Proteolytic Cleavage of Podocin by Matriptase Exacerbates Podocyte Injury

Podocyte injury is a critical step

Chronic kidney disease (CKD) has come to be recognized as a global public health problem (1). An increasing number of CKD patients end up suffering from end stage renal disease, requiring renal hemodialysis, and loading huge financial burden on the shoulder of their home countries. Developing specific treatments for curing CKD are warranted.
The glomerular filtration unit consists of three layers: fenestrated endothelial cells, the glomerular basement membrane, and podocytes (2). Podocytes, a major cell type of renal glomerulus, are terminally differentiated, and interdigitate with adjacent podocytes to form glomerular filtration barrier (3). Interdigitated foot processes of podocytes form a 40-nm-wide cell junction that is composed of several proteins called the slit diaphragm (SD). Podocyte injury is often associated with a loss of SD proteins, and the unique characteristic of several SD proteins with the extracellular domain might be a key to understand podocyte injury. However, the precise mechanism underlying this phenomenon remains unclear.
Proteolysis is a physiological process that is required for the maintenance of renal physiology at multiple levels, including gene transcription, cell trafficking, and extracellular secretion. Aberrant activation of proteolysis is likely involved in the pathogenesis of renal disease. In particular, a number of protease families likely play a key role for the development of renal fibrosis, including, matrix metalloproteinases (MMPs), serine proteases and 2 Cathepsins (4,5). Accumulating evidence demonstrate that intracellular proteolytic processing is required to maintain physiological function in podocytes while aberrant activation causes cellular injury (6). Podocyte injury is often initiated by SD disruption, but its precise mechanism has yet to be clarified. Recently, it was shown that a non-specific serine protease inhibitor, camostat mesilate, slowed the progression of renal disease due to the protection of podocytes (7), suggesting that serine proteases should be involved. Nontheless, the specific type of enzyme involved in this process also remains unknown.

Results
We established a model of Adriamycin (ADR) nephropathy in mice developing albuminuria and podocyte injury with a reduction in mRNA expressions for several markers of podocytes in mouse glomerulus, including Podocin, Synaptopodin, and WT-1 ( Fig. S1 A-D). Podocyte injury was also histologically confirmed by PAS staining and immunohistochemistry. Likewise, Desmin, a marker of podocyte injury, was upregulated along with glomerular injury (data not shown). Interestingly, the injured podocytes still expressed, to some extent, both Nephrin and Podocin in this model (Fig. S1E). A microarray analysis using glomeruli of ADR nephropathy mice identified 15 serine proteases, which were highly upregulated (>2.0 fold) in glomeruli of mice with ADR nephropathy. This is in comparison with control mice at 6 weeks after administration (Fig.1A, Supplemental Table 1). Among these factors, only a type II transmembrane serine protease (TTSP) Matriptase (also known as suppressor of tumorigenicity 14 protein, ST14) is expressed at the epithelial cell junctions in a wide variety of tissues (8).
We examined the expressions of other TTPS and found that Matriptase was induced at mRNA and protein levels in glomeruli of ADR mice at day 42 ( Fig.1B-C, Fig. 2A-C). In particular, Matriptase was predominantly expressed in podocytes (Fig. 2C). Likewise, Matriptase expression was only detected in cultured podocytes, but not in cultured endothelium and mesangial cells (data not shown). Glomerular Matriptase protein was also significantly higher in patients with clinical proteinuria due to diabetic nephropathy or membranous nephropathy, but not in subjects with IgA nephropathy who had no clinical proteinuria (Fig. 2D).
Matriptase is tightly regulated by its cognate inhibitor HAI-1 (also known as SPINT1), a Kunitztype serine protease inhibitor (9). HAI-1 mRNA expression was significantly induced in the glomeruli of ADR mice compared to that of wild type mice (Fig.  1B, D). Subsequently another type of mouse model with podocyte injury (10,11), streptozotocin-induced diabetic endothelial nitric oxide synthase knockout (diabetic eNOS-KO) mice, was examined using a microarray analysis by isolated glomeruli (Supplemental Table 2). Among 13 candidate genes (Fig. S2A), HAI-1 level was significantly upregulated in the podocytes (confirmed by immunohistochemistry) and in serum of diabetic eNOS-KO mice ( Fig. 3A-C, Fig. S2B). A microarray analysis showed that glomerular Matriptase mRNA expression was highly upregulated by 5.3 times in diabetic eNOS-KO mice compared to eNOS-KO mice ( Fig.S2C-D). Analyses of these two independent mouse models suggest that the Matriptase/HAI-1 pathway plays a key role in podocyte injury and could therefore be a potential therapeutic target for kidney disease.
Matriptase and HAI-1 are co-expressed in many epithelial cells (12), and the regulation of Matriptase by HAI-1 is required for epidermal integrity (13). In turn, an imbalance favoring Matriptase over HAI-1 contributes to various diseases (14)(15)(16)(17). According to the Nephroseq database, HAI-1 is induced (1.5-fold), along with two-fold increase in Matriptase expression, in the podocytes of diabetic nephropathy. This indicates that enhancing Matriptase activation over HAI-1 could be a potential mechanism for the progression of podocyte injury. Consistent with these observations, the podocyte-specific depletion of HAI-1 deteriorated podocyte injury in ADR nephropathy ( Fig. 3D and 3E, Fig S3A, S3B and S3C). Podocyte injury could be accounted for by aberrant Matriptase activation due to the absence of HAI-1 in the podocyte.
To develop a therapeutic approach to nephropathy, we subsequently examined whether the selective inhibition of Matriptase could protect kidneys from disease progression. We examined two types of serine protease inhibitors toward Matriptase activity; a selective peptide-mimetic inhibitor of Matriptase (IN-1) (18) and a synthetic non-specific serine protease inhibitor Nafamostat mesilate (NM), which ameliorates rat kidney disease (7). It was found that the two compounds inhibited Matriptase activity in a dose-response manner, and inhibitory effects were nearly equipotent, with half-maximal inhibitory concentration of IN-1 (IC50 IN-1) = 1.3 nM versus IC50 NM = 0.86 nM (Fig. 4A). However, inhibition assays toward thrombin activity showed that IC50 was 2.8 μM for IN-1, whereas it was 99 nM for NM (Fig.  4B), suggesting that IN-1 is more specific to Matriptase than NM. To examine the therapeutic effect of IN-1, this compound was applied to BALB/c mice with ADR nephropathy upon 15 days after ADR treatment. We observed that Matriptase inhibitor IN-1 slowed the progression of ADR nephropathy with blocking podocyte injury and ameliorated albuminuria ( Fig. 4C-F). Likewise, NM-1 also blocked the glomerular injury and suppressed urinary Albumin excretion in mice with ADR nephropathy (Fig. S4). These results indicate that blocking Matriptase could be a potential therapeutic approach against chronic kidney diseases.
Next, we sought to identify the substrates of Matriptase in podocytes. Interestingly, Matriptase was found to directly cleave mouse Podocin (mPodocin) (Fig. S5A, S5I), but not with other slit membrane proteins; Claudin5, β-Catenin or Nephrin (Fig. S5B-D). Fig.5A shows that enzymatic dead G827R Matriptase mutant failed to cleave mPodocin, and multiple bands in anti-His blot indicated that Matriptase was activated by self-cleavage at multiple sites. The mPodocin was cleaved at the membrane but not in the cytoplasm (Fig.  5B). A mutagenesis on mPodocin identified R50 as the specific cleavage site of Matriptase, and not at R36, R45, or R54 (Fig. 5C, Fig. S5H-I). Matriptase cleavage of mPodocin was blocked by co-expression of wild type as well as extracellular domain of HAI-1 (Fig.  5D). We also identified the fragmented cleaved mPodocin in the urine of mice with ADR nephropathy (Fig. S5E), suggesting that Matriptase cleaves Podocin in in vivo.

Discussion
Podocytes have been found to produce other types of proteases, matrix metalloproteases MMP-2 and -9, all of which are believed to have significant roles within the glomerulus as they degrade type IV collagen, a major protein of the glomerular basement membrane (19). It has also been observed that altered MMPs/tissue inhibitors of MMPs (TIMPs) balance lead to increased extracellular matrix deposition or excessive degradation activity -a finding with critical implications for glomerular diseases (2). The upregulation of MMP-9 expression can be found in several glomerular diseases accompanied by proteinuria in human patients (20,21) and diabetic nephropathy murine model (22). Podocytes express CD40 and CD154, the ligand of CD40, to induce MMP-9 expression in an autocrine fashion (23). In various diseases, coagulation proteases also contribute to tissue injuries, including cancer progression and cardiovascular diseases. These injuries are mediated by protease-activated receptors (PARs) (24,25), a family of G protein-coupled receptors consisting of four members (PAR1-PAR4). Consistent with findings from an earlier study which suggested that PARs play a role in kidney disease progression (26), PAR-1 has indeed been found to contribute to the development of podocyte and glomerular injuries (27).
It is important to realize that individual proteases induce divergent signaling pathways that lead to differential functional consequences (28). These studies imply that the strategy to targeting all the proteases via non-specific protease inhibitor may not always be beneficial; the targeting of specific protease(s) by specific inhibitor(s) may be a better approach. Although proteases play key roles in the maintenance of renal filtration barrier and there are indications to suggest that serine protease inhibition is involved in kidney disease progression (7), the precise mechanism(s) underlying the progression of kidney disease, in which serine proteases are involved, as well as the target serine protease(s), remain unknown and to be identified.
In this study, we showed that HAI-1 was essential to the protection of podocytes from nephropathy by blocking of Matriptase-mediated cleavage of Podocin. We also demonstrated that a transmembrane serine protease Matriptase acted as a potential therapeutic target, findings which imply a new class of selective peptide-mimetic inhibitor of Matriptase can be applied as a potential drug for the treatment of chronic kidney diseases. Although careful pre-clinical studies are required before applying the Matriptase inhibitor in clinical practice, our finding that NM, which is commonly used for the treatment of pancreatitis and disseminated intravascular coagulation (29), also protects the kidney from disease progression, might give some reassurance that the inhibition of Matriptase is a safe and promising approach.
In conclusion, we identified Matriptase as a potential cause for podocyte injury. The aberrantly activated enzyme beyond the suppression by HAI-1 potently cleaves Podocin and disturbs podocyte integrity, leading to renal injury. Most surprisingly, Podocin-N migrates to the nucleus and binds nucleoli, indicating its roles in physiological and active cell death process.

Experimental procedures Animal Study
All animal experiments were performed in accordance with either the Animal Experimentation Committee of Kyoto University or Tanabe R&D Service Co., Ltd. (Osaka, Japan), or both. The protocol was approved by the animal care and use committee. Male C57BL/6J-Nos3tm1nc (eNOS-KO) mice at 7 weeks of age were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice with diabetic nephropathy were developed as described previously (10). Diabetes was induced with intraperitoneal injection of 50 mg/kg/day streptozotocin (STZ) for five consecutive days. Diabetes was defined as nonfasting blood glucose > 250 mg/dl using a blood glucose meter. Mice were fed a standard laboratory chow ad libitum. Mean blood pressure was measured using CODA Multi-Channel, Computerized, Non-Invasive Blood Pressure System (Kent Scientific Corporation, Torrington, CT) while blood glucose level was determined by GLUCOCARD MyDIA (Arkray, Edina, MN). Urine was collected overnight using metabolic cages. At 14 weeks of age, all mice were sacrificed. For ADR nephropathy model, ADR at a dose of 10.5 mg/kg body weight was injected via the tail vein of male BALB/c mice at 8 weeks of age. Matriptase inhibitor (IN-1) at a dose of 10 mg/kg body weight was intraperitoneally injected to mice with ADR nephropathy three times a week for 4 weeks. Urine was collected overnight using metabolic cages. At 14 weeks of age, all mice were sacrificed (30). Matriptase inhibitor (IN-1) was synthesized (WuXi AppTec Co., Ltd., Shanghai, China), and the purity of this compound was 97.23% as previously described (18). HAI-1 flox/flox mice were kindly provided from Dr. Hiroaki Kataoka (31). Establishment of Podocin-CreERT2 [Podocin-Cre (+)] mice was reported previously (32). HAI-1 flox/flox: Podocin-Cre (+) mice were produced by crossing HAI-1 flox/flox mice and Podocin-Cre mice. The primers for genotyping Podocin-Cre were TTTGCCTGCATTACCGGTCGATGCAAC and TGCCCCTGTTTCACTATCCAGGTTACGGA. The primers used for genotyping HAI-1 flox were ACCACTGGCTCATTTGGTGTTGGC and TGAAGCCTGGCCACTTCCTGATG.
To induce functional Cre protein, 150 mg/kg Tamoxifen was injected intraperitoneally for 3 consecutive days at the age of 8 weeks. ADR at a dose of 10.5 mg/kg body weight was injected into the tail veins of male mice at 10 weeks of age. At 16 weeks of age, all mice were sacrificed.

Isolation of Glomeruli
Mice were perfused with 8 x 107 Dynabeads (Life technology, Waltham, MA) diluted in 40 mL of Hank's Balanced Salt Solution through the heart under anesthesia. Kidneys were isolated and digested in collagenase solution (1 mg/ml collagenase A and 100 U/ml deoxyribonuclease I) for 30min at 37℃, and then glomeruli containing Dynabeads were gathered by a magnetic particle concentrator. During the procedure, kidney tissues were kept at 4°C except for the collagenase digestion. Finally, glomerular RNA was purified with RNeasy Mini Kit (Qiagen, Chatsworth, CA). For measuring protease activity in glomeruli, glomeruli were lysed with RIPA buffer (Santa Cruz Biotechnology Inc., Dallas, TX).

Microarray Analysis
Glomerular RNA was extracted and purified with RNeasy Mini Kit (Qiagen, Chatsworth, CA). RNA quality was assessed with Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). In accordance with the Agilent Technologies protocol, all samples were processed and hybridized to SurePrint G3 Mouse Gene Expression 8 x 60 K (Agilent Technologies). Fluorescence was detected using Agilent DNA Microarray Scanner. The data was analyzed with GeneSpring GX (Agilent Technologies).

Histological Analysis
Formalin-fixed, paraffin-embedded sections (2 μm) were stained with periodic acid-Schiff reagent (PAS) for the light microscopy. Glomeruli (50-100 per kidney) were examined on coronal sections to evaluate the degrees of glomerulosclerosis. Glomerulosclerosis was defined as obstruction of the capillary lumen caused by mesangial expansion or collapsed capillaries.

Immunohistochemistry and Immunofluorescence
Following deparaffinization, the formalin-fixed, paraffin-embedded section samples were incubated with 3% H2O2 for 20min to inactivate endogenous peroxidase activity. Samples were incubated with citrate buffer (pH 6.0) for retrievals of antigens for Podocin, Nephrin and WT-1. The sections were subsequently incubated with primary antibodies. The following antibodies were used: anti Nephrin antibody (R&D Systems AF3159), anti-HAI-1 antibody (R&D Systems AF1141), anti-WT-1 antibody (Santa Cruz Biotechnology Inc SC192), and anti-Podocin antibody (Sigma Aldrich, St. Louis, MO P0372). For Matriptase antibody, anti-serum was collected from rabbits immunized with peptide corresponding mouse and human Matriptase sequence (single letter code, CAQRNKPGVYTRLP). A validation of the antibody specificity using Basic Local Alignment Search Tool (BLAST) analysis showed that a number of proteins which contained significant identity with CAQRNKPGVYTRLP, including Klk5, Urokinase and Tryptase beta2. However, the microarray analysis showed these proteins were not upregulated in our model, indicating that our antibody was likely specific to Matriptase in this experiment. After reaction with primary antibody, sections were incubated with ImmPRESS Reagent Kit (Vector Labs, Burlingame, CA) for immunohistochemistry or incubated with Alexa Fluor conjugated secondary antibodies (Invitrogen, Carlsbad, CA) for immunofluorescence. Positive areas were measured using MetaMorph (Molecular Devices, Sunnyvale, CA). The percent positive area in the glomerular tuft was determined in 30 glomeruli per section. After permeabilization for 10 min, fixed U2OS was incubated in blocking solution [5% normal goat serum+0.1% Triton X-100 in PBS] and then incubated with primary antibodies diluted in blocking solution overnight at 4 °C. Cells were incubated with Alexa Fluor conjugated secondary antibodies diluted in the blocking solution for 1h at room temperature and mounted on slide glass with VECTASHIELD Mounting Medium with DAPI (Vector Laboratories). The following antibodies were used: anti-FLAG M2 monoclonal Antibody (Sigma Aldrich F1804), anti-Fibrillarin (Cell Signaling Technology). Samples were analyzed using Olympus confocal microscope Fluoview FV-1000 under 100x objective lens.

Cell Culture
HEK293 was used for transfection of plasmids with FuGENE HD Transfection reagent (Promega, Tokyo, Japan). MDCK cells were grown in RPMI-1640 and infected with adenovirus expressing LacZ or FLAG-mPodocin or FLAG-mPodocin  or FLAG-Podocin(51-385). After incubation for 3 days, cells were lysed with RIPA buffer or fixed with 4% paraformaldehyde. For primary podocytes, primary glomerular cells isolated from isolated glomeruli with the Dynabeads methods were seeded in culture in DMEM medium supplemented with 10% fetal bovine serum (33). After podocytes formed the colonies, the cells were harvested with trypsin and used for RNA analysis. Synaptopodin was used as a marker to confirm the characteristics of primary podocytes. RIPA buffer was used for obtaining proteins from whole cells. For separating cytoplasm and membrane proteins, subcellular protein fractionation kit (Thermo Fisher Scientific) was used. U2OS was seeded in cover glass on 6-well plate (Falcon) with DMEM complete medium. Transfection was performed the following day using FuGENE HD Transfection reagent. After 24h, cells were fixed with 4% paraformaldehyde.

Plasmid Constructs
Full-length cDNA of mouse Podocin, mouse Claudin5, and mouse β-Catenin were cloned into pFLAG-CMV-6a (Sigma Aldrich, St.Louis,MO). FLAG-mPodocin or FLAG-mPodocin(2-50) or FLAG-Podocin(51-385) were cloned into pAd-CMV-V5 vector (Invitrogen) using Gateway system. LacZ gene was used as a control included in the kit. The recombinant adenovirus backbones were linearized by PacI digestion and transfected into 293A cells for packaging. Viral titers were estimated using a horseradish peroxidase (HRP) conjugated polyclonal antibody to adenovirus hexon (Thermo Fisher Scientific). cDNA of mutant Podocin (R45A, R50A and R54A) was generated by PCR and cloned into pFLAG-CMV-6a. The full-length cDNA of Matriptase constructed in pcDNA3.1-V5-His (Thermo Fisher Scientific) was provided by Dr. Hiroaki Kataoka. Matriptase mutant (G827R) was generated by PCR and cloned into pcDNA3.1-V5-His. The full-length of cDNA of HAI-1 and the deletion type of transmembrane region cloned into pCIneo were provided by Dr. Kataoka. A human Podocin-N fragment cDNA with 3xFLAG was amplified by PCR using a human kidney cDNA in Human MTC Panel I (TaKaRa) as a template. A 3xFLAG Podocin-N was cloned into pcDNA3.1(+). The adenovirus system was used to express mouse Nephrin-DsRed.

Western Blotting
Cell lysate was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and electrotransferred onto polyvinylidene fluoride membranes. After overnight incubation with primary antibody at 4°C, the membranes were incubated with HRP-conjugated anti-mouse (GE Healthcare, Buckinghamshire, England, NXA931) or anti-rabbit (GE Healthcare, NA934) secondary antibody for 1hr at room temperature, followed by addition of ECL prime (GE Healthcare) to detect bands using Image Quant LAS4000mini (GE Healthcare). The following primary antibodies were used: anti-FLAG antibody (Sigma Aldrich, F1804), anti-His antibody (MBL, Nagoya, Japan, D291-3), anti-Actin antibody (Sigma Aldrich, A2228), anti-NA-K ATPase antibody (Cell Signaling, Danvers, MA, 3010) and anti-HAI-1 antibody (R&D Systems, AF1141). All blots were run under the same experimental conditions. Each image was obtained at a single time point and was not combined into a single image.

Transmission Electron Microscopy
Kidneys were fixed with 4% PFA and 2% glutaraldehyde with 0.1 M phosphate buffer (PB) at 4°C overnight. After post-fixation with 1% Osmium Tetroxide in 0.1 M PB for 2 hr, samples were dehydrated in ethanol and propylene oxide. Then, the samples were penetrated in propylene oxide and Epon after polymerization in pure Epon. Ultrathin sections were cut with an ultramicrotome. The sections were stained in uranyl acetate and lead citrate. The grids were examined with a transmission electron microscope (H-7650; Hitachi).

Analysis of human kidney specimens
All of the human specimens were procured and analyzed after informed consent was obtained and with the approval of the Ethics Committee of Ikeda City Hospital and Kyoto University. Tissue samples were obtained from diagnostic renal biopsies performed at Ikeda City Hospital. We investigated samples from patients who had been diagnosed with IgA nephropathy (N=5), diabetic nephropathy (N=3) and membranous nephropathy (N=3).

Nephroseq database analysis
A kidney transcriptomics data repository Nephroseq (www.nephroseq.org, University of Michigan) was used to analyze various deposited data sets. The data set used in the current study is available under accession number GSE30122 in Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/). The expression of HAI-1, HAI-2 and Matriptase in the diabetic nephropathy dataset was checked against that of healthy living with diabetic nephropathy (34). We checked the gene expressions of glomeruli (threshold; p<0.05, fold change>1.5) and tubulointerstitium (threshold; none).

Statistical Analysis
All values are presented as mean ± SD. Statistical analysis ANOVA was performed and Tukey's method was used to compare groups or twotailed t-test. A level of p<0.05 was considered statistically significant.

Data availability
The raw data sets from the microarray analysis generated in this study have been deposited and available under accession number E-GEAD-374 and E-GEAD-387 in DNA Data Bank of Japan (DDBJ; https://ddbj.nig.ac.jp/). The rest of the data is contained within the manuscript. and at 6 weeks (N=3), and ADR nephropathy mice at 2 weeks (N=3) and at 6 weeks (N=3). Statistical significance (p<0.05) was calculated using moderated Student's t-test followed by the Benjamini-Hochberg false discovery rate correction on GeneSpring GX. Fold change cut-off of > 2 lead to extract 357 probes in ADR nephropathy mice at 6 weeks compared with control mice. Among those genes, 15 genes were categorized as serine protease while only Matriptase is a type II membrane anchored serine protease. Heatmap shows that all serine proteases were upregulated at 42 days in glomerulus of mice with ADR nephropathy. (B) Among 7 probes, which are categorized as membrane anchored serine proteinase and its cognate inhibitors, microarray analysis shows that only Matriptase and HAI-1 were significantly upregulated in ADR-nephropathy mice compared with control mice (B, C, D). Hgfac, hepatocyte growth factor activator; Hpn, hepsin; Prss8, serine protease 8.