A molecular switch from tumor suppressor to oncogene in ER+ve breast cancer: Role of androgen receptor, JAK-STAT, and lineage plasticity

Abstract

Cancers develop resistance to inhibitors of oncogenes mainly due to target-centric mechanisms such as mutations and splicing. While inhibitors or antagonists force targets to unnatural conformation contributing to protein instability and resistance, activating tumor suppressors may maintain the protein in an agonistic conformation to elicit sustainable growth inhibition. Due to the lack of tumor suppressor agonists, this hypothesis and the mechanisms underlying resistance are not understood. In estrogen receptor (ER)-positive breast cancer (BC), androgen receptor (AR) is a druggable tumor suppressor offering a promising avenue for this investigation. Spatial genomics suggests that the molecular portrait of AR-expressing BC cells in tumor microenvironment corresponds to better overall patient survival, clinically confirming AR's role as a tumor suppressor. Ligand activation of AR in ER-positive BC xenografts reprograms cistromes, inhibits oncogenic pathways, and promotes cellular elasticity toward a more differentiated state. Sustained AR activation results in cistrome rearrangement toward transcription factor PROP paired-like homeobox 1, transformation of AR into oncogene, and activation of the Janus kinase/signal transducer (JAK/STAT) pathway, all culminating in lineage plasticity to an aggressive resistant subtype. While the molecular profile of AR agonist-sensitive tumors corresponds to better patient survival, the profile represented in the resistant phenotype corresponds to shorter survival. Inhibition of activated oncogenes in resistant tumors reduces growth and resensitizes them to AR agonists. These findings indicate that persistent activation of a context-dependent tumor suppressor may lead to resistance through lineage plasticity-driven tumor metamorphosis. Our work provides a framework to explore the above phenomenon across multiple cancer types and underscores the importance of factoring sensitization of tumor suppressor targets while developing agonist-like drugs.

Keywords: JAK STAT; androgen receptor; breast cancer; estrogen receptor; tumor suppressor.

Fig. 1.

AR is a ligand-activated tumor suppressor in ER-positive breast cancer. (A) WHIM23 orthotopic patient derived xenograft (PDX) experiment was performed by implanting tumor tissue into the mammary fat pad of 6–8 wks old female NSG mice (n = 8–10/group). Once the tumors reach 100–300 mm³, the animals were randomized and treated orally with dihydrotestosterone (DHT) (30 mg/kg/day), enobosarm (30 mg/kg/day), or vehicle control [DMSO + PEG-300 (20:80)]. Tumor volume was measured by digital caliper twice weekly and the percent change in tumor volume is represented in the graph. (B) WHIM23 tumor-bearing mice treated with vehicle, enobosarm, or DHT reaching euthanasia criteria. (C) Hallmark pathways downregulated in DHT and enobosarm-treated WHIM23 tumors are shown. RNA sequencing was performed with WHIM23 tumors from panel A and GSEA analysis was performed to determine the hallmark pathways differentially downregulated in DHT or enobosarm-treated tumors. (D) Genes up-regulated in DHT and enobosarm-treated WHIM23 tumors from panel A) were used to derive patient survival data from TCGA data. (E) Single nuclei were isolated from vehicle and enobosarm-treated tumors (vehicle, n = 2275 cells; enobosarm, n = 4573 cells) from panel A and sequenced. UMAP and violin plots of pathway signatures (basal signature, luminal signature, and AR signature) are shown. Mean ± S.E.M is shown for all the graphs and the data were analyzed by One Way ANOVA followed by Tukey post-hoc test or paired t-test. *P < 0.05; **P < 0.01; ***P < 0.001. AR—androgen receptor; ER—estrogen receptor; DHT—dihydrotestosterone; KRT18—keratin 18; ER—estrogen receptor; UMAP—Uniform Manifold Approximation and Projection.

Fig. 2.

ER- and AR-positive breast cancers develop resistance to AR agonists. (A) WHIM23 orthotopic PDX experiment was performed by implanting tumor tissue into the mammary fat pad of 6–8 wks old female NSG mice. Once the tumors reach 100–300 mm³, the animals were randomized and treated orally with enobosarm (30 mg/kg/day) or vehicle control (DMSO + PEG-300) for 28 days (n = 8–10/group) or until resistance developed. Tumor volume was measured by digital caliper twice weekly and the percent change in tumor volume is represented in the graph. P-values are a comparison between the vehicle control’s last measurement and the treatment group’s last measurement. The resistant tumors’ last measurements are non-significant (ns) compared to the vehicle’s last tumor measurements. (B) WHIM23 PDX experiment was performed as indicated in panel A, but the tumor-bearing mice were treated with vehicle, 30 mg/kg/day enobosarm or RAD140. Data represented as tumor volume. (C and D) WHIM23 tumor-bearing NSG mice (n = 8–10/group) were treated with enobosarm (30 mg/kg/day) or SARM RAD-140 (30 mg/kg/day) until resistance developed (indicated by the black arrow). Treatment was switched as indicated in the figure (red arrow). Tumor volume was measured twice weekly and the percent change in tumor volume is represented. Mean ± S.E.M is shown for all the graphs and the data were analyzed by One Way ANOVA followed by Tukey post-hoc test. *P < 0.05; **P < 0.01; ***P < 0.001. SARM—tissue selective AR modulator; AR—androgen receptor; DHT—dihydrotestosterone; mpk—mg/kg body weight.

Fig. 3.

Bulk RNA-seq in WHIM23 tumors indicates enrichment of Hallmark interferon and oncogenic pathways. (A) RNA sequencing was performed with WHIM23 tumors (n = 4/group) obtained from animals shown in Fig. 2A. Top hits of Gene set enrichment analysis (GSEA) analysis are shown as heatmap with interferon, inflammation, and JAK/STAT signaling pathways shown in green boxes and EMT and KRAS pathways shown in pink boxes. Gene set from the Molecular Signatures Database of the Broad Institute is reported in resistant tumors compared to the respective sensitive tumors (FDR < 0.25). (B) GSEA enrichment plots for HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION, HALLMARK_INTERFERON_ALPHA, and HALLMARK_KRAS_SIGNALING_UP in resistant WHIM23 tumors compared to sensitive tumors are shown. (C) A heatmap of representative genes from EMT, JAK-STAT, AR, and stem signatures from the RNA-seq data with WHIM23 tumors. Sensitive and resistant tumors were compared and heatmap of genes different in these pathways are shown. (D) Survival plot was obtained with genes that were decreased in resistant WHIM23 tumors compared to sensitive WHIM23 tumors.Genes decreased in resistant tumors were used to obtain Kaplan Meier plot with survival data obtained from TCGA database. EMT—epithelial to mesenchymal transition; JAK/STAT—Janus kinase/signal transducer and activator of transcription; AR—androgen receptor; ARaR—androgen receptor agonist resistant; ARaS—androgen receptor agonist sensitive; NES—normalized enrichment score.

Fig. 4.

AR and ER cistrome are altered in AR agonist-resistant WHIM23 tumors. (A and B) Chromatin immunoprecipitation (ChIP) assay was performed for AR (A) and ER (B) in WHIM23 tumors that are sensitive or resistant to enobosarm (n = 3) (tumors from animals shown in Fig. 2A). Next-generation sequencing was performed to determine the genome-wide binding of AR and ER to the DNA. Heatmap of significantly different peaks (q < 0.05) is shown for individual sensitive and resistant tumor specimens. The top enriched motifs are shown to the right of the heatmap (Motif sequences are provided in SI Appendix, Fig. S10A). (C) Two-factor log-ratio (M) plot showing SARM resistant-induced changes in ER and AR enrichment at AR/ER-target gene loci. Point color denotes AR and ER peak occupancy (orange). Example ARBSs and ERBSs associated with ER and AR target genes are highlighted in orange. AR—androgen receptor; ER—estrogen receptor; SARM—selective androgen receptor modulator; ARBSs—AR binding sites; ERBSs—ER binding sites; GSEA—gene set enrichment analysis; EMT—epithelial to mesenchymal transition.

Fig. 5.

AR agonist-resistant WHIM23 tumor cells exhibit plasticity. (A) Single nuclei were isolated from enobosarm sensitive and resistant WHIM23 tumors (from tumors that are sensitive and resistant to enobosarm shown in Fig. 2A) and sequenced (sensitive, n = 4573 cells; resistant, n = 2292 cells). UMAP and violin plots of pathway signatures (basal, luminal, KRAS, WNT, and E2F signatures) are shown. Luminal keratin 18 (KRT18) UMAP is also shown. (B) PROP1 signature is increased in WHIM23 ARaR tumors. Single nuclei sequencing data from WHIM23 ARaS and ARaR tumors were used to determine the PROP1 signature. For comparison, PROP1 signature of vehicle and enobosarm-treated WHIM23 tumors from Fig. 1A is shown. (C) UMAP plots of single-nuclei transcriptomic profiles of WHIM23 tumor cells isolated from tumors labeled sensitive or resistant; WHIM23 tumors (sensitive, n = 4573 cells; resistant, n = 2292 cells). Cells are colored by predicted cell cycle phases. Bar plot representing the percent distribution of single nuclei in different cell cycle phases in sensitive and resistant cell clusters. Mean ± S.E.M is shown for all the graphs and the data were analyzed by One Way ANOVA followed by Tukey post-hoc test. *P < 0.05; **P < 0.01; ***P < 0.001. UMAP—Uniform Manifold Approximation and Projection; KRT18—keratin 18.

Fig. 6.

AR switches from being a tumor suppressor to being an oncogene (A) AR protein levels are increased in enobosarm-resistant ZR-75-1 cells compared to sensitive cells. Protein was extracted from sensitive and resistant ZR-75-1 cells (ZR-75-1 cells that developed resistance to enobosarm) and Western blot with AR and GAPDH antibodies was performed. (B) AR protein level is increased in enobosarm-resistant WHIM23 tumors compared to enobosarm-sensitive tumors. Protein was extracted from ARaS and ARaR WHIM23 tumors (n = 3/type; tumors sensitive and resistant to enobosarm from Fig. 2A) and Western blot with AR and GAPDH antibodies was performed. (C) UMAP plots of single-nuclei transcriptomic profiles of WHIM23 nuclei isolated from tumors labeled sensitive or resistant to enobosarm (sensitive, n = 4573 cells; resistant, n = 2292 cells). UMAP and violin plots of the AR and ER signatures are shown. (D) UMAP plots of single-nuclei transcriptomic profiles of WHIM23 nuclei isolated from tumors labeled sensitive or resistant; WHIM23 tumors sensitive and resistant to enobosarm (sensitive, n = 4573 cells; resistant, n = 2292 cells). Heatmaps and violin plots of BL1, BL2, and LAR signatures are shown. (E) A schematic of different AR-targeting drugs used in the manuscript, including their binding sites and agonist (green arrow)/antagonist (red blocking arrow) function (biorender.com). (F) SARD UT-105 inhibits enobosarm-resistant WHIM23 tumors. WHIM23 tumor-bearing NSG mice (n = 8–10/group) were treated with enobosarm (30 mg/kg/day) until resistance developed. Treatment was switched and resistant tumors were treated orally with vehicle, enzalutamide, or UT-105. Tumor volume was measured twice weekly and the percent change in tumor volume is represented. P-values are a comparison between the vehicle and the treatment group. Mean ± S.E.M is shown for all the graphs and the data were analyzed by One Way ANOVA followed by Tukey post-hoc test. *P < 0.05; **P < 0.01; ***P < 0.001. AR—androgen receptor; UMAP—Uniform Manifold Approximation and Projection; BL1—basal-like 1; BL2—basal-like 2; LAR—luminal androgen receptor; ARaS—AR agonist sensitive; ARaR—AR agonist resistant; ER—estrogen receptor.

Fig. 7.

JAK/STAT inhibition re-sensitizes WHIM23 ARaR tumor cells to AR agonist. (A) UMAP plots of single-nuclei transcriptomic profiles of WHIM23 nuclei isolated from tumors labeled sensitive or resistant; WHIM23 tumors sensitive and resistant to enobosarm (sensitive, n = 4573 cells; resistant, n = 2292 cells). UMAP and violin plots of the JAK/STAT signature are shown. (B) Enobosarm-resistant tumors express higher phospho-STAT3. Protein was extracted from enobosarm sensitive and resistant WHIM23 tumors from Fig. 2A and Western blot with phospho-STAT3 and total STAT3 antibodies was performed. (C) Survival plot with JAK/STAT signaling genes observed in ARaR tumors. Genes in the JAK-STAT signaling pathways were used to create survival plot using TCGA database. Gene expression was divided into low and high, and survival of patients falling into each category was determined. (D) Ruxolitinib in combination with enobosarm inhibits proliferation of enobosarm-resistant ZR-75-1 cells. Enobosarm sensitive and resistant ZR-75-1 cells were plated in growth medium and treated as indicated in the figure. Cells were treated for twelve days with medium change and retreatment every third day. Sulforhodamine B (SRB) colorimetric assay was performed to measure cell viability. (E) UMAP plots represent the reconstructive trajectory of single nuclei in WHIM23 tumors from Fig. 2A (enobosarm sensitive and resistant). (F) UMAP plots representing the BL1, BL2, and LAR signatures, and lineage scores per cell in each single nuclei within the pseudotime reconstructive trajectory. Mean ± S.E.M is shown for all the graphs and the data were analyzed by One Way ANOVA followed by Tukey post-hoc test. *P < 0.05; **P < 0.01; ***P < 0.001. UMAP—Uniform Manifold Approximation and Projection; BL1—basal-like 1; BL2—basal-like 2; LAR—luminal androgen receptor.

Fig. 8.

A schematic of the proposed mechanism of resistance and lineage plasticity upon sustained activation of a tumor suppressor (biorender.com). AR is a tumor suppressor in ER-positive breast cancer. ER/AR-positive breast cancer exists in a multilineage state (mixture of luminal and basal). ER is the driver of tumor growth and the FOXA1 facilitates this growth. When tumors are treated with an activator of the tumor suppressor AR, AR binds to ARE, and FOXA1 binds to its response elements (FXRE) near AREs. This sequestration of FOXA1 inhibits ER function and tumor growth, driving tumors to a more differentiated luminal phenotype. Sustained activation of tumor suppressor AR causes an activation of the JAK/STAT pathway and switches AR to an oncogene. Tumors then enter into an aggressive less differentiated basal TNBC-like state where AR and ER chromatin binding are dysfunctional and the master transcription factor PROP1 binding site is gained. This lineage plasticity and aggressive growth can be addressed with AR degraders or antagonists or a combination of JAK1/2 inhibitor and AR agonists. The metamorphosis to aggressive less differentiated basal cells could also be prevented by combining AR agonist and JAK inhibitor from the beginning of treatment initiation with AR agonist.