CD8+, HLA-unrestricted, cytotoxic T-lymphocyte line against malignant melanoma

A CD8+ cytotoxic T lymphocyte (CTL) line was derived from the peripheral blood mononuclear cells of a patient with primary melanoma. The CD8+ CTL line specifically lysed the autologous primary melanoma cells and not the natural killer cell-sensitive K562 cells or lymphokine activated killer cell-sensitive DAUDI cells. When a large panel of human leukocyte antigen (HLA)-matched and -unmatched allogeneic melanoma, glioma, breast and colorectal carcinoma cells was tested as targets in cytolysis assays, 4 HLA-matched and two HLA-unmatched allogeneic metastatic melanoma lines were lysed by the CD8+ CTL. Lysis of autologous and allogeneic melanoma cells was dependent on the effector-to-target cell ratio. Lysis of autologous melanoma cells was not blocked by anti-HLA class I or class II antibodies, confirming that the cytolytic activity of the CD8+ CTL was HLA-unrestricted. CTL lysis of autologous melanoma cells was CD3 (T cell receptor) dependent and FAS-FAS-L, and CD1 independent. Identification of the melanoma-associated antigen recognized by the HLA-unrestricted CTL may provide a vaccine for a broad population of melanoma patients.


Background
Cytotoxic T lymphocytes (CTL) have been established from various lymphoid sources of patients with metastatic melanoma [1]. The majority of CTL are of CD8 phenotype and generally lyse tumor cells in a human leukocyte antigen (HLA)-restricted manner. A number of melanomaassociated antigens and peptides defined by CD8 + CTL have been identified and the majority of these CTL are HLA-A2 restricted [1,2]. CD8 + CTL that recognized and lysed melanoma targets of various HLA types also have been described [3][4][5][6]. However, it is unclear whether the CTL are truly HLA-unrestricted, as the allogeneic tumor target cells used were not fully HLA subtyped [3][4][5][6] and, therefore, partial HLA matching of these targets with the CTL cannot be excluded with certainty. Furthermore, the demonstration of the absence of HLA restriction was based on the absence of CTL lysis blocking in the presence of high concentrations of anti-HLA antibodies. However, the high effector-to-target (E:T) cell ratios used in those studies, accompanied by high tumor cell lysis in control cultures [5,6] may be responsible for the absence of CTL blocking by anti-HLA antibodies.
We have established a CD8 + CTL line from the peripheral blood lymphocytes of a patient with primary melanoma. When the cytotoxic activity of the CTL line was tested against a large panel of allogeneic cell lines of melanoma, glioma, breast or colorectal carcinoma, autologous or allogeneic Epstein-Barr virus (EBV)-transformed B cells, or autologous fibroblasts, the autologous and a few allogeneic, HLA non-matched melanoma cells were lysed and induced interferon (IFN)-γ and granulocyte monocytecolony stimulating factor (GM-CSF) secretion by the CTL. Furthermore, the lysis of the autologous or allogeneic tumor cells was not blocked by monoclonal antibodies (MAbs) to HLA-class I or class II, confirming that the CTL lyse targets in a HLA-unrestricted manner.

Patient 793
Patient 793 (male Caucasian, 39 years old) had excision of a "low risk" primary melanoma [superficial spreading type with early vertical growth phase present; the tumor thickness was 0.55 mm and the vertical growth phase had a brisk lymphoid infiltrate, with no evidence of metastases]. The primary lesion was excised ~20 years ago and there has been no recurrence since. The patient did not receive adjuvant chemotherapy after removal of the primary lesion.

Cell lines
Melanoma cell line WM793 was established from the vertical growth phase of a primary lesion of patient 793 [7]. Cell line 1205LU, the metastatic variant of WM793, was established after repeated passages of WM793 cells both in vitro and in vivo in nude mice [8]. Melanoma cell lines WM75, WM98, WM164, and WM1158 were derived from metastatic lesions of melanoma patients [7]. All cell lines were maintained in McCoy's Dulbecco 153-Leibovitz 15 (MCDB153-L15) medium (Sigma, St. Louis, MO) containing 2% fetal bovine serum (FBS). Metastatic melanoma cell line DM196 was obtained from T. L. Darrow (Duke University Medical Center, Durham, NC) and was maintained in Dulbecco's modified Eagle's medium (DMEM; GIBCO-Invitrogen, Carlsbad, CA) supplemented with 5% FBS. Metastatic melanoma cell line ME9874 was obtained from A. Anichini (Istituto Nazionale Tumori, Milan, Italy) and maintained in RPMI 1640 Glutamax medium (GIBCO-Invitrogen) supplemented with 10% FBS. Metastatic melanoma cell line A375 was obtained from American Type Culture Collection (ATCC; Rockville, MD) and maintained in RPMI 1640 medium supplemented with 10% FBS. Rectal carcinoma cell lines WC007 and WC008 were maintained in MCDB20l-L15 (Sigma) medium supplemented with 2% FBS [9]. The glioma cell lines U373MG and U87MG (obtained from Dr. Darell Bigner, Duke University Medical Center) and the breast cancer cell line MDA MB231 (obtained from Dr. Daniela Santoli, The Wistar Institute) were maintained in DMEM medium supplemented with 10% FBS. HLA class I and II expression by tumor cells was determined in complement-dependent lysis assay using HLA typing trays (One Lambda, Inc., Canoga Park, CA). 793 fibroblast cell line [10] was maintained in DMEM containing 10% FBS.
EBV-transformed B cell line was established from freshly isolated peripheral blood mononuclear cells (PBMC) of patient 793 as described before [10]. EBV-B cell lines 888, 1363, 1088, 1102, and 4226 were established from melanoma patients' PBMC. Natural Killer (NK) cell target K562 (human erythroleukemia cell line) and lymphokine-activated killer (LAK) cell target Daudi (human lymphoblastoid cell line) were obtained from ATCC. All lymphoid cell lines were maintained in RPMI 1640 medium supplemented with 10% FBS.

Cytotoxicity assay
This assay was performed as described previously [9,10,13]. Briefly, labeled targets [2 µCi of 51 Cr (sodium chromate, specific activity 400-1200 Ci/g; NEN DuPont, Boston, MA) per 1 × 10 4 cells] were mixed in 96-well Ubottom microtiter plates with effector cells at various E:T ratios and incubated at 37°C for 4 hr. Supernatants were harvested and tested for 51 Cr release (experimental release). For maximal release, target cells were treated with 0.3% Triton X-100 (Sigma). Spontaneous release of radioactivity by target cells was determined in the absence of effector cells. Spontaneous release of the various targets used was 10-15%. The percentage of cytotoxicity was determined by the following formula: The targets (WC008, U373MG and U87MG) that exhibited high spontaneous 51 Cr release were stained with neutral red dye and used in a modified cytotoxicity assay [10].

Statistical analyses
Differences between experimental and control values (triplicates or quadruplicates) were analyzed for significance by Student's two-sided paired t-test.  Table  I). The CD8 + T cell line expressed HLA class I and II, CD3, CD25, FAS, FASL, B7-1, and B7-2, and a low percentage of the cells also expressed CD40 and CD40L (Table I). The T cell line expressed TCR α/β chains. Further, RT-PCR analysis indicated that the T cell line is oligoclonal as it expresses Vα2, Vα4, Vα14, Vα16 and Vβ6 chains. Attempts to clone CD8 + T cell line by limiting dilution were unsuccessful twice. WM793 melanoma cells expressed HLA class I and II and FAS, and a low percentage of the cells expressed FASL, B7-1, and B7-2 (Table I).

Cytolytic activity of CD8 + CTL line against various HLAmatched and non-matched tumor and non-tumor cells
The cytolytic activity of the CD8 + CTL line against autologous and allogeneic partially HLA-matched and unmatched melanoma cells was tested at various E:T ratios (6.25-50) at 8 weeks in culture. As controls, NK susceptible target (K562) and LAK sensitive target (Daudi) were used. The CTL line lysed autologous WM793 melanoma cells as well as partially HLA-matched (DM196, ME9874, A375 and WM75) and HLAunmatched allogeneic melanoma cells (WM164 and WM1158), but not NK-sensitive K562 or LAK-sensitive Daudi cells (Fig. 1). In contrast to autlogous WM793 melanoma cells, which were non-metastatic in nude mice and lysed by CTL 793, 1205LU metastatic variant cells of WM793 cells isolated from nude mice were not lysed by the CTL (Table II). Absence of 1205LU lysis by CTL793 cannot be explained by absence of the expression of known melanoma-associated antigens or HLA class I in 1205LU cells, compared to WM793 cells (Fig. 2). However, we cannot exclude the possibility that other antigens are differentially expressed by the two melanoma cell lines. Thus, the antigen recognized by the CTL may be expressed by primary (WM793), but not metastatic variant (1205LU) melanoma cells, derived from the same patient. However, expression of HLA class II was low (<5%) in 1205LU cells compared to WM793 cells (Fig. 2). Since CTL793 are CD8 + , we do not expect HLA class II to play a role in WM793 lysis by the CTL. Furthermore, there is strong evidence that CTL793 are HLA unrestricted in their capacity to lyse autologous tumor cells (see below).

Cytokine secretion of CD8 + CTL793
Supernatant from CD8 + CTL793 after 24-48 hr of stimulation with autologous melanoma cells WM793 contained significant amounts of IFN-γ, TNF-α, and GM-CSF, but not IL-2 or IL-4 (data not shown, except for IFN-γ release in Table II). IFN-γ, TNF-α, and GM-CSF were produced by the CTL, and not by irradiated autologous melanoma cells, since the supernatants derived from the melanoma cells showed no significant cytokine secretion (data not shown). To determine the specificity of cytokine release by the CTL, IFN-γ and GM-CSF production by the CTL after stimulation with various tumor and non-tumor cells was determined (Table II). All six allogeneic, partially HLA-matched melanoma cells, (DM196, ME9874, A375, WM75, WM164 and WM1158) induced significant IFN-γ and GM-CSF production by the CTL and also were lysed by the CTL (Table II). None of the cell lines derived from gliomas, breast and colorectal carcinomas, or melanocytes, fibroblasts or EBV-B lymphocytes induced IFN-γ release by the CTL (Table II).

Blocking of CTL activity of CD8 + CTL793 using various MAb
Melanoma cells WM793 express HLA-class I and II (97% and 74% of the cells positive, respectively; Table I). CTL activity against autologous WM793 cells was not blocked by pre-incubating target cells with saturating concentrations of MAb to HLA-class I (W6/32) or class II (MAb B33.1 or MAb D1.B6) ( Table III), indicating that CTL activity is HLA-unrestricted. Surprisingly, these MAbs enhanced tumor cell lysis (Table III). In our related studies, these MAbs were able to block the tumor target cell lysis of other HLA-restricted CTLs [10,13]. CTL activity against partially matched allogeneic DM196 melanoma cells was not blocked by MAb to HLA class I (data not shown), indicating that CTL activity is HLA unrestricted.
CTL793 express FAS-L and WM793 melanoma cells express FAS (Table I). To determine the role of FAS and FAS-L interaction in 793 CTL-mediated lysis of WM793 tumor targets, tumor cells were incubated with anti-FAS MAb DX2 before the CTL activity against the cells was tested. Anti-FAS MAb was unable to block 793 CTL lysis of WM793 target cells (Table III), indicating that CTL killing is independent of the FAS and FAS-L mediated pathway.
Low percentage of WM793 cells expresses CD1 (Table I).
The role of CD1 in CTL793-mediated lysis of WM793 tumor targets was tested by incubating tumor cells with anti-CD1 MAb HI149 before the CTL activity against the cell was tested. Anti-CD1 MAb was unable to block 793 CTL lysis of WM793 target cells, indicating that CD1 is not involved in CTL-mediated lysis of tumor cells (Table III).
CTL activity of 793 CTL against WM793 cells was significantly blocked by preincubating effector cells with saturating concentrations of anti-CD3 MAb (Table III). Thus, lysis of target cells is mediated by TCR.

Discussion
There are a number of reports of HLA-unrestricted CD8 + CTL derived from the lymphocytes of melanoma patients [3][4][5][6]14]. Kubo et al [14] described two CTL clones, each expressing a single TCR (αβ) and recognizing HLAmatched tumor targets in an HLA-restricted fashion and HLA-unmatched targets in an HLA-unrestricted manner. HLA-unrestricted CTL from ovarian, breast and pancreatic carcinoma patients also have been reported and they are known to recognize epitopes of mucin (MUC1) tumorassociated antigen [15][16][17]. However, it is unclear whether the CTL against melanomas and carcinomas are truly HLA-unrestricted, as target cells were not fully HLA sub-typed [3][4][5][6] and high E:T cell ratios were used in cytolysis assays [5,6] (see Introduction).
Our study provides strong evidence that the cytolytic activity of the CD8 + 793 CTL line is HLA-unrestricted. We have tested our CTL in two sets of experiments to confirm that the cytolytic activity is HLA-unrestricted. In the first set of experiments we have used a panel of HLA-matched and -unmatched melanoma targets in CTL lysis assays (Table II). The target cells were fully HLA subtyped. In the second set of experiments we have used anti-HLA antibodies to block the cytolytic activity of the CTL at low E:T cell ratios (Table III). CD8 + 793 CTL line lysed autologous melanoma cells, and four different partially HLA-matched (HLA-A1, A29 or HLA-B57 [17] alleles) and two unmatched allogeneic melanoma cells; none of the allogeneic tumor cell lines derived from glioma, breast or colon carcinoma were lysed. These results suggest that the CD8 + CTL lysed melanoma targets in an HLA-unrestricted manner. These results were confirmed in cytokine release assays and in assays where autologous tumor cells were incubated with MAb to HLA-class I or -class II prior to determining tumor cell lysis by the CTL at low E:T ratios. Unexpectedly, pre-incubation of tumor cells with MAb to HLA-class I or -class II resulted in enhancement of CTL tumor lysis (Table III). Similar enhancement of CTL target cell lysis in the presence of anti-HLA antibodies was observed by other investigators [3,4], although the mechanism of this phenomenon is unknown.
Our data rule out a role of Fas/Fas ligand interaction in tumor cell lysis by the CD8 + CTL line as MAb to FAS was unable to block CTL lysis of melanoma cells in spite of expression of FAS-L and FAS by the CTL and tumor cells, respectively (Table I). Cytotoxic reactivity of CD8 + CTL line against autologous melanoma cells The CD8 + CTL described here recognize an antigen(s) preferentially expressed by melanoma cells. This antigen was not present on allogeneic cell lines of glioma (two), breast carcinoma (one), and colorectal carcinoma (two), or allogeneic partially matched melanocytes (three) or autologous fibroblasts (one), and autologous (one) or allogeneic (five) EBV-B cells. A presumably HLA-unrestricted CTL clone recognizing a determinant shared by allogeneic melanoma and lung carcinoma cells has been described [14]. A short-term presumably HLA-unrestricted CTL line established by repeated stimulation of lymph node T cells obtained from a pancreatic carcinoma patient recognized and lysed MUC-1 positive pancreatic, breast, and ovarian carcinoma cells [15,17]. To our knowledge, MUC-1 is not expressed by melanoma cells, and thus it is highly unlikely that MUC-1 is the candidate antigen responsible for CD8 + CTL-mediated lysis.
Identification of the melanoma-associated antigen recognized by the CD8 + , HLA-unrestricted CTL described here will provide a vaccine for a broader population of melanoma patients. The antigen is expressed by the majority of allogeneic, metastatic melanoma cells, although it is lost by metastatic variant cells isolated from primary, autologous WM793 melanoma cells by passaging these cells in nude mice.
Reactivity of anti-melanoma MAbs to melanoma cells