Transgenic kallikrein 5 mice reproduce major cutaneous and systemic hallmarks of Netherton syndrome

KLK5 is one of several proteases known to be overactive in NS, but its precise role in the disease remains to be defined (Descargues et al., 2005, 2006). To investigate the consequences of elevated KLK5 activity in vivo, we developed a transgenic murine model overexpressing human KLK5 under the control of the human involucrin promoter to target transgene expression in the GR of murine epidermis. The cDNA encoding human KLK5 full length sequence was fused to the FLAG tag and cloned downstream of the human involucrin promoter (Fig. 1 A). Animals carrying the transgene were identified using a PCR-based genotyping strategy as shown in Fig. 1 B. Western blot analysis using an anti-human KLK5 antibody confirmed expression of the exogenous protein in epidermal extracts of Tg-KLK5 animals (Fig. 1 C). We identified this protein as being KLK5-Flag by Western blot analysis with an anti-Flag antibody (Fig. 1 E). To confirm correct targeting of KLK5 to the GR, immunostaining of skin sections using the anti–human KLK5 antibody was performed. Murine Klk5 was detected at the GR–SC interface in WT epidermis, indicating cross reactivity of the human antibody with murine Klk5. In Tg-KLK5 epidermis, KLK5 staining was increased in the GR and SC, and faint in suprabasal cells (Fig. 1 D). Detection of the FLAG epitope in Tg-KLK5 epidermis showed predominant expression of exogenous KLK5 in the GR and SC, and confirmed restricted expression of the transgene to the upper epidermis (Fig. 1 F).

NS patients classically present at birth with scaling and generalized erythroderma (Hausser and Anton-Lamprecht, 1996). At birth, Tg-KLK5 animals were slightly smaller than WT littermates and showed mild erythroderma (Fig. 2 A). Tg-KLK5 mice also showed significantly lower body mass than WT littermates (Fig. 2 A). To investigate epidermal barrier function in Tg-KLK5 mice, we first examined the ability of the skin to prevent penetration of an external dye solution in a whole-mount assay. Tg-KLK5 neonates exhibited more patches of blue staining compared with WT, mainly located on the neck, abdomen, muzzle, and ears, suggesting that the skin barrier was defective (Fig. 2 B). To measure the extent of the skin barrier defect, we tested the ability of the skin to prevent fluid loss by measuring transepidermal water loss (TEWL). Tg-KLK5 neonates showed significantly higher TEWL compared with wild-type (Fig. 2 B).

During the first week of life, Tg-KLK5 mice remained smaller than WT littermates with reduced body weight (Fig. 2 C). They developed exfoliative erythroderma and scaling of their entire body as shown at day 8 after birth in Fig. 2 (C and D). Their hair growth was also delayed (Fig. 2, C and D). TEWL measurement revealed that the difference in barrier function between WT and Tg-KLK5 littermates had increased (Fig. 2 D). Although the severity of the phenotype was variable among Tg-KLK5 littermates, growth delay, erythroderma, scaling, and sparse fur were observed in all animals at 1 mo of age (Fig. 2 E). To examine this further, we analyzed disease severity using an adapted Netherton area severity assessment (NASA) scheme which is used in the clinic for NS patients (Oji et al., 2005). Among Tg-KLK5 mice at 1 mo of age, extension of the lesions varied from localized on the head and trunk to more generalized lesions (head, trunk, back, and limbs). The severity of erythema varied between mild and severe, scaling between moderate and severe, and lichenification (of the ears) between moderate and severe. Adult Tg-KLK5 mice showed normal size and body weight but developed intense and continuous scratching (varying from moderate to severe), resulting in chronic and extensive erosions and crusts (Fig. 2 F, left). Another feature of NS is hair defects (brittle and sparse hair), which we could examine in Tg-KLK5 mice as they survive the neonatal period. Body hair appeared thinner, lusterless, and spiky with flakes of dry skin on the fur. Cutting hair on the back revealed scaling underneath the fur (Fig. 2 F, right). Additionally, vibrissae were reduced, disoriented, bent, and short at birth (Fig. 2 G). Contrary to body hair which showed a delay in growth but appeared almost normal in nonlesional adult skin, whiskers stayed short and sparse as shown at 1 yr of age (Fig. 2 G). Finally, sub-cutaneous lumps were noticed around the neck and on the legs of Tg-KLK5 mice (arrow; Fig. 2 H). These corresponded to enlarged lymph nodes, which are examined in further detail below (see also Fig. 7).

In NS patients and Spink5-deficient mice, LEKTI deficiency leads to unopposed activity of several proteases, including KLK5, KLK7, and ELA2 (Descargues et al., 2005, 2006; Bonnart et al., 2010). To study whether overexpression of KLK5 led to global changes in skin proteolytic activity, we next analyzed protein extracts of adult skin by casein gel zymography. Several bands of proteolytic activity were detected and found to be enhanced in Tg-KLK5 extracts (31, 28, 23, and 20 kD; Fig. 3 A). These bands have previously been identified as KLK5 (31 kD; Descargues et al., 2005), ELA2 (28 kD; Bonnart et al., 2010), and KLK7 (23–20 kD; Descargues et al., 2005), whereas KLK14 has been shown to comigrate with KLK7 (Stefansson et al., 2006). These data confirmed that heterologous KLK5 was active in Tg-KLK5 epidermis and led to the activation of several additional protease targets.

To further explore the in vivo proteolytic activation cascade propagated by KLK5, we used peptide substrates known to be cleaved by downstream KLK proteases (KLK7 and KLK14; de Veer et al., 2012, 2013). Cleavage of these substrates was elevated by more than twofold in Tg-KLK5 skin compared with WT (Fig. 3 B). Although we cannot entirely exclude substrate processing by other proteases, these results demonstrate that both trypsin- and chymotrypsin-like proteolytic activities are increased in Tg-KLK5 skin.

Changes in MMP activity were also analyzed using gelatin gel zymography. Two groups of bands were detected, the lowest band for each being at 62 (MMP2) and 92 kD (MMP9). Therefore, it appeared that MMP2 activity was slightly lower and MMP9 activity was enhanced in Tg-KLK5 samples (Fig. 3 C). The MW of these bands together with their sensitivity to EDTA identifies them as pro-, intermediate, and active forms of MMP2 and MMP9.

In situ zymography using BODIPY-FL-casein as a substrate showed that epidermal proteolytic activity was confined to the SC in WT skin, and that this activity was increased and extended to the GR in Tg-KLK5 epidermis (Fig. 3 D). These results confirmed that heterologous KLK5 was active in the most differentiated layers of Tg-KLK5 epidermis. Using BODIPY-FL-elastin as a substrate for in situ zymography, ELA activity was restricted to the GR and SC in WT epidermis, whereas it was slightly elevated and extended to suprabasal layers in Tg-KLK5 (Fig. 3 E). FITC-gelatin in situ zymography showed low activity in WT epidermis, whereas activity was considerably increased in the GR and SC of Tg-KLK5 (Fig. 3 F). In addition to increased proteolytic activity in the epidermis of Tg-KLK5, caseinolytic, elastinolytic, and gelatinolytic activities were also increased in the dermis of these animals (Fig. 3, D–F). Overall, these data show that expression of hKLK5 in the GR of Tg-KLK5 mice is sufficient to enhance KLK7, KLK14, and ELA2 activities in the epidermis, identifying KLK5 as an important contributor to the NS proteolytic cascade.

NS patient skin shows characteristic acanthosis, hyperkeratosis, and SC detachment at the interface with the GR through desmosomal degradation. Histological analysis of Tg-KLK5 epidermis revealed intercellular separation at the GR-SC boundary and detachment of the SC from the GR (Fig. 4 A, top). Compared with WT, Tg-KLK5 epidermis constantly showed marked acanthosis, papillomatosis and hyperkeratosis before SC detachment. Interestingly, hair follicles were significantly reduced in number, hyperplastic and disorganized (Fig. 4 A, bottom). On electron microscopy sections, abnormal detachment of cells was evident in Tg-KLK5 epidermis. Ultrastructural analysis showed that desmosomes were not cohesive at the interface between the GR and SC (Fig. 4 B). These structures were asymmetrically split with both dense plaques staying assembled. Contrary to Spink5^−/− skin in which many remnant nuclei and keratohyalin granules were observed in corneocytes (Descargues et al., 2005), only mild parakeratosis, and no keratohyalin granules in the SC were observed after histological or electron microscopy analyses (Fig. 4 A).

Several studies have shown that KLK5 is able to degrade desmoglein-1 (Dsg-1) and desmocollin-1 (Dsc-1) in vitro, and therefore is regarded to contribute to the detachment of superficial corneocytes during desquamation (Suzuki et al., 1993; Brattsand and Egelrud, 1999; Caubet et al., 2004). We next analyzed the expression of the major desmosomal components including Dsg-1, Dsc-1, and desmoplakin (Dsp). Immunostainings against Dsg-1 and Dsc-1 on WT epidermal sections revealed labeling limited to the cell border in the upper epidermis. In Tg-KLK5 epidermis, cell dissociation correlated with a marked reduction of Dsg-1 and Dsc-1 stainings in suprabasal epidermis compared with WT (Fig. 4, C and F). Western blot analysis confirmed a lower amount of Dsg-1 in epidermal extracts from Tg-KLK5 mice (Fig. 4 D). Transcript levels of Dsg-1 analyzed by quantitative RT-PCR were not reduced in Tg-KLK5 compared with WT mice, supporting the notion that reduced Dsg-1 expression resulted from proteolytic degradation (Fig. 4 E). In Spink5^−/− mice, as well as in NS patients, we have previously shown that increased proteolytic activity leads to SC detachment at the GR–SC interface through degradation of Dsg-1, Dsc-1, and Dsp (Descargues et al., 2005, 2006). However, in Tg-KLK5 epidermis, Dsp was not affected by KLK5 hyperactivity (unpublished data). Altogether, these data provide in vivo evidence that KLK5 hyperactivity initiates desmosomal component degradation and detachment of the SC from the GR.

In addition to premature desquamation, NS skin shows impaired terminal differentiation with lipid abnormalities (Descargues et al., 2005, 2006; Bonnart et al., 2010). We next investigated epidermal terminal differentiation markers including involucrin, loricrin, and Flg in the epidermis of Tg-KLK5 mice. Involucrin, which is an early marker of cornification, was expressed in the upper spinous layer and in the GR of WT epidermis (Fig. 5 A). In Tg-KLK5 epidermis, involucrin showed the same pattern of expression but appeared to be extended due to acanthosis (Fig. 5 A). Immunostaining of loricrin, a later marker of cornification, was restricted to the upper GR and was not different between WT and Tg-KLK5 (Fig. 5 B). Flg, which is mainly present in the GR and SC of WT epidermis, showed the same pattern of expression in Tg-KLK5 epidermis (Fig. 5 C), unlike Flg over-degradation in Spink5^−/− and Tg-ELA2 epidermis (Descargues et al., 2005; Bonnart et al., 2010). By Western blot analysis, no difference could be detected in Flg processing between WT and Tg-KLK5 (Fig. 5 D). We conclude that the skin barrier defect in Tg-KLK5 mice is not associated with an abnormal differentiation pattern.

In adult WT and Tg-KLK5 epidermis, nile red staining revealed uniform, linear lipophilic structures corresponding to corneocyte intercellular spaces. In WT, staining was thin and corresponded to the SC. In Tg-KLK5, multiple layers were seen, corresponding to hyperkeratosis (Fig. 5 E). Contrary to Spink5^−/− and Tg-ELA2 skin, no lipid droplets were observed in Tg-KLK5. Oil Red O, specific for free fatty acids, triglycerides, and esterified cholesterol showed staining only in the SC for both WT and Tg-KLK5 animals with no abnormal deposition in the skin (Fig. 5 F). Filipin, specific for unesterified cholesterol, intensely stained the SC in both WT and Tg-KLK5 animals (Fig. 5 F). Increased intensity of staining in Tg-KLK5 for Oil Red O and filipin could be attributable to marked hyperkeratosis (Fig. 5, F and G).

As epidermal lipid composition analysis requires isolation of the epidermis from the dermis, we used skin samples from 8-d-old mice, before hair emerges from the skin, for high performance-thin layer chromatography (HP-TLC). At this stage of analysis, hyperkeratosis is not as pronounced as in adult skin. No difference in lipid composition for triglycerides, free fatty acids, cholesterol, and ceramides was observed between WT and Tg-KLK5 at day 8 (Fig. 5 H). Overall, these results demonstrate that KLK5 overexpression does not alter terminal differentiation reflected by differentiation marker expression or lipid abnormalities.

Uncontrolled proteolytic activity in NS leads to persistent activation of inflammatory signaling pathways in the epidermis. In NS epidermis, we previously uncovered a cell-autonomous process in which KLK5 directly activates PAR2 and induces NF-κB–mediated overexpression of thymic-stromal lymphopoietin (TSLP), intracellular adhesion molecule (ICAM-1), TNF, and IL-8 (Briot et al., 2009). Furthermore, Spink5^−/−-grafted skin is rich with proinflammatory and proallergic molecules and is highly infiltrated by activated mast cells and eosinophils (Briot et al., 2009). Tg-KLK5 nonlesional skin reproduced similar abnormalities, including infiltrates of T cells, mast cells and presence of eosinophil granulocytes in the dermis whereas none could be seen in WT (Fig. 6, A–C).

Disruption of the epidermal barrier stimulates elevated expression of an array of proinflammatory cytokines and activates signaling pathways that modulate keratinocyte differentiation and the immune response. To profile the proinflammatory environment induced by KLK5 hyperactivity, we performed a targeted transcript analysis using qRT-PCR. First, we examined the expression of prominent proinflammatory and proallergic cytokines in WT and Tg-KLK5 nonlesional skin. Expression of the pro-Th2 cytokine Tslp, was significantly increased in Tg-KLK5 skin, both at the mRNA and protein level (Fig. 6, D and E). Additionally, Tnf-α, Il-1β, and Icam-1 were all significantly increased at the mRNA level (Fig. 6 E). We next investigated the expression of molecules which are known to attract immune cells to the skin (Fig. 6 F). CCL17/TARC and CCL22/MDC are major pro-Th2 mediators that can recruit proallergic Th2 cells via the chemokine receptor CCR4 to the skin (Homey and Zlotnik, 1999). In Tg-KLK5 skin, Ccl17 and Ccl22 mRNA levels showed a slight increase (Fig. 6 F). In contrast, Ccl8 (known to attract actors of allergy and inflammation) and Ccl20 (a chemoattractant for Th17 cells) showed significantly elevated expression in Tg-KLK5 skin (Fig. 6 F). In addition, Tg-KLK5 skin showed elevated expression of Il-6, Il-18, and Il-23 (Fig. 6 G).

Among the allergy/atopy-associated Th1/Th2 and Th17/Th22 cytokines, Il-4, Il-17, and Il-22 mRNA levels showed a significant increase in the skin of Tg-KLK5 mice (Fig. 6 H). Additionally, genes known to be up-regulated by Th17 such as Defb4 and Cxcl1 (Nograles et al., 2008) showed elevated expression at the mRNA level in Tg-KLK5 skin (Fig. 6 I). IL-17 and IL-22 are known to up-regulate the expression of S100A7, S100A8, and S100A9 in human keratinocytes (Boniface et al., 2005; Nograles et al., 2008). Increased levels of Il-17 and Il-22 transcripts in Tg-KLK5 skin coincided with significant increases in S100a7, S100a8, and S100a9 mRNA expression (Fig. 6 I). Finally, the expression of genes up-regulated by Th1/IFN-γ such as Ccl5, Cxcl9, Cxcl10, and Cxcl11 were all considerably reduced in Tg-KLK5 skin (Fig. 6 I).

Collectively, these data demonstrate that dysregulated KLK5 activity in the skin is sufficient to induce cutaneous allergic inflammation. This involves up-regulation of Tslp, Icam-1, and Tnf-α and the production of chemokines known to attract mast cells, eosinophil granulocytes, and Th2/Th17/Th22 cells involved in allergy and inflammation.

NS is a severe disease that is associated with a broad spectrum of severe atopic manifestations with high IgE levels. TSLP has been shown to be elevated in the serum of children with atopic dermatitis (Lee et al., 2010). Increased Tslp expression in Tg-KLK5 skin led us to investigate the expression of this major pro-Th2 cytokine in the serum of adult Tg-KLK5 mice. Tslp levels were significantly elevated in the serum of Tg-KLK5 as compared with WT mice (Fig. 7 A). Additionally, we found significantly increased levels of total IgE in Tg-KLK5 serum, and elevated total IgG, mainly due to increased IgG1 (Fig. 7 B). We next examined the morphology of lymphoid tissues including the spleen, thymus, and draining lymph nodes (cervical and inguinal). The spleen and lymph nodes were enlarged in Tg-KLK5 mice, whereas the thymus appeared slightly smaller (Fig. 7 C). Lymph node enlargement resulted from hyperplasia as shown by the increase in the number of cells in inguinal nodes (Fig. 7 D).

The pro-Th2 cytokine TSLP, which is highly expressed in the skin and the serum of Tg-KLK5 mice, is known to induce a proallergic microenvironment and to promote the differentiation of T helper (Th) 0 cells into proallergic Th2 cells (Soumelis et al., 2002). This led us to investigate whether T cells isolated from lymph nodes were Th1, Th2, and/or Th17 biased in Tg-KLK5. Phenotyping the cell populations from draining lymph nodes revealed an increased number of B cells, monocytes, dendritic cells, and T lymphocytes (unpublished data). Of the CD4⁺ T cells isolated from lymph nodes, a higher proportion were IL-4⁺, IL-13⁺, TNF-α⁺, and IL-17⁺–producing T cells as compared with WT (Fig. 7 E). Collectively, these data establish that the effects of KLK5 hyperactivity are not restricted to the skin. Indeed, overexpression of the protease specifically in the epidermis led to the production of allergy markers found in the serum and to the expansion of activated T cells in draining lymph nodes, as a consequence of skin inflammation and allergy.

Netherton syndrome (NS) is a rare autosomal recessive skin disease with severe skin inflammation and scaling, a specific hair shaft defect and constant allergic manifestations. We and others previously described Spink5-deficient mice, which display key features of NS but die shortly after birth (Yang et al., 2004; Descargues et al., 2005; Hewett et al., 2005). Here, we developed a new model of NS by overexpressing KLK5 in the epidermis. These mice allowed us to both decipher the role of KLK5 hyperactivity in NS pathogenesis in vivo, and to provide a new and viable model to study further aspects of NS physiopathology, particularly those that become prominent after birth. We provide evidence here that overexpression of KLK5 in the epidermis leads to the development of the major NS hallmarks, confirming the role of KLK5 as a key contributor to the NS proteolytic cascade. In addition, we provide a new and viable mouse model recapitulating the hallmarks of NS.

In NS, the absence of LEKTI has been shown to result in unopposed activity of several proteases (Descargues et al., 2005, 2006; Bonnart et al., 2010). We show here that overexpression of human KLK5 in murine epidermis triggers major enhancement of the global proteolytic activity in the skin that overcomes the inhibitory capacities of Lekti. As our construct expresses KLK5 as a proprotease, this implies that it has been activated in vivo. It has recently been shown that matriptase can activate pro-KLK5 and pro-KLK7 in vitro (Sales et al., 2010). KLK5 has been shown to activate pro-KLK7, pro-KLK14, and pro-ELA2, as well as to amplify cascade initiation by activating pro-KLK5 in vitro (Brattsand et al., 2005; Bonnart et al., 2010). Using gel zymography and peptide substrates, we have shown that not only KLK5 activity but also KLK7, KLK14, and ELA2 activities were increased in Tg-KLK5 skin. These results provide in vivo evidence that KLK5 can indeed activate KLK7, KLK14, and ELA2 as described by in vitro experiments (Brattsand et al., 2005; Bonnart et al., 2010). In addition, we observed an increase in MMP9 activity in Tg-KLK5 skin. It is to our knowledge the first description of a functional link between MMP9 and KLK5, which is potentially mediated by PAR2 activation as demonstrated in transgenic CAP1/Prss8 Par2-deficient mice (Frateschi et al., 2011).

NS patients suffer from a profound skin barrier defect. Histopathologic examination of NS skin reveals a psoriasiform epidermis, associated with hyperkeratosis, hyperplasia of the subcorneal epithelium (acanthosis), and varying degrees of epidermal invaginations into the dermis (papillomatosis; Hausser and Anton-Lamprecht, 1996). These observations are consistent with the fact that skin barrier defects can induce compensatory hyperproliferative mechanisms resulting in hyperkeratosis and acanthosis. Another feature of NS epidermis is premature detachment of the SC from the GR as a result of desmosome cleavage (Descargues et al., 2005, 2006). These features are all seen in Tg-KLK5 skin, including marked acanthosis, papillomatosis, and hyperkeratosis before SC detachment. Interestingly, IL-22, which is increased in chronic stages of atopic dermatitis induces epidermal hyperplasia (Nograles et al., 2008). The increased Il-22 expression observed in Tg-KLK5 skin could thus contribute to acanthosis. In Tg-KLK5 skin, the SC detachment occurs at the SC–GR interface as a result of desmosome cleavage through Dsg-1 and Dsc-1 degradation as previously demonstrated in NS skin (Descargues et al., 2005, 2006). NS skin also exhibits impaired epidermal differentiation, with increased involucrin and loricrin expression, fewer keratohyalin granules, decreased Flg immunostaining in the GR and abnormal lipid structure and composition (Descargues et al., 2005, 2006; Bonnart et al., 2010). We have previously shown that ELA2 is hyperactive in NS skin and that transgenic mice overexpressing ELA2 in the GR displayed abnormal (pro-) Flg processing and impaired lipid lamellae structure (Bonnart et al., 2010). Of note, we did not observe major alterations in either epidermal terminal differentiation or Flg processing in Tg-KLK5 mice, even though ELA2 activity was increased. One possibility could be that ELA2 activity did not reach the level of activation observed in the Tg-ELA2 murine model as suggested by comparative elastin in situ zymography between Tg-ELA2 and Tg-KLK5 (unpublished data). Consistent with this, keratohyalin granules in Tg-KLK5 were still present, in contrast to Tg-ELA2 where they were absent (Bonnart et al., 2010). Recently, KLK5 was shown to directly process pro-Flg in vitro (Sakabe et al., 2013). The fact that no significant decrease in pro-Flg was observed in Tg-KLK5 animals suggests that regulation of pro-Flg processing by KLK5 is more complex and involves additional factors in vivo. The presence of Lekti in Tg-KLK5 mice could imply that KLK5 and other pro-Flg processing proteases, such as caspase 14, remained sufficiently regulated during terminal differentiation to prevent Flg overdegradation. The observation that Tg-KLK5 mice reproduce major abnormalities of NS supports an important role for KLK5 in the disease. However, not all features are observed, suggesting that additional proteases could be involved in the complete phenotype.

NS patients show failure to thrive in infancy and short stature. In a similar way, Tg-KLK5 mice displayed a lower weight at birth and a growth delay. However, this delay was only transient and mice had a nearly normal weight and size at 2 mo of age. Several KLKs, including KLK5, are known to cleave growth hormone (GH) and to be present in the pituitary gland (Komatsu et al., 2007). It has been suggested that in NS patients the absence of LEKTI leads to higher proteolytic processing of GH, which may become biologically inactive (Komatsu et al., 2007). As we do not expect transgene expression in the pituitary gland and Lekti is still present in Tg-KLK5, it is possible that other factors contribute to the growth delay in Tg-KLK5 mice. Another characteristic feature of NS patients is a specific hair shaft defect (trichorrhexis invaginata) that leads to sparse, brittle, and spiky hair. In Tg-KLK5 animals, we observed an apparent thinning of body hair, which had a lusterless and spiky appearance. As in newborn Spink5^−/− mice (Descargues et al., 2005), Tg-KLK5 whiskers were scarce, disoriented, bent, and short. Existing murine models such as Cathepsin L^−/− or Matriptase^−/− have shown that hair follicle morphogenesis is a protease-dependent process (Roth et al., 2000; List et al., 2002). This viable murine model will allow the mechanisms underlying hair follicle abnormalities in NS to be deciphered.

In NS patients, atopic features with elevated IgE are constant during infancy and adulthood. These include eczematous-like lesions, allergic asthma, allergic rhinitis, urticaria, and angiodema (Sun and Linden, 2006). Only a few immunological studies have been reported in NS patients to date (Judge et al., 1994; Stryk et al., 1999; Renner et al., 2009). In these studies, a constant finding is elevated specific IgE levels to common allergens, although a proallergic Th2 skew has not been consistently found. Studies on Spink5^−/− skin have been limited to grafting experiments on WT mice and have shown enhanced expression of proinflammatory cytokines, such as Il-1β, Tnf-α, IL-8, and Icam-1 and proallergic mediators, such as Tslp, Ccl17 (Tarc), and Ccl22 (Mdc; Briot et al., 2009). We have previously shown that increased TNF, IL-8, ICAM-1, and TSLP expression is an intrinsic property of NS keratinocytes as a result of the KLK5–PAR2–NF-κB pathway (Briot et al., 2009). The proinflammatory and proallergic cytokines overexpressed in NS epidermis play an active role in recruiting inflammatory cells to the skin, including mast cells and eosinophils whose numbers are significantly increased in Spink5^−/− skin grafts (Briot et al., 2009). These features of severe cutaneous inflammatory and atopic manifestations are prominent in Tg-KLK5 mice. Our results establish that KLK5 overactivity in murine epidermis is sufficient to induce a proallergic and proinflammatory environment.

In skin of atopic dermatitis patients, TSLP expression is associated with the migration of Langerhans cells to the draining lymph nodes. TSLP-activated dendritic cells have been shown to induce naive T cells to differentiate into “allergic” Th2 cells producing IL-4, IL-5, IL-13, and TNF (Soumelis et al., 2002). We have previously shown that NS keratinocytes show high expression of TSLP (Briot et al., 2009). Consistent with this observation, we showed that TSLP expression was also strongly enhanced in Tg-KLK5 skin. Normal human keratinocytes do not express detectable levels of TARC and MDC, contrary to NS and AD keratinocytes (Horikawa et al., 2002; Tsuda et al., 2003; Briot et al., 2009). These two chemokines can recruit Th2 cells via CCR4 to the skin, defining them as major pro-Th2 mediators (Homey and Zlotnik, 1999). In NS keratinocytes, we have previously shown that MDC and TARC are not induced by KLK5 and PAR2 activation (Briot et al., 2009). We have confirmed this observation in vivo because Tg-KLK5 skin exhibits only a slight increase of these two chemokines. However, the presence of IL-4 in Tg-KLK5 skin reflects the presence of Th2 cells. This suggests that slightly elevated Mdc and Tarc levels, when combined with the inflammatory environment are sufficient to recruit Th2 cells to the skin of Tg-KLK5 mice.

We also uncovered the presence of Th17 and Th22 cells in Tg-KLK5 skin. In inflammatory diseases, such as psoriasis, it has been shown that the presence of “inflammasome related” cytokines such as IL-1β and IL-18 could synergize with IL-23 to promote Th17 cell differentiation (Mills et al., 2013). In Tg-KLK5 skin, we found elevated expression of Il-1β, Il-18, and Il-23, which could explain the presence of Th17 cells. The treatment of healthy keratinocytes with TNF induces the expression of CCL20, a molecule involved in Th17 recruitment (Kennedy-Crispin et al., 2012). It is thus possible that enhanced Tnf-α and Il-1β production in Tg-KLK5 contributes to Ccl20 expression leading to Th17 recruitment. Of note, IL-17 has been shown to induce the secretion of proinflammatory cytokines (IL-1β and IL-6), antimicrobial peptides, and several chemokines including IL-8 and CCL20 in human keratinocytes. In addition, the induction of CCL20 by keratinocytes has been suggested as a mechanism that maintains Th17 cells in skin (Perera et al., 2012). We have also found increased expression of Defb4, Cxcl1, S100a7, S100a8, and S100a9 in Tg-KLK5 skin, which are genes known to be induced by Il-17 and Il-22. A similar inflammatory profile is found in atopic dermatitis in which, in addition to Th2, Th22 and Th17 play important roles in the pathogenesis of the disease (Guttman-Yassky et al., 2013; Suárez-Fariñas et al., 2013). These observations support the contribution of Th17 and Th22 cells to skin inflammation in the Tg-KLK5 murine model.

Unlike Spink5^−/− mice, Tg-KLK5 mice are viable, which allowed us to investigate signs of allergy in the serum. We found elevated Tslp and IgE concentrations that reflect the Th2 cytokine-mediated IgE class switch of B cells. Thus, our results show that the effects of dysregulated KLK5 activity are not restricted to Tg-KLK5 skin. In Tg-KLK5 mice, continuous scratching led to alopecic, erosive, and crusty skin lesions. These observations are highly relevant to NS patients who show constant, often severe and disabling pruritus. Persistent itch causes sleeplessness and scratching, which further aggravate skin lesions. A well-established mediator of itching is PAR2 activation (Steinhoff et al., 2003). Several serine proteases capable of activating PAR2 are present in Tg-KLK5 skin, including KLK5, KLK14 in the epidermis, and tryptase from infiltrating mast cells. Tg-CAP1/Prss8 mice display a similar itchy phenotype, which could be rescued by PAR2 knock-down (Frateschi et al., 2011). In Tg-KLK7 mice, which develop chronic dermatitis, itching does not respond to antihistamine therapy and is also potentially attributable to PAR2 activation (Hansson et al., 2002).

In summary, we have generated a new and viable murine model that recapitulates the major cutaneous and systemic hallmarks of NS and validates KLK5 as a promising target for NS treatment. Accordingly, successful inhibition of KLK5 has the potential to reduce the skin barrier defect as well as features of inflammation and allergy. Furthermore, this model will allow the study of further aspects of NS pathophysiology that are currently not well understood, such as growth delay, immune system dysregulation, and hair defects. It will also be instrumental for preclinical testing of therapies for NS and related diseases that involve a skin barrier defect and severe atopic manifestations. Indeed, atopic dermatitis shows several similarities with NS, including increased skin barrier permeability and elevated TSLP, which lead to a Th2-biased response and high serum IgE. Collectively, our results show that the Tg-KLK5 murine model shares important similarities with human atopic dermatitis. Therapies aiming at controlling these dysregulated pathways could be evaluated in vivo using the Tg-KLK5 murine model.

This work was approved by the Commission de Génie Génétique (01–10-08, agreement number 4858) and all experiments were performed in accordance with the required guidelines and regulations. The full-length cDNA encoding human KLK5 (I.M.A.G.E. clone 100020682) was amplified by RT-PCR. The FLAG octopeptide epitope was added to the 3′ end of the resulting product followed by NotI restriction sites to both ends. After digestion with NotI, this construct was inserted into the IVL promoter vector (pH3700-pL2; gift from J. Segre, National Human Genome Research Institute, Bethesda, Maryland; Carroll and Taichman, 1992; Carroll et al., 1993) and validated by sequencing. The transgene (comprising the IVL promoter and human KLK5_FLAG construct) was excised from pH3700-pL2 by digestion with SalI, purified and injected into the pronuclei of fertilized oocytes from (C57BL/6 x CBA) F1 mice, and finally implanted into pseudopregnant females. First generation offspring were genotyped by isolating genomic DNA from mouse tail samples using the Nucleospin Tissue kit (Macherey-Nagel) and performing PCR using allele-specific primers designed to detect the transgene or an internal PCR control. The founder mouse was crossed and backcrossed with wild-type C57BL/6 animals to establish the transgenic line according to the International Committee on Standardized Genetic Nomenclature for Mice. Transgenic mice were named Tg-KLK5 for the following study.

TEWL was measured using an EP1 evaporimeter (ServoMed) as previously described (Descargues et al., 2005).

Euthanized newborn mice were immediately subjected to graded methanol dehydration and subsequent rehydration as previously described (Ekholm et al., 2000). Pups were then washed in PBS for 1 min, stained in 0.1% toluidine blue (Sigma-Aldrich) for 1 h at room temperature, destained for 15 min in PBS, and analyzed.

Murine skin was crushed in RLT buffer (QIAGEN) using an Ultra-Turax, after which RNA was isolated using the RNeasy mini kit (QIAGEN). Complementary DNA was synthesized by reverse transcription from 1 µg total RNA using M-MLV Reverse transcription (Invitrogen). Quantitative real-time PCR was performed on an ABI prism 7500 Sequence Detection system (Applied Biosystems) using qPCR MesaGreen Mastermix (Eurogentec). The expression of target genes was normalized to murine hypoxanthine phosphoribosyl transferase (Hprt). Primer sequences can be obtained on request from the authors. Data were analyzed using 7500 software v2.0.6. (Applied Biosystems).

Antibodies used for Western blot analysis and immunohistochemistry were: Actin (A5441; Sigma-Aldrich), Flag (F7425; Sigma-Aldrich); Desmocollin-1 (sc18115; Santa Cruz Biotechnology, Inc.); Desmoglein-1 (sc20114; Santa Cruz Biotechnology, Inc.); Flg (PRB-417P; Covance); KLK5 (ab7283; Abcam); Involucrin (sc15230; Santa Cruz Biotechnology, Inc.); Loricrin (PRB-145P; Covance); TSLP (AF555; R&D Systems); and CD3 (ab5690; Abcam).

For Western blot analysis, skin was crushed in protein extraction buffer (50 mM Tris-HCl, pH 8.5, 5 mM EDTA, 150 mM NaCl, 0.1% Nonidet-P40, and Protease Inhibitor Cocktail Tablets [Complete, Roche]) using an Ultra-Turax. Lysates were clarified by centrifugation (13,000 g, 4°C for 30 min) to remove insoluble material. For Flg Western blot analysis, proteins were extracted using 50 mM Tris-HCl, pH 8.0, 8 M urea, 10 mM EDTA, and Protease Inhibitor Cocktail Tablets. Protein content in each sample was measured by Bradford assay (Bio-Rad Laboratories) according to the manufacturer’s instructions. Protein fractions were mixed with Laemmli buffer, incubated for 5 min at 90°C, and then separated by SDS-PAGE. For KLK5, Flag, and Dsg-1 Western blots, 5 µg of proteins (for Flg, 1.5 µg of proteins) were loaded on a 4–12% precast gel (Invitrogen). After migration, proteins were transferred to Hybond-ECL membranes (GE Healthcare). After incubation with primary and secondary antibodies, enhanced chemiluminescence detection was performed as recommended by the manufacturer (Thermo Fischer Scientific).

Skin cryosections (5-µm thickness) were mounted on glass slides, rinsed with 2% Tween 20 in PBS and incubated at 37°C overnight with 10 µg ml⁻¹ BODIPY FL casein (Life Technologies), 100 µg ml⁻¹ BODIPY FL elastin (Life Technologies) or 4 µg ml⁻¹ FITC-gelatin (Life Technologies) in 50 mM Tris-HCl, pH 8.0. Sections were rinsed with PBS and visualized using a Leica TCS SP5 AOBS confocal microscope. Data were analyzed using ImageJ (National Institutes of Health) software.

For proteolytic activity assays, skin was crushed in 1 M acetic acid using an Ultra-Turax. After overnight extraction at 4°C, insoluble material was removed by centrifugation (13,000 g, 4°C for 30 min) and the supernatant was dried using a Speed-Vac. Proteins were resuspended in water overnight at 4°C and clarified by centrifugation (13,000 g, 4°C for 30 min). Protein content was determined by Bradford assay (Bio-Rad Laboratories).

For gel zymography, 10 µg of protein extracts were loaded onto casein/acrylamide copolymerized gels (15% acrylamide, 0.1% α-casein; Sigma-Aldrich) or gelatin/acrylamide copolymerized gels (10% acrylamide and 0.1% gelatin; Sigma-Aldrich) for electrophoresis. After separation by SDS-PAGE, gels were washed with 2.5% Triton X-100 (two times for 30 min) and incubated for 24 h at 37°C in reaction buffer (50 mM Tris-HCl, pH 8.0). Buffer for gelatin zymography additionally contained 10 mM CaCl₂. Gels were stained with 1% Coomassie Brilliant blue G250.

Additionally, proteolytic activity was assessed using colorimetric peptide substrates that are preferentially cleaved by different proteases. Acetic acid skin extracts (25 µg) were added to assay buffer (0.1 M Tris-HCl pH 8.0, 0.005% Triton X-100, 0.05% sodium azide) in 96-well plates (final volume 200 µl). Substrates used were 150 µM Ac-YASR-pNA (cleaved by several trypsin-like serine proteases including KLK5, KLK14 and matriptase), 150 µM KHLY-pNA (cleaved by KLK7) and 150 µM Ac-WAVR-pNA (cleaved by KLK14; de Veer et al., 2012, 2013). Plates were incubated at 37°C overnight and activity was analyzed by measuring the increase in absorbance at 405 nm compared with substrate only controls.

Skin samples were snap frozen in liquid nitrogen in Tissue-Tek O.C.T. (Thermo Fischer Scientific) or fixed in 10% buffered formalin and embedded in paraffin. Hematoxylin/eosin/safranin, toluidine blue, and luna staining were performed on paraffin-embedded sections using standard histological techniques. For antigen retrieval, paraffin-embedded sections were boiled in 10 mM citrate buffer, pH 6.0, for 20 min (Flg or TSLP immunostaining) or treated with S1700 solution (Dako) for 45 min (KLK5, Flag, Involucrin, or Loricrin immunostaining). Desmoglein-1, Desmocollin-1, and CD3 immunohistochemistry was performed on skin cryosections. Signal was detected using the appropriate horseradish peroxidase (HRP)–conjugated polymer (En Vision System; Dako).

Skin samples were fixed in 4% paraformaldehyde and 2% glutaraldehyde, rinsed in Sorensen buffer, postfixed in 2% aqueous OsO₄, dehydrated in graded ethanol solution, and embedded in epon. Ultrathin sections were counterstained with uranyl acetate and lead citrate. Examination was performed with a 1011 transmission electron microscope (Jeol Inc., Japan). Acquisition and processing were performed with an Erlangshen CCD camera (Gatan Inc.) and Digital Micrograph software (Gatan Inc.).

For nile red staining, cryosections were incubated with 2.5 µg ml⁻¹ nile red in 75% glycerol for 5 min and mounted. Filipin staining was performed on cryosections using 50 µg ml⁻¹ filipin in 75% glycerol and 25% water (30 min incubation). Slides were rinsed twice with PBS and mounted with aqueous mounting medium. For Oil Red O staining, cryosections were incubated for 10 min in 60% isopropanol followed by staining with 0.3% (wt/vol) Oil Red O in 60% isopropanol (20 min). Sections were subsequently rinsed with water and mounted with aqueous mounting medium. Filipin and nile red stainings were visualized using a Leica TCS SP5 AOBS confocal microscope and data were analyzed using ImageJ. Oil Red O stainings were visualized using a white-light microscope. Lipid composition analysis was performed using HP-TLC as previously described (Bonnart et al., 2010).

Sera were analyzed by ELISA for TSLP, total IgG, IgG1, IgG2a, and IgE levels. Serum TSLP was measured using the murine TSLP Quantikine ELISA kit (R&D Systems) according to the manufacturer’s instructions. For immunoglobulin analysis, ELISA plates were coated overnight with rat anti–mouse IgG (2.5 µg ml⁻¹, LO-MG-COC-2), IgG1 (1.25 µg ml⁻¹, LO-MG1-13) or IgG2a (5 µg ml⁻¹, LO-MG2a-7) antibodies (Southern Biotech). Plates were blocked with 1% BSA in PBS and incubated with serial dilutions of mouse sera, followed by anti-Ig PO HRP-conjugated secondary (1/2000, Southern Biotech). Signal was developed by incubation with 3,3′,5,5′–tetramethylbenzidine substrate (Fluka). For IgE, plates were coated overnight with 3 µg ml⁻¹ rat anti–mouse IgE (LO-ME-3), blocked with 1% BSA in PBS, and incubated with serial dilutions of mouse sera, followed by biotinylated rat anti–mouse IgE antibody (1/2000; Southern Biotech). Signal was developed by addition of streptavidin-HRP conjugate (1/2000; Interchim). All samples were analyzed in duplicate or triplicate by measuring the absorbance at 450 nm using a microplate reader (Versamax; Molecular Devices). Concentrations were calculated using standard curves generated with known dilutions of mouse IgG, IgG2a, IgG1 (Southern Biotech), and IgE (BD).

Single-cell suspensions were prepared from inguinal lymph node extracts. Surface antigen staining was performed using the following antibodies (eBioscience): anti-CD45 (clone 30-F11), anti-CD4 (clone RM4-5), anti-CD8 (clone 53–6.7), and anti-TCRγ (clone H57-597). For intracellular cytokine staining, isolated cells were activated in complete RPMI containing 0.5 µg ml⁻¹ phorbol 12 myristate 13-acetate (PMA), 0.5 µg ml^-1 ionomycin and 10 µg ml⁻¹ brefeldin A (all from Sigma-Aldrich) for 4 h. After surface staining, cells were fixed and permeabilized using Cytofix/Cytoperm solution (BD). Cells were washed in perm/wash buffer and incubated for 30 min at 4°C with different antibody combinations specific for murine cytokines (all from eBioscience) diluted in perm/wash buffer: anti-Il4 (clone 11B11), anti–IL-13 (clone eBio13A), anti-Il17 (clone eBio17B7), and TNF (clone MP6-XT22). Immunostained cells were analyzed by flow cytometry using a FACS Aria flow cytometer (BD).

Experiments were repeated at least three times and p-values were calculated using the Mann-Whitney t test. P < 0.05 was considered significant.

We are grateful to Julia Segre for providing the pH3700-pL2 vector. We thank Anne Huchenq-Champagne and Sylvie Appolinaire-Pilipenko for the transgenesis and cryoconservation platform (IFR150, Toulouse, France).

This work was supported by the French Ministry of Health, the Agence Nationale de la Recherche (ANR-08-GENO-033), and the Fondation pour la Recherche Médicale (FRM-DAL20051205066). S. de Veer received an international travel grant from the Fondation René Touraine.

The authors have no financial conflict of interest to declare.