Kallikrein 5 Inhibition by the Lympho-Epithelial Kazal-Type Related Inhibitor Hinders Matriptase-Dependent Carcinogenesis

Simple Summary Head and neck squamous cell carcinomas (HNSCC) are among the most common cancers worldwide. In contrast to the advances in prevention and treatment of other types of cancer, the five-year survival rate for HNSCC is only about 50% and it has not changed for the past 50 years. This poor prognosis is mainly due to a shortage of suitable markers for early detection, delayed diagnosis and/or referral, and ineffectiveness of chemotherapy. The aim of this study was to explore the inhibitory role of LEKTI in matriptase-dependent squamous cell carcinogenesis and to investigate additional players operating in this pathway. We found that Kallikrein-5 is necessary for PAR-2-mediated IL-8 release, YAP1-TAZ/TEAD activation, and matriptase-mediated oral squamous cell carcinoma migration. This knowledge can contribute for the development of future targeted therapy in HNSCC. Abstract Head and neck squamous cell carcinoma remains challenging to treat with no improvement in survival rates over the past 50 years. Thus, there is an urgent need to discover more reliable therapeutic targets and biomarkers for HNSCC. Matriptase, a type-II transmembrane serine protease, induces malignant transformation in epithelial stem cells through proteolytic activation of pro-HGF and PAR-2, triggering PI3K-AKT-mTOR and NFKB signaling. The serine protease inhibitor lympho-epithelial Kazal-type-related inhibitor (LEKTI) inhibits the matriptase-driven proteolytic pathway, directly blocking kallikreins in epithelial differentiation. Hence, we hypothesized LEKTI could inhibit matriptase-dependent squamous cell carcinogenesis, thus implicating kallikreins in this process. Double-transgenic mice with simultaneous expression of matriptase and LEKTI under the keratin-5 promoter showed a prominent rescue of K5-Matriptase+/0 premalignant phenotype. Notably, in DMBA-induced SCC, heterotopic co-expression of LEKTI and matriptase delayed matriptase-driven tumor incidence and progression. Co-expression of LEKTI reverted altered Kallikrein-5 expression observed in the skin of K5-Matriptase+/0 mice, indicating that matriptase-dependent proteolytic pathway inhibition by LEKTI occurs through kallikreins. Moreover, we showed that Kallikrein-5 is necessary for PAR-2-mediated IL-8 release, YAP1-TAZ/TEAD activation, and matriptase-mediated oral squamous cell carcinoma migration. Collectively, our data identify a third signaling pathway for matriptase-dependent carcinogenesis in vivo. These findings are critical for the identification of more reliable biomarkers and effective therapeutic targets in Head and Neck cancer.


Introduction
Head and neck squamous cell carcinomas (HNSCC) are among the most common cancers worldwide, with over 800,000 new cases diagnosed yearly [1]. In contrast to the advances in prevention and treatment of other types of cancer, the five-year survival rate for HNSCC is only about 50% and is unchanged for 50 years [2,3]. Generally, this poor prognosis is due to a shortage of suitable markers for early detection, delayed diagnosis and/or referral, and ineffectiveness of chemotherapy [3][4][5]. Accordingly, there is an urgent need to identify more reliable therapeutic targets and biomarkers of prognosis in HNSCC.
Several studies demonstrated that dysregulated expression of Type II Transmembrane Serine Proteases (TTSPIIs) is associated with different types of cancers [6]. Matriptase, a TTSPII, is indeed dysregulated in several cancers [7]. List et al. (2005) showed that Matriptase induces malignant transformation when expressed in epithelial stem cells [8]. It has been shown that PI3K-Akt-mTor signaling, elicited by proteolytic activation of pro-hepatocyte growth factor (pro-HGF), is a molecular pathway by which matriptase promotes malignant transformation [9]. Another essential component of matriptase-mediated oncogenesis is the upregulation of NFkB-induced inflammatory cytokines dependent on Protease Activated Receptor 2 (PAR-2) proteolytic cleavage by matriptase [10].
Matriptase was also implicated in Netherton Syndrome (NS), an epidermal disorder caused by mutations in the SPINK5 gene, which encodes for the Lympho-Epithelial Kazal-Type-related Inhibitor (LEKTI). LEKTI inhibits matriptase-dependent skin desquamation in NS by the direct inhibition of Kallikreins 5 (KLK5) and 7 (KLK7), which are known for their role in the degradation of corneodesmosomes at the outermost layers of the stratum corneum [11]. Kallikrein-related peptidases (KLKs) comprise a large family of secreted serine proteases expressed in several tissues [12,13]. Kallikreins are secreted as inactive zymogens and generally depend on proteolytic cleavage at the carboxy-terminal end of either arginine or lysine for activation [14] and are often dysregulated in inflammatory skin disorders and many human cancers [12,15]. Abnormal functional levels of the KLKs in diseased states are a result of unbalanced activation and inhibition events. These peptidases function cooperatively in signaling cascades or complex regulatory networks, often bridging multiple protease families and classes [16] Accordingly, matriptase is able to activate pro-KLK5 and pro-KLK7 [11].
In this study, we found a decrease in LEKTI expression in poorly differentiated OSCCs. This led us to the hypothesis that LEKTI could be a tumor suppressor by regulating KLK activation by matriptase. Indeed, double-transgenic mice with simultaneous expression of matriptase and LEKTI under the keratin-5 promoter led to the inhibition of matriptasedependent carcinogenesis through downregulation of KLK5 expression. Significantly, KLK5 promoted the release of IL-8 and activated the Yap1-TAZ/TEAD transcription network through PAR-2 activation. In conclusion, our study identifies a novel proteolytic pathway that contributes to the matriptase-driven malignant transformation.

Human Tissue Samples
In this study, 127 cases of human Oral Squamous Cell Carcinomas (OSCCs) diagnosed between 2005 and 2016 were selected from Pathology Service (SERPAT) of Ribeirao Preto Medical School, University of São Paulo, to compose a tissue microarray (TMA). Formalin-fixed and paraffin-embedded TMAs were composed of well-differentiated OSCCs, W.D.C., n = 37; moderately-differentiated OSCCs, M.D.C., n = 59; and poorly-differentiated OSCCs, P.D.C., n = 31. Only cases whose diagnoses were made in the following regions of the oral cavity were considered: jaw, floor of mouth, retromolar, mouth, oral cavity, tongue, cheek mucosa, gingiva, gum, larynx/pharynx, hard palate, and lips.

Tissue Microarray Construction
An experienced pathologist (ARS) reviewed the hematoxylin and eosin slides to delineate the most significant area of each tumor to prepare the TMA. From the preselected donor paraffin blocks, 2 mm diameter cylinders were removed and sorted into a paraffin block recipient using the TMA Builder Kit (Histopathology Ltd., Pécs, Hungary).

Ethics Statement
The study was performed in accordance with the Declaration of Helsinki [17]

SPINK5 mRNA Expression Analysis in K5-LEKTI +/0 Mouse Epidermis
Newborn WT and K5-LEKTI +/0 FVB/NJ mice were euthanized by decapitation. Dissected skin was incubated in PBS containing 10 mM EDTA for 5 min at 56 • C to separate the dermis from the epidermis. The epidermis was immediately covered with RNALater (MilliporeSigma, Burlington, MA, USA), and incubated overnight at 4 • C. For the extraction of total RNA, we used TRIzol (Thermo Fisher Scientific, Waltham, MA, USA). Epidermis was added to 1 mL of TRIzol buffer and 100 mg of Precellys ® zirconium oxide beads (Bertin Technologies, Montigny-le-Bretonneux, France). The samples were homogenized using a Precellys-24 tissue homogenizer (Bertin Technologies) for three pulses at 6800 rpm for lysis and homogenization. The samples were then transferred to a new RNase/DNase-free tube for chloroform extraction. The upper phase was transferred to a new tube for isopropanol precipitation. Samples were vortexed and centrifuged at 12,000× g for 30 min. The pellet was then washed with 70% ethanol in DEPC water, and the material was dried. RNA was resuspended in DEPC water. RNA samples were quantified, and cDNA synthesis was performed from 1 µg total RNA using the Kit ™cDNA Synthesis SuperScript ® VILO (Catalog# 11754, Thermo Fisher Scientific). The qRT-PCR reaction was performed in the 7500 Real-Time PCR System according to the protocol of TaqMan Master Mix (Applied Biosystems, Thermo Fisher Scientific) manufacturer. Gapdh and Hprt1 were used as internal controls. LEKTI gene expression was evaluated using mouse SPINK5 TaqMan probes (Mm00486343_cn). All reactions were repeated three times, and the experiments were validated with the use of negative controls.

Histological and Immunohistochemical Analysis
Adult mice were euthanized by CO 2 inhalation, and newborn mice were euthanized by decapitation. Mouse tissues were fixed overnight in 4% paraformaldehyde and embedded in paraffin. Both mice and TMA paraffin blocks were cut into sections 6 µm thick and mounted on glass slides. Tissue sections were processed for histology and stained either with hematoxylin and eosin (H&E), toluidine blue, or immunostained for LEKTI, Matriptase, or KLK5. For IHC staining, antigen retrieval was performed by boiling samples in 0.01 M sodium citrate buffer, pH 6, for 12 min. After non-specific antigen blocking was performed, the sections incubated overnight at 4 • C with the primary antibodies followed by incubation with appropriate biotin-conjugated secondary antibodies and the Vectastain-ABC Elite kit (Vector Laboratories, Burlingame, CA, USA). Staining was developed by incubation with 3,3 -diaminobenzidine (25 mg/mL, Merck) and H 2 O 2 and the sections were counterstained with hematoxylin. Non-immune anti-rabbit IgG was used as negative control (3 µg/mL, Catalog# 011-000-003, Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) for LEKTI, Matriptase, and KLK5. Images were obtained on an Olympus VS120 slide scanner (Olympus Corporation, Tokyo, Japan). Quantification of the total epithelial area, the stained epithelial area of each sample and counts of metachromatically stained mast cells were performed using the Image J software [22]. Briefly, for TMA slides stained for LEKTI or Matriptase, ratios between the total epithelial and immunolabeled epithelial area (µm 2 ) were calculated and plotted as a percentage (%); zero values were included where samples lacked LEKTI staining. For mice skin samples, the height of the epidermis (epithelial thickness) and mast cell count per 10 3 µm 2 were quantified.

Mouse Skin Dissociation
Mouse dorsal skin was extracted from euthanized healthy WT, K5-LEKTI +/0 , K5-Matriptase +/0 , and K5-LEKTI +/0 /K5-Matriptase +/0 FVB/NJ mice. The dorsal skin was shaved and a depilatory cream was applied to remove remaining fur. The depilated skin was cleaned with PBS before extraction. An area of 15 mm × 35 mm of back skin was excised and cut into three pieces before being put in a 50 mL tube with 20 mL of HBSS and quickly washed twice by vortexing. The tissue was drained and placed into a 50 mL conical tube containing 10 mL of pre-dissociation buffer (RT HBSS with phenol red, without calcium or magnesium, 5 mM EDTA, and 10 mM HEPES), incubated for 30 min at 37 • C with agitation, and vortexed for 10 s before being strained through a 70 µm strainer. Tissue pieces were transferred to a Petri dish and finely cut into small fragments (2.2 mm × 2.2 mm) that were transferred to a new tube containing fresh pre-dissociation buffer and incubated again for further 30 min, vortexed vigorously for 10 s, strained and washed with 50 mL HBSS to remove excess EDTA. Collagenase digestion of skin was performed by incubating the strained tissue pieces in a new 2 mL Eppendorf tube containing 1.5 mL fresh digestion buffer (HBSS supplemented with 1 mg/mL collagenase D (Roche, Basel, Switzerland), 1 mg/mL collagenase type IV (Worthington Biochemical Co., Lakewood, NJ, USA), 100 µg/mL DNAse, 1 mM CaCl 2 ) and 100 mg of Precellys ® Zirconium-Oxide Beads of 1.4 mm (Bertin Technologies). The tissue was incubated for 1-2 h at 37 • C with agitation and was briefly vortexed every 15 min. After dissociation, the contents were filtered through a 70 µm strainer into a new 50 mL conical tube, and flow-through containing the digested tissue was washed with 50 mL of DMEM with 10% FBS, centrifuged for 5 min at 350× g, and the cell pellet was resuspended in PBS for flow cytometry analysis.

Flow Cytometry
Skin isolated total cells were counted using an Automated Cell Counter (Countess I, Invitrogen, Thermo Fisher Scientific). The cell suspension (2.1-13 × 10 5 cells/sample) was incubated for 15 min at RT with fixable dead cell stain (Thermo Fisher Scientific) and washed with AutoMACS Rinsing Solution containing 0.5% BSA (Mylteni Biotec, Bergisch Gladbach, Germany). The cell suspensions were then incubated for 10 min at 4 • C with anti-CD16/CD32 monoclonal antibody, followed by incubation for 30 min at 4 • C with anti-CD45, anti-Ly6G, anti-CD11c, anti-MHCII, and anti-CD64. The samples were acquired using a BD FACSCanto II cytometer and CellQuest software (BD Biosciences). Total events per sample were collected and analyzed according to size and granularity, single events, live cells, and fluorescence intensity using FlowJo software (BD Biosciences). The following gating strategy was used: leukocytes (CD45 + ) in total live and single cells; Ly6G + (neutrophils) and Ly6G − in CD45 + cells; dendritic cells (DC-CD11c + MHCII High ) and myeloid cells (MY-CD11c + MHCII + ) in total Ly6G − cells, and macrophages (CD64 + ) in total myeloid cells.

Chemical Carcinogenesis
The shaved dorsal area of mice was treated with five applications of 25 µg of 7,12dimethylbenzanthracene (DMBA) diluted in 100 µL of acetone. Applications started at five weeks of age and were performed every three weeks. The tumor incidence and size in the carcinogen-treated mice were monitored every three weeks. Mice with ulcerating tumors or tumors reaching a diameter of >2 cm were euthanized before the termination of the study at 48 weeks of age. Tumors and organs were collected and fixed and processed for histology.

PAR-2 Activation Assay
The PAR-2 activation assay was performed as previously described [10]. Briefly, HEK293T cells were cultured in DMEM and co-transfected with pCDNA3.1-PAR-2 (50 ng), pRL-Renilla luciferase (20 ng), and SRE-Firefly luciferase (50 ng) plasmids. At 16 h post-transfection, cells were treated for 6 h with 100 nM hrKLK5, or 15 nM hrMatriptase, which was used as a positive control. PAR-2 activation was measured using the Dual-Glo Luciferase Assay System according to the manufacturer's instructions (Promega, Madison, WI, USA). Luminescence was measured using a Victor X3 plate reader (Perkin Elmer, Waltham, MA, USA), and SRE activation was determined as the ratio of firefly to Renilla luciferase counts.

Cytokine Release
WT or KLK5 KO Cal 27 cells were plated at 2 × 10 5 cells per well in 24-well plates and cultivated for 16 h in DMEM containing 10% FBS. Cells were then starved for 2 h before stimulation with hrMatripase for 24 h. Supernatants from cell cultures were collected and measured by ELISA using the DuoSet ELISA kits (R&D Systems) according to manufacturer's protocols for human IL-8/CXCL8 (Catalog# PD8000C), TNF-alpha (Catalog# DY210), and IL1-beta (Catalog# DBL50).

Scratch Assay
The Scratch assay was performed as previously described 21 . Briefly, WT or KLK5 KO Cal 27 cells (2 × 10 5 cells) were cultured on glass coverslips, serum-starved for 16 h and treated or not with hrMatriptase. At time "0" a 10 µL plastic tip was used to scratch the cells in the middle of each coverslip and wound closure was evaluated for up to 48 h. The coverslips were imaged at 0, 24, and 48 h after scratching using an DMI4000 Leica epifluorescence microscope. The FIJI image processing package [22] was used to measure the area between the edges of the scratch in all time points. The following formula was used to calculate the wound closure area: Wound Closure Area = Area [T0] − Area [Tn].

Statistical Analysis
GraphPad Prism 8 (GraphPad Software, Inc., San Diego, CA, USA) was used for data analysis and preparation of graphs. Normality tests were performed. Data from experiments with two groups were analyzed using a t-test. Data from experiments with three or more groups were analyzed using one-way analysis of variance test (ANOVA) and post-test as indicated in the figure legends.

LEKTI but Not Matriptase Is Differentially Expressed in Human OSCCs
Matriptase expression and activity are dysregulated in several human cancers [7]. In particular, matriptase was previously found to be ubiquitously expressed in HNSCCs of different anatomical locations and different presumed etiology [9]. Conversely, a recent study from our group showed that LEKTI was downregulated in OSCCs [21]. Because LEKTI has also been shown to inhibit a matriptase-driven proteolytic pathway in terminal epithelial differentiation [11], we hypothesized that it could play a role in matriptase mediated squamous cell carcinogenesis. To investigate this, we performed an immunohistochemical evaluation of TMAs comprised of human OSCCs of different anatomical locations of the oral cavity and of the following histopathological grades: well-differentiated carcinomas (W.D.C., n = 37), moderately differentiated carcinomas (M.D.C., n = 59), and poorly differentiated carcinomas (P.D.C., n = 31). While LEKTI expression is remarkably re-duced in M.D.C.s and P.D.C.s, where the majority of the samples were negative ( Figure 1A, top panel), matriptase expression was found in 70% of the samples in all three groups ( Figure 1A, bottom panel). Quantification of the immunostained area in all samples confirmed decreased expression of LEKTI in less differentiated samples ( Figure 1B), while matriptase expression remained unaltered among the different groups ( Figure 1C). Matriptase expression was found throughout in both W.D.C. and P.D.C. samples ( Figure 1D-G), and LEKTI expression was limited to well-differentiated cells in W.D.C.s ( Figure 1I,J). These results are consistent with our previous work where increased protease/inhibitor ratio was associated with increased pathological grades and poor prognosis [21].

Generation of K5-LEKTI Mice
To further investigate the role of LEKTI in matriptase-dependent squamous cell carcinogenesis, we generated transgenic mice that express LEKTI in basal keratinocytes. Full-length murine LEKTI cDNA was cloned under the control of bovine keratin-5 promoter in the pBK5 vector ( Figure 2A) [18]. Pronuclear injection of the transgene into multiple two-cell embryos generated five founders ( Table 1). The keratin-5-LEKTI transgene was detected in two mice by PCR amplification of genomic tail DNA ( Figure 2B). The transgenic founders were fertile and transmitted the transgene to the next generation. The mouse line FVB-K5-LEKTI-B1 was selected to establish the colony (K5-LEKTI +/0 , Table 1). K5-LEKTI +/0 mice do not display any particular phenotype as evidenced by representative images of 3 and 11-month-old WT and K5-LEKTI +/0 mice ( Figure 2C-F). qPCR analysis of RNA extracted from the epidermis of newborn WT and K5-LEKTI +/0 ( Figure 2G) mice showed an average five-fold increase of LEKTI mRNA expression in the transgenic mice.

Co-Expression of LEKTI in Basal Keratinocytes Attenuates Matriptase-Dependent Premalignant Phenotype
Matriptase induces malignant transformation when expressed in epithelial stem cells, and this process is preceded by a premalignant phenotype characterized by hyperplasia, dysplasia, and dermal inflammation [8]. To investigate whether LEKTI could play an inhibitory role in this context, we induced concomitant expression of both Matriptase and LEKTI in basal keratinocytes of transgenic mice. For that, transgenic mice expressing matriptase cDNA under the control of Keratin 5 promoter (K5-Matriptase +/0 ) were interbred with the K5-LEKTI +/0 mice ( Figure 3A). Double-transgenic mice (K5-LEKTI +/0 /K5-Matriptase +/0 ) showed a prominent rescue of matriptase-dependent premalignant phenotype. The animals were analyzed at 3 and 11 months of age, and co-expression of LEKTI and matriptase partially rescued matriptase-driven alopecia and ichthyosis (3-month-old mice, Figure 3A, and 11-month-old mice, Figure S1A). In addition, epidermal hyperplasia ( Figure 3B and Figure S1B, top panels, yellow dashed line delimitates epidermal thickness) and dermal mast cell recruitment ( Figure 3B and Figure S1B, bottom panels, black arrows) were also diminished, as verified by H&E histological evaluation and toluidine blue metachromatic staining, respectively. Further quantification of epidermal thickness and dermal mast cell recruitment showed that matriptase-dependent premalignant phenotype is indeed ameliorated by concomitant LEKTI expression ( Figure 3C,D and Figure S1C,D). To further investigate the role of LEKTI in the attenuation of matriptase induced dermal inflammation and myeloid cells infiltration into the skin, we performed flow cytometry analysis of cells isolated from digested dorsal skin samples of 3-month-old mice. In accordance with histopathological data, our results demonstrated that K5-Matriptase +/0 mice present a significant increase in the total number of inflammatory cells in the skin ( Figure 3E). The inflammatory infiltration was confirmed by an elevated percentage of leukocytes (CD45 + cells) in K5-Matriptase +/0 mice, which decreased in K5-LEKTI +/0 /K5-Matriptase +/0 mice ( Figure 3F). Concerning myeloid cells, we analyzed neutrophils, dendritic cells, and macrophages. Although there was no difference in neutrophils (Ly6G + ) among groups ( Figure 3G), K5-Matriptase +/0 mice showed a significant reduction in the percentage of dendritic cells (D11c + MHCII High ), not observed in the double-transgenic group ( Figure 3H). Notably, a significant increase in the percentage of macrophages (CD64 + ) was observed in K5-Matriptase +/0 mice and rescued in double-transgenic mice ( Figure 3J). Considering other myeloid cells (CD11c + MHCII + ), LEKTI expression in both K5-LEKTI +/0 and K5-LEKTI +/0 /K5-Matriptase +/0 mice induced a significant increase of this population in comparison to WT and K5-Matriptase +/0 mice ( Figure 3I). Representative flow cytometry differences among groups are displayed in Figure 3K.
Overall, our data confirms that LEKTI can act as a tumor suppressor downstream of matriptase activation, as evidenced by its ability to modulate matriptase induced hyperplasia and inflammatory cell recruitment.

Co-Expression of LEKTI with Matriptase in Basal Keratinocytes Delays the Onset and Progression of Chemically Induced Carcinogenesis and Decreases KLK5 Expression
Ras-dependent SCC is potentiated by the topical application of the genotoxic agent DMBA in K5-Matriptase +/0 mice [8]. To explore the inhibitory role of LEKTI in the initiation and progression of matriptase driven carcinogenesis, we used a one-stage carcinogenesis model where DMBA was applied to the dorsal skin of WT, K5-LEKTI +/0 , K5-Matriptase +/0 , and K5-Matriptase +/0 /K5-LEKTI +/0 mice every three weeks for a total of five applications ( Figure 4A). While WT and K5-LEKTI +/0 mice did not develop any lesions, all K5-Matriptase +/0 mice developed lesions at around 14-17 weeks of age. More importantly, there was an average 3-week delay in the emergence of lesions in K5-Matriptase +/0 /K5-LEKTI +/0 mice compared to K5-Matriptase +/0 mice ( Figure 4B). Tumor progression was also hampered by co-expression of LEKTI, as evidenced by a smaller number of lesions and the reduced average size compared to lesions of K5-Matriptase +/0 mice ( Figure 4C,D). At 26 weeks of age, Matriptase +/0 mice display an increase in tumor size compared to the double-transgenic mice ( Figure 4E). The observed delay in onset and progression of DMBA-induced carcinogenesis indicates that LEKTI could be working as a suppressor of squamous cell carcinogenesis in the context of dysregulated matriptase.  Since LEKTI is unable to inhibit matriptase directly [11], we hypothesized that there is another matriptase protease substrate in squamous cell carcinogenesis. In this respect, LEKTI inhibition of the matriptase-dependent proteolytic pathway could occur through the inhibition of epithelial kallikreins. In fact, immunohistochemical analysis of KLK5 expression in skin biopsies of 3-month-old mice showed that this protease is increased in K5-Matriptase +/0 mice. Notably, concomitant expression of LEKTI and matriptase in basal layer keratinocytes is sufficient to revert the aberrant KLK5 expression found in K5-Matriptase +/0 mice ( Figure 4F).

KLK5 Activates YAP1-TAZ/TEAD Transcription via PAR-2 and Induces Matriptase-Mediated Release of IL-8 and Cell Migration in OSCC Cells
Because PAR-2 was previously shown to be essential for matriptase-driven premalignant progression and squamous cell carcinogenesis, we next investigated whether KLK5 activates PAR-2. For this purpose, we used a reconstituted cell-based assay in which HEK293 cells were transfected with a PAR-2 expression vector and a serum response element (SRE)-luciferase reporter plasmid. The transfected cells were then exposed to hrKLK5 to determine PAR-2 activation. As expected, KLK5 treatment was able to activate PAR-2 ( Figure 5A, matriptase-dependent activation of PAR-2 was used as positive control). Importantly, KLK5-dependent PAR-2 activation was also responsible for an increase in Hippo-YAP1/TEAD transcriptional activity, as measured by luciferase assay with a reporter containing tandem TEAD-binding sites in HEK293T cells transfected with PAR-2 and treated with hrKLK5 ( Figure 5B). Because PAR-2 activation also leads to NFkB activation, we sought to investigate the extent to which KLK5 contributes to the release of proinflammatory cytokines. To this aim, we used WT and KLK5 KO OSCC cell lines [21] treated or not with hrMatriptase for 24 h, and proinflammatory cytokine release was evaluated by ELISA. Matriptase-dependent release of IL-8 occurred only in cells expressing KLK5 ( Figure 5C). TNF-α release was stimulated by matriptase in both WT and KLK5 KO cells ( Figure 5D). Because both NFkB and Hippo signaling pathways may lead to increased cell migration in cancer [23,24], a scratch wound healing assay was used. WT and KLK5 KO OSCC Cal 27 monolayers were treated or not with hrMatriptase, scratched, and wound closure was evaluated for up to 48 h. Interestingly, matriptase-dependent wound closure after 24 h took place solely in cells expressing KLK5. After 48 h, however, hrMatriptase treatment induced wound closure in both WT and KLK5 KO cells ( Figure 5E, third and fourth columns) and this effect was partially inhibited in KLK5 KO cells ( Figure 5E, fourth column) compared to WT cells ( Figure 5E, third column). Wound closure was also delayed in untreated KLK5 KO cells ( Figure 5E, second column) compared to WT cells ( Figure 5E, first column). Quantification of the wound closure confirmed a significant delay in both matriptase-treated (striped columns) and untreated (unstriped columns) KLK5 KO cells (red columns) when compared to WT cells (white columns) after 24 and 48 h ( Figure 5F).

Discussion
This study aimed to explore the inhibitory role for LEKTI in matriptase-dependent SCCs and investigate additional players operating in this pathway. We showed here that LEKTI expression is either absent or remarkably reduced in less differentiated, more aggressive OSCCs, while matriptase expression is prevalent among OSCCs of different grades. Protease to inhibitor unbalance has been associated with uncontrolled proteolysis and malignant disorders before [25,26], and this result is in agreement with recent findings from our group, where LEKTI was shown to be downregulated in OSCCs in comparison with premalignant lesions and normal oral mucosa, with increased KLK5/SPINK5 mRNA (protease/inhibitor) ratio being associated with worse prognosis [21]. Furthermore, SPINK5 expression levels have been found to be downregulated in head and neck squamous cell carcinomas (HNSCCs) [27][28][29].
Consisting with an inhibitory role for LEKTI in the development of SCCs, we show that co-expression of LEKTI with matriptase in basal keratinocytes partially reverts matriptasemediated premalignant phenotype, as evidenced by reduced alopecia, hyperplasia, and myeloid cell recruitment. In line with the fact that the tumorigenic microenvironment of SCCs often displays reduced quantities of myeloid CD11c + dermal DCs, which indicates disruption in the DC immunostimulatory capacity [30,31], we showed a matriptase-dependent significant decrease in DCs. This was accompanied by increased macrophage infiltration, a phenotype rescued by co-expression of LEKTI in the skin of transgenic mice. Conspicuous influx of tumor-associated macrophages in SCC lesions can contribute to carcinogenesis and tumor growth [32]. More importantly, LEKTI expression also delayed the onset and progression of chemically induced carcinogenesis. Comparable inhibition was previously observed by co-expression of the matriptase cognate inhibitors hepatocyte growth factor inhibitors 1 and 2 (HAI-1 and HAI-2) [8,33].
In agreement with the fact that LEKTI inhibits matriptase-dependent skin desquamation in Netherton Syndrome by the direct inhibition of KLK5 [11] and our data shows that concomitant expression of LEKTI with matriptase rescues aberrant KLK5 expression. This indicates a role for KLK5 as a matriptase substrate. Dysregulated KLK5 has been implicated in a wide range of epithelial cancers [21,[34][35][36][37][38]. Our data supports the existence of a matriptase-activated KLK5-dependent PAR-2 signaling axis in SCCs. PAR-2 is activated by several trypsin-like serine proteases, including matriptase and KLK5, signaling to various downstream pathways that modulate cell proliferation, migration and invasion, and cytokine production [10,21,36,39]. Regarding the potential mechanisms by which KLK5 can contribute to tumorigenesis, we show that KLK5 can induce PAR2 dependent activation of NFKB and Hippo-YAP1. Indeed, it has been shown that PAR-2 can lead to NFκB activation in inflammation and SCCs, and PAR-2 can be activated by KLK5 in OSCC [7,33,36,40,41]. In addition, PAR-2 activation by SLIGRL (PAR-2 peptide agonist) can result in YAP1 activation [42]. Hippo-YAP1/TEAD regulates organ size, tissue homeostasis, and tumorigenesis in mammals. Abnormal upregulation of this pathway occurs in many human malignancies and promotes tumor formation, progression, and metastasis [43,44].
Tumor invasion and metastasis are complex biological processes in which detachment and migration rely on matrix-degrading proteases. Knockdown of matriptase in endometrial cancer cells inhibits migration and invasion ability in vitro [45], and our recent work showed that CRISPR-mediated disruption of KLK5 blocks OSCC cell migration [21]. Our current results showed impaired matriptase-mediated migration in KLK5 KO OSCC cells, supporting a potential role of this pathway in cancer invasion and metastasis.
Taken together, our data support a model wherein matriptase activates PAR-2 via KLK5 and modulates SCC development and progression. This knowledge can contribute for the development of targeted therapeutics in HNSCC.

Conclusions
Our data identify a third signaling pathway for matriptase-dependent carcinogenesis in vivo, wherein matriptase activates PAR-2 via KLK5 and thus modulates SCC development and progression.

Informed Consent Statement:
Only incisional biopsies were included in this study and patient consent was waived since the work was done using paraffin-embedded non-identified samples, which according to Brazilian laws, do not require patient consent forms. Data Availability Statement: Data is contained within the article or supplementary material.