Biology Reference
In-Depth Information
COMPUTATIONAL ANALYSIS
OF SIGNALING NETWORKS
Owing to the large number of components, their varied
molecular functions and their many interactions, it is
impossible to intuit how such a complex system will
function. To obtain such understanding computational
analyses are essential. The goals of computational analyses
are (1) to understand the organization of the signaling
network, and (2) to understand how this organization leads
to the processing of information as signal flows through the
network. This processing of information leads to changes in
the
common in metabolic pathways which are frequently
involved in maintaining homeostasis through end-product
inhibition. Connectivity enables the formation of both
positive and negative feedback loops, depending on the
isoforms of the components present. A typical example in
the neuron is the isoforms of the phosphodiesterase 4D
that are activated by protein kinase A. The presence of
these isoforms enables the formation of a negative feed-
back loop that controls the level of cAMP, thereby regu-
lating signal flow to multiple downstream effectors.
Network motifs display a range of signal processing
capabilities, including the ability to delay or speed up
signal flow, to enable the flow of information between
different timescales, to filter noise, to sort or synchronize
signals, to function as switches and to generate oscillations
[47]
. Incoherent feedback loops can function as sensors
of fold change in both the Wnt
[48]
and EGF signaling
[49]
pathways. The topology of feedback and forward
motifs is shown in
Figure 16.4
A.
One important feature of feedforward motifs is that they
allow the transmission of signals across time scales. This
'temporal bridge' between the fast and slow timescales can
have important effects in controlling several biological
phenomena, for example proliferation and differentiation.
One well-studied example is the MAPK-1,2 signaling
network. MAPK-1 and -2 are activated within minutes in
response to a wide number of external stimuli. In response
to MAPK activation, a set of immediate-early genes (IEGs)
are turned on in order to convert the initial MAPK activa-
tion into different long-lasting biological effects, such as
proliferation or differentiation
[50
output relationships. Thus cell signaling
networks are different from engineered communication
networks. In the latter system the overriding goal is to
maintain the fidelity of the input information so that output
faithfully matches the input. In contrast, in cell signaling
networks the system is configured to process information
such that it can evaluate the information even as signal
flows through the network and modulates output and
responses in accordance with current capabilities of the
cell. Such information processing can lead to many
different kinds of outcome. Sometimes short-duration input
signals are converted into longer-duration outputs so as to
engage multiple cellular machines and change cell physi-
ological functions. At other times sub-threshold signals
may be dissipated, such that even though we may observe
biochemical signaling reactions, no cell physiological
responses are elicited. Two types of computational model
have been widely used to study cell signaling networks, and
these are discussed in the following sections.
input
e
52]
. One crucial factor
in deciphering the signal and translating information into
the proper biological responses is the duration of the signal
itself. In the rat neuronal cell line PC12 short activation of
MAPK-1,2 leads to proliferation, whereas prolonged
MAPK-1,2 signaling induces differentiation
[52
e
Graph Theory-Based Models
The large numbers of components and interactions have
made network analysis a very useful approach for under-
standing the organization of cell signaling networks.
Substantial information about the topology can be gleaned
from both undirected and directed graphs. Characteristics
such as scale-free properties
[44]
and clustering coefficient
[45]
provide insight into the global topology of networks.
At a more microscopic scale network motifs such as
feedback loops, as well as feedforward and bifan motifs,
serve as building blocks of networks as they occur within
networks in a recurrent manner
[46]
.
54]
.
Murphy and co-workers have described a feedforward
motif within the signaling network that decodes the extent
of MAPK activation and consequently affects the expres-
sion of IEGs
[55,56]
. A summary of their results is repre-
sented in
Figure 16.4
B. One of the IEGs that are turned on
in response to MAPK-1,2 activation is c-Fos. Since the
timescale that is necessary for c-Fos transcription, trans-
lation and nuclear transportation is much longer that that
required for transient MAPK-1,2 activation, only an initial
prolonged MAPK signaling promotes c-Fos phosphoryla-
tion, which protects c-Fos against degradation by the
ubiquitin
e
REGULATION BY NETWORK MOTIFS
A motif can be defined as a group of interacting compo-
nents within the network which is capable of signal pro-
cessing. Examples include positive feedback loops that
enable propagation of signals across timescales, whereas
a negative feedback loop limits the signal propagation
through the network. Negative feedback loops are
proteasome pathway
[57,58]
. A stable c-Fos
(c-Fos-P) promotes transcription of c-Fos target genes even
when MAPK signaling is turned off through a negative
feedback loop involving MAPK-phosphatase that requires
about 30
e
60 minutes to become operational. Conse-
quently, the feedforward regulatory motif is responsible for
decoding the signal duration and can therefore bridge the
e
