Introduction
The epidermal growth factor receptor (EGFR), a member of the erbB proto-oncogene family of receptor tyrosine kinases, which also includes erbB2, erbB3 and erbB4, is a transmembrane glycoprotein composed of an extracellular ligand-binding domain and an intracellular domain containing tyrosine kinase activity [
1,
2]. Activation of EGFR results from binding of epidermal growth factor-related growth factors, such as epidermal growth factor, transforming growth factor alpha (TGFα) and amphiregulin, which induce receptor homodimerization and/or heterodimerization with other members of the erbB receptor family [
2]. No direct ligand for erbB2 has yet been identified; however, erbB2 plays a central role in erbB receptor function as it is the preferred dimerization partner for all other erbB family members [
3,
4].
Receptor dimerization stimulates the intrinsic receptor tyrosine kinase activity and promotes autophosphorylation of tyrosine residues within the cytoplasmic domain of the receptor. These phosphotyrosine residues provide docking sites for a variety of adaptor proteins and enzymes involved in the recruitment and activation of downstream intracellular signalling cascades, including the mitogen-activated protein kinase (MAPK) and phosphatidylinositol-3 kinase (PI3K) pathways [
2]. These signalling cascades can promote proliferation, angiogenesis and invasion, and can inhibit apoptosis, key mechanisms underlying tumour growth and progression [
5]. This oncogenic potential in conjunction with the aberrant expression and/or activation of EGFR, which has been reported in a range of human malignancies including breast cancer, provides a strong rationale for targeting this growth factor receptor [
6,
7].
A number of agents targeting EGFR have now been developed and include the monoclonal antibody cetuximab, which targets the extracellular ligand-binding domain of EGFR, and the small molecule tyrosine kinase inhibitors gefitinib (Iressa, ZD1839) and erlotinib (Tarceva, OSI-774), which competitively block binding of ATP to the tyrosine kinase domain of the receptor [
8]. These compounds are proven effective antitumour agents as monotherapies in both the preclinical and clinical setting, and have been shown to enhance the effects of cytotoxic agents and radiation when utilized in combination with these conventional chemotherapies [
8‐
10]. Consequently, cetuximab, gefitinib and erlotinib have now all gained approval for cancer treatment in the clinic. Recent findings from clinical trials, however, have revealed that only a small cohort of patients have derived significant benefit from these therapies, with both
de novo and acquired resistance to these agents being evident [
11‐
13]. Furthermore, evidence of resistance to anti-EGFR therapies has now also been reported in preclinical cell models [
12,
13]. A number of potential resistance mechanisms have now been implicated, including receptor mutation, loss of downstream effector components and activation of alternative oncogenic signalling pathways [
12,
14].
A common resistance mechanism to anti-EGFR therapies identified by a number of research groups in preclinical cancer models of the colon, the breast, the prostate and the brain involves activation of the PI3K/AKT signalling pathway either as a consequence of the loss of phosphatase and tensin homologue or of increased insulin-like growth factor type 1 receptor activity [
12,
15‐
19]. PI3K/AKT signalling pathway activity can also be driven through activation of erbB3 and erbB4 receptors via their ability to directly recruit the p85 regulatory subunit of PI3K [
20,
21]. erbB3 and erbB4 are receptors for the neuregulin family of growth factors, which consists of NRG1 or heregulin (HRG), NRG2, NRG3 and NRG4. These neuregulins have been shown to potently drive cell proliferation, survival, motility and invasion in a range of cancer cell types [
22]. In breast cancer, HRG has been reported to be detectable in clinical tissue [
23,
24] and has been identified as a mitogenic and proinvasive factor in preclinical studies both
in vitro and
in vivo [
25‐
31].
Previous studies from our group have established that growth of a tamoxifen-resistant MCF-7 (Tam-R) breast cancer cell line is driven by the autocrine release of the epidermal growth factor-like ligand amphiregulin and activation of the EGFR/MAPK signalling pathway [
32‐
34]. Furthermore, activation of this signal transduction mechanism has also been implicated in the increased motility and invasiveness observed in this cell line [
35]. Both Tam-R cell growth and invasion are therefore sensitive to the inhibitory actions of the selective EGFR tyrosine kinase inhibitor gefitinib [
32‐
35]. In the present study we have investigated whether HRGβ1 can overcome the inhibitory effects of gefitinib on Tam-R cell growth and invasion, and we have examined the potential pathways activated by this erbB3/4 ligand. We have further assessed the expression levels of HRGβ1 in a series of 77 primary breast cancer samples from patients who had received no prior therapy to assess whether such signalling may be important in the clinical setting.
Discussion
Despite the clear therapeutic promise of anti-EGFR therapies in the preclinical setting, with agents such as gefitinib potently inhibiting growth of a range of human cancer cell lines that express a functional EGFR, there is increasing evidence – from both preclinical and clinical studies – of primary/
de novo and acquired resistance to these inhibitors [
11‐
13]. A key mechanism implicated in resistance to anti-EGFR agents is activation of the PI3K/AKT signalling pathway either as a result of loss of phosphatase and tensin homologue function or of enhanced insulin-like growth factor type 1 receptor activity [
12,
15‐
19]. More recently it has been reported that, in human MKN7 gastric cancer cells, HRGβ1 can also circumvent the antiproliferative action of the selective EGFR-TKI CGP59326 through promotion of erbB3/erbB2 heterodimerization and activation of the PI3K signalling pathway [
44]. In the present study we have, for the first time, examined whether HRGs can similarly promote erbB3/PI3K/AKT signalling and effectively circumvent the inhibitory effects of the anti-EGFR agent gefitinib on the growth and invasion of an EGFR-positive Tam-R breast cancer cell line [
32].
In agreement with our previous findings, the Tam-R cell line was found to express erbB3 at high levels, whereas only minimal expression of erbB4 was observed in this cell line [
32]. erbB3 expression was localized primarily at the cell membrane, although some cytoplasmic staining for this receptor was also observed by immunocytochemistry. We also attempted to assess expression of HRGβ1, a key erbB3 ligand, in this cell line by RT-PCR but were unable to detect a signal. It has previously been reported that MCF-7 cells do not synthesize HRGs [
45], and our findings now indicate that the inability to generate this ligand is maintained through acquisition of tamoxifen resistance in this cell line. The lack of any autocrine release and action of heregulins by the Tam-R cell line would explain the low levels of erbB3/erbB2 heterodimers detected under basal growth conditions. Although low levels of erbB3 activity were also observed under these conditions, this is a result of the formation of erbB3/EGFR heterodimers – most probably a consequence of the autocrine release and action of the EGFR ligand amphiregulin, which we have previously shown to be synthesized and released by this cell line [
34]. Indeed, this hypothesis is supported by the fact that basal erbB3 activity could be reduced by treatment of the cells with the selective EGFR tyrosine kinase inhibitor gefitinib. Basal erbB4 activity and heterodimerization could not be detected in these cells (data not shown).
In agreement with our previous findings, treatment of Tam-R cells with gefitinib (1 μM) blocked the basal phosphorylation of EGFR, reduced the basal formation of active EGFR/erbB3 heterodimers and inhibited the basal activation of both AKT and ERK1/2 [
32,
34,
41]. The reduced basal EGFR/MAPK/AKT signalling activity in response to gefitinib was also associated with a significant reduction in both Tam-R cell proliferation and invasion, again as previously reported, confirming the central role played by EGFR signalling in this cell line [
32,
35]. Interestingly, we also found that gefitinib was capable of reducing phosphorylation levels of erbB2, an observation we have reported on previously [
32]. Gefitinib is highly selective for EGFR (median inhibitory concentration = 0.033 μM) but can inhibit erbB2 tyrosine kinase activity with a median inhibitory concentration of approximately 3–4 μM [
46]. At a concentration of 1 μM, however, gefitinib should have little direct effect on erbB2 tyrosine kinase activity in this study. More recently it has been proposed that gefitinib can indirectly inhibit erbB2 activity through inducing the formation of inactive EGFR/erbB2 heterodimers in erbB2-overexpressing BT-474 cells [
47]. In agreement with Anido and colleagues [
47], we found that gefitinib promoted the formation of inactive EGFR/erbB3 and EGFR/erbB2 heterodimers in our Tam-R cells. This ability of gefitinib to sequestrate erbB2 into an inactive complex with EGFR thus prevented activation of erbB2 through dimerization with other erbB family members.
A possible consequence of the sequestration of erbB2 and erbB3 into inactive heterodimers with EGFR is a reduction in the availability of erbB2 and erbB3 within the cell. Indeed, it has been reported that basal levels of erbB2/erbB3 heterodimers are greatly reduced following treatment of BT-474 breast cancer cells with gefitinib, resulting in a substantial blunting of response to HRGβ1 treatment [
47]. In agreement with these findings we demonstrated in the present study that levels of erbB2/erbB3 heterodimers were reduced following gefitinib treatment in Tam-R cells and that gefitinib prevented HRGβ1-induced activation of EGFR and ERK1/2 while reducing HRGβ1-induced activation of erbB2 and AKT in this cell line. Since this effect on erbB2 and AKT activity cannot be attributed to the tyrosine kinase inhibitory activity of gefitinib because erbB3 is kinase dead [
48] and, as mentioned previously, far higher concentrations of this inhibitor are required to block erbB2 activity, it is more probably a consequence of the reduced erbB2 and erbB3 receptor availability due to gefitinib-induced sequestration with EGFR. It should also, however, be noted that although HRGβ1 signalling is reduced by gefitinib treatment in Tam-R cells, this ligand was still capable of promoting the formation of active erbB2/erbB3, activating AKT signalling and potently driving cell growth.
These findings clearly indicate that although EGFR is the dominant pathway driving growth in Tam-R cells, if this pathway is blocked then these cells are capable of utilizing alternative signalling pathways such as the HRGβ1/erbB3/AKT pathway to circumvent this blockade and to very effectively maintain cell growth. Such a resistance mechanism to anti-EGFR therapy is not unique to Tam-R cells as it has also been shown in MKN7 gastric cancer cells that HRGs can overcome the growth inhibitory actions of the EGFR tyrosine kinase inhibitor CGP59326 through activation of downstream AKT and ERK1/2 signalling pathways [
44]. It is also important to point out that although HRGβ1 only partially recovered Tam-R cell growth in the presence of gefitinib, this ligand fully circumvented the blockade of Tam-R cell invasion generated by this anti-EGFR agent. HRGβ1 signalling therefore appears as effective as EGFR signalling in driving invasion of Tam-R cells, suggesting that targeting EGFR in the presence of HRGs will have minimal effect on the aggressive phenotype of this cell line.
We next examined the effect of targeting components of the HRGβ1 signalling pathway in Tam-R cells to assess whether such a strategy, in combination with gefitinib, would more effectively reduce growth and invasive capacity. We initially targeted erbB2 using the monoclonal antibody trastuzumab. It has previously been demonstrated that combined targeting of EGFR and erbB2 is more effective than targeting either agent alone in inhibiting growth of erbB2-overexpressing breast cancer cells [
44,
49‐
51]. We similarly found, in the present study, that Tam-R cells demonstrated a small reduction in HRGβ1-induced erbB2 and ERK1/2 signalling following treatment with trastuzumab and gefitinib in combination compared with gefitinib alone. This effect of the combination treatment on HRGβ1 signalling activity was also associated with a small but not significant reduction in HRGβ1-driven Tam-R cell growth when compared with the effects of gefitinib alone. This combination strategy, however, was without further significant effect on HRGβ1-induced AKT activity compared with the effects of gefitinib alone, and induced a substantial and significant enhancement of, rather than an inhibition of, HRGβ1-driven Tam-R cell invasion. Targeting erbB2 with trastuzumab in combination with gefitinib to block HRGβ1 signalling therefore had only a minimal effect on cell growth and also had the potential to make the cells substantially more aggressive in nature.
Interestingly, a recent phase I–II clinical study assessing the effect of trastuzumab in combination with gefitinib in patients with erbB2-overexpressing metastatic breast cancer reported that the time to progression was shorter for the combination compared with trastuzumab alone [
52]. Furthermore, this combination treatment was shown to increase erbB3 activity in erbB2-overexpressing breast cancer cells [
52]. Taken in conjunction with the present findings it is possible that HRGβ1/erbB3-mediated signalling in the presence of trastuzumab and gefitinib may maintain cell growth and promote enhanced invasive features within the tumour cells, potentially explaining the reduced time to progression observed in patients receiving this treatment regime.
A more effective combination therapy that inhibited both HRGβ1-driven Tam-R cell growth and invasion was gefitinib in conjunction with the PI3K inhibitor LY294002. Treatment of Tam-R cells with this combination potently knocked out activation of both AKT and ERK1/2. These findings, alongside the inability of trastuzumab and gefitinib to block HRGβ1-induced AKT activity, would suggest that the PI3K/AKT signalling pathway is an important mediator of HRGβ1-driven Tam-R cell invasion and appears to be activated by an EGFR-independent and erbB2-independent mechanism in these cells. Indeed, heregulin-induced PI3K/AKT signalling has been implicated as a dominant signalling pathway driving cell invasion in a range of breast cancer cells [
25,
26,
28,
31]. It should also, however, be noted that the enhanced invasive capacity of the Tam-R cell line in response to trastuzumab and gefitinib in combination was not associated with an increase in AKT activity; indeed, if anything, AKT activity was slightly reduced in the presence of the two inhibitors. As we have found that this enhanced invasive response to HRGβ1 in the presence of both gefitinib and trastuzumab remains sensitive to inhibition by LY294002 (data not shown), it is therefore probable that other PI3K-dependent mechanisms are also involved in mediating this process. Identifying the exact mechanisms involved in the EGFR/erbB2-independent HRGβ1-mediated activation of PI3K-dependent signalling and in the enhanced invasive behaviour in our Tam-R cell line remains a subject of intense investigation within our laboratory.
Finally, we examined whether the HRGβ1-mediated gefitinib-resistant mechanism identified in our Tam-R cells may also be functional within breast tumours by assessing HRGβ1 expression in a set of 77 breast cancer tissue samples from breast cancer patients who had received no previous therapies. We found that all samples expressed predominantly cytoplasmic HRGβ1, with this staining being highly heterogeneous between tumour samples. These findings are in agreement with other reports assessing HRGβ1 expression in primary breast cancer specimens [
23,
24], with both of these studies demonstrating exclusively cytoplasmic staining for this ligand. Interestingly, in the present study the highest expression levels of cytoplasmic HRGβ1 were found in patients with tumours that expressed high levels of membrane EGFR and increased activity of the EGFR-associated signalling elements, membrane erbB2, nuclear MAPK and nuclear AKT.
Interestingly, there was no correlation between expression of cytoplasmic HRGβ1 expression and either membrane erbB2 or erbB3 in these samples. It should be noted that Visscher and colleagues [
23] were unable to identify any correlations between EGFR and HRG expression in their primary tumour specimens; however, this may simply reflect the very small dataset (34 samples) used in the study [
23]. A profile of high EGFR expression and increased activation of downstream MAPK and AKT signalling pathways, which mimics the expression profile observed in our Tam-R model, would suggest that these patients may benefit from anti-EGFR-targeted therapy. Since HRGβ1 is also highly expressed in these samples, however, our model would predict that targeting the EGFR with agents such as gefitinib would not prove effective due to the ability of HRGβ1 to overcome EGFR blockade and to potently drive resistant growth and invasion. Indeed, HRGβ1 levels were also shown to be significantly higher in distant metastatic versus locally advanced patients, suggesting a possible link between this ligand and the invasive capacity of tumour cells.
Competing interests
SEH, JMWG and RIN are in receipt of funding from AstraZeneca, and RIN is also a member of an advisory board for AstraZeneca. The remaining authors declare that they have no competing interests.
Authors' contributions
IRH conceived the study and participated in its design and execution, and in drafting the manuscript. JMK carried out the PCR, participated in the immunoprecipitation and western blotting studies, and helped to draft the manuscript. SEH carried out the invasion assays. DB performed all cell cultures and carried out the growth studies. JMWG carried out the immunocytochemistry and helped to draft the manuscript. JFR and IOE provided the clinical breast cancer samples. RIN participated in the design and coordination of the study and helped to draft the manuscript. All authors read and approved the manuscript.