Biology Reference
In-Depth Information
target genes by transcription factors or micro-RNAs (direct
functional relationship), and similarity of expression
profiles of genes across multiple conditions (profile-based
functional relationship) form interrelated interactome
networks. Study of these interactome networks, individu-
ally or together, is contributing key insights into the cellular
control of gene expression (see Chapters 2 and 4).
Material flow, such as in metabolic reactions, and
information flow, such as in transduction pathways, can be
represented mathematically by edge direction (see Chap-
ters 4, 5 and 11), while edge thickness, or weight, can
symbolize the relative strength of biological relationships.
Additionally, interactome network models can incorporate
logic (see Chapters 10 and 11) or dynamics (see Chapters
12, 13, 14 and 16). Eventually, the aim is to understand
how different interactomes are integrated together to form
the cellular systems that we believe underlie genotype
interaction between the
subunits of G-proteins,
which in turn rapidly switches entire signaling pathways on
or off
[25]
. Many protein interactions are weak and tran-
sient with high dissociation constants, as are associations
between membrane receptors and extracellular matrix
proteins that assist cellular motility
[25]
.
Forthcoming models of protein
a
and
b
protein interactomes
will undoubtedly involve sophisticated network represen-
tations integrating weighted and dynamic protein interac-
tions
[26]
. For protein interactions such as those involved in
signaling, interactome network models can also incorporate
edge direction. Many protein interactions, such as sub-
unit
e
subunit interactions within protein complexes, are
best described with undirected edges
[27
e
29]
. It is not
possible yet to assemble a proteome-scale interactome
network model that integrates strength, dynamics and
direction of edges because available technology is only
beginning to allow experimental measurements of such
interaction properties. Even a catalog of all possible protein
interactions has not yet been compiled for any single
species. Today's challenge lies in obtaining nearly
complete but static, undirected and unweighted reference
protein
e
e
phenotype relationships.
For this aim to be reached, complete single-color
interactome network maps first need to be assembled.
Physical interactions between proteins, or protein interac-
tions, constitute the fundamental backbone of the cell and
are instrumental for most biological processes, including
signaling, differentiation and cell fate determination. This
chapter describes the mapping and modeling of protein
protein interactome maps.
e
e
protein interactome networks, where edges connect pairs of
proteins that physically associate with one another directly
or indirectly.
Strategies for Large-Scale Protein
Protein
e
Interactome Mapping
Three fundamentally different but complementary strat-
egies have been deployed towards this goal: i) curation
of protein interaction data already available in the
scientific literature
[30]
; ii) computational predictions of
protein interactions based on available orthogonal infor-
mation, such as sequence similarity or the co-presence of
genes in sequenced genomes
[31]
; and iii) systematic,
unbiased high-throughput experimental mapping strate-
gies applied at the scale of whole genomes or proteomes
[32]
.
Literature-curated interactome maps present the
advantage of using already available, experimentally
derived information, but are limited by the inherently
variable quality of
TOWARDS A REFERENCE
PROTEIN
PROTEIN INTERACTOME MAP
Most individual proteins execute their biological functions
by interacting with one or several other proteins. Protein
interactions can form large protein complexes such as the
proteasome, in which ~50 protein subunits act together to
degrade other proteins and play a key role in cell protein
homeostasis. The existence of such molecular machines,
performing functions that no single protein can assume,
demonstrates that protein
e
protein interactomes exhibit
emergent properties beyond the sum of all
e
individual
35]
.A
randomly chosen set of literature-curated protein interac-
tions supported by a single publication was shown to be
approximately three times less reproducible than a refer-
ence set of manually curated protein interactions supported
by multiple publications
[36]
. Another caveat of literature-
curated interactome maps is that they mostly derive from
hypothesis-driven research, which often focuses on a few
proteins deemed to be scientifically interesting
[37]
. Some
'star proteins', such as the cancer-associated product of the
TP53 gene
[38]
, are interrogated for protein interactions
much more often than other proteins, resulting in an
artificial increase of their apparent connectivity relative to
the curation process
[33
e
protein interactions.
There is no such thing as a typical protein interaction.
Protein interactions occur in vivo with a wide range of
dissociation constants and dynamics. Proteins associated
by strong and permanent interactions with low dissociation
constants tend to form protein complexes. Protein interac-
tions may also be simultaneously strong and transient,
when controlled by the expression level of either interact-
ing partner using 'just-in-time' assembly, by a change in
subcellular location of one protein or the other, or by
conformation changes induced by post-translational
modification. For example, GTP hydrolysis controls the
