3‐Cl‐AHPC inhibits pro‐HGF maturation by inducing matriptase/HAI‐1 complex formation

Abstract Matriptase is an epithelia‐specific membrane‐anchored serine protease, and its dysregulation is highly related to the progression of a variety of cancers. Hepatocyte growth factor activator inhibitor‐1 (HAI‐1) inhibits matriptase activity through forming complex with activated matriptase. The balance of matriptase activation and matriptase/HAI‐1 complex formation determines the intensity and duration of matriptase activity. 3‐Cl‐AHPC, 4‐[3‐(1‐adamantyl)‐4‐hydroxyphenyl]‐3‐chlorocinnamic acid, is an adamantly substituted retinoid‐related molecule and a ligand of retinoic acid receptor γ (RARγ). 3‐Cl‐AHPC is of strong anti‐cancer effect but with elusive mechanisms. In our current study, we show that 3‐Cl‐AHPC time‐ and dose‐ dependently induces matriptase/HAI‐1 complex formation, leading to the suppression of activated matriptase in cancer cells and tissues. Furthermore, 3‐Cl‐AHPC promotes matriptase shedding but without increasing the activity of shed matriptase. Moreover, 3‐Cl‐AHPC inhibits matriptase‐mediated cleavage of pro‐HGF through matriptase/HAI‐1 complex induction, resulting in the suppression of pro‐HGF‐stimulated signalling and cell scattering. Although 3‐Cl‐AHPC binds to RARγ, its induction of matriptase/HAI‐1 complex is not RARγ dependent. Together, our data demonstrates that 3‐Cl‐AHPC down‐regulates matriptase activity through induction of matriptase/HAI‐1 complex formation in a RARγ‐independent manner, providing a mechanism of 3‐Cl‐AHPC anti‐cancer activity and a new strategy to inhibit abnormal matriptase activity via matriptase/HAI‐1 complex induction using small molecules.

skin cancers, [2][3][4] indicating that up-regulated matriptase proteolytic activity is a common cause of cancer progression. Matriptase autoactivation is greatly increased by an acidic environment, 5 which may explain the high activation of matriptase in solid tumours with a mildly acidic extracellular microenvironment (ECM). Several oncogenic agents and signals including epidermal growth factor (EGF), androgen and ErbB-2 potently activate matriptase in cancer cells and tissues. [6][7][8][9] Therefore, matriptase is a potent oncogenic protein, making it a potential drug target for cancer therapy.
One of the oncogenic actions of matriptase is to activate HGF. [10][11][12] Cell first synthesizes an inactive single chain pro-HGF secreted to ECM. [13][14][15] Pro-HGF is able to bind to but unable to activate its membrane surface receptor c-Met. 16 Activated matriptase catalyzes the cleavage of pro-HGF to produce mature HGF containing α and β chains. [10][11][12] HGF is a pleiotropic growth factor with strong stimulations of cell migration, proliferation, survival and morphogenesis. HGF activation by matriptase and the subsequent induction of c-Met pathway contribute to the progression of cancers. [17][18][19] Thus, a rational strategy to suppress HGF activity in cancers is to inhibit its matriptase-mediated maturation from pro-HGF.
HGF activator inhibitor type-1 (HAI-1) is a cognate partner of activated matriptase. 20 Mice embryogenesis perish as a result of HAI-1 deletions is reversed in matriptase/HAI-1 double-deficient mice, 21,22 whereas ectopically expressed HAI-1 antagonizes the oncogenic properties of matriptase in mice. 23 Thus, HAI-1 is an endogenous inhibitor of matriptase. Upon activation, matriptase is complexed by HAI-1, which largely quenches matriptase proteolytic activity. This is a major approach for cell to avoid excess and uncontrolled activity of matriptase. 1,20 The ratio of matriptase/HAI-1 is generally much higher in cancer cells than in normal cells, [24][25][26] rendering the high activating level of matriptase in cancer cells. Thus, the increase of HAI-I and/or the formation of matriptase/HAI-1 complex should inhibit matriptase activity in cancers.
The high correlation of matriptase activity and cancer progression intrigues the studies of targeting matriptase for cancer treatment. Indeed, some natural and synthetic agents such as curcumin inhibit matriptase activity to exert potent anti-cancer efficacy. 6,[27][28][29] 3-Cl-AHPC, 4-[3-(1-adamantyl)-4-hydroxyphenyl]-3-chlorocinnamic acid, is a selective ligand of retinoic acid receptor γ (RARγ) and has strong anti-tumour effects in both RARγ-dependent and -independent manners. 30 The anti-cancer effect of 3-Cl-AHPC relies on its inhibition of tumour growth and migration, which has been widely established although the underlying mechanisms remain elusive. [31][32][33] Here, we unravelled a novel mechanism of 3-Cl-AHPC anti-cancer effect by inducing matriptase/HAI-1 complex formation to inhibit matriptase proteolytic activity in a RARγ-independent manner. All cell lines were used in less than 6 months of continuous passage after acquisition.

| Western blot analysis
The protein extracts were mixed with protein loading buffer in a
After 10 days, mice were randomly assigned into two groups (6 mice/group): one group receiving 1 mg/kg of 3-Cl-AHPC and the other receiving physiological saline solution by daily intraperitoneal injection. The tumour volume and body weight of each mouse was monitored weekly. After 20 days treatment, mice were sacrificed and individual tumours were taken and weighted, and tumour tissues were used for Western blot analysis and protease activity assay.

| Lentiviral particle preparation and infection
for small hairpin RNA

| Statistical analyses
All experiments were repeated at least three times. Values are given as the mean ± SE. Statistical analyses were performed with Graph-Pad Prism 5.0 (Student's t test or one-way ANOVA analysis) and values with P < 0.05 were considered statistically significant. Unexpectedly, we found that 3-Cl-AHPC, a synthetic ligand of RARγ and a potent cancer inhibitor, strongly induced the formation of matriptase/HAI-1 complex in breast cancer cell MCF-7, skin cancer cell A431 and colon cancer cell SW620 (Figure 1). 3-Cl-AHPC at μM concentrations dramatically promoted matriptase/HAI-1 complex formation in a time-dependent manner ( Figure 1A). When these cells were treated with increasing concentrations of 3-Cl-AHPC, we found 3-Cl-AHPC dose-dependent induction of matriptase/HAI-1 complex ( Figure 1B). The~120 kD band represented matriptase/HAI-1 complex because it was not only recognized by M24 and M19 antibodies but also potently reduced by matriptase siRNA ( Figure 1C). Our co-immunoprecipitation assay also showed that 3-Cl-AHPC significantly induced matriptase/HAI-1 complex formation ( Figure 1D). We then explored whether the effect of 3-Cl-AHPC was because of its induction of matriptase and HAI-1 expressions. D7 is an antibody that recognizes the latent matriptase. Using this antibody, we did not detect significant change of matriptase expression in MCF-7 cells after 3-Cl-AHPC treatment, although the strong induction of matriptase/HAI complex was readily detected by M24, M69 and M19 antibodies (Figure 1E). In addition, our qRT-PCR assay did not show obvious effect of 3-Cl-AHPC on matriptase or HAI-1 mRNA levels ( Figure 1E). Thus, 3-Cl-AHPC was able to induce matriptase/HAI-1 complex formation, which was not because of the altered expression of matriptase or HAI-1.

| 3-Cl-AHPC inhibits rather than activates matriptase
Suramin, DHT and EGF activate matriptase followed by matriptase/ HAI-1 complex induction. 6,8,36 We then investigated whether 3-Cl-AHPC also activated matriptase to induce the complex. We applied matriptase-selective fluorogenic substrate to measure the proteolytic activity of matriptase. 37 To our surprised, matriptase activity in A431 and MCF-7 cell lysates was greatly reduced when cells were treated with 3-Cl-AHPC ( Figure 2A). After activation and complex formation, matriptase shedding from plasma membrane initiates. When we examined cell culture medium, we found that 3-Cl-AHPC also enhanced matriptase shedding from A431 and MCF-7 cells (Figure 2B). However, the proteolytic activity of shed matriptase from 3-Cl-AHPC-treated cells had no increase comparing with vehicle-treated cells in our fluorogenic substrate assay ( Figure 2C) and matrigelenzyme assay ( Figure 2D). In A431 cell-xenografted mouse model, 3-Cl-AHPC dramatically inhibited tumour growth ( Figure 2E). Importantly, 3-Cl-AHPC also inhibited matriptase activity in mouse tumour tissue ( Figure 2F), implying that the anti-matriptase activity of 3-Cl-AHPC contributed to its anti-cancer efficacy. Therefore, 3-Cl-AHPC induced matriptase/HAI-1 complex formation, but it inhibited rather than enhanced matriptase activity. Different from MCF-7, MCF-10A is a non-tumourigenic and nearnormal mammary epithelial cell line. 38 When we compared the expression ratio of matriptase/HAI-1 in these two cell lines, we found that MCF-7 had much higher ratio of matriptase/HAI-1 than MCF-10A ( Figure 3B), which was consistent with the common concept that tumour cells have higher matriptase/HAI-1 ratio. 3,4 The higher ratio of matriptase/HAI-1 also led to the higher proteolytic activity of matriptase in MCF-7 cells ( Figure 3C). We speculated that  Figure 3D). Thus, 3-Cl-AHPC treatment was equivalent to the increase of HAI-1 amount. In A431 cells, 3-Cl-AHPC and EGF induced complex formation, respectively, and their combination induced more complex ( Figure 3E). When we used fluorogenic substrate to measure the activity of matriptase, we found that EGF, as reported previously, induced matriptase proteolytic activity, which was inhibited rather than enhanced by 3-Cl-AHPC ( Figure 3F). Thus, different from EGF activation of matriptase, 3-Cl-AHPC enhanced HAI-1 binding to EGF-activated matriptase followed by blocking matriptase activity.

| 3-Cl-AHPC inhibits pro-HGF-induced cell scattering, migration and invasion
Hepatocyte growth factor is one of the most potent cytokines that promote cell scattering and migration. 39, 40 We further characterized matriptase-mediated 3-Cl-AHPC effect on pro-HGF-induced cell scattering. In normal culture condition, A431 cells normally formed clusters as shown in Figure 5A. When cells were treated with pro-HGF for 24 hours, they became motile and scatter in many directions.
This scattering effect disappeared once matriptase expression was repressed by shRNA, implying that matriptase on the cell surface or in the medium catalyzed the maturation of pro-HGF ( Figure 5A).
When cells were treated with pro-HGF together with 3-Cl-AHPC, pro-HGF-induced cell scattering was largely blocked ( Figure 5A). Similar result was also observed in our cell scratch assay ( Figure 5B). As shown in Figure 5B, pro-HGF strongly promoted A431 cell migration in a matriptase-dependent manner. 3-Cl-APHC potently inhibited pro-HGF-stimulated cell migration, which was also matriptase-dependent because knockdown of matriptase disabled 3-Cl-AHPC (Figure 5B). Moreover, our trans-well experiment also showed that pro-HGF-induced cell migration and invasion was matriptase-dependent  Figure 2B), but it also remains unclear whether the complex induction occurs before or after shedding. Important is that 3-Cl-AHPC-induced shedding does not increase the matriptase activity in the conditioned medium ( Figure 2C and D). Similarly, curcumin potently induces matriptase shedding without increasing shed matriptase activity. 6 The difference is that 3-Cl-AHPC enhances but curcumin inhibits matriptase/HAI-1 complex formation at the plasma membrane, implying different mechanisms of two compounds on inhibiting matriptase activity.
Cell first synthesizes pro-HGF secreting to ECM. Matriptase is involved in the maturation of pro-HGF via cleavage. 48  DHT has been shown to activate AR to promote TMPRSS2 expression, followed by induction of TMPRSS2-mediated matriptase cleavage and activation. 8 3-Cl-AHPC is an analogue of retinoid AHPN and a ligand of RARγ. It binds to RARγ but does not activate RARγ transcriptional activity. 30 It has been shown potent anti-cancer effects in a RARγ-independent manner. 30 Similarly, we find that the antimatriptase activity of 3-Cl-AHPC may not depend on RARγ either ( Figure 6). Thus, 3-Cl-AHPC has other targets in cells to mediate its promotion of matriptase/HAI-1 complex formation. The orphan nuclear receptor small heterodimer partner (SHP) has been identified as a target for 3-Cl-AHPC to exert its anti-cancer effect. 49,50 However, whether 3-Cl-AHPC binding to SHP is responsible for its induction of matriptase/HAI-1 complex remains to be clarified. The robust anti-cancer effect of 3-Cl-AHPC on a variety of cancer types has been widely acknowledged, however, the underlying mechanism is still elusive. Here, we show that inhibiting matriptase also contributes to 3-Cl-AHPC anti-cancer effect, which provides a new direction to optimize 3-Cl-AHPC for cancer treatment.
In all, our research provides a new mechanism underlying 3-Cl-AHPC anti-cancer effect and a new strategy to antagonize matriptase activity through enhancing matriptase/HAI-1 complex formation using small molecules.

ACKNOWLEDG EMENTS
We thank Dr. Chen-Yong Lin at Georgetown University for kindly providing M24, M69 and M19 antibodies, and Dr. Ming-Shyue Lee for kindly providing matriptase shRNA. This work was supported by the grants from the National Nature Science Fund of China (U1405229, 31770811, 31471318, and 31271453).

CONFLI CT OF INTEREST
The authors confirm that there are no conflicts of interest.