Wild-type (WT) BALB/c were obtained from the NCI. WAS ^−/− C57BL/6 mice were provided by Dr. Janis Burkhardt (U. Penn). WAS ^−/− BALB/c were generated by backcross of WAS ^−/− C57BL/6 for at least 10 generations. IL17^−/− BALB/c mice were provided by Dr. Anna Valujskikh (Cleveland Clinic). IL-4^−/− BALB/c animals were purchased from Jackson Animal Laboratories. Rag2^−/− BALB/c mice were obtained from Dr. Terry Wright (U. Rochester). All animals were housed in the specific pathogen-free facility at the University of Rochester. All animal experimentation was reviewed and approved by the University of Rochester’s University Committee on Animal Resources and the Institutional Animal Care and Use Committee.

Fur on mouse flank skin was trimmed manually, then TEWL was measured using TM300 device from Courage-Khazaka Electronics (Cologne, Germany).

For whole mount IF of epidermal sheets, hair was first removed from ears with depilatory cream (VEET). Ear dorsal and ventral sheets were separated and incubated in 0.5 M ammonium thiocyanate at 30°C for 20 min. Epidermis and dermis were separated and epidermal sheets fixed in 4% paraformaldehyde, washed and permeabilized with cold 100% methanol at −20°C. Epidermal sheets were blocked (1% BSA) and stained with polyclonal rabbit anti-claudin 1 Ab (Thermo Fisher; PAD : JAY.8) and species-specific AlexaFluor-tagged secondary Ab (Invitrogen) or AlexaFluor-tagged anti-mouse I-A/I-E (Biolegend: M5/114.15.2) and imaged using laser scanning confocal microscopy (Olympus FV1000). Three to four fields of view per epidermal sheet were captured and analyzed using Imaris v8.3.

Ears and flank skin were digested for 30 min at 37°C with collagenase/dispase. Colonic LP cells were isolated as previously described (33). Single cell suspensions were stained with a combination of the following antibodies: CD3 (BD; 145-2C11), CD4 (RM4-4, RM4-5), CD5 (53-7.3), CD8α (53-6.7), CD8β (53-5.8), CD11b (M1/70), CD11c (N418), CD45R/B220 (RA3-6B2), CD45 (30-F11), CD49b (Dx5), CD103 (2E7), CD127 (A7R34), CD205 (NLDC-145), CD207 (4C7), FcϵR1 (MAR-1), Siglec-F (E50-2440), Gr-1 (RB6-8C5), Ly6G (1A8), I-A/I-E (M5/114.15.2), T1/ST2 (DJ8), ICOS (C398.4A), NKp46 (29A1.4), KLRG1 (2F1), TER-119 (TER-119), γδ TCR(eBioGL3), Foxp3 (FJK-16s), GATA3 (L50-823), RORγt (Q31-378) and Live/Dead (Invitrogen). LSR-II flow cytometer acquisition and analysis using FlowJo software (Tree Star). Cell counts were determined by flow cytometry, from whole ear homogenates. The total homogenate from the ear pinna of each mouse was collected by flow cytometry and the total number of live, CD45⁺ cells calculated using 20,000 AccuCheck (Invitrogen) counting beads/sample for calibration, according to the manufacturers protocol.

Ear tissue was digested in collagenase/dispase and homogenized in 500 μl PBS. Supernatants were assayed using the Milliplex Mouse Cytokine/Chemokine kits, per manufacturer’s instructions. Plates were read using Bio-Plex 200 instrument (Bio-Rad). For LN intracellular cytokine staining, 1 × 10⁶ LN cells in RPMI/10%FCS were stimulated at 37°C with PMA (Sigma; 50 ng/ml) and ionomycin (Calbiochem; 500 ng/ml) for 4 h. Brefeldin A (BD; 1 ug/ml) was added after 1 h of culture, for the remaining 3 h of culture. Cells were washed and surface stained of CD45 and CD4 before fixation and permeabilization (BD Cytofix/cytoperm). Cells were stained for intracellular cytokines IFNγ (eBioscience; XMG1.2), IL-4 (eBioscience; 11B11) and IL-17A (Biolegend; TC11-18H10.1) and analyzed using the LSR-II flow cytometer and FlowJo software.

Serum IgE was measured by ELISA.

Catch-All Sample Collection swabs (Epicentre) of mouse ears were stored in 2 ml Eppendorf Safe-Lock Biopur tubes containing 100 μl of Yeast cell lysis buffer (Epicentre) at −80°C. DNA was extracted using MasterPure Yeast DNA Purification Kit (Epicentre) and PureLink Genomic DNA Mini Kit (Thermo Fisher) and eluted with 40 μl MoBio PCR water (certified DNA-free). 16S rRNA V1-V3 amplicon libraries were sequenced using the Illumina HiSeq2500 Sequencer (San Diego, CA). Illumina reads were assessed for quality and then analyzed using phylogenetic and operational taxonomic unit methods in the Quantitative Insights into Microbial Ecology (QIIME) open source software v1.9.1.

Prism software (GraphPad) was used for all statistical tests except microbiome analyses. All data presented are mean ± SEM unless otherwise indicated. Statistical analysis of microbiome data was performed in R software. Alpha diversity: Shannon index and PD whole tree. Beta diversity: Bray–Curtis dissimilarity index and Unweighted UniFrac. Relative abundance of taxa were binned for WT vs. WAS^−/− ; p-values by Student’s t-test and 999 Monte Carlo permutations, adjusted for multiple comparison using the p.adjust function in R (method = ‘fdr’). *p <.05, **p <.01, ***p <.001, ****p <.0001, not significant (n.s.) p >.05.

We assessed homeostatic immunity in the ear skin of both adult (8-week old) C57BL/6 and BALB/c WAS^−/− mice. No overt changes in immune cell numbers in the skin of WAS^−/− C57BL/6 mice were found, however, WAS^−/− BALB/c mice had significantly elevated numbers of CD45⁺ hematopoietic cells in the steady state ( Figure 1A ). We noted some variability between experiments in the overall magnitude of inflammation, but in all cases the difference between WT and WAS^−/− mice was significant within each individual experiment ( Figure 1B ). Increases in CD45⁺ cells were also observed in the dorsal flank skin of WAS^−/− BALB/c animals ( Figure 1C , left panel). In contrast, no increase in CD45⁺ cells in the cervical skin-draining lymph node (LN) was observed, suggesting immune dysregulation was cutaneously, and not centrally, driven ( Figure 1C , right panel). Moreover, cutaneous inflammation was only observed in the full absence of WASp, WAS^+/− mice showing no overt skin inflammation ( Figure 1D ). Accompanying the increase in leukocyte infiltration was the upregulation of the atopic dermatitis-associated pruritogenic ‘itch’ factor IL-31 ( Figure 1E ).

To determine if the elevated immune cell accumulation in the skin had physiological effects on the skin barrier, we first examined histological sections of mouse ear skin. At eight weeks of age, we observed dysmorphology of WAS^−/− mouse skin architecture, with parakeratosis and significant thickening of the epidermis compared to WT ( Figures 1F, G ). Parakeratosis may indicate altered keratinocyte differentiation and changes to the epidermal tight junction (TJ) network (34–38). We used confocal microscopy to visualize expression patterns of claudin-1 deposition, a major skin TJ protein, in epidermal sheets prepared from WT and WAS^−/− mouse skin ( Figure 1H ). The WT epidermis showed clear claudin-1 stain localized at interepithelial junctions in a typical honeycomb pattern (39), while the epidermis from WASp-deficient mice showed regions of claudin-1 staining with poorly defined intercellular borders. There were no differences in total claudin-1 epidermal expression between WT and WAS^−/− skin (MFI) ( Figure 1I , left panel), but quantification of the contiguous intensity of claudin-1 staining along the epithelial cell borders revealed WAS^−/− claudin-1 expression was significantly more irregular than WT ( Figure 1I , right panel; Supplementary Figure 1 ). In response to TJ disruption, Langerhan cells (LCs) have been reported to extend dendrites out into the TJ area to ‘plug’ holes in a disrupted epidermis. Consistent with a TJ breach, we observed an increase in MHC Class II⁺ projections (predominantly from CD11b⁺CD11c⁺Langerin⁺ cells) between epithelial cells in WAS^−/− epidermal sheets ( Figure 1J , TJ layer panels). Moreover, there was a marked accumulation of MHC-II⁺ cells in the epidermal sheets ( Figure 1J , right panel; Figure 1K ), as previously noted (31, 32).

These data suggest a disruption in the interface between the physical and immune barriers of the skin in the absence of WASp. Disrupted tight junctions in the epidermis in human atopic dermatitis corresponds to alterations in skin permeability to water (37, 39). WAS^−/− mice had significantly elevated trans-epidermal water loss (TEWL) values compared to WT mice, indicating increased skin permeability and epidermal barrier dysfunction ( Figure 1L ). Thus, we find loss of WASp leads to both cutaneous immune dysregulation and skin barrier disruption.

Immune composition by flow cytometry in the skin at eight weeks of age revealed a general increase in many immune cell subsets in WAS^−/− mice (subsets defined in Supplementary Table 1 ). Most innate cell subsets tested were enhanced including neutrophils, basophils, CD11c⁺ dendritic cells (DCs), LCs, NK cells, and CD11b⁺ monocytes/macrophages ( Figure 2A ) and increases in group 2 and 3 innate lymphoid cells (ILC2 and ILC3, Figure 2A ). Eosinophils were, notably, not elevated in the absence of WASp ( Figure 2A ). CD4⁺ T cells, CD8⁺ T cells, and γδ TCR⁺ T cells were also significantly increased in WAS^−/− skin ( Figure 2B ). Foxp3⁺ CD4+ T regulatory cell (Treg) numbers were only modestly enhanced in WAS^−/− skin ( Figure 2C , left panel) resulting in a significant reduction in the proportion of Tregs among CD4⁺ T cells ( Figure 2C , right panel). Aside from Tregs, we found no significant alterations in the relative frequencies of the elevated immune cell subsets we evaluated ( Figure 2D and Supplementary Figure 2 ). Thus, we observed significant increases in multiple skin immune cell populations, indicating broad dysregulation of skin immunity in WASp-deficient mice.

Immune cell cytokines could induce the observed epidermal barrier defects (40–43), or early barrier disruption could enhance immune cell accumulation. To gain mechanistic insight into the skin pathology, we assessed the concordance of immunologic changes and epidermal barrier dysfunction in the skin of WASp-deficient mice kinetically starting at 1 week of age. We detected increased CD45⁺ cell infiltration in WAS^−/− skin as early as one week of age ( Figure 3A ). At this time, a modest, but not significant, increase in TEWL was observed in WAS^−/− mice ( Figure 3B ) but no change was seen in the young (1 wk) skin architecture between WT and WAS^−/− mice ( Figure 3C ). In contrast, physical barrier dysfunctions (TEWL, epidermal thickness) were significantly elevated by 4 weeks of age ( Figures 3B, D ) at the height of immune accumulation ( Figure 3A ). Thus, altered immune cell accumulation appears to precede the epidermal skin barrier dysfunction.

With respect to immune cells, neonatal WT and WAS^−/− had a similar balance of immune cells, heavily skewed to monocyte/macrophage populations ( Figure 3E ). However, at four weeks of age, WAS^−/− mice showed an enhanced frequency of CD4⁺ cells in the skin, with a concomitant decrease in the proportion of neutrophils and monocytes/macrophages present in the skin ( Figure 3E ). Interestingly, these early defects in skin immunity came weeks before the WASp-associated elevation in serum IgE and before lymphocytic infiltration of the colonic lamina propria ( Figure 3F ).

Significant alterations in the skin microbiome have been reported in humans with primary immunodeficiencies and in AD (44–46). To investigate whether the immune and barrier dysregulation in the WAS^−/− skin impacted the skin microbiome, we performed 16S ribosomal RNA (rRNA) sequencing on skin swabs from ear skin. Using the Illumina MiSeq platform, we obtained an average of 143,474 ± 40,444 V1–V3 reads/sample for 10 WT and 10 WAS^−/− mice at eight weeks of age. Rarefaction curves indicated good coverage for dominant species present in the skin ( Supplementary Figure 3 ). We detected no significant differences in alpha diversity between WT and WAS^−/− (WT mean: 4.7 ± 1.8, WAS^−/− mean: 4.5 ± 0.9 p = .773) using the Shannon diversity index (which assesses microbial community richness and evenness within a single sample) ( Figure 4A ). However, we observed significant clustering of WAS^−/− samples separate from WT using the unweighted Unifrac method and principle coordinate analysis for assessing beta diversity ( Figure 4B ) (p = 1.41E−18). Changes in relative abundance of a number of bacterial communities was observed with WASp-deficiency ( Figure 4C ) with enrichment of the Streptococcus and Helicobacter genera, and also the Deferribacteres phylum ( Figure 4D ). Furthermore, there was colonization of a particular Gammaproteobacteria species, Aggregatibacter pneumotropica, on WAS^−/− mouse skin that was not detected on WT skin ( Figure 4D ).

To assess the concordance of dysbiosis with the onset of skin pathology, we analyzed the skin microbiome at 1, 4, and 8 weeks of age ( Figures 4E, F and Supplemental Figure 4 ). At one week of age, a striking change in the representation of microbial taxa in WAS^−/− mice was already evident ( Figures 4E, F ). Elevations in Gammaproteobacteria and members of the Deferribacteres phyla at 1 week, remained elevated above WT at all timepoints ( Figures 4E, F ). The relative abundance of Streptococcus spp. in the WAS^−/− mice was not seen as a neonate but increased in representation with age ( Figures 4E, F ), possibly as a secondary consequence of cutaneous inflammation. Therefore, the dysbiosis in WAS^−/− mice appears coincident with the early immune dysfunction in the skin of WASp-deficient mice. Preliminary co-housing experiments suggest that the altered microbiome of WAS^−/− skin is not sufficient to induce skin inflammation in WAS-sufficient (WT) mice ( Supplementary Figure 5 ). Four week old WT and WAS^−/− mice were co-housed for two weeks and then assessed for ear skin inflammation. After two weeks of co-housing, WT and WAS^−/− skin microbiomes had equilibrated, containing a similar microbial profile ( Supplementary Figure 5 ). This microbial mixing was mainly attributed to the acquisition of elevated WASp-associated Gammaproteobacteria, Deferribacteres, and Streptococcaceae by WT mice, with no significant impact of the WT microbiome on the co-housed WAS ^−/− mice. However, we observed no induction of skin inflammation associated with these changes in the skin microbiome in WT cohoused mice ( Supplementary Figure 5 ), suggesting that the microbiome of WAS^−/− mice is not sufficient to drive cutaneous inflammation in immunocompetent WT hosts.

We next measured inflammatory mediators in the mouse ear skin. We found marked differences in cytokines and chemokines in the skin of WAS^−/− mice using a multiplex Luminex assay ( Figure 5A ). Out of 55 analytes tested ( Supplementary Table 2 ), 18 analytes were significantly upregulated in WAS^−/− versus WT skin ( Figure 5A , highlighted in magenta) and no factors were downregulated. IL-4 and IL-17 were both significantly enhanced in the WAS^−/− skin, but there was no increase in IFN-γ, suggesting a bias toward type 2 and/or 17 responses ( Figure 5B ). Indeed, type 2 cytokines (IL-4, IL-5) and type 17 cytokines (IL-17, TNF, IL-22, IL-23) were among the most upregulated analytes ( Figure 5C ). CCL17, whose cognate receptor CCR4 is expressed by Th2/Th17 CD4⁺ cells, was particularly enriched in WAS^−/− skin ( Figure 5C ). This chemoattractant cue may promote the preferential accumulation of type 2/17 effectors in the skin (47) along with the AD-associated pruritogenic cytokine IL-31 ( Figure 1C ) (48). To determine whether the type 2/17 inflammatory bias present in the skin was associated with changes in LN Th differentiation, we assessed cytokine production by CD4⁺ T cells from the skin-draining LN. Upon in vitro re-stimulation, IL-4 was upregulated 14.62 fold, IL-17 by 4.887 fold and IFN-γ by 2.853 fold in WAS^−/− CD4⁺ T cells compared to WT ( Figure 5D ). Thus, the type 2 cytokine enrichment (and to a lesser extent type 17) we observed in the skin may be associated with enhanced priming or potential for IL-4/IL-17 production in the skin draining LN. In the skin, γδ T cells were also sources of elevated IL-17 (data not shown). Kinetically, the elevated inflammatory cytokines in the skin at 8 weeks ( Figure 5C ), were not enhanced at 1 week of age. However, IL-5 and IL-17 cytokines were upregulated in the WAS^−/− skin by 4 weeks of age ( Figure 5E ).

To assess the role of the adaptive immune compartment in the cutaneous immune dysregulation, we crossed WAS^−/− mice to Rag2^−/− mice on the BALB/c background (WAS^−/−Rag2^−/− ). Loss of T and B cells abolished WAS-associated skin inflammation, as determined by normalizing for the non-T cell CD45⁺ cell (CD45⁺CD4⁻CD8β⁻γδ⁻) compartment between WAS^−/− and WAS^−/−Rag2^−/− ( Figure 5F ). The results suggest the adaptive immune compartment actively drives the enhanced innate cell recruitment in WAS^−/− skin. However, interpretation of these experiments is complicated by the presence of an unexpected skin inflammation in the Rag2^−/− mice themselves. Loss of both T cells and WASp, WAS^−/−Rag2^−/− , led to the amelioration of inflammation. At the cytokine level, elevated cytokine production in the WAS^−/− skin was reduced to WT levels in the WAS^−/−Rag2^−/− mice ( Figure 5G and Supplementary Table 3 ). Thus, lymphocytes appear to be essential for the initiation and/or amplification of cutaneous pathology in WASp-deficient mice.

In the gut, elevated IL-4 has been previously associated with enhanced serum IgE and colitis in WASp-deficient mice (29). Recent studies in an adoptive transfer system of Th1/Th17 colitis have shown elevated IFNγ and IL-17 in the gut when macrophages lack WASp, but the functional significance of IL-17 was not explored (28, 29). Therefore, we assessed the role of IL-4 and IL-17 in the development of skin inflammation using WAS^−/−IL-4 ^-/- and WAS^−/−IL-17^−/− mice ( Figures 6A, B ). IL-4 deficiency failed to alleviate the dysregulated accumulation of immune cells in the WAS^−/− skin ( Figure 6A ). Furthermore, we found little change in composition of immune cell subsets or the cytokine/chemokine milieu in skin of WAS^−/−IL-4^−/− compared to WAS^−/− mice ( Supplementary Figure 6 ). In contrast, we found that total CD45⁺ cells were significantly decreased in WASp-deficient skin in the absence of IL-17 ( Figure 6B ). Lack of IL-17A in WASp-deficient animals led to significant reductions in the accumulation of innate cell types such as monocytes/macrophages/LCs, basophils and CD11c⁺ dendritic cells ( Figure 6C ), suggesting that IL-17A may play a role in the tissue recruitment or expansion of these populations in WAS^−/− animals. CD4+ T cells counts trended down in WAS^−/− groups in the absence of IL-17, but the WAS^−/−IL-17^−/− phenotype was intermediate between WT and WAS^−/− groups, with no significant differences when compared to either group.

Most notably, loss of IL-17 was accompanied by significant reductions in many of the inflammatory mediators found elevated in the WAS^−/− skin ( Figure 6D ), to levels similar to that of immunocompetent WT and IL-17^−/− mice. In particular, we found a striking reduction in the type 2 cytokines IL-4 and IL-5 in the WAS^−/−IL-17^−/− skin compared to WAS^−/− controls ( Figure 6E ). The pruritogenic factor IL-31 was also significantly reduced to WT levels in the skin of WAS^−/−IL-17^−/− mice ( Figure 6E ). Our observations point to IL-17A, but not IL-4, contributing to the development and/or maintenance of the skin inflammation described in WAS^−/− animals. Interestingly, as noted in other murine models of asthma and eczema (49–51), IL-17 appears to support or amplify type 2 cytokines in WASp-deficient animals. Our results uncover a previously unappreciated spontaneous breakdown in cutaneous immune homeostasis in the absence of WASp driven by elevated IL-17. Immune dysregulation occurred soon after birth and was associated with altered skin TJ morphology and barrier function resulting in dysbiosis.