¹Mice & More, Martinistr. 52, D-20251 Hamburg, Germany

²Heinrich-Pette-Institut für Experimentelle Virologie und Immunologie an der Universität Hamburg, D-20251, Hamburg, Germany

³Marienkrankenhaus, Institut für Pathologie, Alfredstr. 9, D-22087 Hamburg, Germany

Correspondence to: Christine Schulze-Garg, Heinrich-Pette-Institut für Experimentelle Virologie und Immunologie an der Universität Hamburg, D-20251, Hamburg, Germany

^aJ Löhler and A Gocht contributed equally to this work

The ductal carcinoma in situ (DCIS) of the mammary gland represents an early, pre-invasive stage in the development of invasive breast carcinoma and is increasingly diagnosed since the introduction of high-quality mammography screening. Uncertainties in the prognosis for patients with DCIS have caused a controversial discussion about adequate treatment, and it is suspected that most patients undergoing mastectomy may be overtreated. In order to improve treatment and treatment decision, it therefore is highly desirable to identify prognostic markers and therapeutic targets for DCIS. We here introduce a set of transgenic mice (WAP-T and WAP-T-NP lines) presenting with various morphological forms of DCIS-like lesions. In these mice the SV40 large tumor antigen is specifically induced in epithelial cells of the terminal duct lobular units (TDLU). As a consequence of continuous expression of the oncogene, the animals develop multifocal DCIS and consequently invasive carcinoma within strain specific periods of latency. DCIS lesions in transgenic mice exhibit distinct architectural and cytological features which closely resemble those commonly present in humans. We therefore propose these transgenic lines as an experimental model to study the underlying molecular events leading to DCIS and its progression to invasive disease. Oncogene (2000) 19, 1028-1037.

DCIS; ductal carcinoma in situ; mouse model; SV40 large T-antigen

Introduction

With an incidence of ~10% in women, mammary carcinoma is among today's most pressing health problems. More than 80% of all mammary carcinomas are ductal carcinomas which originate from the terminal ductal lobular units. The ductal carcinoma in situ (DCIS) most likely represents the earliest screen detectable, non invasive stage in the development of invasive ductal carcinomas. Since the introduction of high quality mammography screening DCIS has been increasingly diagnosed, and presently constitutes 10-20% of screen detected malignancies of the breast (Ernster et al., 1996; Ernster and Barclay, 1997). Furthermore, it turned out that DCIS presents with distinct histological patterns which are associated with different clinical outcomes. However, difficulties in establishing a definite correlation between morphological classification and clinical prognosis caused a controversial discussion about adequate treatment of DCIS (Bellamy et al., 1993; Scott et al., 1997). Although mastectomy offers high cure rates (nearly 99%), it has been recognized as overtreatment for the majority of DCIS patients (Silverstein et al., 1995a; Solin et al., 1996; Talamonti, 1996). Considering the severe medical and psychological problems associated with mastectomy, it is highly desirable to develop criteria, which allow a reliable diagnosis and prognosis of the various forms of DCIS. Such criteria could then form the basis for a rationale decision on the therapeutic strategy to be followed after diagnosis of DCIS.

Currently, the decision for treatment of DCIS is solely based on morphological criteria. Here, the Van Nuys classification (Silverstein et al., 1995b) is the most widely accepted method for risk estimation. Nuclear grading and the presence of intraductal comedo-type necrosis are the major criteria by which DCIS is classified into three groups with low, intermediate, or high risk of local recurrence after breast conserving therapy.

Although several prognostic factors have been elaborated for certain types of DCIS, the clinical outcome of DCIS in an individual patient is not exactly known (Holland et al., 1994; Silverstein et al., 1990, 1995a; Solin et al., 1993). Molecular markers associated with specific forms of DCIS and their progression to invasive ductal carcinoma should provide improved, i.e. more reliable tools for the diagnosis and prognosis of DCIS. Furthermore, knowledge of genetic alterations specific for certain phenotypes of DCIS, and for the subsequent transition to invasive carcinomas, should provide clues for the development of mechanism-based treatments. Therefore, an experimental animal model for DCIS could provide a means to study morphological and biological qualities of the different DCIS lesions more precisely. To address this issue, we developed mouse mutants, presenting with DCIS-like lesions. We introduced the SV40 early genome region, driven by the whey acidic protein (WAP) promoter, into the germline of CB6F2 mice, which then were back-crossed into BALB/c mice (for details see Materials and methods). The rationale for this strategy was based on the following considerations:

WAP-promoter

The WAP promoter is hormonally and developmentally regulated by lactotrophic hormones (e.g. estrogen, prolactin, hydrocortison, insulin) (Burdon et al., 1991; Pittius et al., 1988). Thus expression of the transgene can be induced by mating and is directed to epithelial cells of TDLUs of the differentiating and lactating gland. Consequently, the WAP promoter has been successfully used for the generation of mouse models for breast cancer (Husler et al., 1998; Tzeng et al., 1993).

SV40 early genome region as transgene

The major transforming protein of the DNA tumor virus simian virus 40 (SV40) is the large tumor antigen (T-Ag). T-Ag is a potent inducer of cellular transformation in vitro and of tumor formation in vivo (Butel and Lednicky, 1999). The transforming activity of SV40 T-Ag is largely based on protein-protein interactions with cellular proteins, notably its ability to inactivate the tumor suppressor proteins pRb and p53. pRb controls the entry of cells into S-phase by binding to members of the E2F transcription factor family, thereby maintaining them in an inactive state. Phosphorylation of pRb by cyclin-dependent kinases releases E2F from inhibition by pRb and promotes entry of the cells into S-phase due to the activity of genes transactivated by E2F (Bartek et al., 1997). Similarly, binding of pRb by T-Ag relieves E2F proteins from negative regulation by pRb, and results in entry of the cells into S-phase (Fanning, 1998). As pRb controls the G1 checkpoint, which is the major checkpoint in proliferation control, a direct inactivation through mutations in pRb or indirect inactivation through aberrant expression of proteins regulating pRb activity (e.g. cyclins, cyclin dependant kinases, cyclin/cdk inhibitors) is commonly found in human and murine tumors (Bartek et al., 1999). T-Ag mimics these genetic alterations and thereby can induce proliferation in otherwise quiescent cells.

Another important target of T-Ag is the tumor suppressor p53, whose main function is to maintain the integrity of the genome. Loss of p53 function leads to genomic instability, thereby accelerating the development and progression of neoplastic cells. In accordance, it has been demonstrated that expression of an N-terminal fragment of T-Ag, T121, which only binds pRb, but not p53, is sufficient for tumor induction as a transgene (Sáenz Robles et al., 1994). Crossing such transgenic mice with p53^-/- did neither change the tumor spectrum or phenotype, nor tumor penetrance, but resulted in a significant acceleration of tumor development in the T121´p53^-/- mice, due to abrogation of p53-mediated apoptosis (Symonds et al., 1994).

Use of CB6F2 mice and back-cross into BALB/c mice

BALB/c mice are a well studied model system for the immunology of SV40 induced tumors. In contrast to the `high responder' C57BL/6 mice, BALB/c mice are considered `low responders' in terms of a specific CTL response towards SV40 T-Ag (Pfizenmaier et al., 1980; Schirmbeck et al., 1993). As the CTL response against human tumor antigens also is rather weak, the immune response against SV40 induced tumors in BALB/c mice can be considered a model reflecting the immune response against human tumor antigens. Furthermore, the only transplantable mouse SV40 tumor cell lines are of BALB/c mice origin, and there is a wide body of data describing the immune reaction against such tumors in mice pre-immunized with SV40 or SV40 T-Ag (Bright et al., 1994; Zerrahn et al., 1996, and literature cited therein). Therefore, mice equipped with BALB/c MHC class I will also allow the analysis of questions relevant to the immunology of breast tumors. To be able to further address such questions, we inserted a 33 bp oligomer encoding the MHC class I (H-2^d) restricted T-cell epitope of the LCM virus, which is a dominant epitope in BALB/c mice (Frelinger et al., 1983; Örn et al., 1982; Schulz et al., 1989), into the C-terminal coding region of the SV40 T-Ag (WAP-T-NP transgenic lines). Expression of this hybrid gene allows the comparison of immune responses against `weak' BALB/c T-cell epitopes, as they are provided by SV40 T-Ag, with the `strong' T-cell epitope of the LCM virus.

In this article we describe the properties of in situ carcinoma arising from the epithelial cells of the TDLUs developing in six of the 22 generated transgenic lines, and show by histological analysis that they closely resemble human DCIS. Unlike other experimentally induced murine mammary tumors described so far, our WAP-T and WAP-T-NP lines display variable, but prolonged pre-invasive stages of tumorigenesis with characteristic periods of latency before becoming invasive. This disease model opens the unique opportunity to study and identify distinct steps in mammary tumor progression and the underlying molecular mechanisms promoting it. Furthermore, the identification of prognostic markers in this experimental setting could help to evaluate the risk of DCIS to convert into invasive carcinomas, and to develop adequate therapeutic strategies.

Results

WAP-T and WAP-T-NP transgenic mice develop typical `ductal mammary carcinomas' with characteristic periods of latency

It has been shown previously that a 1.6 kb fragment of the whey acidic protein promoter is able to direct inducible expression of SV40 T-Ag in the mammary gland of transgenic mice (Tzeng et al., 1993). We have used this transgene (WAP-T) and the LCM virus (BALB/c) T-cell epitope tagged version of it (WAP-T-NP) to generate a large set of transgenic founders, and bred these to the BALB/c background (for details see Materials and methods). The majority of the established transgenic lines (15 out of 22) develop inducible DCIS-like in situ carcinomas and invasive ductal mammary carcinomas during their lives. A minority develop tumors of other origin solely (4/22), or both inducible mammary carcinomas and tumors of other origin (7/22). Non-transgenic littermates served as control animals and never developed tumors during the observation period.

The transgenic lines presented here (Table 1) almost exclusively develop DCIS-like carcinoma and invasive ductal mammary carcinomas with characteristic periods of latency after induction. The predominant histological phenotypes of the invasive ductal mammary carcinomas of WAP-T and WAP-T-NP transgenic animals are differentiated papillary and tubular growth patterns or a mixture of both (Figure 1a). Less frequently are moderately or poorly differentiated carcinomas (Figure 1c) disclosing a solid architecture. Advanced carcinomas mostly exhibit tissue portions with variable degrees of differentiation. Adenocarcinomas with lobular differentiation can only rarely be seen (Figure 1b). Although we have not routinely screened for the occurrence of metastases, these have been occasionally found in various organs including lungs and rarely regional lymphnodes, as shown for line WAP-T-NP8 (Figure 1d).

Latency of tumor development was defined as the time between first induction and appearance of the first palpable mammary tumor. Histological analysis of the clinically observed tumors was routinely performed to confirm that tumors were indeed mammary carcinomas (for details see Materials and methods). In females of the transgenic lines WAP-T1 and WAP-T2 invasive ductal carcinomas are first detectable after an average latency period of 7 months following a single induction (Table 1). These carcinomas often show tubular or papillary differentiation, but also poorly differentiated ductal carcinomas can be found. Females of line WAP-T10 mostly present with invasive ductal carcinomas with tubular morphology after an average latency period of 9 months following a single induction period. Females of line WAP-T-NP10 develop invasive ductal carcinomas with the longest average latency, i.e. 12 months, after a single induction phase and both, well differentiated tubular or papillary and poorly differentiated solid carcinomas can be found. In contrast, in females of line WAP-T-NP8 rapidly growing, palpable tumors were evident, on average 5 months after induction. The tumors mostly display a poorly differentiated solid or even anaplastic morphology and well differentiated tumors are rarely found.

Line WAP-T-NP6 is an exception among the transgenic lines described here, in that only multiparous females reproducibly develop invasive ductal carcinoma. When induced only once, just 18% of the transgenic females developed mammary tumors in contrast to 75% of multiparous females. The average latency for tumor formation in multiparous females is 11 months, calculated from first mating. Characteristic periods of latency and penetrance of invasive ductal carcinomas of transgenic lines are summarized in Table 1.

T-Ag expression is necessary but not sufficient to initiate tumorigenesis and to drive progression to invasive ductal carcinoma

Immunostaining of mouse mammary tissue revealed that induced WAP-T and WAP-T-NP transgenic lines express SV40 T-Ag, or the epitope tagged version NP-T-Ag, exclusively in epithelial cells of TDLUs, and expression of T-Ag in these cells can be detected shortly after mating (data not shown). T-Ag expression in epithelial cells may persist for a long time span without any signs of transformation as indicated by the appearance of T-Ag positive, but morphological inconspicuous acinar ductules even 12 months after induction (Figure 2a). Thus, T-Ag clearly is required for initiation of tumorigenesis in these mice but is obviously not sufficient and additional genetic alterations promoting tumorigenesis are required. In addition, T-Ag expression persists throughout the in situ stage of tumorigenesis (Figure 2b) and in well differentiated invasive ductal carcinomas (Figure 2c). In tumors displaying variable degrees of differentiation, such areas that are well differentiated are T-Ag positive, whereas poorly differentiated areas of the same tumor are T-Ag negative (Figure 2d). These findings indicate that continuous T-Ag expression is necessary to drive progression, and that it will only become dispensable after additional genetic alterations, promoting T-Ag independent progression, have occurred.

WAP-T and WAP-T-NP transgenic lines develop in situ carcinoma closely resembling human DCIS

To characterize in situ carcinoma of WAP-T and WAP-T-NP mice by histology, we have collected tumor bearing and macroscopically normal mammary glands of induced transgenic females. Several morphologically distinguishable types of in situ carcinoma could be identified in WAP-T and WAP-T-NP transgenic mice, and we have compared these with the presently known morphological subtypes of human DCIS.

Human DCIS: The various defined forms of DCIS in humans include subtypes with typical cribriform (Figure 3b) or roman arches-like architecture (Figure 3a), as well as micropapillary, clinging (Figure 3c,d) and solid growth patterns. These forms may occur as pure or mixed types, e.g. micropapillary combined with the clinging type (Figure 3d) or a composition of clinging and cribriform patterns (Figure 3c). All these forms may occur with comedo-type necrosis and with microcalcifications (Figure 3e).

WAP-T and WAP-T-NP DCIS: Among the transgenic mouse lines, cribriform morphology of in situ carcinoma is relatively rare, an example is shown for line WAP-T-NP8 (Figure 4a). Representative clinging type forms occurring frequently in our mice are depicted for line WAP-T-NP6 and WAP-T1 (Figure 4d,e). In situ carcinoma with papillary growth pattern are displayed by line WAP-T2 (Figure 4f). Roman arches-like morphology was found in line WAP-T-NP10 in combination with a clinging type growth pattern (Figure 4b). Typical comedo necrosis of murine DCIS-like carcinoma is illustrated in Figure 4c, showing a solid to cribriform in situ carcinoma of line WAP-T-NP6. It should be pointed out that, as in the human condition, pure morphological types are rare and the majority of DCIS-like lesions of our transgenic lines exhibit a mixed growth pattern.

Comparison of diagnostic and prognostic parameters of human and mouse DCIS: DCIS in humans is mostly detected by high resolution mammography, because here intraductal microcalcifications, which are indicative of DCIS, become apparent. They represent calcification products of necrotic cellular debris or they are abnormal metabolic extra- or intracellular products of the malignant cells. By microscopial inspection these calcifications may appear as amorphous masses or onion ring-like psammoma bodies (Figure 3e). Such microcalcifications are likewise detected in murine DCIS (Figure 4d,e). The appearance of necrotic material in human DCIS, which is also frequently seen in corresponding lesions of WAP-T and WAP-T-NP mice (Figure 4a-f), is associated with higher malignancy and is therefore an essential feature for classification and grading.

However, the most important cytological criterion to evaluate the grade of differentiation and thereby the prognosis of DCIS lesions is the nuclear grade of the neoplastic cells. The grade of nuclei is determined by their size, pleomorphism, mitotic index, chromatin pattern, and size and number of nucleoli. A high mitotic index, large pleomorphic nuclei, hyperchromatosis, and prominent nucleoli are characteristics that identify a poorly differentiated lesion with a relatively high risk of recurrence. According to these criteria the human DCIS shown in Figure 3c and the mouse in situ carcinoma in Figure 4a,c represent high-grade lesions, while the other lesions illustrated represent examples for non high-grade forms.

In summary, these histological findings demonstrate that in situ carcinoma observed in our transgenic lines display many of the morphological features of the various human in situ carcinoma forms, commonly described as DCIS.

Mouse DCIS eventually invade the basement membrane and give rise to invasive mammary carcinoma

In transgenic mice DCIS-like lesions could mostly be observed in areas close to smaller or larger invasive ductal carcinomas, but could also be seen in tumor-free glandular tissue. Although it is generally assumed that DCIS progress to invasive carcinoma, the invasive disease and DCIS observed in WAP-T and WAP-T-NP transgenic mice, in theory, could have developed independently. In this case, the observed invasive disease would not represent progression of DCIS, but would rather have directly developed from other pre-neoplastic lesions (e.g. dysplasis, atypical ductal hyperplasia). To address this issue, we have histologically analysed mammary tissue for the appearance of DCIS-like lesions displaying local invasion of the basement membrane. If invasive adenocarcinoma in this transgenic animal model is the consequence of in situ carcinoma invading the basement membrane and surrounding tissue, such lesions should be detectable. As shown in Figure 5 such alterations can indeed be found. In Figure 5a few cells of a small in situ carcinoma are in the process of invading the otherwise intact basement membrane (Figure 5a, arrowhead), thus representing the microinvasive form of ductal carcinoma. The in situ component of the carcinoma can be classified as high-grade with respect to its cytological appearance and can be found frequently in DCIS-lesions of line WAP-T-NP8. Figure 5b shows a significant larger lesion of the same transgenic line with considerable intraductal proliferation and necrosis of the partly preserved preexisting DCIS, while the other part of the tumor extensively invades the surrounding adipose tissue. The in situ component of this invasive carcinoma represents a non high-grade form with micropapillary and clinging type architecture and comedo-type necrosis. These observations support our hypothesis that invasive disease seen in WAP-T and WAP-T-NP mice results from progression of DCIS-like in situ to invasive adenocarcinoma.

Discussion

Among the breast cancers, DCIS is a clearly distinguishable entity. In addition, DCIS itself is not a uniform lesion, as several morphological subtypes with different pathological and clinical implications have been described (Bellamy et al., 1993; Holland et al., 1994). Although prognostic molecular markers for invasive ductal carcinoma are available (e.g. c-erbB-2, estrogen and progesteron receptor, p53) and have been applied to human DCIS (Ravdin, 1997) their relevance as prognostic markers for DCIS remains to be shown. As a consequence, the classification of DCIS today mainly relies on morphological criteria. The identification of risk groups is therefore based on clinical studies that correlate morphological data with the probability of relapse after a specific therapy (e.g. mastectomy, conservative surgery, with or without radiotherapy) (Silverstein, 1997). The search for prognostic markers of DCIS has in part been hampered by the fact that no animal model for DCIS has been available or recognized as such among the existing mammary tumor models.

We here describe mice transgenic for the WAP-SV40 early genome region, which develop DCIS-like carcinoma and invasive mammary adenocarcinoma after induction, and might serve as a model system for the analysis of human DCIS and its progression to invasive carcinoma. With respect to the molecular mechanisms of tumorigenesis in this mouse model, we are aware that the introduction of a viral oncogene does not represent the induction of the disease in humans. However, clinical data obtained by genetic analysis and immunohistochemistry of human mammary carcinoma implicate the functional inactivation of the cellular tumor suppressors pRb and p53, as achieved by SV40 T-Ag expression, as common events in development and progression of mammary carcinoma in humans and in our mouse model (Bartek et al., 1999). It is also evident from our results and those of other groups that tumorigenesis in T-Ag transgenic mice, as in human cancer diseases, is a multistep process that requires multiple genetic alterations to drive progression from pre-neoplastic lesions to invasive and metastasizing carcinoma (Macleod and Jacks, 1999). This conclusion is strongly supported also by our finding that some TDLUs clearly expressing T-Ag did not progress to DCIS-like lesions or invasive carcinoma even after more than one year following induction. Interestingly the requirements for additional genetic alterations seem to differ among the transgenic lines we describe here, as reflected by specific periods of latency.

Histological analysis of DCIS-like neoplasia in WAP-T and WAP-T-NP transgenic mice revealed many similarities between the murine lesion and human DCIS with respect to architectural and cytological characteristics. The common architectural forms of human DCIS, cribriform, papillary, and clinging type, were also frequently found in mouse DCIS, with and without comedo-type necrosis, whereas forms displaying a solid architecture appeared to be underrepresented in the mouse model. A reason for this observation could be the induction of apoptosis in malignant cells by T-Ag (Tzeng et al., 1998), preventing the development of duct filling solid tumor growth. Although the correlation between the morphology of a given in situ carcinoma presented by an individual transgenic mouse line and the prognosis for this line remains to be investigated, there are several hints that the lesions presented by the transgenic line WAP-T-NP8 are highly malignant. Firstly, these animals show the shortest latency period of tumor formation, secondly the in situ carcinomas are mostly of high-grade differentiation and associated with early multifocal microinvasion. Accordingly, we would predict that in situ carcinomas in transgenic lines presenting with low-grade cytology, and more extended latency periods of tumor formation, indicate a more favorable prognosis.

Since we have so far only screened for DCIS in animals already displaying at least one palpable invasive adenocarcinoma, we have not yet found a pure DCIS, i.e. a female individual presenting exclusively with intraductal disease. The screening of induced mice in a time course will help to reveal the time of first appearance of in situ tumors and prove whether invasive adenocarcinoma developed by our transgenic lines is the consequence of in situ tumors progressing to invasive disease, as already suggested by the presence of microinvasive events in our animals (Figure 5), or if the two malignancies develop independently.

In the past several transgenic mouse lines have been established, using either the whey acidic protein promoter (Andres et al., 1987; Husler et al., 1998; Sandgren et al., 1995; Schoenenberger et al., 1988; Tzeng et al., 1993) or the mouse mammary tumor virus promoter (Choi et al., 1988; Daphna-Iken et al., 1998; Guy et al., 1992; Webster et al., 1995; Muller et al., 1988; Sinn et al., 1987) to target the expression of viral and cellular oncogenes to the mammary gland. Although in some cases identical or similar constructs were used to generate transgenic animals in different laboratories, the outcome with respect to mammary tumor incidence, latency of tumor formation, and tumor phenotype differed significantly between the obtained transgenic lines. This also holds true for the WAP-T and WAP-T-NP transgenic lines established by us, as each of the lines analysed was unique with respect to the above criteria. However, considering our observation that expression of the transgene as such is neither sufficient to induce DCIS-like nor the ensuing invasive mammary carcinoma, this heterogeneity may not be too surprising, as the outbreak as well as the outcome of the disease will be determined by additional genetic alterations not so far characterized. In this context we assume that genetic background (genotype of the mouse), integration site, and copy number are important factors that determine the requirement of additional mutations in an individual transgenic line. Therefore, it should be possible to establish a phenotype-genotype correlation, assigning certain genetic alterations to certain morphological types of DCIS and their prognoses. Such a correlation will be extremely helpful not only in understanding the genetic alterations leading to DCIS and invasive mammary carcinoma, but will also provide the basis for rational diagnosis and treatment decisions, and for mechanism-based new treatments of these diseases. Due to the morphological similarities of the DCIS and the invasive mammary carcinoma developing in the mouse lines described here with the respective human diseases, it is hoped that a phenotype-genotype correlation established in this model system can be transferred to human mammary cancer diseases.

Materials and methods

Generation of the WAP-T and WAP-T-NP transgene

The plasmid pWAP-T was kindly provided by Dr Adolf Graessmann. It encodes a 4.3 kb hybrid gene of the 1.6 kb BglII-KpnI fragment of the whey acidic protein promoter fused to the 2.7 kb BglI-BamHI fragment of the SV40 early coding region, inserted into the multiple cloning site of pUC 18 (Tzeng et al., 1993). The singular restriction site for BclI at position 2770 in the SV40 genome was used to insert a 33 bp fragment coding for the MHC class I (L^d) restricted epitope n118 (aa 118-126) in the nucleoprotein of the LCM virus (Schulz et al., 1989). The plasmid pWAP-T was amplified in the E. coli strain GM 2163 (dam, dcm) purified and digested with BclI. Pre-annealed oligonucleotides NPs/CSG-2 (5'-GATCCTAGGCCTCAAGCTTCTGGAGTCTACATG-3') and NPas/CSG-2 (5'-GATCCATGTAGACTCCAGAAGCTTGAGGCCTAG-3') were ligated to the linearized plasmid. Correct orientation and sequence of the inserted fragment was confirmed by PCR and sequencing. Prior to pronuclear injection plasmids were linearized and plasmid sequences removed by digestion with EcoRI and BamHI, purified (QIAquick gel extraction kit, Qiagen Germany) and extensively dialyzed against microinjection buffer (10 mM Tris-HCl pH 7.6, 0.1 mM EDTA).

Generation of WAP-T and WAP-T-NP transgenic mice

Transgenic mice were generated essentially as described elsewhere (reviewed in Hogan et al., 1986). In short, DNA fragments WAP-T and WAP-T-NP were injected at a concentration of 5-10 ng/l into the pronuclei of C57BL/6´BALB/c F2 (CB6F2) hybrid oocytes, followed by implantation of two-cell stage embryos into the oviducts of pseudopregnant females. Potential founder animals were analysed by PCR using the SV40 specific primers AC-1 (5'-TATGTCAGCAGAGCCTGTAGAACCAAAC-3') and DC-2 (5'-GAGAAAGGTAGAAGACCCCAAG-3') which generate a 765 bp fragment. PCR conditions were as follows: 95°C for 4 min, followed by 35 cycles of 45 s at 94°C, 30 s at 57°C and 1 min at 72°C, followed by an additional 10 min at 72°C. Fourteen out of 62 potential founders were positive for the WAP-T-NP transgene and 13 out of 46 potential founders were positive for the WAP-T transgene. The founder mice were bred to BALB/c mice and in the following generations, when available, male transgenic mice were bred to BALB/c females. Offspring were analysed by PCR as described above. Transgenic lines are continuously bred as hemizygotes on the BALB/c background.

Induction of T-Ag expression in transgenic females

In order to induce T-Ag expression in transgenic females, 2-4 animals of matched age were caged together and mated to non transgenic males. Mice were checked for the appearance of a vaginal plug, and males were removed when plug formation had been positive. Females were then allowed to nurse their offspring for 3-4 weeks. Females that did not give birth to offspring were regarded as induced, when they had proven to nurse the common offspring in the cage.

Histopathology and analysis of transgene expression

Mouse mammary tissue specimens were fixed with 4% formaldehyde containing 1% acetic acid and embedded in paraffin. Deparaffinated sections were stained with hematoxylin and eosin.

Immunostaining of SV40 large T-antigen was performed on paraffin sections using a triple-step immunoenzymatic method. Deparaffinated sections were reacted before antibody incubation with a commercial `target unmasking fluid' (Dianova, Hamburg, Germany) in a microwave oven. Subsequently, sections were incubated overnight at 4°C with a 1 : 10 000 dilution of a polyclonal rabbit antiserum against T-Ag (R15 anti-SDS-T, Deppert and Pates, 1979). Specifically bound primary antibody was detected using biotinylated anti-rabbit IgG and phosphatase-conjugated streptavidin from a commercial kit (Super Sensitive Detection System, Biogenex, San Ramon, CA, USA). Phosphatase enzyme activity was revealed with naphthol AS-BI phosphate in combination with hexazotized new fuchsine (Merck, Darmstadt, Germany). Alternatively, mouse monoclonal antibodies pAB101 and KT3, (McArthur and Walter, 1984; Gurney et al., 1986) in the form of mixed, spent tissue culture supernatants were employed for immunostaining of T-Ag. The `HistoMouse-SP' kit (Zymed Laboratories, San Francisco, CA, USA) was used to detect the mouse primary antibodies on the mouse tissue samples without background labelling of endogenous immunoglobulin. The streptavidin-horse radish peroxidase conjugate was visualized by the diamino benzidine method. Naïve rabbit serum and an irrelevant mouse hybridoma supernatant of the same immunoglobulin isotype served as controls. Sections were slightly counterstained with hemalum.

Acknowledgements

We wish to thank Karin Heigl and Roland Gugel for excellent technical assistance and the staff of the mouse facility at the Heinrich-Pette-Institute for the support in breeding and maintenance of the transgenic animals. This work was supported by the Wilhelm-Sander-Stiftung in Munich, Germany. The Heinrich-Pette-Institut is financially supported by Freie und Hansestadt Hamburg and by the Bundesministerium für Gesundheit.

Andres AC, Schonenberger CA, Groner B, Henninhausen L, LeMeur M and Gerlinger P. (1987). Proc. Natl. Acad. Sci. USA 84, 1299-1303. MEDLINE

Bartek J, Bartkova J and Lukas J. (1997). Exp. Cell. Res. 237, 1-6. Article MEDLINE

Bartek J, Lukas J and Bartkova J. (1999). J. Pathol. 187, 95-99. Article MEDLINE

Bellamy COC, McDonald C, Salter DM, Chetty U and Anderson TJ. (1993). Hum. Pathol. 24, 16-23. MEDLINE

Bright RK, Shearer MH and Kennedy RC. (1994). J. Immunol. 153, 2064-2071. MEDLINE

Burdon T, Sankaran L, Wall RJ, Spencer M and Hennighausen L. (1991). J. Biol. Chem. 266, 6909-6914. MEDLINE

Butel JS and Lednicky JA. (1999). J. Natl. Cancer Inst. 91, 119-134. MEDLINE

Choi Y, Lee I and Ross SR. (1988). Mol. Cell. Biol. 8, 3382-3390. MEDLINE

Daphna-Iken D, Shankar DB, Lawshé A, Ornitz DM, Shackleford GM and MacArthur CA. (1998). Oncogene 17, 2711-2717. MEDLINE

Deppert W and Pates R. (1979). Nature 277, 322-324. MEDLINE

Ernster VL, Barclay J, Kerlikowske K, Grady D and Henderson C. (1996). JAMA 275, 913-918. MEDLINE

Ernster VL and Barclay J. (1997). J. Natl. Cancer Inst. Monogr. 22, 151-156. MEDLINE

Fanning E. (1998). Dev. Biol. Stand. 94, 3-8. MEDLINE

Frelinger JA, Orn A, Brayton PR and Hood L. (1983). Transplant. Proc. 15, 2024-2026. MEDLINE

Gurney EG, Tamowski S and Deppert W. (1986). J. Virol. 57, 1168-1172. MEDLINE

Guy CT, Cardiff RD and Muller WJ. (1992). Mol. Cell. Biol. 12, 954-961. MEDLINE

Hogan B, Constantini F and Lacy E. (1986). Manipulating the mouse embryo. A laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.,

Holland R, Peterse JL, Millis RR, Eusebi V, Faverly D, van de Vijver MJ and Zafrani B. (1994). Semin. Diagn. Pathol. 11, 167-180. MEDLINE

Husler MR, Kotopoulis KA, Sundberg JP, Tennent BJ, Kunig SV and Knowles BB. (1998). Transgenic Res. 7, 253-263. MEDLINE

MacArthur H and Walter G. (1984). J. Virol. 52, 483-491. MEDLINE

Macleod KF and Jacks T. (1999). J. Pathol. 187, 43-60. Article MEDLINE

Muller WJ, Sinn E, Pattengale PK, Wallace R and Leder P. (1988). Cell 54, 105-115. MEDLINE

Örn A, Goodenow RS, Hood L, Brayton PR, Woodword JG, Harmon RC and Frelinger JA. (1982). Nature 297, 415-417. MEDLINE

Pfizenmaier K, Pan SH and Knowles BB. (1980). J. Immunol. 124, 1888-1891. MEDLINE

Pittius CW, Sankaran L, Topper YJ and Hennighausen L. (1988). Mol. Endocrinol. 2, 1027-1032. MEDLINE

Ravdin PM. (1997). in: Ductal carcinoma in situ of the breast. Williams & Wilkins, Baltimore, MD, USA,

Sáenz Robles MT, Symonds H, Chen J and Van Dyke T. (1994). Mol. Cell. Biol. 14, 2686-2698. MEDLINE

Sandgren EP, Schroeder JA, Qui TH, Palmiter RD, Brinster RL and Lee DC. (1995). Cancer Res. 55, 3915-3927. MEDLINE

Schirmbeck R, Zerrahn J, Kuhröber A, Deppert W and Reimann J. (1993). Eur. J. Immunol. 23, 1528-1534. MEDLINE

Schoenenberger CA, Andres AC, Groner B, van der Valk M, LeMeur M and Gerlinger P. (1988). EMBO J. 7, 169-175. MEDLINE

Schulz M, Aichele P, Vollenweider M, Bobe FW, Cardinaux F, Hengartner H and Zinkernagel RM. (1989). Eur. J. Immunol. 19, 1657-1667. MEDLINE

Scott MA, Lagios MD, Axelsson K, Roger LW, Anderson TJ and Page DL. (1997). Hum. Pathol. 28, 967-973. MEDLINE

Silverstein MJ, Waisman JR, Gamagami P, Gierson ED, Colburn WJ, Rosser RJ, Gordon PS, Lewinsky BS and Fingerhut A. (1990). Cancer 66, 102-108. MEDLINE

Silverstein MJ, Barth A, Poller DN, Gierson ED, Colburn WJ, Waisman JR and Gamagami P. (1995a). Eur. J. Cancer 31A, 1425-1427.

Silverstein MJ, Poller DN, Waisman JR, Colburn WJ, Barth A, Gierson ED, Lewinsky B, Gamagami P and Slamon DJ. (1995b). Lancet 345, 1154-1157.

Silverstein MJ. (1997). Ductal carcinoma in situ of the breast. Williams & Wilkins, Baltimore, MD, USA.,

Sinn E, Muller W, Pattengale P, Tepler I, Wallace R and Leder P. (1987). Cell 49, 465-475. MEDLINE

Solin LJ, Yeh I, Kurtz J, Fourquet A, Recht A, Kuske R, McCormick B, Cross MA, Schultz DJ and Amalric R. (1993). Cancer 71, 2532-2542. MEDLINE

Solin LJ, Kurtz J, Fourquet A, Amalric R, Recht A, Bornstein BA, Kuske R, Taylor M, Barrett W, Fowble B, Haffty B, Schultz DJ, Yeh IT, Mccormick B and McNeese M. (1996). J. Clin. Oncol. 14, 754-763. MEDLINE

Symonds H, Krall L, Remington L, Saenz-Robles M, Lowe S, Jacks T and Van DT. (1994). Cell 78, 703-711. MEDLINE

Talamonti MS. (1996). Semin. Surg. Oncol. 12, 300-313. MEDLINE

Tzeng YJ, Guhl E, Graessmann M and Graessmann A. (1993). Oncogene 8, 1965-1971. MEDLINE

Tzeng YJ, Zimmermann C, Guhl E, Berg B, Avantaggiati ML and Graessmann A. (1998). Oncogene 16, 2103-2114. MEDLINE

Webster MA, Cardiff RD and Muller WJ. (1995). Proc. Natl. Acad. Sci. USA 92, 7849-7853. MEDLINE

Zerrahn J, Utermoehlen O, Warnecke G, Deppert W and Lehmann-Grube F. (1996). J. Immunol. 156, 3919-3924. MEDLINE

Figure 1 Histopathological features of invasive ductal carcinoma of mammary glands in WAP-T and WAP-T-NP transgenic mice. (a) Well differentiated tubular and papillary carcinoma of a WAP-T1 mouse. (b) Invasive lobular carcinoma of a WAP-T-NP10 mouse. (c) Poorly differentiated ductal carcinoma with solid architecture of a WAP-T-NP8 mouse. (d) Lung metastasis of a poorly differentiated ductal carcinoma of a WAP-TPNP8 mouse. H&E; magnification (a)´155, (b)´195, (c)´195, (d)´40

Figure 2 Immunohistochemical detection in WAP-T and WAP-T-NP transgenic mice of SV40 T-Ag during different stages of tumorigenesis after induction. (a) Positive and negative, nuclear immunostaining of acinar or ductular epithelial cells of TDLUs which appear morphologically inconspicuous. (b) DCIS with mixed cribriform, clinging and roman arch-like growth pattern. (c) Well differentiated tubulo-papillary, invasive ductal carcinoma. (d) Tubular ductal carcinoma with areas of poor differentiation and negative nuclear immunostaining in central parts of the microphotograph, whereas the well differentiated tumor tissue in the periphery exhibits positive antibody decoration. Magnification (a)´315, (b)´195, (c)´195, (d)´195

Figure 3 Characteristic morphological growth patterns of human DCIS. (a) Non high-grade DCIS with roman arch-like architecture. (b) Non high-grade DCIS with cribriform architecture. (c) High-grade DCIS with a mixed clinging and cribriform type architecture. (d) Non high-grade DCIS with predominantly micropapillary and partly clinging type architecture. (e) Non high-grade cribriform DCIS containing a psammoma body (arrow). H&E; magnification (a)´215, (b)´170, (c)´165, (d)´80, (e)´210

Figure 4 Characteristic morphological growth patterns of murine DCIS of transgenic WAP-T and WAP-T-NP mice. (a) High-grade DCIS of a WAP-T-NP8 mouse with cribriform architecture. (b) Non high-grade DCIS of a WAP-T-NP10 mouse with mixed roman arch-like and clinging type architecture. (c) High-grade DCIS of a WAP-T10 mouse with cribriform to solid architecture. A comedo-type necrosis is indicated by an asterisk. (d) Non high-grade DCIS of a WAP-T-NP6 mouse with clinging type architecture and calcification of the central necrosis. (e) Non high-grade DCIS of a WAP-T1 mouse with clinging type architecture. One DCIS profile contains a psammoma body with onion ring-like structure (arrow). (f) Non high-grade DCIS of WAP-T2 mouse with papillary architecture. H&E; magnification (a)´255, (b)´100, (c)´160, (d)´255, (e)´160, (f)´115

Figure 5 Transition of DCIS to invasive carcinoma in mammary gland tissue of transgenic mice. (a) Focal invasion (arrow) of the basement membrane by neoplastic cells in a microinvasive ductal carcinoma of a WAP-T-NP8 mouse. The nuclear grading of the in situ component of the carcinoma is high-grade. Note the strong inflammatory reaction around the neoplasia. (b) This invasive ductal carcinoma of a WAP-T-NP8 obviously originates from a micropapillary type DCIS, the residual in-situ component of which can be recognized in the left part of the microphotograph. H&E, magnification (a)´215, (b)´55

Table 1 Incidence and latency of mammary carcinomas in WAP-T and WAP-T-NP transgenic lines displaying DCIS