Novel tricyclic-α-alkyloxyphenylpropionic acids: Dual PPARα/γ agonists with hypolipidemic and antidiabetic activity

Article abstract of DOI:10.1021/jm010964g

Synthesis and structure-activity relationships of tricyclic α-ethoxy-phenylpropionic acid derivatives guided by in vitro PPARα and PPARγ transactivation data and computer modeling led to the identification of the novel carbazole analogue, 3q, with and dual PPARα (EC₅₀ = 00.36 μM) and PPARγ (EC₅₀ = 0.17 μM) activity in vitro. Ten days treatment of db/db mice with 3q improved the insulin sensitivity, as measured by OGTT, better than that seen with both pioglitazone and rosiglitazone treatment, suggesting in vivo PPARγ activity. Likewise, 3q lowered plasma triglycerides and cholesterol in high cholesterol fed rats after 4 days treatment, indicating in vivo PPARα activity. Investigations of the pharmacokinetics of selected compounds suggested that extended drug exposure improved the in vivo activity of in vitro active compounds.

Full text of DOI:10.1021/jm010964g

J . Med. Chem. 2002, 45, 789-804
789
Novel Tr icyclic-r-a lk yloxyp h en ylp r op ion ic Acid s: Du a l P P ARr/γ Agon ists w ith
Hyp olip id em ic a n d An tid ia betic Activity
Per Sauerberg,*,† Ingrid Pettersson,† Lone J eppesen,† Paul S. Bury,† J ohn P. Mogensen,† Karsten Wassermann,†
Christian L. Brand,† J eppe Sturis,† Helle F. Wo¨ldike,† J an Fleckner,† Anne-Sofie T. Andersen,†
Steen B. Mortensen,† L. Anders Svensson,† Hanne B. Rasmussen,† Søren V. Lehmann,† Zdenek Polivka,‡
Karel Sindelar,‡ Vladimira Panajotova,‡ Lars Ynddal,† and Erik M. Wulff†
Novo Nordisk A/ S, Novo Nordisk Park, 2760 Måløv, Denmark, and RE&D VUFB, s.r.o., Podebradska 56/ 186,
18066 Praha 9, Czech Republic
Received J une 22, 2001
Synthesis and structure-activity relationships of tricyclic R-ethoxy-phenylpropionic acid
derivatives guided by in vitro PPARR and PPARγ transactivation data and computer modeling
led to the identification of the novel carbazole analogue, 3q, with dual PPARR (EC50 ) 0.36
µM) and PPARγ (EC50 ) 0.17 µM) activity in vitro. Ten days treatment of db/db mice with 3q
improved the insulin sensitivity, as measured by OGTT, better than that seen with both
pioglitazone and rosiglitazone treatment, suggesting in vivo PPARγ activity. Likewise, 3q
lowered plasma triglycerides and cholesterol in high cholesterol fed rats after 4 days treatment,
indicating in vivo PPARR activity. Investigations of the pharmacokinetics of selected compounds
suggested that extended drug exposure improved the in vivo activity of in vitro active
compounds.
In tr od u ction
70% of the patients.12 Paradoxically, the improvement
in insulin sensitivity is accompanied with a body weight
gain, which correlates to the effectiveness of the treat-
ment.12,13 This limited, although significant, improve-
ment in insulin sensitivity in type 2 diabetic patients
undergoing TZD (pioglitazone or rosiglitazone) treat-
ment warrants the development of novel and more
effective treatments.
The recent identification of the nuclear receptor
peroxisome proliferator activated receptor-γ (PPARγ)
and PPARR as being the primary targets for the
normoglycaemic TZDs and the lipid lowering fibrates,
respectively, has provided new opportunities for the
identification of novel compounds for the treatment
of type 2 diabetes.14,15 The successful identification of
novel PPARγ selective agonists with good blood glucose
lowering activity, using in vitro PPAR receptor binding
and in vitro activation screening, has already been de-
scribed.16-18
Despite the growing evidence of the positive effects
of fibrate (PPARR agonist) treatment in type 2 diabetic
patients, most reports have been on the identification
of selective PPARγ agonists. Only a few reports have
dealt with selective PPARR agonists,15 and even fewer
compounds have been reported to have both PPARγ and
PPARR agonist activity,19-21 e.g., KRP-297 and (-)-
DRF2725/NNC61-0029. We therefore decided to inves-
tigate if such dual activating receptor agonists would
have improved in vivo efficacy over the aforementioned
PPAR subtype selective agonists. The aim of this work
was therefore to identify compounds with full efficacy
and equal potency on PPARR and PPARγ receptors and
to characterize such compounds in animal models
predictive of clinical activity.
Type 2 diabetes is a metabolic disease characterized
by insulin resistance, hyperglycaemia, and often hyper-
lipidemia. Untreated type 2 diabetes leads to several
chronic diseases such as retinopathy, nephropathy,
neuropathy, and cardiovascular diseases,1 the latter
leading to increased mortality. Two classes of com-
pounds known as the thiazolidinediones (TZDs) and the
fibrates were empirically discovered decades ago to
possess the ability to lower blood glucose and lipids in
rodent models of insulin resistance and hyperlipedimia,
respectively. In humans, fibrates are effective at lower-
ing serum triglycerides and raising HDL cholesterol
levels, primarily through increased clearance and de-
creased synthesis of triglyceride-rich VLDL.2 Fibrates
have been shown to slow the progression of athero-
sclerosis and reduce the number of coronary events in
secondary prevention studies and in patients with
normal levels of LDL cholesterol and lately in diabetic
patients.3-7 Interestingly, improvement in glucose toler-
ance in type 2 diabetic patients has also been shown
with clofibrate and bezafibrate.8-10 Furthermore, fi-
brates have been reported to reduce weight gain in
rodents without effects on food intake.11 This is of
interest since obesity is a major risk factor for the
development of type 2 diabetes. Similarly, clinical trials
have shown that the TZDs lower blood glucose and
insulin levels and improve insulin sensitivity, but they
have only marginal effects on plasma lipids in type 2
diabetic patients.12 The magnitude of the blood glucose
lowering effect corresponds to a lowering of HbA1c of
1-1.9% in responders, which typically accounts for 50-
* Corresponding author: Per Sauerberg, Novo Nordisk A/S, Novo
Nordisk Park E9 2.02, 2760 Måløv, Denmark. Fax: +45 44663939.
Phone: + 45 44434858. E-mail: psa@novonordisk.com.
† Novo Nordisk A/S.
The non-TZD alkoxy-propionic acid class of insulin
sensitizers was chosen as the chemical lead, as this
functional group would be less prone to racemization
‡ RE&D VUFB.
10.1021/jm010964g CCC: $22.00 © 2002 American Chemical Society
Published on Web 01/23/2002

790 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
Sch em e 1
F igu r e 1. Graphical illustration of the chemical structure
evolution of the program.
pure (S)-enantiomers by using the pure (S)-isomer of
1. The basic hydrolysis of the esters (2) leading to the
final compounds 3p , 3q, 3r , and 3s did not cause any
measurable recemization. Compounds for in vivo testing
in the db/db mouse model of type 2 diabetes and
pharmacokinetic measurements in rats were converted
to either the lysine or the arginine salts.
compared to TZD, which undergoes complete racemiza-
tion under physiological conditions.22 This was impor-
tant since only the (S)-enantiomers of the TZDs bind to
the receptor with high affinity.23 Further, recent reports
have suggested very potent in vitro and in vivo activities
of the alkoxy-propionic class of compounds.24-27
The present paper reviews the in vitro transactivation
activity of the compounds obtained by initially combin-
ing structural elements from rosiglitazone (Table 2) and
the ethoxy-propionic acid moiety with tricycles to give
the lead structure A (Figure 1). These compounds were
further revised to compounds B after replacement of the
methylamino with a methine group. Finally, the intro-
duction of a nitrogen atom as an attachment point in
the tricyclic moiety gave the compounds C, which were
both potent and efficacious PPARR and PPARγ agonists
in vitro. In vivo experiments with selected compounds
suggested that continued drug exposure was critical to
the magnitude of the improvement in insulin sensitivity.
The carbazole analogue 3q (Table 2) was identified as
having improved insulin sensitizing and lipid lowering
effects in vivo compared to rosiglitazone and pioglita-
zone due to potent intrinsic PPARR and PPARγ activity
combined with good pharmacokinetics.
Resu lts a n d Discu ssion
The aim of the present work was to identify com-
pounds with full efficacy and equal potency on PPARR
and PPARγ receptors. In vitro receptor transactivation
assays with the binding domains of each of the two
PPAR receptor subtypes were used as the primary
screening tool in that effort. A similar screening strategy
had previously been used in the search for PPARγ
selective agonists.16 To compare the efficacy of com-
pounds from test to test, WY14643 and rosiglitazone
were used as reference agonists in the PPARR and
PPARγ transactivation assays, respectively. Maximum
obtained fold activation with the reference agonist
(approximately 20-fold with WY14643 in PPARR, and
120-fold with rosiglitazone in PPARγ) was defined as
100%.
By combining tricyclic ring fragments with structural
elements from rosiglitazone (Table 2) and from ethoxy-
propionic acids27 the general lead structure A (Figure
1) was obtained. Selecting either the dihydro-dibenzo-
cycloheptene or the dibenzothiazepine as the tricycle
gave the specific compounds 3a and 3b, respectively.
Compound 3a was a potent and selective PPARγ partial
(78%) agonist (Table 1), whereas 3b was equally potent
and efficacious (41-68%) on both PPARR and PPARγ.
Although the potency and efficacy was too low to meet
our criteria for further investigations, this initial data
showed us that it probably would be possible to design
compounds with combined PPARR and PPARγ activity.
Consequently, additional tricyclic analogues were de-
signed and synthesized. Substitution of the methyl-
amine in 3a with carbon (methine) gave 3c. However,
Ch em istr y
Most of the desired compounds were synthesized
by alkylation of ethyl 2-ethoxy-3-(4-hydroxyphenyl)-
propionate (1) (synthesized by a method analogous to
the procedure published by Haigh et al.28) with the
appropriate tricyclic-alcohol under Mitsunobu conditions
using either triphenylphosphine and diethyl azodicar-
boxylate (DEAD) or tributylphosphine and 1,1′-(azodi-
carbonyl)dipiperidine (ADDP) as shown in Scheme 1
(procedures A and B). Alternatively, the alkylation was
performed with the mesylate or the alkyl halide as in
procedures C and D. Aqueous sodium hydroxide hy-
drolysis of 2 in ethanol gave the target products 3. The
more interesting compounds were synthesized as their

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 791
Ta ble 1. In Vitro hPPARR and hPPARγ Transactivation of Novel Racemic Test Compoundsa

792 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Ta ble 1. (Continued)
Sauerberg et al.
a
Compounds were tested in at least three separate experiments in five concentrations ranging from 0.01 to 30 µM. EC50s were not
b
calculated for compounds producing transactivation lower than 25% at 30 µM. Fold activation relative to maximum activation
obtained with WY14643 (approximately 20-fold corresponded to 100%) and with crosiglitazone (approximately 120-fold corresponded to
100%).
3c was still a selective PPARγ partial agonist but much
less potent than 3a . Changing the chain length from
three to two carbon atoms, as in 3d , slightly increased
the PPARγ potency and efficacy but did not improve the
PPARR efficacy. Substituting a carbon for an oxygen
atom in the tricyclic ring of 3c gave 3e, which had some
PPARR and PPARγ activity. Further insertion of oxygen
atoms in the tricyclic ring of 3d as in 3g did not improve
PPARR activity but retained the PPARγ activity. Re-
placing the nitrogen and the sulfur atoms in 3b with
sulfur and oxygen gave a compound, 3f, with a quite
similar in vitro profile.
At this point it was decided to use molecular modeling
in an attempt to improve the PPARγ potency of the
compounds. Using the X-ray structure of rosiglitazone
in complex with the binding domain of the PPARγ
receptor,29 the available pocket around the pyridine ring
in rosiglitazone was calculated with a Grid protocol
using a water probe, Figure 2. These calculations
showed that this part of the pocket was rather narrow
and that a planar ring system would be preferred.
Therefore, rather than expanding the tricyclic ring
system, it was decided to make the center ring smaller
as in 3h , 3i, and 3j. Two of the three compounds (3h
and 3j) were in fact more potent PPARγ agonists than
the previous compounds 3a -g. Furthermore, com-
pounds 3h and 3j were full PPARγ but partial and weak
PPARR agonists. A similar type of modeling calculation
could not be made on the PPARR receptor, as the X-ray
structure of this receptor protein was not available.
In an attempt to improve the PPARR potency and
efficacy, a series of tricyclic analogues was designed with
a nitrogen in the center ring used as the attachment
point, 3k -o. Nitrogen was chosen since that would give
a geometry at the attachment point more similar to that
seen in the potent PPARγ analogues 3h and 3j than in
the less potent 3i. The nitrogen analogue of 3g, 3k , was
close to having the desired profile. The compound was
equally potent on PPARR and PPARγ (EC50 ) 1.2 µM);
it was a full PPARγ agonist and an almost full PPARR
(89%) agonist. The nitrogen analogues of 3c and 3d ,
respectively 3l and 3m , were both, as predicted, potent
full PPARγ agonists, but they had only low PPARR
efficacy. A breakthrough was achieved with the planar
â-carboline analogue 3o, which was a potent and full
agonist on both PPARR and PPARγ. Despite a 10-fold
difference in potency in favor of PPARγ, it was decided
to make the pure (S)-enantiomer (3p , Table 2). The
initial lead optimization had, until this point, been
carried out on racemic mixtures, but the promising
results with 3o prompted us to continue with pure
enantiomers. From the literature,27,30 it was known that
the (S)-enantiomer was the active form, which we later
confirmed with our own compounds (data not shown).
Compound 3p showed the expected increase in po-
tency compared to 3o on both PPARR and PPARγ. In
fact, 3p (Table 2) was approximately 10 times more
potent on PPARγ than rosiglitazone. Replacement of the
â-carboline ring system with the equally planar carba-
zole ring system gave the compound 3q, which had the
desired dual PPARR/PPARγ activity profile. Interest-
ingly, a closely related carbazole containing thiazoli-
dinedione (TZD) analogue had previously been reported
not to have any glucose lowering effect in vivo.31 To
understand the differences between the ethoxypropionic
acid 3q and the TZD analogue, both compounds were
docked into the PPARγ receptor-binding domain. The
results showed that the thiazolidinedione moiety was
slightly smaller than the ethoxypropionic acid group,
resulting in a slightly shorter molecule, which hindered
either the carbazole moiety in reaching the lipophilic
pocked at the right angle or the TZD in making inter-

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 793
Ta ble 2. In Vitro hPPARR, hPPARγ, and hPPARδ Activation of Standard Compounds and Pure (S)-Enantiomeric Test Compoundsa
a
b
See Table 1. Fold activation relative to maximum activation obtained with carbacyclin (approximately 250-fold corresponded to
100%). c One experiment.

794 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
F igu r e 2. Available pocket around the pyridine ring (left) and
around the tricyclic ring (right) in the X-ray structures of the
PPARγ receptor binding domain in complex with either rosi-
glitazone29 or 3q, respectively. The available pocket calculated
with Grid using a water probe is shown as a blue grid at the
energy level 0 kcal/mol, and ligands are shown as space-filling
models.
Ta ble 3.
F igu r e 3. Crystal structure of the PPARγ receptor binding
domain in complex with 3q. The ligand carbon atoms are
shown in magenta. Amino acids of the PPARγ receptor
neighboring the ligand are shown with green-colored carbon
atoms and with their residue type and sequence numbers
written in yellow. Also shown, as a chicken wire net repre-
sentation, are the experimental SigmaA weighted electron
density maps: in blue, the map with the coefficients 2Fo - Fc
at 1σ cutoff; and in brown, the negative side of the Fo - Fc
coefficients map at 3σ cutoff.
X-ray
no. of conformations
structure (Å) within 2.8 kcal/mola
3q
3q
3i
3j
distance 1-2 (9.6-11.6 Å)
distance 1-3 (10.5-12.5 Å)
distance 1-4 (10.5-12.5 Å)
distance 1-2, 1-3, and 1-4
10.6
11.5
11.5
137
110
74
25
7
0
31
19
5
74
0
5
a
Number of conformations within 2.8 kcal/mol which fulfill
distance criteria observed in the crystal structure of the PPARγ
receptor binding domain in complex with 3q.
actions with AF-2 amino acids (data not shown; see also
Table 3). Alternatively, poor pharmacokinetic properties
of the TZD could explain the lack of in vivo activity.
None of the compounds 3p , 3q, 3r , or 3s displayed
any in vitro activity on the PPARδ receptor subtype
(Table 2) or on the RXRR receptor (data not shown).
Activity at other nuclear receptors was not investigated.
To further understand the SAR generated in this
series of tricyclic analogues and to characterize the
receptor interaction, a crystal structure of 3q in complex
with the PPARγ receptor protein was generated. The
crystallographic structure of 3q in complex with the
PPARγ receptor binding domain was determined to 2.5
Å resolution using overnight soaking. A bound ligand
molecule was found in one of the independent PPARγ
receptor molecules. The examination of the ligand-
receptor interactions revealed the following observa-
tions:
The 2-ethoxypropionic acid in 3q showed well-defined
electron density (Figure 3). The four hydrogen bonds
between the propionic acid group of 3q and the PPARγ
receptor protein (Figure 4 and table X in Supporting
Information) had all previously been reported to be
involved in hydrogen bond formation in other PPARγ
receptor/ligand complexes,29,32,33 e.g., rosiglitazone as
shown in Figure 4. The ethoxy group gave further
hydrophobic interactions with Phe 282.
F igu r e 4. Sketch of the relative positions of 3q (magenta)
and rosiglitazone (green29) bound to the PPARγ receptor
binding domain. Indicated are also the residues involved in
hydrogen binding to 3q, with their residue type and sequence
numbers written in yellow, and written in black, the numbers
of the helices in the active site region.
285, Tyr 327, and at a somewhat greater distance, 4.3
Å, the Leu 330 (Figure 3).
The carbazole ring, situated in the large hydrophobic
pocket of the PPARγ receptor protein, showed, however,
less well-defined electron density. The electron density
from the ethyl group atoms connecting the phenoxy and
The phenyl group of the ethoxy-phenyl moiety in 3q
also showed well-defined electron density. This part
formed van der Waals interactions with Met 364, Cys

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 795
F igu r e 5. Stereoview of the crystal structure of the PPARγ receptor binding domain in complex with 3q. The blue grid shows
the interactions between the active site and an aromatic probe at level -2.3 kcal/mol.
the carbazole ring was missing, moreover there were
even some negative difference electron density regions
seen close to this bond. The electron density around the
ring system was also somewhat weak. In addition, for
this part of the ligand, the temperature factors were also
slightly higher, above 70 Å2, compared to the rest of the
molecule that had temperature factors below 70 Å2. This
lack of well-defined electron density indicated than the
carbazole region of 3q was relatively mobile, or alter-
natively had more than one conformation that was not
easily interpreted. To investigate the available cavity
and the interactions between the aromatic rings and the
amino acids in the binding pocket, Grid calculations
with an aromatic probe were performed. During these
calculations the ligand was neglected (Figure 5). From
these calculations it could be seen that the tricyclic ring
system was located where favorable interactions be-
tween an aromatic group and the receptor were ob-
served, mainly Gly 284, Ile 341, Arg 288, Ser 342, Phe
264, His 266, and Leu 333 (Figures 3 and 5). Further-
more, the available binding pocket was, as predicted (see
also Figure 2), narrow but large enough for sideway
movements possibly explaining the less well-defined
electron density.
The importance of the attachment atom (carbon vs
nitrogen) and bond type (single vs double) to the tricyclic
ring system, which had been experimentally observed
(3c vs 3l; 3d vs 3m ; 3i vs 3j vs 3q; Tables 1 and 2), was
also investigated using the crystal structure and model-
ing. The ligands 3q, 3i, and 3j were chosen as model
compounds for these calculations. In this series, both
3q and 3j were potent PPARγ agonists while 3i was a
very weak PPARγ agonist. Two geometric parameters,
which were considered important for the ligands to be
able to adopt the shape of the binding pocket in the
PPARγ receptor, were the distance between the car-
boxylic acid and the tricyclic ring system, and the
u-shape of the molecule. To analyze the possibility for
3q, 3i, and 3j to adopt a shape which was comparable
with the shape of the binding pocket, the distances
between the carbon atom connected to the carboxylic
acid group and three different atoms in the tricyclic ring
system were calculated and compared to the distance
measured in the X-ray structure of the PPARγ receptor
binding domain in complex with 3q, Table 3. To sample
the possible conformations, conformational analyses
were performed using the MMFF force field34-37 and a
systematic pseudo Monte Carlo search in MacroModel
7.0.38,39 The results showed (Table 3) that both 3q and
3j could adopt conformations, which fulfill all three
distance criteria for conformations within 2.8 kcal/mol,
while 3i could not. This means that although the
tricyclic ring system in 3i could reach the lipophilic
binding pocket of the receptor (distance 1-2, Table 3),
the carbon-carbon double bond prevented the ligand
from adopting the necessary curved conformation (dis-
tance 1-4).
In the male db/db mouse, our primary in vivo model
for improvement of insulin sensitivity, the two marketed
PPARγ agonists, rosiglitazone (Avandia) and pioglita-
zone (Actos), showed dose-related reduction of nonfasted
blood glucose (BG) when dosed orally by gavage once
daily for 7 days (Table 4, Figure 6a). The maximum
obtained BG reduction with rosiglitazone maleate, when
treated with the doses 0.2, 0.6, 2.0, and 6.0 mg/kg, was
35% compared to vehicle treated animals (53% normal-
ization compared to lean db/+ mice; see Experimental
Section for calculation). The less potent pioglitazone
gave a 53% reduction (80% normalization) when dosed
at 3.0, 10, 30, and 100 mg/kg. Both compounds also
showed a dose-related reduction of nonfasted plasma
insulin and a non-dose-related reduction in TG (Table
4). TG data were, however, not used as selection criteria,
since the observed TG lowering effect of rosiglitazone
and pioglitazone had not been found in clinical trials.
The animals were dosed for a further 2 days (a total
of 9 days treatment), after which an oral glucose
tolerance test (OGTT) was performed. The reduction of
the blood glucose area under the curve (AUCglu), when
the animals were given an oral glucose dose (3 g/kg),
was considered to be a more direct measure of the
improved insulin sensitivity of the compounds. Surpris-
ingly, rosiglitazone only reduced the AUCglu by 16%
compared to vehicle treated animals (25% normalization
compared to lean db/+ mice), whereas pioglitazone

796 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
Ta ble 4. In Vivo Efficacy in Male db/db Mice after Oral Treatment for 7-9 Daysa
d
BGb
ED50
mg/kg
BG
% max
reduction
TGc
ED50
mg/kg
TG
% max
reduction
insulin
ED50
mg/kg
insulin
% max
reduction
AUCglu
ED50
mg/kg
AUCglu
% max
reduction
,
,
,
,
compd no.
3k
3p
3q
rosiglitazone
pioglitazone
8.11 ( 1.04
0.33 ( 0.80
0.27 ( 0.94
0.87 ( 0.62
2.62 ( 0.19
17 ( 8
46 ( 4
58 ( 7
35 ( 7
53 ( 3
18.28 ( 1.23
2.68 ( 0.62
0.52 ( 0.56
0.14 ( 0.91
26.74 ( 0.60
13 ( 6
50 ( 2
52 ( 10
58 ( 2
21 ( 5
ND
NE
6.57 ( 0.78
3.32 ( 0.71
0.40 ( 0.62
0.77 ( 0.62
12.09 ( 0.79
20 ( 5
12 ( 11
55 ( 13
16 ( 9
39 ( 8
7.74 ( 0.89
0.34 ( 0.78
0.07 ( 1.27
15.05 ( 0.97
77 ( 4
85 ( 3
43 ( 9
54 ( 10
a
Male db/db mice (n ) 6) were treated once a day by oral gavage for 9 days. Compound 3k was tested at the doses 0.3, 1.0, 3.0, and
10.0 mg/kg/day; 3p at 1.0, 3.0, 10.0, and 30 mg/kg/day; 3q at 1.0, 3.0, 5.0, and 20 mg/kg/day; rosiglitazone at 0.2, 0.6, 2.0, and 6.0 mg/
kg/day; and pioglitazone at 3.0, 10.0, 30.0, and 100.0 mg/kg/day. ED50 values were calculated via nonlinear regression using GraphPad
PRISM 3.02 and are expressed as mean ( SEM. “% max reduction” is the maximum achieved reduction relative to vehicle treated control
b
d
group ( SEM. Nonfasting blood glucose after 7 days treatment. c Nonfasting triglycerides after 7 days treatment. Area under blood
glucose time curve after OGTT on the 9th day of treatment.
Ta ble 5. Single Dose Rat Pharmacokinetics after iv and po
Administration of Selected Compoundsa
c
d
f
g
Cmax po,b
AUCpo
,
Fpo
,
CLe
Vss T(1/2)po
compd no.
3k
3p
3q
rosiglitazone 4420
pioglitazone
ng/mL (ng × min)/mL
%
mL/min/kg L/kg min
730
527
4430
ND
79523
8643717
873922
1753075
63
61
>100
83
13.7
12.7
0.75
2.1
1.4
0.76
0.88
0.73 1332
0.38
0.22
ND
162
182
228
4655
112
a
Rats were given either a single dose iv (1.2 mg/kg) (n ) 8) or
a single dose po (2.2 mg/kg) (n ) 8) of each of the test compounds.
At each of the time points (5, 15, 30, 60, 90, 120, 240, and 360
min), one animal was sacrificed, and blood samples were analyzed
b
for compound plasma concentration. Maximum plasma concen-
tration after oral dosing. c Estimated area under the plasma-
d
concentration time curve after oral dosing. Oral bioavailability.
g
e Clearance. f Volume of distribution during steady state. Oral
half-life.
normalization) when dosed at 1.0, 3.0, 10, and 30 mg/
kg of the arginine salt. However, a significant effect was
obtained with 3q. When dosed at 1.0, 3.0, 5.0, and 20
mg/kg of the 3q arginine salt, a dose-related reduction
of both BG (58% of vehicle and 78% normalization) and
AUCglu (55% of vehicle and 89% normalization) was
obtained (Table 4 and Figure 6a,b).
The somewhat surprising db/db mouse results made
us investigate the pharmacokinetics of the compounds
to see if that could explain the differences in the in vivo
effects.
F igu r e 6. (a) Dose-related reduction of the nonfasted blood
glucose in male db/db mice (n ) 6) treated for 7 days with 3q,
rosiglitazone, and pioglitazone orally once a day. Values are
expressed as mean ( SEM. * represents P < 0.05, ** P < 0.01
using one way ANOVA and Dunetts multiple comparison test.
(b) Dose-related reduction of the area under the blood glucose
concentration (AUCglu) vs time curve after oral glucose toler-
ance test (OGTT) in male db/db mice (n ) 6) treated for 9 days
with 3q, rosiglitazone, and pioglitazone orally once a day.
Values are expressed as mean ( SEM. * represents P < 0.05,
** P <0.01 using one way ANOVA and Dunetts multiple
comparison test.
Rats were given either a single dose iv (1.2 mg/kg)
(n ) 8) or a single dose po (2.2 mg/kg) (n ) 8) of each of
the test compounds. One animal was sacrificed at each
of the time points (5, 15, 30, 60, 90, 120, 240, and 360
min), and blood samples were analyzed for compound
plasma concentration. Data showed that the test com-
pounds had very different pharmacokinetic parameters
(Table 5). The two standard compounds rosiglitazone
and pioglitazone had approximately the same maximum
plasma concentration (Cmax ∼ 4400 (ng × min)/mL)
when given the same oral dose. The estimated areas
under the curve (AUCpo) were, however, quite different,
with pioglitazone having an AUCpo twice as big as that
of rosiglitazone (Table 5). The higher pioglitazone
exposure was a consequence of the lower clearance rate,
resulting in a longer plasma half-life (t1/2po ) 228 min
for pioglitazone and 182 min for rosiglitazone).
reduced the AUCglu by 39% (61% normalization) (Table
4, Figure 6b).
Three compounds identified by the in vitro activation
screening, 3k , 3p , and 3q, were tested in db/db mice.
Despite the quite impressive in vitro activity, 3k only
produced a very modest dose-related reduction in BG
(17% compared to vehicle and 25% normalization) and
AUCglu (20% of vehicle and 31% normalization) when
dosed at 0.3, 1.0, 3.0, and 10 mg/kg of the lysine salt.
Compound 3p gave a better reduction in BG (46% of
vehicle and 77% normalization) but surprisingly had
very little effect on the AUCglu (12% of vehicle and 22%
The essentially inactive compound 3k had much lower
Cmax and AUCpovalues; the latter could not be estimated
due to the few data points resulting from the short half-
life. The other low efficacy compound 3p also showed
low Cmax and low AUCpo compared to pioglitazone (Table

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 797
Ta ble 6. In Vivo Effects of Rosiglitazone in Male db/db Mice
after Once-a-Day or Three-Times-a-Day Oral Dosinga
effects after 7 and 9 days dosing
AUCglu
% AUCglu
rosiglitazone
dose
%
after OGTT, red. after
BG, mM
BG red.b mM × min
2954 ( 208
OGTTc
vehicle
2150 ( 1.24
1 mg/kg × 1
16.08 ( 1.55 25 ( 7
2551 ( 180 14 ( 6
2398 ( 338 18 ( 11
2141 ( 165 28 ( 6
0.33 mg/kg × 3 11.35 ( 1.31 47 ( 6
3 mg/kg × 1
11.72 ( 1.29 45 ( 6
13.06 ( 1.88 39 ( 9
1 mg/kg × 3
1423 ( 86
51 ( 3
a
Male db/db mice (n ) 6) were treated for 9 days with
rosiglitazone at 1 mg/kg or 3 mg/kg total daily dosage, given either
b
once a day or three times daily by oral gavage. Nonfasting blood
F igu r e 7. Plasma drug concentration versus time curves
obtained from rats dosed with 3q orally by gavage and
intravenously. At each time point, one rat was sacrificed and
drug plasma concentration was measured.
glucose was measured on day 7, and the percent reduction relative
to vehicle treated group calculated. c On day 9 an oral glucose
tolerance test was performed, and the area under the plasma
glucose-concentration time curve was measured and the percent
reduction relative to vehicle treated group calculated. Values are
expressed as mean ( SEM.
Ta ble 7. Changes in Nonfasted Plasma Triglycerides and
Total Cholesterol in High Cholesterol Fed Male
Sprague-Dawley Rats after 4 Days Oral Treatmenta
TG
total cholesterol
ED50
,
%
ED50
,
%
compd
mg/kg
reduction
mg/kg
reduction
3q
0.06 ( 0.69 56 ( 6
16.51 ( 0.53 63 ( 8
0.34 ( 0.60
16.71 ( 0.57
ND
50 ( 6
74 ( 4
24 ( 6
bezafibrate
rosiglitazone
ND
11 ( 12
a
Six-weeks-old male Sprague-Dawley rats (n ) 6) were fed
on a high cholesterol diet ad libitum (1.25% cholesterol, 0.5% cholic
acid) for 10 days. From day 7 to day 10 the animals were dosed
orally by gavage once a day (rosiglitazone 10 and 30 mg/kg,
bezafibrate 10, 30, 100, and 300 mg/kg, and 3q 0.1, 0.3, 1.0, and
3 mg/kg). On day 10, nonfasting blood samples were collected and
analyzed for triglycerides (TG) and total cholesterol. ED50 values
were calculated via nonlinear regression using GraphPad PRISM
3.02 and are expressed as mean,( SEM. “% reduction” is the
maximum achieved reduction relative to vehicle treated control
group ( SEM.
F igu r e 8. Correlation between area under the plasma com-
pound-concentration versus time curve (AUCcompd) after 2.2 mg/
kg po treatment (3p , 3q, rosiglitazone, and pioglitazone, Table
5) and the percent maximum reduction in area under the
plasma glucose-concentration versus time curve (%AUCglu
obtained in db/db mice (Table 4).
)
5). The carbazole analogue 3q, which had shown ex-
tremely good lowering effects on both BG and on AUCglu
,
cordance with clinical results performed with rosiglita-
zone in diabetic patients. Data showed that rosiglitazone
was more effective in lowering blood glucose and HbA1c
when given twice a day than when given once a day.40
The results suggested that the best improvement of
insulin sensitivity would be obtained with drugs that
give long compound exposure. This could be obtained
by either selecting compounds with long plasma half-
lives, as seen with 3q, or by using sustained release
formulations.
To estimate the in vivo PPARR effect of 3q, we used
the high cholesterol fed rat model, which previously had
been used to select the clinically effective fibrates.41
Male Sprague-Dawley rats were fed a high cholesterol
diet for 10 days, the last 4 days on once a day oral drug
treatment. Nonfasted TG and total cholesterol were
measured and calculated as the percent reduction
relative to vehicle control. Rosiglitazone, even at a high
dose (30 mg/kg po), did not lower plasma TG or
cholesterol significantly (Table 7), suggesting that PPARγ
receptor activation was not involved in plasma lipid
regulation. Bezafibrate, on the other hand, produced the
expected dose-related reduction of both TG (63%) and
cholesterol (74%) when dosed at 10, 30, 100, and 300
mg/kg po for 4 days. At 100 times lower doses, 0.1, 0.3,
1.0, and 3.0 mg/kg po, 3q produced a similar dose-
related reduction of TG (56%) and cholesterol (50%,
had a Cmax similar to that of pioglitazone but an even
higher (approximately 5 times) AUCpo. Again the in-
creased drug exposure was a result of the low clearance
and the resulting long plasma half-life (t1/2po ) 1332
min). The half-life of 3q (Figure 7) was further extended
by entero-hepatic recirculation, which also accounted for
the high oral bioavailability (Fpo, Table 5).
The db/db mouse and rat pharmacokinetic data in
Tables 4 and 5, respectively, suggested that not only
was the in vivo potency dependent on the pharmaco-
kinetics of the compounds, but also the efficacy was
dependent on the pharmacokinetics. As depicted in
Figure 8, the drug exposure (AUCpo) of the compounds
correlated with their maximum ability to reduce the
AUCglu after OGTT.
The data further suggested that it was the extended
drug exposure (long drug half-life) that improved the
efficacy of the compound in vivo. To investigate this, a
separate experiment with rosiglitazone in db/db mice
was designed. Two doses (1 and 3 mg/kg/day po) of
rosiglitazone were given either once a day (1 × 1 mg/kg
and 1 × 3 mg/kg) or over three doses (3 × 0.33 mg/kg
and 3 × 1 mg/kg). The results showed that dividing the
dose over three times per day improved both the BG
lowering and especially the AUCglu lowering ability of
rosiglitazone (Table 6). These findings were in ac-

798 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
refluxed for 48 h. The solid was removed by filtration, and the
filtrate was evaporated. The residue was purified by column
chromatography on SiO2 using benzene:ethyl acetate (10:1) as
eluent. The first fractions gave the title compound in 3.2 g
Table 7). (The shape of the dose-response curve sug-
gested that higher doses would be needed to show the
full potential of 3q, data not shown). These rat data
confirmed that 3q also had potent PPARR activity in
vivo.
1
yield. H NMR (DMSO-d6) δ 1.16 (t, J ) 7 Hz, 3H), 1.22 (t, J
) 7 Hz, 3 H), 2.95 (d, J ) 7 Hz, 2H), 3.3-3.4 (m, 1H), 3.60 (m,
1H), 3.97 (t, J ) 7 Hz, 1H), 4.16 (q, J ) 7 Hz, 2H), 4.25 (t, J
) 7 Hz, 2H), 6.82 (d, J ) 7 Hz, 2H), 7.16 (d, J ) 7 Hz, 2H).
The next fractions gave the starting ester in 5.2 g yield (66%
recovery).
In conclusion, the SAR generated using in vitro
PPARR and PPARγ transactivation assays led to the
identification of the novel carbazole R-ethoxy-phenyl-
propionic acid derivative, 3q, with dual PPARR and
PPARγ activity both in vitro and in vivo. After treat-
ment of db/db mice with 3q, the improvement of the
insulin sensitivity, as measured by OGTT, was greater
than that seen with both pioglitazone and rosiglitazone.
This superior effect was either a consequence of the dual
PPARR and PPARγ activity, a result of extended drug
exposure, or a combination of the two. Furthermore, 3q
lowered plasma triglycerides and cholesterol in high
cholesterol fed rats, a model where PPARγ agonists had
no effect. Although additional experiments are needed
to fully evaluate the potential of 3q, the present data
gives hope for a future drug with improved clinical
efficacy compared to the known TZD insulin sensitizers.
Eth yl 3-(4-[2-(Diben zo[b,f]-1,4-th ia zep in -11-yla m in o)-
et h oxy]p h en yl)-2-et h oxyp r op ion a t e, 2b . A mixture of
ethyl 3-(4-(2-bromoethoxy)phenyl)-2-ethoxypropionate (3.05 g,
8.8 mmol), potassium phthalimide (2.0 g, 10.8 mmol), and
dimethylformamide (20 mL) was heated to 100 °C for 16 h,
benzene (200 mL) and water (200 mL) were added, and the
phases were separated. The organic phase was dried and the
solvent evaporated in vacuo. The residue was dissolved in
ethanol (60 mL), hydrazine hydrate (1.3 mL) was added, the
mixture was refluxed for 2 h and filtered, and the solvent was
evaporated to give 2.4 g (96%) of ethyl 3-(4-(2-aminoethoxy)-
phenyl)-2-ethoxypropionate as an oil. A mixture of dibenzo-
[b,f]-1,4-thiazepin-11(10H)-thione (2.20 g, 9 mmol)42, ethyl 3-(4-
(2-aminoethoxy)phenyl)-2-ethoxypropionate (2.60 g, 8.9 mmol),
and 3-methyl-1-butanol (70 mL) was stirred and heated at 150
°C for 16 h. The solvent was evaporated in vacuo, dichloro-
methane (50 mL) and water (50 mL) were added, the mixture
was filtered, and the phases were separated. The organic phase
was dried (MgSO4), and the solvent was evaporated in vacuo
to give a residue, which was purified by column chromatog-
raphy on silica gel, eluting with chloroform. This afforded 0.7
g of the starting thione and then 1.7 g (38%) of the title com-
Exp er im en ta l Section
Ch em istr y. Melting points were determined on a Bu¨chi
capillary melting point apparatus and are uncorrected. 1H
NMR spectra were recorded at either 200 MHz on a Bruker
Advance DPX 200 or at 300 MHz on a Bruker Advance DRX
300 instrument, and mass spectra were recorded on a Finnigan
5100 mass spectrometer. Column chromatography was per-
formed on silica gel 60 (70-230 mesh, ASTM, Merck). Ele-
mental analyses were performed by the Novo Nordisk Micro-
analytical Laboratory, Denmark, and were within (0.4% of
the calculated values. Optical purity was determined using
capillary electrophoresis performed on a HP3DCE capillary
electrophoresis instrument, Agilent, Waldborn, Germany.
Synthesis of compounds according to reaction Scheme 1.
3-(4-(N-Meth yl-N-(10,11-d ih yd r o-5H-d iben zo[a ,d ]cyclo-
h epten -5-yl)am in oeth oxy)ph en yl)-2-eth oxypr opion ic Acid,
3a , Hyd r och lor id e. A mixture of 5-(methylamino)-10,11-
dihydro-5H-dibenzo[a,d]cycloheptene hydrochloride (1.8 g, 6.94
mmol), ethyl 3-(4-(2-bromoethoxy)phenyl)-2-ethoxypropionate
(2.4 g, 6.95 mmol), potassium carbonate (2.9 g, 21 mmol), and
dimethylformamide (7 mL) was heated at 100 °C for 5 h.
Benzene (200 mL) and water (200 mL) were added, and the
phases were separated. The organic phase was dried and the
solvent evaporated in vacuo. The residue was purified by
chromatography on silica gel (Fluka 60, 40 g, benzene/
chloroform) to give 2.2 g (65%) of ethyl 3-(4-(N-methyl-N-
(10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5-yl)aminoethoxy)-
phenyl)-2-ethoxypropionate, 2a, as an oil. Rf ) 0.60 (chloroform/
ethanol/ammonia ) 20:2:0.1).
Compound 2a (1.5 g, 3.07 mmol) was dissolved in ethanol
(20 mL), and 20% sodium hydroxide (2 mL) was added. After
6 days, the ethanol was evaporated in vacuo, water (50 mL)
and acetic acid (2 mL) were added, and the mixture was
extracted with dichloromethane. The organic phase was dried
(MgSO4) and the solvent evaporated in vacuo. The residue (1.4
g) was dissolved in acetone and neutralized with a solution of
hydrogen chloride in diethyl ether. The solvents were evapo-
rated, and the residue was triturated with diethyl ether,
yielding 1.15 g (74%) of the title compound as an amorphous
solid (hemihydrate). 1H NMR (DMSO-d6) δ 1.07 (t, J ) 7.2
Hz, 3 H), 2.80-3.10 (s + m, 7 H), 3.30-3,60 (m, 4 H), 4.00 (t,
J ) 5.6 Hz; + m, 3 H), 4.49 (bs, 2 H), 5.84 (d, J ) 7.0 Hz, 1 H),
6.87 (d, J ) 7.9 Hz, 2 H), 7.10-7.60 (m, 10 H), 10.50 (bs, 1 H).
Eth yl 3-(4-(2-Br om oeth oxy)ph en yl)-2-eth oxypr opion ate.
A mixture of ethyl 3-(4-hydroxyphenyl-2-ethoxypropionate (7.8
g, 32.7 mmol), 1,2-dibromoethane (37.6 g, 0.2 mol), potassium
carbonate (4.4 g, 31.9 mmol), and 2-butanone (60 mL) was
1
pound as an oil. H NMR (CDCl3) δ 1.16 (t, J ) 7.2 Hz, 3H),
1.22 (t, J ) 7.2 Hz, 3H), 2.95 (d, J ) 6.4 Hz, 2H), 3.59 (m, 1
H), 3.34 (m, 1H), 3.96 (m, 3H), 4.16 (q, J ) 7.2 Hz, 2H), 4.25
(m, 2H), 5.15 (bs, 1H), 6.85-6.95 (m, 3H), 7.05-7.50 (m, 9H).
Gen er a l P r oced u r e A. Eth yl 3-(4-(2-(10,11-Dih yd r o-
d ib en zo[b,f]a zep in -5-yl)et h oxy)p h en yl)-2-et h oxyp r op i-
on a te, 2m . To a solution of 2-(10,11-dihydro-dibenzo[b,f]-
azepin-5-yl)ethanol (120 mg; 0.50 mmol) (generated from
10,11-dihydro-dibenzo[b,f]azepine with BuLi and ethylene
oxide in THF) in THF (20 mL), was added triphenylphosphine
(198 mg; 0.75 mmol). The reaction was cooled to 0 °C, and
diethyl azodicarboxylate (165 mg; 0.75 mmol) and ethyl
2-ethoxy-3-(4-hydroxyphenyl)propionate (1) (179 mg; 0.75
mmol) were added. The reaction was stirred at 0 °C for 2 h
and at room temperature for 16 h. Water (20 mL) was added,
and the mixture was extracted with dichloromethane (2 × 50
mL). The combined organic phases were dried and evaporated.
The residue was purified on column chromatography using
ethyl acetate:dichloromethane (9:1) as eluent to give the title
compound in 205 mg (90%) yield. 1H NMR (CDCl3) δ 1.1-1.25
(m, 6H), 2.92 (d, J ) 7 Hz, 2H), 3.17 (s, 4H), 3.27-3.38 (m,
1H), 3.52-3.63 (m, 1H), 3.95 (t, J ) 7 Hz, 1H), 4.04 (t, 2H),
4.10-4.20 (m, 4H), 6.72 (d, J ) 8 Hz, 2H), 6.88-6.95 (m, 2H),
7.05-7.17 (m, 8H).
The following compounds were made according to procedure
A using the appropriate 2-tricycle substituted ethanol:
Eth yl 3-{4-[3-(6H-Diben zo[b,e]oxep in -11-ylid en e)p r o-
p oxy]p h en yl}-2-eth oxyp r op ion a te, 2e. Isolated as an in-
separable 4:1 mixture of E and Z double-bond isomers, as a
1
pale yellow gum; yield 252 mg (53%). H NMR (CDCl3) δ 1.16
(t, J ) 7 Hz, 3H), 1.23 (t, J ) 7 Hz, 3H), 2.65 (q, J ) 7 Hz,
1.6H, E isomer), 2.90 (q, J ) 7 Hz, 0.4H, Z isomer), 2.94 (d, J
) 7 Hz, 2H), 3.29-3.40 (m, 1H), 3.53-3.67 (m, 1H), 3.97 (t, J
) 7 Hz, 1H), 4.01 (t, J ) 7 Hz, 1.6H, E isomer), 4.08 (t, J ) 7
Hz, 0.4H, Z isomer), 4.17 (t, J ) 7 Hz, 2H), 4.5-5.7 (very broad
m, 2H), 5.82 (t, J ) 7 Hz, 0.2H, Z isomer), 6.12 (t, J ) 7 Hz,
0.8H, E isomer), 6.75-6.90 (m, 4H), 7.1-7.4 (m, 8H). MS: 472
(M+), 426, 341, 326, 235 (100%), 221, 195, 107, 91.
Eth yl 2-Eth oxy-3-(4-[2-(9H-xan th en -9-yl)eth oxy]ph en yl)-
p r op ion a te, 2h . From 2-(9H-xanthen-9-yl)ethanol. Gave the
1
title compound as an oil: yield 1.9 g (45%). H NMR (CDCl3)
δ 1.16 (t, J ) 7 Hz, 3H), 1.21 (t, J ) 7 Hz, 3H), 2.09 (q, J ) 7

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 799
Hz, 2H) 2.94 (bt, J ) 7 Hz, 2H), 3.35 (m, 1H), 3.60 (m, 1H),
3.81 (t, J ) 7 Hz, 2H), 3.97 (t, J ) 7 Hz, 1H), 4.15 (q, J ) 7
Hz, 2H), 4.27 (t, J ) 7 Hz, 1H), 6.75 (bd, J ) 7 Hz, 2H), 6.98-
7.25 (m, 10H), 7.34 (2H, s).
was purified by column chromatography using toluene:ethyl
acetate (19:1) as eluent. The title compound was obtained in
385 mg (89%) yield. 1H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz, 3H),
1.22 (t, J ) 7 Hz, 3H), 2.93 (d, J ) 7 Hz, 2H), 3.32 (m, 1H),
3.57 (m, 1H), 3.93 (t, J ) 7 Hz, 1H), 4.15 (q, J ) 7 Hz, 2H),
4.32 (t, J ) 7 Hz, 2H), 4.70 (t, J ) 7 Hz, 2H), 6.74 (d, J ) 8
Hz, 2H), 7.10 (d, J ) 8 Hz, 2H), 7.27 (m, 2H), 7.50 (m, 4H),
8.12 (d, J ) 8 Hz, 2H).
2-(9H-Xa n th en -9-yl)eth a n ol. To a suspension of 9-xan-
thenylacetic acid43 (4.9 g, 21.5 mmol) in toluene (180 mL) was
added a mixture of sodium dihydrido-bis(2-methoxyethoxy)-
aluminate (60% solution in toluene, 14.5 g, 43.0 mmol) and
toluene (10 mL) dropwise over 10 min under an argon
atmosphere. The reaction mixture was stirred at ambient
temperature for 2 h. The mixture was cooled to 10 °C and
decomposed with water (5 mL) and 15% NaOH (40 mL). The
toluene layer was separated and the water layer extracted with
toluene (2 × 30 mL). The combined toluene solutions were
washed with water (2 × 30 mL) and brine (20 mL), dried
(MgSO4), and evaporated in vacuo. The residue (5.0 g) was
purified by column chromatography (silica gel Fluka 60, 80 g)
using chloroform as eluent. This afforded 3.8 g (78%) of the
title compound. 1H NMR (CDCl3) δ 1.50 (bs, 1 H), 1.92 (q, J )
7 Hz, 2 H), 3.55 (t, J ) 6.5 Hz, 2 H), 4.16 (t, J ) 6.7 Hz, 1 H),
7.11-7.02 (m, 4 H), 7.25-7.15 (m, 4 H).
The following compounds were made according to procedure
B using the appropriate tricyclyl ethanol:
(S)-Eth yl 3-(4-(2-(3-Br om o-ca r ba zol-9-yl)eth oxy)p h en -
yl)-2-et h oxyp r op ion a t e, 2r . From (S)-ethyl 2-ethoxy-3-(4-
hydroxyphenyl)propionate (1) (95.5% ee). Yield 510 mg (33%).
1H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz, 3H), 1.22 (t, J ) 7 Hz,
3H), 2.94 (d, J ) 7 Hz, 2H), 3.33 (m, 1H), 3.58 (m, 1H), 3.93
(t, J ) 7 Hz, 1H), 4.25-4.10 (q, J ) 7 Hz, 2H), 4.33 (t, J ) 7
Hz, 2H), 4.70 (t, J ) 7 Hz, 2H), 6.70 (d, J ) 8 Hz, 2H), 7.10 (d,
J ) 8 Hz, 2H), 7.27 (m, 1H), 7.40 (d, J ) 8 Hz, 1H), 7.60-7.47
(m, 3H), 8.05 (d, J ) 8 Hz, 1H), 8.20 (s, 1H).
(S)-Eth yl 3-(4-(2-(3,6 Dibr om o-ca r ba zol-9-yl)eth oxy)-
p h en yl)-2-eth oxyp r op ion a te, 2s. From (S)-1 (95.5% ee).
Yield 774 mg (100%). 1H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz,
3H), 1.22 (t, J ) 7 Hz, 3H), 2.94 (d, J ) 7 Hz, 2H), 3.31 (m,
1H), 3.58 (m, 1H), 3.94 (t, J ) 7 Hz, 1H), 4.17 (q, J ) 7 Hz,
2H), 4.30 (t, J ) 7 Hz, 2H), 4.80 (t, J ) 7 Hz, 2H), 6.68 (d, J
) 8 Hz, 2H), 7.10 (d, J ) 8 Hz, 2H), 7.40 (d, J ) 8 Hz, 2H),
7.58 (d, J ) 8 Hz, 2H), 8.14 (s, 2H).
Gen er a l P r oced u r e C. E t h yl 3-(4-(2-(10,11-Dih yd r o-
d ib en zo[b,f]a zep in -5-yl)p r op oxy)p h en yl)-2-et h oxyp r o-
p ion a te, 2l. A mixture of ethyl 3-(4-hydroxyphenyl)-2-ethoxy-
propionate (2.26 g, 10.75 mmol), 3-(10,11-dihydro-5H-dibenzo-
[b,f]azepin-5-yl)propanol methane sulfonate (3.55 g, 10.71
mmol), and potassium carbonate (7.65 g, 55.35 mmol) in DMF
(75 mL) was heated at 90 °C for 30 h. The cooled reaction
mixture was poured into water (500 mL) and extracted with
benzene (3 × 100 mL), and the extracts were washed with
water (200 mL), dried (MgSO4), and evaporated in vacuo. The
residue (4.75 g) was purified by column chromatography on
silica gel (Fluka 60, 150 g) using benzene/chloroform 20:1 as
eluent to give the title compound in 1.84 g (36%) yield. 1H NMR
(CDCl3) δ 1.15 (t, J ) 7 Hz, 3H), 1.21 (t, J ) 7 Hz, 3H), 2.03
(q, J ) 6.4 Hz, 2H), 2.92 (d, J ) 6.8 Hz, 2H), 3.14 (s, 4H), 3.33
(m, 1H), 3.59 (m, 1H), 3.87-4.00 (m, 5H), 4.15 (q, J ) 7.2 Hz,
2H), 6.72 (dt, J ) 8.7 and 2.2 Hz, 2H), 6.87-6.95 (m, 2H),
7.05-7.15 (m, 8H).
Eth yl 2-Eth oxy-3-[4-(2-flu or en -9-ylid en e-eth oxy)p h en -
yl]p r op ion a te, 2i. From 2-fluoren-9-ylidene-ethanol.44 Yield
1
280 mg (60%). H NMR (CDCl3) δ 1.17 (t, J ) 7 Hz, 3H), 1.22
(t, J ) 7 Hz, 3H), 2.97 (d, J ) 7 Hz, 2H), 3.29-3.40 (m, 1H),
3.54-3.66 (m, 1H), 3.98 (t, J ) 7 Hz, 1H), 4.17 (q, J ) 7 Hz,
2H), 5.32 (d, J ) 6 Hz, 2H), 6.87 (t, J ) 6 Hz, 1H), 6.92 (d, J
) 8 Hz, 2H), 7.18 (d, J ) 8 Hz, 2H), 7.22-7.45 (m, 4H), 7.57-
7.77 (m, 4H). MS: 428 (M+), 382, 191 (100%).
Eth yl 2-Eth oxy-3-(4-[2-(9H-flu or en -9-yl)eth oxy]ph en yl)-
p r op ion a te, 2j. From 2-(9H-fluoren-9-yl)ethanol.45 Yield 0.20
1
g (47%). H NMR (CDCl3) δ 1.17 (t, J ) 7 Hz, 3H), 1.22 (t, J
) 7 Hz, 3H), 2.46 (q, J ) 7 Hz, 2H), 2.93 (d, J ) 7 Hz, 2H),
3.28-3.40 (m, 1H), 3.52-3.65 (m, 1H), 3.90 (t, J ) 7 Hz, 2H),
3.95 (t, J ) 7 Hz, 1H), 4.16 (q, J ) 7 Hz, 2H), 4.15-4.28 (m,
1H), 6.74 (d, J ) 8 Hz, 2H), 7.11 (d, J 0 8 Hz, 2H), 7.25-7.42
(m, 4H), 7.52 (d, J ) 8 Hz, 2H), 7.78 (d, J ) 8 Hz, 2H), 7.78 (d,
J ) 8 Hz, 2H). MS 430 (M+), 384, 299, 193, 179, 165 (100%),
107.
Eth yl 3-(4-(2-(Diben zo[b,f]a zep in -5-yl)eth oxy)p h en yl)-
2-eth oxyp r op ion a te, 2n . Yield 172 mg (75%). 1H NMR
(CDCl3) δ 1.11-1.29 (m, 6H), 2.93 (d, J ) 7 Hz, 2H), 3.25-
3.91 (m, 1H), 3.50-3.67 (m, 1H), 3.95 (t, J ) 7 Hz, 1H), 4.02-
4.21 (m, 6H), 6.75 (t, J ) 7 Hz, 4H), 6.95-7.30 (m, 10 H).
Eth yl 3-(4-(2-(â-Ca r bolin -9-yl)eth oxy)p h en yl)-2-eth oxy-
1
The following compounds were made according to procedure
C using the appropriate mesylate.
p r op ion a te, 2o. Yield 1.09 g (76%). H NMR (CDCl3) δ 1.12
(t, J ) 7 Hz, 3H), 1.21 (t, J ) 7 Hz, 3H), 2.90 (d, J ) 7 Hz,
2H), 3.24-3.37 (m, 1H), 3.51-3.62 (m, 1H), 3.91 (t, J ) 7 Hz,
1H), 4.14 (q, J ) 7 Hz, 2H), 4.38 (t, J ) 7 Hz, 2H), 4.80 (t, J
) 7 Hz, 2H), 6.74 (d, J ) 8 Hz, 2H), 7.08 (d, J ) 8 Hz, 2H),
7.28-7.70 (m, 4H), 7.96 (d, J ) 7 Hz, 1H), 8.15 (d, J ) 8 Hz,
1H), 8.49 (d, J ) 7 Hz, 1H).
Eth yl 3-(4-(3-(10,11-Dih yd r o-5H-d iben zo[a ,d ]cycloh ep -
ten -5-ylid en e)p r op oxy)p h en yl)-2-eth oxyp r op ion a te, 2c.
From 3-(10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5-ylidene)-
1
propanol.46 Yield 1.5 g (21%). H NMR (CDCl3) δ 0.99 (t, J )
7 Hz, 3H), 1.27 (t, J ) 7 Hz, 3H), 2.69 (q, J ) 7 Hz, 2H), 3.06
(d, J ) 7 Hz, 2H), 3.17 (bs, 4H), 3.45 (m, 1H), 3.68 (m, 1H),
4.07 (m, 3H), 4.27 (q, J ) 7 Hz, 2H), 6.06 (t, J ) 7 Hz, 1H),
6.85 (d, J ) 6 Hz, 2H), 7.10-7.35 (m, 10H).
E t h yl (S)-3-(4-(2-(â-Ca r b olin -9-yl)et h oxy)p h en yl)-2-
eth oxypr opion ate, 2p. Using ethyl (S)-2-ethoxy-3-(4-hydroxy-
phenyl)propionate (95.5% ee). Yield 1.67 g (60%). 1H NMR
(CDCl3) δ 1.12 (t, J ) 7 Hz, 3H), 1.21 (t, J ) 7 Hz, 3H), 2.90
(d, J ) 7 Hz, 2H), 3.24-3.37 (m, 1H), 3.51-3.62 (m, 1H), 3.91
(t, J ) 7 Hz, 1H), 4.14 (q, J ) 7 Hz, 2H), 4.38 (t, J ) 7 Hz,
2H), 4.80 (t, J ) 7 Hz, 2H), 6.74 (d, J ) 8 Hz, 2H), 7.08 (d, J
) 8 Hz, 2H), 7.28-7.70 (m, 4H), 7.96 (d, J ) 7 Hz, 1H), 8.15
(d, J ) 8 Hz, 1H), 8.49 (d, J ) 7 Hz, 1H).
Gen er a l P r oced u r e B. (S)-Eth yl 3-(4-(2-Ca r ba zol-9-yl-
eth oxy)p h en yl)-2-eth oxyp r op ion a te, 2q. To an ice cooled
solution of 2-(carbazol-9-yl)ethanol (211 mg; 1.0 mmol), (S)-
ethyl 2-ethoxy-3-(4-hydroxyphenyl)propionate (1) (99.0% ee)
(238 mg; 1 mmol), and tributylphosphine (370 µL; 1.5 mmol)
in dry benzene (10 mL) was added 1,1′-(azodicarbonyl) dipiperi-
dine (380 mg; 1.5 mmol). The reaction mixture was stirred at
0 °C for 1 h, an additional 10 mL of benzene added, and the
reaction mixture was stirred for another 1 h. Heptane (10 mL)
was added to the reaction mixture, and the resulting precipi-
tate was removed by filtration. The filtrate was evaporated in
vacuo and the residue suspended in heptane. After filtration,
the heptane phase was evaporated to dryness. The residue
Eth yl 2-Eth oxy-3-(4-[2-(11H-5-oxa -10-th ia -d iben zo[a ,d]-
cycloh ep ten -11-yl)eth oxy]p h en yl)p r op ion a te, 2f. Yield
4.6 g (60%). 1H NMR (CDCl3) δ 1.16 (t, J ) 7 Hz, 3H), 1.20
(dt, J ) 0.6 and 7 Hz, 3H), 2.53-2.80 m, 2H), 2.94 (d, J ) 6.6
Hz, 2H), 3.34 (dq, J ) 7.0 and 9.1 Hz, 1H), 3.59 (dq, J ) 7 and
9.1 Hz, 1H), 3.96 (m, 1H), 4.00 (m, 1H), 4.15 (q, J ) 7 Hz,
2H), 4.18 (m, 1H), 4.70 (dd, J ) 6.9 and 8.5 Hz, 1H), 6.80 (t,
J ) 7 Hz, 2H), 6.93 (ddd, J ) 1.5, 6.8 and 7.8 Hz, 1H), 7.01-
7.26 (m, 9H).
Eth yl 3-(4-Diben zo[d ,g]d ioxa zocin -12-yl)-1-p r op oxy)-
p h en yl-2-eth oxyp r op ion a te, 2k . From 3-(4-dibenzo[d,g]-
dioxazocin-12-yl)-1-propanol. Yield 2.45 g (42%). 1H NMR
(CDCl3) δ 1.15 (t, J ) 7.2 Hz, 3H), 1.21 (t, J ) 7.2 Hz, 3H),
1.92 (q, J ) 5.7 Hz, 2H), 3.33 (m, 1H), 3.59 (m, 1H), 3.81 (t,
J
) 5.7 Hz, 2H), 3.97 (t, J ) 5.7 Hz, 3H), 4.15 (q, J )
7.2 Hz, 2H), 5.71 (s, 2H), 6.75 (dt, J ) 8.4 Hz, 2H), 7.20-6.95
(m, 10H).
Gen er a l P r oced u r e D. E t h yl 3-(4-[2-(10,11-Dih yd r o-
d ib en zo[a ,d ]cycloh ep t en -5-ylid en e)et h oxy]p h en yl)-2-

800 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
eth oxyp r op ion a te, 2d . A mixture of 1 (2.38 g, 10.0 mmol),
5-(2-bromo-1-ethylidene)-10,11-dihydro-5H-dibenzo[a,d]cyclo-
heptene47 (2.75 g, 10.0 mmol), and potassium carbonate (5.14
g, 30.0 mmol) in dimethylformamide (30 mL) was heated at
100 °C for 20 h. The reaction mixture was diluted with benzene
(80 mL), washed with 5% aqueous citric acid (3 × 25 mL) and
with saturated NaHCO3 (25 mL), dried (MgSO4), and evapo-
rated. The residue (4.88 g) was purified by column chroma-
tography on silica gel (benzene eluent) to yield the title
Gen er a l P r oced u r e for Hyd r olysis of th e Ester to th e
F in a l Acid . 3-(4-(2-(10,11-Dih yd r o-d iben zo[b,f]a zep in -5-
yl)eth oxy)p h en yl)-2-eth oxyp r op ion ic Acid , 3m . A solution
of ethyl 3-(4-(2-(10,11-dihydro-dibenzo[b,f]azepin-5-yl)ethoxy)-
phenyl)-2-ethoxypropionate (191 mg, 0.42 mmol) in ethanol
(13 mL) and aqueous 1 N sodium hydroxide (4.5 mL) was
stirred at 90 °C for 1 h. The reaction mixture was evaporated
and the residue dissolved in water (7 mL). The aqueous phase
was extracted with ethyl acetate (2 × 50 mL) after acidification
with 1 N HCl (7.5 mL). The combined organic phases were
dried, evaporated, and purified by column chromatography,
using dichloromethane:methanol (9:1) as eluent, to give the
title compound in 176 mg (97%) yield. 1H NMR (CDCl3) δ
1.1 (t, J ) 7 Hz, 3H), 2.72-3.06 (m, 2H), 3.17 (s, 4H), 3.35
(m, 1H), 3.55 (m, 1H), 3.94-4.05 (m, 3H), 4.15 (t, J ) 7 Hz,
2H), 6.69 (d, J ) 8 Hz, 2H), 6.85-6.95 (m, 2H), 7.03-7.15
(m, 8H), 8.5-9.0 (bs, 1H). MS 431 (M+), 222, 208 (100%),
193, 165, 91.
1
compound; 2.3 g (53%). H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz,
3H), 1.95 (t, J ) 7 Hz, 3H), 2.92 (d, J ) 7 Hz, 2H), 3.17 (bs,
4H), 3.28 (m, 1H), 3.58 (m, 1H), 3.94 (m, 1H), 4.14 (q, J ) 7
Hz, 2H), 4.59 (bs, 2H), 6.10 (t, J ) 7 Hz, 1H), 6.71 (m, 2H),
7.05-7.25 (m, 9H), 7.32 (m, 1H).
The following compound was made according to procedure
D using 12-(2-bromoethylidene)-12H-dibenzo[d,g]-1,3-dioxo-
cine:
E t h yl 3-(4-(2-(12H -Dib en zo[d ,g]-1,3-d ioxocin e-12-yli-
d en e)eth oxy)p h en yl)-2-eth oxyp r op ion a te, 2g. From 12-
(2-bromoethylidene)-12H-dibenzo[d,g]-1,3-dioxocine. Yield 0.97
The following compounds were made as described above
using the appropriate starting material, 2.
1
3-(4-(3-(10,11-Dih yd r o-5H-d iben zo[a ,d ]cycloh ep ten -5-
ylid en e)p r op oxy)p h en yl)-2-eth oxyp r op ion ic Acid , 3c, L-
Lysin e Sa lt. The resulting residue (free acid; 1.1 g, 78%) was
dissolved in ethanol and treated with L-lysine monohydrate
(0.41 g), and the ethanol was evaporated. The residue was
triturated with diethyl ether, and the crystalline product was
collected by filtration and air-dried to give the title salt as the
g (62%). H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz, 3H), 1.20 (t, J
) 7.2 Hz, 3H), 2.93 (d, J ) 7.1 Hz, 2H), 3.33 (m, 1H), 3.58 (m,
1H), 3.95 (t, J ) 7.2 Hz, 1H), 4.14 (q, J ) 7.2 Hz, 2H), 4.47 (d,
J ) 6.2 Hz, 2H), 5.90 (s, 2H), 6.21 (t, J ) 6.2 Hz, 1H), 6.73 (d,
J ) 8.2 Hz, 2H), 6.93-7.32 (m, 7H), 7.35 (m, 2H), 7.38 (m,
1H).
12-(2-Br om oet h ylid en e)-12H -d ib en zo[d ,g]-1,3-d ioxo-
cin e. To a solution of vinylmagnesium bromide (prepared from
vinyl bromide (8.65 g, 80.0 mmol) and magnesium turnings
(2.14 g, 88.0 mmol) in tetrahydrofuran (120 mL)), which was
cooled to 10 °C, was added a solution of 12H-dibenzo[d,g]-1,3-
dioxocin-12-one in tetrahydrofuran (30 mL) dropwise over 25
min. The reaction mixture was stirred at room temperature
for 3 h and then cooled to 0 °C, keeping the temperature
between 0 and 10 °C. The mixture was decomposed with a
solution of ammonium chloride (10 g) in water (50 mL). The
mixture was then extracted with benzene (100 mL), the
organic layer separated, and the aqueous layer extracted with
additional benzene (2 × 50 mL). The combined organic extracts
were dried over MgSO4 and evaporated in vacuo. The residue
(13.7 g) was purified by column chromatography on silica gel
(200 g). Benzene and benzene/ethyl acetate fractions afforded
12-vinyl-12H-dibenzo[d,g]-1,3-dioxocin-12-ol (6.0 g), mp 93-
1
dihydrate. Yield 1.45 g, mp 148-150 °C. H NMR (DMSO) δ
1.03 (t, J ) 7 Hz, 3H), 1.66 (m, 6H), 2.51 (m, 2H), 2.70-2,95
(m, 4H), 3.07 (bs, 4H), 3.31-3.59 (m, 2H), 3.76 (m, 1H), 4.02
(t, J ) 6 Hz, 2H), 5.91 (t, J ) 7 Hz, 1H), 6.26 (bs, 8H), 6.75
(dd, J ) 8 and 2H), 7.00-7.35 (m, 10H).
3-(4-[2-(10,11-Dih yd r o-d iben zo[a ,d ]cycloh ep ten -5-yli-
d en e)eth oxy]p h en yl)-2-eth oxyp r op ion ic Acid , 3d . The
residue was crystallized from a mixture of toluene (8 mL) and
n-heptane (8 mL) to give the title compound in 1.60 g (74%)
1
yield; mp 147-150 °C. H NMR (CDCl3) δ 1.15 (t, J ) 7 Hz,
3H), 2.99 (m, 2H), 3.17 (bs, 4H), 3.42-3.58 (m, 2H), 4.01 (dd,
J ) 8 and 4 Hz, 1H), 4.59 (bd, 2H), 6.11 (t, J ) 7 Hz, 1H), 6.72
(m, 2H), 7.02-7.35 (m, 10H).
3-(4-(3-(6H -Dib en zo[b,e]oxep in -11-ylid en e)p r op oxy)-
p h en yl)-2-eth oxyp r op ion ic Acid , 3e. Isolated as an insepa-
rable 4:1 mixture of E and Z double-bond isomers, as a pale
1
1
98 °C. H NMR (CDCl3) δ 2.57 (bs, 1 H), 5.08 (d, J ) 5.0 Hz,
yellow glass; 186 mg (80%). H NMR (CDCl3) δ 1.16 (t, J ) 7
1H); 5.21 (d, J ) 5.0 Hz, 1H), 5.12 (dd, J ) 1.3 and 10.6 Hz,
1H), 5.43 (dd, J ) 1.3 and J ) 17.0 Hz, 1H), 6.47 (dd, J ) 10.6
and J ) 17.0 Hz, 1H), 7.02 (dd, J ) 1.6 and J ) 7.5 Hz, 2H),
7.20 (dt, J ) 1.6 and J ) 7.5 Hz, 2H), 7.28 (dt, J ) 1.9 and J
) 7.5 Hz, 2H); 7.75 (dd, J ) 1.9 and J ) 7.5 Hz, 2H).
Hz, 3H), 2.65 (q, J ) 7 Hz, 1.6H, E isomer), 2.90 (q, J ) 7 Hz,
0.4 H, Z isomer), 2.93 (dd, J ) 14 and 9 Hz, 1H), 3.05 (dd, J
) 14 and 5 Hz, 1H), 3.32-3.70 (m, 2H), 3.94-4.10 (m, 3H),
4.5-5.7 (very broad m, 2H), 5.80 (t, J ) 7 Hz, 0.25H, Z isomer),
6.12 (t, J ) 7 Hz, 0.75H, E isomer), 6.7-6.95 (m, 4H), 7.05-
7.20 (m, 2H), 7.20-7.40 (m, 6H). MS: 444 (M+), 341, 326, 235
(100%), 221, 195, 107, 91.
To a solution of the above dioxocinol (2.2 g, 8.6 mmol) in
dichloromethane (95 mL), was added trimethylbromosilane
(1.50 g, 9.8 mmol) in several portions through a septum inlet
under an atmosphere of nitrogen. The reaction mixture was
stirred at room temperature for 1.5 h and poured into
saturated sodium hydrogen carbonate solution (30 mL). The
organic layer was separated, washed with water (2 × 20 mL)
and brine (20 mL), and dried over MgSO4. After evaporation
of the solvent in vacuo, the crude oily 12-(2-bromoethylidene)-
12H-dibenzo[d,g]-1,3-dioxocine (2.5 g, 91%) was used in the
above step without further purification. 1H NMR (CDCl3) δ
3.89 (d, J ) 8.5 Hz, 2H), 5.89 (s, 2H), 6.22 (t, J ) 8.5 Hz, 1H),
7.14 (m, 8H).
2-Eth oxy-3-(4-(2-(11H-5-oxa -10-th ia -d iben zo[a ,d ]cyclo-
h ep ten -11-yl)eth oxy)p h en yl)p r op ion ic Acid , 3f, L-Lysin e
Sa lt. The resulting residue (3.8 g, 88%) was dissolved in
ethanol and treated with L-lysine (1.25 g), the solvent evapo-
rated, and the residue triturated with diethyl ether. The
resulting crystalline product was collected by filtration and
1
air-dried to give the title salt: 4.35 g; mp 153.5-154.5 °C. H
NMR (DMSO-d6) δ 1.00 (t, J ) 7 Hz, 3H), 1.2-2.0 (m, 6H),
2.4-3.0 (m, 6H), 3.15 (m, 1H); 3.43 (m, 1H), 3.56 (m, 1H), 3.66
(bs, 1H), 4.00 (bs, 1H), 4.13 (bs, 1H), 4.90 (t, J ) 6.7 Hz, 1H),
6.82 (d, J ) 7.9 Hz, 2H), 7.00-7.50 (m, 10H), 7.82 (bs, 5H).
3-(4-[2-(Diben zo[b,f]-1,4-th ia zep in -11-yla m in o)eth oxy]-
p h en yl)-2-et h oxyp r op a n oic Acid , 3b . Ethyl 3-(4-[2-(di-
benzo[b,f]-1,4-thiazepin-11-ylamino)ethoxy]phenyl)-2-ethoxy-
propionate (1.6 g, 3.26 mmol) was dissolved in ethanol (30 mL),
and 20% sodium hydroxide (3 mL) was added. After 6 days
the ethanol was evaporated in vacuo, water (50 mL) and acetic
acid (3 mL) were added, and the product was filtered off and
dried, yielding 1.4 g (87%) of the title compound as the hydrate.
1H NMR (DMSO-d6) δ 1.03 (t, J ) 7.2 Hz, 3H), 2.70-2.82 (m,
1H), 2.91 (m, 1H), 3.25 (m, 1H), 3.54 (m, 1H), 3.78 (bs, 2H),
3.88 (dd, J ) 7.6 and 4.2 Hz, 1H), 4.26 (t, J ) 4.9 Hz, 2H),
6.82-7.15 (m, 4H), 7.10-7.25 (m, 3H), 7.3-7.6 (m, 6H).
3-(4-(2-(12H -Dib en zo[d ,g]-1,3-d ioxocin e-12-ylid en e)-
et h oxy)p h en yl)-2-et h oxyp r op ion ic Acid , 3g, L-Lysin e
Sa lt. Yield 0.79 g (83%). This acid (0.76 g, 1.6 mmol) was
dissolved in acetone (30 mL), L-lysine (0.234 g, 1.6 mmol)
in water (3 mL) was added, and the mixture was stirred at
room temperature for 2 h. The solution was filtered and
evaporated, and the residue was stirred with a mixture of Et2O
(20 mL) and acetone (20 mL) overnight. The resulting solid
was collected by filtration, washed with Et2O (2 × 30 mL),
and dried to give the title compound as a partial hydrate: 0.90
1
g (93.5%); mp 162-168 °C. H NMR (DMSO-d6) δ 1.04 (t, J )
6.8 Hz, 3H), 1.38-1.89 (m, 6H), 2.75 (m, 3H), 2.90 (dd, J )

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 801
14.4 and 4.3 Hz, 1H), 3.27 (m, 3H), 3.58 (m, 1H), 3.75 (m, 1H)
4.49 (d, J ) 6.9 Hz, 2H), 5.89 (bs, 10H), 6.20 (t, J ) 6.3 Hz,
1H), 6.75 (d, J ) 7.7 Hz, 2H), 6.91-7.52 (m, 10H).
2-Eth oxy-3-[4-(2-flu or en -9-ylid en e-eth oxy)p h en yl]p r o-
p ion ic Acid , 3i. Lithium hydroxide (1 M, 1.0 mL, 1.0 mmol)
was added to a suspension of ethyl 2-ethoxy-3-[4-(2-fluoren-
9-ylidene-ethoxy)phenyl]propionate (214 mg, 0.5 mmol) in
ethanol (5 mL), and the resulting mixture was heated to gentle
reflux for 30 min. The cooled mixture was partitioned between
water (30 mL) and dichloromethane (20 mL) and acidified to
pH 1 by adding 1 N HCl (3 mL), and the organic phase was
collected. The aqueous phase was further extracted with
dichloromethane (3 × 20 mL), and the combined organics were
washed with brine, dried (MgSO4), and evaporated to give a
yellow gum. The product was purified by column chromatog-
raphy on silica gel (3% methanol in dichloromethane eluent)
to give 2-ethoxy-3-[4-(2-fluoren-9-ylidene-ethoxy)phenyl]pro-
2-Eth oxy-3-(4-[2-(9H-xa n th en -9-yl)eth oxy]p h en yl)p r o-
p ion ic Acid , 3h . Gave the title compound as an oil. Yield 1.3
1
g (77%). H NMR (CDCl3) δ 1.17 (t, J ) 6.4 Hz, 3H), 2.11 (q,
J ) 6.4 Hz, 2H), 2.80-3.20 (m, 2H), 3.40-3.70 (m, 3H), 3.82
(t, J ) 6.8 Hz, 2H), 4.04 (dd, J ) 4.3 and 7.6 Hz, 1H), 4.28 (t,
J ) 6.8 Hz, 1H), 6.77 (d, J ) 8.6 Hz, 2H), 7.01-7.24 (m, 10H).
2-E t h oxy-3-(4-[2-(9H-flu or en -9-yl)et h oxy]p h en yl)p r o-
1
p ion ic Acid , 3j. Yield 170 mg (95%). H NMR CDCl3) δ 1.17
(t, J ) 7 Hz, 3H), 2.46 (q, J ) 7 Hz, 2H), 2.93 (dd, J ) 16 and
7 Hz, 1H), 3.04 (dd, J ) 16 and 5 Hz, 1H), 3.38-3.50 (m, 1H),
3.50-3.65 (m, 1H), 3.90 (t, J ) 7 Hz, 2H), 4.04 (dd, J ) 7 and
5 Hz, 1H), 4.23 (t, J ) 7 Hz, 1H), 6.74 (d, J ) 8 Hz, 2H), 7.11
(d, J ) 8 Hz, 2H), 7.25-7.42 (m, 4H), 7.52 (d, J ) 8 Hz, 2H),
7.75 (d, J ) 8 Hz, 2H). MS 402 (M+), 299, 193, 178, 165 (100%),
107.
1
pionic acid, as a yellow solid; 0.104 g (51%). H NMR (CDCl3)
δ 1.17 (t, J ) 7 Hz, 3H), 2.98 (dd, J ) 14 and 7 Hz, 1H), 3.10
(dd, J ) 14 and 4 Hz, 1H), 3.40-3.70 (m, 2H), 4.06 (dd, J ) 7
and 4 Hz, 1H), 5.33 (d, J ) 6 Hz, 2H), 6.87 (t, J ) 6 Hz, 1H),
6.98 (d, J ) 8 Hz, 2H), 7.20 (d, J ) 8 Hz, 2H), 7.20-7.47 (m,
4H), 7.55-7.80 (m, 4H). MS: 400 (M+), 435, 297, 235, 209,
191 (100%), 165.
3-(4-Diben zo[d ,g]d ioxa zocin -12-yl)-1-p r op oxy)p h en yl-
2-eth oxyp r op ion ic Acid , 3k . Yield 1.92 g (79%). 1H NMR
(CDCl3) δ 1.15 (t, J ) 7.2 Hz, 3H), 1.95 (q, J ) 5.7 Hz, 2H),
3.1-2.85 (m, 2H), 3.6-3.4 (m, 2H), 3.81 (t, J ) 5.7 Hz, 2H),
3.97 (t, J ) 5.7 Hz, 2H), 4.00 (m, 1H), 5.71 (s, 2H), 6.75 (dt, J
) 8.8 Hz, 2H), 7.20-6.95 (m, 10H).
3-(4-(2-(10,11-Dih ydr o-diben zo[b,f]azepin -5-yl)pr opoxy)-
p h en yl)-2-eth oxyp r op ion ic Acid , 3l. Yield 1.65 g (98%). 1H
NMR (DMSO) δ 1.17 (t, J ) 7 Hz, 3H), 2.04 (q, J ) 7 Hz, 2H),
2.91 (dd, J ) 7.3 and 14.3 Hz, 1H), 3.08 (dd, J ) 4.3 and 14.3
Hz, 1H), 3.15 (s, 4H), 3.61-3.42 (m, 2H), 3.92 (t, J ) 7 Hz,
2H), 3.97 (t, J ) 7 Hz, 2H), 4.04 (dd, J ) 4.3 and 7.3 Hz, 1H),
6.73 (m, 2H), 6.92 (m, 2H), 7.05-7.15 (m, 8H).
Gen er a l P r oced u r e for Cr yst a lliza t ion a s Ar gin in e
Sa lt. (S)-3-(4-(2-Ca r ba zol-9-yl-eth oxy)p h en yl)-2-eth oxy-
p r op ion ic Acid , 3q , L-Ar gin in e. A solution of L-arginine
(2.9 g, 16.67 mmol) in water (10 mL) was dropwise added to a
60 °C warm stirring solution of (S)-3-(4-(2-carbazol-9-yl-
ethoxy)phenyl)-2-ethoxypropionic acid, 3q (7.0 g, 16.6 mmol),
in ethanol (250 mL). The mixture was stirred at room tem-
perature overnight, and the crystals were collected by filtration
1
and dried. Yield 9.5 g (99%). H NMR (CD3OD) δ 1.07 (t, J )
7 Hz, 3H), 1.60-1.73 (m, 2H), 1.78-1.90 (m, 2H), 2.85 (dd, J
) 8 and 16 Hz, 1H), 2.90 (dd, J ) 5 and 16 Hz), 3.10-3.30 (m,
3H), 3.50-3.63 (m, 2H), 3.85 (q, J ) 4 Hz, 1H), 4.34 (t, J ) 4
Hz, 2H), 4.73 (t, J ) 4 Hz, 2H), 6.67 (d, J ) 8 Hz, 2H), 7.09 (d,
J ) 8 Hz, 2H), 7.19 (t, J ) 7 Hz; 2H), 7.45 (t, J ) 7 Hz, 2H),
7.57 (d, J ) 7 Hz, 2H), 8.06 (d, J ) 7 Hz, 2H).
3-(4-(2-(Dibenzo[b,f]azepin-5-yl)ethoxy)phenyl)-2-ethoxy-
1
p r op ion ic Acid , 3n . Yield 151 mg (93%). H NMR (CDCl3) δ
1.12 (t, J ) 7 Hz, 3H), 2.84-3.05 (m, 2H), 3.28-3.40 (m, 1H),
3.50-3.62 (m, 1H), 3.93-4.18 (m, 5H), 6.75 (m, 4H), 6.95-
7.78 (m, 10H), 8.5-9.0 (bs, 1H). MS 429 (M+), 220, 207, 206
(100%), 178, 165, 128, 91.
In Vit r o Tr a n sa ct iva t ion Assa ys. Cell Cu lt u r e a n d
Tr a n sfection . HEK293 cells were grown in DMEM + 10%
FCS. Cells were seeded in 96-well plates the day before
transfection to give a confluency of 50-80% at transfection. A
total of 0.8 µg of DNA containing 0.64 µg of pM1R/γLBD, 0.1
µg of pCMVâGal, 0.08 µg of pGL2(Gal4)5, and 0.02 µg of
pADVANTAGE was transfected per well using FuGene trans-
fection reagent according to the manufacturers instructions
(Roche). Cells were allowed to express protein for 48 h followed
by addition of compound.
3-(4-(2-(â-Ca r bolin -9-yl)eth oxy)p h en yl)-2-eth oxyp r op i-
1
on ic Acid , 3o. Yield 50 mg (30%). H NMR (CDCl3) δ 1.20 (t,
J ) 7 Hz, 3H), 3.03 (d, J ) 7 Hz, 2H), 3.38-3.52 (m, 1H), 3.62-
3.76 (m, 1H), 4.10 (t, J ) 7 Hz, 1H), 4.37 (t, J ) 7 Hz, 2H),
4.70 (t, J ) 7 Hz, 2H), 6.60 (d, J ) 8 Hz, 2H), 7.17 (d, J ) 8
Hz, 2H), 7.35-7.73 (m, 3H), 8.08 (d, J ) 7 Hz, 1H), 8.19 (d, J
) 8 Hz, 1H), 8.41 (d, J ) 7 Hz, 1H), 8.67 (s, 1H), 8.8-9.3 (bs,
1H).
(S)-3-(4-(2-(â-Car bolin -9-yl)eth oxy)ph en yl)-2-eth oxypr o-
p ion ic Acid , 3p . Yield 850 mg (70%). 1H NMR (CDCl3) δ 1.20
(t, J ) 7 Hz, 3H), 3.03 (d, J ) 7 Hz, 2H), 3.38-3.52 (m, 1H),
3.62-3.76 (m, 1H), 4.10 (t, J ) 7 Hz, 1H), 4.37 (t, J ) 7 Hz,
2H), 4.70 (t, J ) 7 Hz, 2H), 6.60 (d, J ) 8 Hz, 2H), 7.17 (d, J
) 8 Hz, 2H), 7.35-7.73 (m, 3H), 8.08 (d, J ) 7 Hz, 1H), 8.19
(d, J ) 8 Hz, 1H), 8.41 (d, J ) 7 Hz, 1H), 8.67 (s, 1H), 8.8-9.3
(bs, 1H). 96.1% ee.
P la sm id s. Human PPARR and PPARγ was obtained by
PCR amplification using cDNA synthesized by reverse tran-
scription of mRNA from liver and adipose tissue, respectively.
Amplified cDNAs were cloned into pCR2.1 and sequenced. The
ligand binding domain (LBD) of each PPAR isoform was
generated by PCR (PPARR: aa 167 - C-terminus; PPARγ: aa
165 - C-terminus) and fused to the DNA binding domain
(DBD) of the yeast transcription factor GAL4 by subcloning
fragments in frame into the vector pM148 generating the
plasmids pM1RLBD and pM1γLBD. Ensuing fusions were
verified by sequencing. The reporter was constructed by
inserting an oligonucleotide encoding five repeats of the GAL4
recognition sequence (5 × CGGAGTACTGTCCTCCG(AG))49
into the vector pGL2 promotor (Promega) generating the
plasmid pGL2(GAL4)5. pCMVâGal was purchased from Clon-
tech and pADVANTAGE was purchased from Promega.
(S)-3-(4-(2-Ca r ba zol-9-yl-eth oxy)p h en yl)-2-eth oxyp r o-
1
p ion ic Acid , 3q. Yield 7.0 g (100%). H NMR (CDCl3) δ 1.15
(t, J ) 7 Hz, 3H), 2.85-3.06 (m, 2H), 3.35 (m, 2H), 3.55 (m,
2H), 3.98 (m, 1H), 4.30 (t, J ) 7 Hz, 2H), 4.70 (t, J ) 7 Hz,
2H), 6.72 (d, J ) 8 Hz, 2H), 7.08 (d, J ) 8 Hz, 2H), 7.25 (m,
2H), 7.47 (m, 4H), 8.07 (m, 2H). 98.6% ee.
(S)-3-(4-(2-(3-Br om o-ca r b a zol-9-yl)et h oxy)p h en yl)-2-
eth oxyp r op ion ic Acid , 3r . Yield 290 mg (87%). 1H NMR
(MeOH) Na-salt δ 1.07 (t, J ) 7 Hz, 3H), 2.74 (m, 1H), 2.88
(m, 1H), 3.20 (m, 1H), 3.57 (m, 1H), 3.74 (m, 1H), 4.35 (t, J )
7 Hz, 2H), 4.75 (t, J ) 7 Hz, 2H), 6.67 (d, J ) 8 Hz, 2H), 7.12
(d, J ) 8 Hz, 2H), 7.22 (t, J ) 8 Hz, 1H), 7.50 (m, 3H), 7.60 (d,
J ) 8 Hz, 1H), 8.06 (d, J ) 8 Hz, 1H), 8.19 (s, 1H). 95.9% ee.
(S)-3-(4-(2-(3,6-Dibr om o-ca r ba zol-9-yl)eth oxy)p h en yl)-
2-eth oxyp r op ion ic Acid , 3s. Yield 75 mg (39%). 1H NMR
(CDCl3) δ 1.17 (t, J ) 7 Hz, 3H), 2.93 (m, 1H), 3.06 (m, 1H),
3.63-3.35 (m, 2H), 4.03 (m, 1H), 4.32 (t, J ) 7 Hz, 2H), 4.67
(t, J ) 7 Hz, 2H), 6.68 (d, J ) 8 Hz, 2H), 7.07 (d, J ) 8 Hz,
2H), 7.38 (d, J ) 8 Hz, 2H), 7.58 (d, J ) 8 Hz, 2H), 8.15 (s,
2H). 94.3% ee.
Lu cifer a se Assa y. Medium including test compound was
aspirated, and 100 µL PBS including 1 mM Mg2+ and Ca2+
was added to each well. The luciferase assay was performed
using the LucLite kit according to the manufacturers instruc-
tions (Packard Instruments). Light emission was quantified
by counting SPC mode on a Packard Instruments top-counter.
To measure â-galactosidase activity, 25 µL of supernatant
from each transfection lysate was transferred to a new micro-
plate. â-Galactosidase assays were performed in the microwell
plates using a kit from Promega and read in a Labsystems
Ascent Multiscan reader. The â-galactosidase data were used

802 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
to normalize (transfection efficiency, cell growth, etc.) the
luciferase data.
LBD-PPARγ protein was crystallized in 0.8 M sodium citrate
and 0.15 M Tris, pH 8.0, at a protein concentration of 5 mg/
mL. The crystal space group and cell parameters obtained are
found in Table X+1, Supporting Information. A crystal storage
solution containing 1.0 M sodium citrate and 0.15 M Tris, pH
8.0, was prepared. Compound 3q was dissolved in 20 µL
storage solution after which crystals were transferred and
soaked for 24 h. Crystals were then transferred over during
10 to 30 s to a cryo-protectant containing the storage solution
mixed with glycerol to a concentration of 20% (v/v). The
crystals were thereafter flash-frozen in a nitrogen gas-stream
cooled to 100 K, their diffraction properties tested and then
stored in liquid nitrogen for subsequent data collection.
Crystallographic data were collected at beamline I711, the
MAX-laboratory, Sweden,51 using a mar345 imaging plate
detector system, and data sets were evaluated by the Xds
program package.52 The structure was subsequently refined
by the Cnx program system,53 in the starting run making use
of the PPARγ coordinates generated by Ebdrup et al. (2002),50
which in turn was based on the coordinates 1PRG of the
Protein Data Bank deposited by Nolte et al.29 Introduction of
3q and corrections to the model according to electron density
maps were made with use of the Quanta program.54 The
program Xplo2d55 was used for creation of ligand Parameter
and Topology files used by the Cnx program. For data
collection, refinement, and model statistics, see Supporting
Information Table X+1. The coordinates of the 3q/PPARγ
structure have been deposited in the Brookhaven Protein Data
Bank, ID 1KNU.
Mod elin g. The Grid calculations were performed with Grid
ver.1856-59 with DPRO equal to 4, DWAT equal to 80, and
EMAX equal to 5. The calculations were performed with 2
planes per ångstrom. All calculations on the complex with 3q
were performed using the structure based on soaked crystals.
The atoms in the ligand were treated as HETATM.
Molecular mechanics calculations were performed with the
MMFF force field34-37 with water as solvation model60 in
MacroModel ver. 7.0.38,39 The Monte Carlo searches were
performed with a systematic pseudo Monte Carlo search.61
P h a r m a cok in etics. The compounds were dosed po and iv
to male SD rats. The compounds were dissolved in 5% ethanol,
10% HPCD, and phosphate buffer pH 7.5-8.0. Blood samples
were collected in EDTA tubes. Each data point represents one
animal.
Plasma samples were analyzed by high turbulence liquid
chromatography (HTLC) combined with tandem mass spec-
trometry (MS/MS).
Com p ou n d s. All compounds were dissolved in DMSO and
diluted 1:1000 upon addition to the cells. Compounds were
tested in quadruple in five concentrations ranging form 0.01
to 30 µM. Cells were treated with compound for 24 h followed
by luciferase assay. Each compound was tested in three
separate experiments. EC50 values were calculated via non-
linear regression using GraphPad PRISM 3.02 (GraphPad
Software, San Diego, CA). The results were expressed as
means ( SD.
In Vivo Mod els. C57BL/KsBom-db/db and lean db/+
male mice, 14 weeks old, were purchased from BOMMICE,
Bomholtga˚rd Breeding & Research centre A/S, Ry, DK. The
mice were housed in groups of six individuals in a room
controlled for temperature (20.0 ( 0.5 °C) and 12/12 h light/
dark cycle (lights on at 6.00 am). The mice had free access to
normal chow and water.
Compounds were dosed as suspensions in 0.2% CMC + 0.4%
Tween-80 in saline. Fresh suspensions were made for 7 days
dosing and kept at +4 °C. The mice (n ) 6 per dose) were dosed
orally by gavage daily at 7.30 am from day 1 to 10. The dose
volume was 10 mL/kg.
Sa m p les a n d An a lyses. A total of 5 µL of A total of non-
fasted full blood was drawn from the tail vein for measuring
baseline glucose. After 7 days of treatment, blood was drawn
from the orbital plexus in nonfasted animals. Blood was
collected in EDTA tubes and centrifuged at 4000g for 10 min
at 4 °C. Plasma was analyzed for triglycerides on a COBAS-
Mirra, and full blood glucose was measured on a EBIO-plus.
Insulin was measured using ELISA.
On day 9 of treatment, an oral glucose tolerance test (OGTT)
was performed on overnight fasted animals. A 5 µL full blood
baseline sample was drawn from the tail vein before glucose
dosing (3 g/kg). After administration of the glucose, 5 µL full
blood samples were drawn at 30, 60, and 120 min from the
tail vein. All samples from OGTT were analyzed for glucose
on an EBIO-plus.
Six weeks old male Sprague-Dawley rats (Charles River,
Germany) were fed on a high cholesterol diet ad libitum (1.25%
cholesterol, 0.5% cholic acid; Research Diets Inc. C13002) for
10 days. From day 7 to day 10, the animals (n ) 6 per dose)
were dosed orally by gavage at 7.30 a.m. Test compounds were
suspended in vehicle (0.2% CMC + 0.4% Tween 80 in sterile
water) and administered in a volume of 2 mL/kg.
Two hours after last dosing on day 10, 2 mL of nonfasting
orbital vein plexus blood was collected and allowed to coagulate
for 30 min on wet ice. Serum was separated by centrifugation
(4000g for 10 min at 4 °C) and stored at -70 °C until analyzed
for triglycerides and cholesterol on a COBAS-Mirra.
Sta tistics. ED50 values were calculated via nonlinear
regression using GraphPad PRISM 3.02 (GraphPad Software,
San Diego, CA). The results were expressed as means ( SEM.
Differences between two groups were evaluated by one way
ANOVA and Dunett’s multiple comparison test *P < 0.05, **P
< 0.01. P values less than 0.05 were considered significant.
Percent reduction was calculated using the equation
Sample preparation included dilution (5% methanol, 1:1),
centrifugation (14500g for 15 min), and aliquotation of mini-
mum 100 µL to 96-well plates. HTLC was performed using a
2300 HTLC System (Cohesive Technologies, Franklin, MA)
consisting of an isocratic pump for sample cleanup and flush
of liquid lines, a binary pump for elution of retained analytes,
and a valve switching module with two Rheodyne six-port
valves. The 2300 HTLC system was operated in single column
mode. The autosampler was a CTC HTS PAL (CTC Analytics,
Zingen, Switzerland). Injection volumes were from 10 to 50
µL. The mass spectrometer was a Sciex API3000 (MDS Sciex,
Toronto, Canada) equipped with a TurboIonSpray and oper-
ated in MRM mode.
((Cv - Ct)/Cv) × 100
and percent normalization using the equation
Ch ir a l An a lysis. The enantiomeric purity of (S)-ethyl
2-ethoxy-3-(4-hydroxyphenyl)propionate (1), 3p , 3q, 3r , and
3s were determined by a chiral capillary electrophoresis (CE)
analytical system developed for the purpose. The CE analyses
were performed on a HP3DCE capillary electrophoresis instru-
ment (Agilent, Waldborn, Germany) equipped with an auto
sampler, a capillary cartridge, a high-voltage power supply, a
diode array detector, electrodes, and a hydrostatic injection
system. The electrophoretic data system was the HP Chem-
station software, and the data were collected with a frequency
of 10 Hz. The CE separations were carried out with untreated
fused-silica capillaries from Agilent with the following dimen-
sions: 48.5 cm total length with 40.0 cm effective length, 50
µm inner diameter, and extended light path with an inner
((Cv - Ct)/(Cv - Cl)) × 100
where Cv was the plasma concentration in the vehicle treated
group, Ct the plasma concentration in the compound treated
group, and Cl the concentration in the lean vehicle treated
group.
Cr ysta llogr a p h y. Ligand binding domain (LBD, amino
acids C165 - Stop) PPARγ was expressed, purified, and
crystallized according to Ebdrup et al., 2002.50 In short, LBD-
PPARγ fused to glutathione S-transferase was expressed in
Escherichia coli, the protein was purified by a GSH-Sepharose
column and thereafter cleaved, and the GST was removed. The

Novel Tricyclic-R-alkyloxyphenylpropionic Acids
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4 803
(11) Vazquez, M.; Merlos, M.; Adzet, T.; Laguna J uan, C. Decreased
susceptibility to copper-induced oxidation of rat-lipoproteins
after fibrate treatment: influence of fatty acid composition. Br.
J . Pharmacol. 1996, 117, 1155-1162.
(12) Aronoff, S.; Rosenblatt, S.; Braithwaite, S.; Egan, J . W.; Mathis-
en, A. L.; Schneider, R. L. Pioglitazone Hydrochloride Mono-
therapy Improves Glycemic Control in the Treatment of Patients
With Type 2 Diabetes. Diabetes Care 2000, 23, 1605-1611.
(13) Mathisen, A. L.; Brockley, M. R. Relationship Between HbA1c
and Weight in the Treatment of Patients with Type 2 Diabetes.
Diabetes 2000, 49 (Suppl. 1), A117.
(14) Willson, T. M.; Cobb, J . E.; Cowan, D. J .; Wiethe, R. W.; Correa,
I. D.; Prakash, S. R.; Beck, K. D.; Moore, L. B.; Kliever, S. A.;
Lehmann, J . M. The Structure-Activity Relationship between
Peroxisome Proliferator-Activated Receptor γ Agonism and the
Antihyperglycemic Activity of Thiazolidinediones. J . Med. Chem.
1996, 39, 665-668.
(15) Brown, P. J .; Winegar, D. A.; Plunket, K. D.; Moore, L. B.; Lewis,
M. C.; Wilson, J . G.; Sundseth, S. S.; Koble, C. S.; Wu, Z.;
Chapman, J . M.; Lehmann, J . M.; Kliever, S. A.; Willson, T. M.
A Ureido-Thioisobutyric Acid (GW9578) Is a Subtype-Selective
PPARR Agonist with Potent Lipid-Lowering Activity. J . Med.
Chem. 1999, 42, 3785-3788.
(16) Henke, B. R.; Blanchard, S. G.; Brackeen, M. F.; Brown, K. K.;
Cobb, J . E.; Collins, J . L.; Harrington, W. W.; Hashim, M. A.;
Hull-Ryde, E. A.; Kaldor, I.; Kliewer, S. A.; Lake, D. H.;
Lesnitzer, L. M.; Lehmann, J . M.; Lenhard, J . M.; Orband-Miller,
L. A.; Miller, J . F.; Mook, R. A., J r.; Noble, S. A.; Oliver, W., J r.;
Parks, D. J .; Plunket, K. D.; Szewczyk, J . R.; Willson, T. M. N-(2-
Benzoylphenyl)-L-tyrosine PPARγ Agonists. 1. Discovery of a
Novel Series of Potent Antihyperglycemic and Antihyperlipi-
demic Agents. J . Med. Chem. 1998, 41, 5020-5036.
(17) Collins, J . L.; Blanchard, S. G.; Boswell, G. E.; Charifson, P. S.;
Cobb, J . E.; Henke, B. R.; Hull-Ryde, E. A.; Kazmierski, W. M.;
Lake, D. H.; Leesnitzer, L. M.; Lehmann, J .; Lenhard, J . M.;
Orband-Miller, L. A.; Gray-Nunez, Y.; Parks, D. J .; Plunkett,
K. D.; Tong, W.-Q. N-(2-Benzoylphenyl)-L-tyrosine PPARγ ago-
nists. 2. Structure-activity relationship and optimization of the
phenyl alkyl ether moiety. J . Med. Chem. 1998, 41, 5037-5054.
(18) Cobb, J . E.; Blanchard, S. G.; Boswell, E. G.; Brown, K. K.;
Charifson, P. S.; Cooper, J . P.; Collins, J . L.; Dezube, M.; Henke,
B. R.; Hull-Ryde, E. A.; Lake, D. H.; Lenhard, J . M.; Oliver, W.,
J r.; Oplinger, J .; Pentti, M.; Parks, D. J .; Plunket, K. D.; Tong,
W.-Q. N-(2-Benzoylphenyl)-L-tyrosine PPARγ agonists. 3. Struc-
ture-activity relationship and optimization of the N-aryl sub-
stituent. J . Med. Chem. 1998, 41, 5055-5069.
(19) Murakami, K.; Tobe, K.; Ide, T.; Mochizuki, T.; Ohashi, M.;
Akanuma, Y.; Yazaki, Y.; Kadowaki, T. A novel insulin sensitizer
acts as a coligand for peroxisome proliferator-activated recep-
tor-R (PPAR-R) and PPAR-γ: effect of PPAR-R activation on
abnormal lipid metabolism in liver of Zucker fatty rats. Diabetes
1998, 47, 1841-1847.
(20) Nomura, M.; Kinoshita, S.; Satoh, H.; Maeda, T.; Murakami, K.;
Tsunoda, M.; Miyachi, H.; Awano, K. (3-Substituted benzyl)-
thiazolidine-2,4-diones as structurally new antihyperglycemic
agents. Bioorg. Med. Chem. Lett. 1999, 9, 533-538.
(21) (a) Brooks, D. A.; Etgen, G. J .; Rito, C. J .; Shuker, A.; Domin-
ianni, S. J .; Warshawsky, A. M.; Ardecky, R.; Paterniti, J . R.;
Tyhonas, J .; Karanewsky, D. S.; Kauffman, R. F.; Broderick, C.
L.; Oldham, B. A.; Montrose-Rafizadeh, C.; Winneroski, L. L.;
Faul, M. M.; McCarthy, J . R. Design and Synthesis of 2-Methyl-
2-{4-[2-(5-methyl-2-aryloxazol-4-yl)ethoxy]phenoxy}propionic Ac-
ids: A New Class of Dual PPARR/γ Agonists. J . Med. Chem.
2001, 44, 2061-2064. (b) Lohray, B. B.; Lohray, V. B.; Bajji, A.
C.; Kalchar, S.; Poondra, R. R.; Padakanti, S.; Chakrabati, R.;
Vikramadithyan, R. K.; Misra, P.; J uluri, S.; Mamidi, N. V. S.
R.; Rajagopalan, R. (-)3-[4-[2-(Phenoxazin-10-yl)ethoxy]phenyl]-
2-ethoxypropanoic Acid [(-)DRF2725]: A Dual PPAR Agonist
with Potent Antihyperglycemic and Lipid Modulating Activity.
J . Med. Chem. 2001, 44, 2675-2678.
(22) Sohda, T.; Mizuno, K.; Kawamatsu, Y. Studies on antidiabetic
agents. VI. Asymmetric transformation of (()-5-[4-(1-methylcy-
clohexylmethoxy)benzyl]-2,4-thiazolidinedione (ciglitazone) with
optically active 1-phenylethylamines. Chem. Pharm. Bull. 1984,
32, 4460-4465.
(23) Parks, D. J .; Tomkinson, N. C. O.; Villeneuve, M. S.; Blanchard,
S. G.; Willson, T. M. Differential activity of rosiglitazone
enantiomers at PPARγ. Bioorg. Med. Chem. Lett. 1998, 8, 3657-
3658.
diameter of 150 µm at the detector window. The electrolyte
was prepared by dissolving 3.0% (w/v) sulfobuthyl ether-â-
cyclodextrin (Advasep 4, Cydex, Inc., Overland Park, KS)
and 0.50% (w/v) dimethyl-â-cyclodextrin (Agilent, Waldborn,
Germany) both in 50 mM borate buffer pH 9.3 (Agilent)
followed by filtering through a 0.45 µm polypropylene filter.
To this solution was added 5% (v/v) acetonitrile to give the
final electrolyte. The electrophoresis was carried out in normal
polarity mode.
The electrophoretic conditions were as follows: voltage, 14
kV; current, 60 µA; capillary temperature controlled at 30 °C;
injection was 25 mbar for 4.0 s; detection, UV at 231 nm with
reference of 350 nm. The sample concentration was 2.5 mg/
mL in 10/90 acetonitrile/5 mM borate buffer pH 9.3. The
capillary was conditioned with 0.1 N NaOH for 20 min daily
and flushed with electrolyte for 1.5 min between each run.
Ack n ow led gm en t. The authors acknowledge the
researchers at Dr. Reddy’s Research Foundation, India,
for their inspiration to this work. The technical as-
sistance from Rikke Burgdorf, Hanne Nielsen, Lene
Priskorn, Helle Bach, Anette Heerwagen, Anette Zen-
eca, Kirsten M. Klausen, Kent Pedersen, Otto Larsson,
Sanne Kold, Per Klifforth, Bjørn Metzler, and Alice
Ravn is highly appreciated.
Su p p or tin g In for m a tion Ava ila ble: Table of hydrogen
bonds and van der Waals interactions between 3q and the
amino acids in the PPARγ receptor protein (Table X). Table
for data collection, refinement, and model statistics (Table
X+1). Illustration of a capillary electrophoresis chromatogram
exemplified by 3p . This material is available free of charge
via the Internet at http://pubs.asc.org.
Refer en ces
(1) Porte, D., J r.; Schwartz, M. W. Diabetes complications: Why is
Glucose Potentially Toxic? Science 1996, 27, 699-700.
(2) Staels, B.; Dallongeville, J .; Auwerx, J .; Schoonjans, K.; Leit-
ersdorf, E.; Fruchart, J .-C. Mechanism of action of fibrates on
lipid and lipoprotein metabolism. Circulation 1998, 98, 2088-
2093.
(3) Oliver, M. F.; Heady, J . A.; Morris, J . N.; Cooper, J . WHO
cooperative trial on the primary prevention of ischemic heart
disease with clofibrate to lower serum cholesterol: final mortal-
ity follow-up. Lancet 1984, 2, 600-604.
(4) Ericsson, C. G.; Nilsson, J .; Grip, L.; Svane, B.; Hamsten, A.
Effect of bezafibrate treatment over five years on coronary
plaques causing 20% to 50% diameter narrowing (the bezafibrate
coronary atherosclerosis intervention trial [BECAIT]). Am. J .
Cardiol. 1997, 80, 1125-1129.
(5) Rutolo, G.; Ericsson, C.-G.; Tettamanti, C.; Karpe, F.; Grip, L.;
Svane, B.; Nilsson, J .; De Faire, U.; Hamsten, A. Treatment
effects on serum lipoprotein lipids, apolipoproteins and low-
density lipoprotein particle size and relationships of lipoprotein
variables to progression of coronary artery disease in the
bezafibrate coronary atherosclerosis intervention trial (BECAIT).
J . Am. Coll. Cardiol. 1998, 32, 1648-1656.
(6) Rubins, H. B.; Robins, S. J .; Collins, D.; Fye, C. L.; Anderson, J .
W.; Elam, M. B.; Faas, F. H.; Linares, E.; Schaefer, E. J .;
Schectman, G.; Wilt, T. J .; Wittes, J . Gemfibrozil for the
secondary prevention of coronary heart disease in men with low
levels of high-density lipoprotein cholesterol. N. Engl. J . Med.
1999, 341, 410-418.
(7) Steiner, G.; Hamsten, A.; Hosking, J .; Stewart, D.; McLaughlin,
P.; Gladstone, P.; Sole, M.; Syvanne, M. Effect of fenofibrate on
progression of coronary-artery disease in type 2 diabetes: the
Diabetes Atherosclerosis Intervention Study,
a randomised
study. Lancet 2001, 357, 905-910.
(8) Kobayashi, M.; Shigeta, Y.; Hirata, Y.; Omori, Y.; Sakamoto, N.;
Nambu, S.; Baba, S. Improvement of glucose tolerance in
NIDDM by clofibrate. Randomized double-blind study. Diabetes
Care 1988, 11, 495-499.
(9) J ones, I. R.; Swai, A.; Taylor, R.; Miller, M.; Laker, M. F.; Alberti,
K. G. Lowering of plasma glucose concentrations with bezafi-
brate in patients with moderately controlled NIDDM. Diabetes
Care 1990, 13, 855-863.
(10) Inoue, I.; Takahashi, K.; Katayama, S.; Akabane, S.; Negishi,
K.; Suzuki, M.; Ishii, J .; Kawazu, S. Improvement of glucose
tolerance by bezafibrate in nonobese patients with hyperlipi-
demia and impaired glucose tolerance. Diabetes Res. Clin. Pract.
1994, 25, 199-205.
(24) Buckle, D. R.; Cantello, B. C. C.; Cawthorne, M. A.; Coyle, P. J .;
Dean, D. K.; Faller, A.; Haigh, D.; Hindley, R. M.; J efcott, L. J .;
Lister, C. A.; Pinto, I. L.; Rami, H. K.; Smith, D. G.; Smith, S.
A. Nonthiazolidinedione antihyperglycemic agents. 1: R-Het-
eroatom substituted â-phenylpropanoic acids. Bioorg. Med.
Chem. Lett. 1996, 6, 2121-2126.

804 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 4
Sauerberg et al.
(25) Buckle, D. R.; Cantello, B. C. C.; Cawthorne, M. A.; Coyle, P. J .;
Dean, D. K.; Faller, A.; Haigh, D.; Hindley, R. M.; J efcott, L. J .;
Lister, C. A.; Pinto, I. L.; Rami, H. K.; Smith, D. G.; Smith, S.
A. Nonthiazolidinedione antihyperglycemic agents. 2: R-Carbon
substituted â-phenylpropanoic acids. Bioorg. Med. Chem. Lett.
1996, 6, 2127-2130.
(41) Krause, B. R.; Princen, H. M. G.; Lack of predictability of
classical animal models for hypolididemic activity: a good time
for the mice? Atherosclerosis 1998, 140, 15-24.
(42) Polivka, Z.; Holubek, J .; Svatek, E.; Dlabac, A.; Pucek, D.; Sedivy,
Z.; Protiva, M. 3-(4-methylpiperazino)dibenzo[b,f]-1,2,4-trazolo-
[4,3-d]-1,4-thiazepine and its 6-chloro and 12-chloro derivatives;
synthesis and pharmacology. Coll. Czech. Chem. Commun. 1983,
48, 1465-1476.
(43) Goldberg, A. A.; Wragg, A. H. Spasmolytics Derived from
Xanthen. J . Chem. Soc. 1957, 4823-4829.
(44) Kuhn, R.; Breyer, U. Alkylation of benzene on a Al2O3-column.
Ann. Chem. 1963, 661, 173-180.
(26) Willson, T. M.; Brown, P. J .; Sternbach, D. D.; Henke, B. R. The
PPARs: From Orphan Receptors to drug Discovery. J . Med.
Chem. 2000, 43, 527-550.
(27) Hulin, B.; Newton, L. S.; Lewis, D. M.; Genereux, P. E.; Gibbs,
E. M.; Clark, D. A. Hypoglycemic Activity of a Series of
R-Alkylthio and R-Alkoxy Carboxylic Acids Related to Ciglita-
zone. J . Med. Chem. 1996, 39, 3897-3907.
(28) Haigh, D.; Birrell, H. C.; Cantello, B. C. C.; Hindley, R. M.;
Ramaswamy, A.; Rami, H. K.; Stevens, N. C. Nonthiazolidinedi-
one antihyperglycaemic agents. Part 4: Synthesis of (()-, R-(+)-
and (S)-(-)-enantiomers of 2-oxy-3-arylpropionic acids. Tetra-
hedron: Asymmetry 1999, 10, 1335-1351.
(29) Nolte, R. T.; Wisely, G. B.; Westin, S.; Cobb, J . E.; Lambert, M.
H.; Kurokawa, R.; Rosenfeld, M. G.; Willson, T. M.; Glass, C.
K.; Milburn, M. V. Ligand binding and co-activator assembly of
the peroxisome proliferator-activated receptor-gamma. Nature
1998, 395, 137-143.
(30) Haigh, D.; Allen, G.; Birrell, H. C.; Buckle, D. R.; Cantello,
B. C. C.; Eggleston, D. S.; Haltiwanger, R. C.; Holder, J . C.;
Lister, C. A.; Pinto, I. L.; Rami, H. K.; Sime, J . T.; Smith, S. A.;
Sweeney, J. D. Nonthiazolidinedione Antihyperglycaemic Agents.
Part 3: The Effects of Stereochemistry on the Potency of
R-Methoxy-â-phenylpropionic Acids. Bioorg. Med. Chem. 1999,
7, 821-830.
(31) Loray, B. B.; Bhushan, V.; Rao, B. P.; Madhavan, G. R.; Murali,
N.; Rao, K. N.; Reddy, A. K.; Reddy, P. G.; Chakrabarti, R.;
Vikramadithyan, R. K.; Rajagopalan, R.; Mamidi, R. N. V. S.;
J ajoo, H. K.; Subramaniam, S. Novel Euglycemic and Hyper-
lipidemic Agents. 1. J . Med Chem. 1998, 41, 1619-1630.
(32) Oberfield, J . L.; Collins, J . L.; Holmes, C. P.; Goreham, D. M.;
Cooper, J . P.; Cobb, J . E.; Lenhard, J . M.; Hull-Ryde, E. A.; Mohr,
C. P.; Blanchard, S. G.; Parks, D. J .; Moore, L. B.; Lehmann, J .
M.; Plunket, K.; Miller, A. B.; Milburn, M. V.; Kliewer, S. A.;
Willson, T. M. A peroxisome proliferator-activated receptor
gamma ligand inhibits adipocyte differentiation. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 6102-6106.
(45) Perumattam, J .; Shao, C.; Confer, W. L. Studies on the Alkyla-
tion and Chlorination of Fluorenes: Preparation of 9-(2-hydroxy-
ethyl)fluorene and 2,7-Dichloro-9-(2-hydroxyethyl)fluorene. Syn-
thesis 1994, 11, 1181-1184.
(46) Hoffsommer, R. D.; Taub, D.; Wendler, N. L. The Homoallylic
Rearangement in the Synthesis of Amitriptyline and Related
Systems. J . Org. Chem. 1962, 27, 4134-4137.
(47) Hoffsommer, R. D.; Taub, D.; Wendler, N. L. Synthesis of a
Cyclopropyl carbinol in the Amitripyline Series. J . Med. Chem.
1964, 7, 392-393.
(48) Sadowski, I.; Bell, B.; Broad, P.; Hollis, M. GAL4 fusion vectors
for expression in yeast and mammalian cells. Gene 1992, 118,
137-141.
(49) Webster, N.; J in, J . R.; Green, S.; Hollis, M.; Chambon, P. The
Yeast UASG Is a Transcriptional Enhancer In Human HeLa
Cells In the Presence of the GAL4 trans-activator. Cell 1988,
52, 169-178.
(50) Ebdrup, S. et al. manuscript in progress.
(51) Cerenius, Y.; Ståhl, K.; Svensson, L. A.; Ursby, T.; Oskarsson,
Å.; Albertsson, J .; Liljas, A. The crystallography beamline I711
at MAX II. J . Synchrotron Radiat. 2000, 7, 203-208.
(52) Kabsch, W. Automatic processing of rotation diffraction data
from crystals of initially unknown symmetry and cell constants.
J . Appl. Crystallogr. 1993, 26, 795-800.
(53) Brunger, A. T.; Krukowski, A.; Erickson, J . W. Slow-cooling
protocols for crystallographic refinement by simulated annealing.
Acta Crystallogr. 1990, A46, 585-593.
(33) Xu, H. E.; Lambert, M. H.; Montana, V. G.; Parks, D. J .;
Blanchard, S. G.; Brown, P. J .; Sternbach, D. D.; Lehmann, J .
M.; Wisely, G. B.; Willson, T. M.; Kliewer, S. A.; Milburn, M. V.
Molecular recognition of fatty acids by peroxisome proliferator-
activated receptors. Mol. Cell. 1999, 3, 397-403.
(54) Oldfield, T. J .; Hubbard, R. E. Analysis of c-alpha geometry in
protein structures. Proteins 1994, 18, 324-337.
(55) Kleywegt, G. J .; J ones, T. A. Databases in protein Crystal-
lography. Acta Crystallogr. 1998, D54, 1119-1131.
(56) Goodford, P. J . A computational procedure for determining
energetically favorable binding sites on biologically important
macromolecules J . Med. Chem. 1985, 28, 849-857.
(57) Boobbyer, D. N.; Goodford, P. J .; McWhinnie, P. M.; Wade, R.
C. New hydrogen-bond potentials for use in determining ener-
getically favorable binding sites on molecules of known structure.
J . Med. Chem. 1989, 32, 1083-1094.
(58) Wade, R. C.; Goodford, P. J . Further development of hydrogen
bond functions for use in determining energetically favorable
binding sites on molecules of known structure. 2. Ligand probe
groups with the ability to form more than two hydrogen bonds.
J . Med. Chem. 1993, 36, 148-156.
(34) Halgren, T. A. Merck molecular-force field. 1. basis, form, scope,
parametrization, and performance of mmff94. J . Comput. Chem.
1996, 17, 490-519.
(35) Halgren, T. A. Merck molecular-force field. 2. mmff94 van der
Waals and electrostatic parameters for intermolecular interac-
tions. J . Comput. Chem. 1996, 17, 520-552.
(36) Halgren, T. A. Merck molecular-force field. 3. molecular geom-
etries and vibrational frequencies for mmff94. J . Comput. Chem.
1996, 17, 553-586.
(37) Halgren, T. A.; Nachbar, R. B. Merck molecular-force field. 4.
conformational energies and geometries for mmff94. J . Comput.
Chem. 1996, 17, 587-615.
(38) Mohamadi, F.; Richards, N. G. J .; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T. and Still,
W. C. MacroModel-An Integrated Software System for Modeling
Organic and Bioorganic Molecules using Molecular Mechanics.
J . Comput. Chem. 1990, 11, 440-467.
(59) Molecular Discovery Ltd., 20a Berkeley Street, Mayfair, London
W1X 5AE, U.K.
(60) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical Treatment of Solvation for Molecular Mechanics
and Dynamics. J . Am. Chem. Soc. 1990, 112, 6127-6129.
(61) Goodman, J . M.; Still, W. C. An Unbounded Systematic Search
of Conformational Space J . Comput. Chem. 1991, 12, 1110-1117.
(39) Schro¨dinger, Inc., 1500 S. W. First Avenue, Suite 1180, Portland,
OR 97201-5815.
(40) Grunberger, G.; Weston, W. M.; Patwardhan, R.; Rappaport, E.
B. Rosiglitazone Once or Twice Daily Improves Glycemic Control
in Patients with Type 2 Diabetes (T2D). Diabetes 1999, 48
(Suppl. 1), A102.
J M010964G