Journal list menu

Volume 277, Issue 23 p. 4888-4900
Free Access

TMPRSS13, a type II transmembrane serine protease, is inhibited by hepatocyte growth factor activator inhibitor type 1 and activates pro-hepatocyte growth factor

Tomio Hashimoto

Tomio Hashimoto

Department of Biological Sciences, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan

Search for more papers by this author
Minoru Kato

Minoru Kato

Advanced Medical Research Laboratory, Mitsubishi Tanabe Pharma Corporation, Kamoshida-cho, Aoba-ku, Yokohama, Japan

Search for more papers by this author
Takeshi Shimomura

Takeshi Shimomura

Advanced Medical Research Laboratory, Mitsubishi Tanabe Pharma Corporation, Kamoshida-cho, Aoba-ku, Yokohama, Japan

Search for more papers by this author
Naomi Kitamura

Naomi Kitamura

Department of Biological Sciences, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan

Search for more papers by this author
First published: 28 September 2010
Citations: 41
N. Kitamura, Department of Biological Sciences, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama 226-8501, Japan
Fax: +81 45 924 5771
Tel: +81 45 924 5701
E-mail: [email protected]

Abstract

Type II transmembrane serine proteases (TTSPs) are structurally defined by the presence of a transmembrane domain located near the N-terminus and a C-terminal extracellular serine protease domain. The human TTSP family consists of 17 members. Some members of the family have pivotal functions in development and homeostasis, and are involved in tumorigenesis and viral infections. The activities of TTSPs are regulated by endogenous protease inhibitors. However, protease inhibitors of most TTSPs have not yet been identified. In this study, we investigated the inhibitory effect of hepatocyte growth factor activator inhibitor type 1 (HAI-1), a Kunitz-type serine protease inhibitor, on several members of the TTSP family. We found that the protease activity of a member, TMPRSS13, was inhibited by HAI-1. A detailed analysis revealed that a soluble form of HAI-1 with one Kunitz domain (NK1) more strongly inhibited TMPRSS13 than another soluble form of HAI-1 with two Kunitz domains (NK1LK2). In addition, an in vitro protein binding assay showed that NK1 formed complexes with TMPRSS13, but NK1LK2 did not. TMPRSS13 converted single-chain pro-hepatocyte growth factor (pro-HGF) to a two-chain form in vitro, and the pro-HGF converting activity of TMPRSS13 was inhibited by NK1. The two-chain form of HGF exhibited biological activity, assessed by phosphorylation of the HGF receptor (c-Met) and extracellular signal-regulated kinase, and scattered morphology in human hepatocellular carcinoma cell line HepG2. These results suggest that TMPRSS13 functions as an HGF-converting protease, the activity of which may be regulated by HAI-1.

Abbreviations

  • BSA
  • bovine serum albumin
  • ERK
  • extracellular signal-regulated kinase
  • HA
  • haemagglutinin
  • HAI-1
  • hepatocyte growth factor activator inhibitor type 1
  • HAI-2
  • hepatocyte growth factor activator inhibitor type 2
  • HGF
  • hepatocyte growth factor
  • HGFA
  • hepatocyte growth factor activator
  • HPAI
  • highly pathogenic avian influenza
  • IC50
  • the concentration of inhibitor that inhibited the enzymatic activity by 50% compared with the uninhibited control
  • LDL
  • low-density lipoprotein
  • MSPL
  • mosaic serine protease large form
  • PBS
  • phosphate-buffered saline
  • TTSP
  • type II transmembrane serine protease
  • Introduction

    Type II transmembrane serine proteases (TTSPs) are structurally defined by the presence of a short N-terminal cytoplasmic domain, a transmembrane domain located near the N-terminus, and a C-terminal extracellular serine protease domain. In addition, TTSPs possess a stem region that may contain a diverse array of protein domains [1,2]. The human TTSP family consists of 17 members, which are classified into four subfamilies [2]. TTSPs are synthesized as inactive single-chain pro-enzymes, the proteolytic cleavage of which is required for the enzymes to exert their activity [2]. Several members of the TTSP family have been shown to have pivotal functions in development and homeostasis [1,2]. Moreover, recent studies revealed that some members are involved in tumorigenesis and viral infections [3]. However, the physiological and pathological functions of most members of the TTSP family remain to be investigated.

    The activities of some members of the TTSP family are regulated by endogenous protease inhibitors, which include Kunitz-type inhibitors and serpins [2]. Hepatocyte growth factor activator inhibitor type 1 (HAI-1), a Kunitz-type serine protease inhibitor, is implicated in the inhibition of two members of the TTSP family, matriptase and hepsin. HAI-1 was originally identified as a potent inhibitor of hepatocyte growth factor activator (HGFA), a blood coagulation factor XII-like serine protease that converts pro-hepatocyte growth factor (pro-HGF) to the active form [4]. HAI-1 was also isolated from human milk in a complex with matriptase, and potentially inhibits the protease activity of matriptase [5]. The physiological role of the inhibition of matriptase by HAI-1 was determined by analysing knockout mice. The homozygous deletion of HAI-1 resulted in embryonic lethality due to impaired formation of the placental labyrinth layer [6,7], whereas matriptase/HAI-1 double-deficient mice formed the placental labyrinth and developed to term, indicating an essential role of the inhibition of matriptase by HAI-1 during placental development in the mouse embryo [8]. Hepsin has an ability to convert pro-HGF to the active form with an activity comparable with HGFA. The HGF-converting activity is inhibited by HAI-1 [9,10].

    The protease inhibitors that regulate the activities of most TTSPs have not been identified yet. Because HAI-1 is a potent inhibitor of matriptase and hepsin, it might also inhibit the protease activities of other TTSPs. To test this possibility, we have searched for TTSPs targeted by HAI-1, and found that the activity of TMPRSS13 is potentially inhibited by HAI-1.

    TMPRSS13 is a splice variant of mosaic serine protease large form (MSPL), and belongs to the hepsin/TMPRSS subfamily of the TTSP family. MSPL and TMPRSS13 were isolated by a PCR-based screening from a human lung cDNA library using degenerate primers designed on the basis of the conserved catalytic motif of known trypsin-type serine proteases [11,12]. The amino acid sequence of TMPRSS13 is identical to that of MSPL except for an insertion of five amino acids in the N-terminal cytoplasmic region and the C-terminal end following the protease domain, in which TMPRSS13 has eight amino acids and MSPL has a different 27 amino acids [12]. MSPL and TMPRSS13 preferentially recognize cleavage sites consisting of paired basic amino acid residues [12]. Recently, MSPL and TMPRSS13 have been shown to be candidates for haemagglutinin (HA)-processing proteases of highly pathogenic avian influenza (HPAI) viruses. Namely, a full-length recombinant HA of an HPAI virus was efficiently converted to mature HA subunits with membrane-fused giant cell formation in MSPL- or TMPRSS13-transfected cells, but not in untransfected cells. Furthermore, infection and multiplication of the HPAI virus were detected in the transfected cells [13]. MSPL and TMPRSS13 are expressed in a variety of tissues, and predominantly in lung, placenta, pancreas and prostate [12]. Therefore, in addition to the function in HA processing, MSPL and TMPRSS13 may have physiological functions in these tissues that remain to be explored.

    Here, we characterize in detail the inhibitory effect of HAI-1 on TMPRSS13. Moreover, we demonstrate a possible physiological function of TMPRSS13, that is its HGF-converting activity.

    Results

    Search for TTSPs targeted by HAI-1

    To search for targets of HAI-1, we constructed Escheri-chia coli expression vectors encoding protease domains with short pro-sequences of six members of the TTSP family. These proteases have been shown to be co-expressed with HAI-1 in various tissues (database of BioExpress System, Gene Logic Inc., Gaithersburg, MD, USA). Then, the putative activation cleavage sequences were replaced with the enterokinase recognition sequence (DDDDK) for activation in vitro. Escherichia coli cells were transformed with the expression vectors, and expressed proteins were purified from cell lysate. The purified proteins were treated with enterokinase, and protease activity was measured using synthetic substrates suitable for each TTSP. TMPRSS3 and TMPRSS4 expressed in this system did not show protease activity. Thus, other TTSPs that did show activity were tested for the inhibitory activity of HAI-1 using the first Kunitz domain of HAI-1 (HAI-1–K1). Among these TTSPs, TMPRSS11A, HAT-like 4 and HAT-like 5 were not inhibited by HAI-1–K1. By contrast, the protease activity of TMPRSS13 was potentially inhibited by HAI-1–K1. We therefore characterized the inhibitory activity of HAI-1 against TMPRSS13 in detail.

    Preparation and activation of a secreted form of pro-TMPRSS13 expressed in mammalian cells

    TMPRSS13 expressed in E. coli showed weak protease activity, probably because of incorrect protein folding. We therefore expressed pro-TMPRSS13 in mammalian cells. To obtain pro-TMPRSS13 from conditioned medium of mammalian cells, we constructed an expression vector encoding a secreted form of this protein that lacked the cytoplasmic and transmembrane domains. In addition, the putative activation cleavage sequence (AMTGR325) was replaced with the enterokinase recognition sequence (DDDDK) for activation in vitro, and the protein was tagged at the C-terminus with myc-His for purification and immunoblot analysis (Fig. 1A). COS-7 cells were transiently transfected with the expression vector. The protein was purified from the conditioned medium of the transfected cells. The immunoblot analysis of the purified protein using an anti-c-Myc IgG showed a band of 63 kDa under reducing and nonreducing conditions (Fig. 1B,C), indicating that pro-TMPRSS13 was highly expressed in this system.

    Details are in the caption following the image

    Production and activation of the recombinant pro-TMPRSS13. (A) Schematic representation of the structure of pro-TMPRSS13 (wild-type), the recombinant pro-TMPRSS13 and the enterokinase-cleaved pro-TMPRSS13. The wild-type pro-TMPRSS13 comprises 567 amino acids. The amino acid numbering starts from the putative N-terminus of the protein. The domain structures are indicated in pro-TMPRSS13 (wild-type). TM, transmembrane domain; LDLA, LDL receptor class A domain; SRCR, scavenger receptor cysteine-rich domain; SPD, serine protease domain. The predicted disulfide linkage is shown as SS. The putative activation cleavage site (indicated by an arrow) and its surrounding sequence are shown in pro-TMPRSS13 (wild-type). The recombinant pro-TMPRSS13 is a secreted form in which the cytoplasmic domain and transmembrane domain (Met1-Gln186) are replaced with the mouse immunoglobulin κ-chain signal peptide. In addition, AMTGR325 in the wild-type protein is replaced with the enterokinase recognition sequence (DDDDK, underlined) for cleavage in vitro before Ile326 (activation cleavage). The recombinant pro-TMPRSS13 is tagged at the C-terminus with myc-His. The enterokinase-cleaved recombinant pro-TMPRSS13, the disulfide-linked two-chain form, is illustrated at the bottom. (B, C) Immunoblot analysis of the recombinant pro-TMPPRSS13 produced in COS-7 cells, and its enterokinase-treated product. Samples of pro-TMPRSS13 (lane 1) and enterokinase-treated pro-TMPRSS13 (lane 2) were separated by SDS/PAGE under reducing conditions (B) or under nonreducing conditions (C), and analysed by immunoblotting with the anti-c-Myc IgG. The protease domain of TMPRSS13 was quantified by scanning densitometry of the immunoblot (B, lane 2) with NIH imagej software using the protease domain of TMPRSS13, which was expressed in E. coli, as a standard.

    To activate pro-TMPRSS13, we treated the protein with enterokinase. The immunoblot analysis of the reaction product using the anti-c-Myc IgG showed a band of 37 kDa under reducing conditions (Fig. 1B), and that of 67 kDa under nonreducing conditions (Fig. 1C). The 37 kDa band probably corresponded to the protease domain of TMPRSS13, suggesting the proteolytic activation of the pro-protein. Detection of the 67 kDa band suggests that the pro-protein was cleaved at a single site, and the cleaved protein is a two-chain form linked by a disulfide bond. The protease domain of TMPRSS13 was quantified by scanning densitometry of the immunoblot, using the protease domain of the TMPRSS13 expressed in E. coli as a standard. The protease activity of the enterokinase-treated pro-TMPRSS13 was measured using a synthetic substrate (Pyr–RTKR–MCA), which has been shown as an efficient substrate of the protease [13]. This substrate was not cleaved by enterokinase itself, or by the untreated pro-TMPRSS13. The enterokinase-treated pro-TMPRSS13 efficiently cleaved the substrate, and thus was used for an assay of inhibition by HAI-1.

    Inhibition of TMPRSS13 protease activity by soluble HAI-1

    Inhibition of the protease activity of TMPRSS13 was assessed using recombinant soluble forms of HAI-1, HAI-1–NK1 and HAI-1–NK1LK2. HAI-1 is first produced as a 66 kDa transmembrane form, and subsequent ectodomain shedding releases two major soluble forms of 40 and 58 kDa from the cell surface into the extracellular space [14]. HAI-1–NK1, which corresponds to the 40 kDa form, consists of the N-terminal region (N) and one Kunitz domain (K1), whereas HAI-1–NK1LK2, corresponding to the 58 kDa form, consists of the N-terminal region (N), two Kunitz domains (K1 and K2), and the low-density lipoprotein (LDL) receptor class A domain (L) between the Kunitz domains (Fig. 2A). Inhibition by aprotinin was compared with that by HAI-1, because aprotinin has been shown to efficiently inhibit the protease activity of TMPRSS13 [12]. TMPRSS13 (100 pm) was incubated with various concentrations of HAI-1–NK1, HAI-1–NK1LK2 and aprotinin, and protease activity was measured using the synthetic substrate. Figure 2C shows the dose dependence of the inhibitory activities. HAI-1–NK1 had the most potent inhibitory effect (IC50 = 2.18 ± 0.18 nm). HAI-1–NK1LK2 and aprotinin showed much weaker inhibitory activity than HAI-1–NK1.

    Details are in the caption following the image

    Dose dependence of the inhibitory activity of soluble forms of HAI-1 and HAI-2 against the protease activity of TMPRSS13. (A) Schematic representation of the structure of the full-length HAI-1 (1) and soluble forms of HAI-1, HAI-1–NK1LK2 (2) and HAI-1–NK1 (3), tagged at the C-terminus with myc-His. SP, signal peptide; N, N-terminal region; K1, Kunitz domain 1; LDLA, LDL receptor class A domain; K2, Kunitz domain 2; TM, transmembrane domain. (B) Schematic representation of the structure of the full-length HAI-2 (1) and a soluble form of HAI-2 tagged at the C-terminals with myc-His (2). (C) Dose dependence of the inhibitory activity of soluble forms of HAI-1 and HAI-2 against the protease activity of TMPRSS13. TMPRSS13 was incubated with various concentrations of HAI-1–NK1 (•), HAI-1–NK1LK2 (▪), aprotinin () or HAI-2 (◆). Then, Pyr-RTKR-MCA was added, and after further incubation, the fluorescence of the reaction mixtures was measured. Data show the mean ± standard deviation for three separate experiments and are expressed as a percentage of TMPRSS13 activity.

    Hepatocyte growth factor activator inhibitor type 2 (HAI-2), also known as placental bikunin, is also a transmembrane Kunitz-type serine protease inhibitor [15,16]. HAI-2 has been shown to inhibit matriptase and hepsin [9,10,17]. Thus, we examined the effect of a soluble form of HAI-2 (Fig. 2B) on the protease activity of TMPRSS13. HAI-2 inhibited TMPRSS13 (IC50 = 1.54 ± 0.01 nm) (Fig. 2C), and the IC50 was similar to that of HAI-1–NK1. However, the inhibition curves were quite different: the inhibition curve of HAI-2 was sigmoidal, whereas that of HAI-1–NK1 was not (Fig. 2C).

    Formation of complexes of TMPRSS13 and HAI-1–NK1

    To confirm the inhibitory effect of HAI-1–NK1 on TMPRSS13, we examined the formation of complexes by the protease–inhibitor pair. HAI-1–NK1 and HAI-1–NK1LK2 were incubated with the activated TMPRSS13 at different molar ratios. The samples were boiled or not boiled, and subjected to an immunoblot analysis. Immunoblotting with an anti-HAI-1 IgG showed that increasing concentrations of TMPRSS13 shifted the HAI-1–NK1 band (40 kDa) to a higher molecular mass species (70 kDa) when the samples were not boiled (Fig. 3A). This shift was confirmed by an immunoblot analysis with an anti-TMPRSS13 IgG (Fig. 3B). When samples were boiled, the band did not shift (Fig. 3A,B). These results indicate the formation of TMPRSS13·HAI-1–NK1 complexes. On the other hand, the HAI-1–NK1LK2 band (58 kDa) did not shift to a high molecular mass species even in the presence of a high concentration of TMPRSS13 (Fig. 3A,B), which is consistent with data showing weak inhibitory activity of HAI-1–NK1LK2 against the protease activity of TMPRSS13.

    Details are in the caption following the image

    TMPRSS13 forms complexes with HAI-1–NK1. TMPRSS13 at the indicated concentrations was incubated with 20 nm HAI-1–NK1 and HAI-1–NK1LK2 at 37 °C for 2 h. After the addition of SDS sample buffer with 100 mm dithiothreitol, each sample was boiled or not boiled (as indicated). Samples were separated by SDS/PAGE under reducing conditions, and analysed by immunoblotting with anti-HAI-1 IgG (A) or anti-TMPRSS13 IgG (B). The asterisks indicate complexes of TMPRSS13 and HAI-1–NK1.

    Proteolytic activation of pro-HGF by TMPRSS13

    Pro-HGF is proteolytically activated by matriptase and hepsin, and the protease activity is inhibited by HAI-1 [9,10,18]. Therefore, we examined whether TMPRSS13 also functions as an HGF-converting protease. The single-chain pro-HGF (2 μm) was incubated with various concentrations of TMPRSS13. The reaction products were separated by SDS/PAGE under reducing conditions and stained with Coomassie Brilliant Blue. The incubation generated two main bands of ∼ 60 and 32 kDa (Fig. 4A). The sizes corresponded to the heavy chain and light chain of activated HGF, suggesting that pro-HGF is activated by TMPRSS13. The intensity of the pro-HGF band on the gel was quantified by scanning densitometry, and the percentage of HGF processed was calculated. Pro-HGF was almost completely converted to the two-chain form by 54 nm TMPRSS13 (Fig. 4B).

    Details are in the caption following the image

    Proteolytic conversion of pro-HGF by TMPRSS13 and its inhibition by HAI-1–NK1. (A) Pro-HGF (2 μm) was incubated with various concentrations of TMPRSS13. The reaction mixtures were separated by SDS/PAGE under reducing conditions. The gel was stained with Coomassie Brilliant Blue. (B) The intensity of the band of pro-HGF was quantified with NIH imagej software and the percentage of HGF processed was calculated. (C) Pro-HGF (2 μm) was incubated with TMPRSS13 (54 nm) pretreated with or without HAI-1–NK1 (5 μm). The reaction mixtures were analysed as described in (A).

    We then analysed the effect of HAI-1–NK1 on the pro-HGF converting activity of TMPRSS13. The single-chain pro-HGF (2 μm) was incubated with TMPRSS13 (54 nm) pretreated with or without HAI-1–NK1 (5 μm). The pretreatment of TMPRSS13 with HAI-1–NK1 did not generate the 60 and 32 kDa bands (Fig. 4C), indicating that HAI-1–NK1 inhibits the pro-HGF converting activity of TMPRSS13.

    Biological activities of HGF converted by TMPRSS13

    To examine the biological activities of the HGF converted by TMPRSS13, we used the human hepatocellular carcinoma cell line HepG2. HGF induces a scattering of cell colonies and inhibition of serum-dependent proliferation in HepG2 cells [19]. These biological responses to HGF are transduced through the activation of a high affinity receptor, the c-met proto-oncogene product (c-Met), and also require strong activation of the extracellular signal-regulated kinase (ERK) [20]. Therefore, we first analysed the activation of c-Met by assessing its tyrosine phosphorylation. HepG2 cells were treated with the TMPRSS13-cleaved pro-HGF, and tyrosine phosphorylation of c-Met was analysed by immunoblotting using an anti-phospho-c-Met IgG. The tyrosine phosphorylation was induced in HepG2 cells treated with the TMPRSS13-cleaved pro-HGF at a level comparable with that in cells treated with the purified active HGF, whereas it was not induced in HepG2 cells treated with the uncleaved pro-HGF (Fig. 5A). Treatment of the cells with TMPRSS13 itself did not induce the phosphorylation (Fig. 5A).

    Details are in the caption following the image

    Biological activity of HGF converted by TMPRSS13. Cells were treated with reaction mixtures of pro-HGF alone (Pro-HGF), TMPRSS13 alone (TMPRSS13) or pro-HGF and 2000 ng·mL−1 TMPRSS13 (Pro-HGF + TMPRSS13) at 50 ng·mL−1 pro-HGF. Cells were also treated with purified active HGF at 50 ng·mL−1 (Active HGF). (A) Cells were cultured for 5 min. Lysate of the cells was immunoblotted with the anti-phospho-c-Met IgG (upper panel) and anti-c-Met IgG (lower panel). (B) Cells were cultured for 5 min. Lysate of the cells was immunoblotted with the anti-phospho-ERK1/2 IgG (upper panel) and anti-ERK1/2 IgG (lower panel). (C) Cells were cultured for 4 days. The morphology of the cells was analysed by light microscopy.

    We then analysed the activation of ERK by assessing its phosphorylation. Immunoblotting using an anti-phospho-ERK1/2 IgG showed that the phosphorylation of ERK1/2 was more enhanced in HepG2 cells treated with the TMPRSS13-cleaved pro-HGF than in HepG2 cells treated with the uncleaved pro-HGF or with TMPRSS13 (Fig. 5B). Finally, we analysed the biological response of HepG2 cells by observing their scattering phenotype. Treatment with the TMPRSS13-cleaved pro-HGF induced a scattering of cell colonies, whereas no scattering was observed in the cells treated with the uncleaved pro-HGF or with TMPRSS13 (Fig. 5C). These results indicate that TMPRSS13 converts the inactive pro-HGF into the active two-chain form of HGF.

    Co-expression of TMPRSS13 and HAI-1 mRNA in cultured cell lines

    Because TMPRSS13 and HAI-1 are both transmembrane proteins, HAI-1 is probably co-expressed with TMPRSS13 in the same cells to function as a physiological inhibitor of the protease. We examined the co-expression of TMPRSS13 and HAI-1 mRNA in cultured cell lines by RT-PCR. We analysed five human carcinoma cell lines: a lung carcinoma cell line A549, a colon carcinoma cell line LoVo, stomach carcinoma cell lines MKN45 and MKN74, and HepG2. A549 and LoVo cells have been shown to express TMPRSS13 mRNA [13]. MKN45 cells were used for identification of HAI-1 proteins [4]. MKN74 and HepG2 cells have been shown to respond to HGF [20]. TMPRSS13 mRNA was detected in MKN45 and MKN74 cells, but not in A549, LoVo and HepG2 cells. On the other hand, HAI-1 mRNA was detected in LoVo, MKN45, MKN74 and HepG2 cells (Fig. 6). These results indicate that HAI-1 mRNA is co-expressed with TMPRSS13 mRNA in MKN45 and MKN74 cells.

    Details are in the caption following the image

    RT-PCR analysis of TMPRSS13 and HAI-1 mRNA in human carcinoma cell lines. Total RNA was isolated from cultured A549, LoVo, MKN45, MKN74 and HepG2 cells, and subjected to RT-PCR analysis. The primers for HAI-1 generate two PCR products of HAI-1 and its splice variant (HAI-1B) [35]. GAPDH mRNA was used as an internal control.

    Discussion

    In this study, we tested the inhibitory effect of HAI-1 on the protease activity of several members of the TTSP family using enzymes expressed in E. coli. We found that the protease activity of TMPRSS13 was inhibited by HAI-1, but that of TMPRSS11A, HAT-like 4 and HAT-like 5 was not. TMPRSS11A, HAT-like 4, and HAT-like 5 belong to the HAT/DESC subfamily [2]. Mouse DESC1, also of the HAT/DESC subfamily, forms stable inhibitory complexes with plasminogen activator inhibitor-1 and protein C inhibitor [21]. Thus, these serpins might be endogenous inhibitors of TMPRSS11A, HAT-like 4 and HAT-like 5.

    The protease activity of TMPRSS13 expressed in E. coli was weak, probably because of incorrect protein folding. Thus, we expressed the enzyme in mammalian cells. To obtain an active TMPRSS13 in mammalian cells, we constructed an expression vector encoding a recombinant protein with two modifications, and transfected COS-7 cells with the vector. One modification was that we deleted the N-terminal cytoplasmic and transmembrane domains and tagged the C-terminus with six His sequences, to simply purify the protein from the conditioned medium of the transfected cells by one-step column chromatography. The other modification was that we replaced the putative activation cleavage sequence with the enterokinase recognition sequence, because the molecular mechanism of the proteolytic activation of pro-TMPRSS13 is unknown. The purified pro-enzyme did not show any protease activity, and the enterokinase treatment generated an active enzyme (Fig. 1). Using this active TMPRSS13, we demonstrated that HAI-1–NK1 had inhibitory activity against the protease (Fig. 2). The activity was much stronger than that of aprotinin, which was previously described as an inhibitor of TMPRSS13 [12]. The inhibitory activity of HAI-1–NK1 against TMPRSS13 was confirmed by in vitro binding assays. HAI-1–NK1 formed complexes with the active TMPRSS13 (Fig. 3). HAI-1–NK1 consists of the N-terminal region and the first Kunitz domain, and corresponds to the 40 kDa form of HAI-1 generated from a transmembrane form by extracellular shedding [4]. TMPRSS13 mRNA is expressed in a variety of human adult tissues, and predominantly in lung, placenta, pancreas and prostate [12]. HAI-1 mRNA is also highly expressed in placenta, pancreas and prostate [4]. Thus, the 40 kDa form of HAI-1 could function as an endogenous regulator of TMPRSS13 in these tissues.

    HAI-1–NK1LK2 had a much weaker inhibitory effect against TMPRSS13 than HAI-1–NK1 (Fig. 2). Moreover, no complex of HAI-1–NK1LK2 and TMPRSS13 was detected in the in vitro binding assays (Fig. 3). These results indicate that HAI-1–NK1LK2 only weakly associates with TMPRSS13. HAI-1–NK1LK2 consists of the N-terminal region, the first Kunitz domain, the LDL receptor class A domain, and the second Kunitz domain, and corresponds to the 58 kDa form of HAI-1 identified in the conditioned medium of cultured carcinoma cells [14]. Weaker inhibitory activity of HAI-1–NK1LK2 against HGFA and matriptase was also observed, and an idea that the second Kunitz domain may obstruct the protease-binding site of the first Kunitz domain was proposed [22,23]. The present results indicate that this idea may also apply to TMPRSS13. The weaker inhibitory activity of HAI-1–NK1LK2 was prominent against TMPRSS13, compared with that against HGFA and matriptase. Thus, the presence of the second Kunitz domain may more strongly affect the binding of the first Kunitz domain to TMPRSS13.

    A soluble form of HAI-2, another Kunitz-type inhibitor, also inhibited the protease activity of TMPRSS13, with an IC50 similar to that of HAI-1–NK1 (Fig. 2C). HAI-2 mRNA is highly expressed in various human adult tissues [15], some of which also express TMPRSS13 mRNA, suggesting that HAI-2 could be an endogenous inhibitor of TMPRSS13 in these tissues.

    The inhibition curve of HAI-1–NK1 was not sigmoidal, which is unusual, compared with the sigmoidal curve of HAI-2. Moreover, a high concentration of HAI-1–NK1 was needed for full inhibition of the protease activity of TMPRSS13 (Fig. 2C). These results suggest the characteristic association of HAI-1–NK1 with TMPRSS13, the mechanism of which remains to be investigated. The in vitro binding assays showed that only small portions of HAI-1–NK1 and TMPRSS13 formed complexes (Fig. 3). The weak complex formation may be related to the characteristic association of the protease–inhibitor pair.

    In the present study we have shown that TMPRSS13 converted the single-chain pro-HGF to a two-chain form in vitro (Fig. 4). We proved that the two-chain form of HGF is biologically active, by three assessments. Its treatment of HepG2 cells induced the tyrosine phosphorylation of c-Met, enhanced the phosphorylation of ERK, and induced the scattering phenotype (Fig. 5). Thus, the proteolytic cleavage of pro-HGF by TMPRSS13 generates a biologically active HGF. The concentration for half-maximal activity of TMPRSS13 was 15 nm (Fig. 4B). This value was 0.17 nm for HGFA under similar reaction conditions [24]. Thus, the specific activity of TMPRSS13 is approximately 90-fold lower than that of HGFA.

    TMPRSS13 preferentially recognizes cleavage sites consisting of paired basic amino acid residues (RR or KR at positions P2 and P1). In addition, the presence of a basic amino acid residue (R or K) at position P4 enhances the efficiency of cleavage [13]. The HA protein of an HPAI virus strain with the KKKR motif at the cleavage site was efficiently converted to mature HA subunits in TMPRSS13-transfected cells [13], supporting the preference for the cleavage sequences in substrates of TMPRSS13. Pro-HGF has the KQLR motif at the cleavage site [25]. Thus, the nonbasic amino acid residue at position P2 may cause the low specific activity of TMPRSS13 for the conversion of pro-HGF.

    HGF is a pleiotropic factor that functions as a mitogen, motogen and morphogen for a variety of cells, particularly epithelial cells [25,26]. HGF is thought to play a crucial role in the regeneration of various tissues following injury [27]. HGF is a mesenchymal cell-derived heparin-binding glycoprotein that is secreted as an inactive single-chain precursor. The secreted HGF normally remains inactive, probably associated with the extracellular matrix in the tissues producing it. In response to tissue injury, such as hepatic and renal injury, the inactive single-chain HGF is converted to a two-chain form exclusively in the injured tissue. This conversion is mediated by serine protease activity, which is induced in the injured tissue [28]. The two-chain form is required for the biological activity of HGF [29,30]. Thus, the biological effects of HGF in injured tissue are regulated through proteolytic processing by a serine protease. HGFA is a serum-derived serine protease that efficiently converts the single-chain HGF to the biologically active two-chain form in vitro [31]. The role of HGFA in the proteolytic activation of HGF in vivo was determined by analysing knockout mice. In HGFA-deficient mice, regeneration of the injured intestinal mucosa and the activation of HGF were impaired, but the injured liver was completely regenerated, suggesting that HGFA is responsible for the activation of HGF in the injured intestinal mucosa, but not in other injured tissues [32]. Thus, other serine proteases are probably involved in the activation of HGF in these tissues.

    Several serine proteases have been shown to convert pro-HGF to the active form in vitro. They include serine proteases involved in blood coagulation, such as plasma kallikrein, and coagulation factors XIa and XIIa [24,33]. These serine proteases might be responsible for the activation of HGF in injured tissues. Matriptase and hepsin, members of the TTSP family, also convert pro-HGF to the active form [9,10,15]. Thus, it is possible that these TTSPs function as HGF-converting proteases in injured tissue. A two-step model for the activation of HGF in injured tissues has been proposed. When tissue injury occurs, circulating plasma serine proteases, such as HGFA, are activated in response to the activation of the coagulation cascade and inflammation. The activated proteases convert pro-HGF to the active form (the first step). Subsequently, the activated HGF functions as a mitogen for the epithelial cells. The proliferating epithelial cells produce TTSPs, such as matriptase. The TTSPs convert pro-HGF to the active form (the second step). The activated HGF is involved in further proliferation of the epithelial cells [32]. TMPRSS13 might also function as an HGF-converting protease in the second step, because it appears to be expressed in epithelial cells [13]. The specific activity of the HGF conversion of TMPRSS13 is much lower than that of HGFA as described above. However, TMPRSS13 localizes to the cell surface, and thus could function in the pericellular activation of HGF.

    The pro-HGF converting activity of TMPRSS13 was inhibited by HAI-1–NK1 (Fig. 4C), suggesting that HAI-1 functions as a regulator for the activation of HGF in injured tissues. RT-PCR analysis showed that TMPRSS13 mRNA is co-expressed with HAI-1 mRNA in MKN45 and MKN74 carcinoma cells (Fig. 6). Thus, the pericellular activation of HGF by TMPRSS13 could be regulated by HAI-1 produced in the same cells. Further characterization is required to clarify the roles of TMPRSS13 and HAI-1 in regulating the activation of HGF in vivo.

    Experimental procedures

    DNA constructs

    The cDNA clones for the protease domains with short pro-sequences of TTSPs were obtained from appropriate human cDNA libraries (Takara, Kyoto, Japan) by PCR, and inserted into an E. coli expression vector, pMAL-c2X (New England BioLabs, Ipswich, MA, USA). The putative activation cleavage sequences were replaced with the enterokinase recognition sequence (DDDDK) using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA).

    The cDNA clone for the full-length TMPRSS13 was obtained from a human placenta cDNA library (Takara) by PCR. The PCR product was further amplified by PCR using a primer containing an EcoRI restriction site and a primer containing an XbaI site, which also had a point mutation replacing the stop codon with a Leu codon. The PCR product was subcloned into a mammalian expression vector, p3xFLAG-CMV14 (Sigma, St. Louis, MO, USA). To construct an expression vector encoding pro-TMPRSS13 lacking the cytoplasmic and transmembrane domains, a cDNA sequence encoding amino acid residues 187–567 was amplified by PCR using p3xFLAG-CMV14-TMPRSS13 as a template. The PCR product was subcloned into the EcoRI and PstI sites of a mammalian expression vector, pSecTag2C (Invitrogen, Carlsbad, CA, USA). The activation cleavage site (A321MTGR325) was replaced with the enterokinase recognition sequence as described above.

    To construct an E. coli expression vector encoding pro-TMPRSS13, the cDNA sequence was excised by digestion with HindIII and XbaI from p3xFLAG-CMV14-TMPRSS13, and subcloned into an expression vector, pcDNA3.1/myc-His-A (Invitrogen). The activation cleavage site was replaced with the enterokinase recognition sequence as described above. A cDNA sequence encoding amino acid residues 315–567 with the C-terminally tagged myc-His sequence was amplified by PCR, and subcloned into the EcoRI and PstI sites of an E. coli expression vector, pMAL-c2X.

    To construct an E. coli expression vector encoding HAI-1–K1, the cDNA sequence encoding amino acid residues 241–305 was amplified by PCR using cDNA of HAI-1 [4] as a template. The PCR product was subcloned into the BamHI and XbaI sites of the vector, pcDNA3.1/myc-His-A. The cDNA sequence encoding HAI-1–K1 with the C-terminally tagged myc-His sequence was amplified by PCR, and subcloned into the NdeI and NotI sites of an E. coli expression vector, pET30a (EMD Chemicals, Gibbstown, NJ, USA).

    To construct expression vectors encoding HAI-1–NK1 and HAI-1–NK1LK2, cDNA sequences encoding amino acid residues 1–314 and 1–436 were amplified by PCR using cDNA of HAI-1 [4] as a template. The PCR products were subcloned into the HindIII and XbaI sites of pcDNA3.1/myc-His-A.

    To construct an expression vector encoding HAI-2, the cDNA sequence encoding amino acid residues 1–194 was amplified by PCR using cDNA of HAI-2 [15] as a template. The PCR product was subcloned into the HindIII and XbaI sites of pcDNA3.1/myc-His-A.

    Preparation and activation of pro-TTSPs expressed in E. coli

    Escherichia coli cells were transformed with the expression vectors encoding pro-TTSPs. The cells were lysed by sonication, and the lysate was applied to an amylose resin (New England BioLabs). After the resin was washed with phosphate-buffered saline (PBS), bound proteins were eluted with 1 mm maltose in PBS. The eluted fraction was treated overnight with enterokinase (EMD Chemicals) at 2 units·100 μL−1.

    Cell culture

    COS-7 cells, A549 cells and HepG2 cells were cultured in Dulbecco’s modified Eagle’s medium, CHO cells and Lovo cells were cultured in Ham’s F12 medium, and MKN45 cells and MKN74 cells were cultured in RPMI1640 medium, supplemented with 10% fetal bovine serum, 100 units·mL−1 penicillin and 100 μg·mL−1 streptomycin at 37 °C in a humidified atmosphere containing 5% CO2.

    Preparation and activation of pro-TMPRSS13 expressed in COS-7 cells

    Cells were seeded on eight 100 mm collagen-coated plates (Iwaki, Chiba, Japan) at a density of 1×106 cells·plate−1. The cells were transfected with the expression vector encoding the secreted form of pro-TMPRSS13 at 6 μg·plate−1 using the FuGENE-6 reagent (Roche Diagnostics, Indianapolis, IN, USA). After 24 h, the medium was replaced with serum-free medium, and cells were further cultured for 3 days. The conditioned medium was applied to a nickel nitrilotriacetic acid resin (EMD Chemicals), and the proteins bound to the resin were eluted with nickel nitrilotriacetic acid buffer (EMD Chemicals). The eluted fraction was treated overnight with enterokinase at 2 units·100 μL−1.

    Quantification of TMPRSS13

    The enterokinase-treated pro-TMPRSS13 was quantified by immunoblotting using the protein expressed in E. coli as a standard. The protease domain with its short pro-sequence and the enterokinase recognition sequence of TMPRSS13 fused at the N-terminus to maltose-binding protein and tagged at the C-terminus with myc-His was expressed in E. coli. Preparation of the cell lysate, purification of the proteins, and treatment with enterokinase were carried out as described above for TTSPs expressed in E. coli. The enterokinase-treated pro-TMPRSS13 was separated by SDS/PAGE under reducing conditions, and stained by Coomassie Brilliant Blue. The intensity of the band of the protease domain was quantified using bovine serum albumin (BSA) as a standard.

    The enterokinase-treated pro-TMPRSS13 obtained from COS-7 cells was separated by SDS/PAGE under reducing conditions. In parallel, various amounts of the enterokinase-treated pro-TMPRSS13 obtained from E. coli were separated by SDS/PAGE. After electrophoresis, the samples were subjected to an immunoblot analysis with the anti-c-Myc IgG. The intensity of the band of the protease domain was quantified using the protease domain of the protein obtained from E. coli.

    Enzyme inhibition assay

    HAI-1–K1 was prepared as follows. Escherichia coli cells were transformed with the expression vector encoding HAI-1–K1. The cells were lysed by sonication. The lysate was centrifuged, and the pellet was dissolved in urea (6 m). To refold proteins, glutathione (oxidized form, 5 mm), glutathione (reduced form, 1 mm) and arginine (100 mm) were added to the solution, and the final concentration of urea was adjusted to 0.5 m. The refolded HAI-1–K1 was purified by column chromatography using a nickel nitrilotriacetic acid resin, followed by dialysis against PBS. The enterokinase-treated pro-TTSPs were mixed with HAI-1–K1 (0.67 μm) and incubated in the assay buffer (50 mm Tris/HCl pH 7.5, 150 mm NaCl, and 0.05% Brij 35) for 10 min at 37 °C. Then each substrate was added to the mixture at a final concentration of 100 μm. After incubation for 3 h at 37 °C, the amount of 7-amino-4-methylcoumarin liberated from the substrate was determined fluorimetrically with excitation and emission wavelengths of 355 and 460 nm, respectively, using a fluorometer (1420 ARVOsx; Perkin Elmer Life Science, Boston, MA, USA).

    HAI-1–NK1, HAI-1–NK1LK2 and HAI-2 were prepared as follows. The expression vectors encoding HAI-1–NK1, HAI-1–NK1LK2 and HAI-2 were introduced into CHO cells using Superfect transfection reagent (Qiagen, Hilden, Germany). Transfected cells were cultured at 37 °C overnight. The medium was replaced with fresh medium containing Geneticin (G418). Neomycin-resistant colonies were selected and further cultured in a roller bottle. When the cells became confluent, the medium was replaced with serum-free medium, and the cells were further cultured for 5 days. The proteins were purified from the conditioned medium by column chromatography using nickel nitrilotriacetic acid and anti-c-Myc IgG resins. Aprotinin was obtained from Nakarai Tesque (Kyoto, Japan). The enterokinase-treated pro-TMPRSS13 (100 pm) and a series of concentrations of inhibitors were mixed and incubated in the assay buffer (50 mm Tris/HCl pH 8.0, 150 mm NaCl, and 0.05% Brij 35) for 10 min at 37 °C. Then, Pyr-RTKR-MCA (Peptide Institute, Osaka, Japan) was added to the mixture at a final concentration of 100 μm. The final volume of each mixture was 200 μL. After incubation for 1 h at 37 °C, the amount of 7-amino-4-methylcoumarin liberated from the substrate was determined as described above. The enzymatic activity without inhibitors was used as an uninhibited control. The IC50 was defined as the concentration of inhibitor that inhibited the enzymatic activity by 50% compared with the uninhibited control. The percentage value relative to the uninhibited control was plotted against the log of inhibitor concentrations. The IC50 value was calculated using the graphpad prism software (GraphPad Software, San Diego, CA, USA).

    Binding assay

    HAI-1–NK1 or HAI-1–NK1LK2 was mixed with various concentrations of TMPRSS13 in the assay buffer. The mixture was incubated at 37 °C for 2 h and SDS sample buffer (20 mm Tris/HCl pH 6.8, 0.5% SDS, 5% glycerol and 0.002% bromophenol blue) with 100 mm dithiothreitol was added. Some of the samples were boiled for 5 min. Twenty microlitres of each sample was analysed by immunoblotting.

    HGF-converting activity of TMPRSS13

    The recombinant pro-HGF was prepared as described previously [34]. Pro-HGF (2 μm) was mixed with various concentrations of TMPRSS13 in 20 μL of 20 mm sodium phosphate (pH 7.3) containing 100 mm NaCl and 0.01% Chaps and incubated at 37 °C for 2 h. The reaction mixture was separated by SDS/PAGE under reducing conditions. Proteins in the gel were stained with Coomassie Brilliant Blue. The intensity of the pro-HGF band was quantified by scanning densitometry using NIH imagej software.

    To examine the inhibitory effect of HAI-1–NK1 on the HGF-converting activity of TMPRSS13, TMPRSS13 (54 nm) was incubated with HAI-1–NK1 (5 μm) in 20 mm sodium phosphate (pH 7.3) containing 100 mm NaCl and 0.01% Chaps at 37 °C for 10 min. Then, pro-HGF (2 μm) was added to the mixture. The final volume of the mixture was 20 μL. After incubation at 37 °C for 2 h, the reaction mixture was analysed by SDS/PAGE, as described above.

    Preparation of cell lysate

    HepG2 cells were seeded at 1×106 cells·100 mm·plate−1. They were treated with reaction mixtures of the assay for HGF-converting activity of TMPRSS13 or with purified active HGF (provided by the Research Center of Mitsubishi Chemical Corp., Yokohama, Japan) for 5 min. The cells were washed twice with ice-cold PBS, and lysed with lysis buffer (137 mm NaCl, 8.1 mm Na2HPO4·12H2O, 2.68 mm KCl, 1.47 mm KH2PO4, 1 mm Na3VO4, 5 mm EDTA, 1% Nonidet-P40, 0.5% sodium deoxycholate, 1 μg·mL−1 leupeptine, 1 μg·mL−1 pepstatin A, 1 μg·mL−1 aprotinin and 1 mm phenylmethylsulfonyl fluoride). The cell lysate was cleared by centrifugation, and the protein concentration of the cleared lysate was determined with the BCA protein assay reagent (Thermo Fisher Scientific, Rockford, IL, USA).

    Antibodies and immunoblotting

    Antibodies were obtained as follows: anti-TMPRSS13 IgG (ab59865), which recognizes the catalytic domain of TMPRSS13, from Abcam (Cambridge, MA, USA); anti-human HAI-1 ectodomain IgG from R&D systems (Minneapolis, MN, USA); anti-phospho-c-Met (Try1234/1235) IgG, anti-phospho-p44/42 mitogen-activated protein kinase (ERK1/2) (Thr202/Tyr204) IgG and anti-p44/42 mitogen-activated protein kinase (ERK1/2) IgG from Cell Signaling Technology (Beverly, MA, USA); anti-c-Met IgG (c-28) and horseradish peroxidase-conjugated anti-goat IgG (sc-2020) from Santa Cruz Biotechnology (Santa Cruz, CA, USA); and horseradish peroxidase-conjugated anti-rabbit and anti-mouse IgG from GE Healthcare UK (Buckinghamshire, UK). The anti-c-Myc IgG was prepared as follows. An anti-c-Myc IgG hybridoma cell line (9E10) was purchased from ATCC (Manassas, VA, USA) and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum. The anti-c-Myc IgG was purified from conditioned medium by column chromatography using protein A sepharose (GE Healthcare UK).

    Equal amounts of protein in the cell lysate were separated by SDS/PAGE. The proteins in the gel were transferred electrophoretically to a poly(vinylidene difluoride) membrane (Pall Corporation, Port Washington, NY, USA). For the detection of HAI-1 and ERK1/2, the blotted membrane was treated with BSA blocking buffer (5% BSA, 20 mm Tris/HCl pH 7.4, 100 mm NaCl, 0.05% Tween20 and 0.02% sodium azide). For the detection of other proteins, the membrane was treated with skim milk blocking buffer (5% skim milk, 20 mm Tris/HCl pH 7.4, 150 mm NaCl, 0.05% Tween20 and 0.02% sodium azide). The membranes were incubated with the primary antibody overnight at 4 °C and then with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. Immunoreactive proteins were visualized with an enhanced chemiluminescence western blotting detection system (ECL; GE Healthcare UK).

    Cell scattering assay

    HepG2 cells were seeded at 2.5×105 cells·100 mm·plate−1. They were treated with reaction mixtures of the assay for HGF-converting activity of TMPRSS13 or with purified active HGF, and cultured for 4 days. The morphology of the cells was analysed by light microscopy.

    RT-PCR

    Total RNA was purified from cultured cells with Isogen (Nippon Gene Co., Tokyo, Japan) followed by treatment with RNase-free DNase I (Takara). The total RNA (2 μg) was subjected to a RT reaction (20 μL) using oligo(dT) primers and Superscript II RT (Invitrogen). To remove RNA complementary to the cDNA, the RT reaction mixture was incubated with RNase H (1 μL). The RT reaction product (1 μL) was amplified by PCR using the following gene-specific primer sets: 5′-TCCCATCTGTAGCAGCAACT-3′ and 5′-GGATTTTCTGAATCGCACCT-3′ for TMPRSS13 (34 cycles), and 5′-ATGGAGGCTGCTTGGGCAACA-3′ and 5′-ACAGGCAGCCTCGTCGGAGG-3′ for HAI-1 (26 cycles). The GAPDH-specific primer set, 5′-AGGTGAAGGTCGGAGTCAAC-3′ and 5′-TACTCCTTGGGAGGCCATGTG-3′, was used for control reactions (20 cycles). The PCR products were run on a 1% (for TMPRSS13 and GAPDH) or 2.5% (for HAI-1) agarose gel and stained with ethidium bromide.

    Acknowledgement

    We thank Mrs M. Kamizono for excellent technical assistance.