Novel FFA1 (GPR40) agonists containing spirocyclic periphery: polar azine periphery as a driver of potency

Abstract A series of nine compounds based on 3-[4-(benzyloxy)phenyl]propanoic acid core containing a 1-oxa-9-azaspiro[5.5]undecane periphery was designed, synthesized and evaluated as free fatty acid 1 (FFA1 or GPR40) agonists. The spirocyclic appendages included in these compounds were inspired by LY2881835, Eli Lilly’s advanced drug candidate for type II diabetes mellitus that was in phase I clinical trials. These polar spirocyclic, fully saturated appendages (that are themselves uncharacteristic of the known FFA1 ligand space) were further decorated with diverse polar groups (such as basic heterocycles or secondary amides). To our surprise, while seven of nine compounds were found to be inactive (likely due to the decrease in lipophilicity, which is known to be detrimental to FFA1 ligand affinity), two compounds containing 2-pyridyloxy and 2-pyrimidinyloxy groups were found to have EC50 of 1.621 and 0.904 µM, respectively. This result is significant in the context of the worldwide quest for more polar FFA1 agonists, which would be devoid of liver toxicity effects earlier observed for a FFA1 agonist fasiglifam (TAk-875) in clinical studies.


Introduction
FFA1 or GPR40 is a cell surface G-protein coupled receptor expressed in pancreatic b-cells and is involved, through its activation, in regulation of insulin release and lowering of glucose levels 1 . The mechanism of glucose level regulation via FFA1 is quite unique as the therapeutic influence can only be exerted in hyperglycemic states when FFA1 expression levels are, in turn, upregulated. Once the normal glycemia is restored due to FFA1 activation, the FFA1 expression goes to basal levels and the insulin levels are no longer affected by the agonists still present in circulation. Therefore, therapeutic agents acting via this mechanism cannot cause hypoglycemia, a condition not less threatening than heightened glucose levels. Therefore, FFA1 agonists, if developed into clinically used drugs, would offer a much safer alternative to the currently available medicines for the treatment of type 2 diabetes mellitus (T2DM) 2 . Unfortunately, none of the compounds of this promising class have yet to claim their place in a clinical setting. The most advanced compound to-date, Takeda's fasiglifam (TAK-875) that had shown very promising efficacy results in phase II and III clinical trials, was stopped in development due to observable adverse liver toxicity in some patients. This has severely affected the field of FFA1 agonists, particularly from industry investment perspective 3 . At the time of writing this manuscript, only one clinical investigation of an FFA1 agonist was underway (Piramal's compound P11187 of unpublished structure; https://clinicaltrials.gov). One of the ways of restoring the dwindled promise of the new class of andidabetic compounds would be to keep in mind that the poof-of-concept was achieved for FFA1 inhibitors in the course of TAK-875 clinical research 4 and to tackle the toxicity profile of next-in-class compounds 5 .
One of the most promising strategies toward reducing liver toxicity profile of FFA1 is to reduce the overly lipophilic character of the advanced FFA1 agonists, including TAK-875 5 . On the one hand, lowering lipophilicity could negatively affect the potency profile of more polar compounds as the receptor has medium-to-long chain fatty acids as endogenous ligands and the majority of the known agonist are considered mimics of the latter. On the other hand, we have already demonstrated that decoration of the 3-phenylpropanoic acid core (which is additionally substituted with a p-benzyloxy substituent in many advanced FFA1 agonists, including TAK-875, Figure 1) with polar heterocyclic appendages 6 or replacing the phenyl ring in this core with heterocyclic motifs 7,8 can result in compounds having potency comparable to that of the most advanced compounds reported. In the more recent study, we continued exploring the former (polar periphery) approach and drew inspiration from Eli Lilly's compounds LY2881835 that contains spirocyclic tertiary amine periphery. We reasoned that if we simplify the pharmacophoric core to the unsubstituted 3-[4-(benzyloxy)phenyl]propanoic acid and decorate aldehyde ester building block 1 (which we had made synthetic available on multigram scale 9 ) with various 1-oxa-9-azaspiro [5.5]undecane building blocks 2 (which are available, in turn, via secondary alcohol manipulation of the earlier reported 10 N-Bocprotected 1-oxa-9-azaspiro [5.5]undecan-4-ol (3), this may provide a series of novel spirocyclic analogs of LY2881835 (4) some of which may be endowed with agonist activity toward FFA1 ( Figure 2). We have already found 11 that if the 1-oxa-9-azaspiro [5.5]undecane periphery is decorated with lipophilic periphery (R ¼ benzyl), the agonist potency falls in the nanomolar range (i.e. becomes comparable to that of TAK-875 that has EC 50 ¼ 0.014 lM 12 ). Such a result was predictable in a sense that lipophilicity is a known driver of potency of free fatty acid receptors 13 . It is also notable that the unsubstituted 1-oxa-9-azaspiro [5.5]undecane moiety (4, R ¼ H) prepared and tested by us earlier was found inactive 11 . In this study, we undertook decoration of the 1-oxa-9-azaspiro [5.5]undecane periphery with diverse polar R groups, including basic heterocyclic and secondary amide moieties. Herein, we disclose our positive findings in this area.

Materials and methods
Chemical synthesesgeneral All reactions were conducted in oven-dried glassware in atmosphere of nitrogen. Melting points were measured with a B€ uchi (Flawil, Switzerland) ffl-520 melting point apparatus and were not corrected. Analytical thin-layer chromatography was carried out on Silufol UV-254 silica gel plates using appropriate mixtures of ethyl acetate and hexane. Compounds were visualized with short-wavelength UV light. 1 H NMR and 13 C NMR spectra were recorded on Bruker MSL-300 spectrometers in DMSO-d 6 using TMS as an internal standard. Mass spectra were recorded using Shimadzu LCMS-2020 system (Kyoto, Japan) with electron impact (EI) ionization. All reagents and solvents were obtained from commercial sources and used without purification.
General procedure for the preparation of compounds 2a-d A solution of tert-butyl 1-oxa-9-azaspiro [5.5]undecan-4-ol (3, 4 g, 14.8 mmol) in DMF (20 mL) was added dropwise to a 0 C suspension of NaH (1.3 g, 32.6 mmol, 60% dispersion in mineral oil) in dry DMF (100 mL) under argon. The resulting mixture was stirred at 0 C for 30 min whereupon a solution of the respective heteroaryl halide (19.2 mmol)in DMF (10 mL) was added dropwise. The resulting mixture was allowed to warm up to r.t. and stirred at that temperature for 18 h. The reaction mixture was poured into water (200 mL) and the aqueous phase was extracted with ethyl acetate (3 Â 200 mL). The combined organic extracts were washed with 3% aqueous citric acid, 5% aqueous NaHCO 3 , brine and water, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was fractionated on silica gel using 0 ! 5% ethyl acetate in hexanes as an eluent and the fractions containing the desired product (according to LC-MS analysis) were combined and concentrated in dryness. Without further purification, the residue was dissolved in CH 2 Cl 2 (3 mL/mmol calculated assuming 100% purity of the material obtained in the previous step), the solution was cooled to 0 C and TFA (1 mL/mmol) was added. The mixture thus obtained was stirred at 0 C for 6 h and then concentrated to dryness to provide, after crystallization from isopropyl alcohol, the target spirocyclic piperidine as a trifluoroacetate salt. Compound 2d was converted to hydrochloride salts by treatment of their r.t. suspensions in 1,4-dioxane with 4 M HCl in 1,4-dioxane followed by stirring for 3 h, evaporation of the volatiles in vacuo and crystallization from isopropyl alcohol. General procedure for the preparation of compounds (2e-f) To a thoroughly stirred solution of 5 (2.0 g, 7.43 mmol) and respective secondary amine (0.58 g, 8.15 mmol) in CH 2 Cl 2 (50 mL) was added, in small portions, sodium triacetoxyborohydride (11.0 g, 52.0 mmol). The reaction mixture was stirred for 18 h, poured into sat. aq. K 2 CO 3 (100 mL) and extracted with CH 2 Cl 2 (3 Â 50 mL). The combined organic extracts were dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was fractionated on silica gel using 0 ! 10% methanol in CH 2 Cl 2 as an eluent. The fractions containing the target compound (according to LC-MS analysis) were pooled and concentrated in vacuo. The residue was dissolved in 1,4-dioxane (15 mL) and treated with 4 M HCl in 1,4-dioxane (1 mL). After stirring at r.t. for 6 h, the reaction mixture was concentrated to dryness in vacuo and the residue was crystallized from isopropyl alcohol to provide analytically pure title compound.

4-(Pyrazin
4-Pyrrolidin-1-yl-1-oxa-9-azaspiro [5.5] 2-(9-(Tert-butoxycarbonyl)-1-oxa-9-azaspiro[5.5]undecan-4yl)acetic acid (6) To a 0 C, vigorously stirred suspension of NaH (6.54 g, 163 mmol, 60% dispersion in mineral oil) in THF (300 mL) thriethylphosphonoacetate (45 g, 200 mmol) was added dropwise under argon. The stirring continued at that temperature for 1 h, whereupon a solution of 10 (40 g, 149 mmol) in THF (100 mL) was added. The reaction mixture was allowed to reach r.t. and was stirred at that temperature for 18 h. The reaction mixture was poured into water (500 mL) and the aqueous phase was extracted with ethyl acetate (3 Â 200 mL). The combined organic extracts were washed with 3% aqueous citric acid, 5% aqueous NaHCO 3 , brine, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was fractionated on silica gel using 0 ! 5% ethyl acetate in hexanes as an eluent. The fractions containing the olefination product (according to LC-MS analysis) were pooled and concentrated to dryness (yielding 32.7 g of the material). The residue (10.9 g) was dissolved in EtOH (200 mL), HCOONH 4 (2.8 g, 0.44 mmol) and 10% Pd on carbon (300 mg) were added and the resulting mixture was heated at reflux for 12 h. The mixture was cooled to r.t. and filtered through a plug of Celite (subsequently washing the latter with EtOH). The combined filtrate and washings were concentrated to dryness. The residue was partitioned between water (150 mL) and ethyl acetate (150 mL). The organic layer was separated and the aqueous layer was additionally extracted with ethyl acetate (2 Â 150 mL). The combined organic extracts were washed with 3% aqueous citric acid, 5% aqueous NaHCO 3 and brine, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was dissolved in MeOH (100 mL) and a solution of KOH (5.42 g, 96.7 mmol) in water (20 mL) was added. The mixture was stirred at r.t. for 18 h and concentrated to dryness in vacuo. The residue was dissolved in water (100 mL), the aqueous solution was extracted with ether (2 Â 50 mL) and then acidified to pH 5.0 with 3% aqueous HCl. The solution thus obtained was extracted with ethyl acetate (3 Â 100 mL) and the combined organic extracts were washed with brine, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo to provide analytically pure 6.
Yield 7.35 g (49%). 1  General procedure for the preparation of compounds 2g-i To a solution of [9-(tert-butoxycarbonyl)-1-oxa-9-azaspiro[5.5]undec-4-yl]acetic acid (6, 0.50 g, 1.59 mmol) in CH 2 Cl 2 (50 mL) carbonyldiimidazole (0.28 g, 1.71 mmol) was added in small portions. The mixture was stirred for 1 h, whereupon a solution of the respective amine (1.75 mmol) in CH 2 Cl 2 (10 mL) was added dropwise, and the stirring continued for 18 h. The mixture was poured into water (200 mL) and the slurry extracted with CH 2 Cl 2 (3 Â 200 mL). The combined organic extracts were washed with 1% aq. citric acid, 5% aq. NaHCO 3 , brine, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was fractionated on silica gel using 0 ! 10% methanol in CH 2 Cl 2 as an eluent. The fractions containing the target compound (according to LC-MS analysis) were pooled and concentrated in vacuo. The residue was dissolved in CH 2 Cl 2 (3 mL/mmol calculated assuming 100% purity of the material obtained in the previous step), the solution was cooled to 0 C and TFA (1 mL/mmol) was added. The mixture thus obtained was stirred at 0 C for 6 h and then concentrated to dryness to provide, after crystallization from isopropyl alcohol, the target spirocyclic piperidines 2g-i as trifluoroacetate salts. Compound 2g was converted to hydrochloride salts by treatment of their r.t. suspensions in 1,4-dioxane with 4 M HCl in 1,4-dioxane followed by stirring for 3 h, evaporation of the volatiles in vacuo and crystallization from isopropyl alcohol. General procedure for preparation of compounds 4a-i A solution of the respective spirocyclic piperidine salt 2a-i (0.46 mmol) in CH 2 Cl 2 (5 mL) was treated with trimethylamine (n Â 0.46 mmol, where n ¼ number of salt parts per molecule) followed by a solution of 1 11 (0.44 mmol) in CH 2 Cl 2 (5 mL). After a brief stirring (15 min), sodium triacetoxyboohydride (STAB, 1.32 mmol) was added and the stirring continued for 12 h at r.t. The reaction was poured into 10% aqueous NaHCO 3 (20 mL). Organic phase was separated and the aqueous phase was extracted with CH 2 Cl 2 (2 Â 10 mL). The combined organic extracts were washed with brine, dried over anhydrous Na 2 SO 4 , filtered and concentrated in vacuo. The residue was fractionated on silica gel using 0 ! 1% MeOH in CH 2 Cl 2 . The fractions containing the reductive amination product (according to LC-MS analysis) were pooled and concentrated in vacuo. The residue was dissolved in CH 2 Cl 2 (3 mL) and treated with TFA (1 mL). The mixture was stirred at r.t. for 18 h and concentrated in vacuo. Two molar HCl in ether (3 mL) was added to the residue and the later was triturated (with occasional sonication) until a crystalline hydrochloride salt formed. The latter was separated by filtration, washed with ether and dried in vacuo to provide analytically pure compounds 4a-i.  13  Determination of agonistic activity of compounds against FFA1 (GPR40) receptor CHO cells stably expressing human GPR40 (stable CHO-GPR40 line created at Enamine Ltd., Kyiv, Ukraine) were seeded (12 500 cells/ well) into 384-well black-wall, clear-bottom microtiter plates 24 h prior to assay. Cells were loaded for 1 h with fluorescent calcium dye (Fluo-8 Calcium Assay kit, Abcam, ab112129, Cambridge, UK) and tested using fluorometric imaging plate reader (FLIPR Tetra V R High Throughput Cellular Screening System, Molecular Devices Corp., Sunnyvale, CA). Maximum change in fluorescence over base line was used to determine agonist response. A potent and selective agonist for FFA1 (GPR40) GW9508 (Selleckchem, S8014) was tested with the test compounds as a positive control. Concentration response curve data were fitted using Molecular Devices ScreenWorks V R System Control Software (Molecular Devices). The half-maximal effective concentration was determined from these curves plotted in "% FFA1 activation À log[drug]" coordinates and % maximum efficacy was related to that of the reference compounds GW9508.

Results and discussion
The spirocyclic building blocks for subsequent use in decorating the pharmacophore core building block 1 10 were synthesized from common precursor 3, which we prepared on multigram scale as described earlier 10 . Sodium alkoxide generated from 3 on treatment with NaH was an effective nucleophile in S N Ar-type reaction with heteroaryl halides furnishing, after Boc group removal, building blocks 2a-d. The secondary alcohol functionality in 3 underwent a facile oxidation with PDC to furnish a good yield of respective ketone 5. The carbonyl group in 5 was reductively aminated with pyrrolidine and morpholine in the presence of sodium triacetoxyborohydride (STAB) and gave, after Boc group removal, building blocks 2e-f. The same keto group a competent partner in Horner-Wadsworth-Emmons olefination, which led, after hydrogenation of the resulting olefin and ethyl ester hydrolysis, to carboxylic acid 6. The latter was a common precursor to amides 2g-i obtained via a standard CDI-promoted amidation followed by Boc group removal (Scheme 1). Spirocyclic piperidine building blocks 2a-i were used in the reductive amination reaction of aldehyde 1 in presence of STAB and final compounds 4a-i were obtained after a facile tert-butyl ester hydrolysis on treatment with TFA followed by salt form exchange, see Material and methods (Scheme 2).
Compounds 4a-i were preliminarily tested at 5 lM concentration for activation of FFA1. Only two compounds (4b and c) demonstrated >50% activation of the receptor at that concentration.
To see how the single-concentration data translate into EC 50 values, compounds 4a-i were also tested in dose-response (% FFA1 activation) mode. As can be seen from the data thus obtained (Table 1), the % activation data (5 lM) were quite predictive of the compounds' ability to activate FFA1 in a broad concentration range as the only meaningful EC 50 values were determined for the same two compounds 4b and c, while the rest of compounds were virtually inactive.
These results are not unexpected as the majority of the compounds containing polar appendages are perhaps too polar to mimic the endogenous ligands of FFA1, that is medium-to-long chain fatty acids. The low-micromolar and even submicromolar EC 50 values obtained for compounds 4b and c are, therefore, very surprising and encouraging. From examination of the superimposed cLogP and FFA1 potency data presented in Table 2 it becomes clear that the unexpectedly high agonist activity of compounds 4b and c is not associated with a lipophilicity increase (a known FFA1 potency driver 13 ) and is likely due to specific interactions of the ligands with the protein, due to the presence of the azin-2-yloxy moieties. We have shown earlier 11 that grafting a benzyl group only a spirocyclic scaffolds of the inactive compound 7 resulted in a highly potent compound 8. While such a potency boost could be rationalized by the nearly 100-fold increase in lipophilicity (as gauged by cLogP values), we demonstrate d 6 2 that the benzyl group in 8 also forms a network of hydrophobic and p-stacking interactions with the target. Compounds 4b and c could, in principle, be considered as more polar isosteres of compound 8 and the observed potency of the former very likely results from similar additional contacts with the receptor. Lipophilicity is clearly not responsible for the observed SAR as the cLogP values of 4b and c are comparable and even lower than the same value for inactive compound 7 ( Table 2).

Conclusion
Highly lipophilic character of the known advanced FFA1 agonists limited their progression through clinical development. In this study, we achieved a significant result in designing new FFA1 agonists, namely, the polar-appendage versions of spirocyclic 1-oxa-9azaspiro [5.5]undecanes, which activated FFA1 in low-micromolar and submicromolar range. This finding significantly broadens the chemistry space and medicinal chemistry freedom-to-operate in today's worldwide quest for more polar and potentially less toxic FFA1 agonists as a fundamentally novel type of therapeutic agents to treat T2DM.