Towards an Understanding of the Role of Matriptase in Normal Physiology

Home , Matriptase

FACULTY OF SCIENCE UNIVERSITY OF COPENHAGEN

PhD Thesis Sine Godiksen

Academic advisors: Lone Rønnov-Jessen and Lotte K. Vogel

Submitted: 17/09/2012

Towards an Understanding of the Role of Matriptase in Normal Physiology

Submitted to:

The Graduate School of Science, Faculty of Science

Department of Biology, University of Copenhagen, Denmark

For the PhD Degree

Sine Godiksen

Department of Biology, University of Copenhagen

Universitetsparken 13, 2100 København Ø

and

Department of Cellular and Molecular Medicine, University of Copenhagen

Blegdamsvej 3, 2200 København N

September 2012 Preface

This thesis is submitted to the Faculty of Science, University of Copenhagen, Denmark as basis for obtaining the PhD degree. I received a personal scholarship from the Faculty of Science in October 2007 and started my PhD studies in January 2008. My PhD thesis entitled: ”Towards an understanding of the role of matriptase in normal physiology” is based on the work performed from November 2009 to August 2011 in the Lab of Lotte Vogel, Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, with Dr. scient Lone Rønnov-Jessen, Department of Biology, Faculty of Science, University of Copenhagen as faculty supervisor and Lotte Vogel as supervisor.

Additionally, I have included work performed in the Lab of PhD Thomas H. Bugge, National Institutes of Health, Bethesda, MD, USA under a leave of absence from my PhD studies from September 2011 – April 2012 (paper II).

The data obtained for this thesis have been presented both at national and international scientific meetings and the results have resulted in three scientific papers; two published manuscripts and a manuscript in progress (see list of papers).

My research was financially supported by the Faculty of Science, Danish Cancer Society, the Augustinus Foundation, Købmand Kristian Kjær og Hustrus Foundation - the Kjær-Foundation, Dagmar Marshall´s Foundation, Snedkermester Sophus Jacobsen og Hustru Astrid Jacobsen´s Foundation, Grosserer Valdemar Foersom og Hustru Thyra Foersom´s Foundation, Fabrikant Einar Willumsens Mindelegat.

Sine Godiksen Cand. Scient. Copenhagen, September 2012 Table of Contents

ACKNOWLEDGEMENTS ...... 3

LIST OF PAPERS ...... 4

ABSTRACT ...... 5

DANSK RESUMÉ ...... 7

LIST OF ABBREVIATIONS ...... 9

INTRODUCTION ...... 11

AIMS AND OBJECTIVES ...... 13

BACKGROUND ...... 14

Membrane-anchored serine proteases ...... 14

Matriptase ...... 16 Matriptase expression and structure ...... 16

Regulation of matriptase ...... 18 Matriptase inhibitors ...... 18 Matriptase proteolytic processing and activation ...... 20

Prostasin ...... 22

Physiological functions of matriptase and its role in pathological processes ...... 23 Matriptase is crucial for epidermal integrity ...... 23 The matriptase-prostasin proteolytic pathway(s) in the epidermis ...... 24 Matriptase has an important role in epithelial homeostasis ...... 25 Importance of matriptase regulation in vivo ...... 26 Matriptase in carcinogenesis ...... 27

TECHNICAL CONSIDERATIONS...... 29

Cell system ...... 29

Protease pull down assays ...... 30

Biotinylation assays...... 32

RESULTS ...... 35 Paper I ...... 36

Paper II ...... 47

Paper III ...... 65

DISCUSSION AND PERSPECTIVES ...... 85

REFERENCES ...... 91

SUPPLEMENTARY I ...... 105

Acknowledgements

During the process of gaining experimental data and knowledge to complete this thesis and obtain my PhD there are many people whom I would like to thank. Without their continuous help, encouragement and support, this work would not have been possible. I would hereby like to express my gratitude to:

Lone Rønnov-Jessen for taking on the job as my supervisor in difficult times. Thank you for your great support and for your invaluable guidance in my writing process.

Lotte K. Vogel for giving me the opportunity to finish my PhD studies, and for your support throughout my PhD.

Thomas Bugge for giving me a very interesting and inspiring stay in his lab in Bethesda and for his excellent scientific guidance.

Jan K. Jensen for valuable discussions throughout my PhD studies.

All of my colleagues at ICMM, University of Copenhagen and NIDCR, National Institutes of Health (NIH), for a very pleasant, helpful and inspiring environment. Stine in particular, for support, scientific discussions, and for establishing the getaway to NIH.

Karen Skriver for your interest and deeply valued support throughout my PhD.

Annette Storgaard for your professionalism and for your guidance.

Family and friends for your constant support and encouragements.

And finally my beloved Peter, “tak for mad” and for being the most supportive and understanding “kæreste” anyone could ask for.

List of papers

Papers included in the thesis:

Paper I Transport via the transcytotic pathway makes prostasin available as substrate for matriptase. Stine Friis, Sine Godiksen, Jette Bornholdt, Joanna Selzer-Plon, Hanne Borger Rasmussen, Thomas H. Bugge, Chen‐Yong Lin and Lotte K. Vogel

Paper II Reduced prostasin (CAP1/PRSS8) activity eliminates HAI-1 and HAI-2 deficiency-associated developmental defects by preventing matriptase activation. Roman Szabo, Katiuchia Uzzun Sales, Peter Kosa, Natalia A. Shylo, Sine Godiksen, Karina K. Hansen, Stine Friis, Silvio Gutkind, Lotte K. Vogel, Edith Hummler, Eric Camerer, and Thomas H. Bugge

Paper III Novel assay for detection of active matriptase. Sine Godiksen and Lotte Vogel. Other authors to be included.

Papers not included in the thesis:

Bornholdt J, Friis S, Godiksen S, Poulsen SS, Santoni‐Rugiu E, Bisgaard HC, Bowitz Lothe IM, Ikdahl T, Tveit KM, Johnson E, Kure E, Vogel LK. The level of claudin‐7 is reduced as an early event in colorectal carcinogenesis. BMC Cancer, 2011, Feb 10; 11:65.

Abstract

Matriptase is a type II membrane-anchored serine protease essential for epithelial integrity and with implication in carcinogenesis. Matriptase is co-expressed in epithelial cells of most tissues with the protease prostasin and their mutual inhibitor hepatocyte growth factor activator inhibitor 1 (HAI-1). All three proteins have crucial functions for epidermal integrity; and in the epidermis matriptase acts upstream of prostasin and is required for its activation. The difference between the subcellular locations of matriptase and prostasin has so far been an enigma for matriptase-prostasin interaction. By mapping the subcellular trafficking of matriptase, prostasin and HAI-1, we demonstrate that the basolateral plasma membrane is part of the subcellular itinerary of both proteases and thus constitutes a location for interaction. In Caco-2 cells, zymogen matriptase is routed to the basolateral plasma membrane where it becomes activation site cleaved and activated. At steady state, prostasin is found to locate mainly to the apical plasma membrane although a minor fraction of prostasin can be detected at the basolateral plasma membrane. We show that prostasin is present in its active form on both the apical and basolateral plasma membrane and that prostasin as well as HAI-1 are endocytosed from the basolateral plasma membrane and transcytosed to the apical plasma membrane. Activation of matriptase on the basolateral plasma membrane is followed by efficient internalization as a matriptase-HAI-1 complex. Deregulated or unopposed matriptase activity causes a perturbation of several biological functions e.g. developmental defects in mice. However, the molecular mechanisms behind these effects are poorly understood. We performed a genetic epistatic analysis to identify new components of matriptase-dependent pathways in embryogenesis. We show that prostasin is an indispensable component of the matriptase-dependent proteolytic cascade that causes early embryonic lethality in mice. In placental tissue prostasin acts upstream of matriptase and is required for the activation of matriptase. During embryonic development both HAI-1 and HAI-2 are essential inhibitors of matriptase. In this study we moreover find that both HAI-1 and HAI-2 are indispensable inhibitors of prostasin in placental tissue. Matriptase activity is a tightly regulated protease; zymogen and HAI-1-complexed matriptase is abundantly present in most epithelial cell lines and tissues. There are no specific substrates or inhibitors of matriptase, which has led to challenges in detecting active matriptase. In order to assess the location and relative amount of active matriptase as compared to the total amount, we have established an assay for detection of active matriptase. This assay is based on a chloromethyl ketone peptide inhibitor with a predicted substrate sequence of matriptase. We show that active matriptase is present on the basolateral plasma membrane of Caco-2 cells and merely comprise a fraction of total matriptase. The conversion of zymogen matriptase to active matriptase can be induced by exposure to slight acidic conditions as well as the levels of HAI-1- complexed matriptase and active matriptase rises under these conditions in Caco-2 cells. Labeling of Caco-2 cells with the chloromethyl ketone peptide inhibitor delayed matriptase-HAI-1 complex formation. In addition to the peptide inhibitor being unable to bind HAI-1-complexed matriptase, we show that the matriptase zymogen has an intrinsic activity that enables it to bind the chloromethyl ketone peptide inhibitor. Moreover, this assay can be easily modified to detect

5 active matriptase in other cell system, and we show that cultured primary murine keratinocytes contain low levels of peptidiolytic active matriptase. Taken together, these results have improved our knowledge of the interplay between matriptase and prostasin towards a understanding of these components roles in epithelial integrity and present an assay for detection of active matriptase.

Dansk resumé

Matriptase er en type II membranbunden serinprotease, der er essentiel for integriteten af epitelvæv og er endvidere involveret i udvikling af kræft. Matriptase ekspreseres sammen med proteasen prostasin og deres fælles proteasehæmmer hepatocyte growth factor activator inhibitor 1 (HAI-1) af epitelceller i mange væv. Alle tre proteiner har en vital funktion for integriteten af epidermis, og matriptase er nødvendig for aktivering af prostasin i dette væv. Forskelle i subcellulær lokalisering mellem matriptase og HAI-1 på basolateralmembranen og prostasin på apikalmembranen af polariserede epitelceller har hidtil været et uafklaret dilemma i forhold til de to proteasers samspil. Kortlægning af den intracellulære transport af matriptase, prostasin og HAI-1 i Caco-2 celler angiver den basolaterale plasmamembran som en mulig subcellulær lokalitet, hvor matriptase og prostasin kan interagere. Zymogen matriptase exocyteres til basolateralmembranen, hvor proteasen kløves proteolytisk til den aktive form. Ved ”steady state” befinder prostasin sig primært på apikalmembranen, omend en mindre fraktion kan detekteres på basolateralmembranen. Vi viser, at aktivt prostasin findes på både apikal -og basolateralmembranen, og at prostasin såvel som HAI-1 internaliseres fra basolateralmembranen og transcyteres til apikalmembranen. Aktivering af matriptase på basolateralmembranen efterfølges af internalisering af matriptase i kompleks med HAI-1. Dereguleret og ukontrolleret matriptase aktivitet er en tilgrundliggende årsag til uligevægt i flere vigtige biologiske processer, f.eks. udviklingsmæssige defekter i mus. Vi udførte genetiske analyser i mus for at identificere nye komponenter i matriptase-afhængige signalveje. Disse resultater viser, at prostasin er en essentiel komponent i den matriptase- afhængige proteolytiske kaskade, der resulterer i embryonal dødelighed i mus. Vi viser, at prostasin ligger opstrøms for matriptase og er nødvendig for aktivering af matriptase i murinet placental væv. Både HAI-1 og HAI-2 er tidligere vist at være vigtige proteasehæmmere af matriptase i placenta. I dette studie viser vi, at dette også er gældende for prostasin. Matriptase er højt reguleret på posttranslationel niveau, hvilket indikeres af, at zymogen matriptase og matriptase i kompleks med HAI-1 let detekteres i mange cellelinier og epitelvæv. Der er ikke identificeret en specifik inhibitor eller et specifikt substrate for matriptase, hvilket har betydet, at detektering af aktivt matriptase har været udfordrende. Vi udviklede et assay til detektering af aktivt matriptase baseret på en klorometylketon peptidinhibitor med en peptidsekvens, der afspejler en foretruknen substratesekvens for matriptase. Vi viser, at aktivt matriptase findes på basolateralmembranen; denne udgør dog kun en fraktion af total matriptase. Zymogen aktivering af matriptase kan induceres ved et let fald i pH, hvormed mængden af aktiv matriptase stiger, ligesom kompleksdannelse med HAI-1 induceres under disse forhold. Ydermere forsinker mærkning med klorometylketon peptidinhibitoren kompleksdannelse mellem matriptase og HAI-1; tilsvarende er inhibitoren ikke i stand til at mærke matriptase-HAI-1 komplekset. Vi viser endvidere, at zymogen matriptase er i stand til at binde klorometylketone peptidinhibitoren. Ydermere kan dette assay let modificeres til andre cellesystemer, og vi viser med dette assay tilstedeværelsen af aktivt matriptase i kulturer af primære murine keratinocytter.

Disse resultater har udvidet vores viden om samspillet mellem matriptase and prostasin mod en forståelse af disse proteases rolle for integriteten af epitel og præsenterer et assay til detektering af aktiv matriptase.

List of Abbreviations

aa Amino acid Arg Arginine ARIH Autosomal recessive ichthyosis with hypotrichosis Asp Aspartic acid cDNA Copy deoxyribonucleic acid CAP Channel activating protease CDCP1 CUB domain-containing protein 1 CUB Cls/Clr, urchin embryonic growth factor, and bone morphogenic protein-1 DESC Differentially expressed in squamous cell carcinoma gene EGFR Epidermal growth factor receptor ENaC Epithelial Na+ channel ER Endoplasmic reticulum Gly Glycine GPI Glycosylphosphatidylinositol Gpld1 GPI-specific phospholipase D1 HAI-1 Hepatocyte growth factor activator inhibitor 1 HAI-2 Hepatocyte growth factor activator inhibitor2 HAT Human airway trypsin-like HGF/SF Hepatocyte growth factor/scatter factor HGFA Hepatocyte growth factor activator His Histidine IGFBP-rP1 Insulin-like growth factor binding protein-related protein1 IP Immunoprecipitation KD1 N-terminal Kunitz domain KD2 C-terminal Kunitz domain kDa Kilo dalton LDLR Low density lipoprotein receptor class A Lys Lysine MAM Meprin, A5 antigen, and receptor protein phosphatase μ MANSC Motif at the N-terminal containing seven cysteines MDCK Mardin Darby canine kidney (cell) mRNA Messenger ribonucleic acid MMP-3 Matrix metalloproteinases 3 MSP-1 Macrophage stimulating protein-1 MSPL Mosaic serine protease large-form MT-SP1 Membrane type serine protease 1 (matriptase) NHS N-hydroxysulfosuccinimide PAR-2 Protease-activated receptor-2 PDGF Platelet-derived growth factor PN1 Protease nexin 1

PRSS14 Protease serine S1 family member 14 PS-SCL Positional scanning synthetic combinatorial library S1P Sphingosine 1-phosphate SA Signal anchor SDS-PAGE Sodium dodecyl sulphate polyacrylamide gel electrophoresis SEA Sea urchin sperm protein, enteropeptidase, and agrin Ser Serine SIMA135 Subtractive immunization M(+)HEp3 associated 135 kDa protein S-NHS-SS-Biotin Sulfosuccinimidyl-2-[biotinamido]ethyl-1,3-dithiopropionate SPD Serine protease domain ST14 Suppressor of tumorigenicity-14 TADG-15 Tumor-associated differentially expressed gene-15 TEER Transepithelial electrical resistance TMPRSS Transmembrane protease, serine tPA Tissue plasminogen activator TTSP Type II transmembrane serine protease uPA Urokinase-type plasminogen activator VEGFR-2 Vascular endothelial growth factor receptor 2

Introduction

Proteolysis is a critical event in many biological processes from simple protein degradation in nutrient digestion to specific alterations in protein activity in highly regulated protease cascades, e.g. the blood coagulation system [1]. Formerly serine proteases represented enzymes that were either secreted or sequestered in cytoplasmic storage organelles awaiting signal-regulated release, but analysis of the human genome at the turn of the millennium revealed a new subclass of serine protease; the membrane-anchored serine proteases. With this family, a whole new range of important biological processes was shown to be regulated by serine proteases [2;3]. Matriptase is a type II membrane-anchored serine protease and is especially interesting due to its role in carcinogenesis and due to its importance for development and maintenance of epithelial integrity [4-8]. However, little is known of matriptase´s role at the molecular level in carcinogenesis. Of importance is the finding that a modest over-expression of wild-type matriptase in the epidermis of transgenic mice is sufficient to promote spontaneous squamous cell carcinoma formation. A simultaneous increase in expression of matriptase´s cognate inhibitor hepatocyte growth factor activator inhibitor 1 (HAI-1) completely negates the oncogenic phenotype of matriptase over-expression suggesting that the unopposed matriptase activity is the underlying cause [6]. Studies of matriptase knock-out mice have revealed critical functions for matriptase in development and maintenance of multiple epithelial tissues, which is in accordance to its widespread epithelial distribution in both mice and humans [4;5;9-12]. In humans, mutations in the ST14 gene encoding matriptase cause congenital ichthyosis [13-15]. Matriptase acts upstream of the glycosylphosphatidylinositol (GPI)-anchored serine protease prostasin in terminal epidermal differentiation and evidence suggests that an impairment of this cascade is the underlying cause of the symptoms observed in these patients [13;15-19]. Moreover, inhibition of matriptase is an essential function of HAI-1 in maintaining epidermal homeostasis in mice and zebrafish, and HAI-1 has been shown to inhibit both matriptase and prostasin in cultured human keratinocytes [20-22]. In contrast to most enzymatic reactions, proteolysis is a “one-way” process, which requires strict regulation of protease activity. Spatial distribution and confinement of proteases to specific cellular compartments is one way to ensure that proteolytic processing occurs under the correct conditions. Matriptase, prostasin and HAI-1 are co-expressed in polarized epithelial cells and HAI- 1 is an inhibitor of both proteases [9;11;21;23;24]. Polarized eepithelial cells have a plasma membrane that is divided into an apical compartment and a basolateral compartment. Both HAI-1 and matriptase is located at the basolateral plasma membrane in numerous epithelial tissues and cell types whereas, prostasin is located at the apical plasma membrane [11;23-26]. This difference in spatial distribution is in conflict with the proposed matriptase-prostasin proteolytic cascade [16]. We have earlier shown in a recombinant system that HAI-1 is transported to the basolateral plasma membrane. After endocytosis, a fraction of HAI-1 is transcytosed across the polarized cell to the apical membrane compartment, suggesting how HAI-1 can access and inhibit both matriptase and prostasin (supplementary I; [27]). However, it is still uncertain how matriptase is

11 able to activate prostasin in light of their different subcellular localization. Nevertheless, the global co-expression of matriptase and prostasin could indicate that this proteases cascade has a general role in maintaining epithelial integrity. Still, it remains unclear what substrates and pathways are involved in matriptase-dependent epithelial homeostasis, and if prostasin is a global downstream target for matriptase. The life cycle of matriptase is very complex and tightly regulated. Matriptase is believed to have a role as a protease at the pinnacle of protease cascades due to its ability to autoactivate [28;29]. Matriptase activation and inhibition by HAI-1 is shown to be intimately linked and has lead to the hypothesis that matriptase only exist in its free active form for a limited period and therefore only has a narrow time window to act on its substrate(s) before being pacified by inhibitor complex formation [21;30;31]. Thus, experimental tools for detection of active matriptase are highly desirable. Hence, matriptase is an important serine protease essential for epithelial integrity and with implications in cancinogenesis. Still, it remains to be elucidated where in the polarized cell matriptase is activated and able to act on downstream substrate(s). Moreover, identification of additional components of matriptase-dependent proteolytic pathways would provide valuable insights into the biology of matriptase-dependent biological processes.

This thesis is written as a synopsis and gives a brief introduction to membrane-anchored serine proteases and matriptase in particular. Mechanisms for matriptase activation and inhibition will be described with an introduction of matriptase´s most important inhibitors, HAI-1 and HAI-2, and its epidermal downstream target prostasin. Finally, a description of matriptase in relation to its physiological functions and its role in pathological processes will be given. The background section is followed by aims and objectives of this thesis and technical considerations. The results obtained are enclosed as two published papers and one manuscript in progress. Finally, the results will be discussed in relation to previous studies.

Aims and objectives

The overall aim of this study was to gain a more comprehensive understanding of matriptase in epithelial biology. We wanted to delineate how matriptase can act as an upstream activator of prostasin in epidermal terminal differentiation, despite their different subcellular localization in polarized epithelial cells at steady state, by mapping the intracellular trafficking of matriptase, prostasin and HAI-1 (paper I). We furthermore wished to determine where in the polarized cell matriptase is activated and able to cleave its substrates (paper I and paper III). More so, we wanted to establish an assay for detection of active matriptase on the surface of living cells (paper III). One way of elucidating the physiological functions of proteins is by identification of interaction partners, upstream regulators and downstream effectors. Accordingly, we aimed at identifying new components of matriptase-dependent proteolytic pathways critical for embryogenesis in mice (paper II).

Background

To understand the importance of matriptase and the relationship between matriptase, prostasin and their inhibitor HAI-1, this chapter gives an introduction to these proteins as well as the mechanisms of matriptase activation. Afterwards, I will outline important physiological functions of matriptase, its interplay with its inhibitors and prostasin in these processes, and finally review the role of matriptase in cancer.

Membrane-anchored serine proteases Proteases are enzymes that conduct proteolysis and were originally discovered a century ago as enzymes involved in nutrient digestion. Sequencing of the human genome at the turn of the millennium revealed a total 569 proteases in humans. On the basis of the mechanism of catalysis, proteases are classified into five distinct classes: aspartic, metallo, cysteine, serine and threonine proteases [32]. Serine proteases are one of the largest and most conserved multigene proteolytic families. This protease family includes the well-known digestive enzymes trypsin and chymotrypsin. Serine proteases are found in a variety of tissues and body fluids with well characterized roles in diverse cellular functions including blood coagulation, wound healing, digestion and immune response. Furthermore, many studies indicate that these proteases contribute to a number of pathological conditions and play a significant role in tumour growth, invasion and metastasis [33;34]. Serine proteases are characterized by the presence of a serine residue in the active site of the catalytic domain. In the active site resides the catalytic triad that is preserved in all serine proteases. This catalytic triad is composed of three amino acids; a histidine, a serine and an aspartic acid, which are essential for the catalytic ability of the protease. The specificity of the protease is determined by the size, shape and charge of the active site cleft [33]. The rate of discovery of new (serine) proteases was greatly aided and accelerated by the publishing of the mouse and human genome sequences and EST databases. This led to the discovery of the hitherto unknown large family of membrane-associated serine proteases; the membrane-anchored serine proteases. Orthologues are found in all vertebrates and in humans 20 members have been identified so far as presented in fig. 1 [35;36].

Membrane-anchored serine proteases are tethered to the membrane either via a C-terminal transmembrane domains (Type I), a GPI-anchor or via an N-terminal transmembrane domain (Type II) as illustrated in fig. 1. The common features of Type II transmembrane serine proteases (TTSP) are a short cytoplasmic anchor, an N-terminal transmembrane domain, a C-terminal extracellular serine protease domain of the trypsin (S1) fold, and a variable length stem region containing modular domains linking the catalytic and the transmembrane domains. This stem region contains an assortment of 1-11 protein domains of six different types [3;35].

Fig 1. Overview and domain structure of the human membrane-anchored serine proteases. Prostasin and testisin is anchored to the membrane by a GPI-anchor, whereas tryptase γ1 is anchored to the membrane through a C-terminal transmembrane domain (type I transmembrane protein). The remaining membrane-anchored serine proteases presented here are type II transmembrane proteins and are attached to the membrane by a signal anchor (SA) located close to the N terminus with a cytoplasmic extension. Type II transmembrane serine proteases are phylogenetically divided into four subfamilies; the largest being the human airway trypsin-like (HAT)/differentially expressed in squamous cell carcinoma gene (DESC) subfamily, which comprises HAT, DESC1, TMPRSS11A, HAT-like 4, and HAT-like 5 (blue shading); the hepsin/transmembrane protease, serine (TMPRSS) subfamily, which comprises hepsin, TMPRSS2, TMPRSS3, TMPRSS4, mosaic serine protease large-form (MSPL), spinesin, and enteropeptidase (yellow shading); the matriptase subfamily, which consists of matriptase, matriptase-2, matriptase-3, and polyserase-1 (green shading); and the corin subfamily, which contains only corin (purple shading). Beside a serine protease domain (SPD) of the S1 fold, the modular structure of the TTSP can contain an assortment of different domains that include sea urchin sperm protein, enteropeptidase, and agrin (SEA); group A scavenger receptor (scavenger); low-density lipoprotein receptor class A (LDLR); Cls/Clr, urchin embryonic growth factor, and bone morphogenic protein-1 (CUB); meprin, A5 antigen, and receptor protein phosphatase μ (MAM); and frizzled. The figure is from [3].

This complex modular structure of most TTSPs is remarkably different from other trypsin-like serine proteases. The multi-domain structure provides the TTSPs with the capacity to interact with several interaction partners and possibly a mean for regulation of proteolytic activity [36;37]. Several TTSPs are involved in fundamental cellular and developmental processes such as morphogenesis, differentiation, epithelial permeability and cellular iron transport [3].

Matriptase Matiptase was first identified in 1993 as a new gelatinolytic activity expressed by cultured breast cancer cells and later 5 different groups individually cloned the cDNA from human prostate cancer cells, human ovarian carcinoma, human breast cancer cells, murine thymic cells, and human colon mucosa [38-43]. Shortly after matriptase was purified from human milk [44]. The molecular cloning of the two closely related proteases matriptase-2 and matriptase-3 was reported later [45;46]. Matriptase is also referred to as MT-SP1, TADG-15, PRSS14, TMPRSS14, SNC19, cleaving activating protease (CAP)-3, prostamin, epithin (mouse ortholog) and the gene designation is suppression of tumorigenicity 14 (ST14). For simplicity, I will use matriptase also when discussing the mouse orthologue.

Matriptase expression and structure Orthologues of matriptase has been found in many vertebrates but unlike most other TTSPs, matriptase is widely expressed in epithelial compartments of many embryonic and adult tissues [10;11]. Expression analysis has been done on both human and murine tissues. mRNA and protein analyses show that matriptase is expressed in all types of epithelium, including columnar, pseudostratified, cubiodal, and squamous in humans and locates to the basolateral plasma membrane [11]. Using enzymatic gene trapping with -galactosidase as a reporter in mice showed an expression pattern of matriptase in mice overall identical to what was found in humans [12]. The high degree of similarity between expression profiles of different species suggests conserved function(s) of matriptase. Full-length human matriptase is an 855 amino acid (aa) glycoprotein that lacks a classical signal peptide of app. 95 kDa. Instead the N-terminal signal anchor, which is not removed during synthesis, functions as a single-span transmembrane domain that orientates the protease in the plasma membrane as a type II integral membrane protein (fig. 1 and 2). Matriptase is a mosaic protein and is composed of a short N-terminal tail (residues 1-54) followed by the transmembrane domain, a sea urchin sperm protein, enterokinase, agrin (SEA) domain (residues 85-193), two C1r/s, urchin embryonic growth factor and bone morphogenetic protein 1 (CUB) domains (residues 214-334 and 340-447), four low-density lipoprotein receptor class A (LDLR) domains (residues 452-486, 487-523, 524-561, and 566-604) and the C-terminal trypsin-fold S1 serine protease domain (SPD; residues 614-855) with the catalytic triad composed of the residues Ser805, Asp711, and His656 [47]. Serine proteases of the S1 fold including matriptase display a pronounced specificity for the positively charged amino acids arginine and lysine at the P1 position [39;46;48-50].

Fig 2. Schematic presentation of matriptase. Matriptase consists of a transmembrane signal anchor domain, a SEA domain; two CUB domains; 4 LDLR domains; and a class S1 serine protease domain.

The extended substrate specificity is determined by the topology of the binding pocket and the extended specificity profile of matriptase was determined with the solution of the crystal structure of the serine protease domain of matriptase and by using positional scanning synthetic combinatorial library (PS-SCL) and substrate phage library. By these methods, the preferred cleavage sequences (P4-P3-P2-P1 P1´) of matriptase was identified to be R/K-XSR A and X-R/K- SR A, where X is a non-basic amino acid [50-52]. Thus unlike trypsin, matriptase does not indiscriminately cleave peptide substrates after Lys or Arg but requires recognition of additional residues. The extended substrate profile correlates very well with the activation motif of matriptase itself (RQAR VVGG) suggesting that matriptase can be activated by a transactivating mechanism.

In vitro and cell culture studies have identified several possible substrates of matriptase; pro- hepatocyte growth factor/scatter factor (HGF/SF) [53;54], pro-macrophage-stimulating protein 1 (MSP-1) [55], pro-urokinase-type plasminogen activator (uPA) [51;53;56], protease activated receptor 2 (PAR-2) [51;57], matrix metalloproteinases 3 (MMP-3) [58], pro-kallikreins [59], immunization M(+)HEp3 associated 135 kDa protein/CUB domain containing protein 1 (SIMA135/CDCP1) [60], insulin-like growth factor binding protein-related protein 1 (IGFBP-rP1) [61], vascular endothelial growth factor receptor 2 (VEGFR-2) [62], platelet-derived growth factor (PDGF) [63;64], epidermal growth factor receptor (EGFR) [65;66], fibronectin [26], laminin [26;67] and gelatine [39]. The function of the intracellular domain is unclear, though data suggest a function in interaction with the cytoskeleton to regulate localization of matriptase on the cell surface to micro domains of the plasma membrane through interaction with the actin-associated protein filamin [30;68]. The extracellular non-catalytic domains are likely to function in protein-protein interactions that modulate localization, activation and inhibition and possibly substrate specificity [28;29;35]. E.g. in the complement protease cascade, CUB domains are important for the formation of C1r/C1s tetramers prior to binding of C1q and activation of the C1r and C1s proteases within the complex [69].

Regulation of matriptase Under physiological conditions proteases are strictly regulated in time and space, thereby restricting their activity to specific subcellular sites and limiting their access to substrates. This regulation is achieved by controlled zymogen activation and interaction with inhibitors [2].

Matriptase inhibitors The serpin family is the largest family of serine protease inhibitors, and complexes between matriptase and the serpins anti-thrombin III, α1-antitrypsin, and α2-antiplasmin have recently been purified from human milk and epithelial cell lines, and protease nexin 1 (PN1) complexes with matriptase was reported in vitro [70-74]. However, the physiological relevance of serpin- mediated inhibition of matriptase has not yet been established. Instead, two members of the family of Kunitz-type serine protease inhibitors have been identified as essential inhibitors in regulation of matriptase. The Kunitz-type inhibitors are a class of serine protease inhibitors that make reversible interactions with their target proteases [31;75;76]. They are members of the type I transmembrane protein family and characterized by having one or two extracellular Kunitz-type serine protease inhibitor domains (KD), a single transmembrane spanning domain and a short C- terminal cytoplasmic domain [75;76]. They have an N-terminal cleavable signal sequence and a stop-transfer sequence that halts further translocation through the ER membrane and acts as a transmembrane anchor [77]. The extracellular part of the membrane bound protein can be released from the plasma membrane by ectodomain shedding (supplementary I [27]; [75]. While, HAI-1 has been purified in a complex with matriptase from human milk, seminal fluid and urine [44;78], complex formation between HAI-2 and matriptase has so far only been observed in vitro [79]. HAI-1 and HAI-2 were originally identified as potent inhibitors of hepatocyte growth factor activator (HGFA) from the conditioned medium of the human stomach carcinoma cell line, MKN45 [80;81]. HAI-1 is the most well studied inhibitor of matriptase and is ubiquitous expressed in epithelia and predominately localizes to the basolateral surface of epithelial cells, especially the simple columnar epithelium of ducts, tubules and mucosal surfaces [10;24]. HAI-1 is also expressed in the endothelial cells of capillaries, venules and lymph vessels [82]. Tissue distribution of HAI-2 has been assessed in both human and mouse. In situ hybridisation of adult human samples showed a distribution of the inhibitor in epithelial layers of e.g. placenta, pancreas, kidney and prostate [81]. Tissue distribution of HAI-2 in mice was determined using enzymatic gene trapping and -galactosidase as a reporter system, and showed a co-localization of HAI-2 with HAI-1 and matriptase in most epithelial tissue [79]. Additionally, HAI-2 and HAI-2 are also co-expressed in developing neural tube [79;83]. Human HAI-1 is composed of 513 amino acids with a 35 amino acid signal peptide, thus, the mature full-length form is composed of 478 amino acids. Full-length HAI-1 has a calculated molecular weight of 53,319 Da [29;84] and has three potential N-glycosylation sites at Asn66, Asn235 and Asn507 [80]. HAI-1 is composed of two extracellular Kunitz domains, KD1 (residues fd 250-300) and KD2 (residues 375-425), an extracellular LDLR domain (residues 319-353), a motif at hg fhj 18 the N-terminal containing seven cysteines (MANSC) (residues 57-147), a transmembrane domain (residues 450-472) and a short C-terminal cytoplasmic domain as outlined in fig. 3 [80].

Fig. 3. Schematic presentation of HAI-1. Domain structure of HAI-1 consists of a MANSC domain, two Kunitz-type domains separated by a LDLR domain and a transmembrane domain. HAI-1 is a type I membrane-anchored protein.

The overall topology of HAI-2 is similar to that of HAI-1. Human HAI-2 is composed of 252 amino acids with a 27 amino acid signal peptide, thus, the mature full-length form is composed of 225 amino acids with a calculated molecular mass of 25,415 Da [81]. HAI-2 is composed of two extracellular Kunitz domains, KD1 (residues 38-88) and KD2 (residues 133-183), and a short C- terminal region (24 aa) as outlined in fig. 4. HAI-2 has two putative N-glycosylation sites at Asn57 and Asn94 [81].

Fig. 4. Schematic presentation of HAI-2. Domain structure of HAI-2 consists of a transmembrane domain and two Kunitz-type domains. HAI-2 is a type I membrane-anchored protein.

The amino acid in the P1 position of the reactive site is of essential importance for the inhibitory activity of KD [85]. For HAI-1 the essential amino acids in the P1 positions are arginine and lysine for KD1 and KD2, respectively, and for HAI-2 arginine is present in P1 of both KD1 and KD2 [86;87]. Site-directed mutagenesis studies suggest that KD1 of both HAI-1 and HAI-2 is responsible for the inhibitory activity against proteases, and for HAI-1 this is also shown for matriptase [84;87;88].

Matriptase proteolytic processing and activation Matriptase is synthesized as an inactive single-chain zymogen and its activation requires two sequential proteolytic processing events. The pro-enzyme is first processed at the N-termini of the stem domain after Gly149 within a conserved G SVIA motif of the SEA domain. This initial cleavage has been proposed to occur by non-enzymatic hydrolysis of the peptide bond, as it has been observed for other SEA containing proteins [89]. The second and actual activation site cleavage event occurs after Arg614 within the highly conserved R VVGG activation motif and is dependent on the first cleavage at Gly149 [28]. Hereby the disulfide-linked proteolytic active matriptase is formed (fig. 5). In spite of these processing events, the catalytic domain still remains attached to the plasma membrane due to the disulfide bond linking it to the stem domain, and strong hydrophobic interactions within the SEA domain [28;39;51;90]. It is the activity of the mature enzyme that is believed to be responsible for both the physiological and pathological functions of the enzyme identified by studies of animal models and human genetics.

For the greater part of serine proteases the activation cleavage depends on another upstream active protease(s), however evidence indicate that matriptase undergoes autoactivation both in solution and in the membrane-bound form caused by weak inherent activity of the matriptase zymogen [28;29;40;91]. First off, the resemblance of the preferred cleavage sequences to the activation site motif of matriptase [39;50]. Second and most importantly, mutations of any of the amino acids in the catalytic triad of matriptase (Ser805, Asp711, or His656) result in matriptase mutants that are processed at Gly149, but unable to undergo the final proteolytic processing at Arg614. This led to a proposed transactivation mechanism for activation of matriptase [28]. Autoactivation, by which two or more zymogen proteases interact and activate one another by virtue of their weak intrinsic proteolytic activity, is believed to be relevant for proteases at the apex of a protease cascades as the active site triad of serine proteases is preformed in the zymogen protease [92-94]. A well studied example of this mechanism is the activation of complement C1r protease zymogen where complex formation induce intramolecular conformational changes that result in activation by a transactivation mechanism [94]. Autoactivation has also been proposed as a mechanism of activation for several other members of the TTSP family including matriptase-2 and hepsin [45;93].

The activation of matriptase is extraordinary complex, and far from fully understood. Evidence suggests that the transactivation mechanism for matriptase rely on additional parameters for matriptase activation to occur including its own multi-domain structure, the plasma membrane as well as HAI-1. Membrane anchorage of full-length matriptase is required but insufficient for activation as shown in studies of matriptase in simple vesicle structure implying that a higher degree of organization within the plasma membrane is indispensable for matriptase activation [29]. A series of deletion -and point mutations demonstrated that activation of matriptase requires glycosylation of the first CUB domain (Asn302) and of the catalytic domain, as well as the presence of intact LDLR domains [28;29;95]. Thus, autoactivation of matriptase may occur by interaction of two or more neighboring SEA domain processed matriptase zymogen molecules, and possibly other cofactors to induce the necessary conformational changes in the substrate binding pocket required for catalysis [28]. 20

Fig. 5. Proteolytic processing of matriptase and inhibition by HAI-1. Matriptase is synthesized as a 95 kDa Full-length protease that undergoes two sequential cleavages to become fully active. The first cleavage is after Gly149 in the SEA domain and the second cleavage is after Arg614 in the conserved activation site motif N-terminal to the serine protease domain. This processing results in a disulfide linked active form of matriptase. Following activation, matriptase forms a SDS-resident complex with HAI-1.

After activation, matriptase is inhibited by HAI-1 within a short time rendering active matriptase little time to act on substrates, see fig. 5 [21;30;31]. Even so, cell culture studies indicate that HAI- 1 also has a role for proper expression, activation and trafficking of matriptase [28;96]. However, the requirement for HAI-1 in these processes is poorly understood. In breast cancer cells that express neither matriptase nor HAI-1 endogenously, matriptase is retained in the ER/Golgi compartments unless the cells are simultaneously transfected with HAI-1. As no difficulties in trafficking of enzymatically dead matriptase (S805A) was observed in the absence of HAI-1, the inability of wild-type matriptase trafficking without HAI-1 was suggested to be caused by the potentially toxic effect of its own unopposed proteolytic activity [96]. Also matriptase activation was impaired when matriptase was co-expressed with HAI-1 mutated in its LDLR domain [28;96]. Thus, HAI-1 appears at all times to be available for matriptase, which could serve to protect the cell against aberrant matriptase proteolysis.

The mechanisms triggering matriptase activation remains largely unclear, although data suggest a role for a number of extracellular stimuli. The signaling sphingolipid sphingosine 1-phosphate

(S1P) was identified as a serum component capable of inducing matriptase activation in immortalized breast epithelial cells and the steroid sex hormone androgen is able to induce a robust but more slow activation of matriptase in human prostate cancer cells [30;31;97]. In spite of the divergence in the nature of matriptase activation in different cell types, the subsequent complex formation with HAI-1 seems to be uniform for both cell types. This is supported by the identification of suramin as a universal inducer of matriptase activation [30]. Also, exposure to a mildly acidic extracellular milieu or lowering of ionic strength induces robust and rapid matriptase zymogen activation in both cell free settings and in several epithelial cell types [29;31;95;98;99].

Although matriptase remains membrane-associated after proteolytic processing and activation it is clear that matriptase is shed from the plasma membrane given that the protease has been purified from human milk [44]. N-terminal sequencing of the matriptase isoforms isolated from conditioned media of an epithelial cell line showed proteolytic cleavage either after Lys189 in the SEA domain or after Lys204 in the linker region between the SEA domain and the CUB1 domain [31]. The mechanism(s) by which matriptase is released from the plasma membrane and the proteases mediating the release remain to be identified. However, cell culture studies suggest that shedding of matriptase is dependent on the protease being fully proteolytic processed and possibly complex-formation with HAI-1 [31;78;100].

Prostasin Prostasin is a GPI-anchored serine protease that was first isolated from human seminal fluid, but displays a widespread tissue distribution in both mice and humans and is co-expressed with matriptase in most murine epithelial tissues [9;101;102]. Prostasin is also termed channel activating protease 1 (CAP-1), following its ability to activate epithelial sodium channel (ENaC) [103;104]. Full-length human prostasin consists of 343 aa with a 32 aa N-terminal signal peptide that is proteolytically removed in the ER. Instead prostasin is modified with a C-terminal GPI- anchor, this result in a 40 kDa mature protein that consists of an extracellular serine protease domain tethered to the outer leaflet of the plasma membrane [102;105], see fig. 6. The serine protease domain contains the catalytic triad that is formed by His53, Asp102 and Ser206 [102;106]. Prostasin is synthesized as an inactive zymogen but unlike matriptase, prostasin is incapable of autoactivation and thus requires the proteolytic activity of a second protease to process the zymogen into the two-chain disulfide-linked active form [105;107]. In vitro matriptase activates prostasin by proteolytic cleavage in the amino-terminal pro-peptide region of prostasin at the Arg44 in human prostasin [16;65]. Like matriptase, prostasin has a preference for poly-basic substrates which also explain the inability of prostasin to autoactivate as the activation site motif of human prostasin is PQAR ITGG [107]. Prostasin is inhibited by PN1, HAI-1 and HAI-2 like matriptase [21;108;109]. Prostasin is shed from the apical plasma membrane. In prostate epithelium this secretion is depended on C-terminal processing at the conserved residue Arg322, while secretion from kidney and lung epithelial cells depends on GPI-anchor cleavage by the endogenous GPI-specific phospholipase D1 (Gpld1) [23;101].

Fig. 6. Schematic presentation of prostasin. Prostasin is posttranslationally modified with a GPI-anchor that tethers the single serine protease domain of prostasin to the membrane.

Physiological functions of matriptase and its role in pathological processes Matriptase has been shown to play a role in a number of different proteolytic processes at the cell surface; however, the mechanisms by which matriptase modulates its effect at the molecular level is still unclear. The importance of matriptase-dependent proteolysis is derived from combined knowledge from genetic disorders and transgenic animal models, as well as cell culture studies. Below I will review physiological and pathological processes in which matriptase plays a crucial role.

Matriptase is crucial for epidermal integrity In humans, a few studies have been reported identifying mutations in the matriptase gene all of which lead to various types of the skin disorder ichthyosis [13-15]. The first documented patients suffer from the rare skin disease autosomal recessive ichthyosis with hypotrichosis (ARIH) characterized by congenital ichthytosis associated with frizzy hair. The syndrome is caused by homozygosity for the missense mutation G827R in the serine protease domain of matriptase. The G827R mutant form of matriptase has a reduced proteolytic activity in vitro as the mutation affects access to the active site binding cleft [17;18]. Further information about the disease was obtained by mouse models of the disorder. Xenografting of matriptase-deficient skin onto adult athymic mice and the generation of a matriptase hypomorphic mouse model both phenocopied the microscopic hallmarks of the skin from ARIH patients [17;110]. Biochemical analysis of ST14 deficient murine and human epidermis revealed highly reduced prostasin activation and reduced processing of the abundant epidermal polyprotein pro-filaggrin into filaggrin [13-17;19;110].

Another example of matriptase´s importance in epidermal integrity is its recently shown implication in Netherton´s syndrome [59]. This disease is a form of ichthyosis characterized by premature stratum corneum shedding resulting in direct exposure of the living surface of the epidermis to the external environment and chronic inflammation [111-113]. The genetic

23 background of Netherton´s syndrome is a deficiency for the protease inhibitor LEKTI [113]. In a mouse model of the syndrome, matriptase initiate a run-away kallikrein proteolytic cascade in the absence of LEKTI that results in an over accumulation of processed filaggrin [59;114]. Additionally, an increase in matriptase expression has been suggested as a common basis for a range of different human skin disorders [115].

The matriptase-prostasin proteolytic pathway(s) in the epidermis Matriptase and prostasin are generally co-expressed in terminally differentiated cells of stratified epithelia that do not possess any proliferative capacity [9]. Studies of both human tissue samples and murine models as well as cell culture studies have revealed that both matriptase and prostasin are important factors in regulating terminal epidermal differentiation and hair follicle development [4;9;12;17;19;116]. The phenotypes observed in mice deficient of prostasin in the skin are nearly indistinguishable from matriptase deficient mice [4;19;116]. Both mouse models display compromised epidermal tight junctions and a generalized disruption of the stratum corneum architecture. The impaired epidermal barrier function is a result of perturbation of several processes that normally takes place during terminal epidermal differentiation. They include lipid matrix formation, formation of the water-impermeable cornified layer, and desquamation (shedding of corneocytes) as well as hyperproliferation of basal keratinocytes, see fig. 7. These defects result in a lack of terminal epidermal differentiation and are accompanied with an impaired skin barrier function resulting fatal dehydration and death within 48-60 hours postnatal [4;19;116]. On the molecular level, matriptase and prostasin deficiency results in greatly reduced proteolytic processing of pro-filaggrin [15;17;19;116;117]. Studies show that matriptase act upstream of prostasin in a matriptase-mediated proteolytic activation cascade most likely by directly activating the prostasin zymogen, which can be seen in fig. 7 [15-17;21]. However, the specific molecular mechanism by which matriptase and prostasin facilitates profilaggrin processing is unclear. The hypothesis is supported by the lack of active prostasin in cultured primary keratinocytes from patients with loss-of-function mutations in ST14 and in the epidermis of matriptase deficient mice, whereas active prostasin can be readily detected in wild-type human keratinocytes and in murine epidermis [4;15;16;21;116]. This correlates with activation of prostasin being suppressed by matriptase ablation in cultured human keratinocytes and that matriptase is able to activate prostasin in vitro [16;21]. Moreover, the two proteases display synchronized developmental onset of expression which correlates with acquisition of epidermal barrier function in mice [16].

Together this suggests that in the skin matriptase and prostasin are functioning in the same pathway and that matriptase act upstream of prostasin in this matriptase-prostasin proteolytic cascade essential for terminal epidermal differentiation. Furthermore, it has been suggested that HAI-1 plays an important role in regulating this proteolytic cascade as inhibitor complexes of HAI- 1 with both matriptase and prostasin have been detected in human cultured keratinocytes induced to differentiate into an organotypic culture model of the skin [21;115].

Fig. 7. Matriptase´s role in the epidermis (A) Matriptase expression is confined to the uppermost living layers of the interfollicular epidermis; the transitional layer (blue color, arrowhead). This is visualized with a knock-in mouse with a promoterless β-galactosidase marker gene inserted into the endogenous matriptase gene. The basal layer is visualized using a keratin-5 antibody (red color, arrow) [12]. Deficiency of either matriptase or epidermal prostasin leads to a range of epidermal defects including loss of tight junctions, lipid extrusion and impaired processing of pro-filaggrin that result in impaired epidermal barrier function (compare B and C). Figure is from [118].

Matriptase and prostasin also co-localize in a number of epithelia other than the epidermis suggestive of a role for a matriptase-prostasin proteolytic cascade in other organs as well [9]. However, although we know that matriptase and prostasin are linked, we have very restricted knowledge of the pathway(s) in which they function. Identification of components acting upstream or downstream of matriptase and prostasin would greatly add to our understanding of the role(s) of these proteases in vivo. Matriptase has also been proposed to be at the apex of a matriptase-prostasin proteolytic cascade activating ENaC that is important for maintenance of salt and water homeostasis by reabsorption of Na+-ions [119;120]. This channel is also important for normal terminal differentiation of the epidermis [121]. Both matriptase and prostasin are able to activate ENaC in Xenopus oocytes [103;120]. So far evidence for a physiological role in channel activation is limited to prostasin, where fluid clearance from the lungs are impaired in mice with alveolar epithelium- specific ablation of prostasin [122]. Intriguingly, the catalytic activity of prostasin is dispensable for its capability to activate ENaC in the Xenopus model system. Catalytically inactive prostasin induce both ENaC cleavage and activation, nonetheless cell surface expression of prostasin is still essential for activity [123;124].

Matriptase has an important role in epithelial homeostasis The generation of ST14 (the gene encoding matriptase) deficient mouse models established that matriptase has a critical role in the development and maintenance of multiple epithelial tissues

25 consistent with its widespread epithelial distribution [4;5;12]. Embryonic tissue specific ablation or acute ablation of matriptase in adult tissues of mice results in severe organ dysfunction and widespread epithelial demise, often associated with increased paracellular permeability, loss of tight junction function and mislocation of tight junction-associated proteins, see fig. 8. Ultimately these defects cause a reduction in body weight and ultimately death [5]. Matriptase deficiency in mice also affects hair follicle development and results in dramatically increased thymocyte apoptosis, and depletion of thymocytes [4;12].

Fig. 8. Matriptase´s role in the intestinal epithelia (A) Matriptase is expressed in goblet cells (arrow) and in surface mucosal cells (arrowhead) in the murine intestine. Sustained matriptase expression is essential for maintenance of the intestinal epithelia. Ablation of matriptase disrupts tight junctions and cell polarity (compare B to C) and causes architectural distortion and compromised barrier function in the large intestine resulting in edema and diarrhea leading to premature death in mice. Figure is from [118].

Matriptase is proposed to promote intestinal barrier recovery in injured intestinal mucosa of inflammatory bowel disease (IBD). Studies of human specimens and mouse models show that matriptase is down-regulated in inflamed colonic tissues from IBD patients and that matriptase has a protective role against colitis in mice [8;125]. Thus, matriptase has an essential role in development as well as maintenance of multiple types of epithelia.

Importance of matriptase regulation in vivo Under normal physiological conditions, proteases are strictly regulated at the protein level. Synthesis of proteases as inactive zymogens and complex formation with protease inhibitors limit their accessibility to substrates. The importance of protease inhibitors to govern protease activity is obvious, as the activation zymogens to active proteases is an irreversible action. Therefore, important knowledge of proteases can be obtained indirectly by knock-out studies of their respective inhibitors. This section presents important knowledge about matriptase obtained by studies of knock-out animal models of its physiological inhibitors, HAI-1 and HAI-2.

Even though matriptase is widely expressed in both embryonic and adult epithelial tissues the protease does not have critical non-redundant functions until after birth [4]. Surprisingly, the proteolytic activity of matriptase must nevertheless be under strict control during embryogenesis for correct placental development to take place. Ablation of HAI-1, HAI-2, or the combined haploinsufficiency for HAI-1 and HAI-2 leads to failure of placental labyrinth formation [15;126- 128]. Correct placental development can be restored with simultaneous ablation of matriptase, indicating that the placental defects in HAI-1-deficient, HAI-2-deficient or HAI-1/HAI-2 haploinsufficient mice are caused solely by unopposed matriptase activity [83;126-128]. Thus although matriptase is completely dispensable for embryogenesis, its activity needs to be strictly regulated by HAI-1 and HAI-2 for placental development to take place. HAI-1-mediated inhibition of matriptase also has an essential role in postnatal epithelial homeostasis. The generation of a chimeric mouse model with sufficient placental HAI-1, but ablation of HAI-1 in the embryo resulted in viable offspring [20;129]. These mice suffer from various defects of the keratinized epithelium and die before adulthood. The same genotype superimposed on hypomorphic expression of matriptase completely eliminates epidermal barrier and hair follicle defects in HAI-1 deficient mice [20;129]. Similar outcome is reported for zebrafish, where deletions of the HAI-1 orthologues Hai-1a and Hai-1b result in skin inflammation and a compromised epithelial integrity caused by detrimental matriptase activity that result in death app. 24 hrs after fertilization [22]. Excess local matriptase activity is detrimental not only to mouse placental labyrinth formation but also for neural tube closure, which became evident with the generation of mice deficient for HAI- 2. HAI-2 and matriptase are expressed in the neural and non-neural ectoderm precisely at the interface where the two neural folds fuse. Simultaneously ablation of matriptase however, only partially rescues this HAI-2 knock-out phenotype suggestive of several in vivo target of HAI-2 in neural tube closure [83]. In humans, a range of mutations in the spint2 gene encoding HAI-2 have been reported. These patients present a wide range of developmental abnormalities including duplication of internal organs and digits, craniofacial dysmorphisms, as well as anal and choanal atresia [130]. It remains to be established if these defects are caused by excess matriptase activity or unregulated activity of other serine proteases. Hence, strict and proper regulation of matriptase is important for postnatal survival, development and maintenance of several epithelia and partially for neural tube closure as these defects were caused by unregulated matriptase activity.

Matriptase in carcinogenesis 85 % of all cancers originate from epithelial cells that line the internal and external surfaces of the body. During transformation of normal epithelia into cancerous tissue a number of events occur; function-altering mutations of oncogenes and tumor suppressor genes, loss of epithelial cell polarity, concomitant tissue disorganization, penetration of the basement membrane and invasion of transformed epithelial cells into the underlying stroma [131]. Unlike most proteases involved in carcinogenesis which are expressed by the connective tissue supporting the epithelia, matriptase is expressed by the transformed epithelial cells themselves [6]. Matriptase is undoubtedly involved in carcinogenesis, however at present its role remains obscure. Knowledge

27 of matriptase´s role in cancer comes from studies of animal models, tissue samples and cancer cell lines. The oncogenic potential of matriptase was directly shown in transgenic mice, where a modest over-expression of wild-type matriptase in the skin caused spontaneous squamous cell carcinoma formation to occur. Moreover, matriptase supports both ras-dependent and independent carcinogenesis and potentiate the effects of genotoxic exposure [6]. Matriptase expression undergoes a dramatically spatial redistribution during the transition of epidermal lesions from hyperplasia to dysplasia and becomes present in the proliferating basal compartment. Hereby matriptase is able to mediate c-Met induced activation of the PI3K-Akt-mTor pathway through matriptase-catalysed HGF activation and binding of the HGF ligand to the c-Met receptor [6;7;12]. Simultaneously ablation of cMet or a corresponding over-expression of matriptase´s cognate inhibitor HAI-1 negated the oncogene potential of matriptase over-expression indicating the impact of unopposed protease activity in malignant transformation [6;7]. In addition, matriptase has been shown in vitro to activate several molecules associated with cancer progression including pro-HGF/SF [53], pro-MSP-1 [55], pro-uPA [51;53;56], PAR-2 [51;57], MMP-3 [58], SIMA135/CDCP1 [60], IGFBP-rP1 [61], VEGFR-2 [62], PDGF [63;64] and EGFR [65;66]. Conversely, a recent study assigns matriptase with a role as tumor suppressor in a murine model of colitis-associated colon cancer. Intestinal-specific epithelial ablation of matriptase resulted in intrinsic intestinal permeability barrier perturbations that progressed into chronic inflammation and subsequently formation of colon adenocarcinoma highlighting the importance of matriptase in epithelial tight junction formation [8]. Multiple studies have also assessed the mRNA and protein levels of matriptase during human carcinogenesis and the potential value of matriptase as a novel prognostic marker, a predictor of patient outcome and as a possible therapeutic target for human cancers. However, the findings from these studies are ambiguous. For some cancers, an increase in matriptase mRNA or protein expression levels has been associated with tumor progression in e.g. breast, cervical, ovarian and prostate cancer [132-139]. In contrast, other studies report a significant downregulation of matriptase expression levels in gastric, colorectal and breast cancer [140-142]. In fact, one of the groups first to discover matriptase did so by identifying genes down-regulated in human colorectal carcinomas, hence matriptase gene annotation; suppressor of tumorigenicity-14 [41;143]. Explanation for these discrepancies could be explained by different roles of matriptase in different epithelia and cancer types and partly by differences in tissue sampling, tissue composition, tumor staging and differences in quantification methodology. Expression studies including both matriptase and HAI-1 may be more informative as HAI-1 has important roles in matriptase- mediated carcinogenesis in mice and for epithelial integrity in general [6;22;128;144]. An imbalance in the matriptase-HAI-1 ratio resulting in a larger proportion of free active matriptase could lead to harmful overactivation of matriptase-mediated pathways contributing to carcinogenesis [145]. This proposal of an imbalance in the matiptase-HAI-1 ratio is supported by studies in gastrointestinal, colorectal, and ovarian cancers where the protease-inhibitor ratio was shifted in favor of matriptase when comparing advanced stage tumors with low-stage lesions or corresponding tissue from control individuals [140;141;145;146].

Technical considerations

This chapter includes methodological considerations for the main methods used for the work presented in three manuscripts; paper I, II and III.

Cell system Matriptase, prostasin and HAI-1 are all expressed in polarized epithelia, yet matriptase and HAI-1 are located to the basolateral plasma membrane at steady state and prostasin locates to the apical plasma membrane at steady state [9;11;24;102]. However, the organization of the epithelium makes it difficult to study the polarity of the cells and the intracellular trafficking of proteins since the basolateral plasma membrane cannot be accessed. To obtain this kind of information it is applicable to utilize model systems of the polarized epithelia. Epithelial cell lines have proven to be useful tools for studying epithelium. We have previously used the Mardin Darby canine kidney (MDCK) cell line as a model system for polarized epithelia to delineate the intracellular trafficking of HAI-1 (supplementary I; [27]). In the present study (paper I and III), we decided to use the well known Caco-2 cell line as a model system for the polarized epithelium. MDCK and Caco-2 cells have both been extensively used as models of the polarized epithelium [147;148]. However combined with the available antibodies, the Caco-2 cell line offers the advantage that we can examine endogenously expressed matriptase, prostasin and HAI-1, which eliminates unwanted effects by recombinant overexpression. Caco-2 cells originate from a human colon carcinoma, yet in culture these cells spontaneously differentiate and form a polarized monolayer of cells that morphologically and functionally resemble the mature enterocytes of the small intestine. For this reason, the Caco-2 cell line has been extensively used as a model of the polarized epithelium [147]. When cultured onto Transwell filters, the apical and basolateral plasma membrane, divided by formation of tight junctions, can be accessed separately, which allows for investigations of intracellular trafficking of proteins (fig. 9). Thus, Caco-2 cells grown on Transwell filters are a good model system for studying the intracellular transport of matriptase, prostasin and their common inhibitor HAI-1. In our studies (paper I and III) we have used 11 days post-confluent Caco-2 cells grown on Transwell filters as the level of HAI-1 complexed matriptase reaches a plateau at this differentiation level (unpublished results, Stine Friis). Tight junction formation and barrier function of Transwell filter grown Caco-2 cells can be assessed in several ways. It can be measured by transepithelial electrical resistance (TEER) also; tracer molecules can be used to measure the permeability of the Caco-2 monolayer, e.g. lucifer yellow or phenol red [149;150]. However, a much simpler way of evaluating barrier function of cells grown on Transwell filters is to assess whether the monolayer of cells is able to maintain a difference in medium level between the inner and outer chamber over night [27].

Fig. 9. Monolayers of polarized Caco-2 cells on porous filters. Caco-2 cells spontaneously form a polarized monolayer of cells when grown on Transwell filters. At confluence, the cells form tight junctions that separate the apical and basolateral extracellular spaces. This setting allows for separate access to the apical and basolateral plasma membrane domains and media. The figure shows a cross section of the Transwell filter insert in the culture dish.

Although cell lines are extremely useful for obtaining new knowledge about molecular mechanism responsible for different phenomena, data from one cell line is not always comparable with data from other cell line nor can one expect data to be transferable to whole organism settings. One reason for this is the immortalized feature of most cell lines. However, cell lines make good model systems to study important biological pathways.

Protease pull down assays Variations over protease inhibitor mediated pull down were applied in paper I and II. These techniques were used to identify activation state and to assay for the presence of active protease. Different set ups were used in the two studies but they rely on the same concept.

Co-immunoprecipitation (co-IP) is a well established method for small scale purification of antigens and their interaction partners and allows for identification of activation status of proteins. In a co-IP, the target antigen has the function of bait and is used to co-precipitate a binding partner/protein complex (prey) from a crude cell lysate or tissue extract. An antibody against the bait is incubated with a lysate either as pre-immobilized onto an insoluble support or as a free un-bound antibody in combination with an insoluble support e.g. protein G sepharose as outlined in fig. 10. The main disadvantage of free antibody protocol for immunoprecipitation and co-IP is that the conditions used to elute the precipitated antigen also release the antibody. Depending on the size of the antigen, the heavy and light chains of the antibody can completely 30 mask the detection of the antigen/interaction partners in Western blot analysis. However this is circumvented by cross-linking the antibody to the resin.

Fig. 10. Immunoprecipitation A suitable antibody is added to a cell lysate or a tissue extract either pre-immobilized onto an insoluble support or as a free unbound antibody to bind the protein of interest. Gentle incubation allows the antigen to by immobilized and purified by means of the antibody and the insoluble support. In the pre-immobilized antibody approach the antibody has been cross-linked to the insoluble support. Immunoprecipitation of intact protein complexes is known as co-IP. Immunoprecipitated proteins and their binding partner(s) can be detected by SDS-PAGE and Western blot analysis. Figure is from [151].

We applied co-IP to determine the presence activated matriptase and activated prostasin in extracts of murine placentas in paper II, as simple detection of the activated proteases in lysates of placentas was unsuccessful by direct Western blotting. We used a free antibody protocol (the latter approach) to co-IP matriptase and prostasin with a HAI-1 antibody and used protein G sepharose that binds to the Fc region of immunoglobulin G as the insoluble support. As unspecific binding to the protease G sepharose can occur, the lysates were pre-incubated with protein G sepharose alone to minimize unspecific binding in the actual co-IP reaction. Placentas from matriptase deficient and prostasin deficient embryos were used as negative controls, respectively. Using co-IP it is often anticipated that associated proteins (prey) is related to the function of bait protein. However, this is merely an assumption that needs further verification. Though, in this case the relationship between matriptase, prostasin and HAI-1 is already well-established [16;21]. Thus, we can use the technique to identify the presence matriptase and prostasin activation in placental extracts from embryos of different genotype.

In paper I, we use a derivation of immunoprecipitation to detect the presence of active matriptase and prostasin. Instead of using antibodies to immobilize matriptase and prostasin, we applied the general serine protease inhibitors aprotenin and leupeptin assuming that only active serine proteases and not their zymogen counterparts will be able to bind to these serine protease inhibitors. As HAI-1 forms reversible inhibitor complexes with proteases, immobilized aprotenin and leupeptin could potentially compete with HAI-1 to bind matriptase and prostasin. This can be ruled out by a complicated experimental setting where the inhibitor-coupled pull down is performed on a sample in which matriptase-HAI-1 complex formation is induced. The major disadvantage of this method is that detection of active matriptase occurs in solution which may deviate from live cell culture settings.

Biotinylation assays The highly specific interaction of biotin with avidin/streptavidin is a useful tool in designing nonradioactive purification and detection systems. The extraordinary affinity of avidin /streptavidin for biotin is the strongest known noncovalent interaction of a protein to a ligand and allows biotin-containing molecules in a complex mixture to be isolated by exploiting this highly stable interaction. For paper I and III included in this thesis, we have exploited the advances of the biotin-avidin interaction to study matriptase, prostasin and HAI-1. In paper I, we have made use of biotinylation assays for studying the subcellular trafficking of matriptase, prostasin and HAI-1 in Transwell filter grown polarized Caco-2 cells. In paper III, we have exploited the streptavidin/avidin–biotin interaction to extract proteases with a peptidolytic activity towards a biotinylated chloromethyl ketone peptide inhibitor designed with a preferred substrate sequence of matriptase.

Fig. 11. Reaction of S-NHS-SS-biotin with primary amine on proteins. Primary amines of proteins react with NHS-esters by nucleophilic attack, whereby the N- hydroxysulfosuccinimide (NHS) is released as a byproduct. The NHS ester group on this reagent reacts with the ε-amine of lysine residues and forms a stable product. Hydrolysis of the NHS-ester competes with the reaction in aqueous solution and increases with increasing pH. α-amine groups present on the N-termini of peptides also react with NHS esters but these α-amines are seldom accessible for conjugation in proteins. Figure is from [152].

Biotin is a relatively small molecule and can thus be conjugated to many proteins without significant affect on the target proteins biological activities. More biotin molecules can bind to a single protein which greatly increases the sensitivity of many assay procedures. Additionally, biotin has been modified in numerous ways to accommodate particular applications. In paper I, we have utilized the membrane impermeable biotin derivative sulfosuccinimidyl-2- [biotinamido]ethyl-1,3-dithiopropionate (S-NHS-SS-biotin) because this conjugate is cell membrane impermeable due to the charged sulfonate group and is thus favorable for cell surface biotinylation. The thiol-cleavable spacer arm of S-NHS-SS-biotin harbors two important properties. First, the extended spacer arm reduces steric hindrance associated with avidin/streptavidin binding and thereby enhances the interaction. Second, the S-S cleavable spacer arm allow for reversible biotinylation, by treatment with reducing agents, which is a required feature when performing transport studies of cell surface proteins. Specifically, we use this feature to determine the endocytosis and transcytosis properties of matriptase, prostain and HAI-1. Finally, S-NHS-SS-biotin reacts with primary amines readily available on the cell surface of proteins (fig. 11).

Fig. 12. Assay for detection of active matriptase in cell culture To establish an assay for the detection of active matriptase, we developed a synthetic peptide with binding preference for the active cleft in matriptase. This peptide is coupled to both a biotin-group and to a chloromethyl ketone (Cmk) group. When the Cmk group is in close proximity to proteases, covalent bond formation occurs by alkylation of the active site histidine. Active proteases with binding preferences for the Cmk inhibitory peptide is labeled and extracted by streptavidin pull down of the biotin group. Specificity for matriptase is obtained via Western blot analysis and matriptase specific antibodies.

For paper III, we make use of the N-terminal biotin moiety of the CMK peptide inhibitor, biotin- RQRR-Cmk, as a tag to extract peptidolytic active matriptase from Caco-2 cell lysates. The peptide sequence of the tetra peptide was designed to obtain the highest specificity towards matriptase and was deduced from a preferred substrate sequence of matriptase [50]. This peptide enables us to biotin-label peptidolytic active matriptase in live and intact Caco-2 cells (fig. 12), as opposed to the inhibitor-coupled sepharose used in paper I that label active matriptase in solution. Once biotin is attached to its target molecule, the molecule can be immobilized using biotin binding molecules. In this study we have made use of streptavidin (paper I and III) and monomeric avidin (paper I). The high affinity biotin-streptavidin binding allows for extraction of biotin from very complex mixtures, and dissociation between biotin and streptavidin requires harsh denaturing conditions e.g. boiling in SDS-PAGE sample buffer. In contrast, monomeric avidin binds biotin in a reversible manner, and gentle recovery of biotinylated molecules is feasible by elution with free excess biotin. Monomeric avidin can facilitate purification of functional proteins and in paper I, we employ this in recovery of the SDS-resistant matriptase-HAI-1 complex in Western blot analysis.

Results

Data included in this PhD thesis is presented in the following three manuscripts:

Manuscript I Transport via the transcytotic pathway makes prostasin available as substrate for matriptase

Manuscript II Reduced prostasin (CAP1/PRSS8) activity eliminates HAI-1 and HAI-2 deficiency associated developmental defects by preventing matriptase activation

Manuscript III Novel assay for detection of active matriptase

Paper I

Transport via the transcytotic pathway makes prostasin available as substrate for matriptase

Stine Friis1, Sine Godiksen2, Jette Bornholdt1, Joanna Selzer‐Plon1, Hanne Borger Rasmussen3, Thomas H. Bugge4, Chen‐Yong Lin5 and Lotte K. Vogel1

1Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark. 2Department of Biology, University of Copenhagen, Copenhagen, Denmark. 3Department of Biomedical Science, University of Copenhagen, Copenhagen, Denmark 4Proteases and Tissue Remodeling Unit, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, USA 5Department of Biochemistry and Molecular Biology, Greenebaum Cancer Centre, University of Maryland, Baltimore, USA

Published in Journal of Biological Chemistry, February 2011.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 286, NO. 7, pp. 5793–5802, February 18, 2011 Printed in the U.S.A.

Transport via the Transcytotic Pathway Makes Prostasin Available as a Substrate for Matriptase * Received for publication, September 29, 2010, and in revised form, December 1, 2010 Published, JBC Papers in Press, December 10, 2010, DOI 10.1074/jbc.M110.186874 Stine Friis‡, Sine Godiksen §, Jette Bornholdt ‡, Joanna Selzer-Plon ‡, Hanne Borger Rasmussen ¶, Thomas H. Bugge ʈ, Chen-Yong Lin** , and Lotte K. Vogel ‡1 From the Departments of ‡Cellular and Molecular Medicine, §Biology, and ¶Biomedical Science, University of Copenhagen, 2200 Copenhagen, Denmark and the ʈProteases and Tissue Remodeling Unit, NIDCR, National Institutes of Health, Bethesda, Maryland 20892, and the ** Department of Biochemistry and Molecular Biology, Greenebaum Cancer Center, University of Maryland, Baltimore, Maryland 21201 Downloaded from

The matriptase-prostasin proteolytic cascade is essential for organ dysfunction associated with increased permeability and epidermal tight junction formation and terminal epidermal loss of tight junctions (1). Knock down of matriptase by differentiation. This proteolytic pathway may also be operative siRNA in a cell model of the intestinal epithelium caused a in a variety of other epithelia, as both matriptase and prostasin leaky barrier, impaired ability to develop transepithelial elec- 2 are involved in tight junction formation in epithelial monolay- trical resistance (TEER) and enhanced paracellular perme- www.jbc.org ers. However, in polarized epithelial cells matriptase is mainly ability through regulation of tight junction proteins (2). To- located on the basolateral plasma membrane whereas prosta- gether these data suggest a key role for matriptase in epithelial sin is mainly located on the apical plasma membrane. To de- barrier function and tight junction assembly. termine how matriptase and prostasin interact, we mapped the Prostasin (also known as CAP1 and PRSS8) is a GPI-an- ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 subcellular itinerary of matriptase and prostasin in polarized chored trypsin-like serine protease. Prostasin is co-expressed colonic epithelial cells. We show that zymogen matriptase is with matriptase in most epithelial tissues including the epi- activated on the basolateral plasma membrane where it is able dermis, kidney, and colon (3). Prostasin proteolytic activity to cleave relevant substrates. After activation, matriptase has also been suggested to promote the development of func- forms a complex with the cognate matriptase inhibitor, hepa- tional tight junctions, TEER and paracellular permeability tocyte growth factor activator inhibitor (HAI)-1 and is effi- (4–6). Unlike matriptase, which undergoes efficient auto- ciently endocytosed. The majority of prostasin is located on activation, the prostasin zymogen is not able to auto-activate, the apical plasma membrane albeit a minor fraction of prosta- and formation of active prostasin requires activation site sin is present on the basolateral plasma membrane. Basolateral cleavage by other trypsin-like serine proteases. Strong data prostasin is endocytosed and transcytosed to the apical plasma suggest that matriptase acts upstream of prostasin in a zymo- membrane where a long retention time causes an accumula- gen cascade in the epidermis. The severe epidermal defects of tion of prostasin. Furthermore, we show that prostasin on the matriptase deficiency appear to be a consequence of lack of basolateral membrane is activated before it is transcytosed. active prostasin. Only the inactive form of prostasin is found This study shows that matriptase and prostasin co-localize for in matriptase-deficient mice and matriptase-deficient and a brief period of time at the basolateral plasma membrane af- prostasin-deficient mice have nearly identical phenotypes ter which prostasin is transported to the apical membrane as with compromised epidermal tight junction formation and no an active protease. This study suggests a possible explanation terminal epidermal differentiation (6–11). Furthermore, it for how matriptase or other basolateral serine proteases acti- has been shown in vitro that the serine protease domain of vate prostasin on its way to its apical destination. matriptase is directly able to cleave the zymogen-form of

prostasin, to generate proteolytically active prostasin (11). 2011 , Matriptase-dependent activation of prostasin was recently The trypsin-like membrane serine protease matriptase is demonstrated in a human organotypic skin model and in essential for maintenance of multiple types of epithelia. Con- matriptase-deficient human epidermis (6, 12). ditional ablation of the St14 gene coding for matriptase in The plasma membrane of a polarized epithelial cell is di- intestine, kidney, and lung of adult mice results in weight loss, vided into an apical and a basolateral plasma membrane do- severe decline in health and death within 2 weeks, caused by main separated by tight junctions. The tight junctions prevent diffusion of membrane proteins between the two membrane domains. In polarized epithelial cells there are two pathways * This work was supported, in whole or in part, by the NIDCR Intramural Research Program and by National Institutes of Health Grant R01-CA- for newly synthesized proteins to reach the apical plasma 123223. This work was also supported by The Harboe Foundation, The membrane: The direct pathway from the trans Golgi network Augustinus Foundation, The Brothers Hartmanns Foundation, The A.P. directly to the apical plasma membrane and the indirect path- Møllers Foundation for the Advancement of Medical Science, The Cluster of Cell Biology at the University of Copenhagen, and the Lundbeck Foundation. 1 To whom correspondence should be addressed: Blegdamsvej 3, Bldg. 6.4, 2 The abbreviations used are: TEER, transepithelial electrical resistance; 2200 Copenhagen N, Denmark. Tel.: 45-35-32-77-87; Fax: 45-35-36-79-80; HAI-1, hepatocyte growth factor activator inhibitor 1; ENaC, epithelial E-mail: [email protected]. sodium channel.

FEBRUARY 18, 2011•VOLUME 286•NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5793 Matriptase Activates Prostasin on the Basolateral Plasma Membrane way, where newly synthesized proteins via the basolateral cells were washed twice with ice-cold PBSϩ. Residual biotin plasma membrane are endocytosed and transcytosed to the was quenched for 5 min at 4 °C with 50 m M glycine in PBS ϩ apical plasma membrane (13). It has been reported that and the cells were washed again with PBSϩ. For internaliza- matriptase is mainly located at the basolateral plasma mem- tion experiments, preheated media (serum-free MEM me- brane in rat enterocytes and other polarized epithelial cells dium containing 20 m M NaHCO3, 2 m M L -glutamine, 100 (14, 15). However, proteolytically shed matriptase in complex units/ml penicillin, and 100 ␮g/ml streptomycin (Invitrogen)) with HAI-1 was purified from human milk suggesting an api- was added and the cells were incubated at 37 °C to regain nor- cal secretion (16). Conversely, the matriptase substrate, pros- mal trafficking. For endocytosis and transcytosis experiments, tasin, is mainly located at the apical plasma membrane of po- surface biotin was removed after incubation as indicated us- larized epithelial cells (17, 18). ing the non-membrane permeable reducing agent glutathione The present study aims to determine where in the polarized (Sigma Aldrich). Surface exposed biotin was removed from epithelial cell active matriptase interacts with its substrate either apical or basolateral side with 16 mg/ml glutathione in

prostasin, in order to explain how a basolateral protease can 75 m M NaCl, 75 m M NaOH, 1 m M EDTA, and 0.5% BSA in Downloaded from ϫ cleave and activate an apicallylocated substrate. Matriptase is H2O under gentle agitation for 2 20 min. Cells were washed synthesized as an inactive, single-chain zymogen. Its activa- with PBSϩ and residual glutathione was inactivated with 5 tion requires two sequential endoproteolytic cleavages. The mg/ml iodoacetamide (Sigma Aldrich) in PBS ϩ for 5 min. A first proteolytic processing cleavage occurs after Gly-149, yet set of parallel samples was left without the glutathione reduc- the processed form remains tightly associated with the tion to monitor the total amount of biotinylated protein pres- membrane. ent through the course of the experiment. Cells were washed www.jbc.org Matriptase is, subsequently (and dependent on the first twice in PBS ϩ, and lysed in PBS containing 1% Triton X-100, cleavage) cleaved after Arg-614 in the serine protease domain 0.5% deoxycholate and protease inhibitors (10 mg/liter benza- to gain proteolytic activity. Shortly after activation, matriptase midine, 2 mg/liter pepstatin A, 2 mg/liter leupeptin, 2 mg/l at DEF ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 forms a complex with HAI-1, whereby matriptase is enzymat- antipain, and 2 mg/liter chymostatin). For inhibitor-Sepha- ically inhibited. Hence, substrates should be present at the rose pull-downs, protease inhibitors were omitted from the same location as matriptase activation takes place. lysis buffer. We present biochemical data showing that a matriptase- Monomeric Avidin/Streptavidin Precipitation of Biotiny- prostasin zymogen cascade is indeed possible in polarized lated Proteins—Lysates from biotin-labeled cells were centri- epithelial cells. We show that matriptase is cleaved to its ac- fuged at 20,000 ϫ g for 20 min to pellet the insoluble material. tive form on the basolateral plasma membrane, subsequently The supernatant was transferred to clean eppendorf tubes inhibited by HAI-1 followed by endocytosis. Importantly, we with either Pierce௡ monomeric avidin agarose (Pierce) (120 find prostasin present on the basolateral plasma membrane ␮l/24 mm filter) or Pierce ௡ streptavidin agarose (50 ␮l/24 during matriptase activation. Furthermore, the basolateral mm filter), prepared as described by manufacturer. After prostasin is active and transcytosed to the apical plasma overnight incubation at 4 °C with end-over-end rotation, the membrane where it accumulates. These results demonstrates agarose was washed three times with 50 m M Tris-HCl, pH that matriptase and prostasin may functionally interact in 6.25. Biotinylated proteins were eluted from the monomeric polarized epithelial cells, despite their, respective, basolateral avidin agarose with 4 m M biotin (Pierce) in PBS for 30 min and apical locations at steady state. followed by addition of SDS sample buffer. For elution from streptavidin-agarose, the samples were boiled in SDS sample EXPERIMENTAL PROCEDURES buffer. Cell Culture—Caco-2 cells were grown in minimal essential Western Blot—The 2ϫ SDS sample buffer (125 mM Tris- medium supplemented with 2 m M L -glutamine, 10% fetal bo- HCl, 25 m M EDTA, pH 6.8, 4% SDS, 5% glycerol, 0.01% brom- vine serum, 1ϫ nonessential amino acids, 100 units/ml peni- phenol blue) did not contain any reducing agent and the sam- cillin, and 100 ␮g/ml streptomycin (Invitrogen) at 37 °C in an ples were not boiled prior to SDS-PAGE to prevent protein 2011 , ϫ 6 atmosphere of 5% CO 2. For all experiments, 2 10 cells complexes from dissociating, unless otherwise specified (0.1 M were seeded into 0.4 ␮m-pore-size 24 mm Transwell ௡ filter dithiothreitol, DTT). All lysates were incubated with the SDS chamber (Corning) allowing separate access to the apical and sample buffer for 10 min at room temperature before gel basolateral plasma membrane. Cells were grown until day 11 loading. The proteins were separated on 7% acrylamide gels postconfluence before they were used for experiments. The made in the laboratory and transferred to Immobilon-P PVDF tightness of filter-grown cells was assayed by filling the inner membranes (Millipore). The membranes were blocked with chamber to the brim and allowing it to equilibrate overnight. 10% nonfat dry milk in PBS containing 0.1% Tween-20 The cell culture medium was changed every day. (PBST) for 1 h at room temperature. The individual PVDF Biotinylation, Internalization, and Biotin-removal—Caco-2 membranes were probed with primary antibodies diluted in cells grown on transwell filters were washed three times with 1% nonfat dry milk in PBST at 4 °C overnight. The next day ϩ ϫ ice-cold PBS (PBS supplemented with 0.7 mM CaCl2 and the membranes were washed 3 with PBST and the binding 0.25 mM MgCl2) on both apical and basolateral side. The cells of primary antibodies was followed by recognition with sec- were biotin-labeled for 30 min at 4 °C, either from the apical ondary horseradish peroxidase (HRP)-conjugated secondary or the basolateral side, with 1 mg/ml EZ-link TM Sulfo-NHS- antibodies (Pierce). After 3ϫ wash with PBST, the signal was SS-Biotin (Pierce) dissolved in PBS ϩ. After biotin labeling the developed using the ECL reagent Super Signal West Femto

5794 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286•NUMBER 7• FEBRUARY 18, 2011 Matriptase Activates Prostasin on the Basolateral Plasma Membrane

Maximum Sensitivity Substrate (Pierce), according to the pro- 4Ј,6-diamidino-2-phenylindole (DAPI) staining. The cells tocol supplied by the manufacturer and visualized with a Fuji were finally mounted with Prolong Gold mounting medium LAS1000-camera (Fujifilm, Sweden AB). For graphs, the free (Invitrogen) and subjected to laser scanning confocal micros- online software ImageJ (created by Wayne Rasband, NIH, copy using the Leica TCS SP2 system and Zeiss LS700 system. Bethesda) was used to quantify the bands on the blot. Values in the graph represent the sum of all the bands visualized on RESULTS the gel. Furthermore, the graph represents the mean of three Steady-state Distribution of Matriptase, HAI-1, and independent experiments and is presented with the standard Prostasin—Where in the cell matriptase cleaves its substrates deviation. is not well understood, however it is fundamental to under- Antibodies—The antibodies used were monoclonal mouse standing how matriptase maintains epithelial integrity. Traf- anti-human antibodies M32 and M69 (19). The antibody M32 ficking and activation of endogenous matriptase, prostasin, detects all forms of matriptase, including zymogen, activated and HAI-1 was therefore studied in Caco-2 cells-a human form, and complexes. Under the conditions used in this study, Downloaded from colon epithelial cell line. Upon reaching confluence, Caco-2 the antibody M69 only detected the matriptase-HAI-1 com- cells spontaneously differentiate into a tight monolayer of plex and M69 reactive material is hereafter referred to as the polarized cells with an apical and a basolateral plasma mem- matriptase-HAI-1 complex. The other antibodies used were brane, separated by tight junctions. Matriptase expression polyclonal rabbit anti-human matriptase raised against the increases during Caco-2 differentiation (2) consistent with the serine protease domain of matriptase (Cat. no. IM1014, Cal-

higher levels of matriptase at the intestinal villous tip (20). www.jbc.org biochem), mouse anti-human HAI-1 antibody M19 (19), and The amount of matriptase-HAI-1 complex also increases dur- mouse anti-human prostasin antibody (Cat. no. 612173, BD ing differentiation and reaches a plateau around day 7–10 in Transduction Laboratories). M32, M69, and polyclonal goat differentiating Caco-2 cells (data not shown). For that reason, anti-human HAI-1 (cat. no. AF1048, R&D) antibodies were Caco-2 cells at day 11 post-confluence were used for the fol- ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 used for immunocytochemistry. Protease Pull-down with Protease Inhibitor-coupled Sepha- lowing experiments. rose 4B —The protease inhibitors aprotinin (5 mg/ml Sepha- We first investigated the steady state distribution of rose), leupeptin (5 mg/ml Sepharose), and soybean trypsin matriptase, HAI-1, and prostasin between the apical and the inhibitor (5 mg/ml Sepharose) were immobilized to CNBr- basolateral plasma membrane domains. Surface biotinylation activated SepharoseTM 4B (GE Healthcare), as specified by the experiments showed that the majority of matriptase was pres- manufacturer’s instructions. Caco-2 cells were biotin-labeled ent among basolateral membrane proteins and most prevalent and biotinylated proteins were pulled down with monomeric in a 70 kDa form corresponding to the extracellular domain of avidin agarose. The biotinylated proteins were gently eluted matriptase cleaved at Gly-149 (Fig. 1, lanes 1 and 2), repre- with 4 m M biotin in 50 m M Tris-HCl, pH 8.5. The biotin-elu- senting either zymogen matriptase or enzymatically active ate was separated from the avidin-agarose and incubated with matriptase. A small fraction of the basolateral matriptase was 60 ␮l protease inhibitor-coupled Sepharose in 50 m M Tris- detected in a 120 kDa form (Fig. 1, lane 2). The 120 kDa form HCl, pH 8.75 at 37 °C for 30 min. The inhibitor Sepharose was was identified as matriptase-HAI-1 complex by detection with washed 3ϫ with 50 m M Tris-HCl, pH 6.5. Proteases were both the M69 and anti-HAI-1 antibodies (Fig. 1, lanes 4 and eluted from the inhibitor-coupled Sepharose using 0.1 M gly- 6). Two matriptase forms below the 120 kDa complex were cine, pH 2.4. Samples were neutralized with 1 M Tris immedi- also detected on the basolateral plasma membrane (Fig. 1, ately after elution. The eluates were added to SDS sample lanes 2 and 4). This could possibly be degradation products of buffer and analyzed by Western blotting. the matriptase-HAI-1 complex or activated matriptase in Gelatin Zymography—Biotinylated monomeric avidin aga- complex with other inhibitors (21). rose-purified proteins were separated on a 7% SDS-polyacryl- Full-length HAI-1 has an estimated molecular weight of 55 amide gel containing 0.1% gelatin. The proteins were re-na- kDa and was detected on both the apical and the basolateral 2011 , tured by washing the gelatin gel 2 ϫ 30 min in 2.5% Triton plasma membrane (Fig. 1, lanes 5 and 6). Apically located HAI-1 displayed a higher mobility than basolaterally located X-100 in H 2O. To wash out excess Triton X-100, the gel was ϫ lanes 5 6 washed 2 10 min in H 2O and thereafter incubated in 50 m M HAI-1 in the presence of SDS (Fig. 1, and ) but the Tris-HCl, pH 8.75 overnight at 37 °C. Gelatinolytic bands on same mobility when the samples were boiled (Fig. 1, lanes 7 the gel were visualized by Coomassie Brilliant Blue staining. and 8). This could indicate a conformational difference mak- Immunofluorescence—Caco-2 cells grown on filters were ing apical HAI-1 more resistant to denaturation with SDS. fixed for 20 min in 4% paraformaldehyde in PBS (Bie & Bernt- Further studies are needed to elucidate the nature of the two sen) at room temperature. The following was performed at forms. A HAI-1 complex around 85 kDa was also detected on 4 °C. Cells were permeabilized with 0.05% Triton X-100 in the basolateral plasma membrane (Fig. 1, lane 6). An 85 kDa PBS for 20 min. Unspecific staining was blocked with PBS HAI-1 complex with matriptase as well as prostasin has previ- containing 3% BSA (PBS/BSA) for 30 min. Cells were incu- ously been reported (22, 23). bated with primary antibody diluted in PBS/BSA for 1.5 h, Prostasin was located mainly on the apical plasma mem- washed 3ϫ in PBS followed by incubation with relevant Alexa brane although minor amounts could be detected on the ba- Fluor-conjugated secondary antibodies (Invitrogen) for 1 h. solateral plasma membrane (Fig. 1, lanes 9 and 10 ). These Where indicated the nuclei of the cells were visualized by data are consistent with the existing literature showing the

FEBRUARY 18, 2011•VOLUME 286•NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5795 Matriptase Activates Prostasin on the Basolateral Plasma Membrane

lateral plasma membrane. Furthermore, the experiments show that large amounts of matriptase-HAI-1 complex are detected in intracellular structures. This suggests that matriptase is endocytosed from the plasma membrane and inhibited by HAI-1 during this process. We examined the endocytosis of basolateral plasma membrane bound matriptase and HAI-1 using biotinylation and internalization techniques (see “Experimental Procedures”). The endocytosis experiments showed that matriptase is very efficiently, and almost completely, endocytosed within 60 min from the basolateral membrane (Fig. 3A). Furthermore, a FIGURE 1. Matriptase is located on the basolateral membrane whereas 3-fold increase in the signal of matriptase-HAI-1 complexes prostasin is apically located. Caco-2 cells grown on Transwell filters were was detected within the first 90 min of incubation, showing Downloaded from biotin-labeled from either the apical (A) ( lanes 1, 3, 5, 7, and 9) or basolateral that the 70 kDa matriptase forms a complex with HAI-1 dur- (B) side ( lanes 2, 4, 6, 8, and 10 ). Biotinylated proteins were precipitated with monomeric avidin, separated by SDS-PAGE and analyzed by Western blot ing the endocytosis (Fig. 3B). Several matriptase-HAI-1 com- with antibodies against total matriptase (M32) (lanes 1 and 2), matriptase- plexes between 95 and 120 kDa were generated from the 70 HAI-1 complex (M69) (lanes 3 and 4), the inhibitor HAI-1 ( lanes 5–8 ) and kDa form during the endocytosis, as detected by both the prostasin (lanes 9 and 10 ) Ϫ/ϩ boiling of samples. Samples analyzed with the anti-prostasin antibody were boiled and reduced with DTT. The posi- matriptase and HAI-1 antibodies (Fig. 3, A–C). After 90 min, tions of molecular weight markers are indicated on the left. Protein and all of the surface-labeled matriptase was found in complex www.jbc.org complex are marked with arrows and size on the right. The majority of the plasma membrane-bound matriptase was located on the basolateral mem- with HAI-1 and was located exclusively intracellularly brane in a 70 kDa form, as detected with the antibody M32 ( lane 2). A small (Fig. 3B). fraction of the basolateral plasma membrane-bound matriptase was found The 55 kDa HAI-1 had an endocytosis pattern different in a 120 kDa form ( lane 2). This form was also detected by the M69 ( lane 4) ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 and anti-HAI-1 (lane 6) antibodies. HAI-1 was found both on apical and ba- from HAI-1 in complex with matriptase. The 55 kDa HAI-1 solateral plasma membrane (lanes 5–8 ). Apical HAI-1 ( lane 5) migrated was only partially endocytosed, with the largest intracellular faster on the SDS-PAGE than the basolateral HAI-1 ( lane 6) when samples were not boiled but displayed the same size after boiling of the samples pool seen after 15 min. This suggests that HAI-1, when not in (lanes 7 and 8). Plasma membrane-bound prostasin was located mainly on complex with matriptase, is recycling to the plasma mem- the apical side (lane 9), although minor amounts were detected on the ba- brane from early endosomes (Fig. 3C). This type of recycling solateral membrane (lane 10 ). Results shown are representative of five independent experiments. of HAI-1 has been shown previously in MDCK cells (24). A HAI-1 complex around 85 kDa was observed in the Total basolateral localization of matriptase and apical localization of panel from time point 0 to 15 min, which was not detected by prostasin in a polarized cell (15, 18). the matriptase antibodies (Fig. 3, compare C to A and B, re- Next, we investigated the subcellular steady state distribu- spectively). This could possibly be a HAI-1-prostasin com- tion of matriptase and HAI-1 in Caco-2 cells by immunocyto- plex. A similar complex has previously been reported in kera- chemistry (Fig. 2). Caco-2 cells were grown on Transwell fil- tinocytes (22). Together these data show that matriptase is ters before fixation, permeabilization, and immunolabeling. efficiently endocytosed from the basolateral side and forms a Matriptase and HAI-1 were both detected on the basolateral complex with HAI-1 during this process. plasma membrane (Fig. 2, A–F). Interestingly, matriptase and Matriptase Zymogen Is Cleaved to the Active Protease on the HAI-1 were also detected in structures near the apical plasma Surface of the Basolateral Plasma Membrane —We have up to membrane (Fig. 2F). Only weak detection of matriptase- now showed that matriptase is present predominantly in a 70 HAI-1 complex was observed on the basolateral plasma mem- kDa form on the basolateral plasma membrane and is effi- brane (Fig. 2D). Surprisingly the majority of the matriptase- ciently endocytosed forming a complex with HAI-1 that accu- HAI-1 complex was detected in structures near the apical mulates in intracellular structures, making it a prerequisite plasma membrane (Fig. 2, D–F). To further investigate the that matriptase activation occurs prior to endocytosis. 2011 , location of the matriptase-HAI-1 complex in the apical region Whether matriptase activation occurs before arrival or at the of the cells, Caco-2 cells were immunolabeled with and with- plasma membrane is not well defined. out permeabilization of the plasma membrane (Fig. 2, G and We examined the activation cleavage of basolateral plasma H). Without permeabilization, almost no matriptase-HAI-1 membrane bound matriptase using surface biotinylation and complex could be detected (Fig. 2 H), however, with permeabi- an incubation assay. Under reducing conditions, matriptase lization a distinct labeling of the vesicular structures was zymogen is a 70 kDa protein while enzymatically-active clearly visible below the apical plasma membrane (Fig. 2G), matriptase separates into two fragments; the stem domain confirming an intracellular localization of the structures con- and the 30 kDa protease domain. The antibody IM1014 reacts taining the matriptase-HAI-1 complex. These data are con- with the serine protease domain of both zymogen and acti- sistent with our biotinylation experiments, as we were only vated matriptase under reducing conditions. This experiment able to label matriptase on the basolateral plasma membrane showed that minor amounts of the 70 kDa matriptase zymo- (Fig. 1, lanes 1 and 3). gen were present on the basolateral plasma membrane to- The Matriptase-HAI-1 Complex Is Generated during gether with 30 kDa cleaved matriptase serine protease domain Endocytosis—Our steady state experiments show that a large (Fig. 4, lane 1). The basolateral matriptase zymogen was rap- portion of matriptase is present in a 70 kDa form on the baso- idly cleaved as only the 30 kDa band could be detected after

5796 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286•NUMBER 7• FEBRUARY 18, 2011 Matriptase Activates Prostasin on the Basolateral Plasma Membrane Downloaded from www.jbc.org at DEF ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15

FIGURE 2. Matriptase-HAI-1 complex accumulates in intracellular structures. Caco-2 cells grown on Transwell filters, were fixed, permeabilized, and immunolabeled with antibodies against total matriptase (M32) and HAI-1 (A–C) or matriptase-HAI-1 complex (M69) and HAI-1 ( D–F). Using a confocal scanning microscope, images were taken in the XY plane showing a single section through the monolayer and the XZ plane showing a cross section of the monolayer. The position of the plane of the XY section is indicated on the XZ plane ( black arrows) on the right. The scale bar represents 10 ␮m. A–C, Matriptase was detected on the basolateral plasma membrane of the Caco-2 cells co-localizing with HAI-1. Matriptase and HAI-1 was also co-localizing in structures near the apical membrane (white arrowheads). D–F, matriptase-HAI-1 complex was observed in structures near the apical plasma membrane ( white arrow-

heads). Low amounts of matriptase-HAI-1 complex were detected on the basolateral plasma membrane. HAI-1 was detected both on apical and basolateral 2011 , plasma membranes as well as in the apical structures co-localizing with matriptase. G and H, Caco-2 cells were treated with or without Triton X-100 prior to immunolabeling with the antibody M69. G, in the cells permeabilized with Triton X-100, a distinct detection of matriptase-HAI-1 complex was observed in apical vesicular structures. H, in cells without permeabilization, the vesicular structures with matriptase-HAI-1 complex were not detected. Only a weak basolateral signal was observed. The nuclei were visualized by DAPI staining shown in blue . Results shown are representative of three independent experiments. just 5 min of incubation (Fig. 4, lane 2). A slight increase in HAI-1. This window of action is assumed to be in between the the 30 kDa band intensity was observed up to 15 min after proteolytic cleavage creating the fully active protease and the labeling, most likely caused by a higher affinity of IM1014 rapid complex formation with its inhibitor HAI-1. Our previ- antibody for the cleaved 30 kDa protease domain than the 70 ous experiments suggest that both matriptase activation and kDa form. These data suggest that matriptase is fast and effi- inhibition takes place on the basolateral membrane and we ciently cleaved to its active form on the basolateral cell would therefore expect to find free active matriptase on the surface. basolateral plasma membrane. To address this, we investi- Active Matriptase Is Present on the Basolateral Plasma gated whether matriptase at the basolateral plasma membrane Membrane—It has previously been reported that the period of was able to bind to the general serine protease inhibitors, time for active matriptase to act on its substrates is very lim- aprotinin and leupeptin, to which we expect only the active ited as matriptase activation is tightly coupled to inhibition by protease to bind. We were able to purify matriptase from the

FEBRUARY 18, 2011•VOLUME 286•NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5797 Matriptase Activates Prostasin on the Basolateral Plasma Membrane Downloaded from www.jbc.org at DEF ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15

FIGURE 3. Matriptase in complex with HAI-1 is generated during the efficient endocytosis of matriptase from the basolateral plasma membrane. Caco-2 cells were grown on Transwell filters, and proteins on the basolateral plasma membrane were biotinylated at 4 °C using a cleavable biotinylation reagent, s-NHS-SS-biotin. The labeled cells were incubated at 37 °C for the time indicated (0–120 min). After incubation, the proteins remaining on the plasma membrane were biotin-stripped, using the membrane non-permeant reducing agent glutathione, leaving only endocytosed proteins biotinylated (endocytosis panels). Biotin reduction was omitted in a parallel set of samples to monitor the degradation of the biotinylated proteins over time (total panels). Biotinylated proteins were precipitated with monomeric avidin agarose, and the avidin pull-downs were analyzed with SDS-PAGE and Western blotting using the antibodies (A) M32 against total matriptase, ( B) M69 against matriptase-HAI-1 complex, and ( C) M19 against HAI-1. The positions of molecular markers (kDa) are indicated on the left. Proteins and complexes are indicated with arrows and size. Quantification of bands from Western blots was done using the software ImageJ. A graphic presentation of three independent endocytosis experiments is shown on the right hand side. The dotted line equals total protein and solid line equals endocytosed protein. The standard deviation is shown with error bars. A, matriptase was endocytosed from the basolateral plasma membrane within 60 min. Approximately 80% of matriptase was endocytosed at 60 min as indicated on the graph. B, during the 120-min incubation there was a 3-fold increase in the matriptase-HAI-1 complexs. C, free 55 kDa HAI-1 was partially endocytosed within 15 min. ( C, endocytosis panel). Results shown are representative of three independent experiments.

basolateral plasma membrane, with both aprotinin- and leu- matriptase-HAI-1 complex is not dissociated during our cell 2011 , peptin-coupled Sepharose (Fig. 5A, lanes 3 and 5). No extraction procedure. matriptase was purified with the negative control soybean Finally, to test the proteolytic activity of matriptase, the two trypsin inhibitor-Sepharose or uncoupled Sepharose (data not biotinylated fractions, containing 70 kDa matriptase and shown). The inhibitor-bound fraction of matriptase was not matriptase-HAI-1 complexes, respectively, were analyzed for detected by the antibody detecting matriptase-HAI-1 com- their ability to display gelatinolytic activity at pH 8.75, where plexes (Fig. 5B, lanes 3 and 5). matriptase has optimal enzymatic activity (25) (Fig. 5C). The We wanted to verify that the purified matriptase detected fraction with the 70 kDa matriptase displayed one gelatino- using the inhibitor-coupled Sepharose indeed was free active lytic band around 70 kDa at pH 8.75, corresponding to the matriptase and not activated matriptase dissociated from the size of free active matriptase (Fig. 5 C, lane 1). The fraction matriptase-HAI-1 complexes. To test this, a sample contain- containing mostly matriptase-HAI-1 complex, showed only ing mostly matriptase-HAI-1 complex was exposed to the weak gelatinolytic activity around 70 kDa (Fig. 5 C, lane 2). inhibitor-coupled Sepharose. No detectable matriptase from Thus, the gelatinolytic activity pattern matches the one ob- the matriptase-HAI-1 complex was purified with the inhibitor served for matriptase immunoreactivity with the antibody Sepharose (Fig. 5, lanes 4 and 6). This verifies that it is free M32 (Fig. 5A) implying that the gelatinolytic activity observed active matriptase that is being pulled down and that the is caused by matriptase.

5798 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286•NUMBER 7• FEBRUARY 18, 2011 Matriptase Activates Prostasin on the Basolateral Plasma Membrane

FIGURE 6. Active prostasin is present on the basolateral as well as on the apical plasma membrane. Caco-2 cells on Transwell filters were surface biotinylated from either the apical or basolateral side at 4 °C. Biotiny- lated proteins were precipitated with monomeric avidin and gently eluted with biotin. The avidin pull-downs were divided into three: No further treatment (lanes 1 and 2), pull-down with aprotinin-Sepharose ( lanes 3 and 4) FIGURE 4. Matriptase is cleaved to the two chain form on the basolat- and pull-down with trypsin inhibitor as negative control ( lanes 5 and 6). The eral cell surface. Caco-2 cells were grown on Transwell filters, biotinylated pull-down fractions were analyzed by SDS-PAGE and Western blotting. Downloaded from from the basolateral side at 4 °C and incubated up to 60 min at 37 °C after Prostasin was purified from both apical plasma membrane (lane 1) and ba- labeling. Biotinylated proteins were pulled down with streptavidin-agarose, solateral plasma membrane (lane 2). The majority of the apical prostasin boiled, reduced and analyzed by SDS-PAGE and Western blotting using the could be pulled down with the serine protease inhibitor aprotinin ( lane 3). A antibody IM1014. The positions of molecular markers (kDa) are indicated on small but significant fraction of basolateral prostasin was also able to bind the left. Bands are indicated with arrows and size. At time 0 the matriptase is aprotinin and hence was in its active form ( lane 4). No prostasin binding was detected as a 70 kDa band, representing the non-cleaved zymogen and a observed when using the negative control trypsin-coupled Sepharose

30 kDa band representing the cleaved protease domain ( lane 1). After 5 min (lanes 5 and 6). Results shown are representative of three independent www.jbc.org of incubation the zymogen could no longer be detected and an increase in experiments. the 30 kDa serine protease domain was detected ( lanes 2–6 ). Results shown are representative of two independent experiments. Active Prostasin Is Present on Both the Apical and the Baso- lateral Plasma Membrane—All of our data suggest that at DEF ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 matriptase is active and able to cleave relevant substrates in a short period of time on the basolateral plasma membrane before it forms a complex with HAI-1 and is endocytosed. If prostasin is indeed a substrate for matriptase in polarized epithelial cells, we would expect to find the cleaved form of prostasin on the basolateral membrane. To test this, we utilized the fact that prostasin is a serine protease and only the cleaved form is proteolytically active and thereby able to bind serine protease inhibitors such as aprotinin. Our experiment showed that prostasin on both the apical and the basolateral plasma membrane was able to bind to aprotinin (Fig. 6, lanes 3 and 4). This suggests that prostasin is active on both the apical and the basolateral side of the cell. Matriptase Co-localizes with Prostasin on the Basolateral Plasma Membrane, Subsequently Prostasin Is Transcytosed to the Apical Plasma Membrane—We know from the surface labeling experiment that the majority of membrane-bound prostasin at steady state is located on the apical side (Fig. 1, lane 9) whereas only a minor fraction is located on the basolateral membrane in Caco-2 cells (Fig. 1, lane 10 ). From our experiments, we also know that prostasin is not only found in 2011 , its active form on the apical plasma membrane but also on the FIGURE 5. Matriptase is active on the basolateral plasma membrane basolateral plasma membrane. This could indicate that pros- but not intracellularly. Caco-2 cells on Transwell filters were surface- tasin is transported to the basolateral plasma membrane to biotinylated from the basolateral side at 4 °C. Some were lysed immedi- ately after biotinylation (0) and some were incubated for 2 h at 37 °C to get activated by matriptase. transform all biotin-labeled matriptase into activated matriptase in com- We wanted to investigate the endocytic transport of the plex with HAI-1 (2). Biotinylated proteins were precipitated with monomeric avidin and gently eluted with biotin. The avidin pull-downs were apical and basolateral prostasin. Initially, we tested if the divided into three groups: No further treatment ( lanes 1 and 2), pull- two membrane fractions were endocytosed using the same down with aprotinin-Sepharose ( lanes 3 and 4) and pull-down with leu- experimental setup as for matriptase and HAI-1 endocytosis. peptin-Sepharose ( lanes 5 and 6). The samples were analyzed by SDS- PAGE and Western blot with antibodies against total matriptase (M32) We found that prostasin was not endocytosed from the apical and matriptase-HAI-1 complex (M69). A, only at time 0 was it possible to plasma membrane but was completely endocytosed from the pull-down M32-detectable matriptase with both aprotinin and leupeptin. B, M69-detectable matriptase was pulled down with the monomeric basolateral plasma membrane within 60 min (data not avidin ( B, lanes 1 and 2), but was lost with additional inhibitor pull-down shown). This made us question if prostasin is initially trans- (B, lanes 3– 6 ). C, two avidin-purified fractions showed gelatinolytic prop- ported to the basolateral membrane and activated by erties with a band around 70 kDa ( C, lanes 1 and 2), matching the size and pattern of A, lanes 1 and 2. Results shown are representative of two matriptase before it is re-routed by transcytosis to the apical independent experiments. membrane. A long residence time at the apical plasma mem-

FEBRUARY 18, 2011•VOLUME 286•NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5799 Matriptase Activates Prostasin on the Basolateral Plasma Membrane Downloaded from www.jbc.org at DEF ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15

FIGURE 7. Prostasin and HAI-1 are transcytosed from the basolateral to the apical plasma membrane. Caco-2 cells were surface biotinylated from the basolateral side at 4 °C. The labeled cells were incubated at 37 °C for the time indicated in the internalization panel (4, 12, or 18 h as indicated). After incubation the proteins remaining on the plasma membrane were biotin-stripped, using the membrane non-permeant reducing agent glutathione, from either apical (A red .), basolateral (B red .), or both sides ( A ϩ B red .). Biotin reduction was omitted in a set of samples to monitor the total amount of biotinylated protein remaining after the incubation (Total). The experiment was performed in duplicates. Biotinylated proteins were precipitated with monomeric avidin agarose, separated with SDS-PAGE and analyzed by Western blotting. A, prostasin could be biotin-stripped from the apical side ( A red .) after 18 h of internalization, as a decrease in signal was observed compared with the total samples. This shows that prostasin has moved from the basolateral to the apical side. When surface proteins were biotin-stripped from the basolateral side no decrease in signal was observed ( B red .), once again showing that no biotinylated prostasin was left on the basolateral side after 18 h. To clarify if some of the prostasin was located intracellularly, the cells were biotin stripped from both apical and basolateral side (A ϩ B red .). Biotin-stripping from both apical and basolateral side removed the majority of the biotinylated prostasin suggesting that all of prostasin had within the 18 h transferred from the basolateral to the apical plasma membrane. B, HAI-1 was transcytosed from the basolateral to the apical plasma membrane within 12 h. Most of HAI-1 could be biotin stripped from the apical side after incubation ( A red .) where a major fraction was left after basolateral reduction (B. red ) showing that a large fraction of biotinylated HAI-1 had moved from the basolateral to the apical plasma membrane. It was demonstrated that the faster migrating form of HAI-1 resides at the apical plasma membrane 12 h after biotinylation and the slower migrating form remained at the basolateral plasma membrane, as these could be biotin-stripped from their respective sites (A red. compared with B red., respectively). C, after just 4 h of incubation most biotinylated matriptase was no longer detectable in the cell. The remaining matriptase was mainly found in the 120 kDa complex with HAI-1. A fraction of the matriptase-HAI-1 complex could be biotin stripped from the basolateral side ( B red .) but not the apical side ( A red. ). D, M69 detectable matriptase showed the same amount of matriptase-HAI-1 complex at time 0 as for 4 h of incubation. A fraction of the complex was re- ducible from the basolateral side after 4 h of incubation (B red. ). Results shown are representative of three independent experiments. , 2011 , brane combined with a short residence time at the basolateral for the faster migrating HAI-1 present on the apical plasma plasma membrane would give a steady state accumulation at membrane (Fig. 7B). the apical plasma membrane. We, therefore, tested if the ba- Matriptase-HAI-1 complex has been purified from milk, solaterally endocytosed fraction was transcytosed to the apical suggesting that matriptase is secreted from the apical membrane. plasma membrane (16). However, we were unable to detect Transcytosis for several incubation times was investigated transcytosis of matriptase from the basolateral to the apical (1, 2, 4, 8, 12, and 18 h) The time points most clearly demon- plasma membrane, as matriptase has a shorter half-life in strating transcytosis are shown in Fig. 7. Transcytosis of pros- the cell than HAI-1 and prostasin (Fig. 7, C and D). The tasin was detectable after 8 h (data not shown) and most of remaining biotin-labeled matriptase was found in the prostasin was transcytosed from the basolateral to the apical matriptase-HAI-1 complexes between 95 and 120 kDa after plasma membrane after 18 h (Fig. 7 A). HAI-1 was also tran- 4 h incubation (Fig. 7, C and D). These complexes were scytosed, however, less efficiently than prostasin (Fig. 7B). also detected with the anti-HAI-1 antibody after 4 h incu- Our results strongly suggest that the slow migrating form of bation (data not shown). Instead, basolaterally biotin-la- HAI-1 at the basolateral plasma membrane is the precursor beled matriptase could be detected in the basolateral media

5800 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286•NUMBER 7• FEBRUARY 18, 2011 Matriptase Activates Prostasin on the Basolateral Plasma Membrane

DISCUSSION 5m 15m 30m 1h 2h 4h 8h 18h It has been shown that matriptase and prostasin are consti- A Total matriptase - 120 kD tutively expressed and co-localize in most epithelia including (M32) the tissues affected by matriptase ablation, suggesting a possi- - 70 kD ble global role for a matriptase-prostasin cascade in epithelial homeostasis (26). In a polarized cell, matriptase is at steady- B Matriptase-HAI-1 - 120 kD state concentrated and localized mainly at the basolateral complex (M69) plasma membrane together with HAI-1 (15, 24) while prosta- - 70 kD sin is mainly concentrated and localized to the apical plasma membrane (17, 18). Our data provide a possible solution to C HAI-1 - 120 kD how matriptase can activate prostasin despite their two different subcellular localizations, by showing that the two molecules meet en route. Downloaded from - 50 kD Matriptase is mainly located on the basolateral plasma membrane where it is activated, inhibited, endocytosed in Apical media complex with its inhibitor HAI-1 and is accumulated in intra- D Prostasin cellular structures. Observation of the transmembrane N-

- 40 kD terminal fragment of matriptase in intracellular compart- www.jbc.org ments has previously been described in rat enterocytes (14). We show here that prostasin is mainly located on the apical FIGURE 8. Matriptase is shed into the basolateral media while basolateral prostasin is transcytosed and shed into the apical media. Caco-2 plasma membrane in its active form. Interestingly, we find a cells were surface-biotinylated from the basolateral side at 4 °C. Cells were small fraction of prostasin on the basolateral plasma mem- ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 incubated at 37 °C after labeling and apical and basolateral media was col- lected at 5, 15, 30 min, 1, 2, 4, 8, and 18 h to reveal any shedding of the bi- brane co-localizing with matriptase, making it possible for otin-labeled matriptase, HAI-1 and prostasin. Biotinylated proteins from the protease and substrate to interact and activation to occur. In media were pulled down with monomeric avidin agarose, separated with agreement with this, we show that part of the basolateral SDS-PAGE and analyzed by Western blotting. A, matriptase was released to the basolateral media within 1 h as three forms: a 70 kDa form and two pro- prostasin is active by its ability to bind to general serine prote- teolytically shed complexes at 85 and 110 kDa. B, two complexes of size 85 ase inhibitors. The basolateral prostasin is endocytosed and and 110 kDa could be detected with the M69 antibody within 1–2 h, suggesting these to be matriptase-HAI-1 complexes. C, no free 55 kDa HAI-1 transcytosed to the apical plasma membrane, where it was found in the media, only HAI-1 in the two complexes at 85 and 110 kDa accumulates. was detected in the basolateral media after 1 h and accumulated up to 18 h. Both matriptase and prostasin has been shown to be able to D, basolateral prostasin was detected in the apical media after 18 h incubation, confirming transcytosis before secretion into the apical media. Results activate the epithelial sodium channel (ENaC) in Xenopus shown are representative of two independent experiments. oocytes (5, 27, 28). ENac is an epithelial membrane-bound sodium channel located in the apical membrane of polarized cells and is required for normal epidermal differentiation (29, after only 1 h and matriptase accumulated in the media 30). Because matriptase is targeted to the basolateral mem- over the 18 h (Fig. 8, A and B). Both the 70 kDa form as brane where it is activated and rapidly inhibited it is more well as two forms of the matriptase-HAI-1 complex around likely that prostasin which co-localizes with ENaC on the api- 110 and 85 kDa were detected in the media. Only HAI-1 in cal membrane is a candidate activator in polarized cells. Both complex with matriptase was detected in the basolateral matriptase and prostasin have been coupled to maintenance media (Fig. 8 C), whereas no free HAI-1 could be detected. of functional tight junctions in a variety of epithelial cell types. None of the biotin-labeled matriptase or HAI-1 could be In a Caco-2 model, loss of matriptase was associated with en- detected in the apical media (data not shown). Basolateral hanced expression and incorporation of the pore-forming 2011 , prostasin could not be detected in the basolateral media protein claudin-2 at tight junctions (2). Prostasin has been (data not shown) but after 18 h of incubation after the bio- reported to regulate tight junctions, paracellular permeability tinylation, prostasin appeared in the apical media (Fig. 8 D). and TEER by a protease activity-dependent mechanism in This is consistent with the finding that prostasin is transcy- renal cells (4, 5). This means that the two proteases are both tosed from the basolateral to the apical plasma membrane involved in cascades important for tight junction formation in within this time frame. Thus, we were able to show that polarized epithelia. prostasin on the basolateral plasma membrane is transcy- Increasing evidence suggests that serine proteases contrib- tosed to the apical plasma membrane from where it is shed ute in a complex way to the regulation of intestinal integrity to the media. This suggests that prostasin is routed via the and barrier function. Protease inhibitors have been shown to basolateral plasma membrane where it is activated before it suppress the formation of tight junctions in gastrointestinal is transcytosed to the apical membrane, thereby providing cell lines suggesting that proteolytic activity is necessary for a a mechanism for activation of the prostasin zymogen in functional epithelial barrier (31). The epithelial barrier is cru- polarized epithelial cells by matriptase or other basolater- cial in the gastrointestinal tract and is often compromised in ally located serine proteases, despite the separate steady- inflammatory bowel diseases like Crohn disease and ulcera- state localization of the proteases. tive colitis. The sodium channel ENaC is down-regulated in

FEBRUARY 18, 2011•VOLUME 286•NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5801 Matriptase Activates Prostasin on the Basolateral Plasma Membrane inflammatory bowel diseases and has been proposed as a ther- and Fushiki, T. (2005) Biochem. J. 388, 679–687 apeutic target (32–34). It will be important for future studies 15. Wang, J. K., Lee, M. S., Tseng, I. C., Chou, F. P., Chen, Y. W., Fulton, A., to identify the role of the matriptase-prostasin cascade and Lee, H. S., Chen, C. J., Johnson, M. D., and Lin, C. Y. (2009) Am. J. Physiol. Cell Physiol. 279, the substrates involved in this pathway to investigate the role C459–470 16. Lin, C. Y., Anders, J., Johnson, M., and Dickson, R. B. (1999) J. Biol. of matriptase in inflammatory diseases of the gastrointestinal Chem. 274, 18237–18242 tract. 17. Chen, M., Chen, L. M., Lin, C. Y., and Chai, K. X. (2008) Biochim. Bio- Together, our data show that active matriptase and its sub- phys. Acta 1785, 896–903 strate prostasin co-localize at the basolateral plasma mem- 18. Selzer-Plon, J., Bornholdt, J., Friis, S., Bisgaard, H. C., Lothe, I. M., Tveit, brane and hereby provide a venue for a matriptase-prostasin K. M., Kure, E. H., Vogel, U., and Vogel, L. K. (2009) BMC. Cancer 9, zymogen cascade to be initiated in polarized epithelial cells. 201 Furthermore, we show how prostasin is transported after acti- 19. Lin, C. Y., Wang, J. K., Torri, J., Dou, L., Sang, Q. A., and Dickson, R. B. (1997) J. Biol. Chem. 272, 9147–9152 vation to the apical membrane to co-localize with its primary 20. Satomi, S., Yamasaki, Y., Tsuzuki, S., Hitomi, Y., Iwanaga, T., and substrate ENaC. Fushiki, T. (2001) Biochem. Biophys. Res. Commun. 287, 995–1002 Downloaded from 21. Tseng, I. C., Chou, F. P., Su, S. F., Oberst, M., Madayiputhiya, N., Lee, Acknowledgment—We thank Dr. Mary Jo Danton for critically M. S., Wang, J. K., Sloane, D. E., Johnson, M., and Lin, C. Y. (2008) reading the manuscript. Am. J. Physiol. Cell Physiol. 295, C423–C431 22. Chen, Y. W., Wang, J. K., Chou, F. P., Chen, C. Y., Rorke, E. A., Chen, L. M., Chai, K. X., Eckert, R. L., Johnson, M. D., and Lin, C. Y. (2010)

REFERENCES J. Biol. Chem. 285, 31755–31762 www.jbc.org 1. List, K., Kosa, P., Szabo, R., Bey, A. L., Wang, C. B., Molinolo, A., and 23. Lin, C. Y., Tseng, I. C., Chou, F. P., Su, S. F., Chen, Y. W., Johnson, Bugge, T. H. (2009) Am. J. Pathol. 175, 1453–1463 M. D., and Dickson, R. B. (2008) Front Biosci. 13, 621–635 2. Buzza, M. S., Netzel-Arnett, S., Shea-Donohue, T., Zhao, A., Lin, C. Y., 24. Godiksen, S., Selzer-Plon, J., Pedersen, E. D., Abell, K., Rasmussen, H. B.,

List, K., Szabo, R., Fasano, A., Bugge, T. H., and Antalis, T. M. (2010) Szabo, R., Bugge, T. H., and Vogel, L. K. (2008) Biochem. J. 413, 251–259 ConsortiumDEF at Københavns- Universitetsbibliotek, onJune 15 Proc. Natl. Acad. Sci. U.S.A. 107, 4200–4205 25. Beliveau, F., Desilets, A., and Leduc, R. (2009) FEBS J. 276, 2213–2226 3. Yu, J. X., Chao, L., and Chao, J. (1995) J. Biol. Chem. 270, 13483–13489 26. List, K., Hobson, J. P., Molinolo, A., and Bugge, T. H. (2007) J. Cell. 4. Steensgaard, M., Svenningsen, P., Tinning, A. R., Nielsen, T. D., Jor- Physiol. 213, 237–245 gensen, F., Kjaersgaard, G., Madsen, K., and Jensen, B. L. (2010) Acta 27. Vallet, V., Chraibi, A., Gaeggeler, H. P., Horisberger, J. D., and Rossier, Physiol. 200, 347–349 B. C. (1997) Nature 389, 607–610 5. Verghese, G. M., Gutknecht, M. F., and Caughey, G. H. (2006) Am. J. 28. Andreasen, D., Vuagniaux, G., Fowler-Jaeger, N., Hummler, E., and Physiol. Cell Physiol. 291, C1258–C1270 Rossier, B. C. (2006) J. Am. Soc. Nephrol. 17, 968–976 6. Leyvraz, C., Charles, R. P., Rubera, I., Guitard, M., Rotman, S., Breiden, 29. Myerburg, M. M., Harvey, P. R., Heidrich, E. M., Pilewski, J. M., and But- B., Sandhoff, K., and Hummler, E. (2005) J. Cell Biol. 170, 487–496 terworth, M. (2010) Am. J. Respir. Cell Mol. Biol . 43, 712–719 7. List, K., Szabo, R., Wertz, P. W., Segre, J., Haudenschild, C. C., Kim, 30. Mauro, T., Guitard, M., Behne, M., Oda, Y., Crumrine, D., Komuves, L., S. Y., and Bugge, T. H. (2003) J. Cell Biol. 163, 901–910 Rassner, U., Elias, P. M., and Hummler, E. (2002) J. Invest. Dermatol. 8. Ovaere, P., Lippens, S., Vandenabeele, P., and Declercq, W. (2009) 118, 589–594 Trends Biochem. Sci. 34, 453–463 31. Bacher, A., Griebl, K., Mackamul, S., Mitreiter, R., Mu¨ckter, H., and 9. Szabo, R., Kosa, P., List, K., and Bugge, T. H. (2009) Am. J. Pathol. 179, Ben-Shaul, Y. (1992) Exp. Cell Res. 200, 97–104 2015–2022 10. List, K., Haudenschild, C. C., Szabo, R., Chen, W., Wahl, S. M., Swaim, 32. Farkas, K., Yeruva, S., Rakonczay, Z., Jr., Ludolph, L., Molnar, T., Nagy, W., Engelholm, L. H., Behrendt, N., and Bugge, T. H. (2002) Oncogene F., Szepes, Z., Schnur, A., Wittmann, T., Hubricht, J., Riederer, B., Ven- 21, 3765–3779 glovecz, V., Lazar, G., Kiraly, M., Zsembery, A., Varga, G., Seidler, U., 11. Netzel-Arnett, S., Currie, B. M., Szabo, R., Lin, C. Y., Chen, L. M., Chai, and Hegyi, P. (2010) Inflamm. Bowel. Dis. in press K. X., Antalis, T. M., Bugge, T. H., and List, K. (2006) J. Biol. Chem. 281, 33. Sullivan, S., Alex, P., Dassopoulos, T., Zachos, N. C., Iacobuzio- 32941–32945 Donahue, C., Donowitz, M., Brant, S. R., Cuffari, C., Harris, M. L., Datta, 12. Alef, T., Torres, S., Hausser, I., Metze, D., Tu¨rsen, U., Lestringant, G. G., L. W., Conklin, L., Chen, Y., and Li, X. (2009) Inflamm. Bowel. Dis. 15, and Hennies, H. C. (2009) J. Invest. Dermatol. 129, 862–869 261–274 34. Zeissig, S., Bergann, T., Fromm, A., Bojarski, C., Heller, F., Guenther, U., 13. Le Bivic, A., Quaroni, A., Nichols, B., and Rodriguez-Boulan, E. (1990) 2011 , J. Cell Biol. 111, 1351–1361 Zeitz, M., Fromm, M., and Schulzke, J. D. (2008) Gastroenterology 134, 14. Tsuzuki, S., Murai, N., Miyake, Y., Inouye, K., Hirayasu, H., Iwanaga, T., 1436–1447

5802 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286•NUMBER 7• FEBRUARY 18, 2011 Paper II

Reduced prostasin (CAP1/PRSS8) activity eliminates HAI-1 and HAI-2 deficiency-associated developmental defects by preventing matriptase activation

1Oral and Pharyngeal Cancer Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD, USA. 2Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark. 3Department of Biology, University of Copenhagen, Copenhagen, Denmark. 4Pharmacology and Toxicology Department, University de Lausanne, Lausanne, Switzerland. 5INSERM U970, Paris Cardiovascular Research Centre, Paris, F-75015, France. 6Université Paris- Descartes, Paris, F-75006, France.

Published in PLoS Genetic, August 2012

Reduced Prostasin (CAP1/PRSS8) Activity Eliminates HAI- 1 and HAI-2 Deficiency–Associated Developmental Defects by Preventing Matriptase Activation

Roman Szabo1, Katiuchia Uzzun Sales1, Peter Kosa1, Natalia A. Shylo1, Sine Godiksen1,2,3, Karina K. Hansen1, Stine Friis1, J. Silvio Gutkind1, Lotte K. Vogel2, Edith Hummler4, Eric Camerer5,6, Thomas H. Bugge1* 1 Oral and Pharyngeal Cancer Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, Maryland, United States of America, 2 Department of Cellular and Molecular Medicine, Faculty of Health Science, University of Copenhagen, Copenhagen, Denmark, 3 Department of Biology, Faculty of Science, University of Copenhagen, Copenhagen, Denmark, 4 Pharmacology and Toxicology Department, University de Lausanne, Lausanne, Switzerland, 5 INSERM U970, Paris Cardiovascular Research Centre, Paris, France, 6 Universite´ Paris-Descartes, Paris, France

Abstract Loss of either hepatocyte growth factor activator inhibitor (HAI)-1 or -2 is associated with embryonic lethality in mice, which can be rescued by the simultaneous inactivation of the membrane-anchored serine protease, matriptase, thereby demonstrating that a matriptase-dependent proteolytic pathway is a critical developmental target for both protease inhibitors. Here, we performed a genetic epistasis analysis to identify additional components of this pathway by generating mice with combined deficiency in either HAI-1 or HAI-2, along with genes encoding developmentally co-expressed candidate matriptase targets, and screening for the rescue of embryonic development. Hypomorphic mutations in Prss8, encoding the GPI-anchored serine protease, prostasin (CAP1, PRSS8), restored placentation and normal development of HAI-1–deficient embryos and prevented early embryonic lethality, mid-gestation lethality due to placental labyrinth failure, and neural tube defects in HAI-2–deficient embryos. Inactivation of genes encoding c-Met, protease-activated receptor-2 (PAR-2), or the epithelial sodium channel (ENaC) alpha subunit all failed to rescue embryonic lethality, suggesting that deregulated matriptase-prostasin activity causes developmental failure independent of aberrant c-Met and PAR-2 signaling or impaired epithelial sodium transport. Furthermore, phenotypic analysis of PAR-1 and matriptase double-deficient embryos suggests that the protease may not be critical for focal proteolytic activation of PAR-2 during neural tube closure. Paradoxically, although matriptase auto-activates and is a well-established upstream epidermal activator of prostasin, biochemical analysis of matriptase- and prostasin-deficient placental tissues revealed a requirement of prostasin for conversion of the matriptase zymogen to active matriptase, whereas prostasin zymogen activation was matriptase- independent.

Citation: Szabo R, Uzzun Sales K, Kosa P, Shylo NA, Godiksen S, et al. (2012) Reduced Prostasin (CAP1/PRSS8) Activity Eliminates HAI-1 and HAI-2 Deficiency– Associated Developmental Defects by Preventing Matriptase Activation. PLoS Genet 8(8): e1002937. doi:10.1371/journal.pgen.1002937 Editor: Hamish S. Scott, SA Pathology, Australia Received April 5, 2012; Accepted July 18, 2012; Published August 30, 2012 This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication. Funding: The study was supported by the NIDCR Intramural Research Program (THB), the Augustinus Foundation, Købmand Kristian Kjær og Hustrus Foundation, the Kjær-Foundation, Dagmar Marshalls Foundation, Snedkermester Sophus Jacobsen og Hustru Astrid Jacobsens Foundation, Grosserer Valdemar Foersom og Hustru Thyra Foersoms Foundation, Fabrikant Einar Willumsens Mindelegat, the Harboe Foundation (SG and LKV), the Lundbeck Foundation (KKH, SG, and LKV), the Swiss National Science Foundation 31003A-127147/1 (EH), the INSERM Avenir, Marie Curie Actions, the French National Research Agency, and the Ile-de-France Region (EC). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction sodium and water homeostasis [12,13,14], iron homeostasis [15,16], and fertility [17,18]. Likewise, mounting evidence suggests Studies conducted within the past two decades have uncovered that excessive or spatially dysregulated membrane-anchored serine a large family of membrane-anchored serine proteases that protease activity contributes to several human disorders, including regulates vertebrate development, tissue homeostasis, and tissue congenital malformations [19], epithelial dysfunction [20,21,22], repair by providing focal proteolysis essential for cytokine and and cancer [3]. growth factor maturation, extracellular matrix remodeling, Matriptase is a modular type II transmembrane serine protease, signaling receptor activation, receptor shedding, regulation of encoded by the ST14 gene, that has pleiotropic functions in ion channel activity, and more (reviewed in [1,2,3]). Individual epithelial development and postnatal homeostasis, at least in part members of this family regulate both vertebrate development and through its capacity to regulate epithelial tight junction formation postnatal tissue homeostasis, including auditory and vestibular in simple and stratified epithelia [2,3]. In the human and mouse system development [4,5,6], differentiation of stratified epithelia epidermis, matriptase appears to function as part of a proteolytic [7,8], loss of epithelial tight junction function [9,10], failure to cascade in which it acts upstream of the GPI-anchored serine activate digestive enzymes [11], thyroid hormone availability [4], protease prostasin (CAP1/PRSS8), most likely by directly activat-

PLOS Genetics | www.plosgenetics.org 1 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Author Summary genetic epistasis analysis aimed at identifying additional components of the matriptase proteolytic pathway. Specifically, we Vertebrate embryogenesis is dependent upon a series of generated mice with simultaneous ablation of either the Spint1 precisely coordinated cell proliferation, migration, and gene (encoding HAI-1) or the Spint2 gene (encoding HAI-2) along differentiation events. Recently, the execution of these with genes encoding candidate matriptase targets that are co- events was shown to be guided in part by extracellular expressed with the protease during development. We then cues provided by focal pericellular proteolysis by a newly screened for the rescue of embryonic lethality or restoration of identified family of membrane-anchored serine proteases. HAI-1 and HAI-2-dependent morphogenic processes in these We now show that two of these membrane-anchored double-deficient mice. This analysis identified prostasin as critical serine proteases, prostasin and matriptase, constitute a to all matriptase-induced embryonic defects in both HAI-1- and single proteolytic signaling cascade that is active at HAI-2-deficient mice. Paradoxically, however, although matrip- multiple stages of development. Furthermore, we show tase autoactivates efficiently and prostasin is incapable of that failure to precisely regulate the enzymatic activity of both prostasin and matriptase by two developmentally co- undergoing autoactivation, we found that prostasin acts upstream expressed transmembrane serine protease inhibitors, of matriptase in the developing embryo and is required for hepatocyte growth factor activator inhibitor-1 and -2, conversion of the matriptase zymogen to active matriptase. causes an array of developmental defects, including clefting Finally, we explored the contribution of this newly identified of the embryonic ectoderm, lack of placental labyrinth prostasin-matriptase pathway to protease-activated receptor formation, and inability to close the neural tube. Our study (PAR)-dependent signaling during neural tube formation [45] also provides evidence that the failure to regulate the and now provide evidence that the pathway may be separate from prostasin–matriptase cascade may derail morphogenesis the proteolytic machinery that mediates focal activation of PAR-2 independent of the activation of known protease-regulated during neural tube closure. developmental signaling pathways. Because hepatocyte growth factor activator inhibitor–deficiency in humans is Results known to cause an assortment of common and rare developmental abnormalities, the aberrant activity of the Developmental defects in HAI-2–deficient mice tightly prostasin–matriptase cascade identified in our study may correlate with matriptase expression levels contribute importantly to genetic as well as sporadic birth 2/2 HAI-2-deficient (Spint2 ) mice were originally reported to defects in humans. display embryonic lethality prior to embryonic day 8 (E8.0), presenting with severe clefting of the embryonic ectoderm at E7.5 ing the prostasin zymogen [23,24,25,26]. Several additional and a failure to progress to the headfold stage [44]. We previously candidate proteolytic substrates have been identified for matrip- reported, however, that approximately 50% of HAI-2-deficient tase in cell-based and biochemical assays, including growth factor mice complete early development but die at midgestation due to precursors [27,28,29,30], protease-activated signaling receptors defective placental branching morphogenesis [43]. However, the [31,32,33], ion channels [34,35], and other protease zymogens genotyping strategy used in the latter study aimed at exploring the besides pro-prostasin [29,36,37]. However, the extent to which contribution of matriptase to this embryonic demise and only cleavage of these substrates is critical to matriptase-dependent allowed for the discrimination of HAI-2-deficient mice on 2 2 + + epithelial development and maintenance of epithelial homeostasis matriptase-sufficient (wildtype, Spint2 / ;St14 / , or haploinsuffi- 2 2 + 2 needs to be established. cient, Spint2 / ;St14 / ) backgrounds from a matriptase-deficient 2 2 2 2 Although matriptase is not required for term development in (Spint2 / ;St14 / ) background. Therefore, to test the possibility humans and most mouse strains ([24,38], and Szabo et al., that early embryonic development of HAI-2-deficient mice is St14 unpublished data), the membrane-anchored serine protease gene dosage-dependent, we first analyzed the offspring of + 2 + 2 nevertheless is expressed in many burgeoning embryonic as well interbred Spint2 / ;St14 / mice at various developmental stages. as extraembryonic epithelia [39,40,41,42]. Furthermore, we have This analysis revealed that the various developmental phenotypes previously shown that matriptase must be tightly regulated at the seen in HAI-2-deficient mice, indeed, were strongly dependent on post-translational level, for successful execution of several devel- St14 gene dosage (Figure 1A). Thus, HAI-2-deficient embryos + + opmental processes. Thus, loss of either of the two Kunitz-type carrying two wildtype matriptase alleles (St14 / ), displayed early 2 2 + + transmembrane serine protease inhibitors, hepatocyte growth lethality, as evidenced by only five percent of Spint2 / ;St14 / factor activator inhibitor (HAI)-1 or -2 or combined haploinsuffi- embryos developing beyond E9.0 and none past E10.5 (Figure 1A, 2 2 ciency for both inhibitors, is associated with uniform embryonic blue diamonds). Inactivation of one matriptase allele (Spint2 / ; + 2 lethality in mice [40,43]. Loss of HAI-1 or combined haploinsuf- St14 / ), however, was sufficient to partially rescue this early ficiency for HAI-1 and HAI-2 causes mid-gestation embryonic embryonic lethality of HAI-2-deficient mice (Figure 1A, red lethality due to failure to develop the placental labyrinth. Loss of squares). As reported previously [43], inactivation of both alleles of 2 2 2 2 HAI-2, in turn, is associated with three distinct phenotypes: a) matriptase (Spint2 / ;St14 / ) completely restored embryonic Early embryonic lethality, b) mid-gestation lethality due to survival and placental development and also reduced the placental labyrinth failure, and c) neural tube defects resulting in occurrence of neural tube defects associated with the loss of exencephaly, spina bifida, and curly tail. All developmental defects HAI-2 (Figure 1A, green triangles and Table 1). Taken together, in HAI-1- and HAI-2-deficient embryos, however, are rescued in these findings show that loss of HAI-2 may lead to three distinct whole or in part by simultaneous matriptase-deficiency, thus developmental phenotypes, dependent on the overall expression demonstrating that a matriptase-dependent proteolytic pathway is level of matriptase (Table 1): (i) early embryonic lethality occurring a critical morphogenic target for both protease inhibitors ([43,44], largely prior to E8.5, which can be partially rescued by matriptase 2 2 + 2 this study). haploinsufficiency (Spint2 / ;St14 / ) and completely by matrip- 2 2 2 2 In this study, we exploited the observation that HAI-1- and tase deficiency (Spint2 / ;St14 / ); (ii) placental defects resulting in 2 2 + 2 HAI-2-deficient mice display matriptase-dependent embryonic mid-gestation lethality, which are observed in Spint2 / ;St14 / 2 2 2 2 lethality with complete penetrance to perform a comprehensive embryos after E9.5, but are absent in Spint2 / ;St14 / embryos,

PLOS Genetics | www.plosgenetics.org 2 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 1. Effect of St14 gene dosage and c-Met activity on embryonic development in HAI-1– and HAI-2–deficient mice. (A) Matriptase haploinsufficiency partially restores early embryonic development of HAI-2 deficient mice. Relative frequency of Spint22/2;St14+/+ (blue diamonds and trend line), Spint22/2;St14+/2 (red squares and trendline), and Spint22/2;St142/2 (green triangles and trendline) embryos in offspring from interbred Spint2+/2;St14+/2 mice at E8.5–E15.5. The expected 25% Mendelian frequency is shown with the dotted trend line. 59–250 embryos were genotyped at each stage. (B) Matriptase haploinsufficiency does not rescue development of HAI-1-deficient mice. Genotype distribution of E8.5–E11.5 embryos and newborn (P1) offspring from interbred Spint1+/2;St14+/2 mice. No living St14+/+;Spint12/2 or St14+/2;Spint12/2 embryos are observed after E9.5. (C) Distribution of Spint1 genotypes in c-Met-expressing (Hgfr+/+ or Hgfr+/2, blue bars) and c-Met-deficient (Hgfr2/2, green bars) embryos from interbred Spint1+/2;Hgfr+/2 mice at E11.5–13.5. Loss of c-Met activity does not improve embryonic survival of HAI-1-deficient mice. (D and E) Distribution of Spint2 genotypes in c-Met-expressing (Hgfr+/+ or Hgfr+/2, blue bars) and c-Met-deficient (Hgfr2/2, green bars) embryos from Spint2+/2;Hgfr+/26Spint2+/2;Hgfr+/2 (D) or Spint2+/2;Hgfr+/26Spint2+/2;Hgfr+/2;St14+/2 (E) breeding pairs at E9.5–10.5. Only St14+/2 embryos are shown in (E). Loss of c-Met does not improve survival of HAI-2-deficient embryos. (F) Frequency of exencephaly observed in 153 control (Spint2+;Hgfr+), 53 c-Met- (Spint2+;Hgfr2/2), 24 HAI-2- (Spint22/2,Hgfr+), and 6 c-Met and HAI-2 double- (Spint22/2;Hgfr2/2) deficient embryos at E9.5. Loss of c-Met activity fails to correct neural tube defects in HAI-2-deficient mice. doi:10.1371/journal.pgen.1002937.g001

Table 1. Developmental defects observed in Spint1- and Spint-2-deficient mice as function of St14 expression.

Genotype Phenotype Penetrance St14+/+ St14+/2 St142/2

Spint12/2 Lack of placental labyrinth, embryonic lethality 100% 100% (no rescue) 0% (complete rescue) at E10.5 Spint22/2 Early embryonic lethality at E9.5 or earlier 100% 45% (partial rescue) 0% (complete rescue) Incomplete differentiation of placental N/A 100% (no rescue) 0% (complete rescue) labyrinth, embryonic lethality at E10.5–E14.5 Neural tube defects N/A Exencephaly 95–100% 18% (partial rescue; P,0.0001) Spina bifida 11% 13% (no rescue) Curly tail 89% 62% (partial rescue?; not significant)

doi:10.1371/journal.pgen.1002937.t001

PLOS Genetics | www.plosgenetics.org 3 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade and (iii) neural tube defects observed at or after E8.5 in most for matriptase in the epidermis of mice and humans (see 2 2 + 2 2 2 Spint2 / ;St14 / embryos, and partially rescued in Spint2 / ; Introduction). To explore the possibility that matriptase acts 2 2 St14 / embryos and term offspring. through prostasin to cause the signature defects in embryonic We next performed a similar analysis of the effect of St14 gene development of HAI-1- and HAI-2-deficient mice, we first dosage on the developmental defects and embryonic lethality performed a detailed immunohistochemical analysis of prostasin associated with HAI-1-deficiency by analyzing the offspring from expression in the developing embryo by staining histological + 2 + 2 + + interbred Spint1 / /St14 / mice (Figure 1B). As shown previ- sections from wildtype (Prss8 / ) and littermate prostasin-deficient 2 2 ously [40], a complete rescue of both the placental defects and (Prss8 / ) embryos with prostasin antibodies. Interestingly, embryonic lethality was observed in HAI-1-deficient mice prostasin was expressed in both the surface ectoderm, specifically 2 2 2 2 expressing no matriptase (Spint1 / ;St14 / ). However, compar- covering the converging neuroepithelium at the time of the neural 2 2 + 2 ison of HAI-1-deficient mice carrying one (Spint1 / ;St14 / )or tube closure (Figure 2A and 2B, compare with 2C), and in the 2 2 + + two (Spint1 / ;St14 / ) wildtype St14 alleles revealed identical developing placenta, where expression was detected as early as on defects in placental labyrinth formation and mid-gestation E8.5 and was present in the placental labyrinth in the entire period embryonic lethality occurring with complete penetrance of placental differentiation (Figure 2D, 2E, 2G, and 2H, compare (Figure 1B, Table 1, and data not shown). with 2F and 2I), thereby displaying co-expression with matriptase, HAI-1 and HAI-2 [40,41,42,43,45]. We, therefore, next directly Activation of hepatocyte growth factor (HGF) does not determined the contribution of prostasin to the matriptase- contribute to placental defects in HAI-1–deficient dependent developmental defects of HAI-1- and HAI-2-deficient embryos or early embryonic lethality and neural tube mice. For this purpose, we exploited the fact that the spontaneous mutant mouse strain, frizzy, recently was described to be defects in HAI-2–deficient mice homozygous for a point mutation in the coding region of the Matriptase is an efficient activator of proHGF [29,30] and Prss8 gene (Prss8fr/fr). This mutation results in a non-conservative dysregulated matriptase activity recently was shown to promote V170D amino acid substitution in the prostasin protein [49]. squamous cell carcinoma through activation of HGF-dependent c- Moreover, this mutant mouse strain completes development, but Met signaling [46]. Furthermore, both proHGF and its cognate displays an epidermal phenotype resembling mice carrying a receptor c-Met are expressed during embryogenesis in both the hypomorphic mutation in St14 [23], suggesting reduced expression placenta and the embryo [47,48]. To investigate the involvement or enzymatic activity of V170D prostasin. Western blot and of aberrant proHGF activation and c-Met signaling in the etiology immunohistochemical analysis of tissues from Prss8fr/fr mice did of the defects observed in HAI-1- and HAI-2-deficient embryos, not reveal an obvious reduction in the level of V170D prostasin + + we took advantage of the fact that c-Met is only required for expression when compared to wildtype prostasin in Prss8 / embryonic development beyond E13.5 [47,48]. This enabled the littermates (data not shown). Therefore, to assess the enzymatic study of key HAI-1- and HAI-2-dependent morphogenic processes 2 2 activity of the mutant prostasin, we generated enteropeptidase- in mice homozygous for a null mutation in Hgfr (Hgfr / ), + 2 activated recombinant V170D prostasin, as well as enteropepti- encoding c-Met. Analysis of embryos from interbred Spint1 / ; + 2 dase-activated wildtype and catalytically inactive (S238A) prostasin Hgfr / mice at E11.5–E13.5 revealed only one surviving 2 2 2 2 variants in HEK293T cells, as described previously [26]. These Spint1 / ;Hgfr / embryo, indicating that the loss of c-Met recombinant proteins were released from the plasma membrane activity does not restore placental development or embryonic by phosphatidylinositol-specific phospholipase C, activated with survival of HAI-1-deficient mice (Figure 1C, P,0.04, Chi-square 2 2 2 2 enteropeptidase, and their enzymatic activity towards a prostasin- Spint2 / ;Hgfr / test, and data not shown). Likewise, no selective fluorogenic peptide substrate (Figure 2J) as well as their embryos were detected beyond E9.5 (Figure 1D, P,0.02, Chi- ability to form enzymatic activity-dependent covalent complexes square test), indicating that the inactivation of c-Met signaling does with the serpin, protease nexin-1 (PN-1) (Figure 2K), were tested. not prevent matriptase-induced early embryonic lethality in HAI- + 2 + 2 + 2 As expected, wildtype recombinant prostasin exhibited easily 2-deficient mice. Interbreeding Spint2 / ;Hgfr / ;St14 / mice detectable hydrolytic activity towards the fluorogenic peptide allowed for the analysis of the impact of c-Met deficiency on the (Figure 2J, red line) and formed SDS-stable complexes with PN-1 formation of neural tube defects in HAI-2-deficient mice by (Figure 2K, compare lanes 3 and 4), while prostasin not activated preventing early embryonic lethality (Figure 1E). However, all of 2 2 2 2 + 2 by enteropeptidase and the catalytically inactive S238A mutant the Spint2 / ;Hgfr / ;St14 / embryos isolated at E9.5 from and exhibited no detectable hydrolytic activity (Figure 2J, black these crosses presented with exencephaly (Figure 1F), suggesting and grey lines) or PN-1 binding (Figure 2K and Figure S1A, lanes that c-Met signaling is not critically involved in the neural tube 2 2 2 and 12). V170D prostasin displayed a low residual enzymatic defects caused by the absence of HAI-2. All Spint2 / ; 2 2 + 2 activity that was above the baseline level, as defined by the Hgfr / ;St14 / embryos displayed synthetic lethality after E9.5, catalytically inactive S238A variant, and corresponded to about which precluded the direct analysis of the impact of c-Met loss on 6% of the activity of wildtype prostasin (Figure 2J, blue line), while the defects in placental differentiation caused by HAI-2 deficiency complex formation with PN-1 could not be detected (Figure 2K (data not shown). Taken together, these findings suggest that and Figure S1A, compare lanes 7 and 8). Taken together, these aberrant HGF-c-Met signaling does not contribute to the data indicated that V170D prostasin, expressed by the Prss8fr matriptase-dependent defects in placentation in HAI-1-deficient allele, displays greatly reduced enzymatic activity. We, therefore, embryos, or early lethality and neural tube closure of HAI-2- + 2 + + 2 next interbred Spint1 / ;Prss8fr/ and Spint1 / ;Prss8fr/fr mice and deficient embryos. analyzed the distribution of Spint1 alleles in the newborn offspring from these crosses. Consistent with our previous findings, loss of Reduced prostasin enzymatic activity prevents HAI-1 was not compatible with embryonic survival of mice 2 2 + developmental defects in both HAI-1– and HAI-2– carrying a wildtype prostasin allele (Spint1 / ;Prss8fr/ ) (Figure 3A, deficient mice blue bars). Interestingly, however, HAI-1-deficient mice carrying 2 2 The GPI-anchored membrane serine protease, prostasin two mutant prostasin alleles (Spint1 / ;Prss8fr/fr) developed to term (CAP1/PRSS8), is a well-validated downstream proteolytic target (Figure 3A, green bars), although they were found at a frequency

PLOS Genetics | www.plosgenetics.org 4 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 2. Prostasin expression in embryonic and extraembryonic tissues. (A–C) Immunohistochemical detection of prostasin at E8.5 in epithelial cells of surface ectoderm (examples with arrows in A and B) overlying the cranial neural tube region. Specificity of staining is shown by the absence of staining of Prss82/2 surface ectoderm (arrow in C). Filled arrowhead shows non-specific staining of yolk sac. No expression was observed

PLOS Genetics | www.plosgenetics.org 5 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

in the neuroepithelium (A and B, open arrowheads). (D–F) Immunohistochemical detection of prostasin in the chorionic ectoderm (examples with arrows) of mouse placenta at E8.5. Specificity of staining is shown by the absence of staining of Prss82/2 chorionic ectoderm (F). Filled arrowheads in D and F shows non-specific staining of trophoblast giant cells. No expression was detected in the trophoblast stem cell-containing chorionic epithelium (open arrowhead in E). (G–I) Immunohistochemical detection of prostasin in the placental labyrinth (examples with arrows in G and H) of mouse placenta at E12.5. Specificity of staining is shown by the absence of staining of the Prss82/2 labyrinth (I). No expression was detected in the trophoblast stem cell-containing chorionic epithelium (open arrowhead in H). Scale bars: A, C, D, F, G, and I, 100 mm; B, E, and H, 25 mm. (J) Enzymatic activity of wildtype (red), V170D (blue), S238A (grey), and zymogen (black) forms of prostasin. Prostasin variants were incubated with 50 mM pERTKR- AMC fluorogenic peptide at 37uC. V170D prostasin exhibited about 6% of the amidolytic activity of wildtype prostasin. No activity of catalytically inactive prostasin or prostasin zymogen was detected. (K) Western blot detection of SDS-stable complexes between prostasin and protein nexin-1 (PN-1). Wildtype zymogen (lanes 1 and 2), activated wildtype (lanes 3 and 4), V170D (frizzy) zymogen (lanes 5 and 6), activated V170D (lanes 7 and 8), S238A zymogen (lanes 9 and 10), and activated S238A (lanes 11 and 12) prostasin variants were incubated with (lanes 2, 4, 6, 8, 10, and 12) or without (lanes 1, 3, 5, 7, 9, and 11) 250 ng of recombinant human PN-1. Wildtype, but not V170D or S238A variants of prostasin formed SDS-stable complexes with PN-1. Positions of pro-prostasin, activated prostasin (migrating slightly faster than the zymogen due to removal of the 12 aa propeptide that is not detected after 4–12% SDS/PAGE with anti-prostasin antibody), and prostasin/PN-1 complexes are indicated. Positions of molecular weight markers (kDa) are shown on left. doi:10.1371/journal.pgen.1002937.g002 that was slightly lower than the expected Mendelian distribution caused by HAI-2 deficiency. Thus, macroscopic (Figure 4D and (20/127, 16% vs. expected 31.75/127, 25%, P,0.05, Chi-square 4E) and histological (Figure 4F and 4G) examination of 2 2 + + 2 2 + 2 test). Furthermore, morphometric analysis showed that reduced Spint2 / ;Prss8fr/fr;St14 / and Spint2 / ;Prss8fr/fr;St14 / embry- prostasin activity fully restored placental labyrinth formation in os showed that, respectively, 5% and 0%, of these embryos HAI-1-deficient embryos, as evidenced by normal histological exhibited exencephaly when analyzed after E9.5 (Figure 4H), and appearance of the labyrinth (Figure 3B–3G), thickness of the no embryos with either spina bifida or curly tail were observed labyrinth layer (Figure 3H) and labyrinth vessel density (Figure 3I) (data not shown). Similarly, histological analysis of placental tissues 2 2 2 2 2 2 + of Spint1 / ;Prss8fr/fr embryos. Furthermore, macroscopic and from E10.5–E13.5 Spint2 / ;Prss8fr/fr or Spint2 / ;Prss8fr/fr;St14 / 2 histological analysis of embryos extracted between E11.5 and embryos did not reveal any of the stereotypic defects associated E13.5 failed to reveal any obvious developmental abnormalities with HAI-2 deficiency (Figure 4I and 4J). Thus, the overall within either embryonic or extraembryonic tissues of appearance of the placental layers (Figure 4I and 4J), the thickness 2/2 fr/fr Spint1 ;Prss8 mice (data not shown), and these mice were of the placental labyrinth (Figure 4K), and the number of fetal fr/fr outwardly indistinguishable from their Prss8 littermates at vessels within the labyrinth (Figure 4L), all were comparable to the weaning and when followed for up to one year (Figure 3J). Taken HAI-2-sufficient littermate controls. In conclusion, these data together, these data show that the matriptase-mediated develop- document an essential role of prostasin in the etiology of all of the mental defects in HAI-1-deficient mice are prostasin-dependent. developmental defects previously observed in HAI-2-deficient To determine the impact of diminished prostasin activity on the mice. developmental defects associated with HAI-2-deficiency, we next analyzed neural tube closure, placental differentiation, and overall + 2 + Prostasin is required for the activation of matriptase survival of the offspring of interbred Spint2 / ;Prss8fr/ mice. during development Analysis of the genotype distribution of embryos at E9.5–11.5 did not identify any HAI-2-deficient embryos carrying at least one Matriptase was previously identified as an essential proteolytic 2 2 + + 2 2 + wildtype Prss8 allele (Spint2 / ;Prss8 / or Spint2 / ;Prss8fr/ ) activator of prostasin in the epidermis, and the near ubiquitous co- (Figure 4A). Interestingly, however, HAI-2-deficient embryos localization of the two membrane serine proteases in the epithelial 2 2 2 carrying two mutant Prss8 alleles (Spint2 / ;Prss8 fr/fr) were found compartment of most other adult tissues indicate that this in the expected Mendelian ratio as late as E13.5–15.5 (Figure 4B, matriptase-prostasin proteolytic pathway may be operating in green bars). Furthermore, genotyping of newborn offspring multiple epithelia to maintain tissue homeostasis [23,24,25,26]. 2 2 revealed the presence of living Spint2 / ;Prss8fr/fr pups The genetic epistasis analysis performed above provided strong (Figure 4C, green bars), although they were found at slightly evidence that matriptase and prostasin also are part of a single lower than expected frequency (15% vs. expected 25%, P,0.06, proteolytic cascade in the context of embryonic development. Chi-square test). These data strongly suggest that matriptase and Furthermore, the striking overlap in expression of the two prostasin act as part of a single proteolytic cascade to cause proteases documented earlier in the surface ectoderm during developmental defects in HAI-2-deficient mice. If this were the neural tube closure (see above) was also observed in the developing case, we hypothesized that lowering the activity of this cascade placenta (compare Figure 5A and 5B). To further investigate the 2 2 even further by eliminating one St14 allele from Spint2 / ;Prss8fr/fr functional interrelationship between the two proteases, we embryos should additionally improve the term survival of HAI-2- analyzed the levels of the activated forms of matriptase and deficient mice. Indeed, genotyping of born offspring from prostasin in embryonic and placental tissues from matriptase- + 2 + + 2 2/2 2/2 interbred Spint2 / ;Prss8fr/ ;St14 / mice showed a normal (St14 ) or prostasin- (Prss8 ) deficient mice at E11.5. + 2 distribution of Spint2 alleles in Prss8fr/fr;St14 / pups (Figure 4C, Trypsin-like serine proteases are activated by autocatalytic or red bars), further suggesting that failure to regulate a proteolytic heterocatalytic cleavage after an arginine or lysine residue, located pathway including matriptase and prostasin accounts for all of the in a conserved activation motif within the catalytic domain. embryonic lethality caused by loss of HAI-2. Activation cleavage severs the bond between the catalytic domain As reported previously, neural tube defects, including exence- and upstream accessory domains, but the activated protease phaly, spina bifida, and curly tail were seen in 95–100% of domain remains connected to upstream accessory domains by a 2 2 + + 2 Spint2 / ;Prss8fr/ ;St14 / mice ([43], this study). Examination of disulfide bond [50]. Zymogen activation, therefore, can be 2 2 embryonic and extraembryonic tissues from Spint2 / ;Prss8fr/fr detected by a mobility shift in reducing SDS-PAGE gels, which 2 2 + 2 and Spint2 / ;Prss8fr/fr;St14 / embryos, however, revealed that breaks the disulfide bond that keeps the two domains together. reduced prostasin activity sufficed to almost completely rescue the Direct detection of active matriptase in placental tissues by western defects in both neural tube closure and placental differentiation blot, however, proved unsuccessful due to low signal intensity and

PLOS Genetics | www.plosgenetics.org 6 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 3. Reduced prostasin activity restores placental development and embryonic survival of HAI-1–deficient mice. (A) Distribution of genotypes of born offspring of intercrossed Spint1+/2;Prss8fr/2 mice. No Spint12/2 mice expressing one or two wildtype prostasin alleles (Prss8+/+ or Prss8+/fr, blue bars) were identified, while Spint12/2 embryos carrying two mutant prostasin alleles (Prss8fr/fr, green bars) were found in near-expected frequency. (B–G) Representative low (B–D) and high (E–G) magnification images showing the histological appearance of H&E-stained placental tissues of (Spint1+;Prss8+) (B and E), (Spint12/2;Prss8+) (C and F), and (Spint12/2;Prss8fr/fr) (D and G) embryos at E11.5. The thickness of the placental labyrinth (two-sided arrows between the dotted lines in B–D), as well as the number of fetal vessels (E–G, arrows) and lacunae filled with maternal blood (E–G, arrowheads) within the labyrinth is markedly reduced in prostasin-sufficient (C and F), but not in prostasin-deficient (D and G) Spint12/2 embryos, when compared to the controls (B and E). (H, I) Quantification of the maximum thickness of the labyrinth layer (H) and the number of fetal vessels in the placental labyrinth (I) of Spint1+;Prss8+, Spint1+;Prss8fr/fr, Spint12/2;Psrr8+, and Spint12/2;Psrr8fr/fr embryos at E11.5. The thickness of the labyrinth and fetal vessel density were strongly diminished in HAI-1-deficient mice but completely restored in HAI-1-deficient mice with low prostasin activity. (J) Outward appearance of one-year-old Spint12/2;Prss8fr/fr and littermate Spint1+;Prss8+ mice. ***, p,0.0001, Student’s t-Test, two tailed. Scale bars: B–D, 100 mm; E–G, 25 mm. doi:10.1371/journal.pgen.1002937.g003 a strong cross reactivity of available anti-matriptase antibodies matriptase-deficient mice (Figure 5C, lane 3). This band was not with unrelated antigens. Similarly, direct detection of active detected in placental extracts from prostasin-deficient embryos prostasin by western blot failed due to the small difference in the (Figure 5C, lane 1) or when anti-HAI-1 antibodies were omitted electrophoretic mobility of the zymogen and the active form of the from the assay (Figure 5D, compare lanes 1 and 2), indicating that enzyme (data not shown). In order to circumvent these problems, it represents the active form of prostasin released from an we instead determined the amount of active matriptase and inhibitory complex with HAI-1. In support of this, when prostasin prostasin that formed inhibitor complexes with endogenous HAI-1 from either matriptase-deficient or littermate wildtype control in embryonic tissues from wildtype, matriptase-, and prostasin- placental tissues was released from the immunoprecipitated HAI- deficient embryos. Immunoprecipitation of protein extracts using 1-prostasin complexes by brief exposure to low pH, it was able to anti-mouse HAI-1 antibodies followed by western blot with form SDS-stable complex with PN-1, which requires the catalytic prostasin antibodies detected the presence of the 38 kDa band activity of prostasin (Figure 5C, compare lane 3 with 5 and lane 4 in the placentas of wildtype mice (Figure 5C, lanes 2 and 4) and with 6). Quantification of the amount of active prostasin in

PLOS Genetics | www.plosgenetics.org 7 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 4. Reduced prostasin activity restores placental differentiation, embryonic survival, and neural tube closure in HAI-2– deficient mice. (A) Distribution of Spint2 genotypes in prostasin-sufficient (Prss8+/+ or Prss8+/fr) E9.5–11.5 offspring from interbred Spint2+/2;Prss8fr/+ mice. No Spint22/2 embryos were observed (P,0.025, Chi-square test). (B) Distribution of Spint2 genotypes in prostasin-sufficient (Prss8+/+ or Prss8+/fr, blue bars) and prostasin-deficient (Prss8fr/fr, green bars) mouse embryos from interbred Spint2+/2;Prss8fr/+ mice at E13.5–15.5. No prostasin-expressing Spint22/2 embryos were observed (P,0.001, Chi-square test), while survival of prostasin-deficient Spint22/2 embryos was restored. (C) Distribution of Spint2 genotypes in newborn prostasin-sufficient, matriptase wildtype (Prss8+/+ or Prss8+/fr;St14+/+, blue bars), prostasin-deficient, matriptase wildtype (Prss8fr/fr;St14+/+, green bars), and prostasin-deficient, matriptase haploinsufficient (Prss8fr/fr;St14+/2, red bars) offspring from Spint2+/2;Prss8fr/2 6Spint2+/2;Prss8fr/+;St14+/2 breeding pairs. Reduced prostasin activity restored embryonic survival of Spint22/2 mice partially in matriptase wildtype and completely in matriptase haploinsufficient mice. (D–G) Macroscopic (D and E) and histological (H&E staining) (F and G) appearance of the HAI-2- deficient, matriptase- and prostasin-sufficient (Spint22/2;Prss8+/+ or Prss8+/fr, St14+/2, D and F) or HAI-2- and prostasin-deficient, matriptase-sufficient (Spint22/2;Prss8fr/fr, St14+/+ or St14+/2) (E and G) embryos at E9.5. HAI-2 deficiency prevents convergence of neural folds in the cranial region of neural tube (D and F, arrows) leading to exencephaly. Convergence and fusion of neural folds are restored in HAI-2-deficient mice with low prostasin activity (E and G, arrows). Presence of medial (F, open arrowhead) and absence of dorsolateral (F, arrowheads) hinge points. (H) Frequency of exencephaly in E9.5–18.5 Spint22/2 embryos with different levels of prostasin activity (Prss8+/+, Prss8fr/+ or Psrr8fr/fr) and matriptase (St14+/+, St14+/2 or St142/2). The frequency of neural tube defects is inversely correlated with the combined number of wildtype Prss8 and St14 alleles. A total of 524 embryos were analyzed. (I–L) Histological appearance (H&E staining) (I and J), thickness of placental labyrinth (K), and number of fetal vessels within the labyrinth (L) in the placentas of HAI-2 and prostasin-sufficient (Spint2+;Prss8+) and HAI-2 and prostasin double-deficient (Spint22/2;Prss8fr/fr) embryos at E12.5. Reduced prostasin activity restores differentiation of placental labyrinth in Spint22/2 mice to levels not significantly (N.S.) different from wildtype littermate controls. Arrows in I and J show examples of fetal vessels. Scale bars: F, 50 mm G, I, and J, 100 mm. doi:10.1371/journal.pgen.1002937.g004

PLOS Genetics | www.plosgenetics.org 8 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 5. Prostasin is required for the activation of matriptase during placental differentiation. (A and B) Expression of prostasin (A) and matriptase (B) in placental tissues of wildtype mice at E11.5. Both proteins were expressed in the chorionic (arrows) and labyrinthine (arrowheads) trophoblasts. (C) Western blot detection of active prostasin in the fetal part of the placenta of wildtype (Prss8+/+ and St14+/+, lanes 2, 4, and 6), prostasin-deficient (St14+/+;Prss82/2) (lane 1), and matriptase-deficient (St142/2;Prss8+/+) (lanes 3 and 5) embryos at E11.5 after immunoprecipitation with anti-mouse HAI-1 antibodies. Immunoprecipitated proteins in lanes 5 and 6 were acid-exposed to dissociate prostasin-HAI-1 complexes, and then incubated with PN-1 prior to western blot analysis. Positions of bands corresponding to active prostasin, prostasin/PN-1 complex, as well as non- specific signals of IgG heavy and light chains are indicated on the right. Positions of molecular weight markers (kDa) are shown on the left. (D) Omission of anti-HAI-1 antibody resulted in loss of detectable prostasin (compare lanes 1 and 2), indicating that the detected prostasin formed complexes with HAI-1. (E) Quantification of the relative amount of active prostasin in wildtype and matriptase placentae by densitometric scanning of prostasin western blots of HAI-1 immunoprecipitated material from (Prss82/2;St14+/+,N=3,Prss8+/+;St142/2, N = 3, and Prss8+/+;St14+/+, N = 6). Data are shown as mean 6 standard deviation (N.S., not significant). (F and G) Western blot detection of active matriptase in the fetal part of the placenta at E11.5 (F) after anti-HAI-1 immunoprecipitation, and in the epidermis of newborn skin (G) of wildtype (Prss8+/+ and St14+/+) (F, lanes 1, and 3, and G, lane 2), prostasin-deficient (St14+/+;Prss82/2) (F, lane 4 and G, lane 1), and matriptase-deficient (St142/2;Prss8+/+) (F, lane 2, and G, lane 3) embryos. A 30 kDa band representing the active serine protease domain of matriptase (Mat SPD) was present in extracts from wildtype (lanes 1 and 3 in F), but not in matriptase- (lane 2 in F) or prostasin-deficient (lane 4 in F) placenta. Zymogen (Mat FL) and active (Mat SPD) forms of matriptase were detected in extracts from both wildtype and prostasin-deficient, but not matriptase-deficient epidermis. (H–H0) Immunohistochemical staining of matriptase in control Prss8+ (H) and prostasin-deficient Prss82/2 (H9) placenta at E11.5. Specificity of staining of chorionic and labyrinthine trophoblasts (examples with arrows) is shown by the absence of staining of corresponding cells in St142/2 placenta (H0). Insets in H and H9 are parallel sections stained with prostasin antibodies. Open arrowheads in H–H0 show examples of non-specific staining. Scale bars: A, B, H, H9, and H0,50mm. doi:10.1371/journal.pgen.1002937.g005

PLOS Genetics | www.plosgenetics.org 9 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade wildtype and prostasin-deficient placentae by densitometric scans immunoprecipitation failed to detect either of the proteases, likely of western blots showed that the loss of matriptase did not affect due to the restricted expression of both proteins (data not shown). the amount of active prostasin (Figure 5E). Taken together, these data suggest that the developing placenta does not require Developmental defects in HAI-2–deficient embryos are matriptase for the activation of prostasin. not caused by aberrant activity of the epithelial sodium Detection of matriptase by western blot after immunoprecipi- channel tation with anti-HAI-1 antibodies revealed the presence of a Both matriptase and prostasin have been reported to activate 30 kDa band corresponding to the activated matriptase serine the epithelial sodium channel (ENaC) in cell-based assays, and protease domain in wildtype placental tissues, but not matriptase- prostasin is a critical regulator of ENaC activity during alveolar deficient placental tissues (Figure 5F, compare lanes 1 and 2). fluid clearance in mouse lungs and likely regulates ENaC activity Surprisingly, however, the active form of matriptase was also in many other adult organs [51,52]. Immunohistological analysis absent in the extracts from prostasin-deficient placentae (Figure 5F, of embryonic tissues at E11.5 revealed strong expression of ENaC compare lanes 3 and 4). The absence of matriptase was observed in the developing labyrinth layer of the placenta (Figure 6D). No in four independent experiments using placentae from a total of ENaC expression was detected in the embryo proper (data not seven prostasin-deficient mice and their prostasin-sufficient litter- shown). To investigate a possible involvement of ENaC in the mate controls (Figure S1B and data not shown). As expected, etiology of prostasin-matriptase-induced developmental defects in + 2 analysis of skin extracts from prostasin-deficient newborn mice and HAI-2-deficient mice, pregnant females from Spint2 / mice bred + 2 + 2 wildtype littermate controls using the same western blot conditions to Spint2 / ;St14 / mice were treated daily between E5.5–8.5 clearly showed the presence of the active form of matriptase in with the pharmacological inhibitor of ENaC activity, amiloride, both the control and prostasin-deficient mice (Figure 5G, compare which is known to cross the feto-maternal barrier [53]. Genotyp- lanes 1 and 2 with lane 3), demonstrating that differences in the ing of embryos extracted at E9.5 from these crosses did not 2 2 + + functional relationship between the two proteases exist in different identify any Spint2 / ;St14 / embryos (Figure S1C), indicating tissues. Immunohistochemistry of placentae from littermate that the inhibition of ENaC activity is not sufficient to prevent control and prostasin-deficient embryos showed no obvious early embryonic lethality resulting from the loss of HAI-2. 2 2 + 2 difference in levels or pattern of matriptase expression (compare Furthermore, all of the seven Spint2 / ;St14 / embryos identi- 9 Figure 5H and 5H ). fied in this experiment exhibited exencephaly, suggesting that To further substantiate the above findings, we next determined ENaC activity is not critically involved in the etiology of neural if prostasin could serve as an activator of matriptase in a tube defects in HAI-2-deficient mice (Figure 6E). Similarly, genetic reconstituted cell-based assay. For this purpose, we transiently inactivation of the a subunit of ENaC (encoded by Scnn1a), which transfected HEK-293 cells with expression vectors encoding HAI- is necessary for channel activity in vivo [54,55], failed to rescue 1 (to allow for efficient matriptase expression) and wildtype or embryonic development of HAI-2-deficient animals, as evidenced 2 2 2 2 catalytically inactive matriptase. The transfected cells were then by a complete absence of any surviving Spint2 / ;Scnn1a / + 2 + 2 exposed to soluble recombinant prostasin or vehicle, and double-deficient embryos at E9.5 from Spint2 / ;Scnn1a / mice + 2 + 2 matriptase activation was analyzed six hours later by western blot bred to Spint2 / ;Scnn1a / mice (Figure 6F) and the failure of 2 2 2 2 of cell lysates (Figure 6A) or conditioned medium (Figure 6B). Spint2 / ;Scnn1a / double-deficient mice to appear in the + 2 + 2 Interestingly, soluble prostasin efficiently activated matriptase, as newborn offspring from Spint2 / ;Scnn1a / mice bred to + 2 + 2 + 2 evidenced by the large increase in the amount of the liberated Spint2 / ;Scnn1a / ;St14 / mice (Figure 6G). Taken together, matriptase serine protease domain (Mat SPD, Figure 6A and 6B, these data do not support the critical involvement of aberrant compare lanes 1 and 2) after reducing SDS/PAGE, and a ENaC activity in the developmental defects resulting from lack of corresponding diminution of the amount of matriptase zymogen HAI-2 regulation of the prostasin-matriptase proteolytic pathway. (Mat SEA, Figure 6A and 6B, compare lanes 1 and 2). Activation site cleavage of matriptase by prostasin did not require matriptase Excess PAR-2 signaling does not cause developmental catalytic activity, as shown by the increased amount of the isolated defects in HAI-2–deficient mice matriptase serine protease domain in prostasin-treated cells Matriptase and prostasin are co-expressed with PAR-2 in expressing a catalytically inactive matriptase (Figure 6A and 6B, surface ectoderm during neural tube closure ([43,45], this study), compare lanes 3 and 4). Similar results were obtained when and matriptase displays extraordinarily favorable activation matriptase-transfected HEK-293 cells were transfected with a kinetics towards PAR-2 in cell-based assays [43,45]. Furthermore, prostasin expression vector, rather than being treated with soluble activation of PAR-2 (encoded by the F2rl1 gene) was recently prostasin (data not shown). To investigate if the prostasin-activated shown to contribute to neural tube closure (see below). These data matriptase displayed functional activity, the HEK-293 cells suggested that some, or all, of the prostasin- and matriptase- described above were also transfected with a PAR-2 expression dependent defects in HAI-2-deficient mice could be caused by vector and a serum response element (SRE)-luciferase reporter excess PAR-2 signaling. To test this hypothesis, we interbred + 2 + 2 plasmid to measure PAR-2 activity (Figure 6C). Exposure of Spint2 / ;F2rl1 / mice and genotyped the ensuing embryos at 2 2 2 2 serum-starved cells to soluble prostasin resulted in a large increase E9.5. This analysis failed to identify any Spint2 / ;F2rl1 / in luciferase activity in cells transfected with wildtype matriptase embryos (Figure 7A). Thus, the loss of PAR-2 activity is not (Figure 6C, left panels), but not in cells transfected with sufficient to overcome matriptase- and prostasin-dependent early catalytically inactive matriptase (Figure 6C, second panels from embryonic lethality in HAI-2-deficient mice. When the early left), with HAI-1 alone (Figure 6C, second panels from right) or embryonic survival was improved by matriptase haploinsufficiency with empty vector (Figure 6C, right panels). Taken together, the (see above), analysis of neural tubes at E9.5 revealed exencephaly 2 2 2 2 + 2 data indicate that prostasin can proteolytically activate matriptase in 100% of Spint2 / ;F2rl1 / St14 / embryos, identical to the and is critical for the generation of active matriptase during frequency of defects observed in littermate HAI-2-deficient 2 2 + + + 2 placental development. Detection of active matriptase and embryos expressing PAR-2 (Spint2 / ;F2rl1 / or F2rl1 / ; + 2 prostasin in the embryo by western blot or by anti-HAI-1 St14 / ) (Figure 7B). Thus, excess PAR-2 activation does not

PLOS Genetics | www.plosgenetics.org 10 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 6. Prostasin activates matriptase on the surface of HEK293 cells. (A and B) Western blot detection of matriptase in cell lysates (A) and in the conditioned medium (B) from HEK293 cells transiently transfected with wildtype recombinant human matriptase and HAI-1 expression vectors (lanes 1 and 2), catalytically inactive (S805A) matriptase and HAI-1 (lanes 3 and 4), HAI-1 alone (lanes 5 and 6), and cells transfected with a control empty vector (lanes 7 and 8) that were incubated with (lanes 2, 4, 6, and 8) or without (lanes 1, 3, 5, and 7) 100 nM soluble recombinant human prostasin. Addition of prostasin promoted conversion of the matriptase zymogen to its activated two-chain form. Positions of bands corresponding to full length matriptase (Mat FL), matriptase pro-enzyme processed by autocatalytic cleavage within the SEA domain (Mat SEA), and activated matriptase serine protease domain (Mat SPD) are indicated on the right. Positions of molecular weight markers (kDa) are shown on left. (C)

PLOS Genetics | www.plosgenetics.org 11 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Quantification of the activation of PAR-2 in HEK293 cells expressing recombinant human PAR-2 in combination with wildtype (WT) or inactive (S805A) variants of matriptase and HAI-1, HAI-1 alone, or transfected with an empty vector, incubated without (blue bars) or with (red bars) 100 nM soluble recombinant human prostasin. Prostasin induced matriptase activity-dependent activation of PAR-2. (D) Immunohistochemical analysis of the expression of the gamma subunit of the epithelial sodium channel (ENaC) in placenta of control mice at E11.5. The expression was detected in the populations of chorionic (arrow) and labyrinthine (arrowhead) trophoblasts. Scale bar: 50 mm. (E) Frequency of exencephaly in amiloride-treated wildtype (Spint2+/+, N = 56), untreated HAI-2-deficient (Spint22/2, N = 12) and amiloride-treated HAI-2-deficient (Spint22/2, N = 7) embryos at E9.5. Amiloride treatment failed to rescue neural tube defects in Spint22/2; St14+/2 embryos. (F) Distribution of Spint2 genotypes in ENaC-expressing (Scnn1a+/+ or Scnn1a+/2, blue bars) and ENaC-deficient (Scnn1a2/2, green bars) offspring from Spint2+/2;Scnn1a+/26Spint2+/2;Scnn1a+/2 breeding pairs at E9.5. Loss of ENaC expression did not rescue early embryonic lethality in Spint22/2 mice. (G) Distribution of Spint2 genotypes in matriptase- haploinsufficient ENaC-expressing (St14+/2;Scnn1a+/+ or Scnn1a+/2, blue bars) and ENaC-deficient (St14+/2; Scnn1a2/2, green bars) offspring from Spint2+/2;Scnn1a+/2, St14+/26Spint2+/2;Scnn1a+/2;St14+/+ breeding pairs at birth. Loss of ENaC expression did not rescue overall embryonic survival in Spint22/2; St14+/2 mice. doi:10.1371/journal.pgen.1002937.g006 appear to be critically involved in the etiology of neural tube mice were generally only obvious after E10.5 (compare Figure 7H defects in HAI-2-deficient mice. with 7I and 7J), and were generally restricted to the hindbrain region of the neural tube, with less than five percent exhibiting Loss of matriptase does not cause neural tube defects in exencephaly that extended to the midbrain region (Figure 7I and PAR-1–deficient embryos Table 2). In contrast, HAI-2 deficiency was generally associated with a failure of cranial neural tube closure that was obvious at Rac1 activation in surface ectoderm through Gi, initiated by either PAR-1 or PAR-2 activation was recently shown to be E9.5 or earlier, and was due to the inability of neural folds to elevate properly, and to come into juxtaposition necessary for the required for neural tube closure. Thus, mice with combined, but 9 9 not single, deficiency in PAR-1 and PAR-2 display exencephaly fusion (compare Figure 7F with 7K, and 7F with 7K ). The fusion with high frequency [45]. As matriptase and prostasin are co- at closure point 1 of HAI-2-deficient mice was completed in all embryos analyzed at E9.5 or later, and no case of craniorachis- expressed with PAR-2 in surface ectoderm during neural tube 2/2 closure ([43,45], this study), we next investigated if the prostasin- chisis was observed (Table 2). However, 10 percent of Spint2 matriptase cascade identified in the current study contributes to embryos failed to initiate fusion at closure point 2, resulting in exencephaly that extended from forebrain region to the hindbrain- physiological PAR-2 activation during neural tube closure. To test 9 this, we generated mice with combined deficiency in PAR-1 cervical boundary (Figure 7K and 7K ). In addition, even in the 2 2 2 2 (encoded by the F2r gene) and matriptase (F2r / ;St14 / ). If embryos that successfully initiated the fusion at closure point 2, the matriptase was essential for the activation of PAR-2 during neural exencephaly was more extensive than the one observed in PAR-1 tube closure, these mice should phenocopy mice with a combined and PAR-2 double-deficient embryos, typically spanning the entire PAR-1 and PAR-2 deficiency, including embryonic lethality and midbrain and hindbrain regions (compare Figure 7L and 7I). high susceptibility to cranial neural tube defects [45]. However, Finally, histological analysis of affected embryos at E9.5 showed that dorsolateral hinge points (DLHPs) critical for the final stages analysis of the midgestation embryos from intercrossed 2 2 + 2 + 2 / F2r / ;St14 / mice yielded the expected distribution of all of the neural tube closure were absent in 97 percent of Spint2 2/ embryos (Figure 4D and 4F, Figure 7K and 7K9, and Table 2), genotypes (Figure 7C). Furthermore, none of 20 observed F2r 2 2 2 2 2 2 2 / / ;St14 / mice displayed cranial neural tube defects, although the while F2r ;F2rl1 embryos generally exhibited DLHP PAR-1 deficiency alone or in combination with matriptase formation indistinguishable from wildtype littermate controls (Figure 7G, 7G9, and 7M, compare to Figure 4E, 4G, Figure 7F deficiency occasionally led to defects in the closure of the posterior 9 neural tube, resulting in spina bifida and curly tail (Figure 7D). and 7F ). Thus, substantial differences are observed in the location, frequency, extent, and onset of the neural tube defects of HAI-2-deficient mice and PAR-1 and PAR-2 double-deficient HAI-2 and PAR-1/PAR-2 regulate different stages of mice, further indicating the independent roles of, respectively, neural tube development repression and activation of the two protease-regulated pathways The lack of functional interaction between prostasin-matriptase in distinct stages of neural tube formation. and PAR-1/PAR-2 regulated signaling pathways evidenced from the above experiments suggested that the two pathways are either Discussion involved in two essential, non-redundant mechanisms regulating the same steps of neural tube closure, or that they may regulate In this study, we exploited the uniform matriptase-dependent different stages of the process. To distinguish between the two embryonic lethality of mice deficient in hepatocyte growth factor possibilities, we performed a detailed morphologic comparison of activator inhibitors as a means to genetically identify novel the neural tube defects caused by loss of HAI-2 and by the molecules and pathways regulating and being regulated by combined loss of PAR-1 and PAR-2. Macroscopic analysis showed matriptase in the developing embryo by epistasis analysis. This significant differences in the types of neural tube defects in the two analysis resulted in a number of unexpected findings. First, we mutant mouse strains. In PAR-1 and PAR-2 double-deficient found that prostasin is an essential component of the matriptase- mice, the defects were almost exclusively restricted to the dependent molecular machinery that causes early embryonic hindbrain region of the cranial neural tube (Figure 7E and lethality, derails placental labyrinth formation, and causes defects Table 2). In addition, PAR-1 and PAR-2 double-deficient embryos in neural tube closure in these mice. This shows that both proteins did not exhibit any obvious abnormalities during the early stages are expressed, are active, functionally interact, and must be of neural tube closure, as all of the embryos analyzed before E10.5 regulated by hepatocyte growth factor activator inhibitors already showed normal elevation and conversion of the opposing neural during early development. Surprisingly, however, rather than folds, as well as the completion of the neural fold fusion at initial being a downstream effector of matriptase function, as previously closure points 1 and 2 at hindbrain/cervix and forebrain/ established for both mouse and human epidermis ([23,24,25,26], midbrain boundaries, respectively (compare Figure 7F with 7G, this study), prostasin acts upstream of matriptase during embryo- and 7F9 with 7G9). As a result, the neural tube defects in these genesis and is essential for activation of the matriptase zymogen.

PLOS Genetics | www.plosgenetics.org 12 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

Figure 7. Neural tube defects and embryonic lethality in HAI-2–deficient mice are not dependent on PAR-2, and combined PAR-1 and matriptase deficiency does not phenocopy combined PAR-1 and PAR-2 deficiency. (A) Distribution of Spint2 genotypes at E9.5 in PAR-2-expressing (F2rl1+/+ or F2rl1+/2, blue bars) and PAR-2-deficient (F2rl12/2, green bars) offspring from interbred Spint2+/2,F2rl1+/2 mice. No Spint22/2 embryos were detected irrespective of PAR-2 expression. (B) Frequency of exencephaly observed in HAI-2 and PAR-2-sufficient (Spint2+;F2rl1+ N = 366), PAR-2-deficient (Spint2+;F2rl12/2, N = 164), HAI-2-deficient (Spint22/2,F2rl1+, N = 18), and PAR-2 and HAI-2 double- (Spint22/2;F2rl12/2, N = 12) deficient embryos extracted at E9.5–E11.5. Loss of PAR-2 activity fails to correct neural tube defects in HAI-2-deficient embryos. (C) Distribution of St14 alleles at E11.5–15.5 in PAR-1-expressing (F2r+/+ or F2r+/2, blue bars) and PAR-1-deficient (F2r2/2, green bars) embryos from interbred St14+/2;F2r+/2 mice. Loss of PAR-1 activity does not affect embryonic survival of matriptase-deficient mice. (D) Frequency of exencephaly (Ex), spina bifida (SB), and curly tail (CT) in E9.5–18.5 embryos with different levels of expression of PAR-1 (F2r+ or F2r2/2) and matriptase (St14+or St142/2). A total of 326 embryos were analyzed. Loss of matriptase does not significantly increase the incidence of neural tube defects in PAR-1-deficient embryos. (E) Comparison of the severity of exencephaly in HAI-2-deficient (Spint22/2, N = 29) and PAR-1 and PAR-2 double-deficient (F2r2/2;F2rl12/2, N = 39) embryos. 95% of affected F2r2/2;F2rl12/2 embryos exhibited exencephaly that was confined to hindbrain region of the

PLOS Genetics | www.plosgenetics.org 13 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

cranium (HB only, green bars), with the remaining 5% extending to the midbrain region (MB-HB, blue bars). In contrast, only 10% of exencephalies observed in Spint22/2-deficient mice were confined to the hindbrain, with 59% extended to midbrain, and 31% to forebrain region (FB-HB, red bars). (F–G9) Ventral (F and G) and dorsal (F9 and G9) view of non-affected control (F and F9) and affected PAR-1 and PAR-2 double-deficient (F2r2/2;F2rl12/ 2) (G and G9) embryos at E9.5. The initial stages of neural tube closure all appear to be unaffected by the combined absence of PAR-1 and PAR-2. (H– J) Appearance of control (H) and PAR-1 and PAR-2 double-deficient embryos with exencephaly (I and J) at E14.5. Exencephaly in 95% of the affected PAR-1 and PAR-2 double-deficient embryos was restricted to hindbrain region (HB, two-sided arrow in I) and extended to midbrain (MB-HB, two-sided arrow in J) in only 5% of the cases. (K and K9) Ventral (K) and dorsal (K9) view of the macroscopic appearance of HAI-2-deficient (Spint22/2) embryos at E9.5. Divergence of neural folds (arrows) and defects in neural tube closure extending from forebrain region to cervix are obvious. Open arrowheads show normal formation of medial hinge points. (L) Macroscopic appearance of a HAI-2-deficient embryo with exencephaly at E14.5. 90% of embryos presented with exencephaly that included at least midbrain and hindbrain regions of the developing cranium. (M) Histological appearance (nuclear fast red staining) of PAR-1 and PAR-2 double-deficient embryo with exencephaly at E9.5. Defined medial (arrow) and dorsolateral (arrowheads) hinge points are clearly visible. Scale bar: 150 mm. doi:10.1371/journal.pgen.1002937.g007

This finding is perplexing, as matriptase is well-established to be long been recognized in the coagulation, fibrinolytic, complement, able to auto-activate, as most clearly evidenced by the inability of and digestive systems. The current findings, thus, serve to recombinant matriptase protein with the catalytic triad serine underscore that our knowledge of the molecular workings of mutated to alanine (S805A matriptase) to undergo activation site membrane-anchored serine proteases is still fragmentary, due to cleavage [56]. Furthermore, prostasin shows no catalytic activity their quite recent emergence as a protease subfamily. towards peptide sequences derived from the prostasin pro-peptide The outcome of our epistasis analysis querying the contribution [57] and no reports of prostasin auto-activation have appeared to of proHGF, PAR-2, and ENaC to the prostasin and matriptase- date. Substantiating this finding, however, we found that prostasin dependent embryonic demise of HAI-1- and HAI-2-deficient mice efficiently activated the matriptase zymogen in a reconstituted cell- also was unanticipated. Each of the three proteins has been based assay. These findings are aligned with a recent study genetically validated as a substrate for either matriptase or showing that PAR-2 activation in some cultured cells, caused by prostasin in developmental or post-developmental processes, has exposure of cultured cells to exogenously added activated established functions in embryonic development, and is develop- prostasin, was blunted by a neutralizing antibody directed against mentally co-expressed with both proteases. Nevertheless, their matriptase [45], providing further evidence that complex and genetic elimination failed to prevent or alleviate any of the context-specific relationships between the two membrane-an- abnormalities caused by the loss of HAI-1 or HAI-2. Importantly, chored serine proteases may exist in vivo. Another important our analysis does not exclude that cleavage of either of the three finding relating to the developmental prostasin-matriptase cascade proteins must be suppressed by HAI-1 or HAI-2 at later stages of identified in this study emanated from our biochemical analysis of development that cannot be analyzed by the current experimental placental tissues, which revealed that activated forms of both approach. Also, the possibility that the lethality of HAI-1- or HAI- matriptase and prostasin were present in a complex with HAI-1 in 2-deficient embryos is caused by the simultaneous cleavage of placental tissues. This indicates that the regulation of the prostasin- more than one of these substrates cannot be formally excluded. It matriptase cascade by HAI-1 (and likely HAI-2) may occur by was particularly surprising that the neural tube defects associated controlling both prostasin and matriptase proteolytic activity. with HAI-2-deficiency were unrelated to either excessive or Furthermore, as both HAI-1 and HAI-2 are very promiscuous and reduced (through desensitization) PAR-2 activity, despite the display potent inhibitory activity towards a number of trypsin-like unequivocal contribution of PAR-2 signaling to neural tube serine proteases in vitro [58,59,60,61,62,63,64], it is throughout closure, and the wealth of strong circumstantial evidence that plausible that they may also regulate the activity of as yet prostasin and matriptase contribute to PAR-2 activation in this unidentified proteases that act upstream of, downstream of or process [45,65]. Equally surprising in this regard, the combined between prostasin and matriptase. Such profound complexities in loss of PAR-1 and matriptase failed to cause the neural tube zymogen activation relationships between trypsin-like serine closure defects observed in PAR-1 and PAR-2 double-deficient proteases and for the promiscuity of their cognate inhibitors have embryos, showing that matriptase is not essential for initiation of physiological PAR-2 signaling during neural tube formation. Previous analysis has identified five other membrane-anchored Table 2. Comparison of morphologic features of neural tube 2 2 2 2 2 2 serine proteases and fourteen secreted trypsin-like serine proteases defects observed in Spint2 / and F2r / ; F2rl1 / mice. that are expressed during neural tube formation, some of which can activate PAR-2 in cell-based assays [43,45]. It is therefore

Spint22/2 F2r2/2; F2rl12/2 possible that the prostasin-matriptase cascade does contribute to PAR-2 activation during neural tube closure, but sufficient Process residual activation of PAR-2 by other developmentally co- Formation of medial hinge point 100% (28/28) 100% (38/38) expressed serine proteases takes place in its absence to allow for Formation of dorsolateral hinge points 3% (1/31) 97% (28/29) completion of this developmental process. Nevertheless, the careful Completion of C1 fusion 100% (28/28) 100% (66/66) comparison of the morphology of neural tube defects in PAR-1 and PAR-2, and HAI-2-double deficient embryos performed here Completion of C2 fusion 90% (27/30) 100% (66/66) revealed distinct differences in terms of their anatomical location Extent of neural tube defect and the stage of developmental failure. Taken together, these data Hindbrain only 10% (3/29) 95% (37/39) suggest that promotion of neural tube closure by HAI-2 Hindbrain and midbrain 59% (17/29) 5% (2/39) suppression of the prostasin-matriptase cascade and promotion Forebrain to cervix 31% (9/29) 0% (0/39) of neural tube closure by PAR-1/PAR-2 signaling may be Craniorachischisis 0% (0/29) 0% (0/39) temporally and spatially distinct morphogenic processes. In conclusion, this study identifies a prostasin-matriptase cell doi:10.1371/journal.pgen.1002937.t002 surface protease cascade whose activity must be suppressed by

PLOS Genetics | www.plosgenetics.org 14 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

HAI-1 and HAI-2 to enable early embryonic ectoderm formation, then homogenized in ice-cold 50 mM Tris/HCl, pH 8.0; 1% NP- placental morphogenesis, and neural tube closure. 40; 500 mM NaCl buffer and incubated on ice for 10 minutes. The lysates were centrifuged at 20,000 g for 10 min at 4uCto Materials and Methods remove the tissue debris and the supernatant was used for further analysis as described below. Mouse strains All experiments were performed in an Association for Assess- Detection of active matriptase and prostasin in mouse ment and Accreditation of Laboratory Animal Care International- embryonic tissues accredited vivarium following Standard Operating Procedures. The Lysates from two placentae of the same genotype were studies were approved by the NIDCR Institutional Animal Care combined and pre-incubated with 100 ul GammaBind G and Use Committee. All studies were littermate controlled. 2 2 2 2 2 2 2 2 2 2 2 2 Sepharose beads (GE Healthcare Bio-Sciences, Uppsala, Sweden) Spint1 / , Spint2 / , St14 / , Hgfr / , F2r / , F2rl1 / , 2 2 for 30 minutes at 4uC with gentle agitation. The samples were Scnn1a / , and Prss8fr/fr mice have been described 2 2 spun at 5,000 g for 1 min to remove the beads, and the [38,40,43,49,66,67,68]. Prostasin-deficient (Prss8 / ) mice were supernatant was then incubated with 5 mg goat anti-mouse HAI- generated by standard blastocyst injection of C57BL/6J-derived 1 antibody (R&D Systems) and 100 ul of GammaBind G embryonic stem cells carrying a gene trap insertion in the Prss8 gene Sepharose beads for 3 hours at 4uC. The samples were spun at (clone IST10122F12, Texas A&M Institute for Genomic Research, 5,000 g for 1 min, the supernatant was removed, and the beads College Station, TX). were washed 3 times with 1 ml ice-cold 50 mM Tris/HCl, pH 8.0; 1% NP-40; 500 mM NaCl buffer. The beads were then Extraction of embryonic and perinatal tissues mixed with 30 ul of 16SDS loading buffer (Invitrogen, Carlsbad, Breeding females were checked for vaginal plugs in the morning CA) with 0.25 M b-mercaptoethanol, incubated for 5 min at and the day on which the plug was found was defined as the first 99uC, and cooled on ice for 2 minutes. The samples were spun at day of pregnancy (E0.5). Pregnant females were euthanized in the 5,000 g for 1 min and the released proteins were resolved by SDS- mid-day at designated time points by CO2 asphyxiation. Embryos PAGE (4–12% polyacrylamide gel) and analyzed by western blot were extracted by Caesarian section and the individual embryos using mouse anti-human prostasin (1:250, BD Transduction Labs) and placentae were dissected and processed. Visceral yolk sacs of or sheep anti-human matriptase (1:500, R&D Systems) primary individual embryos were washed twice in phosphate buffered antibodies and goat anti-mouse (DakoCytomation) or donkey anti- saline, subjected to genomic DNA extraction and genotyped by sheep (Sigma-Aldrich) secondary antibodies (both 1:1000) conju- PCR (see Table S1 for primer sequences). Newborn pups were gated to alkaline phosphatase, and visualized using nitro-blue u euthanized by CO2 inhalation at 0 C. For histological analysis, the tetrazolium and 5-bromo-4-chloro-39-indolyphosphate. embryos and newborn pups were fixed for 18–20 hrs in 4% paraformaldehyde (PFA) in PBS, processed into paraffin, sec- Amiloride injections tioned, and stained with hematoxylin and eosin (H&E), or used for Five ug per g of body weight of amiloride (Sigma-Aldrich) in immunohistochemistry as described below. For histomorphomet- ric analysis of placental labyrinth, the midline cross sections of 10% DMSO in PBS was administered to pregnant females by plancetal tissues were stained with H&E and the thickness of the intraperitoneal injection every 24 hours starting on E5.5. Embryos labyrinth was determined as the maximum perpendicular distance were extracted on E9.5 by Caesarian section and genotyped as of fetal vessel from the chorionic trophoblast layer. described, and scored for neural tube closure defects.

Immunohistochemistry Generation of soluble recombinant wildtype, catalytically Antigens from 5 mm paraffin sections were retrieved by inactive S238A, and V170D prostasin zymogens incubation for 10 min at 37uC with 10 mg/ml proteinase K The generation of pIRES2-EGFP-prostasin has been described (Fermentas, Hanover, MD) for HAI-1 staining, or by incubation [26]. Substitution of the native prostasin activation site (APQAR) for 20 min at 100uC in 0.01 M sodium citrate buffer, pH 6.0, for by the enteropeptidase-dependent cleavage site (DDDDK), and all other antigens. The sections were blocked with 2% bovine either the S238A or V170D point mutations were introduced serum albumin in PBS, and incubated overnight at 4uC with using the QuickChange Kit (Stratagene, La Jolla, CA) and the 9 rabbit anti-human CD31 (1:100, Santa Cruz Biotechnology, Santa following primers, respectively: 5 -GCTCCCTGCGGTGTGG- Cruz, CA), goat anti-mouse HAI-1 (1:200, R&D Systems, CCCCCCAAGCACGCATCACAGGTGGCAGC-39,59-GAC- Minneapolis, MN), mouse anti-human prostasin (1:200, BD GCCTGCCAGGGTGACGCTGGGGGCCCACTCTCCTGC- Transduction Laboratories, San Jose, CA), sheep anti-human 39, and 59-GGCCTCCACTGCACTGACACTGGCTGGGGT- matriptase (1:200, R&D Systems) or ENaCc subunit (1:100, CAT-39. Successful mutagenesis was verified by sequencing of Sigma-Aldrich, St. Louis, MO) primary antibodies. Bound both strands of the resulting cDNA. Expression plasmids carrying antibodies were visualized using biotin-conjugated anti-mouse, - individual mutations were transiently transfected into HEK-293T rabbit, -sheep or -goat secondary antibodies (all 1:400, Vector cells using Turbofect (Fermentas). The cells were grown for two Laboratories, Burlingame, CA) and a Vectastain ABC kit (Vector days and soluble recombinant prostasin was prepared by Laboratories) using 3,39-diaminobenzidine as the substrate (Sig- treatment of cells with phosphatidylinositol-specific phospholipase ma-Aldrich). All microscopic images were acquired on an C (Sigma-Aldrich) as described previously [26]. Olympus BX40 microscope using an Olympus DP70 digital camera system (Olympus, Melville, NY). Determination of enzymatic activity of recombinant prostasin variants Protein extraction from mouse tissues Recombinant wildtype, V170D Frizzy or catalytically inactive Placentae were extracted from embryos at E10.5 or E11.5. The S238A prostasin zymogen variants were first incubated with 5.1 U embryonic portion of each placenta was manually separated from recombinant bovine enteropeptidase (Novagen, Cambridge, MA) maternal decidua using a dissecting microscope. The tissues were overnight at 37uC in enterokinase buffer (Novagen). Following

PLOS Genetics | www.plosgenetics.org 15 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade enteropeptidase removal using the Enterokinase Removal Kit Supporting Information (Sigma-Aldrich), the protein concentration was estimated by western blot of serially diluted proteins using a reference with Figure S1 (A) Western blot detection of protein nexin-1 (PN-1). known protein concentration. For substrate hydrolysis assays, the Wildtype zymogen (lanes 1 and 2), activated wildtype (lanes 3 and activated prostasin variants (62.5 nM) were incubated with the 4), V170D (frizzy) zymogen (lanes 5 and 6), activated V170D fluorogenic substrate pERTKR-AMC (50 mM final concentration) (lanes 7 and 8), S238A zymogen (lanes 9 and 10), and activated (R&D systems) at 37uC in 50 mM NaCl, 50 mM Tris-HCl S238A (lanes 11 and 12) prostasin variants were incubated with (lanes 2, 4, 6, 8, 10, and 12) or without (lanes 1, 3, 5, 7, 9, and 11) pH 8.8, 0.01% Tween-20 buffer, and the fluorescence was 250 ng of recombinant human PN-1. Position of PN-1, and measured using a Wallac plate reader (Perkin Elmer, Waltham, predicted position of prostasin/PN-1 complexes (not detected by MA). Each measurement was performed in triplicate. For serpin anti-PN-1 antibody presumably due to significant molecular complex formation, prostasin variants were diluted in 50 mM rearrangement of PN-1 in the complex with the protease) are Tris-HCl, pH 9.0, 50 mM NaCl, 0.01% Tween 20 to a final indicated. Positions of molecular weight markers (kDa) are shown concentration of 150 nM, incubated with 250 ng recombinant on left. (B) Western blot detection of active matriptase in the fetal human protease nexin-1 (PN-1) (R&D Systems) for 1 h at 37uC, part of the E11.5 placentas of one matriptase-deficient 2 2 + + + + + + and analyzed using 12% reducing SDS-PAGE and western (St14 / ;Prss8 / ) (lane 1), three wildtype (Prss8 / and St14 / ) + + 2 2 blotting, using a monoclonal anti-prostasin antibody (BD Trans- (lanes 2,3, and 4), and three prostasin-deficient (St14 / ;Prss8 / ) duction Laboratories). (lanes 5, 6, and 7) embryos after anti-HAI-1 immunoprecipitation. A 30 kDa band representing the active serine protease domain of Matriptase activation and SRE–luciferase assay matriptase (Mat SPD) was present in extracts from wildtype, but HEK 293 cells were plated in 24-well plates and grown in not in matriptase- or prostasin-deficient placentas. (C) Distribution + 2 DMEM supplemented with 10% FBS for 24 h. Cells were co- of Spint2 genotypes at E9.5 in offspring from interbred Spint2 / transfected with pSRE-firefly luciferase (50 ng), pRL-Renilla breeding pairs treated with the ENaC inhibitor, amiloride, at 2 2 luciferase (20 ng), pcDNA 3.1 Par2 (100 ng) (Missouri S&T E5.5–8.5. No Spint2 / embryos were observed. cDNA Resource Center) using Lipofectamine and Plus reagent (TIF) (Invitrogen), pcDNA 3.1 expression vectors containing wildtype Table S1 Sequences of PCR primers used for mouse genotyp- human matriptase or catalytically dead matriptase (S805A), full ing. length human HAI-1 [69] and empty pcDNA 3.1 vector to (DOCX) equalize the total amount of transfected DNA. After 36 h the cells were serum starved over night and then stimulated with 100 nM Acknowledgments recombinant human soluble prostasin (R&D) or vehicle for 6 h. Cell were lysed and luciferase activity was determined using the We thank Dr. Mary Jo Danton for critically reviewing this manuscript. dual luciferase assay kit (Promega, Madison, WI) according to the Histology was performed by Histoserv, Germantown, Maryland, United manufacturer’s instructions. Chemiluminiscence was measured States of America. using Microtiter Plate Luminometer (Dynex Technologies, Chantilly, VA) and the SRE activation was determined as the Author Contributions ratio of firefly to Renilla luciferase counts. The assay was Conceived and designed the experiments: RS THB JSG. Performed the performed two times in duplicates. experiments: RS KUS PK NAS SG EC KKH SF. Analyzed the data: RS KUS PK NAS EC THB KKH. Contributed reagents/materials/analysis tools: SG LKV EH. Wrote the paper: RS THB.

References 1. Szabo R, Bugge TH (2011) Membrane anchored serine proteases in cell and 10. List K, Kosa P, Szabo R, Bey AL, Wang CB, et al. (2009) Epithelial integrity is developmental biology. Annu Rev Cell and Developmental Biology 27: 213–235. maintained by a matriptase-dependent proteolytic pathway. Am J Pathol 175: 2. Bugge TH, Antalis TM, Wu Q (2009) Type II transmembrane serine proteases. 1453–1463. J Biol Chem 284: 23177–23181. 11. Zheng XL, Kitamoto Y, Sadler JE (2009) Enteropeptidase, a type II 3. Antalis TM, Buzza MS, Hodge KM, Hooper JD, Netzel-Arnett S (2010) The transmembrane serine protease. Front Biosci (Elite Ed) 1: 242–249. cutting edge: membrane-anchored serine protease activities in the pericellular 12. Dries DL, Victor RG, Rame JE, Cooper RS, Wu X, et al. (2005) Corin gene microenvironment. Biochem J 428: 325–346. minor allele defined by 2 missense mutations is common in blacks and associated 4. Guipponi M, Tan J, Cannon PZ, Donley L, Crewther P, et al. (2007) Mice with high blood pressure and hypertension. Circulation 112: 2403–2410. deficient for the type II transmembrane serine protease, TMPRSS1/hepsin, 13. Wang W, Liao X, Fukuda K, Knappe S, Wu F, et al. (2008) Corin variant exhibit profound hearing loss. Am J Pathol 171: 608–616. associated with hypertension and cardiac hypertrophy exhibits impaired 5. Scott HS, Kudoh J, Wattenhofer M, Shibuya K, Berry A, et al. (2001) Insertion of zymogen activation and natriuretic peptide processing activity. Circ Res 103: beta-satellite repeats identifies a transmembrane protease causing both congenital 502–508. and childhood onset autosomal recessive deafness. Nat Genet 27: 59–63. 14. Yan W, Wu F, Morser J, Wu Q (2000) Corin, a transmembrane cardiac serine 6. Fasquelle L, Scott HS, Lenoir M, Wang J, Rebillard G, et al. (2010) Tmprss3, a protease, acts as a pro-atrial natriuretic peptide-converting enzyme. Proc Natl transmembrane serine protease deficient in human DFNB8/10 deafness, is Acad Sci U S A 97: 8525–8529. critical for Cochlear hair cell survival at the onset of hearing. J Biol Chem 286: 15. Du X, She E, Gelbart T, Truksa J, Lee P, et al. (2008) The serine protease 17383–17397. TMPRSS6 is required to sense iron deficiency. Science 320: 1088–1092. 7. List K, Szabo R, Wertz PW, Segre J, Haudenschild CC, et al. (2003) Loss of 16. Finberg KE, Heeney MM, Campagna DR, Aydinok Y, Pearson HA, et al. proteolytically processed filaggrin caused by epidermal deletion of Matriptase/ (2008) Mutations in TMPRSS6 cause iron-refractory iron deficiency anemia MT-SP1. J Cell Biol 163: 901–910. (IRIDA). Nat Genet 40: 569–571. 8. Leyvraz C, Charles RP, Rubera I, Guitard M, Rotman S, et al. (2005) The 17. Netzel-Arnett S, Bugge TH, Hess RA, Carnes K, Stringer BW, et al. (2009) The epidermal barrier function is dependent on the serine protease CAP1/Prss8. glycosylphosphatidylinositol-anchored serine protease PRSS21 (testisin) imparts J Cell Biol 170: 487–496. murine epididymal sperm cell maturation and fertilizing ability. Biol Reprod 81: 9. Buzza MS, Netzel-Arnett S, Shea-Donohue T, Zhao A, Lin CY, et al. (2010) 921–932. Membrane-anchored serine protease matriptase regulates epithelial barrier 18. Kawano N, Kang W, Yamashita M, Koga Y, Yamazaki T, et al. (2010) Mice formation and permeability in the intestine. Proc Natl Acad Sci U S A 107: lacking two sperm serine proteases, ACR and PRSS21, are subfertile, but the 4200–4205. mutant sperm are infertile in vitro. Biol Reprod 83: 359–369.

PLOS Genetics | www.plosgenetics.org 16 August 2012 | Volume 8 | Issue 8 | e1002937 Embryonic Cell Surface Serine Protease Cascade

19. Heinz-Erian P, Muller T, Krabichler B, Schranz M, Becker C, et al. (2009) 43. Szabo R, Hobson JP, Christoph K, Kosa P, List K, et al. (2009) Regulation of Mutations in SPINT2 Cause a Syndromic Form of Congenital Sodium cell surface protease matriptase by HAI2 is essential for placental development, Diarrhea. Am J Hum Genet, 84: 188–196. neural tube closure and embryonic survival in mice. Development 136: 2653– 20. Szabo R, Kosa P, List K, Bugge TH (2009) Loss of matriptase suppression 2663. underlies spint1 mutation-associated ichthyosis and postnatal lethality. 44. Mitchell KJ, Pinson KI, Kelly OG, Brennan J, Zupicich J, et al. (2001) Am J Pathol 174: 2015–2022. Functional analysis of secreted and transmembrane proteins critical to mouse 21. Nagaike K, Kawaguchi M, Takeda N, Fukoshima T, Sawaguchi A, et al. (2008) development. Nat Genet 28: 241–249. Defect of hepatocyte growth factor activator inhibitor type 1/serine protease 45. Camerer E, Barker A, Duong DN, Ganesan R, Kataoka H, et al. (2010) Local inhibior, Kunitz type 1 (Hai-1/Spint1) leads to ichthyosis-like condition and protease signaling contributes to neural tube closure in the mouse embryo. Dev abnormal hair development in mice. American Journal of Pathology 173: 1–12. Cell 18: 25–38. 22. Kawaguchi M, Takeda N, Hoshiko S, Yorita K, Baba T, et al. (2011) 46. Szabo R, Rasmussen AL, Moyer AB, Kosa P, Schafer J, et al. (2011) c-Met- Membrane-bound serine protease inhibitor HAI-1 is required for maintenance induced epithelial carcinogenesis is initiated by the serine protease matriptase. of intestinal epithelial integrity. Am J Pathol 179: 1815–1826. Oncogene 30: 2003–2016. 23. List K, Currie B, Scharschmidt TC, Szabo R, Shireman J, et al. (2007) 47. Bladt F, Riethmacher D, Isenmann S, Aguzzi A, Birchmeier C (1995) Essential Autosomal ichthyosis with hypotrichosis syndrome displays low matriptase role for the c-met receptor in the migration of myogenic precursor cells into the proteolytic activity and is phenocopied in ST14 hypomorphic mice. J Biol Chem limb bud. Nature 376: 768–771. 282: 36714–36723. 48. Uehara Y, Minowa O, Mori C, Shiota K, Kuno J, et al. (1995) Placental defect 24. Alef T, Torres S, Hausser I, Metze D, Tursen U, et al. (2009) Ichthyosis, and embryonic lethality in mice lacking hepatocyte growth factor/scatter factor. Follicular Atrophoderma, and Hypotrichosis Caused by Mutations in ST14 Is Nature 373: 702–705. Associated with Impaired Profilaggrin Processing. J Invest Dermatol 129: 862– 49. Spacek DV, Perez AF, Ferranti KM, Wu LK, Moy DM, et al. (2010) The mouse 869. frizzy (fr) and rat ‘hairless’ (frCR) mutations are natural variants of protease 25. Chen YW, Wang JK, Chou FP, Chen CY, Rorke EA, et al. (2010) Regulation of serine S1 family member 8 (Prss8). Exp Dermatol 19: 527–532. the matriptase-prostasin cell surface proteolytic cascade by hepatocyte growth 50. Hedstrom L (2002) Serine protease mechanism and specificity. Chem Rev 102: factor activator inhibitor-1 during epidermal differentiation. J Biol Chem 285: 4501–4524. 31755–31762. 51. Rossier BC, Stutts MJ (2009) Activation of the epithelial sodium channel (ENaC) 26. Netzel-Arnett S, Currie BM, Szabo R, Lin CY, Chen LM, et al. (2006) Evidence by serine proteases. Annu Rev Physiol 71: 361–379. for a matriptase-prostasin proteolytic cascade regulating terminal epidermal 52. Planes C, Randrianarison NH, Charles R-P, Frateschi S, Cluzeeaud D, et al. differentiation. J Biol Chem 281: 32941–32945. (2009) ENaC-mediated alveolar fluid clearance and lung fluid balance depend 27. Kim C, Lee HS, Lee D, Lee SD, Cho EG, et al. (2011) Epithin/PRSS14 on the channel-activating protease 1. EMBO Molecular Medicine 2: 26–37. proteolytically regulates angiopoietin receptor Tie2 during transendothelial 53. Miller TA, Scott WJ, Jr. (1992) Abnormalities in ureter and kidney development migration. Blood 117: 1415–1424. in mice given acetazolamide-amiloride or dimethadione (DMO) during 28. Ustach CV, Huang W, Conley-LaComb MK, Lin CY, Che M, et al. (2010) A embryogenesis. Teratology 46: 541–550. novel signaling axis of matriptase/PDGF-D/ss-PDGFR in human prostate 54. Canessa CM, Schild L, Buell G, Thorens B, Gautschi I, et al. (1994) Amiloride- cancer. Cancer Res 70: 9631–9640. sensitive epithelial Na+ channel is made of three homologous subunits. Nature 29. Lee SL, Dickson RB, Lin CY (2000) Activation of hepatocyte growth factor and 367: 463–467. urokinase/plasminogen activator by matriptase, an epithelial membrane serine 55. Hummler E, Barker P, Gatzy J, Beermann F, Verdumo C, et al. (1996) Early protease. J Biol Chem 275: 36720–36725. death due to defective neonatal lung liquid clearance in alpha-ENaC-deficient 30. Owen KA, Qiu D, Alves J, Schumacher AM, Kilpatrick LM, et al. (2010) mice. Nat Genet 12: 325–328. Pericellular activation of hepatocyte growth factor by the transmembrane serine 56. Oberst MD, Williams CA, Dickson RB, Johnson MD, Lin CY (2003) The proteases matriptase and hepsin, but not by the membrane-associated protease activation of matriptase requires its noncatalytic domains, serine protease uPA. Biochem J 426: 219–228. domain, and its cognate inhibitor. J Biol Chem 278: 26773–26779. 31. Takeuchi T, Harris JL, Huang W, Yan KW, Coughlin SR, et al. (2000) Cellular 57. Shipway A, Danahay H, Williams JA, Tully DC, Backes BJ, et al. (2004) localization of membrane-type serine protease 1 and identification of protease- Biochemical characterization of prostasin, a channel activating protease. activated receptor-2 and single-chain urokinase-type plasminogen activator as Biochem Biophys Res Commun 324: 953–963. substrates. J Biol Chem 275: 26333–26342. 32. Bhatt AS, Welm A, Farady CJ, Vasquez M, Wilson K, et al. (2007) Coordinate 58. Denda K, Shimomura T, Kawaguchi T, Miyazawa K, Kitamura N (2002) expression and functional profiling identify an extracellular proteolytic signaling Functional characterization of Kunitz domains in hepatocyte growth factor pathway. Proc Natl Acad Sci USA 104: 5771–5776. activator inhibitor type 1. J Biol Chem 277: 14053–14059. 33. Bhatt AS, Erdjument-Bromage H, Tempst P, Craik CS, Moasser MM (2005) 59. Kirchhofer D, Peek M, Lipari MT, Billeci K, Fan B, et al. (2005) Hepsin Adhesion signaling by a novel mitotic substrate of src kinases. Oncogene 24: activates pro-hepatocyte growth factor and is inhibited by hepatocyte growth 5333–5343. factor activator inhibitor-1B (HAI-1B) and HAI-2. FEBS Lett 579: 1945–1950. 34. Vuagniaux G, Vallet V, Jaeger NF, Hummler E, Rossier BC (2002) Synergistic 60. Kirchhofer D, Peek M, Li W, Stamos J, Eigenbrot C, et al. (2003) Tissue activation of ENaC by three membrane-bound channel-activating serine expression, protease specificity, and Kunitz domain functions of hepatocyte proteases (mCAP1, mCAP2, and mCAP3) and serum- and glucocorticoid- growth factor activator inhibitor-1B (HAI-1B), a new splice variant of HAI-1. regulated kinase (Sgk1) in Xenopus Oocytes. J Gen Physiol 120: 191–201. J Biol Chem 278: 36341–36349. 35. Andreasen D, Vuagniaux G, Fowler-Jaeger N, Hummler E, Rossier BC (2006) 61. Herter S, Piper DE, Aaron W, Gabriele T, Cutler G, et al. (2005) Hepatocyte Activation of epithelial sodium channels by mouse channel activating proteases growth factor is a preferred in vitro substrate for human hepsin, a membrane- (mCAP) expressed in Xenopus oocytes requires catalytic activity of mCAP3 and anchored serine protease implicated in prostate and ovarian cancers. Biochem J mCAP2 but not mCAP1. J Am Soc Nephrol 17: 968–976. 390: 125–136. 36. Jin X, Yagi M, Akiyama N, Hirosaki T, Higashi S, et al. (2006) Matriptase 62. Fan B, Wu TD, Li W, Kirchhofer D (2005) Identification of hepatocyte growth activates stromelysin (MMP-3) and promotes tumor growth and angiogenesis. factor activator inhibitor-1B as a potential physiological inhibitor of prostasin. Cancer Sci 97: 1327–1334. J Biol Chem 280: 34513–34520. 37. Kilpatrick LM, Harris RL, Owen KA, Bass R, Ghorayeb C, et al. (2006) 63. Delaria KA, Muller DK, Marlor CW, Brown JE, Das RC, et al. (1997) Initiation of plasminogen activation on the surface of monocytes expressing the Characterization of placental bikunin, a novel human serine protease inhibitor. type II transmembrane serine protease matriptase. Blood 108: 2616–2623. J Biol Chem 272: 12209–12214. 38. List K, Haudenschild CC, Szabo R, Chen W, Wahl SM, et al. (2002) 64. Qin L, Denda K, Shimomura T, Kawaguchi T, Kitamura N (1998) Functional Matriptase/MT-SP1 is required for postnatal survival, epidermal barrier characterization of Kunitz domains in hepatocyte growth factor activator function, hair follicle development, and thymic homeostasis. Oncogene 21: inhibitor type 2. FEBS Lett 436: 111–114. 3765–3779. 65. Copp AJ, Greene ND (2010) Defining a PARticular pathway of neural tube 39. List K, Szabo R, Molinolo A, Nielsen BS, Bugge TH (2006) Delineation of closure. Dev Cell 18: 1–2. matriptase protein expression by enzymatic gene trapping suggests diverging 66. Hummler E, Merillat AM, Rubera I, Rossier BC, Beermann F (2002) roles in barrier function, hair formation, and squamous cell carcinogenesis. Conditional gene targeting of the Scnn1a (alphaENaC) gene locus. Genesis Am J Pathol 168: 1513–1525. 32: 169–172. 40. Szabo R, Molinolo A, List K, Bugge TH (2007) Matriptase inhibition by 67. Connolly AJ, Ishihara H, Kahn ML, Farese RV, Jr., Coughlin SR (1996) Role of hepatocyte growth factor activator inhibitor-1 is essential for placental the thrombin receptor in development and evidence for a second receptor. development. Oncogene 26: 1546–1556. Nature 381: 516–519. 41. Fan B, Brennan J, Grant D, Peale F, Rangell L, et al. (2007) Hepatocyte growth 68. Lindner JR, Kahn ML, Coughlin SR, Sambrano GR, Schauble E, et al. (2000) factor activator inhibitor-1 (HAI-1) is essential for the integrity of basement Delayed onset of inflammation in protease-activated receptor-2- deficient mice. membranes in the developing placental labyrinth. Dev Biol 303: 222–230. J Immunol 165: 6504–6510. 42. Tanaka H, Nagaike K, Takeda N, Itoh H, Kohama K, et al. (2005) Hepatocyte 69. Oberst MD, Chen LL, Kiyomiya K-I, Williams CA, Lee M-S, et al. (2005). Growth Factor Activator Inhibitor Type 1 (HAI-1) Is Required for Branching HAI-1 regulates activation and expression of matriptase, a membrane-bound Morphogenesis in the Chorioallantoic Placenta. Mol Cell Biol 25: 5687–5698. serine protease. Am J Physiol Cell Physiol 289: C462–C470.

PLOS Genetics | www.plosgenetics.org 17 August 2012 | Volume 8 | Issue 8 | e1002937 Paper III

Novel assay for detection of active matriptase Sine Godiksen1,2, and Lotte K. Vogel2

More authors to be added

1Department of Biology, University of Copenhagen, Copenhagen, Denmark 2Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark

Manuscript in progress

Novel assay for detection of active matriptase

Sine Godiksen1,2 and Lotte K. Vogel2 Other authors to be added

1Department of Biology, University of Copenhagen, Copenhagen, Denmark 2Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark

Matriptase is a member of the family of type II transmembrane serine proteases that is crucial for development and maintenance of several epithelial tissues. Matriptase is synthesized as a zymogen that is converted by activation site cleavage to a disulfide-linked active form. This occurs by autoactivation, which has led to the hypothesis that matriptase functions at the pinnacle of several protease induced signal cascades involved in epithelial tissue formation and maintenance. Matriptase can be detected in either its zymogen form or in a complex with its cognate inhibitor hepatocyte growth factor activator inhibitor 1 (HAI-1) in cells, whereas the active form has been difficult to detect. In this study, we have established an assay to detect active matriptase using a peptide substrate-based chloromethyl ketone (Cmk) inhibitor. Our assay is based on the assumption that matriptase able to react with a Cmk peptide inhibitor is equivalent to active matriptase. In order to detect covalently Cmk peptide-bound matriptase, we designed a Cmk peptide inhibitor with a biotin moiety in the N-terminal of the small peptide that allows for streptavidin pull-down and subsequent analysis by Western blotting. This study presents a novel assay for detection active matriptase in living cells. The assay can be easily applied in other cell systems and other species.

Abbreviations used: Cmk, chloromethyl ketone; HAI-1, hepatocyte growth factor activator inhibitor-1; RT, room temperature; SPD: serine protease domain

Introduction

Matriptase (also known as MT-SP1, epithin, TADG-15 and SNC19) is a member of the matriptase subfamily of type II transmembrane serine proteases. This protease is expressed in most epithelial cells and has pleiotropic roles in mouse epithelial homeostasis [1-5]. In humans, mutations in the ST14 gene expressing matriptase, is the underlying cause of congenital ichthyosis [6-8]. More recently, matriptase was identified as an initiator of the runaway kallikrein protease cascade leading to an autosomal recessive form of ichthyosis referred to as Netherton syndrome [9].

Ablation of the ST14 gene in mice shows that the protease is essential for postnatal survival. Matriptase knock-out mice display defects in epidermal barrier function, hair follicle development, pro-filaggrin processing, and in thymic homeostasis and die within 48 hrs of birth [5;10]. Tissue-specific or postnatal ablation of matriptase in mice cause severe organ dysfunction, generalized epithelial demise, and demonstrate a principal and global function of matriptase in promoting the formation of paracellular permeability barriers in simple and stratified epithelia [4].

Important knowledge about matriptase has also been derived from ablation of the physiological inhibitors of matriptase; hepatocyte growth factor activator inhibitor-1 (HAI-1) and -2 (HAI-2). Genetic inactivation of either HAI-1 or HAI-2 leads to failure of placental labyrinth formation [11;12]. This defect can be completely rescued by simultaneously reducing or eliminating matriptase expression [12;13]. HAI-1 deficiency in both zebrafish and chimeric mice leads to fatal defects in epidermal integrity [14-17]. Simultaneous ablation of matriptase in both model systems completely eliminates these defects [14;16]. Additionally, ablation of HAI-2 in mice leads to defects in neural tube closure that are partially rescued by genetic inactivation of matriptase [12;18].

Deregulated matriptase can promote carcinogenesis, as a modest overexpression of wild-type (WT) matriptase in the epidermis of transgenic mice is sufficient to induce spontaneous squamous cell carcinoma formation and to strongly potentiate chemical-induced skin carcinogenesis. The oncogenic effect is mediated through activation of the c-Met-Akt-mTor pathway by matriptase- dependent activation of the cMet ligand; pro-HGF/SF [5;19;20]. A simultaneous increase in expression of HAI-1 completely negates the oncogenic effect of matriptase overexpression [19].

In colitis-associated colorectal cancer, matriptase was recently assigned as a critical tumor- suppressor gene. Selective genetic inactivation of matriptase in the intestinal epithelium lead to a compromised intestinal barrier function associated with chronic inflammation and finally formation of colon adenocarcinomas in the mouse gastrointestinal tract [21].

Matriptase is synthesized as an 855 amino acid single chain zymogen that undergoes two successive cleavages to become active. First, matriptase is cleaved in the sea urchin sperm protein, enteropeptidase, and agrin (SEA) domain by non-enzymatic hydrolysis and subsequently proteolytically processed at its canonical activation site motif after Arg614 in the serine protease domain. The serine protease domain (SPD) remains attached to the stem domain by a disulfide bridge. Shortly hereafter matriptase is inhibited by HAI-1 which leaves a short window of activity for the disulfide-linked active form of matriptase [22-24]. Using surface biotinylation, we have previously shown that matriptase undergoes activation site cleavage on the plasma membrane, and shortly after matriptase is endocytosed in complex with HAI-1 [25].

Despite our knowledge about the important roles played by matriptase, both under normal physiological conditions as well as in a range of pathological conditions, our understanding of matriptase and the mechanisms regulating its proteolytic activity is poor. Many epithelial cells express matriptase that can be detected either in the zymogen or the HAI-1-complexed form [1;2;22;23;26]. However, the active non-inhibited form, which is the presumably biological active form of matriptase, has been difficult to detect.

In the present study, we have established an assay to detect active matriptase based on the assumption that active matriptase is equivalent to matriptase capable of binding an inhibitor. The assay employs a chloromethyl ketone (Cmk) peptide inhibitor which combined with Western blotting specifically detects active matriptase.

Materials and methods

Chromogenic assay 0.2 μM matriptase SPD was prepared with 20 mM HEPES pH 7.4, 140 mM NaCl supplemented with 0.1% BSA (Sigma-Aldrich) in wells of a 96 well plate. Stocks of varying concentrations (5 nM or 50 μM) of biotin-RQRR-Cmk peptide (American Peptide) were incubated in the same buffer at 37°C for up to 3 hours. At specific time points 100 μl of inhibitor solution was added to the well containing the indicated concentration of the active catalytic domain of matriptase [27]. Following additional 10 min incubation at 37°C 10 μl 6.3 mM chromogenic substrate H-D-Isoleucil-L-prolyl-L- arginine-p-nitroaniline (cat. no. S2288, Chromogenix) was added and substrate conversion was followed in a standard plate reader by 30 min continuous measurements of the absorbance at 405 nm at 37°C. The rate of substrate turnover was determined from the color development resulting from a pseudo first order reaction due to a substrate concentration far greater than the expected nM range of protease. To test pH effects, HEPES was replaced by 20 mM citric acid buffer pH 6.0.

Cell culture Caco-2 cells were grown in minimal essential medium supplemented with 2 mM L-glutamine, 20% fetal bovine serum (Gibco), 1 x non essential amino acids, 100 units/ml penicillin and 100 μg/ml 6 streptomycin (Invitrogen) at 37°C in an atmosphere of 5% CO2. For all experiments, 1-2 × 10 cells were seeded into 35 mm tissue culture plates or 0.4 μm-pore-size 24 mm Transwell® filters (Corning) allowing separate access to the apical and the basolateral plasma membrane. The cell culture medium was changed every day. Cells were grown until day 11 days post confluence, as indicated by the tightness of the cell monolayer for filter grown cells, before they were used in experiments. The tightness of filter-grown cells was assayed by filling the inner chamber to the brim and allowing it to equilibrate overnight.

Isolation and short-term culture of primary keratinocytes from newborn mice Epidermis was isolated from newborn mice (d1-2) and grown in culture as described in [28]. In short, newborn pups were euthanized by decapitation and the torso was submerged in betadine and ethanol to sterilize the skin. The skin was incubated in 0.25% trypsin w/o EDTA (Sigma-

Aldrich) o/N at 4°C. The dermal portion was discarded and epidermis was minced to release keratinocytes. The epidermis was resuspended in 45 μM Ca2+/10%FBS/Keratinocyte-SFM (Invitrogen) media and was filtered through a 100 μm cell strainer and centrifuged to remove stratum corneum pieces. The cell pellet was resuspended in low calcium medium (45 μM Ca2+/Keratinocyte-SFM media) and plated on collagen (BD Biosciences) coated culture plates. Cell were grown in low calcium medium to sub-confluence and cell culture medium was changed every second day.

Labeling with biotin-Arg-Gln-Arg-Arg-chloromethyl ketone (biotin-RQRR-Cmk) peptide inhibitor and S-NHS-SS-biotin Cells were washed twice; filter grown Caco-2 cells with PBS++ (PBS supplemented with 0.7 mM

CaCl2 and 0.25 mM MgCl2) and primary murine keratinocytes with PBS. For labeling of active matriptase, cells were incubated with 50 μM biotin-RQRR-Cmk (American Peptide) in serum-free

MEM eagle with Earle´s supplemented with 0.2% NaHCO3, 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen) at 37°C for the times indicated from the basolateral side. For acid- induced activation of matriptase, cells were labeled in a physiological phosphate buffer (25 mM

Na2HPO4, 175 mM NaH2PO4) pH 6 or pre-treated with physiological phosphate buffer pH 6 before labeling in serum-free MEM eagle with Earle´s supplemented with 0.2% NaHCO3, 100 units/ml penicillin and 100 μg/ml streptomycin. As a negative control, cells were labeled with 50 μM of a corresponding peptide without a CMK group; biotin-Arg-Gln-Arg-Arg (biotin-RQRR). For labeling of surface proteins, cells were biotinylated from the basolateral side with 1 mg/ml EZ-link™ Sulfo- NHS-SS-Biotin (Pierce) dissolved in PBS++ for 30 min at 4°C. After peptide- and/or ordinary biotin- labeling, the cells were washed four times with ice-cold PBS++. In case of biotin-labeling, residual biotin was quenched with 50 mM glycine/PBS++ for 5 min at 4°C and the cells were washed twice with PBS++. Cells were lysed in PBS containing 1% Triton X-100, 0.5% deoxycholate and protease inhibitors (10 mg/l benzamidine, 2 mg/l pepstatin A, 2 mg/l leupeptin, 2 mg/l antipain, and 2 mg/l chymostatin). Insoluble material was precipitated at 20,000×g for 20 min at 4°C and the supernatants were transferred to clean eppendorf tubes.

Streptavidin pull down Cleared lysates were incubated for 2 hrs with end-over-end rotation at 4°C with 50 μl/24 mm filter pre-washed streptavidin-coated resin (Pierce), prepared as described by manufacturer. The streptavidin-coated resin was washed four times with 25 mM TRIS-HCl, 500 mM NaCl, 0.5% Triton X-100, pH 7.8, and three times with 10 mM TRIS-HCl, 150 mM NaCl, pH 7.8, and biotinylated proteins were eluted from the streptavidin-coated resin by boiling in SDS sample buffer.

SDS-PAGE and Western blot Proteins were separated on 10% acrylamide gels and transferred to Immobilon-P PVDF membranes (Millipore). PVDF membranes were blocked with 10% non-fat dry milk in PBS containing 0.1% Tween-20 (PBST) for 1 hr at RT and were probed with primary antibody diluted in

1% non-fat dry milk in PBST at 4°C o/N. The next day the membranes were washed 3 times with PBST, followed by detection of bound primary antibody with horseradish peroxidase (HRP)- conjugated secondary antibody (Pierce) or alkaline phosphatase (AP)-conjugated secondary antibody (Sigma-Aldrich). After 3 washes with PBST the signal was developed using the ECL reagent Super Signal West Femto Maximum Sensitivity Substrate (Pierce) for HRP-conjugated secondary antibodies according to the protocol supplied by the manufacturer and visualized with a Fuji LAS1000 camera (Fujifilm Sweden AB) or by nitro-blue tetrazolium and 5- bromo-4-chloro- 3´-indolyphosphate (Pierce) AP-conjugated secondary antibody.

Antibodies The antibodies used for detection of matriptase in Western blotting were monoclonal mouse anti- human matriptase antibody M32 that reacts with the A chain recognizing total matriptase as a 70 kDa band (both active and zymogen matriptase in boiled samples) and the 120-130 kDa complex of matriptase with HAI-1 (only in non-boiled samples) (3ug antibody/blot) [29]; monoclonal mouse anti-human matriptase antibody M69, which recognizes the 120-130kDa complex of matriptase with HAI-1 [29]; monoclonal mouse anti-human HAI-1 antibody M19, which recognizes free 55kDa HAI-1 and the 120-130kDa matriptase-HAI-1 complex [29]; polyclonal rabbit anti- human matriptase raised against the SPD of matriptase recognizing a 70 kDa band (both active and zymogen matriptase) under non-reducing conditions and the 70 kDa zymogen form and the 30 kDa protease domain of cleaved matriptase under reducing conditions (Cat. no. IM1014, Calbiochem). For detection of matriptase in murine cells, sheep anti-matriptase was used (AF3946 diluted 1:1000, R&D). Secondary antibodies include goat anti-mouse HRP-conjugated (10 ng/blot) (Pierce), goat anti-rabbit HRP-congugated (10 ng/blot) (Pierce) and donkey anti-sheep AP- congugated (10 ng/blot) (Sigma-Aldrich).

Results

Biotin-RQRR-Cmk peptide inhibitor designed to react with active matriptase

In order to detect active matriptase, we designed a peptide inhibitor biotin-RQRR-Cmk consisting of a tetra peptide; RQRR with an N-terminal biotin moiety and a C-terminal Cmk group. This inhibitor was designed based on a predicted substrate recognition sequence of matriptase [30]. The Cmk group ensures that the protease-peptide interaction results in the formation of a covalent bond between the peptide and the protease by alkylation of the active site histidine attaching a biotin moiety to the active enzyme [31]. The biotin group allows for efficient concentration of active protease from a complex media by streptavidin precipitation [32].

Biotin-RQRR-Cmk inhibits the proteolytic activity of matriptase SPD in vitro

To verify that biotin-RQRR-Cmk binds matriptase, we tested whether different concentrations of the peptide inhibitor were able to inhibit the proteolysis of a chromogenic substrate by purified recombinant matriptase serine protease domain (SPD). 50 μM biotin-RQRR-Cmk renders matriptase SPD unable to cleave the chromogenic substrate (Fig. 1A, black triangle). Although chloromethyl ketones are known to be hydrolysed in aqueous solutions with half lives in the order of 5-20 min [31], we found that an initial concentration of 50 μM biotin-RQRR-Cmk was still efficient after 180 min pre-incubation at 37°C and able to quench the activity of a sample containing 0.2 nM matriptase SPD (Fig. 1A, closed circle). A biotin-RQRR-Cmk concentration as low as 5nM is capable of inhibiting the peptidolytic activity of 0.2 nM matriptase SPD (fig. 1B, black triangle) even after 60 min of pre-incubation at 37°C (fig. 1B, open circle ), showing that biotin- RQRR-Cmk is very efficient near-stoichiometric inhibitor of matriptase SPD. Thus, biotin-RQRR- Cmk is a very efficient inhibitor of matriptase activity and relative stable in aqueous solutions.

Biotin-RRQR-Cmk reacts with a subset of matriptase molecules on the surface of cultured cells

Next, we wanted to test whether the biotin-RQRR-Cmk peptide inhibitor is able to bind active matriptase on the surface of cells in culture. Differentiated Caco-2 cells have an endogenous expression of matriptase that can be detected mainly as a zymogen form but also as a two-chain form in complex with HAI-1 [25;33]. We have previously shown that activation site cleavage of matriptase takes place on the basolateral plasma membrane of 11 days post-confluent Caco-2 cells [25], indicating that these cells contains active matriptase on the basolateral plasma membrane.

11 days post-confluent filter grown Caco-2 cells were treated with biotin-RQRR-Cmk from the basolateral side at 37°C for 2-180 min. Next, cells were lysed and biotin-RQRR-Cmk-labeled proteases were extracted from cleared lysates using streptavidin-coated resin. Proteins were released from the resin by boiling in SDS sample buffer. Due to substrate overlap of matriptase with other trypsin-like serine proteases [30], specificity of the assay is obtained by Western blot analysis using a matriptase specific antibody. To assess the steady state level of total surface- associated matriptase, parallel cultures were surface-biotinylated with S-NHS-SS-biotin at 4°C.

Matriptase binds biotin-RQRR-Cmk as demonstrated by Western blotting (Fig. 2, lanes 4-6), and no matriptase could be detected when labeling with a control peptide; biotin-RQRR (Fig. 2, lane 2: CTRL). Biotin-RQRR-Cmk labeling displayed time dependence at 37°C as an increasing amount of matriptase was detected with increasing time of incubation with biotin-RQRR-Cmk, indicating that active matriptase is continuously generated. Comparison of steady state surface-associated matriptase by standard biotinylation techniques to the accumulated biotin-RQRR-Cmk labeled matriptase shows that only a fraction of surface-associated matriptase on Caco-2 cells is able to

bind biotin-RQRR-Cmk (compare lane 2 to lanes 3-6). Incubation with biotin-RQRR-Cmk for 180 min at 4°C gave very low signals (data not shown) indicating that the steady state level of matriptase that is able to bind to biotin-RQRR-Cmk is low. Biotin-RQRR-Cmk also react with other serine proteases, as prostasin could be detected in Western blotting (data not shown), emphasizing that specificity of the assay depends on the antibody used for the Western blot analysis.

Biotin-RRQR-Cmk does not react with the matriptase-HAI-1 complex

After activation site cleavage, matriptase forms a reversible, but SDS-resistant complex with HAI-1 that can be analyzed by SDS-PAGE under non-reducing conditions. In order to investigate whether biotin-RRQR-Cmk binds to the matriptase-HAI-1 complex, we took advantage of the fact that cells exposed to slightly acidic conditions has been reported to rapidly convert matriptase from the zymogen form into the matriptase-HAI-1 complex [24;26;34]. First, we tested whether exposure to slightly acidic conditions also in Caco-2 cells converts zymogen matriptase into a complex of activation site cleaved matriptase inhibited by HAI-1.

11 days post-confluent filter grown Caco-2 cells were treated with physiological phosphate buffer pH 6.0 for 20 min (Fig. 3A, lanes 2, 4, 6, and 8) or left untreated (Fig. 3A, lanes 1, 3, 5, and 7). The two different lysates were investigated by Western blotting under non-boiled and non-reduced conditions (Fig. 3A, lanes 1-4 and 7-8) and under reducing conditions (Fig. 3A, lanes 5 and 6) using different antibodies. Untreated Caco-2 cells mainly contain 70 kDa zymogen matriptase (Fig. 3A, lanes 1 and 5) and to a minor degree the 130 kDa matriptase-HAI-1 complex (Fig. 3A, lane 1), whereas matriptase is detected as the 130 kDa matriptase-HAI-1 complex and to a minor degree 70 kDa zymogen matriptase under non-boiling conditions after treatment with physiological phosphate buffer pH 6.0 (Fig. 3A, lane 2). The generation of the pH 6.0 induced 130 kDa matriptase-HAI-1 complex is confirmed by antibodies against the matriptase-HAI-1 complex (M69) (Fig. 3A, compare lanes 3 and 4) and a HAI-1 antibody that also recognizes the 130 kDa matriptase-HAI-1 complex (Fig. 3A, compare lanes 7 and 8). Analysis of the same samples under reducing conditions confirms the pH 6.0 induced activation site cleavage of matriptase, as matriptase was primarily detected as a band migrating at 30 kDa representing the released disulfide-linked SPD of matriptase after this treatment (Fig. 3A, lane 6). Thus, treatment with physiological phosphate buffer pH 6.0 induces activation site cleavage of matriptase and subsequent matriptase-HAI-1 complex formation in Caco-2 cells.

Active matriptase has been shown to have pH optimum at pH 9 [30], we therefore tested the activity of recombinant matriptase SPD and the inhibitory capacity of biotin-RQRR-Cmk at pH 6 in the chromogenic assay previously described in Fig. 1. Matriptase SPD is able to cleave the chromogenic substrate at pH 6.0, although at a much lowered level, as compared to neutral pH (compare Fig. 3B, square to Fig. 1, square). When 50 μM biotin-RQRR-Cmk was added to the reaction, no cleavage of the substrate was observed (Fig. 3B, black triangle). Thus, matriptase SPD

is able to bind biotin-RQRR-Cmk at this lowered pH. This enabled us to test whether biotin-RRQR- Cmk binds to the matriptase-HAI-1 complex.

To examine whether biotin-RQRR-Cmk reacts with the matriptase-HAI-1 complex, we employed physiological phosphate buffer pH 6.0 to induce matriptase activation. 11 days post-confluent filter grown Caco-2 cells were treated with 50 μM biotin-RQRR-Cmk from the basolateral side at 37°C for 30 min under different conditions; labeling with biotin-RQRR-Cmk at pH 7.4 (Fig. 4, lanes 3 and 7), labeling with biotin-RQRR-Cmk in physiological phosphate buffer pH 6.0 (Fig. 4, lanes 4 and 8), or pre-treatment with physiological phosphate buffer pH 6.0 for 30 min, before labeling with biotin-RQRR-Cmk at pH 7.4 for 30 min (Fig. 4, lanes 5 and 9). Additional controls were lysates of untreated cells (Fig. 4, lane 1) and cells treated with physiological phosphate buffer pH 6.0 (Fig. 4, lane 2) that show pH 6.0 induced matriptase activation and complex formation with HAI-1 (Fig. 4, compare lane 1 and 2). In all cases, aliquots of the total lysates (Fig. 4, lanes 1-5) as well as the boiled streptavidin pull downs were analyzed in Western blot analysis (Fig. 4, lanes 6-9).

Under the conditions used here, biotin-RQRR-Cmk is unable to dissociate the matriptase-HAI- complex formed by pH 6 treatment (Fig. 4, compare lane 9 with lanes 4, 6 and 8). On the other hand, simultaneous treatment of Caco-2 cells with biotin-RQRR-Cmk and physiological phosphate buffer pH 6.0 hinders complex formation between matriptase and HAI-1 (Fig. 4, lanes 4 and 5) and instead more matriptase binds biotin-RQRR-Cmk as detected by Western blot analysis (Fig. 4, lanes 7 and 8). This observation is supporting the mutual sterical block of the active site in a competitive manner by either covalent binding to biotin-RQRR-Cmk or non-covalent, SDS- resistant complex formation with HAI-1 [31;35]. Thus, HAI-1 and biotin-RQRR-Cmk competes for binding to active matriptase.

Biotin-RRQR-Cmk reacts with zymogen matriptase

CMK peptides have been widely used in studies of serine proteases and in most cases Cmk- peptides react only with the active form of the protease and not with the zymogen form except in a few cases of autoactivating proteases that have an intrinsic activity enabling the zymogen to reacts with the Cmk-peptide, one example is tPA [36;37]. Matriptase is also an autoactivating serine protease implying that the zymogen form has an intrinsic activity [26;38]. To investigate for an intrinsic activity of matriptase, we labeled 11 days post-confluent filter grown Caco-2 cells with biotin-RQRR-Cmk from the basolateral side for 180 min at 37°C. Biotin-RQRR-Cmk labeled proteases were precipitated with streptavidin-coated resin, and subjected to reducing SDS-PAGE. Samples were analyzed by Western blot analysis using an antibody against matriptase SPD. Under reducing conditions, matriptase could be detected both as the 70 kDa form (representing zymogen matriptase) and the 30 kDa form (representing activation site cleaved matriptase) with a stronger signal intensity of the 70 kDa form (Fig. 5, lane 2). No matriptase could be detected when labeling Caco-2 cells with a control peptide; biotin-RQRR (Fig. 5, lane 1: CTRL). Thus, both

activation site cleaved and zymogen matriptase is able to bind biotin-RQRR-Cmk indicating that zymogen matriptase has an intrinsic activity.

Biotin-RRQR-Cmk also reacts with matriptase in other cell systems

The described assay may easily be modified to detect active matriptase from other species as the specificity of the assay depends on the antibody used. To verify this, the presence of active matriptase was investigated in cultured primary murine keratinocytes. Keratinocytes from the skin of matriptase wild-type (WT) and matriptase deficient (KO) newborn pups were isolated, plated on collagen coated plastic and labeled with 50 μM biotin-RQRR-Cmk at subconfluence for 180 min at 37°C or labeled with S-NHS-SS-biotin to assess surface-associated matriptase. As a negative control, keratinocytes were labeled with biotin-RQRR. Equal aliquots of cleared lysates were analyzed by Western blot for total matriptase, prior to streptavidin pull down of biotinylated proteins. Proteins were released from the streptavidin-coated resin by boiling in SDS sample buffer and analyzed by Western blotting. By labeling of keratinocytes isolated from mice expressing WT matriptase with biotin-RQRR-Cmk, we were able to detect active matriptase (Fig. 6, lane 2). No matriptase was detected when labeling with a corresponding control peptide; biotin-RQRR (Fig. 6, lane 3). No matriptase could be detected in lysates or pull downs of keratinocytes from matriptase-deficient mice (Fig. 6, lanes 4-6 and 10-12), whereas matriptase was easily detected in all lysates of the 3 differently treated keratinocyte cultures from mice expressing WT matriptase (Fig. 6, lanes 7-9). Evaluation of total surface biotinylation and biotin- RQRR-Cmk labeling of WT murine keratinocytes showed that only a fraction of surface-associated matriptase on WT keratinocytes cells could be detected by means of biotin-RQRR-Cmk (compare Fig. 6 lanes 1 and 2). Thus, we have established an assay for detection of active matriptase that can be easily applied in other cell systems and other species.

Discussion

Matriptase is essential for postnatal survival and tight regulation of matriptase is crucial for a number of physiological functions including development and maintenance of epithelia [3-5;12- 14;16;17]. Deregulated matriptase activity can also promote skin carcinogenesis [19]. It is therefore desirable to obtain methods to detect active matriptase to be able to assess active matriptase under these processes. This has proven difficult as no specific matriptase inhibitor and/or substrate is commercially available.

For these reasons, we combined antibody specificity with the high affinity of biotin-streptavidin interaction in the design of a peptide inhibitor-based assay for detection of active matriptase. We

engineered a chloromethyl ketone based tetra-peptide inhibitor that allows for extraction of inhibitor bound matriptase from complex media by means of an N-terminal biotin moiety. Specificity of the assay is obtained by Western blot analysis employing specific antibodies against matriptase. Chloromethyl ketone based inhibitors have been extensively used as inhibitors of proteases and are active site-directed inhibitors that irreversible bind serine proteases by alkylation the active site histidine residue of serine proteases rendering it catalytic inactive [31]. The assay is based on the assumption that active matriptase is equivalent to inhibitor bound matriptase. We show here that matriptase SPD activity is efficiently inhibited by biotin-RQRR-Cmk and that biotin-RQRR-Cmk is able to label active matriptase on the cell surface of cultured cells.

Upon activation, matriptase rapidly forms a complex with HAI-1 [26;34;39]. HAI-1 is a Kunitz-type transmembrane serine protease inhibitor that binds to active matriptase in a reversible and competitive manner as is typical for this family of protease inhibitors [24;40]. The matriptase-HAI- 1 complex comprises a docking of the Kunitz domain I of HAI-1 onto the active site and close surroundings thereby “capping” the substrate accessibility to the active site of the protease [35]. We found that HAI-1 and biotin-RQRR-Cmk compete for active matriptase and although HAI-1 binds matriptase in an irreversible, but yet very stable manner, biotin-RQRR-Cmk is not able to dissociate the matriptase-HAI-1 complex under the conditions used in the study. Moreover, the presence of biotin-RQRR-Cmk hinders matriptase-HAI-1 complex formation.

In most epithelial cell lines and tissues a large fraction of matriptase is found in its zymogen form [23;26]. In accordance with this, we show that only a fraction of surface-associated matriptase could be labeled with biotin-RQRR-biotin from cell cultures of primary murine keratinocytes and unstimulated Caco-2 cells. The zymogen form of trypsin-like proteases is considered to exist in equilibrium between two conformations, an inactive conformation and an active conformation. Although most serine proteases are strongly in favor of the inactive form, intrinsic activity has been reported for a number of autoactivating proteases [36]. Matriptase is an autoactivating protease implying that the zymogen form has an inherent protease activity. We show that zymogen matriptase is able to bind biotin-RQRR-Cmk. This finding is supported by a recent study where a recombinant form of zymogen matriptase in an in vitro assay reacts with an inhibitor similar to the one used in the present study [41].

Acid-induced activation of matriptase is ubiquitous among epithelial and carcinoma cells and is followed by rapid complex formation with HAI-1 [24;26;34]. Concurrent with this, we find that treatment of cells with physiological phosphate buffer pH 6.0 induces matriptase-HAI-1 complex formation in Caco-2 cells. We also find that a lowering of pH induces a rise in the level of matriptase that can be labeled with the inhibitor suggesting that there is a brief window between activation of matriptase and complex formation with HAI-1. This may be of physiological relevance, as the pericellular environment turns acidic in the transitional layer of the epidermis, the precise location where matriptase initiates a proteolytic cascade crucial for terminal epidermal differentiation [3;5;42].

In summary, we have established an assay for detection active matriptase in cell culture. The availability of an assay for detection of active matriptase can greatly contribute to the further understanding of the complex biology of matriptase.

Figures

Fig.1 Inhibition of matriptase SPD-mediated chromogenic substrate cleavage by biotin-RQRR- Cmk. (A) 0.2 μM matriptase SPD was incubated for 10 min at 37°C with (black triangle) or without (square) 50 μM biotin-RQRR-Cmk before addition the chromogenic substrate to a final concentration of 300μM. The reactivity of biotin-RQRR-Cmk was tested after 180 min of pre- incubation at 37°C (closed circle). Matriptase SPD was unable to cleave the chromogenic substrate, when biotin-RQRR-Cmk was added (black triangle) even after 180 min pre-incubation before addition to reaction buffer (closed circle). (B) 0.2nM matriptase SPD was incubated for 10 min at 37°C with (black triangle) or without (square) 5 nM biotin-RQRR-Cmk before addition the chromogenic substrate to a final concentration of 300μM. The reactivity of biotin-RQRR-Cmk was tested with 60 min, 120 min, and 180 min of pre-incubation at 37°C before addition to reaction buffer. 5nM Biotin-RQRR-Cmk is able to completely inhibit substrate cleavage after 60 min pre- incubation at 37°C (open circle) and is still able to partly inhibit substrate cleavage after 180 min pre-incubation at 37°C (closed circle). All measurements were performed in 20 mM TRIS pH 7.4, 140 mM NaCl supplemented with 0.1% BSA at 37°C. The reaction buffer was used as an internal control to verify that the peptidolytic activity originated from matriptase SPD (grey triangle) and not unspecific degradation of the substrate over time. Each plot shows the change in optical density at 405 nm of the reaction mixture as a function of reaction time. The presence of active protease results in a continued release of a yellow cleavage product resulting in a linear color development in agreement with a pseudo 1st order reaction due to the high molar excess of substrate to protease. Results shown are representative of 3 independent experiments.

Fig. 2. Biotin-RRQR-Cmk reacts with a subset of matriptase molecules on the surface of Caco-2 cells. 11 day post-confluent Caco-2 cells grown on Transwell filters were labeled with 50μM biotin- RQRR-Cmk from the basolateral side for the times indicated (2-180 min) at 37°C. As a measure of steady state levels of matriptase, cells were incubated on the basolateral membrane with S-NHS- SS-biotin at 4°C (lane 1) to biotinylate cell surface proteins. As a negative control, cells were labeled from the basolateral side with 50 μM control peptide; biotin-RQRR (lane 2). Biotinylated proteins were precipitated using streptavidin-coated resin and the streptavidin pull downs were analyzed by non-reducing SDS-PAGE and Western blotting using the monoclonal matriptase antibody; M32. A tenth of the surface biotinylated sample was loaded (lane 1); whereas total sample volume was loaded for the other samples (lanes 2-6). Surface biotinylation with S-NHS-SS- biotin resulted in a band of high intensity (lane 1), indicating that matriptase is abundantly present on the basolateral membrane. There is a clear time dependence of biotin-RQRR-Cmk binding to matriptase with weak detection of matriptase after 30 min to the most prominent band after 180 min (lanes 3-6). No matriptase could be detected with the control peptide; biotin-RQRR (CTRL, lane 2). Position of the molecular weight markers (kDa) is indicated on the left. Results shown are representative of 3 independent experiments.

Fig. 3: Matriptase activation, activity and inhibitor complex formation at pH 6.0. (A) 11 days post-confluence filter grown Caco-2 cells were either treated at pH 6.0 (lanes 2, 4, 6, and 8) or left untreated (Lanes 1, 3, 5, and 7) and lysates were analyzed by Western blotting with antibody against total matriptase (M24; lane 1 and 2), matriptase SPD (IM1014; lanes 5 and 6), matrpitase- HAI-1 complex (M69; lanes 3 and 4) and HAI-1 (lanes 7 and 8). Samples in lane 1-4, 7 and 8 were not boiled, while samples in lanes 5 and 6 were boiled and reduced to dissociate the S-S bridged SPD from the stem domain of activated matriptase. Non-boiled samples of untreated cells reveal that matriptase is primarily present as a prominent band of 70 kDa and as a minor band at 130 kDa representing the matriptase-HAI-1 complex form (lane 1). Treatment at low pH reveals matriptase primarily in the 130 kDa complex form and a minor fraction in the 70 kDa form (lane 2). The same rise in matriptase-HAI-1 complex can be detected with the M69 antibody (lanes 3 and 4) and the HAI-1 antibody M19 (lanes 7 and 8). Under reducing conditions, matriptase can be detected mainly as the 70 kDa form, but also as two band of app. 30 kDa in size (lane 5). These 30 kDa forms are more prominent than the 70 kDa form when cells were treated at pH 6.0 (lane 6). Treatment with phosphate buffer pH 6.0 and DTT is indicated by +/-. Position of the molecular weight markers (kDa) is indicated on the left. (B) 0.2 μM SPD were incubated for 10 min at 37°C with (black triangle) or without (square) 50 μM biotin-RQRR-Cmk before addition the chromogenic substrate to a final concentration of 300μM. All experiments were performed in 20 mM citric acid buffer pH 6.0, 140 mM NaCl and 0.1% BSA at 37°C. Matriptase SPD is able to cleave the chromogenic substrate resulting in color development of the sample at pH 6.0, when biotin-RQRR- Cmk was omitted (square) and an initial concentration of 50μM biotin-RQRR-Cmk is able to quench matriptase SPD activity towards the chromogenic substrate also at pH 6.0. A control containing only buffer verified that the peptidolytic activity originated from matriptase SPD (grey triangle) and not unspecific degradation of the substrate over time. Results shown are representative of 3 independent experiments.

Fig. 4. HAI-1 and biotin-RQRR-Cmk competes in inhibition of matriptase

11 days post-confluent Caco-2 cells grown on Transwell filters were labeled with 50 μM biotin- RQRR-Cmk at neutral pH 7.4 (lane 3 and 7), in physiological phosphate buffer pH 6.0 (Lane 4 and 8), or at neutral pH with a 30 min pre-incubation treatment with physiological phosphate buffer pH 6.0 (Lanes 5 and 9) for 30 min at 37°C. Samples of lysates were analyzed under non-boiled and non-reducing conditions (lanes 1-5). Labeled proteases were precipitated using streptavidin coated resin and released from the beads by boiling. As a negative control, lysate of cells treated with only physiological phosphate buffer pH 6.0 for 30 min was precipitated (CTRL, lanes 6). The streptavidin pull downs and non-boiled lysates of the same samples were analyzed by Western blotting using the monoclonal M32 antibody. Zymogen activation and matriptase-HAI-1 complex formation is induced with pH 6.0 treatment (lanes 1 and 2, respectively). Likewise, the amount of matriptase able to bind biotin-RQRR-Cmk also increases after pH 6.0 treatment (compare lanes 7 and 8) and the presence of biotin-RQRR-Cmk hinders complex formation (compare lanes 4 and 5). Matriptase activation induced prior to biotin-RQRR-Cmk treatment result in matriptase-HAI-1 complex formation (lane 5) and only low levels of matriptase binds biotin-RQRR-Cmk under this condition (lane 9) comparable to the level detected when treating cells with biotin-RQRR-Cmk at neutral pH (lane 7). Position of the molecular weight markers (kDa) is indicated on the left. Results shown are representative of 1 experiment.

Fig. 5. Biotin-RQRR-Cmk inhibits both matriptase zymogen and activation site cleaved matriptase. 11 days post-confluent Caco-2 cells grown on Transwell filters were labeled with 50 μM biotin-RQRR-Cmk from the basolateral side for 180 min at 37°C. As a negative control, cells were labeled from the basolateral side with 50 μM control peptide; biotin-RQRR (CTRL), under the same conditions. Labeled proteases were precipitated using streptavidin-coated resin and the streptavidin pull downs were analyzed by reducing SDS-PAGE and Western blotting using the IM1014 antibody raised against matriptase SPD. The biotin-RQRR-biotin peptide labels both the 70 kDa zymogen matriptase and activation site cleaved matriptase, which under these conditions can be detected as a 30 kDa band. Position of the molecular weight markers (kDa) is indicated on the left and position of matriptase zymogen and SPD is indicated on the right. Results shown are representative of 3 independent experiments.

Fig. 6: Detection of active matriptase in cultured primary murine keratinocytes. Murine keratinocytes were isolated from newborn wild-type (WT) or matriptase-deficient pups and plated on collagen-coated plastic. The cells were grown until sub-confluence and then labeled with the S-NHS-SS-biotin (B: lane 1, 4, 7, and 10), with 50 μM biotin-RQRR-Cmk (lane 2, 5, 8, and 11), or with 50 μM control peptide; biotin-RQRR (lanes 3, 6, 9, and 12). Labeled proteases were precipitated using streptavidin-coated resin, and released from the beads by boiling and analyzed by SDS-PAGE and Western blotting using the matriptase antibody AF3946. Matriptase is detected in all lysates of WT keratinocytes (lanes 7-9), whereas no matriptase is detected in the lysates or pull downs from matriptase-deficient keratinocytes (lanes 4-6 and 10-12). The pull down fraction of S-NHS-SS-biotin labeled cells (lane 1) as well as the loaded aliquots of WT lysates (lanes 1-3 and 7-9) shows that matriptase is abundantly present at the plasma membrane and that the fraction of peptidolytic active matriptase only constitute a small fraction of total surface-associated matriptase (lane 2). Results shown are representative of 2 independent experiments.

References

[1] Oberst MD, Singh B, Ozdemirli M, Dickson RB, Johnson MD, & Lin CY (2003) Characterization of matriptase expression in normal human tissues. J. Histochem. Cytochem., 51, 1017-1025.

[2] Oberst M, Anders J, Xie B, Singh B, Ossandon M, Johnson M, Dickson RB, & Lin CY (2001) Matriptase and HAI-1 are expressed by normal and malignant epithelial cells in vitro and in vivo. Am. J. Pathol., 158, 1301-1311.

[3] List K, Szabo R, Molinolo A, Nielsen BS, & Bugge TH (2006) Delineation of matriptase protein expression by enzymatic gene trapping suggests diverging roles in barrier function, hair formation, and squamous cell carcinogenesis. Am. J. Pathol., 168, 1513-1525.

[4] List K, Kosa P, Szabo R, Bey AL, Wang CB, Molinolo A, & Bugge TH (2009) Epithelial integrity is maintained by a matriptase-dependent proteolytic pathway. Am. J. Pathol., 175, 1453-1463.

[5] List K, Haudenschild CC, Szabo R, Chen W, Wahl SM, Swaim W, Engelholm LH, Behrendt N, & Bugge TH (2002) Matriptase/MT-SP1 is required for postnatal survival, epidermal barrier function, hair follicle development, and thymic homeostasis. Oncogene, 21, 3765-3779.

[6] Basel-Vanagaite L, Attia R, Ishida-Yamamoto A, Rainshtein L, Ben AD, Lurie R, Pasmanik-Chor M, Indelman M, Zvulunov A, Saban S, Magal N, Sprecher E, & Shohat M (2007) Autosomal recessive ichthyosis with hypotrichosis caused by a mutation in ST14, encoding type II transmembrane serine protease matriptase. Am. J. Hum. Genet., 80, 467-477.

[7] Avrahami L, Maas S, Pasmanik-Chor M, Rainshtein L, Magal N, Smitt J, van MJ, Shohat M, & Basel- Vanagaite L (2008) Autosomal recessive ichthyosis with hypotrichosis syndrome: further delineation of the phenotype. Clin. Genet., 74, 47-53.

[8] Alef T, Torres S, Hausser I, Metze D, Tursen U, Lestringant GG, & Hennies HC (2009) Ichthyosis, follicular atrophoderma, and hypotrichosis caused by mutations in ST14 is associated with impaired profilaggrin processing. J. Invest Dermatol., 129, 862-869.

[9] Sales KU, Masedunskas A, Bey AL, Rasmussen AL, Weigert R, List K, Szabo R, Overbeek PA, & Bugge TH (2010) Matriptase initiates activation of epidermal pro-kallikrein and disease onset in a mouse model of Netherton syndrome. Nat. Genet., 42, 676-683.

[10] List K, Szabo R, Wertz PW, Segre J, Haudenschild CC, Kim SY, & Bugge TH (2003) Loss of proteolytically processed filaggrin caused by epidermal deletion of Matriptase/MT-SP1. J. Cell Biol., 163, 901-910.

[11] Tanaka H, Nagaike K, Takeda N, Itoh H, Kohama K, Fukushima T, Miyata S, Uchiyama S, Uchinokura S, Shimomura T, Miyazawa K, Kitamura N, Yamada G, & Kataoka H (2005) Hepatocyte growth factor activator inhibitor type 1 (HAI-1) is required for branching morphogenesis in the chorioallantoic placenta. Mol. Cell Biol., 25, 5687-5698.

[12] Szabo R, Hobson JP, Christoph K, Kosa P, List K, & Bugge TH (2009) Regulation of cell surface protease matriptase by HAI2 is essential for placental development, neural tube closure and embryonic survival in mice. Development, 136, 2653-2663.

[13] Szabo R, Molinolo A, List K, & Bugge TH (2007) Matriptase inhibition by hepatocyte growth factor activator inhibitor-1 is essential for placental development. Oncogene, 26, 1546-1556.

[14] Carney TJ, von der HS, Sonntag C, Amsterdam A, Topczewski J, Hopkins N, & Hammerschmidt M (2007) Inactivation of serine protease Matriptase1a by its inhibitor Hai1 is required for epithelial integrity of the zebrafish epidermis. Development, 134, 3461-3471.

[15] Nagaike K, Kawaguchi M, Takeda N, Fukushima T, Sawaguchi A, Kohama K, Setoyama M, & Kataoka H (2008) Defect of hepatocyte growth factor activator inhibitor type 1/serine protease inhibitor, Kunitz type 1 (Hai-1/Spint1) leads to ichthyosis-like condition and abnormal hair development in mice. Am. J. Pathol., 173, 1464-1475.

[16] Szabo R, Kosa P, List K, & Bugge TH (2009) Loss of matriptase suppression underlies spint1 mutation- associated ichthyosis and postnatal lethality. Am. J. Pathol., 174, 2015-2022.

[17] Mathias JR, Dodd ME, Walters KB, Rhodes J, Kanki JP, Look AT, & Huttenlocher A (2007) Live imaging of chronic inflammation caused by mutation of zebrafish Hai1. J. Cell Sci., 120, 3372-3383.

[18] Szabo R, Uzzun SK, Kosa P, Shylo NA, Godiksen S, Hansen KK, Friis S, Gutkind JS, Vogel LK, Hummler E, Camerer E, & Bugge TH (2012) Reduced Prostasin (CAP1/PRSS8) Activity Eliminates HAI-1 and HAI-2 Deficiency-Associated Developmental Defects by Preventing Matriptase Activation. PLoS. Genet., 8, e1002937.

[19] List K, Szabo R, Molinolo A, Sriuranpong V, Redeye V, Murdock T, Burke B, Nielsen BS, Gutkind JS, & Bugge TH (2005) Deregulated matriptase causes ras-independent multistage carcinogenesis and promotes ras-mediated malignant transformation. Genes Dev., 19, 1934-1950.

[20] Szabo R, Rasmussen AL, Moyer AB, Kosa P, Schafer JM, Molinolo AA, Gutkind JS, & Bugge TH (2011) c- Met-induced epithelial carcinogenesis is initiated by the serine protease matriptase. Oncogene.

[21] Kosa P, Szabo R, Molinolo AA, & Bugge TH (2011) Suppression of Tumorigenicity-14, encoding matriptase, is a critical suppressor of colitis and colitis-associated colon carcinogenesis. Oncogene.

[22] Chen YW, Wang JK, Chou FP, Chen CY, Rorke EA, Chen LM, Chai KX, Eckert RL, Johnson MD, & Lin CY (2010) Regulation of the matriptase-prostasin cell surface proteolytic cascade by hepatocyte growth factor activator inhibitor-1 during epidermal differentiation. J. Biol. Chem., 285, 31755-31762.

[23] Lee MS, Kiyomiya K, Benaud C, Dickson RB, & Lin CY (2005) Simultaneous activation and hepatocyte growth factor activator inhibitor 1-mediated inhibition of matriptase induced at activation foci in human mammary epithelial cells. Am. J. Physiol Cell Physiol, 288, C932-C941.

[24] Benaud C, Dickson RB, & Lin CY (2001) Regulation of the activity of matriptase on epithelial cell surfaces by a blood-derived factor. Eur. J. Biochem., 268, 1439-1447.

[25] Friis S, Godiksen S, Bornholdt J, Selzer-Plon J, Rasmussen HB, Bugge TH, Lin CY, & Vogel LK (2010) Transport via the transcytotic pathway makes prostasin available as a substrate for matriptase. J. Biol. Chem..

[26] Lee MS, Tseng IC, Wang Y, Kiyomiya K, Johnson MD, Dickson RB, & Lin CY (2007) Autoactivation of matriptase in vitro: requirement for biomembrane and LDL receptor domain. Am. J. Physiol Cell Physiol, 293, C95-105.

[27] Yuan C, Chen L, Meehan EJ, Daly N, Craik DJ, Huang M, & Ngo JC (2011) Structure of catalytic domain of Matriptase in complex with Sunflower trypsin inhibitor-1. BMC. Struct. Biol., 11, 30.

[28] Lichti U, Anders J, & Yuspa SH (2008) Isolation and short-term culture of primary keratinocytes, hair follicle populations and dermal cells from newborn mice and keratinocytes from adult mice for in vitro analysis and for grafting to immunodeficient mice. Nat. Protoc., 3, 799-810.

[29] Lin CY, Anders J, Johnson M, & Dickson RB (1999) Purification and characterization of a complex containing matriptase and a Kunitz-type serine protease inhibitor from human milk. J. Biol. Chem., 274, 18237-18242.

[30] Beliveau F, Desilets A, & Leduc R (2009) Probing the substrate specificities of matriptase, matriptase-2, hepsin and DESC1 with internally quenched fluorescent peptides. FEBS J., 276, 2213-2226.

[31] Powers JC, Asgian JL, Ekici OD, & James KE (2002) Irreversible inhibitors of serine, cysteine, and threonine proteases. Chem. Rev., 102, 4639-4750.

[32] Anderson PJ & Bock PE (2001) Biotin derivatives of D-Phe-Pro-Arg-CH2Cl for active-site-specific labeling of thrombin and other serine proteinases. Anal. Biochem., 296, 254-261.

[33] Buzza MS, Netzel-Arnett S, Shea-Donohue T, Zhao A, Lin CY, List K, Szabo R, Fasano A, Bugge TH, & Antalis TM (2010) Membrane-anchored serine protease matriptase regulates epithelial barrier formation and permeability in the intestine. Proc. Natl. Acad. Sci. U. S. A.

[34] Tseng IC, Xu H, Chou FP, Li G, Vazzano AP, Kao JP, Johnson MD, & Lin CY (2010) Matriptase activation, an early cellular response to acidosis. J. Biol. Chem., 285, 3261-3270.

[35] Friedrich R, Fuentes-Prior P, Ong E, Coombs G, Hunter M, Oehler R, Pierson D, Gonzalez R, Huber R, Bode W, & Madison EL (2002) Catalytic domain structures of MT-SP1/matriptase, a matrix-degrading transmembrane serine proteinase. J. Biol. Chem., 277, 2160-2168.

[36] Williams EB, Krishnaswamy S, & Mann KG (1989) Zymogen/enzyme discrimination using peptide chloromethyl ketones. J. Biol. Chem., 264, 7536-7545.

[37] Ranby M, Bergsdorf N, & Nilsson T (1982) Enzymatic properties of the one- and two-chain form of tissue plasminogen activator. Thromb. Res., 27, 175-183.

[38] Oberst MD, Williams CA, Dickson RB, Johnson MD, & Lin CY (2003) The activation of matriptase requires its noncatalytic domains, serine protease domain, and its cognate inhibitor. J. Biol. Chem., 278, 26773- 26779.

[39] Benaud C, Oberst M, Hobson JP, Spiegel S, Dickson RB, & Lin CY (2002) Sphingosine 1-phosphate, present in serum-derived lipoproteins, activates matriptase. J. Biol. Chem., 277, 10539-10546.

[40] Laskowski M, Jr. & Kato I (1980) Protein inhibitors of proteinases. Annu. Rev. Biochem., 49, 593-626.

[41] Inouye K, Yasumoto M, Tsuzuki S, Mochida S, & Fushiki T (2010) The optimal activity of a pseudozymogen form of recombinant matriptase under the mildly acidic pH and low ionic strength conditions. J. Biochem., 147, 485-492.

[42] Netzel-Arnett S, Currie BM, Szabo R, Lin CY, Chen LM, Chai KX, Antalis TM, Bugge TH, & List K (2006) Evidence for a matriptase-prostasin proteolytic cascade regulating terminal epidermal differentiation. J. Biol. Chem., 281, 32941-32945.

Discussion and perspectives

In the following section the data presented in the three manuscripts enclosed in this thesis will be discussed starting with a short summary of the key findings

In paper I, we delineate the subcellular trafficking of matriptase, prostasin and HAI-1 to address how matriptase can act as an upstream activator of prostasin in the epidermis, despite their different subcellular localizations in polarized epithelial cells. We show that matriptase is routed to the basolateral plasma membrane, where it is cleaved into its active form. We also found that both prostasin and HAI-1 are transported to the basolateral plasma membrane from where they are internalized and transcytosed to the apical membrane. No transcytosis of matriptase could be detected although matriptase is endocytosed in a complex with HAI-1 from the basolateral plasma membrane. Interestingly, we could detect active prostasin on both the apical and the basolateral plasma membrane, as well as active matriptase on the basolateral plasma membrane by their ability to bind inhibitor-coupled beads. This study propose the basolateral plasma membrane as the site for matriptase-prostasin interaction and hereby suggest how the apically located substrate, prostasin, can be cleaved and activated by a basolateral localized protease, matriptase, in terminal epidermal differentiation. To gain more knowledge of matriptase-mediated proteolysis in normal physiological processes, we performed genetic epistasis analyses to identify new component(s) of matriptase-dependent proteolytic pathways critical to embryogenesis (paper II). We found that a hypomorphic mutation in Prss8, the gene encoding prostasin, in a similar manner to ablation of ST14, encoding matriptase, restored nearly all developmental defects associated with HAI-1 and HAI-2 deficiency indicating that matriptase and prostasin function in the same pathway(s) crucial for survival of newborn mice. Interestingly, we show that activation of matriptase does not occur in prostasin deficient placental tissues, whereas activated prostasin is readily detectable in matriptase deficient placental tissues. Thus, contrary to studies of the epidermis, we demonstrate that prostasin acts upstream of matriptase and is requires for its activation in placental tissues. In an attempt to obtain tools for detection of active matriptase, we designed a chloromethyl ketone peptide inhibitor of matriptase and were hereby able to optimize conditions and establish an assay for detection of active matriptase in live cell cultures. We found that matriptase is active on the basolateral plasma membrane by its ability to bind the chloromethyl ketone peptide inhibitor, biotin-RQRR-Cmk, however only a fraction of total surface matriptase could by extracted by means of the inhibitor. We show that the chloromethyl ketone peptide inhibitor competes with HAI-1 for binding to matriptase and that the matriptase zymogen has an intrinsic activity that enables it to bind the chloromethyl ketone peptide inhibitor. Moreover, we are able to modify the assay to confirm the presence of active matriptase in cultures of primary murine keratinocytes.

Within recent years, the conception of a matriptase-prostasin proteolytic pathway has been established in epidermal homeostasis. The existence of such a cascade is based on the facts that active prostasin is absent in matriptase-deficient epidermis, and that matriptase-deficient mice

85 and mice deficient of prostasin in the skin display nearly indistinguishable phenotypes. Also the two proteases are co-expressed in a number of epithelial tissues besides the epidermis [4;9;12;15;16;116]. We have previously delineated the intracellular transport of human HAI-1 recombinantly expressed in a different model of the polarized epithelia, namely the MDCK cell line. This study show that HAI-1 has a complex subcellular itinerary and that HAI-1 is transcytosed from the basolateral plasma membrane to the apical membrane (supplementary I; [27]). The fact that matriptase in complex with HAI-1 has been purified from human milk, and unpublished results showing that a 80 kDa protein was co-precipitated with HAI-1 lead us to the hypothesis that HAI- 1, in addition to its protease inhibitor function, could play a role in transporting matriptase in a matriptase–HAI-1 complex from the basolateral plasma membrane to the apical plasma membrane, where matriptase could activate prostasin (Sine Godiksen, unpublished results; supplementary I [27], [44]). In the current studies (paper I and III), we have used the human Caco-2 cell line as a model for the polarized epithelial cell. Paper I reports a polarized distribution of endogenous expressed matriptase, prostasin and HAI-1 in Caco-2 cells, which is in accordance to findings in other cell types and tissues, showing matriptase and HAI-1 on the basolateral plasma membrane and prostasin on the apical plasma membrane [9;11;21;23-26]. This makes the Caco-2 cell line a suitable model system for the studies conducted in paper I and paper III. The results obtained in paper I also confirmed the previous results obtained in MDCK cells on subcellular trafficking of HAI-1. However, we were unable to detect transcytosis of matriptase from the basolateral to the apical plasma membrane as hypothesized, although shedding of HAI- 1-complexed matriptase to the apical media has been observed in polarized Caco-2 cells by us and others (Stine Friis, unpublished result; [78]). Instead the results from paper I point to a mechanism where prostasin is routed to the basolateral plasma membrane followed by transcytosis to the apical membrane as an activated protease. Thus, the complex itinerary of prostasin, position matriptase and prostasin on the same subcellular domain and explain how matriptase can activate prostasin in terminal epidermal differentiation. Whether matriptase act as a direct activator of prostasin in Caco-2 cells, or if other proteases are responsible for the activation is not shown. On the other hand, our proposed mechanism for the matriptase-prostasin interaction does not specify any order of the interaction between the two proteases.

With the generation of matriptase-deficient mice it became clear that matriptase plays a role in global epithelial development and maintenance [4;5]. Matriptase is believed to be at the pinnacle of protease cascade(s) because of its ability to autoactivate [28;29;91]. However, it is still unclear which pathways and components that are involved in matriptase- dependent proteolysis. Matriptase and prostasin are co-expressed in a variety of epithelia besides the epidermis, which has led to speculations of a role for the matriptase-prostasin proteolytic cascade in other epithelia [9]. In genetic analyses to identify new components of matriptase-dependent proteolytic pathway(s) necessary for survival of newborn mice, we identified prostasin to be critical to all matriptase- induced embryonic defects in both HAI-1 and HAI-2 deficient mice (paper II). Paradoxically, our study reports that in placental tissue prostasin acts upstream of matriptase and is required for the

86 conversion of zymogen matriptase to active matriptase, which is contrary to what is previously described for epidermis [15-17;21]. To consolidate this in vivo finding, we show that prostasin is able to efficiently cleave and activate zymogen matriptase in a recombinant cell based assay. Interestingly, our data is supported by cell culture studies where addition of active prostasin is sufficient to activate zymogen matriptase expressed recombinantly in KOLF cells as well as in HaCaT cells that have an endogenous expression of matriptase [57].

Fig. 13. Outline of matriptase-prostasin interaction in the epidermis and prostasin-matriptase interaction in placental tissue. In the epidermis, matriptase acts upstream of prostasin and is required for its action. In placental tissue the relationship between the two proteases is reversed; thus prostasin is the upstream protease and required for matriptase activation.

Thus, prostasin may act both as a downstream target but also as an upstream activator of matriptase in a tissue specific manner, see fig. 13, and/or perhaps constitute a positive feedback loop similar to the amplification protease cascades operating in coagulation [153]. Further studies in different epithelial tissues are highly desirable to determine the tissue-specific interplay between the two proteases, and to identify new components and pathways dependent on matriptase proteolytic activity in global epithelial homeostasis. It would be interesting to assess the presence of active matriptase and active prostasin in prostasin-deficient and matriptase-

87 deficient intestinal tissue, respectively, as this is the epithelial tissue to which Caco-2 cells has the highest degree of resemblance. Still regardless of whether matriptase activates prostasin or vice versa, the fact that both active matriptase and active prostasin locate on the basolateral plasma membrane of Caco-2 cells as outlined in paper I can still apply to explain how two proteases with different steady state localization may interact. Thus it is feasible that in placental tissue, prostasin could, upon activation on the basolateral plasma membrane, activate matriptase before being transcytosed to the apical plasma membrane.

The finding that prostasin act upstream of matriptase and is required for its activation is puzzling in regards to the well-documented ability of matriptase to autoactivate and the fact that prostasin is incapable of autoactivation due to an unfavorable isoleucine in the P1' position of the activation cleavage motif of prostasin [107]. It also raises the obvious question of the upstream activator of prostasin in this setting. Another activator of prostasin is the membrane-anchored serine protease hepsin [65]. However, hepsin deficient mice exhibit normal embryogenesis and adult mice display no aberrant phenotype suggesting that another protease is involved in prostasin activation [154;155]. Establishment of biological functions of the rest of the members of the family of membrane-anchored serine proteases could provide important information for determining an upstream activator of prostasin during embryogenesis.

Another interesting aspect and additional layer of complexity of matriptase-prostasin interplay comes from studies on ENaC activation. Both matriptase and prostasin are able to activate ENaC in model systems [103;120]. Although membrane anchorage of prostasin is needed for activation of ENaC in xenopus oocytes, the catalytic activity of prostasin has been reported to be dispensable for the activational cleavage of ENaC [123;124;156]. This could suggest a role for prostasin as a co-factor important for the proteolytic activity of another protease towards ENaC. One could speculate that matriptase participate in prostasin-dependent activation of ENaC. In paper I we were able to detect matriptase on the apical membrane, and we and others have observed shedding of matriptase into the apical media, as well as matriptase is present in human milk (unpublished results, Stine Friis; [44;78]). These findings all indicate that matriptase in fact is present on the apical plasma membrane. Moreover, by means of the assay established in paper III, we have shown that active matriptase is present on the apical membrane (unpublished results, Sine Godiksen). This could position matriptase on the apical plasma membrane to activate ENaC in a prostasin-dependent manner. Perhaps prostasin could have a general role as a co-factor, as prostasin has been shown to enhances matriptase cleavage of EGFR in FT293 cells with no direct cleavage of EGFR by active prostasin [65]. In this regard, it would also be interesting to examine the necessity for prostasin catalytic activity in the dependence of GPI-anchored prostasin in plasmin-stimulated ENaC activation [157]. The role of prostasin as a co-factor could be addressed by the generation of tissue-specific knock in mouse models of catalytic dead prostasin by mutation of the active site serine. Thus, the association between matriptase and prostasin is complex and a great amount of work is needed to unravel the interrelationship of the two proteases in different tissues and cells types.

This thesis also aimed at detecting active matriptase and determining in what subcellular compartment matriptase is proteolytically processed and activated, and thus able to cleave downstream substrate(s). In most cell lysates and tissue extracts, matriptase is found in either its zymogen form or in a complex with its inhibitor HAI‐1 [30;98]. In cell cultures of epithelial and carcinoma origin, matriptase activation can be induced under slight acidic conditions. This acid- induced activation of matriptase is fast and efficient and is followed by HAI-1 inhibition within few minutes [30;31;98;99]. This intimate link between activation and inhibition of matriptase leaves a small window of action for the protease to cleave its substrates and poses a challenge in detecting matriptase in its free active form. Different methods have been used in the three enclosed manuscripts to assess activation of matriptase and active matriptase. In paper II, we assay for the presence of active matriptase and active prostasin by co-IP with HAI-1. However, this does not give an assessment of active protease, but merely inform of whether activation site cleavage has occurred, as we precipitate HAI-1-complexed prostasin and HAI-1-complexed matriptase. In paper I, we use inhibitor-coupled sepharose to assess the level of active matriptase. The non-inhibitory LDLR domain of HAI-1 is important for the matriptase-HAI-1 interaction [28]. For this reason, we used non-endogenous inhibitors of matriptase and prostasin to avoid possible non-inhibitory interactions and thus to target only free active matriptase and prostasin. A shortcoming of this procedure is that the interaction between matriptase and the inhibitor-coupled beads occurs in cell lysates, as matriptase is sensitive to autoactivation in cell free systems [29].

This prompted us to establish an assay for the detection of active matriptase in live cell cultures (paper III). The difficulties in setting up a specific assay for the detection of matriptase come from the great substrate overlap within this family and the absence of a specific inhibitor and/or specific substrate of matriptase [48;50]. We have based our assay on a chloromethyl ketone peptide inhibitor with a tetra peptide sequence based on a preferred substrate sequence of matriptase [50]. CMK peptide inhibitors have long been used in protease research and one of the great advantages is that they are active site inhibitors and bind proteases in a covalent manner by alkylation of the active site histidine [158;159]. Although some specificity can be obtained by altering the peptide sequence of this type of inhibitor, it is very difficult to get absolute specificity towards a single protease [158]. To circumvent the issue of substrate overlap, we employ specific antibodies and Western blot analysis in the detection of active matriptase. We show that active matriptase is present on the basolateral plasma membrane but only constitute a minor fraction of total surface-associated matriptase although we use vast amounts of chloromethyl ketone peptide inhibitor (paper III). However, the absence of a linker arm in our design of chloromethyl ketone peptide inhibitor could cause a reduced affinity for streptavidin and hereby a reduced detection of biotin-RQRR-Cmk labeled matriptase. For the serine protease thrombin, a similar assay set-up showed that a linker arm of 7-14 atoms produced the most sensitive detection in Western blotting [160]. Moreover, our results indicate that biotin-RQRR-Cmk competes with HAI-1 for matriptase binding however; the precipitation step used prevents us from directly assessing this. Instead, applying monomeric avidin that bind biotin in a reversible manner would allow us to examine this. Interestingly, we also find that the zymogen matriptase has intrinsic activity. This is not an uncommon feature of serine proteases and has been shown for i.a. single-chain tPA [161]. The

89 increase in catalytic activity after zymogen activation varies widely among the different members of the trypsin protease family, however for single-chain and two-chain tPA the catalytic activities vary by a factor of only 3-9 [161-165]. It would be interestingly to determine the catalytic capacity of zymogen matriptase; if strong the zymogen form of matriptase could perhaps be the biological relevant form in light of the intimate link between activation site cleavage of matriptase and subsequent inhibition by HAI-1.

Whereas expression studies give valuable insights on matriptase localization and matriptase mRNA/protein levels, these findings does not reflect the biological active state of the protease. Having an assay that enables us to assess the level of active matriptase would be valuable in determining the activation degree of matriptase under different physiological and pathological states. This is particular the case in assessment of matriptase´s contribution in progression of cancer. However, applying the assay presented in paper III on tissue extracts would require some adjustments and also present challenges as matriptase is prone to spontaneous autoactivation in cell free systems [29]. Additionally it would be desirable with a specific molecular probe for active matriptase that would also allow visualization of matriptase activity in pathological conditions.

There have been other attempts to identify inhibitors of matriptase. These inhibitors include bis- benzamidines, sun flower seed trypsin inhibitor-1 derived peptides, antibody-derived inhibitors, and small molecule peptidyl- derivatives inhibitors [149;166-175]. However, not all inhibitors are specific for matriptase, e.g. the chemical inhibitor (CVS-3983) of matriptase that was reported to suppress the prostate cancer growth in nude mice has a high selectivity but is not specific for matriptase [167]. On the other hand, antibodies constructs have been used for detection of matriptase positive cancer cells in a mouse xenografts model [166]. Recently an assay was reported that measured the activity of matriptase on epithelial airway cells. This study use a different approach than ours and assign proteolytic activity to matriptase by the difference in measured activity with and without addition of a non-commercial matriptase inactivating antibody [149]. Even so, despite the methods described above there is still a shortage for specific biochemical assays for detection of active matriptase.

This thesis reports how complex subcellular trafficking of prostasin enables this protease to interact with matriptase on the basolateral plasma membrane of polarized epithelial cells and consolidate how HAI-1 can function as an inhibitor of both proteases. We have shown the subcellular location for matriptase activation and established an assay for detection of active matriptase on the surface of living cells. Moreover, this thesis describes a prostasin-matriptase cascade important for morphogenesis in mice. Together the data presented here has improved our understanding of matriptase´s role in epithelial physiology.

References

[1] Vine AK (2009) Recent advances in haemostasis and thrombosis. Retina, 29, 1-7.

[2] Szabo R, Wu Q, Dickson RB, Netzel-Arnett S, Antalis TM, & Bugge TH (2003) Type II transmembrane serine proteases. Thromb. Haemost., 90, 185-193.

[3] Szabo R & Bugge TH (2011) Membrane-anchored serine proteases in vertebrate cell and developmental biology. Annu. Rev. Cell Dev. Biol., 27, 213-235.

[4] List K, Haudenschild CC, Szabo R, Chen W, Wahl SM, Swaim W, Engelholm LH, Behrendt N, & Bugge TH (2002) Matriptase/MT-SP1 is required for postnatal survival, epidermal barrier function, hair follicle development, and thymic homeostasis. Oncogene, 21, 3765-3779.

[5] List K, Kosa P, Szabo R, Bey AL, Wang CB, Molinolo A, & Bugge TH (2009) Epithelial integrity is maintained by a matriptase-dependent proteolytic pathway. Am. J. Pathol., 175, 1453-1463.

[6] List K, Szabo R, Molinolo A, Sriuranpong V, Redeye V, Murdock T, Burke B, Nielsen BS, Gutkind JS, & Bugge TH (2005) Deregulated matriptase causes ras-independent multistage carcinogenesis and promotes ras-mediated malignant transformation. Genes Dev., 19, 1934- 1950.

[7] Szabo R, Rasmussen AL, Moyer AB, Kosa P, Schafer JM, Molinolo AA, Gutkind JS, & Bugge TH (2011) c-Met-induced epithelial carcinogenesis is initiated by the serine protease matriptase. Oncogene.

[8] Kosa P, Szabo R, Molinolo AA, & Bugge TH (2011) Suppression of Tumorigenicity-14, encoding matriptase, is a critical suppressor of colitis and colitis-associated colon carcinogenesis. Oncogene.

[9] List K, Hobson JP, Molinolo A, & Bugge TH (2007) Co-localization of the channel activating protease prostasin/(CAP1/PRSS8) with its candidate activator, matriptase. J. Cell Physiol, 213, 237-245.

[10] Oberst M, Anders J, Xie B, Singh B, Ossandon M, Johnson M, Dickson RB, & Lin CY (2001) Matriptase and HAI-1 are expressed by normal and malignant epithelial cells in vitro and in vivo. Am. J. Pathol., 158, 1301-1311.

[11] Oberst MD, Singh B, Ozdemirli M, Dickson RB, Johnson MD, & Lin CY (2003) Characterization of matriptase expression in normal human tissues. J. Histochem. Cytochem., 51, 1017-1025.

[12] List K, Szabo R, Molinolo A, Nielsen BS, & Bugge TH (2006) Delineation of matriptase protein expression by enzymatic gene trapping suggests diverging roles in barrier function, hair formation, and squamous cell carcinogenesis. Am. J. Pathol., 168, 1513-1525.

[13] Basel-Vanagaite L, Attia R, Ishida-Yamamoto A, Rainshtein L, Ben AD, Lurie R, Pasmanik-Chor M, Indelman M, Zvulunov A, Saban S, Magal N, Sprecher E, & Shohat M (2007) Autosomal recessive ichthyosis with hypotrichosis caused by a mutation in ST14, encoding type II transmembrane serine protease matriptase. Am. J. Hum. Genet., 80, 467-477. 91

[14] Avrahami L, Maas S, Pasmanik-Chor M, Rainshtein L, Magal N, Smitt J, van MJ, Shohat M, & Basel-Vanagaite L (2008) Autosomal recessive ichthyosis with hypotrichosis syndrome: further delineation of the phenotype. Clin. Genet., 74, 47-53.

[15] Alef T, Torres S, Hausser I, Metze D, Tursen U, Lestringant GG, & Hennies HC (2009) Ichthyosis, follicular atrophoderma, and hypotrichosis caused by mutations in ST14 is associated with impaired profilaggrin processing. J. Invest Dermatol., 129, 862-869.

[16] Netzel-Arnett S, Currie BM, Szabo R, Lin CY, Chen LM, Chai KX, Antalis TM, Bugge TH, & List K (2006) Evidence for a matriptase-prostasin proteolytic cascade regulating terminal epidermal differentiation. J. Biol. Chem., 281, 32941-32945.

[17] List K, Currie B, Scharschmidt TC, Szabo R, Shireman J, Molinolo A, Cravatt BF, Segre J, & Bugge TH (2007) Autosomal ichthyosis with hypotrichosis syndrome displays low matriptase proteolytic activity and is phenocopied in ST14 hypomorphic mice. J. Biol. Chem., 282, 36714- 36723.

[18] Desilets A, Beliveau F, Vandal G, McDuff FO, Lavigne P, & Leduc R (2008) Mutation G827R in matriptase causing autosomal recessive ichthyosis with hypotrichosis yields an inactive protease. J. Biol. Chem., 283, 10535-10542.

[19] List K, Szabo R, Wertz PW, Segre J, Haudenschild CC, Kim SY, & Bugge TH (2003) Loss of proteolytically processed filaggrin caused by epidermal deletion of Matriptase/MT-SP1. J. Cell Biol., 163, 901-910.

[20] Szabo R, Kosa P, List K, & Bugge TH (2009) Loss of matriptase suppression underlies spint1 mutation-associated ichthyosis and postnatal lethality. Am. J. Pathol., 174, 2015-2022.

[21] Chen YW, Wang JK, Chou FP, Chen CY, Rorke EA, Chen LM, Chai KX, Eckert RL, Johnson MD, & Lin CY (2010) Regulation of the matriptase-prostasin cell surface proteolytic cascade by hepatocyte growth factor activator inhibitor-1 during epidermal differentiation. J. Biol. Chem., 285, 31755-31762.

[22] Carney TJ, von der HS, Sonntag C, Amsterdam A, Topczewski J, Hopkins N, & Hammerschmidt M (2007) Inactivation of serine protease Matriptase1a by its inhibitor Hai1 is required for epithelial integrity of the zebrafish epidermis. Development, 134, 3461-3471.

[23] Verghese GM, Gutknecht MF, & Caughey GH (2006) Prostasin regulates epithelial monolayer function: cell-specific Gpld1-mediated secretion and functional role for GPI anchor. Am. J. Physiol Cell Physiol, 291, C1258-C1270.

[24] Kataoka H, Suganuma T, Shimomura T, Itoh H, Kitamura N, Nabeshima K, & Koono M (1999) Distribution of hepatocyte growth factor activator inhibitor type 1 (HAI-1) in human tissues. Cellular surface localization of HAI-1 in simple columnar epithelium and its modulated expression in injured and regenerative tissues. J. Histochem. Cytochem., 47, 673-682.

[25] Kataoka H, Meng JY, Itoh H, Hamasuna R, Shimomura T, Suganuma T, & Koono M (2000) Localization of hepatocyte growth factor activator inhibitor type 1 in Langhans' cells of human placenta. Histochem. Cell Biol., 114, 469-475.

[26] Satomi S, Yamasaki Y, Tsuzuki S, Hitomi Y, Iwanaga T, & Fushiki T (2001) A role for membrane-type serine protease (MT-SP1) in intestinal epithelial turnover. Biochem. Biophys. Res. Commun., 287, 995-1002.

[27] Godiksen S, Selzer-Plon J, Pedersen ED, Abell K, Rasmussen HB, Szabo R, Bugge TH, & Vogel LK (2008) Hepatocyte growth factor activator inhibitor-1 has a complex subcellular itinerary. Biochem. J., 413, 251-259.

[28] Oberst MD, Williams CA, Dickson RB, Johnson MD, & Lin CY (2003) The activation of matriptase requires its noncatalytic domains, serine protease domain, and its cognate inhibitor. J. Biol. Chem., 278, 26773-26779.

[29] Lee MS, Tseng IC, Wang Y, Kiyomiya K, Johnson MD, Dickson RB, & Lin CY (2007) Autoactivation of matriptase in vitro: requirement for biomembrane and LDL receptor domain. Am. J. Physiol Cell Physiol, 293, C95-105.

[30] Lee MS, Kiyomiya K, Benaud C, Dickson RB, & Lin CY (2005) Simultaneous activation and hepatocyte growth factor activator inhibitor 1-mediated inhibition of matriptase induced at activation foci in human mammary epithelial cells. Am. J. Physiol Cell Physiol, 288, C932- C941.

[31] Benaud C, Dickson RB, & Lin CY (2001) Regulation of the activity of matriptase on epithelial cell surfaces by a blood-derived factor. Eur. J. Biochem., 268, 1439-1447.

[32] Lopez-Otin C & Matrisian LM (2007) Emerging roles of proteases in tumour suppression. Nat. Rev. Cancer, 7, 800-808.

[33] Hedstrom L (2002) Serine protease mechanism and specificity. Chem. Rev., 102, 4501-4524.

[34] Bugge TH, Antalis TM, & Wu Q (2009) Type II transmembrane serine proteases. J. Biol. Chem., 284, 23177-23181.

[35] Netzel-Arnett S, Hooper JD, Szabo R, Madison EL, Quigley JP, Bugge TH, & Antalis TM (2003) Membrane anchored serine proteases: a rapidly expanding group of cell surface proteolytic enzymes with potential roles in cancer. Cancer Metastasis Rev., 22, 237-258.

[36] Hooper JD, Clements JA, Quigley JP, & Antalis TM (2001) Type II transmembrane serine proteases. Insights into an emerging class of cell surface proteolytic enzymes. J. Biol. Chem., 276, 857-860.

[37] Szabo R & Bugge TH (2008) Type II transmembrane serine proteases in development and disease. Int. J. Biochem. Cell Biol., 40, 1297-1316.

[38] Kim MG, Chen C, Lyu MS, Cho EG, Park D, Kozak C, & Schwartz RH (1999) Cloning and chromosomal mapping of a gene isolated from thymic stromal cells encoding a new mouse type II membrane serine protease, epithin, containing four LDL receptor modules and two CUB domains. Immunogenetics, 49, 420-428.

[39] Lin CY, Anders J, Johnson M, Sang QA, & Dickson RB (1999) Molecular cloning of cDNA for matriptase, a matrix-degrading serine protease with trypsin-like activity. J. Biol. Chem., 274, 18231-18236.

[40] Takeuchi T, Shuman MA, & Craik CS (1999) Reverse biochemistry: use of macromolecular protease inhibitors to dissect complex biological processes and identify a membrane-type serine protease in epithelial cancer and normal tissue. Proc. Natl. Acad. Sci. U. S. A, 96, 11054-11061.

[41] Cao J, Cai X, Zheng L, Geng L, Shi Z, Pao CC, & Zheng S (1997) Characterization of colorectal- cancer-related cDNA clones obtained by subtractive hybridization screening. J. Cancer Res. Clin. Oncol., 123, 447-451.

[42] Underwood LJ, Tanimoto H, Wang Y, Shigemasa K, Parmley TH, & O'Brien TJ (1999) Cloning of tumor-associated differentially expressed gene-14, a novel serine protease overexpressed by ovarian carcinoma. Cancer Res., 59, 4435-4439.

[43] Shi YE, Torri J, Yieh L, Wellstein A, Lippman ME, & Dickson RB (1993) Identification and characterization of a novel matrix-degrading protease from hormone-dependent human breast cancer cells. Cancer Res., 53, 1409-1415.

[44] Lin CY, Anders J, Johnson M, & Dickson RB (1999) Purification and characterization of a complex containing matriptase and a Kunitz-type serine protease inhibitor from human milk. J. Biol. Chem., 274, 18237-18242.

[45] Velasco G, Cal S, Quesada V, Sanchez LM, & Lopez-Otin C (2002) Matriptase-2, a membrane- bound mosaic serine proteinase predominantly expressed in human liver and showing degrading activity against extracellular matrix proteins. J. Biol. Chem., 277, 37637-37646.

[46] Szabo R, Netzel-Arnett S, Hobson JP, Antalis TM, & Bugge TH (2005) Matriptase-3 is a novel phylogenetically preserved membrane-anchored serine protease with broad serpin reactivity. Biochem. J., 390, 231-242.

[47] Ge W, Hu H, Ding K, Sun L, & Zheng S (2006) Protein interaction analysis of ST14 domains and their point and deletion mutants. J. Biol. Chem., 281, 7406-7412.

[48] Antalis TM, Buzza MS, Hodge KM, Hooper JD, & Netzel-Arnett S (2010) The cutting edge: membrane-anchored serine protease activities in the pericellular microenvironment. Biochem. J., 428, 325-346.

[49] Hobson JP, Netzel-Arnett S, Szabo R, Rehault SM, Church FC, Strickland DK, Lawrence DA, Antalis TM, & Bugge TH (2004) Mouse DESC1 is located within a cluster of seven DESC1-like genes and encodes a type II transmembrane serine protease that forms serpin inhibitory complexes. J. Biol. Chem., 279, 46981-46994.

[50] Beliveau F, Desilets A, & Leduc R (2009) Probing the substrate specificities of matriptase, matriptase-2, hepsin and DESC1 with internally quenched fluorescent peptides. FEBS J., 276, 2213-2226.

[51] Takeuchi T, Harris JL, Huang W, Yan KW, Coughlin SR, & Craik CS (2000) Cellular localization of membrane-type serine protease 1 and identification of protease-activated receptor-2 and single-chain urokinase-type plasminogen activator as substrates. J. Biol. Chem., 275, 26333- 26342.

[52] Friedrich R, Fuentes-Prior P, Ong E, Coombs G, Hunter M, Oehler R, Pierson D, Gonzalez R, Huber R, Bode W, & Madison EL (2002) Catalytic domain structures of MT-SP1/matriptase, a matrix-degrading transmembrane serine proteinase. J. Biol. Chem., 277, 2160-2168.

[53] Lee SL, Dickson RB, & Lin CY (2000) Activation of hepatocyte growth factor and urokinase/plasminogen activator by matriptase, an epithelial membrane serine protease. J. Biol. Chem., 275, 36720-36725.

[54] Owen KA, Qiu D, Alves J, Schumacher AM, Kilpatrick LM, Li J, Harris JL, & Ellis V (2010) Pericellular activation of hepatocyte growth factor by the transmembrane serine proteases matriptase and hepsin, but not by the membrane-associated protease uPA. Biochem. J., 426, 219-228.

[55] Bhatt AS, Welm A, Farady CJ, Vasquez M, Wilson K, & Craik CS (2007) Coordinate expression and functional profiling identify an extracellular proteolytic signaling pathway. Proc. Natl. Acad. Sci. U. S. A, 104, 5771-5776.

[56] Kilpatrick LM, Harris RL, Owen KA, Bass R, Ghorayeb C, Bar-Or A, & Ellis V (2006) Initiation of plasminogen activation on the surface of monocytes expressing the type II transmembrane serine protease matriptase. Blood, 108, 2616-2623.

[57] Camerer E, Barker A, Duong DN, Ganesan R, Kataoka H, Cornelissen I, Darragh MR, Hussain A, Zheng YW, Srinivasan Y, Brown C, Xu SM, Regard JB, Lin CY, Craik CS, Kirchhofer D, & Coughlin SR (2010) Local protease signaling contributes to neural tube closure in the mouse embryo. Dev. Cell, 18, 25-38.

[58] Jin X, Yagi M, Akiyama N, Hirosaki T, Higashi S, Lin CY, Dickson RB, Kitamura H, & Miyazaki K (2006) Matriptase activates stromelysin (MMP-3) and promotes tumor growth and angiogenesis. Cancer Sci., 97, 1327-1334.

[59] Sales KU, Masedunskas A, Bey AL, Rasmussen AL, Weigert R, List K, Szabo R, Overbeek PA, & Bugge TH (2010) Matriptase initiates activation of epidermal pro-kallikrein and disease onset in a mouse model of Netherton syndrome. Nat. Genet., 42, 676-683.

[60] Bhatt AS, Erdjument-Bromage H, Tempst P, Craik CS, & Moasser MM (2005) Adhesion signaling by a novel mitotic substrate of src kinases. Oncogene, 24, 5333-5343.

[61] Ahmed S, Jin X, Yagi M, Yasuda C, Sato Y, Higashi S, Lin CY, Dickson RB, & Miyazaki K (2006) Identification of membrane-bound serine proteinase matriptase as processing enzyme of insulin-like growth factor binding protein-related protein-1 (IGFBP- rP1/angiomodulin/mac25). FEBS J., 273, 615-627.

[62] Darragh MR, Bhatt AS, & Craik CS (2008) MT-SP1 proteolysis and regulation of cell- microenvironment interactions. Front Biosci., 13, 528-539.

[63] Hurst NJ, Jr., Najy AJ, Ustach CV, Movilla L, & Kim HR (2012) Platelet-derived growth factor-C (PDGF-C) activation by serine proteases: implications for breast cancer progression. Biochem. J., 441, 909-918.

[64] Ustach CV, Huang W, Conley-Lacomb MK, Lin CY, Che M, Abrams J, & Kim HR (2010) A Novel Signaling Axis of Matriptase/PDGF-D/{beta}-PDGFR in Human Prostate Cancer. Cancer Res., 70, 9631-9640.

[65] Chen M, Chen LM, Lin CY, & Chai KX (2009) Hepsin activates prostasin and cleaves the extracellular domain of the epidermal growth factor receptor. Mol. Cell Biochem..

[66] Chen M, Chen LM, Lin CY, & Chai KX (2008) The epidermal growth factor receptor (EGFR) is proteolytically modified by the Matriptase-Prostasin serine protease cascade in cultured epithelial cells. Biochim. Biophys. Acta, 1783, 896-903.

[67] Tripathi M, Potdar AA, Yamashita H, Weidow B, Cummings PT, Kirchhofer D, & Quaranta V (2011) Laminin-332 cleavage by matriptase alters motility parameters of prostate cancer cells. Prostate, 71, 184-196.

[68] Kim C, Cho Y, Kang CH, Kim MG, Lee H, Cho EG, & Park D (2005) Filamin is essential for shedding of the transmembrane serine protease, epithin. EMBO Rep., 6, 1045-1051.

[69] Gal P & Zavodszky P (1998) Structure and function of the serine-protease subcomponents of C1: protein engineering studies. Immunobiology, 199, 317-326.

[70] Janciauskiene S, Nita I, Subramaniyam D, Li Q, Lancaster JR, Jr., & Matalon S (2008) Alpha1- antitrypsin inhibits the activity of the matriptase catalytic domain in vitro. Am. J. Respir. Cell Mol. Biol., 39, 631-637.

[71] Tseng IC, Chou FP, Su SF, Oberst M, Madayiputhiya N, Lee MS, Wang JK, Sloane DE, Johnson M, & Lin CY (2008) Purification from human milk of matriptase complexes with secreted serpins: mechanism for inhibition of matriptase other than HAI-1. Am. J. Physiol Cell Physiol, 295, C423-C431.

[72] Chou FP, Xu H, Lee MS, Chen YW, Richards OX, Swanson R, Olson ST, Johnson MD, & Lin CY (2011) Matriptase is inhibited by extravascular antithrombin in epithelial cells but not in most carcinoma cells. Am. J. Physiol Cell Physiol, 301, C1093-C1103.

[73] Yamasaki Y, Satomi S, Murai N, Tsuzuki S, & Fushiki T (2003) Inhibition of membrane-type serine protease 1/matriptase by natural and synthetic protease inhibitors. J. Nutr. Sci. Vitaminol. (Tokyo), 49, 27-32.

[74] Myerburg MM, McKenna EE, Luke CJ, Frizzell RA, Kleyman TR, & Pilewski JM (2008) Prostasin expression is regulated by airway surface liquid volume and is increased in cystic fibrosis. Am. J. Physiol Lung Cell Mol. Physiol, 294, L932-L941.

[75] Kataoka H, Itoh H, & Koono M (2002) Emerging multifunctional aspects of cellular serine proteinase inhibitors in tumor progression and tissue regeneration. Pathol. Int., 52, 89-102.

[76] Laskowski M, Jr. & Kato I (1980) Protein inhibitors of proteinases. Annu. Rev. Biochem., 49, 593-626.

[77] Goder V & Spiess M (2001) Topogenesis of membrane proteins: determinants and dynamics. FEBS Lett., 504, 87-93.

[78] Wang JK, Lee MS, Tseng IC, Chou FP, Chen YW, Fulton A, Lee HS, Chen CJ, Johnson MD, & Lin CY (2009) Polarized epithelial cells secrete matriptase as a consequence of zymogen activation and HAI-1-mediated inhibition. Am. J. Physiol Cell Physiol, 297, C459-C470.

[79] Szabo R, Hobson JP, List K, Molinolo A, Lin CY, & Bugge TH (2008) Potent inhibition and global co-localization implicate the transmembrane Kunitz-type serine protease inhibitor hepatocyte growth factor activator inhibitor-2 in the regulation of epithelial matriptase activity. J. Biol. Chem., 283, 29495-29504.

[80] Shimomura T, Denda K, Kitamura A, Kawaguchi T, Kito M, Kondo J, Kagaya S, Qin L, Takata H, Miyazawa K, & Kitamura N (1997) Hepatocyte growth factor activator inhibitor, a novel Kunitz-type serine protease inhibitor. J. Biol. Chem., 272, 6370-6376.

[81] Kawaguchi T, Qin L, Shimomura T, Kondo J, Matsumoto K, Denda K, & Kitamura N (1997) Purification and cloning of hepatocyte growth factor activator inhibitor type 2, a Kunitz-type serine protease inhibitor. J. Biol. Chem., 272, 27558-27564.

[82] Akiyama Y, Nagai M, Komaki W, Marutsuka K, Asada Y, & Kataoka H (2006) Expression of hepatocyte growth factor activator inhibitor type 1 in endothelial cells. Hum. Cell, 19, 91-97.

[83] Szabo R, Hobson JP, Christoph K, Kosa P, List K, & Bugge TH (2009) Regulation of cell surface protease matriptase by HAI2 is essential for placental development, neural tube closure and embryonic survival in mice. Development, 136, 2653-2663.

[84] Denda K, Shimomura T, Kawaguchi T, Miyazawa K, & Kitamura N (2002) Functional characterization of Kunitz domains in hepatocyte growth factor activator inhibitor type 1. J. Biol. Chem., 277, 14053-14059.

[85] Girard TJ, Warren LA, Novotny WF, Likert KM, Brown SG, Miletich JP, & Broze GJ, Jr. (1989) Functional significance of the Kunitz-type inhibitory domains of lipoprotein-associated coagulation inhibitor. Nature, 338, 518-520.

[86] Denda K, Shimomura T, Kawaguchi T, Miyazawa K, & Kitamura N (2002) Functional characterization of Kunitz domains in hepatocyte growth factor activator inhibitor type 1. J. Biol. Chem., 277, 14053-14059.

[87] Qin L, Denda K, Shimomura T, Kawaguchi T, & Kitamura N (1998) Functional characterization of Kunitz domains in hepatocyte growth factor activator inhibitor type 2. FEBS Lett., 436, 111-114.

[88] Kojima K, Tsuzuki S, Fushiki T, & Inouye K (2008) Roles of functional and structural domains of hepatocyte growth factor activator inhibitor type 1 in the inhibition of matriptase. J. Biol. Chem., 283, 2478-2487.

[89] Levitin F, Stern O, Weiss M, Gil-Henn C, Ziv R, Prokocimer Z, Smorodinsky NI, Rubinstein DB, & Wreschner DH (2005) The MUC1 SEA module is a self-cleaving domain. J. Biol. Chem., 280, 33374-33386.

[90] Cho EG, Kim MG, Kim C, Kim SR, Seong IS, Chung C, Schwartz RH, & Park D (2001) N-terminal processing is essential for release of epithin, a mouse type II membrane serine protease. J. Biol. Chem., 276, 44581-44589.

[91] Miyake Y, Yasumoto M, Tsuzuki S, Fushiki T, & Inouye K (2009) Activation of a membrane- bound serine protease matriptase on the cell surface. J. Biochem., 146, 273-282.

[92] Bode W, Schwager P, & Huber R (1978) The transition of bovine trypsinogen to a trypsin-like state upon strong ligand binding. The refined crystal structures of the bovine trypsinogen- pancreatic trypsin inhibitor complex and of its ternary complex with Ile-Val at 1.9 A resolution. J. Mol. Biol., 118, 99-112.

[93] Qiu D, Owen K, Gray K, Bass R, & Ellis V (2007) Roles and regulation of membrane-associated serine proteases. Biochem. Soc. Trans., 35, 583-587.

[94] Thielens NM, Bersch B, Hernandez JF, & Arlaud GJ (1999) Structure and functions of the interaction domains of C1r and C1s: keystones of the architecture of the C1 complex. Immunopharmacology, 42, 3-13.

[95] Miyake Y, Tsuzuki S, Mochida S, Fushiki T, & Inouye K (2010) The role of asparagine-linked glycosylation site on the catalytic domain of matriptase in its zymogen activation. Biochim. Biophys. Acta, 1804, 156-165.

[96] Oberst MD, Chen LY, Kiyomiya K, Williams CA, Lee MS, Johnson MD, Dickson RB, & Lin CY (2005) HAI-1 regulates activation and expression of matriptase, a membrane-bound serine protease. Am. J. Physiol Cell Physiol, 289, C462-C470.

[97] Kiyomiya K, Lee MS, Tseng IC, Zuo H, Barndt RJ, Johnson MD, Dickson RB, & Lin CY (2006) Matriptase activation and shedding with HAI-1 is induced by steroid sex hormones in human prostate cancer cells, but not in breast cancer cells. Am. J. Physiol Cell Physiol, 291, C40-C49.

[98] Tseng IC, Xu H, Chou FP, Li G, Vazzano AP, Kao JP, Johnson MD, & Lin CY (2010) Matriptase activation, an early cellular response to acidosis. J. Biol. Chem., 285, 3261-3270.

[99] Inouye K, Yasumoto M, Tsuzuki S, Mochida S, & Fushiki T (2010) The optimal activity of a pseudozymogen form of recombinant matriptase under the mildly acidic pH and low ionic strength conditions. J. Biochem., 147, 485-492.

[100] Cho EG, Schwartz RH, & Kim MG (2005) Shedding of membrane epithin is blocked without LDLRA4 and its protease activation site. Biochem. Biophys. Res. Commun., 327, 328-334.

[101] Yu JX, Chao L, & Chao J (1994) Prostasin is a novel human serine proteinase from seminal fluid. Purification, tissue distribution, and localization in prostate gland. J. Biol. Chem., 269, 18843-18848.

[102] Yu JX, Chao L, & Chao J (1995) Molecular cloning, tissue-specific expression, and cellular localization of human prostasin mRNA. J. Biol. Chem., 270, 13483-13489.

[103] Vallet V, Chraibi A, Gaeggeler HP, Horisberger JD, & Rossier BC (1997) An epithelial serine protease activates the amiloride-sensitive sodium channel. Nature, 389, 607-610.

[104] Adachi M, Kitamura K, Miyoshi T, Narikiyo T, Iwashita K, Shiraishi N, Nonoguchi H, & Tomita K (2001) Activation of epithelial sodium channels by prostasin in Xenopus oocytes. J. Am. Soc. Nephrol., 12, 1114-1121.

[105] Chen LM, Skinner ML, Kauffman SW, Chao J, Chao L, Thaler CD, & Chai KX (2001) Prostasin is a glycosylphosphatidylinositol-anchored active serine protease. J. Biol. Chem., 276, 21434- 21442.

[106] Rickert KW, Kelley P, Byrne NJ, Diehl RE, Hall DL, Montalvo AM, Reid JC, Shipman JM, Thomas BW, Munshi SK, Darke PL, & Su HP (2008) Structure of human prostasin, a target for the regulation of hypertension. J. Biol. Chem., 283, 34864-34872.

[107] Shipway A, Danahay H, Williams JA, Tully DC, Backes BJ, & Harris JL (2004) Biochemical characterization of prostasin, a channel activating protease. Biochem. Biophys. Res. Commun., 324, 953-963.

[108] Fan B, Wu TD, Li W, & Kirchhofer D (2005) Identification of hepatocyte growth factor activator inhibitor-1B as a potential physiological inhibitor of prostasin. J. Biol. Chem., 280, 34513-34520.

[109] Chen LM, Zhang X, & Chai KX (2004) Regulation of prostasin expression and function in the prostate. Prostate, 59, 1-12.

[110] Scharschmidt TC, List K, Grice EA, Szabo R, Renaud G, Lee CC, Wolfsberg TG, Bugge TH, & Segre JA (2009) Matriptase-deficient mice exhibit ichthyotic skin with a selective shift in skin microbiota. J. Invest Dermatol., 129, 2435-2442.

[111] Suzuki Y, Nomura J, Koyama J, & Horii I (1994) The role of proteases in stratum corneum: involvement in stratum corneum desquamation. Arch. Dermatol. Res., 286, 249-253.

[112] Komatsu N, Takata M, Otsuki N, Ohka R, Amano O, Takehara K, & Saijoh K (2002) Elevated stratum corneum hydrolytic activity in Netherton syndrome suggests an inhibitory regulation of desquamation by SPINK5-derived peptides. J. Invest Dermatol., 118, 436-443.

[113] Chavanas S, Bodemer C, Rochat A, Hamel-Teillac D, Ali M, Irvine AD, Bonafe JL, Wilkinson J, Taieb A, Barrandon Y, Harper JI, de PY, & Hovnanian A (2000) Mutations in SPINK5, encoding a serine protease inhibitor, cause Netherton syndrome. Nat. Genet., 25, 141-142.

[114] Hewett DR, Simons AL, Mangan NE, Jolin HE, Green SM, Fallon PG, & McKenzie AN (2005) Lethal, neonatal ichthyosis with increased proteolytic processing of filaggrin in a mouse model of Netherton syndrome. Hum. Mol. Genet., 14, 335-346.

[115] Chen CJ, Wu BY, Tsao PI, Chen CY, Wu MH, Chan YL, Lee HS, Johnson MD, Eckert RL, Chen YW, Chou F, Wang JK, & Lin CY (2010) Increased matriptase zymogen activation in inflammatory skin disorders. Am. J. Physiol Cell Physiol.

[116] Leyvraz C, Charles RP, Rubera I, Guitard M, Rotman S, Breiden B, Sandhoff K, & Hummler E (2005) The epidermal barrier function is dependent on the serine protease CAP1/Prss8. J. Cell Biol., 170, 487-496.

[117] Mildner M, Ballaun C, Stichenwirth M, Bauer R, Gmeiner R, Buchberger M, Mlitz V, & Tschachler E (2006) Gene silencing in a human organotypic skin model. Biochem. Biophys. Res. Commun., 348, 76-82.

[118] Miller GS & List K (2012) The matriptase-prostasin proteolytic cascade in epithelial development and pathology. Cell Tissue Res..

[119] Ovaere P, Lippens S, Vandenabeele P, & Declercq W (2009) The emerging roles of serine protease cascades in the epidermis. Trends Biochem. Sci., 34, 453-463.

[120] Vuagniaux G, Vallet V, Jaeger NF, Hummler E, & Rossier BC (2002) Synergistic activation of ENaC by three membrane-bound channel-activating serine proteases (mCAP1, mCAP2, and mCAP3) and serum- and glucocorticoid-regulated kinase (Sgk1) in Xenopus Oocytes. J. Gen. Physiol, 120, 191-201.

[121] Mauro T, Guitard M, Behne M, Oda Y, Crumrine D, Komuves L, Rassner U, Elias PM, & Hummler E (2002) The ENaC channel is required for normal epidermal differentiation. J. Invest Dermatol., 118, 589-594.

[122] Planes C, Randrianarison NH, Charles RP, Frateschi S, Cluzeaud F, Vuagniaux G, Soler P, Clerici C, Rossier BC, & Hummler E (2010) ENaC-mediated alveolar fluid clearance and lung fluid balance depend on the channel-activating protease 1. EMBO Mol. Med., 2, 26-37.

[123] Andreasen D, Vuagniaux G, Fowler-Jaeger N, Hummler E, & Rossier BC (2006) Activation of epithelial sodium channels by mouse channel activating proteases (mCAP) expressed in Xenopus oocytes requires catalytic activity of mCAP3 and mCAP2 but not mCAP1. J. Am. Soc. Nephrol., 17, 968-976.

[124] Vallet V, Pfister C, Loffing J, & Rossier BC (2002) Cell-surface expression of the channel activating protease xCAP-1 is required for activation of ENaC in the Xenopus oocyte. J. Am. Soc. Nephrol., 13, 588-594.

[125] Netzel-Arnett S, Buzza MS, Shea-Donohue T, Desilets A, Leduc R, Fasano A, Bugge TH, & Antalis TM (2011) Matriptase protects against experimental colitis and promotes intestinal barrier recovery. Inflamm. Bowel. Dis..

[126] Fan B, Brennan J, Grant D, Peale F, Rangell L, & Kirchhofer D (2007) Hepatocyte growth factor activator inhibitor-1 (HAI-1) is essential for the integrity of basement membranes in the developing placental labyrinth. Dev. Biol., 303, 222-230.

[127] Tanaka H, Nagaike K, Takeda N, Itoh H, Kohama K, Fukushima T, Miyata S, Uchiyama S, Uchinokura S, Shimomura T, Miyazawa K, Kitamura N, Yamada G, & Kataoka H (2005) Hepatocyte growth factor activator inhibitor type 1 (HAI-1) is required for branching morphogenesis in the chorioallantoic placenta. Mol. Cell Biol., 25, 5687-5698.

[128] Szabo R, Molinolo A, List K, & Bugge TH (2007) Matriptase inhibition by hepatocyte growth factor activator inhibitor-1 is essential for placental development. Oncogene, 26, 1546-1556.

[129] Nagaike K, Kawaguchi M, Takeda N, Fukushima T, Sawaguchi A, Kohama K, Setoyama M, & Kataoka H (2008) Defect of hepatocyte growth factor activator inhibitor type 1/serine protease inhibitor, Kunitz type 1 (Hai-1/Spint1) leads to ichthyosis-like condition and abnormal hair development in mice. Am. J. Pathol., 173, 1464-1475.

[130] Heinz-Erian P, Muller T, Krabichler B, Schranz M, Becker C, Ruschendorf F, Nurnberg P, Rossier B, Vujic M, Booth IW, Holmberg C, Wijmenga C, Grigelioniene G, Kneepkens CM, Rosipal S, Mistrik M, Kappler M, Michaud L, Doczy LC, Siu VM, Krantz M, Zoller H, Utermann G, & Janecke AR (2009) Mutations in SPINT2 cause a syndromic form of congenital sodium diarrhea. Am. J. Hum. Genet., 84, 188-196.

[131] Hanahan D & Weinberg RA (2000) The hallmarks of cancer. Cell, 100, 57-70.

100

[132] Tanimoto H, Underwood LJ, Wang Y, Shigemasa K, Parmley TH, & O'Brien TJ (2001) Ovarian tumor cells express a transmembrane serine protease: a potential candidate for early diagnosis and therapeutic intervention. Tumour. Biol., 22, 104-114.

[133] Santin AD, Cane' S, Bellone S, Bignotti E, Palmieri M, De Las Casas LE, Anfossi S, Roman JJ, O'Brien T, & Pecorelli S (2003) The novel serine protease tumor-associated differentially expressed gene-15 (matriptase/MT-SP1) is highly overexpressed in cervical carcinoma. Cancer, 98, 1898-1904.

[134] Tsai WC, Chu CH, Yu CP, Sheu LF, Chen A, Chiang H, & Jin JS (2008) Matriptase and survivin expression associated with tumor progression and malignant potential in breast cancer of Chinese women: tissue microarray analysis of immunostaining scores with clinicopathological parameters. Dis. Markers, 24, 89-99.

[135] Kang JY, Dolled-Filhart M, Ocal IT, Singh B, Lin CY, Dickson RB, Rimm DL, & Camp RL (2003) Tissue microarray analysis of hepatocyte growth factor/Met pathway components reveals a role for Met, matriptase, and hepatocyte growth factor activator inhibitor 1 in the progression of node-negative breast cancer. Cancer Res., 63, 1101-1105.

[136] Jin JS, Cheng TF, Tsai WC, Sheu LF, Chiang H, & Yu CP (2007) Expression of the serine protease, matriptase, in breast ductal carcinoma of Chinese women: correlation with clinicopathological parameters. Histol. Histopathol., 22, 305-309.

[137] Riddick AC, Shukla CJ, Pennington CJ, Bass R, Nuttall RK, Hogan A, Sethia KK, Ellis V, Collins AT, Maitland NJ, Ball RY, & Edwards DR (2005) Identification of degradome components associated with prostate cancer progression by expression analysis of human prostatic tissues. Br. J. Cancer, 92, 2171-2180.

[138] Saleem M, Adhami VM, Zhong W, Longley BJ, Lin CY, Dickson RB, Reagan-Shaw S, Jarrard DF, & Mukhtar H (2006) A novel biomarker for staging human prostate adenocarcinoma: overexpression of matriptase with concomitant loss of its inhibitor, hepatocyte growth factor activator inhibitor-1. Cancer Epidemiol. Biomarkers Prev., 15, 217-227.

[139] Lee JW, Yong SS, Choi JJ, Lee SJ, Kim BG, Park CS, Lee JH, Lin CY, Dickson RB, & Bae DS (2005) Increased expression of matriptase is associated with histopathologic grades of cervical neoplasia. Hum. Pathol., 36, 626-633.

[140] Zeng L, Cao J, & Zhang X (2005) Expression of serine protease SNC19/matriptase and its inhibitor hepatocyte growth factor activator inhibitor type 1 in normal and malignant tissues of gastrointestinal tract. World J. Gastroenterol., 11, 6202-6207.

[141] Vogel LK, Saebo M, Skjelbred CF, Abell K, Pedersen ED, Vogel U, & Kure EH (2006) The ratio of Matriptase/HAI-1 mRNA is higher in colorectal cancer adenomas and carcinomas than corresponding tissue from control individuals. BMC. Cancer, 6, 176.

[142] Welman A, Sproul D, Mullen P, Muir M, Kinnaird AR, Harrison DJ, Faratian D, Brunton VG, & Frame MC (2012) Diversity of matriptase expression level and function in breast cancer. PLoS. One., 7, e34182.

[143] Zhang Y, Cai X, Schlegelberger B, & Zheng S (1998) Assignment1 of human putative tumor suppressor genes ST13 (alias SNC6) and ST14 (alias SNC19) to human chromosome bands 22q13 and 11q24-->q25 by in situ hybridization. Cytogenet. Cell Genet., 83, 56-57. 101

[144] Kawaguchi M, Takeda N, Hoshiko S, Yorita K, Baba T, Sawaguchi A, Nezu Y, Yoshikawa T, Fukushima T, & Kataoka H (2011) Membrane-bound serine protease inhibitor HAI-1 is required for maintenance of intestinal epithelial integrity. Am. J. Pathol., 179, 1815-1826.

[145] Oberst MD, Johnson MD, Dickson RB, Lin CY, Singh B, Stewart M, Williams A, al-Nafussi A, Smyth JF, Gabra H, & Sellar GC (2002) Expression of the serine protease matriptase and its inhibitor HAI-1 in epithelial ovarian cancer: correlation with clinical outcome and tumor clinicopathological parameters. Clin. Cancer Res., 8, 1101-1107.

[146] Baba T, Kawaguchi M, Fukushima T, Sato Y, Orikawa H, Yorita K, Tanaka H, Lin CY, Sakoda S, & Kataoka H (2012) Loss of membrane-bound serine protease inhibitor HAI-1 induces oral squamous cell carcinoma cells' invasiveness. J. Pathol..

[147] Sambuy Y, de A, I, Ranaldi G, Scarino ML, Stammati A, & Zucco F (2005) The Caco-2 cell line as a model of the intestinal barrier: influence of cell and culture-related factors on Caco-2 cell functional characteristics. Cell Biol. Toxicol., 21, 1-26.

[148] Misfeldt DS, Hamamoto ST, & Pitelka DR (1976) Transepithelial transport in cell culture. Proc. Natl. Acad. Sci. U. S. A, 73, 1212-1216.

[149] Nimishakavi S, Besprozvannaya M, Raymond WW, Craik CS, Gruenert DC, & Caughey GH (2012) Activity and Inhibition of Prostasin and Matriptase on Apical and Basolateral Surfaces of Human Airway Epithelial Cells. Am. J. Physiol Lung Cell Mol. Physiol.

[150] Buzza MS, Netzel-Arnett S, Shea-Donohue T, Zhao A, Lin CY, List K, Szabo R, Fasano A, Bugge TH, & Antalis TM (2010) Membrane-anchored serine protease matriptase regulates epithelial barrier formation and permeability in the intestine. Proc. Natl. Acad. Sci. U. S. A.

[151] Thermo Scientific. Immunoprecipitation Methodology. 2012. Ref Type: Online Source

[152] Thermo Scientific. Thermo Scientific Avidin-Biotin Technical Handbook. 2012. Ref Type: Online Source

[153] Hoffman M & Monroe DM, III (2001) A cell-based model of hemostasis. Thromb. Haemost., 85, 958-965.

[154] Yu IS, Chen HJ, Lee YS, Huang PH, Lin SR, Tsai TW, & Lin SW (2000) Mice deficient in hepsin, a serine protease, exhibit normal embryogenesis and unchanged hepatocyte regeneration ability. Thromb. Haemost., 84, 865-870.

[155] Wu Q, Yu D, Post J, Halks-Miller M, Sadler JE, & Morser J (1998) Generation and characterization of mice deficient in hepsin, a hepatic transmembrane serine protease. J. Clin. Invest, 101, 321-326.

[156] Bruns JB, Carattino MD, Sheng S, Maarouf AB, Weisz OA, Pilewski JM, Hughey RP, & Kleyman TR (2007) Epithelial Na+ channels are fully activated by furin- and prostasin-dependent release of an inhibitory peptide from the gamma-subunit. J. Biol. Chem., 282, 6153-6160.

[157] Svenningsen P, Uhrenholt TR, Palarasah Y, Skjodt K, Jensen BL, & Skott O (2009) Prostasin- dependent activation of epithelial Na+ channels by low plasmin concentrations. Am. J. Physiol Regul. Integr. Comp Physiol, 297, R1733-R1741.

102

[158] Powers JC, Asgian JL, Ekici OD, & James KE (2002) Irreversible inhibitors of serine, cysteine, and threonine proteases. Chem. Rev., 102, 4639-4750.

[159] Schoellmann G & Shaw E (1963) Direct evidence for the presence of histidine in the active center of chymotrypsin. Biochemistry, 2, 252-255.

[160] Anderson PJ & Bock PE (2001) Biotin derivatives of D-Phe-Pro-Arg-CH2Cl for active-site- specific labeling of thrombin and other serine proteinases. Anal. Biochem., 296, 254-261.

[161] Ranby M, Bergsdorf N, & Nilsson T (1982) Enzymatic properties of the one- and two-chain form of tissue plasminogen activator. Thromb. Res., 27, 175-183.

[162] Boose JA, Kuismanen E, Gerard R, Sambrook J, & Gething MJ (1989) The single-chain form of tissue-type plasminogen activator has catalytic activity: studies with a mutant enzyme that lacks the cleavage site. Biochemistry, 28, 635-643.

[163] Tate KM, Higgins DL, Holmes WE, Winkler ME, Heyneker HL, & Vehar GA (1987) Functional role of proteolytic cleavage at arginine-275 of human tissue plasminogen activator as assessed by site-directed mutagenesis. Biochemistry, 26, 338-343.

[164] Tachias K & Madison EL (1995) Variants of tissue-type plasminogen activator which display substantially enhanced stimulation by fibrin. J. Biol. Chem., 270, 18319-18322.

[165] Madison EL, Kobe A, Gething MJ, Sambrook JF, & Goldsmith EJ (1993) Converting tissue plasminogen activator to a zymogen: a regulatory triad of Asp-His-Ser. Science, 262, 419-421.

[166] Darragh MR, Schneider EL, Lou J, Phojanakong PJ, Farady CJ, Marks JD, Hann BC, & Craik CS (2010) Tumor detection by imaging proteolytic activity. Cancer Res., 70, 1505-1512.

[167] Galkin AV, Mullen L, Fox WD, Brown J, Duncan D, Moreno O, Madison EL, & Agus DB (2004) CVS-3983, a selective matriptase inhibitor, suppresses the growth of androgen independent prostate tumor xenografts. Prostate, 61, 228-235.

[168] Farady CJ, Sun J, Darragh MR, Miller SM, & Craik CS (2007) The mechanism of inhibition of antibody-based inhibitors of membrane-type serine protease 1 (MT-SP1). J. Mol. Biol., 369, 1041-1051.

[169] Enyedy IJ, Lee SL, Kuo AH, Dickson RB, Lin CY, & Wang S (2001) Structure-based approach for the discovery of bis-benzamidines as novel inhibitors of matriptase. J. Med. Chem., 44, 1349- 1355.

[170] Long YQ, Lee SL, Lin CY, Enyedy IJ, Wang S, Li P, Dickson RB, & Roller PP (2001) Synthesis and evaluation of the sunflower derived trypsin inhibitor as a potent inhibitor of the type II transmembrane serine protease, matriptase. Bioorg. Med. Chem. Lett., 11, 2515-2519.

[171] Li P, Jiang S, Lee SL, Lin CY, Johnson MD, Dickson RB, Michejda CJ, & Roller PP (2007) Design and synthesis of novel and potent inhibitors of the type II transmembrane serine protease, matriptase, based upon the sunflower trypsin inhibitor-1. J. Med. Chem., 50, 5976-5983.

[172] Sun J, Pons J, & Craik CS (2003) Potent and selective inhibition of membrane-type serine protease 1 by human single-chain antibodies. Biochemistry, 42, 892-900.

103

[173] Napp J, Dullin C, Muller F, Uhland K, Petri JB, van de Locht A, Steinmetzer T, & Alves F (2010) Time-domain in vivo near infrared fluorescence imaging for evaluation of matriptase as a potential target for the development of novel, inhibitor-based tumor therapies. Int. J. Cancer, 127, 1958-1974.

[174] Steinmetzer T, Schweinitz A, Sturzebecher A, Donnecke D, Uhland K, Schuster O, Steinmetzer P, Muller F, Friedrich R, Than ME, Bode W, & Sturzebecher J (2006) Secondary amides of sulfonylated 3-amidinophenylalanine. New potent and selective inhibitors of matriptase. J. Med. Chem., 49, 4116-4126.

[175] Farady CJ, Egea PF, Schneider EL, Darragh MR, & Craik CS (2008) Structure of an Fab-protease complex reveals a highly specific non-canonical mechanism of inhibition. J. Mol. Biol., 380, 351-360.

104

Supplementary I

Hepatocyte growth factor activator inhibitor-1 has a complex subcellular itinerary

Sine Godiksen1, Joanna Selzer‐Plon1, Esben D. K. Pedersen1, Kathrine Abell1, Hanne Borger Rasmussen2, Roman Szabo3, Thomas H. Bugge3, and Lotte K. Vogel1

1Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark. 2Department of Biomedical Science, University of Copenhagen, Copenhagen, Denmark 3Proteases and Tissue Remodeling Unit, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, USA

Published in Biochemical Journal, July 2008.

105