Novel Series of Potent Glucokinase Activators Leading to the Discovery of AM-2394

Abstract

Glucokinase (GK) catalyzes the phosphorylation of glucose to glucose-6-phosphate. We present the structure–activity relationships leading to the discovery of AM-2394, a structurally distinct GKA. AM-2394 activates GK with an EC₅₀ of 60 nM, increases the affinity of GK for glucose by approximately 10-fold, exhibits moderate clearance and good oral bioavailability in multiple animal models, and lowers glucose excursion following an oral glucose tolerance test in an ob/ob mouse model of diabetes.

Type 2 diabetes mellitus (T2DM) is a disease characterized by elevated plasma glucose in the presence of insulin resistance and inadequate insulin secretion. Glucokinase (GK), a member of the hexokinase enzyme family, catalyzes the phosphorylation of glucose to glucose-6-phosphate in the presence of ATP.¹ Expression of GK is restricted primarily to hepatocytes, pancreatic α- and β-cells, enteroendocrine cells, and the hypothalamus. GK exhibits low substrate affinity (S_0.5 or K_m = 7.5 mM), positive substrate cooperativity (Hill slope h = 1.7), and a lack of product inhibition, characteristics that allow it to serve as a physiologic glucose sensor and tightly regulate plasma glucose levels.^2,3 Small molecule activators of GK activity (glucokinase activators or GKAs) were first described in 2003⁴ and proposed as potential therapeutic agents for treating diabetes.⁵ In fact, several compounds have entered clinical trials, although the potential for hypoglycemia, elevated plasma triglycerides and blood pressure, and lack of durable response remain as considerable obstacles to this therapeutic class.⁶⁻¹⁰

Glucokinase activators are allosteric binders that can increase the enzyme’s affinity for glucose (S_0.5) and/or the maximal velocity or number of glucose molecules phosphorylated per unit time, expressed as V_max. Physiologically, these parameters impact the degree to which the activated enzyme phosphorylates glucose at a given blood glucose concentration and reduces blood glucose levels compared to the nonactivated enzyme.

In this report we describe efforts toward the discovery of 4,5-substituted-2-pyridyl ureas as GKAs. Two key assays were used to characterize compounds as GKAs and establish the structure–activity relationships (SAR).

The first assay determined the EC₅₀ of the GKA in the presence of 5 mM glucose. In addition, this assay was also run in the presence of 4% human serum albumin (HSA) as a surrogate measure of plasma protein binding. The second assay generated S_0.5 and V_max values.¹¹ Our goals were to maintain the S_0.5 between 0.6 and 1.0 mM to balance efficacy with risk of hypoglycemia, and V_max close to 100%, but no less than 90%, to protect against potential of the GKA inhibiting enzyme activity at high, but still physiologic, glucose concentrations, a possibility in severely diabetic patients.

We began by modeling possible structural replacements of the amido-heterocycle found in a number of GKAs¹²⁻¹⁵ and determined that a 2-pyridyl urea could be a suitable replacement, particularly due the potential to maintain the key bidentate hydrogen bonding interaction to Arg63, which is observed in all GKAs (Figure 1). X-ray crystal structure of compound 5 confirmed the formation of the hydrogen bonds, and thus, we focused our efforts on this series.

Figure 1.

X-ray cocrystal structure of 5 with glucokinase.

Attachment of a phenoxy substituent at position C-4, resulted in compound 1 (see Table 1). Addition of fluorine at position 2 of the phenylether resulted in a 5-fold increase in potency, hypothesized in part to result from an improved π stacking of the aryl ring and Tyr214 (vide infra). Further installation of a chlorine atom at position 3 (3) led to a significant increase in potency but, in general resulted in increased metabolism in both rat and human liver microsomes as shown in Table 1. However, the 2,6-difluoro substitution pattern (4) improved the potency relative to 2 and reduced the clearance in human liver microsomes compared to 3.

Table 1. Initial C-4 Substituent SAR of 1-methyl-3-(pyridine-2-yl)-urea^a,b,c.

^aMean standard deviation for the EC₅₀, S_0.5, and V_max are ±30%, 17%, and 4%, respectively (n = 2).

^bFor the EC₅₀ assay, the glucose was fixed at 5 mM.

^cMaximal velocity of the glucose phosphorylation reaction catalyzed by activated GK/unactivated GK.

It was hypothesized that introduction of substituents at the C-5 position based upon the crystal structure would lead to improvements in potency (Table 2).

Table 2. SAR of C-5 Substitution of Pyridyl Urea GKAs^a,b,c.

^aMean standard deviation for the EC₅₀, S_0.5, and V_max are ±30%, 17%, and 4%, respectively (n = 2).

^bFor the EC₅₀ assay, the glucose was fixed at 5 mM.

^cMaximal velocity of the glucose phosphorylation reaction catalyzed by activated GK/unactivated GK.

Replacement of the hydrogen with a methyl substituent (6) resulted in a large increase in potency and improvement in V_max. Switching the methyl group to a cyclopropyl group (7) led to a further 3-fold improvement in potency with desired increase in V_max. Switching to a phenyl ring (8) was tolerated with little effect on potency; however, an undesired drop in V_max was observed. Introduction of a 2-pyridyl group, in the case of 9, resulted in a 10-fold loss in potency. The potency was restored when a 4-pyridyl group was incorporated (10), but the V_max of 88% was still lower than desired. The 3-pyridyl substituent (11) exhibited a 4-fold loss in potency compared to the phenyl substituent (8), but more importantly the V_max increased from 79% to 95%.

In general, introduction of a pyridyl ring helped improve the intrinsic clearance of the molecule. For instance, comparing 8 to 10, reductions in both plasma protein-shift (from 50-fold to 10-fold) and clearance in rat liver microsomes (from 355 μL/(min × mg) to 35 μL/(min × mg)) were achieved by introduction of a pyridyl ring at the C-5 position. Addition of a methoxy group to the pyridyl ring, exemplified by 12, further increased the potency and improved the S_0.5 in comparison to 11. However, despite an attractive biochemical profile, 12 had poor solubility (27 μM in SIF; pH 6.8) and moderate clearance in rat microsomes (116 μL/(min × mg)).

With compounds displaying good in vitro potency and kinetics, we returned to addressing the in vitro rat metabolism. To this end, we found that replacement of the methoxy group with an ethoxy group (13) was tolerated and improved the rat microsomal clearance, however, with a substantial loss in potency in our shift assay. To address this loss in potency, the cLogP was reduced through installation of the hydroxyethyl group, 14. This achieved a reduction in cLogP from 3.9 (13) to 2.5, with only a moderate loss of potency and change in kinetic profile, while maintaining low rat liver microsomal clearance (human liver microsomal clearance for compounds containing the hydroxyethyl group were in general consider low; <10 μL/(min × mg)). However, 14 still displayed a 7-fold shift in potency in the presence of 4% HSA (EC₅₀ = 0.65 μM) and exhibited no improvement in solubility over 12 (5 μM in SIF; pH 6.8) and 13 (<1 μM in SIF; pH 6.8), respectively.

Efforts to improve the overall free fraction of our molecules were investigated using a 4% HSA assay. We focused on replacements of the 2,6-difluoro phenyl ring (Table 3). Initial introduction of a 2-pyridyl or 4-pyridyl displayed a large loss in potency (15 and 16, EC₅₀ = 6.25 and >50 μM, respectively). Although the 3-pyridyl 17 also exhibited a loss of potency, it was only 5-fold (EC₅₀ = 0.5 μM), but more importantly, there was only a 2-fold shift in potency in the presence of 4% HSA. From the X-ray structure (Figure 2), it appeared that a slight increase in steric bulk and lipophilicity of R₁ would be tolerated. Further SAR of the 3-pyridyl group led to the finding that, in general, addition of a methyl group around the pyridyl ring led to only minor improvements in potency (18, 19, and 20).

Table 3. Further SAR of C-4-Position of the Pyridyl Urea^a,b,c.

^aMean standard deviation for the EC₅₀, S_0.5, and V_max are ±30%, 17%, and 4%, respectively (n = 2).

^bFor the EC₅₀ assay, the glucose was fixed at 5 mM.

^cMaximal velocity of the glucose phosphorylation reaction catalyzed by activated GK/unactivated GK.

Figure 2.

X-ray crystallographic structure of 14 bound to glucokinase. Key hydrogen bonding between the urea and Arg-63 and the π-stacking between the difluorophenoxy group and Tyr-214 are shown. The substituted pyridine occupies an induced pocket with the hydroxyl group in close proximity to Ser-69 and Trp-99.

However, the 5-methyl-3-pyridyl analogue 21 displayed a 10-fold increase in potency. Importantly, 21 had only a 1.6-fold shift presence of 4% HSA. Switching from methyl (21) to ethyl (22) or methoxy (23) resulted in very little change in potency and kinetics. A further increase in size to an ethoxy substituent (24) was not tolerated, resulting in a large loss in potency.

Unfortunately, despite attractive biochemical properties, 21 exhibited poor pharmacokinetic properties in mice (%F = 6) and poor solubility (<1 μM) in SIF (pH 6.8). The free hydroxy was targeted as an area to attenuate these properties, and the results are described in Table 4. Synthesis of a secondary alcohol after chiral separation resulted in both 25 and 26. Although both compounds had similar potency to 21, they also had an undesired decrease in S_0.5. Because the chiral center appeared to have no effect on potency, it was replaced with a geminal dimethyl substituent, leading to tertiary alcohol 27 (AM-2394), which had excellent biochemical properties (4% HSA EC₅₀ = 0.22 μM), similar to that of 21 and displayed improved solubility and pharmacokinetic properties (see Table 5).

Table 4. SAR of the Ethoxy Alcohol^a,b,c.

^aMean standard deviation for the EC₅₀, S_0.5, and V_max are ±30%, 17%, and 4%, respectively (n = 2).

^bFor the EC₅₀ assay, the glucose was fixed at 5 mM.

^cMaximal velocity of the glucose phosphorylation reaction catalyzed by activated GK/unactivated GK.

Table 5. Pharmacokinetic Properties of 27 (AM-2394)^a.

species	oral Cmax (μM)	tmax (h)	Cl (L/h/kg)	Vdss (L/kg)	iv t1/2 (h)	F (%)	Fub
mouse	0.35	0.33	0.37	0.59	2.0	25	0.13
rat	0.49	0.83	1.64	2.18	1.4	60	0.13
cyno	1.15	1.67	0.60	2.29	6.7	70	n/a
dog	3.52	1.33	0.25	1.49	6.5	99	n/a

^aAdministered at a dose of 1 mg/kg, iv; 5 mg/kg, PO, in mice. Administered at a dose of 0.5 mg/kg, iv; 2 mg/kg, PO, in rats, cynomolgous monkeys, and dogs. Data are expressed as mean values (n = 3). Vehicle adminstration; found in the Supporting Information.

Importantly, 27 (AM-2394) exhibits good-to-moderate cross species plasma clearance, volume of distribution, and oral bioavailability, allowing for further evaluation in animal models (Table 4).

To supply material for further evaluation, a general scalable route to deliver all 4,5-substituted-2-pyridyl ureas and more specifically 27 (AM-2394) was devised (see Scheme 1). Under palladium catalysis, 2-chloro-4-fluoro pyridine (32) was coupled with methyl urea followed by bromination to deliver bromide 34.

Scheme 1. Synthesis of AM-2394.

Reagents and conditions: (a) 1.5 equiv of ethyl glycolate, 1.5 equiv of NaH (60%), DMF, 23 °C, 99%; (b) 2.2 equiv of MeMgBr, THF, 23 °C, 93%; (c) 1.5 equiv of bis(pinacolato)diboron, PdCl₂(dppf)·DCM, 3 equiv of KOAc, 1,4-dioxane, 100 °C, 84%; (d) methyl urea, Xantphos, Pd₂(dba)₃, K₃PO₄, DME, 95 °C, 51%; (e) NBS, DMF, 73%; (f) 5-methylpyridin-3-ol, Cs₂CO₃, DMF, 80 °C, 98%; (g) 31, Pd₂(dba)₃, PCy₃, K₃PO₄, 1,4-dioxane, 100 °C, 65%.

The reactivity and electronics of 34 are such that a S_NAr reaction with 3-hydroxy-5-methylpyridine, followed by a Suzuki coupling of boronate ester 31 (synthesized through a S_NAr of 2-fluoro-5-bromo pyridine 28 with ethyl glycolate, followed by Grignard addition to the ester to install the gem dimethyl group and palladium mediated borate ester formation) delivered 27 (AM-2394) in good efficiency and overall yield.

In order to determine the effect of 27 (AM-2394) in an animal model of type 2 diabetes, it was administered per os (PO) to male ob/ob mice 30 min prior to performing an oral glucose tolerance test (OGTT). Doses of 1, 3, 10, and 30 mg/kg each reduced glucose excursion, with maximal efficacy seen at 3 mg/kg (Figure 3).¹⁶

Figure 3.

Oral glucose tolerance test in ob/ob mice with 27 (AM-2394). Statistical significance compared to vehicle treatment (0.5% methylcellulose, 1% tween, pH 2 with MSA) is denoted by *(p < 0.05), **(p < 0.01), ***(p < 0.001), and ****(p < 0.0001), as determined by two-way ANOVA, and is color coded to the treatment in the figure legends.

In conclusion, we have described the SAR leading to the discovery of 27 (AM-2394), a structurally distinct glucokinase activator that displays a robust reduction in plasma glucose during an OGTT in ob/ob mice at a dose of 3 mg/kg.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.6b00140.

• Experimental procedures and analytical data (PDF)

Author Present Address

^∥ (T.D.A.) Lycera Corp., 2800 Plymouth Road, NCRC, Building 26, Ann Arbor, Michigan 48109, United States.

Author Present Address

^⊥ (W.V.) Alexion Pharmaceuticals 352 Knotter Drive, Cheshire, Connecticut 06410, United States.

Author Present Address

^# (K.R.C.) Celgene Corporation, 10300 Campus Point Drive, Suite 100, San Diego, California 92121, United States.

Author Present Address

^∇ (S.A.B.) 30 Spring Mill Drive, Malvern, Pennsylvania 19355, United States.

Author Present Address

^○ (X.D. and J.M.) ORIC Pharmaceuticals, 407 Cabot Road, South San Francisco, California 94080, United States.

Author Present Address

^◆ (E.T.) Biogen, 225 Binney Street, Cambridge, Massachusetts 02142, United States.

The authors declare no competing financial interest.