Design, synthesis and biological evaluation of 6‐substituted quinolines derived from cabozantinib as c‐Met inhibitors

Based on the cabozantinib scaffold, novel c‐Met inhibitors were rationalized from the limited knowledge of structure‐activity relationships for the quinoline 6‐position. Emphasis was given to modifications capable of engaging in additional polar interactions with the c‐Met active site. In addition, ortho‐fluorinations of the terminal benzene ring were explored. Fifteen new molecules were synthesized and evaluated in a c‐Met enzymatic binding assay. A wide range of substituents were tolerated in the quinoline 6‐position, while the ortho‐fluorinations performed were shown to give considerable reductions in the c‐Met binding affinity. The antiproliferative effects of the compounds were evaluated in the NCI60 cancer cell line panel. Most notably, compounds 15b and 18b were able to inhibit cell proliferation more efficiently than cabozantinib in leukemia, CNS, and breast cancer cell lines. The in vitro data agreed well with the in silico docking results, where additional hydrogen bonding was identified in the enzymatic pocket for the para‐amino substituted 15b and 18b.

cabozantinib. [11][12][13] For the hinge-binding quinoline moiety, modifications have primarily been made at the 7-position, these frequently being groups intended to increase solubility. [14] With other heterocycles as hinge-binders, substituents known to participate in more direct interactions with the enzyme have been explored, exemplified with amines, anilines, and nitrogen containing heterocycles in Figure 2. [15][16][17][18]  was of interest since this would block one of the main metabolic pathways for this compound class. [19,20] In this study, we present the synthesis and in vitro evaluation of these novel kinase inhibitors.

| Chemistry
The two fluorinated analogs of the terminal aromatic ring, 6a and 6b, were prepared starting from cyclopropane-1,1-dicarboxylic acid by reaction with the corresponding aniline, as shown in Scheme 1.
The 6-hydroxy-7-methoxyquinoline scaffold in 5 was synthesized in a similar manner to earlier reported work, [21] although starting from 3hydroxy-4-methoxy acetophenone, which was first benzylated to 7, then nitrated to 8, and further reduced to the aniline 9. Cyclization into the 4-hydroxy quinoline 10 was achieved using ethyl formate and further reacted with 1-fluoro-4-nitrobenzene to give diaryl ether 11, which was reduced to the aniline 12. The acids 6 were then coupled with aniline 12 into the main scaffold 13, which could then be F I G U R E 1 Examples of known inhibitors of c-Met, including the regulatory approved inhibitors cabozantinib (1) and crizotinib (2) F I G U R E 2 Examples of known variations around the hinge binding heterocycles and our proposed target scaffold 5. Atoms with the capability of engaging in polar interactions are displayed in red deprotected to the phenols 14. Cleavage of amide bonds was observed using hydrogen gas, and 1,4-cyclohexadiene was therefore applied as a milder hydrogen source. Synthesis of 19 was performed in a similar manner, using steps f-h, starting from 6,7-dimethoxyquinolin-4-ol.
These reagents were prone to produce several side-products, so the yields were correspondingly low. Using 2-chloro-5-nitro-pyridine, the nitropyridyls 18a and 18c were achieved, which then were reduced to the corresponding aminopyridyls 18b and 18d.

| Biology
The novel compounds were evaluated for enzymatic c-Met binding affinity, and the results are reported in Table 1.
The introduction of two additional fluorine atoms on the terminal aromatic ring reduced the inhibition of c-Met, and this is observed for all six pairs of compounds shown in Table 1 In summary, substituents in the quinoline 6-position capable of engaging in polar interactions seem to augment c-Met affinity.
The analogs 14a, 15a, 15b, 18b, and 18d were progressed for further evaluation in a cancer cell proliferation inhibition assay. This selection was based on low IC 50 values for c-Met, while simultaneously maintaining a structural diversity to further explore the efficacy in cell-based assays. These studies were performed using the NCI60 program at the National Cancer Institute (NCI). [22,23] Here, the compounds were tested at 10-µM concentration in a broad range of cell lines from nine different tumor types. The results are presented in Table 2 as observed growth percent.
From Table 2, it can be deduced that the tested compounds were able to inhibit growth in a wide range of tumor cell lines. The most potent compounds were 15b and 18b, which is consistent with the observed trend in Table 1. The compound 14a, the 6-O-demethylated analog of cabozantinib, is seen to have a markedly reduced ability for the inhibition of cell proliferation, even though c-Met affinity is comparable with cabozantinib. The importance of the para-amino group in 15b and 18b is evident from the notably reduced capability of the para-nitro analog 15a to influence the growth rates, despite comparable c-Met IC 50 values. These observations indicate that the 6-position on the quinoline ring is important for the interaction with other kinases in addition to c-Met, as can be expected for this class of multikinase inhibitors. The same trend is also seen with the trifluorinated compound 18d performing overall better in the cellbased assay than compounds 14a and 15a, despite its lower c-Met affinity. This observation is in compliance with known SAR on related structures that have shown that c-Met affinity is more sensitive to modifications on the terminal benzene ring than is VEGFR. [24] Compared with cabozantinib, higher growth suppressive effects are S C H E M E 2 Synthesis of compounds 15-18. Reagents and conditions: (a) 4-nitrobenzoyl chloride, Cs 2 CO 3 , DMF, rt, 5 hr, 37%; (b) Fe, NH 4 Cl, EtOH, H 2 O, 70°C, 3 hr, 60%; (c) 4-(trifluoromethyl)benzoic acid, HATU, DMAP, DMA, rt, 16 hr, 53%; (d) chlorodifluoromethylbenzene, Cs 2 CO 3 , DMF, 100°C, 20 hr, 6%; (e) 1,1,1-trifluoro-2-iodoethane, Cs 2 CO 3 , DMF, 110°C, 5 hr (17a in 33%, 17b in 4%); (f) 2-chloro-5-nitro-pyridine, Cs 2 CO 3 , DMF, rt, 1.5 hr (18a in 75%, 18c in 93%); (g) Fe, NH 4 Cl, EtOH, H 2 O, 70°C, 3 hr (18b in 54%, 18d in 59%) seen with 15b and 18b in several of the cell lines, particularly in cells derived from leukemia, CNS, and breast cancer. 15b and 18b were progressed for 5-dose testing, and the results for selected cancer cell lines are reported in Table 3.
The results from the 5-dose assay corroborate compounds 15b and 18b as potent inhibitors of cancer cell proliferation. All mean GI 50 , TGI, and LC 50 values for 15b and 18b are lower than for cabozantinib, except for the TGI value for 15b. The lowest GI 50 values observed were 50 and 40 nM with 18b in the HOP-92 and KM-12 cell lines, respectively. The reported means in Table 3 are for all NCI60 cancer cell lines, and the complete data are given in the Supporting Information.
To examine whether the structural modifications affected the kinase selectivity profile, a screen was performed on six kinases in addition to c-Met. Kinase selectivity was assessed for the analogs 15b, 18b, and 18d, and is presented in Table 4.
As can be seen in Table 4, the novel analogs have a similar kinase selectivity profile as cabozantinib, albeit with a lower affinity toward c-Kit. Interestingly, 18b exhibits a stronger inhibition of ALK. In light of the different cell proliferation results among the compounds in Table 2, additional modes of action cannot be ruled out. were, therefore, further evaluated by molecular docking using AutoDock Vina [25] via the PyRx [26] interface. The experimental crystal structure with the c-Met inhibitor foretinib (PDB: 3LQ8) was employed.

| In silico evaluation
By overlaying the docked structures of cabozantinib, 15b and 18b, it is seen in Figure 3a          and then filtered through Celite, which was then washed with EtOAc (20 ml). The organic filtrate was washed with water (10 ml) and brine (10 ml), dried over MgSO 4 , and concentrated on a rotary evaporator.

| Enzymatic c-Met assay
The enzymatic c-Met assay was purchased from Cyclex and used following the manufacturer's instructions. Briefly, while placed on ice, recombinant c-Met was added to wells precoated with a substrate, and the reaction was started by adding buffer containing the inhibitors in appropriate dilutions. The plate was incubated at 30°C for 60 minutes. After washing with buffer, a horseradish peroxidaseconjugated detection antibody PY-39 was added to each well, and then incubated at ambient temperature for 60 min. The TMB substrate was added after another round of washing and then incubated at ambient temperature for 10 min. Stop solution was then added, and absorbance was measured using a spectrophotometric plate reader (Perkin Elmer VICTOR™ X3). The results were analyzed using GraphPad Prism 7.04.

| Cell proliferation
Testing was performed by the Developmental Therapeutics Program,

Division of Cancer Treatment and Diagnosis, National Cancer
Institute. The studies were performed using the NCI60 panel and performed according to their internal procedures. [29] 4.3 | Molecular docking Dockings were performed using AutoDock Vina [25] via the PyRX [26] interface. The experimental crystal structure of foretinib in the enzymatic site of c-Met was downloaded from the Protein Data Bank (PDB: 3LQ8). This was prepared for docking using AutoDock Tools (ligand and water removed and polar hydrogens added). The ligands were build using Avogadro, [30] and initial geometrical optimization was done using the same software. After docking, visualization of the conformations and binding interactions were done in PyMol. [31] Initially, the performance of the docking method was validated by the redocking of the experimental ligand.