Cyclohexyl-linked tricyclic isoxazoles are potent and selective modulators of the multidrug resistance protein (MRP1)

Article abstract of DOI:10.1016/j.bmcl.2005.08.075

Structure-activity relationship (SAR) studies on the tricyclic isoxazole series of MRP1 modulators have resulted in the identification of potent and selective inhibitors containing cyclohexyl-based linkers. These studies ultimately identified compound 21b

Full text of DOI:10.1016/j.bmcl.2005.08.075

Bioorganic & Medicinal Chemistry Letters 15 (2005) 5526–5530
Cyclohexyl-linked tricyclic isoxazoles are potent and
selective modulators of the multidrug resistance protein (MRP1)
Bryan H. Norman,a,* Peter A. Lander,a Joseph M. Gruber,a Julian S. Kroin,a
Jeffrey D. Cohen,a Louis N. Jungheim,a James J. Starling,b Kevin L. Law,b Tracy D. Self,b
Linda B. Tabas,b Daniel C. Williams,b Donald C. Paulb and Anne H. Dantzigb
aDiscovery Chemistry Research, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA
bCancer Research, Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA
Received 11 July 2005; revised 23 August 2005; accepted 25 August 2005
Available online 3 October 2005
Abstract—Structure–activity relationship (SAR) studies on the tricyclic isoxazole series of MRP1 modulators have resulted in the
identification of potent and selective inhibitors containing cyclohexyl-based linkers. These studies ultimately identified compound
21b, which reverses drug resistance to MRP1 substrates, such as doxorubicin, in HeLa-T5 cells (EC50 = 0.093 lM), while showing
no inherent cytotoxicity. Additionally, 21b inhibits ATP-dependent, MRP1-mediated LTC4 uptake into membrane vesicles prepared
from the MRP1-overexpressing HeLa-T5 cells (EC50 = 0.064 lM) and shows selectivity (1115-fold) against the related transporter,
P-glycoprotein, in HL60/Adr and HL60/Vinc cells. Finally, when dosed in combination with the oncolytic MRP1 substrate vincris-
tine, 21b showed tumor regression and growth delay in MRP1-overexpressing tumors in vivo.
ꢀ 2005 Elsevier Ltd. All rights reserved.
The effective treatment of many types of cancer contin-
ues to be a significantly unmet medical need. Oftentimes,
the failure of a specific course of therapy is a result of the
tumor cells developing resistance to the agent(s) used in
the therapy.1 In the clinic, it is quite common for cancer
patients to initially respond to chemotherapy, but ulti-
mately experience relapse, due to the fact that their tu-
mors have developed resistance. In some cases,
patients develop multidrug resistance (MDR). These pa-
tients are resistant to several classes of oncolytics in
addition to the treatment drug.1 We have been interested
in understanding the molecular mechanisms by which
tumor cells become resistant to chemotherapeutic
agents.2,3 One mechanism involves the efflux of the
oncolytic (or a metabolite) out of the tumor cell, result-
ing in a lower intracellular concentration of the agent. It
has been shown that P-glycoprotein (Pgp, ABCB1)4 and
the multidrug resistance protein (MRP1, ABCC1)5,6 can
transport a variety of chemotherapeutics out of cancer
cells and these transport proteins are often overexpres-
sed in MDR tumors. Thus, one approach to improving
treatment would be to inhibit the action of these trans-
porters, thus restoring sensitivity to the tumor cells.
This should result in an improved response to therapy.
The Pgp transporter has been studied over the past two
decades and numerous modulators of Pgp have been de-
scribed.7 Many of these agents are presently being tested
in the clinic to determine the role of Pgp in clinical resis-
tance. The related transporter, MRP1, was discovered
much later.8 Thus, the number of selective modulators
of MRP1 is far less than for Pgp.9 We recently reported
on a novel class of selective MRP1 modulators called tri-
cyclic isoxazoles, characterized by 1 and 2.3 These com-
pounds demonstrated reversal of MRP1-mediated drug
resistance (in the presence of a sublethal dose of doxoru-
bicin) in the MRP1-transfected cell line, HeLa-T5
(EC50 = 0.90 and 1.13 lM, respectively). This is a direct
measure of a compoundÕs ability to reverse the MRP1-
mediated resistance in a relevant cell line which overex-
presses MRP1. Additionally, 1 demonstrated selectivity
for MRP1 versus Pgp in cells and inhibited ATP-depen-
dent, MRP1-mediated uptake of LTC4 into membrane
vesicles (Fig. 1). Finally, when dosed in combination
with the MRP1 substrate vincristine, 1 delayed the
growth of MRP1-overexpressing tumors in vivo. This
was the first description of an enhanced antitumor effect
Keywords: Multidrug resistance; MRP; Tricyclic; Isoxazole.
*
Corresponding author. Tel.: +1 317 277 2235; fax: +1 317 277
3652; e-mail: norman@lilly.com
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.08.075

B. H. Norman et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5526–5530
5527
OMe
One can imagine this class of molecules as having
three specific regions: tricyclic, linker, and aryl regions
(Fig. 2). Our early SAR efforts indicated that the aryl
substitution had very significant effects on the activity
of these compounds. Additionally, we have shown
that the tricyclic functionality is required for activity.
Thus, we have hypothesized that the primary purpose
of the linker region is to present the aryl and tricyclic
moieties in the proper three-dimensional space for
optimal binding to the MRP1 protein. At first, we
tried the simplest linker replacements—straight carbon
chains. The chemistry to prepare these compounds is
shown in Scheme 1.10 Beginning with either 5-amino-
valeric acid–HCl (3) or BOC-1,4-diaminobutane (7),
the targets (6 and 10) were prepared as previously de-
scribed,3 with the key synthetic step being an intramo-
lecular nucleophilic aromatic substitution reaction to
form the tricyclic isoxazole. Interestingly, these flexible
linker analogs 6 and 10 had HeLa-T5 EC50 values
similar to those of the phenyl linker compounds 1
and 2.
OMe
MeO
OMe
O
MeO
OMe
Cl
Cl
HN
N
N
O
NH
O
O
N
N
O
O
2
1
HeLa-T5 EC50 = 1.13 µM
HeLa-T5 EC50 = 0.90 µM
HL60/ADR EC50 = 0.07 µM
HL60/Vinc EC50 = 1.54 µM
MRP1/P-gp Selectivity Ratio = 22-fold
LTC4 Uptake inhibition IC50 = 1.8 µM
Figure 1. Aryl-linked tricyclic isoxazole inhibitors of MRP1.
with a selective MRP1 modulator and represented the
beginning of an SAR effort to find molecules which dem-
onstrated greater potency and selectivity at the target.
Aryl
OMe
In an effort to continue to explore alternative linker
functionalities, we considered saturated, 1,3-disubstitut-
ed carbocyclic linkers. We envisioned that these mole-
cules would be significantly less flexible than the
straight chain linkers 6 and 10, but with a slightly differ-
ent presentation of the aryl and tricyclic moieties, rela-
tive to the aryl linkers, such as 1 and 2. Additionally,
due to the sp3 nature of the saturated carbocyclic link-
ers, we sought to explore both cis and trans analogs of
the cyclohexyl linkers. Due to the expected 1,3-diequato-
rial orientation of the cis and 1,3-axial/equatorial orien-
tation of the trans isomers, we felt that the cyclohexyl
MeO
OMe
O
Cl
Tricyclic
HN
N
O
N
O
Linker
Figure 2. Tricyclic isozazoles. SAR regions.
OMe
OMe
O
MeO
Cl
Cl
F
Cl
CO2Me
3
N
CO2Me
a
c
CO2H
b
N
H
O
NH2
N
HN
N
O
N
O
O
N
O
4
O
5
6
HeLa T5 EC50 = 1.72 µM
OMe
OMe
MeO
Cl
Cl
F
Cl
N
NHBOC
d
NHBOC
a
b
NHBOC
N
H
N
O
NH2
N
O
N
O
NH
O
O
N
O
7
8
O
9
10
HeLa T5 EC50 = 1.22 µM
Scheme 1. Reagents and conditions: (a) 3-(2-chloro-6-fluorophenyl)-5-methyl-4-isoxazolyl chloride, Et3N, CH2Cl2, 25 ꢁC, 1 h (85%); (b) 2 N NaOH/
MeOH, DMF, 25 ꢁC, 2 h (94%); (c) i—(COCl)2, CH2Cl2, 1 h (100%); ii—3,4,5-trimethoxyaniline, Et3N, CH2Cl2, 15 h (SCX column) (89%); (d) i—
TFA, 1 h; ii—3,4,5-trimethoxybenzoyl chloride, Et3N, CH2Cl2, 25 ꢁC, 1 h (81%).

5528
B. H. Norman et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5526–5530
Cl
We became interested in pursuing a different amide ori-
entation in the linker region. Thus, we prepared targets
19–22 (Scheme 3) in a similar fashion and separated
the enantiomers of the more active cis analogs.11,12
The in vitro data are shown in Table 2.13
EtO
O
N
CO2Et
CO2Et
O
NO2
NH2
N
a
b,c
O
12
(~ 3:1 syn:anti)
13
11
Once again, cis diastereomers 19 and 21 proved more
potent than the trans isomers 20 and 22 in both
HeLa-T5 and LTC4 transport assays. Additionally,
we found that in this amide series, simple phenyl
was more potent than the trimethoxyphenyl substitu-
tion. For both series, the cis diastereomers were sepa-
rated into their enantiomers 19a/b and 21a/b.14 In
addition to its exquisite activity in these assays, the
more active phenyl substituted enantiomer 21b showed
excellent selectivity (>1000-fold) in the HL60 panel of
MRP1- overexpressing (HL60/Adr) and Pgp-over-
expressing (HL60/Vinc) cells.15
d,e,f
OMe
OMe
MeO
OMe
O
MeO
OMe
Cl
Cl
HN
HN
O
N
N
O
O
N
O
N
O
14
15
Scheme 2. Reagents and conditions: (a) H2, Rh on C, 60 psi, EtOH,
60 ꢁC, 18 h (72%); (b) 3-(2-chloro-6-fluorophenyl)-5-methyl-4-isoxaz-
olyl chloride, Et3N, CH2Cl2, 25 ꢁC, 1 h (85%); (c) KO-t-Bu, DMF,
25 ꢁC, 18 h (90%); (d) 1 N NaOH, MeOH (98%); (e) (COCl)2, CH2Cl2,
1 h (100%); (f) 3,4,5-trimethoxybenzoyl chloride, Et3N, CH2Cl2, 25 ꢁC,
1 h (86%).
NHBOC
NHBOC
NO2
NH2
a
linkers would provide a significant harvest of ligand
orientation information.
b,c
16
(~ 3:1 syn:anti)
We prepared the cyclohexyl analogs 14 and 15, starting
with ethyl 3-nitrophenylacetate (11, Scheme 2).11 Satu-
rating hydrogenation conditions provided the cyclohex-
yl amino ester 12 as a 3:1 mixture (syn:anti) of
inseparable isomers.12 Acylation of the amine mixture,
followed by intramolecular nucleophilic aromatic substi-
tution, produced the tricyclic ester 13. Standard sapon-
ification conditions, followed by acid chloride
formation and amide preparation, provided the targets
14 and 15 as a 3:1 mixture of diastereomers. These mate-
rials were easily separated by flash chromatography.
Additionally, each diastereomer was separated into its
pure enantiomers (14a and b; 15a and b) via chiral
HPLC (Chiralpak AD).
Cl
Cl
Ar
N
N
O
NH
NHBOC
d,e
O
O
N
N
O
O
19, Ar = 3,4,5-(OMe)3Ph
21, Ar = Ph
17
Cl
Cl
N
Ar
N
NHBOC
O
NH
O
d,e
O
N
N
O
O
18
20, Ar = 3,4,5-(OMe)3Ph
22, Ar = Ph
The in vitro data on cyclohexyl-linked compounds 14
and 15 are shown in Table 1.13 While all four com-
pounds inhibited MRP1, 14b showed greater potency
in the HeLa-T5 assay and significantly greater activity
in the transport assay, which is a direct measure of the
compoundÕs ability to inhibit the ATP-dependent trans-
port of an MRP1 substrate into membrane vesicles.
Additionally, we noted a clear enantiospecificity in the
in vitro results. Specifically, 14b was ꢀ4· more potent
in cells and up to 23· more potent in the LTC4 transport
assay, relative to 14a.
Scheme 3. Reagents and conditions: (a) H2, Rh on C, 60 psi, EtOH,
60 ꢁC, 18 h (81%); (b) 3-(2-chloro-6-fluorophenyl)-5-methyl-4-isoxaz-
olyl chloride, Et3N, CH2Cl2, 25 ꢁC, 1 h (82%); (c) KO-t-Bu, DMF,
25 ꢁC, 18 h (86%); (d) TFA, 1 h (100%); (e) aroyl chloride, Et3N,
CH2Cl2, 25 ꢁC, 1 h (71–86%).
Table 2. In vitro properties of 19–22a
Compound
EC50 (lM) HeLa-T5 IC50 (lM) LTC4 transport
19 (racemic) 0.25
20 (racemic) 0.42
0.27
NT
Table 1. In vitro properties of 14 and 15a
21 (racemic) 0.18 ( 0.016)
22 (racemic) 0.64
0.038 ( 0.004)
1.014
0.78
Compound EC50 (lM) HeLa-T5 IC50 (lM) LTC4 transport
14a
14b
15a
15b
1.10 ( 0.13)
0.256 ( 0.021)
1.63
1.86 ( 0.545)
0.079 ( 0.038)
0.849
19a
19b
21a
21b
0.56
0.16
0.17 ( 0.07)
0.545 ( 0.159)
0.064 ( 0.022)
0.43 ( 0.015)
0.093 ( 0.004)
0.344
0.637
a Standard errors are shown in parentheses. Others tested in duplicate.
a Standard errors are shown in parentheses. Others tested in duplicate.

B. H. Norman et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5526–5530
5529
21b & Vincristine Against HeLa T5 in CD1 NU/NU Mice
1400
1200
1000
800
600
400
200
0
Control
Vincristine @ 0.5 mg/kg
21b @ 10 mg/kg
21b @ 8 mg/kg + Vin. 0.5 mg/kg
21b @ 4 mg/kg + Vin. 0.5 mg/kg
21b @ 2 mg/kg + Vin. 0.5 mg/kg
21b @ 1 mg/kg + Vin. 0.5 mg/kg
RX
0
5
10
15
20
25
30
DAYS POST IMPLANT
Figure 3. Efficacy of 21b and vincristine in HeLa-T5 antitumor model. Compound 21b: BID · 5, PO, 30 min before and after vincristine dosing.
Vincristine: qd · 5, iv.
useful tools, which may hasten the understanding of
MRP1-mediated drug resistance.
21b
HL60/Adr
HL60/Vinc
Selectivity = 1115·
EC50 = 0.002 lM
EC50 = 2.23 lM
References and notes
Due to its interesting in vitro properties, we decided to
explore 21b in our in vivo model for MRP1-mediated
drug resistance. As previously described,3 we have devel-
oped an MRP1-dependent in vivo model by implanting
the drug resistant HeLa-T5 cell line into nude mice, pro-
ducing tumors which are resistant to the MRP1-associ-
ated oncolytic, vincristine.13 The vector control cell
line HeLa-C1, which does not overexpress MRP1, was
similarly implanted into nude mice and found to be
responsive to vincristine treatment. Thus, a successful
MRP1 inhibitor should have a synergistic effect on
reducing tumor growth when dosed in combination with
vincristine in the HeLa-T5 in vivo model. Figure 3
shows the results of an in vivo study using 21b and vin-
cristine. Animals were dosed orally with 8, 4, 2, and
1 mg/kg 21b, 30 min before and after a bolus iv infusion
(0.5 mg/kg) of vincristine, for five days. Additionally,
several controls were included in this study. The vehicle
control shows the tumor growth with no therapy, while
the vincristine-treated animals (0.5 mg/kg, iv) showed
the effect of drug without inhibitor. A final control
was performed where the animals were dosed with inhib-
itor alone. This curve indicates that the inhibitor has no
antitumor activity of its own. As shown in Figure 3, the
combination treatment of vincristine and 21b gave a sta-
tistically significant improvement in efficacy, relative to
the controls, at all doses. The effect was dose responsive.
Even more noteworthy is the demonstration of tumor
regression at the higher doses of 21b: 63% regression
at the 8 mg/kg dose (green line) and 40% regression at
the 6 mg/kg dose (blue line).
1. Bellamy, W. T.; Dalton, W. S. Adv. Clin. Chem. 1994,
31, 1.
2. Norman, B. H.; Dantzig, A. H.; Kroin, J. S.; Law, K. L.;
Tabas, L. B.; Shepard, R. L.; Palkowitz, A. D.; Hauser, K.
L.; Winter, M. A.; Sluka, J. P.; Startling, J. J. Bioorg. Med.
Chem. Lett. 1999, 9, 3381.
3. Norman, B. H.; Gruber, J. M.; Hollinshead, S. P.; Wilson,
J. W.; Starling, J. J.; Law, K. L.; Self, T. D.; Tabas, L. B.;
Williams, D. C.; Paul, D. C.; Wagner, M. M.; Dantzig, A.
H. Bioorg. Med. Chem. Lett. 2002, 12, 883.
4. Endicott, J. A.; Ling, V. Ann. Rev. Biochem. 1989, 58, 137.
5. Cole, S. P. C.; Sparks, K. E.; Fraser, K.; Loe, D. W.;
Grant, C. E.; Wilson, G. M.; Deeley, R. G. Cancer Res.
1994, 54, 5902.
6. Jedlitschky, G.; Leier, I.; Buchholz, U.; Center, M.;
Keppler, D. Cancer Res. 1994, 54, 4833.
7. Robert, J. Drugs Future 1997, 22, 149.
8. Cole, S. P. C.; Bhardwaj, G.; Gerlach, J. H.; Mackie, J. E.;
Grant, C. E.; Almquist, K. C.; Stewart, A. J.; Kurz, E. U.;
Duncan, A. M. V.; Deeley, R. G. Science 1992, 258, 1650.
9. Norman, B. H. Drugs Future 1998, 23, 1001.
10. Gruber, J. M.; Norman, B. H. U.S. Patent 6 417 193, 2002.
11. Bonjouklian, R.; Cohen, J. D.; Gruber, J. M.; Johnson, D.
W.; Jungheim, L. N.; Kroin, J. S.; Lander, P. A.; Lin, H.;
Lohman, M. C.; Muehl, B. S.; Norman, B. H.; Patel, V.
F.; Richett, M. E.; Thrasher, K. J.; Vepachedu, S.; White,
W. T.; Xie, Y.; York, J. S.; Parkhurst, B. L.; Wang, Q.
U.S. Patent 6 743 794, 2004.
12. The relative stereochemistry of the syn and anti bis-
substituted cyclohexyl derivatives (14, 15, and 19–21) was
determined using spectroscopic methods and will be
described in a forthcoming publication.
13. Assay conditions have been previously described. See
Ref. 3 (and references cited therein).
14. An enantiospecific synthesis has been developed and the
absolute stereochemistries have been determined for
compounds in the cyclohexyl-linked series. This work will
be described in a forthcoming publication.
In conclusion, we have shown that tricyclic isoxazoles
continue to be a promising scaffold for selective MRP1
modulation. Specifically, compounds, such as 21b, are

5530
B. H. Norman et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5526–5530
15. Modulator EC50 in HL60/Adr and HL60/Vinc cells were
determined by assessing the concentration-dependent abil-
ity of the modulator to enhance the antiproliferative
response of the cells to an IC20 concentration of doxorubicin
using alamarBlueꢂ reduction as a surrogate measure for cell
number. The IC20 concentrations for doxorubicin used are
1.0 lg/ml for HL60/Adr line and 0.3 lg/ml for the HL60/
Vinc line. The dox IC50 values for these cell lines are 4.0 lg/
ml for HL60/Adr and 1.0 lg/ml for HL60/Vinc. The
selectivity ratio was calculated by dividing the average
EC50 for HL60/Vinc cells bythe average EC50 for HL60/Adr
cells.