Loss of chromosomal integrity drives rat mammary tumorigenesis

Abstract

Breast cancer incidence varies with diet and other environmental influences, including carcinogen exposure. However, the effects of carcinogens on cell growth control pathways are poorly understood. Here, we have examined processes that are activated in the mammary glands of rats treated with 1-methyl-1-nitrosourea (MNU). This synthetic carcinogen was used to study events occurring during mammary tumor initiation and development. In female Wistar-Furth rats, given 1 dose of MNU beginning at 50 days of age, 84% of the rats developed tumors by 46 weeks of age (latency 13–15 weeks). Changes in the gland occurred as early as 1-day post-MNU. Cells exhibited DNA damage, leading to chromosomal instability, supernumerary centrosomes and higher levels of Aurora A; these events correlated with the appearance of preneoplasia in the glands. In mammary tumors, elevated numbers of centrosomes coincided with genomic instability. Tumors were transplanted into syngeneic hosts and subsequent tumor generations displayed the same marker chromosomes in mostly aneuploid metaphases with hyperdiploid numbers of chromosomes, suggesting that clonality and aneuploidy were passed on from one generation to the next. Collectively, these data suggest that the carcinogen MNU induces changes resulting in genetic instability detectable before hyperplasia and tumors develop in the rat mammary gland. © 2006 Wiley-Liss, Inc.

Experimental models in laboratory rodents have provided a great tool to gain insights about many factors involved in the genesis, prevention and treatment of breast cancer.1, 2, 3, 4 A limited number of studies have addressed the nature of the disease from its premalignant to malignant stages.1, 5, 6 In 1-methyl-1-nitrosourea (MNU)-induced and 7,12-dimethylbenz[a] anthracene (DMBA)-induced models, the stepwise evolution of mammary lesions was demonstrated.1, 6, 7, 8

Cancers arise via multistage carcinogenesis; intermediate stages, termed preneoplastic or premalignant, are used to identify the essential properties and molecular alterations for neoplastic development. Histopathologically, the premalignant steps in the rat include interductal proliferation also called ductal hyperplasia (DH), and ductal carcinomal in situ (DCIS) (reviewed in Ref.6). In general, it appears that epithelial cells in mammary terminal end buds or terminal ductules are the targets of carcinogenic initiation. Several studies on rat mammary carcinogenesis have focused on identifying genetic alterations in primary MNU-induced mammary tumors.9, 10 Cytogenetic analyses indicate that these tumors start to develop as diploid lesions with cytogenetically “normal” karyotypes; tumors showed coexistence of normal diploid clones with abnormal clones bearing specific chromosomal aberrations.11 One study identified and compared AI (allelic imbalance) in primary rat mammary tumors10; the results of the AI genome scan revealed low levels of genetic instability. A mutation in codon 12 of the Ha-ras gene (G to A) is frequently observed in MNU-induced mammary carcinomas.12 However, this mutation is also present in mammary glands from nonexposed animals, indicating that Ha-ras-1 gene mutation appears to be neither necessary nor sufficient for neoplastic transformation.13, 14 Mutations in the p53 gene were rare and no gross gene alteration or amplification was seen in mdm2, a negative regulator of p53 transcription.15 Collectively, previously published data suggested that MNU-induced tumors are diploid.

Earlier we have demonstrated that tumors induced by MNU (96-days-old animals treated with two doses of MNU)16 contain supernumerary centrosomes that was accompanied by the overexpression of the centrosome-associated kinase Aurora A. These events were found to occur at early premalignant stages (40-days post-MNU) prior to detectable lesions suggesting a link between centrosome amplification, mitotic spindle aberrations, genomic instability and rat mammary tumorigenesis.

Our data here demonstrate that the multistage model of carcinogenesis emphasizes DNA damage and genomic instability as the initiating event, followed by accumulation of genetic and epigenetic changes, in the initiated cell or its progeny. Collectively, these events lead to the emergence of a clonal-dominant cell population. We established the temporal sequence of events in the occurrence of the initiation of premalignant lesions following carcinogen (MNU) exposure. We focused on molecular processes that interfere with mitotic spindle dynamics, centrosome anomalies and higher expression levels of the centrosome-associated kinase Aurora A. As a tool to study more advanced stages of malignant progression, tumors were transplanted into syngeneic hosts. Cytogenetic analyses of mammary tumors demonstrated that marker chromosomes were present in mostly aneuploid metaphases with hyperdiploid numbers of chromosomes.

AMV, age-matched virgin; DCIS, ductal carcinoma in situ; DH, ductal hyperplasia; MEC, mammary epithelial cells; MNU, 1-methyl-1-nitrosourea; PAC, papillary cystic adenocarcinoma; TG, tumor generation.

Material and methods

Animals

Female Wistar-Furth (WF) rats, 50 days old, were purchased from Harlan (Indianapolis, IN). The WF rat is an inbred strain of rats sensitive to both MNU- and DMBA-induced mammary carcinogenesis.17 The animals were maintained in a clean facility in the Center for Comparative Medicine at Baylor College of Medicine ad libitum of a 12/12 h light–dark cycle.

Tumorigenesis studies: Conventional rat mammary carcinoma induction model

In rat mammary carcinogenesis induction protocols 50-days-old virgin female animals are injected with MNU.18 At this age, the rat is experiencing normal estrous cycles and the mammary gland is rapidly growing and undergoing branching morphogenesis in response to hormonal stimulation. In experimental regimen for carcinogenesis studies MNU was prepared,3 and rats were injected with one dose of MNU (50 mg/kg body weight). Tissue was collected at 1, 2, 10 days up to 100 days postcarcinogen. In tumor regimens, animals were palpated weekly for the length of the experiment, which in this study was 46 weeks. Tumors were excised when they were 1–2 cm in diameter. Tissue was digested for mammary epithelial cell (MEC) isolation; a portion of each tissue was fixed for paraffin embedding and whole mount, or frozen at −80°C.

Primary rat MEC culture

The procedure for the culture of rat MECs was modified from the mouse MEC-culture method, previously described.19 Tissue (0.7–1 g) was digested with 0.1% hyaluronidase and 0.15% collagenase in DMEM/F12 containing 1% penicillin/streptomycin. Tissue suspension was washed once with 5% fetal bovine serum (FBS) in PBS, and 3 times with PBS.

To enrich for MECs, cells were grown for 3.5 hr in 75-cm² plastic flasks containing DMEM/F12 (for glands) and RPMI 1640 (for tumor), respectively. Medium was supplemented with 2% FBS, 1% HEPES, 10 μg/ml insulin and 5 ng/ml EGF. After 3.5 hr supernatant was centrifuged, cells were resuspended and transferred to collagen-coated (0.05 mg/ml) flasks, fed with fresh medium after 48 hr and subsequently twice a week. Cultures were treated with dispase (0.4% in HBSS) once a week to enrich for the MEC population and to eliminate fibroblastic growth. To passage, cells were incubated with dispase for 30–45 min at 37°C.

Preparation of mammary fat pads, orthotopic injection and transplantation

The inguinal mammary fat pads (4th pair of mammary anlagen) of female rats were used as the site of transplantation. Fat pads of 20- to 23-days-old females were cleared by a modification of the method described in Ref.20; the fat pad lymph nodes were also removed.

Transplantation of mammary tumors. Twelve MNU-induced primary tumors, each from a different rat, were used for transplantation. Tumors were transplanted as either (i) MECs or as (ii) dissociated tumors.

(i) MECs grown for 2–8 days in culture were dislodged from the plastic flask by incubation with dispase (0.4% in HBSS), washed once with PBS and resuspended in DMEM/F12; 10–15 μl of cell suspension (2.5 × 10⁷–5.0 × 10⁷ cells/ml) were injected into each cleared fat pad. (ii) Alternatively, when dissociated tumors were used for transplantation, 15 μl of digested and washed cells (4 times with PBS), resuspended in DMEM/F12, were injected.

Each tumor was injected into contralateral fat pads in at least 2 syngeneic rats; some glands were injected with PBS as a control. Mammary fat pads were palpated for outgrowth once a week, up to 20 weeks; tumors were allowed to grow until they were at least 1–2 cm in diameter before termination. At this time, samples of tumors were removed for histological evaluation, or snap frozen in liquid nitrogen for further examination. Tumor tissue was digested for MEC culture for serial transplantation and cytogenetic analyses of tumor outgrowths (up to 5 tumor generations (TG)). The remaining fat pads were removed intact and fixed for histological evaluation.

Apoptosis

Paraffin sections were processed according to the manufacturer's instructions using the Apoptag Plus Flourescein In Situ Apoptosis Detection Kit (Chemicon, Billerica, MA) and counterstained with DAPI. The percentage of apoptotic cells was determined from a count of 500 epithelial cells acquired from fluorescent images. (Positive control section was a 3-days involuted mouse mammary gland provided by manufacturer.)

FACS analysis

Sections (50 μm thick) from paraffin embedded tissue were analyzed by FACS.21 Cells were harvested from 4 × 50 μm² paraffin sections and prepared for PI staining. DNA content analysis (Becton-Dickinson: Cycle test plus DNA reagent kit) was performed on a FACS Calibur Flow Cytometer (Becton Dickinson, San Jose, CA). The percentage of cells in various stages of the cell cycle was determined using the ModFitLT software program.

Chromosome preparation and banding analysis

For metaphase chromosome preparations, actively dividing MECs were grown to 70–80% confluency, following previously described procedure. The fixed cell suspension was dropped on the acetone-cleaned, wet slides and allowed to air dry. Some slides were conventionally stained in Giemsa for chromosome counts; others were aged at room temperature for 10–12 days for G-banding.22

G-banding

Optimally aged slides were Giemsa (G)-banded following the procedure described previously.23 Metaphase spreads were captured using a NIKON Eclipse 800 microscope. The metaphase preparation was processed using the Quantitative Image Processing System (Applied Imaging, Santa Clara, CA). Karyotypes were arranged following the guidelines of the Committee for a standardized karyotype of Rattus norvegicus nomenclature.24

Immunohistochemistry

Paraffin-embedded sections were deparaffinized as previously described.25 For antigen retrieval, sections were placed in 2100 Retriever (PickCell Laboratories, Leiden, the Netherlands) according to the manufacturer's protocol.

Peroxidase staining

Staining of sections was performed as described previously.25 The rabbit-anti-human Progesterone Receptor (DAKO, Denmark, A0098) was diluted 1:200, and rabbit-anti-mouse Estrogen Receptor (Santa Cruz Biotechnology, MC-20) was diluted 1:400. For immunodetection of Aurora A the same primary antibody was used as for Western blots (1:1,000). A biotinylated donkey–anti-rabbit secondary antibody (Amersham Biosciences, Buckinghamshire, England) was diluted 1:500 in 10% normal donkey serum.

Immunofluorescence staining for histological sections

Mouse–anti-human Cytokeratin 8 (ARP, Belmont, MA) was diluted 1:50; rabbit–anti-mouse Cytokeratin 14 (Covance, Berkeley, CA), mouse–anti-porcine Vimentin (Santa Cruz Biotechnology, Santa Cruz, CA) and mouse–anti-smooth muscle actin (SMA, Sigma-Aldrich, St. Louis, MO) were diluted 1:1,000. The secondary antibody was a goat–anti-rabbit Alexa Fluor 568 (Molecular Probes, Eugene, OR) or a goat–anti-mouse Alexa Fluor 488 (Molecular Probes, OR) diluted 1:1,600. Slides were mounted using Prolong Gold anti-fade media with DAPI (Molecular Probes, OR).

Immunofluorescence staining of rat MEC cultures

Indirect immunofluorescence techniques were performed as described previously.26 Same primary antibody dilutions as described earlier were used to detect vimentin, K14, K8 and SMA. For double staining mouse–anti-chicken α-tubulin (Sigma-Aldrich, St. Louis, MO) (1:200) and a human antiserum to centrosomes (a gift from Rattner and coworkers27) (1:5,000) were used. Cells were viewed under a deconvolution microscope (Applied Precision, Issaquah, WA).

H&E staining

Mammary gland was dissected, fixed in 4% paraformaldehyde for 3 hr at 4°C, transferred to 70% ethanol at 4°C and embedded in paraffin following standard procedures.28

Whole mount analysis

Glands were dissected and fixed in 10% Formalin for 48 hr. Whole mounts were prepared following protocols described in Ref.25.

Confocal imaging and scoring of centrosomes

Staining, imaging and scoring of centrosomes was performed as previously described.16

Western blot

Mammary gland tissue lysates were prepared following the procedure as previously described.29 Sixty microgram of total protein were loaded for Western blot analysis30 using the anti-Aurora A-antibody (BL656, Bethyl Labs, Montgomery, TX) (1:5,000). Immunospecific protein bands at 50 kDa were normalized against total protein (Ponceau staining) and quantitated using Image Quant.

Results

Conventional rat mammary carcinoma induction model

For studies on early tumorigenesis we used 50-days-old rats. To ascertain any major morphological or biological alterations in mammary cells in the days, weeks and months, following a single MNU treatment, we examined selected sets of developmental characteristics by whole mount preparations and immunohistochemical staining. Glands were dissected 1, 2, 10 days to 60 days post-MNU treatment.

In our model, preneoplastic lesions (i.e., DH) were detected as early as 10-days postcarcinogen, and the majority of mammary carcinomas emerged over a 4-month period (Fig. 1). The incidence of mammary premalignant lesions and adenocarcinomas in rats, as a function of time, postcarcinogen are shown in Figures 1a and 1b. About 10 days, 3% of animals (1/29) were found to have one or more mammary intraductal proliferation (DH) (Fig. 1d), and this increased to 20% at 20-days postcarcinogen, to 55% at 30-days and 83% at 50-days postcarcinogen (Fig. 1a). DCIS (Fig. 1e) were first observed 30-days postcarcinogen. Mammary tumors (in glands no. 1 to no. 4) arose after 7 weeks up to 48 weeks. Most of the tumors were papillary cystic adenocarcinoma (PAC; Fig. 1f); one tumor was a fibroadenoma. The tumor incidence was 84%, 27 of 32 rats developed tumors with a latency of 13–15 weeks (Fig. 1b).

Figure 1. Incidence of mammary premalignant lesions and tumors after exposure to MNU. (a) The highest incidence of premalignant lesions was detected 50-days post-MNU with n = 5; (b) Tumors arose as early as 7 weeks. H&E staining on paraffin-embedded tissue: (c) Gland, 1-day post-MNU, with sloughing apoptotic cells in lumen of duct. Premalignant lesions and tumors were confirmed by subgross morphology: (d) for a DH (20-days post-MNU), (e) for a DCIS (60-days post-MNU) and (f) for a tumor (PAC).

Transplantation of selected mammary tumors

To specify the clonal composition of tumors, MECs and dissociated tumor tissue, respectively, were serially transplanted into cleared fat pads and analyzed. The 12 primary tumors that provided the transplants were mammary adenocarcinomas (PAC), taken between 15 and 48 weeks postcarcinogen (Fig. 1). For orthotopic transplantation, isolated MECs were grown under selective culture conditions to enrich for epithelial cells and to deselect for fibroblasts (see Material and Methods). Characterization of the cell type was of particular importance for the transplantation experiments. Before transplantation, cells were characterized to assure that only epithelial cells were used for the orthotopic injection. To identify and determine the relative content of each cell type in culture, immunofluorescence was performed using the cell markers vimentin, K-14, K-8 and SMA. Fibroblasts and mesenchymal cells were positive for vimentin, but not for K14, K8 and SMA; epithelial cells stained positive for K8; basal carcinoma cells stained positive for K14 and vimentin but not SMA (Fig. 2, lower panels). There was a correlation between the cell type and the growth rate; epithelial cells grew slower compared to fibroblasts and mesenchymal cells.

Figure 2. Immunofluorescence of tumor tissues using specific antibodies to distinguish the tumor type (upper panels). Staining was performed on paraffin sections using anti-cytokeratin 8 (K8) (green), -cytokeratin 14 (K14) (red), -smooth muscle actin (SMA) (green) and -vimentin antibodies (green). Panels depict positive and negative cross-reaction to antibody. The H&E stain of each tumor type is shown in the panels on the left. Lower panels: Immunofluorescence of mammary cells in culture using specific antibodies (K8 and K14 in red, and vimentin in green) to distinguish the cell type. Left panels depict cells growing in culture dish with typical growth morphology. Right panels depict positive stained cells for the particular antibody used.

Among the 9 of 12 transplanted tumors that grew out, the first TG developed into PAC (1/9), 1 grew out as a normal gland (1/9), 1 as a DH (1/9), 1 as a DCIS (1/9), 4 transplants grew out as mesenchymal tumors (4/9) and 1 as a basal carcinoma (1/9). The tumor type was confirmed morphologically by H&E staining on paraffin-embedded tissue. In cases where morphology was nondistinctive, immunofluorescence staining was performed using vimentin, K14, K8 and SMA antibodies on paraffin-embedded tissue (Fig. 2). PACs were positive for K8, K14, SMA and vimentin. Basal carcinomas were positive for K14 and vimentin but not for SMA and K8. Mesenchymal tumors were positive for vimentin and in some areas for SMA but not for K14 and K8 (Fig. 2, upper panels). We also investigated the status of ER and PR. In general, PACs were positive for both while in mesenchymal tumor, ER and PR were not expressed; basal carcinomas lost their hormone-dependency and were negative for ER and PR.

Subsequent serial transplants (TG2 to TG5), with dissociated tumors instead of cultured MECs, gave rise to mesenchymal tumors in all cases. The latencies for palpability ranged from 1 to 12 weeks. The outgrowths, scored from stained whole mounts of mammary fat pad varied in size and morphology resembling normal tissue in their ductal and ductal-alveolar appearance. Hyperplasia and DCIS were best visible when whole-mounted glands were paraffin-embedded and stained with H&E (Fig. 3). Tumors were palpable, some grew without forming outgrowths, especially in the case of mesenchymal tumors. Tumor latency was shorter with each TG and in two cases, tumor cells metastasized into the lung giving rise to mesenchymal tumors.

Figure 3. Whole mount image of an outgrowth after injecting MECs from a primary tumor (46-weeks postcarcinogen) into the orthotopic site of syngeneic rats. The image depicts hyperplastic outgrowth on the left, tumor outgrowth on the right, and normal outgrowth in-between. Morphologically these areas were identified as DCIS (left lower panel) and PAC (see right lower panel) after the tissue was paraffin-embedded, sectioned and stained for H&E.

Centrosome anomalies in glands, tumors and cells

Earlier, we demonstrated that centrosome anomalies coincide with tumorigenesis.16 To elucidate how early during carcinogenesis in this rat mammary model centrosome alterations occur, we performed IF confocal microscopy on frozen glands and tumors, and scored nuclei with corresponding centrosomes. From each time point (1, 2, 10 days to 60 days post-MNU) 3 glands were evaluated for their centrosome numbers (n = 3) (Fig. 4a). Counting mean values of 300 cells/gland demonstrated that, as early as 1 and 2 days after MNU exposure, centrosome numbers are elevated (>10% of cells) compared to untreated control (below 8%); 20- and 40-days post-MNU, 18% of cells showed abnormal centrosome numbers.

Figure 4. Centrosome amplification during early carcinogenesis. (a) MNU-treated glands and (b) tumors demonstrate significant changes compared to control (AMV). AMV mean values are as follows: 50 days (6%), 70 days (5%) and 96 days (6%). Cells with more than 2 centrosomes (amplified) were increased in treated glands, mean values are as follows: 1 day (13%), 2 days (11%), 10 days (9%), 20 days (18%), 30 days (11%), 40 days (17%), 50 days and 60 days (15%) (panel a); in tumors, 9-weeks post-MNU (14%), 11 weeks (18%), 15 weeks (17%) and 46 weeks (14%) (panel b). Differences among the three groups were statistically significant (p < 0.001, t-test), and results are expressed as mean ± standard error of the mean. (c) Electron micrographs of tumor cells demonstrating centrosome anomalies. One centrosome usually consists of a pair of centrioles. One pair of centrioles is oriented at right angles (left panel). In mammary tumors centrioles were detected that missed the typical structure (right panel, arrow); centrioles were more condensed and showed aggregates of fibrogranular material. (d) Deconvolution images showing MECs with aberrant centrosome numbers, stained for microtubules (green), centrosomes (red), and DNA (blue). The mitotic cell (left) contains several centrosomes (at least 4) that give rise to a multipolar spindle. The anaphase cell (right) will generate 3 daughter cells instead of the normal 2 after cell division is completed leading to aneuploidy.

Tumors showed elevated abnormal centrosome numbers ranging from 13 to 15% (n = 3) (Fig. 4b) that confirmed our earlier findings. Many centrosomes were irregular in shape, exposing smaller or larger fluorescent dots compared to control (70-days virgin). In electron micrographs, centrosomes appeared normal throughout early carcinogenesis (10- to 60-days post-MNU). However, in mammary tumors (PACs) that arose after 15 weeks postcarcinogen, we observed unusual centrosome-associated structures (Fig. 4c). These masses were juxtaposed to the parent centrioles and were not found in normal tissue. These data suggest that mammary tissue after MNU exposure exhibits abnormalities in centrosome size, number and morphology in up to 18% of the cells.

Staining of MECs for centrosomes, microtubules and DNA confirmed the data from our above described centrosome counts. In culture, MECs exhibited abnormal centrosome numbers leading to multipolar spindle formation that, in turn, gave rise to abnormal mitoses (Fig. 4d). Cells that progressed through mitosis, illustrated here in an anaphase cell, displayed more than 2 daughter cells (Fig. 4d).

Exposure to MNU changes the expression pattern of the centrosome-associated kinase Aurora A

Earlier we reported that mRNA levels of Aurora A increase during tumorigenesis.16 Here, we were interested in Aurora A protein expression levels during early carcinogenesis. In Western blot analyses using an Aurora A antibody, expression levels in tissues from rats treated with MNU for 10–60 days were compared. Each treatment set represents 3 glands from 3 different animals. Figure 5 depicts the positive bands at 50 kDa that represent Aurora A expression in glands normalized to total protein (Ponceau). Levels were very variable between animals and exposure time (10- to 40-days post-MNU) with values below or above control (see standard deviation) suggesting an association with the early premalignant changes in glands occurring at this time. Aurora A levels were significantly higher than control (50-days age-matched virgin (AMV); 3.08 with SD 0.09) after MNU exposure at 50 days (3.53 with SD 0.14) and 60 days (3.48 with SD 0.15) (with p < 0.03) with an incidence of premalignant lesions of more than 60% (Fig. 5b).

Figure 5. Upregulation of Aurora A kinase during early tumorigenesis. (a) Western blot analysis of Aurora A kinase in mammary tissue lysates from 3 rats exposed to MNU (0–60 days). Ponceau staining was used as a loading control of total protein. (b) Quantification of the 50 kDa band demonstrated that levels are significantly increased by 50- and 60-days post-MNU (p < 0.03). Spatial distribution of Aurora A in (c) control tissue, in (d) DH and (e) DCIS with mean values of 72 (±3.3)%, 76 (±17.8)% and 93 (±8.3)% of positive cells, respectively.

To investigate the spatial distribution of Aurora A expression in the topology of the gland we used the anti-Aurora A-antibody in IHC on paraffin sections. Spatially, Aurora A was found to be localized in epithelial cells of alveoli and ducts, in stroma cells and myoepithelial cells (Fig. 5c). Comparing Aurora A expression in tissues harboring premalignant lesions (DH and DCIS) with control (50-days AMV) demonstrated elevated Aurora A levels in DCIS confirming that Aurora A expression levels are increased during early tumorigenesis (Figs. 5d and 5e).

Centrosome anomalies and Aurora A overexpression can coincide with errors occurring during cell cycle and cell division, and are therefore a good indicator for genomic instability and aneuploidy. Published data reported that these primary tumors are thought to exhibit a low frequency of LOH (loss of heterozygocity) and no apparent genetic instability.31, 32 Because of the inconsistency of our results to published data it was necessary to further investigate the ploidy and chromosome composition of these tumor samples.

Ploidy

The carcinogen-induced mammary glands were assessed for their aneuploid frequency by FACS analyses. As expected, normal mammary glands (50, 70 and 96 days virgin) exhibited a normal diploid frequency with a DNA-index of 1 (DI = 1). Our findings demonstrate that after MNU exposure epithelial cells were highly aneuploid; after 1-day post-MNU mammary glands exhibited 52% aneuploid cells (Table I). During the preneoplastic phase, 10- to 60-days post-MNU, mammary cells were detected with DI-indexes between 1 and 1.6, with 14/16 samples containing cells that were genomically instable; the mean percentages of aneuploid cells ranged from 35 up to 82% (Table I). In some glands, a second peak, occurring in the G2 phase of the cell cycle, was found with 9–33% aneuploid cells (Table I). However, we also found a small proportion of mammary gland samples with a normal diploid DNA distribution with an index of 1.00 (DI = 1). Results from the tumor samples demonstrated that tumors were diploid (DI = 1), near diploid (DI = 1.2) or aneuploid (DI = 1.4) with a fraction of aneuploid cells up to 82% (Table I). These data indicate that glands and tumors were frequently highly aneuploid.

Table I. Ploidy Analyses of Mammary Glands and Tumors

Regimen	Number of animals	Morphology	DNA index in G1/S FACS	DNA index in G2 FACS	% of aneuploid cells first peak FACS	% of aneuploid cells second peak FACS
The number of animals, morphology, DNA-index and % of aneuploid cells (FACS) in mammary glands (1d to 60d post-MNU), and tumors (7 to 46 wks post-MNU) are shown. Listed are groups of control animal, 1 and 2d, 20 to 60d, and 11 to 46 wks post-MNU describing a normal morphology, premalignant lesions in glands, or tumors, respectively. Values are mean of all experiments, with the low and high means of individual experiments presented in parentheses. (DI = channel of aneuploid/channel of diploid cells).
Untreated	2	Normal	1
1 day, 2 days post-MNU	6	Lots of alveoli, vacuoles, with apoptotic cells	1.30 (1.2–1.35)	1.65 (n = 1)	52 (22–94)	33 (n = 1)
10–60 days post-MNU	1	No detectable lesions	1
	3	No detectable lesions	1.4
	8	DH and papillary hyperplasia DCIS	1.6 (1.2–1.74)		51 (4–92)
	3		1.5 (1.3–1.75)	1.7 (n = 1)	35 (22–46)	9 (n = 1)
	2	Papillarycystic adenocarcinoma	1.2 (1.0–1.4)		82
11–46 wks post-MNU	1	Papillary hyperplasia	1
	1	Papillary DCIS	1.5		24
	5	Papillarycystic adenocarcinoma	1
	5	Papillarycy stic adenocarcinoma	1.4 (1.1–1.85)		41 (7–78)

Levels of apoptosis were low

One possible mechanism that could explain how cells manage a stage of high genomic instability and aneuploidy leading to diploid or near diploid tumors, as proposed by others,10, 33 is apoptosis. Two glands of each time point (0- to 60-days post-MNU) were analyzed by TUNEL assay. The detected levels of apoptosis ranged from 0.2 to 1.26% (60-days post-MNU) that is not very different to control levels of 0.8% (50-days AMV). In histological sections, especially 1-day post-MNU, ducts were found with sloughing apoptotic cells accumulating in the lumen (Fig. 1c). However, the low values of apoptosis detected after 10–60 days of carcinogen treatment could not account for the changes from highly aneuploid cells converting into diploid or near diploid tumors. According to FACS analyses, 5 out of 10 tumors showed normal DI-indexes; in the other 5 tumors DI values were above normal (DI = 1.4) with 82% aneuploid cells.

Cytogenetic analyses revealed nonclonal and clonal tumorigenic origin

Because ploidy data produced by FACS analyses were variable, we decided to cytogenetically analyze MECs of glands and tumors. For cytogenetic analyses metaphases were G-banded.

Out of a total of 65 tumors, 6 tumors and 5 glands were chosen for karyotyping. The tumors were PAC with a latency of 11–20 weeks. The contralateral glands of sample G1, 70-days virgin, and samples G2 and G3, 6 and 8.5 weeks post-MNU, respectively, were without detectable lesions; G4 and G5, 11 and 14 weeks post-MNU, respectively, both harbored premalignant lesions (DH). The percent cells with aberrations, range of chromosome distribution, modal number of chromosomes and presence or absence of marker chromosome(s) for each sample was investigated (Table II). All of the tumors listed here had chromosomal aberrations, expressed in the range of chromosomal number or in the presence of marker chromosomes. Markers present in gland G5 were nonclonal, whereas markers present in tumor T3 were clonal. Clonality was determined if identical markers were present in more than one metaphase spread.34 All 4 tumor samples showed cells with aberrations (8–18%) compared to control G1 (Table II).

Table II. Cytogenetic Analyses of Mammary Glands and Tumors

Sample #	Tumor generation	Regimen	Histology	Number of cells analyzed	% cells with aberrations	Range	Modal number	Marker (s)	Marker chromosomes
The histology, number of cells analyzed, % of cells with aberrations, the range and modal number of chromosomes, and the presence of marker chromosomes are shown. Group G describes control or premalignant lesions in glands, and Group T describes primary tumors that arose between 14 and 42 wks. Group TP describes tumors that derived from transplantation of primary tumors (T5 and T6, respectively).
G1		70-d virgin	gland	25	4	42–43	42	Absent
G2		6-wks post-NMU	gland	25	44	37–42	42	Absent
G3		8.5-wks post-NMU	gland	25	48	37–42	42	Absent
G4		11-wks post-NMU	gland, DH	17	17.6	42–43	42	Absent
G5		14-wks post-NMU	gland, DH	18	15.7	36–44	42	Present	m = t(5q; 9q); m = t(6q;8q) nonclonal
T1		14-wks post-NMU	PAC	68	14.7	41–84	42	Absent
T2		14-wks post-NMU	PAC	74	18	32–44	42	Absent
T3		11-wks post-NMU	PAC	10	15.1	32–44	42	Present	m = t(3q;8q) clonal
T4		14-wks post-NMU	PAC	74	8.1	39–84	42	Absent
T5	TG0	42-wks post-NMU	PAC	13	92	38–45	42	Present	M1 = t(1q;2q); M2 = del(2q); clonal
TP1	TG1	7-wks transplant	mesench. tum	37	100	39–55	52	Present	M1 = t(1q;2q); M2 = del(2q);M3 = iso(4q) clonal
TP2	TG2	7-wks transplant	mesench. tum	31	100	40–103	52	Present	M1 = t(1q;2q); M2 = del(2q); M3 = iso(4q) clonal
T6	TG0	28-wks post-NMU	PAC	22	100	31–101	60, 71	Absent
TP3	TG1	20-wks transplant	basal carcinoma	20	100	48–109	60	Present	M1, M2 clonal
TP4	TG2	8-wks transplant	mesench. tum	23	100	35–85	42	Present	M1, M2 clonal

Comparing the results of FACS analyses with the cytogenetic analyses demonstrated several discrepancies in ploidy. For example tumors T2 and T4 were diploid by FACS with an DI-index of 1; however, cytogenetic analysis demonstrated 8 and 18% of cells with aberrations, respectively, with a range of chromosomes up to 84.

To examine the clonal relationship of tumors, different generations of transplanted tumors were analyzed. Tumor T5, a PAC, with a latency of 15 weeks was taken 42 weeks after MNU injection; tumor T6, a PAC, with a latency of 20 weeks was taken 28 weeks after MNU injection. These two tumors were serially transplanted into cleared fat pad of syngeneic hosts and cytogenetic analysis was performed. Tumor T5 was serially transplanted 5 times. Tumor generation 0 (TG0) was a PAC, 42 weeks post-MNU treatment. Tumor generation 1 (tumor TP1) transformed to a mesenchymal tumor after transplantation (Fig. 2), TG2 (tumor TP2) to TG5 were transplanted as mesenchymal tumors. Tumor T6 was serially transplanted 5 times; TG0 was a PAC, 28 weeks post-MNU treatment. Tumor generation 1 (tumor TP3) transformed to a basal carcinoma after transplantation (Fig. 2); TG2 transformed into a mesenchymal tumor (tumor TP4), and TG3 to TG5 were transplanted as mesenchymal tumors. The chromosome numbers varied from 35 to 109 in different cell samples, with a modal number ranging from 42 to 71 (Table II). With exception of the parental tumor T6, all cell lines showed characteristic marker chromosomes. Since some of these markers were shared by all 6 cell lines, they are numbered serially. All marker chromosomes found were clonal. A tentative identification of these markers are as follows: M1= t(1q; 2q), M2= del(2q), M3 = t(4q;4q) (Fig. 6a). Karyotypic analysis of the in vivo passages allowed confirmation of the data from chromosomal counting, indicating that specific trisomies started to develop mainly after subsequent in vivo transplantations (Fig. 6b), and marker chromosomes of TP3 and TP4 (Fig. 6c). There was no clear relationship between the histopathological pattern of the tumor and the karyotypic abnormalities observed.

Figure 6. Cytogenetic Analyses of the banded and unbanded chromosomes of a primary tumor (induced by MNU) and of different TGs, derived from in vivo passaging of primary tumor 1 (T5) and primary tumor 2 (T6). (a) Metaphase spreads from primary tumor T5 (left) and its tumor generation 1 (TP1) (right) demonstrating the presence of marker chromosomes M1, and M1 and M3, respectively. (b) A Giemsa-banded karyotype of tumor TP2, TG2 of tumor T5, with 3 marker chromosomes M1, M2 and M3 and variable copy numbers of different autosomes. (c) In tumor TP3 and TP4, TG1 and 2 of the primary tumor T6, the same marker chromosomes, M1 and M2, were present (see text for the identification of marker chromosomes).

In conclusion, mammary tumors induced by MNU were genomically instable. Serial transplantation of tumor cells into cleared fat pads demonstrated that tumors clonally expand as confirmed by the presence of marker chromosomes. Aneuploidy is carried over from one TG to the next; with each in vivo passage, tumors accumulated more chromosomal aberrations, also reflected in the aggressiveness with which the tumors grew out in the rats.

Discussion

Using the rat mammary gland as a model for tumorigenesis, it was possible to recapitulate the early stages of tumorigenesis occurring in a time frame of 1–100 days after treatment with MNU. An early event was the damage of cells and their DNA, captured 1 day after MNU exposure. The alkylating agent MNU causes DNA-damage; initial genomic instability was accompanied by a population of cells which was highly aneuploid. Cells tolerate only a certain degree of damage and undergo cell death if the damage cannot be repaired; as a result, during the first 48 hr, a portion of cells underwent cell death. Morphologically, glands treated with MNU for 1 and 2 days depicted proximal and distal ducts with sloughing apoptotic, luminal epithelial cells. Evidence for that comes also from Anderson et al., who demonstrated that the percentage of labeled mitosis in rat mammary glands exposed to MNU for 24 hr decreased significantly indicating that MNU is either immediately cytotoxic or that DNA synthesis is temporarily halted.35 Thereafter, cells seem to be able to cope with the MNU insult and survive. These events are coinciding with amplified centrosomes as early as 1-day postcarcinogen. The rapid increase of centrosomes in response to DNA damage might be associated with abortive or delayed mitosis,36 or an increase in cyclin E level and Cdk2 activity.37 There is accumulating evidence that the centrosome might be part of the DNA damage response network, with Chk1 and Chk2, and the centrosomal kinases Plk1 and Plk3, inducing cell cycle arrest or apoptosis.38 Complementary findings from two research groups suggest that the DNA damage checkpoint is activated in a wide variety of human preneoplastic and neoplastic lesions promoting tumor progression and genetic instability.39, 40 In practice, centrosome amplification causes highly variable outcomes of mitosis: some cells partition chromosomes equally, others mis-segregate one or a few chromosomes and some fail cleavage.41, 42 Here, centrosome amplification initially leading to the loss of spindle bipolarity induced the formation of multipolar spindles to regulate cell division and the distribution of fragmented chromosomes into new daughter cells. The kinase Aurora A, associated with centrosomes, was expressed higher during these early changes (50-days post-MNU), and levels remained high throughout tumorigenesis. In MNU-induced glands, as early as 10 days and throughout 100-days post-MNU, DNA-indexes were greater than one, indicating that glands carry an aneuploid cell population. In these aneuploid cells, through events of rearrangement and redistribution of chromosomes we postulate that cells gained a higher growth potential responsible for the induction of tumorigenesis. Increased levels of ploidy coincided with centrosome amplification peaking at 20-days postcarcinogen, and with preneoplastic development peaking at 50-days postcarcinogen. Glands developed DH, DCIS and adenocarcinoma. The tumor incidence was highest between 13 and 15 weeks. These data strongly indicate that genomic instability and centrosome aberration precede the development of preneoplastic lesions, and hence, are a potential cause of tumorigenesis.

According to published data, primary rat tumors are thought to exhibit a low frequency of LOH and apparent genetic instability. Here, cytogenetic analyses using metaphase spreads and G-banding proved to be sensitive and detected specific chromosomal aberrations.

In our model, in all tumors investigated, cytogenetic analyses clearly demonstrate that tumor cells are aneuploid, and structural alterations of chromatid, iso-chromatid breaks and translocations involving chromosome 1, 2 and 4 are detectable. The serial transplantation experiments emphasize that primary tumors carried chromosomal aberrations that are passed on to the next generations. Change of tumor phenotype is rather a secondary event which might be due to transplantation techniques used.

Further evidence came from serial transplants, performed on 12 primary MNU-induced adenocarcinomas. Despite the similarities among tumors induced by MNU, there were inherent differences in their transplantability and fat-pad growth patterns. Cellular heterogeneity may account in part for the diversity in outgrowth morphology as well as in the proliferative potential of selected tumors. It is possible that the nontumorous outgrowths were a reflection of the heterogeneous cell composition of the tumors, and therefore the result of proliferation of preexisting normal and hyperplastic cells. Transplants derived from primary MECs and from enzyme-dissociated tissue gave rise to DH, DCIS and tumors (PAC, basal carcinoma and mesenchymal tumors). The frequency of outgrowths that carried mesenchymal tumors was relatively high. In all transplantation experiments, tumors of generation 2 (TG2) and beyond generation 2 (TG2–TG5) were of mesenchymal phenotype. The development of anaplastic areas could be observed in all tumor series at some point in the progression, a phenomenon also described by other investigators.43, 44 In Aldaz's study, the highly invasive fibroblastoid (mesenchymal) cells appeared to be derived from the basal-myoepithelial cell compartment because they were positive for the expression of Keratin-14.43

In our model, when cells were injected into the orthotopic site of syngeneic rats, metastasizing cells were observed in lung resulting in mesenchymal tumors in this organ. Advanced tumorigenesis was also reflected in the latency of tumor outgrowths; tertiary tumors (TG2) appeared after 5 weeks, and in generation 5 (TG5) tumor cells grew more aggressively and needed only 1 week to produce an outgrowth of similar size. Results of the karyotypes of tumors and their TGs confirm that in carcinomas during the invasive and metastatic phase chromosomes and marker chromosomes accumulate, leading to highly aneuploid cells (Fig. 6).

In humans and mice, studies have demonstrated a monoclonal composition of advanced neoplasms, the clonal composition of smaller, antecedent lesions revealed that mammary ducts and alveoli are polyclonal and putative early preneoplastic lesions are frequently polyclonal.45 In our model, clonality was tested by serially transplanting mammary tumors. In metaphase spreads of transplanted tumors together with their corresponding TGs, similar chromosomal aberrations were observed. Marker chromosomes were found in many metaphases of the same tumor, and these same marker chromosomes were observed in subsequent transplanted tumors; by definition, these tumor cell populations are clonal. In addition, in subsequent TGs, cells accumulated chromosomes, resulting in a change of the modal number of chromosomes and the number of marker chromosomes. Basal carcinomas lost their hormone-dependent growth and were negative for ER and PR. These events occurred in parallel with tumor aggressiveness during the course of transplantation, as discussed earlier.

The fact that similar marker chromosomes were found in more than one primary tumor is striking and raises the question whether MNU targets “hot spots” in the genome leading to only a subset of chromosomal aberrations. Proof could come from thorough cytogenetic analyses of a greater number of mammary tumors through the identification of marker chromosomes.

In conclusion, our data strongly indicate that changes on the genome level occur rapidly after exposure to MNU preceding preneoplastic transformation. It will be interesting to elucidate changes on the gene level that are most likely responsible for altered proliferation leading to changes in tissue architecture.

Acknowledgements

We gratefully acknowledge Mr. Mohammed Mamdani for perfect technical assistance in procedures of immunohistochemistry, Mr. Himo O. Garricks for expert assistance in histology, Ms. Claire M. Haueter for expert help in EM procedures and Ms. Rebbecca V. Moore for laboratory organization. We are grateful to Dr. Susan G. Hilsenbeck for statistical analyses. Thanks are extended to Drs. Orla M. Coneely, Jeffrey M. Rosen and Bert W. O'Malley for comments on the article.