RUNX2 correlates with subtype-specific breast cancer in a human tissue microarray, and ectopic expression of Runx2 perturbs differentiation in the mouse mammary gland

Abstract

RUNX2, a master regulator of osteogenesis, is oncogenic in the lymphoid lineage; however, little is known about its role in epithelial cancers. Upregulation of RUNX2 in cell lines correlates with increased invasiveness and the capacity to form osteolytic disease in models of breast and prostate cancer. However, most studies have analysed the effects of this gene in a limited number of cell lines and its role in primary breast cancer has not been resolved. Using a human tumour tissue microarray, we show that high RUNX2 expression is significantly associated with oestrogen receptor (ER)/progesterone receptor (PR)/HER2-negative breast cancers and that patients with high RUNX2 expression have a poorer survival rate than those with negative or low expression. We confirm RUNX2 as a gene that has a potentially important functional role in triple-negative breast cancer. To investigate the role of this gene in breast cancer, we made a transgenic model in which Runx2 is specifically expressed in murine mammary epithelium under the control of the mouse mammary tumour virus (MMTV) promoter. We show that ectopic Runx2 perturbs normal development in pubertal and lactating animals, delaying ductal elongation and inhibiting lobular alveolar differentiation. We also show that the Runx2 transgene elicits age-related, pre-neoplastic changes in the mammary epithelium of older transgenic animals, suggesting that elevated RUNX2 expression renders such tissue more susceptible to oncogenic changes and providing further evidence that this gene might have an important, context-dependent role in breast cancer.

Keywords: Breast cancer; Mammary development; RUNX2; Transgenic model.

Fig. 1

Expression of RUNX2 correlates with ER/PR/HER2-negative human breast cancer. Invasive breast carcinomas from a tumour tissue microarray (TMA-1) were stained for RUNX2. (A) Scatterplot showing the range of positive histoscores. Expression was divided into RUNX2-negative (histoscore 0), -low (histoscore 1–24) or -high (histoscore ≥25) in 416 breast cancers. The dotted line at histoscore 25 demonstrates the cut-off for RUNX2-high patients. Position on the y-axis reflects the order in which samples were analysed. (B) Kaplan-Meier of patient survival for 384 patients in A for which follow-up data was available. Survival is plotted for patients with high-RUNX2 tumours (n=21) and negative/low-RUNX2 tumours (n=363). (C) Examples of individual tumours stained for RUNX2 and ER, depicting the reciprocal expression pattern. Boxed areas are shown at higher magnification. (D) Significantly more RUNX2-high cancers were ER-negative (P=0.005; chi-square) and specifically associated with the triple-negative (ER/PR/HER2-negative) group (P=0.008; chi-square).

Fig. 2

Transgenic expression of Runx2 perturbs pubertal mammary development. (A) qRT-PCR of Runx2 throughout murine mammary development. Expression levels relative to wild-type 12-week-old virgins; data are means ± s.d. (P, pregnant; L, lactating; inv, involution, d, day). (B) qRT-PCR of Runx expression in basal/myoepithelial and luminal epithelial cell populations sorted by FACS based on CD29 and CD24 surface markers (see text for details). Runx1 is plotted on a different y-axis owing to the higher levels of expression. Runx3 expression was not detectable in either population. Expression normalised to Gapdh is relative to luminal Runx2; data are means of three independent samples ± s.d. (C) Whole-mounts of 6-week-old mammary glands. Elongation from lymph node (LN) in weight- and litter-matched WT glands is greater than in MMTV-Runx2 glands. Ductal elongation lengths as represented by the red arrows are quantified in the box plot (P=0.002; WT n=28; Runx2 n=23). Dots represent outliers. (D) Whole-mounts of 8-week-old MMTV-Runx2 glands reveal a reduction in tertiary side-branching in Runx2 glands compared with WT glands. Number of ducts per H&E section were counted for each sample (inset images); graph shows a significant reduction in average duct number in Runx2 mammary glands (P=0.01; WT n=15; Runx2 n=16). Dots represent outliers. Whole-mounts, 6.5× magnification; H&Es, 40× magnification.

Fig. 3

Ectopic RUNX2 expression leads to delayed differentiation and a lactation defect. (A) Whole-mount and histological analysis of MMTV-Runx2 glands (n=11) during late pregnancy (D17 pregnant) reveals a reduction in side-branching and alveolar expansion compared with WT (n=14), and failure to form mature alveolar units during lactation (WT n=13; Runx2 n=11). (B) Ki67 staining of WT and MMTV-Runx2 glands at day 12 (D12) pregnancy (n=4 each), late pregnancy (D17P; n=6 each) and day 1 of lactation (D1; WT n=6; Runx2 n=7). (C) Immunohistochemistry of whey acidic protein (WAP) and RUNX2 on serial sections at day 1 lactation in WT (n=6) and MMTV-Runx2 (n=7) glands illustrates their reciprocal expression pattern. (D) Quantification of Wap mRNA with dramatically less Wap in Runx2 glands at late pregnancy (d17P) and lactation (d1L); data are means ± s.d. normalised to HPRT relative to WT virgin. Whole-mounts, 8× magnification; H&Es, 200× magnification; IHC images, 100× magnification in B and 200× magnification in C.

Fig. 4

Ectopic Runx2 inhibits ELF5 and p-STAT5 during lactation. (A) Prolactin receptor (Prlr) mRNA expression at pregnancy (d8P, d17P) and lactation (d1L) in Runx2 and WT glands; data are means ± s.d., normalised to HPRT and relative to WT virgin. (B) ELF5 and p-STAT5 protein levels are not affected in MMTV-Runx2 glands at late pregnancy (day 17) but, during lactation, Runx2 overexpression causes a reduction in levels of ELF5 and p-STAT5 (C). Each image is representative of n≥3 mice. Boxed areas are shown at higher magnification. Scale bars: 10 μm (5 μm in higher-magnification images).

Fig. 5

Runx2-induced lactation failure is partially rescued by multiple parities. Transgenic females were housed with co-fostered WT females to achieve continued suckling from pups. Representative images for single parous (A) and multiparous (≥4 pregnancy) (B) co-fostered females at day 1 lactation. Whole-mounts show the retarded development in transgenics, although Runx2 multiparous glands have more mature alveolar formation, with glands evidently fuller and producing more WAP (B) than Runx2 single-parity females (A). Multiparous transgenic glands proportionally express less RUNX2 than single-parity glands, which is reciprocal to the higher levels of WAP as shown in serial sections in A and B. Boxed areas are shown at higher magnification. Single-parous females: WT n=6, Runx2 n=7; multiparous females: WT n=3, Runx2 n=3. Quantification of RUNX2 positivity per total epithelium in multiparous (grey) and single-parous (black) glands is given in the bar graph (P=0.0001; single parous n=5; multiparous n=3). Scale bars in wholemounts: 2 mm in A and 1 mm in B; scale bars in IHC images: 20 μm (10 μm in higher-magnification images).

Fig. 6

Mammary glands of aged MMTV-Runx2 females display abnormal hyperplastic and pre-neoplastic changes. (A) Representative whole-mounts of aged MMTV-Runx2 and WT littermate controls (8× magnification). (B) Representative images of H&E sections showing diffuse hyperplasia in two independent MMTV-Runx2 transgenic glands with evidence of secretory hyperplastic lesions and dilated ducts (100× magnification). (C) Abnormal features observed in aged MMTV-Runx2 glands, such as distorted acini with lobular fibrosis and chronic inflammatory cell infiltrate (I), and alveolar hyperplasia with luminal cells exhibiting large nuclei and prominent nucleoli (II,III) (images shown at 400× magnification). (D) Ductal carcinoma in situ (DCIS) in an MMTV-Runx2 female; middle panel is higher magnification of boxed area. Smooth muscle actin (SMA) staining shows an intact basal/myoepithelial layer. (E) Hyperplastic lesions are negative for ER, PR and HER2 but show positivity for MYC as determined by immunohistochemistry (400× magnification).