Introduction

The epidermal growth factor (EGF) receptor (also known as c-erbB1/HER1) and the product of the neu oncogene (also known as c-erbB2/HER2) are the members of the EGF receptor superfamily. Clinical studies indicate that overexpression of EGF receptors or HER2, which occurs commonly in human cancer, often correlates with poorer prognosis (Slamon et al., 1987; 1989; Harris et al., 1989). These observations have stimulated preclinical investigations targeting on inhibiting the function of human EGF receptors or HER2 as novel cancer therapies (Fan and Mendelsohn, 1998; Mendelsohn, 1990; Lewis et al., 1993; Divgi et al., 1991; Baselga et al., 1996; Pegram et al., 1995).

Both EGF receptors and HER2 are transmembrane proteins consisting of an extracellular binding domain, a transmembrane fragment and an intracellular domain with tyrosine kinase activity. In spite of extensive sequence homology between EGF receptors and HER2, the ligands for EGF receptors, EGF and transforming growth factor-α (TGF-α), do not bind to HER2 (Ullrich and Schlessinger, 1990). No direct ligand for HER2 has been identified yet. However, HER2 can be transactivated by EGF through heterodimerization with EGF receptors (Goldman et al., 1990; Wada et al., 1990) or by heregulin through heterodimerization with HER-3 or HER-4 receptors (Sliwkowski et al., 1994; Plowman et al., 1993; Peles et al., 1993). The activation of EGF receptors or HER2 results in a cascade of down-stream substrate activation, leading to cell proliferation. Blockade of either receptor results in growth inhibition (Kawamato et al., 1983; Lewis et al., 1993). Many steps in receptor-mediated signal transduction pathways are shared by EGF receptors and HER2 (Earp et al., 1995). It is possible that when one receptor is blocked, the other may provide `cover' if the appropriate ligand is present. The convergence of down-stream pathways mediated by these two receptors presents new opportunities for improving the efficacy of growth inhibition achieved by dual receptor blockade.

C225 is a human-mouse chimeric anti-EGF receptor mAb derived murine mAb 225 (Goldstein et al., 1995). The chimeric mAb 225 fully retains the activity of murine mAb 225 in competing with EGF for receptor binding and produces a similar, or even improved, spectrum of anti-tumor activities on a variety of cultured and xenografted human cancer cell lines (Goldstein et al., 1995; Prewett et al., 1996). C225 can repeatedly be administered to patients for up to 3 months without eliciting a host anti-antibody response, and is currently being investigated in several clinical phase I trials (Bos et al., 1996; Falcey et al., 1997). Anti-HER2 mAb 4D5 (rhuMAb HER2 or Herceptin) is the humanized version of murine 4D5, which is an mAb directed against the extracellular domain of HER2 and has been shown to inhibit the proliferation of several cancer cell lines stimulated by HER2 signaling (Lewis et al., 1993). Clinical trials with humanized 4D5 have shown no toxicities after weekly administration of the mAb for many months (Baselga et al., 1996; Pegram et al., 1995). In a clinical phase II trial, 10% of patients with advanced breast cancer showed objective partial responses to mAb 4D5 (Baselga et al., 1996). Recent phase III trial with mAb 4D5 as a single agent revealed a 16% (34 out of 213 patients) overall response rate including eight patients experiencing a complete response and 26 patients experiencing a partial response (Cobleigh et al., 1998). When treated with 4D5 in combination with chemotherapy, approximately 28% of the patients did not show evidence of tumor progression at one year compared to 14% of the patients treated with chemotherapy alone (Slamon et al., 1998).

Simultaneous blockade of these two related receptors presents an interesting approach to cancer therapy. In the present study, we examined the effect of mAb C225 and 4D5, either alone or in combination, on the proliferation of OVCA420 human ovarian cancer cells. We demonstrate that the proliferation of OVCA420 cells can be inhibited by mAb C225 and by mAb 4D5. Inhibition by either mAb is accompanied by an increase in G1 phase cells, increased levels of cyclin-dependent kinase (CDK) inhibitor p27Kip1, increased association of p27Kip1 with CDKs, and decreased activities of CDKs. Concurrent treatment of the cells with the two mAbs results in augmentation of inhibition. The enhanced inhibitory effect on cell proliferation is accompanied by higher levels of p27Kip1 and greater inhibition of CDK activities. Our data suggest the potential fruitful cooperation of anti-EGF receptor mAb C225 and anti-HER2 mAb 4D5, two mAbs that are already showing evidence of efficacy in clinical trials.

Results

Growth inhibition of OVCA420 cells by mAb C225, mAb 4D5 and the combination

Exposure of OVCA420 cells to either mAb alone for 7 days resulted in 55% growth inhibition by mAb C225 and 35% growth inhibition by mAb 4D5, respectively, compared with the control group with no additions (Figure 1a). Concurrent exposure of the cells to the two mAbs resulted in significantly augmented growth inhibition (80%). Monoclonal antibody W6/32 binds to the core HLA protein on human cells. When mAb W6/32 was combined with mAb C225 or 4D5, no such additive effect on proliferation was observed. Statistical analyses indicated significant differences in growth inhibition among four treatment conditions: control, C225, 4D5, and C225 plus 4D5 (P<0.01), as well as significant differences between any two cultures from the four treatment conditions.

Figure 1
figure 1

Inhibition of cell proliferation and cell cycle arrest induced by mAb C225 and/or mAb 4D5. (a) OVCA420 cells (2×104 cells per well) were cultured for 7 days as described in the Materials and methods in the absence or presence of 20 nM mAb C225, 20 nM mAb 4D5, 20 nM mAb W6/32 or the combination of any two mAbs. The media and the antibodies were replenished once on day 4. Triplicate cultures were counted with a Coulter counter on day 7. (b) Exponentially growing OVCA420 cells were incubated with no addition (upper left panel), 20 nM mAb C225 (upper right panel), 20 nM mAb 4D5 (lower left panel) or the combination of 20 nM mAb C225 and 20 nM mAb 4D5 (lower right panel), respectively, for 24 h. Cell cycle distributions were analysed by flow cytometry as described in the Materials and methods

Full size image

Flow cytometric analysis demonstrated that the reduced cell proliferation was associated with an increased G1 cell population (Figure 1b). Concurrent treatment of the cells with mAb C225 and mAb 4D5 resulted in greater accumulation in G1 phase.

Progression through G1 phase in mammalian cells is controlled by sequential activation of CDK kinases (Morgan, 1995). We examined the activity of CDK2 after 24 h exposure to mAb C225, mAb 4D5 or the combination of the two mAbs, using the assay of histone H1 phosphorylation in cell lysates. Treatment of OVCA420 cells resulted in inhibition of CDK2 kinase activity by either antibody. Consistent with the observed effects on cell proliferation and cell cycle arrest, concurrent treatment of the cells with mAb C225 and mAb 4D5 resulted in greater inhibition of CDK2 activity (Figure 2).

Figure 2
figure 2

Inhibition of CDK2 activity by mAb C225, mAb 4D5 or the combination of mAb C225 and mAb 4D5. OVCA420 cells were cultured for 24 h in the absence or presence of 20 nM mAb C225, 20 nM mAb 4D5 or combination of 20 nM mAb C225 and 20 nM mAb 4D5. CDK2-associated histone H1 kinase activity was assayed by immunoprecipitation of the cell extracts with antibody against CDK2, followed by incubating the immunoprecipitates with a reaction solution for the in vitro histone H1 assay. Radiolabeled histone H1 bands were separated by 12% SDS-polyacrylamide gel electrophoresis (inset). The bands were quantitated with a PhosphoImager and the data are presented in bar graphs. The data are representative results from three similar individual experiments

Full size image

Involvement of CDK inhibitor, p27Kip1, in mAb-induced G1 arrest

Rb and Rb related protein, p130, are substrates of CDKs, and their phosphorylation levels may reflect CDK activity in vivo. We therefore measured the amounts of Rb and p130 protein and their levels of phosphorylation following mAb C225 or mAb 4D5 treatment. Figure 3a shows that treatment of OVCA420 cells with either mAb C225 or mAb 4D5 resulted in a decrease in the level of hyperphosphorylated p130, with a concomitant increase in the hypophosphorylated form. In contrast to this change in p130 phosphorylation status, the total level of Rb was found to be decreased. A similar response to EGF receptor blockade has been observed in other cell lines such as A431 human squamous carcinoma cells and immortalized MCF10A mammary epithelial cells (Fan et al., manuscript submitted; Chou et al., 1996) The significance of Rb protein reduction in response to mAb 225 or 4D5 treatment is not known, but a similar result was observed in TGF-β-arrested mink epithelial cells (Laiho et al., 1990).

Figure 3
figure 3

Effects of mAb C225 and mAb 4D5 on the phosphorylation of Rb and p130 protein. OVCA420 cells were treated with mAb C225 or mAb 4D5 for indicated time periods. The levels of p130 (a) and Rb (b) were assayed by immunoblotting with antibodies against p130 and Rb. Hyperphosphorylated p130 or Rb protein displays an upward shift on SDS gel electrophoresis

Full size image

We previously reported that treatment of several cancer cell lines with anti-EGF receptor mAb 225 resulted in G1 arrest, which was associated with induction of the CDK inhibitor, p27Kip1. The induced p27Kip1 protein was physically associated with CDK2 molecules and contributed to the observed kinase inhibition following mAb 225 treatment (Fan et al., 1997; Peng et al., 1996; Wu et al., 1996). In the present study, we further investigated the effect of mAb C225 as well as the effect of mAb 4D5 on the induction of p27Kip1 on OVCA420 cells. Figure 4 shows that mAb C225 and mAb 4D5 both induced markedly increased expression of p27Kip1. The kinetics for p27Kip1 induction was similar for mAb C225 and mAb 4D5, with the maximal levels at approximately 24 h after antibody exposure. The association of p27Kip1 with CDKs in these cells and the effects of combined mAb treatment upon p27Kip1 levels are described below.

Figure 4
figure 4

Regulation of p27kip1 protein levels by mAb C225 or mAb 4D5. OVCA420 cells were cultured in the absence or presence of 20 nM mAb C225 or 20 nM mAb 4D5 in 0.5% FBS media for indicated time periods. The levels of p27kip1 were assayed by immunoblotting with antibodies against p27kip1

Full size image

The early and mid-G1 cyclins, including cyclins D1, D3, and E, did not show any changes in protein levels after 24 h exposure to mAb C225 or mAb 4D5, alone or in combination (Figure 5). In contrast, cyclin A showed moderately reduced levels following mAb C225 or mAb 4D5 treatment and, especially after combined treatment with the two mAbs. The synthesis of cyclin A typically peaks in S phase (Sherr, 1996). Exposure of the OVCA420 cells to the mAbs resulted in accumulation of cells in G1 phase, with a reduced per cent in S cells (Figure 1b). This may explain the reduced level of cyclin A observed in these cultures. Taking these observations together, we conclude that mAb 225 and mAb 4D5 are acting through a similar mechanism, in attenuating mitogenic growth signals from the EGF receptor and the HER2.

Figure 5
figure 5

Expression of cyclins after mAb C225 and/or mAb 4D5 treatment. OVCA420 cells were cultured in 0.5% FBS media in the absence or presence of 20 nM mAb C225 or 20 nM mAb 4D5 for indicated time periods. The levels of cyclins D1, D3, E and A were assayed by immunoblotting with corresponding antibodies

Full size image

Reversal of mAb C225 and mAb 4D5-induced growth inhibition by EGF

We explored whether EGF can reverse the growth inhibition induced by mAb C225 or mAb 4D5. Addition of exogenous EGF to OVCA420 cultures did not stimulate proliferation, presumably due to optimum levels of autocrine activity mediated by endogenous TGF-α. However, the growth inhibition resulting from exposure of cells to mAb C225, mAb 4D5, or a combination of the two mAbs could be prevented when saturating concentrations of EGF were administered with the mAbs (Figure 6).

Figure 6
figure 6

Capacity of EGF to reverse the growth inhibition induced by mAb C225, mAb 4D5, or combination of mAb C225 and mAb 4D5. One set of OVCA420 cells (2×104 cells per well in triplicate) were cultured for 7 days in 0.5% FBS media with no additions, 20 nM mAb C225, 20 nM mAb 4D5 or combination of 20 nM mAb C225 and 20 nM mAb 4D5. The other parallel set of the cultures were concurrently grown in the presence of 10 nM EGF for each group. The media and the various additions were replenished once on day 4. On day 7, the cells were harvested and counted with a Coulter counter

Full size image

We then determined whether the reversal of the antibody-induced growth inhibition by EGF is accompanied by changes in p27Kip1 levels and the association of p27Kip1 with CDKs. In these experiments, the cells were treated for 24 h with either the antibody alone or with combination of the two mAbs, in the presence or absence of EGF. Figure 7a confirms the induction of p27Kip1 by antibody treatment. Combination treatment with mAb C225 and mAb 4D5 resulted in a greater amount of p27Kip1 (lane 7) than either antibody alone (lanes 3 and 5). In parallel with the increased levels of p27Kip1 in cells treated with either mAb, increased amounts of CDK2, CDK4 and CDK6 were coimmunoprecipitated with p27Kip1 (Figure 7b, c and d, lanes 3 and 5). This indicates that there was an increased physical association of p27Kip1 with these CDKs in vivo. Consistent with the data on cell proliferation and cell cycle distribution, concurrent treatment of the cells with mAb C225 and mAb 4D5 resulted in a greater level of p27Kip1 (Figure 7a, lane 7) and increased association of p27Kip1 with CDK2, CDK4 and CDK6 (Figure 7b – d, lane 7). The increased levels of p27Kip1 in association with CDKs could be prevented by addition of EGF to the antibody-treated cultures (Figure 7b – d, lanes 4, 6 and 8).

Figure 7
figure 7

Effect of EGF on the levels of p27Kip1 and the association of p27Kip1 with CDKs in culture treated with mAb C225, mAb 4D5, or combination of mAb C225 and mAb 4D5. One set of OVCA420 cells was treated in 0.5% FBS media in the absence or presence of 20 nM mAb C225, 20 nM mAb 4D5, or combination of 20 nM mAb C225 and 20 nM mAb 4D5 for 24 h. The other parallel set of cells was concurrently treated with 10 nM EGF for each group. Cell extracts were immunoprecipitated (IP) with antibodies against p27kip1, followed by electrophoresis resolution and Western blot (WB) assays with antibodies against p27kip1, CDK2, CDK4 and CDK6

Full size image

To further demonstrate the capacity of EGF to rescue antibody-induced growth inhibition, we examined the activities of CDKs under these conditions, using GST-Rb as a substrate for CDK in an in vitro kinase assay. In these experiments, the cells were again treated for 24 h with either antibody alone or with a combination of the two mAbs, in the presence or absence of EGF. Figure 8 demonstrates a moderate decrease in the activities of CDK2, CDK4 and CDK6 in phosphorylating Rb in lysates from cells treated with mAb C225 or 4D5 and reversal of the kinase inhibition when these cultures were concurrently exposed to EGF. Treatment with the combination of the two mAbs resulted in greater inhibition of CDK2, CDK4 and CDK6 activity, which also was prevented by addition of EGF.

Figure 8
figure 8

Effect of EGF on reversing CDK inhibition induced by mAb C225, mAb 4D5, or the combination of mAb C225 and mAb 4D5. One set of OVCA420 cells was cultured in 0.5% FBS media in the absence or presence of 20 nM mAb C225, 20 nM mAb 4D5, or the combination of 20 nM mAb C225 and 20 nM mAb 4D5 for 24 h. The other parallel set of the cells was concurrently cultured with 10 nM EGF for each group. Cell extracts were immunoprecipitated with antibodies against CDK2, CDK4 or CDK6 and the immunoprecipitates were incubated in a kinase reaction solution containing GST-Rb and [32P]-γ-ATP for the in vitro kinase activity assay. Radiolabeled GST-Rb proteins bands were separated by 8% SDS-polyacrylamide gel electrophoresis. The gels were dried and subjected to autoradiography

Full size image

Mechanisms for EGF-mediated reversal of mAb 4D5-induced growth inhibition

While the capacity of EGF to reverse mAb C225 induced-inhibition of kinase activity and proliferation can be explained by competitive binding of EGF to receptors, the reversal of mAb 4D5 induced-inhibition of kinase activity and proliferation by EGF can not be explained by a similar mechanism, since EGF does not bind to HER2 directly. There were low basal levels of receptor tyrosine phosphorylation at the location of the HER2 protein (p185), but hardly any basal phosphorylation of the EGF receptor (at p170) was detected by Western blot analysis with ECL visualization (Figure 9a). To further explore how EGF reverses mAb 4D5-induced growth inhibition, we examined the levels of EGF receptor and HER2 in cells treated with mAb C225, mAb 4D5 or the combination of mAbs. The levels of tyrosine-phosphorylated EGF receptor and HER2 were also measured in these antibody-treated cells. After 24 h culture of OVCA420 cells with the described antibody additions, the cells were lysed and subjected to immunoprecipitation with antibodies against EGF receptor or HER2. The EGF receptor or HER2 immunoprecipitates were then split into two equal parts and separated by electrophoresis, followed by immunoblotting with anti-individual receptor (EGF receptor or HER2) antibody or with anti-tyrosine phosphorylated protein antibody. Treatment of OVCA420 cells with mAb C225 or mAb 4D5 alone, or with the combination of the two mAbs, did not significantly change the levels of EGF receptor (Figure 9b, anti-EGFR blot, lanes 1 – 4). However, treatment of the cells with mAb 4D5, or with the combination of the two mAbs, significantly down-regulated the levels of HER2 receptor, while treatment of the cells with mAb C225 alone did not cause HER2 receptor down-regulation (Figure 9c, anti-HER2 blot, lanes 1 – 4).

Figure 9
figure 9

Levels of receptor protein and tyrosine phosphorylation of EGF receptors and HER2, after antibody treatment. One set of OVCA420 cells was cultured in 0.5% FBS media in the absence or presence of 20 nM mAb C225, 20 nM mAb 4D5, or the combination of 20 nM mAb C225 and 20 nM mAb 4D5 for 24 h. The other parallel set of antibody-treated cells was stimulated with 80 nM EGF for 5 min, right after removing the antibodies and washing the cells once with PBS. Cell extracts in RIPA buffer were subjected to electrophoretic resolution by SDS – PAGE for Western blot (WB) analysis with anti-phosphotyrosine antibodies (a), or to immunoprecipitation (IP) with antibodies against EGF receptor or HER2 first, followed by electrophoretic resolution and Western blot analyses with antibodies against the EGF receptor, HER2 and tyrosine phosphorylated proteins as indicated (b and c)

Full size image

In a parallel separate group of antibody treated-cultures, prior to harvesting the cells the antibodies (mAb C225, mAb 4D5, or combination of the two mAbs) were removed from the cultures, and the cells were washed once with PBS to remove free mAb C225 or 4D5, followed by immediate stimulation with 80 nM EGF for 5 min. Analyses of EGF receptor tyrosine autophosphorylation showed that 5 min of stimulation by exogenous EGF induced a high level of tyrosine autophosphorylation of EGF receptors, which was prevented in the cells pretreated with either mAb 225 alone or mAb C225 and mAb 4D5 in combination, but not in the cells pretreated with mAb 4D5 alone (Figure 9b, anti-P-Tyr blot, lanes 5 – 8). We know from previous studies that after mAb 225 has bound to EGF receptors, incubation for 5 min (even in the presence of high levels of EGF) does not allow adequate time for release of antibody, which would allow access of ligand to receptors (Kawamoto et al., 1983; Fan et al., 1994). This explains the inability of EGF to simulate receptor phosphorylation after culture for 24 h in the presence of mAb C225 (Figure 9b, anti-P-Tyr blot, lane 6). The baseline level of EGF receptor activation by endogenous ligand (autocrine stimulation) remained below the threshold detected by Western blot analysis with ECL visualization in the immunoprecipitated receptors (Figure 9b, anti-P-Tyr blot; lane 1).

In contrast, analyses of HER2 receptor tyrosine autophosphorylation in these EGF-stimulated cells showed that, although the levels of HER2 were significantly down-regulated in the cells that had been treated with mAb 4D5 or the combination of mAbs 4D5 and C225, stimulation of these cells for 5 min with EGF produced HER2 receptor phosphorylation to levels similar to those in untreated or mAb C225-treated cells (Figure 9c, anti-P-Tyr blot, lanes 5 – 8).

It should be stressed that the levels of HER2 in cells treated with mAb 4D5 and with mAbs C225 plus 4D5 for 24 h are much lower than in the untreated cells (Figure 9c, HER2 blot, lane 5 versus lanes 7 and 8) due to mAb 4D5-mediated down-regulation of HER2 receptors. However, their phosphotyrosine levels after brief exposure to EGF stimulation are comparable (Figure 9c, anti-P-Tyr blot, lane 5 versus lanes 7 and 8). This observation indicates that the per cent of HER2 receptors that are phosphorylated in mAb 4D5 treated cells and in mAb C225 plus 4D5 treated cells must be much higher than in untreated cells. These findings suggest that the growth inhibitory signal resulting from mAb 4D5-induced HER2 receptor down-regulation in these cells can be compensated by EGF, which can stimulate the HER2 signaling pathway by forming EGF receptor-HER2 heterodimers with the residual (not down-regulated) HER2 molecules remaining on the cell surface. Together, these observations indicate that only modest levels of HER2 activation, and even lower levels of EGF receptor activation, suffice to stimulate optimal proliferation of the OVCA420 cells.

Discussion

Growth factors play vital roles in regulating cell proliferation by binding to their cognate cell membrane receptors and eliciting a cascade of biological responses in target cells. EGF receptors and HER2 appear to be two of the most important growth factors to be involved in epithelial cancer cell proliferation, especially in human breast and ovarian cancer (Jadines et al., 1993; Reese and Slamon, 1997; Slamon et al., 1987). Therefore, it is reasonable to combine mAb C225 and mAb 4D5, in order to inhibit the proliferation of cancer cells stimulated by both EGF receptor and HER2 signals. Furthermore, HER2 is known to be cross-activated upon the formation of heterodimers between EGF receptors and HER2, which is triggered upon the binding of EGF to EGF receptors (Goldman et al., 1990; Wada et al., 1990). EGF receptors can be activated by either EGF receptor homodimerization or heterodimerization with HER2 (Goldman et al., 1990; Wada et al., 1990). In addition, there is evidence that receptor heterodimers have higher ligand affinity and activity (Stern and Kamps, 1988; Qian et al., 1994).

We demonstrated previously that interruption of the EGF receptor/TGF-α autocrine pathway by mAb 225 results in perturbation of cell cycle progression, by inducing a reversible cell cycle arrest in G1 phase associated with p27Kip1-mediated inhibition of CDK2 activity. This sequence of events has been demonstrated with several malignant and non-malignant cell lines, including DiFi colon adenocarcinoma cells (Wu et al., 1996), DU145 prostatic adenocarcinoma cells (Peng et al., 1996), A431 vulvar squamous carcinoma cells (Fan et al., 1997) and MCF10A nonmalignant mammary epithelial cells (Chou et al., 1996). In the present study, we further demonstrated the induction of p27Kip1 protein and its association with CDKs by mAb C225 in an ovarian cancer cell line, OVCA420. Treatment of OVCA420 cells with mAb C225 resulted in growth inhibition accompanied by an increased G1 population, increased levels of p27Kip1 protein, increased association of p27Kip1 with CDKs, and reduced CDK activities. In contrast to previously studied cell lines, mAb C225-induced p27Kip1 in OVCA420 cells showed increased association not only with CDK2, but also with CDK4 or CKD6. This resulted in reduction in kinase activity of all three CDKs. This observation indicates that different cell lines may display differences in the responses of cell cycle regulators to EGF receptor blockade with mAb C225.

Interestingly, similar changes in cell cycle perturbation and induction of p27Kip1 were also observed in OVCA420 cells treated with mAb 4D5, which is directed against HER2. Furthermore, concurrent treatment of OCVA420 cells with mAb C225 and 4D5 resulted in a greater degree of G1 arrest than with either antibody alone. This enhanced G1 arrest was accompanied by greater increases in the levels of both p27Kip1 protein and p27Kip1/CDK complexes, and greater reduction in kinase activity of the CDKs. Therefore, the EGF receptor and HER2 pathways appear to be two individual mitogenic stimulants for OVCA420 cells that share at least one common mechanism involving CDKs for regulating OVCA420 cell proliferation.

We have previously shown that mAb C225 binds to EGF receptor with a Kd similar to EGF and blocks EGF binding and EGF-mediated activation of the receptor (Kawamoto et al., 1983; Fan et al., 1993, 1994). We have found that receptor overexpression is not required for mAb 225 to inhibit the proliferation of a number of malignant cell lines and all non-malignant cells tested. Our current data suggest that in OVCA420 cells mAb C225 is mainly working by blocking the binding of ligand to EGF receptors since no significant EGF receptor downregulation was detected in response to treatment with antibody (Figure 9b, anti-EGFR blot, lane 2). In contract, mAb 4D5 appears to induce HER2 downregulation in the OVCA420 cells (Figure 9c, anti-HER2 blot, lane 3), which may contribute to its mechanism of inhibition. Since there is no identified ligand for HER2 so far, it would be logical that an antibody specific for HER2 might inhibit receptor function by downregulation, although the antibody might also work by steric inhibition of receptor heterodimerization. There is published evidence that mAb 4D5 paradoxically activates HER2 autophosphorylation as a partial agonist (Scott et al., 1991). However, our own studies suggest that anti-receptor mAb-induced autophosphorylation could be an artifact, occurring after cell lysis if the cells are lysed in a buffer containing only mild detergent, such as NP40 or Triton X-100. In a study with anti-EGF receptor mAb-induced EGF receptor autophosphorylation, we have reported that following culture of cells with several anti-EGF receptor mAbs, when cells were lysed in a buffer containing greater than 0.1% SDS or a tyrosine kinase inhibitor such as genistein, this apparent EGF receptor autophosphorylation in the presence of anti-EGF receptor mAbs was prevented (Fan et al., 1993). We did not find that mAb 225 or 4D5 mediated EGF receptor or HER2 autophosphorylation in our current studies, when the cells were lysed in a buffer containing 0.1% SDS (RIPA buffer).

In summary, these studies provide novel experimental data demonstrating significantly augmented anti-proliferative effects on OVCA420 ovarian carcinoma cells resulting from concurrent treatment with mAbs directed against EGF receptors and HER2, respectively. The data also show that EGF can reverse mAb 4D5-induced growth inhibition by activating the HER2 signal presumably through the previously described heterodimerization of HER2 with EGF receptor. Binding of mAbs to both receptors may prevent the formation of active receptor heterodimers. The results of our present study provide novel experimental evidence that the combination of anti-EGF receptor mAb and anti-HER2 mAb may be useful in inhibiting cancer cells expressing both EGF receptor and HER2 pathways. Since both mAbs are currently under investigation in clinical trials and have demonstrated no severe toxicity (Prewett et al., 1996; Bos et al., 1996; Cobleigh et al., 1998; Slamon et al., 1998), the possibility of combining them in patients should be strongly considered.

Materials and methods

Materials

Anti-EGF receptor human/mouse chimeric mAb 225 (C225) has been described previously (Goldstein et al., 1995) and was provided by ImClone Systems, Inc. (New York, NY, USA). Anti-HER2 humanized mAb 4D5 (rhumAb HER2) was provided by Genetech, Inc (San Francisco, CA, USA). The hybridoma producing mAb W6/32 was ordered from American type culture collection. The W6/32 mAb was purified by our laboratory. Antibodies against human cyclin D1, cyclin D3, cyclin E, cyclin A, CDK2, CDK4, CDK6, p130, p27Kip1 and glutathione S-transferase-Rb fusion protein (GST-Rb) were purchased from Santa Cruz Biotechnology Inc (Santa Cruz, CA, USA). Antibody against retinoblastoma protein was ordered from PharMingen (San Diego, CA, USA). Anti-phosphotyrosine antibody (PY69) was obtained from ICN Biomedical Inc (Irvine, CA, USA). Anti-EGF receptor antibody (RK2) was generously provided by Dr J Schlessinger (New York University Medical Center, New York, NY, USA). Anti-HER2 antibody (Ab3) was purchased from CalBiochem/Oncogene Research Products (Cambridge, MA, USA). EGF was obtained from Collaborative Research Inc (Bedford, MA, USA). Protein A-agarose beads was purchased from Repligen Corp (Cambridge, MA, USA). [γ-32P]ATP was obtained from New England Nuclear Inc. (Boston, MA, USA). Other chemicals are ordered from Sigma.

Cell cultures and proliferation analyses

Human ovarian cancer cell line, OVCA420, was kindly provided by Dr Robert Bast's laboratory (UT MD Anderson Cancer Center, Houston, TX, USA). The cells were maintained in MEM supplemented with 10% fetal bovine serum (FBS) at 37°C in a 5% CO2/95% air atmosphere. Studies of cultured cells were initiated by seeding OVCA420 cells in low density into six-well plates in triplicate for each group with 10% FBS medium. On the next day, the cells were switched to low serum (0.5% FBS) medium and exposed to various additions as described in the figure legends. Media, antibodies or EGF was replenished once on day 3 or day 4. Cells were harvested on day 3 and/or day 7 by trypsinizing and counted with a Coulter counter. Cell cycle distribution analyses were performed as previously described (Fan et al., 1995). Briefly, cells were stained for DNA by suspending them in 2 ml PBS containing 50 μg propidium iodide/106 cells and 100 μg/ml RNase I for 6 h at room temperature. The samples were analysed on a FACScan flow cytometer.

Immunoprecipitation and Western blot

Immunoprecipitation and Western blot analysis were performed as described previously (Fan et al., 1993, 1994, 1995). Briefly, OVCA420 cells were lysed with NP40 lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP40, 50 mM NaF, 1 mM sodium orthovanidate, 1 mM phenylmethylsulfonyl fluoride and 25 μg/ml of leupetin and 25 μg/ml of aprotinin) and sonicated at 4°C. In the studies to detect the levels of EGF receptor and HER2, and the levels of the receptor phosphorylation, a RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM NaF, 1 mM sodium orthovanidate, 1 mM phenylmethylsulfonyl fluoride and 25 μg/ml of leupetin and 25 μg/ml of aprotinin) was used. Equal amounts of protein were subjected to various primary antibodies for 1 h at 4°C, followed by another 1 h incubation with protein A sepharose beads for precipitation. Immunoprecipitates or whole cell extracts were resolved by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose and blotted with specific primary antibodies. The specific bands were visualized with the ECL assay (Amersham Life Science Inc., Arlington Heights, IL, USA).

CDK kinase activity assay

Cells were lysed and sonicated in 0.5% NP40 lysis buffer as described above. Centrifugation cleared extracts of freshly prepared cell lysates (200 μg of protein) were subjected to immunoprecipitation in the lysis buffer for 1 h at 4°C in the presence of antibodies against to CDK2, CDK4 or CDK6. The in vitro histone H1 or GST-Rb phosphorylation assay was performed as previously described (Fan et al., 1995; Wu et al., 1996).

Statistical analysis

Statistical analyses were performed using the ANOVA program for one-way analysis of variance, to examine the differences among four treatment conditions: control, mAb C225 alone, mAb 4D5 alone and mAb C225 plus mAb 4D5; and using the Tukey method for multiple comparison to examine the differences between any two cultures from the four treatment conditions (Shott, 1990).