Systems Biology and Metabolic Engineering in Bacteria - Systems Biology of Metabolic and Signaling Networks - page 347

concentration and enzyme activities. MCA and its extensions remain very useful
tools for network understanding and they may gain in importance as more quanti-
tative experimental data become available.
13.4 Dynamical Models
The natural habitat of most microorganisms is far from constant, forcing the
microbe to adapt its metabolism to ever changing conditions. Even a typical
batch culture in the laboratory puts the bacterium through cycles of feast and
famine (Ferenci 2001 ). The same problem arises for production strains in industrial
fermentors. In order to understand and optimize phenotypes under these conditions,
we need to assess the system behavior not only during a particular steady state but
also during transitions between different growth conditions. Otherwise stated, we
need a dynamic description of the metabolic network.
Historically, such models derive from enzyme kinetics, a detailed description of
an enzymatic reaction. The challenge for systems biology is to consider all enzy-
matic reactions of the cell simultaneously. Two approaches are possible for
attaining this goal: (1) bottom-up, where we do precisely that: assemble the
individual component reactions of a metabolic network, and (2) top-down, where
we attempt to put dynamics into a genome-wide steady state model.
Generally, the first modeling approach is formulated as a system of differential
equations, one equation for each enzymatic reaction (Fig. 13.3 ). This formalism has
the advantage of being very general and leading to very precise predictions of
system dynamics. However, this approach requires knowledge of (1) the topology
of the metabolic and regulatory network, as well as (2) good estimates of all
parameters of the kinetic equations. Such models are therefore limited to well-
known organisms such as E. coli . Once all parameters and equations are assembled,
the system behavior is calculated, “emerging” from the interactions of the individ-
ual components.
Because it is currently impossible to measure all the parameters of all the
enzymatic reactions of an organism, such models focus on specific parts of the
global network, for example, central metabolism. Kremling et al. ( 2008 ) have
succeeded in compiling a complete model of glycolysis in E. coli . Such detailed
models enable answering very specific questions. The Kremling model, for exam-
ple, led to the discovery of specific metabolic regulatory mechanisms for system
functioning. They could demonstrate the importance of a feed-forward loop linking
the upper part of glycolysis to pyruvate kinase. Heinemann et al. have extended this
approach and constructed a complete model of the central catabolism of E. coli
(Kotte et al. 2010 ). One major advantage of such models is that metabolic and
genetic regulatory mechanisms can be integrated seamlessly into a unified system
description (also see below). The surprising result of Kotte and coworkers was that
metabolic adaptations rely on distributed sensing of metabolic fluxes, where
metabolites such as fructose-biphosphate or cAMP play key roles as flux sensors.