Introduction distinct functional domains such as epithelial cells


Most body cells are not uniform in size, shape and
function, but have distributed particular functions to specific areas of the
cell. In order to do that, cells need to establish an axis of polarity. Cell
polarity an essential feature of cells, required for developmental processes
such as asymmetric cell division, and specification and functioning of distinct functional
domains such as epithelial cells and neurons. Loss of polarity is associated
with development of cancer and other diseases. The nematode Caenorhabditis elegans is used as a
model system to understand the
molecular processes of polarity establishment and maintenance. Many
mechanisms driving cell polarity are conserved between this nematode and

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now



Over the past decades a number of
cortically localized proteins involved in polarity establishment have been found. Certain
protein complexes promote apical domain identity or basolateral identity and formation
of two distinct domains, together with other components establishing polarity. Mutual
exclusion of polarity regulators is a crucial mechanism in polarity
establishment. For example, the RhoGAP protein PAC-1 is involved in radial
polarization of the C. elegans early
embryo. PAC-1 localizes to sites of cell-cell contact and locally inactivates
CDC-42, thereby limiting active CDC-42 to outer cell surfaces. Active CDC-42
binds and recruits the polarity protein PAR-6, also restricting PAR-6
localization and leading to establishment of radial polarity. However, more
detailed mechanistic understanding of these kinds of processes is lacking. In addition, relatively
little is known about how cortical polarity integrates with other aspects of cell
polarity, such as cytoskeleton organisation. A full understanding of cell
polarity requires a comprehensive overview of protein-protein interactions that
are involved.



Koorman et al. (2016) studied the
protein network involved in cell polarity in C. elegans by coupling large-scale protein interaction mapping by
yeast two-hybrid (Y2H) screens to phenotypic characterization using RNA-mediated
interference (RNAi).

To generate the polarity interaction network using Y2H,
69 proteins were selected for
screening that are known to be essential for cell polarity in C. elegans or are homologs of polarity regulators in
other organisms. Besides a full-length construct for each protein, in total 338 fragment constructs were included
to map interaction domains and increase completeness of the interaction
network. Each construct was screened against a C. elegans AD-cDNA
library containing all genes and an AD-fragment library containing full-length genes and fragments involved in early embryonic
development. 439 interactions between 296 proteins were identified. 359/439 interactions were
detected using protein fragments and 54 interactions were previously described
(of which 19 in detail), indicating this approach indeed increased network

The quality of the network was validated
using computational and experimental approaches. Enrichment in GO terms, presence
in WormNET of protein pairs and enrichment for similar mRNA expression profiles
indicates functional associations. With co-affinity purification of 33
interactions 48% could be reproduced, comparable to literature values for
similar assays. Mammalian-protein-protein interaction trap (MAPPIT) showed the
interaction network retested at a similar level as literature-curated
interactions and a previously generated C.
elegans interaction dataset, confirming network quality.

The protein domains mediating the
interactions (minimal region of interaction, MRI) had an average length of 408
amino acids or 60% of the full protein length. 93% of 29 previously identified
MRIs contained the domain described in literature. The identified MRIs could
also be reproduced with co-affinity purification (10/19 tested positive),
confirming accuracy of the
MRIs. Some literature MRIs were smaller, as the approach used in this
study was not as sophisticated as literature methods. Or a short linear MRI could
be part of a larger and folded protein that was identified here.  

To select interaction pairs that are
most likely to be biologically relevant for cell polarity, the interaction map was combined
with phenotypic profiling by RNAi. Genes corresponding to proteins in the
network were inactivated and phenotypic effects were examined using nine
different fluorescent
markers of cell polarity. Markers were involved in early embryonic polarity,
epithelial cell
polarity, correct polarization of neurons and processes depending on polarity (yolk-protein uptake and receptor
in oocytes). 44 of the 69 bait proteins tested for 40 phenotypic defects
in all strains showed a defect in at least one strain. Some proteins
affected most tissues and others only one, so many genes have tissue-specific
roles. Alternatively, it could result from incomplete RNAi depletion. Next,
interacting proteins were screened in marker strains where bait RNAi caused a
defect. This revealed 100
protein pairs that displayed a similar phenotype in the same polarity-related

To examine whether overall in phenotype
is predictive of a true functional association, several analysis were
performed. Unbiased clustering of bait protein phenotypes showed several genes
known to function together, clustered together. Interacting proteins had 2.3
times more frequent
overlap in phenotype than non-interacting proteins in the network. In addition,
protein pairs with overlap in phenotype were more likely to share a similar
GO-term, less likely to share a non-similar GO-term and more likely to be
described in literature (9% versus 2% of the residual interactions).

The combination of Y2H and RNAi revealed
many interactions with proteins that were not known to act together. Some interacting
proteins led to a defect in the same tissue, but in distinct ways. Several
phenotypes were identified that have not been reported before.

Finally, the predictive capabilities of
the network were demonstrated by further examining the physical interaction and
clustered phenotypes of PAC-1 and PAR-6. Y2H with fragments showed the PDZ
domain of PAR-6 and PAC-1 region downstream the RhoGAP domain mediate binding. The
interaction was confirmed by co-affinity purification and co-localisation when
co-expressing PAR-6 and PAC-1, supporting a direct interaction. A mutant of PAC-1
lacking the PAR-6 binding domain could not mediate radial polarization, although
expression levels of mutant/non-mutant PAC-1 were similar. Thus, physical
interaction is important to establish radial polarization.



Overlapping phenotypes clustering with
known functional interactions, GO-term similarity and higher literature
description of the 100 protein pairs suggests a functional association in vivo, although the exact function
might not be immediately obvious. While not all phenotypes were the same, the
fact that they affect the same process is an indication that these interactions
are physiologically relevant.

With 9 markers used, not all polarity-related processes
were studied. Therefore, using other markers should lead to discovery of other protein
pairs with overlapping

Only 12 of the 19 interactions studied before
in detail were previously identified in large-scale studies. Hence, sophisticated
small-scale approaches can detect additional interactions not revealed in this
type of study.

The identification of candidate interactions
and their domains involved could serve as a starting point for further studies.
As many polarity establishment mechanisms are conserved between various
organisms, this interaction network can be used in other polarity systems as
well. Altogether, this may contribute to the understanding of the mechanisms
controlling cell polarity.