momordin-Ic

Natural triterpenoid saponin Momordin Ic suppresses HepG2 cell invasion via COX-2 inhibition and PPARγ activation

A B S T R A C T
We previously reported that Momordin Ic, a natural triterpenoid saponin from the fruit of Kochia scoparia (L.) Schrad., exerts good anti-invasive activity on liver cancer partly by altering E-cadherin, VCAM-1, ICAM-1 and MMP-9. The JNK and p38-MAPK pathways differentially altered the four molecules to some extent. However, MMP-9, which is greatly suppressed by Momordin Ic, was affected by neither p38-MAPK nor JNK. Therefore, we further investigated how other signals previously found to regulate cell growth, such as COX-2 and PPARγ, function in the process of cell invasion by western blot. The results demonstrated that COX-2 and PPARγ play a significant role in Momordin Ic-inhibited cell invasion. However, COX-2 only regulated E-cadherin and ICAM-1. PPARγ was not involved in VCAM-1alteration but was significant for the expressions of other proteins. Akt, a kinase upstream of COX-2 and PPARγ, did not influence ICAM-1 but directly mediated the expression of E- cadherin, VCAM-1 and MMP-9. Momordin Ic weakens HepG2 cell invasion through PPARγ activation and COX-2 inhibition. These findings provide evidence for the anti-invasion mechanism of Momordin Ic.

1.Introduction
Fructus Kochiae is the fruit of Kochia scoparia (L.) Schrad., a broom cypress fruit that belongs to the goosefoot family and is of edible and pharmaceutical origin. First described in “Shennong’s Herbal Classic”, Fructus Kochiae is attractive for various bioactivities, including in- secticide, anti-inflammation, anti-diabetic, anti-anaphylaxis, anti- pruritic as well as anti-rheumatoid arthritis activities. Momordin Ic (oleanolic acid-3-O-β-D-xylopyranose(1 → 3) β-D-pyranoid glucose) (Fig. 1), an oleanolic acid oligoglycoside isolated from dried Fructus Kochiae, has been identified as the principle effective constituent. Momordin Ic is a natural triterpenoid saponin that has already been reported to enhance antioxidant capacity, inhibit ethanol-induced gastric mucosal lesions, alleviate carbon tetrachloride-induced hepa- totoxicity, accelerate gastrointestinal transit and prevent glucose-in- duced blood sugar increase. Furthermore, we recently discovered that Momordin Ic is also an effective anticancer candidate, which had not previously been shown (Wang et al., 2013; Mi et al., 2016; Wang et al., 2019). Despite the well-documented pharmacological activities of Mo- mordin Ic, reports of its functional mechanisms, especially for antic- ancer activities, are scarce. We reported for the first time that Mo- mordin Ic is a promising anticancer candidate in liver cancer treatment (Wang et al., 2013).

We have been continuing to explore the anticancer mechanism of Momordin Ic and have found that Momordin Ic can antagonize liver cancer progress by stimulating cancer cell apoptosis and weakening invasive capacity (Wang et al., 2013; Mi et al., 2016; Wang et al., 2019). MAPK pathways activated by Momordin Ic work effectively to mediate cell growth. MAPK pathways monitor cell invasion by reg- ulating some essential cell factors, such as E-cadherin (epithelial cad- herin), VCAM-1 (vascular cell adhesion molecule-1), and ICAM-1 (in- tercellular adhesion molecule-1). However, MMP-9 (matrix metalloproteinase-9) is not regulated by MAPK pathways, although its expression is greatly suppressed by Momordin Ic (Wang et al., 2019). It is likely that MMP-9 is involved in other potential signal transductions. Cyclooxygenase-2 (COX-2) represents as a possible signal transduction that we first became interested in since it has been shown to participate in Momordin Ic-induced cell death (Wang et al., 2014). Its role in cell invasion also merits further study. Inducible COX-2 is frequently over-expressed in different tumour types. Accumulating evidence has shown that elevated levels of COX-2 correlate directly with carcinogenic processes and contribute to poor outcomes in multiple malignancies (Xu et al., 2018; Lin and Wu, 2018). Some studies have found that COX-2 can facilitate phenotypic transition of epithelial cells to mesenchymal cells (epithelial-mesenchymal tran- sition, EMT) and tumour-associated macrophages (TAMs), which in turn cause augmented motility and cancer cell invasion (Han et al., 2019).

The proliferative and invasive potential of epithelial cancer cells are much greater than those of normal cells that activate COX-2 sig- nalling (Zhu et al., 2012). These findings make COX-2 a nonnegligible point for clinical cancer treatment. In addition to COX-2, peroxisome proliferator-activated receptor γ (PPARγ) is another interesting mole- cule because it regulates Momordin Ic-stimulated cell death and influ- ences COX-2, as suggested in our previous research (Wang et al., 2014). Thus, it is important to explore the functions of PPARγ in cell invasion. PPARγ is a member of the nuclear receptor superfamily and has a contradictory role in tumourigenicity. Some studies have suggested a tumourigenic role for PPARγ in cancers, such as bladder tumours, renal pelvic tumours, haemangiomas, lipomas, skin fibrosarcomas and he- patic tumours, although the tumourigenic mechanisms of PPARs have not been fully clarified. By contrast, others have reported that activa- tion of PPARγ can suppress the progression of liver, pancreatic, skin, oral, gastric and colorectal tumour cells (Yousefnia et al., 2018; Borland et al., 2018). Moreover, PPARγ is neither tumour suppressive nor on- cogenic in advanced clear cell renal cell carcinoma (Sanchez et al., 2018). Therefore, PPARγ may be recognized as a promising therapeutic target for cancer treatment, but the particular cell type is also im- portant. The specificity of PPARγ in HepG2 cell invasion process is still unknown. Based on the above, in this study we explored the role of PPARγ and COX-2 in cell invasion. Their effects on MMP-9 and adhesive molecules were also analysed according to our previous studies to further eluci- date the complex anti-invasive mechanism of Momordin Ic.

2.Materials and methods
Momordin Ic (98% purity, Chengdu Purechem-Standard Co., Ltd., China). Modified RPMI 1640 medium and foetal bovine serum (FBS) were purchased from Thermo Fisher (Shanghai, China). LY294002, insulin and NS398 were obtained from Beyotime Institute of Biotechnology (China). Rabbit polyclonal antibodies for Akt (9272) and phospho-Akt (p-Akt) (9271) were obtained from Cell Signalling Technology, Inc. Rabbit polyclonal antibodies against PPARγ (sc-7196); goat polyclonal antibodies against COX-2 (sc-1747), α-tubulin (sc- 5286), GAPDH (sc-25,778), E-cadherin (sc-8426), VCAM-1 (sc-13,160),ICAM-1 (sc-8439) and MMP-9(sc-13,520); as well as goat anti-rabbit IgG-horseradish peroxidase (sc-2004), goat anti-mouse IgG-horseradish peroxidase (sc-2005), and GW9662 (22978–25-2) were from Santa Cruz. Human VEGF was obtained from PeproTech (Rocky Hill, NJ USA), and Matrigel was obtained from BD Biosciences. All otherreagents were of analytical reagent grade.The HepG2 cell line (human hepatocyte carcinoma cell) was ob- tained from Collection of Cell Cultures of the Fourth Military Medical University (Shaanxi, Xi’an, China) and incubated in RPMI 1640 medium supplemented with 10% FBS, benzylpenicillin (100 kU/L) and strep- tomycin (100 mg/L) at 37 °C in a humidified incubator (5% CO2). When the cells reached 70% confluence, they were washed three times with phosphate-buffered saline, digested with trypsin, collected after cen- trifugation (5 min, 1000 rpm) and then re-suspended in fresh medium for subculture.After incubation with Momordin Ic or VEGF for 24 h, cells were washed three times with phosphate-buffered saline, digested with trypsin, followed by centrifugation for 5 min (1000 rpm) and finally re- suspension in fresh medium. Cells were seeded at a density of 2 × 104 cells/well in 6-well plates pre-coated with Matrigel followed by in- cubation for 2 h at 37 °C.

Cell morphology changes in the VEGF and Momordin Ic co-treatment group were captured by microscopy (Olympus, TH4–200, Japan) compared to the control and VEGF-treated groups.Cell lysate preparation was performed according to our previous reports (Wang et al., 2019). After treatment, cells were lysed with 100 ml of lysis buffer (with 1% PMSF (100 μg/ml)) on ice for 10 min. After centrifugation at 15,000 g at 4 °C for 15 min, the supernatant was harvested and proteins quantified via a bicinchoninic acid (BCA) pro- tein assay (Thermo Fisher, Shanghai, China). Protein samples were mixed with one fourth of 5 × SDS-PAGE loading buffer (34% Tris-HCl (pH 6.8), 0.15 g/ml SDS, 0.1 mg/ml bromophenol blue, 25% glycerol and 10% β-mercaptoethanol), denatured, and separated by sodium dodecylsulfate–polyacrylamide gel electrophoresis (3% of concentra- tion gel for 10 min under 80 V and 10% of separation gel for 90 min under 120 V. Running buffer: 25 mM Tris, 192 mM glycine, 0.1% SDS). Proteins were transferred onto a polyvinylidene difluoride membrane using a semi-dry electroblotting system (Bio-Rad, USA) and blocked with non-fat milk (5%) in TBST (Tris-HCl, 100 mM; pH 7.4; NaCl, 150 mM containing 0.1% Tween-20) for 2 h. Membranes were then washed and incubated with the primary specific antibodies (1:500, diluted in TBST) at 4 °C overnight followed by HRP-conjugated sec- ondary antibody (1:2000, diluted in TBST) for 3 h. The bands were visualized by enhanced chemiluminescence (ECL, Bio-RAD ChemiDoc XRS) according to the manufacturer’s instructions.Each independent experiment was repeated three times. All the data analysed from three independent experiments are expressed as the means ± standard deviation (SD). Protein bands were quantified by Quantity One 4.6.2 software. Statistical differences were evaluated using ANOVA, and significant differences were determined using Duncan’s multiple-range test (DPS 9.50). Values of p < .05 were considered statistically significant, and p < .01 was considered ex- tremely significant. 3.Results Many growth factors, including EGF, VEGF, and HGF, playimportant roles in tumour development and metastasis. Vascular en- dothelial growth factor (VEGF) functions critically in tumour neo- vascularization as one of the most important angiogenic factors. Recent studies have also found that VEGF promotes angiogenesis and enhances vascular permeability during tumour progression and participates di- rectly in tumourigenesis, adhesion and tumour invasion (Roskoski Jr, 2017; Su et al., 2017; Zhao et al., 2018). VEGF, together with other growth factors, is more likely to bind to receptors on the surface of tumour cells, initiating downstream signalling pathways. Therefore, we applied VEGF to promote HepG2 cell spreading and invasion upon addition of Momordin Ic to better verify the anticancer function of Momordin Ic.The results suggested that VEGF-stimulated cells, compared with the control group (cells cultured in medium only), showed stretched morphologies and reduced contact between cells, indicating a trend towards easier spread. However, Momordin Ic-treated cells showed closer contact between cells, and the stretched-like morphology was less obvious compared to VEGF-stimulated cells (Fig. 2b, c), suggesting an inhibitory effect of Momordin Ic on spreading.Cell spreading together with cell-matrix adhesion generally pro- motes homotypic cells to detach, invading the extracellular matrix. Acell-matrix adhesion assay was already done in our previous report, and it is suggested that Momordin Ic could prevent cell-matrix adhesion (Wang et al., 2019). Apart from adhesion, subsequent cell invasion into surrounding tissues and the vasculature is essential for tumourigenesis. Tumour invasion is highly significant for metastasis as it is the first rate limiting step in metastasis (Perlikos et al., 2013; Carey et al., 2013). A cell invasion was already performed by a transwell culture system in our previous research (Wang et al., 2019). In that study, approximately 3 × 105 cells in 0.5% BSA medium were seeded into the upper chamber of a Matrigel (2.5 mg/ml) pre-coated transwell insert, while medium with 10% FBS was added to the lower chamber. Cells invading the lower surface were calculated by an MTT assay. We found that 10 μM of Momordin Ic could reduce cell invasion by 32% compared to the con- trol. A wounding-healing test and cell spreading assay provided further evidence that Momordin Ic may potentially suppress cell invasion.It has been widely reported that COX-2 represents an important downstream target of various kinases in regulating inflammation, neurological diseases, carcinogenesis, metastatic and other stresses(Tung et al., 2010; Mercau et al., 2014). COX-2 is highly expressed in HepG2 cells but is downregulated by Momordin Ic, as indicated in our previous reports (Wang et al., 2014). In the present research, COX-2 rose again after co-incubation with VEGF (Fig. 3A). We also detected that a COX-2 inhibitor (NS398) downregulated MMP-9 levels, although NS398 and Momordin Ic co-treatment exerted little effect on MMP-9 levels. However, E-cadherin expression, to some extent, unexpectedly decreased by NS398 co-treatment with Momordin Ic compared to Mo- mordin Ic alone. Single NS398 treatment showed little effect on ICAM-1 expression, and co-treatment with Momordin Ic restored the level of ICAM-1 (Fig. 3B). VCAM-1 was not modified regardless of COX-2 in- hibition. Therefore, COX-2, which mainly regulates protein expression of E-cadherin, MMP-9 and ICAM-1, plays an essential role in Momordin Ic-inhibited cell invasion.PPARγ, which is generally involved in insulin sensitization and adipocyte differentiation, has also been revealed to mediate cancer progression, such as breast and gastric adenocarcinoma or lung carci- noma. The results indicated that PPARγ is activated by Momordin Ic but inhibited by VEGF (Fig. 4A). Inhibition of PPARγ by GW9662 could counteract the inductive effect of Momordin Ic on E-cadherin. Similarly,the decreased MMP-9 and ICAM-1 returned to initial levels when GW9662 was used. Furthermore, VCAM-1 expression was maintained at comparable levels regardless of the addition of GW9662 (Fig. 4B). Therefore, PPARγ works effectively in mediating Momordin Ic-inhibited cell invasion mainly by regulating E-cadherin, MMP-9 and ICAM-1.Akt signalling is of great significance in regulating cell growth, differentiation, apoptosis as well as cell adhesion and metastasis. We found that Momordin Ic inhibited Akt phosphorylation, as shown in Fig. 5A. Meanwhile, we applied VEGF to confirm the regulatory role of Akt signalling in Momordin Ic-suppressed cell invasion. VEGF co- treatment with Momordin Ic (10 μM) remarkably restored p-Akt ex- pression, although VEGF alone did not affect p-Akt levels (Fig. 5B).The results in Fig. 6A showed that MMP-9 protein levels further decreased with LY294002 (specific inhibitor for PI3K) alone or Mo- mordin Ic co-treatment. Similar results were observed for VCAM-1. LY294002 upregulated E-cadherin expression and further enhanced thelevels together with Momordin Ic. Exceptionally, there was no apparent difference in ICAM-1 between the LY294002 alone or Momordin Ic co- treatment. Insulin is generally used to activate the PI3K-Akt pathway, and we found that insulin together with Momordin Ic rescued MMP-9 and VCAM-1, while it reversed the level of E-cadherin compared to the Momordin Ic-treated group. However, ICAM-1 expression was hardly influenced, the result of which is consistent with that of the LY294002 group (Fig. 6B). 4.Discussion Distant metastasis is one of the greatest challenges for HCC (hepa- tocellular carcinoma) treatment. Momordin Ic works effectively at blocking liver cancer cell adhesion, migration and invasion. Its anti- invasive property was primarily due to stimulating the homogeneous adhesion factor, E-cadherin. E-cadherin, functional loss of which dis- turbs epithelial cell-cell adhesion and promotes oncogenic migration, is considered a broad-acting tumour suppressor. Presently, E-cadherin activation was also detected when COX-2 was inactivated by Momordin Ic. COX-2 is generally rare in normal tissues but highly expressed in tumours. Documented evidence suggests that COX-2 may be a clinically useful target for future anti-metastasis therapy (Lin and Wu, 2018; Han et al., 2019; Zhang et al., 2018). Tumours that usually exhibit COX-2 are likely to be negative for E-cadherin (Miladi-Abdennadher et al., 2012). For example, in epithelial ovarian cancer, celecoxib inhibits COX-2, which subsequently attenuates E-cadherin suppression (Wang et al., 2018). Interestingly, E-cadherin upregulation is somewhat re- versed when Momordin Ic and a COX-2 inhibitor are used together. There is no reasonable explanation for this unexpected finding, and we hope to explore this phenomenon in the future. Although the COX-2 pathway is complicated, PPARγ regulates E-cadherin more simply. PPARγ activation by Momordin Ic tends to promote E-cadherin ex- pression. Considering the contradictory role of PPARγ on tumour- igenicity, one must specifically differentiate the possible mechanisms in various tumours (Yousefnia et al., 2018). Han (Han et al., 2018) and Cao et al. (Cao et al., 2015) reported that, in hepatocellular carcinoma growth and metastasis, PPARγ usually acts as an antagonist, which is in accordance with the Momordin Ic results. Some reports have suggested a PPARγ-independent pathway for the expression of E-cadherin in human pancreatic cancer cells (Kumei et al., 2009). However, in our present study, GW9662, a PPARγ antagonist, succeeded in blocking the increased expression of E-cadherin, indicating a PPARγ-dependent pathway. We previously reported that the MAPK pathway, which re- presents a general upstream kinase of COX-2 and PPARγ, positively upregulates E-cadherin expression after Momordin Ic treatment (Wang et al., 2019). Similarly, Akt is also progressive in mediating E-cadherin. Active Akt is frequently demonstrated in a large number of tumour tissues. Sustained Akt activation contributes to anchorage-independent growth and EMT (Akca et al., 2011; Singh et al., 2018). Lau et al. de- monstrated that insulin-like growth factor 1 suppresses E-cadherin ex- pression in ovarian cancer cells through activation of PI3K/Akt sig- nalling (Lau and Leung, 2012). Similar results were reported in nasopharyngeal carcinoma cells by Yip et al. (Yip and Seow, 2012). These findings are consistent with ours, which emphasizes the im- portance of AKT blockade for enhancing cell-cell homogeneous adhe- sion. It also indicates that Momordin Ic may stimulate E-cadherin via an AKT/PPARγ and AKT/COX-2-dependent pathway. MMP-9 (matrix metalloproteinases-9), which assists with the invasion process and matrix destruction (He et al., 2018), was greatly suppressed by Momordin Ic in HepG2 cancer cells, and COX-2 strongly upregulated MMP-9 levels. Both COX-2 and MMP-9 expressions are likely to increase or decrease simultaneously in tumour tissues (Kim et al., 2009; Kamaraj et al., 2010; Lu et al., 2015), but the correlation between them is unclear. Our present finding suggests that MMP-9 might be expressed in a COX-2-dependent manner. PPARγ had a clear antagonistic effect on the expression of MMP-9 only when Momordin Ic was added, as shown in Fig. 4. Similarly, PPARγ overexpression could significantly diminish MMP-9 upregulation promoted by SEPT2 in HCC cells (Cao et al., 2015). In HTR-8/SVneo cells, suppression of the PPARγ pathway rescued the MEHP-inhibited MMP-9 expression, which is ac- companied by the recovery of invasion (Gao et al., 2017). PPARγ agonists could also decrease MMP-9 mRNA expression and release in macrophages (Xue et al., 2010). Thus, Momordin Ic may inhibit MMP-9 expression through the COX-2 and PPARγ-mediated pathways, and the two pathways may be partly overlapped and even interrelated (Ravi Kiran Ammu et al., 2019). Interestingly, MMP-9 was affected by the AKT pathway, although it seems to not be related to MAPK pathways, as shown in our previous report (Wang et al., 2019). Li et al. reported that AKT phosphorylation promotes tumourigenesis and metastasis in colon cancer cells via MMP-9 upregulation (Li et al., 2019b, 2019a). Sup- pressed MMP-9 was also detected with blockade of the AKT signalling pathway in angiogenesis of malignant melanoma and vascular smooth muscle cell migration (Ma et al., 2015; Li et al., 2019b, 2019a). Fur- thermore, MMP-9 secreted by TAMs (tumour associated macrophages) that promoted distant gastric cancer metastasis was also associated with the AKT pathway (Lin et al., 2019). These results emphasize that Momordin Ic inhibits cell invasion through AKT/COX-2 and AKT/ PPARγ-mediated MMP-9 inactivation, thus maintaining extracellular matrix integrity. The anti-invasive properties of Momordin Ic are also associated with suppressing heterogeneous adhesion factors, such as ICAM-1 (inter- cellular adhesion molecules-1) and VCAM-1 (vascular cell adhesion molecules-1). ICAM-1 is widely expressed in various cell surfaces (such as lymphocytes, endothelial cells, and cancer stem cells), while VCAM-1 is usually detected in activated endothelial cells, smooth muscle cells, macrophages, fibroblasts as well as variety of cancer cells rather than normal tissues and cells. ICAM-1 and VCAM-1 on the surface of en- dothelial cells could be specifically recognized and bound by cancer cells, accelerating cell attachment to the extracellular matrix. ICAM-1 and VCAM-1 on the surface of tumour cells could also combine with receptor molecules embedded in neutrophils, which subsequently ad- here to endothelial cells. Additionally, tumour cells readily bind to lymphocytes or peripheral blood lymphocytes via ICAM-1 and VCAM-1 to escape immune killing and migrate within lymphatic circulation (Ding et al., 2003; Liu et al., 2013). Specifically, ICAM-1 but not VCAM- 1 is regulated by COX-2 and PPARγ during Momordin Ic treatment. There is some evidence of synchronization between ICAM-1 and COX-2 (Uehara et al., 2016), but no studies have unveiled this relationship. Our findings may imply a potential relationship between them. Al- though PPARγ activation has been reported to inhibit VCAM-1 other than ICAM-1 in endothelial cells (Mun et al., 2011), PPARγ in response to Momordin Ic appears more important for ICAM-1, but dispensable for VCAM-1, in HepG2 cells. There is some evidence to suggest a close relationship between adhesion molecules and kinases (Hou et al., 2014; Sung et al., 2018), and here we found that VCAM-1 rather than ICAM-1 was mediated by the AKT pathway, differing from the previous results in which both ICAM-1 and VCAM-1 were strongly downregulated by p38 in response to Momordin Ic (Wang et al., 2019). COX-2 and PPARγ are generally considered to be downstream of the AKT pathway, and it seems reasonable that ICAM-1 might be indirectly altered via an AKT/ COX-2 or AKT/PPARγ-mediated pathway. However, the result that ICAM-1 was maintained at consistent levels with AKT alteration was not in line with our hypothesis. There may be other undiscovered pathways that are interrupted or initiated when the AKT pathway is disturbed, and these possible pathways might coordinate or counteract with COX-2 (PPARγ), thus leading ICAM-1 to perform the initial level. Another inexplicable phenomenon was that ICAM-1 levels recovered when cells were co-incubated with Momordin Ic and a COX-2 inhibitor compared to Momordin Ic treatment alone. This inspired us to pursue more in-depth and comprehensive mechanisms in the future. To summarize, E-cadherin acts sensitively to enhance cell-cell ad- hesion in response to Momordin Ic, and it is also widely mediated by various signals. ICAM-1 expression may involve a complicated process, as AKT does not directly influence its expression. VCAM-1 is affected by AKT rather than COX-2 (PPARγ). MMP-9, which is hardly altered by MAPK pathways, is clearly a target of AKT, COX-2 and the PPARγ pathway. Momordin Ic weakens HepG2 cell invasion through a PPARγ- dependent momordin-Ic manner, and COX-2 inhibition may have beneficial effects in the treatment of HCC, as proposed in Fig. 7A (Fig. 7B illustrates the roles of MMP-9, ICAM-1, VCAM-1 and E-cadherin in cancer metastasis according to previous reports (Ding et al., 2003; Carey et al., 2013; Liu et al., 2013; Wang et al., 2019), which is a simple explanation to better understand Fig. 7A). Momordin Ic might also represent a promising inhibitor for COX-2 and an activator of PPARγ.