Gemcitabine receptor As part of a program to

As part of a program to explore the key binding interactions between the core of 1 and DGAT-1 a systematic analysis of the aminopyrimidine subunit was initiated. A potential approach to improving passive permeability within this series would be to increase lipophilicity via incorporation of substituents in the 3-position of the pyrimidine ring. Towards this end a 3-methyl analog 5 was prepared and shown to be a potent inhibitor of DGAT-1 activity and triglyceride synthesis in HT-29 Gemcitabine receptor (Table 1). While low projected oxidative clearance was observed for this analog, the reduction in potency relative to 1 suggested that further efforts based on modification of the 3-position were unlikely to provide an improved balance of potency, permeability and microsomal stability. An alternative approach to improving permeability was to alter the amino group of 1 by replacement with less polar isosteric replacements or masking as a secondary/tertiary amine, which would result in reduced polar surface area (PSA). The corresponding hydroxy (6) or methoxy (7) analogs were devoid of DGAT-1 inhibitory activity. Conversion of the primary amine in 1 into the N-methyl (8) or N,N-dimethylamines (9) also led to a loss of activity. These results demonstrate that the 4-amino substituent is critical for DGAT-1 inhibition. With a reduced PSA relative to 1, the regioisomeric aminopyridines were prepared to determine if they offered an advantage. 2-Aminopyridine 10 had substantially reduced potency, though the 4-amino analog 11 was determined to have an IC50 value of 23 nM against hDGAT-1 and inhibited triglyceride synthesis with an IC50 value of 99nM. While 11 exhibited a projected clearance profile comparable to 1, its passive permeability (Papp=1×10−6cm/s) is not improved relative to the lead. The structural core of 1 is similar to that in the cannabinoid receptor-1 receptor antagonist 15, which showed a substantial erosion of the therapeutic index in canine safety studies. Based on mechanism work it was hypothesized that formation of 16 was responsible for the observed toxicology (Scheme 1). Based on the structural overlap between 15 and 1 a two-pronged strategy to address this potential reactive metabolite issue was pursued. The first of these was a design approach based on steric blocking of the ether bearing carbon in 1 was pursued. If tolerated from a potency standpoint, incorporation of lipophilicity at this position would also have the potential to improve passive permeability. Substituents in this position would roughly overlap space occupied by the methyl groups on the quaternary carbon of 2. The direct homolog 12 was less active than the 1, however, preparation of the homochiral monomethyl derivatives revealed that each of these methyl groups was having a differential impact on DGAT-1 inhibitory activity. S-Isomer 13 had substantially reduced activity (IC50=625 nM), while its enantiomer 13 was equipotent to 1. These results are consistent with the good overlap of the methyl group of 13 with the pro-R methyl of 2 while the methyl of 12 occupies novel space, likely resulting in a negative steric interaction with DGAT-1. Monomethyl derivative 14 would be expected to dramatically reduce the potential for metabolic activation of the substituted carbon center, as well as any metabolic activation leading to a much less reactive keto-metabolite. In parallel to this design effort, a more detailed evaluation of the metabolic fate of 1 in human microsomal preparations was initiated. Acid 1 is very stable in human liver microsomes, with 93% of parent remaining after 1h. Of the portion that is turned over there was no evidence of the putative aldehyde metabolite being formed based on trapping experiments with methoxymethylamine. This result gave us confidence in advancing 1 to human clinical studies. With the abosve studies providing confidence that the pyrimidooxazepinone core of 1 represented a highly optimized scaffold, attention turned to expanding the SAR for the phenylcyclohexylacetic acid sidechain. With 1 possessing a lipophilic carboxylic acid, it is reasonable to propose a binding site overlap with that occupied by the acyl CoA substrate in DGAT-1. However, kinetic analysis of inhibition of hDGAT-1 utilizing decanoyl CoA as the acyl donor revealed 1 to be a noncompetitive inhibitor with respect to this substrate (R2=0.98). As for efforts on the bicyclic core, a key goal was to drive designs into physiochemical space that would be predicted to lead to improved passive permeability. In order to gain an understanding of the minimum pharmacophore required to maintain potency in this system a series of truncated analogs of 1 were prepared. Both the parent bicyclic heterocycle 17 and N-methyl derivative 18 exhibited weak inhibition at 30μM (Table 2). Reestablishment of the N-aryl motif afforded substantial improvements in DGAT-1 inhibitory activity, with 4-alkyl functionalized analogs 19 and 20 possessing ligand efficiencies modestly improved relative to 1. Attempts to incorporate ionizable/polar functionality in the vicinity of the aryl ring (21) resulted in loss of DGAT-1 inhibitory activity. In order to understand whether general lipophilicity in this region of the inhibitor structure was sufficient for potency, a series of N-cycloalkyl analogs were evaluated. Cyclopentyl derivative 22 was found to have a reduced potency and ligand efficiency relative to neutral N-aryl analogs. For a more direct comparison the aromatic ring saturated analogs of 20 were prepared. The more active N-4-t-butylcyclohexyl isomer 24 was 20-fold less active than the corresponding aryl homolog. These results support a conclusion that the N-aryl motif present in 1 is optimal for DGAT-1 inhibitory activity.