Dynamics at the Horsetooth Volume 2A, 2010, Focused Issue: Asymptotics and Perturbations

A Mathematical Model for Enzyme Kinetics: Multiple Timescales

Analysis
Roberto Muñoz-Alicea
Department of Mathematics
Colorado State University [email protected]

Report submitted to Prof. J. Oprea for Math 676, Fall 2010

Abstract. We summarize some of the work on enzyme kinetics presented by J. D.

Murray in his Mathematical Biology I book. We study a basic model for enzyme-
substrate reaction. Through several assumptions and via nondimensionalization, we
simplify the equations. Then we use a multiple timescales method to derive asymptotic
solutions to the model.

Keywords: enzyme, substrate, multiple timescales

1 Introduction
An enzyme is “a protein molecule that catalyzes a biochemical reaction...” [1]. These molecules
regulate many biological processes and may act as activators or inhibitors. Enzymes act on
substances called substrates to create a product. An example of an enzyme is the viral enzyme
reverse transcriptase. Retroviruses use this enzyme to make a copy of DNA from their RNA [2].
Our goal here is to study a mathematical model of enzyme-substrate reactions in order to better
understand them.

2 Michaelis and Menten’s Model

A basic enzyme reaction model by Michaelis and Menten can be represented with the following
diagram:

k1 k2
S + E
SE → P + E (1)
k−1

where S is the substrate, E is the enzyme, SE is the substrate-enzyme complex, P is the product,
and k1 , k−1 , and k2 are constants [3]. This diagram states that one molecule of the enzyme
combines with one molecule of the substrate to form a molecule of the substrate-enzyme complex.
The substrate-enzyme complex could decompose back into enzyme and substrate or may give rise
to one molecule of the product and one molecule of the enzyme. Let s = [S], e = [E], c = [SE],
A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

and p = [P ], where [ ] denotes the concentration of a substance. The Law of Mass Action states
that “the rate of a reaction is proportional to the product of the concentration of the reactants”
[3]. We use this law to derive the following equations for the enzyme reaction described above:

ds
= −k1 es + k−1 c,
dt
de
= −k1 es + (k−1 + k2 )c,
dt
dc (2)
= k1 es − (k−1 + k2 )c,
dt
dp
= k2 c,
dt
s(0) = s0 , e(0) = e0 , c(0) = 0, p(0) = 0.

where k−1 , k1 , k2 , s0 , and e0 are positive constants [3].

Notice that
dc de
+ = 0 ⇒ c(t) + e(t) = e0 .
dt dt

Hence, we can write

e(t) = e0 − c(t). (3)

Moreover,
∫ t
dp
= k2 c ⇒ p(t) = k2 c(τ )dτ.
dt 0

Thus, we only need the equations for s and c. Plugging equation (3) into the first three equations
of system (2) we obtain the following reduced system

ds
= −k1 e0 s + (k1 s + k−1 )c,
dt
dc (4)
= k1 e0 s − (k1 s + k−1 + k2 )c,
dt
s(0) = s0 , c(0) = 0.

For enzyme-substrate reactions, it is usually assumed that the initial stage of enzyme-substrate
complex formation occurs very fast, and subsequently the reaction goes to a quasi-equilibrium or
dt ≈ 0) [3]. In this
quasi-steady state, in which the concentration of the complex barely changes ( dc
case, solving the second equation of system (4) for c gives

Dynamics at the Horsetooth 2 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

k1 e0 s(t)
c(t) =
k1 s(t) + k−1 + k2
e0 s(t)
= (5)
s(t) + k−1k+k
1
2

e0 s(t)
=
s(t) + Km

where Km denotes the Michaeli’s constant.

Plugging the above result for c(t) into the first equation in system (4) we obtain the following:

ds e0 s(t)
= −k1 e0 s(t) + (k1 s(t) + k−1 )
dt s(t) + Km
−k1 e0 s(t)(s(t) + Km ) + (k1 s(t) + k−1 )e0 s(t)
=
s(t) + Km
(6)
(k−1 − k1 Km )e0 s(t)
=
s(t) + Km
−k2 e0 s(t)
=
s(t) + Km

Now, we can solve for s(t) by using separation of variables and integrating both sides of equation
(6) from 0 to t to obtain

s(t) + Km ln(s(t)) = s0 + Km ln(s0 ) − k2 e0 t. (7)

Notice that the solution we obtained here does not satisfy the initial conditions in system (4); for,
if c(0) = 0, then either s0 = s(0) = 0 or e0 = 0, by equation (5). This result would imply that
either there is no substrate or no enzyme (or neither) in the beginning of the process. However,
this quasi-steady state solution might be “a reasonable approximation for most of the time” [3].

3 Multiple Timescales Analysis

There are two timescales in the enzyme reaction:

dt ≈ 0 and dt ̸= 0, under the usual

1. First case (transitional stage): near t = 0, where ds dc

assumption that the amount of enzyme present is much smaller than that of the substrate.

2. Second case: t ≥ δ > 0, for some constant δ, where the substrate’s concentration changes
dt ̸= 0) and the substrate-enzyme complex’s concentration is almost at
significantly ( ds
equilibrium ( dt ≈ 0). The quasi-steady state approximation, equations (5) and (7), applies
dc

in this case [3].

Dynamics at the Horsetooth 3 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

3.1 Nondimensionalization
As done by Murray in [3], we use the substitutions

s(t) c(t)
τ = k1 e0 t, u(τ ) = , v(τ ) =
s0 e0
(8)
k2 k−1 + k2 Km e0
λ= , K= = , ϵ=
k1 s0 k1 s0 s0 s0
to rewrite system (4) in the form

du dv
= −u + (u + K − λ)v, ϵ = u − (u + K)v
dτ dτ (9)
u(0) = 1, v(0) = 0.

We will focus our discussion on this non-dimensional system. The equations in system (9) are not
easy to solve analytically. In our analysis, we must keep in mind that ϵ is a very small, positive
dv
parameter. Notice that ϵ is multiplied by dτ . As we saw before, we cannot expect to find a
uniformly valid approximate solution of the system by simply setting ϵ = 0, because that would
reduce the order of the equation. This is, in fact, what we did before when we derived the
approximate solution given by equations (5) and (7), which do not satisfy the initial conditions in
system (4). This situation is characteristic of a singular perturbation problem.

3.2 Outer and Inner Solutions

To proceed with our analysis, let us suppose that
∑ ∑
u(τ ; ϵ) = ϵn un (τ ), v(τ ; ϵ) = ϵn vn (τ ). (10)
n=0 n=0

We are interested in the case where 0 < ϵ = e0 /s0 << 1 (ϵ → 0), since this is the usual
assumption e0 << s0 , i.e., that the initial concentration of enzyme is much smaller than the
initial concentration of the substrate. Substituting equations (10) into system (9) gives a sequence
of differential equations for un (τ ) and vn (τ ).

The O(1) equations are

du0
= −u0 + (u0 + K − λ)v0 , 0 = u0 − (u0 + K)v0 ,
dτ (11)
u0 (0) = 1, v0 (0) = 0.

Solving the second equation in system (11) for v0 we get

u0 (τ )
v0 (τ ) = (12)
u0 (τ ) + K
and then
du0 u0 u0
= −u0 + (u0 + K − λ) = −λ ,
dτ u0 + K u0 + K

Dynamics at the Horsetooth 4 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

which has the same form as equation (6) and, thus, has solution

u0 (τ ) + Kln(u0 (τ )) = u0 (0) + Kln(u0 (0)) − λτ = 1 − λτ. (13)

1
As before, we see that v0 (0) = 1+K ̸= 0, so this solution does not work for τ = 0. In other words,
the approximate solution found here is not uniformly valid for all τ ≥ 0. In our case, we say that
there is a boundary layer in the immediate neighborhood of τ = 0.
dv
In deriving system (11) we neglected the term ϵ dτ . We must then include this term in our
analysis, since without it the initial condition on v0 (τ ) cannot be satisfied. Put
τ
σ= , u(τ ; ϵ) = U (σ; ϵ), v(τ ; ϵ) = V (σ; ϵ). (14)
ϵ
This transformation to a new timescale allows us to “magnify” the time neighborhood around
τ = 0, since, as ϵ approaches 0, for any fixed value of τ with 0 < τ << 1, σ becomes much larger
than 1. With the new transformation (14), system (9) becomes

dU dV
= −ϵU + ϵ(U + K − λ)V, = U − (U + K)V,
dσ dσ (15)
U (0) = 1. V (0) = 0.
Write ∑ ∑
U (σ; ϵ) = ϵn Un (σ), V (σ; ϵ) = ϵn Vn (σ). (16)
n=0 n=0

Plugging equations (16) into system (15) and equating powers of ϵ gives, for n = 0,

dU0 dV0
= 0, = U0 − (U0 + K)V0 ,
dσ dσ (17)
U0 (0) = 1, V0 (0) = 0.
System (17) has solution
1
U0 (σ) = 1, V0 (σ) = (1 − exp[−(1 + K)σ]). (18)
1+K
This solution works for σ = 0 and, hence, τ = 0, but cannot be expected to work for all τ ≥ 0. If
it did, that would imply that u(τ ) = s(t)
s0 ≈ 1 for all t ≥ 0 (since the contribution from the higher
order terms in the expansion for u(τ ) for small ϵ are negligible [3]). In other words, the
concentration of substrate would remain virtually constant throughout the enzyme reaction,
which contradicts our assumption that the substrate’s concentration changes significantly after
the transitional stage. We call (18) the inner solution for u and v, which is valid for 0 ≤ τ << 1.
The solution given by equations (12) and (13) is called the outer solution, which is valid for
values of τ not close to 0 [3].

We can continue this way, by using equations (9), (10), (15), and (16) to derive the O(ϵ), O(ϵ2 ),
... equations for the outer and inner solutions. However, as stated above, since we are interested
here in the case when 0 < ϵ << 1, the O(1) equations will suffice.

The matching principle of inner and outer solutions requires that, to all orders of ϵ,

lim [U (σ; ϵ), V (σ; ϵ)] = lim [u(τ ; ϵ), v(τ ; ϵ)] [3]. (19)
σ→∞ τ →0

Dynamics at the Horsetooth 5 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

From equations (13) and (18) we see that

lim U (σ) = 1 = lim u0 (τ )

σ→∞ τ →0

and
1
lim V (σ) = = lim v0 (τ ).
σ→∞ 1+K τ →0

Hence, our solution satisfies the matching principle of inner and outer solutions for the equations
of O(1). Thus, we obtain a uniformly valid asymptotic solution, to O(1), of system (9) for
0 < ϵ << 1 [3].

The overall asymptotic solution, which is valid for all τ ≥ 0 (t ≥ 0) can be summarized as follows:

 u(τ ; ϵ) = u0 (τ ) + O(ϵ), u0 (τ ) + K ln(u0 (τ )) = 1 − λτ,

(20)

 v(τ ; ϵ) = v0 (τ ) + O(ϵ), v0 (τ ) =
u0 (τ )
, 0 < ϵ << τ
u0 (τ ) + K


 u(τ ; ϵ) = U0 (τ ) + O(ϵ), U0 = 1,
(21)

 v(τ ; ϵ) = V0 (σ) + O(ϵ), V0 (σ) = 1 τ
(1 − exp[−(1 + K) ]), 0 < τ << 1.
1+K ϵ

4 Numerical and Asymptotic Solutions

The following graphs of numerical and asymptotic solutions of system 9 were generated using
Matlab. The numerical solutions were found using the ode45 routine. Parameters used were:
ϵ = 0.1, 0.001, s0 = 10, e0 = 1, k−1 = 1, k1 = 2, and k2 = 3.

Notice that the asymptotic solutions for v tend to be closer to its numerical solutions in
comparison to the asymptotic and numerical solutions for u. Even when ϵ = 0.1, the asymptotic
inner and outer solutions are very close to the numerical solutions for v. For both functions,
though, as ϵ approaches zero, the asymptotic outer solutions and the numerical solutions become
almost identical.

Numerical and Asymptotic Approximations for u: epsilon = 0.1

1.05
numerical u
outter u
inner u
1

0.95
y

0.9

0.85

0.8
0 0.2 0.4 0.6 0.8 1
x

Dynamics at the Horsetooth 6 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

Numerical and Asymptotic Approximations for v: epsilon = 0.1

1

0.9

0.8
numerical v
0.7 outter v
inner v
0.6

0.5

y
0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
x

Numerical and Asymptotic Approximations for u: epsilon = 0.001

1.04
numerical u
1.02 outter u
inner u
1

0.98

0.96
y

0.94

0.92

0.9

0.88

0.86
0 0.2 0.4 0.6 0.8 1
x

Numerical and Asymptotic Approximations for v: epsilon = 0.001

1

0.9

0.8
numerical v
0.7 outter v
inner v
0.6

0.5
y

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
x

5 Summary and Conclusions

We studied a system of differential equations for enzyme-substrate reactions. We used
nondimensionalizations and asymptotic expansions to find approximate solutions to the system.
The approximate solutions were derived using a singular perturbation method utilizing two
different timescales. The inner solution is valid in the immediate neighborhood of the initial time
τ = 0, corresponding to the initial stage of the enzyme reaction, when there is little change in the
substrate’s concentration. The outer solution is valid for larger values of τ > 0 (not in the
immediate neighborhood of τ = 0), when the substrate’s concentration changes significantly.

Dynamics at the Horsetooth 7 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

In addition to the asymptotic solutions, we also computed numerical solutions for our enzyme
reaction model equations. We compared both types of solutions graphically and observed that the
solutions get closer to each other as ϵ → 0. This property of the solutions suggests that the
perturbation method of multiple timescales gives a very accurate approximation to the solution of
our model when the general assumption that the initial concentration of enzyme is much smaller
than the initial concentration of substrate.

Dynamics at the Horsetooth 8 Vol. 2A, 2010

A Mathematical Model for Enzyme Kinetics: Multiple Timescales Analysis Roberto Muñoz-Alicea

References
[1] W.G. Hale, and J.P. Demeler, The HarperCollins Dictionary of Biology, HarperCollins, New
York, N.Y. (1991).

[2] P.H. Raven, G.B. Johnson, J.B. Losos, K.A. Mason, and S.R. Singer, Biology, 8th edition,
McGraw-Hill, New York, N.Y. (2008), p. 279.

[3] J.D. Murray, Mathematical Biology I: An Introduction, 3rd edition, Springer, New York, N.Y.
(2002), pp. 175-188.

Dynamics at the Horsetooth 9 Vol. 2A, 2010