High Avidity Melanoma-reactive Cytotoxic T Lymphocytes Are ...

Vol. 6, 1459 –1467, April 2000

Clinical Cancer Research 1459

High Avidity Melanoma-reactive Cytotoxic T Lymphocytes Are Efficiently Induced from Peripheral Blood Lymphocytes on Stimulation by Peptide-pulsed Melanoma Cells1 Nadine Gervois,2 Nathalie Labarriere,2 Soizic Le Guiner, Marie-Christine Pandolfino, Jean-François Fonteneau, Yannik Guilloux, Elisabeth Diez, Brigitte Dreno, and Francine Jotereau3 Institut de Biologie, INSERM U463, 44093 Nantes Ce´dex 1 [N. G., N. L., S. L., J-F. F., Y. G., E. D., F. J.]; Unite´ de the´rapie cellulaire et geńique, CHRU de Nantes, 44093 Nantes [M-C. P.]; and Departement de dermatologie du CHRU de Nantes, 44093 Nantes [B. D.], France

ABSTRACT To design an efficient procedure to expand high avidity melanoma reactive T cells and to perform immunotherapies, we compared conditions of peripheral blood lymphocyte (PBL) stimulation by Melan-A/MART-1 peptides. Avidity of induced CTLs was evaluated by measuring their lysis and cytokine secretion to peptide-pulsed transporter-associated protein-deficient cells and to melanoma cells. In side-by-side experiments, we show that melanoma cells, either allogeneic or autologous, induced the growth of high avidity Melan-Areactive CTLs from all donors, whereas essentially low avidity T cells were induced by peptide-pulsed PBLs. We also show that at least two cytokines, interleukin-6 and interleukin-2, were required to promote the growth of high avidity CTLs. Once sorted by tetramer labeling or cloning, the specificity and reactivity to tumor cells of peptide-specific T cells induced by allogeneic melanoma cells were confirmed. We then describe a relatively simple and efficient procedure that allowed us to obtain systematically high amounts (in the range of billion) of high avidity Melan-A/ MART-1-specific T cells from the PBLs of HLA-A2 melanoma patients and healthy donors in 3 months. Because this antigen is expressed by most melanoma tumors, this procedure should be useful for checking the efficiency of adoptive

Received 10/20/99; revised 12/15/99; accepted 12/16/99. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported by contract grant sponsors (Institut National de la Sante´ et de la Recherche Me´dicale; Grant 6494 from the Association pour la Recherche contre le Cancer) and funds from the Ligue Nationale contre le Cancer “axe immunologie des tumeurs” and the Ligue De´partementale de Loire Atlantique. 2 The two first authors contributed equally to this work. 3 To whom requests for reprints should be addressed, at Institut National de la Sante´ et de la Recherche Me´dicale U463, 9 Quai Moncousu, 44093 Nantes Ce´dex 1, France. Phone: 33-2-40-08-47-20; Fax: 33-2-40-3566-97; E-mail: [email protected].

immunotherapy of melanoma tumors and using functionally well-defined Melan-A/MART-1-specific CTLs in a large group of patients.

INTRODUCTION The identification of melanoma-associated antigens recognized by CTL (for reviews, see Refs. 1 and 2) has opened the way to performing active and adoptive therapies of these tumors. Adoptive therapies consist of infusing high numbers of ex vivo-expanded antigen-specific T cells. In the mouse, the transfer of antigen-specific CTLs was shown to induce immunity against tumors and viruses (3, 4). More recent studies in this animal further established that high avidity/affinity and high cytokine-producing CTLs were the effectors of such therapies (5–7). In humans, the efficiency of adoptive therapy has been demonstrated thus far for viral diseases (8, 9) by the transfer of monoclonal or polyclonal virus-specific CTLs. Nonetheless, the follow-up of treated patients revealed a CD4 lymphocyte dependency of transferred CTL for long-term survival (8). The reason of this dependency, although not addressed, could be a limited capacity of CTLs to produce cytokines and especially IL4-2, hence their need for a T helper 1 cell help. Immunotherapeutic trials using ex vivo-expanded TILs have reported relative success (10). However, we have shown recently that a majority of melanoma TILs are either not tumorspecific or not avid enough to secrete cytokines in response to melanoma cells (data not shown). The development of strategies allowing to generate high amounts of high avidity tumor-reactive CTLs is therefore now a critical issue for the development of melanoma-specific adoptive therapies. The discovery of antigenic tumor epitopes has offered a strategy to obtain antigen-specific CTLs. This relies on performing repeated in vitro stimulations of PBLs by peptide-pulsed cells to obtain specific lymphocytes. Although most melanoma antigenic peptides stimulate the growth of peptide-specific PBLs in vitro (11–15), a large fraction of these, as well as of in vivo-detectable tumor antigen-specific T cells, were of low avidity and exhibited limited responses to tumor cells endogenously expressing the antigen (16). It was therefore necessary to derive methods allowing the expansion of high avidity T cells. Recently, peptide-MHC tetramers have been used to sort specific T cells, and by this approach, two groups already reported the isolation of high avidity tumor-reactive clones from

4

The abbreviations used are: IL, interleukin; TIL, tumor-infiltrating lymphocyte; PBL, peripheral blood lymphocyte; FACS, fluorescenceactivated cell sorting; mAb, monoclonal antibody; PHA, phytohemagglutinin; GM-CSF, granulocyte macrophage colony-stimulating factor.

1460 Ex Vivo Induction of High Avidity Tumor-reactive CTL

either TILs or PBLs (17, 18). However, FACS sorting is not yet applicable at a clinical grade to perform immunotherapies. We describe here an alternative procedure to isolate MelanA/MART-1 (thereafter Melan-A)-specific CTL clones of high reactivity toward melanoma cells endogenously expressing this antigen. To design this procedure, we have analyzed how conditions of ex vivo peptide presentation, i.e., the peptide presenting cell type, the nature of the peptide, its concentration, and the different cytokines added during PBL stimulation, could affect the growth of low and high avidity-specific CTLs. It has been shown before that cytokine secretion by T cells requires higher avidity interactions with target cells than with lysis (19, 20). Therefore, specific CTLs that developed only lysis in response to peptide were considered to be of low avidity, and those also producing cytokines were considered to be of high avidity. We show that, with selected conditions of peptide presentation, high avidity tumor-reactive CTL clones could be generated from the blood of all HLA-A*0201 donors, including melanoma patients. This procedure opens the possibility of Melan-A-targeted CTL-based therapy for a large group of melanoma patients.

MATERIALS AND METHODS Cell Lines. Melanoma cell lines HA, n44, n74, n77, n113, and n119 were established in our laboratory from metastatic tumor fragments as previously described (21), melanoma cells IPC 277/5 and IGR 1/54 were a gift from C. Aubert (INSERM U119, Marseilles, France), and melanoma lines Mel260 and Mel290 were a gift from J. C. Cerottini (Ludwig Institute for cancer research, Lausanne, Switzerland). Fn3 clones 25 and 29 were a gift from J. Zeuthen (Danish Cancer Society Research Center, Copenhagen, Denmark). The human mutant cell line CEMx721 T2 (T2) used as the presenting cell and the mouse fibrosarcoma WEHI 164 clone 13 used for the tumor necrosis factor assay were generous gifts from T. Boon (Ludwig Institute for Cancer Research, Brussels, Belgium). The murine cell line CTLL-2, used for the IL-2 assay, was obtained from the American Type Culture Collection. All cell lines were cultured in RPMI 1640 supplemented with 10% FCS, antibiotics, and for CTLL-2 cells, with ␤-mercaptoethanol (5.10⫺5 M; Sigma) and recombinant IL-2 (150 units/ml; Cetus). Synthetic Peptides. The Melan-A peptides27–35 AAGIGILTV and 26 –35EAAGIGILTV and the decamer analogue ELAGIGILTV were purchased from Genosys (Cambridge, United Kingdom). Purity (⬎70%) was controlled by reversed-phase high-performance liquid chromatography. Peptides were lyophilized, dissolved in DMSO at 10 mg/ml, and stored at ⫺80°C. mAbs. Phycoerythrin-conjugated MQ1–17H12 (antiIL-2) and FITC-conjugated 4S.B3 (anti-IFN-␥) purified antibodies were obtained from PharMingen SA (San Diego, CA). A103 (anti-Melan-A) antibody was obtained from DAKO (Trappes, France). Immunolabeling of Melanoma Cells with Monoclonal Anti-Melan-A Antibody. Cells (105 cells/well) were incubated with A103 mAb at appropriate an concentration in PBS containing 0.1% BSA and 0.1% Saponin for 30 min at 4°C. After two washes with the same buffer, cells were incubated for 30 min with the secondary FITC-labeled antibody. After wash-

ing, fluorescence analysis was immediately performed on a FACScan flow cytometer using CellQuest software (Becton Dickinson, Grenoble, France). Five thousand cells were gated with forward scatter/side scatter parameters and analyzed. PBL Stimulation. Stimulated cells were total PBLs from HLA-A2 donors, either healthy or melanoma patients. Peptidepulsed irradiated HLA-A2 melanoma lines or PBLs were used as stimulator cells. Loading with Melan-A peptides was done by incubating stimulator cells with the peptide (50 ␮M for the natural peptides or 1 ␮M for the peptide analogue) at 37°C in serum-free medium for 2 h. Stimulator cells were washed twice to eliminate unbound peptide. Stimulations were performed either in 24-well culture plates by mixing 5.105 peptide-pulsed stimulator cells and 1.106 responder PBLs or in 96-well culture plates by mixing peptide-pulsed stimulator cells, either 2.104 melanoma cells or 105 PBLs, with 105 responder PBLs in RPMI medium containing 8% human serum and cytokines. Irradiated stimulator cells were added again twice at 7-day intervals. Three cytokine combinations were used: either 5 ng/ml IL-6 (Sigma, St. Louis, MO) and 5 ng/ml IL-12 (Sigma) for the first week and 10 units/ml IL-2 and 5 ng/ml IL-7 thereafter, or IL-6 for the first week and IL-2 (50 U/ml) thereafter or no cytokine for the first week and IL-2 (50 units/ml) thereafter. 51 Chromium Microcytotoxicity Assay. Cytotoxic activity was measured in a standard 4-h assay against 51Cr-labeled cells. Briefly, target cells (peptide-pulsed T2 or melanoma cells) were incubated with 100 ␮Ci Na251CrO4 (Oris Industrie, Gifsur-Yvette, France) at 37°C for 1 h. Various amounts of effector cells were then added to 103 target cells in a final volume of 100 ␮l. For peptide recognition assays, T2 cells were preincubated with Melan-A27–35 peptide (10 ␮M) for 30 min at room temperature and then washed before the addition of effector cells. After 4 h, 25 ␮l of supernatant were mixed with 100 ␮l of scintillation cocktail (Optiphase Supermix, Wallak, United Kingdom) for measurement of radioactive content. A culture was considered positive when the percentage of lysis of peptide-pulsed T2 cells exceeded the percentage of specific lysis of T2 cells by at least 15%. Flow Cytometric Analysis of Intracellular Cytokines. Antigen-specific lymphocytes were quantified among stimulated PBLs by intracellular cytokine labeling using a method described by Jung et al. (22). PBLs were stimulated by peptidepulsed or unpulsed T2 cells (or by antigen expressing melanoma cells) at the ratio of 1:2 in RPMI 10% FCS in the presence of Brefeldin A (10 ␮g/ml, Sigma, St. Louis, MO). Peptide-pulsed T2 cells were prepared by a 1-h incubation at 37°C with 50 ␮M of Melan-A27–35 peptide in RPMI 1640 medium without FCS, followed by two washings. After 6 h, cells were fixed for 10 min at room temperature in a solution of PBS 4% paraformaldehyde. T cells (5.104) were stained with anticytokine mAbs at a concentration of 5 ␮g/ml for 30 min at room temperature. Reagent dilutions and washes were done with PBS containing 0.1% BSA (A-9647, Sigma) and 0.1% saponin (S-2149, Sigma). After staining, cells were resuspended in PBS, and 5000 events were analyzed on a FACScan (Becton Dickinson). Cloning and Selection of Melan-A-specific CTLs. PBLs from microcultures containing a significant fraction of Melan-A peptide-reactive cells were cloned in U-bottomed 96well plates (Falcon) by limiting dilution. Growth was induced


using PHA (1 ␮g/ml) and irradiated allogeneic feeder cells: EBV-transformed LAZ cells (2.104/well) and allogeneic PBLs (105/well). Microcultures showing ⬎95% probability of monoclonality, according to the single-hit Poisson law, were transferred into new plates with freshly irradiated feeder and PHA. Melan-A-specific CTL clones were selected on the basis of their capacity to lyse Melan-A peptide-pulsed T2 cells but not T2 cells alone. Expansion of T-Cell Clones. Melan-A-specific clones were expanded using a polyclonal T-cell stimulation protocol described previously (23). Briefly, 2000 clone cells/well were distributed in 96-well plates with 200 ␮l of culture medium (RPMI with 8% human serum and IL-2 150 units/ml) and irradiated feeder cells: LAZ EBV-B cells (2.104/well) and allogeneic PBLs (105/well). PHA was added (1 ␮g/ml). Culture medium was changed 2 days later. Clone T cells were split when their number exceeded 2.105/well. Sorting of Melan-A-specific Cells Using HLA-A2/ELA Tetramers. HLA-A0201/ELAGIGILTV tetramer was a generous gift from F. Romanier (Immunotech, Marseilles, France). T cells (1.106) were incubated for 1 h at room temperature in 200 ␮l of PBS/0.1% BSA containing diluted HLA/peptide tetramers (25 ␮g/ml). After two washes with 200 ␮l of PBS/0.1% BSA, labeled T cells were sorted using a FACS (Vantage, Becton Dickinson, Mountain View, CA).

RESULTS Peptide-pulsed Melanoma Cells Are More Efficient than Peptide-pulsed Autologous PBL in Stimulating Highly Reactive Melan-A-specific T Cells from PBL. PBLs from three HLA-A2 healthy donors were stimulated by allogeneic melanoma cells or autologous PBLs pulsed with Melan-A peptides 27–35. After three rounds of stimulation, PBLs were tested for their lytic activity against T2 cells pulsed with the same peptides and, as a negative control, against unpulsed T2 cells. Similar percentages of the microcultures stimulated by peptidepulsed melanoma cells or by peptide-pulsed PBLs developed a specific lysis, and levels of lysis were similar (data not shown). Microcultures exhibiting specific lysis were then analyzed individually to precise the fraction of high avidity Melan-A-specific CTLs able to secrete IFN-␥. As shown in Fig. 1a, microcultures stimulated by PBLs did not contain a significant fraction of IFN-␥-labeled cells. In contrast, significant fractions of IFN␥-labeled cells (ranging from 1.1 to 37%) were found in microcultures stimulated by melanoma cells (Fig. 1b). Loading Melanoma Cells with High Peptide Concentration Is Required to Efficiently Induce Highly Reactive Melan-A-specific T Cells. To establish the relative efficiency of different allogeneic melanoma cells and to determine the influence of peptide density, we compared the ability of five HLA-A2 melanoma lines pulsed or not pulsed with peptide to induce highly reactive Melan-A-specific CTLs from HLA-A2 PBLs. As shown in Table 1, three cell lines (M17, M119, and M113) efficiently induced lysis and IFN-␥ secretion in all of the microcultures containing 106 stimulated PBLs. Nonetheless, this high stimulation capacity was dependent upon peptide addition, although two of these cell lines (M113 and M119) expressed the Melan-A antigen endogenously, as shown by Melan-A protein

Fig. 1 Efficiency of allogeneic melanoma cells to stimulate Melan-A/ MART-1-specific CTLs from PBLs. IFN-␥ production was measured in polyclonal cultures obtained from three healthy donor PBLs stimulated by autologous PBLs (a) or allogeneic melanoma cells (b) pulsed with 50 ␮M of the nonamer peptide. The stimulation was performed in 96 wells containing 105 stimulated PBLs/well. Intracellular IFN-␥ labeling was measured in response to T2 cells preincubated (f) or not preincubated (䡺) for 1 h at 37°C with 50 ␮M Melan-A (27–35) peptide in microcultures containing specific lysis ⬎15%. T cells were stimulated for 6 h in the presence of Brefeldin A before being fixed, permeabilized, stained for cytokine, and analyzed on a FACScan. Each histogram represents the percentage of positive cells in each microculture tested.

labeling (data not shown) and by reverse transcription-PCR (24). Two melanoma lines, M44 and IGR1/54, were inefficient in stimulating the growth of Melan-A-specific CTLs. We then investigated the efficiency of different peptide concentrations to stimulate the growth of highly reactive MelanA-specific T cells. We used M17 allogeneic melanoma cells pulsed with 0.5, 5, and 50 ␮M of the Melan-A 27–35 peptide. As shown in Fig. 2, the fraction of IFN-␥-positive cells increased proportionally with the peptide concentration and passed beyond 10% in some wells only at the highest peptide concentration (50 ␮M). Melan-A-modified Peptide, ELAGIGILTV, Is More Efficient than the Natural Nonapeptide in Inducing the Growth of High Avidity Melan-A-specific CTLs. A Melan-A peptide analogue (ELAGIGILTV) of improved antigenicity has been described recently (25). Because we have observed that responses to this peptide of all Melan-A-specific clones or tetramer-sorted polyclonal TILs reached a plateau at a peptide concentration of 1 ␮M, we compared the efficiency of stimulations by M17 melanoma cells pulsed either with this concentration of the ÇELAE` peptide or with 50 ␮M of the natural nonamer AAGIGILTV. Fig. 3 shows that 1 ␮M of the peptide analogue induced higher fractions of high avidity CTLs than 50 ␮M of the Melan-A peptide 27–35. These fractions induced by the peptide analogue and by peptide 27–35 ranged respectively from 3.8 to 39% and from 1.3 to 9.2% for donor H4 and from 1.1 to 5.4% and from 0.4 to 2.7% for donor H5.


Table 1 Efficiency of different HLA-A*0201 melanoma cell lines to stimulate the growth of Melan-A-specific CTLs from allogeneic HLA-A*0201 PBLs Stimulating cells MelanFraction of Fraction of specific (Melan-A/MART1 A/MART-1 specific lytic IFN-␥-producing expression)a peptide 27–35b microculturesc microculturesd M17 (⫺) M44 (⫹) M113 (⫹) M119 (⫹) IGR1/54 (⫹⫹)

⫹ ⫺ ⫹ ⫺ ⫹ ⫺ ⫹ ⫺ ⫹ ⫺

7/7 0/7 1/5 0/5 5/7 1/7 7/7 0/7 0/7 0/6

5/5 0/5 0/3 0/3 3/7 1/7 5/5 0/4 0/4 0/6

a Expression of Melan-A/MART-1 antigen was estimated in melanoma lines by intracellular immunolabeling with a monoclonal specific antibody. b Stimulator melanoma cells were pulsed or not pulsed with MelanA/MART-1 peptide 27–35. c Seven days after the third stimulation, microcultures were tested for lytic activity in a chromium release assay against unpulsed T2 cells or T2 cells loaded with Melan-A/MART-1 27–35 peptide. Cytotoxicity was detected at an E:T ratio of 20:1. Specific lysis was considered significant when the percentage of lysis on peptide-pulsed T2 cells was ⬎15% than the percentage of lysis on unpulsed T2 cells. d Fraction of wells containing T cells secreting IFN-␥ upon stimulation by Melan-A/MART-1 peptide 27–35-pulsed T2 cells (50 ␮M). IFN-␥ secretion was detected by intracellular cytokine labeling.

Therefore, the peptide analogue induced a higher expansion of high avidity Melan-A-specific CTLs per microculture. However, it did not improve the percentage of positive microcultures. Efficiency of Melanoma Cells in Inducing Highly Reactive Melan-A-specific Lymphocytes Is Not Related to an Allorestricted Effect. To establish whether highly reactive Melan-A-specific T cells could also be obtained from melanoma patient PBLs and to ascertain that melanoma cells efficiency did not depend on an allorestricted effect, we stimulated PBLs of melanoma patients with the ELAGIGILTV peptide pulsed on autologous PBLs and autologous melanoma cells. Fractions of IFN-␥-producing lymphocytes were measured after three stimulations. Results confirmed that autologous PBLs were poorly efficient (Fig. 4a), as shown before with PBLs from healthy donors. These results also show that autologous melanoma cells were as efficient as allogeneic melanoma lines (Fig. 4b). Therefore, the efficiency of melanoma cells in stimulating highly reactive Melan-A-specific lymphocytes is not due to an allorestricted effect. IL-2 Alone Is Not Effective to Grow Highly Reactive Melan-A-specific T Cells. To be used in clinical trials, Melan-A-specific lymphocytes must be produced in GMP conditions. One difficulty of our stimulation protocol was the use of four recombinant cytokines, two of which, IL-7 and IL-12, were unavailable at clinical grade. To check the possibility to induce high avidity CTLs with a reduced number of cytokines, we compared stimulations with IL-2 alone from the third day and stimulations with IL-6 during the first week and IL-2 thereafter, with stimulations performed as previously with four cytokines.

Fig. 2 Influence of peptide densities on the induction of Melan-A/ MART-1-specific T cells. Three concentrations of peptide 27–35 were used to stimulate PBLs from one healthy donor: 0.5, 5, and 50 ␮M. This experiment was performed in five wells of a 24-well plate (with 106 stimulated PBLs/well). IFN-␥ production was measured in each well by intracellular labeling in response to T2 cells preincubated (f) or not preincubated (䡺) for 1 h at 37°C with 50 ␮M Melan-A (27–35) peptide.

PBLs from two melanoma patients (P3 and P4) were stimulated with the ELAGIGILTV peptide loaded on allogeneic M17 melanoma cells in the presence of these different cytokine combinations. As shown in Fig. 5, results obtained using Il-6 and IL-2 were similar to those obtained with the four cytokines. Therefore, IL-7 and IL-12 did not appear to be essential for the generation of highly reactive Melan-A-specific T cells from in vitro-stimulated PBLs. In contrast, IL-2 alone was clearly less efficient to induce a high growth of IFN-␥-producing T cells. High Avidity Tumor-reactive CTL Clones Are Generated from PBL Stimulated by Peptide-pulsed Melanoma Cells. Because alloreactive T cells could be generated in PBL microcultures stimulated by allogeneic melanoma cells, the capacity of these polyclonal PBLs to recognize Melan-A peptides on tumor cells was difficult to establish. This question was addressed using clones and Melan-A peptide-specific T cells sorted from the cultures using HLA-A2/ELA tetramers. Six microcultures from four different donors that contained a high percentage of Melan-A-specific T cells were cloned by limiting dilution. Melan-A peptide-specific clones (ranging from 8 to 95% of tested clones) were obtained from each cloning. On the basis of specific V␤ expression and/or on T-cell receptor sequencing, we found that one to five different Melan-A-specific clones were derived from each microculture. These clones were


Fig. 3 Comparison of the induction of Melan-A/MART-1-specific T cells by the natural peptide (AAGIGILTV) and the modified peptide (ELAGIGILTV) from two healthy donors (H4 and H5). This experiment was performed in 48 wells of a 96-well plate (105 stimulated PBLs/ well). IFN-␥ production was measured in eight wells containing specific lysis ⬎15% by intracellular labeling in response to T2 cells preincubated (f) or not preincubated (䡺) for 1 h at 37°C with 50 ␮M of Melan-A (27–35) peptide.

tested for their capacity to kill a panel of melanoma lines expressing or not expressing Melan-A and HLA-A2 molecules. All of the clones reacted with the peptide 27–35 pulsed onto T2 cells and with the HLA-A2⫹/Melan-A⫹ melanoma cells but not with HLA-A2⫺/Melan-A⫹ cell lines or with HLA-A2⫹/ Melan-A⫺ melanoma cell lines. Results of a representative experiment done with four of these clones are shown on Fig. 6. To estimate the tumor reactivity of these clones, we compared the level of their lytic and cytokine responses to melanoma lines expressing Melan-A with that of our most reactive Melan-A-specific TIL clones. The data on Fig. 7 show that these PBL-derived clones lysed melanoma lines at similar levels to those of melanoma lines lysed by TIL-derived clones. However, all of them secreted significantly higher amounts of GM-CSF and IFN-␥ than did TIL clones, and two of the four clones shown on Fig. 7 also secreted higher amounts of IL-2 than did TIL clones. With the perspective to use such highly reactive tumorspecific CTL for adoptive immunotherapy, the growth potential of these clones was analyzed using a culture protocol described and used for therapeutic TIL expansion (23, 26). With this approach, we could repeatedly obtain an amplification of each clone until we obtain 109 T cells within 1 month.

Fig. 4 The efficiency of melanoma cells to induce Melan-A-specific CTLs does not depend on an allorestricted effect. PBLs from two melanoma patients were stimulated by autologous PBLs (a) or autologous melanoma cells (b) pulsed with 1 ␮M of the modified peptide (ELAGIGILTV). This experiment was performed in 48 wells of a 96-well plate (105 stimulated PBLs/well). Intracellular IFN-␥ production was measured in response to T2 cells preincubated (f) or not preincubated (䡺) for 1 h at 37°C with 50 ␮M Melan-A (27–35) peptide in microcultures containing specific lysis ⬎15%. T cells were stimulated for 6 h in the presence of Brefeldin A before being fixed, permeabilized, stained for cytokine, and analyzed on a FACScan. Each histogram represents the percentage of positive cells in each microculture tested.

Highly Tumor-reactive T-Cell Populations Can Be Sorted by Specific Tetramer from Stimulated PBL. MelanA-reactive CTL stimulated either by the natural peptide 27–35 or by the ELA peptide were sorted by FACS using HLA-A2/ ELAGIGILTV peptide tetramers and were tested for their capacity to produce cytokines upon stimulation by HLA-A2⫹/ Melan-A⫹ cell lines. Fractions of tetramer-labeled cells found among the PBLs stimulated by peptide 27–35 and by the ELA analogue were respectively of 2 and 39% before sorting and of 94 and 99% after (data not shown). We measured the production of IFN-␥ and IL-2 of these two sorted populations in response to melanoma cells. These responses were identical in terms of percentage of labeled cells and mean of fluorescence (Fig. 8). This shows an equal capacity of individual lymphocytes stimulated by one of these two peptides to secrete IFN-␥ and IL-2.

DISCUSSION Although in vitro peptide presentation has been shown to be capable of stimulating peptide-specific T cells from PBLs, the relative efficiency of different presenting cells and also the tumor reactivity/avidity of specific T cells obtained have rarely been evaluated. To achieve adoptive therapy, we tried to design an efficient and simple procedure to derive high amounts of high avidity tumor-reactive T cells. To this end, we have taken advantage of the intracellular cytokine labeling technique to


Fig. 6 Demonstration of the capacity of four Melan-A-specific PBL clones to recognize melanoma cells naturally expressing this antigen. Lytic activity of these clones was measured simultaneously against peptide-pulsed T2 cells and a panel of melanoma lines expressing the antigen: M44, M74, M77, M113, and Mel290. Clones were also tested against the K562 class I-negative cell line, the HLA-A2-negative Melan-A-positive cell line Mel260, and an HLA-A2-positive Melan-Anegative subclone of melanoma line IPC277/5.

Fig. 5 Influence of cytokine addition on the induction of Melan-A/ MART-1-specific T cells. PBLs from two melanoma patients (P3 and P4) were stimulated by the allogeneic M17 melanoma cell line pulsed with 1 ␮M of the modified peptide (ELAGIGILTV). Stimulations were performed in 32 wells of a 96-well plate (105 stimulated PBLs/well) using three different cytokine combinations. I, first stimulation with 5 ng/ml of IL-6 and IL-12 and the next stimulations with 5 ng/ml of IL-7 and 10 units/ml of IL-2. II, first stimulation with 5 ng/ml of IL-6 and the next stimulations with 50 units/ml of IL-2. III, first stimulation without a cytokine addition, and at day 3 and for the next stimulations, the addition of 50 units/ml of IL-2. Intracellular IFN-␥ labeling was measured in microcultures containing ⬎15% of the specific lysis in response to T2 cells preincubated (f) or not preincubated (䡺) for 1 h at 37°C with 50 ␮M Melan-A (27–35) peptide.

quantify high avidity peptide-specific and tumor-reactive lymphocytes growing in individual cultures. Here we show that most melanoma cells, once loaded with Melan-A peptides, systematically induced the growth of high avidity Melan-A-specific and tumor reactive CTLs from only 106 PBLs. After three peptide stimulations, the percentage of high avidity-specific CTLs detected in microcultures ranged from 0.5 to 37%. Such an efficient generation of MelanA-specific CTLs can be related to the high frequency of Melan-A-specific effectors in the blood of many donors, as recently shown using HLA-A2-peptide tetramers (27) As reported before (14, 28), stimulation by peptide-pulsed PBLs also induced Melan-A-specific CTLs, as shown by the specific lysis that developed in most of these microcultures. Nonetheless, we show that CTLs generated by this way were of too low avidity to secrete cytokines in response to natural Melan-A peptide. We also showed that a density of MHC-peptide complexes, higher than the naturally occurring density, was required for efficient CTL generation by melanoma cells. Indeed, in the absence of a peptide addition, Melan-A-expressing melanoma

Fig. 7 Comparison of the lytic activity and cytokine production of five CTL clones derived from stimulation of healthy donor PBL clones (F) with that of two TIL-derived clones (E) against two Melan-A-expressing cell lines, M77 and M113. For the cytotoxicity assays, the target cells were 51Cr-labeled and incubated for 4 h with CTLs at an effector/ target ratio of 3:1. For cytokine production, clones were stimulated by Melan-A-expressing cell lines M77 and M113 for 6 h. The supernatants were collected, and cytokines were measured by ELISA assay.

lines were inefficient. In addition, although the lowest concentration of the natural peptide (0.5 ␮M) induced high avidity CTLs in some microcultures, 50 ␮M induced a much higher fraction of such CTLs in all of the microcultures. A minimal


Fig. 8 Comparison of cytokine production of polyclonal populations stimulated either with the natural peptide 27–35 (a) or the peptide analogue (b) and sorted with HLA-A2/ELA tetramers (see “Materials and Methods”). Intracellular IFN-␥ and IL-2 labeling was measured for each sorted population in response to melanoma cells expressing (FM3-clone 29 and M113) or not expressing (FM3-clone 25) the Melan-A/MART-1 antigen.

density of MHC peptides on melanoma cells is therefore required for efficient stimulation and expansion of high avidity Melan-A-specific PBLs in vitro, and there was a clear-cut dose response relationship from 0.5 to 50 ␮M for the generation of high avidity CTLs. This may explain why a melanoma line, M44, with relatively low HLA-A2 expression (24) was inefficient in inducing high avidity CTLs (Table 1). It has been previously reported that high concentrations of viral peptides preferentially induced low avidity T cells by inducing apoptosis of high avidity T cells (29). This apparent discrepancy with our results may be explained by the low affinity of natural Melan-A peptide for the HLA-A*0201 molecule. This likely limits the density of HLA-A2/peptide complexes expressed by melanoma cells even in presence of a high concentration of this peptide (50 ␮M). Therefore, high avidity CTLs generated in the present study have been probably induced, as in the study of AlexanderMiller et al. (29), by melanoma cells bearing relatively low MHC peptide densities. In support of this interpretation, we recently obtained results similar to those of Alexander-Miller et al. (29) using the Melan-A peptide analogue ELA of high avidity for HLA-A2, i.e., more efficient induction of high avidity Melan-A-specific CTLs with 1 ␮M than with 10 or 50 ␮M (data not shown). This suggests that high ELA peptide concentrations, like viral peptides used by Alexander-Miller et al. (29), induced apoptosis of high avidity CTLs. In addition to the density of the MHC peptide expression, the density of the ICAM-1 molecule on melanoma cells also seemed critical, as suggested by the complete inefficiency of a melanoma line (IGR1/54) lacking ICAM-1 to stimulate specific CTLs. This is in accordance with our previous results showing that melanoma cells lacking ICAM-1 induced a suboptimal activation of Melan-A- and NA17-A-specific CTL clones (24). Although not shown in

this study, the same was true for the LFA-3/CD2 interaction as far as a melanoma line that expressed LFA-3 at a low density was found inefficient in activating CTL clones. Furthermore, we had shown before that increasing LFA-3 density on melanoma cells by transfection always resulted in higher cytokine responses of melanoma-associated antigenspecific CTL clones (30). A possible explanation for the inefficiency of PBLs in generating high avidity-specific CTLs may be the low expression of ICAM-1 and LFA-3 by most of these cells compared with melanoma cells (data not shown). In contrast, the classical costimulatory interaction between B7 molecules and CD28, often considered critical for T-cell activation, was not required in the present stimulations as far as melanoma cells lack B7 molecules. Nonetheless, T-cell stimulations independent of the CD28-CD80 interaction have already been reported especially for memory CD8 and even for naive T cells in stimulations with high MHC-peptide densities (31–33). The role of cytokines in inducing the growth of specific T cells during peptide stimulation is thus far unclear. Various cytokine combinations have been used previously with success, but neither their requirement nor their relative efficiency was established. Van der Bruggen et al. (12) used IL-6 and IL-12 for the first week of stimulation, then IL-7 and IL-2. Others used IL-7 and IL-2 (11, 34). IL-1␤, IL-4, and IL-10 were also used for human CTL induction (15, 35, 36). In side-by-side experiments, we found no significant difference between the efficiency of the four cytokine combination (IL-6/IL-12 then IL-7/ IL-2) and that of the two cytokine combination (IL-6 then IL-2) in generating high avidity CTLs. In contrast, the use of IL-2 alone dramatically reduced the frequency of these cells. Furthermore, cytokines secreted by melanoma cells could contributed to the differentiation and activation of Melan-A-specific


PBLs. Indeed, we have observed differences between melanoma lines in cytokine transcription that could also account for their different efficiency to stimulate CTLs. Among cytokine RNAs, those differentially expressed by melanoma lines were IL-1, tumor necrosis factor ␣ and ␤, IL-6, IL-7, IL-10, and 1L-15.5 The efficiency of the M17 melanoma line in stimulating specific CTLs in the absence of exogenous Il-7 could be due to the endogenous expression of this cytokine by M17 cells (detected by reverse transcription-PCR). To improve the generation of high avidity Melan-Aspecific CTLs, we also tested the efficiency of a recently described Melan-A peptide analogue ELAGIGILTV, which was designed by Cerottini’s group as one of the most interesting variants of Melan-A 27–35 and 26 –35 natural epitopes (25). We show that, when loaded on M17 melanoma cells, this peptide induced a higher expansion of high avidity CTLs than the natural nonamer, although it did not increase the frequency of positive cultures. Furthermore, this peptide analogue was as efficient as the natural peptide at a concentration 50 times lower. With the perspective to derive CTL for use in immunotherapy, it was particularly critical to control if Melan-A peptide reactive T cells derived with allogeneic melanoma cells and with this peptide analogue were not alloreactive and of sufficient avidity to recognize tumor cells endogenously expressing the natural epitopes. To this end, Melan-A peptide reactive lymphocytes were sorted from the cultures, either by cloning or by using HLA-A2/ELA tetramers. We observed that these lymphocytes exclusively recognized melanoma cells expressing both HLA-A2 and Melan-A and were of high avidity toward these cells as shown by the secretion of not only IFN-␥ but also IL-2. Using V␤ labeling, we found that several clones (between one and five) were derived from 105 or 106 stimulated PBLs. All these clones had a similar lysis but significantly higher capacity to secrete IFN-␥ and GM-CSF than TIL-derived clones, and for a significant fraction, a higher capacity to secrete IL-2 in response to melanoma cells. Regarding the recent demonstration that IFN-␥ and GM-CSF secretion by tumor-specific CTLs plays a critical role in inducing tumor-specific responses (7), these PBL-derived clones represent interesting effectors to be used for antigen adoptive immunotherapy. Although not shown, Il-2 secretion could be a critical function for adoptive therapy, especially for the survival of transferred CTLs in vivo. Finally, we could systematically expand all melanoma-specific clones by a factor of 1000 within 1 month. In conclusion, stimulation conditions designed here were found efficient to derive high amounts of high avidity and highly tumor-reactive CTL clones from all melanoma donors. Another procedure to obtain high avidity tumor-reactive CTL clones was reported recently by Yee et al. (18) using autologous peptide pulsed dendritic cells to stimulate PBLs and HLA-A2Melan-A peptide tetramers to sort high avidity CTLs that were then cloned. Another group also succeeded to derive such clones from patient TIL or stimulated PBL using FACS sorting with peptide MHC tetramers and then cloning (17). However, clinical

5 P. Lemarre, S. Raher, N. Labarriere, B. Dreno, F. Jotereau, and A. Godard, unpublished data.

grade MHC class I peptide tetramers are not yet available. In the procedure described here, IFN-␥ labeling was an alternative way to select populations sufficiently enriched in high avidity T cells to allow successful cloning. Another interest of the present procedure is the use of a single established melanoma line to stimulate PBLs of all donors, which obviates the need to grow high amounts of autologous dendritic cells as done in other recently described protocols (17, 18). Therefore, it is feasible to grow high amounts of high avidity CTL clones specific for Melan-A antigen from the PBLs of all melanoma patients. Reinfusion of such cells to melanoma patients should allow adequate determination of the efficiency of adoptive immunotherapy against this melanoma antigen.

REFERENCES 1. Van den Eynde, B. J., and Van der Bruggen, P. T. Cell-defined tumor antigens. Curr. Opin. Immunol., 9: 684 – 693, 1997. 2. Rosenberg, S. A. Cancer vaccines based on the identification of genes encoding cancer regression antigens. Imunol. Today, 18: 175– 182, 1997. 3. Rouse, B. T., Wagner, H., and Harris, A. W. In vivo activity of in vitro immunized lymphocytes. I. Tumor allograft rejection mediated by in vitro activated mouse thymocytes. J. Immunol., 108: 1353–1361, 1972. 4. Melief, C. J. M., and Kast, W. M. T-cell immunotherapy of tumors by adoptive transfer of cytotoxic T lymphocytes and by vaccination with minimal essential epitopes. Immunol. Rev., 145: 167–177, 1995. 5. Alexander-Miller, M. A., Leggatt, G. R., and Berzofsky, J. A. Selective expansion of high- or low-avidity cytotoxic T lymphocytes and efficacy for adoptive immunotherapy. Proc. Natl. Acad. Sci. USA, 93: 4102– 4107, 1996. 6. Speiser, D. E., Miranda, R., Zakarian, A., Bachmann, M. N., Mc Kall-Faienza, K., Odermatt, B., Hanahan, D., Zingernagel, R. M., and Ohashi, P. S. Self antigens expressed by solid tumors do not efficiently stimulate naive or activated T cells: implications for immunotherapy. J. Exp. Med., 186: 645– 653, 1997. 7. Nagoshi, M., Goedegebuure, P. S., Burger, U. L., Sadanaga, N., Chang, M. P., and Eberlein, T. J. Successful adoptive immunotherapy is dependent on induction of a host immune response triggered by cytokine (IFN-␥ and granulocyte/macrophage colony-stimulating factor) producing donor tumor-infiltrating lymphocytes. J. Immunol., 160: 334 –344, 1998. 8. Riddell, S. R., Watanabe, K. S., Goodrich, J. M., Li, C. R., Agha, M. E., and Greenberg, P. D. Restoration of viral immunity in immunodeficient humans by the adoptive transfer of T cell clones. Science (Washington DC), 257: 238 –241, 1992. 9. Heslop, H. E., Ng, C. Y., Li, C., Smith, C. A., Loftin, S. K., Krance, R. A., Brenner, M. K., and Rooney, C. M. Long-term restoration of immunity against Epstein-Barr virus infection by adoptive transfer of gene-modified virus-specific T lymphocytes. Nat. Med., 2: 551–555, 1996. 10. Boon, T., and Old, L. J. Cancer tumor antigens. Curr. Opin. Immunol., 9: 681– 683, 1997. 11. Celis, E., Tsai, V., Crimi, C., DeMars, R., Wentworth, P. A., Chesnut, R. W., Grey, H. M., Sette, A., and Serra, H. M. Induction of anti-tumor cytotoxic T lymphocytes in normal humans using primary cultures and synthetic peptide epitopes. Proc. Natl. Acad. Sci. USA, 91: 2105–2109, 1994. 12. Van der Bruggen, P., Bastin, J., Gajewski, T., Coulie, P. G., Boel, P., De Smet, C., Traversari, C., Townsend, A., and Boon, T. A peptide encoded by human gene MAGE-3 and presented by HLA-A2 induces cytolytic T lymphocytes that recognize tumor cells expressing MAGE-3. Eur. J. Immunol., 24: 3038 –3043, 1994. 13. Visseren, M. J. W., van Elsas, A., van der Voort, E. I. H., Ressing, M. E., Kast, W. M., Schrier, P. I., and Melief, C. J. M. CTL specific for


the tyrosinase autoantigen can be induced from healthy donor blood to lyse melanoma cells. J. Immunol., 154: 3991–3998, 1995. 14. Rivoltini, L., Kawakami, Y., Sakaguchi, K., Scouthwood, S., Sette, A., Robbins, P. F., Marincola, F. M., Salgaller, M. L., Yannelli, J. R., Appella, E., and Rosenberg, S. A. Induction of tumor-reactive CTL from peripheral blood and tumor-infiltrating lymphocytes of melanoma patients by in vitro stimulation with an immunodominant peptide of the human melanoma antigen MART-1. J. Immunol., 154: 2257–2265, 1995. 15. Tsai, V., Southwood, S., Sidney, J., Sakaguchi, K., Kawakami, Y., Appella, E., Sette, A., and Celis, E. Identification of subdominant CTL epitopes of the GP100 melanoma-associated tumor antigen by primary in vitro immunization with peptide-pulsed dendritic cells. J. Immunol., 158: 1796 –1802, 1997. 16. Gervois, N., Guilloux, Y., Diez, E., and Jotereau, F. Suboptimal activation of melanoma infiltrating lymphocytes (TIL) due to low avidity of TCR/MHC-tumor peptide interactions. J. Exp. Med., 183: 2403– 2407, 1996. 17. Dunbar, P. R., Chen, J. L., Chao, D., Rust, N., Teisserenc, H., Ogg, G. S., Romero, P., Weynants, P., and Cerundolo, V. Rapid cloning of tumor-specific CTL suitable for adoptive immunotherapy of melanoma. J. Immunol., 162: 6959 – 6962, 1999. 18. Yee, C., Savage, P. A., Lee, P. P., Davis, M. M., and Greenberg, P. D. Isolation of high avidity melanoma-reactive CTL from heterogeneous populations using peptide-MHC tetramers. J. Immunol., 162: 2227–2234, 1999. 19. Fonteneau, J. F., Le Drean, E., Le Guiner, S., Gervois, N., Diez, E., and Jotereau, F. Heterogeneity of biologic responses of melanomaspecific CTL. J. Immunol., 159: 2831–2839, 1997. 20. Valitutti, S., Muller, S., Dessing, M., and Lanzavecchia, A. Different responses are elicited in cytotoxic T lymphocytes by different levels of T cell receptor occupancy. J. Exp. Med., 183: 1917–1921, 1996. 21. Gervois, N., Heuze´, F., Diez, E, and Jotereau, F. Selective expansion of a specific anti-tumor CD8⫹ cytotoxic T lymphocyte clone in a bulk culture of tumor-infiltrating lymphocytes from a melanoma patient: cytotoxic activity and T cell receptor gene rearrangements. Eur. J. Immunol., 20: 825– 831, 1990. 22. Jung, T., Schauer, U., Heusser, C., Neumann, C., and Rieger, C. Detection of intracellular cytokines by flow cytometry. J. Immunol. Methods, 159: 197–207, 1993. 23. Jotereau, F., Pandolfino, M. C., Boudart, D., Diez, E., Dreno, B., Douillard, J. Y., Muller, J. Y., and Le Mevel, B. High fold expansion of human cytotoxic T-lymphocytes specific for autologous melanoma cells for use in immunotherapy. J. Immunother., 10: 405– 411, 1991. 24. Labarrie`re, N., Diez, E., Pandolfino, M. C., Viret, C., Guilloux, Y., Le Guiner, S., Fonteneau, J. F., Dreno, B., and Jotereau, F. Optimal T cell activation by melanoma cells depends on a minimal level of antigen transcription. J. Immunol., 158: 1238 –1245, 1997. 25. Valmori, D., Fonteneau, J. F., Lizana, C. M., Gervois, N., Lienard, D., Rimoldi; D., Jongeneel, V., Jotereau, F., Cerottini, J. C., and Ro-

mero, P. Enhanced generation of specific tumor-reactive CTL in vitro by selected Melan-A/MART-1 immunodominant peptide analogues. J. Immunol., 160: 1750 –1758, 1998. 26. Tessier, M. H., Pandolfino, M. C., Jotereau, F., Boudart, D., Litoux, P., and Dreno, B. Home therapy with autologous tumour-infiltrating lymphocytes and subcutaneous interleukin-2 in metastatic melanoma. Eur. J. Cancer, 4: 735–736, 1996. 27. Pittet, M. J., Valmori, D., Dunbar, P. R., Speiser, D. E., Lienard, D., Lejeune, F., Fleischhauer, K., Cerundolo, V., Cerottini, and Dreno, B. High frequencies of naive Melan-A/MART-1-specific CD8⫹ T cells in a large proportion of human histocompatibility leukocyte antigen (HLA)-A2 individuals. J. Exp. Med., 190: 705–715, 1999. 28. Ostankovitch, M., Le Gal, F. A., Connan, F., Chassin, D., Choppin, J., and Guillet, J. G. Generation of MelanA/MART-1-specific CD8⫹ cytotoxic T lymphocytes from human naive precursors: helper effect requirement for efficient primary cytotoxic T lymphocyte induction in vitro. Int. J. Cancer, 72: 987–994, 1997. 29. Alexander-Miller, M. A., Derby, M. A., Sarin, A., Henkart, P. A., and Berzofsky, J. A. Supraoptimal peptide-major histocompatibility complex causes a decrease in Bcl-2 levels and allows tumor necrosis factor ␣ receptor II-mediated apoptosis of cytotoxic T lymphocytes. J. Exp. Med., 188: 1391–1399, 1998. 30. Le Guiner, S., Le Drean; E., Labarriere, N., Fonteneau, J. F., Viret, C., Diez, E., and Jotereau, F. LFA-3 co-stimulates cytokine secretion by cytotoxic T lymphocytes by providing a TCR-independent activation signal. Eur. J. Immunol., 28: 1322–1331, 1998. 31. Harding, F. A., Mc Arthur, J. G., Gross, J. A., Raulet, D. H., and Allison, J. P. CD28-mediated signaling co-stimulates murine T cells and prevents induction of anergy in T-cell clones. Nature (Lond.), 356: 607– 609, 1992. 32. Green, J. M., Noel, P. J., Sperling, A. I., Walunas, T. L., Gray, G. S., Bluestone, J. A., and Thompson, C. B. Absence of B7-dependent responses in CD28-deficient mice. Immunity, 1: 501–508, 1994. 33. Cai, Z., and Sprent, J. Influence of antigen dose and costimulation on the primary response of CD8⫹ T cells in vitro. J. Exp. Med., 183: 2247–2257, 1996. 34. Bakker, A. B., Marland, G., de Boer, A. J., Huijbens, R. J., Danen, E. H., Adema, G. J., and Figdor, C. G. Generation of antimelanoma cytotoxic T lymphocytes from healthy donors after presentation of melanoma-associated antigen-derived epitopes by dendritic cells in vitro. Cancer Res., 55: 5330 –5334, 1995. 35. Mehta-Damani, A., Markowicz, S., and Engleman, E. G. Generation of antigen-specific CD8⫹ CTLs from naive precursors. J. Immunol., 153: 996 –1003, 1994. 36. Stevens, E. J., Jacknin, L., Robbins, P. F., Kawakami, Y., El Gamil, M., Rosenberg, S. A., and Yannelin, J. R. Generation of tumor-specific CTLs from melanoma patients by using peripheral blood stimulated with allogeneic melanoma tumor cell lines. J. Immunol., 154: 762–777, 1995.