Introduction
There is strong experimental and clinical evidence implicating the insulin-like growth factor type I receptor (IGF-IR) in breast cancer development and growth [
1‐
3]. The IGF-IR, which belongs to a family of receptor tyrosine kinases that includes the insulin receptor (IR), has been found to be expressed in a high percentage of breast tumours, where its expression is positively correlated with oestrogen receptor (ER) status and is usually coexpressed with markers of a better overall prognosis [
2,
4‐
6]. Expression of the IGF-IR has also been demonstrated in the majority of ER+ breast cancer cell lines [
7,
8]. Indeed, in MCF-7 cells, IGF-IR has been shown not only to be a key receptor in mediating hormone-sensitive growth but also to engage in significant cross-talk with ER [
9,
10]. Importantly, this leads to synergistic interactions between ER and IGF-IR signalling to promote efficient growth responses [
2].
However, conversely, increased expression and activation of IGF-IR and its associated downstream signalling components have also been reported in some clinical breast cancers and have been linked to disease progression and recurrence [
11,
12]. On the basis of these data, IGF-IR has been identified as a potential therapeutic target for the treatment of breast cancer [
13]. Activation of the IGF-IR promotes binding of insulin receptor substrate (IRS) members, a family of structurally related adaptor molecules which have classically been identified as key signalling intermediates of the IR and IGF-IR [
14]. Binding results in phosphorylation of their carboxyl termini at multiple tyrosine residues, and these phosphotyrosine residues provide docking sites for the recruitment of key signalling pathways, such as the mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase 1/2 (ERK1/2) and phosphatidylinositol 3-kinase (PI3K) pathways [
15]. These signalling cascades can mediate mechanisms underlying tumour growth and progression, implicating a potential role for IRS members in oncogenesis [
1,
15‐
18]. Indeed, IRS-1 has been reported to be overexpressed and constitutively phosphorylated in breast tumours [
18,
19], and high expression of this adaptor protein has been associated with lymph node metastases and poor patient prognosis [
11,
20,
21]. Furthermore, IRS-1 and IRS-2 have been implicated in the regulation of proliferation, survival and metastatic potential in a range of breast cancer cell lines [
17].
However, there is now increasing evidence that IRS-1 is not restricted to binding to IR/IGF-IR but is also capable of associating with a variety of other signalling-related proteins [
17]. One such protein is the epidermal growth factor receptor (EGFR), a member of the erbB receptor tyrosine kinase family also comprising erbB2, erbB3 and erbB4 and which has been shown to play a central role in driving both
de novo and acquired anti-hormone-resistant growth and invasion in breast cancer [
22‐
25]. Evidence of an EGFR-IRS-1 interaction arises from reports by Fujioka and colleagues [
26,
27], who reported that the phosphorylated NPXY motifs in activated IR and IGF-IR to which IRS molecules bind are also present in the C-terminus region of activated EGFR and were indispensable for EGF-induced IRS-1 tyrosine phosphorylation in EGFR-transfected CHO cells [
27]. Furthermore, a potential interaction between EGFR and IRS-1 has been predicted on the basis of the binding of peptides, representing the physical sites of EGFR tyrosine phosphorylation, to protein microarrays comprising all Src homology 2 and phosphotyrosine binding domains encoded in the human genome [
28]. Recently, we provided strong evidence that IRS-1 can function as a key signalling intermediate for EGFR, a receptor that drives the growth of a tamoxifen-resistant MCF-7 breast cancer cell line [
29]. In these cells, we showed that IRS-1 physically complexes with EGFR and is preferentially phosphorylated on Y896, a Grb2-binding/MAPK recruitment site [
15]. Moreover, EGFR was the dominant recruiter of IRS-1, which thus served to limit the availability of IRS-1 to associate with IGF-IR in these cells and, as a result, suppressed IGF-IR signalling via this receptor.
Other erbB receptors are also prevalent in breast cancer, and their interplay with IRS-1 remains unknown. Of note, we have previously shown [
30] that erbB3 is commonly expressed in clinical breast tumours alongside one of its ligands, heregulin β1 (HRGβ1). Interestingly, erbB3 also possesses the NPXY motifs recognized by IRS proteins [
31] and as such may bind IRS-1 in breast cancer cells. Indeed, such an association has again been predicted on the basis of protein microarray studies [
28]. A role for erbB3 in breast cancer has only recently become appreciated [
32,
33], with overexpression of erbB3 shown to be positively associated with metastasis [
34], increased histological grade [
35] and tumour recurrence [
36]. There is also growing awareness of the importance of the erbB2/erbB3 heterodimer in breast cancer progression. Heterodimers between these two receptors have been shown to form the most potent mitogenic and transforming receptor complex
in vitro [
37], and coexpression of erbB2 and erbB3 have been shown to be significantly associated with decreased survival in breast cancer patients [
38]. Interestingly, erbB3 signalling has also been implicated in mediating resistance to IGF-IR-targeted agents in hepatocellular carcinoma cells [
39], but whether it plays a similar role in breast cancer remains to be determined. In the present study, using a panel of ER+ breast cancer cell lines, we examined for the first time whether IRS-1 can contribute to erbB3 signalling in breast cancer and what impact this may have on IGF-IR signalling. We show that IRS-1 is recruited to erbB3 following HRGβ1 treatment in these cells and demonstrate that this novel interaction can serve to reduce the association between IRS-1 and IGF-IR and inhibits signalling via this receptor. We show in turn that suppression of IGF-IR by the use of a tyrosine kinase inhibitor and siRNA technology can promote erbB3 downstream signalling by reinforcement of erbB3 interplay with IRS-1. This provides a potential novel resistance signal, which, when targeted, may generate more effective inhibition of cell growth compared to IGF-IR treatment alone.
Materials and methods
Cell culture
MCF-7 cells (a gift from AstraZeneca Pharmaceuticals, Cheshire, UK) and T47D cells (American Type Culture Collection (ATCC), Manassas, VA, USA), which are both nonamplified erbB2 breast cancer cell lines, were grown in RPMI medium containing 5% FCS and glutamine (4 mM). Both cell lines were maintained at 37°C in a humidified 5% CO2 atmosphere. BT-474 over amplified erbB2 breast cancer cells (ATCC) were grown in RPMI medium containing 10% FCS and glutamine (4 mM).
Experimental procedures
The cell lines were grown to 70% confluence prior to transfer into phenol red/steroid-and serum growth factor-free dendritic cell conditioned medium (Biosynergy Europe, Cambridge, UK) for 24 hours followed by exposure for up to 20 minutes to either 0.1 to 10 ng/ml HRGβ1 or 10 ng/ml IGF-I in 10 mM acetic acid/0.1% BSA or appropriate vehicle control. To examine the effects of pharmacological blockade of IGF-IR, cells were incubated in phenol red-free (white) RPMI medium supplemented with 5% FCS and either the IGF-IR/IR tyrosine kinase inhibitor 4-anilino-5-bromo-2-[4-(2-hydroxy-3-(
N,
N-dimethylamino)propoxy)anilino]pyrimidine (ABDP) (1 μM in dimethyl sulphoxide, AstraZeneca, Macclesfield, UK) [
40] or appropriate vehicle control for 1 to 2 days. All experiments were performed at least three times. Cells were then lysed to measure protein expression.
siRNA studies
Dharmacon SMARTpool siRNA Design specific for IRS-1 (IRS si), erbB3 (3 si) or IGF-IR (IGF si; all 20 mM, Dharmacon RNAi Technologies, Lafayette, CO, USA) were mixed with DharmaFECT 1 transfection reagent (lipid; Dharmacon RNAi Technologies) at a ratio of 10 μl of siRNA to 1 μl of lipid and incubated at room temperature for 20 minutes. The mix was added to the cells, which were maintained in white RPMI medium containing 5% FCS to give a final siRNA concentration of 100 nM per dish. Control experiments consisted of transfection with the ON-TARGETplus Nontargeting siRNA control pool (100 nM; Dharmacon RNAi Technologies), medium only (nontransfected cells) or lipid. All experimental arms were set up in duplicate. Cells were incubated in growth medium containing either IRS si, 3 si, IGF si or control (C si) (100 nM for each) for 4 days prior to treatment with either 10 ng/ml HRGβ1 or vehicle alone for 5 minutes. To examine the effect of IRS-1 knockdown and IGF-IR blockade, cells were incubated in medium containing either 100 nM C si, 100 nM IRS si, 1 μM ABDP or a combination of these treatments prior to a 5-minute incubation with HRGβ1 (10 ng/ml). The cells were then lysed, and protein extracts were examined by Western blot analysis.
Immunoprecipitation and Western blot analysis
Fresh cell lysates containing 500 μg of protein were immunoprecipitated using 1 μg of specific antibody as described previously [
24]. Protein samples from either immunoprecipitation or total cell lysates (20 to 50 μg) were separated on a 7.5% polyacrylamide gel and then transblotted onto nitrocellulose membrane as described previously [
24]. The antibodies used were directed against total EGFR (SC-03), total erbB2 (SC-284), total erbB3 (SC-285), total IGF-IR (SC-712), total IRS-1 (SC-7200; Insight Biotechnology Ltd, Wembley, UK), total and phosphorylated Akt YS473, ERK1/2 and phosphorylated c-erbB3 Y1289 (Cell Signaling Technology, Hitchin, UK), phosphorylated EGFR (Y1068), phosphorylated IRS-1 (Y612 and Y896; BioSource International, Camarillo, CA, USA), β-actin (Sigma-Aldrich, Dorset, UK) and specific phosphorylated IGF-IR Y1316 (a kind gift from AstraZeneca, Macclesfield, UK). The Western blots were then scanned by densitometry to provide data for semiquantification. Each experiment was performed at least three times with representative gels shown in figures.
Cell proliferation studies
Cells were seeded at 40, 000 cells per well overnight in phenol red-free RPMI medium supplemented with 5% FCS and then incubated in fresh medium containing 0.1 to 1 μM ABDP, 10 ng/ml HRGβ1, vehicle control or a combination of these agents for 4 days. Cell population growth was evaluated by means of trypsin dispersion of the cell monolayers, and cell number was measured using a COULTER COUNTER (Beckman Coulter (UK) Ltd, High Wycombe, UK). All experiments were performed in triplicate.
Clinical series
A small historical series of 50 primary tumours were excised from ER+ patients with histologically proven breast cancer who had presented for surgery at Nottingham City Hospital (Nottingham, UK) from 1984 to 1987. Representative tissue samples from these tumour samples were fixed routinely in 4% formal saline and embedded in paraffin. No patient had previously received any form of adjuvant endocrinological or cytotoxic therapy. The use of these samples for research purposes, without the requirement of further patient informed consent, was approved by Nottingham Research Ethics Committee 2 under the title 'Development of a molecular genetics classification of breast cancer' (C2020313).
Immunocytochemical assays
Immunocytochemical assays for phosphorylated Akt, Ki-67 and total erbB3 and specific phosphorylated IGF-IR Y1316 were performed on 50 primary ER+ breast tumours as previously described [
6,
30,
41] and the clinicopathological parameters for the clinical set of these tumours are as shown (Additional file
1, Table S1). For the detection of phosphorylated IRS-1 Y612, paraffin wax sections from each tumour sample were dewaxed using xylene treatment and then rehydrated through graded ethanols to PBS. Endogenous peroxidases were destroyed by immersing the sections in 3% hydrogen peroxide prepared in methanol for 5 minutes, followed by rinsing with distilled water for 5 minutes. Antigen retrieval was achieved by pressure cooking the slides in 0.01 M sodium citrate buffer, pH 6.0, for 4 minutes. Slides were then immersed in slowly running tap water for 10 minutes before being transferred to PBS for 5 minutes. Sections were blocked in 1% l BSA for 5 minutes prior to incubation overnight at 37°C in anti-phosphorylated IRS-1 Y612 rabbit primary antibody diluted 1:50 in PBS. Sections were washed for 3 minutes in PBS, washed twice for 5 minutes in 0.02% PBS-Tween 20 and then incubated for 2 hours at room temperature in a peroxidase-labelled polymer secondary antibody EnVision Kit (Dako Ltd, Ely, UK). Slides were then washed for 3 minutes in PBS, washed twice for 5 minutes in 0.02% PBS-Tween 20 and incubated for 10 minutes at room temperature in EnVision DAB chromogen solution (diaminobenzidine; Dako Ltd). Slides were then rinsed twice for 2 minutes in distilled water, incubated in 0.5% methyl green for 25 seconds as a counterstain, rinsed in distilled water and allowed to air-dry before DPX mountant (a mixture of distyrene, a plasticizer and xylene) was applied. Combined cytoplasmic and plasma membrane staining intensity and percentage positivity were assessed by HScore analysis as described previously [
41]. Expression of phosphorylated membrane and cytoplasmic IRS-1 Y612 has previously been verified by immunocytochemistry in ER+ breast cancer cell lines [
29].
Statistics
For immunocytochemical analysis of clinical material, Hscores were compared using Spearman's rank-correlation and Mann-Whitney U tests for nonparametric data. For experimental growth studies, overall differences between the control and treatment groups were determined by analysis of variance with post hoc t-tests with the Bonferroni adjustment factor. A two-sided t-test was performed on the densitometry values obtained following Western blot analysis. Differences were considered significant at the P ≤ 0.05 level.
Discussion
IRS-1 is not restricted to binding to IR/IGF-IR but also has the capacity to interact with a variety of other proteins [
21]. Recently, we reported that IRS-1 can interact with EGFR, resulting in loss of recruitment of IRS-1 by IGF-IR and reducing signalling via this receptor in an ER+, tamoxifen-resistant MCF-7 breast cancer cell line [
29]. In the present study, we examined whether IRS-1 can associate with other erbB family members, notably erbB3, and whether this has a direct impact on IGF-IR signalling in three ER+ breast cancer cell lines (MCF-7, T47D and BT-474) previously shown to express IRS-1 protein [
42‐
44].
Initial characterisation of these cell lines showed that EGFR, erbB2, erbB3 and associated downstream signalling elements MAPK and Akt were activated following HRGβ1 treatment, with this ligand having a more potent effect on phosphorylation levels in MCF-7 and T47D cells that on BT-474 cells. Interestingly, HRGβ1 treatment also increased levels of IRS-1 phosphorylation at both the Y612 and Y896 residues, with this effect being greater in MCF-7 and T47D cells than in the BT-474 cell line. The more modest effect of HRGβ1 priming of such activity in BT-474 cells most likely reflects the fact that these cells constitutively overexpress erbB2 and consequently have higher basal phosphorylation levels of all these signalling elements. As such, any increase in activity is harder to distinguish compared to the erbB2 low-expressing MCF-7 and T47D cell lines [
45]. Using immunoprecipitation and Western blot analysis, we confirmed that HRGβ1-induced phosphorylation of IRS-1 was a result of IRS-1's complexing with erbB3/EGFR and erbB3/erbB2 heterodimers in both MCF-7 and T47D cells. The ability of erbB3 to heterodimerise with both EGFR and erbB2 in response to HRGβ1 stimulation explains the increased phosphorylation of IRS-1 at Y896 in these two cell lines. We have previously described the recruitment and phosphorylation of IRS-1 at this tyrosine residue by EGFR/erbB2 heterodimers in a tamoxifen-resistant MCF-7 breast cancer cell line [
29]. We have previously reported that phosphorylation of IRS-1 Y612 results from recruitment and activation by IGF-IR. In the present study, however, HRGβ1-induced IRS-1 Y612 phosphorylation appeared to be IGF-IR-independent. There was no effect of this ligand on IGF-IR phosphorylation, as verified by the use of a specific pY1316 IGF-IR antibody in these cell lines [
40]. Indeed, HRGβ1 treatment reduced the association of IRS-1 with IGF-IR in both cell lines. This leaves association of IRS-1 with erbB3 as the likely mediator of HRGβ1-induced IRS-1 Y612 phosphorylation in these cells.
It has previously been reported in other systems that IRS-1-erbB3 interactions can occur, as erbB3 possesses NPXY motifs within its C-terminal domain, like those observed in IGF-IR/IR, which are recognized by IRS proteins and would enable this adaptor molecule to potentially bind to this receptor [
31]. Furthermore, in a study of the binding of peptides representing the physical sites of erbB3 tyrosine phosphorylation to protein microarrays comprising all Src homology 2 and phosphotyrosine binding domains encoded in the human genome, researchers predicted a potential interaction between erbB3 and IRS-1 [
28]. Importantly, our studies reveal that IRS-1 has a significant functional role in erbB3 signalling in MCF-7 and T47D cells, as erbB3 knockdown using siRNA potently inhibited basal and HRGβ1-induced IRS-1, Akt and ERK1/2 phosphorylation, whilst IRS-1 siRNA similarly reduced HRGβ1-induced Akt and, to a modest degree, ERK1/2 activity in these cells. As ERK1/2 activity was not significantly altered following IRS-1 knockdown, this would suggest that an IRS-1-independent mechanism underlying HRGβ1-induced ERK1/2 activity was at work in our cell lines. Consequently, the remainder of our study focused primarily on IRS-1 Y612/Akt phosphorylation, as this appeared to be the IRS-1-dependent pathway in response to HRGβ1 in our cell models. In BT-474 cells, there was a strong basal association between IRS-1 and erbB3, as observed in immunoprecipitation studies, which could not be enhanced further by exogenous ligand stimulation. Again, this could be due to the high constitutive erbB2 activity present within these cells masking the exogenous stimulatory effects of HRGβ1 treatment. Moreover, IRS-1 itself may potentially be more freely available to interact with erbB3 in these cells, as they have somewhat less IGF-IR protein with which to associate compared to MCF-7 and T47D cells, as shown in this study and as shown elsewhere previously [
46]. As this HRGβ1-induced association between IRS-1 and erbB3 was not evident in the BT-474 cells, these cells were omitted from further study.
As mentioned previously, another interesting phenomenon noted in these studies is the finding that whilst HRGβ1 treatment enhanced erbB3-IRS-1 interactions, it also promoted a decrease in the association between IRS-1 and IGF-IR, an effect that was clearly apparent in the MCF-7 and T47D cell lines. This finding suggests that the enhanced physical interaction between erbB3 and IRS-1 following HRGβ1 treatment may serve to limit the availability of IRS-1 to associate with IGF-IR, potentially resulting in inhibition of signalling via this receptor. Indeed, as mentioned above, we have previously reported that EGFR can similarly suppress IGF-IR signalling through such a mechanism in a tamoxifen-resistant MCF-7 cell line [
29]. A potential consequence of the ability of HRGβ1/erbB3 signalling to suppress IGF-IR signalling activity is that such a mechanism could severely affect the efficacy of IGF-IR-targeted agents in these breast cancer cells. Indeed, there is now evidence emerging from experimental breast cancer cell models implicating a role for erbB receptors in resistance to IGF-IR blockade, with Haluska and colleagues [
47] showing that EGFR/erbB2 signalling can confer resistance to the IGF-IR tyrosine kinase inhibitor BMS-536924 in MCF-7 cells. In our present study, a role for erbB3 signalling in resistance to IGF-IR blockade is also clearly implicated, as HRGβ1 readily overcame the growth-inhibitory effects of the IGF-IR/IR tyrosine kinase inhibitor ABDP in the MCF-7 and T47D cell lines. ABDP is a novel dual IGF-IR/IR tyrosine kinase inhibitor that has previously been reported to potently inhibit IGF-IR signalling in breast and prostate cancer cell lines [
40]. Blockade of IGF-IR signalling in these cells using ABDP also enhanced responses to HRGβ1 in both cell lines, with phosphorylation of IRS-1 Y612, Akt and ERK1/2 apparent at lower concentrations of this ligand and with a greater magnitude of phosphorylation also observed at the highest concentrations of HRGβ1 in ABDP-treated compared to untreated cells. Similar results were observed when IGF-IR signalling was blocked using an IGF-IR siRNA. This rapid enhancement of HRGβ1 signalling by IGF-IR inhibition is likely a consequence of two mechanisms. The first is an IRS-1-mediated mechanism, which immunoprecipitation and Western blot analysis revealed a loss of IRS-1 association with IGF-IR and an increased association of this adaptor protein with erbB3 in MCF-7 and T47D cells following treatment with ABDP, mirroring results observed for HRGβ1 treatment alone in these cell types. Thus, HRGβ1 signalling was enhanced as a result of increased availability and association of IRS-1 with erbB3. This is further supported by the finding that knockdown of IRS-1 protein levels by siRNA not only reduced HRGβ1-primed Akt phosphorylation but also prevented the ABDP-induced sensitisation of the cells to this ligand, greatly reducing signalling via Akt in particular. The second is that an erbB3-dependent mechanism appears to play a role, as IGF-IR inhibition by either ABDP or IGF-IR siRNA knockdown also enhanced HRGβ1-induced erbB3 phosphorylation, with this effect being most apparent in the MCF-7 cells. The reasons behind this effect remain unclear, although similar findings have been reported in hepatocellular carcinoma cells treated with the novel IGF-IR monoclonal antibody AVE1642 [
39]. One possible mechanism was recently identified by Gijsen and colleagues [
48], who reported that blockade of Akt can activate ADAM17 (ADAM metallopeptidase domain 17) in erbB2-overexpressing breast cancer cells, leading to release of heregulins, which can act in an autocrine manner to activate erbB3. As IGF-IR blockade can acutely inhibit Akt activity in our cell lines, such a mechanism may explain the subsequent phosphorylation of erbB3; however, further studies are required to confirm this hypothesis. Interestingly, IRS-1 knockdown was not as effective in reducing HRGβ1-induced ERK1/2 activity compared to Akt activity in ABDP-treated MCF-7 and T47D cell lines. One possible explanation for this is that the increased erbB3 phosphorylation observed in response to ABDP may provide the input maintaining ERK1/2 phosphorylation in these cells; however, further investigation into this mechanism is required and is currently ongoing.
To determine whether this novel association between IRS-1 and erbB3 identified in our ER+ cell lines could also have clinical relevance, an exploratory study was performed in a small series of ER+ clinical breast tumours. An immunocytochemical assay was developed to detect phosphorylated IRS-1 Y612 and associations with erbB3, and other clinical markers were assessed. Importantly, a significant positive correlation between IRS-1 Y612 phosphorylation and total erbB3 expression was observed in these ER+ primary breast tumours. As the majority of these ER+ tumours were found to express low and/or negative erbB2 levels, these findings directly support our cell line work and suggest that an association between erbB3 and IRS-1 may well occur within ER+ breast tumours. The link between IRS-1 Y612 phosphorylation levels and Akt activity identified in the cell lines was also observed in the clinical samples, with significant correlations between phosphorylated levels of IRS-1 Y612 and Akt in ER+ patients. Moreover, there was a significant correlation between IRS-1 Y612 and the proliferation marker Ki-67 in these ER+ tumours, suggesting that the potential interplay between erbB3, IRS-1 and Akt in these tumours may culminate in driving cell proliferation. However, for such signalling to arise, heregulins must be synthesized and accessible within the cancer milieu. Importantly, our previous findings based on the same clinical breast cancer series used in this study, as well as others [
30,
49], clearly demonstrate that neuregulins such as HRGβ1 are ubiquitously expressed in clinical breast tissue, thus making such interplay a distinct possibility and warranting a more extensive study to be carried out in a larger breast cancer series. In light of these findings, the recent suggestion that IRS-1 should be considered as a biomarker for IGF-IR activity in cancers susceptible to IGF-IR targeting [
50] should be viewed with a degree of caution, especially in cancer types that also express erbB receptors and their ligands.
Competing interests
IRH, JMWG and RIN are in receipt of funding from AstraZeneca, and JFR is in receipt of funding from Amgen. RIN is also a member of an advisory board for AstraZeneca. All other authors have no conflicts of interest.
Authors' contributions
IRH conceived of the study, participated in its design and execution and helped draft the manuscript. JMK drafted the manuscript, carried out all the Western blot analyses and conducted all siRNA and cell culture studies, including growth studies with the help and support of DB. JMWG carried out the immunocytochemistry on a small series of 50 primary breast tumours excised from ER+ patients recruited by JFR and IOE. RIN participated in the design and coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.