Identification of a pharmacologically tractable Fra-1/ADORA2B axis promoting breast cancer metastasis

Abstract

Metastasis confronts clinicians with two major challenges: estimating the patient's risk of metastasis and identifying therapeutic targets. Because they are key signal integrators connecting cellular processes to clinical outcome, we aimed to identify transcriptional nodes regulating cancer cell metastasis. Using rodent xenograft models that we previously developed, we identified the transcription factor Fos-related antigen-1 (Fra-1) as a key coordinator of metastasis. Because Fra-1 often is overexpressed in human metastatic breast cancers and has been shown to control their invasive potential in vitro, we aimed to assess the implication and prognostic significance of the Fra-1-dependent genetic program in breast cancer metastasis and to identify potential Fra-1-dependent therapeutic targets. In several in vivo assays in mice, we demonstrate that stable RNAi depletion of Fra-1 from human breast cancer cells strongly suppresses their ability to metastasize. These results support a clinically important role for Fra-1 and the genetic program it controls. We show that a Fra-1-dependent gene-expression signature accurately predicts recurrence of breast cancer. Furthermore, a synthetic lethal drug screen revealed that antagonists of the adenosine receptor A2B (ADORA2B) are preferentially toxic to breast tumor cells expressing Fra-1. Both RNAi silencing and pharmacologic blockade of ADORA2B inhibited filopodia formation and invasive activity of breast cancer cells and correspondingly reduced tumor outgrowth in the lungs. These data show that Fra-1 activity is causally involved in and is a prognostic indicator of breast cancer metastasis. They suggest that Fra-1 activity predicts responsiveness to inhibition of pharmacologically tractable targets, such as ADORA2B, which may be used for clinical interference of metastatic breast cancer.

Fig. 1.

Gene-expression profiling of a metastasis model system identifies Fra-1 as a candidate metastasis gene. (A) Microarray gene-expression analysis of RK3E^TB and RIE^TB cells. The top 10 genes that are up- or down-regulated in both cell systems are shown in a heat map. (B) Fra-1 and E-cadherin expression levels measured by Western blotting in RK3E^TB cells expressing independent shRNAs targeting Fra-1 as indicated. α-Tubulin serves as loading control. (C) Representative images of macroscopic pulmonary metastases and hematoxylin-eosin staining of histological lung sections from mice injected s.c. with control or Fra-1–depleted RK3E^TB cells analyzed 3 wk postinoculation. (Scale bars: 200 μm.) M, metastasis. (D) Macroscopic quantification of pulmonary metastases in mice described in C. Data in B–D are representative of three independent experiments.

Fig. 2.

Suppression of Fra-1 abrogates the metastatic potential of human breast cancer cells and restores epithelial characteristics. (A) Expression levels of Fra-1 and epithelial proteins in LM2 cells as a function of Fra-1 depletion measured by Western blotting. α-Tubulin serves as loading control. (B) Representative bioluminescence images of mice injected i.v. with 2 × 10⁵ LM2 cells expressing a control or sh-Fra1 vector at different time points as indicted. (C) Quantification of the luminescence signal in the lungs of mice described in B as a function of time. n = 6. Error bars indicate SE. *P < 0.05, **P < 0.01, ***P < 0.005 vs. control based on a two-tailed Wilcoxon signed-rank test. (D) Kaplan–Meier curves for the survival of the mice described in B and C. Mice were killed when clinical symptoms became apparent. Displayed P value is based on a log-rank test. (E) Immunohistochemical analysis of human vimentin expression (which was retained by Fra-1–depleted cells) identifying metastases formed in the lungs of recipient mice by control or Fra-1–depleted LM2 cells. (Upper) Composite image of the entire left lung. (Lower) Image taken at 20× magnification. (Scale bars: 100 μm.) Data are representative of two (B–E) or three (A) independent experiments.

Fig. 3.

A Fra-1–associated gene-expression profile accurately predicts clinical outcome of human breast cancer. (A) Outline of the procedure used to identify a Fra-1–dependent transcriptome signature in LM2 cells and to derive a prognostic Fra-1 classifier. (B) A heat map representing one-sided Cox proportional hazards model P values for time to distant metastasis (if available) or relapse on breast cancer datasets not used to train the Fra-1 classifier (SI Appendix, Methods) for the indicated gene-expression signatures. (C) A heat map representing one-sided log-rank P values between signature-high samples and signature-low samples for time to distant metastasis (if available) or relapse on samples of indicated breast cancer subtypes from datasets not used to train the Fra-1 classifier (SI Appendix, Methods).

Fig. 4.

A screen for small molecules identifies adenosine receptor inhibitors as compounds preferentially targeting Fra-1–expressing breast tumor cells. (A) Screen design and protocol. (B) Dose–response curves of control and Fra-1–depleted MDA-MB-231 cell lines treated with the indicated compounds. n = 4. Error bars indicate SD. (C) Dose–response curves of control and Fra-1–depleted LM2 cell lines treated with the indicated compounds. n = 4. Error bars indicate SD. The P values shown in B and C are for the two sh-Fra-1 groups vs. control based on a repeated measures ANOVA followed by Bonferroni’s multiple comparison test. Data in B and C are representative of three independent experiments.

Fig. 5.

ADORA2B is a Fra-1 target gene contributing to metastatic activity of breast cancer cells. (A) Quantification of the luminescence signal in the lungs of mice injected i.v. with 2 × 10⁵ LM2 cells expressing a control or sh-ADORA2B vector at different time points as indicated. n = 6. Error bars indicate SE. *P < 0.05, **P < 0.01 for the two sh-ADORA2B groups vs. control based on a two-tailed Wilcoxon signed-rank test. (B) Migration (Left) and invasion (Right) capacities as a function of ADORA2B depletion. n = 3. Error bars indicate SD. *P < 0.002 vs. control based on a one-way ANOVA followed by a partial least-squares difference test. (C) Representative immunofluorescence imaging of filopodia in LM2 cells depleted for ADORA2B or treated with 100 μM theophylline. Control cells were treated with excipient. Merged images of DNA (blue), F-actin (red), and α-tubulin (green) staining are shown. Individual images are provided in SI Appendix, Fig. S14C. (Scale bars: 10 μm.). (D) Quantification of the luminescence signal in the lungs of mice injected i.v. with 2 × 10⁵ LM2 cells and receiving the indicated treatment (docetaxel, 4 mg/kg i.p. weekly, and/or theophylline, 10 mM in drinking water) at different time points as indicated. n = 6. Error bars indicate SE. *P < 0.05, **P < 0.005, combined treatment vs. docetaxel-only group based on a two-tailed Wilcoxon signed-rank test. Data are representative of two (A–C) or three (D) independent experiments.