Genome-Scale Models of Metabolic and Regulatory Networks - Systems Biology: Networks, Models, and Applications - page 246

amino acid and phospholipid metabolism coupled with reactions
involved in nucleotide metabolism. Thus, Co-Sets in metabolic networks
identify functional and not necessarily intuitive groups of reactions.
Co-Sets in E. coli for growth on a variety of different carbon sources
were also enumerated from alternate optimal solution calculations [32].
The flux coupling finder method can be directly used to define
Co-Sets that constitute functional modules in the network [45]. The
percentage of reactions in such Co-Sets under aerobic glucose minimal
medium growth conditions was found to be ~60% for H. pylori, ~30%
for E. coli , and ~20% for S. cerevisiae . The high fraction of coupled
reactions in H. pylori provides another indication of the low degree
of flexibility that the metabolic network of this pathogen has. The
directional couplings obtained from flux coupling analysis can be used
to establish the core set of reactions required for biomass formation in
all possible flux distributions within the allowed flux space for a
particular media condition. In H. pylori , 59% of the reactions in the
model were required for biomass formation whereas for E. coli and
yeast these percentages were 28% and 14% respectively.
The flux distributions obtained by random sampling methods
described above can also be used to establish functional modules in
metabolic networks [47,48]. By calculating pairwise correlation coef-
ficients between all reaction fluxes based on the sampled uniform flux
distributions, the degree of co-utilization of fluxes can be quantified.
Sampling can also be used to establish other types of general features
of metabolic network function such as the distribution of flux magni-
tudes allowed through each reaction in the metabolic network.
This type of analysis identified a high-flux reaction backbone in the
genome-scale E. coli metabolic network that carries the vast majority
of the metabolic flux through the network [46]. This backbone remains
relatively intact under a variety of environmental conditions, whereas
the remaining low-flux pathways are used differently in different
conditions.
STRAIN DESIGN AND METABOLIC ENGINEERING
The high degree of functional connectivity in metabolic networks
implied by the results described in the previous sections also poses a
challenge for designing strains for overproduction of desired metabolic
byproducts. Accounting for all the cofactor balancing requirements, as
well as cellular growth requirements, is essential for determining the
minimal genetic modifications needed to increase desired byproduct
secretion compared to a wild-type strain. Genome-scale models
provide an attractive starting point for strain design, because they
automatically include all the necessary requirements as comprehen-
sively as is possible based on our current understanding of the
metabolic physiology of a given organism.