Biomedical Engineering Reference
In-Depth Information
lism and small-molecule building block synthesis in
E. coli
(15,57). A substrate
graph was defined by the nodes representing all metabolites, two substrates be-
ing considered linked if they occurred in the same reaction. They found the sub-
strate graph to be scale-free, with
glutamate
,
coenzyme A
,
2-oxoglutarate
,
pyruvate
, and
glutamine
having the highest degree, which were viewed as an
evolutionary core of the
E. coli
.
At the same time, Jeong et al. analyzed the metabolic networks of 43 organ-
isms representing all three domains of life (26), finding that the power-law de-
gree distribution for both incoming and outgoing edges holds for organisms of
all kingdoms. Furthermore, the average separation between nodes has the same
value for all organisms under consideration, regardless of the number of sub-
strates found in the given species. Interestingly, the ranking of the most con-
nected substrates is largely identical for all organisms. A recent study comparing
the system-level properties of metabolic networks in various organisms indicates
that the structural features of these networks are more conserved than the com-
ponents themselves (44,61).
4.2. Protein Interaction Networks
Protein interactions offer another opportunity to study cellular networks,
considering proteins as nodes and physical interactions (binding) as links. It has
been shown that interaction networks of
S. cerevisiae
and
H. pylori
proteins
exhibit distinct scale-free behavior (24,56; see also this volume, Part III, chapter
1.3, by Wagner). Although protein interaction data are derived from different
sources and retrieved by different methods, the emergence of the scale-free
property appears to be a robust feature. As previously discussed, scale-free net-
works are vulnerable to targeted attack on their highly connected nodes. There-
fore, mutations of highly interacting proteins are expected to be lethal for the
cell. This prediction is supported by explicit measurements (25). Figure 5 repre-
sents the yeast protein interaction network, illustrating the basic feature that
hubs keep many sparsely connected nodes together.
4.3. Protein Domain Networks
The domain architecture of proteins was studied by considering protein do-
mains as nodes and their co-occurrence in proteins as links (4,62,63), document-
ing again the emergence of a scale-free architecture. Although the methods and
sources of domain information were different, the scale-free features of the net-
works were found to be robust. Domains that appear in cellular functions crucial
for the maintenance of multicellular organisms, such as signal transduction and
cell-cell contacts, were found to be the most connected. Thus, domains like
