| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cell Growth & Differentiation |
Articles |
Center for Immunology and Microbial Disease, Albany Medical College, Albany, New York 12208
Abstract
The 44-amino acid bovine papillomavirus E5 protein induces tumorigenic transformation of immortal rodent fibroblasts by binding to and activating the platelet-derived growth factor ß receptor (PDGFßR). Here E5 was expressed in mortal human diploid fibroblasts (HDFs), which lack the accumulated genetic changes that are present in immortal rodent cells. E5 induced focus formation and morphological transformation of HDFs without inducing anchorage independence or immortalization. Similar effects were observed with the v-sis and neu* oncogenes. E5-PDGFßR complexes were observed in the E5-expressing HDFs, as was constitutive PDGFßR activation, which was required for the transforming activity of E5. The E5 HDFs attained a higher saturation density than the control cells, expressing increased levels of hyperphosphorylated retinoblastoma protein at subconfluent densities. However, when these cells reached confluence, growth inhibition accompanied by dramatic down-regulation of the PDGFßR, and retinoblastoma protein was induced apparently by a factor secreted into the medium. This may represent a novel negative feedback mechanism controlling PDGFßR-induced proliferation and thereby protecting against complete transformation.
Introduction
The E5 protein of BPV3 rapidly induces tumorigenic transformation of certain immortalized rodent fibroblasts and is the primary transforming protein of this virus (1–4) . The hydrophobic 44-amino acid E5 protein forms a disulfide-linked homodimer with a subunit size of Mr 7000 and localizes primarily to intracytoplasmic membranes in BPV-transformed cells (5, 6) . An important cellular target for the BPV E5 protein in fibroblasts is the PDGFßR. In E5-transformed rodent fibroblasts, the E5 protein forms a stable complex with PDGFßR and induces constitutive activation of the PDGFßR (7, 8) , an event that is required for cell transformation (9, 10) . The E5 protein is thought to activate the PDGFßR by binding to the receptor as a homodimer and facilitating receptor dimerization, which then leads to receptor activation (11) .
Studies of E5-mediated transformation have been performed primarily in mouse C127 cells, a line of immortalized fibroblasts derived from a mouse mammary carcinoma (12) , and the effect of E5 expression alone in mortal and completely untransformed fibroblasts has not been investigated. Analysis of mortal cells has been limited to primary rodent or bovine fibroblasts expressing the entire BPV genome. For example, BPV induced morphological transformation of primary hamster embryo fibroblasts at a low frequency, with the transformed cells being anchorage independent and tumorigenic (13–15) . BPV infection of primary bovine dermal fibroblasts resulted in morphological transformation, with the transformed cells possessing activated PDGFßR associated with the E5 protein (16) . Transfection of primary mouse fibroblasts with BPV DNA resulted in immortalization after serial subcloning, but not in focus formation (17, 18) . Finally, transfection of rat embryo fibroblasts with BPV mutants resulted in immortalization that was dependent on the BPV E6 gene (19) . From these studies, it was not possible to determine the traits that were directly attributed to E5 expression because of the presence of the entire viral genome and/or the high spontaneous transformation rates of rodent cells (20) .
Here we assessed the effects of E5 expression in mortal human fibroblasts to determine the transforming characteristics that can be ascribed solely to this small but potent oncogene. Specifically, we used a retroviral vector containing the E5 gene to introduce E5 into HDFs. We chose human cells because rodent cells are susceptible to spontaneous immortalization and tumorigenic transformation and because young mortal HDFs lack a background of accumulated genetic changes that might modify the effects of E5. We found that the E5 gene can morphologically transform and induce focus formation of these fibroblasts, which are notoriously difficult to transform. In agreement with several earlier studies (21–24) , we also found that v-sis, the viral homologue of the PDGF B gene, can induce a similar transformed phenotype. Unlike these previous studies, however, here we show that the transformed phenotype of HDFs expressing E5 or v-sis is associated with and dependent on activation of the PDGFßR. We also found that neu* (the oncogenic form of neu which encodes the p185neu receptor tyrosine kinase, activated by a single point mutation resulting in a Val to Glu substitution at position 664 in transmembrane domain) could induce focus formation and morphological transformation of these cells. Although the E5-expressing HDFs lost their contact inhibition for growth and continued to proliferate beyond the saturation density of the normal cells, their growth was still impeded once they reached confluence. This growth inhibition at confluence was accompanied by down-regulation of the PDGF receptor, and both events appeared to be induced by a factor secreted into the medium. We propose that the E5-expressing HDFs activate a negative feedback mechanism to limit the proliferative effects of sustained PDGFßR activation.
Results
Phenotypic Effects of Stable E5 Expression in HDFs.
E5 was stably expressed in HSF4012 cells (NHDF4012; Clonetics, San
Diego, CA), a human foreskin fibroblast cell strain with a normal
karyotype4
that typically undergoes senescence after 55
PDs.5
E5 was introduced into these cells by infecting low-passage
(approximately 15 PDs) HSF4012 cells with a high titer of an
E5-expressing retrovirus containing a puromycin resistance marker.
Cells stably expressing E5 were established after 1 week of selection
in puromycin-containing medium and then verified for expression of the
E5 protein by immunoblotting (Fig. 2A
). Using the analogous
retroviral construct without an insert and one containing the
v-sis gene, control and v-sis-expressing HDFs,
respectively, were established in parallel. The control HDFs were
indistinguishable from normal HSF4012 cells (Fig. 1A
). In contrast, the E5-expressing HDFs displayed an altered
morphology similar to that of murine fibroblast cell lines transformed
by this oncogene (25)
, appearing more elongated,
refractile, and dense than the control cells (compare Fig. 1B
with Fig. 1A
). Virtually every cell that
survived selection acquired the transformed morphology within just a
few PDs (by 5–6 days after retroviral infection and just before
confluence), suggesting that morphological transformation was a direct
effect of E5 expression. Generally, the E5-expressing cells appeared
the most refractile and elongated at higher cell densities,
i.e., near and at confluence (Fig. 1E
). The
E5-expressing cells also were multilayered and growing on top of each
other (Fig. 1B
), whereas the control cells appeared as a
flat monolayer (Fig. 1A
). Moreover, the E5-expressing HDFs
resembled HDFs stably expressing v-sis, as shown here (Fig. 1, C and F)
and in earlier studies (22, 24)
. A similar morphology was observed previously for HDFs
overexpressing c-sis (26)
or HDFs treated with
PDGF B
(21)
.6
Therefore, E5, like v-sis, can induce morphological
transformation of HDFs.
|
|
). This indicates that the E5 protein can form a stable
complex with the PDGFßR in HDFs. Next, the tyrosine phosphorylation
status of PDGFßRs in the E5-expressing HDFs was examined by
antiphosphotyrosine immunoblotting of PDGF receptor immunoprecipitates
from subconfluent cells. As expected, when compared with the control
cells, tyrosine phosphorylation of both mature and precursor forms of
the PDGFßR was dramatically increased in the E5-expressing HDFs (Fig. 2
C, top left panel), indicating that this receptor is
constitutively activated in these cells. A similar increase in PDGF
receptor tyrosine phosphorylation was observed in the
v-sis-expressing HDFs. In both the E5- and v-sis
expressing cells, the increase in tyrosine phosphorylation was not due
to a proportionate increase in the total amount of receptor expressed
(Fig. 2
C, bottom left panel). Thus, the E5 protein complexes
with the PDGFßR and induces its constitutive activation in mortal
human fibroblasts.
Focus Formation of HDFs by E5.
A focus-forming assay was performed by infecting low-passage
(approximately 15 PDs) HSF4012 cells with a low concentration of
retrovirus expressing E5 or v-sis. As a control, cells were
also infected with the analogous retrovirus without an insert. After
infection, cells were split and then maintained in the absence of
biochemical selection and observed for the development of foci. Fig. 3
shows that no foci were induced after infection with the control
retroviral vector (Fig. 3, Puro)
. In contrast, E5 readily
induced the formation of many large, dense foci of piled up cells. The
foci induced by E5 had a star-shaped morphology similar to that
documented previously for foci induced by v-sis or
overexpression of c-sis (the cellular gene encoding the PDGF
B chain) in HDFs (21, 23, 24, 26) . Focus formation
induced by E5 was typically observed by 14 days after infection,
suggesting that it was a direct effect of E5 expression rather than a
result of mutation during long-term culture. Corroborating the results
of previous studies (21, 23, 24)
, we showed that
v-sis also could induce the formation of numerous star-like
foci in HDFs (Fig. 3)
. In addition, the neu* oncogene also
was assessed for focus-forming activity in HDFs. neu*
encodes a mutated form of the epidermal growth factor receptor family
member p185neu that is constitutively activated
by a Val to Glu substitution at position 664 in the transmembrane
domain (27, 28)
. We found that retrovirus expressing the
activated p185neu* receptor readily induced focus
formation of HDFs, whereas the corresponding empty retroviral vector
(LXSN) did not (Fig. 3)
. Thus, constitutive activation of a receptor
tyrosine kinase other than the PDGF receptor can result in focus
formation in HDFs.
|
C, top panel). AG1296 treatment before confluence also
reversed the morphology change of the E5- (Fig. 1D
) and
v-sis (data not shown)-expressing HDFs without preventing
continued growth to confluence. Specifically, the AG1296-treated cells
became flatter and less refractile, resembling the control cells. It is
important to note that the effects of AG1296 in the E5-expressing cells
were reversible; after its removal, tyrosine phosphorylation of the
PDGF receptor was restored within an hour, and the characteristic
transformed morphology of the cells was reestablished within 1–2 days
(data not shown). This suggests that E5 expression was not lost in the
AG1296-treated cells. AG1296 treatment also inhibited focus formation
by E5 or v-sis (Fig. 3)
. This inhibition was not simply due
to inefficient retroviral infection because cells treated with AG1296
were taken from the same infections that resulted in focus formation.
Thus, these results indicate that constitutive activation of the PDGF
receptor is required for E5- and v-sis-induced morphological
transformation and focus formation of HDFs. To ensure that AG1296
inhibitor was specific for the PDGF receptor, we assessed whether it
could inhibit morphological transformation and focus formation induced
by constitutive activation of a different receptor tyrosine kinase,
p185neu*. HDFs stably expressing the activated
p185neu* were established by retroviral-mediated
gene transfer as described in "Materials and Methods." These cells
appeared to be morphologically transformed (data not shown) and
expressed abundant amounts of tyrosine-phosphorylated
p185neu* as assessed by
p185neu immunoprecipitation followed by
antiphosphotyrosine immunoblotting (Fig. 2
C, top panel).
AG1296 treatment before confluence could not inhibit tyrosine
phosphorylation of p185neu* (Fig. 2C
),
nor could it reverse morphological transformation by neu*
(data not shown). AG1296 was also unable to inhibit focus formation
induced by neu* (Fig. 3)
and therefore was not a general
inhibitor of focus formation. Taken together, these data indicate that
the AG1296 inhibitor is specific for the kinase activity of the PDGF
receptor. A structurally similar inhibitor, AG1295, was shown to be
specific for E5-activated PDGF receptors and not for activated
p185neu* in mouse C127 cells (30)
.
Growth Kinetics of E5-expressing HDFs.
HDFs stably expressing E5, v-sis, or the empty retroviral
vector were seeded into multiple dishes at the same cell concentration
and then counted at intervals thereafter. As illustrated by
representative growth curves (Fig. 4
) the E5- and v-sis-expressing HDFs consistently reached a
2–3-fold higher saturation density than the control cells. During the
initial phase of exponential growth, the growth rate of the E5-and
v-sis-expressing cells was not significantly different than
that of the control cells. However, the E5- and
v-sis-expressing cells continued to proliferate
exponentially after growth of the control cells had slowed down.
Notably, both control and transformed cells continued to grow, albeit
slowly, for many days after exponential growth. Therefore, although the
E5- and v-sis-expressing cells did not display an increased
growth rate initially, the exponential growth phase of these cells was
extended, allowing them to attain a higher saturation density than the
control cells.
|
). Control or E5-expressing HDFs were seeded into medium containing
either 1% or 10% FBS and then counted at various times thereafter.
Growth of both the control and E5-expressing cells in 1% FBS was
severely restricted compared with growth in 10% FBS. This indicates
that the E5-expressing cells, like the normal cells, required higher
serum concentrations for optimal growth. However, even in 1% FBS, the
E5-expressing HDFs were able to attain a higher final cell density than
the control cells. Thus, although the E5 HDFs were serum dependent for
growth, they still maintained a growth advantage under low serum
conditions.
|
The E5-expressing cells were also tested for the ability to bypass senescence. We found that the E5-expressing cells, like the normal cells, senesced after serial passaging (data not shown), indicating that E5 was unable to directly induce immortalization of HDFs. Furthermore, E5 expression did not appear to induce premature senescence of HDFs, as was reported previously for ectopic expression of activated ras in these cells (31) .
Down-Regulation of the PDGF Receptor in E5-expressing HDFs.
For the PDGFßR and other receptor tyrosine kinases, ligand binding
not only incites receptor activation but also triggers the clustering
of receptors in coated pits, followed by endocytosis and lysosomal
degradation of receptor-ligand complexes (32, 33) , a
process referred to as receptor down-regulation. To determine
the effect of E5 on PDGFßR down-regulation in HDFs, we examined PDGF
receptor levels at different cell densities. An equal number of normal
or E5-expressing cells was seeded into multiple dishes, and, at various
times during expansion of the cultures, monolayers were lysed for
immunoblot analysis or trypsinized in parallel for cell counts. To
detect PDGF receptor levels in the cells, whole cell lysates were
subjected to PDGF receptor immunoblotting. Fig. 6A
shows that at any given cell density, the level of PDGF
receptors in the E5-expressing cells was significantly less than that
in the control cells. This is likely due to receptor down-regulation in
response to E5-induced receptor activation. Strikingly, when
E5-expressing HDFs had just reached confluence (at approximately 2 x 106 cells/60-mm dish; Fig. 6B
),
PDGF receptor levels were nearly abolished (Fig. 6
A, upper panel,
E5 Lane 3). A similar decrease in the amount of activated,
tyrosine-phosphorylated receptor was observed when the cells reached
confluence (data not shown). Thereafter, the growth rate of these cells
was greatly reduced, and PDGF receptor levels remained diminished for 3
days (Fig. 6A
, upper panel, E5 Lanes 4 and
5). Dramatic receptor down-regulation also was observed when
v-sis-expressing cells reached confluence (data not shown).
In contrast, the levels of receptor in the control cells remained
unchanged after the cells attained confluence and exhibited a reduced
growth rate (Fig. 6
A, control Lane 3). Receptor
down-regulation at confluence in the E5 HDFs was inhibited by AG1296
treatment (Fig. 9)
, suggesting that down-regulation required the kinase
activity of the PDGF receptor. Therefore, constitutive activation of
the PDGFßR by E5 in HDFs appears to lead to dramatic down-regulation
of the receptor when the cells reach confluence. Others have
demonstrated down-regulation of the PDGFßR in response to activation
by E5 (34)
, v-sis (35)
, or
v-fps (36)
. However, none of these studies
related PDGFßR down-regulation to cell density.
|
|
(top panel), the conversion of
distinct PDGF receptor bands to faint, smeary bands suggested that this
receptor was being rapidly degraded. This degradation was specific for
the PDGF receptor because the levels of actin remained constant at
different cell densities, as assessed by an anti-actin immunoblot of a
lower portion of the same gel. A pulse-chase experiment confirmed that
the half-life of the PDGF receptor was markedly reduced in the
E5-expressing HDFs when they reached confluence (Fig. 7
). Twelve h before confluence (72 h after plating), a significant amount
of receptor was still present after a 1-h chase, and complete
degradation occurred after a 4.5-h chase. It is important to note that
at this time, the PDGF receptor form with the longest half-life is one
with an electrophoretic mobility between the mature (Fig. 7, m)
and precursor (Fig. 7, p)
forms. This may
represent a more stable form of the receptor that is either a specific
degradation product of the mature form or a different, incompletely
processed precursor. In contrast to the subconfluent cells, in
confluent cells (84 h after plating) metabolically labeled PDGF
receptor is almost completely degraded after a 1-h chase. Thus,
increased degradation accounts for down-regulation of the PDGF receptor
in confluent E5 HDFs.
|
(bottom panel) shows that in subconfluent control and E5
HDFs (Lanes 1 and 2) both hyper- and
hypophosphorylated forms of Rb (corresponding to the upper
and lower bands, respectively) were present. This is to be
expected because the hyperphosphorylated form should be prevalent in
dividing cells, allowing for cell cycle progression by releasing
the transcription factor E2F. The amount of Rb in subconfluent
E5-expressing cells appeared to be significantly increased compared
with that in the control cells, particularly when the cells were
approximately 50% confluent (Lanes 1). This was due
primarily to an increase in hyperphosphorylated Rb (indicated by a
lighter exposure of the Rb blot in Fig. 6A
) and could
therefore account for the extended exponential growth of the
E5-expressing HDFs compared with the control cells. When both the
control and E5 HDFs reached confluence (Lanes 3), a
significant reduction in hyperphosphorylated Rb was observed. In
addition, when the E5-expressing HDFs reached confluence, a faster
migrating Rb form of approximately Mr
68,000–70,000 appeared that was not evident in the control cells. The
presence of this smaller Rb form may indicate specific cleavage of Rb,
implying that both Rb and the PDGF receptor are degraded in the E5 HDFs
when confluence is reached.
We next determined whether the state of confluence was the direct cause
of PDGF receptor down-regulation in the E5-expressing cells. For
example, it is possible that the increased number of cell-cell contacts
in a confluent culture might trigger a signal for receptor degradation.
E5-expressing HDFs were plated at a high density (2.7 x
106 cells/dish, which is at or slightly higher
than confluence) or at a low density (2 x
105 cells/dish). At various times after plating,
cells were lysed and examined for PDGF receptor levels by
immunoblotting. Fig. 8
shows that the PDGF receptor was not immediately degraded after plating
at saturation density. Instead, receptor degradation began some time
between 1 and 3 days after plating. Thus, in addition to confluence,
there appears to be a time requirement for PDGF receptor
down-regulation in the E5-expressing cells. In another experiment, we
measured whether the E5 HDFs could continue to grow after being plated
at confluence (approximately 2.3 x 106
cells/dish). Indeed, these cells continued to grow, although not at the
exponential rate, reaching a final density of approximately 3 x
106 cells/plate 2 days after plating (data not
shown). This corresponded to the time that a more transformed
morphology and degradation of the PDGF receptor became evident (data
not shown). Continued growth after plating at confluence may be due to
the ability of these cells to grow on top of each other.
|
shows that serum deprivation clearly inhibited PDGF receptor
down-regulation, even when serum was removed just as the cells were
starting to degrade their receptors (i.e., at 84 h). If
the media on the cells were replaced with serum-containing media,
receptor down-regulation still occurred, although the response was
delayed by 1 day and reduced. These results indicate that serum factors
are required for receptor down-regulation and argue against the
possibility that receptor down-regulation is caused by a depletion of
factors in the medium. If this were the case, down-regulation should
not occur after replacement with fresh serum-containing medium. A
more likely possibility is that the cells secreted a factor that
induced receptor down-regulation and that media replacement removed
such a factor. Media replacement with serum-containing medium would
still remove such a factor, but the presence of serum might allow cells
to secrete more of this factor and eventually induce receptor
down-regulation.
To test the hypothesis that a secreted factor induces PDGF
receptor degradation, we determined whether conditioned media from
confluent E5 HDFs that were degrading their PDGF receptors could induce
receptor down-regulation in subconfluent E5 cells that had not yet
begun to degrade their receptors (Fig. 10A
). Briefly, the media on subconfluent E5 cells was replaced
with media from E5 cells that had just reached confluence and were
down-regulating their PDGF receptors (S lane in Fig. 10A
). Both treated and untreated cells were lysed 2.5 and
20 h later, and PDGF receptor levels were examined by
immunoblotting. Strikingly, a 20-h treatment with conditioned media
from confluent cells dramatically down-regulated the PDGF receptors in
subconfluent cells (Fig. 10A
). Even a short treatment of
2.5 h induced some receptor degradation. Media taken from
confluent control cells not expressing E5 could not induce receptor
down-regulation (Fig. 10B
). Receptor down-regulation was not
a result of the subconfluent cells becoming confluent and
down-regulating their receptors independent of the treatment because
receptor degradation was not evident in the untreated cells. Moreover,
the fact that subconfluent cells could degrade their PDGF receptors in
response to the conditioned media provides further evidence that the
confluent state is not a direct cause of receptor degradation, although
it may be required for secretion of a factor. Interestingly, a 20-h
treatment with this conditioned medium also induced down-regulation of
hyperphosphorylated Rb (Fig. 10, A and B)
. Thus,
down-regulation of hyperphosphorylated Rb and degradation of the PDGF
receptor are induced by the same stimulus, presumably a factor secreted
into the medium at confluence.
|
). After
20 h of treatment, there was minimal receptor down-regulation, but
by 43 h, there was significant receptor down-regulation in these
cells. Conditioned medium did not induce receptor activation in the
control cells (data not shown). By 43 h, hyperphosphorylated forms
of Rb were also down-regulated in the control cells (data not shown).
We next began to characterize the factor in the conditioned medium of
confluent E5 HDFs. First, boiling the conditioned media before
treatment did not inhibit the ability of the media to induce
down-regulation of the PDGF receptor in the E5 HDFs (Fig. 10D
), suggesting that the down-regulating activity in the
medium was not enzymatic. To estimate the size of the factor, we
treated subconfluent E5 HDFs with medium that was fractionated through
a centrifugal filter device with a molecular weight cutoff of 3000. The
majority of the down-regulating activity in the medium fractionated to
the filtrate containing low molecular weight solutes
(Mr less than
3000) and was depleted from the retentate
containing the high molecular weight solutes (Fig. 10E
).
Taken together, these results suggest that when E5 HDFs reach
confluence, they secrete a small, heat-stable soluble factor that
induces down-regulation of the PDGF receptor and hyperphosphorylated
Rb.
We also determined the effect of this conditioned medium on DNA
synthesis and cell proliferation. Subconfluent E5 or control HDFs were
either treated with conditioned medium from confluent E5 HDFs or left
untreated. At various times after treatment, cells were examined for
incorporation of [3H]thymidine and counted in
parallel. Fig. 11
shows that by 43 h of treatment, DNA synthesis in the treated E5
HDFs was decreased 7-fold compared with the untreated cells. In
addition, the treated E5 HDFs could only reach a peak density that was
half that of the untreated cells, suggesting that cell growth also was
inhibited. By 65 h after treatment, the E5 HDFs showed a 3-fold
reduction in cell number compared with the untreated cells. Thus,
conditioned medium from confluent E5 HDFs suppressed DNA synthesis and
cell proliferation in subconfluent E5 HDFs. It is important to note
that this effect occurred after the time (20 h) when down-regulation of
the PDGF receptor and Rb was typically observed. We expected to see a
decrease in DNA synthesis in the treated control cells 1 day after
receptor down-regulation typically occurred (e.g., as shown
in Fig. 10C
). In this particular experiment, however, these
cells probably initiated the normal mechanism of density-dependent
growth arrest before factor-dependent growth arrest could occur.
|
In this report, we demonstrate that the 44-amino acid BPV E5 protein can induce morphological transformation and focus formation and increase the saturation density for growth of mortal HDFs. Similar effects were observed for expression of the v-sis oncogene, as shown previously (21–24) . Unlike these earlier studies, here we provide evidence that constitutive activation of the PDGF receptor mediates the transforming effects of E5 or v-sis in HDFs. We also found that constitutive activation of another receptor tyrosine kinase, p185neu*, imparts similar phenotypic changes in these cells. Therefore, two different receptor tyrosine kinases can apparently use common signaling pathways to elicit transforming phenotypes in human fibroblasts.
One pathway shared by receptor tyrosine kinases is the Ras/MAPK pathway (37) . However, instead of inducing transformation, exogenous expression of activated ras in HDFs has been shown to induce premature senescence associated with increased p16INK4a and p53 expression (31) . In this study, it was proposed that premature senescence occurred to protect against an initial mitogenic stimulus by ras that could lead to extensive proliferation and tumorigenesis. Similar results were observed after exogenous expression of constitutively activated forms of Raf or MAPK kinase in HDFs (38, 39) , suggesting that the Ras/MAPK pathway is involved in the induction of p16INK4a and p53 expression, which, in turn, promotes senescence. Therefore, activated ras has a very different effect in HDFs than constitutive activation of the PDGFßR or p185neu, although Ras is a downstream effector of these receptors. This may be because the phenotypic changes induced by sustained activation of a receptor tyrosine kinase represent the combined effect of several interconnected signaling pathways, which may be very different from the effect of activating only the Ras/MAPK pathway. In the E5-, v-sis-, and neu*-expressing HDFs, perhaps another receptor tyrosine kinase signaling pathway inhibits senescence by suppressing MAPK activation. For example, in certain cell systems, the Raf-MAPK kinase-MAPK pathway is inhibited by activation of the phosphatidylinositol 3'-kinase-protein kinase B pathway through inhibitory phosphorylation of Raf by protein kinase B (40, 41) . We have preliminary evidence suggesting that MAPK activity is only transiently increased in the E5- and v-sis-expressing cells; compared with the control cells, it is only increased at very low densities and is significantly reduced at higher densities (data not shown). A lack of sustained MAPK activity may explain why these cells do not undergo premature senescence. Nonetheless, a different growth inhibitory pathway, which leads to quiescence rather than senescence, is activated in the E5-expressing HDFs (see below). Importantly, HDFs appear to have multiple growth control pathways that may serve to protect against complete transformation.
Despite the focus-forming activity of E5, which indicates a loss of contact inhibition, the growth rate of the E5-expressing HDFs was still was greatly impeded once they reached confluence. We also found that the PDGF receptor was rapidly and dramatically down-regulated in the E5-expressing cells and not in the control cells when confluence was reached. We propose that the E5-expressing HDFs activate an alternative growth inhibitory pathway at confluence, resulting in density-dependent growth arrest in the absence of contact inhibition. Specifically, our data suggest that a soluble factor secreted by these cells induces degradation of both the PDGFßR and hyperphosphorylated Rb, leading to subsequent growth inhibition. Because of its insensitivity to heat and its small size, the factor secreted by E5 HDFs is not likely to be a protease that directly degrades the PDGFßR and Rb. Instead, such a factor is more likely to stimulate an intracellular degradation pathway. This is substantiated by the fact that only cycling cells and not quiescent cells can degrade their receptors in response to this factor (data not shown). Thus, we present the possibility that an intracellular proteolysis pathway may be part of a back up densitydependent growth control mechanism in HDFs. Additional experiments are required to identify the secreted factor, determine the basis for its expression, ascertain the degradation pathway it induces, and determine the role of this pathway in growth inhibition.
The confluent state is probably involved in secretion of the factor rather than the degradation pathway per se because degradation was activated in subconfluent cells in response to the factor. However, the confluent state is unlikely to directly cause secretion of the factor because plating cells at a confluent density did not immediately induce PDGF receptor down-regulation. Under these conditions, the ability of the cells to assume the typical elongated and refractile "transformed" morphology (data not shown) was also delayed and coincided with PDGF receptor down-regulation. Therefore, it may be the change in cell shape associated with the confluent state that stimulates the secretion of a factor.
The inability of the control cells to down-regulate the PDGF receptor
at confluence may reflect their inability to secrete a factor rather
than respond to one. Indeed, a soluble factor from the E5 HDFs could
induce down-regulation of the PDGF receptor and Rb in control cells,
albeit more slowly, despite the fact that the control cells have
only a low basal level of activated PDGF receptor. Therefore, the
down-regulation pathway probably does not require a high level of PDGF
receptor activation, although the increased level of activated
receptors in the E5 HDFs may expedite this process. This seems to be
inconsistent with previous studies suggesting that internalization and
degradation of the PDGFßR require the receptor to be active and able
to phosphorylate one of its substrates, phosphatidylinositol 3'-kinase
(42, 43)
. It is possible that instead of receptor-mediated
endocytosis, the E5 HDFs use another pathway such as ubiquitin-mediated
proteolysis to degrade the PDGF receptor and Rb at confluence. Because
conditioned medium from confluent control cells could not induce
receptor down-regulation in the E5 cells (Fig. 10)
, receptor activation
is probably required for secretion of the factor. This may explain why
AG1296 inhibits receptor down-regulation at confluence in the E5 cells
(Fig. 9)
. Thus, receptor activation appears to be required for
secretion of the factor, but receptor activation over a basal level may
not be required for the down-regulation process itself.
Like the PDGF receptor, Rb is probably down-regulated by proteolytic digestion when the E5 HDFs reach confluence, as evidenced by the disappearance of hyperphosphorylated Rb and the appearance of a smaller Mr 68,000–70,000 form. This raises the intriguing possibility that degradation of Rb and degradation of the PDGF receptor occur by the same mechanism. This is supported by the fact that down-regulation of Rb and the PDGF receptor in subconfluent control and E5 HDFs occurred in response to the same stimulus (i.e., a secreted factor from confluent E5 HDFs). A secreted factor could stimulate the activation of a specific protease that acts on the PDGF receptor and Rb as substrates, suggesting that the PDGF receptor and Rb share a common proteolytic recognition sequence. Similarly, the secreted factor could activate the specific targeting of these proteins for ubiquitinization. Alternatively, a proteolytic product of one protein could stimulate the degradation of the other. Whatever the mechanism, this represents the first report linking posttranslational processing of Rb to that of the PDGF receptor.
The Mr 68,000–70,000 Rb form is
likely to be a specific COOH-terminal cleavage product because the
antibody that detected it was raised against a COOH-terminal peptide.
Others have detected a similar Mr
68,000 Rb form in human leukemia and lung cancer cell lines
(44–46)
. One group reported that this Rb form results
from specific cleavage by a caspase-like activity during apoptosis
(44, 45)
. It is possible that degradation of Rb and the
PDGF receptor in the E5 HDFs is also associated with apoptosis because
a drop in cell number was observed after treatment of cells with
conditioned medium from confluent E5 HDFs (Fig. 11)
. A different group
also detected a COOH-terminal Mr
68,000 Rb form but found that it was not associated with apoptosis
(46)
. Additional experiments are required to characterize
this Mr 68,000–70,000 Rb form and
assess its role in growth control of the E5 HDFs.
We believe that that down-regulation of Rb and the PDGF receptor plays a role in growth inhibition at confluence because it stands to reason that a dramatic decrease in the amount of activated PDGFßR and hyperphosphorylated Rb might serve to limit cell growth. Furthermore, when subconfluent cells were treated with the secreted factor, down-regulation of both proteins appeared to precede growth inhibition. However, our data do not rule out the possibility that down-regulation of these proteins and growth inhibition result from parallel pathways induced by the same stimulus, and additional experiments are required to definitively demonstrate that the two events are causally related.
The E5-expressing cells attained a higher saturation density than the control cells under both high and low serum conditions, indicating that constitutive PDGFßR activation allowed a proliferative advantage. However, because the growth rate of the E5-expressing cells was significantly impaired under low serum conditions, constitutive activation of the PDGFßR is not sufficient for optimal growth of HDFs. This may explain why exponentially growing E5 HDFs had a growth rate similar to that of exponentially growing control cells. We believe that sustained PDGFßR activation provides a proliferative advantage by delaying the onset of a density-dependent growth control mechanism rather than by increasing the rate of cell cycle progression. In both the E5 and control HDFs, growth inhibitory mechanisms are activated once confluence is reached. However, the E5 HDFs attain confluence at a higher density than the control cells, thereby raising the threshold density for onset of growth inhibition. Two characteristics of the E5 HDFs may account for this: (a) reduced cell spreading at higher densities, which could cause each cell to occupy less space; and (b) the ability of the cells to grow on top of each other. Both factors may be related to alterations in cell attachment. The fact that a back up growth control mechanism is activated in the E5 HDFs at confluence despite a loss of contact inhibition suggests that confluence is an important growth-restricting state. Thus, sustained PDGFßR activation may effect changes in cell attachment, which allow the cells to attain a higher density before reaching a growth-restricting state.
Unlike previous studies that showed that v-sis, c-sis, or PDGF could induce anchorage-independent growth of HDFs (22, 26, 47) , we were unable to demonstrate this activity for either E5 or v-sis. One might argue that the level of E5 or v-sis expressed in our system was not sufficient to induce detectable anchorage-independent growth. By preselecting cells that expressed high levels of c-sis, Stevens et al. (26) showed that a threshold level of c-sis expression was required for efficient anchorage-independent growth of HDFs. In our system, there appeared to be sufficient E5 and v-sis expression to induce abundant tyrosine phosphorylation of the PDGFßR, yet anchorage independence was not induced. Alternatively, the discrepancy between previous studies and our study may be explained by our more rigorous assessment of anchorage independence. First, we used stricter criteria (>100 µ) for assigning anchorage-independent growth and were therefore less likely to consider false positives significant. Even when the E5- and v-sis-expressing cells displayed a low level of growth in semisolid medium (which was over 100-fold less than that of an anchorage-independent cell line), the same low level was also observed for the control cells. Second, by using cells that were derived from a transformed colony and expanded, the previous studies may have selected for additional mutations conferring anchorage independence. It is possible that during the expansion and passage of clonal populations, mutations occurred that led to increased numbers of anchorage-independent colonies. In our study, the cells were not cloned and were used soon after infection. In any event, our data suggest that E5 and v-sis are unable to directly induce anchorage-independent growth. This implies that constitutive activation of the PDGFßR in HDFs is not sufficient to induce anchorage independence. It is possible that the growth inhibitory pathway activated at confluence also may be activated when these cells are suspended in semisolid medium.
Because E5 was unable to induce anchorage-independent growth, immortalization, or complete growth in low serum of HDFs, we expect that it would be unable to fully transform these cells to tumorigenicity. These observations are consistent with those of several other studies that reported that v-sis could not impart an extended life span or make cells tumorigenic (22, 23, 24) . Thus, although E5 has a potent transforming effect in HDFs, it provides only some of the steps necessary for the multistage process of oncogenesis. Recent evidence suggests that one requirement for tumorigenic transformation of human cells is telomere maintenance and immortalization promoted by expression of the telomerase catalytic subunit hTERT (48) . Hahn et al. (48) showed that ectopic expression of hTERT, together with activated Ras and the SV40 large T antigen, was sufficient to allow human epithelial and fibroblast cells to form tumors when injected into nude mice. In this study, expression of hTERT with only activated Ras resulted in senescence, as shown previously for expression of activated Ras alone (31) . Because E5 did not appear to induce this premature senescence phenotype, it would be interesting to assess the effect of coexpressing E5 with hTERT and determine what additional oncogene functions, if any, would be necessary and sufficient to induce tumorigenic transformation. For example, if the Mr 68,000–70,000 form of Rb observed in the E5 HDFs plays a role in growth inhibition at confluence, human papillomavirus E7 might be able to contribute to complete transformation by inactivating this Rb form.
In conclusion, we provide evidence that constitutive activation of the PDGFßR can elicit transforming characteristics in mortal human fibroblasts. This stands to reason because the PDGFßR is capable of eliciting signals to induce cell proliferation and is abundantly expressed in fibroblasts. However, it appears that HDFs adopt a strategy to abate the proliferative effect of PDGFßR activation, perhaps as a protective mechanism against further transformation. We have evidence to suggest that part of this strategy may be to activate an intracellular proteolysis pathway. Additional experiments are required to identify this pathway and determine how it may function in normal cells. Thus, the E5-expressing HDFs may allow characterization of a novel growth regulatory mechanism. These cells should also be useful for examining the mechanistic relationship between loss of contact inhibition and morphological transformation apart from other transforming phenotypes.
Materials and Methods
Plasmids.
pRVH-E5 and pRVH-v-sis plasmids were obtained from Daniel
DiMaio (Yale University, New Haven, CT). Standard subcloning
procedures were used to subclone the BPV E5 and v-sis
oncogenes from these plasmids into the pBabepuro retrovirus vector (a
gift from Jeff Settleman; Harvard University, Cambridge, MA) to
generate pBabepuro-E5 and pBabepuro-sis, respectively. The
neu* cDNA was subcloned into the LXSN retrovirus vector from
pSR
-NeuNT (a gift from David Stern; Yale University) using standard
methods. High-titer amphotropic retrovirus was obtained after transient
transfection of the retroviral plasmids into the Phoenix amphotropic
producer cell line (American Type Culture Collection with
permission from G. Nolan; Stanford University, Palo Alto, CA) as
described previously (49)
.
Stable Expression of E5, v-sis, and
neu*.
Young HSF4012 (NHDF4012) human foreskin fibroblasts were purchased from
Clonetics at passage 1 and maintained in MEM--10 [MEM- (Life
Technologies, Inc.) supplemented with 10% FBS, 50 units/ml penicillin,
and 50 µg/ml streptomycin]. Cell strains stably expressing
E5, v-sis, or neu* were established using
high-titer recombinant retroviruses expressing these oncogenes to
infect low-passage HSF4012 cells. Briefly, approximately 1 x
106 colony-forming units of retrovirus and 16
µg of Polybrene in 4 ml were added to HSF4012 cells at 80%
confluence in a T75 flask. After 1 day, the infected cells were
trypsinized and split into eight to ten 100-mm dishes. After allowing
the cells to adhere to the dishes, puromycin (Sigma) or G418 (Life
Technologies, Inc.) was added to the cells at a final concentration of
0.5 or 400 µg/ml, respectively. After 1 week of selection, stable
cell strains were established. Control cells were established by
infection with the corresponding retrovirus without an insert.
Focus-forming Assay.
Passage 7 HSF4012 cells, which were at 90% confluence in 60-mm dishes,
were infected with approximately 1 x 103
colony-forming units of recombinant retrovirus. One day after
infection, the cells in each dish were split 1:3. Three days later,
when the cells were 90% confluent, 7 µl of DMSO or Tyrphostin AG1296
(Calbiochem; 17.5 µM, final concentration) were added to
the cells. The cells were then maintained at confluence in MEM--10,
with a weekly media change containing fresh AG1296 or DMSO. Foci were
visible 10–14 days after infection, and 3 weeks after infection,
monolayers were fixed and stained with 1% crystal violet (w/v) in 70%
ethanol.
Growth Curves.
HDFs expressing E5, v-sis, or the empty retroviral vector
were seeded into multiple 60-mm dishes at 1 x
105 cells/dish. At intervals thereafter, cells in
triplicate dishes were trypsinized and counted using a Coulter counter.
The number of cells per dish was converted to the ln, and the results
from the triplicate dishes were used to determine the mean and SD
at each time point. Cell doublings were calculated using the formula
 |
Growth in Low Serum.
Cells were diluted into media containing 1.0% FBS and counted. Cells
were then added to media containing either 1% or 10% FBS and plated
at a density of 105 cells/60-mm dish. Cells were
trypsinized and counted in triplicate at intervals thereafter. Growth
curves were plotted as described above.
Anchorage Independence Assay.
Control, E5- or v-sis-expressing cells (3 x
105) were suspended in 10 ml of 1.5%
methylcellulose in complete media as described previously
(50)
. This suspension was mixed and distributed equally to
three gridded 60-mm dishes that had been precoated with sterile 1.0%
agarose in PBS. HT1080, a cell line derived from a human fibrosarcoma,
was used as a positive control and found to be virtually 100%
anchorage independent, based on daily observation of colony growth.
Dishes were checked for clumps on the day after initiation of the
assay, and no significant clumping problems were detected. After
approximately 3 weeks, colonies that were greater than 100 µm were
considered positive for anchorage-independent growth.
Immunoprecipitation and Immunoblotting.
For the experiment shown in Fig. 2C
, HSF4012 cells stably
expressing E5, v-sis, or neu* in 60-mm dishes at
80% confluence were treated with 10 µl of DMSO or 10 µl of 10
mM AG1296 (final concentration, 25
µM). Twenty-four h later, these cells as well
as the corresponding untreated control cells (expressing the empty
retroviral vectors pBabepuro or LXSN) were lysed in cold EBC
buffer as described previously (7, 8, 15)
. For the
experiment shown in Fig. 9
, cells were treated with AG1296 in the same
manner but lysed 2 days after treatment in protein sample buffer as
described below. For Fig. 2, A and B
, cells were
lysed in cold
3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid
extraction buffer as described previously (29)
. PDGF
receptor was immunoprecipitated as described previously (8, 9, 15)
by adding approximately 1 µl of anti-PR-C3a antibody (a
gift from Daniel DiMaio; this antibody recognizes the
COOH-terminal 13 amino acids of the human PDGFßR) per 100–200 µg
of protein extract. p185neu* was
immunoprecipitated from EBC cell extracts by adding 10 µl of
anti-p185neu antibody Ab-4 (Calbiochem) to
approximately 600 µg of protein extract. Immunoprecipitation of the
E5 protein and its associated proteins was performed as described
previously (8, 15) by adding approximately 1 µl of
anti-E5 antibody (a gift from Daniel DiMaio; this antibody
recognizes the COOH-terminal 16 amino acids of the E5 protein) per
100–200 µg of protein extract. Washed immunoprecipitates were
resuspended in 2x Laemmli sample buffer. To prepare whole cell
extracts for the experiments shown in Figs. 6–10
, the monolayers were
washed twice with PBS and then lysed by adding 200–250 µl of hot 2x
Laemmli sample buffer. Whole cell extracts or immunoprecipitates were
boiled; electrophoresed on a SDS-7%, 7.5%, or 15% (for E5
immunoblotting) polyacrylamide gel; and transferred to nitrocellulose
or polyvinylidene difluoride (for E5 immunoblotting) at 100 V for 2
hours. Immunoblotting was performed as described previously (8, 15)
using a 1:1000 dilution of antiphosphotyrosine monoclonal
antibody 4G10 (Upstate Biotechnology, Inc.), a 1:1000 dilution of
anti-PDGF receptor antiserum (PR-C3a), a 1:100 dilution of
anti-p185neu monoclonal antibody (Ab-3,
Calbiochem), a 1:500 dilution of anti-E5 antiserum, a 1:750 dilution of
anti-Rb antibody (C-15; Santa Cruz Biotechnology), or a 1: 500 dilution
of anti-actin antiserum (Sigma). Proteins were detected by enhanced
chemiluminescence using a protein A- or antimouse IgG (for
p185neu immunoblot)-horseradish peroxidase
conjugate (Pierce).
Pulse-Chase Experiment.
Monolayers of E5 or control HDFs that were either 80% (72 h after
plating) or 100% (84 h after plating) confluent in 60-mm dishes were
washed twice with PBS and preincubated for approximately 30 min in
serum-free, methionine-free, and cysteine-free DMEM. Cells were then
incubated in 1 ml of fresh serum-free, methionine-free, and
cysteine-free medium containing approximately 418 µCi of
Expre35S35S label (New
England Nuclear; 71% [35S]methionine and 29%
[35S]cysteine) for 45 min. Cells were then
washed twice with PBS and either lysed immediately in EBC buffer or
incubated in fresh medium containing 150 µg/ml
L-methionine and 120 µg/ml L-cysteine for
various times before lysis. After preclearing the extracts by
incubation with protein A-Sepharose beads, PDGF receptor was
immunoprecipitated and subjected to SDS-PAGE, followed by fluorography
and autoradiography.
Fractionation of Conditioned Medium.
Conditioned medium from E5 HDFs that had just reached confluence was
fractionated by centrifugation at 1600 x g for 3 h through a Centriplus (Millipore) centrifugal filter device with a
molecular weight cutoff of 3000. The retentate, which was concentrated
down to a smaller volume and contained solutes with molecular weights
of >3000, was added directly to the existing medium of subconfluent E5
HDFs to restore its original concentration. The filtrate, which
contained solutes with molecular weights of <3000, was added to
subconfluent E5 HDFs after removal of the existing medium.
DNA Synthesis Assay.
For the experiment shown in Fig. 11, E5
or control HDFs were seeded at
1–2 x 104 cells/well in 24-well plates,
and 2 days later, when the cells were still subconfluent, they were
either left untreated or treated with conditioned medium from E5 HDFs
that had just reached confluence. At the time of treatment and various
times thereafter, the cells were assayed for
[3H]thymidine incorporation or trypsinized for
cell counts. For [3H]thymidine incorporation,
cells were incubated for 2 h with 100 µl/well of 1.5 µCi/ml
[3H]thymidine (New England Nuclear). Triplicate
wells were assayed for cold trichloroacetic acid-precipitable, hot
perchloric acid-soluble radioactivity as described previously
(30)
. Cell counts were performed in parallel from
triplicate wells using a hemocytometer.
Acknowledgments
We gratefully acknowledge Daniel DiMaio, David Stern, and Jeff Settleman for essential reagents and Julia Schaefer and Daniela Drummond-Barbosa for subcloning neu* and E5, respectively, into the appropriate retroviral vectors. We also thank Daniel DiMaio, Judith Laffin, Tom Friedrich, John Lehman, Ying Zhang, and Jeff Banas for invaluable discussions and advice.
Footnotes
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.
1 Supported by National Cancer Institute Grant
1R29 CA73682-01A1 (to L. M. P.) and a Schaffer Fellowship from
Albany Medical College (to F. A. R.). 
2 To whom requests for reprints should be
addressed, at Center for Immunology and Microbial Disease, MC-151,
Albany Medical College, 47 New Scotland Avenue, Albany, NY 12208.
Phone: (518) 262-6285; Fax: (518) 262-5748; E-mail: pettil{at}mail.amc.edu
3 The abbreviations used are: BPV, bovine
papillomavirus; PDGF, platelet-derived growth factor; PDGFßR; PDGF
ß receptor; HDF, human diploid fibroblast; ln, natural logarithm;
FBS, fetal bovine serum; Rb, retinoblastoma protein; MAPK,
mitogen-activated protein kinase; PD, population doubling.
4 F. A. Ray, unpublished observations.
5 E. Okubo and F. A. Ray, unpublished
observations.
6 Y. Zhang and L. M. Petti, unpublished
observations.
Received for publication 3/17/00. Revision received 5/12/00. Accepted for publication 5/16/00.
References
This article has been cited by other articles:
 |
|
 |
|
  L. M. Petti, E. C. Ricciardi, H. J. Page, and K. A. Porter Transforming signals resulting from sustained activation of the PDGF{beta} receptor in mortal human fibroblasts J. Cell Sci., April 15, 2008; 121(8): 1172 - 1182. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Narechania, M. Terai, Z. Chen, R. DeSalle, and R. D. Burk Lack of the canonical pRB-binding domain in the E7 ORF of artiodactyl papillomaviruses is associated with the development of fibropapillomas J. Gen. Virol., May 1, 2004; 85(5): 1243 - 1250. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Zhang, J. M. Lehman, and L. M. Petti Apoptosis of Mortal Human Fibroblasts Transformed by the Bovine Papillomavirus E5 Oncoprotein Mol. Cancer Res., December 1, 2002; 1(2): 122 - 136. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. M. Nappi, J. A. Schaefer, and L. M. Petti Molecular Examination of the Transmembrane Requirements of the Platelet-derived Growth Factor beta Receptor for a Productive Interaction with the Bovine Papillomavirus E5 Oncoprotein J. Biol. Chem., November 27, 2002; 277(49): 47149 - 47159. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. M. Nappi and L. M. Petti Multiple Transmembrane Amino Acid Requirements Suggest a Highly Specific Interaction between the Bovine Papillomavirus E5 Oncoprotein and the Platelet-Derived Growth Factor Beta Receptor J. Virol., August 15, 2002; 76(16): 7976 - 7986. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Penington, I. Bryant, and D. J. Riese II Constitutively Active ErbB4 and ErbB2 Mutants Exhibit Distinct Biological Activities Cell Growth Differ., June 1, 2002; 13(6): 247 - 256. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cell Growth & Differentiation |
),
v-sis (•), or empty vector (control, *) were seeded
at equal cell densities into multiple dishes and then counted 48 h
later and every 24 h thereafter. For each time point, cells were
counted in triplicate. Growth curves were generated after plotting the
ln of the cell number versus the hours after plating.
The mean value of the ln of the cell number with the SD is shown.
, control cells; 
) or treated (
) with conditioned
medium from E5 HDFs that had just reached confluence. At the time of
treatment and various times thereafter, the cells were assayed for DNA
synthesis by measuring [3H]thymidine incorporation as
described in "Materials and Methods" or trypsinized and counted in
parallel. cpm/100,000 cells (top panels) and cell number
(bottom panels) were determined in triplicate. Mean
values and SD are shown.