First we synthesized inhibitor following the patent procedur
First, we synthesized inhibitor following the patent procedure reported by Shewchuk et al. to provide its first complete characterization (see ). Two different synthetic approaches were used to obtain compounds –. The first method relied on the alkylation of the common 5-phenol-2,4-diaminopyrimidine intermediate (). We envisioned that this fragment could also be used as nonradioactive precursor for the synthesis of [F] and [F] via simple alkylation employing F-fluorobenzyl halides. Our initial strategy towards involved the direct catalytic hydrogenolysis of . Disappointingly, this approach only delivered very low yields of under the different conditions tested. Alternatively, the synthesis of was envisioned via the deprotection of either the -MOM- or -THP-protected 2,4-diaminopyrimidine intermediates and . Those compounds were synthesized via condensation/cyclization using the protected vanillin and with either the 3-morpholinopropionitrile/aniline exchange strategy or with 3-ethoxypropionitrile followed by treatment with guanidine (). Overall, the cyclization with the β-morpholinopropionitrile intermediate proved far more useful than the synthesis of . MOM deprotection of afforded the phenol intermediate in 28% overall yield from vanillin. Attempts to combine the use of the more labile THP protecting group with the 3-morpholinopropionitrile/aniline exchange strategy failed to yield or , presumably due to a66 promoted aminolysis of the THP fragment in the presence of aniline hydrochloride. Alkylation of with suitable benzyl halides afforded – in good yields (, ). Yet, the high polarity imparted by the common diaminopyrimidine moiety shared by and – often led to the isolation of poorly separable residual starting material/alkylated product mixtures. We thus adapted a linear synthesis similar to the approach used to obtain which is exemplified by the synthesis of and as depicted in . The synthesized compounds were then evaluated for their inhibitory activity against TrkA, TrkB, TrkC and CSF-1R in comparison to GW2580 (). The resulting ICs along with relevant physico-chemical properties are summarized in (dose–response curves are presented in , ). Under assayed conditions, lead inhibitor was shown to display moderate intra Trk isoform selectivity compared to values reported with binding assays. Fluorine-for-methoxy substitution had a negligible impact on the potency towards TrkB (IC=119±38.7nM for versus IC=132±12.0nM for ). This highly potent fluorinated TrkB inhibitor also displayed similar potencies for TrkC (135±5.66nM) and CSF-1R (169±27.6nM) and slightly improved selectivity towards TrkA. Therefore, has a suitable affinity for PET imaging of tumors. Fluorinated compound also showed a reduced surface polar area (TSPA) and increased Log/Log compared to in a range which is favorable considering ideal physico-chemical properties for PET radiotracers. Inhibitor was 2–4.5-fold less potent towards Trk receptors and 28-fold less potent for CSF-1R as compared to . Interestingly, replacement of one hydrogen for a fluorine atom in -position in had a dramatic negative impact on the potency for all four targets despite being the smallest structural modification of all tested derivatives (>100-fold decrease in potency). As expected, derivative , lacking the tail benzyloxy fragment did not display kinase inhibition (). In order to rationalize the unexpected potency leap between and , a molecular modeling study was performed using the X-ray co-crystal structure of TrkB-GW2580 complex (PDB ID: ) and CSF-1R co-crystal complex (PDB ID: ) with FITTED (FORECASTER platform). The binding modes of compounds and in the ATP-binding cavity of TrkB (DFG-out), which as expected overlaid significantly with the resolved crystal structure of , are depicted in A and B. While key hydrogen bonds and hydrophobic interactions at the hinge and within the DFG motif do not differ between inhibitors, discrepancies occur at the hydrophobic back pocket regarding the spatial orientation of the tail benzyl moiety. The fluorobenzyl ring in displays the same perpendicular orientation relative to the central methoxybenzyl ring as the PMB ring in . This conformation is also favoured considering the lowest energy conformation in this motif and presumably allows for the optimal interaction with the hydrophobic pocket residues. In contrast, the -fluoro PMB fragment in is distorted from this conformation (C). The -hydrogen atoms (position 2- and 6-) in are positioned in close proximity (2.8Å) to the oxygen atom from the carbonyl groups from both sides, namely residues Asp710 and Val617—which is inferior to the sum of the van der Waals radii of fluorine and oxygen (2.99Å). Those unfavorable electrostatic interactions, exacerbated in as compared to when considering the longer C–F bond compared to C–H, results in the distorted and seemingly disfavored orientation of the tail group of (D and E)—similar interactions occur with CSF-1R (, ). Moreover, even in this conformation, the fluorine substituent is potentially forced to lie in the vicinity of Asp710 only, due to the overlapping proximity of the side chain from Val617 if oriented towards the back of the hydrophobic cavity. In addition, conformational factors involving intermolecular hydrogen bonding of the solvated ligand, reminiscent of the intramolecular interactions observed in -fluorobenzylic alcohol structures and -fluoro arylamides, may also contribute to the poor relative potency of through the stabilisation of suboptimal conformations for binding (, ).