br Results br Discussion br Experimental
In the preceding paper, we reported 3-amido-4-anilinoquinolines as highly selective inhibitors of CSF-1R kinase, along with our initial efforts to optimize the series. Examples from the 6,7-dimethoxyquinoline series were very potent in our enzyme and cell assays, but typically had poor physical properties and rat PK, with short half-lives. Activity in our mouse PD model was achieved with Missed Opportunities (), but only at the high dose of 100mpk. With the goal of optimizing the physical properties and PK profile of the series, basic side chains were installed at the 6- and 7-positions of the quinoline scaffold, linked through oxygen and nitrogen atoms. Compounds of this type (e.g., , ) retained the good enzyme and cellular potency of the dimethoxyquinoline leads, and also possessed high aqueous solubility and low levels of plasma protein binding. For these compounds, however, the in vivo clearance in rats was extremely high, and oral exposure was minimal. The amidoquinolines displayed a remarkable kinase selectivity profile, however, with examples found to have activity against almost no kinases other than CSF-1R. This observation encouraged our continued efforts to optimize the series. The introduction of a basic side chain has been used to improve the physical properties and PK profiles of other series of kinase inhibitors, to deliver marketed drugs such as Gleevec and Iressa. However, this structural feature proved to be a metabolic liability for the amidoquinoline scaffold. We next installed cyclic amines such as methylpiperazine directly at the 6- and 7-positions. Biological and pharmacokinetic data for these compounds are given in . In the 7-MeO series, the dichloroanilino compounds, including the very potent 2,3-Cl analogue , were highly cleared, but still showed an improvement over compounds in this series with basic side chains. Encouragingly, the 2,4-F example had excellent rat PK, including good oral bioavailability, although it had an IC of only 1μM in our cell proliferation assay. From a small number of matched pairs, and influenced by earlier data on related compounds, we felt that potency was better with the piperazine substituent at the 6-position rather than the 7-position, and focused subsequent efforts on compounds of this type. We speculated that oxidation of the electron-rich quinoline ring might contribute to the high rate of clearance, and compounds were prepared without the 7-methoxy group, and with a 7-F substituent in its place. Compounds without the 7-MeO substituent () showed reduced activity in both enzyme and cell. The two 7-F compounds (, ) did in fact show moderate in vivo clearance, but were also less potent. The 2,4-F example was only weakly active in our cell assay (4.2μM). The 2,3-Cl example was closer to the level of potency we felt was required (0.44μM in the cell), but still significantly less active than the corresponding 7-MeO compound . We had observed previously improved potency for 6-aminoquinolines when 7-MeO was replaced by 7-EtO. The 2,4-F analogue achieved both good in vitro activity, 0.23μM in the cell, and excellent in vivo PK. The 2,3-Cl compound had higher clearance in rats than the 2,4-F example, but was very potent in the cell assay (0.06μM). With a combination of good cell potency and encouraging PK, compounds and represented excellent advanced leads for further optimization of the series. We next explored a range of cyclic amines at the 6-position, with the goal of identifying compounds with the best overall balance of potency and PK, as determined by our mouse PD model. Sets of compounds were prepared with both 2,3-Cl aniline – (best in vitro potency) and 2,4-F aniline – (best physical properties and good PK). Data for these compounds is shown in . To assess the in vivo CSF-1R activity of the amidoquinolines, compounds were dosed orally in a mouse pharmacodynamic (PD) model. 3T3 cells were engineered to express human mutant full length CSF-1R (301–969) (3T3/CSF-1R in which the kinase activity was constitutively on. Female nude mice were implanted with 5×10 3T3/CSF-1R cells subcutaneously and grown in vivo until tumors were >250mm in size. Tumors were analyzed for pCSF-1R levels by ELISA 2 and 6h after dosing, and blood plasma samples were assessed for drug concentrations. Earlier compounds had been dosed at 50 or 100mpk, but the potency and other properties of these amidoquinolines allowed us to drop the screening dose to 25mpk. Compounds with 2,3-Cl aniline were typically ∼10-fold more potent in the cell assay than the 2,4-F aniline examples, but had higher levels of plasma protein binding, and higher rat in vivo clearance. The choice of amine at the 6-position had relatively little effect on cell potency, but a significant impact on physical properties and PK. Basic groups here typically resulted in compounds with a superior profile to those with neutral substituents such as morpholine (), or the piperazine amides (, ). Examples with both 2,3-Cl and 2,4-F anilines showed good PD activity out to 6h.