Allosteric Induction of Estrogen Receptor Ligand Binding Domain Tetramerization by a Distinct Complete Estrogen Receptor Antagonist

Abstract

Complete estrogen receptor antagonists (CERANs) are effective against advanced estrogen receptor-positive (ER+) breast cancers, but current chemical scaffolds limit our ability to explore the full range of ER pharmacology. We report the synthesis of OP-1690 (2), a CERAN featuring a distinct unconstrained core. Structural and biophysical studies reveal that 2 uniquely promotes estrogen receptor alpha (ERα) ligand binding domain (LBD) tetramer formation, which goes beyond the conventional homodimer. This study shows how new CERAN scaffolds can reveal unrecognized mechanisms of action.

Breast cancer (BC) accounts for over 30% of female cancers, and ERα drives disease progression in ∼75% of patients. Current endocrine therapies (ETs) target ERα either indirectly through aromatase inhibitors (AIs) or directly via selective estrogen receptor modulators (SERMs) or degraders (SERDs). However, many breast cancers will recur or exhibit de novo resistance, highlighting the need for continued improvements. Approximately 85% of resistant tumors retain ERα signaling, and next-generation antiestrogens show promising anticancer activities, but responses are nondurable for many patients. −

CERANs demonstrate superior clinical efficacy in refractory ER+ BC compared to standard-of-care ETs. They represent a subclass of SERDs, which antagonize ERα before inducing its proteasomal degradation. While many SERDs like elacestrant (RAD1901) show partial agonism at low doses or others like bazedoxifene display tissue-specific partial agonism, CERANs such as fulvestrant (ICI), palazestrant, , giredestrant, imlunestrant, amcenestrant, and camizestrant act as full antiestrogens across all hormone-responsive tissues. Next-generation CERANs offer improved pharmacokinetic profiles over ICI and suppress both ligand-dependent and ligand-independent ERα activation, mechanisms that contribute to standard-of-care ET-resistance. ,,−

Most CERANs comprise a fused-ring heterocyclic core that mimics estrogen and an extended arm that imposes antagonism through allosteric modulation of the ERα ligand binding domain (LBD). These conformational ensembles rewire coregulator specificity to reprogram transcriptional outputs toward cellular quiescence and induce receptor turnover. ,, Importantly, targeting noncanonical ERα conformations can elicit neomorphic and potentially therapeutically relevant receptor activities. , Therefore, expanding beyond traditional antiestrogen scaffolds will broaden the scope of possible CERAN mechanisms of action. In this study, we evaluated whether an unconventional unconstrained ERα-binding scaffold could be developed into a CERAN and whether it exerts unique influences on the receptor structural states that dictate anticancer activities.

We hypothesized that an unconstrained core could provide new opportunities for unique ERα-engagement modes through additional rotational degrees of freedom. In 2021, J. S. Disch et al. reported a compound comprised of a 2-chloro-3-hydroxybenzyl (steroidal A-ring mimic), a (1-(2-fluorophenyl)­cyclopentyl)­methyl (D-ring), and 4-phenoxyaniline central ring (B-ring) (1) that could be modified to antagonize ERα (Figure A). In the previous study, piperidine and proteolysis-targeting chimera moieties were used to engage antagonistic and degrading activities, respectively. In this study, we appended a fluoropropylazetidine group to the core to produce compound 2 (Figure A). This moiety is characteristic of clinical-stage CERANs, including palazestrant, , giredestrant, imlunestrant, and camizestrant (Figure S1). We first used the alkaline phosphatase (AP) assay to indirectly measure partial ERα agonism in endometrial cells. This assay measures the activity of AP, an ER responsive gene, as a readout of ER transcriptional activity, where SERMs act as weak estrogens while CERANs/SERDs act as antiestrogens. ,, Here, 2 exhibited a CERAN-like profile by inhibiting AP activity with pIC₅₀ = 7.5 in the presence of 500 pM 17β-estradiol (E2), and no stimulation was measured in its absence (Figure B,C). Next, a LanthaScreen, a time-resolved FRET assay that measures competitive binding to ERα, showed a favorable pIC₅₀ = 7.9 for compound 2 (Figure D).

1.

Structure and CERAN activities of compound 2. (A) Modifications to the scaffold occur at the A-ring (pink) and basic side chain (blue). A representative steroid, 17-β-estradiol, is shown with rings labeled in the top inset, and the parental compound 1 is shown in the bottom inset. (B) Alkaline phosphatase (AP) activity of Ishikawa cells in estrogen-depleted media following incubation for 72 h with 2 and 500 pmol/L E2. (C) Basal AP activity of Ishikawa cells in estrogen-depleted media following incubation for 72 h with 2. N/A = not applicable. (D) LanthaScreen competitive binding assay of 2 using wild-type ERα LBD.

The AP assay showed a CERAN profile for 2, but its potency was inferior to palazestrant and fulvestrant (pIC₅₀ = 8.5 for both). Therefore, optimization efforts were undertaken to improve the antagonistic potency of this scaffold, starting with the basic side arm (O-azetidine-alkyl (O-R³)) (Table , Figure S2). Removal of the fluorine in 3 was well tolerated and maintained the antagonist activity. To evaluate the effect of −F substitution in the D-ring, we prepared an analog of 3, lacking the −F substitution (R¹ = −H) (4). Compound 4 displayed potency comparable to that of 2, suggesting that the −F substitution in the D-ring is not crucial for the activity (Table ). Subsequent fluoropropyl­pyrrolidine analogs 5 and 6 showed reduced antiestrogenic potencies. Thus, propyl azetidine and 3-fluoropropylazetidine were privileged basic side arms in terms of antagonistic potencies within this series.

1. Exploration of the Side Chain Activities.

Given that SERDs elicit antiestrogenic transcriptional activities while also inducing ERα degradation, whereas SERMs attenuate receptor turnover in breast cancer cells, we next assessed the receptor degradation activities of compound 2. ,,, Western blot analysis confirmed that compound 2 induces ERα degradation, using the SERM, 4-hydroxytamoxifen (4OHT), the SERD, ICI, the weak SERD, RAD1901, and the agonist, E2, as comparator compounds (Figure A,B). Treatment with compound 2 reduced ERα protein levels in two ERα-positive breast cancer cell lines, with a 44% reduction in MCF7 cells and a 74% reduction in CAMA-1 cells compared to the vehicle control. In CAMA-1 cells, the ERα protein remaining after treatment with 2 (26%) was similar to that seen with ICI (16%). In contrast, compound 2 was less effective than ICI in MCF7 cells, leaving 56% and 26% ERα protein remaining, respectively. Compound 2 achieved greater ERα degradation compared to RAD1901, in MCF7 (56% vs 84% remaining) and CAMA-1 (26% vs 46% remaining) cell lines. As expected, 4OHT stabilized ERα, increasing protein levels by 63% in MCF7 cells and 39% in CAMA-1 cells. Compound 2 was more effective at degrading ERα than the agonist E2 (26% vs 54% remaining) in CAMA-1 cells, while in MCF7 cells, its activity was slightly lower than E2 (48% vs 44% remaining).

2.

Compound 2 induces ERα degradation. Representative simple Western blots and quantification of ERα protein levels with the percentage of ERα remaining after treatment listed in (A) MCF7 cells or (B) CAMA-1 cells after 24 h of treatment with 300 nM compound or 1 nM E2. ERα levels are normalized to those of untreated controls (dotted line). Data represent the mean across at least three independent replicates.

To evaluate whether compound 2’s ERα degradation translated to functional inhibition of ER+ breast cancer growth, proliferation assays were performed on MCF7 and CAMA-1 cell lines (Figure ). Compound 2 demonstrated a lower efficacy profile than comparator compounds in MCF7 (pIC₅₀ = 6, E _max = 47%) and CAMA-1 (pIC₅₀ = 5.9, E _max = 80%) cells.

3.

Compound 2 inhibits cellular proliferation with a reduced efficiency. Cellular proliferation assay of (A) MCF7 and (B) CAMA-1 cells treated over 7 days with the listed compounds in the presence of E2 (100 pM). Proliferation is assessed by CyQUANT reagent and normalized to the DMSO vehicle, the value of which is indicated by the horizontal dotted line.

Cellular proliferation was inhibited with reduced efficiency with 2, despite its favorable biochemical binding affinity (pIC₅₀ = 7.5), complete blockade of ER-induced AP activity in Ishikawa cells (pIC₅₀ = 7.6), and effective ERα degradation. To investigate the basis for the reduced antiproliferative efficacy, quantitative PCR (qPCR) was used to determine whether and to what extent 2 affected canonical ERα target gene expression in BC cells. Two different ER-positive BC cell lines, MCF7-WS8 and T47D, were serum-starved for 48 h, then treated with E2 alone or ICI, 4OHT, RAD1901, or 2 in the presence of E2 for 24 h alongside vehicle control (Figure ). ERα target genes included CDK1, CCND1, cMyc, CA12, E2F1, PR, and GREB1. Compound 2 had comparable effects to those of ICI and RAD1901 on the transcription of these target genes. Only CA12 showed a slightly significant increase in the MCF7-WS8 cells, but that was also observed for ICI and RAD1901. Therefore, the reduced antiproliferative efficacy of 2 does not appear to be related to its capacity to antagonize ERα transcriptional activities in breast cancer cells.

4.

Analysis of the impact of 2 on canonical ERα target gene transcription by qPCR. MCF7-WS8 and T47D were serum-starved for 48 h, then treated with 1 nM E2 or 1 nM E2 + 1 μM ICI, 4OHT, RAD1901, or 2 for 24 h alongside vehicle control before harvest and analysis. Representative data are the mean of three replicates ± sd, and error bars show sd, ΔΔCt = ΔCt (drug + E2) – ΔCt­(E2). Significance determined by unpaired t test where ns = not significant, *p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0005.

We further evaluated 2 using in vitro ADME assays and in vivo mouse intravenous and oral pharmacokinetic (PK) studies (Table ). Compound 2 has high mouse and human plasma protein binding (>99.93 and >99.98, respectively), moderate in vitro clearance in human and mouse hepatocytes (human CL_hep = 9.4 μL/min/million cells; mouse CL_hep = 10.1 μL/min/million cells). In vivo, PK evaluation of 2 in mice shows a high plasma clearance (CL_total = 55.9 mL/min/kg) and moderate volume of distribution (12.1 L/kg) with a half-life of 5.7 h. However, 2 exhibits low exposure (AUC_inf = 298 h·ng/mL) and oral bioavailability (F) of 0.7% (Table ), due to first pass clearance by the liver. The phenol group in most ERα ligands is known to be critical for potency but also a metabolic liability. , Together, these data point to potential pharmaceutical restrictions that may limit 2’s antiproliferative efficacy and in vivo anticancer activities.

2. In Vitro Hepatocyte Stability and in Vivo Pharmacokinetics of Compound 2 .

in vitro ADME			in vivo PK
species	PPB	CLhep	route	T max	C max	T 1/2	AUClast	AUCinf	Cl	V ss	F
	% bound	(μL/min/106 cells)		(h)	(ng/mL)	(h)	(h·ng/mL)	(h·ng/mL)	(mL/min/kg)	(L/kg)	(%)
human	>99.98	9.4	IV	0.08	406	5.6	291	298	55.9	12.1
mouse	>99.93	10.1	PO	0.25	4.95	2.8	9.19	10.6			0.7

^aCompound 2 was formulated in 1 M HCl:DMSO:5% HPbCD (1:100:3900 v/v) and 2.5% DMSO in 0.5% CMC-Na for intravenous (IV) injection and oral (PO) gavage, respectively. 2 was administered to female Balb/c mice through IV at 1 mg/kg (5 mL/kg of 0.2 mg/mL dose formulation) and through PO at 5 mg/kg (10 mL/kg of 0.5 mg/mL).

To address the measured pharmaceutical restrictions, a variety of A-ring modifications were explored to identify efficacious ERα antagonists with improved ADME and oral PK properties (Table , Figure S3). The des-hydroxy modification in compound 7 is tolerated in the ERα competitive binding assay. However, it leads to loss of antagonist activity, suggesting that the phenol group is necessary for anticancer efficacy. The chlorophenol A-ring of 2 was the most active antiestrogenic compound, and removal of the chloro-substituent (8) leads to a loss of antagonism and ERα binding activity. Replacing the chloro-substituent with a methyl group (9) was well tolerated, suggesting that substitution at the ortho position is not important for activity. Compound 10 was designed to sterically protect the hydroxy group from glucuronidation with a substitution at the position adjacent to the hydroxy group, but it lost considerable activity. Several phenol isostere derivatives were investigated to circumvent the high clearance and low bioavailability issues of 2 (Table ). The single-ring phenol isostere boronic acid analog (11) lost substantial antagonistic activities. Additionally, the high lipophilicity of compound 2 (Chemaxon cLogD = 7.19) was a concern, as high LogD correlates with lower solubility. To reduce the LogD, the A-ring was replaced with heterocyclic phenol isosteres: indazole (12) (chemaxon calcd LogD = 6.36), and carboxylic acid (13) (chemaxon calcd LogD = 4.08). However, both indazole and carboxylic acid analogs lacked ERα antagonistic activity, with 12 showing partial agonism in the absence of E2 (agonist mode) in the AP assay. Neither modification was productive, and compound 2 remained the best CERAN discovered in this series.

3. A-Ring Modifications of Compound 2 .

^aThe asterisk * indicates that log­(inhibitor)–response curves were fit using a variable-slope (four-parameter) model by least-squares regression; for at least one parameter, confidence intervals could not be calculated, and best-fit values should be interpreted with caution.

To determine the molecular basis of 2’s antiestrogenic and ERα-degrading activities, we used X-ray crystallography to understand how it impacted the structure of the ligand binding domain (LBD). An X-ray cocrystal structure of 2 in complex with ERα LBD was solved to 2.1 Å. Unexpectedly, the asymmetric unit is composed of an ERα LBD tetramer (Figure A). Here, tetramerization is facilitated by an allosteric structural interaction via the c-terminal helix 12 (H12), the main allosteric sensor of the LBD. Specifically, two H12s form symmetric contacts with grooves between H4 and H10 (Figure B). Hydrogen bonds are observed between the carboxylic acid side chain of D545, the imidazole side chain of H547, and the backbone amide of R548 of H12 with the backbone amide of L372, the carboxylic acid side chain of E470, and the backbone carbonyl of D369 in the other monomer, respectively (Figure B). The core of 2 participates in a hydrogen bond network with E353, R394, and a water molecule within the hormone binding pocket, while the cyclopentyl group dips into a hydrophobic pocket near H8, and the fluorobenzyl group sits perpendicular to H11 (Figure C). The interaction with E353/R394 likely explains the loss in potency for the des-hydroxyl derivative. The fluoropropylazetidine-containing side arm forms a hydrogen bond with D351 and extends near P535 within the H11–12 loop (Figure C). In three of the four monomers, the H11–12 loop is disordered, a common observation in SERD-bound X-ray cocrystal structures ,,

5.

Structural basis of 2’s CERAN activities. (A) Overview of ERα LBD-compound 2 X-ray cocrystal structure. Helix 12 (H12) for each monomer is highlighted in yellow. Compound 2 is shown as sticks, with the 2mF _o – DF _c difference density map shown as a blue mesh, contoured to 1σ. (B) H12 interactions that mediate tetramer formation. (C) Compound 2 interactions within the hormone binding pocket and near H12.

Interestingly, there are other examples of ERα LBD tetramers with clinically relevant SERMs and SERDs in the Protein Data Bank (PDB). Giredestrant (GDC-9545) shows a H12-mediated tetramer within the crystallographic asymmetric unit. However, tetramerization appears to be mediated by H12 of diagonal monomers rather than side-by-side monomers 2 and 4, like 2 (Figure S4A,B). Other structures, like the 4OHT complex, show two sets of dimers that are too far apart to interact or show packing not mediated by H12 (Figure S4C). Such tetramers are likely crystallographic artifacts rather than bona fide ligand-induced tetramers due to their lack of H12 involvement. Together, these structures suggest that SERDs can induce H12-mediated ERα LBD tetramerization with distinct ligand-specific symmetries.

Analytical size-exclusion chromatography (SEC) was used to test whether ligand-induced ERα ligand-binding domain (LBD) tetramers persist in solution. SEC separates proteins based on their hydrodynamic radius, allowing an estimation of molecular weight from retention times relative to protein standards. All ligand conditions produced a predominant peak with estimated molecular weights of 72.3 ± 28.2 kDa (2), 88.8 ± 0.6 kDa (ICI), 87.9 ± 34.4 kDa (4OHT), and 88.4 ± 36.0 kDa (apo) (Figure S5). The expected dimer mass of the ERα LBD is ∼56 kDa, which falls within one standard deviation for all conditions. Notably, an earlier-eluting peak consistent with a tetramer of 173.4 ± 33 kDa was measured in two out of three replicates for the ERα LBD–compound 2 complex, suggesting that this tetramer can be formed in solution.

Mass photometry (MP) was used as an orthogonal approach to measure changes in ERα LBD oligomerization in solution. This method measures the mass of individual unlabeled proteins based on light interference upon surface landing. MP revealed distinct ligand-dependent oligomerization profiles (Figure A–C). Compound 2 induced a larger oligomeric population, with up to 10% of the binding events corresponding to species consistent with a tetramer. ICI- and 4OHT-bound ERα LBD showed minor populations consistent with low-abundance oligomers (<3% of binding events), supported by subtle inflections in the SEC traces (Figure S5A). While compounds 2 and 4OHT contain unrestrained cores, ICI has a steroidal core with an extended fluorinated alkyl side chain. In the ligand-binding pocket (LBP), all three ligands adopt similar binding poses with their cores binding within the orthosteric hormone binding pocket and side arms influencing H12 conformational dynamics (Figure S6). Compound 2 is similar to GDC-9545, imlunestrant, and camizestrant, which also contain fluoropropylazetidine side arms. However, only GDC-9545 has a published crystal structure, which likewise exhibits a tetramer in the asymmetric unit (Figure S4). Because tetramer formation is ligand-specific and the ligand does not directly contact the dimer–dimer interface, this behavior suggests an allosteric mechanism of higher-ordered oligomerization. Together, these findings indicate that certain ligands, particularly compound 2, promote higher-order oligomerization of the ERα LBD in solution.

6.

Compound 2 increases the proportion of ERα LBD tetramers in solution compared with ICI and 4OHT. (A–C) Mass photometry spectra of 15 nM ERα LBD in the presence of saturating SERM or SERDs (mass calibration error: 2.4%). (A) ERα LBD bound to compound 2 with 4 replicates shown each displaying distinct dimer and tetramer peaks. (B) ERα LBD bound to ICI. Among three replicates, only one exhibited a detectable tetramer peak. (C) ERα LBD bound to 4OHT. Among three replicates, only one showed a detectable tetramer peak. (D) Thermal shift assay of ERα with saturating ligand. Data are the mean of three independent replicates, and significance is determined by unpaired t test: ns = not significant, ****p < 0.00005.

Next, a thermal shift assay was used to measure how 2 affected the overall stability of ERα LBD compared to E2, 4OHT, and ICI. Ligand binding and other protein–protein interactions can provide favorable stabilizing energies that manifest as an increased protein melting temperature (T _M). Compound 2 showed a significantly increased T _M = 61.7 ± 0.6 °C compared to that of unliganded LBD (T _M = 52.2 ± 0.2 °C). E2, 4OHT, and ICI also increased the melting temperatures at T _M = 61.8 ± 0.5, 66.7 ± 0.4, and 62.8 ± 0.4 °C respectively (Figure D). Therefore, while 2 can favor the formation of higher ordered ERα LBD oligomers, it does not appear to appreciably impact the overall thermal stability of the protein compared with other ERα ligands.

Together, our data show that CERAN, 2, is a structurally unconventional antiestrogen that impacts ERα LBD oligomerization. It does not induce AP activity in Ishikawa cells, has ERα degrading capabilities comparable to ICI, and can downregulate ERα target genes. Despite these favorable activities, it shows reduced potency and efficacy in inhibiting cellular proliferation relative to other SERMs and SERDs. Moreover, its high plasma binding and low oral bioavailability point to pharmaceutical liabilities that would limit its clinical utility. Structural and biophysical analyses suggest that 2 increases the propensity of ERα LBD to form oligomers larger than a canonical homodimer in solution. The glucocorticoid receptor (GR), a homologous steroid receptor, forms higher-order oligomers that result in phase-separated condensates within the nucleus that modulate gene transcription. , While GR condensates colocalize with RNA polymerase II in transcriptional initiation and elongation complexes, the overall function of steroid hormone receptor condensates in cells requires further study. In this case, our data suggest that ERα ligands can engage discrete LBD structural states to favor higher-ordered oligomers even in the absence of DNA. We would expect tetramers to form upon ligand activation after displacement from heat shock proteins, and tetramers may function through both direct and indirect chromatin interactions. ERα forms biomolecular condensates in vivo, and condensate formation is required for superenhancer function and gene activation. , If compound 2-induced tetramerization impacts ERα condensate partitioning, it would represent a valuable chemical tool to understand the functional role of these condensates. While the anticancer activities of 2 were limited, this study has revealed a new and exciting structure-based mechanism to modulate ERα activities that should be studied further.

Glossary

Abbreviations

BC

breast cancer

ER+

estrogen receptor-positive

SERM

selective estrogen receptor modulator

SERD

selective estrogen receptor degrader

H12

helix 12

AF-2

activating function 2

RAD1901

elacestrant

CERAN

complete estrogen receptor antagonist

DEL

DNA-encoded library

4OHT

4-hydroxytamoxifen

ICI

fulvestrant

E2

17β-estradiol

qPCR

quantitative polymerase chain reaction

DMSO

dimethyl sulfoxide

PK

pharmacokinetic

AUC_inf

area under the curve from the time of dosing extrapolated to infinity

CL_total

total clearance

CL_hep

hepatic clearance

IV

intravenous

PO

oral

HPbCD

hydroxypropyl-β-cyclodextrin

CMC-Na

sodium carboxymethyl cellulose

LogD

distribution constant

LBD

ligand binding domain

PDB

Protein Data Bank

GDC-9545

giredestrant

laso

lasofoxifene

MP

mass photometry

T _M

melting temperature

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.5c00667.

• Detailed descriptions of the synthesis of each compound, methods section, reaction schemes, HPLC traces, and crystallographic information (PDF)

‡.

E.C.F. and R.C. contributed equally. S.W.F. and B.R.H. conceived of the study. E.C.F. performed protein expression and purification, thermal shift assay, size exclusion chromatography, and mass photometry. R.C. and K.S.W. performed synthetic chemistry. E.C.F. and S.W.F. performed the X-ray crystallography studies. G.R.H. performed qPCR. G.P., S.B., and B.W.R. performed mammalian cell culture and pharmacokinetics experiments. E.C.F. and R.C. wrote the manuscript with edits and intellectual contributions from S.W.F., R.C., and D.C.M. All authors read and approved the manuscript.

Safety Statement. No unexpected or unusually high safety hazards were encountered.

The authors declare the following competing financial interest(s): Brian R. Hearn, Susanna Barratt, Reena Chawla, Brandon W. Robello, Guadalupe Pena, Kylie S. Wells, and David C. Myles are employees of Olema Oncology. Sean W. Fanning received sponsored research funds from Olema Oncology.