Erlotinib (Tarceva), is an orally available, reversible inhibitor of epidermal growth factor receptor (EGFR; HER1) that exhibits inhibitory activity on purified HER2 kinase at much higher concentrations. Despite the minimal activity on purified protein in vitro, in vivo studies show that erlotinib inhibits the growth of HER2-driven systems effectively. Several hypotheses have been put forward to explain this discrepancy. In particular, it has been suggested that erlotinib might indirectly suppress the activity of HER2 by blocking the ability of EGFR to transactivate it when the two receptors are part of a heterodimer complex. However, an alternative possibility that has not been adequately addressed is whether the direct inhibitory action of erlotinib on the HER2 kinase might account for the observed biological responses. To distinguish between a direct effect of erlotinib on HER2 kinase in intact cells or an indirect effect of erlotinib on HER2 activity that is mediated through EGFR, we generated cell lines that express either EGFR-H2 chimeric receptor or HER2 and HER3 receptors in an EGFR-negative background. We show that dose-dependent inhibition of HER2 was achieved at the receptor level, on downstream signaling molecules, and more importantly was also translated into inhibition of cell growth. Our findings imply that the inhibitory effect of erlotinib in HER2-expressing cells may in part be mediated through direct interaction with HER2 rather than indirectly through a process that requires the presence of EGFR. [Cancer Res 2007;67(3):1228–38]
The epidermal growth factor receptor (EGFR) family consists of four members: EGFR/HER1, HER2/ErbB2, HER3/ErbB3, and HER4/ErbB4. These receptors are essential in modulating cell proliferation, cell differentiation, and cell survival in many tissue types (1). On ligand binding, the receptors form homodimers and heterodimers, and the subsequent activation of the intrinsic kinase then results in receptor autophosphorylation and the activation of downstream signaling mediators (2). Aberrant expression of EGFR and HER2 has been implicated in the development of many types of human cancers, including lung, head and neck, colorectal, pancreatic, breast, and ovarian cancers, and patients with tumors bearing dysregulated EGFR or HER2 receptors have been linked to poor clinical outcome (3, 4). Therefore, EGFR as well as HER2 have been intensely pursued as therapeutic targets.
Erlotinib (OSI-774, Tarceva) is a selective, orally available low molecular weight inhibitor that binds competitively to the ATP-binding site at the kinase domain of EGFR. The IC50 measured in in vitro kinase assays is 2 nmol/L, whereas the potency in intact cells is ∼20 nmol/L (5). Preclinical studies with erlotinib have shown potent antitumor activity in a variety of cultured tumor cell lines as well as human tumor xenografts (6, 7), and favorable clinical studies have led to the approval of erlotinib for the use in the treatment of advanced non–small cell lung cancer (NSCLC) and advanced or metastatic pancreatic cancer (8).
Identifying predictors of responsiveness to EGFR tyrosine kinase inhibitors has been a focus of recent research, and a variety of efforts are ongoing to elucidate what confers sensitivity to these inhibitors. In contrast to the HER2 therapeutic antibody trastuzumab that is effective only on HER2-amplified tumors, the role of EGFR expression as a predictor for clinical outcome is less defined (9–11). Consequently, further studies were designed to determine if EGFR gene amplification rather than EGFR overexpression as assessed by immunohistochemistry is a better indicator to identify patients who might benefit from EGFR-directed therapies (12). The recently identified EGFR kinase mutations have proven to be very sensitive to erlotinib (13–15). Yet, tumors with mutated EGFR are not the only tumors that respond to kinase inhibitors (16–18), suggesting that factors other than EGFR mutations contribute to the sensitivity.
The presence or absence of other members of the EGFR family alters the responsiveness of tumor cells to EGFR kinase inhibitors (19) because these receptors modify the signaling properties of EGFR (20). In fact, HER2 is known to form heterodimers with EGFR, and several studies have recently reported that epithelial tumor cell lines and human tumor xenograft models overexpressing HER2 are particularly sensitive to erlotinib and to another EGFR tyrosine kinase inhibitor, gefitinib (21–23). Additionally, a recent clinical study suggests that EGFR-positive NSCLC patients with increased HER2 gene copy numbers respond better to gefitinib than patients that do not overexpress HER2 (24).
Why HER2-overexpressing cells are particularly sensitive to an EGFR inhibitor is still unclear. In vitro kinase assays suggest that erlotinib is not effective in inhibiting HER2 kinase directly because the sensitivity of this kinase is 40- to 3,000-fold less when compared with the EGFR kinase (25–27). However, HER2 phosphorylation decreases significantly upon inhibitor treatment (22, 28). Therefore, we wanted to explore whether erlotinib had the capability to directly block HER2 kinase function in intact cells.
HER2 activation arises through either a ligand-dependent or a ligand-independent process. Ligand-independent activation occurs when receptor expression levels are very high, leading to spontaneous receptor homodimerization and kinase activation. In contrast, ligand-dependent HER2 kinase activation requires the formation of a heterodimeric complex between HER2 and another member of the HER receptor family. In this case, the ligand-occupied partner transactivates HER2.
In the present study, we explored the inhibitory effects of erlotinib on both modes of HER2 activation. We present evidence that erlotinib directly inhibits the HER2 kinase in the absence of EGFR. The inhibition of HER2 kinase at submicromolar levels also translates into the effective inhibition of Akt and mitogen-activated protein kinase (MAPK) signaling pathways. We further show that, in the absence of EGFR, erlotinib effectively inhibits HER2/HER3-driven signaling and cell proliferation. These findings indicate that direct inhibition of HER2 may account for a portion of the overall responsiveness to erlotinib in HER2-driven and HER2/HER3-driven signaling.
Materials and Methods
Materials. NR6 cells were kindly provided by Mark Pegram (University of California at Los Angeles, Los Angeles, CA; ref. 29) and maintained in RPMI 1640 supplemented with 10% calf serum. The NCI-H1666 cell line was purchased from the American Type Culture Collection and maintained in a 50:50 mixture of F12 Ham's medium and DMEM supplemented with 10% fetal bovine serum.
Cetuximab was purchased from St. Mary's pharmacy. The EGF-like domain of HRGβ177-241 was expressed in Escherichia coli and purified as described previously (30).
Antibodies. Anti-phosphotyrosine antibody (RC20H) conjugated to horseradish peroxidase was purchased from BD Transduction Laboratories (Lexington, KY). Anti-HER2 (Ab-17) was obtained from Lab Vision Corp. (Fremont, CA). Anti-HER3 (C-17) and anti-EGFR (1005) were from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against pHER3, pAkt (Ser473), pAkt (Thr308), Akt, pMAPK, and p44/42MAPK were purchased from Cell Signaling.
Expression constructs and infection of NR6 cells. Full-length HER2 cDNA was subcloned into the LNCX retroviral vector, HER3 cDNA was introduced into the LHCX vector, and the EGFR cDNA was subcloned into MSCVneo vector. All retroviral vectors were purchased from BD Biosciences Clontech (Palo Alto, CA). The expression construct for the chimeric EGFR-H2 receptor encoding a molecule containing the extracellular and transmembrane portion of EGFR and the intracellular domain of HER2 was generated by PCR. In detail, amino acids 1 to 680 of EGFR were fused to amino acids 689 to 1,255 of HER2, creating the fusion sequence HIVRKRTLR/RLLQETEL. The entire coding sequence of the chimeric receptor was subcloned into MSCVneo vector, and the accuracy of the construct was verified by DNA sequencing. The retroviral constructs were transfected into the Phoenix amphotropic packaging cell line using LipofectAMINE Plus reagent according to the recommendation of the manufacturer (Invitrogen, Carlsbad, CA). Viral particles were harvested 48 h after transfection and NR6 cells were infected. Stable populations were selected using the appropriate selection markers.
Western blot analysis. For biochemical characterization, cells were plated in six-well plates. Following serum starvation, EGFR and EGFR-H2 chimera-expressing cells were stimulated with the indicated concentrations of transforming growth factor α (TGFα) for 12 min. HER2/HER3-expressing cells were stimulated with the indicated concentrations of heregulin for 8 min. Total cell lysates were run on SDS-PAGE, and Western blots were probed with various antibodies. Western blots were quantitated by densitometry, and the results are shown as percentage of control. H1666 cells were plated in 12-well plates and treated as above.
Cell proliferation assay. Cells were plated in 96-well plates (NR6, 3,000 per well; H1666, 5,000 per well) and incubated overnight at 37°C. The next day, medium was removed and cells were stimulated with indicated concentrations of ligand in 1% (NR6) or 0.1% (H1666) serum-containing growth medium. After 3 to 4 days, Alamar Blue (Trek Diagnostic Systems, Cleveland, OH) was added to the wells and fluorescence was read using a 96-well fluorometer with excitation at 530 nm and emission of 590 nm. The results are expressed in relative fluorescence units (RFU). For growth inhibition assays, erlotinib or cetuximab was administered with ligand, and assay was done as stated above.
Because constitutive, homodimeric activation of HER2 requires very high HER2 levels, we examined the effects of erlotinib on the HER2 kinase using a chimeric receptor composed of the extracellular and transmembrane domains of EGFR fused to the intracellular domain of HER2 (31). This allowed us to activate HER2 using the EGFR ligand TGFα. The chimeric receptor, which we called EGFR-H2, was stably expressed in the NR6 derivative of the NIH3T3 cell line. NR6 cells lack endogenous EGFR and therefore do not respond to EGFR ligands. Although ligand stimulation was required for activation of the EGFR-H2 chimera, the activation did not result from transactivation by a different member of the HER receptor family. Therefore, EGFR-H2 activation was similar to a homodimeric activation that occurs when full-length HER2 is expressed at high levels. An EGFR-expressing NR6 line was generated as a control (Fig. 1A).
Biochemical and in vitro characterization of EGFR-H2-expressing cells. To validate the use of our engineered cells for drug studies, we first determined whether the EGFR-H2-expressing line responded appropriately to ligand treatment. Cells were stimulated with TGFα and then lysed, and the levels of receptor autophosphorylation were assessed using anti-phosphotyrosine immunoblots. The Western blots were quantitated by densitometry, and the results were then plotted. As shown in Fig. 1, receptor phosphorylation of EGFR-H2 (Fig. 1B) as well as EGFR (Fig. 1C) was seen in a TGFα dose-dependent manner. The EC50 for EGFR-H2 receptor phosphorylation was 0.49 nmol/L, ∼2.5-fold higher than the EC50 value measured for EGFR of 0.19 nmol/L.
The two major downstream signaling pathways that are activated in response to EGFR phosphorylation are the Ras/MAPK and the phosphatidylinositol 3-kinase (PI3K)/Akt pathways. We examined whether the EGFR-H2 chimeric receptor was capable of activating these pathways in NR6 cells. As shown in Fig. 1B, TGFα treatment of cells expressing EGFR-H2 induced phosphorylation of p44/42MAPK in a dose-dependent manner. Similar to the trend for the autophosphorylation of the chimeric receptor, the measured EC50 for the EGFR-H2 cell line was 0.29 nmol/L, whereas the EC50 for activation of p44/42MAPK in the EGFR-expressing cells was 0.17 nmol/L (Fig. 1C). Akt phosphorylation also occurred in EGFR-H2 cells in response to TGFα stimulation. An EC50 of 0.21 nmol/L was observed in the EGFR-H2 cells compared with an EC50 of 0.08 nmol/L in the control EGFR-expressing cells (Fig. 1B and C, respectively).
These data confirmed that the EGFR-H2 kinase can be activated by TGFα and that, once activated, the chimeric receptor transduced a signal similar to that of EGFR. TGFα was ∼2.5 times less potent in activating the EGFR-H2 receptor kinase versus wild-type EGFR based on receptor phosphorylation. It was about 1.7 and 2.6 times less potent in activating p44/42MAPK and Akt, respectively, in EGFR-H2 NR6 cells versus the EGFR NR6 control cells.
Results from cell proliferation experiments were consistent with these biochemical data. EGFR-H2, EGFR, and the parental NR6 cells were treated with various concentrations of TGFα for 3 days and their growth rates were measured using Alamar Blue reduction. As shown in Fig. 1D, the EGFR-H2 cells as well as EGFR-expressing cells proliferated on ligand stimulation with EC50s of 2.1 and 0.59 nmol/L, respectively. As expected, parental NR6 cells did not grow on addition of TGFα.
Erlotinib inhibits the HER2 kinase directly. The biochemical and biological responses of EGFR-H2 cells to TGFα stimulation were very similar to those of the EGFR line, and we therefore considered it as a good model system for examining the effects of the inhibitor erlotinib on the HER2 kinase. EGFR-H2 and control EGFR-expressing cells were pretreated with various concentrations of erlotinib prior to ligand stimulation, and receptor phosphorylation status was analyzed by immunoblots (Fig. 2A). Interestingly, despite the 40- to 3,000-fold difference in activity in in vitro kinase assays (26, 27), erlotinib had a profound effect on the HER2 kinase in our cell-based assays. EGFR-H2 receptor phosphorylation was inhibited by erlotinib with an IC50 of 230 nmol/L. As expected, erlotinib blocked EGFR receptor phosphorylation with an IC50 of 20 nmol/L (Fig. 2B). This measured inhibitory concentration for the EGFR cell line was similar to previously published IC50 data obtained from high EGFR-expressing cancer cell lines (25).
Because erlotinib efficiently blocked receptor phosphorylation in both cell lines, albeit with less potency for the EGFR-H2 kinase, we next examined the effect of erlotinib on the downstream MAPK and Akt signaling pathways. As shown in Fig. 2, erlotinib inhibited phosphorylation of p42/p44 MAPK in EGFR-H2-expressing cells and also in EGFR-expressing cells in a dose-dependent manner. The IC50 values were 290 and 170 nmol/L, respectively. Interestingly, despite the >10-fold difference in IC50 values for receptor phosphorylation for the two cell lines, the IC50 values for p42/p44 MAPK differed by only ∼2-fold. Similar experiments done on breast cancer lines have revealed that inhibition of receptor activation and downstream signaling are not necessarily directly proportional (6). We also measured the inhibitory effects of erlotinib on Akt, a major focal point in the cell survival pathway. The measured IC50 value in EGFR-H2-expressing cells was 210 nmol/L, and the IC50 value in the EGFR line was 40 nmol/L.
Inhibition of cell proliferation. After we established that erlotinib prevented signaling by the HER2 kinase, we examined whether this inhibitory effect was also translated into growth inhibition. EGFR and EGFR-H2 cells were treated with various concentrations of erlotinib in the presence of 3 nmol/L TGFα, and cell proliferation was measured after 3 days using Alamar Blue staining. Consistent with what has previously been shown for various cancer cell lines (32), erlotinib strongly inhibited the growth of EGFR-expressing cells with an IC50 of 260 nmol/L. Erlotinib also inhibited the proliferation of the EGFR-H2 line despite the fact that these cells did not express EGFR. As expected, the potency was lower compared with what was observed for the EGFR-expressing line. Cell growth was inhibited with an IC50 of 1.4 μmol/L in the EGFR-H2 cells (Fig. 2C). In spite of its lower potency against the HER2 kinase, erlotinib effectively blocked proliferation of EGFR-H2 cells.
Generation of HER2/HER3-expressing NR6 cells. As mentioned earlier, the HER2 kinase can also be indirectly activated by ligands through the formation of heterodimeric complexes with other HER receptors that are capable of binding their cognate ligand. We chose to study the effect of erlotinib on the HER2/HER3 heterodimer because HER3 is the preferred partner of HER2 and the resulting complex is more active than any other HER receptor combination in terms of signaling (33) and transforming potential (34, 35).
We engineered NR6 cells to coexpress HER2 and HER3 and confirmed that they would respond biochemically and biologically as expected as we described earlier for the EGFR-H2 cells. In this case, however, the cells were stimulated with the HER3 ligand heregulin. To detect an active receptor complex in the HER2/HER3-expressing line, we stimulated the cells with various concentrations of heregulin for 8 min and then determined HER3 phosphorylation by immunoblots using a HER3 phospho-specific antibody. As shown in Fig. 3A, ligand treatment led to a dose-dependent increase in HER3 phosphorylation. Quantification of the Western blot by densitometry revealed an EC50 of 0.09 nmol/L. Heregulin-induced activation of the receptor complex also resulted in the phosphorylation of the downstream molecules MAPK and Akt. The measured EC50 values were 0.17 and 0.22 nmol/L, respectively (Fig. 3B). The ability of HER2/HER3-expressing cells to proliferate in response to heregulin was also assessed. The cells were treated with various concentrations of heregulin for 3 days, and their growth response was determined by Alamar Blue staining. HER2/HER3-expressing cells responded to heregulin with an EC50 of 0.38 nmol/L, whereas the parental cells did not grow under these conditions (Fig. 3C).
Erlotinib inhibits heregulin-stimulated HER2 kinase activity. HER3 lacks intrinsic kinase activity and cannot transduce a signal by itself (33). Instead, it requires the presence of a partner receptor, such as HER2, which is catalytically active and can be transactivated. For this reason, any inhibitory effects of erlotinib on the HER2/HER3 cell line result from the effects of erlotinib on the HER2 kinase. We first explored whether erlotinib could block ligand-induced receptor phosphorylation in the HER2/HER3-expressing cells. Cells were pretreated with various concentrations of erlotinib prior to ligand stimulation, and HER2/HER3 phosphorylation status was analyzed by immunoblots (Fig. 4A). Erlotinib inhibited the HER2 kinase in the HER2/HER3-expressing cells with an IC50 of 150 nmol/L (Fig. 4B). Erlotinib also inhibited the phosphorylation of p42/p44 MAPK in HER2/HER3-expressing cells in a dose-dependent manner with an IC50 value of 160 nmol/L. Moreover, it also inhibited Akt phosphorylation in HER2/HER3 cells with a measured IC50 value of 720 nmol/L (Fig. 4B).
Inhibition of heregulin-stimulated cell proliferation. We next investigated if the growth of HER2/HER3-expressing cells was also inhibited by erlotinib. Cells were stimulated with 1 nmol/L heregulin in the presence of various concentrations of erlotinib. The extent of cell proliferation was measured after 3 days using Alamar Blue staining. As with the EGFR-H2 line, erlotinib inhibited the proliferation of the HER2/HER3-expressing line with an IC50 value of 690 nmol/L despite the fact that these cells did not express EGFR (Fig. 4C).
Effect of erlotinib on the NSCLC line H1666. After we established that erlotinib was capable of inhibiting the HER2 kinase in the engineered cell lines, we asked whether the inhibitory effect could also be seen in a nonengineered cancer cell line. H1666 cells were chosen because this NSCLC line expresses HER2 and HER3 and responds well to heregulin stimulation (Fig. 5A,, lane 2). Like most other lung cancer lines, H1666 cells also express EGFR and respond to EGFR ligands (Fig. 5A,, lane 11). To examine the effects of erlotinib on HER2 in the absence of EGFR signaling, we blocked the ligand binding and dimerization capability of EGFR with the EGFR ectodomain-specific antibody cetuximab. Cetuximab is a monoclonal antibody that sterically hinders the EGFR ectodomain and prevents it from adopting the extended conformation necessary for ligand binding and dimerization (36). Treating H1666 cells with saturating amounts of cetuximab eliminated EGFR signaling, although the receptor was still physically present on the cells (Fig. 5A , lane 15).
Erlotinib inhibits heregulin-dependent activation of H1666 cells in the presence or absence of cetuximab. Heregulin treatment induced receptor phosphorylation in H1666 cells regardless of whether cetuximab was present or not. As shown in Fig. 5A, addition of 2 nmol/L heregulin resulted in a strong increase of HER3 phosphorylation (Fig. 5A,, lanes 2 and 3) in both the absence and the presence of 20 μg/mL cetuximab. In contrast, the TGFα-induced EGFR receptor phosphorylation was fully inhibited by the presence of 20 μg/mL cetuximab (Fig. 5A , lane 15). These results confirmed that heregulin-mediated signaling, in contrast to TGFα-mediated signaling, does not depend on EGFR.
We next incubated the cells with various concentrations of erlotinib prior to ligand stimulation both in the presence and in the absence of cetuximab. HER3 receptor phosphorylation was completely abrogated with 5 μmol/L erlotinib (Fig. 5A,, lane 6), whereas 1 μmol/L erlotinib decreased receptor phosphorylation by 76% (Fig. 5A,, lane 5). As anticipated and shown previously, erlotinib inhibited TGFα-induced receptor phosphorylation more potently. Treatment with 0.1 μmol/L erlotinib decreased EGFR phosphorylation by 65% (Fig. 5A,, lane 12), and treatment with 1 μmol/L erlotinib abolished receptor phosphorylation completely (Fig. 5A,, lane 13). The presence of saturating concentrations of cetuximab in heregulin-treated cells did not change the activity of erlotinib (Fig. 5A , lanes 7–9), again indicating that EGFR is not participating in this aspect of the inhibitory activity of erlotinib on the HER2/HER3 complex.
The inhibitory effect of erlotinib was also assessed on the downstream target MAPK using phospho-specific antibodies. MAPK was inhibited by erlotinib in the absence (Fig. 5A,, lanes 4–6) or presence (Fig. 5A,, lanes 7–9) of cetuximab, indicating that EGFR signaling is not required for drug sensitivity in heregulin-stimulated cells. Cetuximab alone, however, completely inhibited MAPK activation in TGFα-stimulated cells (Fig. 5A , lane 15).
Erlotinib abrogates heregulin-induced growth in H1666 cells in the presence of cetuximab. Ultimately, we wanted to determine whether erlotinib could inhibit heregulin-driven growth in H1666 cells. H1666 cells were plated in 96-well plates in 0.1% serum-containing medium. The cells underwent different treatments, and cell growth was measured after 5 days using Alamar Blue staining.
As shown in Fig. 5B, heregulin treatment stimulated the growth of H1666 cells, and cetuximab had essentially no effect on this stimulation. TGFα also induced proliferation of the H1666 cells but, as anticipated, the effect was abrogated by the presence of cetuximab. To obtain IC50 values, we treated H1666 cells with various concentrations of erlotinib and measured ligand-dependent cell growth in the presence or absence of cetuximab (Fig. 5C). Heregulin-driven proliferation was inhibited by erlotinib with an IC50 of 700 nmol/L regardless of whether cetuximab was present or not. TGFα-stimulated growth was repressed by erlotinib with an IC50 of 300 nmol/L. As expected, significant cell growth did not occur when TGFα-stimulated cells were treated additionally with cetuximab. The heregulin-stimulated proliferation of H1666 cells in the presence of saturating concentrations of cetuximab indicated that HER2/HER3 heterodimer complexes play an important role in the growth and survival of these cells. The fact that erlotinib suppressed the proliferation in a dose-dependent manner showed that this inhibitor effectively blocked the HER2 kinase, albeit with a lower potency in comparison with its effect on the EGFR kinase.
The engineered EGFR-H2 and HER2/HER3 cell lines enabled us to study the effect of erlotinib on the HER2 kinase in the absence of endogenous EGFR. With this system, we were able to clearly distinguish a direct effect of erlotinib on HER2 from possible indirect effects that might be facilitated through EGFR/HER2 cooperation. Using the EGFR-H2 chimera, we showed that erlotinib can directly inhibit the HER2 kinase when it is activated through the formation of a homodimeric complex. We also showed that erlotinib effectively inhibits the most prominent heterodimeric complex in the HER receptor family—the HER2/HER3 heterodimer. The dose-dependent inhibition was seen at the receptor level, on downstream signaling molecules, and was also translated into inhibition of cell growth.
EGFR tyrosine kinase inhibitor treatment in HER2-overexpressing cells results in decreased basal receptor phosphorylation of EGFR and HER2 (21, 28). Thus far, it is unclear whether this decreased phosphorylation is due to direct inhibition of both HER2 and EGFR receptors or to direct inhibition of EGFR and the subsequent loss of HER2 transactivation.
In this study, we set out to determine whether some portion of the antiproliferative effects of erlotinib can be attributed to direct inhibition of the HER2 kinase. We first showed that, in intact cells, erlotinib was capable of inhibiting an EGFR-H2 chimeric receptor at submicromolar levels following activation by an EGFR ligand. The inhibition at the receptor level translated into the potent inhibition of downstream signaling molecules and resulted in a marked decrease in cell proliferation. The EGFR-H2 chimeric receptor enabled us to recapitulate homodimeric activation of the HER2 kinase; the mode of activation that is predominantly seen in HER2-overexpressing cells. Because NR6 cells lack EGFR expression, we showed that erlotinib blocked the HER2 kinase directly. We found that erlotinib was only ∼12-fold less active against HER2 kinase versus the EGFR kinase in intact cells. These results are in strong contrast to the 40- to 3,000-fold difference in potency that has been seen in purified enzymatic assays. One potential explanation for the substantial difference may be that erlotinib is a far better inhibitor of the HER2 kinase in intact cells because the in vitro kinase format may not recapitulate how HER2 is presented in the cell membrane. Hence, higher IC50 values are obtained in in vitro kinase assays.
The potent inhibition of HER2 kinase by EGFR tyrosine kinase inhibitors has been reported previously (21–23). However, since the in vitro kinase assays suggested that erlotinib or gefitinib was incapable of inhibiting the HER2 kinase at submicromolar levels, an alternative mechanism of action was proposed. In the presence of EGFR, HER2 forms functional heterodimers with EGFR upon ligand stimulation that in turn result in the transphosphorylation of HER2. It was suggested that EGFR inhibitors efficiently hinder the transphosphorylation of HER2 but do not block HER2 function directly. However, based on our data, the activity of EGFR kinase inhibitors on HER2-overexpressing cells can be attributed to the direct inhibition of HER2 kinase. In accordance with our findings, a study by Wels et al. (37) shows that an EGFR-neutralizing antibody is unable to inhibit cell proliferation of HER2-overexpressing SKBR3 cells, yet SKBR3 cells are very sensitive to EGFR tyrosine kinase inhibitors. Because SKBR3 cells express HER2 in large excess over EGFR, it is unlikely that the potent down-regulation of HER2 in SKBR3 cells is solely due to the inhibition of cross-talk between EGFR and HER2. It is more likely that the direct inhibition of HER2, in conjunction with some disruption of EGFR/HER2 cross-talk, leads to the profound antiproliferative effect of EGFR tyrosine kinase inhibitors on HER2-overexpressing cells.
Although HER2 overexpression is observed in 20% to 30% of all breast cancers, the frequency of HER2 overexpression in most other tumor types is low (38). When HER2 levels are low, HER2 signaling is mediated by a second receptor in a heterodimeric complex. Because HER3 is the preferred heterodimeric partner of HER2 and is the most common mediator of ligand-stimulated HER2 activation (2, 33), we examined the inhibitory effects of erlotinib on the heregulin-dependent activation of HER2/HER3 heterodimers in cells that lack EGFR. Inhibition of receptor phosphorylation as well as Akt and MAPK activation was achieved using submicromolar concentrations of erlotinib. Heregulin-induced cell growth was also effectively inhibited in these cell lines. Because HER3 is catalytically inactive and cannot signal by itself, these results indicated that erlotinib directly blocked HER2 kinase and, consequently, prevented HER3 transphosphorylation.
Our findings agree with a previously published report that gefitinib potently inhibits ligand-driven HER2/HER3 signaling in various breast cancer cells (28). Contrary to our interpretation, Anido et al. postulate that inhibition of HER2 occurs through the formation of inactive EGFR/HER2 and EGFR/HER3 heterodimers and not through direct inhibition of HER2. These inactive heterodimers sequester HER2 and prevent the formation of active HER2/HER3 heterodimers. This model, however, implies that only cells that express significant amounts of EGFR in relation to HER2 will respond to EGFR tyrosine kinase inhibitors. Our findings and a recent report describing the potent effect of gefitinib on the inhibition of HER2/HER3 signaling in NSCLC cells that have no detectable EGFR (39) do not support this model.
As mentioned previously, EGFR, HER2, and HER3 activate the PI3K/Akt signaling pathway. The activation of this pathway is most efficiently mediated through HER3 because only HER3 has multiple binding sites for the p85 regulatory subunit of PI3K (40, 41). In contrast, EGFR predominantly activates PI3K via the adaptor protein GrbB2-associated binder 1 (42, 43). Additionally, in the presence of EGFR and HER3, PI3K can be stimulated via ligand-induced EGFR/HER3 heterodimers (44). Gefitinib sensitivity in NSCLC lines has been strongly linked to the down-regulation of the PI3K/Akt pathway, and loss of HER3 activation seems to be an important prerequisite for the sensitivity (45, 46). The authors propose the direct inhibition of EGFR/HER3 as the mechanism of action. Interestingly, all of the sensitive lines also express HER2. Given our data, one can imagine that the potent down-regulation of the survival pathway in HER2-overexpressing cells may also occur via the inhibition of HER2/HER3 heterodimers especially in cases where EGFR expression is low.
In conclusion, we have shown that erlotinib directly blocks HER2 activity regardless of whether HER2 was activated by either homodimeric association or heterodimeric transactivation with HER3. These modes of inhibition are summarized schematically in Fig. 6. In both cases, erlotinib also inhibited the activation of downstream signaling intermediates and suppressed cell proliferation in the absence of EGFR.
Based on our study, we speculate that erlotinib may have a much wider therapeutic potential and that, in addition to EGFR-expressing tumors, it may also be useful against HER2-overexpressing tumors. Pharmacokinetic studies determined that a daily oral dose of 150 mg erlotinib, which is the maximal tolerated dose, translates into a mean serum concentration of 3 to 10 μmol/L (47, 48). Because ∼90% of erlotinib is bound to serum protein, the free drug concentration is approximately 0.3 to 1 μmol/L. Therefore, it is unlikely that erlotinib effectively inhibits HER2 activity in a wide range of patients. Yet, some individuals might benefit from this therapy because plasma concentrations of erlotinib vary substantially among individuals (49). If higher drug exposure could be achieved potentially through pulse dosing or i.v. administration, erlotinib may very well efficiently block HER2 function in a substantial number of patients bearing tumors that are driven by either EGFR or HER2 activation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank David Wood for assistance with graphics.