DREAM, a possible answer to the estrogen paradox of the Women's Health Initiative Trial

Abstract

Estrogen is thought to cause proliferation of all estrogen receptor positive (ER+) breast cancers. Paradoxically, in the Women's Health Initiative Trial, estrogen-only hormone replacement therapy reduced the incidence and mortality of low grade, ER+, HER2- breast cancer. We gave estradiol to 19 post-menopausal women with newly diagnosed low-grade, ER+, HER2- breast cancer in a prospective window of opportunity clinical trial and examined the changes in proliferation and gene expression before and after estradiol treatment. Ki67 decreased in 13/19 (68%) patients and 8/13 (62%) showed a decrease in Risk of Recurrence Score. We chose three prototypical estrogen responders (greatest decrease in ROR) and non-responders (no/minimal change in ROR) and applied a differential gene expression analysis to develop pre-treatment (PRESTO-30^core) and post-treatment (PRESTO-45^surg) gene expression profiles. The PRESTO-30^core predicted adjuvant benefit in a published series of tamoxifen, the partial estrogen agonist. Of the 45 genes in the PRESTO-45^surg, thirty contain the Cell cycle genes Homology Region (CHR) motif that binds the class B multi-vulva complex (MuvB) a member of the DREAM (Dimerization partner, retinoblastoma-like proteins, E2F, MuvB) complex responsible for reversible cell cycle arrest or quiescence. There was also near uniform suppression (89%) of the remaining DREAM genes consistent with estrogen induced activation of the DREAM complex to mediate cell cycle block after a short course of estrogens. To our knowledge, this is the first report to show estrogen modulation of DREAM genes and suggest involvement of DREAM pathway associated quiescence in endocrine responsive post-menopausal ER+ breast cancers.

Keywords: Breast cancer; DREAM; Estrogen; Hormone therapy.

Figure 1

Validation of the RNA-seq data using BC360™. Correlation based unsupervised hierarchical cluster analysis of fold change using BC360™ gene expression data (Nanostring Technologies, Inc. Seattle, Washington, USA) and RNA-seq data. Samples' pairs are highlighted, showing all corresponding patients' consistently cluster together.

Figure 2

Estrogen decreases Ki67 and Risk of Recurrence (ROR) Score. Changes in Ki67 (A) (P = 0.025) and ROR (B) (P = 0.07) from the pre-estrogen core biopsy (Bx) to the post-estrogen surgical specimen (Sx) with responders (red solid lines) and non-responders (red dotted lines) shown. Ki67 immunohistochemical (brown) staining for patient CCI-006 before (C) and after (D) estrogen. Scale bar: 100μm.

Figure 3

Characteristics of estrogen induced gene transcription changes. (A) Contingency table analysis of the 38 key estrogen induced apoptosis genes (Ariazi et al. [27]) with a log₂ fold change in responders vs. non-responders (Chi-square test with Yate's correction, P = 0.2383, ns). Kaplan Meier curves for distant relapse free survival for node negative (B) tamoxifen treated (P = 0.001) or (C) untreated patients (P = ns) scored using PRESTO-30^core. See text for data sources. Statistical analysis using log-rank. ns – not significant.

Figure 4

Overlap between PRESTO-45^surg and DREAM genes in responders. (A) Venn diagram showing overlap between genes repressed through the p53-LIN37/DREAM pathway (Uxa et al. [32]) and PRESTO-45^surg. (B) Volcano plot of the gene changes in responding patients after estrogen. PRESTO-45^surg genes are indicated by blue circles. Genes repressed through the DREAM pathway are marked by an X.

Figure 5

Changes in key DREAM mediators and hypothetical model. (A) p53 protein expression in the surgical specimens of responders and non-responders is similar (p = ns, t-test). p53 protein was detected immunohistochemically using the DO-7 antibody. Expression was calibrated using image analysis and then scored manually using a histoscore calculation (see ref. [42]). (B) Log2 fold change (surgical over core samples) in responders or non-responders for TP53 (p53), CDKN1A (p21), FOXM1 and MYBL2, after estradiol. ∗∗P < 0.01, ns – not significant. (C) In this hypothetical model, estrogen therapy is associated with increases in p21 and decreases in FOXM1 and/or MYBL2. The mechanism for these changes in responding patients is currently unknown and this is indicated by dashed lines. Increases in p21 (or cyclin dependent kinase inhibitors, not shown) will lead to hypophosphorylated forms of the retinoblastoma family proteins (RB, p107/RBL1 and p130/RBL2). Hypophosphorylated p130 binds with the LIN52 member of MuvB core proteins resulting in a stable DREAM repressor complex which binds to the CHR element of cell cycle genes through another MuvB core protein (LIN54). The DREAM complex suppresses cell cycle genes causing reversible cell cycle arrest (quiescence). The DREAM complex can be destabilized by increases in pro-proliferative MuvB co-factors, FOXM1 and MYBL2. When MuvB is complexed with these proteins and bound to the CHR motif of the same cell cycle genes, it activates transcription resulting in proliferation.