| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Oak Ridge Poster Sessions |
1
IA, Inc., Box 1306, Ann Arbor, MI, 48106, and
2
University of
Louisville Medical School, Hormone Receptor Laboratory, Louisville, KY 40223-2211
a Author for correspondence: fax 734-995-6869,
Ligand-binding assays are ubiquitous in biochemical research and
clinical determinations. In most common assay methods, binding proceeds
until an equilibrium condition has been obtained, which leads to
relatively long incubation times. The requirement for separating bound
from free reagents in the reaction mixture before signal detection
precludes direct observation of binding events. To reduce the duration
of an assay and to facilitate kinetic analysis, methods based on
evanescent field technology have recently been used in assay
development. These include fiber optic fluorometric sensors
(1)(2)(3) and surface plasmon resonance (4).
Specifically, evanescent field technology is based on the observation
that when light travels through a waveguide at angles approaching the
critical angle for total internal reflection, an evanescent field is
produced on the surface of the waveguide. This field falls off
exponentially with distance from the surface and is exquisitely
sensitive to the refractive indices of the waveguide surface and the
medium in which the surface resides. In surface plasmon resonance, this
sensitivity to refractive index is used to measure the amount of
substance that binds to the waveguide surface. In evanescent field
fluorometry, the field stimulates fluorophores, which become attached
to the surface through specific binding interactions (2). We
describe here the use of evanescent fiber optic fluorometric sensors to
characterize ligand binding of IgG and monovalent Fab, both specific
for estrone-1-glucuronide (E1g), and of the human estrogen receptor 
(hER-) to fibers bearing E1g or the specific estrogen response
element (ERE) and demonstrate the determination of apparent association
and dissociation binding constants.
Fused silica optical fibers obtained from Polymicro Technologies were
cleaved in 11.5-cm lengths, and the cladding was removed from a 7.0-cm
portion of the fiber using FluorinertTM (3 M). To assess the binding
kinetics of IgG, Fab, and hER-, fibers were sensitized by a
modification of the method of Bhatia et al. (5). Reagents
were obtained from Sigma-Aldrich unless otherwise noted. Fibers were
placed in a 20 mL/L solution of 3-(mercaptopropyl)-trimethoxysilane in
dry toluene for 2 h at room temperature, and then rinsed in
toluene, creating a glass surface bearing thiol groups. Thiols were
reacted with the heterobifunctional agent, 
-maleimidobutyric
acid-N-hydroxysuccinimide ester, 2 mmol/L in reagent alcohol
for 1 h. To make E1g fibers, the succinimide ester was reacted for
1 h with 0.05 g/L casein in 20 mmol/L carbonate, pH 9.3. To
prepare ERE fibers, 4 mg/L 5'-amino-ERE (nucleotide sequence, 5'
GTCCAAAGTCAGGTCACAGTGACCTGATCAAAGTT 3'; Research Genetics) was
substituted for the protein. E1g was activated using
1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide, and then incubated with
casein-derivatized fibers for 1 h. After extensive washing, fibers
were air-dried and stored under dessicant. For binding assays, fibers
were mounted inside capillary tubes having a 1.1-mm inner diameter and
80 µL final volume, and ferrules were applied to the ends to permit
sample injection and continuous fluid flow through the reaction cell.
Monoclonal antibody specific for E1g was obtained from Dr. Fortune
Kohen (Weizmann Institute, Rehovat, Israel). Fab was prepared by papain
digestion followed by purification using protein G to remove unreacted
IgG and IgG fragments. hER- was prepared as described by Wittliff et
al. (6). Proteins were labeled with Cy5
(1,3,3,3',3',1',1'-heptamethyl-indodicarbocyanine sulfonic acid;
Amersham) according to manufacturer's instructions. IgG was used at a
final concentration of 2 x 10-8 mol/L, and
Fab was used at a final concentration of 4 x
10-8 mol/L. The concentration and affinity of
hER- were estimated by radioligand-binding assay (7), and
Cy5-labeled hER- was used in fiber optic assays at 5 x
10-10 mol/L. Binding of fluorophore-labeled
protein to coated fibers was determined using an evanescent fiber optic
fluorometer (Threefold Sensors) (8). To maximize the
strength of the evanescent field, the fluorometer directed light from a
637 ± 2 nm laser diode into the fiber optic sensor cartridge at
an angle just less than the critical angle. A holographic band-stop
filter centered at 637 nm allowed the Stokes-shifted fluorescence
emitted from Cy5 at wavelengths longer than 650 nm to pass with high
efficiency. Light was detected by a photodiode, and the resulting
currents were measured by a lock-in amplifier (model SR810; Stanford
Research Systems) and collected on a computer using a program written
in Labview (National Instruments).
To determine rate constants for association and dissociation, kinetic data were fitted to a mathematical model derived under the following assumptions: (a) first order reaction kinetics applied; (b) there were no interactions between binding sites; (c) the association rate constant was much larger than the dissociation rate constant kon >> koff; (d) the number of binding sites on the fiber was limiting; and (e) the amount of binder consumed in the reaction was small enough to ignore diffusion.
Under assumption (a), the differential equation describing
the binding is:
 | (1) |
where P represents the number of bound labeled species and the over-dot represents its time-rate of change; A represents the concentration of labeled binder in solution; B represents the effective concentration of unbound sites on the fiber; kon is the apparent association constant; and koff is the apparent dissociation constant.
Applying the other assumptions and solving Eq. 1
gives:
 | (2) |
Off rates are fitted to a first order kinetic model that assumes
negligible re-binding:
 | (3) |
Both models were equipped with detrending terms for fitting to
data. Data were fitted using the custom curve fitting functionality
within DeltaGraph (DeltaPoint). As an alternative to this
noncompetitive means to determine Kd,
a competitive binding protocol was also used to determine some binding
constants for hER- to various estrogenic ligands.
Sensor responses were measured by incubating a solution containing the
binder in the sensor cartridge for either 12.5 min (Fig. 1
, A and B) or 30 (Fig. 1C
) min and collecting the fluorescent
response every 5 s throughout the incubation. For dissociation
measurements, incubation was followed by flowing a buffer solution over
the sensor for 17.5 min (Fig. 1
, A and B) or 30 (Fig. 1C
) min at 0.5
mL/min and measuring sensor response every 5 s. Nonspecific
binding was minimized by including 200 mL/L fetal bovine serum in the
buffer. Specificity of hER- binding was established by observing the
absence of response from a similar preparation of cell lysate from
yeast not transfected with the her- gene (not shown).
|
Sensorgrams refer to the real-time output of evanescent sensors
(9)(10). Sensorgrams for the binding of labeled
IgG, Fab, and hER- demonstrated monotonic association kinetics;
dissociation kinetics were generally monotonic, although the slow
dissociation rates made measurements susceptible to noise (Fig. 1
).
Curve fitting with the above models typically gave correlation
coefficients r2 
0.99, with residual
errors also shown in Fig. 1
. Using this method, the apparent
Kd for hER- to E1g fibers was
2.5 x 10-9 mol/L. For the antibody to E1g,
IgG gave an apparent Kd of 1.6 x
10-8 mol/L, whereas Fab gave 1.1 x
10-8 mol/L.
The apparent dissociation constants determined for hER- and estrogen
mimics obtained from sensor data via the competitive protocol are
consistent with radioligand binding data (11)(12)(13)(14)(15). The
Kd values obtained using competition were as
follows: estradiol-17ß, 2 x 10-10 mol/L;
estrone, 2 x 10-8 mol/L; estriol, 2
x 10-8 mol/L; diethylstilbestrol, 2 x
10-9 mol/L; zearalenone, 2 x
10-8 mol/L; and tamoxifen, 2 x
10-9 mol/L. All first digits of the results
obtained for the biosensor are expressed as 2 in these preliminary
experiments because the solutions used to determine
K consisted of concentrations of candidate
estrogen mimics that were 2 times various powers of 10. In parallel
experiments, fiber optic sensors were used to observe binding of
hER- to the ERE. Binding data were consistent with the observed
agonist or antagonist activity of the aforementioned compounds. Graphs
of 
vs concentration of ligand generally displayed an increase
in as
concentrations increased. This type of response would be expected if
hER possesses multiple ligand-binding sites displaying cooperative
behavior as has been reported (14)(16). At low
ligand concentrations, the likelihood of receptor binding only one
ligand in solution dominates. When such binding occurs, a
conformational change producing positive cooperativity would increase
the on-rate of the receptor binding to the fiber. As the concentration
of ligand in solution increases, it becomes more likely that all
receptor binding sites are occupied; therefore, the rate of binding to
the fiber would decrease, reflecting a decreased concentration of
available receptor. Thus, the data obtained with the fiber optic sensor
appear to be consistent with expected results.
Acknowledgments
This work was supported by NIH grants NIEHS 2R44ES007471-02 to J.L.W. and J.L.E. and 5R44AG12322 to J.L.E. J.L.W. is a consultant to IA, Inc. J.G.D. holds equity interest in IA, Inc. We thank Bradford Henderson and Ingrid Picazo for technical assistance.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
