As part of our discovery
As part of our discovery efforts searching for small molecule GPR119 agonists, we analyzed active pharmacophores of known agonists from the literature, and found that -methylsulfonylphenyl and substituted 4-hydroxyl piperidine are common fragments. In our initial efforts, we linked these two fragments via the 1,4-positions of a phenyl group and generated a set of compounds of general structure (). These compounds exhibited moderate GPR119 agonism activity in vitro. In an effort to improve potency, we focused on a strategy that conformationally constrained the molecule, exemplified by the spiro dihydrobenzofuran analogs . Unfortunately, several analogs of showed no GPR119 agonism activity. Conformational searches of several molecules were performed using MacroModel (Schrödinger, LLC, New York, NY, 2008) with the OPLS2005 force field, followed by energetic analysis of low energy structures using Jaguar DFT B3LYP/6-31G (Schrödinger, LLC, New York, NY, 2008). A low energy conformation of literature Nystatin APD597, was used as the basis for a pharmacophore and low energy conformations of our compounds were matched to the pharmacophore using Phase (Schrödinger, LLC, New York, NY, 2008). The analogs clearly demonstrated that compound was not only too short to match the desired pharmacophore of APD597, but also possessed a completely different piperidine ring orientation (). This led us to conceive a slightly longer version of the dihydrobenzofuran, , by simple insertion of a bond between dihydrobenzofuran and piperidine. This change simultaneously offers greater molecular length and the appropriate orientation of the piperidine ring via release of the spiro fusion. The modeling results suggest a better overlay of compound with APD597 than compound (). Therefore, a set of analogs of compound was synthesized. To our delight, moderate GPR119 activity was regained. In order to improve the in vitro activity of , the left-hand phenyl was replaced with piperidinyl or piperazinyl. To our surprise, such modifications resulted in significant improvement of in vitro potency, which led to the evolution of the dihydrobenzofuran series to the general structure (). We developed a general route to access dihydrobenzofuran analogs as described in . Selective lithiation at the α-position of the commercially available benzofuran was achieved using -BuLi at −78°C. The resulting benzofuran-2-yllithium reacted with -butyl 4-oxopiperidine-1-carboxylate to afford . Attempts to convert to using Pd catalyzed hydrogenation in a single step did not prove to be successful. Therefore, a two-step approach was adopted. Treatment of with EtSiH and TFA under reflux conditions resulted in the removal of the benzylic hydroxyl group and the deprotection of the Boc group, yielding . The amine functional group was then reprotected with Boc, and the subsequent Pd-catalyzed hydrogenation afforded . Selective bromination at the position - to the oxygen group of dihydrofuran afforded in high yield. The Boc-protecting group was removed under acidic conditions and the resulting amine was then reacted with various reagents (such as aldehydes, chloroformates, sulfonyl chlorides, and 2-chloropyrimidines) under various conditions to form compound . The synthesis of compound from compound was achieved in three steps following the standard reaction sequence of Suzuki coupling, deprotection, and sulfonamide formation. To explore the SAR of the nitrogen attachment in the right-hand piperidine (), the -propyl sulfonamide was used as the standard left-hand substitution. The basic amine has moderate activity (521nM), as do the carbamates ( and ) and the sulfonamide (, much weaker agonism activity with of 43%). Pyrimidine and pyridine analogs (–) have better in vitro potency, as indicated by their reasonable EC numbers and the overall higher . For example, 5-trifluoromethylpyrimidine () shows the best potency in vitro (14nM), but its extremely poor aqueous solubility precluded it from further testing. 5-Chloropyrimidine () and 5-(-propyl)pyrimidine () are equally potent (78 vs 82nM), but surprisingly, 5-(-propyl)pyrimidine () has better stability in both human and mouse liver microsomes. Besides poor rodent microsomal stability (only 3% remaining), the 5-cyclopropylpyrimidine analog (EC 67nM, 85%) also suffers from glutathione adduct formation due to the cyclopropyl ring opening, precluding it from further consideration.