Oncogene (1998) 16, 1443 ± 1453
 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00
Suppression of invasive properties of colon cancer cells by a metastasis
suppressor KAI1 gene
Akinori Takaoka, Yuji Hinoda, Shuji Satoh, Yasushi Adachi, Fumio Itoh, Masaaki Adachi
and Kohzoh Imai
First Department of Internal Medicine, Sapporo Medical University, Sapporo 060, Japan
KAI1 is a potential metastatic suppressor gene for
prostate cancer. We found by Northern blot analysis that
six of ten (60%) gastric and colon cancer cell lines
exhibited undetectable or very low expression level of
KAI1 mRNA. The e€ects of KAI1 on the adhesion,
motility and invasiveness of colon cancer cells was
therefore investigated by using two kinds of stable
transfectants, i.e., antisense transfectants of BM314
cells whose KAI1 mRNA expression was suppressed by
transfer of antisense KAI1 cDNA and sense transfectants of DLD-1 cells with the enhanced KAI1 mRNA by
sense cDNA transfer. The following results were
obtained: (1) KAI1 gene expression had no signi®cant
e€ect on in vitro cell growth rate of colon cancer BM314
and DLD-1 cells; (2) Cell aggregation assay showed that
KAI1 enhanced the Ca++-independent aggregatability of
those colon cancer cells; (3) It was revealed by cell
motility and invasion assays that KAI1 suppressed both
the motility and in vitro invasiveness of those cells and
(4) Furthermore, both the binding to ®bronectin and the
migration on ®bronectin-coated plates of those cells were
inhibited by KAI1 expression. These suggest that
reduced KAI1 gene expression may contribute to the
invasiveness and metastatic ability of colon cancer cells.
Keywords: colon cancer; KAI1 gene; invasion; metastasis
Introduction
Colon cancer has recently been increasing as one of
the most common cause of cancer death of both men
and women in Japan (Ministry of Health and Welfare,
1996). Although endoscopic treatments have been
greatly improved in early stage of colon cancer, there
is no established treatment for the late stage which is
characteristic of invasion into neighboring organs, or
distant metastasis. Tumor metastasis, a leading cause
of death for cancer patients, involves a multistep
process which is determined by both properties of
metastatic cancer cells and their interactions with the
host (Poste and Barsky, 1980; Fidler and Hart, 1982;
Liotta and Fidler, 1983; Liotta and Stetler-Stevenson,
1991; Aznavoorian et al., 1993). The genetic events that
suppress or induce tumor metastatic process remains
least understood aspects of cancer. Evidence on genetic
alterations in oncogenes and tumor-suppressor genes
had encouraged the search for genes that may induce
Correspondence: Y Hinoda
Received 24 March 1997; revised 13 October 1997; accepted 15
October 1997
or suppress tumor invasion and metastasis. A number
of negative e€ector genes have been identi®ed as being
implicated in invasion and metastasis. Among them,
the nm23 gene, with potential tumor-suppressor
function, had been reported to be negatively correlated with metastatic potential in certain spontaneous
and experimental tumors (Barnes et al., 1990; Leone et
al., 1991). Recent con¯icting results, however, have
revealed no signi®cant relationship between Nm23
expression and metastatic ability of colon cancer cells
(Haut et al., 1988; Cohn et al., 1991; Myero€ and
Markowitz, 1993; Wang et al., 1993; Whitelaw and
Northover, 1993). DCC also seemed to be a candidate
tumor suppressor gene, deletions and somatic mutation
of which have been observed in a number of colorectal
cancers (Fearon et al., 1990; Ookawa et al., 1993). Our
previous report showed some relationship between
colon cancer DCC mRNA expression and the
presence of liver metastasis (Itoh et al., 1993). Its
structural homologies with the neural cell adhesion
molecule (N-CAM) and other related cell surface
glycoproteins suggest that DCC might be involved in
cell adhesion, but its function is yet to be completely
elucidated. Recently, E-cadherin, a putative invasion
suppressor gene, has been reported to modulate the
invasiveness of various carcinoma cells (Behrens et al.,
1989; Frixen et al., 1991; Ueda et al., 1991). However,
a clear-cut relationship between invasiveness and Ecadherin expression was not found in all tumor-cell
systems (Behrens et al., 1993).
A recent study by Dong et al. (1995) has identi®ed a
potential marker for metastatic prostate cancer, KAI1,
a metastatic suppressor gene localized to chromosome
11p11.2. KAI1 expression is reduced in human cell lines
derived from metastatic prostate cancers, compared
with its expression in normal prostate tissue. The
introduction of KAI1 into a rat prostate cancer cell
line, AT6.1 signi®cantly suppressed pulmonary metastases in nude mice. KAI1 speci®es 267 amino acids, and
has a characteristic polypeptide chain containing four
hydrophobic, presumably membrane-spanning segments and a single major presumably extracellular
domain with three potential N-glycosylation sites.
Thus, KAI1 belongs to a structurally related family
of membrane glycoproteins named `transmembrane 4'
(TM4) or `tetra spans transmembrane' (TST) family
(Boucheix et al., 1991; Horejsi and Vlcek, 1991; Gil et
al., 1992; Wright and Tomlinson, 1994), most of which
have been identi®ed as leukocyte surface proteins
(Horejsi and Vlcek, 1991). KAI1 cDNA sequence
revealed that KAI1 is identical to three cDNA clones
from human leukocytes, designated by C33 (Fukudome
et al., 1992; Imai et al., 1992), R2 (Gaugitsch et al.,
1991), and IA4 (Gil et al., 1992), which has been
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1444
assigned the CD-names : CD82. KAI1 shows its
expression not only in the prostate but in a wide
variety of tissues including the intestine and colon. It is
thus intriguing to hypothesize that KAI1 may play a
universal role in suppressing metastases of other tissuederived cancers.
Our research question concerning KAI1 is whether its
reduced expression is associated with or causes highly
metastatic behavior in colon cancer cells. In the present
study, we describe the transfection of human KAI1
cDNA into two kinds of colon cancer cell lines both in
the sense direction and in the antisense direction, and
analyse their resultant phenotypic changes with
bidirectional regulations of KAI1 gene expression. We
demonstrate the potential involvement of KAI1 in the
invasion and metastasis of colon cancer cells.
Results
Expression of KAI1 mRNA in gastric and colon cancer
cell lines
To evaluate the expression of KAI1 message in several
gastric and colon cancer cell lines, we performed
Northern blot analyses using the cDNA as probe,
which exhibited various di€erences in the level of
expression among those cell lines examined. In gastric
cancer cell lines (Figure 1a, lanes 1 to 5), the 2.4 kb
KAI1 transcript was detected most abundantly in
MKN74 (lane 4): moderately abundant in JRST (lane
2), while MKN45 (lane 3) showed relatively slight
expression of KAI1 mRNA, and AZ521 (lane 1) and
NUGC3 (lane 5) revealed loss of KAI1 mRNA
expression. In colon cancer cell lines (Figure 1a, lanes
6 to 10), KAI1 message was observed in BM314 (lane
6) and Colo201 (lane 8), but not in CHCY-1 (lane 7),
Colo320DM (lane 9) and DLD-1 (lane 10). BM314 and
DLD-1 cells were used for the following antisense and
sense KAI1 cDNA transfection, respectively.
Preparation of antisense and sense KAI1 transfectants
To analyse the role of KAI1 in motility and metastatic
process of colon cancer cells, stable transfectants in both
the sense and the antisense directions were established.
BM314 cells with endogenous KAI1 mRNA expression
and DLD-1 cells with its least expression were transfected
with antisense and sense KAI1 cDNAs, respectively. Cells
transfected with only an expression vector pRc/CMV
served as a control. Expression of KAI1 gene in resultant
transfectant clones was analysed by reverse transcriptasepolymerase chain reaction (RT ± PCR) (Figure 2),
showing the 801 bp band predicted for primer pair 1
and 2 with di€erent expression levels. In antisense
transfectants (Figure 2a), clone 3 with highly suppressed
endogenous KAI1 mRNA expression named as BM/
as77 and clone 5 with moderate suppression named as
BM/as7 were chosen as negatively regulated clones for
KAI1 gene expression and were analysed for further
functional experiments. In sense KAI1 transfectants
(Figure 2b), clone 8 with high expression level of KAI1
mRNA named as DLD/s++ and clone 9 with moderate
expression named as DLD/s+ were chosen as positively
regulated clones for KAI1 expression. Control clones for
sense and antisense KAI1 transfectants were named as
DLD/con and BM/con, respectively.
Detection of KAI1 protein in transfectants by ¯ow
cytometry
To con®rm the expression of KAI1 gene at the protein
level, ¯ow cytometry using permeabilized transfectants
b
a
BM314
1 2 3 4 5 6 7
a
1
2 3 4 5 6
KAI1
7 8 9 10
β-actin
28S —
18S —
DLD1
8 9 10 11 12 13 14
KAI1
b
28S —
18S —
Figure 1 Detection of KAI1 mRNA in gastric and colon cancer
cell lines by Northern blot. Ten mg per lane of total cellular RNA
extracted from various human gastric and colon cancer cell lines
were electrophoresed and hybridized with a 32P-labeled full-length
human KAI1 cDNA as a probe. The ®lter was exposed to an Xray ®lm. The following cell lines were examined: Gastric cancer
cells AZ521 (lane 1), JRST (lane 2), MKN45 (lane 3), MKN74
(lane 4) and NUGC3 (lane 5), and colon cancer cells BM314 (lane
6), CHCY1 (lane 7), Colo201 (lane 8), Colo320DM (lane 9) and
DLD-1 (lane 10). The autoradiograph of the ®lter (a) and the
photograph of the gel (b) stained with ethidium bromide for RNA
loading control are shown. The size markers of 18S and 28S
rRNAs were indicated
Figure 2 KAI1 mRNA expression of transfected colon cancer
cell clones. Total RNA of KAI1-transfectants derived from two
di€erent tumor cell lines were analysed. RT ± PCR was carried
out as described in Materials and methods. Each PCR product
was separated electrophoretically on a 1% agarose gel. After
transfer to nylon membranes, DNA was probed with a full-length
KAI1 cDNA. (a) Primers KAI1/F and KAI1/R2 were used to
amplify cDNAs from BM314 transfectants and BM314 parental
cells to evaluate the e€ect of transfection of antisense KAI1
cDNA on the endogenous KAI1 mRNA level. BM314 colon
cancer cells transfected with antisense cDNA of KAI1 (lane 1 to
lane 5), vector-alone (lane 6) and untransfected BM314 cells (lane
7). (b) Primers KAI1/F and KAI1/R1 were used to amplify
cDNAs from DLD-1 transfectants and DLD-1 parental cells.
DLD-1 colon cancer cell clones transfected with sense cDNA of
KAI1 (lane 8 to lane 12), vector-alone (lane 13) and untransfected
DLD-1 cells (lane 14). The KAI1 transfectant clones with di€erent
expression levels of KAI1 mRNA (lane 3, a highly suppressed
clone; lane 5, a moderately suppressed clone; lane 8, a highly
expressed clone; lane 9, a moderately expressed clone) were used
for the following experiments. BM314 (lane 6) and DLD-1 (lane
13) transfectant clones with vector (pRc/CMV) alone were used as
a control. Expression of b-actin mRNA is shown in lower panel
to control for equal RNA amounts
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
was performed. DLD/s+ and DLD/s++ were shown
to react with an anti-KAI1 polyclonal antibody where
the relative ¯uorescence intensity of DLD/s++ was
higher than that of DLD/s+. DLD/con was negative
for the antibody (Figure 3). Percentages of positive
cells of DLD/con, DLD/s+ and DLD/s++ were
1.1%, 85.5% and 95.1%, respectively. BM/con also
clearly reacted with the anti-KAI1 antibody whereas
BM/as7 and BM/as77 were weakly and faintly
positive, respectively (data not shown).
E€ect of KAI1 on in vitro growth of colon cancer cells
The in vitro cell growth rates of both antisense and
sense KAI1 transfectants and control clones measured
by MTT assay are graphed in Figure 4. The growth
rates of antisense KAI1 transfectants (BM/as77,
BM/as7) and control cells (BM/con) were virtually
superimposable, as were the curves for sense KAI1
transfectants (DLD/s++, DLD/s+) and control cells
(DLD/con). These revealed that neither expression nor
suppression of KAI1 message a€ected the in vitro
growth rate of those two colon cancer cell lines. Cell
viability was greater than 98% in all cultures.
E€ect of KAI1 on aggregatability of colon cancer cells
Since some members of the TM4 family have been
shown to be related with intercellular adhesion, cell
aggregation assay was performed to reveal if KAI1 is
also involved in it. Single cell suspensions of sense and
antisense transfectants were prepared and allowed to
Figure 3 Detection of KAI1 protein in transfectants of DLD-1
cells by ¯ow cytometry. The reactivity of DLD/con (a), DLD/s+
(b) and DLD/s++ (c) with an anti-KAI1 polyclonal antibody
are shown. In CONTROL and KAI-1, diluted rabbit serum and
anti-KAI1 polyclonal antibody were used as the primary
antibody, respectively
aggregate in saline with either 0 or 5 mM Ca++ at 378C
for 1 or 2 h. As shown in Figure 5a and Figure 5b,
antisense KAI1 transfectants BM/as77 cells showed
signi®cantly decreased cell aggregation as compared
with BM/con cells at 2 h after culture irrelevant of the
addition of 5 mM Ca++ (**P50.01). On the contrary,
a signi®cant increase of cell aggregation was observed
in sense KAI1 transfectants DLD/s++ and DLD/s+
cells at 1 h and 2 h after culture under the Ca++ free
conditions (Figure 5c) (**P50.01). This trend was not
a€ected by the addition of 5 mM Ca++ (Figure 5d),
although a signi®cant di€erence was obtained only in
DLD/s++ at 2 h after culture. These results indicate
that KAI1 is involved in the cell to cell adhesion of
colon cancer BM314 and DLD-1 cells.
E€ect of KAI1 on motility and invasive ability of colon
cancer cells
To explore if KAI1 has some roles as an invasion
suppressor in colon cancer cells, phagokinetic motility
of sense and antisense transfectants was investigated
by using a gold-particle coating method. Areas of the
particle-clear zone for each clone were measured after
a 24 h incubation, and the average area was
calculated. As shown in Figure 6a, antisense KAI1
transfectants BM/as77 and BM/as7 showed increased cell motility dependent on the degree of KAI1
mRNA suppression (**P50.01). BM/as77 and BM/
as7 cells exhibited a ®vefold and a 2.7-fold higher
basal phagokinetic activity than BM/con cells,
respectively. Reversely, phagokinetic motility was
suppressed in sense KAI1 transfectants DLD/s++
and DLD/s+ in an expression level dependent
manner, which were 0.36 and 0.67 times compared
with that of DLD/con cells, respectively (Figure 6b)
(*P50.05, **P50.01). The results obtained revealed
that KAI1 negatively regulates the phagokinetic
motility of those colon cancer cells.
Figure 4 In vitro growth analyses of antisense or sense KAI1
transfectants. Approximately 16104 viable tumor cells/well were
plated in 96-well tissue culture plates and incubated at 378C for
the periods indicated, and cell growth was quanti®ed by MTT
assay. (a) BM314 transfectant clones with antisense KAI1 cDNA.
Open circles, BM/as77 cells; open squares, BM/as7 cells; open
triangles, BM/con cells. (b) DLD-1 transfectant clones with sense
KAI1 cDNA. Closed circles, DLD/s++ cells; closed squares,
DLD/s+ cells; closed triangles, DLD/con cells. Data represent
the mean+s.e.m. of quadruplicate cultures. Using the Student's ttest, no signi®cant di€erence was observed both among
transfectants with various KAI1 mRNA expression levels and
between control cells and KAI1 transfectants
1445
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1446
Figure 6 E€ect of antisense or sense KAI1 gene transfer on
motility of colon cancer cells as determined by phagokinesis on
gold colloid method. BM314 transfectants with antisense KAI1
cDNA (a) and DLD-1 transfectants with sense KAI1 cDNA (b)
were plated on colloidal-gold-coated coverslips in culture medium.
After 24 h, cells were photographed and areas cleared by
migration of each clone were measured for cell motility. Shown
in the ®gure are means+s.e.m.s (n=30). Asterisks indicate that
the di€erences were statistically signi®cant (*P50.05, **P50.01)
Figure 5 E€ect of antisense or sense KAI1 gene transfer on cell
aggregatability of colon cancer cells. Cell aggregation assay was
carried out by generally following the method by Rojas et al.
(1990). 16105 cells were used at the initiation of the assay. Each
tumor cell clone transfected with KAI1 cDNA was incubated in
Puck's saline containing no Ca++ or 5 mM Ca++: BM314 cells
transfected with antisense cDNA of KAI1 in the absence (a) and
presence (b) of 5 mM Ca++; DLD-1 cells transfected with sense
cDNA of KAI1 in the absence (c) and presence (d) of 5 mM
Ca++. The degree of aggregate was examined under a
microscope, and results were expressed as a percent of total cell
numbers forming aggregates at speci®c time points as indicated.
The percent values represent a mean of triplicates with standard
error. Asterisks indicate that the di€erences were statistically
signi®cant (*P50.05, **P50.01)
We then performed in vitro cell invasion assay. The
number of migrated cells through the ®lter coated by
reconstituted basement membrane was counted.
Results of three independent experiments for comparative invasiveness among sense and antisense
transfectants are presented in Figure 7. Antisense
KAI1 transfectant BM/as77 cells showed significantly enhanced invasiveness as compared with
control BM/con cells (*P50.05), whereas a significant suppression of invasion was observed in sense
KAI1 transfectant DLD/s++ cells compared with
DLD/con cells (**P50.01). These indicate that KAI1
suppresses the in vitro invasive ability of those colon
cancer cells.
E€ect of KAI1 on binding of colon cancer cells to
extracellular matrices
Tumor cells interact with extracellular matrix (ECM)
components and basement membrane, which is an
essential initial event during the process of invasion
and metastasis. To analyse the mechanism for the
decreased invasion of BM314 and DLD-1 cells, we
examined the e€ect of KAI1 gene expression on the
Figure 7 E€ect of antisense or sense KAI1 gene transfer on
invasive ability of colon cancer cells. BM314 transfectants with
antisense KaI1 cDNA (a) and DLD-1 transfectants with sense
KAI1 cDNA (b) were incubated using transwell chambers
equipped with reconstituted basement membrane-coated ®lters
for 24 h. Cells which had migrated through the membrane and
stuck to the lower surface of the ®lter were counted under a
microscope in ®ve pre-determined ®elds at a magni®cation of
6200. The bars represent the standard error of the mean of three
separate experiments. The di€erences observed between BM/
as77 or DLD/s++ and each control were statistically
signi®cant (*P50.05, **P50.01)
binding of sense and antisense transfectants to
ECMs, ®bronectin, type IV collagen, type I collagen
and laminin by using cell attachment assay. As a
result, only ®bronectin showed a signi®cant e€ect on
the binding of KAI1 transfectants. Figure 8 displays
the results for antisense and sense KAI1 transfectants
where BM/as77 cells exhibited signi®cantly increased binding to ®bronectin compared with control
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1447
Figure 8 E€ect of antisense or sense KAI1 gene transfer on cell
attachment assay to ®bronectin. 46104 cells in 200 ml of medium
were allowed to attach to each well of 96-well plates that had
been preincubated with ®bronectin. After incubation for 1 h,
unbound cells were aspirated, and the number of attached cells
were measured by crystal violet staining. Each experiment was
performed in quadruplicated wells, and data are shown as means
of triplicate determinations. Bars indicate standard errors.
Asterisks indicate that the di€erences were statistically significant
(**P50.01)
BM/con cells (**P50.01), whereas a signi®cant
decrease of binding to ®bronectin of DLD/s++
cells was observed when compared with DLD/con
cells (**P50.01). To analyse the e€ect of KAI1 gene
transfer on the expression of ®bronectin receptors in
BM314 and DLD-1 cells, ¯owcytometry with
monoclonal antibodies (mAbs) was performed. There
was no signi®cant di€erence on CD49c and CD49e
expression levels between transfectants and their
controls, i.e., sense and antisense KAI1 transfectants
equally expressed a3b1 integrin, whereas none of
them expressed a5b1 integrin (data not shown).
These data show that expression of KAI1 gene
prevents those colon cancer cells from binding to
®bronectin.
E€ect of KAI1 on locomotion of colon cancer cells on
®bronectin-coated plates
Modi®ed wound assay using ®bronectin-precoated
culture plates was then performed to further study
the signi®cance of decreased binding of KAI1expressed colon cancer cells to ®bronectin. The cell
migration was quantitated by calculating the percentage of width of wounded area ®lled by the migrated
cells to that of wounded area at the beginning of the
assay. As shown in Figure 9a, antisense KAI1
transfectant BM/as77 cells demonstrated a significant increase in cell migration on the ®bronectincoated surface as compared with control cells
(**P50.01). No signi®cant di€erence was seen
among antisense KAI1 transfectants on non-treated
plates. Figure 9b presents the results for sense KAI1
transfectants where the converse trend was apparent,
in comparison with those for antisense transfectants.
Figure 9 E€ect of antisense or sense KAI1 gene transfer on the
®bronectin-mediated migration of colon cancer cells. Con¯uent
cultures of sense or antisense KAI1 transfectants on ®bronectincoated dishes were wounded with a disposable pipette tip as
described in Materials and methods. Then, photographs were
taken, compared to the photos after 24 h-incubation, and
migration was quantitated by calculating the ®lling rate of the
wounded area as described. Each sample was tested in triplicate
and data represent the mean plus or minus the standard error.
Both antisense KAI1 transfectants (a) and sense KAI1 transfectants (b) were examined in this assay. Asterisks indicate that the
di€erences were statistically signi®cant (**P50.01). FN indicates
®bronectin
Sense KAI1 transfectant DLD/s++ cells showed a
signi®cantly decreased cell migration on ®bronectincoated surfaces compared to control DLD/con cells,
but not on non-treated surfaces (**P50.01). These
®ndings indicate that KAI1 suppresses the ®bronectinmediated locomotion of colon cancer BM314 and
DLD-1 cells at least partly through decreasing their
motility and binding ability to ®bronectin.
Immunohistochemical detection of KAI1 protein in colon
cancer tissues
We attempted the preliminary immunohistochemical
studies of 14 primary and four liver-metastatic colon
cancer tissue specimens ®xed with AMeX method
(Sato et al., 1986) using anti-KAI1 antibody against
C-terminal peptide to study the clinical signi®cance of
KAI1 expression in colon cancer. Primary tumors were
histologically classi®ed into three groups, i.e. four well
di€erentiated, seven moderately di€erentiated and
three poorly di€erentiated adenocarcinomas. The
entire area of one section per each tissue specimen
was evaluated. The staining intensity of well or
moderately di€erentiated adenocarcinoma was almost
the same as that of normal epithelial cells. Interestingly, a total loss of KAI1 protein expression was
observed in one out of 14 primary colon cancer tissue
specimens whose histological type was poorly differentiated adenocarcinoma. Furthermore, two of four
metastatic cancer tissue specimens did not express
KAI1 protein in spite of the fact that the corresponding primary lesions were positive for the antibody.
Their representative ®ndings are shown in Figure 10.
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1448
The cytoplasm except for mucus was positively
immuno-stained in normal colon epithelial cells
(Figure 10a). Some mononuclear cells in the lamina
propria of the colon (Figure 10a) and in the noncancerous region of the liver (Figure 10c) were also
positive.
a
b
c
Figure 10 Immunohistochemical detection of KAI1 protein in a
case of colon cancer with liver metastasis. Normal colonic
epithelium adjacent to cancer region (a) (6190) and the primary
cancer tissue (b) (6190) were positively immunostained whereas
the metastatic lesion in the liver (c) (6190) was negative
Discussion
The ®rst report about the cloning of KAI1 gene
indicated that KAI1 gene has suppressive e€ects on
the metastasis of the rat prostate cancer. They also
showed that the introduction of KAI1 gene by
microcell-mediated chromosome transfer method resulted in no e€ect on their metastatic ability of highly
metastatic rat mammary cancer cells despite the certain
amounts of KAI1 expression (Rinker-Schae€er et al.,
1994; Dong et al., 1995). Furthermore, a recent
immunohistochemical study has shown that the
presence of KAI1 protein in prostate cancer has a
negatively correlated with advanced clinical stages
(Ueda et al., 1991). These ®ndings seem to suggest
that KAI1's capability of metastasis suppression may
have a tumor-speci®c action and be limited to prostate
cancers (Marx, 1995). However, there have recently
been several studies upon the association of KAI1
expression with clinicopathological characteristics,
which support the possibility that KAI1 could have a
metastasis suppressor e€ect in other kinds of tumors as
well as in prostatic cancer. It was documented that
conserved KAI1 gene expression is associated with
good prognosis in patients with non-small cell lung
cancer, suggesting that the levels of expression of KAI1
gene may be a potent prognostic factor (Adachi et al.,
1996). Evaluation of KAI1 mRNA expression in
relation with clinical parameters of pancreatic cancer
patients revealed that the levels of KAI1 mRNA
expression is signi®cantly down-regulated in pancreatic cancer patients with advanced tumor stages with
lymph node metastasis compared with those without it
(Guo et al., 1996). These observations and the broad
tissue distribution of KAI1 message (Dong et al., 1995)
suggest a more universal role of KAI1 as a metastasis
suppressor. Normal small intestine and colon tissues
were found to express KAI1 message (Dong et al.,
1995), but there have been no reports on the expression
and functional analyses of KAI1 in gastrointestinal
cancers. In the present study, we therefore examined
the expression of KAI1 mRNA in various gastric and
colon cancer cell lines. Northern blot analysis revealed
undetectable or very low expression level of KAI1
message in six of ten (60%) gastric and colon cancer
cell lines. To study the e€ect of KAI1 on the invasive
and metastatic properties of colon cancer cells, we then
prepared stable KAI1 transfectants using human colon
cancer cell lines. A key aspect of our study has been to
use both the antisense and sense KAI1 transfectants to
ensure the reproducibility of obtained data, i.e.,
antisense transfectants of BM314 cells whose KAI1
mRNA expression was suppressed by transfer of
antisense KAI1 cDNA and sense transfectants of
DLD-1 cells with the enhanced KAI1 mRNA by sense
cDNA transfer.
In cell aggregation assay, antisense and sense KAI1
transfectants exhibited decreased and increased aggregatability, respectively, as compared with each control.
This suggests an important role of KAI1 in intercellular
adhesion. There have been several lines of evidences
showing the involvement of TM4 family members in
cell adhesion. C33 antigen which is identical to KAI1
had been identi®ed as the target antigen of a mAb
inhibitory to HTLV-1-induced syncytium formation
(Fukudome et al., 1992). This antigen is thought not to
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
be the HTLV-1 receptor itself, but to be involved in
cell adhesion in the process of HTLV-1-induced
syncytium formation in human T cells. Also, other
members of the TM4 family have been shown to be
involved in the regulation of cellular adhesion;
stimulation of CD9 by mAb induces homotypic
adhesion of pre-B cell lines (Straus et al., 1989) and
platelet aggregation (Schaepfer et al., 1994), antibody
against CD81 triggers homotypic adhesion of various
lymphoid cell lines (Takahashi et al., 1990), and rat
CD53/OX-44 is associated with CD2, the adhesion
receptor in activated T cell and NK cell lines (Bell et
al., 1992). Considering these observations, our data
suggest that reduced expression of KAI1 gene in colon
cancer cells results in the suppressive e€ect on cell-tocell adhesion. These data also propose a hypothesis
that detachment from primary lesion might be one of
the aspects where KAI1 would a€ect as a suppressive
regulator of the metastatic process. Further studies are
required to determine whether KAI1 participates
directly or indirectly in the homotypic aggregation of
colon cancer cell lines.
To see the e€ect of KAI1 expression on the invasive
properties of colon cancer cells, we carried out
phagokinetic track assay (Figure 6) and in vitro
invasion assay (Figure 7). They revealed the negative
correlation between the expression level of KAI1
mRNA of sense transfectants and their motility or
invasiveness, which was con®rmed by the assays with
antisense KAI1 transfectants. Of interest is the
observation on the CD9/MRP-1 antigen and CD63/
ME491 that belong to other members of the TM4
family. CD9/MRP-1 has been reported to have little
e€ect on the proliferation of some tumor cell lines
including lung adenocarcinoma MAC10 (Miyake et al.,
1991). The introduction of CD9/MRP-1 into a mouse
malignant melanoma cell line showed its diminished
cell motility and pulmonary metastatic potential
(Ikeyama et al., 1991). Furthermore, the CD9/MRP-1
expression has been reported to be inversely correlated
with the progression of lymph node metastasis in
breast cancer (Miyake et al., 1996). Likewise, the
CD63/ME491, a stage-speci®c melanoma-associated
antigen, has been described as being not detected in
normal melanocyte, but abundantly expressed in
human malignant melanoma and precancerous regions
of melanoma, and further reduced in dermis-invading
and metastatic melanomas. The reduced expression of
CD63/ME491 antigen is thought to be associated with
enhanced invasive and metastatic abilities of human
malignant melanoma (Hotta et al., 1988; Kondoh et
al., 1993; Azorsa et al., 1991). These suggest that antiinvasive and metastatic e€ect may be shared by some
members of the TM4 family. In this context, it is
obviously important to observe if KAI1 a€ects the in
vivo invasive and metastatic abilities of tumor cells.
However, we have not thus far obtained any data on
the in vivo behavior of KAI1 cDNA-transfected BM314
and DLD-1 cells, because those cell lines did not show
metastasis at all when transplanted in immunode®cient
mice. We have examined the e€ect of KAI1 on the
experimental metastasis of KAI1 cDNA-transfected
mouse melanoma B16 cells. As a result, the number
of metastatic nodules of lungs were signi®cantly
decreased in KAI1 transfectants compared to mock
transfectants and parental cells (unpublished data).
Furthermore, preliminary immunohistochemical staining was performed for surgically resected 14 human
colon cancer tissue specimens. A total loss of KAI1
protein expression was observed in one primary colon
cancer tissue specimen whose histological type was
poorly di€erentiated adenocarcinoma, and in two livermetastasis specimens in spite of the fact that the
corresponding primary lesions were positive for the
anti-KAI1 antibody. When some tumor cell clones
acquire the lower expression level of KAI1, they might
metastasize to the liver. Taken together, these data may
suggest the possible involvement of KAI1 in the in vivo
metastasis of tumor cells.
Attachment of cancer cells to ECM and subsequent
cellular movement are essential steps for invasion. To
analyse the mechanism for the decreased invasiveness
of BM314 and DLD-1 cells by KAI1 expression, we
examined the e€ects of KAI1 on the adhesive ability of
colon cancer cells to several ECMs. With increasing the
level of KAI1 expression, sense KAI1-transfected clones
showed the reduced ability to bind to ®bronectin,
whereas the inverse trend was observed in the antisense
KAI1-transfected clones. For the interaction with other
ECMs, type IV collagen, type I collagen and laminin,
no signi®cant e€ect on the binding ability of KAI1
transfectants was observed. These data facilitated us to
investigate whether KAI1 may in¯uence the regulation
of the expression level of ®bronectin receptors.
However, the ¯ow cytometric analysis showed that
there was no di€erence of the expression level of a3b1
integrins known as ®bronectin receptors among those
transfectants (Chintala et al., 1996; Wu et al., 1993;
Ruoslahti, 1996). To further examine the e€ect of
KAI1 on the interaction of colon cancer cells with
®bronectin, the ®bronectin-mediated locomotion of
KAI1 transfectants was observed by using wound
assay (Figure 9). A negative e€ect of KAI1 on the
locomotion of those cells on ®bronectin-coated surfaces
was consequently obtained, suggesting the importance
of ®bronectin for the expression of anti-invasion e€ect
of KAI1.
It should be noted here that KAI1 could a€ect the
binding and locomotion of BM314 and DLD-1 cells
on ®bronectin-coated surfaces without any in¯uences
on the expression level of ®bronectin receptors. A
presumption from these results is that KAI1 might
have some involvement in regulating the function of
®bronectin receptors. Previous reports showed that
®bronectin receptors function not only in attachment
or cellular adhesion, but in an intracellular signaling
system where the binding of ®bronectin to a cellsurface ®bronectin receptor conveys controlling
signals to cell locomotion (Liotta et al., 1986; Straus
et al., 1989; Fearon, 1993; Schaepfer et al., 1994). As
reported for CD82/KAI1 expressed on hematopoietic
cells, CD82 may play an essential role as a regulator
molecule in some signal transduction pathways, and
engagement of CD82/KAI1 delivers di€erent signals
on the cell type; for example, a costimulatory signal
for full activation for the human T cell line, Jurkat
(Lebel-Binay et al., 1995a), or an enhancing signal in
FcR-mediated activation of monocyte/macrophage
cell lines (Lebel-Binay et al., 1995b). Some members
of the TM4 family including CD82/KAI1 have been
demonstrated to form multimolecular membrane
complexes by structurally or functionally associating
1449
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1450
with each member and with other molecules, many of
which appear to be involved in signal transduction.
CD9/MRP-1 have been demonstrated to be associated with VLA integrins in the pre-B cell and
megakaryocytic cell lines, and this association a€ected
aggregation and migration of the two cell lines
(Rubinstein et al., 1994). In accordance with a
signaling function, CD9/MRP-1 is reported to
associate with small GTP-binding proteins (Seehafer
and Shaw, 1991). This report is intriguing with a
view point that both CD9/MRP-1 and CD82/KAI1
belong to the TM4 family, and are furthermore
involved in cell migration, invasion, and metastasis.
These ®ndings provide some speculations that KAI1
alone or with other member(s) might be involved to
some extent in modulating the intracellular signaling
for cell locomotion. Further studies are in progress in
our laboratory to investigate some involvement of
KAI1 with integrins.
Our current data using sense and antisense KAI1
transfectants clearly showed the suppressive e€ect of
KAI1 on the intercellular adhesion, attachment to
ECMs, cell motility and invasiveness of colon cancer
BM314 and DLD-1 cells. However, if the functions of
KAI1 are of real importance in those cells, the e€ect of
KAI1 would also be observed in parental cells. Indeed,
in cell aggregation assay, BM/con cells showed a
tendency of higher aggregatability under the Ca++-free
conditions than DLD/con cells as shown in (Figure 5).
DLD-1 cells (DLD/con) which has by far least
expression of KAI1 message (Figure 1) tended to
demonstrate better migratory and invasive responses
than BM314 cells (BM/con) which express certain
amounts of KAI1 mRNA (Figure 1) as shown in
Figures 6 and 7. Also, a tendency of higher ®bronectinmediated locomotion of DLD/con cells was observed
as compared to that of BM/con cells (Figure 9). These
data may strengthen the ®ndings obtained by using
KAI1 transfectants.
Materials and methods
Tumor cell lines and culture conditions
Both gastric and colon cancer cell lines used for these
studies were obtained from the Japanese Cancer Research
Resources Bank. Gastric cancer cell lines, AZ521, JRST,
MKN45, MKN74, NUGC3, were maintained in RPMI
1640 supplemented with 10% fetal bovine serum (FBS).
Colon cancer cell lines, CHCY1, Colo201, Colo320DM,
DLD-1 except BM314 were maintained in the same way.
BM314 cells were maintained in Dulbecco's modi®ed
minimum essential medium containing 10% FBS. Cell
cultures were maintained and incubated in 5% CO2-95%
air at 378C.
Northern blot analysis of tumor cell lines
Total RNA was extracted from the indicated cell lines
using ISOGEN kit (Nippon Gene, Toyama, Japan)
according to the manufacturer's protocol. Ten mg of each
RNA sample per lane was fractionated through a 1.0%
agarose gel containing 2.2 M formaldehyde, and transferred
to a nitrocellulose membrane in 106SSC. Hybridization to
32
P-labeled full-length of KAI1 cDNA was carried out in
50% formamide, 56SSPE (16SSPE=0.18 M NaCl,
10 mM sodium phosphate, pH 7.7, 1 mM EDTA),
56Denhart's solution, 5 mM EDTA, 0.1% SDS, and
100 mg/ml denatured salmon sperm DNA at 428C for
18 h. The ®lters were washed twice in 26SSC, 0.1% SDS
at room temperature for 10 min and twice in 0.16SSC,
0.1% SDS at 558C for 10 min. KAI1 cDNA probe was
radiolabeled with a-32P-dCTP by random priming. After
hybridization, all blots were exposed to Kodak XAR ®lms
with an intensifying screen at 7708C. As a control for the
amount of mRNA loaded, rRNAs were stained with
ethidium bromide.
RT ± PCR and Southern blot analysis of stable transfectants
According to the nucleotide sequence of the R2 cDNA
identi®ed by Gaugitsch et al. (1991), speci®c PCR primers
were designed to recognize the full coding region of KAI1
cDNA: forward `HindIII-MGSACIK' primer, 5'-CC-AAGCTT-ATG-GGC-TCA-GCC-TGT-ATC-AAA-G-3', primer
1; reverse `HindIII-TCA-YKPVKS' primer, 5'-CC-AA-GCTTTCA-GTA-CTT-GGG-GAC-CTT-GCT-G-3', primer 2.
One mg of total RNA was reverse-transcribed with random
nanomers and avian myeloblastosis virus reverse transcriptase using RT ± PCR kit (Takara, Otsu, Japan) following
the conditions of the manufacturer. The template cDNAs
were ampli®ed with Taq polymerase in the presence of
primers 1 and 2 on the basis of PCR protocol (Saiki et al.,
1985). The thermocycling parameters used in the PCR were
as follows: denaturation, 30 s at 948C; annealing, 30 s at
608C; extension, 1.5 min at 728C. These reactions were
repeated for 30 cycles. The PCR products were electrophoresed through a 2.0% agarose gel and blotted onto
nylon membranes (Hybond-N, Amersham, Arlington
Heights, IL) in alkaline solution (0.25 M NaOH, 0.4 M
NaCl) by capillary transfer. The full-length KAI1 cDNA
probe was labeled with a-32P-dCTP using a random primer
labeling kit (Takara, Otsu, Japan). The ®lters were
hybridized with 32P-labeled KAI1 cDNA probe in 50%
formamide, 56Denhart's solution, 36SSC, 100 mg/ml
salmon sperm DNA, 1% sodium dedecyl sulfate (SDS) at
428C overnight. The membranes were then washed three
times in 26SSC, 0.1% SDS at room temperature for
10 min and twice in 0.16SSC, 0.1% SDS at 558C for
15 min, and exposed to Kodak XAR ®lms at 7708C with
an intensifying screen. Similarly, b-actin was ampli®ed and
hybridized with 32P-labeled b-actin cDNA probe for an
internal control. The e€ect of antisense RNA on
endogenous KAI1 mRNA expression level was investigated by RT ± PCR analysis using primer 1 and primer
3, 5'-CCTCCTCCCCAATTTTTCATACAGTTGC-CC-3',
which was designed to be corresponded to the part of
3'-untranslated region from nucleotide position 1284 to
1313 (Gaugitsch et al., 1991) to speci®cally amplify the
endogenous KAI1 mRNA.
Flow cytometric analysis
The transfectants were removed from plastic surface by
5 min incubation at 378C with 5 mM EDTA-PBS. After
washing with cold phosphate-bu€ered saline (PBS), the
cells were permeabilized with cold ethanol for 5 min, and
washed again. They were then incubated with an
appropriate dilution of anti-KAI1 polyclonal antibody
(Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or of
rabbit serum as a negative control for 30 min at 48C. After
being washed three times with cold PBS, they were
incubated with ¯uorescein isothiocyanate-labeled goat
antibody to rabbit immunoglobulins (Dakopatts, Copenhagen, Denmark) for 30 min at 48C. Washing was repeated in
the same manner, and cell surface immuno¯uorescence was
analysed by ¯ow cytometry (Cytron, Ortho, Coulter
Electronics, FL). To detect integrins on transfectants, the
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
unpermeabilized cells were incubated with an appropriate
dilution of a monoclonal antibody against either human
a5b1 integrin (CD49e) (0.2 mg/ml, Pharmingen, San Diego,
CA) or human a3b1 integrin (CD49c) (0.5 mg/ml,
Immunotech, Marseille, France) for 30 min at 48C. After
incubation with the primary antibody, the cells were
washed and incubated with ¯uorescein isothiocyanatelabeled goat antibody to mouse IgG as described above,
and were subjected to ¯ow cytometry.
Constructs
The eukaryotic expression vector (pRc/CMV) was obtained
from Invitrogen (San Diego, CA). The vector was designed
for high-level stable expression, containing the cytomegalovirus promoter and enhancer and neomycin-resistant
gene for the selection of stable transformants. In order to
clone the full-length cDNA encoding human KAI1 into the
expression vector pRc/CMV, the entire coding region of
KAI1 cDNA was ampli®ed from total RNA derived from
normal colon mucosa by RT ± PCR using primers 1 and 2
as described above. The resulting 803 bp PCR product was
digested with HindIII, and subcloned into the HindIII site
of pRc/CMV vector. Both sense (KAI1/CMV-s) and
antisense (KAI1/CMV-as) orientations were prepared.
Con®rmation of the constructs was obtained by dideoxy
sequencing.
Transfection procedures and selection of stable clones
Sense cDNA of KAI1 (KAI1/CMV-s) was transfected into
DLD-1 cells, which constitutively expressed KAI1
mRNA, while antisense cDNA of KAI1 (KAI1/CMV-as)
was introduced into BM314 cells, which showed least
expression of endogenous KAI1 mRNA. Transfections
were initiated following the Lipofection (GIBCO BRL,
Grand Island, NY) technique. Brie¯y, media was changed
to OPTI-MEM I, reduced serum medium (GIBCO BRL).
Two mg of plasmid was mixed with 10 ml of LIPOFECTAMINE, reagent, incubated in a six-well tissue culture
plate of 70% con¯uent cells for 5 h at 378C in a CO2
incubator. Following incubation, FBS was added to the
plates (®nal FBS concentration was 10%), and cell
cultures were incubated for 48 h. Cells were detached
by trypsin-ethylene diamine tetraacetic acid (EDTA)
solution and passaged 1 : 10 into the selective media
containing 700 mg/ml GENETICIN (GIBCO BRL).
Surviving clones were characterized by RT ± PCR
analysis for their expressions of KAI1 mRNA, and the
clones with di€erent levels of KAI1 expression were used
for the functional experiments described. Similarly, a
clone transfected with a vector (pRc/CMV) alone was
used as a control.
MTT assay
Cell growth was measured using a modi®ed 3-[4,
5 - dimethylthiazol - 2 - yl] -2,5-diphenyltetrazolium bromide
(MTT) method (Mosmann, 1983; Tada et al., 1986). Cells
were seeded into 96-well plates at 16104 cells per well in
culture media (100 ml). The plates were incubated for up to
4 days. In the MTT assay, 10 ml of the MTT solution
(5 mg/ml) were added to each well and the cells were
cultured for another 4 h and then 100 ml of isopropanol0.04 N HCl were added to each well, mixed vigorously to
solubilize colored crystals produced within the cells. The
ratio of absorbance at 590 nm to absorbance at 630 nm as
reference wave was measured by a multiwell scanning
spectrophotometer (Titertek multiskan, Flow Laboratories). Cell viability was examined by trypan blue dye
exclusion test.
Cell aggregation assay
Cell aggregation experiments were performed following the
procedure by Rojas et al. (1990) with slight modi®cation.
Adherent tumor cells were rendered single cell suspensions
by a 5 min incubation at 378C with 5 mM EDTA in
phosphate bu€ered saline (PBS) lacking Ca++ and Mg++
and by 2 passes through a 27-gauge needle. Twenty fourwell plates were used for the assay of cell aggregation.
Suspensions of single cells were washed three times with
Puck's saline (5 mM KCl, 140 mM NaCl, 8 mM NaHCO3,
pH 7.4), resuspended at a concentration of 16105 cells/ml
in 1 ml of Puck's saline plus 0.8% FBS, 50 U/ml
deoxyribonuclease I (Sigma Co., St. Louis, MO), and
incubated in an atmosphere of 5% CO 2-95% air at 378C,
agitated at 70 ± 80 r.p.m. with a gyratory shaker. Samples
were taken at the points of 1 h- and 2 h-incubation, and
both the total cell numbers (A) and the number of cells
remaining as single cells (B) were counted under a
microscope on three predetermined ®elds. The results
were shown as the percentage of cells that formed
aggregates as follows: (A ± B)/A6100[%]. To determine
the Ca++ dependence of aggregation, suspensions of single
cells were resuspended according to the general procedure
at 16105 cells/ml of Puck's saline plus 0.8% FBS, 50 U/ml
DNase I with either 5 mM Ca++ or 0 mM Ca++, these
suspensions were stirred at 378C, and the aggregation rate
was calculated as above.
Phagokinetic track motility assay
Cell motility was determined on the basis of phagokinetic
tracks on a gold particle-coated plate. Brie¯y, uniform
carpets of gold particles were prepared on cover slips coated
with bovine serum albumin, as previously described by
Albrecht-Buchler (1977). The cover slips were rinsed
extensively to remove nonadhering gold particles before
cell plating. Colloidal-gold-coated coverslips were placed in
35 mm tissue culture dishes containing 2 ml culture medium,
and then 56103 freshly trypsinized cells were added to each
plate and left in the incubator for 24 h. Photographs were
taken, and areas cleared by a single cultured cell were traced
on semitransparent paper of uniform thickness. The
particle-free areas of at least 40 individual cells were
measured, and average areas were calculated.
Cell invasion assay
This assay was based on the methods described by Albini
et al. (1987), Repesh (1989), and Parish et al. (1992). We
used 6-well BIOCOAT MATRIGEL Invasion Chambers
(Becton Dickinson, Bedford, MA) that consist of Cell
Culture Inserts (Falcon, Oxnard, CA) containing an 8 mm
pore size polycarbonate membrane that has been coated
with a solubilized tissue basement membrane `MATRIGEL' (Collaborative Research, Bedford, MA), 100 mg/cm2.
Brie¯y, after rehydration of the membrane, 1.56104 cells
were added into each upper compartment. After 24 h
culture, the non-invasive cells and matrigel were removed
with a cotton swab, and the cells which had migrated
through the membrane and stuck to the lower surface of
the polycarbonate membrane were ®xed with methanol and
stained with Giemsa's staining solution (GIBCO BRL).
For quanti®cation, cells that had migrated to the lower
surface were counted under a microscope in ®ve predetermined ®elds at a magni®cation of 6200.
Attachment assay
The cell attachment assay was based on the method
described by Furukawa et al. (1993). Fibronectin (10 mg/
well) was added to 96-well plates, which were incubated for
1451
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
1452
24 h, to allow binding to the surface of the well. The plates
were then washed by PBS and, bovine serum albumin
(BSA) (1.5% w/v) was added to them, which were
incubated for 1 h at 378C to block remaining binding
sites. Cells were detached with 5 mM EDTA/PBS and
resuspended in culture media containing 0.02% BSA and
counted with a hemocytometer. The number of cells was
adjusted to 26105/ml in culture media with 0.02% BSA; an
aliquot (200 ml) of cell suspension was added to each
®bronectin-coated well. Cells were incubated at 378C in a
humidi®ed atmosphere of 95% air and 5% CO 2 for 1 h.
Cell attachment was veri®ed by crystal violet staining
(Gillies et al., 1986). Brie¯y, after unbound cells were
removed by aspiration, 100 ml of 0.04% crystal violet
solution was added and incubated for 10 min at room
temperature, followed by one wash with PBS. The crystal
violet that absorbed onto the cells was solubilized with
0.1% TritonX-100 (Sigma). The absorbance was measured
at 590 nm with a microplate reader. Each experiment was
performed in quadruplicate wells, and the cell attachment
to ®bronectin was determined as follows: cells bound
(%)=(mean absorbance of cells attached to wells/mean
absorbance of total number of cells)6100.
Wound assay
56105 transfectants were seeded in 2 ml of culture medium
containing 10% FBS in 35 mm plastic plates which had
been preincubated with ®bronectin or non-treated dishes.
After 24 h, a wound was made in the con¯uent monolayer
with a disposable pipette tip as described by Burk (1973)
with slight modi®cation. The medium and debris were
aspirated away and replaced by 2 ml of fresh medium. For
the evaluation, photographs of wounded areas were taken
both just after making a wound and after 24 h-incubation.
The migration was quanti®ed by calculating ®lling rate
(%)=(A ± B)/A6100 [A, width of a wounded area at the
starting; B, width of a wounded area after 24 hincubation]. Each sample was tested in triplicate and data
represent the mean plus or minus the standard error of the
mean (+s.e.m.).
Immunohistochemistry
Surgically resected 14 primary and four metastatic colonic
cancer tissues were obtained from the patients treated at
Sapporo Medical University Hospital. Immunohistochemical staining was done on AMeX-®xed (Sato et al., 1986)
and paran-embedded tissue sections by an immunoperoxidase method as described previously (Ohe et al.,
1994). Brie¯y, each section was deparanized, rehydrated
and incubated with fresh 0.3% hydrogen peroxide in
methanol for 20 min at room temperature and then
washed with PBS. Normal goat serum (5%) was applied
for 20 min and removed by blotting. The sections were
incubated with anti-KAI1 polyclonal antibody (Santa Cruz
Biotechnology, Inc.) for 60 min at room temperature,
washed three times in PBS, and incubated with secondary
antibody (peroxidase-conjugated goat anti-rabbit Ig diluted
1/50 in PBS) for 30 min at room temperature. After
washing three times, the sections were incubated with
diaminobenzidine tetrahydrochloride in 0.03% hydrogen
peroxide for 5 ± 10 min, washed, counterstained with
hematoxylin, rinsed in tap water and mounted. Diluted
rabbit serum was used as a negative control for the
primary antibody. Histological classi®cation of colon
cancer was according to Jass and Sobin (1989).
Statistical evaluation
Di€erences between control and experimental groups were
evaluated using the Student's t-test, by the Statworks
program. A P-value50.05 was considered signi®cant.
Acknowledgements
We would like to thank the Japanese Cancer Research
Resources Bank for providing cell lines. This work was
supported by Grant-in-Aid for Scienti®c Research on
Priority Areas from the Ministry of Education, Science,
Sports and Culture [KI, YH] and by Grants for Cancer
Research from the Ministry of Health and Welfare [KI,
YH], Japan.
References
Adachi M, Taki T, Ieki Y, Huang CL, Higashiyama M and
Miyake M. (1996). Cancer Res., 56, 1751 ± 1755.
Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson
SA, Kozlowski JM and McEwan RN. (1987). Cancer Res.,
47, 3239 ± 3245.
Albrecht-Buehler G. (1977). Cell, 11, 359 ± 404.
Aznavoorian S, Murphy AN, Stetler-Stevenson WG and
Liotta LA. (1993). Cancer, 71, 1368 ± 1383.
Azorsa DO, Hyman JA and Hildreth JE. (1991). Blood, 78,
280 ± 284.
Barnes R, Masood S, Barker E, Rosengand AM, Coggin DL,
Crowell T, King CR, Porter-Jordan K, Wargotz ES,
Liotta LA and Steeg PS. (1990). Am. J. Pathol., 139, 254 ±
250.
Behrens J, Mareel MM, Van Roy FM and Birchmeier W.
(1989). J. Cell. Biol., 108, 2435 ± 2447.
Behrens J, Vakaet Jr, L, Friis R, Winterhager E, Van Roy
FM, Mareel MM and Birchmeier W. (1993). J. Cell. Biol.,
120, 757 ± 766.
Bell GM, Seaman WE, Niemi EC and Imboden JB. (1992). J.
Exp. Med., 175, 527 ± 536.
Boucheix C, Benoit P, Frachet P, Billard M, Worthington
RE, Gagnon J and Uzan G. (1991). J. Biol. Chem., 266,
117 ± 122.
Burk RR. (1973). Proc. Nat. Acad. Sci. USA, 70, 369 ± 372.
Chintala SK, Sawaya R, Gokaslan ZL and Rao JS. (1996).
Cancer Lett., 103, 201 ± 208.
Cohn KN, Wang F, Desoto-Lapaix F, Solomon WB,
Paterson LG, Arnold MR, Wemar J, Feldman JG, Levy
AT, Leone A and Steeg PS. (1991). Lancet, 338, 722 ± 724.
Dong JT, Lamb PW, Rinker-Schae€er CW, Vukanovic J,
Ichikawa T, Isaacs JT and Barrett JC. (1995). Science, 268,
884 ± 886.
Fearon DT. (1993). Curr. Opin. Immunol., 5, 341 ± 348.
Fearon ER, Cho KR, Nigro JM, Kern SE, Simons JW,
Ruppert JM, Hamilton SR, Preisinger AC, Thomas G,
Kinzler KW and Vogelstein B. (1990). Science, 247, 49 ±
56.
Fidler IJ and Hart IR. (1982). Science, 217, 998 ± 1001.
Frixen UH, Behrens J, Sachs M, Eberle G, Voss B, Warda A,
Lohner D and Birchmeier W. (1991). J. Cell. Biol., 113,
173 ± 185.
Fukudome K, Furuse M, Imai T, Nishimura M, Takagi S,
Hinuma Y and Yoshie O. (1992). J. Virol., 66, 1394 ± 1401.
Furukawa T, Watanabe M, Kubota T, Kase S, Fujita S,
Yamamoto T, Saikawa Y, Kuo T-H, Tanino H, Kurihara
N, Kuwano Y, Kawamoto K, Suto A, Teramoto T,
Ishibiki K and Kitajima M. (1993). J. Surg. Oncol., 53,
10 ± 16.
Invasion suppression of colon cancer cells by KAII gene
A Takaoka et al
Gaugitsch HW, Hofer E, Huber NE, Schnabl E and
Baumruker T. (1991). Eur. J. Immunol., 21, 377 ± 383.
Gil LM, Vita N, Lebel-Binay S, Miloux B, Chalon P,
Kaghad M, Marchiol-Fournigault C, Conjeaud H, Caput
D, Ferrara P, Fradelizi D and Quillet-Mary A. (1992). J.
Immunol., 148, 2826 ± 2833.
Guo X, Friess H, Graber HU, Kashiwagi M, Zimmermann
A, Korc M and Buchler MW. (1996). Cancer Res., 56,
4876 ± 4880.
Haut M, Steeg PS, Willson JKV and Markowitz SD. (1988).
J. Natl. Cancer Inst., 83, 712 ± 716.
Horejsi V and Vlcek C. (1991). FEBS Lett., 288, 1 ± 4.
Hotta H, Ross AH, Huebner K, Isobe M, Wendeborn S,
Chao MV, Ricciardi RP, Tsujimoto Y, Croce CM and
Koprowski H. (1988). Cancer Res., 48, 2955 ± 2962.
Ikeyama S, Koyama M, Yamaoko M, Sasada R and Miyake
M. (1991). J. Exp. Med., 177, 1231 ± 1237.
Imai T, Fukudome K, Takagi S, Nagira M, Furuse M,
Fukuhara N, Nishimura M, Hinuma Y and Yoshie O.
(1992). J. Immunol., 149, 2879 ± 2886.
Itoh F, Hinoda Y, Ohe M, Ban T, Endo T, Imai K and Yachi
A. (1993). Int. J. Cancer, 53, 260±263.
Jass JR and Sobin LH. (1989). Histological typing of
intestinal tumors (WHO), 2nd edn. Berlin: SpringerVerlag.
Kondoh M, Ueda M, Ichihashi M and Mishima Y. (1993).
Melanoma Res., 3, 241 ± 245.
Lebel-Binay S, Lagaudriere C, Fradelizi D and Conjeaud H.
(1995a). J. Immunol., 155, 101 ± 110.
Lebel-Binay S, Lagaudriere C, Fradelizi D, Conjeaud H.
(1995b). J. Leukocyte Biol., 57, 956 ± 963.
Leone A, Flatow U, King CE, Sandeen MA, Margulies MK,
Liotta LA and Steeg PS. (1991). Cell, 65, 25 ± 35.
Liotta LA, Rao CN and Wewer UM. (1986). Annu. Rev.
Biochem., 55, 1037 ± 1057.
Liotta LA and Stetler-Stevenson WG. (1991). Cancer Res.,
51, 5054 ± 5059.
Marx J. (1995). Science, 268, 799 ± 800.
Ministry of Health and Welfare. (1996). Statistics and
Information Department Vital Statistics 1995 Japan, 52 ±
54.
Miyake M, Koyama M, Seno M and Ikeyama S. (1991). J.
Exp. Med., 174, 1347 ± 1354.
Miyake M, Nakano K, Ieki Y, Adachi M, Huang CL, Itoi S,
Koh T and Taki T. (1996). Cancer Res., 55, 4127 ± 4131.
Mosmann T. (1983). J. Immunol. Methods, 65, 55 ± 63.
Myero€ LL and Markowitz SD. (1993). J. Natl. Cancer Inst.,
85, 147 ± 152.
Ohe Y, Hinoda Y, Irimura T, Imai K and Yachi A. (1994).
Jpn. J. Cancer Res., 85, 400 ± 408.
Ookawa K, Sakamoto M, Hirohashi S, Yoshida Y,
Sugimura T, Terada M and Yokota J. (1993). Int. J.
Cancer, 53, 382 ± 387.
Parish CR, Jakobsen KB and Coombe DR. (1992). Int. J.
Cancer, 52, 378 ± 383.
Repesh LA. (1989). Invasion Metastasis, 9, 192 ± 196.
Rinker-Schae€er CW, Hawkins AL, Ru N, Dong J, Stoica
G, Grin CA, Ichikawa T, Barrett JC and Isaacs JT.
(1994). Cancer Res., 54, 6249 ± 6256.
Rojas M, Fuks A and Stanners CP. (1990). Cell Growth &
Di€er., 1, 527 ± 533.
Rubinstein E, Le Naour F, Billard M, Prenant M and
Boucheix C. (1994). Eur. J. Immunol., 24, 3005 ± 3013.
Ruoslahti E. (1996). Tumor Biol., 17, 117 ± 124.
Saiki RK, Scharf S, Faloona F, Mullis KB, Horn GT, Erlich
H and Arnheim N. (1985). Science, 230, 1350 ± 1354.
Sato Y, Mukai K, Watanabe S, Goto M and Shimosato Y.
(1986). Am. J. Pathol., 125, 431 ± 435.
Schaepfer DD, Hank SK, Hunter T and van der Geer P.
(1994). Nature, 372, 786 ± 791.
Seehafer JG and Shaw AR. (1991). Biochem. Biophys. Res.
Commun., 179, 401 ± 406.
Straus AH, Carter WG, Wayner EA and Hakomori S.
(1989). Exp. Cell Res., 183, 126 ± 139.
Tada H, Shiho O, Kuroshima K, Koyama M and Tsukamoto
K. (1986). J. Immunol. Methods, 93, 157 ± 165.
Takahashi S, Doss C, Levy S and Levy R. (1990). J.
Immunol., 145, 2207 ± 2213.
Ueda T, Ichikawa T, Tamura J, Mikata A, Akakura K,
Akimoto S, Imai T, Yoshie O, Shiraishi T, Vleminckx K,
Vakaet Jr, L, Mareel M, Fiers W and Van Roy F. (1991).
Cell, 66, 107 ± 120.
Wang L, Patel U, Ghosh L, Chen HC and Banerjee S. (1993).
Cancer Res., 53, 717 ± 720.
Whitelaw SC and Northover JMA. (1993). Gut, 34, 289 ± 292.
Wright MD and Tomlinson MG. (1994). Immunol. Today,
15, 588 ± 594.
Wu C, Bauer JS, Juliano RL and McDonald JA. (1993). J.
Biol. Chem., 268, 21883 ± 21888.
1453