Estrogen Receptor Microarrays: Subtype-Selective Ligand Binding

Article abstract of DOI:10.1021/ja039586q

We present the first example of a nuclear hormone receptor microarray, using for illustration the ligand-binding domains of the two estrogen receptors, ERα-LBD and ERβ-LBD. The proteins are printed and allowed to attach to aldehyde slides; the efficiency

Full text of DOI:10.1021/ja039586q

Published on Web 03/25/2004
Estrogen Receptor Microarrays: Subtype-Selective Ligand Binding
Sung Hoon Kim, Anobel Tamrazi, Kathryn E. Carlson, Jonathan R. Daniels, In Young Lee, and
John A. Katzenellenbogen*
Department of Chemistry, UniVersity of Illinois, 600 South Matthews AVenue, Urbana, Illinois 61801
Received November 13, 2003; E-mail: jkatzene@uiuc.edu
The two estrogen receptors, ERR and ERâ, are ligand-regulated
transcription factors of the nuclear receptor (NR) class whose
activity is modulated by the binding of ligands and interaction with
coregulators.1,2 Various methods have been developed to assay these
interactions,3,4 but only some of these are sufficiently convenient
for high-throughput analysis. Here, we describe the preparation of
protein microarrays of ERR and ERâ ligand-binding domains
(LBDs) on glass slides; the ERs on these arrays remain active and
can be used to assay the specific binding of different ligands in a
rapid and convenient manner. Although DNA microarrays are well
known, there are only a few examples of protein microarrays,5-7
because proteins are structurally diverse and often sensitive to
denaturation.6 The NRs are considered to be very sensitive proteins.
To follow the tethering of the ERs to glass, we used engineered
ERR and ERâ LBDs labeled site specifically at a singly reactive
cysteine residue between helices 7 and 8 (C417 in ERR and C369
in ERâ) with thiol-reactive fluorophores (tetramethylrhodamine
(MTMR) or Cy3, and Cy5) in a manner that preserves their binding
activity.8 We evaluated various commercially available slides
(amine, aldehyde, epoxide, nickel) and found attachment to aldehyde
slides to be most efficient (Figure 1A). Attachment was rapid (2-5
h), reproducible (CV < 5%), and linear over an extended
concentration range. Curiously, the level of ER attachment depended
on the liganded state of the ER-LBD, with more apo-ER and
antagonist-liganded ERR and ERâ (i.e., hydroxytamoxifen, TOT,
Figure 1. Attachment of the ER-LBD-ligand complexes to an aldehyde
or ICI 182,780) becoming attached than the agonist-liganded ERs
slide. Panel A: A microarray showing the differential attachment of ERR-
(i.e., either estradiol, E2, or diethylstilbestrol, DES; Figure 1A and
B, left).9 The attachment of the ERs was irreversible, and the level
Cy5 (red) and ERâ-MTMR (green) bound with various ligands. Agonist-
bound ERs (E2 and DES) show less attachment as compared to antagonist-
liganded (TOT and ICI) or apo-ER. Panel B: Quantification of the attach-
of attached receptor was unaffected by subsequent ligand treatment.
ment of ERR-Cy5 and ERâ-MTMR, and of nonfluorescent ERR-pre-bound
Most of the 11 lysines in ERR-LBD are folded deeply within
the protein,2,4 but two, K529 and K531, appear exposed and are
the most likely attachment sites to the aldehyde surface (Figure
1C). These lysines are in a region where conformation is markedly
ligand regulated: In agonist structures, both are tightly folded back
onto the protein surface (Figure 1C, green), whereas in antagonist
structures, K531 projects outward toward solvent and K529 remains
exposed but is somewhat closer to the protein surface (Figure 1C,
red).2,4 The exposure of K531 in ER-antagonist structures is
supported by its high sensitivity to trypsin proteolysis,10 and apo-
ER is believed to resemble antagonist-bound ER.10 Thus, the
attachment of ER to aldehyde slides appears to be specific, with
K531 in ERR and the corresponding K482 in ERâ the most likely
sites of attachment.
with [3H]E2 or [3H]TOT. Dots show the level of attachment to plain glass.
Panel C: A ribbon structure of the ERR-LBD, showing the position of the
two important lysines (K529 and K531 in ERR) when the ligand is the
agonist DES (green) or the antagonist TOT (red).
The ligand-binding activity of ER-LBDs, tethered to aldehyde
slides, can be measured using a fluorophore-estradiol conjugate in
which a Cy3 fluorophore has been attached through an oligoethylene
glycol spacer to a 17R-ethynylestradiol derivative (EE2-Cy3). EE2-
Cy3 binds to both ERR and ERâ in a manner that shows good
linearity with the concentration of spotted ER (Figure 2).
Good attachment was evident with apo ER or [3H]TOT-liganded
ER, but very little activity was recovered when [3H]E2-prebound
ER was used (Figure 1B, right). The greater attachment of apo and
antagonist-bound ER than agonist-bound ER, determined here
radiometrically, is consistent with the results using fluorophore-
tagged ERs (Figure 1B, left) and indicates that the attached ER
retains ligand-binding activity.
The specificity of this binding is evident from Figure 2, in which
EE2-Cy3 binding to both ERR and ERâ is blocked by either
9
4754
J. AM. CHEM. SOC. 2004, 126, 4754-4755
10.1021/ja039586q CCC: $27.50 © 2004 American Chemical Society

C O M M U N I C A T I O N S
binding to ERâ than to ERR. The ER subtype selectivity of these
compounds mirrors quite closely their selectivity measured in
radiometric assays, as does the affinity of various antiestrogens (not
shown).11,13
Thus, microarrays of ERR- and ERâ-LBDs printed on aldehyde
slides retain good binding activity that can be assessed using a
fluorescent ligand, and the specific and ER subtype-selective
binding of ligands can be determined conveniently in a competitive
binding assay. Elsewhere, we will describe the use of NR-LBD
microarrays in coactivator recruitment assays.
Acknowledgment. We are grateful to Dr. Mark Band for
assistance with array imaging and to Drs. Paul Hergenrother and
Gavin MacBeath for advice. This work was supported through a
grant and a traineeship from the National Institutes of Health (PHS
5R37 DK015556 to J.A.K., and T32 GM007283 to J.R.D.).
Supporting Information Available: Details of conditions for
printing ERR- and ERâ-LBD protein arrays on aldehyde slides and
for ligand-binding assays on these arrays. Synthesis and characterization
of 17R-ethynylestradiol-Cy3 (EE2-Cy3) and table of ligand structures
and receptor binding affinities (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
Figure 2. Microarrays showing EE2-Cy3 binding to ERR and ERâ. The
ERs were printed in quadruplicate at six concentrations, denoted by the
fmol values that correspond to the amount of ER-LBD applied to each spot.
Panel A, left: Microarray showing the concentration-dependent binding of
EE2-Cy3 to ERR and ERâ. The fluorescence from the highest concentration
of ERR saturates the detector. Panel A, middle: Microarray showing the
blocking of 700 nM EE2-Cy3 binding to both ERR and ERâ by 7 µM ICI.
Panel A, right: The selective blocking of only ERR by the ERR-selective
ligand PPT. Panel B: Quantification of the binding of 700 nM EE2-Cy3 to
3.6 fmol of ERs with no competitor, or when blocked by 7 µM of E2, ICI,
or subtype-selective ligands, DPN, PPT, and MPP, or by the ERâ-selective
phytoestrogen genistein.
References
(1) Hall, J. M.; Couse, J. F.; Korach, K. S. J. Biol. Chem. 2001, 276, 36869-
36872.
(2) Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard, R. E.; Bonn, T.;
Engstro¨m, O.; O¨ hman, L.; Greene, G. L.; Gustafsson, J.-A.; Carlquist,
M. Nature 1997, 389, 753-758.
(3) Zhou, G. C.; Cummings, R.; Li, Y.; Mitra, S.; Wilkinson, H. A.; Elbrecht,
A.; Hermes, J. D.; Schaeffer, J. M.; Smith, R. G.; Moller, D. E. Mol.
Endocrinol. 1998, 12, 1594-1604.
(4) Bolger, R.; Wiese, T. E.; Ervin, K.; Nestich, S.; Checovich, W. EnViron.
Health Perspect. 1998, 106, 551-557.
(5) MacBeath, G.; Schreiber, S. L. Science 2000, 289, 1760-1763.
(6) Kodadek, T. Chem. Biol. 2001, 8, 105-115.
(7) Kumble, K. D. Anal. Bioanal. Chem. 2003, 377, 812-819.
(8) Tamrazi, A.; Katzenellenbogen, J. A. Methods Enzymol. 2003, 364, 37-
53.
(9) We printed the ER LBDs onto the aldehyde slides (TeleChem International
Inc, Sunnyvale, CA) using a SpotBot Protein Edition Microarray Robot
(TeleChem International Inc.), under conditions of controlled humidity
(65-70%) and temperature (15 °C). Each slide was then blocked by
placing it carefully, face down, in a solution of 3% bovine serum albumin
(BSA) in 50 mM MOPS (3-[N-morpholino]propanoic sulfonic acid) buffer,
pH 8.0, 0.01% sodium azide, and then incubated in 50 mL of fresh BSA-
MOPS buffer at room temperature, with gentle agitation for 1 h. The slide
was dried by a brief centrifugation (1 min) at 1500 rpms (400g). Printed
slides can be stored for at least 3 months in a solution of 3% BSA at 4 °C
before probing. In ligand-binding experiments, separate subarrays on the
slide were isolated by streaking with a hydrophobic pen (Pap Pen) and
were then incubated with the fluorescent ligand, EE2-Cy3 alone (700 nM
in 3% BSA solution at room temperature) or together with different
concentrations of nonlabeled ligands. After incubation (1 h in humidity
chamber, at room temperature), arrays were washed as noted above, dried,
and scanned using a GenePix 4000 dual-laser at 535 and 635 nm with
33% power and PMT gain to give a signal slightly below saturation (Axon
Instruments, Union City, CA), and the arrays were quantified using the
GenePix Pro 4.0 software (Axon Instruments). Typically, the fluorescence
intensity was considered to be the mean value of a constant diameter circle.
Gain adjustments can usually be made so that the highest signal
corresponds to ∼45 000 fluorescence intensity units (intensity per pixel);
background intensity is typically 500 units, giving a 90:1 signal to
background.
(10) Tamrazi, A.; Carlson, K. E.; Katzenellenbogen, J. A. Mol. Endocrinol.
2003, 17, 2593-2602.
(11) 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, 4934-4947.
(12) Sun, J.; Huang, Y. R.; Harrington, W. R.; Sheng, S.; Katzenellenbogen,
J. A.; Katzenellenbogen, B. S. Endocrinology 2002, 143, 941-947.
(13) Meyers, M. J.; Sun, J.; Carlson, K. E.; Marriner, G. A.; Katzenellenbogen,
B. S.; Katzenellenbogen, J. A. J. Med. Chem. 2001, 44, 4230-4251.
(14) Katzenellenbogen, J. A.; Johnson, H. J., Jr.; Myers, H. N. Biochemistry
1973, 12, 4085-4092.
Figure 3. Plot of the fluorescence intensity of EE2-Cy3 bound to ERR
and ERâ on a microarray, showing competition by four ligands. E2 is the
strongest competitor but is not subtype selective; competition by PPT is
strongly ERR selective, and DPN and genistein are ERâ selective.
unlabeled ICI or E2, and EE2-Cy3 binding to ERR is blocked
selectively by the ERR-selective agonist propylpyrazole triol (PPT)11
and methyl piperidinopyrazole (MPP),12 whereas the binding to ERâ
is selectively blocked by the ERâ-selective agonist diarylpropioni-
trile (DPN)13 and genistein.
By adding varying amounts of competitors, we can use these
fluorometric competition experiments to quantify ligand-binding
affinity, making them operationally the equivalent of a relative
binding assay (RBA).14 The percent of the remaining fluorescence
intensity can be plotted against the competitor concentration to
generate a displacement curve (Figure 3).
The nonselective ligand, E2, competes with the binding of EE2-
Cy3 to both ERR and ERâ with a similar affinity, whereas the
ERR-selective ligand, PPT, competes ca. 100-fold more effectively
with EE2-Cy3 for binding to ERR than to ERâ. Conversely, both
ERâ-selective ligands, DPN and genistein, compete better for
JA039586Q
9
J. AM. CHEM. SOC. VOL. 126, NO. 15, 2004 4755