Inhibition of transforming growth factor α (TGF-α)-mediated growth effects in ovarian cancer cell lines by a tyrosine kinase inhibitor ZM 252868

The modulating effects of the epidermal growth factor (EGF) receptor-specific tyrosine kinase inhibitor ZM 252868 on cell growth and signalling have been evaluated in four ovarian carcinoma cell lines PE01, PE04, SKOV-3 and PE01CDDP. Transforming growth factor α (TGF-α)-stimulated growth was completely inhibited by concentrations ≥ 0.3 μM in the PE01 and PE04 cell lines and by ≥ 0.1 μM in SKOV-3 cells. TGF-α inhibition of PE01CDDP growth was reversed by concentrations ≥ 0.1 μM ZM 252868. TGF-α-stimulated tyrosine phosphorylation of both the EGF receptor and c-erbB2 receptor in all four cell lines. The inhibitor ZM 252868, at concentrations ≥ 0.3 μM, completely inhibited TGF-α-stimulated tyrosine phosphorylation of the EGF receptor and reduced phosphorylation of the c-erbB2 protein. EGF-activated EGF receptor tyrosine kinase activity was completely inhibited by 3 μM ZM 252868 in PE01, SKOV-3 and PE01CDDP cells. These data indicate that the EGF receptor-targeted TK inhibitor ZM 252868 can inhibit growth of ovarian carcinoma cells in vitro consistent with inhibition of tyrosine phosphorylation at the EGF receptor. © 1999 Cancer Research Campaign


Growth assays
Log-phase PE01, PE01 CDDP , PE04 and SKOV-3 ovarian cancer cells were harvested by trypsinization and seeded in 24-well plates (Falcon) at a density of 2.5 × 10 4 cells per well, or in the case of the SKOV-3 cells at 1.25 × 10 4 cells per well, in quadruplicate in RPMI-1640 (Gibco BRL, Paisley, UK) containing 10% heatinactivated FCS and penicillin (100 U ml Ð1 ) and streptomycin (100 µg ml Ð1 ). Cells were maintained routinely at 37°C in a humidified atmosphere of 5% carbon dioxide in air. After 24 h, the media were removed and, after two phosphate-buffered saline (PBS) washes, replaced by phenol-red-free RPMI-1640 containing 5% double charcoal-stripped fetal calf serum (DCS FCS), penicillin (100 U ml Ð1 ), streptomycin (100 µg ml Ð1 ) and glutamine (2 mmol l Ð1 ). Phenol-red-free medium was used because the PE01 and PE04 are growth-stimulated by oestrogen, and the absence of the oestrogenic indicator provides a cleaner system in which to observe TGF-α modulations. After a further 24 h, media were removed and replenished by fresh phenol-red-free RPMI plus additives. Phenol-red-free medium was used because the PE01 and PE04 cell lines are growth stimulated by oestrogen and phenol red has oestrogenic activity. The tyrosine kinase inhibitor ZM 252868 was added 30 min before the addition of TGF-α (10 Ð10 M); this time point was designated day 0. Thereafter, fresh medium was added on day 2. Cells were harvested on day 5 and counted using a Coulter counter (Coulter Electronics, Luton, UK).

Phosphotyrosine Western blotting
PE01, PE01 CDDP , PE04 and SKOV-3 ovarian cancer cells were grown to 80% confluence in 25 cm 2 flasks (Falcon) in the presence of RPMI-1640 containing 10% FCS. After two washes with PBS, phenol-red-free RPMI-1640 containing 5% DCS FCS in the presence or absence of ZM 252868 (0.03Ð3.0 µM) was added, and the cells were incubated overnight at 37°C in a humidified atmosphere of 5% carbon dioxide in air. Cells were then incubated for 30 min in fresh RPMI ± ZM 252868. RPMI ± ZM 252868 was replenished and incubated for a further 5 min in the presence or absence of TGF-α (10 Ð10 M). Cells were then washed twice in ice cold PBS, lysed, detached from the flasks and spun at 15 000 r.p.m. at 4°C. Protein content of the resulting supernatant was determined by Bradford assay (Bradford, 1976) and stored at Ð80°C before phosphotyrosine Western blotting. Protein samples were denatured at 95°C for 5 min in buffer containing sodium dodecyl sulphate (SDS) and mercaptoethanol, then 75 µg of protein was loaded into each lane on a 7.5% polyacrylamide gel. After polyacrylamide gel electrophoresis, proteins were transferred to Immobilon-P polyvinylidene fluoride (PVDF) membrane using a wet transfer system (BioRad Trans Blot Cell). The transfer was carried out overnight at 4°C with 30 V followed by 60 V for 2 h. Proteins were detected using a specific anti-phosphotyrosine mouse monoclonal antibody, p-Tyr (PY20) (Santa Cruz Biotechnology), in conjunction with a chemiluminescence Western Blotting Kit (Boehringer Mannheim). Membranes were blocked in 1% blocking reagent (supplied with kit and containing 10% purified casein protein in maleic acid) diluted in TBS (Tris-buffered saline pH 7.5) for 1 h at room temperature. The membrane was incubated with primary antibody (PY20) diluted to 1 µg ml Ð1 in 0.5% blocking reagent in TBS overnight at 4°C, washed three times for 5 min in TBS-T (TBS with 0.1% Tween 20), twice for 5 min in TBS, incubated twice for 5 min in 0.5% blocking agent in TBS and treated with secondary antibody for 1 h at room temperature. The secondary antibody (anti-mouse IgG-POD/anti-rabbit IgG-POD) was diluted in 0.5% blocking solution in TBS to 160 mU ml Ð1 . Finally, the membrane was washed once again, three 5-min washes in TBS-T, then three 5-min washes in TBS. After incubation in luminescence substrate solution, light emission was detected on radiographic film.

Epidermal growth factor receptor tyrosine kinase enzyme assay
Epidermal growth factor receptor tyrosine kinase enzyme activity was measured in the ovarian cancer cell lines using a Biotrak assay (Amersham, UK). Briefly, membrane preparations were obtained from cells grown to 70% confluence in 175 cm 2 flasks. Lysis buffer (50 mM Tris, 1 mM magnesium chloride, 2 mM EDTA, 20 µg ml Ð1 soybean trypsin inhibitor, 50 µg ml Ð1 PMSF) was added (7 ml per 175 cm 2 flask) to cell monolayers. Lysates were homogenized and centrifuged at 3000 r.p.m. for 10 min and supernatants then centrifuged at 100 000 g for 30 min to obtain membrane preparations. These were resuspended in solubilization buffer [50 mM Hepes, 20% glycerol, 1% Triton-X, 0.1% bovine serum albumin (BSA), 0.05% sodium azide] and stored at Ð80°C until assayed.
In the presence of enzyme sample and EGF, the 32 P counted on the papers is the sum of non-specific [ 32 P]ATP binding, specific binding of phosphorylated peptide and binding of phosphorylated proteins in the cell extract (A). In the absence of EGF and presence of enzyme, the 32 P counted on the papers is the sum of non-specific [ 32 P]ATP binding and non-EGF-dependent tyrosine kinase phosphorylation of the peptide and cell extract proteins (B). EGF-dependent tyrosine kinase activity is, therefore, obtained from (A Ð B).

Determination of EGF-R, c-erbB2 and c-erbB3 by immunofluorescence
The presence of EGF receptor, c-erbB2 and c-erbB3 proteins was identified on PE01, PE01 CDDP , PE04 and SKOV-3 cells by immunofluorescence using a flow cytometer. The following antibodies were used: EGF receptor, clone EGFR1 (ICRF, Clare Hall, London UK); c-erbB2, clone CB11 (Novocastra); and c-erbB3, clone RTJ1 (Novocastra). Cells were harvested by trypsinization (in pilot experiments found to be less damaging than cell scraping), washed in cold PBS containing 5% FCS, and aliquots of approximately 10 6 cells were then incubated for 60 min with antibody. For c-erbB2 and c-erbB3 staining, 1% saponin (BDH, Poole, Dorset, UK) was added to the cells before antibody addition (Brotherick et al, 1995). Cells were then washed in PBS/FCS and incubated with sheep anti-mouse fluorescein isothiocyanate (FITC, 1:20) for 60 min and washed twice with PBS/FCS. Cells were resuspended in PBS and analysed on the FACScan flow cytometer.

Effects of tyrosine kinase inhibitor on TGF-α-modulated growth of ovarian cancer cells
The PE01, PE04, SKOV-3 and PE01 CDDP cell lines were treated with ZM 252868 for 5 days in the absence or presence of TGF-α (10 Ð10 M) (Figures 1 and 2). In the absence of TGF-α, ZM 252868 inhibited growth of PE01 cells at concentrations equal to or greater than 0.03 µM. Addition of TGF-α produced an approximately 150% increase in control cell number; this increase was abolished by ZM 252868 at concentrations equal to or greater than 0.3 µM ( Figure 1A). For PE04 cells, ZM 252868 inhibited growth at 0.3 µM ( Figure 1B). Addition of TGF-α produced a 100% increase in cell number and this stimulation was partially reversed at 0.03 µM and completely blocked at 0.3 µM ( Figure 1B). For SKOV-3 cells, ZM 252868 inhibited growth at 0.1 µM. Addition of TGF-α produced an approximately 130% increase in control cell number; this increase was abolished by ZM 252868 at concentrations equal to or greater than 0.1 µM ( Figure 1C).
Addition of TGF-α to PE01 CDDP cells produces growth inhibition ( Figure 2). ZM252868 (0.03Ð3.0 µM) reversed the TGF-αinduced growth inhibition, with a total reversal of this effect observed at concentrations ³ 0.1 µM. Basal cell growth was also significantly increased (P < 0.001) after the addition of ZM 252868 at 0.03 µM, but was not significantly different from control at concentrations ³ 0.1 µM (Figure 2).

Activation of EGF receptor and c-erbB2 by TGF-α
Western blotting with an anti-phosphotyrosine antibody (PY20) showed that, after exposure to TGF-α, both the EGF receptor and c-erbB2 protein were phosphorylated on tyrosine residues in PE01 cells. The identity of bands was confirmed in parallel experiments using antibodies specific for the EGF receptor and c-erbB2 (data not shown). The time course of activation by TGF-α is illustrated in Figure 3 for the PE01 cell line and occurred within 1 min of addition of the growth factor ( Figure 3A). Investigation of the concentration range 10 Ð12 Ð10 Ð8 M indicated that phosphorylation increased with increasing concentrations of TGF-α after a 5-min exposure ( Figure 3B). This latter concentration (10 Ð8 M) was selected for experiments with ZM 252868.

Inhibition of tyrosine phosphorylation of the EGF receptor by ZM 252868
The cell lines were treated with ZM 252968 for 5 min in the presence or absence of 10 Ð8 M TGF-α. TGF-α increased tyrosine  (Figure 4). At concentrations of 0.3, 1 and 3 µM, the EGF receptor phosphorylation was completely blocked. The c-erbB2 phosphorylation was also decreased, but to a lesser degree in all four lines (Figure 4). In the absence of TGF-α, only c-erbB2 phosphorylation was observed in untreated cells (Figure 4). This was also decreased after addition of ZM 252868, but only at higher concentrations of inhibitor.

Epidermal growth factor receptor tyrosine kinase inhibitory activity of ZM 252868
The EGF receptor tyrosine kinase enzyme activity was investigated by an assay which measures the transfer of the γ-phosphate of adenosine 5′ triphosphate to the tyrosine group on a peptide which is specific for EGF receptor tyrosine kinase (Amersham). EGF, which is known to be equipotent with TGF-α in these systems, was used to activate the receptor. Addition of EGF resulted in a 22% increase in phosphorylation in PE01 cells, a 41% increase in phosphorylation in PE01 CDDP cells and a 49% increase in phosphorylation in SKOV-3 cells (Table 1). For PE04 cells, a 9% increase in phosphorylation was observed, but this did not achieve statistical significance. These increases were completely reversed by the addition of ZM 252868 (3 µM) (Table 1). In a subsequent experiment with PE01 cells, the 33% increase in phosphorylation produced by EGF was blocked by ZM 252868 at concentrations of 0.03, 0.3 and 3 µM ( Table 2). The basal level of EGF receptor activity was lower in this experiment, reflecting an unknown variable, however intra-assay changes in activity after addition of ligand and blockade by inhibitor were reproducible.  Figure 4 Western blotting of phosphotyrosine in ZM 252868-treated ovarian cancer cells. Blots are shown for the PE01, PE04, SKOV-3 and PE01 CDDP cell lines. The upper signal shown corresponds to the c-erbB2 protein and the lower signal to the EGF receptor. In untreated cells, only c-erbB2 is tyrosine phosphorylated, but after addition of TGF-α phosphotyrosine signals are seen on the EGF receptor and are increased on c-erbB2 compared with no treatment. ZM 252868 was added 30 min before TGF-α addition. Lysates were collected after 5 min exposure to TGF-α. At the concentrations shown, ZM 252868 eliminated the EGF phosphotyrosine signal associated with the EGF receptor and reduced the signal associated with c-erbB2 Values shown are means ± standard errors of 4-6 replicates in a typical experiment. *Significantly different from control P < 0.05 (Student's t-test).

Expression of erbB receptors in ovarian cancer cells
Because the activated EGF receptor has the potential to heterodimerize with other members of the c-erbB family, it was of interest to know the relative levels of expression of these proteins in the cell lines under study. The relative expression of EGF receptors, c-erbB2 and c-erbB3 in PE01, PE04, SKOV-3 and PE01 CDDP was measured by immunofluorescence on a FACscan, and representative profiles are shown in Figure 5 and mean values obtained from three independent experiments are shown in Figure 6. The SKOV-3 cell line possessed the greatest number of EGF receptors and PE04 the least. SKOV-3 cells also had a two-to threefold greater expression of c-erbB2 than the other three line, whereas all four lines had comparable levels of c-erbB3 ( Figure 6).

DISCUSSION
ZM 252868 (also known as PD 153035) is a potent inhibitor of the EGF receptor, with an IC 50 value of 29 pM against the purified tyrosine kinase in A431 cells (Fry et al, 1994). Against other human tumour cell lines, including breast, prostatic, cervical and colon cancer cell lines, the compound has been reported to inhibit growth at concentrations ³ 75 nM (Bos et al, 1996;Jones et al, 1997), and this is similar to the observations in these ovarian cancer cell lines wherein 100 nM inhibited the TGF-α-stimulated growth of PE01, PE04 and SKOV-3 cells. All three ovarian cancer lines have comparable levels of EGF receptor and ZM 252868 showed a similar degree of potency, with complete inhibition of growth being obtained at concentrations between 0.1 and 0.3 µM. At these concentrations, ZM 252868 inhibited TGF-α-activated tyrosine phosphorylation at the EGF receptor and partially inhibited phosphorylation at c-erbB2 consistent with its proposed mechanism of action. Because the activated EGF receptor interacts with and transphosphorylates the c-erbB2 receptor, it is not surprising that phosphorylation levels are reduced at this protein as well as at the EGF receptor, though direct effects by ZM 252868 on c-erbB2 cannot be ruled out and this is in line with previous studies (Bos et al, 1996). Although we did not investigate the specificity of ZM 252868 for the EGF receptor in this study, previous investigations have demonstrated its ability to specifically block TGF-α or EGFstimulated growth as opposed to platelet-derived growth factor or insulin-like growth factor-stimulated growth (Fry et al, 1994b;Jones et al, 1997). Data obtained from the EGF receptor tyrosine kinase activity assay confirmed that inhibition of kinase activity occurred at growth inhibitory concentrations in PE01 cells.
Although it cannot always be assumed that EGF receptor activation necessarily leads to a mitogenic response, the data obtained in the PE01, PE04 and SKOV-3 lines suggests that it can be and that the inhibitor blocks this action. However, the data obtained using the PE01 CDDP model indicate that TGF-α activation may also be associated with growth inhibitory effects. TGF-α treatment produced growth inhibition in this cell line (Simpson et al, 1998), and ZM 252868 was able to antagonize this growth effect with associated reduction in EGF receptor phosphotyrosine and kinase activity. This cell line was derived from the PE01 cell line after exposure to cisplatin (Beattie et al, 1993) and, as demonstrated here, the changed growth response is not simply a result of altered erbB receptor numbers. We are currently investigating downstream responses that can be specifically associated with either growth stimulation or growth inhibition because these may help indicate the type of growth response elicited by TGF-α.
In the absence of TGF-α, ZM 252868 inhibited the growth of PE01, PE04 and SKOV-3 cells and stimulated growth of PE01 CDDP cells compared with no treatment. This would be consistent with inhibition of the action of autocrine production of TGF-α or other EGF receptor-activating ligands which are stimulating the EGF receptor in the absence of added factors. Evidence to support such a process has been obtained from other ovarian cancer models, wherein antibodies to the EGF receptor or to TGF-α have been shown to block growth consistent with TGF-α/EGF receptor autocrine pathways being functional (Morishige et al, 1991).
Blockade of signalling via the EGF receptor appears a promising growth inhibitory strategy. Using these cell lines, we have recently investigated other approaches to neutralizing this receptor. Antibody blockade of the EGF receptor produces similar growth-reversing effects in the PE01 and PE01 CDDP cell lines (Simpson et al, 1998), whereas antisense knockout of EGF receptor mRNA in PE01 cells also produces growth inhibition . Relatively small structures such as ZM 252868 may, however, have improved pharmacokinetic properties compared with antibody or antisense delivery. Although this compound has demonstrated transient reduction of EGF receptor tyrosine phosphorylation in A431 xenografts, this was insufficient to produce growth inhibition in vivo (Kunkel et al, 1996). Closely related analogues from this class of agents with improved pharmacokinetic properties and which are active against in vivo models have now been developed (Woodburn et al, , 1997, and an analogue is currently undergoing phase I studies in the UK. These data are the first to report that inhibitors of this class could have activity in ovarian cancer systems, and would indicate that this disease would be a suitable system for clinical investigations. It will be important to define tumours in the clinical setting whose growth is dependent on the EGF receptor and which are being driven by activating ligands such as TGF-α and EGF. Tyrosine phosphorylation of the EGF receptor in clinical specimens of ovarian cancer can be readily identified, and it seems likely that those tumours in which activation is found represent the target population for this inhibitor.

ACKNOWLEDGEMENT
We are grateful to Dr Alan Wakeling, Zeneca Pharmaceuticals, Macclesfield for supplies of ZM252868. c-ErbB receptor expression in ovarian cancer cell lines. Cells were incubated with anti-erbB antibodies (EGF receptor, clone EGFR1; c-erbB2, clone CB11; anti-erbB3, clone RTJ1) followed by sheep anti-mouse fluorescein isothiocyanate. Cells were then analysed on a FACscan. The ratio of median fluorescence in the presence of antibody compared with that obtained in the absence of antibody was determined. The means of at least three independent experiments ± standard deviation are shown