Allosteric Protein Sensors of Hormone Binding
A R T I C L E S
such compounds virtually undetectable by similar nontranscrip-
tional assays in yeast or bacteria. For example, estrone, 17R-
estradiol, and bisphenol A were unable to enhance cell growth
in our prototype bacterial estrogen-sensing system,12 while
genistein, estriol, and other lower-affinity binders were found
inactive in a similar yeast assay based on a chimeric fusion of
ERR with dihydrofolate reductase.24 All of these compounds,
however, produced clear changes in growth in the pMIT::ERâ*
sensor system.
saturating concentration of E2 (500 nM as opposed to 10 µM
used in Figure 2) was used to allow a clear determination of
any growth enhancement arising from agonistic behavior of the
test compounds. As it was previously observed with our
prototype estrogen sensor,12 the addition of all of the estrogen
agonists in E2-supplemented -THY medium had an additive
effect on growth, while the addition of the antiestrogens
tamoxifen, 4-hydroxytamoxifen, and clomiphene was found to
have an inhibitory effect on growth (Figure 4A). In TTM
medium, growth phenotypes were inverted: the presence of
estrogen agonists suppressed growth, whereas antagonist addi-
tion enhanced bacterial growth (Figure 4B). Notably, with the
ERâ sensor, raloxifene and ICI 182,780 were also able to
antagonize E2, although they could not be detected and/or
correctly identified as antagonists by the ERR-based sensor.12
However, the full antagonistic effect of ICI 182,780 could not
be observed at the tested concentrations, presumably reflecting
that its affinity of binding to ERâ, as determined using standard
methodology (see Experimental Section for details), is 40-fold
lower than that of E2, whereas that of, e.g., 4-hydroxytamoxifen
is only 4.5-fold lower. One cannot exclude, though, that
secondary effects (e.g., low diffusivity through the E. coli
membrane) might also modulate the inhibitory effect. Addition-
ally, antagonistic effects in this system could be detected at
concentrations above 5-10 µM with the pure estrogen antago-
nist ZK 164,015, as well as with the synthetic SERMs GW 5638,
tetrahydrochrysene and PPT (data not shown). PPT exhibits a
subtype-selective pharmacological profile, activating ER signal-
ing via ERR, while acting as an estrogen antagonist through
ERâ at high concentrations.8,34 These observations, together with
the fact that estrogen analogues have been shown not to interfere
with cell growth in the absence of the ER sensor protein,12
provide strong evidence that antiestrogens directly antagonize
the ability of E2 to bind to ERâ and enhance the TS activity of
the chimeric sensor.
Molecular Basis of Agonist-Antagonist Discrimination.
Crystallographic studies have revealed the structural basis of
the functional differences between NHR agonists and antago-
nists. These studies suggest that agonist binding induces a
structural shift in the LBD that allows the C-terminal helix 12
to close as a lid over the hormone-binding cavity and expose a
surface of the receptor that can recruit coactivators and initiate
signaling (Figure 5A).5,6,8,35 Antagonistic compounds on the
other hand, whose backbone structures resemble that of agonists
but contain bulky side chains that protrude from the binding
cavity, prevent helix 12 from closing properly over the binding
pocket. Instead, they yield a LBD structure where helix 12
extends outward (Figure 5A) and prevents coactivator recruit-
ment, while allowing weak interactions with corepressors.5,6,8,35
Note that no crystal structure of the apo form of either ER
subtype has yet been determined, and therefore the structural
model of the unliganded form of ER shown in Figure 5A is
conceptual and based on the determined structure of the apo
LBD of the retinoid X receptor R.
The sensitivity of our biosensor was further evaluated by
determining dose-response curves for various estrogenic
compounds at 34 °C (Figure 3A). This test revealed a half-
maximal effective concentration (EC50) for E2 of approximately
100 nM, about 100-fold lower than the concentration required
by the ERR-based sensor.12 In addition, all of the tested
compounds could be detected at nanomolar concentrations and
the higher-affinity ones at sub-nanomolar concentrations (Figure
3A). This sensitivity is comparable to that of previously reported
chimeric sensors for estrogen binding in yeast25,26 or in vitro,27
and to the sensitivity of other simple screening systems, such
as recently developed NHR microarrays of coactivator recruit-
ment.28 Interestingly, for low-affinity ligands, the sensitivity of
our system converges with that of highly sensitive in vitro
binding assays,29 and with some transcriptional activation assays
in genetically engineered yeast30,31 and mammalian cells.9,31,32
These features may render this system a particularly attractive
tool for the rapid screening of samples and extracts derived from
natural products or environmental sources. Because the EC50
values correlate well with binding affinity (Figure 3B), the
intensity of a given ligand-receptor interaction can be estimated
by comparing its EC50 value to those of known compounds.
Recognition of Pharmacological Properties. A general
limitation of simple screening systems for NHR modulators is
the inability to predict the pharmacological effect of a particular
hormone mimic. For example, in vitro competitive binding
assays without coactivator recruitment, as well as transcriptional
activation assays in yeast, have been unable to discriminate
between known NHR agonists and antagonists.33 This is a very
important characteristic for a screening system since most of
the clinically valuable hormone analogues that target ER, such
as tamoxifen, raloxifene, and ICI 182,780, exert their anticancer
therapeutic effects through their ability to antagonize estrogen
signaling.11
To investigate the ability of the ERâ-TS sensor to reliably
differentiate between agonistic and antagonistic effects, our
sensor strain was incubated in E2-containing -THY medium
at 34 °C, or TTM medium at 37 °C, and exposed to increasing
concentrations of known estrogen agonists, SERMs, or pure
estrogen antagonists (Figure 4). In these experiments, a sub-
(24) Tucker, C. L.; Fields, S. Nat. Biotechnol. 2001, 19, 1042-6.
(25) Muddana, S. S.; Peterson, B. R. ChemBioChem 2003, 4, 848-55.
(26) Kohler, F.; Zimmermann, A.; Hager, M.; Sippel, A. E. Gene 2004, 337,
113-9.
(27) De, S.; Macara, I. G.; Lannigan, D. A. J. Steroid Biochem. Mol. Biol. 2005,
96, 235-44.
(28) Kim, S. H.; Tamrazi, A.; Carlson, K. E.; Katzenellenbogen, J. A. Mol.
Cell. Proteomics 2005, 4, 267-77.
It has been proposed that the hormone-regulated activities
of steroid-LBD fusions with heterologous proteins expressed
(29) Kim, S. H.; Tamrazi, A.; Carlson, K. E.; Daniels, J. R.; Lee, I. Y.;
Katzenellenbogen, J. A. J. Am. Chem. Soc. 2004, 126, 4754-5.
(30) Breinholt, V.; Larsen, J. C. Chem. Res. Toxicol. 1998, 11, 622-9.
(31) Breithofer, A.; Graumann, K.; Scicchitano, M. S.; Karathanasis, S. K.; Butt,
T. R.; Jungbauer, A. J. Steroid Biochem. Mol. Biol. 1998, 67, 421-9.
(32) Miksicek, R. J. J. Steroid. Biochem. Mol. Biol. 1994, 49, 153-60.
(33) Joyeux, A.; Balaguer, P.; Germain, P.; Boussioux, A. M.; Pons, M.; Nicolas,
J. C. Anal. Biochem. 1997, 249, 119-30.
(34) Stauffer, S. R.; Coletta, C. J.; Tedesco, R.; Nishiguchi, G.; Carlson, K.;
Sun, J.; Katzenellenbogen, B. S.; Katzenellenbogen, J. A. J. Med. Chem.
2000, 43, 4934-47.
(35) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner, P. J.; Agard,
D. A.; Greene, G. L. Cell 1998, 95, 927-37.
9
J. AM. CHEM. SOC. VOL. 129, NO. 27, 2007 8447