Anti-invasion and anti-proliferation effects of 3-acetyl-5,8-dichloro-2-((2,4-di- chlorophenyl)amino)quinolin-4(1H)-one (ADQ) via regulation of Akt and Twist1 in liver cancer cells

When primary cancer faces limited oxygen and nutrient supply, it undergoes an epithelial-mesenchymal transition, which increases cancer cell motility and invasiveness. The migratory and invasive cancer cells often exert aggressive cancer development or even cancer metastasis. In this study, we investigated a novel compound, 3-acetyl-5,8-dichloro-2-((2,4-dichlorophenyl)amino)quinolin-4(1H)-one (ADQ), that showed signicant suppression of wound healing and cellular invasion. This compound also inhibited anchorage-independent cell growth, multicellular tumor spheroid survival/invasion, and metalloprotease activities. The anti-proliferative effects of ADQ were mediated by inhibition of the Akt pathway. In addition, ADQ reduced the expression of mesenchymal markers of cancer cells, which was associated with the suppressed expression of Twist1. In conclusion, ADQ successfully suppressed carcinogenic activity by inhibiting the Akt signaling pathway and Twist1, which suggests that ADQ may be an ecient candidate for cancer drug development.


Introduction
Liver cancer is the seventh most commonly diagnosed cancer and the third leading cause of death from cancer worldwide in 2018 1 . The primary cancer cells undergo physiological changes, which are mediated by suppression of E-cadherin and upregulation of N-cadherin along with the massive expression of matrix metalloproteinases (MMP) that can degrade the extracellular matrix (ECM) 2 . The different expression pattern of E-cadherin and N-cadherin is called "cadherin switching," which is a hallmark of molecular alterations associated with cancer development 3 . As a result, primary cancer develops the potential to escape from the primary site to other parts of the body and eventually develops into secondary cancers.
In normal cells, the Akt pathway participates in the regulation of cell proliferation via balancing cell cycle progression and apoptosis 4 . When the Akt pathway is activated, it transduces intracellular signaling that induces cell proliferation 5 . The activities of forkhead box O (FoxO) transcription factors are downregulated via phosphorylation by Akt 6 . The FoxO family exerts tumor-suppressive functions by inducing the expression of cyclin-dependent kinase inhibitors (CDKI), which ensure strict CDK regulation during the cell cycle process 7 . Therefore, the suppression of FoxO transcription factors induced by hyperactivation of Akt leads to the inhibition of CDKI expression, resulting in continuous cell proliferation. Akt is often found highly phosphorylated in most liver cancer cell lines and liver cancer tissues from patients 8 .
Therefore, the proper regulation of the Akt pathway is one of the crucial goals for anti-cancer strategies.
Several transcription factors that function in the development of cancer have been reported in recent years 9,10 . The functions of the Twist family have been extensively studied in physiology and pathology, including organogenesis, cell stemness, senescence, angiogenesis, chemoresistance, and metastasis 11 .
As a transcription factor, Twist1 regulates the expression of E-cadherin and N-cadherin that are associated with the progression of cancer 9 . It has been reported that high expression of Twist1 is associated with aggressive cancers such as breast cancer, gastric cancer, pancreatic cancer, and liver cancer 12 . Therefore, the regulatory mechanism of Twist1 needs to be researched to identify proper therapeutic strategies. To identify and con rm the anti-cancer effect of a chemical, 3-acetyl-5,8-dichloro-2-((2,4-dichlorophenyl)amino)quinolin-4(1H)-one (ADQ), on a liver cancer cell line, various cell-based assays were performed in this study. The scaffold structure of ADQ was previously reported to have activity toward several proteases 13 . In this study, the effects of ADQ suppressing cancer cell migration and invasion were evaluated comprehensively using various liver cancer-derived cell lines, including several HCC such as Huh7, Hep3B, and PLC-PRF-5 and a liver endothelial cancer cell line, SK-Hep1.

Results
ADQ suppressed the cancerous physiologies of liver cancer cells The chemical compound library (~3,300 compounds) was screened through wound healing assays with SK-Hep1 cells. The compound ADQ (Fig. 1A) was identi ed as one of the most potent inhibitors among the 3,300 compounds screened (Fig. S1A). SK-Hep1 cells treated with 10 μM ADQ showed a signi cant reduction in cell viability (Fig. S1B). Therefore, ADQ was applied at a maximum concentration of 5 μM in the following assays. The ability of ADQ to suppress the migration of cancer cells was examined using wound healing assays (Fig. 1B). The migratory ability of the cells was signi cantly decreased by ADQ treatment. The wound closure of the cells treated with the ADQ was only around 50% of the control, which represents signi cant suppression of cell migration. In addition, the inhibitory effect of ADQ on wound closure was similar to that of emodin (20 μM), a well-known tumor suppressor compound for various cancer types 14 . The invasion assays were performed using Matrigel to investigate the effect of ADQ (Fig.   1C). SK-Hep1 cells treated with increasing concentrations of ADQ showed less invasion than the ADQuntreated control. ADQ showed dramatic suppression of cell invasion at a relatively lower concentration than emodin. The invasion of other liver cancer cells, Huh7 and PLC-PRF-5, was also notably inhibited by ADQ (Fig. S1C). Gelatin zymography was performed to evaluate the effect of ADQ on the activities of MMPs (Fig. 1D). The protease activities of MMP-2 and -9 were inhibited by treatment with ADQ. In particular, the MMP-2 activity was almost entirely suppressed by ADQ. The inhibition of MMP activities was also con rmed by MMP enzymatic activity assay (Fig. 1E). Cancer cells that are going through metastatic progression also show anchorage-independent growth 15 . Therefore, a soft-agar colony formation assay was performed to observe the effect of ADQ on the anchorage-independent growth of SK-Hep1 cells (Fig. 1F). During colony formation in the 0.3% soft agar, ADQ showed notable inhibition of colony formation. These data indicate that ADQ inhibits the cancerous physiologies of SK-Hep1 cells, including cell migration, invasion, anchorage-independent growth, and MMP activities.
Invasion of multicellular tumor spheroid (MTS) was inhibited by ADQ MTS growth or invasion assays have been utilized to study cancer cell growth and screen anti-cancer drugs because this three-dimensional (3D) culture system mimics the microenvironment of cancer cells that grow in multi-layers and invade into adjacent tissues [16][17][18] . In our study, an MTS invasion assay was performed to identify the effect of ADQ in conditions similar to the in vivo cancer microenvironment. In comparison to ADQ-untreated MTS, which invaded into the surrounding Matrigel, MTS treated with ADQ showed lesser invasion (Fig. 2). Comparing the yellow outline of MTS observed at the 0 h-time point to the spheroids at each subsequent time point, the MTS invasions were dose-dependently suppressed by ADQ treatment. When the MTS was incubated with 5 μM ADQ, the invasion of MTS was almost completely repressed, which was a more substantial effect than 40 μM emodin treatment. These results suggest that ADQ has the potential to inhibit the invasion of cancer cells in vivo.
ADQ showed anti-cancer effects via the Akt pathway The various intracellular signaling pathways were analyzed to sort out the target pathway of ADQ, and ADQ suppressed the Akt pathway (Fig. 3A). The SK-Hep1 cells were treated with LY294002, a PI3K inhibitor, to inhibit the Akt activation speci cally. The cells treated with LY294002 showed suppression in wound healing, migration, and invasion as similar to the cells treated with ADQ (Fig. S2). In addition, The Akt pathway has been reported to regulate cell survival 4 . The apoptosis-related proteins were evaluated to investigate whether the suppressed cell growth was due to apoptosis. The levels of the pro-apoptotic protein BAX 19 were not signi cantly affected by ADQ treatment (Fig. 3B). The levels of p53 that induce apoptosis 19 were also not altered by ADQ, indicating that the anti-cancer effects of ADQ are not signi cantly correlated with apoptosis.
The FoxO transcription factor is regulated by Akt, and it induces the expression of CDKIs 7 . The promoter activity of the manganese superoxide dismutase (MnSOD) gene (sod2), a FoxO target gene, was evaluated by luciferase reporter assays. The SOD promoter activity was signi cantly up-regulated by 5 μM ADQ, which suggests that FoxO activity is increased in response to ADQ (Fig. 3C). In addition, the mRNA levels of the CDKI genes that are FoxO target genes were evaluated by RT-PCR after treatment with ADQ. The transcriptional levels of CDKN1A, CDKN1B, CDKN2B, and CDKN2D were increased by ADQ (Fig.  3D). The levels of the p21 protein, which is encoded by the CDKN1A gene, were also increased by ADQ (Fig. 3E). The cell cycle distribution was evaluated during treatment with ADQ. Cells treated with ADQ showed an increased population in the G0/G1 phase and a decreased population in the G2/M phase ( Fig.   3F). When cells were treated with 5 μM ADQ, cell populations in both G0/G1 and the G2/M phase showed signi cant changes. Cell proliferation was observed to evaluate the anti-cancer effect of ADQ further. Treatment with ADQ successfully suppressed the growth of SK-Hep1 cells (Fig. 3G). Cell growth was almost completely repressed by 5 μM ADQ. Furthermore, the effect of ADQ on cellular growth was also evaluated in 3D culture using MTS (Fig. S3). The MTS growth was measured without Matrigel to observe only the growth. When ADQ was added, the growth of the MTS was suppressed compared to the control.
Taken together, these data suggest that the Akt pathway is inhibited by ADQ, which results in cell cycle arrest, not apoptotic cell death.
The Twist1 transcription factor was down-regulated by ADQ To further investigate the regulatory mechanisms of ADQ, luciferase reporter assays were performed (Fig.  S4). Various promoter activities were measured via the reporter system, including the urinary-type plasminogen activator (uPA), uPA receptor (uPAR), heparanase-1, and E-cadherin promoters. Notably, the E-cadherin promoter activity was increased (Fig. S4D). As the alteration in E-cadherin expression is a hallmark of cancer development, ADQ was further analyzed for its molecular regulatory mechanisms.
ADQ was applied for the indicated time, and Twist1 levels were observed along with the level of Akt phosphorylation by immunoblot analysis (Fig. 4A). Twist1 and phospho-Akt levels were decreased by ADQ treatment. The decrease in p-Akt and Twist1 levels was also observed in other liver cancer cell lines, such as Hep3B and Huh7 cells (Fig. S5). In addition, a reduction of the Twist1 expression level was observed earlier than the inhibition of Akt phosphorylation (Fig. S6), which suggests the possibility that Twist1 regulates the phosphorylation of Akt. The overexpression or knockdown of Twist1 showed no dramatic effects on either Akt expression or its phosphorylation level ( Fig. 4B and S7A). Furthermore, when Akt wild-type or inactive mutant was overexpressed, the endogenous expression levels of Twist1 were not altered (Fig. 4C). These data indicate that the effects of ADQ on Akt phosphorylation and Twist1 expression should be independent.
The suppressive effects of ADQ on Twist1 expression were investigated at the transcriptional and translational levels. The transcriptional levels of TWIST1 were not altered by ADQ treatment (Fig. 4D). When the SK-Hep1 cells were treated with ADQ, the Twist1 protein expression level was down-regulated as observed previously, but its down-regulation was abolished by treatment with MG132, a potent inhibitor of the proteasome (Fig. 4E). Altogether, these data indicate that the decreased Twist1 by ADQ is mediated by the regulation of Twist1 protein degradation rather than its gene expression. When the Twist1 protein level was reduced in the cells treated with ADQ dose-dependently, E-cadherin expression was induced, and N-cadherin expression was suppressed (Fig. 4F). It was also observed that when the endogenous Twist1 level was decreased with siRNA transfection, cell migration and invasion were inhibited as shown with ADQ treatment (Fig. S7). These data demonstrate that ADQ treatment leads to the inhibition of the Twist1 expression, which results in the alteration of cadherin expressions and the suppression of further progression of cancer (Fig. 4G).

Discussion
Most cancer mortality can be attributed to metastasis that occurs during cancer progression 20 . Thus, targeting developing cancer provides a novel therapeutic approach. In this study, we identi ed a potent cancer inhibitor, ADQ, and evaluated its anti-cancer mechanisms at cellular and molecular levels.
The anti-proliferative effects of ADQ were shown in our study to be mediated by suppression of the Akt pathway. Activation of the Akt pathway generally leads to increased proliferation and suppressed apoptosis or autophagy 21 . The FoxO transcription factor is phosphorylated by Akt and then exported from the nucleus 6 . The suppression of Akt phosphorylation by ADQ led to up-regulated FoxO activity, which was observed in luciferase reporter assays and by the expression levels of CDKIs. Increased expression of p21 encoded by the CDKN1A gene was observed at both the transcriptional and translational levels. It is a crucial CDKI in FoxO-induced G1 cell cycle arrest 7 . Therefore, the alteration of the cell cycle distribution and subsequent suppression of cell proliferation by ADQ was possibly mediated via the Akt pathway regulation. Although ADQ did not show signi cant cytotoxicity up to ~5 μM, the cell proliferation was inhibited, which showed notably different pattern that might be affected by the starting cell density of each assay 22 . Interestingly, this cell cycle arrest induced by Akt inhibition/FoxO activation often leads cells to a quiescent state rather than to cell deaths 23,24 . As FoxO has a potential role in suppressing liver brosis via cell cycle arrest 25 , ADQ may have effects on preventing hepatic brosis.
ADQ rapidly reduced the Twist1 protein level, which raised the possibility that the regulation of Twist1 by ADQ is not dependent on the Akt pathway. The expression of Twist1 did not alter Akt expression and vice versa, suggesting that ADQ regulates the Akt activation and Twist1 stability independently on each other. The protein level of Twist1 was signi cantly reduced by ADQ treatment in various liver cancer cell lines, including SK-Hep1, Huh7, and Hep3B. Since Twist1 is associated with drug resistance in various cancer types [26][27][28][29] , the application of ADQ may have bene t as a novel therapeutic, targeting liver cancer by inhibiting Twist1. Interestingly, a previous study reported an inhibitor that suppressed Twist1 expression level at working concentrations over 10 μM in oncogene-driven lung cancer 30 . Since ADQ showed a su cient reduction of Twist1 at 5 μM in our study, this suggests that ADQ is e cient in suppressing Twist1 levels and can be developed as a drug candidate for treating multiple cancer types.
ADQ showed potential anti-cancer effects through regulation of the Akt pathway and Twist1 protein level in our study. To further evaluate ADQ in the molecular regulation of the metastasis-associated pathways, the regulatory mechanisms of the Akt pathway and Twist1 by ADQ needs to be veri ed in different cancer types, which is our next research goal.

Methods
Chemical screening A chemical compound library (~3,300 compounds) contains 1) compounds extracted from natural products and 2) re-positioning chemical drugs. This library was provided by the Korea Chemical Bank (www.chembank.org). The library was screened through wound healing assays of the human liver cancer cell line, SK-Hep1. SK-Hep1 cells (2 × 10 4 cells/well) were seeded in 96-well plates and incubated overnight. The wound was made by scratching with a pipette tip, and the detached oating cells were removed. Each compound was applied to the cells at 1 μM concentration in 1% FBS media, and the cells were incubated at 37 ℃ for 24 h. The wound area at 0 h and 24 h was observed with a JuLI Stage Real-Time Cell History Recorder. Wound closure was calculated as (X 0 -X 24 ) / X 0 * 100 (%) , where X N indicates the wound area at N-h point. The values were plotted as a heatmap using "ComplexHeatmap" package in R (version 4.0.1). A literature search was followed using the chemical structures of initially screened compounds provided by the Korea Chemical Bank upon request. ADQ was selected for further evaluation because there were no previous reports of the anti-cancer effects of ADQ.

Wound healing assay
The cells were seeded in 6-well plates (1.2 × 10 6 cells/well). The wound was made with a scratcher tip (0.5 mm; SPL Life Sciences, Gyeonggi-do, Korea) 24 h after cell seeding. Then the cells were treated with ADQ in 1% FBS media for 24 h. The wound area was observed with a JuLI Stage Real-Time Cell History Recorder (NanoEnTek Inc., Seoul, Korea). The wound closure is represented as the percent of wound recovery. All experiments were performed in triplicate.

Migration and invasion assay
For invasion assay, transwells (24-well type, pore size 8 µm) were coated with Matrigel (0.5 mg/ml; BD Biosciences, Franklin Lakes, NJ, USA) for 2 h. The cells (2 × 10 5 cells/well) in 1% FBS media were added to the upper chamber along with ADQ, and the lower chamber was lled with 10% FBS media. After incubation at 37 ℃ for 6 h (for migration assay) or 21 h (for invasion assay), the membranes were xed for 10 min in 3.7% paraformaldehyde. The migrated or invaded cells were stained with 0.5% crystal violet for 10 min. The cells on the upper surface of the membrane were removed with a cotton swab. After being dried, ve random elds of the membrane were observed at ×100 magni cation.

Gelatin zymography
Cells were seeded in 12-well plates (2 × 10 5 cells/well) for 16 h. ADQ was added to the cells in serum-free media for 24 h. The collected media were mixed with 5× sample buffer without β-mercaptoethanol. The samples were analyzed by gelatin zymography using 0.1% gelatin-containing 10% acrylamide SDS-PAGE.
MMP activity assay SK-Hep1 cells were seeded in 12-well culture plates. The cultured medium changed to serum free medium containing ADQ with indicated concentration. The medium was collected after 24 h. The collected medium was incubated with an assay system according to the manufacturer's instructions of EnzChek™ Gelatinase/Collagenase Assay Kit (E12055, Thermo Fisher Scienti c, Inc., USA). The uorescence was measured with (microplate reader).
Soft-agar colony formation assay SK-Hep1 cells (4500 cells/well) were mixed with 0.3% agarose (Sigma-Aldrich, St. Louis, MO, USA), and the mixture was seeded on 0.5% agarose in 12-well plates. The media with ADQ was replaced every 3 d. After incubation for 14 d, the colonies were stained with a 0.05% crystal violet solution. The stained samples were photographed by a digital camera (Olympus SP-350; Cam2Com).

MTS invasion assay
SK-Hep1 cells were seeded in low-attachment 96-well plates (2000 cells/well) and incubated for 3 days. After media removal, Matrigel (3 mg/ml; BD Biosciences, Franklin Lakes, NJ, USA) was added. The plate was centrifuged at 300 g for 3 min at 4℃ and then incubated at 37℃ and 5% CO 2 overnight. The spheroids were treated with ADQ in 10% FBS media. All images were captured by a JuLI Stage Real-Time Cell History Recorder (NanoEnTek Inc., Seoul, Korea).

Immunoblot analysis
After the cells were treated with ADQ for 24 h or transfected with the indicated plasmid for 48 h, cell lysates were harvested with lysis buffer [20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.5% Triton X-100, 0.5% IGEPAL® CA-630, 1 mM EDTA, 1% glycerol, 1 mM phenylmethylsulfonyl uoride, 10 mM NaF, and 1 mM Na 3 VO 4 ]. The protein concentrations were determined by a Bradford protein assay (Bio-Rad Laboratories, Inc.) according to the manufacturer's protocol. Equal amounts of protein samples were boiled in 5× SDS sample buffer [12 mM Tris-HCl (pH 6.8), 5% glycerol, 0.4% SDS, 1% β-mercaptoethanol, and 0.02% bromophenol blue] at 100℃ for 5 min. Samples were separated by 12% SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were blocked with 5% nonfat-dried skimmed milk for 1 h at 25℃ and then they were incubated for 16 h at 4℃ with a speci c primary antibody diluted in 3.3% BSA in 1× TBST [Tris-buffered saline with 0.05% Tween 20]. The membranes were washed several times with 1× TBST and then incubated with the appropriate secondary antibody for 2 h at 25℃. Protein bands were detected by enhanced chemiluminescence immunoblotting detection reagent (Pierce; Thermo Fisher Scienti c, Inc.). All immunoblot analysis was performed at least three times. The band intensities of each protein were analyzed with ImageJ (https://imagej.nih.gov/ij/).
MTS growth assay SK-Hep1 cells (2000 cells/well) were seeded on 1% agarose-coated 96-well plates. The cells were incubated for 16 h, and then the formed spheroids were treated with ADQ in 10% FBS media for 96 h. Images were captured with 24-h interval by a JuLI Stage Real-Time Cell History Recorder.

Statistical analysis
All data are expressed as means ± standard error of the mean (SEM). Statistical analysis of the data was performed by one-way ANOVA using Prism 3.0 (GraphPad Software, San Diego, CA, USA). P<0.05, p<0.01, and p<0.001 were considered to be statistically signi cant.