Mapping the Genotype–Phenotype Relationship in Cellular Signaling Networks: Building Bridges Over the Unknown - Systems Biology: Networks, Models, and Applications - page 151

involves the use of microarrays, as mentioned above. With microar-
rays, genome-scale gene expression profiles are obtained and used to
characterize global function. For example, genes that are expressed
together are likely functionally related [33]. Simply, cellular mRNA
transcripts are extracted from cells and hybridize to cDNA or oligonu-
cleotides corresponding to the genes within the genome on extremely
dense arrays. The resulting data provide details of which genes are
activated or inactivated under various experimental conditions.
Although this approach is affected by the method of mRNA isolation,
among other factors, it has proven to be effective in constructing
complete phenotypic states [34,35]. Previous studies, for instance, have
been able to distinguish between different types of lymphoma via
expression profiling [13].
Experimental techniques based on the principle of RNAi may be the
most promising for perturbing a signaling network and measuring
subsequent intermediate and endpoint phenotypes [64]. These tech-
niques take advantage of the cell's natural response to double-stranded
RNA (dsRNA) (e.g., from a virus). After introduction of dsRNA with
a sequence similar to the target mRNA, an RNA-inducing silencing
complex (RISC) is activated and cleaves mRNA molecules with a com-
plementary sequence. Consequently, the function of a specific protein
corresponding to the targeted mRNA can be inhibited. The endpoint
phenotypes are then measured without the function of the target pro-
tein. This technique allows for tailored investigation of a signaling
network. One recent application of this technique categorized the
growth and viability of 91% of Drosophila genes [65].
Several of the technologies listed above can be integrated to evaluate
intermediate and endpoint phenotypes and characterize cellular sig-
naling networks. Two examples are presented here. First, the yeast
galactose utilization system was perturbed with a series of environmental
(e.g., growth with and without galactose) and genetic (e.g., gene deletion)
modifications [66]. Subsequently, global measurements of network
function were made (e.g., analysis of corresponding galactose utiliza-
tion genes with mRNA expression arrays). This integrated approach led
to several novel hypotheses and an extensive refinement of the galac-
tose utilization pathway in yeast. The second example integrated RNA
interference, mass spectrometry, and other techniques to generate a
model of the human TNF-a/NF-kB signaling reactions that accounted
for 221 molecular associations [29]. These studies are examples of how
several techniques described above can be integrated to probe cellular
signaling networks and how they are coupled with endpoint phenotypes.
What questions do these methods answer? All these techniques meas-
ure global effects of a protein's function, that is, how a signaling network
responds to the absence of a given protein. Certainly, combinations of
proteins may be perturbed leading to characterizations of how groups