Antihormone induced compensatory signalling in breast cancer: an adverse event in the development of endocrine resistance1)

Abstract

Using MCF7 breast cancer cells, it has been shown that anti- hormones promote expression/activity of oestrogen-repressed tyrosine kinases, notably EGFR, HER2 and Src. These inductive events confer responsiveness to targeted inhibitors (e.g., gefitinib, trastuzumab, saracatinib). We observed that these antihormone-induced phenomena are common to ERqHER2– and ERqHER2q breast cancer models in vitro, where targeting of EGFR, HER2 or Src alongside anti- hormone improves antitumour response and delays/prevents endocrine resistance. Such targeted inhibitors also subvert acquired endocrine resistant cells which retain increased EGFR, HER2 and Src (e.g., TAMR and FASR models derived after 6–12 months of tamoxifen or Faslodex treat- ment). Thus, antihormone-induced tyrosine kinases comprise ‘‘compensatory signalling’’ crucial in limiting maximal ini- tial antihormone response and subsequently driving acquired resistance in vitro. However, despite such convincing pre- clinical findings from our group and others, clinical trials examining equivalent antigrowth factor strategies have proved relatively disappointing. Our new studies deciphering underlying causes reveal that further antihormone-promoted events could be pivotal in vivo. Firstly, Faslodex induces HER3 and HER4 which sensitise ERq cells to heregulin, a paracrine growth factor that overcomes endocrine response and diminishes antitumour effect of agents targeting EGFR, HER2 or Src alongside antihormone. Secondly, extended antihormone exposure (experienced by ERq cells prior to adjuvant clinical relapse) can ‘‘reprogramme’’ the compen- satory kinase profile in vitro, hindering candidate antigrowth factor targeting of endocrine resistance. Faslodex resistant cells maintained with this antihormone for 3 years in vitro lose EGFR/HER2 dependency, gaining alternative mitogenic/ invasion kinases. Deciphering these previously unrecognised antihormone-induced events could provide superior treat- ments to control endocrine relapse in the clinic.

Keywords: antihormone; breast cancer; ERBB; resistance; Src.

Introduction

Antihormones (AHs) encompass antioestrogens we.g., non- steroidal selective oestrogen receptor modulators such as tamoxifen; the steroidal agent Faslodex (fulvestrant)x and oestrogen deprivation we.g., aromatase inhibitors in oestrogen receptor positive (ERq) postmenopausal breast cancer; luteinising hormone-releasing hormone agonist Zoladex in ERq premenopausal breast cancerx. Such strategies inhibit oestrogen-dependent gene transcription and growth of ERq breast cancer cells. Experimental data indicate that AHs fur- thermore subvert productive oestrogen/ER crosstalk with ‘‘favoured’’ growth factor pathways such as insulin-like growth factor 1 receptor signalling which can be prominent in ERq AH responsive breast cancer models w1, 2x. It is believed that in total such inhibitory events underlie the clin- ical observations that AHs promote worthwhile responses and significant survival benefit for many ERq breast cancer patients both in the adjuvant and advanced disease setting. However, despite these important benefits, a pervading prob- lem exists with AH treatment of ERq breast cancers as along with up to 40% of ERq patients being intrinsically resistant to AH therapy, antitumour responses are often incomplete in initially responsive patients w3x. Thus, some breast cancer cells clearly escape AH inhibition, as evi- denced by the observation that resistant growth commonly emerges during adjuvant AH treatment in the clinic culmi- nating in disease relapse w4x. Indeed, despite differences in response duration according to AH type, acquisition of resis- tance remains inevitable for approximately 40% initial responders in the ERq adjuvant setting and for almost all patients with advanced metastatic disease w5x. Moreover, in vitro data indicate that acquisition of endocrine resistance can be paralleled by a gain in breast cancer cell invasiveness w6x, whereas clinically endocrine relapse invariably heralds metastasis and poorer prognosis for the breast cancer patient w7x.

Interestingly, previous in vitro studies from our laboratory using ERq HER2– AH responsive MCF7 breast cancer cells and, primarily, 4-OH-tamoxifen challenge reveal that AHs are not passive but actively promote the acquired endocrine resistant phenotype via induction of candidate growth factor signalling, i.e., tyrosine kinases EGFR, HER2 and Src w8–10x. Importantly, this drug-induced ‘‘compensatory sig- nalling’’ serves to limit the magnitude of initial AH antitu- mour response, and ultimately such growth factor signalling promotes and, as shown by several groups, maintains endo- crine resistant breast cancer growth in preclinical models, as well as facilitating disease progression. Here, we review the phenomenon of AH-induced candidate growth factor signal- ling in ERq breast cancer and the impact and limitations of its therapeutic targeting in vitro and in vivo in antihormone responsive and resistant states, drawing on research from our group and others. Furthermore, we illustrate from our new experimental studies that further deciphering of the breadth of AH-induced mechanisms, notably those permitting para- crine growth factor influences as well as those promoted by extended AH exposure, could potentially provide superior therapeutic targets to control emergence, growth and pro- gression of acquired endocrine resistant breast cancer in the clinic.

AH induction of candidate growth factor signalling and its adverse impact in ERH breast cancer cells

The mechanism of AH-induced signalling has its roots in the phenomenon of repression by oestrogen. Along with classical promotion of ER-regulated genes, oestrogens can also trans- criptionally repress many genes in ERq AH responsive MCF7 cells, as evidenced by microarray studies from the laboratory of Frasor et al. w11x, and also by our application of BD Clontech Atlas Human Cancer 1.2K gene cDNA arrays (nylon platform) and subsequently HGU133A Affy- metrix microarrays examining the impact of 7-day oestradiol treatment (10–9 M) on such cells w9x. Along with reported oestrogen repression of known antiproliferative genes such as cyclin G1 w12x, paradoxically several adverse growth fac- tor signalling elements (e.g., the receptor tyrosine kinase HER2 and the additional erbB family member EGFR) are subject to oestrogen inhibition w13–15x. The underlying molecular biology of this phenomenon remains poorly understood w16x, although there are emerging data indicating that such repressive events can occur at oestrogen response element-bearing gene promoters within breast cancer cells w17, 18x. For both EGFR and HER2, some reports have indi- cated that repression by oestrogens can involve events at intron 1 of these receptors w13, 14, 19x. For HER2, repression could involve competition for the coactivator steroid receptor coactivator 1 (SRC1) between the oestrogen/ER complex and the HER2 intron 1 enhancer w14x. The ER complex is also able to recruit corepressors to the HER2 promoter to inhibit transcription of this gene: an importance for binding of paired box 2 (PAX2) to an ER binding site within the HER2 gene has been described in this transcriptional inhibition, with PAX2 competing with the coactivator SRC3/AIB1 for promoter binding w20x. A further corepressor implicated in oestrogen/ER suppressive events is the DEAD-box RNA helicase DP97 w21x, and additional transcriptional corepres- sors which can be recruited to agonist-bound ER include LCoR, RIP140 and ZNF366 w22x.

Not surprisingly therefore, AHs such as 4-OH-tamoxifen (Tam) or Faslodex (10–7 M) that inhibit oestrogen/ER sig- nalling are capable of inducing significant expression of such oestrogen-repressed genes from as early as 1 week in vitro w8, 9x. Although the consequences of these AH-induced events for the tumour cell have not been fully explored, Fra- sor et al. w23x and others have reported that a proportion of oestrogen-suppressed genes are antiproliferative/pro-apopto- tic w12, 24x, and hence it is probable that their expression contributes to antitumour effect when induced by AHs. How- ever, we have confirmed at the protein level using Western blotting and immunocytochemistry that there are also marked increases in the receptor tyrosine kinases EGFR and HER2, coupled with increased activity of the key downstream non- receptor tyrosine kinase Src, in Tam or Faslodex treated MCF7 cells versus oestradiol-treated controls from 1 week into treatment w8–10x. Increases in HER2 or Src activity have similarly been reported by others during aromatase inhibitor treatment of aromatase-transfected MCF7 cells in vitro and in vivo w25–27x.

Crucially, we have been able to demonstrate that the AH- induced EGFR, HER2 and Src (kinases whose signalling is heavily implicated in oncogenic cellular events and progres- sion w28x) adversely impacts on behaviour of MCF7 cells in vitro. This induced growth factor signalling maintains resid- ual ERq cell survival and proliferation, allowing some breast cancer cells to tolerate initial AH treatment thereby limiting maximal antitumour response w8, 9, 29, 30x. Indeed, it is known that AHs exert considerable antiproliferative effects but generally promote only modest cell kill w8x. EGFR, HER2 and Src signalling are subsequently further AH-induced (in the case of HER2, potentially involving PAX2 reduction during extended tamoxifen exposure w20x), culminating in growth inhibitory responses of finite duration and emergence of acquired resistant growth and a stable resistant phenotype from approximately 6 months continuous AH exposure in culture w8, 9x. The markedly induced HER2, EGFR and Src activity and the increased downstream sig- nalling (notably MAPK and AKT) and its crosstalk with ER and also IGF1R in such acquired resistant cells drives their increased proliferation, cell survival and invasive behaviour, as evidenced by our extensive studies in TAMR and FASR lines derived after 6–12 months 10–7 M Tam or Faslodex exposure, respectively w3, 31–36x. Further antioestrogen or oestrogen deprivation resistant in vitro/in vivo ERq breast cancer models have invariably been reported to have increased erbB receptors and/or downstream kinases (e.g., MAPK, PI3K, mTOR and AKT) and Src signalling w37–47x, including in vivo aromatase-transfected MCF7 models from the Brodie group where aromatase inhibitor-induced HER2 is again maximised on acquisition of resistance w26, 27x.

Although the data as yet remain controversial, some studies also indicate there could indeed be increases in various com- ponents of EGFR, HER2, downstream kinases and associated signalling following tamoxifen relapse in a proportion of breast cancer patients w3, 47–50x.

Therapeutic targeting of AH-induced candidate growth factor signalling in ERH breast cancer

Supportive evidence for the crucial role of AH-induced signalling in these adverse events can be derived from cotreatment experiments in vitro. Firstly, targeting of the AH- induced ‘‘compensatory’’ candidate growth factor signalling (e.g., EGFR, HER2 or Src) alongside the AH substantially improves initial antitumour response in MCF7 cells. Second- ly, such cotreatment is able to prevent the emergence of acquired endocrine resistance in this cell line. Thus, we have observed a superior antitumour effect of Faslodex cotreat- ment with the EGFR tyrosine kinase inhibitor (TKI) gefitinib (Iressa 1 mM) w8, 9, 51x or with the humanised HER2 mono- clonal antibody trastuzumab (Herceptin 100 nM) vs. the AH alone. Such cotreatment markedly depleted proliferation w70%–80% fall measured by Ki67 (MIB1 clone) stainingx and increased cell death (30%–50% increase measured by ApoAlert MitoSensor microscopy and trypan blue assays), effects significantly superior to those achieved with AH alone. This also contrasts the lack of impact of monotherapy with trastuzumab or gefitinib because MCF7 cells express very little target HER2 or EGFR in the absence of AH treat- ment w8, 31x. Furthermore, our in vitro passage data tracking continuous cotreatment of MCF7 cells with Faslodex plus inhibitors of the ‘‘compensatory’’ candidate growth factor signalling weither gefitinib, trastuzumab or the Src/Abl TKI saracatinib (AZD0530; 1 mM)x revealed continuous cotreat- ment subsequently prevented development of acquired resis- tant growth, contrasting the resistance that ultimately evolved following an initial growth inhibitory response to Faslodex as a single agent w8, 9, 29, 51x. Our studies with Tam plus either gefitinib or saracatinib in AH responsive MCF7 and T47D cells in vitro w8, 30x indicate the superiority of AH/ antigrowth factor cotreatment extends to further AH meas- ures. Tam plus either gefitinib or saracatanib cotreatment depleted key downstream kinases (e.g., MAPK), reduced proliferation-related genes (e.g., cyclin D1, myc), cells in S phase and Ki67 staining, and promoted total cell loss by 12–13 weeks continuous culture, preventing emergence of resistance. Furthermore, we observed that as Tam plus sara- catinib cotreatment efficiently blocks Src activity in MCF7 and T47D, it was also able to abrogate invasive behaviour during AH treatment w30x, including in an ERq model defi- cient in E-cadherin (and thereby intercellular adhesion) where antioestrogen-induced invasiveness was greatly aug- mented w10x. In total, such in vitro studies from our group and others w52, 53x, as well as ERq xenograft studies mon- itoring impact of various AHs plus anti-EGFR or anti-HER2 treatment w54x and anastrazole plus Src inhibitor studies in aromatase-transfected MCF7 cells w25x, are highly supportive of clinical breast cancer trials combining candidate anti- growth factors targeting ‘‘compensatory’’ tyrosine kinase signalling with AHs to improve initial antitumour response and hinder development of aggressive acquired endocrine resistance.

In addition, considerable model system research has impli- cated use of such candidate antigrowth factor treatments as second-line agents to treat acquired endocrine resistant breast cancer growth and its associated progression. This concept is supported by our extensive studies demonstrating signifi- cant growth and invasion/migration inhibition with agents targeting EGFR, HER2, Src or downstream MAPK/AKT kinases in our acquired resistant model TAMR both in vitro w6, 31, 33–35, 55x and in vivo w56x, as well as in our acquired resistant FASR model w29x. Other research groups have sim- ilarly shown relevance of candidate growth factor signalling inhibitors in vitro and in vivo using various ERq endocrine resistant models. Such studies have encompassed resistance to antioestrogens and oestrogen deprivation w37, 39, 44, 57, 58x, with candidate growth factor signalling inhibitors in some instances also restoring AH responsiveness in the endo- crine resistant models w27, 45–47, 59–63x.

Based on these and further promising preclinical findings, diverse inhibitors of candidate growth factor signalling are now under clinical investigation in breast cancer (reviewed in w64x), including in combination with AHs. Indeed, there are several completed, just closed or ongoing trials examin- ing AH cotreatment with inhibitors of EGFR or HER2 (e.g., gefitinib, erlotinib, trastuzumab or lapatinib) or their down- stream signalling, including MAPK inhibitors, farnesyltrans- ferase inhibitors or mTOR antagonists, where the aim is again to treat endocrine resistance or (alongside AHs) delay this state, also potentially restoring AH responsiveness (reviewed in w65, 66x). Further studies are investigating Src inhibitor plus AH cotreatment, for example, Phase II studies are ongoing with dasatinib plus Faslodex, exemestane or letrozole and with the further agent bosutinib plus aromatase inhibitors (reviewed in w67, 68x).

In keeping with the experimental cotreatment findings in ERq AH responsive cells, and with studies in our TAMR and FASR lines and further antioestrogen or oestrogen-dep- rivation resistant models from others, there is possible benefit emerging from some of these clinical trials. Thus, there was an improvement in progression-free survival (PFS) in endo- crine naive ERq metastatic breast cancer patients treated with tamoxifen plus gefitinib in the randomised Phase II study reported by Osborne et al. w69x. There was also an improved PFS for anastrazole plus gefitinib vs. AH alone in a randomised Phase II ERq postmenopausal metastatic breast cancer trial reported by Cristofanilli et al. w70x. Such data are suggestive that cotreatment might be able to delay development of endocrine resistance in some breast cancer patients. Among elements downstream in growth factor path- ways, cotargeting of mTOR alongside AHs might also have potential. Of note, the mTOR inhibitor RAD001 (everoli- mus) plus the AH letrozole in the ERq neoadjuvant setting resulted in increased tumour shrinkage and reduction in pro- liferation w71x.

There are also reports that erbB inhibitors could benefit a proportion of patients who have a tamoxifen resistant breast cancer phenotype. Thus, in a collaborative Phase II study with Nottingham City Hospital we observed a 53.6% res- ponse rate to gefitinib (500 mg/day) in ERq acquired tamoxifen resistant patients. Immunohistochemistry per- formed in sequential biopsies taken pretreatment and 8 weeks into gefitinib treatment revealed such responses were com- monly associated with significant decreases in Ki67 staining during treatment, in some instances paralleled by depletion of EGFR and MAPK activity w56x. Of note, gefitinib responses were more frequent in the ERq acquired tamoxi- fen resistant patients than in the ER– AH insensitive cohort also examined within this study w56, 72x, presumably a con- sequence of some ERq acquired resistant tumours being driven by growth factor receptor/ER crosstalk in equivalence with models such as the TAMR line w34x. Also in keeping with gefitinib having potential to influence ERq tamoxifen resistant tumour phenotypes and associated EGFR/ER cross- talk, Polychronis et al. w73x reported in a preoperative study a reduction in tumour size with this agent, either alone or in combination with anastrazole, in a proportion of newly diag- nosed ERq EGFRq patients. There were declines in pro- liferation, MAPK and ER activity with such treatment, and a greater fall in Ki67 was observed in tumours treated with gefitinib plus anastrazole. Johnston et al. w74x also reported that treatment with the aromatase inhibitor letrozole in com- bination with the dual EGFR/HER2 TKI lapatinib numeri- cally improved PFS versus AH alone within a subgroup of ERqHER2– patients who had prior antioestrogen therapy and experienced relapse less than 6 months since prior tamoxifen discontinuation.

However, despite extremely encouraging preclinical data and the above positive clinical observations, benefit of such agents targeting candidate growth factor signalling in patients remains controversial w75, 76x. Invariably, the clinical trials have proved disappointing in that many breast cancer patients fail to benefit from, or at best exhibit only a limited response to, candidate antigrowth factor therapeutic strate- gies (reviewed in w65, 66, 77x). For example, although clin- ical benefit was commonly seen in acquired tamoxifen resistant patients with the anti-EGFR agent gefitinib in our Phase II study, this generally comprised disease stabilisation which invariably culminated in relapse w56, 72x. Further- more, although of significantly improved value when com- bined with chemotherapy, responses to trastuzumab as a single agent only occur in approximately 30% of HER2q patients w78x and again, despite initial benefit and significant improvement in prognosis in the adjuvant and metastatic HER2q setting (reviewed in w64x), disease relapse remains a common ultimate outcome of all trastuzumab regimes in breast cancer w79x. When considering AH/antigrowth factor combination treatment, it is notable that the majority of ERq HER2– patients failed to respond to lapatinib plus letrozole in the study reported by Johnston et al. w74x. We clearly have a long way to go to stratify patients adequately for AH/anti- growth factor treatment and to fully understand and effectively target the signalling driving AH relapse in vivo.

Mechanisms limiting responses to therapies targeting candidate growth factor signalling in ERH breast cancer: potential role for further AH-induced events

There remain many challenges if we are to substantially improve the impact of antigrowth factor approaches in clin- ical breast cancer, not least of which is determining the sig- nalling mechanisms that can adversely impact on outcome of such treatments. For example, PTEN deficiency w80x and hence constitutive AKT activity, p27 downregulation w81x and elevated IGF1R de novo w82, 83x have all been shown to adversely influence response to gefitinib or trastuzumab in cancer cells. Recruitment of alternative signalling path- ways actively promoted by antigrowth factor treatment has furthermore been implicated in acquired resistance to such agents. This is exemplified by observations that gefitinib can induce insulin receptor substrate-1/IGF1R signalling in TAMR cells and also alternative erbB ligand/receptor and downstream AKT signalling in further cancer types w84–86x. Similarly, acquisition of Src inhibitor resistance has been associated with recruitment of EGFR, mTOR and MEK/ MAPK signalling as alternative mitogenic routes during sara- catinib treatment of ERq breast cancer cells w25x.
However, to our knowledge it has not been considered if there could also be a contribution for further AH-induced compensatory events in limiting clinical impact of candidate antigrowth factor treatments in ERq AH responsive and resistant breast cancer. This aspect comprises the focus of our group’s new research, where through in vitro studies within various ERq AH responsive breast cancer cell lines, as well as by developing new acquired endocrine resistant models, we are addressing three questions that are yielding interesting findings with potentially important therapeutic implications, as detailed below.

Does ERH tumour genetic background (e.g., HER2 status) influence the profile of AH-induced candidate growth factor signalling and thus outcome of EGFR, HER2 or Src inhibitor approaches?

Building further on MCF7 cell data, we have recently extended our AH and AH/antigrowth factor cotreatment studies to multiple ERq models to more fully reflect the breadth of ERq breast cancer subtypes, in the first instance considering HER2 status. Interestingly, our in vitro immu- nocytochemical studies monitoring impact of Faslodex (10–7 M) in such models have shown that by day 7 treatment, further detectable induction of EGFR and HER2 expression by this AH is common to both ERq HER2– (e.g., MCF7, T47D) and ERq HER2q (e.g., BT474, MDAMB361)
breast cancer cells in vitro. Moreover wherever Faslodex resistance is acquired, the resultant emerging phenotype retains increases in these receptors vs. the parental counter- part, again irrespective of HER2 status. Interestingly, Yarden et al. w15x also reported detectable EGFR upregulation both in ERq HER2– and ERq HER2q lines with oestrogen withdrawal, and an increased resultant sensitivity to EGFR ligands. Our studies monitoring Src activity using immunocytochemistry have similarly revealed that significant increa- ses occur with Faslodex treatment irrespective of ERq HER2– or ERq HER2q status.

We have subsequently tracked the growth impact of con- tinuous cotreatment in vitro with Faslodex plus inhibitors of this candidate growth factor signalling (versus AH or anti- growth factor alone) using these ERq HER2– and ERq HER2q models. We observed that although the ERq HER2q models showed an initial tumour inhibitory response to Faslodex monotherapy (1–4 months duration in vitro), equating with clinical observations that ERq HER2q advanced breast cancer does respond to this AH w87x, this response was shortened vs. the ERq HER2– lines (at least 6 months in vitro), findings equating with the reported asso- ciation in some clinical studies between HER2 and poorer endocrine responsiveness w3, 88x. However, as in our previ- ous ERq HER2– MCF7 studies w8, 29x, cotreating ERq HER2q cells with Faslodex plus the antigrowth factor gefi- tinib or trastuzumab to target the AH-induced EGFR and HER2, respectively, further depleted proliferation and cell survival versus AH or antigrowth factor monotherapy. More- over, despite the shortened response to AH alone, cotreat- ment of ERq HER2q cells again either significantly delayed (AHqtrastuzumab in MDAMB361; AHqsaracati- nib in BT474 cells) or completely prevented emergence of resistance (AHqtrastuzumab or AHqgefitinib in BT474; AHqgefitinib or AHqsaracatinib in MDAMB361 cells) vs. AH alone. Of note, the duration of response for cotreatment was also superior to antigrowth factor monotherapy, highly suggestive that an effective antitumour response required impact on AH-induced as well as basal EGFR, HER2 or Src in these models. In keeping with our findings of an improved antitumour impact of AH plus antigrowth factor cotreatment in ERq HER2q cells, Kunisue et al. w89x have reported that trastuzumab improved Faslodex impact in a further ERq
/HER2 expressing cell line, ML20. Furthermore, Kurokawa et al. w90x observed that the EGFR antagonist AG1478 improved the antitumour activity of tamoxifen in HER2 over- expressing MCF7 xenografts, where lapatinib plus tamoxifen w59x and gefitinib plus tamoxifen w57x were also superior cotreatments vs. AH alone. The Slingerland group have also demonstrated that saracatinib can improve initial antitumour response and delay resistance in the presence of the AHs Faslodex or tamoxifen in ERq HER2q MDAMB361 xeno- grafts w91x. The mechanism of inhibition involved depletion of Src, MAPK and AKT activity, increased p27, and a reduc- tion in cyclin E-cdk2 activity and proliferation.

Clearly, therefore, AH induction of candidate (EGFR, HER2 and Src) growth factor signalling is a shared feature of ERq HER2– and ERq HER2q breast cancer models, with little impact of HER2 status on the superior outcome of AH/candidate antigrowth factor cotreatment vs. AH or antigrowth factor alone. Interestingly, there are some emerg- ing clinical studies supportive of this concept. Kaufman et al. w92x have reported significant improvement in PFS with a first-line anastrazole plus trastuzumab regime vs. AH alone in ERq HER2q postmenopausal metastatic breast cancer patients in the randomised Phase III ‘‘TAnDEM’’ study. Furthermore, Johnston et al. w74x examining letrozole in com- bination with the EGFR/HER2 TKI lapatinib observed significantly improved PFS and clinical benefit rate in ERq HER2q postmenopausal metastatic breast cancer patients vs. aromatase inhibitor alone. However, despite the promis- ing preclinical model data and these observations in ERq HER2q patients, it is clear from such clinical studies that subsequent emergence of resistance during cotreatment with AH plus agents targeting candidate growth factor signal- ling remains a pervading problem. Such data infer that addi- tional crucial mechanisms are contributory to endocrine resistance in vivo, which could be dependent on further inherent aspects of the tumour genetic background or involve as yet unknown AH-induced events.

Are there further AH-induced events that enable ERH breast cancer cells to harness paracrine growth factors to drive endocrine resistance via alternative pathways in vivo?
There are emerging data indicating that endocrine resistant breast cancers in the clinic have heterogeneous gene expres- sion profiles w93x, suggesting that resistance in vivo can involve further mechanisms and associated pathways. Through profiling for additional tyrosine kinases induced by AHs in MCF7 cells, we observed that 7 day Faslodex treat- ment increased expression and phosphorylation of the further erbB receptors HER3 and HER4 w94x. These receptors have also previously been reported as oestrogen-suppressed w95x. Interestingly, we observed that Faslodex treatment sensitised these ERq cells to exogenous heregulin b1 (HRGb1) in vitro, a HER3/HER4 ligand not produced by ERq breast cancer cells such as MCF7 but readily detectable in clinical tissue and reported to be a potent paracrine promoter of breast cancer growth and metastasis w96, 97x. Faslodex treat- ment enhanced the HER3/HER4 activity and downstream kinase signalling (e.g., AKT) triggered by HRGb1, convert- ing this weakly mitogenic growth factor in MCF7 cells into a potent growth promoter in vitro. Indeed, HRGb1 entirely overcame Faslodex antitumour response in these ERq cells w94x. An ability of HRGb1 to circumvent Faslodex response has also recently been reported in MCF7 cells by Sonne- Hansen et al. w98x. In keeping with this, some established endocrine resistant lines have increased HER3 signalling, with HER3 also implicated in early AH relapse of some breast cancer patients w98–101x. Heregulin can similarly overcome tamoxifen response in MCF7 cells, triggering a gene profile that interestingly is shared by poor prognosis (luminal B) ERq breast cancers w102x. Our studies have revealed that significant induction of HER3/HER4 signalling activity occurs with Faslodex treatment irrespective of HER2 status in a panel of in vitro ERq breast cancer models, including ERq HER2– MCF7 and ERq HER2q BT474 and MDAMB361 cells, with exogenous HRGb1 promoting Faslodex resistant growth in all instances w94x. Furthermore, exogenous HRGb1 was able to substantially diminish anti- tumour responses to cotreatments targeting candidate growth factor signalling EGFR, HER2 or Src (e.g., gefitinib, tras- tuzumab or saracatinib, respectively) alongside Faslodex in all these ERq models. Clearly, therefore, there is an addi- tional layer of complexity to AH-induced events in ERq breast cancer cells, as further growth factor signalling ele- ments are induced by AHs that might allow breast cancer cells to harness available paracrine ligands in vivo and pro- mote an alternative endocrine resistant phenotype w103x. Our experimental findings suggest that such an event could have considerable potential to adversely impact on outcome of candidate antigrowth factors in ERq breast cancer treated with AHs in the clinic.

Could duration of AH exposure influence the profile of AH-induced growth factor signalling elements and thus outcome of candidate antigrowth factors?
Although patients are invariably AH-treated in the adjuvant setting for several years prior to detectable relapse, in general only short-term (up to 12 months) AH exposure has been modelled experimentally in vitro. To our knowledge, the Jor- dan laboratory also remains the only group addressing AH duration in their in vivo studies w104x. However, it is feasible that the extended duration of adjuvant AH exposure could significantly impact on the profile of AH-induced events in ERq breast cancer.
To study this concept, we are developing new acquired endocrine resistant breast cancer cell lines following contin- uous AH exposure for at least 3 years in vitro. This is exem- plified by our FASRLT cell model, derived following such prolonged Faslodex treatment of MCF7 cells w36x. Interest- ingly, this model is particularly proliferative and invasive, indicating that extended AH exposure can promote a highly aggressive resistant phenotype. Of note, studies with gefiti- nib or trastuzumab challenge have revealed that growth and invasion of the FASRLT line is independent of EGFR and HER2, contrasting inhibitory responses to blockade of such signalling exhibited in the short-term FASR model (derived after 6–12 months of AH exposure w29x). Crucially, FASRLT cells have lost a significant proportion of their EGFR w36x, and all HER2 expression, compared with elevated levels of these erbB receptors in the short-term FASR cells vs. the parental MCF7 line w29x, explaining the lack of significant gefitinib or trastuzumab response exhibited by FASRLT. It thus appears that duration of AH exposure does indeed have potential to influence the candidate growth factor signalling elements and thus outcome of inhibitory agents. There are possible clinical parallels because a considerable number of patients on AH relapse fail to show obvious increases in expression of such erbB receptors w47, 49x. Similarly, pre- clinical data from the Jordan group indicate that additional factors other than such erbB receptors are likely to contribute during prolonged tamoxifen treatment w104x.

However, our profiling studies have also revealed that although prolonged Faslodex treatment appears to drive a decline in these erbB receptors, the AH is also capable of inducing prominent expression of alternative adverse recep- tor tyrosine kinases never previously implicated in endocrine resistant breast cancer. Our group has used Affymetrix HG- U133A GeneChip microarray technology to determine deregulated members of the ‘‘kinome’’ (as defined in w105x) in FASRLT cells, particularly tyrosine kinases as these have commonly been implicated in neoplastic development and progression and are key targets for anticancer drug devel- opment w106x. Our gene array studies w103x, in some instanc- es with subsequent RT-PCR verification, have revealed substantial deregulation of several kinases in the FASRLT line, and subsequently in subclones that we have derived to further understand this model. These include significant increases in the tyrosine kinases MET, TYRO3 and STYK1. We have shown that MET is able to contribute to disease progression in these long-term Faslodex resistant cells, with their increased invasiveness further enhanced by exogenous hepatocyte growth factor, a fibroblast-derived ligand for MET readily available in the breast cancer paracrine envi- ronment w107x. TYRO3 (SKY) is an AXL family member. The AXL/TYRO3 family trigger downstream kinase signal- ling and can be increased in expression in breast cancer w108x. AXL family signalling is also implicated in patholog- ical thrombosis and in stabilising b-catenin and thus poten- tially tumour growth and progression w109x. Hence, TYRO3 targeting could prove relevant to breast cancer patients. STYK1 is a tyrosine kinase with 30% similarity to fibroblast growth factor receptor 3, signalling through MAPK and PI3K/AKT to promote aggressive, proliferative tumour behaviour w110x. Kimbro et al. w111x have suggested that STYK1 could be oestrogen-repressed in MCF7 cells, find- ings that are in keeping with our observation of substantially increased STYK1 within cells treated long-term with the AH Faslodex. This kinase has also been reported in breast cancer and the observation that STYK1 siRNA is growth inhibitory in T47D breast cancer cells implies an adverse contribution to this disease w112x. Moreover, studies in mice have revealed that STYK1 overexpression is strongly transform- ing, with stably transfected cells readily able to induce tumourigenesis in vivo, metastasis and a poorer resultant sur- vival. As such, increased STYK1 could be a key driver of growth and progression in our long-term Faslodex treated cells.

Summary and therapeutic implications

Our in vitro studies have demonstrated that AH-induced compensatory growth factor signalling is a crucial adverse event that can limit initial AH antitumour response and, together with experimental studies from other groups, have revealed such growth factor signalling can ultimately drive acquired endocrine resistance. Promising model system data have been derived targeting candidate AH-induced signalling (EGFR, HER2, Src) in ERq HER2– and ERq HER2q cells using antigrowth factor/AH cotreatment strategies to hinder development of resistance, as well as extensive stud- ies monitoring the candidate antigrowth factors within acquired resistant models. However, although some value is emerging for such strategies clinically, there is clearly a need for substantial improvement in response rate, requiring fur- ther deciphering of the endocrine resistance mechanism.

Based on our emerging data, it is our expert opinion that priority future research areas in this regard should be (i) to determine the breadth of AH-induced growth factor signal- ling mechanisms that permit alternative paracrine resistance mechanisms and (ii) to fully detail how prolonged adjuvant AH treatment ‘‘reprogrammes’’ breast cancer cells to pro- mote an adverse phenotype, loss of candidate growth factor receptors and gain of alternative adverse AH-induced sig- nalling elements. To successfully address these issues, it is important that such studies should focus not only on exper- imental models, taking on board different genetic back- grounds, but wherever possible researchers should also investigate matched clinical material taken during AH treat- ment and at subsequent disease relapse to ensure relevance of induced signalling mechanism within the breast cancer patient. Our experimental studies are beginning to demon- strate that these concepts of AH-induced signalling have con- siderable potential to adversely impact on outcome of candidate antigrowth factor approaches. Hence, it is feasible that deciphering this previously unrecognised AH induced signalling could ultimately yield new therapeutic approaches for endocrine relapse, potentially to be used alongside AHs and candidate antigrowth factors to provide superior responses and improve patient prognosis.

Of particular therapeutic interest is future development/ study of agents able to inhibit both the candidate growth factor pathways and additional antihormone-induced signal- ling mechanisms. For example, using ‘‘pan-erbB’’ inhibition alongside AHs to subvert recruitment of multiple erbB recep- tors (over and above EGFR and HER2), thereby hindering paracrine heregulin impact, could prove a valuable avenue. Indeed, xenograft studies demonstrated that ‘‘pan-erbB’’ inhibition with AH (e.g., multiple cotreatment with gefitinib plus trastuzumab plus pertuzumab together with either tamoxifen or oestrogen withdrawal) exerted a superior anti- tumour effect vs. the individual anti-EGFR or HER2 agents with AH in ERq HER2q breast cancer w113x, as well as in endocrine resistant models w98x. Of further interest could be targeting, alongside AHs, of crucial ‘‘hubs’’ regulating AH- induced candidate growth factor signalling as well as alter- native AH-induced pathways. Of potential relevance in this regard is targeting of heat shock protein 90 (HSP90), a molecular chaperone with multiple client proteins including the erbB receptors and additional key proliferation and cell survival kinases w86, 114x. The HSP90 inhibitor 17-allyl- amino-17-demethoxy geldanamycin (17-AAG) can inhibit tamoxifen refractory breast cancer models w115x and a further HSP90 inhibitor has recently been shown to be effective against aromatase inhibitor resistant cells via depletion of multiple clients including HER2 and AKT w116x. Excitingly, our ongoing in vitro studies using 17-AAG (50 nM) along- side the AH Faslodex (10–7 M) in ERq HER2– (MCF7) and ERq HER2q (MDAMB361, BT474) cell lines reveal that such cotreatment is able to overcome HRGb1-promoted AH resistance in all instances, restoring a substantial 65%– 85% growth inhibition in such models. Promisingly, we also observed in these experiments that cotreatment with 17-AAG plus Faslodex gave a more effective antitumour response in such models vs. either trastuzumab, gefitinib or saracatanib plus AH cotreatment which were generally only poorly growth inhibitory in the presence of exogenous HRGb1.

Finally, although EGFR or HER2 blockade was ineffective in controlling growth or aggressive behaviour of long-term AH-treated FASRLT cells, our recent siRNA studies target- ing two of the alternative kinases induced by prolonged AH exposure in this model, TYRO3 or STYK1 w103x, have con- firmed significant dependence on their signalling. Such gene knockdown resulted in ;40% and ;70% inhibition of FASRLT cell growth and invasiveness, respectively. These encouraging findings are not only supportive of investigation of TYRO3 and STYK1 within clinical acquired endocrine resistant breast cancer but also provide tantalising experi- mental evidence that continued deciphering of kinases induced by long-term AH exposure could yield future targets relevant to the treatment of aggressive AH relapse in the adjuvant clinical setting.