Synthesis and anti-hepatocellular carcinoma activity of aminopyridinol–sorafenib hybrids

Abstract Sorafenib is recommended as the primary therapeutic drug for patients with hepatocellular carcinoma. To discover a new compound that avoids low response rates and toxic side effects that occur in sorafenib therapy, we designed and synthesized new hybrid compounds of sorafenib and 2,4,5-trimethylpyridin-3-ols. Compound 6 was selected as the best of 24 hybrids that inhibit each of the four Raf kinases. The anti-proliferative activity of 6 in HepG2, Hep3B, and Huh7 cell lines was slightly lower than that of sorafenib. However, in H6c7 and CCD841 normal epithelial cell lines, the cytotoxicity of 6 was much lower than that of sorafenib. In addition, similar to sorafenib, compound 6 inhibited spheroid forming ability of Hep3B cells in vitro and tumour growth in a xenograft tumour model of the chick chorioallantoic membrane implanted with Huh7 cells. Compound 6 may be a promising candidate targeting hepatocellular carcinoma with low toxic side effects on normal cells.


Introduction
Protein kinase is an enzyme catalysing the phosphorylation of a hydroxyl group of a particular protein residue and classified into three broad groups: serine/threonine protein kinases (e.g. Raf, MAPK), tyrosine protein kinases (e.g. VEGFR, PDGFR), and dual-specificity protein kinases (e.g. MEK) 1 . Raf is a serine/threonine kinase, and three different Raf isoforms originating from three independent genes can be distinguished in mammals: A-Raf, B-Raf, and C-Raf (also known as Raf-1) 2 . B-Raf/C-Raf, the predominant heterodimer, activated by growth factor receptor signalling or constitutive Ras activity 3 play important roles in cell proliferation, differentiation, and survival 4 . Somatic mutations in B-Raf frequently with valine 600 mutation found in some types of cancer cause constitutive activation of the enzyme without heterodimerization 5 to induce abnormal cell proliferation and survival 6 . The most frequent B-Raf gene mutation is translated into a mutation of valine 600 to glutamic acid (V600E), which has been observed in more than 90% of melanoma 7 . Several inhibitors of B-Raf(V600E) such as vemurafenib and dabrafenib are currently used to treat melanoma in patients with a B-Raf(V600E) mutation 8,9 . However, patients develop resistance to B-Raf(V600E) inhibitors due to enhanced Raf dimerisation potential by the inhibitors 10,11 , leading to paradoxical activation of the MAPK signalling pathway. One way to overcome the paradoxical MAPK activation is by developing pan-Raf monomer/dimer inhibitors. The other way to achieve ultimate suppression of cancer growth detouring the Raf dimer dilemma is to block both Raf kinases (B-Raf, C-Raf, and mutant B-Raf) and receptor tyrosine kinases linked to activation of MAPKs, such as VEGFR-2. Sorafenib (Figure 1) is an oral multikinase inhibitor targeting intracellular serine/threonine kinases (C-Raf, B-Raf, and B-Raf(V600E)) and receptor tyrosine kinases (mainly VEGFR-2 and PDGFR-b) which is currently approved for the second-line treatment of advanced renal cell carcinoma (RCC), hepatocellular carcinoma (HCC), acute myeloid leukaemia (AML), and radioactive iodine resistant differentiated advanced thyroid carcinoma (DTC) 12,13 .
Molecular hybridisation is a structural modification approach that chemically fuses two or more pharmacophores into one compound of a new structure 14 . In most cases, considering new drug design using a molecular hybridisation strategy mainly aims to increase efficacy with broad-spectrum, avoid drug resistance, reduce side effects and toxicity, or improve druggability. Not a few successful cases demonstrated the high feasibility of this strategy and also made molecular hybridisation be recognised as one of the reliable and useful strategies for new drug discovery 15,16 . We recently reported a study on the design and synthesis of new hybrid compounds (e.g. B) that fuses sunitinib with 4,6-dimethyl-5-hydroxy-7-azaindolin-2-one (A) (Figure 1) 17 . Sunitinib is an oral multitargeted tyrosine kinase inhibitor and is approved for the treatment of renal cell carcinoma (RCC) and imatinib-resistant gastrointestinal stromal tumour (GIST) 18 . Compound B showed in vitro cytotoxicity comparable to sunitinib against several cancer cells, and surprisingly, it showed very little toxicity to normal cells. Compared to sunitinib which showed similar levels of cytotoxicity between cancer cells and normal cells, the hybrid compound B has much higher safety windows than sunitinib, which can be an advantageous point in the development of a new anticancer drug with lesser toxicity.
As an extension of our efforts for the discovery of new promising compounds by molecular hybridisation, we chose sorafenib as the target compound to fuse with our pyridin-3-ol skeleton as in the structure of C for this study. Pyridoxine-derived 2,4,5-trimethylpyridin-3-ol derivatives with various nitrogen-containing functional groups at para-position to the hydroxyl group have been designed and developed by us in the last decade [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33] . This series of compounds C showed good pharmacological actions against reactive oxygen species, angiogenesis, inflammatory bowel disease, autoimmune disease, and cancer. In the structure of new hybrid compounds 4, the central aminophenol substructure in sorafenib is replaced by the 2,4,5-trimethylpyridin-3-ol moiety in compound C.

Synthesis of the sorafenib-aminopyridinol hybrid compound 4
A key feature of the synthesis of compound 4, a hybrid structure of sorafenib and aminopyridinol, is to combine the moieties at both ends of sorafenib, i.e. 4-chloro-N-methylpicolinamide and 4chloro-3-(trifluoromethyl)phenyl isocyanate, with centrally located 6-amino-2,4,5-trimethylpyridin-3-ol (2) (Scheme 1). The formation of the ethereal bond of 3 by nucleophilic aromatic substitution reaction between the oxygen atom of the hydroxyl group in 2 and the carbon atom with chloride in 4-chloro-N-methylpicolinamide was carried out under microwave irradiation with potassium tert-butoxide base in a sealed tube. Then, the other substructure of sorafenib was linked to the aminopyridinol centre by nucleophilic addition reaction of the nitrogen atom in the amino group of 3 to the carbon atom in the isocyanate group in 4-chloro-3-(trifluoromethyl)phenyl isocyanate to form ureido compound 4. The key intermediate 2 was readily prepared by the well-established method which was developed by us starting from pyridoxine hydrochloride 20 .

Comparison of inhibitory activity of the two compounds, sorafenib and 4, against four Raf kinases
Inhibitory effects of the new hybrid compound 4 on four key target kinases of sorafenib, namely A-Raf, B-Raf, B-Raf(V600E), and C-Raf, were measured by cell-free kinase assay systems at a fixed concentration (1 lM). As shown in Figure 2, the assay results, against our expectations, showed general low inhibitory activities of compound 4 compared to sorafenib. The inhibitory activity of compound 4 against C-Raf showed a bit less than that of sorafenib, but it was far behind in the cases of the other three Rafs. In particular, the inhibition of B-Raf with compound 4 was seriously low. These somewhat disappointing results have strongly driven the need for the design and synthesis of derivatives with more diverse structures.

Structural modification of compound 4
For structural modification of compound 4, we first noted a paper published in 2002 by the Bayer Corporation, the original developer of sorafenib 34 . This paper summarises the history of design and discovery from the hit compound found through highthroughput screening targeting C-Raf kinase to the final sorafenib using the medicinal chemistry approach. Considering the three key facts which are (i) the existence of carboxamide functionality in the pyridine ring was found to be the essential factor for potency, (ii) the primary amides showed better activity than the secondary amides due to providing a hydrogen-bond doner in this region, and (iii) the N-methyl substituent was the best compared to other bulkier ones, we decided to design new derivatives that mainly modified the structure of the urea-side part without touching the N-methylpicolinamide part. Many of the new derivatives designed have N-phenyl group with various substituents as in compounds 5-17, and 23. Included were compounds with Nbenzyl 18, N-alkyl 19 and 20, N-benzoyl 21, and N-tosyl 22 in the list of new derivatives as well. Compound 24 where N-methylpicolinamide moiety of compound 4 is replaced by benzyl group was picked for biological evaluation to see the effect of that moiety on potency. Moreover, the free OH analogue 25 in which N-methylpicolinamide was entirely removed was also included in the compound list. Lastly, we wanted to examine how the dimethylpyrimidine ring in 28 instead of the trimethylpyridine ring in 4 affects inhibitory activity. The 18 derivatives 5-17 were readily prepared by the coupling reaction between the common substrate 3 and the commercially available corresponding isocyanates 28 . The aminophenyl derivative 23 was obtained from the corresponding nitrophenyl compound 10 by reduction with tin(II) chloride. The benzyloxy analogue 24 was transformed to free hydroxy compound 25 by boron trichloride along with pentamethylbenzene as a non-Lewis basic cation scavenger. For the synthesis of the pyrimidine-centred analogue 28, 2-amino-4,6dimethylpyrimidin-5-ol (26) was needed and prepared by slightly modifying the known methods 35,36 . The synthetic strategy used to prepare compound 4 was also applied here to synthesise compound 28 finally.

Raf inhibitory abilities of new compounds
The new derivatives were subjected to cell-free kinase assay to examine their inhibitory activities against the four Raf kinases at a fixed concentration of 1 lM. The assay results are shown in Figure  3 and the mean value of % inhibition for each compound was obtained from two independent experiments.
As shown in Figure 3, sorafenib (S) was observed to have similar potent inhibitory effects on all four Raf kinases. On the other hand, looking at the sorafenib-trimethylpyridinol hybrid compounds with inhibitory activity, the degree of inhibition against B-Raf was particularly low compared to the inhibition tendency against other Rafs, and the inhibition against C-Raf showed the highest. This trend was most noticeably observed in compound 4, which had the structure closest to sorafenib. Its level of c-Raf inhibition was 84%, which contrasted sharply with 12% inhibition against B-Raf. The inhibitory effect of compound 3, which was used as the key intermediate in the preparation of many derivatives, was also measured. This was to confirm the correlation between the presence of the urea-side moiety and the Rafs inhibitory activity. In the absence of the ureido-part, it was observed that the inhibitory effect almost disappeared. As in compounds 5 and 6, the inhibitory activity was significantly restored if only one of the two substituents, chloro and trifluoromethyl, remained in the urea-side moiety of sorafenib structure. However, as in compound 7, when both substituents of the phenyl group were removed, the inhibitory ability considerably deteriorated. In the series of derivatives in which various substituents were introduced at the para-position of the urea-side phenyl ring, it was confirmed that the inhibitory activity tends to vary depending on the type of the substituent. First, as in compounds 8-11, it was observed that the inhibitory activity was maintained to some extent by the substituents, F, Br, NO 2 , and CF 3 , that withdraw electrons. Second, in the case of substituents in which oxygen is directly connected to the phenyl ring, such as trifluoromethoxy 12 and methoxy 13, the inhibitory activity was remarkably reduced, whereas in the cases of alkyl group such as methyl 14 and isopropyl 15 there was no significant decrease in inhibitory activity. Lastly, in the case of compound 23 in which an amino group was introduced, the inhibitory activity almost disappeared. As in compounds 16 and 17, when steric bulkiness occurred at the ortho-position of the urea in the phenyl ring, the inhibitory ability dropped sharply. The benzyl compound 18, which has a methylene linker between the ureido nitrogen and the phenyl ring, was much less effective than compound 7, in which ureido nitrogen is directly connected to phenyl ring. Compound 19, which has no phenyl group and npropyl group directly connected to ureido nitrogen had almost zero inhibitory activity, and the cyclopropyl-containing compound 20 showed slightly better results than n-propyl one, but still had little activity. Compounds 21 and 22, each containing a carbonyl group and a sulphonyl group between urea and phenyl ring, respectively, showed weak or inhibitory activity. The importance of the existence of the N-methylpicolinamide moiety was briefly investigated. Both compound 24, which replaced it with benzyl moiety, and compound 25, which completely removed it, caused a serious loss of inhibitory activity. in compound 4, when the trimethylpyridine core ring was changed to dimethylpyrimidine ring as in compound 28, the inhibitory activity was slightly improved.
Compounds 4 and 6 were further examined for half maximal inhibitory concentration (IC 50 ) against four different Raf kinases and the results are summarised in Table 1. The IC 50 values showed a tendency to be roughly similar to the percent inhibition values measured at the 1 lM concentration. The IC 50 values of compound 4 were 6.9-90 times higher than those of sorafenib, whereas compound 6 showed IC 50 values of 1.8-3.3 times higher than that of sorafenib. The difference in the inhibitory activity against each of the four Raf kinases was the same as sorafenib which showed the weakest inhibitory activity against B-Raf.

Molecular docking
Eight kinase structures containing sorafenib have been known. B-Raf, CDK8, KDR, and P38 belong to the kinases. When we align the complex structures with an identical direction, all sorafenib conformations are almost invariant, suggesting the critical intermolecular interactions are shared. To understand the difference of the inhibitory activities in the synthesised molecules at the atomic level, we performed docking studies using sorafenib, compound 4, and compound 6. To improve two key factors in the docking, poses and energies, we undertook a series of tasks. First, we selected the template coordinate that reproduces the crystal poses of the ligands the most. The RCSB protein data bank includes 45 inhibitors complexed with B-Raf proteins. We docked the inhibitors into the RCSB-deposited 86 B-Raf coordinates using Glide-SP 37 and counted the poses reproduced within less than 2 Å rmsd (root mean square deviation in the heavy atoms) compared to the crystal ligands. The coordinate of 1UWJ-A faithfully reproduced the poses the most in 22 cases with reasonable geometries ( Figure S1 in Supplemental material). Interestingly, the co-crystalized ligand with 1UWJ-A is sorafenib (BAY43-9006). The value of rmsd between crystal and docked sorafenib was 0.374 Å. We, second, generated the physicochemically matched but topologically different 50 decoys per inhibitor from the 45 crystal inhibitors using the DUD-E server 38   indicating the early enrichments ( Figure S2 in Supplemental material). Note that the higher values of AUC and LogAUC can reflect the faithfulness of the docking 39 . Third, we performed MMPBSA refinements using three different 20 ns molecular dynamics simulations per inhibitor to approximate the free energies upon the binding 40 . Sorafenib, compound 4, and compound 6 showed the values of À15.0 (± 7.0), À9.3 (± 6.7), and À17.3 (± 1.2) kcal/mol, respectively ( Figure S3 in Supplemental material). The results qualitatively agree with the observation that the inhibitory activities of compound 4 deviated from those of sorafenib and compound 6. Then can the docked poses explain the difference in the inhibitions? Three molecules shared the apparently similar poses. Small but noticeable changes between sorafenib and the other two were the phenoxy moiety of sorafenib deviated slightly from compounds 4 and 6 ( Figure 4(A)). The trimethyl groups of compounds 4 and 6 seemingly cause the difference. The moiety participates in two intermolecular interactions with Lys-483 and Phe-595 of B-Raf. Lys-483 forms the cation-p contact, whereas Phe-595 is involved in the aromatic-aromatic interaction. The subtle change of the phenoxy moiety in compound 4, and compound 6 resulted in the loss of the aromatic-aromatic interaction through Phe-595 ( Figure 4(C,D)). It may explain the decreased inhibition of compounds 4 and 6 from sorafenib, though the difference between sorafenib and compound 6 was not distinguishable in MMPBSA.

Inhibitory activities of compounds 4 and 6 against VEGFR-2 tyrosine kinase
The anticancer effect of sorafenib is also associated with inhibition of angiogenesis in HCC which is a highly vascularised tumour, and the antiangiogenic effect of sorafenib is mediated through blocking VEGFR-2 and other receptor tyrosine kinases. We also compared the inhibitory activity of the hybrid compounds 4 and 6 against VEGFR-2 tyrosine kinase with sorafenib. At a fixed concentration (1 mM), sorafenib showed about 95% inhibitory activity against VEGFR-2 tyrosine kinase, whereas inhibitory activities of compounds 4 and 6 were about 32% and À6%, respectively.

Comparison of antiproliferative and cytotoxic activities of the hybrid compounds and sorafenib
Anti-proliferative activities of compounds 4 and 6 were examined by using three representative HCC cell lines, HepG2, Hep3B, and Huh7, in which Ras/Raf/MAPK signalling pathway is abnormally activated 41 in along with overexpression of EGFR, IGFR, VEGFR, and c-Met 42 . In the three HCC cell lines, the growth inhibitory effects of compounds 4 and 6 were slightly lower than that of sorafenib (Table 2).
In patients receiving sorafenib, adverse events have been reported, and the events are predominantly diarrhoea, weight loss, and hand-foot skin reactions 43 . Since the side effects of the orally available anticancer drug sorafenib seem to be closely related to epithelial cell injury in the gastrointestinal tract and  skin, the cytotoxic effects of the hybrid compounds and sorafenib in normal epithelial cells were compared. Cytotoxic activities of compounds 4 and 6 turned out to be much lower than that of sorafenib. In H6c7, a normal pancreatic ductal epithelial cell line, and CCD841, a normal colon epithelial cell line, the half maximal cytotoxic concentrations (CC 50 ) of sorafenib were about 4.4 mM and 4.1 mM, respectively, whereas CC 50 of compounds 4 and 6 were over 10 mM in both H6c7 and CCD841 cell lines (Table 3).

Effects of hybrid compounds and sorafenib on sphere forming ability of HCC cell lines
Cancer stem cells (CSCs), a subset within tumour tissues, possess the ability to self-renewal and change their properties depending on their surrounding environments, resulting in resistance to conventional therapies, recurrent growth, and distant metastasis 44,45 . We compared the effects of hybrid compounds and sorafenib on the maintenance of CSC population of HCC cell lines by using a sphere-forming assay. In the anchorage-independent sphere-forming assay with serum-free, non-adherent, and nutrition-deficient culture conditions, differentiated tumour cells undergo apoptosis, while CSCs survive, adapt, and proliferate, resulting in the enrichment of the potential CSC subpopulation 46,47 . Because the presence of hepatitis B virus X protein (HBx) is responsible for resistance to targeted therapies in HCC 48 , Hep3B (HBx-positive) and Huh7 (HBx-negative) cells were selected to investigate their spheroid forming ability and reactivity to the hybrid compounds and sorafenib. Huh7 cells formed more and bigger spheroids than Hep3B cells ( Figure 5(A)), and a spheroid-suppressing effect of sorafenib was stronger in Huh7 than in Hep3B cells. However, the inhibitory effect of compound 4 on spheroid formation was similar in both cell lines, whereas compound 6 showed a better inhibitory effect on Hep3B than on Huh7 spheroid formation ( Figure 5(B)).

Anti-tumour effects of hybrid compounds on a CAM tumour model implanted with Huh7 cells
Next, the in vivo efficacy of compounds 4 and 6 on the growth of Huh7 cells xenografted onto CAM was compared with that of  sorafenib. In the CAM tumour model, implanted Huh7 cells developed a tumour mass and tumour-induced angiogenesis ( Figure  6(A)). The inhibitory effect of compound 4 on tumour weight was weaker than that of sorafenib, whereas compound 6 showed an inhibitory effect similar to sorafenib ( Figure 6(B,C)). The tumourinduced angiogenesis was also significantly blocked by sorafenib, and compounds 4 and 6 showed a similar degree of inhibition as sorafenib ( Figure 6(C)).

Conclusion
We designed and synthesised a series of novel compounds by molecular hybridisation of 6-amino-2,4,5-trimethylpyridin-3-ol and sorafenib. From the in vitro kinase assay using four different Raf kinases, compound 6 was found to have similar inhibitory activity to sorafenib. The detailed analysis using docking simulation may provide a clue for the quantitative understanding of B-Raf and inhibitors' interaction. The in vitro inhibitory activity of the hybrid compounds 4 and 6 on the proliferation of HCC cell lines were correlated with the inhibitory activity against Raf kinases, and the degree of inhibitory activity was strong in the order of sorafenib, 6, and 4. HCC cell lines had different spheroid-forming abilities, with Huh7 forming larger spheroid faster. Sorafenib showed stronger inhibitory activity on the spheroid forming ability of Huh7 than Hep3B, whereas compound 6 more strongly inhibited the spheroid formation of Hep3B. The results suggest that compound 6 has good inhibitory activity against CSC formation of  HBx-associated HCC. In conjunction with the previous report that HBx is responsible for resistance to targeted therapies in HCC 49 , the stronger action of compound 6 against Hep3B than Huh7 also suggest that compound 6 may be a better choice than sorafenib for the treatment of a type of HCC, HBx-positive, which develops resistance easily. In conclusion, although inhibitory activities of compound 6 against Raf kinases and VEGFR2 were weaker than those of sorafenib, the similar degree of in vivo anti-tumour efficacy suggests that compound 6 may be considered as an effective compound targeting HBx-positive HCC while causing a less toxic effect on normal epithelial cells. Our current study did not dissect the difference in the inhibitory activities of the small molecules for B-Raf, B-Raf(V600E), and C-Raf. To know how the change of polypharmacological profiles at the enzyme level leads to the different phenotypes in cell and animal models will be an interesting and challenging topic. Our data may be a meaningful addition to this direction.

General
Unless noted otherwise, materials were purchased from commercial suppliers and used without further purification. Air or moisture-sensitive reactions were carried out under an inert gas atmosphere. The reaction progress was monitored by thin layerchromatography (TLC) using silica gel F 254 plates. The products were purified by flash column chromatography using silica gel 60 (70-230 mesh) or by using the Biotage "Isolera One" system with indicated solvents. Melting points were determined using a Fisher-Johns melting point apparatus and were not corrected. Low-resolution mass spectra (LRMS) were obtained using a Jeol ColdSpray-LC-TOF-MS and recorded in a positive ion mode with an electrospray (ESI) source. NMR spectra were obtained using a Bruker-250 spectrometer (250 MHz for 1 H-NMR, and 62.5 MHz for 13 C-NMR) and a Bruker Avance Neo 400 spectrometer (400 MHz for 1 H-NMR, and 100 MHz for 13 C-NMR). Chemical shifts (d) were expressed in ppm using a solvent as an internal standard and the coupling constant (J) in Hertz. HPLC analyses were performed on a system consisted of an LC-20AD pump, a CBM-20A communication bus module, an SPD-20A UV-visible detector, and a DGU-20A5 degasser from Shimadzu Corporation (Kyoto, Japan). A Phenomenex Luna V R C18 column (250 Â 4.6 mm, 5.0 mm) was used with a gradient solvent system consisted of acetonitrile and water (from 10% to 100% of acetonitrile over 15 min, then 100% of acetonitrile for 10 min) at a flow rate of 1.0 ml/min at 254 nm UV detection. The purity of compound was recorded as a percentage (%) and the retention time was given in minutes.

Raf kinase assay
Kinase assays with 4 different Raf kinases were performed by Reaction Biology Corporation using Kinase HotSpot TM assay platform (www.reactionbiology.com). Each human Raf kinase, A-Raf, B-Raf, B-Raf (V600E), or C-Raf (5-10 mU), was separately prepared in 25 ml reaction buffer solution with substrate (MEK1) in a concentration of 5, 3, and 1 mM, respectively. Compounds were delivered into each Raf reaction, followed $20 min later by addition of a mixture of ATP and [c-33 P-ATP] (specific activity approx. 500 cpm/ pmol, concentration as required) to a final concentration of 10 mM. After incubation for 40 min at 25 C, the reaction was stopped by the addition of a 3% phosphoric acid solution. Then, the reaction was spotted onto a P30 filtermat, and unbound phosphate was removed by washing three times for 5 min in 75 mM phosphoric acid and once in methanol prior to drying and scintillation counting. The background counting derived from control reactions containing inactive enzyme was subtracted, and specific kinase activity data were expressed as the percent remaining kinase activity in test compounds compared to vehicle (dimethyl sulphoxide) reactions. IC 50 values and curve fits were obtained using Prism (GraphPad Software, La Jolla, CA).

Vegfr-2 kinase assay
The VEGFR-2 (KDR) kinase assay was performed with a KDR (Promega, WI, USA) kinase enzyme system using the ADP-glow kinase assay kit (Promega) in accordance with the manufacturer's instructions. Kinase activity was detected by the addition 50 mM ATP to a mixture of 0.2 mg/ml poly (Glu4, Tyr1), test drug, and 1.5 ng/ml KDR enzyme. The reaction was carried out at 25 C for 1 h in a total volume of 25 ml. Subsequently, 25 ml of ADP-Glow reagent was added to the mixture, and then incubated for 40 min at 25 C. After the addition of 50 ml of ADP detection reagents for 30 min at 25 C, luminescence was measured using a Fluostar Omega microplate reader (BMG LABTECH GmbH, Ortenberg, Germany).

Cell proliferation assay
Cells were seeded in a 96-well plate and allowed to adhere for 24 h. The cells were serum starved in 1% serum containing medium for 12 h, and then treated with drugs in 10% serum-containing medium. After incubation for 48 h, cell proliferation was measured using MTT assay. Optical density was measured at 540 nm using a Spectrostar Nano microplate reader (BMG LABTECH GmbH, Ortenberg, Germany).

Cytotoxicity assay
Cells were seeded in a 96-well plate and allowed to adhere for 24 h. Cells in low serum (1%)-containing medium were treated with drugs for 48 h. In case of H6c7 cells, cells in keratinocyte serum free media were treated with drugs for 48 h. Then, cell viability was measured using MTT assay. Optical density was measured at 540 nm using a Spectrostar Nano microplate reader (BMG LABTECH).

Sphere formation assay
Sphere forming ability of Hep3B and Huh7 cells were examined after incubation with drugs for 48 h. Briefly, 3 Â 10 3 cells which were pre-treated with or without drugs in prEGM media (Lonza, Basel, Switzerland) were seeded on an ultra-low adhering 24 well plate (Corning Incorporated Costar, Corning, NY). Twice a week, 500 mL of used medium was replaced with fresh prEGM media. At 13 days after plating, images of spheres were captured using Celloger Mini (Curiosis Inc., Seoul, South Korea). The number of spheroids over 50 mm in diameter was counted by using Image J software (National Institute of Health, Bethesda, MD).
4.2.7. Antitumour activity measurement in Huh7-xenografted CAM tumour model Cancer cell xenograft onto chick chorioallantoic membrane (CAM), an in vivo tumour model, was used to evaluate anti-tumour efficacy of compounds. Fertilised chicken eggs purchased from Byeolbichon Farm (Gyeongbuk, South Korea) were incubated at 37 C with 55% relative humidity. On day 9 of egg incubation (embryonic day 9, E9), a small hole was made in the shell over the air sac after the selection of bifurcated vessels. Another hole was made on the broadside using a needle by applying negative pressure from the wider part, creating false air sac. Using a grinding wheel (Dremel, Racine, WI), a small window (1 cm 2 ) was created in the eggshell above the false air sac. For generation of xenograft CAM tumour model, Huh7 cells (1.5 Â 10 6 cells/CAM) were inoculated on the CAM with or without test drugs (10 mM final concentration). On day 13 of egg incubation (E13), tumour tissues attached to CAM were resected from the embryo and harvested. Photographs of tumour as well as blood vessel branches were taken using optical microscope (Olympus Corporation, Tokyo, Japan). Tumour tissues were weighted and the number of vessel branch points within the tumour region was counted.
Chick embryo experiments were approved beforehand by the Institutional Animal Care and Use Committee of Yeungnam University and were performed accordingly the guidelines issued by the Institute of Laboratory Animal Resources (1996) and Yeungnam University (The care and use of animals 2009).

Docking study
Small molecules were prepared into 3D coordinates considering tautomerization and electrostatic charge using LigPrep and Epik from Schr€ odinger. Docking screen with sorafenib, compound 4, and compound 6 against the coordinates of B-Raf followed. Glide-SP 36 of Schr€ odinger package (Schr€ odinger, LLC, New York, NY) was used as a docking engine and score. Intermolecular interactions were compared and visualised using the Grapheme Toolkit of Openeye package. The protocols in the previous study were employed for MMPBSA approach 50 . In brief, the SQM and LEaP programs in AMBER 18 generated the GAFF force fields of small molecules for the molecular dynamics (MD) simulation, whereas the ff14SB force field was employed for B-Raf (A chain of PDB code, 1UWJ). Four stages of minimisation, heating, equilibrium, and production comprised a single run. After 1500 cycles of minimisation, the heat of the system was elevated from 0 K to 300 K at a constant volume for 50 ps. The equilibrium step consisted of a constant pressure (1 atm) and temperature (300 K) continued for the next 100 ps. Of the 20 ns MD simulation, the latter 10 ns trajectories were employed for "MMPBSA.py" analyses 39 to extract the change of free energy from the data. The GPU version of PMEMD generated the trajectory through MD simulation.