not directly mean that these tumors are susceptible to NK-cell–mediated lysis but at least indicates that these HLA molecules are not involved in the immunobiology of this tumor. 

In conclusion, the results of the study indicate that 50% of the uveal melanomas had lost one or more HLA-A and/or -B allotype and that HLA-C and -G molecules were practically absent in these tumors. For planning immunotherapy, these results are of importance and indirectly suggest that NK-cell–based lysis may be more effective in destroying uveal melanoma cells than T-cell–mediated cytotoxicity. 

²  Contributed equally to the work and therefore should be considered as equivalent authors.

 Submitted for publication August 12, 2002; revised October 16, 2002; accepted December 9, 2002.

 Disclosure: G. Anastassiou, None; V. Rebmann, None; S. Wagner, None; N. Bornfeld, None; H. Grosse-Wilde, None

 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.

 

Western blot analysis of 1D-IEF analysis summarizing the expression profile of HLA-A, and -B antigens in uveal melanoma from 18 patients. Missing HLA-A or -B allotypes are circled. M1 and M2 are reference cell lines with known HLA type positions, which served as markers in 1D-IEF. Left: isoelectric points (IP) of HLA-A2.2 and HLA-B8.

Table 1.

 

Tumor Characteristics and Expression Analysis of Classic HLA Class I Antigens in 18 Uveal Melanomas

Specimen	Tumor Characteristics							Allotypic HLA Class I Expression Analysis
		Location	Histology	Diameter (mm)	Extraocular Growth	1D-IEF*							SDS-PAGE			IHC, †
										HLA-A	HLA-A	HLA-B	HLA-B	HLA-C	HLA-C	HLA-A2
1	Choroid	Spindle	10	No	2	29	18	44	07	—	+
2	Choroid/ciliary body	Mixed	13	No	1	2	7	44	0702	—	Neg
3	Choroid/ciliary body	Mixed	10	No	1	29	8	44	0201	—
4	Choroid/ciliary body	Spindle	17	No	3	23	49	—	n.t.	n.t.
5	Choroid	Spindle	14	No	11	26	27	55	0202	0303
6	Choroid	Mixed	19	Yes	3	26	7	18	0701
7	Choroid/ciliary body	Spindle	11	No	2	11	62	—	01	0302	+++
8	Ciliary body/iris	Spindle	—	No	2	24	13	35	04	0702
9	Choroid/ciliary body	Spindle	11	No	1	—	8	60	0304	0701
10	Choroid/ciliary body	Spindle	12	No	3	28	35	62	0304	04
11	Choroid	Mixed	10	No	26	32	41	14	0302	1701
12	Choroid/ciliary body	Spindle	11	No	23	25	18	49	0701	—
13	Choroid/ciliary body	Spindle	12	No	2	11	52	55	0303	—	+++
14	Choroid/ciliary body	Spindle	13	No	1	28	8	55	0202	0701
15	Choroid/ciliary body	Spindle	12	No	28	—	55	8	01	—
16	Choroid	Spindle	10	No	11	25	18	—	0701	—
17	Choroid	Mixed	15	Yes	2	—	7	35	04	0702	+++
18	Choroid/ciliary body	Epithelioid	20	No	2	3	35	44	04	0501	+