Design, synthesis, and structure-activity analysis of isoform-selective retinoic acid receptor β ligands

Article abstract of DOI:10.1021/jm801532e

We recently discovered the isoform selective RARβ2 ligand 4′-octyl-4-biphenylcarboxylic acid (3, AC-55649). Although 3 is highly potent at RARβ2 and displays excellent selectivity, solubility issues make it unsuitable for drug development. Herein we descr

Full text of DOI:10.1021/jm801532e

1540
J. Med. Chem. 2009, 52, 1540–1545
Design, Synthesis, and Structure-Activity Analysis of Isoform-Selective Retinoic Acid Receptor
ꢀ Ligands
Birgitte W. Lund,† Anne Eeg Knapp,† Fabrice Piu,‡ Natalie K. Gauthier,‡ Mikael Begtrup,§ Uli Hacksell,‡ and Roger Olsson*,†
ACADIA Pharmaceuticals AB, Medeon Science Park, S-205 12 Malmo¨, Sweden, ACADIA Pharmaceuticals Inc., 3911 Sorrento Valley
BouleVard, San Diego, California, 92121, and Department of Medicinal Chemistry, The Faculty of Pharmaceutical Sciences, UniVersity of
Copenhagen, UniVersitetsparken 2, DK-2100 Copenhagen, Denmark
ReceiVed December 4, 2008
We recently discovered the isoform selective RARꢀ2 ligand 4′-octyl-4-biphenylcarboxylic acid (3, AC-
55649). Although 3 is highly potent at RARꢀ2 and displays excellent selectivity, solubility issues make it
unsuitable for drug development. Herein we describe the exploration of the SAR in a biphenyl and a
phenylthiazole series of analogues of 3. This ultimately led to the design of 28, a novel, orally available
ligand with excellent isoform selectivity for the RARꢀ2.
Chart 1. Chemical structures
Introduction
Retinoids, biologically active metabolites and synthetic
analogues of vitamin A, e.g., tretinoin (all-trans-retinoic acid,
ATRA,a 1) and 9-cis-retinoic acid (9CRA, 2) (Chart 1), have
not yet reached their full potential in drug discovery mainly
because of severe toxicity when administered systemically. In
part, this toxicity is believed to originate from their nonselective
activation of retinoic acid receptor subtypes (RARs, R, ꢀ, γ),
retinoic X receptor subtypes (RXRs, R, ꢀ, γ), or both and further
at the respective subtype isoforms denoted as R1, R2, ꢀ1-ꢀ5,
γ1, and γ2.1,2 Recently, it was reported that the isoforms have
specific expression patterns and thus discrete pharmacology.3
For example, the RARꢀ2 isoform, which is one of five
discovered isoforms (ꢀ1-ꢀ5) for the RARꢀ subtype, has been
associated with the induction of neural differentiation, motor
axon outgrowth, and neural patterning displayed by ATRA
(1).3a,4 Consequently, the RARꢀ2 isoform might have a role in
the treatment of neurodegenerative diseases, e.g., Alzheimer’s
disease and Parkinson’s disease.5 Thus, isoform selective ligands
would be of interest as leads in drug discovery and as
pharmacological tools in exploratory biology. Designing isoform
selective RARꢀ ligands is, however, an intricate endeavor, as
the ligand-binding domain (LBD) of RARꢀ subtype only differs
by one residue from that of its paralogue RARR and by two
residues from that of RARγ.6 Further complicating the design
of RARꢀ2 isoform selective ligands is the fact that the structural
difference between the RARꢀ1 and RARꢀ2 isoforms lies within
the N-terminal domain that encompasses the ligand independent
activation domain, and thus, the isoforms have identical
LBDs.5,7-11
alkylbiphenyl hit 3 was transformed into a more potent
alkoxythiazole series of RARꢀ2 agonists with retained isoform
selectivity and improved physicochemical properties. Herein,
we report on establishing the underlying structure-activity
relationship (SAR) and the optimization of the drug properties
of the initial hit compound.
Evaluation of Hits
Screening of an in-house chemical library using a cell-based
functional assay, R-SAT, led to the discovery of the isoform
selective RARꢀ2 agonist 4′-octylbiphenyl-4-carboxylic acid
(3).12,13 Compound 3 was originally developed in the context
of liquid crystal research and consequently has low aqueous
solubility (<0.001 mg/mL) and high lipophilicty (log P ) 7.9).
However, 3 displays 100-fold selectivity for RARꢀ2 versus the
other retinoid receptors including RARꢀ1 and toward all of the
other nuclear receptors tested.14-17 The RARꢀ2 isoform is
known to stimulate neurite outgrowth.3,4 In line with this, the
RARꢀ2 isoform selective agonist 3 induced neurite outgrowth
of NTERA-2 cells, indicating that it mediates neuronal dif-
ferentiation similar to that observed with retinoic acid.13 The
compound was further evaluated in a reporter gene assay, and
again 3 was demonstrated to be a potent activator of RARꢀ2.
The activity of 3 was further established in an assay of inhibition
of proliferation of the breast cancer cell line MCF-7.18-21
We recently reported on the discovery of the first agonists
displaying isoform selectivity (RARꢀ2) using the high-
throughput cell-proliferation assay, R-SAT.12 Through a hit to
lead optimization effort, the lipophilic and poorly soluble
* To whom correspondence should be addressed. Phone: +46406013466.
Fax: +46406013405. E-mail: roger@acadia-pharm.com.
†
ACADIA Pharmaceuticals AB.
ACADIA Pharmaceuticals Inc.
University of Copenhagen.
‡
§
a Abbrevations: ATRA, all-trans-retinoic acid; BB, building block;
cLogP, calculated log P; 9CRA, 9-cis-retinoic acid; LBD, ligand-binding
domain; MW, microwave; RAR, retinoic acid receptor; R-SAT, receptor
selection and amplification technology; RXR, retinoic X receptor; SAR,
structure-activity relationship.
10.1021/jm801532e CCC: $40.75
2009 American Chemical Society
Published on Web 02/24/2009

Retinoic Acid Receptor ꢀ Ligands
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 6 1541
Scheme 2. Reaction Details for the Negishi Cross-Coupling
Reactions and the Basic Hydrolysisa
Figure 1. Compound 3 was divided into four building blocks
BB1-BB4.
Scheme 1. Synthesis of Electrophiles for the Negishi Reactionsa
a (a) (i) KOH, CH3CN, 60 °C, 2 h; (ii) RX, KI, 60 °C, 3 h, 34%; (b)
t-BuLi, THF, -20 °C, 1 h; (c) ZnBr2, THF, -20 °C to room temp, 0.5 h;
(d) XAr1CO2Pg, Pd2(dba)3, tfp, THF/NMP 2:1, room temp, 16 h; (e) LiOH,
H2O, THF, 60 °C, 14 h; (f) LiOH, H2O, THF, MW, 160 °C, 5 min. (#)
Resynthesis. (*) Amount after purification of 20 mg of crude material by
preparative LC/MS; X)halogen.
a (a) E1, E2, and E4, PgOH, EDCI ·HCl, DIPA, HOBt, CH3CN, room
temp, 16 h; (b) E3, EtOH, H2SO4, MW, 120 °C, 5 min.
In summary, it was concluded that the hit 3 exhibits retinoid
properties and because of its unique selectivity profile, 3 was
selected as a starting point for a hit-to-lead optimization effort.
Scheme 3. O-Alkylation Procedurea
Chemistry
A number of synthetic strategies were developed in the hit-
to-lead optimization effort. To facilitate the synthetic investiga-
tions and subsequent structure activity-relationship analysis,
biphenyl 3 was divided into four building blocks BB1-BB4
(Figure 1), each of which could be modified independently one
by one or several at a time. To the extent possible the reactions
were conducted in parallel using a 48-well format or, as in the
case of the microwave (MW) assisted reactions, in a sequence
using an automated system. The libraries based on Negishi cross-
coupling reactions were purified by a mass-triggered preparative
LC-MS system collecting one fraction per sample (20 mL).
The goal was to get 2 mg of each compound with >90% purity,
enough to generate the initial activity profile. Carboxylic acids
displaying interesting activities were resynthesized using the
same protocol as for the library synthesis to confirm activity
and in some cases to get a physicochemical profile. Around
200 compounds were synthesized with a success rate of 50%,
meeting the above criteria.
To effectively explore the SAR around the biphenyl system,
libraries of compounds based on Negishi cross-coupling reac-
tions were synthesized. This provided compounds with different
alkyl chains and biphenyl systems (BB2, BB3, and BB4).
Electrophiles used in the Negishi reactions (E1-E4) that were
not commercially available were synthesized from the corre-
sponding carboxylic acids by two methods (Scheme 1).
The synthesis of 5-12 and 16 is outlined in Scheme 2. In
the synthesis of 5-12, a halogen-lithium exchange reaction
followed by transmetalation gave the required zinc reagents.
After the subsequent Negishi cross-coupling reactions, the crude
products were subjected to a basic ester hydrolysis under MW
irradiation conditions, giving the free acids (5-12, Scheme 2).
The synthesis of 16 included an O-alkylation step prior to the
Negishi reaction and the ester hydrolysis.
a (a) EtOH, H2SO4, MW, 170 °C, 1 min, 77%; (b) RX, K2CO3, KI,
CH3CN, MW, 180 °C, 25 min; (c) LiOH, H2O, THF, MW, 160 °C, 5 min.
(#) Resynthesis. (*) Yield after purification of 20 mg of crude material by
preparative LC/MS; X)halogen.
Scheme 4. Synthesis of 4′-(2-Butoxyethoxy)-3-fluorobiphenyl-
4-carboxylic Acid (18)a
a (a) 1-(2-Bromoethoxy)butane, Cs2CO3, DMF, MW, 180 °C, 25 min,
95%. (b) (i) t-BuLi, THF, -20 °C, 1 h; (ii) ZnBr2, THF, -20 °C to room
temp, 0.5 h; (iii) 2-(pyridin-2-yl)ethyl 4-bromo-2-fluorobenzoate, Pd2(dba)3,
tfp, THF/NMP 2:1, room temp, 16 h, 83%; (c) LiOH, H2O, THF, MW,
160 °C, 5 min, 95%.
halides. The subsequent MW assisted hydrolysis gave the
desired acids 15, 17, and 21.
The synthesis of 18 was based on the procedures previously
developed for the O-alkylations, Negishi cross-coupling reaction,
and basic hydrolysis and yielded the product in 75% yield over
the three steps (Scheme 4).
Conversion of commercially available heteroaromatic biaryls
into analogues of 3 was further investigated. Disappointingly,
only a limited number of compounds were available for the
purpose; e.g., 4-(5-heptyl-2-pyrimidinyl)benzonitrile was con-
verted into carboxylic acid 2221 in 85% yield by heating under
acidic conditions (Scheme 5).
A small library of alkoxythiazoles was synthesized by a two-
step protocol (Scheme 6). As previously observed for the
O-alkylation of ethyl 4′-hydroxybiphenyl-4-carboxylate (Scheme
Another route to obtain biphenyl analogues of 3 is depicted
in Scheme 3. The commercially available 4′-hydroxybiphenyl-
4-carboxylic acid was transformed into the corresponding ethyl
ester followed by O-alkylation with a set of different alkyl

1542 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 6
Lund et al.
Table 1. Activities at RARꢀ2 and RARꢀ1a
RARꢀ2
Scheme 5. Synthesis of 4-(5-Heptyl-2-pyrimidinyl)benzoic Acid
(22)
RARꢀ1
compd
Eff (%)
pEC50
Eff (%)
pEC50
Am-580
1 (ATRA)
4
3
13
14
19
20
100
7.7 ( 0.4
8.2 ( 0.4
7.6 ( 0.3
6.9 ( 0.4
7.8 ( 0.4
7.5 ( 1.0
7.2 ( 0.3
7.0 ( 0.2
103 ( 11
230 ( 13
nab
7.3 ( 0.4
7.6 ( 0.4
127 ( 10
64 ( 19
92 ( 24
100 ( 10
72 ( 44
62 ( 24
70 ( 21
29 ( 13
58 ( 11
nab
5.7 ( 0.1
6.5 ( 0.2
nab
Scheme 6. Synthesis of Alkoxythiazolesa
49 ( 15
6.0 ( 0.1
a AM-58012 was used as reference and set to 100% Eff. pEC50 and
efficacy values are the mean values of at least three experiments ( SD.
b na: no activity at pEC50 > 5.0. c Max efficacy at concentrations of <10
µM.
receptor, the nonyl analogue 13 also showed significant activity
at the RARꢀ1 isoform. Furhermore, 4′-alkoxybiphenyl ana-
logues (14, 19, and 20; see Figure 1) were also found to display
activity at the RARꢀ2, and with regard to both activity and
selectivity, the same trend was seen as for the 4′-alkylbiphenyl
analogues with six, seven, and eight carbon chains, 14, 19, and
20, respectively. Consequently, we decided to direct the
medicinal chemistry effort to mainly synthesizing compounds
having a 4′-alkyl or 4′-alkoxy chain length mimicking the chain
length of the heptyl and octyl biphenyl analogues. The main
focus of the medicinal chemistry effort was on the synthesis of
compounds with improved physicochemical properties while
retaining the isoform selectivity.
a (a) RX, K2CO3, CH3CN, MW, 180 °C, 25 min; (b) LiOH, H2O, THF,
MW, 160 °C, 5 min.
Scheme 7. Synthesis of a Fluorinated Alkoxythiazole 28a
Attention was drawn to the low solubility of 3. The crystal
packing and the lipophilic properties of the biphenyl structure
of 3 were likely to cause the low solubility of <0.001 mg/mL
at physiological pH 7.4. Therefore, we speculated that the
solubility would probably be improved by substituting at least
one of the phenyl groups in the biphenyl system with a
heteroaromatic ring. In addition, the introduction of heteroatoms,
substituents, or both in the biphenyl system might alter the
crystal packing through disruption of the symmetry, which also
could lead to improved solubility. This hypothesis was supported
by melting point changes indicative of solubility changes.
Introducing a fluorine substituent in 3 gave a slight decrease of
the melting point from 149-150 °C to 142-143 °C (8), and
introduction of a heteroaryl in the form of a thiazole (28) gave
a melting point of 109-110 °C. The modifications of the
aromatic system were obtained through focused libraries based
on Negishi cross-coupling reactions. In total, using all proce-
dures described, more than 200 compounds were synthesized
and tested in R-SAT for their activities at RARꢀ2, and the results
for representative compounds are listed in Table 2.
The in vitro RARꢀ2 data indicated that when BB2 was a
phenyl group, fluorine was the only reasonably tolerated
substituent ortho to the carboxylic acid functionality (see 5 and
8). The o-fluorine substituent led to increased activity with the
octyl 4′-chain (compare 3 and 8, pEC50 of 6.9 and 7.4,
respectively), whereas in contrast, the potency dropped with the
heptyl 4′-chain (compare 4 and 5, pEC50 of 7.6 and 6.9,
respectively). In addition with an o-methyl substituent (9) the
activity was considerably reduced. One might hypothesize that
the increase in potency in 8 could be due to the inductive
properties of fluorine which would lower the pKa (calculated
pKa of 3 was 4.2 and that of 8 was 3.3) of 8, thereby stabilizing
the ligand-receptor interaction by reinforcing the salt bridge.
However, inspection of several RAR cocrystallized complexes
shows that crystal water is situated about 2.8 Å from where the
fluorine atom of 8 would be positioned.23 This water molecule
is involved in the H-bonding network, which includes the
a (a) EtOH, EDCI·HCl, DIPEA, HOBt, CH3CN, room temp, 16 h, 93%;
(b) 2-mercaptopropanoic acid, pyridine, MW, 180 °C, 15 min, 67%; (c)
1-(2-bromoethoxy)butane, KI, Cs2CO3, CH3CN, MW, 180 °C, 25 min; (d)
LiOH, H2O, THF, MW, 160 °C, 5 min, 68% (two steps).
3), the O-alkylation of methyl 4-(4-hydroxy-5-methylthiazol-
2-yl)benzoate did not go to completion unless the reaction was
performed under forcing conditions using MW irradiation at
180 °C for 25 min. The second step consisted of a basic
hydrolysis, and overall products were obtained in good to
excellent yields (Scheme 6).
The synthesis of 2812 (Scheme 7) was based on a literature
procedure where the benzonitrile, 2-mercaptopropionic acid, and
pyridine were reacted to form a hydroxythiazole under con-
ventional heating.22 However, we found the reaction sluggish
using the reported conditions, while MW irradiation gave a
cleaner reaction. Subsequently, the hydroxythiazole was O-
alkylated under microwave irradiation conditions. It soon
became apparent that the purification of the alkyoxythiazoles
was hampered by the instability of the compounds toward
chromatography; however, purification by distillation, crystal-
lization, or both gave pure compounds. Kugelrohr distillation
of the crude reaction mixture was used to remove low boiling
byproducts. The residue was then subjected to an ester hydroly-
sis, and carboxylic acid 28 was obtained in 42% yield over four
steps.12
Results and Discussion
As previously reported, we discovered that biphenyl ana-
logues substituted with 4′-alkyl carbon chains displayed activity
at the RARꢀ2 (Table 1). From the in vitro pharmacology of
these analogues, which have 4′-alkyl chain lengths ranging from
two to nine carbons, it was apparent that the aliphatic chain
length had a profound effect on the activity and selectivity for
RARꢀ2. While heptyl, octyl, and nonyl analogues (4, 3, and
13, respectively) displayed full agonist activity at the RARꢀ2

Retinoic Acid Receptor ꢀ Ligands
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 6 1543
Table 2. Activities of RARꢀ2 Selective Agonistsa
RARꢀ2
compd
R1
Ar2
Ar1
Eff (%)
pEC50
4
5
6
7
3
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
CH3(CH2)6
CH3(CH2)6
CH3(CH2)6
CH3(CH2)6
CH3(CH2)7
CH3(CH2)7
CH3(CH2)7
CH3(CH2)7
CH3(CH2)7
CH3(CH2)7
CH3(CH2)8
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
2-F-C6H3
3-CH3-C6H3
2,5-thiophene
Ph
2-F-C6H3
2-CH3-C6H3
3-F-C6H3
3-CH3-C6H3
2,5-thiophene
Ph
Ph
Ph
Ph
Ph
2-F-C6H3
Ph
64 ( 19
50 ( 18
nab
7.6 ( 0.3
6.9 ( 0.2
nab
92 ( 24
114 ( 33
84 ( 15
85 ( 23
37 ( 22
nab
100 ( 10
72 ( 44
75 ( 29
41 ( 9
78 ( 19
108 ( 5
62 ( 24
70 ( 21
nab
6.9 ( 0.4
7.4 ( 0.2
6.2 ( 0.7
7.0 ( 0.4
7.0 ( 0.2
7.8 ( 0.4
7.5 ( 1.0
7.2 ( 0.2
6.6 ( 0.3
7.2 ( 0.2
7.7 ( 0.1
7.2 ( 0.3
7.0 ( 0.2
CH3(CH2)5O
(CH3)2CH(CH2)4O
CH3(CH2)3O(CH2)2O
CH3(CH2)3O(CH2)2O
CH3(CH2)3O(CH2)2O
CH3(CH2)6O
CH3(CH2)7O
CH3(CH2)5NH(CO)CH2O
CH3(CH2)6
2,5-(CH3)2-C6H2
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
pyrimidinyl
93 ( 21
6.0 ( 0.2
a AM-580 was used as reference and set to 100% Eff. pEC50 and efficacy values are the mean values of at least three experiments ( SD. b na: no activity
at pEC50 > 5.0.
Table 3. Activities of RARꢀ2 Selective Alkoxythiazolesa
ligand’s carboxylate, Ser280, and the backbone carbonyl group
of Leu224 and other water molecules. The fluorine would
occupy a very favorable position, which allows for H-bond
formation with the water, and hence does not require its
displacement. In the case of reduced potency incorporating an
o-fluorine in the heptyl 4′-chain analogue (5), 4 and 5 show
partial agonist efficacies of 64% and 50%, respectively. The
efficacy has been shown to be dependent on the length of the
4′-chain (Table 1). Therefore, a tighter interaction in the
carboxylic acid pharmacophore part would decrease an already
suboptimal interaction of the 4′-chain with the receptor R-helix
12.12 However, the presented rationale needs further clarification,
preferable in the form of crystal data.
RARꢀ2
compd
R1
R2
Eff (%)
pEC50
23
24
25
26
27
28
CH3(CH2)3O(CH2)2O
CH3(CH2)2O(CH2)2O
CH3CH2O(CH2)2O
CH3(CH2)5NH(CO)CH2O
CH3(CH2)6O
H
H
H
H
H
F
99 ( 30
100 ( 41
78 ( 5
85 ( 22
nab
7.1 ( 0.3
6.9 ( 0.2
7.2 ( 0.2
7.4 ( 0.3
When BB2 is a phenyl, substituents in the meta-position were
not as well tolerated as the o-fluoro substituent. This was
indicated by the reduced activity of compounds with fluorine
and methyl as meta-substituents (6, 10, and 11, respectively).
Surprisingly, when a thienyl was introduced as BB2 (7 and 12),
no activity was observed; in fact none of the compounds
containing a heteroaromatic group as BB2 were active.24
With regard to modifications in BB3, it appeared that the
activity at the RARꢀ2 was sensitive to steric hindrance as seen
when R1 is butoxyethoxy. This characteristic was observed for
16 having the 2,5-dimethyl substituted benzene ring (BB3), as
potency and efficacy dropped, compared to the unsubstituted
BB3 analogue 17. Although 22, pyrimidine (BB3), diplays full
agonist activity, the potency was reduced 50-fold compared with
4 at RARꢀ2.
The unsubstituted biphenyl analogue 17 with an oxygen atom
as a methylene substitute (R1 ) butoxyethoxy) was potent and
efficacious, which shows that the receptor tolerates heteroatoms
in the aliphatic chain part (BB4). In addition, branching of the
terminal end of the 4′-alkyl chain (15) provided a potent
derivative. When the activities of the heptyl (4-6) and the octyl
(3, 8, and 11) chain analogues are compared, it is apparent that
compounds with the octyl chain are more efficacious than the
heptyl analogues.
CH3(CH2)3O(CH2)2O
106 ( 26
8.1 ( 0.4
a AM-580 was used as reference and set to 100% Eff. pEC50 and efficacy
values are the mean values of at least three experiments ( SD. b na: no
activity at pEC50 > 5.0.
The encouraging RARꢀ2 agonist activity of 8 prompted us
to design and synthesize 18 which would combine the activity
enhancing property of 8 (2-F substituent) with the less lipophilic
chain of 17. Compound 18 did indeed show an increased
potency compared to 8 and 17, and the solubility in phosphate
buffer at pH 7.4 was significantly higher than that for 3 (18,
0.02 mg/mL; 3, <0.001 mg/mL).
However, the solubility data suggested that to synthesize
analogues of biphenyl 3 with dramatically altered physicochem-
ical properties, we had to modify the biphenyl moiety. A
5-methylthiazole unit was identified as an interesting BB3
building block, which could easily be synthesized from com-
mercially available starting materials (Schemes 6 and 7). The
in vitro activity at RARꢀ2 of the thiazole analogues 23-28
showed some interesting results (Table 3). Most surprisingly,
the alkoxythiazole containing a heptyloxy chain 27 displayed
no agonist activity which is in contrast to the corresponding
compound 19 (pEC50 ) 7.2, Eff ) 58%) in the alkoxybiphenyl

1544 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 6
Lund et al.
Table 4. Biological Activities at the RAR Subtypes and Isoforms for Selected Compoundsa
RARR
RARꢀ1
RARꢀ2
RARγ
compd
Eff (%)
pEC50
Eff (%)
pEC50
Eff (%)
pEC50
Eff (%)
pEC50
AM-580
1 (ATRA)
3
8
15
17
18
19
23
26
28
100 ( 4
121 ( 11
32 ( 18
65c
7.1 ( 0.4
8.2 ( 0.4
5.6 ( 0.0
5.7c
103 ( 11
230 ( 13
29 ( 13
54 ( 20
28 ( 30b
nad
7.3 ( 0.3
7.6 ( 0.4
5.7 ( 0.1
6.4 ( 0.1
100 ( 3
127 ( 10
92 ( 24
114 ( 33
75 ( 29
78 ( 19
108 ( 5
62 ( 24
99 ( 30
85 ( 22
106 ( 26
7.7 ( 0.4
8.2 ( 0.4
6.9 ( 0.4
7.4 ( 0.2
7.2 ( 0.2
7.2 ( 0.2
7.7 ( 0.1
7.2 ( 0.3
7.1 ( 0.3
7.4 ( 0.3
8.1 ( 0.4
99 ( 5
116 ( 5
31 ( 14b
84c
7.6 ( 0.6
7.3 ( 0.3
6.8c
nad
31 ( 25b
nad
nad
55c
5.8c
36 ( 10b
37 ( 22b
nad
nad
nad
45 ( 17
42 ( 20b
52 ( 16
nad
5.9 ( 0.2
6.4 ( 0.3
54 ( 16b
63 ( 23
48 ( 23
35 ( 25b
46 ( 25
6.2 ( 0.3
6.3 ( 0.6
6.2 ( 0.3
a AM-58012 was used as reference and set to 100% Eff. pEC50 and efficacy values are the mean values of at least three experiments ( SD. b Max efficacy
at concentrations of <10 µM. c Result of one experiment. d na: no activity at pEC50 > 5.0.
Table 5. Metabolic Stability and Solubility of Selected Compounds
series. Likewise the alkoxythiazole analogue of 15 showed 100
a
a
solubilityb
times less activity compared with 15 at the RARꢀ2.24 Benefi-
cially, by use of the parallel synthesis strategy, a number of
alkoxythiazole analogues had been prepared simultaneously;
otherwise, the negative result with the obvious analogue 27
would probably have discouraged us from further work on this
series. In contrast, while the biphenyl 21 showed no activity,
the thiazole analogue 26 is quite potent and selective at RARꢀ2.
Although the biphenyls and the thiazoles look similar, it is
apparent from the molecular modeling that the 4′-chains of the
two scaffolds occupy different areas in the receptor but converge
in the interaction with R-helix 12;12 therefore, it is not given
that the 4′-chains are interchangeable between the two scaffolds.
Four of the alkoxythiazoles 23-26 displayed comparable
activity to biphenyl 3. In these compounds the third sp3
hybridized methylene in the alkyl chain of 27 has been
exchanged with a heteroatom (O in 23-25 and NH in 26) which
slightly alters the preferred low energy conformations of the
4′-chain. This again clearly demonstrates the influence of this
part of the molecule on the activity. These compounds showed
that a substantial reduction of the lipophilicity from that of 3
(cLogP ) 7.9) could be achieved (23, 24, and 26 had cLogP
values of 5.0, 4.5, and 4.8, respectively). This prompted us to
further focus on the alkoxythiazoles; however, before we
embarked on the synthesis of additional alkoxythiazole deriva-
tives, the selectivity profile of the most active compounds was
investigated.
ClInt,human
ClInt,rat
compd (µL/(min·mg)) (µL/(min·mg)) (mg/mL) cLogD7.4 cLogP
3
18
28
5
8
36
5
0
10
<0.001 6.1 (4.5)c 7.7 (7.9)c
0.02
4.8
0.8
4.6
0.7 (1.4)c 5.2 (5.1)c
a The hepatic extraction was calculated without taking the fraction of
unbound drug in plasma into account. b The solubility was measured in
phosphate buffer, pH 7.4. c Experimental value.
and human microsomes was determined to be 10 and 36 µL/
(min·mg), respectively. In vitro clearance values of less than
75 and 50 µL/(min·mg) represent moderate to low clearance
compounds (less than 30% liver blood flow) in rat and human,
respectively. The solubility in phosphate buffer at pH 7.4 was
increased almost 250 times from 0.02 mg/mL for biphenyl 18
to 4.8 mg/mL for 28 by changing the BB3 from a phenyl to a
thiazole moiety (Table 5). Compound 28 showed good oral
bioavailability (F) of 52% with an in vivo clearance of 41 mL/
(min·kg) in rats.
Conclusion
A synthetic strategy involving parallel synthesis was suc-
cessfully applied to rapidly establish a SAR starting from a
unique isoform-selective RARꢀ2 agonist 3. An alkoxythiazole
series that retained the selectivity profile with improved phys-
icochemical properties was designed and synthesized. An orally
available potent RARꢀ2 agonist 28 was ultimately designed,
which exhibits promising druglike properties and represents an
excellent starting point for further lead optimization efforts. In
addition, the discovered retinoid ligands are potential research
tools that may shed light on the functions and therapeutic
potential of the RARꢀ2 isoform.
The compounds were tested on all of the RARs revealing a
good to excellent selectivity for the RARꢀ2 over RARꢀ1,
RARR, and RARγ (Table 4). The biphenyl analogues 8, 15,
17, and 18 were equally selective as 3. In addition, alkoxythia-
zoles 23 and 26 showed good selectivity profiles at all the tested
RARs. The selectivity profile of butoxyethoxy derivative 23 was
slightly superior to that of acetamide 26. On the basis of these
findings and the indication that a fluorine ortho to the carboxylic
acid on the phenyl ring in BB2 would enhance the activity (as
observed for 8 and 18), we synthesized 28 (Scheme 7). When
tested, the o-fluoro compound 28 displayed a significantly
improved activity at RARꢀ2 (Eff ) 106%, pEC50 ) 8.1)
compared to all synthesized alkoxythiazole analogues. The
increase in potency of o-fluoro substituted compounds was more
pronounced in the thiazole series as compared to the phenyl
series, about 10 and 3 times, respectively, also correlating with
the activity of the 2-F analogue of 24 (pEC50 ) 7.8).24 The
selectivity of 28 was also tested, and as expected, it was
Experimental Section
Reagents, starting materials, and solvents were purchased from
commercial suppliers and used as received. Spectroscopic data were
recorded on a Varian XL 400 MHz spectrometer. Purities were
determined by liquid chromatography/mass spectrometry performed
on a Waters/Micromass ZQ2000 LC/MS instrument consisting of
a ZQ single quadropole mass spectrometer equipped with an
electrospray ionization interface and on a Waters Alliance HT with
a 2795 separation module and 996 photodiode array detector
(190-450nm). Elemental analyses were done in the microanalytical
laboraties of Fakulta¨t fu¨r Chemie, Universita¨t Wien, Austria, and
all compounds had a purity of >95%. High-resolution mass
spectrometry was performed at the University of Lund, Sweden.
General Negishi Cross-Coupling Method: 2-(2-Pyridinyl)ethyl
4′-(2-Butoxyethoxy)-3-fluorobiphenyl-4-carboxylate. In a dry and
argon flushed vial 1-(2-butoxyethoxy)-4-iodobenzene (1.5 mmol,
480 mg) was taken up in dry THF (2.5 mL). The vial was cooled
12
comparable to that of 3, both with regard to the RAR and
RXR subtypes.12
The microsomal clearance and solubility of 3, 8, 18, and 28
were measured (Table 5). The in vitro clearance of 28 in rat

Retinoic Acid Receptor ꢀ Ligands
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 6 1545
retinoic acid recetor ꢀ2 promotes recovery of function after corti-
cospinal tract injury in the adult rat spinal cord. Hum. Mol. Genet.
2006, 21, 3107–3108.
to -20 °C before slow addition of t-BuLi in hexane (1.6 M, 3.00
mmol, 1.88 mL). After 1 h at -20 °C ZnBr2 (1.5 M, 1.65 mmol,
1.10 mL) was added and the cooling stopped. In another dry and
argon flushed vial Pd2dba3 (0.075 mmol, 70 mg) and tfp (1.0 mmol,
70 mg) were taken up in dry NMP (2.5 mL). The activated catalyst
was added to the zinc reagent via a syringe. Finally, 2-(2-
pyridinyl)ethyl 4-bromo-2-fluorobenzoate (1.0 mmol, 324 mg)
dissolved in dry THF (2.5 mL) was added to the mixture. The
mixture was left at room temperature for 16 h, then quenched with
NH4Cl (aq) and poured onto a HM-N SPE column, then extracted
with EtOAc. The extract was concentrated in vacuo. Purification
was by flash chromatography (heptane/EtOAc 4:1 to 3:1). Yield:
370 mg (84%). 1H NMR (300 MHz, CDCl3) δ 8.60-8.54 (m, 1H),
7.94-7.85 (m, 1H), 7.68-7.58 (m, 1H), 7.58-7.47 (m, 2H),
7.40-7.22 (m, 3H), 7.21-7.12 (m, 2H), 4.73 (t, J ) 6.7 Hz, 2H),
4.17 (t, J ) 4.2 Hz, 2H), 3.82 (t, J ) 4.7 Hz, 2H), 3.56 (t, J ) 6.6
Hz, 2H), 3.28 (t, J ) 6.4 Hz, 2H), 1.70-1.56 (m, 2H), 1.49-1.32
(m, 2H), 0.94 (t, J ) 7.3 Hz, 3H). 13C NMR (100 MHz, CD3OD)
δ 165.4 (d, JCF ) 257.0 Hz), 161.2, 159.3, 149.9 (2C), 149.0 (d,
JCF ) 9.0 Hz), 138.7 (2C), 133.5, 132.1 (d, JCF ) 1.9 Hz), 129.4,
125.4, 123.4, 122.9 (d, JCF ) 3.3 Hz), 117.2 (d, JCF ) 9.9 Hz),
116.2, 115.3 (d, JCF ) 23.3 Hz), 72.2, 70.3, 68.7, 65.4, 37.9, 32.8,
20.3, 14.2. Anal. Calcd for C26H28FNO4: C, 71.4; H, 6.5; N, 3.2.
Found: C, 70.8; H, 6.4; N, 3.1.
4′-(2-Butoxyethoxy)-3-fluoro[1,1′-biphenyl]-4-carboxylic Acid
(18). A MW vial was charged with 2-(2-pyridinyl)ethyl 4′-(2-
butoxyethoxy)-3-fluorobiphenyl-4-carboxylate (0.3 mmol, 131 mg),
THF (1.5 mL), and water (0.8 mL). Then LiOH ·H2O (0.9 mmol,
38 mg) was added and the mixture was microwave irradiated at
160 °C for 5 min. After cooling, the mixture was transferred to a
separation funnel with EtOAc and was washed with NaOH (1 M).
The aqueous phase was acidified with HCl (1 M) and extracted
with EtOAc. The organic phase was washed with brine, dried over
Na2SO4, filtered, and concentrated in vacuo, yielding the desired
product. Yield: 95 mg (95%). Mp 122.7-123.3 °C. 1H NMR (300
MHz, CDCl3) δ 10.59 (s, 1H), 8.10-7.99 (m, 1H), 7.60-7.49 (m,
2H), 7.46-7.38 (m, 1H), 7.38-7.28 (m, 1H), 7.07-6.96 (m, 2H),
4.19 (t, J ) 4.3 Hz, 2H), 3.84 (t, J ) 4.6 Hz, 2H), 3.58 (t, J ) 6.7
Hz, 2H), 1.71-1.56 (m, 2H), 1.50-1.33 (m, 2H), 0.95 (t, J ) 7.4
Hz, 3H). 13C NMR (75 MHz, CDCl3) δ 169.3 (d, JCF ) 3.5 Hz),
163.0 (d, JCF ) 261.0 Hz), 159.7, 148.4 (d, JCF ) 9.2 Hz), 133.2,
130.9 (d, JCF ) 1.6 Hz), 128.3 (2C), 122.0 (d, JCF ) 3.3 Hz), 115.2
(2C), 115.2 (d, JCF ) 9.2 Hz), 114.7 (d, JCF ) 23.1 Hz), 71.6,
69.2, 67.7, 31.8, 19.5, 14.2. Anal. Calcd for C19H21FO4: C, 68.7;
H, 6.4. Found: C, 68.4; H, 6.4.
(5) Krezel, W.; Ghyselinck, T. A.; Dupe´, V.; Kastner, P.; Borrelli, E.;
Chambon, P. Impaired Locomotion and Dopamine Signaling in
Retinoid Receptor Mutant Mice. Science 1998, 279, 863–867.
(6) Ge´hin, M.; Vivat, V.; Wurtz, J.-M.; Losson, R.; Chambon, P.; Moras,
D.; Gronemeyer, H. Structural Basis for Engineering of Retinoic Acid
Receptor Isotype-Selective Agonists and Antagonists. Chem. Biol.
1999, 6, 519–529.
(7) Peng, X.; Maruo, T.; Cao, Y.; Punj, V.; Mehta, R.; Das Gupta, T. K.;
Christov, K. A Novel RAR ꢀ Isoform Directed by a Distinct Promoter
P3 and Mediated by Retinoic Acid in Breast Cancer Cells. Cancer
Res. 2004, 64, 8911–8918.
(8) Mendelsohn, C.; Larkin, S.; Mark, M.; Lemeur, M.; Clifford, J.; Zelent,
A.; Chambon, P. RARꢀ Isoforms. Distinct Transcriptional Control
by Retinoic Acid and Specific Spatial Patterns of Promoter Activity
during Mouse Embryonic-Development. Mech. DeV. 1994, 45, 227–
241.
(9) Giguere, V.; Shago, M.; Zirngibl, R.; Tate, P.; Rossant, J.; Varmuza,
S. Identification of a Novel Isoform of the Retinoic Acid Receptor γ
Expressed in the Mouse Embryo. Mol. Cell. Biol. 1990, 10, 2335–
2340.
(10) Kastner, P.; Krust, A.; Mendelsohn, C.; Garnier, J. M.; Zelent, A.;
Leroy, P.; Staub, A.; Chambon, P. Murine Isoforms of Retinoic Acid
Receptor γ with Specific Patterns of Expression. Proc. Natl. Acad.
Sci. U.S.A. 1990, 87, 2700–2704.
(11) Sommer, K. M.; Chen, L. I.; Treuting, P. M.; Smith, L. T.; Swisshelm,
K. Elevated Retinoic Acid Receptor ꢀ4 Protein in Human Breast
Tumor Cells with Nuclear and Cytoplasmic Localization. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 8651–8656.
(12) Lund, B. W.; Piu, F.; Gauthier, N. K.; Eeg, A.; Currier, E.; Sherbukhin,
V.; Brann, M. R.; Hacksell, U.; Olsson, R. Discovery of a Potent,
Orally Available, and Isoform-Selective Retinoic Acid ꢀ2 Receptor
Agonist. J. Med. Chem. 2005, 48, 7517–7519.
(13) Piu, F.; Gauthier, N. K.; Olsson, R.; Currier, E. A.; Lund, B. W.;
Croston, G. E.; Hacksell, U.; Brann, M. R. Identification of Novel
Subtype Selective RAR Agonists. Biochem. Pharmacol. 2005, 71, 156–
162.
(14) Laudet, V.; Gronemeyer, H. The Nuclear Receptor Factsbook;
Academic Press: San Diego, CA, 2002.
(15) Mangelsdorf, D. J.; Thummel, C.; Beato, M.; Herrlich, P.; Schutz,
G.; Umesono, K.; Blumberg, B.; Kastner, P.; Mark, M.; Chambon,
P.; Evans, R. M. The Nuclear Receptor Superfamily: The Second
Decade. Cell 1995, 83, 835–839.
(16) Parks, D. J.; Blanchard, S. G.; Bledsoe, R. K.; Chandra, G.; Consler,
T. G.; Kliewer, S. A.; Stimmel, J. B.; Willson, T. M.; Zavacki, A.;
Moore, D. D.; Lehmann, J. M. Bile Acids: Natural Ligands for an
Orphan Nuclear Receptor. Science 1999, 284, 1365–1368.
(17) Staudinger, J. L.; Goodwin, B.; Jones, S. A.; Hawkins-Brown, D.;
MacKenzie, K. I.; LaTour, A.; Liu, Y.; Klaassen, C. D.; Brown, K. K.;
Reinhard, J.; Willson, T. M.; Koller, B. H.; Kliewer, S. A. The Nuclear
Receptor PXR is a Lithocholic Acid Sensor That Protects against Liver
Toxicity. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 3369–3374.
(18) Toma, S.; Isnardi, L.; Raffo, P.; Dastoli, G.; DeFrancisci, E.; Riccardi,
L.; Palumbo, R.; Bollag, W. Effects of All-trans Retinoic Acid and
13-cis Retinoic Acid on Breast-Cancer Cell Lines: Growth Inhibition
and Apoptosis Induction. Int. J. Cancer 1997, 70, 619–627.
(19) Liu, Y.; Lee, M. O.; Wang, H. G.; Li, Y.; Hashimoto, Y.; Klaus, M.;
Reed, J. C.; Zhang, X. Retinoic Acid Receptor ꢀ Mediates the Growth-
Inhibitory Effect of Retinoic Acid by Promoting Apoptosis in Human
Breast Cancer Cells. Mol. Cell. Biol. 1996, 16, 1138–1149.
(20) Shao, Z.; Dawson, M.; Li, X.; Rishi, A.; Sheikh, M.; Han, Q.; Ordonez,
J.; Shroot, B.; Fontana, J. p53 independent G0/G1 arrest and apoptosis
induced by a novel retinoid in human breast cancer cells. Oncogene
1995, 11, 493–504.
(21) Boller, A.; Cereghetti, M.; Scherrer, H. Phenylpyrimidine. De Patent
DE2547737, 1976; Hoffmann-La Roche.
(22) Kerdesky, F. A. J.; Holms, J. H.; Moore, J. L.; Bell, R. L.; Dyer,
R. D.; Carter, G. W.; Brooks, D. W. 4-Hydroxythiazole inhibitors of
5-lipoxygenase. J. Med. Chem. 1991, 34, 2158–2165.
(23) Ser280 and Leu224 refer to the RARꢀ binding site (RCSB-PDB code
1XAP, http://www.pdbbind.org cn/quickpdb.asp?quickpdb)1xap).
Germain, P.; Kammerer, S.; Perez, E.; Peluso-Iltis, C.; Tortolani, D.;
Zusi, F. C.; Starrett, J.; Lapointe, P.; Daris, J. P., Marinier, A., De
Lera, A. R.; Rochel, N.; Gronemeyer, H. Rational design of RAR-
selective ligands revealed by RARꢀ crystal structure. EMBO Rep.
2004, 5 877-882.
(24) Lund, B. W. Design, Synthesis and Structure-Activity Analysis of
Isoform-Selective Retinoic Acid Receptor ꢀ Ligands. Ph.D. Thesis,
University of Copenhagen, Denmark, 2007.
Acknowledgment. We thank Agnete Dyssegaard, Sine
Mandrup Bertozzi, Susanna Henriksson, and Tina Gustafsson
for expert analytical services.
Supporting Information Available: General experimental
methods, biological assays, the synthesis of 5-27, their analytical
data, and spectra. This material is available free of charge via the
Internet at http://pubs.acs.org.
References
(1) Camacho, L. H. Clinical Applications of Retinoids in Cancer Medicine.
J. Biol. Regul. Homeostatic Agents 2003, 17, 98–114.
(2) Orfanos, C. E.; Zouboulis, C. C.; Almond, R. B.; Geilen, C. C. Current
Use and Future Potential Role of Retinoids in Dermatology. Drugs
1997, 53, 358–388.
(3) (a) Maden, M. Retinoic Acid in the Development, Regeneration and
Maintenance of the Nervous System. Nat. ReV. Neurosci 2007, 8, 755–
765. (b) Swift, C. B.; Hays, J. L.; Petty, J. W. Distinct Functions of
Retinoic Acid Rcceptor Beta Isoforms: Implications for Targeted
Therapy. Endocr., Metab. Immune Disord.: Drug Targets 2008, 8,
47–50.
(4) (a) Corcoran, J.; So, P. L.; Barber, R. D.; Vincent, K. J.; Mazarakis,
N. D.; Mitrophanous, K. A.; Kingsman, S. M.; Maden, M. Retinoic
Acid Receptor ꢀ2 and Neurite Outgrowth in the Adult Mouse Spinal
Cord in Vitro. J. Cell Sci. 2002, 115, 3779–3786. (b) Yip, P. K.; Wong,
L.-F.; Pattinson, D.; Battaglia, A.; Grist, J.; Bradbury, E. J.; Maden,
M.; McMahon, S. B.; Mazarakis, N. D. Lentiviral vector expressing
JM801532E