Design and synthesis of novel RXR-selective modulators with improved pharmacological profile

Article abstract of DOI:10.1016/j.bmcl.2003.08.048

New RXR-selective modulators possessing a 6-fluoro trienoic acid moiety (6Z olefin) or a fluorinated/heterocyclic-substituted benzene core ring, were synthesized in an expedient and selective way. A subset of these compounds was evaluated for their metabo

Full text of DOI:10.1016/j.bmcl.2003.08.048

Bioorganic & Medicinal Chemistry Letters 13 (2003) 4071–4075
Design and Synthesis of Novel RXR-Selective Modulators with
Improved Pharmacological Profile
Pierre-Yves Michellys,a,*,y Marcus F. Boehm,a,{ Jyun-Hung Chen,a Timothy A. Grese,b
Donald S. Karanewsky,a,y Mark D. Leibowitz,c Sha Liu,c Dale A. Mais,d
Christopher M. Mapes,a,x Anne Reifel-Miller,e Katheen M. Ogilvie,e,k Deepa Rungta,d
Anthony W. Thompson,a,{ John S. Tyhonas,a Nathan Yumibee and Robert J. Ardeckya
aDepartment of Medicinal Chemistry, Ligand Pharmaceuticals, Incorporated, 10275 Science Center Drive, San Diego, CA 92121, USA
bDiscovery Chemistry Research, Lilly Research Laboratories, Indianapolis, IN, USA
cDepartment of Pharmacology, Ligand Pharmaceuticals, Incorporated, 10275 Science Center Drive, San Diego, CA 92121, USA
dDepartment of New Leads Discovery, Ligand Pharmaceuticals, Incorporated, 10275 Science Center Drive, San Diego, CA 92121, USA
eDivision of Endocrine Research, Lilly Research Laboratories, Indianapolis, IN, USA
Received 31 March 2003; accepted 18 August 2003
Abstract—New RXR-selective modulators possessing a 6-fluoro trienoic acid moiety (6Z olefin) or a fluorinated/heterocyclic-sub-
stituted benzene core ring, were synthesized in an expedient and selective way. A subset of these compounds was evaluated for their
metabolic properties (exposure in IRC male mice) and show a dramatic increase of exposure compared to our reference compound,
3 (LG101506).
# 2003 Elsevier Ltd. All rights reserved.
RXR/PPAR heterodimers play a major role in the reg-
ulation of both glucose (RXR/PPARg) and lipid (RXR/
PPARa) metabolism.1 Recent reports have shown that
RARs and RXR/LXR heterodimers while RXR ago-
nists like 1 are not. Because of this selective profile
against various RXR heterodimers, we have called
compounds like 3 RXR modulators. Moreover, we also
have shown that 3 can decrease plasma glucose levels by
the same magnitude as seen for BRL49653 (rosiglita-
zone1) in the db/db mouse model3 while displaying a
distinct side-effect profile in a mouse and a rat model
compared to classic RXR agonists like 1.3 3 possesses a
(2E,4E,6Z) trienoic acid moiety coupled to a 3,5-di-tert-
butyl-6-(2,2-difluoroethoxy) phenyl core. Historical
studies have established the critical role that the trienoic
6Z olefin geometry plays in determining the selectivity
of these compounds for RXR over RAR.4 SAR studies
have demonstrated that a fluorinated alkoxy side chain
improved the in vivo activity of this type of molecules.3
Here, we report the synthesis of new RXR-selective
modulators possessing fluorinated or heterocyclic sub-
stitutions on the 3- and 5-positions of the core benzene
ring as well as on the potentially labile 6 olefin of the
trienoic acid moiety.
synthetic RXR agonist (rexinoids) such as
1
(LG100268) and 2 (AGN194204) (Fig. 1) are insulin
sensitizing agents. They bind to RXR and functionally
stabilize the RXR/PPAR heterodimers. These com-
pounds exhibit the capacity to control both hyperglyce-
mia and hyperinsulinaemia.2 Unfortunately the rexinoid
class of RXR agonists also produce various undesirable
side effects. The most serious of these adverse effects is a
severe increase of triglyceride levels and a suppression
of the thyroid hormone axis2b that prohibits their use
for chronic therapy of type 2 diabetes. We have pre-
viously demonstrated that 3 is capable of selectively
activating the RXR/PPAR heterodimers versus RXR/
*Corresponding author. E-mail: pmichellys@gnf.org
yCurrent address: Genomic Institute of the Novartis Foundation, San
Diego, CA, USA.
{Current address: Conforma Therapeutics, San Diego, CA, USA.
xCurrent address: Johnson and Johnson, San Diego, CA, USA.
kCurrent address: Pfizer Inc, San Diego, CA, USA.
{Current address: Roche Bioscience, Palo Alto, CA, USA.
The derivatives 22, 23a,b and 24 possessing a fluorine in
the 6-position of the trienoic acid moiety (60-position,
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.08.048

4072
P.-Y. Michellys et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4071–4075
Figure 1. Selected RXR-selective scaffold.
Fig. 1) were synthesized in a selective and straightfor-
ward manner. The 3-fluorocoumarins 7–9 were readily
prepared from the acetylphenols 4–6 using 2.5 equiv of
triethyl-2-fluoro-2-phosphonoacetate and NaH in DMF
at 60 ꢀC in good to moderate yield. Interestingly, the
yield of the reaction decreases dramatically when the
substitution of the benzene ring increases (iso-pro-
pyl>tert-butyl>tert-amyl). Reduction of the 3-fluor-
ocoumarins 7–9 appeared to be delicate. None of the
alcohol 10–12 was isolated when 7–9 was treated with
common reductive agents like LAH, NaAlH4 or Dibal-
H. Instead, the only products observed were the over-
reduced phenolic alcohols. However, when treated with
a large excess of NaBH4 in methanol at 0 ꢀC, the cou-
marins 7–9 reduced smoothly into the desired phenolic
alcohols 10–12 in good yields (>70%) except for 9 that
produced 12 in 35% yield. The phenol present in 10–12
was then selectively alkylated using standard conditions
(Cs2CO3, DMF, rt). The remaining allylic alcohols were
then oxidized into the aldehydes 16–18 with TPAP/
NMO without isomerization of the double bond. Then,
phonocrotonate in THF–DMPU, ꢁ78 ꢀC) to afford the
corresponding esters 19–21 in excellent yields (>95%).
Hydrolysis of 19–21 using 2 N aqueous LiOH in reflux-
ing THF/MeOH afforded the crude acids 22–24.
Recrystallization of crude 22–24 from CH3CN released
the isomerically pure 60-fluorotrienoic acids in excellent
purity (>99%). Scheme 2 describes the introduction of
various aromatic groups and a pentafluoroethyl moiety
in the 3-position of the core aromatic ring (para position
of the alkoxy side chain, Fig. 1). We chose to introduce
the desired groups on a coumarin-based substrate for
convenience of synthesis. Fluoro-aromatic or hetero-
aromatic groups were introduced using Suzuki coupling
while the pentafluoroethyl group was introduced using
copper chemistry on the coumarin 28 (easily synthesized
from the commercially available 2-hydroxy-3-iso-propyl
benzoic acid 25). 29a–f were synthesized in good yield
by treatment of 28 with various commercially available
boronic acids in the presence of Pd(PPh3)4 (5%) in
refluxing toluene/ethanol. Introduction of the penta-
fluoroethyl group was realized using condensation of
pentafluoroethyl iodide and 28 in the presence of CuBr
in DMF at 60 ꢀC in a pressure tube.5 When subjected to
these conditions; the corresponding coumarin 35 was
isolated in 80% yield. The remaining steps of the
16–18 were subjected to
a
Horner–Wadsworth–
Emmons reaction with the anion of triethyl-3-methyl-
phosphonocrotonate (generated from slow addition of
n-BuLi to
a
solution of triethyl-3-methylphos-
Scheme 1. Reagents: (a) 2-fluorotriethylphosphono acetate, NaH, DMF, 60 ꢀC; (b) NaBH4, EtOH, rt; (c) Cs2CO3, R1-Br, DMF, rt; (d) TPAP,
NMO, CH2Cl2; (e) triethyl-3-methyl-phosphonocrotonate, n-BuLi, THF–DMPU, ꢁ78 ꢀC; (f) LiOH, THF–MeOH, reflux then HCl and recrystalli-
zation from CH3CN.

P.-Y. Michellys et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4071–4075
4073
Scheme 2. Reagents: (a) MeLi, THF, 0 ꢀC; (b) NIS, TsOH (cat.), CH2Cl2, reflux; (c) ethyltriphenyl phosphoranilidene acetate, toluene, reflux; (d)
ArB(OH)2, Pd(PPh3)4, toluene–ethanol, 2 N aq Na2CO3, reflux; (e) NaAlH4, Et2O, 0 ꢀC; (f) Cs2CO3, CHF2CH2Br, DMF, rt; (g) TPAP, NMO,
CH2Cl2; (h) triethyl-3-methyl-phosphonocrotonate, n-BuLi, THF–DMPU, ꢁ78 ꢀC; (i) LiOH, THF–MeOH, reflux then HCl and recrystallization
from CH3CN; (j) CF3CF2I, CuBr, DMF, 60 ꢀC, sealed tube; (k) Cs2CO3, R1-Br, DMF, rt.
synthesis of 34a–f and 42a–c follows exactly the same
synthetic route described in Scheme 1. The final trienoic
acids 34a–f and 42a–c were recrystallized from CH3CN
and isolated in excellent isomeric purity (>99%).
acterizing activity at both RXR/PPARg or RXR/RAR
heterodimers7 has been used to characterize the com-
pounds described in this communication (22, 23a,b, 24,
34a–f, 42a–c, 56, and 57). The results are summarized in
Table 2.
Scheme 3 describes the synthesis of analogues posses-
sing similar functional groups in the 5-position of the
core aromatic moiety (ortho position of the alkoxy side
chain, Fig. 1). Again, the iodo-coumarin (46 and 47)
route proved to be very useful and allowed a quick and
efficient synthesis of compounds 56 and 57.
All the compounds from Table 1 show very good selec-
tivity for RXRs over RARs. They bind with high affi-
nity (Ki<45 nM) to all RXR receptor sub-types, and
show only weak RARa, RARb or RARg binding
(Table 1). In some cases, a trend of selectivity for a
specific RXR sub-type could be noted. For example, 22,
23a, 23b, 34a, and 34c (Table 1, entries 3, 4, 5, 7, and 9)
bind better to the RXRa receptor over RXRb and
RXRg, 23c and 42c show better binding to RXRb
(Table 1, entries 6 and 15) while 34d has the same
binding affinity for all three RXR sub-types (Table 1,
entry 10). Table 2 shows co-transfection data for the
compounds 22, 23a,b, 24, 34a–f, 42a–c, 56, and 57. All
the compounds exhibit RXR homodimer antagonist
activity (except 42a, Table 2, entry 13) that has little
RXR agonist activity. Previous SAR on similar struc-
tures possessing a 6-substituted trienoic acid moiety has
shown that the RXR activity of such compounds can be
altered by elongation of the 6-alkoxy side chain.3 As this
side chain gets bulkier, RXR homodimer agonist activ-
ity decreases and the RXR homodimer antagonist
activity increases. While the same paradigm applies to
the compounds presented in this communication (data
The competitive binding to RXRa, b and g and RARa,
b and g of each compound has been characterized by
using 3[H]-9-cis-retinoic acid for RXRs and 3[H]-all-
trans-retinoic acid for RARs (shown as Ki, Table 1).
The RXRa transcriptional activation profile of each
compound was determined in CV-1 cells (the AOX
response element used as the reporter).6 For RXR, the
efficacy is measured relative to LGD1069 (Table 2). We
have already reported that 1 efficiently activates the
RXR/PPARg and RXR/RAR heterodimers alone or in
combination (synergy assay) with 100 nM of BRL49653
(RXR/PPARg synergy assay) or 3 nM of TTNPB
(RXR/RAR synergy assay).3,7 In the case of structures
like 3, the activation of the RXR:PPARg heterodimer is
of much lower amplitude and generally the use of a
PPARg agonist (e.g., BRL49653) amplifies the response
of the modulator alone. The same protocol for char-

4074
P.-Y. Michellys et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4071–4075
Scheme 3. Reagents: (a) NIS, TsOH (10%), CH2Cl2, rt; (b) n-BuLi, THF, then N,N-dimethylacetamide, ꢁ78 ꢀC to rt; (c) NIS, TsOH (10%),
CH2Cl2, reflux. Steps (d), (e), (f), (g), (h), (i), and (j) identical to Scheme 2.
Table 1. Binding data of RXR modulators (CV-1 cells). Ki calculated using [3H]-9-cis-RA for RXR and [3H]-ATRA for RAR6 (all data shown
in nM)
Entries
Compd
RXRa (Ki)
RXRb (Ki)
RXRg (Ki)
RARa (Ki)
RARb (Ki)
RARg (Ki)
1
2
3
4
5
6
7
8
1
3
22
23a
23b
24
34a
34b
34c
34d
34e
34f
42a
42b
42c
56
12.9
1.5
75(17)
4
12
84(6.0)
1550
54(8.6)
3730
8300
4663
6592
6756
>10,000
4555
>10,000
NT
4555
>10,000
2500
2750
9901
9513
4623
5681
6008
5325
3927
3597
9901
9513
4.4ꢂ2.1
7.8ꢂ1.2
4.1ꢂ2.4
28.7ꢂ1.3
2.3ꢂ1.5
1.0ꢂ0.2
0.8ꢂ0.2
2.6ꢂ1.2
3.0ꢂ1.3
0.8ꢂ0.3
6.3ꢂ2.1
25.6ꢂ15.2
21.4ꢂ12.3
6.1ꢂ4.2
85.8ꢂ18.3
24.4ꢂ12.6
11.5ꢂ2.3
15.5ꢂ3.1
5.2ꢂ2.3
16.0ꢂ5.3
7.4ꢂ3.2
3.9ꢂ2.1
2.4ꢂ1.4
NT
3.9ꢂ1.5
8.5ꢂ3.1
17.2ꢂ3.1
8.2ꢂ1.3
12.3ꢂ6.5
9.9ꢂ2.5
27.8ꢂ14.1
13.2ꢂ2.6
12.4ꢂ3.5
40.3ꢂ4.5
10.9ꢂ4.1
2.9ꢂ0.2
5.6ꢂ2.2
2.8ꢂ1.5
NT
5.6ꢂ2.4
7.7ꢂ2.5
27.6ꢂ10.3
45.7ꢂ14.2
22.5ꢂ10.3
23.6ꢂ5.6
>10,000
>10,000
>10,000
>10,000
8764
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
>10,000
9
2112
1956
NT
10
11
12
13
14
15
16
17
2112
>10,000
1984
2270
2612
>10,000
4659
>10,000
57
not shown), these are close analogues to our lead com-
pound 3, and the in vitro profile of compound 3 was
used as reference for their profiling and only RXR
homodimer antagonists are shown in Table 2. All the
compounds described in Table 2 strongly activate the
RXR/PPARg heterodimer in combination with
BRL46953. Some compounds like 34a, 34b and 42a
have remarkable sub-nanomolar EC50s in this assay. No
RXR/RAR synergy was observed when tested in com-
bination with 3 nM TTNPB except for 38a (Table 1).
This compound possesses a partial agonist profile,
explaining the synergy observed for RXR/RAR. Pre-
vious experimental results have already shown that
RXR/RAR synergy was generally correlated with the
degree of RXR agonist activity observed with the tested
compounds. We have described extensively in a pre-
vious communication the pharmacological relevance of
compounds possessing the in vitro profile of 3.3 Here,
all the compounds described have an in vitro profile as
good as or better than the reference compound 3.
Previous studies in our laboratories (unpublished data)
have shown that molecules like 3 possessing a trienoic
acid moiety have sometimes low PK exposure and the
use of the Na salt was generally required to ensure a
minimal exposure. We have evaluated the PK exposure
of a subset of compounds described here in IRC mice.
Each compound was dosed as the free acid in CMC/
SLS/Povidone at 30 mg/kg (single dose). Table 2 shows
the AUC results obtained with compounds 20a, 31a and
37a compared to 3 as well as their respective Tmax (data
given in h) and Cmax (expressed in mg/mL of plasma or

P.-Y. Michellys et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4071–4075
Table 2. In vitro evaluation of RXR modulators (CV-1 cells)a
4075
Entries
Compds
RXRa
Agonist Eff
RXRa
Agonist EC50
RXRa
Antagonist Eff
RXRa
Antagonist IC50
RXRa/PPARg
Agonist Eff (EC50
RXRa/PPARg
RARg
Agonist Syn.
)
Agonist (EC50
)
1
2
3
4
5
6
7
8
1
2
22
23a
23b
24
34a
34b
34c
34d
34e
34f
42a
42b
42c
56
75ꢂ5
17ꢂ3
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
92ꢂ15
NC
NC
NC
NC
12
161ꢂ25
54ꢂ12
47ꢂ15
63ꢂ18
30ꢂ11
51ꢂ14
46ꢂ17
36ꢂ9
18.6ꢂ5.1
8.6ꢂ3.2
4.1ꢂ2.2
6.4ꢂ5.3
12.2ꢂ5.0
2.7ꢂ1.9
0.4ꢂ0.2
0.8ꢂ0.5
4.7ꢂ2.4
2.7ꢂ1.2
7.4ꢂ5.6
1.3ꢂ1.0
0.7ꢂ0.4
2.7ꢂ1.9
7.4ꢂ4.3
16.3ꢂ16.3
15.1ꢂ9.4
6.9
1.9
1.5
1.8
1.5
1.5
1.5
1.6
1.4
0.8
1.2
1.1
2.8
1.4
1.2
1.0
1.0
4
84ꢂ12
92ꢂ10
87ꢂ8
6.0ꢂ2.1
4.7ꢂ1.6
6.8ꢂ2.3
7.2ꢂ2.5
13.7ꢂ3.0
3.1ꢂ1.0
3.9ꢂ1.1
1.9ꢂ0.5
4.6ꢂ1.5
3.6ꢂ1.5
2.2ꢂ1.2
5.3ꢂ2.0
5.1ꢂ2.2
4.0ꢂ1.2
57.9ꢂ8.3
10.7ꢂ6.4
1
2
1
2
1
1
3
2
92ꢂ11
90ꢂ11
91ꢂ9
91ꢂ11
91ꢂ14
92ꢂ9
9
15ꢂ5
10
11
12
13
14
15
16
17
40ꢂ11
37ꢂ18
32ꢂ8
3
2
89ꢂ6
94ꢂ13
39ꢂ8
44ꢂ13
99ꢂ13
59ꢂ11
71ꢂ14
78ꢂ12
29ꢂ11
3
1
5
5
91ꢂ12
93ꢂ14
83ꢂ15
80ꢂ11
57
aEfficacy for RXRa activity was measured against LGD1069.6 RXR/PPARg synergy mode calculated using 100 nM BRL49653, efficacy relative to
BRL49653. RXR/RAR synergy calculated using 3 nM TTNPB, fold elevation over DMSO background7 (all data shown in nM).
Table 3. Oral exposure study of selected RXR modulators in male ICR mousea
Compd
Oral AUC (0–8 h)
(mg ꢃ h/mL)
Tmax (h)
Cmax
(mg/mL)
(mM)
3
3.8ꢂ1.20
17.7ꢂ4.95
15.17ꢂ2.97
14.41ꢂ3.35
1
3
1
1
0.76ꢂ0.46
3.69ꢂ1.89
4.14ꢂ0.51
4.43ꢂ2.82
1.81ꢂ1.09
8.41ꢂ4.31
9.31ꢂ1.15
10.24ꢂ6.52
23a
34a
42a
aDose formulation of the free acid in CMC/SLS/Povidone (30 mg/kg). Timepoints: 1, 3, and 8 h (serial sacrifice, n=3/timepoint).
2. (a) Schulman, I. G.; Crombie, D.; Bissonnette, R. P.;
Cesario, R.; Roegner, K.; Shao, G.; Heyman, R. A. Handbook
of Experimental Pharmacology; Springer: Berlin, 1999; Vol.
139, p 215, and references therein. (b) Stumvoll, M.; Haring,
H. Horm. Res. 2001, 55, 3. (c) Mukherkjee, R.; Davies, P. J. A.;
Crombie, D. L.; Bischoff, E. D.; Cesario, R. M.; Jow, L.;
Hamann, L. G.; Boehm, M. F.; Mondon, C. E.; Nazdan,
A. M.; Paterniti, J. R.; Heyman, R. A. Nature 1997, 386, 407.
3. Michellys, P. Y.; Ardecky, R. J.; Chen, J. H.; Crombie,
D. L.; Etgen, G.; Faul, M.; Faulkner, A. L.; Grese, T. A.;
Heyman, R. A.; Karanewsky, D. S.; Klausing, K.; Leibowitz,
M. D.; Liu, S.; Mais, D. A.; Mapes, C. M.; Marschke, K.;
Reifel-Miller, A.; Ogilvie, K. M.; Rungta, D.; Thompson,
A. W.; Tyhonas, J. S.; Boehm, M. F. J. Med. Chem. 2003, 46,
2683.
mM). All the compounds displayed in Table 2 show a
dramatic increase in exposure (>3 times) and compound
20a has a longer half-life (3 h instead of 1 h for 3).
In conclusion, we have expediently synthesized a new
series of RXR-selective modulators possessing a fluori-
nated trienoic moiety or various fluorinated, aromatic
or heteroaromatic-substituted core ring systems from
inexpensive starting materials. All the compounds
selectively bind with high affinity to RXRs versus
RARs. They also possess little or no RXR/RAR
synergy (<2-fold) and activate the RXR/PPARg het-
erodimer when used in combination with a PPARg
agonist (BRL49653). The PK profiles (AUC calcula-
tion) for a subset of this new series of compounds have
been examined in IRC mice. The exposure data pre-
sented in Table 3 show a dramatic exposure increase
compared to our lead compound 3.
4. Zhang, L.; Nazdan, A. M.; Heyman, R. A.; Love, D. L.;
Mais, D. E.; Croston, G. E.; Lamph, W. W.; Boehm, M. F. J.
Med. Chem. 1996, 39, 2659.
5. Carini, D. J.; Chiu, A. T.; Wong, P. C.; Johnson, A. L.;
Wexler, R. R.; Timmermans, L. Bioorg. Med. Chem. Lett.
1993, 3, 895.
6. Boehm, M. F.; Zhang, L.; Badea, B. A.; White, S. K.;
Mais, D. A.; Berger, E.; Suto, C. M.; Goldman, M. E.; Hey-
man, R. A. J. Med. Chem. 1994, 37, 2930.
References and Notes
7. Michellys, P.-Y.; Ardecky, R. J.; Chen, J.-H.; D’Arrigo, J.;
Grese, T. A.; Karanewsky, D. S.; Leibowitz, M. D.; Liu, S.;
Mais, D. A.; Mapes, C. M.; Montrose-Rafizadeh, C.; Ogilvie,
K. M.; Reifel-Miller, A.; Rungta, D.; Thompson, A. W.;
Tyhonas, J. S.; Boehm, M. F. J. Med. Chem. 2003, 46, 4087.
1. (a) Willson, T. M.; Brown, P. J.; Sternbach, D. D.; Henke,
B. R. J. Med. Chem. 2000, 44, 527 and references therein. (b)
Lehnard, J. M. Recept. Channels 2001, 7, 249. (c) Jones, A. B.
Med. Res. Rev. 2001, 21, 540. (d) Isserman, I.; Green, S. Nat-
ure 1990, 347, 645.