Activation of hepatocyte growth factor activator zymogen (pro-HGFA) by human kallikrein 1-related peptidases


H. Kataoka, Section of Oncopathology and Regenerative Biology, Department of Pathology, Faculty of Medicine, University of Miyazaki, 5200 Kihara, Kiyotake, Miyazaki 889-1692, Japan
Fax: +81 985 85 6003
Tel: +81 985 85 2809


Hepatocyte growth factor activator (HGFA) is a serine protease and a potent activator of prohepatocyte growth factor/scatter factor (pro-HGF/SF), a multifunctional growth factor that is critically involved in tissue morphogenesis, regeneration, and tumor progression. HGFA circulates as a zymogen (pro-HGFA) and is activated in response to tissue injury. Although thrombin is considered to be an activator of pro-HGFA, alternative pro-HGFA activation pathways in tumor microenvironments remain to be identified. In this study, we examined the effects of kallikrein 1-related peptidases (KLKs), a family of extracellular serine proteases, on the activation of pro-HGFA. Among the KLKs examined (KLK2, KLK3, KLK4 and KLK5), we identified KLK4 and KLK5 as novel activators of pro-HGFA. Using N-terminal sequencing, the cleavage site was identified as the normal processing site, Arg407–Ile408. The activation of pro-HGFA by KLK5 required a negatively charged substance such as dextran sulfate, whereas KLK4 could process pro-HGFA without dextran sulfate. KLK5 showed more efficient pro-HGFA processing than KLK4, and was expressed in 50% (13/25) of the tumor cell lines examined. HGFA processed by these KLKs efficiently activated pro-HGF/SF, and led to cellular scattering and invasion in vitro. The activities of both KLK4 and KLK5 were strongly inhibited by HGFA inhibitor type 1, an integral membrane Kunitz-type serine protease inhibitor that inhibits HGFA and other pro-HGF/SF-activating proteases. These data suggest that KLK4 and KLK5 mediate HGFA-induced activation of pro-HGF/SF within tumor tissue, which may thereafter trigger a series of events leading to tumor progression via the MET receptor.


human α1-antichymotrypsin




Chinese hamster ovary


glyceraldehyde-3-phosphate dehydrogenase


hepatocyte growth factor activator inhibitor


secreted form of hepatocyte growth factor activator inhibitor type 1 consisting of the first Kunitz domain


hepatocyte growth factor/scatter factor


hepatocyte growth factor activator


human kallikrein 1-related peptidase


Madin–Darby canine kidney


polyclonal antibody


plasminogen activator inhibitor-1


processing concentration 50%



In the pericellular microenvironment of tumor tissues, growth factors and proteases are critically important for tumor progression. These factors enhance tumor cell proliferation, survival, motility, invasion, and angiogenesis. Proteolytic activities are also essential for degrading components of the extracellular matrix or initiating coagulation and fibrinolytic systems. In addition, several growth factors require proteolysis to gain full biological activity. Hepatocyte growth factor/scatter factor (HGF/SF) is a multifunctional growth factor known to play an important role in tumor progression via its specific receptor tyrosine kinase MET, the c-met proto-oncogene product [1,2]. HGF/SF is secreted primarily by stromal cells as an inactive single-chain precursor (pro-HGF/SF) that lacks biological activity and requires proteolytic cleavage to become the active, two-chain mature form [3]. To date, several proteases have been reported to be HGF/SF-converting enzymes [3]. Hepatocyte growth factor activator (HGFA) has been identified as a very potent serum activator of pro-HGF/SF [4,5]. The membrane-anchored, cellular surface serine proteases, matriptase and hepsin, have also been reported as cellular activators of pro-HGF/SF [6–8]. HGFA is primarily synthesized by the liver, and circulates in blood as an inactive zymogen (pro-HGFA) at a concentration of approximately 40 nm [3,9]. Thrombin proteolytically activates pro-HGFA through cleavage of the Arg407–Ile408 bond in response to tissue injury, generating the two-chain active form consisting of a disulfide-linked 66 kDa long chain and 32 kDa light chain [10,11]. The light chain exhibits the enzymatic activity, and the long chain is further cleaved by proteases [10,11]. Activation can also occur in tumor tissue, and we have reported enhanced activity of HGFA and its involvement in the activation of HGF/SF in tumors [12–15]. Although thrombin is considered to be an activator of pro-HGFA [3,10], alternative pathways in the activation of pro-HGFA in the tumor microenvironment remain to be clarified. As the endogenous inhibitor of HGFA, HGFA inhibitor type 1 (HAI-1), an integral membrane Kunitz-type serine proteinase inhibitor, has been identified as a cell surface regulator of HGFA activity [3,16,17]. Protein C inhibitor may also act as a serum inhibitor of HGFA [18].

Human kallikrein 1-related peptidase (KLK) genes consist of 15 homologous serine protease genes located in tandem on chromosome 19q13.4 [19–21]. The KLK proteins are translated as single-chain preproproteases, with cleavage of the signal peptide prior to secretion. An additional cleavage of the propeptide is required for activation of KLKs. After activation, KLK3, KLK7 and KLK9 have chymotrypsin-like activity, whereas other KLKs have trypsin-like activity in proteolysis [20].

Autoactivation of KLK2 and KLK5 has been confirmed, and these enzymes can activate other coexpressed KLKs [22]. Many members of the KLK family are associated with various human tumors, such as prostate, breast, ovary, colon, urothelial and renal cancers [23–28]. For example, KLK3 (prostate-specific antigen) is a well-known tumor marker for prostatic cancer [23], and the expression of KLK5 is associated with unfavorable prognosis and invasiveness in breast, ovary and urothelial cancers [26–28]. However, little is known regarding the underlying biological significance of KLK expression in tumors.

In this study, we examined the possible roles of KLKs (KLK2, KLK3, KLK4, and KLK5) in the HGF/SF signaling axis, focusing on their ability to generate active HGF/SF in the pericellular microenvironment via activation of HGFA. We found that KLK5 is an efficient activator of pro-HGFA, and KLK4 also activates pro-HGFA to a lesser degree.


Expression of KLKs in cancer cell lines

We characterized the expression of KLK2, KLK3, KLK4 and KLK5 mRNAs in a panel of human tumor cell lines. The expression of KLKs such as KLK4 and KLK5 had already been extensively studied in ovarian and breast cancers. For that reason, we examined cell lines of human tumors that were derived from other organs, in which activation of the HGF/SF signaling axis has been reported [1–3,8,12,13]. Thus, we characterized colon (RCM-1, HT29, CaCo-2, HCT116, SW837, DLD-1, LoVo, Colo205), pancreas (S2-007, SUIT-2, AsPC1, Panc1, MiaPaCa), lung (HLC1, LC-2, LC-1, T3M11, LU139), kidney (MRT-1, Caki-1), prostate (LNCap, PC3, DU145), and urinary bladder (KU-1 and UMK-1). In addition, expression of MET, HGF/SF, HGFA and HAI-1 was examined. As shown in Fig. 1, KLK2 was expressed only by the prostatic cancer cell line LNCap. The expression of KLK3 (also known as PSA) was also limited, and was observed in LNCap. Interestingly, DLD-1 derived from colon cancer also expressed low level of KLK3. KLK4 was abundantly expressed by LNCap and MRT-1, and to a lesser degree by DLD-1. On the other hand, low but distinct expression of KLK5 was observed in 13 out of 25 cell lines examined. MET and HAI-1 were detectable in most (22/25 and 20/25, respectively) cell lines. As the primer set for HAI-1 was designed to detect both HAI-1 and its splicing variant with similar inhibitory properties, HAI-1B [29], two PCR products (289 and 337 bp products for HAI-1 and HAI-1B, respectively) were observed. None of the cell lines expressed notable levels of HGF/SF mRNA. Although pro-HGFA is produced mainly by the liver and can be supplied from plasma in vivo [5,9], some tumor cell lines, such as DLD-1, LU139, MRT-1, and LNCap, also expressed endogenous HGFA.

Figure 1.

 RT-PCR analyses for the expression of KLKs and HGF-related molecules in various human tumor cell lines.

Processing of pro-HGFA by KLKs

We directly tested the possibility that KLKs could activate pro-HGFA. Before assaying, the enzymatic activity of each recombinant KLK was confirmed with synthetic chromogenic substrates (Fig. 2A). KLK3 and KLK4 required activation by a processing protease, whereas KLK2 and KLK5 were autoactivated [22]. As reported previously [10], thrombin efficiently processed pro-HGFA in the presence of a negatively charged substance (dextran sulfate, 10 μg·mL−1) (Fig. 2B). Under the same assay conditions, KLK4 and KLK5 also processed pro-HGFA, generating a 32 kDa light-chain band similar to thrombin-treated pro-HGFA under reducing conditions (Fig. 2B). The N-terminal amino acid sequences of the 32 kDa processed bands generated by KLK4 and KLK5 were identified as Ile-Ile-Gly-Gly-Ser. Therefore, thrombin, KLK4 and KLK5 cleaved pro-HGFA at the same sites (Arg407–Ile408) [10]. The processing activity of KLK2 was weak, and generated a 34 kDa product similar to that generated by plasma kallikrein (data not shown). KLK3 did not show HGFA-processing activity (Fig. 2B).

Figure 2.

 Cleavage of pro-HGFA by KLKs. (A) Enzymatic activities of recombinant KLKs measured with chromogenic substrates. (B) Cleavage of pro-HGFA by KLKs. Pro-HGFA (52 nm) was incubated for 12 h at 37 °C in the presence of 10 μg·mL−1 dextran sulfate with 5 nm of one of the following: thrombin, plasma kallikrein (p-kallikrein), KLK3, KLK4, or KLK5. Recombinant KLK3 and KLK4 were preincubated with 0.04 nm thermolysin to convert them to the active forms, and this was followed by addition of 4 mm phosphoramidon. Thermolysin also activated pro-HGFA, and the addition of phosphoramidon completely inhibited the activity of thermolysin. Processing of pro-HGFA was determined by SDS/PAGE under reducing conditions followed by immunoblot analysis using a mAb to HGFA light chain (A-1). Both KLK4 and KLK5 generated 32 kDa fragments of HGFA. The N-terminal amino acid sequences of the 32 kDa fragments were identical (Ile-Ile-Gly-Gly-Ser) for KLK4-mediated and KLK5-mediated processing.

Thrombin, a known activator of pro-HGFA, requires negatively charged substances such as dextran sulfate [10]. Therefore, we checked the effect of dextran sulfate on KLK4 and KLK5 activation of pro-HGFA. In the presence of dextran sulfate, KLK5 processed pro-HGFA more efficiently than did KLK4 (Fig. 3A). The approximate concentrations of thrombin, KLK4 and KLK5 required to activate 50% of 52 nm pro-HGFA in the presence of 10 μg·mL−1 dextran sulfate after 12 h at 37 °C (PC50%) were 0.035 nm, 0.45 nm, and 0.085 nm, respectively. Thus, KLK5 appeared to be much more potent at activating pro-HGFA than KLK4, and its specific activity was about half that of thrombin. In the absence of dextran sulfate, the processing of pro-HGFA by KLK5 and by thrombin was markedly attenuated (Fig. 3B). In contrast, KLK4 activated pro-HGFA even in the absence of dextran sulfate (PC50%: 0.60 nm and 0.45 nm in the absence and presence of dextran sulfate, respectively) (Fig. 3B). As reported previously [10], an alternatively cleaved 80 kDa product of pro-HGFA (cleavage at the Arg88–Ala89 bond) was generated by thrombin in the absence of dextran sulfate. This 80 kDa product was not apparent in the case of KLK5.

Figure 3.

 Dose-dependent and time-dependent processing of pro-HGFA by KLK4 and KLK5 and the effect of dextran sulfate. (A, B) Pro-HGFA (52 nm) was incubated with various concentrations of KLK4 (0.1–10 nm), KLK5 (0.05–5 nm) or thrombin (0.05–5 nm) in the presence (A) or absence (B) of 10 μg·mL−1 dextran sulfate. The mixtures were subjected to SDS/PAGE under reducing conditions, and analyzed by immunoblot analysis. The extent of processing (%) is shown. (C) Time course of pro-HGFA processing in the presence of dextran sulfate. Pro-HGFA (52 nm) was incubated with 5 nm KLK4, 5 nm KLK5 or 5 nm thrombin in the presence of 10 μg·mL−1 dextran sulfate for the indicated period at 37 °C. The mixtures were subjected to SDS/PAGE under reducing conditions, and then subjected to immunoblot analysis.

Further degradation of HGFA light chain was not observed even in the presence of high concentrations of KLK4 or KLK5, confirming that both KLKs are activators of pro-HGFA (Fig. 3A,B). We also generated a time course for pro-HGFA processing by KLK. In the presence of dextran sulfate, KLK5 was a more efficient activator than KLK4 (Fig. 3C). Taken together, these findings show that in the presence of dextran sulfate, KLK5 was five to 10 times more potent than KLK4 in the processing of pro-HGFA.

By using antibodies (A-1, N19, and C20) that recognize different epitopes of HGFA, we further analyzed the cleavage patterns of pro-HGFA by KLKs (Fig. 4). The epitope of each antibody is indicated in Fig. 4B. KLK5 cleaved the activation site in the first step to generate the active light chain, and then cleaved the heavy chain, generating 41–32 kDa fragments (Fig. 4A). Similar, but less efficient, cleavage patterns were also observed in KLK4 (not shown). A schematic representation of each band observed in immunoblot is indicated in Fig. 4A, and that of the cleavage sites is shown in Fig. 4B.

Figure 4.

 Analysis of cleavage sites of pro-HGFA by KLKs. (A) Time-dependent cleavage patterns of pro-HGFA (52 nm) were analyzed by using three kinds of antibody to HGFA (A-1, C20, and N19) under reducing or nonreducing conditions. The epitope of each antibody is indicated in (B). A schematic representation of each band in immunoblot analysis is also indicated (right and lower panels). The results obtained under nonreducing conditions indicate the existence of multiple disulfide bonds in the heavy chain, as reported previously [5,10]. (B) Schematic representation of the cleavage sites of pro-HGFA and epitopes of the antibodies. A-1 recognizes the light chain of active HGFA (hatched bar), but the precise position of the epitope is not known. The intra-heavy-chain disulfide bonds are not indicated.

Inhibition of KLK-mediated pro-HGFA activation by serpins and HAI-1

The pro-HGFA-processing activity of KLK4 was inhibited by plasma serine protease inhibitors such as human α1-antichymotrypsin (ACT), α1-antitrypsin (AT), and α2-antiplasmin (α2-AP), and to a lesser degree by plasminogen activator inhibitor-1 (PAI-1) (Fig. 5). On the other hand, KLK5 was inhibited strongly by PAI-1 and α2-AP, but not by ACT and AT (Fig. 5). Pro-HGFA processing by both KLKs was potently inhibited by HAI-1KD1, a truncated form of recombinant HAI-1 containing the first Kunitz domain, which is the major functional inhibitor domain against cognate proteases [29,30] (Fig. 5). As mature HAI-1 is a membrane-anchored inhibitor expressed on the surface of various epithelial cells and tumor cells [3,12,31,32], the activities of these KLKs could be regulated by HAI-1 within the pericellular microenvironment.

Figure 5.

 Inhibition of KLK-mediated pro-HGFA activation by serpins and HAI-1. KLK4 or KLK5 was preincubated with each protease inhibitor (ACT, PAI-1, AT, α2-AP, or HAI-1KD1) at a 1 : 10 molar ratio, and each mixture was used for pro-HGFA activation assay. The final concentrations of pro-HGFA, KLK4 and KLK5 were 52 nm, 5 nm, and 1 nm, respectively. The mixtures were subjected to SDS/PAGE under reducing conditions, and then subjected to immunoblot analysis.

Degradation of pro-HGF/SF by KLK4 and KLK5

We examined whether KLK4 and KLK5 were able to activate not only pro-HGFA but also pro-HGF/SF. However, as shown in Fig. 6, pro-HGF/SF was degraded by KLK4 and KLK5 in a dose-dependent manner. Although, the physiological cleavage of pro-HGF/SF may occur very inefficiently when the pro-HGF/SF is incubated with a low concentration of KLK5, pro-HGF/SF was almost completely degraded into small fragments when the KLK5/pro-HGF/SF ratio was set higher than 0.1. Similar findings were obtained with KLK4 (Fig. 6). On the other hand, HGFA showed very efficient activation of pro-HGF/SF (Fig. 6) without any degradation, even at high enzyme/substrate ratios (data not shown).

Figure 6.

 Degradation of pro-HGF/SF by KLK4 and KLK5. (A, B) Pro-HGF/SF (53 nm) was incubated at various concentrations of KLK4 (A) or KLK5 (B) at 37 °C for 12 h, and the mixtures were subjected to SDS/PAGE and analyzed by immunoblot. For positive control of processing, the same amount of pro-HGF/SF was incubated with 0.05 nm HGFA at 37 °C for 12 h and simultaneously analyzed. hc, heavy chain of mature active form HGF/SF.

Biological roles of KLK-dependent activation of HGFA

We wanted to investigate whether KLK-mediated activation of pro-HGFA and subsequent processing of pro-HGF/SF induced cellular responses via the MET receptor tyrosine kinase. Thus, we examined the phosphorylation of MET. The addition of pro-HGF/SF, which was preincubated with KLK5-treated pro-HGFA, rapidly activated the cellular MET receptor (Fig. 7A,B). We also examined the effects on cellular scattering. Madin–Darby canine kidney (MDCK) cells were treated with pro-HGF/SF in the presence or absence of KLK5-treated pro-HGFA. Enhanced cellular scattering was observed at 12 h after the treatment when the cells were treated concomitantly with pro-HGF/SF and KLK5-activated HGFA (Fig. 7C). Therefore, HGFA activated by KLK5 appears to be functional. On the other hand, at 24 h after treatment, cells treated with pro-HGF/SF and pro-HGFA also showed cellular scattering. This may be due to processing of pro-HGFA or pro-HGF/SF by endogenous protease of MDCK cells.

Figure 7.

 Biological activity of HGFA processed by KLK. (A) Processing of pro-HGF/SF by KLK5-treated pro-HGFA. Pro-HGF/SF (1660 nm) was incubated with 1 nm pro-HGFA pretreated (6 h at 37 °C) with 0.02 nm KLK5 for 1 h in the presence of 10 μg·mL−1 dextran sulfate in a final volume of 20 μL. Control reaction mixtures lacking pro-HGFA, or KLK5, or both were also prepared. The mixtures were subjected to SDS/PAGE under reducing conditions and analyzed by immunoblotting using a mAb to human HGF/SF. (B) Phosphorylation of MET induced by HGF/SF processed by KLK5-treated pro-HGFA. PC3 cells were cultured in DMEM with 0.1% BSA for 48 h. Then, 1 μL of the mixture described in (A) was added and incubated for 10 min. Cellular proteins were extracted at the indicated time points, and the samples were analyzed by immunoblotting using antibody to human phosphorylated MET (p-MET). The same blot was analyzed with antibody to human MET (total MET). The band densities were measured, and the ratio of p-MET to the corresponding total MET was calculated. (C) MDCK cells were cultured in DMEM and 0.1% BSA for 48 h. Then, 1 μL of the mixture described in (A) was added, and the cells were further cultured for 24 h. The culture morphology was photographed under phase-contrast microscopy at the indicated time points.

Finally, we used a KLK5-negative tumor cell line to determine the effect of engineered expression of KLK5 on pericellular activation of the HGFA–HGF/SF axis. For this purpose, we selected SUIT-2, because this cell line did not express HGFA (Fig. 1) and the expression levels of the HGF/SF activators such as matriptase and hepsin were low (data not shown). As shown in Fig. 8A, cellular KLK5 induced pro-HGF/SF activation via processing of pro-HGFA. It should be noted that, although HGF/SF could be degraded by KLK5 in an in vitro tube assay (Fig. 6B), the cell-based assay revealed that KLK5/HGFA-mediated activation of pro-HGF/SF worked without HGF/SF being degraded. In a migration assay (Fig. 8B), pro-HGF/SF enhanced migration of SUIT-2 cells even in the absence of HGFA; this may be caused by a trace of activated HGF/SF contaminating the pro-HGF/SF preparation (as shown in the immunoblot in Fig. 8A) or by endogenous pro-HGF/SF-activating protease of SUIT-2. However, in the co-presence of pro-HGFA and pro-HGF/SF, the expression of KLK5 significantly upregulated cellular migratory capability as compared with corresponding control cells (Fig. 8B). The data suggested that pericellular pro-HGFA derived from plasma or tumor cells may be activated by tumor cell-derived KLK5, which may thereafter trigger a series of events leading to cellular invasion via HGF/MET signaling.

Figure 8.

 Effects of endogenous KLK5 on the HGFA–HGF/SF activation cascade. (A) Effect of engineered KLK5 expression on HGFA-mediated processing of pro-HGF/SF. SUIT-2 cells were transiently transfected with pCMV6-XL4-KLK5, and maintained with or without pro-HGFA (1 ng/mL) for 3 h; this was followed by the addition of pro-HGF/SF (40 ng·mL−1). After 8 h of cultivation, the processing of pro-HGF/SF was analyzed by immunoblot analysis. Pro-HGF/SF incubated for the same period without any cells is shown as ‘cell free’. hc, heavy chain of mature active form of HGF/SF. (B) Effects of engineered expression of KLK5 on cellular migration of SUIT-2. Values are mean number ± standard deviation of migrated cells per high-power field in triplicate experiments. Representative photographs of migrating cells are also shown. #P < 0.001 as compared to corresponding mock control. *P < 0.001.


In the present study, we showed that KLK5 activates pro-HGFA, resulting in activation of pro-HGF/SF and MET-mediated cellular responses. KLK4 also showed pro-HGFA-processing activity, although the specific activity was lower than that of KLK5. As negatively charged substances, particularly dextran sulfate, stimulated the activation of HGFA by thrombin [10], we also examined the effect of dextran sulfate on KLK-mediated HGFA activation. We found that KLK5 required dextran sulfate to efficiently activate pro-HGFA, whereas KLK4 activated pro-HGFA even in the absence of dextran sulfate. The mechanism by which dextran sulfate enhanced KLK5-mediated pro-HGFA processing remains to be determined. As the pericellular microenvironment is rich in negatively charged substances such as glycosaminoglycans, and both HGFA and HGF/SF also show affinity for negatively charged substances [3], it is reasonable to postulate that, in tumors expressing KLK4 and/or KLK5, efficient HGFA-activating machinery would be generated in the pericellular microenvironment. Pro-HGFA is abundant in plasma [9], and is also expressed by certain human cancers [12,14,33–35], whereas pro-HGF/SF is produced by stromal cells and is significantly increased in tumor tissues via interactions between tumor cells and stromal cells [36]. Therefore, these results reveal a novel mechanism in the control of cellular invasiveness that involves an upstream tumor cell-derived activator and downstream stromal effectors in tumor tissue.

Kallikrein 1-related peptidases are expressed in various tissues, and are implicated in several physiological and pathological conditions. KLK5 is expressed in normal skin as stratum corneum tryptic enzyme and in the prostate [22,37]. In the epidermis, KLK5 activates pro-KLK7 (known as stratum corneum chymotryptic enzyme), and cleaves the components of corneodesmosomes, leading to desquamation in a coordinated manner with KLK7 [38]. In the prostate, KLK5 is secreted into the prostatic fluid. Self-activated KLK5 converts KLK2 and KLK3 into their active forms. After ejaculation, these KLKs degrade semenogelins, the components of seminal clots, to release sperm [22]. KLK4 was first characterized as enamel matrix serine protease 1, and was reported to be involved in enamelogenesis by processing enamelin [39]. KLK4 is also expressed abundantly in the prostate and secreted into the prostatic fluid; however, its physiological function in the prostate is not understood, except for the proteolytic activation of KLK3.

Recently, KLKs have been studied in terms of their diagnostic and prognostic values in some cancers [23]. KLK5 appears to be a potential biomarker of ovarian and breast cancers [23,26,27], and may be involved in the progression of prostate cancer [23–25]. Moreover, overexpression of the KLK5 gene is associated with invasiveness of urinary bladder carcinoma cells [28]. In this study, KLK5 was expressed in more than 50% of the tumor cell lines examined. However, the mechanisms underlying the role of KLK5 in tumor progression are poorly understood. KLK5 and other KLKs activated by KLK5 may degrade extracellular matrix proteins such as fibronectin, laminin, and type IV collagen, which facilitate the cellular invasiveness at the invasion front [37]. Insulin-like growth factor-binding proteins are also possible targets of KLKs [21,40]. Degradation of extracellular insulin-like growth factor-binding proteins would increase the pericellular concentration of insulin-like growth factor, and this could eventually stimulate invasive growth of tumor cells [21,40]. Our present observations offer another means by which KLK5 could contribute to the malignant progression of cancer cells, as the HGF/SF–MET signaling axis is important for the invasive growth of several types of tumors [1,2].

The activation of pro-HGF/SF is catalyzed by serum and cellular proteases [3]. As a serum activator, HGFA is the most potent HGF/SF-converting enzyme [3–5]. Factor XIIa, factor XIa and plasma kallikrein possess converting activity to lesser degrees [41]. On the other hand, cellular surface serine proteases such as matriptase and hepsin also show HGF/SF-converting activity, and serve as cellular activators of pro-HGF/SF [6,7]. Clearly, there are several pathways for the activatation and utilization of this multifunctional growth factor in vivo, depending on the cellular type and tissue microenvironment. In fact, knocking out murine HGFA was not lethal for embryos [42], whereas knocking out HGF/SF was lethal, due to impaired development of the placenta and liver [43,44]. Therefore, another pro-HGF/SF-activating system compensates for the loss of HGFA during tissue development and morphogenesis. Moreover, both matriptase and hepsin knockout mice also failed to show developmental anomalies, similar to HGF/SF knockouts [45,46]. These observations obtained using mutant mice indeed indicate the redundancy in the pro-HGF/SF activation system in vivo.

Although complex and redundant, these processes may be regulated by a single molecule, cell surface HAI-1, as it potently inhibits all of these pro-HGF/SF-converting proteases and is expressed on the surface of epithelial and tumor cells [30,31]. Our results revealed that KLK4 and KLK5 were also sensitive to HAI-1. Therefore, HAI-1 might be a key molecule in the regulation of pericellular HGF/SF bioactivity and also of matrix remodeling and processing of other bioactive molecules mediated by these proteases. In fact, HAI-1 knockout mice cannot survive, due to impaired development of placental tissue, indicating the critical and nonredundant function of this cell surface protease inhibitor [47–49]. Our data also showed that excessive and prolonged exposure to KLK4 and KLK5 leads to degradation of HGF/SF. However, our cell-based assay indicates that this HGF/SF-degrading activity does not affect the pericellular HGF/SF activity. We speculate that, in the pericellular microenvironment, the activity of KLKs might be regulated by HAI-1, and the limited availability of KLKs is not enough to degrade HGF/SF. On the other hand, as the processing activity of HGFA on pro-HGF/SF is very potent, only a trace amount of HGFA generated by KLKs might be enough to generate active HGF/SF, which thereafter triggers a series of events via the MET receptor. Thus, regulation of KLKs, HGFA and other HGF/SF-processing proteases by HAI-1 on the cell surface may be important for optimal availability of HGF/SF. We suggest that complex pericellular interactions between proteases and inhibitors occur in vivo, where they are required for tissue homeostasis. It is likely that they also play important roles in pathological phenomena. In Fig. 9, we summarize our hypothesis regarding the pericellular activation system of HGF/SF in tumors and the possible relationship of KLKs to this system.

Figure 9.

 Hypothetical model for the pericellular activation of HGF/SF in tumors. Tumor cell–stroma interactions result in increased accumulation of extracellular pro-HGF/SF in tumor tissue. There may be two pathways for the activation of pro-HGF/SF in the extracellular space. One is mediated by membrane-bound serine proteinases (cellular activators), such as matriptase and hepsin [6–8]. The second pathway is mediated by HGFA (serum activator), a very efficient activator of pro-HGF/SF [3–5]. Pro-HGFA is circulating in the blood, and is activated in response to tissue injury by thrombin (serum pro-HGFA activator) [10,11]. Certain tumor cells also produce pro-HGFA by themselves [12,14]. Pericellular pro-HGFA may be activated by KLK4 or KLK5 (cellular pro-HGFA activator). HAI-1 might be an important regulator of pericellular pro-HGF/SF activation.

In summary, our results revealed possible roles for KLK4 and KLK5 in the activation of pro-HGFA, which activates the HGF/SF and HGF/SF–MET signaling axis. This finding may shed light on novel functions of these KLKs in pathophysiological conditions, including tumor growth and progression.

Experimental procedures

Cell lines and reagents

Eight human colon cancer cell lines (RCM-1, HT29, CaCo-2, HCT116, SW837, DLD-1, LoVo, Colo205), five human pancreatic cancer cell lines (S2-007, SUIT-2, AsPC1, Panc1, MiaPaCa), five human lung cancer cell lines (HLC1, LC-2, LC-1, T3M11, LU139), two human renal cell carcinoma cell lines (MRT-1, Caki-1), three human prostate cancer cell lines (LNCap, PC3, DU145), the MDCK cell line and the Chinese hamster ovary (CHO) cell line were used in this study. RCM-1, LC-2, LC-1, MRT-1 and UMK-1 were established in our laboratory. S2-007 and SUIT-2 were kindly provided by T. Iwamura (Junwakai Memorial Hospital, Miyazaki, Japan). KU-1 was kindly provided by M. Oya (Keio University, Tokyo, Japan). DLD-1, LoVo, T3M11, LU139 and Caki-1 were obtained from the Riken Cell Bank (Tsukuba, Japan), and HT29, CaCo-2, SW837, Colo205, HCT116, AsPC1, Panc1, MiaPaCa, LNCaP, PC-3 and DU145 were obtained from Dainihon Seiyaku (Osaka, Japan). The cells were maintained in DMEM containing 10% fetal bovine serum at 37 °C in a humidified atmosphere containing 5% CO2.

Preparations of recombinant pro-HGFA, pro-HGF/SF and HAI-1KD1 have been described previously [4,10,30]. Dextran sulfate, Chaps, phosphoramidon and thrombin were purchased from Sigma-Aldrich (St Louis, MO, USA). ACT, AT, α2-AP and PAI-1 were obtained from Calbiochem (San Diego, CA, USA). Chromogenic substrates S-2586 (methoxysuccinyl-Arg-Pro-Tyr-p-nitroaniline.HCl), S-2266 (D-Val-Leu-Arg-p-nitroaniline.2HCl) and S-2302 (D-Pro-Phe-Arg-p-nitroaniline.2HCl) were purchased from Chromogenix (Milan, Italy). Plasma kallikrein and thermolysin were obtained from R&D Systems (Minneapolis, MN, USA).

A mouse mAb to human HGF/SF (P-1), which recognized the α-chain of the two-chain active form of HGF/SF, was used [14]. For detection of HGFA protein, a mAb to human HGFA (A-1), which recognized the light chain of the two-chain active form of HGFA [14], a goat polyclonal antibody (pAb) C-20, which recognized the C-terminal end of the HGFA light chain (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and N-19, which recognized the N-terminal end of the 34 kDa two-chain active form of HGFA (Santa Cruz Biotechnology), were used. A rabbit pAb to human MET (C-12) was obtained from Santa Cruz Biotechnology, and a rabbit pAb to antiphosphorylated MET (pYpYpY1230/1234/1235) was obtained from Biosource (Camarillo, CA, USA).

Preparation and activation of KLKs

Recombinant KLK3, KLK4 and KLK5 were purchased from R&D Systems. For the preparation of pro-KLK2, the cDNA encoding the whole coding region of human pro-KLK2 was amplified by Pyrobest DNA polymerase (Takara, Shiga, Japan), using cDNA generated from normal human prostate (Invitrogen, Carlsbad, CA, USA). The sequences of the primers were 5′-CAGGAATTCAGCATGTGGGACCTG-3′ (forward) and 5′-ATACAGCTGCACTCAGGGGTTGGC-3′ (reverse). The product was subcloned into pCRII (Invitrogen), sequenced, and used as a template for subsequent PCR, using the Gateway recombination system (Invitrogen) according to the manufacturer’s instructions, finally generating pcDNA-DEST40-KLK2 with a histidine tag at the C-terminal end of KLK2. The plasmid was linearized by ScaI and transfected into cultured CHO cells by electroporation (MicroPorator MP-100; NanoEn Tek, Korea). After transfection, colonies resistant to G418 (Sigma, 0.5 mg·mL−1) were selected and screened for the expression of KLK2. For purification of recombinant KLK2, KLK2-transfected CHO cells were cultured in serum-free medium (CHO-S-SFM II; Invitrogen), and recombinant KLK2 was affinity purified with a HisTrap chelating column (GE Healthcare Bio-Science, Uppsala, Sweden), according to the manufacturer’s instructions.

KLK3 and KLK4 were proforms, and required proteolytic activation using thermolysin followed by addition of phosphoramidon to inhibit excess thermolysin activity before use. One microgram of recombinant pro-KLK3 was incubated with 10 ng of thermolysin (molar ratio, 119 : 1) in KLK activation buffer (50 mm Tris/HCl, 150 mm NaCl, 10 mm CaCl2, pH 7.5). The reaction was incubated at 37 °C for 3 min, and terminated by addition of phosphoramidone (4 mm, final concentration). Recombinant pro-KLK4 was also mixed with thermolysin at the same molar ratio as for KLK3 activation, and incubated at 37 °C for 2 h; this was followed by the addition of phosphoramidon. KLK2 and KLK5 were autoactivated during incubation. To confirm the enzymatic activity, activated KLKs were incubated with the chromogenic substrates S-2586 (for KLK3), S-2266 (KLK2, KLK4, and KLK5) or S-2302 (KLK2, KLK4, and KLK5). The chromogenic substrates were diluted in 50 mm Tris/HCl, 0.05% Chaps (pH 7.5) to a final concentration of 1 mm, and this was followed by the addition of 100 ng of each KLK. The final volume of each mixture was 150 μL. The enzymatic activity was monitored by the release of p-nitroaniline (absorbance at 405 nm), using a SpectraMax microplate reader (Applied Biosystems, Foster City, CA, USA), at 37 °C for 50 min.

RNA extraction and RT-PCR analyses

Total cellular RNA was extracted from the cultured cells with Trizol™ reagent (Invitrogen). For RT-PCR, 3 μg of total RNA was reverse transcribed with a mixture of oligo(dT) and random primers, using 200 units of SuperScript reverse transcriptase (Invitrogen), and 1/30 of the resultant cDNA was processed for each PCR with 0.1 μm both reverse and forward primers and 2.5 units of HotStar Taq DNA polymerase (Qiagen, Tokyo, Japan). For KLKs and HAI-1, the following primers were designed: KLK2 forward, 5′-CAGAGCCTGCCAAGATCACAGATG-3′; KLK2 reverse, 5′-CCATTACAGACAAGTGGACCCCCA-3′; KLK3 forward, 5′-ATGACGTGTGTGCGCAAGTTCACCC-3′; KLK3 reverse, 5′-GATCCACTTCCGGTAATGCACCACC-3′; KLK4 forward, 5′-AATCATAAACGGCGAGGACTGCAG-3′; KLK4 reverse, 5′-TTAGCGAGCAAGGGTCTGTTGTAC-3′; KLK5 forward, 5′-TGTGCTCTGATCACAGCCTTGCTT-3′; KLK5 reverse, 5′-CCAGCATTTTAGCATTACTT-3′; HAI-1 forward, 5′-AAGAGTTTCGTTTATGGAGG-3′; HAI-1 reverse, 5′-TGTGCATATCGCAGTCGGATCCAT-3′. For HGFA, MET, HGF/SF and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), the following primer sequences were used, as previously described [13]: HGFA forward, 5′-CCACTTGGATGAGAACGTGA-3′; HGFA reverse, 5′-ATGATGCCGTAGAGGTAAGC-3′; MET forward, 5′-TTGCCAGAGACATGTATGATAAAGAATACT-3′; MET reverse, 5′-TTGTCACTGGGGAAATGGAT-5′; HGF/SF forward, 5′-GGGAAATGAGAAATGCAGCCA-3′; HGF/SF reverse, 5′-AGTTGTATTGGTGGGTGC-3′; GAPDH forward, 5′-GTGAAGGTCGGAGTCAACG-3′; GAPDH reverse, 5′-GGTGAAGACGCCAGTGGACTC-3′. The PCR products were analyzed by 1.5% agarose gel electrophoresis.

Immunoblot analysis

The reaction samples were mixed with SDS/PAGE sample buffer and heated for 15 min at 75 °C. SDS/PAGE was performed under reducing conditions, using 4–12% gradient gels. After electrophoresis, the sample proteins were transferred electrophoretically to Immobilon membranes (Millipore; Billerica, MA, USA). After blocking of the nonspecific binding site with 5% nonfat dry milk in 50 mm Tris/HCl (pH 7.5), 150 mm NaCl, and 0.05% Tween-20, the membranes were incubated with primary antibody in buffer containing 1% BSA at 4 °C overnight; this was followed by four washes with the buffer, and incubation with peroxidase-conjugated secondary antibody diluted in the buffer with 1% BSA for 1 h at room temperature. The labeled proteins were visualized with chemiluminescence reagent (PerkinElmer Life Sciences, Boston, MA, USA).

Activation of pro-HGFA by KLK

Recombinant pro-HGFA (52 nm, final concentration) was incubated with varying concentrations of one of the test proteins (thrombin, plasma kallikrein, activated KLK2, KLK3, KLK4, or KLK5) in 50 mm Tris/HCl and 0.05% Chaps (pH 7.5) in the presence or absence of dextran sulfate (10 μg·mL−1) for 12 h at 37 °C. Final volumes of the reaction mixtures were 20 μL. The reactions were subjected to SDS/PAGE under reducing conditions, and then subjected to immunoblot analysis for HGFA. The extent of the processing was verified by image analysis using photoshop software (Adobe Systems, San Jose, CA, USA). The specific activity of each enzyme was expressed as the enzyme concentration required for processing 50% of 52 nm pro-HGFA and designated as PC50%. To analyze the effects of protease inhibitors on KLK-mediated pro-HGFA activation, each KLK was preincubated with a 10-fold greater concentration of each protease inhibitor (ACT, AT, α2-AP, PAI-1, HAI-1KD1) and used for the processing assay described above.

N-terminal amino acid sequences of cleaved HGFA

Pro-HGFA (357 nm) was incubated with KLK4 (50 nm) or KLK5 (37.5 nm) in 50 mm Tris/HCl and 0.05% Chaps (pH 7.5) in the presence of 10 μg·mL−1 dextran sulfate at 37 °C for 12 h. The total volume of each reaction was 40 μL. After incubation, each sample was subjected to SDS/PAGE. After electrophoresis, the sample proteins were transferred electrophoretically to an Immobilon membrane and stained with 0.1% Coomassie Brilliant Blue in a water/methanol/acetic acid solution (4.5 : 4.5 : 1, v/v). The cleaved HGFA protein band was cut and processed for N-terminal amino acid sequences by automated Edman degradation using a Procise 494 HT Protein Sequencing System (Applied Biosystems).

Activation of pro-HGF/SF

The activation of HGFA by KLK treatment was verified by a pro-HGF/SF activation assay as described previously [10,17]. Briefly, 1.66 μm pro-HGF/SF was incubated for 12 h at 37 °C with 0.001 μm pro-HGFA that had been pretreated with KLK4 or KLK5 in 50 mm Tris/HCl and 0.05% Chaps (pH 7.5) for 2 h at 37 °C. The processing of pro-HGF/SF was determined by immunoblot analysis under reducing conditions, and the extent of processing was semiquantified by photoshop software as described above. To test the direct effect of KLK on the activation of pro-HGF/SF, 53 nm pro-HGF/SF was incubated with various concentrations of KLK4 or KLK5 in the assay buffer in a final volume of 20 μL for 12 h at 37 °C. The reaction was analyzed by immunoblot.

Cell scattering assay and phosphorylation of MET

Eight nanograms of pro-HGFA was incubated with 0.06 ng of KLK5 in 50 mm Tris/HCl and 0.05% Chaps (pH 7.5) for 6 h at 37 °C. Then, 3.2 μg of pro-HGF/SF was added to the mixture and further incubated for 2 h in a final volume of 20 μL. Control reaction mixtures lacking pro-HGFA or KLK5 were also prepared. For cell scattering assays, MDCK cells were cultured in 25 cm2 culture flasks for 48 h in 4 mL of serum-free DMEM containing 0.1% protease-free BSA at 37 °C. During the incubation, the cells were washed with serum-free medium every 24 h. Then, 1 μL of the reaction mixture described above, which should contain 160 ng of activated HGF/SF, was added to the serum-free culture medium (i.e. 40 ng·mL−1 HGF/SF), and the cells were further cultured for 24 h. The morphological change of the cells was photographed under phase-contrast microscopy at each desired time point. For the detection of MET phosphorylation, PC3 cells (cultured under the same conditions as MDCK cells) were treated with 1 μL of the reaction mixture described above, equivalent to 160 ng of activated HGF/SF in 4 mL of serum-free medium, for 10 min at 37 °C. Then, the cells were washed twice with ice-cold NaCl/Pi, and this was followed by the addition of 1.5 mL of 10% trichloroacetic acid on ice. The degenerated cells were scraped and collected into microcentrifuge tubes, and centrifuged at 21 480 g at 4 °C for 3 min. The pellet was dissolved in the extraction solution, consisting of 7 m urea, 2% Triton X-100, and 5% 2-mercaptoethanol. The extracted protein was analyzed by immunoblot analyses.

Engineered expression of KLK5 and cell migration assay

Engineered expression of human KLK5 was performed by transient transfection of the KLK5 expression plasmid, pCMV6-XL4-KLK5 (OriGene, Rockville, MD, USA), using the SUIT-2 human pancreatic cancer cell line, which lacks endogenous KLK5. To test the effect of engineered KLK5 expression on HGFA-mediated processing of pro-HGF/SF, the pCMV6-XL4-KLK5-transfected SUIT-2 cells (SUIT-2KLK5) were maintained in DMEM and 0.1% BSA with or without 1 ng·mL−1 pro-HGFA for 3 h. Then, pro-HGF/SF was added (final concentration in culture 40 ng·mL−1), and cultivation was continued for 8 h. The cellular proteins were extracted, and the processing of pro-HGF/SF was analyzed by immunoblot analysis. The migration capability of SUIT-2KLK5 cells was analyzed using poly(vinyl pyrrolidone)-free polycarbonate filters with a pore size of 8 μm (Chemotaxicells; Kurabo, Osaka, Japan), which were coated with type IV collagen, as described previously [15]. The cells were harvested from culture by incubation with 0.125% trypsin and 0.5 mm EDTA in NaCl/Pi, and this was followed by neutralization with an excess volume of 1% AT in DMEM. After neutralization, the cells were collected by centrifugation at 55 g and rinsed three times in serum-free medium. The cells (2 × 105 cells per 200 μL of serum-free medium, 0.1% BSA, with or without 1 ng·mL−1 pro-HGFA) were then added to the Chemotaxicells and incubated for 3 h. One microliter of pro-HGF/SF solution (final concentration in culture; 40 ng·mL−1) or NaCl/Pi was added and incubated for an additional 21 h at 37 °C in 5% CO2. At that point, the filters were fixed with 3.7% formaldehyde in NaCl/Pi and stained with hematoxylin. The cells on the upper surface of the filter were wiped off with a cotton swab. Migration was quantified by counting the migrant cells on the lower surface in 10 randomly selected high-power fields (200-fold magnification). The cell count was performed blind in triplicate Chemotaxicells. Three independent experiments were performed to confirm the tendency. Statistical analyses were carried out using the statview 5.0 program (Brainpower, Calabass, CA, USA). P-values less than 0.05 were considered statistically significant.


This work was supported in part by Grant-in-Aid for Scientific Research (B) No. 17390116 from the Ministry of Education, Science, Sports and Culture, Japan. We are grateful to Dr S. Uchinokura, Department of Neurosurgery, Faculty of Medicine, University of Miyazaki, for his kind suggestions, and Ms Tobayashi for her skillful technical assistance.