Carcinogen-Induced Inflammation and Immunosuppression Are Enhanced in Xeroderma Pigmentosum Group A Model Mice Associated with Hyperproduction of Prostaglandin E21

Xeroderma pigmentosum group A (XPA) gene-deficient mice easily developed skin cancers by the application of topical chemical carcinogens as well as by UV irradiation. As certain chemical carcinogens have been shown to be immunosuppressive, we examined the inflammatory and immunosuppressive effects of dimethylbenz(a)anthracene (DMBA) on XPA mice. Compared with wild-type mice, XPA mice showed greater ear swelling and reduction of epidermal Langerhans cells after DMBA application. Topical application of DMBA impaired the induction of contact hypersensitivity, initiated either locally or at distant sites. These DMBA-induced local and systemic immunosuppressions were more greatly enhanced in XPA mice than in wild-type mice. DMBA application induced pronounced production of PGE2, IL-10, and TNF-α in the skin of XPA mice. Treatment with indomethacin, a potent inhibitor of PG biosynthesis, inhibited DMBA-induced inflammation and local immunosuppression. In XPA mice, increased serum IL-10 was detected after DMBA treatment. Excess production of PGE2, TNF-α, and IL-10 after DMBA application may be involved in the enhanced local and systemic immunosuppression in DMBA-treated XPA mice. Susceptibility to DMBA-induced skin tumors in XPA mice may be due to easy impairment of the immune system by DMBA in addition to a defect in the repair of DMBA-DNA adduct. Enhanced immunosuppression by chemical carcinogens as well as the mutagenicity of these mutagens might be associated with the high incidence of internal malignancies seen in XP patients. Moreover, these results supported the hypothesis that persistent DNA damage is a trigger for the production of immunoregulatory cytokines.

X eroderma pigmentosum (XP) 3 is an autosomal recessive disorder clinically characterized by extreme sensitivity to sunlight and a Ͼ1000-fold higher incidence of skin cancer (1,2). XP has a defect in an early step of the nucleotide excision repair of bulky DNA lesions, such as the cyclobutane pyrimidine dimer (CPD) and the pyrimidine-pyrimidone 6-4 photoproduct caused by UV radiation (1)(2)(3). However, the defect in nucleotide excision repair cannot completely explain the clinical futures of XP patients, although investigations at the cellular and molecular levels have been extensively performed. Group A XP (XPA) is the most common form among eight genetic complementation groups of this disease in Japan and shows the most severe clinical symptoms. We have generated XPA gene-deficient mice by gene targeting (4), and the model mice have defects in nucleotide excision repair without any obvious physical abnormalities or pathologic alterations. Similar to humans with XP, these XPA mice easily develop UVB-induced skin cancer (4) and show a strong and long-lasting inflammation after UVB radiation (5).
UVB radiation not only acts as a carcinogen, but also induces specific suppression within the cutaneous immune system (6,7). The immunosuppressive effects of UVB, as well as its mutagenic properties, may contribute to the mechanisms of sunlight-induced skin cancer (6 -8). We have already studied the effects of UVB on the immunologic function of XPA mice. Although contact hypersensitivity (CHS) was induced equally in XPA and wild-type mice, there was a greater impairment of sensitization through UVB-irradiated skin in XPA mice than in wild-type mice. Moreover, irradiation of XPA mice with UVB greatly suppressed CHS systemically, i.e., when a sensitizer was applied to a site that had not been irradiated (5). A high incidence of UVB-induced skin cancer in XPA mice may be mainly due to a defect in the repair of UVBinduced DNA lesions of cutaneous cells. Moreover, the enhancement of UVB-induced local and systemic immunosuppression in XPA mice could be involved in the high incidence of skin cancer in XP. Although not completely delineated, several mechanisms have been implicated to explain these immunosuppressive effects of UVB. UVB directly impairs the activity of APCs in the skin. It also increases the production of cytokines or soluble factors with immunomodulatory activity, including cis urocanic acid (9), TNF-␣ (10), IL-4 and IL-10 (11), and PGE 2 (12). Among them, PGE 2 is also known to be an inflammatory mediator. We have reported that the amount of PGE 2 significantly increased after UVB radiation in XPA mice in contrast to that in wild-type mice (13). This result suggested that the increased PGE 2 production may be involved in the enhancement of inflammation and immunosuppression caused by UVB radiation in XPA mice.
As in the case of UVB radiation, certain chemical carcinogens, such as dimethylbenz(a)anthracene (DMBA), have been shown to be immunosuppressive. The following parameters have been reportedly affected in DMBA-exposed mice: proliferative response to T cell mitogens, unidirectional mixed lymphocyte response, generation of cytotoxic lymphocytes, and NK cell tumor cytolysis (14,15). Application of DMBA to the skin depletes epidermal Langerhans cells (LCs) (16) and suppresses CHS locally, when a sensitizer is applied to DMBA-treated skin (17)(18)(19). Similar to UVB-induced immunosuppression, DMBA-induced local immunosuppression is reported to be associated with the development of Ag-specific suppressor T lymphocytes, which inhibit the induction of both cellular and humoral immune responses when adoptively transferred to naive syngeneic recipients (17,18). In contrast to UVB, topically applied DMBA inhibits cutaneous immunological functions only at the site of application, because when DMBA is applied at one skin site, and animals are immunized to the hapten at another site, a normal CHS response is observed (18). Although these findings suggested that DMBA exerts immunosuppressive effects by reducing LC numbers and/or function, the precise mechanisms by which DMBA induces immunosuppression have not been clarified. XPA mice also showed a high incidence of skin tumors by topical application of DMBA as well as by UVB radiation (4). It is well established that the formation of DMBA adducts with DNA is a necessary precondition for the development of skin cancer and that the level of adducts correlates closely with tumorigenicity (20,21). Because XPA protein is required to repair DMBA-DNA adducts, the high incidence of DMBA-induced skin cancer in XPA mice may be mainly due to a defect in the repair of DMBA-induced DNA lesions of cutaneous cells. We hypothesized that the immunosuppression by DMBA in addition to its mutagenicity may contribute to the development of skin tumors initiated by the application of DMBA to XPA mice. DNA damage caused by UVB radiation has been demonstrated to be one of the primary molecular mechanisms for initiation of UVB-induced immunosuppression. However, it is not known whether other forms of DNA damage, including the DMBA-DNA adduct, also cause immunosuppression, or whether the same immunosuppressive pathway is activated by chemical carcinogens and UVB radiation. We investigated the effects of DMBA on the immunologic function of XPA mice and the mechanisms of DMBA-induced immunosuppression.

Materials and Methods
Animals XPA gene-deficient mice with CBA, C57BL/6, and CD-1 chimeric genetic background (4) were backcrossed with hairless albino mice of the inbred strains Hos/HR-1, which were supplied by Hoshino Experimental Animal Farm (Saitama, Japan), and the resultant hairless XPA Ϫ/Ϫ and XPA ϩ/ϩ mice were used in this study. All mice were female and 8 -10 wk of age at the beginning of each experiment, but within a single experiment, all mice were age-matched. Each experimental panel consisted of 5-14 mice.

Ear swelling response to DMBA application
The ears of mice were treated with 20 l of different concentrations of DMBA in acetone. Ear thickness was measured with a dial thickness gauge (Peacock, Tokyo, Japan) immediately before and 1, 2, 3, 4, and 5 days after application. An ear swelling response was represented by an increment in thickness above the value before application.

ADPase staining
The epidermal sheets were prepared from skin samples using EDTA and were stained with ADP according to the method of Chaker et al. (22). Stained LCs were counted in 30 randomly selected interfollicular fields per sheet with the aid of a calibrated ocular grid.

Sensitization and elicitation of CHS
The mice were sensitized using an epicutaneous application of 25 l of 1% DNFB solution in acetone/olive oil (4/1) on abdominal or ventral skin. CHS was elicited after using an application of 20 l of a 0.2% DNFB solution on the surface of each left ear 6 days after sensitization. Ear thickness was measured before and 24 h after application of the challenge dose, and the difference between the two readings was recorded as ear swelling.

Assay for DMBA-induced immunosuppression of CHS
To assay the induction of local immunosuppression by DMBA, 60 l of a 0.5 or 0.1% DMBA in acetone was applied on the dorsal area of the mice on day 0 unless otherwise stated, and sensitization with DNFB was performed at the DMBA-applied site on day 4. Elicitation at the left ear was performed on day 10.
To assay the induction of systemic immunosuppression by DMBA, 100 l of 0.05 or 0.1% DMBA was applied on the dorsal area of the mice on day 0, and sensitization with DNFB was performed at the abdominal region on day 4. Elicitation at the left ear was performed on day 10.
To determine whether any tolerance was induced, we sensitized all mice again by applying 25 l of 1% DNFB on the nontreated dorsal skin immediately after measurement of ear swelling. The second challenge test was performed on the right ear 6 days after the second sensitization. Suppression of the ear swelling response was calculated as follows: 100 Ϫ (ear swelling of test mice/ear swelling of DMBA-nontreated control mice) ϫ 100%.

Determination of PGE 2 in mouse skin
The amount of PGE 2 in mouse ears at 0, 1, 2, 3, and 4 days after application of 20 l of 0.5% DMBA was determined by enzyme immunoassay, as described previously (23). In brief, both ears of each mouse were cut off and immediately dropped into liquid nitrogen. Four frozen ears were combined into one sample, and then weighed and homogenized with a Polytron homogenizer (Brinkmann, Westbury, NY) in 10 ml of ethanol containing 0.1 ml of 5 N HCl, which was precooled at Ϫ20°C. [ 3 H]PGE 2 (10,000 dpm each) was added to the homogenates as a tracer for estimation of the recovery. After centrifugation, PGE 2 in the ethanol extract was applied to Sep-Pak C 18 cartridges (Waters Associates, Milford, MA) and purified by HPLC. The amount of PGE 2 was measured in duplicate using an enzyme immunoassay kit (Cayman Chemical, Ann Arbor, MI).

Immunohistochemical analysis for IL-10 and TNF-␣
IL-10 and TNF-␣ in the skin were examined by an immunoperoxidase method using monoclonal rat anti-mouse IL-10 (BD PharMingen, San Diego, CA) and monoclonal rat anti-mouse TNF-␣ (BD PharMingen). DMBA-treated mouse skin was removed 2 days after treatment, embedded in an OTC compound, and frozen in liquid nitrogen; 4-m cryostat sections were fixed with 2% paraformaldehyde. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide; slides were incubated with a protein blocking solution (10% normal horse serum and 1% normal goat serum) for 20 min, followed by treatment with rat anti-mouse IL-10 or TNF-␣ mAb overnight at 4°C. The slides were washed with PBS and incubated for 30 min with peroxidase-labeled streptavidin and for 20 min with diaminobenzidine. As a negative control, isotype rat IgG1 (BD PharMingen) was used instead of mAbs.

RT-PCR for IL-10
Whether the application of DMBA induced IL-10 mRNA signals was determined according to Enk and Katz (24) with some modification. At various times after application of DMBA, the mice were killed, and ears were excised, split, and floated dermal side down in 10,000 U of Dispase (Godo Shusei,Tokyo, Japan) in 10 ml of PBS for 2 h at 37°C. Epidermal cell suspensions were prepared, and total epidermal RNA was extracted by the method of Chomczynski and Sacchi (25).
The cDNA was synthesized from 1 g of mRNA in a 20-l volume containing 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 2.5 mM MgCl 2 , 10 mM DTT, 1 mM dNTPs, 0.5 g of oligo(dT), and 2.5 U of SuperScript II RT (Life Technologies Oriental, Tokyo, Japan). The mixture was incubated at 42°C for 50 min and heated at 70°C for 10 min to inactivate the enzyme. The mixture was incubated with 2 U of Escherichia coli RNase H at 37°C for 20 min.
PCR amplification was performed according to Saiki et al. (26). The cycling conditions chosen were 1 min at 94°C, 1 min at 58°C, and 2 min at 72°C, and 30 cycles were used. Primers for IL-10 were purchased from Clontech (Palo Alto, CA). Primer sequences for G3PDH were ACCA CAGTCCATGCCATCAC (5Ј primer) and TCCACCACCCTGTTGCT GTA (3Ј primer). After the final cycle, the temperature was maintained at 72°C for 10 min. PCR products were then electrophoresed though 1% agarose gels in Tris/acetate/EDTA buffer containing ethidium bromide, and the gels were photographed under UV light.

Measurement of TNF-␣ protein
Samples of ear tissue extract for ELISA were prepared as described by Ferguson et al. (27) with some modification. Briefly, at various times after application of DMBA, ears were excised and immediately homogenized with a 5-fold volume of 0.1% Tween 20 in PBS. Samples were then quickly frozen in liquid nitrogen, thawed in a 37°C water bath, sonicated for 15 s, and centrifuged for 5 min at 13,000 ϫ g, and supernatants were used for ELISA (Genzyme, Minneapolis, MN).

Determination of serum IL-10 by ELISA
Peripheral blood was obtained by venipuncture 2 days after the epicutaneous application of 100 l of 1% DMBA and was anticoagulated with heparin. Following centrifugation, serum IL-10 was measured in duplicate using an ELISA kit (BD PharMingen), according to the manufacturer's procedures.

Statistical analysis
Student's t test was used to determine the statistical difference between means, with p Ͻ 0.05 considered significant.

XPA mice, but not wild-type mice, developed significant ear swelling response by DMBA application
A single application of 20 l of 0.5% DMBA resulted in significant ear swelling in XPA mice 2 days after application, and edema was still increasing on day 5 ( Fig. 1). XPA mice developed lesser ear swelling using a 0.1% DMBA application than with 0.5% DMBA, reaching a plateau on day 3. The application of 0.05% DMBA produced no significant ear swelling in XPA mice. In contrast, the wild-type mice did not develop any significant ear swelling even with a 2.5% DMBA application.
Next, we examined the susceptibility of the mice to other primary irritating reagents and/or carcinogens. Croton oil is known to be a tumor promoter and includes 12-O-tetradecanoylphorbol-13acetate, which increases the amount of intracellular oxidative damage. Application of 0.2 and 1% croton oil produced a similar degree of ear swelling in XPA mice and wild-type mice (data not shown). Application of 10 and 20% phenol, which is a primary irritating reagent, also produced ear swelling in XPA mice as well as in wild-type mice (data not shown). These results indicate that XPA mice are highly susceptible to DMBA, but are no more susceptible to these other agents than are normal mice.

XPA mice developed stronger damage to ADPase ϩ LCs after DMBA application
In nontreated skin, approximately the same numbers of ADPase ϩ LCs were found in the epidermal sheet of XPA and wild-type mice (614.4 Ϯ 104.32 and 577.6 Ϯ 155.52/mm 2 , respectively). The time course of change in the number of ADPase ϩ LCs after application of 0.5% DMBA was similar in both types of mice (Fig. 2). The number of ADPase ϩ LCs significantly decreased 2 days after application of 0.5% DMBA, reached a minimum on day 4 and then recovered, but was still significantly lower than the pretreated level on day 9. By day 16, the density of ADPase ϩ LCs in both types of mice recovered and reached a higher level than that before treatment. During the period from days 2-9, the number of ADPase ϩ LCs in XPA mice was smaller than that in wild-type mice. The DMBA application of a lower concentration (0.1%) significantly reduced ADPase ϩ LCs in XPA mice 2 and 4 days after application ( p Ͻ 0.0001 compared with the pretreated level, on days 2 and day 4), but not in the wild-type mice. Morphologically, dendritic cells in untreated XPA mice skin were indistinguishable from those in wild-type mice; both showed intense staining and multiple branched dendrites. The ADPase ϩ LCs remaining after the 0.1% FIGURE 1. XPA mice, but not wild-type mice, developed significant ear swelling in response to DMBA application dependent on the concentration. Twenty microliters of various concentrations of DMBA were applied to the ears of mice, and ear thickness was measured immediately before and then 1, 2, 3, 4, and 5 days after the application. Data are expressed as the mean intensity of ear swelling (Ϯ SD) in seven mice (14 ears) per group. The XPA mice developed an increase in ear swelling after the application of 0.1 and 0.5% DMBA, whereas no significant ear swelling occurred in the wild-type mice, even following the application of 2.5% DMBA. Significant ear swelling was observed in XPA mice after 0.5% DMBA application (p Ͻ 0.025 on day 1; p Ͻ 0.0001 from days 2-5) and after 0.1% DMBA application (p Ͻ 0.005 on day 2; p Ͻ 0,0001 from days 3-5) as compared with the pretreated level. XPA mice developed a significant reduction in the number of ADPase ϩ LCs after DMBA application. Groups of mice received applications of 0.1 or 0.5% DMBA, and skin samples were removed from the treated areas 2, 4, 9, and 16 days after application. Epidermal sheets were prepared from three mice per group and stained for ADPase. The ADPase ϩ LCs were counted in 10 random fields per sheet. Data are expressed as the mean number of LC per square millimeter (Ϯ SD). A statistically significant reduction of LC was induced by 0.1 and 0.5% DMBA applications on XPA mice. In contrast, the number of LC decreased with 0.5% DMBA, but not with 0.1% DMBA, application in wild-type mice. ‫,ء‬ p Ͻ 0.0001 compared with the respective pretreatment level. Statistically significant differences were present between XPA mice and wild-type mice after a 0.5% DMBA application from days 2-9 (p Ͻ 0.0001).
DMBA application were rounded and lacked dendrites in the XPA mice, but those in the wild-type mice had multiple branched dendrites.

DMBA-induced local immunosuppression was enhanced in XPA mice
The immune status of mice sensitized to DNFB through DMBAtreated skin was investigated. Both XPA and wild-type mice developed a strong CHS response when nontreated skin was sensitized with DNFB (Fig. 3a). Sensitization with DNFB on skin that had received a 0.1 or 0.5% DMBA application resulted in almost complete suppression of the CHS response in XPA mice (96 and 94% suppression, respectively). In contrast, less suppression was induced in the wild-type mice with pretreatment using 0.1 or 0.5% DMBA (24 and 36% suppression, respectively). In these mice, the induction of tolerance was confirmed by the second challenge tests, which revealed almost the same reduced responses as those of the first challenge (Fig. 3b).

DMBA-induced systemic immunosuppression was enhanced in XPA mice
Application of 0.1% DMBA on the back 4 days before sensitization with DNFB on the abdominal skin induced a stronger suppression of CHS in the XPA mice than in wild-type mice (82 vs 45% suppression; Fig. 4a). Although 0.05% DMBA induced only slight suppression in wild-type mice (32% suppression), it produced pronounced suppression in XPA mice (59% suppression). In these mice, the induction of tolerance was also confirmed by the second challenge tests. Because the intensity of the second challenge tests was almost the same as that of the first challenge (Fig.  4b), it was thought that systemic suppression induced by DMBA application may be an active tolerance, but not a transient unresponsiveness.
Production of PGE 2 in DMBA-treated skin increased in XPA mice but not in wild-type mice PGE 2 is known to be not only an inflammatory mediator, but also an immunomodulator. To investigate the molecular mechanisms of enhanced DMBA-induced inflammation and immunosuppression in XPA mice, the amounts of PGE 2 in mouse ears were measured by enzyme immunoassay at various times after 0.5% DMBA application (Table I). In nontreated skin, approximately the same amounts of PGE 2 were detected in XPA (0.97 Ϯ 0.02 ng/g tissue) and wild-type mice (2.3 Ϯ 0.88 ng/g tissue). Although the amount of PGE 2 in the skin of the wild-type mice had not changed by 4 days after application of 0.5% DMBA, the amount in the skin of XPA mice had increased at 2 days after application (15.9 Ϯ 6.1 ng/g tissue) and was still increasing on day 4 (53.9 Ϯ 2.2 ng/g tissue).

Treatment of indomethacin after DMBA application inhibited DMBA-induced ear swelling response in XPA mice
To determine whether the DMBA-induced edema was mediated by PGE 2 , the effect of indomethacin (a potent inhibitor of PG biosynthesis) on the DMBA-induced ear swelling response was examined. Twenty microliters of 1% indomethacin was applied to the ears of XPA mice immediately after the application of 0.1% DMBA, and the thickness of the ears was measured (Fig. 5). A slight ear swelling response developed 1 day after application in the mice that only received DMBA (3.67 Ϯ 0.47 ϫ 10 Ϫ3 cm), but there was no observable swelling in the mice that also received indomethacin after DMBA application (0.58 Ϯ 0.76 ϫ 10 Ϫ3 cm; The second challenge test was performed on the right ear 6 days after the second sensitization. Data are expressed as the mean intensity of ear swelling (Ϯ SD) of 5-10 mice/group 24 h after the first challenge test (a) and after the second challenge tests (b). a, Sensitization with DNFB through 0.5 and 0.1% DMBA-treated skin was significantly suppressed in XPA mice. b, In these mice, the induction of tolerance was confirmed by the second challenge test. ‫,ء‬ p Ͻ 0.001 compared with the positive control group (animals sensitized without DMBA application). Statistically significant differences in the intensity of suppression were present between XPA and wild-type mice that had received the same concentration of DMBA treatment (p Ͻ 0.0001 at 0.1% DMBA, p Ͻ 0.005 at 0.5% DMBA).
p Ͻ 0.0005). Although the ear swelling in both groups of mice increased after treatment, reaching maximum swelling 3-4 days after application, indomethacin inhibited DMBA-induced ear swelling by 40 -70% at all of the time points studied.

Treatment with indomethacin after DMBA application inhibited DMBA-induced local immunosuppression in XPA mice
To determine whether PGE 2 is involved in the DMBA-induced immunosuppression, we studied the effect of indomethacin on DMBA-induced local immunosuppression. Immediately after the application of 40 l of 0.1 or 0.05% DMBA, 40 l of 1% indomethacin was applied to the DMBA-treated area, and sensitization with DNFB was performed on the DMBA-and indomethacintreated skin 4 days later. Epicutaneous application of indomethacin on DMBA-treated skin abrogated the suppressive effect of DMBA in both XPA and wild-type mice (Fig. 6). In XPA mice, sensitization with DNFB on skin that had been exposed to 0.1 or 0.05% DMBA resulted in a significantly decreased CHS response (98 and 75.3% suppression, respectively). However, application of 1% indomethacin immediately after exposure to 0.1 or 0.05% DMBA inhibited the degree of suppression (43.2 and 30% suppression, respectively). Similarly, the degree of suppression decreased after treatment with indomethacin in wild-type mice, although less suppression was induced using DMBA in wild-type mice than in XPA mice.

DMBA-induced IL-10 and TNF-␣ production in the epidermis was enhanced in XPA mice
IL-10 and TNF-␣ have been implicated as participants in the immunosuppressive cascade. The ability of DMBA to induce IL-10 and TNF-␣ production in mouse skin in vivo was analyzed by mAb and immunoperoxidase staining of skin sections. Constitutive expression of IL-10 was not detected in the skin of XPA (Fig.  7A) and wild-type mice (data not shown). IL-10 could be detected in the epidermis 2 days after 0.1% DMBA application in XPA mice (Fig. 7B), but not in wild-type mice (Fig. 7C). There was no detectable staining using isotype control rat IgG1 as a negative control in epidermis of XPA mice treated with 0.1% DMBA (Fig.  7D). Similarly, application of 0.1% DMBA induced detectable TNF-␣ production in the epidermis of XPA mice, but not in that of wild-type mice (data not shown).
To confirm that DMBA application induces IL-10 production more easily in XPA mice than in wild-type mice, the effect of DMBA application on IL-10 gene expression in epidermis was examined. We studied mRNA expression for IL-10 at various time points after DMBA application in both types of mice. There was no demonstrable mRNA expression for IL-10 in nontreated skin of  XPA and wild-type mice (Fig. 8A, lane 1, and Fig. 8B, lane 1, respectively). In XPA mice, IL-10 mRNA expression was observed at 9 h after 0.1% DMBA application (Fig. 8A, lane 3), peaked at 12 h (Fig. 8A, lane 5), and disappeared at 15 h (Fig. 8A,  lane 7). Furthermore, strong expression of IL-10 mRNA was induced in XPA mouse skin treated with 0.5% DMBA at 12 h (Fig.   8A, lane 4), the signal was decreased at 15 h (Fig. 8A, lane 6), and disappeared at 24 h (Fig. 8A, lane 8). In contrast, IL-10 gene expression was undetectable in wild-type mice treated with 0.1 and 0.5% DMBA throughout the observation period (Fig. 8B). Next, to confirm that DMBA application induces TNF-␣ production more easily in XPA mice than wild-type mice, production of TNF-␣ at protein level in the DMBA-treated skin of both types of mice. There was no detection of TNF-␣ protein in nontreated skin of XPA and wild-type mice. In the skin of XPA mice, TNF-␣    production at the protein level was detected 24 h after the application of 0.1% DMBA and increased at 48 h (Table II). The level of TNF-␣ production was enhanced by 0.5% DMBA application in XPA mice. However, production of TNF-␣ protein was undetectable in wild-type mice treated with 0.1 and 0.5% DMBA throughout the observation period.

IL-10 release in serum was detectable only in DMBA-treated XPA mice
IL-10 has been reported to play a role in the induction of systemic immunosuppression by alteration of the Ag-presenting property of APC in draining lymph nodes. The release of IL-10 in peripheral blood was examined after DMBA application. IL-10 was not detected in serum constitutively in either XPA or wild-type mice. However, in all XPA mice that received an application of 100 l of 1% DMBA, IL-10 was detected in serum 2 days after application (Table III). In contrast, there was no detectable serum level of IL-10 by ELISA in wild-type mice treated with DMBA. In animals treated with 100 l of 0.5% DMBA 2 days previously, serum IL-10 was detectable in two of five XPA mice (22 and 27 pg/ml, respectively), but not in five wild-type mice. Application of 100 l of 0.1% DMBA did not induce a detectable serum level of IL-10 by ELISA in either type of mouse.

Discussion
Immunological factors play an extremely important role in controlling the growth and metastasis of tumors (28). As well as UVB radiation, certain chemical carcinogens, including polyaromatic hydrocarbons, have been shown to be immunosuppressive. Although much is known about how UVB radiation modulates immunologic responses, relatively little is known about interactions between the immune system and the carcinogenesis pathway caused by chemical carcinogens. Moreover, a complete picture of the molecular steps leading to UV-induced immune suppression is not yet available. Kripke and colleagues proposed UVB-induced DNA damage as a primary molecular target for initiation of immune suppression. We previously demonstrated that UVB-induced local and systemic immunosuppression was enhanced in XPA mice (5). Our results supported the hypothesis proposed by Kripke and colleagues, because XPA mice are different from wild-type mice only in a defect in the repair of UVB-induced DNA lesion. However, it is unclear whether cutaneous DNA damage caused by other than UVB radiation might also induce immunosuppression. DMBA, one of the prototype carcinogenic polyaromatic hydrocarbons, is a chemical carcinogen and causes DNA adduct, and XPA protein is needed for its repair. We have reported that XPA mice showed a high incidence of skin tumors initiated by topical application of DMBA as well as by UVB radiation (4). Based on this background, we investigated the effects of DMBA on inflammation and the immune system in XPA mice.
A single cutaneous application of 0.1 or 0.5% DMBA induced significant ear swelling in XPA mice, but not in wild-type mice. In contrast, the intensities of ear swelling responses after treatment with croton oil or phenol were almost the same in both types of mouse. Croton oil and phenol do not form DNA adduct, or XPA protein is not required to repair DNA damage, if any, caused by these chemicals. Thus, a defect in the repair of DMBA-DNA adduct in XPA mice might be involved in the enhancement of DMBA-induced inflammation. In XPA mice, LCs are impaired at a lower DMBA concentration than in wild-type mice. Moreover, DMBA-induced local immunosuppression was enhanced in XPA mice, but not in wild-type mice. In XPA mice, enhanced reduction of LC number after application of DMBA might be involved in the enhancement of DMBA-induced local immunosuppression. Addition to the cytotoxic effects of DMBA, DMBA-induced inflammation might contribute to the reduction of the number of LCs, because inflammation will prompt LCs to emigrate from skin and to home to regional lymph nodes. Furthermore, the fact that LC numbers in DMBA-treated skin remain low for several days may indicate that DMBA and/or DMBA-induced inflammation inhibits the immigration of LC precursors or their terminal development in the skin. Kinetics of epidermal ADPase ϩ LC in XPA mice after DMBA application were different from those after UVB radiation with regard to the following points: 1) the number of ADPase ϩ LCs after DMBA application decreased and then increased to higher level than those before treatment, whereas the density of ADPase ϩ LCs after UVB radiation recovered only to the pretreatment level (5). 2) Although the reappearance of ADPase ϩ LC after UVB radiation in XPA mice was delayed compared with that in wild-type mice, that after DMBA application in XPA mice was not. The reasons for these findings were not clarified. Mice rendered unresponsive by immunization with DNFB through DMBAtreated dorsal skin remained unresponsive when reimmunized through nontreated abdominal skin. Therefore, contact with hapten through DMBA-treated skin might have stimulated suppressor cells that inhibited the development of CHS to that hapten. Actually, Halliday et al. reported that DMBA-induced immunosuppression was associated with the development of Ag-specific suppressor T lymphocytes that inhibited the induction of cellular immune  responses when adoptively transferred to naive syngeneic recipients (17,18). However, the mechanisms by which DMBA induces suppressor T cells have not been clarified. We assumed that, similar to UVB radiation, DMBA may exert its immunosuppressive effects by altering APC function in skin. In our experiments although the number of epidermal ADPase ϩ LCs was significantly decreased in both XPA and wild-type mice 4 days after a 0.5% DMBA application (Fig. 2), local immunosuppression was greatly enhanced in XPA mice when DNFB sensitization was performed 4 days after 0.5% DMBA application (Fig. 3). This result might indicate that LC depletion alone cannot explain DMBA-induced inhibition of the CHS. It is possible that mechanisms similar to UVB-induced immunosuppression might be involved in DMBAinduced local immunosuppression. Exposure to DMBA was demonstrated to suppress the IL-2 production of Th cells in vivo (29) and in vitro (30). These observations suggest the dysfunction of Th1 cells in establishing DMBA-induced immunosuppression. It has been demonstrated that UV-treated LCs are unable to stimulate Th1 cells, whereas their ability to stimulate Th2 cells is unaffected. UVB radiation induces the production and secretion of a variety of immunomodulatory factors in the skin. Among them, PGE 2 is known to be an inflammatory mediator as well as an immunosuppressive factor. We previously reported that the amount of PGE 2 in skin significantly increased after UVB irradiation in XPA mice, which might account for the enhanced inflammation and local immunosuppression induced by UVB radiation in XPA mice (13). In the present study, we demonstrated that the application of DMBA also induced the production of a pronounced amount of PGE 2 in the skin of XPA mice. Moreover, in XPA mice, the DMBA-induced ear swelling response and local immunosuppression were partially inhibited by treatment with indomethacin, a potent inhibitor of PG biosynthesis. Therefore, enhanced production of PGE 2 after DMBA application, similar to that after UVB radiation, might cause enhancement of inflammation and local immunosuppression in XPA mice. Unrepaired DMBA-induced DNA lesions might cause an excess production of PGE 2 in XPA mice. In addition to PGE 2 , IL-10 and TNF-␣ production in the epidermis were detected by 0.1% DMBA application in XPA mice, but not in wild-type mice. Previously, Nishigori et al. provided direct evidence that UVB-induced DNA damage initiates the production of IL-10 in keratinocytes (31). They showed that treatment with liposomal T4 endonuclease V, an excision repair enzyme specific for CPD in DNA, reduced UVB-induced production of IL-10 in mice in vivo and in vitro. Similarly, it is possible that unrepaired DMBA-DNA adduct in XPA mice stimulates keratinocytes to produce IL-10 and TNF-␣, thereby modifying critical steps in the immunological pathway and ultimately leading to the suppression of the immune response and generation of Ag-specific suppressor T cells.
In this study, DMBA-induced systemic immunosuppression was enhanced in XPA mice. These mice became tolerant against DNFB, because they could not be resensitized against the same hapten. Therefore, DMBA-induced systemic immunosuppression might be mediated by the development of Ag-specific suppressor T cells, similar to DMBA-induced local immunosuppression. A key element in UVB-induced systemic immunosuppression is the production of epidermal cytokines, PGs, and cis-urocanic acid, which mediate the transfer of signals from UV-irradiated skin to lymphoid cells. Shreedhar et al. suggested that the cytokine cascade, including PGE 2 , IL-4, and IL-10, is responsible for UVinduced systemic immunosuppression (12), because PGE 2 can induce the secretion of IL-10 from PBMC (32) or from keratinocytes via IL-4 (12). We demonstrated that DMBA treatment induced a detectable release of serum IL-10 in XPA mice. Thus, it is possible that PGE 2 derived from DMBA-treated skin induces the secretion of IL-10 from keratinocytes and/or PBMC. Consequently, the APC function in the lymph nodes is altered by PGE 2 and/or IL-10 to develop systemic immunosuppression. Although systemic suppression was induced by application of 0.05 and 0.1% DMBA in XPA mice, there was detectable serum level of IL-10 by ELISA only in 1% DMBA-treated XPA mice. There was also detectable serum IL-10 in two of five XPA mice treated with 0.1% DMBA, but not in 0.05% DMBA-treated XPA mice. These findings might be due to the limitation of the detectability of this assay. Similarly, elevation of serum IL-10 was detectable only after high dose UVB radiation (33). Therefore, even an undetectable amount of IL-10 in serum seems to play a crucial role in the induction of systemic immunosuppression.
The results presented here suggest that DMBA-DNA adduct formation is a primary molecular event for the initiation of immunosuppression induced by DMBA, because XPA mice, having a defect in the repair of DMBA-DNA adduct, showed enhanced local and systemic immunosuppression in response to DMBA. Regarding UVB-induced immunosuppression, DNA damage caused by UVB has been demonstrated to be one of the primary molecular mechanisms for initiation of immunosuppression. Kripke and colleagues reported that reducing the number of CPDs in UV-irradiated skin abrogated the systemic suppression of CHS and delayedtype hypersensitivity (34,35). The action spectra for UV-induced suppression of the mixed lymphocyte and mixed epidermal celllymphocyte reactions were shown to closely resemble those for the induction of CPD and 6-4 photoproducts (36). Action spectra for local and systemic suppression of CHS are also consistent with DNA damage as initiators of these effects (37,38). Our previous study, in which XPA mice showed increased local and systemic immunosuppression induced by UVB radiation, supported this hypothesis (5). Moreover, Kripke and colleagues indicated that DNA damage other than CPD also causes immune suppression. They reported that the liposomal HindIII restriction enzyme, which causes double-strand breaks at the 5Ј-AAGCTT-3Ј base sequence in DNA, induced local and systemic immunosuppression (39,40). However, treatment of HindIII failed to induce suppressor T cells, suggesting that the mechanisms inducing immunosuppression were different for UVB and HindIII treatment. Our results support the possibility that DNA damage by other than UVB can also induce immune suppression and trigger the production of immunomodulatory cytokines. Recently, Yarosh et al. reported that UVB-induced DNA damage leading to an increase in and activation of FKBP12-rapamycin-associated protein kinase, resulted in gene expression of TNF-␣ (41). Moreover, it was demonstrated that TNF-␣ induced cyclooxygenase-2 expression, leading to the production of PGE 2 in human lung epithelial cells (42). In our model it is speculated that DMBA-induced DNA damage might lead to the production of TNF-␣ mediated by DNA protein kinase, resulting in the expression of cyclooxygenase-2 and production of PGE 2 . Further studies are necessary to elucidate the precise molecular mechanisms by which DNA damage in cells induces production of immunomodulately cytokines and PGE 2 .
Clinically, XP patients show increased rates of neoplasms in organs that are not exposed to UV radiation. Kraemer and colleagues reported that XP patients had a 10-to 20-fold increase in the occurrence of neoplasms in sites not exposed to UV radiation compared with the general population (43). We have previously shown that NK function is easily suppressed by UVB radiation in XPA mice, suggesting that impairment of NK function by UVB radiation might contribute to an increased internal malignancy in XP patients (44). Exposure to environmental carcinogens, endogenous mutagens, and/or oxygen-free radicals (45) may be important for the development of these internal neoplasms in XP patients. XP cells are hypersensitive to mutagenic tryptophan pyrolysis products in charbroiled meat and fish (46,47) and are also hypersensitive to killing and are hypermutable to DNA-damaging chemical carcinogens, such as benzopyrene, found in cigarette smoke (48,49). Moreover, oral treatment of XPA mice with benzopyrene, acetylaminofluorene, and aminomethylphenylidazopyridine resulted in lymphomas, liver and bladder tumors, and intestinal adenomas and lymphomas, respectively (50). DMBA is a prototype carcinogenic polyaromatic hydrocarbon. Polyaromatic hydrocarbons are ubiquitous environmental contaminants generated as byproducts of the incomplete combustion of fossil fuels, wood, and other organic materials (51). The high incidence of internal cancers in XP patients may be due to a defect in the repair of such chemical carcinogen-damaged DNA. Moreover, the present study suggested that easy impairment of the immune surveillance system by chemical carcinogens demonstrated in XPA mice could also be involved in the high incidence of internal malignancy in XP patients. The role of cyclo-oxygenase and PGs in tumorigenesis is well established (52,53), and cyclo-oxygenase inhibitors are effective in tumor patients in both prophylactic and therapeutic settings (52,54). Overexpression of PGE 2 induced by UVB and chemical carcinogens may also be responsible for the increased incidence of neoplasms in XPA patients. If this is true, cyclo-oxygenase inhibitors may be beneficial for XPA patients.