The transcription factor ATF3 acts as an oncogene in mouse mammary tumorigenesis

Abstract

Background: Overexpression of the bZip transcription factor, ATF3, in basal epithelial cells of transgenic mice under the control of the bovine cytokeratin-5 (CK5) promoter has previously been shown to induce epidermal hyperplasia, hair follicle anomalies and neoplastic lesions of the oral mucosa including squamous cell carcinomas. CK5 is known to be expressed in myoepithelial cells of the mammary gland, suggesting the possibility that transgenic BK5.ATF3 mice may exhibit mammary gland phenotypes.

Methods: Mammary glands from nulliparous mice in our BK5.ATF3 colony, both non-transgenic and transgenic, were examined for anomalies by histopathology and immunohistochemistry. Nulliparous and biparous female mice were observed for possible mammary tumor development, and suspicious masses were analyzed by histopathology and immunohistochemistry. Human breast tumor samples, as well as normal breast tissue, were similarly analyzed for ATF3 expression.

Results: Transgenic BK5.ATF3 mice expressed nuclear ATF3 in the basal layer of the mammary ductal epithelium, and often developed squamous metaplastic lesions in one or more mammary glands by 25 weeks of age. No progression to malignancy was seen in nulliparous BK5.ATF3 or non-transgenic mice held for 16 months. However, biparous BK5.ATF3 mice developed mammary carcinomas with squamous metaplasia between 6 months and one year of age, reaching an incidence of 67%. Cytokeratin expression in the tumors was profoundly disturbed, including expression of CK5 and CK8 (characteristic of basal and luminal cells, respectively) throughout the epithelial component of the tumors, CK6 (potentially a stem cell marker), CK10 (a marker of interfollicular epidermal differentiation), and mIRSa2 and mIRSa3.1 (markers of the inner root sheath of hair follicles). Immunohistochemical studies indicated that a subset of human breast tumors exhibit high levels of nuclear ATF3 expression.

Conclusion: Overexpression of ATF3 in CK5-expressing cells of the murine mammary gland results in the development of squamous metaplastic lesions in nulliparous females, and in mammary tumors in biparous mice, suggesting that ATF3 acts as a mammary oncogene. A subset of human breast tumors expresses high levels of ATF3, suggesting that ATF3 may play an oncogenic role in human breast tumorigenesis, and therefore may be useful as either a biomarker or therapeutic target.

Figure 1

ATF3 expression in myoepithelial cells of BK5.ATF3 mice. (a). A paraffin section (4 μm) of a mammary gland from an 8 week old BK5.ATF3 virgin female was stained with fluorescently labeled antibodies to CK5 (Texas Red, red) and CK8 (FITC, green) and examined by fluorescence microscopy. (b-e). Sections were also treated with a primary antibody to ATF3, then stained using horseradish peroxidase-coupled secondary antibody and a chromogenic substrate that produces a brown stain, and counterstained with hematoxylin. Mammary gland from a BK5.ATF3 transgenic female (b,d) was compared with a non-transgenic littermate (c,e). Scale bars = 100 μm; scale bar in panel c applies to panels b and c, scale bar in panel e applies to panels d and e. (f). Nuclear extracts were analyzed for ATF3 protein expression by immunoblotting as described in Materials and Methods. Equal amounts of protein from mammary glands of non-transgenic mice (lane WT) or BK5. ATF3 mice of line 1 (lane 1), line 2 (lane 2), line 3 (lane 3), or line 4 (lane 4) were analyzed. As a positive control, an extract from RAW 264.7 cells was analyzed (lane C). β-actin was used as loading control. An ATF3-specific band was detected in the 20–25 KD size range in the BK5.ATF3 mammary gland extracts and in the positive control extract; this band was barely detectable in extracts from non-transgenic mice (lane 1). The ATF3 band was abolished by preincubation with the ATF3 blocking peptide (not shown).

Figure 2

Squamous metaplasia in virgin transgenic animals. Mammary glands from 21 nulliparous, BK5.ATF3 line 1 mice between 14 and 32 weeks of age and 18 age-matched, nulliparous non-transgenic littermates were examined histologically and by IHC. Transgenic mice from three other BK5.ATF3 lines that express the transgene were also examined (Table 1). (a) A low power view of multiple squamous metaplastic lesions in a single gland of a BK5.ATF3 line 4 female. (b-f) Two typical, cystic lesions from BK5.ATF3 line 1 mammary glands are shown, stained for (b) CK5; (c) CK6; (d) CK8; (e) ATF3; (f) CK10. Scale bar in a = 200 μm; scale bar in b = 100 μm, applies to panels b, c, and d; scale bar in f = 100 μm, applies to panels e and f.

Figure 3

Mammary tumors in parous BK5.ATF3 mice. (a). Female wild-type mice (n = 12, closed squares) or BK5.ATF3 mice (n = 15, closed triangles) were allowed to mate and raise pups twice between 6 and 24 weeks of age, then observed until 16 months of age. Mice were euthanized when tumors reached 1.5 cm. Groups of nulliparous wild type (n = 20) and BK5.ATF3 (n = 13) females were also maintained for 16 months (data not shown); survival of both of these groups was 100% at the conclusion of the experiment. Sections of two representative tumors arising in BK5.ATF3 mice are shown at low power (b,c; scale bar = 200 μm); sections from five different tumors are shown at higher power (d-h; scale bar = 50 μm). KC, keratinaceous core.

Figure 4

ATF3, Ki67 and cytokeratin expression in BK5.ATF3 mammary tumors. Immunohistochemistry of representative tumors arising in BK5.ATF3 mice utilizing primary antibodies directed against ATF3 (a), Ki67 (b), CK5 (c), CK8 (d), and CK6 (e). Scale bars = 100 μm. KC, keratinaceous core; S, stroma.

Figure 5

Co-expression of CK5 and CK8 in tumor cells. A paraffin section of a mammary tumor arising in a BK5.ATF3 parous female was stained with fluorescently labeled antibodies to CK5 (Texas Red, red) and CK8 (FITC, green) and examined by fluorescence microscopy. (a), CK5 (b), CK8 (c), merged image.

Figure 6

Deregulation of cytokeratin expression in BK5.ATF3 mammary tumors. Immunohistochemistry of representative mammary tumors arising in BK5.ATF3 mice (a,c,e) or uninvolved mammary ducts of tumor-bearing animals (b,d,f) was performed with primary antibodies to CK10 (a,b), mIRS.a2 (c,d) and mIRS.a3.1 (e,f), and stained with the chromogenic substrate. Scale bar = 100 μm. KC, keratinaceous core; S, stroma.

Figure 7

Expression of ERα and ErbB2 in BK5.ATF3 mice. Immunohistochemistry for ERα (a,b,c) or for ErbB2 (d,e,f) was performed with representative sections from mammary tumors arising in BK5.ATF3 mice (a,d), from a normal appearing mature mammary gland (b), from a mammary tumor that developed in an MMTV.neu transgenic animal (e), or from squamous metaplastic lesions (c, f) that occurred in the mammary glands of nulliparous BK5.ATF3 females. Scale bar = 100 μm.

Figure 8

Expression of ATF3 in human mammary tumors. Paraffin sections from human invasive ductal carcinomas that were less than 2 cm at presentation were examined by immunohistochemistry with an antibody directed against ATF3 (a-c,e,f). In (b), a blocking peptide for ATF3 (sc-188p, Santa Cruz) was added prior to the ATF3 antibody; the same region of the tumor is shown in (a) and (b). Panels a, c, and e represent tumors from three different patients. An adjacent section to that shown in (c) was analyzed for CK5 expression, and the corresponding region of the tumor is shown in (d). (f) shows a hyperplastic duct adjacent to an ATF3-positive tumor. (g-h), paraffin sections from normal mammoplasty tissues were examined by IHC with the same ATF3-specific antibody. Scale bar = 100 μm.