A structure-guided approach to an orthogonal estrogen-receptor-based gene switch activated by ligands suitable for in vivo studies

Article abstract of DOI:10.1021/jm060516e

A strategy to obtain a fully orthogonal estrogen-receptor-based gene switch responsive to molecules with acceptable pharmacological properties is presented. From a series of tetrahydrofluorenones active on the wild-type estrogen receptor (ER) an inactive

Full text of DOI:10.1021/jm060516e

5404
J. Med. Chem. 2006, 49, 5404-5407
Letters
A Structure-Guided Approach to an Orthogonal
Estrogen-Receptor-Based Gene Switch Activated
by Ligands Suitable for in Vivo Studies
Olaf Kinzel,* Daniela Fattori, Ester Muraglia,
Paola Gallinari, Maria Chiara Nardi, Chantal Paolini,
Giuseppe Roscilli, Carlo Toniatti, Odalys Gonzalez Paz,
Ralph Laufer, Armin Lahm, Anna Tramontano,
Riccardo Cortese, Raffaele De Francesco,
Figure 1. Inactive chosen lead compound 1a and active analogue 1b.
Gennaro Ciliberto, and Uwe Koch
IRBM (Merck Research Laboratories Rome),
Via Pontina km 30,600, 00040 Pomezia, Rome, Italy
ReceiVed May 2, 2006
Abstract: A strategy to obtain a fully orthogonal estrogen-receptor-
based gene switch responsive to molecules with acceptable pharma-
cological properties is presented. From a series of tetrahydrofluorenones
active on the wild-type estrogen receptor (ER) an inactive analogue is
chosen as a new lead compound. Coevolution of receptor mutants and
ligands leads to an ER-based gene switch suitable for studies in animal
models.
Figure 2. (a) Superposition of estradiol and 1a and (b) residues most
likely to interfere with binding of 1a (bold). Conserved polar interactions
used to define the position of the tetrahydrofluorenone are also shown
(italic).
In the field of gene therapy an important goal is to have tight
temporal and dose control over the expression of an introduced
gene (transgene). This can be achieved by controlling the
expression of the transgene with a small molecule (inducer) with
well-behaved pharmacokinetic and pharmacodynamic properties.
In the presence of the inducer the ligand-dependent transcription
system should be activated in a dose-dependent manner, whereas
in its absence a low basal expression of the transgene should
occur.1,2 We recently published the development of an estrogen
receptor (ER) based gene switch fulfilling all the desired criteria
for such a system, namely, the complete orthogonality of the
artificial receptor and the exogenous ligand to the endogenous
ligand/receptor pair and a low immunogenicity.3 In this paper
we describe the medicinal chemistry effort of this work
culminating in the identification of highly selective ligands,
active on an estradiol-unresponsive ER mutant and suitable for
studies in vivo.
estradiol; (6) second round optimization of the ligands in terms
of selectivity for the final mutant.
Compound 1a (Figure 1) belongs to a series of tetrahydro-
fluorenones, which were prepared as part of our â-selective ER
modulator (SERM) project.4,5 1a shows weak binding affinity
for hERR and hERâ (IC50 > 10 µM for both subtypes).
Members of this series of benzopyrazole analogues had dem-
onstrated good oral bioavailability and an acceptable clearance
in several species (data not shown). In contrast, 1b (Figure 1)
is an active tetrahydrofluorenone with high affinity for hERâ.
A comparison of both compounds shows that the replacement
of the ethyl group in position 9a of 1b by the benzyl group is
responsible for the loss of affinity for 1a. This replacement
increases the volume of the ligand by 54 Å3.
The strategy applied in order to obtain an orthogonal gene
switch consisted of six steps: (1) identification of inactive
compounds within an active series, focusing on compounds
where the activity was abolished because of minor structural
changes; the compounds should also have favorable pharma-
cological properties; (2) determination of the residues of the
protein responsible for the loss of activity of the chosen
compound through molecular modeling; (3) generation of a
mutant library of the ER ligand binding domain (LBD),
containing mutations at the previously identified amino acid
positions and selection of mutants with increased affinity for
the low-affinity ligand;3 (4) optimization of the chosen lead
compound in terms of selectivity profile; (5) introduction of an
additional mutation to abolish the remaining affinity for
To identify the residues that interfere with binding of the
benzyl group, we built models of 1a and 1b bound to the hERR
and hERâ LBDs using X-ray structures of both LBDs with
bound agonists.6,7 For this purpose the tetrahydrofluorenone
scaffold was superimposed with the estradiol scaffold (Figure
2a). Severe steric clashes between the benzyl group of 1a and
five hydrophobic amino acid residues in the binding pocket were
identified. These amino acids were selected as sites of replace-
ments by other hydrophobic amino acids in the generation of
an ERR-LBD mutant library (together with the mutation L384M
to mimic the ERâ subtype) (Figure 2b). Recently a crystal
structure of a similar tetrahydrofluorenone has been published
that is in good agreement with the superposition shown in Figure
2.8
The designed LBD-mutant library was expressed in yeast,
and sequence analysis of mutants which conferred sufficient
* To whom correspondence should be addressed. Phone: +39 069 109
3334. Fax: +39 069 109 3654. E-mail: olaf_kinzel@merck.com.
10.1021/jm060516e CCC: $33.50 © 2006 American Chemical Society
Published on Web 08/05/2006

Letters
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 18 5405
Figure 4. Compound 7f and raloxifene.
Figure 3. Model of 1a in the estrogen pocket. The solvent accessible
surface area was calculated for Gly421, showing the large pocket that
becomes available to accommodate the benzyl moiety of 1a. Met421
of the wild-type receptor is shown for comparison.
as 2-naphthylmethyl were not tolerated (7e). Replacements of
the pyrazole ring were investigated as well. Compound 15a,
the triazole analogue of 7a, showed an 8-fold lower affinity,
but the introduction of a 4-fluoro substituent on the benzyl ring
largely restored affinity and resulted in 15b with an IC50 of
300 nM on MG-ERR and no activity on hERR and hERâ up to
10 µM. Finally, replacement of the pyrazole by a 7-hydroxy
group resulted in compounds with the highest affinity and
selectivity. Introduction of a chloro substituent in the 4-position
on the benzyl ring yielded 20b as the most potent and selective
compound (>300-fold selective for MG-ERR over wtERR-
LBD).
Table 1. SAR of Tetrahydrofluorenones
To abolish the remaining affinity of MG-ERR for estradiol,
the additional mutation G521R was introduced, leading to a
triple-mutated ERR-LBD (called MGR-ERR). It is known in
the literature that estrogen receptors that contain the mutation
G521R in the LBD exhibit a strongly reduced affinity for
estradiol but remain responsive to certain ER antagonists such
as 4-OH-tamoxifen and raloxifene.9,10
To test our compounds on MGR-ERR, a functional assay in
mammalian cells was set up. (Affinity determination of our
ligands on MGR-ERR by a competitive radiometric binding
assay was hampered by the loss of affinity for radiolabeled
estradiol.) For this purpose MGR-ERR was grafted into our
previously described gene switch HEA-111 (replacement of the
HEA-1-LBD with MGR-ERR), which was transfected in HeLa
cells.3 In the presence of the ligands the expression of a reporter
protein (SEAP, human-secreted alkaline phosphatase) was
activated, protein concentrations were measured at different
ligand concentrations, and the EC50 values were determined.
The MGRR-containing gene switch was unresponsive to estra-
diol up to 10 µM, but to our disappointment, when 7a was tested
in this assay, the gene switch also remained unresponsive up to
2 µM, suggesting a substantial loss of affinity of 7a for MGR-
ERR.
a IC50 values are the mean of three or more experiments. b Full-length
receptor. c Estradiol. d Not determined. e IC50 on the wtERR-LBD. f Only
one experiment.
growth in the presence of 1a revealed the presence of one
predominant mutation in the ligand binding pocket. Methionine
421 was replaced by the much smaller glycine (ERR-LBD
containing mutations L384M/M421G, called MG-ERR). The
IC50 of 1a on the isolated ligand binding domain MG-ERR was
found to be 3.9 µM. This increase in affinity can be rationalized
by an increase of the volume of the binding pocket by 66 Å3,
matching the 54 Å3 increase of the volume of 1a compared to
1b (Figure 3). As anticiptated, MG-ERR still exhibited high
affinity for estradiol, although being reduced by at least 1 order
of magnitude (Table 1).
Keeping in mind that MG-ERR would not be our final mutant,
we started an effort to improve affinity and selectivity of lead
compound 1a for MG-ERR. Meanwhile additional mutations
were evaluated for insertion into MG-ERR to abolish the
remaining affinity for estradiol. We reasoned that a compound
with high affinity and selectivity for MG-ERR would most likely
serve as a lead for a SAR study on the final mutant.
A summary of the SAR on MG-ERR with 1a as the lead
compound is given in Table 1. The introduction of a methyl
group at the 6-position resulted in 7a with a 17-fold increase in
affinity for MG-ERR, being inactive on hERR and displaying
an 8-fold selectivity over hERâ. The exploration of the SAR
around the benzyl substituent resulted in the identification of
7b with an IC50 of 82 nM on MG-ERR and a slightly improved
selectivity over hERâ (>12-fold). More bulky substituents such
Undeterred by this result, we synthesized a hybrid molecule
of 7a and raloxifene, yielding 7f (Figure 4).
When 7f was tested on MGR-ERR in the cell-based assay,
an EC50 of 300 nM was obtained. But at the same time 7f
exhibited an increased affinity for hERR and hERâ compared
to 7a (Table 2). Encouraged by this result, we concentrated on
compounds with bulky R2 side chains, and a SAR study was
performed to restore selectivity versus hERR/hERâ.
For practical reasons we continued to measure affinity for
MG-ERR, while interesting compounds were further tested in
the cell-based assay on MGR-ERR. A summary of the results
is given in Table 2. The hydroxy compound 20c showed a much
improved selectivity (defined as IC50 ratios of MG-ERR/hERR
or MG-ERR/hERâ) compared to 7f (17-fold/8-fold against
6-fold/1.2-fold), and a small SAR study around the side chain
R2 in the hydroxy series identified 20g with still further
improved selectivity (30-fold/14-fold). As expected, installation
of a chloro substituent at the 4-position of the benzyl ring (20h)
gave the most selective compound (38-fold/76-fold), confirming

5406 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 18
Table 2. SAR of Tetrahydrofluorenones with Bulky Side-chains R2
Letters
Scheme 2a
a Reagents and conditions: (a) Br2, HOAc, 10 °C; (b) CrO3, HOAc, 15-
17 °C; (c) HNO3 (100%), -40°C; (d) 6 N HCl, MeOH, reflux; (e) Pd/C,
H2, NaOAc, EtOAc; (f) NaNO2, HCl, EtOH, 0-5°C; (g) ArCHO, NaOMe,
MeOH; (h) Pd/C, H2; (i) ethyl vinyl ketone, NaOMe, MeOH, 60°C; (j) 6
N aqueous HCl/HOAc, 1:1, 80 °C.
Scheme 3a
a IC50 values are the mean of three or more experiments. b Full-length
receptor. c Not determined.
Scheme 1a
a Reagents and conditions: (a) ArCHO, NaOMe, MeOH; (b) Pd/C, H2,
EtOAc for 17a,c; Se, NaBH4, EtOH/THF, 0-50 °C for 17b,h; (c) ethyl
vinyl ketone, NaOMe, MeOH, 60 °C for 17a,b; methyl vinyl ketone, DBU,
THF, 55 °C for 17c,h; (d) 6 N aqueous HCl/HOAc, 1:1, 80 °C, for 17a,b;
pyrrolidine, HOAc, THF, 55 °C, for 17c,h; (e) Br, NaHCO3, CCl4, 0 °C;
(f) aryl-SnBu3, Pd(PPh3)4, toluene, 100 °C; (g) MeOH/2 N aqueous HCl,
5:1, 60°C; (h) amino alcohol, DIAD, PPh3, THF, 0-25°C; (i) AlCl3,
2-propanethiol, CH2Cl2.
a Reagents and conditions: (a) NBS, CH3CN, 60°C; (b) PdCl2(PPh3)2,
PPh3, LiCl, SnMe4, DMF, 100°C; (c) ArCHO, NaOMe, MeOH; (d) Pd/C,
H2; (e) ethyl vinyl ketone, NaOMe, MeOH, 60°C; (f) 6 N aqueous HCl/
HOAc, 1:1, 80°C; (g) (i) NOBF4, CH2Cl2, -35 to 4 °C; (ii) KOAc, db-
18-crown-6, CH2Cl2, -35°C to room temp.
The synthesis of the pyrazole-containing ligands followed the
strategy developed for the synthesis of the hER-â-active
4,5
compounds
and is summarized in Scheme 1. Selective
bromination of N-protected 5-amino-1-indanone (2) at the
4-position with NBS and subsequent replacement of the bromo
substituent by a methyl group via Stille cross-coupling led to
intermediate 4. Introduction of the alkyl substituents R1 was
achieved by an aldol condensation with the corresponding
arylaldehydes and a subsequent catalytic reduction of the double
bond. A two-step Robinson annulation with ethyl vinyl ketone
and a concomitant N-deacetylation led to intermediates 5a-g.
The final pyrazole ring formation was achieved under mild
conditions by formation of the diazonium tetrafluoroborate
intermediates, which were then cyclized under phase-transfer
conditions.12
the results obtained in the series lacking the bulky side chain.
20c, 20e, and 20h were tested in the cell-based assay on MGR-
ERR (Table 2). Their EC50 values correlated well with their
IC50 values measured on MG-ERR, suggesting that the ad-
ditional mutation present in MGR-ERR has no detrimental effect
on the affinity of ligands with bulky side chains R2. The data
obtained in mammalian cells also suggest a good cell penetration
of the molecules.
Compounds 20c, 20e, and 20h were evaluated for their
suitability for in vivo studies: 20c and 20e were dosed iv in
rats (dose 3 mg/kg). 20c displayed a moderate clearance of 22
mL min-1 kg-1 and a terminal half-life of 2 h. Meanwhile, 20e
displayed a clearance of 39 mL min-1 kg-1 and a terminal half-
life of 1.5 h. The plasma concentration of 20c after 6 h was
more than 10-fold its EC50 in mammalian cells, and that of 20e
was more than 4-fold. Compound 20h was dosed in male Balb-c
mice (4 mg/kg, ip). The plasma concentration after 4 h was
1.5-fold its EC50 in mammalian cells with a Cmax of 0.5 µM
after 30 min. These data demonstrate that the obtained class of
compounds is suitable for in vivo studies in animal models,
ideally using ip dosing.
The triazole-containing ligands were synthesized as shown
in Scheme 2. To obtain the key intermediate 12, a transient
protection of the 6-position of 8 with a bromo substituent was
necessary because the alternative direct nitration of 8 or 2 gave
mixtures of regioisomers that were difficult to separate. After
formation of the triazole ring the introduction of the alkyl
substituents R1 was achieved in the same manner as in the
pyrazole series by aldol condensation with the corresponding
arylaldehydes and subsequent reduction of the double bond. A
two-step Robinson annulation led to the final triazole com-

Letters
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 18 5407
receptor-based transcription switch specifically induced by a non-
steroid synthestic ligand. Chem. Biol. 2005, 12, 883-893.
(4) Parker, D. L., Jr.; Wildonger, K. J.; Meng, D.; Ratcliffe, R. W.
Preparation of 1,2,9,9a-tetrahydro-3H-fluoren-3-ones and related
compounds as estrogen receptor modulators. PCT Int. Appl. WO
2002041835 A2, 2002.
(5) Parker, D. L., Jr.; Ratcliffe, R. W.; Wilkening, R. R.; Wildonger, K.
J. Preparation of 1,2,9,9a-tetrahydro-3H-fluoren-3-ones and related
compounds as estrogen receptor modulators. PCT Int. Appl. WO
2001082923 A1, 2001.
(6) Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard, R. E.; Bonn,
T.; Engstrom, O.; Ohman, L.; Greene, G. L.; Gustafsson, J. A.;
Carlquist, M. Molecular basis of agonism and antagonism in the
oestrogen receptor. Nature 1997, 389, 753-758.
(7) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner, P. J.;
Agard, D. A.; Greene, G. L. The structural basis of estrogen receptor/
coactivator recognition and the antagonism of this interaction by
tamoxifen. Cell 1998, 95, 927-937.
(8) Wilkening, R. R.; Ratcliffe, R. W.; Tynebor, E. C.; Wildonger, K.
J.; Fried, A. K.; Hammond, M. L.; Mosley, R. T.; Fitzgerald, P. M.
D.; Sharma, N.; McKeever, B. M.; Nilsson, S.; Carlquist, M.;
Thorsell, A.; Locco, L.; Katz, R.; Frisch, K.; Birzin, E. T.; Wilkinson,
H. A.; Mitra, S.; Cai, S.; Hayes, E. C.; Schaeffer, J. M.; Rohrer, S.
P. The discovery of tetrahydrofluorenones as a new class of estrogen
receptor â-subtype selective ligands Bioorg. Med. Chem. Lett. 2006,
16, 3489-3494.
(9) Danielian, P. S.; White, R.; Hoare, S. A.; Fawell, S. E.; Parker, M.
G. Identification of residues in the estrogen receptor that confer
differential sensitivity to estrogen and hydroxytamoxifen. Mol.
Endocrinol. 1993, 7, 232-240.
(10) Littlewood, T. D.; Hancock, D. C.; Danielian, P. S.; Parker, M. G.;
Evan, G. I. A modified oestrogen receptor ligand-binding domain as
an improved switch for the regulation of heterologous proteins.
Nucleic Acids Res. 1995, 23, 1686-1690.
(11) Roscilli, A. G.; Rinaudo, C. D.; Cimino, M.; Sporeno, E.; Lamartina,
S.; Ciliberto, G.; Toniatti, C. Long-term and tight control of gene
expression in mouse skeletal muscle by a new hybrid human
transcription factor. Mol. Ther. 2002, 6, 653-663.
(12) Bartsch, R. A.; Yang, I.-W. Phase transfer catalyzed synthesis of
indazoles from o-alkylbenzenediazonium tetrafluoroborates. J. Het-
erocycl. Chem. 1984, 21, 1063-1064.
(13) Schwimmer, L. J.; Rohatgi, P.; Azizi, B.; Selely, K. L.; Doyle, D. F.
Creation and discovery of ligand-receptor pairs for transcriptional
control with small molecules. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 14707-14712.
(14) Chockalingam, K.; Chen, Z.; Katzenellenbogen, J. A.; Zhao, H.
Directed evolution of specific receptor-ligand pairs for use in the
creation of gene switches. Proc. Natl. Acad. Sci. U.S.A. 2005, 102,
5691-5696.
(15) Chockalingam, K.; Zhao, H. Creating new specific ligand-receptor
pairs for transgene regulation. Trends Biotechnol. 2005, 23, 333-335.
pounds, although in low yields because of a partial conjugate
addition of ethyl vinyl ketone to the unprotected triazole ring.
The series of hydroxytetrahydrofluorenones was prepared
mostly according to the synthesis of the hER-â-active com-
pounds (Scheme 3).4,5
In summary we have successfully demonstrated an approach
for the construction of an orthogonal gene switch, responsive
to molecules with acceptable pharmacological properties. The
essence of this approach lies in the early consideration of the
properties of the ligand and the subsequent construction of a
suitable receptor for the chosen ligand. Similar to our work,
directed evolution approaches have been successfully applied
in the creation of highly orthogonal gene switches.13-15 In these
studies though the pharmacological properties of the ligands
were not taken into consideration. By contrast, our concept
consists of the coevolution of the receptor and of the ligand.
To our knowledge the work presented here is the first example
where the ligands of a completely orthogonal estrogen-receptor-
based gene switch are small molecules with modest clearance
and acceptable half-life in vivo, suitable for studies in animal
models.
Acknowledgment. The authors thank Ronald W. Ratcliffe
and Robert R. Wilkening for their helpful collaboration regard-
ing the synthesis of the ligands, Fabio Bonelli and Vincenzo
Pucci for the recording of HR MS spectra, and Michael Rowley
and Philip Jones for their critical review of the manuscript. This
work was funded in part by a grant from the MIUR.
Supporting Information Available: Experimental procedures
and characterization of intermediates and final compounds. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Toniatti, C.; Bujard, H.; Cortese, R. Gene therapy progress and
prospects: transcription regulatory systems. Gene Ther. 2004, 11,
649-657.
(2) Weber, W.; Fussenegger, M. Artificial mammalian gene regulation
networkssnovel approaches for gene therapy and bioengineering.
J. Biotechnol. 2002, 98, 161-187.
(3) Gallinari, P.; Lahm, A.; Koch, U.; Paolini, C.; Nardi, M. C.; Roscilli,
G.; Kinzel, O.; Fattori, D.; Muraglia, E.; Toniatti, C.; Cortese, R.;
De Francesco, R.; Ciliberto, G. A functionally orthogonal estrogen
JM060516E