ESR1 F404 Mutations and Acquired Resistance to Fulvestrant in ESR1-Mutant Breast Cancer

Abstract

Fulvestrant is used to treat patients with hormone receptor-positive advanced breast cancer, but acquired resistance is poorly understood. PlasmaMATCH Cohort A (NCT03182634) investigated the activity of fulvestrant in patients with activating ESR1 mutations in circulating tumor DNA (ctDNA). Baseline ESR1 mutations Y537S are associated with poor outcomes and Y537C with good outcomes. Sequencing of baseline and EOT ctDNA samples (n = 69) revealed 3/69 (4%) patients acquired novel ESR1 F404 mutations (F404L, F404I, and F404V), in cis with activating mutations. In silico modeling revealed that ESR1 F404 contributes to fulvestrant binding to estrogen receptor-alpha (ERα) through a pi-stacking bond, with mutations disrupting this bond. In vitro analysis demonstrated that single F404L, E380Q, and D538G models were less sensitive to fulvestrant, whereas compound mutations D538G + F404L and E380Q + F404L were resistant. Several oral ERα degraders were active against compound mutant models. We have identified a resistance mechanism specific to fulvestrant that can be targeted by treatments in clinical development.

Significance: Novel F404 ESR1 mutations may be acquired to cause overt resistance to fulvestrant when combined with preexisting activating ESR1 mutations. Novel combinations of mutations in the ER ligand binding domain may cause drug-specific resistance, emphasizing the potential of similar drug-specific mutations to impact the efficacy of oral ER degraders in development. This article is featured in Selected Articles from This Issue, p. 201.

Figure 1.

Baseline ESR1 mutations and fulvestrant efficacy. A, % Incidence of mutations in indicated genes at baseline in Cohort A (n = 79 assessable patients). B, Incidence of baseline ESR1 alterations within Cohort A (n = 79 assessable patients). C, PFS of patients in Cohort A, divided by baseline ESR1 Y537C mutation status (left) and ESR1 Y537S mutation status (right). P values from the log-rank test. HR >1 denotes worse PFS for that group. WT, wild-type; mt, mutant. D, MCF7 cells were cotransfected with the indicated ESR1 expression constructs and treated with the indicated concentration of fulvestrant in the presence of 1 nmol/L estradiol for 24 hours and estrogen response element-luciferase reporter activity determined. Two independent experiments.

Figure 2.

Acquired mutations on fulvestrant. A, Incidence of acquired alterations (n = 69 assessable patients), colored by targetability of the alterations (Methods). Level 2B denotes the highest level of supporting evidence (“Standard care biomarker recommended by the National Comprehensive Cancer Network or other professional advice guidelines predictive of response to an FDA-approved drug”), whereas level 4 is the lowest (“Compelling biochemical evidence supports the biomarker as being predictive of response to a drug”). B, Incidence of acquired ESR1 mutations (n = 14 patients) and resultant amino acid changes. C,ESR1 F404 locus in the DNA-binding domain of the estrogen receptor. The number of base changes identified within the data set that result in the three different missense mutations are illustrated using https://proteinpaint.stjude.org/ (36). D,cis/trans analysis of F404 and E380Q in the three patients with assessable targeted sequencing data. Both alleles of chromosome 6 are represented, with annotated locations of the F404 and E380Q on each respective allele representing the cis/trans relationship of the variants. E, Mutations at phenylalanine 404 result in the substitution of amino acid residues without an aromatic ring. F,In silico modeling predicts the aromatic ring of F404 contributes to a pi-stacking bond between the receptor and both estrogen and fulvestrant.

Figure 3.

F404 does not activate estrogen signaling. A, CRISPR clones of MCF7 cells expressing ESR1 F404L (1210T>C, CRISPR edit indicated by red arrows) or D538G (1613A>G; CRISPR edit indicated by black arrows) were identified by RT-PCR followed by Sanger sequencing (left-hand panels). Similarly, a second round of CRISPR was used to introduce ESR1 F404L (1210T>C) into a clone (D6C) that expressed D538G (1613A>G; right-hand panels). B, Estrogen-dependent growth was assessed in colony formation assay. Parental MCF7 cells and indicated ESR1-mutant models were grown in either the absence or presence of estradiol (1 nmol/L) for 14 days. C, Quantification of colony formation assays of ESR1-mutant models treated with and without estradiol (1 nmol/L). Sulforhodamine B (SRB)-stained colonies were dissolved, and absorbance at 565 nm was measured. Mean with SEM, n = 3 independent experiments, nonparametric one-way ANOVA with Dunn multiple comparisons test; **, P < 0.01. D, Expression of estrogen target genes, progesterone receptor (PgR), and trefoil factor-1 (TFF1), assessed by western blot in parental MCF7 cells and indicated ESR1-mutant models grown in either the absence or presence of estradiol (1 nmol/L) for 24 hours. E, MCF7 cells were transfected with ESR1 expression constructs with indicated ESR1 variants. Expression of ERα was determined by western blot. F, MCF7 cells were cotransfected with the indicated ESR1 expression constructs ERE-luciferase reporter and control construct. Cells were treated in either the absence or presence of estradiol (1 nmol/L) for 24 hours, and ERE-luciferase activity was assessed. Two-way repeated-measures ANOVA with Dunnett multiple comparisons test, n = 4 mean with SD; *, P < 0.05. WT, wild-type.

Figure 4.

Compound F404L mutations induce resistance to fulvestrant. A, Compound mutations of D538G-F404L in MCF7 cells, along with single mutations and wild-type (WT), with sensitivity to fulvestrant assessed after 6 days treatment with Cell-Titer Glo viability assay. n = 4 mean with SD. B, Representative images of clonongenic assays grown in indicated concentrations of fulvestrant for 14 days. C, Quantification of colony formation assays for ESR1-mutant models treated with the indicated concentrations of fulvestrant for 14 days. EC₅₀ and IC₅₀ values were calculated from the response curves. SRB-stained colonies were dissolved, and absorbance at 565 nm was measured. Mean with SEM, n = 3 independent experiments. D, Expression of estrogen target genes, progesterone receptor (PgR), and trefoil factor-1 (TFF1) was assessed by western blot in parental MCF7 cells and indicated ESR1-mutant models grown in the presence of 1 nmol/L estradiol or 1 μmol/L fulvestrant. E, MCF7 cells were cotransfected with the indicated ESR1 expression constructs ERE-luciferase reporter and control construct. Cells were treated with 1 nmol/L estradiol either in the absence or presence of fulvestrant (1 μmol/L) for 24 hours, and ERE-luciferase activity was assessed. Two-way repeated-measures ANOVA with Sidak multiple comparisons test, n = 4 mean with SD; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 5.

Transcriptomic analysis of ESR1-mutant models. A, Gene set enrichment analysis (GSEA) for D538G + F404L models compared with D538G D6C cells maintained in 1 nmol/L estradiol. Pathways are highlighted red; false discovery rate-adjusted q value <0.05. B, GSEA for D538G + F404L models compared with D538G D6C cells treated with 1 μmol/L fulvestrant for 24 hours. Pathways are highlighted red; false discovery rate-adjusted q value <0.05. C, GSEA for ESR1-mutant models. Normalized enrichment score (NES) is shown for the indicated pathways. *, False discovery rate-adjusted q value <0.05. D, Heat map of “Estrogen response late” genes (log₂ expression) for ESR1-mutant models maintained in 1 nmol/L estradiol. E, Heat map of “Estrogen response late” genes (log₂ expression) for ESR1-mutant models treated with 1 μmol/L fulvestrant in the presence of 1 nmol/L estradiol.

Figure 6.

Compound F404 mutations are sensitive to novel SERDs. A–D, Compound mutations of D538G-F404L in MCF7 cells, along with single mutations and wild-type, with sensitivity to elacestrant (A), camizestrant (B), 4OH tamoxifen (C), and giredestrant (D), assessed after 6 days treatment with CellTiter Glo viability assay. n = 4 mean with SD. E, Representative clonongenic assays grown in indicated concentrations of elacestrant, camizestrant, 4OH tamoxifen, and giredestrant for 14 days. F, MCF7 cells were cotransfected with the indicated ESR1 expression constructs ERE-luciferase reporter and control construct. Cells were treated with indicated concentrations of fulvestrant, elacestrant, camizestrant, 4OH tamoxifen, and giredestrant in the presence of 1 nmol/L estradiol for 24 hours, and ERE-luciferase activity was assessed. Two-way repeated-measures ANOVA with Sidak multiple comparisons test, n = 3 mean with SD; *, P < 0.05.