Synthesis and biological and structural characterization of the dual-acting peroxisome proliferator-activated receptor α/γ agonist ragaglitazar

Article abstract of DOI:10.1021/jm021027r

A new and improved synthesis of the peroxisome proliferator-activated receptor (PPAR) agonist ragaglitazar applicable for large-scale preparation has been developed. The convergent synthetic procedure was based on a novel enzymatic kinetic resolution step

Full text of DOI:10.1021/jm021027r

1306
J . Med. Chem. 2003, 46, 1306-1317
Syn th esis a n d Biologica l a n d Str u ctu r a l Ch a r a cter iza tion of th e Du a l-Actin g
P er oxisom e P r olifer a tor -Activa ted Recep tor r/γ Agon ist Ra ga glita za r
Søren Ebdrup,* Ingrid Pettersson, Hanne B. Rasmussen, Heinz-J osef Deussen, Anette Frost J ensen,
Steen B. Mortensen, J an Fleckner, Lone Pridal, Lars Nygaard, and Per Sauerberg
Novo Nordisk A/ S, Novo Nordisk Park, 2760 Ma˚løv, Denmark
Received August 20, 2002
A new and improved synthesis of the peroxisome proliferator-activated receptor (PPAR) agonist
ragaglitazar applicable for large-scale preparation has been developed. The convergent synthetic
procedure was based on a novel enzymatic kinetic resolution step. The conformation of
ragaglitazar bound to the hPPARγ receptor was quite different compared to the single-crystal
structures of the L-arginine salt of ragaglitazar. In particular, the phenoxazine ring system
had varying orientations. Ragaglitazar had high affinity for the hPPARR and -γ receptors with
IC50 values of 0.98 and 0.092 µM, respectively. The lack of hPPARδ activity could be explained
by the absence of binding in the tail-up pocket in the hPPARδ receptor, in contrast to the
hPPARδ agonist GW2433, which was able to bind in both the tail-up and tail-down pockets of
the receptor.
In tr od u ction
EC50 ) 0.8 µM),15 and LY465608 (6; R, EC50 ) 0.15 µM;
γ, EC50 ) 0.88 µM),16 Figure 1, have recently been
described. Ragaglitazar is a selective, potent, and ef-
ficacious agonist of the human PPARR (hPPARR) and
PPARγ (hPPARγ) receptors with an R/γ activation ratio
of 5.6.11 Clinical phase 2 studies showed significant
lowering of plasma triglycerides, total cholesterol, blood
glucose, and HbA1C as well as increasing levels of HDL
cholesterol at pharmacological relevant doses.17,18
In this paper, an improved and scalable synthetic
procedure, which includes an enantioselective enzymatic
resolution step, is presented for ragaglitazar. The
absolute stereochemistry of ragaglitazar, as well as the
crystal structure of the hPPARγ-ligand binding domain
(LBD)-protein complex, is also described. Finally,
modeling approaches explaining the hPPARR, -γ, and
-δ profiles of ragaglitazar are provided.
Type 2 diabetes is a chronic multifactorial metabolic
disease characterized by insulin resistance, hypergly-
cemia, and hyperinsulinimia, leading to impaired secre-
tion of insulin in the later stages. The disease is often
associated with obesity, dyslipidemia, and hypertension,
leading to increased cardiovascular risks.1 Owing to the
forecasted epidemic in type 2 diabetes, the increasing
financial and social costs, and the complicated pathology
of the disease, new therapies are needed that address
both the insulin resistance and dyslipidemic components
of the disease.2-4 Different types of PPAR (peroxisome
proliferator-activated receptor) agonists have been shown
to have beneficial effects on the described characteristics
of type 2 diabetes.5 Fibrates (PPARR agonists, e.g.,
fenofibrate and clofibrate) primarily decrease serum
triglyceride levels and increase HDL cholesterol (HDLc)
levels, but they also improve glucose tolerance in type
2 diabetic patients.5,6 Furthermore, fibrates have been
reported to reduce weight gain in rodents without effects
on food intake.7 Insulin sensitizers (PPARγ agonists,
e.g., pioglitazone (1) and rosiglitazone (2) (Figure 1)) also
have a range of clinical effects including improvement
of insulin sensitivity and glucose tolerance and lowering
of blood glucose levels in type 2 diabetic patients.8,9
New dual-acting PPARR and -γ agonists, designed to
combine the beneficial effects seen with insulin sensitiz-
ers and fibrates, have received increased attention.
Further, the dual PPARR and -γ agonists might also
reduce the weight gain associated with adipogenesis
resulting from PPARγ activation through the simulta-
neous stimulation of lipid oxidation and decreased
adiposity observed after PPARR activation. Different
dual-acting agonists, e.g., ragaglitazar (NNC 61-0029,
(-)DRF 2725, 3; R, EC50 ) 3.2 µM; γ, EC50 ) 0.6
µM),10-12 tesaglitazar (AZ 242, 4; R, EC50 ) 1.2 µM; γ,
EC50 ) 1.3 µM),13,14 KRP 297 (5; R, EC50 ) 1.0 µM; γ,
Ch em istr y
A novel and improved synthesis of (S)-2-ethoxy-3-(4-
[2-(phenoxazine-10-yl)ethoxy]phenyl)propanoic acid (S-
3, ragaglitazar) applicable for large-scale production has
been developed. The convergent synthetic procedure is
based on the condensation of the two key intermediates
2-phenoxazin-10-yl-ethyl methanesulfonate (12) and (S)-
2-propyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate (S-18)
(Scheme 1). The earlier described12 linear synthesis of
ragaglitazar is based on a six-step procedure involving
five chromatographic separation steps giving an overall
yield of 7% and a chiral purity of 94.6% ee. The optical
resolution was performed at the last step in the syn-
thesis as racemic ragaglitazar was converted to a
diastereomeric (S)-(+)-2-phenylglycinolamide, which was
chromatographed and hydrolyzed to give optical pure
S-3.
In the new and improved synthesis of ragaglitazar,
enantiomerically pure S-18 was alkylated with the
mesylate 12 to give (S)-2-propyl 2-ethoxy-3-(4-[2-(phe-
noxazine-10-yl)ethoxy]phenyl)propanoate (S-19).11 Al-
kaline hydrolysis of crude S-19 in 2-propanol with
* To whom correspondence should be addressed. Fax: +45 44663939.
Phone: + 45 44434830. E-mail: sebd@novonordisk.com.
10.1021/jm021027r CCC: $25.00 © 2003 American Chemical Society
Published on Web 03/01/2003

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1307
F igu r e 1. Structures of PPAR agonists: (1) pioglitazone; (2) rosiglitazone; (3) ragaglitazar; (4) tesaglitazar (AZ 242); (5) KRP
297; (6) LY465608; (7) 3q;11 (8) GW2433; (9) GW501516,
sodium hydroxide yielded (S)-2-ethoxy-3-(4-[2-(phenox-
azine-10-yl)ethoxy]phenyl)propanoic acid (S-3, ragagli-
tazar) without significant racemization in 81% yield
starting from the two key intermediates 12 and S-18.
tion with n-BuLi/ethylenoxide in tert-butyl methyl ether
to give 2-phenoxazine-10-yl-ethanol (11).23,24 The alcohol
11 was then mesylated with methanesulfonyl chloride
to give 1225 in 70% yield over the two steps.
Different approaches to the key intermediate S-18
have been considered, but because of access to a collec-
tion of hydrolases, we decided to investigate if S-18
could be prepared by enzymatic resolution from racemic
ethyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate (rac-16).
More than 80 different hydrolases were initially screened
in order to find a highly enantioselective enzyme for this
substrate.19 It was found that racemic rac-16 could be
resolved by kinetic resolution with the commercially
available enzyme preparation Pectinex Ultra-SP-L from
Novozymes A/S. Compound S-17 was obtained, and the
enantiomeric ratio for the process was found to be E >
200. The enzymatic process for the preparation of S-17
has meanwhile been further optimized and scaled and
was recently performed on a 44 kg scale.20 rac-16 was
prepared from commercially available 4-benzyloxybenz-
aldehyde (13), which was reacted with triethyl 2-ethoxy-
phosphonoacetate (14)21 in a Horner-Emmons-Wads-
worth22 reaction to give ethyl 3-(4-benzyloxyphenyl)-2-
ethoxyacrylate (15) as an E/Z mixture. 15 was then
hydrogenated to ethyl 2-ethoxy-3-(4-hydroxyphenyl)-
propanoate (rac-16).
Thus, the overall yield starting from phenoxazine 10
could be improved to 56% with an enantiomeric excess
of 98.2%.
To identify a suitable salt of ragaglitazar for phar-
maceutical development, various bases that are com-
monly used in pharmaceutical products (NaOH, KOH,
Mg(OH)2, L-arginine, L-lysine, and N-methyl-D-glucam-
ine (meglumine)) were tested for salt formation with the
carboxylic acid (pKa ) 3.7) of ragaglitazar. The precipi-
tates formed with the bases were subjected to analysis
by powder X-ray diffraction, differential scanning cal-
orimetry, and thermogravimetric analysis.26
The L-arginine salt of ragaglitazar (S-3, arginine) that
was crystallized from ethanol or 2-propanol showed good
crystallinity and a melting point suitable for tablet
formulation. Moisture sorption studies showed ragagli-
tazar, L-arginine to be nonhygroscopic. Stability studies
at different storage conditions, including elevated tem-
perature and humidity, demonstrated the integrity of
the crystal form and satisfactory chemical stability.
On the basis of these results, the L-arginine salt of
ragaglitazar was selected as the best candidate for drug
development and ragaglitazar was converted to this salt.
The other key intermediate 12 was prepared from
commercially available phenoxazine (10) by N-alkyla-

1308 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8
Ebdrup et al.
Sch em e 1. Synthesis of Ragaglitazar, L-Arginine [(S)-3, L-Arginine]
Str u ctu r a l Ch a r a cter iza tion
[3H]-NNC 61-4655 in the hPPARγ and -R assays,
respectively, were further developments of the SPA
assay described by Nichols.29 The binding profile of
ragaglitazar was compared to the profiles of two mar-
keted insulin sensitizers rosiglitazone and pioglitazone,
to the structurally related tesaglitazar, and to the
(R)-enantiomer of ragaglitazar. Ragaglitazar had the
highest affinity for both the hPPARR (IC50 ) 0.98 µM)
and the hPPARγ receptors (IC50 ) 0.092 µM) of the five
compounds tested. The hPPARγ affinity was consider-
able higher than the hPPARγ-selective ligands piogli-
tazone (7.4 µM) and rosiglitazone (0.44 µM), and neither
pioglitazone nor rosiglitazone (up to 10 µM) had affinity
for the hPPARR receptors (Table 1). Tesaglitazar had a
binding profile similar to that of ragaglitazar but with
lower affinity (Table 1). The R isomer of ragaglitazar
had less than 10 µM affinity for both PPAR receptors.
The crystal structure of the dimethyl sulfoxide solvate
of ragaglitazar, L-arginine was determined by single-
crystal X-ray crystallography.27,28 The absolute config-
uration of ragaglitazar was independently determined
to be the (S)-enantiomer by refining the crystal struc-
ture using the anomalous X-ray dispersion from sulfur.
This finding is in agreement with the configuration of
the active enantiomer of related ligands (e.g., tesagli-
tazar).14 The crystal structure contained two crystallo-
graphically independent ragaglitazar molecules with
different conformations, two identical L-arginine, and
two DMSO molecules in the asymmetric unit (Figure
2). The two ragaglitazar molecules (A and B of Figure
3) had almost identical orientations of the phenyleth-
oxypropanoic acid groups. This moiety could be aligned
with a root-mean-square (rms) deviation of 0.12 Å; i.e.,
the moieties are similar. The conformational difference
of the two ragaglitazar molecules could be described as
a rotation of the ethoxy-linked phenoxazine group from
this moiety. The torsion angles describing this rotation
differ by 157(3)° between molecules A and B.
Resu lts a n d Discu ssion
In in vitro hPPARR, -γ, and -δ transactivation assays,
ragaglitazar was shown to be a potent and efficacious
hPPARR and hPPARγ agonist with no hPPARδ activity
(R, EC50 ) 3.2 µM, 97%; γ, EC50 ) 0.6 µM, 117%; δ,
7%).11 Animal studies in db/db mice and high-cholesterol-
fed rats further showed that ragaglitazar had excellent
blood glucose lowering and insulin sensitizing as well
as plasma lipid lowering activities, confirming PPARR
Bin d in g Stu d ies
Ragaglitazar was further characterized by determin-
ing the binding affinity to hPPARR and -γ receptors.
The ligand binding assays, using [3H]-ragaglitazar and

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1309
F igu r e 2. Packing diagram27 in stereo of the ragaglitazar, L-arginine DMSO solvate, showing the unit cell determined by single-
crystal X-ray crystallography. The crystallographic c axis is horizontal, the a axis is inclined to this in the plane of the paper,
while the b axis is perpendicular to the paper plane. Chemical bonds are indicated by full lines, and hydrogen bonds are indicated
by stipulated lines.
F igu r e 3. Drawing27 of the two crystallographically independent ragaglitazar molecules A (left) and B (right), determined by
single-crystal X-ray crystallography. Thermal ellipsoids are shown at the 50% probability level for non-hydrogen atoms. Hydrogen
atoms are shown as small spheres.
Ta ble 1. Receptor Binding Affinities to HPPARR and -γ
animal data showing significant lowering of HbA1c and
plasma lipids.11,17 Thus, ragaglitazar belongs to a new
class of dual PPARR and -γ agonists intended to restore
insulin sensitivity and correct dyslipidemia.
Receptors of Selected PPAR Agonists
PPARRa,c
IC50, µM
PPARγb,c
IC50, µM
compound
pioglitazone (1)
rosiglitazone (2)
ragaglitazar (S) (3)
ragaglitazar (R)
tesaglitazar (4)
>10
7.4 ( 0.3
To further understand the interaction with the
hPPARγ receptor, crystals of the hPPARγ-LBD-
ragaglitazar complex were prepared by co-crystallization
and by soaking. Despite the very different crystallization
conditions for the co-crystallized and soaked crystals,
the crystals belong to the same space group with similar
cell parameters and overall similar structures; only
minor differences in the binding of the ligand was
detected. The asymmetric unit contained two hPPARγ-
LBD proteins and one ragaglitazar molecule. The overall
structure of hPPARγ-LBD was very similar to the apo
structures reported by Nolte et al.32 and Uppenberg et
>10
0.44 ( 0.04
0.092 ( 0.003
>10
0.98 ( 0.04
>10
3.8 ( 0.3
0.35 ( 0.04
a
b
[3H]-NNC 0061-4655 used as radioligand. [3H]-Ragaglitazar
used as the radioligand. c Results are expressed as the mean (
SEM (n ) 5-6).
and PPARγ activity in vivo.10 Furthermore, clinical
phase 1 data with ragaglitazar showed favorable phar-
macokinetics in both healthy subjects and type 2
diabetic patients,30,31 and phase 2 data confirmed the

1310 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8
Ebdrup et al.
F igu r e 5. Alignment of ragaglitazar (green carbon atoms) and
compound 7 (magenta carbon atoms) in complex with the
hPPARγ receptor.
F igu r e 4. Drawing of the crystal structure of soaked hPPARγ-
LBD in complex with ragaglitazar. The final 2Fo - Fc electron
density map covering the ligand is shown at the 1σ level (blue)
and 2σ level (orange). The ligand is shown in bold stick, and
the coordinating residues of the protein are shown in thin stick.
lized and the soaked structures. Sauerberg et al.11 also
reported weak ed definition for the carbazole part of 7
(Figure 1).
The structural differences in the binding pocket of the
nonbinding receptor protein (inactive) relative to the
binding receptor protein (active) were detected in the
hydrophilic part coordinating the carboxylic acid moiety
of the ligand. Superimposing the active and the inactive
receptor proteins showed that His323 and His449 were
situated at similar positions while Tyr473 in the inactive
receptor was not in place for interaction with the ligand.
However, even more striking was that the C-terminal
of the inactive receptor was positioned at approximately
the same position as the carboxylic acid moiety of the
ligand in the active receptor. Not only was the inactive
receptor protein lacking a coordinating residue Tyr473
in its binding pocket but also the binding pocket was
partly sterically blocked by its own C-terminal residue.
The conformation of the nonbinding PPARγ-LBD mono-
mer might be an effect of crystal packing and not
biologically relevant. However, it also shows the receptor
to be dynamic in the C-terminal part. Furthermore, the
location of the C-terminal in the nonbinding monomer
of the receptor shows a strong preference for an acidic
group at this position.
To detect hard bound water molecules, which interact
with the ligand, and to understand their importance for
ligand binding and/or activation of the receptor, Grid
calculations with a water probe were performed. In
particular, two water molecules (water7 and -14 located
in the soaked complex) were found to be of importance
(Figure 5). Calculations with a water probe, with the
ligand included and water14 excluded, gave an energy
minimum close to the position where water14 is located,
with an interaction energy of -9.1 kcal/mol. This energy
minimum was due to hydrogen bonds to the oxygen
atom in the phenoxazine ring system and to water7.
al.33 and was similar to the complex structures reported
by Oberfield et al.,34 Cronet et al.,14 and Sauerberg et
al.,11 i.e., a homodimer containing one receptor protein
in the binding conformation (active) holding a ragagli-
tazar molecule in the binding site and one receptor
protein in the nonbinding conformation (inactive) with
an empty binding site.
A close-up examination of the binding pocket, com-
paring the ragaglitazar complex to the deposited
hPPARγ-LBD apo structures,32,33 and the complex
structures11,14,34 revealed a binding pocket where certain
side chains were able to adopt different conformations.
In particular, Phe363, Tyr473, and Arg288 showed
variations.
The ligands in the two complex structures reported
here (soaked and co-crystallized) were placed overall at
the same position, with the same relative orientation
with respect to the hydrophilic and hydrophobic ends
of the ligands (Figure 4). The phenoxazine part, situated
in a large hydrophobic cavity of the binding pocket, was
oriented slightly differently in the structures based on
the co-crystallized and the soaked crystals, indicating
that this part of the binding pocket could be explored
even further to fully understand its numerous possibili-
ties for ligand accommodation. The different orienta-
tions might be explained by the very different crystal-
lization conditions for the co-crystallized specimen
relative to the soaked one, both in terms of pH and ionic
strength. The possibilities for various orientations in the
hydrophobic cavity of the binding pocket were further
supported by weak but significant electron density (ed)
maps of the phenoxazine moiety in both the co-crystal-

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1311
Ta ble 2. Alignment of Amino Acids in the Hydrophobic Part of
the hPPAR Ligand Binding Domain
PPARR
PPARγ
PPARδ
Ile272
Ile281
Val281
Ile277
Leu356
Leu255
Phe352
Ile249
Ala268
Leu347
Leu247
Phe343
Ile241
Val277
Leu356
Leu255
Phe352
Ile249
Ile339
Met348
Ile341
Val348
Val341
Leu339
Phe368
Ile364
Val334
Ile333
Met228
Leu330
Met329
Phe226
Val332
Met330
Phe359
Met355
Met325
Val324
Met220
Leu321
Met320
Phe218
Val339
Phe368
Met364
Met334
Leu333
Leu228
Leu330
Met329
Phe226
F igu r e 6. Superimposition based on the phenyl ring of
ragaglitazar from four crystal structures: (magenta) confor-
mation A from ragaglitazar, L-arginine single crystal; (cyan)
conformation B from ragaglitazar, L-arginine single crystal;
(green) conformation from co-crystallized protein structure;
(gray) conformation from soaked protein structure.
protein ligands for these torsion angles were negligible,
but the angles of both ligands differed significantly from
the torsion angles of the ragaglitazar, L-arginine salt
molecules (Figure 6). The most important difference
between the conformations of the two protein ligands
was described by the torsion angle extending from the
phenoxazine ring to the ethoxy linker N(1x)-C(13x)-
C(14x)-O(2x) (Figure 6). This angle was -169° for the
soaked-in ligand and -131° for the co-crystallized
ligand, a significant difference of 38°. The correspond-
ing values for molecules A and B in the salt were
-177.1(9)° and 175.5(9)°, which were both quite similar
to the soaked-in ligand. The only other major confor-
mational difference between the ragaglitazar, L-arginine
salt molecules and the protein ligands was the torsion
angle describing the twist of the terminal ethoxy group
from the chiral center, C(22x)-O(5x)-C(24x)-C(25x)
(Figure 6). This angle was -78.6(10)° and -77.4(10)°
for ragaglitazar, L-arginine salt molecules A and B,
respectively, whereas it was 180° and 179° for the
soaked-in and the co-crystallized ligands, respectively,
i.e., a difference in orientation of approximately 100°.
Owing to the conformational differences described above,
the relative orientation of the phenyl ring with respect
to the phenoxazine ring system differed for the ragagli-
tazar, L-arginine salt molecules and the protein ligands.
The phenyl ring plane was almost parallel to the N(1x)-
O(1x) axis in the ragaglitazar, L-arginine salt molecules,
whereas this plane was almost perpendicular to the
N(1x)-O(1x) axis in the protein ligands (Figure 6).
Selected torsion angles are listed in Table S1 (Support-
ing Information).
Calculations with a water probe, with the ligand in-
cluded and water7 excluded, gave an energy minimum
close to that of water7 with a value of -7.4 kcal/mol.
This energy minimum was based on a strong hydrogen
bond interaction with the oxygen atom in water14 and
a weak hydrogen bond with the oxygen atom in water9.
The two water molecules (water7 and -14) stabilize the
conformation of the hydrophobic moiety observed in the
soaked structure. The crystal structure of 7 soaked into
the hPPARγ receptor11 was superimposed with ragagli-
tazar soaked into the hPPARγ receptor (Figure 5). In
the alignment, the main differences were found in the
tricyclic part of the ligands. The carbazole ring system
in 7 had no possibility of making hydrogen bonds to
water molecules, whereas the phenoxazine ring system
in ragaglitazar had this possibility, thus affecting the
binding mode of this ligand.
A structural comparison between the apo structure
and the two complexes of hPPARγ-LBD showed that the
protein had changed only slightly to make the best fit
to ragaglitazar. To study the structural flexibility of the
ligand in different environments, the two different
conformations of ragaglitazar in the single-crystal struc-
ture of the L-arginine salt (A and B, Figure 3) were
compared to the conformations observed in the protein
complex crystal structures obtained by soaking and by
co-crystallization, respectively. Thus, in total four dif-
ferent molecular conformations of ragaglitazar were
observed (Figure 6).
In conclusion, a comparison of the structures of
ragaglitazar observed in the protein complex to those
observed in the single-crystal structure of the ragagli-
tazar, L-arginine salt showed that the ligand undergoes
quite drastic changes dependent on its environment and
adapts to the hPPARγ-LBD protein.
The difference between the four conformations ob-
served for ragaglitazar was to a large extent described
by the two torsion angles extending from the ethoxy
linker to the phenyl group, C(14x)-O(2x)-C(15x)-
C(16x) and C(14x)-O(2x)-C(15x)-C(20x) where x is
denoting A or B in Figure 3. The values for the angles
were 170.2(10)° and -8.9(15)° for molecule A and
13.7(17)° and -167.3(11)° for molecule B in the ragagli-
tazar, L-arginine salt crystal, respectively. Thus, the
torsion angle differences between molecules A and B
were 157(3)° and 158(3)° (Figure 6). The corresponding
torsion angles for the protein ligands were 73° and
-107° for the soaked-in ligand and 70° and -110° for
the co-crystallized ligand. The differences between the
To understand the hPPARR and -γ activity and the
lack of hPPARδ activity for ragaglitazar at a molecular
level, the binding pockets in the different receptors were
compared. The ligand binding domains of the hPPARγ,
hPPARR, and hPPARδ receptors that were crystallized
with ragaglitazar (3), tesaglitazar (4),14 and GW2433
(8)35 (R, EC50 ) 0.17 µM; γ, EC50 ) 2.5 µM; δ, EC50
0.19 µM)5 (Figure 1), respectively, were superimposed
)

1312 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8
Ebdrup et al.
F igu r e 7. Interactions between a DRY probe (blue grid) and the ligand-binding domain of the hPPARγ receptor. The amino
acids are shown as stick models, and the ligand is shown as a ball-and-stick model. The crystallized ragaglitazar is shown with
green carbon atoms, and the FlexX docked ligand is shown with magenta carbon. The amino acids shown are participating in
interactions with the ligand.
F igu r e 8. Interactions between a DRY probe (blue grid) and the ligand binding domain of the hPPARR receptor. The amino
acids are shown as stick models, and the ligands are shown as ball-and-stick models. The crystallized tesaglitazar is shown with
green carbon atoms and the FlexX docked ragaglitazar is shown with magenta carbon atoms. The amino acids shown are
participating in interactions with the ligand.
and the ligands removed. To investigate and compare
the characteristics of the hydrophobic pocket, Grid
calculations36-39 with a DRY (hydrophobic) probe were
performed in the part of the pocket that was formed by
the amino acids in Table 2. These calculations showed
significant differences between the binding pockets
(Figures 7-9). The binding pocket in the hPPARγ
receptor showed its main interaction with the DRY
probe in the area where the phenoxazine group in
ragaglitazar was located (Figure 7). The Grid calcula-
tions in the hydrophobic pocket of the hPPARR receptor
showed the same pattern as for the hPPARγ receptor.
The main interactions were found in the area where the
phenyl ring substituted with a methylsulfoxy group was
located, (Figure 8). The binding pocket in the hPPARδ
receptor showed two areas in which interactions be-
tween the DRY probe and the binding pocket were found
(Figure 9). These correspond to the tail-up and tail-down
configurations of eicosapentaenoic acid (EPA) crystal-
lized with the ligand binding domain of the hPPARδ
receptor.40 The amino acids that formed the hydrophobic
pocket in the different receptors are listed in Table 2.
One part of the hydrophobic pocket that was found to
interact with the DRY probe was common for the

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1313
F igu r e 9. Interactions between a DRY probe (blue grid) and the ligand-binding domain of the hPPARδ receptor. The amino
acids are shown as stick models, and the ligands are shown as ball-and-stick models. The crystallized GW2433 is shown with
green carbon atoms, and the FlexX docked ragaglitazar is shown with magenta carbon atoms. The amino acids shown are
participating in interactions with the ligand.
Ta ble 3. Results from FlexX Docking of Ragaglitazar, Tesaglitazar, and GW2433
CScore ) 5
rms, Å
CScore ) 4
rms, Å
3.4
no. of solutions
FlexX score
no. of solutions
FlexX score
-12.7
-19.1 to -16.9
-16.8 to -16.5
-21.5 to -18.8
-22.5 to -20.5
-17.5 to -14.1
-24.5 to -20.7
PPARγ and ragaglitazar (3)
PPARγ and tesaglitazar (4)
PPARR and tesaglitazar (4)
PPARR and ragaglitazar (3)
PPARδ and GW2433 (8)
PPARδ and ragaglitazar (3)
PPARδ and GW501516 (9)
3
2
0
2
2
1
4
1.9-2.3
-17.1 to -15.5
-21.6 to -20.4
1
4
2
3
3
0.8-1.3
3.1-6.8
-23.2 to -22.4
-24.4 to -23.6
-16.2
3.1-6.8
2
10
-26.4 to -24.3
hPPARR and the hPPARγ receptors and the tail-down
pocket in the hPPARδ receptor. The DRY-probe tail-up
pocket in the hPPARδ receptor was not found in the two
other receptors. Two of the amino acids in the hPPARδ
receptor that, according to Grid calculations, gave this
pocket hydrophobic character (Val334 and Ile364) were
substituted with Met325 and Met355 in the hPPARR
receptor and with Met334 and Met364 in the hPPARγ
receptor, which resulted in a different interaction pat-
tern between the DRY probe and the binding pocket.
To further understand the activity of ragaglitazar,
FlexX41-44 was used to dock ragaglitazar into the
binding pockets of the hPPARR, hPPARγ, and hPPARδ
receptors. A combination of the rms values to the
relevant crystallized ligand, consensus scoring (CScore)
values,45 and FlexX scores were used to analyze the
results (Table 3). To validate the FlexX dockings,
ragaglitazar was docked into the crystal structure of the
hPPARγ receptor, tesaglitazar was docked into the
crystal structure of the hPPARR receptor, and GW2433
was docked into the crystal structure of the hPPARδ
receptor (Table 3). The calculations show that FlexX
is able to reproduce the experimental binding modes
with high CScore values (Table 3). To understand the
hPPARR activity and investigate possible binding modes,
ragaglitazar was docked into the hPPARR receptor
(Table 3). FlexX gave solutions with high CScore values
and favorable FlexX scores. Ragaglitazar was predicted
to bind to the hPPARR receptor in a mode similar to
that of tesaglitazar (Figure 8). To further validate the
predicted binding modes, tesaglitazar was docked into
the hPPARγ receptor (Table 3). This gave solutions with
high CScore values and favorable FlexX scores.
To understand the lack of hPPARδ activity, ragagli-
tazar was docked into the hPPARδ receptor (Table 3).
When the results from the docking of GW2433 and
ragaglitazar into the hPPARδ receptor were compared
(Table 3), docking of ragaglitazar gave fewer solutions
with high CScore compared to GW2433 and less favor-
able FlexX scores. The selective and potent hPPARδ
agonist GW501516 (R, EC50 > 1 µM; γ, EC50 > 1 µM; δ,
EC50 ) 0.001 µM) (9)40 (Figure 1) was also docked into
the hPPARδ receptor (Table 3). This gave high CScore
values and favorable FlexX scores. In the proposed
docking modes for GW501516, the hydrophobic tail
could bind both in the tail-up pocket formed by Phe226,
Met228, Met329, Leu330, Ile333, and Val334 and in the
tail-down pocket that was formed by Ile249, Leu255,
Ile277, Val281, Val341, Val348, Phe352, and Leu356.
Thus, on the basis of the FlexX dockings, we propose
that the absence of hPPARδ activity for ragaglitazar can
be explained by a lack of favorable interactions between
the phenoxazine ring system in ragaglitazar and the
hydrophobic part of the binding pocket in the hPPARδ
receptor. Furthermore, on the basis of Grid calculations
and the crystal structure of the hPPARδ receptor

1314 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8
Ebdrup et al.
crystallized with GW2433,35 we propose that the hy-
drophobic part of the binding pocket consists of two
pockets and that GW2433 makes interactions with both
at the same time as also proposed by Xu et al.35 On the
basis of FlexX docking, GW501516 was predicted to be
able to bind to either pocket in the hPPARδ receptor.
Because the tail-down pocket is common to all three
receptors and since ragaglitazar lacks hPPARδ activity,
we propose that interactions between the tail-up pocket
in the hPPARδ receptor and a ligand are of major
importance for hPPARδ activity.
was added dropwise to a mixture of tert-butyl methyl ether
(55 mL) and potassium tert-butoxide (4.65 g, 41.4 mmol) under
a nitrogen atmosphere at 20-30 °C. 4-Benzyloxybenzaldehyde
13 (4.61 g, 21.7 mmol) was added in portions to this mixture
at 5 °C followed by the addition of tert-butyl alcohol (6.70 g).
The reaction mixture was allowed to reach 15 °C and stirred
at this temperature for approximately 30 min, after which the
reaction was completed (as judged by TLC). Water (30 mL)
was added at 5-10 °C, and the phases were allowed to
separate. The organic phase was concentrated in vacuo, and
ethanol (30 mL) was added to the stirred solution. After
crystallization had occurred, water (18 mL) was added to the
suspension. The light-yellow title compound was filtered off,
washed with ethanol/water (1:1 v/v), and dried in vacuo to yield
1
Con clu sion
6.52 g (92% yield). H NMR, 400 MHz (acetone-d6): δ 1.12 (t,
J ) 7 Hz), 1.32 (t, J ) 7 Hz), 1.33 (t, J ) 7 Hz), 3.91 (q, J )
7 Hz), 4.00 3.91 (q, J ) 7 Hz), 4.12 (q, J ) 7 Hz), 4.24 (q, J )
7 Hz), 5.12 (s), 5.17 (s), 6.10 (s), 6.93 (s), 6.94 (d, J ) 9 Hz),
7.05 (d, J ) 9 Hz), 7.15 (d, J ) 9 Hz), 7.32-7.42 (m), 7.46-
7.50 (m), 7.81 (d, J ) 9 Hz).
A new and improved synthesis of ragaglitazar ap-
plicable for large-scale preparation was developed. The
absolute stereochemistry of ragaglitazar was, not sur-
prisingly, established to be S. By comparison of the
structures of ragaglitazar, L-arginine in the single
crystal to the structures in the hPPARγ-LBD protein
complex, it was shown that the ligand undergoes quite
drastic changes to adopt to its environment. Ragagli-
tazar had high affinity for the hPPARR and -γ receptors,
while the lack of hPPARδ activity could be explained
by the absence of binding to the tail-up pocket in the
hPPARδ receptor. The more pronounced hPPARR activ-
ity observed for ragaglitazar and other dual-acting
PPARR and -γ agonists gives hope for a new class of
drugs with improved properties compared to the known
insulin sensitizers (TZDs).
Eth yl 2-Eth oxy-3-(4-h yd r oxyp h en yl)p r op a n oa te, r a c-
16. Ethyl E/Z-3-(4-benzyloxyphenyl)-2-ethoxyacrylate (15) (20.0
g, 61.3 mmol) dissolved in tert-butyl methyl ether (40 mL)
charged with palladium on carbon (5%) (1.0 g, Engelhard Tech
code no. 4531) was hydrogenated with vigorous stirring at
atmospheric pressure at room temperature for 2-3 days. The
catalyst was filtered off and washed with a few milliliters of
tert-butyl methyl ether. The combined filtrates were concen-
trated in vacuo to yield 14.5 g (99% yield) of the title compound
as a viscous oil, which crystallizes upon standing. 1H NMR,
400 MHz (acetone-d6): δ 1.09 (t, J ) 7 Hz, 3H), 1.17 (t, J ) 7
Hz, 3H), 2.83-2.91 (m, 2H), 3.35 (dq, J ) 7 and 14 Hz, 1H),
3.55 (dq, J ) 7 and 14 Hz, 1H), 3.98 (dd, J ) 4,7 Hz, 1H), 4.10
(q, J ) 7 Hz, 2H), 6.74 (d, J ) 9 Hz, 2H), 7.06 (d, J ) 9 Hz,
2H), 8.08 (s, 1H). 13C NMR (acetone-d6), 100 MHz: δ 4.9, 15.9,
39.5, 61.3, 66.6, 81.5, 116.2, 129.3, 131.6, 157.3, 173.0.
Exp er im en ta l Section
(S)-2-Eth oxy-3-(4-h ydr oxyph en yl)pr opan oic Acid, S-17.
Ethyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate (50.0 g, 210.0
mmol) (rac-16) was mixed with an aqueous phosphate buffer,
pH 7.0 (0.1 M, 100 mL). Pectinex Ultra SP-L (Novozymes A/S,
Denmark) (150 mL) was added, and the mixture was stirred
vigorously for approximately 44 h at room temperature. During
that time, the pH of the reaction mixture was kept constant
at pH 6.5-7.5 by addition of NaOH (5 M, 19.0 mL). After 45%
conversion, most of the water was evaporated in vacuo
(approximately 200 mL) and methanol (500 mL) was added
to the remaining slurry. The precipitate that formed was
filtered off, and the methanol was evaporated in vacuo. The
remaining oil was dissolved in water (150 mL) followed by
extraction of the unreacted ester with tert-butyl methyl ether
(4 × 100 mL). The water phase was acidified to pH 3 and
extracted with tert-butyl methyl ether (3 × 150 mL). After the
mixture was dried over Na2SO4 and after evaporation of the
solvent, 17.0 g (39%) of the title compound was obtained as
an oil that crystallized on standing (mp 105 °C, ee ) 99.6%).
1H NMR, 400 MHz (acetone-d6): δ 1.10 (t, J ) 7 Hz, 3H), 2.85
(dd, J ) 7 and 14 Hz, 1H), 2.96 (dd, J ) 4 and 14 Hz, 1H),
3.37 (dq, J ) 7 and 14 Hz, 1H), 3.62 (dq, J ) 7 and 14 Hz,
1H), 4.01 (dd, J ) 4,7 Hz, 1H), 6.74 (d, J ) 8 Hz, 2H), 7.10 (d,
J ) 8 Hz, 2H). 13C NMR (acetone-d6), 100 MHz: δ 15.4, 38.9,
66.4, 80.8, 115. 8, 129.3, 131.2, 156.9, 173.5.
Ch em ist r y. Melting points were determined either on a
capillary melting point apparatus and are uncorrected or on
a DSC instrument. NMR data were recorded on a 300 MHz
and on a 400 MHz spectrometer. The mass spectrum was
recorded on a quadrupole ion trap mass spectrometer with an
electrospray ion source (source voltage 3.52 kV, capillary
voltage 3.66 V, capillary temperature 175 °C). Elemental
analyses were within 0.1% of the calculated values. Enantio-
meric purities were determined using capillary electrophoresis
performed on a HP3D capillary electrophoresis instrument
(80.5/72.0 cm, 50 µm HP extended light path capillary),
Agilent, Waldborn, Germany. The electrolyte used for com-
pounds S-17 and S-18 was HS-â-CD (Regis) (2% w/v) and TM-
â-CD (SIGMA) (2% w/v) in 24 mM borate buffer, pH 9.3 (HP).
The electrolyte used for compounds 3 and 3, L-arginine was
10%/90% v/v ACN (Rathburn)/(2.0% SB-â-CD (CyDex) and
0.70% DM-â-CD (Agilent) (w/w) in 25 mM phosphate buffer,
pH 8.0. The applied voltage was 30 kV. The infrared spectrum
was recorded in KBr by a Perkin-Elmer One FTIR IR spec-
trophotometer equipped with a diffuse reflectance accessory.
Syn th esis of Com p ou n d s Accor d in g to Sch em e 1.
2-P h en oxa zin -10-yleth yl Meth a n esu lfon a te, 12. Triethyl-
amine (24.3 g, 240 mmol) was added to a solution of 2-phe-
noxazine-10-ylethanol (11) (36.0 g, 158 mmol) in dry dichlo-
romethane (180 mL). Methanesulfonyl chloride (22.5 g, 196
mmol) was added dropwise at 5-10 °C. The reaction mixture
was then heated to 35 °C for 3 h followed by washing with
water (3 × 150 mL). The organic phase was then dried over
MgSO4 and filtered, and most of the solvent was removed in
vacuo. n-Hexane (300 mL) was added and the precipitate was
filtered and dried in vacuo to yield 42.5 g (88% yield) of the
(S)-2-P r opyl 2-Eth oxy-3-(4-h ydr oxyph en yl)pr opan oate,
S-18. (S)-2-Ethoxy-3-(4-hydroxyphenyl)propanoic acid (S-17)
(16.52 g, 78.6 mmol) was dissolved in dry 2-propanol (80 mL).
Thionyl chloride (9.35 g, 78.6 mmol) was added slowly to this
solution at room temperature. The mixture was heated to 60
°C and stirred for 2 h at this temperature. The solvent was
distilled off, and the remainder was dissolved in tert-butyl
methyl ether (100 mL). The solution was washed with aqueous
NaHCO3 (10%, 2 × 50 mL), dried over Na2SO4, and filtered,
and the solvent was removed in vacuo. A yield of 17.45 g (88%)
of the title compound was obtained as an oil, which crystallized
1
title compound (mp 88-92 °C). H NMR, 400 MHz (acetone-
d6): δ 3.11 (s, 3H), 4.06 (t, J ) 8 Hz, 2H), 4.50 (t, J ) 8 Hz,
2H), 6.63-6.66 (m, 2H), 6.69-6.72 (m, 2H), 6.79-6.86 (m, 4H).
13C NMR (acetone-d6), 100 MHz: δ 37.3, 44.1, 66.4, 113.2,
116.2, 122.4, 124.8, 133.8, 145.6.
1
Eth yl E/Z-3-(4-Ben zyloxyp h en yl)-2-eth oxya cr yla te, 15.
Triethyl 2-ethoxyphosphonoacetate (14) (11.11 g, 41.4 mmol)
on standing (ee ) 99.4%). H NMR, 400 MHz (acetone-d6): δ
1.10 (t, J ) 7 Hz, 3H), 1.12 (d, J ) 6 Hz, 3H), 1.19 (d, J ) 6

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1315
Hz, 3H), 2.84-2.87 (m, 2H), 3.36 (m, 1H), 3.55 (m, 1H), 4.94
(m, 1H), 6.74 (d, J ) 8 Hz, 2H), 7.07 (d, J ) 8 Hz, 2H). 13C
NMR (acetone-d6), 75 MHz: δ 15.9, 22.3, 22.4, 39.4, 66.5, 68.8,
81.6, 116.1, 129.3, 131.6, 157.3, 172.6.
from adipose tissue using the following primers: sense, 5′-
TGCACTCGAGCATGACCATGGTTGACACAGAG-3′; antisense,
5′-TCAGTCTAGACTAGTACAAGTCCTTGTAGAT-3′. Ampli-
fied cDNAs were cloned into pCR2.1-TOPO (Invitrogen) and
sequenced. For expression in E. coli, the ligand binding domain
(LBD, amino acids C165 - Stop) of hPPARγ was generated by
PCR using the primers (sense) 5′-Gatcggatcctctcacacaac-
gcgattcgtttt-3′ and (antisense) 5′-TCAGTCTAGACTAGTA-
CAAGTCCTTGTAGAT-3′ and fused in frame to glutathione
S-transferase in the vector pGEX-3X (Amersham Pharmacia
Biotech). The ensuing construct was verified by sequencing.
A 5 mL culture inoculated with E. coli harboring this construct
was grown overnight at 37 °C. After dilution (1:100) and
regrowth to OD600 ) 0.5, the culture was cooled to 25 °C.
Recombinant protein was produced by adding IPTG to the
culture to a final concentration of 0.1 mM.
(S)-2-Eth oxy-3-(4-[2-(ph en oxazin e-10-yl)eth oxy]ph en yl)-
p r op a n oic Acid , S-3. A mixture of 2-phenoxazin-10-ylethyl
methanesulfonate (12) (12.80 g, 41.9 mmol), potassium car-
bonate (8.70 g, 62.9 mmol), and toluene (65 mL) was heated
to reflux. (S)-2-Propyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate
(S-18) (10.10 g, 40.0 mmol) dissolved in toluene (30 mL) was
added dropwise over 2 h, and the mixture was then refluxed
for 20 h. The reaction mixture was cooled to room temperature,
and water (50 mL) was added. The organic phase was
separated and washed with acetic acid (5%, 50 mL). Evapora-
tion of the solvent in vacuo yielded crude (S)-2-propyl 2-ethoxy-
3-(4-[2-phenoxazine-10-yl)ethoxy]phenyl)propanoate (S-19),
which was dissolved in a mixture of water (35 mL) and
2-propanol (35 mL). Aqueous sodium hydroxide (30%, 12 g)
was added, and the mixture was stirred vigorously for 16 h at
room temperature. Toluene (50 mL) and fumaric acid (5 g)
were added to the mixture, and the organic phase was
separated. The organic phase was washed with brine (20 mL),
dried over Na2SO4, decolorized with charcoal, and filtered. The
title compound was then crystallized by addition of n-heptane.
The crystalline title compound was filtered off and dried in
vacuo to give 13.59 g (81%) of the title compound (mp 89-90
°C, ee ) 98.2%).46 1H NMR, 300 MHz (acetone-d6): δ 1.09 (t,
J ) 7 Hz, 3H), 2.87 (dd, J ) 8 and 14 Hz, 1H), 2.98 (dd, J )
5 and 14 Hz, 1H), 3.36 (dq, J ) 7 and 14 Hz, 1H), 3.62 (dq, J
) 7 and 14 Hz, 1H), 4.01 (dd, J ) 5,7 Hz, 1H), 4.06 (t, J ) 6
Hz, 2H), 4.27 (t, J ) 6 Hz, 2H), 6.60-6.71 (m, 4H), 6.80-6.86
(m, 4H), 6.86 (d, J ) 9 Hz, 2H), 7.18 (d, J ) 9 Hz, 2H). 13C
NMR (acetone-d6), 75 MHz: δ 15.8, 39.2, 45.2, 65.5, 66.8, 80.9,
113.8, 115.4, 116.34 122.4, 125.1, 131.4, 131.7, 134.7, 146.0,
158.7. Anal. (C25H25NO5) C, H, N.
Cells were harvested and resuspended in 50 mM imidazole,
pH 7.2, 5 mM EDTA, 10% glycerol, 0.1% PMSF, 0.1% DNase,
and 0.1% MgCl2. After stirring for 1 h, the cells were broken
by cell disruption at 150 bar. The material was frozen, thawed,
and centrifuged at 40000g for 30 min. The supernatant was
loaded on a GSH-Sepharose column (Pharmacia) in 50 mM
imidazole, pH 7.2, 1 mM EDTA, 10% glycerol, and 150 mM
NaCl and eluted with 250 mM imidazole, pH 7.0, and 300 mM
NaCl. The material was subjected to factor Xa digestion (10
units FXa/mg protein) for 16 h. Cleaved GST protein was
removed by passage over a GSH-Sepharose column. The
hPPARγ-LBD was further purified by Superose 200 size
exclusion chromatography (Pharmacia) in 20 mM Tris, pH 8.0,
0.5 mM EDTA, 5 mM DTT. After SDS-PAGE and MALDI-
TOF MS analysis, fractions containing hPPARγ-LBD were
pooled and concentrated by Centricon-30 (Amicon) to 15 mg/
mL. N-terminal sequencing verified the cleaved protein to
consist of a.a. 188-476.
Cr yst a lliza t ion of t h e h P P ARγ-LBD-R a ga glit a za r
Com p lex. Crystals of the hPPARγ-LBD-ligand complex were
produced in two different ways: by co-crystallization and by
soaking.
L-Ar gin in iu m (2S)-2-Eth oxy-3-(4-[2-(10H-p h en oxa zin e-
10-yl)eth oxy]p h en yl)p r op a n oa te, S-3, L-Ar gin in e. (S)-2-
Ethoxy-3-(4-[2-(phenoxazine-10-yl)ethoxy]phenyl)propanoic acid
(S-3) (12.00 g, 28.6 mmol) was dissolved in ethanol (260 mL),
and the solution was filtered. L-Arginine (4.98 g, 28.6 mmol)
dissolved in water (35 mL) at 50-60 °C was added dropwise
to the ethanol solution at 75-78 °C with stirring. The solution
was slowly cooled to room temperature overnight and finally
to 0-5 °C with vigorous stirring. The crystalline precipitate
was filtered off, washed with ethanol (2 × 30 mL), and dried
in vacuo to give 15.28 g (90%) of the title compound as an off-
1. Co-cr ysta lliza tion . The free acid form of the ligand
ragaglitazar was dissolved in DMSO and mixed with hPPARγ-
LBD in a molar ratio of 3:1. The mixture was left for 60 min
before crystallization setup. Crystals were grown by the sit-
ting drop vapor diffusion method, mixing equal amounts of
hPPARγ-LBD-ragaglitazar and reservoir (0.1 M Tris buffer,
pH 7.0, 30% PEG monomethylether (mme) 5000, and 0.5 M
ammonium sulfate). The final concentration of DMSO in the
drop was 1% due to the ragaglitazar solution. Crystals grew
to a size of 0.15 mm × 0.15 mm × 0.02 mm over 14 days.
Crystals were flash-frozen at 100 K prior to data collection.
Cryoconditions were 35% PEGmme 5000, 0.1 M Tris, pH 7,
and 0.2 M ammonium sulfate including ragaglitazar. Diffrac-
tion data to 2.9 Å resolution were collected at beamline 71147
equipped with a Mar345 image plate detector at the MAX-lab
synchrotron facilities. Data processing and reduction were
performed using the programs DENZO and SCALEPACK.48
Cell parameters, space group, and data statistics are listed in
Table S2 (Supporting Information).
white powder (mp 180 °C (DSC), ee ) 98.2%). Anal. (C31H39
-
N5O7) C, H, N. IR (KBr), cm-1: 3240, 3063, 2971, 2871, 3500-
2500 br, 2643, 1708, 1688, 1615, 1587, 1510, 1489, 1374, 1347,
1
1312, 1273, 1244, 1129, 1045. MS m/z: 420 (MH+). H NMR,
400 MHz (dmso-d6): δ 0.97 (t, J ) 7 Hz, 3H), 1.57 (m, 2H),
1.65 (m, 1H), 1.75 (m, 1H), 2.64 (dd, J ) 14 and 8.5 Hz, 1H),
2.82 (dd, J ) 14 and 4 Hz, 1H), 3.05 (m, 2H), 3.14 (dq, J ) 9
and 7 Hz, 1H), 3.23 (t, J ) 5.5 Hz, 1H), 3.52 (dq, J ) 9,7 Hz,
1H), 3.59 (dd, J ) 4 and 8.5 Hz, 1H), 3.99 (t, J ) 6 Hz, 2H),
4.17 (t, J ) 6 Hz, 2H), 6.64 (m, 2H), 6.67 (m, 2H), 6.82, (m,
2H), 6.83 (m, 2H), 6.79 (d, J ) 8.5 Hz, 2H), 7.11 (d, J ) 8.5
Hz, 2H), 7.90 (br. s, 8H). 13C NMR (DMSO-d6), 100 MHz: δ
15.1, 24.6, 28.5, 38.3, 40.3, 43.5, 53.5, 63.8, 64.1, 81.9, 112.5,
113.8, 114.9, 121.0, 123.9, 130.0, 132.1, 132.9, 143.9, 156.3,
157.5, 171.5, 175.7. Powder XRD diffractogram is given in
Supporting Information. pKa ) 3.7. Solubility in water is
greater than 40 mg/mL.
Cr ysta lliza tion of Ra ga glita za r , L-Ar gin in e. Growth of
single crystals of the ragaglitazar, L-arginine salt from solution
in 2-propanol or ethanol was not successful. Therefore, ragagli-
tazar, L-arginine (28 mg) was dissolved in dimethyl sulfoxide
(0.3 mL) and large (up to 0.5 mm) plate-shaped (0.01 mm thick)
single crystals suitable for X-ray diffraction were grown by
slow evaporation at 40 °C over 1-2 days. Crystal data are
listed in Supporting Information.
2. Soa k ed Cr ysta ls. Crystals of the apo protein were grown
using the sitting drop vapor diffusion method at slightly
modified conditions reported by Nolte et al.32 Crystals were
grown by mixing equal amounts of hPPARγ-LBD and reservoir
solution (0.8 M sodium citrate and 0.15 M Tris, pH 8.0). Within
10 days, crystals grew to a size of 0.15 mm × 0.15 mm × 0.1
mm. The ligand was introduced to the drop by addition of 0.15
µL of 1 mM arginine salt of ragaglitazar dissolved in water.
The crystals were left to soak for 10 days. Crystals were flash-
frozen at 100 K prior to data collection. Cryoconditions were
40% glycerol, 0.6 M sodium citrate, and 7% PEGmme 5000
including the arginine salt of ragaglitazar. Diffraction data to
2.65 Å resolution were collected on a Mar345 image plate
detector mounted at a Rigaku RU-300 rotating anode genera-
tor. Data processing and reduction were performed using the
programs DENZO and SCALEPACK.48 Cell parameters, space
Exp r ession a n d P u r ifica tion of h P P ARγ-LBD. A full-
length hPPARγ cDNA was obtained by PCR amplification
using cDNA synthesized by reverse transcription of mRNA

1316 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8
Ebdrup et al.
group, and data statistics are listed in Table S2 (Supporting
Information).
L-arginine salt molecules and the protein ligands (Table S1),
data collection for the co-crystallized and soaked hPPARγ-LBD
crystal (Table S2), statistical data for the co-crystallized and
soaked hPPARγ-LBD crystal (Table S3), and powder X-ray
diffraction results (Figure S1). This material is available free
of charge via the Internet at http://pubs.acs.org.
Str u ctu r e Deter m in ation of th e h P P ARγ-LBD-Ragagli-
ta za r Com p lex. Both structures were solved by the molecular
replacement method with the apo hPPARγ-LBD structure
1PRG (Nolte et al.32) as search model using the program
AMORE.49,50 Refinement was performed using the programs
X-PLOR51 and CNX52 and model building with the program
Quanta (Molecular Simulations). Parameter and topology files
were generated using the program Xplo2D.53 Refinement
statistics are listed in Table S3 (Supporting Information). After
an initial cycle of refinement and model building, the ligand
was introduced in the Fo - Fc difference electron density (ed)
map. Water molecules were introduced in the structure based
on the diffraction data from the soaked crystals using a 2.5σ
cutoff in the Fo - Fc difference ed map. The coordinates have
been deposited in the Brookhaven Protein Data Bank, ID
1NYX.
Mod elin g. The ligand binding domains of hPPARR,
hPPARγ, and hPPARδ were superimposed with the backbone
atoms in the amino acids shown in Table 2, Tyr314 and His440
(PPARR), His323 and His449 (PPARγ), and His323 and His449
(PPARδ). The rms between the hPPARR and hPPARγ recep-
tors was 0.67 Å, and the rms between hPPARδ and hPPARγ
receptors was 0.65 Å.
The Grid calculations were performed with Grid, version
20,36-38 with DPRO equal to 4, DWAT equal to 80, and EMAX
equal to 5. The calculations with the DRY probe were
performed with two planes per ångstrom and with the water
probe with five planes per ångstrom. All calculations on the
complexes between ragaglitazar and the hPPARγ receptor
were performed using the structure based on soaked crystals.
FlexX version 1.10.0 was used.41-44 Assignment of formal
charges in the ligands was used. Consensus scoring was
applied on the obtained solutions.45 FlexX was used with the
Sybyl 6.8 interface.54
Refer en ces
(1) Ginsberg, H.; Plutzky, J .; Sobel, E. Review of metabolic and
cardiovascular effects of oral antidiabetic agents: beyong glucose-
level lowering. J . Cardivasc. Risk 1999, 6 (5), 337-346.
(2) Cobb, J .; Dukes, I. Recent Advances in the Development of
Agents for Treatment of Type 2 Diabetes. Annu. Rep. Med.
Chem. 1998, 33, 213-221.
(3) Grover, S. A.; Coupal, L.; Zowall, H.; Dorais, M. Cost-Effective-
ness of Treating Hyperlipidemia in the Presence of Diabetes.
Circulation 2000, 102, 722-727.
(4) Bailey, C. J . Potential new treatments for type 2 diabetes. TIPS
2000, 21, 259-265.
(5) 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.
(6) Staels, B.; Dallongeville, J .; Auwerx, J .; Schoonjans, K.; Leit-
ersdorf, E.; Fruchart, J .-C. Mecanism of Action of Fibrates on
Lipid and Lipoprotein Metabolism. Circulation 1998, 98, 2088-
2093.
(7) Chaput, E.; Saladin, R.; Silvestre, M.; Edgar, A. D. Fenofibrate
and Rosiglitazone Lower Triglycerides with Opposing Effects on
Body Weight. Biochem. Biophys. Res. Commun. 2000, 271, 445-
450.
(8) Boyle, P. J .; King, A. B.; Olansky, L.; Marchetti, A.; Lau, H.;
Magar, R.; Martin J . Effects of Pioglitazone and Rosiglitazone
on Blood Lipid Levels and Glycemic Control in Patients with
Type 2 Diabetes Melitus: A Retrospective Review of Randomly
Selected Medical Records. Clin. Ther. 2002, 24 (3), 378-396.
(9) Chilcott, J .; Tappendem, P.; J ones, M. L.; Wight, J . P. A
Systematic Review of the Clinical Effectiveness of Pioglitazone
in the Treatment of Type 2 Diabetes Mellitus. Clin. Ther. 2001,
23, 1792-1823.
The ligands used for docking were built with Maestro 4.1,55
and the global energy minimum was found using MacroModel
7.256 and the SUMM method.57 The calculations were per-
formed with the MMFF force field58-64 and with water as the
solvation model.65
Bin d in g Assa ys. The hPPAR binding assay described by
Nichols et al.29 was used with a few minor adjustments. The
ligand binding assays were based on SPA (scintillation prox-
imity assay) particles.
Ragaglitazar (Kd ) 85 nM) and NNC 61-4655 (Kd ) 175 nM)
were labeled with 3H and used as a tracer for the receptors
GST-PPARγ-LBD and HIS-PPARR-LBD, respectively. [3H]-
Ragaglitazar was prepared by tritiation of an acrylic acid
precursor,66 and [3H]-NNC 61-4655 was prepared by catalytic
tritium dehalogenation of an aryl bromide precursor.67
(10) Wulff, E. M.; J eppesen, L.; Bury, P. S.; Mogensen, J . P.; Fleckner,
J .; Andersen, A.-S. T.; Wassermann, K.; Sauerberg, P.; Charac-
terisation of a Future Generation Insulin Sensitizers in Vitro
and in Vivo. Diabetes 2001, 50 (Suppl. 2), A524.
(11) Sauerberg, P.; Pettersson, I.; J eppesen, L.; Bury, P. S.; Mogensen,
J . P.; Wasserman, K.; Brand, C. L.; Sturis, J .; Wo¨ldike, H. F.;
Fleckner, J .; Andersen, A.-S. T.; Mortensen, S. B.; Svensson, L.
A.; Rasmussen, H. B.; Lehman, S. V.; Polivka, Z.; Sindelar, K.;
Panajotova, V.; Ynddal, L.; Wulff, E. M. J . Novel tricyclic-R-
alkyloxyphenylpropionic acids: Dual PPARR/γ agonists with
hypolidemic and antidiabetic activity. J . Med. Chem. 2002, 45,
789-804.
(12) Lohray, B. B.; Lohray, V. B.; Bajii, A. C.; Kalchar, S.; Poondra,
R. R.; Padakanti, S.; Chakrabarti, R.; Vikramadithyan, R. K.;
Misra, P.; J uluri, S.; Namidi, N. V. S. R.; Rajagopalan, R. (-)3-
[4-[2-(phenoxazine-10-yl)ethoxy]phenyl]-2-ethoxypropanoic acid
[(-)DRF 2725]: A Dual PPAR Agonist with Potent Antihyper-
glucemic and Lipid Modulating Activity. J . Med. Chem. 2001,
44, 2675-2678.
Ack n ow led gm en t. We are indebted to Centre for
Crystallographic Studies, University of Copenhagen, for
collection of X-ray diffraction data from the ragaglitazar,
L-arginine salt single crystal. We gratefully acknowledge
useful discussions and technical support from Dr. Reddy
Research Foundation, India, and Siegfried CMS LTD,
Switzerland. The preparation of [3H] tracers by J . B.
Kristensen and S. K. J ohansen and the preparation of
the ragaglitazar, L-arginine salt by Petra Wiede Lug-
stein are gratefully acknowledged. The technical as-
sistance from K. M. Klausen, V. Weil, F. Gundertofte,
A. Hansen, and A. Ryager, Novo Nordisk A/S, is highly
appreciated. Scientific assistance from and fruitful
discussions with M. Zundel, H. Olsen, and I. T. Chris-
tensen at Novo Nordisk A/S and with P. Østergaard at
Novozymes A/S is also gratefully acknowledged.
(13) Kjellstedt, A.; Oakes, N.; Svensson, L.; Carlson, A.-C.; Ljung,
B. AZ 242, a Novel PPARR/γ Agonist, Has Potent and Efficient
Insulin-Sensitizing and Antidyslipidemic Properties in ob/ob
Mice and fa/fa Zucker Rats. Diabetes 2001, 50 (Suppl. 2), A121.
(14) Cronet, P.; Petersen, J . F. W.; Folmer, R.; Blomberg, N.; Sjo¨blom,
K.; Karlsson, U.; Lindstedt, E.-L.; Bamberg, K. Structure of the
PPARR and -γ ligand binding domain in complex with AZ 242;
Ligand selectivity and agonist activation in the PPAR family.
Structure 2001, 9, 699-706.
(15) Murakami, K.; Tobe, K.; Ide, T.; Mochizuki, T.; Ohashi, M.;
Akanuma, Y.; Yazaki, Y.; Kadowaki, T. A Novel Insulin Sensi-
tizer Acts as a Coligand for Peroxisome Proliferator-Activated
Receptor-R (PPAR-R) and PPAR-γ. Effects of PPAR-R Activation
on Abnormal Lipid Metabolism in Liver of Zucker Fatty Rats.
Diabetes 1998, 47, 1841-1847.
(16) Etgen, G. J .; Oldham, B. A.; J ohnson, W. T.; Broderick, C. L.;
Montrose, C. R.; Brozinick, J . T.; Misener, E. A.; Bean, J . S.;
Bensch, W. R.; Brooks, D. A.; Shuker, A. J .; Rito, C. J .;
McCarthy, J . R.; Ardecky, R. J .; Tyhonas, J . S.; Dana, S. L.;
Bilakovics, J . M.; Paterniti, J . R., J r.; Ogilvie, K. M.; Liu, S.;
Kauffman, R. F. A Tailored Therapy for the Metabolic Syndrome,
The Dual Peroxisome Proliferator-Activated Receptor-R/γ Ago-
nist LY465608 Ameliorates Insulin Resistance and Diabetic
Hyperglycemia While Improving Cardiovascular Risk Factors
in Preclinical Models. Diabetes 2002, 51, 1083-1087.
Su p p or tin g In for m a tion Ava ila ble: Crystallographic
data and analysis for the single crystal of ragaglitazar,
L-arginine, comparison of torsion angles in the ragaglitazar,

PPARR/ γ Agonist Ragaglitazar
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 8 1317
(17) Saad, M. F.; Osel, K.; Lewin, A. J .; Patel, N.; Edwards, C. R.;
Greco, S. Nunez, M.; Huang, W.-C.; Reinhardt, R. R. Ragagli-
tazar Improves Lipid Profile and Glycemic Control in Hyper-
triglyceridemic Type 2 Diabetes. Diabetes 2002, 51 (Suppl. 2),
A35-A36.
(40) Oliver, W. R.; Shenk, J . L.; Snaith, M. R.; Russel, C. S.; Plunket,
K. D.; Bodkin, N. L.; Lewis, M. C.; Winegar, D. A.; Sznaidman,
M. L.; Lambert, M. H.; Xu, H. E.; Sternbach, D. D.; Kliewer, S.
A.; Hansen, B. C.; Willson, T. M. A selective peroxisome
proliferator-activated receptor γ agonist promotes reverse cho-
lesterol transport. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 5306-
5311.
(41) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A Fast Docking
Method Using an Incremental Construction Algorithm. J . Mol.
Biol. 1996, 261, 470-489.
(42) Kramer, B.; Rarey, M.; Lengauer, L. Evaluation of the FlexX
incremental construction algorithm for protein-ligand docking.
Proteins: Struct., Funct., Genet. 1999, 37, 228-241.
(43) Rarey, M.; Wefing, S.; Lengauer, T. Placement of medium-sized
molecular fragments into active sites of proteins. J . Comput.-
Aided Mol. Des. 1996, 10, 41-54.
(44) Rarey, M.; Kramer, B.; Lengauer, T. Multiple automatic base
selection: Protein-ligand docking based on incremental con-
struction without manual intervention. J . Comput.-Aided Mol.
Des. 1997, 11, 369-384.
(45) Clark, R. D.; Strizhev, A.; Leonard, J . M.; Blake, J . F.; Matthew,
J . B. Concensus scoring for ligand/protein interactions. J . Mol.
Graphics Modell. 2002, 20, 281-295.
(46) The enantiomeric purity can be significantly increased by
diastereomer crystallization with (-)-ephedrine.
(18) Strand, J .; Sandeman, D.; Skovsted, B.; Edsberg, B.; Mu¨ller, P.
Ragaglitazar (NNC 61-0029, (-)DRF 2725): The Efficacy and
Safety of a Novel Dual PPARR and PPARγ Agonist in Patients
with Type 2 Diabetes. Diabetes 2002, 51 (Suppl. 2), A141-A142.
(19) Ebdrup, S.; Deussen, H.-J .; Zundel, M. Process for the prepara-
tion of substituted 3-phenyl-propanoic acid esters and substi-
tuted 3-phenyl-propanoic acids. PCT Appl. WO 0111072, 2001.
(20) Deussen, H. J .; Zundel, M.; Valdois, M.; Lehmann, S. V.; Weil,
V.; Hjort, C. M.; Østergaard, P. R.; Marcussen, E.; Ebdrup, S.
Process development on the enantioselective enzymatic hydroly-
ses of the (S)-ethyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate.
Org. Proc. Res. Dev. 2003, 7, 82-88.
(21) Grell, W.; Machleidt, H. PO-Aktivierte Alkoxyolifinierung. J ustus
Liebigs Ann. Chem. 1966, 53-67.
(22) Wadsworth, W. S., J r. Synthetic Applications of Phosphoryl-
Stabilized Anions. Org. React. 1977, 25, 73-253.
(23) Linde, H. Untersuchungen an Phenoxazinderivaten. Arch. Pharm.
Ber. Dtsch. Pharm. Ges. 1960, 293, 537-542.
(24) Cauquil, G.; Casadeval, A.; Casadevall, E. Recherches dans la
se´rie da la phenothiazine (3e me´moire) Reaction de metallation
(Research in the phenothiazine series (part 3). Metalation
reaction). Bull. Soc. Chim. Fr. 1960, 1049-1066.
(47) Cerenius, Y.; Stahl, K.; Svensson, L. A.; Ursby, T.; Oskarsson,
A.; Albertsson, J .; Liljas, A. The crystallography beamline I711
at MAX II. J . Synchrotron Radiat. 2000, 7, 203-208.
(48) Otwinowski, Z.; Minor, W. Processing of X-ray Diffraction Data
Collected in Oscillation Mode. Methods Enzymol. 1997, 276,
307-326.
(25) Kim, S. H.; Martin, A. R. Tetracyclic Phenothiazines IV (1).
Synthesis of 1,2,3,4-Tetrahydroazepino[3,2,1-kl]phenothiazine-
4-one. J . Heterocycl. Chem. 1978, 15, 1503-1504.
(26) Nygaard, L.; Wiede, P.; J ensen, A. F.; Olsen, H.; Ebdrup, S.
Unpublished results.
(49) Collaborative Computational Project, N4. The CCP4 Suite:
Programs for Protein Crystallography. Acta Crystallogr. 1994,
D50, 760-763.
(27) SHELXTL. Software package for single crystal structure deter-
mination, version 5.1; Bruker AXS, Inc.: Madison, WI, 1997.
(28) Crystallographic data for the structure has been deposited with
the Cambridge Crystallographic Data Centre as Supplementary
Publication No. CCDC 197689. Copies of the data can be
obtained free of charge on application to CCDC, 12 Union Road,
Cambridge CB2 1EZ, U.K. or on e-mail, deposit@ccdc.ac.uk.
(29) Nicols, J . S.; Parks, D. J .; Consler, T. G.; Blanchard, S. G.
Development of a Scintilation Proximity Assay for Peroxisome
Proliferator-Activation. Anal. Biochem. 1998, 112-119.
(30) Skrumsager, B. K.; Nielsen, K. K.; Pedersen, P. C. NNC 61-0029
((-)DRF2725): The multiple dose pharmacokinetics of a novel
dual acting PPARalpha and PPARgamma agonist in healthy
subjects and type 2 diabetic patients. Diabetes. 2001, 50 (Suppl.
2), A123-A132.
(31) Skrumsager, B. K.; Nielsen, K. K.; Reinhardt, R.; Pedersen, P.
C. NNC 61-0029 ((-)DRF2725): The single dose pharmacoki-
netics of a novel dual acting PPARalpha and PPARgamma
agonist in caucasian and J apanese healthy subjects. Diabetes.
2001, 50 (Suppl. 2), A445.
(32) 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-γ. Nature 1998,
395, 137-143.
(33) Uppenberg, J .; Svensson, C.; J aki, M.; Bertilsson, G.; J endeberg,
L.; Berkenstam, A. Crystal Structure of the Ligand Binding
Domain of the Human Nuclear Receptor PPARγ. J . Biol. Chem.
1998, 273, 31108-31112.
(34) 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.
(35) Xu, H. E.; Lambert, M. H.; Montana, V. G.; Parks, D. J .;
Blanchard, S. G.; Brown, P. J .; Sternbach, D. D.; Lehman, 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.
(36) Goodford, P. J . A computational procedure for determining
energetically favorable binding sites on biologically important
macromolecules J . Med. Chem. 1985, 28, 849-857.
(50) Navaza, J . Amoresan automated package for molecular replace-
ment. Acta Crystallogr. 1994, A50, 157-163.
(51) Bru¨nger, A. T.; Kuriyan, J .; Karplus, M. Crystallographic
R-factor refinement by molecular dynamics. Science 1987, 235,
458-460.
(52) Brunger, A. T.; Adams, P. D.; Clore, G. M.; Delano, W. L.; Gros,
P.; Grosse-Kunstleve, R. W.; J iang, J .-S.; Kuszewski, J .; Nilges,
M.; Pannu, N. S.; Read, R. J .; Rice, L. M.; Simonson, G. L.;
Warren, T.; Badger, J .; Berard, D.; Kumar, R. A.; Szalma, S.;
Yip, P.; Griesinger, C.; J unker, J . Crystallography and NMR
System: A New Software Suite for Macromolecular Structure
Determination. Acta Crystallogr. 1998, D54, 905-921.
(53) Kleywegt, G. J .; J ones, T. A. Databases in Protein Crystal-
lography. Acta Crystallogr. 1998, D54, 1119-1131.
(54) SYBYL, version 6.8; Tripos Inc., 1699 South Hanley Road, St.
Louis, Missouri, 63144.
(55) Schrodinger Inc. (www.schrodinger.com).
(56) Mohamadi, F.; Richards, N. G. J .; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W.
C. MacroModelsAn Integrated Software System for Modeling
Organic and Bioorganic Molecules Using Molecular Mechanics.
J . Comput. Chem. 1990, 11, 440-467.
(57) Goodman, J . M.; Still, W. C. An Unbounded Systematic Search
of Conformational Space. J . Comput. Chem. 1991, 12, 1110-
1117.
(58) Halgren, T. A. Merck Molecular Force Field. I. Basis, Form,
Scope, Parametrization, and Performance of MMFF94. J . Com-
put. Chem. 1996, 17, 490-519.
(59) Halgren, T. A. Merck Molecular Force Field. II. MMFF94 van
der Waals and Electrostatic Parameters for Intermolecular
Interactions. J . Comput. Chem. 1996, 17, 520-552.
(60) Halgren, T. A. Merck Molecular Force Field. III. Molecular
Geometries and Vibrational Frequencies for MMFF94. J . Com-
put. Chem. 1996, 17, 553-586.
(61) Halgren, T. A.; Nachbar, R. B. Merck Molecular Force Field. IV.
Conformational Energies and Geometries for MMFF94. J .
Comput. Chem. 1996, 17, 587-615.
(62) Halgren, T. A. Merck Molecular Force Field. V. Extension of
MMFF94 Using Experimental Data, Additional Computational
Data, and Empirical Rules. J . Comput. Chem. 1996, 17, 616-
641.
(63) Halgren, T. A. MMFF VI. MMFF94s Option for Energy Mini-
mization Studies. J . Comput. Chem. 1999, 20, 720-729.
(64) Halgren, T. A. MMFF VII. Characterization of MMFF94,
MMFF94s, and Other Widely Available Force Fields for Con-
formational Energies and for Intermolecular-Interaction Ener-
gies and Geometries. J . Comput. Chem. 1999, 20, 730-748.
(65) 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.
(66) Kristensen, J . B.; J ohansen, S. K.; Valsborg, J . S.; Martiny, L.;
Foged, C. Unpublished results.
(37) Boobbyer, D. N. A.; Goodford, P. J .; McWhinnie, P. M.; Wade,
R. C. New hydrogen-bond potentials for use in determining
energetically favorable binding sites on molecules of known
structure. J . Med. Chem. 1989, 32, 1083-1094.
(38) 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.
(67) J ohansen, S. K.; Martiny, L. Unpublished results.
(39) Molecular Discovery Ltd., 20a Berkeley Street, Mayfair, London
W1X 5AE, U.K.
J M021027R