New Self-assembled Supramolecular Bowls as Potent Anticancer Agents for Human Hepatocellular Carcinoma

We report herein on the design, synthesis and biological activity of Ru-based self-assembled supramolecular bowls as a potent anticancer therapeutic in human hepatocellular cancer. The potent complex induces production of reactive oxygen species (ROS) by higher fatty acid β-oxidation and down-regulation of glucose transporter-mediated pyruvate dehydrogenase kinase 1 via reduced hypoxia-inducible factor 1α. Also, overexpressed acetyl-CoA activates the tricarboxylic acid cycle and the electron transport system and induces hypergeneration of ROS. Finally, ROS overexpressed through this pathway leads to apoptosis. Furthermore, we demonstrate that the naphthalene derived molecular bowl activates classical apoptosis via crosstalk between the extrinsic and intrinsic signal pathway. Our work into the mechanism of Ru-based self-assembled supramolecular bowls can provide valuable insight into the potential for use as a promising anticancer agent.

Synthesis and characterization of molecular bowls 6-9. The new dipyridyl benzamide ligand 1 was synthesized by Sonagashira coupling of 3,5-dibromobenzamide and fully characterized by 1 H 13 ,C NMR and high-resolution mass spectroscopy. The pure dipyridyl benzamide ligand 1 with proper dinuclear Ru(II) acceptors 2-5 was used for self-assembly of molecular bowls 6-9 in CH 3 NO 2 /CH 3 OH (1/1) cosolvent (Fig. 1). The reaction mixtures were stirred at room temperature for 6 h to obtain clear solutions. Molecular bowls 6-9 were purified by precipitation and filtration and then fully characterized by 1 H and 13 C NMR, ESI-MS, and elemental analysis. The solid-state structure of complex 6 was determined by single-crystal X-ray analysis. The 1 H NMR peaks associated with building blocks were shifted after self-assembly reaction in CD 3 OD/CD 3 NO 2 (1/1) cosolvent. This result indicated the formation of molecular bowls 6-9 via metal-ligand coordination bonding (Figures S3-S10). The α-pyridyl protons of ligand 1 shifted upfield by 0.2-0.6 ppm on complexation with acceptors 2-5. The formation of molecular bowls 6-9 was further supported by ESI-MS analysis. The isotopic distribution peaks for molecular bowls 6, 7, 8 and 9 observed at m/z 638.09, 671.45, 704.80 and 771.51, respectively, corresponded to [M-3OTf] 3+ ( Figure S11). These peaks matched well with the theoretical distributions. Single-crystal X-ray diffraction (SCXRD) analysis of 6 unambiguously confirmed its molecular structure ( Fig. 1 and S12). Slow vapor diffusion of diethyl ether into the nitromethane/methanol solution of 6 yielded light yellow single crystals suitable for SCXRD analysis. The structure of macrocycle 6 was refined in the triclinic Pspace group and showed a bowl-shaped geometry. Interestingly, the benzamide moieties interacted with displaced π−π stacking at a distance of 3.671 1 Å; thus, the stacked semicircles resulting in a bowl-shaped architecture.  Cytotoxicity of molecular bowl 8 against human cancer cell lines. We first investigated the cytotoxic potential of MB8 on a variety of human cancer cell lines by determining their IC 50 . The inhibition effect was assessed in AGS, A549, HCT-15, SK-HEP-1 and HepG2 cells, respectively. MB8 showed the strongest anticancer effect in all of the tested cell lines (Table 1). In particular, MB8 had the greatest inhibitory effect in another human HCC line, HepG2. Based on these results, we selected the HepG2 hepatic cell line for further analysis of the anticancer potential of MB8.

Molecular bowl 8 and β-oxidation by activating acyl-CoA dehydrogenases.
To identify genes that are actively expressed by treatment with MB8, we compared mRNA expression using ACP-based GeneFishing PCR technology. When we treated MB8, 2 bands were increased, and 1 band was decreased in comparison to the control by ACP9 primer (Fig. 2a). As shown in Fig. 2a, the bands of treated MB8 increased compared to the control (lane 1). We next confirmed the DNA sequence of actively expressed genes via comparison with GenBank (NIH, MD, USA) and then estimated very-long-chain acyl-CoA dehydrogenase (VLCAD), which was related to β-oxidation. There are four distinct acyl-CoA dehydrogenases: short-chain acyl-CoA dehydrogenase (SCAD), medium-chain acyl-CoA dehydrogenase (MCAD), long-chain acyl-CoA dehydrogenase (LCAD) and very-long-chain acyl-CoA dehydrogenase (VLCAD). We performed qRT-PCR to further confirm the β-oxidation-related genes and to determine if VLCAD was involved. Contrary to the expectation, MB8 significantly increased expression of SCAD, MCAD and LCAD mRNA in comparison with the untreated cells while the expression of VLCAD mRNA did not show a significant change. In particular, MB8 at concentrations of 1, 2 and 4 µM produced significant up-regulation of MCAD mRNA expression by 1.48, 1.87 and 3.85 times, respectively, compared to the untreated cells (Fig. 2b). Furthermore, MB8 at a concentration of 4 µM significantly up-regulated the expression of MCAD protein by 2.52-fold compared to untreated cells (Fig. 2c,d). The protein level of MCAD and LCAD was significantly decreased under hypoxia in the cancer cell lines including HepG2, Hep3B and SK-HEP-1, and HIF-1 inhibited MCAD and LCAD, resulting in decreased ROS levels and enhanced tumor proliferation 38 . In this regard, our data indicate that MB8 induces fatty acid oxidation via the up-regulation of acyl-CoA dehydrogenase, especially MCAD, and then affects HepG2 cancer cell survival.
Molecular bowl 8 and suppression of HIF-1α and PDHK-1. Next, we investigated the level of HIF-1α protein in HepG2 cells treated with MB8, which induced acetyl-CoA by acyl-CoA dehydrogenase-mediated β-oxidation or PDHK-1 (Fig. 3a). As shown in Fig. 3b, MB8 significantly decreased the level of HIF-1α protein in a dose-dependent manner by 0.95-, 0.79-, and 0.002-fold compared with the untreated cells. HIF-1α suppressed the MCAD and LCAD, followed by tumor progression via fatty acid oxidation and PTEN 38 .Contrary to our expectation, PTEN did not show significant changes from MB8 treatment at any of the concentrations tested when compared to the untreated cells (Fig. 3c). Accordingly, we assumed that MB8 should act on cancer cell death via a different route. The HIF-1α is stabilized and translocates to the nucleus under hypoxic conditions, where it dimerizes with HIF-1β and activates the expression of a variety of target genes including pyruvate dehydrogenase kinase-1 (PDHK-1), also known as PDK-1, and glucose transporters during tumorigenesis and cancer progression 45,46 . PDHK-1 is a key rate-limiting enzyme for pyruvate conversion to acetyl-coA, after which it enters into the TCA cycle. Under hypoxic conditions, the conversion of pyruvate to acetyl-coA is decreased by PDHK-1-mediated inhibition of PDH, resulting in the reduced flow of glucose-derived pyruvate into the TCA cycle 34,47 . The expression of PDHK-1 protein is significantly decreased by 0.49-, 0.68-, and 0.0007-fold after treatment with 1, 2, and 4 µM MB8 in comparison to untreated cells, respectively (Fig. 3d). Our data suggest that MB8 may play a role in inducing HepG2 cancer cell death by up-regulating the conversion of pyruvate to acetyl-CoA via the reduced HIF-1α expression-mediated suppression of PDHK-1 expression.

Molecular bowl 8 and glucose transporter.
Next, we used qRT-PCR to analyze the expression of glucose transporter-1 (GLUT-1) and glucose transporter type-4 (GLUT-4) mRNA in MB8-treated HepG2 cancer cells. The glucose transporter is a key rate-limiting factor in the transport and metabolism of glucose in cancer cells, which promotes higher glucose uptake to support increased cellular respiration. GLUT-1 is up-regulated in various malignant tumors, which depend on glycolysis for ATP generation 48 . GLUT-4 is up-regulated when the level of OXPHOS complex and the respiration rate increase [49][50][51] . None of the MB8 concentrations showed a significant change in GLUT-1 mRNA expression compared to the untreated cells (Fig. 3e). However, the expression of GLUT-4 mRNA significantly increased by 1.8-fold when cells were treated with 4 µM MB8, in comparison to untreated cells (Fig. 3f). These findings suggest that MB8 treatment predominantly involves acetyl-CoA and electron transport chain (ETC), followed by higher ROS production in HepG2 cancer cells.

Molecular bowl 8 and ROS production by activated cytochrome c oxidase. To examine whether
MB8 regulates ETC-mediated ROS production, we analyzed the mRNA expression of cytochrome c oxidase subunit 1-8 and cellular ROS production. Expression of mRNA from the cytochrome c oxidase (COX) subunits (COX-1, 2, 3 and 6) significantly increased after treatment with MB8 at all concentrations (1, 2 and 4 µM) in comparison to the untreated cells (Fig. 4). In particular, the mRNA expression of COX-2 showed dose-dependent increases of 1.6-, 1.8-and 2.5-fold for 1, 2 and 4 µM of MB8, respectively. In succession, MB8 significantly increased ROS levels in a concentration-dependent manner to 4.2, 8.8 and 15.1 at concentrations of 1, 2 and 4 µM compared to the untreated cells (Fig. 5). COX, the terminal enzyme in the respiratory electron transport chain of mitochondria, is a large integral membrane protein and consists of 3 mtDNA-encoded and 10 genomic DNA-encoded subunits 48 . COX activity regulates the overall rates of mitochondrial respiration and electron transport, resulting in the production of superoxide (O 2 − ) as a by-product and a pro-oxidant to promote ROS-mediated signaling pathways [52][53][54] . Our data provide evidence that MB8 induces cancer cell death by increasing the ETC, followed by increasing the ROS production.

Molecular bowl 8 and classical apoptosis.
We determined whether the ROS-mediated cell death by MB8 was apoptosis or necrosis. We confirmed that the number of HepG2 cells was decreased due to the cytotoxicity of MB8 and more apoptotic cells were detected. Via quantification of the apoptotic cells and necrotic cells, we found the apoptotic cell population increased to 5%, 16%, and 50% at concentrations of 1, 2 and 4 µM MB8, respectively (Fig. 6a). These results show that MB8 mediated HepG2 cancer cell death through apoptosis rather than necrosis. Next, we investigated the signal transduction pathway underlying the apoptosis-mediated cell death induced by MB8 and found that it activated caspase-8 cleavage, and the ratio of cleaved caspase-8 and caspase-8 was significantly enhanced by 2.5-, 2.6-, and 2.5-fold at concentrations of 1, 2, and 4 µM of MB8 compared with the untreated cells (Fig. 6c). In succession, the expression of Bid pre-form in the MB8-treated HepG2 cells significantly decreased in comparison with the untreated cells, while the expression of cleaved Bid was significantly induced. The ratio of cleaved Bid and Bid increased by 13.8-fold at a concentration of 4 µM compared to the untreated cells (Fig. 6d). The ratio of cleaved caspase-9 and caspase-9 also showed a significant increase of 47.2-fold at a concentration of 4 µM MB8 compared to the untreated cells (Fig. 6e). Sequentially, MB8 significantly induced the activation of caspase-3 and PARP by 42.5-fold and 4,581-fold, respectively, at a concentration of 4 µM in comparison with the untreated cells (Fig. 6f,g). As shown in Fig. 6h, there was no significant change in the expression of p53 protein after treatment with MB8. We also confirmed that MB8 significantly increased cytochrome c release by 4.3-fold at a concentration of 4 µM in comparison with the untreated cells, and continued for 48 h (Fig. 6i). Based on these results, MB8 induced HepG2 cancer cell death by activating both intrinsic and extrinsic apoptosis, followed by crosstalk between the intrinsic and extrinsic pathway through caspase-8 mediated cleavage of the Bcl-2 family member Bid, suggesting that the potential of cancer cell death by MB8 was not associated with cell arrest via p53. We have reported the synthesis and characterization of a new dipyridyl benzamide ligand and its subsequent coordination-driven self-assembly with Ru (II) p-cymene acceptors to obtain four molecular bowls. All of the new compounds have been characterized by 1 H NMR 13 ,C NMR, ESI-MS and the structure of the oxalate derived molecular bowl has been established by single-crystal X-ray analysis. The naphthalene derived molecular bowl facilitates cytotoxicity in HepG2 human HCC by HIF-1α-mediated cellular ROS production and glucose metabolism via 1) increased glycolysis through up-regulation of GLUT-4, 2) stimulated conversion of acetyl-CoA by suppression of PDHK 1, and 3) enhanced β-oxidation by stimulation of MCAD. Also, our in vitro experimental evaluations revealed that the MB8 promotes apoptosis via crosstalk between intrinsic and extrinsic cell death pathways via Bid activation in HepG2 cancer cells (Fig. 7). We suggest that the MB8 has development potential as a therapeutic agent. Further research should investigate the exact mechanisms by which the molecular bowl acts and assess the safety of the molecular bowl in vivo for further development as an anti-cancer therapeutic agent.

Materials and Methods
Materials. All chemicals used in this work were purchased from commercial sources. All solvents were distilled via standard methods prior to use. The starting arene−ruthenium acceptor clips were prepared as previously described 10,11 . The 1 H and 13 C NMR spectra were recorded with a Bruker 300 MHz spectrometer. Mass spectra for the self-assembled architectures were recorded using electrospray ionization with a MassLynx operating system  General procedure for the self-assembly of molecular bowls 6-9. Benzamide donor 1 and arene-Ru(II) acceptor 2, 3, 4, or 5 were stirred in 1.5 mL nitromethane/methanol (1:1) at room temperature for 6 h to obtain a clear solution, to which diethyl ether was added drop-wise to precipitate the product, which was washed twice with diethyl ether via centrifugation. The resulting crystalline powders were dried to obtain pure samples of molecular bowls.
Cell culture. Human AGS, A549 and HCT-15 cells were cultured in RPMI-1640. SK-HEP-1 and HepG2 cells were cultured in Dulbecco's modified essential medium (DMEM). All media were supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin in a 37 °C incubator with a 5% CO 2 atmosphere.   Cells were harvested using Trypsin/EDTA reagent, and stained using the Tali ® Apoptosis Kit. For the determination of apoptotic cells, cells were stained using the annexin V-Alexa Flour ® 488 conjugate and necrotic cells were estimated by propidium iodide (PI) staining. The population was assessed using the Tali cytometer.

Cell viability assay. AGS
Quantitative Real-Time PCR. Total RNA from HepG2 cells was isolated with Trizol reagent. The yield, and purity of the RNA were checked by measuring the absorbance values at 260 nm and the ratio of 260 to 280 nm, respectively. Then, 2 µg of total RNA in a 20 µL volume was transcribed using the PrimeScriptII 1st strand cDNA Synthesis kit (Takara, Japan). Quantitative real-time polymerase chain reactions (qRT-PCR) were performed with a MX3005P (Stratagene, USA) using the following primers (Table 2). For real-time PCR, SYBR Premix Ex Taq II (Takara, Japan) was used. The final volume of the reaction mixture was 25 µL containing 2 µL cDNA template, 12.5 µL master mix, 1 µL each primer (10 µM stock solution), and 9.5 µL sterile distilled water. The thermal cycling profile consisted of a pre-incubation step at 95 °C for  Cytochrome c measurement. To determine cytochrome c release, we used the ELISA kit (Enzo Life Science). Briefly, HepG2 cells were treated with MB8 4 µM for 12, 24 and 48 h. HepG2 cells were harvested after centrifugation, resuspended with permeabilization buffer, vortexed and then incubated on ice for 5 min and centrifuged. The supernatants, which contained the cytosolic fraction of cytochrome c, were retained, and RIPA and cell lysis buffer 2 were used to resuspend the pellet. The lysate was incubated on ice for 5 min. After centrifugation, the supernatant containing the mitochondrial fraction of cytochrome c was saved and measured by a multi-reader at 405 nm.
Statistics. All data are presented as the mean ± standard error of the mean (SEM). The 50% inhibitory concentrations (IC 50 ) was calculated by analyzing the log of the concentration-response curves by nonlinear regression analysis. The results were analyzed for statistically significant experimental differences by one-way analysis of variance (ANOVA) and post-hoc Duncan's multiple range tests (DMRT). Statistical analysis was done with SPSS software (version 12.0). A P < 0.05 was considered statistically significant.

Data Availability
The datasets generated during and/or analysed during the current study are not publicly available due to patent registration but are available from the corresponding author on reasonable request.