(Received for publication, November 12, 1996, and in revised form, April 10, 1997)
From the § Departments of Medicine and Cell Biology, Vanderbilt University and Veterans Affairs Medical Center, Nashville, Tennessee 37232 and the ¶ Department of Pharmacology and Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina 27599
We recently have shown that activated Ras, but
not Raf, causes transformation of intestinal (RIE-1, IEC-6) epithelial
cells, whereas both activated Ras and Raf transform NIH 3T3 fibroblasts (Oldham, S. M., Clark, G. J., Gangarosa, L. M., Coffey, R. J., and Der,
C. J. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 6924-6928). The observations that conditioned medium from Ras-, but
not Raf-, transfected RIE-1 cells, as well as exogenous transforming
growth factor (TGF
), promoted morphological transformation of
parental RIE-1 cells prompted us to identify epidermal growth factor
(EGF) receptor (EGFR) ligands produced by Ras-transformed RIE-1 cells responsible for this autocrine effect. Since studies in fibroblasts have shown that v-Src is transforming, we also determined if v-Src could transform RIE-1 cells. H- or K-Ras-transformed cells secreted significant amounts of TGF
protein, and mRNA transcripts for TGF
, amphiregulin (AR), and heparin-binding EGF-like growth factor (HB-EGF) were induced. Like Ras, v-Src caused morphological and growth
transformation of parental RIE-1 cells. However, TGF
protein was not
secreted by RIE-1 cells stably expressing v-Src or activated Raf, and
only minor increases in EGFR ligand mRNA expression were detected
in these cells. A selective EGFR tyrosine kinase inhibitor PD153035
attenuated the Ras-, but not Src-, transformed phenotype. Taken
together, these observations provide a mechanistic and biochemical basis for the ability of activated Ras, but not activated Raf, to cause
transformation of RIE-1 cells. Finally, we suggest that an
EGFR-dependent mechanism is necessary for Ras, but not Src, transformation of these intestinal epithelial cells.
A remarkable convergence of biological, biochemical and genetic evidence has established that Ras proteins mediate many of their actions via recruitment of Raf-1 to the cell surface, which allows this serine/threonine kinase then to become activated. Raf-1 causes activation of a mitogen-activated protein kinase (MAPK)1 kinase (MEK), which in turn phosphorylates p42 and p44 MAPKs. These two MAPKs then translocate to the nucleus where they activate the Elk-1 transcription factor. The ability of dominant negative mutants of Raf-1, MEK, and MAPKs to block oncogenic Ras transformation of rodent fibroblasts demonstrates the essential role of this kinase cascade in Ras function (1-4). However, recent studies also support the importance of Raf-independent effector pathways in mediating Ras transformation (4-6).
Although mutated ras genes are most frequently associated
with human tumors of epithelial origin (7, 8), most of our knowledge on
the signaling pathways that mediate oncogenic Ras function is based on
studies of rodent fibroblasts. Therefore, we have been interested in
examining transformation by oncogenes such as ras,
raf, and src in epithelial cell systems. We have observed that activated Ras, but not Raf, causes transformation of the
rat intestinal epithelial cell line RIE-1, as determined by altered
morphology, growth in soft agar, and rapid appearance of tumors in nude
mice (9). Thus, activation of the Raf/MEK/MAPK cascade alone was not
sufficient to cause transformation. Furthermore, our observation that
Ras-, but not Raf-, expressing cells secreted factors that promoted
RIE-1 morphological transformation suggested that Ras transformation of
these epithelial cells was mediated, at least in part, via an autocrine
mechanism. Previous studies in other systems have found increased
expression of transforming growth factor- (TGF
) associated with
Ras transformation (10-12).
In the present study, we have evaluated the role of EGF receptor (EGFR)
ligands in mediating Ras transformation of RIE-1 cells. We observed
that Ras-, but not Raf-, expressing cells exhibited increased
expression and secretion of TGF protein. In contrast, Src
transformation did not cause up-regulation of TGF
production. Furthermore, inhibition of EGFR function impaired the morphological and
growth characteristics of Ras-, but not Src-, transformed cells. These
observations distinguish the activities of Ras versus Raf
and suggest that Ras transformation is mediated by a Raf-independent, EGFR-dependent mechanism. Transformation by v-Src, however,
is not dependent on an EGFR autocrine mechanism.
Cell Lines and Reagents
RIE-1 cells were obtained from Dr. Kenneth Brown (Cambridge, UK) and are a diploid, nontransformed, EGF-responsive cell line derived from rat small intestine (13, 14). Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (Invitrogen). All assays were done before passage 30. The EGFR kinase inhibitor PD153035 was obtained from Parke-Davis (15).
Constructs and Transfections
The pZIP-K-ras(12V) and
pZIP-raf22W retrovirus expression vector
constructs, which encode transforming mutants of human K-Ras 4B and
Raf, respectively, have been described (9, 16). The pSV2-H-ras (12V) expression vector construct contains the
genomic human sequences encoding the transforming H-Ras(12V) protein
and was provided by Dr. Jorge Filmus (Sunnybrook Health/Science Ctr., Toronto, Canada) (17). The psrc construct encodes viral Src and was a gift from Dr. Mark Kamps (VCSD, San Diego, CA) (18). The
constructs pZIP-K-ras(12V),
pZIP-
raf22W, and psrc with
pZIP-NeoSV(x)1, as well as the pZIP-NeoSV(x)1 vector control, were each
transfected into the RIE-1 cells (1-3 µg of plasmid DNA/60-mm dish).
Transfections were done using 5 µl of LipofectAMINE (Life
Technologies, Inc.) for 16-20 h on cells seeded at 1-5 × 105/60-mm dish. Transfected cells were selected and
maintained in medium containing 400 µg/ml G418 (Life Technologies,
Inc.). Multiple G418-resistant colonies (>50) were pooled together for
further studies. The pSV2-H-ras (12V) as well as the pSV2neo
vector control were each transfected into the RIE-1 cells by calcium
phosphate precipitation as described previously (19). Transfected cells were selected in medium containing 500 µg/ml of G418 (Life
Technologies, Inc.) and subcloned by limiting dilution.
TGF Radioimmunoassay
Parental RIE-1 cells and each of the transfected RIE-1 lines
were grown to confluence in 24-well dishes, washed in isotonic buffer
twice, and switched to serum-free DMEM (1 ml/well) for 48 h, at
which time media and lysates were harvested. The cells were washed
twice with isotonic buffer and then lysed at room temperature in 25 mM Tris-HCl (pH 8.0), 50 mM NaCl, 0.5% sodium deoxycholate, and 0.5% Nonidet P-40 (1 ml/well) on a rocker for 1 h. The rat TGF antibody used for the RIA was developed in
collaboration with East Acres Biologicals (Southbridge, MA). The RIA
has been described previously in detail (20) and was used to measure TGF
in both the conditioned media and lysates of each transfected cell line. Representative wells were trypsinized, and the cells were
counted with a hemacytometer to normalize the data. These experiments
were performed on cell lines prior to passage 11.
Isolation of Poly(A)+ RNA and Northern Blot Analysis
Cells were grown to near confluence, washed twice with isotonic
buffer, and then switched to serum-free DMEM for 72 h. Then total
cellular RNA was extracted by the method of Schwab et al. (21). Oligo(dT)-selected RNA was separated by electrophoresis in 1.2%
agarose/formaldehyde gels, and Northern blotting was performed as
described previously (22, 23). Hybridizations with species-specific probes labeled by RNA polymerase-directed reverse transcription (EGF,
TGF, amphiregulin (AR), betacellulin (BTC), and 1B15) or random
primer extension (heparin-binding EGF-like growth factor (HB-EGF)) were
performed in hybridization ovens as described previously (24, 25). 1B15
is a constitutively expressed sequence previously described (26).
PhosphorImager analysis (Molecular Dynamics) was performed to quantify
band intensities.
Growth Assays
Anchorage-dependentParental, H-Ras, K-Ras, and v-Src RIE-1 cells were plated in 6-well cluster plates at a density of 1 × 106 cells/well. Cells were counted 24 h later to determine plating efficiency. Cells were then treated with the indicated concentrations of PD153035 (10 nM, 25 nM, and 50 nM) or Me2SO (1 µl/ml). Cells were treated every other day (days 2 and 4), and triplicate wells were counted by hemacytometer on days 3 and 6.
Anchorage-independent1-20 × 103 cells/ml of each cell line (H-ras-, K-ras-, or v-src-transfected RIE-1 cells) were plated in 0.4% SeaPlaque agarose (FMC Corp. BioProducts) over a hardened layer of 0.8% agarose. Growth medium containing Me2SO alone or PD153035 (dissolved in Me2SO) was added prior to plating. After 7-10 days, colonies in triplicate (>50 µm) were counted with a colony counter (Bausch & Lomb). Paired t test was used to test for significant differences compared with control.
Morphology Experiments
Morphological reversion experiments were performed by adding 1 µM PD153035 or Me2SO in growth medium to
~50% confluent ras-and v-src-transfected RIE-1
cells growing in 6-well dishes. Morphological changes were monitored
over the next 72 h and photographs were taken. Conditioned media
experiments were performed by harvesting 48-h serum-free conditioned
medium from the Ras-transformed and control cell lines. The conditioned
media were then filtered through a 0.2 micron filter and added to 50%
confluent parental RIE-1 cells at a 3:1 ratio with fresh serum-free
medium alone or with 1 µM PD153035 or 3 µl/ml
anti-TGF antibody S-574. The cells were examined for morphological
changes over the next 48 h and photographed.
EGFR Analysis
Parental and H-Ras RIE-1 cells were plated in 24-well cluster
dishes at a density of 1 × 105 cells/well and allowed
to grow to 70-80% confluence. Cells were then treated for 24 h
with 1 µM PD153035 or Me2SO followed by a
5-min pulse with TGF (10 ng/ml) where indicated. Cells were then
lysed with buffer containing 20 mM Hepes, pH 7.4, 1% (w/v) Triton X-100, 2 mM EGTA, 2 mM EDTA, 500 µM Na3VO4, 50 µM
PMSF, 10 µg/ml aprotinin, and 10 µg/ml leupeptin. Samples of equal
protein concentrations were run on SDS-polyacrylamide gel
electrophoresis and Western blotted onto polyvinylidene difluoride
membranes. The Western blot was probed with anti-human EGFR sheep
polyclonal (Upstate Biotechnology) and detected with a donkey
anti-sheep horseradish peroxidase-conjugated IgG secondary antibody
(Jackson Laboratories) followed by ECL chemiluminescence (Amersham Life Science Inc.) and autoradiography. The phosphorylated tyrosine content
of the receptor was determined by subsequently probing the Western blot
with an horseradish peroxidase-conjugated anti-phosphotyrosine, clone
4G10 (Upstate Biotechnology), followed by ECL chemiluminescence detection (Amersham Life Science Inc.) and autoradiography.
We recently
have reported that constitutively activated Ras, but not Raf, caused
transformation of RIE-1 cells and that Ras-conditioned medium
morphologically transformed parental RIE-1 cells (9). Since studies in
fibroblasts have demonstrated that v-Src causes transformation, we
examined whether RIE-1 cells were also transformed by v-Src. A line of
RIE-1 cells was generated with a v-src construct. We
initially compared the morphological appearance of the parental RIE-1
cells (Fig. 1A) with that of those
transfected with pSV2-H-ras(12V) (Fig. 1D),
pZIP-K-ras(12V) (Fig. 1E), and psrc
(Fig. 1F). RIE-1 cells stably expressing oncogenic H-Ras,
K-Ras, and v-Src showed morphological alterations from a flat monolayer
of cuboidal cells to spindle-shaped cells with a poorly adherent,
highly refractile subpopulation. The H-Ras- and v-Src-expressing RIE-1
cells, like the recently described K-Ras-expressing RIE-1 cells (9),
also grew in soft agar and formed tumors rapidly in nude mice (data not
shown). The RIE-1 cells transfected with
pZIP-raf22W (Fig. 1C), the pSV2neo
vector (Fig. 1B), and the pZIP-Neo SV(x)1 vector (data not
shown) appeared identical in morphology to parental cells. Thus, in
contrast to activated Raf, v-Src caused a fully transformed phenotype
that was similar to that caused by activated Ras.
TGF
Studies then were directed toward
identification of factors that contribute to the morphological changes
induced by Ras-conditioned medium. Since administration of TGF to
parental RIE-1 cells led to morphological alterations similar to
Ras-conditioned medium (9), we initially examined TGF
protein
production by Ras-transformed cells and compared the results with
raf- and src-transfected cells. RIE-1 cells
transfected with activated H-ras and K-ras
secreted ~3-6 ng of TGF
/106 cells/48 h into
conditioned media, and ~500-1000 pg of TGF
/106 cells
were measured in cell lysates (Fig. 2). Neither the
parental RIE-1 cells nor the cells transfected with activated
raf, v-src, or control vectors had detectable
TGF
in their conditioned media (Fig. 2). Less than 100 pg of
TGF
/106 cells was detected in cell lysates of pZIPneo-,
raf-, and v-src-transfected RIE-1 cells. Thus,
production of TGF
clearly distinguishes Ras- from Raf-expressing
RIE-1 cells. Furthermore, v-Src-mediated transformation was not
associated with increased TGF
protein levels, and thus, production
of TGF
is not required for RIE-1 cellular transformation. Administration of a neutralizing TGF
antibody to Ras-conditioned medium was able to partially revert morphological transformation of
parental RIE-1 cells (Fig. 6C) although it was able to
completely block the effects of exogenous TGF
(data not shown),
suggesting that additional factors participate in Ras morphological
transformation.
Ras, Src, and Raf Exhibit Differential Expression of EGFR Ligands
We have reported previously that EGFR ligand expression
is coordinantly regulated in parental RIE-1 cells (27), and
up-regulation of several EGFR ligands has been observed in
keratinocytes infected with v-H-ras (12). Inasmuch as
reliable protein assays are not available for AR and HB-EGF, we
examined expression of mammalian EGFR ligands by Northern blot
analysis. Poly(A)+ RNA was isolated from near confluent
cultures of each cell line maintained serum-free for 72 h. Primary
data are shown in Fig. 3A and phosphoimager
quantification of band intensities are depicted in Fig. 3B.
Results are expressed as -fold increase over appropriate neo control,
after normalization to the constitutively expressed 1B15 mRNA of
each lane. The only transcript detected in parental RIE-1 cells and
RIE-1 cells transfected with vector controls was HB-EGF. RIE-1 cells
transfected with H-ras and K-ras constructs exhibited a greater than 10-fold induction of signals for AR and TGF
and a 2-3-fold increase in HB-EGF mRNA expression. In contrast to
Ras-transformed cells, raf- and v-src-transfected
cells had less than a 2-fold increase in AR and HB-EGF expression. A
TGF
transcript was observed in v-Src-transformed cells although no TGF
protein was detected as noted above. Transcripts for EGF and BTC
were not seen in any of the cell lines. Thus, selective overexpression
of TGF
, AR, and HB-EGF distinguishes Ras RIE-1 cells from both
nontransformed Raf and transformed v-Src RIE-1 cells.
EGFR Kinase Inhibition Selectively Alters the Morphology and Growth of Ras-transformed RIE-1 Cells
PD153035 has been demonstrated to
be a highly specific EGFR tyrosine kinase inhibitor (15). The ability
of this compound to block EGFR function in RIE-1 cells is shown in Fig.
4. Pretreatment with PD153035 (1 µM)
blocked TGF-induced EGFR tyrosine phosphorylation in serum-starved
parental RIE-1 cells (Fig. 4). This agent was then used to determine
the contribution of up-regulated EGFR ligands to the Ras-transformed
phenotype under four experimental conditions. First, PD153035 (1 µM) reduced basal EGFR tyrosine phosphorylation in
H-Ras-transformed RIE-1 cells (Fig. 4). Based on the previous data, we
suggest that this enhanced basal EGFR phosphorylation is likely due to
increased levels of endogenous EGF-like ligands activating the EGFR in
Ras-transformed cells. Second, addition of 1 µM PD153035
to RIE-1 cells expressing activated K-Ras caused morphological
reversion to a more normal appearance within 72 h (Fig.
5). Third, addition of the compound to conditioned
medium from K-Ras-transformed cells prevented morphological
transformation of parental RIE-1 cells caused by K-Ras-conditioned
medium alone (Fig. 6). Finally, PD153035 caused a
dose-dependent reduction in the ability of H- and
K-Ras-transformed RIE-1 cells to proliferate on plastic (Fig.
7). We observed, however, that PD153035 treatment of
Src-transformed cells failed to cause any significant morphological reversion (Fig. 5) and did not inhibit the monolayer growth of these
Src-transformed cells or parental RIE-1 cells (Fig. 7). This agent also
significantly reduced soft agar growth of H- and K-Ras-, but not
v-Src-, transformed RIE-1 cells (data not shown). Thus signaling
through the EGFR contributes significantly to Ras transformation of
RIE-1 cells, but Src transformation does not appear dependent on this
pathway.
These results provide a mechanism to explain why activated Ras,
but not activated Raf, transforms RIE-1 cells (9). Herein, we have
shown that there is up-regulation of EGFR ligand expression in
activated ras-, but not raf-, transfected RIE-1
cells and that autocrine growth factor signaling through the EGFR
contributes significantly to the Ras-transformed phenotype. Additional
recent support for the importance of EGFR signaling in mediating the Ras-transformed phenotype comes from studies using EGFR "knock-out" keratinocytes infected with H-ras retroviral constructs. As
compared with H-ras-infected, wild-type keratinocytes,
H-ras-infected EGFR /
cells form smaller tumors in a
primary engrafted papilloma model.2 The
failure of Raf-expressing cells to cause induction of EGFR ligand
expression suggests that Ras activates a Raf-independent pathway that
promotes EGFR ligand expression. Studies are underway to identify the
Ras signaling pathway that is responsible for EGFR ligand
overexpression.
These studies do not delineate the role of individual EGFR ligands in
contributing to the Ras-transformed phenotype. Filmus et al.
(28) studied H-ras transformation of an immature rat epithelial crypt cell line (IEC-18) and also found TGF mRNA
induction and increased protein production. Soft agar growth of the
H-ras-transfected IEC-18 cells was attenuated with both
TGF
neutralizing antibodies and anti-sense TGF
construct
transfection although transfection of IEC-18 cells with TGF
was not
sufficient to transform these cells. Expression of other EGFR ligands
was not examined in this study. The observation that EGFR blockade (via
pharmacological inactivation of its tyrosine kinase (Fig.
6D)), but not neutralization of TGF
(via antibody
neutralization (Fig. 6C)), is able to fully revert
morphological alterations induced by Ras-conditioned medium suggests
that secreted AR and HB-EGF contribute to this effect. It is possible
that TGF
along with AR and/or HB-EGF act in concert to mediate
morphological transformation of RIE-1 cells. Future studies will
address whether overexpression of AR or HB-EGF may be sufficient to
transform RIE-1 cells or whether a combination of EGFR ligands is
necessary.
v-Src transformation of RIE-1 cells appears to be mediated by a
mechanism distinct from that of Ras. In contrast to events related to
Ras transformation of RIE-1 cells, v-Src cells do not highly express
EGFR ligand transcripts nor produce TGF protein, and administration
of a specific EGFR tyrosine kinase inhibitor did not lead to
morphological reversion or a decrease in growth of these
Src-transformed cells. Collectively, these data indicate that Src
transformation of RIE-1 cells occurs independently of EGFR-mediated
events. Thus, whereas Src transformation of rodent fibroblasts is
dependent on Ras function, a similar requirement may not be involved in
Src transformation of RIE-1 cells. Ras-independent Src signaling events
have been observed recently (29, 30). Whether Ras activation is
required for Src transformation awaits future analyses.
These results in intestinal epithelial cells underscore the potentially
important distinct pathways and mechanisms underlying oncogenic
transformation of epithelial cells and fibroblasts. This is not
unanticipated in view of marked differences in the growth regulation of
these two cell types. For example, platelet-derived growth factor is a
mitogen for fibroblasts, but epithelial cells are nonresponsive due to
their lack of platelet-derived growth factor receptors. TGF
stimulates growth of fibroblasts, whereas it is a potent epithelial
cell growth inhibitor. Moreover, half-lives of farnesylated Ras
proteins markedly differ between fibroblasts and epithelial cells in
the presence of farnesyltransferase
inhibitors.3 Finally, activated
ras and raf constructs transform NIH 3T3
fibroblasts, whereas only the former is able to transform intestinal
and mammary epithelial cells (6). An important lesson from these
studies is that one cannot necessarily extrapolate results from growth regulation and oncogene transformation of fibroblasts to epithelial cell systems.
We thank Mark Kamps for the v-src expression vector and Dick Leopold for providing the PD153035 EGF receptor inhibitor.