Deciphering Steroid Receptor Crosstalk in Hormone-Driven Cancers

Abstract

Steroid hormone receptors (SRs) have a multitude of functions in human biology and disease progression. The SR family of related ligand-activated transcription factors includes androgen, estrogen, glucocorticoid, mineralocorticoid, and progesterone receptors. Antiestrogen or estrogen receptor (ER)–targeted therapies to block ER action remain the primary treatment of luminal breast cancers. Although this strategy is successful, ∼40% of patients eventually relapse due to endocrine resistance. The majority of hormone-independent tumors retain some level of SR expression, but sidestep hormone ablation treatments. SRs are known to crosstalk extensively with kinase signaling pathways, and this interplay has been shown to bypass ER-targeted therapies in part by providing alternative proliferation and survival signals that enable hormone independence. Modified receptors adopt alternate conformations that resist antagonism or promote agonism. SR-regulated transcription and SR-binding events have been classically studied as single receptor events using single hormones. However, it is becoming increasingly evident that individual steroids and SRs rarely act alone. Emerging evidence shows that coexpressed SRs crosstalk with each other in hormone-driven cancers, such as breast and prostate. Crosstalk between related SRs allows them to modulate signaling and transcriptional responses to noncognate ligands. This flexibility can lead to altered genomic binding and subsequent changes in SR target gene expression. This review will discuss recent mechanistic advances in elucidating SR crosstalk and the implications for treating hormone-driven cancers. Understanding this crosstalk (i.e., both opposing and collaborative) is a critical step toward expanding and modernizing endocrine therapies and will ultimately improve patient outcomes.

Steroid hormone receptors (SRs) of the nuclear receptor superfamily are ligand-activated transcription factors that regulate diverse physiological functions ranging from development, reproduction, homeostasis, and metabolism (1). SR family members include the mineralocorticoid receptors, androgen receptors (ARs), estrogen receptors (ERs), glucocorticoid receptors (GRs), and progesterone receptors (PRs). These SRs are modular proteins that share a common structure and functional domains, which include the N-terminal domain, DNA-binding domain (DBD), and C-terminal ligand-binding domain (LBD). Binding of ligands to the LBD of SRs acts as a classical molecular switch that alters receptor conformation, effectively shifting the receptor to a transcriptionally active conformation capable of binding coactivators with high affinity.

In addition to their roles in normal physiology, SRs also have well-established functions in hormone-driven cancers. For example, primary diagnosis of breast cancer is subtyped by expression of ERs, PRs, and the human epidermal growth factor 2. ER is expressed in ∼50% to 88% of all breast cancers (2) and is a primary therapeutic target due to its well-documented role in breast cancer cell proliferation, survival, and tumorigenesis. PR is a classical ER-target gene and is used as a biomarker of functional ER that, when present, predicts good response to ER-targeted therapies (3, 4). Currently, these cases are most effectively treated with endocrine therapies aimed at blocking ER action (e.g., tamoxifen and fulvestrant) or estrogen synthesis (e.g., aromatase inhibitors). As ER⁺/PR⁺ tumors progress, they are more likely to become hormone independent, yet retain their SR expression. In the context of elevated or activated signaling pathways that typify breast cancer progression, ligand-independent SR action can ultimately lead to emergence of endocrine-resistant tumors. Previous work has demonstrated that extensive crosstalk occurs between SRs and growth factor receptors (5–8), intracellular and stress-activated kinases within the MAPK superfamily (9–11), as well as downstream signaling components of these kinase pathways (reviewed in Refs. 12–14). Specifically, this interplay has been shown to bypass ER-targeted therapies by providing alternative proliferation and survival signals that promote hormone independence. In addition, modified SRs undergo conformational changes in response to phosphorylation events that alter protein-protein interactions and promote agonist actions of antagonists.

Transition of initially hormone-responsive ER⁺ breast cancers to endocrine-resistant status is a considerable clinical problem. Notably, up to 40% of women with SR⁺ breast tumors exhibit de novo resistance or fail on ER-targeted therapies (i.e., acquired resistance) and eventually progress to metastatic disease (15). Breast cancer recurrence has recently been reported to occur at a steady rate ranging from 10% to 41% and is tightly correlated with nodal status (16). In addition to ERs and PRs, a growing number of studies have also implicated ARs and GRs in breast cancer. AR⁺/ER⁺/PR⁺ breast tumors have a better prognosis (17) when compared with AR⁺/ER⁻/PR⁻ tumors (18). Discovery of the emerging role of GR in breast cancer is more recent. Interestingly, ER⁺/PR⁺ luminal breast tumors with high GR expression have increased overall disease-free survival (i.e., good outcome), whereas high GR expression predicts poor outcome in triple-negative breast cancer (TNBC; subtype that lacks ER, PR, and human epidermal growth factor 2 expression) (19). Trends drawn from studies of coexpression of SRs suggest that SRs may either antagonize or cooperate with each other; understanding this crosstalk may lead to the development of improved approaches to treat hormone-driven cancers.

Most studies of SR-regulated transcription have been performed using a single hormone, and SR-binding events have been classically studied as single-receptor events. However, most cells contain multiple SR family members and are typically bathed in complex mixtures of dozens, if not hundreds, of hormones and other chemical messengers. It is becoming increasingly evident that individual steroids and SRs rarely act in isolation. An emerging paradigm in SR biochemistry is crosstalk between different receptor types, which allows receptors to sense activated signaling pathways and modulate transcriptional responses to noncognate ligands. This review will focus on mechanistic advances in deciphering this context-dependent SR crosstalk, and ultimately, the pharmacological implications for modernizing and expanding therapeutic intervention in hormone-driven cancers. Emphasis will be placed on discussing SR crosstalk in breast cancer; however, other hormone-driven cancers (e.g., prostate cancer) will also be discussed.

ER/PR Crosstalk in Breast Cancer

Seminal clinical studies have investigated the effects of estrogen and progestin on breast cancer incidence. These large clinical trials, the Women’s Health Initiative (20–23) and Million Women Study (24), examined the use of hormone-replacement therapy (HRT) when delivered as estrogen-only HRT or as HRT that contained estrogens and a progestin in combination (i.e., combined HRT). The Women’s Health Initiative study concluded that postmenopausal women receiving combined HRT had increased risk of developing invasive breast cancer compared with a placebo-treated group with an incidence of 0.38% and 0.3%, respectively (20, 21). The Million Women Study reported similar observational findings and concluded that postmenopausal women receiving combined HRT had increased risk of breast cancer compared with women that had not received or were not currently using these therapies (24). Breast tumors that arose in women receiving combined HRT relative to placebo or estrogen alone were larger and of higher grade (21). These findings led to decreased use of combined HRT that has resulted in reduced breast cancer incidence, but remain controversial due to the use of MPA (synthetic progestin with androgenic actions) in postmenopausal women enrolled in the trials. Synthetic progestins may modify breast cancer risk by prolonged activation of PRs or elicit off-target effects on other SRs such as AR or GR (25). More recently, it has been suggested that estrogen-only HRT may not contribute to breast cancer risk in younger women and may provide protection in older women via estrogen-induced apoptosis in breast tumors (26, 27). Conversely, studies (albeit with relatively small cohort sizes) comparing combined HRT containing synthetic (i.e., MPA) and natural progestins found that only synthetic progestins significantly increased breast cancer risk, whereas no increased risk was associated with natural progestins (28, 29). The relative instability of natural progestins may account for the differential outcomes compared with long-lived synthetic progestins such as MPA. This raises interesting questions regarding the appropriate duration and exposure level of PR agonists; HRT is typically delivered as a constant hormone exposure rather than as regulated cycles (4). Taken together, these studies argue the importance in delineating the complexity underlying SR signaling pathways and transcription programs in hormone-driven cancers.

Recent studies demonstrate that PR is a binding partner and major modifier of ER actions, which has consequences for hormone-driven breast cancer growth and implications for endocrine therapy responses (30–34). Progesterone signaling is mediated by two PR isoforms, full-length PR-B and N-terminally truncated PR-A (−164 amino acids; termed the B-upstream segment), which are coexpressed from the same gene (PGR) in breast and other tissues. The first mechanistic studies of ER/PR crosstalk were reported in the context of rapid signaling cascades in hormone-treated breast cancer cell lines (9, 35). Progestins were shown to stimulate proliferation in human breast cancer cells (T47D) by rapid induction of the c-Src/p21^ras/Erk signaling pathway mediated by cytoplasmic ER/PR/c-Src complexes; proliferation was blocked using either antiprogestins or antiestrogens. These authors reported that PR does not interact directly with c-Src, but rather with ER through direct interaction of two ER interaction domains located in the PR N-terminus with the ER LBD. These studies provided the first evidence of ER/PR crosstalk by which ER transmits signals to the c-Src/p21^ras/Erk pathway in response to ligand-activated PR-B. Conversely, a polyproline-rich motif in the N-terminal domain of PR-B was subsequently reported to interact with the Src homology 3 domain of c-Src, and this interaction was required for progestin-induced c-Src kinase activation (10, 36). Progestin-induced rapid activation of c-Src was only observed when ER and PR-B were cotransfected into Cos-7 cells (9), suggesting that both receptors are needed to coordinate the proliferative signal that activates MAPK signaling.

Giulianelli et al. (37) reported ER and PR coassociated at target gene promoters to drive progestin-induced cell proliferation in breast cancer cell models (C4-HD and T47D). The authors demonstrated that the synthetic progestin MPA induced expression and activation of ER as well as rapid nuclear colocalization of ER with PR-B in human breast tumor samples. Chromatin immunoprecipitation (ChIP) studies indicated ER and PR were both required for MPA-induced expression of CCND1 and MYC via binding of ER/PR at their respective promoter regions, which was blocked with antiestrogens (fulvestrant) in part by preventing ER/PR association. RNA interference–mediated silencing of ER inhibited ER, but not PR binding to regulatory sequences, demonstrating that ER and PR interact directly at chromatin to coregulate PR-target gene expression and PR-mediated cancer cell proliferation. Although global gene expression studies were not performed, it is likely that many genes (in addition to CCND1 and MYC) require both ER and PR. Indeed, ER/PR crosstalk appears to occur in both rapid signaling and transcriptional contexts to impact global gene expression and breast cancer cell fate.

Similarly, reciprocal studies from the Lange laboratory [i.e., PR regulation of estrogen/ER-mediated functions (30)] reported global PR-dependent regulation of ER target gene sets associated with endocrine resistance (Fig. 1A). In ER⁺ luminal breast cancer models, PR-B was shown to participate in estrogen-induced ER signaling and transcriptional complexes independent of exogenously added progesterone. This constitutive complex, comprised of ERα, PR-B, and the coactivator and scaffolding molecule PELP1, was observed in human breast tumor samples and multiple breast cancer cell lines. The consequences of this interaction in the presence of estrogen included enhanced ER phosphorylation, increased cellular proliferation, decreased sensitivity to tamoxifen treatment, and altered ER promoter selectivity (Fig. 1A). Global gene expression analysis of estrogen-treated ER⁺ breast cancer cells revealed that the presence of PR-B (but not progestin) was required for estrogen induction of gene expression on select promoters (e.g., CTSD) previously thought to be entirely ER driven. Notably, PR antagonists (onapristone) or PR knockdown blocked expression of these ER target genes. Interestingly, PELP1 has also been shown to mediate estrogen-driven ER and AR complex formation to induce androgen-independent AR activation in prostate cancers (38). Taken together, these studies (30, 37) demonstrate that breast cancer cells harboring both ER and PR-B may be sensitive to exposure of either proliferative hormone (i.e., estrogen or progesterone). It is possible that SRs can readily substitute functionally for each other via crosstalk. Utilization of protein-protein interactions with related SRs (i.e., a form of sensing alternative ligands) allows SR pairs to drive oncogenic gene programs and thereby escape inhibition of one receptor type to promote endocrine resistance.

Figure 1.

ER/PR crosstalk in breast cancer. (A) PELP1, ER, and PR-B form a constitutive signaling and transcriptional complex. PR-B expression, but not progestin, is required for estradiol (E2) stimulated regulation of novel ER/PR/PELP1 target genes (e.g., CTSD) associated with breast cancer progression to endocrine resistance. PR antagonists [onapristone (ona)] blocked expression of these E2-driven target genes (30). (B, top) E2-stimulated ER recruits cofactors within the genome to activate target genes that promote cell proliferation [e.g., cell cycle activators such as cyclin D1 (CCND1) and MYC]. (B, bottom) P4-stimulated PR interacts with and reprograms E2-stimulated ER to novel binding sites that contain progesterone response elements (PREs). Target gene expression is shifted away from proliferation and toward a new profile associated with cell differentiation and apoptosis (e.g., cell cycle inhibitors such as CDKN2C and CDKN1B) (31). IGFR, IGF receptor.

Large-scale studies have recently confirmed that ER/PR complexes mediate global changes in chromatin binding and ER-driven gene expression (31, 32). Using ChIP-sequencing (ChIP-seq), Mohammed et al. (31) demonstrated that progestin treatment (natural and synthetic) induced PR association with ER complexes at chromatin in breast cancer cell models (MCF-7 and T47D) when combined with estrogen (Fig. 1B). The authors observed that a net gain of ER binding occurs at novel transcriptionally active loci, many of which coincided with nearby PR-binding events, suggesting the presence of a functional ER/PR complex. Specifically, this progestin-driven ER/PR association was found to induce global reprogramming of ER genomic recruitment; ER was shifted away from estrogen-response elements (EREs) and onto progesterone-response elements (Fig. 1B). This study also reported the growth inhibitory actions of progestin in estrogen-treated xenograft models and primary ER⁺ breast tumor explants, which were additive with an ER antagonist (tamoxifen). This observation could be due to altered ER binding moving from mitogenic loci to PR-controlled cell death and apoptotic genes. The authors suggest that this is associated with good clinical outcome in a cohort of 1959 patients with breast cancer and concluded that women with breast cancer should take progesterone or progestins (i.e., agonists) in combination with antiestrogen therapies (31). A caveat of this study was that high hormone doses (i.e., 10 times higher than physiological or pharmacological norms) were used to demonstrate progestin antagonism of estrogen. Prior to the advent of antiestrogens such as tamoxifen, high-dose estrogens (diethylstilbestrol) were historically used as breast cancer treatments (39). Additionally, it is becoming clear that progesterone is a potent driver of normal as well as breast cancer stem cell outgrowth (40–42). Hence, the use of progestins as a therapeutic agent for luminal breast cancers must be carefully considered and will require further investigation with the aid of genome-wide technologies and heterogeneous breast tumor models.

In corroboration with the Mohammed study, Singhal et al. (32) reported that estrogen or progesterone alone regulate 85% similar gene sets in breast cancer cell lines or primary breast tumor explants. Furthermore, combined treatment (estrogen plus progesterone) led to decreased expression of oncogenic gene programs involved in cellular proliferation, survival, and metastasis. The authors demonstrated that PR dominantly controlled ER genomic binding in response to dual hormone treatment consistent with progestin-only phenotypes, suggesting a net antagonism of estrogen-related gene expression. T47D tumor xenograft studies showed that a selective PR antagonist (CDB4124 and telapristone) abrogated estrogen-induced tumor growth and was synergistic with the ER antagonist tamoxifen. These studies also used high hormone doses (estrogen; 10 nM), but made opposite conclusions; women with breast cancer should take antiprogestins in concert with antiestrogen therapies. Taken together, these studies (30–32) confirm that extensive ER/PR crosstalk occurs at the genome regulatory level and illustrate the profound impact of PR and ligands on ER actions (Fig. 1). They also reveal major knowledge gaps with regard to the inclusion of progestins/antiprogestins in breast cancer treatment regimens. Notably, progestin treatment of breast cancer cells that had been pretreated with estrogen shifted the cells from luminal A toward a basal-like phenotype (43). In addition, a recent study demonstrated that sustained progestin treatment altered up to ∼50% of the estrogen-dependent transcriptomes in ER⁺/PR⁺ breast cancer patient-derived xenografts (34). Interestingly, PR was shown to physically associate with the PolIII complex and decrease the expression of tRNAs needed for translation in progestin-treated tumors. These data indicated that progestins might indirectly modulate ER-driven tumor growth by restricting the tRNA-driven translation of ER-regulated genes related to breast cancer growth. Given the renewed interest in utilizing natural progestins or antiprogestins, it will be necessary to perform comparable studies in heterogeneous breast tumor models and with diverse ligands. This will contribute to fully resolving the context dependence of SR action to define more durable therapy regimens.

Although PR-A and PR-B share structural and sequence identity downstream of the B-upstream segment, a growing number of studies have begun to tease apart isoform-specific differences to demonstrate that these isoforms transcriptionally regulate unique gene sets distinct of each other (44–46). Specifically, Singhal et al. (45) demonstrated that PR-A and PR-B are recruited to distinct but overlapping genomic sites and interact with different sets of coregulators to differentially modulate estrogen signaling. Of the two isoforms, PR-A inhibited gene expression and ER chromatin binding significantly more than PR-B. In contrast, PR-B dominantly activated more genes relative to PR-A and changed the location of ER chromatin binding. The authors observed that PR-A–rich gene signatures in patient tumors were correlated to poorer survival outcome compared with PR-B–rich patient tumors. However, genetic data acquired using cell lines did not accurately predict tumor behavior (45). Rojas et al. (47) reported that PR-B–rich tumors were correlated to poorer prognosis relative to PR-A–rich tumors. These opposing conclusions suggest that additional factors contribute to PR isoform-specific influence on breast cancer biology, particularly in the context of high tumor heterogeneity and multiple signaling aberrations that typify human breast cancers. Ultimately, PR isoform-specific actions contribute an additional layer of complexity toward delineating ER/PR crosstalk in breast cancer.

ER/AR Crosstalk in Breast Cancer

AR is a primary therapeutic target in prostate cancers; more recently, however, AR has also been implicated in breast cancer progression. Indeed, AR is expressed in ∼80% and 60% of primary and metastatic breast carcinomas, respectively (48). As with ER and PR, expression of AR correlates to better prognosis in ER⁺/PR⁺ breast tumors (17) when compared with AR⁺/ER⁻/PR⁻ tumors (18). Moreover, patients with AR⁺/ER⁺/PR⁺ tumors typically have smaller tumors, decreased Ki-67 expression (i.e., proliferation marker), and better disease-free survival compared with patients with AR⁻/ER⁺/PR⁺ tumors (49, 50). Mechanistic studies examining ER/AR crosstalk indicated that androgen-mediated AR signaling generally has an antagonistic effect on estrogen-stimulated proliferation in ER⁺ breast cancer models (51–53).

Additional studies reported the antagonistic effect of AR on ER transcriptional activity (Fig. 2A) (54, 55, 57). For example, AR overexpression was shown to potently suppress ER-mediated proliferation and block output of an ERE-driven luciferase reporter in T47D cells (54). Overexpression of functionally impaired AR variants or the AR DBD indicated that the AR DBD was sufficient to block ER activity, suggesting that AR occupancy of DNA regulates ER-mediated gene expression. Electrophoretic mobility shift assays, in parallel with molecular modeling studies, further corroborated these findings by demonstrating AR binding to ERE sequences. The authors demonstrate that ER and AR are both recruited to estradiol (E2)-responsive promoters [i.e., CTSD and PGR (PR)] and conclude that AR can prevent activation of estrogen-driven proliferative genes by competitive binding to a subset of EREs. Treatment with bicalutamide (first-generation AR antagonist) blocked AR binding to EREs. Moreover, AR was shown to significantly associate with disease outcome in ER⁺ breast cancer; low AR levels conferred a 4.6-fold increased risk of breast cancer-related death. Follow-up large-scale studies (i.e., ChIP-seq and microarrays) investigated ER/AR crosstalk at the genomic and transcriptional levels in luminal breast cancer cell models (ZR-75-1) (55). These studies demonstrated reciprocal interference between DHT (potent AR ligand) and estrogen-induced transcriptomes. Cotreatment with DHT significantly affected 26% of the estrogen-regulated gene programs (e.g., proliferation and survival); conversely, 15% of DHT-induced target genes were affected by estrogen cotreatment. A majority of these changes were antagonistic (78% to 83%). Moreover, androgen response elements at ER binding sites were enhanced following DHT cotreatment, and AR binding was also enriched at estrogen-responsive genes. Taken together, these studies (54, 55) demonstrate that AR competes with ER for a subset of estrogen-driven binding sites and provides mechanistic insight into the antagonistic interplay of ER/AR crosstalk and the importance of the relative AR/ER expression ratio in luminal breast cancer.

Figure 2.

ER/AR crosstalk in breast cancer. (A) Antagonistic ER/AR crosstalk. (1) Estradiol (E2) stimulates ER binding to EREs to drive cell proliferation. (2) Ligand-activated AR or overexpression of constitutively active AR competes with ER for binding at regulatory regions of ER target genes (3) to block E2-stimulated cell growth (54, 55). (B) Cooperative ER/AR crosstalk. (1) E2 stimulation induces both (2) ER and AR binding at EREs to mediate cellular proliferation. (3) Enzalutamide (Enza; second-generation AR antagonist), but not bicalutamide, inhibits E2-mediated proliferation of ER⁺/AR⁺ breast cancer cells (56).

Although AR⁺ expression is associated with improved outcome in ER⁺ breast cancer, AR has been shown to promote oncogenic signaling in tamoxifen-resistant settings (58). More recent preclinical studies from other investigators (56, 59, 60) have drawn contrasting interpretations regarding ER/AR crosstalk and suggest that ER and AR are cooperative rather than antagonistic of each other (Fig. 2B). Cochrane et al. (56) examined the effect of enzalutamide in AR⁺ breast cancer models (both ER⁺ and ER⁻). Enzalutamide is a selective second-generation AR antagonist that binds AR with fivefold higher affinity than bicalutamide, impairs AR nuclear translocation, and lacks agonist activity at effective doses. This study demonstrated that enzalutamide, but not bicalutamide, inhibited estrogen-mediated proliferation in breast cancer cells and ER⁺/AR⁺ MCF-7 xenografts. Interestingly, enzalutamide was observed to inhibit estrogen-driven tumor growth as effectively as tamoxifen. Enzalutamide inhibited androgen-driven tumor growth regardless of ER status (ER⁺ or ER⁻) in xenograft models by increasing apoptosis. The authors also demonstrated that high ratios of AR/ER (more than twofold) in ER⁺ patients with breast cancer indicated an over fourfold increased risk of recurrence/tumor progression while on tamoxifen, suggesting that AR is involved in endocrine resistance. Interestingly, separate studies demonstrate that AR can collaborate with ER to enhance its transcriptional activity in aromatase inhibitor-resistant breast cancer cell models (59). Follow-up studies examined global AR chromatin binding and observed that estrogen induced AR binding at unique sites compared with DHT treatment (60). Specifically, estrogen treatment enriched for AR binding sites at EREs and had considerable overlap with ER binding sites, which included gene targets like GREB1 and PGR. Enzalutamide attenuated this response by reducing overall genomic ER binding and estrogen-mediated proliferative genes and was observed to be synergistic with tamoxifen and fulvestrant. Additionally, enzalutamide significantly reduced tumor growth in tamoxifen-resistant MCF-7 xenograft and ER⁺/AR⁺ patient-derived xenograft models. Taken together, these studies provided preclinical evidence that second-generation AR antagonists (enzalutamide) are effective with current breast cancer therapies such as tamoxifen and can potentially be used in tumors that have become resistant to traditional endocrine therapies. Although the studies discussed previously appear to make conflicting conclusions, they highlight the complexities underlying ER/AR crosstalk and demonstrate the context dependence of diverse AR ligands [i.e., agonists or antagonists (54, 55) and the influence of hormone ablation conditions (56, 60)]. Cotreatment with androgen and estrogen may play a more protective role in which AR antagonizes ER action, whereas hormone ablation via AR antagonists disrupts cooperative actions between AR/ER. A possible explanation that reconciles conflicting reports regarding the role of AR in breast cancer relates to hormonal influences on the breast in premenopausal vs postmenopausal women. Studies suggesting the protective effect of androgens were performed in the presence of estrogen, which more closely models the premenopausal state (54, 61). In contrast, circulating levels of estrogen are low in postmenopausal women with ER⁺ breast cancer, particularly in those being treated with aromatase inhibitors (62). Interestingly, AR and ER were recently shown to coregulate two androgen-metabolizing enzymes in breast cancer: UDP-glucuronosyltransferases UGT2B15 and UGT2B17 (63). These enzymes can functionally affect both ER and AR activity by directly or indirectly reducing levels of their respective activating ligands. Thus, AR action is highly context dependent and reliant on a number of factors: menopausal status (pre- vs post-), relative AR/ER expression ratio, and endogenous androgenic/estrogenic hormone levels. Clearly, modulating one SR will affect the other(s); this crosstalk has substantial implications regarding the pharmacology of antiandrogens and antiestrogens when applied as combination therapies in disease treatment.

SR/GR Crosstalk in Breast Cancers

As with ER, PR, and AR, GR activity is context dependent in breast cancer cells. Indeed, GR signaling appears to differ according to ER expression, having distinct functions in ER⁺ vs ER⁻ breast cancer. High GR expression in TNBC is significantly associated with chemoresistance and disease recurrence and serves as a poor prognostic marker in ER⁻ breast cancers (19). However, in luminal breast cancer subtypes that are ER⁺/PR⁺, high GR expression is associated with an increased overall disease-free survival and a more favorable prognosis (19). Thus, ER/GR crosstalk may account for the differential impact of GR expression and activity across breast cancer subtypes (luminal vs TNBC). Treating breast cancer cells with corticosteroids (e.g., dexamethasone) and estrogen can cause distinct effects from either hormone treatment alone and suggests there is a complex interaction between ER and GR in which multiple hormones are present in physiological systems. Mechanistic studies performed in ER₊ breast cancer models suggest that ER/GR crosstalk may mirror ER/PR crosstalk through SR global reprogramming of the chromatin landscape (64). Notably, GR is capable of binding progestins at similar affinities to its own endogenous ligands (65). Many antiprogestins (e.g., RU486) exhibit similar binding affinities for GR and PR (66, 67). Similarly, PR is responsive to corticoid steroids and able to bind to glucocorticoid response elements and vice versa for GR, thus distinguishing ER/PR vs ER/GR crosstalk can be challenging. Understanding the mechanisms underlying ER/GR crosstalk has great potential for studying multiple disease states, including inflammatory disease (68) and cancers (e.g., breast, endometrial, and prostate) (69).

Early studies reporting ER/GR interaction used a dynamic-assisted loading model (70). This model demonstrated in vivo that two SRs are unable to compete for the same binding site; instead, one receptor serves as a pioneer factor to assist chromatin binding of the secondary receptor (71). This model was used to demonstrate ER/SR crosstalk using engineered ERE/steroid-response element (SRE) repeats bound by PR, AR, and GR derived from the rat prolactin (PRL) promoter (70). Using this system, ER and GR actions were shown to be reciprocal to each other. Both GR and ER were recruited to the ERE-rich arrays only in response to dual ligand treatment (estrogen plus dexamethasone); dexamethasone alone was insufficient to recruit GR. Thus, glucocorticoid treatment may offer an alternative way to modulate estrogen response in patients with breast cancer. Deletion mutant analyses demonstrated that corecruitment of GR to the PRL array was dependent on the ER coregulator interaction domain (located in the LDB). The authors also queried the effect of ER/GR chromatin interplay at the transcriptional level and found that ER/GR crosstalk has cooperative and antagonistic gene-specific outcomes.

Miranda et al. (64) extended these studies by describing a novel mechanism for direct ER/GR crosstalk at the genome level, by which coactivation of the two receptors leads to reorganization of chromatin structure and global rearrangement of receptor binding (Fig. 3). Chromatin architecture can markedly restrict SR access to SREs to enable tightly controlled cell type–specific activity in response to hormone. Coligand treatment (estrogen plus dexamethasone), vs single ligand treatment, reprogrammed ∼25% of ER binding sites enriched at SREs and GR binding sites enriched at EREs, but also repressed binding sites for each SR, respectively (64). Additionally, DNaseI digestion demonstrated that cotreatment with estrogen and dexamethasone promoted reprogramming to new ER and GR binding sites by inducing changes in DNA accessibility (e.g., via enrichment of AP1 motifs). These findings from Miranda et al. (64) demonstrate that ER/GR crosstalk functions in both directions; that is, GR reprograms accessibility for ER, and ER modulates the chromatin landscape for GR access. Separate studies (33, 72, 73) also confirmed that AP1 is required at ER/GR binding sites, specifically at the TFF1, PGR, and CCND1 promoters (72). ER activation increased GR chromatin association at ER, FOXO, and AP1 response elements, and ER also associated with GR response elements (73). Taken together, these studies suggest that the time of induction for each respective SR during disease progression may be critical in dictating specific binding patterns of other SRs across the genome in response to hormones.

Figure 3.

ER/GR crosstalk in breast cancer. SRs reorganize local nucleosome structures to recognize and bind to their respective response elements in chromatin. (1, 3, and 4) A small percentage of potential binding elements are available for binding for each respective receptor. (2) Most binding sites are blocked for direct receptor interaction. (5) For a subset of elements, the action of one receptor can dramatically modulate chromatin access for an alternate receptor. For example, ER can modulate chromatin to allow binding at a (5) site normally closed to GR or vice versa. (6) Access for a given receptor can also be reduced by local chromatin remodeling (64).

Profiling of the SR landscape in primary human male breast cancer tumors has indicated extensive overlap among ER, PR, AR, and GR chromatin binding (74). Posttranslational modifications of GR, such as SUMOylation, have also been shown to mediate transcriptional repression of ER. Dexamethasone treatment significantly repressed estrogen-activated genes via recruitment of SUMOylated GR to ER-bound enhancers; this was shown to block select ER-bound enhances such as the MegaTrans complex (75). This event was associated with poorer metastasis-free outcomes in patients with breast cancer. GR SUMOylation has also been shown to fine-tune GR chromatin occupancy on a genome-wide level and suggests that SUMOylation can have a profound impact on SR crosstalk (76).

AR/GR crosstalk also occurs extensively in prostate cancer. AR and GR cistromes and transcriptional programs exhibit considerable overlap. Ligand-activated GR acts as a partial antiandrogen and attenuates AR-dependent transcriptional programs, suggesting that GR may act as a tumor suppressor in prostate cancer (77). Standard therapies for prostate cancer include drugs that target AR signaling such as abiraterone or enzalutamide. Resistance to enzalutamide is now a major clinical problem due to widespread use. Emerging evidence demonstrates that ligand-activated GR may confer enzalutamide resistance by shifting from an antiproliferative to pro-cell survival role when AR is blocked (78). Comparative AR and GR transcriptomes studies showed that GR bypasses enzalutamide-mediated AR blockade without the need for restored AR function. ChIP-seq analyses revealed that GR binds to >50% of all AR binding sites in enzalutamide-resistant cells. A separate study from Isikbay et al. (79) corroborated these findings and concluded that GR inhibition may be a useful adjunct to AR antagonists for treating prostate cancer. Taken together, these findings establish a mechanism of escape from AR blockade via GR/AR crosstalk [much like that proposed for escape from ER blockade by ER/PR (30) or ER/AR (58, 60) crosstalk] and demonstrate the need to fully understand SR crosstalk to delineate mechanisms driving therapy resistance in hormone-driven cancers.

Conclusion

SRs provide exciting opportunities for translational research aimed at exploring both monotherapy and combinatorial endocrine therapies. An overwhelming majority of past work investigating SR signaling and transcriptional programs in hormone-driven cancers has focused on individual SRs and single hormone treatment conditions. Although this has provided an invaluable knowledge base to our understanding of SR function, it is seldom physiologically relevant in the biological context because tissues are bathed in a complex milieu of hormones, and cells coexpress multiple SRs that are activated concurrently. This high degree of complexity and functional redundancy perhaps ensures highly precise SR sensing of hormonal fluxes over the lifespan. In SR⁺ cancers subjected to long-term endocrine therapy, long-lived disseminated SR⁺ tumor cells sidestep hormone ablation and ultimately progress to endocrine resistance. In this review, we have provided an in-depth overview and analysis of recent literature reporting opposing as well as collaborative SR crosstalk in hormone-driven cancers. Indeed, the strategy of targeting multiple SRs or SR combinations such as ER/PR, ER/AR, ER/GR, and AR/GR is likely to prove itself clinically useful. However, challenges include maintaining a high quality of life in patients who are deprived of all cognate hormones. Selected antihormone combinations are not without risks to other normal hormone-regulated organ systems. For example, although tamoxifen is predominantly an ER antagonist in the breast, it has ER agonist actions in the uterus, and long-term use is associated with increased endometrial cancer risk (80). Thus, the combination of tamoxifen with an antiprogestin (i.e., a PR antagonist in the uterus) may improve breast cancer outcomes at the substantial expense of promoting reproductive cancers. Clearly, a new generation of highly SR-specific and organ site–selective ligands is urgently needed to overcome these challenges. Finally, better public support for basic scientific discovery is imperative for translation. A deep mechanistic and global understanding of context-dependent SR crosstalk at both the genomic and proteomic levels is necessary to pinpoint the subset of tumors appropriate for intervention and which combinations of hormones/antihormones are required to achieve therapeutic efficacy and durably prevent therapy resistance.

Glossary

Abbreviations:

AR

androgen receptor

ChIP

chromatin immunoprecipitation

ChIP-seq

chromatin immunoprecipitation sequencing

DBD

DNA-binding domain

E2

estradiol

ER

estrogen receptor

ERE

estrogen-response element

GR

glucocorticoid receptor

HRT

hormone-replacement therapy

LBD

ligand-binding domain

PR

progesterone receptor

SR

steroid hormone receptor

SRE

steroid-response element

TNBC

triple-negative breast cancer