Matriptase is inhibited by extravascular antithrombin in epithelial cells but not in most carcinoma cells

Antithrombin, a major anticoagulant, is robustly transported into extravascular compartments where its target proteases are largely unknown. This serpin was previously detected in human milk as complexes with matriptase, a membrane-bound serine protease broadly expressed in epithelial and carcinoma cells, and under tight regulation by hepatocyte growth factor activator inhibitor (HAI)-1, a transmembrane Kunitz-type serine protease inhibitor that forms heat-sensitive complexes with active matriptase. In the current study, we detect, in addition to matriptase-HAI-1 complexes, heat-resistant matriptase complexes generated by nontransformed mammary, prostate, and epidermal epithelial cells that we show to be matriptase-antithrombin complexes. These findings suggest that in addition to HAI-1, interstitial antithrombin participates in the regulation of matriptase activity in epithelial cells. This physiological mechanism appears, however, to largely be lost in cancer cells since matriptase-antithrombin complexes were not detected in all but two of a panel of seven breast, prostate, and ovarian cancer cell lines. Using purified active matriptase, we further characterize the formation of matriptase-antithrombin complex and show that heparin can significantly potentiate the inhibitory potency of antithrombin against matriptase. Second-order rate constants for the inhibition were determined to be 3.9 × 10(3) M(-1)s(-1) in the absence of heparin and 1.2 × 10(5) M(-1)s(-1) in the presence of heparin, a 30-fold increase, consistent with the established role of heparin in activating antithrombin function. Taken together these data suggest that normal epithelial cells employ a dual mechanism involving HAI-1 and antithrombin to control matriptase and that the antithrombin-based mechanism appears lost in the majority of carcinoma cells.     

Na+ transport in normal and CF human bronchial epithelial cells is inhibited by BAY 39-9437

To test the hypothesis that Na+ transport in human bronchial epithelial (HBE) cells is regulated by a protease-mediated mechanism, we investigated the effects of BAY 39-9437, a recombinant Kunitz-type serine protease inhibitor, on amiloride-sensitive short-circuit current of normal [non-cystic fibrosis (CF) cells] and CF HBE cells. Mucosal treatment of non-CF and CF HBE cells with BAY 39-9437 decreased the short-circuit current, with a half-life of approximately 45 min. At 90 min, BAY 39-9437 (470 nM) reduced Na+ transport by approximately 70%. The inhibitory effect of BAY 39-9437 was concentration dependent, with a half-maximal inhibitory concentration of approximately 25 nM. Na+ transport was restored to control levels, with a half-life of approximately 15 min, on washout of BAY 39-9437. In addition, trypsin (1 microM) rapidly reversed the inhibitory effect of BAY 39-9437. These data indicate that Na+ transport in HBE cells is activated by a BAY 39-9437-inhibitable, endogenously expressed serine protease. BAY 39-9437 inhibition of this serine protease maybe of therapeutic potential for the treatment of Na+ hyperabsorption in CF.     

Characterization of senescence-associated proteases in postharvest broccoli florets.

We characterized the senescence-associated proteases of postharvest broccoli (Brassica oleracea L. var Green King) florets, using class-specific protease inhibitors and gelatin-polyacrylamide gel electrophoresis. Different classes of senescence-associated proteases in broccoli florets were partially characterized for the first time. Protease activity of broccoli florets was depressed by all the inhibitors and showed different inhibition curves during postharvest. The hydrolytic activity of metalloprotease (EC 3.4.24. - ) and serine protease (EC 3.4.21. - ) reached a maximum, 1 day after harvest (DAH), then decreased, while the hydrolytic activity of cysteine protease (EC 3.4.22. - ) and aspartic protease (EC 3.4.23. - ) increased throughout the postharvest senescence based on the calculated inhibition percentage of protease activity. The senescence-associated proteases were separated into seven endoprotease (EP) groups by gelatin-polyacryamide gel electrophoresis and classified into EP1 (metalloprotease), EP2 (metalloprotease and cysteine protease), EP3 (serine protease and aspartic protease), EP4, EP5, EP7 (cysteine protease), and EP6 (serine protease) based on the sensitivity of class-specific protease inhibitors. The proteases EP2, EP3, and EP4 were present throughout the postharvest stages. EP3 was the major EP at all times during senescence; EP4 intensity of activity increased after 2 DAH; EP6 and EP7 clearly increased after 4 DAH. Our results suggest that serine protease activity contributes to early stage (0-1 DAH) and late stage (4-5 DAH) of senescence; metalloprotease activity was involved in the early and intermediate stages (0-3 DAH) of senescence; and cysteine protease and aspartic protease activities participated in the whole process of broccoli senescence.     

Characterization of recombinant human protein C inhibitor expressed in Escherichia coli

The serine protease inhibitor (serpin) protein C inhibitor (PCI; also named plasminogen activator inhibitor-3) regulates serine proteases in hemostasis, fibrinolysis, and reproduction. The biochemical activity of PCI is not fully defined partly due to the lack of a convenient expression system for active rPCI. Using pET-15b plasmid, Ni(2+)-chelate and heparin-Sepharose affinity chromatography steps, we describe here the expression, purification and characterization of wild-type recombinant (wt-rPCI) and two inactive mutants, R354A (P1 residue) and T341R (P14 residue), expressed in Escherichia coli. Wild-type rPCI, but not the two mutants, formed a stable bimolecular complex with thrombin, activated protein C and urokinase. In the absence of heparin, wt-rPCI-thrombin, -activated protein C, and -urokinase inhibition rates were 56.7, 3.4, and 2.3 x 10(4) M(-1) min(-1), respectively, and the inhibition rates were accelerated 25-, 71-, and 265-fold in the presence of 10 mug/mL heparin for each respective inhibition reaction. The stoichiometry of inhibition (SI) for wt-rPCI-thrombin was 2.0, which is comparable to plasma-derived PCI. The present report describes for the first time the expression and characterization of recombinant PCI in a bacterial expression system and demonstrates the feasibility of using this system to obtain adequate amounts of biologically active rPCI for future structure-function studies.     

Crystal structure of enteropeptidase light chain complexed with an analog of the trypsinogen activation peptide.

Enteropeptidase is a membrane-bound serine protease that initiates the activation of pancreatic hydrolases by cleaving and activating trypsinogen. The enzyme is remarkably specific and cleaves after lysine residues of peptidyl substrates that resemble trypsinogen activation peptides such as Val-(Asp)4-Lys. To characterize the determinants of substrate specificity, we solved the crystal structure of the bovine enteropeptidase catalytic domain to 2.3 A resolution in complex with the inhibitor Val-(Asp)4-Lys-chloromethane. The catalytic mechanism and contacts with lysine at substrate position P1 are conserved with other trypsin-like serine proteases. However, the aspartyl residues at positions P2-P4 of the inhibitor interact with the enzyme surface mainly through salt bridges with the Nzeta atom of Lys99. Mutation of Lys99 to Ala, or acetylation with acetic anhydride, specifically prevented the cleavage of trypsinogen or Gly-(Asp)4-Lys-beta-naphthylamide and reduced the rate of inhibition by Val-(Asp)4-Lys-chloromethane 22 to 90-fold. For these reactions, Lys99 was calculated to account for 1.8 to 2.5 kcal mol(-1) of the free energy of transition state binding. Thus, a unique basic exosite on the enteropeptidase surface has evolved to facilitate the cleavage of its physiological substrate, trypsinogen.     

Hepatocyte growth factor activator inhibitor type 1 inhibits protease activity and proteolytic activation of human airway trypsin-like protease

Hepatocyte growth factor activator inhibitor type 1 (HAI-1) is a Kunitz-type transmembrane serine protease inhibitor initially identified as a potent inhibitor of hepatocyte growth factor activator (HGFA), a serine protease that converts pro-HGF to the active form. HAI-1 also has inhibitory activity against serine proteases such as matriptase, hepsin and prostasin. In this study, we examined effects of HAI-1 on the protease activity and proteolytic activation of human airway trypsin-like protease (HAT), a transmembrane serine protease that is expressed mainly in bronchial epithelial cells. A soluble form of HAI-1 inhibited the protease activity of HAT in vitro. HAT was proteolytically activated in cultured mammalian cells transfected with its expression vector, and a soluble form of active HAT was released into the conditioned medium. The proteolytic activation of HAT required its own serine protease activity. Co-expression of the transmembrane full-length HAI-1 inhibited the proteolytic activation of HAT. In addition, full-length HAI-1 associated with the transmembrane full-length HAT in co-expressing cells. Like other target proteases of HAI-1, HAT converted pro-HGF to the active form in vitro. These results suggest that HAI-1 functions as a physiological regulator of HAT by inhibiting its protease activity and proteolytic activation in airway epithelium.     

Epi p 1, an allergenic glycoprotein of Epicoccum purpurascens is a serine protease

Epicoccum purpurascens (EP) is a ubiquitous saprophytic mould, the inhalant spores and mycelia of which are responsible for respiratory allergic disorders in 5-7% of population worldwide. The diagnosis/therapy of these disorders caused by fungi involves the use of standardized and purified fungal extracts. A 33.5 kDa glycoprotein, Epi p 1 released histamine from whole blood cells of EP allergic patients at a concentration of 50-ng protein. The high specific IgE values detected in EP hypersensitive sera indicated that Epi p 1 is capable of mediating type I hypersensitive reaction in predisposed individuals. It also showed protease activity by virtue of its dose dependent cleavage of serine protease specific synthetic substrate, N-benzoyl arginine ethyl ester hydrochloride (BAEE). The serine protease nature of Epi p 1 was confirmed by its N-terminal sequence (ADG/FIVAVELD/STY) homology to a subtilisin like serine protease. The protease activity of Epi p 1 may be responsible for making its way into the system of pre-disposed individuals through epithelial cell detachment and the histamine releasing ability by cross-linking of IgE antibodies on cell surface is the cause of its allergenic nature.     

Chemical replacement of P1' serine residue at the second reactive site of soybean protease inhibitor C-II

The P1' Ser(50) at the second reactive site of soybean protease inhibitor C-II was replaced with arginine to confirm the contribution of this residue to the inhibition. The Arg derivative had less trypsin inhibitory activity (Ki = 1 X 10(-7) M) than the Ser derivative (Ki = 2 X 10(-8) M), in contrast to the results obtained from studies on peanut protease inhibitor B-III reported in the previous paper (J. Biochem. 101, 723-728 (1987)). These results suggest that each Bowman-Birk type inhibitor has an amino acid at the P1' position inherently best suited to maintaining its inhibitory activity, and that serine is not unique for the P1' amino acid in Bowman-Birk type inhibitors.     

Structural basis of trypsin inhibition and entomotoxicity of cospin, serine protease inhibitor involved in defense of Coprinopsis cinerea fruiting bodies

Cospin (PIC1) from Coprinopsis cinerea is a serine protease inhibitor with biochemical properties similar to those of the previously characterized fungal serine protease inhibitors, cnispin from Clitocybe nebularis and LeSPI from Lentinus edodes, classified in the family I66 of the MEROPS protease inhibitor classification. In particular, it exhibits a highly specific inhibitory profile as a very strong inhibitor of trypsin with K(i) in the picomolar range. Determination of the crystal structure revealed that the protein has a β-trefoil fold. Site-directed mutagenesis and mass spectrometry results have confirmed Arg-27 as the reactive binding site for trypsin inhibition. The loop containing Arg-27 is positioned between the β2 and β3 strands, distinguishing cospin from other β-trefoil-fold serine protease inhibitors in which β4-β5 or β5-β6 loops are involved in protease inhibition. Biotoxicity assays of cospin on various model organisms revealed a strong and specific entomotoxic activity against Drosophila melanogaster. The inhibitory inactive R27N mutant was not entomotoxic, associating toxicity with inhibitory activity. Along with the abundance of cospin in fruiting bodies of C. cinerea and the lack of trypsin-like proteases in the C. cinerea genome, these results suggest that cospin and its homologs are effectors of a fungal defense mechanism against fungivorous insects that function by specific inhibition of serine proteases in the insect gut.     

Hybrid protease inhibitors for pest and pathogen control – a functional cost for the fusion partners?

Fusion proteins integrating dual pesticidal functions have been devised over the last 10 years to improve the effectiveness and potential durability of pest-resistant transgenic crops, but little attention has been paid to the impact of the fusion partners on the actual activity of the resulting hybrids. Here we assessed the ability of the rice cysteine protease inhibitor, oryzacystatin I (OCI), to retain its protease inhibitory potency when used as a template to devise hybrid inhibitors with dual activity against papain-like proteases and carboxypeptidase A (CPA). C-terminal variants of OCI were generated by fusing to its C-terminal end: (i) the primary inhibitory site of the small CPA inhibitor potato carboxypeptidase inhibitor (PCI, amino acids 35-39); or (ii) the complete sequence of PCI (a.a. 1-39). The hybrid inhibitors were expressed in E. coli and tested for their inhibitory activity against papain, CPA and digestive cysteine proteases of herbivorous and predatory arthropods. In contrast with the primary inhibitory site of PCI, the entire PCI attached to OCI was as active against CPA as free, purified PCI. The OCI-PCI hybrids also showed activity against papain, but the presence of extra amino acids at the C terminus of OCI negatively altered its inhibitory potency against cysteine proteases. This negative effect, although not preventing dual binding to papain and CPA, was correlated with an increased binding affinity for papain presumably due to non-specific interactions with the PCI domain. These results confirm the potential of OCI and PCI for the design of fusion inhibitors with dual protease inhibitory activity, but also point out the possible functional costs associated with protein domain grafting to recipient pesticidal proteins.     

Intracellular localization of protein C inhibitor (PCI) and urinary plasminogen activator in renal tubular epithelial cells from humans and human PCI gene transgenic mice

Urinary plasminogen activator (uPA) is a serine protease that plays important roles in various extracellular proteolytic processes. In humans, protein C inhibitor (PCI) is known to regulate the activity of the serine proteases involved in blood coagulation, wound healing, and tumor metastasis, whereas PCI is not present in murine plasma or tissues other than the reproductive tissues. The large amount of uPA–PCI complexes found in human urine suggests that these complexes are formed in the kidneys. In the present study, we performed immunofluorescence double labeling and electron microscopic immunocytochemistry using renal tissues from humans and human PCI gene transgenic (PCI-TG) mice. In human renal tissues, PCI and uPA colocalized in the cytoplasm of renal proximal tubular epithelial cells (RPTECs), and juxtaposition of PCI and uPA immunoreactive particles was detected in the microvilli and lysosomes in the RPTECs. The intracellular distributions of PCI and uPA in the RPTECs from PCI-TG mice were similar to those observed in human RPTECs. These findings hint at the physiological roles of uPA and PCI in human kidneys, and also suggest that the PCI-TG mice will be useful for evaluating the roles of PCI in human physiological and pathological conditions.     

Behavioural and Electroantennogram Responses of Male Cigarette Beetle (Lasioderma serricorne F.) to Optically Active Serricornins

Pievious work with MAPI, a serine protease inhibitor, has shown that inactivation of membrane bound protease by MAPI resulted in inhibition of normal sporulation of Bacillus subtilis IFO 3027 [Shimizu et al, Agric. Biol. Chem., 48, 365 (1984)]. In the cells cultured with MAPI, the cellular amount of IP-I, a cytoplasmic serine protease which is sensitive to EDTA was lower than the control cells. An endogenous proteinaceous inhibitor having specific inhibitory activity against IP-I was produced during the sporulation and its amount in the MAPI-treated cells was higher than that of control cells. The proteinaceous inhibitor was inactivated only by membrane bound protease. Consequently, IP-I was activated through degradation of proteinaceous inhibitor by membrane bound protease. It seems probable that the proteinaceous inhibitor and membrane bound protease are involved in the regulation of a protease system in sporulating cells of B. subtilis.     

Mutation of the H-helix in antithrombin decreases heparin stimulation of protease inhibition

Blood clotting proceeds through the sequential proteolytic activation of a series of serine proteases, culminating in thrombin cleaving fibrinogen into fibrin. The serine protease inhibitors (serpins) antithrombin (AT) and protein C inhibitor (PCI) both inhibit thrombin in a heparin-accelerated reaction. Heparin binds to the positively charged D-helix of AT and H-helix of PCI. The H-helix of AT is negatively charged, and it was mutated to contain neutral or positively charged residues to see if they contributed to heparin stimulation or protease specificity in AT. To assess the impact of the H-helix mutations on heparin stimulation in the absence of the known heparin-binding site, negative charges were also introduced in the D-helix of AT. AT with both positively charged H- and D-helices showed decreases in heparin stimulation of thrombin and factor Xa inhibition by 10- and 5-fold respectively, a decrease in affinity for heparin sepharose, and a shift in the heparin template curve. In the absence of a positively charged D-helix, changing the H-helix from neutral to positively charged increased heparin stimulation of thrombin inhibition 21-fold, increased heparin affinity and restored a normal maximal heparin concentration for inhibition.     

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

HAI-1 [HGF (hepatocyte growth factor) activator inhibitor-1] is a Kunitz-type transmembrane serine protease inhibitor that forms inhibitor complexes with the trypsin-like serine protease, matriptase. HAI-1 is essential for mouse placental development and embryo survival and together with matriptase it is a key regulator of carcinogenesis. HAI-1 is expressed in polarized epithelial cells, which have the plasma membrane divided by tight junctions into an apical and a basolateral domain. In the present study we show that HAI-1 at steady-state is mainly located on the basolateral membrane of both Madin-Darby canine kidney cells and mammary gland epithelial cells. After biosynthesis, HAI-1 is exocytosed mainly to the basolateral plasma membrane from where 15% of the HAI-1 molecules are proteolytically cleaved and released into the basolateral medium. The remaining membrane-associated HAI-1 is endocytosed and then recycles between the basolateral plasma membrane and endosomes for hours until it is transcytosed to the apical plasma membrane. Minor amounts of HAI-1 present at the apical plasma membrane are proteolytically cleaved and released into the apical medium. Full-length membrane-bound HAI-1 has a half-life of 1.5 h and is eventually degraded in the lysosomes, whereas proteolytically released HAI-1 is more stable. HAI-1 is co-localized with its cognate protease, matriptase, at the basolateral plasma membrane. We suggest that HAI-1, in addition to its protease inhibitory function, plays a role in transporting matriptase as a matriptase-HAI-1 complex from the basolateral plama membrane to the apical plasma membrane, as matriptase is known to interact with prostasin, located at the apical plasma membrane.     

Protein C inhibitor regulates both cathepsin L activity and cell-mediated tumor cell migration

Background

Protein C inhibitor (PCI) is a plasma serine protease inhibitor (serpin) that regulates several serine proteases in coagulation including thrombin and activated protein C. However, the physiological role of PCI remains under investigation. The cysteine protease, cathepsin L, has a role in many physiological processes including cardiovascular diseases, blood vessel remodeling, and cancer.

Methods and results

We found that PCI inhibits cathepsin L with an inhibition rate (k₂) of 3.0 × 10⁵ M⁻¹ s⁻¹. Whereas, the PCI P1 mutant (R354A) inhibits cathepsin L at rates similar to wild-type PCI, mutating the P2 residue results in a slight decrease in the rate of inhibition. We then assessed the effect of PCI and cathepsin L on the migration of human breast cancer (MDA-MB-231) cells. Cathepsin L was expressed in both the cell lysates and conditioned media of MDA-MB-231 cells. Wound-induced and transwell migration of MDA-MB-231 cells was inhibited by exogenously administered wtPCI and PCI P1 but not PCI P14 mutant. In addition, migration of MDA-MB-231 cells expressing wtPCI was significantly decreased compared to non-expressing MDA-MB-231 cells or MDA-MB-231 cells expressing the PCI P14 mutant. Downregulation of cathepsin L by either a specific cathepsin L inhibitor or siRNA technology also resulted in a decrease in the migration of MDA-MB-231 cells.

Conclusions

Overall, our data show that PCI regulates tumor cell migration partly by inhibiting cathepsin L.

General significance

Consequently, inhibiting cathepsin L by serpins like PCI may be a new pathway of regulating hemostasis, cardiovascular and metastatic diseases.     

Purification from human milk of matriptase complexes with secreted serpins: mechanism for inhibition of matriptase other than HAI-1

Matriptase, a type 2 transmembrane serine protease, is predominately expressed by epithelial and carcinoma cells in which hepatocyte growth factor activator inhibitor 1 (HAI-1), a membrane-bound, Kunitz-type serine protease inhibitor, is also expressed. HAI-1 plays dual roles in the regulation of matriptase, as a conventional protease inhibitor and as a factor required for zymogen activation of matriptase. As a consequence, activation of matriptase is immediately followed by HAI-1-mediated inhibition, with the activated matriptase being sequestered into HAI-1 complexes. Matriptase is also expressed by peripheral blood leukocytes, such as monocytes and macrophages; however, in contrast to epithelial cells, monocytes and macrophages were reported not to express HAI-1, suggesting that these leukocytes possess alternate, HAI-1-independent mechanisms regulating the zymogen activation and protease inhibition of matriptase. In the present study, we characterized matriptase complexes of 110 kDa in human milk, which contained no HAI-1 and resisted dissociation in boiling SDS in the absence of reducing agents. These complexes were further purified and dissociated into 80-kDa and 45-kDa fragments by treatment with reducing agents. Proteomic and immunological methods identified the 45-kDa fragment as the noncatalytic domains of matriptase and the 80-kDa fragment as the matriptase serine protease domain covalently linked to one of three different secreted serpin inhibitors: antithrombin III, 1-antitrypsin, and 2-antiplasmin. Identification of matriptase-serpin inhibitor complexes provides evidence for the first time that the proteolytic activity of matriptase, from those cells that express no or low levels of HAI-1, may be controlled by secreted serpins. protease; type 2 transmembrane serine protease; protease inhibitor; ST-14; hepatocyte growth factor activator inhibitor 1     

Time-dependent inhibition of tuberculin-induced lymphocyte DNA synthesis by a serine protease inhibitor

Diisopropylfluorophosphate (DFP), a group-specific irreversible inhibitor of serine proteases, has been shown to exert time-dependent inhibition of DNA synthesis of lymphocytes stimulated by three different B lymphocyte mitogens: purified protein derivative of tuberculin (PPD), endotoxin protein (EP), and lipopolysaccharide (LPS). The time-dependent inhibition profile found in B lymphocytes is absent in concanavalin A (Con A)-stimulated T lymphocytes. Structural analogs of DFP, which have lost the phosphorylating ability, are not inhibitory. Inhibition of DNA synthesis by DFP is reversible in the first 8 hr of mitogenic stimulation. Maximal and irreversible inhibition by DFP occurs around the 16th hour of stimulation. These data support the postulate that a mitogenesis-linked protease, or proteases, in B lymphocytes is absent in the resting cells but is made available several hours before the initiation of DNA synthesis in the late G1 phase of the cell cycle.     

Assessment of the Interaction Between Urokinase and Reactive Site Mutants of Protein C Inhibitor

Urokinase-type plasminogen activator (uPA) is a serine protease involved in pericellular proteolysis and tumor cell metastasis via plasmin-mediated degradation of extracellular matrix proteins. Plasma uPA is inhibited by the serine protease inhibitor protein C inhibitor (PCI) by the insertion of PCI's reactive site loop into the active site of the protease. To better understand the structural aspects of this inhibition, 15 reactive-site mutants of recombinant PCI (rPCI) were assayed for differences in uPA inhibition. These assays revealed that substitutions at the P1 Arg354 and P3 Thr352 sites of rPCI were detrimental to inhibitory activity, while P3 Arg357 mutations had little effect upon the inhibition rate. However, replacement of the P2 Phe353 with small residues like Ala and Gly increased the effectiveness of rPCI three- to four fold. To explain these altered rates of inhibition, a computer-derived molecular model of uPA was generated and docked to a model of PCI to simulate complex formation. The changes made by mutagenesis were then recreated in the model of uPA–PCI. In accordance with the kinetic data, the poor performance of P3 variants is primarily attributable to charge repulsion, while alleviation of steric hindrance at P2 produces the observed increase in uPA inhibition. In the model, residues at P3 interact with PCI rather than uPA, consistent with P3 variants demonstrating that little variation from wild-type activity. Ultimately, this combination of mutagenesis and molecular modeling will further refine our understanding of the interaction between PCI and uPA.     

Requirement of the activity of hepatocyte growth factor activator inhibitor type 1 for the extracellular appearance of a transmembrane serine protease matriptase in monkey kidney COS-1 cells

Hepatocyte growth factor activator inhibitor type I (HAI-1) is a membrane-bound, serine protease inhibitor with two protease-inhibitory domains (Kunitz domain I and II). HAI-1 is known as a physiological inhibitor of a membrane-bound serine protease, matriptase. Paradoxically, however, HAI-1 has been found to be required for the extracellular appearance of the protease in an expression system using a monkey kidney COS-1 cell line. In the present study, we show using COS-1 cells that co-expression of recombinant variants of HAI-1 with the inhibition activity toward matriptase, including a variant consisting only of Kunitz domain I (the domain responsible for inhibition of matriptase), allowed for the appearance of this protease in the conditioned medium, whereas that of the variants without the activity did not. These findings suggest that the inhibition activity toward matriptase is critical for the extracellular appearance of protease in COS-1 cells.