KLK5 and KLK7 Ablation Fully Rescues Lethality of Netherton Syndrome-Like Phenotype

Netherton syndrome (NS) is a severe skin disease caused by the loss of protease inhibitor LEKTI, which leads to the dysregulation of epidermal proteases and severe skin-barrier defects. KLK5 was proposed as a major protease in NS pathology, however its inactivation is not sufficient to rescue the lethal phenotype of LEKTI-deficient mice. In this study, we further elucidated the in vivo roles of the epidermal proteases in NS using a set of mouse models individually or simultaneously deficient for KLK5 and KLK7 on the genetic background of a novel NS-mouse model. We show that although the ablation of KLK5 or KLK7 is not sufficient to rescue the lethal effect of LEKTI-deficiency simultaneous deficiency of both KLKs completely rescues the epidermal barrier and the postnatal lethality allowing mice to reach adulthood with fully functional skin and normal hair growth. We report that not only KLK5 but also KLK7 plays an important role in the inflammation and defective differentiation in NS and KLK7 activity is not solely dependent on activation by KLK5. Altogether, these findings show that unregulated activities of KLK5 and KLK7 are responsible for NS development and both proteases should become targets for NS therapy.


Author Summary
Netherton syndrome (NS) is a genetic skin disorder caused by the loss of protease inhibitor LEKTI, which leads to the dysregulation of epidermal proteases and severe skin-barrier defects. In this work, we aimed to explore the molecular mechanisms underlying this disease using a novel mutant mouse model for NS, which is based on mimicking a causative mutation known from human patients. This novel model reproduces the symptoms of NS and thus provides a useful tool to study the NS pathology in a complex in vivo environment. Most importantly, by combination of this NS-mouse model with mutant mice individually or simultaneously deficient for proteases KLK5 and KLK7, we elucidated the a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
Netherton syndrome (NS) is a life-threatening autosomal recessive disorder that affects approximately one in 200 000 newborn children [1,2]. Newborns suffering from NS exhibit congenital ichthyosiform erythroderma with scaly and peeling skin, resulting in severe disruption of epidermal barrier, which in some cases is fatal. These conditions may improve with age and older patients often show less severe ichthyosis exhibiting erythematous plaques with double-edged scales at the periphery [1][2][3]. The hair of NS patients is usually thin, fragile and the patients often develops "bamboo hair", a hair shaft defect where the distal part of the hair shaft is invaginated into its proximal part [4]. NS may be also associated with growth retardation, asthma, food allergies, and elevated serum levels of IgE [1,5].
NS is caused by mutations in SPINK5 gene (serine protease inhibitor Kazal-type 5) that encodes LEKTI (lympho-epithelial Kazal-type related inhibitor), an inhibitor of serine proteases expressed in the epidermis and other stratified epithelia [6]. Full length LEKTI consists of 15 inhibitory domains (D1-D15) and upon synthesis undergoes proteolytic processing into multiple bioactive fragments containing one to six domains with distinct inhibitory specificities [7,8]. LEKTI has been reported to inhibit several proteases including plasmin, trypsin, subtilisin A, cathepsin G, elastase, caspase-14 [9][10][11] as well as members of the family of kallikrein-related peptidases (KLK), mainly KLK5, KLK7 and KLK14 [12][13][14]. Unregulated activity of KLK5 and possibly also KLK7 is considered a major source of pathology in NS. Spink5 deficient mice show increased proteolytic activities of KLK5 and KLK7 [15], which corresponds to elevated tryptic and chymotryptic activities described in NS patients [16,17]. KLK5 also initiates a proteolytic cascade by proteolytic activation of KLK7 and KLK14, that leads to degradation of corneodesmosomal proteins desmoglein1 (DSG1), desmocollin1 (DSC1), and corneodesmosin (CDSN) [18]. Premature degradation of corneodesmosomes results in detachment of the stratum corneum (SC) and disruption of the epidermal barrier in NS patients [17]. Upregulated proteolytic activity can further contribute to skin barrier defects by abnormal processing of profilaggrin, a precursor protein which is proteolytically converted into physiologically active filaggrin monomers. Filaggrin is one of key players in maintaining skin hydratation and water retention of the epidermis [19]. Recently, KLK5 was shown to promote profilaggrin processing either via proteolytic activation of elastase 2, which cleaves filaggrin precursor proteins [20] or by direct degradation of profilaggrin [21]. In addition, previous studies using mouse models and human NS patients suggest that the unregulated activity of KLK5 contributes to the inflammatory response in the LEKTI-deficient epidermis by activation of protease-activated receptor 2 (PAR2) [22][23][24].
In this study, we revealed the in vivo roles of KLK5 and KLK7 using a set of mouse models that are simultaneously deficient for KLK5 and KLK7 on the genetic background of Netherton syndrome-like mouse model based on a mutation found in human patients. The close proximity of these genes (on the same locus) has so far prevented the generation of suitable animal models and therefore the in vivo roles of KLK5 and KLK7 could not be studied concurrently. Our study shows that individual functional ablation of KLK5 or KLK7 is not sufficient to rescue the lethal effect of Spink5 mutation. In contrast, simultaneous deficiency of both KLK5 and KLK7 completely rescues the lethality allowing adult mice to survive to adulthood with a fully functional skin barrier.

Results
A135X mutation in LEKTI causes severe skin phenotype that leads to early postnatal lethality To study NS pathology in vivo, we generated a new mouse model mimicking a causative mutation of SPINK5 gene (398delTG; p.A134X) previously described in human patients [25]. Due to the similarity between the human and murine SPINK5 nucleotide sequences, deletion of TG nucleotides at positions 402 and 403 of murine Spink5 (402delTG) produces the premature termination codon (PTC) at a similar position as described in human patients (p.A135X) ( Fig  1A). To introduce the mutation into mouse genomic DNA, we prepared TALE nucleases (TALENs) specific for the critical region of Spink5 in combination with a single stranded oligonucleotide (ssODN) carrying the desired mutation ( Fig 1B). Founders were screened for targeted incorporation of ssODN by RFLP analysis using XbaI restriction site as a marker ( Fig  1C). Heterozygous mice carrying A135X mutation (hereafter referred to as Sp5 +/A135X ) did not show any obvious phenotype and were used to obtain Sp5 A135X/A135X mice, which showed dramatic downregulation of Spink5-RNA expression (Fig 1D). The presence of PTC in the Spink5 transcript was confirmed by sequencing (S1 Fig). Sp5 A135X/A135X mice were born in normal Mendelian ratios, however they exhibited severe skin phenotype with exfoliating epidermis, predominantly localized in the abdominal and facial area ( Fig 1E) and died within 12 hours after delivery. These phenotypical features mimic NS characteristics and correspond to previously published mouse models of Netherton syndrome [15,[26][27][28].

Hyperactivity of KLK5 and KLK7 is responsible for the barrier disruption of LEKTI-deficient skin
The integrity of epidermal barrier in Sp5 A135X/A135X newborn P0 pups was analysed using the toluidine blue (TB) penetration assay and showed severe skin barrier disruption marked by penetration of TB through large areas of the body, mainly the abdomen, paws, and head. Ablation of KLK7 on the Sp5 A135X/A135X background did not improve the barrier and Klk7 -/-Sp5 A135X/A135X newborns showed similar barrier disruption to Sp5 A135X/A135X . In contrast, Klk5 -/-Sp5 A135X/A135X newborns developed less severe barrier phenotype characterised by multiple small stained patches and in Klk5 -/-Klk7 -/-Sp5 A135X/A135X pups, the barrier integrity was almost completely recovered and the mice showed TB staining only in the area of nostrils ( Fig 4A).
To understand why Klk5 -/-Sp5 A135X/A135X died at P5, mice were stained with TB at P5. Interestingly, Klk5 -/-Sp5 A135X/A135X mice showed clear penetration of the dye in the epidermis adjacent to hair shafts whereas this barrier defect was completely rescued in Klk5 -/-Klk7 -/-Sp5 A135X/A135X (Fig 4C). Detailed analysis of the epidermis using SEM revealed that P5 Klk5 -/-Sp5 A135X/A135X mice had dramatic epidermal defects manifested by defective separation of hair shafts from the surrounding tissues and subsequent loss of infundibular epidermis in the upper part of hair follicles. These defects were not observed in Klk5 -/-Klk7 -/-Sp5 A135X/A135X
Both, KLK5 and KLK7 contribute to abnormal differentiation of LEKTIdeficient epidermis Histological analysis of the epidermis from NS patients together with previously published data on LEKTI-deficient models describe an abnormally differentiated epidermis [15]. In line with these observations, analysis of non-lesional skin of P0 pups showed a reduced granular layer, acanthosis and sporadic SC detachment and parakeratosis in Sp5 A135X/A135X pups ( Fig 5A and S6 Fig). Although Klk7 -/-Sp5 A135X/A135X pups showed a similar phenotype to Sp5 A135X/A135X mice, no differentiation defects were observed in the epidermis of newborn pups apart from the occasional focal detachment of SC (Fig 5A and S6B Fig). Klk5 -/-Sp5 A135X/A135X and Klk5 -/-Klk7 -/-Sp5 A135X/A135X newborn pups exhibited well differentiated epidermal layers.
NS is also associated with systemic inflammation, allergy and elevation of IgE levels. We examined the serum levels of TNFα, Il-1β, IL-9 and IL-17 in Klk5 -/-Klk7 -/-Sp5 A135X/A135X 6 weeks old mice, however no signs of systemic inflammation were found when compared to wt mice (Fig 7C). These observations are in line with the rescue of other NS-like symptoms in adult Klk5 -/-Klk7 -/-Sp5 A135X/A135X mice.

Discussion
Netherton syndrome is a severe genetic disorder associated with unregulated proteolytic activity, caused by the absence of functional LEKTI, a protease inhibitor encoded by SPINK5 gene. In this study, we elucidated the roles of KLK5 and KLK7 proteases in the disease by genetic inactivation of these proteases on the background of a mouse model for NS. This novel model was generated by mimicking a SPINK5 p.A134X mutation found in human patients [25] and recapitulates the phenotype of previously described Spink5-deficient mouse models [15,26,27]. However, our Sp5 A135X/A135X mice in combination with the ablation of KLK5 and KLK7 reveal the complexity of the LEKTI-KLK network. We showed, that although single inactivation of KLK5 or KLK7 rescues a number of NS-like pathological manifestations, only simultaneous ablation of both proteases fully rescues the lethal phenotype of Sp5 A135X/A135X mice.
It has been proposed that the barrier defects observed in LEKTI-deficient skin are caused by proteolytic hyperactivity leading to premature degradation of corneodesmosomal proteins.
In vitro assays showed that three putative LEKTI targets are able to promote corneodesmosome degradation namely KLK5, KLK7, and KLK14 [18]. The in vitro study of the KLK proteolytic activation cascade proposed that KLK5 acts upstream from KLK7 and KLK14 and therefore, KLK5 hyperactivity should contribute to barrier defects either directly or indirectly via activation of the remaining KLKs. Indeed, significant improvement of skin-barrier defects by inactivation of KLK5 in Sp5 A135X/A135X mice was observed, however the rescue was incomplete as toluidine blue staining in Klk5 -/-Sp5 A135X/A135X mice revealed patches of disruptedbarrier distributed all over the body surface. This observation implicates the role of another protease whose activity contributes to barrier defects in the absence of LEKTI and does not depend on KLK5. This was identified as KLK7, since Klk5 -/-Klk7 -/-Sp5 A135X/A135X newborn mice did not show any major barrier defects of epidermis. Interestingly, single inactivation of KLK7 on Sp5 A135X/A135X background did not significantly improve the barrier defects. , IL-1β and IL33 in the skin of E18.5 dpc wt, Sp5 A135X/A135X , Klk5 -/-Sp5 A135X/A135X , Klk7 -/-Sp5 A135X/A135X , and Klk5 -/-Klk7 -/-Sp5 A135X/A135X analysed by qRT-PCR, n!5. Error bars represent standard deviations from mean; Data were analysed by One-way ANOVA followed by Bonferroni posthoc tests. (B) Cytokine expression in the skin of P5 wt, Klk5 -/-Sp5 A135X/A135X , Klk5 -/-Klk7 -/-Sp5 A135X/A135X and Klk5 -/analysed by qRT-PCR, n!3. Error bars represent standard deviations from mean; Data were analysed by One-way ANOVA followed by Bonferroni post-hoc tests. (C) Analysis of serum levels of TNFα, IL-1β, IL-9 and IL-17 in 6 weeks old wt and Klk5 -/-Klk7 -/-Sp5 A135X/A135X mice, n!4. Error bars represent standard deviations from mean; Means were compared with a t-test, ns means "not significant"; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. Therefore we assume that barrier properties of LEKTI-deficient neonatal epidermis are compromised mainly by direct activity of KLK5 and only to a lesser extent by KLK5-mediated activation of KLK7. The significant contribution of KLK5 to NS pathology is in line with a recent study of Furio et al. showing amelioration of skin barrier-phenotype in Spink5-deficent newborns upon KLK5 inactivation [24]. Nevertheless, the remaining activity of KLK7 still contributes to the defective barrier and further intensifies with age as Klk5 -/-Sp5 A135X/A135X show severe epidermal defects manifested by loss of infundibular epidermis at P5. Klk5 -/-Klk7 -/-Sp5 A135X/A135X mice exhibit no skin-barrier defects at P5 and most importantly, in contrast to Klk5 -/-Sp5 A135X/A135X mice, the triple mutants survive to adulthood. The skin defects in Klk5 -/-Sp5 A135X/A135X P5 pups markedly resemble those observed in mice deficient for corneodesmosomal proteins CDSN and DSC1 [30,31]. Indeed, Klk5 -/-Sp5 A135X/A135X show reduced CDSN expression at P5, which indicates that unregulated activity of KLK7 results in degradation of corneodesmosomes. The mechanism of pro-KLK7 activation in the absence of KLK5 remains unclear. KLK7 can be activated by matriptase [32] and a recent study also suggests a role of mesotrypsin in pro-KLK7 activation [33].
Although the lethal phenotype is fully rescued in Klk5 -/-Klk7 -/-Sp5 A135X/A135X mutants and the mice do not show any signs of skin barrier-defects leading to dehydration, we observed minor barrier disruptions in the nostril area of newborn pups, which suggests the activity of another protease physiologically inhibited by LEKTI. As the toluidine blue -stained area overlaps with the expression of KLK14 in late embryonic development (S9A Fig), we propose that KLK14 could be responsible for the remaining pathology of LEKTI-deficient mice even in the absence of KLK5 and KLK7. Moreover, we and others also observed expression of KLK14 in hair follicles (S9B Fig)[34], which makes KLK14 a candidate protease responsible for the development of the bamboo hair defect in Klk5 -/-Klk7 -/-Sp5 A135X/A135X animals up to the age of 3 weeks. As reported, the defects of cell adhesion proteins in hair follicles result in "lanceolate hair"-a hair shaft phenotype in mice that strongly resembles the bamboo hairs of Klk5 -/-Klk7 -/-Sp5 A135X/A135X mutants and NS patients [35,36]. This further supports a possible role of KLK14 in the formation of bamboo-hair, as KLK14 is linked to the degradation of desmosomal proteins in LEKTI-deficient epidermis [8]. Nevertheless, any targets of LEKTI inhibition present in hair follicles, such as caspase-14 [37] or other, currently unidentified proteases, should be considered as a potential cause of bamboo hairs.
Association of NS with abnormal epidermal differentiation accompanied by acanthosis, parakeratosis, and hyperproliferation of keratinocytes was previously reported in Spink5-deficient mouse models [15,26,27] and our Sp5 A135X/A135X confirms the previous findings. We found clear overexpression of keratin6 in Sp5 A135X/A135X E18.5 dpc epidermis, suggesting that events leading to hyperproliferation of keratinocytes are triggered prior to the exposure to the external environment and are a result of unregulated proteolytic activity in the epidermis. In light of the fact that single inactivation of either KLK5 or KLK7 completely rescues the differentiation defects in LEKTI-deficient embryos as well as in newborn mice, we believe that the signalling events resulting in keratinocyte hyperproliferation in neonates depend on the presence of both, KLK5 and KLK7 together. Moreover, we observed that aggravated cutaneous inflammation, which is found in E18.5 Sp5 A135X/A135X embryos, fully depends on simultaneous activity of both, KLK5 and KLK7. KLK5 was previously shown to initiate inflammation in LEKTI-deficient epidermis by activation of PAR2, which results in the induction of pro-Th2 and pro-inflammatory cytokines [22][23][24]. In this study we show that KLK7 is also required for the induction of inflammation in LEKTI-deficient mice as P5 Klk5 -/-Sp5 A135X/A135X pups developed severe acanthosis together with significantly increased expression of TNFα, TSLP, Il-33, Il-1β and ICAM1 while Klk5 -/-Klk7 -/-Sp5 A135X/A135X showed no major defects in the epidermis and increased levels of pro-inflammatory cytokines. Altogether, this suggests that inflammation and differentiation changes in older LEKTI-deficient pups (P5) are initiated by KLK7 activity which is independent of KLK5. Indeed, KLK7 was previously shown to induce inflammation and keratinocyte proliferation in the epidermis [38,39] and a recent study identified KLK7 as a proliferative factor in a mouse model of colon cancer and in human cells in vitro [40]. The mechanism by which KLK7 induces inflammation and differentiation changes remains to be elucidated. In contrast to KLK5, KLK7 cannot directly activate PAR2 as shown in vitro [41] and thus, the inflammation is likely to be triggered by a different mechanism. One possible pathway is the KLK7-mediated conversion of pro-IL1β to active IL1β [42], which could affect the inflammatory phenotype of NS-epidermis.
In summary, we show that the individual inactivation of KLK5 or KLK7 only partially rescues the defective skin barrier but not the lethal phenotype of Sp5 A135X/A135X . Only the concurrent ablation of both KLK5 and KLK7 can fully rescue the lethal phenotype of Sp5 A135X/A135X mice, therefore both proteases should be investigated as clinical targets. We show that KLK7 plays an important role in the inflammation and defective differentiation in NS and its activity is not dependent on activation by KLK5. We also show that the pathological effects of unregulated KLK activities are remarkably age dependent. Altogether, this study expounds the complexity of the proteolytic network and its regulation, which are especially important to understand Netherton syndrome and its treatment.

Generation of targeted mice
All animal studies were ethically reviewed and performed in accordance with European directive 2010/63/EU and were approved by the Czech Central Commission for Animal Welfare.

Generation of Klk5 -/mice
Knock-out first allele of Klk5 was produced by introduction of targeting construct (vector PRPGS00082_A_A10 obtained from NIH Knock-out Mouse Program, KOMP) via homologous recombination in embryonic stem cells (ESC). Positively targeted ESC were injected into developing wt embryos, to produce chimeric mice, which were used to establish Klk5 -/line.

TEWL measurement
Newborn pups from at least two independent litters were separated from mothers to prevent fluid intake. The rate of water loss was analyzed by measuring the reduction of initial body weight at 1h, 2h, 3h and 4 h.

Barrier penetration assay
Newborn mice were euthanized and then dehydrated by incubation for 5 min in 25, 50, 75, and 100% methanol. After rehydration in PBS, mice were incubated for 4 hours in 0.1% toluidine blue (Sigma-Aldrich), washed in PBS and imaged.

Histology
Newborn pups or skin tissues were fixed in 3.6% formaldehyde for 24 h and embedded in paraffin. 5-μm sections were prepared using microtome were stained by hematoxylin/eosin (H&E) or by 0.5% toluidine blue using standard protocols. Images were obtained using Zeiss Axioscan Z1 (Carl Zeiss AG).

Immunohistochemistry
Dorsal skin of P5 pups was fixed in 3.6% formaldehyde for 24 hours and embedded in paraffin. 5-μm paraffin sections were used for antigen retrieval with Discovery Ultra automated IHC/ ISH system (Ventana) and stained with antibodies against Dsg-1 (Santa Cruz, 1:100 dilution, retrieval at pH6) and CDSN (Abcam, 1:100 dilution, retrieval at pH6). After 1 hour incubation at room temperature, anti-rabbit peroxidase conjugated polymer (Zytomed GmBH) was applied for 30 min and the reaction was developed using DAB (DAKO) as a chromogen. Images were obtained using Zeiss Axioscan Z1 (Carl Zeiss AG).

Gene expression analyses
Dorsal skin was obtained from newborn pups, crushed in liquid nitrogen and total RNA was isolated using TRIzol (Thermo-Scientific) according to the manufacturer's instructions. Residuals of genomic DNA were removed using 1 U of DNAse I (Roche) per 1 μg of RNA by 15 min incubation at 37˚C. 1 μg of total RNA was used for reverse transcription by M-MLV Reverse Transcriptase (Promega) using oligo (dT) primers. RT-PCR was performed in a 20-μl reaction mixture containing SYBR Green JumpStart™ Taq ReadyMix with MgCl 2 (Sigma-Aldrich) and 0.25mM of each primer. Respective gene expression was normalized to the expression of TATA-binding protein (TBP). Normalized expression levels were then re-expressed relative to the mean expression level of the respective target in the wt mice. Primer sequences are detailed in S1 Table. Scanning electron microscopy The samples on cellulose filter paper strips were fixed with 3% glutaraldehyde in cacodylate buffer overnight at 4˚C. After fixation, extensively washed samples were dehydrated through ascending alcohol concentrations followed by absolute acetone and critical point drying from liquid CO 2 in a K 850 unit (Quorum Technologies Ltd). The dried samples were sputtercoated with 20 nm of gold in a Polaron Sputter-Coater (E5100) (Quorum Technologies Ltd). The final samples were examined in a FEI Nova NanoSem 450 scanning electron microscope (FEI) at 5 kV using secondary electron detector.