File: /General/MLAB-Text/Papers/enzyme/enzyme.

tex

Chemical Kinetics: Simple Enzyme Reactions

Gary D. Knott, Ph.D.
Civilized Software, Inc.
12109 Heritage Park Circle
Silver Spring, MD 20906 USA
Tel. (301) 962-3711
Email: csi@civilized.com
URL: http://www.civilized.com
Let S be a substrate material which is converted by the addition of an
enzyme, E, into a molecular complex C which dissociates to yield a product
P . We have:
K1
K
S+E*
) C 3 E + P.
K2

This is, of course, an idealization of a more complex schema, since, in

fact, even when S and P have the same molecular weight, energy must be
used to, at least momentarily, change and restore the conformation of the
enzyme E. But for many situations, our simple model will suffice.
Let S(t), E(t), C(t), and P (t) be the amounts of S, E, C, and P respectively at time t. Then we have:
dC/dt(t) = K1 S(t)E(t) K2 C(t) K3 C(t),
dP/dt(t) = K3 C(t),

C(0) = 0,

P (0) = 0,

S(t) = S0 C(t) P (t),

with

E(t) = E0 C(t),

E0 = E(0).

with

S0 = S(0),

Typically the reaction C P + E is much slower than the complexformation reaction S + E *

) C. Thus after the initial rise of C within time
t0 , we have dC/dt 0, so that then
C(t) K1 S(t)(E0 C(t))/(K2 + K3),

or

C(t) E0 S(t)/[(K2 + K3 )/K1 + S(t)],

for t > t0 .

Thus, from the equation for dP/dt, we have the so-called MichealisMenten equation:
dP/dt(t) K3 E0 S(t)/((K2 + K3 )/K1 + S(t)),
1

for t > t0 .

Now the slope dP/dt(t) is measurable by fitting a straight-line to a segment of the kinetic curve for P (t) with t0 < t te , where te is the time
such that S(te ) 0, e.g., when almost all the substrate material has been
consumed. Note S(t) S0 for t0 < t te if S0 is large enough, so one can
fit the Michealis-Menten equation to the single data point (S0 , dP/dt) to
try to obtain K3 and Km = (K2 + K3 )/K1 , the so-called Michealis-Menten
constant.
The parameter K3 may be independently resolved by observing that
when S0 is very large we have S(t)/(Km + S(t)) 1 for t0 < t te , so
dP/dt(t) K3 E0 . Thus the maximum rate of formation of P for a fixed E0
value is K3 E0 . Let K3 E0 = Vm . Vm can be determined by measuring the
slope, dP/dt(t), of the linear region of the kinetic curve obtained when S 0
is very large. Now, the Michealis-Menten equation becomes:
dP/dt(t)/Vm S(t)/(Km + S(t))

for t0 < t te .

Note dP/dt(t)/Vm is the relative rate of formation of P , i.e., the proportion of the maximum rate which is achieved. Hence, having measured Vm ,
one can then measure dP/dt(t) for a relatively-large amount of S, and then
obtain Km from the Michealis-Menten equation as S0 Vm /(dP/dt(t)) S0 .
The Michealis-Menten constant is the amount of substrate which will
yield a product formation rate of Vm /2. It thus is the point at which the
formation of product becomes increasingly sensitive to a decreasing amount
of substrate. The activity of the enzyme for a given amount of substrate
is determined directly as dP/dt(t) computed from the Michealis-Menten
equation for a given Km .
The Lineweaver-Burke form of the Michealis-Menten equation is often
used because of its linear form. It is:
1/(dP/dt(t)) (Km /Vm )(1/S(t)) + (1/Vm ).
The Eadie and Dixon form is also often used. It is:
S(t)/(dP/dt(t)) S(t)/Vm + Km /Vm .
Actually, as noted above, the reaction S + E *
) C P + E is a fiction.
It is commonly used to approximate the situation:
K1

K3

K5

K2

K4

K6

S+E*
)C *
)D*
)P +F
U +F *
)B*
)A*
)V +E
2

where C is ES-complex, and D is EP -complex and E and F are co-factors.

Typically K4 is negligible, but as the amount of P increases, it may block an
appreciable amount of enzyme if K6 is not nearly zero, and this can render
the Michealis-Menten equation useless.
The following is an MLAB tutorial sequence for studying the MichaelisMenten model relative to the simplified kinetic model.
First define the kinetic model by typing:

*FUNCTION C DIFF T(T) = K1*(S0-C-P)*(E0-C)-(K2+K3)*C

*FUNCTION P DIFF T(T) = K3*C
*INITIAL C(0)=0
*INITIAL P(0)=0
*S0 = 10; E0 = 1
*K1 = .2; K2 = .025; K3 = .025

Thus we have assumed the true situation is:

.2

.025
S+E *
) C P +E
.025

starting with 10 moles of substrate and 1 mole of enzyme. We may look at

the kinetic behavior of this system over 900 seconds by typing:

* Q = INTEGRATE(P DIFF T, C DIFF T, 0:900:5)

* TYPE Q ROW 1:180:10

The first column of Q is time, t, the second is P (t), the third is dP/dt(t),
the fourth is C(t), and the fifth is dC/dt(t). We can look at the graph of P
vs. t by typing:

* DRAW Q COL 1:2, LINETYPE dashed

* VIEW
3

Now, let us generate some laboratory data about our reaction. We

shall use the MLAB normal random number generator to generate normally
distributed random numbers. Type:

* M = Q COL 1:2 ROW 2:162:5

* E = (NORMRAN ON 0^^NROWS(M))/4
* TYPE E

E is a vector of normal errors. Now type:

* M COL 2 = (M COL 2) + E
* DRAW M, POINTTYPE triangle LINETYPE none
* view
4

M is a matrix of P vs. t points with error, as might have been measured

in an actual laboratory situation. Now, let us guess K1 , K2 , and K3 and
try to determine them as functions of M . Type:

*
*
*
*

K1 = .3; K2 = .01; K3 = .02

CONSTRAINTS CX={K1>K2, K2>0, K3>0}
METHOD = GEAR; ERRFAC = .002; MAXITER = 12
FIT(K1,K2,K3), P to M, CONSTRAINTS CX

The control variables METHOD and ERRFAC are set based on prior experience; this problem is stiff and runs slowly! Our curve-fit predicts that
K1 , K2 , and K3 are 2.4295, .7975, and .025296 respectively, and resets them
accordingly. Note K1 and K2 are not even close to .2 and .025, but K3 is
approximately correct. We may observe the graph of this fit by typing:

* Q1 = INTEGRATE(P DIFF T, C DIFF T, 0:900:5)

* DRAW Q1 COL 1:2 color green
* VIEW
5

Q1 is now a matrix of kinetic concentration and velocity curves for

our reaction as determined by curve-fitting. Let us discard our picture by
typing:

* DELETE W

Now, let us analyze the same reaction using the Michaelis-Menten model.
Type:

*
*
*
*
*
*
*

K1 = .2; K2 = .025; K3 = K2
FUNCTION MM(S)=VM*S/(KM+S)
KM = (K2+K3)/K1; VM = K3*E0
DRAW Q COL (1,3)
SM = S0-(Q COL 2)-(Q COL 4)
DRAW (Q COL 1)&(MM ON SM), LINETYPE dashed
view
6

Recall Q is the matrix of true curves corresponding to K1 = .2, K2 =

.025, and K3 = .025. The matrix SM is computed as the amounts of S at
time t = 0, 5, 10, . . . , 900. The curves we see are the rate of change, dP/dt
vs. t and its Michaelis-Menten approximation.
Now, let us generate two runs of laboratory data using the error vector
E and obtain the two constants Vm and Km . Type:

*
*
*
*
*
*
*
*
*
*
*
*
*

DELETE W
S0 = 500
Z = INTEGRATE (P DIFF T, C DIFF T, 100:300:10) COL 1:2
DRAW Z
Z COL 2 = (Z COL 2)+(NORMRAN ON 0^^NROWS(Z))/4
DRAW Z, LINETYPE none, POINTTYPE crosspt
FUNCTION Y(T)=A*T+B
CONSTRAINTS QS = {A > 0, B > 0}
A = 1; B = 1
FIT(A,B), Y TO Z, CONSTRAINTS QS
VM =A
DRAW POINTS (Y, 50:350!2), LINETYPE dashed
VIEW
7

We have generated a straight-line segment of the P vs. t curve for S 0 =

500, drawn it, added some noise, shown the simulated points obtained, fit
a straight line to these points, set Vm as the slope of this line, and drawn
the straight-line fit.
Now, we proceed in the same manner to do another experiment to
help compute Km . Type:

*
*
*
*
*
*
*

DELETE W
S0 = 10
Z = INTEGRATE(P DIFF T, C DIFF T,100:300:10) COL 1:2
Z COL 2 = (Z COL 2)+(NORMRAN ON 0^^NROWS(Z))/4
FIT(A,B),Y TO Z, CONSTRAINTS QS
KM = VM*S0/A-S0\
TYPE VM,KM

Now, Vm and Km are computed. Let us look at the result. Type:

*
*
*
*

DRAW Q COL (1,3)

DRAW (Q COL 1)&(MM ON SM) LINETYPE dotted color red
DRAW Q1 COL (1,3) LINETYPE DASHED color green
VIEW

Note the dP/dt curve predicted from the kinetic differential equation
model is much better than the consistent underestimate predicted by the
Michaelis-Menten model.
There is another approach to estimating the Michaelis-Menten constants,
Vm and Km , based on the intersections of various linear plots. This scheme is
due to R. Eisenthal and A. Cornish-Bowden (Biochemistry Journal, Vol. 139,
pp. 715:730). It is robust and, at the cost of more experiments, allows a confidence region for Vm and Km to be obtained, without the usual restrictive
assumptions. Unfortunately, it often produces poor extimates of Vm and
Km .
Given observations (S0i , Hi ) of substrate concentrations and corresponding product-formation velocities (obtained by linear-regression), we can construct lines defined by Vm /Hi + Km /S0i = 1, which may be plotted in Km ,
Vm space. The line { (Km , Vm ) | Vm /Hi + Km /S0i = 1 } is the locus of
all (Km , Vm ) pairs which could produce the observation (S0i , Hi ). Each
of the (Km , Vm ) points obtained by the intersections of all pairs of these
lines is an estimate of the true Km , Vm values. The arithmetic median
of the Km -estimates is the Eisenthal-Cornish-Bowden estimate of Km , and
the arithmetic median of the Vm -estimates is the Eisenthal-Cornish-Bowden
estimate of Vm .
We shall simulate ten experiments for S0 = 50 : 500 : 50 and compute
the Michaelis-Menten velocity curve based on Km and Vm as estimated by
the Eisenthal-Cornish-Bowden procedure.
* FUNCTION VMF(I,J) = (SV[I]-SV[J])/(SV[I]/VV[I]-SV[J]/VV[J])
* FUNCTION KMF(I,J) = (VV[J]-VV[I])/(VV[I]/SV[I]-VV[J]/SV[J])
* FOR I = 1:10 DO \
{S0 = 50*I;
9

Z = INTEGRATE(P DIFF T, C DIFF T, 100:300:10) COL 1:2;

Z COL 2 = (Z COL 2) + (NORMRAN ON 0^^NROWS(Z))/4;
LSQRPT = 8;
FIT(A,B),Y to Z, CONSTRAINTS QS;
SV[I] = S0; VV[I] = A;
};
* D = 1:9^^9
* D = COMPRESS((LIST(D)&LIST(D))*LIST(D<=D))
* D COL 1 = (D COL 1) +1
* VM = MEDIAN(VMF ON D)
* KM =MEDIAN(KMF ON D)
* TYPE VM,KM
* DELETE W
* DRAW Q COL (1,3)
* DRAW (Q COL 1)&(MM ON SM),LINETYPE DASHED
* VIEW

Overall the best approach to enzyme kinetics is to try to measure enough

points on the kinetic curves of several species, so that direct curve-fitting
using the appropriate differential equation model can permit the association
and dissociation constants to be found. The Michealis-Menten equation
is used only due to the difficulty of obtaining data other than P (t) for
t0 < t < t2 . Even then, concurrent use of the kinetic model is useful. An
excellent source for mathematical models in enzyme kinetics is: Enzyme
Kinetics by Kent Plowman, published by McGraw-Hill. Another is Enzyme
Kinetics by Irwen Segal, published by Wiley.

10