Structural development of tetrachlorophthalimides as liver X receptor β (LXRβ)-selective agonists with improved aqueous solubility

Article abstract of DOI:10.1016/j.bmcl.2017.12.024

LXRβ-selective agonists are promising candidates to improve atherosclerosis without increasing plasma or hepatic TG levels. We have reported a series of tetrachlorophthalimide analogs as an LXRβ-selective agonist. However, they exhibited poor aqueous solubility probably due to its high hydrophobicity and highly rigid and plane structure. In this report, we present further structural development of tetrachloro(styrylphenyl)phthalimides as the LXRβ-selective agonists with improved aqueous solubility.

Full text of DOI:10.1016/j.bmcl.2017.12.024

Accepted Manuscript
Structural Development of Tetrachlorophthalimides as Liver X Receptor β
(LXRβ)-Selective Agonists with Improved Aqueous Solubility
Sayaka Nomura, Kaori Endo-Umeda, Shinya Fujii, Makoto Makishima, Yuichi
Hashimoto, Minoru Ishikawa
PII:
S0960-894X(17)31184-8
https://doi.org/10.1016/j.bmcl.2017.12.024
BMCL 25480
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
21 November 2017
8 December 2017
11 December 2017
Please cite this article as: Nomura, S., Endo-Umeda, K., Fujii, S., Makishima, M., Hashimoto, Y., Ishikawa, M.,
Structural Development of Tetrachlorophthalimides as Liver X Receptor β (LXRβ)-Selective Agonists with
Improved Aqueous Solubility, Bioorganic & Medicinal Chemistry Letters (2017), doi: https://doi.org/10.1016/
j.bmcl.2017.12.024
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Structural Development of Tetrachlorophthalimides as Liver X Receptor β
(LXRβ)-Selective Agonists with Improved Aqueous Solubility
Sayaka Nomuraa, Kaori Endo-Umedab, Shinya Fujiia, Makoto Makishimab, Yuichi
Hashimotoa and Minoru Ishikawaa*
a Institute of Molecular and Cellular Biosciences, The University of Tokyo, 1-1-1 Yayoi,
Bunkyo-ku, Tokyo 113-0032, Japan
b Nihon University School of Medicine, 30-1 Oyaguchi-kamicho, Itabashi-ku, Tokyo
173-8610, Japan
Corresponding author: Minoru ISHIKAWA
Institute of Molecular and Cellular Biosciences, The University of Tokyo
1-1-1, Yayoi, Bunkyo-ku, Tokyo, 113-0032, Japan
Tel.: +81 3 5841 7853; fax: +81 3 5841 8495
E-mail address: m-ishikawa@iam.u-tokyo.ac.jp
1

Table of Contents
Abstract
LXRβ-selective agonists are promising candidates to improve atherosclerosis without
increasing plasma or hepatic TG levels. We have reported a series of
tetrachlorophthalimide analogs as an LXRβ-selective agonist. However, they exhibited
poor aqueous solubility probably due to its high hydrophobicity and highly rigid and
plane structure. In this report, we present further structural development of
tetrachloro(styrylphenyl)phthalimides as the LXRβ-selective agonists with improved
aqueous solubility.
2

Liver X receptors (LXRs) are members of the nuclear receptor (NR) superfamily,1,2 and
ligand-dependent transcription factors. The physiological ligands for LXRα/β are
oxysterols, including 22(R)-hydroxycholesterol (1) and 24(S),25-epoxycholesterol (2)
(Figure 1).3 Upon binding of an agonist to the ligand-binding domain (LBD) of LXR,
gene transcription occurs. The products of LXR-regulated genes, such as ABCA1,
ABCG1, ABCG5, ABCG8, ApoE and GLUT44-6 are involved in lipid metabolism,
reverse cholesterol transport,7 and glucose transport, so LXRs are considered to be
potential drug targets for atherosclerosis, hyperlipidemia, and metabolic syndrome.8
However, LXRs agonists also induce genes involved in lipogenesis, such as SREBP-1c
(sterol regulatory binding element protein 1c)9 and FAS (fatty acid synthase), 8 resulting
in increased plasma and hepatic triglyceride levels,10 which in turn might lead to fatty
liver and atherosclerosis as possible side effects.
3

Figure 1. Chemical structures of LXR agonists
LXRs include two subtypes with different tissue distribution, LXRα and LXRβ. LXRα
is highly expressed in liver, intestine and macrophages, while LXRβ has a more
widespread pattern of expression, being almost ubiquitous. LXRα contributes to
lipogenesis in liver, while selective LXRβ activation improves RCT in LXRα-knockout
mouse.11,12 Therefore, LXRβ-selective agonists are expected to improve atherosclerosis
via induction of RCT and cholesterol efflux from liver, without increasing plasma or
hepatic TG levels. However, LXRα and LXRβ are highly related and share 78% amino
acid sequence identity in the ligand-binding domains (LBDs), especially in the vicinity
of the ligand-binding pocket.13 Consequently, most LXR ligands, including T0901317
4

(3)1 and GW3965 (4)14,15 do not show subtype selectivity.
To date, a few LXRβ-selective agonists 5-8 have been reported (Figure 1).16 During our
continual research of LXR ligands,17 we have also found that 9 exhibited >100-fold
selective LXRβ agonistic activity in a full-length LXRβ reporter gene assay system.18
Compound 9 showed high selectivity over other NRs, and induced only ABCA1 mRNA
expression but not SREBP-1c mRNA expression. However, 9 exhibited poor aqueous
solubility. In this report, we present further structural development of
tetrachloro(styrylphenyl)phthalimides as the LXRβ-selective agonists with improved
aqueous solubility.
5

O
N
Ar
H
N
NO2
Br
NO2
PPh3
a
a
b
c
d
e
14a-g
S
S
b
Br
N
N
R
f
11
10
NO2
a
16a-h
b
g
Br
PPh3
S
Br
O
12
13
H
h
O
NO2
15
Cl
Cl
Cl
Cl
O
O
Cl
Cl
Cl
Cl
R
c
d
N
N
+
NH2
O
O
17a-h
9, 18a-g
19h
R
Cl
Cl
Cl
Cl
O
O
Cl
Cl
Cl
Cl
d
d
N
N
H2N
H2N
R
O
R
O
22-24
25-27
20
21
R = phenyl (22)
R = benzyl (23)
R = phenoxy (24)
R = phenyl (25)
R = benzyl (26)
R = phenoxy (27)
Scheme 1. Reagents and conditions: (a) PPh3, CH3CN, reflux; (b) benzaldehydes, 18-crown-6,
K2CO3, DCM, reflux; (c) SnCl2·2H2O, AcOEt, reflux; (d) tetrachlorophthalic anhydride, AcOH,
reflux.
Styrylphenylphthalimide analogs were synthesized as shown in Scheme 1. Benzyl
bromides 10, 11 were treated with PPh3 to generate phosphonium ylides 11, 13. Wittig
6

reaction of ylides 11 and aldehydes 14a-g, and ylide 13 and aldehyde 15 afforded
ortho-stilbenes 16a-h. Reduction of the nitro group of 16a-h with SnCl2·2H2O,
cyclization with tetracholorophthalic anhydride, and separation of the EZ isomers gave
the E-isomers 9 and 18a-g, and Z-isomer 19h. Various amines 20, 22-24 were cyclized
with tetracholorophthalic anhydride to give 21, 25-27.
Our previous SAR studies indicated that chloro atoms at phthalimide are necessary for
selective LXRβ agonistic activity. In addition, introduction of various substituents or
changing position of methoxy substituent of the terminal benzene ring at styryl group
did not lead to improve LXRβ agonistic activity. These results indicated that
substituent(s) at the terminal benzene ring would interfere binding to LXRβ binding
pocket because of its bulkiness. Therefore, other approaches except for introduction of
substituent(s) are required for further structural development. On the other hand, the %
efficacy for LXRβ varied depending on the space that was occupied by the terminal
benzene ring. These results led us to change the space that was occupied by the terminal
benzene ring by changing the linker.
7

Table 1. SAR for the suitable spatial arrangement of terminal benzene ring
agonistic activity
Compound
Ar
LXR
LXR
EC50 (M)a
% efficacyb
100
EC50 (M)a
0.006
N.A.
0.021
3
9
143
0.27±0.02
N.A.
N.A.
N.A.
117
28
21
4.98±0.29
N.A.
-
>10
104c
19h
8

N.A.
153
25
1.73±0.78
N.A.
N.A.
109
99
26
27
8.00±0.59
8.98±1.39
a N.A.: no activity at 30 M; data are the mean±SD.
b % efficacy:percent of maximal activity relative to 3.
c % efficacy at 30 M
We fixed the non-substituted terminal phenyl group, and synthesized analogs with
C0-C2 linker length to examine the suitable spatial arrangement of terminal benzene
ring. Naphthyl analog 25 and (Z)-styrylphenyl analog 19h lacked LXRβ agonistic
activity indicating that LXRβ would not have enough space near meta-position. This
hypothesis was supported by our previous SAR that meta-phenethyl analog lacked
LXRβ agonistic activity. As for analogues possessing C0-C2 linker (25, 26 and 28), the
order of EC50 value was C0 (25) < C2 (28) < C1 (26). However, (E)-styrylphenyl
analog 9 showed more potent EC50 than biphenyl analog 25. On the other hand, linker
with heteroatom (27) showed less potent EC50 and % efficacy than the carba-analog 26.
9

This result might suggest that hetero linker does not contribute to LXRβ agonistic
activity. These results mentioned above led us to fix the spacer as (E)-vinyl linker.
Second approach for structural development was to substitute the terminal benzene
ring with hetero aromatic rings to improve aqueous solubility. Compared to pyridine
analogues (18a-18c), 2-pyridinyl (18a) > 3-pyridinyl (18b) > 4-pyridinyl (18c)
analogues showed potent activity (EC50) and % efficacy in the indicated order, but these
analogues were weaker than phenyl analogue (9). We also synthesized analogs bearing
five-membered heterocycles 18d-g. This idea based on our previous reports which
showed the disruption of molecular symmetry (bending molecules) can increase the
aqueous solubility of molecules even if their hydrophobicity is not concomitantly
increased.19,20 2-Furyl (18d), 2-thiazolyl (18e) analogues also exhibited weaker LXR
agonistic activity than compound 9. These results suggest that decrease of
hydrophobicity of molecules by introduction of the heteroatom(s) decreased LXRβ
agonistic activity. Then, we hypothesized that substitution of the terminal phenyl ring
with a heterocycle possessing higher hydrophobicity might maintain LXR agonistic
10

activity. Actually, 2-thienyl (18f) and 3-thienyl (18g) analogues, especially, 18f
exhibited potent activity (EC50: 0.559 μM, 149% efficacy) close to phenyl analogue (9).
Next, we analyzed the solubility of compound 9 and 18f in EtOH: 1/15 M phosphate
buffer (pH7.4) 1:1 by HPLC. Compound 18f showed about 4 times improved solubility
(0.015 ug/mL) over 9 (0.0036 ug/mL). Melting point of 18f (257-258ºC) was lower than
that of 9 (264-265ºC). This data support our idea that converting the terminal benzene
ring to five-membered hetero aromatic ring could disrupt the molecular symmetry that
increase the improvement of aqueous solubility.
Table 2. SAR of the terminal hetero aromatic rings
agonistic activity
cLogPd
Compound
Ar
LXR
LXR
% efficacyb
143
EC50 (M)a
EC50 (M)a
0.27±0.02
N.A.
7.0
9
11

N.A.
N.A.
N.A.
N.A.
N.A.
N.A.
N.A.
148
141
35c
5.6
5.7
5.5
6.6
5.2
6.8
6.8
18a
18b
18c
18d
18e
18f
4.77±2.51
4.96±2.21
>10
167
151
149
133
4.67±1.53
3.61±0.89
0.559±0.12
0.915±0.02
18g
a N.A.: no activity at 30 m; data are the mean:SD.
b % efficacy:percent of maximal activity relative to 3.
c % efficacy at 30 M
d ACD/Chem sketch 2012 (ver 14.01)
To understand the relationships between LXR agonistic activity and hydrophobicity of
the compounds, CLogP and –LogEC50 of the compounds shown in Table 2 was plotted.
As shown in Figure 2, CLogP and LXR agonistic activity showed a high correlation
12

(R2 = 0.63). This result might indicate that the binding pocket of LXR occupied with
the terminal aromatic ring is hydrophobic character.
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
y = 0.9531x + 0.7858
R² = 0.63239
4
4.5
5
5.5
6
6.5
7
-LogEC50
Figure 2. Relationships between LXR agonistic activity and hydrophobicity.
Next, we investigated further biological studies whether 18f exhibit similar activity to
parent 9. First, we considered the difference of agonistic activities towards human and
mouse LXRs by means of mouse full-length LXRs reporter gene assay. Compounds 26
and 27 showed 3-fold weaker LXRβ-agonistic activity than 9 but showed no
13

LXRα-agonistic activity at 30 μM. And % efficacies of 18f and 18g against LXRβ were
similar to that of 9. Thus, 9, 18f and 18g are LXRβ selective agonists for both human
and mouse.
Table 3. Selectivity for LXR in mouse LXRs.
agonistic activity
mouse LXR
EC50 (M)a
N.A.
mouse LXR
% efficacyb
EC50 (M)a
0.30±0.01
0.79±0.18
0.82±0.09
2.93±0.15
Compound
26
24
23
22
9
N.A.
18f
18g
18e
N.A.
N.A.
a N.A.: no activity at 30 m; data are the mean:SD.
b % efficacy:percent of maximal activity relative to 3.
The selectivity of 18f over other NRs (PPARα/γ, RARα/β/γ, RXRα/β/γ, FXR) was
evaluated. Compound 18f showed weak agonistic activity towards FXR (9% efficacy)
14

and RARα (6% efficacy), but did not show activity towards other NRs. (supplementary
Figure S1).
We next investigated whether 18f binds directly to LXRα/β LBD as 9 does by means of
TR-FRET assay18 (Figure 3). Compound 18f showed dose-dependent LXRβ-partial
agonistic activity (EC50 = 40.4 nM, 14% efficacy compared to T0901317 (3) at 3 μM).
On the other hand, compound 18f showed very weak agonistic activity towards LXRα
(6% efficacy at 0.3 μM compared to T0901317 (3) 3 μM). These EC50 values and %
efficacies were similar to those of 9. These results may indicate that 18f binds directly
to both the LXRβ LBD and LXRα LBD, but recruits the coactivator peptide
preferentially to LXRβ rather than LXRα, at least under our conditions. There is another
possibility that T0901317 (3) and 18f recruit the different coactivators. In that case, 18f
could show the partial agonistic activity in TR-FRET assay, whereas the full agonistic
activity in reporter gene assay. The difference of recruited coactivators may cause the
selective activation of the target gene transcription.
15

Figure 3. Results of TR-FRET LXR binding assays. Data are the mean ± SD. The
binding efficacy was compared using an unpaired, two-sided Student’s t-test. The
P-value is indicated by asterisk *p<0.05, **p<0.01, N.S.=not significant relative to
DMSO control. %eff was calculated by comparison to T0901317.
Figure 4. Docking simulation of LXRs and 18f. Compound 18f (pink) was docked into
the X-ray crystal structures of complexes between GW3965 (4, green) and the LBDs
(violet mesh) of a) LXR (PDB ID: 3IPQ) and b) LXR (PDB ID: 1PQ6).
16

To understand the binding mode between 18f and LXR, 18f was computationally
docked into the cocrystal structures of LXRα LBD and LXRβ LBD taken from the
complexes with GW3965 (4) (Figure 4). The most favorable conformation had a free
energy of binding of -12.02 kcal/mol (LXRα) or -11.62 kcal/mol (LXRβ), indicating
that 26 would bind to both LXRs. In addition, the results indicated that 26 would bind
similarly to the binding pockets of both LXRs but the thiophene ring in 26 faced to the
opposite direction, which might cause the selectivity. These results were also consistent
with the idea that the LXRβ selective agonistic activity of 26 might be a result of
post-binding conformation change or differential coactivator recruitment, rather than
binding preference.18 Our SARs indicated that the hydrophobic terminal aromatic ring
would be suitable for LXR agonistic activity. The amino acids located near the bound
thiophene ring in 26 were also hydrophobic character, that is, Leu274, Ala275, Met312,
Ile327 and Phe329. Thus, the docking study supported this SAR, and the hydrophobic
terminal aromatic ring might be able to have hydrophobic interaction with these
hydrophobic amino acids.
17

We next examined the agonistic activity of 18f by means of mRNA expression analysis
of ABCA1 and SREBP-1c. In THP-1 cells, 18f induced expression of only ABCA1
mRNA, but not SREBP-1c mRNA expression. Compound 18f showed more potent
activity than our previous reported 9 in 1 M treatment, which reflect the result of %eff
of reporter gene assay (Figure 5). This is consistent with the idea that 18f works as an
LXRβ specific agonist which would improve atherosclerosis without lipogenic side
effects.
Figure 4. Real-time PCR analysis of ABCA1 and SREBP-1c mRNA expression in
THP-1 cells. Data are the mean ± SD. The mRNA level was compared using an
unpaired, two-sided Student’s t-test. The P-value is indicated by asterisk *p<0.05,
**p<0.01, N.S.=not significant relative to DMSO control.
In
summary,
we
achieved
further
structural
development
of
tetrachloro(styrylphenyl)phthalimides as the LXRβ-selective agonists with improved
18

aqueous solubility compared to compound 9 which we have reported. 2-Thienyl
analogue 18f exhibited potent LXRβ selective agonistic activity (EC50: 0.559 μM, 149%
efficacy) close to phenyl analogue 9 and similar biological identity in several assays. In
addition, Compound 18f showed about 4 times improved solubility (0.015 μg/mL) over
9 (0.0036 μg/mL). Then, Compound 18f could to be a promising lead compound for the
development of agents to treat atherosclerosis without the side effects.
References and notes
(1)
Annicotte JS, Schoonjans K, Auwerx J. Anat. Rec., Part A.
2004;277A;312-316.
(2)
(3)
Schultz JR, Tu H, Luk A, et al. Genes Dev. 2000;14;2831-2838.
(a) Janowski BA, Grogan MJ, Jones SA, et al. Proc. Natl. Acad. Sci. U.S.A.
1999;96;266-271.
(b) Lehmann JM, Kliewer SA, Moore LB, et al. J. Biol. Chem.
1997;272;3137-3140.
(4)
(5)
Venkateswaran A, Laffitte BA, Joseph SB, et al. Proc. Natl. Acad. Sci. U.S.A.
2000;97;12097-12102.
Kennedy MA, Venkateswaran A, Tarr PT, et al. J. Biol. Chem.
2002;277;17375.
(6)
Repa JJ, Berge KE, Pomajzl C, et al. J. Biol. Chem. 2002;277;18793-18800.
Zhao C, Dahlman-Wright K. J. Endocrinol. 2010;204;233-240.
Joseph SB, Laffitte BA, Patel PH, et al. J. Biol. Chem. 2002;277;11019-11025.
Repa JJ, Mangelsdorf DJ. Nat. Med. 2002;8;1243-1248.
Schultz JR, Tu H, Luk A,et al. Genes Dev. 2000;14;2831-2838.
Bradley MN, Hong C, Chen M, et al. Clin. Invest. 2007;117;2337-2346.
(7)
(8)
(9)
(10)
(11)
19

(12)
(13)
Lund EG, Peterson LB, Adams AD, et al. Biochem. Pharmacol.
2006;71;453-463.
(a) Faernegardh M, Bonn T, Sun S, et al. J. Biol. Chem.
2003;278;38821-38828.
(b) Williams S, Bledsoe RK, Collin J, et al. J. Biol. Chem.
2003;278;27138-27143. ꢀ
(14)
(15)
Collins JL, Fivush AM, Watson MA, et al. J. Med. Chem. 2002;45;1963-1966.
Joseph SB, McKilligin E, Pei L, et al. Proc. Natl. Acad. Sci. U.S.A
2002;99;7604-7609.
(16)
(a) Kick E, Martin R, Xie Y, et al. Bioorg. Med. Chem. Lett. 2015;25;372-377.
(b) Hu B, Unwalla RJ, Goljer I, et al. J. Med. Chem. 2010;53;3296. (c)
Molteni, V, Li X, Nabakka J, et al. J. Med. Chem. 2007;50;4255-4259. (d)
Zheng Y, Zhuang L, Fan KY, et al. J. Med. Chem. 2016;59;3264-3271.
(a) Noguchi-Yachide T, Miyachi H, Aoyama H, et al. Chem. Pharm. Bull.
2007;55;12;1750-1754. (b) Motoshima K, Noguchi-Yachide T, Sugita K, et al.
Bioorg. Med. Chem. 2009;17;14;5001-5014. (c) Aoyama A, Endo-Umeda K,
Kishida K, et al. J. Med. Chem. 2012;55;7360-7377. (d) Nomura, S.,
Endo-Umeda, K, Aoyama A, Makishima M, et al. ACS Med. Chem. Lett.
2015;6; 902-907.
(17)
18
Nomura S, Endo-Umeda K, Makishima M, Hashimoto Y, Ishikawa M.
ChemMedChem. 2016;11;2347-2360.
19
20
Ishikawa M, Hashimoto Y. J. Med. Chem. 2011;54;1539–1554.
Ishikawa M, Hashimoto Y. in Pract. Med. Chem. (Eds.: C. G. Wermuth, D.
Aldous, P. Raboisson, D. Rognan), Academic Press, Massachusetts,
2015;pp;747–765.
20