In anticipation of an approach
In anticipation of an approach based on modification of the cyclohexylacetic Cholera Toxin subunit of 1 as a means of improving permeability, a set of analogs were prepared to map out the optimum display of the acidic functionality. Homologated acids 25 and 26 were found to be equipotent to 1, although they had substantially reduced whole cell activities (Table 3). Tetrazole 27 was also a potent inhibitor of DGAT-1 but was right-shifted around 10-fold in the cell assay, due to poor permeability (Papp<0.1×10−6cm/s) resulting from increased PSA. These results led us to retain the core structure present in 1 in efforts to optimize permeability. From the very early stages of investigations on 1, the potent DGAT-1 inhibitory activity of neutral analogs such as methyl ester 28 and primary amide 29 demonstrated that an acidic functionality in the cyclohexylphenyl pharmacophore was not obligate for activity. Because 29 had improved passive permeability relative to 1 it was advanced to a rat pharmacokinetics. Though it had low/no detectable turnover in rat liver microsomes, 29 did have moderate clearance in the rat (Table 4). This enhanced turnover in vivo was consistent with the observation that 29 was unstable in plasma, presumably through conversion to 1 by plasma amidases/esterases. To more fully explore the opportunity in this chemical space a series of compounds in which the carboxylic acid was replaced with neutral moieties were pursued. A number of design criteria were employed in selection of targets, including calculated properties cutoffs for polar surface area (PSA <140), molecular weight (<500), predictive models of human microsomal clearance (<30mL/min/kg) and passive permeability (Papp >5×10−6cm/s). Based on the excellent DGAT-1 inhibitory profile of primary amide 29, a library of substituted amides was pursued. The primary goals for this set of compounds were to improve passive permeability relative to 1, while also increasing in vivo half-life relative to 29. Representative examples detailed in Table 3 (30–38) show that a range of secondary/tertiary amides with or without polar functionalities retain good to excellent levels of DGAT-1 enzyme and cell inhibitory activities. While introduction of modest lipophilicity (31/33) resulted in good passive permeability, there was an associated increase in oxidative clearance. Attempts to rebalance these two properties through incorporation of additional polarity was challenging given the high PSA starting point of the core structure. Incorporation of hydroxyl (32/34/38) or amide (36) functionalities led to enhanced microsomal stability, but decreased permeability and DGAT-1 potency. These results led us to focus efforts on amides with a PSA value intermediate between 33 and 34, such as the morpholine and 4-methoxypiperidine amides, 35 and 37. While the latter of these had high human microsomal clearance, 35 had a good balance of DGAT-1 activity, low clearance and good permeability. Rat pharmacokinetic analysis revealed this amide to have a half-life reduced relative to 1 and primary amide 29. During the course of this work both 29 and 35 were found to be positive in an in vitro micronucleus assay performed in the absence of metabolic activation. Based on these results efforts were shifted away from this series of amides. In parallel with the above efforts, a library of amides in which the cyclohexylacetic acid moiety was replaced with a N-acylpiperizine were evaluated. Data for representative members of this library are detailed in Table 5. Compounds possessing small N-acyl substituents (e.g., 39–41) had modest DGAT-1 inhibitory activity. Attempts to increase potency through the incorporation of more lipophilic substituents, such as in 42, led to modest improvements in potency, but were accompanied by significant metabolic instability. Attenuating this lipophilicity through addition of polarity (43 and 44) were successful in reducing clearance potential (HLM <8mL/min/kg), but resulted in poor passive permeability. A set of five-membered heterocycle-based amides were included in the initial library and of these oxadiazole 45 was identified as a potent inhibitor of DGAT-1 (IC50=7nM). While this analog was 10-fold more potent than benzoyl derivative 46, it suffered from a ∼1000-fold disconnect with respect to inhibition of triglyceride synthesis in HT-29 cells. This poor cellular activity was likely the result of poor cellular penetration driven by the high PSA of 46. Based on their predicted improved PSA and passive permeability a series of related oxazole and isoxazoles were prepared. While this approach did restore high passive permeability (e.g., 47 and 48), all of these analogs had reduced potency and microsomal stability. These results suggest that it would be very difficult to generate a balanced pharmacology/ADME profile within this series.