High copy amplification of the Aurora-A gene is associated with chromosomal instability phenotype in human colorectal cancers.

©2007 L
ANDES B
IOSCIENC
E. DO NO
T DISTRIB
UTE.
Research Paper
High Copy Amplification of the Aurora-A Gene is Associated
with Chromosomal Instability Phenotype in Human Colorectal Cancers
[Cancer Biology & Therapy 6:4, e1-e9 , EPUB Ahead of Print: http://www.landesbioscience.com/journals/cbt/abstract.php?id=3817; April 2007]; ©2007 Landes Bioscience
Naoshi Nishida1,2
Takeshi Nagasaka1
Kazuhiro Kashiwagi1
C. Richard Boland1
Ajay Goel1,*
1Department of Internal Medicine; Division of Gastroenterology and Charles A
Sammons Cancer Center, Baylor University Medical Center; Dallas, Texas USA
2Department of Gastroenterology and Hepatology; Graduate School of Medicine;
Kyoto University; Kyoto, Japan
*Correspondence to: Ajay Goel; Cancer Research Laboratory; 250 Hoblitzelle;
Baylor University Medical Center; 3500 Gaston Avenue; Dallas, Texas USA 75246;
Tel.: 1.214.820.2603; Fax :+1.214.818.9292; Email: Ajayg@BaylorHealth.edu
Original manuscript submitted: 10/31/06
Manuscript accepted: 01/05/07
This manuscript has been published online, prior to printing for Cancer Biology &
Therapy, Volume 6, Issue 4. Definitive page numbers have not been assigned. The
current citation is: Cancer Biol Ther 2007; 6(4):
http://www.landesbioscience.com/journals/cc/abstract.php?id=3817
Once the issue is complete and page numbers have been assigned, the citation
will change accordingly.
Key woRds
chromosomal instability, colorectal cancer,
Aurora-A, CDC4, cyclin E
ABBReviATioNs
CIN chromosomal instability
CRC colorectal cancer
MSI microsatellite instability
CIMP CpG island methylator
phenotype
COBRA combined bisulfite restriction
analysis
LOH loss of heterozygosity
ABsTRACT
Chromosomal instability (CIN) is a common but not universal feature of colorectal
cancer (CRC); however, the molecular basis for CIN is controversial and poorly under‑
stood. There are many plausible mechanisms proposed for CIN, including disruption of
G1/S and G2/M checkpoint regulation, and alterations in the spindle checkpoint genes.
However, mutations in individual growth regulatory genes are not commonly observed in
CRC. Therefore, a more comprehensive analysis of the genes involved in each cell cycle
checkpoint regulatory pathway might be required to evaluate a possible role for involve‑
ment in CIN. We investigated the presence of high copy amplification of the cyclin E,
Aurora‑A, Skp2 genes, mutation of ubiquitin ligase CDC4, and promoter methylation of
Mad2L1, as well as the expression of the gene products in a panel of 11 human CRC
cell lines as well as 48 human CRC specimens. In the cell lines with CIN, we found
amplification of the Aurora‑A, cyclin E and Skp2 genes, and a mutation in the CDC4
gene, all of which resulted in altered expression of the cognate proteins. In the human
CRC tissues, amplification of Aurora‑A was frequent (29%), while alterations were rarely
observed in cyclin E, Skp2 or CDC4. Aurora‑A amplification was strongly associated
with a high fractional allelic loss score (p = 0.0001), but not with microsatellite instability,
nor with the promoter methylation phenotype in these tumors. Our data confirm involve‑
ment in the CDC4‑cyclin E pathway of the development of the CIN phenotype in human
CRC, and find that amplification of the Aurora‑A is a common target for disruption of
this pathway.
iNTRoduCTioN
Mounting evidence indicates that almost every cancer has some degree of genetic or
epigenetic instability that generates a large number of nuclear alterations.1 Perhaps the
most common of these is chromosomal instability (CIN), in which a tumor cell experi-
ences the progressive accumulation of chromosomal losses, gains, and rearrangements.
CIN results in aneuploidy, is a common feature of most human cancers, and is present
in as many as 85% of colorectal cancers (CRCs).1 Concerted efforts from numerous
laboratories have been focused on solving the molecular basis of CIN, however a unifying
mechanism to account for CIN has not been found.
It has been proposed that genetic alterations in the mitotic or spindle checkpoint
genes, such as hBUB1 and hBUBR1 or ZW10 and Zwitch might be the cause for CIN, but
mutations are only rarely observed in these particular genes in human CRC specimens.2,3
Disruption of G1/S and G2/M control checkpoint genes has also been suggested as a
potential cause of CIN by in vitro studies;4 amplification of cyclin E has been observed
in many human cancers, and this has been proposed as a possible mechanism for CIN.5,6
Loss of the E3 ubiquitin ligase CDC4, which is responsible for the degradation of cyclin E,
has been reported to induce checkpoint failure and CIN.7,8 Induction of another ubiquitin
ligase, Skp2, which is involved in the degradation of the cyclin dependent kinase inhibitors
p27kip1 and p21Cip1, is also capable of inducing the accumulation of cyclin E, and conse-
quently CIN.9-11 Overexpression of Aurora-A in association with CDC4 insufficiency
has also been speculated as a potential mechanism of CIN, and Aurora‑A amplification is
reported in several cancers.12-14 In addition to these genetic events, the epigenetic inactiva-
tion of the Mad2L1, which is a target of cyclin E signaling via the retinoblastoma pathway,
has been reported as another mechanism that can lead to CIN.15
Collectively, these data suggest that dysregulation of cell-cycle checkpoints by acti-
vation or inactivation of G1/S and G2/M regulatory proteins, especially those in the
CDC4-cyclin E cascade, may be sufficient to induce CIN in some cell lines. However,
e1 Cancer Biology & Therapy 2007; Vol. 6 Issue 4

©2007 L
ANDES B
IOSCIENC
E. DO NO
T DISTRIB
UTE.
www.landesbioscience.com Cancer Biology & Therapy e2
CDC4 Regulatory Genes and CIN in Colorectal Cancer
no comprehensive analysis of these alterations
has been undertaken in human CRC tissues as
well as CRC cell lines to better understand the
importance of individual targets in this growth
regulatory pathway, and their association with
the induction of CIN.
Since several CRC cell lines have been
well characterized for their CIN status, these
cells provide us with suitable in vitro models
to analyze the interplay between genetic and
epigenetic alterations, and to investigate the
relationship between these alterations and
the presence of CIN in a pure population of
neoplastic cells. Therefore, we first performed
a comprehensive analysis of genomic altera-
tions present in the cell-cycle checkpoint
regulatory genes, and determined protein
expression in a panel of 11 CRC cell lines.
Specifically, we investigated the mutational
status of CDC4, genomic amplifications of
cyclin E, Aurora‑A and Skp2, and promoter
methylation of Mad2L1, and correlated these
findings with the changes in protein expres-
sion. Second, we corroborated the in vitro
findings by examining for these alterations
in clinical specimens of human CRC tissues,
and performed a detailed analysis of the rela-
tionship between alterations in these genes
and CIN, microsatellite instability (MSI), as
well as the CpG island methylator pheno-
type (CIMP) status. This permitted us to
take a comprehensive look at the relationship
between CIN and genetic or epigenetic altera-
tions in several genes involved in this pathway.
We found that amplification of the Aurora‑A
gene is frequently present, and was strongly
associated with CIN. Equally importantly,
by examining genetic/epigenetic alterations
in several genes involved in the control of a
single cell cycle checkpoint in human tumor
samples, we have found that alterations in different cell-cycle check
point genes may give rise to CIN in CRCs through a single common
pathway, which is a recurring theme in tumor genetics.
MATeRiAls ANd MeThods
Human CRC cell lines. Eleven CRC cell lines (RKO, SW48,
HCT116, HCT116-p53-/-, LoVo, SW480, CaCo-2, LS174T,
SW837, WiDr, LIM6) were used in this study. Among the 11 CRC
cell lines, six (RKO, SW48, LS174T, HCT116, HCT116-p53-/- and
LoVo) have been categorized as having no CIN, but instead have
defects in the DNA mismatch repair system that cause MSI. These
cells are all near-diploid. The remaining five cell lines (SW480,
CaCo-2, SW837, WiDr, LIM6) all have CIN.16-18 For deter-
mination of methylation status of these cell lines, we quantified
methylation density at six CIMP-related loci (MINT1, MINT2,
MINT31, MLH1, p14 and p16) using combined bisulfite restriction
analysis (COBRA) and classified as CIMP-positive when there was
methylation of >5% density at three or more loci.19-22 The primer
pairs, PCR conditions and restriction enzymes used for COBRA
were described previously.22,23 According to methylation analysis of
six CIMP-related loci, four MSI cell lines (RKO, SW48, HCT116,
HCT116-p53-/-) and three CIN cell lines (CaCo-2, WiDr, SW837)
were classified as CIMP-positive, some of which were confirmed by
previous reports.24,25 Mutations of the K‑ras and BRAF genes of these
cell lines were also described previously.17,26 Each cell line was grown
in the appropriate culture medium, with 10% FBS. All cell lines were
harvested at ~70–80% confluency for the analyses. Control RNA
and protein lysates from normal colon tissues were obtained from
CLONTECH Laboratories, Inc. (Mountain View, CA).
Human CRC tissues. Forty-eight CRC tissues were selected
randomly from a larger pool of CRCs obtained from Okayama
University Hospital (Okayama, Japan). The samples were fresh-frozen
and stored at -80˚C after surgery until extraction of genomic DNA.
The demographics and other clinical data from these patients
and their genetic and epigenetic alterations are listed in Table 1.
The CIN and MSI status, methylation profiles, and mutations of
the K‑ras (codons 12 and 13) and BRAF (codon 600) genes were
determined previously.22,27 Loss of heterozygosity (LOH) analyses
were performed at 15 loci where frequent LOH events occur, and
aFAL score is expressed as the percentage of LOH events of all loci assayed. bCRCs were classified as CIMP-positive when methylated with >5%
density at >3 loci at six CIMP-related loci (MINT1, MINT2, MINT31, MLH1, p14 and p16). cThe MSI status of each sample was determined using
12 microsatellite markers. The criteria for MSI-H, MSI-L and MSS are described in the text.
Table 1 Profile of 48 CRC cases examined in this study

©2007 L
ANDES B
IOSCIENC
E. DO NO
T DISTRIB
UTE.
CDC4 Regulatory Genes and CIN in Colorectal Cancer
e3 Cancer Biology & Therapy 2007; Vol. 6 Issue 4
fractional allelic loss (FAL) scores were calculated. LOH was deter-
mined if there was that >50% reduction in the relative intensity of
one allele in tumor compared with normal tissues.27 LOH markers
included MYCL1 (1p32), D1S211 (1p34), D1S228 (1p36), D2S123
(2p16), D5S107 (5q11-13), D5S346 (5q21-22), D7S480 (7q31),
D8S87 (8p12), D8S254, D8S258 (8p22), D17S250 (17q12-21),
D17S261 (17p11-12), D17S960 (17p12-13), D18S35 (18q21) and
D18S58 (18q22-23). In order to classify CRCs tissues according
to CIMP status, we used the same criteria which was applied for
CRC cell lines.19-22 The MSI status of each sample was determined
using 12 microsatellite markers: BAT25, BAT26, BAT40, D2S123,
D5S107, D5S346, D8S87, D17S261, D17S250, D18S35, D18S58
and MYCL1. The criteria for MSI-high (H), MSI-low (L) and MSS
were as follows: MSI-H when ≥40% of loci had allelic shifts, MSI-L
when <40% of loci had allelic shifts, and MSS when there were
no mutations at any of the loci.28 Written informed consent was
obtained from each patient, and Institutional Review Board approval
was obtained from all institutions involved this study.
Mutational analysis of the CDC4 gene. For mutational analysis of
the CDC4 gene in the cell lines, exons 2 to 11 (including intron-exon
junctions) were amplified using primers described previously,29 since
previous studies had not reported any mutations in exon 1 of CDC4
in CRC.8,30 Direct sequencing was performed to confirm muta-
tions in the amplified PCR products. We specifically focused on the
WD40-region (exon 7 to exon 10) of the CDC4 gene for mutational
analysis of tissue samples, because the only CDC4 mutation observed
in the CRC cell lines was located in the WD40-region. Furthermore,
more than 70% of previously reported mutations in the literature
have been confined to this region.30
Detection of gene amplification of the cyclin E, Aurora‑A and
Skp2 genes. We used real-time PCR methodology to determine copy
number alterations in each of the genes. Genomic DNA was isolated
using the QIAamp DNA Mini Kit (QIAGEN, Valencia, CA).
Real-time quantitative PCR was performed using an ABI Prism 7000
sequence detection system (Applied Biosystems, Foster City, CA).
The primer and probe sequences for genomic real-time PCR for each
of the genes are as follows. Cyclin E forward primer; 5'-TGCCTCTG
GGAGGTGTTCTC-3', reverse primer: 5'-CAAAATGCAGCCATG
TTCCA-3', probe: 5'-AGCTGGACACATTGTG-3': The Aurora‑A
forward primer: 5’-TCTTTTATAGAAATGTGTGGAAGTTCC
T-3', reverse primer: 5'-CAATAAAAAAGTACAGACGCATAAAC
CA-3', probe: 5'-TCTGTCCTTAGAAATAACCACTAC-3'. The
Skp2 forward primer: 5'-CGGCTTTCGGATCCCATT-3', reverse
primer: 5'-TAAAATCAACTGCCTTTGTGACACT-3', probe:
5'-TCAAGTGAGTGGCAGGCA-3'. All probes were labeled with
FAM as the reporter dye and T MRA as the quencher. For quantita-
tive purposes, the RNaseP gene, which is located on 14q11.2-q12,
was used as the endogenous control gene (Applied Biosystems Inc.,
Foster City, CA). In addition, as loss of one copy of the endogenous
control gene would cause erroneous estimates of amplification of all
the target genes, a second reference gene, the 18S rRNA gene (located
on 12p12) was used for further confirmation, if all of the target genes
showed amplification compared to RNaseP. Each PCR amplification
reaction was performed in duplicate. The ratios of target/control
genes from CRC cell lines or tissues were normalized using the mean
target/control ratios of five normal genomic DNA samples obtained
from diploid cells that included two normal lymphocytic popula-
tions, two frozen normal human liver tissues, and one fibroblast cell
line (CCD18CO). As a normal diploid cell carries two copies of the
gene, two fold ratios following normalization were considered to be
the absolute copy number of the target gene. To ensure the reproduc-
ibility and optimal cut-off values for the detection of copy number
alterations, gene copy numbers of each target in normal diploid cells
were determined from fourteen individual real-time PCR amplifi-
cation reactions using normal DNA from five different cell types,
which is theoretically 2.0. We found that the copy numbers for target
genes in genomic DNA samples from normal specimens were 2.0 ±
0.18 (mean ± SD) for the Aurora‑A gene, 2.0 ± 0.22 for the cyclin E
gene, and 2.0 ± 0.12 for the Skp2 gene. Therefore, a copy number of
≥4.0 was considered to represent amplification of two-fold or greater
of a given target gene and defined as “amplification”, as has been
suggested previously.10
COBRA analysis for methylation of the Mad2L1 gene
promoter. To analyze the methylation status of the Mad2L1
gene, we designed three different primer pairs in the
promoter region for COBRA. The primer sequences of each
of the pairs were, 5'-AGGAGGGTTTTAAGTTTAGGATA-
GA-3' and 5'-CTAAAAAACCAAAAAACACAATTTC-3',
5'-GGAAATTGTGTTTTTTGGTTTTTTA-3' and
5'-CTACCCTTTCTCTCAACCTTCCTATA-3', 5'-TATAG-
GAAGGTTGAGAGAAAGGGTAG-3' and 5'-TAACCAAAAACA
CAAACAAAAACAC-3', respectively. These primers spanned the
entire promoter region of Mad2L1 as previously described.15 PCR
products were subsequently digested using 0.5 U of the restriction
enzymes Nru1, BstU1 and HpyCH4IV (New England BioLabs,
Ipswich, MA) respectively. PCR amplified digested products were
electrophoresed on 2.5% agarose gels and subjected to subsequent
densitometric analysis for purposes of quantitation using a Kodak
Gel-Logic 200 imaging system (Eastman Kodak Co., Rochester,
NY).
mRNA expression analysis of Mad2L1 by real‑time RT‑PCR.
Quantitative changes in mRNA expression for Mad2L1 were
determined for each of the CRC cell lines, by comparing the expres-
sion levels with that of normal colon tissue, for which the total
RNA was obtained commercially (CLONTECH Laboratories, Inc.,
Mountain View, CA). Total RNA was extracted using TRIZOL®
(Invitrogen, Carlsbad, CA) from each cell line grown in a 100 mm
dish. Total RNA was subjected to a real-time PCR quantitative assay
using the ABI PRISM 7000 Sequence Detection System with gene
specific primer pairs and TaqMan probes for the Mad2L1 messages
obtained commercially from TaqMan® Gene Expression Assays
(Applied Biosystems, Foster City, CA). Expression of the GAPDH
gene was used as an internal control. For relative quantification, total
RNA extracted from WiDr cells was used as an internal calibrator,
and the quantity of RNA was determined as a ratio of the target gene
to that of the calibrator using a standard curve.
Western blot analysis. Expression of cyclin E, Aurora-A and Skp2
proteins was analyzed by western immunoblotting. Briefly, total
protein was extracted from each cell line by direct lysis in RIPA buffer
(1X TBS, 1% nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS
and 0.004% sodium azide) containing 10 ng/ml PMSF, 1% protease
inhibitor and 1% sodium orthovanadate. Subsequently, 50mg of total
protein extract was electrophoresed on 10% SDS-polyacrylamide
gels. Proteins were transferred to polyvinylidene difluoride (PVDF)
membranes, which were thereafter probed with specific primary anti-
bodies as follows: anti-cyclin E mouse monoclonal antibody (1:1000;
Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-Aurora A/AIK
rabbit polyclonal antibody (1:1000; Cell Signaling Technology,
Beverly, MA) and anti-Skp2 mouse monoclonal antibody (1:500;
Santa Cruz Biotechnology, Inc.). This was followed by incubation

©2007 L
ANDES B
IOSCIENC
E. DO NO
T DISTRIB
UTE.
www.landesbioscience.com Cancer Biology & Therapy e4
CDC4 Regulatory Genes and CIN in Colorectal Cancer
of the membranes with a horseradish peroxide-
conjugated secondary antibody and chemilumi-
nescence reagents (Amersham, Piscataway, NJ).
For the detection of phosphorylated-cyclin E at
threonine 380, an anti-thr380 phosphorylated
cyclin E rabbit polyclonal antibody (1:2000;
Santa Cruz Biotechnology Inc.) was used. All
membranes were finally probed with a-actin
(1:5000; Sigma, St. Louis, MO) to control for
variable protein loading. A PhosphoCruzTM
Protein Purification System was used to concen-
trate the phosphoprotein fraction (Santa Cruz
Biotechnology, Inc.). For each western blot used
to determine the steady-state levels of protein
levels in CRC cell lines, control protein lysates
obtained from normal colon tissue (CLONTECH
Laboratories, Inc.: Mountain View, CA) were also
analyzed under similar conditions. The inten-
sity of each band was quantified using a Storm
840 imaging system. Protein expression of each
cell line was normalized against the expression
in the normal colonic tissue, and the values
are shown as a ratio of the density of the CRC
cell line divided by that from the normal colon
(Amersham Bioscience, Piscataway, NJ).
Statistical analysis. For comparisons of catego-
rized data and the presence or absence of genetic
alterations, a Chi-Square test or a Fisher’s exact
test were used. The Wilcoxon rank-sum test was
also applied to compare the FAL scores between
any two categorical variables. To examine the
relationship between the copy number of the
Aurora‑A gene and its expression, Pearson’s rank
and Spearman’s rank correlation test were applied.
All statistical analyses were calculated using JMP
version 4.05J software (SAS Institute Inc. Cary,
NC). All p-values were two-sided and p < 0.05
was considered statistically significant.
ResulTs
Gene amplification of Aurora‑A, Cyclin E and
Skp2 and mutation of CDC4 in CIN cell lines.
Among the 11 human CRC cell lines, three CIN
cell lines (SW480, CaCo-2 and WiDr) demon-
strated amplification of the Aurora‑A gene. The
copy numbers of the Aurora‑A gene in SW480,
CaCo-2 and WiDr were 4.0, 4.2 and 4.0, respec-
tively. WiDr cells also showed more than four
copies of the cyclin E (6.6 copies) and Skp2 genes
(4.0 copies). On the other hand, none of the MSI
cell lines showed amplification of any these three
cell-cycle related genes (Table 2).
All CRC cell lines showed higher protein
expression of cyclin E, Aurora‑A and Skp2 compared
to normal colonic tissues. In addition, CRC
cell lines which showed gene amplification at a
specific target also demonstrated higher protein
expression, suggesting that the gene amplifica-
tions were biologically relevant (Fig. 1A–C).
aCopy number of each gene. Bold shows gene amplification. b”c” indicates the codon with the mutation. “W” denotes a wild type sequence.
cMSI cell lines were considered as MSI-H according to the previous reports 17. dSamples are sorted according to FAL score. N.D, not deter-
mined.Methylation, MSI and mutations of the K‑ras and BRAF genes in CRC cell lines were described previously.16-18,24-26
Table 2 Genetic alterations of cell cycle regulators correlated with CIN,
CIMP, MSI and K‑ras/BRAF mutation status for CRCs

©2007 L
ANDES B
IOSCIENC
E. DO NO
T DISTRIB
UTE.
CDC4 Regulatory Genes and CIN in Colorectal Cancer
e5 Cancer Biology & Therapy 2007; Vol. 6 Issue 4
Although amplification of gene copy number and protein expression
was observed for most targets, the protein expression of Aurora-A
significantly correlated with the increased gene copy number
(Pearson correlation, r = 0.81 and p = 0.0023; Spearman correlation,
r = 0.70 and p = 0.0158) (Fig. 1D).
Sequence analysis of the CDC4 gene revealed that among the 11
CRC cell lines investigated, only SW837 cells (with a CIN pheno-
type) carried an insertion mutation at codon 403 in exon 7, which
results in the loss of the wild-type allele (Fig. 2 and Table 2). This
mutation created a new stop codon in the downstream coding region,
and was located within the WD40 repeat domain, which is respon-
sible for substrate binding and critical for CDC4 function. Western
blot analysis also revealed accumulation of the phosphorylated form
of cyclin E in the SW837 cell line (Fig. 1E).
Next, we analyzed the promoter methylation status of the Mad2L1
gene using three pairs of COBRA primers, since Mad2L1 is a regu-
latory target of G1/S and mitotic checkpoint signaling. A previous
study had suggested that inactivation of Mad2L1 by promoter meth-
ylation may be an important mechanism of transcriptional silencing
of this gene in cancer cells, which may lead to defects in the mitotic
checkpoint.15 However in our study, we did not observe promoter
methylation for Mad2L1 in any of the CRC cell lines. To determine
whether Mad2L1 might be inactivated by other mechanisms, we
examined expression of Mad2L1 in CRC cell lines and compared it
with the normal colon using real-time PCR. Expression analysis of
Mad2L1 using real-time PCR revealed increased Mad2L1 transcripts
in every cell line compared with that from normal colon tissue.
As Mad2L1 is reportedly a direct target of E2F, and is aberrantly
expressed in cells with suppression of the Rb pathway or activation
of E2F,31 our result suggests that Mad2L1 is probably up-regulated
in CRC cell lines as a consequence of activation of E2F through an
enhanced cyclin E system and cell cycle progression. In other words,
the promoter function of Mad2L1 is maintained and methylation-in-
duced inactivation of the Mad2L1 gene was not observed in the CRC
cell lines we analyzed (data not shown), raising doubt that it acts as a
tumor suppressor in the colon.
Frequent amplification of the Aurora‑A gene in human CRC
tissues. To follow up upon our observation of genomic amplifica-
tion in Aurora‑A, cyclin E, Skp2 and mutation in the CDC4 gene
in CRC cell lines, we looked for similar alterations in human CRC
tissues. Distributions of copy numbers (mean ± SE) of the three
genes ranged from 1.0 to 9.4 (3.1 ± 0.2) for Aurora‑A, from 1.2
Figure 1. Representative western blots for Aurora‑A (A), cyclin‑E (B) and Skp2 (C) in CRC cell lines. Each band specific for Aurora‑A, cyclin E and Skp2
(arrows) was quantified by densitometry and normalized to that of a‑actin. The relative expression of each protein is denoted by the fold increase of
expression compared to that in normal colon tissue. (A) Three CRC cell lines (SW480, CaCo‑2, WiDr) which were found to have ≥4 copies of the Aurora‑A
gene showed increased expression of Aurora‑A compared with the others. SW837 cells, which carry a CDC4 mutation also showed accumulation of
Aurora‑A. (B) One CRC cell line (WiDr) had ≥4 copies of the cyclin E gene, and had a correspondingly higher expression of cyclin E. (C) The CRC cell
line WiDr showed ≥4 copies of the Skp2 gene, and demonstrated a correspondingly higher expression of Skp2 protein. (D) A significant positive correla‑
tion is demonstrated between Aurora‑A gene copy number and protein expression in CRC cell lines. [Pearson’s rank correlation (r = 0.81, p = 0.0023),
and Spearman’s rank correlation (r = 0.70, p = 0.0158)]. (E) Representative western blots for thr380‑phosphorylated cyclin E using a specific polyclonal
antibody. Among the colon cancer cell lines, SW837 showed the highest expression, suggesting that accumulation of thr380‑phosphorylated cyclin E, which
is the putative substrate of CDC4, has occurred.