Reduced Ultraviolet-Induced Carcinogenesis in Mice with a Functional Disruption in B7-Mediated Costimulation

Mice

CTLA-4 Tg mice were generated as described (24). The murine CTLA-4/IgG₂aFc hybrid gene was cloned into a human keratin 14 (K14) promoter cassette to generate the Tg construct that was microinjected into C3H/HeN background oocytes. Founder mice expressed the soluble murine CTLA-4Ig predominantly in the skin and also had significant CTLA-4Ig serum concentrations (10–30 μg/ml). C3H/HeN wild-type mice (H-2^k) were used as controls and were purchased, along with BALB/c (H-2^d) and C57BL/6 (H-2^b) mice from Harlan (Borchen, Germany). Mice were kept under conventional housing conditions and were utilized in the experiments according to institutional guidelines.

UVR, tumor induction, and histology

Within the solar spectrum, the UVB range (290–320 nm) is responsible for carcinogenesis and immunosuppression. Therefore, a bank of four Philips Ultraviolet-B TL40W/12 sunlamps (Philips, Hamburg, Germany) with an emission spectrum from 280 to 350 nm and a peak at 306 nm were used for irradiation. These lamps deliver an average dose of 8 W/m² as measured with an IL-1700 UV detector and an SED 24 (#3124) filter (both from International Light, Newburyport, MA). The mice were placed on a shelf 20 cm below the light bulbs for irradiation. The cage order was systematically rotated before each treatment to compensate for uneven lamp output along the shelf as described before (6, 7, 25, 26). The mice, 20 for each group (10 males and 10 females), were shaved with electric clippers on the entire dorsum once per week. Beginning at 10 wk of age, mice were irradiated three times per week with 2.5 kJ/m² for 4 wk, with 5 kJ/m² for 4 wk and, then with 10 kJ/m² for 6 mo. Afterward, all mice were observed at weekly intervals for tumor development for an additional 4 mo. The location and growth of each tumor exceeding 2 mm in diameter was recorded. Excision biopsies from all tumors were fixed in paraformaldehyde and embedded in parablast. Sections were stained with hematoxylin and eosin, and documented by a video-computer-assisted digital image-processing technique (DISKUS version 3.99 for Windows 95, Hilgers, Königswinter, Germany).

Modulation of the induction of delayed-type hypersensitivity responses by UVR

A commonly used high-dose UVR protocol was employed, as described elsewhere (12, 27), in which mice were exposed to 30 kJ/m² UVR on the shaved back and immunized to alloantigens 5 days later by s.c. injection of 1 × 10⁸ allogeneic (BALB/c, C57BL/6) nucleated spleen cells at the nonirradiated abdominal site. These mice were challenged 5 days later by injection of 1 × 10⁷ allogeneic (BALB/c, C57BL/6) spleen cells into one hind footpad. Footpad swelling was assessed at 24 h with a micrometer (Mitutoyo, Tokyo, Japan) as a measure of delayed-type hypersensitivity (DTH) responses. Groups of control mice were either irradiated but not immunized before challenge, or only challenged without prior immunization.

Generation and culture of bone-marrow derived dendritic cells

Bone marrow-derived dendritic cells (BmDC) were generated by culture of bone marrow (BM) cells in the presence of GM-CSF and IL-4, as described by Inaba et al. (28) with modifications. Briefly, BM was collected from tibias and femurs of BALB/c mice using PBS and a syringe with a 25-gauge needle, and suspended by vigorous pipetting. Erythrocytes were lysed by incubating cells in lysing buffer (Ortho, Neckargmünd, Germany) for 2 min. Remaining cells were passed through nylon mesh to remove small pieces of bone and debris. The cells were washed twice with cold PBS, resuspended in RPMI 1640 medium (5% FCS, 2 mM l-glutamine, 0.1 mM essential and nonessential amino acids, 50 μM 2-mercaptoethanol, 20 μg/ml gentamicin), and cultured in petri dishes (Becton Dickinson, Heidelberg, Germany) at a density of 0.5 × 10⁶ cells/ml for 4 h. Nonadherent cells were collected, then 1 × 10⁶ cells were placed in 24-well plates (Becton Dickinson) and adjusted to 1 ml with BM medium supplemented with 150 U/ml GM-CSF (R&D Systems, Wiesbaden, Germany) and 75 U/ml IL-4 (PharMingen, Hamburg, Germany). After 2 days of incubation (37°C, 5% CO₂), 600 μl of medium was removed and the same volume of fresh BM medium containing 150 U/ml GM-CSF and 75 U/ml IL-4 was added. Cells were incubated for an additional 3 days, and nonadherent cells were harvested by pipetting and subsequently subcultured in 6-well plates in medium containing 150 U/ml GM-CSF and 75 U/ml IL-4. After 2 days of incubation, more than 70% of the nonadherent cells in culture had acquired typical dendritic morphology. These cells were harvested and used as a source of BmDC in subsequent experiments.

BmDC-lymphocyte alloreaction

BmDC were incubated in RPMI 1640 supplemented with 10% heat-inactivated FCS (PAA, Linz, Austria), 100 U/ml penicillin, 100 mg/ml streptomycin, 0.1 mM essential and nonessential amino acids, 2 mM l-glutamine, 1 mM sodium pyruvate, and 0.01 M HEPES buffer (complete medium); and applied in serial dilutions to 2 × 10⁵ allogeneic T cells from CTLA-4Ig Tg or wild-type mice in 96-well round bottom plates. T cells were obtained from spleen cells of mice by nylon-wool purification. After 4 days, T cell proliferation was measured by adding 1 μCi [³H]thymidine; incorporated [³H]thymidine was quantified, and supernatants were taken for cytokine quantification.

T cell stimulation and cytokine quantification

Nylon-wool-enriched splenic T cells were used for proliferation experiments. Stimulation of T cells was induced with PMA (3 ng/ml) and ionomycin (300 ng/ml) (both from Sigma, Deisenhofen, Germany). Supernatants were harvested at 48 h for cytokine quantification. IL-4 concentrations were measured using the CT.4S cell line as described (29). IFN-γ and IL-10 were quantitated using a commercially available ELISA (R&D Systems). The limits of detection were 10 pg/ml IFN-γ and 10 pg/ml IL-10.

For quantification of serum IL-10 concentrations induced by UV exposure, CTLA-4Ig Tg and C3H/HeN mice were shaved on the back and exposed to 30 kJ/m² UVR. Serum was harvested at 0, 1, 2, 3, 4, and 5 days and IL-10 content was determined using an IL-10 ELISA (R&D Systems). Control animals were shaved but not exposed to UV.

Immunohistochemistry

Dorsal skin of CTLA-4Ig Tg and C3H/HeN mice was shaved and exposed to 30 kJ/m² UVR. The skin was biopsied at 0, 1, 2, 3, and 4 days after UV treatment and snap frozen in liquid nitrogen. Cryostat sections were fixed on aminoalkysilane-treated slides (Sigma-Aldrich, Steinheim, Germany) and stained for IL-10 with a rat monoclonal anti-mouse IL-10 IgG1 (clone JES5-2A5), biotinylated anti-rat IgG1 (clone R3-34) (both from PharMingen), peroxidase-labeled streptavidin, and diaminobenzidine substrate. The tissue specimens were counterstained with hematoxylin and examined microscopically.

Evaluation of LC, dendritic epidermal T cells, and sunburn cells

Epidermal sheets were prepared from the ears of CTLA-4Ig Tg and wild-type control mice by separation with EDTA 20 mM (pH 7.3). LC in these epidermal sheets were stained indirectly with a monoclonal anti-I-A Ab (from clone M5/114; American Type Culture Collection, Manassas, VA) and fluorescein-conjugated goat anti-rat IgG (PharMingen). Dendritic epidermal T cells were visualized indirectly by staining with anti-CD3 Ab (from clone 145-2C11; American Type Culture Collection) and fluorescein-conjugated goat anti-rat IgG. Positively stained cells were counted using a net micrometer grid. Five 0.04 mm² sections were counted and averaged on each sample. Samples from three mice of each strain were counted.

For the generation of sunburn cells, CTLA-4Ig Tg and control mice (n = 3) were UVB-irradiated on their shaved backs with 1 kJ/m². Twenty-four hours later skin biopsies were obtained, fixed in buffered formaldehyde, embedded in paraffin, sectioned (7 μm), and stained with hematoxylin and eosin. Sunburn cells were defined as apoptotic cells within the epidermis exhibiting an eosinophilic cytoplasm and a dense nucleus. Sunburn cells were counted per mm length of the epidermis.

Statistical analysis

The method of Kaplan and Meier was used to describe the probability of tumor development in the carcinogenesis study. This is a life table analysis and also takes into account animals that die before developing a tumor. Statistical differences for the development of tumors between the two strains of mice were determined using a logrank test by Peto et al. (30). The differences in tumor latent periods were analyzed by a Mann-Whitney U test. The data from the DTH experiments were analyzed using the Student t test for independent samples.

Reduced incidence of UV-induced skin tumors in CTLA-4Ig Tg mice

Although the importance of the B7/CD28-CTLA-4 signaling pathway in the generation of Ag-specific immune responses is well established, the relevance of this mechanism for tumor-specific immune responses is still controversial. To investigate the role of the B7/CD28-CTLA-4 pathway in the development of UV-induced skin cancer in vivo, CTLA-4Ig Tg and littermate control mice were chronically irradiated with UV and skin tumors were documented over time. CTLA-4Ig Tg mice are healthy, breed well, and show no obvious signs of altered T cell development. Their skin has a normal morphology as well as a normal number of LC and dendritic epidermal T cells (Table I⇓). Additionally, the formation of sunburn cells after UVR was not altered in CTLA-4Ig Tg mice when compared with controls. Sunburn cells are keratinocytes within the epidermis that become apoptotic after UVR. They can be used as an indicator for UV damage of the epidermis after light exposure. However, as seen in Fig. 1⇓, the period of time between the first UVR and the appearance of the first visible skin tumor was significantly different in the two groups. Although both Tg and wild-type mice were susceptible to photocarcinogenesis, UV-induced tumor development was delayed in CTLA-4Ig Tg mice. Furthermore, the incidence of UV-induced skin tumors was significantly reduced in the CTLA-4Ig Tg group; fewer mice developed tumors and the overall number of tumors observed was lower compared to wild-type control mice (p < 0.04). Groups of wild-type (n = 15) and CTLA-4Ig Tg mice (n = 15) that were not irradiated but were otherwise treated identically did not develop any skin tumors. These results suggest that functional disruption of the B7/CD28-CTLA-4 pathway leads to reduced photocarcinogenesis.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Reduced photocarcinogenesis in CTLA-4Ig Tg mice compared with wild-type controls. Rate of UV-induced skin tumor development in CTLA-4Ig Tg and wild-type control mice. Mice were initially treated with 2.5 kJ/m² UVR for 4 wk, then with 5 kJ/m² UVB for 4 wk, and then with 10 kJ/m² UVR for 6 mo. Their dorsal hair was shaved weekly. This experiment was performed once. For both groups n = 20. ∗, p < 0.002 for wild-type vs CTLA-4Ig Tg mice.

The data depicted in Table II⇓ demonstrate that of 20 UV-irradiated wild-type mice, 16 developed a total of 21 skin tumors (4 mice died without a tumor). In the CTLA-4Ig Tg group, 7 mice of 20 developed a total of 8 skin tumors (3 mice died without a tumor). Remarkably, none of the wild-type mice but 50% of the CTLA-4Ig Tg mice remained free of tumors 1 year after the onset of the experiment. A total of 76% of the tumors occurring in wild-type mice and 88% of the tumors in the CTLA-4Ig Tg group were located on the ears and the back which, in accordance with other reports, are the primary sites of UV-induced tumor development in mice (7).

CTLA-4Ig Tg mice develop fewer poorly differentiated UV-induced skin tumors

Histopathological examination revealed that the tumors arose uniformly from the epidermis and showed typical features of squamous cell carcinoma in both the CTLA-4Ig Tg mice and the wild-type controls. The differentiation state of UV-induced skin tumors was assessed using three categories: 1) a carcinoma was interpreted as well-differentiated if regular endo-exophytic proliferation of squamous cells with mild cytological atypia and central orthokeratotic keratinization was noted; 2) a moderately differentiated carcinoma was defined by irregular dermal nests of pleiomorphic epithelia with individual cell keratinization and horn pearl formation; and 3) a poorly differentiated carcinoma was composed of highly mitotic spindle cells with large vesicular, pleiomorphic nuclei and scanty eosinophilic cytoplasms. Based on this classification, 37.5% of the tumors seen in the Tg mice were determined to be well differentiated, 12.5% moderately differentiated, and 50% poorly differentiated squamous cell carcinomas. The wild-type controls developed 19% well differentiated, 14.3% moderately differentiated, and 66.7% poorly differentiated squamous cell carcinomas. Normalized to tumor numbers, this result translated into a 3.5-fold decrease in poorly differentiated carcinomas (n = 4) in Tg mice compared with wild-type controls (n = 14). The proportion of well differentiated, usually less aggressive tumors vs poorly differentiated, more malignant carcinomas was increased in the CTLA-4Ig Tg group, though this difference was less striking. Although there was a trend toward later tumor onset in the CTLA-4Ig Tg group, differences in the latencies for the development of poorly, moderately, or well differentiated skin tumors between wild-type and Tg mice were not statistically significant.

UV-induced immunosuppression is reduced in CTLA-4Ig Tg mice

Besides TCR-MHC I and/or II interactions, costimulation via the B7/CD28-CTLA-4 pathway seems to be important for the induction of optimal immune responses, including antitumor immunity. Thus, it appeared surprising that functional blocking of this pathway by soluble CTLA-4Ig resulted in reduced rather than enhanced UV-induced skin tumor development. Because UVR has been shown to induce systemic immunosuppression, we were interested in investigating the effects of CD28/CTLA-4 blockade on UV-induced suppression of DTH responses to alloantigens. Wild-type and CTLA-4Ig Tg mice were exposed to a single dose of 30 kJ/m² UVR on the shaved back before sensitization by s.c. injection of allogeneic spleen cells. Control groups of mice were sensitized but not UV-exposed. Five days after sensitization, all groups of mice were challenged at one hind footpad by injection of allogeneic spleen cells (H-2^d). Footpad swelling was assessed 24 h later as a measure of DTH response to alloantigens. Wild-type mice that were immunized and challenged developed a significant footpad swelling response (Fig. 2⇓A). Exposure of wild-type mice to UVR before sensitization suppressed the DTH response to alloantigens by 50%, indicating that UVR suppressed the induction of DTH. In contrast, UV-treated CTLA-4Ig Tg mice were still able to mount a significant DTH response to alloantigens, suggesting that these mice were resistant to UV-induced suppression of DTH responses (Fig. 2⇓A). In another experiment CTLA-4Ig Tg mice as well as littermates were immunized to alloantigen (H-2^b) and challenged 5 days later at one hind footpad. The data in Fig. 2⇓B show an enhanced footpad swelling response in the CTLA-4Ig Tg mice compared to controls, suggesting a hyperresponsiveness of the CTLA-4 Tg mice to alloantigens.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

A, Impaired immunosuppressive responses to UVR in CTLA-4Ig Tg mice. Wild-type and CTLA-4Ig Tgs were UV-irradiated as described. Five days later, indicated groups were immunized to alloantigens by s.c. injection of nucleated BALB/c spleen cells. Another 5 days later, all groups of mice were challenged at one hind footpad with nucleated BALB/c spleen cells and footpad swelling was assessed as a measure of DTH. Nonimmunized mice served as negative controls. For all groups n = 5. Error bars, mean ± SEM. The data show one representative experiment of three. ∗, p < 0.02 for immunized and challenged vs UV-treated, immunized, and challenged wild-type mice; and NS for UV-treated, immunized and challenged CTLA-4Ig Tg mice. B, Enhanced DTH responses to alloantigens in CTLA-4Ig Tg mice. Wild-type (H-2^k) and CTLA-4Ig Tg (H-2^k) mice were immunized to alloantigens (H-2^b) by s.c. injection of nucleated C57BL/6 spleen cells. Five days later, all groups of mice were challenged at one hind footpad with nucleated C57BL/6 spleen cells and footpad swelling was assessed as a measure of DTH. Nonimmunized mice served as negative controls. For all groups n = 5. Error bars show the mean ± SEM. The data show one representative experiment of three. ∗, p < 0.04 for immunized and challenged CTLA-4Ig Tg mice vs immunized and challenged wild-type mice.

Impaired Th2-type immune responses in CTLA-4Ig Tg mice

The finding that CTLA-4Ig Tg mice have a predisposition toward Th1 responses might explain impaired photocarcinogenesis for several reasons. First, a competent Th1 response seems to be crucial for the induction of tumor immunity (31, 32, 33). Second, UV light impairs cellular immune reactions by inducing a shift toward Th2-type responses (12). This might also explain why CTLA-4Ig Tg mice are resistant, or at least less susceptible, to UV-induced suppression of DTH, a classical Th1 immune reaction. It has been suggested that UV light skews immunity toward Th2 by inducing the release of IL-4 and/or IL-10, because increased levels of IL-4 and IL-10 were detected in the sera of UV-exposed mice (12, 27, 34). In addition, treatment of animals with anti-IL-4 or anti-IL-10 mAb abrogates UV-induced immunosuppression. To address the issue that the differences in the tumor incidence and in the UV susceptibility of DTH in these Tg animals might be consequences of altered cytokine production, cytokine analysis was performed. Mitogen-stimulated splenic T cells from CTLA-4Ig Tg mice produced significantly less IL-4 compared to their wild-type counterparts (Fig. 3⇓A). In contrast, IFN-γ secretion by mitogen-stimulated T cells from CTLA-4Ig Tg mice was enhanced compared to wild-type T cells (Fig. 3⇓B). In addition, CTLA-4Ig Tg T cells that were stimulated by allogeneic BmDC produced significantly more IFN-γ and less IL-4 compared to wild-type T cells (Fig. 3⇓, C and D). Additionally, CTLA-4Ig Tg T cells demonstrated an enhanced proliferative response to alloantigens compared to controls (Fig. 4⇓). There was no difference in the mitogen- or alloantigen-induced production of IL-10 by splenic T cells from Tg and wild-type mice (data not shown). Because UVR itself is able to induce elevated serum levels of IL-10, we analyzed the serum of irradiated wild-type and Tg mice. No differences in the serum concentrations of IL-10 between the two groups of mice was observed (data not shown). Furthermore, immunohistochemical studies revealed equal staining of IL-10 in the skin of UV-treated Tg and wild-type mice (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Production of IL-4 and IFN-γ by mitogen-stimulated splenic T cells from CTLA-4Ig Tg mice. Various concentrations of splenic T cells were stimulated with PMA and ionomycin for 48 h as indicated. Subsequently, culture supernatants were harvested and assayed for IL-4 (A) and IFN-γ (B) content as described. The data presented show one representative experiment of three. The p value was <0.001 for IL-4 and <0.02 for IFN-γ. In another experiment, splenic T cells were stimulated by various concentrations of BmDC for 4 days in the mixed BmDC lymphocyte alloreaction as indicated. Subsequently, culture supernatants were harvested and assayed for IL-4 (C) and IFN-γ (D) content as described. The p value was <0.003 for IL-4 and <0.04 for IFN-γ.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Enhanced stimulation of CTLA-4Ig Tg T lymphocytes by allogeneic BmDC. BmDC were prepared and cultured as described above and used at various concentrations in the mixed BmDC-lymphocyte reaction. After 4 days the reaction was pulsed with [³H]thymidine and incorporation was measured 18 h later in a cell harvester. Data show one of three representative experiments. ∗, p < 0.008 for CTLA-4Ig Tg T cells vs wild-type T cells.