(Received for publication, December 10, 1996, and in revised form, January 30, 1997)
From the Schepens Eye Research Institute, Harvard
Medical School, Boston, Massachusetts 02114, § University
of Colorado Health Sciences Center, Department of Pharmacology,
Denver, Colorado 80262, and ¶ Section of Virology and Oncology,
Division of Biology, Kansas State University,
Manhattan, Kansas 66506
Although it has been well established that
constitutive activation of receptor tyrosine kinases leads to cellular
transformation, the signal relay pathways involved have not been
systematically investigated. In this study we used a panel of
platelet-derived growth factor (PDGF) receptor mutants (
-PDGFR),
which selectively activate various signal relay enzymes to define which
signaling pathways are required for PDGF-dependent growth
of cells in soft agar. The host cell line for these studies was Ph
cells, a 3T3-like cell that expresses normal levels of the
-PDGFR
but no PDGF-
receptor (
-PDGFR). Hence, this cell system can be
used to study signaling of mutant
PDGFRs or
/
chimeras. We
constructed chimeric receptors containing the
PDGFR extracellular
domain and the
PDGFR cytoplasmic domain harboring various
phosphorylation site mutations. The mutants were expressed in Ph cells,
and their ability to drive PDGF-dependent cellular
transformation (growth in soft agar) was assayed. Cells infected with
an empty expression vector failed to grow in soft agar, whereas
introduction of the chimera with a wild-type
-PDGFR cytoplasmic
domain gave rise to a large number of colonies. In contrast, the
N2F5 chimera, in which the binding sites for
phospholipase C
(PLC-
), RasGTPase-activating protein,
phosphatidylinositol 3 kinase (PI3K), and SHP-2 were eliminated, failed
to trigger proliferation. Restoring the binding sites for
RasGTPase-activating protein or SHP-2 did not rescue the
PDGF-dependent response. In contrast, receptors capable of associating
with either PLC-
or PI3K relayed a growth signal that was comparable
to wild-type receptors in the soft agar growth assay. These findings
indicate that the PDGF receptor activates multiple signaling pathways
that lead to cellular transformation, and that either PI3K or PLC-
are key initiators of such signal relay cascades.
Several lines of evidence implicate PDGF1 and its
receptor (PDGFR) as key members in the genesis of
certain forms of cancer. First, the B chain of the PDGF ligand is
identical to the transforming protein of the v-sis oncogene
(1). Second, several studies have shown that co-expression of PDGF and
its receptor results in cellular transformation (2), whereas expression
of dominant negative constructs of the PDGF reagents can reverse the
transformed phenotype of naturally occurring tumor cell lines (3, 4). Third, a fusion between the -PDGFR and tel (an
ets-like transcription factor) is implicated in the
progression of chronic myelogenous leukemia patients to an acute
chromic myelomonocytic leukemic state (5). Collectively these findings
suggest that deregulation of PDGF-dependent pathways leads
to cellular transformation.
There are three isoforms of the PDGF ligand, PDGF AA, PDGF BB, and PDGF
AB, which differ in their transforming efficiencies (6, 7). PDGF AA
activates only the -PDGFR isoform, whereas the PDGF BB ligand
activates both
-PDGFRs and
-PDGFRs (8). The PDGF BB ligand is
functionally identical to v-sis (1) and can drive cellular
transformation in NIH3T3 cells. In contrast, PDGF AA is less potent at
mediating this response. The ability of the PDGF BB ligand to drive
cellular transformation of NIH3T3 cells more efficiently than the PDGF
AA ligand is believed to be due to either activation of the
-PDGFR
or due to the simultaneous activation of both types of receptors.
The majority of efforts to elucidate -PDGFR signal relay has focused
on the role of the receptor-associated proteins in mediating regulated
growth. In contrast, relatively little is known regarding the signal
transduction pathways important for driving abnormal, unregulated
growth akin to that which would be present in a cancerous or
transformed state. Careful studies comparing the
- and
-PDGFRs suggest that the
-PDGFR more efficiently transforms cells and that a
region in the tail of the
-PDGFR is critical for this effect (6, 9).
It has not yet been determined what signaling pathways are being
modulated by this region of the
-PDGFR to enhance the transformation
response.
To better understand the role of PI3K, RasGAP, SHP-2, and PLC- in
PDGF-dependent cellular transformation, we constructed a
panel of chimeric PDGFR mutants. Each construct contained the extracellular
-PDGFR domain and the intracellular
-PDGFR domain with a tyrosine to phenylalanine substitution at the tyrosine(s) required for binding the receptor-associated proteins. The chimeric constructs were stably expressed in fibroblast cells lacking
-PDGFRs and evaluated for their ability to promote PDGF-dependent
transformation by assaying growth in soft agar. The chimeric receptor
that binds all of the PDGFR-associated proteins, N2WT, was
capable of driving PDGF-dependent foci formation, whereas the chimera in which the binding sites for PI3K, RasGAP, SHP-2, and
PLC-
were mutated failed to promote growth in soft agar. These
observations suggested that activation of the kinase activity of the
receptor was not enough to drive cellular transformation, and that one
or more of the signaling enzymes recruited to the receptor are
required. Using the panel of
-PDGFR mutants, we found that
activation of either the PI3K or PLC-
signaling cascades restored
PDGF-dependent growth in soft agar. We conclude that similar signaling cascades are used to drive normal as well as deregulated growth of cells.
The Ph cell line is derived from mouse embryos
homozygous for the Ph/Ph deletion that includes the
-PDGFR gene (10) and was kindly provided by Dan Bowen-Pope
(University of Washington). These are 3T3-like cells that express
endogenous
-PDGFR at approximately 1 × 105
-PDGFRs/cell (11) and no
-PDGFR. Ph cells were maintained in
Dulbecco's modified Eagle's medium (DMEM) medium supplemented with
5% calf serum, and 1 mg/ml G418 was added to cultures of cells
expressing introduced chimeric constructs. The BALB/C
3T3/v-sis-transformed cells were maintained in DMEM
supplemented with 5% calf serum.
A SacII site was
introduced in the human -PDGFR at position 1972 using site-specific
mutagenesis according to the protocol supplied with the Amersham Corp.
oligonucleotide-directed in vitro mutagenesis system. The
mutagenic oligonucleotide had the following sequence:
5
-TAGTCCATCCCGCGGAAACTCCCA-3
. This unique SacII site was just upstream of the kinase domain and was introduced to facilitate the construction of the chimeras with the human
-PDGFR, which has a
naturally occurring SacII site at the analogous position (position 2145). The chimeras were constructed by replacing the entire
kinase, kinase insert, and tail regions of the
-PDGFR (from the
SacII site at position 1972 to the BamHI site at
position 3519 of the human
-PDGFR) with the corresponding portion of
the
-PDGFR (from the SacII site at position 2145 to the
XbaI site at position 4466 of the human
-PDGFR). In
addition to making the wild-type chimera, we also constructed a panel
of chimeras in which the intracellular
-PDGFR domain was one from
the series of mutants (F5, Y740/51, Y771, Y1009, or Y1021), the
construction of which has been previously described (12). The
full-length chimeric constructs were then subcloned into the
pLXSN2 retroviral vector, which is a modification of the
pLXSN vector (13), such that the polylinker contains the following
restriction sites: EcoRI, NotI, HpaI,
SalI, and BamHI. The DNA constructs were
introduced first into GP+E and then PA317 virus-producing cell lines,
or alternatively the DNA was transiently co-transfected with
SV-
-env
and SV-A-MLV-env constructs into
the 293T virus-producing cell line (14). The resulting virus was used
to infect Ph cells as described previously (12). The
receptor-expressing cells were then selected in DMEM, 5% calf serum,
and 1 mg/ml G418. The Ph cells expressing the chimeric receptors were
subjected to fluorescence-activated cell sorting analysis using an
antibody against the extracellular domain of the
-PDGFR (PR292)
followed by staining with an anti-mouse secondary antibody coupled with
fluorescein isothiocyanate fluorescent dye. The cells were sorted so
that the chimeric receptors were expressed at approximately the same
number as the endogenous receptors, 1 × 105
receptors/cell (10). Periodic assessment of the level of receptor expression by Western blot analysis indicated that the levels of
expression were stable for at least several months.
Subconfluent cultures of Ph cells expressing the various receptor constructs were washed gently with phosphate-buffered saline, trypsinized briefly, and resuspended in DME, 5% calf serum, and 0.45% low melting point agarose (SeaPlaque) supplemented with buffer (10 mM acetic acid and 2 mg/ml bovine serum albumin), 200 ng/ml PDGF AA, or 200 ng/ml PDGF BB. These cells were plated into 35-mm tissue culture plates containing a solidified bottom layer of DMEM, 5% calf serum, and 0.6% low melting point agarose. The soft agar cultures were then placed at 37 °C in 5% CO2 for 8-10 days, after which time the foci were counted. Foci larger than eight cells were scored as a colony, and the number of foci in a defined fraction of the dish were counted and used to calculate the number of foci per dish. Three areas of a dish were selected at random for counting the foci, and each condition was assayed in triplicate.
Metabolic Labeling of Cells with myo-[3H]InositolTo determine cellular levels of inositol metabolites, cells were plated in triplicate at a density of 6 × 105 cells/35-mm well or 1.2 × 106 cells/60-mm tissue culture dish. When nearly confluent, cultures were washed twice with inositol-free DMEM containing 5 µg/ml transferrin and then incubated in this medium for 4-6 h to deplete cellular stores of inositol. Cells were then metabolically labeled by addition of 10-40 µCi of myo-[2-3H]inositol (115 mCi/mmol, Amersham) for 18-24 h. When used for inositol turnover analysis, cultures were treated with LiCl (20 mM) for 30 min prior to addition of growth factor. Cultures were incubated in the presence or absence of human recombinant PDGF AA or BB (40 ng/ml) for 30 min to measure inositol turnover or for 3 min to quantitate D-3-phosphoinositide levels.
Extraction of Inositol Metabolites from CellsRadioactively
labeled cultures were rapidly rinsed with ice-cold phosphate-buffered
saline and then treated with 0.6 ml of perchloric acid (4.5%, v/v) at
4 °C. After 15 min, cells were scraped from the dishes with a rubber
policeman, transferred to microfuge tubes, and centrifuged at 14,000 rpm for 10 min at 4 °C. Supernatant fluids (containing the inositol
phosphates) were transferred to tubes containing 0.12 ml of 10 mM EDTA and 0.5 ml of tri-N-octylamine/freon (1:1, v/v) (15). These tubes were mixed vigorously and centrifuged, and
the upper phase (containing inositol phosphates) was transferred to
clean tubes and stored at 20 °C.
Perchloric acid-insoluble material (containing phosphoinositides) was
washed once with 100 mM EDTA at 4 °C and then
resuspended in 50 µl of H2O. Lipids were deacylated by
incubating these samples with 1 ml of methanol, 40% methylamine, and
n-butanol (4:4:1, v/v) for 45 min at 56 °C (15). After
drying, samples were resuspended in 1 ml of H2O, and
aliquots were removed to determine the total lipid-associated
radioactivity. These glycerophosphoinositide-containing preparations
were then extracted twice with butanol/petroleum ether/ethyl formate
(20:4:1), dried, and frozen at 20 °C.
Glycerophosphoinositide composition of the deacylated lipid preparations was analyzed by anion exchange HPLC using a Whatman Partisphere 5-SAX column. Briefly, glycerophosphoinositide preparations were reconstituted with water, filtered, and then fractionated by HPLC using a series of ammonium phosphate elution gradients designed to isolate isomers within glycerophosphoinositide classes (16). Fractions (0.4 ml) eluting from the column were collected directly into scintillation vials, and 1.2 ml of scintillation mixture (Uniscint BD; National Diagnostics, Manville, NJ) was added to each. Radioactive content of the samples was determined using a Beckman Instruments LS 6000 counter calibrated to correct for quenching and background. Sample recovery using this procedure was greater than 90%. Peaks of interest were identified by co-migration with 32P-labeled lipid standards, prepared as described previously (17).
Analysis of Inositol TurnoverInositol phosphate-containing preparations were diluted 10-fold with water and applied to columns containing 0.5 ml of anion-exchange resin AG-1, formate form (X8, 200/400 mesh; Bio-Rad). Columns were washed with 30 ml of H2O and 15 ml of 5 mM disodium tetraborate and 60 mM sodium formate to remove free inositol and glycerophosphoinositol, respectively, and then inositol phosphates were eluted with 15 ml of 1 M ammonium formate and 0.1 M formic acid (18). HPLC analysis revealed the presence of inositol phosphate, inositol bisphosphate, and inositol trisphosphate isomers in the inositol phosphate preparations (data not shown). Radioactivity associated with the inositol phosphate fractions was quantitated by liquid scintillation counting and normalized to the total uptake of label into cellular phosphoinositides, to compensate for any differences in inositol uptake or pool size between the different cell lines. Statistical significance was determined by t test using the StatView statistics program (BrainPower, Inc., Calabasas, CA).
Immunoprecipitation and Western Blot AnalysisSubconfluent
(85-90%) Ph cells were starved for 18-24 h in DMEM and 0.1% calf
serum and then stimulated with 50 ng/ml PDGF AA for 5 min. The cells
were then washed and lysed, and the chimeric receptors were
immunoprecipitated using a mouse monoclonal antibody directed against
the extracellular domain of the human -PDGFR (PR292). The
immunoprecipitates were then washed as described previously (19) and
subjected to Western blot analysis (12). The blots were developed with
Western blot detection reagents (ECL), and the signal was detected by
autoradiography.
Immunoprecipitates were incubated in
the presence of 20 mM PIPES, pH 7.0, 10 mM
MnCl2, 20 µg/ml aprotinin, and 10 µCi of
[-32-P]ATP for 10 min at 30 °C in the presence or
absence of 0.5 µg of an exogenous substrate, glutathione
S-transferase-PLC-
. The fusion protein included amino
acid residues 550-850 of rat PLC-
. The reaction was stopped by
adding an equal volume of 2 × sample buffer (10 mM
EDTA, 4% SDS, 5.6 mM 2-mercaptoethanol, 20% glycerol, 200 mM Tris-HCl, pH 6.8, and 1% bromphenol blue). The samples were then incubated for 3 min at 95 °C, spun, and resolved on 7.5%
SDS-polyacrylamide gel electrophoresis, and the radiolabeled proteins
were detected by autoradiography.
One limitation to
studying the role of the PDGFR-associated proteins in driving cellular
transformation has been the lack of a suitable cell line. To date, most
studies using mutant -PDGFRs have been done in cell lines that do
not naturally express the PDGFR. The ideal cell line to assay cell
transformation would be NIH3T3 cells; however, they have endogenous
- and
-PDGFRs, which limits their suitability for studying
introduced
-PDGFRs. Ph cells, generated from the embryos of the
Ph/Ph mouse, are 3T3-like cells that express the
-PDGFR
but not the
-PDGFR. Since PDGF AA binds only
-PDGFRs,
-PDGFRs
introduced in Ph cells can be selectively activated with PDGF AA.
However, since we were interested in studying the role of the
-PDGFR-associated proteins in driving PDGF-dependent
transformation, we constructed chimeric PDGFRs in which the
extracellular, transmembrane, and juxtamembrane regions of the
-PDGFR were fused with the intracellular domain of the
-PDGFR, as
shown in Fig. 1A. The panel of chimeras we
constructed included the wild-type PDGFR (N2WT), as well as
constructs in which the intracellular
-PDGFR domain harbored the
add-back series of
-PDGFR mutants (Fig. 1B and Ref. 12).
This series of mutant receptors was constructed by mutating the
intracellular tyrosine residues required for binding PI3K, RasGAP,
SHP-2, and PLC-
from tyrosine to phenylalanine to generate the
F5 receptor, which is unable to associate with any of these proteins.
The tyrosine to phenylalanine mutations were then restored one at a
time to construct the panel of mutants that associate with one of the
signaling molecules (PI3K, RasGAP, SHP-2, or PLC-
) (Fig.
1B).
Although PDGF AA will not activate the endogenous -PDGFRs, these
receptors could potentially contribute to signaling if the introduced
receptors were able to activate the endogenous receptor. To investigate
this possibility we examined whether the endogenous
-PDGFR was
activated following stimulation of an introduced
-PDGFR. These
studies were performed with Ph cells (Ph
WT) expressing an introduced
-PDGFR instead of the chimeric receptor, because we were unable to
locate suitable antibodies that distinguish between the chimeric
receptor and the endogenous
-PDGFR. Cultures of Ph
WT cells (11)
expressing an introduced
-PDGFR as well as the endogenous
-PDGFR
were either left resting or stimulated with 50 ng/ml PDGF AA for 5 min.
The cells were lysed, and the lysates were divided into two equal parts
and immunoprecipitated with antibodies specific for either the
- or
-PDGFR, and the immunoprecipitates were analyzed by
anti-phosphotyrosine Western blotting. PDGF AA stimulated robust
tyrosine phosphorylation of the
-PDGFR but no detectable
phosphorylation of the
-PDGFR (Fig. 2, lanes
3-6). To verify that the endogenous
-PDGFR could be activated,
we exposed the Ph parental cell lines to PDGF BB, immunoprecipitated the
-PDGFR, and subjected it to anti-phosphotyrosine Western blotting. Consistent with previous findings (11), the endogenous
-PDGFR underwent extensive tyrosine phosphorylation (Fig. 2, lanes 1 and 2). These studies indicate that
exposure of Ph cells to PDGF AA does not result in detectable
activation of the endogenous
-PDGFR, suggesting that the
-PDGFR
is not directly activated by PDGF AA or trans-activated by the
activated
-PDGFR.
The chimeras shown in Fig. 1B were constructed and subcloned
into the retroviral expression vector pLXSN2, and
recombinant retroviruses were used to infect Ph cells. Mass populations
of G418-resistant cells were grown out and sorted by
fluorescence-activated cell sorting analysis using a PDGFR antibody
(PR292) against the extracellular domain of the chimera. Receptor
levels were adjusted to the expression level of the N2WT
cell line, which was comparable to the endogenous level of -PDGFRs
in Ph cells (10). Receptor expression in the resulting cell lines was
verified by Western blot analysis of total cell lysates using an
antibody (80.8) directed against the extracellular domain of the
-PDGFR (Fig. 3A). No chimera was detected
in Ph cells expressing an empty vector, whereas all of the other cell lines expressed similar levels of the panel of the chimeric receptors. Fig. 3A, lower panel, is a RasGAP Western blot of the same
samples and demonstrates that a similar amount of total cell lysate was present in all lanes.
Characterization of the Chimeras
To assess the intrinsic
kinase activity of the chimeric constructs, we analyzed their protein
kinase activity in receptor immunoprecipitates from resting or
stimulated cells. Ph cells expressing the chimeric receptors were grown
to 85-90% confluence, starved in DMEM and 0.1% calf serum for 18-24
h, stimulated for 5 min with 50 ng/ml PDGF AA, lysed, and the chimeras
were immunoprecipitated using PR292. The immunoprecipitates were
subjected to an in vitro kinase assay in the presence of the
exogenous substrate, glutathione S-transferase-PLC-, the
proteins were resolved by SDS-polyacrylamide gel electrophoresis, and
the gel was exposed to film. No detectable receptor kinase activity was
recovered from the empty vector-expressing cells, indicating that PR292
did not recognize the endogenous
-PDGFR (Fig. 3B, lanes 1 and 2). Immunoprecipitates from all of the
chimera-expressing cell lines had readily detectable kinase activity,
as indicated by phosphorylation of the receptor itself, as well as the
exogenous substrate. In addition, the kinase activity of all of the
receptor mutants was stimulated by the addition of PDGF prior to cell
lysis. The unequal kinase activity of the different receptors was due
to variability in the amount of receptor in the samples (data not
shown) and was not routinely observed. These findings indicate that
creation of a chimeric receptor construct did not impair the intrinsic
kinase activity of the receptor.
The autoradiogram in Fig. 3 also addresses the ability of the PDGFR
mutants to recruit Src homology 2 domain-containing signal relay
enzymes, since the receptor-associated proteins which
co-immunoprecipitate with the receptor are phosphorylated in the
in vitro kinase assay. After stimulation of cells with PDGF,
the N2WT chimera co-immunoprecipitated with p148, p124,
p85, and p64 species, which are most probably PLC-, RasGAP, the p85
subunit of PI3K, and SHP-2 (Fig. 3, lanes 3 and
4). In contrast, the N2F5 chimera bound PLC-
,
p85, and RasGAP poorly (Fig. 3, lanes 13 and 14).
Although Tyr1009, which has been previously shown to be
required for SHP-2 binding, was mutated in the N2F5
receptor, SHP-2 bound to the N2F5 receptor to near
wild-type levels. Restoring the tyrosine residue at position
Tyr1009 did not greatly augment binding of SHP-2, and all
of the chimeras were able to bind SHP-2 (Fig. 3B). Restoring
tyrosines at position 1021 or 771 or both 740 and 751 selectively
restored the ability of the PDGFR to associate with PLC-
, RasGAP, or
PI3K, respectively. Similar results were obtained when the presence of
PLC-
, RasGAP, PI3K, and SHP-2 in these receptor immunoprecipitates
was monitored by Western blot analysis (data not shown). We conclude
that with the exception of SHP-2, the binding of the other
PDGFR-associated proteins examined can be modulated by mutating
receptor phosphorylation sites.
To ensure that the chimeric
receptors were able to selectively activate the signaling enzyme with
which they associated, PDGF-dependent production of PI3K
and PLC- products were examined. To assay production of PI3K
products, Ph cells expressing mutant receptors were labeled with
myo-[3H]inositol and serum-starved. After 24 h, the
cells were stimulated with PDGF AA for 3 min, and the
D-3-phosphoinositides were extracted and quantitated as
described under "Materials and Methods." In unstimulated cells,
there was a comparable basal level of
phosphatidylinositol-3,4,5-trisphosphate detected in all of the cell
lines examined (Fig. 4A). Exposure to PDGF AA
resulted in a marked increase in the level of
phosphatidylinositol-3,4,5-trisphosphate and
phosphatidylinositol-3,4-bisphosphate in cells expressing the
N2WT and N2Y40/51 chimeras, but not the
N2F5 or N2Y1021 receptors (Fig. 4A
and data not shown). These studies demonstrate that although there is
some difference in the relative amounts of products formed, only the
receptors that recruit PI3K are able to mediate this event.
We next examined PDGF-dependent PLC- activation by
measuring the accumulation of PLC-
products in intact cells. Cells
were incubated with myo-[3H]inositol for 18-24 h and
then stimulated with PDGF AA for 30 min in the presence of LiCl.
Inositol phosphates were extracted and purified by anion-exchange
chromatography. PDGF stimulated a robust increase in inositol turnover
in the N2WT and the N2Y1021 chimera-expressing
cells (Fig. 4B). In contrast, the N2F5 and
N2Y40/51 receptors were unable to activate PLC-
. Note
that the level of inositol phosphates produced in response to
stimulation of the N2Y1021 chimera was consistently greater
than that observed on stimulation of the N2WT chimera. A
possible explanation is that some of the other proteins that associate
with the N2WT receptor but not the N2Y1021
receptor negatively regulate PLC-
activation (20). Taken together
these studies indicate that the chimeric receptor mutants that bind
PI3K or PLC-
selectively activate the PI3K or PLC-
signaling
pathways, respectively.
To examine the contribution of the various signaling pathways in PDGF-dependent cellular transformation, we determined the ability of the receptor mutants to promote growth of cells in soft agar. Cells were plated in soft agar containing buffer alone, PDGF AA, or PDGF BB. PDGF AA was used to measure foci formation in response to activation of the introduced chimeric receptors, whereas PDGF BB activates both the endogenous and introduced PDGF receptors, so the resulting PDGF-dependent soft agar growth reflects the contribution of all PDGFRs. Foci were photographed and quantitated after 8-10 days of incubation at 37 °C and 5% CO2.
Cells expressing the empty vector did not grow in agar containing
buffer or PDGF AA, whereas an average of 3113 colonies were detected in
a 35-mm plate of cells supplemented with 200 ng/ml PDGF BB (Fig.
5). Like the empty vector-expressing cells, the cells
harboring the N2WT chimeras failed to grow into detectable
colonies when plated in the absence of PDGF (Fig. 5). Unlike the
control cells, however, PDGF AA stimulated a robust response in these
WT receptor-expressing cells, (5894 colonies/35-mm dish). Thus like the
endogenous -PDGFRs, the N2WT chimeric PDGFR was able to
trigger growth in soft agar. In contrast, no colonies were observed
when cells expressing the N2F5 receptor were cultured in
the presence of PDGF AA (Fig. 5), indicating that the proteins that
still bind to N2F5 are not sufficient to drive
PDGF-dependent transformation.
In response to activation using PDGF BB, Ph cells expressing an N2 or N2F5 receptor were able to drive growth in soft agar to similar levels (3113 and 3487 colonies in the N2 and N2F5 dishes, respectively). Also, the morphology of the colonies in both instances was similar. In contrast, the N2WT-expressing cells grow better in response to PDGF BB than either the N2- or N2F5-expressing cells, with approximately 6990 foci observed. The enhanced colony number is likely to be due to the contribution of the chimeric receptor. The morphology of the N2WT colonies was similar to that observed with the N2- and N2F5-expressing cells. Although these studies were done using 200 ng/ml PDGF, smaller doses (100 or 50 ng/ml PDGF AA or PDGF BB) yielded similar results (data not shown).
To compare the response of Ph cells with those of other cell types that
have been used by other groups, we analyzed PDGF-dependent soft agar growth of NIH3T3 cells and BALB/C 3T3/v-sis cells.
In the absence of PDGF, the NIH3T3 cells did not grow in soft agar, whereas PDGF BB and PDGF AA induced 4162 and 2773 colonies,
respectively (Fig. 6). These findings are consistent
with reports from other groups showing that PDGF BB is more potent than
PDGF AA (6). Unlike the NIH3T3 and Ph cells, BALB/C
3T3/v-sis cells grew well in soft agar without exogenously
applied PDGF (Fig. 6). Addition of PDGF to the culture media led to a
modest increase in the number of colonies that grew, but the
exogenously applied PDGF did not alter size or morphology of the
colonies. In comparison, the PhN2WT cells grew better in
response to PDGF than the NIH3T3 cells but produced fewer colonies than
the BALB/C 3T3/v-sis cells. The morphology of the colonies
in all three cell types was similar, except that PhN2WT
cells formed larger colonies (Fig. 6). These studies indicated that Ph
cells grow in soft agar in response to PDGF stimulation to form
colonies with a frequency, size, and morphology comparable to those
observed in other cell types. Furthermore, one or more of the signaling
pathways engaged by the N2WT but not the N2F5
receptor is required for the soft agar growth response.
We next examined whether restoring the binding sites for RasGAP or SHP-2 rescued the ability of the N2F5 chimera to promote growth in soft agar. Activation of chimeric receptors with intact RasGAP or SHP-2 binding sites did not rescue the ability of the receptor to form foci in soft agar (data not shown). Furthermore, in both cell lines, PDGF BB promoted foci formation to a level comparable to the N2F5-expressing cells.
Contrary to the results with N2Y771 and N2Y1009
receptors, restoring the binding sites for either PI3K or PLC-
rescued PDGF-dependent growth in soft agar. In the presence
of PDGF AA, both the N2Y1021 and N2Y40/51
receptor constructs produced an increase in foci formation over that of
buffer alone, with approximately 5494 and 5277 of the cells forming
foci, respectively (Fig. 7). The numbers of foci formed
were comparable to those observed with the N2WT chimera,
suggesting that restoring the binding sites for either PLC-
or PI3K
can fully restore the ability to grow in soft agar to near wild-type
levels. Colonies of N2WT and N2Y1021 cells were
morphologically indistinguishable, whereas colonies of
N2Y40/51 cells were consistently larger than those produced
by the N2Y1021- or N2WT-expressing cells (Fig.
7). Collectively these findings suggest that PI3K or PLC-
is
required for PDGF-dependent growth in soft agar.
In this study we have evaluated the importance of RasGAP,
SHP-2, PLC-, and PI3K in promoting PDGF-dependent
growth of mouse Ph fibroblasts in soft agar. We observed that receptors
capable of associating with PI3K and PLC-
, but not RasGAP or
SHP-2, initiate pathways that result in foci formation.
There is considerable controversy concerning the role of PI3K and
PLC- in PDGF-dependent cell cycle progression. One
important variable is that, depending on the type of PDGF used, one or
more isoform of the PDGF receptor can be activated. Different signaling pathways are used by the
- and
-PDGFRs; hence, the downstream effects are not identical. Studies by Yu et al. (21, 22)
suggest that stimulation of chimeric receptor constructs that contained colony-stimulating factor-1 receptor extracellularly and
-PDGFR receptor intracellularly promoted growth in soft agar in response to
colony-stimulating factor stimulation. Mutation of one or both of the
sites required for binding of PI3K and PLC-
to the
-PDGFR did not
compromise the ability of these chimeric constructs to promote growth
in soft agar (21, 22). These findings suggest that for the
-PDGFR,
PI3K and PLC-
are dispensable for PDGF-dependent growth
in soft agar. In contrast, using the
-PDGFR chimera system described
here, we find that the
-PDGFR initiates multiple pathways that lead
to growth in soft agar, and that PI3K and PLC-
are required to
engage these events. The extent to which these pathways converge
downstream of PI3K and PLC-
remains to be investigated.
In comparison with the -PDGFR, the
-PDGFR more efficiently
stimulates growth of cells in soft agar, and recent studies (9) identified a region in the
-PDGFR tail that is important for conferring the enhanced transforming activity of the
-PDGFR. These
studies were carried out by swapping a piece of the
-PDGFR into
-PDGFR, which included the PLC-
binding site for both of the
receptors. As a result the swapped receptor is an
-PDGFR with a
-PDGFR PLC-
binding site. The
-PDGFRs and
-PDGFRs have been
compared with respect to PLC-
binding and activation (23), and based
on these findings, one would predict that replacing the PLC-
binding
site of the
-PDGFR with the PLC-
binding site of the
-PDGFR
would decrease binding of PLC-
and increase tyrosine phosphorylation
and activation of PLC-
. Although the ability of the swapped receptor
to activate PLC-
has not been reported, the hybrid receptor should
activate PLC-
better than wild-type
-PDGFR. This is consistent
with our observation that PLC-
is required for soft agar growth by
the
-PDGFR. However, it is not consistent with findings by Obermeier
et al. (24), which suggest that PLC-
activation inversely
correlates with transformation. Although all of these studies indicate
that PLC-
is important in cellular transformation, the exact role of
this signaling molecule, as well as whether it has a positive or
negative effect, has not yet emerged.
An unexpected observation that arose from our studies was the ability
of the chimeric receptors to bind SHP-2 independent of phosphorylation
of tyrosine 1009 (12, 25). Previous studies have shown that mutating
the tyrosine at position 1009 in the WT receptor largely eliminates
SHP-2 binding, whereas restoring tyrosine at 1009 in the F5 receptor
enables the phosphorylated -PDGFR to bind SHP-2 stably (12). One
difference between these studies is the host cell line in which these
receptor constructs were expressed. The chimeric receptors were
expressed in Ph cells, whereas the
-PDGFR mutants were expressed in
HepG2 cells. Consequently, we tested whether the unusual binding of
SHP-2 was cell line-dependent and found that this was not
the case.2 Although SHP-2 binding is still
dependent on tyrosine phosphorylation of the receptor, it appears that
the chimera is able to bind SHP-2 in a Tyr1009-independent
manner. It is possible that creation of the chimeric receptors alters
receptor conformation such that some sequences that were previously
unavailable for SHP-2 binding are now accessible. Studies are currently
under way to understand better the binding of SHP-2 to the chimeric
PDGFRs independent of an intact binding site.
Although SHP-2 associated with all of the receptors used in these
studies, we do not think that this changes the interpretation of the
soft agar data. Previous studies from our laboratory have suggested
that SHP-2 does not play a central role in relaying signals that lead
to a proliferative response (20, 26). Furthermore, restoring the
binding site for SHP-2 to -PDGFRs capable of associating with
PLC-
had no effect on PLC-
binding, activation of PLC-
, or the
DNA synthesis response (20). Similarly, restoration of the SHP-2
binding site to receptors capable of associating with PI3K did not
inhibit PDGF-dependent activation of PI3K (27). Thus SHP-2
binding does not negatively affect signaling by either PLC-
or PI3K,
and binding of SHP-2 itself does not mediate DNA synthesis. These data
are consistent with the idea that SHP-2 may not play a pivitol role in
regulating the PDGF-dependent growth of cells in soft
agar.
Although we have focused on binding of PI3K, PLC-, SHP-2, and RasGAP
to the chimeric receptor mutants described in these experiments, it is
likely that several other proteins associate with these receptors as
well. For instance, Src associates with the F5 receptor to near-WT
levels when expressed in HepG2 or A431 cell
lines.3 The chimeric N2F5
receptor was unable to mediate PDGF-dependent growth of
cells in soft agar, indicating that Src and other proteins that may associate with the N2F5 receptor are not sufficient to
trigger a biological response. However, it is possible that Src or
these other receptor-associated proteins play a co-operative role in
mediating this response.
In summary, these studies suggest that PI3K or PLC- binding to
chimeric
/
receptors is required for PDGF-dependent
growth of cells in soft agar. Previous reports (12) have suggested that
activation of PI3K or PLC-
is sufficient to promote
PDGF-dependent DNA synthesis. The data present herein
indicate that PI3K or PLC-
are required for
PDGF-dependent transformation and suggest that the early
steps in the pathways leading to normal or cancerous growth are
surprisingly similar.
We thank Charlie Hart for the PDGF AA, Dan Bowen-Pope for the Ph cells, Chuck Stiles for the BALB/C 3T3/v-sis cells, and members of the Kazlauskas laboratory for critically reviewing the manuscript.