The NFkB pathway and endocrine-resistant breast cancer

Endocrine therapy with an estrogen receptor (ER)-targeted antiestrogen, such as tamoxifen, or estrogen ablation by aromatase inhibitors is clinically indicated for the management of all forms of ER-positive breast cancer. However, 30–50% of ER-positive breast cancer cases fail to benefit clinically from endocrine therapy alone, and recent molecular evidence suggests that ‘crosstalk’ pathways originating from activated receptor tyrosine kinases and/or other proliferative and survival signals may be contributing to this endocrine resistance. Molecular identification and validation of candidate ER crosstalking pathways will likely lead to clinically important prognostic markers and targets for the application of novel therapeutics in combination with standard endocrine agents. This review focuses on a critical survival and proliferation pathway involving activation of nuclear factorkB (NFkB), a family of ubiquitously expressed transcription factors that for nearly two decades have been known to be critical regulators of mammalian immune and inflammatory responses, and more recently have been associated with chemotherapy resistance. With the demonstration that activation of NFkB is absolutely required for normal mammary gland development, NFkB involvment in human breast cancers was initially explored and linked to the development of hormone-independent (ER-negative) breast cancer. Newer clinical evidence now implicates NFkB activation, particularly DNA-binding by the p50 subunit of NFkB, as a potential prognostic marker capable of identifying a high-risk subset of ER-positive, primary breast cancers destined for early relapse despite adjuvant endocrine therapy with tamoxifen. Furthermore, initial preclinical studies suggest that treatment strategies designed to prevent or interrupt activation of NFkB in cell-line models of these more aggressive, ER-positive breast cancers can restore their sensitivity to such standard endocrine agents as tamoxifen. Endocrine-Related Cancer (2005) 12 S37–S46 Receptor crosstalk and endocrine resistance Levels of estrogen receptor (ER) (alpha isoform) overexpression, as well as the coexpression of ERassociated gene products (e.g. PR, pS2), have long been recognized as markers of breast cancer prognosis and, more importantly, predictors of response to endocrine therapy and clinical outcome. In fact, the clinical responsiveness of ER-positive breast cancers to the antiestrogen tamoxifen correlates positively with the absolute expression level (fmol/mg protein) of tumor ER (McGuire 1980, Elledge & Fuqua 2000). Gene microarray and other studies now indicate that ER-positive breast cancers can be divided into clinical subsets with extremely different outcomes, ranging from tumorswith goodprognosis and endocrine responsiveness to others with de novo or acquired endocrine resistance and risk of early relapse (Gruvberger et al. 2001, Sorlie et al. 2001, Benz 2004a). Additionally, a growing body of preclinical and clinical reports link antiestrogen resistance with tumor overexpression This paper was presented at the 1st Tenovus/AstraZeneca Workshop, Cardiff (2005). AstraZeneca has supported the publication of these proceedings. Endocrine-Related Cancer (2005) 12 S37–S46 Endocrine-Related Cancer (2005) 12 S37–S46 DOI:10.1677/erc.1.00977 1351-0088/05/012–S37 g 2005 Society for Endocrinology Printed in Great Britain Online version via http://www.endocrinology-journals.org of one or more members of the ErbB/HER family of receptor tyrosine kinases (reviewed in Benz 2004a,b). In particular, up to 15% of newly arising breast cancers are not only ER-positive but also overexpress the ErbB2 receptor as a result of oncogene amplification. Several clinical studies have shown that ErbB2positive breast cancers that are also ER-positive have significantly lower ER and PR content than ER-positive breast cancers that are ErbB2-negative (Eppenberger-Castori et al. 2001, Konecny et al. 2003). Supporting these clinical observations, ER-positive breast cancer cell lines engineered to overexpress ErbB2 retain their ER positivity but show marked reductions in their ER content (Konecny et al. 2003, Benz et al. 1992). While this downregulation of ER and PR expression may partially explain the reduced antitumor activity of antiestrogens against ERpositive/ErbB2-positive breast cancers relative to ER-positive/ErbB2-negative cancers, how ErbB2 activation downregulates ER and PR expression remains a mechanistic mystery. Recent studies are beginning to elucidate how signaling pathways activated by ErbB-related membrane receptor tyrosine kinases (RTK) crosstalk with ER pathways (reviewed in Benz 2004a,b). Membrane receptor-initiated signaling through the mitogenactivated protein kinase (MAPK) pathway and the phosphatidylinositol-3-kinase (PI3K)-Akt pathway results in phosphorylation of ER on its various serine (S167, S118, S104, S106) residues and leads to both ligand-dependent and ligand-independent ER-mediated gene activation via ‘classical’ (direct ER DNA-binding at promoters containing estrogen-response elements) and ‘non-classical’ (ER tethering and coactivation of other DNA-bound transcription factor complexes such as AP-1, Sp1, C/EBPb and CREB) gene induction. Importantly, the net transcriptional effect of crosstalk that phosphorylates ER and its many transcriptional coregulators includes the functional conversion of an ER-binding antiestrogen such as tamoxifen into an ER agonist, capable of driving ER-positive cancer growth almost as efficiently as the potent estrogen, estradiol (Benz et al. 1992). Promising preclinical results from our well-characterized MCF-7/HER2 model of ERpositive/ErbB2-positive breast cancer are now fueling clinical studies involving patients with endocrineresistant, ER-positive breast cancers, in which ErbB RTK inhibitors are being administered in combination with tamoxifen (Benz et al. 1992, Kurokawa et al. 2000, Shou et al. 2004, Schiff et al. 2004). Alterations in ER crosstalk pathways clinically linked with tamoxifen resistance but not necessarily originating from ErbB2 amplification and overexpression have also been described; these include enhanced activation of the gene-regulating transcription factor complex, AP-1 (Johnston et al. 1999, Schiff et al. 2000), dysregulated PI3/Akt (Campbell et al. 2001), protein kinase Ca (Chisamore et al. 2001), and the insulin-like growth factor I (Parisot et al. 1999) signaling pathways. Notably, all of these signaling pathways leading to tamoxifen resistance share a common mechanistic link with activation of another gene-regulating complex, nuclear factor-kB (NFkB) (Vertegaal et al. 2000, Zhou et al. 2000, Bhat-Nakshatri et al. 2002, DeGraffenried et al. 2004). NFkB in organ development and disease The NFkB complex is composed of a family of inducible transcription factors found in almost all cells (Baeuerle & Baltimore 1996, Ghosh et al. 1998, Allen & Tresini 2000); and this complex is generally recognized as an essential cell mediator acting ‘at the crossroads of life and death’ (Karin & Lin 2002). Activation of NFkB occurs in response to extracellular chemical stresses, various cytokines and growth stimuli, resulting in the direct induction of hundreds of genes whose cellular influences extend well beyond those of the immune system, where its essential role was first appreciated nearly two decades ago (Pahl 1999). In fact, the antiapoptotic, proliferation-, motilityand invasion-promoting roles of NFkB appear to be critical for normal organ development, including the mammary gland (reviewed in Cao & Karin 2003). NFkB activation can become abnormal during organ aging, with development and progression of various chronic inflammatory disorders, and in malignancies such as B and T cell lymphoma and leukemia, and thyroid, head and neck, gastrointestinal, and breast carcinoma (Baldwin 2001, Giardina & Hubbard 2002, Feinman et al. 2004, Veiby & Read 2004). The NFkB family consists of five mammalian members: p50 (NFkB1), p52 (NFkB2), p65 (relA), c-rel and relB. These all share a conserved 300-aminoacid N-terminal Rel homology domain (homologous to that encoded by the avian oncogene, v-Rel) that is responsible for dimerization, nuclear translocation, DNA binding, and association with IkB inhibitory proteins (Dixit & Mak 2002, Ghosh & Karin 2002). These Rel family members exist as homoor heterodimers, although the most abundant form of intracellular NFkB is generally thought to be the p50/p65 heterodimer. In resting cells, NFkB is cytoplasmically sequestered as a latent complex bound to one or more members of the IkB protein family (IkBa, IkBb, IkBe, IkBg, Bcl-3, and the precursor Rel proteins Zhou et al.: NFkB and breast cancer S38 www.endocrinology-journals.org p100 and p105). Diverse cell stimuli (e.g. tumor necrosis factor (TNF)a, CD40 ligand, interleukin (IL)-1, TRANCE, epidermal growth factor (EGF), phorbol esters, peroxides, ionizing radiation) induce phosphorylation (via activation of the IkB kinase complex, IKK) and subsequent proteasomal degradation of IkB inhibitory proteins, activating NFkB for nuclear translocation, where it binds promoter-specific kB consensus DNA elements that direct transcription of over 180 known NFkB target genes. While phosphorylation and degradation of IkB inhibitory proteins are considered the rate-limiting if not obligate mechanisms by which NFkB is activated, novel NFkB phosphorylating kinases and IKK-independent pathways leading to IkB proteasomal degradation have recently been described. Most activated forms of NFkB induce gene transcription, although specific NFkB subunits lack transactivation domains; thus, activation and nuclear translocation of p50/p50 and p52/p52 homodimers can result in repression of NFkBdependent genes (Ghosh & Karin 2002). However, when either NFkB p50 or p52 products of the p105 and p100 Rel precursor proteins are bound to the oncogenic and noninhibitory IkB family member, Bcl-3, they become transcriptionally competent and stimulate expression of NFkB-dependent genes (Cogswell et al. 2000, Ghosh & Karin 2002). Since NFkB regulates so many known survival and proliferation genes, it is not surprising that NFkB activation has generally been implicated in cancer chemotherapy resistance mechanisms (Wang et al. 1999). When first studied in human breast cancer cell lines and breast cancer samples, however, constitutive activation of NFkB was associated only with hormoneindependent (ER-negative) breast cancers, and this was thought to be due to its known inhibitory effects on almost all steroid receptors, including ER (Nakshatri et al. 1997). Importance of NFkB in hormonedependent breast cancer Less than a decade ago, NFkB activation was initially linked with the etiology and progression of hormoneindependent breast cancers, where it was shown transcriptionally to induce genes mediating cell proliferation and invasion, such as cyclin D1 and urokinase-type plasminogen activator (uPA).Measured by DNA-binding, transactivation and immunoblot assays, NFkB activation was first evaluated in several samples and cell lines, where it was found to be minimal in ER-positive breast cancers and cell lines but constitutively elevated in ER-negative breast cancers and cell lines (Nakshatri et al. 1997, Sovak et al. 1997). A subsequent study compared a small number of breast cancers with normal adjacent breast tissue (and also a few breast cancer cell lines) by measuring total NFkB DNA-binding activity and subunit (p65, c-rel, p52 and p50) protein and transcript expression levels (Cogswell et al. 2000). The breast cancer samples all showed greater total NFkB DNAbinding activity than the normal mammary gland tissue, but the increased tumor activity did not correlate with tumor ER status. In contrast, the cell line results again confirmed low NFkB activation in ER-positive cell lines and high NFkB activation in ER-negative breast cancer cells; moreover, these breast cancer cell lines showed predominantly increased p65 subunit expression and p65/p50 NFkB DNA-binding activity, while the breast tumor samples showed selective upregulation of p50, p52 and c-rel expression (as well as Bcl-3) and increased DNA-binding by complexes composed mostly of these subunits and with relatively little p65 (Cogsell et al. 2000). We recently performed the most extensive evaluation to date of NFkB activation in primary human breast tumor samples in order to clarify the extent and clinical importance of NFkB activation in hormonedependent breast cancer (Zhou et al. 2005). With a new ELISA-based method to quantitate specific p65 and p50 NFkB DNA-binding subunits, these subunit activities were independently measured in 81 ERpositive, primary breast cancer sample extracts with a wide range of ER content (group A samples:>100 fmol/ mg protein; group B samples: 21-87 fmol/mg protein). NFkB p50 and p65 subunit DNA-binding activities were also evaluated for their prognostic association with clinical outcome in the subset of 59 group B cases that were comparably staged, characterized for a number of other prognostic biomarkers (Quong et al. 2002), uniformly treated with adjuvant tamoxifen, and clinically followed until metastatic relapse to determine disease-free patient survival (Zhou et al. 2005). Among the entire collection of 81 breast cancer samples (groups A+B), DNA-binding complexes with the p50 NFkB subunit were almost twofold more abundant than those with the p65 NFkB subunit, although these two independently measured parameters were tightly correlated (rs=0.86, P<0.0001). As illustrated in Fig. 1 (panel A), the group B breast cancers with a median under 0.5-fold lower ER content showed significantly higher NFkB DNA-binding than the group A tumors with higher ER content, indicating that hormone-dependent breast cancers might be subset according to NFkB activity and ER content. Endocrine-Related Cancer (2005) 12 S37–S46 www.endocrinology-journals.org S39 Metastatic relapse rates and disease-free survival (DFS) status were available only for the group B cases; despite uniform adjuvant treatment with tamoxifen, the 13/59 primary breast tumors destined for later relapse possessed significantly higher NFkB p50 DNAbinding than the 46/59 similarly staged, ER-positive cases not destined for relapse. The generally lower NFkB p65 DNA-binding activities followed a similar trend that did not reach statistical significance. Regression tree analyses were performed on both the p50 and p65 DNA-binding values to establish statistical cutpoints (0.95 and 0.75 respectively) that would optimally separate the Kaplan–Meier DFS curves for high versus low NFkB subsets within the group B cases. As shown in Fig. 1 (panel B), the higher NFkB p50 DNA-binding values were associated with significantly reduced DFS (P=0.04). Likewise, the p65 DNA-binding DFS curves showed a similar trend, but their separation did not achieve statistical significance (P=0.09). Among the numerous other biomarkers previously determined in the group B tumors, only AP-1 DNAbinding and uPA expression also showed significant prognostic associations with patient outcome assessed by Kaplan–Meier DFS plots (P=0.009 and P=0.001 respectively). Furthermore, we found that tumor ErbB2 and uPA protein levels, as well as AP-1 DNA-binding activities, correlated significantly with both NFkB p50 and p65 DNA-binding values (Zhou et al. 2005). Mechanistically, the observed NFkB correlation with ErbB2 expression could have resulted from the reported activation of the NFkB pathway by ErbB2 RTK signaling (Romieu-Mourez et al. 2002, Biswas et al. 2004). Likewise, the NFkB correlations with uPA expression and AP-1 DNA-binding probably reflect the fact that the uPA gene is known to be transcriptionally activated by both NFkB and AP-1 transcription factor complexes working in concert (Hansen et al. 1992, Sliva et al. 2002). Given these mechanistic links and the fact that large clinical trials have consistently demonstrated that uPA expression independently identifies a high-risk subset of earlystage breast cancers (Eppenberger et al. 1998), it is tempting to conclude that tumor cell invasiveness and motility mediated by increased uPA expression and induced by activated NFkB and AP-1 complexes contributed to the relapse rate and reduced DFS survival observed in our group B, ER-positive primary breast cancer cases that failed to benefit from adjuvant tamoxifen. NFkB inhibition can reverse endocrine resistance NFkB p50 and p65 subunit DNA-binding activities were also assessed in a panel of breast cancer cell lines representing four different clinical phenotypes (ERpositive/ErbB2-negative, ER-positive/ErbB2-positive, Group A (n=22):ER ≥100 fmol/mg Group B (n=59): ER 21-87 fmol/mg