Reduced androgen receptor expression accelerates the onset of ERBB2 induced breast tumors in female mice

Abstract

Androgen receptor (AR) is commonly expressed in both the epithelium of normal mammary glands and in breast cancers. AR expression in breast cancers is independent of estrogen receptor alpha (ERα) status and is frequently associated with overexpression of the ERBB2 oncogene. AR signaling effects on breast cancer progression may depend on ERα and ERBB2 status. Up to 30% of human breast cancers are driven by overactive ERBB2 signaling and it is not clear whether AR expression affects any steps of tumor progression in this cohort of patients. To test this, we generated mammary specific Ar depleted mice (MARKO) by combining the floxed allele of Ar with the MMTV-cre transgene on an MMTV-NeuNT background and compared them to littermate MMTV-NeuNT, Ar(fl)/+ control females. Heterozygous MARKO females displayed reduced levels of AR in mammary glands with mosaic AR expression in ductal epithelium. The loss of AR dramatically accelerated the onset of MMTV-NeuNT tumors in female MARKO mice. In this report we show that accelerated MMTV-NeuNT-dependent tumorigenesis is due specifically to the loss of AR, as hormonal levels, estrogen and progesterone receptors expression, and MMTV-NeuNT expression were similar between MARKO and control groups. MMTV-NeuNT induced tumors in both cohorts displayed distinct loss of AR in addition to ERα, PR, and the pioneer factor FOXA1. Erbb3 mRNA levels were significantly elevated in tumors in comparison to normal mammary glands. Thus the loss of AR in mouse mammary epithelium accelerates malignant transformation rather than the rate of tumorigenesis.

Figure 1. Characterization of mice with conditional knockout of AR in mammary glands (MARKO).

A. Breeding strategy used to create experimental cohorts. Two generation breeding was used to produce tumor-prone MMTV-NeuNT (activated rat ERBB2) transgenic MARKO (male and females underlined on the left side) and Control (underlined on the right) mice. MARKO mice are positive for MMTV-cre transgene, specifically expressed in mammary glands. B. Recombination events in the genomic DNA from mammary ductal tissue dissociated from fat cells. Top band (952 bp) is derived from the non-recombined floxed Ar^fl allele, middle band (855 bp) is from the wild-type Ar⁺ allele, and the lower band (404 bp) is an amplicon derived from the Ar^Δ allele, resulting from Cre/LoxP induced deletion of exon 2. Ar^fl/+ is a positive control for the floxed allele and WT is wild-type (Ar⁺/Ar⁺). Left panel is recombination in male mice and the right panel is recombination in female mice. No recombination is observed in mice lacking MMTV-cre (lanes 1 and 4), the deleted allele is present in mice with MMTV-cre (lanes 2 and 3) C. Reduced number of AR positive luminal epithelial cells in MARKO mammary glands. AR positive luminal epithelial cells in the mammary glands of Control (n = 5) and MARKO (n = 5) mice were counted and determined as a percentage of the total number of cells per gland (p = 0.0019). An average of 260 cells were counted per individual mouse. Immunohistochemical staining for AR expression in Control (D) and MARKO (E and F) mammary glands. Scale bar = 20 µm.

Figure 2. Control and MARKO mice show similar proliferation in the normal mammary gland.

Representative pictures of normal Control (A) (n = 4) and MARKO (B) (n = 4) mammary glands stained for Ki67. Arrowheads denote Ki67 positive cells. C. Ki67 staining was quantified and is shown as the percentage of total epithelial cells positive and negative for Ki67 staining. Scale bar = 20 µm.

Figure 3. MARKO mice show an earlier onset of breast tumors in MMTV-NeuNT mice.

A. Mice were examined two to three times per week starting at 6 months for the presence of palpable tumors. Age of incidence was recorded as the first day at which a palpable tumor was detected. Percentage of tumor free mice was plotted versus the age of the mice in days. Hazard ratio = 0.2340 and 95% CI ratio of = 0.06054 to 0.9047. B. Mean age of tumor incidence was calculated for tumor bearing MARKO (273+/−18.36 days, n = 7) and control (352.5+/−8.263 days, n = 10) mice, p = 0.0005. C. Survival of MARKO and Control MMTV-NeuNT females. Percentage of surviving mice was plotted versus age (p = 0.3499 at 400 days). Hazard ratio = 0.6023 and 95% CI of ratio = 0.2080 to 1.744. D. Average time in days between tumor detection and sacrifice, due to tumor burden prior to 400 days of age, p = 0.5000. Representative hematoxylin and eosin staining of Control (E) and MARKO (F) tumors. Scale bar = 50 µm.

Figure 4. Elevated ERBB2 and ERBB3 expression in Female MMTV-NeuNT tumors.

A. Serum obtained at sacrifice from control (n = 11) and MARKO (n = 10) mice was analyzed for estradiol levels (E2). B. Serum from A was analyzed for testosterone (T) levels. C. MMTV-NeuNT expression. At the time of sacrifice, tumors were harvested and non-tumor bearing mammary glands were dissociated from the fat pad and RNA was prepared as described in the methods section. NeuNT expression was analyzed by quantitative RT-PCR and normalized to 18S expression. Expression of NeuNT in tumors (Control n = 9 and MARKO n = 7) for MARKO and Control mice and normal mammary glands (Control n = 26, MARKO = 15) was normalized to expression in non-tumor bearing mammary glands of the Control group. D-G. Immunohistochemical detection of ERBB2 in tumors, Control (D) and MARKO (E) at 40X magnification, and in normal mammary ducts, Control (F) and MARKO (G) at 100X magnification. Gene expression analysis of Erbb3 (H) and Erbb4 (I) in the same samples as in C. *denotes P<0.05, **P<0.01 and ***P<0.005.

Figure 5. Steroid receptor expression is reduced in MMTV-NeuNT tumors.

At the time of sacrifice, tumors were harvested and non-tumor bearing mammary glands were dissociated from the fat pad. RNA was prepared as described in the methods section. Levels of Ar (A), Esr1 (B), Pr (C) and 18S were analyzed by quantitative RT-PCR (Control normal = 26, Control tumor = 9, and MARKO normal = 15 and MARKO tumor = 7). Expression of each receptor was normalized to 18S. Receptors in tumors for MARKO and Control mice and normal MARKO glands were normalized to expression in non-tumor bearing mammary glands from the Control group. *denotes P<0.05, **P<0.01 and ***P<0.005. Immunohistochemical detection of ERα in the normal mammary gland (D) of Control mice and tumors from Control (E) and MARKO (F) mice. Immunohistochemical detection of AR in the normal mammary gland of Control mice (G) and tumors from Control (H) and MARKO (I) mice. Scale bar = 50 µm.

Figure 6. Downstream targets of steroid receptors are reduced in MMTV-NeuNT tumors.

Tumors and non-tumor bearing mammary glands were evaluated for the expression of ERα and PR downstream target genes. Expression of each gene was normalized to 18S. Gene expression in tumors for MARKO and Control mice and normal MARKO glands were normalized to expression in non-tumor bearing control mammary glands (Control normal = 26, Control Tumor = 9, MARKO normal = 15 and MARKO tumor = 7). Reduced expression was seen in tumors compared to normal tissue for Areg (A), Rankl (B), Wnt4 (C) and Foxa1 (D). ***P<0.005, for normal versus tumor expression.

Figure 7. MMTV-NeuNT does not affect steroid receptor expression in normal mammary glands.

MMTV-NeuNT expressing (NeuNT) and non-expressing (WT) mammary glands were dissociated from the fat pad. RNA was prepared as described in the materials and methods section. Expression of NeuNT (A), Ar (B), Esr1 (C), Pr (D), and Foxa1 (E) were analyzed by qRT-PCR (WT n = 8 and MMTV-NeuNT n = 10). Expression of each gene was normalized to 18S. (*denotes P<0.05).