Accepted Manuscript Stereoselective Synthesis, Biological Evaluation, and Modeling of Novel Bile Acid-Derived G-Protein Coupled Bile Acid Receptor 1 (GP-BAR1, TGR5) Agonists Donna D. Yu, Kyle M. Sousa, Daniell L. Mattern, Jeffrey Wagner, Xianghui Fu, Nagarajan Vaidehi, Barry M. Forman, Wendong Huang PII: DOI: Reference:

S0968-0896(15)00072-3 http://dx.doi.org/10.1016/j.bmc.2015.01.048 BMC 12051

To appear in:

Bioorganic & Medicinal Chemistry

Received Date: Revised Date: Accepted Date:

1 October 2014 19 January 2015 27 January 2015

Please cite this article as: Yu, D.D., Sousa, K.M., Mattern, D.L., Wagner, J., Fu, X., Vaidehi, N., Forman, B.M., Huang, W., Stereoselective Synthesis, Biological Evaluation, and Modeling of Novel Bile Acid-Derived G-Protein Coupled Bile Acid Receptor 1 (GP-BAR1, TGR5) Agonists, Bioorganic & Medicinal Chemistry (2015), doi: http:// dx.doi.org/10.1016/j.bmc.2015.01.048

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Stereoselective Synthesis, Biological Evaluation, and Modeling of Novel Bile Acid-Derived G-Protein Coupled Bile Acid Receptor 1 (GP-BAR1, TGR5) Agonists

Donna D. Yu,a* Kyle M. Sousa,a,d* Daniell L. Mattern,c Jeffrey Wagner,b Xianghui Fu,a Nagarajan Vaidehi,b Barry M. Forman, a § and Wendong Huang a §* a

Department of Diabetes and Metabolic Diseases Research, b Department of Immunology, The

Beckman Research Institute, City of Hope National Medical Center, Duarte, California 91010, USA. c

Department of Chemistry and Biochemistry, The University of Mississippi, University, MS

38677, USA. d

Department of Pharmaceutical Sciences, West Coast University School of Pharmacy, Los

Angeles, California 90004

AUTHOR INFORMATION *Corresponding Authors *D. D. Yu, phone, (626)-359-8111 x 65993, Fax (626)256-8704, E-mail: [email protected] Present Addresses Department of Diabetes and Metabolic Diseases Research, The Beckman Research Institute at The City of Hope National Medical Center, Duarte, California 91010, USA. *K. M. Sousa, phone (818)-232-4117, Fax (323)-661-0935, E-mail: [email protected] *W. Huang, phone, (626)-359-8111 x 65203, Fax (626)256-8704, E-mail: [email protected] §

These authors (Barry M. Forman and Wendong Huang) contributed equally.

1

ABSTRACT GP-BAR1 (also known as TGR5), a novel G-protein coupled receptor regulating various nongenomic functions via bile acid signaling, has emerged as a promising target for metabolic disorders, including obesity and type II diabetes. However, given that many bile acids (BAs) are poorly tolerated for systemic therapeutic use, there is significant need to develop GP-BAR1 agonists with improved potency and specificity and there also is significant impetus to develop a stereoselective synthetic methodology for GP-BAR1 agonists. Here, we report the development of highly stereo-controlled strategies to investigate a series of naturally occurring bile acid derivatives with markedly enhanced GP-BAR1 activity. These novel GP-BAR1 agonists are evaluated in vitro using luciferase-based reporter and cAMP assays to elucidate their biological properties. In vivo studies revealed that the GP-BAR1 agonist 23(S)-m-LCA increased intestinal GLP-1 transcripts by 26-fold. Additionally, computational modeling studies of selected ligands that exhibit enhanced potency and specificity for GP-BAR1 provide information on potential binding sites for these ligands in GP-BAR1. 1. Introduction Diabetes is a burgeoning, worldwide health problem, affecting almost twenty-six million people in the United States alone, with obesity-associated type II diabetes (T2D) accounting for 95% of all diabetes cases.1 The growing incidence and cost of this disease and its complications have driven intense efforts in generating novel pharmacological agents to treat and/or prevent this epidemic. One approach to improve existing therapies involves simultaneously targeting multiple pathways implicated in obesity and its associated co-morbidities. Recent discoveries have demonstrated that bile acid signaling may offer a viable therapeutic option for the treatment of metabolic disorders.2 To date, two bile acid-activated receptors have 2

been identified and are attractive targets for drug discovery efforts: the nuclear Farnesoid X Receptor (FXR), 3 and more recently, G-protein-coupled bile acid receptor 1, (GP-BAR1, or TGR5).4-5 GP-BAR1 (TGR5) is a cell surface receptor and is expressed in monocytes, gall bladder, brown adipose tissue, muscle, liver, and intestine, where its activation by bile acids triggers an increase in energy expenditure and attenuates diet-induced obesity.6 Through activation of the FXR and GP-BAR1 pathways, bile acids can regulate their own metabolism. However, unlike FXR, which mediates downstream activation via a genomic pathway, activation of GP-BAR1 constitutes a gateway to the non-genomic functions of bile acids. Interestingly, the activation of GP-BAR1 by bile acids, such as lithocholic acid (LCA, 1, Figure 1) and chenodeoxycholic acid (CDCA, 2), has profound effects on glucose homeostasis and insulin sensitivity, suggesting that treatment of T2D and obesity could be improved by targeting GP-BAR1 with potent and well-tolerated bile acid analogs that would more effectively modulate GP-BAR1 activity in vivo.

3

21

20 22 23

12

16 3

COOH

7

HO

COOH

HO

6

1 LCA

OH

COOH

OH

HO

3 UDCA

2 CDCA

HO COOH

COOH

Cl

OH OH

HO

HO

4 INT-777

O N O

OH

5 6e-16-epi-avicholic acid

N

Cl

6 GSK lead HO

H N N H2N

COOH

OCF3

N

COOH

N O

O

7 Pfizer lead

HO

OH

8 23(S)-Me-CDCA

HO

OH

9 23(S)-Me-CA

Figure 1. Chemical structures of some natural bile acids and previously disclosed GP-BAR1 synthetic agonists.

Naturally-occurring bile acids are physiological ligands for GP-BAR1.6 However, from a pharmacological perspective; endogenous bile acids are weak GP-BAR1 ligands in terms of both potency and specificity. Notably, bile acids not only activate GP-BAR1, but also trigger activation of FXR, providing a significant challenge for identifying GP-BAR1 specific agonists. Although the search for potent, selective GP-BAR1 agonists has intensified, only a few novel agonists have been identified (Figure 1). These include semi-synthetic bile acid derivatives, exemplified by INT-777 (4), which activates GP-BAR1 both in vitro and in vivo.7 Importantly, by its activation of GP-BAR1, INT-777 is able to stimulate type 2 iodothyronine deiodinase (D2) 4

activity in brown adipose tissue (BAT) and muscle, as well as induce the release of glucagon-like protein 1 (GLP-1) in enteroendocrine cells.8 These results provide the first proof-of-concept that modulating GP-BAR1 may provide a novel, viable strategy for the treatment of metabolic disorders such as T2D and obesity. Further chemical modifications of natural bile acids resulted in a new avicholic acid derivative, 6α-ethyl-16-epi-avicholic acid (5), that exhibited enhanced potency for both the GP-BAR1 and FXR receptors.9 The identification of avicholic acid derivatives suggests that chemical modification of natural bile acids can provide additional lead compounds for GP-BAR1 modulation. In addition to bile acid-derived agonists, several non-bile acid agonists have recently been reported that offer potential for higher potency and specificity than bile acid-mediated pathways. High-throughput screening (HTS) techniques have identified potent GP-BAR1 agonists, including GlaxoSmith Klines's isoxazole 610a and Pfizer's tetrahydropyridopyrimidine 7.10b A key aspect of the latter work was the calibration of human and dog in vitro assay systems that could be linked to data from a human ex vivo peripheral blood monocyte assay that expresses receptor at endogenous levels. Such GP-BAR1 modulators may not only be useful agents to treat T2D with concurrent management of glucose levels and body weight, but may also address other aspects, such as the inhibition of hepatic inflammatory responses that exacerbate metabolic disease.11 In devising chemical modifications of the bile acid side chain and nucleus to afford more potent and selective GP-BAR1 modulators, the role of stereochemistry on the specific biological activities of bile acids should be kept in mind. The absolute configuration of alkyl-substituted bile acid derivatives has a significant influence on their specificity towards GP-BAR1, which allows, for the first time, pharmacologically directed separation of genomic and nongenomic

5

activities of bile acid derivatives.7 Computational modeling studies indicate a narrow ligand binding pocket in GP-BAR1 that preferentially recognizes the (S) configuration of 23-methylsubstituted bile acids, that is predicted to provide improved selectivity for GP-BAR1 over FXR.6 Introduction of a methyl group at the C23(S) position of the natural bile acids CDCA to give 8 and cholic acid (CA) to give 9 confers selectivity for GP-BAR1 over FXR activation, while the (R) epimers of these C23-methyl-substituted bile acids display very low activity and no specificity towards GP-BAR1. This documents that a minor stereochemical modification of bile acid side chain moieties, namely methylation at C23 to provide the (S) configuration, can improve their potency, affinity, selectivity, and bioavailability.7

Relative to other bile acids, both LCA (1) and its C7-hydroxy epimer ursodeoxycholic acid (UDCA, 3) remain less explored. Among the naturally-occurring bile acids and their glycol derivatives, LCA is the most potent endogenous GP-BAR1 agonist.12 Even so, it is insufficiently potent to be used at low doses and is potentially toxic if used at high doses. UDCA is the only drug in the bile acid series that has been approved by the FDA for the treatment of primary biliary cirrhosis (PBC), an autoimmune disease characterized by progressive cholestasis.13 UDCA, in contrast to toxic hydrophobic bile acids, has also been shown to suppress the inflammatory response and, as such, is increasingly being employed for the treatment of hepatic and intestinal inflammatory diseases.13 While UDCA also binds FXR, its affinity is extremely weak and it acts, instead, as a partial agonist.14 The presence of the equatorial 7β-hydroxyl group in UDCA renders it more hydrophilic than CDCA (with its axial 7α-hydroxyl), and this may contribute to UDCA’s improved liver-function activity.15 However, due to its weak activity and low specificity, UDCA doses at concentrations up to 200 µM have been shown to exert only

6

mild effects on metabolic disease. More recent studies have examined the ability of eleven UDCA analogs, including 3- or 7-methylated UDCAs and amino acid conjugates, to activate GPBAR1 in luciferase-based reporter assays.13 However, these UDCA modifications failed to significantly improve affinities and activation of GP-BAR1. Oxidation or epimerization of UDCA’s C3 hydroxyl did not achieve impressive improvements.16a To further investigate the specificity of the UDCA, the most recent studies open the way to provide evidence that UDCA may be a useful scaffold to generate novel and selective steroidal ligands for GP-BAR1.16b-c These reports underscore the need for a combination of knowledge-based and structure-based design to achieve stereochemical modifications of the bile acids. Stereoselective introduction of substituents α to a carbonyl can be readily achieved via alkylation using chiral auxiliaries or catalysts.17 However, the development of stereoselective syntheses to produce alkyl-substituted bile acid derivatives in high yields and few steps remains a formidable challenge. Taking advantage of three-dimensional quantitative structure-activity relationship (3D QSAR) studies that show how regions around the bile acid scaffold affect GP-BAR1 activity,18 and also taking advantage of our previous experience in stereoselective methodology development,19 we designed processes that achieve stereoselective alkylation at the C23(S) position, and stereoselective Meerwein-Ponndorf-Verley reduction of the 7-keto group. Here, we present the results for the preparation of a series of alkyl-substituted LCA and UDCA derivatives. Their syntheses, biological appraisals, and structure-activity relationships (SAR) are described. Additionally, using computational modeling studies of binding of selected ligands in GP-BAR1, we have rationalized and discussed the enhanced potency and specificity for GPBAR1.

7

2. Results and discussion 2.1. Chemistry In attempting to devise a stereoselective strategy for the formation of 23(S)-methyl derivatives of LCA, CDCA, and UDCA, we considered the use of the extremely bulky base potassium hexamethyldisilazide (KHMDS) to form enolates of methyl esters of the parent acids. Each

ester’s α carbon (C23) is two flexible bonds distant from the nearest stereocenter at C20, and thus it was an open question whether the bulky complex, shown in Figure 2 during the deprotonation, could achieve adequate stereodifferentiation. In oorder rder to achieve 23(S) alkylation,

the KHMDS complex must be favored on the re face of the α enolate carbon, exposing the si face for preferential attack on the alkylating agent iodomethane. In order for the bulky complex

to control the face of attack, the KHMDS must not completely dissociate from the enolate after proton transfer, but must remain in proximity until alkylation occurs; a free enolate after dissociation would not be expected to show much preference for re or si attack, as shown by

molecular mechanics calculations (MMFF, Spartan Student) indicating that re- and si-exposed uncomplexed conformations are within about a kcal/mol in steric energy of each other.

8

Figure 2. The complex of KHMDS deprotonating CDCA ester. Alkylation on the si face of the incipient enolate would give the desired (S) methyl at C23.

We designed a two-step procedure for the synthesis of 23(S)-methyl-substituted CDCA, UDCA, and LCA (Figure 3). Since methyl esters of these acids are very expensive, we began with the readily available bile acids CDCA, UDCA, and LCA. We performed acid-catalyzed esterification with 1% sulfuric acid in methanol to obtain the corresponding methyl esters in quantitative yield. The hydroxyls of the crude methyl esters were then protected by treatment with dihydropyran and a catalytic amount of p-toluenesulfonic acid to give the corresponding tetrahydropyranyl (THP) ethers 10 in quantitative yields.

Treatment of the protected CDCA (10a) or UDCA (10b) with KHMDS in THF/HMPA at -78 oC produced enolates that reacted with the electrophile iodomethane. The THP groups were then removed by treatment with a catalytic amount of pyridinium p-toluenesulfonate (PPTS) (20 mol %),20 or an overnight reaction with 2N HCl in MeOH (25 mL).7 The methyl esters were subsequently hydrolyzed with NaOH in methanol at room temperature, affording the 23(S)methyl-substituted CDCA (8) and new UDCA derivative (12). The crude products were purified by crystallization or in some case by flash chromatography eluted with CH2Cl2/MeOH (9:1).

9

Figure 3. Syntheses of 23(S)-methyl-CDCA (8), 23(S)-methyl-UDCA (12), and 23-methyl-LCA (13).

Both the (S) and (R) epimers of 8 and 13 are known, and the configuration at C23 of a close analog has been determined by X-ray crystallography.6 The stereoisomers can be distinguished

by the increased polarity of the (R R)) isomers on chromatography and by characteristic chemical shifts in the 1H NMR spectrum, in particular the shift of the C23 hydrogen. In the (S) isomers, this signal appears > 2.45 ppm, whereas in (R) isomers it is < 2.45 ppm. The methodology shown in Figure 3 resulted in exclusive production of the 23(S)-epimers of 8 and 12. Interestingly, in contrast to the complete stereoselectivity in the alkylations forming 8 and 12, the corresponding

sequence on LCA 1 produced a mixture of (R) and (S) epimers of the product 13. These 10

stereoisomers could be separated chromatographically, but the reaction was not very stereoselective. This sequence of esterification/protection followed by deprotonation/methylation and deprotection/saponification achieved the desired products 8, 12, and 13, in 65%, 10% and 35% overall yields, respectively, starting from the native CDCA, UDCA, and LCA. We then investigated the effect of varying reaction conditions on the yield and stereoselectivity of the reaction of protected CDCA (10a). In the presence of KHMDS and absence of HMPA, a reduced yield (