Dominant negative knockout of p53 abolishes ErbB2-dependent apoptosis and permits growth acceleration in human breast cancer cells

We previously reported that the ErbB2 oncoprotein prolongs and amplifies growth factor signalling by impairing ligand-dependent downregulation of hetero-oligomerised epidermal growth factor receptors. Here we show that treatment of A431 cells with different epidermal growth factor receptor ligands can cause growth inhibition to an extent paralleling ErbB2 tyrosine phosphorylation. To determine whether such growth inhibition signifies an interaction between the cell cycle machinery and ErbB2-dependent alterations of cell signalling kinetics, we used MCF7 breast cancer cells (which express wild-type p53) to create transient and stable ErbB2 transfectants (MCF7-B2). Compared with parental cells, MCF7-B2 cells are characterised by upregulation of p53, p21WAF and Myc, downregulation of Bcl2, and apoptosis. In contrast, MCF7-B2 cells co-transfected with dominant negative p53 (MCF7-B2/Δp53) exhibit reduced apoptosis and enhanced growth relative to both parental MCF7-B2 and control cells. These data imply that wild-type p53 limits survival of ErbB2-overexpressing breast cancer cells, and suggest that signals of varying length and/or intensity may evoke different cell outcomes depending upon the integrity of cell cycle control genes. We submit that acquisition of cell cycle control defects may play a permissive role in ErbB2 upregulation, and that the ErbB2 overexpression phenotype may in turn select for the survival of cells with p53 mutations or other tumour suppressor gene defects. British Journal of Cancer (2002) 86, 1104–1109. DOI: 10.1038/sj/bjc/6600219 www.bjcancer.com © 2002 Cancer Research UK

We previously reported that growth arrest of 3T3 cells is associated with catalytic activation of ErbB2 (Epstein et al, 1990), and more recently demonstrated that ErbB2 lengthens and intensifies mitogenic signalling by impairing ligand-dependent EGFR downregulation (Huang et al, 1999). In addition, we have shown that the functionally distinct EGFR ligands, EGF and transforming growth factor-alpha (TGFa), exert differing effects on EGFR downregulation and hence on the duration of ErbB2 co-activation: high concentrations of EGF initially cause prolonged EGFR activation associated with ErbB2 heterodimerisation, followed by eventual EGFR downregulation and signal cessation; whereas TGFa fails to downregulate EGFR, leading to sustained signalling (Gulliford et al, 1997;Ouyang et al, 1999a). The possibility is thus raised that ErbB2 could mimic the tumorigenic effects of TGFa in cancer cells by its similar ability to prolong EGFR signalling.
The above-mentioned differential induction of growth stimulation or inhibition by EGFR (Filmus et al, 1985;Polet, 1990;Armstrong et al, 1994;Gulli et al, 1996) and ErbB2 (Tagliabue et al, 1991;Harris et al, 1995;Kita et al, 1996) strongly suggests an interaction between downstream signal duration (e.g. of MAP kinase) and cell cycle control proteins (Traverse et al, 1992;Marshall, 1995). To address the possibility that ErbB2-dependent changes in signal duration may contribute to such differences in cell fate, it is necessary to create cell systems in which the effects of ErbB2 expression can be correlated with the function or dysfunction of a given cell cycle regulatory molecule. Here we show that the effects of ErbB2 on cell signalling kinetics are selectively associated with induction of apoptosis in oestrogenresponsive MCF7 human breast cancer cells -which, like most hormone-sensitive cancers (Caleffi et al, 1994;Elledge et al, 1995;Berns et al, 2000), express wild-type p53 (Casey et al, 1991;Balcer-Kubiczek et al, 1995;Furuwatari et al, 1998) but normally do not overexpress ErbB2 (Wright et al, 1997;Ferrero-Pous et al, 2000;Pinto et al, 2001). Dominant negative knockout of p53 converts growth inhibition to growth enhancement in these ErbB2-transfected cells, suggesting that a p53 mutational pathway could favour selection for ErbB2 gene amplification during tumour progression.

MATERIALS AND METHODS
Cell lines, reagents, antibodies, and immunoblotting MCF7 and A431 cells were obtained from the American Type Culture Collection (Rockville, MD, USA). Synthetic human EGF and TGFa were purchased from Sigma. Activation-state-specific EGFR antibodies, and antibodies to p53, Myc, Bcl2 and p21 WAF , were purchased from Cambridge BioScience (Cambridge, UK). Polyclonal antibodies to Tyr 1248 -and Tyr 1222 -phosphorylated ErbB2 were developed and validated for receptor-specificity as described previously (Epstein et al, 1992;Ouyang et al, 1998). For immunoblotting studies, cells were lysed as previously described (Gulliford et al, 1997): protein lysates were immediately boiled for 5 min in sample buffer (6.7% sodium dodecyl sulfate, 30% glycerol, 62.5 mM Tris base pH 6.8, 0.01% bromophenol blue) then loaded onto a 7.5% SDS-polyacrylamide gel. Samples were electrophoresed and transblotted onto nitrocellulose as described (Towbin et al, 1979).

Growth curves and apoptosis assays
Cell growth was measured using a multiwell colorimetric assay based on sulphorhodamine B (SRB) spectrophotometric detection. Confirmation and quantification of morphologic apoptosis was performed using a Tdt-mediated dUTP nick-end labelling (TUNEL) kit to directly detect DNA fragmentation in situ. Briefly, cells were plated and grown on glass slides, treated with ligand for the required period, then fixed in 4% paraformaldehyde for 30 min at room temperature. The slides were washed with PBS three times, after which the cells were permeabilised with 0.1% Triton-X-100 in 0.1% sodium citrate for 10 min. After washing, the cells were covered in 50 ml of equilibration solution for 10 min, then covered with 50 ml of labelling solution (Biovation) and incubated at 378C for 1 h while light-protected. The slides were then washed, covered in 10 ml counterstain for 10 min, and analysed using fluorescence microscopy.

Cell transfection
For calcium phosphate transfection, cells were seeded in 90 mm diameter cell culture dishes at 5610 5 cells ml 71 24 h before the transfection. One plate was required for each transfection experiment; the monolayer normally grew to 80% confluence by the following day, and the medium was changed 3 h before the transfection. Two sterile microfuge tubes were labelled for each transfection experiment: to one tube was added 500 ml of 26BBS (pH 6.95) and to the other tube was added 125 ml of 1 M CaCl 2 , 10-20 mg of recombinant plasmid DNA which contained the relevant cDNA; distilled H 2 O was added to give a final volume of 500 ml. This was added to equal the volume of 26concentrated BBS using a sterile Pasteur pipette. At the same time, filtered air was passed through the 26BBS buffer (pH 6.95) with a second Pasteur pipette, and the DNA mixture was then incubated at room temperature for 20 min to allow precipitation. The DNA/CaPO 4 precipitate was mixed by inverting the tube, and was added directly to a 10 ml cultured cell dish dropwise with gentle shaking, and the cell culture incubated at 378C with 3% CO 2 overnight followed by washing with PBS and re-culturing in fresh medium at 378C with 5% CO 2 .

Constructs and selection procedures
The well-characterised temperature-sensitive dominant negative p53 construct (Kuerbitz et al, 1992;Slichenmeyer et al, 1993;Zhang et al, 1994;Vasey et al, 1996) was kindly provided by Dr B Vogelstein (Baker et al, 1990). For selection, transfected cells were plated at 5610 4 cells/9 cm tissue culture dish with relevant reagents: dominant negative p53 was selected with neomycin. The wild-type ErbB2 construct, which is under the control of the Moloney murine leukaemia virus LTR and contains the Ecogpt selectable marker from E. coli (Di Fiore et al, 1987), was selected with HAT (hypoxanthine, aminopterin and thymidine) as described by Mulligan and Berg (1981). For double transfection a pool of six p53 dominant negative clones (Dp53) or p53 empty vector clones were transfected with either ErbB2 or ErbB2 empty vector, and selected with HAT medium for at least 6 weeks. Resistant colonies were cloned and a pool of six clones was cultured with HAT medium to amplify the cell number. For analysis, the cells were cultured in normal medium for at least 2 weeks before the experiments were performed. For morphologic analysis, cells were grown in plastic 8chamber containers (LabTek; Gibco) and the monolayers photographed using a Zeiss microscope. Growth experiments were carried out in 96-well plates using quantification of Hoechst dye immunofluorescence in six matched samples following 3 days growth to assess cumulative DNA content.

RESULTS
Consistent with earlier reports (Gill and Lazar, 1981;Polet, 1990), ligand stimulation experiments confirm EGF-dependent growth inhibition of sparsely-plated A431 cells ( Figure 1A, upper panel). The extent of growth inhibition correlates with the intensity of equimolar ligand-dependent ErbB2 tyrosine phosphorylation as detected by site-specific phosphoantibodies  which confirm greater ErbB2 tyrosine phosphorylation following EGF stimulation ( Figure 1A, lower panel). As reported previously, this initial difference in ligand-dependent signal intensity is maintained and further exaggerated over the subsequent 12 h (Ouyang et al, 1999a). Correlation of light microscopy with TUNEL assay indicates that the growth-inhibitory effects of EGF in this context are associated with increased apoptosis ( Figure 1B).
The foregoing data do not distinguish whether the observed growth inhibition is induced by ligand-dependent ErbB2 co-activation per se or, alternatively, by the downstream consequences of growth factor signal prolongation induced by ligand-dependent ErbB2 heterodimerisation. However, since our previous work documented a marked prolongation of EGFR signalling by ErbB2 expression (Huang et al, 1999), we elected to test the latter hypothesis by creating ErbB2 transfectants in cell lines differing solely in terms of cell cycle control functionality. To this end, MCF7 human breast cancer cells known to express both copies of the wild-type p53 gene (Casey et al, 1991;Balcer-Kubiczek et al, 1995;Furuwatari et al, 1998) were transiently transfected with ErbB2. As shown in Figure 2A, ErbB2 expression in these cells induces increased immunoreactivity of both activated ErbB2 and EGFR, consistent with previous studies (Huang et al, , 1999, while also inducing increased expression of p53, p21 WAF and Myc. Of note, ErbB2 expression is associated with reduced Bcl2 expression -an effect reported previously following primary overexpression of p53 (Haldar et al, 1994). These effects on protein expression are accompanied by morphologic changes (membrane blebbing, chromatin condensation) typical of apoptosis in ErbB2-transfected, but not vector control, cells ( Figure 2B). These ErbB2-dependent changes in protein expression and morphology directly implicate ErbB2 in the activation of an apoptotic pathway.
To clarify whether the apoptosis-triggering effect of ErbB2 might be at least partly related to its effects on signalling kinetics (i.e. as opposed to an exclusive cell-killing effect of ErbB2 kinase activity), stable MCF7 cell transfectants were created using either the wildtype ErbB2 gene, the dominant-negative p53 mutant gene, or both. As in the ErbB2 transient transfectants, stable overexpression of ErbB2 selectively induces endogenous (wild-type) p53 protein overexpression ( Figure 3A, upper panel, left three lanes); as expected, In the upper figure, cells were plated at 1.5610 4 ml 71 seeding density and stimulated for 6 days with the respective ligand (EGF, solid columns; TGFa, open columns) prior to counting using a sulphorhodamine-based assay as described. Nanomolar ligand concentrations are represented on the abscissa. The results are expressed as a percentage change relative to control cell growth; error bars are based on six measurements. The lower figure shows the corresponding short-term effects of EGF and TGFa on ErbB2 Y 1222 phosphorylation: cells were treated for 5 min with EGF or TGFa at the indicated nanomolar concentration prior to lysis, electrophoresis and immunoblotting using aPY 1222 . The bands were visualised using ECL. (B) Visualisation of cell death by light microscopy and TUNEL assay (see Materials and Methods) associated with ligand treatment. Twenty-four hours following attachment, cells were treated with the respective ligands (2 nM) in serum-free medium. Typical low-power views of triplicate plates are shown after 48 h treatment using light microscopy (above) and fluorescence microscopy (below). 1248, ErbB2 tyrosine-phosphorylated at position 1248, detected by the activation-specific aPY 1248 antibody . (B) Effects of ErbB2 expression on morphology and apoptosis of MCF7 human breast cancer cells assessed using light microscopy. ErbB2 transfectants (at right) were generated using a standard calcium phosphate transfection procedure followed by neomycin selection. Mock transfectants containing empty plasmids are shown at left.

Molecular and Cellular Pathology
dominant-negative mutant p53 (Dp53) cells grossly overexpress immunoreactive p53 ( Figure 3A, upper panel, right three lanes). Irradiated control and ErbB2-transfected MCF7 cells exhibit a normal increase in p21 WAF expression following X-irradiation ( Figure 3B, lower panel, left 6 lanes). In contrast, MCF7-Dp53 cells sustain no immunodetectable rise in p21 WAF levels ( Figure 3B, lower panel, right 6 lanes), thus validating the functionality of the dominant-negative p53 construct used in these experiments. Of note, p21 WAF was not detectably overexpressed in stable ErbB2-overexpressing cells ( Figure 3B) unlike in transient transfectants (Figure 2A), raising the possibility that prolonged ErbB2 overexpression induces clonal selection. The four transfectant cell lines of interest -parental MCF7, MCF7-B2, MCF7-Dp53, and MCF7-B2/Dp53 -were then compared with respect to morphology and growth. Unlike parental MCF7 cells which adopt a spread-out cell appearance suggesting density-dependent growth inhibition ( Figure 4A, upper left), all of the other transfectants exhibit a crowded morphology. MCF7-B2 cells also exhibit striking apoptosis ( Figure 4A, lower left), however, a feature which is absent from both the MCF7-Dp53 and MCF7-B2/Dp53 cells ( Figure 4A, right upper and lower panels, respectively). Cell growth as measured by Coulter counting was increased in MCF7-Dp53 cells and reduced in MCF7-B2 cells relative to parental cell growth: MCF7-B2/Dp53 cells exhibit more rapid growth than parental cells, though slower than MCF7-Dp53 cells ( Figure 4B). Given the foregoing results, these data indicate that the observed ErbB2-dependent effects on cell fate vary with the functional status of p53, suggesting in turn that p53 may act as a sensor for ErbB2-induced changes in cell signalling kinetics.

DISCUSSION
We previously showed that ErbB2 expression causes constitutive EGF signalling by retarding downregulation of hetero-oligomerised EGFR (Huang et al, 1999). This effect most likely relates to the absence of motifs in the ErbB2 C-terminal tail for receptor internalisation and degradation (Sorkin et al, 1993;Baulida et al, 1996). Since human tumours exclusively overexpress the wild-type ErbB2 rather than the transforming point mutant (Lemoine et al, 1990), a reasonable hypothesis is that tumour cells acquire a growth advantage from wild-type ErbB2 overexpression, but that this phenotype does not represent the primary transforming event -implying the co-existence, that is, of at least one other molecular defect within the tumour cells. This hypothesis is consistent with numerous reports linking tumour cell ErbB2 overexpression and p53 dysfunction (Horak et al, 1991;Mehta et al, 1995;Li et al, 1997) and identifying poor-prognosis clinical subgroups based on concurrence of these phenotypes (Tsuda et al, 1998). Moreover, our recent documentation of differential survival outcomes in ErbB2-overexpressing breast cancers associated with different phos-  phorylation patterns (Ouyang et al, 1999b(Ouyang et al, , 2001 supports the notion of multiple signalling pathways governing tumour growth phenotypes.

Molecular and Cellular Pathology
Given that the p53 checkpoint prevents cell-cycle progression when activated (Casey et al, 1991;Yin et al, 1992;Wyllie et al, 1995) and that the duration of growth factor signalling influences whether cells proliferate or arrest (Traverse et al, 1992;Marshall, 1995), the present study suggests a model of cell signal sensing which is differentially perturbed by ErbB2 depending upon the functional p53 status. Other studies have concluded that the main in vitro and in vivo consequences of p53 mutation on cell growth relate to enhanced proliferation rather than to reduced apoptosis (Nikiforov et al, 1996;Tyner et al, 1999). Our data suggest a more complex interpretation of p53 function as a co-variable within the cell growth machinery; this is consistent with the surprising finding in human tumours that p53 mutation is often associated with increased, rather than decreased, apoptotic indices (van Slooten et al, 1999). In the context of tumour progression, it is important to note that apoptosis could represent a mechanism of clonal selection for more aggressive cell lineages, rather than simply indicating a benign tumour-suppressive function.
Reductions in mitogenic signal intensity may normally cause cells to arrest and/or differentiate, whereas signal prolongation may trigger differentiation or death (Traverse et al, 1992;Dolmetsch et al, 1997). According to this paradigm, apoptosis may be inducible by forced cell cycle progression in the presence of activated checkpoints (Polet, 1990). Abrogation of p53 function by mutation could thus prevent cells from sensing an abnormally prolonged signal, leading to loss of growth arrest, reduced apoptosis and differentiation, and consequent outgrowth of less differentiated cells. In contrast, ErbB2-dependent impairment of EGFR downregulation both prolongs and intensifies growth factor signalling (Huang et al, 1999), an outcome associated with the increased apoptosis reported here. Such an effect of ErbB2 might be expected to be short-lived, given that selection for apoptotic resistance should be rapid (Balcer-Kubiczek et al, 1995). Acquisition of a p53 defect in this context would cause mutant cells to 'perceive' mitogenic signals as short despite ErbB2-dependent signal prolongation -leading to apoptotic resistance, dedifferentiation and clonal outgrowth.
Human tumours could thus evolve from an interplay between progressive ErbB2 overexpression and acquisition of cell-cycle control defects including, though not necessarily limited to, p53 mutations. We therefore submit that human tissues with cell-cycle control defects (De Cremoux et al, 1999;Prevo et al, 1999) may gain a growth advantage by prolonging and intensifying ambient growth factor signals via ErbB2 upregulation, and that tumour cells overexpressing ErbB2 may in turn clonally select for cell-cycle checkpoint loss (Li et al, 1997).