The Journal of Immunology, 1998, 160: 3225-3235.
Copyright © 1998 by The American Association of Immunologists
Kinetics of Multivalent Antigen DNP-BSA Binding to IgE-Fc
RI in Relationship to the Stimulated Tyrosine Phosphorylation of FcRI1
Keli Xu*,
Byron Goldstein
,
David Holowka2,* and
Barbara Baird2,*
*
Department of Chemistry, Baker Laboratory, Cornell University, Ithaca, NY 14853 and
Theoretical Biology and Biophysics Group, Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM 87545.
 |
Abstract
|
|---|
Multivalent DNP-BSA is commonly used to cross-link anti-DNP IgE
bound to Fc
RI to stimulate cellular responses, although key features
of the binding process are unknown. Fluorescence quenching can be used
to study the kinetics of DNP-BSA binding to FITC-IgE. We observe that
DNP-BSA binds more slowly to IgE than does an equimolar amount of a
monovalent DNP ligand, suggesting that the average effective number of
DNP groups per BSA is less than one. The binding data are well
described by a transient hapten exposure model in which most of the DNP
groups are unavailable for binding but have some probability of
becoming exposed and available for binding during the time of the
binding measurement. Additional experiments indicate that, for
suboptimal to optimal concentrations of DNP-BSA, most of the FITC
fluorescence quenching on the cell surface is due to cross-linking
events. With these concentrations at 15°C, the kinetics of FITC
fluorescence quenching by DNP-BSA correlates with the kinetics of
DNP-BSA-stimulated tyrosine phosphorylation of FcRI. At 35°C, the
phosphorylation kinetics are biphasic during the time period in which
cross-linking continues to increase. Our results establish a
quantitative relationship between the timecourse for cross-linking by
multivalent Ag and FcRI-mediated signaling, and they provide the
means to predict the kinetics of cross-linking under a wide variety of
conditions.
|
Introduction
|
|---|
Antigen-mediated aggregation of
immunoreceptors
initiates signal transduction leading to cellular activation, and this
has been best characterized for FcRI, the high affinity receptor for
IgE (1). RBL-2H3 mast cells, which express 
3 x
105 FcRI/cell, often serve as a model system for
examining this process in detail. Monomeric IgE binding to FcRI does
not stimulate a cellular response but does create a bivalent cell
surface receptor for Ag with the specificity of the IgE. Binding of
monovalent ligand to IgE-FcRI causes no response. However,
aggregation of IgE-FcRI by multivalent Ag generates a complex
cascade of cellular responses, including tyrosine phosphorylation of
multiple proteins, inositol phosphate production, Ca2+
mobilization, and protein kinase C activation, culminating in
degranulation and secretion of mediators of allergy and inflammation
(2). Steady progress has been made in elucidating the binding
properties of bivalent ligands, particularly from studies with the
symmetric bivalent ligand,
N,N'-bis(-N-(2,4-DNP)aminocaproyl-L-tyrosine)-
cystine
((DCT)2-cys)3
(3). However, synthetic bivalent ligands, such as
(DCT)2-cys, typically stimulate only weak cellular
responses (4, 5). Multivalent ligands, such as BSA conjugated with
multiple DNP groups (DNP-BSA), are used extensively in experimental
studies to stimulate strong cellular responses (1, 3), but the features
of Ag binding that are critical for signaling have not been fully
defined.
A major reason that DNP-BSA binding and cross-linking have not been
examined in detail previously is the complexity encountered in
describing these processes. Whereas bivalent ligands are limited to
forming linear or cyclic complexes with bivalent IgE, multivalent
ligands can also form a variety of complicated branched structures (6, 7). We have initiated a detailed study of multivalent ligand binding to
surface IgE by using a simple mathematical model to analyze the
observed kinetics of DNP-BSA binding to IgE-FcRI. We find that these
data can be explained by a transient hapten exposure model, in which
individual DNP groups are mostly inaccessible for binding to IgE but
transiently become exposed, allowing binding and cross-linking to
proceed.
Protein tyrosine phosphorylation is known to be a primary signaling
event for FcRI and other members of the multichain immune
recognition receptor family (8, 9). As established in RBL-2H3 cells,
tyrosine phosphorylation of FcRI ß and 
subunits is the
earliest detectable cytoplasmic signaling event resulting from
Ag-mediated aggregation of IgE-FcRI (10). Therefore, we compared the
kinetics of DNP-BSA binding to, and cross-linking of, IgE-FcRI with
those of DNP-BSA-stimulated tyrosine phosphorylation of FcRI to
relate these two processes directly.
|
Materials and Methods
|
|---|
Reagents
FITC was obtained from Molecular Probes (Eugene, OR). The
protein-reactive carbocyanine dye Cy3 was purchased from Biological
Detection Systems (Pittsburgh, PA). Monoclonal mouse IgE specific for
DNP (11) was purified (12) and modified with FITC (13) as previously
described. The monovalent DNP ligand {-[(2,4
DNP)amino]caproyl}-L-tyrosine (DCT) was purchased
from Biosearch (San Rafael, CA). BSA, conjugated with an average of 15
DNP groups (DNP-BSA), and bovine -globulin (BGG), conjugated with an
average of 25 DNP groups (DNP-BGG), were prepared as previously
described (14). DNP-BSA was chromatographed on a Sepharose 12 HPLC gel
filtration column (Phamacia Fine Chemicals, Piscataway, NJ) to remove
trace amounts of aggregates. Conjugation of DNP-BSA with Cy3 (yielding
2.4 DNP groups per Cy3 bound) was conducted as recommended by the Cy3
manufacturer. Concentrations of BSA and BGG were determined by
absorption at 280 nm using extinction coefficients of 0.69
ml/(mg · cm) and 1.4 ml/(mg · cm), respectively; absorption by
DNP and Cy3 at this wavelength was corrected. ImmunoPure Immobilized
Protein A was purchased from Pierce Chemical Co. (Rockford, IL).
Polyclonal rabbit anti-IgE was purified by affinity chromatography
(15). Horseradish peroxidase-conjugated mouse mAb 4G10 and recombinant
Ab RC20H, both specific for phosphotyrosine, were purchased from
Upstate Biotechnology (Lake Placid, NY) and Transduction Laboratories
(Lexington, KY), respectively.
Cell preparation
RBL-2H3 cells (16) were grown in stationary culture and
harvested as described (17), then resuspended in buffered saline
solution (BSS: 135 mM NaCl, 5 mM KCl, 20 mM HEPES, 1.8 mM
CaCl2, 1 mM MgCl2, 1.8 mM glucose, 0.1%
gelatin, pH 7.4). For spectrofluorometric measurements, suspended cells
(1 x 107 cells/ml) were mixed with a three- to
fivefold molar excess of FITC-anti-DNP-IgE over FcRI and rotated
at 37°C for 75 min to saturate FcRI, then washed three times with
BSS. For tyrosine phosphorylation experiments, cells were sensitized
overnight or for 75 min at 37°C with three- to fivefold excess of
anti-DNP-IgE over FcRI. For flow cytometric measurements, cells
were sensitized with three- to fivefold excess of unlabeled
anti-DNP-IgE over FcRI. Cells were resuspended in BSS at 1 to
2 x 106 cells/ml for all experiments.
Spectrofluorometric measurements
Steady state fluorescence measurements were made with an SLM
8000 fluorescence spectrophotometer (SLM, Urbana, IL) operated in ratio
mode. For each experiment 1.3 to 2.0 ml RBL-2H3 cells saturated with
FITC-IgE were stirred continuously in a 10x10x40-mm acrylic cuvette
and thermostatically controlled at 15°C or 35°C as indicated. FITC
was excited at 495 nm, and emission was monitored at 520 nm. Indicated
concentrations of DNP-BSA or DNP-BGG solution were added, and the FITC
quenching was monitored. Dissociation was induced by adding a large
excess of DCT and monitored by the partial recovery of FITC that occurs
when DCT replaces DNP-BSA in the FITC-IgE combining sites, causing less
fluorescence quenching per site (18). The data were collected at 1-s
intervals with an AST386 computer (AST Products, Billerica, MA). For
experiments at 35°C, 2 µM cytochalasin D, which is known to inhibit
aggregation-dependent internalization (19), was added to the cells
before the addition of DNP-BSA. This prevents additional fluorescence
quenching resulting from the acidic environment encountered by
internalized FITC. Independent experiments showed that cytochalasin D
has no significant effects on the kinetics of DNP-BSA binding and
cross-linking (Ref. 20 and data not shown).
Flow cytometric measurements
Cell-associated Cy3 fluorescence was measured by a Coulter Epics
Profile flow cytometer (Coulter Electronics, Hialeah, FL). Excitation
was provided by an air-cooled argon ion laser with spectral lines at
458 nm, 488 nm, and 514 nm (intensities at 458 nm and 514 nm are about
20% of that at 488 nm). Emitted light was passed through a 457- to
515-nm laser blocking glass interference filter, followed by a 550-nm
dichroic short pass glass filter. The reflected light then passed
through a 590-nm long pass glass filter and was detected as Cy3
fluorescence. For each experiment, 1 ml of RBL-2H3 cells saturated with
unlabeled IgE was maintained at 15°C, and the cells were gently
agitated occasionally to maintain suspension.
Tyrosine phosphorylation measurements
IgE-saturated cells were stirred in a 25-ml Celstir spinner
flask (Wheaton, Millville, NJ) and maintained at 15°C or 35°C while
indicated concentrations of DNP-BSA were added to stimulate the cells.
Cells (750 µl) were removed from the flask at indicated times and
mixed with an equal volume of ice-cold 2x lysis buffer (20 mM Tris, 2
mM Na3VO4, 30 mM
Na4P2O7, 10 mM glycerolphosphate,
0.04 U/ml aprotinin, 0.02% NaN3, 6 mM EDTA, 2 mM
4-(2-aminoethyl)-benzenesulfonylfluoride (Calbiochem, La Jolla, CA), pH
8.0) containing 0.13% TX-100. This was followed by addition of 10 µM
DCT to dissociate bound DNP-BSA, incubation on ice for 10 min, then
centrifugation at 14,000 x g for 5 min at 4°C in a
Hermle Z 230 MR microfuge (B. Hermle, AG, Gosheim, Germany) to sediment
the nuclei. The supernatants were incubated with 1 µg/ml polyclonal
rabbit anti-IgE on ice for 1 h, followed by rotating with 35
µl protein A beads at 4°C for 1 h to immunoprecipitate the
IgE-FcRI. The beads were washed once in lysis buffer containing
0.06% TX-100 followed by another wash with lysis buffer without
detergent, then boiled in 33 µl sample buffer (10% glycerol, 1%
SDS, 0.05 M Tris, pH 6.8) for 3 min, followed by centrifugation,
removal of the supernatant, and reextraction of the beads with the same
volume of sample buffer. Proteins in the combined sample buffer
supernatants were separated by SDS-PAGE under nonreducing conditions,
then transferred to immobilon-polyvinylidene difluoride (PVDF)
(Millipore, Bedford, MA) using a semidry transfer apparatus (Integrated
Separation Systems, Hyde Park, MA). Membranes were blocked with 5% BSA
then probed with horseradish peroxidase-conjugated
anti-phosphotyrosine Abs (4G10 or RC20H). Blots were developed
using ECL chemiluminescence and Hyperfilm-ECL (Amersham, Arlington
Heights, IL). In most of these experiments, the only bands detected
with apparent molecular mass less than 200 kDa were
tyrosine-phosphorylated FcRI ß and , as previously identified
(21). Densitometry was performed with a Quick Scan Flur-Vis
Densitometer (Helena Laboratories, Beaumont, TX). Peak intensities were
integrated after subtracting the background.
Time-dependent tyrosine phosphorylation of the ß subunit of FcRI
was plotted together with the time-dependent quenching of FITC-IgE
fluorescence obtained under the same conditions of DNP-BSA binding and
stimulation. The relative tyrosine phosphorylation obtained at 15°C
was scaled to maximize overlap with the fluorescence-quenching curves
at the longer time points measured. Tyrosine phosphorylation data
obtained under the same stimulating conditions for three separate
experiments were included in this analysis. Unlike the phosphorylation
data, which are subject to scatter, the fluorescence-quenching curves
are highly reproducible. Therefore, a single fluorescence-quenching
curve is compared with the combined phosphorylation data.
Data analysis
Two models were used to analyze DNP-BSA binding data, a
monovalent ligand model and a transient hapten exposure model. For both
models, we make the following simplifying assumptions: 1) all the Fab
binding sites have the same intrinsic binding properties; 2) all the
DNP groups conjugated to BSA fall into two categories: they are either
not available for binding (buried DNP) or available for binding and
have the same intrinsic binding properties (exposed DNP); 3) FITC
fluorescence quenching is proportional to the Fab binding site
occupancy, i.e., the fluorescence quenching per Fab site bound remains
the same regardless of whether the Fab site is occupied by monovalently
bound DNP-BSA or by DNP-BSA that bridges two or more Fab sites.
Monovalent ligand model (see Fig. 2
A)
This model further assumes that each BSA has a single exposed
DNP group. The rate of binding is described by the following
differential equation:
 | (1) |
where [C] and [R] are the molar
concentrations of nonbound DNP-BSA and IgE Fab sites, respectively, and
[Y] is the molar concentration of bound DNP-Fab complexes.
k+1 and k-1 are the rate
constants for the reversible binding of a DNP-group to a Fab site. The
conservation equations for total Fab sites
([R]total) and total DNP-BSA
([C]total) are
 | (2) |
 | (3) |
Substituting for [C] and [R] into Eq. 1
we obtain a single equation for [Y], which we solve
numerically. We note that an analytic solution could also have
been used.
We define
fb=[Y]/[R]total
as the fraction of Fab sites in the bound state at a time t
after DNP-BSA is added to the FITC-IgE-saturated cells.
fb is related to the experimentally measured
fluorescence by the expression
 | (4) |
where F is the relative fluorescence at time
t, Fmax is the relative fluorescence
of FITC-IgE-saturated cells before addition of DNP-BSA, and
Fmin is the relative fluorescence when all the
Fab sites are saturated with DNP-BSA.
Transient hapten exposure model (see Fig. 3A)
This model postulates that most of the DNP groups on BSA are
buried at any time. However, in any time interval, the DNP groups have
some probability of becoming exposed and available for binding to IgE.
The binding stabilizes the exposed state. The simplest form of this
model is considered here, and we make the following two additional
assumptions: 1) the exposure and burying of DNP on BSA can be described
by fixed rate constants, 
+ and -,
respectively; and 2) the maximal DNP groups per BSA that can become
exposed (n) is 2. (It is straightforward to
generalize assumption 2 and take n to be any value less than
or equal to the total number of DNP per BSA.) We assign the rate
constants k+1 and k-1
for a free ligand (with one or two exposed DNP groups) binding
reversibly to a free Fab site, and the rate constants
k+2,2D and k-2 for a
ligand (with one exposed DNP bound and the other exposed DNP free)
binding reversibly to a free Fab site, where
k+2,2D is a two dimensional rate constant. In
the fitting program, we use the spatially dimensionless parameter
k+2[R]total, where
[R]total is three dimensional molar
concentration of total Fab sites, and k+2 is the
apparent three dimensional cross-linking rate constant corresponding to
the same amount of Fab sites as that on the cell surface but spread
uniformly over available three dimensional space. Because
k+2 and [R]total have
units of M-1 · s-1 and M, respectively,
k+2[R]total has units
of s-1. By dividing
k+2[R]total by
rt, the total Fab site density on the cell surface (in units
of sites/cm2), we get k+2,2D.
Five differential equations describe DNP-BSA binding and cross-linking
in this model (when n = 2), and there are two
conservation equations for total Fab sites
([Ctotal]) and total DNP-BSA
([Rtotal]) (See Fig. 3A):
 | (5) |
 | (6) |
 | (7) |
 | (8) |
 | (9) |
 | (10) |
 | (11) |
[C0], [C1],
and [C2] are the molar concentrations of free
DNP-BSA with 0, 1, and 2 exposed DNP groups, respectively.
[Ya] and [Yc] are the
molar concentrations of bound DNP-BSA to IgE on the cell surface, with
one and two DNP groups bound, respectively; these species contain no
exposed DNP groups that are not bound. [Yb] is
the molar concentration of bound DNP-BSA with one DNP group bound to
IgE and one exposed DNP group that is not bound. In this model
fb=([Ya]+[Yb]+2[Yc])/[R]total,
and it is related to the experimentally measured fluorescence by Eq. 4.
Based on the conservation equations and the equilibrium constant for
the hapten exposure step, = +/-, we
derive the following equations, which determine the initial
distribution of the Ag species, with 0, 1, or 2 exposed haptens.
 | (12A) |
 | (12B) |
 | (12C) |
Correspondingly, the average number of exposed hapten per Ag
before the encounter of the FITC-IgE
(Dav) is given by
 | (13) |
In general, for any n, Dav =
n/(1 + ).
If, at t=tz, a large excess of DCT is
added to dissociate DNP-BSA, free Fab sites fill up quickly with DCT,
and [R] 
0. Then Eqs. 7–9 become
 | (14) |
 | (15) |
 | (16) |
If we set
[Yab]=[Ya]+[Yb],
and add Eq. 14 to Eq. 15, we get
 | (17) |
Eqs. 16, and 17 can be integrated to yield
 | (18) |
 | (19) |
where [Yab]t=tz and
[Yc]t=tz are the
concentrations of DNP-BSA singly and doubly bound to cell surface IgE,
respectively, immediately after DCT is added. Note that for DCT-induced
dissociation (Eqs. 18, and 19) the exponential decay constants depend
only on k-1 and k-2.
Thus, in the dissociation phase the dynamics of DNP exposure on the Ag
surface plays no role.
We define f as the fraction of Fab sites originally bound to
DNP-BSA at tz that are still bound at subsequent
time t,
 | (20) |
f is related to the experimentally measured
fluorescence by
 | (21) |
where F'max is the value of the relative
fluorescence after dissociation has gone to completion and
F'min is the value of the relative fluorescence
immediately after addition of DCT. Substituting Eqs. 18, and 19 into Eq. 20, we obtain
 | (22A) |
 | (22B) |
 | (22C) |
i.e., the fluorescence recovery can be fit by a double
exponential equation. If k-1 =
k-2, then Ai = 1, and
F in Eq. 22a reduces to a single exponential equation.
The parameter estimates of Fmax,
Fmin, Fmax',
Fmin', k+1,
k+2,2D, k-1,
k-2, +, and -
were obtained with a subroutine, DNLSI, from the Common Los Alamos
Software Library, which is based on a finite difference,
Levenberg-Marquardt algorithm for solving nonlinear least-squares
problems. For fitting both models to the association data, the
differential equations Eq. 1 and Eqs. 5–9 were numerically solved with
a standard algorithm. The dissociation data were fit with the
exponential equation Eq. 22a. The zero time point for binding was
determined by the addition of DNP-BSA or DNP-BGG;
tz for the induced dissociation of DNP-BSA or
DNP-BGG was determined by the addition of DCT;
[R]total was calculated by multiplying
measured cell density (cells/ml) by 2x the estimated number of FcRI
on the cell surface (3 x 105/cell) and converting to
molar concentration; [C]total was determined
by absorption spectroscopy measurements of the stock ligand solutions
as described above.
|
Results
|
|---|
We examined DNP-BSA binding to FITC-labeled anti-DNP IgE in
solution or bound to FcRI on RBL cells with our previously
established fluorescence quenching method (13). Figure 1 shows a kinetic quenching curve for
cells saturated with FITC-IgE after addition of DNP-BSA at 15°C.
Binding was evaluated at this temperature to avoid complications due to
cross-linking-dependent internalization of IgE-receptor complexes (22).
Although maximal quenching varies somewhat with the levels of FITC and
DNP conjugations, these quenching curves are very reproducible for the
same preparations of DNP-BSA and FITC-IgE. The saturation level of
fluorescence quenching remains the same (35% of the total
fluorescence) for DNP-BSA additions ranging from concentrations where
cross-linking dominates to conditions where monovalent binding
dominates (22), and this supports our assumption that the FITC
fluorescence quenching is proportional to the fraction of Fab binding
sites occupied by DNP groups. We monitored the dissociation of DNP-BSA
by adding an excess of monovalent DCT, which quenches FITC-IgE
fluorescence less than DNP-BSA; fluorescence recovery occurs as DNP
from DCT replaces DNP from DNP-BSA in the Fab site.
View larger version (26K):
[in this window]
[in a new window]
|
FIGURE 1. Binding of monovalent DCT (inset) and multivalent
DNP-BSA to cell-bound FITC-anti-DNP-IgE, and dissociation of
DNP-BSA in the presence of large excess of DCT at 15°C. DNP-BSA (11
nM; DNP:BSA = 15) or DCT (5.6 nM) were added to FITC-IgE-saturated
RBL cells (1 x 106 cells/ml; 0.5 nM FITC-IgE) at the
arrows. At t = 2093 s, 77 µM DCT was added to
the DNP-BSA-containing sample (arrowhead); the quenching occurring
immediately after this addition is due to filling of unoccupied IgE
combining sites, inner filter effect, and a small dilution effect. The
partial FITC fluorescence recovery indicates the dissociation of
DNP-BSA, which cannot rebind in the presence of sufficiently excess
DCT. The dashed line represents the level of relative fluorescence
observed for a separate but identical sample of cells to which 77 mM
DCT was added to saturate the FITC-IgE.
|
|
Analysis of DNP-BSA binding data with monovalent ligand model
Fluorescence quenching caused by binding and cross-linking of
DNP-BSA to cell-bound FITC-IgE (Fig. 1) is surprisingly slow compared
with that observed when similar amounts of DCT are added (Fig. 1, inset). For this experiment the DCT concentration was 5.6 nM
and the DNP-BSA concentration was 11 nM (corresponding to 165 nM DNP
groups). These results raise the possibility that the effective number
of DNP groups available for binding on BSA is relatively small. Because
DNP-BSA triggers cellular responses, we know that a signficant fraction
of these ligands can cross-link IgE, i.e., are effectively multivalent.
However, for comparison purposes we began our analysis of the DNP-BSA
binding data with a monovalent ligand model (Eqs. 1–4 and Fig. 2A). The data from
Figure 1 for DNP-BSA binding to cell-bound FITC-IgE were analyzed with
this model as illustrated in Figure 2B. In the analysis, the
effective DNP concentration was assumed to be the same as the DNP-BSA
concentration (11 nM). We see that the monovalent ligand model predicts
faster leveling off of fluorescence quenching than the experimental
data (Fig. 2B). As expected, the binding of
monovalent ligand DCT to cell-bound FITC-IgE (Fig. 1, inset)
is well fit with the monovalent ligand model (Fig. 2C). Furthermore, this analysis of DCT binding yields
association and dissociation rate constants
(k+1 2 x 107
M-1 · s-1, k-1
1 x 10-2 s-1) that are consistent with
those determined previously for DCT binding to cell-bound IgE
(23).
Variations of this simple model also do not fit the DNP-BSA binding
data. In particular, a bivalent model, which assumes that there are two
effective DNP groups per BSA that can cross-link IgE, predicts even
faster binding (data not shown). This shows that expanding the
monovalent ligand model to higher effective valency does not improve
the fit to the data. On the other hand, simply reducing the average
effective valency to less than one without including a cross-linking
step in the model also fails to fit the data (data not shown). We
confirmed that this DNP-BSA preparation is a potent stimulus for RBL
cellular degranulation (50% ß-hexosaminidase release at 0.2–1.1
nM DNP-BSA for adherent RBL cells), verifying productive cross-linking
of IgE-FcRI by this ligand.
Analysis of DNP-BSA binding data with a transient hapten exposure
model
We developed this model (Eqs. 5–22; Fig. 3A)
to take into account both the low effective valency of DNP-BSA and its
capacity for cross-linking on the cell surface. With it we analyzed the
DNP-BSA binding and dissociation data shown in Figure 1. Figure 3B shows that DCT-induced DNP-BSA dissociation from
cell-bound FITC-IgE is fit by a single exponential indicating
k-1 = k-2 (Eqs. 14–22). Consequently, we set k-1 =
k-2 as an adjustable parameter in fitting the
forward binding. As shown in Figure 3C, this model readily
fits the kinetics of DNP-BSA binding to the cell-bound IgE and yields
values of k+1 1 x 107
M-1 · s-1,
k+2[R]total 1
x 10-1 s-1, k-1 =
k-2 1 x 10-2
s-1, and = +/-
1 x 10-1. The molar concentration of total Fab is 1
nM, yielding the apparent three dimensional k+2
1 x 108 M-1 · s-1,
which is about 10-fold higher than k+1. This
indicates that the cross-linking step is accelerated, presumably due to
the high local concentration of IgE confined to the cell surface.
We tested whether this transient hapten exposure model is consistent
with binding of DNP-BSA to cell-bound IgE over a range of
concentrations corresponding to biologic responses, i.e., 1 to 11 nM
DNP-BSA. For all of these experiments, a similar amount of FITC-IgE
quenching (35%) is achieved when binding reaches saturation. We
find that these binding data are all well fit, and values for
parameters derived from the best fits of representative data are listed
in Table I. Also included in this table
is the lumped parameter nk+1/(1+) for
n = 2 which corresponds to the initial rate of DNP-BSA
binding to IgE. The dissociation data were consistently well fit with a
single exponential, and the values for k-1 =
k-2 derived from various Ag concentrations are
in good agreement. Furthermore, values for k-1
= k-2 derived from fits of the association data
are consistent with those derived from fitting the dissociation data.
For the data summarized in Table I, the arithmetic average of the
parameter estimates for DNP-BSA are k+1 =
8.0 x 106 M-1 · s-1,
k+2[R]total = 1.1
x 10-1 s -1, k-1 =
k-2 = 9.4 x 10-3
s-1, = +/- = 1.6
x 10-1, and 2k+1/(1+) = 2.1
M-1 s-1. Notably, the estimates of
k+1 and k-1 derived from
DNP-BSA binding are comparable to those derived from DCT binding (Table I and 23 , indicating that the slower binding of DNP-BSA compared
with DCT is not because DNP groups on BSA intrinsically bind
significantly slower to IgE than does DCT or that the binding of
DNP-BSA is diffusion limited. The model implies that the DNP groups on
BSA are not readily available for binding because of their low exposure
probability as indicated by values for and
Dav that are less than 1 (see below).
Uncertainties in the absolute values for the various parameters derived
with this model are considered in the Discussion.
DNP-BGG binds somewhat less slowly to IgE-FcRI on cells
DNP-BGG is also commonly used experimentally as an Ag, therefore,
we compared its binding properties with those of DNP-BSA. Figure 4 shows binding of 0.66 nM of DNP-BGG and
1.1 nM of DNP-BSA to two identical samples of RBL-2H3 cells saturated
with FITC-IgE (0.5 nM bulk IgE). Because the degree of modification
for DNP-BGG and DNP-BSA are 25 and 15, respectively, the bulk DNP
concentrations for the two Ags are the same under these conditions.
FITC-IgE combining sites are occupied faster by DNP-BGG than by DNP-BSA
although still substantially slower than by the monovalent ligand DCT
(Fig. 1 and data not shown). As with DNP-BSA, DCT-induced dissociation
of DNP-BGG is well-fit with a single exponential (data not shown), and
the derived dissociation rate constants are similar to those for
DNP-BSA and DCT (Table I). Binding of DNP-BGG can also be fit with the
transient hapten exposure model as shown in Figure 4. As indicated for
these and other representative data presented in Table I, the rate
constants for monovalent binding and dissociation,
k+1 and
k-1(=k-2), are similar
for DCT, DNP-BSA, and DNP-BGG. Parameters derived from the model
indicate that DNP-BSA and DNP-BGG binding kinetics are limited by
hapten exposure as reflected by the equilibrium constant .
Specifically, values for DNP-BSA and DNP-BGG are 0.16 and
0.59, respectively, and corresponding values for
Dav (average number of exposed hapten per Ag
before encounter with IgE; Eq. 13) are 0.28 and 0.74,
respectively. Thus, the observation that DNP-BGG binds and cross-links
faster than DNP-BSA is consistent with its larger value for and
therefore for Dav.
View larger version (21K):
[in this window]
[in a new window]
|
FIGURE 4. Comparison of the kinetics of DNP-BSA binding (upper trace) with
the kinetics of DNP-BGG binding (lower trace; offset) to cell-bound
FITC-anti-DNP-IgE and their fits with slow hapten exposure model
(Eqs. 5–22; solid lines). The cells were saturated with
FITC-anti-DNP-IgE (0.5 nM bulk IgE) and maintained in suspension
at 15°C. Amounts equal to 1.1 nM DNP-BSA (DNP:BSA = 15) and 0.66
nM DNP-BGG (DNP:BGG = 25) were added to two identical aliquots of
cells, as indicated by the arrows, so that the final DNP concentrations
in both samples was 16.5 nM. The parameters derived from DNP-BSA
binding are k+1 = 7.82 x 106
M-1 · s-1, k-1
= k-2 = 1.4 x 10-2
s-1,
k+2[R]total = 1.3
x 10-1 s-1, + = 1.7 x
10-4 s-1, - = 5.4 x
10-4 s-1. The parameters derived from DNP-BGG
binding are k+1 = 8.1 x 106
M-1 · s-1, k-1
= k-2 = 6.3 x 10-3
s-1,
k+2[R]total = 1.8
x 10-1 s-1, + = 1.1 x
10-3 s-1, - = 1.9 x
10-3 s-1.
|
|
Fluorescence-quenching curves reflect cross-linking of
cell-bound FITC-IgE by DNP-BSA at concentrations that are
suboptimal to optimal for cell activation
Because FITC fluorescence quenching caused by DNP-BSA binding is
proportional to the Fab site occupation, this method cannot directly
distinguish between that caused by monovalent binding and that caused
by cross-linking to additional IgE-FcRI. To dissect the contribution
of these two processes, we fluorescently labeled DNP-BSA with Cy3 and
monitored the binding of DNP-BSA to cell-bound IgE with flow cytometry.
Monovalent binding of Cy3-DNP-BSA to cell-bound IgE causes
cell-associated Cy3 fluorescence, and subsequent cross-linking steps
cause no additional fluorescence. Therefore, the rate of increase of
cell-associated Cy3 fluorescence reflects the rate of the monovalent
binding of DNP-BSA to cell-bound IgE. A control experiment conducted
with excess DCT showed that nonspecific binding of Cy3-DNP-BSA is
negligible (data not shown). Fluorescence-quenching curves confirmed
that Cy3-DNP-BSA binds to FITC-IgE on cells with kinetics identical to
DNP-BSA (Fig. 1 and data not shown).
Figure 5 compares the time courses from
these two fluorescence methods. The increase in cell-associated Cy3
fluorescence caused by binding of DNP-BSA is substantially faster
(t1/2 
15 s) than the FITC
fluorescence quenching caused by the binding and cross-linking of the
same concentrations of DNP-BSA (t1/2
150 s). These results indicate that the cross-linking of DNP-BSA
occurs significantly slower than the initial association of DNP-BSA
with the cells. Furthermore, it appears that the fraction of FITC
fluorescence quenching caused by monovalent binding is smaller than
that caused by subsequent cross-linking events. We estimate that <20%
of fluorescence quenching after 50 s is due to monovalent binding
under these conditions.4 That
the fluorescence quenching curve after 50 s reflects primarily
cross-linking is valuable for comparison to the time course of the
cellular response, as is described below. Implications of this
numerical estimate for our minimal model are considered in the
Discussion.
We further evaluated fluorescence quenching caused by monovalent
binding vs cross-linking by comparing DNP-BSA-dependent quenching for
FITC-IgE in solution with that for cell-bound FITC-IgE. Cross-linking
on the cell surface is expected to be accelerated by the high local IgE
concentrations. Therefore, if fluorescence quenching on cell surface is
mostly due to cross-linking, one would expect that fluorescence
quenching for DNP-BSA binding to IgE in solution would approach a
plateau more slowly than that for DNP-BSA binding to IgE on the cell
surface. The data shown in Figure 6 are
consistent with this prediction. For FITC-IgE in solution (upper
trace), there is a small, fast phase of fluorescence quenching just
after the addition of DNP-BSA, presumably due to the monovalent binding
of the preexisting exposed DNP groups to FITC-IgE. Thirty minutes after
its addition to IgE in solution, 5.5 nM DNP-BSA has quenched less than
half of the maximum quenching (25%) that occurs with excess (55 nM)
DNP-BSA, suggesting that equilibrium is approached at the lower
concentration with a significant fraction of IgE sites unoccupied. In
contrast, DNP-BSA binding to IgE on cells (lower trace) causes rapid
quenching in the first 100 s, and this levels off to the same
maximal quenching (35%) observed with higher concentrations of Ag
within 30 min, indicating that occupancy of the IgE combining sites is
nearly complete by this time.
Results from Figures 5 and 6 are consistent with the conclusion that
facilitated cross-linking of IgE on the cell surface serves to
stabilize the exposure of DNP groups on DNP-BSA. Furthermore, these
results indicate that the time course of DNP-BSA-dependent FITC
fluorescence quenching on the cell surface is dominated by
cross-linking after a short period of time (1 min for these Ag
concentrations), such that this quenching curve can serve as a
cross-linking curve at the later time points. We emphasize that FITC
fluorescence quenching is parallel to Ag cross-linking only at the low
concentrations of DNP-BSA that are suboptimal to optimal for cell
activation. As the concentration of Ag increases, the concentration of
immediately exposed DNP groups will increase, such that the fraction of
the fluorescence quenching due to monovalent binding will also
increase. This prediction is consistent with our flow cytometric
measurements, which showed that the maximal cell-associated Cy3
fluorescence increases as the concentration of Cy3-DNP-BSA increases
(data not shown).
DNP-BSA-mediated IgE-FcRI cross-linking compared with stimulated
tyrosine phosphorylation of FcRI
A major motivation for our binding studies is to understand the
relationship between Ag-induced receptor aggregation and consequent
signal transduction. For this purpose we examined tyrosine
phosphorylation of FcRI, which is the earliest known cytoplasmic
signaling event. We matched experimental conditions to compare directly
the kinetics of DNP-BSA-stimulated tyrosine phosphorylation with the
kinetics of DNP-BSA-induced IgE-FcRI cross-linking as indicated by
the DNP-BSA-induced fluorescein fluorescence quenching. We previously
showed that cross-linking-dependent tyrosine phosphorylation of FcRI
occurs to a similar extent, albeit more slowly, at lower temperatures
(4°C) as it does at 37°C (24). Figure 7A shows a representative
western blot of time-dependent tyrosine phosphorylation of FcRI
stimulated by 3 nM DNP-BSA at 15°C. The probing Ab (4G10) detects
phosphorylated ß subunit more strongly than phosphorylated
subunit (see Fig. 8a),
and, therefore, phosphorylation of ß was monitored for comparison to
the binding kinetics. Densitometric measurements of ß tyrosine
phosphorylation stimulated by 3 nM DNP-BSA at 15°C in three
independent experiments are plotted in Figure 7B, together
with fluorescence quenching occurring under the same conditions. These
phosphorylation and binding curves show a strong correlation. Similar
experiments were conducted with 1.1 nM DNP-BSA at 15°C (Fig. 7C). At this lower concentration, both fluorescence
quenching and tyrosine phosphorylation are slower
(t1/2 230 s for binding by 1.1
nM DNP-BSA vs t1/2 120 s for binding by
3.0 nM DNP-BSA), but the correlation between these two processes
remains similar for both concentrations of DNP-BSA. For both, the
curves overlap at all but the earliest phosphorylation time points when
monovalent binding dominates. Thus, these results at 15°C show
parallel time courses for IgE-FcRI aggregation caused by DNP-BSA and
consequent cytoplasmic signaling.
Because most tyrosine phosphorylation experiments documented in the
literature are performed at higher, more physiologic temperatures, we
also compared binding and phosphorylation at 35°C. We find that
tyrosine phosphorylation of FcRI stimulated by 1.1 nM DNP-BSA at
this higher temperature (Fig. 8A) is somewhat
stronger than that observed at 15°C, and the phosphorylation level
peaks at about 5 min and then declines. This time course is similar to
that observed by others at 35°C (25), and it is very different from
what we observed at 15°C where tyrosine phosphorylation is sustained
for at least 20 min (Fig. 7). The time courses for fluorescence
quenching and ß tyrosine phosphorylation caused by 1.1 nM DNP-BSA at
35°C are directly compared in Figure 8B. The kinetics of
the cross-linking of 1.1 nM DNP-BSA at 35°C are not very different
from those of 3 nM DNP-BSA at 15°C, while the corresponding kinetics
of ß tyrosine phosphorylation are dramatically different (Figs. 7 and 8).
|
Discussion
|
|---|
These studies represent part of our continuing effort to determine
the features of IgE-FcRI cross-linking that are critical for
initiating the signaling cascade in RBL-2H3 cells. Multivalent ligands
such as DNP-BSA are commonly used as potent stimuli. However, they have
complicated structures, and, unlike the simpler bivalent ligands,
examination of their binding to cell-associated Igs has been very
limited (26, 27). The present study analyzed the kinetics of DNP-BSA
binding with plausible mathematical models. The data indicate that the
time course of FITC quenching after the first minute of binding is
dominated by cross-linking events in the Ag concentration range
examined. Thus, we could compare the kinetics of DNP-BSA-mediated
receptor cross-linking to the kinetics of DNP-BSA-stimulated
phosphorylation of receptor tyrosines. This comparison at 15°C shows
that these two processes occurs at the same rate, indicating that
cross-linking is likely to be the rate-limiting step and providing
strong evidence for a direct relationship between FcRI aggregation
and the earliest known signaling events.
Time-dependent binding of DNP-BSA to FITC-IgE revealed that binding of
multivalent Ag DNP-BSA to cell-bound anti-DNP-IgE is surprisingly
slow compared with the binding of an equimolar concentration of
monovalent ligand DCT (Fig. 1). Neither monovalent nor bivalent ligand
models can account for the data, and models that invoke Ag
heterogeneity with respect to absolute DNP exposure are also
inadequate. For example, a model in which a small subpopulation of the
total Ag concentration has all DNP groups available for binding while
the DNP groups on most Ags are irreversibly buried could not account
for the capacity of 0.2 nM DNP-BSA (3 nM DNP) to fully occupy the
FITC-IgE combining sites (1 nM) as we have observed (unpublished
results). As shown in Figure 5, comparison of combining site occupancy
monitored by fluorescence quenching and flow cytometric measurement of
DNP-BSA association with cells reveals that a large fraction of the
DNP-BSA-induced fluorescence quenching occurs more slowly than DNP-BSA
association with cells and must therefore be due to cross-linking
events. Consistent with this, we observed that DNP-BSA-induced
fluorescence quenching is significantly slower for IgE in solution than
for IgE-FcRI on the cell surface where cross-linking is facilitated
by the local high concentration of IgE (Fig. 6). The observations that
DNP-BSA stimulates robust cellular responses (28) and also causes
immobilization of most of the IgE-FcRI (5, 29) further indicate that
most IgE-FcRI can be cross-linked into aggregates by this
Ag.
To account for both the initially low effective valency of DNP-BSA and
the efficient cross-linking of DNP-BSA to IgE-FcRI on the cell
surface, we developed a transient hapten exposure model (Fig. 3A). In this model most of the n DNP
groups are generally unavailable for binding to IgE, but each has a
finite probability of transient exposure, and binding to IgE stabilizes
the exposed state. For a first approximation, we truncated the model at
n = 2 DNP groups that can be exposed. This truncated
model effectively describes the association and dissociation of DNP-BSA
on the cell surface (Fig. 3), and the dissociation rate constants
derived from the dissociation data are consistent with those derived
from the association data (Table I). Furthermore, we find that
association and dissociation data can be fit simultaneously with this
model (our unpublished results). Also, several sets of association or
dissociation data obtained at different concentrations of DNP-BSA can
be fit simultaneously as effectively as they are fit individually, and
similar rate constants are obtained (our unpublished results). The
estimated k+1 and k-1 =
k-2 for DNP-BSA derived from this model are
comparable to those obtained for DCT (Table I), consistent with our
interpretation that the slow association of DNP-BSA to IgE is not
because the DNP group has a slower binding rate constant or lower
intrinsic affinity when conjugated to BSA, but, rather, that the
availability of DNP groups is limited.
This model quantitatively accounts for the faster binding of DNP-BGG
than DNP-BSA at equimolar DNP, and it predicts that, on average, the
availability of DNP groups on DNP-BGG is somewhat greater, but still
significantly restricted. The small differences observed may be related
to the secondary structures of these two different proteins (30, 31),
or, more likely, to more subtle structural features near the surfaces
of these proteins where the lysine-conjugated DNP groups are expected
to be located (32). The faster binding by DNP-BGG compared with DNP-BSA
indicates that diffusion of these Ags is not likely to play a major
role in their binding to the cell surface, as D20,w for BSA
and -globulin are 5.9 x 10-7 cm2/s
and 3.8 x 10-7 cm2/s, respectively (33, 34). The restricted availability of DNP groups on both BSA and BGG to
Ab binding is likely to be a general property of DNP-conjugated
proteins, although chemically unrelated haptens conjugated to proteins
will not necessarily share this property.
As described in Materials and Methods, the transient hapten
exposure model includes several simplifying assumptions. Our assumption
that IgE can be considered in terms of two equivalent Fab binding sites
is supported by our observation that the binding of DNP-BSA to these
Fab fragments is slow and quite similar to DNP-BSA binding to intact
IgE in solution (our unpublished results). This similar binding
behavior indicates it is unlikely that limited accessibility or
intramolecular cross-linking of IgE binding sites by DNP-BSA
contributes substantially to the overall binding.
With our initial approximation that a maximum of two DNP groups become
exposed (n = 2; Fig. 3A), each
cross-linking event must be preceded by a monovalent binding event.
This would predict that cross-linking can cause no more than 50% of
FITC fluorescence quenching. However, we calculate from the data of
Figure 5 that more than half of the DNP-BSA-induced quenching on cell
surface is due to cross-linking. The fact that DNP-BSA stimulates much
more robust cellular responses than does bivalent ligand
(DCT)2-cys also makes it appear unlikely that DNP-BSA only
forms IgE-FcRI dimers (3). Therefore, a more realistic description
of the binding of DNP-BSA to cell-bound FITC-IgE requires that the
model be expanded to n > 2. Because the Figure 5
analysis estimates that monovalent binding contributes about 30% to
the FITC quenching, the model with n = 2 is probably
not too far off. Expansion of the model should not change the estimated
value of the dissociation rate constant k-1
(k-2) as it is determined by the single
exponential of the dissociation data. Trends in the estimated values of
the other parameters due to expanding the model further can be
predicted from changes occurring with expanding the model from
n = 1 to n = 2. We predict that
expanding the model from n = 2 to n = 3
will decrease somewhat the estimated values of ,
k+1, and k+2. However,
consistency with the DCT results suggest that more accurate values for
n > 2 will not be very different. The lumped
parameter, nk3+1/(1+), corresponds to the
initial rate for binding of DNP-BSA to IgE, and an algorithm
fitting the initial slope of the binding data will yield an accurate
value for this parameter. Table I shows that, as the DNP-BSA
concentration decreases, the values for this parameter
(n = 2) fluctuate about a constant of 2
x 106 M-1 · s-1, while the
values of k+1 decrease and the values for
increase. Thus, it will be possible to determine an accurate value for
this lumped parameter as well as for k+1 and
for the correct choice of n.
Whereas the values for listed in Table I are likely to be close to
accurate, the absolute values of + and -
are much less well defined. The absolute values we have extracted for
+ and - are probably too small because
the characteristic times associated with these values are on the order
of 103 to 104 s whereas the fluorescence
quenching we observe occurs on the time scale of 0 to 103
s. Better estimates of the absolute values of + and
- will require more experiments and development of
better fitting procedures. For present purposes, the ratio =
+/- provides the most useful parameter,
the equilibrium constant for the exposure of DNP groups.
We observe a direct correlation between the time course for receptor
cross-linking and receptor tyrosine phosphorylation at 15°C (Fig. 7, B and C). This correlation does not appear
at 35°C where there is a biphasic tyrosine phosphorylation response
(Fig. 8). A similar biphasic time course for receptor tyrosine
phosphorylation at similar Ag concentrations and physiologic
temperatures was observed by Pribluda and Metzger (25), who also showed
that the onset of tyrosine phosphorylation after addition of Ag was the
same for ß and subunits. In their experiments, addition of excess
hapten quickly reversed the receptor phosphorylation, suggesting this
is regulated by a dynamic interplay of kinases and phosphatases (25, 35). The observation that the phosphatase inhibitor phenylarsine oxide
inhibited the time-dependent decrease in phosphorylation (25) supports
this model. Our results at 35°C that tyrosine phosphorylation
declines as cross-linking continues to increase are also consistent
with the involvement of regulatory phosphatases; at 15°C these
regulatory processes appear to be suppressed. Thus, this lower
temperature provides a simpler situation for examining the relationship
between receptor aggregation and receptor tyrosine phosphorylation.
Wofsy et al. (36) compared at 37°C the time course for binding
of covalently linked dimers of IgE with the time course for the
resulting phosphorylation of tyrosines on FcRI. They found that
phosphorylation reached a plateau much more rapidly than dimer binding
and cross-linking. They proposed that the leveling off of
phosphorylation while receptor aggregation continued to increase
occurred because receptor aggregates competed for limited amounts of
the initiating kinase, presumably Lyn. They further suggested that, as
receptors were aggregated, new high affinity sites (phosphotyrosines on
receptors) that bind the initiating kinase were created and these sites
also entered the competition for Lyn. Competition experiments
subsequently demonstrated that the initiating kinase can rapidly
redistribute in response to the formation and dissociation of receptor
aggregates (37). Interpreting our DNP-BSA binding and receptor
phosphorylation data in terms of this model suggests that
redistribution of Lyn is slowed and the effects of competition for Lyn
is diminished at the lower temperature of 15°C.
In summary, our observations that DNP-BSA binds substantially more
slowly to cell-bound IgE than does equimolar DCT reveal that the
average effective number of DNP groups per BSA is less than one. After
eliminating simpler models, we found that this slow binding of DNP-BSA
to cell-bound IgE is consistent with a transient hapten exposure model
in which most DNP groups are unavailable for binding to FITC-IgE, but
they have a finite probability of exposure. At the DNP-BSA
concentrations that are suboptimal to optimal for cell activation,
Top
Abstract
Introduction
, 
, 
).
