Biology Reference
In-Depth Information
and even substructures of organelles, such as the contact
sites between outer and inner mitochondrial membranes,
can be distinguished
[126]
. Another study integrated such
data with other MS-based datasets to comprehensively
identify chromosome-associated proteins across different
phases of the cell cycle
[127]
.
For systems biological applications it is attractive to
generate datasets of sufficient size to capture a reasonably
large part of the interactome. Because isotopic labeling,
which helps to ensure accurate quantitative data, is more
challenging to perform at a very large scale, recent high-
throughput protein
possible way. For protein
protein interaction data, one
critical parameter is the expression level of the bait protein.
Ideally, this should be adjusted to near-physiologic levels to
avoid aberrant localization and to ensure that bona fide
interaction partners are present in appropriate amounts
compared to the bait
[138]
. This can be achieved by tagging
the endogenous locus encoding the bait, which is straight-
forward in lower organisms such as yeasts, but much more
complicated in human cell lines. A recent method based on
bacterial artificial chromosome (BAC) transgenes allevi-
ates these limitations and allows the expression of GFP-
tagged proteins under fully endogenous gene regulatory
control
[139]
. In addition to providing a subcellular local-
ization tool via the fluorescent tag, this method can easily
be combined with quantitative interaction analysis, for
example to allow the splice isoform-specific interaction
partners of a bait protein to be identified
[140]
.BAC
e
protein interaction datasets mostly
employed simple label-free quantification methods, such as
counting the number of times peptides belonging to
a certain protein have been sequenced as a proxy for its
abundance (see
Box 1.1
). Based on this technology, large
scale protein
e
protein interaction datasets of human
[128,129]
and Drosophila
[130]
have been published.
Today, high-resolution MS data are routinely available and
can be analyzed with very sophisticated label-free quanti-
fication algorithms. As a result, datasets with much higher
true positive and lower false negative rates should become
available for systems biological modeling.
Beyond the goal of accurate and comprehensive
mapping of the interactome into lists of proteins associated
in complexes, the next challenge is to provide additional
functionally relevant information such as topology and
stoichiometry. One future direction involves the use of
chemical cross-linkers in combination with bioinformatic
algorithms to help deduce the three-dimensional architec-
ture of protein complexes
[131,132]
(
Figure 1.4
B). This
technology is still under development and currently limited
to purified complexes, but it has the potential to be
extended towards more complex samples, ultimately
offering the vision of 'the interactome in a single experi-
ment'. MS-based interaction proteomics is also uniquely
suited to measuring the dynamics of protein interactions in
response to stimuli. Such effects range from subtle modu-
lations of the interaction network, e.g. in the case of
autophagy-related proteins when that process was triggered
[133]
, or the changing composition of Wnt signaling
complexes
[134]
, to extensive disruption of complexes, e.g.
of the Bcr-Abl kinase complex after drug treatment
[135]
.
Accurate quantification is paramount when it comes to
dynamic interactions, and various groups have addressed
this with isotope-labeled reference peptides and absolute
quantification, which also allows estimation of the stoi-
chiometries of interacting proteins
[136,137]
. The ultimate
challenge in interaction proteomics is to achieve high
throughput and coverage while maintaining very high
quality standards. With the proteomics methods evolving
and quantification being increasingly accurate, the biolog-
ical samples from which the interactions are determined
should also represent
e
GFP
interaction data have also been combined with phenotypic
data from RNA interference screens to place genes
involved in mitosis into the context of protein complexes
[141]
. This study showed how physical interactions derived
from proteomics integrate beautifully with other omics
data, providing functional relationships of genes or
proteins. Consolidating physical with functional interaction
data will ultimately allow the placing of proteins into
complexes and arranging complexes into dynamic path-
ways and networks.
In this way ever-growing large-scale datasets will
become increasingly useful for biologists and systems
biologists alike (
Figure 1.4
C). Systems biologists will
better understand the intricate interplay of molecules inside
the cell, while biologists will find new interaction partners
of their protein of interest, and they will be able to place
specific genes into pathways, helping to explain observed
phenotypes.
However, the complete characterization of a mamma-
lian protein
e
protein interactome and its integration with
other omics data of the same scale is a vision of the future
and is just coming into reach for the most primitive
organisms
[142
e
144]
.
e
LARGE-SCALE DETERMINATION OF POST-
TRANSLATIONAL MODIFICATIONS
Post-translational modifications (PTMs) of proteins are
a key regulatory mechanism in signal transmission that
controls nearly all aspects of cellular function. Tradition-
ally, signaling processes are perceived as discrete linear
pathways that transduce external signals via the post-
translational modification of a few key sites. For example,
a specific phosphorylation event might regulate the func-
tion of a crucial pathway. These pathways have typically
been studied in the conventional, reductionist manner, with
the in vivo situation in the best
