Int. J. Cancer: 70, 98–105 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer DENDRITIC EPIDERMAL T-CELL INVOLVEMENT IN INDUCTION OF CD81 T CELL-MEDIATED IMMUNITY AGAINST AN ULTRAVIOLET RADIATION-INDUCED SKIN TUMOR Lois L. CAVANAGH1*, Ross StC. BARNETSON1, Antony BASTEN2 and Gary M. HALLIDAY1 of Medicine (Dermatology), University of Sydney at Royal Prince Alfred Hospital, Camperdown, Australia 2Centenary Institute of Cancer Medicine and Cell Biology, Camperdown, Australia 1Department Murine epidermis contains 2 distinct cell populations which contribute to the skin immune system, Langerhans cells (LC), and dendritic epidermal T cells (DETC). LCs are important in the induction of immunity against a wide range of antigens; however, the function of DETC is unclear. To investigate the roles of these epidermal cells (EC) in protective antitumor immunity, an in vivo model of an ultraviolet radiation-induced fibrosarcoma, UV-13-1, was used. Mice were immunized with tumor antigen-pulsed EC followed 10 days later by an injection into the ear of 105 tumor cells, which did not lead to formation of a detectable tumor, but was intended to simulate the influence of a developing tumor on the ensuing immune response. The mice were then challenged with 2 3 106 viable tumor cells in each flank, sufficient to result in growth of a measurable tumor. Protective immunity was induced by DETC, and shown to be long-lasting, with tumors inoculated 160 days after immunization being effectively rejected. The effector cells responsible for protective immunity were CD81 T cells. Delayed-type hypersensitivity generated by tumor antigen-pulsed EC was dependent on LCs, with no involvement of DETCs. This response, in contrast to that of DETC, required prior culture of EC with GM-CSF, but failed to inhibit tumor growth or incidence. Thus DETC and LC can both activate antitumor immune responses, although only the DETC-dependent response results in protective immunity in the presence of a developing tumor. Int. J. Cancer, 70:98–105, 1997. r 1997 Wiley-Liss, Inc. The immune system is thought to play a critical role in immune surveillance against developing cutaneous tumors. The initiation of immune responses against these tumors requires both the recognition of the tumor by local antigen-presenting cells (APC), and the subsequent presentation of tumor-specific antigens to T lymphocytes (Elmets et al., 1995). Dendritic APC, which are the most potent cells for induction of primary T cell-dependent immune responses (Steinman et al., 1993), have been shown to be involved in the induction of antitumor immunity (Grabbe et al., 1991). However, the cellular events involved remain unknown, although they are likely to be influenced by certain properties of the tumor itself, including the stage at which it expresses tumor antigens, as well as other factors such as the cytokines it may produce. Most human skin tumors arise within the epidermis. Hence it is reasonable to assume that epidermal cells (EC) will play a role in the induction of skin cancers. Murine epidermis contains 2 populations of cells belonging to the immune system, i.e., dendritic Langerhans cells (LC), and dendritic epidermal T cells (DETC), which comprise 1–5% of total epidermal cells, depending on mouse strain (Stingl et al., 1978; Sullivan et al., 1985). DETC express CD3, Thy-1.2 and the gd T-cell receptor (Vg3/Cg1/Vd1) for antigen (Kuziel et al., 1987). The function of these cells has not been fully established; however, in vitro studies have shown that DETC cell lines, unlike splenic T cells, can lyse a range of skin-tumor cell lines, but not normal keratinocytes (Kaminski et al., 1993). Moreover, when admixed with the melanoma cell line K1735, DETC have been shown to prevent tumor growth after intradermal injection (Love-Schimenti and Kripke, 1994b). By contrast, other studies have suggested that DETC exert inhibitory effects on the growth of cultured tumor cells (Welsh et al., 1992; Love-Schimenti and Kripke, 1994a). LC, which function as professional antigen-presenting cells, have been shown to be important in induction of contact hypersensitivity and delayed-type hypersensitivity (DTH) (Stingl et al., 1978; Sullivan et al., 1986). There has been much speculation as to their role in antitumor immunity, although LC have been shown to be involved in induction of protective immunity against a chemically-induced tumor (Grabbe et al., 1991). As ultraviolet (UV) radiation is the major cause of skin cancers in humans (Dubin et al., 1990), and the immunogenicity, tumor antigens and biological behavior of UV-induced tumors is likely to be different than in tumors induced by other carcinogenic agents (Prehn and Main, 1957; Kripke, 1974), it is important to determine the part played by the immunocompetent cells of the epidermis in initiation of immune responses against such tumors. In an attempt to simulate the effects exerted by a tumor during the early stages of development, we immunized mice with tumor antigen-pulsed EC followed by a dose of tumor cells insufficient to form a detectable tumor, but sufficient to elicit a DTH response. Protective antitumor immunity was then assessed by inoculating the mice with a large enough number of tumor cells to produce a growing tumor. The tumor selected for this purpose was the UV-induced fibrosarcoma, UV-13-1, which grows progressively in immunocompetent mice, thus enabling the investigation of immunity which protects against tumor growth. MATERIAL AND METHODS Mice Inbred male and female mice of the C3H/HeNCr strain were between 8–12 weeks old at the beginning of the experiments. Mice were used in accordance with University of Sydney Animal Experimental Ethics Committee guidelines. Tumor cell line The UV-induced fibrosarcoma UV-13-1, originally derived in a C3H/HeNCr mouse, was kindly provided by Dr. M.L. Kripke, M.D. Anderson Cancer Center (TX). Both the tumor and mice were tested periodically by isoenzyme analysis to ensure genetic drift had not occurred. The UV-13-1 tumor line was maintained in tissue culture at 37°C and 5% CO2 in Dulbecco’s modification of Eagle’s media (DMEM; Trace Biosciences, Castle Hill, Australia) containing 10% newborn calf serum (NCS; CSL, Parkville, Australia), 20 mM HEPES buffer and 8 mM L-glutamine. UV-13-1 grows progressively when injected subcutaneously or intradermally in immunocompetent syngeneic recipients. *Correspondance to: University of Queensland Department of Medicine, Princess Alexandra Hospital, Woolloongabba, Queensland 4102, Australia. Fax: 617 3240 5399. E-mail: firstname.lastname@example.org Abbreviations: DETC, dendritic epidermal T cells; DTH, delayed-type hypersensitivity; EC, epidermal cells; LC, Langerhans cells; TE, tumor extract; UV, ultraviolet. Received 17 June 1996; revised 14 September 1996. ANTITUMOR IMMUNITY AND DENDRITIC EPIDERMAL T CELLS Preparation of tumor extract (TE) UV-13-1 tumors were grown in mice which had been immunosuppressed by prior exposure to an acute dose of UV radiation over 5 days, giving a total dose of 6.82 J/cm2 UVA, and 0.59 J/cm2 UVB. The tumors were excised when approximately 10–15 mm in diameter, placed into sterile PBS, disaggregated mechanically and then sonicated to disrupt cell membranes. The supernatant, collected after centrifugation at 1,500g for 30 min at 4°C, was dialyzed against PBS (12–14 kDa cutoff, SpectraPor Membrane Tubing, Spectrum Medical Industries, Houston, TX), and used as 99 the source of tumor antigen. Protein concentration was determined by the Lowry method. Preparation of epidermal cells (EC) EC were prepared according to a modification of a method previously described (Sullivan et al., 1985). Briefly, mice were sacrificed, and the dorsal trunk shaved and chemically depilated (Veet, Reckitt and Coleman, West Ryde, Australia). The excised skin was cut into small pieces, placed into Ca21/Mg21 free-HBSS supplemented with 20 mM HEPES (HEPES-BSS) containing 0.3% trypsin (Boehringer-Mannheim, Germany) and 300 U/ml DNase (Amersham, Takara, Japan) and incubated for 16 hr at 4°C. The epidermis was mechanically separated from the underlying dermis, and reincubated in fresh 0.3% trypsin solution containing 300 U/ml DNase for 20 min at 37°C, after which an equal volume of DMEM containing 10% NCS was added. EC were dislodged by gentle agitation, filtered through 250-µm nylon mesh (Swiss Screens, Seven Hills, Australia) and washed twice. Following preparation, EC were incubated for 2 hr at 37°C in DMEM containing 10% NCS, 20 mM HEPES, 8 mM L-glutamine and 0.05 mM 2-mercaptoethanol (complete medium) prior to use. In some experiments, 10 U/ml murine recombinant granulocytemacrophage colony-stimulating factor (GM-CSF; Genzyme, Cambridge, MA) were added during this culture. In vitro depletion of EC subsets MHC class II1 EC were removed prior to the 2-hr culture by incubation with monoclonal anti-Ia antibody (TIB 93; ATCC, Rockville, MD) in the form of hybridoma supernatant diluted 1:4 in complete medium, for 30 min at 4°C. Alternatively, Thy-11 EC were removed by incubation in purified monoclonal anti-Thy-1.2 antibody (T-24-31.7; Dennert et al., 1980) diluted 1:500 in complete medium. The EC were washed and incubated at 37°C for 35 min in low-toxicity rabbit complement (Cedarlane, Hornby, Canada) diluted 1:30 in PBS containing 5% fetal calf serum (FCS) (CSL, Parkville, Australia). Dead cells were removed by treatment with 0.05% trypsin and 300 U/ml DNase in HEPES-BSS for 10 min at 37°C, and then washed. Depletion of class II1 or Thy-11 cells was confirmed by immunofluorescence. Antigen pulsing of EC and immunization protocol EC were washed twice with HEPES-BSS containing 10% NCS, and once with HEPES-BSS prior to exposure to TE at a concentration of 1,000 µg/ml in HEPES-BSS, for 30 min at 37°C. They were then washed as before to remove any remaining TE, and resuspended for immunization in HEPES-BSS. Mice were immunized with 104 viable TE-pulsed EC by s.c. injection into the dorsal trunk. A sample of the EC pulsed with TE were killed by freeze-thawing for use as nonviable controls; microscopic examination ensured that no intact cells remained. Ten days following immunization with TE-pulsed EC, mice were injected with 105 viable UV-13-1 cells in PBS in the dorsal pinna of one ear; this was a dose of tumor FIGURE 1 – Antitumor protective immunity is dependent on immunization with TE-pulsed EC and a low dose of tumor cells. EC cultured in 10 U/ml GM-CSF for 2 hr were pulsed with TE and then injected s.c. into 2 groups of mice (104 EC/mouse). One group was injected with 105 UV-13-1 tumor cells into the ear 10 days later (group a) and the DTH induced by EC measured. Another group, given only tumor cells, served as control (group c). (a) DTH response (mean ear swelling response 6 SEM) was measured 24 hr later. These groups were then inoculated with 2 3 106 UV-13-1 cells i.d. into each flank 7 days later. Mice immunized with TE-pulsed EC only (group b) and naive mice (group d) were inoculated with tumor cells 10 days after immunization with EC. (b) Tumor growth (mean tumor diameter) was measured over the 28 days following tumor inoculation (p , 0.001, least significant difference 5 2.28). (c) The incidence of tumors (% mice per group with tumors) was determined at each time point. All mice were compared with the naive control group. *p , 0.05; n 5 7/group. 100 CAVANAGH ET AL. cells insufficient to form a tumor (subsequently referred to as ‘‘tumor cell immunization’’). Assessment of delayed-type hypersensitivity DTH responses induced by TE-pulsed-EC were determined 24 hr after tumor-cell immunization as the difference in thickness between the injected and uninjected ears, measured using an engineer’s micrometer. A control group of mice which received tumor-cell immunization without EC was included in each experiment. Assessment of protective antitumor immunity Protective antitumor immunity was assessed by both incidence and growth of subsequently inoculated tumor cells. Mice immunized with TE-pulsed EC plus tumor cells, or those given the tumor-cell immunization only, received an intradermal inoculation of 2 3 106 UV-13-1 live tumor cells in 50 µl PBS into each flank between 17–25 days after exposure to TE-pulsed EC. The development of tumors was assessed by manual palpitation 7 days after inoculation of tumors, and then every 3–7 days until the end of the experiment (4–5 weeks). In those mice which developed tumors, tumor diameter was measured with vernier calipers. In all experiments, a naive group of unimmunized mice was included, which received tumor inoculation at the same time as immunized mice. In vivo depletion of T-cell subsets Mice were depleted of T-cell subsets according to the protocol described by Cobbold et al. (1984). Briefly, mice were injected intraperitoneally with the optimal concentration of purified antibody (Ab), previously determined to result in maximal depletion of T-cell subsets, on 3 consecutive days, and then at weekly intervals throughout the experiment. Antibodies included were anti-CD4 (GK1.5; ATCC), and anti-CD8 (YTS169.4; Cobbold et al., 1984) used at 1.0 and 4.0 mg protein per injection, respectively. Anti-CD4 Ab removed approximately 80% of splenic CD41 T cells, and anti-CD8 Ab removed approximately 75% of CD81 T cells. The anti-mycobacteria Ab (L22, 1.0 mg protein per injection; a kind gift from Dr. W. Britton, Centenary Institute of Cancer Medicine and Cell Biology, Camperdown, Australia), used as a control, did not alter either CD8 or CD4 T-cell subpopulations. Statistical evaluation DTH results were analyzed using the unpaired Student’s t-test, the results from all groups being compared to the control group within each experiment. Significance was determined at 95% confidence limits (p , 0.05). Tumor growth measurements were taken every 3–7 days after tumor inoculation until the end of the experiment, and the average size of tumors per mouse at each time point over the entire length of the experiment were included in the statistical analyses. This was done by multivariate analysis of variance (ANOVA), taking into account all individual tumor measurements throughout the duration of the experiment. Results are presented as mean tumor diameter of the 2 tumors possible at each time point. Fisher’s protected least significant difference was used to test for significant differences between groups within the analysis. As the incidence data indicate the number of mice without tumors, any tumors which had been rejected were excluded from the growth curve. The incidence of tumors which was determined throughout the experiment was analyzed using the Mann-Whitney U test for nonparametric analysis. Significance was determined at 95% (p , 0.05) for both tumor growth and incidence. FIGURE 2 – Live EC induce protective antitumor immunity. EC were cultured with (groups a and c) or without (groups b and d) 10 U/ml GM-CSF for 2 hr prior to pulsing with TE. Viable EC were injected s.c. immediately (groups a and b); a sample of EC was killed by freeze-thawing before injection (groups c and d). These mice, and a group which did not receive EC (group e), were reimmunized with 105 tumor cells into the ear 10 days after immunization with EC. (a) DTH (mean ear swelling response 6 SEM) elicited by tumor cells was measured 24 hr later. All mice were compared with control group e. All groups of mice, plus naive, unimmunized group f, were inoculated with 2 3 106 UV-13-1 tumor cells i.d. into each flank 7 days later. (b) Tumor growth (mean tumor diameter) was measured throughout the duration of the experiment (p , 0.01, least significant difference 5 1.97). (c) The incidence (% mice per group with tumors) of tumors was determined at each time point. All groups were compared with the naive, unimmunized control mice (group f). *p , 0.05; n 5 6/group. ANTITUMOR IMMUNITY AND DENDRITIC EPIDERMAL T CELLS RESULTS Protective anti-tumor immunity is dependent upon immunization with both TE-pulsed EC and tumor cells Two groups of mice (a and b) were immunized with 104 EC pulsed in vitro with UV-13-1 TE. One group (a) was given 105 viable UV-13-1 tumor cells into the pinna of one ear 10 days later. This and a tumor-cell-immunized control group (c) were inoculated with 2 3 106 UV-13-1 tumor cells s.c. into each flank after 7 days. The other EC-immunized group (b) and a naive control group (d) 101 were also inoculated with UV-13-1 tumor cells 10 days after immunization with EC. DTH was assessed 24 hr after immunization with tumor cells. Immunization of mice with TE-pulsed EC induced a DTH response significantly greater than that of the control (Fig. 1A). Immunization with TE-pulsed EC and a low dose of tumor cells induced significant protective immunity against subsequent inoculation of UV-13-1 tumor cells, illustrated by a significantly reduced tumor growth rate compared to those tumors growing in naive mice (Fig. 1B). Protection was dependent upon immunization with both TE-pulsed EC and tumor cells, as neither EC immunized mice (b) nor mice given only a low dose of tumor cells (c) demonstrated significantly reduced tumor growth compared with naive mice (d). Immunization with TE-pulsed EC and tumor cells (a) also induced significant tumor rejection compared to naive control mice (Fig. 1C). However neither cell population alone affected tumor incidence. Live EC induce protective antitumor immunity Prior to pulsing with TE, EC were cultured for 2 hr either with (groups a and c) or without (groups b and d) 10 U/ml GM-CSF. Mice were immunized with these viable EC (a and b) or with EC killed by freeze-thawing (c and d). These 4 groups of mice, and a further group (e), were injected in the ear with 105 live UV-13-1 tumor cells 10 days later. When DTH was measured 24 hr later, only viable EC cultured with GM-CSF induced a significant response, whereas nonviable EC, or EC cultured without GM-CSF, did not (Fig. 2A). These groups, plus a naive control group (f), were then inoculated with 2 3 106 tumor cells 7 days later. Mice immunized with viable EC cultured with GM-CSF and tumor cells (a) rejected all tumors by 3 weeks (Fig. 2C). Immunization with viable EC cultured without GM-CSF and tumor cells (b) also exerted significant protective effects. Tumor growth (Fig. 2B) was significantly less than the growth of tumors in naive mice (f), and tumor incidence was significantly reduced to 20% by 4 weeks. Immunization with nonviable EC (c and d) did not alter tumor growth or incidence, nor did tumor cells in the absence of EC immunization (e). Thus, viable EC were required for the induction of protective immunity. Using multiple immunizations, however, we found that viable EC were not required for the induction of protective immunity. Viable EC were necessary for the induction of DTH, as this aspect of immunity was totally dependent on EC viability (data not shown). Culture of EC with GM-CSF was necessary for induction of DTH. However, GM-CSF was not required for protective antitumor immunity, although it did slightly enhance antitumor immunity in some experiments. In all subsequent experiments presented, EC were cultured for 2 hr with 10 U/ml GM-CSF prior to pulsing with TE. Protective antitumor immunity is long-lived To investigate whether the antitumor immunity induced by TE-pulsed EC is long-lived, mice were immunized with either viable or nonviable EC pulsed with TE, followed by reimmunization with 105 tumor cells in the ear, as in previous experiments. Once again, viable but not killed EC induced a significant antitumor DTH (Fig. 3A). The mice then received 2 3 106 tumor cells, approximately 5 months (160 days) after immunization. FIGURE 3 – Antitumor immunity induced by immunization with EC is long-lived. Mice were immunized s.c. with either 104 viable (group a) or killed (group b) EC pulsed with TE. These mice and an untreated group were given 105 UV-13-1 tumor cells into the ear 10 days later. (a) DTH response (mean ear swelling response 6 SEM) was assessed at 24 hr. Immunized mice were compared with control group c. All immunized, and a group of naive, unimmunized mice (group d), were inoculated with 2 3 106 tumor cells i.d. in each flank 160 days after EC immunization. (b) Tumor growth (mean tumor diameter) was measured throughout the experiment ( p , 0.01, least significant difference 5 1.15). (c) Tumor incidence (% mice per group with tumors) was determined at each time point. EC-immunized mice were compared with naive control group d. *p , 0.05; n 5 7/group. 102 CAVANAGH ET AL. Compared with naive mice, the tumors in mice immunized with viable EC and tumor cells displayed a reduction in both growth and incidence (Fig. 3B, C). On the other hand, immunization with killed EC and tumor cells did not affect growth, although it did lead to reduced incidence. However, the reduction in incidence was not as great as that in mice immunized with viable EC (Fig. 3C). This demonstrates that the protective antitumor immunity induced by viable EC pulsed with TE, followed by a low dose of tumor cells, is long-lived and capable of inducing tumor rejection at least 5 months after immunization. Induction of antitumor immunity by EC subsets Prior to culture with GM-CSF, class II (Ia)1 or Thy-11 EC were depleted by monoclonal antibody-mediated complement lysis. Class II1 EC were depleted from approximately 4% to 1.3%, and Thy-11 EC from approximately 4% to 0.5%, as determined by immunofluorescence. These subset-depleted EC, and unfractionated EC, were pulsed with TE, and used to immunize mice. These and unimmunized mice were injected in the ear with 105 UV-13-1 tumor cells. On measuring ear thicknesses 24 hr later, it was found that the class II1 EC were responsible for induction of DTH against the UV-13-1 tumor, as class II1-depleted EC were unable to induce this response, whereas the groups given unfractionated, or Thy-1depleted EC, had DTH responses significantly higher than the control (Fig. 4A). Hence, class II1 EC were responsible for the induction of DTH against the UV-13-1 tumor, whereas Thy-11 EC were not. Mice immunized with EC and tumor cells, plus groups of naive mice, were injected intradermally with 2 3 106 UV-13-1 tumor cells in each flank. In 3 of 4 repeat experiments, it was found that depletion of class II1 EC (group b; Fig. 4B) had no effect on the incidence of tumors compared with mice immunized with unfractionated EC (group a). The tumor incidence in both these groups was significantly reduced compared with naive control group c. Hence, class II1 EC are not responsible for inducing protective antitumor immunity against the UV-13-1 tumor, although this cell type did induce DTH against the same tumor. However, depletion of Thy-11 EC did affect tumor incidence (Fig. 4C). Mice immunized with unfractionated EC (group d) again showed marked reduction in tumor incidence compared to the naive control mice (group f), whereas Thy-1 depletion (group e) prevented the rejection of tumors. Hence, although uninvolved in the induction of DTH against the UV-13-1 tumor, Thy-11 EC were responsible for inducing the rejection of this tumor in immunized mice. CD81 T cells are required for protective antitumor immunity The role of T lymphocytes in antitumor immunity was investigated by depleting the mice of T-cell subsets by repeated injections of anti-CD4, anti-CD8 or a control antibody at the time of inoculation with 2 3 106 viable UV-13-1 tumor cells. The injections of Ab continued throughout each experiment to ensure adequate depletion of T cells. Treatment with anti-CD8 Ab resulted in a reduction of approximately 71% of splenic CD81 T cells, with no effect on CD41 T cells, and anti-CD4 Ab removed approximately 70% of splenic CD41 T cells without affecting CD8 cells. Treatment with the control Ab did not influence either T-cell subset. The growth of tumors in T-cell-depleted mice was compared with control mice, which were not treated with Ab. In unimmunized mice, depletion of CD81 T cells significantly enhanced tumor growth, whereas depletion of CD41 T cells or use of the control Ab had no effect. On the other hand, there was no difference in the incidence of tumors in any group (data not shown). The next step was to investigate the effects of T-cell subset depletion in mice immunized with TE-pulsed EC given 10 days before injection of 105 tumor cells into the ear. This form of tumor cell immunization elicited the expected antitumor DTH response in all EC-immunized mice (Fig. 5A). Seven days later, the mice were inoculated with 2 3 106 viable tumor cells in the flank, and at the same time treatment with purified antibodies was commenced. In mice injected with the control Ab (group a), tumor growth was significantly reduced compared with that in naive unimmunized FIGURE 4 – Role of class II1 and Thy-11 EC in induction of antitumor immunity. Mice were immunized s.c. with 104 TE-pulsed EC or EC depleted of class II1 or Thy-11 cells by Ab-mediated complement lysis. These mice and an unimmunized group were given 105 tumor cells into the ear 10 days later, and DTH were assessed at 24 hr. (a) DTH (mean ear swelling response 6 SEM) responses induced by immunization were compared with control mice. Mice immunized with unfractionated EC (groups a and d), class II-depleted (group b), or Thy-1-depleted (group e) EC, together with naive, unimmunized mice (groups c and f), were inoculated with 2 3 106 tumor cells i.d. in each flank 7 days later. (b,c) Tumor incidence (% mice per group with tumors) was determined at each time point. Immunized mice were compared with naive, unimmunized mice. *p , 0.05; n 5 7/group. ANTITUMOR IMMUNITY AND DENDRITIC EPIDERMAL T CELLS mice (group d; Fig. 5B). This effect persisted following depletion of CD41 cells (group c). However, tumor growth in mice depleted of CD81 lymphocytes (group b) was comparable to that of the naive group. The incidence of tumors in EC-immunized mice given control Ab was significantly reduced compared to the naive control group, as was the incidence in CD41 T-cell-depleted mice (Fig. 5C). By contrast, depletion of CD81 T cells completely abolished protective immunity induced by immunization with EC and tumor cells. In other words, CD81 but not CD41 effector cells are required for protective immunity in this system. 103 DISCUSSION The studies reported here made use of a model in which immunization with tumor antigen-pulsed EC was followed by injection of a low number of viable tumor cells in an attempt to simulate induction of antitumor immunity during the early stages of a developing tumor. The protective effect of this immunization protocol was then determined by challenging the mice with a sufficient number of tumor cells to form a detectable tumor. The results pointed to involvement of distinct EC populations in induction of DTH and protective immunity, with gd1 DETC playing a role in protection, and with MHC class II1 LC in DTH. In other words, 2 different epidermal cell subpopulations were responsible for the 2 different types of antitumor immunity observed. Induction of protective antitumor immunity depended on both DETC and live tumor cells. Neither immunization with a single injection of TE pulsed-EC nor a low dose of tumor cells alone initiated a protective response. Moreover, a second injection of EC pulsed with tumor extract failed to substitute for viable tumor cells (data not shown). Thus, induction of protective immunity in this system appears to be dependent on 2 separate and different steps, the first mediated by DETC and the second by tumor cells per se. The part played by DETC in cutaneous immunity in vivo is not clearly understood. Here we have presented in vivo evidence of a role for DETC in protective antitumor immunity. DETC cell lines have been shown to influence the growth of tumor cells in vitro, by directly lysing a range of tumor cell lines (Kaminski et al., 1993). Based on this evidence it has been postulated that DETC may play a role in host resistance to neoplastic cells. An in vivo study has shown that coinjection of cultured DETC and K1735 melanoma cells abrogates normal growth of these tumors (Love-Schimenti and Kripke, 1994b), although if the DETC and tumor cells were injected simultaneously at different sites there was no inhibitory effect on tumor growth, indicating a direct effect of the DETC on tumor cells. By contrast, in our experiments the initial subcutaneous injection of EC, containing DETC, was followed by a low dose of tumor cells given intradermally at a remote site 10 days later. Thus, DETC do not appear to be functioning in a similar way in our system. CD81 T lymphocytes were involved in rejection of the UV-13-1 tumor in EC-immunized mice. Depletion of CD81 T cells in unimmunized mice also led to an accelerated growth rate, indicating that there was an immune response generated by the tumor itself which limited growth without inducing tumor rejection. These results are in agreement with previous studies demonstrating a role for CD81 T cells in limiting the growth of UV-induced tumors (Fortner and Kripke, 1977). On the other hand, CD81 T cells are not present in DETC (Bergstresser et al., 1983), which means that the protective effect of DETC may be mediated by activation of CD81 T cells. Activation of CD81 T cells depends on presentation of antigens in association with MHC class I, for example by bone-marrow-derived, nonprofessional antigen- FIGURE 5 – Antitumor immunity induced by TE-pulsed EC and tumor cells in mice depleted of T-cell subsets. Mice were immunized s.c. with EC pulsed with TE, and then these and another group were given 105 tumor cells into the ear 10 days later. (a) DTH response (mean ear swelling response 6 SEM) was measured 24 hr later, and EC-immunized mice were compared with mice immunized with tumor cells only. These mice, and naive group d, were inoculated with 2 3 106 tumor cells i.d. in each flank 7 days later. At the same time, the EC-immunized mice were divided into 3 groups, one being injected repeatedly with anti-mycobacteria Ab (group a), the second injected with anti-CD8 Ab (group b), and the third with anti-CD4 Ab (group c). (b) Tumor growth (mean tumor diameter) was measured over the next 35 days (p , 0.001, least significant difference 5 2.30). (c) Tumor incidence (% mice per group with tumors) was determined at each time point. All EC-immunized groups were compared with naive, unimmunized control mice (group d). *p , 0.05; n 5 7/group. 104 CAVANAGH ET AL. presenting cells in the case of B16 melanoma (Huang et al., 1994). This implies that tumor antigens can be reprocessed by the host’s APC after immunization. However, this does not appear to apply to our model, since neither injection of DETC (tumor extract pulsedEC), nor a low dose of tumor cells alone, was sufficient to induce protective immunity. CD41 T lymphocytes may also be involved in the immunity induced by DETC. When CD41 T cells were depleted in the immunized animals, the immune response against the transplanted tumor cells was enhanced, as all tumors were rejected by 14 days. Although we did not observe such an absolute response in 3 other similar experiments, a subsequent reduction in tumor growth still occurred. Hence the data suggest a potentially suppressive role for CD41 T cells in antitumor immunity generated by DETC. Moreover, in vivo depletion of CD41 T cells caused regression of a tumor which required CD81 cytolytic T lymphocytes for rejection (Koeppen et al., 1993). By contrast, Kosugi et al. (1987) have shown that depletion of CD41 T cells diminished the CD81 T-cell-dependent response to X5563 plasmacytoma, an effect which could be reversed by addition of T-cell-derived cytokines. In our study CD41 T cells were not depleted until mice were inoculated with tumor cells to assess tumor growth; thus, a role for CD41 cells in providing help to CD81 effector cells early in the antitumor immune response could not be excluded. Indeed, they may well be the explanation for the result observed by Kosugi et al. (1987). Protective antitumor immunity generated by immunization with DETC followed by tumor cells is still as effective 160 days after the immunization as at 7 days, demonstrating that the protective effect of DETC is long-lived. Previous studies using EC or splenic APC to induce protective immunity have not investigated the longevity of the antitumor immune response (Grabbe et al., 1991; Flamand et al., 1994; Shimizu et al., 1989). Immunization against the UVinduced melanoma K1735, using tumor cells transfected with the costimulatory molecule B7, revealed a discrete window of responsiveness following immunization, as protective immunity was observed only when tumor cells were transplanted 25 days after immunization, and not at 5 or 90 days (Townsend et al., 1994). However, when the same immunization protocol was used for the non-UV-induced EL-4 thymoma, protective immunity was observed up to 92 days postimmunization, the latest time tested in that study. It is possible that in the case of our immunization protocol, the low dose of tumor cells used for secondary immunization persisted without developing into a tumor, thus providing a continual source of tumor antigen. LC have long been known to be integral to induction of cutaneous immunity against a wide variety of antigens (Stingl et al., 1978), and here we have confirmed their role in initiation of antitumor DTH. The DTH response elicited by the exposure to a low dose of tumor cells was dependent on immunization with LC, not DETC, with only a single injection of live LC being required provided they had been precultured with GM-CSF. By contrast, the capacity of DETC to induce protective antitumor immunity was unaffected by culture with GM-CSF, consistent with the findings that protection and DTH were initiated by 2 different cell types in this model. LC have been shown to induce protective immunity against the chemically-induced S1509a fibrosarcoma (Grabbe et al., 1991), providing valuable information on the range of mechanisms by which protective antitumor immunity can be induced. The EC used in this study were depleted of DETC, as the presence of these cells prevented induction of protective immunity, possibly by exerting a suppressive effect on the antitumor response. The reasons for the difference in antitumor mechanisms in this, and in our system, may be due to the nature of the tumors used. Thus, tumors induced by UV radiation express unique antigens and do not lead to crossprotective immunity against other tumors, irrespective of their origin (Kripke, 1974), whereas some non-UV-induced tumors may induce crossreactive responses (Prehn and Main, 1957). In addition, the repertoire of cytokines produced by each tumor, or by the host in response to a developing tumor, may influence the mechanism responsible for immunity. Different cytokines have been shown to influence tumor growth in distinct ways, and can broadly be divided into those which are tumor-inhibitory, or tumor-promoting (reviewed by Schreiber, 1993). The UV-13-1 tumor used in these studies produces high levels of GM-CSF but low levels of TNFa and no detectable IL-3 compared with other UV-induced murine skin tumor lines which we have examined (D. Rubel and G. Halliday, personal communication). Hence the nature of immune responses, and the mechanisms by which immunity is induced against tumors, can be affected by the tumor itself. It is therefore likely that a variety of immunological mechanisms can be activated by different EC populations, and that the tumor itself can play a role in controlling the development of the responses involved in protective anti-tumor immunity. ACKNOWLEDGEMENTS We thank Dr. G. Berry, Department of Public Health, University of Sydney, for statistical advice, and Mr. R. Sluyter for technical assistance and helpful discussions. The support of the National Health and Medical Research Council, and of the University of Sydney Cancer Research Fund, is also gratefully acknowledged. REFERENCES BERGSTRESSER, P.R., TIGELAAR, R.E., DEES, J.H. and STREILEIN, J.W., Thy-1 antigen-bearing dendritic cells populate murine epidermis. J. Invest. Dermatol., 81, 286–288 (1983). COBBOLD, S.P., JAYASURIYA, A., NASH, A., PROSPERO, T.D. and WALDMANN, H., Therapy with monoclonal antibodies by elimination to T-cell subsets in vivo. Nature (Lond.), 312, 548–551 (1984). DENNERT, F., HYMAN, R., LESLEY, J. and TROWBRIDGE, I.S., Effects of cytotoxic monoclonal antibody specific for T200 glycoprotein on functional lymphoid cell populations. Cell Immunol., 53, 350–364 (1980). DUBIN, N., PASTERNACK, B.S. and MOSESON, M., Simultaneous assessment of risk factors for malignant melanoma and non-melanoma skin lesions with emphasis on sun exposure and related variables. Int. J. Epidemiol., 19, 811–819 (1990). ELMETS, C.A., ZEPTER, K. and HAFFNER, A.C., The role of cutaneous immunity in skin cancer. In: H. Mukhtar (ed.), Skin cancer: mechanisms and human relevance, pp. 223–236, CRC Press, Boca Raton (1995). FLAMAND, V., SORNASSE, T., THIELEMANS, K., DEMANET, C., BAKKUS, H., BAZIN, H., TIELEMANS, F., LEO, O., URBAIN, J. and MOSER, M., Murine dendritic cells pulsed in vitro with tumor antigen induce tumor resistance in vivo. Europ. J. Immunol., 24, 605–610 (1994). FORTNER, G.W. and KRIPKE, M.L., In vitro reactivity of splenic lymphocytes from normal and UV-irradiated mice against syngeneic UV-induced tumors. J. Immunol., 118, 1483–1487 (1977). GRABBE, S., BRUVERS, S., GALLO, R.L., KNISELY, T.L., NAZARENO, R. and GRANSTEIN, R.D., Tumor antigen presentation by murine epidermal cells. J. Immunol., 146, 3656–3661 (1991). HUANG, A.Y.C., GOLUMBEK, P., AHMADZADEH, M., JAFFEE, E., PARDOLL, D. and LEVITSKY, H., Role of bone marrow-derived cells in presenting MHC class I-restricted tumor antigens. Science, 264, 961–965 (1994). KAMINSKI, M.J., CRUZ, P.D., BERGSTRESSER, P.R. and TAKASHIMA, A., Killing of skin-derived tumor cells by mouse dendritic epidermal T-cells. Cancer Res., 53, 4014–4019 (1993). KOEPPEN, H.K.W., SINGH, S., STAUSS, H.J., PARK, B.H., ROWLEY, D.A. and SCHREIBER, H., CD4-positive and B lymphocytes in transplantation immunity. I. Promotion of tumor allograft rejection through elimination of CD4-positive lymphocytes. Transplantation, 55, 1349–1355 (1993). KOSUGI, A., YOSHIOKA, T., SUDA, T., SANO, H., TAKAHAMA, Y., FUJIWARA, H. and HAMAOKA, T., The activation of L3T41 helper cells assisting the generation of anti-tumor Lyt-21 cytotoxic T lymphocytes: requirement of Ia-positive antigen-presenting cells for processing and presentation of tumour antigens. J. Leukocyte Biol., 42, 632–641 (1987). ANTITUMOR IMMUNITY AND DENDRITIC EPIDERMAL T CELLS KRIPKE, M.L., Antigenicity of murine skin tumors induced by ultraviolet light. J. nat. Cancer Inst., 53, 1333–1336 (1974). KUZIEL, W.A., TAKASHIMA, A., BONYHADI, M., BERGSTRESSER, P.R., ALLISON, J.P., TIGELAAR, R.E. and TUCKER, P.W., Regulation of T-cell receptor g-chain RNA expression in murine Thy-11 dendritic epidermal cells. Nature (Lond.), 328, 263–266 (1987). LOVE-SCHIMENTI, C.D. and KRIPKE, M.L., Dendritic epidermal T cells inhibit T cell proliferation and may induce tolerance by cytotoxicity. J. Immunol., 153, 3450–3456 (1994a). LOVE-SCHIMENTI, C.D. and KRIPKE, M.L., Inhibitory effect of a dendritic epidermal T cell line on K1735 melanoma cells in vivo and in vitro. J. Leukocyte Biol., 55, 379–384 (1994b). PREHN, R.T. and MAIN, J.M., Immunity to methylcholanthrene-induced sarcomas. J. nat. Cancer Inst., 18, 769–778 (1957). SCHREIBER, H., Tumor immunology. In: W.E. Paul (ed.), Fundamental Immunology, pp. 1143–1178, Raven Press, New York (1993). SHIMIZU, J., SUDA, T., YOSHIOKA, T., KOSUGI, A., FUJIWARA, H. and HAMAOKA, T., Induction of tumor-specific in vivo protective immunity by immunization with tumor antigen-pulsed antigen-presenting cells. J. Immunol., 142, 1053–1059 (1989). 105 STEINMAN, R.M., WITMER-PACK, M. and INABA, K., Dendritic cells: antigen presentation, accessory function and clinical relevance. In: E.W.A. Kamperdijk (ed.), Dendritic cells in fundamental and clinical immunology, pp. 1–9, Plenum Press, New York (1993). STINGL, G., KATZ, S.I., CLEMENT, L., GREEN, I. and SHEVACH, E.M., Immunologic functions of Ia-bearing epidermal Langerhans cells. J. Immunol., 121, 2005–2013 (1978). SULLIVAN, S., BERGSTRESSER, P.R., TIGELAAR, R.E. and STREILEIN, J.W., FACS purifaction of bone marrow-derived epidermal populations in mice: Langerhans cells and Thy-11 dendritic cells. J. invest. Dermatol., 84, 491–495 (1985). SULLIVAN, S., BERGSTRESSER, P.R., TIGELAAR, R.E. and STREILEIN, J.W., Induction and regulation of contact hypersensitivity by resident, bone marrow-derived, dendritic epidermal cells: Langerhans cells and Thy-11 epidermal cells. J. Immunol., 137, 2460–2467 (1986). TOWNSEND, S.E., SU, F.W., ATHERTON, J.M. and ALLISON, J.P., Specificity and longevity of antitumor immune responses induced by B7-transfected tumors. Cancer Res., 54, 6477–6483 (1994). WELSH, E.A., LOVE-SCHIMENTI, C. and KRIPKE, M.L., Studies on the mechanism of immunologic tolerance induction by murine dendritic epidermal Thy-11 cell lines. J. Leukocyte Biol., 52, 425–432 (1992).