Selective Inhibition of Prostasin in Human Enterocytes by the Integral Membrane Kunitz-Type Serine Protease Inhibitor HAI-2.

Mutations of hepatocyte growth factor activator inhibitor (HAI)-2 in humans cause sodium loss in the gastrointestinal (GI) tract in patients with syndromic congenital sodium diarrhea (SCSD). Aberrant regulation of HAI-2 target protease(s) was proposed as the cause of the disease. Here functional linkage of HAI-2 with two membrane-associated serine proteases, matriptase and prostasin was analyzed in Caco-2 cells and the human GI tract. Immunodepletion-immunoblot analysis showed that significant proportion of HAI-2 is in complex with activated prostasin but not matriptase. Unexpectedly, prostasin is expressed predominantly in activated forms and was also detected in complex with HAI-1, a Kunitz inhibitor highly related to HAI-2. Immunohistochemistry showed a similar tissue distribution of prostasin and HAI-2 immunoreactivity with the most intense labeling near the brush borders of villus epithelial cells. In contrast, matriptase was detected primarily at the lateral plasma membrane, where HAI-1 was also detected. The tissue distribution profiles of immunoreactivity against these proteins, when paired with the species detected suggests that prostasin is under tight control by both HAI-1 and HAI-2 and matriptase by HAI-1 in human enterocytes. Furthermore, HAI-1 is a general inhibitor of prostasin in a variety of epithelial cells. In contrast, HAI-2 was not found to be a significant inhibitor for prostasin in mammary epithelial cells or keratinocytes. The high levels of constitutive prostasin zymogen activation and the selective prostasin inhibition by HAI-2 in enterocytes suggest that dysregulated prostasin proteolysis may be particularly important in the GI tract when HAI-2 function is lost and/or dysregulated.

Plasminogen-Dependent Matriptase Activation Accelerates Plasmin Generation by Differentiating Primary Human Keratinocytes.

Pericellular plasmin generation, an important pathophysiological process, can be initiated and accelerated by the autoactivation of the type 2 transmembrane serine protease matriptase and subsequent activation of urokinase plasminogen activator. The link between matriptase and plasminogen was initially thought to be one-directional: from matriptase, through plasminogen activator, to plasminogen. However, in the current study, we now show that primary human keratinocytes that are undergoing calcium-induced differentiation can rapidly activate matriptase in response to serum treatment via a mechanism dependent on intracellular calcium, protein kinase C, and phosphatidylinositol 3-kinases-based signaling. The serum factor, responsible for the induction of matriptase zymogen activation, was shown to be plasminogen. A sub-pM concentration of plasminogen (but not plasmin) acting at the cell surface is sufficient to induce matriptase activation, suggesting high potency and specificity of the induction. After matriptase zymogen activation, a proportion of active matriptase is shed into extracellular milieu and returns to the cell surface to accelerate plasmin generation. The ability of plasminogen to induce matriptase zymogen activation and the subsequent acceleration of plasmin generation by active matriptase reveals a feed-forward mechanism that allows differentiating human keratinocytes to rapidly and robustly activate pericellular proteolysis.

Matriptase Complexes and Prostasin Complexes with HAI-1 and HAI-2 in Human Milk: Significant Proteolysis in Lactation.

Significant proteolysis may occur during milk synthesis and secretion, as evidenced by the presence of protease-protease inhibitor complex containing the activated form of the type 2 transmembrane serine protease matriptase and the transmembrane Kunitz-type serine protease inhibitor HAI-1. In order to identify other proteolysis events that may occur during lactation, human milk was analyzed for species containing HAI-1 and HAI-2 which is closely related to HAI-1. In addition to the previously demonstrated matriptase-HAI-1 complex, HAI-1 was also detected in complex with prostasin, a glycosylphosphatidylinositol (GPI)-anchored serine protease. HAI-2 was also detected in complexes, the majority of which appear to be part of higher-order complexes, which do not bind to ionic exchange columns or immunoaffinity columns, suggesting that HAI-2 and its target proteases may be incorporated into special protein structures during lactation. The small proportion HAI-2 species that could be purified contain matriptase or prostasin. Human mammary epithelial cells are the likely cellular sources for these HAI-1 and HAI-2 complexes with matriptase and prostasin given that these protease-inhibitor complexes with the exception of prostasin-HAI-2 complex were detected in milk-derived mammary epithelial cells. The presence of these protease-inhibitor complexes in human milk provides in vivo evidence that the proteolytic activity of matriptase and prostasin are significantly elevated at least during lactation, and possibly contribute to the process of lactation, and that they are under tight control by HAI-1 and HAI-2.

Natural Endogenous Human Matriptase and Prostasin Undergo Zymogen Activation via Independent Mechanisms in an Uncoupled Manner.

The membrane-associated serine proteases matriptase and prostasin are believed to function in close partnership. Their zymogen activation has been reported to be tightly coupled, either as a matriptase-initiated proteolytic cascade or through a mutually dependent mechanism involving the formation of a reciprocal zymogen activation complex. Here we show that this putative relationship may not apply in the context of human matriptase and prostasin. First, the tightly coupled proteolytic cascade between matriptase and prostasin might not occur when modest matriptase activation is induced by sphingosine 1-phospahte in human mammary epithelial cells. Second, prostasin is not required and/or involved in matriptase autoactivation because matriptase can undergo zymogen activation in cells that do not endogenously express prostasin. Third, matriptase is not required for and/or involved in prostasin activation, since activated prostasin can be detected in cells expressing no endogenous matriptase. Finally, matriptase and prostasin both undergo zymogen activation through an apparently un-coupled mechanism in cells endogenously expressing both proteases, such as in Caco-2 cells. In these human enterocytes, matriptase is detected primarily in the zymogen form and prostasin predominantly as the activated form, either in complexes with protease inhibitors or as the free active form. The negligible levels of prostasin zymogen with high levels of matriptase zymogen suggests that the reciprocal zymogen activation complex is likely not the mechanism for matriptase zymogen activation. Furthermore, high level prostasin activation still occurs in Caco-2 variants with reduced or absent matriptase expression, indicating that matriptase is not required and/or involved in prostasin zymogen activation. Collectively, these data suggest that any functional relationship between natural endogenous human matriptase and prostasin does not occur at the level of zymogen activation.

N-Glycan Branching Affects the Subcellular Distribution of and Inhibition of Matriptase by HAI-2/Placental Bikunin.

The gene product of SPINT 2, that encodes a transmembrane, Kunitz-type serine protease inhibitor independently designated as HAI-2 or placenta bikunin (PB), is involved in regulation of sodium absorption in human gastrointestinal track. Here, we show that SPINT 2 is expressed as two species of different size (30-40- versus 25-kDa) due to different N-glycans on Asn-57. The N-glycan on 25-kDa HAI-2 appears to be of the oligomannose type and that on 30-40-kDa HAI-2 to be of complex type with extensive terminal N-acetylglucosamine branching. The two different types of N-glycan differentially mask two epitopes on HAI-2 polypeptide, recognized by two different HAI-2 mAbs. The 30-40-kDa form may be mature HAI-2, and is primarily localized in vesicles/granules. The 25-kDa form is likely immature HAI-2, that remains in the endoplasmic reticulum (ER) in the perinuclear regions of mammary epithelial cells. The two different N-glycans could, therefore, represent different maturation stages of N-glycosylation with the 25-kDa likely a precursor of the 30-40-kDa HAI-2, with the ratio of their levels roughly similar among a variety of cells. In breast cancer cells, a significant amount of the 30-40-kDa HAI-2 can translocate to and inhibit matriptase on the cell surface, followed by shedding of the matriptase-HAI-2 complex. The 25-kDa HAI-2 appears to have also exited the ER/Golgi, being localized at the cytoplasmic face of the plasma membrane of breast cancer cells. While the 25-kDa HAI-2 was also detected at the extracellular face of plasma membrane at very low levels it appears to have no role in matriptase inhibition probably due to its paucity on the cell surface. Our study reveals that N-glycan branching regulates HAI-2 through different subcellular distribution and subsequently access to different target proteases.

Differential subcellular localization renders HAI-2 a matriptase inhibitor in breast cancer cells but not in mammary epithelial cells.

The type 2 transmembrane serine protease matriptase is under tight control primarily by the actions of the integral membrane Kunitz-type serine protease inhibitor HAI-1. Growing evidence indicates that HAI-2 might also be involved in matriptase inhibition in some contexts. Here we showed that matriptase inhibition by HAI-2 depends on the subcellular localizations of HAI-2, and is observed in breast cancer cells but not in mammary epithelial cells. HAI-2 is co-expressed with matriptase in 21 out of 26 human epithelial and carcinoma cells examined. HAI-2 is also a potent matriptase inhibitor in solution, but in spite of this, HAI-2 inhibition of matriptase is not observed in all contexts where HAI-2 is expressed, unlike what is seen for HAI-1. Induction of matriptase zymogen activation in mammary epithelial cells results in the formation of matriptase-HAI-1 complexes, but matriptase-HAI-2 complexes are not observed. In breast cancer cells, however, in addition to the appearance of matriptase-HAI-1 complex, three different matriptase-HAI-2 complexes, are formed following the induction of matriptase activation. Immunofluorescent staining reveals that activated matriptase is focused at the cell-cell junctions upon the induction of matriptase zymogen activation in both mammary epithelial cells and breast cancer cells. HAI-2, in contrast, remains localized in vesicle/granule-like structures during matriptase zymogen activation in human mammary epithelial cells. In breast cancer cells, however, a proportion of the HAI-2 reaches the cell surface where it can gain access to and inhibit active matriptase. Collectively, these data suggest that matriptase inhibition by HAI-2 requires the translocation of HAI-2 to the cell surface, a process which is observed in some breast cancer cells but not in mammary epithelial cells.