Download vitroandremaintumorigenic. However, late

CANCER RESEARCH55. 6215-6221, December 5. 19951
Loss of the Tumorigenic Phenotype with in Vitro, but not in Vivo, Passaging of a
Novel Series of Human Bronchial Epithelial Cell Lines: Possible Role of
an a5/fll-Integrin-Fibronectin
Interaction'
Joan H. Schiller@ and Gerard Bittner
William S. Middleton Veterans Administration
Madison. Wisconsin 53792
Hospital. Madison.
Wisconsin 53705, and University of Wisconsin Comprehensive
ABSTRACT
We established an immortalized, nontumorigenic human bronchial
epithelial cell line by transfection with the origin of replication-defective
SV4O large T plasmid. This line spontaneously became tumorigenic at
passage 184 (NL2OT),although subsequent passages (passages 189, 200,
and 205) failed to form tumors. The tumongenic cell line NL2OT was
reinoculated back into athymic nude mice, and the two subsequently
Cancer Center, Department
of Medicine.
primary tumor; (b) adherence and attachment of tumor cells to the
basement membrane; (c) invasion of tumor cells through the basement
membrane, with local proteolysis associated with the breakdown of
the basement membrane components; and (d) migration of tumor cells
through the defect in the extracellular matrix. This sequence of events
implicates clearly the adhesion of tumor cells to the extracellular
matrix as a crucial event in the progression of cancer.
Integrins are a family of heterodimers consisting of an a chain and a
@3chain that mediate both cell-substratum and cell-cell adhesion (2).
Functionally, integrins can be divided into three groups: those that func
lion as cell-cell adhesion molecules (primarily found on leukocytes);
those that bind primarily to the major constituents of the basement
membrane, such as collagen and laminin; and those that bind primarily to
the extracellular matrix proteins found primarily during early develop
ment, inflammation, and wound healing, such as fibronectin, fibrinogen,
vitronectin, and thrombospondin (3). The availability of monocbonal
antibodies has made it possible to determine the expression of these
molecules on cultured cells and tissues. In general, the pattern of integrin
expression on normal and transformed epithelial cells seems complex,
heterogeneous, and dependent on cell type. Only a few studies have
reported a clear correlation between the degree of transformation or
malignancy and the pattern of integrin expression (4). However, most
studies have demonstrated that high levels of a5/(3l (the fibronectin
receptor) integrin expression are correlated negatively with transforma
tion and tumor expression (4—9).
Lung cancer is the most common cause of cancer-related deaths in
the United States (10). The study of the genetic and molecular events
associated with the transformation and progression of normal HBE3
cells to the malignant phenotype in human systems has been limited
by the difficulty in obtaining preneoplastic tissue. However, we have
established a unique in vitro and in vivo model system of HBE cell
progression, which allows us to study the events associated with lung
cancer tumorigenesis (I l—l3).@In this report, we describe the spon
taneous acquisition of the tumorigenic phenotype in an immortalized
HBE cell line and loss of this phenotype when cultured in vitro but not
vitro
and
remain
tumorigenic.
However,
late-passage
NL2OT
cells
consis
derived
cell lines (NL2OT-A
and NL2OT-B)
have been passaged
85 times in
tently lose their tumongenicity when passaged in vitro on tissue culture
plastic dishes (NL2OT-n cells). Thus, two of the cell lines, NL2O and
NL2OT,
reverted
spontaneously
whereas
to the
following
cells passaged
nontumongenic
serial
phenotype
passage
reproducibly
on plastic
in mice (NL2OT-A
tissue
and
culture
plates,
and -B) did not.
We used these nontumorigenic (NL2Oand NL2OT-n)and tumorigenic
(early passage NL2OT,NL2OT-A,and NL2OT-B)cells to study the role of
the aS/fJl-integnn and attachment to fibronectin in tumorigenicity. The
two nontumongemc cell lines (NL2O and NL2OT-n)attached slower to
fibronectin-coated
plates than the two tumorigenic
cell lines in a cellular
extracellular matrix adhesion assay. Attachment was abrogated by expo
sure to a blocking antibody to the a5/@31-integrin,the fibronectin receptor,
in the two tumorigenic cell lines. Cell surface expression of the aS/fM cell
surface protein by flow cytometry
was highest in the tumorigenic
NL2OT
and NL2OT-Acells.
NL2OT-Acells were cultured with an antibody to aS/fM and inoculated
s.c. into athymic nude mice; tumongenicity of the NL2OT-Acells was
inhibited
in a dose-dependent
manner.
Tumorigenicity
was also inhibited
partially with monoclonal antibodies to either aS or flu.
A mixture of 10% tumorigenic NL2OT-Acells and 90% nontumori
genic NL2Ocells was cultured on plastic, type IV collagen, lambda, and
fibronectin for 9 weeks. Only cells cultured on fibronectin formed tumors
when inoculated s.c. into athymic nude mice.
We conclude that these data are consistent with the hypothesis that
neoplastic
transformation
in our
original
cell
line
arose
from
in vivo
selection of a small mutant clone, which had arisen in culture and was
selected subsequently
in vivo but was lost with in vitro culture
in NL2O
cells, and that a5/fll-integrin interaction with the extracellular matrix
may play a role in tumorigenicity in our system.
in vivo.
We
INTRODUCTION
One of the classic definitions of cancer in model systems is the
ability of cells to form tumors in animals. The steps involved in this
process are unknown but are postulated to involve multiple host
tumor interactions in which tumor cells adhere to one another to form
a primary tumor and then desegregate and penetrate the basement
membrane of the host organ and the vascular endothelium (1). These
initial steps include: (a) loss of intracellular adhesion within the
MATERIALS
work
was supported
in part
by the Veterans
Administration,
NIH
Grant
whom
requests
importance
of the a5/(31-
for
Wisconsin Comprehensive
reprints
should
be
addressed,
at
in tumor
AND METHODS
and gentamicin
(50
@tg/m1;GIBCO).
The cells grew attached,
as
a monolayer, on plastic. They were cultured in T75 flasks at 37°Cin a
humidified5% CO2atmosphereand were passagedonce per week.
3 The
CA14520, and a grant from the Elsa U. Pardee Foundation.
2 To
the possible
Cell Culture. Cells were culturedin Ham's F12 medium(GIBCO,Grand
Island, NY) supplemented with 4% FBS (Sigma Chemical Co., St. Louis,
MO), insulin(250 milliunits/ml;Sigma),transferrin(5 @.tg/ml;
Sigma),epider
mal growth factor (10 ng/ml; Sigma) dextrose (15 mM;Sigma), hydrocortisone
(1 ng/ml; Sigma), a mixture of nonessential amino acids (each amino acid, 0.1
mM; GIBCO)
Received 6/I 3/95: accepted I0/16/95.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
I This
also describe
integrin cell surface protein interaction with fibronectin
development in our system.
abbreviations
used
are:
HBE,
human
bronchial
epithelial;
FBS,
fetal
bovine
serum; MR. mean fluorescence intensity; SCID, severe combined immunodeficient.
K4/666
CSC.
University
4 i. Schiller,
of
G. Bittner,
K. Jankowski,
S. Wu,
and L. Meisner.
Spontaneous
acquisition
ofthe neoplasticphenotypeinanimmortalized
bronchialepithelialcellline,submittedfor
Cancer Center, 600 Highland Avenue, Madison, WI 53792.
Phone: (608) 263-8600: Fax: (608) 263-8613.
publication, 1995.
6215
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGENICITY AND ThE a5/@3IRECEPTOR
took 6 weeks to reach 1 cm in diameter; at passage 10, they took 12
weeks, and at passages 20 and 25, no tumors formed.
Inc. (Temecula, CA; catalog numbers MAB 1969,MAB 1956,and MAB 1965,
To confirm that this model system was reproducible, we thawed
respectively).
The ct5/f31-integrin
antibody is an IgG2A murine monoclonal
cryopreserved, early passage NL2OT cells and inoculated them into
antibody (clone JBS5; Ref. 14) produced against the intact a5/@31-integrin,
nude mice at serial passages (Fig. 2). Similar to our first observations,
which has been adsorbed with both aS- and (31-integrins individually to
NL2OT
cells formed s.c. tumors at passages 7, 10, and 14. At passage
remove any cross-reactivity, and the resultant product is thus specific for
20,
very
small tumors, which later regressed, formed in two of four
a5/f3l-integrin.5 The antilaminin antibody is an IgG2A murine monoclonal
mice. At passage 23, no tumors formed in three of three mice.
antibody (GIBCO).
Tumorigenicity. Immunodeficient 3—4-week-oldfemale BALB/c nude Therefore, this loss of tumorigenicity is reproducible and occurs after
mice were inoculated
s.c. with 5 X 106 cells and observed
1 year for tumor
20—25passages on plastic with NL2OT cells but not with the second
development. Tumorigenic HCT I 16 human colorectal carcinoma cells and ary NL2OT-A cells.
A549 human lung cancer cells were also inoculated into mice as positive
Expression of a5/fll-Integrin.
We evaluated the expression of
controls.
Antibodies.
murine
The anti-a5/f31-integrin
monoclonal
Attachment
antibodies,
Assay.
antibody,
were obtained
Fibronectin-coated,
anti-aS,
and anti-(3l, all
from Chemicon
six-well
International,
trays (Collaborative
integrins
Bio
medical Products) were incubated at 37°Cfor 2 h with PBS plus 5% BSA to
block nonspecific protein-binding sites. Cells were dissociated nonenzymati
cally and counted,
and 106 cells/well
were plated
in 2 ml Ham's
Fl2 medium
plus 4% FBS at time 0 and incubated at 37°C.After the specified times, plates
were aspirated, rinsed two times with PBS to remove nonadherent cells,
dissociated nonenzymatically as described below, and counted on a hemacy
tometer. Results are presented as the percentage of the initial number plated
that attached. Each experiment was repeated at least twice; results between
experiments varied by less than 5%.
Flow Cytometry. Cells were dissociated using a nonenzymatic method to
avoid degradation of cell surface antigens. Cells were incubated for 12 minutes
in Ca2@- and Mg@ k-free
HBSS
with 0.2%
EDTA
and 8% dialyzed
FBS at
37°C.Cells were stained with the primary antibody (5 @g)
for 30 mm on ice
in PBS,
1% BSA, and 0.1% sodium
azide,
30 mm on ice with an FITC-conjugated,
(2.5
@g)to the primary
antibody,
washed
washed
two times, and stained
goat antimouse
secondary
two times, and evaluated
for
antibody
on a Becton
a5/g3l,
as, and f31 on the nontumorigenic
tant NL2OT (NL2OT-n)
NL2O and rever
cell lines and the tumorigenic
early passage
NL2OT and NL2OT-A cell lines using flow cytometry. Table 1 shows
the fold increase in MFI of antibody-stained cells over cells stained
with isotype-matched, negative control antibodies as measured by
flow cytometry. The tumorigenic, early passage NL2OT and
NL2OT-A cell lines had more cell surface expression of these proteins
than the nontumongenic NL2OT-n cells and greater or equal amounts
as the parent nontumorigenic NL2O cell line. For example, the MR of
@l-integrinon the NL2OT-n cells was about 48-fold brighter than the
negative control, whereas on the NL2OT-A cells, the MR of @31integrin was 100-fold brighter than the negative control antibody. This
indicates that there is twice as much expression of f31-integrin on the
tumorigenic NL2OT-A cells as there is on the nontumorigenic
NL2OT-n cells, because 100 is approximately twice as bright as 48.
Attachment to Fibronectin and Reversal by Blocking Antibod
Dickinson FACScan flow cytometer. Nonspecific isotype control antibodies
len to a5431 in the Tumorigemc Cell Lines. We characterizedthe
(5 ,.@g) were
ability of the different cell lines to attach to fibronectin and found a
marked difference between the cell lines (Fig. 3). The two nontumori
genic cell lines, NL2O and NL2OT-n, attached the slowest. Ninety
percent of the parent NL2O cell line and NL2OT-n were attached by 90
and 60 mm, respectively; whereas 90% of both the tumorigenic
NL2OT and NL2OT-A cell lines were attached by 30 mm.
These experiments were repeated using a blocking antibody to
used
as negative
controls.
Propidium
iodide
(I
p@g/ml) was
included to gate out dead cells.
RESULTS
We have reported previously the establishment of six HBE cell
strains and one immortalized HBE cell line (NL2O) by transfection
with either the SV4O virus or an origin of a replication-defective large
T plasmid (1 1, 12). NL2O cells were nontumongenic at passages 11,
20, 32, and 56 and had been passaged for more than 3 years when one
of four mice inoculated with passage 184 NL2O cells formed a s.c.
tumor approximately 12 weeks after inoculation (1 1). After an addi
tional 12 weeks of growth, the tumor was dissociated and placed into
culture. The resulting cell line (NL2OT) has been passaged 85 times
and cultured in vitro for more than 18 months (1 1, 13). Passage 3
NL2OT cells formed tumors in eight of eight athymic nude mice,
whereas the parent NL2O cell line, which has continued to be passaged
on plastic tissue culture plates, has not formed tumors in mice inoc
ulated with passage 189, 200, and 205 cells (Fig. 1). Thus, the NL2O
cells formed tumors in passage 184 cells only.
Establishment
of Tumorigemc NL2OT-A and NL2OT-B Cell
Lines. To mimic the in vivo development of human lung cancer, we
passaged NL2OT cells in mice. NL2OT cells, derived from the disso
ciation of the original tumor, were cultured for 21 days and then
reinoculated s.c. into four mice. All four mice developed s.c. tumors;
two of the four tumors were dissociated and placed into culture. The
two subsequent cell lines (NL2OT-A and NL2OT-B) have been pas
saged 75 times in vitro and remain tumorigenic at passages 15, 3 1, and
68 (Fig. 1).
Loss of Tumorigenicity of NL2OT CelLs. NL2OT cells cultured on
the surface of tissue culture plastic dishes were reinoculated into mice
at passages 3, 10, 20, and 25. At passage 3, it was noted that the cells
5Chemicon,
personal
communication.
a5/@3l
(Fig.
4). The
a5/j31
antibody
blocked
attachment
of the
two
tumorigenic cell lines to levels equivalent to or less than those of the
two nontumorigenic cell lines. For example, whereas 80—90%of the
tumorigenic NL2OT and NL2OT-A cell lines had attached to fibronec
tin by 30 mm, in the presence of the antibody, it took 90 mm or longer
before 90% of the NL2OT and NL2OT-A cells attached (Fig. 4, C and
D).
Of
mote,
the
cx51f31 antibody
did
not
affect
attachment
of the
parental nontumorigenic NL2O cell line appreciably and affected it
only minimally in the derivative montumorigenic NL2OT-n (Fig. 4, A
and B). We conclude that the residual attachment is due to other
imtegrins, which also bind to fibronectim as well as other extracellular
matrix proteins.
Reversal of the Tumongenic Phenotype in NL2OT-A Cells by
Blocking Antibodies to aS/fl!. We incubated NL2OT-A cells in
vitro for 30 mm on ice with monoclonal
antibodies
to human
integrin
a5/@l or with nonspecific mouse immunogbobulins as controls. Cells
were then inoculated s.c. into athymic nude mice.
Six of six mice inoculated with NL2OT-A cells exposed to nonspe
cific antibodies formed tumors compared with one of four mice
inoculated with cells exposed to the a543l-integrin monoclonal anti
body. Antibodies to a5/f3l blocked tumorigenicity in a dose
dependent manner (Fig. 5).
Mice were inoculated with NL2OT-A cells exposed to momoclonal
antibodies to either aS or f3l (Fig. 6). The anti-aS antibody inhibited
tumor growth in one mouse completely and decreased growth in the
remaining three mice; the anti-/35 antibody inhibited tumor growth in
6216
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGENICITY AND THE a5/@1 RECEPTOR
Flow Chart
of HBE Cell Line System
inoculate
No tunsor
growth
clone 1(80%)
clone 2(20%)
(P189)
Passage
1 inoculate
$ (P184)
I inoculate
, (P189,
205)
clone Li
No tumorgrowth
inoculate
.
Gone 2
growth
Clone Li
(iOO%)/(P3)
one Li
@-A (iOO%)/(P3)
(100%)/(P3)
I inoculate
I inoculate
I inoculate
t (P15,
31)
t (‘@‘@,
31)
* (P20,
25)
*
*
No tumor growth
Tumor growth
Tumor growth
Fig. 1. Schematic diagram outlining the establishment of cell lines.
three of four mice completely and inhibited tumor growth nearly
completely in the fourth.
To rule out possible immunological effects in blocking tumor
formation
of anti-a5/(3l
-coated
cells,
tumorigenic
experiments
were
repeated in athymic nude mice and SCID mice using NL2OT-A cells
coated with antihuman laminin. These two isotype-matched antibod
ies coat the NL2OT-A cells to very similar levels as measured by flow
cytometry. The antihuman lamimin did not inhibit NL2OT-A tumor
formation in either the nude or SCID mice, whereas the a5/g3l
antibody did inhibit tumor formation. These data indicate that immu
nological factors such as complement-mediated cytotoxicity do not
play a role in tumor suppression by anti-a5/fll-integrin
antibody.
Tumorigenic NL2OT-A cells were plated in triplicate on plastic and
fibronectin-coated, six-well trays. Following attachment 24 h later, the
anti-a5/f3l-integrin
antibody
was
added
at a concentration
of 60
@g/l06cells, 30 times the concentration that inhibited tumor forma
tion. Following 3 days of antibody exposure, the cells were dissoci
ated and counted. There was no difference between antibody-treated
and untreated cells on either plastic or fibronectin, indicating no direct
cytotoxicity of the antibody.
6217
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGENICITY AND THE a5I@3lRECEPTOR
1
j31, the fibronectin receptor, implicating an adhesion protein-extracel
lular matrix interaction in tumor formation in our system.
We reported previously the specific genetic alterations identified
before and after neoplastic transformation.6 Of these, the most notable
alterations that were observed in NL2OT cells and not in the parent
NL2O cells or in the reverting NL2OT-n cells were amplification of
9q2l.2—@sq34,
the site of the c-abl oncogene, and loss of one chromo
some 18, the site of the tumor suppressor DCC gene. Because the
genetic changes associated with tumorigenicity occurred in the ab
sence of any chemical or viral exposures, in one mouse only, and at
only one passage number (passage 184), we suspect that the tumor
arose from in vivo selection of a small mutant clone, which had arisen
in culture and was selected subsequently in vivo but was lost with in
vitro culture in NL2O cells. The fact that the tumongenic karyotype
was lost following only 20—25in vitro passages of the original tumor,
whereas it was stabilized after inoculation into two mice prior to
long-term passage, emphasizes the role of in vivo selective forces in
establishing the tumorigenic genotype. This is supported by experi
ments in which a mixture of 10% tumorigenic NL2OT-A and 90%
nontumorigenic NL2O cells formed tumors after 9 weeks in culture in
vitro on fibronectin but not on plastic, laminin, or type IV collagen.
These data suggest that the minor population of tumorigenic
NL2OT-A cells expanded to form tumors when cultured on fibronectin
and not on the other three substrates.
The a5/@l-integrin is a well-characterized receptor for fibronectin.
Interaction of the receptor with the ligand results in signal transduc
tion with multiple outcomes, including regulation of cell adhesion,
CV)
E
E
ti)
E
0
>
0
E
I—
0
5
15
10
Day
Fig. 2. Influence of culture conditions on tumor formation of NL2OT cells in athymic
@
nudemice.NL2OTtumorformationfollowing7 passagesin vitro( ), 23 passagesin
vitro (0), and 4 passages in nude mice (0). Five X lO@@
cells were injected s.c. into groups
of three athymic nude mice, and subsequent tumor volumes were followed. NL2OT cells
passaged in vitro for an extended period of time lost the ability to form s.c. tumors when
inoculated in nude mice. One of three representative experiments is shown.
Table I Fold increase in MFI
Increase in mean fluorescent intensity over isotype-matched
negative control anti
bodies.
migration,
matrix
assembly,
and cytoskeletal
organization
induction of collagenase and stromelysin gene expression
aS
lineTumorigenicityaS/@31-IntegrinIntegrinjll-IntegrinNL2OT-n—25.715.447.8NL2O—28.421.679.9NL2OT+34.023.780.6NL2OT-A+37.226.4100.9
Cell
f31 subunit
functions
in the
initial
attachment
of fibroblasts
(15) and
( 16). The
on
fi
bronectin but not in the later steps of cell spreading (15). The a5/f3l
receptor also may be regulated by other integrins, such as the avlb3vitronectin receptor (17).
In our study, we observed that the rate of attachment on fibronectin
correlated with tumorigenicity; antibodies to the a5/(3l-integrin
blocked binding to fibronectin in the two tumorigenic cell lines and
also blocked tumorigenicity. These data implicate a key role for the
Role of Fibronectin in Expanding the Tumorigenic Phenotype.
A mixture of 10% tumorigenic NL2OT-A cells and 90% nontumori
genic NL2O cells was cultured on plastic, collagen type IV, laminin,
and fibronectin, and 5 X 106 cells of the mixture were injected s.c.
into each of four nude mice. One of four mice developed s.c. tumors
after 8 weeks. Following 9 weeks of culture in vitro on each of the
four substrates, mice were injected s.c. again with S X 106 cells from
each substrate (Fig. 7). Four of four mice injected with cells cultured
on fibronectin developed tumors at 6 weeks, whereas no tumors
developed in mice injected with cells cultured on tissue culture plastic,
laminin, and type IV collagen.
C
tI)
E
.C
0
DISCUSSION
C
We have established and characterized a series of nontumorigenic
and tumorigenic HBE cell lines, all derived from an immortalized
HBE nontumorigenic cell line (NL2O) established by transfection with
an origin of the replication-defective SV4O large T plasmid ( 11, 13)•6
Two of the cell lines, NL2O and NL2OT, reverted to the nontumori
genic phenotype reproducibly and spontaneously following serial
passage on plastic tissue culture plates, whereas cells passaged in nude
mice (NL2OT-A and -B) do not. Attachment to fibronectin correlated
with tumongenicity, and tumorigenicity was abrogated by exposing
the NL2OT-A cells to a monoclonal antibody to human integrin-cs5/
6 J@ Schiller,
G.
Bittner,
K.
Jankowski,
S.
Wu,
and
L.
Meisner.
Spontaneous
e
a-
0
30
45
60
75
90
105
Minutes
Fig. 3. Cellular/extracellular adhesion assay. Cells were dissociated, and lIP cells were
allowed to attach onto fibronectin. At the specified times, cells were rinsed twice with
PBS to remove nonadherent cells, dissociated, and counted. One hundred percent attach
ment represents the 1O@cells that were plated. Nontumorigenic NL2O (0) and NL2OT-n
acquisition
of the neoplastic phenotype in an immortalized bronchial epithelial cell line, submitted for
publication, 1995.
15
(K@
) cells;tumorigenicNL2OT(0) andNL2OT-A(@)cells.One of two representative
experiments is shown.
6218
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGENICITY AND THE a5/@1 RECEPTOR
A.
Nontumorigenic
a
NL2O
C
C
E
U
E
.c
Nontumorigenic
NL2OT-n
U
.(
C
C
e
0
a.
a-
0
Fig. 4. Effects of monoclonal antibody to a5/@l
(I @.tg;0) or nonspecific immunoglobulin (LI) on
attachment to fibronectin in the nontumorigenic
NL2O (A ) or NL2OT-n (B) cells and the tumori
genic NL2OT (C) and NL2OT-A (D) cells. One of
two representative experiments is shown.
15
30
45
60
75
90
0
105
15
30
Minutes
C
t:i
Tumorigenic NL2OT
.c
U
.5
‘C
‘C
C
C
0
0
e
0
Tumorigenic
15
30
45
60
75
90
105
0
15
30
90
105
NL2OT-A
strongly.
These
data
conflict with those reported by others in Chinese hamster ovary cells
in which Chinese hamster ovary transformed cells or subclones over
expressing a5/@l were less tumorigenic than their counterparts cx
pressing low levels of a543l (18, 19). In human colon cancer, de
creased expression of integrins, including the f3l chain, was
associated with transformation from normal colonic epithelium to
benign adenomas to malignant neoplasms (20). In a model of multi
stage carcinogenesis of mouse skin, ceSIg3lwas down-regulated in the
progression from benign to malignant skin tumors (21). However,
none of these reports looked at the functional ability of the a5/(3l-
45
60
75
Minutes
Minutes
system
105
S
0
in our
90
a.
a.
in tumorigenicity
75
E
E
protein
60
C
0
C
0
a5/13l
45
Minutes
presence of the a5/@31antibody may be mediated by other receptors,
such as the a3/fll- or a4/(31-integrins, which also bind to fibronectin.
In our system, tumor cells cultured in vitro lose their tumorigemc
ity,
whereas
those
cultured
in
vivo
do
not.
Tumorigenicity
was
also
blocked by antibodies to the a5/fll-fibronectin receptor. These results
are consistent with the hypothesis that the interaction of the extracel
lular matrix with neoplastic cells expressing specific adhesion pro
teins results in a selection or growth advantage for these cells, with
subsequent initiation of the invasion and metastasis cascade. How this
interaction results in a selection or expansion of tumorigenic cells is
unclear. The interaction between the underlying extracellular matrix
integrin, merely its level of expression.
and the preneoplastic or neoplastic cells may result in an expansion of
In our system, cell surface expression of the a51f31 receptor also did
a subclone of tumorigemc cells expressing a specific attachment
not correlate with tumorigenicity, whereas attachment to fibronectin
protein phenotype; alternatively, the underlying extracellular matrix
did; thus, these data suggest that cell surface expression of this protein
or associated growth factors produced by surrounding stromal cells
may not correlate with tumorigenicity. Furthermore, attachment to
such as transforming growth factor-fl may regulate the expression of
fibronectin was not blocked, or only minimally so, by the a5/(3l
the adhesion protein profile of these cells.
antibody in the two nontumorigenic cell lines, whereas attachment
Our results also do not exclude the potential role of other integrins
was blocked in the two tumorigenic cell lines. These data suggest that
in mediating tumorigenicity in our system. Additional studies will
expression of a5/@l on the surface of the nontumorigenic cells was
need to be performed to determine whether tumorigenicity is related
either of a nonfunctional protein (NL2O cells) or was too low to
specifically to the a5/(31-integrin, or whether an interaction with other
contribute significantly to attachment to the fibronectin-coated plates
integrins or components of the extracellular matrix is also important.
(NL2OT-n cells). The slow attachment of the two nontumorigenic cell
These data also have implications for studies of in vitro culture
lines to fibronectin and of the two tumorigenic cell lines in the
systems as models of carcinogenesis and tumor biology. Much of
6219
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGENICITY AND THE a5/@1 RECEPTOR
traditional cancer research has involved the study of neoplastic or
preneoplastic cells growing in vitro on plastic tissue culture plates,
because these systems are generally less costly, more reproducible,
and easier to manipulate than studies of cancer cells growing in mice
or humans. Our data suggest that these in vitro systems may not
always reflect the growth and development of cancer in vivo accu
rately and, indeed, may give false or misleading results. The use of in
vitro
culture
systems
for studies
of tumorigenesis
and invasion
in
particular should be approached with caution.
Last, in this study, we describe the spontaneous transformation of
an immortalized nontumorigenic human bronchial epithelial cell line
to the tumorigenic phenotype. Although transformation of immortal
Cs)
E
E
a,
E
0
>
0
E
I-
0
50
E
Day
E
0
>
Fig. 7. Tumor formation in nude mice. A mixture of 90% nonwmorigenic NL2O cells plus
10% tumorigenic NL2OT-A cells was cultured on plastic, collagen IV, laminin, and fibronectin
for 9 weeks. Groups of four mice were then injected with cells from the various culture
conditions. Each mouse received S X l0'@cells. Tumor volume is plotted versus time for cells
0
cultured on plastic (L@),
collagen IV ( ). laminin (El).and fibronectin(0).
E
0
@
150
100
E
I—
0
10
20
30
ized human cell lines to the tumorigenic phenotype by carcinogens or
oncogenes is not uncommon (22—27),spontaneous transformation is
rare. The parent NL2O cell line and tumorigenic NL2OT cell line add
to the small list of immortalized human cell lines that have undergone
spontaneous malignant transformation and thus are important models
of tumorigenesis (28—30).
We have developed a unique, well-defined model system of HBE
40
Day
cell progression.
Thesecellsareuniquein theirgeneticstabilityand
Fig. 5. Dose-response effects of antibody to a5/@l in blocking tumorigenicity in the
NL2OT-A cell line. Cells were incubated in vitro with increasing concentrations of
monoclonal antibody for 30 mm on ice, and then 5 X lO'@cells were inoculated into the
mice. Control cells, 0; a5/@3l monoclonal antibody, 0.2 @sg,0 ; I @.sg,
0; 2 @.sg,
[email protected]
of two representative experiments is shown.
homogeneity,
making
them an excellent
cell progression.
Furthermore,
we have
.
.
.
velop
a senes
of cell lines, all denved
which differ reproducibly
in their ability
system
extended
for studies
of HBE
our studies
to de
from the same parent
to form tumors
in nude
line,
mice
based on whether they have been cultured in vitro or in vivo. Blocking
antibodies to the fibronectin receptor abrogates both attachment to
fibronectin in the derived cell lines and tumorigenicity. An isotype
matched control antibody did not prevent tumor formation in either
athymic nude or SCID mice, indicating that the antitumor effect of the
a5/fll antibody is unlikely to be due to immunological mechanisms.
These data are consistent with the hypothesis that a subclone of
(f)
E
E
tumorigenic
cells
that had arisen
in vitro
was selected
for in vivo
culture and suggest that the aS/fll-integrin protein-fibronectin
action may play a role in tumorigenicity in our system.
a,
E
inter
0
>
ACKNOWLEDGMENTS
0
E
We thank Drs. Lorraine Meisner and Shi-Qi Wu for their helpful advice and
critical comments regarding this manuscript. We also thank Kathy Edge for her
help in the preparation of the manuscript.
I-
REFERENCES
0
10
20
30
40
50
1. Liona,L.,andKohn,E.Grandroundsattheclinicalcenterof theNationalInstitutes
ofHealth.JAMA,263:
1123-1126, 1990.
2. Albelda, S., and Buck, C. Integrins and other cell adhesion molecules. FASEB (Fed.
Day
Am.Soc.Exp.Biol.)i. 4: 2868—2880.
1990.
Fig. 6. Effects of blocking antibodies to aS or fll in NL2OT-A cells. Mice were
inoculated with 5 x l0@NL2OT-A cells incubated with 1 p.g anti-aS ( 0 ), I p@ganti-f3l
(0), 1 @.tg
anti-aS/@3l (L:@),or nonspecific immunoglobulins (0).
3. Albelda, S. Biology of disease: role of integrins and other cell adhesion molecules in
tumor progression and metastasis. Lab. Invest.. 68: 4—17,1993.
4. Juliano, R., and Varner, J. Adhesion molecules in cancer: the role of integrins. Curr.
Opin. Cell Biol.. 5: 812—818. 1993.
6220
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
LOSS OF TUMORIGEN1CITYAND ThE a5/@I RECEPTOR
5. Vamer, J., Fisher. M.. and Juliano, R. Ectopic expression of integrin a3b1 suppresses
in vitro growth and tumorigenicity of human colon carcinoma cells. Mol. Biol. Cell,
3: 232a, 1992.
6. Plantefaber, L.. and Hynes. R. Changes in integrin receptors on oncogenically
transformed cells. Cell. 56: 281—290,1989.
7. Stallmach, A., Lampe, B. V., Matthes, H.. Bornhoft, G., and Riecken, E. Diminished
expression of integrin adhesion molecules on human colonic epithelial cells during
the benign to malignant tumour transformation. Gut, 33: 342—346,1992.
8. Giancotti, F., and Ruoslahti,E. Elevatedlevels of the a3b, fibronectin receptor
suppress the transformed phenotype of Chinese hamster ovary cells. Cell. 60: 849—
859,1990.
9. Schreiner, C., Fisher, M., Hussein, S., and Juliano, R. Increased tumorigenicity of
fibronectin receptor deficient Chinese hamster ovary cell variants. Cancer Res., 51:
1738—1740,
1991.
10. Boring. C., Squires. T.. Tong. T.. and Montgomery. S. Cancer statistics, 1994. CA
Cancer J. Clin.. 44: 7—26,1994.
II. Schiller, J. H.. Bittner. G., Oberley. T. D., Kao, C., Harris, C., and Meisner, L. F.
Establishment and characterization of an SV4O T-antigen immortalized human bron
chial epithelial cell line. In Vitro Cell Dcv. BioI., 28A: 461—464, 1992.
12. Schiller, J. H.. Sabatini, L., Bittner, G., Pinkerman, C., Mayotte, J., Meisner, L., and
Levitt, M. Phenotypic, molecular, and genetic characterization of transformed human
bronchial epithelial cell strains. Int. J. Oncol., 4: 461—470. 1994.
13. Schiller, J.. Bittner, G.. Jankowski, K., and Meisner, L. Neoplastic transformation of
an immortalized human bronchial epithelial cell line. Proc. Am. Assoc. Cancer Res.,
35: 163, 1994.
14. Caixia,
S., Stewart,
S., Wayner.
E., Carter,
W., and Wilkins,
J. Antibodies
19. Schreiner, C., Fisher, M., Hussein, S., and Juliano, R. Increased tumorigenicity of
fibronectin receptor deficient Chinese hamster ovary cell variants. Cancer Res., 51:
1738—1740,
1991.
20. Stallmach, A., Lampe, B. V., Matthes, H., Bornhoft, G., and Riecken, E. Diminished
expression of integrin adhesion molecules on human colonic epithelial cells during
the benign to malignant tumour transformation. Gut, 33: 342—346,1992.
21. Tennenbaum, T., Yuspa, S., Grover, A., Castronovo, B., Sobel, M., Yamada, Y., and
de Luca. L. Extracellular matrix receptors and mouse skin carcinogenesis: altered
expression linked to appearance of early markers of tumor progression. Cancer Res.,
52: 2966—2976, 1992.
22. Reznikoff, C., Loretz, L., Christian, B., Wu, S., and Meisner, L. Neoplastic transfor
mation of SV4O-immortalized human urinary tract epithelial cells by in vitro exposure
to 3-methylcholanthrene. Carcinogenesis (Lond.), 9: 1427—1436,1988.
23. Bookland, E. A., Swaminathan, S., Oyasu, R., Gilchrist, K. W., Lindstrom, M., and
Reznikoff, C. A. Tumorigenic transformation and neoplastic progression of human
uroepithelial cells after exposure in vitro to 4-aminobiphenyl or its metabolites.
Cancer Res., 52: 1606—1614, 1992.
24. Rhim, J., Fujita, J., Arnstein, P., and Aaronson, S. Neoplastic conversion of human
keratinocytes by adenovirus 12-SV4O virus and chemical carcinogens. Science
to different
(Wash. DC), 232: 385—388,1986.
25. Pfeifer, A. M., Mark, III, 0. E., Malan, S. L., Graziano, S., Amstad, P., and Harris,
C. C. Cooperationof c-raf-1and c-mycprotooncogenesin the neoplastictransfor
26.
members of the @l(CD29) integrins induce homotypic and heterotypic cellular
aggregation.
Cell.lmmunol..138:216—228,
1991.
IS. Akiyama. S., Yamada. S. S.. Chen, W. T., and Yamada. K. M. Analysis of fibronectin
27.
receptor function with monoclonal antibodies: roles in cell adhesion, migration,
matrix assembly and cytoskeletal organization. J. Cell Biol., 109: 863—875, 1989.
16. Werb, Z., Tremble, P., Behrendtson, 0., Crowley, E.. and Damsky, C. Signal
28.
transduction through the fibronectin receptor induces collagenase and stromelysin
gene expression. J. Cell Biol.. 109: 877—889, 1989.
17. Blystone. S.. Graham. I. L.. Lindberg. F. P., and Brown, E. J. Integrin av @33
29.
differentially regulates adhesive and phagocytic functions of the fibronectin receptor
aS @3l.J. Cell Biol.. 127: 1129—1137,1994.
18. Giancotti. F.. and Ruoslahti, E. Elevated levels of the a@b1tibronectin receptor
suppress the transformed phenotype of Chinese hamster ovary cells. Cell, 60: 849—
859. 1990.
30.
mation of simian virus 40 large tumor antigen-immortalized human bronchial epi
thelial cells. Proc. Natl. Acad. Sci. USA, 86: 10075—10079, 1989.
Klein-Szanto, A., lizasa, T., Momiki, S., Garcia-Palazzo, J., Caamano, J., Metcalf, R.,
Welsh, J., and Harris, C. A tobacco-specific N-nitrosamine or cigarette smoke
condensate causes neoplastic transformation of xenotransplanted human bronchial
epithelial cells. Proc. Natl. Acad. Sci. USA, 89: 6693—6697, 1992.
Milo, G., Shuler, C., Stoner, G., and Chen, J. Conversion of premalignant human cells
to tumorigenic cells by methylmethane sulfonate and methylnitronitrosoguanidine.
Cell Biol. Toxicol., 8: 193—205,1992.
Reddel, R., Salghetti, S., Willey, J., Ohnuki, Y., Ke, Y., Gerwin, B., Lechner, J., and
Harris, C. Development of tumorigenicity in simian virus 40-immortalized human
bronchial epithelial cell lines. Cancer Res., 53: 985—991,1993.
Koprowski. H., and Croce, C. Tumorigenicity of Simian virus 40-transformed human
cells and mouse-human hybrids in nude mice. Proc. Natl. Acad. Sci. USA, 74:
1142—1146,
1977.
Brown, K., and Gallimore, P. Malignant progression of an SV4O-transformed human
epidermic keratinocyte cell line. Br. J. Cancer, 56: 545—554,1987.
6221
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.
Loss of the Tumorigenic Phenotype with in Vitro, but not in Vivo
, Passaging of a Novel Series of Human Bronchial Epithelial Cell
Lines: Possible Role of an α5/β1-Integrin-Fibronectin
Interaction
Joan H. Schiller and Gerard Bittner
Cancer Res 1995;55:6215-6221.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/55/24/6215
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 12, 2017. © 1995 American Association for Cancer Research.