Kallikrein-5 Promotes Cleavage of Desmoglein-1 and Loss of Cell-Cell Cohesion in Oral Squamous Cell Carcinoma*

Oral squamous cell carcinoma (OSCC) ranks among the top 8 causes of cancer death worldwide, with only a 60% 5-year survival rate, highlighting the need for discovery of novel biomarkers and therapeutic targets. We have previously reported that expression of a panel of serine proteinase kallikreins (KLK 5, 7, 8, and 10) is correlated with formation of more aggressive OSCC tumors in a murine orthotopic OSCC model and is elevated in human OSCC. Current studies focus on understanding the potential role of KLK5 in OSCC progression. In initial studies, KLK levels in malignant OSCC cells (SCC25) were compared with cells from normal oral mucosa (OKF/6) and pre-malignant oral keratinocytes (pp126) using qPCR. A marked elevation of all KLKs was observed in aggressive SCC25 cells relative to OKF/6 cells. In normal skin, KLKs are involved in desquamation during epidermal differentiation via proteolytic cleavage of the desmosomal cadherin component desmoglein 1 (Dsg1). As loss of cell-cell cohesion is prevalent in tumor metastasis, Dsg1 integrity was evaluated. Results show that SCC25 cells exhibit cleavage of Dsg1, which is blocked by proteinase inhibitor treatment as well as by siRNA silencing of KLK5 expression. Furthermore, cell-cell aggregation assays demonstrate that silencing of KLK5 enforces cell-cell adhesion; conversely, overexpression of KLK5 in normal oral mucosal cells (OKF/6) enhances cell dispersal. These data suggest that KLK5 may promote metastatic dissemination of OSCC by promoting loss of junctional integrity through cleavage of desmoglein 1.

Oral cavity cancer results in over 200,000 deaths annually and is one of the top eight most frequently diagnosed cancers worldwide (1). The most common malignancy of the oral cavity is oral squamous cell carcinoma (OSCC), 3 including tumors affecting the tongue, floor of the mouth, buccal mucosa, lips, palate, and gingiva, causing more deaths than any other oral disease (2)(3). According to National Cancer Institute statistics, about 23,110 new cases of oral cancer were diagnosed in 2009 in the United States (4). Unfortunately, despite advancements in surgery, chemotherapy, and radiation, only 60% of these individuals will survive for five years. The high mortality associated with OSCC is due primarily to the detection of late-stage disease after the primary tumor has metastasized. Treatment of advanced OSCC is also associated with high morbidity and mortality, resulting from local, regional, and distant metastasis (2)(3). The cellular and biochemical factors that underlie locoregional and distant spread of the disease are poorly understood. Invasion and metastasis of OSCC require multiple cellular events including disruption of cell-cell adhesive contacts, cytoskeletal alterations, and basement membrane attachment, matrix protein proteolysis, and migration (5)(6). Thus a more detailed analysis of the molecular events that contribute to OSCC metastasis is a necessary prerequisite for the development of novel early detection and treatment strategies that have a favorable impact on the survival of OSCC patients.
Human tissue kallikreins or kallikrein-related peptidases (KLKs) constitute a single family of 15 highly conserved trypsin-or chymotrypsin-like serine proteases encoded by a large uninterrupted cluster of protease-encoding genes (KLK1-15) co-localized on chromosome 19q13.4 (7). In initial studies, we identified expression of a panel of kallikreins (KLK-5, -7, -8, and -10) that have a marked elevation in malignant OSCC relative to normal oral mucosa and normal tongue (8). The physiologic roles and natural substrates for most KLKs have not been defined; however expression of multiple KLKs in a single tissue compartment suggests participation in proteolytic cascades (9 -12). As KLK5 can auto-activate and contributes to activation of pro-KLK2, -3, -6, -7, -11, -12, and -14 zymogens, KLK5 may be considered an initiator of putative KLK proteolytic cascades (13). In normal skin, KLK5 is involved in desquamation during epidermal differentiation via proteolytic cleavage of the desmosomal cadherin component desmoglein 1 (14 -15).
Desmosomes are intercellular junctions that provide strong adhesion between cells. They are particularly abundant in tissues such as epithelia and cardiac muscle that are continually assailed by mechanical forces. Extracellularly, desmosomal adhesion is mediated by the desmosomal cadherins comprised of desmogleins (Dsg) and desmocollins (Dsc) (16 -19). Previous studies have shown that loss of desmosomal adhesion can significantly disrupt tissue integrity (20).
In the current study, the effect of modulating KLK5 levels on the expression and integrity of desmoglein 1 (Dsg1) was evaluated in normal oral keratinocytes and OSCC cells. Our data demonstrate that malignant OSCC cells exhibit cleavage of Dsg1 which can be blocked by treatment with proteinase inhibitors as well as by silencing of KLK5. Modification of KLK5 expression also alters cell-cell aggregation and cohesion. These results suggest KLKs may contribute to metastatic dissemination of OSCC via a mechanism involving KLK5catalyzed Dsg1 cleavage.

EXPERIMENTAL PROCEDURES
Antibodies-Rabbit polyclonal antibodies to kallikrein 5, generated against a synthetic peptide based on the kallikrein loop area of kallikrein 5, were obtained from Abcam (Cambridge, MA). This antibody reacts with KLK5, but has no reactivity against KLKs 1-7 and 9 -15. Mouse monoclonal anti-desmoglein 1 antibodies were purchased from BD Transduction Laboratories (San Jose, CA), and do not cross-react with Dsg2 or Dsg3 according to manufacturer's specifications. Mouse monoclonal anti-E-cadherin antibody was obtained from Invitrogen (Carlsbad, CA). Anti-GAPDH was purchased from Santa Cruz Biotechnology.
Immunohistochemical Analysis-Immunohistochemical analysis was performed by analyzing tongue-derived specimens of microarrayed human OSCC and normal oral tissues (US Biomax). Paraformaldehyde fixed, paraffin-embedded sections (4 m) were de-paraffinized with xylene and rehydrated in a series of ethanol washes. Endogenous peroxidase activity was quenched with 3.3% hydrogen peroxide in methanol for 30 min. Antigen retrieval was enhanced by microwaving in 10 mM sodium citrate pH 6.0. Nonspecific binding was blocked with 3% normal horse serum in PBS for 30 min. Sections were incubated for 1 h at room temperature with primary antibody (1:15-1:25 dilution) in 1% BSA in PBS. Staining was detected using an avidin-biotin horse radish peroxidase system (Vectastain Universal Elite ABC kit, Cat. PK-6200, Vector Laboratories, Burlingame, CA), with positive cells staining brown using diaminobenzidine chromogen and hydrogen peroxide substrate (Liquid DAB substrate pack HK153-5K, BioGenex, San Ramon, CA). Analysis of tissue staining was done by light microscopy and scoring of KLK5 was assigned according to the average overall intensity of staining and was graded as follows: 0, no staining; 1, fine granular staining; 2, somewhat coarse staining; 3, very coarse staining.
In some experiments, cells were cultured for up to 7 days in medium containing the proteinase inhibitors leupeptin (Sigma, 100 M final concentration) and chymostatin (Sigma, 100 M final concentration) (24) to block KLK activity.
To generate SCC25 cells with reduced KLK5 expression (KLK5 knockdown, designated SCC25-KLK5-KD), pGFP-V-RS plasmid containing a gene-specific shRNA sequence silencing KLK5 (GTG TTG GTG CAT CCA CAG TGG CTG CTC AC) (OriGene Technologies, Rockville, MD) was used to transfect SCC25 cells using the Human Keratinocyte Nucleofector kit (Amaxa GmbH, Germany). The resulting clonal cell lines were routinely maintained in the medium described above supplemented with puromycin and were assessed for expression of KLK5 by qPCR and immunofluorescence microscopy (see below). To generate cells that overexpress KLK5 in the OKF/6 normal oral epithelial cell background (designated OKF/6-KLK5ϩ), pCEFL-KLK5 plasmid (gift from Dr. Thomas Bugge, National Institutes of Dental and Craniofacial Research, Bethesda, MD) was used to transfect OKF cells using the Human Keratinocyte Nucleofector kit (Amaxa GmbH, Germany). The resulting clonal cell lines were routinely maintained in the medium described above supplemented with G418 and were assessed for expression of KLK5 by qPCR and immunofluorescence microscopy (see below).
Quantitative Real Time PCR (qPCR) Analysis-Total RNA was extracted with Trizol Reagent (Invitrogen) and reverse transcription was performed with 10 g of the total RNA from each specimen using SuperScript II reverse transcriptase (Invitrogen) according to the manufacturer's protocol. The cDNA products were then diluted 1:10 and 5 l of each were used for PCR templates. The following primer sequences were used at final concentrations of 167 nM for each: for KLK5 forward, 5Ј-GCA GGT AGA GAC TCC TGC CA-3Ј and reverse 5Ј-CAC AAG GGT AAT CTC CCC AG-3Ј; for KLK7 forward, 5Ј-ACC CTC AGT GCT GGA GAA GA-3Ј and reverse 5Ј-ACT GGG TCA AAG GTG GTG AA-3Ј; for KLK8 forward, 5Ј-AAG TGC ACC GTC TCA GGC-3Ј and reverse 5Ј-TCC TCA CAC TTC TTC TGG-3Ј; for KLK10 forward, 5Ј-CTC TGG CGA AGC TGC TG-3Ј and reverse 5Ј-ATA GGC TTC GGG GTC CA-3Ј; for Dsg1 forward, 5Ј-AGA TGC CTG ACT TGC GAG AT-3Ј and reverse 5Ј-CTA TCA TGC CGG AAG TTG GT-3Ј; for endogenous control gene phosphoglycerate kinase (PGK) forward, 5Ј-GGG CTG CAT CAC CAT CAT AGG-3Ј and reverse 5Ј-GAG AGC ATC CAC CCC AGG AAG-3Ј. DNA oligos were custom synthesized (Integrated DNA Technologies). Real-time PCR was performed with SYBR green Master Mix (Aplied Biosystems, Foster City, CA). PCR cycling conditions were 95°C for 13.5 min followed by 40 cycles of 95°C for 15 s and 60°C for 30 s. Melt curve cycling consisted of 81 30 s cycles beginning at 55°C increasing by 0.5°C to 95°C. All reactions were carried out on an iCycleriQ real-time PCR detection system (Bio-Rad). Each sample was analyzed in triplicate for each PCR measurement. Melting curves were checked to ensure specificity. Relative quantification of mRNA expression was calcu-lated using the standard curve method with the endogenous housekeeping gene PGK level as normalizer and control sample as calibrator.
Western Blot Analysis-Cells were harvested by scraping in lysis buffer (modified RIPA buffer, with an added protease inhibitor mixture tablet) and samples were centrifuged at 6,000 rpm for 10 min at 4°C to produce supernatant. Protein concentration of samples was determined by using a detergent compatible protein assay kit (Bio-Rad). Proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE), and then transferred onto a polyvinylidene fluoride (PVDF) microporous membranes (Millipore). After blocking nonspecific binding to membranes in 3% BSA in TBST for 1 h at room temperature, membranes were incubated with primary antibodies for 3 h at room temperature or overnight at 4°C, and then with HRP-conjugated secondary antibodies. Immunoreactivity was determined by SuperSignal West Dura Extended Duration Substrate kit (Thermo Scientific). Blots were also probed with antibodies against GAPDH to ascertain equal protein loading. Images were acquired with Fuji Film LAS-4000 Luminescent Image Analyzer. Semi-quantitative analysis of bands was performed by densitometry to evaluate the relative optical density of bands on a given gel. The relative optical densities are expressed on an arbitrary scale.
Immunofluorescence Microscopy-Cells were placed on coverslips in 6-well dishes and were cultured to 80% confluence. Adherent cells were washed once with 37°C phosphate-buffered saline (PBS) once and fixed with 4% paraformaldehyde in PBS containing 0.12 M Sucrose for 20 min. For intracellular antibodies, cells were permeabilized in 0.3% Triton X-100 in PBS for 5 min at room temperature. After blocking with 10% bovine serum albumin (BSA) in PBS for 1 h, the cells were labeled with primary antibodies for 1 h, followed by the respective Alexa-Fluor-labeled secondary antibodies for 30 min. Coverslips were removed from 6-well plates, air dried, and mounted to slide with Gelvatol ϩ DABCO (sigma d2522) with the cell side face down. Specimens were examined and images were acquired with an Olympus IX81 Inverted Spinning Disk Confocal Microscope system using the widefield fluorescence function.
Transmission Electron Microscopy (TEM)-Samples were fixed using 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer, washed with 0.1 M cacodylate buffer; then with 1% OsO 4 , 0.8% potassium ferricyanide in 0.1 M cacodylate buffer for 30 min at room temperature for secondary fixation. After rinsing with ultra pure water, 2% aqueous uranyl acetate was used for 30 min for tertiary fixation. After washing, samples were dehydrated in ascending concentrations of EtOH followed by dehydrating with 100% acetone, then infiltrated with Epon/Spurr's Resin. The samples were embedded and polymerized overnight. Coverslips were removed using liquid nitrogen and samples were returned to the oven to complete polymerization (overnight). Ultrathin sections were cut with Reichert Ultracut R ultramicrotome and collected on celloidine-coated 100-mesh nickel grids. Thin sections were examined with a JEOL 1400 transmission electron microscope (Jeol, Tokyo, Japan).
Cell Aggregation Assay-Semi-quantitative analysis of cellcell aggregation was performed by monitoring aggregation of single-cell suspensions as previously described (25). Cells were grown in 100-mm dishes and removed from the substrate with 0.5% EDTA in PBS. The cells were then counted, rinsed in PBSϩ containing 0.5% BSA, and incubated in singlecell suspension (5 ϫ 10 5 cells/ml) in culture medium containing 15 mM HEPES and 0.5% BSA. Cell suspensions were then placed in 1.5 ml Eppendorf tubes and rotated at 37°C. At various time points up to 7 h, samples (50 l) were removed and phase-contrast photomicrographs of living cell suspensions obtained. The number of single cells was enumerated and compared with the number obtained at the initial time point. were processed for immunostaining using antibodies directed against Dsg1 (1:25 dilution) followed by a biotinylated secondary antibody. Although the sample size was not sufficiently large for robust statistical power, staining trends indicate strong junctionally localized staining in normal tongue (scores: 1 ϭ 12%, 2 ϭ 24%, 3 ϭ 63%), and weaker cytoplasmic staining in OSCC (scores: 1 ϭ 56%, 2 ϭ 35%, 3 ϭ 8%). Arrows denote strong junctional Dsg1 staining. Magnification ϫ100.  Dispase-based Dissociation Assay-Relative dissociation of cell-cell adherent monolayers was evaluated using a dispasebased dissociation assay as previously described (26). Paired cell lines for analysis (SCC25 and SCC25-KLK5-KD cells; or OKF/6 and OKF/6-KLK5ϩ) were cultured in triplicate in 60-mm dishes to 80% confluence, washed twice with PBS and then incubated in 2 ml of dispase for 30 min to release cells as a monolayer. Following careful washing with PBS, cell sheets were transferred to 15 ml conical tubes in a final volume of 5 ml PBS and were subjected to 50 inversion cycles using a rocking platform. The resulting cellular fragments were placed in tissue culture plates and the number of fragments (irrespective of fragment size) was enumerated. Statistical analysis (two-tailed Student's t test) was performed on the

Expression of Kallikreins in Normal and Malignant Oral
Cells-Microarray-based profiling of aggressive OSCC identified a panel of serine proteinase KLKs (KLK-5, -7, -8, and -10) that are highly expressed in poorly differentiated murine OSCC tumors and are abundant in human OSCC (8,27). To evaluate the potential function of KLKs in OSCC progression, KLK gene expression levels in malignant OSCC cells (SCC25) were compared with cells from normal oral mucosa (OKF/6) and pre-malignant oral keratinocytes cells (pp126) using real time qPCR. A significant elevation of all four KLKs was observed in SCC25 cells relative to OKF/6 ( Table 1). With the exception of KLK10, expression of all four KLKs was also higher in SCC25 relative to pre-malignant pp126 ( Table 1).
Expression and Processing of Desmoglein 1 (Dsg1) in Normal and Malignant Oral Cells and Tissues-In stratified epithelia, strong junctional staining for Dsg1 is found in the more differentiated upper layers (Fig. 1A, arrows, 63% strong positive) (28 -29), while both overall staining and junctional localization are reduced in OSCC (Fig. 1B, 8% strong positive) (30 -31). In normal skin, numerous studies have supported a role for KLK-mediated proteolysis of Dsg1 in the cellular desquamation process in the human stratum corneum (32)(33)(34)(35), suggesting that KLKs may play a role in loss of cellular cohesion seen in aggressive tumors. Indeed, analysis of Dsg1 by Western blotting of SCC25 whole cell lysates shows the presence of both full-length (165 kDa) Dsg1 as well as a lower molecular mass Dsg1-immunoreactive band migrating at 130 kDa (Fig.  2, lanes 1, 3, and 5). Cells cultured for 7 days with chymostatin and leupeptin, inhibitors that block KLK5 and KLK7 activities (36 -38), exhibited only the 165 kDa Dsg1 protein (Fig. 2,  lanes 2, 4, and 6), suggesting that the 130 kDa species represents a proteolytic Dsg1 degradation product.
Silencing of KLK5 Modulates Dsg1 Cleavage and Desmosome Formation-To evaluate the effect of altered KLK5 expression on Dsg1 cleavage, KLK5 expression was silenced in SCC25 cells using shRNA (designated SCC25-KLK5-KD). Down-regulation of KLK5 was confirmed using immunofluorescence microscopy and qPCR, showing a 60 -85% knockdown of expression (Fig. 3, A-C). Examination of Dsg1 integrity by Western blotting shows a significant loss of the Dsg1 cleavage fragment in SCC25-KLK5-KD cells (Fig. 3D, lanes 3  and 4; Fig. 3E) relative to parental SCC25 cells (Fig. 3D, lanes  1 and 2). Controls show that the classic cadherin E-cadherin is not differentially processed in response to KLK-5 modification (Fig. 3, D, middle panel and F). GAPDH blots demonstrate equal protein loading (Fig. 3D, lower panel and F).
To evaluate whether cleavage of Dsg1 results in altered ultrastructural integrity, TEM was used to examine cell-cell junctions and the number of desmosomes/field in parental SCC25 versus SCC25-KLK5-KD cell images was enumerated. While electron dense desmosomal cadherin plaques are relatively rare in the cell-cell junctions of SCC25 cells (Fig. 4A). SCC25-KLK5-KD cells form abundant desmosomal contacts (Fig. 4B). Quantitative analysis shows SCC25 cells form 0.65 desmosome clusters per field, whereas SCC25-KLK5-KD cells form 2.06 demosome clusters per field, indicating that reducing KLK5 expression results in a 3-fold increase in desmosome number (Fig. 4B).
Overexpression of KLK5 Induces Dsg1 Cleavage in Normal Oral Mucosal Cells-To evaluate whether induced expression of KLK5 can initiate Dsg1 cleavage, KLK5 expression was increased in OKF/6 normal oral mucosal cells (designated OKF/ 6-KLK5ϩ). Up-regulation of KLK5 was confirmed using immunofluorescence microscopy and qPCR (Fig. 5, A-C). Examination of Dsg1 integrity by Western blotting demonstrates the appearance of the 130 kDa Dsg1 cleavage product in KLK5-transfected OKF/6 cells (Fig. 5, D, lane 3 and E) whereas only intact 165 kDa Dsg1 is observed in parental and vector-transfected cells (Fig. 5, D, lanes 1-2 and E). Controls show that the classic cadherin E-cadherin is not differentially processed in response to KLK-5 modification (Fig. 5, D, middle panel and F). GAPDH blots demonstrate equal protein loading (Fig. 5, D, lower panel and F).
Modulation of KLK5 Expression Alters Cell-Cell Aggregation Kinetics and Monolayer Dissociation-In normal keratinocytes, the formation of functional intercellular adhesion complexes is dependent upon the relative expression and function of desmosomal cadherins (25)(26). As the results above demonstrate a loss of Dsg1 structural integrity and reduced desmosomal cadherin number in cells expressing KLK5, the potential functional consequences of altered desmosomal integrity on cell-cell aggregation dynamics was evaluated. Ro- tating single cell suspensions of SCC25 cells aggregated slowly (Fig. 6, A and E) and were present as small aggregates and single cells at 7 h of incubation (Fig. 6F). In contrast, KLK5 knockdown resulted in significant differences in aggregation kinetics detectable as early as 3 h (Fig. 6, B and E (p Ͻ .05), G), with highly significant differences observed at longer time points (p Ͻ 0.001). At 7 h, large multicellular aggregates were present with no visible single cells. Conversely, normal oral keratinocytes rapidly form robust aggregates (Fig. 6, C, H, I). However, introduction of KLK5 expression significantly de-lays aggregation kinetics as early as 3 h (p Ͻ .05), with highly significant differences observed at longer time points (p Ͻ .001). Similar to results observed in parental SCC25 cells, the majority of the OKF/6-KLK5ϩ population remains as single cells at 7 h (Fig. 6J).
To examine the relative resistance of epithelial cell sheets to dissociation into single cells, a dispase-based dissociation assay was used. Harvested cell monolayers were pulsed with dispase and subjected to controlled mechanical agitation followed by semi-quantitative analysis of sheet fragmentation   (26,39). SCC25 monolayers dissociated into numerous smaller fragments consistent with reduced epithelial cohesion (Fig. 7, A and C). In contrast, SCC25-KLK-KD cells retained better epithelial integrity, as evidenced by a 14-fold reduction in fragment number (Fig. 7, B and C). Conversely, OKF/6 cells remained largely clustered following dispase treatment (Fig. 7,  D and F). However, introduction of KLK5 expression resulted in a relative 6-fold loss of cohesivity of OKF/6-KLK5ϩ aggregates (Fig. 7, E and F).

DISCUSSION
Human tissue KLKs are aberrantly expressed in multiple malignancies including those of the prostate, breast, and ovary (9 -12). While accumulating data support the clinical utility of various KLKs as prognostic biomarkers in these cancers, KLK expression has not been extensively evaluated in OSCC. Using cDNA microarray screening of aggressive OSCC cells, combined with qPCR and immunohistochemical methods, we previously reported that KLK-5, -7, -8, and -10 are abundantly expressed in OSCC (8,27). Further, a recent study correlates up-regulated KLK-4 and -7 expression with poor survival in OSCC patients (40). Although the potential functions of these KLKs are not fully elucidated, emerging data suggest that analysis of KLK expression may provide a useful clinical parameter for molecular classification of oral SCC lesions. Concurrent with up-regulation of KLKs in OSCC, aberrant staining of desmosomal components including desmogleins and desmocollins is also observed in human OSCC relative to normal oral mucosa (30,31,40). Data in the current study support these findings, showing both reduced staining intensity and loss of junctional localization for Dsg1 in a panel of human tongue tumors.
To test the hypothesis that KLK-5 may contribute to OSCC progression via modulation of cell-cell junctional integrity, in the current study the effect of KLK5 expression on Dsg1 processing was evaluated. Our results show that acquisition of KLK5 expression results in cleavage of Dsg-1, while processing of functionally related cell adhesion molecules including desmocollin-1, desmoglein-3 and E-cadherin was not observed. Dsg-1 proteolysis was associated with a corresponding loss of cell-cell adhesive function as assessed by transmission electron microscopy together with functional assays of cellcell aggregation and dispersal. Indeed, down-regulation of KLK5 expression in malignant SCC25 cells significantly enforced cell-cell adhesion, as demonstrated by both cell-cell aggregation and dissociation assays. In contrast, overexpression of KLK5 in normal oral mucosal OKF/6 cells led to reduced cell-cell cohesion.
Cell-cell adhesion is an essential function that contributes to the maintenance of tissue integrity and barrier function (41,42). Within the normal epidermis, desmosomes are present at extensive keratinocyte cell-cell contacts and provide strong intercellular adhesion. Desquamation, the process of shedding of superficial corneocytes, has been shown to involve KLK-mediated proteolytic processing of corneodesmosin, Dsg1, and desmocollin 1 (9,14,15). Direct cleavage of recombinant Dsg1 by KLK5 and identification of KLK5 cleavage sites in the Dsg1 extracellular domain have also been reported (32). KLK5-catalyzed Dsg1 cleavage occurs within the first three extracellular cadherin repeat domains (ECI-ECIII), which mediate heterophilic binding to desmocollins (32), suggesting that KLK5 action may destabilize Dsg1 structure and abolish adhesion to desmocollins on neighboring cells. Moreover, extracellular Dsg1 proteolytic fragments generated by KLK5 may further perturb intercellular adhesion by competing with intact Dsg1 and desmocollin 1 molecules for binding. Desquamation is further regulated by the protease inhibitor lymphoepithelial Kazal-type related inhibitor (LEKTI) encoded by SPINK5 (serine protease inhibitor Kazal-type 5) (43). Loss of LEKTI expression and/or function leads to dysregulated FIGURE 7. Effect of KLK5 expression on monolayer cohesion. Cell-cell adherent monolayers were separated from culture dishes by pulsing with dispase as described under "Experimental Procedures," transferred to conical tubes affixed to a rocking platform, and subjected to 50 inversion cycles. A, B, D, E, aliquots were photographed to visualize relative monolayer cohesion or dissociation. C and F, total number of fragments present following mechanical disruption was quantified.
proteolysis in the outer epidermal layers, loss of cell-cell cohesion, and defective barrier function (15). Interestingly, transcriptomic profiling has shown that expression of the SPINK5 gene is significantly down-regulated in human tongue tumors relative to normal tongue (44). Together with our report of up-regulated KLK5 expression in human tongue tumors (8) and the current results showing KLK5mediated Dsg1 cleavage, these data suggest that KLK5 may potentiate metastatic dissemination of OSCC by contributing to dissolution of Dsg1-containing cell-cell contacts, thereby modifying epithelial cohesion.
Although our data implicate KLK5 activity in Dsg1 processing, the potential involvement of a downstream proteinase(s) that is activated by KLK5 cannot be ruled out using the current approach. Pro-KLK5 has been shown to undergo autoactivation and may process downstream zymogens including pro-KLK7 or other pro-KLKs (13,45) that could functionally contribute to Dsg1 cleavage. Furthermore, it was recently demonstrated that the transmembrane serine proteinase matriptase, the zymogen of which undergoes highly efficient autolysis, functions to activate pro-KLKs in vitro and in vivo (46). As these data suggest the potential presence of a regulatory node further upstream in the zymogen activation cascade, future analysis of matriptase expression in OSCC is warranted.