Genome-Scale Models of Metabolic and Regulatory Networks - Systems Biology: Networks, Models, and Applications - page 252

consumption of metabolites and the production of biomass X BM is deter-
mined by the following set of differential equations derived from eq. (1):

BM

dX
dt

= µ X

BM

(8)

dX
dt

ext

= vX

BM

ext

where m is the growth rate obtained from the FBA calculation (7)
and v ext are the exchange fluxes corresponding to all extracellular
metabolites that trigger regulatory effects (negative fluxes correspond
to uptake and positive fluxes to secretion of metabolites). Equations
(6)-(8) can be solved iteratively starting from a particular initial set
of extracellular metabolite concentrations to obtain a time course of
biomass and extracellular metabolite concentrations. Overall the rFBA
approach allows accounting for environment- and time-dependent
constraints acting on the metabolic network to limit its phenotypic
capabilities (figure 8.5). Similarly to the rFBA method, other constraint-
based methods, such as extreme pathway analysis can be extended to
allow accounting for regulatory effects [108].

Applications of Integrated Models
The kind of integrated models that combine genome-scale metabolic
and regulatory networks discussed above have so far been formulated
for E. coli [57,109] and yeast [102]. The dynamic time-course analysis
discussed above was used to predict metabolite concentration profiles
for a number of metabolic shifts using a regulated E. coli core metabolic
model [109]. Building on this model, a genome-scale regulated meta-
bolic model of E. coli was established [57]. This model, which accounts
for a total of 1010 genes, represents the first genome-scale integrated
model of multiple cellular functions in a microbial organism (table 8.3).
The major advantage of such integrated models is that even when the
modeling of the regulatory network function is done at the qualitative
level, the integrated regulatory/metabolic model can be used to quan-
titatively predict phenotypes such as growth rates. These predictions
can then be compared to experimentally measured regulatory or meta-
bolic gene knockout strain growth rates.
In addition to the growth rate predictions, the qualitative gene
expression change predictions made by the integrated model can be
compared with experimentally measured expression profiles [57]. This
type of systematic comparison allowed iterative improvement of the
regulatory and metabolic network models to establish models with
improved predictive capability for both growth rates and expression
changes (figure 8.5). The integrated models discussed here are a