Biology Reference
In-Depth Information
connectivity and metabolic capacity of the network from genome annotation
(Pitk¨nen et al.
2010
). The preferred model organisms were
E. coli
and
S. cerevisiae
. The latest, integrated model of
E. coli
comprises almost 12,000
network components and over 13,000 reactions (Thiele et al.
2009
). This model
was created by the Palsson group, who has pioneered and is still at the forefront of
the reconstruction of metabolic networks in
E. coli
and other organisms (Feist
et al.
2009
).
The second step after the reconstruction of network components and their
connectivity is the comparison with available experimental data in order to validate
the proposed structure. A prerequisite for the faithful reconstruction of a metabolic
network is thus the availability of high-throughput, quantitative techniques for
measuring the network components and their interactions (Yamada and Bork
2009
). Even though the reconstruction of network connectivity is greatly facilitated
by software tools, the automatic detection and repair network inconsistencies
remain very difficult (Pitk¨nen et al.
2010
). The best models still rely on manual
curation.
Furthermore, even for very well studied organisms, such as
E. coli
, the experi-
mental exploration of the metabolic network still yields surprises. Recently,
Nakahigashi et al. (
2009
) have detected significant differences between the
predictions derived from the very well established metabolic network of
E. coli
and the observed growth phenotypes of double knockout mutants. These additional
reactions of the central carbon metabolism provide an alternative pathway for
glucose breakdown. The remarkable fact about these new reactions is that their
activation does not require any changes in gene expression. Such alternative
pathways are certainly part of the features that convey robustness to metabolic
networks. A purely bioinformatic analysis will almost certainly miss such reactions,
reiterating the need for experimental validation of predicted network structures
even for the best-studied organism. Despite certain shortcomings, these network
reconstruction methods have been applied with great success to well-studied
organisms (Kim et al.
2012
), such as
E. coli
and
S. cerevisiae
, but also to less
well studied organisms of particular fundamental or biotechnological interest, such
as photosynthetic cyanobacteria (Montagud et al.
2010
).
Even though the completeness of the metabolic network cannot be assured, the
quality of current network reconstructions allows to pass on to the next step:
calculating the phenotype produced by a metabolic network. This task consists
essentially in predicting the metabolic fluxes through the network in different
growth conditions. From these fluxes, we can calculate the phenotypes, such as
growth rate, metabolic capacities, and yield of particular metabolite. There are two
major modeling approaches for calculating the “behavior” of a metabolic network
(Chen et al.
2012
): (1) steady state, and (2) kinetic. The former can be applied to
microorganisms during balanced growth, whereas the latter allows to assessing
time-dependent responses of a microbe to changes in the environment.
Search WWH ::
Custom Search