Biology Reference
In-Depth Information
pathway with known effects at the translational rather than
the transcriptional level.
provides a unique, context/signal specific 'signature' or
'phenoprint' for each perturbed network component.
Network components that are deployed in the same or
similar manner in response to the incoming signal would
tend to have similar phenoprints
[227,228]
.
QUANTITATIVE RNAI SIGNATURES
OR PHENOPRINTS TO INFER CONTEXT
DEPENDENT INFORMATION FLOW
THROUGH CELLULAR SIGNALING
NETWORKS
The network maps discussed so far are primarily static and
do not provide information regarding the direction of
information flow through the nodes; instead, they provide
the framework required to begin to dissect the functional,
logical and dynamical nature of cellular signaling
networks. The topological features of signaling networks
reflect the need to process multiple input cues received by
a cell, interpret them correctly and transmit the information
to coordinate cellular activity and generate the proper
phenotypic response
[224
Direction of Information Flow from Gene
Expression Signatures
Transcriptional signatures resulting from the loss/reduction
of individual network components by RNAi can be used to
infer the flow of information through proteins that are
interconnected within a cellular network. This approach
was used successfully in analyzing the response of
Drosophila cells to microbial infection and lipopolysac-
charides (LPS)
[229]
. In these studies, the topology of
network connections was retrieved from experimentally
measured global transcriptional responses to successive
perturbations in pathway components. Genome-wide
expression profiling and loss-of-function experiments using
RNAi were used to determine the identity of the signaling
pathways that control microbial challenge-induced cellular
responses. Differential gene expression signatures
appeared with discrete temporal patterns after LPS stimu-
lation and septic injury, and could be assigned to the acti-
vation of distinct signaling pathways by impairing
pathway-specific components using RNAi. Specifically, the
results indicated that in addition to signaling through the
Toll and Imd pathways, microbial agents induce signal
transmission through the JNK and JAK/STAT pathways.
Altogether, this demonstrated how data obtained from
microarray expression profiling combined with the RNAi
technology could be used to extract interconnections
between different signaling pathways downstream of an
extracellular stimulus.
Whole genome expression profiling
[230]
has identified
gene expression signature-based analysis of signaling
networks in a number of different model systems. One of the
first successful applications of this approach was the
generation of a compendium of gene expression profiles in
yeast for 300 different mutations and small molecule treat-
ments
[231]
. The 300 different mutations and chemical
treatments specifically included 276 deletion mutants, 11
tetracycline-inducible essential genes and 13 small molecule
inhibitors (data available from Rosetta Inpharmatics). The
assumption was that the cellular state can be deduced from
the global gene expression response, and the transcriptional
profile of a gene in response to a change in cellular state
(disease, cellular activity such as cell division, response to
drugs or genetic perturbation) constitutes a unique quanti-
tative molecular phenotype
[210,232
226]
. Thus, signaling networks
are highly dynamic, exist in distinct states, and are capable
of deploying the same or overlapping set of signaling
molecules in different ways, depending on the context and
the input cues received by the cell. It has been demonstrated
that, depending on the cumulative effects of the signals
received by a cell, JNK activity, for instance, can be either
pro- or anti-apoptotic
[224, 226]
. The challenge of future
studies is to gain a mechanistic understanding of the
direction of information flow, the dynamic nature of cross-
talk between signaling pathways, and the hierarchical
relationship between network components in response to
a distinct set of stimuli. Such mechanistic insights will
allow the generation of predictive (testable) models of how
this information processing capacity of signaling networks
is coopted in disease conditions to produce aberrant
phenotypes, and will lead to the identification of novel drug
targets and the development of more effective therapeutics.
In recent years it has been demonstrated, albeit on
a small scale, that systematically perturbing the compo-
nents of the network and simultaneously measuring
multiple quantitative phenotypes in the presence or absence
of specific input cues can be used to infer information flow
through signaling networks. The phenotypes measured can
include changes in gene expression, in phosphorylation of
key signaling or target proteins and/or in cellular
morphology. These quantitative phenotypes result from
multiplexed assays and are therefore different from the cell-
based assays used in RNAi HTS (described above), such
that, instead of measuring a single transcriptional reporter
or changes in the phosphorylation status of a single protein
in response to the knockdown of a gene, RNAi signatures/
phenoprints are composed of multiple measurements
ranging anywhere from tens to hundreds of genes or
proteins. A compilation of such quantitative phenotypes
e
239]
. The study
showed that genes known to be co-regulated could be easily
detected, and mutations in genes or treatments with small
molecules that regulate similar cellular processes displayed
e
