Biology Reference
In-Depth Information
Interaction (direct or mediated via a third protein) can be
inferred from a high coincidence rate, as the co-diffusion of
independent proteins is statistically rare.
WhileFRAPandFCSaimtoobtainmolecularinfor-
mation by observing the ensemble, they still cannot
observe different conformational states directly, or
distinguish between direct interactions and those mediated
via a large protein complex. Foerster resonance energy
transfer (FRET) measures molecular proximity by moni-
toring the far-field photophysical effects of the near-field
dipole
gene expression is a substantial factor of population
heterogeneity and a cause of variability in the cellular
phenotype. Interestingly, the noise in the allele transcrip-
tion in diploid cells is gene specific and not dependent on
the regulatory pathway or absolute rate of expression.
Moreover, mutations can alter the noise in gene expression,
suggesting that it is an evolvable trait that can be optimized
to balance fidelity and diversity in eukaryotic gene
expression
[28,29]
.
Methods that rely on the measurement of cell pop-
ulation averages provide a weighted summation over all
possible states. In contrast, single cell techniques such as
microscopy, flow cytometry and recently developed mass
cytometry can assess such variation by providing cellular
and, in the case of microscopy, subcellular information.
The first clear advantage is that multimodal populations can
be properly characterized. More importantly, single cell
observations of the state of a set of intracellular compo-
nents can provide information about the network connec-
tivity between them.
A common feature of almost all signaling networks is
the presence of feedback loops, and these are the basis for
any regulated response
[30,31]
. An aspect of this network
motif is that the response of any given protein within
a feedback loop contains information about the dynamics
of the network as a whole. Each protein can be considered
as an embedded probe that relays the coherent response of
the network. If the state of two proteins is monitored in
a single cell, their relationship will be dictated by the
connectivity in the network to which they belong. The
paired observations in multiple cells will therefore repre-
sent a statistical sample of the landscape of possible rela-
tions compatible with the underlying network. While causal
information cannot be derived from such correlations
(correlation does not imply causation), the number of
possible networks that can describe the system can be
significantly reduced.
To obtain causal connections from observational data,
extra information is needed. A priori knowledge, such as
partial causal connectivity (or lack of), can be used to derive
topological motifs. This has proved useful to detect positive
feedback loops in tyrosine phosphorylation networks even
in the case of the elusive autocatalytic loops
[32]
. Other
means of deriving causality are by acquiring and analyzing
richer datasets in which the time evolution of the system is
followed
[33]
. In such time-dependent datasets the obser-
vation of subsequent events allows the analysis of flow of
information and therefore enables the reconstruction of
causality that is missing from static data.
While causality can be derived from correlative data in
certain restricted cases, in general it must be derived from
data after perturbation of the system. Here, one or more
parameters of individual elements in a signaling network
are perturbed, such as the activity level or concentration of
dipole coupling between two fluorophores
(usually called donor and acceptor). Such effects include
the quenching of the donor fluorescence
[20]
, the sensi-
tized emission of the acceptor and the reduction in the
donor fluorescence lifetime
[21]
, and others
[22]
.As
energy transfer only occurs when the distance between
fluorophores is in the order of a few nanometers, FRET
effectively senses proximity in a volume (10
e
5
fL) rele-
vant to uncover interactions between proteins. In addition,
FREThasbeenusedasabasis for sensors that use
conformational changes to relay information about
activity, pH or concentration of molecular species
[23]
.
In summary, functional fluorescence microscopy allows
a cellular dynamic topographic map of proteins to be
overlaid with topological information on the causality that
determines protein state
[24]
. Such a state consists of
mobility and population evolution of the different inter-
acting or modified proteins. Altogether, this state is the
molecular basis of the cell-spanning patterns that generate
functionality on the micrometer scale
[25]
.
e
CAUSALITY FROM VARIATION
AND PERTURBATION ANALYSIS
Each cell is an individual entity that may respond to a signal
differently from its neighbors, even in a clonal population
exposed to the same environmental conditions. One of the
sources of such variation are the stochastic properties of
chemical reactions at low concentrations
[26]
. Many of the
components involved in signaling are in such low numbers
that actual reaction speeds can differ significantly from the
average. The other source of variation arises from extrinsic
factors such as its microenvironment. The accumulation of
small differences in a series of events leads to an overall
cell-to-cell variation in the internal state determining their
response properties
[27]
. Cellular heterogeneity has been
observed in a variety of cell types, ranging from bacteria to
mammalian cells.
For example, in gene expression the process by which
mRNAs and proteins are synthesized is inherently
stochastic. This stochastic nature introduces fluctuations
around the mean level of mRNA, causing identical copies
of a given gene to express at different levels. This noise in
