See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257417959

The Hill equation and the origin of quantitative pharmacology

Article in Archive for History of Exact Sciences · July 2012

DOI: 10.1007/s00407-012-0098-5

CITATIONS READS

58 1,262

6 authors, including:

Rudolf Gesztelyi Judit Zsuga

University of Debrecen University of Debrecen
102 PUBLICATIONS 1,087 CITATIONS 53 PUBLICATIONS 423 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Limbal cell monolayer regeneration: a corneal surface regeneration modell View project

EIM in IBD View project

All content following this page was uploaded by Rudolf Gesztelyi on 10 February 2016.

The user has requested enhancement of the downloaded file.

The Hill equation and the origin of
quantitative pharmacology

Rudolf Gesztelyi, Judit Zsuga, Adam

Kemeny-Beke, Balazs Varga, Bela Juhasz
& Arpad Tosaki

Archive for History of Exact Sciences

ISSN 0003-9519
Volume 66
Number 4

Arch. Hist. Exact Sci. (2012) 66:427-438

DOI 10.1007/s00407-012-0098-5

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag. This e-offprint is for personal use only
and shall not be self-archived in electronic
repositories. If you wish to self-archive your
work, please use the accepted author’s
version for posting to your own website or
your institution’s repository. You may further
deposit the accepted author’s version on a
funder’s repository at a funder’s request,
provided it is not made publicly available until
12 months after publication.

1 23
 Author's personal copy
Arch. Hist. Exact Sci. (2012) 66:427–438
DOI 10.1007/s00407-012-0098-5

The Hill equation and the origin of quantitative

pharmacology

Rudolf Gesztelyi · Judit Zsuga · Adam Kemeny-Beke ·

Balazs Varga · Bela Juhasz · Arpad Tosaki

Received: 20 October 2011 / Published online: 31 March 2012

© Springer-Verlag 2012

Abstract This review addresses the 100-year-old Hill equation (published in

January 22, 1910), the first formula relating the result of a reversible association
(e.g., concentration of a complex, magnitude of an effect) to the variable concentra-
tion of one of the associating substances (the other being present in a constant and
relatively low concentration). In addition, the Hill equation was the first (and is the
simplest) quantitative receptor model in pharmacology. Although the Hill equation is
an empirical receptor model (its parameters have only physico-chemical meaning for
a simple ligand binding reaction), it requires only minor a priori knowledge about the
mechanism of action for the investigated agonist to reliably fit concentration-response
curve data and to yield useful results (in contrast to most of the advanced receptor
models). Thus, the Hill equation has remained an important tool for physiological
and pharmacological investigations including drug discovery, moreover it serves as a
theoretical basis for the development of new pharmacological models.

Communicated by : Manfred Laubichler.

R. Gesztelyi (B) · B. Varga · B. Juhasz · A. Tosaki

Department of Pharmacology, Faculty of Pharmacy, University of Debrecen,
P.O. Box 8, Debrecen 4012, Hungary
e-mail: [email protected]

J. Zsuga
Department of Neurology, Faculty of Medicine, University of Debrecen,
P.O. Box 31, Debrecen 4012, Hungary

A. Kemeny-Beke
Department of Ophthalmology, Faculty of Medicine, University of Debrecen,
P.O. Box 30, Debrecen 4012, Hungary

123
 Author's personal copy
428 R. Gesztelyi et al.

1 Introduction

As to methods there may be a million and then some, but principles are few. The
man who grasps principles can successfully select his own methods. The man
who tries methods, ignoring principles, is sure to have trouble.
Ralph Waldo Emerson (1803–1882)

This review addresses the Hill equation, published first in its presently known form
on January 22 1910. This equation possesses three of the most important features of
scientific models, i.e.,: it offers a didactical and mnemotechnic aid for understanding
and memorizing knowledge and/or information; it can be used to resolve practical
problems; and it can serve as a starting point for the development of more detailed
and particular models. The Hill equation is the first milestone in quantitative pharma-
cology, being the first formula that relates a consequence of a reversible association
(as an effect) to the variable concentration of one of two associating substances, as
long as the other substance is present in a constant and relatively low concentra-
tion. Therefore, the Hill equation was the first exact (quantitative) receptor model in
pharmacology. Beyond its historical role, the Hill model has theoretical and practical
significance even today. In this paper we will trace how the Hill equation was first
introduced, how it was modified and how it is currently used within pharmacology. The
more than 100 year history of this equation is a good illustration of a rather common
feature of theoretical models; the fact that they undergo multiple transformations in
light of expanding knowledge of the underlying phenomenon. These transformations
are generally of two kinds (1) a refinement of the original model/equation and (2) a
more precise understanding of its applicability in specific situations. The history of
the Hill equation shows both of those trends and is therefore a good case study for
understanding the development of quantitative models in the life sciences.

2 Historical background

The history of pharmacology, as a separate scientific discipline, goes back to the mid-
and late 19th century, to the works of Rudolf Buchheim (1820–1879) and his student
Oswald Schmiedeberg (1838–1921), generally recognized as the founders of modern
pharmacology (Scheindlin 2001; Rang 2006). Similarly to the evolution of physiol-
ogy from anatomy, pharmacology emancipated from physiology when it developed its
own methods to examine the living organism. Although the subject of investigation is
nearly the same for physiology and pharmacology, they are guided by different prin-
ciples (leitmotifs), as they address different questions: while physiology asks: how,
pharmacology questions: how to influence with chemical agents (Table 1)?
In the late 19th century, inspired by the spectacular advances in chemistry and phys-
ics, some scientists turned to apply mathematical approaches to phenomena in the life
sciences. In physiology, Carl Ludwig (1816–1895) and his followers, the “organic
physicists,” rejected the concepts of vitalism, and wanted to describe living nature
using mathematical formalism (Zimmer 1996). The mathematical approach in phar-
macology (more precisely: in a topic that is thought to be pharmacological nowadays)

123
 Author's personal copy
The Hill equation and the origin of quantitative pharmacology 429

Table 1 Principles and leitmotifs of some disciplines in life sciences

Discipline Principle

Anatomy The body is a machine (or like a machine, this question is of course out of the
scope of anatomy)
Physiology Every function of a living organism aims the maintenance of an equilibrium
(complex of equilibria is homeostasis)
Biochemistry The same laws apply to molecules in inanimate nature as those in living
organisms (metabolism is based on chemical reactions)
Pharmacology Functions of a living organism can be affected by certain exogenous agents
(called pharmacons if the influence is beneficial), with special regard to those
that can evoke or inhibit effects of endogenous regulatory molecules)
Quantitative Interaction between a pharmacon and a living organism can be described in a
pharmacology mathematical manner

Fig. 1 Archibald Vivian Hill,

British physiologist and
pharmacologist (1886–1977)

only emerged at the beginning of the 20th century. This breakthrough occurred in the
context of the receptor theory, a hypothesis that had emerged just before the birth of
quantitative pharmacology.
The suggestion that both exogenous and endogenous agents must be specifically
bound to structures inherent to the living organisms before exerting an effect was first
proposed by John Newport Langley (1852–1925) in 1878, based on his experiments
on salivary secretion in the dog (Maehle 2004). Paul Ehrlich (1854–1915) developed a
similar hypothesis starting from the results of his bacteriological investigations in 1897
(Maehle 2009). They refined these ideas in close competition, which is also reflected
in the terms they used to describe the specific chemical structure on the cell receiv-
ing the agent evoking a cellular response: “side chain” (Ehrlich in 1897), “receptor”
(Ehrlich in 1900; the term proved to be enduring), and “receptive substance” (Langley
in 1905) (Maehle 2004, 2009). The receptor concept can be summarized by the classic
maxim of Ehrlich: “Corpora non agunt nisi fixata” (agents do not act unless they are
bound) (Pelner 1972).

123
 Author's personal copy
430 R. Gesztelyi et al.

At the beginning of the 20th century, Archibald Vivian Hill (1886–1977; Fig.1) was
the first who quantified drug induced biological responses by introducing an equation
(Hill 1910) that has since been named after him. This relationship was then incorpo-
rated by Alfred Joseph Clark (1885–1941) into the first extensive model of receptor
function (Kenakin 2004; Colquhoun 2006; Rang 2006).

3 The Hill equation described for ligand binding (Hill–Langmuir equation)

Although the relationship, revealed by the Hill equation, is already implicit in Hill’s
first paper dealing with actions of nicotine and curare (or curari) on skeletal muscle
(Hill 1909), the equation, in the form known today, was first introduced in his sec-
ond publication describing the interaction of hemoglobin with oxygen (Hill 1910).
In the form it later became widely accepted, Hill’s formulation is a good approxi-
mation for every specific and saturable binding. In this context the term “specific,”
in its broadest sense, means that binding occurs between those particular substances
that are of interest for the examiner. In a narrower sense, a binding is specific when
there is a considerable affinity between the particular substances relative to the sin-
gle binding ability of both of these substances to other materials. In turn, the “sat-
urability” stems from the fact that one of the particular substances is present in a
significantly lower amount than the other(s). Thus, above a certain level, increasing
concentrations of this/these later can only slightly increase the amount of resulting
complexes. This formal relationship between binding substances was also published,
independently from Hill, a few years later by Irving Langmuir in relation to the adsorp-
tion of gases on plain surfaces (Langmuir 1918), therefore the Eq. 2 expressing this
link (see below) is referred to as Hill, Langmuir, Langmuir–Hill, and Hill–Langmuir
equation, respectively. In a recent proposal, the relationship between concentration
and effect has been defined as “Hill equation,” while the term “Hill–Langmuir equa-
tion” has been suggested for quantifying the ligand–receptor interaction (Neubig et al.
2003). For the sake of simplicity, we will use the term “Hill equation” throughout this
paper.
The Hill equation can be derived from the law of mass action (more precisely: from
the mass action model for equilibrium):

k1
−→
nL + R L R (1)
←− n
k2
[L]n [L]n
[L n R] = [R0 ] · = [R 0 ] · (2)
[L]n + K d [L]n + (K A )n

where L—the ligand that can be present in variable concentration; R—the receptor,
amount of which is constant and is significantly exceeded by the amount of the ligand;
[L n R]—the concentration of the ligand–receptor complex; [R0 ]—the total receptor
concentration (receptor number); [L]—the concentration of the free ligand (in practice,
however, it is treated as the total ligand concentration); k1 and k2 —the rate constants of
association and dissociation, respectively; K d = k2 /k1 —the equilibrium dissociation

123
 Author's personal copy
The Hill equation and the origin of quantitative pharmacology 431

constant of the ligand–receptor complex; K A —the ligand concentration, at which half

the receptors are ligand-bound (if n = 1, it equals the K d ); n—originally, the number
of binding sites for the given ligand in one receptor (as indicated in the Eq. 1), but
its interpretation is more complex than it would seem at first glance (it is commonly
referred to as the Hill coefficient or Hill slope factor, see below). K A serves as a mea-
sure for affinity of the ligand to the receptor: the smaller the KA , the greater the affinity
is. The [L]n /([L]n + (K A )n ) quotient (on the right side of the Eq. 2) is referred to as
fractional receptor occupancy.
There are some assumptions concerning Eq. 2. As mentioned above, an absolute
requirement is that the binding reaction reaches equilibrium, otherwise the rate of
association and dissociation affects the result (Kenakin 2009). Furthermore, the free
ligand concentration is assumed not to differ significantly from the total ligand con-
centration (this is fulfilled if total ligand concentration significantly exceeds the total
receptor concentration, which is common in biological systems). In addition, it is
important to point out that the reaction described by the Eq. 1 requires that, during
ligand binding, all binding sites of a receptor should be simultaneously occupied; in
other words, only the free L, the free R, and the fully bound L n R forms are permit-
ted. So, if incompletely ligand-bound receptors accumulate to a significant degree,
Eq. 2 fails to provide proper (=physico-chemically correct) equilibrium concentra-
tions of the permitted forms and/or constants characterizing the interaction (Weiss
1997). Consistently, although the Hill coefficient stems from the equation of the bind-
ing reaction (Eq. 1), it has been observed for a long time that the Hill equation only
yields an accurate estimate of the binding sites in one receptor, if there is excessive
positive cooperativity among these binding sites. This is due to the fact that an increas-
ing affinity of the receptor after the binding of the first ligand accelerates the binding
of the further ligands and thereby reduces the amount of incompletely ligand-bound
receptors (Weiss 1997). If a receptor has more than one binding sites and there is no
extreme positive cooperativity among them, the Hill coefficient underestimates the
number of binding sites and, in parallel, it overestimates the K d (or K A ). So, when
fitting the Hill equation to the data of a binding reaction, the Hill coefficient should be
considered as a compound measure of molecularity and cooperativity. Furthermore, if
the Hill coefficient significantly differs from unity, K A values should be treated with
caution (Weiss 1997; Goutelle et al. 2008). In conclusion, because of the complex
nature of the Hill coefficient, if its value is other than unity, fitted parameters provided
by the Hill equation for ligand binding cannot considered to be physico-chemically
meaningful without reservation. This fact that adds an empirical dimension to the use
of this equation.

4 The receptor theory and the Hill equation

The receptor theory is widely thought to be the concept most characteristic for phar-
macology (Kenakin 2004; Colquhoun 2006; Rang 2006; Maehle 2009). Following
preliminary suggestions by Langley and Ehrlich, Joseph Clark established the first
comprehensive receptor theory, the so-called occupation (or occupancy, or traditional
receptor-stimulus) theory (Clark 1926). Clark incorporated the Hill equation into the

123
 Author's personal copy
432 R. Gesztelyi et al.

occupation theory (referred to in this context as Langmuir’s adsorption isotherm). As

a result of Clark’s theoretical work, the Hill equation became the first quantitative
receptor model linking concentration of an agonist (a ligand capable of exerting a
cellular response) to the effect evoked.
In the context of the occupation theory, the Hill equation was used to evalu-
ate concentration-response (E/c) curves in a form, in which [L n R] and [R0 ] were
replaced by the effect of the agonist concentration and by the maximal effect achiev-
able with the given agonist in the given system, respectively. However, as it turned
out, features of a ligand–receptor binding cannot be simply extrapolated to the ago-
nist-effect relationship, i.e., the whole process of the receptor function. As a conse-
quence, the early version of the occupation theory often conflicted with experimental
observations.
To avoid these conflicts, an advanced form of the occupation theory was developed
assuming that receptor function consists of two major components: signal-recognition
(via formation of an agonist–receptor complex serving as a stimulus for the subsequent
structures), and signal-transduction (by means of changes produced by the stimulus)
(Ruffolo 1982; Black and Leff 1983; Kenakin 2004). The Hill equation (more pre-
cisely, Eq. 2 with n = 1) was used to describe the stimulus; thereby this part remained
the exact (parametric) element of the modified occupation theory. To describe the sig-
nal-transduction, a nonparametric function was introduced that related the stimulus to
the biological response (Stephenson 1956). Because this nonparametric function could
hardly be determined, the major use of this semiparametric occupation theory was to
study comparative assays, in which the mutual signal-transduction characterized by
the nonparametric function can be canceled out by generating ratios. This procedure
forms the basis for the so-called null method in pharmacology (Ruffolo 1982; Black
and Leff 1983).
To overcome the limitations of the occupation theory, attempts were made to develop
parametric receptor models more sophisticated than the Hill equation. The first of these
was the operational model of agonism that was derived from the following general
reaction equation (Black and Leff 1983):

k1 k1
−→ −→
mA + mR + ε m AR + ε (A R)m ε (3)
←− ←−

k2 k2

where A—the agonist; R—the corresponding receptor; ε—the effector having low
affinity for the free receptor but high affinity for the agonist–receptor complex (ε repre-
sents all postreceptorial elements playing a role in the generation of an effect); m—the
number of the agonist–receptor complexes that is necessary to bind to the effector to
evoke an effect (the so-called operational slope factor, see Eqs. 5 and 6 below); k1 and
k2 —the rate constants of association and dissociation for the agonist–receptor com-
plex, respectively; k1 and k2 —the rate constants of association and dissociation for the
agonist–receptor–effector complex, respectively; AR—the agonist–receptor complex;
(A R)m ε—the agonist–receptor–effector complex.
It should be noted that equations containing the equilibrium concentrations for the
first and the second part of the Eq. 3 are none other than the Hill equation described for

123
 Author's personal copy
The Hill equation and the origin of quantitative pharmacology 433

ligand binding and an analogue of the Hill equation described for signal-transduction,
respectively:

[A]
[A R] = [R0 ] · (4)
[A] + K A
[A R]m
E(∼ [(A R)m ε]) = E m · (5)
[A R]m + (K E )m

Substituting Eq. 4 into the Eq. 5 and then rearranging it slightly yields the following
equation:

([A] · [R0 ])m

E = Em · (6)
(K E · (K A + [A]))m + ([R0 ] · [A])m

where [A]—the agonist concentration; [R0 ]—the total receptor concentration; [AR]—
the concentration of the agonist–receptor complex; [(A R)m ε]—the concentration of
the agonist–receptor–effector complex (assumed directly proportional to the effect);
K A —the agonist concentration producing half-maximal receptor occupancy; E m —
the maximal effect achievable by stimulating the given receptor in the given system;
K E —the agonist–receptor complex concentration producing half-maximal E m ; E—
the effect elicited by the agonist at [ A].
In Eqs. 5 and 6, the m operational slope factor characterizes the signal-transduc-
tion as a whole rather than purely the molecularity of this process (similarly to the
Hill coefficient concerning the molecularity of the binding reaction). Because of this
general approach, the operational model preserves the comprehensive nature of the
earlier receptor models (Black and Leff 1983).
Several important methods in pharmacology are based on the occupation theory
(e.g., Furchgott’s method to determine K A as a measure of affinity: Furchgott and
Bursztyn 1967) as well as the operational model (to assess K A and parameters charac-
terizing efficacy of the agonist and effectiveness of the receptor-holding system, e.g.,:
Van der Graaf and Danhof 1997). Since the operational model contains both physico-
chemically meaningful (K A ) and arbitrary (m, K E , E m ) constants, some authors treat
it as a hybrid model that possesses features of both the so-called mechanistic (mech-
anism-based) and the empirical (phenomenological) models (Giraldo et al. 2002).
However, these models of agonist action have been challenged on theoretical
grounds (Colquhoun 1998, 2006). The essence of this objection is that the funda-
mental reaction equation of these models (e.g., the Eq. 3 in case of the operational
model) does not contain the following step already described in 1957 (Del Castillo
and Katz 1957):

k1
−→
AR A R∗ (7)
←−
k2

where: AR and AR*—the inactive and active agonist–receptor complex (before

and after a conformational change that enables the receptor to generate an effect),

123
 Author's personal copy
434 R. Gesztelyi et al.

respectively; k1 and k2 —the rate constants of the conformational change activating
and inactivating the receptor, respectively.
It should be noted that the above-mentioned objection does not apply to the use of
the Hill equation per se, because the Hill model does not divide the signal-transduc-
tion process, i.e., it does not provide separate constants for the different parts of the
signaling.
Subsequently, more recent receptor models have been developed on the basis of par-
ticular rather than general reaction equations. These mechanism-based models are far
more specific for a given sort of receptors, e.g., ternary complex models for G-protein
coupled receptors (Giraldo et al. 2002; Kenakin 2004). Despite being more accurate
in describing particular processes and molecular interactions, these specific models
are not the most suitable approaches for some practical purposes (e.g., curve fitting)
due to their high complexity. Thus, the simpler models did not lose their significance
in current pharmacology.

5 The Hill equation described for E/c curves

The Hill equation, described for E/c (agonist concentration) curves, is generally the
first choice to phenomenologically characterize the link between agonist concentra-
tion and effect. The most frequent way to do this is fitting the Hill equation to E/c
curve data using non-linear regression (Motulsky and Christopoulos 2003).
In agreement with the considerations discussed above, when fitting the Hill equa-
tion to E/c curves representing both the “cognitive” and the “transducing” functions
of the receptors, the provided parameters should be treated as best-fit values that geo-
metrically describe the curve. A classic form of the Hill equation defined for E/c
curves is:

cn
E = E max · (8)
cn + (EC50 )n

where c—the concentration of an agonist; E—the effect elicited by the agonist at

c; E max —the maximal effect that can be evoked by the given agonist in the given
system (geometrically: upper asymptote); EC50 —the concentration producing half-
maximum effect (geometrically: midpoint location); n—the Hill coefficient (geomet-
rically: a factor characterizing the slope of the E/c curve at the midpoint).
In this three-parametric (E max , EC50 , n) form of the Hill equation, the lower
asymptote is implicitly zero and therefore is not present in the equation. For sim-
plicity, this form is presented throughout this review. If the system has a basal activity
(E = 0 at c = 0), the four-parametric form (containing a lower asymptote) can be
used to determine precisely the baseline effect. On the other hand, subtraction of the
effect measured at c = 0 from every particular effect value enables the proper fitting
of the three-parametric Hill equation as well (Motulsky and Christopoulos 2003).

123
 Author's personal copy
The Hill equation and the origin of quantitative pharmacology 435

A B
100 100
E max E max

75 75

50 50
Emax /2

25 25

EC50
0 0
0 0.05 0.1 0.00001 0.001 0.1

Fig. 2 The Hill equation in linear and semilogarithmic plotting. Parameters for the plotted Hill equation
were chosen to represent a typical E/c curve (E max = 100%; EC50 = 700µM; n = 1). The x axis shows
the molar concentration (A using a linear scale; B using a logarithmic scale), while the y axis denotes the
effect in percentage of its maximal value. For this function (and for typical E/c curves), the midpoint of the
curve is the same as the point of symmetry (the so-called point of inflection) in the case of semilogarithmic
plotting (B)

6 Curve fitting with use of the Hill equation

When analyzing ligand binding or E/c curves, Eqs. 2 or 8 can be fitted to the curve
data, respectively. This procedure will supply regression parameters (best-fit values)
that define the shape and location (along the x axis) for the curve. When plotting the
concentration on a linear scale, the fitted Hill equation will generally follow a hyper-
boloid shape (Fig.2a). Geometrically, if n > 1, the fitted function will be sigmoid, but
this is only conspicuous at very great Hill coefficients. At the same time, when plotting
the logarithm of the concentration (or, equivalently, depicting the concentration on a
logarithmic scale), the fitted Hill equation will follow a sigmoid shape at any value of
n (Fig.2b) (Motulsky and Christopoulos 2003; Goutelle et al. 2008).
In order to get symmetrical confidence intervals and correct standard errors for the
obtained parameters (that allow proper comparison between different ligand binding
or E/c curves), it is recommended to plot the [L n R] or E versus the logarithm of
ligand concentration, and to fit a rearranged form of the Hill equation, in which the
agonist concentration and the concentration-like constant (K A or EC50 ) are expressed
as logarithm (Motulsky and Christopoulos 2003):

[R0 ]
[L n R] = (9)
1 + 10n·(log K A −log c)
E max
E= EC50 −log c)
(10)
1 + 10 n·(log

The reason for this is the fact that distribution of concentration-like constants is asym-
metric, but the logarithm of these constants follows a symmetric (Gaussian) distribu-
tion (Motulsky and Christopoulos 2003).

123
 Author's personal copy
436 R. Gesztelyi et al.

7 Features of the Hill equation

The only formal difference between the Hill equation and the Michaelis–Menten equa-
tion published three years later (Michaelis and Menten 1913) is the Hill coefficient,
i.e., the variable nature of the slope factor in the Hill equation. This suggests that the
equation of Leonor Michaelis (1875–1949) and Maud Leonora Menten (1879–1960),
relating velocity of an enzyme mediated reaction to the substrate concentration, is
equivalent with the Hill equation if n = 1. Indeed, both ligand–receptor binding and
substrate-enzyme association are specific, reversible and saturable. However, the equa-
tion of an enzymatic reaction is more complex than the Eq. 1, so the Michaelis-Menten
constant (K M ) is only apparently equivalent with K d in the Eq. 2, as K M depends on
the rate constant of product formation in addition to affinity of the substrate to the
enzyme.
An important feature of the Hill equation is that it exclusively determines symmet-
ric functions, for which the midpoint and the point of inflection is the same (Giraldo
et al. 2002) (when plotting semilogarithmically (see Fig.2b), point of inflection of
an E/c curve is the point having the greatest steepness, while the midpoint is asso-
ciated with half the maximal effect). Accordingly, use of the Hill model to evaluate
markedly asymmetric E/c curves may lead to incorrect estimates (Van der Graaf and
Schoemaker 1999). To resolve this problem, it is worth comparing the fit of a model
defining symmetric functions with the fit of a corresponding model accounting for
asymmetry (such a counterpart of the Hill equation is the Richards equation: Richards
1959). The simplest way for model comparison is to perform an F test. If the asymmet-
ric model fits the E/c data better, that should be used (Giraldo et al. 2002; Motulsky
and Christopoulos 2003).
Today, the Hill equation is classified as an empirical model (for E/c curves), sim-
ilarly to several other relatively simple equations (such as the so-called modified
Hill equation, Gompertz equation, Hodgkin equation, Douglas equation and Richards
equation). Parameters yielded by empirical models may lack physico-chemical mean-
ing, and in this case they only geometrically characterize the E/c curve (Giraldo et al.
2002; Keller et al. 2002). In contrast, the mechanistic models contain physico-chemical
constants allowing deeper insight to the signal amplification cascade and more proper
extrapolation from a test system (investigated experimentally) to the target system
(set up to be modeled by the test system) (Mager et al. 2003; Danhof et al. 2007).
To offer these benefits, however, mechanism-based models require detailed a priori
knowledge about particular mechanisms of the signal-transduction (Kenakin 2006). In
addition, too many parameters in the equation to be fitted can yield highly correlated
estimates with large standard errors (Van der Graaf and Stam 1999; Motulsky and
Christopoulos 2003). These considerations indicate that, under certain circumstances,
it is reasonable to use empirical models, especially the prototype Hill equation, which
is simple, flexible, and requires only minor previous knowledge about the properties of
the agonist and the pharmacological system investigated. Therefore, when screening
the biological activity of pharmacological agents, the first step is to take E/c curves
and then to fit them to the Hill equation (Kenakin 2006).

123
 Author's personal copy
The Hill equation and the origin of quantitative pharmacology 437

8 Conclusions

The Hill equation was the first quantitative (parametric) model that related the response
of a system to the concentration of a pharmacological agonist. Subsequently developed
receptor models generally incorporated the Hill equation, typically in more and more
complex formulae. Further research revealed that the parameters of the Hill equation
(or its simple derivatives) only have a precise physico-chemical meaning when applied
to ligand binding reactions (especially when the Hill coefficient is unity). However, in
contrast to most of advanced models developed since, the Hill equation requires only
minor a priori knowledge or assumptions (e.g., the symmetry of the E/c relationship)
about the mechanism of action for the investigated agonist in the given system to
reliably fit the E/c data and to provide useful results. In summary, simplicity, flexi-
bility and reliability of the Hill equation make it an essential tool for the first step of
E/c data analysis in physiological and pharmacological investigations, furthermore a
theoretical starting point for the development of new pharmacological models.

Acknowledgements This study was supported by grants of Hungarian Scientific Research Fund Programs
(OTKA-PD 78223; OTKA-K 72315) and the PhD Fellowship of Gedeon Richter Ltd.

References

Black, J.W., and P. Leff. 1983. Operational models of pharmacological agonism. Proceedings of the Royal
Society B: Biological Sciences 220: 141–162.
Clark, A.J. 1926. The antagonism of acetylcholine by atropine. Journal of Physiology (London) 61: 547–556.
Colquhoun, D. 1998. Binding, gating, affinity and efficacy: The interpretation of structure-activity rela-
tionships for agonists and of the effects of mutating receptors. British Journal of Pharmacology 125:
924–947.
Colquhoun, D. 2006. The quantitative analysis of drug-receptor interactions: A short history. Trends in
Pharmacological Sciences 27: 149–157.
Danhof, M., J. de Jongh, E.C. De Lange, O. Della Pasqua, B.A. Ploeger, and R.A. Voskuyl. 2007. Mech-
anism-based pharmacokinetic-pharmacodynamic modeling: Biophase distribution, receptor theory, and
dynamical systems analysis. Annual Review of Pharmacology and Toxicology 47: 357–400.
Del Castillo, J., and B. Katz. 1957. Interaction at end-plate receptors between different choline derivatives.
Proceedings of the Royal Society B: Biological Sciences 146: 369–381.
Furchgott, R.F., and P. Bursztyn. 1967. Comparison of dissociation constants and of relative efficacies of
selected agonists acting on parasympathetic receptors. Annals of the New York Academy of Sciences 144:
882–899.
Giraldo, J., N.M. Vivas, E. Vila, and A. Badia. 2002. Assessing the (a)symmetry of concentration-effect
curves: Empirical versus mechanistic models. Pharmacol Ther 95: 21–45.
Goutelle, S., M. Maurin, F. Rougier, X. Barbaut, L. Bourguignon, M., Ducher, and P. Maire. 2008. The
Hill equation: A review of its capabilities in pharmacological modelling. Fundamental and Clinical
Pharmacology 22: 633–648.
Hill, A.V. 1909. The mode of action of nicotine and curari, determined by the form of the contraction curve
and the method of temperature coefficients. Journal of Physiology (London) 39: 361–373.
Hill, A.V. 1910. The possible effects of the aggregation of the molecules of haemoglobin on its dissociation
curves. Journal of Physiology (London) 40: Proceedings iv–vii.
Keller, F., M. Giehl, D. Czock, and D. Zellner. 2002. PK-PD curve-fitting problems with the Hill equation?
Try one of the 1-exp functions derived from Hodgkin, Douglas or Gompertz. International Journal of
Clinical Pharmacology and Therapeutics 40: 23–29.
Kenakin, T. 2006. Data-driven analysis in drug discovery. Journal of Receptors and Signal Transduction
Research 26: 299–327.

123
 Author's personal copy
438 R. Gesztelyi et al.

Kenakin, T. 2004. Principles: Receptor theory in pharmacology. Trends in Pharmacological Sciences 25:
186–192.
Kenakin, T. 2009. Quantifying biological activity in chemical terms: A pharmacology primer to describe
drug effect. ACS Chemical Biology 4: 249–260.
Langmuir, I. 1918. The adsorption of gases on plane surfaces of glass, mica and platinum. Journal of the
American Chemical Society 40: 1361.
Maehle, A.H. 2004. Receptive substances”: John Newport Langley (1852–1925) and his path to a receptor
theory of drug action. Medical History 48: 153–174.
Maehle, A.H. 2009. A binding question: The evolution of the receptor concept. Endeavour 33: 135–140.
Mager, D.E., E. Wyska, and W.J. Jusko. 2003. Diversity of mechanism-based pharmacodynamic models.
Drug Metabolism and Disposition 31: 510–518.
Michaelis, L., and M.L. Menten. 1913. Die Kinetik der Intertinwerkung. Biochemische Zeitschrift 49:
333–369.
Motulsky, H.J., and A. Christopoulos. 2003. Fitting models to biological data using linear and nonlinear
regression. A practical guide to curve fitting. Oxford: Oxford Press (Corrected online version: http://
www.graphpad.com/manuals/Prism4/RegressionBook.pdf).
Neubig, R.R., M. Spedding, T. Kenakin, and A. Christopoulos. 2003. International union of pharmacology
committee on receptor nomenclature and drug classification. XXXVIII. update on terms and symbols in
quantitative pharmacology. Pharmacological Reviews 55: 597–606.
Pelner, L. 1972. Corpora non agunt nisi fixata. Maxim behind all of Ehrlich’s great discoveries. New York
State Journal of Medicine 72: 620–624.
Rang, H.P. 2006. The receptor concept: Pharmacology’s big idea. British Journal of Pharmacology 147:
S9–16.
Richards, F.J. 1959. A flexible growth function for empirical use Journal of Experimental Botany 10:
290–300.
Ruffolo, R.R. Jr. 1982. Review important concepts of receptor theory. Journal of Autonomic Pharmacology
2: 277–295.
Scheindlin, S. 2001. A brief history of pharmacology. Modern Drug Discovery 4: 87–88.
Stephenson, R.P. 1956. A modification of receptor theory. British Journal of Pharmacology 11: 379–393.
Van der Graaf, P.H., and M. Danhof. 1997. Analysis of drug-receptor interactions in vivo: A new approach
in pharmacokinetic-pharmacodynamic modelling. International Journal of Clinical Pharmacology and
Therapeutics 35: 442–446.
Van der Graaf, P.H., and Schoemaker, R.C. 1999. Analysis of asymmetry of agonist concentration-effect
curves. Journal of Pharmacological and Toxicological Methods 41: 107–115.
Van der Graaf, P.H., and W.B. Stam. 1999. Analysis of receptor inactivation experiments with the operational
model of agonism yields correlated estimates of agonist affinity and efficacy. Journal of Pharmacological
and Toxicological Methods 41: 117–125.
Weiss, J.N. 1997. The Hill equation revisited: Uses and misuses. FASEB Journal 11: 835–841.
Zimmer, H.G. 1996. Carl Ludwig: the man, his time, his influence. Pflugers Archiv 432: R9–22.

123
View publication stats