Assuming proteins are assembled in a modular fashion, it follows that one can apply principles from modular systems engineering. Thus, by redesigning or reverse engineering of known modular proteins, new synthetic, modular and functional polypeptides could be constructed. By attempting this we can fully explore whether modules have served as building blocks in the evolution of protein function, something that has been suggested from bioinformatics and experimental analysis, but remains to be substantiated for reverse engineered proteins. Such experiments will address the extent to which biological systems actually conform to this paradigm of modularity.
A system with logical, dynamical and programmable behaviour could
be assembled from well-defined components/parts (e.g. domains and
linear motifs). This approach is different from the classical molecular
biology method of creating chimeric proteins by changing an existing
system.
The biological potential of modularity in signaling proteins might be
explored by designing new proteins in a modular manner (i.e. not at an
atomic level). In other words, modules could be joined in an in silico
predicted sequence background . This is in contrast with more
traditional protein engineering approaches such as “rational design”,
“de novo” and “molecular evolution” An integrated approach of in silico and in vivo design, coupled with
directed evolution, might allow us to determine fundamental building
blocks of selected interaction networks. Engineered proteins can be
expressed and tested for predicted functions, refer to Figure 3, such
as activation of specific signalling pathways using phospho-specific
antibodies (for example to MAP kinases and STAT proteins). This could
be performed at the single cell level using flow cytometry with
phospho-specific antibodies. Another Approach would be to look at
markers of DNA synthesis/apoptosis. These synthetic proteins could
function as control units (gates and switches) that allow for switching
and gating in cellular or in vitro systems, thus allowing us to rewire
or probe signaling pathways.
Emergence in Protein Function We have previously discussed the
possibility that the juxtaposition of interaction domains and motifs in
novel combinations may contribute to the evolution of new biological
functions. The addition or removal of interaction modules can also be
achieved in real time through alternative splicing, which together with
post-translational modifications can greatly increase the numbers of
protein isoforms in any cell with different binding properties, and
potentially with different effects on regulatory networks. It remains
unclear how the addition/elimination of domains or motifs might modify
the overall complexity and function of a biological system, or might
this might give rise to emergence in a cellular system. Informatics
studies have shown that a general functional classification of a
protein sequence can be performed by looking at its set of “features”,
which includes post-translational modifications and other derived
parameters [47]. Complexity appears to arise partly as a result of
emergent properties of a system, but it is not clear whether our
current inventory of modules is sufficiently well defined to describe
this in quantitative terms. In this context, building new synthetic
systems with chimeric proteins, using the currently known set of
signalling components and interaction modules, will allow us to monitor
their ability to perturb complexity. Standard network parameters can be
measured experimentally, for instance connectivity and average cluster
size.
Specificity and Cross-talk in Signalling Pathways Transient
interactions between proteins play an important role in preventing
aberrant interactions between pathways in normal cells, and in
stimulating cross-talk when this is physiologically desirable. Since
these interactions frequently involve the recognition of peptide motifs
by modular interaction domains, it is critical to understand how short
linear peptides can confer specificity towards their cognate domain
partners, especially in cases where their binding affinity is modest.
In any one cell, many different members of a particular domain family
are likely expressed (yeast for example have 28 SH3 domains), which
might then compete for related ligands. Several mechanisms likely act
in combination to build specificity. One is expression and subcellular
localization – only when two proteins are co-expressed and co-localized
will an interaction take place. This allows for combinatorial effects,
since one domain may localize a protein, for example to a specific
membrane site, thereby directing a second domain to specifically
recognize a binding partner. Second, selectivity is driven by both
permissive and inhibitory forces, and a steric restriction on binding
non-physiological binding partners may therefore be just as important
as an ability to engage the appropriate target. Thus, a linear
proline-rich motif in the yeast Pbs2 protein, a MAP kinase scaffold,
engages the SH3 domain of the Sho1 osmosensor, but does not undergo
aberrant crosstalk with other yeast SH3 domains . However the Pbs2
motif does interact promiscuously with SH3 domains from non-yeast
species [48]. This has led to the suggestion that an interaction motif
need only discriminate between a set of domains that it meets in the
context of the cell in which it is expressed, or the subcellular
compartment in which it is localized [48]. Although some interaction
domains show a very specific interaction with a single target, this
usually applies only to proteins with highly specialized functions in
one or a few cell types. In a biological setting, most interaction
domains likely have numerous partners, which may differ according to
the cell type, or within a single cell depending on the environmental
conditions. Thus, it may be dangerous to assign an interaction domain
to a unique function in a cellular network, when its connectivity is
likely undergoing constant flux.
|