Liver-selective glucocorticoid antagonists: A novel treatment for type 2 diabetes

J . Med. Chem. 2004, 47, 4213-4230

4213

Liver -Selective Glu cocor ticoid An ta gon ists: A Novel Tr ea tm en t for Typ e 2

Dia betes

Thomas W. von Geldern,* Noah Tu,^|Philip R. Kym,^†J ames T. Link,^†Hwan-Soo J ae,^†Chunqiu Lai,^†

Theresa Apelqvist,^§Patrik Rhonnstad,^§Lars Hagberg,^§Konrad Koehler,^§Marlena Grynfarb,^§

Annika Goos-Nilsson,^§J ohnny Sandberg,^§Marie O¨ sterlund,^§Tomas Barkhem,^§Marie Ho¨glund,^§J iahong Wang,^†

Steven Fung,^†Denise Wilcox,^†Phong Nguyen,^†Clarissa J akob,^‡Charles Hutchins,^‡Mathias Fa¨rnegårdh,^§

Bjo¨rn Kauppi,^§Lars O¨ hman,^§and Peer B. J acobson^†

Metabolic Disease Research and Structural Biology Departments, Global Pharmaceutical Discovery, Abbott Laboratories,

Abbott Park, Illinois 60064, and KaroBio AB, Huddinge, Sweden

Received J anuary 8, 2004

Hepatic blockade of glucocorticoid receptors (GR) suppresses glucose production and thus

decreases circulating glucose levels, but systemic glucocorticoid antagonism can produce adrenal

insufficiency and other undesirable side effects. These hepatic and systemic responses might

be dissected, leading to liver-selective pharmacology, when a GR antagonist is linked to a bile

acid in an appropriate manner. Bile acid conjugation can be accomplished with a minimal loss

of binding affinity for GR. The resultant conjugates remain potent in cell-based functional

assays. A novel in vivo assay has been developed to simultaneously evaluate both hepatic and

systemic GR blockade; this assay has been used to optimize the nature and site of the linker

functionality, as well as the choice of the GR antagonist and the bile acid. This optimization

led to the identification of A-348441, which reduces glucose levels and improves lipid profiles

in an animal model of diabetes.

In tr od u ction

use in the treatment of T2DM. There is evidence to

suggest that this is indeed the case. Thiazolidinediones

such as rosiglitazone (Avandia) or pioglitazone (Actos)

improve glycemic status by increasing the insulin

response in peripheral tissues, while the antidiabetic

properties of metformin (Glucophage) are derived at

least in part through a reduction of hepatic glucose

production.

Hepatically derived glucose plays a key role in main-

taining circulating glucose levels, especially in the

fasting state. Changes in the activity of individual

enzymes in this pathway can have dramatic metabolic

consequences. For example, loss-of-function mutations

in key gluconeogenic enzymes (e.g. glucose-6-phos-

phatase, G6Pase, or fructose-1,6-bisphosphatase, FB-

Pase) trigger human genetic diseases characterized by

a dramatic reduction in HGO and resulting episodes of

hypoglycemia.^1,2Conversely, the overexpression of phos-

phoenolpyruvate carboxykinase (PEPCK) in the liver of

transgenic mice is sufficient to produce hyperglycemia

and results in an overall phenotype that closely re-

sembles human type 2 diabetes.³

Th e Role of Glu cocor ticoid s. While glucocorticoids

have antiinflammatory properties and produce multiple

effects on protein, carbohydrate, lipid, and nucleic acid

metabolism, they are named in recognition of their

primary role in modulating glucose metabolism. Glu-

cocorticoids raise blood glucose levels by functionally

antagonizing the action of insulin, thereby inhibiting

glucose disposal and promoting hepatic glucose produc-

tion. The change in HGO is primarily driven by an

increase in gluconeogenesis, a result of increased mo-

bilization of gluconeogenic precursors, direct transcrip-

tional stimulation of gluconeogenic enzymes (e.g. PEP-

Glycemic control depends on a precise match between

glucose inputs and outputs; any disturbance in this

balance can result in hypo- or hyperglycemia. Glucose

is taken in as dietary carbohydrate or is produced in

the liver (hepatic glucose output (HGO)); it is consumed

in tissues as a primary energy source. Following meals,

a glucose-driven increase in circulating insulin levels

leads to a stimulation of glucose disposal (primarily in

skeletal muscle) and a reduction of HGO. Through the

combination of these insulin effects, blood glucose levels

show only a small and transient increase after meals.

In the fasting state, HGO serves as the primary source

of glucose and thus is critical for survival. These

responses are tightly regulated; any drop in glucose

levels triggers a counter-regulatory hormonal response

that will increase hepatic glucose output and restore

normoglycemia.

These counter-regulatory responses are out of balance

in patients with Type 2 diabetes mellitus (T2DM).

Diabetics fail to move glucose into tissues efficiently as

circulating levels rise, and these elevated circulating

glucose levels fail to appropriately suppress hepatic

glucose production. Each of these factors contributes to

the hyperglycemia that is the hallmark of this disease.

Thus, it is expected that an agent that restores periph-

eral insulin sensitivity, or decreases HGO, would find

* Address correspondence to D-47H, AP10-208, Abbott Laboratories,

100 Abbott Park Rd., Abbott Park, IL 60064-6100. E-mail:

thomas.vongeldern@abbott.com.

^†Metabolic Disease Research, Abbott Laboratories.

^‡Structural Biology, Abbott Laboratories.

^§KaroBio AB.

^|Current address: S*BIO, 1 Science Park Rd., Singapore.

Published on Web 07/13/2004

4214 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

CK and G6Pase), and interference with insulin signal

transduction in the liver.⁴The potent effects of gluco-

corticoids on glucose metabolism can be observed in

human diseases characterized by increased or decreased

cortisol production. Patients with an excess of cortisol

production (Cushing’s disease) have a cluster of symp-

toms that include impaired glucose tolerance due to

insulin resistance and increased gluconeogenesis.^5,6

Overt diabetes occurs in approximately 20% of these

patients. On the other hand, a deficit of cortisol produc-

tion (Addison’s disease) results in decreased HGO and

occasionally to hypoglycemia.⁷It is worth noting that

these changes in cortisol levels modulate global glucose

metabolism even when the secretion and action of other

hormones involved in glucose homeostasis is unim-

paired.

F igu r e 1. Some compounds reported to exhibit in vivo liver

selectivity.

Rosuvastatin has been reported to be unique among

HMG-CoA reductase inhibitors in demonstrating liver-

selective pharmacology.¹²While it is apparent that

specific structural features of these molecules are

related to their “intrinsic targeting” properties (Figure

1), the structure-function correlations do not provide

a clear strategy to identify liver-selective variants of

other drugs.

Efforts to rationally develop generic liver targeting

strategies have generally employed an “address-and-

message” approach. The bioactive agent in question is

attached (“conjugated”) to a vector intended to direct

distribution selectively to the liver. As a rule, these

vectors are selected based on their high affinity for

specific hepatic structures; sialo- and asialoglycopro-

teins^13,14and neoglyco- and neopeptide albumins¹⁵have

all been studied as targeting agents. Liposomes have

been galactosylated to select for their uptake into

hepatocytes.¹⁶Triglyceride hydrolysis has been used as

a means to release active drug in the liver.¹⁷Except in

the latter case, these conjugates can only be adminis-

tered through intravenous injection.

Glucocorticoid receptor (GR) antagonism has been

validated as a strategy for regulating HGO in vitro, in

animal disease models, and in man. Most of these

validation studies have been performed using the

prototype GR antagonist mifepristone (RU-486, 1).

Thus, acute treatment of genetically obese and diabetic

db/db mice (which lack the satiety hormone leptin)

reduces circulating glucose levels by half and normalizes

hepatic levels of G6Pase and PEPCK.³Longer-term

treatment of obese diabetic ob/ob mice (which lack leptin

receptor) normalizes postprandial glucose levels and

improves insulin sensitivity.⁸Glucose metabolism im-

proves in Cushing’s patients who receive mifepristone

treatment, as indicated by a 40% reduction in HbA1c

levels.⁶And a single dose of a GR antagonist can reduce

HGO and fasting glucose levels in normal subjects.⁹

However, long-term systemic GR antagonism may not

be a viable approach for the treatment of type 2

diabetes. Experience with extended (up to 12 months)

mifepristone treatment suggests that generalized GR

antagonism can lead to symptoms of adrenal insuf-

ficiency (nausea, vomiting, exhaustion). It is also ex-

pected that systemic receptor blockade will be problem-

atic in situations where increased cortisol levels are an

advantageous physiologic adaptation to stress,¹⁰and

because peripheral glucococorticoid responsiveness is

closely monitored by the hypothalamus, generalized

receptor antagonism results in the activation of the

hypothalamic-pituitary-adrenal (HPA) axis, causing

stimulation of the adrenal cortex (adrenal hyperplasia)

and increased cortisol secretion. Neither of these com-

peting outcomes (adrenal insufficiency, hypercorti-

solemia) is desirable from a patient perspective.

In principle, bile acids offer a unique solution to the

problem of liver targeting which is compatible with oral

delivery of a therapeutic agent. Naturally occurring bile

acids are tightly constrained within the enterohepatic

loop through the action of a series of transporter

proteins.¹⁸These highly efficient enteric and hepatic

uptake mechanisms ensure that less than 2% of the

total bile acid pool is excreted daily, even as it makes

10 or more passes through the intestine, portal vein,

liver, and bile duct. A bile acid conjugate that retained

these properties would be expected to have exceptionally

high bioavailability as well as substantial hepatic levels

and a high liver-to-plasma (or -tissue) ratio.

A number of researchers have attempted to probe this

mechanism for hepatic targeting.^19-23Of particular note

is the work of Kramer and co-workers at Aventis. These

workers have attached bile acids to dye molecules

(allowing them to track distribution), to small peptide

fragments, and to oligonucleotides (Figure 2). They have

demonstrated in a series of elegant experiments that

these conjugate molecules are recognized by bile acid

transport systems,^19,20and that they are trafficked

through the liver in a manner similar to a bile acid.^21,22

But none of these conjugates has been shown to deliver

intact conjugate when dosed orally. In fact, Borchardt

and co-workers have questioned the role of transporters

in moving molecules of this type, noting that while bile

acid-peptide conjugates can bind to transporters, they

are not themselves transported.²³Thus, several key

questions regarding the value of bile acid transport as

a means to deliver oral pharmaceuticals selectively to

the liver remain unanswered. The primary goal of the

In contrast, a liver-specific derivative of mifepristone

would be expected to decrease HGO and improve glucose

metabolism without the risk of these peripherally driven

side effects. The above considerations have led us to

propose liver-selective glucocorticoid antagonism as a

novel approach to the treatment of T2DM. Compounds

with the target profile would retain the beneficial effects

of glucocorticoid blockade to suppress hepatic glucose

production; but would avoid the complications associ-

ated with the loss of peripheral glucocorticoid response.

Liver Ta r getin g. An extensive literature describes

the efforts of researchers to “target” drugs to the liver.

Leeson and co-workers report that certain substructural

features lead to a reduction in cardiac vs. hepatic

activity in a family of thyromimetic T₃derivatives.¹¹

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4215

F igu r e 2. Bile acid conjugation has been employed to alter the distribution of a number of pharmacologic agents, including

chlorambucil and an inhibitor of prolyl-4-hydroxylase (22).

Sch em e 1. Derivatization of Mifepristone. Compound 2 Is a Key Intermediate in the Preparation of Bile

Acid-Conjugated Mifepristone Derivatives^a

a

Reagents and conditions: i. TPAP, NMO, CH₂Cl₂; ii. aq HCl, MeOH; iii. Ac₂O, pyr, 60 °C.

Sch em e 2. C17-Linked Conjugate 15 Is Prepared from 2 through a Sequence of Acylations^a

a

Reagents and conditions: i. Boc-Glu-OMe, TBTU, i-Pr₂NEt, DMF; ii. HCl/EtOAc; iii. cholic acid, TBTU, i-Pr₂NEt, DMF; iv. LiOH,

H₂O-THF.

present study was to address these questions in the

context of preparing a liver-targeted derivative of mife-

pristone (1).

glycocholate derivative 4 is prepared by coupling the

acid with a suitably protected glutamate analogue,

followed by selective removal of the side chain protecting

group (Scheme 2). Other derivatives (see Scheme 3) take

advantage of the known reactivity order of the three

hydroxyl groups, C3-OH > C7-OH > C12-OH. For

example, the C3-OH may be selectively acylated to give

intermediates such as 5 and 6, or selectively activated

via sulfonylation and displaced with a diol, as originally

reported by Kramer and co-workers,¹⁹to give the

inverted ether 7 (after activation of the remaining

primary alcohol). Protection of the C3-alcohol, via

formation of the mono-acetate, allows for acylation at

C7 to provide derivatives such as 10 and 12. When the

C7-OH is also protected, for example by formation of

the 3,7-diacetate, reaction occurs at the hindered C12

site to give intermediates such as 13. The formation of

C3R ether-linked intermediate 8, retaining the natural

Ch em istr y. In general, the bile acid conjugates

described herein are assembled from two primary frag-

ments, derived respectively from a glucocorticoid an-

tagonist and a naturally occurring bile acid. GR antago-

nist fragment 2 was initially prepared from commercially

available 1 using the radical dealkylation procedure of

Acosta et al.²⁴In our hands this procedure gave variable

results and was particularly problematic when scaled

up. For this reason we have developed a new procedure²⁵

for dealkylating this N,N-dimethylaniline (Scheme 1).

Treatment of 1 with TPAP cleanly oxidizes one methyl

group to give an N-formyl derivative, which can be

hydrolyzed under mildly acidic conditions to provide 2

in 60% overall yield. Bile acid derivatives are generally

prepared from cholic acid as a starting material. The

4216 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Sch em e 3. Synthesis of Bile Acid-Based Intermediates^a

Tu et al.

a

Reagents and conditions: i. p-NP-OCOCl, pyr, 60 °C; ii. 2-aminoethanol, NMM, dioxane/water, 50 °C; iii. NBS, PPh₃, DMF; glycine,

NMM, dioxane/water, 50 °C; iv. glycine, NMM, dioxane/water, 50 °C; v. Ac₂O (1.1 equiv), DMAP, pyr; vi. Ac₂O (2.2 equiv), DMAP, pyr; vii.

Ag₂CO₃, Celite, tol, reflux/-H₂O; viii. CH₃NH₂, NaCNBH₃, AcOH, MeOH/CH₃CN; ix. R1-Cl, i-Pr₂NEt, CH₃CN; x. ref. 19; xi. Zn, allyl

iodide, THF; xii. TMS-Cl, imidazole, DMF; xiii. 9-BBN, THF; Me₃NO, 60 °C; xiv. TsCl, pyr; xv. MOM-Cl, i-Pr₂NEt, DCE, 45 °C; xvi. allyl

bromide, KOtBu, THF, 0 °C; xvii. NaBH₄, MeOH.

configuration of the bile acid, is also possible, but

ironically this alkylation is most successful when the

C7- and C12-hydroxyls are protected as acetates.¹⁹

The C3-OH can also be selectively oxidized, using

Tserng’s modification of the Fetizon oxidation.²⁶The

resultant ketone can be used as the substrate for a

reductive amination, which in the case of methylamine

provides the R-epimer 9 as the major product (>5:1).

Alternatively, the enolate can be alkylated with allyl

bromide (alkylation occurs with high selectivity on the

â-face of C4); or the ketone can be reacted with an

allylzinc reagent (approach of the nucleophile again

occurs from the â-face, reestablishing the R-alcohol

stereochemistry). In each of these latter cases, further

elaboration of the allyl group provides the coupling

intermediate (14 and 11, respectively).

Coupling of GR antagonist and bile acid-based frag-

ments is accomplished in straightforward fashion

(Scheme 4). Anilide-based conjugates such as 15, 16a ,

19a , and 20 are formed by in situ activation of the acid

fragment under standard amide bond-forming condi-

tions (generally EDCI/HOBt or TBTU). Aniline-based

couplings (i.e. to give 16b-k , 17, 18, and 19b) are

accomplished by warming an acetonitrile solution of the

coupling partners in the presence of an amine base.

When the electrophile is a sulfonate ester, the addition

of an iodide source to the reaction mixture catalyzes the

process; alternatively the alkyl iodide may be preformed

and used directly in the reaction. Hydrolysis of the ester

functionality under alkaline conditions completes the

synthesis of the bile acid conjugates.

derivatives. Conjugates based on glycocholic or tauro-

cholic acids are prepared through an alternative strat-

egy that involves assembly of the intact cholate conju-

gate 16câ, followed by coupling with glycine or taurine

to generate the final product (Scheme 5).

Com p ou n d Eva lu a tion . All compounds are evalu-

ated for their ability to bind to the glucocorticoid

receptor, employing a standard radioligand binding

assay and using [³H]-dexamethasone, a potent agonist,

as the reference radioligand. A similar format is used

to measure the affinity of selected compounds for other

steroid hormone receptors. A typical “receptogram”

includes androgen (AR), estrogen (ERR and ERâ), min-

eralocorticoid (MR), progesterone (PR), and thyroid

hormone (TRR and TRâ) receptors.

Functional activity is evaluated using two distinct

cell-based assay systems. GRAF (in Swedish, “GR

linked to Alkaline F osfatase”) cells are a genetically

engineered mammalian cell line expressing GR. Based

on a Chinese hamster ovarian (CHO) cell line, GRAF

cells also contain a stably integrated artificial transcrip-

tion unit composed of a glucocorticoid hormone response

element (GRE) and core promoter sequences fused to a

downstream reporter gene encoding a secreted form of

alkaline phosphatase (ALP). Compounds are evaluated

for antagonism of dexamethasone-induced ALP expres-

sion. GRAF cells can also be used to determine intrinsic

agonist activity of analogues. Compounds may also be

evaluated for activity in hepatocytes, the intended target

cell. Primary rat hepatocytes are stimulated with dex-

amethasone (“dex”), leading to the increase in expres-

sion in a number of GR-regulated proteins. The enzyme

tyrosine aminotransferase (TAT) is chosen as a marker

for GR activation. TAT is known to be regulated by

hepatic glucocorticoid, but is not directly on the pathway

leading to increased hepatic glucose production. It is

Other bile acids can provide the basis for the address

component of these conjugate molecules. Thus, for

example, deoxycholic acid, chenodeoxycholic acid, and

lithocholic acid can be subjected to the synthetic pro-

cedures described above to produce the corresponding

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4217

Sch em e 4. Bile Acid Conjugates Are Assembled from Intermediate Fragments via Acylation or Alkylation Strategies

a

Reagents and conditions: i. compound 2, TBTU, DMF; ii. LiOH, H₂O/THF; iii. compound 2, i-Pr₂NEt, NaI, CH₃CN, 100 °C

Sch em e 5. Glycocholate and Taurocholate Conjugates 21a and 21b Are Prepared by Coupling 16câ with Glycine and

Taurine, Respectively

selected for practical reasons; dex increases TAT levels

2-3-fold over the course of a 4-h incubation, a far

greater window than is observed with key gluconeogenic

enzymes such as PEPCK or G6Pase.

levels but also the lymphopenic response. A liver-

selective agent would modulate TAT and glycogen

responses but have no effect on lymphocyte levels.

Resu lts a n d Discu ssion

The use of surrogate markers of glucocorticoid activa-

tion also provides a basis for the in vivo evaluation of

efficacy and liver selectivity. In a rat prednisolone

challenge (RPC) experiment, normal fasted Sprague-

Dawley rats are given an oral dose of prednisolone,

another glucocorticoid agonist. Prednisolone (pred) is

systemically available and produces GR-mediated re-

sponses in peripheral tissues as well as in the liver. In

particular, it increases hepatic levels of TAT, stimulates

the deposition of glycogen in the liver, and blocks the

production of circulating lymphocytes (part of the basis

for its antiinflammatory effect). A systemic glucocorti-

coid antagonist like 1, given prior to the pred challenge,

will block not only the increases in TAT and glycogen

For the “address-and-message” strategy embodied in

the production of conjugate molecules such as 15-21

to be successful, each component of the conjugate must

retain most, if not all, of its desirable characteristics.

Thus, for example, compounds 15-21 must bind tightly

to the glucocorticoid receptor, despite the attachment

of a large bile acid moiety. Similarly, the bile acid

fragment of 15-21 must be capable of directing liver-

selective distribution and/or pharmacology, even while

it is attached to a large GR antagonist unit. It is clear

that the details of the attachment of these two func-

tionalities, as well as the nature of the linking unit

itself, will play a critical role in defining the properties

4218 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

F igu r e 3. Computer model of 16câ bound to the GR ligand-binding domain (GR-LBD). Model is based on the X-ray crystal

structure of mifepristone (1)/GR-LBD complex.²⁷

of the conjugates. Initial structure-activity studies

probed this question in some detail. We began with the

question of how to attach a bile acid fragment to a

glucocorticoid antagonist while retaining GR affinity.

The X-ray crystal structure of the GR ligand-binding

domain (GR-LBD) recently reported by our group,²⁷and

including mifepristone as ligand, provided valuable

assistance in addressing this issue. The major structural

features of GR-LBD are consistent with those reported

for other nuclear hormone receptors such as PR,²⁸

defined by an antiparallel sandwich of R-helical layers.

The ligand sits within a large hydrophobic binding

cavity, with key hydrogen-bonding contacts involving

the C3-ketone (to Gln-570 and Arg-611) and C17â-

hydroxyl (to Gln-642 and to a tightly bound water

molecule held in place by Cys-736). The p-dimethy-

lamino group on the C-11 aromatic ring protrudes from

this binding pocket, accessing solvent-exposed space. By

analogy with earlier studies on PR,²⁸it is this group

which displaces helix 12 from its characteristic agonist

position sitting over (and capping) the binding cavity;

this forces the receptor/ligand complex into an antago-

nist conformation. The solvent-accessibility of this

amine also suggests it as a logical site for attachment

of a large molecular fragment, such as a bile acid (see

model in Figure 3, based on X-ray structure of mife-

pristone/GR-LBD complex). In practice, we have ex-

plored two chemical strategies for linking through the

C-11 aniline. Compounds such as 15, 16a , 19a , and 20,

which are coupled through an activated acid to give an

anilide-containing linker, generally exhibit GR affinities

between 1 and 7 nM (K_I) range, with an average for this

subclass of 2.3 nM (data summarized in Table 1). While

these analogues are active, supporting the premise that

the C-11 substituent can tolerate a large modification,

they are nonetheless substantially less potent than the

parent 1 (K_I) 0.1 nM). These conjugate molecules have

two features that distinguish them from the parent: a

large bile acid substituent has been attached, and the

nature of the aniline group has been altered by the

presence of the amide group. How much of the change

in binding affinity is due to each of these modifications?

A comparison of 1 with the corresponding N-acetyl

analogue 3 (IC₅₀) 0.4 nM) suggests that the N-acyl

linkage contributes substantially to the observed loss

of potency. To test this hypothesis, a series of N-alkyl-

modified analogues (16b-k , 17, 18, 19b) have been

prepared by alkylation of the methylaniline 2. These

compounds recover the ∼4-fold loss of activity due to

the presence of the amide; they exhibit GR affinities in

the 0.11-1 nM range, with a mean K_Iof 0.33 nM for

this group. When properly linked, the cost (in GR

binding affinity) of attaching a bile acid substituent can

be 3-fold or less.

Conjugation with the bile acid also has relatively little

effect on the affinity of these analogues for other nuclear

hormone receptors (Table 1). In addition to its role as a

GR antagonist, mifepristone has potent interactions

with both the progestin (K_I) 0.64 nM) and androgen

(K_I) 0.65 nM) receptors. Acylation of the C-11 aniline

nitrogen, as in 3, 15, 16a , 19a , or 20, reduces PR binding

affinity as was observed in the GR binding assay. Most

importantly, there seems to be no particular advantage

(nor disadvantage) in terms of GR:PR selectivity to

adding the bile acid moiety to these GR antagonists.

Similar results are observed with ER, MR, and TR,

against which selectivity is generally quite high. The

exception to these trends comes in the case of androgen

receptor. The modest selectivity of mifepristone for GR

over AR (6-fold) improves significantly upon conjugation.

Conjugates have GR:AR ratios which vary from 10 to

70, averaging 25-fold. No obvious SAR trends emerge.

The role of the bile acid moiety is examined obliquely,

by determining the potency of analogues in several

functional assays in intact cells. GRAF cells, which do

not contain bile acid transporters, are expected to

provide some indication of the ability of the various

conjugates to penetrate cells passively. A comparison

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4219

F igu r e 4. Bile acid-conjugated glucocorticoid receptor antagonists.

of mifepristone, the acetamide 3, and the conjugated

analogues produces the following conclusions:

analogues. In the single instance where a direct com-

parison is possible, 16bR is ∼35× more potent than

16a R.

1. All of the compounds prepared for this study are

full functional antagonists in this assay, with minimal

agonist activity. The maximal blockade of dexametha-

sone-induced alkaline phosphatase production produced

by the conjugates varies from 75 to 100%, and averages

>90%.

2. None of the conjugates is superior to the parent 1

in the GRAF assay. Most analogues are 3-50 times less

potent than 1.

5. The data set is not complete enough to allow a clear

evaluation of the role of the linkage site on GRAF

activity. C3-linked conjugates tend to be more potent

than those linked at other positions on the bile acid

fragment; the mean K_Ifor the C3-linked compounds is

3.7 nM, while the average for those linked at other sites

(15, 18, 19, and 20) is 75 nM. This result is strongly

biased by the fact that all non-C3-linked conjugates are

anilides; which have been demonstrated to be less

potent in the GRAF assay (point 3 above). When this

confounding factor is removed, the trends are less

apparent. The C3-amide-linked 16a R is no different in

activity than C7- and C12- variants (19a and 20,

respectively); on the other hand, the C4-aniline 18 (52

nM) is far less active than a typical C3-aniline (average

K_I) 3.1 nM).

3. Analogues 21a and 21b, based on the “conjugated”

bile acids glycocholate and taurocholate, are relatively

poor antagonists in the GRAF assay; K_Ivalues are 7-

and 30-fold higher than the parent 16câ. This loss of

activity may be due to decreased cell permeability; 21b

exhibits a very low flux rate in caco-2 cells (Papp < 5 ×

10^-8cm/s, unpublished results).

4. Following the trend previously observed in the GR

binding assay, anilides are less active than the corre-

sponding anilines; the mean K_Ifor 15, 16a , 19a , and

20 is 88 nM, vs 4.4 nM for the remainder of the

6. It is clear that, within the C3-linked analogues,

conjugates linked on the â-face of the bile acid are

superior to those linked on the R-face. The average

4220 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Ta ble 1. In Vitro Activity of GR Antagonist Molecules

Tu et al.

compd

GR^a

PR^a

MR^a

AR^a

ERR^a

ERâ^a

TRR^a

TRâ^a

GRAF^a

0.44

HepTAT^a

1

0.10

0.64 (1)

640 (1)

0.65 (1)

>200 (1)

>750 (1)

>1,250 (1)

>2250 (1)

120

(0.09-0.11, 2)

(0.4-0.46, 2) (39-180, 8)

3

0.4

6.4 (1)

1.2 (1)

>1000 (1)

8.9 (1)

34 (1)

21 (1)

6.2 (1)

8.9

34 (1)

72 (1)

11 (1)

>200 (1)

11

570 (1)

240 (1)

160 (1)

>750 (1)

89

790 (1)

>2250 (1)

1500

1.4

ND

350

(0.24-0.66, 2)

(1.1-1.6, 2)

15

3.0

590 (1)

1200 (1)

1100 (1)

>1250 (1)

670

250

(1.0-6.3, 8)

(190-310, 3) (200-610, 2)

16a R 2.1

2.4

310

50

340

(1.8-2.8, 4)

16bR 0.39

(0.35-0.43, 2)

16cR 0.67

(0.3-1.4, 4)

16câ 0.27

(2.4-2.5, 2)

0.39 (1)

(210-460, 2)

(26-99, 2)

(170-540, 3)

>1000 (1)

440

1.4 (1)

330

(330-340, 2)

3.3

11

220

(0.79-14, 2)

(6.3-12, 2) (7.4-15, 2) (47-170, 2)

(2.7-73, 4)

(150-390, 3)

0.81

8.1

12

110 (1)

13 (1)

190 (1)

39

1.1

56

(0.1-1.5, 20) (0.18-2.8, 15) (220-1800, 8) (4.4-29, 8) (9-74, 7)

(340-1700, 7) (900-5400, 7) (0.45-2.3, 9) (34-100, 10)

16d R 0.27

(0.16-0.46, 2)

16eR 0.30

(0.27-0.33, 2)

16fR 0.42

(0.32-0.61, 4)

16gâ 0.22

(0.18-0.26, 2)

16h R 0.21

(0.20-0.23, 2)

16h â 0.17

0.96 (1)

3.5 (1)

1.9 (1)

1.1 (1)

1.7 (1)

0.48

190 (1)

850 (1)

300

7.4 (1)

16 (1)

17

6.2 (1)

140 (1)

19

>1250 (1)

ND

>2250 (1)

ND

1.2 (1)

15 (1)

60

(47-81, 2)

34

(21-60, 2)

15

60

(270-350, 2) (13-22, 2) (14-28, 2) (24-63, 2)

(14-16, 2)

(56-64, 2)

80 (1)

170 (1)

150 (1)

ND

4.6 (1)

4.8 (1)

3.3 (1)

ND

8.8 (1)

13 (1)

ND

>1250 (1)

ND

1900 (1)

ND

1.1 (1)

12 (1)

100

(81-130, 2)

ND

90

(90-90, 2)

14 (1)

ND

51 (1)

ND

>1250 (1)

ND

>2250 (1)

ND

1.1

56

(0.11-0.27, 6) (0.39-0.64, 3)

(1.0-1.3, 3)

(51-60, 2)

16iâ 0.16

(0.14-0.19, 2)

16jâ 0.18

(0.17-0.20, 2)

16k â 0.22

(0.20-0.25, 2)

17â 0.11

(0.06-0.16, 3)

18â 0.80

(0.7-0.9, 2)

19a R 1.1

(1.1-1.2, 2)

19bR 1.2

(0.87-1.7, 3)

20R 6.5

(4.8-8.2, 3)

21a â 0.42

(0.42-0.43, 2)

21bâ 0.27

(0.17-0.38, 4)

ND

0.38 (1)

0.76 (1)

2.7 (1)

ND

4.3

(4.0-4.6, 2)

2.1

>1000 (1)

310 (1)

ND

53 (1)

37 (1)

18 (1)

ND

46 (1)

94 (1)

ND

48 (1)

>750 (1)

ND

52 (1)

49 (1)

37 (1)

100 (1)

8.6 (1)

340

(1.9-2.2, 2)

(300-380, 2)

4.4 (1)

4.6 (1)

ND

>1250 (1)

ND

>2250 (1)

ND

230

(150-330, 2)

730

(360-1300, 5)

ND

500

(460-550, 2)

1.0 (1)

0.53 (1)

510 (1)

140 (1)

34 (1)

8.9 (1)

ND

>1250 (1)

>2250 (1)

180

(150-300, 3)

10 (1)

30 (1)

37 (1.9-67, 2) 160

(130-220, 4)

a

All K_Ivalues reported in nM, as geometric mean, (range, N). Each N ) 1 refers to the average of two experiments performed on the

same sample; each experimental value is measured in duplicate.

GRAF K_Ifor the R-linked analogues is 11 nM; while the

â-linked conjugates average 1.2 nM. In the two cases

where a direct comparison is possible, both 16câ and

16h â (each 1.1 nM) are 10× more potent than their

R-linked comparators 16cR and 16h R (11 nM and 12

nM). This paradoxical observation, that the more active

conjugate series is attached on the face opposite that of

a native bile acid, cannot readily be explained on the

basis of physicochemical properties such as logD or polar

surface area (PSA). It is, however, consistent with

previous reports by Kramer and co-workers.^20-22

7. A variety of bile acids are tolerated. While most of

the analogues shown are built from cholic acid, the

predominant native bile acid, the GRAF activity of

conjugates incorporating deoxycholic (16iâ, 0.38 nM),

chenodeoxycholic (16jâ, 0.76 nM), and lithocholic (16k â,

2.7 nM) have similar activity to the cholate comparator

16câ (1.1 nM).

The hepatocyte TAT assay provides a cellular readout

that factors in several additional parameters. These

cells carry functional bile acid importers and exporters,

so that intracellular drug concentrations are likely to

be regulated through a balance of active and passive

uptake and secretion pathways. In addition, primary

hepatocytes are metabolically competent, so that me-

tabolism may play a role in determining the availability

of active drug within the cells. As a consequence, all of

the compounds in this study (including 1) are less potent

in the TAT assay than in GRAF cells, and the differ-

ences in activity between compounds is less dramatic

(only a 20-fold potency difference between the best and

worst compounds). Several structure-activity trends

are observed, however:

1. In contrast to the results in the GRAF assay, a

number of the bile acid conjugates are superior to

mifepristone in hepatocytes. This result may provide

indirect evidence that the analogues are actively trans-

ported into these cells; alternatively it might represent

a change in the metabolic profile of the conjugates. In

fact, while we have no direct evidence on the first of

these points, we have been able to demonstrate that the

conjugates are more stable than the parent in hepato-

cytes. In these experiments (representative results

shown in Figure 5) compounds are incubated with

primary rat hepatocytes, and the consumption of start-

ing drug is evaluated by HPLC. While mifepristone is

converted rapidly under these experimental conditions,

the bile acid conjugates (represented here by compound

15) are essentially unchanged over the course of the

incubation.

2. Consistent with the trend in the GRAF assay,

conjugates linked at C4, C7, C12, and C21 are inferior

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4221

Ta ble 2. In Vivo Activity of GR Antagonist Molecules

% inhibition of pred response

compd

lymphocytes

TAT

glycogen

1

104

14

-14

9

101

-16

49

72

11

-19

64

66

60

77

-20

46

86

20

9

43

50

45

38

5

16bR

16cR

16câ

16d R

16fR

16gâ

16h R

16h â

16iâ

17â

4

9

-5

-2

43

22

0

-6

-16

69

6

21b

22

-63

3

ND

5

gen deposition. For each of these hepatic responses, the

greatest blockade produced by a conjugate molecule

comes from 16câ, which inhibits TAT by 72% and

glycogen levels by 86%. Several structure-activity

relationships emerge from this study. Compounds that

are attached to the bile acid through a C3-ether linkage

(16cR, 16câ, 16gâ, 16h R, and 16h â) are clearly superior

to those linked via nitrogen (16d R, 16fR) or through a

carbamate (16bR). The failure of the N-linked ana-

logues, which were among the most potent in hepato-

cytes, is particularly notable. Among the ether-linked

derivatives, there is relatively little difference between

linkers of differing lengths (compare 16câ with 16gâ

and 16h â) or between R- and â-face linked conjugates

(16cR vs 16câ; 16h R vs 16h â). While the former trend

mirrors the in vitro cellular potencies of the analogues,

the latter does not.

The data from these rat prednisolone challenge stud-

ies support the premise that bile acid conjugation might

lead to a functionally liver-targeted glucocorticoid an-

tagonist. Conjugate molecules such as 16cR, 16câ, 16gâ,

16h R, and 16h â are effective in blunting pred’s hepatic

effects, while leaving its systemic lymphopenic response

intact. Compound 16câ is particularly effective in this

model. To test whether these surrogate markers of

glucocorticoid activity accurately capture the com-

pounds’ properties, 16câ has been characterized further

in in vivo models that directly assess its antidiabetic

activity and its propensity for producing undesirable,

peripherally mediated side effects such as HPA activa-

tion.

An tid ia betic Effica cy a n d Sid e Effect P r ofilin g.

The model selected for the evaluation of compound 16câ

(A-348441) as an antidiabetic agent is the ob/ob mouse.

These animals, which have a genetic defect blocking the

production of the satiety hormone leptin, are obese,

hyperphagic, and insulin resistant and thus serve as

an excellent model for human Type 2 diabetes. They

have been used to characterize a number of commercial

antidiabetic and antilipidemic agents including met-

formin, fenofibrate, and the thiazolidinediones. In the

experiment at hand (data shown in Figure 6), the

compound is dosed orally, once a day, at 10, 30, and 100

mg/kg, over a period of 6 weeks. Postprandial plasma

glucose levels are measured every week, and are com-

pared with normal animals, with untreated ob/ob mice,

and with ob/ob mice treated with metformin (500 mg/

kg qd). Compound 16câ clearly lowers plasma glucose

levels in a time-dependent and dose-dependent fashion.

F igu r e 5. Hepatocyte stability studies of bile acid-conjugated

GR antagonist 15. The bile acid conjugate is substantially

protected against metabolism upon incubation with primary

rat hepatocytes (T_1/2. 24 h), as compared to parent 1 (T_1/2

2 h).

<

to C3-linked conjugates. The average K_Ifor all C3-linked

analogues is 110 nM, which is substantially lower than

any of the non-C3’s.

3. Among the C3-linked analogues, the amine-linked

compounds are notably active in the TAT assay. Com-

pounds 16d R, 16eR, and 16fR, which are linked on the

R-face through amide, urea, and sulfonamide linkages,

comprise three of the top five most active analogues;

though it is important to note that many of the differ-

ences in potency recorded for the TAT assay in Table 1

are not statistically significant.

4. Of the oxygen-linked conjugates, â-linked conju-

gates are superior to those linked on the R-face of the

bile acid. The mean TAT activity for 16a R, 16bR, and

16cR is 290 nM; for 16câ, 16gâ, 16h â, 21a â, and 21bâ

the average is 98 nM. In the one case where a direct

comparison is possible, 16câ (56 nM) is 4-fold more

potent than 16a R (220 nM).

Compounds with superior profiles in the GRAF and

hepatocyte assays were next subjected to in vivo evalu-

ation. The rat prednisolone challenge is designed as a

primary screening assay to provide a simultaneous look

at two key parameters. Hepatic activity is demonstrated

by the blockade of pred-stimulated TAT production and

through inhibition of pred-induced glycogen deposition.

Systemic selectivity is measured by the ability (or

inability) of the conjugate molecules to inhibit the

antiinflammatory effect of pred, as determined through

evaluation of circulating lymphocyte levels. As sum-

marized in Table 2, there is little doubt that our

conjugation strategy has succeeded in suppressing the

peripheral GR antagonist properties of these com-

pounds. Of the analogues tested, only 16h â shows any

significant ability to block pred-induced lymphopenia,

while mifepristone itself completely blocks this response

at the test dose of 100 mg/kg.

None of the bile acid conjugates is as active as 1 at

inhibiting hepatic glucocorticoid responses. 1 normalizes

pred-elevated TAT levels at a 100 mg/kg oral dose and

completely suppresses pred-induced increases in glyco-

4222 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

F igu r e 6. Chronic effects of A-348441 (16câ) in the diabetic ob/ ob mouse model. Compound is administered orally, qd to ob/ob

mice (N ) 8/group); plasma glucose determinations are made at the start of the study and on a weekly basis. Glucose values are

recorded 16 h postdose (a). Measurements of HbA_1cand lipid levels are made at beginning and end of study (b).

The onset of the effect is relatively slow, but by the end

of the dosing period, glucose levels have actually nor-

malized in the 100 mg/kg group. The lower doses

produce smaller responses, which still reach statistical

significance by the end of the study. It is important to

note that glucose levels are measured 18 h after the

once-daily dose; so the effects observed clearly indicate

a long-term improvement in glycemic control.

The side effect profile of 16câ is addressed by compar-

ing it with 1 in a hypothalamic-pituitary-adrenal

(HPA) activation model, which measures the normal

response of the body to reduced glucocorticoid signaling.

In this model, normal mice receive a single oral dose of

antagonist and are examined 2 h later to determine the

status of their HPA axis. As shown in Figure 7, a

systemic glucocorticoid antagonist like 1 stimulates

production of adrenocorticotrophic hormone (ACTH) by

the pituitary; this elevation in ACTH then drives

increased production of corticosterone from the adrenal

glands. While 1 causes substantial increases in both

parameters (∼7-fold) in a dose-responsive fashion, even

higher doses of 16câ leave ACTH and corticosterone

values unchanged at baseline levels. Importantly, the

levels of the hormones remain unaltered at doses that

demonstrated glucose-lowering activity in the ob/ob

mouse. Thus, it appears that 16câ demonstrates a

significant degree of functional hepatic selectivity in

clinically relevant models demonstrating antidiabetic

efficacy and systemic antiglucocorticoid side effects.³¹

The overall pharmacology profile of this agent indicates

that it might be useful for treating Type 2 diabetes

through the reduction of hepatic glucose production and

that it is amenable to long-term dosing with a mini-

mized mechanism-based side effect profile.

Th e Role of Bile Acid Tr a n sp or t. The bile acid

component of 16câ clearly plays a key role in targeting

the antiglucocorticoid activity of this conjugate molecule

to the liver. This observation would seem to confirm our

initial hypothesis regarding the role of bile acid trans-

porters in moving these conjugates selectively through

the enterohepatic loop. However, additional studies

probing this issue further have generated conflicting

results. We note several of these below; in the end we

find ourselves unable to reach a clear conclusion regard-

ing the specific role of transporters in moving these

molecules.

This observation is reinforced by end-of-study mea-

surements of glycated hemoglobin (HbA_1c) levels. Gly-

cated protein levels are a good indication of long-term

glycemic status and are used as such in clinical practice.

An HbA_1cchange of 1% is considered to be clinically

relevant and has been shown to correlate with a

substantial improvement in long-term outcomes for a

diabetic patient.²⁹Ob/ob mice have levels of HbA_1c

(typically 8-10%) that are clearly elevated when com-

pared to their lean littermates (4-5%). These values are

in line with those recorded for human diabetic patients

and nondiabetic control subjects, respectively. In the

study at hand, six weeks of once-daily treatment with

a 100 mg/kg dose of 16câ reduces HbA_1cby 4.7% units,

or more than 80% of the difference between lean and

diabetic animals. This result is clearly superior to that

resulting from treatment with a maximal dose (500 mg/

kg) of metformin in the same model (data not shown).

16câ produces desirable changes in other relevant

metabolic parameters as well. Like many human Type

2 diabetics, ob/ob mice have elevated levels of several

key plasma lipids, including cholesterol, triglycerides,

and free fatty acids (FFAs). These elevations are sup-

pressed by treatment with 16câ at 100 mg/kg orally.

Serum cholesterol decreases by 60 mg/dL, though it does

not return to lean levels; circulating free fatty acids are

normalized, and triglyceride concentrations drop below

those of the control group. These changes, which are

consistent with a reduction in hepatic glucocorticoid

signaling,³⁰occur in the absence of significant weight

loss. All are expected to provide additional benefit to

the typical Type 2 diabetic patient.

1. The pharmacokinetic profile of 16câ is quite differ-

ent from a typical bile acid. One primary motivation for

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4223

F igu r e 7. Comparison of A-348441 (16câ) with mifepristone (1) in the hypothalamic-pituitary-adrenal activation (HPA) model.

A single oral dose of each agent is given to normal Sprague-Dawley rats (N ) 10/group); measurements of ACTH (a) and

corticosterone levels (b) are recorded 2 h postdose.

F igu r e 9. Bile acid conjugation of RU-43044, a selective GR

antagonist.

F igu r e 8. Plasma and liver concentrations following oral

2. The address-and-message concept inherent in the

dosing of A-348441 (10 mg/kg) in ob/ob mice. Note: each data

design of 16câ is not generally applicable. Two sets of

point represents a single animal.

examples are relevant. (a) Numerous naturally occur-

ring bile acids are able to engage the transporter

our decision to use bile acids as a targeting device is

systems in a manner similar to cholic acid. We expected

the high efficiency with which they are constrained

that replacement of the cholic acid component of 16câ

within the enterohepatic system. Steady-state kinetic

analysis¹⁸has determined that the bulk absorption

with these alternative bile acids would lead to other

analogues with similar in vivo properties. In practice

efficiency of bile acids during a single pass though the

this has proven not to be the case. Of particular note is

intestine is 97-99% (F_A∼ 0.98); similarly the efficiency

the taurocholate conjugate 21b. While taurocholic acid

is one of the most efficient substrates for the bile acid

specific transporters, 21b has no beneficial effect in a

of hepatic capture of cholic acid from the portal circula-

tion is >90% (F_H< 0.1). As a result, the bile acid pool

is highly conserved, with <1% of the total excreted daily

rat prednisolone challenge experiment; and no drug can

be detected in the liver after oral dosing. (b) Replace-

ment of the mifepristone-derived fragment of 16câ is

also not tolerated. In an attempt to prepare an analogue

of 16câ that lacked affinity for the progesterone recep-

tor, we conjugated the known GR-selective steroid RU-

43044 using the same â-ether linkage to give 22 (Figure

9). This compound retains most of the potency and GR-

selectivity of the parent but despite this has no activity

in the rat prednisolone challenge model (see Table 1).

These experiments make it clear that it is na¨ıve to

interpret the pharmacology of these bile acid-conjugated

despite multiple passes through the enterohepatic loop.

By comparison,³³segmental absorption and hepatic

extraction studies of 16câ indicate that the single-pass

intestinal absorption efficiency is 20-30% (F_A∼ 0.25),

and that the hepatic capture efficiency is ∼50% (F_H

∼

0.5). Thus, the systemic oral bioavailability (F ) F_A*[1

- F_H]) of this compound is relatively low, at 10-15%;

though liver levels are substantially higher (Figure 8).

As a consequence, >98% of the drug dose is excreted

within 24 h, largely through the bile. The absence of

an efficient enterohepatic recirculation mechanism makes

16câ distinctly different from a typical bile acid.

4224 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

filtered solution of NMO (2.6 g, 22 mmol) in CH₂Cl₂(10 mL)

which had been dried over Na₂SO₄. The resultant solution was

stirred under nitrogen atmosphere in an ice bath. A solution

of TPAP (160 mg) in CH₂Cl₂(2 mL) was added dropwise over

10 min. The resultant solution was stirred at 0 °C for 20 min;

reaction was quenched by addition of 10% sodium bisulfite (20

mL). The solution was warmed to room temperature and

stirred an additional 10 min, then the product was partitioned

between EtOAc and water. The aqueous layer was back-

extracted with EtOAc; the combined organics were washed

with pH 7 phosphate buffer and brine. The organic layer was

filtered through Celite; the filtrate was concentrated in vacuo

and the product precipitated. The product was filtered and

dried to give 1.3 g (63%) of formamide.

Meth od A2. The crude formamide was dissolved in metha-

nol (25 mL); 20 mL of 10% aqueous HCl was added slowly,

and the resultant reaction mixture stirred at ambient tem-

perature for 40 h. The pH of the reaction mixture was adjusted

to pH 7 by slow addition of 10% Na₂CO₃solution; a solid

product precipitates. The product was isolated by filtration,

dissolved in CH₂Cl₂(20 mL), and dried over Na₂SO₄. The

solvent was removed in vacuo to give 1.1 g (86%) of 2 as a

white solid.

Meth od A3. A sample of this material (0.050 g, 0.12 mmol)

in pyridine (1 mL) was treated with acetic anhydride (0.014

g, 0.14 mmol) and heated at 60 °C for 2 h. The reaction was

cooled to room temperature and solvent was removed in vacuo.

The residue was dissolved in 1.5 mL of a 1:1 mixture of DMSO/

MeOH and purified by preparative reversed-phase HPLC (CH₃-

CN/H₂O) to give 3.

Meth od B. P r ep a r a tion of Ch olic Acid Asp a r ta m id e

4. Meth od B1. A solution of compound 2 (0.90 g, 2.2 mmol)

and N-tert-butoxycarbonylglutamic acid methyl ester (0.63 g,

2.4 mmol) in DMF (10 mL) was cooled at 0 °C; diisopropyl-

ethylamine (0.83 mL, 4.8 mmol), and TBTU (0.77 g, 2.4 mmol)

were added sequentially. The reaction was allowed to reach

room temperature, and the stirring was continued for 2 days.

The mixture was poured into ice-water with stirring. The

precipitate was filtered and washed with water. The dried solid

was purified by silica gel chromatography (2-5% MeOH/CH₂-

Cl₂) to give 0.93 g of the desired product 4.

Meth od B2. The above compound (0.92 g, 1.4 mmol) was

treated with freshly prepared saturated HCl/EtOAc for 1 h,

until no starting material remained when monitored by TLC.

All volatile materials were removed under vacuum to give 0.75

g the desired product as a yellowish solid. The crude product

(0.74 g, 1.2 mmol) was combined with cholic acid (0.65 g, 1.6

mmol) in DMF (10 mL); the resultant solution was cooled to 0

°C, and diisopropylethylamine (0.83 mL, 4.8 mmol) and TBTU

(0.51 g, 1.6 mmol) were added. The reaction was allowed to

reach room temperature, and the stirring was continued for 2

days. The mixture was poured into ice-water with stirring.

The precipitate was filtered, washed with water, and dried.

The residue was purified by silica gel column (5-10% MeOH/

CH₂Cl₂) to give 0.29 g of a yellowish solid. This material (0.29

g, 0.30 mmol) was treated with aqueous lithium hydroxide (2

M, 8 mL) and THF (8 mL) for 2 h, until no starting material

remained as monitored by TLC. All solvents were removed in

vacuo, and the residue was purified by silica gel chromatog-

raphy (10-15% MeOH/CH₂Cl₂, in the presence of 2% acetic

acid) to give 0.23 g of 15.

Meth od C. P r ep a r a tion of Ca r ba m a te-Lin k ed Der iva -

tives of Ch olic Acid . Meth od C1. Preparation of compound

5. To a solution of cholic acid methyl ester (2.0 g, 4.7 mmol)

stirring in pyridine (5 mL) at 0 °C was added 4-nitrophenyl-

chloroformate (1.4 g, 7.1 mmol). The mixture was stirred at 0

°C for 30 min and at room temperature for 1 h. The reaction

mixture was partitioned between EtOAc (50 mL) and 1 N HCl

(50 mL). The aqueous layer was extracted with EtOAc (3 ×

30 mL). The combined organic layers were washed with brine

(50 mL), dried over MgSO₄, and concentrated in vacuo. The

crude carbonate was dissolved in 1,4-dioxane (5 mL) and water

(5 mL) and was treated with ethanolamine (0.58 g, 9.5 mmol)

and 4-methylmorpholine (0.96 g, 9.5 mmol). The mixture was

GR antagonists purely in the context of our original

address-and-message hypothesis.

Con clu sion s

Through a strategy of bile acid conjugation, it has

been possible to prepare molecules that are potent

inhibitors of glucocorticoid binding in vitro, demonstrate

GR antagonist activity in cellular assays, and are

selective for hepatic versus systemic glucocorticoid

blockade in several animal models. The optimization of

these properties requires compound screening in both

in vitro and in vivo models and is facilitated by informa-

tion from an X-ray crystallographic study of mifepris-

tone binding to a GR fragment. From this analysis,

conjugate 16câ (A-348441) has been selected for in-

depth evaluation as an antidiabetic agent acting through

the reduction of hepatic glucose output.

The ability of 16câ to lower plasma glucose levels in

diabetic ob/ob mice provides the first direct evidence that

hepatic glucocorticoid blockade is sufficient to suppress

hepatic glucose production. While this hypothesis was

central to our efforts in this area, the absence of a liver-

selective agent has made it impossible to prove until

now. The glucose-lowering effect of 16câ persists through-

out the day, even when the drug is given once-daily, and

is reflected in profound changes in HbA1c levels, which

provide the best assessment of long-term glycemic

control. In addition, the compound reduces the hyper-

lipidemia that is often associated with Type 2 diabetes.

As such, it presents a compelling profile as a novel

treatment for diabetes and the metabolic syndrome.

Exp er im en ta l Section

Abbr evia tion s. 9-BBN, 9-borabicicylo[3.3.1]nonane; DBU,

1,8-diazabicyclo[5.4.0]undec-7-ene; DMAP, 4-(dimethylamino)-

pyridine; DMF, dimethylformamide; EtOAc, ethyl acetate;

KOtBu, potassium tert-butoxide; NBS, N-bromosuccinimide;

PPh₃, triphenylphosphine; TBTU, O-(1H-benzotriazol-1-yl)-

N,N,N′,N′-tetramethyluronium tetrafluoroborate; TFA, tri-

fluoroacetic acid; TPAP, tetra-n-propylammonium perruthen-

ate.

Com p ou n d P r ep a r a tion . Unless otherwise specified, all

solvents and reagents were obtained from commercial suppli-

ers and used without further purification. All reactions were

performed under nitrogen atmosphere unless specifically

noted. Normal-phase flash chromatography was done using

Merck silica gel 60 (230-400 mesh) from E. M. Science or was

performed on a hybrid system employing Gilson components

and Biotage prepacked columns. Reversed-phase chromatog-

raphy was performed using a Gilson 215 solvent handler-

driven HPLC system with a Zorbax SB-C18, 5 µm 21.2 × 250

mm column attached or using a Waters system with a Zorbax

SB-C8, 5 µm 20 × 250 mm column. ¹H NMR spectra were

recorded at 300 and 500 MHz; all values are referenced to

tetramethylsilane as internal standard and are reported as

shift (multiplicity, coupling constants, proton count). Mass

spectral analysis is accomplished using fast atom bombard-

ment (FAB-MS), electrospray (ESI-MS), or direct chemical

ionization (DCI-MS) techniques. Analytical LC-MS was per-

formed on a Shimadzu HPLC system with a Zorbax SB-C8, 5

µm 2.1 × 50 mm (Agilent technologies) column, with a PE

Sciex, API 150EX single quadropole mass spectrometer, at a

flow rate of 1 mL/min (0.05% NH₄OAc-buffer:MeCN and 0.05%

HCOOH in H₂O:MeCN). All elemental analyses are consistent

with theoretical values to within (0.4% unless indicated.

Meth od A. Dem eth yla tion /Acetyla tion of Mifep r iston e.

Meth od A1. To a solution of (11â,17R)-17-hydroxy-11-(4-

(dimethylamino)phenyl)-17-prop-1-ynylestra-4,9-dien-3-one (1,

mifepristone; 2.0 g, 4.7 mmol) in CH₂Cl₂(10 mL) was added a

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4225

stirred at 50 °C for 16 h. The reaction mixture was partitioned

between EtOAc (20 mL) and 1 N HCl (20 mL). The aqueous

layer was extracted with EtOAc (3 × 20 mL). The combined

organic layers were washed with brine (15 mL), dried over

MgSO₄, and concentrated in vacuo. A portion of the resultant

alcohol (0.45 g, 0.80 mmol) was dissolved in DMF (5 mL) at

0°C; PPh₃(0.32 g, 1.2 mmol) was added, followed by NBS (0.22

g, 1.2 mmol). The mixture was stirred at 0 °C for 1 h and at

room temperature for 2 h. The reaction mixture was parti-

tioned between EtOAc (10 mL) and 1 N HCl (10 mL). The

aqueous layer was extracted with EtOAc (3 × 10 mL). The

combined organic layers were washed with brine (15 mL),

dried over MgSO₄, and concentrated in vacuo. The crude

material was purified by flash chromatography on silica gel

eluting with 20% acetone/hexanes and yielded bromide 5 as a

white solid (0.31 g).

Meth od C2. P r ep a r a tion of Com p ou n d 6. To a solution

of the crude carbonate described in Method C1 (0.50 g, 0.85

mmol) in 1,4-dioxane (2 mL) and water (2 mL) were added

glycine (0.13 g, 1.7 mmol) and 4-methylmorpholine (0.17 g, 1.7

mmol). The mixture was stirred at 50 °C for 16 h. The reaction

mixture was partitioned between EtOAc (10 mL) and 1 N HCl

(10 mL). The aqueous layer was extracted with EtOAc (3 ×

10 mL). The combined organic layers were washed with brine

(15 mL), dried over MgSO₄, and concentrated in vacuo to yield

acid 6 as a light yellow oil.

Meth od C3. P r ep a r a tion of Com p ou n d 10. To a solution

of cholic acid methyl ester (4.0 g, 9.5 mmol) in pyridine (10

mL) were added DMAP (1.3 g, 10 mmol) and acetic anhydride

(1.1 g, 10 mmol). The mixture was stirred at 0 °C for 2 h and

at room temperature for 8 h. The reaction mixture was

partitioned between EtOAc (50 mL) and 1 N HCl (50 mL). The

aqueous layer was extracted with EtOAc (3 × 30 mL). The

combined organic layers were washed with brine (50 mL),

dried over MgSO₄, and concentrated in vacuo. The crude

product, a light yellow solid, was dissolved in pyridine (15 mL)

at 0 °C and treated with 4-nitrophenylchloroformate (2.1 g,

10 mmol). The reaction was stirred at 0 °C for 30 min and at

room temperature for 1 h. The reaction mixture was parti-

tioned between EtOAc (50 mL) and 1 N HCl (50 mL). The

aqueous layer was extracted with EtOAc (3 × 30 mL). The

combined organic layers were washed with brine (50 mL),

dried over MgSO₄, and concentrated in vacuo. The crude

carbonate was dissolved in 1,4-dioxane (15 mL) and water (15

mL) and was treated with ethanolamine (1.2 g, 19 mmol) and

4-methylmorpholine (1.9 g, 19 mmol). The reaction was stirred

at 50 °C for 16 h. The reaction mixture was partitioned

between EtOAc (50 mL) and 1 N HCl (50 mL). The aqueous

layer was extracted with EtOAc (3 × 20 mL). The combined

organic layers were washed with brine (50 mL), dried over

MgSO₄, and concentrated in vacuo. The crude material was

purified by flashing chromatography on silica gel eluting with

40% acetone/hexanes and yielded the alcohol as a white solid

(2.6 g, 50% yield over three steps). Reaction of the alcohol

under the NBS/PPh₃conditions described above yielded the

corresponding bromide 10 (1.2 g, 45% yield).

dried over MgSO₄, and concentrated in vacuo, yielding the C3-/

C7-diacetate as a light yellow solid. This material was dis-

solved in pyridine (15 mL) at 0 °C and was treated with

4-nitrophenylchloroformate (2.1 g, 10 mmol). The reaction was

stirred at 0 °C for 30 min and at room temperature for 1 h.

The reaction mixture was partitioned between EtOAc (50 mL)

and 1 N HCl (50 mL). The aqueous layer was extracted with

EtOAc (3 × 30 mL). The combined organic layers were washed

with brine (50 mL), dried over MgSO₄, and concentrated in

vacuo. The resultant crude carbonate 24 was dissolved in 1,4-

dioxane (15 mL) and water (15 mL) and was treated with

glycine (1.4 g, 19 mmol) and 4-methylmorpholine (1.9 g, 19

mmol). The reaction was stirred at 50 °C for 16 h. The reaction

mixture was partitioned between EtOAc (50 mL) and 1 N HCl

(50 mL). The aqueous layer was extracted with EtOAc (3 ×

20 mL). The combined organic layers were washed with brine

(50 mL), dried over MgSO₄, and concentrated in vacuo, yielding

crude acid 13 as a yellow oil.

Meth od D. P r ep a r a tion of Am in e-Lin k ed Der iva tives

of Ch olic Acid . Meth od D1. P r ep a r a tion of Com p ou n d

9a . 3-Oxo-cholic acid methyl ester, prepared according to the

method of Tserng²⁶(550 mg, 1.3 mmol), was dissolved in 10

mL of acetonitrile and stirred with 5.0 mL of 2.0 M methanolic

methylamine for 30 min at ambient temperature. 110 mg (1.6

mmol) of NaCNBH₃was added, and stirring was continued

for 2 h at room temperature. 5 drops of acetic acid was added;

stirring was continued for an additional 1h, until TLC indi-

cated that no starting ketone remained. 10 mL of water was

added; the resultant mixture was extracted with ethyl acetate

and dried over Na₂SO₄. This crude product (540 mg, 95%) was

carried forward without further purification. 300 mg (0.69

mmol) of the crude amine is combined with 3 equiv of Hunig’s

base and 130 mg (0.69 mmol) of 4-bromobutyryl chloride in

acetonitrile (6.0 mL); the resultant mixture was stirred

overnight in a sealed bottle at ambient temperature. The

resultant solution was carried forward without workup or

purification.

Compounds 9b and 9c are prepared using the above

procedures, substituting 2-bromoethyl isocyanate or 3-bro-

mopropanesulfonyl chloride for 4-bromobutyryl chloride.

Meth od E. P r ep a r a tion of Eth er -Lin k ed Der iva tives

of Ch olic Acid . Meth od E1. P r ep a r a tion of Com p ou n d

8a . A solution of 1.4 g (3.0 mmol) of 7R,12R-dihydroxy-3R-2-

hydroxyethoxy-5â-cholanic acid (prepared according to the

methods of Wess et al.¹⁹) in 5 mL of DMF, stirring at 0 °C,

was treated with PPh₃(0.63 g, 3.7 mmol) followed by the

addition of NBS (0.95 g, 3.7 mmol). The mixture was stirred

at 0 °C for 1 h and at ambient temperature for 2 h. The

reaction mixture was partitioned between EtOAc (20 mL) and

1 N HCl (20 mL). The aqueous layer was extracted with EtOAc

(3 × 20 mL). The combined organic layers were washed with

brine (50 mL), dried over MgSO_4,and concentrated in vacuo.

The crude material was purified by flash chromatography on

silica gel eluting with 20% acetone/hexanes and yielded

bromide 8a as a white solid (0.95 g, 74% yield).

Meth od E2. Preparation of compound 8b. A mixture of

7R,12R-dihydroxy-3R-2-hydroxyethoxy-5â-cholanic acid (pre-

pared according to the methods of Wess et al.¹⁹) (7.1 g, 13

mmol), DMF (20 mL) and allyl bromide (20 mL, 230 mmol)

was heated to 120 °C for 16 h. The reaction mixture was

partitioned between EtOAc (100 mL) and 1 N HCl (100 mL).

The aqueous layer was extracted with EtOAc (3 × 100 mL).

The combined organic layers washed with brine (150 mL),

dried over MgSO₄, and concentrated in vacuo. The crude

product was dissolved in water (10 mL) and diethyl ether (10

mL) and treated with osmium tetroxide (30 mg) and sodium

periodate (5.4 g, 25 mmol). The mixture was stirred at ambient

temperature for 16 h. The reaction mixture was partitioned

between EtOAc (50 mL) and sodium thiosulfate (2 M solution,

50 mL). The aqueous layer was extracted with EtOAc (3 × 10

mL). The combined organic layers were washed with brine (50

mL), dried over MgSO₄, and concentrated in vacuo. The crude

material was purified by flash chromatography on silica gel

eluting with 20% acetone/hexanes and yielded the aldehyde

Meth od C4. P r ep a r a tion of Com p ou n d 12. To a solution

of the crude carbonate described in Method C3 (0.50 g, 0.79

mmol) in 1,4-dioxane (2 mL) and water (2 mL) were added

glycine (0.13 g, 1.7 mmol) and 4-methylmorpholine (0.17 g, 1.7

mmol). The reaction was stirred at 50 °C for 16 h. The reaction

mixture was partitioned between EtOAc (10 mL) and 1 N HCl

(10 mL). The aqueous layer was extracted with EtOAc (3 ×

10 mL). The combined organic layers were washed with brine

(15 mL), dried over MgSO₄, and concentrated in vacuo, yielding

crude acid 12 as a light yellow oil.

Meth od C5. P r ep a r a tion of Com p ou n d 13. To a solution

of cholic acid methyl ester (4.0 g, 9.5 mmol) in pyridine (10

mL) were added DMAP (2.5 g, 21 mmol) and acetic anhydride

(2.1 g, 21 mmol). The reaction was stirred at 0 °C for 2 h and

at room temperature for 8 h. The reaction mixture was

partitioned between EtOAc (50 mL) and 1 N HCl (50 mL). The

aqueous layer was extracted with EtOAc (3 × 30 mL). The

combined organic layers were washed with brine (50 mL),

4226 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

as a white solid (6.4 g, 85% yield over two steps). The aldehyde

was combined with sodium borohydride (0.54 g, 14 mmol) in

THF (30 mL) and ethanol (10 mL) at ambient temperature.

After stirring overnight, the reaction mixture was concentrated

in vacuo and the crude alcohol was purified by flash chroma-

tography on silica gel eluting with 40% acetone/hexanes. The

resultant alcohol (4.3 g, 61% yield) was converted to the

corresponding bromide 8b using the procedures of Method E1.

1:9 to obtain tosylate 11 as a clear oil. The product was used

immediately in the next reaction step.

Meth od G. Preparation of compound 14. 3-Oxo-cholic acid

methyl ester, prepared according to the method of Tserng²⁶

(17 g, 39 mmol), was dissolved in 125 mL of dichloromethane;

17 mL of Hu¨nig’s base was added with stirring, followed the

slow addition of 8.7 mL (110 mmol, 3 eq) of chloromethyl

methyl ether. The resultant mixture was warmed overnight

at 45 °C. Solvents were removed in vacuo; the residue was

partitioned between EtOAc and aqueous 1 N H₃PO₄. The

organic phase was washed with brine, dried over Na₂SO₄, and

concentrated. The crude product was purified by silica gel

chromatography (gradient elution with 25% to 35% ethyl

acetate/hexane) to afford the bis-MOM ether (17 g, 86% yield).

A sample of this material (9.6 g) was dissolved in 100 mL of

dry THF and cooled in an ice bath. Allyl bromide (2.3 mL) was

added, followed by 2.6 g of KOtBu. The mixture was stirred

for 90 min, then the reaction was quenched by addition of

aqueous 1 N H₃PO₄. The mixture was extracted with EtOAc;

the organic phase was washed with brine, dried over Na₂SO₄,

and concentrated. The crude product was dissolved in 120 mL

of MeOH, 1.1 g of NaBH₄pellets was added, and the resultant

mixture was stirred at ambient temperature for 4 h. Solvents

were removed in vacuo; the residue was partitioned between

EtOAc and aqueous 1 N H₃PO₄. The organic phase was washed

with brine, dried over Na₂SO₄, and concentrated. The crude

product was purified by silica gel chromatography (gradient

elution with 33% to 40% ethyl acetate/hexane). The resultant

C3R-alcohol was dissolved in 50 mL of dichloromethane; 5 mL

of Hu¨nig’s base was added with stirring, followed the slow

addition of 1.8 mL of chloromethyl methyl ether. The resultant

mixture was warmed for 3 h at 45 °C. Solvents were removed

in vacuo; the residue was partitioned between EtOAc and

water. The organic phase was washed with brine, dried over

Na₂SO₄, and concentrated. The crude product was purified by

silica gel chromatography (25% ethyl acetate/hexane) to afford

the tris-MOM ether (3.8 g). To a well-stirred solution of this

compound (1.2 g, 2.0 mmol) in diglyme (4 mL) at 0 °C was

added borane-THF solution (0.66 mL, 1.0 mmol). The result-

ant solution was stirred at 0 °C for 1 h and warmed to ambient

temperature for 3 h. The reaction mixture was then reacted

with trimethylamine N-oxide dihydrate (0.44 g, 4.0 mmol) at

60 °C for 2 h and 70 °C overnight. The mixture was partitioned

between H₂O and EtOAc. The organic layer was washed with

brine, dried over Na₂SO₄, and concentrated. The crude product

was purified by silica gel chromatography (50% ethyl acetate/

hexane) to afford the corresponding alcohol (0.80 g, 66% yield).

A sample of this compound (20 mg, 0.033 mmol) in acetonitrile

(0.5 mL) was stirred overnight at ambient temperature with

methanesulfonyl chloride (4.5 mg, 0.039 mmol) and Hunig’s

base (30 mg, 0.23 mmol). The resultant solution was used in

subsequent procedures without further workup or purification.

Meth od E3. P r ep a r a tion of Com p ou n d 7a . A solution

of 7R,12R-dihydroxy-3â-2-hydroxyethoxy-5â-cholanic acid (pre-

pared according to the methods of Wess et al.¹⁹) (2.5 g, 5.4

mmol) in DMF (10 mL) at 0 °C was treated with PPh₃(2.1 g,

8.0 mmol) followed by the addition of NBS (1.4 g, 8.0 mmol).

The mixture was stirred at 0 °C for 1 h and at ambient

temperature for 2 h. The reaction mixture was partitioned

between EtOAc (20 mL) and 1 N HCl (20 mL). The aqueous

layer was extracted with EtOAc (3 × 20 mL). The combined

organic layers were washed with brine (50 mL), dried over

MgSO₄, and concentrated in vacuo. The crude material was

purified by flash chromatography on silica gel eluting with 20%

acetone/hexanes and yielded bromide 7a as a white solid (2.2

g, 79% yield).

Compounds 7b, 7c, and 7d are also prepared according to

method E3, by substituting chenodeoxycholic, lithocholic, and

deoxycholic acid methyl esters for cholic acid methyl ester in

the procedure of Wess et al.¹⁹Compounds 7e and 7f are also

prepared according to Method E3, by substituting propylene

glycol and diethylene glycol for ethylene glycol in the procedure

of Wess et al.¹⁹

Meth od F . P r ep a r a tion of Com p ou n d 11. A solution of

allyl iodide (10 mL, 110 mmol) in dry THF (45 mL) was

dropwise added to commercial zinc-dust (7.5 g, 110 mmol) at

ambient temperature and stirred for 90 min. A solution of

3-oxo-cholic acid methyl ester (16 g, 38 mmol) in dry THF (45

mL) was added slowly; the resultant mixture was stirred at

ambient temperature for 3 days. Some water was added; the

mixture was filtered and concentrated. The residue was taken

up in EtOAc and partitioned with water; the organic layer was

washed with brine and dried over Na₂SO₄. The crude product

was purified by flash chromatography on silica gel, eluting

with 1:1 EtOAc/n-heptane, to yield a white solid. The stereo-

chemistry of addition (allyl group on â-face) was confirmed by

NOE studies. The resultant triol (11 g, 23 mmol) was combined

with imidazole (7.8 g, 110 mmol) in dry DMF (150 mL) and

flushed with dry nitrogen. Trimethylsilyl chloride (12 mL, 92

mmol) was added, and the mixture was stirred overnight at

ambient temperature. Some water was added and the reaction

mixture was concentrated in vacuo. The residue was taken

up in EtOAc/Et₂O (1:1), diluted with water, and extracted, and

the organic layer was dried over Na₂SO₄then concentrated.

The crude product was purified by flash chromatography

(dichloromethane/n-heptane 4:6) to afford the tris-silyl ether

as a clear oil. This material (0.61 g, 0.90 mmol) was dissolved

in dry THF (10 mL); 9-BBN (0.5 M in THF, 3.7 mL, 1.8 mmol)

was added dropwise, and the resultant solution was stirred

at ambient temperature until hydroboration was nearly com-

plete, as judged by TLC. The reaction was cooled in a water

bath; a mixture of aq KOH (5 mL, 9.0 mmol) and aq H₂O₂(1.8

mL, 27 mmol) was added slowly, keeping the reaction tem-

perature below ambient. After 30 min, the mixture was diluted

with water and pH adjusted to ∼7 with sat. NH₄Cl followed

by extraction with dichloromethane. The organic layer was

concentrated in vacuo. The crude product was purified by flash

chromatography on silica gel (EtOAc/n-heptane, gradient 0:1-

3:7) to afford the alcohol as a clear oil. This material (1.0 g,

1.4 mmol) was dissolved in dry dichloromethane (25 mL) and

cooled to 0 °C. DMAP (cat.) and triethylamine (2.0 mL, 14

mmol) were added, followed by p-toluenesulfonyl chloride (0.55

g, 2.9 mmol) added portionwise over 2 h. After 5 h stirring,

the reaction was quenched with sat. aqueous NH₄Cl and water,

followed by extraction with dichloromethane. After solvent

removal in vacuo, the crude product was purified by flash

chromatography on silica gel, eluting with EtOAc/n-heptane

Meth od H. Cou p lin g of F r a gm en ts. Meth od H1. Acy-

la tion -Ba sed Cou p lin gs. P r ep a r a tion of Com p ou n d 16Ra .

To a solution of acid 6 (0.53 g, 1.0 mmol) in DMF (5 mL) was

added compound 2 (0.50 g, 1.2 mmol), followed by TBTU (0.48

g, 1.5 mmol). The mixture was warmed to 50 °C for 5 h. The

reaction mixture was partitioned between EtOAc (20 mL) and

1 N HCl (20 mL). The aqueous layer was extracted with EtOAc

(3 × 20 mL). The combined organic layers were washed with

brine (20 mL), dried over MgSO₄, and concentrated in vacuo.

The crude product was dissolved in 1,4-dioxane (5 mL) and

was treated with aqueous LiOH solution (2 M, 2 mL); the

resultant mixture was stirred at ambient temperature for 16

h. The reaction mixture was partitioned between EtOAc (10

mL) and 1 N HCl (30 mL or until acidic). The aqueous layer

was extracted with EtOAc (3 × 10 mL). The combined organic

layers were washed with brine (20 mL), dried over MgSO₄,

and concentrated in vacuo. The crude product was purified

with reversed phase HPLC (Zorbax XDB-C₁₈, 21.2 × 250 mm,

water/acetonitrile/0.1% TFA) and yielded conjugate 16Ra .

Compounds 15, 19a , and 20 were also prepared according

to method H1.

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4227

1

Meth od H2. Alk yla tion -Ba sed Cou p lin gs. P r ep a r a tion

of Com p ou n d 16âc. A solution of bromide 7a (1.0 g, 1.8 mmol)

and compound 2 (0.87 g, 2.1 mmol) in acetonitrile (5 mL) in a

dry pressure tube equipped with a stir bar was treated with

sodium iodide (0.03 g, 0.20 mmol) and diisopropylethylamine

(0.68 g, 5.23 mmol). The reaction was heated to 100 °C with

an oil bath for 16 h. The reaction mixture was partitioned

between EtOAc (20 mL) and 1 N HCl (20 mL). The aqueous

layer was extracted with EtOAc (3 × 20 mL). The combined

organic layers were washed with brine (50 mL), dried over

MgSO₄, and concentrated in vacuo. The crude product was

dissolved in 1,4-dioxane (10 mL) and was treated with LiOH

solution (2M, 5 mL); the mixture was stirred at room temper-

ature for 16 h. The reaction mixture was partitioned between

EtOAc (50 mL) and 1 N HCl (100 mL or until acidic). The

aqueous layer was extracted with EtOAc (3 × 20 mL). The

combined organic layers were washed with brine (50 mL),

dried over MgSO₄, and concentrated in vacuo. The crude

product was purified with reversed phase HPLC (Zorbax XDB-

Com p ou n d 16Rc. H NMR (400 MHz, CD₃OD) δ 7.47 (s, 4

H), 5.76 (s, 1 H), 4.59 (d, J ) 7.6 Hz, 1 H), 3.96 (t, J ) 2.4 Hz,

1 H), 3.79 (q, J ) 2.8 Hz, 1 H), 3.74 (m, 2 H), 3.71 (t, J ) 5.2

Hz, 1 H), 3.32-3.53 (m, 2 H), 3.26 (m, 1 H), 2.68-3.09 (m, 4

H), 2.12-2.68 (m, 12 H), 1.86 (s, 3 H), 1.66-2.12 (m, 13 H),

1.21-1.66 (m, 14 H), 1.14 (m, 1 H), 1.03 (d, J ) 6.6 Hz, 3 H),

0.89 (s, 3 H), 0.86 (m, 1 H), 0.72 (s, 3 H), 0.49 (s, 3 H); MS

(ESI) m/z 850 (M + H)⁺; HRMS calcd for C₅₄H₇₆NO₇

:

850.5616; Found: 850.5606.

1

Com p ou n d 16Rd . H NMR (300 MHz, CD₃OD) δ 7.03 (d,

J ) 9 Hz, 2H), 6.70 (d, J ) 9 Hz, 2H), 5.73 (s, 1H), 4.39 (d, J

) 7.5 Hz, 1H), 4.97 (s, 2H), 4.80 (s, 2H), 3.35 (m, 2H), 2.88 (d,

J ) 4.5 Hz, 3H), 2.84 (d, J ) 9 Hz, 3H), 2.65-1.05 (envelope,

46H), 1.85 (s, 3H), 1.03 (d, J ) 6 Hz, 3H), 0.94 (d, J ) 4.5 Hz,

+

3H), 0.72 (s,3H), 0.54 (s, 3H). MS (ESI) m/z 905 (M + H)

.

HRMS calcd for C₅₇H₈₁N₂O₇: 905.6038. Found: 905.6046.

Com p ou n d 16Re. ¹H NMR (300 MHz, CD₃OD) δ 7.02 (d, J

) 9 Hz, 2H), 6.73 (d, J ) 9 Hz, 2H), 5.72 (s, 1H), 4.48 (d, J )

7.5 Hz, 1H), 3.97 (s, 2H), 3.80 (s, 2H), 3.90 (m, 1H), 2.92 (s,

3H), 2.67 (s, 3H), 2.65-1.06(envelope, 46H), 1.85 (s, 3H), 1.03

(d, J ) 6 Hz, 3H), 0.93 (s, 3H), 0.72 (s, 3H), 0.54 (s, 3H). MS

(ESI) m/z 906 (M + H)⁺. Anal. Calcd for C₅₆H₇₉N₃O_7.1.0 H₂O:

C, 72.77; H, 8.83; N, 4.55. Found: C, 72.93; H, 9.07; N, 4.58.

HRMS calcd for C₅₆H₈₀N₃O₇: 906.5991; Found: 906.5966.

C

₁₈, 21.2 × 250 mm, water/acetonitrile/0.1% TFA) yielding

conjugate 16âc.

Compounds 16Rb, 16Rc, 16Rd , 16Re, 16Rf, 16Rh , 16âg,

16âh , 16âi, 16âj, 16âk , 17, 18, and 19b were also prepared

according to method H2.

1

Com p ou n d 16Rf. H NMR (300 MHz, CD₃OD) δ 7.05 (d, J

Meth od I. P r ep a r a tion of Glycoch ola te a n d Ta u r o-

ch ola te Con ju ga tes. Meth od I1. P r ep a r a tion of Com -

p ou n d 21b. To a solution of compound 16âc (10 mg, 0.01

mmol) in DMF (1 mL) were added TBTU (7 mg, 0.02 mmol)

and taurine (3 mg, 0.02 mmol). The mixture was stirred at 50

°C for 1 h. The reaction mixture was partitioned between

EtOAc (2 mL) and 1 N HCl (2 mL). The aqueous layer was

extracted with EtOAc (3 × 2 mL). The combined organic layers

were washed with brine (5 mL), dried over MgSO₄, and

concentrated in vacuo. The crude product was purified with

reverse phase HPLC (Zorbax XDB-C₁₈, 21.2 × 250 mm, water/

acetonitrile/0.1% TFA) yielded 4 mg of compound 21b.

) 9 Hz, 2H), 6.72 (d, J ) 9 Hz, 2H), 5.73 (s, 1H), 4.40 (d, J )

7.5 Hz, 1H), 3.96 (s, 1H), 3.80 (s, 1H), 3.53-3.42 (m, 3H), 3.03

(t, J ) 7.5 Hz, 2H), 2.88 (s, 3H), 2.85 (m, 1H), 2.79 (s, 3H),

2.66-1.67 (envelop, 33H), 1.59-1.27 (m, 14H), 1.03 (d, J ) 6

Hz, 3H), 0.93 (s, 3H), 0.72 (s, 3H), 0.53 (s, 3H); MS(ESI) m/z

941 (M + 1)⁺, 939 (M - 1)^-. Anal. Calcd for C₅₆H₈₀N₂O₈S. 1.0

EtOAc: C, 70.01; H, 8.62; N, 2.72. Found: C, 69.65; H, 8.58; N,

2.64. HRMS calcd for C₅₆H₈₁N₂O₈S : 941.5708; Found: 941.5715.

Com p ou n d 16Rh . ¹H NMR (400 MHz, CD₃OD) δ 7.32-7.66

(m, 4 H), 5.77 (s, 1 H), 4.61 (m, 1 H), 3.96 (m, 2 H), 3.83 (m, 2

H), 3.73 (m, 3 H), 3.63 (m, 5 H), 3.53 (m, 5 H), 3.27 (m, 4 H),

2.85 (m, 2 H), 2.66 (m, 4 H), 2.50 (m, 3 H), 2.12-2.42 (m, 9

H), 1.92-2.12 (m, 2 H), 1.69-1.92 (m, 5 H), 1.58 (m, 7 H),

1.26-1.50 (m, 6 H), 1.13 (m, 2 H), 1.02 (d, J ) 5.2 Hz, 3 H),

0.94 (s, 3 H), 0.72 (s, 3 H), 0.48 (s, 3 H); MS (ESI) m/z 894 (M

+ H)⁺892 (M - H)^-; HRMS calcd for C₅₆H₈₀NO₈: 894.5878;

Found: 894.5901.

Compound 21a was prepared using method I.

Com p ou n d Ch a r a cter iza tion . Compound 3. ¹H NMR (300

MHz, CDCl₃) δ 7.24 (d, J ) 8.5 Hz, 1 H), 7.10 (d, J ) 8.5 Hz,

1 H), 5.80 (s, 1 H), 4.44 (d, J ) 6.2 Hz, 1 H), 3.26 (s, 3 H), 2.80

(dt, J ) 7.0 Hz, 2.5 Hz, 1 H), 2.64-2.59 (m, 2 H), 2.50-2.10

(m, 6 H), 2.08-1.25 (m, 7 H), 1.91 (s, 3 H), 1.85 (s, 3 H), 0.50

(s, 3 H). MS (ESI(+)Q1) MS m/z 458 (M + H)⁺. Anal. Calcd

for C₃₀H₃₅NO₃‚0.8 H₂O: C, 76.34; H, 7.82; N, 2.97. Found: C,

76.47; H, 7.88; N, 2.93. HRMS calcd for C₃₀H₃₆NO₃: 458.2695.

Observed: 458.2698.

Com p ou n d 16âc. ¹H NMR (500 MHz, CD₃OD) δ 7.07-7.71

(m, 4 H), 5.76 (s, 1 H), 4.57 (d, J ) 7.8 Hz, 1 H), 3.95 (s, 1 H),

3.78 (d, J ) 2.2 Hz, 1 H), 3.74 (t, J ) 4.4 Hz, 1 H), 3.44 (m, 1

H), 3.23 (m, 3 H), 2.86 (m, 1 H), 2.66 (m, 1 H), 2.50 (m, 2 H),

2.16-2.41 (m, 7 H), 2.15 (m, 2 H), 2.08 (m, 1 H), 1.95 (m, 3

H), 1.85 (m, 5 H), 1.75 (m, 5 H), 1.57 (m, 7 H), 1.40 (m, 7 H),

1.31 (m, 6 H), 1.01 (d, J ) 6.6 Hz, 3 H), 0.92 (m, 5 H), 0.71 (s,

3 H), 0.48 (s, 3 H); MS (ESI) m/z 850 (M + H)⁺848 (M - H)^-;

Anal. Calcd for C₅₄H₇₅NO₇‚1.0 H₂O: C, 74.70; H, 8.94; N, 1.61.

Found: C, 74.65; H, 8.58; N, 1.78. HRMS calcd for C₅₄H₇₆NO₇:

850.5616; found 850.5620:

Com p ou n d 15. ¹H NMR (500 MHz, CDCl₃) δ 7.18-7.35

(2×d, 4H, arom.), 5.76 (s, 1H, H-4), 4.55 (d, 1H), 4.18 (d, 1H),

3.94 (s, 1H), 3.79 (s, 1H), 3.25 (s, 3H, NCH₃), 1.86 (s, 3H), 0.99

(d, 3H, CH₃), 0.92 (s, 3H, CH₃), 0.70 (s, 3H, CH₃), 0.49 (s, 3H).

MS (ESI) m/z 936 (M + 1)⁺. Anal. Calcd for C₅₇H₇₈N₂O₉‚2.0

H₂O. 0.3 ETOAc: C, 70.06; H, 8.53; N, 2.81. Found: C, 69.85;

H, 8.30; N, 2.96. HRMS calcd for C₅₇H₇₉N₂O₉: 935.5780. Found:

935.5785.

Com p ou n d 16Ra . ¹H NMR (500 MHz, CD₃OD) δ 7.38 (d, J

) 7.0 Hz, 2 H), 7.26 (d, J ) 8.1 Hz, 2 H), 5.76 (s, 1 H), 4.56 (d,

J ) 6.6 Hz, 1 H), 4.36 (m, 1 H), 3.96 (m, 1 H), 3.79 (m, 1 H),

3.55 (m, 2 H), 3.23 (m, 3 H), 2.87 (m, 1 H), 2.54-2.74 (m, 3

H), 2.20-2.54 (m, 9 H), 2.17 (m, 2 H), 2.09 (m, 1 H), 1.77 (m,

6 H), 1.66 (m, 3 H), 1.55 (m, 6 H), 1.27-1.48 (m, 11 H), 1.25

(m, 3 H), 1.12 (m, 2 H), 1.02 (m, 5 H), 0.89 (m, 3 H), 0.80 (s, 3

H), 0.72 (s, 3 H); MS (ESI) m/z 924 (M + NH₄)⁺; 905 (M -

H)^-. Anal. Calcd for C₅₅H₇₄N₂O₉‚1.0 H₂O: C, 71.40; H, 8.28;

N, 3.03. Found: C, 71.54; H, 8.43; N, 3.08. HRMS calcd for

Com p ou n d 16âg. ¹H NMR (400 MHz, CD₃OD) δ 6.92 (d, J

) 8.5 Hz, 2 H), 6.86 (d, J ) 8.0 Hz, 2 H), 5.28 (s, 1 H), 4.08 (m,

1 H), 3.74 (m, 1 H), 3.60 (m, 1 H), 3.47 (m, 1 H), 3.29 (m, 1 H),

3.13 (m, 2 H), 3.01 (m, 1 H), 2.91 (m, 2 H), 2.72 (m, 3 H), 2.25-

2.45 (m, 1 H), 2.17 (m, 2 H), 2.02 (m, 3 H), 1.88 (m, 4 H), 1.72

(m, 4 H), 1.60 (m, 1 H), 1.48 (m, 4 H), 1.37 (m, 5 H), 1.24 (m,

6 H), 1.08 (m, 6 H), 0.78-1.00 (m, 12 H), 0.75 (m, 1 H), 0.65

(m, 1 H), 0.53 (d, J ) 6.6 Hz, 3 H), 0.42 (s, 3 H), 0.23 (s, 3 H);

MS (ESI) m/z 864 (M + H)⁺862 (M - H)^-; HRMS calcd for

C

₅₅H₇₈NO₇: 864.5773; Found: 864.5751.

Com p ou n d 16âh . ¹H NMR (300 MHz, CD₃OD): δ 7.40 (bd

s, 4H), 5.77 (s, 1H), 4.56 (d, J ) 8 Hz, 1H), 3.96 (bd s, 1H),

3.80 (bd q, J ) 3 Hz, 1H), 3.71 (m, 2H), 3.57 (bd s, 1H), 3.49

(bd s, 2H), 3.23 (s, 3H), 2.85 (dt, J ) 6,15 Hz, 1H), 2.64 (m,

2H), 2.6-1.1 (envelope, 45H), 1.86 (s, 3H), 1.03 (d, J ) 7 Hz,

3H), 0.92 (s, 3H), 0.73 (s, 3H), 0.48 (s, 3H). MS (ESI) m/z 894

(M + H)⁺, m/z 892 (M - H)^-. HRMS calcd for C₅₄H₈₀NO₈:

894.5878. Found: 894.5893.

C

₅₅H₇₄N₂O₉Na: 929.5287. Found: 929.5281.

1

Com p ou n d 16Rb. H NMR (500 MHz, CD₃OD) δ 7.31 (d,

J ) 7.8 Hz, 2 H), 7.17 (m, 2 H), 5.75 (s, 1 H), 4.38 (m, 1 H),

3.94 (m, 1 H), 3.81 (m, 2 H), 3.39 (m, 2 H), 3.23 (m, 2 H), 3.12

(m, 2 H), 2.83 (m, 3 H), 2.66 (m, 2 H), 2.12-2.58 (m, 11 H),

1.62-2.12 (m, 16 H), 1.56 (m, 6 H), 1.45 (m, 4 H), 1.02 (m, 6

H), 1.02 (d, J ) 6.6 Hz, 3 H), 0.93 (s, 3 H), 0.72 (s, 3 H), 0.50

(s, 3 H); MS (ESI) m/z 893 (M + H)⁺891 (M - H)^-; HRMS

calcd for C₅₅H₇₇N2O₈: 893.5674; found: 893.5649.

1

Com p ou n d 16âi. H NMR (300 MHz, CD₃OD): δ 7.4-7.3

(m, 4H), 5.76 (s, 1H), 4.55 (d, J ) 8 Hz, 1H), 3.95 (bd s, 1H),

3.71 (m, 2H), 3.49 (bd s, 1H), 3.21 (s, 3H), 2.85 (m, 1H), 2.64

4228 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

1

(m, 2H), 2.6-2.4 (m, 3H), 2.4-1.2 (envelope, 41H), 1.86 (s, 3H),

1.00 (d, J ) 7 Hz, 3H), 0.92 (s, 3H), 0.70 (s, 3H), 0.48 (s, 3H).

MS (ESI) m/z 834 (M + H)⁺, m/z 832 (M - H)^-. HRMS calcd

for C₅₄H₇₆NO₆: 834.5667. Found: 834.5661.

Com p ou n d 21b. H NMR (500 MHz, CD₃OD) δ 7.55 (d, J

) 9.0 Hz, 2 H), 7.50 (d, J ) 8.7 Hz, 2 H), 5.77 (s, 1 H), 4.61 (m,

1 H), 3.95 (m, 1 H), 3.80 (m, 3 H), 3.59 (m, 2 H), 3.44 (m, 1 H),

2.97 (m, 2 H), 2.86 (m, 1 H), 2.64 (m, 2 H), 2.53 (m, 3 H), 2.39

(m, 4 H), 2.03-2.30 (m, 10 H), 1.96 (m, 6 H), 1.77 (m, 6 H),

1.46-1.68 (m, 7 H), 1.39 (m, 8 H), 1.27 (m, 3 H), 1.12 (m, 2

H), 1.02 (d, J ) 6.6 Hz, 3 H), 0.94 (s, 3 H), 0.71 (s, 3 H), 0.47

(s, 3 H); MS (ESI) m/z 864 (M + H)⁺, 862 (M - H)^-; HRMS

calcd for C₅₆H₈₁N₂O₉S: 957.5657; Found: 957.5666.

1

Com p ou n d 16âj. H NMR (300 MHz, CD₃OD): δ 7.42 (bd

s, 4H), 5.76 (s, 1H), 4.57 (d, J ) 8 Hz, 1H), 3.8-3.7 (m, 3H),

3.43 (bd s, 1H), 3.25 (s, 3H), 2.87 (m, 1H), 2.65 (m, 2H), 2.6-

2.4 (m, 3H), 2.4-1.1 (envelope, 41H), 1.86 (s, 3H), 0.95 (d, J )

7 Hz, 3H), 0.93 (s, 3H), 0.69 (s, 3H), 0.47 (s, 3H). MS (ESI)

m/z 834 (M + H)⁺, m/z 832 (M - H)^-. HRMS calcd for C₅₄H₇₆

-

In Vitr o Com p ou n d Eva lu a tion . For all replicated dose-

response data recorded using the assays below, averages were

computed as the geometric mean.

NO₆: 834.5667. Found: 834.5666.

1

Com p ou n d 16âk . H NMR (300 MHz, CD₃OD): δ 7.02 (d,

J ) 9 Hz, 2H), 6.70 (d, J ) 9 Hz, 2H), 5.73 (s, 1H), 4.38 (d, J

) 7 Hz, 1H), 3.6-3.5 (m, 7H), 2.95 (s, 3H), 2.92 (m, 1H), 2.65

(m, 2H), 2.6-1.1 (envelope, 41H), 1.86 (s, 3H), 0.94 (d, J ) 7

Hz, 3H), 0.86 (s, 3H), 0.66 (s, 3H), 0.55 (s, 3H). MS (ESI) m/z

818 (M + H)⁺, m/z 816 (M - H)^-. Anal. Calcd for C₅₄H₇₅NO₅:

C, 79.27; H, 9.24; N, 1.71. Found: C, 79.06; H, 9.44; N, 1.68.

HRMS calcd for C₅₄H₇₆NO₅: 818.5718. Found: 818.5729.

GR Bin d in g Assa y. ³H-Dexamethasone (TRK 645, “³H

-dex”) was purchased from Pharmacia Amersham, Uppsala,

Sweden. Dexamethasone (“dex”) was purchased from Sigma.

The Costar 96-well polypropylene plates (3794 or 3365) were

purchased from Life Technologies AB, Ta¨by, Sweden. The

GF/B filter (1450-521), filter cassette (1450-104), MeltiLex

scintillating wax (1450-441), sample bag (1450-42), Microbeta

1450-PLUS and Microsealer 1495-021 were all purchased from

Wallac Oy, Turkku, Finland. Human glucocorticoid receptors

were extracted from Sf9 cells infected with a recombinant

baculovirus transfer vector containing the cloned hGR gene.

Recombinant baculovirus was generated utilizing the BAC-

TO-BAC expression system (Life Technologies) in accordance

to instruction from the supplier. The hGR coding sequences

were cloned into a baculovirus transfer vector by standard

techniques. The recombinant baculoviruses expressing hGR

were amplified and used to infect Sf9 cells. Infected cells were

harvested 48 h post infection. The receptors were extracted

from the cell pellet with a phosphate buffer (1 mM EDTA, 20

mM KPO₄(pH8), 8.6% Glycerol, 12 mM MTG, 20 mM Na₂-

MoO₄). The concentration of hGR in the extract was measured

1

Com p ou n d 17. H NMR (300 MHz, CD₃OD): δ 7.02 (d, J

) 8.5 Hz, 2H), 6.66 (d, J ) 8.5 Hz, 2H), 5.73 (s, 1H), 4.39 (d,

J ) 6.9 Hz, 1H), 3.96-3.92 (m, 1H), 3.81 - 3.76 (m, 1H), 3.26

(t, J ) 7.3 Hz, 2H), 2.88 (s, 3H), 2.85-2.77 (m, 1H), 2.62 (m,

2H), 2.55-2.10 (m, 11H), 2.10-1.60 (m, 14), 1.86 (s, 3H), 1.60-

1.24 (m, 15H), 1.16-1.05 (m, 1H), 1.01 (d, J ) 6.0 Hz, 3H),

0.94 (s, 3H), 0.71 (s, 3H), 0.54 (s, 3H); MS (ESI) m/z (M + H)⁺

864, (M - H)^-862. Anal. Calcd for C₅₅H₇₇NO₇‚0.5H₂O: C,

75.66; H, 9.00; N,1.61. Found: C, 75.59; H, 9.32; N, 1.43.

Com p ou n d 18. ¹H NMR (300 MHz, CDCl₃) δ 6.93-7.05 (m,

2 H), 6.57-6.70 (m, 2 H), 5.76 (s, 1 H), 4.34 (d, J ) 6.1 Hz, 1

H), 3.96 (m, 1 H), 3.83 (m, 1 H), 3.13-3.34 (m, 3 H), 2.89 (s, 3

H), 0.80-2.83 (m, 48 H), 1.90 (s, 3 H), 0.99 (d, J ) 5.4 Hz, 3

H), 0.87 (s, 3 H), 0.69 (s, 3 H), 0.55 (s, 3 H); MS (ESI) m/z 864

(M + H)⁺, 862 (M - H)^-_.HRMS calcd for C₅₅H₇₇NO₇: 864.5773;

found 864.5785.

via specific binding of H-dexamethasone (“³H-dex”) with the

3

G25-assay as described in J . Steroid Biochem. Molec. Biol. 50,

No. 5/6, 313-318, 1994 and estimated to approximately 25 nM.

The extract was aliquoted and stored at -70 °C. Dilution series

of the test compounds and dexamethasone (“dex”) as reference

were made from 10 mM (1 mM dex) stock solutions in DMSO.

10 µL of each dilution was added in duplicate to the wells.

The cell extracts were diluted 10-fold in EPMo + MTG buffer

(1 mM EDTA, HPO₄20 mM (pH8), 6 mM MTG). The diluted

extract was added to the wells (110 µL). ³H-dex was diluted

from the stock solution to 10 nM in EPMo + MTG buffer. 110

µL of the diluted ³H-dex were added to the wells. The final

concentration of hGR in the experiment was estimated to be

1 nM. All preparations were made in ambient temperature

(20-25 °C) on ice and with buffers at +4 °C. The plates were

incubated overnight at +4 °C (15-20 h). The incubation was

stopped by filtration through GF/B filter on a Tomtec Cell-

Harvester. The GF/B filters were dried for at least 1 h at 65

°C. A MeltiLex scintillation wax was melted onto filters with

the Microsealer. Filters were placed in a sample bag, which

was thereafter trimmed with scissors to fit the filter cassette.

The cassette were placed in the Microbeta and measured for

1 min/position, returning ccpm (corrected counts per minute).

1

Com p ou n d 19a . H NMR (500 MHz, CD₃OD) δ 7.38 (d, J

) 8.1 Hz, 1 H), 7.27 (d, J ) 12.5 Hz, 2 H), 5.76 (s, 1 H), 4.66

(s, 1 H), 4.57 (s, 1 H), 4.57 (d, J ) 6.6 Hz, 1 H), 4.56 (s, 1 H),

3.96 (s, 1 H), 3.59 (m, 1 H), 3.42 (m, 1 H), 3.25 (m, 2 H), 2.86

(m, 1 H), 2.52-2.73 (m, 3 H), 2.47 (m, 2 H), 2.33 (m, 5 H),

2.18 (m, 6 H), 2.08 (m, 2 H), 1.97 (m, 3 H), 1.80 (m, 10 H),

1.61 (m, 9 H), 1.31-1.52 (m, 7 H), 1.01 (d, J ) 6.6 Hz, 3 H),

0.92 (s, 3 H), 0.71 (s, 3 H), 0.52 (s, 3 H); MS (ESI) m/z 907 (M

+ H)⁺905 (M - H)^-Anal. Calcd for C₅₅H₇₄N₂O₉0.0.6 TFA:

;

C, 69.19; H, 7.71; N, 2.87. Found: 69.08; H, 7.60; N, 2.67.

1

Com p ou n d 19b. H NMR (500 MHz, CD₃OD) δ 7.27 (m, 2

H), 7.10 (m, 2 H), 5.74 (s, 1 H), 4.72 (m, 1 H), 4.49 (m, 1 H),

3.97 (m, 1 H), 3.56 (m, 2 H), 3.44 (m, 2 H), 3.25 (m, 2 H), 3.15

(m, 3 H), 3.09 (m, 1 H), 2.85 (m, 2 H), 2.67 (m, 3 H), 2.48 (m,

3 H), 2.18-2.39 (m, 9 H), 1.90-2.13 (m, 9 H), 1.80 (m, 4 H),

1.60 (m, 5 H), 1.44 (m, 6 H), 1.28 (m, 6 H), 1.14 (m, 2 H), 1.01

(d, J ) 6.6 Hz, 3 H), 0.93 (s, 3 H), 0.72 (s, 3 H), 0.51 (s, 3 H);

MS (ESI) m/z 893 (M + H)⁺891 (M - H)^-; Anal. Calcd for

C

₅₅H₇₆N₂O₈. 1.6 TFA: C, 64.99; H, 7.27; N, 2.60. Found: C,

64.82; H, 7.13; N, 2.62. HRMS calcd for C₅₅H₇₇N₂O₈: 893.5674.

Found: 893.5653.

3

For compounds able to displace (H)-dexamethasone from the

receptor, an IC₅₀value (the concentration required to inhibit

50% of the binding of ³(H)-dex) was determined by a nonlinear

four-parameter logistic model:

Com p ou n d 20. ¹H NMR (400 MHz, CD₃OD) δ 7.38 (d, J )

6.1 Hz, 2 H), 7.27 (d, J ) 8.0 Hz, 2 H), 6.1 (m, 3 H), 5.76 (s, 1

H), 4.90 (m, 2 H), 4.58 (m, 1 H), 4.00 (m, 8 H), 3.84 (m, 3 H),

3.21 (m, 3 H), 2.41-2.69 (m, 4 H), 2.33 (m, 2 H), 2.17 (m, 5

H), 2.06 (m, 5 H), 1.95 (m, 6 H), 1.86 (m, 2 H), 1.75 (m, 2 H),

1.46-1.68 (m, 3 H), 1.39 (m, 3 H), 1.28 (m, 6 H), 0.80-1.02

(m, 4 H), 0.74 (m, 3 H), 0.34-0.61 (m, 3 H); MS (ESI) m/z 907

(M + H)⁺905 (M - H)^-)^-_;Anal. Calcd for C₅₅H₇₄N₂O₉. 0.5 TFA:

C, 71.24; H, 8.00; N, 2.99. Found: C, 71.40; H, 6.74; N, 2.65.

b ) ((b_max- b_min)/(1 + (I/IC₅₀)^S)) + b_min

I

where I is the added concentration of binding inhibitor, IC₅₀

is the concentration for inhibitor at half-maximal binding, and

S is a slope factor. For determinations of the concentration of

³H-dex in the solutions, regular scintillation counting in a

Wallac Rackbeta 1214 was performed using the scintillation

cocktail Supermix (Wallac). The Microbeta instrument gener-

ates the mean cpm (counts per minute) value/minute and

corrects for individual variations between the detectors, thus

generating corrected cpm values. Counting efficiency between

detectors differed by less than 5%. Similar protocols were

employed to measure affinity of the compounds for progest-

1

Com p ou n d 21a . H NMR (300 MHz, CD₃OD) δ 7.02 (d, J

) 9 Hz, 2H), 6.70 (d, J ) 9 Hz, 2H), 5.72 (s, 1H), 4.38 (d, J )

7.5 Hz, 1H), 3.94 (s, 1H), 3.75 (s, 1H), 3.55-3.97 (m, 5H), 2.95

(s, 3H), 2.85-1.24 (envelope, 49H), 1.03 (d, J ) 6 Hz, 3H), 0.85-

(s, 3H), 0.70 (s, 3H), 0.55 (s, 3H); MS (ESI) m/z 907(M + 1)⁺,

905(M - 1)^-. Anal. Calcd for C₅₆H₇₈N₂O₈‚1.8H₂O: C, 71.58; H,

8.75; N, 2.98. Found: C, 71.53, H, 8.66; N, 3.09. HRMS calcd

for C₅₆H₇₉N₂O₈: 907.5831. Found: 907.5834.

Liver-Selective Glucocorticoid Antagonists

J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17 4229

erone receptor (PR), mineralocorticoid receptor (MR), androgen

receptor (AR), estrogen receptors (ERR and ERâ), and thyroid

hormone receptors (TRR and TRâ). IC₅₀values were converted

to K_iusing the standard equation

euthanized with CO₂and bled by cardiac puncture, and plasma

is analyzed for corticosterone by mass spectroscopy and for

adrenocorticotropic hormone (ACTH) levels by ELISA. Brains

and plasma are also removed for analysis of GR antagonist

levels. At doses of 100 mg/kg, RU-486 significantly increases

ACTH and corticosterone levels compared to vehicle controls.

K_i) IC₅₀/(1 + [ref]/K_d(ref))

Ack n ow led gm en t. The authors thank Peter Agback

where ref is the reference radioligand

of KaroBio, and the Abbott Analytical Department, for

1

GRAF Assa y. Chinese hamster ovarian cells (CHO-K1)

were stably transfected with an expression plasmid encoding

hGR and a reporter construct containing a glucocortiocoid

response element driving expression of alkaline phosphatase

(ALP) to produce the GRAF cell line. GRAF cells were grown

in HAM’s F12 media supplemented with 10% FCS and 1%

L-glutamine. Induction media was Opti-MEM with 1% L-

glutamine, 50 µg/mL gentamycin. GRAF cells were seeded in

growth medium in 96-well plates at ∼50 000 cells per well.

After 24 h of incubation, the cells were induced with test

compounds as well as dexamethasone, as a positive control,

at serial dilutions, to measure agonist response. To examine

antagonist effects GRAF cells were treated with increasing

amounts of test compounds in the presence of 5 nM dexa-

methasone. After 48 h of induction, disodium 3-(4-methoxy-

spiro{1,2-dioxetane-,2′-(5′-chloro)tricyclo[3.3.1.13.7]decan}-4-

yl)phenyl phosphate (CSPD; Applied Biosystems) was added

to the medium, and levels of secreted alkaline phosphatase

were analyzed by chemiluminescence on a MicroBeta Trilux.

Hep a tocyte TAT Assa y. Primary rat hepatocytes were

plated in 96-well collagen-coated plates at 50 000 cells per well.

Cells were incubated in DMEM with 10% charcoal-stripped

serum at 37 °C for 4 h. Cells were pretreated with serial

diluted compounds for 30 min, followed by 100 nM of pred-

nisolone for 4 h. Cells were then washed with PBS and lysed

with lysis buffer (50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 10

µg/mL PMSF, and 0.1% Lubrol PX). The tyrosine aminotrans-

ferase assay was performed according to standard protocols.³⁴

Basically, the substrate mixture containing 6 mM L-tyrosine,

11 µM R-ketoglutarate, and 56 µM pyridoxal-5′ phosphate in

potassium phosphate buffer was added to the lysates and

incubated for 30 min at 37 °C. The reaction was stopped by

10 N KOH and plates were read on a Spectrophotometer at

340 nm. Wells that contained substrate without compound or

Dex were used as the background, while the wells that

contained substrate and Dex without any compound were

considered as maximal signal. Percent inhibition of each

compound was calculated relative to the maximal signal, and

IC₅₀curves were generated.

assistance in acquiring H-NMR and mass spectra. We

note the valuable contributions of Bryan Sorensen to

the synthesis of intermediates such as 11, and greatly

appreciate the efforts of Rob Kimura and David Beno

to study A-348441 in the chronically catheterized rat

model.

Refer en ces

(1) van den Berghe, G. Disorders of gluconeogenesis. J . Inherit.

Metab. Dis. 1996, 19(4), 470-477.

(2) Rallison, M. L.; Meikle, A. W.; Zigrang, W. D. Hypoglycemia and

lactic acidosis associated with fructose-1,6-diphosphatase defi-

ciency. J . Pediatr. 1979, 94(6), 933-936.

(3) Friedman, J . E.; Sun, Y.; Ishizuka, T.; Farrell, C. J .; McCormack,

S. E.; Herron, L. M.; Hakimi, P.; Lechner, P.; Yun, J . S.

Phosphoenolpyruvate carboxykinase (GTP) gene transcription

and hyperglycemia are regulated by glucocorticoids in genetically

obese db/db transgenic mice. J . Biol. Chem. 1997, 272(50),

31475-31481.

(4) Baxter, J . D. Glucocorticoid hormone action. Pharmacol. Ther.

[B] 1976, 2(3), 605-669.

(5) Nieman, L. K.; Chrousos, G. P.; Kellner, C.; Spitz, I. M.; Nisula,

B. C.; Cutler, G. B.; Merriam, G. R.; Bardin, C. W.; Loriaux, D.

L. Successful treatment of Cushing’s syndrome with the gluco-

corticoid antagonist RU 486. J . Clin. Endocrinol. Metab. 1985,

61(3), 536-540.

(6) Chu, J . W.; Matthias, D. F.; Belanoff, J .; Schatzberg, A.;

Hoffman, A. R.; Feldman, D. Successful long-term treatment of

refractory Cushing’s disease with high-dose mifepristone (RU

486). J . Clin. Endocrinol. Metab. 2001, 86(8), 3568-3572.

(7) Armstrong, L.; Bell, P. M. Addison’s disease presenting as

reduced insulin requirement in insulin-dependent diabetes. Br.

Med. J . 1996, 312(7046), 1601-1602.

(8) Gettys, T. W.; Watson, P. M.; Taylor, I. L.; Collins, S. RU-486

(mifepristone) ameliorates diabetes but does not correct deficient

beta-adrenergic signaling in adipocytes from mature C57BL/6J -

ob/ ob mice. Int. J . Obes. Relat. Metab. Disord. 1997, 21(10),

865-873.

(9) Garrel, D. R.; Moussali, R.; De Oliveira, A.; Lesiege, D.; Lariviere,

F. RU 486 prevents the acute effects of cortisol on glucose and

leucine metabolism. J . Clin. Endocrinol. Metab. 1995, 80(2),

379-385.

(10) Lamberts, S. W.; Koper, J . W.; deJ ong, F. H. The endocrine

effects of long-term treatment with mifepristone (RU 486). J .

Clin. Endocrinol. Metab. 1991, 73(1), 187-191.

(11) Leeson, P. D.; Emmett, J . C.; Shah, V. P.; Showell, G. A.; Novelli,

R.; Prain, H. D.; Benson, M. G.; Ellis, D.; Pierce, N. J .;

Underwood, A. H. Selective thyromimetics. Cardiac-sparing

thyroid hormone analogues containing 3′-arylmethyl substitu-

ents. J . Med. Chem. 1989, 32, 2(2), 320-336.

In Vivo Com p ou n d Eva lu a tion . All protocols involving

animals were approved in advance by the Abbott Institutional

Animal Care and Use Committee (IACUC). All in vivo data

were analyzed for statistical significance (two-tailed t test)

using GraphPad Instat. Drug-treated groups were compared

to vehicle controls at the same day or time point.

(12) Nezasa, K.-I.; Higaki, K.; Matsumura, T.; Inazawa, K.; Hase-

gawa, H.; Nakano, M.; Koike, M. Liver-specific distribution of

rosuvastatin in rats: Comparison with pravastatin and sim-

vastatin. Drug Metab. Distr. 2002, 30(11), 1158-1163.

(13) Biessen, E. A. L.; Beuting, D. M.; Roelen, H. C. P. F.; van de

Marel, G. A.; van Boom, J . H.; van Berkel, T. J . C. Synthesis of

cluster galactosides with high affinity for the hepatic asialogly-

coprotein receptor. J . Med. Chem. 1995, 38(9), 1538-1546.

(14) Lu, X.; Fischman, A. J .; J yawook, S. L.; Hendricks, K.; Tompkins,

R. G.; Yarmush, M. L. Antisense DNA delivery in vivo: liver

targeting by receptor-mediated uptake. J . Nucl. Med. 1994, 35-

(2), 269-75.

Ra t P r ed n isolon e Ch a llen ge Mod el. Overnight fasted

150 g male Sprague Dawley rats are orally dosed with vehicle,

RU-486 (30-100 mg/kg), or selected glucocorticoid receptor

antagonists (30-100 mg/kg), 60 min prior to an oral challenge

with prednisolone at 10 mg/kg. Six hours following the

prednisolone challenge, rats are euthanized with CO₂, and bled

via cardiac puncture for evaluation of blood lymphocytes and

plasma drug levels. 7 mm liver biopsy punches are harvested

for evaluation of tyrosine aminotransferase (TAT), hepatic

glycogen, and GR antagonist levels. Additional liver tissue,

retroperitoneal fat, skeletal muscle, kidney, and skin from an

ear biopsy are also removed for isolation and evaluation of

mRNA. Prednisolone increases hepatic TAT and glycogen

levels and induces severe lymphopenia during the 6 h chal-

lenge interval. At doses of 100 mg/kg, RU-486 completely

antagonizes these hepatic and peripheral responses.

(15) Beljaars, L.; Melgert, B. N.; Meijer, D. K. F.; Molema, G.;

Poelstra, K. Neoglyco- and neopeptide albumins for cell-specific

delivery of drugs to chronically diseased livers. Drugs Pharm.

Sci. 2001, 115(Drug Targeting Technology), 189-240.

(16) Hatori, Y.; Kawakami, S.; Yamashita, F.; Hashida, M. Controlled

biodistribution of galactosylated liposomes and incorporated

probucol in hepatocyte-selective drug targeting. J . Controlled

Release 2000, 69, 369-377.

Mou se Hyp oth a la m ic-P itu ita r y-Ad r en a l (HP A) Ac-

tiva tion Mod el. Nonfasted CD-1 male mice, weighing ap-

proximately 25 g, are dosed with vehicle, RU-486 (30-100 mg/

kg), or GR antagonist (30-100 mg/kg) at 0800 h when

corticosterone levels are low. Two hours later, mice are

(17) Bebernitz, G. R.; Dain, J . G.; Deems, R. O.; Otero, D. A.; Simpson,

W. R.; Strohschein, R. J . Reduction in glucose levels in STZ

Diabetic Rats by 4-(2,2-dimethyl-1-oxopropyl)benzoic acid:

A

prodrug approach for targeting the liver. J . Med. Chem. 2001,

44, 512-523.

4230 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 17

Tu et al.

(18) Carey, M. C.; Cahalane, M. J . Enterohepatic Circulation. In The

Liver: Biology and Pathobiology, 2nd ed.; Raven Press: Ltd,

New York, 1988; Chapter 33, pp 573-616. Data derived from

other referenced sources.

Soc. Chem. Commun. 1994, 1985-1986.

(25) Hagberg, L.; Apelqvist, T. Manuscript in preparation.

(26) Tserng, K.-Y. A convenient synthesis of 3-keto bile acids by

selective oxidation of bile acids with silver carbonate-Celite. J .

Lipid Res. 1978, 19(4), 501-504.

(19) Wess, G.; Kramer, W.; Bartmann, W.; Enhsen, A.; Glombik, H.;

Mu¨llner, S.; Bock, K.; Dries, A.; Kleine, H.; Schmitt, W. Modified

bile acids: Preparation of 7R, 12R-dihydroxy-3â- and 7R, 12R-

dihydroxy-3R-(2-hydroxyethoxy)-5â-cholanic acid and their bio-

logical activity. Tetrahedron Lett. 1992, 33(2), 195-198.

(20) Kullak-Ublik, G.-A.; Glasa, J .; Boker, C.; Oswald, M.; Grutzner,

U.; Hagenbuch, B.; Stieger, B.; Meier, P. J .; Beuers, U.; Kramer,

W.; Wess, G.; Paumgartner, G. Chlorambucil-taurocholate is

transported by bile acid carriers expressed in human hepato-

cellular carcinomas. Gastroenterology 1997, 113, 1295-1305.

(21) Starke, D.; Lischka, K.; Pagels, P.; Uhlmann, E.; Kramer, W.;

Wess, G.; Petzinger, E. Bile acid-oligodeoxynucleotide conju-

gates: Synthesis and liver excretion in rats. Bioorg. Med. Chem.

Lett. 2001, 11, 945-949.

(22) Kramer, W.; Wess, G.; Schubert, G.; Bickel, M.; Girbig, F.;

Gutjahr, U.; Kowalewski, S.; Baringhaus, K.-H.; Enhsen, A.;

Glombik, H.; Mu¨llner, S.; Neckermann, G.; Schulz, S.; Petzinger,

E. Liver-specific drug targeting by coupling to bile acids. J . Biol.

Chem. 1992, 267(26), 18598-18604. See also Holzinger, F.;

Schteingart, C. D.; Ton-Nu, H.-T.; Eming, S. A.; Monte, M. J .;

Hagey, L. R.; Hofman, A. F. Fluorescent bile acid derivatives:

Relationship between chemical structure and hepatic and in-

testinal transport in the rat. Hepatology 1997, 26(3), 1263-1271.

(23) Kim, D.-C.; Harrison, A. W.; Ruwart, M. J .; Wilkinson, K. F.;

Fisher, J . F.; Hidalgo, I. J .; Borchardt, R. T. Evaluation of the

bile acid transporter in enhancing intestinal permeability to

renin-inhibitory peptides. J . Drug Targeting 1993, 1, 347-359.

(24) Acosta, K.; Cessac, J . W.; Rao, N.; Kim, H. K. Oxidative

demethylation of 4-substituted N, N-dimethylanilines with

iodine and calcium oxide in the presence of methanol. J . Chem.

(27) Kauppi, B.; J akob, C.; Fa¨rnegårdh, M.; Yang, J .; Ahola, H.;

Alarcon, M.; Calles, K.; Engstro¨m, O.; Harlan, J .; Much-

more, S.; Ramqvist, A.-K.; Thorell, S.; O¨ hman, L.; Greer, J .;

Gustafsson, J .-Å.; Carlstedt-Duke, J .; Carlquist, M. The three-

dimensional structures of antagonistic and agonistic forms of

the glucocorticoid receptor ligand-binding domain: RU-486

induces a transconformation that leads to active antagonism.

J . Biol. Chem. 2003, 278(25), 22748-22754.

(28) Williams, S. P.; Sigler, P. B. Atomic structure of progesterone

complexed with its receptor. Nature 1998, 393, 392-396.

(29) Crofford, O. B. Diabetes Control and Complications. Annu. Rev.

Med. 1995, 46, 267-279.

(30) J over, E.; Paradinas, C.; Prieto, J .; Arranz, M. T.; Para, J .;

Velasco, R. Study of the effect of ACTH and cortisol on the

plasma lipids in man using thin-layer chromatography. J . Med.

1976, 7(2), 131-142.

(31) The hepatic selectivity of A-348441 also reduces its ability to

block peripheral progesterone receptors. We have determined

(P. J acobson et al., manuscript in preparation) that A-348441

is much less effective than mifepristone in

progestin model.

(32) Osterlund, M.; Barkhem, T., et al. Manuscript in preparation.

(33) Studies performed by R. Kimura and D. Beno using a chronically

catheterized rat model; data not shown here.

(34) Granner, D. K.; Tomkins, G. M. Tyrosine Aminotransferase (Rat

Liver). Methodology 1970, 17A, 633-637.

a mouse anti-

J M0400045

Article Doi

Liver-selective glucocorticoid antagonists: A novel treatment for type 2 diabetes

DOI: 10.1021/jm0400045

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jm0400045

Full text of DOI:10.1021/jm0400045

Products guided by the article

R&D Labs maybe for 298186-92-2

Relevant to this article

Hot Product