Synthesis and evaluation of anilinohexafluoroisopropanols as activators/modulators of LXRα and β

Article abstract of DOI:10.1016/j.bmcl.2006.06.081

A series of branched and unbranched anilinohexafluoroisopropanols related to the known sulfonamide T0901317 were prepared and evaluated as activators/modulators of both LXRα and LXRβ. A structure-activity relationship was established and compounds with high potency on both the receptors were identified. Many compounds showed a tendency toward selectivity for LXRβ versus LXRα. Several analogues were evaluated for effects on plasma lipoprotein levels in mice. A few of these significantly raised HDL-cholesterol levels in plasma but showed markedly different effects on liver triglyceride content, suggesting that this series may yield candidates with improved efficacy/safety profiles compared to existing molecules.

Full text of DOI:10.1016/j.bmcl.2006.06.081

Bioorganic & Medicinal Chemistry Letters 16 (2006) 5231–5237
Synthesis and evaluation of anilinohexafluoroisopropanols
as activators/modulators of LXRa and b
Narendra Panday,* Jo¨rg Benz, Denise Blum-Kaelin, Vanessa Bourgeaux,
Henrietta Dehmlow, Peter Hartman, Bernd Kuhn, Hassen Ratni,
Xavier Warot and Matthew B. Wright
F. Hoffmann-La Roche Ltd., Pharmaceuticals Division, Preclinical Research, CH-4070 Basel, Switzerland
Received 12 June 2006; revised 27 June 2006; accepted 28 June 2006
Available online 31 July 2006
Abstract—A series of branched and unbranched anilinohexafluoroisopropanols related to the known sulfonamide T0901317 were
prepared and evaluated as activators/modulators of both LXRa and LXRb. A structure–activity relationship was established and
compounds with high potency on both the receptors were identified. Many compounds showed a tendency toward selectivity for
LXRb versus LXRa. Several analogues were evaluated for effects on plasma lipoprotein levels in mice. A few of these significantly
raised HDL-cholesterol levels in plasma but showed markedly different effects on liver triglyceride content, suggesting that this series
may yield candidates with improved efficacy/safety profiles compared to existing molecules.
Ó 2006 Elsevier Ltd. All rights reserved.
The Liver-X-receptors, LXRa and LXRb, are nuclear
hormone receptors that function as oxysterol regulated
transcription factors and activate the expression of genes
regulating cholesterol and lipid metabolism.1 The LXRs
directly induce the expression of the transmembrane lip-
id/cholesterol transporters ABCA1 and ABCG1 in cho-
lesterol-loaded macrophages, liver, and intestinal cells,
promoting cholesterol efflux and formation of high densi-
ty lipoprotein particles (HDL).2 Numerous clinical and
epidemiological studies have shown that HDL-cholester-
ol levels are inversely related to the risk for coronary
artery disease (CAD).3 Drugs that activate LXR have
the potential to increase HDL-C and cellular cholesterol
efflux and are thus expected to be atheroprotective.4 Full
LXR agonism however leads to the undesired activation
of triglyceride synthesis in the liver by upregulation of
Srebp-1c.5 The identification of LXR modulators, devoid
of this undesired side effect remains a major challenge for
drug development. Current efforts focus on the identifica-
tion of LXRb-selective agonists, partial or gene-specific
LXR activators, or compounds with more favorable
PK/PD properties.5c,6
One of the most extensively studied LXR activators is
T0901317 (Fig. 1), a potent but pathway unselective
LXRa/b coagonist developed by Tularik.7 T0901317
has been shown to increase HDL-C and to reduce
atherosclerotic plaques in mouse models4,8 and thus
was considered a starting point for the design of new ana-
logues with potentially improved properties. The X-ray
structure of T0901317 complexed with the LXRb-ligand
binding domain revealed the key contacts of the ligand re-
quired to stabilize the active conformation of the recep-
tor.9 Apart from a strong hydrogen bond between the
˚
hydroxyl group of the ligand and His435 (d = 2.6 A),
numerous lipophilic receptor–ligand contacts exist that
lead to strong binding. The structure revealed that the tri-
fluoroethyl and the phenylsulfonyl groups reach into two
pockets marked as P1 and P2 in Figure 1. A third pocket
P3 could be discerned at the opposite end of the ligand
which was unoccupied and could offer opportunities for
the introduction of additional functionalities. As
illustrated in Figure 1, we reasoned that replacing the –
SO2– moiety of T0901317 by a methine group –CHRa–
would allow the positioning of a new functional group
Ra into P3 and possibly improved affinity and activity.
However, according to molecular modeling10 the aniline
resulting from a replacement of –SO2– by a –CHRa–
(Ra = H or functional group) has a significantly smaller
dihedral angle s of approximately 27° compared to the
experimentally observed s = 62° for T0901317, which
Keywords: Hexafluoroisopropanol; Aniline; Liver-X-receptor activa-
tors/modulators; SAR; Molecular modeling.
*
Corresponding author. Tel.: +41 0 61 688 26 72; fax: +41 0 61 688 64
59; e-mail: npanday@epo.org
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.06.081

5232
N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
His 435
OH
His 435
F3C
CF3
Met 312
Arg 319
P1
P3
empty
R'
CF3
τ
P2
N
Phe 329
X
Aniline
T0901317
X = CHRa, R’ = H (τcalc = 27 )
R’ = small subst. (τcalc = 60-70 )
X = SO2, R’ = H, τx-ray = 62
Figure 1. Left: definition of the dihedral angle s discussed in the text. Middle: binding mode of T0901317 complexed to LXRb. The hydrogen bond
between the OH group of the ligand and His 435 is shown as a red, dashed line. Right: model of anilines illustrating the possibility to reach the P3
pocket and interact with different hydrophobic and polar side chains. Ra is a suitably chosen substituent reaching into the P3 pocket.
could result in a potentially detrimental reorientation of
the N-substituents. To compensate for this, introduction
of a substituent R0 in the ortho position to the aniline-ni-
trogen or the quaternization of the –CHRa– to a –
CRaRb– may be necessary to increase the s angle and pre-
vent loss of affinity/activity. We report here the synthesis
and SAR of (i) ‘unbranched’ anilines for which X is –
CH2– and (ii) ‘branched’ anilines for which X is –CRaRb–
with at least one of Ra or Rb not being H.11 In both series,
analogues with a chloro substituent ortho to the aniline-
nitrogen were prepared.
29–32 by EDCI/HOBT-promoted coupling.14 LiHM-
DA-promoted deprotonation in a position to the ester
functionality of 26 followed by quenching with iodome-
thane or -ethane, led, after TBAF-mediated deprotec-
tion, to the quaternary derivatives 33 and 34. Using
the same method, the re-O-silylated derivative of
dimethylamide 30 was converted to the a-methyl ana-
logue 36. Treatment of ester 34 with N-chloro succinimide
gave the o-chloroaniline 35. The bridged derivatives 37
and 38 were obtained by a double deprotonation of 25
with LiHMDA, treatment with either 1,3-diiodopro-
pane or 1,4-diiodobutane, followed by TBAF-mediated
deprotection. In the presence of lithiumperchlorate,15
the O-protected N-ethyl aniline 24 reacted with styrene
epoxide mainly to give the racemic hydroxymethyl
derivative 39, though small amounts of the alcohol 40
resulting from nucleophilic attack of the aniline on the
b-carbon of the epoxide were also isolated. The enantio-
merically pure S- or R-derivatives of 39 were obtained
by using the corresponding optically pure styrene epox-
ides. The o-chloroaniline 41 was prepared from 39 by
treatment with N-chlorosuccinimide. TBAF-mediated
deprotection of 39–41 yielded the corresponding diols
42–44. Both 39 and its o-chlorinated analogue 41 were
further converted to a series of O-functionalized deriva-
tives (45–54) by either O-alkylation or Mitsunobu-type
substitution reactions and optional subsequent transfor-
mations (e.g., LiAlH4-mediated reduction of 47 to 48,
LiOH-promoted hydrolysis of 47 to 49 and 52–54 to
55–57, and amide couplings of 49 to 50 and 51).
The syntheses started from the commercially available
2-(4-amino-phenyl)-1,1,1,3,3,3-hexafluoro-propan-2-ol
1.12 The unbranched analogues (Scheme 1) were ob-
tained by acetylation or trifluoroacetylation (2 and 3)
followed by BH3–THF-complex promoted reduction (4
and 5). Heating in the presence of an arylalkylhaloge-
nide13 optionally followed by o-chlorination of the ani-
line with NCS led to the final compounds (6–22). For
the synthesis of the branched analogues (Scheme 2),
we used the O-silyl protected derivative 23 of 1 which
was converted to the N-ethyl derivative 24 using the
same method as for the conversion of unprotected 2–4.
Treatment of either 23 or 24 with a-phenylbromoacetic
acid methyl ester led to phenylmethylesters 25 and 26,
respectively, of which the latter was deprotected with
TBAF to give 27. Lithium hydroxide mediated hydroly-
sis of 26 was accompanied by desilylation and gave the
deprotected acid 28, which was converted to amides
OH
OH
OH
F3C
CF3
OH
F3C
CF3
F3C
CF3
F3C
CF3
c
b
a
R3
R2
6 - 22
N
R1
R1
HN
R1
HN
4, 5
R3 = H
R3 = Cl
NH2
d
O
1
2, 3
Scheme 1. Reagents and conditions: (a) for R1 = CH3/Ac2O, Py, followed by treatment with aq NaOH, >90% of 2. For R1 = CF3, TFAA, CH2Cl2,
t
DIPEA, >90% of 3; (b) BH3ÆTHF, 96% of 4 (R1 = CH3), 73% of 5 (R1 = CF3); (c) R2-Cl or R2-Br, DMF or BuOH, heating, 11–90%; (d) NCS,
n-propanol, >64%.

N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
5233
OR1
OH
N
OR1
CF3
OTES
CF3
F3C
CF3
F3C
CF3
F3C
F3C
f, g, h
c
c
R5
Ph
R4
N
Ph
Et
HN
R2
Et
CO2R3
26 R1 = TES, R3 = Me
27
Ph
NH
COR6
1 R1 = R2 = H
23 R1 = TES, R2 = H
24 R1 = TES, R2 = Et
CO2Me
29 - 32
33 - 34
35
a
b
d
e
R4, R5, R6
as in Tab. 2
R1 = H, R3 = Me
25
k
28 R1 = R3 = H
i
l
36
OH
OR1
OR1
OH
F3C
CF3
F3C
CF3
F3C
CF3
F3C
CF3
m
n, o
R5
+
R5
Ph
N
Et
Ph
N
Ph
N
N
Et
Et
(
)
CO2Me n
OR6
OH
Ph
OH
40 R1 = TES
42 R1 = H
52 - 54
55 - 57
45 - 47
37 n = 1
38 n = 2
39 R1 = TES, R5 = H
41 R1 = TES, R5 = Cl
43/44 R1 = H, R5 = H/ Cl
e
e, p
48 - 49
50 - 51
k
d
f
d
Scheme 2. Reagents and conditions: (a) TESCl, DBU, DMF, >90%; (b) (1) Ac2O, Py, 2—BH3ÆTHF, 96%; (c) a-phenylbromoaceticacidmethylester,
NaOAc, tBuOH, >90%; (d) TBAF, >90%; (e) LiOH in THF/H2O 1:1, >90% of 28 from 26, of 49 from 47, of 55–57 from 52–54; (f) amine or amine
hydrogen chloride salt, EDCI, HOBt, NMO, from 28: 73 % of 29, 70% of 30, 27% of 31, 44% of 32, from 49: >90% of 50, 76% of 51; (g) (1)
LiHMDA, (2) MeI, (3) TBAF, 67% of 33 from 26; (h) (1) LiHMDA, (2) EtI, (3) TBAF, 30% of 34 from 26; (i) (1) TESCl, DBU, DMF, >90% of
o-silylated 30, (2) LiHMDA, (3) MeI, (4) TBAF, 30% of 36 from 30; (k) NCS, propanol, 50% of 35 from 34, 69% of 41 from 39; (l) (1) 2 equiv
LiHMDA, (2) 1,3-diiodopropane or 1,4-diiodobutane, (3) TBAF, ca. 90% of 37, 3% of 38; (m) Li2ClO4, rac-, R-, or S-styreneoxide, 50–63% of rac-,
R- or S-39, 5% of rac-40; (n) (1) BuLi, (2) MeI, BnBr or BrCH2CO2Me, (3) TBAF, 56–95% of 45, 46 or 47 from 39; (o) (1) methyl-4-
hydroxybenzoate, methyl-4-hydroxyphenylacetate or methyl-3-(4-hydroxyphenyl)-propionate, DIAD, Ph3P, (2) TBAF, 7–13% of 52–54; (p) LiAlH4,
>90% of 48 from 47.
The compounds were evaluated for binding affinity and
the ability to transcriptionally activate LXRa and
LXRb in in vitro radioligand displacement and cellular
transcriptional transactivation assays.16
however, the o-chloro substituent did not further im-
prove the potency of the already highly potent pheny-
loxadiazolyl derivatives (17/20 vs 18/21). The
phenyloxadiazolyl substituent is predicted to be too
large to fit into the P2 pocket and presumably reaches
instead into the P3 pocket both in the non-chlorinated
and the o-chlorinated analogues. Favorable interactions
between the phenyloxadiazole moiety and hydrophobic/
aromatic amino acid side chain surrounding the P3
pocket probably account for the very high affinity and
activity observed for 17–22. Moreover, the values for
17/20 and 18/21 also suggest that interactions of the
receptor with the chlorine contribute little if anything
to the binding affinity. Thus, the improvement in affinity
observed for the o-chloro derivatives 7, 9, 11, and 16 in-
deed results primarily from the reorientation of the N-
substituents. Quite interestingly, the trifluoroethyl group
led to a significant loss in affinity and activity of the
phenyloxazolyl series (17/20 vs 19/22). Perhaps the reori-
entation of the N-substituents to properly accommodate
the large phenyloxazolyl group forces the trifluoroethyl
group into repulsive interactions with the receptor, while
the slightly smaller ethyl group is tolerated (Table 1).
Comparison of potency in binding and transactivation
of the sulfonamide T0901317 and the benzyl analogue
6 showed that a slight decrease in affinity and activity to-
ward LXRa (but not b) resulted from the replacement of
the SO2-moiety by a –CH2–. It is likely that the unfavor-
able reorientation of the phenyl group due to the smaller
s-angle in 6 is compensated by improved interactions
with the more lipophilic methylene group as compared
to those with the rather polar SO2-moiety of
T0901317. Quite remarkably, there was no marked dif-
ference in potency between the trifluoroethyl and the
ethyl anilines 6 and 8. Introduction of an ethylene spacer
(15) instead of the methylene (8) did not significantly af-
fect the affinity and activity. Replacement of the phenyl
ring of 7 by a less lipophilic heterocycle (10–14) resulted
in loss of affinity and activity, with the 3-pyridyl proving
particularly unfavorable. On the other hand, introduc-
tion of a thiazolyl group (10 and 11) led to two com-
pounds with modest selectivity for LXRb versus
LXRa. The importance of the s-angle and, hence, of
the proper orientation of the aniline substituents toward
the P1 and P2 pockets is evidenced by the marked in-
crease in affinity and activity of the o-chloro anilines 7,
9, 11, and 16, as compared to the corresponding unchlo-
rinated counterparts 6, 8, 10, and 15. Very significantly,
The receptor binding and transactivation data for the
branched anilines are presented in Tables 2 and 3. As
compared to the parent aniline 8 the derivative with a
methoxycarbonyl substituent on the methylene moiety
(27) showed reduced affinity and activity. The loss in
affinity/activity was even more pronounced for 28–32

5234
N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
Table 1. Binding affinity (IC50) and transactivation potency (EC50) of unbranched anilines
Compound
R1
—
R2
—
R3
—
LXRa
EC50, lM (% eff )
LXRb
EC50, lM (% eff )
IC50, lM
IC50, lM
T0901317
0.03
0.25 (100)
0.07
0.3 (100)
6
7
8
9
CF3
CF3
CH3
CH3
H
0.07
0.9 (87)
0.1 (89)
1.1 (100)
0.2 (88)
0.06
0.2 (79)
0.02 (55)
0.4 (71)
0.08 (91)
Cl
H
0.001
0.05
0.003
0.003
0.03
0.003
Cl
N
10
11
CH3
CH3
H
Cl
0.6
0.02
3.3 (132)
0.5 (122)
0.05
0.002
0.2 (84)
0.16 (112)
S
12
13
14
CH3
CH3
CH3
H
H
H
1.5
2.7
1.7
0.7 (70)
4.3 (88)
2.1 (108)
0.46
2.32
1.68
0.3 (74)
2.3 (106)
0.5 (93)
2
3
4
4
3
N
2
15
16
CH3
CH3
H
Cl
0.12
0.006
1.0 (80)
0.8 (78)
0.61
0.003
0.2 (76)
0.09 (72)
17
18
19
CH3
CH3
CF3
H
Cl
H
0.005
0.002
0.03
0.2 (90)
0.3 (78)
1.0 (117)
0.005
0.004
0.05
0.06 (86)
0.04 (94)
0.3 (89)
Cl
Me
O
N
20
21
22
CH3
CH3
CF3
H
Cl
H
0.006
0.006
0.013
0.2 (96)
0.1 (130)
0.7 (130)
0.003
0.007
0.02
0.06 (72)
0.03 (100)
0.1 (83)
F3C
Me
O
N
Table 2. Binding affinity (IC50) and transactivation potency (EC50) of branched anilines derived from N-alkylation with a-phenylbromoacetic acid
methyl ester
Compound
R4
R5
R6
LXR a
EC50, lM (% eff )
LXR b
EC50, lM (% eff )
IC50, lM
IC50, lM
rac-27
rac-28
rac-29
rac-30
H
H
H
H
H
H
H
H
OMe
OH
0.7
5.2
2.3 (33)
9.4 (77)
10.9 (24)
9.3 (18)
0.9
13.8
22
2.2 (50)
5.4 (95)
11 (32)
2.3 (12)
NHMe
NMe2
11.2
13.8
6.3
X = CH2
X = O
rac-31
rac-32
H
H
H
H
2.2
5.4
2.2 (49)
8.5 (47)
1.2
1.4
2.2 (26)
2.3 (88)
X
N
rac-33
rac-34
rac-35
rac-36
rac-37
rac-38
Me
Et
H
H
Cl
H
H
H
OMe
OMe
OMe
NMe2
OMe
OMe
0.05
0.05
0.06
0.2
0.1 (81)
0.3 (94)
0.5 (99)
0.4 (67)
0.2 (92)
0.09 (116)
0.008
0.04
0.03
0.06
0.03
0.03
0.03 (105)
0.07 (89)
0.07 (64)
0.07 (63)
0.04 (58)
0.04 (75)
Et
Me
—
0.05
0.06
—
bearing a polar carboxy or amido group. The lipophilic
piperidino and morpholino moieties of amides 31 and
32, respectively, are probably not properly oriented in
the P3-pocket and therefore only slightly compensate
for the repulsive interactions involving the amido func-
tionality. An improvement of up to 100-fold in affinity
was achieved by methylation or ethylation in the a-posi-
tion to the ester or amide functionality (33–36). Analo-
gous to the above-discussed ‘o-Cl-effect,’ this is most
likely due to the increased s-angle resulting from the
quaternization of the a-carbon with additional binding
interactions between the receptor and the introduced al-
kyl group probably playing a minor role. This interpre-
tation is supported by the similar increase in affinity
observed upon o-chlorination of unquaternized anilines
(see above and 43 vs 44), in contrast to the lack of an
effect when the o-chloro substituent was introduced after
the a-carbon was already quaternized (34 vs 35). Quat-
ernization of the a-carbon also proved effective when
the a-substituent and the former N-ethyl substituent
were fused to form a cycle (37 and 38).
Compared to the carboxy or amido groups of 28–30, the
still quite polar hydroxymethyl group of 43 was rather
well tolerated. The isomeric diol 42, obtained as side
product, showed somewhat higher affinity and activity.
Only minor differences were observed between the
respective enantiomers R- and S-43, though the latter,
in agreement to modeling predictions, was slightly more
potent. Consistent with the mostly lipophilic character
of the P3 pocket, O-methylation, O-benzylation or O-
methoxycarbonylmethylation of 43–45, 46, and 47
slightly improved affinity. Modeling suggested that the
benzyl group of 46 probably cannot orient in a way to

N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
5235
Table 3. Binding affinity (IC50) and transactivation potency (EC50) of branched anilines obtained by opening of styrene epoxide
Compound
R5
R6
LXRa
EC50, lM (% eff )
LXRb
EC50, lM (% eff )
IC50, lM
IC50, lM
rac-42
rac-43
rac-44
R-43
—
H
Cl
H
H
H
H
H
H
H
H
H
—
H
0.5
1.1
0.2
1.3
0.5
0.4
0.2
0.6
12
1.0 (63)
1.9 (87)
0.4 (75)
1.5 (65)
0.5 (73)
0.9 (111)
1.9 (55)
2.2 (54)
>40 (nd)
10 (10)
0.5
0.9
0.09
0.5
0.4
0.3
0.3
0.5
32
0.2 (97)
0.7 (53)
0.2 (98)
0.4 (91)
0.1 (92)
0.1 (73)
0.6 (85)
0.9 (65)
>40 (nd)
>40 (nd)
>40 (nd)
0.26 (67)
H
H
S-43
H
Me
rac-45
rac-46
rac-47
rac-48
rac-49
rac-50
rac-51
Bn
CH2CO2Me
CH2CH2OH
CH2CO2H
CH2CONHMe
CH2CONMe2
61
18
90
31
>40 (nd)
0.8 (87)
0.9
0.5
rac-52
rac-53
rac-54
rac-55
rac-56
rac-57
Cl
Cl
Cl
Cl
Cl
Cl
n = 0, R = Me
n = 1, R = Me
n = 2, R = Me
n = 0, R = H
n = 1, R = H
n = 2, R = H
0.06
0.2
0.7 (94)
1.1 (65)
1.4 (43)
3.7 (61)
5.5 (65)
4.5 (72)
0.02
0.03
0.05
0.02
0.02
0.007
0.2 (79)
0.3 (98)
0.6 (98)
1.1 (89)
2.4 (54)
1.2 (55)
0.1
0.1
0.1
0.06
optimally interact with the hydrophobic/aromatic side
chains surrounding P3, explaining the only modest gain
in affinity. The smaller but rather polar methoxycarbon-
yl and dimethylaminocarbonyl groups of 47 and 51,
respectively, are tolerated in P3, but further reduction
of the lipophilicity in this position led to practically inac-
tive compounds (48–50).
pounds showed significant PXR activity in a range
similar to T0901317. However, none of the compounds
activated FXR, in contrast to T0901317 which showed
low but significant activation in our assay.
To evaluate the pharmacological effects on HDL-C and
triglyceride levels, C57BL/6J mice were dosed once per
day for five days with the compounds 9, 20, and 33,
and plasma and liver were collected 2 h following the fi-
nal dose.19 The choice of daily dose for each compound
was based upon initial studies of compound stability
during incubation in the presence of mouse microsomes.
The results revealed distinct effects on plasma HDL-cho-
lesterol levels and liver triglyceride content (Table 4).
Compound 33, at 100 mg/kg, significantly and dramati-
cally increased liver triglyceride content by 520% com-
pared to control, but surprisingly, produced only a
marginal and insignificant increase in HDL-C. By con-
trast, compound 20, at 10 mg/kg, significantly increased
plasma HDL-C by 35% with a comparatively modest in-
crease in liver triglyceride content (70%). Finally, the
chloroaniline 9, at 100 mg/kg, significantly increased
HDL-C by 23% with almost no effect on liver triglycer-
ide content. Plasma triglyceride levels were not signifi-
cantly affected by any of the compounds. This is
consistent with literature reports that the increase in
plasma TGs after dosing with LXR agonists is transient
and normalizes after repeated dosing.20 Our experience
suggests that this is due to a secondary increase in turn-
over of TG-rich lipoprotein particles which may be med-
iated by increased lipoprotein lipase expression, a direct
target of LXR (unpublished data). Thus, steady-state
plasma TG levels in mice appear not to reliably reflect
the effects of LXR agonists on increased liver TG
synthesis.
In contrast to an arylmethyl group (e.g. benzyl group in
46), an aryl group attached to the hydroxymethyl moiety
of 43 or 44 was expected to orient in a way to interact
more favorably with the hydrophobic and aromatic side
chains surrounding P3. In agreement with these expecta-
tions, most of the O-phenylated derivatives 52–57
showed an improved affinity compared to the parent
44. This improvement was particularly pronounced for
the propionic acid 57 and could well be due to interac-
tions of the carboxy group with the Arg 319 side chain
at the entrance to the binding pocket (see Fig. 1). The
rather weak potencies measured for carboxylic acids
55–57 compared to those of the corresponding esters
in the cellular transactivation assay were likely the result
of reduced membrane permeability of these com-
pounds.17 Though 52–57 showed markedly improved
affinities and activities compared to the other branched
analogues, they did not reach the potency of the corre-
sponding unbranched aniline 9, indicating that the addi-
tional binding interactions in P3 were not sufficient to
compensate for the entropically unfavorable fixation of
the tested ‘–CH2–O–R6’ groups and/or for the potential-
ly reduced binding interactions of the other ligand moi-
eties which could slightly reorient upon accommodation
of a large ‘–CH2–O–R6’ group in P3.
The anilines 9, 20, and 33 were selected for evaluation in
animal studies. Since T0901317 has been reported to
have significant crossreactivity with PXR and FXR,
we evaluated these compounds in transcriptional trans-
activation assays for these receptors.18 All three com-
Compound levels were monitored by LC–MS in plasma
and liver extracts (Table 4). All compounds were highly
exposed, reaching plasma levels ranging between 6- and

5236
N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
Table 4. Effects on plasma HDL-C and triglyceride levels and liver triglyceride content in male C57Bl6J mice after 5 days treatment
Compound
Dose (mg/kg/day)
HDL-C
Plasma TG
Liver TG
Plasma exposure (ng/ml)
Liver exposure (ng/g)
9
20
33
100
10
+23%*
+35%*
+12%
+7%
ꢀ12%
ꢀ11%
+28%
+70%
+520%*
3400
700
13000
8200
6300
100
2300
Values are expressed as % change versus vehicle-treated group; *p > 0.05; Anova followed by Student’s t-test.
3. (a) Krasuski, R. A. Curr. Opin. Lipidol. 2005, 16, 652; (b)
Brousseau, M. E. Drug Discovery Today 2005, 10, 1095.
4. Terasaka, N.; Hiroshima, A.; Koieyama, T.; Ubukata, N.;
Morikawa, Y.;Nakai, D.;Inaba, T. FEBSLett. 2003, 536, 6.
5. (a) Grefhorst, A.; Elzinga, B. M.; Voshol, P. J.; Plosch, T.;
Kok, T.; Bloks, V. W.; van der Sluijs, F. H.; Havekes, L.
M.; Romijn, J. A.; Verkade, H. J.; Kuipers, F. J. Biol.
Chem. 2002, 277, 34182; (b) Schultz, J. R.; Tu, H.; Luk, A.;
Repa, J. J.; Medina, J. C.; Li, L.; Schendner, S.; Wang, S.;
Thoolen, M.; Mangelsdorf, D. J.; Lustig, K. D.; Shan, B.
Gene Dev. 2000, 14, 2831; (c) Miao, B.; Zondlo, S.; Gibbs,
S.; Cromley, D.; Hosagrahara, V. P.; Kirchgessner, T. G.;
Billheimer, J.; Mukherjee, R. J. Lipid Res. 2004, 45, 1410.
6. Cao, G.; Liang, Yu; Xian-Cheng, J.; Eacho, P. L. Drug
News Perspect. 2004, 17, 35.
50-fold higher and liver levels between 30- and 150-fold
higher than the EC50 in transcriptional transactivation
assays. These compounds show significant binding to
plasma proteins suggesting that the free fraction avail-
able for receptor activation may be somewhat lower.
The contrast in the relative effects on HDL-C-raising
versus the undesired induction of liver triglyceride syn-
thesis for these compounds is striking, however it is cur-
rently unclear if this is due to differential activation of
LXRa versus LXRb or to gene-specific or tissue-specific
regulatory functions. Further studies will be required to
fully dissect these mechanisms.
In summary, the design of new LXR ligands, suggested
partly by the published T091317/LXRb-cocrystal struc-
ture, has led to the identification of unbranched and
branched anilines with markedly improved affinity and
activity compared to the sulfonamide T0901317. The
modifications that led to improved potency were the
introduction of suitable functional groups into the P2
and/or P3 pockets as well as proper orientation of the
N-substituents. This was accomplished by either intro-
duction of a chloro substituent ortho to the anilino
nitrogen or by quaternizing the carbon atom that re-
placed the corresponding SO2 group of T0901317. The
identification of representative compounds that raised
HDL-C with different effects on activation of liver tri-
glyceride synthesis in mice, and of compounds with
selectivity for LXRb versus LXRa, suggests that this
class may hold promise to find LXR modulators with
reduced side effects.
7. Li, L.; Liu, J.; Zhu, L.; Cutler, S.; Hasegawa, H.; Shan, B.;
Medina, J. C. Bioorg. Med. Chem. Lett. 2006, 16, 1638,
and references cited therein.
8. Jiang, X.-C.; Beyer, T. P.; Li, Z.; Liu, J.; Quan, W.;
Schmidt, R. J.; Zhang, Y.; Bensch, W. R.; Eacho, P. L.;
Cao, G. J. Biol. Chem. 2003, 278, 49072.
9. Hoerer, S.; Schmid, A.; Heckel, A.; Budzinski, R.-M.;
Nar, H. J. Mol. Biol. 2003, 334, 853.
10. Protein structure analysis and molecular modeling were
performed with the program Moloc Gerber, P. R.;
Mueller, K. J. Comput. Aided Mol. Des. 1995, 9, 251.
11. Related branched and unbranched anilinohexafluoroiso-
propanols as LXR receptor modulators have also been
claimed by Van Camp, J.; Malecha, J.; Miyashiro, J. M.;
Decrescenzo, G. A.; Collins, J. T.; Kalman, M. J. Int.
Patent Appl. WO2003099769, 2003.
12. Details on the synthetic methodologies can in part be
found in Int. Patent Appl. WO200600323, 2006.
13. The phenyloxazolylmethylchlorides used to synthesize
derivatives 17–22 were prepared according to a procedure
described by Binggeli, A.; Bo¨hringer, M.; Grether, U.;
Hilpert, H.; Maerki, H.-P.; Meyer, M.; Mohr, P.; Ricklin,
F. Int. Patent Appl. WO200292084, 2002 (example 21,
steps a and b).
14. Vogel, Ch.; Jeschke, U.; Kramer, S.; Ott, A. J. Liebigs
Ann. Chem. 1997, 4, 737.
15. Chini, M.; Crotti, P.; Macchia, F. J. Org. Chem. 1991, 56,
5939.
16. Binding (IC50) was determined in a radioligand displace-
ment assay and transactivation (EC50) by means of a
cellular luciferase transcriptional reporter assay. The
details of the two assays can be found in Int. Patent
Appl. WO2006000323, 2006. The value ‘%-effect’ given in
brackets in addition to the EC50 indicates the maximally
achievable transactivation as compared to the reference
compound T0901317 which is set at 100%.
Acknowledgments
The authors thank A. Araujo del Rosario, A. van der
´
Klooster, B. Wolf, and K. Guitre for performing bind-
ing and transactivation assays, H. Isel for the formula-
tions, H. Meyer, S. Weiss, and T. Traendlin for
performing the in vivo studies, M. Mangold for the Hit-
achi measurements, S. Masur and I. Walter for LC–MS
measurements, and L. Forzy and P. Dietiker for sup-
porting the synthesis of several compounds.
References and notes
1. (a) Geyeregger, R.; Zeyda, M.; Stulnig, T. M. Cell. Mol.
Life Sci. 2006, 63, 524; (b) Collins, Jon L. Curr. Opin.
Drug Discov. Dev. 2004, 7, 692.
2. (a) Wang, N.; Ranalletta, M.; Matsuura, F.; Peng, F.;
Tall, A. R. Arterioscler. Thromb. Vasc. Biol. 2006, 26,
1310; (b) Crestani, M.; De Fabiani, E.; Caruso, D.; Mitro,
N.; Gilardi, F.; Vigil Chacon, A. B.; Patelli, R.; Godio, C.;
Galli, G. Biochem. Soc. Trans. 2004, 32, 92.
17. The rather modest EC50-values of esters 52–54 could be
the result of their partial hydrolysis to the acids 55–57
under the conditions of the cellular transactivation assay
(probably esterase promoted). Other esters (e.g., 33 and
47) seem to be less readily hydrolyzed.
18. Houck, K. A.; Borchert, K. M.; Hepler, C. D.; Thomas, J.
S.; Bramlett, K. S.; Michael, L. F.; Burris, T. P. Mol.
Genet. Metab. 2004, 83, 184.

N. Panday et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5231–5237
5237
19. Compounds were administered to C57Bl/6J mice (Charles
River, France), N = 6 per group, in a microemulsion of
solutol, labrafil, and oleic acid once per day for five days.
Animals were sacrificed and plasma and liver were
collected 2 h following the last dose. Liver triglycerides
were extracted by the Folch method Bligh, E. G.; Dyer, W.
J. A. Can. J. Biochem. Physiol. 1959, 37, 911, The
concentrations of cholesterol and triglycerides in plasma
and liver extracts were determined by standard enzymatic
methods on an automated clinical analyzer (Hitachi 912
Roche Diagnostics, AG). All experiments involving ani-
mals were performed in full compliance with the relevant
laws and institutional guidelines.
20. Joseph, S. B.; Laffitte, B. A.; Patel, P. H.; Watson, M. A.;
Matsukama, K. E.; Walczak, R.; Collins, J. L.; Osborne,
T. F.; Tontonoz, P. J. Biol. Chem. 2002, 13, 11019.