|Home | About | Journals | Submit | Contact Us | Français|
Epidemiological studies indicate that parity enhances HER2/ErbB2/Neu-induced breast tumorigenesis. Furthermore, recent studies using multiparous, ErbB2/Neu-overexpressing mouse mammary tumor virus (MMTV-Neu) mice have shown that parity induces a population of cells that are targeted for ErbB2/Neu-induced transformation. Although parity accelerates mammary tumorigenesis, the pattern of tumor development in multiparous MMTV-Neu mice remains stochastic, suggesting that additional events are required for ErbB2/Neu to cause mammary tumors. Whether such events are genetic in nature or reflective of the dynamic hormonal control of the gland that occurs with pregnancy remains unclear. We postulated that young age at pregnancy initiation or chronic trophic maintenance of mammary epithelial cells might provide a cellular environment that significantly increases susceptibility to ErbB2/Neu-induced tumorigenesis. MMTV-Neu mice that were maintained pregnant or lactating beginning at 3 weeks of age demonstrated accelerated tumorigenesis, but this process was still stochastic, indicating that early pregnancy does not provide the requisite events of tumorigenesis. However, bitransgenic mice that were generated by breeding MMTV-Neu mice with a luteinizing hormone-overexpressing mouse model of ovarian hyperstimulation developed multifocal mammary tumors in an accelerated, synchronous manner compared to virgin MMTV-Neu animals. This synchrony of tumor development in the bitransgenic mice suggests that trophic maintenance of the mammary gland provides the additional events required for tumor formation and maintains the population of cells that are targeted by ErbB2/Neu for transformation. Both the synchrony of tumor appearance and the ability to characterize a window of commitment by ovariectomy/palpation studies permitted microarray analysis to evaluate changes in gene expression over a defined timeline that spans the progression from normal to preneoplastic mammary tissue. These approaches led to identification of several candidate genes whose expression changes in the mammary gland with commitment to ErbB2/Neu-induced tumorigenesis, suggesting that they may either be regulated by ErbB2/Neu and/or contribute to tumor formation.
Although it is well established that pregnancy and lactation provide long-term protection against breast cancer, epidemiological studies also indicate that risk for this disease is transiently elevated with parity (Kelsey et al., 1993; Lambe et al., 1994; Robertson et al., 1997). There is a 15-year period of increased risk postpregnancy that peaks 5 years after parturition in uniparous women and 3 years after delivery in biparous women compared to nulliparous women (Liu et al., 2002). This increased parity-associated risk suggests that the hormones that are elevated during pregnancy can adversely affect the mammary gland, possibly promoting growth of cells that have already undergone malignant transformation. Alternatively, epidemiological studies suggest that the hormonal milieu of the mammary gland during pregnancy and lactation promotes acquisition of a cell population that is particularly susceptible to transformation by HER2, also known as ErbB2 or Neu. HER2 expression is associated with increased parity-induced risk of breast cancer in women (Reed et al., 2003), and women that have been pregnant and have breastfed acquire a 4.2-fold increased risk of developing HER2-positive breast cancers than their counterparts that never breastfed (Treurniet et al., 1992). The mechanisms responsible for the increased risk of developing HER2-positive breast cancer following pregnancy and lactation remain to be elucidated.
Mice harboring an oncogenic allele of HER2/Neu in the germ-line have revealed that transformation requires the formation of a susceptible mammary epithelial cell population (Andrechek et al., 2004). A further report by Wagner and co-workers (Henry et al., 2004) suggested that induction of susceptible mammary epithelial cells during pregnancy and retention of these cells following involution may account for the parity-associated acceleration of ErbB2/Neu mammary tumorigenesis. This later study involved the use of the mouse mammary tumor virus (MMTV)-Neu mouse model of ErbB2/Neu-induced breast cancer, which overexpresses the proto-oncogenic form of rat ErbB2, Neu, under control of the MMTV promoter. These transgenic mice acquire mammary tumors with long latency and eventually develop pulmonary metastases (Guy et al., 1992). The mammary tumors are estrogen-receptor negative (Wu et al., 2002), solid, nodular lesions composed of intermediate cells that histologically resemble a subset of human breast tumors (Cardiff et al., 2000). Although parity accelerates mammary tumorigenesis in MMTV-Neu mice compared to their nonparous littermates, tumor development remains stochastic in nature (Guy et al., 1992; Anisimov et al., 2003; Henry et al., 2004).
Breast tumorigenesis is a multistep process, as reflected by the stochastic kinetics of tumor development (Beckmann et al., 1997; Hanahan and Weinberg, 2000). The stochastic nature of tumor development in multiparous, MMTV-Neu mice suggests that mechanisms above and beyond overexpression of ErbB2/Neu and the stimulus of pregnancy hormones are required to cause malignant transformation of mammary epithelial cells. What are these additional events or ‘hits’ that contribute to ErbB2/Neu-induced tumorigenesis and can they be attributed to factors other than random genetic events? We postulated that the timing of pregnancy may significantly impact the formation of tumor-susceptible cells because it has previously been shown that terminal end buds, which form during puberty, are particularly susceptible to oncogenic insults (Singletary et al., 1991; Russo and Russo, 1996). Alternatively, we surmised that chronic hormonal maintenance, rather than broad fluctuations that occur during the brief mouse pregnancy, may provide a milieu that contributes more significantly to susceptibility.
To examine these possibilities, we carried out two independent tumor palpation studies. To assess the effects of timing of pregnancy on ErbB2/Neu-mediated tumor development, we maintained MMTV-Neu female mice in a pregnant or lactating state beginning at 3 weeks of age. This correlates with the onset of puberty when terminal end buds are just beginning to form (Hennighausen and Robinson, 2001). To determine whether chronic trophic maintenance of the mammary epithelial cells would modulate ErbB2/Neu-initiated tumorigenesis, we bred the MMTV-Neu mice with a model of ovarian hyperstimulation (LH-overexpressing mice; Risma et al., 1995). The LH-overexpressing mice have elevated serum levels of estradiol and progesterone by 5 weeks of age. This results in an ovary-dependent hyperproliferative mammary gland phenotype reflective of a mid-pregnant gland at both the morphological and molecular level (Milliken et al., 2002). In addition to permitting an assessment of chronic trophism on ErbB2/Neu-induced tumor susceptibility, the bitransgenic mice generated by these breedings allowed us to identify critical time points for evaluation of molecular events that contribute to early events of the ErbB2/Neu-mediated tumorigenic cascade.
During puberty, the developing mammary gland is particularly susceptible to carcinogenesis (Tokunaga et al., 1979; Hancock et al., 1993; Russo and Russo, 1998; Land et al., 2003). As adult pregnancy increases susceptibility to ErbB2/Neu-induced tumorigenesis, we postulated that chronic pregnancy/lactation initiated at puberty may provide a further increase in the susceptible cell population for ErbB2/Neu-induced transformation, possibly leading to synchronous tumor formation. To determine whether timing of pregnancy affects ErbB2/ Neu-mediated tumorigenesis, MMTV-Neu female mice were superovulated and then continuously housed with male mice to maintain the females in a pregnant or lactating state beginning at 3 weeks of age. Following weekly palpation to detect developing mammary tumors, tumor latency in the multiparous, superovulated MMTV-Neu mice was compared to that of virgin MMTV-Neu mice. The multiparous, superovulated MMTV-Neu mice demonstrated accelerated tumor development (Figure 1, 25.1 ± 5.0 weeks) compared to virgin MMTV-Neu mice (34.0 ± 10.1 weeks). The acceleration of tumorigenesis by early pregnancy is consistent with the window of increased vulnerability during early reproductive age identified by chemical carcinogenic studies in rodent models (reviewed in Russo and Russo, 1998) and human epidemiological studies (Tokunaga et al., 1979; Hancock et al., 1993; Land et al., 2003). However, tumor development remained somewhat stochastic within the population, suggesting that other rate limiting steps are necessary for the formation of an ErbB2/Neu tumor. We speculated that such events may simply involve sustained trophism of the gland as opposed to the vast hormonal fluctuations that occur during repeated pregnancy and lactation.
To determine whether chronic trophic maintenance of mammary epithelial cells would provide the necessary support to bypass the rate limiting, stochastic step in tumor formation, we generated bitransgenic mice by breeding the MMTV-Neu mice with a model of ovarian hyperstimulation (LH-overexpressing mice), which have elevated circulating estradiol, progesterone, and prolactin, display a chronic mid-pregnancy-like mammary gland phenotype, and ultimately develop pathologically diverse mammary tumors at a late age (Milliken et al., 2002). Bi- and single-transgenic mice were assessed for mammary tumor development by weekly palpation. Tumors were 3–4 mm in diameter upon palpation and confirmed histologically. Tumor development in the bitransgenic mice was accelerated (Figure 2a, 17.7 ± 2.0 weeks) compared to virgin (34.1 ± 10.1 weeks) or multiparous, superovulated MMTV-Neu mice (25.1 ± 5.0 weeks) or LH-overexpressing mice (43.0 ± 7.4 weeks). The bitransgenic mice also had an increase in the multiplicity of tumors, yielding a greater tumor burden compared to the MMTV-Neu single transgenics (Figure 2b) The increased multiplicity and short range of tumor development (s.d. = 2.0 weeks) indicates that tumors are forming in the bitransgenic mice in a much more synchronous manner. These results suggest that trophic maintenance of the mammary gland is sufficient to promote ErbB2/Neu-induced tumorigenesis and removes the requirement for additional stochastic insults before tumor development.
It was possible that accelerated tumor development in the bitransgenic mice was due to activation of distinct tumorigenic pathways compared to those that occurred in mice with just the MMTV-Neu transgene. If so, one would expect that either the morphological appearance or the molecular profile of the tumors from these two strains of mice would be distinct. MMTV-Neu mice develop signature solid, nodular tumors composed of intermediate cells (Cardiff and Wellings, 1999; Cardiff et al., 2000). However, more aggressive tumors have been reported for MMTV-Neu mice that were chronically treated with high levels of 17β-estradiol (Yang et al., 2003). To determine whether the bitransgenic mammary tumors were indeed ErbB2/Neu-induced tumors or whether they were derived by alternative, perhaps hormonally-induced, pathways we examined both the histopathology and the molecular profiles of these tumors. Tumor pathology was evaluated according to the guidelines of the Annapolis Pathology Panel (Cardiff et al., 2000). Bitransgenic mice developed solid, nodular mammary tumors that were morphologically identical to the characteristic MMTV-Neu tumors (Figure 3a). Whereas histologically both MMTV-Neu and bitransgenic tumors are well circumscribed, pulmonary metastases provides evidence for the malignant nature of these tumors. The incidence of pulmonary metastases occurring in both MMTV-Neu and bitransgenic mice was highly variable and no obvious difference in the number of metastases was observed between these groups (data not shown). Gene expression profiling provided further evidence that the bitransgenic tumors were very similar to the MMTV-Neu tumors. Affymetrix MGU74Av2 arrays were used to evaluate the gene expression pattern of mammary tumors from bitransgenic, MMTV-Neu, and LH-over-expressing mice. Microarray data were generated for three individual bitransgenic tumors and compared to the microarray data of two tumors from LH-over-expressing mice and five MMTV-Neu tumors that we reported previously (Landis et al., 2005). Hierarchical clustering analysis of this data positioned the bitransgenic tumor samples interspersed among the MMTV-Neu tumor samples within the same arm of the hierarchical dendrogram (Figure 3b), indicating that the bitransgenic tumors are no more different from MMTV-Neu tumors than MMTV-Neu tumors are from each other. The tumors from the LH-overexpressing mice were placed on an entirely separate arm of the hierarchical dendrogram from the ErbB2/Neu-derived tumors, indicating their very distinct molecular signature. Hence, the bitransgenic mice develop characteristic MMTV-Neu mammary tumors, signifying that ErbB2/ Neu-induced mechanisms of tumorigenesis dominate in bitransgenic tumors and that the accelerated tumorigenesis in this model is not due to alternative mechanisms of tumorigenesis induced by the hormonal milieu in the bitransgenic mice.
The hormonally responsive MMTV promoter is induced during late pregnancy and lactation (reviewed in Gunzburg and Salmons, 1992). Thus, accelerated tumor development in bitransgenic mice could simply be due to increased expression of the transgene in these animals. To determine whether elevated transgene expression is responsible for the more synchronous development of mammary tumors in the bitransgenic mice compared to parous MMTV-Neu mice, we examined MMTV-Neu transgene expression by Northern blot analysis of RNA from bitransgenic mammary glands and pregnant MMTV-Neu mammary glands (Figure 4a). MMTV-Neu transgene expression was similar or even lower in the bitransgenic mammary glands compared to the pregnant MMTV-Neu mammary glands. This suggests that higher transgene expression is not responsible for the shift in latency between these hormonally driven states.
In considering alternative mechanisms for enhanced tumorigenesis in bitransgenic mice, we explored the possibility that bitransgenic mice might accumulate mutations in the MMTV-Neu transgene. Activating mutations within the MMTV-Neu transgene have previously been reported by Siegel et al. (1994). These mutations involve a deletion within the juxtamembrane domain and have been reported to occur in ~65% of tumors. This has led to the assumption that tumor formation in the majority of MMTV-Neu mice requires such mutations. To determine if the altered hormonal milieu in bitransgenic mice might promote mutations within the transgene, we assessed tumor mRNA from both bitransgenic and MMTV-Neu mice for presence of these activating mutations. From sequencing (data not shown) and RT–PCR analysis of the MMTV-Neu transgene (Figure 4b), we found that no MMTV-Neu single transgenic tumors contained evidence of somatic mutations in the transgene and only one of four bitransgenic tumors contained a deletion in the MMTV-Neu transgene. These data reveal that mechanisms other than altered transgene expression or somatic mutations within the transgene are responsible for the synchrony of tumor development in the bitransgenic mice. We propose that such mechanisms involve trophic stimulation and maintenance of the susceptible cell population that has previously been identified by Wagner and co-workers (Henry et al., 2004).
To determine how long chronic hormonal stimulation is required for enhancing tumor susceptibility, we performed ovariectomy experiments on the bitransgenic mice to remove hormonal contributions of the ovary and assessed mammary gland morphology as well as tumor development. Mammary glands from bitransgenic mice that were ovariectomized at 8 weeks of age had sustained hyperplasia 2 weeks following the surgery (Figure 5). This suggests that irreversible events have already occurred by this age, maintaining hyperplasia in an ovarian hormone-independent state after only a few weeks of chronic hormonal input. Furthermore, animals ovariectomized at 8-weeks developed mammary tumors more rapidly (25.0 ± 21.8) than the virgin MMTV-Neu mice (34.1 ± 10.0), but delayed compared to sham-operated bitransgenic mice (19.5 ± 2.2; Figure 6). Importantly, tumor development across the cohort regained a stochastic pattern. This further supports the conclusion that chronic trophism by ovarian hormones can provide an environment that eliminates the need for numerous random events before tumor formation.
In a similar experiment involving ovariectomy at 5 weeks of age, only 30% of mice developed palpable tumors by 2 years of age (Figure 6). This indicates that the mammary epithelial cells from 5-week-old bitransgenic mice have not yet become transformed at the time of ovariectomy or that additional tumorigenic insults fail to occur in these mice following removal of the ovaries. Furthermore, these ovariectomy/palpation experiments suggest that the bitransgenic mammary gland becomes committed to tumorigenesis between 5 and 8 weeks of age. Attempts to narrow this window further by ovariectomizing mice at 6 and 7 weeks of age generated intermediate tumor curves between the 5- and 8-week-ovariectomy tumor curves (data not shown). From these data, we conclude that only a 3-week window of trophic support (i.e. from 5 to 8 weeks of age) is required for commitment to ErbB2/Neu-induced mammary tumorigenesis. Defining this window permitted the identification of two temporally distinct physiological stages of the bitransgenic mammary gland: noncommitted and committed to tumorigenesis.
Although significant progress has been made regarding understanding the processes of late stages of breast tumorigenesis and characterization of tumor types, mechanisms underlying earlier steps ranging from normal to preneoplastic and ultimately to overt tumors are not well understood. The window of commitment to ErbB2/Neu-induced tumorigenesis that we identified herein provided two critical time points in which to examine molecular changes that occur in the progression from normal (before 5-weeks) to preneoplastic (after 8-weeks) mammary glands.
To identify genes that may facilitate early steps of ErbB2/Neu-mediated mammary tumorigenesis, we performed comparative microarray analysis of 5- and 10-week bitransgenic mammary glands from ovary-intact mice in triplicate. Ten week glands were used as the preneoplastic time point because preliminary experiments with glands from 8-week-old mice failed to detect a significant number of expression changes. This is probably due to the small size of the preneoplastic cell population at 8 weeks of age. Importantly, 10 week glands do not contain overt tumors as determined by whole-mount analysis (data not shown). We analysed three pooled RNA samples that represented three animals for each time point so that a total of nine bitransgenic animals were analysed per time point. From this analysis, 2793 of 45 101 analysed probe sets were identified as changed according to the Affymetrix change call parameter in all of the 10-week samples compared to the 5-week bitransgenic samples. This data has been submitted to GEO omnibus (http://www.ncbi.nlm.nih.gov/geo).
We compared the list of genes that changed expression during the transition from noncommitted to committed mammary tissue in this report with those identified in our previous study of ErbB2/Neu-expressing preneoplastic mammary glands and tumors (Landis et al., 2005) and identified the expression changes that are consistent between the two approaches. Previously, we identified an ErbB2/Neu-induced mammary tumor molecular signature by comparing the partial transcriptomes of tumors from MMTV-Neu mice to those of mammary glands from age-matched, wild-type animals (Landis et al., 2005). Of the 324 genes contained in the molecular signature for ErbB2/Neu-induced mammary tumors, 119 were also changed in the comparison of 10-week committed mammary glands vs the 5-week noncommitted glands reported herein (Supplementary Table 1), indicating that these genes are altered during early events of tumorigenesis and maintained in overt tumors. We, also previously identified genes whose expression was altered in preneoplastic mammary gland tissue and retained in the ErbB2/Neu tumor signature. In that study, the adjacent mammary gland that housed, but did not directly contact, a palpable ErbB2/Neu-induced mammary tumor in MMTV-Neu mice was used to represent preneoplastic mammary tissue. Of the 82 preneoplasia genes identified previously, 32 were observed in the current comparison of committed (10-week-old) vs noncommitted (5-week-old) mammary glands (Table 1). Identification of these genes by two independent approaches using different mouse models strongly supports the notion that these genes may be transcriptional targets of ErbB2/Neu and/or that they contribute to early neoplastic events of ErbB2/ Neu-induced mammary tumorigenesis.
Several (Idb2, Tpd52, Ghr, Ppp2r) of the 32 genes identified by both approaches have previously been associated with human breast cancer (Calin et al., 2000; Gebre-Medhin et al., 2001; Boutros et al., 2004; Stighall et al., 2005), corroborating this method for identifying genes involved in early stages of mammary gland tumorigenesis. In addition, 44% of the genes identified by this approach have known roles in cellular metabolism. Acyl-CoA synthetase (Acsl4) expression is elevated in both hepatocellular carcinoma and colon adenocarcinoma and has been shown to block apoptosis and promote colon carcinogenesis (Cao et al., 2000, 2001; Sung et al., 2003; Kurokawa et al., 2004; Liang et al., 2005). The downregulation of several genes involved in fatty acid oxidation (Decr1, Adipor2), electron transfer (Etfb, Nr1h3), triglyceride synthesis (Dgat1), diversion from glycolysis (Pgm2), and inhibition of insulin-like growth factor (Igfbp6) may be reflective of the well described alteration of energy metabolism in human tumors (Warburg, 1956; Dutu et al., 1980; Hennipman et al., 1987; Board et al., 1990; Mazurek et al., 2002; Macheda et al., 2005). Malignant cells have increased glycolysis, suppression of mitochondrial energy production, increased nucleogenesis, increased de novo fatty acid synthesis with decreased triglyceride production, activated glutaminolysis, and activated serinolysis (Mazurek et al., 2002), which is consistent with the early expression changes we have observed in tissues committed to form an ErbB2/Neu tumor. Moreover, these metabolic changes precede morphological changes in human carcinogenesis (McDermott et al., 1990). The identification of altered expression of these metabolic enzymes in preneoplastic mammary glands substantiates the ability of this microarray approach to identify early events of tumorigenesis. Further investigation of the genes described herein that undergo a change in expression with commitment to tumorigenesis should improve our understanding of early events of ErbB2/ Neu-induced neoplasia.
The data presented herein reveal that trophic maintenance of mammary epithelial cells is sufficient to generate synchronous growth of ErbB2/Neu-mediated mammary tumors, suggesting that hormonal input provides the secondary events necessary for tumor formation in this model. How does chronic trophism cause synchronous tumor formation? Two experiments in this report oppose the presumption that elevated transgene expression is the sole contributing factor. Northern blot analysis indicated that the transgene is expressed at similar levels in the bitransgenic mammary gland compared to pregnant MMTV-Neu glands, yet multiparous MMTV-Neu mice develop tumors in a stochastic manner (Guy et al., 1992; Anisimov et al., 2003; Henry et al., 2004). Furthermore, early pregnancy, which would induce the transgene in the multiparous/ superovulated MMTV-Neu mice at a vulnerable age, accelerated tumorigenesis, but, again, the tumor curve remained stochastic in nature. These results indicate that hormonal stimulation of transgene expression does not provide the requisite mechanisms for synchronous tumor development. Alternatively, Wagner and coworkers have identified a parity-induced target population of cells that are susceptible to ErbB2/Neu transformation (Henry et al., 2004). Since the mammary glands of LH-overexpressing mice are highly similar to mid-pregnancy glands (Milliken et al., 2002), it is likely that a similar susceptible cell population is induced and maintained in the hormonal environment of these mice. Elevated expression of whey acidic protein, the marker of this ErbB2/Neu-targeted cell population, occurs in the mammary glands of LH-overexpressing mice (Milliken et al., 2002), further corroborating this supposition. Determining whether this parity-induced population of cells exists in LH-overexpressing mice is the subject of ongoing studies.
In conclusion, we have found that chronic trophic input to the mammary gland is sufficient to convert the stochastic pattern of ErbB2/Neu-induced tumorigenesis in virgin mice to a more rapid and synchronous pattern. This indicates that secondary events required for ErbB2/ Neu-induced tumor development may take the form of hormonal stimulation rather than complex multistep genetic events that are followed by natural selection. Hormonal stimulation for just 3 weeks is sufficient for acceleration of tumor formation. However, chronic stimulation is required for the apparent synchronous formation of tumors. This suggests that in addition to pregnancy hormones inducing a susceptible cell population as described by Henry et al. (2004), chronic maintenance of this cell population contributes substantially to overt tumor susceptibility.
Radiolabeled nucleotides were purchased from Perkin-Elmer Life Sciences (Boston, MA, USA). All chemicals were purchased from Sigma (St Louis, MO, USA). Primers were synthesized by Genosys (The Woodlands, TX, USA).
All animals were housed in microisolator plus units under pathogen-free conditions with a 12-h light/dark cycle. Food and water were provided ad libitum. Mice harboring the MMTV-Neu transgene (FVB/N-Tg(MMTVneu)202 Mul/J) (Guy et al., 1992) were purchased from Jackson Laboratories (Bar Harbor, ME, USA) and bred with luteinizing hormone (LH)-over-expressing mice (Risma et al., 1995) to generate a colony of bitransgenic mice. The bitransgenic mice were hemizygous for the MMTV-Neu transgene. The LH-overexpressing mice (CF-1 genetic background) were bred at least four generations into the MMTV-Neu (FVB/N) strain. The genotypes of the mice were determined by PCR amplification of tail DNA with transgene-specific primers as previously reported (Kero et al., 2000; Landis et al., 2005). Mice were palpated weekly to detect mammary tumors. By caliper assessment, tumors were approximately 3–4 mm in diameter when detected. Graphs of tumor data were generated and statistically evaluated using Kaplan–Meier survival analysis. Most mice that developed tumors were euthanized to isolate tissues following palpation. In some cases, mice were kept for 4–6 weeks after initial tumor detection for collection of larger tumors, assessment of multiplicity, and analysis of metastatic progression.
For tumor multiplicity and lung metastastes evaluation, mice with a primary tumor volume ranging from 340 to 1654 mm2 were killed as described above and total primary tumor numbers were counted. Lungs were harvested and fixed overnight in 4% paraformaldehyde at 4°C and held in PBS until processed for paraffin sectioning. Every tenth section of lung was collected and stained hematoxylin and eosin until 10 slides were obtained from each lung. Total number of metastases and pulmonary emboli were counted in each section. All animal studies were approved by the Case Western Reserve University Institutional Animal Care and Use Committee.
To induce early/simultaneous ovulation, pregnant mares’ serum gonadotropin (PMSG) (Sigma, St Louis, MO, USA; 5 i.u. PMSG and 0.05 mg bovine serum albumin per ml sterile saline) followed by human chorionic gonadotropin (5 USP units in sterile saline; Ayerst APL, NY, USA) 48 h later was administered to 3-week-old female mice. The superovulated mice were then housed with male mice and copulation was verified by presence of vaginal plugs. To maintain these mice in a pregnant or lactating state, they were continuously housed with male mice.
Ovariectomy and sham surgeries were performed under avertin anesthesia as described (Milliken et al., 2002). The mice were 5, 6, 7, or 8 weeks of age. Subsequently, mice were either killed by CO2 asphyxiation 2 weeks following surgery to isolate tissues for histological and whole-mount analysis or palpated weekly to detect developing mammary tumors.
All microarray data has been submitted to Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo; Accession # GSE3501) according to Minimum Information About a Microarray Experiment (MIAME) guidelines (Brazma et al., 2001). Microarray analyses were carried out as described (Landis et al., 2005). Briefly, for each mammary gland age group, RNA was isolated from the thoracic mammary glands of nine animals and pooled into three groups, each representing three individual animals. Biotinylated cRNA was synthesized from 10 μg of total RNA and hybridized to the Affymetrix Murine 430v2 GeneChip Arrays (Santa Clara, CA, USA), which contains 45 101 probe sets. For tumor data, total RNA was isolated from three individual tumors and the biotinylated cRNA probes were hybridized to three Affymetrix Murine U74Av2 GeneChip Arrays (Santa Clara, CA, USA), which contains 12 488 probe sets. Computational analyses were performed with Microarray Suite (v.5.0, Affymetrix), Data Mining Tool (DMT v.3.0, Affymetrix), MicroDB (v.3.0, Affymetrix), and GeneSpring (v.6.0, Silicon Genetics) software. For comparison to our previous study (Landis et al., 2005), we identified probe sets that were included on both microarray platforms and determined which of these genes exhibited changes in expression level. Probe sets (10 038) on the MOE430v2 array were represented by probe sets on the Murine U74Av2 according to Affymetrix ‘good match’ criteria (http://www.affymetrix.com/support/technical/). As these studies were performed on different microarray platforms, the fold change values cannot be directly compared.
Northern blot analysis was performed as described (Landis et al., 2005). Briefly, 20 μg of TRIzol purified (Invitrogen, Carlsbad, CA, USA) total RNA was separated by gel electrophoresis on a 1% denaturing agarose gel and then transferred to Hybond-N + Nylon membrane (Amersham Pharmacia Biotech, Piscataway, NJ, USA). Membranes were hybridized with a radio-labeled, double-stranded DNA probe to rat Neu (α-32P-dCTP; DECAprime II, Ambion, Austin, TX, USA) in QuikHyb solution (Stratagene, Cedar Creek, TX, USA) according to the manufacturer’s protocol. The DNA probe was PCR amplified with the Neu transgene primers (Landis et al., 2005).
The synthesis of single-stranded cDNA from tumor samples was performed as previously described (Siegel et al., 1994) with minor modifications. PCR was performed to amplify a 237 base pair region encompassing the area of multiple deletion mutations in the rat Neu cDNA (Siegel et al., 1994) using the following primers: NeuF1842: 5′-GAAACCGGACCTCTCC TACA-3′ and NeuR2079: 5′-CGGATCTTCTGTCTCCTTC G-3′. The PCR cycling conditions were: 95°C for 5 min; (95°C for 1.5 min, 60°C for 2 min, 72°C for 3 min) ×35 cycles, 72°C for 8 min for elongation and held at 4°C until electrophoresis. PCR products were separated on an 8% acrylamide gel, dried, and exposed to film for 18 h.
Whole mounts and histology of abdominal (#4 or #9) mammary glands were prepared as described (Milliken et al., 2002). Briefly, for whole-mount analysis, glands were preserved in Kahle’s fixative, stained with Carmine Alum stain (2% carmine (w/v), 5% aluminum potassium sulfate (w/v) in water) overnight, cleared in xylene, and mounted on glass slides with Permount. For histological analysis, glands were fixed in 4% (w/v) paraformaldehyde/PBS overnight, paraffin-embedded, cut into 5 μm sections, and stained with hematoxylin and eosin.
We extend our gratitude to John Nilson for providing the LH-overexpressing mice and to Kristen Lozada for her dedicated technical support. Histology and microarray hybridization were provided by the core facilities of the CASE Comprehensive Cancer Center (P30-CA43703). This work was supported by a National Institutes of Health Grant (RO1-CA90398, RAK), a USAMRMC Breast Cancer Research Program Predoctoral Traineeship Award (DAMD17-03-1-0302, MDL), and the National Institutes of Health Molecular Therapeutics Training Program (GM08803, MDL).