Aurora A and B Overexpression and Centrosome Amplification in Early Estrogen-Induced Tumor Foci in the Syrian Hamster Kidney: Implications for Chromosomal Instability, Aneuploidy, and Neoplasia

Estrogen-induced Syrian hamster tumors of the kidney have been intensively studied in an effort to gain a better understanding of the role of estrogenic hormones in oncogenic processes (1–4). Importantly, 100% tumor incidence is achieved in animals with a relatively modest elevation in serum estradiol-17β (E₂) concentrations (∼2.3 ± 0.4 nmol/L/mL) and very low E₂ concentrations in renal tissue (4.6 pg/mg protein; ref. 5). These E₂ levels are comparable with those reported in human breast tissue in cycling women (∼5.2–5.8 pg/mg protein; refs. 6, 7).

These E₂-elicited tumors arise in the kidney from ectopic uterine-like germinal stem cells found in the interstitium and primarily located in the corticomedullary region, the earliest region in kidney development (8–10). Moreover, complimentary cytogenetic techniques have clearly shown that both E₂-induced early tumorous lesions and primary tumors of the kidney are highly aneuploid (92–94%; refs. 8, 11). They occur coincident with, and may be downstream consequences of c-myc/MYC overexpression, c-myc amplification, and deregulation of the cell cycle; i.e., overexpressed cyclins D1, E, cdk2 and cdk4, and MDM2 driven by estrogen receptor α (ERα; refs. 12, 13). However, it is presently unclear how E₂-induced deregulation of c-myc and cell cycle entities result in large-scale genomic instability and tumorigenesis.

Recently, it has been proposed that the elevated expression of mitotic kinases, especially Aurora A and B, are crucial events in the cascade leading to centrosome amplification, chromosomal instability, and aneuploidy, thus, driving oncogenesis (14, 15). The Aurora family of serine/threonine kinases is essential for the orderly progression of mitosis (16, 17). Aurora A, the most intensively studied mitotic kinase, localizes to centrosomes, has a critical role in regulating centrosome duplication, maturation, and separation, spindle assembly and stability, as well as chromosome condensation, segregation, and cytokinesis (14–17). Importantly, Aurora A overexpression has been shown to elicit neoplastic transformation in mammalian cells, both in vivo and in vitro (18–20), thus behaving as a bona fide oncogene. It is noteworthy that Aurora kinase (A/B) expression is nominal in resting cells and transiently detectable in normal proliferating cells (21, 22). The levels of Aurora B mRNA and protein expression, like those of Aurora A, peak at G₂-M phase, and their maximum kinase activity is attained at the transition from metaphase to the completion of mitosis (14, 16). Present evidence indicates that Aurora B, considered a passenger protein, targets the centromere at metaphase to phosphorylate histone H3, regulates proper chromosome segregation, and translocates the inner centromere protein (INCENP) and possibly survivin to the spindle mid-zone required for the regulation of cytokinesis. Thus, Aurora B is a key regulator of accurate chromosome segregation via the control of microtubule-kinetochore attachment and cytokinesis (16, 23–25).

In the current study, we show for the first time a concerted and sustained overexpression of both Aurora A and B and coincident centrosome amplification in solely E₂-induced early tumorous foci and in large, well-established tumors in the hamster kidney. These findings further support the concept that E₂-mediated oncogenesis occurs by a common pathway via sustained Aurora kinase overexpression, which in turn, generates amplified centrosomes, chromosomal instability, aneuploidy, and tumor formation.

Animals. Adult castrated male Syrian golden hamsters (LAK/LVG), outbred strain, weighing 90 to 100 g were purchased from Harlan Sprague-Dawley, Inc. (Indianapolis, IN). Hamsters were housed in facilities certified by the American Association for the Accreditation of Laboratory Animal Care. They were acclimated for at least 1 week prior to use, maintained on a 12-h light/12-h dark cycle, fed certified rodent chow (Ralston-Purina 5002), and tap water ad libitum. The animal studies were carried out in adherence to the guidelines established in the “Guide for the Care and Use of Laboratory Animals,” U.S. Department of Health and Human Resources (NIH 1985). Hamsters in the treatment groups were implanted s.c. with 20 mg E₂ pellets as described previously (26). Age-matched control animals were implanted with 20 mg pellets of cholesterol. To maintain constant levels, new pellets were implanted every 3 months. Their mean daily absorption was 96 ± 4 μg/d. The pellets were prepared by Hormone Pellet Press (Shawnee Mission, KS). Over a 6-month period of E₂ treatment, the average E₂ concentration in serum was 2.28 ± 0.43 ng/mL, and 4.57 ± 1.04 pg/mg protein in the kidney (5). Groups of five to eight castrated cholesterol-treated and E₂-treated hamsters were used for 3 to 8 months in most of the studies described. For the withdrawal studies, hamsters bearing tumors treated for 6 months with two 20-mg pellets of E₂, were divided into four groups (n = 3): animals in group 1 were maintained on E₂ pellets; group 2 had their two pellets removed; and those in group 3 had the initial E₂ pellet removed, whereas the second one remained and were additionally implanted with two pellets of 20 mg tamoxifen citrate; group 4 included age-matched cholesterol-treated controls. All animals were killed 10 days after the removal of the E₂ pellets and tamoxifen citrate treatment. Tumors were individually harvested from groups 1 to 3 (and kidneys from group 4) for subsequent analysis of Aurora A expression and centrosome amplification.

Western blot. Kidneys and tumors were homogenized in lysis buffer containing 50 mmol/L of Tris-HCl (pH 7.4), 0.2 mol/L of NaCl, 2 mmol/L of EDTA, 0.5% NP40, 50 mmol/L of NaF, 0.5 mmol/L of Na₃VO₄, 20 mmol/L of Na-pyrophosphate, 1 mmol/L of phenylmethylsulfonyl fluoride, 10 μg/mL of aprotinin, 10 μg/mL of leupeptin, and 1 mmol/L of DTT. The supernatant fractions were collected and their protein content determined with bicinchoninic acid reagents (Pierce, Rockford, IL). Proteins were separated by gel electrophoresis on PAGEr Gold Precast Gels (Cambrex Bio Science Rockland, Inc., Rockland, ME) and transferred to a nitrocellulose membrane. Primary antibodies against Aurora A BL656 (1:1,000; Bethyl Labs, Montgomery, TX), Aurora B ab2254 (1:1,000; Abcam, Cambridge, MA), Centrin MCI (1:25,000; Salisbury, JL, Mayo Clinic, Rochester, MN), and γ-tubulin MMR58 (1:10,000; Salisbury, JL), were incubated overnight at 4°C. Appropriate secondary antibodies were incubated for 2 h and protein expression was visualized with enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ). Densitometry was done using Alpha Imager 2000 (Alpha Innotech, San Leandro, CA).

In vitro protein kinase assay. Proteins were extracted as described for Western blotting. Aliquots, 200 μg of total protein extracts, were incubated for 2 h at room temperature with 1 μg of Aurora A BL469 (Bethyl Labs) followed by a 2-h incubation with 30 μL of A/G agarose slurry. Normal IgG was used as a negative control. Immunoprecipitated complexes were washed several times with PBS and immediately used for kinase assays (Upstate, Inc., Charlottesville, VA) following the manufacturer's standard protocol with modifications. Immunoprecipitates were incubated for 30 min at 30°C with 5× reaction buffer [40 mmol/L MOPS (pH 7.0), 1 mmol/L EDTA], 10 μCi/μL of [³²P]γATP (Amersham Biosciences) diluted in Upstate Mg/ATP cocktail. Reactions were spotted onto P81 paper, subjected to a series of washes in 0.75% phosphoric acid and acetone, and counts per minute measured using a scintillation counter (Beckman Coulter, Fullerton, CA).

Reverse transcriptase PCR and real-time PCR. All reagents for reverse transcriptase and real-time PCR were purchased from Invitrogen Corp., Carlsbad, CA. Total RNA was extracted using the protocol supplied by Life Technologies (Gaithersburg, MD) for Trizol Reagent. Aliquots, ∼100 mg of hamster kidney tissue and E₂-induced primary tumors were immediately frozen in liquid nitrogen and pulverized while still frozen. The pulverized tissue was then placed into 3 mL of Trizol and homogenized using a Polytron. All samples were centrifuged at 12,000 × g for 10 min at 4°C. The supernatant was then transferred to a fresh tube and incubated at room temperature for 5 min before adding 0.2 mL of chloroform/1 mL of Trizol. The tubes were capped, shaken vigorously, and incubated for 3 min at room temperature. The tubes were centrifuged at 12,000 × g for 15 min at 4°C. Total RNA was then precipitated with 0.5 mL of isopropanol/1 mL of Trizol. The tubes were incubated at −80°C for 30 min and then centrifuged at 12,000 × g for 10 min at 4°C. The RNA pellet was washed in 1 mL of 75% ethanol/diethyl pyrocarbonate and centrifuged at 7,500 × g for 5 min at 4°C. After briefly drying, the pellet was resuspended in 100 μL of diethyl pyrocarbonate water and stored at −80°C. RNA integrity was evaluated by agarose gel electrophoresis and the concentration of the RNA was determined by spectrometry.

Reverse transcription was done using 5 μg of total RNA/5 μL. The RNA was combined with 250 ng of random primers, 10 mmol/L of deoxynucleotide triphosphate mix, and sterile distilled water. This mixture was heated at 65°C for 5 min and quick-chilled on ice. 5× first strand buffer, 0.1 mol/L of DTT, and Moloney murine leukemia virus reverse transcriptase was added to each sample and the reverse transcription reaction was done using an iCycler thermocycler (Bio-Rad Laboratories, Hercules, CA). The thermocycling protocol consisted of 1 h incubation at 37°C followed by 10 min incubation at 96°C. The samples were stored at 4°C until used for PCR analysis.

The cDNA synthesized during the reverse transcription reaction was used to perform the real-time PCR reaction. cDNA was amplified using the Platinum SYBR Green qPCR SuperMix UDG and gene-specific primers for β-actin (FWD 5-CAGCCGAGAGGGAAATTGTG-3′; REV 5-TCGTTGCCAATGGTGATGAC-3′; 101 bp amplicon) and Aurora A (FWD 5-TGGGTGTGTGCCTCGAAA; REV 5-GATTGAAGGCCGGATGCA; 102 bp amplicon) in a 7300 Real-time PCR System (Applied Biosystems, Foster City, CA). Myc gene amplification was used as a positive control (FWD 5-TGATGTCGTGTCTGTGGAAAAGA; REV 5-GTTTGCTGTGGCCTCCTGAT; 83 bp amplicon). The baseline was set automatically, as was the threshold Ct value measured during the exponential phase of the amplification. The thermocycling protocol consisted of 2 min incubation at 50°C followed by 10 min incubation at 95°C. Then, the samples underwent 40 cycles at 95°C for 45 s, 60°C for 45 s, and 72°C for 45 s. The specificity of the amplicon was verified by running a dissociation step.

Immunohistochemical analysis. Kidneys were excised, trimmed, and fixed in 5% paraformaldehyde, followed by a rapid paraffin-embedding process. Tissue sections (6 μm) were prepared and dewaxed. Antigens were retrieved (Dako Target Retrieval Solution; Dako, Carpinteria, CA) by heating in a water bath set at 97°C for 20 min or a digital decloaking chamber (Biocare Medical, Concord, CA) and treated with 3% H₂O₂ for 15 min to block endogenous peroxidases. After blocking with 6% of the appropriate serum in 1% bovine serum albumin, the primary antibodies, ERα clone 1D5 (Dako), progesterone receptor (PR) C-19 (Santa Cruz Biotechnology, Santa Cruz, CA), Aurora A BL656 (Bethyl Labs), and Aurora B ab2254 (Abcam), diluted 1:50 (1:1,000 for Aurora A), were applied to the sections overnight at 4°C. Appropriate secondary antibodies were incubated for 1 h at 25°C followed by 1 h with Vector Laboratories Elite ABC (Burlingame, CA). As negative controls, similar tissue sections were incubated replacing the primary antibodies with the appropriate normal serum. The slides were counterstained with hematoxylin, dehydrated in alcohol, and mounted in Permount medium (1:1 permount/xylene) before being examined under the microscope.

Enrichment of amplified tumor centrosomes. The isolation of amplified E₂-induced tumor centrosomes was done according to Moudjou and Bornens (27) with minor modifications. Primary tumors (8–10 g) from the kidney were minced and brought to 50 mL in culture medium containing 10 μL cytochalasin, 10 μL nocodazole, and 50 μL protease inhibitor cocktail, all obtained from Sigma (St. Louis, MO). The minced tumor was incubated for 30 min at 4°C, washed and then resuspended in 0.1× TBS/8% sucrose containing 2 μL of cytochalasin, 2 μL of nocodazole, and 10 μL of protease inhibitor cocktail and pressed through a stainless steel sieve and filtered through a 105-μm nylon mesh monofilament cloth (Small Parts, Inc., Miami Lakes, FL). The resulting cell suspension was lysed in a buffer containing 1 mmol/L of Tris-HCl (pH 8.0), 0.1% 2-mercaptoethanol, 0.5% Triton X-100, and 1× protease inhibitors, homogenized in a glass dounce, and centrifuged at 2,500 × g for 10 min. The lysate was placed in a 50 mL tube, underlaid with 60% sucrose, and centrifuged at 10,000 × g for 30 min. The upper 25 mL fraction was carefully aspirated off, and the remaining sample was vortexed, overlaid onto a vegetable-dyed sucrose gradient (70%, 50%, and 40% sucrose), and centrifuged at 40,000 × g for 1 h. The gradient was fractionated into 25 0.5-mL portions [Fraction Collector and Econo Pump (Bio-Rad Laboratories) Fractionation System (Brandel, Gaithersburg, MD)]. The fraction volumes were brought to 1.5 mL with gradient buffer [10 mmol/L PIPES (pH 7.2), 0.1% Triton X-100 and 0.1% 2-mercaptoethanol], vortexed, and centrifuged at 50,000 × g for 1 h. The supernatant was discarded and the pelleted fractions were stored at −80°C for further analysis.

Centrosome amplification: size and number. Centrosome size and number were analyzed by confocal microscopy of paraffin sections immunolabeled with a monoclonal antibody against the centrosomal protein γ-tubulin (28). Kidney tissues from age-matched control and E₂-treated hamsters for 3, 4, and 8 months were analyzed (n = 3/group). For centrosome evaluation in tumor lesions, tumor foci were first identified in serial sections stained with H&E and PR (C19) from Santa Cruz Biotechnology. Once the foci were identified, subsequent serial sections were immunolabeled with γ-tubulin and Hoechst 33342, to label the centrosomes and the nuclei, respectively (29). Confocal image stacks of seven 5-μm/L-thick slices were captured with a Zeiss 510 confocal microscope using a 63× oil immersion objective. Image stacks were then processed to yield a single maximum intensity projection image. The resulting image was imported into ImageJ (NIH Image) for analysis. The γ-tubulin signal was thresholded consistently for all the images collected for each tissue. The thresholded images were then binarized and the area of each γ-tubulin–labeled spot measured and counted. Cells were counted manually according to the number of nuclei present within an image. On average, 500 cells per tissue category were analyzed; 106 cells was the minimum number analyzed. The size of the epithelial cell centrosomes were normalized against centrosomes in nearby fibroblasts. This normalization process is based on the assumption that the size of fibroblast centrosomes is consistent between the individual animals; therefore, it allows valid comparisons of centrosome size among different tissues in separate animals immunolabeled at various times, and among images captured in separate microscopy sessions. Centrosome size per cell was calculated as the combined region of all γ-tubulin–labeled areas in the image divided by the total number of cells present. Centrosomes in normal tissues are an order of magnitude smaller than nuclei; therefore, this calculation returns a value of >1 because sections through cells pass through the centrosomes less frequently than sections pass through the cell nucleus. In the average cell, the centrosome will appear in only 1/4 of the sections, whereas the nucleus, on average, will appear in essentially all sections.

Statistical analyses. One-way ANOVA with Dunn or Tukey post hoc tests were used for statistical evaluation of the data presented, with the exception of the Aurora A kinase activity and the centrosome size and number that were analyzed using t test. Values were expressed as the mean ± SE. Statistical significance was assumed when P < 0.05 were obtained.

Western blot analysis of Aurora A and B in whole kidneys and tumors during E₂-induced oncogenesis. Protein expression of Aurora A and B kinase was determined in lysates of whole hamster kidney samples after 3, 4, 5, and 6 months, and primary tumors after 6 months of E₂ treatment (Fig. 1A). Although modest increases in Aurora A protein were seen in early E₂ treatment periods, a significant 8.7-fold increase was detected in primary tumors (Fig. 1B) when compared with control cholesterol-treated kidneys. A slower migrating Aurora A band, presumably the phosphorylated form of Aurora A, was observed only in E₂-induced tumors of the kidney. No detectable changes in Aurora B protein expression (Fig. 1A) were seen after 3 to 5 months of E₂ treatment. A slight increase in Aurora B expression was detected in 6-month E₂-treated kidneys. However, a significant 4.6-fold increase in Aurora B expression was seen in primary E₂-induced tumors of the kidney compared with age-matched cholesterol-treated control kidneys (Fig. 1B).

Modulation of Aurora kinase expression in E₂-induced tumors by either estrogen withdrawal or concomitant tamoxifen citrate treatment. Tumor samples from groups of 6-month tumor-bearing animals that were continuously E₂-treated or had either their E₂ pellets withdrawn or concomitantly treated with tamoxifen citrate for 10 days were analyzed for Aurora A and B expression by Western blot. The withdrawal/treatment period was selected because we have previously shown that 72 h are required to completely clear estrogens from the serum of treated animals (5). Compared with age-matched control cholesterol-treated kidneys, an 8.0-fold increase in Aurora A expression was detected in all tumors receiving sustained E₂ treatment (Fig. 1D). Upon 10-day estrogen withdrawal, Aurora A expression was significantly reduced by 78% compared with tumors continuously treated with E₂. Similarly, coadministration with tamoxifen citrate also resulted in a significant 79% decline in Aurora A expression in these tumors. Correspondingly, E₂-elicited Aurora B overexpression also markedly declined when E₂ was withdrawn (81%) or tamoxifen citrate (64%) was coadministered. Cell proliferation, assessed by Ki-67 labeling was not significantly altered in any of the tumor groups (data not shown). Although a significant decline in the expression of Aurora A and B was observed upon E₂ withdrawal and E₂ + tamoxifen citrate treatment, a corresponding reduction in amplified centrosomes was not observed in any of these treatment groups.

Aurora A activity and mRNA levels in E₂-induced tumors. Additionally, the increased overexpression of Aurora A led to a concomitant 2.6-fold increase in its activity as detected using a synthetic construct kemptide as a substrate (Fig. 2A). Additionally, Aurora A mRNA also showed a significant 6.0-fold increase in primary tumors in the kidney compared with age-matched cholesterol-treated control kidneys (Fig. 2B).

Overexpressed Aurora A and B in early tumorous foci. In order to assess the precise cellular location of Aurora A and B expression after various E₂-treatment intervals, serial kidney sections containing early tumorous foci (Fig. 3A) were examined by H&E staining and immunohistochemistry. Serial sections of 3.5-month E₂-treated whole kidneys show that the expression of both Aurora A (Fig. 3B) and B (Fig. 3C) was confined essentially to cells present in early tumorous foci. No positive stained cells were detected in similar tumor foci sections in the absence of either primary antibodies (Fig. 3D). Positive stained Aurora A (Fig. 3E) and B (Fig. 3F) cells were also present in well-established large tumor foci derived from 6-month E₂-treated hamsters, but not in adjacent uninvolved epithelial kidney cells.

Localization of Aurora A and centrosome proteins to isolated amplified tumor centrosomes. Preparations of enriched amplified centrosomes were obtained from tumors induced after 6 to 8 months of E₂ treatment employing a discontinuous sucrose gradient fractionated into 25 aliquots (Fig. 4). Western blot analysis established that fractions 8 to 10 expressed high levels of γ-tubulin and centrin. Both centrosome proteins consistently exhibited the highest level of expression in fraction 9, whereas their expression in other fractions was variable and in reduced amounts. Notably, Aurora A expression was expressed exclusively in fraction 9, indicating that this mitotic kinase was associated solely with the amplified tumor centrosome fraction. On the other hand, Aurora B expression was distributed at least throughout fractions 6 to 10 (data not shown), as expected of a passenger protein kinase.

Centrosome amplification in early estrogen-induced tumorous foci. Small early E₂-induced tumorous foci were first detected in H&E-stained renal serial sections from hamsters treated with E₂ for 3 (Fig. 5A) and 4 months (Fig. 5E). These cell clusters of incipient tumor foci were characterized by their large pale-staining nuclei (Fig. 5A and E), PR-positive staining (Fig. 5B and F), and were not evident in age-matched cholesterol-treated kidneys. E₂-induced tumor foci, at all time periods examined (Fig. 5C,, G, and K), showed a markedly elevated number of centrosomes, which were unevenly distributed in the cytoplasm compared with cholesterol-treated, age-matched, and adjacent uninvolved kidney cells (Fig. 5D,, H, and L). In adjacent uninvolved kidney tissue sections from 3-, 4- and 8-month E₂-treated hamsters (Fig. 5D,, H, L), the pattern of immunofluorescence staining of the centrosomal protein γ-tubulin was confined to the pair of centrioles apical to the nucleus. Similarly, cholesterol-treated age-matched kidney samples showed comparable levels and location of γ-tubulin immunostaining (data not shown). When 3-month E₂-induced tumorous foci were compared with age-matched untreated control kidneys, a significant 1.8-fold increase in centrosome number was observed (Fig. 6). A further significant increase in centrosome number (2.8-fold) was detected after 4 months of E₂ treatment. Well-established tumor foci also exhibited a 1.8-fold increase in centrosome number per cell. When total centrosome area per cell was determined, 2.6-, 5.4-, and 2.5-fold increases were found in 3- and 4-month early tumorous foci, and in large tumor foci after 8 months of E₂-treatment, respectively (Fig. 6). An increase in the number and total volume of centrosomes was indicative of centrosome amplification.

Tumors arising from interstitial uterine-like germinal stem cells in the kidney are driven by estrogen interacting with ERα, and the neoplastic progression occurs in the absence of any discernible histopathologic stages (8). This E-ERα–mediated process is completely prevented by concomitant administration of either tamoxifen citrate, progesterone, or androgen competing with ERα or respective hormone binding to their specific receptors (26, 30, 31). These findings strongly indicate that tumorigenesis is mediated by an ERα-mediated process consistent with a uterine-like tissue behavior because progesterone opposes the action of estrogen in the endometrium.

The detection of high aneuploid frequencies, employing cytogenetic methodologies, nuclear image cytometry, G-banding karyotyping, comparative genomic hybridization, and fluorescence in situ hybridization in early tumorous lesions and in large, well-established tumors (8, 11, 13) have clearly established that chromosomal instability and aneuploidy are early events in solely E₂-induced tumors. Moreover, non-random or consistent whole chromosome gains and losses have been detected in primary tumors with a frequency of 50% to 88%, depending on the specific chromosome of all tumors examined (11). Interestingly, similar high aneuploid frequencies have been detected in E₂-induced ductal carcinoma in situ (DCIS) and primary breast tumors in female ACI rats, and in E₂ plus androgen-induced primary tumors in male Noble rats (32). Taken together, these findings indicate that these molecular alterations represent a common early pathway in E₂-driven oncogenic processes.

Persistent overexpression of mitotic kinases, notably Aurora A and B, has been associated with centrosome amplification, which in turn, is believed to elicit chromosomal instability and aneuploidy; key molecular changes leading to tumor formation (14–17). Nevertheless, no definitive evidence has yet shown that these events are in fact causally related. Persistent Aurora A and B overexpression in concert with centrosome amplification in early tumorous foci and in primary tumors elicited by E₂, however, is consistent with previous findings that these molecular events invariably occur together (21, 33). The early detection of Aurora A and B overexpression and centrosome amplification in 3-month tumorous foci, as reported herein, provides additional evidence that these mitotic kinases may be involved in the incipient events leading to tumor development. The discovery that Aurora A and B are persistently overexpressed in early tumorous foci in the kidney and Aurora A in dysplasias and DCIS in female ACI rats (34), both induced by E₂, suggest that these kinases are under direct or indirect estrogen control. This notion is now supported by our finding that upon estrogen withdrawal or the coadministration of tamoxifen citrate in the presence of E₂, a marked decline in Aurora A expression was found in tumors residing in the kidney compared with corresponding tumors maintained only on E₂.

Not unexpectedly, the relative brief period of E₂ withdrawal, or E₂ + tamoxifen citrate treatment, did not result in a decrease in centrosome amplification in tumors in the kidney, even though Aurora A expression declined significantly. Centrosomes are complex structures, and the γ-tubulin and centrin present in centrosomes are relatively stable proteins. Therefore, the tumor cells likely require many cell cycles in order to deplete these proteins from the centrosome structure. Moreover, when E₂ is withdrawn or blocked, the rate of cell proliferation will eventually decrease, reducing the opportunity for changes in the contribution of these centrosome proteins to centrosome size and number.

Because estrogens are the major causative agent in human sporadic breast cancer, it is not surprising that our findings in two E₂-induced tumor rodent models replicate the high frequencies of Aurora A overexpression (∼94%) and amplified centrosomes (∼80%) reported in both human premalignant DCIS lesions and primary invasive ductal breast cancer (22, 28, 35). The precise relationship between the sustained overexpression of Aurora A and B and the generation of amplified centrosomes leading to multipolar spindle formation and unequal segregation of chromosomes both in E₂-induced early tumorous focal lesions in male Syrian hamster kidneys and in dysplasias/DCIS in female ACI rats remains to be resolved. Combined Aurora A and B phosphorylate more than 20 centrosomal and mitotic protein substrates (21), and the phosphorylation of individual or cassettes of their substrates might affect the deregulation of the centrosome cycle, particularly centrosome duplication.

In summary, we have shown that Aurora A overexpression and activity and Aurora B overexpression are linked to centrosome amplification, leading to chromosome instability, aneuploidy, and tumor development. These critical events occur during incipient stages of E₂-induced oncogenesis and indicate a possible common mechanism whereby estrogens elicit tumor development in target tissues.

Grant support: National Cancer Institute grants RO1 CA102849 (J.J. Li), DAMD 17-01-1-0753, The Breast Cancer Research Foundation (W.L. Lingle), and RO1 CA072836 (J.L. Salisbury).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.