Membrane-anchored serine protease matriptase regulates epithelial ...

1 downloads 0 Views 667KB Size Report
Mar 2, 2010 - Membrane-anchored serine protease matriptase regulates epithelial barrier formation and permeability in the intestine. Marguerite S. Buzzaa ...
Membrane-anchored serine protease matriptase regulates epithelial barrier formation and permeability in the intestine Marguerite S. Buzzaa, Sarah Netzel-Arnetta, Terez Shea-Donohueb, Aiping Zhaob, Chen-Yong Linc, Karin Listd, Roman Szaboe, Alessio Fasanob, Thomas H. Buggee, and Toni M. Antalisa,1 a Center for Vascular and Inflammatory Diseases and Department of Physiology, bMucosal Biology Research Center, and cDepartment of Biochemistry and Molecular Biology, University of Maryland School of Medicine, Baltimore, MD 21201; dDepartment of Pharmacology, Wayne State University and Karmanos Cancer Institute, Detroit, MI 48201; and eProteases and Tissue Remodeling Section, National Institute of Dental and Cranofacial Research, National Institutes of Health, Bethesda, MD 20892

Edited by Masatoshi Takeichi, RIKEN, Kobe, Japan, and approved January 14, 2010 (received for review April 11, 2009)

The intestinal epithelium serves as a major protective barrier between the mammalian host and the external environment. Here we show that the transmembrane serine protease matriptase plays a pivotol role in the formation and integrity of the intestinal epithelial barrier. St14 hypomorphic mice, which have a 100-fold reduction in intestinal matriptase mRNA levels, display a 35% reduction in intestinal transepithelial electrical resistance (TEER). Matriptase is expressed during intestinal epithelial differentiation and colocalizes with E-cadherin to apical junctional complexes (AJC) in differentiated polarized Caco-2 monolayers. Inhibition of matriptase activity using a specific peptide inhibitor or by knockdown of matriptase by siRNA disrupts the development of TEER in barrierforming Caco-2 monolayers and increases paracellular permeability to macromolecular FITC-dextran. Loss of matriptase was associated with enhanced expression and incorporation of the permeabilityassociated, “leaky” tight junction protein claudin-2 at intercellular junctions. Knockdown of claudin-2 enhanced the development of TEER in matriptase-silenced Caco-2 monolayers, suggesting that the reduced barrier integrity was caused, at least in part, by an inability to regulate claudin-2 expression and incorporation into junctions. We find that matriptase enhances the rate of claudin-2 protein turnover, and that this is mediated indirectly through an atypical PKCζdependent signaling pathway. These results support a key role for matriptase in regulating intestinal epithelial barrier competence, and suggest an intriguing link between pericellular serine protease activity and tight junction assembly in polarized epithelia.

| |

claudin-2 intestinal barrier protease tight junction

| St14 | type II transmembrane serine

T

he intestinal epithelium provides a critical protective barrier against enteric pathogens, food antigens, and physiochemical stresses caused by digestive and microbial products, and yet must be selectively permeable to beneficial nutrients and fluids. Tightly regulated control of barrier function and integrity is critical, as the pathogenesis of intestinal diseases such as Crohn's disease, ulcerative colitis, inflammatory bowel diseases (IBD), and autoimmune diseases are linked to intestinal barrier dysfunction and increased intestinal permeability (1). The intestinal epithelium is a single layer of linked columnar epithelial cells that regulates, through the paracellular pathway, the selective passage of ions, fluid, and macromolecules from the intestinal lumen into the underlying tissues. This paracellular pathway is controlled by intercellular apical junctional complexes (AJC) comprising apically located tight junctions and lateral adherens junctions and desmosomes (1). Intercellular AJCs are extremely dynamic structures that readily adapt to a variety of physiological and pathological stimuli. They are composed of transmembrane proteins, including occludin, claudins, and cadherins, which are linked intracellularly to cytoplasmic adaptor proteins, including the family of zonula occludins proteins (e.g., ZO-1) and catenins. Despite significant

4200–4205 | PNAS | March 2, 2010 | vol. 107 | no. 9

progress in our knowledge of the components of these structures, their functional regulation remains incompletely understood. Matriptase [membrane-type serine protease-1 (MT-SP1), TADG-15, epithin, SNC19] is an integral membrane trypsin-like serine protease that is a member of the type II transmembrane serine protease (TTSP) family (2, 3). Matriptase has a multidomain structure, consisting of a short cytosolic domain, a transmembrane domain, a stem region, and a C-terminal serine protease (catalytic) domain, which is linked to the rest of the molecule by a disulfide bond (4). Matriptase is widely expressed in virtually all epithelium and is specifically found in the epithelial cells lining the esophagus, stomach, jejunum, ileum, and colon of the GI tract (5). The physiological function of matriptase in the GI tract is not known. Studies of individuals with homozygosity for null and hypomorphic mutations in the St14 gene encoding matriptase, and studies of St14 null and hypomorphic mice have revealed a critical physiological role for matriptase in skin barrier formation and epidermal differentiation (6–8), however the role of matriptase in other epithelia is less well defined. Using the matriptase-deficient St14 hypomorphic mouse strain (6) to investigate matriptase function in intestinal epithelia, we have identified a critical role for matriptase in the formation and regulation of the integrity of the intestinal epithelial barrier. Loss of matriptase, resulting either from genetic depletion in St14 hypomorphic mice, via siRNA knockdown in the Caco-2 model of intestinal epithelium, or by chemical inhibition causes a “leaky” barrier, manifested by the impaired ability to develop transepithelial resistance (TEER) and enhanced paracellular permeability. This is mechanistically linked to the inappropriate expression of claudin-2, a tight junction protein associated with increased intestinal permeability and barrier disruption in IBD. Results Matriptase Hypomorphic Mutant Mice Have a Leaky Gut. St14

hypomorphic mice were found consistently to express less than 1% of the matriptase mRNA levels detected in intestinal tissues of littermate control mice (Fig. 1A). The effect of this marked matriptase deficiency on intestinal barrier function was investigated by measurement of transepithelial electrical resistance (TEER) of ex vivo intestinal tissues. TEER reflects paracellular resistance imparted by tight junctions and the lateral paracellular space, and is

Author contributions: M.S.B., S.N.-A., T.S.-D., A.F., and T.M.A. designed research; M.S.B., S.N.-A., T.S.-D., and A.Z. performed research; M.S.B., T.S.-D., A.Z., C.-Y.L., K.L., R.S., and T.H.B. contributed new reagents/analytic tools; M.S.B., S.N.-A., T.S.-D., A.Z., A.F., T.H.B., and T.M.A. analyzed data; and M.S.B., T.H.B., and T.M.A. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1

To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0903923107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0903923107

changed during this period (Fig. S2A), suggesting matriptase protein expression is regulated at a posttranslational level during Caco-2 differentiation. A similar increase in matriptase protein levels (Fig. S2B), in the absence of changes in mRNA (Fig. S2C) was also observed in colonic T84 cells.

a sensitive measure of barrier integrity. Measurement of the mean baseline TEER revealed a 35% reduction in TEER in intestinal tissue segments of St14 hypomorphic mice compared with littermate controls (Fig. 1B). Other measures of intestinal function, including smooth muscle contractibility and glucose absorption, were not found to be different between St14 hypomorphs and control littermates. In addition, microscopic evaluation of formalin-fixed tissues taken from the ileal and colonic regions of the hypomorph GI tract showed no significant structural abnormalities in villus architecture, the number of goblet cells, or mucosa and crypt features. Together these data suggested a specific intestinal epithelial barrier deficiency in St14 hypomorphic mice. Matriptase Increases During Differentiation and Formation of Polarized Caco-2 Monolayers. The molecular mechanisms underlying

matriptase activity in intestinal epithelium were investigated using polarized Caco-2 monolayers (9). When grown on transwell filters, Caco-2 cells spontaneously assemble tight junctions and display markers of intestinal epithelial cell differentiation, including the brush border membrane protease, dipeptidyl peptidase IV (DPPIV) (Fig. 2A). Cell-associated matriptase was detected as a 70-kDa band that represents the mature latent protein, and a faster-migrating 30-kDa band that represents the C-terminal serine protease domain released after reduction of the single disulfide bridge that links the two peptide chains following proteolytic activation of matriptase (4). Caco-2 cells constitutively expressed low levels of matriptase before reaching confluence (day 1; Fig. 2A), which increased dramatically as the cells became confluent (by day 4) and developed TEER, with a 5fold increase in expression detected by day 8 that remained elevated through day 21 (Fig. 2 A and B). These data demonstrate a correlation between matriptase expression and AJC assembly. Matriptase mRNA levels were not correspondingly

Fig. 2. Matriptase expression increases during Caco-2 epithelial cell polarization and barrier formation. (A) Immunoblot of cell lysates, prepared at the indicated times, showing increased levels of DPPIV and matriptase compared with GAPDH. Matriptase (∼70 kDa) and the matriptase serine protease domain (∼30 kDa) are indicated. NS indicates a ∼38-kDa band nonspecifically recognized by the IM1014 antibody (Fig. S1). (B) TEER measured during formation of Caco-2 epithelial barrier (solid line). Graph represents the mean ± SEM TEER from three separate experiments. Dotted lines show densitometric quantitation of changes in DPPIV and matriptase from A. Fold increase is calculated relative to day 1 after normalizing for GAPDH levels.

Buzza et al.

Inhibition of Matriptase Activity or siRNA Silencing of Matriptase Expression Impairs Caco-2 Epithelial Barrier Formation. Exposure of

Caco-2 monolayers to the broad-spectrum serine protease inhibitor AEBSF during Caco-2 differentiation substantially inhibited the development of TEER (Fig. 4A), suggesting a role for serine protease enzymatic activities during barrier formation. When Caco-2 monolayers were exposed to the synthetic inhibitor, CVS3983, a selective inhibitor of matriptase proteolytic activity (10), Caco-2 monolayers exhibited a similar inhibition of TEER, with values measuring ≈50% less on day 8 than in control cultures (Fig. 4A). Discontinuation of inhibitor treatment after day 8 resulted in restoration of TEER to the levels of media alone–treated cultures (day 12; Fig. 4A), showing that the viability of the cultures was unaffected. Together, these data implicate matriptase proteolytic activity in epithelial barrier formation and AJC assembly. Because chemical inhibitors can sometimes display off-target activities, we developed an alternative approach to specifically deplete matriptase using siRNA silencing. For these experiments, Caco-2 cells were transfected with siRNAs at a subconfluent density, and then added to transwell filters at higher density to accelerate the attainment of confluence and processes of epithelial differentiation. Under these conditions, and as reported by others (11), intercellular tight junctions are assembled within 4 days. Knock-down of matriptase mRNA and protein expression was

Fig. 3. Matriptase colocalizes with E-cadherin at apical junctional complexes in polarized Caco-2 monolayers. Caco-2 monolayers were grown on transwell filters for 21 days, immunostained with the indicated antibodies, and analyzed by confocal analysis. XY images show a single section through the monolayer; XZ images show a three-dimensional reconstruction of a cross-section of the monolayer. Position of plane of XY sections shown is indicated by arrow on right of XZ section. (A) Double labeling for matriptase using the antimatriptase antibody M32 (green) and rhodamine-conjugated phalloidin (red). (B) Double labeling for matriptase using M32 antibody (green) and anti–E-cadherin antibody (red).

PNAS | March 2, 2010 | vol. 107 | no. 9 | 4201

CELL BIOLOGY

Fig. 1. Matriptase hypomorph mice possess a leaky gut. (A) Matriptase mRNA levels in intestinal tissue segments of matriptase hypomorph mice (Hypo) compared with control littermates (Control) analyzed by quantitative PCR (Q-PCR). Shown is mean from three mice of each genotype. (B) Intestinal permeability measured by TEER in segments of small intestine from matriptase hypomorph (n = 6) and control littermates (n = 4). Mean ± SEM at 90 min are shown. *P < 0.05.

Matriptase Localizes to AJCs in Polarized Caco-2 Monolayers. Confocal microscopic examination of matriptase in the polarized Caco-2 monolayers revealed specific localization of matriptase to sites of intercellular contacts (XY sections, Fig. 3A). Matriptase was confined to intercellular apically located junctional complexes, whereas polymerized actin was present around the periphery of the cells (XZ sections, Fig. 3A). Negligible matriptase staining was detected at either the apical or basal cell surfaces of polarized Caco-2 monolayers (Fig. 3A, XZ merge panel). Matriptase specifically colocalized with the adherens junction marker E-cadherin at the intercellular contacts (Fig. 3B), and was detected below the apically associated tight junction proteins, ZO-1 and occludin (Fig. S3), linking matriptase to the adherens junctions.

Matriptase Depletion Increases Permeability to Macromolecular FITCDextran. TEER is influenced both by paracellular ion flux and

permeability to macromolecules (13). Measurement of the permeability of matriptase-silenced Caco-2 monolayers to the nonionic macromolecular tracer, FITC-dextran (4kD), which can only traverse the monolayer via the paracellular route, was assessed (Fig. 4D). By day 5, when the development of TEER was substantially delayed in siM1 cultures, monolayers displayed a 3-fold increase in permeability to FITC-dextran compared with control monolayers, linking matriptase to the paracellular pathway. Matriptase Contributes to AJC Reassembly After Barrier Disruption.

Fig. 4. Matriptase is required for development of TEER in Caco-2 epithelial monolayers. (A) Inhibition of matriptase serine protease activity inhibits development of TEER in Caco-2 monolayers. Confluent Caco-2 cultures (TEER ∼250 ohm·cm2) were treated with the broad-spectrum serine protease inhibitor AEBSF (25 μM) or the specific matriptase inhibitor CVS-3983 (25 μM) for 8 days, after which the inhibitors were removed. TEER development recovered to that of untreated cultures by day 12 after removal of CVS-3983 (right axis). (B) RNAi silencing of matriptase in Caco-2 epithelial monolayers. Immunoblot of matriptase protein levels analyzed 24 h after transfection with the two matriptase siRNAs (siM1, M2) or control siRNA (siCtl) on the indicated days after plating on transwell filters. Matriptase protein levels are suppressed by >90% by day 1, whereas levels of the related membrane proteases prostasin and DPPIV are unaffected by matriptase siRNAs. (C) RNAi silencing of matriptase inhibits de novo formation of Caco-2 epithelial barriers. Measurement of TEER development on the indicated days from the cultures depicted in B. (D) Matriptase-depleted monolayers display increased paracellular permeability to macromolecular 4 kDa FITC–dextran. Apparent permeabilities (Papp) were measured on day 2 (before barrier formation) and on day 5 postplating onto transwells. Graph represents mean ± SEM from triplicate wells, **P = 0.01. Corresponding average of TEER measured for each culture is shown in parentheses above each bar.

achieved with either of two independent siRNAs (siM1, siM2) by day 1 after addition to transwell filters, relative to the control siRNA (siCtl) (Fig. S4 and Fig. 4B). Matriptase protein levels remained effectively suppressed through day 7, after which time matriptase levels slowly began to increase, most notably with siM1 (Fig. 4B). The expression of other membrane-associated serine proteases, DPPIV or prostasin, was unaffected by the matriptase targeted siRNAs (Fig. 4B). Over the 7-day period, the matriptasesilenced Caco-2 monolayers (siM1, siM2) exhibited impaired development of TEER, resulting in a ∼70% lower TEER compared with siCtl monolayers at day 7 (Fig. 4C). Extended analysis over a 21-day period showed recovery of TEER as matriptase expression in siM1-silenced monolayers reappeared (Fig. S5). These data show that the depletion of matriptase did not irreversibly arrest barrier development, and indicate that matriptase is essential for differentiation processes required for normal intestinal epithelial barrier formation. Caco-2 Cell Proliferation in Polarized Monolayers Is Unaffected by Matriptase Depletion. Matriptase activity has been associated with

cell proliferation via activation of growth factors in some cellular contexts (12), raising the possibility that the impaired ability to develop TEER in matriptase-silenced Caco-2 monolayers could be due to reduced cell proliferation. Cell proliferation assays revealed that there were similar cell numbers in matriptasesilenced and control Caco-2 monolayers (Fig. S6), showing that decreased cell proliferation did not account for the inability of matriptase depleted monolayers to develop normal TEER. 4202 | www.pnas.org/cgi/doi/10.1073/pnas.0903923107

The contribution of matriptase to Caco-2 barrier formation and the paracellular pathway, independent of cell proliferation, suggested a functional role during AJC assembly. To test this possibility, assembled tight junction complexes formed in confluent mature Caco-2 monolayers were disrupted by low calcium media and then allowed to reassemble by replacing the media with calcium-containing growth media (calcium switch). In the presence of the matriptase inhibitor CVS-3983, barrier reassembly was significantly delayed (Fig. 5A). Depletion of matriptase by siRNA in Caco-2 monolayers similarly delayed TEER development; within 7 h postcalcium switch, the siM1-treated monolayers developed only ∼50% of the TEER of the siCtl-treated cultures (Fig. 5B). The delayed development of TEER was associated with a corresponding increase in permeability to FITC-dextran (4kDa) of ∼9fold relative to control monolayers (Fig. 5C). Together these data demonstrate the importance of matriptase to AJC assembly and the paracellular pathway. Known Matriptase Substrates Do Not Account for Matripase Regulation of Epithelial Barrier Integrity. The natural substrates of

matriptase that may be proteolytically processed in the lateral paracellular space of polarized intestinal epithelia are not known.

Fig. 5. Matriptase contributes to AJC reassembly after barrier disruption. (A) Inhibition of matriptase activity delays the development of TEER in Caco2 monolayers disrupted by low calcium. Calcium switch assay performed on Caco-2 cultures (TEER > 1,000 ohm·cm2). Cells were cultured in the presence or absence of the matriptase inhibitor CVS-3983 (25 μM) during barrier reformation after calcium switch. (B) RNAi silencing of matriptase inhibits Caco-2 barrier reassembly after disruption. Transfected Caco-2 cells (siCtl or siM1) were grown on transwells for 5 days before calcium switch. (Upper) Immunoblot analysis of matriptase protein levels analyzed before calcium depletion (−15 h), after incubation in low-calcium medium (0 h), and after calcium restoration at 8 and 24 h. (Lower) Development of TEER at the indicated times after calcium restoration. (C) Permeabilities to macromolecular FITC–dextran were assessed at 0 and 7 h postcalcium switch. Graph shows mean ± SEM from triplicate wells, and data are plotted using a log scale. *P < 0.05. Corresponding average of TEER measured for each culture is shown in parentheses above each bar.

Buzza et al.

Known matriptase substrates present in intestinal epithelia, e.g., the zymogens of prostasin and urokinase plasminogen activator (uPA), or the epidermal growth factor (EGF) receptor, proteaseactivated receptor–2 (PAR-2), or HGF signaling pathways, were not found to be direct targets of matriptase in polarized intestinal epithelial cells (Fig. S7), indicating an alternative molecular activity for matriptase during epithelial barrier formation and AJC assembly. Permeability-Associated Tight Junction Protein Claudin-2 Is Selectively Deregulated in Matriptase-Silenced Caco-2 Monolayers. Examination

of the expression of AJC proteins involved in barrier assembly in formed matriptase-silenced Caco-2 monolayers compared with control cultures revealed no significant changes in the levels of the tight junction proteins occludin and claudin-1, claudin-3, claudin4, and claudin-8, the adapter protein ZO-1, or the adherens junction proteins, E-cadherin, and β-catenin (Fig. 6A and Fig. S9A). The localization of several of these AJC components at intercellular contacts similarly appeared to be unaffected by matriptase depletion either during barrier assembly (Fig. S8A) or following calcium switch (Fig. S8B). In contrast, there was a substantial 3.5- to 4-fold increase in the levels of the permeabilityassociated tight junction protein claudin-2 in the matriptasesilenced Caco-2 monolayers (Fig. 6A). The claudins are a family of tight junctional proteins that form ion specific channels, the majority of which, including claudin-1, function to enhance the “tightness” of epithelial barriers. Some of the claudins however, such as claudin-2, form ion specific channels that result in the “loosening” of epithelial barriers and reduction of epithelial resistance (13). Investigation of the time course of claudin-2 expression during Caco-2 barrier formation revealed that claudin2 protein and mRNA expression were highest in Caco-2 cells before formation of the epithelial barrier, and then were substantially down-regulated as Caco-2 monolayers differentiate and

TEER develops (Fig. S2 and Fig. 6B, siCtl, claudin-2: day 3 vs. day 6). In contrast, claudin-1 protein levels increase over this same period (Fig. 6B, siCtl, claudin-1: day 3 vs. day 6) as expected. Silencing of matriptase expression prevented the down-regulation of claudin-2, with matriptase-depleted monolayers displaying the elevated claudin-2 levels associated with the less differentiated cells (Fig. 6B, siM1 and siM2, claudin-2: day 3 vs. day 6), whereas claudin-1 levels were unaffected by matriptase depletion (Fig. 6B, siM1 and siM2, claudin-1: day 3 vs. day 6). Interestingly, loss of claudin-2 is observed predominantly in the insoluble fraction of extracted lysates (Fig. 6B; day 6 siCtl, claudin-2: IS vs. S), suggesting that matriptase facilitates claudin-2 loss from formed tight junctional complexes. Indeed, confocal microscopic examination of claudin-2 expression revealed elevated claudin-2 at the intracellular junctions of matriptase-silenced Caco-2 monolayers compared with control monolayers (Fig. S8C). Matriptase Hypomorphic Leaky Gut Is Associated with Enhanced Claudin-2 Expression in Surface Villous Epithelial Cells. In the small

intestine, claudin-2 expression is normally restricted to the relatively leaky epithelial cells of the crypts, with its expression decreasing substantially in differentiating epithelium along the crypt–villus axis (14). Examination of claudin-2 expression by immunostaining of jejunal tissues from St14 hypomorphic mice compared with control mice showed normal claudin-2 staining in the crypt epithelium of both genotypes (Fig. 6C, arrows). Of significance, the surface villous epithelial cells of St14 hypomorph small intestines alone displayed substantial claudin-2 expression that was not evident in the control animals (Fig. 6C, arrowheads). Claudin-2 immunostaining was localized to intercellular regions of the surface epithelium (Fig. 6C, Right). These in vivo findings are consistent with the in vitro findings and support a role for matriptase in regulating intestinal epithelial barrier integrity through the regulation of claudin-2 protein and its incorporation into intercellular junctional complexes.

Fig. 6. Claudin-2 is deregulated in matriptase-depleted intestinal epithelial cells. (A) Immunoblot analysis of junctional proteins present in total cell lysates on day 6 posttransfection. Shown are lysates from two independent transfections (A and B). (B) Immunoblot analysis of claudin-2, claudin-1, and matriptase protein levels in lysates collected on days 3 and 6 posttransfection. Triton-soluble fraction (S), and the Triton-insoluble (IS), junction-associated fraction are shown. Average TEERs, day 3: 518 (siCtl), 197 (siM1), and 205 (siM2) ohm·cm2; day 6: 1,275 (siCtl), 517 (siM1), and 452 (siM2) ohm·cm2. (C) Detection of claudin-2 on surface of villous epithelium of St14 hypomorphic animals. Frozen sections of jejunum from control and hypomorph mice immunostained with antimouse claudin-2 antibody (green), and counterstained with DAPI for detection of nuclei (blue). Arrows indicate intestinal crypt cells, arrowheads indicate villous epithelium. At high power, claudin-2 staining can be seen to localize to intercellular junctions in the hypomorph intestine (Right). Micrographs are representative of four animals of each genotype. Original magnifications are 200× (Left and Center) and 1,000× (Right).

Buzza et al.

determinant of transepithelial resistance and functions to decrease the “tightness” of the epithelial intercellular tight junctions (13), we reasoned that the persistent expression of claudin-2 could contribute to the impaired ability of matriptase-depleted Caco-2 monolayers to develop an effective TEER. Indeed, knockdown of claudin-2 by cotransfection of claudin-2 siRNA in matriptase-depleted monolayers enhanced TEER development to normal levels (Fig. 7A). Interestingly, knockdown of claudin-2 alone in the Caco-2 monolayers increased TEER development, as has been observed previously (15), suggesting that the low level expression of claudin-2 in control Caco-2 monolayers (Fig. 7A, day 5 siCtl, claudin-2) prevents complete tightening of the Caco-2 epithelial barrier. These findings support a relationship between claudin-2 levels and the tightness of AJC complex formation, and further implicate the permeability-associated activities of claudin2 in the paracellular pathway regulated by matriptase. Matriptase Does Not Cleave Claudin-2 Directly but Facilitates Claudin2 Protein Turnover. Persistent claudin-2 expression was not asso-

ciated with changes in claudin-2 mRNA levels in matriptase-silenced Caco-2 monolayers during barrier formation (Fig. S9B) or in intestinal epithelial scrapings from St14 hypomorph mice compared with control littermates (Fig. S9C). However, pulse chase analyses comparing claudin-2 protein stability during barrier formation in matriptase-silenced Caco-2 monolayers revealed that matriptase depletion stabilized claudin-2 levels by greater than 3-fold (Fig. 7B), indicating that matriptase facilitates the posttranscriptional turnover of claudin-2 protein during formation of epithelial barriers. As claudin-2 is not proteolytically processed directly by matriptase (Fig. S10), and as we found no evidence for claudin-2 cleavage peptides in PNAS | March 2, 2010 | vol. 107 | no. 9 | 4203

CELL BIOLOGY

Knockdown of Claudin-2 Enhances TEER Development in MatriptaseSilenced Caco-2 Monolayers. Because claudin-2 expression is a

Fig. 7. Enhanced turnover of claudin-2 is functionally associated with matriptase activity. (A) Knockdown of claudin-2 enhances the development of TEER in Caco-2 monolayers. (Left) TEER development in Caco-2 monolayers after silencing of matriptase (siM1), claudin-2 (siCL2), or both together (siCL2+ iM1). (Right) Immunoblot analysis of the levels of the indicated proteins on day 3 and day 5. Results are representative of three independent experiments using two independent claudin-2 siRNAs. (B) Matriptase mediates enhanced turnover of claudin-2 protein. siCtl- and siM2-transfected cells were cultured on transwells for 7 days before metabolic labeling and pulse chase. Representative autoradiogram is shown (Upper); plot (Lower) shows average rates of claudin-2 and claudin-1 decay from two independent experiments. Time required to attain 50% of starting claudin-2 levels is greater than 3-fold slower in matriptase-silenced cells.

Caco-2 lysates under a variety of conditions, these findings implicated the indirect activation of a signaling pathway by matriptase that affects claudin-2 protein turnover. Matriptase Regulates Claudin-2 Turnover via Activation of Atypical PKCζ Signaling Pathway. The constitutive turnover of claudin-2 is

thought be mediated by an E3 ubiquitin ligase that targets claudins for subsequent lysosomal degradation (16). However, the signaling pathways that regulate claudin-2 turnover have not been established. Atypical protein kinase C (PKC) isoforms are involved in several signal transduction pathways and are essential for the formation of epithelial tight junctions and establishment of cell polarity (17). We found that matriptase silencing during barrier formation results in decreased activation of PKCζ (Fig. 8A), detected by phosphorylation at Thr-410 which is diagnostic for PKCζ activation (18), indicating that matriptase-mediated barrier formation involves activation of a PKCζ signaling pathway. To investigate whether activation of PKCζ regulates the rate of claudin-2 turnover, PKCζ activity was functionally inhibited using a cell-permeable PKCζ pseudosubstrate inhibitor (PKCζ-psi). TEER development was inhibited in a dose-dependent manner by PKCζ-psi (Fig. 8B and Fig. S11), which was accompanied by the substantial accumulation of claudin-2 protein (Fig. 8C), in the absence of changes in claudin-2 mRNA levels (Fig. S11B). These data support a mechanism by which matriptase regulates claudin-2 turnover and intestinal epithelial barrier integrity through a PKCζdependent signaling pathway. Discussion The studies presented here provide molecular insight into matriptase function in intestinal epithelia and reveal a fundamental role for matriptase during epithelial barrier formation and AJC assembly. Matriptase depletion, either by genetic down-regulation in St14 hypomorphic mice or by RNAi silencing in epithelial monolayers, alters the permeability of tight junctions, resulting in decreased epithelial resistance and increased permeability to macromolecules. The molecular basis of the leaky barrier involves the stabilization and aberrant incorporation of the permeabilityassociated protein claudin-2 into intercellular junctions mediated indirectly through an atypical PKC cell-signaling pathway. 4204 | www.pnas.org/cgi/doi/10.1073/pnas.0903923107

Fig. 8. Matriptase activates the PKCζ pathway during epithelial barrier formation. (A) Decreased activation of PKCζ in matriptase-depleted Caco-2 monolayers. Cell lysates prepared on day 6 were immunoblotted for phosphorylated PKCζ (P-threonine-410), followed by total PKCζ. (B) Inhibition of PKCζ pathway delays TEER development. Caco-2 cells plated on transwells were treated with 50 μM PKCζ-pseudosubstrate inhibitor (PKCζ-psi) for 48 h. Graph shows delay in TEER development in the presence of PKCζ-psi (mean ± SE from triplicate wells). (C) Inhibition of the PKCζ pathway is associated with sustained expression of claudin-2, whereas claudin-1 is unaffected. Immunoblot analysis was performed after 48 h treatment with 50 μM PKCζ-psi.

Matriptase expression increases during Caco-2 barrier formation and differentiation, consistent with the higher levels of matriptase associated with differentiated epithelia at the intestinal villous tips (19). In polarized Caco-2 monolayers, matriptase localizes to intercellular junctional complexes in association with E-cadherin along the lateral membrane. Activated PKCζ expression and localization at intercellular junctions is similarly detected in differentiated intestinal epithelia (18, 20). In contrast, claudin-2 expression decreases with Caco-2 differentiation (21); this is paralleled in the intestine, where expression decreases substantially in villous epithelium (14). Members of the claudin family that are present at epithelial tight junctions regulate TEER by a mechanism that does not appear to be related to structural changes in the tight junctions but, rather, through the formation of pores that selectively control the passage of cationic or anionic solutes (13). Claudin-2 decreases the tightness of epithelial barriers through the formation of cationselective ion channels with particular affinity for sodium ions at the tight junctions, but does not contribute to the paracellular permeability of small noncharged molecules (13, 15). In addition to decreased TEER, matriptase loss from epithelial monolayers is associated with increased permeability to macromolecules, suggesting that matriptase may regulate additional pathways associated with tight junction assembly and barrier formation. Increased claudin-2 expression and loss of barrier integrity in other cell systems is typically in response to cytokines (22) and is characterized by increased claudin-2 mRNA. The pathways that regulate claudin-2 protein turnover and target it for degradation via lysosomes (16) remain to be established; however, our data implicate a matriptase-mediated proteolytic pathway that facilitates the turnover of claudin-2 during AJC assembly that is associated with PKCζ activation. Interestingly, PKCζ signaling has been implicated recently in trypsin and chymotrypsin-enhanced barrier formation (23). Little is known regarding the regulation of claudin function by atypical PKCs, although numerous studies demonstrate their involvement in the association of intercellular signaling complexes that trigger AJC assembly. Atypical PKC activation induces increased phosphorylation of the barrier-forming claudin-4 and its localization at cell junctions (24). The present study links PKCζ activity with increased turnover of the leaky claudin-2 during the assembly of tight junctions, further implicating atypical PKCs in the regulation of this family of tight junction proteins. Serine proteases have long been recognized to contribute to protein degradation associated with nutrient digestion in the GI tract, and increasing evidence suggests that these enzymes also Buzza et al.

Methods Reagent details and additional methods are provided in SI Methods. Animals. The St14 (matriptase) hypomorphic mice have been described (6). All experiments were littermate and sibling controlled. Institutional Animal Care and Use Committee approval was obtained for all experimental procedures. For ex vivo analyses, the small intestines were flushed, and the was muscle removed and then cut into 1-cm sections. Directly adjacent tissue segments were used for measurements of ex vivo TEER and isolation of RNA. Enriched preparations of epithelial cells were obtained by the coverslip scraping method.

monolayers was measured using “chopstick: probes. Baseline resistance readings were determined in wells containing membrane inserts only, subtracted from sample values and expressed in ohm·cm2. Flux of 4-kDa FITC-conjugated dextran (Fluka) across Caco-2 monolayers was assayed as described in SI Text. Caco-2 Cell Culture and Transfection. Caco-2 cells (ATCC, passage 35–45), cultured under standard cell culture conditions (27), were grown on Transwell filters (Costar) for up to 21 days for formation of polarized epithelial monolayers. For calcium switch experiments, 5-day filter-grown Caco-2 monolayers were exposed to low calcium medium (DMEM + 5 μM CaCl2) for 15 h, after which medium was replaced with complete DMEM (1.6 mM CaCl2) for up to 24 h. When used, the chemical inhibitors AEBSF (Calbiochem), CVS-3983 (10) or PKCζ myristolated-psi (Calbiochem), were added to culture media and replaced daily. Transfections were performed using Dharmafect 1 transfection reagent (Dharmacon) using primers listed in the SI Methods. Cells were plated onto filters 24 h posttransfection at high density (5 × 105 cells/well). Results presented are representative of at least three to five independent transfection experiments. Immunofluorescence Microscopy. Caco-2 monolayers grown on transwell filters for 21 days were methanol fixed at room temperature. Rhodamineconjugated phalloidin was used for detection of actin. XY sections (0.5-μm steps) were captured using a Bio-Rad Radiance 2100 confocal microscope and reconstructed using the Velocity Image Analysis program for XZ images. Intestinal tissues removed from mice were cryopreserved, fixed in methanol, and stained for claudin-2 and DAPI for detection of nuclei. Fluorescence was visualized using a Nikon Eclipse E800 microscope and captured with Axiovision (Zeiss) image capture software. Statistical Analyses. The two-tailed Student’s t test was used to compare averages of normally distributed data with equal variance. A threshold of P < 0.05 was considered significant.

Measurements of Barrier Integrity. TEER of small intestine segments mounted in microsnapwells apical side up was measured in triplicate using an EVOM Voltohmmeter (World Precision Instruments) as described (27). TEER of Caco-2

ACKNOWLEDGMENTS. We thank E. Smith, J. Stiltz, and R. Sun for technical assistance. CVS-3983 was provided by Dr. Ed Madison and Corvas International (San Diego, CA). This work was supported in part by grants from the National Institutes of Health (NIH): DK081376, CA098369, and HL084387 (to T.M.A.); DK48373 (to A.F.); AI/DK49316 (to T.S.D.); HL07698 (to S.N.A.); CA096851 (to C.Y.L.) and the NIH Intramural Program (T.H.B.). M.B. was supported by Australian National Health and Medical Research Council CJ Martin Fellowship 384359.

1. Laukoetter MG, Nava P, Nusrat A (2008) Role of the intestinal barrier in inflammatory bowel disease. World J Gastroenterol 14:401–407. 2. Netzel-Arnett S, et al. (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. 3. 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. 4. Lin CY, et al. (2008) Zymogen activation, inhibition, and ectodomain shedding of matriptase. Front Biosci 13:621–635. 5. Bugge TH, List K, Szabo R (2007) Matriptase-dependent cell surface proteolysis in epithelial development and pathogenesis. Front Biosci 12:5060–5070. 6. List K, et al. (2007) Autosomal ichthyosis with hypotrichosis syndrome displays low matriptase proteolytic activity and is phenocopied in ST14 hypomorphic mice. J Biol Chem 282:36714–36723. 7. List K, et al. (2002) Matriptase/MT-SP1 is required for postnatal survival, epidermal barrier function, hair follicle development, and thymic homeostasis. Oncogene 21: 3765–3779. 8. Alef T, et al. (2009) Ichthyosis, follicular atrophoderma, and hypotrichosis caused by mutations in ST14 is associated with impaired profilaggrin processing. J Invest Dermatol 129:862–869. 9. Sambuy Y, et al. (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. 10. Galkin AV, et al. (2004) CVS-3983, a selective matriptase inhibitor, suppresses the growth of androgen independent prostate tumor xenografts. Prostate 61:228–235. 11. Clayburgh DR, et al. (2004) A differentiation-dependent splice variant of myosin light chain kinase, MLCK1, regulates epithelial tight junction permeability. J Biol Chem 279: 55506–55513. 12. Förbs D, et al. (2005) In vitro inhibition of matriptase prevents invasive growth of cell lines of prostate and colon carcinoma. Int J Oncol 27:1061–1070. 13. Van Itallie CM, Anderson JM (2006) Claudins and epithelial paracellular transport. Annu Rev Physiol 68:403–429.

14. Rahner C, Mitic LL, Anderson JM (2001) Heterogeneity in expression and subcellular localization of claudins 2, 3, 4, and 5 in the rat liver, pancreas, and gut. Gastroenterology 120:411–422. 15. Hou J, Gomes AS, Paul DL, Goodenough DA (2006) Study of claudin function by RNA interference. J Biol Chem 281:36117–36123. 16. Takahashi S, et al. (2009) The E3 ubiquitin ligase LNX1p80 promotes the removal of claudins from tight junctions in MDCK cells. J Cell Sci 122:985–994. 17. Suzuki A, et al. (2001) Atypical protein kinase C is involved in the evolutionarily conserved par protein complex and plays a critical role in establishing epitheliaspecific junctional structures. J Cell Biol 152:1183–1196. 18. Hirai T, Chida K (2003) Protein kinase Czeta (PKCzeta): Activation mechanisms and cellular functions. J Biochem 133:1–7. 19. Oberst MD, et al. (2003) Characterization of matriptase expression in normal human tissues. J Histochem Cytochem 51:1017–1025. 20. Verstovsek G, Byrd A, Frey MR, Petrelli NJ, Black JD (1998) Colonocyte differentiation is associated with increased expression and altered distribution of protein kinase C isozymes. Gastroenterology 115:75–85. 21. Escaffit F, Boudreau F, Beaulieu JF (2005) Differential expression of claudin-2 along the human intestine: Implication of GATA-4 in the maintenance of claudin-2 in differentiating cells. J Cell Physiol 203:15–26. 22. Capaldo CT, Nusrat A (2009) Cytokine regulation of tight junctions. Biochim Biophys Acta 1788:864–871. 23. Swystun VA, et al. (2009) Serine proteases decrease intestinal epithelial ion permeability by activation of protein kinase Czeta. Am J Physiol Gastrointest Liver Physiol 297:G60–G70. 24. Aono S, Hirai Y (2008) Phosphorylation of claudin-4 is required for tight junction formation in a human keratinocyte cell line. Exp Cell Res 314:3326–3339. 25. Bacher A, et al. (1992) Protease inhibitors suppress the formation of tight junctions in gastrointestinal cell lines. Exp Cell Res 200:97–104. 26. Mankertz J, Schulzke JD (2007) Altered permeability in inflammatory bowel disease: Pathophysiology and clinical implications. Curr Opin Gastroenterol 23:379–383. 27. El Asmar R, et al. (2002) Host-dependent zonulin secretion causes the impairment of the small intestine barrier function after bacterial exposure. Gastroenterology 123:1607–1615. 28. Angelow S, Schneeberger EE, Yu AS (2007) Claudin-8 expression in renal epithelial cells augments the paracellular barrier by replacing endogenous claudin-2. J Membr Biol 215:147–159.

Buzza et al.

PNAS | March 2, 2010 | vol. 107 | no. 9 | 4205

CELL BIOLOGY

contribute in a complex way to the regulation of intestinal integrity and barrier function. Low levels of trypsin and certain other endopeptidases are potent inducers of intercellular AJC formation in intestinal epithelial monolayers (23, 25), whereas higher levels of these enzymes disrupt epithelial barriers. Epithelial barrier formation may also be suppressed by serine protease inhibitors (25), further implicating serine protease activities in AJC assembly. These findings, together with the barrier-protective activity of matriptase, implicate pericellular proteolytic processing as an important component of epithelial barrier formation. In summary, we have identified a proteolytic mechanism in the lateral paracellular space of polarized epithelia important for the regulation of intestinal epithelial barrier integrity. Compromise of the epithelial barrier is an early event in disorders such as Crohn’s disease and IBD, and the resulting functional change in barrier permeability exacerbates inflammation. Claudin-2 is frequently up-regulated in active Crohn’s disease and in the inflamed mucosa of patients with ulcerative colitis, where it has been proposed to contribute to increased intestinal permeability and pathogenesis (26). It will be important in future studies to identify the unique substrates of matriptase involved in this pathway, and to investigate matriptase dysregulation in the pathogenesis of inflammatory diseases of the GI tract.