Enhanced genetic instability and dasatinib sensitivity in mammary tumor cells lacking NEDD9.

  
2010;70:8907-8916. Published OnlineFirst October 12, 2010.Cancer Res
 
Mahendra K. Singh, Eugene Izumchenko, Andres J. Klein-Szanto, et al.
 
Mammary Tumor Cells Lacking NEDD9
Enhanced Genetic Instability and Dasatinib Sensitivity in
 
 
 
 
Updated Version
10.1158/0008-5472.CAN-10-0353doi:
Access the most recent version of this article at:
 
 
Cited Articles
http://cancerres.aacrjournals.org/content/70/21/8907.full.html#ref-list-1
This article cites 29 articles, 8 of which you can access for free at:
 
 
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
.pubs@aacr.orgDepartment at
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
.permissions@aacr.orgDepartment at
To request permission to re-use all or part of this article, contact the AACR Publications
American Association for Cancer Research Copyright © 2010
on March 20, 2012cancerres.aacrjournals.orgDownloaded from
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353

Tumor and Stem Cell Biology
Enhanced Genetic Instability and Dasatinib Sensitivity in
Mammary Tumor Cells Lacking NEDD9
Mahendra K. Singh1, Eugene Izumchenko1,4, Andres J. Klein-S
Marina Wolfson4, and Erica A. Golemis1
Abstract
Elevated expression of the NEDD9/HEF1/Cas-L scaffo
metastasis in multiple cancer cell types. Conversely, gen
mary tumor virus (MMTV)-polyoma virus middle T (PyV
type. These activities arise from the role of NEDD9 in
cancer signaling proteins, including FAK, Src, Shc, and
expression as an unambiguous biomarker for tumor ag
om M
in an
fibrob
r tail
in an
al, an
wth. I
vels o
EDD
. ©201
enhanced
mors and
brain (rev
that a Ne
formed i
(MMTV)-
mammar
of integri
FAK and
involved
EDD9 in
nt resis-
NEDD9
, as has
ay have
and me-
ies have
cell mi-
egatively
pithelial
t Nedd9
tasis of
2). This
function
tered ex-
f tumor
lates the activity of the Aurora-A (AurA) kinase (5, 6, 13). Be-
Authors' A
2Pathology
Pennsylvan
the Negev,
Note: Curr
University o
Road, Gain
2860; Fax: 215-728-3616; E-mail: EA_Golemis@fccc.edu
Cancer
Research
8907
on March 20, 2012crjournals.org
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353 American Association for Cancer Research Copyright © 2010
cancerres.aaDownloaded from www.aacrjournals.orgcause of its role in timing mitotic progression, AurA levels
and activity are highly regulated within a narrow range, with
doi: 10.1158/0008-5472.CAN-10-0353
©2010 American Association for Cancer Research.growth. In potential support of the latter interpretation, we
have previously shown that NEDD9 directly binds and regu-
Corresponding Author: Erica A. Golemis, Fox Chase Cancer Center,
W406, 333 Cottman Avenue, Philadelphia, PA 19111. Phone: 215-728-tosis (7). Upregulation of NEDD9 is associated with
invasion and metastasis in human and murine tu-
cell lines including melanoma, lung, breast, and
iewed in ref. 8). Reciprocally, we recently reported
dd9−/− genotype increased the latency of tumors
n the aggressive mouse mammary tumor virus
polyoma virus middle T antigen (PyVT) mouse
y tumor model (9). NEDD9 is a central component
n-dependent signaling cascades that activate the
Src kinases to promote cell migration and is also
in lateral communication through Shc and other
support the straightforward interpretation that NE
tumor-promoting factor and that high levels of N
tumors correlate with poor prognosis and treatme
tance. As such, this would support adaptation of
expression as a biomarker for aggressive disease
been proposed (10).
However, some observations argue that NEDD9 m
a more complex role in tumor formation, invasion,
tastasis. For example, whereas the majority of stud
shown a positive effect of NEDD9 on regulation of
gration, two works have reported that NEDD9 can n
regulate the migration of MCF10A normal breast e
cells, and mRNA profiling experiments suggest tha
expression negatively correlates with the metas
MDA-MB-231 breast cancer cells to bone (11, 1
may reflect cell type–specific differences in NEDD9
or, alternatively, indicate that cellular response to al
pression of NEDD9 is dynamic at different stages o
ffiliations: Departments of 1Developmental Therapeutics,
, and 3Biostatistics, Fox Chase Cancer Center, Philadelphia,
ia and 4Faculty of Health Sciences, Ben Gurion University of
Beer Sheva, Israel
ent address for E. Izumchenko: Department of Biochemistry,
f Florida, Genetics/Cancer Research Complex, 1376 Mowry
esville, FL 32610.coordinates signaling cascades that control migration and
invasion (1–3), cell cycle transition through M-phase (4–6),
and apop
significantly reduced, as was the migratory and invasive
behavior of isolated cell lines (9). Taken in sum, these data
DD9 is athe initial delay in tumor growth, cells derived fr
hyperaggressive versus MMTV-PyVT;Nedd9+/+ cells
dimensional matrix produced by tumor-associated
orthotopic reinjection and of lung metastases afte
selection of MMTV-PyVT;Nedd9−/− cells for growth
Nedd9−/− cells have increased cell cycle, centrosom
increased selection of these cells for aggressive gro
MMTV-PyVT;Nedd9−/− cells persistently have low le
kinase inhibitor dasatinib. These studies identify N
mammary tumor growth. Cancer Res; 70(21); 8907–16
Introduction
The NEDD9/HEF1/Cas-L noncatalytic scaffolding proteinzanto2, Brian L. Egleston3,
lding protein promotes tumor cell invasion and
eration of mammary tumors in the mouse mam-
T) genetic model is delayed by a Nedd9−/− geno-
assembling complexes and supporting activity of
AKT, and would support evaluation of NEDD9
gressiveness. However, we here show that despite
MTV-PyVT;Nedd9−/− tumors are characteristically
chorage-independent growth, in growth on three-
lasts, and in formation of tumors after mammary
vein injection. This reversal suggests the specific
in vivo microenvironment. Indeed, MMTV-PyVT;
d mitotic defects, phenotypes compatible with the
ntriguingly, in spite of their aggressive phenotype,
f Src activation and are hypersensitive to the Src
9 as a complex modulator of different aspects of
0 AACR.
proteins to the Ras signaling cascades (8, 9). In MMTV-
PyVT;Nedd9−/− tumors, activation of FAK, Src, AKT, and
extracellular signal–regulated kinase (ERK) was typically

both overexpression and depletion causing centrosomal and
mitotic defects and contributing to genomic instability (14).
In cell culture experiments, we have determined that exoge-
nous overexpression or siRNA depletion of NEDD9 results in
similar cell cycle phenotypes, based on aberrant regulation
of 50,000 cells per well in 12-well plates directly onto tissue
culture plastic, NIH3T3-derived three-dimensional (3D)
Orthotopic and tail vein injections
Care of mice and injection protocols were approved by the
Fox Chase Cancer Center Institutional Animal Care and Use
Committee and followed the NIH guidelines. For orthotopic
injection, 1 × 106 cells in 200 μL of PBS were injected (bilat-
Singh et al.
8908
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353matrix, or TAF-derived 3D matrix. Cells were grown for 24,
48, 72, and 96 hours and then treated with 10% (v/v) Alamar
Blue solution (Invitrogen); fluorescence was measured with a
plate reader. For drug treatment experiments, cells were seed-
ed into 48- or 96-well plates. After 24 hours, vehicle (0.1%
DMSO) or drugs [dasatinib and erlotinib (obtained from the
Fox Chase Cancer Center pharmacy) and C1368 (Sigma)] were
added to themedium. After 72 hours, cell viability was assessed
by Alamar Blue assay. All assays were performed aminimum of
three times in triplicate. Cell cycle compartmentalization was
measured using a Guava Personal Cell Analysis-96 System
(Guava Technologies). Soft-agar experiments were performed
using standard techniques, as described in ref. 17. For rescue
experiments, cells were transfected with the plasmids pEGFP-
C1 vector (Clontech) containing Nedd9 and pCDNA3.1-mRFP
(Invitrogen) containing AurA to overexpress GFP-Nedd9 and
RFP-AurA, respectively.
Cancer Res; 70(21) November 1, 2010of AurA (5).
These data led us to hypothesize that although the ab-
sence of NEDD9 might initially delay tumor appearance, it
may also increase in vivo selection pressures affecting tumor
growth, thus reconciling contradictory observations. In this
study, we have analyzed early-passage cells derived from
the mammary tumors of MMTV-PyVT;Nedd9−/− and MMTV-
PyVT;Nedd9+/+ mice to test this idea. The results of this study
strongly support the idea that constitutive absence of Nedd9
in tumors induces defects in cell cycle associated with re-
duced function of AurA and a more aggressive tumor pheno-
type. Intriguingly, our data also indicate that the absence of
NEDD9 causes a persistent deficiency in Src activation and
influences the susceptibility of mammary cancer cell lines
to growth inhibition by dasatinib, a drug targeting Src kinase.
These studies in sum should significantly inform the use of
NEDD9 as a biomarker for clinical response.
Materials and Methods
Cell lines and cell growth measurements
The protocol for derivation of mammary tumor cell lines
from MMTV-PyVT;Nedd9−/− and MMTV-PyVT;Nedd9+/+ mice
was described in ref. 9. Basal proliferation rates for all cell lines
weremeasured by seeding 1 × 105 cells into 0.1% gelatin–coated
T25 flasks and then propagating them for up to 96 hours, after
which cells were trypsinized and counted using trypan blue
exclusion analysis at the times indicated. To contrast cell
proliferation on tissue culture plastic or on matrix produced
by NIH3T3 cells or tumor-associated fibroblasts (TAF) to
simulate an in vivo microenvironment, matrix was prepared
and measurements were made using protocols described in
detail in refs. 15, 16. Typically, cell lines were plated at a densityAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from eral inguinal) into the fourth mammary fat pad of severe
combined immunodeficient (SCID) mice (five mice per cell
line). Mice were palpated twice weekly for tumor onset. Tu-
mors were measured by caliper beginning 6 days after injec-
tion, and volume was calculated as width × length × 0.4. The
mice were euthanized by methoxy-fluorane (Metofane) inha-
lation when the longest dimension of the largest tumor
reached 1.5 cm or, alternatively, if mice exhibited signs of ill-
ness or distress. For each mouse, the tumor and lungs were
excised, divided in half, and processed either for Western
blot or pathologic analysis. Xenografted tumors and both
lungs were fixed in 10% phosphate-buffered formaldehyde
for 24 hours, embedded in paraffin, sectioned, and stained
with H&E. Three sections of each lung separated by 1 mm
were evaluated for metastases. Metastases were counted by
a pathologist (A.K.S.), using a Nikon Eclipse 50i microscope.
The surface area of the lungs was determined with a plani-
metric software (Image Pro Pus, Media Cybernetics). Metas-
tases were expressed as number of metastases per square
centimeter.
For tail vein injections into SCID mice, 0.35 × 106 cells sus-
pended in 200 μL of PBS were injected per mouse (five mice
per cell line). The SCID mice were monitored daily for signs
of developing tumor burden, such as weight loss, reduced
mobility, hunched posture, and ruffled fur. All mice were sac-
rificed at the end of week 3, when two mice showed signs of
breathing problems. For each mouse, the lungs were excised,
divided in half, and processed for Western blot or pathologic
analysis.
Biochemical analysis
For Western blot analysis, tumor sections histologically
confirmed to contain >90% tumor tissue were harvested, ho-
mogenized, and lysed in PBS-TDS buffer (1× PBS, 1% Triton
X-100, 0.1% SDS, 20% glycerol) containing complete protease
and phosphatase inhibitor cocktail (Roche Diagnostic).
Whole-cell lysates from the MMTV-PyVT;Nedd9−/− and
MMTV-PyVT;Nedd9+/+ cell lines were prepared using CelLytic
MT Mammalian Tissue Lysis/Extraction Reagent (Sigma).
The primary antibodies used were targeted to NEDD9 (2G9;
ref. 5; diluted 1:1,000), p130Cas (Santa Cruz Biotechnology;
1:1,000), Aurora (BD Biosciences; 1:1,000), phospho-Aurora
A T288 (Cell Signaling; 1:500), phospho-FAK-Y397 (1:2,000),
ShcA-Y317 (1:1,000), Src-Y418 (1:1,000), FAK (1:2,000), Src
(1:1,000), ShcA (1:1,000; Abcam), phospho-ERK1/2-T202/Y204
(1:500), AKT-S475 (1:1,000), AKT-T308 (1:500), ERK1/2
(1:1,000), AKT (1:1,000; Cell Signaling), phospho-Lyn-Y507
(1:500), Lyn (1:1,000; Cell Signaling), and β-actin (1:20,000;
Sigma). Secondary horseradish peroxidase–conjugated anti-
bodies were from Pierce Biotechnology. Proteins were visual-
ized using the West-Pico system (Pierce). Image analysis was
done using NIH ImageJ (NIH), with signal intensity normal-
ized to β-actin.
Cancer Research ciation for Cancer Research
on March 20, 2012rnals.org

For in vitro kinase assays, cell lysates prepared from
MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines
were immunoprecipitated using anti–Aurora A antibody
(BD Transduction Laboratories) and incubated with
[γ-32P]ATP and histone H3 (Upstate) substrate, and stan-
mor growth. However, continuing analysis of the data from
the study indicated that although the initial appearance of
MMTV-PyVT;Nedd9−/− tumors was delayed, once tumors
had attained palpable mass, the average time until eutha-
nasia was required because of tumor size was similar to
Nedd9 Modulates Tumor Aggressiveness
8909
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353dard methods were followed for kinase assay as described
in ref. 5. Aliquots of the reaction mix were used for SDS-
PAGE and Western blot analysis to confirm levels of AurA
or T288-AurA [using T288 phospho-specific antibody (1:500),
Cell Signaling].
Immunofluorescence
Epifluorescence microscopy was done using an inverted
Nikon TE300 microscope. All images were acquired as
12-bit images with a Spot RT monochrome camera (Diagnostic
Instruments). Secondary antibodies labeled with Alexa-488,
Alexa-568, and 4′,6-diamidino-2-phenylindole (DAPI; to
stain DNA) were from Molecular Probes/Invitrogen. To
measure cell spreading, cells were fixed in 4% paraformal-
dehyde for 10 min, permeabilized in 0.2% Triton X-100 for
5 min, and blocked with 3% bovine serum albumin (BSA) in
PBS. Anti-paxillin monoclonal antibodies (BD Transduction
laboratories) were used to mark focal adhesions and outline
cells. Cell area measurements were made for ∼100 cells per
experiment using MetaMorph or MetaVue software (Molec-
ular Devices, Universal Imaging) to score pixels within cell
perimeters.
To count centrosomes and analyze mitotic spindles, cells
were fixed with 4% paraformaldehyde followed by methanol;
permeabilized with 1% Triton-X100 in PBS; blocked in
1× PBS, 3% BSA; and incubated with antibodies to γ-tubulin
(Sigma; 1:200) and α-tubulin (Sigma; 1:800) to visualize cen-
trosomes and mitotic spindles, respectively.
Statistical analysis
We used Kaplan-Meier curves for survival outcomes. For
figures, we used generalized linear models with restricted
cubic splines estimated by generalized estimating equa-
tions. We also used semiparametric regressions and gener-
alized linear mixed effects models. Approaches are similar
to those detailed in ref. 9. Analyses were conducted in R
(R Foundation for Statistical Computing) and STATA 10
(Stata Corp.).
Results
Lack of NEDD9 increases the aggressiveness of selected
MMTV-PyVT mammary tumor cell lines
We previously compared mammary tumor development
in MMTV-PyVT;Nedd9+/+ and MMTV-PyVT;Nedd9−/− mice
(9). This work indicated that tumor development was very
significantly delayed in mice with an MMTV-PyVT;Nedd9−/−
genotype (P = 0.0001). Based on this observation and addi-
tional analysis indicating reduced activation of FAK, Src,
AKT, and ERK1/2 in MMTV-PyVT;Nedd9−/− versus MMTV-
PyVT;Nedd9+/+ tumor lysates (9) and reduced cell migration
by mammary tumor–derived cell lines lacking Nedd9, we
concluded that NEDD9 acted positively in promoting tu-
www.aacrjournals.orgAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from that of MMTV-PyVT;Nedd9+/+ animals (average, 129 versus
133 days; median, 128 versus 134 days; Fig. 1A).
This adaptation suggested that the absence of NEDD9
might select for compensatory changes that increase
tumor aggression. To test this idea, we used three MMTV-
PyVT;Nedd9+/+ and three MMTV-PyVT;Nedd9−/− tumor-
derived cell lines for orthotopic reinjection into mammary
fat pads and again monitored the growth of tumors (Fig.
1B and C). All three MMTV-PyVT;Nedd9−/− cell lines formed
palpable tumors within 14 days (median, 11 days), reaching
maximal size within 40 days after appearance. In contrast,
two of the three MMTV-PyVT;Nedd9+/+ cell lines yielded
mammary tumors that were detected within 55 days after
injection (median, 33 days) and grew more slowly, whereas
the third cell line failed to form any tumors over a
4-month period (Fig. 1C). In a complementary analysis,
we used four cell lines of each genotype to perform tail
vein injections followed by measurement of metastases in
the lung after 3 weeks. Strikingly, four of four injected
MMTV-PyVT;Nedd9−/− cell lines generated a large number
of metastases in 20 of 20 injected animals (with an average
of 73 metastases/cm2). In contrast, a single micrometasta-
sis was detected in 1 of 20 mice injected with one of the
four MMTV-PyVT;Nedd9+/+ cell lines assessed, whereas 19
of 20 mice had none at the same time point (Fig. 1D).
These data clearly supported the idea that the absence
of NEDD9 ultimately selected for more aggressive tumor
growth in vivo.
MMTV-PyVT;Nedd9−/− cells have a consistent growth
advantage on 3D matrices and in soft agar
We had previously determined that MMTV-PyVT;Nedd9−/−
cell lines were less migratory and more sensitive to
detachment-induced anoikis than MMTV-PyVT;Nedd9+/+
cell lines (9). Given these observations, the most plausible
explanation for the increased aggressiveness of the
MMTV-PyVT;Nedd9−/− cells in vivo would be if they had
undergone specific selection for proliferation in a tumor
microenvironment. To begin assessing this possibility, we
first compared the coefficient of variation in growth rate
in a larger panel of cell lines derived from the two geno-
types (Fig. 2A). This analysis indicated that among a group
of 12 cell lines, the range of doubling times of cells cultured
on plastic was significantly greater with the MMTV-PyVT;
Nedd9−/− genotype (P < 0.001), with some growing very rap-
idly and others growing relatively slowly.
Second, the growth of cells on 3D matrices secreted
either by immortalized fibroblast cell lines or by primary
fibroblasts that have been “primed” to support the growth
of tumor cells (i.e., TAFs) can simulate a stromal tumor
microenvironment (18). We compared the proliferation
rates of a set of MMTV-PyVT;Nedd9−/− versus MMTV-
PyVT;Nedd9+/+ cell lines cultured on tissue culture plastic,
Cancer Res; 70(21) November 1, 2010 ciation for Cancer Research
on March 20, 2012rnals.org

Singh et al.
8910
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353American Asso Copyright © 2010
cancerres.aacrjouDownloaded from on 3D matrix prepared from NIH3T3 cells, and on 3D
matrix generated from TAFs (Fig. 2B). Strikingly, although
there was no overall statistically significant difference
between the two genotypic groups cultured on plastic (as
both fast-growing and slow-growing MMTV-PyVT;Nedd9−/−
lines were assessed), all MMTV-PyVT;Nedd9−/− cell lines grew
much faster on NIH3T3- or TAF-derived matrix (P < 0.001).
Third, we also compared the growth of MMTV-PyVT;
Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines cultured in
soft agar to fully assess anchorage-independent growth. In
this case, 3 of 4 MMTV-PyVT;Nedd9−/− cell lines efficiently
formed large colonies in soft agar, whereas 0 of 4 MMTV-
PyVT;Nedd9+/+ cell lines formed large colonies (Fig. 2C). The
increased growth rate of MMTV-PyVT;Nedd9−/− versus MMTV-
PyVT;Nedd9+/+ cells on 3D matrices, in soft agar, and in vivo,
coupled with the greater growth variability of MMTV-PyVT;
Nedd9−/− cell lines on tissue culture plastic, suggested that
MMTV-PyVT;Nedd9−/− cell lines may have been more subject
to selective pressures for growth in the in vivo microenviron-
ment, particularly as the effects were seen in early-passage
(<6) populations of cells recovered from tumors.
MMTV-PyVT;Nedd9−/− cells have more cell cycle,
spindle, and centrosome abnormalities than
MMTV-PyVT;Nedd9+/+ cells
To begin to establish the basis for the increased variabil-
ity in the growth of MMTV-PyVT;Nedd9−/− cells, we directly
assessed whether these cells had signs of increased ge-
nomic instability and aneuploidy. Indeed, flow cytometric
analysis (Fig. 3A and data not shown) indicated a greater
proportion of >4N cells and more abnormal profiles in
cells with the MMTV-PyVT;Nedd9−/− genotype. These cells
typically had a much greater number of abnormal mitoses
than MMTV-PyVT;Nedd9+/+ lines, including both monopolar
(19% versus 13%, P = 0.002) and multipolar (67% versus
38%, P = 0.001) spindles (Fig. 3B), with consistent increases
observed in all cell lines of the MMTV-PyVT;Nedd9−/−
genotype. Commensurate with this observation, a higher
proportion of these cells had supernumerary centrosomes
(Fig. 3C).
We have previously shown that NEDD9 binds and
controls the activation of the mitotic kinase AurA (5, 6).
Defects in AurA activation are often associated with
Figure 1. Comparison of the in vivo growth of MMTV-PyVT;Nedd9+/+ versus MMTV-PyVT;Nedd9−/− primary tumors and derivative cell lines. A, Kaplan-Meier
curve showing insignificant increase from time of first appearance of tumors to maximal growth before euthanasia in MMTV-PyVT;Nedd9−/− (n = 26)
versus MMTV-PyVT;Nedd9+/+ (n = 21) mice (P = 0.181). B, independent cell lines derived from MMTV-PyVT;Nedd9−/− and MMTV-PyVT;Nedd9+/+ primary
mammary tumors (3 per genotype) were injected orthotopically into the mammary pads of SCID mice (n = 5 per cell line; 30 mice total). Data shown
represent significant difference in time until appearance of tumors (P = 0.001). C, significant decrease in total tumor burden in SCID mice with tumors
derived from MMTV-PyVT;Nedd9+/+ versus MMTV-PyVT;Nedd9−/− cell lines (P ≤ 0.001). D, MMTV-PyVT;Nedd9−/− cell lines more effectively form lung
metastases following tail vein injection. Left, quantification of lung metastases per square centimeter, averaged from five mice for each of eight cell lines
21 d after tail vein injection. Difference between genotypic groups is highly significant (P < 0.0001). Right, H&E-stained lungs derived from SCID mice
3 wk after tail vein injection of representative MMTV-PyVT;Nedd9−/− and MMTV-PyVT;Nedd9+/+ cell lines (magnification, ×10).
Cancer Res; 70(21) November 1, 2010 Cancer Research ciation for Cancer Research
on March 20, 2012rnals.org

centrosomal abnormalities and aneuploidy in human can-
cers. Intriguingly, we found that the levels of AurA kinase
activity were sharply reduced in MMTV-PyVT;Nedd9−/− cells
(Fig. 3D, top). Further, and unexpectedly, this activity drop
kinase in this genetic background, rather than a propor-
tional reduction in AurA activity (Fig. 3D, middle). A
NEDD9-dependent decrease in AurA might provide an ini-
tiating event for the genomic instability manifested by the
(>150 pixels); diamonds, small colonies (<150 pixels). Difference between genotypic groups is highly significant (P = 0.001). D, representative pictures of
colonies formed on soft agar by MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines.
Nedd9 Modulates Tumor Aggressiveness
8911
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353www.aacrjournals.orgAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from Cancer Res; 70(21) November 1, 2010correlated with a reduction in the overall level of AurA MMTV-PyVT;Nedd9−/− cells. However, overexpression of
Figure 2. Comparative growth of MMTV-PyVT;Nedd9+/+ versus MMTV-PyVT;Nedd9−/− cell lines in vitro. A, doubling time of seven MMTV-PyVT;Nedd9−/−
versus five MMTV-PyVT;Nedd9+/+ cell lines grown on tissue culture plastic indicates greater overall variability in growth rates associated with absence
of Nedd9 (P < 0.0001). B, the ratio of growth rate of three MMTV-PyVT;Nedd9−/− versus three MMTV-PyVT;Nedd9+/+ cell lines grown on tissue
culture plastic versus matrix derived from NIH3T3 cells or TAFs indicates a significant difference at 90 h (for each, P < 0.001). C, colony formation on
soft agar 3 wk after plating of 2,500 cells per well. Experiments were done in quadruplicate at least three times independently. Squares, large colonies ciation for Cancer Research
on March 20, 2012rnals.org

Singh et al.
8912
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353American Asso Copyright © 2010
cancerres.aacrjouDownloaded from neither GFP-Nedd9 nor RFP-AurA in these cell lines re-
verses the centrosomal abnormalities, implying that at
minimum, other changes in the regulatory machinery
had become critical (Fig. 3D, bottom).
The MMTV-PyVT;Nedd9−/− genotype is sensitized
to dasatinib
These data in sum showed that the absence of NEDD9
resulted in a very different profile of mammary tumor cell
Figure 3. Cell cycle abnormalities in MMTV-PyVT;Nedd9+/+ versus MMTV-PyVT;Nedd9−/− cell lines. A, fluorescence-activated cell sorting analysis of
asynchronously growing MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines indicates more cells with 4N or greater DNA content. B, mitotic
profiles in asynchronously growing cells show fewer normal bipolar spindles (P ≤ 0.001) and more multipolar (P = 0.001) and monopolar (P = 0.002)
spindles in MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines. More than 50 mitoses were counted per cell line in each of three independent
experiments. Representative cells were visualized with DAPI (blue), α-tubulin (red), and γ-tubulin (green). Bar, 5 μm. C, increased frequency of
supernumerary centrosomes in MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines (P ≤ 0.001). More than 100 cells were counted per cell
line in each of three independent experiments. Images indicate representative cells with normal centrosome count (left) and supernumerary centrosomes
(right), visualized with DAPI (blue) and γ-tubulin (green). Bar, 10 μm. D, IP:AurA, immunoprecipitation of total AurA from cell lines followed by in vitro
kinase assay with phosphorylated histone H3 as substrate visualized by [γ-32P]ATP and with immunoprecipitated, autophosphorylated (T288) AurA, total
AurA, and H3 substrates visualized by Western blot analysis. WCL, Western blot analysis of T288-phospho-AurA and total AurA in whole-cell lysates
(WCL). Bottom, the change in the proportion of MMTV-PyVT;Nedd9−/− cells with >2 centrosomes following expression of GFP (G) versus GFP-Nedd9
(G-Nedd9) or RFP (R) versus RFP-AurA (R-AurA) is insignificant (P > 0.1).
Cancer Res; 70(21) November 1, 2010 Cancer Research ciation for Cancer Research
on March 20, 2012rnals.org

growth, and signs of genomic instability were associated with
loss of AurA. We had previously shown that Src and FAK had
consistently reduced activity in the tumors and derivative
cell lines of MMTV-PyVT;Nedd9−/− mice, whereas Shc, ERK,
mammary tumor cells positive or negative for NEDD9
displayed different profiles for response to drugs targeting
proteins with which NEDD9 functionally interacts.
Strikingly, MMTV-PyVT;Nedd9−/− cell lines were much more
(not shown). C, immunoblots of tumor lysates prepared from tumors generated by mammary orthotopic reinjection of the cell line indicated on top. Each
lane represents an independent tumor arising from the noted cell line. D, 3D TAF matrix rescues the dasatinib sensitivity of MMTV-PyVT;Nedd9−/− cell
atrix g
erence
ell lines
Nedd9 Modulates Tumor Aggressiveness
8913
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353lines. Top, experiment as described in A in eight cell lines plated on 3D m
genotypes and conditions indicated. Taking 1 μmol/L as an example, diff
on 3D TAF matrix (P = 0.033). Bottom, fold difference in IC50 values for c
www.aacrjournals.orgAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from enerated from TAFs. Center, IC50 curves represented as averages for the
s in IC50 between genotypes are significant on plastic (P = 0.002) and
indicated in plating on TAF matrix versus tissue culture plastic.
Cancer Res; 70(21) November 1, 2010and AKT had statistically significant reduced activity in
primary MMTV-PyVT;Nedd9−/− tumors, but greater tumor-
to-tumor variability (9). We therefore assessed whether
sensitive to dasatinib, an inhibitor of the Src kinases, thanwere
MMTV-PyVT;Nedd9+/+ cells (Fig. 4A). This correlated not only
with a significantly decreased basal ability of MMTV-PyVT;
Figure 4. Response to dasatinib predicted by Nedd9 status and compensated by 3D TAF matrix. A, IC50 curve calculations for four dasatinib-treated
MMTV-PyVT;Nedd9−/− versus four MMTV-PyVT;Nedd9+/+ cell lines. Cell lines analyzed in quadruplicate in three independent experiments. B, comparative
rate of attachment of MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell lines plated in dasatinib. Dasatinib-dependent decrease in spreading
differs significantly at 6 h (P = 0.001) and 24 h (P = 0.016). At 24 h, proportionately fewer MMTV-PyVT;Nedd9−/− cells remained attached to the plate ciation for Cancer Research
on March 20, 2012rnals.org

or EG
Singh et al.
8914
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353Figure 5. Nedd9 status does not affect the response to inhibitors of AurA
C1368 for two cell lines of each genotype; resulting data were averagedAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from FR. A, experiment as described in Fig. 4A with the AurA inhibitor
11 versus 2.21 μmol/L; not significant). Similar results were obtainedNedd9−/− cell lines to spread but also with enhanced inhibition
of spreading following plating in low doses of dasatinib, indi-
cating an inhibition of cytoskeletal rearrangement (Fig. 4B), as
well as with the depressed activation of FAKwe had previously
noted in these cell lines (9). Reanalysis of FAK and Src activa-
tion in lysates from the primary tumors arising from orthoto-
pic reinjection (Fig. 1B) indicated that activation of Src
remained consistently significantly depressed in tumors of
the MMTV-PyVT;Nedd9−/− genotype, although FAK activation
was similar in tumors of both genotypes (Fig. 4C).
These data suggest that MMTV-PyVT;Nedd9−/− cells might
be specifically vulnerable to additional dasatinib-based
targeting of Src because of predisposing prior reduction in
pathway activity arising from the absence of the Src partner
NEDD9. If so, and given the prior data suggesting that
MMTV-PyVT;Nedd9−/− cells were more adapted to extrinsic
microenvironment, 3D matrix cues might be expected to
rescue cell growth. Strikingly, plating of cell lines on 3D TAF
matrix produced markedly different results between MMTV-
PyVT;Nedd9−/− and MMTV-PyVT;Nedd9+/+ cell lines (compare
Fig. 4D, top and center, with 4A). Although the MMTV-PyVT;
Nedd9−/− lines remained more sensitive to dasatinib, overall
IC50 values were significantly increased in 3 of 4 lines (Fig.
4D, bottom). In contrast, MMTV-PyVT;Nedd9+/+ cell lines as a
group had little responsiveness to 3D TAF matrix, with 3 of 4
lines instead becoming more sensitive to dasatinib.
We also compared the responsiveness of MMTV-PyVT;
Nedd9−/− and MMTV-PyVT;Nedd9+/+ cells to C1368, which tar-
gets the AurA kinase (Fig. 5A), and erlotinib, which targets
epidermal growth factor receptor (EGFR)–dependent signal-
ing (Fig. 5B). In neither case was a specific difference ob-
served, suggesting that the sensitization we observed was
specific for Src. Interestingly, although the activity of the
EGFR effectors ERK, Shc, and AKT had also been depressed
in primary tumors, as in the case with FAK, this was not
observed in tumors arising from orthotopic reinjection (Fig.
5C), in which levels were comparable or higher than that in
MMTV-PyVT;Nedd9−/− cells, and some tumors showed
evidence of recovery of AurA levels (not shown).
Finally, although Src is often considered the primary
physiologic target of dasatinib in solid tumors (19), this
drug can affect other relevant cellular kinases including
Src-family kinases such as LYN (20) as well as c-Kit (19),
EGFR (19), Ephrin (21), and others. Given the close physical
interactions between Nedd9 and Src kinases (1–3), one pos-
sibility was that tumors developing in the absence of
NEDD9 might affect the basal expression or activity of
Src-related kinases. We determined that Src and Lyn were
(IC50 1.
with a second AurA inhibitor, PHA 680632 (results not shown). B, experiment as described in A with the EGFR inhibitor erlotinib (IC50 1.29 versus
1.37 μmol/L; not significant) indicates no cell genotype–dependent difference in IC50 value. C, immunoblots of tumor lysates as described in Fig. 4C.
D, expression and activation (reflected by levels of autophosphorylated forms) of Src and Lyn in MMTV-PyVT;Nedd9−/− versus MMTV-PyVT;Nedd9+/+ cell
lines treated for 2 h with 200 nmol/L dasatinib.
Cancer Res; 70(21) November 1, 2010 Cancer Research ciation for Cancer Research
on March 20, 2012rnals.org

similarly expressed in MMTV-PyVT;Nedd9−/− and MMTV-
PyVT;Nedd9+/+ cells (Fig. 5D) and also established that the
activation of these kinases was similarly inhibited by dasa-
tinib. The expression and activity of a number of other
dasatinib targets (including EGFR and c-Kit) were also
The variant proliferation rates, increased percentage of
cells to the Src-family targeting agent dasatinib is particularly
significant, given the increasing use of dasatinib in the clinic
for treatment of breast (28) and other cancers. MMTV-PyVT;
Nedd9−/− cells maintain persistently low levels of active Src,
suggesting the “double hit” of dasatinib and that a MMTV-
whereas other populations may be more vulnerable. Con-
Nedd9 Modulates Tumor Aggressiveness
8915
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353cells characterized by supernumerary centrosomes, and
anomalous cell cycle compartmentalization in MMTV-
PyVT;Nedd9−/− tumor-derived cell lines suggest that in each
case, the Nedd9−/− genotype has provided a selection for
mutations in the tumor population. The most direct expla-
nation for this selection lies in the deficient expression and
activation of AurA in Nedd9−/− cells. AurA activation is re-
quired for mitotic entry and transit, and we have shown
that siRNA depletion of NEDD9 impairs AurA activation
and mitotic progression, leading to the accumulation of an-
euploid binucleate cells (4, 5). Our present data show that
genetic loss of NEDD9 results in a similar phenotype, in
which context it is remarkable that the consequences of a
Nedd9−/− genotype are predominantly restricted to effects
on tumorigenesis. We suggest that this may reflect a specif-
ic requirement for Nedd9 scaffolding function in sustaining
the activity and expression of some of its partner proteins
in the constitutive growth environment of a tumor. By con-
trast, the transient activation of focal adhesion–associated
proteins, such as Src or FAK, in a normally growing or qui-
escent cellular context may not require NEDD9 interactions
but may be supported by the NEDD9 paralogue BCAR1/
p130Cas (discussed in ref. 22), whereas AurA activation
may be supported by its other partners, such as Ajuba
(23) and Tpx2 (24). The fact that very significant NEDD9
upregulation has been observed in a large subset of aggres-
sive tumors from the lung, breast, and brain (refs. 10, 25, 26;
discussed in ref. 22) and head and neck (27), whereas
BCAR1, Ajuba, FAK, and Tpx2 have not been reported to
be thus overexpressed, is compatible with such a model.
Identifying effective biomarkers for cancer prognosis and
drug responsiveness is of great importance in improving the
clinical management of cancer. Whereas earlier studies fo-
cusing on NEDD9 upregulation in tumors suggest that high
NEDD9 levels are specifically associated with poor prognosis
and metastasis, the work presented here and in ref. 9 indi-
cates a more nuanced role for NEDD9 in mammary tumor
biology, in which both elevated or absent levels of Nedd9
may be associated with aggressive tumor phenotypes. This
argues against the straightforward consideration of high
NEDD9 levels in tumors as a predictive biomarker. However,
the observation that the absence of NEDD9 greatly sensitizes
References
1. Fashena SJ, Einarson MB, O'Neill GM, Patriotis CP, Golemis EA. Dis-
section of HEF1-dependent functions in motility and transcriptional
regulation. J Cell Sci 2002;115:99–111.
www.aacrjournals.orgAmerican Asso Copyright © 2010
cancerres.aacrjouDownloaded from versely, the subset of metastatic tumors overexpressing
NEDD9 (8, 10, 25, 26, 29) may be particularly resistant to
Src-targeting agents and independent of microenviron-
ment-associated survival cues.
MMTV-PyVT;Nedd9−/− cells do not have increased sensitiv-
ity to EGFR- or AurA-targeting agents. In contrast to the
situation with Src, the cells have overcome the inhibition
of the EGFR effectors AKT and ERK observed in primary tu-
mor lysates. Whereas the levels of AurA remain depressed in
these tumors, much of the actual cell killing associated with
Aurora-inhibiting small molecules has been proposed to be
associated with inhibition of Aurora-B, rather than AurA
kinase (30), compatible with our observation that reductions
in AurA expression do not correlate with increased sensitivity
to Aurora-targeting inhibitors. Based on these data, we sug-
gest that NEDD9 may be an excellent candidate to examine
as a modulator of response to dasatinib and potentially other
Src kinase inhibitors in the ongoing clinical development of
this and other Src-targeting agents.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Acknowledgments
We thank Sharon Howard, Anthony Lerro, and Jackie Valvardi for technical
help with cell line generation and animal work; Sachiko Seo for the gift of
original Nedd9−/− mice, which allowed derivation of the cell lines; Nicolas G.
Day for assistance with soft agar and xenograft assays and graphical rendering
of data; Ilya Serebriiskii, Edna Cukierman, and Natalia Skobeleva for guidance
in some experiments and analysis; and Denise Connolly and Joy Little for
critical comments on the manuscript.
Grant Support
NIH grants R01-CA63366 and R01-CA113342; Army Materiel Command
grant W81XWH-07-1-0676; the Pennsylvania Tobacco Settlement funding
(E.A. Golemis); the Israel Cancer Association and Stanley Abersur Research
Foundation (M. Wolfson); NIH core grant CA-06927 and the Pew Charitable
Trust (Fox Chase Cancer Center); and the American Associates, Ben-Gurion
University of the Negev (E. Izumchenko).
The costs of publication of this article were defrayed in part by the payment
of page charges. This article must therefore be hereby marked advertisement in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 01/28/2010; revised 06/14/2010; accepted 08/23/2010; published
OnlineFirst 10/12/2010.
2. van Seventer GA, Salman HJ, Law SF, et al. Focal adhesion kinase
regulates β1 integrin dependent migration through an HEF1 effector
pathway. Eur J Immunol 2001;31:1417–27.
Cancer Res; 70(21) November 1, 2010Discussion sensitivity by the 3D matrix produced by TAFs suggests that
in vivo, specific niches of tumor cells may be protected,unaffected by Nedd9 genotype (not shown). PyVT;Nedd9−/− genotype is particularly deleterious to cell sur-
vival signaling networks. In this context, the modulation of ciation for Cancer Research
on March 20, 2012rnals.org

3. Liu X, Elia AEH, Law SF, Golemis EA, Farley J, Wang T. A novel ability
of Smad3 to regulate proteasomal degradation of a Cas family mem-
ber, HEF1. EMBO J 2000;19:6759–69.
4. Dadke D, Jarnik M, Pugacheva EN, Singh MK, Golemis EA. Deregu-
lation of HEF1 impairs M-phase progression by disrupting the RhoA
activation cycle. Mol Biol Cell 2006;17:1204–17.
5. Pugacheva EN, Golemis EA. The focal adhesion scaffolding protein
HEF1 regulates activation of the Aurora-A and Nek2 kinases at the
centrosome. Nat Cell Biol 2005;7:937–46.
6. Pugacheva EN, Golemis EA. HEF1-aurora A interactions: points of
dialog between the cell cycle and cell attachment signaling networks.
Cell Cycle 2006;5:384–91.
7. O'Neill GM, Golemis EA. Proteolysis of the docking protein HEF1 and
implications for focal adhesion dynamics. Mol Cell Biol 2001;21:
5094–108.
8. O'Neill GM, Seo S, Serebriiskii IG, Lessin SR, Golemis EA. A new
central scaffold for metastasis: parsing HEF1/Cas-L/NEDD9. Cancer
Res 2007;67:8975–9.
9. Izumchenko E, Singh MK, Plotnikova OV, et al. NEDD9 promotes on-
cogenic signaling in mammary tumor development. Cancer Res
2009;69:7198–206.
10. Kim M, Gans JD, Nogueira C, et al. Comparative oncogenomics
identifies NEDD9 as a melanoma metastasis gene. Cell 2006;125:
1269–81.
11. Simpson KJ, Selfors LM, Bui J, et al. Identification of genes that
regulate epithelial cell migration using an siRNA screening ap-
proach. Nat Cell Biol 2008;10:1027–38.
12. Minn AJ, Gupta GP, Siegel PM, et al. Genes that mediate breast can-
cer metastasis to lung. Nature 2005;436:518–24.
13. Pugacheva EN, Jablonski SA, Hartman TR, Henske EP, Golemis EA.
HEF1-dependent Aurora A activation induces disassembly of the pri-
mary cilium. Cell 2007;129:1351–63.
14. Marumoto T, Zhang D, Saya H. Aurora-A—a guardian of poles. Nat
Rev Cancer 2005;5:42–50.
15. Serebriiskii I, Castello-Cros R, Lamb A, Golemis EA, Cukierman E.
in vivo synergy of MCP compounds with mitogen-activated protein
kinase pathway- and microtubule-targeting inhibitors. Mol Cancer
Ther 2007;6:898–906.
18. Amatangelo MD, Bassi DE, Klein-Szanto AJ, Cukierman E. Stroma-
derived three-dimensional matrices are necessary and sufficient to
promote desmoplastic differentiation of normal fibroblasts. Am J
Pathol 2005;167:475–88.
19. Lombardo LJ, Lee FY, Chen P, et al. Discovery of N-(2-chloro-
6-methyl-phenyl)-2-(6-(4-(2-hydroxyethyl)-piperazin-1-yl)-2-methyl-
pyrimidin-4-ylamino)thiazole-5-carboxamide (BMS-354825), a dual
Src/Abl kinase inhibitor with potent antitumor activity in preclinical
assays. J Med Chem 2004;47:6658–61.
20. Choi YL, Bocanegra M, Kwon MJ, et al. LYN is a mediator of
epithelial-mesenchymal transition and a target of dasatinib in
breast cancer. Cancer Res 2010;70:2296–306.
21. Huang F, Reeves K, Han X, et al. Identification of candidate molec-
ular markers predicting sensitivity in solid tumors to dasatinib: ratio-
nale for patient selection. Cancer Res 2007;67:2226–38.
22. Tikhmyanova N, Little JL, Golemis EA. CAS proteins in normal and
pathological cell growth control. Cell Mol Life Sci 2010;67:1025–48.
23. Hirota T, Kunitoku N, Sasayama T, et al. Aurora-A and an interacting
activator, the LIM protein Ajuba, are required for mitotic commitment
in human cells. Cell 2003;114:585–98.
24. Bayliss R, Sardon T, Vernos I, Conti E. Structural basis of Aurora-A
activation by TPX2 at the mitotic spindle. Mol Cell 2003;12:851–62.
25. Natarajan M, Stewart JE, Golemis EA, et al. HEF1 is a necessary and
specific downstream effector of FAK that promotes the migration of
glioblastoma cells. Oncogene 2006;25:1721–32.
26. Ji H, Ramsey MR, Hayes DN, et al. LKB1 modulates lung cancer dif-
ferentiation and metastasis. Nature 2007;448:807–10.
27. Lucas JT, Jr., Salimath BP, Slomiany MG, Rosenzweig SA. Regula-
tion of invasive behavior by vascular endothelial growth factor is
HEF1-dependent. Oncogene Epub 2010 May 24.
Singh et al.
8916
Published OnlineFirst October 12, 2010; DOI:10.1158/0008-5472.CAN-10-0353Fibroblast-derived 3D matrix differentially regulates the growth and
drug-responsiveness of human cancer cells. Matrix Biol 2008;27:
573–85.
16. Castello-Cros R, Cukierman E. Stromagenesis during tumorigenesis:
characterization of tumor-associated fibroblasts and stroma-derived
3D matrices. Methods Mol Biol 2009;522:275–305.
17. Skobeleva N, Menon S, Weber L, Golemis EA, Khazak V. In vitro and
Cancer Res; 70(21) November 1, 2010American Asso Copyright © 2010
cancerres.aacrjouDownloaded from 28. Finn RS, Dering J, Ginther C, et al. Dasatinib, an orally active small
molecule inhibitor of both the src and abl kinases, selectively inhi-
bits growth of basal-type/“triple-negative” breast cancer cell lines
growing in vitro. Breast Cancer Res Treat 2007;105:319–26.
29. Bui LC, Tomkiewicz C, Chevallier A, et al. Nedd9/Hef1/Cas-L
mediates the effects of environmental pollutants on cell migration
and plasticity. Oncogene 2009;28:3642–51.
30. Keen N, Taylor S. Mitotic drivers—inhibitors of the Aurora B kinase.
Cancer Metastasis Rev 2009;28:185–95.
Cancer Research ciation for Cancer Research
on March 20, 2012rnals.org