The histology of atopic dermatitis (AD) is characterized by epidermal alterations and a dermal inflammatory infiltrate containing eosinophils. Although tissue eosinophilia is not striking in AD, infiltrating eosinophils together with other inflammatory cells are suggestive for an allergic reaction similar to that seen in bronchial asthma, allergic rhinitis, or in allergic gastrointestinal diseases. In this review, we summarize our current knowledge regarding the mechanisms of eosinophil skin infiltration as well as the potential role of eosinophils in the pathogenesis of AD.
In spite of the progress regarding the description of immunological phenomena associated with atopic dermatitis (AD), the pathogenesis of this disease still remains unclear. The presence of eosinophils in the inflammatory infiltrate of AD has long been established. Eosinophil numbers as well as eosinophil granule protein levels in peripheral blood are elevated in most AD patients and appear to correlate with disease activity. Moreover, eosinophil granule proteins, which possess cytotoxic activity, are deposited in the skin lesions. These observations indicate a role of eosinophils in the pathogenesis of AD. Furthermore, AD is associated with increased production of T helper 2 cytokines including interleukin (IL)-5, which specifically acts on eosinophils, resulting in accelerated eosinophilopoiesis, chemotaxis, cell activation, and delayed apoptosis. Therefore, IL-5 is an interesting target for experimental therapy in this inflammatory disorder of the skin. Such studies might result in new insights into the pathogenetic role of eosinophils in AD.
Eosinophils and eosinophil-derived proteins in AD
Eosinophils are typically characterized by a bilobar nucleus with highly condensed chromatin and cytoplasm containing two major types of granules, specific and primary granules, and lipid bodies. Specific granules contain a number of cationic proteins that give the eosinophils the unique staining properties. About 120 years ago, Paul Ehrlich described the affinity of a coarsely granular leukocyte for the acid dye eosin and called this cell eosinophil. Because of the characteristic staining property, eosinophils can also be detected in paraformaldehyde fixed and paraffine embedded tissues. Under normal conditions, eosinophils are almost exclusively limited to the digestive tract (1, 2) and are not present in most other tissues including the dermis. Eosinophils are however part of the mixed perivascular inflammatory infiltrate within the dermis of AD patients (3).
Skin eosinophil numbers are particularly increased in patients with onset of AD before adulthood (4). In adults with AD, eosinophil skin infiltration is often only modest (5), but tissue eosinophilia as well as eosinophilic granule protein deposition can be found in nearly all biopsies of AD lesions with a median eosinophil count of 2.8 cells/mm2 (range 0–90.3) (6). The maximum deposits of eosinophilic granule proteins are located in the upper dermis, 0.47–0.93 mm from the epidermis, whereas in the lower dermis, below a depth of 1.39 mm, no deposition can be detected. Tissue eosinophilia has been shown to be a feature in both acute and chronic stages of AD and it correlates with disease severity. In chronic AD, eosinophilia appears to be more pronounced in lesions with marked epidermal hyperplasia compared with those with no or slight hyperplasia. Moreover, a correlation between eosinophilia and the degree of spongiosis was noticed in acute dermatitis or acute exacerbations of chronic AD (6). Moreover, in a mouse model of AD, tissue eosinophilia correlated with an increase in the thickness of the epidermal and dermal layers and skin hypertrophy was suggested to result from repair processes following cytotoxic effects of eosinophil major basic protein (MBP) or eosinophil cationic protein (ECP) (7).
In addition to tissue eosinophilia, blood eosinophilia is present in most patients with AD correlating roughly with the severity (8). Blood eosinophilia was described to be more pronounced if the AD was associated with respiratory allergic diseases (9) as well as in patients with extrinsic AD compared with those with intrinsic AD (10). As some patients exhibit normal blood eosinophil counts despite active AD and because increased eosinophil numbers might be the consequence of additional allergic disorders, the determination of eosinophil number in blood is not a reliable tool in establishing the diagnosis of AD.
Besides eosinophils, eosinophil-derived products are present in increased amounts in the blood and the skin of AD patients. In particular, the basic proteins ECP, eosinophil-derived neurotoxin (EDN, EPX), and MBP have been analyzed in clinical studies. Although EDN and ECP might also be synthesized in small amounts by neutrophils (11), all these proteins can be considered as specific eosinophil proteins in most clinical and experimental settings. They are usually detected by immunoassays using specific monoclonal antibodies. For instance, eosinophils and release phenomena in tissues can be analyzed using immunohistochemistry or immunofluorescence techniques, and eosinophil activation in blood can be measured with a modified enzyme-linked immunosorbent assay technique.
The measurement of ECP in serum is a frequently used tool in monitoring AD activity (12). The decline of serum ECP level correlates with clinical improvement upon corticosteroid (8, 13, 14), cyclosporin A (15), and interferon-γ (16) therapies. In addition, successful high-altitude climate therapy (17) and UVA1 phototherapy (18) are also associated with decreasing levels of serum ECP. However, serum ECP levels do not correlate with total immunoglobulin (Ig)E levels (19). Moreover, ECP levels are found elevated in both extrinsic and intrinsic type of AD, and so do not differ between these two groups (8). Besides ECP, serum EDN (14), serum MBP (20), and urine EPN levels (21–23) have also been used as markers for monitoring AD activity.
In the absence of eosinophilic-specific surface markers (24), MBP and ECP have also been popular molecular targets in immunohistochemical studies of skin from patients with AD. These studies demonstrated that eosinophil granule proteins are not only present inside eosinophils but also in the extracellular spaces, suggesting eosinophil degranulation. Extracellular MBP deposition is primarily localized in the upper dermis and was detected in all biopsies obtained from patients with chronic AD lesions (25). Another striking observation of this study was the near absence of intact eosinophils in the presence of extensive extracellular MBP staining. Intact eosinophils, however, were located predominantly within the perivascular mononuclear cell infiltrate. Interestingly, dermal eosinophil granule protein deposits have also been observed during the cutaneous late phase reaction that preceded the maximal expression of clinical symptoms (26). The presence of mostly disrupted eosinophils in the dermis of AD patients was confirmed by an electron-microscopy study, in which disrupted eosinophils and/or free eosinophil granules were detected in seven of 10 specimens (27, 28). Various degrees of eosinophil degeneration were observed ranging from intact eosinophils with granule abnormalities, to intact eosinophils with abnormal granules and pseudopod-like extensions, to eosinophils with degenerated cell and/or nuclear membranes to free eosinophil granules in proximity to, or in the absence of eosinophils. It remains to be investigated how the eosinophil cytolysis in AD is initiated.
Taken together, there is clear evidence for eosinophil infiltration and activation in AD skin lesions. In experimental models, the eosinophils are present before clinical symptoms occur. Severe disease correlates with elevated eosinophil numbers and increased release of eosinophil-derived proteins. Clinical improvement because of therapeutic interventions is associated with markedly reduced eosinophilic inflammation. Although these observations make it likely that the eosinophil plays an important pathogenic role in AD, its exact function(s) remains to be determined.
Molecular mechanisms contributing to eosinophilic responses
Increased eosinophil production
Eosinophils are derived from a CD34+ hematopoietic progenitor cell in the bone marrow. Eosinophils share this progenitor with basophils, defined as the eosinophil/basophil-colony forming unit (Eo/B-CFU) (29). In the peripheral blood of atopic individuals, the Eo/B-CFU is elevated and correlates with the severity of the atopic disease. Allergen exposure of patients with allergic rhinitis during the pollen season causes a decline in the number of eosinophil/basophil progenitors in the peripheral blood suggesting that these progenitors are trafficking through the peripheral blood into the tissue, where they mature (30). Moreover, the α-subunit of the interleukin-5 receptor (IL-5Rα) is up-regulated on bone marrow CD34+ progenitors after allergen challenge (31), indicating increased sensitivity towards the eosinophil differentiation factor IL-5 after allergen exposure.
The importance of IL-5 for the generation of eosinophils is evident from studies of IL-5-deficient mice, which were unable to develop eosinophilia upon allergen sensitization and challenge. On the contrary, IL-5-transgenic mice exhibited large eosinophil production in the bone marrow, and tissue eosinophilia in multiple organs (32, 33). Besides IL-5, the cytokines IL-3 and granulocyte/macrophage colony-stimulating factor (GM-CSF) have also been shown to stimulate eosinophil production in the bone marrow (34). An experimental mouse model of allergic rhinitis indicated that even an apparently isolated allergic response within the nasal mucosa is associated with increased progenitor cell production in the bone marrow, resulting in an IL-5-dependent increase in eosinophil and basophil numbers (35). In conclusion, there is evidence from human and mouse studies that accelerated eosinophilopoiesis plays a critical role during allergic eosinophilic responses.
Preferential eosinophil recruitment
Under physiological conditions, eosinophils are located in the gastrointestinal tract but not in other tissues (2). Eosinophil mobilization from the bone marrow was suggested to be under the control of IL-5 and eotaxin, an important chemotactic factor for eosinophils, in a selective and concentration dependent manner (36, 37). The question arises, what triggers the eosinophils to migrate into tissues, which they usually do not enter? This appears to be quite complex and many groups are performing intense research in this field. Under normal conditions, the lumenal surface of blood vessels does not express sufficient levels of adhesion molecules to allow leukocytes to adhere. However, their expression can be enhanced by cytokines such as IL-1, IL-4, and tumor necrosis factor (TNF) at the sites of allergic inflammation. Animal studies have demonstrated that IL-1 and TNF receptor expression on endothelial cells is important in both mediating eosinophil rolling and adhesion to the endothelium (38). The IL-1β release has been detected at sites of allergic reactions in the skin (39), and the cutaneous late phase reaction can be inhibited by soluble IL-1 receptors (40).
Which are the adhesion molecules responsible for the sequential events rolling, adhesion, and transmigration of eosinophils into allergic tissues? The following molecules have been identified as important players: E-selectin, P-selectin, intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1 as well as the corresponding ligands on eosinophils including L-selectin, P-selectin glycoprotein (PSGL)-1 and integrins (α4β1 and α4β7 integrins) (38). In the skin of AD the number of eosinophils as well as the deposits of eosinophil granule proteins correlate with VCAM-1 expression and were found to be more pronounced in acute lesions than in chronic lesions (41). Also other studies suggested a key role of VCAM-1 for the specific recruitment of eosinophils into inflamed tissues (42–46) (Fig. 1).
Several chemokines important for eosinophil recruitment have been demonstrated to be expressed at sites of allergic inflammation. For instance, eotaxin and regulated upon activation, normal T-cell expressed and secreted (RANTES) are two important chemotactic factors for eosinophils and largely contribute to the movement of eosinophils from the blood to the sites of inflammation (47). The complement anaphylatoxins C3a and C5a have also been implicated in eosinophil recruitment (48). Moreover, leukotrienes and prostaglandins, in particular LTB4 and PGD2, were found to induce eosinophil chemotaxis (49, 50), whereas lipoxin A4 blocked eosinophil trafficking (51). In contrast to eotaxin, IL-5 alone has only weak chemotactic activity, but might increase eotaxin-mediated chemotaxis (38). The T helper 2 (Th2) cytokines, IL-4, IL-13, and IL-9 are believed to promote eosinophilia by regulating local IL-5 and/or eotaxin synthesis and/or by suppressing IFN-γ production. For instance, IL-4 induces eotaxin mRNA expression in dermal fibroblasts in a dose- and time-dependent manner supporting this concept (52). Moreover, Th2 cytokines induce the expression of the adhesion molecules VCAM-1 and PSGL-1 that may additionally support eosinophil influx into allergic tissues. In contrast, IL-6, IL-11 and IL-12 seem to inhibit Th2 cytokine expression, VCAM-1 expression, and eosinophilia (53–55).
Additional studies revealed that eotaxin (56) and monocyte chemotactic protein (MCP)-3 (57) are important in the early recruitment of eosinophils following allergen challenge, whereas at later time points eosinophil recruitment is largely eotaxin-independent and chemokines such as RANTES, MCP-5, and MIP-1α play important roles (58). Several CXC chemokines such as CXCL9, CXCL10, and CXCL12 have also been shown to induce eosinophil chemotaxis (59). The biological effects of chemokines are mediated by their interaction with specific receptors that belong to the seven-transmembrane G-protein-coupled receptors (60). The principal receptor involved in eosinophil attraction seems to be CCR-3 (61). The major ligands for CCR-3 are eotaxin, eotaxin-2, eotaxin-3, RANTES, MCP-2, MCP-3, and MCP-4 in humans (62, 63).
Eotaxin levels are increased in the blood from AD patients (64) and eotaxin is expressed in human AD skin lesions (65). In a mouse model of AD, CCR-3 was found to be essential for eosinophil recruitment into the skin at sites of repeated antigen sensitization with ovalbumin and into the lung (66). After transmigration of blood vessels, eosinophils enter the extracellular matrix, where they bind to matrix proteins such as fibronectin. This binding is mediated by VLA-4 on eosinophils. Eotaxin decreases the affinity of eosinophil-expressed VLA-4 to its counterligand, the CS-1 region of fibronectin (67). This de-adhesion seems to be a prerequisite for further tissue migration, in which metalloproteases are involved (68).
Intradermal injection of eotaxin and eotaxin-2 has been shown to cause an eosinophil infiltrate within 1 h, which further increased after 6 and 24 h. Surprisingly, eotaxin also recruited neutrophils and macrophages into the skin of atopic and nonatopic individuals (69). A similar fast recruitment of eosinophils was seen in sensitized AD patients following positive patch testing. The influx of eosinophils into the dermis started within 2 h and reached its maximum 6–24 h after patch testing (70). A quantification of the infiltrating cells in positive patch test reactions revealed 9% eosinophils (71). Moreover, atopy patch testing with house dust mite allergens was performed in AD patients sensitized to house dust mites. Eosinophils were detected in postcapillary venules in the dermis 2 h upon allergen challenge, followed by eosinophil infiltration at 6 h, which peaked at 24 and 48 h. The adhesion molecules E-selectin and ICAM-1 were upregulated as eosinophils increased in numbers (72). The rapid recruitment of eosinophils into the skin of patients is also reflected by the fact that eosinophil numbers in blood can decrease following allergen challenge (73). Taken together, the recruitment of eosinophils upon an adequate trigger is rapid and involves the increased expression of adhesion molecules including VCAM-1 and also involves specific chemotactic factors such as eotaxin (Fig. 2).
Delayed eosinophil apoptosis
Eosinophilia and high IL-5 expression are often associated in chronic allergic diseases such as bronchial asthma or AD. Beside increased production of eosinophils, inhibition of eosinophil apoptosis by IL-5 appears to play an important role at sites of allergic inflammation (74). Delayed eosinophil apoptosis as a mechanism of tissue eosinophil accumulation has been demonstrated in nasal polyps (75). In addition to IL-5, IL-3 and GM-CSF are also known to increase eosinophil viability in vitro (76, 77). Recently, CCR3-reactive chemokines such as eotaxins have been demonstrated to prolong eosinophil survival (78).
Purified blood eosinophils from AD patients that were cultured ex vivo had a reduced death kinetic compared with normal eosinophils (79). This observation might reflect that eosinophils were exposed to survival factors in vivo before isolation. However, the intracellular mechanisms, which mediate increased in vitro survival in the absence of survival cytokines remain to be investigated. It is possible that anti-apoptotic proteins of the Bcl-2 family play a role (80–82). Eosinophils from AD patients have also been demonstrated to be resistant to Fas-induced apoptosis, a phenomenon, which was not related to decreased Fas receptor surface expression (83). Although additional mechanisms might play a role (84), these data support the idea that eosinophils from AD patients express increased amounts of anti-apoptotic proteins.
All the mentioned studies in AD were performed using blood eosinophils. Whether delayed eosinophil apoptosis occurs in the skin of AD patients has not been demonstrated. As eosinophil cytolysis was present in about 70% of the cases (27), it is possible that unknown death triggers operate in the skin of these patients and kill the cells even in the presence of increased amounts of anti-apoptotic proteins. Figure 3 demonstrates two patients with AD. In one patient the phenomenon of eosinophil degeneration is suggestive, whereas in the second patient normal eosinophils are present. Clearly, it would be interesting to precisely dissect these two patient groups in terms of pathogenic mechanisms and clinical characteristics.
Potential pathogenic roles of eosinophils in AD
Mechanisms of activation
As mentioned earlier in this article, the release of eosinophil basic proteins in the context of an inflammatory response suggests eosinophil activation. As the eosinophil granule proteins may also cause tissue damage, the process of eosinophil activation needs to be tightly controlled. There have been a number of studies describing eosinophil activation mechanisms. Hematopoietins, such as IL-3, IL-5, and GM-CSF (85) increase functional responses of eosinophils to various agonists, including lipid mediators, complement factors, and chemokines (85–88). This effect of hematopoietins, called ‘priming’, is also observed in other granulocyte subtypes (89). Recently, IL-2 has also been reported to act as a priming factor in CD25+ eosinophils (90). That effector cells of the immune system do not usually release toxic proteins immediately upon stimulation might be part of a safety mechanism, which prevents accidental degranulation.
Second signals that trigger functional responses after previous priming of eosinophils are provided by a number of various agonists, including lipid mediators, complement factors and chemokines (85, 91). The same factors might also release preformed cytokines from eosinophils (92, 93). According to their physiological function in host defense surveillance of mucosal surfaces, eosinophils can also be activated by IgG, IgA or soluble IgA (94). Although the high-affinity IgE receptor had been proposed as stimulus for activation (95), recent studies did not obtain evidence for functional high-affinity IgE receptors on human eosinophils (96, 97).
Eosinophils express a variety of receptors for immunoglobulins, cytokines, chemokines and other chemotactic factors that upon activation result in degranulation and the release of inflammatory mediators, e.g. cationic proteins, leukotrienes, and immunoregulatory cytokines (47, 68). In particular, because of the production and release of cytokines, eosinophils appear to have immunoregulatory properties. Although eosinophils are terminally differentiated cells, their capacity to generate cytokines can be quite intriguing. Eosinophils are able to produce a wide spectrum of cytokines, including TNF, transforming growth factor, IL-1, IL-3, IL-4, IL-5, IL-8, and GM-CSF (68). By secretion of these cytokines eosinophils are capable of enhancing the inflammatory processes, including T-cell differentiation, but also of initiating tissue repair processes. The IL-13 was recently described to be expressed by peripheral blood eosinophils derived from patients with atopic diseases, including AD, and to be released upon stimulation with eotaxin (93). Eosinophil-derived IL-13 was functional, because it increased the surface expression of the low affinity IgE receptor on purified B cells. Besides IL-13, RANTES and eotaxin have also been shown to be able to release IL-4 from eosinophils (98). In addition, eosinophils are able to produce IL-10 (99).
Besides cytokines, eosinophils contain lipid bodies, which play a role in the generation of eicosanoid mediators (100). They are a major source of the cysteinyl leukotriene LTC4 and its active metabolites LTD4 and LTE4. The generation of LTC4 by blood eosinophils of AD patients was enhanced when compared with healthy controls and did not depend on the presence or absence of associated bronchial asthma (101). In asthma, a specific polymorphism within the promoter of the LTC4 synthase has been identified that may contribute to increased LTC4 synthesis (102). Released leukotrienes may amplify the inflammatory cascade, for instance by acting as chemotactic factors or by triggering the release of cytotoxic proteins. Eosinophils have also been described as antigen-presenting cells. However, compared with professional antigen-presenting cells, they are relatively inefficient in activating T cells (103).
Eosinophils or factors participating in the development of eosinophilia, such as chemokines and cytokines, in particular IL-5, are interesting therapeutic targets in AD. Glucocorticoids, cyclosporin A and tacrolimus significantly inhibit IL-5 production by peripheral blood mononuclear cells from atopic patients (104). Glucocorticoids have been shown to suppress IL-5 synthesis by targeting CD4+ T cells activated via the T-cell receptor or by IL-2 (105). The reduction of IL-5 expression by corticosteroids was associated with both reduced eosinophil production and increased eosinophil apoptosis (106–108). Cyclosporin A and tacrolimus have also been shown to be clinically effective in AD. Both drugs inhibit cytokine production of T cells, including IL-5, confirming the critical role of IL-5 and T cells in the pathogenesis of AD (10, 109). In a recent study, a decrease of eosinophils and Th2 cytokine expressing T cells in lesional skin after treatment with topical tacrolimus was observed (110). Also IFN-γ therapy was reported to decrease IL-5 levels in the peripheral blood of AD patients (111). The beneficial therapeutic effect of phototherapy such as UVA irradiation is associated with a marked decrease of CD4+ T cells and eosinophils (112).
Because of the pivotal role of IL-5 for eosinophilia and its selective activity on eosinophils and basophils, specific neutralization of this cytokine is a promising strategy in the treatment of eosinophilic diseases (113). For instance, the administration of a neutralizing anti-IL-5 antibody to ovalbumin-sensitized mice during repeated allergen challenge prevented the development of airway hyperresponsiveness (114). In patients with bronchial asthma, the therapy with an anti-IL-5 antibody was associated with a reduction of eosinophil numbers in both blood and sputum, although it had no effect on bronchial hyperreactivity (115). In contrast, anti-IL-5 antibody therapy was therapeutically effective in patients with eosinophilic dermatitis (116). In order to rule out exactly the role of eosinophils in AD pathogenesis, we believe that an experimental trial using the anti-IL-5 antibody as a tool to decrease eosinophil numbers in warranted. Such trial might result in new insights into the role of eosinophils in AD.