The Regulatory Design of Metabolism An Exercise in Computational Systems Biology

The Regulatory Design of Metabolism
An exercise in computational systems

biology
Jan-Hendrik S. Hofmeyr,
Centre for Studies in Complexity
and
Triple-J Group for Molecular Cell Physiology,
Dept. of Biochemistry,
University of Stellenbosch,
South Africa

c 2009
April 13, 2009
1
Contents
1 The Kinetic Model for a Network of Coupled Reactions 4

1.1 The basic linkages and structures in metabolism . . . . . . . . . . . . . . . . . . . 4
1.2 The kinetic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 A hierarchy of kinetic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 The mass-action form of the kinetic model . . . . . . . . . . . . . . . . . . 11
1.3.2 The rate law form of the kinetic model . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 The power-law form of the kinetic model . . . . . . . . . . . . . . . . . . . 13
1.4 Metabolic states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.1 Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.2 Steady state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.3 Transient state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5 The kinetic model in steady state . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6 Time hierarchy in metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7 Parameters and variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 The kinetics and energetics of chemical reactions 18

2.1 Reaction rates—how fast? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.1 Reaction order and sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2 The equilibrium constant—how far? . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.1 The mass-action ratio and the distance from equilibrium . . . . . . . . . . 22
2.3 ∆G, Γ and Keq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4 Coupled reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5 Kinetic and energetic aspects of reaction rate . . . . . . . . . . . . . . . . . . . . . 29
3 Enzyme kinetics 30
3.1 The reversible Michaelis-Menten equation . . . . . . . . . . . . . . . . . . . . . . . 30
3.1.1 The Haldane relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2 Uncompetitive inhibition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3 Cooperativity and the reversible Hill equation . . . . . . . . . . . . . . . . . . . . 34
3.3.1 The reversible Hill equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.2 Modifier effects in the reversible Hill equation . . . . . . . . . . . . . . . . 36
4 Control Analysis 39
4.1 Quantifying metabolic control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.1.1 Local coefficients: Elasticity coefficients . . . . . . . . . . . . . . . . . . . . 40
4.1.2 Global coefficients: Response and control coefficients . . . . . . . . . . . . 43
4.2 Control Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.1 Summation properties of control coefficients . . . . . . . . . . . . . . . . . 46
4.2.2 Connectivity properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2
Structural Analysis, Kinetics, Control and Regulation of Cellular Systems
5 Obtaining elasticity coefficients from the logarithmic form of rate equations 50

5.1 Example: The reversible Michaelis-Menten equation . . . . . . . . . . . . . . . . . 51
5.2 Tutorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6 Metabolic Regulation: Supply-Demand Analysis 54

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2 Metabolic regulation, organisation and function . . . . . . . . . . . . . . . . . . . 54
6.3 Quantitative analysis of supply-demand systems . . . . . . . . . . . . . . . . . . . 55
6.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3
1 The Kinetic Model for a Network of Coupled
Reactions
When faced with the metabolic map we are struck by its seemingly bewildering complexity. Just
as we search for structural motifs to try to make sense of protein structure, so should we look for
underlying structure in metabolic networks. When wanting to study the dynamic behaviour of
metabolic networks, we are not so much interested in reaction chemistry and mechanism (the
main concern of metabolic studies thus far), but rather in the different ways in which reactions
are coupled. In the first section of this chapter we show that the metabolic network is built
up from a few basic linkage types and structures. This provides a good background against
which a kinetic model of metabolism can be developed. We are of course primarily interested in
metabolic structures in which reactions are catalysed by enzymes, but it should be emphasised
that the kinetic model we develop in this chapter is applicable to any system of coupled chemical
reactions. Although coupled reaction networks show many of the properties of electrical and
hydrodynamic networks, there is one property of reaction networks that makes them unique,
richer in behaviour, and therefore more interesting—this property is stoichiometry, the fixed
ratios in which molecules react with each other. The existence of stoichiometry complicates the
study of network behaviour, but this is more than compensated for by the new possibilities it
opens up for novel behaviour. Although stoichiometry will of course be incorporated into the
kinetic model right from the start, a systematic study of stoichiometry is beyond the scope of
this course.
1.1 The basic linkages and structures in metabolism

A fundamental feature of metabolic pathways is that any two functional steps, usually enzyme-
catalysed reactions or transport steps, are linked by metabolites common to both (the product
of one step becomes the substrate of the next). The two different types of linkages are shown in
Fig. 1.1:
An important difference between these two types of linkage is that a type 1 linking metabolite
is part of the material being metabolised and is passed on, whereas a type 2 linking metabolite
alternates between two or more forms of a chemical group (called a moiety ); it forms a permanent
part of the system. The concentration of type 1 linker is in theory free to assume any value
(however, there may be physical constraints on the concentrations of metabolic intermediates
due to limits in the solvent capacity of water). On the other hand, the total concentration of
the different forms of a type 2 linking metabolite is often constant. The condition is that the
formation and degradation of the moiety be slow in comparison to the interconversion of the
different forms. Such type 2 structures are called moiety-conserved cycles. For example, ATP,
ADP and AMP form an adenylate-conserving system, whereas pyridine nucleotide is conserved
in the NAD+ /NADH couple. The concept of moiety conservation is so important that Reich
and Sel’kov described energy metabolism as ‘open flow through networks of moiety-conserved
cycles’.
4
A.. X Y Z
B..
X Y
Figure 1.1: The two ways in which enzymic reactions are linked. A. The second enzyme takes
the linking metabolite, Y, on to form a new substance, Z. B. The second enzyme
reverses the action of the first enzyme as far as the linking metabolite is concerned.
For example, if the first enzyme oxidises X, the second reduces it, or if the first
enzyme phosphorylates X to form Y, the second dephosphorylates Y back to X. Of
course, the second reaction is not a simple reversal of the first, but a different reaction
catalysed by a different enzyme (e.g., the first enzyme may be a kinase that uses
ATP to phosphorylate X, in which case the second enzyme usually is a phosphatase
that hydrolyses the phosphoester bond). This is the well-known metabolic feature of
reactions that are coupled by the interconversion of different forms of cofactors, such
as NAD+ /NADH or ATP/ADP. These reactions are usually bimolecular reactions in
which some or other chemical group (such as a hydride or phosphoryl group in the
above examples) is transferred from a donor to an acceptor compound.
In metabolism there is a tendency to form chains consisting of only one type of linkage
(Fig. 1.2). Many of the familiar metabolic pathways have as backbone a chain of type 1 linkages.
On the other hand, in the electron transfer chain all linkages are of the second type (each carrier
existing in two forms, the total amount being conserved). Often a type 2 chain might start from
one of the steps of a type 1 chain and perhaps terminate in another type 1 chain or even in a later
step of the same type 1 chain.
The two fundamental types of linkage are combined to form four basic metabolic structures
(Fig. 1.3). In branched pathways depicted in Fig. 1.3B, a branchpoint occurs where a metabo-
lite is committed by different enzymes to more than one end-product. A central branchpoint
metabolite, such as acetyl-CoA or glutamine, can be the junction of a number of branches.
Fig. 1.3B shows a branch with diverging fluxes. Reversing the direction of fluxes changes the
A..
B..
C..
Figure 1.2: Chains of reactions consisting of: A. type 1 linkages, B. type 2 linkages, C. type 2
linkages starting from and ending in a type 1 chain.
5
A..
B..
C..
D.. 1
2
Figure 1.3: The four basic metabolic structures: A. Linear chain B. Branched chain C. Loop D.
Cycle.
system to one with converging branches. When two branches in a pathway reconverge a loop is
formed (Fig. 1.3C). Flux through the two limbs of the loop can be either parallel or opposed. The
simplest forms of loops are, first, reactions catalysed by two isozymes and, second, the substrate
loop (also called the futile loop), which consists of two enzymes catalysing opposite directions
of the same transformation (e.g. the kinase/phosphatase loops of glycolysis). Note that we do
not, as is common in the literature, call such a structure a substrate or futile cycle; the term
cycle is reserved for the distinct structural type depicted in Fig. 1.3D. Whereas linear, branched
and looped structures can be formed by monomolecular reactions, the numerous bimolecular
metabolic reactions can give rise to cycles (note that cycles consisting of monomolecular reactions
are identical to loops without input and output; they form closed systems with no metabolic
significance).
The main difference between substrate loops and cycles is that the input to and output from
loops are at branchpoints formed by metabolites common to more than two reactions, while in
cycles they are through bimolecular reactions (reactions 1 and 3 in Fig. 1.3D). The simplest type
of cycle is the 2-member cycle mentioned above under type 2 linkages.
1.2 The kinetic model

In general, therefore, a metabolic system consists of a network of coupled enzymic reactions and
membrane-associated transport steps. The structure of such networks is commonly displayed on
metabolic maps, where each reaction is described in terms of the participating enzyme, metabo-
lites and cofactors and the reaction stoichiometry. These chemical reactions and transport steps
6
can be thought of as the primary connections between metabolite pools through which the pools
affect each other by mass action. Metabolites and enzymes can also interact through regula-
tory loops, i.e., feedback and feedforward interactions; these can be thought of as secondary
connections for transmitting information through the network. Another important feature of
many metabolic systems is that they are divided into different compartments with the same
metabolite sometimes occurring in two or more compartments. For the purpose of analysing the
dynamic behaviour of such systems the compartmentalised pools of the same metabolite must
be regarded as separate metabolites even if they have the same chemical structure.
In large compartments and when diffusion processes are slow there may be a concentration
gradient of a specific metabolite, which therefore cannot be represented by a uniform pool
with a measurable size. However, as this complicates the analysis considerably and is, more
often than not, inapplicable, we make the simplifying assumption that metabolites are at all
times uniformly distributed through the space occupied by the compartment in which they
occur, i.e., no concentration gradients exist within the system, except across membranes (this
assumption has its counterpart in the chemical engineer’s concept of a ‘well-stirred reactor’).
Such metabolites can be regarded as thermodynamically defined in the sense that they can be
assigned a chemical potential.
v1 v4
v2 v5
S
v3 v6
Figure 1.4: A metabolite S that is produced and consumed by various reactions
Consider a specific metabolite pool, S with concentration s (Fig. 1.4)1 . The rate at which s
changes depends on the magnitudes of individual rates of the reactions that produce or consume
S. The rate at which s changes at any specific instance can be written as
ds
= v1 + v2 + v3 − v4 − v5 − v6 = (v1 + v2 + v3 ) − (v4 + v5 + v6 ) (1.1)
dt
where v1 is the net rate of reaction 1, v2 that of reaction 2, etc. These equations symbolise the
statement that the rate at which s changes is equal to the sum of the rates of the reactions for
which S is a product minus the sum of the rates of the reactions for which S is a substrate.
The net rate of each individual reaction can be expressed as the rate of change of the concen-
tration of any reactant or product of that reaction. Thus, for a simple reaction, A + B ⇋ C, the
rates of consumption of A and B are da/dt and db/dt, while the rate of production of C is dc/dt.
For this reaction one can write
da db dc
v=− =− = , (1.2)
dt dt dt
Because one molecule of A must react with one molecule of B to form one molecule of C, it is
immaterial which of these derivatives is used as a measure of the v, as long as one keeps the
signs correct.
1
Names of chemical species are denoted by a capital letters E, X, S, . . . and their concentrations by lowercase italic
letters e, x, s, . . . rather than by the usual square brackets [E], [S] or [X].
7
However, for a reaction such as 2A + B ⇋ 3C the situation is more complicated. Here, for
example, a changes twice as fast as b, while c changes three times as fast as b. The three time
derivatives are related by
1 da db 1 dc
− =− = . (1.3)
2 dt dt 3 dt
It should be clear that, in order to be unambiguous about what we mean by the ‘rate’ of a chemical
reaction, we should state which time derivative we have chosen to represent the rate (e.g. −da/dt
or dc/dt), and we should supply the balanced chemical equation for the reaction. Note that if the
stoichiometries are not one, the value of the rate depends on which time derivative is chosen.
This procedure is sufficient if only one isolated reaction is studied, but as soon as reactions
are coupled by common intermediates that can participate in these reactions with different
stoichiometries, we need a more consistent definition of reaction rate. This can be given in
terms of the degree of advancement, ξ (the Greek letter xi), of the reaction, also called the extent
of reaction. Consider a simple reaction A ⇋ B. If an infinitesimal amount dξ moles of A is
converted into B, we can write
change in the amount of A = −dξ

change in the amount of B = dξ.
The more complex reaction 2A + B ⇋ 3C serves to illustrate that, in general, any reaction can be
treated in the same way. We could write the reaction in an alternative way:
0 = −2A − B + 3C.
The numbers −2, −1 and +3 are called the stoichiometric coefficients, the negative signs going
with the reactants and the positive ones with the products. We shall use the symbol c for a
stoichiometric coefficient, so that in general we can write for any equation:
0 = cA A + cB B + cC C + cD D + . . .
When the reaction proceeds by an amount dξ, the amounts of the reactants and products change
as follows:
change in the amount of A = −2dξ, or cA dξ in general
change in the amount of B = −dξ, or cB dξ in general
change in the amount of C = +3dξ, or cC dξ in general.
Now that we understand what ξ means, we can define the rate, v, of the reaction as
dξ
v= , (1.4)
dt
the rate of change of advancement of the reaction. In the above we saw that, say, a changed
by an amount cA dξ in the time increment dt; its rate of change da/dt = cA dξ/dt or, keeping
in mind the definition of reaction rate in eq. 1.4, da/dt = cA v. Therefore, reaction rate can be
expressed as the rate of change of a reagent scaled by the inverse of its stoichiometric coefficient
(i.e. v = (1/cA )da/dt). For our example v can be seen to be any of the three scaled time derivatives
in eq. 1.3.
In Fig. 1.4 we assumed that the stoichiometries with which S participates in the different
reactions are all equal to one. Of course this need not be so and with the above general definition
of reaction rate we have a way of accounting for differing stoichiometries in such equations.
8
Let us explore this by considering a simple system that consists of two reactions, one of which
produces compound S 1 while the other consumes it.
v1
X0 ⇋ 2S1 (1.5)
v2
S1 ⇋ X 2 (1.6)
The individual reaction rates are

−dx0 1 ds1 −ds1 dx2
v1 = = and v2 = = (1.7)
dt 2 dt dt dt
It follows that for reactions 1.5 and 1.6 individually
ds1 ds1
= 2v1 and = −v2 (1.8)
dt dt
If the two enzymes occur together, the rate at which s 1 changes at any time will be the sum of
the individual rates at which the enzymes produce or consume S 1 .
ds
1
= 2v1 − v2 (1.9)
dt
In general, therefore, the rates which affect the concentration of a metabolite will, in differential
equations such as eq. 1.1, be weighted by the stoichiometric coefficients; c > 0 if the metabolite
is a product of the reaction, c < 0 is it is a substrate, and, of course, c = 0 if the metabolite does
not participate in the reaction.
With the above background we can now develop a general way of describing the rates at which
the metabolite concentrations in a system change. Consider a system that consists of n steps that
interconvert m metabolites. We use j as a counter for the 1 to n steps and i as a counter for the
1 to m metabolites. Therefore S i will be any metabolite (with concentration si ) and vj the rate of
any reaction or transport step. The stoichiometric coefficient c i j describes the stoichiometry by
which Si participates in step j. For any metabolite S i we can now write
dsi
= ci1 v1 + ci2 v2 + ci3 v3 + · · · + ci(n−1) vn−1 + cin vn (1.10)
dt
In shorthand summation notation this differential equation becomes
n
dsi
= ci j vj (1.11)
dt
j=1
Such an equation can be written for each metabolite so that the dynamic behaviour of the whole
system is described by the following system of m differential equations which may be regarded
as is the general kinetic model for a system of coupled reactions ):
n
dsi
= ci j vj for i = 1, . . . , m (1.12)
dt
j=1
In general, Si can denote enzymic intermediates, metabolite pools or groups of metabolite pools;
vj denotes the rate of functional steps, which can be elementary, enzyme, translocator, non-
catalysed, or groups of reactions. The stoichiometric coefficients, c i j describe the mass-action
9
S1 S3
c11 c31
1
c21 c41
S2 + S4
X0
Figure 1.5: A schematic representation of the enzymatic reaction c 11 S1 + c21 S2 = c31 S3 + c41 S4
which is imbedded in a metabolic network. Step numbers are boxed and metabolites
represented by either a subscripted X (for external metabolites) or a subscripted S (for
internal variable metabolites). Stoichiometric coefficients are associated with the line
connecting a metabolite to an enzyme box. Arrowheads are in accordance with the
direction defined in the rate equation, and do not imply irreversibility. Effectors such
as Xe are linked to the enzyme box by arrows and the type of interaction indicated by
+ (activation) or − (inhibition).
linkages in the reaction network, i.e. its topology. In general, the rates v j are non-linear functions
of the Si and, except for very simple networks, the system has no analytical solution. However,
starting from specified initial conditions, the time-course behaviour of S i can be simulated by
numerical integration of the differential equations, usually by digital computer.
Let us again consider the simple system described by reactions 1.5 and 1.6. Fig. 1.5 explains the
schematic method that will be used throughout to depict the way in which reactions are coupled.
One of its useful features is that it allows stoichiometric coefficients to be shown unambiguously.
2
X0 1 S1 2 X2
Figure 1.6: A schematic representation of the reactions in eq. 1.5 and 1.6
Using this schematic method our simple model is depicted in Fig. 1.6. The full kinetic model
for this system is (according to eq. 1.12)
dx0 /dt = (−1)v1 + (0)v2 = −v1

ds1 /dt = (2)v1 + (−1)v2 = 2v1 − v2 (1.13)
dx2 /dt = (0)v1 + (1)v2 = v2
Given explicit rate equations for reactions 1 and 2, this is a complete description of how x 0 , s1
and x2 change with time.
1.3 A hierarchy of kinetic models

Up to now we have been rather vague about the identity of the metabolites, about the reactions
that produce or consume them and about the equations that describe the rates of reaction v. The
kinetic model is extremely versatile in that it can be used to describe many different levels of
complexity; these levels differ in terms of how reactions are aggregated; in turn this aggregation
is inextricably bound up with the time hierarchy of chemical interactions. At one extreme,
10
chemical interactions can be described on the quantum level where the players are elementary
particles. Through the atomic level one progresses to the individual molecular level to the
level of populations of molecules. Here the kinetic model becomes a valid way of describing
dynamics. The lowest rung on this part of the hierarchy is where reaction are described in terms
of elementary reactions where two molecules associate to form a complex (e.g. protein-ligand
association), a complex dissociates into two molecules, or a molecule or complex isomerises.
The rates of these reactions are described in terms of first or second-order rate constants in
combination with reactant concentrations. This can be called the mass-action level where
molecular events happen in the micro to millisecond time-scale. Biochemists usually aggregate
the elementary steps associated with an enzyme catalytic process into a single reaction, the rate
of which is described by a rate-law, the most familiar being the Michaelis-Menten rate equation.
At this level of the time scale events happen within seconds to minutes—this we regard as the
intermediary metabolic time scale. The processes of transcription, translation, protein synthesis
and degradation are slower (minutes to hours) and can be thought of as the genetic time-scale.
We now describe three important approaches that have been used to describe the dynamic
behaviour of metabolic systems; they all depend on the way that enzymic reactions and their
rates are represented mathematically.
1.3.1 The mass-action form of the kinetic model

In the mass-action approach each enzyme reaction is written in terms of elementary mechanistic
steps. Reaction rates of each step are then described in terms of mass-action kinetics. The
si in the kinetic model then signify all metabolites and intermediary enzyme complexes (with
a differential equation for each of them); the rates vj of the elementary steps in each enzyme
mechanism are then usually expressed by simple first or second order rate equations.
For the system in Fig. 1.6 the mass-action approach will be based on the scheme in Fig. 1.7,
which assumes a simple reversible mechanism for each enzyme reaction
E 1.X 0 1b
1a E 1 .2S1
E 2. S 1 2b
X0 E1 1c
2 2a E 2 .X 2
S1
E2 2c
X2
Figure 1.7: The mass-action representation of the reaction system in Fig. 1.6.
The kinetic model is expressed in the following set of differential equations (here we use
11
square brackets to depict concentrations):
d[X0 ]/dt = −v1a (1.14)

d[E1 X0 ]/dt = v1a − v1b (1.15)
d[E1 2S1 ]/dt = v1b − v1c (1.16)
d[S1 ]/dt = 2v1c − v2a (1.17)
d[E2 S1 ]/dt = v2a − v2b (1.18)
d[E2 X2 ]/dt = v2b − v2c (1.19)
d[X2 ]/dt = v2c (1.20)
d[E1 ]/dt = v1c − v1a (1.21)
d[E2 ]/dt = v2c − v2a (1.22)
Each individual net rate v is the difference between the forward and reverse rates of that reaction:
v1a = k1a [X0 ][E1 ] − k−1a [E1 X0 ] (1.23)

v1b = k1b [E1 X0 ] − k−1b [E1 2S1 ] (1.24)
2
v1c = k1c [E1 2S1 ] − k−1c [E1 ][S1 ] (1.25)
v2a = k2a [S1 ][E2 ] − k−2a [E2 S1 ] (1.26)
v2b = k2b [E2 S1 ] − k−2b [E2 X2 ] (1.27)
v2c = k2c [E2 X2 ] − k−2c [E2 ][X2 ] (1.28)
The mass-action approach formed the basis of the pioneering simulation studies of Garfinkel,
Chance, Hess and others. These studies furnished the conceptually important general result
that the time behaviour of complex metabolic systems can essentially be explained by the basic
kinetic model; correct predictions and refutation of particular hypotheses were possible. The
mass-action approach thus played an important role in establishing the validity of the basic
kinetic model. The model also showed that observed metabolic phenomena can be explained in
terms of physico-chemical principles.
However, the mass-action approach often meets with practical, conceptual and numerical
difficulties. Not only is it cumbersome, but there is seldom enough information on the values of
rate constants to make the approach feasible; even if the rate constants were known, relatively
simple metabolic pathways would lead to huge differential equation systems. Other problems
arise when rate constants differ widely in magnitude; this causes ‘stiffness’ in the differential
equations and results in difficulties of numerical integration (however, this problem has now
largely been overcome by modern numerical integration routines).
1.3.2 The rate law form of the kinetic model

In the rate-law approach, the rate of each enzymic reaction is described by a aggregated rate law
(a net-flux equation) obtained from steady-state kinetic studies. This most useful approximation
follows from the assumption that enzyme concentrations are in general much lower than the
concentrations of intermediary metabolites. However, this assumption only has to be made when
the concentrations of substrates and products that enter the rate equations are total concentrations
(i.e. the sum of free plus bound metabolite). If only free metabolite concentrations enter the rate
equation then it is valid even at high enzyme concentration. Well-known examples of such rate
laws are the Michaelis-Menten, Hill or Monod-Wyman-Changeux equations (or, more generally,
12
the type of equations described by Cleland). Each rate law is a non-linear function of substrate,
product and effector concentration (but linear in enzyme concentration), in conjunction with
either fundamental rate constants (King-Altman form) or phenomenological constants such as
kcat , Keq , Km and Ki (Cleland form). An important point to note is that any allosteric feedback or
feedforward effects are accounted for in these rate equations. In practice, the rate-law approach
gives results that agree satisfactorily with those obtained by solving the mass-action form of the
model and simplifies analysis considerably. If we assume that rate laws are known for reactions
1 and 2 in the system in Fig. 1.6, then the differential equations that form the kinetic model are
those in eq. 1.13, while the rate laws themselves could be something like:
s21

V1
x0 −
K1(X0 ) Keq(1)
v1 = x0 s (1.29)
1+ + 1
K1(X0 ) K1(S1 )

V2 x2
s1 −
K2(S1 ) Keq(2)
v2 = s x2 (1.30)
1+ 1 +
K2(S1 ) K2(X2 )
We shall mostly use the rate-law approach in our modelling exercises.
1.3.3 The power-law form of the kinetic model

In this form of the kinetic model, pioneered and used extensively by Savageau, all the reactions
that produce a metabolite are aggregated as are those that consume the metabolite. The rates of
each of these aggregated reactions are then described by a power-law equation. Any differential
equation in the kinetic model therefore contains only two rate terms which simplifies the math-
ematical and numerical analysis considerably. For the system in Fig. 1.6, the detailed kinetic
model in power-law notation will be (as this is just illustrative, irreversibility of both reactions
is assumed):
g1X0
dx0 /dt = −α1 x0 (1.31)
g1X g2S
ds1 /dt = α1 x0 0 − α2 s1 1 (1.32)
g2S
dx2 /dt = α2 s1 1 (1.33)
The α-terms are rate constants and the g-terms apparent kinetic orders. Although use of this
approach has its advantages, these aggregated reactions are often difficult to interpret in physical
terms. We shall not use this approach at all.
In summary it can be seen that the essential ingredients of the description of a metabolic
network are the stoichiometric reactions and an algebraic expression for the rate of each reaction.
Any constraint on the concentrations of a group of metabolites is described by a conservation
equation (this will be discussed fully in a later section).
1.4 Metabolic states

With the background of the previous sections we are now well equipped to understand the three
possible states a metabolic system can be in:
13
1.4.1 Equilibrium
All metabolic pools have time-invariant values and the individual net reaction rates are zero (each
reaction is itself in equilibrium). This situation, where there is detailed balance in each reaction,
is, biochemically speaking, of little interest, as only isolated systems can achieve equilibrium.
Systems in equilibrium are dead.
1.4.2 Steady state

Functioning metabolic systems are open, with several or all reactions being either ‘pushed’ or
‘pulled’ from the outside. As these pushing and pulling forces are not in equilibrium with
each other, they ensure that a metabolic system rather approaches a steady state (also called a
stationary state ). Here the metabolite pools are also time-invariant (their net rates of change
are zero), but the individual reaction rates are not zero, so that there are constant fluxes of
matter through the system. In a non-growing system an additional condition for steady state
is that of conservation of mass; the net import of mass into the system per unit time is equal
to the net export per unit time. In a growing system in steady state the metabolite pools will
not be constant because of a constant ‘flux to expansion’; after correcting for this the normal
time-invariant steady state will be seen to obtain.
Theoretically, a steady state is never actually reached, only approached asymptotically. How-
ever, in practice one can regard a system as being ‘in’ the steady state. When for a certain set of
conditions a steady state is ‘unique’, it will be approached no matter what the initial concentra-
tions of the metabolite pools are. This should be distinguished from situations where more than
one steady state is possible for a given set of parameters (multistationarity or multiple steady
states); here, which steady state is approached depends on the initial conditions, i.e., on the
history of the system. Metabolite pools may also show constant periodic (oscillatory) behaviour
around a steady-state point; the oscillating motion of metabolite concentrations is then said to
be in a ‘limit cycle’.
The concept of stability is of great importance in connection with the steady state. A metabolic
system is continually subjected to fluctuations in its variables (even the parameters, which we
regard as constant, fluctuate). A steady state cannot exist or be maintained if it is not stable with
respect to these fluctuations and perturbations. In this discussion we assume the steady state to
be asymptotically stable. This means that if we regard the set of steady-state concentrations as
describing a point in a multidimensional space, then there exists a neighbourhood around that
point from which the steady-state point is always approached in time, i.e. if the initial concen-
trations correspond to any point in that neighbourhood, the system eventually reaches the same
steady state. Besides this dynamic stability we also assume structural stability, which means
that after a perturbation in one of the parameters the system relaxes to a closely-neighbouring
steady state.
1.4.3 Transient state

A metabolic system could be moving towards a steady-state from some initial condition, from
one steady state to another after a perturbation in some parameter of the system, or back to the
old steady state after a fluctuation in one of the internal metabolites; it is then said to be in the
transient state.
Let us now review our kinetic model in terms of the steady state.
14
1.5 The kinetic model in steady state

Biochemical systems have two important properties: the functional units of the system (such as
enzymes and translocators) display non-linear kinetics and the systems are open, that is, they
interact with their environment through the exchange of matter and energy. We should know
how to define the system so that it is open.
Open systems are not always easy to set up experimentally. One possibility is to inject pathway
substrate(s) at a constant rate and remove pathway product(s) by some method (precipitation,
adsorption). This means we supply the system with constant fluxes into and out of the system
(source fluxes and sink fluxes respectively). Another possibility is to ensure that the concentra-
tions of source substrates are so high or the extra-system volume so large that, for all practical
purposes, they remain constant within the time scale of the experiment; this is the situation
most frequently encountered. This means providing the system with a practically inexhaustible
supply of pathway substrate and a practically unsaturable pool of waste products. Sometimes a
substrate or product of a pathway is a solid such as glycogen. As a solid has constant chemical
potential, its ‘concentration’ is automatically constant.
Fortunately, for the purposes of modelling such experimental detail can be ignored. We just
define the concentrations of all pathway substrates, products and external effectors as constant
(we thus regard these ‘external’ metabolites as ‘fixed’). All those metabolites in the system that
are either only produced or only consumed must be regarded as external and fixed, so that a
steady state solution is at least possible, in other words, that the system is thermodynamically
open. The other ‘internal’ metabolites can be regarded as variable metabolites.
In steady state the total rate of production of each variable metabolite is balanced by the total
rate of consumption, so that the net rate is zero and the metabolite concentrations are time-
invariant (i.e. dsi /dt = 0). If our general kinetic model is in steady state, we call the resulting
system of m equations the set of balance equations, one for each variable metabolite:
n

ci j J j = 0 for i = 1, . . . , m (1.34)
j=1
Note that we use the symbol J to remind us that the reaction rates are now steady-state fluxes.
The term ‘flux’ is often used rather loosely as a general synonym for ‘reaction rate’. We prefer
to be more specific and define a flux as a steady-state reaction rate; in terms of this definition,
‘flux’ has no meaning in the transient state.
Here we note that the numerical value of a flux could seem to depend on where it is measured.
Consider the algebraic nature of the balance equations. Each balance equation is a linear
function of reaction rates, whereas the individual rate equations are nonlinear functions of
metabolite concentrations and the system parameters. It follows that the balance equations are
implicit nonlinear functions of constant quantities (parameters) and variable quantities (metabo-
lite concentrations) and thus form a set of nonlinear equations for which no analytical solution
is possible (except in the simplest cases). Therefore, numerical solution by computer is the only
way to obtain the steady state metabolite concentrations and rates. What is important is that the
steady state is fully determined by the parameters.
For our simple model in Fig. 1.6 there is only one balance equation, i.e. for S 1 , since X0 and
X2 respectively serve as external substrate and product and must therefore be regarded as fixed
constants in order to allow a steady state to be reached. The balance equation for S 1 is
2J1 − J2 = 0 (1.35)
15
The balance equations allow us to determine the number of steady-state fluxes in the system
and how they are related to each other. The above balance equation shows that, in steady-state,
2J1 = J2 . Although there is really only one flux of matter, its numerical value could be measured
as either that of J2 or that of J1 ; these values stand in a fixed relationship in that they differ by
a factor of 2. Although x0 and x2 are clamped, one can measure flux in terms of the rate of
disappearance of a radio-active tracer x∗0 or of appearance of a radio-active tracer of x∗2 :
dx∗0 /dt = J1 = 0.5J2 (1.36)
or dx∗2 /dt = 2J1 = J2 (1.37)
Which flux-value is used in practice depends on which rate of change is measured. However, in
the section on control analysis we shall see that we usually are interested in relative changes in
flux-values, which will be the same in the above example wherever the flux is measured.
1.6 Time hierarchy in metabolism

We have already touched upon the grouping of metabolic interactions on the basis of time-scale.
Very rapid reactions, such as protein-ligand interactions, usually reach near-equilibrium within
milliseconds and can, when observed from longer time-scales, be regarded as frozen in this
state. The reactions normally depicted on the metabolic map occupy an intermediate position
in the time-hierarchy of metabolic events; this ‘metabolic’ time-scale of seconds to minutes is
applicable in most studies on the behaviour and control of intermediary metabolism where
changes in enzyme rates and metabolite concentrations are observed. Slower reactions such as
synthesis and degradation of enzymes may take much longer (hours) and cause a slow drift
of the steady-state metabolite pools. In terms of the metabolic time-scale one sees a system of
slowly drifting equilibria and slowly drifting steady states.
1.7 Parameters and variables

Within a chosen time-scale, certain quantities can thus be regarded as essentially constant, unless
they are manipulated by the investigator. These are called the system parameters, which we
shall also call the controllers. Obvious examples are environmental factors such as temperature
and, if the system is well-buffered, pH. From the viewpoint of the metabolic time-scale other
constant quantities fall into three groups.
• The concentrations of those chemical species that are synthesised and degraded very
slowly in relation to the metabolic reactions: enzymes, translocators and certain moieties
in cofactors.
• The concentrations of the external metabolites of the system: initial substrates, final prod-
ucts and other external effectors such as inhibitors, activators and hormones. These external
metabolites are considered to be buffered by the environment. They must be constant to
keep the system open and allow a steady state to be approached.
• Equilibrium constants, Keq , and enzymic constants such as kcat , KM and Ki .
The system parameters, as well as the stoichiometry and rate function of each reaction, de-
termine the steady-state values of the intermediate metabolite pools and fluxes. These are
the dependent variables of the metabolic system; they are controlled by the controllers. The
variables therefore are:
16
• Fluxes
• Metabolite concentrations
• Functions of metabolite concentrations, e.g. Gibbs-energy changes, chemical and mem-

brane potentials, mole fractions, ratios of concentrations.
17
2 The kinetics and energetics of chemical
reactions
As biochemists and molecular biologists, we want to understand what happens in a living
cell, because (i) we are scientists, and understanding and explaining is what science is all
about, and (ii) we have this desire to manipulate things, which is not always a good thing, but
which is nevertheless there. This understanding we seek is not any old understanding, but
understanding at a deep level, the level of molecular reaction and interaction, i.e., a physico-
chemical understanding. At the fundamental level of chemical mechanism, the three basic
building blocks for molecular events that occurs in living processes are:
1. an association between two molecules to form a non-covalently bound complex,
A + B −→ A · B (2.1)
2. a dissociation of a complex into two molecules,
A · B −→ A + B (2.2)
3. an interconversion where one molecule is chemically transformed into another (an iso-
merisation ).
A −→ B (2.3)
This reaction could also depict the excitation of A by, for example, light.
These elementary reactions are defined to be irreversible (denoted by −→). They never exist
in isolation, but always in combination with each other. So, what we usually describe as a
chemical reaction can always be broken down into a mechanism that consists of combinations of
these three elementary processes. For example, the probable mechanism of the chemical reaction
A + B ⇋ C would be
A+B⇋A·B⇋C (2.4)
where C is A—B, the condensation product of A and B. Each half of the double arrow (⇋)
denotes one of the elementary reactions. In principle, all chemical reactions are reversible and
consist of a forward reaction and a reverse reaction.
2.1 Reaction rates—how fast?

A fundamental question of reaction chemistry is “how fast does an elementary process occur
and what determines this rate?” All else follows from this question. So, let us ask it. Given the
incredible diversity of the molecular world, the answer is astoundingly simple: the rate v of an
elementary reaction is proportional (∝) to the concentration of molecules involved, irrespective
18
of the identity of the molecules. This fundamental law of chemistry is of course the famous law
of mass action. For example, for the elementary reaction A + B −→ A · B this boils down to
v∝a and v∝b (2.5)
where v is the rate of reaction, and a and b the concentrations of A and B. The rate is therefore
proportional to the product of a and b
v ∝ ab (2.6)
(you may ask why the product of concentrations and not the sum: think of it as a matter of chance
encounters between A and B. The probability of a collision between an A and a B depends on
how much A and B there is, not on how much A irrespective of how much B, or how much B
irrespective of how much A. When probabilities depend on each other like this, the combined
probability is always the product of the individual probabilities).
In general, reaction rate is defined as minus the rate of change of reactant concentration, or
plus the rate of change of product concentration. Translating this to the example above, the
reaction rate tells us how fast the concentration of reactant A or reactant B decreases with time,
or how fast the concentration of the product A · B increases with time. The two rates will of
course be the same, because for every molecule A and molecule B that disappear, one molecule
of A · B appears. Mathematically the rate equals
da db d(a · b)
v=− =− = (2.7)
dt dt dt
The reaction rate v thus has units of concentration·time −1 .
Remember that since we are considering an elementary reaction, the concentration of product
A·B does not affect the rate (obviously, if we are considering the reverse reaction, the dissociation
of A·B into A and B, its rate will be proportional to the concentration of A·B).
The above relationship between rate v and concentrations a and b can be transformed into a
rate equation by inserting a constant, called the rate constant :
v = kab (2.8)
The value of the rate constant is unique for every reaction. It is not, however, a fundamental
constant, because it varies with, e.g., temperature.
Imagine that in the above example we measure the reaction rate at different concentrations of
a while keeping b constant. If we plot v against a then, because of the proportionality, we obtain
a straight line through the origin. The slope of this line (kb, where b is fixed) is of fundamental
importance as it tells us something about the reactivity of A and B. The steeper the slope at some
fixed b, the more vigorous the reaction between A and B. The same argument holds when b is
varied at fixed a.
The linear relationship between v and either a or b is called a first-order mathematical rela-
tionship (the order of any polynomial function is the highest exponent to which the variable is
raised). This term has been adopted in the language of reaction kinetics: the reaction is said to
be first-order in a and first-order in b. In general, the order of a reaction with respect to some
chemical species is the exponent to which the concentration of that species is raised to in the rate
equation. The overall order of a reaction is the sum of all the exponents of the concentration
terms in its rate equation (in our example the overall order is therefore two, and the reaction is
called a second-order reaction: if both a and b are varied in the same proportion the rate curve
would be parabolic). It should be clear that for elementary reactions the overall order can only
be one (for dissociation and isomerisation reactions) or two (for association reactions).
19
2.1.1 Reaction order and sensitivity

In the light of the above, one may expect reaction order to be an integer, but in general, for
ordinary chemical reactions, there is no such requirement, the reaction orders being complex
functions of elementary orders. Reaction order is therefore an experimentally determined entity.
Consider a reaction between two species A and B where the rate equation is of the form
v = kap bq (2.9)
and where p and q are the unknown orders. Taking logarithms on both sides we obtain
ln v = ln k + p ln a + q ln b (2.10)
I have used natural logarithms (base e = 2.718 . . .), but I could also have used common logarithms
with base 10. The logarithmic form of the equation immediately suggests a way of determining
p and q experimentally. To obtain p we measure the rate v at fixed b for different concentrations
of a. When ln v is plotted against ln a a straight line is obtained with slope p and an intercept on
the ln v-axis of (ln k + q ln b). The inverse experiment where b is varied at fixed a yields q.
In this example we obtained straight lines when plotting ln v against ln(concentration). This
does not have to be so. In fact, for us as biologists, who virtually always study enzyme-catalysed
reactions, the plot of ln v against ln s, where s is substrate concentration, always yields a curve.
This means that there is no fixed reaction order (no constant slope), but that the order (slope)
varies with substrate concentration. Nevertheless, it is still possible to calculate the order at any
given concentration as the value of the slope at that concentration. Now, if you recall your first-
year calculus, you will remember that the slope of a curve at a given point is the first derivative
of the function that describes the slope evaluated at that point. This leads to the most general
definition of reaction order with respect to species a as
d ln v
(2.11)
d ln a
evaluated at a given a. (Because v is a function
of both a and b, the reaction order should, strictly
∂ ln v
speaking, be a partial derivative ∂ ln a . Remember that the subscript outside the brackets lists
b
the variables that are kept constant—here b—while a is varied). Don’t let the symbolism put you
off. It is just a short-hand notation for the slope of the curve obtained when plotting ln v against
ln a.
The concept of reaction order is extremely important and we will return to it later in Sec-
tion 4.1.1, when discussing metabolic control in a system of coupled reactions.
You may feel uncomfortable with this definition of reaction order as the slope of the curve
of v versus a in log-log space; in fact, at a deeper level, you may feel uncomfortable with the
introduction of the concept of logarithms into the discussion (in the days before calculators
logarithms were introduced as a mathematical trick for calculation; you know, adding the logs
of two quantities is the same as multiplying those quantities; this easily leads to the mistaken
view that as we now use calculators we can safely ignore logs). The one answer that I’ve heard
as a counter to this discomfort is: the definition of reaction order in terms of logarithms falls out
of the mathematical derivation; just accept it and quit worrying. I distrust this type of reasoning.
There must be a deeper and clearer explanation for why logarithms appear at all. We will cover
this topic extensively when we examine rate characteristics of enzyme-catalysed reactions. You
will see that we shall also draw those rate characteristics as logarithmic graphs.
20
To conclude this section, let us focus more closely on the distinction between the terms “molec-
ularity” of a reaction and “reaction order”. The molecularity of a reaction has to with the number
of molecules that react, which is given by the so-called stoichiometric coefficients of the reac-
tants in the balanced reaction. In terms of the three elementary reactions, the isomerisation and
dissociation reactions are unimolecular, while the association reaction is bimolecular. For these
elementary reactions we have seen that the molecularity is equal to the overall reaction order.
For any real chemical reaction this does not have to be so: reaction order is an experimentally
determined entity and can in general not be obtained from the stoichiometric coefficients.
We have seen that the answer to the question “How fast does a reaction go?” is embodied
in an experimentally characterised rate equation. The main aim of reaction kinetics is to obtain
such a rate equation for any reaction. Further on in this course we shall specifically discuss the
special rate equations that describe enzyme-catalysed reactions.
2.2 The equilibrium constant—how far?

The second fundamental question to be asked about any reaction is “How far does a reaction
go?”. This question forms the starting point for what is known as chemical thermodynamics or,
better, chemical energetics. I have already stated that all reactions are in principle reversible.
How do we reconcile this statement with what we discussed above? The statement implies that
every reaction can be thought of as a combination of a forward reaction and a reverse reaction,
each described by an irreversible rate equation. For example, the reversible reaction A + B ⇋ C
is a combination of the forward reaction
A+B→C with rate equation vf = kf ab (2.12)
and the reverse reaction
C→A+B with rate equation vr = kr c (2.13)
The net rate of reaction is the difference between the forward and reverse rates
v = vf − vr = kf ab − kr c (2.14)
If one starts off with everything in the form of A and B, then the reaction can only go forward.
With everything in the form of C the reaction can only go backward. It follows that there must
always be some point in between these two extremes when the concentrations of reactants and
products are such that the forward and reverse reaction rates are equal so that the net rate is
zero. This point is called the dynamic equilibrium point (“dynamic” because the forward and
reverse reactions are still happening even though the concentrations of reactants and products
are constant). This also implies that, when a reaction proceeds in isolation, the net direction of
reaction is always towards the equilibrium point. No reaction can spontaneously move away
from equilibrium
At equilibrium:
kf (a)eq (b)eq − kr (c)eq = 0 (2.15)
so that
kf (c)eq
= = Keq (2.16)
kr (a)eq (b)eq
21
The subscripts eq refer to equilibrium concentrations. The concentrations of A, B and C in Eqs. 2.15
and 2.16 do not refer to any arbitrary condition, but only to the case where the reaction is in
equilibrium.
The ratio of forward and reverse rate constants is called the equilibrium constant, K eq , and its
value is arguably the most important piece of information that one can have about a reaction.
Although the ratio of forward and reverse rate constants is an important way of describing the
equilibrium constant, it is the function in terms of equilibrium reactant and product concentra-
tions which is the most useful in practice.
The first thing to notice is that the equilibrium constant for the reaction A + B → C has units,
namely that of (concentration)−1 . The equilibrium constant for the same reaction written in the
opposite direction C → A + B is of course the inverse of that for the original direction, and has
units of concentration. Here it is important to note that the so-called thermodynamic equilibrium
constant (see Section 2.3) looks the same as the one above but is always dimensionless because
each concentration term is divided by a standard concentration of 1 M.
To end this section we expand the expression for the equilibrium constant to that of the general
reaction
mA + nB ⇋ pC + qD (2.17)
where m, n, p, and q are the stoichiometric coefficients. From the rate equations for the forward
and reverse reactions
vf = kf am bn and vr = kr cp dq (2.18)
and using the equilibrium condition we obtain
(c)eq p (d)eq q
Keq = (2.19)
(a)eq m (b)eq n
Note that the exponents in the equilibrium expression are stoichiometric coefficients that describe
the molecularity of the reaction. The above derivation using forward and reverse rate equations
may have given the impression that they should be reaction orders, but we have already seen that
molecularity does not necessarily equal reaction order. The form of the equilibrium expression
should actually not be derived from rate equations as above, but from basic thermodynamic
considerations where kinetic parameters such as reaction orders have no role to play. We will
focus in more detail on the relationship between thermodynamic quantities (Gibbs energy) and
the equilibrium constant in Section 2.3.
2.2.1 The mass-action ratio and the distance from equilibrium

With this background we can now refine our original question of “How far?” to “For a given
set of concentrations of reactants and products, how far is the reaction from equilibrium and on
which side?” For anyone studying living systems this question is of great importance because
the hallmark of living systems is that they are out of equilibrium—being in equilibrium means
being dead. The classical view of metabolic regulation sets great store by the classification
of reactions into near-equilibrium and far-from-equilibrium (also referred to non-equilibrium)
reactions. Where to draw the line has been the subject of much (and rather futile) discussion.
Let us rather derive an unambiguous quantitative measure for distance from equilibrium which
simultaneously given the direction in which the reaction proceeds.
The relationship that we seek must have something to do with the relationship between the
forward and reverse rates: the greater the difference between them, the further away from
22
equilibrium. Although this algebraic difference between forward and reverse rate (i.e., the net
rate), may seem to be a quantity that measures the distance from equilibrium, there is a much
better measure, namely the ratio of forward and reverse rates (which is dimensionless):
vr kr c c

= = Keq (2.20)
vf kf ab ab
Note that the concentrations a, b, and c can have any value. In the special case where they are
equilibrium concentrations then the fraction c/ab is the equilibrium constant and the ratio vf /vr
is equal to one (which of course it must be as at equilibrium the two are equal).
The quantity c/ab is so important that it has been given a special name, the mass-action ratio,
usually symbolised by Γ (capital Greek gamma). Note that Γ is always exactly the same algebraic
function of concentrations that one finds in the expression for the equilibrium constant. In fact,
the equilibrium constant is just a special value of the general mass-action ratio, namely that
when the concentrations are equilibrium concentrations.
2.3 The relationship between Gibbs energy, the mass action ratio
and the equilibrium constant
For any process (and hence, also for chemical reactions) the change in Gibbs free energy (∆G)
can be viewed as the driving force for that process and will determine whether it will occur
spontaneously:
• ∆G < 0 — the process proceeds from what we regard as the initial state to the final state
(for a chemical reaction from reactants to products).
• ∆G = 0 — the process is at equilibrium.
• ∆G > 0 — the process cannot proceed from what we regard as the initial state to the final
state, but it rather occurs in the reverse direction (for a chemical reaction from products to
reactants).
The above can be summarised in the following important statement:
The change in Gibbs energy of a process at constant pressure and temperature is an

indication of the tendency of that process to occur spontaneously.
The quantity ∆G leads to a new way for interpreting the thermodynamic driving force for a
process—the more negative ∆G, the greater is the driving force for that process to occur.
Consider again the general balanced reaction in Eq. 2.17. The change in Gibbs energy for a
small change in the extent of the reaction (also called the affinity of the reaction) is given by:
(c)p (d)q
∆G = ∆G◦ + RT ln (2.21)
(a)m (b)n
The term ∆G◦ refers to all reagents and products being in their standard state (a concentration
of 1 M for dissolved substances; a pressure of 1 atm for gases). The Gibbs energy change for
a reaction under any given conditions is thus given by a term for all reagents and products in
their standard state, and a term that reflects the activities (concentrations) in the system under
the specific conditions considered.
23
Remember that when the reaction is in equilibrium,
∆G = 0. (2.22)
Hence, it follows that

◦
(c)eq p (d)eq q
∆G = −RT ln (2.23)
(a)eq m (b)eq n
As above, the eq subscript indicates that we consider concentrations in the system at equilibrium.
Because ∆G◦ is a constant quantity (standard conditions are defined constants), it follows that
the term containing the concentrations is also a constant for this equilibrium state. In fact,
comparing Eq. 2.19 with Eq. 2.23, we see that
∆G◦ = −RT ln Keq (2.24)
Eq. 2.24 enables us to use thermodynamic data (in particular standard Gibbs energies) to
calculate the equilibrium constant for any real or imaginary chemical reaction. Put differently,
this equation describes the driving force towards equilibrium when proceeding from standard
conditions. Eq. 2.21 gives the driving force towards equilibrium when proceeding from any
other conditions.
Remember that the ratio of concentrations that occurs in the expression for the equilibrium
constant is called the mass-action ratio, Γ. Importantly, in the mass-action ratio these concen-
trations can refer to any conditions. When the system is at equilibrium, the concentrations in the
ration are equilibrium concentrations, and the equilibrium constant is just a special value of the
general mass-action ratio. For the reaction in Eq. 2.17, the mass-action ratio is given by:
(c)p (d)q
Γ= (2.25)
(a)m (b)n
We can now write Eq. 2.21 as

∆G = −RT ln Keq + RT ln Γ (2.26)
or
Γ
∆G = RT ln (2.27)
Keq
Eq. 2.27 provides an excellent summary of ∆G as a measure for the driving force of a reaction
towards equilibrium. The ratio Γ/Keq indicates how far a reaction is away from equilibrium. ∆G
is merely this term on a log-scale, multiplied by a constant to convert it to energy units.
The ratio Γ/Keq is also known as the disequilibrium ratio and evidently is very important for
the description of the thermodynamics of chemical reactions. Fig. 2.1 indicates schematically
how this quantity can be interpreted as a measure for the “distance” from equilibrium and the
direction of a reaction. This scheme also indicates clearly that the value of the disequilibrium
ratio under standard conditions is the reciprocal of the equilibrium constant.
Remember that ∆G◦ indicates the direction of a reaction only in the special case where Γ = 1
(standard conditions). Under any other conditions, the corresponding value of ∆G has to be
used. I emphasise this because ∆G ◦ and ∆G are often confused in biochemistry; because ∆G◦ can
be measured more easily, it is (wrongly!) used as the measure for the tendency of a reaction to
occur. From the above it should be clear that, for conditions differing from the standard state
(such as those present in living cells), the values of ∆G and ∆G ◦ will differ; they will sometimes
even differ in sign.
24
Only Equil. Only

reactants mixture products
Standard
conditions eq.
Γ
0 1 Keq ∞
Γ
Keq ∞
0 1 1
Keq
Γ Γ
0< <1 1< <∞
Keq Keq
Γ
ln
Keq
−∞ − ln Keq 0 ∞
Γ
∆G = RT ln
Keq
−∞ ∆G◦ = −RT ln Keq 0 ∞
∆G < 0 ∆G > 0
Figure 2.1: How the disequilibrium ratio measures the “distance” from equilibrium and the
direction in which a reaction proceeds from any specified value of the mass-action
ratio. The relationship between the disequilibrium ratio and the Gibbs energy of a
reaction is also shown.
The relationship between ∆G, ∆G ◦ , Keq and the direction of the reaction is illustrated in Figs. 2.2
and 2.3. The curves for both figures were calculated for the simple reaction A ⇋ B. The abscissa
shows the mole fraction of B in the mixture, while the vertical scale indicates the Gibbs energy
for a system containing both A and B, with a total amount A + B of 1 mol. The curve thus
describes the relative Gibbs energy of such a system as a function of the changing mole fractions
of its components. The slope of the curve at any point equals ∆G (in J mol −1 ) for the reaction
A ⇋ B. One might imagine the reaction as a ball rolling inside the curve towards the direction
of equilibrium (at the bottom where the slope equals zero); the tendency to roll becomes greater
and greater the further the ball is away from equilibrium (higher up against the curve), because
the slope of the curve is constantly increasing.
In Fig. 2.2 the equilibrium constant Keq equals 1. ∆G◦ , the slope of the curve when Γ = 1
(and thus, when the mole fraction equals 0.5), equals zero here because at equilibrium the
concentrations of reactant and product are equal so that Γ = 1 (if K eq = 1, the mass action
ratio of the reaction mixture at equilibrium is the same as that under standard conditions:
0.5/0.5 = 1/1 = 1). The slope drawn at arbitrary mole fraction a illustrates the fact that the
slope of the curve at any point equals the molar Gibbs energy change of the reaction for that
25
1600
1400
1200
Relative Gibbs energy (J)

1000 Keq = 1
800
600
∆G
400
200
0
a ∆G◦ =0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
[B]
Mole fraction of B, [A]+[B]
Figure 2.2: Relative Gibbs energy of a system containing the compounds A and B in a total
amount of 1 mol, plotted as a function of the mole fraction of B. ∆G, the molar change
in Gibbs energy for the reaction at an arbitrary composition of the A-B mixture,
is given by the slope of the curve at the point referring to that reaction mixture
composition (e.g., at point a).
composition of the reaction mixture.

Fig. 2.3 illustrates curves for two reactions with equilibrium constants different from 1. When
Keq = 4, the system is at equilibrium if the mole fraction of B equals 0.8. If the mole fraction is
less than 0.8, the slope is negative and the reaction from A to B is thermodynamically possible.
The standard Gibbs energy change ∆G ◦ is, as before, the slope at a mole fraction of 0.5, i.e. when
Γ = 1.
The other curve in Fig. 2.3 shows the reciprocal case where K eq = 0.25 and ∆G◦ is positive.
Because the relationship between Γ and ∆G of a reaction only has one degree of freedom, the
whole curve is specified by the value of the slope at any value of the mole fraction (or any value
of Γ for the more general case of a reaction having more than one substrate and/or product). It is
for this reason that ∆G◦ values are so convenient. The value of ∆G◦ by itself is not so significant,
but it can be used to calculate the equilibrium point (the minimum of the curve), as well as the
value of ∆G (the slope of the curve) at any other value of Γ: the further removed the standard
state is from equilibrium, the greater is the slope at that point (the value of ∆G ◦ ).
Remember that G does not have an absolute value, so that the position of zero on the vertical
axis is arbitrary. For convenience’s sake, in order to compare the curves, we assigned a value of
26
4000
3500
3000
25
K eq
= 0.
Relative Gibbs energy (J)
=4
2500
Keq
2000
1500
−1
∆G
◦
ol
Jm
=
−3
35
1000 43
34
5J
=
◦
m
∆G
ol
−1
500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
[B]
Mole fraction of B, [A]+[B]
Figure 2.3: The same as Fig. 2.2, except that the relative Gibbs energy changes are shown for the
cases where Keq = 0.25 and Keq = 4. The standard change in Gibbs energy, ∆G ◦ , is
the slope of the curve at a reaction mixture composition that gives a value of 1 for the
mass-action ratio Γ.
zero to the relative Gibbs energy of the equilibrium mixture (the minimum of the curve).
2.4 Coupled reactions

Two reactions are said to be coupled when the product of one reaction serves as a reactant for
the next reaction; this substance is then called the common intermediate. Such a situation is
common in metabolic pathways. An example is:
A⇋B⇋C
where B is the common intermediate. For such a system of coupled reactions to be in equilibrium,
each individual reaction has to be in equilibrium.
Therefore, if Keq1 and Keq2 are the two equilibrium constants,
b c
Keq1 = and Keq2 = (2.28)
a b
27
where a, b, and c are equilibrium concentrations.

For the sequence as a whole one can also write an equilibrium constant K eq12 as
c
Keq12 = (2.29)
a
It follows that the overall equilibrium constant must be the product of the individual equilib-
rium constants:
bc c
Keq1 Keq2 = = = Keq12 (2.30)
ab a
We know, however, that the equilibrium constant is related to the standard Gibbs energy
change for a reaction (∆G◦ ). For the coupled system, we have from the definition of ∆G ◦
(Eq. 2.24):
∆G◦1 = −RT ln Keq1 (2.31)

∆G◦2 = −RT ln Keq2 (2.32)
∆G◦12 = −RT ln Keq12 (2.33)
Since Keq12 = Keq1 Keq2 , we can derive the following by taking logarithms on both sides and
multiplying by −RT throughout:
−RT ln Keq12 = −RT ln Keq1 − RT ln Keq2 (2.34)
Hence,
∆G◦12 = ∆G◦1 + ∆G◦2 (2.35)
This relationship emphasises what you know already: ∆G ◦ -values can be added, so that if ∆G◦
is known for two reactions, we can calculate, by addition, the ∆G ◦ of the net reaction when the
two reactions are coupled.
This additivity is of course not only a property of ∆G ◦ -values, but of any ∆G-values in general.
The following example illustrates an important point which can easily be overlooked, i.e., two
reaction series with the same global overall Keq (and thus the same overall ∆G◦ ) do not necessarily
lead to the same equilibrium concentration of the final product. Consider the following:
If the initial concentration of reactant A is 1 M, which of the following two coupled reaction
series will lead two the greatest equilibrium concentration of the product C? (a) series 1; (b) series
2; or (c) both series lead to the same equilibrium concentration of C.
• Series 1
A ⇋ B ∆G◦ = +18, 85 kJ mol−1 ; Keq = 5 × 10−4
B ⇋ C ∆G◦ = −18, 85 kJ mol−1 ; Keq = 2 × 103
• Series 2
A ⇋ B ∆G◦ = −18, 85 kJ mol−1 ; Keq = 2 × 103
B ⇋ C ∆G◦ = +18, 85 kJ mol−1 ; Keq = 5 × 10−4
The sum of the ∆G◦ values (or the product of the K eq values) is the same for both reactions
(∆G◦ = 0 and Keq = 1). Perhaps your immediate response is that as a result of this the equilibrium
concentration of C should be the same in both cases. However, you have to distinguish carefully
between an equilibrium concentration and an equilibrium ratio. Although the final ratio c/a
needs to be the same for both series, the equilibrium concentration (c) eq differs between the two
cases. In series 1, approximately 49.99 % of the A that was originally present is converted to C.
28
In series 2, however, the equilibrium levels of A and C are each only approximately 0.05 % of
the original a, while 99.9 % accumulates as B. Evidently, series 1 converts A to C more efficiently
than series 2.
To generalise: a reaction with a low Keq for A ⇋ B can thus be effectively “pulled over” to
C by a following reaction B ⇋ C with a high Keq ; however, a high Keq reaction A ⇋ B cannot
“push over” a low Keq reaction B ⇋ C to C.
2.5 Kinetic and energetic aspects of reaction rate

We have seen how the equilibrium constant for a chemical reaction is related to ∆G ◦ and how
the driving force ∆G is related to the disequilibrium ratio Γ/K eq . We have also seen that the rate
of a reaction depends on the concentrations of its reactants and products, as well as the forward
and reverse rate constants k f and kr (see Eq. 2.14). How can we unite these thermodynamic and
kinetic aspects of reaction rate?
In principle, the rate of any reversible reaction can be written in a form where the kinetic aspects
are clearly separated from the thermodynamic aspects. Consider a possible rate equation for the
reaction A + B ⇋ C + D:
v = kf ab − kr cd (2.36)
When we factor out the forward rate term we obtain

kr cd
v = kf ab 1 − (2.37)
kf ab

1 cd
= kf ab 1 − (2.38)
Keq ab

Γ
= kf ab 1 − (2.39)
Keq
A reversible rate equation is therefore always the product of a kinetic term k f and an energetic
term ab(1 − Γ/Keq ), which can also be written as (ab − Kcdeq ). The kinetic term always contains
time−1 as a unit, and its functional form depends on the mechanism of the reaction: here it is very
simple, just a rate constant, but it can become quite complicated, e.g., for an enzyme catalysed
reaction.
The energetic term, on the other hand, consists only of concentrations and the equilibrium
constant and its form is always the same. Clearly, when the reaction is in equilibrium (Γ/Keq = 1)
the energetic term is zero; otherwise, when Γ/K eq < 1 the energetic term is positive, indicating a
net reaction in the forward direction, or when Γ/K eq > 1 it is negative, indicating a net reaction
in the reverse direction. The greater the distance from equilibrium, the faster the rate of reaction.
29
3 Enzyme kinetics
Up to now, we only considered reactions following mass-action kinetics. Clearly, the situation
in the living cell is more complex, as cellular reactions are catalysed by enzymes and follow
enzyme kinetics. In this section, we discuss how enzymatic catalysis affects the shape of rate
characteristics; this is the field of enzyme kinetics.
3.1 The reversible Michaelis-Menten equation

In most standard biochemistry textbooks, enzyme kinetics is introduced with the irreversible
Michaelis-Menten equation:
Vmax s
v=
KM + s
where Vmax is the maximal rate of the enzyme and KM is the Michaelis constant of the enzyme
for its substrate. This form of the equation has the serious shortcoming that it only considers
irreversible reactions. However, in principle all reactions are reversible (see the general rate
equation given above in Eq. 2.39). In this section we derive how incorporating reversibility
changes the form of the Michaelis-Menten equation. We will see that the reversible Michaelis-
Menten equation is a much more useful tool for describing the kinetic behaviour of reactions in
living cells than the irreversible equation, which ignores the product.
Consider the following generalised reaction scheme for the enzyme-catalysed conversion from
S to P:
k2
E + S⇋ES⇋EP⇋E + P (3.1)
Ks k−2 Kp
Remember that the Michaelis-Menten treatment assumes that the two binding equilibria above
(with dissociation constants Ks and Kp ) are rapid when compared to the interconversion of ES
and EP (defined by the rate constants k 2 and k−2 ). Thus, substrate S is in equilibrium with
enzyme-substrate complex ES, and product P is in equilibrium with enzyme-product complex
EP.
If S is reacting in the initial absence of P, the concentration of EP is nearly zero; the forward
rate reaches its limiting value, Vf , when the catalytic sites of the enzyme are saturated so that es
may be assumed to be equal to the total enzyme concentration e 0 .
Vf = k2 es = k2 e0 (3.2)
Similarly, if P is reacting in the initial absence of S, the concentration of ES is nearly zero; the
reverse rate reaches its limiting value, Vr , when the catalytic sites of the enzyme are saturated
so that ep may be assumed to be equal to e0 .
Vr = k−2 ep = k−2 e0 (3.3)
However, if both S and P are present, neither can saturate the enzyme. For any given concen-
tration of S, the fraction of S bound to the enzyme is reduced by increasing concentrations of P.
30
For any given concentration of P, the fraction of P bound to the enzyme is reduced by increasing
concentrations of S.
This is a kinetic inhibition of the forward and reverse rates of reaction. It is an additional and
quite different effect from the normal mass-action effect experienced by all reactions (catalysed
or not) that when they approach equilibrium their net rate decreases because of the increasing
rate of the reverse reaction (as shown in the term 1 − Γ/K eq above in Eq. 2.39).
With this background one can construct the reversible Michaelis-Menten rate equation.
1. We write the conservation equation for enzyme forms. The enzyme can exist either as free
enzyme E, or bound to substrate (ES) or product (EP). The total enzyme concentration e 0
is thus the sum of all the different forms of E:
e0 = e f + es + ep (3.4)
where e f denotes the concentration of free enzyme E, bound neither to S nor to P. As

throughout this course, we use lowercase italics to denote concentrations.
2. The dissociation equilibria of S and P from the enzyme are characterised by the dissociation
constants Ks and Kp respectively:
ef · s ef · p
Ks = and Kp = (3.5)
es ep
In principle, s and p above refer to the free concentrations of S and P respectively. How-
ever, most of the time substrate and product concentrations are much larger than enzyme
concentrations, so that the fraction of S or P that is bound to the enzyme (in the form of ES
or EP) is negligible. For practical purposes, one can therefore equate the total substrate or
product concentration (s0 or p0 ) with the free concentration (s f or p f ), and we simply write
s or p.
Using the above expressions for K s and Kp , we rearrange the conservation equation (Eq. 3.4)
into an expression for Y ES = es/e0 , i.e. the fractional saturation of the enzyme with S:
es es
YES = =
e0 e f + es + ep
ef s
Ks
= ef s ef p (from Eq. 3.5)
e f + Ks + Kp
s
Ks
= p (3.6)
1 + Kss + Kp
Using the same approach, one can derive the expression for Y EP = ep/e0 , i.e. the fractional
saturation of the enzyme with P:
p
ep Kp
YEP = = s p (3.7)
e0 1+ +
Ks Kp
3. The net rate of the catalysed reaction is given by
v = k2 es − k−2 ep
31
Substituting es and ep from Eqs. 3.6 and 3.7,
s p
Ks Kp
v = k2 e0 p − k−2 e0 p
1 + Kss + Kp 1 + s
Ks + Kp
p
k2 e0 Kss − k−2 e0 Kp
= p (3.8)
1 + Kss + Kp
Using the definitions of V f and Vr (Eqs. 3.2 and 3.3), we obtain the reversible Michaelis-
Menten equation:
p
Vf Kss − Vr Kp
v= s p (3.9)
1+ Ks + Kp
3.1.1 The Haldane relationship

Recall that the net rate of any chemical reaction (and thus also of an enzyme-catalysed reaction)
at equilibrium is zero. Assume that the reaction above is in equilibrium. From Eq. 3.9 above it
follows that:
(s)eq (p)eq
Vf = Vr
Ks Kp
(p)eq Vf Kp
Hence, = Keq = (3.10)
(s)eq Vr Ks
As before, (s)eq and (p)eq denote the equilibrium concentrations of S and P. Eq. 3.10 is known
as the Haldane relationship, named after the scientist who first described it. The Haldane
relationship relates the kinetic constants of the reversible Michaelis-Menten equation to the
equilibrium constant. Because the equilibrium constant for a reaction only depends on substrates
and products (and not on the enzyme; a catalyst does not alter the equilibrium), the Haldane
relationship further shows that the four kinetic constants V f, Vr , Ks and Kp are not independent. If
three of them are known, the fourth can be calculated from their values and from the equilibrium
constant.
Using the Haldane relationship, we can rewrite the reversible Michaelis-Menten equation:
p/s
Vf Kss − Vf Kss Keq
v = p (substitute Vr and × ss )
1 + Kss + Kp

Vf Kss 1 − KΓeq
= s p
1+ Ks + Kp

Vf 1 Γ
v= × s p ×s 1− (3.11)
Ks 1 + + Keq
Ks Kp
Eq. 3.11 illustrates the important point that any rate equation can be written in the following
form (see also Section 2.5):
32

Γ
v = k × fkin × s 1 −
Keq
p
Here, k = Vf /Ks is the overall rate constant for the reaction. fkin = 1/(1 + Kss + Kp ) is the kinetic
term describing the effect of saturation binding of S and P to the enzyme, and s(1 − Γ/K eq ) is
the thermodynamic (energetic) term which depends only on the concentration of substrates and
products (compare Eq. 2.39).
Because Eq. 3.11 makes the equilibrium constant explicit, it separates kinetic and thermody-
namic aspects. This is not true of the form of the equation given in Eq. 3.9.
3.2 Uncompetitive inhibition

As pointed out in the previous section, the product P acts as a competitive inhibitor for the
substrate S in an enzyme following reversible Michaelis-Menten kinetics. The competitive
inhibition is due to S and P binding at the same active site of the enzyme: binding of P blocks
the active site and prevents S from binding. The extent of inhibition is quantified by the p/K p
term in the denominator of the rate equation (Eqs. 3.9 and 3.11), and is part of the kinetic term
fkin .
Sometimes, however, a product can also inhibit an enzyme uncompetitively by binding to the
ES-complex, causing both S and P to be bound to the enzyme concomitantly and preventing
catalysis. In this section, we investigate how uncompetitive inhibition will affect the rate equation
for the reaction.
Consider the following reaction scheme:
k2
E+S ⇋ ES ⇋ EP ⇋ E + P
Ks k−2 Kp
+P (3.12)
Kiu ⇃↾
ESP
Because the enzyme can now exist in an additional form, ESP, we have to include this in the
conservation equation for the different enzyme forms:
e0 = e f + es + ep + esp (3.13)
The two binding equilibria for S and P to the free enzyme are the same as before (Eq. 3.5).
The additional binding equilibrium for P to ES is characterised by a dissociation constant Kiu (Ki
stands for inhibition constant, and the u stands for uncompetitive ):
es · p
Kiu = (3.14)
esp
We solve the above equation for esp, using the definition of K s in Eq. 3.5:
es · p e·s·p
esp = = (3.15)
Kiu Ks · Kiu
33
As previously, we rearrange the conservation equation (Eq. 3.13) into expressions for the
fractional saturation of the enzyme with S, P, and both S and P:
s
es Ks
YES = = s p s·p (3.16)
e0 1+ + Kp + Ks Kiu
Ks
p
ep Kp
YEP = = s p s·p (3.17)
e0 1+ + Kp + Ks Kiu
Ks
s·p
esp Ks Kiu
YESP = = s p s·p (3.18)
e0 1+ + Kp + Ks Kiu
Ks
The net rate of the catalysed reaction is again given by
v = k2 es − k−2 ep
As before, we substitute es and ep from Eqs. 3.16 and 3.17,

p
k2 e0 Kss − k−2 e0 Kp
v= s p s·p (3.19)
1+ Ks + Kp + Ks Kiu
Using the same definitions of V f and Vr as before (Eqs. 3.2 and 3.3), and factorising the de-
nominator, we obtain the reversible Michaelis-Menten equation with uncompetitive product
inhibition:
p
Vf Kss − Vr Kp
v= p
p (3.20)
1 + Kss 1 + Kiu + Kp
or, in the form separating the thermodynamic and kinetic terms:

Vf Kss 1 − KΓeq
v= p
p
(3.21)
1 + Kss 1 + Kiu + Kp
p

The additional term 1 + Kiu in the denominator of Eqs. 3.20 and 3.21 quantifies the extent of
uncompetitive inhibition by P. The above equations also show that when K iu = ∞ (P does not
bind to ES), the inhibition is purely competitive (given by the term p/K p ). Likewise, if Kp = ∞
(P does not bind to the free enzyme), the inhibition is purely uncompetitive (given by the term
p
1 + Kiu ). In general, neither K p nor Kiu is infinitely large, so that the inhibition by P will have a
competitive and an uncompetitive component (this is called mixed inhibition ).
3.3 Cooperativity and the reversible Hill equation

The reversible Michaelis-Menten equation discussed above is a vast improvement on the com-
monly used irreversible form, because it successfully takes into account the reversibility of
enzyme-catalysed reactions and separates the thermodynamic and kinetic contributions to reac-
tion rate. However, it fails to explain the following phenomena:
34
Figure 3.1: Kinetic model for a two-site enzyme.
• Many enzymes follow cooperative kinetics. These enzymes have multiple binding sites for
substrates, and the binding of one substrate molecule changes the affinity of the other
binding sites so that subsequent substrate molecules bind better (positive cooperativity)
or worse (negative cooperativity). Positive cooperativity leads to a sigmoidal dependence
of reaction rate on substrate concentration, whereas the v vs. s curve is flattened for
an enzyme displaying negative cooperativity. The reversible Michaelis-Menten equation
cannot explain cooperativity.
• Many enzymes in biosynthetic metabolic pathways are subject to allosteric feedback inhibi-
tion, where an end-product further down in the metabolic pathway inhibits the committing
enzyme by binding to a separate (allosteric) site, which is distinct from the active site. Such
feedback inhibition usually also follows sigmoidal kinetics, and can likewise not be ex-
plained by the reversible Michaelis-Menten equation.
3.3.1 The reversible Hill equation

Consider the scheme in Fig. 3.1. It shows a kinetic model for a two-site enzyme. Each active site
can either bind a molecule of S or a molecule of P. The enzyme can also exist in states where one
or both sites are empty.
The arrows show all the possible transitions between the different enzyme states. The hori-
zontal arrows represent catalytic conversions between S and P; the forward rate constant for the
conversion from S to P is kf , and the reverse constant (P to S) is k r . The arrows linking the vertical
levels represent binding/dissociation equilibria for S or P.
In the bottom level of Fig. 3.1 the rate constants k f and kr are multiplied by 2 because there
are two molecules of S bound to the enzyme in the ES 2 species (and two molecules of P in
EP2 ), whereas there is only one S or P in each of ES, EP or ESP. This makes the likelihood S or
P reacting in the ES2 or EP2 species twice as high as in the other species (because any one of
the two molecules can react). The 2 thus is a “statistical factor” to take into account that two S
molecules or two P molecules are bound to the enzyme at the same time.
In an analogous manner, by taking all the horizontal arrows into account, the net rate v of
conversion from S to P can be given by:
v = kf (es + 2es2 + esp) − kr (ep + 2ep2 + esp) (3.22)
35
Note how the concentrations es 2 and ep2 are multiplied by 2 because they have two molecules
of S or P bound. The term esp occurs in both the forward and the reverse reaction rates, because
the species ESP can react to form either EP 2 (forward reaction) or ES2 (reverse reaction).
Eq. 3.22 can be solved for v (we will not give the derivation here):

p

s Γ s
Vf s0.5 1 − Keq s0.5 + p0.5
v= (3.23)
p 2

s
1 + s0.5 + p0.5
where s0.5 is the concentration of S giving half-maximal forward reaction rate in the absence of
P, and p0.5 is the concentration of P giving half-maximal reverse reaction rate in the absence of S.
Vf , Γ and Keq are defined as previously.
We now introduce the following shorthand: σ = s/s 0.5 and π = p/p0.5 . We use this to rewrite
the above equation in a simpler form:

Γ
Vf σ 1 − Keq (σ + π)
v= (3.24)
1 + (σ + π)2
Eq. 3.24 is the reversible Hill equation for an enzyme with two active sites. The general form
reads
Γ
Vf σ 1 − Keq (σ + π)h−1
v= (3.25)
1 + (σ + π)h
Self-study exercises
1. The parameter h is known as the Hill coefficient and can vary between 1 and the number of
active sites on the enzyme. The greater h, the stronger is the cooperativity of the enzyme.
What happens if h = 1?
2. How does Eq. 3.25 reduce when p = 0? (This reduced equation is known as the irreversible
Hill equation and was first described in 1910. It is widely used in biochemistry to describe
cooperative enzymes.) Incidentally, the reversible Hill equation was developed by Prof
Hofmeyr from our department in 1997.
3. How does Eq. 3.25 reduce when s = 0? Why is the reaction rate negative?
4. How many reasons can you think of why the reversible Hill equation is better than the
irreversible one?
3.3.2 Modifier effects in the reversible Hill equation

Although the reversible Hill equation quantitatively describes cooperativity and gives a sig-
moidal rate vs. substrate concentration graph, Eq. 3.25 in its present form cannot account for
modifier effects, i.e. inhibition or activation by an allosteric modulator. However, these allosteric
modifier effects are central to understanding metabolic regulation because of the ubiquitous feed-
back loops present throughout metabolism. Fortunately there is a way to extend the reversible
Hill equation to incorporate a modifier.
36
Assume that X is a modifier binding to the enzyme, x is its concentration and x 0.5 the concen-
tration at which X exerts half of its total effect. Let ξ = x/x 0.5 . The reversible Hill equation with
modifier then reads:
Γ
Vf σ 1 − Keq (σ + π)h−1
v= (3.26)
1+ξh
(σ + π)h + 1+αξh
Eq. 3.26 applies equally well to allosteric inhibition and activation. Check this for yourself: if
α < 1, then X is an inhibitor; if α > 1, then X is an activator; if α = 1, then X has no effect on the
reaction rate.
1 (a) 1 (b)
0.8
0.1
0.6
v
v
0.01
0.4
0.001
0.2
0 0.0001
0 1 2 3 4 5 0.01 0.1 1 10 100 1000
x x
Figure 3.2: Graphs of the reversible Hill equation with modifier. (a) linear coordinates; (b) loga-
rithmic coordinates. The graphs show the dependence of reaction rate v on modifier
concentration x. The fixed parameters are: V f = 2, s = s0.5 = 1, p = 1, p0.5 = 104 ,
Keq = 104 , h = 4, α = 10−4 , and x0.5 = 1.
Fig. 3.2 shows graphically how the reversible Hill equation (Eq. 3.26 above) depends on the
concentration of the modifier X, while keeping the other parameters constant. We plot the rate as
a function of x here, because the modifier plays the role of allosteric inhibitor in many metabolic
pathways.
Fig. 3.2 (a) shows v vs. x in linear space, whereas Fig. 3.2 (b) shows the same graph in double-
logarithmic space. We show the logarithmic graph, because you can directly read off from the
graph the reaction order with respect to X (see Section 2.1.1).
For now, it is sufficient to note that both graphs in Fig. 3.2 show a sigmoidal dependence of v
on x, and X acts as an inhibitor of the enzyme (v decreases as x increases). The dependence of v
on s (the substrate concentration) is also sigmoidal, but in that case v increases with increasing s
(this graph is not shown here).
Meaning of parameters You might be wondering what is the effect of changing the parameters
in the reversible Hill equation, and how this affects the shape of the graph.
Changing the value of the hill coefficient h changes the degree of cooperativity in the binding
of X (Fig. 3.3(a)). The lower h, the less steep is the sigmoidal curve and the weaker is the
cooperativity. When h = 1, there is no cooperativity and the curve is no longer sigmoidal.
Changing the value of α varies the extent to which X inhibits the enzyme (Fig. 3.3(b)). If we
lower α, the plateau of maximum inhibition shifts down and the enzyme is inhibited more. If
37
(a) (b)
1 1
0.1 0.1
α = 0.01
h=4 2 1
v
v
0.01 0.01
0.001
0.001 0.001
0.0001
0.0001 0.0001
0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000

x x
(c) (d)
1 1
0.1 0.1
x0.5 = 0.1 1 10
v
v
0.01 0.01
0.001 Vf = 6 0.001
0.0001 0.67 0.0001
0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000

x x
Figure 3.3: The effect of changing the parameters in the reversible Hill equation. The reference
parameters were as in Fig. 3.2 and are indicated by the bold line in each graph. In
each case, one of the parameters is varied around its reference value while keeping
the others constant at their reference values. The parameters varied are (a) h; (b) α;
(c) V f ; and (d) x0.5 .
α = 1, there is no effect of X on the reaction rate, and if α > 1, X activates the enzyme (the latter
two cases are not shown on the graph).
Changing the Vf of the enzyme varies its total capacity. As a consequence, the whole curve
shifts up when Vf is increased and down when V f is decreased (Fig. 3.3(c)).
Finally, changing x0.5 varies the concentration at which X is an effective inhibitor (Fig. 3.3(d)).
As a consequence, the curve shifts to the left if x 0.5 is decreased and to the right if x 0.5 is
increased. Observe from the graph that the enzyme is halfway inhibited (v = 0.5) at a point
where x is approximately equal to x0.5 .
We now need to return to coupled reactions and rate characteristics and learn yet another tool
(the last one!), which we need for our functional view of metabolic regulation. This tool is called
metabolic control analysis.
38
4 Control Analysis
The aim of control analysis is twofold: First, we want to understand how sensitive the steady-
state variables (the fluxes and concentrations) are to variations in the parameters. As these
sensitivities depend on all the interactions within a cellular system, we call them global or
systemic properties. Second, we want to relate these global sensitivities to local properties of the
reactions (enzymes) themselves. How these local and global properties are expressed remains
to be seen.
The reason that this type of analysis is called ‘metabolic control analysis’ arises from a specific
meaning that we attach to the concept of control. The term ‘metabolic control’ is often used
rather loosely and qualitatively, and we need to define it in a precise and quantitative way.
Generally, ‘to control’ is understood to mean ‘to be able to influence something’. When we say
that an enzyme controls the flux through a pathway, we mean that a change in the activity of the
enzyme results in a change of the flux. Unfortunately, the word ‘control’ seems to have acquired
an ‘all or none’ connotation: something exerts either complete control or no control at all. This
qualitative meaning has been taken over in the concept of a ‘rate-limiting’ enzyme, which would
(completely) control the flux. We shall see that, on the contrary, control of flux is often shared
by all the enzymes in a metabolic system; in fact, control of any steady-state variable is shared
by all the enzymes. From the following treatment it will become clear exactly how control can
be defined quantitatively in this context. Remember that we made a careful distinction between
parameters and variables in cellular systems. This is important, as the parameters must be
regarded as the ‘controllers’, the variables as the ‘controlled’.
The aim of this chapter is to give more insight into what control analysis tells us about cellular
systems, and also to show the valuable insight that can be gained about metabolic behaviour by
performing ‘thought experiments’.
4.1 Quantifying metabolic control

As we have seen in Chapter 1 the relationship between the steady-state variables (fluxes and
concentrations) and parameters that affect the activities of the enzymes in a system is non-
linear and cannot be obtained analytically, except in the simplest cases. At present, no theory
will therefore be able to predict quantitatively what the change in, say, flux would be for a
large change in enzyme activity (unless we make the simplifying and dangerous assumption
that flux responds linearly or logarithmically to enzyme concentration). In order to arrive at a
mathematically exact and consistent formulation of metabolic control, control analysis therefore
concerns itself only with small changes (more precisely, infinitesimal changes). In mathematical
terms this means we linearise the system equations around a steady-state. From our definition
of the kinetic model and a knowledge of the time frame we are working in we know what the
controllers (parameters) and the controlled (variables) are. Now we need some mathematical
function that quantifies control. Traditionally, control analysis uses ratios of relative changes
(also called fractional changes ), which have the advantage that they are dimensionless. This
function is used to define the ‘coefficients’ which describe metabolic control. As mentioned
39
above, two types of coefficients, local and global, are distinguished.
4.1.1 Local coefficients: Elasticity coefficients

We start off our exploration of the world of control analysis by focusing on the functional steps
of the system (here we regard each functional step to be catalysed by an enzyme, but this is
not strictly necessary; any reaction or transport step can be treated in this way). As we have
seen in our kinetic models, each enzyme is directly connected to specific metabolites, i.e., its
substrates, products and effectors (which can be either fixed external inhibitors or activators, or
variable internal effectors, i.e., internal metabolites that directly affect the enzyme by feedback
or feedforward interactions). These metabolites are the ones that will appear in the rate equation
of the enzyme-catalysed reaction.
We now answer the following question: How can we describe the sensitivity of the local
rate through an enzyme to small changes in these directly-connected metabolites? Note that,
to answer this question, we must regard each enzyme separately, as if it were isolated (in
fact, although we answer this question by conducting a thought experiment, it could easily be
studied experimentally with the isolated enzyme in a test tube, using the well-known procedures
of steady-state enzyme kinetics).
So, let us conduct the thought experiment: Imagine the metabolic system ‘frozen’ in the
steady-state. We now ‘remove’ the enzyme we are interested in from the system, together with
the surrounding steady-state concentrations of all the metabolites that interact directly with it.
In this ‘frozen’ state, we put it into a test tube in which the general conditions are the same
as in the intact system (i.e., pH, ionic strength, protein concentration, etc.). First, it should be
clear is that if we ‘unfreeze’ the enzyme, the initial rate of the reaction must be the same as the
steady-state flux through the enzyme when it was part of the system. Second, it is possible to
imagine adding a small amount δs of, say, the substrate S, before unfreezing the enzyme, while
keeping all the other metabolites constant at their prevailing concentrations. When we unfreeze
the enzyme at this new concentration of the substrate a slight change in initial rate δv will be
measured. We can repeat this experiment at other concentrations s around the steady-state
concentration. If we were to plot these results as initial rate v against s we would obtain a curve
around the original steady-state concentration (Fig. 4.1). The slope of the tangent to the curve at
the steady-state point δv/δs is an indication of how sensitive the local enzyme rate is to changes
in the concentration of S when the system is in steady state. The value of the slope will, however,
depend on which units we choose to express rate and concentration; we can make the slope
dimensionless be dividing δv and δs by the steady-state values of v and s respectively. We call
this scaled slope the elasticity coefficient of v with respect to s:
δv/v
εvs = (4.1)
δs/s
In words, the elasticity coefficient can be described in five ways:
1. The ratio of the fractional change δv/v in the local rate of enzyme and the fractional
change δs/s in the concentration of a metabolite that interacts directly with the enzyme
at a specified value of s (and, of course, specified values of all the other metabolites that
directly affect v).
2. Multiplying both fractional changes by 100 will give the percentage change in each, so that
we can also describe the elasticity coefficient as the % change measured in the rate of an
enzyme in response to a 1% change in a metabolite concentration.
40
3. The product of s/v and the slope of the tangent δv/δs at a specified value of s on a plot of v
against s.
4. An even more elegant way of expressing elasticity coefficients takes into account that,
mathematically speaking, δx/x = δ ln |x|. This means that we can also define the elasticity
coefficient as
δ ln v
εvs = (4.2)
δ ln s
in other words, the slope of the tangent to the curve obtained by plotting ln v against ln s.
This is shows explicitly in Fig. 4.1
5. The apparent kinetic order of the rate with respect to s at a specified value of s. This
description implies that an elasticity coefficient is a variable enzymological property; like
all other enzymic properties (KM , V, etc.) its value is constant for a prescribed set of
conditions but will vary as conditions vary.
S ε vs = δδlnln[S]v > 0
ln v
ln [S]
δ ln v
ε v
I
=
δ ln [I]
<0
I ε vE = δδlnln[E]v
ln v
= 1
ln [E]
ln v
ln v
ln [I]
ln [P]
ε vp = δδlnln[P]v
P <0
Figure 4.1: Common elasticities of a enzyme-catalysed reaction.
For those who like a more formal mathematical definition, we must consider the limit where
the changes become infinitesimally small, i.e., when δ → 0. Then the elasticity coefficient of a
rate vi for any metabolite, S j , that directly interacts with the enzyme that catalyses reaction i is
41
equivalently defined as
s j ∂vi

∂vi /vi ∂ ln vi
εvs ji = = = (4.3)
∂s j /s j sk ,sl ,...
vi ∂s j s ,s ,... ∂ ln s j s ,s ,...
k l k l
The sk and sl outside the brackets indicate that any other metabolites that interact with the
enzyme are kept constant at the prevailing steady-state values; that is why we define an elasticity
coefficient as a partial derivative.
Note that elasticity coefficients describe the properties of the ‘isolated’ functional units in the
system and, individually, are uninformative about system behaviour. Nevertheless we shall see
that they enter in important ways into control analysis.
Above we considered the elasticity coefficient of an enzyme with respect to a substrate. An
enzyme will of course have elasticity coefficients with respect to all other metabolites that interact
with it directly (i.e. that appear in its rate equation). For example, keeping the substrate constant,
we can now vary the product concentration around its steady-state value to obtain:
δv/v
εvp = (4.4)
δp/p
It should also be clear that there are other entities that directly affect the rate of an enzyme
reaction, e.g., the enzyme concentration e, so that
δv/v
εve = (4.5)
δe/e
When the rate equation for the reaction is known, an explicit expression for an elasticity
coefficient can be obtained as the scaled partial derivative of the rate function towards an
interacting metabolite. For instance, the net-rate equation for the reaction S⇋P could be that of
a reversible Michaelis-Menten mechanism:

Vf p
s−
Ks Keq
v= p (4.6)
1+ s +
Ks Kp
where V f denotes the limiting forward velocity, K eq the equilibrium constant and Ks and Kp the
respective Michaelis constants for S and P. Partial differentiation with respect to s or p and
scaling of the partial derivative with s/v and p/v respectively gives the elasticities towards s and
p:
s
1 K s
εvs = − p (4.7)
1− Γ 1+ s +
Keq Ks Kp
p
− Γ
Keq Kp
εvP = − (4.8)
1− Γ 1+ s + p
Keq Ks Kp
where Γ is the mass-action ratio p/s.
42
An enzymic rate equation is usually a proportional to enzyme concentration e, i.e., e is a

multiplier of some function of s, p, etc. This can be seen in the above reversible Michaelis-
Menten rate equation if one remembers that V f = kcat e. If this is so then partial differentiation
with respect to e and scaling with e/v gives
εve = 1 (4.9)
4.1.2 Global coefficients: Response and control coefficients

Let us now perform another thought experiment, this time not with an isolated enzyme, but with
the intact system in steady state. We already know that this steady state is completely determined
by the parameters of the system. Changing any parameter will affect the steady-state to some
degree. However, the parameters of interest in metabolic studies, i.e., pathway substrates and
products, external inhibitors and activitors, all exert their effects through specific steps in the
pathways, usually acting on the enzymes that catalyse these steps. Enzyme concentrations are
of course also very important and they certainly affect the steady-state through the steps that
they catalyse.
In the light of this we now ask a fundamental question: How can we describe the sensitivity
of a steady-state variable to a perturbation of the local rate (activity) through a specific step
caused by a change in a parameter that affects this step directly? To make things more real
we shall consider as the steady-state variable a flux J and as the parameter an external effector
X with concentration x that acts only on the enzyme that catalyses step i. Our argument will
hold, however, for any other steady-state variable, e.g. an internal concentration, and any other
parameter that only affects step i.
Now we perform the thought experiment to answer our question. First we make a small
change δx in the parameter X. What happens? Well, a whole chain of things happen. The
change in x causes a change in the rate of step i, which in turn affects the concentrations of its
substrates and products, which then affect the other rates with which these metabolites interact,
and so the effect of the perturbation in x spreads out via changes in the metabolites that link the
reactions in the network. The initial change in the rate of step i therefore reverberates through
the whole system which goes into a transient state. Eventually though, the system will again
settle into a steady state, but this steady state will differ from the original steady state because
one of the parameters has changed. Therefore we should be able to measure a difference δJ
between the old and new steady-state flux-values. Using fractional changes as before (and for
the same reasons), we can summarise by saying that a fractional parameter change δx/x caused
a fractional steady-state flux change δJ/J. Therefore, although the parameter acted locally on
step i, the change we observe is systemic in the sense that the properties of all steps in the system
contributed to the observed change in the steady state.
The ratio of these fractional changes we call a response coefficient, equivalently defined as:
J δJ/J δJ x δ ln J
Rx = = · = (4.10)
δx/x δx J δ ln x
Note, that although the type of mathematical function in this definition is the same as that
in the definition of an elasticity coefficient, the two types of coefficient are conceptually and
operationally quite different, elasticity coefficients being local (isolated) enzymic properties and
response coefficients being systemic (global) properties.
We have, however, still not answered our original question. We now know how to describe the
response of a steady-state variable to a change in a parameter, but we actually want to describe
43
X0
ln J
δ ln J
R J2 = δ ln [E2 ]
1 ln [E2 ]
S1
ln [S1 ]
δ ln [S ]
δ [E2 ]
R S21 = δ ln [E21 ]
2 ln [E2 ]
S2
ln [S2 ]
δ ln [S ]
R S22 = δ ln [E22 ]
ln [E2 ]
3
S3 δ ln [S ]
ln [S3 ]
R S23 = δ ln [E23 ]
ln [E2 ]
4
Figure 4.2: Response coefficients.
the response of a steady-state variable to a change in the rate through a step. Well, seeing that
the change in x initially affected only step i before anything else happened, we could ask by
how much vi changed. This we can already answer: The effect of δx on the local rate v i seen in
isolation is described by the elasticity coefficient
δvi /vi
εvxi = (4.11)
δx/x
so that
δvi δx
= εvxi (4.12)
vi x
Now, if we take the ratio of the systemic effect of δx/x on J and the local effect of δx/x on v i we
get
J
δJ/J δvi /vi R

/ = vxi (4.13)
δx/x δx/x εx
As the value of δx/x is the same in both the numerator and the denominator it cancels from the
expression, leaving
J
δJ/J R
= vxi (4.14)
δvi /vi εx
44
It should be clear that this ratio is the key to the answer to our original question: How can we
describe the sensitivity of a steady-state variable to a perturbation of the local rate (activity)
through a specific step caused by a change in a parameter that affects this step directly? It is this
ratio that we call the control coefficient of step i, more specifically the flux-control coefficient of
step i:
J δJ/J
Cvi = (4.15)
δvi /vi
In general, therefore, the control coefficient of steady-state variable y with respect to the rate
of step i is equivalenty defined as
y δy/y δy vi δ ln y
Ci = = = (4.16)
δvi /vi δvi y δ ln vi
Note that we usually just use a subscript i in the symbol of a concentration control coefficient,
knowing that it refers to v i . If y refers to a steady-state concentration, the control coefficient is
called a concentration-control coefficient.
A very important general relationship arises from eq. 4.14:
y
y Rx y y
Ci = or Rx = Ci εvxi (4.17)
εvxi
where x is any parameter that acts on step i. A response coefficient is therefore always a product
of a control coefficient and an elasticity coefficient. This last relationship is called the combined
response relationship.
Above we used the external concentration x as the parameter. However, the parameter could
be anything that affects step i specifically. Let us now consider the special case where the
parameter is the enzyme concentration ei of step i. The combined response of steady-state
variable y to ei can be written as
y y
Rei = Ci εveii (4.18)
In the section on elasticity coefficients we have shown that the elasticity of a step towards its
own enzyme is usually one, εveii = 1. In this special case
y y
R ei = C i (4.19)
In most of the early papers on control analysis (and even in some recent ones) a control
coefficient is defined not in terms of v i , but in terms of enzyme concentration ei . Eq. 4.19 shows
that in mnay cases the two definitions are equivalent. However, most workers in the field now
accept the more general definition in eq. 4.16. However, although accepting the conceptual
y y
difference between R ei and Ci , many papers still call both of these entities control coefficients.
This is understandable in terms of the nomenclature recommendations made in 1985 [TIBS 10,
16], but it is our hope that response and control coefficients will become formally accepted as
distinctive concepts. Calling two different things by the same name can only cause confusion.
n

y y i
Rp = Ci εvp (4.20)
i=1
45
4.2 Control Properties

Now that we know what control, response and elasticity coefficients are, we come to the bit
that makes metabolic control analysis so powerful: the relationships between the various types
of coefficients. We have already discovered the combined response property that expresses a
response coefficient as a product of a control coefficient and an elasticity coefficient. Here we
discover the summation and connectivity properties by means of thought experiments.
We shall explore the relationships between the various coefficients of control analysis by
analysing a specific metabolic system, shown in Fig. 4.3, which is in a stable steady state.
Xe
εvx1e
εvx10 εvs11 εvs12 εvs22 εvs23 εvx33
X0 E1 S1 E2 S2 E3 X3
v1
εs2
Figure 4.3: A 3-enzyme linear system with a feedback loop from S 2 onto E1 , and an external
effector, Xe , that interacts with E1 .
4.2.1 Summation properties of control coefficients

Thought experiment: What would happen if we simultaneously made the same
fractional change, α, in the local rates of all the steps in the system?, i.e. if
δv1 δv2 δv3

= = =α (4.21)
v1 v2 v3
Answer: The flux, J, must increase fractionally by α, but, since all the rates increased
in the same proportion, the concentrations of the variable metabolites s 1 and s2 remain
unchanged.
The combined effect of all the changes in local rates on the systemic variables J, s1 and s2 can
be written as the sum of the individual effects caused by each change in local rate. For the flux J:
δJ J δv1 J δv2 J δv3

= C1 + C2 + C3 (4.22)
J v1 v2 v3
J J J
α = α(C1 + C2 + C3 ) (4.23)
Therefore
C1J + C2J + C3J = 1 (4.24)
Similarly, for s1
δs1 δv1 δv2 δv3

= Cs11 + Cs21 + Cs31 (4.25)
s1 v1 v2 v3
s1 s1 s1
0 = α(C1 + C2 + C3 ) (4.26)
46
Therefore
Cs11 + Cs21 + Cs31 = 0 (4.27)
Similarly, for s2
Cs12 + Cs22 + Cs32 = 0 (4.28)
These two properties hold in general. The first, namely that for any steady-state flux in the
system Jm the Jm -control coefficients of all the steps of the metabolic system sum to one, is called
the summation property of flux-control coefficients and can be generalized as:
n

J
Ci m = 1 (4.29)
i=1
where n denotes the number of enzymes in the system.

The important conclusion to be reached from the flux-summation property is that enzymes
share the control of a flux.
The second property, namely that with respect to any steady-state concentration s j the s j -
control coefficients of all the steps of the metabolic system sum to zero, is called the summation
property of concentration-control coefficients and can be generalized as:
n
sj

Ci = 0 (4.30)
i=1
From the summation to zero it should be clear that whereas some enzymes act to decrease
metabolite concentrations, others increase them.
These summation equations show how global control properties of metabolic pathways are
related. The summation property for flux-control coefficients played an important role in dis-
pelling the established notion that flux-control resides in one so-called ‘rate-limiting’ enzyme.
On the contrary, as this summation property shows, flux-control can be shared by all enzymes
in the system. Furthermore, as the conditions change the distribution of control between the
enzymes of the system will vary within the confines of the summation property.
4.2.2 Connectivity properties

The combined response property describes the relationship between local and systemic coeffi-
cients for external metabolites. Are there similar relationships between control coefficients and
elasticities towards variable metabolites?
Thought experiment: How can we bring about a small change in the concentration
of a variable metabolite concentration, say δs2 , without changing the flux J or any of
the other metabolite concentrations (here s 1 )?
Answer: The immediate effect of the change δs 2 is to change the local rates of all
the reactions to which it is directly connected, in this case reaction 3 (for which it
is a substrate), reaction 2 (for which it is a product), and reaction 1 (for which it is
an effector). If we could cancel these local rate changes by making compensating
changes in some parameter, such as enzyme concentration, of each of the directly
affected reactions, then the flux through each of these rates would remain constant,
thereby ensuring that no other steady-state concentrations in the system change. So,
47
we make these compensating rate changes by δe 1 , δe2 , and δe3 (any three parameters
each acting specifically on one reaction will do, but enzyme concentration is the most
convenient).
That δe1 compensates for the effect of δs 2 on v1 is expressed by

δv1 δe1 δs2
= εve11 + εvs21 = 0. (4.31)
v1 e1 s2
Similarly,
δv2 δe2 δs2
= εve22 + εvs22 =0 (4.32)
v2 e2 s2
δv3 δe3 δs2
= εve33 + εvs23 =0 (4.33)
v3 e3 s2
It follows that
δe1 δs2 δe1 −εvs21 δs2
εve11 = −εvs21 or = v1 (4.34)
e1 s2 e1 εe1 s2
δe2 δs2 δe2 −εvs 2 δs2
εve22 = −εvs22 or = v22 (4.35)
e2 s2 e2 εe2 s2
δe3 δs2 δe3 −εvs23 δs2
εve33 = −εvs23 or = v3 (4.36)
e3 s2 e3 εe3 s2
Although the system is in a new steady state (some parameters have changed), only the value
of s2 is different; J and s1 are still the same. Therefore
δJ J δe1 J δe2 J δe3
= R e1 + R e2 + R e3 =0 (4.37)
J e1 e2 e3
Substituting the above expressions for δe i /ei yields
J εvs21 δs2 v2
J εs2 δs2
v3
J εs2 δs2
−Re1 − R e2 v2 − R e3 v3 =0 (4.38)
εve11 s2 εe2 s2 εe3 s2
From eq. 3.16 we recognise the response coefficient divided by the elasticity coefficient as a
control coefficient, so that
J δs2 J δs2 J δs2

−C1 εvs21 − C2 εvs22 − C3 εvs23 =0 (4.39)
s2 s2 s2
Dividing by −δs2 /s2 yields
J J J
C1 εvs21 + C2 εvs22 + C3 εvs23 = 0 (4.40)
Similarly, as s1 did not change, it follows that
δs1 δe1 δe2 δe3
= Rs11 + Rs21 + Rs31 =0 (4.41)
s1 e1 e2 e3
so that, by the same manipulations as above
Cs11 εvs21 + Cs21 εvs22 + Cs31 εvs23 = 0 (4.42)
48
However, δs2 /s2 0, so that

δs2 δe1 δe2 δe3
= Rs12 + Rs22 + Rs32 (4.43)
s2 e1 e2 e3
Substitution as above yields
Cs12 εvs21 + Cs22 εvs22 + Cs32 εvs23 = −1 (4.44)
The same thought experiment done for S 1 would yield
J J
C1 εvs11 + C2 εvs12 = 0 (4.45)
Cs11 εvs11 + Cs21 εvs12 = −1 (4.46)
Cs12 εvs11 + Cs22 εvs12 = 0 (4.47)
Note that since S1 does not interact with E 3 in any way, there is no term that contains ε vs13 in any
of these three expressions.
These so-called connectivity properties can be generalized as the
• Connectivity between flux-control coefficients and elasticities

n

J
Ci m εvs ji = 0 (4.48)
i=1
where S j is a variable metabolite pool and Jm a specific system flux.
• Connectivity between concentration-control coefficients and elasticities

n
s

Ci j εvski = −δ jk (4.49)
i=1
where δ jk is the Kronecker delta (δ jk = 1 when j = k; δ jk = 0 when j k).
49
5 Obtaining elasticity coefficients from the
logarithmic form of rate equations
One of the main aims of metabolic control analysis is to understand systemic properties such
as flux and concentration-control in terms of the local properties of the steps in the system.
These local properties are expressed in terms of so-called elasticity coefficients. If we want to
understand how thermodynamic reaction properties such as distance from equilibrium, and
kinetic properties such as degree of saturation and degree of cooperativity affect the value of
an elasticity coefficient, we need an analytical expression for that elasticity coefficient. This
is essential if we are to use metabolic control analysis as an analytical tool for understanding
metabolic behaviour.
In the past, analytical expressions for elasticity coefficients have usually been obtained by par-
tial differentiation and appropriate scaling of the rate equation. This follows from the following
definition for an elasticity coefficient:
∂v S
εvS = (5.1)
∂S v
where the partial derivative ∂v/∂S is multiplied by the scaling factor S/v in which v is replaced
by the rate expression.
In practice, however, this procedure leads to rather tedious algebra. I describe here an
alternative procedure which is considerably simpler and less error-prone. It is based on the
logarithmic definition for an elasticity coefficient
∂ ln v
εvS = (5.2)
∂ ln S
In order to understand this procedure, we should first consider the general structure of rate
equations.
For the general catalysed reaction aA + bB ⇋ cC + dD any physically realisable rate equation
must be the product of a rate constant k, a kinetic term Θ, and a thermodynamic term (also called
a mass-action term):
C c Dd

a b
v= k·Θ· A B − (5.3)
Keq
The rate constant (units concentration.time −1 ) is a function of the forward specificity constant
and the enzyme concentration, while the kinetic term Θ is a function of kinetic constants and
the concentrations of substrates, products, and effectors. Examples are given further on. For an
uncatalysed reaction Θ is of course 1, while k is a (a + b)th-order rate constant.
To obtain analytical expressions for elasticity coefficients the rate equation is written in loga-
rithmic form as a function of logarithmic concentrations using the equality y = eln y :
ec ln C ed ln D

a ln A b ln B
ln v = ln k + ln Θ + ln e e − , (5.4)
Keq
50
and then partially differentiated with respect to one of ln A, ln B, ln C, or ln D. The resulting

expressions for the four elasticity coefficients with respect to substrates A and B and products C
and D are
∂ ln v ∂ ln Θ a · Aa Bb ∂ ln Θ a
εvA = = + = + (5.5)
∂ ln A ∂ ln A C c Dd ∂ ln A Γ
Aa Bb − 1−
Keq Keq
∂ ln v ∂ ln Θ b· Aa B b ∂ ln Θ b
εvB = = + = + (5.6)
∂ ln B ∂ ln B C c Dd ∂ ln B Γ
Aa Bb − 1−
Keq Keq
C c Dd Γ
c· c·
∂ ln v ∂ ln Θ Keq ∂ ln Θ Keq
εvC = = − = − (5.7)
∂ ln C ∂ ln C c
CD d ∂ ln C Γ
Aa Bb − 1−
Keq Keq
C c Dd Γ
d· d·
∂ ln v ∂ ln Θ Keq
∂ ln Θ K eq
εvD = = − = − (5.8)
∂ ln D ∂ ln D c
CD d ∂ ln D Γ
Aa Bb − 1−
Keq Keq
In the right-hand formulation the thermodynamic terms have been written in terms of the
c d
mass-action ratio Γ = CAaDBb . The form of Θ depends on the kinetic mechanism; the form of the
thermodynamic terms is fixed by the reaction stoichiometry: for substrates as in eqs. 5.5 and 5.6
and for products as in eqs. 5.7 and 5.8.
Note that differentiating the logarithmic form of the rate equation in terms of the logarithm of
a substrate or product concentration is very easy. The two main differentiation rules needed are
dy dy
(i) the chain rule, which states that if y = g(u) and u = f (x) then dx = du du
dx , and (ii) the rule for
exponential functions
d
(aebx ) = abebx (5.9)
dx
The following set of examples illustrates the form of the kinetic part of the elasticity coefficient
(e.g., ∂∂ ln Θ
ln A ) for a number of different kinetic mechanisms.
5.1 Example: The reversible Michaelis-Menten equation

The rate equation for a uni-uni reversible Michaelis-Menten catalysis of reaction S ⇋ P is usually
written as:
Vf p
s−
Ks Keq
v= s p (5.10)
1+ +
Ks Kp
where Vf is the limiting forward rate, and Ks and Kp the Michaelis constants for substrate and
product.
51
V kcat
Using the equality Ksf = kf E, where k f = Ks f is the specificity constant and E the enzyme
concentration, the expression splits up into the general form of eq. 5.3:

1 p
v = kf E · s p · s − Keq (5.11)
1+ +
Ks Kp
which, in logarithmic form, becomes
eln s eln p eln p

ln s
ln v = ln kf E − ln 1 + + + ln e − (5.12)
Ks Kp Keq
Partial differentiation with respect to ln s and ln p gives

s
−
∂ ln v Ks s
εvs = = p + p (5.13)
∂ ln s 1 + s + s−
Ks Kp Keq
p p
−
∂ ln v Kp Keq
εvp = = p − p (5.14)
∂ ln p 1 + s + s−
Ks Kp Keq
5.2 Tutorial
1. Consider the elasticity expressions for the reversible Michaelis-Menten equation.
a) What value does Γ/K approach in the equilibrium and far-from-equilibrium states?
b) What values do the thermodynamic terms of ε vs and εvp approach in the equilibrium
and the far-from-equilibrium states?
c) How do the kinetic terms of ε vs and εvp vary between saturation and unsaturation?
2. What are the values for εvs and εvp in the following states
a) Unsaturated, far-from-equilibrium.
b) Saturated, far-from-equilibrium.
c) Near-equilibrium (how does the degree of saturation affect this value?).
3. In the following rate equation the product is a mixed-type inhibitor of an enzyme catalysing
a reversible reaction with Michaelis-Menten kinetics:

Vf 1 p
v= · · s− (5.15)
Ks 1 + s (1 + p ) + p Keq
Ks Kup Kcp
where Vs is the limiting forward rate, Keq the equilibrium constant, Ks the substrate
Michaelis constant, and Kcp and Kup the competitive (specific) and uncompetitive (cat-
alytic) inhibition constants for the product. When K cp → ∞, P is a pure uncompetitive
52
inhibitor, while when Kup → ∞, P is a pure competitive inhibitor. When Kcp = Kup , P is a
non-competitive inhibitor.
Obtain an expression for the elasticity coefficients of this enzyme with respect to S and
P. Convince yourself that kinetic effects of P are only pronounced in conditions far-from-
equilibrium.
How does ε vs change in response to changes in p when the reaction is far-from-equilibrium?
To answer this, obtain the expression for ∂ ln ε vs /∂ ln p, taking into account that far-from-
equilibrium the thermodynamic term in εvs equals 1. What is the difference for pure
competitive, pure uncompetitive, and non-competitive inhibition?
4. Consider an enzyme that responds cooperatively to a change in the concentration of its

substrate S. A rate equation for this reaction could be the Hill equation
Vsh
v=
sh0.5 + sh
where V is the limiting rate, s0.5 the concentration of s that gives 0.5V, and h is the Hill
coefficient.
Obtain an expression for εvs . What limiting value does εvs approach and at which values of
s?
5. Consider an enzyme inhibited by an allosteric effector I. If the inhibition is cooperative the

reaction rate is given by
v0
v= h
i0.5 + ih
where v and v0 are the reaction rates at inhibitor concentrations of 0 and i respectively, i 0.5
is that concentration of i that gives 0.5v0 , and h, the Hill coefficient, is the slope of the curve
of ln(v0 /(v − 1)) versus ln i.
Obtain an expression for εvi . What limiting value does εvi approach and at which values of
i?
53
6 Metabolic Regulation: Supply-Demand
Analysis
6.1 Introduction
How highly would one rate an economic analysis of a factory that ignored the consumer demand
for its products? 1 Ludicrous as it may sound, this is precisely what most metabolic studies of
the past century have been doing. If this seems far-fetched, consider for example that we have
yet to find a textbook analysis of, say, biosynthetic flux to an amino acid that takes into account
the rate of protein synthesis. This state of affairs is perhaps understandable: faced with the
huge complexity of the cellular reaction network the only way to proceed was to chop it up into
manageable parts and study the parts separately in terms of stoichiometric structure, enzymes,
and transporters. However, although all these parts are undoubtedly connected, the current
view of metabolism and its regulation still seems to be that the parts behave the same whether
in isolation or in cellular context. A telling example is the continued insistence of modern
biochemistry textbooks on the purported rate-limiting role of the kinases in glycolysis, despite
clear evidence that over-expression of these (and other) glycolytic enzymes either on their own
or in combination has no effect on the carbon flux in vivo from glucose to ethanol in yeast [1, 2].
Here we outline a quantitative theory called metabolic supply-demand analysis that addresses
this problem by allowing the integration of the different parts of metabolism with each other and
with other intracellular processes. Within this framework the concepts of metabolic regulation
and function acquire a clear and quantitative meaning. In addition, a number of concepts central
to the classical view of metabolic regulation are shown to be fallacious.
6.2 Metabolic regulation, organisation and function

We consider metabolic regulation to be inextricably linked to function: to say a system is
regulated is to mean that its intrinsic properties have been moulded by evolution to fulfil
specific functions [3, 4]. Because mass action is the intrinsic driving force for self-organisation of
reaction networks, we broadly define metabolic regulation as the alteration of reaction properties
to augment or counteract the mass-action trend in a network of reactions [4, 5]. A corollary to
this definition is that regulatory performance should always be measured in terms of a specified
function. Reaction properties can be regulated by altering the concentrations and the catalytic
and binding properties of enzymes; a host of such regulatory mechanisms have evolved [4, 6].
Enzymes lift the metabolic network from the underlying network of thermodynamically feasible
reactions onto a different timescale and therefore act as the primary ‘handles’ through which
evolution can create function.
Central to any understanding of metabolic function is our knowledge of the organisation of
the metabolic network. Its core consists of a catabolic block that provides phosphorylation and
1
This is the text of Hofmeyr, J.-H.S. and Cornish-Bowden, A. (2000) Regulating the cellular economy of supply and
demand. FEBS Lett. 476, 46–51
54
reducing power plus carbon skeletons, a biosynthetic block that makes building blocks for macro-
molecular synthesis, and a ‘growth’ block that makes and maintains the cellular structure and the
gene and enzyme machinery. These blocks are coupled by either one common intermediate (e.g.,
an amino acid or nucleotide) or a pair of common intermediates that form a moiety-conserved
cycle in which the sum of the cycle members remains constant (e.g., NAD(P)H-NAD(P), ATP-
ADP or, in the presence of adenylate kinase, ATP-ADP-AMP [7]). To remind ourselves that the
living process is intrinsically a molecular economy (cf. [8]) we call the producing block in these
linkages the supply and the consuming block the demand (Fig. 6.1). Although in this article
we restrict the discussion to the simplest case (Fig. 6.1) the same general approach applies with
almost no differences to supply-demand systems that involve moiety-conserved cycles [2].
Supply P Demand
Figure 6.1: A metabolic supply-demand system.
The metabolic network is an open system that can exist in either a transient or a steady
state. Although equilibrium is excluded as a possible state for living systems, it is an important
reference state: the distance ρ of any reaction or reaction block from equilibrium, defined as
Γ/Keq , where Γ is the mass-action ratio and Keq is the equilibrium constant, is an important factor
in determining the behaviour of any reaction network.
Metabolic function is a multi-level concept. At the lowest level the function of an enzyme is to
catalyse a reaction. At the level of the integrated system of coupled enzyme-catalysed reactions
its function may be to control a steady-state metabolite concentration. Enzymes are regulated to
perform these higher-level systemic functions, namely: (i) the determination of the steady-state
itself, (ii) control over the steady-state fluxes and intermediate concentrations, (iii) the steady
state response to a perturbation in some system parameter or a fluctuation in some intermediate
concentration (structural and dynamic stability), (iv) the time of transition from one steady state
to another [9], and (v) the dynamic form of the transient or steady states (e.g., point, monotonic,
oscillatory, trigger, chaotic [10, 11]). Here we only consider the first three functions, although
the others are also important in a complete supply-demand analysis.
6.3 Quantitative analysis of supply-demand systems

We now describe a theory that allows a visual and quantitative analysis of how the properties of
the supply and demand blocks determine the behaviour and control of the steady-state flux and
concentration of P. As our main tools we use rate characteristics [4] and control analysis [12, 13].
The graph of combined rate characteristics (Fig. 6.2) is a powerful tool for visualising how
the steady state in a supply-demand system is formed, and how the distribution of flux and
concentration control depends on the properties of the supply and demand blocks. On the
graph the natural logarithms of the supply and demand rates are plotted as a function of the
natural logarithm of the concentration variable that links them. If the supply and demand
were catalysed by single enzymes these curves would represent, for example, the familiar
Michaelis-Menten or Hill responses of a rate with respect to a product or a substrate. In general,
however, the supply and demand are reaction blocks, so that the rate curves actually represent
the variation in the local steady-state fluxes of the isolated supply and demand blocks as they
respond to variation in the concentration of P. The use of logarithmic rather than linear scales
has a number of advantages [4], the most important being that it allows direct comparison of
55
vsupply ∂ ln vsupply
Supply εp =
ln v
∂ ln p
∂ ln vdemand
εvpdemand = ∂ ln p
ln J
Steady state
Demand
ln p̄
ln p
Figure 6.2: The rate characteristics of a supply-demand system plotted in double logarithmic
space.
the magnitude of steady state responses to perturbations at different positions of the rate and
concentration scale.
The intersection of the supply and demand rate characteristic represents the steady state,
which is characterised by a flux, J, and concentration of P, p̄. From the graph it should be clear
that the response in the steady state to small perturbations in the activities of supply or demand
depend completely on the elasticity coefficients, i.e., the slopes of the tangents to the double
logarithmic rate characteristics at the steady-state point.
Supply
3
2
d ln v
d ln J2
1
0
d ln J1
−d ln p̄ d ln p̄
ln v
Demand
ln p
Figure 6.3: How the steady state (0) responds to a small increase d ln v in the activities of either
supply (leading to a new steady state at 1) or demand (leading to a new steady state
at 2) or both supply and demand (leading to a new steady state at 3).
Fig. 6.3 shows how flux and concentration control can be quantified [14]. Consider a small
increase d ln v in the activity of the supply, caused by, say, an increase in the concentrations of
the supply enzymes. The system moves from the original steady state 0 to a new steady state 1;
flux increases by d ln J1 and p̄ by d ln p̄. Similarly, if the demand activity is increased by d ln v
the system moves from steady state 0 to 2 with a flux increase of d ln J 2 and a decrease in p̄ of
d ln p̄. The degrees to which supply and demand control J and p̄ are given by the flux-control
56
coefficients
J d ln J1 J d ln J2
Csupply = , Cdemand = (6.1)
d ln vsupply d ln vdemand
and the concentration-control coefficients
p d ln p p −d ln p
Csupply = , Cdemand = (6.2)
d ln vsupply d ln vdemand
If both supply and demand are both increased by d ln v the system moves to steady state 3 in
which the flux has increased by d ln J2 + d ln J2 = d ln v while p̄ remains unchanged. Using the
definition of control coefficients given in eqns. eq:ccJdef and eq:ccpdef it follows that
J J
Csupply + Cdemand = 1 (6.3)
p p
Csupply + Cdemand = 0 (6.4)
These are specific cases of the so-called summation theorems of control analysis [12].
Furthermore, using the definitions of the elasticities of supply and demand given in Fig. 6.2
the connectivity theorems [12] can also be derived:
J vsupply J v
Csupply εp + Cdemand εpdemand = 0 (6.5)
p vsupply p v
Csupply εp + Cdemand εpdemand = −1 (6.6)
The summation and connectivity theorems provide enough information to express the control
coefficients in terms of elasticities of supply and demand [3]. The flux-control coefficients are
v
J
εpdemand
Csupply = vsupply (6.7)
εvpdemand − εp
and vsupply
J
−εp
Cdemand = v vsupply (6.8)
εpdemand − εp
and the concentration-control coefficients:
p p 1
Csupply = −Cdemand = vsupply (6.9)
εvpdemand − εp
vsupply
Note that εp is typically a negative quantity, i.e., product inhibits supply. The ratio of elastici-
vsupply v
ties determines the distribution of flux-control between supply and demand (if |ε p /εpdemand | >
vsupply
1 the demand has more control over the flux than the supply; if |ε p /εvpdemand | < 1 the demand
has less control over the flux than the supply). With regard to p̄ it is not the distribution of
p p
p̄-control that is of interest (Csupply always being equal to −Cdemand no matter what the values
of the elasticities), but what determines the magnitude of the variation in p̄ (and, therefore, its
v vsupply
homeostatic maintenance): the larger εpdemand − εp , the smaller the absolute values of both
p p
Csupply and Cdemand . This algebraic analysis is clearly illustrated by the different configuration of
rate characteristics around the steady state shown in Fig. 6.4.
57
A. B. C.
s s s
d d
s s s
d d
ln v
d
ln p
Figure 6.4: The effect on the steady-state of varying the demand (upper half of the figure) or
supply (lower half of the figure). The slope of each line is an elasticity of either
supply (s) or demand (d) at the steady state. The dotted lines show a set percentage
increase or decrease in activity. The shaded regions show the magnitude of the
response in the steady-state flux (horizontal) and concentration of P (vertical).
Fig. 6.4A shows a situation where the elasticities of supply and demand are equal, so that
the functions of flux and concentration control are equally distributed: the same percentage
J
change in the activity of either supply or demand causes the same change in the flux (C supply =
J
Cdemand = 0.5). The magnitude of the variation in p̄ is determined to the same degree by supply
and demand.
In Fig. 6.4B the elasticity of demand in decreased to zero (the demand becomes saturated with
P): it is clear that the demand now has complete control over the flux, while the supply has none.
However, the elasticity of supply now completely determines the magnitude of the variation
p p vsupply
in p̄ (Csupply = −Cdemand = −1/εp ). The steeper the slope of the supply characteristic, the
narrower the band of variation in p̄ and, therefore, the better the homeostatic maintenance of
p̄. The opposite would obtain if the supply elasticity were zero whereas the demand elasticity
remained finite: supply would completely control flux, and the elasticity of demand would
completely determine the magnitude of variation in p̄.
In Fig. 6.4C not only is the elasticity of demand zero (as in Fig. 6.4B) but in addition that of
supply is −∞. The homeostatic maintenance of p̄ in the face of changes in the maximal activity of
either supply or demand is now perfect; the only way in which p̄ can change is if the half-limiting
concentration p0.5 of the supply block changes, as in the bottom half of Fig. 6.4C.
Supply-demand analysis therefore shows that the functions of flux and concentration control
are mutually exclusive in the sense that if one block controls the flux it loses any influence over
the magnitude of variation in the linking product p̄: this becomes the sole function of the other
block. This finding has profound consequences for any view of metabolic regulation.
Up to now the analysis has been limited to the response of the steady state to small variations
in the activity of supply or demand without considering either the form of the full rate charac-
teristics or the position of the steady state in relation to equilibrium. We now expand the picture
to obtain an overall view of the limits within which the system can fulfil its functions.
Like any factory, a supply pathway must be able to fulfil two primary functions: to meet
increasing demand for its product at least up to some limit, and to cope with low demand in
such a way that its product and intermediate metabolite concentrations do not tend towards
58
their equilibrium concentrations (most biosynthetic pathways have huge equilibrium constants
so that near-equilibrium conditions would cause fatally high accumulation of supply pathway
intermediates and product [7]). Textbook wisdom has it that allosteric feedback inhibition of
supply by its product is responsible for satisfying demand, while it has little to say about low
demand. What can supply-demand analysis teach us?
ln v
1
2
Demand
Supply
ln p
Figure 6.5: The steady-state behaviour of a supply-demand system with (solid) and without
(dashed) inhibition of supply by its product P. The grey lines represent different
demand activities. The four marked steady states are discussed in the text. The
rate characteristics were generated with Gepasi [15] for the supply-demand system
described in [3] using the reversible Hill [16] and reversible Michaelis-Menten rate
equations with realistic parameter values.
Fig. 6.5 gives a bird’s-eye view of a hypothetical set of supply-demand rate characteristics
spanning the full range of p to its equilibrium value (assuming that the substrate for the supply
pathway is buffered and therefore constant). For the supply to be able meet a specific range
of variation in demand activity it cannot have any flux control in that range. Focussing for
the moment on the solid supply curve, it is clear that only in the shaded band between steady
states 2 and 3 will the supply be able to meet the variation in demand while keeping p̄ reasonably
constant. When demand becomes higher that 2 it loses control over the flux (steady state 1) with
a concomitant sharp decrease in p̄. An increase in the maximal activity of the supply (the
plateau at 1) would extend the range in which the supply can meet the demand. However,
it is also clear that the presence of allosteric feedback inhibition is not a prerequisite for flux
control by demand: in the shaded band on the right demand also controls the flux in the
absence of allosteric feedback (the dashed supply characteristic), and the supply is equally
effective in keeping p̄ homeostatic. The dramatic difference between the two situations is in the
concentration where P is homeostatically maintained: without feedback inhibition it can only be
near equilibrium (with all the accompanying disadvantages), whereas with feedback inhibition
it can be maintained orders of magnitude away from equilibrium (at a concentration around the
p0.5 of the allosteric enzyme). Clearly, therefore, when demand controls flux the functional role
of feedback inhibition is homeostatic maintenance of p̄ at a concentration far from equilibrium.
In general each elasticity coefficient is the sum of a thermodynamic term that depends only
on Γ/Keq and a kinetic term that is determined by the binding properties of the enzyme. The
thermodynamic term in the supply elasticity approaches 0 in conditions far from equilibrium
and −∞ near equilibrium, where it completely swamps the kinetic term, which typically varies
between 0 and the Hill coefficient [4]. Kinetic effects such as allosteric feedback inhibition can
59
therefore only play any regulatory role far from equilibrium where the thermodynamic term in
negligible. This is also shown by the solid curve in Fig. 6.5: there is a lower limit (around 3) to
the range in which p̄ can be kinetically regulated; below this limit p̄ jumps to the region where
the thermodynamic term dominates the supply elasticity.
6.4 Discussion
The central regulatory problem of metabolism is to be able to satisfy a varying demand for its
products from low to high values while maintaining these products within narrow concentration
ranges far from equilibrium. Supply-demand analysis shows that these two functions are
inextricably linked: the more control either block has over flux, the less it determines the
degree of homeostasis and the distance from equilibrium where homeostasis is maintained,
which becomes the function of the other block. A common solution to this design problem in
living cells is that the flux is largely controlled by the demand block, whereas the supply block
determines homeostasis of the linking metabolite. Direct experimental evidence for control by
demand exists (see, for example, [1, 17, 18, 19, 20]), while it can be deduced for many systems
on the basis of known kinetics (in general, for example, aminoacyl-tRNA transferases have K m -
values for their amino-acid substrates at least an order of magnitude lower than the intracellular
concentrations of amino acids, thereby ensuring that protein synthetic demand is saturated,
giving a demand elasticity of zero [21]). By identifying the elasticities of supply and demand
as the keys to a quantitative understanding of the integrated cellular process, supply-demand
analysis provides a framework for further experimentation. A number of experimental strategies
for measuring block elasticities are already available [22, 23].
Supply-demand analysis also has major implications for biotechnology [2, 24] and biomedicine
and drug design [25, 26] because it shows that what were thought to be ‘rate-limiting’ steps
catalysed by allosteric enzymes actually have nothing to do with flux control, but are responsible
for homeostasis of metabolites. It opens a new window on our understanding of metabolic design
and regulation [27].
60
Bibliography
[1] Schaaff, I., Heinisch, J. and Zimmermann, F.K. (1989) Yeast 5, 285–290.
[2] Hofmeyr, J.-H.S. (1997) in: New Beer in an Old Bottle: Eduard Buchner and the Growth of
Biochemical Knowledge (Cornish-Bowden, A., Ed.) pp. 225–242, Universitat de València,
Valencia.
[3] Hofmeyr, J.-H.S. and Cornish-Bowden, A. (1991) Eur. J. Biochem. 200, 223–236.
[4] Hofmeyr, J.-H.S. (1995) J. Bioenerg. Biomembranes 27, 479–490.
[5] Reich, J.G. and Sel’kov, E.E. (1981) Energy metabolism of the cell, Academic Press, London.
[6] Cornish-Bowden, A. (1995) Fundamentals of enzyme kinetics, Portland Press, London, 2nd
edition.
[7] Atkinson, D.E. (1977) Cellular energy metabolism and its regulation, Academic Press, New
York.
[8] Cascante, M. and Martı́, E. (1997) in: New Beer in an Old Bottle: Eduard Buchner and the
Growth of Biochemical Knowledge (Cornish-Bowden, A., Ed.) pp. 199–214, Universitat de
València, Valencia.
[9] Lloréns, M., Nuño, J.C., Rodrı́guez, Y., Melendéz-Hevia, E. and Montero, F. (1999) Biophys.
J. 77, 23–36.
[10] Heinrich, R. and Schuster, S. (1996) The regulation of cellular systems, Chapman & Hall,
New York.
[11] Goldbeter, A. (1996) Biochemical oscillations and cellular rhythms, Cambridge University
Press, Cambridge, UK.
[12] Kacser, H., Burns, J.A. and Fell, D.A. (1995) Biochem. Soc. Trans. 23, 341–366.
[13] Heinrich, R. and Rapoport, T.A. (1974) Eur. J. Biochem. 42, 97–105.
[14] Mazat, J.P., Letellier, T. and Reder, C. (1990) Biomed. Biochim. Acta 49, 801–810.
[15] Mendes, P. (1993) Comp. Appl. Biosci. 9, 563–571.
[16] Hofmeyr, J.-H.S. and Cornish-Bowden, A. (1997) Comp. Appl. Biosci. 13, 377–385.
[17] Scopes, R.K. (1973) Biochem. J. 134, 197–208.
[18] Scopes, R.K. (1974) Biochem. J. 142, 79–86.
[19] Koebmann, B.J., Nilsson, D., Snoep, J.L., Westerhoff, H.V. and Jensen, P.R. (1998) in: Bio-
ThermoKinetics in the post-genomic era (Larsson, C., Påhlman, I.L. and Gustafsson, L.,
Eds.) pp. 205–210, Chalmers Reproservice, Göteborg.
61
[20] Buttgereit, F. and Brand, M.D. (1995) Biochem. J. 312, 163–167.
[21] Hershey, J.W.B. (1987) in: Escherichia coli and Salmonella typhimurium. Cellular and molec-
ular biology (Neidhardt, F.C., Ingraham, J.L., Brooks, K.L., Magasanik, B., Schaechter, M.
and Umbarger, H.E., Eds.) volume 1, pp. 613–647, American Society for Microbiology,
Washington.
[22] Hofmeyr, J.-H.S. and Cornish-Bowden, A. (1996) in: BioThermoKinetics of the Living Cell
(Westerhoff, H.V., Snoep, J.L., Wijker, J.E., Sluse, F.E. and Kholodenko, B.N., Eds.) pp.
155–158, BioThermoKinetics Press, Amsterdam.
[23] Brand, M.D. (1996) J. Theor. Biol. 182, 351–360.
[24] Cornish-Bowden, A., Hofmeyr, J.-H.S. and Cárdenas, M.L. (1995) Bioorg. Chem. 23, 439–449.
[25] Eisenthal, R. and Cornish-Bowden, A. (1998) J. Biol. Chem. 273, 5500–5505.
[26] Cornish-Bowden, A. and Eisenthal, R. (2000) in: Technological and Medical Implications of
Metabolic Control Analysis (Cornish-Bowden, A. and Cárdenas, M.L., Eds.) pp. 165–172,
Kluwer Academic Publishers, Dordrecht.
[27] Hofmeyr, J.-H.S., Olivier, B.G. and Rohwer, J.M. (2000) in: Technological and Medical
Implications of Metabolic Control Analysis (Cornish-Bowden, A. and Cárdenas, M.L., Eds.)
pp. 299–308, Kluwer Academic Publishers, Dordrecht.
62

The Regulatory Design of Metabolism An Exercise in Computational Systems Biology

Uploaded by

Copyright:

Available Formats

The Regulatory Design of Metabolism An Exercise in Computational Systems Biology

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Regulatory Design of Metabolism An Exercise in Computational Systems Biology

Uploaded by

Copyright:

Available Formats

The Regulatory Design of Metabolism

An exercise in computational systems

April 13, 2009

1 The Kinetic Model for a Network of Coupled Reactions 4

2 The kinetics and energetics of chemical reactions 18

5 Obtaining elasticity coefficients from the logarithmic form of rate equations 50

6 Metabolic Regulation: Supply-Demand Analysis 54

1.1 The basic linkages and structures in metabolism

1.2 The kinetic model

Figure 1.4: A metabolite S that is produced and consumed by various reactions

change in the amount of A = −dξ

The individual reaction rates are

dx0 /dt = (−1)v1 + (0)v2 = −v1

1.3 A hierarchy of kinetic models

1.3.1 The mass-action form of the kinetic model

square brackets to depict concentrations):

d[X0 ]/dt = −v1a (1.14)

v1a = k1a [X0 ][E1 ] − k−1a [E1 X0 ] (1.23)

1.3.2 The rate law form of the kinetic model

We shall mostly use the rate-law approach in our modelling exercises.

1.3.3 The power-law form of the kinetic model

1.4 Metabolic states

1.4.2 Steady state

1.4.3 Transient state

1.5 The kinetic model in steady state

1.6 Time hierarchy in metabolism

1.7 Parameters and variables

• Functions of metabolite concentrations, e.g. Gibbs-energy changes, chemical and mem-

1. an association between two molecules to form a non-covalently bound complex,

2. a dissociation of a complex into two molecules,

2.1 Reaction rates—how fast?

v∝a and v∝b (2.5)

2.1.1 Reaction order and sensitivity

2.2 The equilibrium constant—how far?

A+B→C with rate equation vf = kf ab (2.12)

and the reverse reaction

C→A+B with rate equation vr = kr c (2.13)

2.2.1 The mass-action ratio and the distance from equilibrium

• ∆G = 0 — the process is at equilibrium.

The above can be summarised in the following important statement:

The change in Gibbs energy of a process at constant pressure and temperature is an

Remember that when the reaction is in equilibrium,

Hence, it follows that

∆G◦ = −RT ln Keq (2.24)

We can now write Eq. 2.21 as

Only Equil. Only

Relative Gibbs energy (J)

composition of the reaction mixture.

2.4 Coupled reactions

where a, b, and c are equilibrium concentrations.

∆G◦1 = −RT ln Keq1 (2.31)

−RT ln Keq12 = −RT ln Keq1 − RT ln Keq2 (2.34)

2.5 Kinetic and energetic aspects of reaction rate

3.1 The reversible Michaelis-Menten equation

Vr = k−2 ep = k−2 e0 (3.3)

where e f denotes the concentration of free enzyme E, bound neither to S nor to P. As

3. The net rate of the catalysed reaction is given by

Substituting es and ep from Eqs. 3.6 and 3.7,