Preparation and Cytotoxic Evaluation of PGV-1 Derivative, CCA-1.1, as a New Curcumin Analog with Improved-Physicochemical and Pharmacological Properties

Purpose: This study aimed to challenge the anticancer potency of pentagamavunone-1 (PGV- 1) and obtain a new compound (Chemoprevention-Curcumin Analog 1.1, CCA-1.1) with improved chemical and pharmacological properties. Methods: CCA-1.1 was prepared by changing the ketone group of PGV-1 into a hydroxyl group with NaBH4 as the reducing agent. The product was purified under preparative layer chromatography and confirmed with HPLC to show about 93% purity. It was tested for its solubility, stability, and cytotoxic activities on several cancer cells. The structure of the product was characterized using 1HNMR, 13C-NMR, FT-IR, and HR-mass spectroscopy. Results: Molecular docking analysis showed that CCA-1.1 performed similar or better interaction to NF-κB pathway-related signaling proteins (HER2, EGFR, IKK, ER-alpha, and ER-beta) and reactive oxygen species (ROS) metabolic enzymes (NQO1, NQO2, GSTP1, AKC1R1, and GLO1) compared with PGV-1, indicating that CCA-1.1 exhibits the same or better anticancer activity than PGV-1. CCA-1.1 also showed better solubility and stability than PGV-1 in aqueous solution at pH 1.0–7.4 under light exposure at room temperature. The cytotoxic activities of CCA-1.1 against several (10) cancer cell lines revealed the same or better potency than PGV-1. Conclusion: In conclusion, CCA-1.1 performs better chemical and anticancer properties than PGV-1 and shows promise as an anticancer agent with high selectivity.


Molecular docking analysis
To compare the binding interaction between CCA-1.1 and PGV-1 toward several cancer markers, we performed a molecular docking analysis of several proteins involving ROS metabolic enzymes and NFB-pathway related proteins based on previously reported works. The computational simulation was performed using licensed-software MOE 2010.10 to simulate molecular binding, calculate RMSD, and visualize protein-ligand interaction. The ROS metabolic enzymes were represented by NQO1 (PDB ID: 1D4A), NQO2 (PDB ID: 4FGL), AKR1C1 (PDB ID: 1MRQ), GST-P1 (PDB ID: 5J41), and GLO1 (PDB ID: 1QIP) enzymes, and the binding interaction targeted on the cofactor site 1 . The molecular docking study on NFB-pathway related proteins were conducted toward HER2 (PDB ID 3PP0), EGFR (PDB ID: 1XKK), and IKK (PDB ID: 4KIK), focusing on the native ligand binding site 16 . We also performed molecular docking on ER-alpha (PDB ID: 3ERT) and ER-beta (PDB ID: 5TOA) as the representation of NFB-pathway related proteins and used site finder mode from MOE due to the unknown reported binding site. The default settings were utilized as long as there is no further explanation. The chemical structure of CCA-1.1 and PGV-1 was created in ChemDraw software and then minimized the structural energy generated for conformational structure in MOE. The docking simulation setting used the MOE default mode, such as triangle matcher and London dG, as the placement setting and scoring method. The force field method was used to refine the docking results from 10 retain settings. The molecular docking results described the affinity represented by the docking score and the binding interaction of each compound with the target proteins.

Absorption, distribution, metabolism, and excretion (ADME) prediction
Considering that certain compounds' interaction to the protein targets also depended on their bioavailability profile, we used pkCSM software (http://biosig.unimelb.edu.au/pkcsm/prediction) to compare the ADME profile prediction between PGV-1 and CCA-1.1. CCA-1.1 and PGV-1 have encoded the SMILES code and then generated it for pkCSM analysis. The parameters collected in this study were water solubility, Caco2 permeability, intestinal absorption, VDss, and clearance.

Main materials and general analytical procedures
PGV-1 and curcumin were obtained from the Cancer Chemoprevention Research Center (CCRC), Faculty of Pharmacy, UGM. The reagents and solvents in this experiment were classified as an analytical grade unless otherwise stated. The 1 H-and 13 C-NMR spectra of CCA-1.1 were recorded in DMSO-D6 with a JNM-ECZ500R (500 MHz, JEOL Ltd., Tokyo, Japan) spectrometer. FT-IR spectrum (KBr disc) was determined using an FT-IR spectrophotometer (Perkin Elmer, USA) at 4000-400 cm -1 . The ESI-MS spectrum of CCA-1.1 was obtained by using LCMS-2010EV (Shimadzu, Japan). The purity of CCA-1.1 was determined by using a C18 column (Chromosorb) with mobile phase acetonitrile: water (70:30 v/v) in UV-Vis HPLC (Hitachi, Japan). The melting point was obtained by the B-545 instrument (Buchi, Switzerland).

Synthesis of CCA-1.1
Synthesis of CCA-1.1 was conducted by reducing the ketone into a hydroxyl group. An amount of 100 mg PGV-1 (0.1 mmol) was dissolved in 10 mL of ethanol. The 11 mg of NaBH4 (0.3 A c c e p t e d M a n u s c r i p t mmol) was added then stirred for 1 h. The crude product was purified using PLC Silica gel 60 F254 (Merck) with a mobile phase system of chloroform: methanol (90:10).

Dissolution and stability test in buffer solution
A stock of curcumin, PGV-1, and CCA-1.1 solution was prepared in DMSO (Merck) at the concentration of 100 mM. The tested solution of curcumin, PGV-1, and CCA-1.1 was prepared by diluting the stock solution in distillate water, phosphate buffer saline pH 1.0, and pH 7.4 up to 100 M. Each solution in the respected buffer was then evaluated for its stability at ambient temperature and light exposure. It measured the absorbance change at 428, 416, and 404 nm, subsequently over a time-course using a UV-1800 spectrophotometer (Shimadzu, Japan). Cells were grown in 24 or 96-well microplate and then cultured for 24 h. Subsequently, cells were treated with PGV-1 or CCA-1.1 at concentration series up to 10 M, except for NIH-3T3 up to 100 M and incubated for 24 h. Untreated cells were used as control. After treatment, viable cells were enumerated using the trypan blue exclusion or quantified under MTT assay 17 . The viable cells were represented as % cell viability vs. concentration of the sample, and the IC50 value or the concentration that inhibits 50% of cell growth was calculated.

Data Analysis
Molecular docking results were validated by determining the RMSD value of conformation bearing the lowest docking score representing the G (kkal/mol). The validity of the molecular docking method was represented as RMSD value < 2. Cytotoxic potencies against several cell lines were statistically analyzed based on the IC50 values through linear regression with a pvalue of > 0,05.

CCA-1.1 binds to targeted-proteins of PGV-1
The binding interaction of PGV-1 to several ROS metabolic enzymes and NFB-pathway related proteins contributed to the potent cytotoxic effect. Molecular docking of several types of target proteins such as HER2, EGFR, IKK, and ER showed that CCA-1.1 exhibited good or A c c e p t e d M a n u s c r i p t better affinity than PGV-1 to all target proteins (Figure 2). Similar to PGV-1, CCA-1.1 also interacted appropriately with ROS metabolic enzymes, such as AKC1R1, GSTP1, GLO1, NQO1, and NQO2 (Figure 3). The molecular docking analysis revealed that the docking score of CCA-1.1 was lower than that of PGV-1 on all target proteins, except IKK, indicating improved binding affinity ( Table 1; Supplementary Table 1). The hydroxyl group on the cyclic structure formed hydrogen bonds on NQO1, NQO2, GLO1, and ER-alpha, which possibly contributed to the higher affinity of CCA-1.1 than PGV-1. Overall, the molecular docking analysis predicted the higher affinity of CCA-1.1 than PGV-1, making it a more potent cytotoxic agent than PGV-1.

ADME profile of CCA-1.1 and PGV-1
To predict whether or not the modification of ketone into a hydroxy group improves the bioavailability or not, we expected the ADME profile of PGV-1 and CCA-1.1 by using pkCSM software. The ADME prediction from pkCSM showed that CCA-1.1 exhibited lower log P and higher water solubility than PGV-1. Despite the lower Caco2 permeability and intestinal absorption of CCA-1.1 than PGV-1, the VDss of both compounds were similar (Table 2). Summarizing the ADME profile prediction, it can be indicated that CCA-1.1 performed better bioavailability than PGV-1.  In the 13 C-NMR spectrum, the resonance of C-1, C-2/C-5, and C-3/C-4 of the cyclopentanol ring was represented by the resonance signals  124.6, 134.5, and 25.9 ppm, respectively. The chemical shift of C-2/C-5 was high because they are two quaternary sp2 carbon atom of a vinyl moiety. The other vinyl carbon (C-6 and C-7) signal appeared at  126.7 ppm. The carbon resonance at  132.5, 131.3, and 155.1 ppm were assigned to C-1', C-2'/C-3'/C-5'/C-6', and C-4' of the benzene ring, respectively. The aromatic methyl carbons were observed at  14.8 ppm. Basing on the above spectral analysis, we confirmed the synthesis product was 2,5-bis[(4'-hydroxy-3',5'-dimethyl)benzylidene)]cyclopentanol or CCA-1.1.

Solubility and stability of CCA-1.1 compared with PGV-1
The evaluation of solubility and stability of CCA-1.1 were conducted in the buffer and nonbuffer medium. We used buffer in pH 1.0 (which represents pH in the stomach), pH 7.4 (which is defined as pH situation in lumen also in blood and colon), and distilled water to describe the differences of proton environment. The result showed that CCA-1.1 was more soluble than PGV-1 and curcumin at a concentration of 100 M (Figure 6). We then observed all compounds' stability in storage times under a light environment monitored by a UV spectrophotometer. We found that curcumin and PGV-1 tended to decrease the absorbance at 3 h, but not for CCA-1.1, especially in pH 1.0 (acidic), but not in pH 7.4. We also determined A c c e p t e d M a n u s c r i p t the stability of both PGV-1 and CCA-1.1 after being stored at room temperature for 48 h. We realized that the CCA-1.1 solution still looked clear, but the PGV-1 solution appeared with some sediment or crystal (Figure 6D), showing that PGV-1 was more unstable than CCA-1.1.

Cytotoxicity of CCA-1.1 compared with PGV-1
The cytotoxic effect of CCA-1.1 was investigated in various breast cancer cell lines (MCF-7/HER2, 4T1, MCF-7, HCC1954, T47D), human leukemic cells (K562), human colon carcinoma cells (Caco2 & WiDr), and immortalized fibroblast cells (NIH-3T3). Cytotoxic assay on several cancer cell lines showed that the IC50 values of both PGV-1 and CCA-1.1 were revealed to be less than 10 M (Table 3). Interestingly, CCA-1.1 was found to be more cytotoxic than PGV-1 on 4T1 and K562 cells. Although both compounds exhibited superior cytotoxic activities against cancer cells, it was still less toxic on non-cancerous NIH-3T3 cells with a concentration up to 50 M indicating that both compounds provide high selectivity.

Discussion
Curcumin is already known as a natural compound that shows cytotoxic activities against several types of cancer cells but unstable in aqueous solution resulting in low bioavailability 12,18 . Curcumin is easily degraded through breaking in alpha-beta diene bonds to generate aldehyde (vanillin) compound 10 . In summary, curcumin possesses limitations such as low water solubility, slight absorption in the gut, and less stable in vivo experiment 19,20 . Numerous attempts have been carried out to overcome curcumin problems by synthesizing the modified curcumin, like its analogs, derivatives, and pro-drugs, but without sufficient results 1 . Moreover, although curcumin analog PGV-1 has superior cytotoxic properties than curcumin, it can have less perfect stability. The present research developed a new derivative compound from PGV-1 by reducing the ketone group into a hydroxyl group while retaining alpha-beta diene to achieve better physical-chemical properties and cytotoxic effects than PGV-1 and curcumin since the diene part has contributed to cytotoxic effect 21,22 . The synthesis of CCA-1.1 by reducing the ketone group on PGV-1 produced a good result with a yield of 10% under PLC separation. Analysis using FT-IR showed a dramatic decrease in the peak at around 1620 cm -1 , which indicates a change in the ketone to a hydroxy group without a change in the diene bonds. This synthesis result, although still not perfect, can already show their repeatability so that this method can be applied for further synthesis. The results collected can also be sufficient for other experiments that are needed in anticancer in vitro studies. However, the reaction conditions must be optimized to increase the yield of the product. The A c c e p t e d M a n u s c r i p t method can be improved by optimizing reaction time, reaction techniques, and concentrations, including the separation technique 15 . Theoretically, NaBH4 can reduce the ketone group up to >90% with more than 40% efficiency. Therefore, the yield may increase and reach up to 40% 14 . Furthermore, in solvent-free environments with a contribution of wet SiO2 (30%), the reduction using NaBH4 can be performed effectively 15 . Ireson et al. 23 also achieved several metabolic reductions of curcumin for metabolism studies of curcumin by reducing the ketone groups in curcumin. This new compound shows several advantages compared to PGV-1 in several aspects. CCA-1.1 has better solubility in water solvents in various pH compared with PGV-1. During observation at pH 1.0; 7,4; also in distilled water, CCA-1.1 was two times more soluble than PGV-1 at observations of up to 48 h where at a concentration of 100 M, PGV-1 dissolved showed the presence of insoluble aggregates while the CCA-1.1 solution was still apparent (Figure 6). The PGV-1 solution at a 50 M concentration even appeared clear (data are not shown). This phenomenon is comparable with the solubility of curcumin at 40 M 10 . Further chemical stability test exhibited that CCA-1.1 performed unchanged absorbance up to 3 h but not for PGV-1, indicating that CCA-1.1 was more stable in an aqueous solution in an acidic environment than PGV-1. However, both compounds' stability is still better than that of curcumin, which decreases in absorbance in less than 60 min 10,11 . This result indicates that CCA-1.1 significantly increases the solubility and stability of PGV-1, making it a promising anticancer drug with minimal side effects. Previous molecular docking studies highlighted the contribution of the two benzenes or its substituent of PGV-1 on the formation of hydrophobic or hydrogen bonds. Still, none of them mentioned the role of its cyclic ketone 1,24,25 . Furthermore, hydrogen bonds' appearance on the structure between the two benzenes increased the affinity of many curcumin derivatives on several cancer marker proteins 18 . In the present study, we synthesized compound CCA-1.1 by modifying the cyclic ketone into cyclic alcohol to improve the binding affinity of PGV-1. CCA-1.1 performed comparable and even better in protein target interaction than PGV-1 on several marker proteins in NF-B signaling and ROS metabolic enzymes. The NF-B signaling pathway controls the expression of numerous genes that regulate cell proliferation, stress responses, and apoptosis 26 . Inhibiting NF-B signaling has potential therapeutic applications in cancer therapy 27 . The docking result provides a compelling reason to develop this new compound (CCA-1.1) as an anticancer agent. Moreover, its ability to interact with ROS metabolic enzymes reinforces its anticancer mechanisms' rational basis because increased ROS in the cell is believed to be one of the selective targets of anticancer agents 28 . Several research types realize that a significant increase in ROS in cells can cause cancer cell death and rarely found in normal cells 29 . These binding affinities of CCA-1.1 against several ROS metabolic enzymes in silico should be interesting to be clarified in a laboratory experiment with several physiological implications. These molecular interacting models of CCA-1.1 would be essential to contribute its cytotoxic activities against cancer cells with the related markers. Our data support the phenomenon that CCA-1.1 performs onefold to twofold better cytotoxic activities on several cancer cells than PGV-1, which agrees with previous findings [1][2][3]30 . In this regard, the two synthesized compounds were evaluated in vitro against some samples of breast cancer cell lines (MCF-7 This phenomenon should be noted as an important finding as consistent with the previous result for PGV-1 that should be explored further for more deep studies. Last but not least, we may also look at the cytotoxic results against colon and breast cancer cells, especially against TNBC and non TNBC cancer cells. This result is not extraordinarily different but essential to consider the alternative or choices for some specific cancer types. As various cells had other characteristics for their cell signaling, CCA-1.1 needed to be elaborated more for its molecular mechanism towards cancer cells. Cancer cells with HER2+ feature found in more than 30% cancer cases 31 . Human Epidermal Growth Factor Receptor-2 (HER2) is a notable tyrosine kinase receptor responsible for the progression, proliferation, and metastasis of cancer 32,33 . Regarding these results, whether CCA-1.1 modulates the HER2 signaling to inhibit cell growth will be the exciting focus for further research. Besides, the strong cytotoxic effect of CCA-1.1 against triple negative breast cancer cells, though only 15%-20% of cases led to elevated mortality in patients due to lack of specific target to eradicate these cancers, 34,35 also need to be explored accordingly. Although using a limited number of cells (10 types of cell lines), this cytotoxic evaluation is sufficient to represent the target of cancer types as in the PGV-1 test as a reference compound. Later, we also identified possible mechanism activity from CCA-1.1 in breast cancer and colorectal cancer cells, including cell cycle arrest at the mitotic phase, induced a high amount of intracellular ROS that led to cell senescence, as well as attenuated cancer cells migration through inhibition of MMP-9 activity 36-40 . Additionally, bioinformatic analysis of CCA-1.1 revealed several possible target genes, including TP53, MAPK1, and ERBB2 in colorectal cancer 41 . The overall results showed that CCA-1.1 might replace PGV-1. It has been reported that PGV-1 and being selective in vitro also show minimal side effects in vivo; therefore, CCA-1.1 is also expected to be safe. Even more, we should consider the superior solubility and stability of CCA-1.1 in the aqueous solution that may be important for the development of dosage form for clinical application. The ADME prediction of CCA-1.1 that looked better than PGV-1 also needs to be proven in the real experiment, especially with a specific formulation. Therefore, we still have to challenge ourselves to prepare a formula that performs much more stable in aqueous solution for intravenous administration.

Conclusion
The CCA-1.1 can be prepared by reducing PGV-1 using NaBH4. This new compound exhibited better solubility and stability in aqueous solution with better cytotoxic activity towards several cancer cells than PGV-1. CCA-1.1 can be developed as a potent anticancer agent with fewer adverse effects.

Ethical Issues
Not applicable

Conflict of interest
The authors declare that there is no competing interest in preparing and publishing the manuscript.