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.

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.

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.

No abstract