ARTICLES
Identification of genes that regulate epithelial cell
migration using an siRNA screening approach
Kaylene J. Simpson
1,3
, Laura M. Selfors
1
, James Bui
1
, Angela Reynolds
2
, Devin Leake
2
, Anastasia Khvorova
2
and Joan S. Brugge
1,4
.
To provide a systematic analysis of genes that regulate epithelial cell migration, we performed a high throughput wound healing
screen with MCF-10A breast epithelial cells, using siRNAs targeting 1,081 human genes encoding phosphatases, kinases and
proteins predicted to influence cell migration and adhesion. The primary screen identified three categories of hits: those that
accelerate, those that inhibit and those that impair migration with associated effects on cell proliferation or metabolism. Extensive
validation of all the hits yielded 66 high confidence genes that, when downregulated, either accelerated or impaired migration;
42 of these high confidence genes have not been previously associated with motility or adhesion. Time-lapse video microscopy
revealed a broad spectrum of phenotypic changes involving alterations in the extent and nature of disruption of cell–cell adhesion,
directionality of motility, cell polarity and shape, and protrusion dynamics. Informatics analysis highlighted three major signalling
nodes, β-catenin, β1-integrin and actin, and a large proportion of the genes that accelerated migration impaired cell–cell adhesion.
Cell migration is central to development and tissue remodelling and
has a major role in cancer and metastasis. Many genes regulate mam-
malian cell migration; some are clearly involved in the mechanics
of the process, whereas others mediate receptor-driven regulatory
pathways. These genes have been identified by targeted investigations
or through unbiased genetic screens in Caenorhabditis elegans and
Drosophila melanogaster
1,2
. The development of small interfering RNA
(siRNA) technology has made it feasible to perform genetic screens
in mammalian cells.
We used a high throughput siRNA screening approach to identify
genes involved in epithelial cell migration, focusing on the generation
of a highly validated dataset. This contrasts with the general strategy
of selective analysis of a few strong hits with minimal validation of the
majority of primary hits
3–6
. We used an immortalized, non-tumorigenic
mammary epithelial cell line, MCF-10A, because of its strong migratory
response to EGF, its relative homogeneity, compared with tumour cell
lines, and its ability to form an epithelial sheet when confluent. The
screen was based on the classical scratch-wound assay
7
. This assay is
well recognized for assessing epithelial cell motility. It allows visualiza-
tion of morphological features and has the potential to identify siRNAs
that promote escape from the epithelial sheet, a property associated with
tumour-cell dissemination.
Three siRNA libraries were screened; the human phosphatase and kinase
libraries were chosen because these enzyme families regulate many cellular
pathways. The third, a custom library designed in collaboration with the
Geiger Laboratory (Weizmann Institute, Israel) targeted expanded family
members of genes with known or predicted roles in migration or adhesion
(migration and adhesion related, MAR).
We identified 34 genes that have a negative regulatory role in migra-
tion, 32 that have a positive role and 29 that affected cell metabolism.
Forty-two of the high confidence hits have not previously been impli-
cated in cell migration. Informatics analysis implicated many of the hits
in cell–cell and cell–matrix adhesion through involvement in β-catenin
and β1-integrin pathways, regulation of the actin cytoskeleton and
EGFR signalling. This screen provides a resource of high confidence
data, annotated at the level of migration effects, cell morphology fea-
tures, patterns of motility over time and pathway interactions. A fully
interactive database is hosted by the Cell Migration Consortium (www.
cellmigration.org/pubs/wound_rnai.htm).
RESULTS
Screen overview
To screen for motility in high throughput, we developed a robotic-driven
pin to deliver a precise scratch in confluent cell monolayers. The assay
conditions were established using siRNAs targeting RHOA and RAC1
(refs 8, 9) and wound healing was evaluated after 12 h when the mock-
transfected cells, migrating collectively as an epithelial cell sheet, close
the wound by 50–60% (Fig. 1; Supplementary Information, Fig. S1a).
The extent of motility, termed the ‘area score’, was numerically quanti-
fied and a visual evaluation was also performed to provide a secondary
1
Department of Cell Biology, Harvard Medical School, Boston, Massachusetts, 02115.
2
Dharmacon, Thermo Fisher Scientific, Colorado, 80026, USA.
3
Current
address: The Peter MacCallum Cancer Institute, Smorgon Family Building, St Andrews Place, East Melbourne 3002, Victoria, Australia
4
Correspondence should be addressed to J.S.B. (e-mail joan_brugge@hms.harvard.edu)
Received 21 April 2008; accepted 1 July 2008; published online 10 August 2008; DOI: 10.1038/ncb1762
NATURE CELL BIOLOGY VOLUME 10 | NUMBER 9 | SEPTEMBER 2008 1027
© 2008 Macmillan Publishers Limited. All rights reserved.