Asthma is often preceded by atopic dermatitis (AD), a phenomenon known as ‘atopic march’ [1, 2]. It has been recognized that sensitization to common inhalant allergens is developed in a majority of patients with AD, often during the course of AD [1, 3]. Moreover, an early onset and the severity of AD are associated with the degree of sensitization to aeroallergens, and with the risk of developing asthma . Although these studies have suggested that AD skin may deliver signal(s) that could promote sensitization to inhalant allergens , thus triggering the onset of ‘atopic march’, the involved mechanism remains to be clarified.
Thymic stromal lymphopoietin (TSLP) produced by skin keratinocytes has emerged as a key player in AD pathogenesis [4, 5]. Increased expression of TSLP has been revealed in epidermal keratinocytes from patients with AD , and elevated serum TSLP level was reported in children with AD . TSLP overproduction has been also reported in the skin of patients with Netherton disease, which is highly associated with AD and atopy . Our previous studies using mouse models have shown that TSLP overexpression in keratinocytes not only triggers a spontaneous AD-like dermatitis, but also aggravates experimental allergic asthma induced upon intraperitoneal (i.p.) sensitization to ovalbumin (OVA) and airway OVA challenge . This was supported by the study of Demehri, et al. . Importantly, employing a TSLPover protocol in which TSLP expression in keratinocytes is induced by skin topical application of MC903 (a vitamin D3 analog) [11, 12], we showed that overproduction of keratinocytic TSLP during the i.p. sensitization phase could aggravate OVA asthma . This study suggested that TSLP produced by AD skin enters the circulation and acts as a skin-derived signal to promote allergen sensitization, thereby contributing to the ‘atopic march’ . However, it remained unclear whether these findings have clinical relevance. Indeed, the i.p. sensitization of OVA complexed with exogenous Th2 adjuvant (alum) is far from a ‘real-life’ situation, and moreover, OVA is not a natural inhalant allergen, and it does not prime airway sensitization the same way as house dust mites (HDM), pollens and other aeroallergens do in human asthma [13, 14].
We thus set off to determine whether TSLP overproduced in AD skin could impact sensitization to inhaled aeroallergens and thereby triggers the development of allergic asthma. The experimental protocol is outlined in Fig. 1A, in which an initial sensitization via intranasal (i.n.) exposure of the clinically relevant aeroallergen HDM, followed by i.n. challenge of HDM 3 weeks after. This protocol allowed us to temporally dissect the phases of airway sensitization and challenge. To effectively reveal the effect of skin-derived TSLP on airway sensitization, we used here a ‘sub-asthmatic’ mild HDM dose (0.5 μg for sensitization and 0.05 μg HDM for challenge), which induced in wild-type Bab/c mice a weak sensitization to HDM (see Fig. 1B), and generated a very mild allergic asthma upon i.n. HDM challenge (see Fig. 2A–F). During the phase of airway sensitization, TSLP production in skin keratinocytes was induced by MC903 topical application on mouse ears  (Fig. 1A). Note that MC903 treatment induced TSLP production in the skin, with simultaneously increased serum TSLP levels , whereas ending the MC903 treatment normalized skin and serum TSLP levels within 3 days [ref  and data not shown].
We first examined the effect of skin MC903 treatment on the sensitization to HDM. At D25 (a time-preceding airway challenge), the i.n.HDM + MC (i.n. HDM-sensitized and skin MC903 treated) mice exhibited much higher HDM-specific IgE and IgG1 in sera, compared with i.n. HDM + EtOH (i.n.HDM-sensitized and skin EtOH treated) mice (Fig. 1B). In vitro HDM stimulation of splenocytes from these mice induced much higher production of Th2 cytokines (IL4 and IL13), as well as Th17 cytokine (IL17), but not Th1 cytokine (IFNγ) (Fig. 1C). Therefore, skin topical MC903 application enhances Th2 and Th17 sensitization to inhaled HDM.
We then analyzed pulmonary inflammation upon HDM challenge. Whereas i.n.HDM + EtOH/i.n.HDM mice exhibited only a mild airway inflammation, the i.n.HDM + MC/i.n.HDM mice developed a robust asthmatic phenotype at D35. Indeed, bronchoalveolar lavage (BAL) cell counting showed a significant increase of eosinophils, neutrophils and lymphocytes in the i.n.HDM + MC/i.n.HDM mice (Fig. 2A). These mice exhibited much more severe pulmonary infiltration of eosinophils (Fig. 2B) and CD4+ T cells (but not CD8+ T cells, data not shown), as well as stronger mucus production (Fig. 2C). They produced much higher HDM-specific IgE and IgG1 in sera (Fig. 2D). RT-PCR analyses of lungs of these mice showed that the expression of Th2 cytokines (IL4, IL5, IL13), eosinophil-attractant chemokine (eotaxin-2), and receptor (CCR3) was all increased. The expression of Th17 cytokine (IL17A) showed a tendency to increase (although the statistic analyses did not reach significance), whereas Th1 cytokine (IFNγ) expression remained unchanged (Fig. 2E). Moreover, the i.n.HDM + MC/i.n.HDM mice also exhibited an enhanced airway hyperresponsiveness (Fig. 2F). Notably, in the absence of HDM sensitization, MC903 did not induce any lung inflammation upon i.n. HDM challenge (see i.n.sal + MC/i.n.HDM in Fig. 2A–F). It suggests that the promotion of allergic asthma by skin MC903 requires the copresence of allergen sensitizing. Together, these results indicate that MC903 application in the skin during the HDM airway sensitization phase triggers an aggravation of Th2- and Th17- (to a lesser extent) type allergic asthma upon HDM airway challenge.
To examine whether the MC903-induced aggravation of HDM asthma is mediated by keratinocytic TSLP, we employed TSLPiep−/− mutant mice in which TSLP gene is selectively and inducibly ablated in epidermal keratinocytes  (see Fig. S1). As expected, MC903-enhanced accumulation of eosinophils, neutrophils, and lymphocytes in BAL fluid, increase of HDM-specific IgE and IgG1, pulmonary infiltrate of eosinophils, as well as mucus production in HDM-treated wild-type (WT) mice, were all abolished in HDM-treated TSLPiep−/− mutant mice (Fig. 2G–H). Moreover, MC903-enhanced AHR in HDM-treated WT mice was also diminished upon TSLP ablation in keratinocytes (see Fig. S2). These results thus demonstrate that TSLP overproduced by skin keratinocytes promotes airway sensitization to HDM and thereby triggers subsequently an HDM allergic asthma.
Collectively, we provide here for the first time the experimental proof that TSLP represents an AD skin-delivered signal to promote airway sensitization to inhaled aeroallergen like HDM, which may account for one mechanism underlying the ‘atopic march’. We find that skin TSLP enhances not only HDM-primed Th2 but also Th17 immune responses , while barely affecting Th1 response. Of interest, we observed additionally that TSLP could promote sensitization to very low ‘non-asthmatic’ dose of HDM (e.g. 0.1 μg) (data not shown). Given that HDM is the most clinically relevant aeroallergen for allergic asthma and that its levels in the environment have been correlated with prevalence of asthma , our finding is of great clinical implication. It suggests that when exposed to common environment, AD individuals bearing increased skin TSLP expression may have a higher risk for developing sensitization to inhalant common allergens through airways, which contributes to the onset of the ‘atopic march’. It will be necessary to examine whether serum levels of TSLP in patients with AD, particularly in the infants with early-onset AD (in which sensitization often develops at early stage) , correlate with sensitization to environmental allergens in these patients. Our study also suggests that treatment or prevention strategy of asthma only targeting on airway may be insufficient, and blocking TSLP production in skin could be therapeutically useful in preventing or limiting airway sensitization, which is commonly developed in patients with AD, and halting the progress of the ‘atopic march’.