Systems Biology Approaches to Cancer Energy Metabolism - Systems Biology of Metabolic and Signaling Networks - page 214

pentose phosphate pathway; (2) the incorporation of a complete PFK-1 rate equa-
tion which includes the effect of allosteric modulators; and (3) the inclusion of
phosphate in the GAPDH rate equation. These modifications are further elaborated
below.
In a first attempt, the model rendered lower levels of G6P than those determined
in vivo, and in silico titration of the activity of each enzyme indicated that the rate
of HPI was too high. This prompted us to search for physiological inhibitors for this
enzyme. We found that physiological concentrations of F1,6BP (glycolytic metab-
olite), and 6-phosphogluconate (6PG) and erythrose-4-phosphate (Ery4P)
(metabolites of the PPP oxidative and non-oxidative sections, respectively) could
be potential candidates. Although the inhibitory effect of these metabolites on HPI
was described earlier (Zalitis and Oliver 1967 ; Chirgwin et al. 1975 ; Gaitonde
et al. 1989 ), its relevance was only evident after the glycolytic flux-controlling
property of HPI was elucidated.
Another difficulty found in our preliminary studies was that the simple rate
equation used for PFK-1 (i.e., hyperbolic kinetics or Hill equation) was unable to
describe its in vivo kinetic behavior, because it did not include the interaction with
regulatory metabolites. Given the surprisingly few kinetic studies on PFK-1, it was
necessary to thoroughly characterize its kinetic behavior to formulate a rate equa-
tion in both normal and tumor cells (Moreno-S ´ nchez et al. 2012 ). The general form
of the new PFK-1 equation obeys the concerted transition model of Monod,
Wyman, and Changeux (Segel 1975 ) for exclusive ligand binding (F6P, activators,
and inhibitors) together with mixed-type hyperbolic activation by F2,6BP or AMP
or Pi, and simple Michaelis-Menten terms for ATP and the reverse reaction (bi-bi
random). The inhibitory allosteric effect of ATP (at high concentrations) and citrate
could be reproduced with this equation. Due to the prevalence of the F2,6BP
activating effect over the inhibitory effect exerted by ATP and citrate, a 50-fold
decreased F6P level was obtained with the model.
Another finding of the model was that although GAPDH had a negligible control
on the glycolytic flux, it exerted significant control on the concentration of F1,6BP
and DHAP, depending on the cellular concentration of free Pi. The mechanistic
rationale underlying this homeostatic GAPDH behavior is given by the low affinity
of the enzyme for Pi. The relevance of this finding is readily apparent in the context
of previous models (Bakker et al. 1999 ; Teusink et al. 2000 ; Saavedra et al. 2007 )in
which it was assumed that Pi was saturating thus irrelevant for regulating GAPDH,
or any other enzyme activity. A recently published kinetic model indeed explored
the important role of Pi in the regulation of glycolysis in bacteria (Levering
et al. 2012 ). These examples are illustrative of the power of implementing iterative
strategies of modeling-experimentation to generate validated kinetic models that
accurately simulate available experimental data while predicting new behaviors
that help understanding the control and regulatory mechanisms of metabolic
pathways in vivo.
After incorporating into the model the three main changes mentioned above, it

;

predicted that the main flux and ATP concentration control steps are HK C J
E i

¼ 0 : 44