The RUNX family in breast cancer: relationships with estrogen signaling

Abstract

The three RUNX family members are lineage specific master regulators, which also have important, context-dependent roles in carcinogenesis as either tumor suppressors or oncogenes. Here we review evidence for such roles in breast cancer (BCa). RUNX1, the predominant RUNX family member in breast epithelial cells, has a tumor suppressor role reflected by many somatic mutations found in primary tumor biopsies. The classical tumor suppressor gene RUNX3 does not consist of such a mutation hot spot, but it too seems to inhibit BCa; it is often inactivated in human BCa tumors and its haploinsufficiency in mice leads to spontaneous BCa development. The tumor suppressor activities of RUNX1 and RUNX3 are mediated in part by antagonism of estrogen signaling, a feature recently attributed to RUNX2 as well. Paradoxically, however RUNX2, a master osteoblast regulator, has been implicated in various aspects of metastasis in general and bone metastasis in particular. Reciprocating the anti-estrogenic tumor suppressor activity of RUNX proteins, inhibition of RUNX2 by estrogens may help explain their context-dependent anti-metastatic roles. Such roles are reserved to non-osseous metastasis, because ERα is associated with increased, not decreased skeletal dissemination of BCa cells. Finally, based on diverse expression patterns in BCa subtypes, the successful use of future RUNX-based therapies will most likely require careful patient selection.

Figure 1

RUNX1, highly expressed in breast epithelial cells, is frequently mutated in BCa. (a) RNA-sequencing data for RUNX1, RUNX2 and RUNX3 in human brain, heart, lymph node and breast tissues and in Human Mammary Epithelial Cells (HMEC) were produced by Wang et al. and displayed using the UCSC Genome Browser (http://genome.ucsc.edu). (b) Lollipop plot of RUNX1 somatic mutations identified by The Cancer Genome Atlas project and the Washington University BRC77 study. These studies indentified only one synonymous mutation in RUNX2 and none in RUNX3. Blue box represents the Runt DNA-binding domain.

Figure 2

Working model for the reciprocal antagonism between ERα and RUNX signaling. (a) During tumor initiation, RUNX proteins act as tumor suppressors in part by inhibiting ERα. (b) The beneficial effect of ERα in late-stage disease is attributable in part to repression of the pro-metastatic properties of RUNX2.

Figure 3

Roles of RUNX2 and ERα in BCa metastasis. According to this working model, the RUNX2/SNAI2 axis promotes early metastatic events (e.g., EMT, invasion, intravasation) and bone metastasis. ERα antagonizes RUNX2 during the early metastatic events but operates along with the RUNX2/SNAI2 axis during bone metastasis.

Figure 4

RUNX genes in BCa subtypes. Each tumor from a heterogeneous cohort of 557 BCa patients was subtyped according to the PAM-50 algorithm (courtesy of Dr Charles Perou). The values for RUNX1, RUNX3, HER2 and ESR1 represent the signal intensity associated with probe sets 210805_x_at, 204197_s_at, 216836_s_at and 205225_at, respectively (Affymetrix, Santa Clara, CA, USA). Because of the ambiguity with the RUNX2 probe sets in the early Affymetrix arrays used to profile mRNA expression, the RUNX2 values are those of a metagene defined as the average normalized expression of the top 100 RUNX2-stimulated genes in MCF7/Rx2^dox BCa cells.