Regulatory mechanisms of steroid hormone receptors on gene transcription through chromatin interaction and enhancer reprogramming

Abstract

Regulation of steroid hormone receptors (SHRs) on transcriptional reprogramming is crucial for breast cancer progression. SHRs, including estrogen receptor (ER), androgen receptor (AR), progesterone receptor (PR), and glucocorticoid receptor (GR) play key roles in remodeling the transcriptome of breast cancer cells. However, the molecular mechanisms by which SHRs regulate chromatin landscape in enhancer regions and transcription factor interactions remain largely unknown. In this review, we summarized the regulatory effects of 3 types of SHRs (AR, PR, and GR) on gene transcription through chromatin interactions and enhancer reprogramming. Specifically, AR and PR exhibit bi-directional regulatory effects (both inhibitory and promoting) on ER-mediated gene transcription, while GR modulates the transcription of pro-proliferation genes in ER-positive breast cancer cells. In addition, we have presented four enhancer reprogramming mechanisms (transcription factor cooperation, pioneer factor binding, dynamic assisted loading, and tethering) and the multiple enhancer-promoter contact models. Based on these mechanisms and models, this review proposes that the combination of multiple therapy strategies such as agonists/antagonists of SHRs plus endocrine therapy and the adoption of the latest sequencing technologies are expected to improve the efficacy of ER positive breast cancer treatment.

Introduction

Transcription factors (TFs) play a crucial role in regulating gene expression, holding significant implications for health and disease. Various regulatory elements, including core promoters, promoter-proximal elements, and enhancers, are essential for fine-tuning gene expression levels [1]. Enhancers, as key DNA regulatory elements, activate gene transcription by long distance looping to interact with target gene promoters, regardless of their orientation, distance, and genomic location [2]. The identification of potential enhancers requires profiling the characteristic histone modifications (such as H3K4me1 and H3K27ac) and the binding of transcriptional activators and pioneer factors to enhancers [3, 4]. Recently, enhancer RNAs (eRNAs) have been reported as a sharp feature of active enhancers [5], and thus it has been utilized for the systematic discovery of enhancers across the genome [6].

Estrogen receptor alpha (ERα) is a master transcription factor that predominantly occupies enhancer regions under estradiol (E2) treatment in breast cancer. Distinct ER enhancer occupancy and different ER-DNA binding patterns are defined as ER-involved enhancer reprogramming, which results in different expression profiles and prognoses of breast cancer patients [7]. ER-mediated enhancer reprogramming drastically activates or represses typical ER-target genes, which in turn influence tumorigenesis, tumor progression and endocrine resistance [8, 9]. Several types of proteins participate in regulating ER-mediated enhancer reprogramming progression. Pioneer factors, such as Forkhead Box A1 (FOXA1) and GATA Binding Protein 3 (GATA3), facilitate reprogramming by remodeling chromatin landscape and recruiting other proteins [10, 11]. Chromatin-remodeling complexes such as polycomb complexes (PcGs) and trithorax complexes (TrxGs) could modulate the chromatin accessibility for TF recruitment [12]. The eRNAs play a role in ER-mediated enhancer-promoter contacts and loop formations. Except the above-mentioned participants, steroid hormone receptors (SHRs) including progesterone receptor (PR), glucocorticoid receptor (GR), and androgen receptor (AR) have been identified as crucial factors modulating ER-mediated enhancer reprogramming.

In this review, we discussed the characteristics and molecular functions of ER-mediated enhancer reprogramming in breast cancer, and explored the roles of other steroid hormone receptors (PR, GR and AR) in ER-mediated enhancer reprogramming. Additionally, we focused on the TF-enhancer interactions shaping chromatin landscapes, spotlighting the effects of enhancer reprogramming on the development of novel treatment strategies for breast cancer.

Breast cancer and estrogen receptor (ER)

Breast cancer is the most prevalent cancer among women, accounting for about one-third of all newly diagnosed cancer cases worldwide [13]. With a global mortality rate ranging from 10 to 20%, breast cancer surpassed lung cancer as the primary cause of death for women in 2020 [14]. The heterogeneity of breast cancer necessitates targeted biological therapies and treatments aimed at minimizing adverse effects. The standard approach for treating breast cancer is neoadjuvant therapy, a chemotherapy primarily using targeted agents [15]. Approximately 70–80% of patients with early-stage cancer that has not spread beyond the axillary lymph nodes can be cured through neoadjuvant therapy and surgical intervention [16]. However, advanced cancer remains uncurable within current therapeutic options. The treatment for advanced breast cancer patients aims at extending survival, alleviating symptoms, reducing side effects, and maintaining or enhancing quality of life.

In 2000, Perou and Sorlie classified breast cancer into four main subtypes, based on the 4 indicators ER, PR, human epidermal growth factor receptor 2 (HER2), and Ki67, including (1) luminal A – ER⁺ and/or PR⁺, HER2^- and Ki67^low, (2) luminal B – ER⁺, PR⁺ and/or HER2⁺ and Ki67^high, (3) HER2-enriched – ER^-, PR^- and HER2⁺, and (4) basal-like or triple-negative breast cancer (TNBC) – ER^-, PR^- and HER2^- [17]. The luminal type (ER⁺) accounts for approximately 50–80% of all cases, underscoring the significant role of ER and its downstream signalling pathways in breast cancer treatment [18, 19]. Consequently, endocrine therapy targeting ER/E2 signaling has become a preferred strategy for treating luminal breast cancer. Current endocrine therapies mainly include selective ER modulators (SERMs, such as tamoxifen), selective ER degraders (SERDs, such as fulvestrant), and aromatase inhibitors (AIs, such as letrozole, anastrozole, and exemestane). Although single or combined endocrine therapies can substantially decrease the risk of recurrence and mortality [20, 21], approximately 25% of ER-positive breast cancer patients develop endocrine resistance, leading to incurable metastasis and 90% of breast cancer-related deaths [22]. Endocrine resistance is predominantly driven by the upregulation of anti-endocrine genes, and the resistance to targeted drugs suggests potential changes in ER-mediated transcription and interaction patterns.

Multiple hormones such as 17β-estradiol (E2), progesterone (P4), glucocorticosteroid (GC), and dihydrotestosterone (DHT) can regulate the proliferation, invasion, and endocrine sensitivity of breast cancer cells. Their corresponding hormone receptors constitute a nuclear receptor superfamily which play dominant roles in modulating gene transcription. ERa belonging to the nuclear receptor superfamily experiences conformational changes and translocation from the cytosol to the nucleus under E2 treatment. The conserved DNA-binding domain (DBD) of ER proteins can recognize and bind to specific estrogen response elements (EREs) on DNA to regulate gene transcription. This review focuses on ERa rather than ERb, in light of the unknown functions of ERb in reprogramming enhancers (Hereafter, following ER refers to ERa). Genomic recruitment of ER proteins is complex and variable across breast cancer cell lines and human breast tissues. Both chromatin immunoprecipitation sequencing (ChIP-seq) and chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) of ER proteins reveal high variability of genome-wide ER-chromatin interactants, including pioneer factors, chromatin remodelors, eRNAs, and TFs [23–25].

ER interactants

Most of the ER recruitments occur in distal enhancer regions within the genome, particularly at binding sites of pioneer factors, including FOXA1 and GATA3 (Fig. 1A). FOXA1 is essential for ER-chromatin binding and estrogen-induced gene activation in enhancer regions [23]. An effective ERE can potentially be predicted based on the co-occurrence of FOXA1 and ER bindings, the presence of the H3K4me1 marks, and open chromatin configurations [26]. FOXA1 preferentially binds to the regions marked by active histone modifications such as H3K4me1 and H3K4me2, and exhibits less affinity to the regions marked by H3K9me2 associated with transcriptional repression [27]. Moreover, FOXA1 enhances chromatin accessibility at these specific sites by directly removing nucleosomes, thereby facilitating the recruitment of transcription factors and cofactors.

Fig. 1.

Reprogramming of enhancers by steroid receptors in breast cancer. A Upon estrogen treatment, ER forms homodimers and translocates to the nucleus, where it forms protein complexes with RNA polymerase II and transcriptional co-activators such as BRD4 and CARM1 to co-recruit to classical EREs. For non-canonical EREs, ER can form heterodimers with TFs like AP-1 and SP1. B PRs are capable of recruiting ER to the classical PREs in breast cancer cells following exposure to both estrogen and progesterone; PR can also form protein complexes with TFs like ER, ELK4, and BRCA1, and bind to non-classical sites to activate the transcription of BRCA1 target genes. C GR has a dual function on ER-mediated gene transcription. While Sumo-GRs recruit corepressors to inhibit ER-mediated eRNA transcription, unmodified GR can form protein complexes with ER and AP-1 to activate transcription of downstream target genes. D AR exerts effects on ER-mediated gene transcription. It can replace ER binding on the classical EREs to repress transcription and also recruit ER and co-activators to classical AREs to promote ER-mediated gene transcription

Interestingly, FOXA1 tends to access proximal opening enhancers, whereas GATA3 prefers to modulate distal enhancer-promoter interactions. GATA3 is critical for optimal ER-enhancer binding and the redistribution of ER-chromatin interactions. The ectopic expression of ER-FOXA1-GATA3 complex can induce estrogen-responsive growth in ER-negative breast cancer cell lines such as MDA-MB-231 and BT-549. The previously mentioned tripartite enhanceosome complex proteins, especially GATA3, is necessary for optimal enhancer transactivation [28]. Loss of functional GATA3 remodels the chromatin binding of ER, FOXA1, and EP300 in enhancer regions (such as Enhancer^TFF1 and Enhancer^TFF3) [29]. Additionally, silencing GATA3 also results in the formation of different chromatin loops, thus indirectly enhancing chromatin binding of ER and FOXA1.

Chromatin is constantly remodeled to regulate gene transcription and TF-chromatin binding. The chromatin-remodeling complex proteins primarily include polycomb group proteins (PcGs, also known as PRC complexes) dynamically repressing transcription and trithorax group proteins (TrxGs) promoting transcription activation [30, 31]. RING1B, a core subunit of the PRC1 complex, is crucial for luminal breast cancer development and ER-RING1B co-occupancy on active enhancers. Specifically, RING1B is recruited in a cyclical manner to co-bound sites of ER, FOXA1, and GRHL2, where it regulates estrogen-induced enhancer transcription through nucleosome binding domain of RING1B [32]. DNA-binding proteins (such as the RE1-silencing transcription factor (REST), zinc-finger protein SNAI1 (also known as SNAIL), RUNT-related transcription factor 1 (RUNX1), and proteins containing zinc-finger domains (such as ZNF518A and ZNF518B)) interact with PRC1 or PRC2, thus functionally regulating ER-mediated enhancer transcription [33–36]. Notably, the recruitment of the PRC complexes to chromatin is independent of repressive histone modifications such as H3K27me3 and H2AK119ub1, suggesting that in addition to transcriptional repression function, PRC complexes can also regulate protein interaction in ER-involved enhancer reprogramming [37–40].

One group of TrxG complexes comprise BAF and PBAF which are formed from SWI/SNF and BRM or BRG1, respectively, with each containing up to 15 additional subunits highly conserved throughout evolution. COMPASS complexes, another group of TrxG complexes, are associated with histone modification such as histone acetylation and methylation, and they are highly conserved with WDR5, ASH2, RBBP5, and DPY30 as their core proteins. The COMPASS complexes regulate ER-mediated enhancer reprogramming progress in Enhancer^TFF1 and Enhancer^MYC. The chromatin-interacting protein BAP18, part of COMPASS complex, has been reported to modulate ER recruitment to specific EREs and transactivation of oncogenic enhancers by remodelling chromatin accessibility [41, 42]. In addition, other chromatin-remodeling proteins have been reported to modulate ER-mediated enhancer activation. The epigenetic writer DOT1L is channelled to multiple regulatory sites in the genome by interacting with transcription factors including ER. DOT1L interacts with MEN1 and BAZ1B to catalyze histone H3K79 methylation on the EREs of target genes, thus increasing accessibility of chromatin-ER binding sites, eventually enhancing ER recruitment [43, 44].

Enhancer RNAs contribute to ER-mediated enhancer-promoter interactions and loop formation. eRNA-ER-cohesin complexes are crucial for E2-dependent gene activation, with eRNAs facilitating the formation of enhancer-promoter loops [45, 46]. Subsequent studies reveal that eRNA^TM4SF1 and eRNA^EEMP1 directly interact with the DNA-binding domain of ER, thereby enhancing the interaction between KDM2A and ER at Pol II binding sites [47]. eRNA^PRRX2 and eRNA^UBE2E also recruit ER to specific enhancers with abundant H3K27ac modifications, highlighting their transactivating roles in enhancing ER recruitment in response to E2 treatment [48]. Multiple proteins, such as ELF5, JMJD6, CARM1 and BRD4, have been reported to be involved in regulating eRNA transcription and ER chromatin recruitment. For example, ELF5 binds to ER-chromatin binding sites, channeling ER to a new region with a motif of ER/FOXA1 cis-element, resulting in an acquired resistance to endocrine therapy [49]. However, the precise identification and functional exploration of enhancer RNAs are scarce. Therefore, more advanced techniques for detecting eRNA-regulated enhancer reprogramming remain to be developed.

Other SHRs including PR, androgen receptor (AR), and glucocorticoid receptor (GR) participate in regulating cell division, organ development, and metabolism [50, 51]. Approximately 90%, 60%, and 86% of human breast cancer samples are positive for AR, GR, and PR, respectively, and the co-expression of ER and other SHRs is significantly correlated with favourable clinical outcomes [52]. Notably, patients with AR⁺/ER⁺/PR⁺ tumors exhibit significantly better prognosis than those with AR^-/ER^-/PR^- tumors [53]. Furthermore, high GR expression in luminal breast tumors is associated with improved prognosis, underscoring an antagonistic effect of GR on ER⁺ breast cancer [54, 55]. Since all the SHRs contain analogous DNA-binding domains, SHRs can bind to similar genomic regions in cell lines and breast cancer tumor samples. These findings imply that the crosstalk between SHRs may be competitive or cooperative, resulting in different clinical outcomes (Table 1). Revealing the nuances of crosstalk-induced gene expression allows insights into breast cancer tumorigenesis, development, and responses to endocrine therapy.

Table 1.

Brief summaries of ER-involved enhancer reprogramming

Factors involved in ER-mediated enhancer reprogramming	Function	Reference
SHR: Progesterone receptor (PR)	Tethers ER to PREs with pioneer factors and coactivators to promote PR-target genes transcription	[70, 71]
	Tethers ER to PR-BRCA1-ELK4 binding sites with FOXA1 to reduce aggressive phenotype	[72]
	Interacts with CTCF on PR-orchestrated insulated regions	[75]
SHR: Glucocorticoid receptor (GR)	Cooperatively interacts with ER, AP-1, FOXO1 at GREs and adjacent EREs to facilitate differentiation and reduce EMT	[98, 99]
	Tethers ER to AP-1 binding sites to inhibit ER activity and ER-related eRNAs	[101, 103]
	Hijacks ER on ZBTB16 promoter to induce epigenetic remodelling and loss of histone acetylation	[102]
	Dynamic rapidly chromatin binding on GREs with ER to reduce classical ER-mediated enhancer transactivation	[104, 105, 107]
	Cooperatively with ER to facilitate the recruitment of FOXA1 in an assiated loading manner	[168]
SHR: Androgen receptor (AR)	Alters genomic distribution of ER and coactivators (such as p300 and SRC-3) to reduce ER-regulated cell cycle genes, while upregulated of AR target genes	[110, 122, 123]
	Tethers ER and displaces ER binding at the same EREs on chromatin to reduce ER-mediated gene transcription	[118–120, 124]
Pioneer factors	Forms MegaTrans supercomplex (including RARα/γ, GATA3, AP2γ, STAT1, AP1, and FOXA1) to regulate ER-mediated enhancer activation and enhancer-promoter loopings	[189]
	FOXA1, GATA3, AP2g, and PBX1 dynamicly regulate ER-target gene transcription	[149, 150]
	GATA3 upregulates ER-induced enhancer/promoter activation	[160, 161]
Cofactors, eRNAs and other TFs	RING1B facilitates ER-mediated enhancer-promoter loops by regulating R-loop formation and POL2 elongation with p300	[191]
	ER cooperates with RARa on EREs of ER-target genes (such as TFF1 and GREB1) to activation gene transcription	[137]
	TEAD4/YAP1 act as coactivators of ER and upregulate ER-mediated enhancer activation	[138]
	NURF complex and CTCF participate in enhancer activation of ER-target genes (such as TFF1 and GREB1)	[42]
	GREB1 stabilizes the interaction between ER and other cofactors (such p300 and CBP) on ER binding regions	[139]
	ARID1A-mediated SWI/SNF at ER-bound enhancer regulates the interaction between ER and FOXA1 or FOS/JUN	[143]
	MegaTrans, condensin and eRNAs form megaloops between enhancers and promoters of ER target genes by phase separation	[145]
	ER tethers SP1, AP-1, CEBPb, RUNX1 and PITX1 to regulate target genes of these TFs	[177–181]

Enhancer reprogramming induced by SHR crosstalks

ER-chromatin interaction reprogramming in enhancer regions, also known as enhancer reprogramming, facilitates the development of extrinsic resistance [56]. The crosstalk between ER and other SHRs (including PR, GR and AR) is a key mechanism of enhancer reprogramming.

Progesterone receptor (PR)

PRA and PRB are the two principal isoforms of PR, transcribed from two distinct promoters of the same gene. PRA is characterized by the absence of 164 amino acids at its N-terminus, resulting in a diminished response to progesterone and a weaker interaction with progesterone response elements (PREs), compared to PRB [57]. PRB functions as the primary receptor mediating progesterone-induced transactivation [58]. PRA and PRB exhibit distinct biological functions, specifically, depletion of PRA results in severe reproductive anomalies, including ovarian or uterine dysfunctions and infertility, and PRB facilitates mammary gland development by upregulating the paracrine factor RANKL [59, 60]. Thus, PRB is often regarded as the predominant receptor type involved in the ER-PR crosstalk in breast cancer.

The post-translational modifications (PTMs) of PR, such as phosphorylation, ubiquitination, SUMOylation, and acetylation, significantly influence its capacity to regulate gene transcription [61–63]. Upon ligand binding and growth factor stimulation, various protein kinases phosphorylate serine residues of PRB, such as MAPK, CSNK2A1, and CDK2 [64]. Notably, phosphorylation of PRB-Ser294 residue modulates the nuclear translocation of PRB and enhances the transcription of target genes involved in cell cycle progression (such as CCND1), proliferation (such as MYC), and survival (such as BCL2L1) [65, 66]. Furthermore, phosphorylation of PRB-Ser294 residue antagonizes SUMOylation of PRB-Lys388 residue, thus stabilizing PRB in response to progesterone treatment, further elevating levels of growth factors and activating MAPK pathway [67]. In contrast, PRA protein is constantly SUMOylated due to low phosphorylation level at Ser294 residue [67]. CDK2-mediated PR phosphorylation modifications at Ser345 and Ser400 residue are also critical for PR-mediated transcriptional activation, especially for the activation of HSP8 enhancer during cell cycle progression [68, 69].

PR participates in ER-mediated enhancer reprogramming. Upon progesterone treatment, PR can switch the ER-mediated regulatory network from high proliferation state to low proliferation state by tethering ER to classical PREs, hence attenuating ER recruitment to highly activated enhancers (such as Enhancer^ST6GAL1, Enhancer^CCND1, and Enhancer^MYC) [70, 71]. ER proteins interact with PR and coactivators such as p300, which enhances ER binding to canonical PREs rather than the EREs with progresterone [71]. PR activation can also remodel nucleosomes and recruit ER to enhancers containing BRCA1 motifs, while PR depletion reduces ER binding to DNase-hypersensitive regions [72]. Since less than half of the altered ER binding sites contain FOXA1 motifs, PR-mediated ER chromatin reprogramming does not exclusively depend on FOXA1 [72]. Selective PR modulators (SPRMs) can synergistically enhance the response to SERM. When both ER and PR are activated, PR predominates the transcriptional landscape rather than ER, resulting in a less aggressive tumor phenotype (Fig. 1B) [72]. Indeed, gene expression co-driven by ER and PR is associated with a favorable prognosis in breast cancer patients, suggesting that PR-mediated gene activation may inhibit breast cancer progression [70]. These findings indicate the synergistic inhibitory effect of PR and progesterone on ER-mediated gene transcription, which is consistent with the reports on desirable clinical outcomes (ER⁺/PR⁺) versus poor outcomes (ER⁺/PR⁻) [7, 73].

However, recent studies have demonstrated the positive role of PR and progesterone in promoting ER⁺/PR⁺ breast cancer, since PR is associated with estrogen-independent growth and metastasis. In premenopausal breast cancer patients, estrogen alone and progesterone alone can facilitate tumor growth, while their combined treatment enhances metastastasis of ER⁺/PR⁺ breast cancer [74]. Furthermore, PR alone is sufficient to drive tumor growth and metastasis in ER-negative tumor cells [74]. The induction of basal cytokeratins is dependent on PR-regulated chromatin reprogramming in CTCF binding regions, indicating a positive role of PR in the metastasis ER+/PR + breast cancer [75]. Although ER⁻/PR⁻ luminal cells exhibit increased metastatic capacity and resistance to endocrine therapy and chemotherapy, these cells remain dormant in the micrometastasis state under hormone-deficient conditions, but they revert to the macrometastasis state under estrogen and progesterone combined treatment, suggesting a positive role of progesterone in metastasis [76]. Collectively, these results highlight bi-directional regulatory role (inhibitory and promoting) of PR and progesterone in ER-mediated enhancer reprogramming, which requires further exploration.

Glucocorticoid receptor (GR)

The glucocorticoid receptor (GR) is encoded by the nuclear receptor subfamily 3 group C member 1 (NR3C1) gene, and its two isoforms GRα (777 amino acids) and GRβ (742 amino acids), are produced through alternative splicing [77]. GRβ lacks a 35-amino-acid segment at the C terminus, limiting its interaction with glucocorticoids (GCs) and suppressing endogenous gene expression [78].

Upon binding to glucocorticoid response elements (GREs), GR recruits co-regulators and chromatin remodeling complexes to modulate chromatin accessibility and transcriptional activity of target genes [79, 80]. The transcription of genes mediated by GR can either be activated or repressed, which is affected by GR-DNA binding patterns, GR tethering by other TFs, and post-translation modifications (PTMs) of GR [81]. Several GR-DNA binding patterns in the genome have been identified, such as directly binding to consensus GREs through GR homodimers [82]. A consensus GRE contains two palindromic sequences, enabling binding of two GR monomers to DNA, while the hybrid or half GRE contains only one palindromic sequence, allowing binding of one GR monomer to DNA, with or without other TFs [83]. The hybrid site tends to be in proximity to DNA-bound TFs except for GR, such as NFKB1 and AP-1 [84]. Additionally, negative GREs can repress gene transcription upon GR binding in a specific manner through the recruitment of corepressors and histone deacetylases via transrepression [85, 86]. Aside from post-translation modifications, GR-mediated gene transcription is decreased when other TFs tether GR to their binding sites [82, 87]. For instance, TFs such as IκBα [88], STATs [89], C/EBPs [90], PPARs [91], and LXR [92] tether monomeric GR to their binding sites, thus diminishing GR-mediated gene transcription. GR-mediated gene transcription is also influenced by various PTMs of GR proteins, including phosphorylation, ubiquitination, SUMOylation, acetylation, methylation, nitrosylation, and nitration [93]. The enzymes MAPKs, CDKs, GSK-3, and JNK are involved in GR phosphorylation [94, 95]. Breast tumor kinase phosphorylates the Ser134 residue of GR, thus up-regulating HIF-2a transcription in breast cancer cells [96]. SUMO modification of GR also significantly influences GR-mediated transcriptional activity. GR has two transcription-repressive SUMOylation sites, K297 and K313, in its N-terminal domain (NTD), which limits synergistic transactivation of target genes. Conversely, SUMOylation of GR at K721 facilitates the interactions between GR and the RSMUE protein, thus up-regulating target gene transcription (including FKBP51) [97].

Crosstalk between ER and GR contributes to the development and progression of breast cancer. Co-activation of ER and GR reprograms the chromatin landscape, resulting in a genome-wide rearrangement of SHR in mouse mammary cells. GR facilitates the chromatin accessibility in regions containing specific EREs and GREs, thus enhancing the recruitment of cofactors, ultimately promoting the binding of GR and ER on the enhancers [98]. In the ER⁺/GR⁺ MCF7 cells, GR interacts with ER, AP-1, and FOXO1 in chromatin regions which contain GREs adjacent to EREs, thereby promoting cell differentiation and reducing epithelial-mesenchymal transition (EMT) [98, 99]. The coordination of GR and ER induces pro-differentiation gene transcription such as KDM4B and VDR, while inhibiting Wnt signaling pathway genes such as IGFBP4 [99].

However, the crosstalk between GR and ER has a negative effect on the transcription of proliferation genes. Dynamic assisted loading of GR and ER enhances the binding of ER to GRE on enhancers, and the dominance of GR adjacent to ER impairs ER-mediated gene transcription [100]. GR represses ER-mediated gene transcription by tethering ER to enhancer regions binding AP-1, thus switching co-activator recruitments to co-repressor ones [101]. GR hijacks ER protein to chromatin redistribution regions with low-level histone acetylation modification [102]. Global run-on sequencing (GRO-seq) reveals that GR decreases eRNA expression of ER target genes under specific agonist dexamethasone treatment, and this finding is employed to distinguish luminal A breast cancer from ER⁺/lymph node⁻ breast cancer by comparing eRNA expression (such as eRNA^TFF1 and eRNA^FOXC1) in patient with standard expression profile [103]. These studies demonstrate that GR can block enhancer activation of classical ER target pro-proliferation genes. However, recent studies surprisingly suggest that GR and ER do not compete for binding events due to their rapid chromatin binding kinetics [104, 105]. Rapid number change in ER-bound cofactors in the transcriptome due to the tethering effect of GR affects classical ER target gene activation (Fig. 1C) [106, 107]. Overall, GR may influence ER-mediated anti-proliferative enhancer activation and the progression of ER⁺ breast cancer.

Androgen receptor (AR)

It is well established that androgens effectively inhibit the proliferation of ER⁺/AR⁺ breast cancer cell lines, and AR antagonists have also been found to stimulate the proliferation of these cells [108, 109]. Therefore, AR is considered to function as a tumor suppressor in ER⁺ breast cancer cells, and theoretically, AR agonists can enhance therapeutic responses when they are combined with endocrine and CDK4/6 inhibitor therapies [110]. However, the clinical application of AR agonists against ER⁺ breast cancer is limited, indicating a complex role for AR in ER⁺ breast cancer.

Upon nuclear translocation, AR dimers bind to androgen response elements (AREs) to regulate the transcription of target genes. In approximately 70% of breast cancers, the wild-type AR (wt-AR) and its several variants are found. Over 20 AR variants (AR-Vs) have been identified, of which AR-v7 is one of the most commonly detected and extensively studied in breast cancer [111]. The presence of AR-v7 in the circulating tumor cells (CTCs) of breast cancer patients is positively correlated with cellular response to aromatase inhibitors [112]. In some breast cancer cell lines, such as MCF7, T47D, and MDA-MB-453, the expression of membrane-bound AR lowers nuclear androgen sensitivity and impairs the response to AR antagonists [113]. Several extracellular kinases have been reported to regulate AR functions. For example, the STATs, PKA, PKC, and MAPK phosphorylate AR, thus facilitating AR binding to target gene promoters [114, 115]. STAT3 and cytokines phosphorylate AR, thereby inducing AR nuclear translocation and inflammatory responses [116]. Furthermore, epidermal growth factors (EGFs) and insulin-like growth factors (IGFs) can activate MAPK signaling, hence facilitating AR activation and nuclear localization [117]. These findings illustrate androgen-independent role of AR, partially explain the poor performance of AR antagonists and limited clinical application.

Whether the ER-AR interaction is cooperative or antagonistic is a controversial issue. Some evidence suggests that AR antagonizes ER signalling [118, 119]. For example, AR may suppress the ER signaling pathway in postmenopausal or lower-level circulating estrogen ER⁺ breast cancer patients, which provides potential therapeutic targets [120]. Like tamoxifen, enzalutamide (a second-generation AR antagonist) can significantly inhibit estrogen-induced proliferation and downregulate ER-induced gene transcription [121]. Notably, AR directly occupies EREs and inhibits ER-chromatin interactions, thereby suppressing ER-mediated cell proliferation [122, 123]. AR displaces ER and its co-activators from enhancer regions of proliferation-related genes, suggesting the competitive relationship between AR and ER [120, 124].

However, the cooperative relationship between AR and ER has also been reported in some studies [125, 126]. ER and AR co-occupy target enhancers (such as Enhancer^GATA3 and Enhancer^AQP3) in a manner dependent on multiple essential cofactors such as cAMP responsive element binding protein (CREB)-binding protein (CBP) and p300. FT-6876, a selective inhibitor of CBP/p300, can significantly repress ER/AR co-regulated gene expression, decrease enrichment of CBP/p300, and lower H3K27ac modification at Enhancer^GATA3 and Enhancer^AQP3 in both TNBC cells and ER⁺/AR⁺ breast cancer cells [127]. Taken together, the interplay between AR and ER in breast cancer underscores the complexity of hormone receptor signalling, holding significant implications for therapeutic strategies for different patient populations (Fig. 1D).

Enhancer reprogramming models

Currently, there are four major mechanisms underlying TF binding and enhancer reprogramming in breast cancer, involving distinct high-order assemblies of TFs on the enhancers, as known as transcription factor cooperation, pioneer factor binding, dynamic assisted loading, and tethering.

Transcription factor cooperation

Transcription factors form multiprotein complexes through direct or indirect cooperation to access binding sites in chromatin regions [128]. Direct cooperation involves the physical binding of TFs, while indirect cooperation, also known as collaborative competition, occurs when TFs bind to DNA without direct protein-protein interactions. Several large superfamily TFs in humans form protein-level complexes to compete for the same histone octamer so as to access the DNA segments [129, 130]. AP-1, CREBs, MAFs, and CEBPs from the basic-leucine zipper (bZIP) family form dimers through the leucine zipper domains [131]. Similarly, basic helix-loop-helix (bHLH) proteins, such as MYC, MAX, MAD, E47-NeuroD, and BMAL-CLOCK, form dimers through interactions between their helical domains [132]. These TFs can also form heterodimers, resulting in collaborative regulatory control [133, 134]. The above reports reflect two types (direct and indirect) of TF cooperation.

The cooperative TFs, including SHRs and other receptors (such as AP-1 and MYC), regulate the proliferation, metastasis, and treatment resistance of breast cancer cells (Fig. 2A) [135]. For instance, PRB reprograms the binding of ER to specific EREs in a cooperative mode [136]. Several EREs are modulated by retinoic acid receptor alpha (RARa) in response to E2 treatment [137]. In addition to SHRs, non-classical nuclear TFs also cooperate with ER. For example, TEAD4 proteins interacting with YAP1 serve as co-regulators of ER to induce estrogen/ER target genes and oncogenic progression [138].

Fig. 2.

Illustration of enhancer reprogramming models. A TF cooperative binding model: the binding region of TFs is unaccessible until two TFs concomitantly bind and recruit chromatin remodelling factors. B Pioneer factor binding: pioneer factors such as FOXA1 bind to adjacent regions of chromatin and remove nucleosomes, allowing a secondary factor to access without ATP. C Dynamic assisted loading: the initiating factor binds to a closed chromatin region, followed by the binding of a secondary factor to an unaccessible region once chromatin remodeling factors are recruited. Depending on the chromatin landscape and enhancer region, it is usually a bimodal switch between two TFs. D Tethering model: one TF binds to a chromatin region, and the secondary TF is recruited upon binding, or is initially tethered to the first TF

Various techniques have been developed to detect ER-interacting proteins in vivo or in vitro. For example, rapid immunoprecipitation mass spectrometry (RIME) has been developed to detect endogenous proteins, and GREB1 has been identified as a chromatin-bound coactivator that stabilizes interactions between ER and additional co-factors [139]. Based on this technique, quantitative multiplexed-RIME (qPLEX) has been further developed to validate the interactions between ER and three proteins (CBX3, NIPBL and FOXK1) [140]. By using tandem affinity purification (TAP) coupled with mass spectrometry (MS) techniques, 1,222 interactants of ER in MCF7 cells were identified [141], and in follow-up studies, 1,843, 1,580, and 1,763 interactants of ER under treatment with estrogen, aromatase inhibitors, and tamoxifen were discovered, respectively [142]. Notably, ATP-dependent chromatin accessibility determines the interaction between ER and their interactants, highlighting the critical role of chromatin remodeling in their interaction. For instance, in the SWI/SNF complex, ARID1A functions as a tumor suppressor, and its loss leads to increased breast cancer cell proliferation and endocrine resistance [143]. The recruitment of SWI/SNF by ARID1A to distal enhancers disrupts the interaction between ER and FOXA1 or FOS/JUN, suggesting that chromatin remodeling proteins play a more important role than pioneer factors in the cooperative mode [143].

A potential mechanism underlying TF cooperation in the nucleus is phase separation formed on the intrinsically disordered domains of TFs and specific proteins (usually co-regulators) in transcriptional regions. Phase separation facilitates proximity between specific TFs, thus accelerating transcriptional activation. Estrogen promotes the interaction between ER and MED1 by binding to the ligand-binding domain (LBD) of ER, allowing ER to form heterotypic droplets with MED1-IDR recombinant protein [144]. These findings suggest that structure interactions may contribute to the formation of transcriptional condensates. Additionally, estrogen treatment recruits MegaTrans and condensin complexes to enhancers, thus initiating eRNA transcription and the formation of ‘megaloops’ between these enhancers, ultimately leading to the assembly of an eRNA-dependent ribonucleoprotein (eRNP) through phase separation [145]. AR also participates in phase separation mediated by the LBD of MED1. AR forms high-concentration condensates through intrinsically disordered region (IDR)-mediated multivalent interactions in response to androgen stimulation, and the disruption of AR condensates impairs its capability of enhancer assembly [146, 147]. These findings highlight high-order assemblies of SHRs on enhancers through complicated structures and protein interaction, suggests potential clinical applications for targeting assembly processes.

Pioneer factor binding

The capability of pioneer factors FOXA1 and GATA3 to modulate ER-mediated enhancer reprogramming makes them valuable targets in endocrine therapies [148]. Other pioneer factors, such as AP2g and PBX1, have also been found to be involved in ER-mediated enhancer reprogramming [149, 150]. Unlike most transcription factors, pioneer factors have a unique ability to directly bind to their target regions on inaccessible chromatin [151]. FOXA1 has the capacity to recognize histones in a wrapped state and to interact with specific DNA sequences through its winged helix domain [152]. Specifically, FOXA1 exhibits high-affinity interactions with histones H3 and H4, displacing linker histones to access unique DNA sequences on the nucleosome core. This displacement allows for chromatin loosening and facilitates the recruitment of other transcription factors [153, 154]. FOXA1 rapidly reprograms ER binding in the genome in response to endocrine treatment, particularly in enhancer regions [7]. Furthermore, FOXA1 is largely unaffected by estrogen activation, consistent with its function as an innate pioneer factor that acts upstream the ER-cofactor complex [155]. GATA3 interacts with DNA via its zinc finger domain, serving as a desirable prognostic marker for human luminal breast cancer [156, 157]. Approximately 10 to 15% of luminal breast cancer patients are subjected to GATA3 mutation, which leads to transcriptional defects [158, 159]. Moreover, GATA3 and FOXA1 jointly upregulate ER-responsive genes and induce chromatin reprogramming [160, 161]. Collectively, these findings suggest that distinct pioneer proteins influence genomic ER-chromatin binding in luminal breast cancer.

FOXA1 and GATA3 are also implicated in AR-mediated enhancer reprogramming [27, 162]. For instance, FOXA1 cooperates with AR to regulate ER-targeted gene transcription, while GATA3 prevents AR genomic binding on EREs [162]. Additionally, AP-1 functions as a pioneer factor to facilitate GR binding in the whole genome [163]. AP-1-mediated chromatin binding accounts for over 80% of GR chromatin binding [164]. Overall, pioneer factors facilitate interactions between TFs and chromatin (Fig. 2B).

Dynamic assisted loading

DNA loci can become saturated when increasing numbers of transcription factors bind to specific genomic regions. Consequently, these TFs may be displaced by other TFs with similar DNA-binding domains, which is known as loading in vitro [165, 166]. Instead of competing for the same binding site, two TFs may be engaged in dynamic assisted loading, in which one TF aids in the binding of the other TF [167]. For example, GR facilitates the binding of the ER to GREs, which is a case in point of dynamic assisted loading in the cellular context. ATP-dependent ER and GR jointly recruit FOXA1 in an assisted manner, and ligand-activated ER or GR facilitates significant and transient interactions between FOXA1 and DNA [168]. During this dynamic assisted loading, the chromatin-remodeling complexes can reorganize nucleosomes, thus facilitating TF binding. In addition, chromatin-remodeling complexes such as the SWI/SNF complex can also displace bound TFs from DNA through nucleosome replacement [169, 170]. Thus, dynamic assisted loading results in dynamic chromatin alterations in an ATP-dependent manner.

Mechanistically, the dynamic assisted loading may exhibit symmetrical characteristics, where one TF acts as an initiator at certain sites, while the other TF functions as a follower at different sites, in spite of there being no direct physical interaction between these two TFs (Fig. 2C) [90, 171–173]. Notably, TF NF-κB can facilitate STAT3-chromatin interactions, particularly at Enhancer^Saa1/2, leading to synergistic gene expression [171]. The assisted loading model between GR and cAMP response element-binding protein 1 (CREB1) can be found in response to glucocorticoid. Specifically, GR assists CREB1 in binding to CREB-targeted enhancers, thereby promoting glucose synthesis and gluconeogenic gene expression. Conversely, CREB1 assists GR in binding to the PPARA enhancer, hence enhancing PPARA expression and activating fatty acid oxidation and ketogenic genes [172]. Since most examples of dynamic assisted loading involve GR, it is necessary to further explore the mechanisms underlying GR-related dynamic assisted loading.

Tethering

In the tethering model, one TF is recruited to a specific region of the chromatin by another TF, rather than directly binding to regulatory elements, which is commonly referred to as a non-canonical TF binding event (Fig. 2D). Selective estrogen receptor modulators (usually used as antagonists) have been found to act as agonists to activate ER target genes, indicating the presence of this non-canonical TF binding event of ER [174–176]. ER can regulate the transcription of target genes after being tethered to the binding sites of other TFs such as SP1, AP-1, CEBPb, and PITX1 [177–180]. The chromatin immunoprecipitation-exonuclease (ChIP-Exo) assays and DNA-binding domain (DBD) mutation assays collectively validated the presence of this non-canonical ER tethering mode [181–184]. After being tethered by RUNX1 and AP-1 to specific EREs, a DBD-mutant ER directly activates gene transcription in MCF7 and MDA-MB-231 cells [181]. Similarly, GR and FOXA1 tether ER to GRE, thus activating non-classical E2-target genes in MCF7 cells [183].

GR can also tether pro-inflammatory factors such as NF-κB1 and STAT3, thereby suppressing their activity [89, 185]. GREs have been frequently identified in the regulatory regions of NF-κB and AP-1, suggesting that GR can interact with these proteins to repress their target genes [89]. Typically, GR exhibits non-reciprocal interactions. For example, GR can repress transcription after being tethered to STAT3 binding sites, while GR can promote transcription by tethering STAT3 to GREs, indicating synergistic effects of GR and STAT3 [182]. Additionally, GR can also tether GRIP1 to chromatin binding sites of this protein, thus decreasing chromatin accessibility and histone acetylation levels [186].

It is worth noting that the tethering mode involves only one TF binding to regulatory element, which distinguishes it from the above-mentioned transcription modes (cooperative mode and dynamic assisted loading). In contrast, the cooperative model usually involves two TFs binding to DNA. Moreover, tethering mode requires no involvement of additional chromatin remodeling proteins, which is necessary for dynamic assisted loading. Unlike other modes, in the tethering mode, the involvement of pioneer factors is not accompanied by the fixed binding pattern of TFs to DNA, and gene transcription is regulated by non-canonical complexes, rather than canonical ones.

Enhancer-promoter contact in breast cancer

Typically, enhancers can interact with promoters to exert their function through physical contact in looping, a complex and dynamic process, in which transcription factors, RNA polymerase II, cofactors, and nucleosome-remodeling enzymes are involved. There at least three proposed models of enhancer-promoter contact (Fig. 3A and C): (1) The stable contact model is a compound cis-regulatory element consisting of a stable complex of TFs, coactivators, and transcriptional machinery; (2) The kiss-and-run model posits that enhancer-bound coactivators undergo post-translational modifications at the promoter, with TFs being transferred during transient contact; (3) The diffusion model refers to the model in which TFs are activated at the enhancer and subsequently diffused to the promoter. Most enhancer-promoter contacts mediated by SHRs follow the above-mentioned three models. For instance, the enhancer-promoter contacts of AR-target genes such as HOXC6 and DLX1, as a stable contact, require the presence of FOXA1 and GRHL2 [187]. CCAAT/enhancer-binding protein alpha (C/EBPα) maintains an open chromatin conformation, which is conducive to enhancer-promoter contact following the diffusion model and loading of ligand-activated PR [188].

Fig. 3.

Mechanisms of enhancer–promoter communication. Distinct enhancer-promoter contact models are concluded. A Stable contact model: a ‘compound cis-regulatory’ element is established through a stable complex comprising TFs (blue and red), co-activators (cyan), and the transcriptional machinery (brown). B Kiss-and-run model: Upon transient contact, enhancer-bound histone writers or erasers deposit PTMs at the promoter. C Communication by diffusion model: TFs (orange) are activated at the enhancer, and the originally interacted TFs (blue) diffuse to the promoter and contact the promoter-TFs (red). D Working model of a ‘MegaTrans’ enhancer: DNA-bound ER and FoxA1 dynamically recruit the MegaTrans complex of DNA-binding TFs, including RAR, GATA3, AP2γ, STAT1, and AP1, to the ER active enhancers that contain EREs and FoxA1 motifs

Notably, ER-mediated enhancer-promoter contact does not strictly conform to the aforementioned models. Estrogen receptor is recruited to enhancer region containing estrogen response elements to form a ‘MegaTrans’ complex comprising TFs such as RARα/γ, GATA3, AP2γ, STAT1, AP-1, and FOXA1. This MegaTrans complex is recruited to ER-bound enhancers to regulate enhancer activation. Knockdown of RARα/γ and GATA3 significantly reduces the induction of eRNAs and activation of target genes [188–190].

ER-mediated enhancer-promoter contacts are influenced not only by TFs interacting with ER but also by co-regulatory factors. For instance, the interaction between RING1B and ER regulates the enhancer-promoter contact of the GREB1 gene by modulating R-loop formation and Pol II elongation. Transcription induced by the RING1B-ER complex is independent of acetylation processes involving factors such as p300 [191].

Differential ER binding has been associated with breast cancer endocrine resistance and clinical outcomes. Increasing evidence indicates that ER-mediated enhancer activation and enhancer-promoter contact can be enhanced by multiple proteins, which may induce endocrine resistance [42, 192, 193]. Overexpression of FOXA1 in ER-positive breast cancer cells drives genome-wide enhancer reprogramming to activate pro-metastatic transcriptional programs. Upregulation of FOXA1 triggers transcriptional reprogramming through super-enhancers (SEs) and increases enhancer-promoter contact in endocrine-resistant breast cancer cells [192, 194]. GATA3 and AP-1 can significantly alter transcriptional programs in breast cancer cells [193]. GATA3 modulates the luminal lineage-specific genes, while AP-1 regulates the genes associated with cancer invasion, including basal/mesenchymal marker genes. The loss of GATA3 results in the increased AP-1 activity and content, thus reducing enhancer-promoter contacts in luminal lineage-specific genes and facilitating phenotypic switch in cancer [193, 195]. Chromatin structural factors such as CTCF and topoisomerases (TOPases) are also engaged in ER-mediated enhancer-promoter contacts in breast cancer [196, 197]. Chromatin-remodeling complex NURF, as a coactivator of CTCF, can recruit CTCF to ER-related enhancers, thereby propelling enhancer transactivation and enhancer-promoter contact of GREB1 and TFF1 [42].

Discussion and future perspectives

The ER signaling pathway plays an essential role in the incidence and progression of breast cancer. ER is a master regulator of transcription, and its functions depend on its interacting proteins that are highly variable in breast cancer cells. The in vitro models and human tumor models demonstrate that the variability of the ER-interacting proteins is closely related to malignancy and therapeutic response. Endocrine therapy is the primary therapy for ER⁺ breast cancer, and reactivation of ER signaling can induce treatment resistance.

The higher-order assembly of TFs at enhancers of ER-target genes triggers genome-wide enhancer reprogramming, thus leading to transcriptional changes and phenotypic plasticity enhancement of breast cancer cells, eventually inducing treatment resistance [193, 198, 199]. ER-mediated enhancer reprogramming is regulated by pioneer factors (such as FOXA1) and TFs (such as CTCF). FOXA1 upregulation has been observed in endocrine resistant cell lines and ER⁺ metastatic breast cancer cells [7]. AP-1 can modulate FOXA1/ER-dependent enhancer activation by its accumulation in FOXA1-bound genomic regions in tamoxifen-resistant cells [200, 201]. Recent studies indicate that three-dimensional (3D) chromatin interactions, particularly those in CTCF-bound anchors, are frequently disrupted in endocrine-resistant tumors, thus impacting ER-dependent transcriptional activity [141, 202]. Therefore, identification of key proteins regulating ER-mediated enhancer activation, such as BRD4 will contribute to improving the efficacy of endocrine therapy. JQ-1, a selective inhibitor of BRD4, is used for breast cancer therapy in combination with endocrine therapy [203, 204]. PI3K and CDK4/6 proteins can modulate chromatin accessibility, especially in enhancer regions. PI3K inhibitors increase the enhancer accessibility in ER-FOXA1-PBX1 complex binding regions [205], while CDK4/6 inhibitors enhance the chromatin accessibility at AP-1 binding regions [206]. These findings reveal the dynamic alterations in the chromatin landscape of ER-mediated enhancers and the critical role of 3D genomic architecture in developing strategies to address endocrine drug resistance.

Considering the contribution of SHRs, including PR, GR and AR to ER-mediated enhancer reprogramming, SHRs may be utilized as the hallmark for accurate prediction of patient prognosis and therapy response. SHR-based cytomic and transcriptomic classification can assist in clinical decision for breast cancer therapy. Furthermore, the pro-oncogenic effects of ER can be weakened by pharmacologically targeting its interacting proteins, leading to better clinical outcomes. Currently, clinical trials are conducted to evaluate the efficacy of PR agonists in combination with aromatase inhibitors in ER⁺/AR⁺ breast cancer patients (clinical trials.gov: NCT03306472 and NCT02910050). Large-scale preclinical studies have indicated that the AR/ER ratio in breast tumors determines whether patient can effectively respond to AR-targeted therapies [110, 207]. The other factors affecting the TF interaction mode, such as the time of interaction, the proportion of proteins, chromatin binding sites, and complex structure remain to be further investigated. The recent advancements in sequencing technology have enabled large-scale comparison of the DNA, RNA, DNA methylation, chromatin accessibility, and 3D chromatin architecture in paired primary and metastatic tumors before and after treatments. It is suggested that single-cell RNA sequencing, ATAC sequencing, patient-derived organoid models should be employed for exploring intra-tumor heterogeneity and developing personalized treatment approaches in future studies. The identification of novel epigenetic regulators that drive therapy resistance will be conducive to the development of effective epigenetic-based therapies. Combining multiple therapy strategies such as agonists/antagonists of SHRs plus endocrine therapy is expected to improve the efficacy of ER⁺ breast cancer treatment.

Author contributions

Ge Sun completed the entire manuscript and Shaoya Wu and Yan Chen completed the revisions. The remaining authors and responsible authors completed the refinement of the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (32100440 for Li Li); Wuhan Knowledge Innovation Special Project ‘Dawning Program’(2023020201020352 for Li Li); Postdoctoral Fellowship Program of CPSF (GZC20230907 for Ge Sun)

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

Not applicable.

Competing interests

The authors declare no competing interests.