Mouse models of atopic dermatitis: a critical reappraisal

Mouse models for atopic dermatitis (AD) are an indispensable preclinical research tool for testing new candidate AD therapeutics and for interrogating AD pathobiology in vivo. In this Viewpoint, we delineate why, unfortunately, none of the currently available so-called "AD" mouse models satisfactorily reflect the clinical complexity of human AD, but imitate more "allergic" or "irriant" contact dermatitis conditions. This limits the predictive value of AD models for clinical outcomes of new tested candidate AD therapeutics and the instructiveness of mouse models for human AD pathophysiology research. Here, we propose to initiate a rational debate on the minimal criteria that a mouse model should meet in order to be considered relevant for human AD. We suggest that valid AD models should at least meet the following criteria: (a) an AD-like epidermal barrier defect with reduced filaggrin expression along with hyperproliferation, hyperplasia; (b) increased epidermal expression of thymic stromal lymphopoietin (TSLP), periostin and/or chemokines such as TARC (CCL17); (c) a characteristic dermal immune cell infiltrate with overexpression of some key cytokines such as IL-4, IL-13, IL-31 and IL-33; (d) distinctive "neurodermatitis" features (sensory skin hyperinnervation, defective beta-adrenergic signalling, neurogenic skin inflammation and triggering or aggravation of AD-like skin lesions by perceived stress); and (e) response of experimentally induced skin lesions to standard AD therapy. Finally, we delineate why humanized AD mouse models (human skin xenotransplants on SCID mice) offer a particularly promising preclinical research alternative to the currently available "AD" mouse models.