Mass Spectrometry in Systems Biology - Systems Biology: Networks, Models, and Applications - page 28

across four tissues tested the null hypothesis of no correlation between
detection of protein and mRNA expression level and rejected it for
426 out of 569 gene pairs. Further statistical modeling estimated the
specificity of mitochondrial gene products in all four tissues. While the
results of this study complemented the atlas of mitochondrial proteins,
the proteomics platform (based on single separation by reversed-phase
chromatography) emphasized the need for adequately tailored multi-
dimensional chromatography for unbiased sampling of the analytes.
Furthermore, the authors moved beyond mitochondrial gene products
and defined a metric (called the neighborhood index [92]) to explore
coregulated genes.
In addition to proteomics, many metabolomics studies include
relative quantitative data. An example of such a study was reported by
Weckwerth et al. [93]. In this study, using a robust GC-TOF/MS
system, more than 1000 metabolites were monitored by relative quan-
tification and compared between three different plant lines [93]. Three
parameters— m / z values, RT and chromatographic profile (that is, a
chromatospectrogram as defined under Platforms for Proteomics), and
in some cases MS/MS spectra of reference compounds—were used for
identification and quantification of plant metabolites without stable
isotope labeling. Processing of the data included deconvolution of
peaks, peak identification, and statistics of metabolite peaks. Based on
a large pool of metabolite pairs showing medium ( r

> 0.60) to strong

correlation ( r 0.80), the authors were able to distinguish between dif-

ferent plant lines or phenotypes.
These studies clearly outline the potential for integration of MS data
into systems biology approaches that include mass spectrometry data.
Washburn et al. [43] proved that integrative approaches can be success-
ful in building knowledge from protein pathways. Integration of
individual protein pathways (i.e., signaling, trafficking, etc.) in a compu-
tational model is limited only by our ability to test them experimentally
[94]. In this framework, signaling pathways detach themselves as real-
istic and intermediate steps in modeling a whole biological system.

>

Signaling Pathways
Cellular signaling usually translates information from extracellular
stimuli to the cell nucleus through specific pathways. Compensation
of signaling pathways (through interaction or crosstalk) defines the
biological response of the cell [1]. Proteins assemble in dynamic macro-
molecular complexes regulating the specificity of individual signaling
pathways (i.e., usually on the same scaffolding protein). Protein phos-
phorylation plays a prominent role in many pathways. Cascades of
kinase phosphorylation and phosphatase dephosphorylation act as
on/off switches in formation and dissociation of protein-protein inter-
action and in functional and enzymatic activity and consequently in