Expression of eotaxin, an eosinophil-selective chemokine, parallels eosinophil accumulation in the vesiculobullous stage of incontinentia pigmenti

Abstract

Incontinentia pigmenti (IP) is an X-linked dominant genodermatosis primarily affecting female children. The initial vesiculobullous stage of IP is characterized clinically by inflammatory papules, blisters, and pustules, and histopathologically by acanthosis, keratinocyte necrosis, epidermal spongiosis and massive epidermal eosinophil infiltration. The cause of this multisystem disease is attributed to the mutations of an X-linked regulatory gene, termed nuclear factor-κB essential modulator (NEMO). The exact mechanism of epidermal eosinophil accumulation has not yet been determined. We explored the possible role of an eosinophil-selective, nuclear factor-κB-activated chemokine, eotaxin, in the accumulation of eosinophils in the initial stage of the disease. Monoclonal antibody (6H9) specific for human eotaxin strongly labelled the suprabasal epidermis of IP skin, paralleling the upper epidermal accumulation of eosinophils, but did not label the epidermis of normal skin or lesional skin from patients with other inflammatory skin diseases not characterized by prominent eosinophil accumulation, namely dermatitis herpetiformis and selected cases of atopic dermatitis lacking significant numbers of eosinophils. In addition, endothelial cells in lesional skin of IP also exhibited strong expression of eotaxin, which correlated with perivascular and intravascular eosinophil infiltration. We also examined the in vitro effects on epidermally derived eotaxin of several cytokines that were nuclear factor-κB-activated and/or known to induce eotaxin expression. In normal human keratinocytes, proinflammatory cytokines either independently (IL-1α) or synergistically (tumour necrosis factor-alpha (TNF-α)/ interferon-gamma (IFN-γ) and TNF-α/IL-4) up-regulated eotaxin expression. These studies suggest that release of cytokines during the initial inflammatory stage of IP induces epidermal expression of eotaxin, which may play a role in the epidermal accumulation of eosinophils.

INTRODUCTION

Incontinentia pigmenti (IP) is an X-linked dominant multisystem disease with prominent cutaneous manifestations [1,2]. The typical cutaneous symptoms comprise the four classic stages of disease. The first stage of IP is manifested clinically by inflammatory papules, blisters, and pustules, and is usually called the vesiculobullous stage. This stage is histopathologically characterized by acanthosis, keratinocyte necrosis, epidermal spongiosis and massive epidermal eosinophil accumulation. Peripheral eosinophilia is usually associated, and may run as high as 70 000 eosinophils per mm³. The vesiculobullous stage occurs concurrently with or is followed by the hyperkeratotic, hyperpigmented and, in a minority of patients, atrophic stages. Genetic linkage studies map the ‘IP gene’ to the locus of the factor VIII gene region of the X chromosome (Xq28) [3–5]. Recently genetic studies demonstrated that the cause of IP is genomic rearrangement of an X-chromosome-encoded regulatory gene, termed nuclear factor-κB essential modulator (NEMO), which is also known as inhibitor κB kinase γ-subunit (IKKγ), located 200 kilobase proximal to the factor VIII locus [6]. Genetic disruption of this X-linked regulatory gene IKKγ in knockout mice results in massive liver apoptosis and midgestational death of the IKKγ^– male mice. The female mice heterozygous for this gene (IKKγ^+/–) survive and develop a clinical phenotype similar to IP in human patients, further supporting the role of this regulator gene in the development of IP [7,8].

Nevertheless, the mechanism by which eosinophils accumulate in IP is not yet fully understood. Investigations into the pathomechanism of IP uncovered an unidentified eosinophil chemotactic factor in the blister fluid of patients with IP and also in the extracts of their crusted scales [9,10]. Eotaxin, a CC class chemokine, is a recently identified eosinophil-selective chemokine [11–16]. Eotaxin is produced by specific leucocytes (including eosinophils, macrophages, and T cells) and some structural cells (including endothelial cells, fibroblasts, and epithelial cells) [16]. The chemoattractive capacities of eotaxin are more potent and more selective for eosinophils than any other known chemokines, for example RANTES (regulated on activation, normal T cell expressed and secreted) or macrophage inflammatory protein-1α[16]. Administration of eotaxin in vivo to guinea pigs, rodents, and primates results in strong and exclusive eosinophil recruitment, indicating the strong selectivity of eotaxin for eosinophils [16]. Eotaxin induces eosinophil chemotaxis and degranulation apparently through the activation of mitogen-activated protein kinases [17]. We have examined the possible role of eotaxin in epidermal eosinophil recruitment in IP.

MATERIALS AND METHODS

Patient selection

All IP patients included in our study had clinical findings and histopathological documentation that fulfilled the diagnostic criteria for IP [1,2]. The histology specimens from 13 IP patients were studied, including nine female patients and four male patients who were either mosaics or had Klinefelter’s syndrome (XXY). The initial blistering occurred at less than 2 weeks of age in all patients. Histological specimens from patients with bullous pemphigoid that showed prominent eosinophil infiltration served as positive controls (n = 3), while normal adult skin (n = 6) and neonatal foreskin (n = 2) were included as negative controls. To compare the findings observed in IP with those of inflammatory skin conditions, lesional skin of patients with dermatitis herpetiformis (n = 3) and selected cases of atopic dermatitis (without significant numbers of eosinophils, n = 7) were also included.

Immunohistochemistry

Immunohistochemical staining of eotaxin was carried out as previously described [18]. Sections (4 μm thick) were prepared from formalin-fixed paraffin-embedded lesional skin biopsies. After deparaffinization and rehydration, the sections were treated for antigen retrieval and the native peroxidase was removed by incubation of the sections in 0·03% H₂O₂ solution. After blocking with normal serum of the host animal used for the generation of the second antibody (1:10 diluted in 1% bovine serum albumin (BSA) in PBS), the sections were incubated with a monoclonal anti-eotaxin antibody (6H9) [13] diluted in 1% BSA in PBS at room temperature for 30 min. The second and subsequent immunohistochemical steps were carried out according to protocols of a commercially available Vector Elite ABC Universal kit (Vector Labs, Burlingame, CA). The immunoreactions were visualized with diaminobenzidine. The sections were counterstained with haematoxylin, mounted for microscopic examination, and photographed under an Olympus BX60 microscope equipped with an automatic camera system. Normal mouse serum was used as a negative control for the primary antibody.

Immunofluorescence

Immunofluorescence studies were performed as previously described [18]. Cryosections of lesional skin (6 μm thick) and normal skin were first incubated with monoclonal anti-eotaxin antibodies (6H9), followed by fluorescein-labelled goat anti-mouse IgG in one set and rhodamine-labelled goat anti-mouse IgG in another set. The sections were mounted with PBS:glycerol (1:1v/v) and examined under an Olympus BX60 immunofluorescence microscope equipped with epi-illumination and photographed.

Immunoblotting

In order to determine whether normal human skin epithelial cells can be induced to express eotaxin by proinflammatory cytokines, primary normal human cultured keratinocytes, derived from neonatal foreskin and grown in low-calcium, serum-free medium (SFM; Gibco-BRL, Gaithersburg, MD) were assessed [19]. The early confluent monolayer of keratinocytes was incubated for 12 h with or without addition of either recombinant human IL-1α (5 or 10 ng/ml; Sigma, St Louis, MO), recombinant human tumour necrosis factor-alpha (TNF-α) (10 ng/ml; PeproTech, Rocky Hill, NJ), interferon-gamma (IFN-γ) (10 ng/ml; PeproTech), IL-4 (20 ng/ml; PeproTech), a combination of TNF-α and IL-4, or a combination of TNF-α and IFN-γ. Cytosol was extracted in SDS sample buffer [20] and the protein concentrations were measured. Equal amounts of protein were loaded onto a 4% stacking gel over a 14% running gel (Novex, San Diego, CA) and the proteins were separated by SDS–PAGE under reducing conditions. The proteins were then transferred to nitrocellulose membranes, transfer verified by reversible Ponceau S staining (Sigma), blocked by 5% powdered milk, and reacted with rabbit antibody to human eotaxin (PeproTech). The immunoreaction was visualized by incubating the nitrocellulose membranes with peroxidase-labelled goat anti-rabbit antibody, enhanced chemiluminescence substrate (ECL; Dupont, NEN, Boston, MA), and then exposing to radiographic film (Kodak, Rochester, NY). Prokaryote-expressed recombinant human eotaxin (PeproTech) was included in the immunoblot as a positive control. The relative band density was determined by densitometric scanning. In an additional set of cultures, total RNA was extracted from the stimulated and unstimulated cultured keratinocytes and the concentrations measured for Northern blot analysis as described below [21].

Northern blot analysis

Northern blot hybridization was performed as previously described [20]. Equal amounts of total RNA (20 μg/lane) were size-fractionated, transferred to nylon membranes (Gene-Screen Plus; DuPont), and hybridized with a ³²P-labelled (random-primer labelling kit; BMB, Indianapolis, IN) human eotaxin cDNA [13].

RESULTS

Eotaxin is strongly expressed in the epidermis of IP lesional skin

In all IP lesional skin, epidermal eotaxin staining was detected, with moderate to strong intensity. By contrast, all normal skin (adult and neonatal) showed only negative or minimum epidermal eotaxin staining. Figure 1 depicts the immunohistochemical staining of eotaxin in IP lesional skin, in comparison with normal skin. The epidermis in the three different IP patients shown in Fig. 1 demonstrated strong expression of eotaxin (Fig. 1e,h,k), whereas such expression was not observed in the normal skin epidermis (Fig. 1b). Eotaxin staining was also observed in epidermal infiltrating cells, which were predominantly eosinophils. Normal mouse serum controls did not label the epidermis (Fig. 1a,d,g,j). In our study, we did not observe any significant eotaxin expression in the dermal fibroblasts of IP lesional skin or normal skin.

Fig. 1.

Expression of eotaxin is up-regulated in the subrabasal epidermis of lesional IP skin. Formalin-preserved and paraffinized skin sections from a normal individual (a–c), IP patient 1 (d–f), IP patient 2 (g–i), and IP patient 3 (j–l) were reacted with either normal mouse serum (a,d,g,j) or monoclonal anti-eotaxin antibodies (b,e,h,k), followed by a subsequent biotinylated secondary antibody reaction, streptavidin–peroxidase, and diaminobenzidine reaction. c,f,i and l were stained with haematoxylin and eosin. Bar = 25 μm (a–l).

Eotaxin expression is specific for eosinophil-predominant skin diseases

In order to demonstrate that the eotaxin expression in IP is specific for eosinophil-associated disease, and not a result of non-specific inflammation, immunohistochemical staining to detect eotaxin was performed with lesional skin of patients with bullous pemphigoid, atopic dermatitis and dermatitis herpetiformis. All of the selected cases of atopic dermatitis and dermatitis herpetiformis showed negative or minimum epidermal eotaxin staining, similar to that of normal skin. Figure 2 depicts the immunohistochemical staining of eotaxin in IP lesional skin, in comparison with lesional skin of bullous pemphigoid, atopic dermatitis and dermatitis herpetiformis. As illustrated, the lesional IP skin showed strong epidermal staining of eotaxin, throughout most of the epidermal layers (Fig. 2b). Lesional bullous pemphigoid skin, rich in eosinophils, showed strong epidermal staining of eotaxin, confined only to the basal epidermal layer (Fig. 2e). By contrast, the lesional atopic dermatitis (characterized by predominantly lymphocytic infiltration without a significant number of eosinophils, Fig. 2h) and dermatitis herpetiformis (characterized by predominantly neutrophilic infiltration, Fig. 2k) skin did not show significant epidermal or dermal eotaxin staining. Just as illustrated in Fig. 1, normal mouse serum did not stain epidermis or dermis (Fig. 2a,d,g,j). In order to confirm that the eotaxin staining in bullous pemphigoid lesional skin was restricted to basal keratinocytes, an immunofluorescence study was performed which verified the localization of eotaxin to basal keratinocytes (data not shown).

Fig. 2.

The up-regulated expression of eotaxin in IP skin is specific for eosinophil-prominent skin disease. Formalin-preserved and paraffinized skin sections from a patient with IP (a–c), a patient with bullous pemphigoid (d–f), a patient with atopic dermatitis (g–i), and a patient with dermatitis herpetiformis (j–l) were reacted with either normal mouse serum (a,d,g,j) or monoclonal anti-eotaxin antibodies (b,e,h,k), followed by a subsequent biotinylated secondary antibody reaction, streptavidin–peroxidase, and diaminobenzidine reaction. Sections c,f,i and l were stained with haematoxylin and eosin. Bar = 25 μm (a–l).

The location of epidermal eotaxin expression parallels the location of eosinophil accumulation

In order to examine closely the relationship between epidermal eotaxin expression and eosinophil accumulation, higher magnification photomicrographs of Fig. 1e,e and Fig. 2e,f were obtained (Fig. 3). As illustrated, the strong basal epidermal eotaxin expression in bullous pemphigoid lesional skin correlated with the location of basement membrane-infiltrating eosinophils (Fig. 3a,b). Similarly, the strong suprabasal eotaxin expression in IP lesional skin paralleled the upper epidermis-infiltrating eosinophils (Fig. 3c,d). The relationship between epidermal eotaxin expression and eosinophil accumulation in IP is further illustrated in Fig. 3e,f, which contains both peri-lesional and lesional skin from a single biopsy specimen. The peri-lesional skin showed moderate eotaxin staining, whereas the lesional skin of the same patient showed strong eotaxin staining.

Fig. 3.

The location of up-regulated epidermal eotaxin expression in IP skin parallels the location of eosinophil accumulation. Formalin-preserved and paraffinized skin sections from a patient with bullous pemphigoid (a,b) and a patient with IP (c–f) were either stained with haematoxylin and eosin (b,d,f) or reacted with monoclonal anti-eotaxin antibodies (a,c,e), followed by a subsequent biotinylated secondary antibody reaction, streptavidin–peroxidase, and diaminobenzidine reaction. Arrowheads point to skin basement membrane-localized eosinophils in bullous pemphigoid lesional skin (b). In e and f, lesional skin is on the left whereas peri-lesional skin is on the right. Bar = 15·3 μm (a–d), 61·5 μm (e,f).

Endothelial cells in IP lesional skin demonstrate strong expression of eotaxin

In order to examine whether endothelial cells might play a role in eosinophil recruitment in IP, endothelial cells in IP lesional skin were examined for eotaxin expression and inflammatory cell infiltration. As illustrated in Fig. 4, endothelial cells in IP lesional skin showed strong eotaxin staining (Fig. 4a) which paralleled perivascular and intravascular infiltration by eosinophils and mononuclear cells in the same patient (Fig. 4b). Most of these infiltrates of inflammatory cells also expressed eotaxin (Fig. 4a), as did endothelial cells from bullous pemphigoid lesional skin. Eotaxin expression was not detected in the endothelial cells of normal skin (data not shown), however.

Fig. 4.

The IP lesional skin demonstrates induced expression of endothelial cell eotaxin which parallels perivascular and intravascular infiltration of eosinophils. Formalin-preserved and paraffinized skin sections from a patient with IP were either stained with haematoxylin and eosin (b) or reacted with monoclonal anti-eotaxin antibodies (a), followed by a subsequent biotinylated secondary antibody reaction, streptavidin–peroxidase, and diaminobenzidine reaction. Arrowheads point to eotaxin-expressing endothelial cells (a) and perivascular and intravascular eosinophils (b). Bar = 12·3 μm (a,b).

Keratinocyte expression of eotaxin is induced in vitro by the proinflammatory cytokine IL-1α

Unstimulated, constitutive keratinocyte expression of eotaxin was barely detected. After incubation with 5 ng/ml and 10 ng/ml of IL-1α for 12 h., expression of eotaxin protein (9 kD) in the keratinocyte cytosol was increased by 1·75- and two-fold, respectively (Fig. 5). This IL-1-induced eotaxin expression in human keratinocytes was consistent with the reported findings in other structural cell types [11,22] and repeated experiments confirmed these findings.

Fig. 5.

Eotaxin expression in normal human keratinocytes is up-regulated by the proinflammatory cytokine IL-1α. Primary cultured human keratinocyte monolayers were either untreated or treated with recombinant human IL-1α (5 or 10 ng/ml) for 12 h. Equal amounts of cytosol proteins were vertically separated by SDS–PAGE, horizontally transferred to nitrocellulose membrane, and then immunoreacted with rabbit anti-human eotaxin antibodies (upper panel). The relative density of the immunoreactive eotaxin protein bands, depicted in the upper panel, is shown in the lower panel.

Keratinocyte eotaxin mRNA and protein expression is induced synergistically in vitro by TNF-α/IFN-γ and TNF-α/IL-4

After incubation with IFN-γ alone, IL-4 alone or TNF-α alone, no significant eotaxin mRNA or protein induction was detected, compared with the minimum protein expression in the untreated group. However, the combination of TNF-α and IFN-γ induced strong eotaxin mRNA and protein expression. The combination of TNF-α and IL-4 also induced eotaxin protein expression, although to a lesser extent (Fig. 6). This induction synergy by TNF-α in combination with IFN-γ or IL-4 in human keratinocytes was consistent with the reported findings in other structural cell types [11,22,23].

Fig. 6.

Eotaxin mRNA and protein in normal human keratinocytes are synergistically up-regulated by the proinflammatory cytokines TNF-α/IFN-γ and TNF-α/IL-4. Primary cultured human keratinocyte monolayers were either untreated or treated with recombinant human IFN-γ alone, IL-4 alone, TNF-α alone, a combination of TNF-α and IFN-γ or a combination of TNF-α and IL-4 overnight. Equal amounts of total RNA were separated by horizontal agarose gel, transferred to nylon membrane, and then hybridized with an eotaxin cDNA labelled with ³²P. Equal amounts of cytosol proteins were also vertically separated by SDS–PAGE, horizontally transferred to nitrocellulose membrane, and then immunoreacted with rabbit anti-human eotaxin antibodies. The relative quantity of RNA from each experimental group is illustrated by the ethidium bromide-stained and UV light-visualized 28S bands. Rec., Recombinant human eotaxin protein control. Although the resolution in some of the protein bands was not optimal, the location of the cellular eotaxin protein bands was confirmed by the recombinant eotaxin control.

DISCUSSION

In this study, we have demonstrated epidermal expression of eotaxin, an eosinophil-selective chemokine, in the epidermis of lesional IP skin and shown a correlation between eotaxin expression and epidermal accumulation of eosinophils. This epidermal expression of eotaxin was apparently specific for eosinophil-prominent skin diseases and was location-specific with regard to eosinophil recruitment. In our study, fibroblasts of lesional IP skin did not show prominent eotaxin expression, suggesting that this cell type is not significantly involved in the recruitment of epidermal eosinophils in IP [14]. Other investigators have demonstrated the enhanced expression of eotaxin primarily in the dermal infiltrating mononuclear cells in lesional atopic dermatitis characterized by eosinophil infiltration [24]. Our control samples of selected cases of atopic dermatitis did not contain significant numbers of eosinophils, and this may explain the absence of significant dermal expression of eotaxin in our samples (Fig. 2). Another interesting finding with regard to eotaxin expression was with the bullous pemphigoid control sample. Whereas in our samples eotaxin expression was restricted to the basal epidermis (confirmed by both immunohistochemical and immunofluorescence methods), other investigators have demonstrated expression throughout the entire epidermis [25]. The reason for this discrepancy is not known. The different antibodies used in these two projects might explain the differential staining patterns. The endothelial cells in the IP lesional skin exhibited induced expression of eotaxin, suggesting a possible role for the endothelium in eosinophil recruitment in IP. Our findings, however, did not suggest that eotaxin plays an exclusive or even major role in recruiting eosinophils to the epidermis of IP, as discussed below. Unfortunately, as IP is a rare disease, fresh skin or blister fluid samples are not readily available to allow the exact amount of eotaxin present in the skin of these patients to be measured by quantitative methods such as ELISA. One might hypothesize that in other inflammatory skin diseases in which eosinophils accumulate in the dermis, rather than in the epidermis, dermal fibroblasts may be a cellular source of eosinophil chemoattractant. One clinical example which seems to support this hypothesis is atopic dermatitis, in which the dermal fibroblast eotaxin expression parallels the dermal accumulation of eosinophils [24].

While the detailed mechanism of eosinophil accumulation in IP remains to be further elucidated, recently generated animal models of IP may shed light on the mechanism [7,8] Female mice heterozygous for IKKγ gene deficiencies (IKKγ^+/–) demonstrate a random X inactivation of IKKγ gene (in about 50% keratinocytes) and exhibit a multisystem disease similar to human IP [7,8]. The IKKγ gene is essential for the activation of a transcription factor termed nuclear factor-κB (NF-κB) which regulates multiple genes including chemokines, cytokines, adhesion molecules and enzymes that produce secondary inflammatory mediators. NF-κB is also essential for protecting cells from apoptosis caused by members of the TNF family of death cytokines and it negatively regulates keratinocyte proliferation [7]. In the mouse model of IP observed in heterozygous IKKγ^+/– female mice, skin expression of numerous chemokines and cytokines is increased, including eotaxin IL-1α, IL-1β, TNF-α, IFN-γ, RANTES, macrophage inflammatory proteins-1α, -1β, -2, IP-10, monocyte chemoattractant protein (MCP)-1, lymphotactin, TCA-3, and transforming growth factor-beta1 and beta2. Most of these cytokines and chemokines are known to be regulated by NF-κB [26,27] and are probably produced by the adjacent IKKγ⁺ cells, since NF-κB is not activated in IKKγ^– cells [7].

Eotaxin expression by keratinocytes may be up-regulated by cytokines that show increased expression in the mouse model of IP. The expression of eotaxin in normal human lung epithelial cells is up-regulated by either IL-1α or TNF-α in a time- and dose-dependent manner [22]. Furthermore, the expression of eotaxin in normal human keratinocytes is up-regulated by TNF-α, particularly in combination with the T helper cell subset 2 cytokines of either IL-4 or IL-13 [23]. This TNF-α augmentation of IL-13-induced eotaxin production is also observed in normal human lung epithelial cells [28]. Moreover, the eotaxin promoter in mice and humans has an NF-κB binding site, a binding element for a T helper cell subset 2 mediator termed STAT-6 (signal transducer and activator of transcription-6), an IFN-γ-response element and a glucocorticoid response element. These observations may account for the observed regulation of eotaxin expression by TNF-α, IL-4, IFN-γ and glucocorticoids, respectively [16,29]. We also demonstrated that eotaxin expression by primary cultured human keratinocytes was induced by the proinflammatory cytokines IL-1α[30], TNF-α/IL-4, and TNF-α/IFN-γ (Figs 5 and 6). Additionally, human endothelium eotaxin expression is also induced by IL-1α and TNF-α[11,16].

Based on the animal model, a mechanism has been proposed to explain the inflammatory reactions that occur during the initial stage of IP [7]. With our additional finding of increased epidermal and endothelial eotaxin expression that parallels eosinophil accumulation, we propose the following mechanism, as diagrammatically illustrated in Fig. 7. In the IKKγ^– keratinocytes, NF-κB remains inactivated. The inactivation of NF-κB leads to abnormal epidermal hyperproliferation, since NF-κB negatively regulates keratinocyte proliferation. Keratinocyte apoptosis and necrosis then follow as a result of abnormal keratinocyte proliferation. The necrotic keratinocytes then act as an activating signal to neighbouring IKKγ⁺ keratinocytes, leading to activation of NF-κB and subsequent inflammatory responses. This cascade results in the induced synthesis and release of chemokines like RANTES [31], MCP-1 [32], and eotaxin. In addition, the release of cytokines such as IL-1 and TNF-α can further stimulate eotaxin production by keratinocytes and endothelial cells. The released eotaxin, along with RANTES and MCP-1, recruits eosinophils to the epidermis. Eosinophilic degranulation and release of proteases, along with intercellular oedema (spongiosis), lead to the blistering process of the first stage of IP [33]. The death cytokines (such as TNF-α) produced by the IKKγ⁺ keratinocytes then kill the IKKγ^– cells, perhaps accounting for the transient nature of the first stage of IP. If a few IKKγ⁻ cells escape this elimination process and survive, a second episode of keratinocyte hyperproliferation, followed by an inflammatory response, can be initiated. This may explain the occurrence of second episodes of the first stage in some human patients with IP [1,34]. The observation of the lessened severity and duration of the second episode is also consistent with this proposed mechanism since there would be fewer IKKγ⁻ keratinocytes, which survive from the inflammation of the first episode, to induce the inflammatory response of the second episode [1]. Eosinophil recruitment in IP may also involve the participation of other cytokines, such as the differentiation and maturation of eosinophils in bone marrow by IL-3, IL-5, and granulocyte-macrophage colony-stimulating factor and eosinophil migration from the bone marrow to the circulation by IL-5, leading to the peripheral eosinophilia in infants with IP [16].

Fig. 7.

Proposed mechanisms of intraepidermal eosinophil accumulation and blister formation in IP. Solid lines represent intracellular signal/action pathways. Dotted lines represent extra-cellular signal/action pathways. This schematic presentation is derived in part from: Makris C, Godfrey VL, Krahn-Senftleben G et al. Female mice heterozygous for IKK gamma/NEMO deficiencies develop a dermatopathy similar to the human X-linked disorder incontinentia pigmenti. Mol Cell 2000; 5: 969–79.

Lastly, if normal human keratinocytes in fact produce and secrete eotaxin in vivo in response to IL-1α, combined TNF-α/IL-4 and combined TNF-α/INF-γ, one might propose that these cytokines may also be involved in other skin diseases such as erythema toxicum neonatorum, pemphigus vegetans, and eosinophilic pustular folliculitis, in which prominent epidermal accumulation of eosinophils is a histological hallmark. Although at the present time no experimental data are available to explain the cause of eosinophil accumulation in these diseases, these are certainly interesting questions to be answered in the future.