Research ArticleSKIN

Humidity-regulated CLCA2 protects the epidermis from hyperosmotic stress

See allHide authors and affiliations

Science Translational Medicine  09 May 2018:
Vol. 10, Issue 440, eaao4650
DOI: 10.1126/scitranslmed.aao4650

Humidity—Harbinger of health?

Low environmental humidity can aggravate symptoms of atopic dermatitis (AD), an inflammatory skin disease. Seltmann et al. investigated the link between humidity and epidermal barrier function. Using a mouse model that exhibits AD-like symptoms, they found that high humidity reduced epidermal thickness and skin inflammation. Chloride channel accessory 2 protein was highly expressed in keratinocytes in response to hyperosmotic stress, and it increased cell-cell adhesions and protected cells from apoptosis. This osmoregulated protein was up-regulated in skin samples from patients with AD, which suggests a compensatory mechanism to maintain epidermal barrier function.


Low environmental humidity aggravates symptoms of the inflammatory skin disease atopic dermatitis (AD). Using mice that develop AD-like signs, we show that an increase in environmental humidity rescues their cutaneous inflammation and associated epidermal abnormalities. Quantitative proteomics analysis of epidermal lysates of mice kept at low or high humidity identified humidity-regulated proteins, including chloride channel accessory 3A2 (CLCA3A2), a protein with previously unknown function in the skin. The epidermis of patients with AD, organotypic skin cultures under dry conditions, and cultured keratinocytes exposed to hyperosmotic stress showed up-regulation of the nonorthologous human homolog CLCA2. Hyperosmolarity-induced CLCA2 expression occurred via p38/c-Jun N-terminal kinase–activating transcription factor 2 signaling. CLCA2 knockdown promoted keratinocyte apoptosis induced by hyperosmotic stress through impairment of cell-cell adhesion. These findings provide a mechanistic explanation for the beneficial effect of high environmental humidity for AD patients and identify CLCA3A2/CLCA2 up-regulation as a mechanism to protect keratinocytes from damage induced by low humidity.


Chronic inflammatory skin diseases affect a large percentage of the population, and their incidence is continuously increasing. Particularly frequent is atopic dermatitis (AD), which has a lifetime prevalence of more than 15% in many countries and in particular in affluent settings (13). AD severely affects quality of life through itching, sleep deprivation, and social embarrassment due to visible lesions. The underlying pathological mechanisms are still incompletely understood but include epidermal barrier deficiency and immunological alterations, such as immunoglobulin E–mediated sensitization to various allergens (3). A large number of studies imply that impaired keratinocyte function drives AD in a considerable proportion of patients (13).

Inherited loss-of-function mutations in the gene encoding the structural protein filaggrin are present in 15 to 20% of all AD patients and in up to 50% of patients with severe disease (13). The gene encoding the tight junction component claudin-1 has also been suggested as a susceptibility gene for AD, and expression of claudin-1 was strongly reduced in the epidermis of AD patients (4). This is likely functionally important because a reduction in claudin-1 expression induced AD-like features in mice (5). Mice lacking fibroblast growth factor receptor 1 (FGFR1) and FGFR2 in keratinocytes (K5-R1/R2 mice) also exhibit defective epidermal barrier function, which results at least in part from reduced expression of tight junction components (6, 7). This causes a phenotype that shares several features with chronic AD, including increased transepidermal water loss, epidermal thickening, keratinocyte hyperproliferation, and inflammation (6, 7). Reduced FGFR1 and FGFR2 expression was detected in the skin of AD patients (8), and single-nucleotide polymorphisms in the FGFR1 and FGFR2 genes were linked to atopy (9), indicating a role for the FGFR–tight junction axis in AD pathogenesis. Although this requires further investigation, and with the caveat that mouse models never completely reflect human disease because of the anatomical differences between murine and human skin, K5-R1/R2 mice are a suitable model to study the consequences of an impaired epidermal barrier and the associated enhanced water loss, which are hallmarks of AD.

The prevalence of AD is markedly lower in geographic regions with high humidity, and aggravation of AD symptoms often occurs under conditions of low environmental humidity, such as during the winter season (10, 11). In hairless mice, low humidity increased interleukin-1 (IL-1) expression, mast cell number, and histamine content (12, 13). However, it also enhanced epidermal barrier function under steady-state conditions and promoted repair of the barrier after its disruption by acetone treatment or tape stripping, possibly as a compensatory mechanism (14). In organotypic and ex vivo human skin cultures, a reduced hydration status induced the expression of the S100A8/S100A9 cytokines and consequent activation of the underlying fibroblasts (15). These studies demonstrate a remarkable influence of environmental humidity on the skin, but the underlying molecular mechanisms are still largely unknown.

Here, we show that the phenotype of K5-R1/R2 mice improves at high environmental humidity. In addition, we identify chloride channel accessory 2 (CLCA2) as a cytoprotective protein in keratinocytes under conditions of hyperosmotic stress.


High humidity normalizes keratinocyte proliferation and differentiation in K5-R1/R2 mice

Mice lacking Fgfr1 and Fgfr2 in keratinocytes (K5-R1/R2 mice) show AD-like signs when kept at 50% relative humidity (normal housing conditions) as a result of their epidermal barrier defect (6, 7). To investigate whether alterations in humidity affect the phenotype, K5-R1/R2 and control mice were transferred to 40 or 70% humidity. The epidermal thickening that was seen in the K5-R1/R2 mice at 40% humidity was almost completely rescued after 3 weeks of maintenance at 70% humidity (Fig. 1A). This correlated with a strong reduction of keratinocyte proliferation and of interfollicular expression of keratin 6 (K6) and K16 (Fig. 1, B and C). Expression of these proteins in the interfollicular epidermis is a hallmark of abnormal differentiation, hyperproliferation, and stress of keratinocytes (16). The increase in humidity did not significantly affect epidermal thickness, keratinocyte proliferation, or K6/K16 expression in control mice (Fig. 1, A to C).

Fig. 1 High humidity reduces the epidermal phenotype of K5-R1/R2 mice.

Four-month-old mice were kept at 40 or 70% humidity for 24 days. (A) Epidermal thickness determined in hematoxylin and eosin (H&E)–stained sections of dorsal skin. Scale bars, 50 μm. D, dermis; E, epidermis; HF, hair follicle; ctrl, control. (B) Keratinocyte proliferation was determined by 5-bromo-2′-deoxyuridine (BrdU) staining (green). Nuclei were counterstained with propidium iodide (red). Scale bars, 50 μm. (C) Immunofluorescence staining of K6 and K16 (red) and counterstaining of nuclei with 4′,6-diamidino-2-phenylindole (DAPI; blue). Staining in the stratum corneum (indicated by an asterisk) is unspecific background. Scale bar, 100 μm. Dotted white line marks the basement membrane. At least 5 mm of skin was analyzed per mouse. Scatter plots in (A) and (B) show means ± SD. n = 5 to 6 mice per genotype and treatment group. Improvement of the phenotype of K5-R1/R2 mice was observed in five independent experiments. At least 5 mm of skin was analyzed per mouse. **P ≤ 0.01, Mann-Whitney U test.

High humidity reduces skin inflammation in K5-R1/R2 mice

Numbers of total epidermal and dermal immune cells were also mildly reduced at high humidity in K5-R1/R2 mice (Fig. 2, A and B) because of a mild reduction of epidermal αβ and γδ T cells and Langerhans cells and of dermal αβ T cells and dendritic cells (Fig. 2, A and B). However, only the reduction in total dermal immune cells was statistically significant. Analysis of the expression of proinflammatory cytokines showed a more robust effect of high humidity, and there was a significant reduction in the expression of the genes encoding IL-1β, IL-36β, S100A8, and S100A9 in the skin of K5-R1/R2 mice at 70% versus 40% humidity (Fig. 2C). The reduction of the inflammatory phenotype was not a consequence of normalization of tight junction gene expression because the down-regulation of claudin-1 and claudin-3 in K5-R1/R2 mice was independent of humidity (Fig. 2C). These tight junction components are direct targets of FGF signaling (6), and thus, their expression was only affected by the genotype but not by alterations in humidity. Therefore, the increase in humidity does not affect the barrier directly but the consequences of the barrier defect, such as epidermal thickening, keratinocyte hyperproliferation, and inflammation.

Fig. 2 High humidity decreases inflammation in the skin of K5-R1/R2 mice.

Four-month-old male mice were kept for 24 days at 40 or 70% humidity. (A and B) Flow cytometry analysis of dissociated epidermal (A) and dermal cells (B) for quantification of total CD45+ immune cells, γδ T cells (CD3+ TCRδ+), αβ T cells (CD3+ TCRβ+), Langerhans cells (epidermal CD45+ CD11c+), and dendritic cells (dermal CD45+ CD11c+). The percentage of individual cells among all cells is shown. (C) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) of epidermal RNA for Il1b, Il36b, S100a8, S100a9, claudin-1 (Cldn1), and Cldn3 relative to Rps29. Error bars indicate means ± SD. n = 4 to 7 mice per genotype and treatment group. Data are representative of two experiments. *P < 0.05, **P < 0.01, Mann-Whitney U test.

Amelioration of the phenotype by high humidity was still observed in aged mice (18 months; fig. S1, A and B). A 10-day incubation at high humidity significantly reduced the number of epidermal γδ T cells and Langerhans cells at this age (fig. S1B).

Quantitative proteomics identifies humidity-regulated proteins in mice with a barrier function defect

To gain insight into the molecular mechanisms underlying the suppression of the skin phenotype of K5-R1/R2 mice at high humidity, we performed a quantitative isobaric tags for relative and absolute quantitation (iTRAQ)–based proteomics analysis of epidermal proteins 3 days after the change in humidity to identify the early alterations. At this time point, the reduction of epidermal thickening, keratinocyte proliferation, and inflammation was already detectable, although most parameters were not yet statistically significant (fig. S2). We identified and quantified the relative expression of 1920 proteins (false discovery rate < 1%) in the epidermis from mice of both genotypes kept at either low or high humidity (table S1). Using a reference approach, we quantitatively compared samples from both genotypes and humidity conditions (Fig. 3A) in two 8plex iTRAQ experiments (fig. S3A). We then filtered those proteins that were only significantly higher in abundance in samples from K5-R1/R2 mice under low humidity but, at high humidity, reverted their expression to the expression seen in control epidermis (Fig. 3B). Consequently, we obtained a group of 23 proteins (table S2) associated with the amelioration of the AD-like phenotype in K5-R1/R2 mice in response to an increase in humidity.

Fig. 3 Quantitative proteomics identifies proteins that are controlled by alterations in humidity.

(A) Three-month-old control and K5-R1/R2 mice were kept at 40 or 70% humidity for 3 days. Epidermal lysates were subjected to iTRAQ-based quantitative proteomics. n = 4 to 5 control and n = 3 K5-R1/R2 mice per condition. LC-MS/MS, liquid chromatography–tandem mass spectrometry. (B) Heat map of proteins whose increased abundance in K5-R1/R2 animals was reverted to normal expression by high humidity. Proteins were analyzed by two-way analysis of variance (ANOVA) (raw P < 0.05) for interaction of genotype and treatment and were filtered by fuzzy clustering for high abundance only in K5-R1/R2 mice at low humidity. Abundance values were normalized to total peptide amounts, scaled to the reference channel, and, for fuzzy clustering, standardized to an average abundance value of zero and 1 SD (table S2). (C) K16/K6 immunofluorescence (red) and counterstaining of nuclei (DAPI; blue). Staining in the stratum corneum (indicated with an asterisk) is unspecific. Scale bar, 100 μm. Dotted white line indicates basement membrane. (D) qRT-PCR of RNA from K5-R1/R2 or control mice maintained at 40 or 70% humidity for Rps19, Rps25, and Clca3a2 relative to Gapdh. n = 6 to 8 mice per genotype and treatment group. Error bars indicate means ± SD. *P < 0.05, **P < 0.01, Mann-Whitney U test.

Humidity-regulated proteins are involved in protein synthesis and keratinocyte differentiation

K16 was among the proteins that were rapidly regulated by high humidity. We verified that the abnormal interfollicular expression of K16 and its dimerization partner K6 was partially rescued by high humidity conditions after 3 days (Fig. 3C). Therefore, K6 and K16 belong to a group of proteins that are up-regulated in K5-R1/R2 mice and rapidly respond to high humidity. Most of these proteins are components of ribosomes or other proteins involved in protein translation (Fig. 3B and table S1), which suggests that they are regulated in a coordinated manner in response to humidity alterations. We confirmed the humidity-dependent expression of ribosomal proteins S19 (RPS19) and S25 (RPS25) at the RNA level (Fig. 3D).

CLCA3A2 is a humidity-regulated protein in keratinocytes

We next performed a literature search to determine whether any of the humidity-regulated proteins had previously been associated with a defect in epidermal barrier function. Clca3a2 (previously designated Clca2) (17), a gene with previously unknown functions in keratinocytes, is overexpressed in mice lacking the transcription factor Kruppel-like factor 4 (KLF4), and these mice display a disturbed epidermal barrier due to a defect in the cornified envelope (18). Therefore, increased expression of Clca3a2 may be a general consequence of a barrier defect. Up-regulation of Clca3a2 in the epidermis of K5-R1/R2 mice was confirmed at the RNA level, and its expression declined at 70% humidity (Fig. 3D). Therefore, we further studied the regulation of CLCA3A2 and its role in the response of the skin to altered environmental humidity.

Clca3a2 is an osmoregulated gene

Exposure of the skin to low humidity and subsequent water loss through a defective barrier causes hyperosmotic stress, which results in an efflux of water from the cytoplasm to the extracellular environment (19). This can be mimicked in vitro by addition of sorbitol, sucrose, or high concentrations of NaCl to the medium (Fig. 4A). Expression of Clca3a2 was strongly induced by sorbitol in murine keratinocytes, whereas the related Clca2 gene (previously called mClca5), which is also expressed in keratinocytes albeit at much lower amounts, was not regulated (Fig. 4B). FGFR deficiency did not induce Clca3a2 expression as shown by analysis of cultured keratinocytes from K5-R1/R2 and control mice (Fig. 4C). This finding further suggests that the up-regulation of CLCA3A2 in the epidermis results from the hyperosmolarity and not from the FGFR deficiency and subsequent reduction in claudin expression.

Fig. 4 Low humidity and hyperosmolarity induce Clca3a2 and CLCA2 expression in keratinocytes.

(A) Schematic drawing of cells subjected to dryness (hyperosmolarity), which leads to water efflux and cell shrinking. (B) Primary murine keratinocytes were treated for 3 hours with 100 mM sorbitol and analyzed by qRT-PCR for Clca3a2 (n = 5 to 9) and Clca2 (n = 4) relative to Rps29. ns, not significant. (C) RNA from primary keratinocytes of K5-R1/R2 and control mice was analyzed by qRT-PCR for Clca3a2 (n = 7 to 10). WT, wild type. (D) RNA from the epithelium of 3D organotypic cultures of human foreskin fibroblasts and HaCaT keratinocytes kept at high or low humidity was analyzed by qRT-PCR for CLCA2 relative to RPS27 (n = 5) and by Western blotting for CLCA2 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). CLCA2 expression after small interfering RNA (siRNA)–mediated knockdown is shown as a control for antibody specificity. scr, scrambled siRNA. (E) RNA from the epithelium of 3D organotypic cultures of human fibroblasts and keratinocytes, which had been treated for 24 hours with 200 mM sorbitol, was analyzed by qRT-PCR for CLCA2 relative to RPS27 (n = 6). (F) Primary human keratinocytes were treated for 3 hours with 100 mM sorbitol and analyzed by qRT-PCR for CLCA2 relative to RPS27 (n = 5). (G and H) HaCaT keratinocytes were treated for different time periods with 200 mM sorbitol and analyzed by qRT-PCR for CLCA2 relative to RPS27 (n ≥ 4) (G) or by Western blotting for CLCA2 or α-tubulin (H). Bars show means ± SEM. Expression in control cells was set to 1. **P ≤ 0.01, ***P ≤ 0.001, Mann-Whitney U test.

Human CLCA2 is an osmoregulated protein in keratinocytes

The proposed human ortholog of Clca3a2 is CLCA3, which is most likely a pseudogene that cannot give rise to a functional protein and is not even expressed at the RNA level (17). Data from the Human Protein Atlas ( confirmed these results and demonstrated that the only highly expressed protein of the CLCA family in human skin is CLCA2 (fig. S3B). We therefore examined whether CLCA2 is also regulated by low humidity using three-dimensional (3D) organotypic human skin cultures (20). CLCA2 was expressed in the 3D epithelium formed by immortalized but nontumorigenic HaCaT keratinocytes under normal culture conditions (100% humidity), and the amounts of CLCA2 mRNA and protein markedly increased when the 3D cultures were exposed to silica gel (21) to reduce the humidity (Fig. 4D).

A rapid up-regulation of CLCA2 mRNA also occurred in the epithelium of 3D cultures maintained under 100% humidity in the presence of sorbitol (Fig. 4E), whereas expression of RPS19, whose murine ortholog was regulated by humidity in K5-R1/R2 epidermis (Fig. 3D), was not affected (fig. S4A). Induction of CLCA2 expression by sorbitol was confirmed in primary human foreskin keratinocytes (HFKs) (Fig. 4F).

The sorbitol-induced CLCA2 up-regulation occurred in a time- and dose-dependent manner (Fig. 4G and fig. S4B). Only low amounts of full-length CLCA2 were detected in keratinocytes, whereas a 35-kDa protein, a putative carboxyterminal proteolytic product (22), was abundant and increased in response to sorbitol (Fig. 4H). Induction of CLCA2 expression was also observed in response to sucrose or high salt concentration, confirming the role of hyperosmolarity in its regulation (fig. S4C).

Hyperosmotic stress induces CLCA2 expression via p38/JNK-ATF2 signaling

Hyperosmotic stress was shown to activate the p38 and c-Jun N-terminal kinase (JNK) stress pathways, resulting in activation of different members of the activator protein 1 (AP-1) family (23). Activation of p38, JNK, and activating transcription factor 2 (ATF2), an AP-1 family member, occurred when HaCaT cells were treated with sorbitol (Fig. 5, A and B). Inhibition of either p38 or JNK, but not of extracellular signal–regulated kinase (fig. S4D), prevented the up-regulation of CLCA2 in response to sorbitol (Fig. 5C), and a similar inhibition was seen with two different AP-1 inhibitors or upon siRNA-mediated knockdown of ATF2 (Fig. 5, C and D). Sorbitol also caused an increase in ATF2 gene expression (Fig. 5D). These findings demonstrate that sorbitol-induced CLCA2 expression involves p38/JNK-ATF2 signaling.

Fig. 5 Hyperosmotic stress induces CLCA2 expression via a p38/JNK-ATF2 signaling pathway in HaCaT cells.

(A and B) Cells were treated for different time periods with sorbitol and analyzed by Western blotting for total and phosphorylated (p) p38, JNK/SAPK, and ATF2. (C) Cells, which had been pretreated for 1 hour with the p38 inhibitor SB202190, the JNK inhibitor SP600125, the Erk1/2 inhibitor U0126, or the AP-1 inhibitors SR11302 or T-5524, were cultured in the presence or absence of sorbitol and analyzed 3 hours after sorbitol addition by qRT-PCR for CLCA2 relative to RPS27. (D) Cells were transfected with ATF2 or scrambled (control) siRNAs. Forty-eight hours later, they were treated with sorbitol for 3 hours and analyzed by qRT-PCR for ATF2 or CLCA2 relative to RPS2 (n = 3 to 6). Bars show means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, Mann-Whitney U test.

CLCA2 protects keratinocytes from hyperosmotic stress–induced cell death

To determine whether the up-regulation of CLCA2 in response to hyperosmotic stress is functionally relevant, we performed siRNA-mediated knockdown of CLCA2 using four different siRNAs (Fig. 6A). Knockdown of CLCA2 affected neither proliferation nor survival of keratinocytes under normal culture conditions (Fig. 6, B and C). However, survival of CLCA2 knockdown cells, but not of control cells, was reduced upon treatment with sorbitol, as shown by measurement of lactate dehydrogenase (LDH) in the supernatant, which reflects cell lysis, and by staining for the apoptosis marker cleaved caspase-3 (Fig. 6, D and E). By contrast, ultraviolet B (UVB) irradiation did not induce CLCA2 expression in human keratinocytes (Fig. 6F), and Clca3a2 expression was not significantly up-regulated by UVB irradiation of mouse skin (Fig. 6G). Furthermore, cell death induced by UVB irradiation or incubation with the reactive oxygen species producing compound menadione was not enhanced by CLCA2 knockdown (Fig. 6H and fig. S4E), suggesting that CLCA2 protects cells from hyperosmotic stress.

Fig. 6 Knockdown of CLCA2 reduces keratinocyte survival in response to hyperosmotic stress.

(A to E) HaCaT keratinocytes were transfected with different CLCA2 siRNAs and analyzed 48 hours later. (A) RNA or protein samples were analyzed by qRT-PCR for CLCA2 relative to RPS27 or by Western blotting for CLCA2 and GAPDH, respectively (n = 4). (B) Cells were analyzed by BrdU incorporation (n = 4; at least 5000 cells per slide). (C) Cells were analyzed for LDH release (n = 4). (D) Cells were treated with 200 mM sorbitol as indicated and analyzed for LDH release (n = 3). (E) Cells were treated with 200 mM sorbitol for 6 hours, and 5000 cells per slide were analyzed by immunofluorescence for cleaved caspase-3 (n = 3). (F) HaCaT cells were irradiated with UVB (100 mJ/cm2) or treated with 200 mM sorbitol and analyzed 24 hours later for CLCA2 relative to RPS27 (n = 4). (G) Mice were irradiated with UVB (100 mJ/cm2). Twenty-four hours after irradiation, epidermal RNA was analyzed by qRT-PCR for Clca3a2 relative to Rps29 (n = 4 to 5 mice per treatment group). (H) siRNA-transfected cells were irradiated with UVB (100 mJ/cm2) and analyzed for LDH release 24 hours after irradiation (n = 4 to 8). Bars indicate means ± SEM. **P < 0.01, ***P < 0.001, Mann-Whitney U test.

CLCA2 promotes cell-cell adhesion under conditions of hyperosmotic stress

We next determined whether the protective effect of CLCA2 under these conditions results from enhanced cell-cell or cell-matrix adhesion. Although adhesion of CLCA2 knockdown cells to different matrices was not affected (Fig. 7A), cell-cell adhesion was reduced in sorbitol-treated cells transfected with CLCA2 siRNA as seen in a dispase assay under conditions of mechanical stress (Fig. 7B). This was reflected by the significant increase in the number of dispersed particles and their reduced size when the dispase-treated cells were subjected to high forces (Fig. 7, B to D). The assay was performed 1 hour after sorbitol treatment and thus before cleaved caspase-3–positive cells or other signs of cell damage/apoptosis appeared (Fig. 7E). Immunofluorescence staining of the cell adhesion protein E-cadherin showed strong membrane staining in control cells but a broader distribution in cells with CLCA2 knockdown. Upon sorbitol treatment, the membrane localization of E-cadherin seemed even more pronounced in control cells, whereas the staining was less restricted in CLCA2 knockdown cells (Fig. 7F). Western blot analysis of HaCaT cells with doxycycline-inducible expression of CLCA2 small hairpin RNA (shRNA) confirmed the CLCA2 knockdown (fig. S4F) and concomitant reduction of the E-cadherin amounts in the membrane fraction (Fig. 7, G and H). The effect of CLCA2 knockdown on E-cadherin is unlikely the consequence of epithelial-mesenchymal transition (EMT) because our cell-cell adhesion experiments were performed 2 days after CLCA2 knockdown and because even a 14-day knockdown of CLCA2 in HaCaT cells did not affect expression of major EMT markers (fig. S5).

Fig. 7 Knockdown of CLCA2 decreases cell-cell contact strength.

(A to D) HaCaT keratinocytes were transfected with scrambled (control) or CLCA2 siRNA and analyzed 48 hours after transfection. (A) After trypsinization, they were allowed to adhere for 45 min on plates coated with or without collagen type IV or fibronectin. Adhesion of control cells to uncoated dishes was set as 1. Values above 1 indicate stronger adhesion (n = 3 to 5). (B to D) Transfected cells were treated with sorbitol for 1 hour, followed by a 30-min dispase I treatment with or without orbital shaking (low force or high force). The number of cell sheets (B) and the size of the fragments (C) were quantified in the mechanically stressed (high force) samples (n = 4). Representative pictures are shown in (D). (E) At the same time point, cells were stained for cleaved caspase-3 combined with DAPI. (F) Immunofluorescence staining of E-cadherin in control and CLCA2 siRNA transfected HaCaT keratinocytes before and 30 min after sorbitol treatment. Scale bar, 10 μm. Data shown are representative of three experiments. (G and H) Western blot analysis of membrane and cytoplasmic fractions of HaCaT cells stably transfected with a vector allowing doxycycline-induced expression of CLCA2 shRNA or empty vector. Two days after induction of the knockdown, cells were treated with sorbitol for 30 min and analyzed for E-cadherin, α-tubulin, and epidermal growth factor receptor (EGFR) (loading controls). Quantification of the data from three independent experiments is shown in the bar graph. *P < 0.05, **P < 0.01, ***P < 0.001, Mann-Whitney U test.

Sorbitol treatment increased the amount of E-cadherin in the membrane of control cells (Fig. 7, G and H), as also suggested by the immunofluorescence data (Fig. 7F). However, this increase was not seen in CLCA2 knockdown cells (Fig. 7, G and H). In the absence of sorbitol, cytoplasmic E-cadherin expression was higher in CLCA2 knockdown cells but decreased upon sorbitol treatment, possibly as a result of proteolytic degradation (Fig. 7, G and H). Overall, these results demonstrate that the up-regulation of CLCA2 in response to hyperosmotic stress is beneficial and promotes cell-cell adhesion, which involves maintenance of E-cadherin in the plasma membrane.

Finally, we tested whether enhanced expression of CLCA2 in keratinocytes is beneficial. However, overexpression of CLCA2 in HaCaT cells also enhanced sorbitol-induced cell death (fig. S6, A and B), suggesting that CLCA2 expression must be tightly regulated to maintain cell viability under stress conditions.

CLCA2 is overexpressed in skin lesions of patients with AD

To determine the potential relevance of our findings for AD, we performed immunofluorescence staining of sections from lesional skin of AD patients and from the same body site of healthy control patients. CLCA2 was detected throughout the epidermis in control and AD patients (Fig. 8), with the strongest staining intensity observed in suprabasal cells as reported for porcine CLCA2 (24). The signals were much stronger in the hyperthickened epidermis of all tested AD patients (Fig. 8 and fig. S7).

Fig. 8 Overexpression of CLCA2 in the epidermis of lesional skin of AD patients.

Immunofluorescence staining of CLCA2 on paraffin sections of skin from three different healthy donors and AD patients. Samples from control and AD patients were photographed using the same exposure time. Scale bar, 100 μm.

Together, our results demonstrate that hyperosmotic stress that occurs in an epidermis with impaired barrier results in up-regulation of CLCA2 expression via a p38/JNK-ATF2 signaling pathway. The increased expression of CLCA2 stabilizes cell-cell contacts, which is likely a compensatory effect to maintain epidermal integrity (fig. S8).


We showed that increased humidity alleviates skin abnormalities in mice with a defective epidermal barrier. This is medically relevant because the clinical symptoms of AD often worsen during the dry winter season (25) and because lower prevalence rates of AD are observed in geographical regions of high humidity (26). Although normal skin can adapt to a dry environment over time, the increased water loss hampers the adaptation in AD skin (10, 27, 28).

Our findings are consistent with previous data demonstrating that low humidity amplifies the hyperproliferative response to barrier disruption in hairless mice (29). Furthermore, low humidity induced the expression of S100A8 and S100A9 in human ex vivo skin cultures or skin equivalents (15), and tape stripping of human ex vivo skin followed by exposure to low humidity caused up-regulation of IL-1β, TNF-α, and IL-8 (30). This resulted from the increase in extracellular sodium that was sensed by the sodium channel Nax (31). These studies focused on selected humidity-regulated cytokines, whereas our unbiased proteomics approach allowed us to identify a large panel of proteins that are regulated by alterations in environmental humidity in mice with an impaired epidermal barrier. Many components of the protein translation machinery are among the panel of proteins that we identified; these alterations may allow keratinocytes to rapidly adjust their proliferation rate.

Here, we focused on murine CLCA3A2 and human CLCA2, members of a protein family that includes four isoforms in humans and eight in mice (32). CLCA proteins are named for their ability to induce calcium-dependent chloride conductance through regulation of chloride channel activity (33). As shown for CLCA1, this activation requires the metalloprotease function of the protein, resulting in autocatalytic cleavage and subsequent activation of the chloride channel by an N-terminal–secreted fragment (34). However, it is not clear whether chloride channel activation is the major activity/function of these proteins. Additional functions, which may be independent from the effect on chloride channels, have been described for CLCA2, including inhibition of proliferation and promotion of senescence, induction of apoptosis, and stabilization of cell-cell contacts of normal mammary epithelial and breast cancer cells (3538).

Expression of Clca3a2 and CLCA2 is regulated by dryness in vivo or in 3D culture, respectively, and by sorbitol in cultured keratinocytes, indicating a common osmoregulation of both genes. This regulatory mechanism also provides a likely explanation for the up-regulation of Clca3a2 in Klf4 knockout mice, which have a skin barrier defect due to alterations in the cornified envelope (18). Therefore, any defect in the epidermis that affects the barrier and thereby enhances skin dryness may result in the overexpression of this gene. Expression of bovine CLCA2 was also up-regulated by hyperosmotic stress in vertebral disc cells (39), demonstrating the conservation of this mechanism across species. Clca2 (previously called mClca5) rather than Clca3a2 is considered the mouse ortholog of human and bovine CLCA2 (17, 40). However, Clca2 mRNA was hardly detectable in the murine epidermis, including K5-R1/R2 mice, and its expression was not regulated by sorbitol. On the other hand, the proposed human ortholog of Clca3a2 is the pseudogene CLCA3 (17). Therefore, it seems likely that murine CLCA3A2 and human CLCA2 fulfill similar functions in the epidermis, particularly in response to hyperosmotic stress, and that CLCA2 is a nonorthologous homolog of Clca3a2.

Up-regulation of CLCA2 expression by hyperosmotic stress involved the p38/JNK-ATF2 signaling pathway, which also regulates several other stress response genes (41). However, it seems likely that additional signaling molecules contribute to this regulation because UVB irradiation, which also activates p38 and JNK, did not affect CLCA2 expression in keratinocytes.

Upon exposure to a dry/hyperosmotic environment, cell shrinkage occurs as a result of water efflux. A rapid induction of adaptive responses by the affected cells is essential for their survival and involves not only up-regulation of various transporters to restore the cell volume but also cytoskeletal alterations and increased expression of adhesion molecules (19). Our data strongly suggest that the enhanced expression of CLCA2 contributes to this adaptive response and allows survival of keratinocytes under conditions of hyperosmotic stress, as seen in patients with AD, especially in dry environments.

Knockdown of CLCA2 strongly affected cell-cell adhesion, which is consistent with the role of CLCA2 in adhesion of mammary epithelial cells through its interaction with the cell junctional protein EVA1 (epithelial V-like antigen 1) (38). In keratinocytes, CLCA2 knockdown reduced the amounts of E-cadherin, a major cell adhesion protein, in the cell membrane. This effect is unlikely the consequence of EMT, which occurred in breast cancer cells after CLCA2 knockdown (42), suggesting cell-type specific functions of CLCA2. The reduction of the amounts of E-cadherin in the cell membrane provides a possible explanation for the impaired cell-cell adhesion of CLCA2 knockdown cells upon sorbitol treatment. The cell shrinkage that occurs at hyperosmotic conditions may impose an additional stress on the cell-cell junctions, and under these conditions, the reduction of CLCA2 and of membrane E-cadherin is likely to affect the adhesive capacity. Finally, CLCA2 knockdown may also affect cell volume regulation because of its effect on chloride channels (43). This could further contribute to the enhanced cell death upon sorbitol treatment.

Overall, our results suggest that up-regulation of CLCA2 in response to low humidity/hyperosmolarity promotes the integrity of the epidermal sheet, thereby allowing cell survival. This seems to be similar for mouse and human keratinocytes, although the consequence of CLCA3A2 knockdown in mouse keratinocytes could not be further studied because of rapid death of the knockdown cells.

Our results also demonstrate that K5-R1/R2 mice serve as a useful model to study the effect of humidity on homeostasis of skin with a compromised barrier, which is a hallmark of AD and occurs in response to different genetic and/or immunological alterations. Proteins that are overexpressed in AD and in K5-R1/R2 mice, such as K6, K16, and IL-36β, were down-regulated by high humidity in the mouse model. We found overexpression of CLCA2 in affected skin of AD patients. This finding suggests that up-regulation of CLCA2 may be a compensatory and protective response in AD patients to prevent an even more severe skin phenotype. It remains to be determined whether compounds, which further increase CLCA2 expression, or compounds/peptides that mimic the effect of CLCA2 on cell adhesion, may have a therapeutic potential for AD. However, enhanced expression of CLCA2 in a healthy epidermis may not be beneficial because overexpression of CLCA2 in HaCaT cells also enhances sorbitol-induced cell death.

Together, we have identified CLCA2 as a regulator of cell-cell adhesion in the epidermis. Our in vitro data suggest that the overexpressed CLCA2 exerts a protective function under hyperosmotic conditions, such as in AD. Future work should determine whether the other humidity-regulated proteins identified in this study are also abnormally expressed in AD and may thus serve as putative targets for therapeutic intervention.


Study design

The goal of this study was to determine the effect of humidity on chronic skin inflammation at the histological, cellular, and molecular level. Mice with an impaired epidermal barrier, which develop dermatitis with similarities to AD, were used. Data obtained by quantitative proteomics analysis in the mouse model were validated using human keratinocytes grown in 2D and 3D cultures, as well as biopsies from AD patients. For all experiments, we used the largest possible sample size, and there was no exclusion of any animal. All analyses of mouse samples except proteomics were performed blinded with regard to genotype and treatment of mice.

Mice and humidity conditions

Mice were housed under optimal hygiene conditions and received food and water ad libitum. Mouse maintenance and all animal experiments had been approved by the veterinary authorities of Zurich, Switzerland. K5-R1/R2 and control mice on the C57BL/6 genetic background (6) were maintained at a relative humidity of about 50 to 55%. For experiments, they were exposed to 40 or 70% relative humidity in TSE PhenoMaster cages (TSE Systems GmbH). The temperature was maintained at 22° to 25°C. For UVB radiation, mice were anesthetized by inhalation of 5% isoflurane, shaved, and irradiated with UVB (100 mJ/cm2).

Human skin samples

Skin biopsies were obtained under local anesthesia from the upper extremities of adult AD patients and healthy controls as part of an investigator-initiated clinical study to determine the molecular signatures of AD. Procedures were conducted according to the Declaration of Helsinki principles. Informed written consent was obtained from all participants under a protocol approved by the local ethics board at the University Hospital Schleswig-Holstein, Campus Kiel, Germany (reference: A100/12). Inclusion criteria were a dermatologist-confirmed diagnosis of active AD according to standard criteria (3). Exclusion criteria were the presence of any other chronic skin disease, systemic treatment with immune-efficient medication ever, and topical treatment within 1 week before material sampling. All patients were negative for the four most common filaggrin mutations in the European population.

Cell-cell adhesion assay

HaCaT cells (provided by P. Boukamp, Düsseldorf, Germany) (ca. 3 × 105 cells per six-well plate) were seeded, grown to confluency, and pretreated as described (44). Cells were washed with phosphate-buffered saline (PBS) and incubated with 2 ml per well of dispase II [2.4 U/ml diluted in Dulbecco’s modified Eagle’s medium (DMEM)] for 30 min at 37°C. Detachment of the monolayers was facilitated by orbital rotation of the plates (150 rpm) for 5 min at 37°C (low force). For high-force treatment, sheets in 3 ml of PBS in a 15-ml tube were subjected to 50 inversions. Resulting fragments and their size were counted with ImageJ.

Organotypic skin culture

Organotypic 3D cultures were established as described previously (45). Briefly, HFK (provided by H.-D. Beer, Department of Dermatology, University of Zurich) or HaCaT cells were seeded onto bovine collagen type I gels (3 mg/ml; Advanced BioMatrix Inc.) containing 5 × 105/ml human fibroblasts (provided by H.-D. Beer, Department of Dermatology, University of Zurich) in transwell plates. Cells were lifted to the air-liquid interface 24 hours after seeding, and supplemented media were added to the outside only. After lifting, cells were left to differentiate for up to 14 days. For dry cultures, silica gel was used as a desiccant, and for sorbitol stimulation, DMEM high-glucose medium (Sigma-Aldrich) was exchanged with medium containing 200 mM sorbitol (Sigma-Aldrich).

Proteomics analysis

Frozen epidermal samples were homogenized and subjected to iTRAQ (AB SCIEX) labeling. iTRAQ labels were assigned to samples according to the principles of randomized block design (46). Upon peptide fractionation, they were subjected to mass spectrometry analysis. Peptides were analyzed on a Q Exactive Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Scientific) coupled to an ultra high-pressure nano-LC system (EASY nLC 1000, Thermo Scientific). Details of the experimental procedure and data analysis are described in Supplementary Materials and Methods.

Statistical analysis

Statistical analysis was performed using PRISM software v5 (GraphPad Software Inc.). Mann-Whitney U test was used for comparison between two different groups. Error bars represent SD in all graphs. All cell culture data show biological replicates and were obtained from studies conducted on at least three different days with independent batches of cells, except data shown in figs. S5 and S6, which were obtained with the same batch of cells in one experiment. Statistical analysis of the proteomics data is described in Supplementary Materials and Methods. Raw, individual subject-level data are shown in table S3.


Materials and Methods

Fig. S1. High humidity decreases epidermal thickness, keratinocyte proliferation, and the number of cutaneous immune cells in aged mice.

Fig. S2. Rapid reversal of the skin phenotype of K5-R1/R2 mice at high humidity.

Fig. S3. iTRAQ labeling scheme for quantitative proteomics and Human Protein Atlas data for tissue-specific expression of CLCA proteins.

Fig. S4. Induction of CLCA2, but not of RPS19, expression by hyperosmotic stress and verification of the functionality of signaling inhibitors.

Fig. S5. Long-term CLCA2 knockdown does not induce EMT in HaCaT cells.

Fig. S6. Overexpression of CLCA2 induces apoptosis in HaCaT cells.

Fig. S7. Overexpression of CLCA2 in additional AD patients.

Fig. S8. Schematic representation of CLCA2 regulation and function in dry skin.

Table S1. List of 1920 proteins identified and quantified in epidermal samples from K5-R1/R2 and control mice at low and at high humidity.

Table S2. List of all 23 proteins whose increased abundance in K5-R1/R2 animals was reverted to normal amounts by high humidity.

Table S3. Raw data.

References (4855)


Acknowledgments: We thank A.-K. Müller, J. Yang, L. Maddaluno, B. Siegenthaler, S. Yang, A. Fernandes, and R. Muff (previously or currently at ETH Zurich) for invaluable experimental help; J. Partanen (University of Helsinki) and D. Ornitz (Washington University, St. Louis) for Fgfr1 or Fgfr2 floxed mice, respectively; A. Ramirez and J. Jorcano (Centro para Investigaciones Energéticas, Medioambientales y Tecnológicas, Madrid) for K5-Cre mice; P. Boukamp (University of Düsseldorf, Germany) for HaCaT keratinocytes; and H.-D. Beer (University of Zurich, Switzerland) for HFKs and primary human fibroblasts. Funding: This work was supported by grants from the Swiss National Science Foundation (310030_132884 and 31003A_169204 to S. Werner), ETH Zurich and Leopoldina postdoctoral fellowships (to K.S.), and a doctoral fellowship from the Ernst Schering Foundation (to J.S.). Author contributions: K.S., M.M., and J.S. designed and performed the experiments and analyzed the data. U.a.d.K. and T.K. designed and analyzed the proteomics experiments. U.W. and S. Weidinger provided human skin sections of healthy donors and AD patients. S. Werner designed the study, wrote the manuscript, and provided the funding. All co-authors made important suggestions to the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium ( via the PRIDE (PRoteomics IDentifications) partner repository (47) with the data set identifier PXD007124.

Stay Connected to Science Translational Medicine

Navigate This Article