Cables 1, a cyclin‑dependent kinase binding protein, is primarily
involved in cell cycle regulation. Loss of nuclear Cables 1 expres‑
sion is observed in human colon, lung and endometrial cancers.
We previously reported that loss of nuclear Cables 1 expression
was also observed with high frequency in a limited sample set
of human ovarian carcinomas, although the mechanisms under‑
lying loss of nuclear Cables 1 expression remained unknown.
Our present objective was to examine Cables 1 expression in
ovarian cancer in greater detail, and determine the predominant
mechanisms of Cables 1 loss. We assessed potential genetic and
epigenetic modifications of the Cables 1 locus through analyses of
mutation, polymorphisms, loss of heterozygosity and DNA meth‑
ylation. We observed a marked loss of nuclear Cables 1 expression
in serous and endometrioid ovarian carcinomas that correlated
with decreased Cables 1 mRNA levels. Although we detected no
Cables 1 mutations, there was evidence of LOH at the Cables 1
locus and epigenetic modification of the Cables 1 promoter region
in a subset of ovarian carcinomas and established cancer cell lines.
From a functional perspective, over‑expression of Cables 1 induced
apoptosis, whereas, knockdown of Cables 1 negated this effect.
Together these findings suggest that multiple mechanisms underlie
the loss of Cables 1 expression in ovarian cancer cells, supporting
the hypothesis that Cables 1 is a tumor suppressor in human
Cables 1 is a novel cyclin‑dependent kinase (CDK) binding
protein that maps to human chromosome 18q11‑12.1,2 This gene
was independently cloned and the encoded protein identified as
interacting kinase 3 (ik3‑1).3 Cables 1 interacts with CDKs 2, 3 and
5, p53 and p73, c‑abl, Trap and Pctaire2,1‑5 and was recently shown
to link Robo‑bound Abl kinase to N‑cadherin bound b‑catenin.6
Studies to date provide evidence that Cables 1 markedly augments
Wee‑1‑mediated tyrosine 15 phosphorylation of CDK2,1 which
inhibits CDK2 and slows cell proliferation. Furthermore, over‑expres‑
sion of mouse Cables 1 in human cervical (HeLa) and endometrial
(Ishikawa) cancer cell lines markedly slows cell proliferation.1,7,8
In addition to slowing cell proliferation, Cables 1 over‑expression
augments wild‑type p53 induced cell death4 suggesting Cables 1
may contribute to apoptosis. This idea is again supported, albeit
indirectly, by the finding that mouse embryonic fibroblasts (MEFs)
derived from Cables 1‑/‑ mice are more resistant to cell death induced
by serum withdrawal when compared to MEFs derived from Cables
1+/+ mice.9 Furthermore, Cables 1‑/‑ MEFs proliferate at a faster rate
when compared to their Cables 1+/+ counterparts and exhibit delayed
replicative senescence.9 Taken together, these results suggest that
Cables 1 serves as a negative regulator of cell proliferation, and that
loss of Cables 1 function can lead to uncontrolled growth in vivo.
In support of this concept, Cables 1‑/‑ mice have evidence of
endometrial hyperplasia and develop endometrial cancer in response
to unopposed estrogen.7 Likewise, in response to 1,2‑dimethylhy‑
drazine (DMH), Cables 1‑/‑ mice have an increased incidence of
colorectal cancer and a reduced survival rate when compared to
Cables 1+/+ mice.10 Consistent with this hypothesis, loss of nuclear
Cables 1 expression has also been observed in many human cancers,
such as colon, lung and gynecological malignancies including ovarian
and endometrial cancers.1,7,8,11 Collectively, these observations
suggest that Cables 1 may function as a tumor suppressor.
We previously determined that nuclear Cables 1 expression is lost
at high frequency in ovarian serous carcinoma,11 the most common
histological subtype of ovarian cancer. Despite the small number
of samples evaluated in this early study, the high frequency of loss
provided the impetus for analysis of Cables 1 expression in a larger
cohort of samples. In this study, we have examined the extent of loss
of Cables 1 expression in a much larger cohort of ovarian cancer
samples, including clear cell, serous, endometrioid and transitional
cell carcinomas. We sought to define potential mechanisms of Cables 1
gene inactivation, focusing our analyses on the incidence of Cables 1
mutations, loss of heterozygosity (LOH) at the Cables 1 locus and
DNA methylation, in serous ovarian cancer and established cell lines.
Mechanisms of Cables 1 gene inactivation in human ovarian cancer
Hideo Sakamoto1,2, Anne M. Friel1,2, Antony W. Wood1,2, Lankai Guo1, Ana Ilic1, Michael V. Seiden2,5, Daniel C. Chung3,
Maureen P. Lynch1,2, Takehiro Serikawa1,2, Elizabeth Munro1,2, Esther Oliva1,4, Sandra Orsulic4, Sandra D. Kirley4,
Rosemary Foster5, Lawrence R. Zukerberg1,4 and Bo R. Rueda1,2,*
1Vincent Center for Reproductive Biology; 2Vincent Obstetrics and Gynecology Service; 3Department of Gastroenterology; 4Department of Pathology; 5Cancer Center; Division of
Cancer Medicine Oncology; Massachusetts General Hospital; Boston, Massachussetts USA
Key words: ovarian cancer, tumor suppressor, cell cycle, gene inactivation
*Correspondence to: Bo R. Rueda, Ph.D.; Vincent Center for Reproductive Biology;
Massachusetts General Hospital; THR 901A; 55 Fruit St; Boston, Massachusetts
02114 USA; Email: firstname.lastname@example.org
Submitted: 04/30/07; Revised: 10/22/07; Accepted: 11/15/07
Previously published online as a Cancer Biology & Therapy E-publication:
[Cancer Biology & Therapy 7:2, 180‑188; 1 February 2008]; ©2008 Landes Bioscience
180Cancer Biology & Therapy 2008; Vol. 7 Issue 2
www.landesbioscience.comCancer Biology & Therapy 181
To explore the functional role of human Cables 1, we then examined
the consequences of both over‑expression and knockdown of Cables
1 in established ovarian cell lines.
Immunohistochemical microarray study. In order to confirm
and expand our early findings,11 we analyzed Cables 1 expression
in a larger sample of ovarian carcinomas by immunohistochemical
analysis of an ovarian tissue microarray (Fig. 1). We observed a
significant loss of nuclear Cables 1 expression (Fig. 1B) in serous
and endometrioid ovarian carcinomas, which are the more common
ovarian cancer histological subtypes. Loss of Cables 1 nuclear expres‑
sion was much less frequent in clear cell carcinomas. By contrast, all
of the analyzed mucinous (n = 6) and transitional cell (n = 4) carci‑
nomas showed loss of Cables 1 nuclear expression.
Relative Cables 1 mRNA expression in benign and malignant
human ovarian cells and cancer tissue. Figure 2A illustrates the
quantitative PCR (qPCR) analysis of Cables 1 mRNA expression in a
benign HOSE cell line and in the malignant ovarian cancer cell lines
SKOV3, OVCAR5 and OVCAR8, as well as its relative expression
in primary ovarian serous carcinomas or short term cultured primary
ovarian cancer cells (Fig. 2B). Cables 1 expression was dramatically
greater (4‑ to 11‑fold) in HOSE cells relative to its expression in the
ovarian cancer cell lines. Cables 1 mRNA levels among the analyzed
primary serous carcinomas varied slightly; however their overall levels
were an order of magnitude lower than those observed in benign
Cloning of the 5'‑ region of the human Cables 1 gene and iden‑
tification of the transcription start site. We previously reported a
partial sequence of the human Cables 1 mRNA (AF348525) which
lacked complete sequence information in the 5' region. Using RNA
ligase‑mediated 5' rapid amplification of cDNA ends (RACE), we
have now determined the complete sequence of the 5' region of
human Cables 1 and identified the TSS. The full‑length human Cables 1
sequence contains an open reading frame of 1902 bp, encoding a
633 amino acid protein (Fig. 3A) and shares 77% (487/633) iden‑
tity with mouse Cables 1, and 59% identity with zebrafish (Danio
rerio) Cables 1 (EF105292) (Fig. 3A), indicating that Cables 1
has been conserved in vertebrate evolution. All known vertebrate
Cables 1 peptides contain one or more SH3‑binding domains
(PXXP), four putative nuclear localization sequences (NLS), and a
highly‑conserved CDK‑binding domain in the C‑terminal region;
the latter domain is also present in the putative Cables 1 orthologs of
Drosophila (NP610890) and C. elegans (NP491031), confirming its
importance for Cables 1 function (Fig. 3B). Indeed, we have previ‑
ously reported that deletion of the CDK‑binding domain impairs
Cables 1‑mediated negative regulation of CDK2 activity and leads
to increased cell proliferation.15
It is generally observed that nuclear proteins harbor at least
one NLS and proteins containing a NLS are imported into the
nucleus.16‑18 Our earlier studies determined that Cables 1 is
primarily localized to the nucleus in healthy cells.1 Analysis of the
Cables 1 amino acid sequence by PSORT (Prediction of Protein
Sorting Signals and Localization Sites in Amino Acid Sequences,
http://psort.hgc.jp/) identified four putative NLS in mouse, human
and zebrafish Cables 1 (Fig. 3A). These signals are similarly located
in all three primary sequences.
Mutation assay. The results of our microarray and qPCR
analyses suggest that Cables 1 expression is significantly decreased
in primary human ovarian tumors and cell lines. To assess whether
Cables 1 is a frequent target of mutation in ovarian serous carci‑
noma, we extracted genomic DNA from ovarian serous carcinoma
samples with patient‑matched DNA obtained from whole blood.
Genomic DNA corresponding to the entire Cables 1 coding region
was sequenced following PCR amplification of all exon regions. A
number of sequence alterations were identified in a subset of the
blood and cancer samples. To determine whether these changes were
specific to ovarian cancer, we sequenced the relevant region of the
Cables 1 gene in genomic DNA isolated from healthy volunteers.
None of the identified sequence alterations were restricted to the
ovarian cancer samples and were thus classified as polymorphisms
(Supplemental Fig. 1A). These included two single nucleotide
polymorphisms (SNPs), one of which has been previously reported
(rs2304301). Additionally, we identified a single base‑pair insertion
and a 9 bp deletion in samples from healthy volunteers. Interestingly,
40% (16/40) of healthy samples were heterozygous for the 9 bp
deletion, and 7.5% (3/40) were homozygous for this deletion. By
contrast, 26.6% of the ovarian cancer samples were heterozygous for
the 9 bp deletion (26.6%; 4/15), but none were homozygous for the
deletion (Supplemental Fig. 1B).
Loss of heterozygosity analysis. Potential LOH at the Cables 1
locus on chromosome 18q was examined using the highly polymor‑
Mechanisms of Cables 1 gene inactivation in ovarian cancer
Figure 1. Loss of Cables 1 expression in human ovarian cancer in specific
subtypes. An ovarian cancer tissue array was stained with an anti human
Cables 1 polyclonal antibody and nuclear expression of Cables 1 was
assessed as previously described.7,8,11 (A) Representative example of
ovarian serous cancer. Solid arrow points out positive nuclear Cables
1 stain in ovarian inclusion cysts. Whereas, open arrow highlights focal
cancerous area where nuclear Cables 1 positive staining is no longer evident.
(B) The subtype of ovarian carcinoma, number of samples evaluated and the
percentage of samples demonstrating some loss of nuclear Cables 1 expres-
sion in the ovarian cancer tissue array. It is important to note that appropriate
controls were strategically placed within the array to confirm effectiveness
of positive staining.
www.landesbioscience.comCancer Biology & Therapy182
phic markers D18S44 and D18S1107 which map to 18q11. Allelic
loss was observed in 9.1 % and 22.7% of the tumor samples utilizing
the D18S44 and D18S1107 polymorphic markers respectively
(Supplemental Fig. 2).
Methylation assay. To explore the potential role of CpG site
methylation in transcriptional silencing of the Cables 1 gene, we
analyzed the methylation status of the Cables 1 promoter region in
human ovarian cancer tissue, ovarian cancer cell lines and HOSE
cells. We attempted to analyze the methylation status of five different
CpG islands, however, the abundance of GC‑rich regions in the
promoter region of Cables 1 made it difficult to generate primers
that would provide optimum sequence for the specific regions.
Consequently, we focused our efforts on a region where accurate
and complete sequence (EF028204) could be obtained consistently.
There was no evidence of methylation of the specific CpG sites within
the Cables 1 promoter region in genomic DNA isolated from HOSE
cells (Fig. 4A). In contrast, six of eight ovarian serous carcinomas
samples had evidence of methylation in a region corresponding to
627 and 397 bp upstream of the TSS. To extend these studies, we
also analyzed the methylation status of the Cables 1 promoter in the
ovarian cancer cell lines OVCAR5 and OVCAR8. OVCAR8 cells
had evidence of methylation similar to that which was observed in
the primary ovarian cancer tissue. We directly tested the effect of this
methylation on Cables 1 expression by incubating OVCAR8 cells
in the presence of the de‑methylating agent 5‑Aza‑2'‑deoxycytidine
(5‑aza‑dC). Treatment of OVCAR8 cells with 5 mM 5‑aza‑dC for
three days resulted in a loss of methylation and significantly increased
Cables 1 mRNA expression relative to untreated cells (Fig. 4B).
Evaluating samples for evidence of LOH and methylation.
Tumors having evidence of LOH often show some evidence of
mutation, methylation or histone modification of the other allele.
Figure 2. Relative expression of human Cables 1 mRNA in benign and malig-
nant cell lines and ovarian tumor tissue. (A) Cables 1 cDNA copy number
in benign ovarian surface epithelium cells (HOSE) and ovarian cancer cell
lines as determined by qPCR. (B) cDNA copy number from serous ovarian
cancer tissue as determined by qPCR. (Note different scale of Y axis in the
right panel). Each experiment was replicated three times and data represent
the mean and SE.
Figure 3. (A) Comparative analysis of Cables 1 amino acid sequence in
mouse, zebrafish and human. Using PSORT, we identified the putative SH3
binding domain and NLS. The SH3 domain binding motifs (PXXP) are shown
in bold type and the NLS are demarcated in gray highlighted sections. The
homology of the CDK binding region (underlined) was determined by using
the NCBI conserved domain database (rpsblast, www.ncbi.nlm.nih.gov/
Structure/cdd/wrpsb.cgi). (B) Comparative analysis of Cables 1 CDK binding
region across multiple species. Aligned sequence illustrates the cross species
homology of human Cables 1 CDK binding region amongst mouse,
zebrafish and the predicted sequence for other species using a ClustalW
However, after analyzing 8 primary ovarian cancer samples for LOH,
mutation and/or methylation (Supplemental Fig. 3), only one had
evidence of both LOH and methylation.
Overexpression of Cables 1 in ovarian cancer cells. To further
study the role of human Cables 1 in ovarian cell proliferation, we
examined the effects of Cables 1 overexpression on the proliferation
of a benign HOSE cell line, and SKOV3, OVCAR5 and OVCAR8
ovarian cancer cell lines. Overexpression of Cables 1 in SKOV3 and
OVCAR8 cells resulted in a decreased (p = 0.036 and p < 0.001
respectively) proliferation rate by day 6 relative to their controls (Fig.
5A, right panels). A significant decrease in OVCAR5 cell prolif‑
eration rate was also observed (Fig. 5A, lower left panel), but not
Mechanisms of Cables 1 gene inactivation in ovarian cancer
183Cancer Biology & Therapy2008; Vol. 7 Issue 2
Cables 1, a regulator of cell prolifera‑
tion and a candidate tumor suppressor, is
lost in a variety of cancers including
head and neck squamous cell carcinoma,
endometrial adenocarcinoma and ovarian
carcinoma.1,7,11 Our analysis of a larger
cohort of samples support our previous
findings,11 confirming a high incidence
of loss of Cables 1 nuclear expression in
serous, endometrioid and clear cell ovarian
carcinomas. Overall, the percentage of
loss (25 of 26 cases, 96.2%) in the more
common ovarian serous carcinomas was
more extensive than previously described
(11 of 14 cases, 78.6%,11). In addition,
we report herein a significant loss of
nuclear Cables 1 in mucinous and transi‑
tional ovarian carcinoma.
In human tissues, decreased nuclear
Cables 1 protein expression in ovarian
carcinoma is associated with a marked
decrease in mRNA encoding Cables 1.
For example, benign HOSE cells had
between 4‑and 11‑fold more copies of
Cables 1 mRNA relative to malignant
ovarian cancer cells. The relative levels
of Cables 1 mRNA were also very low in
primary ovarian cancer samples, and were
lower than those observed in the cancer
cell lines. Again these levels corresponded
to relative differences observed in the
Cables 1 protein. We hypothesized that
the low levels of mRNA encoding Cables
1 in ovarian carcinoma may be due, in
part, to alterations in the upstream regulatory regions of the gene,
either by mutation or epigenetic alteration. However, efforts to
determine mechanisms by which Cables 1 gene expression might be
regulated have thus far been hindered by the fact that the complete
coding sequence and 5' regulatory regions of the gene were previ‑
A partial mRNA sequence encoding human Cables 1 was reported
over five years ago.1 Following the initial report, the Mammalian
Gene Collection Program Team described what was believed to be
full length mouse (BC043661) and human (BC037218) Cables 1
cDNA sequences.19 This sequence differs from ours in that it has a
truncated exon 1, while exons 2 through 10 are in complete agree‑
ment with our previous report.1,2 The original reported sequence
encodes a 368 amino acid protein, markedly shorter than the 633
amino acid predicted by our primary sequence, and confirmed
by Western blot analysis.1,2 A separate, more recent, report20
(AK093243) contributed additional Cables 1 sequence information
which included more 5' sequence than we previously reported.1 but
their predicted peptide was only 136 amino acids and their exon 1
still lacked 5' regulatory region and a TSS. Given the discrepancies
and the apparent lack of complete sequence, we conducted 5' RACE
Mechanisms of Cables 1 gene inactivation in ovarian cancer
until day 9 (p = 0.0002 data not shown). In contrast, although the
proliferation rate of HOSE cells was decreased in response to over‑
expression of Cables 1, the effect was much less dramatic compared
to the malignant cell lines (Fig. 5A, upper left panel). The relative
overexpression of Cables 1 in each of the analyzed cell lines was
evaluated by Western blotting (Fig. 5B).
Infection with Lenti‑Cables resulted in increased levels of apop‑
tosis in HOSE, SKOV3, OVCAR5 and OVCAR8 cells, relative to
lenti‑GFP‑infected cells. Figure 6A shows apoptosis in OVCAR8
cell line, which was by far the most responsive to Lenti‑Cables infec‑
To determine further the role of Cables 1 in the induction of
apoptosis, each cell line was transfected with one of two different
siRNAs specific for Cables 1. The efficacy of siRNAs in targeting
Cables1 was confirmed in HOSE cells as evidenced by suppression
of endogenous Cables 1 mRNA and protein expression (Fig. 6B).
Suppression of Cables 1 in OVCAR8 cells by siRNA significantly
rescued the cells from Lenti‑Cables‑induced apoptosis (Fig. 6C).
Figure 4. (A) Cables 1 gene promoter methylation status in ovarian cancer tissue and benign HOSE cells.
CpG sites are represented by vertical lines spanning from ‑1 kb to the end of exon 1 (+1238). Ten clones
representing a 385 bp PCR product (nucleotides ‑762 to ‑378) area were evaluated from each specimen.
The circles running horizontal are from a single clone. Black circles represent a methylated CpG site and
white circles represent un‑methylated CpG sites. (B) Change in Cables 1 mRNA expression following treat-
ment with a de‑methylating agent. OVCAR8 cells were treated with 5 mM 5‑aza‑dC for three days. The
graph represents Cables 1 cDNA copy number before and after treatment. The analysis was repeated a
minimum of three times in three different experiments. (C) Change in Cables 1 gene promoter methylation
status in treatment with a de‑methylating agent. All experiments were repeated a minimum of three times
in three different experiments.
www.landesbioscience.comCancer Biology & Therapy 184
to isolate the complete human Cables 1 sequence. The sequence
reported herein encodes a 633 amino acid protein and encompasses
the TSS. The 5' region of exon 1 that was missing from our previous
human Cables 1 sequence is 78% GC rich, which likely accounts
for the difficulty in obtaining accurate sequence data for this region.
The human coding sequence shares regions of high homology
with the mouse Cables 1 mRNA, which encodes a 583 amino acid
protein. We also cloned the homolog of Cables 1 in the zebrafish
(Danio rerio), representing a vertebrate group that diverged from the
mammalian lineage over 450 million years ago. These efforts revealed
high conservation of the Cables 1 amino acid sequences (59%
between zebrafish and humans), with even greater sequence identities
in domains corresponding to the CDK‑binding region and putative
nuclear localization signals, suggesting a fundamentally important
role for this gene in basic cellular function.
Also of interest, the sequence reported by Ota and colleagues
(AK093243) is similar to both our full length Cables 1 sequence
and other previously reported sequences BC037218.19,20 Regions
of similarity among our reported sequence,
BC037218 and AK093243 include exon 2
to exon 4 and exon 5 to exon 7 (and end of
exon 8 in AK093243). These two regions
have 100% identity with our cloned Cables
1 and exons 2 through 4 and 8 through
10 of BC037218. On the other hand, as
reported by Strausberg et al., 2002,19 exon
1 of BC037218 is different from our cloned
Cables 1; it is located downstream of our
cloned exon 1. Exons 2 through 10 are iden‑
tical in all sequences. It is not clear whether
the exon 1 variants identified in the different
reports represent alternatively spliced Cables
1 isoforms, as had been identified in endo‑
metrial and colon cancer.15
Having obtained the full length Cables 1
we investigated several potential mechanisms
of Cables 1 dysregulation and/or inactiva‑
tion. Although no significant mutations were
found in any of the ten exons encoding Cables
1 in the analyzed ovarian tumors, we did find
2 SNPs, one of which is novel. Additionally,
we identified a heterogeneous single base
insertion and a 9 bp in frame deletion in the
cancer tissue samples. Subsequent analysis
of genomic DNA isolated from healthy
volunteers suggests that the 9 bp deletion
is a polymorphism of no functional conse‑
Our previous work argued that LOH in
the proximal region of chromosome 18q
could contribute to loss of Cables 1 expres‑
sion1 in human colon cancers. In contrast,
there was no evidence of LOH in the proximal
region of 18q in endometrial cancer samples
suggesting that LOH is not the primary
mechanism of Cables 1 loss in endometrial
cancer.7 Although a significant number of
ovarian cancers have evidence of allelic loss in the proximal and/or
distal portions of chromosome 18, most research to date has focused
on the distal portion of 18q which encompasses the DCC, SMAD4,
and SMAD2 genes.21,22 In this study we determined that approxi‑
mately 27% of the analyzed ovarian cancer samples had evidence
of LOH in the proximal region of 18q which encompasses the
Cables 1 locus. Although LOH did occur in a subset of the samples,
it cannot fully account for the high degree of loss of Cables 1 expres‑
sion in ovarian cancer observed in both by immunohistochemistry
analyses presented here and our previous studies.10 Because of this
discordance, we explored DNA methylation as another possible
mechanism of Cables 1 inactivation.
It is well known that aberrant methylation can silence gene expres‑
sion.23,24 Although the putative promoter region of Cables 1 contains a
number of CpG islands, only a small CpG rich cluster was methylated
in 75% of the analyzed primary ovarian cancer samples. There was no
evidence of methylation in benign HOSE cells, whereas OVCAR8
cells had a methylation pattern that mirrored what was observed
Figure 5. (A) Relative proliferation rate of HOSE, SKOV3, OVCAR5, and OVCAR8 cells infected with
either 1 MOI of Lenti‑Cables virus or Lenti‑GFP virus (control). The proliferation rate of cells were deter-
mined by seeding 5 x 103 cells/well and counting the cells in triplicate for up to six days as described.
Each experiment was replicated three times and data represent the mean and SE of the three separate
experiments. (B) Protein levels of Cables 1 and actin were confirmed by Western blot.
Mechanisms of Cables 1 gene inactivation in ovarian cancer
185Cancer Biology & Therapy2008; Vol. 7 Issue 2
in the primary ovarian cancer samples. To ascertain the functional
significance of this methylated region, we treated OVCAR8 cells
with the de‑methylation agent 5‑aza‑dC. De‑methylation resulted in
a marked increase in Cables 1 mRNA expression relative to its expres‑
sion in control or nontreated OVCAR8 cells.
Although most tumors that have LOH often show some evidence
of mutation, methylation or histone modification of the other allele,
this is not a prerequisite for reduction in expression. Where LOH is
present it is possible that haploinsufficiency alone could contribute
to the reduction in Cables 1 expression. In cases where there is
no evidence of LOH, biallelic methylation remains a possibility.
However the heterogeneity of the tumor itself (i.e., tumor cells,
endothelial cells, fibroblast and/or inflammatory cells) precludes us
from discriminating from between monoallelic and biallelic methyla‑
tion. Thus, epigenetic modification remains a viable mechanism by
which Cables 1 gene is downregulated/activated, thereby contrib‑
uting to the progression of ovarian cancer.
Similar to what has been observed in cervical and endometrial
cancer cells, overexpression of Cables 1 in ovarian cancer cells can
markedly reduce cell proliferation. However, unlike the previous
experiments, in this study we used human instead of mouse Cables
1, and it was introduced via lentiviral infection. Furthermore, the
changes in Cables 1 mRNA levels in ovarian cancer cells and benign
HOSE were reflected by corresponding reductions in protein levels,
suggesting that Cables 1 expression is inversely associated with cell
Our in vitro findings are supported in part by the phenotype of the
Cables 1 mutant mouse. Loss of Cables 1 is associated with increased
incidence of both endometrial hyperplasia and endometrial cancer
following prolonged exposure to unopposed estrogen.7 Moreover,
the Cables 1 mutant mice develop colon cancer at an accelerated rate
when exposed to DMH.10 In addition, the median survival rate was
significantly shorter for the Cables 1‑/‑ mice compared to Cables 1+/+
littermates following DMH induced tumor formation.
While Cables 1‑/‑ mice develop endometrial cancer,7 there is no
evidence of ovarian neoplasia in these mice, despite the fact that we
find strong correlation between loss of nuclear Cables 1 expression
and human ovarian epithelial carcinoma. Aside from the fact that no
mouse model has been shown to develop spontaneous ovarian epithe‑
lial neoplasia, most ovarian epithelial malignancies in human are not
normally observed until the peri‑ or post‑menopausal years,25,26
corresponding to the period of loss of ovarian function. Interestingly,
the ovaries of aged Cables 1‑/‑ mice possess significantly more devel‑
oping follicles compared to wild type sibling ovaries,27 suggesting
that a lack of Cables 1 may contribute to extended functional ovarian
lifespan and may indirectly delay ovarian neoplasia.
Herein the present study we examined the effects of Cables 1 over‑
expression on the proliferation of a benign human ovarian epithelial
surface cell line, and several ovarian cancer cell lines. Overexpression
of Cables 1 resulted in a significant decrease in cell proliferation rate.
The decrease in cell proliferation was associated with an increase in
apoptosis, whereas the lenti‑GFP had no effect on cell viability.
To further support Cables 1 role in the induction of apoptosis,
the cells were treated with one of two different siRNAs specific
for Cables 1. The siRNAs significantly suppressed the endogenous
mRNA and protein of Cables 1 and there was no effect observed
with the nonspecific siRNA. Moreover the transfection of the
Cables 1 siRNA 24 hr after infection of the Lenti‑Cables virus rescued
Mechanisms of Cables 1 gene inactivation in ovarian cancer
Figure 6. (A) OVCAR8 cells were infected with Lenti-Cables or Lenti-GFP virus at an MOI of 0.3, 1 or 3 for 72 hr and the percentage of cells undergoing
apoptosis was assessed by Hoechst staining. The experiment was replicated three times and the graph represents the mean and SE of relative percentage of
apoptosis. * represents a significance of p < 0.05 and ** represents a significance of p < 0.01 when compared to the same MOI of Lenti-GFP. Protein levels
of Cables 1 and actin were confirmed by Western blotting (lower panel). (B) HOSE cells were transfected with one of two siRNAs specific for Cables 1 or a
nonspecific control siRNA for 48 hr. The expression of Cables 1 and b actin mRNA and protein in HOSE cells were confirmed by RT-PCR (upper panel) and
Western blotting (lower panel). (C) An siRNA specific for Cables 1 reversed Lenti-Cables-induced apoptosis. OVCAR8 cells were infected with Lenti-Cables
virus at an MOI of 1. Twenty-four hours post-infection, the cells were transfected with the indicated siRNAs and apoptosis was analyzed as above. The
experiment was replicated three times and the graph represents the mean and SE of relative percentage to of apoptosis. * represents a significance of p <
0.05 when compared to the siRNA control. Protein levels of Cables 1 and actin were confirmed by Western blot in lower panel.
www.landesbioscience.comCancer Biology & Therapy186
them from Lenti‑Cables induced apoptosis suggesting that Cables 1
knockdown reduces their overall death susceptibility. Although
Cables 1 is considered a candidate tumor suppressor, the degree of
loss of nuclear Cables 1 observed within the tumor samples does
not correlate with any one or more classic epigenetic modifications
including LOH and/or methylation status. Interestingly, of eight
primary samples assessed for multiple parameters (i.e., LOH, meth‑
ylation and mutation) only one sample had evidence of both LOH
and methylation (Supplemental Fig. 3). Therefore our data suggest
that the loss of nuclear Cables 1 expression may also be regulated,
in part, by alternate mechanisms including post‑translational modi‑
fication resulting in accelerated degradation of Cables 1 outside the
nucleus. Regardless, when expressed, it has the ability to significantly
slow tumor cell proliferation or induce apoptosis. Alternatively, its
loss may contribute to the cells survival mechanisms. Thus, loss
or inactivation of Cables 1, may contribute to the pathobiology of
Materials and Methods
Primary tissue, cell and cell line cultures. All primary human
tissue and cells were collected in accordance with policies of the
Massachusetts General Hospital (MGH) IRB, MGH Pathology
and/or MGH Ovarian Tumor Bank (M.V.S. B.R.R and R.F.). The
primary carcinoma cells, whether derived from dissociated ovarian
cancers or ascites, were all used in passage three. Established cancer
cell lines (SKOV3, OVCAR5, OVCAR8) and human ovarian surface
epithelial (HOSE, between passage 15 and 20) were maintained
in Dulbecco’s modified Eagle’s medium (Mediatech, Herdon, VA)
supplemented with 10% fetal bovine serum, penicillin (100 mg/ml),
and streptomycin (100 mg/ml) in humidified 5 % CO2 atmosphere
Tissue microarray and immunohistochemistry. Two tissue
microarrays (TMA) were constructed from paraffin‑embedded
blocks of 70 different ovarian surface epithelial carcinomas. The
ovarian carcinoma samples included 26 serous, 17 endometrioid,
17 clear, 6 mucinous and 4 transitional cell carcinomas, none of
which was part of the previous study.11 Representative areas were
carefully selected from hematoxylin‑eosin stained slide sections
and marked on matching paraffin blocks. Liver core samples were
used as a control and were flanking the carcinoma core samples.
Two tissue cores (1 mm diameter) were obtained from each sample.
The TMA final blocks were constructed using a TMA workstation
(Beecher Instruments, Sun Prairie, WI) and sectioned. The sections
were stained with an affinity‑purified polyclonal antibody raised
against a GST‑tagged human Cables 1 at a 1:200 dilution, using
a microwave‑enhanced avidin‑biotin staining method as described
elsewhere.1,7 Cables 1 staining was scored as positive only when
strong nuclear staining was present in the tumor cells. As every
sample was represented twice within the array, only those samples in
which both representative samples scored the same (positive or nega‑
tive) were incorporated into the final analysis.
Cloning of full length human Cables 1. The Gene Racer kit
(Invitrogen, Carlsbad, CA)12 was used to clone the full length human
Cables 1 and determine the Cables 1 putative transcription start site
(TSS), using a partial template previously reported.1 Briefly, 5 mg of
total RNA from human secretory phase endometrial tissue was used
to generate cDNA. PCR was performed using the manufacturer’s
GeneRacer 5' forward primer (5'‑CGACTGGAGCACGAGGACA
CTGA‑3') and Cables 1 gene specific primer reverse (5'‑ TAGTTCT
GCGAAGTCGGCCTGGAGAA ‑3'). The cDNA transcripts were
sequenced by the MGH DNA Sequencing Core Facility.
Generation of lentiviral vectors and infection. Lentiviral vectors
were generated by the Harvard Gene Therapy Core facility.13
Briefly, cDNA corresponding to the entire Cables 1 open reading
frame containing a Kozak translational sequence was cloned into
pHAGE‑CMV‑MCS‑IZsGreen‑W. Lentivirus was generated from
Cables 1 cloned plasmid and the empty plasmid as a control, using
standard methods. Lentiviral infection was carried out by incubating
cells with virus supernatant for six hours in the presence of 8 mg/ml
polybrene (Sigma), after which the medium was replaced by regular
medium. The Cables 1 lentivirus‑infected cells simultaneously
express green fluorescent protein (GFP) and Cables 1, whereas those
infected with the control virus express only GFP. Optimal infection
rate was determined by infecting OVCAR8 cells with 0. 0.1, 1 or 10
multiplicity of infection (MOI) and assessed for their level of fluores‑
cence. To confirm Cables 1 protein expression, we evaluated lysates
by Western immunoblot. Cell lysates were collected utilizing the
M‑PER mammalian protein extraction reagent (Pierce, Rockford,
IL); 20 mg of lysate were immunostained with anti‑Cables 17 and
anti‑actin (mAb C4; Labvision, Fremont, CA).
Cables 1 effects on proliferation of benign and ovarian cancer
cells. Benign HOSE cells and SKOV3, OVCAR5 and OVCAR8
ovarian cancer cell lines were infected with 1 MOI of Cables 1
lentivirus (Lenti‑Cables) or the GFP control lentivirus (Lenti‑GFP)
as described; cell proliferation rates were determined by seeding 5 x
103 cells/well and counting cell numbers, in triplicate, for up to six
days as described.7
Assessment of apoptosis. Hoechst nucleic staining was used to
obtain percentage of apoptosis as described previously.14 Briefly, cells
were seeded and cultured in 24 well plates. The wells were infected
with increasing concentrations of Lenti‑Cables or Lenti‑GFP virus.
At the end of each time point the media was removed and the plates
were rinsed with PBS and fixed for 10 min with 4% paraformal‑
dehyde. Nuclei were stained by adding 2 mg/ml Hoechst 33258
(Sigma). The number of cells having evidence of nuclear condensa‑
tion or fragmentation were recorded as a percent of the total counted.
Three wells were assessed per experimental group and at least 400
cells in random fields counted for three independent experiments.
The percentage of apoptosis evident in intact untreated or vehicle
cells was used as the control value. To further support our assessment
of apoptosis, DNA was isolated and subjected to 2% agarose gel
electrophoresis. Fragmented DNA (DNA ladder) was visualized by
ethidium bromide staining.
RNA interference. Small‑interfering RNA (siRNA) corresponding
to bases 987–1005 and 1509–1527 of Cables 1 or control siRNA
oligonucleotides were transfected by Lipofectamine RNAiMAX
(Invitrogen) according to the manufacturer’s instructions for final
concentration of 10 nM. Nucleotide sequences corresponding to
siRNAs were as follows: siRNA1, 5'‑GCAACACGAUACCAGGAA
U‑3'; siRNA2, 5'‑GGAGAAGUUUCCUCACAUU‑3'; and siRNA
control, 5'‑AUUGUCCGGUUAUUGCUGC‑3'. Expression of
Cables 1 was confirmed by reverse transcription (RT) PCR and
Western blot for protein.
Mechanisms of Cables 1 gene inactivation in ovarian cancer
www.landesbioscience.comCancer Biology & Therapy187
Quantitative PCR (qPCR). Total RNA was isolated and reverse
transcribed into cDNA as previously described.7 Standard curves
containing a known number of cDNA copies were generated in
10‑fold increments (1 x 104–1 x 108 cDNA copies) for both Cables 1
and b actin. qPCR was performed on the Cepheid Smart Cycler
II using the Lightcycler® TaqMan® Master kit (Roche, Basel,
Switzerland) using commercially synthesized probes (TaqMan,
Biosearch Technologies, Novato, CA): Cables 1, 5' 6‑FAMd [CTG
ATGGGAAGACTGTTTCCTATACCCAA] BHQ‑1 3'; b actin,
5'6‑FAMd [ATCCACGAAACTACCTTCAACTCCATCA] BHQ‑1
3'. Specific primers utilized included Cables 1 sense, 5'‑GGACGGA
GGAAGACAATCAA‑3'; Cables 1 antisense, 5'‑CAGGTTACGGA
ACTGGGAGA‑3'; b actin sense, 5'‑CTTCCAGCCTTCCTTCC
TG‑3';and b actin antisense, 5'‑TTGGCGTACAGGTCTTTGC‑
3'. qPCR cycling conditions were as follows: Cables 1, 95˚C for 10
min, 40 cycles of 95˚C for 10 s, 60˚C for 45 s, 72˚C for 35 s and
1 cycle of 40˚C for 10 s; b actin, 95˚C for 10 min, 40 cycles of
95˚C for 10 s, 65˚C for 45 s, 72˚C for 35 s and 1 cycle of 40˚C for
10 s. Relative expression levels were calculated automatically using
the cycle threshold value from each individual sample, and values
were compared to standards containing known copy numbers. Target
gene copy numbers were expressed as relative fold change normalized
to the internal reference standard gene (b actin).
Mutation analysis. The Cables 1 coding exons and their flanking
intronic regions were amplified by PCR with gene‑specific primers
(Table 1), using genomic DNA extracted from ovarian serous cancer
tissue. In addition, we sequenced genomic DNA isolated from blood
of healthy volunteers. Amplified products were purified by QIAquick
PCR purification Kit (Qiagen, Valencia, CA) and verified by auto‑
Loss of heterozygosity assay. LOH at the Cables 1 gene locus
(chromosome 18q) was examined as previously described.1,7 Archival
DNA was extracted from formalin‑fixed ovarian tissue using stan‑
dard methods, and LOH was assayed using the highly polymorphic
markers D18S44 and D18S1107, which map to 18q11.
Methylation assay. The methylation status of the human Cables 1
promoter was analyzed in ovarian cancer tissue samples (n = 8),
benign HOSE cells (n = 3) and established ovarian cancer cell lines
(OVCAR5 and OVCAR8) by bisulfite sequencing analyses. Briefly,
500 ng of genomic DNA was bisulfite treated using the EZ DNA
methylation kit (Zymo Research, Orange, CA) in accordance with
the manufacturer’s instructions. The PCR primers used were as
follows: sense, 5'‑ AGTTTATGAGGGAGAAGAGAT ‑3'; antisense,
5'‑TAAATAAATCCTTTTCATAATAACA ‑3'. PCR was carried out
using the following conditions: 95˚C for 10 min, 45 cycles of 95˚C
for 30 s, 50˚C for 30 s, 68˚C for 30 s and 1 cycle of 68˚C for 10 min.
PCR products were cloned into pCR2.1 (Invitrogen); ten clones per
reaction were submitted to the MGH Core Facility for sequencing.
This work was funded in part by the Advanced Medical Research
Foundation (BRR), NIH RO1 CA098333 (BRR, LRZ) andthe
Ovarian Cancer Research Foundation (BRR, SO). Dr. Michael
Seiden is supported in part by the NCI mid career award K24‑
CA109416. The MGH Ovarian Cancer Tumor Bank is supported
by a grant from the OCEAN Foundation. The authors would like to
thank Dr. Ronny Drapkin for providing the HOSE cells used in the
analyses of Cables 1 mRNA. Dr. Samuel Mok generously provided
genomic DNA isolated from HOSE cells for the methylation assay.
We thank Dr. Jonathan L. Tilly for critical reading of the manuscript
before its submission. We would also like to thank Dr. Do Youn
Park for his assistance in the design of a bisulfate sequencing primer.
Finally, we would like to thank Dr. Susan Slaugenhaupt for expert
assistance related to mutation analysis.
1. Wu CL, Kirley SD, Xiao H, Chuang Y, Chung DC, Zukerberg LR. Cables enhances cdk2
tyrosine 15 phosphorylation by Wee1, inhibits cell growth, and is lost in many human colon
and squamous cancers. Cancer Res 2001; 61:7325‑32.
2. Zukerberg LR, Patrick GN, Nikolic M, Humbert S, Wu CL, Lanier LM, et al. Cables links
Cdk5 and c‑Abl and facilitates Cdk5 tyrosine phosphorylation, kinase upregulation, and
neurite outgrowth. Neuron 2000; 26:633‑46.
3. Matsuoka M, Matsuura Y, Semba K, Nishimoto I. Molecular cloning of a cyclin‑like protein
associated with cyclin‑dependent kinase 3 (cdk 3) in vivo. Biochem Biophys Res Commun
4. Tsuji K, Mizumoto K, Yamochi T, Nishimoto I, Matsuoka M. Differential effect of ik3‑1/
Cables on p53‑ and p73‑induced cell death. J Biol Chem 2002; 277:2951‑7.
Mechanisms of Cables 1 gene inactivation in ovarian cancer
Table 1 Primers used for PCR amplification and sequencing of the human Cables 1 gene
Exon 1 and 10 were divided into three portions to cover the entire exon and flanking intronic region.
188Cancer Biology & Therapy2008; Vol. 7 Issue 2
5. Yamochi T, Nishimoto I, Okuda T, Matsuoka M. ik3‑1/Cables is associated with Trap and
Pctaire2. Biochem Biophys Res Commun 2001; 286:1045‑50.
6. Rhee J, Buchan T, Zukerberg L, Lilien J, Balsamo J. Cables links Robo‑bound Abl kinase
to N‑cadherin‑bound beta‑catenin to mediate Slit‑induced modulation of adhesion and
transcription. Nat Cell Biol 2007; 9:883‑92.
7. Zukerberg LR, DeBernardo RL, Kirley SD, D’Apuzzo M, Lynch MP, Littell RD, et al. Loss
of Cables, a cyclin‑dependent kinase regulatory protein, is associated with the development
of endometrial hyperplasia and endometrial cancer. Cancer Res 2004; 64:202‑8.
8. Debernardo RL, Littell RD, Luo H, Duska LR, Oliva E, Kirley SD, et al. Defining the
extent of Cables loss in endometrial cancer subtypes and its effectiveness as an inhibitor of
cell proliferation in malignant endometrial cells in vitro and in vivo. Cancer Biol Ther 2005;
9. Kirley SD, Rueda BR, Chung DC, Zukerberg LR. Increased growth rate, delayed senescense
and decreased serum dependence characterize Cables‑deficient cells. Cancer Biol Ther 2005;
10. Kirley SD, D’Apuzzo M, Lauwers GY, Graeme‑Cook F, Chung DC, Zukerberg LR. The
Cables gene on chromosome 18Q regulates colon cancer progression in vivo. Cancer Biol
Ther 2005; 4:861‑3.
11. Dong Q, Kirley S, Rueda B, Zhao C, Zukerberg L, Oliva E. Loss of Cables, a novel gene
on chromosome 18q, in ovarian cancer. Mod Pathol 2003; 16:863‑8.
12. Maruyama K, Sugano S. Oligo‑capping: A simple method to replace the cap structure of
eukaryotic mRNAs with oligoribonucleotides. Gene 1994; 138:171‑4.
13. Mostoslavsky G, Kotton DN, Fabian AJ, Gray JT, Lee JS, Mulligan RC. Efficiency of trans‑
duction of highly purified murine hematopoietic stem cells by lentiviral and oncoretroviral
vectors under conditions of minimal in vitro manipulation. Mol Ther 2005; 11:932‑40.
14. Pru JK, Hendry IR, Davis JS, Rueda BR. Soluble Fas ligand activates the sphingomyelin
pathway and induces apoptosis in luteal steroidogenic cells independently of stress‑activated
p38MAPK. Endocrinology 2002; 143:4350‑7.
15. Zhang H, Duan HO, Kirley SD, Zukerberg LR, Wu CL. Aberrant splicing of Cables gene,
a CDK regulator, in human cancers. Cancer Biol Ther 2005; 4:1211‑5.
16. Garcia‑Bustos J, Heitman J, Hall MN. Nuclear protein localization. Biochim Biophys Acta
17. Hicks GR, Raikhel NV. Protein import into the nucleus: An integrated view. Annu Rev Cell
Dev Biol 1995; 11:155‑88.
18. Moroianu J, Blobel G, Radu A. Previously identified protein of uncertain function is
karyopherin alpha and together with karyopherin beta docks import substrate at nuclear
pore complexes. Proc Natl Acad Sci USA 1995; 92:2008‑11.
19. Strausberg RL, Feingold EA, Grouse LH, Derge JG, Klausner RD, Collins FS, et al.
Generation and initial analysis of more than 15,000 full‑length human and mouse cDNA
sequences. Proc Natl Acad Sci USA 2002; 99:16899‑903.
20. Ota T, Suzuki Y, Nishikawa T, Otsuki T, Sugiyama T, Irie R, et al. Complete sequencing and
characterization of 21,243 full‑length human cDNAs. Nat Genet 2004; 36:40‑5.
21. Arnold N, Hagele L, Walz L, Schempp W, Pfisterer J, Bauknecht T, et al. Overrepresentation
of 3q and 8q material and loss of 18q material are recurrent findings in advanced human
ovarian cancer. Genes Chromosomes Cancer 1996; 16:46‑54.
22. Pere H, Tapper J, Wahlstrom T, Knuutila S, Butzow R. Distinct chromosomal imbalances
in uterine serous and endometrioid carcinomas. Cancer Res 1998; 58:892‑5.
23. Balch C, Huang TH, Brown R, Nephew KP. The epigenetics of ovarian cancer drug resis‑
tance and resensitization. Am J Obstet Gynecol 2004; 191:1552‑72.
24. Freitag M, Selker EU. Controlling DNA methylation: Many roads to one modification.
Curr Opin Genet Dev 2005; 15:191‑9.
25. Partridge EE, Barnes MN. Epithelial ovarian cancer: Prevention, diagnosis, and treatment.
CA Cancer J Clin 1999; 49:297‑320.
26. Hensley ML, Robson ME, Kauff ND, Korytowsky B, Castiel M, Ostroff J, et al. Pre and
postmenopausal high‑risk women undergoing screening for ovarian cancer: Anxiety, risk
perceptions, and quality of life. Gynecol Oncol 2003; 89:440‑6.
27. Lee HJ, Sakamoto H, Luo H, Skaznik‑Wikiel ME, Friel AM, Niikura T, et al. Loss of
CABLES1, a cyclin‑dependent kinase‑interacting protein that inhibits cell cycle progres‑
sion, results in germline expansion at the expense of oocyte quality in adult female mice.
Cell Cycle 2007; 6:2678‑84
Mechanisms of Cables 1 gene inactivation in ovarian cancer