Biomedical Engineering Reference
In-Depth Information
The recent availability of information about large protein and gene net-
works (containing thousands of components) in baker's yeast (
S. cerevisiae
)
made possible by new large-scale, high-throughput biochemical methods has
stimulated investigations into the natural architecture of biological regulatory
networks (see also this volume, Part II, chapter 4, by Wuchty, Ravasz, and
Barabási). These studies revealed that the almost genome-wide protein-
interaction network and gene regulatory network are indeed sparsely connected
in these cells. Moreover, several interesting features of the network architecture,
such as a near power-law distribution of connectivity (number of interaction
partners per molecule), a propensity to modularity, and use of hierarchical struc-
ture, were found to be present (41,46,70). Interestingly, a power-law architecture
appears to have beneficial consequences for system-wide dynamics (1,20). Spe-
cifically, the regime in the space of possible network architectures for "biologi-
cally reasonable" networks (i.e., those that exhibit ordered behavior with small
attractors) is larger because the networks tolerate higher connectivity without
becoming chaotic.
4.2.3. Biological Implications of Attractor States
As in the case of structural networks and the tensegrity model that allows
prediction of some macroscopic mechanical properties of the cell based on
emergent features of the model, the generic global behavior of the cell is pre-
dicted by the model of an attractor landscape. In fact, the existence of distinct
stable cell fates (proliferation, apoptosis, quiescence, etc.) and of different dif-
ferentiated cell types (liver, skin, neuron, etc.) that are robust to many perturba-
tions, yet can switch between distinct states under restricted conditions, is itself
a strong indication that they are attractors of an underlying molecular network.
Similarly, robust developmental trajectories, corresponding to long valleys lead-
ing to lowest points in the landscape, can be explained as emergent properties of
the genome-wide network of genetic interactions. However, the dynamic net-
works approach and the attractor landscape idea may also provide new insight
into other cell biological phenomena that have previously resisted straightfor-
ward explanation by the conventional paradigm, which emphasizes the role of
individual signal-transduction pathways.
Cell fate regulation in tissue homeostasis
. As predicted by the dynamic
network model, cell fates represent discrete, mutually exclusive, stable states
that require specific signals to transition to each other, when such a transition is
possible. For instance, differentiation and proliferation are well known to be
mutually exclusive and robust (27,67); in many cell systems just quitting the
proliferation state by overexpressing the cell cycle inhibitor protein p21 forces
the cell to automatically enter the differentiation program (14,52,61). That cell
fates are robust and can be realized just by "placing" cells in the corresponding
