Animal models for atopic dermatitis: are they relevant to human disease?

Journal of Dermatological Science (2004) 36, 1—9

REVIEW

Animal models for atopic dermatitis: are they relevant to human disease? Tetsuo Shiohara*, Jun Hayakawa, Yoshiko Mizukawa Department of Dermatology, Kyorin University School of Medicine, 6-20-2 Shinkawa, Mitaka, Tokyo 181-8611, Japan Received 28 January 2004; received in revised form 18 February 2004; accepted 18 February 2004

KEYWORDS Atopic dermatitis; NC/Nga mice; Hapten; Th1 and Th2 cytokines; Corticosteroids; Tacrolimus; IL-18; RelB

Summary Over the last decade, animal models of atopic dermatitis (AD) have received increasing attention. They include NC/Nga mice, a hapten-induced mouse model, and transgenic and knockout mouse models. Although the pathogenesis of skin inflammation elicited in these models and that in AD are not quite the same, it is pertinent to ask what these animal models really tell us about the pathogenesis and possible therapies for the disease. NC/Nga mice may yield information relevant to the dissection of the crucial components of the pathophysiology of AD rather than the assessment of potentially therapeutic agents for its treatment. A hapten-induced mouse model created by repeated application of 2,4,6-trinitrochlorobenzene (TNCB) is a simple and reproducible one. This model offers several advantages over others: by changing hapten and the mouse strain used, various types of chronic inflammation, probably reflecting heterogeneity in clinical presentation of AD, can be induced; this model is also of enormous value in its high reproducibility as well as the ease of quantitative assessment by measuring ear thickness. Among various transgenic and knockout mouse models, the IL-18-transgenic mouse is one of the closest available mouse models of human AD, although the onset of the AD-like lesions in the IL-18transgenic mice is such a late event. Although these mice all have significant disadvantages, it is important to review the current literature on the models in the hope that one may identify useful areas for investigation. ß 2004 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.

1. Introduction There has been a striking rise in the incidence of atopic dermatitis (AD) during the past two decades, which is not simply due to an increased recognition of the disease. Population studies suggest that, in most countries, AD now affects 10—20% of children

*Corresponding author. Tel.: þ81-422-47-5511; fax: þ81-422-41-4741. E-mail address: [email protected] (T. Shiohara).

at some point during childhood [1]. In particular, higher prevalences have been recorded in urban regions than in rural regions of developed countries, and the disease is more common in higher social class groups, suggesting that environmental factors associated with more industrialized and urban living determine expression of AD [2]. Because the use of antibiotics and topical corticosteroids has increased in the Western World at the same time that the incidence of AD has increased, exposure to these agents early in life has been also suggested to be a risk factor for the development of AD [3]. Although

0923–1811/$30.00 ß 2004 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.jdermsci.2004.02.013

2

T. Shiohara et al.

topical corticosteroids remain one of the most important treatments available for AD, patients’ irrational fears about using topical corticosteroids have become the greatest barrier to effective longterm management of severe AD [4]. Indeed, there is limited knowledge on long-term efficacy and safety of topical corticosteroids for the treatment of AD; this lack of information causes many dermatologists to be wary of their extensive use. Thus, research on the etiopathogenesis of AD and testing of potential therapeutic agents has been hampered by the paucity of reproducible and histopathologically relevant animal models of AD. Animal models allow detailed study of the early stages of the illness and facilitate more rapid trials of possible therapies for the disease; clinical trials in patients with AD are intrinsically cumbersome, requiring the enrollment of large numbers of patients over a long period. Over the last decade, animal models of AD have received increasing attention. The earlier models included NC/Nga mice [5] and a hapten-induced mouse model [6]. More recently, other mouse models of AD have been developed using knockouts or transgenes [7—10]. Although each of these models has some features that are characteristic of human AD, they all have significant disadvantages, such as limited reproducibility, a requirement for repeated treatments, a long prodromal phase preceding overt disease and limited availability of these genetically-engineered mice. Despite these disadvantages, it would seem pertinent to review the current literature on animal models of AD in the hope that useful areas for investigation can be identified. This review examines what has been learned about the pathogenesis and treatment of AD from these animal models. The advantages and disadvantages of these models are summarized in Table 1. Table 1

2. NC/Nga mice The NC/Nga mouse is the most extensively studied animal model of AD. The NC/Nga strain originated from Japanese fancy mice and was established as an inbred strain by Kondo et al. in 1957 [11]. This strain has been reported to have features such as a high susceptibility to X-irradiation and to anaphylactic shock induced by ovalbumin. NC/Nga mice have also been reported to develop AD-like eczematous skin lesions when kept in an air-uncontrolled conventional room but not when maintained under specific pathogen-free (SPF) conditions [5]. Clinical symptoms begin with itching, erythema, hemorrhage, scaling, dryness, and alopecia at the age of 8 weeks. These eczematous skin lesions are typically observed on the face, nose, ears, neck and back, suggesting that they are caused by hind limb scratching. The clinical severity of the dermatitis, as determined by a scoring system which has been established for human AD, increases with age and reaches a maximum at around 17 weeks of age [5]. Serum IgE levels begin to increase at 8 weeks of age, coinciding with the appearance of the skin lesions and peak at 17 weeks, thus correlating with clinical severity. In contrast, IgG levels are unaffected until 17 weeks of age. Recently, Matsuda et al. [12] have demonstrated that the NC/Nga mice kept under SPF conditions are able to respond with increased IgE on immunization with ovalbumin and aluminun hydroxide in vivo, despite no apparent clinical symptoms and an absence of IgE hyperproduction. They have further shown that B cells isolated from NC/Nga mice possess a higher potency for IgE synthesis due to a higher sensitivity to CD40L and IL-4 compared with those from BALB/c mice, suggesting that constitutive and enhanced Janus kinase 3 phosphorylation in B cells highly sensitive to CD40L and IL-4 is attributable to IgE

Comparison of mouse models for AD

Variables

NC/Nga mouse model

Hapten-induced model

IL-18 Tg mouse model

Incidence of AD-like lesions Need for additional treatment to develop AD-like lesions Age of onset months

50% Hapten application

100% No

(0)a 100% Hapten applicationa

8—17 weeks

6

Reproducibility Need for particular conditions Variations

Poor—fair Conventional conditions Impossible

Availability

Limited

Anytime 30 days after starting hapten application Excellent No Possible by changing haptens or mouse strains employed Excellent

a

According to ref. [8].

Fair No Impossible

Limited

Animal models for atopic dermatitis

hyperproduction in NC/Nga mice kept in conventional conditions [12]. Histologically, there are significant changes such as an increased number of mast cells with mild degranulation, infiltration of numerous eosinophils and a small number of mononuclear cells, even at 7 weeks of age under conventional conditions, when only minimal clinical symptoms can be detected. Dramatic changes can be detectable in the NC/Nga mice at 17 weeks of age, including thickening of the epidermis with marked hyperkeratosis and parakeratosis, and a marked increase in mast cell numbers and numerous eosinophils with marked degranulation [5]. Immunohistochemical analysis of involved skin at 17 weeks under conventional conditions showed infiltration of numerous CD4þ T cells and macrophages with a few CD8þ T cells, a finding seen in human AD. In 17-week-old NC/Nga mice kept under conventional conditions, IL-4 and IL-5 are produced by mast cells and CD4þ T cells in lesional skin, while IFN-g is also produced by a small number of these cells. These results suggest that defective production of IFN-g by T cells in the NC/Nga mice could result in IgE hyperproduction. Matsuda and co-workers [13] therefore examined the effect of IL-12 and IFN-g on IgE synthesis in NC/Nga mice. The results showed that rIL-12 induces defective production of IFN-g due to low phosphorylation of STAT 4, resulting in failure of rIL-12 to inhibit IgE synthesis in NC/Nga mice immunized with ovalbumin. In a study using NC/Nga mice, Vestergaard et al. [14] provided evidence to indicate that thymus- and activation-regulated chemokine (TARC) produced by basal keratinocytes and macrophage-derived chemokine (MDC) produced by dermal dendritic cells (DC) may play a significant role in recruiting Th2-type T cells expressing CCR-4 to the AD-like skin lesions. Impairment of water retention properties and barrier function with decreased levels of ceramide in the skin, findings also seen in AD, are reported in NC/Nga mice [15]. Given the resemblance of these immunological alterations in NC/ Nga mice to AD in humans, these data have important implications for the understanding and treatment of AD; they strongly suggest that Th2 cells are primarily involved in the development of AD-like lesions in the NC/Nga mice. However, not in accordance with this notion is the recent observation that STAT6-deficient NC/Nga mice, that cannot mount Th2 cytokine responses and produce serum IgE responses, develop AD-like skin lesions at an equivalent frequency and time of onset as normal NC/Nga littermates [16]. In addition, the skin lesions elicited in STAT6-deficient NC/Nga mice displayed histological features indistinguishable from those

3

in control NC/Nga mice. These findings clearly indicate that accumulation of Th2 cells and increased IgE levels are not prerequisites for the development of AD-like lesions in NC/Nga mice, and these alterations may be consequences of the progression of the pathogenesis of AD-like inflammation. Sophisticated genetic studies in NC/Nga mice performed by Yagi et al. [16] and by Kohara et al. [17] showed linkage to chromosome 9. This region is the syntenic homolog of human chromosome 11q23, which is linked in some human studies to eosinophil counts and specific IgE responses [18]. Thus, the NC/Nga mice are now inbred, yet the strain is not 100% concordant for the development of AD-like skin lesions even under conventional conditions. Approximately 50% of NC/Nga mice spontaneously develop AD-like skin lesions around the age of 6 months or later, implying that environmental factors may be essential for their development. Keeping these considerations in mind, and based on clinical observations in AD, Sasagawa et al. [19] examined whether mite antigens could cause AD-like skin lesions in NC/Nga mice under conditions in which other environmental factors are not involved. To test this possibility, mite antigens were intradermally injected into the ventral side of the ears of NC/Nga mice kept in SPF conditions. They demonstrated that mite antigens can induce not only AD-like skin lesions but also Th2-dominated immune responses in NC/Nga mice maintained under SPF conditions, while they cannot induce such skin lesions in BALB/c mice. Although these results indicate that environmental allergens such as mite antigens contribute to the development of AD-like lesions in NC/Nga mice [19], as suggested in human AD, it would be difficult to attribute a primary role to mite antigens rather than accepting them as secondary invaders at the onset of overt disease in NC/Nga mice. The NC/Nga mice can be also used for evaluating an effective therapeutic strategy. The major drawback of using this strain as a tool for the investigation of therapeutic strategy is the relatively low incidence of AD-like lesions even in NC/ Nga mice kept under conventional conditions. Additional treatments such as hapten application is often needed to be 100% concordant for the development of AD-like lesions (Table 1). Another drawback is the difficulty in quantitatively evaluating the severity of skin lesions. To overcome this difficulty, Matsuda et al. employed a clinical severity score previously described for human AD. As an alternative approach, Vestergaard et al. [14] counted the number of cells infiltrating the skin; they showed that topical corticosteroids markedly reduce the infiltration of CD4þ and CD8þ

4

T. Shiohara et al.

T cells while leaving macrophages and mast cells unaffected. Hiroi et al. [20] also reported a remarkable ameliorating effect of tacrolimus ointment on spontaneously-developed AD-like skin lesions in the NC/Nga mice. In summary, this model may yield information relevant to the dissection of the crucial components of the pathophysiology of AD rather than the assessment of potentially therapeutic agents for its treatment.

3. Hapten-induced model Although animal models spontaneously developing AD-like lesions, such as NC/Nga mice, are useful for research on the etiopathogenesis of AD, this strain is not widely available, thus limiting its usefulness. We were prompted by the great need to develop a simple and reproducible animal model of AD to try to do so by using laboratory animals that are widely available. During the process of developing a mouse model, we noted earlier findings that antigenspecific IgE antibody is preferentially produced in mice repeatedly painted with a hapten [21], although this study did not demonstrate that the appearance of a serum IgE response to the inducing agents was associated with the development of an immediate-type hypersensitivity (ITH) response. At that time, however, no attempts had been made to induce an ITH to a hapten by its repeated epicutaneous application. In 1995, we reported that repeated application of 2,4,6-trinitrochlorobenzen (TNCB) at 2-day intervals for 24 days to the same skin site results

in a site-restricted shift in the time course of antigen-specific hypersensitivity responses from a typical delayed-type to an ITH followed by a late reaction [6], a finding often seen in skin lesions of AD patients. This shift is associated with epidermal hyperplasia, accumulation of large numbers of mast cells and CD4þ T cells beneath the epidermis, and elevated serum levels of antigen-specific IgE. In this model, immune responses initially induced by epicutaneous introduction of a hapten through normal skin shifts to those induced by chronic introduction of the hapten into the damaged skin, as the antigeninduced hypersensitivity response progresses into chronicity [6]. When sequential cytokine mRNA expression after hapten application is assessed in the acute versus chronic lesions, the former is driven by the production of Th1 cytokines (IFN-g and IL-12) while the latter is driven by the production of Th2 cytokines IL-4 and IL-10 [22]. These results clearly indicate that chronic epicutaneous exposure to a hapten, which can predominantly induce Th1 cytokines during the primary response, induces a shift in the pattern of antigen-induced cytokine expression toward the induction of Th2 cytokines in a site-restricted fashion. The inflammatory response in chronic lesions shares many of the histopathological, immunological and clinical features of human AD. This model also allows the study of events characterizing the progression from acute to chronic inflammation, and has important implications for the understanding of the mechanism by which chronic antigenic exposure drives the immune system. In an area subjected to chronic antigenic exposure, the most appropriate type of response to

Hapten

Hapten

Epidermis

DC1

DC1

Epidermis

Th1 cell Th1 cytokine DC1

DC1

Th2 cell

Endothelial cell

Mast cell mast cell-derived granules

Th2 cytokine

Fig. 1 Conversion from a Th1- (left) to a Th2-dominated (right) response upon repeated hapten application in a hapten-induced mouse model. In the chronic lesions mimicking AD lesions (right), Th1 cells would be silenced through a combination of an increase in Th2 cells and a paucity of IL-12-producing dendritic cells (DC1).

Animal models for atopic dermatitis

limit tissue damage must be chosen; the Th2-dominated immune responses observed in chronic lesions may be beneficial for maintaining immunologic homeostasis in the skin chronically exposed to a hapten. Thus, the conversion from a Th1 to a Th2 response as the consequence of inflammation progressing into chronicity might be interpreted as representing a natural evolution process directed towards reducing the more deleterious Th1 response upon repeated hapten application, as Th2 responses have been shown to minimize the tissue-damaging effects of Th1 responses. We reason that if this notion is true, CD4þ T cells and mast cells capable of producing Th2 cytokines would accumulate beneath the epidermis, serving a regulatory function in opposing the tissue-damaging effects of Th1 cells (Fig. 1). This model may provide important insights into the way Th2 cells can regulate the function of Th1 cells in AD lesions. Severe pathological reactions observed in AD were thought to result from defective cross-regulation by Th1 cells, that can normally inhibit Th2 cells. Nevertheless, several lines of evidence argue against an essential role of Th2 cells in the pathogenesis of AD: the lesional skin of patients with AD has been shown to contain significant numbers of Th1 cells; and studies on cytokine mRNA expression in lesional skin has shown that IFN-g production, which is thought to be mediated by Th1 cells, is higher than that observed in contact dermatitis, and that IFN-g production increases following the evolution of individual lesions [2,23,24]. In interpreting the reported data, one must appreciate that the frequencies, as determined by PCR analyses and allergen-specific T cell clones established from lesional skin, do not necessarily reflect the actual frequency in the lesions: all of these reports illustrate the controversies and difficulties in providing clear and specific evidence on this subject. To circumvent part of this difficulty and to allow a detailed analysis of the kinetics of the pathological changes accompanied by profound alterations in cytokine production patterns within the tissue sites, we asked whether the strong skewing to Th2 responses in the chronic lesions could be due either to a decrease in the frequency of Th1 cells or a decrease in the amount of Th1 cytokines produced per cell. Our ex vivo intracellular cytokine expression analysis demonstrated that equivalent frequencies of IFN-g producing T cells are present in acute and chronic lymph node (LN) cells although the chronic LN contains a higher frequency of IL-4 producing T cells than the acute LN. There is no significant difference in the levels of IFN-g and IL-4 produced on a per cell basis between the acute and chronic LN. Because a significant decrease in the

5 frequencies of CD11cþ dendritic cells capable of producing IL-12 is noted in the chronic LN, many of IFN-g-producing T cells in the chronic LN would be silenced through a combination of an increase in IL-4-producing T cells and a paucity of IL-12-producing dendritic cells. This model offers several advantages over others. By changing hapten and mouse strain, various types of chronic inflammation, probably reflecting heterogeneity in clinical presentation of AD, can be induced. Indeed, at least two types of AD have been identified: an extrinsic type associated with IgE-mediated sensitization, which affects 70—80% of patients; and an intrinsic type without IgE-mediated sensitization, which affects 20—30% of patients [25]. This model had originally been developed using BALB/c mice, which are known to have an intrinsic tendency to develop towards theTh2 response, repeatedly exposed to TNCB. However, when another hapten, oxazolone, is used for repeated application, it differs from TNCB-induced dermatitis by the fact that a mixed Th1 þ Th2 response can be obtained even in the acute lesions, although repeated application of oxazolone results in a shift to Th2 responses. In contrast, when C57BL/6 mice, known to be biased toward mounting a Th1 response, are repeatedly exposed to oxazolone, Th2-dominated chronic inflammation, which is neither associated with an increase in serum IgE levels nor the development of ITH, can emerge. It would be reasonable to assume that BALB/c mice repeatedly exposed to TNCB represent an extrinsic type of AD, while C57BL/6 mice repeatedly exposed to oxazolone represent an intrinsic type. Thus, variability in the clinical presentation of AD can be reproduced by using different haptens and mouse strains. This simple and reproducible model is of enormous value in the assessment of potentially therapeutic agents for the treatment of AD. The major advantage of using this model in trials of possible therapies over other models is its reproducibility as well as the ease of quantitative assessment by measuring ear thickness. The short time taken to develop the AD-like lesions (within approximately 30 days after starting hapten application on day 7) is an additional advantage of this model (Table 1). On the other hand, it has significant disadvantages inherent to inducible models, such as the requirement for previous sensitization to a hapten (mice must be sensitized 7 days before starting repeated hapten application), and a concern about potential interactions between the hapten and therapeutic agents. Nevertheless, this model allows us to test large numbers of potential treatments that cannot be performed, for ethical reasons, in patients

6

with AD. The efficacy of therapeutic agents on this model is evaluated under two different conditions. To evaluate its protective effect on the development of chronic AD-like lesions, a treatment regimen is started at the same time as the commencement of repeated hapten application; to evaluate its curative effect on the development of chronic AD-like lesions, the application of the test agent is started on day 24, by which time AD-like immunological alterations, such as the development of ITH followed by a late reaction, increased serum IgE levels, and a Th2-dominated immune response, has developed. Although the efficacy of the test agents can be examined by several parameters, such as sequential measurement of total ear thickness, histology, serum IgE levels and scratch behavior, our previous studies showed that sequentially measuring total ear thickness is the most reliable parameter [26,27]. By using this model, we can compare the efficacy of various therapeutic agents with that of topical corticosteroids. However, because in this model corticosteroid-induced skin atrophy makes it impossible to precisely evaluate the effect of anti-inflammatory agents, we have to use prednisolone (in a dose of 0.001 mg per ear) as a positive control, at which dose no atrophic effect on the ear is observed after its daily epicutaneous application for 2 weeks. Because several double-blind, placebo-controlled trials in patients with AD have reported that topical tacrolimus is as effective as topical corticosteroids [28,29], we studied the effect of topical tacrolimus using this model. Indeed, topical tacrolimus was as effective as, or significantly more effective than, topical corticosteroids depending on the evaluation parameters employed. Although in this model the rebound phenomenon, defined as a greater and more rapid deterioration than the non-treated group (receiving only repeated applications of TNCB), can be typically observed during the maintenance phase with vehicle alone after cessation of topical corticosteroids, this steroid rebound was effectively prevented by the subsequent use of tacrolimus (Shiohara, unpublished observations). Furthermore, deterioration observed during the maintenance phase after cessation of topical tacrolimus were much less than that upon cessation of topical corticosteroids. Because the capacity of topical tacrolimus to inhibit neutrophil infiltration and to reduce scratching was much superior to that of topical corticosteroids in this model, these properties may explain the relatively long-lived remissions achieved with topical tacrolimus as compared with the duration of remission achieved with topical corticosteroids. In short, maintenance therapy with topical tacrolimus plus intermittent topical corti-

T. Shiohara et al.

costeroids (one to two times per week) appears to be more effective than the continuous use of tacrolimus. Thus, this model provides a promising in vivo assay system that allows us to test large numbers of potential treatments and evaluate the efficacy of novel anti-inflammatory agents. Indeed, we have recently demonstrated that CX-659S, a novel diaminouracil derivative, has inhibitory activities against the rebound phenomenon following withdrawal of topical corticosteroids [27]. The effectiveness of this agent on existing disease and its effect on the prevention of reccurrence are two important characteristics supporting its potential as an anti-inflammatory agent for AD. The concept of this model has recently extended to ovalbumin-elicited models using mice with targeted deletions of the IL-4, IL-5, and IFN-g cytokine genes [30]. These data also suggest that both Th1 and Th2 cytokines play important roles in inflammatory responses in these models, consistent with the findings observed in AD.

4. Lessons from transgenic and knockout mice A number of animal models are available, but they have only limited usefulness in studying the pathogenesis of AD. The development of transgenic and knockout models is therefore desirable. The sudden and unexpected discovery that genetically engineered mice with specific immune abnormalities are associated with the development of AD-like inflammatory skin lesions represents a welcome novel approach to the study of AD pathogenesis, although this is not possible at present. Instead, a series of transgenic mice, characterized by a susceptibility to AD-like inflammatory skin lesions, has been created. Among them, the disease described in IL-18-transgenic mice is one of the closest available animal models of human AD. Although IL-18 (originally named IFN-g-inducing factor) is a potent inducer of IFN-g, particularly when acting in concert with IL-12 [31,32], it can also potentially induce Th2 cytokines and IgE and IgG1 production [33,34]. Thus, IL-18 can act as a strong cofactor for both Th1 and Th2 cell development [8]. These considerations prompted Konishi et al. [7] to generate transgenic mice overexpressing mature IL-18 in their skin. These IL-18-transgenic (KIL-18 Tg) mice have been shown to develop AD-like skin lesions at about 6 months after birth under SPF conditions. They also generated another transgenic mouse, named KCASP1Tg, which overexpresses caspace-1 in its keratinocytes, and

Animal models for atopic dermatitis

demonstrated that KCASP1Tg mice also spontaneously release biologically active IL-18 and develop AD-like skin lesions within 8 weeks with much faster kinetics than in KIL-18Tg mice. The histology of the AD-like lesions in both Tg mice showed acanthotic epidermis, accumulation of mast cells, and prominent infiltration of lymphocytes and neutrophils. A marked elevation of serum IgE levels, associated with the development of skin lesions, is observed in both Tg mice. However, STAT6-deficient KCASP1Tg mice, created by depleting the STAT6 gene, also develop AD-like skin lesions at the same time as observed in KCASP1Tg mice, despite no detectable levels of IgE. These results indicate that high levels of IL-18, and not high levels of IgE, are responsible for the development of ADlike skin lesions. However, if this notion is true, one critical question is why the onset of the AD-like lesions in KIL-18Tg mice is such a late event. An answer to this question was provided by the investigators: because IL-1-deficient KCASP1Tg mice also required the same long lag period as did KIL-18Tg mice to develop the AD-like lesions, the authors concluded that AD-like lesions can be initiated by overrelease of IL-18 and accelerated by IL-1 [7]. Kawase et al. [8] and Hoshino et al. [35] also demonstrated with the use of IL-18-Tg mice that targeted overexpression of IL-18 in the skin can induce not only an aggravated TNCB-induced chronic contact hypersensitivity reaction with a high serum IgE but also an exacerbated and prolonged croton oil-induced non-allergic irritant contact dermatitis. In IL-18 Tg mice repeatedly treated with TNCB, an acanthotic epidermis with marked hyperkeratosis, an influx of mononuclear cells into the epidermis, and dermal edema were observed [8]. These results indicate that overproduction of IL-18 by keratinocytes alone cannot result in the spontaneous development of AD-like lesions but aggravates chronic AD-like inflammation induced by repeated TNCB application. Unlike the conclusion drawn by Konishi et al. [7], they suggest that IL-18 acts as a strong cofactor that can cause deterioration, rather than induction, of allergic and nonallergic skin inflammation. Consistent with these results, Tanaka et al. [36] reported that serum IL-18 levels are significantly elevated prior to the onset and during the development of AD-like lesions in NC/Nga mice as well as in patients with AD. Great caution is needed, however, in concluding that IL-18 is a critical cytokine that can trigger AD-like lesions, because weekly injection of anti-IL-18 antibody into NC/Nga mice starting at 4 weeks of age failed to inhibit the onset and development of AD-like lesions and elevation of serum IgE [37]. Thus, these transgenic mice are of enormous importance as

7

a research tool to develop new approaches to therapy. Another gene-targeted mouse with AD-like lesions is that lacking the transcription factor RelB, a member of the NF-kB/Rel family [38]. RelB-deficient mice display a complex phenotype, including impaired development of lymphoid organs, reduced numbers of mature dendritic cells, multiorgan inflammation and multifocal defects in immune responses. Between 4 and 10 weeks of age, 70— 80% of RelB/ mice develop a dermatitis characterized clinically by thickened and reddened skin, scaling, itching and loss of hair. Histological and immunohistochemical evaluation of RelB/ skin lesions reveals marked epidermal hyperplasia with hyperkeratosis and infiltration of many CD4þ T cells, eosinophils and lesser numbers of CD8þ T cells. There is markedly increased expression of IL-4, IL-5 and IFN-g in the skin lesions of RelB/ mice. RelB/ mice also have an eosinophilia, IgE overproduction and mast cell degranulation. Interestingly, RelB/ mice with a moderate to marked skin phenotype also develop several features of pulmonary inflammation, which is reminiscent of the association of AD and asthma in humans. Unfortunately, as yet the skin lesions of RelB/ mice have not been extensively studied.

5. Conclusion Finally, it is pertinent to ask what these animal models really tell us about the pathogenesis of AD. Although the pathogenesis of the skin inflammation elicited in these models and that in AD are not quite the same, these studies show quite clearly that many of the immunologic alterations associated with the development of skin lesions, such as Th2-dominated immune responses and increased serum IgE levels, may be a secondary process, rather than the cause of the disease. An important implication of these models is that homeostasis is not determined solely by the characteristics of individual cells, but also by the total number and diversity of competing cells, and that the timing of immune responses is also a critical factor in determining the outcome of inflammation. In this regard, the increasing availability of transgenic mice in which Cre recombinase expression is exquisitely controlled in a tissue- and/or age-dependent manner will allow for the elucidation of the normal physiological role of these multifunctional molecules in postnatal life. Once large numbers of these mice are made available to investigators, an effective treatment for human AD may be closer at hand.

8

T. Shiohara et al.

Acknowledgements This work was supported in part by grants from Ministry of Education, Sports, Science and Culture of Japan (J.H. and Y.M.) and the Ministry of Health, Labor and Welfare of Japan (T.S.).

References [1] Schultz-Larsen F, Hanifin JM. Epidemiology of atopic dermatitis. Immunol Allergy Clin North Am 2002;22:1—24. [2] Leung DY. Atopic dermatitis. Lancet 2003;361:151—60. [3] Mckeever TM, Lewis SA, Smith C, Collins J, Heatlie H, Frisher M, et al. Early exposure to infections and antibiotics and the incidence of allergic disease: a birth cohort study with the west Midlands General Practice Research Database. J Allergy Clin Immunol 2002;109:43—50. [4] Charman CR, Morris AD, Williams HC. Topical corticosteroid phobia in patients with atopic eczema. Br J Dermatol 2000;142:931—6. [5] Matsuda H, Watanabe N, Geba GP, Sperl J, Tsudzuki M, Hiroi J, et al. Development of atopic dermatitis-like skin lesion with IgE hyperproduction in NC/Nga mice. Int Immunol 1997;9:461—6. [6] Kitagaki H, Fujisawa S, Watanabe K, Hayakawa K, Shiohara T. Immediate-type hypersensitivity response followed by a late reaction is induced by repeated epicutaneous application of contact sensitizing agents in mice. J Invest Dermatol 1995;105:749—55. [7] Konishi H, Tsutsui H, Murakami T, Yumikura-Futatsugi S, Yamanaka K, Tanaka M, et al. IL-18 contributes to the spontaneous development of atopic dermatitis-like inflammatory skin lesion independently of IgE/stat6 under specific pathogen-free conditions. Proc Natl Acad Sci USA 2002;99:11340—5. [8] Kawase Y, Hoshino T, Yokota K, Kuzuhara A, Kirii Y, Nishiwaki E, et al. Exacerbated and prolonged allergic and non-allergic inflammatory cutaneous reaction in mice with targeted interleukin-18 expression in the skin. J Invest Dermatol 2003;121:502—9. [9] Barton D, HogenEsch J, Weih F. Mice lacking the transcription factor REIB develop T cell-dependent skin lesions similar to human atopic dermatitis. Eur J Immuol 2000; 30:2323—32. [10] Chan LS, Robinson N, Xu L. Expression of interleukin 4 in the epidermis of transgenic mice results in a pruritic inflammatory skin disease: an experimental animal model to study atopic dermatitis. J Invest Dermatol 2001;117: 977—83. [11] Kondo K, Nagami T, Tadokoro S. Differences in haematopoeitic death among inbred strains of mice. In: Bond P, Sugawara S, editors. Comparative cellular and species radiosensitivity. Tokyo, Japan: Igakushoin; 1969. p. 20. [12] Matsumoto M, Ra C, Kawamoto K, Sato H, Itakura A, Sawada J, et al. IgE hyperproduction though enhanced tyrosine phosphorylation of Janus Kinase 3 in NC/Nga mice, a model for human atopic dermatitis. J Immunol 1999;162:1056— 63. [13] Matsumoto M, Itakura A, Tanaka A, Fujiwara C, Matsuda H. Inability of IL-12 to down-regulate IgE synthesis due to defective production of IFN-g in atopic NC/Nga mice. J Immunol 2001;167:5955—62. [14] Vestergaard C, Yoneyama H, Murai M, Nakamura K, Tamaki K, Terashima Y, et al. Overproduction of Th2-specific

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

chemokines in NC/Nga mice exhibiting atopic dermatitislike lesions. J Clin Invest 1999;104:1097—105. Aioi A, Tonogaito H, Suto H, Hamad K, Ra C, Ogawa H, et al. Impairment of skin barrier function in NC/Nga mice as a possible murine model for atopic dermatitis. Br J Dermatol 2001;144:12—8. Yagi R, Nagai H, Iigo Y, Akimoto T, Arai T, Kubo M. Development of atopic dermatitis-like skin lesions in STAT6deficient NC/Nga mice. J Immunol 2002;168:2020—7. Kohara Y, Tanabe K, Matsuoka K, Kanda N, Matsuda H, Karasuyama H, et al. A major determinant quantitativetrait locus responsible for atopic dermatitis-like lesions in NC/Nga mice is located on Chromosome 9. Immunogenetics 2001;53:15—21. Daniels SE, Bhattacharrya S, James A, Leaves NI, Young A, Hill MR, et al. A genome-wide search for quantitative trait loci underlying asthma. Nature 1996;383:247— 50. Sasagawa T, Higashi Y, Sakuma S, Hirayama Y, Sasagawa Y, Ohkubo Y, et al. Atopic dermatitis-like skin lesions induced by topical application of mite antigens in NC/Nga mice. Int Arch Allergy Immunol 2001;126:239—47. Hiroi J, Sengoku T, Morita K, Kishi S, Sato S, Ogawa T, et al. Effect of tacrolimus hydrate (FK506) ointment on spontaneous dermatitis in NC/Nga mice. Jpn J Pharmacol 1998;76:175—83. Thomas WR, Watkins MC, Jouhal SS, Asherson GL. Induction and modification of anti-TNP reaginic and IgG antibody response by reactive trinitrophenyl derivatives. Immunology 1978;35:673—82. Kitagaki H, Ono N, Hayakawa K, Kitazawa T, Watanabe K, Shiohara T. Repeated elicitation of contact hypersensitivity induces a shift in cutaneous cytokine milieu from a T helper cell type 1 to a T helper cell type 2 profile. J Immunol 1997;159:2484—91. Leung DYM. Atopic dermatitis: new insights and opportunities for therapeutic intervention. J Allergy Clin Immunol 2000;105:860—76. Akdis CA, Akdis M, Trautmann A, Blaser K. Immune regulation in atopic dermatitis. Curr Opin Immunol 2000; 12:641—6. Johansson SG, Hourihane JO, Bouswuet J, BruijnzeelKoomen C, Dreborg S, Haahtela T, et al. A revised nomenclature for allergy. An EAACI position statement from the EAACI nomenclature task force. Allergy 2001;56: 813—24. Inoue Y, Isobe M, Shiohara T, Goto Y, Hayashi H. Protective and curative effects of topically applied CX-659S a novel diaminouracil derivative, on chronic picryl chloride-induced contact hypersensitivity responses. Br J Dermatol 2002;147:675—82. Inoue Y, Isobe M, Shiohara T, Hayashi H. Inhibitory activity of CX-659S, a novel diaminouracil derivative, against the rebound phenomenon following withdrawal of corticosteroid therapy for chronic contact hypersensitivity responses. Int Arch Allergy Immunol 2003;131:143—52. Boguniewicz M, Fiedler VC, Raimer S, Lawrence ID, Leung DY, Hanifin JM. A randomized, vehicle-controlled trial of tacrolimus ointment for treatment of atopic dermatitis in children. J Allergy Clin Immunol 1998;102:637—44. Hanifin JM, Ling MR, Langley R, Breneman D, Rafal E. Tacrolimus ointment for the treatment of atopic dermatitis in adult patients. Part 2. Efficacy. J Am Acad Dermatol 2001;44:S28—38. Spergel JM, Mizoguchi E, Oettgen H, Bhan AK, Geha RS. Roles of Th1 and Th2 cytokines in a murine model of allergic dermatitis. J Clin Invest 1999;103:1103—11.

Animal models for atopic dermatitis

[31] Ahn HJ, Maruo S, Tomura M, Mu J, Hamaoka T, Nakanishi K, et al. A mechanism underlying synergy between IL-12 and IFN-g-inducing factor in enhanced production of IFN-g. J Immunol 1997;159:2125—31. [32] Robinson D, Shibuya K, Mui A, Zonin F, Murphy E, Sana T, et al. IGIF does not drive Th1 development but synergizes with IL-12 for interferon-g production and activates IRAK and NFkB. Immunity 1997;7:571—81. [33] Hoshino T, Yagita H, Ortaldo JR, Wiltrout RH, Young HY. In vivo administration of IL-18 can induce IgE production through Th2 cytokine induction and up-regulation of CD40 ligand (CD154) expression on CD4þ T cells. Eur J Immunol 2000;30:1998—2006. [34] Xu D, Trajkovic V, Hunter D, Leung BP, Schultz K, Gracie JA, et al. IL-18 induces the differentiation of Th1 or Th2 cells depending upon cytokine milieu and genetic background. Eur J Immunol 2000;30:3147—56. [35] Hoshino T, Kawase Y, Okamoto M, Yokota K, Yoshino K, Yamanura K, et al. IL-18-transgenic mice: in vivo evidence of a broad role for IL-18 in modulating immune function. J Immunol 2001;166:7014—8. [36] Tanaka T, Tsutsui H, Yoshimoto T, Kotani M, Matsumoto M, Fujita A, et al. Interleukin-18 is elevated in the sera from patients with atopic dermatitis and from atopic dermatitis model mice, NC/Nga. Int Arch Allergy Immunol 2001; 125:236—40.

9

[37] Higa S, Kotani M, Matsumoto M, Fujita A, Hirano T, Suemura M, et al. Administration of anti-interleukin 18 antibody fails to inhibit development of dermatitis in atopic dermatitismodel mice. Br J Dermatol 2003;149:39—45. [38] Barton D, HogenEsch H, Weih F. Mice lacking the transcription factor ReIB develop T cell-dependent skin lesions similar to human atopic dermatitis. Eur J Immunol 2000; 30:2323—32. Tetsuo Shiohara received the MD degree from Keio University School of Medicine, Tokyo, Japan, and the PhD degree from Keio University Graduate School of Medicine, in 1973 and 1977, respectively. He has been at Kyorin University School of Medicine, Tokyo, Japan since 1979, and is currently a professor and chairman in the Department of Dermatology. During 1983— 1985, he worked as a research associate at the Department of Dermatology, Yale University School of Medicine, New Haven, CT, USA, where he received the Rita Annerberg Awards for Excellence in Clinical research. His interests include immunodermatology, drug eruption, atopic dermatitis, skin-homing of T cells, and innate immunity.