ESR1 fusions and therapeutic resistance in metastatic breast cancer

Breast cancer is the most frequent female malignant tumor, and the leading cause of cancer death in women worldwide. The most common subtype of breast cancer is hormone receptor positive that expresses the estrogen receptor (ER). Targeting ER with endocrine therapy (ET) is the current standard of care for ER positive (ER+) breast cancer, reducing mortality by up to 40% in early- stage disease. However, resistance to ET represents a major clinical challenge for ER+ breast cancer patients leading to disease recurrence or progression of metastatic disease. Salient drivers of ET resistance are missense mutations in the ER gene (ESR1) leading to constitutive transcriptional activity and reduced ET sensitivity. These mutations are particularly prominent and deleterious in metastatic breast cancer (MBC). In addition to activating ESR1 point mutations, emerging evidence imposes that chromosomal translocation involving the ESR1 gene can also drive ET resistance through the formation of chimeric transcription factors with constitutive transcriptional activity. Although these ESR1 gene fusions are relatively rare, they are enriched in ET resistant metastatic disease. This review discusses the characteristics of ER fusion proteins and their association with clinical outcomes in more aggressive and metastatic breast cancer. The structure and classification of ER fusion proteins based on function and clinical significance are also addressed. Finally, this review summarizes the metastatic phenotypes exhibited by the ER fusion proteins and their role in intrinsic ET resistance.


ESR1 fusions are acquired and enriched in MBC
RNA-seq analysis conducted by Veeraraghavan and colleagues on 990 primary TCGA breast samples identified the first ESR1 gene fusion, ESR1-e2>CCDC170 (Table 1), in a subset (2.1%) of Luminal B breast tumor samples (75,82). This fusion is formed by tandem-duplication, it retains the first two non-coding exons of ESR1 (ESR1-e2) connected to various sequences from the coiled-coil domain containing 170 (CCDC170) gene ( Figure 1). The promoter trap drives aberrant expression of CCDC170 and produces N-terminal truncated forms of the CCDC170 protein (DCCDC170) (75). The authors also provided functional evidence

Structure and function of ESR1-e6>fusion proteins in MBC
The ESR1-e2>CCDC170 fusion protein consists of the 5′ untranslated region of ESR1 to the coding region of CCDC170, generating N-terminally truncated CCDC170 proteins (DCCDC170) expressed under the ESR1 promoter ( Figure 1) (75, 82). Structural studies have revealed that this structure is distinct from the ESR1-e6>fusions identified from ET resistant MBC. Despite the diversity among the ESR1-e6>fusions, they share a common structure whereby the first 6 exons of ESR1 (ESR1-e6) are preserved, retaining the hormone-independent transactivation domain (TAD) as well as the DNA-binding domain of ER whereas the LBD is lost and replaced with a functional domain of the 3' fusion partner ( Figure 1) (76, 81, 84). This structure is strongly associated with estrogen independent growth and ET resistant metastatic ER+ breast tumors. The loss of a functional LBD suggests a clear pathological impact, leading to complete resistance to the activity of current ER antagonists, which all bind to the LBD. As expected, several ESR1-e6>fusions (ESR1-e6>YAP1, ESR1-e6>PCDH11X, ESR1-e6>SOX9 and ESR1-e6>ARNT2-e18) remained stably expressed in the presence of fulvestrant and promoted ET-resistant growth of T-47D and MCF7 cells (80,81). In contrast, the expression of ER mutant constructs that lack the LBD had decreased transcriptional activity, suggesting that the presence of the 3' partner is essential for the ER fusion activity (80, 81, 84). The fact that multiple different 3' partners have the same effect and drive ET resistance and malignant phenotypes, indicates that the enhanced activity of the ER fusions is not dependent on a specific 3' partners. These findings suggest that the 3' partner may be important for the stability of ER and possibly the dimerization of ER, however, an intact LBD is not required for ER activity in the context of the ER fusions.

Activating ESR1 fusion proteins drive endocrine resistance and metastatic phenotypes
When first described, ESR1-e2>CCDC170 in ER+ breast cancer cells led to enhanced growth and reduced sensitivity to tamoxifen (75) suggesting a role for ESR1-e2>CCDC170 in ET resistance. Additional pre-clinical studies (74,75,79) showed that the expression of ESR1-e2>CCDC170 fusions in ERpositive breast cancer cells resulted in increased cell migration, increased colony formation, and increased cell proliferation as evidenced by the increase in the number of cells in S-G2/M phase. Li and colleagues (74) provided detailed evidence supporting the function of ESR1-e2>CCDC170 in promoting breast cancer cell survival and endocrine resistance both in vitro and in xenograft models. Their mechanistic study suggests that ESR1-e2>CCDC170 fusions bind and stabilize the HER2/HER3/ SRC complex and enhance the activation of SRC/PI3K/AKT signaling during ET in vitro and in vivo. This study also suggested a potential strategy to manage ESR1-e2>CCDC170 positive patients by combining the HER2 inhibitor lapatinib and/or SRC inhibitor dasatinib with ET.
A series of publications clearly demonstrated that ESR1-e6>fusions were identified from ER+ MBC patients and most ESR1-e6>fusion proteins are drivers of ET resistance (69, 76,80,81,84). Functional properties of these ESR1-e6>fusions include estradiol-independent growth and constitutive expression of ER target genes leading to ET-resistant proliferation and epithelialmesenchymal transition (EMT) genes facilitating metastasis.

ESR1 fusions as potential biomarkers and novel therapeutic vulnerabilities in breast cancer
Next generation sequencing (NGS) methods such as DNAbased comprehensive whole genome (WGS) or RNA-based transcriptome (WTS) sequencing have been extensively used to describe gene fusions in multiple cancer types (86-90). WGS detects gene fusions based on hybrid-capture methods and still considered the most unbiased approach to identify fusion events, especially in large gene panels. WGS is highly sensitive and can be used on fresh, snap frozen and formalin fixed specimens. Still and all, WGS does not indicate the expression of the gene fusions, and the detection of fusion variants involving large DNA intronic regions is poor (86, 91). WTS have the overall advantage of detecting transcriptionally expressed gene fusions and the sequencing is not affected by intronic regions. Moreover, WTS does not require a priori knowledge of gene fusion partners, can distinguish splicing isoforms, quantify fusion transcripts, and it requires low input material. WTS approaches can be based on hybrid-capture or amplicon-based methods using classical or anchor multiplex PCR (86, 92-94).
Matissek and colleagues reported anchored multiplex PCR (AMP) as an effective approach to identify gene fusions in cancer, including ER+ metastatic breast cancer (71). AMP was also validated in their study and applied to cohorts of (1) 110 of early-stage and advanced ER+ breast cancer patients (Clinical Genotyping Cohort) and (2) 63 of advanced ER+ breast cancer patients with matched primary and metastatic samples (Matched Primary/Metastasis Cohort). 14 patients in the Clinical Genotyping Cohort harbored intergenic exon-exon fusions, including the in-frame fusion of ESR1 to CCDC170. The authors emphasized that the identified fusion junction sequences involved at least one precise exon boundary. 10 of 63 patients in the Matched Primary/Metastasis Cohort harbored gene fusions in either the primary or metastatic samples. Collectively, AMP detected fusions in 24 of 173 breast cancer patients (14%) in this study, including 11 primary tumors. Combined with complementary "break-apart" fluorescence in situ hybridization (FISH) analysis, they further validated AMP as detection technique for clinically relevant fusions. As example, break-apart of ESR1 was present in primary tumors and metastases from a patient whose tumors demonstrated the ESR1-e4>COA5 fusion upon AMP analysis. Additionally, the ESR1-e2>CCDC170 fusion detected by AMP was also confirmed upon FISH analysis for ESR1. A disadvantage of WTS is that it only identifies expressed fusion genes and not adequate for gene fusion analysis at DNA level. Moreover, biological material is often short, resulting in poor quality RNA and false positive sequencing results. For multiplex PCR approach, the primer design and PCR bias like allele dropout can also impact analysis result (86, 92-94).
ESR1-e6>fusion genes have been detected by NGS methods ( Table 1), techniques that are not yet employed routinely in the clinic. Analysis of plasma circulating tumor DNA to detect ESR1 point mutations by droplet digital PCR (ddPCR) is now done in the clinic and clinical trials are investigating the use of these assays in real-time (95-97). Obtaining circulating DNA from liquid biopsies conserves the genomic landscape of the tumor suggesting that this less invasive detection methods may efficiently identify ESR1 fusions, particularly in metastatic ET resistant ER+ breast cancer. Indeed, Hartmaier and colleagues (84) used this approach to provide additional evidence of ESR1 fusion recurrence following extensive ER-targeted endocrine therapies. They obtained target capture sequencing data and examined a cohort of 9542 solid breast tumors and a cohort of 254 ctDNA samples from patients with advanced breast cancer. They successfully identified the ESR1-e6>SOX9, ESR1-e7>MTHFD1L, ESR1-e6>PLEKHG1, and ESR1-e6>TFG fusions (Table 1) in four solid tumors and the ESR1-e6>NKAIN2, ESR1-e6>AKAP12, and ESR1-e7>CDK13 (Table 1) fusions in 3 ctDNA samples. While there are several commercial platforms available and the cost of ctDNA assay is acceptable in the clinic, the concentration of ctDNA in plasma correlates with tumor size and stage, thus this assay is likely to be useful for late-stage breast cancer patients only (98).
As aforementioned, several ESR1-e6>fusion proteins are inactive and therefore not clinically actionable. As a potential efficient approach for screening samples for the presence of ESR1 fusions that drive ET failure in MBC, Gou and colleagues (81) developed a 24-gene expression signature that is specific for the presence of transcriptionally active ESR1 fusion proteins. Specifically, they identified 24 Hallmark genes, including 19 genes in the estrogen response gene set (CHST8, MAPT Gou and colleagues further confirmed the overlap in the transcriptional properties of ESR1-e6>fusions and ESR1 LBD point mutants in several PDX models and MBC cases. ETresistant PDXs harboring LBD point mutations (e.g. BCM15100, WHIM20, WHIM40, and HCI013 for ESR1-Y537S; WHIM37 and WHIM43 for ESR1-D538G) highly expressed the MOTERA signature, similar to the PDX naturally expressing the ESR1-e6>YAP1 (WHIM18) fusion. The expression levels of the MOTERA genes were not affected by E2 supplementation in the ESR1-e6>YAP1 expressing PDX or PDXs harboring the LBD point mutations. Furthermore, the MOTERA scores of PDXs expressing WT ESR1 was significantly lower than those of expressing the LBD mutations or the ESR1-e6>YAP1 fusion. Similar to the PDX models, MOTERA gene expression was significantly elevated in MBC tumors harboring the Y537S and D538G point mutations or the ESR1-e6>ARNT2-e18 fusion, and the signature score distinguished the LBD point mutations and the ESR1-e6>ARNT2-e18 fusion from WT ESR1.
In addition to the mechanistic studies, evaluating the potential of targeting these fusion proteins for the development of new targeted therapies is critical. Due to the formation and unique structure of ESR1-e6>fusions, all known ET options that target the LBD are ineffective. Lei and colleagues (80) targeted ER signaling regulated by ESR1 fusions by using Palbociclib, a CDK4/6 inhibitor for MBC. ESR1-e6>YAP1 and ESR1-e6>PCDH11X induced cell proliferation was sensitive to a CDK4/6 inhibition, and a PDX naturally expressing the ESR1-e6>YAP1 fusion was also responsive to Palbociclib treatment. Since ESR1 fusion driven growth of ER-positive breast cancers remained sensitive to CDK4/ 6 inhibition, the presence of an ESR1-e6>fusion could be a putative biomarker to stratify patients for CDK4/6 inhibitor therapy after resistance to endocrine treatment or continued CDK4/6 inhibitor therapy with a second targeted therapy after resistance to first line treatment for metastatic disease with endocrine therapy in combination with a CDK4/6 inhibitor. It is also hypothesized that CDK4/6 inhibition could be beneficial for patients with ESR1 fusions. Further mechanistic and preclinical studies are expected to introduce additional ESR1 fusions sensitive to CDK4/6 inhibitor therapy and expand on the data from patients harboring ESR1 fusions to examine their CDK4/6 inhibitor responses.
To further explore therapeutic strategies that target ESR1 fusions, Gates and colleagues (99) showed that pharmacological inhibition of ESR1-e6>YAP1 fusion with the proteosome inhibitor MG132, blocked ESR1-e6>YAP1 mediated activation of ER target genes. In the same study, bortezomib, a specific 26S proteasome inhibitor, also suppressed growth driven by the ESR1-e6>YAP1 fusion. Bortezomib was tested in a phase II clinical trial in postmenopausal women with ER+ MBC who had progressive disease after prior aromatase inhibitor therapy. The patients were randomized to fulvestrant and bortezomib versus fulvestrant alone groups. In this study, there was no significant difference in progression free survival, which was the primary end point. However, the combination was overall well tolerated and may have enhanced activity in patients who have an ESR1 fusion (100).

Conclusion and future directions
ET resistance in ER+ breast cancer patients remain a significant clinical problem. The ESR1 fusion proteins are emerging as a mechanism of ET resistance and the studies discussed in this review, deepened our understanding of the prevalence of the ESR1 fusion proteins and the mechanisms by which they drive resistance. The most prevalent and clinically significant ESR1 fusions can be divided into the ESR1-e2>CCDC170 and ESR1-e6>fusion genes. DCCDC170, identified in Luminal B breast cancer and generated by ESR1-e2>CCDC170 led to enhanced growth and reduced sensitivity to ET in MBC. ESR1-e6>fusions were identified in ET-resistant MBC and are formed by inter-chromosomal translocation fusing ESR1 exons 1 to 6 into a 3' fusion partner, replacing the LBD of ESR1. ESR1-e6>fusions drive estradiol-independent growth and constitutive expression of ER target genes leading to ET-resistance. ESR1 fusions were identified in more aggressive forms of breast cancer (ET resistant MBC and Luminal B breast cancer) and can guide the diagnosis and the development of therapeutic strategies to treat a subset of patients with tumors that harbor these ESR1 alterations. As for the ESR1-e6>fusion genes, only a handful of functionally active ESR1 fusion proteins have been studied to date and therefore ESR1 fusion events remain an understudied form of somatic mutation in breast cancer. The incidence of these ESR1 fusions is still not well understood, but the studies discussed here collectively suggest that the frequency of ESR1 fusions may be higher in heavily pre-treated metastatic samples and when using more sensitive detection techniques.
The discovery of the ESR1 LBD point mutations has sparked enthusiasm for the development of a new generation of compounds that not only combat existing ER mutants but also inhibit secondary mutations in ER. Indeed, novel oral SERDs and SERCAs are being developed, and likely to be approved in the clinic. Similar to the emergence of ESR1 LBD mutations that render ligand independent activity, it is likely that the adoption of more potent SERDs and SERCAs will lead to adaptive mechanisms of resistance that are either ER independent or ER dependent but independent of the LBD. It is currently unclear which mechanisms of resistance may emerge following novel SERD and SERCA treatments.
Although, ESR1 gene fusions are rare, the frequency of these fusions may increase under the selective pressure of more effective SERDs and SERCAs. Therefore, better understanding of the mechanism of action of these fusions that lack the LBD, yet drive tumor progression in ER+ MBC, will be critical for the identification of vulnerabilities to target these fusions.

Author contributions
All authors listed have made a substantial contribution to the work and approved it for publication.

Conflict of interest
RJ received research funding from Pfizer and Lilly and is on an advisory board for GE Healthcare.
The remaining author declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.