Biochemical Engineering Journal 53 (2011) 234–238

Contents lists available at ScienceDirect

Biochemical Engineering Journal

journal homepage: www.elsevier.com/locate/bej

Short communication

Evaluation of enzyme kinetic parameters using explicit analytic approximations

to the solution of the Michaelis–Menten equation
Marko Goličnik ∗
Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia

a r t i c l e i n f o a b s t r a c t

Article history: The time-course of product accumulation during an enzyme-catalyzed reaction that conforms to the
Received 14 September 2010 Michaelis–Menten rate equation is expressed by an explicit closed-form equation in terms of the Lambert
Received in revised form 5 October 2010 W(x) function. Unfortunately, the use of direct solution of the Michaelis–Menten equation is limited,
Accepted 16 October 2010
because the W(x) function is not widely available in curve-fitting software. The present commentary
suggests an alternative to surmount this difficulty. The Lambert W(x) function can be approximated in
terms of the elementary mathematical functions that enable the fitting of particular equations on time-
Keywords:
course data of the Michaelis–Menten enzyme reaction by any nonlinear regression computer program.
Michaelis–Menten equation
Enzyme-kinetics analysis
Three different demonstrated approximations of the W(x) with relatively high accuracies are shown here
Progress curve to be appropriate for use when progress curves are analyzed by the direct solution of the integrated
Nonlinear regression Michaelis–Menten equation. The elementary and precise nature of these approximations makes them
Lambert W function the most user-friendly open candidates for simple, but operative, kinetic parameter estimations from
experimental time-course data.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction The ideal simplification of kinetics parameter evaluation from

progress-curve data can be carried out when algebraic integration
Characterization of enzymes according to details of their kinetic of the rate equation results in an explicit mathematical equation
parameters is crucial to several research fields, including bio- that describes the dynamics of the model system of the reaction.
chemistry, biotechnology, pharmacy and medicine [1–3]. Enzyme Fitting explicit mathematical equations to data using nonlinear
activities are typically characterized in terms of initial rates that regression can be carried out with any spreadsheet curve-fitting
are determined at various substrate concentrations. However, software package that is user-friendly (e.g. GraphPad Prism), and
analyses of complete progress curves over time can be advanta- therefore this is more appealing to a wider group of life-science
geous, because the time-courses contain additional information researchers. Schnell and Mendoza [5] defined a truly explicit solu-
relating to the properties of the enzyme. The important point tion of the Michaelis–Menten equation, which has since been used
about monitoring the progress of a reaction in this way is that to investigate enzyme kinetics [6–8] and pharmacodynamics [9].
in any single experiment, kinetics data can be obtained not only The solution is expressed in terms of the Lambert W(x) function
at the initial concentrations of the substrate, but also at every [10]. However, most of the available curve-fitting packages are not
concentration between this value and that at the end of the set up to handle the equations that involve the W(x) function, and
reaction. Therefore, time-course analyses of reactions provide hence they are of little use for direct progress-curve analysis.
the same information as that obtained by classical initial-rate Therefore, the objective of this study was to investigate explicit
approaches, but with only a fraction of the number of separate mathematical equations that will enable easy, yet accurate, analy-
experiments [4]. Despite the obvious advantages of progress-curve ses of progress-curve data by any nonlinear regression curve-fitting
analyses, one of the major barriers to its wide applications is program. The Lambert W(x) function was approximated using dif-
that it can be algebraically complex or computationally intensive ferent composites of known elementary mathematical functions
to extract the full information from progress-curve experiments [11,12], whereby no numerical procedures were required for cal-
[4]. culating the theoretical progress curves. The present study briefly
demonstrates and tests the fitting of direct solution of the inte-
grated Michaelis–Menten equation expressed by W(x) and three
different approximations to the experimental data of single time
∗ Tel.: +386 1 5437669; fax: +386 1 5437641. course and multiple progress curves from two different enzyme
E-mail address: [email protected] reactions, respectively.

1369-703X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.bej.2010.10.012
 M. Goličnik / Biochemical Engineering Journal 53 (2011) 234–238 235

2. Theory and calculation

2.1. Mathematical analysis of the Michaelis–Menten

enzyme-reaction model

The kinetics of enzymes are generally governed by the

Michaelis–Menten equation, as follows in Eq. (1):
d [P] d [S] V · [S]
v= =− = (1)
dt dt Km + [S]

where v, [S], V and Km represent the rate of reaction, the substrate

concentration, the limiting rate, and the Michaelis constant, respec-
tively. An explicit closed-form solution of the Michaelis–Menten
equation was described using the Lambert W(x) function for that
purpose [5], as Eq. (2):
  
[S]0 [S]0 − V · t
[P]t = [S]0 − Km · W · exp (2)
Km Km
It should be elucidated that Eqs. (1) and (2) are accurate when
the criterion for total enzyme concentration [E]0 < <([S]0 + Km )
guarantees the validity of the standard quasi-steady-state approx-
imation (sQSSA) [5]. In the case of most in vitro assays, the latter
criterion is satisfied easily. However, the sQSSA condition breaks
down under in vivo conditions, where the intracellular concen-
trations of enzymes are usually higher or at least of the same
magnitude as their substrates [13,14]. Therefore, Tzafriri extended
the Michaelis–Menten kinetics at high enzyme concentrations and Fig. 1. (A) The three real branches of the Lambert W(x) function for the correspond-
ing values of argument x. Region 1 (- - -) x > 0, region 2 (· · ·) −exp(−1) < x < 0 and
derived the following equation, Eq. (3) for the time-course of prod-
0 > W(x) > −1; region 3 (-) −exp(−1) < x < 0 and W(x) < −1. (B) Relative errors of the
uct accumulation [13]: three different approximations of the Lambert W function for x > 0 on the exponen-
tial x-scale: W2 (- - -), 10 × W2 * (-·· -·· -··), and Wω (· · ·). The infinite x range interval
[P]t = [S]0 − (Km + [E]0 ) (0,∞) has been mapped onto the interval (0,1) with the aid of the exponential x-scale
    = 1 − 1/ln(x + exp(1)).
[S]0 [S]0 − k2 · [E]0 · t
·W · exp (3)
(Km + [E]0 ) (Km + [E]0 )
approximations, with the second term of this sequence as in Eq.
which is valid for the following condition [E]0 + Km > >[S]0 . As bio- (6):
chemists usually analyze enzyme kinetics indeed within the sQSSA  6·x

framework, the use of Eq. (3) exceeds the scope of this article, and W 2 (x) = ln (6)
5 · ln[(12/5) · (x/ln(1 + (12/5) · x))]
all further expressions and discussions refer to the sQSSA only.
where the 2 in Eq. (6) is a counter, and not a power. The maximum
2.2. The Lambert W(x) function and its calculation relative error computed for Eq. (6) is 2.4% (see Fig. 1B). Linear inter-
polation between the first two approximation terms W1 and W2
The Lambert W(x) function is a mathematical function (see produces a significant improvement (additionally marked with *),
Fig. 1A) with numerous well-documented applications in math- which is given as in Eq. (7):
ematics, physics and chemical engineering [10,11,15]. It is defined

6·x
as the solution to the transcendental equation that is given as in Eq. W 2∗
(x) = (1 + ε) · ln  
(4): 5 · ln (12/5) · (x/ln(1 + (12/5) · x))

W (x) · exp (W (x)) = x (4)

 2·x

−ε · ln (7)
ln(1 + 2 · x)
An examination of Eq. (2) for x gives Eq. (5):
[S]0
 [S] − V · t  where ε = 0.4586887. For Eq. (7), the maximum relative error is
0
x= · exp (5)
Km Km only 0.2% (see Fig. 1B). Winitzki [12] providing another simple ana-
lytic function that can approximate W(x) with a maximum relative
which indicates that x is positive at all times, as the Michaelis
error of less than 2% (see Fig. 1B), as Eq. (8):
constant Km , the initial substrate concentration [S]0 , and the expo-
 ln(1 + ln(1 + x))

nential function are always positive. Therefore, a unique solution to
W ω (x) = ln(1 + x) · 1− (8)
Eq. (2) exists for any progress curve of interest when this solution to 2 + ln(1 + x)
the Michaelis–Menten type kinetics is applied. However, this solu-
tion needs to be calculated computationally [16] when fitting the 3. Materials and methods
explicit Eq. (2) to data, where evaluating W(x) is the main problem,
since it is not an implemented function in most nonlinear regres- 3.1. Progress-curve experimental data
sion computer programs. Simple, yet accurate, approximations of
W(x) that are useful for quickly generating results for progress- The data analysis was carried out on time-courses from two
curve analyses of Michaelis–Menten reaction systems are based on reactions obtained from the literature: from Fig. 1 of Moss et al. [17],
the principle that W(x) can be expressed by a composite of natu- for the reduction of testosterone to dihydrotestosterone by human
ral logarithms. Barry et al. [11] reduced the W(x) to a sequence of steroid 5␣ reductase; and from Fig. 1B of Kuzmič et al. [18], for the
236 M. Goličnik / Biochemical Engineering Journal 53 (2011) 234–238

hydrolysis of a fluorogenic peptide by HIV-1 proteinase. These data

are a part of example problems included in the DynaFit academic
free nonlinear regression software [19].

3.2. Nonlinear regression fitting using explicit equations

The Lambert W(x) function is built into Wolfram Mathematica

7.0 software tools under the name of ProductLog. ProductLog[x] give
the principal solution for W(x) in Eq. (4). Therefore, the explicit
closed-form solution for the product concentration as a function of
the reaction time was expressed in Mathematica as the model in
Eq. (9):
 S  S − V × t 
0 0
model = P × S0 − Km × ProductLog × Exp
Km Km
+baseline (9)

where P stands for the specific molar instrumental response of the

reaction product, and the baseline is the instrumental offset param-
eter that accounts for the possible systematic error of the detection
method. The sum of the squares of the differences between the
product concentrations of the calculated model (Eq. (9)) and the
experimental data was minimized with the Mathematica Nonlin-
earModelFit routine (see Supplementary Material), which finds the
numerical values of the fitted parameters that provide the model
equation with the best least-squares fit to the data as a function of
Fig. 2. Time-courses of product concentrations. The symbols represent the exper-
the independent variable, time t. imental data for the percentage of testosterone reduction by human steroid 5␣
Approximations of Eqs. (6)–(8) to the Lambert W(x) function of reductase [Ref. 17] (A), and for the fluorescence changes during HIV-1 proteinase-
modified Eq. (2) for product accumulation were implemented into catalyzed hydrolysis of fluorogenic substrate [Ref. 18] (B). The lines are the
theoretical concentrations obtained from the approximation W2 * of W(x) (see Eq.
the GraphPad Prism 5 software package as user-defined built-in
(7) in Theory) of modified Eq. (2), and the parameter values shown in Tables 1 and 2.
explicit model equations (see Table A1 in Appendix A) for calculat- Only 20% of the data points are shown in (B) for clarity.
ing theoretical product concentrations. This standard-curve-fitting
computer program has an all-user interface that allows users to
easily set-up a global least-squares nonlinear regression curve fit
tion to the Michaelis–Menten rate equation is an accurate model
so as to adjust the values of the model parameters to find the curves
of the enzyme action within the sQSSA framework that can be
that best predict the data.
fit to experimental time-course data, but it involves the Lam-
bert W-omega function. Although several powerful mathematical
3.3. Nonlinear regression fitting using numerical integration software packages (e.g. Mathematica) have been made available
for this purpose [5–9], the requirement for a specialized soft-
Data analyses were performed using the available DynaFit com- ware algorithm for computing W(x) [6] is a serious drawback in
puter program [19], which combines numerical integration with many circumstances. Therefore, W(x) in Eq. (2) has been replaced
nonlinear regression. The classic Michaelis–Menten model mech- by three different approximate functions (Eqs. (6)–(8)) expressed
anism (E + S ⇔ ES → E + P) and the initial estimates of the fitting as composites of elementary mathematical functions only. These
kinetic constants were entered into the program input script files equations are given in Table A1 (see Appendix A) and can now
(see Supplementary Material), while the value of the bimolecular be used to simulate and fit experimental data without relying on
association rate constant kon = 108 M−1 s−1 remained fixed in the highly specialized algorithms and software, but simply by encoding
model. them in any standard data-fitting computer program.
This new analytical approach was tested by selecting the data
4. Results and discussion from two metabolism reactions from the literature (Fig. 2) that were
catalyzed by human steroid 5␣ reductase and HIV-1 proteinase
The objective of this study was to determine the reliability [17,18]. The time-course data analyses for these two enzyme-
with which the kinetic parameters of reactions displaying hyper- catalyzed reactions were carried out using nonlinear regression,
bolic kinetics can be estimated by analyzing progress curves in where the theoretical curves were computed by different calculat-
terms of different explicit time-dependent equations. The solu- ing techniques, and the fitted parameters for each reaction are given

Table 1
Parameters acquired by single progress-curve fitting. Comparison of the fitted parameters for the reaction of human steroid 5␣ reductase obtained using numerical integration
methods (DynaFit), model Eq. (9) with the Lambert W function (Mathematica), and approximations of W(x) (Eqs. (6)–(8)) of modified Eq. (2) (Prism 5). The initial substrate
concentration was 31 nM and was set as constant. Data are means ± SD.

Numerical integration Lambert W function Lambert W approx.-W2 Lambert W approx.-W2 * Lambert W approx.-Wω

Km (nM) 20.5 ± 3.4 20.5 ± 3.4 20.8 ± 3.3 20.7 ± 3.4 17.3 ± 3.2
V (nM/min) 1.83 ± 0.14 1.83 ± 0.14 1.84 ± 0.14 1.85 ± 0.15 1.68 ± 0.15
P (a.u./nM) 3.21 ± 0.04 3.21 ± 0.04 3.14 ± 0.04 3.21 ± 0.04 3.22 ± 0.06
Baseline (a.u.) −1.0 ± 0.6 −1.0 ± 0.6 −1.3 ± 0.6 −1.1 ± 0.6 −1.8 ± 1.1
SSQ 7.618 7.618 7.590 7.624 7.823

SSQ, absolute sum of squares. Arbitrary units (a.u.) are the percentage of the product formed.
 M. Goličnik / Biochemical Engineering Journal 53 (2011) 234–238 237

Table 2
Parameters acquired by global (simultaneous) multiple progress-curve fitting. Comparisons of the fitted parameters for the reaction of HIV-1 proteinase acquired using
numerical integration methods (DynaFit), model Eq. (9) with the Lambert W function (Mathematica), and approximations of W(x) (Eqs. (6)–(8)) of modified Eq. (2) (Prism 5).
The specific molar instrumental response of the reaction product P was 2.57 and it was set as constant. Data are means ± SD.

Numerical integration Lambert W function Lambert W approx.-W2 Lambert W approx.-W2 * Lambert W approx.-Wω

Km (␮M) 1.786 ± 0.025 1.796 ± 0.025 1.798 ± 0.024 1.805 ± 0.025 1.621 ± 0.021
V (␮M/s) 0.0909 ± 0.0006 0.0910 ± 0.0006 0.0897 ± 0.0005 0.0914 ± 0.0006 0.0862 ± 0.0005
[S]0 1 (␮M) 0.638 ± 0.007 0.639 ± 0.007 0.629 ± 0.007 0.638 ± 0.007 0.648 ± 0.007
Baseline1 3.291 ± 0.016 3.290 ± 0.016 3.316 ± 0.015 3.292 ± 0.016 3.262 ± 0.016
[S]0 2 (␮M) 0.857 ± 0.07 0.857 ± 0.007 0.843 ± 0.007 0.856 ± 0.007 0.870 ± 0.007
Baseline2 3.522 ± 0.015 3.522 ± 0.015 3.560 ± 0.015 3.524 ± 0.015 3.483 ± 0.015
[S]0 3 (␮M) 1.375 ± 0.006 1.375 ± 0.006 1.348 ± 0.006 1.374 ± 0.006 1.397 ± 0.006
Baseline3 3.580 ± 0.014 3.560 ± 0.014 3.652 ± 0.013 3.581 ± 0.014 3.517 ± 0.014
[S]0 4 (␮M) 2.927 ± 0.006 2.927 ± 0.006 2.861 ± 0.006 2.930 ± 0.006 2.946 ± 0.006
Baseline4 3.449 ± 0.011 3.448 ± 0.011 3.621 ± 0.011 3.442 ± 0.012 3.391 ± 0.011
[S]0 5 (␮M) 3.727 ± 0.006 3.727 ± 0.006 3.642 ± 0.006 3.731 ± 0.006 3.729 ± 0.006
Baseline5 3.739 ± 0.011 3.738 ± 0.011 3.959 ± 0.011 3.726 ± 0.011 3.723 ± 0.012
[S]0 6 (␮M) 5.637 ± 0.008 5.637 ± 0.008 5.513 ± 0.008 5.648 ± 0.008 5.584 ± 0.009
Baseline6 3.750 ± 0.015 3.749 ± 0.015 4.071 ± 0.014 3.721 ± 0.015 3.874 ± 0.017
SSQ 2.865 2.865 2.869 2.865 2.867

SSQ, absolute sum of squares. Superscript numbers correspond to progress curves in Fig. 1B (bottom to top).

in Tables 1 and 2, respectively. Progress curves were first analyzed to Eq. (2). The latter can be used for progress-curve analyses that
by numerical integration, which has been the most precise and decrease the number of experimental assays by at least a factor of
widely used fully general method among enzymologists. Therefore, five [6]. However, it could be argued that Eq. (2), and consequently
the best fit to the data obtained using DynaFit numerical integration its approximations, is of little relevance to most real enzyme reac-
[17–19] yielded the reference values for comparing to the results tions, as these apply only to enzymes that show Michaelis–Menten
obtained by the fitting of Eq. (2) or its modified forms to time- kinetics behavior. In fact, a real enzyme usually obeys more com-
course data. The Mathematica NonlinearModelFit routine on model plex rate equations than Eq. (1), but experimental conditions can
Eq. (9) yielded values almost identical to the reference values. The sometimes be manipulated such that Eq. (1) is a very good starting-
best parameter values shown in Tables 1 and 2 yielded a surpris- point approximation [5]. Finally, it should be remembered that
ingly good fit to the experimental data of the integrated rate-law the starting substrate concentration is very important for reliable
approximations W2 and W2 * (Eqs. (6) and (7)) by applying stan- determination of the kinetic parameters. There is no way to distin-
dard nonlinear regression software, although Wω approximation guish V and Km constants when the substrate concentration is too
(Eq. (8)) resulted in slightly lower values for V and Km . Despite an low, because the time-course shows apparent first-order kinetics,
accuracy of all approximation functions of around 1% only [11,12], following the single-exponential shape that reflects the ratio V/Km
which is in the range of usual experimental error, it appears that only. An appropriate choice for the initial substrate concentration
these approximations can serve as simple analytic tools that can has been suggested as two to three times the Km [21].
excellently substitute the Lambert W(x) function in Eq. (2). The type In conclusion, this study has described an easy and quick
of approximation (Wω , W2 , W2 *) does not have a significant role for analysis of progress curves of enzymatic reactions within the
the uncertainties. The Km values always have the highest uncertain- Michaelis–Menten framework, to extract the kinetic parameters.
ties, but this should be expected according to the progress-curve The primary improvement of this study is the use of approxima-
robustness analysis [8]. tions to the Lambert W(x) function, thereby permitting the analysis
It has been shown that there are no technical limits for direct fit- to be carried out using standard nonlinear regression software. This
ting of model equations to two or more progress curves of the same is an improvement that could greatly facilitate characterization of
reaction simultaneously, because most nonlinear regression curve- enzyme dynamics in several life-science research fields.
fitting software tools now have an all-user interface that allows
users to easily set-up a global curve fit. The latter is without doubt
the better choice, and often it is even necessary, because differ-
Acknowledgements
ent kinetic models cannot be distinguished from a single progress
curve only. While the derivation of the explicit solution (Eq. (2))
The author gratefully acknowledges Prof. J. Stojan for valuable
has been applied to the Michaelis–Menten equation, it is equally
discussions and critical reading of the manuscript. This work was
applicable to several other kinetic models that can be reduced to
supported by the Slovenian Research Agency (grant P1-170).
forms that are analogous to the Michaelis–Menten equation. For
instance, the approaches to equilibrium of the enzyme and inhi-
bition reaction mechanisms, such as competitive, uncompetitive
and mixed-type inhibition, can all be reduced to forms that are Appendix A.
analogous to Eq. (1), with different definitions of V and Km [20].
Hence, they all have explicit closed-form solutions that are similar See Table A1.

Table A1
Software-user-defined built-in approximations of W(x) (Eqs. (6)–(8)) of modified Eq. (2) for product accumulation in GraphPad Prism 5. The expression [P]t = [S]0 − [S]t holds
for both enzyme-catalyzed reactions.

GraphPad Prism 5

Wω (x) P × (S0 − Km × ln(1 + S0 /Km × exp((S0 − Vm × x)/Km )) × (1 − ln(1 + ln(1 + S0 /Km × exp((S0 − Vm × x)/Km )))/(2 + ln(1 + S0 /Km × exp((S0 − Vm × x)/Km ))))) + baseline
W2 (x) P × (S0 − Km × ln(1.2 × (S0 /Km × exp((S0 − Vm × x)/Km ))/ln(2.4 × (S0 /Km × exp((S0 − Vm × x)/Km ))/ln(1 + 2.4 × (S0 /Km × exp((S0 − Vm × x)/Km )))))) + baseline
W2 *(x) P × (S0 − Km × (1.4586887 × ln(1.2 × (S0 /Km × exp((S0 − Vm × x)/Km ))/ln(2.4 × (S0 /Km × exp((S0 − Vm × x)/Km ))/ln(1 + 2.4 × (S0 /Km × exp((S0 − Vm × x)/Km ))))) −
0.4586887 × ln(2 × (S0 /Km × exp((S0 − Vm × x)/Km ))/ln(1 + 2 × (S0 /Km × exp((S0 − Vm × x)/Km )))))) + baseline
238 M. Goličnik / Biochemical Engineering Journal 53 (2011) 234–238

Appendix B. Supplementary data [10] R.M. Corless, G.H. Gonnet, D.E.G. Hare, D.J. Jeffrey, D.E. Knuth, On the Lambert
W function, Adv. Comput. Math. 5 (1996) 329–359.
[11] D.A. Barry, J.Y. Parlange, L. Li, H. Prommer, C.J. Cunningham, F. Stagnitti, Ana-
Supplementary data associated with this article can be found, in lytical approximations for real values of the Lambert W-function, Math. Comp.
the online version, at doi:10.1016/j.bej.2010.10.012. Simul. 53 (2000) 95–103.
[12] S. Winitzki, Uniform approximations for transcendental functions, Lecture
Notes Comp. Sci. 2667 (2003) 780–789.
References [13] A.R. Tzafriri, Michaelis–Menten kinetics at high enzyme concentrations, Bull.
Math. Biol. 65 (2003) 1111–1129.
[1] E.F. Murphy, S.G. Gilmour, M.J.C. Crabbe, Effective experimental design: enzyme [14] M. Jevšek, T. Marš, K. Miš, Z. Grubič, Origin of acetylcholinesterase in the neu-
kinetics in the bioinformatics era, Drug Discov. Today 7 (2002) S187–S191. romuscular junction formed in the in vitro innervated human muscle, Eur. J.
[2] N.G. Walter, Michaelis–Menten is dead, long live Michaelis–Menten, Nat. Chem. Neurosci. 20 (2004) 2865–2871.
Biol. 2 (2006) 66–67. [15] D. Clamond, Efficient resolution of Colebrook equation, Ind. Eng. Chem. Res. 48
[3] J. Bjerre, C. Rousseau, L. Marinescu, M. Bols, Artificial enzymes, “chemzymes”: (2009) 3665–3671.
current state and perspectives, Appl. Microbiol. Biotechnol. 81 (2008) 1–11. [16] D.A. Barry, S.J. Barry, P.J. Culligen-Hensley, Algorithm 743: a Fortran routine for
[4] R.G. Duggleby, Quantitative analysis of the time-courses of enzyme-catalyzed calculating real values of the W function, ACM Trans. Math. Softw. 21 (1995)
reactions, Methods 24 (2001) 168–174. 172–181.
[5] S. Schnell, C. Mendoza, Closed-form solution for time-dependent enzyme kinet- [17] M.L. Moss, P. Kuzmič, J.D. Stuart, G. Tian, A.G. Peranteau, S.V. Frye, S.H. Kadwell,
ics, J. Theor. Biol. 187 (1997) 207–212. T.A. Fost, L.K. Overton, I.R. Patel, Inhibition of human steroid 5␣ reductase type
[6] C.T. Goudar, J.R. Sonnad, R.G. Duggleby, Parameter estimation using a direct I and II by 6-aza-steroids, Biochemistry 35 (1996) 3457–3464.
solution of the integrated Michaelis–Menten equation, Biochim. Biophys. Acta [18] P. Kuzmič, A.G. Peranteau, C. Garcia-Echeverria, D.H. Rich, Mechanical effects
1429 (1999) 377–383. on the kinetics of the HIV proteinase deactivation, Biochem. Biophys. Res. Com-
[7] S. Schnell, C. Mendoza, Time-dependent closed form solution for fully compet- mun. 221 (1996) 313–317.
itive enzyme reactions, Bull. Math. Biol. 62 (2000) 321–336. [19] P. Kuzmič, Program DYNAFIT for the analysis of enzyme kinetic data, Anal.
[8] C.T. Goudar, S.K. Harris, M.J. McInerney, J.M. Suflita, Progress curve analysis for Biochem. 237 (1996) 260–273 (http://www.biokin.com/dynafit/).
enzyme and microbial kinetic reactions using explicit solutions based on the [20] I.H. Segel, In Enzyme Kinetics, John Wiley & Sons, New York, 1993.
Lambert W function, J. Microbiol. Methods 59 (2004) 317–326. [21] R.G. Duggleby, R.B. Clarke, Experimental designs for estimating the parameters
[9] S. Tang, Y. Xiao, One-compartment model with Michaelis–Menten elimination of the Michaelis–Menten equation from progress curves of enzyme-catalyzed
kinetics and therapeutic window: an analytical approach, J. Pharmacokinet. reactions, Biochim. Biophys. Acta 1080 (1991) 231–236.
Pharmacodyn. 34 (2007) 807–827.