Animal models of toxic epidermal necrolysis


Hiroaki Azukizawa, M.D., Ph.D., 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan. Email:


Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are severe acute exfoliative skin diseases developing extensive epidermal detachment and mucosal damage. Although SJS and TEN are mostly caused by drugs, an animal model of TEN using drugs has not been established yet. We have established an autoimmune skin disease model mouse reproducing the devastating skin damage of TEN by a combination of transgenic mice expressing an epidermal model antigen and its specific CD8+ T-cell receptor. In this model mouse, we found that the thymus-derived CD4+CD25+ regulatory T cell (Treg) is a critical regulator of cytotoxic T lymphocytes (CTL) causing TEN. Indeed, loss of Treg function was recently demonstrated by human studies of TEN patients. Although how drug-reactive CTL is activated in vivo is still unknown, this model elucidated the immunological pathomechanism of TEN after CTL obtained cytotoxicity against epidermal keratinocyte. In this review, roles of CTL, Treg, cytotoxic granules and antigen-presenting cells were discussed on pathogenesis of TEN.


Cutaneous adverse drug reaction (cADR) is a frustrating problem for the patient and physician, because withdrawal of the causative drug is often necessary to end the symptom. Allergic drug reaction is usually idiosyncratic in both immediate reaction and non-immediate reaction, classified as type I allergy and type IV allergy by Coombs and Gell,1 respectively. Physicians sometimes encounter mild cases of type IV delayed allergic reaction of cADR, such as maculopapular rash, in daily clinical practice. On the other hand, severe cADR, especially Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN), is a rare and unanticipated event.2 Although drugs are the major cause of SJS/TEN, herpes simplex virus or Mycoplasma infection is also occasionally associated with the acute phase.3 In addition, certain causative drugs or infections cannot be identified in some SJS/TEN cases, suggesting that part of SJS/TEN cases is related to unknown factors.

As a defense system, the immune system protects our bodies against microbes, whereas it sometimes creates an unfavorable situation by being involved in the pathogenesis of various diseases. Innate immunity and adaptive immunity are the two major defense systems, in particular, T-cell immunity plays a key role on pathogenesis of many inflammatory diseases.4 For example, donor-derived T cell, which is non-self, kills recipient cells expressing the host self-antigen in acute graft-versus-host disease (GVHD) after hematopoietic cell transplantation.5 On the other hand, autoreactive T cells cause immune reaction to the self-antigen and lead to organ destruction in autoimmune diseases. Obviously, T-cell-mediated, delayed drug hypersensitivity reaction can be classified into neither GVHD nor autoimmune disease.6 In drug hypersensitivity reaction, the drug is a non-self-antigen recognized by drug-reactive “native” T cells. Conversely, clinical symptom of acute GVHD are often difficult to distinguish from drug eruption, because maculopapular rash and TEN can be seen in both acute GVHD and delayed drug hypersensitivity reaction.7 Unlike acute GVHD or autoimmune diseases, the target of T-cell immunity is not peptides derived from tissue, but low molecular weight chemical substances in the delayed drug hypersensitivity reaction, suggesting that the culprit drug may haptenize self-antigen like in contact hypersensitivity reaction. Moreover, drug-reactive T cells can be activated by drugs binding to major histocompatibility complex (MHC) molecules without self-peptides, and this phenomenon is so-called the pharmacological interaction (p-i) concept.8,9 Nevertheless, T cells activated by a drug cannot remove a drug from the body, but induce skin eruption, and drug eruption unconsciously alarms patients and physicians against continuation of the causative drug.

Adoptive transfer models of ten

In humans, TEN is a most severe clinical symptom in delayed drug hypersensitivity and acute GVHD. In mice, acute GVHD, identical to the human disease, can be induced by allo-specific T-cell transfer. Asagoe et al.10 demonstrated that BALB/c nude mice that have been adoptively transferred T cells from C57BL/6 mouse spleen develop a lethal acute GVHD. This mouse develops bodyweight loss, diarrhea, erythematous skin changes and erosions with positive Nikolsky’s sign. Because the target of the T cells in acute GVHD is MHC molecules, allo-specific T cells attack multiple organs distributed through the whole body. Thus, adoptive transfer is a simple and established method to induce cytotoxic immune response.

Animal model of TEN induced by drug antigen have not been established so far. The major target organs of this severe cADR is skin and mucosa including the cornea, although liver dysfunction and intestinal symptoms are sometimes observed. As previously reported, the effector cell of TEN was assumed to be CD8+ T cells (cytotoxic T cells) akin to acute GVHD,11 however, there was no direct evidence demonstrating that the epidermotropic cytotoxic T lymphocyte (CTL) is crucial for pathogenesis of TEN. By using an ovalbumin (OVA) model antigen system that never expresses in wild-type mice, we established an adoptive transfer model of TEN without using drugs. In K5-mOVA transgenic mice, OVA was expressed in the epidermal keratinocyte under the human keratin 5 promoter.12 As an epidermal self-antigen-specific T cell, OT-I transgenic mouse having MHC class I restricted OVA-specific T-cell receptor on CD8 T cells was used.13

OT-I cells proliferated in the skin-draining lymph node, infiltrated into the epidermis, and induced apoptosis of keratinocytes in the K5-mOVA mouse, however, the K5-mOVA mouse never developed TEN even when an extremely large number of OT-I cells were transferred. Interestingly, when OT-I cells were transferred into K5-mOVA on an athymic nude mouse background, a large sheet of epidermal detachment was induced (Fig. 1). Because athymic nude mice lack T cells that normally mature in the thymus, transferred T cells can proliferate rapidly by homeostatic expansion.

Figure 1.

 Adoptive transfer model and autoimmune disease model of toxic epidermal necrolysis. (a) Athymic K5-mOVA nude mouse transferred OT-I cells developed erosion with positive Nikolsky’s sign in dorsal skin and tail skin at day 7 after transfer. (b) Athymic K5-mOVA.OT-I double transgenic nude mouse spontaneously developed epidermal detachment on dorsal skin and tail skin at 9 weeks of age.

Autoimmune disease model of ten

Obviously, TEN is not an autoimmune disease. Nassif et al.11 demonstrated that keratinocyte added to the culprit drug, but not keratinocyte alone, is the target of drug-specific CTL from blister fluid of TEN patients. This result confirms that drug-reactive T cells are not autoreactive against the keratinocyte itself. However, we could also establish an animal model of TEN as an autoimmune disease.14 When K5-mOVA.OT-I double transgenic mice were generated in either euthymic mouse or athymic nude mice, only the athymic double transgenic mouse spontaneously developed a lethal exfoliative skin disease closely resembling human TEN at approximately 8–12 weeks of age (Figs 1,2). Interestingly, this athymic double transgenic mouse developed not only large sheets of detachment but also mucosal damage in the eyes, mouth and genital area. Because TEN was induced by a “native” transgenic T cell without adoptive transfer, an autoimmune disease model may be a more favorable model than an adoptive transfer model of TEN.

Figure 2.

 (a) Athymic K5-mOVA.OT-I double transgenic nude mice showed microscopic skin changes just before developing toxic epidermal necrolysis (TEN)-like devastating skin disease. Neither K5-mOVA or OT-I single transgenic athymic nude mouse developed any skin diseases. (b) Autoimmune disease model of TEN. Athymic K5-mOVA.OT-I nude mice developed large area of skin erosion at ∼8–12 weeks of age. Histopathologically, full thickness of epidermal necrosis and blister formation were observed (hematoxylin–eosin, original magnification ×200).

By analyzing the autoimmune model of TEN, we could find out a new pathomechanism of TEN that determines the severity of delayed drug hypersensitivity. Interestingly, a mouse having naturally-occurring regulatory T cell (Treg), named the euthymic K5-mOVA OT-I double transgenic mouse, did not develop erosion, whereas the mouse lacking CD4+CD25+Treg, named the athymic K5-mOVA OT-I double transgenic mouse, developed lethal large sheets of epidermal detachment clinically and histopathologically similar to TEN (Fig. 3). Surprisingly, adoptive transfer of OT-I cells separated from the euthymic K5-mOVA OT-I double transgenic mouse into the athymic K5-mOVA mouse lacking OT-I transgene induced TEN-like devastating skin damage, indicating that tolerance is induced on CTL by thymus-derived T cells. Furthermore, in vivo depletion of CD4+ T cells, including Treg, induced TEN in the euthymic K5-mOVA OT-I double transgenic mouse, suggesting that thymus-derived Treg prevents a runaway reaction of CTL that develops robust apoptosis of keratinocytes.

Figure 3.

 The differences between euthymic and athymic K5-mOVA.OT-I double transgenic mice. Thymus-derived regulatory T cells prevent expansion and activation of cytotoxic T lymphocytes causing toxic epidermal necrolysis.

Treg dysfunction in human ten

Clinically, lymphocytopenia is often seen in TEN patients. Roujeau et al.15 reported that CD4+ T-cell number is decreased in TEN patient, however, the number of CD8+ T cells is relatively maintained. Recently, Takahashi et al.16 nicely demonstrated a dysfunction in Treg of the patient peripheral blood of TEN. They showed that the percentage of Treg in TEN patients was within normal range at both the acute stage and resolution stage, although a drug-induced hypersensitivity syndrome patient showed an increased percentage of Treg. Very interestingly, the suppressive function of Treg is profoundly impaired in acute-stage TEN patients. These findings strongly support our findings in mouse TEN, and clearly demonstrated that loss of Treg function is a new exacerbating factor of cADR (Fig. 4).

Figure 4.

 Regulatory T cell (Treg) function may be a “turning point” between mild and severe delayed drug hypersensitivity. CTL, cytotoxic T lymphocytes; TEN, toxic epidermal necrolysis.

Cytolytic granules in ten

Our understanding of the effector mechanism of keratinocyte apoptosis in TEN has been dramatically changed in the last decade. Although inflammatory cells infiltrating into the epidermis are relatively few, robust apoptosis of keratinocytes is observed in the epidermis. This contrasting finding leads us to assume that keratinocyte apoptosis in TEN might be mediated by cytolytic granules without cell–cell contact. Indeed, several candidates for the cytotoxic molecules of TEN have been reported. Viard et al.17 showed that high concentration of soluble Fas-L is detected in the serum of TEN patients; epidermal keratinocytes of TEN patients expressed not only Fas, that is expressed in the normal keratinocytes, but also Fas-ligand (FasL). In this report, they demonstrated that pooled human intravenous immunoglobulins contained antibodies that can directly block Fas–FasL interaction. Posadas et al.18 demonstrated that mRNA of tumor necrosis factor (TNF)-α, perforin and granzyme B from blisters or peripheral blood mononuclear cells of acute-stage SJS/TEN patients were increased.

Recently, Chung et al.19 found that granulysin produced by CTL and natural killer (NK) cells can be detected in vesicles of SJS/TEN patients, and that granulysin concentration is 100–10 000-fold higher than in Fas-L and granzyme B. Interestingly, when granulysin is injected into the skin of a nude mouse or shaved C3H mouse, epidermal necrosis similar to TEN was induced within several hours. This model is also a very important animal model of TEN, demonstrating a final effector molecule produced by drug-reactive CTL or NK cell.

Role of antigen-presenting cell in ten

Delayed drug hypersensitivity reaction including TEN is induced by a drug-specific T cell.20 T cells recognize peptides presented in the context of MHC molecules on antigen-presenting cells (APC). Although the drug must be presented by certain APC in delayed drug hypersensitivity reaction, the APC responsible for drug-antigen presentation have not been identified yet. In contact dermatitis, APC activating a hapten-specific immune response are well studied. Because hapten applied onto the skin is presented by migratory dendritic cells (DC) that migrate from the skin to draining lymph node, then hapten-reactive T cells infiltrate into the skin where hapten originally existed and induce epidermal damage. On the contrary, the sensitization site of the culprit drug in delayed drug hypersensitivity reaction is still unknown. In delayed drug hypersensitivity reaction, most causative drugs are administrated p.o. or i.v., and spread into organs and tissues of the whole body through the bloodstream. However, it is still unclear whether p.o. or i.v. administrated medicine really binds to epidermal self-antigen like contact dermatitis.6

Before drug-reactive T-cell immunity attacks epidermal keratinocytes, many immunological steps, such as T-cell activation, extravasation and skin infiltration, are required. Indeed, the vast majority of patients who receive drugs never develop skin eruption, probably because activation of the T cells does not occur. DC are APC which can activate T cells very efficiently, and it has been reported that certain drugs induce maturation of DC.21 Unexpectedly, in our autoimmune disease model of TEN, a naive CD8+ T cell became CTL and induced epidermal damage without any exogenous trigger of inflammation or maturation stimuli. In other words, drug-induced maturation of DC seems to be not always necessary for inducing TEN.

When T cells migrate to the target organ (e.g. skin, intestinal tract) by responding to the foreign antigen, recognizing the destination where the foreign antigen is originally engulfed by DC is important for T cells to reach the inflammation site. If drug hypersensitivity is induced by epidermal self-antigen haptenized by causative drugs, skin-migratory DC may play a major role in the sensitization phase and elicitation phase. Recent mouse studies showed that epidermal self-antigen is presented by Langerin+CD103+ dermal DC, but not by epidermal Langerhans cell, that migrate from skin to the draining lymph node, indicating that dermal DC might be important for activation of the effector T cell of drug eruption.22,23


Animal models of TEN effectively help us to understand the immunological background of this life-threatening cADR, that is difficult to study in humans because of rare incidence. Especially, the importance of Treg in TEN was recently confirmed in human studies by great efforts using Treg separated from TEN patients. According to recent clinical and genetic studies, drug, virus infection and human leukocyte antigen (HLA) are known as the three major factors of severe drug eruption. Our animal model and the clinical study of Shiohara’s group strongly suggest that Treg function should be included in major factors of severe drug eruption. Although the animal model of TEN using drugs have not been established yet, further understanding of TEN for developing effective prophylaxis and treatment should be achieved by continuous efforts in this field through basic and clinical research.


This work was partly supported by Health and Labor Sciences Research Grants (Research on Intractable Diseases) from the Ministry of Health, Labor and Welfare of Japan