role in this process by promoting the establishment of multivalent
interactions. Highlighting the need for dynamic interactions within
RNP assemblies, mutations in the disordered regions of various
RBPs have been shown to alter phase separation by generating
toxic, aggregation-prone proteins inducing the abnormal compac-
tion of RNPs into pathological inclusions [133, 163].
Remarkably, RNA granules not only set the basis for efficient
and flexible compartmentalization of the cell cytoplasm, but are
themselves organized into subdomains that result from the differ-
ential clustering of RNAs and proteins [159, 164–167]. InDro-
sophilaoocyte, for example, in situ hybridization combined with
electron microscopy revealed that gurken and bicoid mRNAs
occupy distinct positions within P bodies: whilegurkenmRNA
was found enriched at the periphery,bicoidmRNA was found in
the central domain [165]. Interestingly, such a differential distribu-
tion reflects mRNA translational state, asgurkenmRNA associates
with its translational activator Orb at the edge of P bodies, where it
is translated. Furthermore, centrally localized and repressedbicoid
mRNA is released from P-bodies upon egg activation to become
actively translated.
Together, our understanding of the molecular bases of RNA–
protein recognition and assembly into RNP complexes has dramat-
ically improved over the last past years. Efforts have however to be
done to study RNA–protein interactions with a high spatiotempo-
ral resolution, in living cells or organisms [168]. As a first step
toward this goal, Singer and coworkers have combined endogenous
single RNA and protein detection with two-photon fluorescence
fluctuation analysis to directly measure the association of the trans-
lational repressor ZBP1 withβ-actinmRNA in living fibroblasts.
This revealed a stronger association between ZBP1 andβ-actin
mRNA at the nuclear periphery than at the leading edge, consistent
with the localized translation ofβ-actinat the front of migrating
cells [130].5 Perspectives
With the advent of transcriptomic methods and the concomitant
implementation of functional single-molecule imaging, our view
on the “central dogma of molecular biology” has changed dramat-
ically. Although it is now clear that RNA regulation is much more
complex that initially anticipated, and that RNA has a large range of
functions, methodological challenges are still ahead to continue
improving spatiotemporal detection of RNA molecules. Optimiza-
tion of spatially resolved fourth-generation sequencing technolo-
gies, for example, is needed to improve sensitivity and data
interpretation [169]. Furthermore, improvements have to be
made to visualize RNA molecules and regulatory partners in their18 Caroline Medioni and Florence Besse