Molecular Biology of the Cell

Vol. 21, 4061– 4067, December 1, 2010

MBOC TECHNICAL PERSPECTIVE

A Guide to Simple and Informative Binding Assays
Thomas D. Pollard

Departments of Molecular Cellular and Developmental Biology, of Molecular Biophysics and Biochemistry
and of Cell Biology, Yale University, New Haven, CT 06520-8103

Submitted August 9, 2010; Revised August 18, 2010; Accepted August 26, 2010
Monitoring Editor: Douglas Kellogg

The aim of binding assays is to measure interactions between two molecules, such as a protein binding another protein,
a small molecule, or a nucleic acid. Hard work is required to prepare reagents, but flaws in the design of many binding
experiments limit the information obtained. In particular many experiments fail to measure the affinity of the reactants
for each other. This essay describes simple methods to get the most out of valuable reagents in binding experiments.

INTRODUCTION affinity. Journals should insist (where practical) that all binding
reactions be carried out at equilibrium and that reagents be
One of the most common experiments in biochemistry and used to measure (where practical) the equilibrium constant.
cellular and molecular biology is testing whether two mol- Otherwise precious reagents and hard work go to waste.
ecules interact with each other. This includes two proteins This essay explains proven experimental strategies to
binding to each other, a small molecule binding to a protein, maximize the information obtained from reagents painstak-
or a protein binding to a nucleic acid. In spite of apparent ingly acquired by biologists for binding studies. The tools
simplicity, many binding experiments reported in this and for powerful binding assays are available in most laborato-
other journals are poorly designed and fall short of extract- ries. The essay also points out features of well-designed
ing all of the useful information available from the valuable assays, which greatly simplify analysis of binding data.
reagents that were collected to do the experiment. In partic-
ular many experiments fail to measure the affinity of the
reactants for each other. This is a loss, because binding THE BASIC BINDING REACTION
reactions are not digital, with black/white, yes/no answers. Let’s begin with a basic, reversible, bimolecular binding reac-
Rather, the answer to whether two molecules interact with tion: molecule A binds to molecule B forming a complex AB:
each other should always be quantitative, with a number
that describes the affinity, or if binding is weak, a limit on the A ⫹ B % AB (Reaction 1)
possible affinity.
Conceptually such reactions are straightforward. In the
Many attempts to measure affinities fail to meet one or
forward reaction (indicated by the right facing arrow) one
both criteria for a successful binding experiment: the reac- molecule of A and one molecule of B collide and bind
tions must be at equilibrium at the time of measurement, together. In this case the stoichiometry is 1:1, which is true
and the concentration of one reactant must be varied. For for most biological binding reactions. Where the final prod-
example, after much labor an investigator prepares the re- uct consists of more than two molecules, the process usually
actants and mixes them to allow binding; so far, so good. But involves a series of reactions each with 1:1 stoichiometry. In
then the investigator makes the fatal mistake of disturbing the reverse reaction (indicated by the left facing arrow) one
the equilibrium by separating the reactants and products AB complex dissociates into A and B.
before measuring the extent of the reaction. For example, The association reaction A ⫹ B 3 AB involves two reac-
think of typical “GST pull down” or “immunoprecipitation” tants, so it is by definition a second order reaction. The rate
assays where pelleted material is washed free of the reac- of such a second order reaction is
tants. Even if the extent of the reaction is measured at
equilibrium, many forget to vary the concentration of one of Rate ⫽ k⫹ (A) (B) (Equation 1)
the reactants and thus miss the opportunity to measure the
The parameter k⫹ is a second order “association rate
constant” (lower case k) with units of M⫺1s⫺1 (pronounced
DOI: 10.1091/mbc.E10-08-0683 per molar per second). (A) and (B) are the free concentra-
Address correspondence to: Thomas D. Pollard (thomas.pollard@ tions of the molecules available for reacting at the given
yale.edu). moment in time. A useful rule of thumb is that association
rate constants for molecules the sizes of typical proteins are
© 2010 T. D. Pollard. This article is distributed by The American
Society for Cell Biology under license from the author(s). Two
often in the range of 106 to 107 M⫺1s⫺1. These reactions are
months after publication it is available to the public under an typically limited by the rates of collisions between mole-
Attribution–Noncommercial–Share Alike 3.0 Unported Creative cules, which are determined by both the rates of diffusion of
Commons License (http://creativecommons.org/licenses/by-nc- the molecules and the sizes of their interaction surfaces.
sa/3.0). Small molecules diffuse faster than large molecules, but this

4061
T. D. Pollard

is offset by smaller interaction surfaces, restricting most negative free energy change, which drives the reaction to-
protein association constants to this relatively small range. ward products.
For more detail see Chapter 4 in Pollard and Earnshaw
(2007).
When a complex of two molecules, AB, dissociates, AB 3 TWO STRATEGIES TO MEASURE EQUILIBRIUM
A ⫹ B, AB is the only reactant, so by definition this is a CONSTANTS
first-order reaction. The rate such a first-order dissociation Two general options are available to measure affinities. 1) In
reaction is an equilibrium experiment, one determines the extent of the
reaction as a function of the concentration of one of the
Rate ⫽ k⫺ (AB) (Equation 2) reactants. Analysis of these data gives the equilibrium con-
The parameter k⫺ is a first-order rate constant called a stant. 2) In a kinetic experiment, one determines the rates of
“dissociation rate constant” with units of s⫺1 (pronounced the forward and reverse reactions as a function of the con-
per second) and (AB) is the concentration of AB. The disso- centration of one of the reactants. Analysis of these data
ciation rate constant is simply the probability that the com- gives the rate constants for the forward and reverse reac-
plex will fall apart in a unit of time. tions. The ratio of these rate constants gives the equilibrium
constant.
The kinetic approach is more revealing, because it gives
EQUILIBRIUM CONSTANTS not only the thermodynamic parameter, the equilibrium
A number called the equilibrium constant characterizes the constant, but also the rate constants, which characterize the
affinity of molecules for each other. A few simple relation- dynamics of the system. Note that the street is one way from
ships define equilibrium constants and form the basis for rate constants to equilibrium constants. Knowing only the
determining their values experimentally. By definition at affinity from an equilibrium experiment reveals nothing
equilibrium the forward and reverse rates of a reaction are about the rates of the forward or reverse reactions. Although
equal, so at equilibrium the forward and reverse rates of our generally stronger than equilibrium experiments, kinetic ex-
binding reaction are equal: periments often require more reagents and more sophisti-
cated assays.
rate of binding ⫽ k⫹ (A)(B) ⫽ k⫺ (AB) ⫽ rate of dissociation
(Equation 3) EQUILIBRIUM BINDING EXPERIMENTS

By definition the equilibrium constant (capital K) is equal This section outlines the strategy to measure affinity with an
to the ratio of the forward and reverse rate constants or the equilibrium binding experiment. The following sections pro-
ratio of the concentration of products to the concentrations vide practical advice about each step.
of free reactants at equilibrium. The reaction in question is simple reversible binding of
molecules A and B:
Keq ⫽ k⫹/k⫺ ⫽ (ABeq)/(Aeq)(Beq) (Equation 4)
A ⫹ B % AB,
Defined in this way the equilibrium constant Keq for the
binding reaction has the units of M⫺1. This definition has the where
advantage that the value of Keq is proportional to the affin-
ity. Thus the larger the value of Keq, the stronger the reaction Keq ⫽ (ABeq)/(Aeq)(Beq) and Kd ⫽ (Aeq)(Beq)/(ABeq).
and the more completely the reactants A and B are con- (Equations 4 and 5)
verted into the product AB.
The reciprocal of this expression for a binding reaction We will assume that the stoichiometry is 1:1. Other more
describes the “dissociation equilibrium constant” or Kd complicated outcomes are possible, but this is a reasonable
starting assumption.
Kd ⫽ k⫺/k⫹ ⫽ (Aeq)(Beq)/(ABeq) (Equation 5) The task is to measure the equilibrium concentrations of
A, B, and AB over a range of concentrations of one of the
This form of the equilibrium constant has the units of M,
reactants, A or B. With good design, this experiment is
moles per liter. Biologists favor this definition owing to the
simple compared with the effort required to make the re-
familiar units. Defined thus, the lower the value of Kd, the
agents. With a powerful assay the whole experiment may
stronger the reaction, and the more completely reactants A
take only a few minutes. Five steps yield the affinity.
and B are converted into the product AB.
Step 1: Develop a sensitive assay to measure the concen-
Given the small range of association rate constants, the value
tration of Aeq, Beq, or AB. Note that (Aeq) and (Beq) are the
of the dissociation rate constant often determines the affinity.
concentrations of free A and free B. To reinforce this essen-
Low affinity interactions with Kds in the micromolar range
tial point, this essay will use (Afree) and (Bfree). Measuring
have dissociation rate constants around 1 s⫺1. For a first-order
one of these concentrations usually suffices for this simple
reaction this corresponds to a halftime for dissociation of 0.7 s
type of reaction, because one generally knows the total
(t1/2 ⫽ ln 2/k ⫽ 0.693/k). On the other hand, high-affinity
concentrations of A and B and therefore can calculate the
interactions with Kds in the nanomolar range have dissociation
unknown concentrations by difference from the total con-
rate constants around 0.001 s⫺1 and half lives ⬎10 min.
centrations and measurements of (Afree), (Bfree), or (AB).
The equilibrium constant is a thermodynamic parameter
Step 2: Design an experiment with the lowest concentra-
directly related to the free energy change (⌬G) for the reac-
tion of Afree, Bfree, or AB that can be measured accurately.
tion under a particular set of conditions.
Put some thought into which reactant to fix and which to
⌬G ⫽ ⫺RT ln Keq (Equation 6) vary. Make the decision based on the abundance of the
reactants available and the ease of the measurements The
where R is the gas constant and T is the absolute temper- following examples use a low concentration of A. This key
ature. Favorable reactions have a large Keq and a large decision greatly simplifies the experiment and its analysis.

4062 Molecular Biology of the Cell

 Binding Reactions

Step 3: Set up a series of reactions, each with a fixed, low

concentration of total A and a wide range of concentrations
of B. The concentrations of Bfree should span a range from
less than the Kd to high enough to convert essentially all of
A into AB. Pilot experiments will reveal the appropriate
concentrations.
Step 4: Let the reactions equilibrate. The time required is
often much less (seconds) than often assumed. Then mea-
sure either (Afree), (Bfree), or (AB) without disturbing the equi-
librium.
Step 5: Plot the concentration of AB versus the concentra-
tion of Bfree and calculate the Kd from the shape of the curve
(as explained below) or from the concentration of Bfree re-
quired for half of Atotal to be converted to AB.

ADVICE ON ASSAYS TO MEASURE BINDING Figure 1. A quantitative GST pull down assay. Supernatant deple-
REACTIONS tion assay for GST-2/3/A (part of the tail of Acanthamoeba myo-
sin-I) binding amoeba Arp2/3 complex. Conditions: 0.3 mM
The assay for Afree, Bfree, or AB should be sensitive and as amoeba Arp2/3 complex in 65 mM NaCl, 10 mM Tris (pH 7), 1 mM
simple as possible. Note again that Afree and Bfree are the DTT was titrated with (F) GST-2/3/A immobilized on glutathione
concentrations of free A and free B, not the total concentra- beads or (E) the same volumes of glutathione beads without GST-
tions. The assay can be an optical measurement, a chemical 2/3/A. Beads were pelleted at 16,000 ⫻ g, and supernatant proteins
were separated by SDS-PAGE, stained with Coomassie, and
measurement, the density of the band on a gel or an immu-
scanned to measure the fraction bound. The binding isotherm gives
noblot or anything else, but the signal must be directly a Kd of ⬃3 ␮M. (Inset) Coomassie-stained gel of the supernatants
proportional to the concentration. The following paragraphs showing Arp2/3 complex depletion. Numbers above each lane
explain how to use chemical and optical assays for binding indicate total GST-2/3/A ␮M concentration on the beads (Lee et al.,
reactions. 2000).

Chemical Assays
To measure affinity carry out this assay with a low, fixed
When properly calibrated, chemical assays including gel
concentration of A and a range of concentrations of B at-
electrophoresis, ELISA, and immunoblots are excellent ways
tached to beads (by varying the concentration of beads). The
to measure concentrations with materials available in most
equilibrium concentration of AB as a function of Bfree (⬃total
labs, but these assays must be done without altering the
B) is the data required to calculate the affinity. Figure 1
equilibrium. Making chemical measurements at equilibrium
shows an example of such a quantitative pull down assay,
is generally possible but rarely achieved.
where gel electrophoresis and Coomassie blue staining were
used to measure the concentration of a purified Arp2/3
Faults with Nonequilibrium Chemical Assays
complex in the supernatant after equilibration with a range
Typical assays prepare an equilibrium mixture of reactants of concentrations of a GST-ligand bound to beads. In this
and products but then separate products from reactants, at case the ligand was a region from the tail of myosin-I.
which point the products start to dissociate to achieve a new Note that this strategy can be applied to a crude system, if
equilibrium. Examples of this mistake are assays where an assay such as a quantitative immunoblot or ELISA is
products are pelleted, the supernatant containing the free available to measure free A. For example one can titrate a
reactants is removed, and the pellet is washed one or more crude cellular extract, diluted to contain a low concentration
times with buffer without reactants. Because the solution of A, with a range of concentrations of B bound to beads and
containing the resuspended pellet has no reactants, the for- measure free A remaining in the supernatant.
ward reaction rate is zero and the products will dissociate. If If one finds it necessary to measure the complex of AB in
the affinity is low, the dissociation reaction is typically fast, the pellet, this can be done with minimal washing using a
for example ⬎1 s⫺1 for a Kd of 1 ␮M. Therefore the products cushion of glycerol or sucrose in the bottom of the tube. By
will dissociate with a half-time of ⬍1 s, making it impossible keeping the washing time brief (seconds) the equilibrium
to measure the equilibrium concentration of products. will not drift too far. Washing through a cushion is done by
1) equilibrating the reaction, 2) pelleting the beads
Equilibrium Chemical Assays through a small volume of buffer with 5–10% glycerol or
One way to use a chemical assay at equilibrium is to sepa- sucrose, 3) removing the supernatant including most of
rate products from reactants without disturbing the equilib- the cushion, and 4) measuring bound B in the pellet.
rium (Figure 1). If the products differ in sedimentation co-
efficient from one of the reactants, the products can be The Advantages of Optical Assays
pelleted leaving an equilibrium mixture of reactants in the The rewards for developing an optical assay outweigh the
supernatant. For example, reactant B can be attached to a effort, because optical measurements have the advantages of
bead (through a GST tag, for example) and equilibrated with speed and simplicity. Speed allows for both equilibrium and
a solution of a low concentration of reactant A. Simply kinetic measurements. Good optical assays measure the con-
pelleting the beads does not change the equilibrium but centration of one component in equilibrium with the other
separates all of the B attached to the beads along with any A components of the reaction.
bound to B from the free A in the supernatant. Rather than
examining the pellet, simply measure the concentration of Fluorescence Intensity
free A in the supernatant and calculate by difference the The simplest optical assay is a difference in the fluorescence
concentration of AB in the pellet. intensity of the product compared with the reactants. In a

Vol. 21, December 1, 2010 4063

T. D. Pollard

Figure 2. Equilibrium binding of poly-L-proline oligomers to pro-

filin measured by the increase in intrinsic fluorescence. These ex-
periments tested the effects of peptide size and ionic strength on this
simple bimolecular binding reaction. Conditions: 5 ␮M Acan-
thamoeba profilin I in 10 mM Tris (pH 7.5), 75 mM KCl, and 3.1 mM
NaN3, 22°C. Excitation 295 nm, emission intensity 318 nm. All data
sets are normalized (F/Fmax) to their maximum fluorescence
Figure 3. Equilibrium binding of fission yeast cofilin to pyrene-
change. Equilibrium dissociation constants (Kd) were obtained by
labeled ADP-actin filaments measure by titrating 2 ␮M pyrenyl
fitting data directly using nonlinear least squares analysis. (F) L-
actin filaments with 50 nM to 10 ␮M S. pombe cofilin. Cofilin binding
proline10, Kd ⫽ 433 ␮M proline; (*) L-proline15, Kd ⫽ 344 ␮M
quenches the fluorescence and was used to measure the extent of the
proline; (O) L-proline82, Kd ⫽ 120 ␮M proline; and (䉬) L-(proline-
reaction. This plot of binding density (fraction of actin bound de-
proline-glycine)5, Kd ⫽ 34.1 mM proline; and (
) L-(proline-hy-
termined from the extent of quenching of fluorescence) versus free
droxyproline-glycine)5 (Petrella et al., 1996).
cofilin concentration is sigmoidal owing to positive cooperativity.
The apparent Kd ⫽ 1/w Ka, where w is the cooperativity factor (8.5
in this case) and Ka ⫽ k⫹/k⫺ measured in kinetics experiments to be
0.12 ␮M⫺1, so the apparent Kd is 1.0 ␮M (Andrianantoandro and
few favorable cases the intrinsic tryptophan fluorescence of Pollard, 2006).
a protein changes when a ligand binds. For example, poly-
L-proline binding increases the intrinsic tryptophan fluores-
cence of profilin (Figure 2). Unfortunately, few reactions
give an intrinsic fluorescence signal, but do not fail to check these products. However, the signal is usually noisy compared
whether your system gives such a signal, because it will save with fluorescence assays. Keep in mind that light scattering
much time and effort. A variant of this approach is a change scales with the inverse fourth power of the wavelength, so
in the fluorescence intensity of a dye bound to one of the using short wavelengths gives much higher signals.
reactants when the product forms, but again such changes
are not common. One example is the decrease in fluores- ADVICE ON EXPERIMENTAL DESIGN
cence of pyrene-labeled actin filaments when cofilin binds.
This assay was used to collect the binding data for Figure 3. An ideal binding experiment has a fixed total concentration
of one of the two reactants, let’s choose A, that is lower than
Fluorescence Anisotropy the Kd. Ten-fold lower than the Kd is fine. Using a low
A universally useful approach is to measure the fluorescence concentration of total A simplifies the execution of the ex-
anisotropy of one reactant tagged with a fluorescent dye. periment and interpretation of the data, because essentially
Fluorescence anisotropy measures the rotational diffusion of all of the other reactant B is free under all conditions. The
a molecule, so when the reactant binds its partner, the larger practical limitation is that the assay must be able to measure
product has a lower rotational diffusion coefficient and a the concentration of free or bound A.
higher fluorescence anisotropy. One example is the increase
in fluorescence anisotropy when a rhodamine-labeled frag-
ment of the Wiskott Aldrich syndrome protein (the VCA
region of WASp) binds Arp2/3 complex (Figure 4). One
typically uses a very low concentration of the tagged mole-
cule and titrates in a range of concentrations of the untagged
partner molecule. This gives the affinity of the tagged mol-
ecule for its partner. The fluorescent dye may influence the
binding reaction, so an essential control with any tagged
molecule is a competition experiment with the untagged
molecule. Fixed concentrations of the tagged molecule and
its partner are titrated with a range of concentrations of the
untagged molecule to measure the affinity of the untagged
molecule for the partner.

Other Optical Assays

Fluorescence energy transfer is also useful to measure bind- Figure 4. Fluorescence anisotropy assay of equilibrium binding of
ing but requires fluorescent tags on both reactants, doubling a rhodamine-labeled ligand to Arp2/3 complex. Rhodamine-
the work in reagent preparation and doubling the chances of WASp-VCA (0.025 ␮M) was titrated with the Arp2/3 complex in the
artifacts due to the labels. Some products scatter more light presence (F; Kd ⬃ 0.05 ␮M) or absence (
; Kd ⫽ 0.25 ␮M) of 5 ␮M
than the reactants, so light scattering can be used to measure actin filaments (Marchand et al., 2001).

4064 Molecular Biology of the Cell

 Binding Reactions

If one chooses a low concentration of total A, then the and a half saturating value of Btotal is equal to Kd. No math,
experiment is to determine the extent of the reaction over a software, or even much thinking is required with good
wide range of concentrations of B. When the concentration experimental design.
of B is below the Kd, most of B is free, because, given the low On the other hand, if the assay requires that the experi-
concentrations of both reactants, little binding will occur. At ment be set up with a concentration of Atotal in the range of
higher concentrations of B more AB forms, but the concen- the Kd, a considerable fraction of total B will be bound to A.
tration of A is so low that virtually all of B is free. Because Therefore an additional calculation is required to determine
most of B is free both below and above the Kd, Bfree ⬃ Btotal Bfree. Under these circumstances a quadratic equation can
one does not have to compensate for any B bound to A when used to fit the binding data and calculate the Kd.
analyzing the data.
If the assay lacks the sensitivity to do the experiment with 关LR兴/关L兴 ⫽ ((关R兴 ⫹ 关L兴 ⫹ Kd) ⫺ ((关R兴 ⫹ 关L兴 ⫹ Kd) ˆ 2 ⫺
a concentration of A below the Kd, then use the lowest
practical concentration of A. Later, when analyzing the data 4*关R兴*关L兴) ˆ 0.5))/2*关L兴 (Equation 8)
(see below), compensate for the finite amount of B bound to
where [LR] is the concentration ligand bound to the recep-
A at each concentration of total B in the sample.
tor, [R] is the total concentration of receptor (Btotal in our
example, which is varied), [L] is the total concentration of
ANALYSIS OF EQUILIBRIUM BINDING DATA ligand (Atotal in our example), and [LR]/[L] is the fraction of
ligand bound to receptor. This equation can be fit to the data
One plots the concentration of AB as a function of the to obtain the Kd. Using a low concentration of Atotal avoids
concentration of Bfree. This plot will reveal the extent of the having to deal with this equation.
reaction. Providing that the experiment included high con- If the concentrations of both A and B are far above the Kd,
centrations of B, B will saturate A and the concentration of then essentially all of the added B will bind to A up to the
AB will reach a plateau. Without reaching this plateau, one point where A is saturated. Little or nothing is learned about
does not know the amplitude of the reaction and, unfortu- the affinity in such an experiment, except that the Kd is less
nately, without the amplitude, one cannot calculate the equi- than the concentrations of A and B.
librium constant. Many binding experiments reported in
reputable journals fail to saturate the reaction and reach a
plateau of AB, compromising the value of the hard work. A COOPERATIVE BINDING REACTIONS
Scatchard plot is an alternative method to plot the data as
(bound)/(free) versus (bound). This formulation was useful Sometimes the plot of (AB) versus (Bfree) is not a hyperbola.
in the days before computers, because such plots of the data This indicates something more complicated than a simple
for a simple bimolecular reaction are linear. However, Scat- bimolecular binding reaction. For example, when the protein
chard plots are now discouraged, because they can easily cofilin binds to actin filaments, the binding curve is sigmoi-
hide the fact that a binding curve did not reach a plateau. dal, evidence for cooperativity in the reaction (Figure 3).
Let’s take a closer look at a binding curve such as Figure Cooperativity means that binding of one ligand molecule to
1. In a successful binding reaction, when Bfree is zero, all A a receptor influences the affinity of subsequent ligand mol-
is free and (AB) ⫽ 0. At the other extreme, when the con- ecules to the same receptor. Binding of oxygen to the four
centration of Bfree is high, all active A is bound to B. The sites on hemoglobin is the classic example (Morgan and
binding curve can reveal whether some A is inactive, be- Chichester, 1935), where each successive bound oxygen in-
cause at very high concentrations of B some A is still free. It creases the affinity for subsequent oxygens. Both of these
is important to document in this way the amount of inactive examples exhibit positive cooperativity.
A, which can be ignored in the analysis. Some A may be Analysis of cooperativity requires models based on physical
inactive, because it is denatured, but the existence of inactive interactions, an interesting subject beyond the scope of this
A may be interesting and important from a biological per- essay. Figuring out mechanisms in cooperative systems can be
spective. For example, some A may be inactive because a complicated. Investigators have been working on coopera-
posttranslational modification compromises binding to B. tive binding of oxygen to hemoglobin for more than 50 years
Two methods are available to determine the equilibrium and cooperative binding of cofilin to actin filaments for
constant from the binding curve. For a simple bimolecular nearly two decades. For the purposes of this essay, the
binding reaction a plot of the fraction of total molecule A that important point is that a binding curve from a well-designed
is bound (the fractional saturation) versus Bfree is a hyperbola: assay will distinguish a simple bimolecular reaction from a
cooperative reaction.
Fractional saturation of A ⫽ (Bfree)/关Kd ⫹ (Bfree)兴
(Equation 7) KINETIC BINDING EXPERIMENTS
The shape of the hyperbola depends on the equilibrium As noted above kinetic experiments yield more information
constant. Readily available software can be used to fit the about binding reactions than equilibrium experiments, but
equation to the data and calculate the equilibrium constant. kinetic experiments are much less common. One reason is
Inspection of the equation reveals that the concentration of that more materials, better assays, and specialized equip-
Bfree required to convert half of A to AB (fractional satura- ment are sometimes required. Another reason is that many
tion ⫽ 0.5) is equal to the Kd, which can be read directly investigators have no classroom or practical experience with
from a binding curve (Figure 1). the type of kinetic assays used to evaluate binding reactions.
Practically speaking, experiments are set up with concen- Most biochemistry courses cover only steady state enzyme
trations of Atotal either below the Kd or higher. If Atotal is kinetics. Although useful in some circumstances, neither the
well below the Kd, essentially all B is free at every concen- experiments nor the methods of analysis are appropriate for
tration of B used in the experiment, so Bfree ⬃ Btotal. Under binding experiments. To make matters worse, most students
these circumstances a plot of (AB) versus Btotal is a hyper- find analysis of steady state enzyme kinetics by Michaelis–
bola with a shape dependent on the equilibrium constant Menton equations baffling or irrelevant to their work.

Vol. 21, December 1, 2010 4065

T. D. Pollard

Analysis of binding reactions requires a different type of Note that if one knows Kd from an equilibrium experi-
kinetic experiments called presteady state or transient state ment and measures k⫺ or k⫹, one can calculate the missing
kinetic experiments. Fortunately, the concepts behind and rate constant, because Kd ⫽ k⫺/k⫹.
the execution of presteady experiments are not complicated.
One simply changes conditions and observes the time course
LINKED REACTIONS
of the reequilibration of the system. For example, mixing A
and B changes the conditions. Then one watches what hap- Many binding reactions actually occur in two steps:
pens over time. Pretty simple.
A detailed explanation of the methods used in presteady A ⫹ B % AB % AB* (Reaction 2)
state kinetics is beyond the scope of this essay, but a few
where the conformation of AB* differs from AB and the
general points are useful for any biologist. Let’s consider our
second reaction is a pair of reversible first-order reactions.
binding reaction from a kinetics perspective. The reaction is
Often such a conformational change is favorable and will
A ⫹ B % AB. The goal is to measure the rate constants, k⫹
pull the overall reaction to the right. Most chemical assays
and k⫺, for the two reactions. Some reactions can be mea-
will not distinguish AB and AB*, but the signals from AB
sured manually by using low concentrations of reactants to
and AB* differ in some optical assays, so kinetics can be used
slow the reactions. For fast reactions that take place on a
to characterize the second reaction.
subsecond time scale mechanical mixing in stopped flow or
Everyone trying to understand molecular mechanisms, even
quenched flow devices is required.
experts on kinetics and thermodynamics, is tempted to violate
fundamental principles by proposing mechanisms that sound
Forward Binding Reaction nice but are impossible thermodynamically. One can avoid this
The second order reaction is A ⫹ B 3 AB, with rate ⫽ trouble by paying attention to something called detailed bal-
k⫹(A)(B). One can measure the rate constant by observing ance or “microreversibility,” which outlaws perpetual motion
the time course of the formation of AB after mixing A and B. machines. The concept of detailed balance is that the free
To simplify the experiment one usually uses “pseudofirst- energy change around any cycle of reactions must be zero.
order conditions” with a low concentration of one reactant, Thus the product of the equilibrium constants around the cycle
let’s say A, and a range of higher concentrations of the other must be one. An “energy square” with the linked binding
reactant, let’s say B. As in the equilibrium binding experi- reactions among molecules A, B, and C illustrates the concept.
ments, the low concentration of A means that most of B is
free and its concentration does not change as A is converted K1 ⫽ k⫹1/k⫺1
to AB. For the purpose of this argument, we will assume that
binding is irreversible. With B in excess, the time course of B % AB
the formation of AB (⫽ the time course of the depletion of A)
is an exponential with an observed rate constant kobs ⫽ k⫹ K4 ⫽ k⫺4/k⫹4 ( ( K2 ⫽ k⫹2/k⫺2
(B). By varying (B) one can evaluate k⫹. BC % ABC
Dissociation Reaction K3 ⫽ k⫺3/k⫹3
The rate of a first-order dissociation reaction AB 3 A ⫹ B is
Note that direction of each reaction is defined in one
equal to k⫺(AB). One can observe this reaction in several
direction around this square, so reactions 1 and 2 are given
ways. Simplest, one can dilute an equilibrium mixture of A,
as association reactions with Keq ⫽ k⫹/k⫺ while reactions 3
B, and AB and observe the time course of the dissociation of
and 4 are given as dissociation reactions with Kd ⫽ k⫺/k⫹.
AB to establish new equilibrium concentrations of A and B.
If K1 ⫻ K2 ⫻ K3 ⫻ K4 does not equal 1, something is wrong.
Another strategy useful with optical assay is to compete a
One can use energy squares to calculate unknown equi-
fluorescent ligand from its receptor with a large excess of
librium constants. For example, if one knows that A binds to
unlabeled ligand. The fluorescent ligand dissociates at a rate
B with Keq ⫽ 10 ␮M⫺1 (association reaction), B binds to C
determined by k⫺ and does not rebind, because the unla-
with Kd ⫽ 10 ␮M (dissociation reaction), and AB binds to C
beled ligand takes its place.
with Keq ⫽ 1 ␮M⫺1 (association reaction), one can calculate
Because many association reactions have rate constants in
the affinity of BC for A.
the range of 106 to 107 M⫺1s⫺1, the rate constant for dissocia-
tion of the products is usually the main determinant of the
affinity. For example, dissociation rate constants will be on the MAIN POINTS TO CONSIDER WHEN PLANNING A
order of 1 s⫺1 for a low affinity reaction with a Kd of 1 ␮M and BINDING EXPERIMENT
0.001 s⫺1 for a high-affinity reaction with a Kd of 1 nM.
• Create a good assay to measure reactants or products.
Dealing with Reversible Reactions • Consider the advantages of optical assays.
• Use a low concentration of one of the reactants.
These examples ignore the reality that the binding reaction is
reversible, which is legitimate for pseudofirst-order condi- • Vary the concentration of the other reactant, including
tions or the competition dissociation experiment. On the concentrations that saturate the first reactant.
other hand, many kinetics experiments need to take both • Do not disturb the equilibrium when measuring con-
reactions into account—a more advanced topic. An impor- centrations of reactant or products.
tant but surprising feature of a reversible binding reaction is • Calculate the equilibrium constant from the dependence
that kobs for the formation of AB is equal to the sum of the of complex formation on the concentration of the varied
rate constants for the two reactions, so for our reaction under reactant.
pseudofirst-order conditions kobs ⫽ k⫹ (B) ⫹ k⫺. One way to • Consider measuring rate constants to learn more about the
reconcile this in your mind is that the time required for reaction.
equilibration after mixing A and B is less, because the re- • Check detailed balance to avoid pitfalls in linked reac-
verse reaction limits the extent of the reaction. tions.

4066 Molecular Biology of the Cell

 Binding Reactions

ACKNOWLEDGMENTS Lee, W.-L., Bezanilla, M., and Pollard, T. D. (2000). Fission yeast myosin-I,
Myo1p, stimulates actin assembly by Arp2/3 complex and shares functions
The author thanks his laboratory colleagues Jon Goss, Irene Reynolds Tebbs, with WASp. J. Cell Biol. 151, 789 – 800.
and Shih-Chieh Ti for their suggestions on the text. This work in the author’s Marchand, J.-B., Kaiser, D. A., Pollard, T. D., and Higgs, H. N. (2001).
laboratory is supported by National Institutes of Health research grants Interaction of WASp/Scar proteins with actin and vertebrate Arp2/3 com-
GM-026132, GM-026338, and GM-066311. plex. Nat. Cell Biol. 3, 76 – 82.
Morgan, V. E., and Chichester, D. F. (1935). Properties of the blood of the
domestic fowl. J. Biol. Chem. 110, 285–298.
REFERENCES Petrella, E. C., Machesky, L. M., Kaiser, D. A., and Pollard, T. D. (1996).
Structural requirements and thermodynamics of the interaction of proline
Andrianantoandro, E., and Pollard, T. D. (2006). Mechanism of actin filament peptides with profilin. Biochemistry 35, 16535–16543.
turnover by severing and nucleation at different concentrations of ADF/ Pollard, T. D., and Earnshaw, W. C. (2007). Cell Biology. Second Edition. W. B.
cofilins. Mol. Cell 24, 13–23. Saunders, New York. pp. 902.

Vol. 21, December 1, 2010 4067