Human epidermal growth factor receptor 2 (HER2) is overexpressed in 25–30% of ovarian carcinoma cases and a correlation between increased HER2 expression and decreased survival has been demonstrated. HER2 is a ligand-less member of the HER family that functions as the preferred coreceptor for epidermal growth factor receptor (EGFR), HER3, and HER4.
An approach was developed to target HER2's role as a coreceptor using a monoclonal antibody, 2C4, which sterically hinders HER2's recruitment into a functional HER complex.
HER2 was robustly expressed in all eight ovarian carcinoma cell lines; expression of EGFR and HER3 was variable. Even though four of the eight cell lines responded to EGF, 2C4 antibody moderately inhibited in vitro proliferation of only two cell lines (OVCA433 and SK-OV-3). Furthermore, ligand-stimulated p-MAPK expression was inhibited by 2C4 only in these two cell lines after exposure to EGF. Immunoprecipitation and eTag analysis revealed that OVCA433 expressed heterodimers of EGFR/HER2, and these heterodimers were disrupted after treatment with 2C4, whereas OVCA432 cells did not have these heterodimers. In murine xenograft experiments, the in vivo growth of OVCA433, but not of OVCA432, ovarian carcinoma cells was significantly inhibited by 2C4 treatment of the mice.
Receptor tyrosine kinases (RTKs) are major mediators of growth and differentiation signals.1, 2 The human epidermal growth factor receptor (HER) signaling network is a receptor-ligand system composed of at least 11 ligands and 4 receptors: epidermal growth factor receptor (EGFR), HER2, HER3, and HER4.1, 2 The ligands can be divided into two groups based on their binding specificities.3 EGF-like ligands bind primarily to EGFR, and those include EGF, transforming growth factor (TGF)-α, amphiregulin, betacellulin, heparin-binding EGF, and epiregulin.3 The second group consists of the heregulin (HRG) growth factor family that binds to HER3 and HER4.4, 5 In contrast with the other HER receptor members, a soluble ligand of HER2 has not yet been identified.3 After ligand binding, the HER receptors become activated by dimerization between either two identical receptors (homodimerization) or different receptors of the same family (heterodimerization).4, 6 After receptor dimerization, activation of the intrinsic protein kinase activity and tyrosine autophosphorylation occurs, recruiting and phosphorylating several intracellular substrates involving the Ras-Raf-MAPK, the PI3k-Akt, and other signaling pathways that regulate multiple biological processes including apoptosis and cellular proliferation.7, 8 HER2 is known to be the preferred coreceptor for EGFR, HER3, and HER4, largely due to the extremely high signaling potency of HER2-containing heterodimers.9, 10 This preference is further biased upon overexpression of HER2, as seen in many human cancers.9, 10 HER2 overexpression confers a growth advantage to tumor cells.2, 3 This is due to the ability of HER2 to remarkably reduce the rate of ligand dissociation, which in turn prolongs and enhances signaling by growth factors.11 Thus, HER2 overexpression in tumor cells provides a selective advantage, due to better utilization of stroma-derived EGF-like growth factors.
In this study, we investigated the basal expression of the three major HER family members in ovarian carcinoma cell lines and examined the activation status of the associated downstream signaling components. These results were correlated with receptor homo- and heterodimerization patterns that exist between them. A novel monoclonal antibody, 2C4, which targets HER2 and disrupts ligand activation,12 was used to attempt to block this activation. We demonstrate that 2C4 is effective in selectively suppressing the proliferation of clonogenic ovarian carcinoma cells in vitro as well as xenografted ovarian tumors in vivo. 2C4 may be most efficient in those tumors in which ligand induces HER2 heterodimers that transduce proliferative signals.
MATERIALS AND METHODS
2C4 mAb was a generous gift from Genentech (South San Francisco, CA). Recombinant human EGF was obtained from BD Bioscience (Carlsbad, CA) and HRG was from LabVision (Fremont, CA).
The human cell lines OVCAR-3, SK-OV-3, TOV-21G, OV-90, and TOV-112D were from the American Type Culture Collection (Manassas, VA). OVCA420, OVCA432, and OVCA433 were kindly provided by Robert C. Bast Jr. (M. D. Anderson Cancer Center, Houston, TX). Cells were maintained as monolayers at 37 °C in 5% CO2/air in RPMI 1640 (Gibco, Rockville, MD) containing 10% heat-inactivated fetal bovine serum (FBS) (Omega, Tarzana, CA).
Cells were washed three times with phosphate-buffered saline (PBS) and lysed using ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 1 mM NaF, 15 mM sodium molybdate, 1 mM phenylmethylsulfonylfluoride, 10 μg/mL leupeptin, 3 μg/mL pepstatin, 10 μg/mL aprotinin, and 2 mM Na3VO4). The cellular contents were transferred to microcentrifuge tubes, clarified by centrifugation at 13,000 rpm for 15 minutes at 4 °C, and supernatant aliquots were stored at −80 °C until required. Total protein concentrations were determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories, Hemel Hempstead, UK).
Protein samples from total cell lysates (10 μg) and after immunoprecipitation were subjected to electrophoresis separation on a 7.5% polyacrylamide gel and then blotted onto an immobilon polyvinylidene difuride membrane (Amersham Pharmacia Biotech, Piscataway, NJ). Blots were blocked at room temperature for 1 hour in 5% skimmed milk powder made up in Tris-buffered saline (TBS)-Tween 20 (0.05%) and then incubated for a minimum of 1 hour in primary antibody diluted 1/10,000 for GAPDH (reference control) (Research Diagnostics, Flanders, NJ) or 1/1,000 for EGFR (sc-03, Santa Cruz Biologicals, Santa Cruz, CA), HER2 (06-562, Upstate Biotechnology, Lake Placid, NY), HER3 (sc-285, Santa Cruz), and phospho-p44/42 MAPK (Thr202/Tyr204) (Cell Signaling, Beverly, MA) in TBS-Tween solution. The membranes were washed three times in TBS-Tween and then incubated for 1 hour with the required secondary horseradish peroxidase-labeled donkey antirabbit or sheep antimouse IgG antibody (Amersham Biosciences, Buckinghamshire, UK), diluted 1/20,000 in TBS-Tween solution. The blots were developed using the enhanced chemoluminescence (ECL) kit (Amersham Pharmacia Biotech).
To assess MAPK phosphorylation, cells (5 × 105/well) were plated in serum-containing media in 6-well culture plates. On the next day, media were removed and fresh medium without serum was added to each well. On the following day prior to assay, the medium was replaced again with serum-free media. Cells were incubated for 2 hours with 100 nM 2C4. Cells were then treated with either 4 nM EGF at 37 °C for 10 minutes or 30 nM HRG at room temperature for 30 minutes. The reaction was stopped by aspirating the cell medium and then adding 0.5 mL RIPA buffer. MAPK activation was assessed by Western blotting using an antiphosphorylated MAPK antibody (Cell Signaling). The intensity of bands on films was quantified using the NIH Image Program (Bethesda, MD) and the results were graphed.
EGFR was immunoprecipitated using 2 μg of a monoclonal antibody (sc-120) (Santa Cruz) from 200 μg of total cell lysates overnight at 4 °C with slow agitation. Immunocomplexes were captured with 20 μl of protein A gel slurry and the samples were allowed to mix at 4 °C for 2 hours. The beads were then collected by centrifugation and washed three times with RIPA lysis buffer to remove unbound materials. SDS sample buffer was added and the samples were boiled for 5 minutes to release the bound proteins before gel loading. Supernatants were run on 5% polyacrylamide gels and electroblotted onto nitrocellulose membranes. The presence of EGFR and HER2 was assessed by probing the blots with the respective antibodies. The blots were observed using a chemiluminescent substrate (ECL, Amersham Pharmacia Biotech).
Proximity-Based eTag Assay
The eTag assay (Virologic, San Francisco, CA) is multiplexed proximity-based test that evaluates protein dimerization using specific antibodies to interacting protein partners.13, 14 Briefly, specific antibody to EGFR is conjugated with a unique fluorescent eTag moiety via a cleavable linker. HER2-specific antibody is coupled to a “molecular scissor” which can emit singlet oxygen. If eTag locates within 200 nm of the “molecular scissor,” singlet oxygen emitted from the “molecular scissor” can break the linker releasing the eTag from the antibody. The signal of the released eTag can be detected by capillary electrophoresis (CE) and the results are analyzed using eTag Informer software. The quantification is based on the CE peak. Therefore, when EGFR dimerizes, eTag is released and detected.
OVCA432 and OVCA433 cells were treated with either EGF alone or 2C4 and EGF as described above. The cell lysates were prepared by adding 500 μL of ice-cold ACLARA lysis buffer (Virologic) per 10-cm dish containing phosphatase and protease inhibitors. The analysis was performed in triplicate. Background for the eTag signal was determined by omitting the cell lysates in the assay and later subtracting it from each sample. The eTag signals are reported as Peak Area, expressed in relative fluorescence units (PA-RFU) for 30 μg total protein/assay.
Clonogenic Assay in Soft Agar
The effect of 2C4 on clonogenic growth of ovarian carcinoma cells was determined by dose–response studies in soft agar as previously described.15 Ovarian carcinoma cells from 60–80% confluent liquid cultures were trypsinized. Washed single-cell suspensions of cells were enumerated, mixed with soft agar, and plated at a total of 1 × 103 cells/well in a volume of 400 μL/well into 24-well, flat-bottomed plates. Prior to this step, 2C4 antibody was pipetted into wells. After 14 days of culture, colonies (>50 cells) were counted with an inverted microscope. All experiments were done independently at least three times in triplicate dishes per experimental point.
Xenograft studies were performed as previously described.12 Four- to six-week-old athymic BNX nu/nu female mice were obtained from Harlan Sprague Dawley, (Indianapolis, IN) and maintained in pressurized ventilated cages at the Cedars-Sinai Medical Center Vivarium.
Six female mice were subcutaneously inoculated with 1 × 107 of either OVCA432 or OVCA 433 cells. Cells were injected in both flanks together with reconstituted basement membrane (Matrigel; Collaborative Research, Bedford, MA), as described previously.12 Three mice began treatment 2 days later with intraperitoneal injections of 20 mg/kg 2C4 in PBS twice weekly for 4.5 weeks. Three control mice were given vehicle alone. Tumors were measured every 3–4 days with vernier calipers and tumor volumes were calculated by the formula: π/6 × larger diameter × (smaller diameter)2 in treated groups and controls.
At the end of the experiments (after 5 weeks), blood was collected from the orbital sinus for serum chemistries and blood analysis using the Dupont Analyst Benchtop Chemistry System (Dade International, Newark, DE) and by Serono-Baker 9000 Diff (Biochem Immuno-Systems, Allentown, PA), respectively. Animals were sacrificed by carbon dioxide asphyxiation and tumor weights were measured after their careful resection. Tumors, livers, lungs, spleens, and kidneys were fixed and stained for histologic analysis. All animal experiments were in compliance with the National Institutes of Health guidelines.
Tumors and normal organs from sacrificed mice were fixed in 10% neutral buffered formalin and embedded in paraffin wax prior to histologic sectioning. Sections were stained with hematoxylin and eosin and tumor necrosis and fibrosis were examined. Normal organs were evaluated for evidence of toxic damage. Controls consisted of tumors and organs from mice not subjected to treatment.
Statistical Analysis of the Xenograft Experiments
Pairwise differences between the tumor volumes of the treatment groups were compared over time. Statistical differences between means were analyzed by the t-test. P values of less than 0.05 were considered statistically significance.
EGFR, HER2, and HER3 Protein Expression in Eight Ovarian Carcinoma Cell Lines and pMAPK Expression before and after Exposure to EGF/HRG and 2C4
To examine expression of HER family proteins (EGFR, HER2, and HER3) in ovarian carcinoma cell lines, we performed Western blot analysis using the specific antibodies (Fig. 1). EGFR was strongly expressed in SK-OV-3, OVCA420, OVCA432, and OVCA433 cell lines, whereas OVCAR-3, TOV-21G, OV-90, and TOV-110 cells weakly expressed EGFR. HER2 was prominently expressed in all eight cell lines, with a marked expression in SK-OV-3. HER3 was strongly expressed in OVCAR-3, TOV-112D, OVCA420, and OVCA432 cell lines. Conversely, SK-OV-3, TOV-21G, and OV-90 showed weak expression of HER3.
The cell lines were treated with either EGF or HRG to examine whether their receptors transduced cellular proliferative signals (Fig. 2). Since these signals activate the HER receptors resulting in phosphorylation of MAPKs (ERK1 [p44] and ERK2 [p42]),7 we used a phospho-MAPK specific antibody to examine the transduction of the signal from HER family receptors in these cell lines. EGF induced the phosphorylation of MAPKs in four cell lines: SK-OV-3, OVCA420, OVCA432, and OVCA433 (Fig. 2), whereas the other cell lines did not have an apparent response to EGF stimulation (data not shown). HRG induced phosphorylation of MAPKs in SK-OV-3, OVCA 432, and OVCA433 (Fig. 2). We used these four cell lines for further analysis since proliferation of these cell lines can be stimulated by EGF and/or HRG.
To examine if 2C4 blocked signal transduction from the HER complex in these cell lines, we analyzed the phosphorylation status of MAPKs using a phospho-specific MAPK antibody after the cells were cultured with ligand ± 2C4 (Fig. 2). 2C4 blocked EGF-induced phosphorylation of MAPKs in SK-OV-3 and OVCA433 cells and HRG-induced phosphorylation of MAPKs in OVCA433 (Fig. 2). Interestingly, 2C4 had no effect on either the EGF- or HRG-induced phosphorylation of MAPKs in OVCA432 cells. All blots were reprobed with GADPH to confirm equal loading of proteins (data not shown).
Heterodimers of HER2 and EGFR Detected in OVCA433, but Not in OVCA432
EGF binds to EGFR, and promotes the formation of either EGFR/EGFR homodimers, EGFR/HER2 heterodimers, or, less frequently, EGFR/HER3 and EGFR/HER4 heterodimers.16, 17 To clarify the preferred partners of EGFR in each of these ovarian carcinoma cell lines, cell lysates were immunoprecipitated with monoclonal antibody directed to EGFR, and immunoblots were probed for EGFR or HER2 (Fig. 3A). In OVCA433 cells, HER2 was co-immunoprecipitated with EGFR, whereas HER2 was not co-immunoprecipitated with EGFR in OVCA432 cells. This result suggests that EGFR/HER2 heterodimers are expressed on OVCA433 cells, but not on OVCA432 cells.
We also performed eTag assay to confirm the heterodimerization of EGFR and HER2 in OVCA433. In the presence of EGF, the dimer of EGFR and HER2 was detected (Fig. 3B). Addition of 2C4 in the media disrupted this dimerization (Fig. 3B). Using the same assay, dimers of EGFR and HER2 were not detected in OVCA432 cells even before their exposure to 2C4 (data not shown).
2C4 Diminishes Clonal Proliferation of Ovarian Carcinoma Cell Lines In Vitro
We performed clonogenic growth assays in soft agar to test the antiproliferative activity of 2C4 against the above-mentioned four ovarian carcinoma cell lines against 2C4, which blocks HER2 from forming heterodimers (Fig. 4). Two cell lines (OVCA420 and OVCA432) showed minimal response to 2C4; the other two cell lines (SK-OV-3, OVCA433) showed moderate sensitivity to 2C4 (approximately 40% inhibition of clonal growth occurred in the presence of 1 × 10−6 M 2C4).
2C4 Inhibits Growth of Human Ovarian Carcinoma Xenografts
To examine the effect of 2C4 treatment on tumor cell growth in vivo, nude mice were injected in each flank with either OVCA433 (in vitro responder to 2C4) or OVCA432 (in vitro nonresponder to 2C4) ovarian carcinoma cells; the mice were treated with 2C4 monoclonal antibody by intraperitoneal injection, significant growth inhibition of the OVCA433 ovarian carcinoma xenografts occurred with 2C4 administration as compared with the size of the tumors in the diluent-treated animals (Fig. 5A,B). In contrast, no significant difference in tumor growth of OVCA432 cells was detected in the mice receiving 2C4 compared to the control cells (Fig. 5A).
During the study, the total body weights of all treated mice were 91–101% of those of the control groups (data not shown), and all the mice in each of the cohorts appeared healthy. No significant differences in the mean weights, histology of internal organs, mean blood chemistries (including hepatic parameters as well as hematopoietic values) were found between diluent-treated mice and those that received 5 weeks of 2C4 antibody treatment (data not shown). Histologic analysis of OVCA433 tumors from untreated mice revealed poorly differentiated carcinomas with small foci of necrosis and fibrosis, which constituted approximately 20% of the area of the tumor sections. In contrast, 50–60% of each of the tumor sections from mice treated with 2C4 revealed necrosis and histologic changes of apoptosis, including formation of apoptotic bodies, and fibrosis involved approximately 30% of the tumor area (data not shown).
An interesting feature of the HER system is that it is frequently coopted by various hyperproliferative diseases including cancer,18 psoriasis,19 and vascular restenosis.20 As a result, two of these receptors, EGFR and HER2, are targets for drug development, particularly in a number of solid tumors.21 Disrupting ligand-activated HER2 may complement other strategies devised to target HER signaling in human neoplasms. The current study examined the role of ligand-activated HER2 signaling in ovarian tumor models.
HER2 is known to be the preferred coreceptor for EGFR, HER3, and HER4, largely due to the extremely high signaling potency of HER2-containing heterodimers.9, 10 Strong expression levels of HER2 were noted in all eight ovarian carcinoma cell lines examined, with variable expression of EGFR and HER3.
We demonstrated that 2C4, a monoclonal antibody that sterically hinders recruitment of HER2 into HER ligand complexes, is able to inhibit the in vitro and in vivo growth of ovarian carcinoma cell lines. We suggest that this growth inhibition occurs as a result of the disruption of MAPK and other signaling pathways occurring downstream of receptor dimerization. We previously suggested that 2C4 behaves in a very similar manner to trastuzumab, another antibody against HER2, with regard to its pharmacokinetic properties; however, 2C4 has a growth inhibitory effect on breast and prostate cancers that were not inhibited by trastuzumab.12 Crossblocking experiments, as well as epitope-mapping studies, have shown that trastuzumab and 2C4 bind to distinct epitopes on the extracellular domain of the HER2 receptor.22 More recently, trastuzumab was demonstrated to inhibit HER2 ectodomain cleavage by matrix metalloproteases, whereas 2C4 did not.23 Of note, 2C4-mediated disruption of this receptor complex is independent of HER2 expression levels, in that it is effective in blocking receptor complex formation in both low- and high-HER2-expressing cell lines.12
Signal transduction by the HER family of receptors is mediated by two major pathways: the Ras-Raf-MAPK pathway, which drives cell proliferation and additional processes, and the PI3k-Akt pathway, which primarily mediates cellular survival and antiapoptotic signals.7, 8
In our experiments, 2C4 had only a moderate antiproliferative activity against the ovarian carcinoma cells in vitro (Fig. 4). These cells were not grown in the presence of EGF. Perhaps the addition of this growth factor would have enhanced proliferation of these cells and resulted in a larger difference of proliferation during treatment with 2C4.
We observed that 2C4 inhibits the EGF-induced activation of p42/44 MAPK in SK-OV-3 and OVCA433 cells, suggesting that HER2 is the dimerization partner involved in MAPK signaling in response to these growth factors. This is supported by our immunoprecipitation data, which indicate that in OVCA433 cells, EGFR/HER2 heterodimers are formed. In contrast, 2C4 had no effect on either EGF- or HRG-induced MAPK activation in OVCA432 cells. This is not surprising, given that our immunoprecipitation data showed that EGFR does not form EGFR/HER2 heterodimers upon ligand binding in these cells. These cells may express EGFR homodimers even though they also express HER2. 2C4 was also able to block the HRG-induced MAPK activation in OVCA433 cells (Fig. 2). We suggest that in cells where 2C4 is able to inhibit MAPK signaling, recruitment of HER2 into HER receptor ligand complexes is necessary for maximal signaling response. This has been reported previously for EGFR, where blocking the recruitment of HER2 into EGFR-ligand complexes does not ablate signaling but dampens the diversity, intensity, and duration of signaling.24 It is not clear why some cancer cells preferentially express heterodimer and others do not.
The growth of SK-OV-3 and OVCA433 cells in vitro was inhibited by 2C4 in a dose-dependent fashion. This mirrors the inhibition of MAPK phosphorylation by 2C4 noted in our experiments, suggesting that the growth inhibition is a result of a disruption of the MAPK signaling cascade as well as other downstream pathways. In our xenograft experiment, growth of OVCA433 was significantly inhibited (Fig. 5A). Growth of OVCA432 was also inhibited by 2C4, but because of the large standard deviation between the sizes of tumors, these results were not significantly different from diluent-treated control animals (Fig. 5A). The in vivo environment around the cancer cells, including stromal and vascular cells, may contribute to the proliferation of cancer cells. Nevertheless, our results suggest that the cancer cells could respond to 2C4 in vivo even though their growth was not robustly suppressed in vitro.
In conclusion, we have shown that 2C4, an antibody directed against HER2, is able to disrupt the HER signaling pathway and partially inhibit the in vitro and in vivo growth of selective ovarian carcinoma cell lines that actively transduce proliferation signals through their EGFR/HER2 receptors. HER2 is overexpressed in 25–30% of ovarian carcinoma cases and a correlation between increased HER2 levels and decreased survival has been demonstrated.25 Disrupting ligand-activated HER2 may complement other strategies devised to target HER signaling in human neoplasms. We may select patients with ovarian carcinomas who benefit from 2C4 treatment by quantification of blocking the signaling of activated EGFR/HER2 heterodimer in the tumor cells.
The authors thank Dr. Sharat Singh, Virologic (formerly Aclora Biosciences), for performing the eTag assays.