Spontaneous Vitiligo in an Animal Model for Human Melanoma: Role of Tumor-Specific CD8 T Cells

[CANCER RESEARCH 64, 1496–1501, February 15, 2004] Spontaneous Vitiligo in an Animal Model for Human Melanoma: Role of Tumor-specific CD8؉ T Cells

Rene´e Lengagne,1 Fre´de´rique-Anne Le Gal,2 Maryle`ne Garcette,1 Laurence Fiette,3 Patrick Ave,3 Masashi Kato,4 Jean-Paul Briand,5 Christian Massot,6 Izumi Nakashima,4 Laurent Re´nia,1 Jean-Ge´rard Guillet,1 and Armelle Pre´vost-Blondel1 1De´partement d’Immunologie, Institut Cochin, INSERM U567, CNRS UMR 8104, Laboratoire membre de l’IFR 116, Universite´ R. Descartes, Paris, France; 2Laboratory of Tumor Immunology, Division of Oncology, Geneva, Switzerland; 3Unite´ de Recherche et d’Expertise en Histotechnologie et Pathologie, Institut Pasteur, Paris, France; 4Department of Immunology, Nagoya University School of Medicine, Nagoya, Japan; 5UPR 9021 CNRS, Strasbourg, France; and 6INSERM CRI/IFR69, Hopital P. Brousse, Villejuif, France

ABSTRACT ated with positive clinical response. Indeed, it has been observed specifically in patients with metastatic melanoma who respond favor- Tumor antigen-reactive T cells can be detected in a large proportion of ably to interleukin (IL)-2 (4). Vitiligo also occurs in melanoma melanoma patients, but their efficacy on tumor control in vivo remains ϩ patients after infusion of tumor-specific CD8 T cells (5, 6). The unclear. On the other hand, vitiligo, a skin disorder characterized by patchy depigmented macules, may occur spontaneously or after antitumor prevalence of antibodies and T cells that are MDA specific in both therapies. Moreover, vitiligo is significantly associated with positive clin- melanoma and vitiligo patients highlights the similarities between ical response, but the mechanism is not understood. Therefore, the estab- autoimmunity observed in vitiligo and antitumor immunity observed lishment of a relevant animal model in which melanoma and vitiligo in melanoma immune surveillance (1). However, the exact role of the spontaneously develop stepwise may be useful for better understanding of immune response specific for MDA in vitiligo development and the parameters involved in the destruction of both benign and malignant melanoma control remains unclear. melanocytes. In a previous work, we established a mouse model for Progressive skin depigmentation occurs spontaneously in several melanoma in which MT/ret transgenic mice express the ret oncogene fused animal models of nonmelanoma-associated vitiligo, including swine, to the metallothionein promoter. Here we report that melanoma leads to Smyth chickens, and the C57BL6/Ler-vit/vit mouse strain (7). It has spontaneous vitiligo. We further investigate, for the first time in this been suggested that the vitiligo pathogenesis is similar in humans and model, the natural antitumor T-cell response and evaluate the role of cellular immunity in the development of the disease. Interestingly, the animals. Indeed, antibodies against the pigmented cell surface anti- occurrence of spontaneous tumor nodules in MT/ret mice with melanoma- gens have been found in dogs, cats, and horses with vitiligo (8), and associated vitiligo is significantly delayed when compared in melanoma T cells infiltrating feather tissue are abundant before visible signs of mice without vitiligo. Moreover, a significant proportion of mice with vitiligo and in tissues undergoing active loss of pigment in Smyth melanoma-associated vitiligo resisted a challenge with syngeneic mela- chickens (9). Frequent combination of melanoma and vitiligo has been noma cells in contrast to animals without vitiligo. Our results confirm that observed in Arabian horses and Sinclair swine (10), heralding regres- vitiligo is associated with clinical benefit and further demonstrate the sion or slow progression of the melanoma, but no relevant mouse ؉ crucial role of CD8 T cells for tumor control in melanoma-associated model is currently available. Therefore, the establishment of a suitable vitiligo. animal model in which melanoma and vitiligo spontaneously develop stepwise would contribute to a better understanding of parameters INTRODUCTION involved in the destruction of both benign and malignant melanocytes. By introducing the human ret oncogene fused to the murine metallo- A large proportion of melanoma patients spontaneously develop a thionein (MT) promoter-enhancer, Iwamoto et al. (11) produced MT/ strong response to melanocyte differentiation antigens (MDAs), pro- ret transgenic mouse lines with a mixed strain background (C57BL/ viding direct evidence for induction of antitumor immunity during 6 ϫ BALB/c). In these mice, skin melanosis and benign melanocytic cancer progression. However, the efficiency of tumor-reactive T cells tumors developed slowly. More recently, a new line was established in controlling tumor progression in vivo remains unclear. The onset of by back-crossing one MT/ret line with C57BL/6 mice. In this line, the vitiligo, a skin disorder characterized by patchy depigmented macules, process of tumor development and malignant transformation bears has been thought to be the result of immune-mediated destruction of multiple resemblances with that of the human giant congenital mela- melanocytes, although other mechanisms may lead to their destruction nocytic nevus that gives rise to cutaneous melanoma during aging (12). (1). Recent studies suggest the involvement of cellular immunity in In the present report, we show that a large proportion of MT/ret vitiligo development. Ogg et al. (2) were the first to identify, in animals spontaneously develop melanoma-associated vitiligo. We fur- patients with non-melanoma-associated vitiligo, skin-homing autore- ther used this model to evaluate the role of cellular immunity in the active circulating T cells that specifically recognize melanoma cells in disease development and to address the following questions: Can we vitro. Moreover, we recently studied T cells from vitiliginous skin ϩ detect melanoma-specific T cells in peripheral blood lymphocytes? margins, and our data clearly support the role of CD8 T cells specific What are their antigen specificities? Does T-cell frequency correlate for MDAs in vitiligo occurring naturally in melanoma patients (3). with disease severity? Does the occurrence of spontaneous vitiligo Antimelanoma therapies sometimes lead to the development of viti- correlate with a higher frequency of melanoma-specific circulating T ligo, and this permanent loss of skin pigment is significantly associ- cells? Does vitiligo correlate with regression of the spontaneous melanoma or protect mice against challenge with syngeneic mela- Received 9/8/03; revised 11/18/03; accepted 12/5/03. ϩ Grant support: French Ligue Nationale contre le Cancer, “axe immunologie des noma cells? Are CD8 T cells crucial for melanoma control in mice tumeurs”; and a fellowship from the Association Franc¸aise de Recherche contre le Cancer with melanoma-associated vitiligo? (to A. Pre´vost-Blondel). 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 MATERIALS AND METHODS 18 U.S.C. Section 1734 solely to indicate this fact. Requests for reprints: Armelle Pre´vost-Blondel, Institut Cochin, De´partement d’Immunologie, 27 rue du Faubourg Saint Jacques, 75014 Paris, France. Phone: 33-1-40- Mice. MT/ret transgenic 304/B6 mice (12) were established by crossing b/d 51-65-24; Fax: 33-1-40-51-65-35; E-mail: [email protected]. line 304 MT/ret mice (originating from a BCF1 mouse, H-2 ) five times with 1496

C57BL/6J (H-2b) mice (Harlan, Gannat, France). The founder has been derived anti-mouse CD8 or purified rat IgG as described above. Mice were checked for from a cross between C57BL/6 and BALB/c ϫ (BALB/c ϫ C57BL/6). Thus, the presence of a palpable tumor, and tumor growth was measured twice a all of the mice used in the present report have been back-crossed six times with week. Animals were scored positive when the mean tumor diameter was Ͼ3 C57BL/6 mice and contain Ͻ1% of BALB/c genes. Mice were kept under mm, measured using a gauge caliper. standard conditions except that zinc (115 ␮g/l) was present in their drinking Statistical Analysis. Differences in the incidence of disease signs in ani- water. Clinical signs were assessed twice a month, and development of mals 6 months of age were evaluated using the ␹2 test. Associations between exophthalmus, facial or dorsal tumor nodules, and vitiligo was recorded. Mice disease sign and mortality were evaluated using Pearson correlation statistics. were used for experiments at different time points in the course of malignancy. The Mann-Whitney test was used to compare the frequency of IFN-␥-secreting Control mice were nontransgenic littermates. cells in response to Melan-ret cells in MT/ret mice with or without vitiligo. The Histology. Skin samples were taken from the periphery of vitiligo patches. ␹2 test and Fisher’s exact test were used to evaluate the significance between Skin tissues were fixed in a solution containing zinc acetate (0.5%), zinc responder and nonresponder mice groups. The difference in tumor develop- chloride (0.5%), and calcium acetate (0.05%) in Tris buffer at pH 7 for 72 h. ment in protection assays was compared using the log-rank test. Statistical They were then embedded in low-melting-point paraffin (37°C; polyethyleneg- analysis was performed using the SPSS or Graphpad software. lycol distearate; Sigma-Aldrich, St. Louis, MO). Sections (5–6 ␮m) were deparaffinized in absolute ethanol, air dried, and stained with Fontana-Masson for melanin pigments. RESULTS Peptides and Cell Lines. Peptides corresponding to murine TRP2 181–188 Treatment with Heavy Metal Leads to Vitiligo Development in (VYDFFVWL), TRP2 (SVYDFFVWL), and GP100 (EGSRN- 180–188 25–33 MT/ret Mice Spontaneously Developing Melanoma. In the present QDWL) melanocyte lineage normal differentiation proteins were synthesized and purified by high-performance liquid chromatography to Ͼ90% purity. study, we used transgenic mice expressing the human ret oncogene Lyophilized peptides were diluted to 2 mg/ml in 10% DMSO and 90% water back-crossed five times with C57BL/6 mice. After 3 months, all and were aliquoted and stored at Ϫ20°C. MT/ret mice developed benign melanocytic tumors that slowly pro- A Melanoma cell line (Melan-ret) was established from a facial nodule that gressed to cutaneous melanoma in 65% of the mice within 10 months developed in a MT/ret mouse backcrossed six times with the C57BL/6 strain. (12). Tumor cells metastasized mostly in lymph nodes, lung, and This line was grown in RPMI 1640 with 10% FCS, 2 mML-glutamine, 50 ␮M brain. These are also privileged metastatic sites in melanoma patients. ␤-mercaptoethanol, 100 units/ml penicillin, and 100 ␮g/ml streptomycin (com- Because the transgene construct contained a MT promoter, we thought plete medium). For ELISPOT assays, this line was cultured for 24 h in we could promote melanoma development by adding zinc to the water ␥ complete medium containing 100 units/ml mouse recombinant IFN- (Boeh- given to mice from birth. In a first series of 88 mice, the tumor ringer Mannheim, Mannheim, Germany). IFN-␥ treatment of Melan-ret cells incidence increased, and melanoma progressed more rapidly in mice significantly enhanced MHC expression (data not shown), as already reported for B16 melanoma cells (13), and thereby increased their probability of being bred under these conditions compared with those described previously recognized by melanoma-specific T cells. Cells were harvested and washed (12). Indeed, 80% of our MT/ret mice displayed exophthalmus or before use for in vitro assays. EL-4 thymoma (H-2b) cells (American Type cutaneous nodules 6 months after birth. Exophthalmus was in most Culture Collection, Manassas, VA) were cultured in complete medium and cases the first sign of tumor development. It was followed by the used as controls. development of cutaneous melanocytic nodules on the face (cheek, Biopsy skin specimens were minced and placed in 6-well plates in complete nose, ear, and neck) and on the back of the head (back, tail, leg medium supplemented with IL-2 (100 IU/ml; Boehringer Mannheim) added muscles, and genitals). Sixty-eight % of mice 6 months of age dis- once a week. Vitiligo infiltrating lymphocytes were harvested within 4 weeks played exophthalmus, in contrast to the 35 and 26% of mice that of culture for ELISPOT assay. displayed facial and dorsal nodules, respectively. Exophthalmus was ␥ ELISPOT-IFN- . Nitrocellulose microplates (Millipore, Molsheim, diagnosed earlier than facial (P ϭ 0.004) and dorsal (P ϭ 0.001) France) were coated with rat anti-mouse IFN-␥ antibody (R4–6A2; Phar- nodules. No significant difference was observed between the onset of Mingen, Le Pont de Claix, France) and then washed and saturated with ϭ complete medium. Blood samples were collected from the retroorbital sinus in facial and dorsal nodules (P 0.656). heparin, and peripheral blood lymphocytes (PBLs) were separated on lympho- A clear-cut correlation between mortality and exophthalmus or cyte separation medium (NycoPrep A 1.077, Axis-Shield, Norway). Triplicate nodules was observed. The development of dorsal nodules was a aliquots of freshly isolated PBLs (1–5 ϫ 105 cells/well) were added to the worse prognostic factor (Pearson coefficient, r ϭ 0.841; P Ͻ 0.001; wells containing complete medium with 30 units/ml human recombinant IL-2 n ϭ 36). The correlation coefficient with mortality decreased to 0.73 (Boehringer Mannheim) with either Melan-ret cells (103 cells/well) or EL-4- (P Ͻ 0.001) for exophthalmus and facial nodules. These observations 4 Ϫ6 presenting cells (10 cells/well), loaded or not with 10 M TRP2181–188, allowed us to define a severity grade of melanoma, where exophthal- TRP2180–188, and GP10025–33 peptides. After 20 h at 37°C, biotinylated rat mus and facial nodules corresponded to a less aggressive disease ␥ anti-mouse IFN- mAb (XMG1.2; PharMingen) was added for overnight compared with dorsal nodules. Skin of MT/ret mice exhibited a black incubation. After washing, alkaline phosphatase-labeled ExtrAvidin (Sigma) color attributable to melanosis as illustrated on newborn mice (Fig. was added, and wells were incubated for1hatroom temperature. Finally, alkaline phosphatase conjugate substrate (Bio-Rad Laboratories, Hercules, 1A) and on depilated adult mice (Fig. 1B, left mouse) in contrast to the CA) was added for 30 min. Spot-forming cells were counted using a computer- pink color of control nontransgenic mice (data not shown). Unexpect- assisted video image analyzer (Zeiss-Kontron, Jena, Germany). edly, vitiligo was observed in nearly one-half of MT/ret animals (45 CD8 T-Cell Depletion in Vivo. Rat IgG anti-mouse CD8 (clone 2.43; TIB of 88), and this depigmentation was permanent. The site and pattern 210; American Type Culture Collection) was purified as described previously of pigment loss differed from mouse to mouse [face, back (right (14). Antibody was injected i.p. (250 ␮g) 1 day before the tumor challenge and mouse in Fig. 1B) or tail (Fig. 1, C and D)]. Vitiligo occurred as every 4 days (500 ␮g/injection) until day 40 after tumor challenge. More than depigmentation surrounding tumors (Fig. 1D) or distant from tumor ϩ 98% of circulating CD8 T cells were depleted by this procedure, as verified nodules (Figs. 1, B and C). Histopathological analysis of skin samples by FACS analysis using an anti-CD8 (clone 53–6.7; PharMingen) rat mono- taken from the margin of vitiligo patches confirmed the loss of clonal antibody that recognizes an epitope different from that recognized by melanocytes in the epidermis and dermis. Fontana-Masson staining of the depleting mAb. Cells were analyzed on a Coultronics Epics Elite instru- ment. Purified rat IgG (Sigma-Aldrich) was used as control. skin sections (Fig. 1, E, middle part and F, right part) revealed the Antitumor Activity. The protection against tumor growth was assessed by absence of pigments in the depigmented area of the tail (Fig. 1C). injecting 5 ϫ 105 Melan-ret cells in 100 ␮l of PBS into the right flank of Only few pigments were still detectable in the deep dermis (Fig. 1E). control (nontransgenic) and MT/ret mice. To evaluate the role of CD8ϩ T cells Interestingly, in sections of some cutaneous nodules, pigments were in the melanoma control, MT/ret mice with vitiligo were treated with rat IgG absent in a large proportion of the dermis above the tumor nodule 1497

1H). To assess such a hypothesis, we monitored the tumor-specific IFN-␥ response in MT/ret melanoma mice and compared it with the response in mice with concurrent vitiligo. Many reports have indicated that IFN-␥ plays a critical role in the development of tumor immunity, and recent reports including ours using IFN-␥ knock-out mice demonstrated a crucial role for IFN-␥ in the control of B16 melanoma growth in vivo (15–17). To evaluate natural tumor-specific response, ex vivo IFN-␥-ELISPOT assays were performed by using PBLs isolated from MT/ret mice as effector cells and a Melan-ret cell line as melanoma antigen-presenting cells. This Melan-ret cell line was established from a nodule of a MT/ret mouse. IFN-␥ secretion by PBLs in response to Melan-ret cells revealed that naturally processed melanoma-associated antigens were presented to and recognized by T cells (Fig. 2A). IFN-␥ was not secreted when EL-4 thymoma cells were used as antigen-presenting cells (not shown). The number of melanoma-specific PBLs was significantly different from mouse to mouse and may be as high as 1 PBL of 1000 PBLs in mice without vitiligo (Fig. 2A, left part) and as high as 1 PBL of 680 in mice with melanoma-associated vitiligo (Fig. 2A, right part). Reactivity was found in 61% (28 of 46) of melanoma mice. Thirty-nine % (18 of 46) of them displaying Ͻ1 IFN␥-secreting PBL of 104 in response to Melan-ret cells were classified as nonresponders. Interestingly, the number of responders decreased as the severity of symptoms increased. Indeed, 71% (10 of 14) and 69% (11 of 16) of melanoma mice with exophthalmus and facial nodules, respectively, had IFN-␥- secreting cells in response to melanoma cells, whereas only 44% (7 of 16) of animals with dorsal nodules showed this response (Fig. 2B). Fig. 1. Spontaneous vitiligo in MT/ret melanoma mice. Features of vitiligo and Antitumor reactivity was found in 93% (25 of 27) of mice with correlation with tumor growth suppression. A–D, photographs of MT/ret transgenic mice. Pigmentation of the whole skin after birth (A, 1–5-day-old littermate mice). The melanosis melanoma-associated vitiligo, and in contrast to mice without vitiligo, is homogeneous in newborn (A) and in adult (B, left mouse) mice. In B–D, depigmentation this reactivity did not significantly depend on disease severity (Fig. occurs on the back (B, right mouse) and on the tail (C and D). Note a small tumor nodule on the depigmented tail (D). E and F, skin biopsy specimens at the margin between the 2B). To further determine the immune effector cells responsible for normal and vitiliginous tail skin of the 30-week-old mouse in C were stained with melanocyte destruction, biopsy samples taken from the margin of a ϫ Fontana-Masson ( 125). Melanin-containing cells are abundant in the dermis and epi- vitiligo patch were minced and cultured as we reported recently in dermis of the pigmented skin, and only a few such cells are detectable in the deep dermis of the depigmented zone. G, biopsy of a cutaneous nodule consisting of melanoma cells humans (3). Interestingly, vitiligo-infiltrating lymphocytes that pro- stained with Fontana-Masson (ϫ125). The upper dermis nodule shows melanin-producing liferated expressed CD8 antigen and specifically recognized Melan- cells on the right, in contrast with absence of pigments on the left. H, comparison of the percentage of 6-month-old mice with melanoma alone (n ϭ 43) or with melanoma- ret cells (Fig. 2C). associated vitiligo (n ϭ 45) that presented defined clinical signs (exophthalmus, facial and One of our recent studies in human has highlighted the natural dorsal nodules). No significant difference in exophthalmus incidence was found when ϩ melanoma mice (n ϭ 31), and mice with melanoma-associated vitiligo (n ϭ 26) were induction of CD8 T cell-mediated cellular immunity in the course of compared (␹2 test, P ϭ 0.095). In contrast, facial nodules were significantly more frequent vitiligo and melanoma progression (3). Therefore, we focused our in melanoma mice (n ϭ 20) than in mice with melanoma-associated vitiligo (n ϭ 10, investigations on CD8ϩ T cell-mediated antitumor immunity. As a P ϭ 0.012). Similar data were obtained with dorsal nodules that were also significantly more frequent in melanoma mice (n ϭ 15) than in mice with melanoma-associated vitiligo prelude to studying the fine antigen specificities of tumor-specific (n ϭ 10, P ϭ 0.03). CD8ϩ T cells, the expression of MDA transcripts was studied in the tumors by reverse transcription-PCR. TRP2 and GP100 transcripts were expressed in most organs (skin, lymph nodes, and lungs) mac- (Fig. 1G, left part). The area with no pigments bordered a pigmented roscopically invaded by tumor cells (data not shown). TRP2 area (right part) in which melanocytes were present from the epider- 181–188 (18) and TRP2 (19) epitopes are presented by the Kb MHC mis down to the deep dermis. 180–188 class I molecule, whereas GP100 epitope (20) is presented by Db. Finally, the relationship between vitiligo and melanoma is summa- 25–33 Interestingly, pooled PBLs from mice responding with a high fre- rized in Fig. 1H. Exophthalmus was not significantly delayed in MT/ret animals with melanoma-associated vitiligo compared with quency to the Melan-ret cell line also responded with a high frequency mice without vitiligo. In contrast, animals 6 months of age with to a pool of these TRP2 and GP100 epitopes (Fig. 2D). To ensure that melanoma-associated vitiligo displayed significantly less facial and both epitopes were recognized, pools of PBLs isolated from MT/ret dorsal nodules than animals of the same age without vitiligo. This mice were restimulated with TRP2- or GP100-derived peptides sep- difference was even more significant in vitiligo mice with dorsal arately. T cells specific for each epitope were detected in PBLs from nodules. responder mice (Fig. 2D). PBLs from C57BL/6 mice immunized with ␥ Spontaneous Antitumor T-Cell Response in MT/ret Mice. Some human GP10025–33 or murine TRP2180–188 peptides secreted IFN- MT/ret mice exhibited a melanoma with cutaneous nodules that both in response to these peptides and to the Melan-ret cells, confirm- progressed very slowly (over 6 months), suggesting that a melanoma- ing that these peptides are naturally presented by the Melan-ret cell induced immune response may have a protective effect in vivo. line (not shown). As expected, no significant response against

Moreover, the reduced number of melanoma symptoms in MT/ret GP10025–33 and TRP2180–188 peptides was found when pooled PBLs mice with vitiligo compared with mice with melanoma alone sug- isolated from MT/ret mice that exhibited no antitumor response were gested that self-reactivity can be useful for melanoma control (Fig. used (data not shown). 1498

Spontaneous Vitiligo Development Is Associated with Mela- noma Protection. To assess whether self-reactivity can be useful for melanoma control, MT/ret mice were challenged s.c. with Melan-ret cells. The outgrowth of the s.c. tumor (referred as s.c. tumor) was significantly suppressed in mice with melanoma-associated vitiligo. Indeed, 5 of 12 MT/ret mice in this group (42%) were protected 40 days after transplantation (Fig. 3A). In contrast, s.c. tumor developed in most control littermates (82%). Tumor incidence was even higher in MT/ret mice that had already spontaneously developed a melanoma (88%). The difference in s.c. tumor development was more significant 140 days after tumor challenge, when 25% of MT/ret mice with vitiligo were still protected, in contrast to 100% of MT/ret mice with melanoma alone that developed a s.c. tumor (data not shown, Fisher’s exact test, P ϭ 0.0261). Interestingly, 6 of 6 mice protected against tumor at day 40 displayed a significantly higher level of melanoma- specific PBLs, in contrast to mice that developed a s.c. tumor (Fig. 3B). To further assess the role of the CD8ϩ T-cell subpopulation in controlling melanoma in MT/ret mice with melanoma-associated vitiligo, depletion experiments were conducted starting 1 day before the challenge with Melan-ret cells. Two of 6 mice (33%) injected with the rat immunoglobulin non-depleting antibody controlled the s.c. tumor (Fig. 4A), consistent with the number of mice resistant in the absence of antibody treatment (Fig. 3A). In contrast, CD8ϩ T-cell depletion favored the transplanted tumor outgrowth. In this latter group, tumor incidence was increased and tumor development was more rapid. Thirty-three % of MT/ret mice treated with control antibody were still free of tumor at day 100 after tumor challenge in contrast to CD8ϩ T-cell depleted mice, who were all dead by day 65 (Fig. 4B).

DISCUSSION In this study, we used MT/ret transgenic mice, an animal model for human cutaneous melanoma recently established by some of us. We investigated for the first time in this model the natural antitumor T-cell response and further evaluated the role of cellular immunity in the course of disease development. Interestingly, our data show that a large proportion of MT/ret animals spontaneously developed melanoma-associated vitiligo. To our knowledge, this is the first mouse model in which melanoma leads to vitiligo. The pattern we observed bears strong resemblance to the vitiligo that occurs spontaneously in some melanoma patients. In humans, melanoma therapy also sometimes leads to the development of vitiligo that is significantly associated with a positive clinical response (4–6, 21, 22). Similarly, we obtained a good correlation between vitiligo development and melanoma control in our MT/ret mouse model. Indeed, mice with vitiligo and melanoma displayed significantly fewer tumor nodules than mice of the same age with melanoma alone (Fig. 1H). Vitiligo was rarely observed without signs of melanoma. However, we cannot exclude Fig. 2. Spontaneous tumor-specific T-cell response in MT/ret melanoma animals. A, specific recognition by PBLs of Melan-ret cells of individual MT/ret mice with melanoma the possibility that some nodules may be too small to be detectable by (n ϭ 46) or with melanoma-associated vitiligo (n ϭ 27). IFN-␥-producing cell number gross clinical examination. Therefore, it remains difficult to determine was assessed in mice displaying an exophthalmus, facial nodules, or dorsal nodules. All whether vitiligo delays melanoma outgrowth or whether melanoma of the mice used in this experiment are 1–14-month-old with no significant difference of mean age in mice with or without vitiligo. All mice with facial nodules (2–4 mm) have precedes vitiligo in this model. a similar tumor load. Similarly, dorsal nodules (2–5 mm) are comparable from one mouse Autoreactive T cells have been identified in the blood of patients to another in terms of tumor load. Background values obtained with control nontransgenic with progressive autoimmune vitiligo (2) and in the depigmented skin mice were subtracted from the mean values obtained with MT/ret mice. Mice were considered responders if the number of IFN-␥-secreting cells per 106 PBLs was over 100 of patients with spontaneous melanoma-associated vitiligo (3). Viti- spots (dotted line). Ninety-three % of mice with melanoma-associated vitiligo were ligo also occurs in melanoma patients after infusion of melanoma- responders, which contrasts significantly with the 61% of melanoma mice without vitiligo (Fisher’s exact test, P ϭ 0.008). B, percentage of responders as a function of disease severity in mice with melanoma (f) or with melanoma-associated vitiligo (Ⅺ). C, specific recognition of melanoma cells by vitiligo infiltrating lymphocytes. The number of cells that secrete IFN-␥ was assessed using vitiligo infiltrating cells (6 ϫ 105 cells/well) as and recombinant IL-2 at a final concentration of 30 units/ml. EL-4 cells were coated with 3 effector cells and Melan-ret cells (10 cells/well) as targets. EL-4 cells were used as a pool of TRP2180–188, TRP2181–188, and GP10025–33 peptides or coated with the peptides control nonmelanoma tumor cells. D, comparison between the recognition of peptide- individually. The values obtained with EL-4 cells in the absence of peptides were always 6 derived tumor antigens and that of Melan-ret cells. Specific recognition of TRP2180–188, below 10 spots (data not shown). Results are expressed as specific spots per 10 PBLs (in TRP2181–188, and GP10025–33 murine CD8 T-cell epitopes by pools of PBLs from mice A and C) (mean of triplicate tests; bars, SD). These results are representative of two displaying tumor reactivity. Freshly isolated PBLs were incubated for 24 h with EL-4 cells independent experiments. 1499

by PBLs from nontransgenic littermates. T cells that secrete IFN-␥ in response to Melan-ret cells were statistically more frequent in melanoma mice that developed vitiligo (mean, 0.058% of PBLs in 27 mice) than in mice that did not (mean, 0.029% of PBLs in 46 mice). This result suggests a crucial role for IFN-␥-secreting T cells in tumor control. Moreover, in contrast to melanoma mice, the number of responder mice with melanoma-associated vitiligo did not decrease with disease progression. The antitumor reactivity is mediated by ϩ CD8 T cells that are specific for TRP2- and GP100-derived epitopes, the only B16 tumor rejection antigens that have been characterized to date. The number of IFN-␥-secreting PBLs was as high as 1 PBL of 1686 in response to Melan-ret cells and 1 of 2673 in response to cells pulsed with TRP2- and GP100-derived peptides (Fig. 2D). Given an average of 10% CD8ϩ T cells in PBLs, the mean number of circulating CD8ϩ T lymphocytes specific for melanoma-derived peptides from responder mice corresponded to 1 of 270 (0.37%). Interestingly, this is relatively close to the proportion found in humans. Indeed, the ϩ circulating CD8 T-cell population specific for Mart27–35 and for tyrosinase368–376 may represent 0.014–0.5% and 0.19–2% of total CD8ϩ T cells, respectively, in HLA-A*0201-positive metastatic melanoma patients (27–29). Tumor-specific circulating lymphocytes in

MT/ret mice are more frequent than cells specific for TRP2181–188, TRP2180–188, and GP10025–33, suggesting recognition of other CD8 T-cell epitopes presented at the Melan-ret cell surface and which remain to be identified. Melanoma-specific T cells in freshly isolated Fig. 3. The spontaneous vitiligo onset correlates with melanoma protection. A, anti- PBLs of MT/ret mice are currently used to identify naturally pro- tumor protection assay. 5 ϫ 105 Melan-ret cells were injected s.c. into control nontrans- cessed and MHC class I-presented peptides derived from the Melan- genic mice (triangles; n ϭ 16) or into MT/ret mice displaying melanoma-associated ϭ ϭ ret line. In this study, we focused our investigations on T-cell- clinical signs with (squares; n 12), or without vitiligo (circles; n 17). The mice with ϩ vitiligo were 2–15 months of age (mean age, 9.7 months) at the time of the tumor cell mediated antitumor immunity and more particularly on CD8 T cells. transplantation. They all displayed exophthalmus, 4 of 12 (33%) displayed facial nodules, and no mice displayed dorsal nodules attributable to the rare occurrence of such a nodule in this group. The mice without vitiligo were 2–16 months of age (mean age, 7.8 months). They all displayed exophthalmus, 14 of 17 (82%) had facial nodules, and 10 of 17 (59%) had dorsal nodules. As expected, the selection of animals of the same age in both groups results in different tumor load. The s.c. transplanted tumor growth was monitored twice a week. Values are expressed as percentage of s.c. tumor-free mice at the indicated time after tumor cell transplantation (results of two separate experiments). Mice with vitiligo showed a significant delay in s.c. tumor development compared with mice without vitiligo (log-rank test, P ϭ 0.0463). s.c. tumor development was not significantly different in mice with melanoma alone compared with control littermate mice (P Ͼ 0.05). B, specific recognition of Melan-ret cells by PBLs of MT/ret mice displaying melanoma-associated symptoms with vitiligo (squares; n ϭ 5) or without vitiligo (circles; n ϭ 4). Tested mice are either protected (6 mice) or not protected (3 mice) against tumor challenge. Results are expressed as specific spots per 106 PBLs (mean of triplicate tests; bars, SD). specific CD8ϩ T cells (5, 6). In existing animal models, immunotherapy against melanoma may favor the appearance of vitiligo. Indeed, vaccine consisting of granulocyte/macrophage-colony stimulating factor-producing irradiated B16 cells and CTLA-4 blockade success- fully treated mice carrying a small load of B16 melanoma cells. In the surviving mice, a melanoma-specific CD8ϩ T cell-dependent mechanism produces depigmented lesions (19). Similarly, adoptive transfer of T cells specific for MDA-derived peptides favors induction of vitiligo (17, 23). CTL-mediated depigmentation has also been reported in mice immunized with DNA encoding human TRP-2 (24). Moreover, the antitumor effect of locally secreted IL-12 on B16 melanoma cells mediated by CD8ϩ T cells also leads to vitiligo-like coat color alteration (25). Finally, the tissue damage observed after Fig. 4. Role of CD8ϩ T cells in the control of melanoma in MT/ret mice with vitiligo. melanoma treatment may result from an immune response independ- A, MT/ret mice displaying melanoma-associated vitiligo received injections of the fol- ϩ lowing rat antibodies: control IgG (Ⅺ; n ϭ 6) or anti-CD8 (f; n ϭ 6). One day later, mice ent of CD8 T cells. Indeed, mice vaccinated with recombinant were injected s.c. with 5 ϫ 105 Melan-ret cells. Antibody treatments were repeated every vaccinia virus encoding murine TRP1 display extensive skin depig- 4 days until day 40. All of the mice displayed vitiligo and exophthalmus without facial or mentation, and this melanocyte-directed autoimmunity depends on dorsal nodules; therefore, they displayed no significant difference of tumor load. Mice in ϩ both groups were 4–7 months of age. Tumor growth was monitored twice a week. Values both TRP1-specific antibodies and CD4 T cells (26). are expressed as a percentage of tumor-free mice at the indicated time after tumor cell Our results show a significant frequency of PBLs derived from transplantation. The CD8 T-cell-depleted group had a tumor incidence higher than the ␥ group treated with control non-depleting IgG, and tumor development was enhanced MT/ret mice that secrete IFN- ex vivo on recognition of Melan-ret (log-rank test, P ϭ 0.0271). B, survival curve of mice in A treated with the control rat IgG melanoma cells, in marked contrast to the absence of IFN-␥ secretion (Ⅺ) or anti-CD8 (f). 1500

However, cells that secrete IFN-␥ in response to Melan-ret cells were 10. Cui, J., Chen, D., Misfeldt, M. L., Swinfard, R. W., and Bystryn, J. C. Antimelanoma also found in PBLs depleted of CD8ϩ lymphocytes (data not shown), antibodies in swine with spontaneously regressing melanoma. Pigment Cell Res., 8: 60–63, 1995. suggesting the recognition of tumor-derived CD4 T-cell epitopes. 11. Iwamoto, T., Takahashi, M., Ito, M., Hamatani, K., Ohbayashi, M., Wajjwalku, W., Such naturally processed and MHC class II-presented peptides de- Isobe, K., and Nakashima, I. Aberrant melanogenesis and melanocytic tumour derived from murine melanoma cells have yet to be identified. velopment in transgenic mice that carry a metallothionein/ret fusion gene. EMBO J., 10: 3167–3175, 1991. Thirty % of mice with melanoma-associated vitiligo were totally 12. Kato, M., Takahashi, M., Akhand, A. A., Liu, W., Dai, Y., Shimizu, S., Iwamoto, T., resistant to challenge with syngeneic Melan-ret cells (Fig. 3A), and Suzuki, H., and Nakashima, I. Transgenic mouse model for skin malignant melanoma. this effect depended on CD8ϩ T cells (Fig. 4A). Interestingly, 4 of 4 Oncogene, 17: 1885–1888, 1998. 13. Bohm, W., Thoma, S., Leithauser, F., Moller, P., Schirmbeck, R., and Reimann, J. T vitiligo mice that were protected had a higher frequency of melanoma- cell-mediated, IFN-␥-facilitated rejection of murine B16 melanomas. J. Immunol., specific T cells than the vitiligo mouse that developed a s.c. tumor as 161: 897–908, 1998. ␥ ϩ 14. Belnoue, E., Kayibanda, M., Vigario, A. M., Deschemin, J. C., van Rooijen, N., assessed by IFN- ELISPOT. Our data indicate that CD8 T cells Viguier, M., Snounou, G., and Renia, L. On the pathogenic role of brain-sequestered mediate, at least in part, spontaneous development of vitiligo, which ␣␤ CD8ϩ T cells in experimental cerebral malaria. J. Immunol., 169: 6369–6375, can be seen as an early sign of protective immunity against melanoma. 2002. 15. Dobrzanski, M. J., Reome, J. B., and Dutton, R. W. Type 1 and type 2 CD8ϩ effector This model enables antitumor response and tumor escape to be T cell subpopulations promote long-term tumor immunity and protection to progres- studied in parallel. The reduced number of responders in mice with sively growing tumor. J. Immunol., 164: 916–925, 2000. dorsal nodules may be attributable to poor expansion of tumor- 16. Prevost-Blondel, A., Neuenhahn, M., Rawiel, M., and Pircher, H. Differential re- ␥ quirement of perforin and IFN-␥ in CD8 T cell-mediated immune responses against specific cells or to the loss of their capacity to secrete IFN- .Itis B16.F10 melanoma cells expressing a viral antigen. Eur. J. Immunol., 30: 2507–2515, therefore crucial to characterize in longitudinal in vivo studies the 2000. different functional stages (homing and effector capacities) of T 17. Winter, H., Hu, H-M., McClain, K., Urba, W. J., and Fox, B. A. Immunotherapy of melanoma: a dichotomy in the requirement for IFN-␥ in vaccine-induced antitumor lymphocytes and their frequency. In melanoma patients, tumor- ϩ immunity versus adoptive immunotherapy. J. Immunol., 166: 7370–7380, 2001. specific CD8 T cells may display an antigen-experienced phenotype 18. Bloom, M. B., Perry-Lalley, D., Robbins, P. F., Li, Y., El-Gamil, M., Rosenberg, both in the periphery and in metastatic lymph nodes (27–31). Alter- S. A., and Yang, J. C. Identification of tyrosinase-related protein 2 as a tumor ϩ rejection antigen for the B16 melanoma. J. Exp. Med., 185: 453–459, 1997. natively, melanoma-specific CD8 T cells may become functionally 19. van Elsas, A., Hurwitz, A. A., and Allison, J. P. Combination immunotherapy of B16 unresponsive as shown for tyrosinase368–376-specific T cells in one melanoma using anti-cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) and patient (28). The MT/ret model will be essential for in vivo study of granulocyte/macrophage colony-stimulating factor (GM-CSF)-producing vaccines induces rejection of subcutaneous and metastatic tumors accompanied by autoim- the functional status of antimelanoma effector cells at different stages mune depigmentation. J. Exp. Med., 190: 355–366, 1999. of tumor development, because such longitudinal studies are not 20. Overwijk, W. W., Tsung, A., Irvine, K. R., Parkhurst, M. R., Goletz, T. J., Tsung, K., possible in humans. The MT/ret model will also be helpful for further Carroll, M. W., Liu, C., Moss, B., Rosenberg, S. A., and Restifo, N. P. gp100/pmel ϩ ϩ 17 is a murine tumor rejection antigen: induction of “self”-reactive, tumoricidal T assessment of the role of autoantibodies and CD4 or CD8 T cells cells using high-affinity, altered peptide ligand. J. Exp. Med., 188: 277–286, 1998. in the development of spontaneous melanoma-associated vitiligo 21. Duhra, P., and Ilchyshyn, A. Prolonged survival in metastatic malignant melanoma compared with treatment-induced vitiligo. associated with vitiligo. Clin. Exp. Dermatol., 16: 303–305, 1991. 22. Nordlund, J. J., Kirkwood, J. M., Forget, B. M., Milton, G., Albert, D. M., and Lerner, A. B. Vitiligo in patients with metastatic melanoma: a good prognostic sign. J. Am. ACKNOWLEDGMENTS Acad. Dermatol., 9: 689–696, 1983. 23. Colella, T. A., Bullock, T. N. J., Russell, L. B., Mullins, D. W., Overwijk, W. W., Luckey, C. J., Pierce, R. A., Restifo, N. P., and Engelhard, V. H. Self-tolerance to the We thank Professor I. Bourgault and Dr. A. Carlotti for helpful discussions, murine homologue of a tyrosinase-derived melanoma antigen: implications for tumor A. Strickland for editing the English text, J-C. Deschemin for antibody immunotherapy. J. Exp. Med., 191: 1221–1232, 2000. purification, and M. Plut for animal care and disease monitoring. 24. Bowne, W. B., Srinivasan, R., Wolchok, J. D., Hawkins, W. G., Blachere, N. E., Dyall, R., Lewis, J. J., and Houghton, A. N. Coupling and uncoupling of tumor immunity and autoimmunity. J. Exp. Med., 190: 1717–1722, 1999. REFERENCES 25. Nagai, H., Hara, I., Horikawa, T., Oka, M., Kamidono, S., and Ichihashi, M. Elimination of CD4(ϩ) T cells enhances anti-tumor effect of locally secreted inter- 1. Das, P. K., van den Wijngaard, R., Wankowicz-Kalinska, A., and Le Poole, C. A leukin-12 on B16 mouse melanoma and induces vitiligo-like coat color alteration. symbiotic concept of autoimmunity and tumour immunity: lessons from vitiligo. J. Investig. Dermatol., 115: 1059–1064, 2000. Trends Immunol., 22: 130–136, 2001. 26. Overwijk, W. W., Lee, D. S., Surman, D. R., Irvine, K. R., Touloukian, C. E., Chan, 2. Ogg, G. S., Dunbar, P. R., Romero, P., Chen, J. L., and Cerundolo, V. High frequency C. C., Carroll, M. W., Moss, B., Rosenberg, S. A., and Restifo, N. P. Vaccination with of skin-homing melanocyte-specific cytotoxic T lymphocytes in autoimmune vitiligo. a recombinant vaccinia virus encoding a “self” antigen induces autoimmune vitiligo J. Exp. Med., 188: 1203–1208, 1998. and tumor cell destruction in mice: requirement for CD4(ϩ) T lymphocytes. Proc. 3. Le Gal, F. A., Avril, M. F., Bosq, J., Lefebvre, P., Deschemin, J. C., Andrieu, M., Natl. Acad. Sci. USA, 96: 2982–2987, 1999. Dore, M. X., and Guillet, J. G. Direct evidence to support the role of antigen-specific 27. Anichini, A., Molla, A., Mortarini, R., Tragni, G., Bersani, I., Di Nicola, M., Gianni, CD8(ϩ) T cells in melanoma-associated vitiligo. J. Investig. Dermatol., 117: 1464– A. M., Pilotti, S., Dunbar, R., Cerundolo, V., and Parmiani, G. An expanded 1470, 2001. peripheral T cell population to a cytotoxic T lymphocyte (CTL)-defined, melanocyte- 4. Rosenberg, S. A., and White, D. E. Vitiligo in patients with melanoma: normal tissue specific antigen in metastatic melanoma patients impacts on generation of peptide- antigens can be targets for cancer immunotherapy. J. Immunother. Emphasis Tumor specific CTLs but does not overcome tumor escape from immune surveillance in Immunol., 19: 81–84, 1996. 5. Dudley, M. E., Wunderlich, J. R., Robbins, P. F., Yang, J. C., Hwu, P., Schwartzen- metastatic lesions. J. Exp. Med., 190: 651–668, 1999. truber, D. J., Topalian, S. L., Sherry, R., Restifo, N. P., Hubicki, A. M., Robinson, 28. Lee, P. P., Yee, C., Savage, P. A., Fong, L., Brockstedt, D., Weber, J. S., Johnson, D., M. R., Raffeld, M., Duray, P., Seipp, C. A., Rogers-Freezer, L., Morton, K. E., Swetter, S., Thompson, J., Greenberg, P. D., Roederer, M., Davis, M. M., van Elsas, Mavroukakis, S. A., White, D. E., and Rosenberg, S. A. Cancer regression and A., Hurwitz, A. A., and Allison, J. P. Characterization of circulating T cells specific autoimmunity in patients after clonal repopulation with antitumor lymphocytes. for tumor-associated antigens in melanoma patients. Nat. Med., 5: 677–685, 1999. Science (Wash. DC), 298: 850–854, 2002. 29. Pittet, M. J., Valmori, D., Dunbar, P. R., Speiser, D. E., Lienard, D., Lejeune, F., 6. Yee, C., Thompson, J. A., Roche, P., Byrd, D. R., Lee, P. P., Piepkorn, M., Kenyon, Fleischhauer, K., Cerundolo, V., Cerottini, J. C., and Romero, P. High frequencies of ϩ K., Davis, M. M., Riddell, S. R., and Greenberg, P. D. Melanocyte destruction after naive Melan-A/MART-1-specific CD8( ) T cells in a large proportion of human histocom- antigen-specific immunotherapy of melanoma: direct evidence of T cell-mediated patibility leukocyte antigen (HLA)-A2 individuals. J. Exp. Med., 190: 705–715, 1999. vitiligo. J. Exp. Med., 192: 1637–1644, 2000. 30. Pittet, M. J., Zippelius, A., Speiser, D. E., Assenmacher, M., Guillaume, P., 7. Lerner, A. B., Shiohara, T., Boissy, R. E., Jacobson, K. A., Lamoreux, M. L., and Valmori, D., Lienard, D., Lejeune, F., Cerottini, J. C., and Romero, P. Ex vivo Moellmann, G. E. A mouse model for vitiligo. J. Investig. Dermatol., 87: 299–304, IFN-␥ secretion by circulating CD8 T lymphocytes: implications of a novel 1986. approach for T cell monitoring in infectious and malignant diseases. J. Immunol., 8. Naughton, G. K., Mahaffey, M., and Bystryn, J. C. Antibodies to surface antigens of 166: 7634–7640, 2001. pigmented cells in animals with vitiligo. Proc. Soc. Exp. Biol. Med., 181: 423–426, 31. Romero, P., Dunbar, P. R., Valmori, D., Pittet, M., Ogg, G. S., Rimoldi, D., Chen, 1986. J. L., Lienard, D., Cerottini, J. C., and Cerundolo, V. Ex vivo staining of metastatic 9. Erf, G. F., Trejo-Skalli, A. V., and Smyth, J. R., Jr. T cells in regenerating feathers lymph nodes by class I major histocompatibility complex tetramers reveals high of Smyth line chickens with vitiligo. Clin. Immunol. Immunopathol., 76: 120–126, numbers of antigen-experienced tumor-specific cytolytic T lymphocytes. J. Exp. 1995. Med., 188: 1641–1650, 1998. 1501

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2004 American Association for Cancer Research. Spontaneous Vitiligo in an Animal Model for Human Melanoma: Role of Tumor-specific CD8+ T Cells

Renée Lengagne, Frédérique-Anne Le Gal, Marylène Garcette, et al.

Cancer Res 2004;64:1496-1501.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/64/4/1496

Cited articles This article cites 31 articles, 17 of which you can access for free at: http://cancerres.aacrjournals.org/content/64/4/1496.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/64/4/1496.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/64/4/1496. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.