Deregulated Estrogen Receptor α and p53 Heterozygosity Collaborate in the Development of Mammary Hyperplasia

Breast cancer is the second leading cause of cancer death in women with prevalence increasing with age (1). Breast cancer develops as molecular changes accumulate in the ductal epithelium, giving rise to precursor lesions such as atypical ductal hyperplasias (DH), which may progress to ductal carcinoma in situ (DCIS) and invasive breast cancer (2). Breast cancer is associated with somatic genetic and epigenetic alterations in the breast tissue, such as tumor suppressor gene mutation, or other molecular changes that compromise their function.

The tumor suppressor p53 plays a role in mediating cell response to various stresses by inducing or repressing genes involved in cell cycle arrest, senescence, apoptosis, DNA repair, and angiogenesis (3). Alterations to p53 are commonly detected in primary human breast tumors (4), reported in 30% to 40% of human breast cancers (5) and ∼25% of all preinvasive DCIS lesions (6). Disruption of p53 function may be involved in earlier rather than later stages of breast cancer progression, such as initiation of breast carcinogenesis and impaired differentiation of DCIS (7, 8). Alterations to p53 function include mutation, changes in upstream regulators, transcriptional target genes, and coactivators (9). p53 detection in benign lesions, indicative of possible mutation, is associated with elevated cancer risk (10). In DCIS, p53 is associated with more advanced lesions (11) and is a predictor for local recurrence (12, 13). In cancers, loss or mutation of p53 is correlated with increased aggressiveness, poor prognosis (14), and chemotherapy resistance (15). In addition to p53 somatic mutation in sporadic cancers, germline mutation of one allele of this gene in humans causes an inborn predisposition to cancer known as Li-Fraumeni syndrome (16), where early-onset female breast cancer is the most prevalent tumor type (17).

Hormone receptor status is one of the main differentiating characteristics of human breast cancers and modifies therapeutic response. About 60% to 70% of human breast cancers are estrogen receptor α (ERα) positive and estrogen dependent (18). Increased ERα expression in normal breast epithelium is found in conjunction with breast cancer, leading to the concept that loss of the normal regulatory mechanisms that control expression levels of ERα in normal breast epithelium may increase the risk for the development of breast cancer (19). Increased and deregulated ERα expression in the mammary epithelial cells of transgenic mice (CERM) results in the development of DCIS and increased cell proliferation (20). Expression of ERα is increased 2-fold in the mammary epithelial cells of these mice and is considered deregulated because it is not downregulated by estrogen exposure.

Reproductive history is the strongest and most consistent risk factor outside of genetic background and age (21). Early pregnancy in reproductive life reduces breast cancer lifetime risk in women by up to 50% (22, 23). In mouse models, p53 is required for hormonal protection from mammary tumorigenesis (24). Early exposure to estrogen and progesterone, designed to mimic pregnancy, has been found to enhance p53-dependent responses, increase resistance to carcinogenesis by blocking proliferation of ERα-positive cells (25), and suppress mammary tumor formation in BALB/c-Trp53^+/− mice (26).

Different observations point to potential cross-talk between p53 and ERα. Human breast cancers with p53 mutations are more frequently ER negative (27). In serial transplant studies, the absence of p53 in mammary epithelium is associated with DCIS lesions and invasive cancer that progress from an ERα-positive to ERα-negative state (28, 29). Studies have shown that p53 can regulate ERα expression and transcriptional activity, but both positive and negative effects have been shown (30, 31). ERα can also be regulated at the protein level. c-Src phosphorylation has been shown to stimulate ERα ubiquitylation and proteasome-dependent degradation (32), and p53 has been reported to downregulate some Src functions (33).

The effects of loss of p53 and ERα deregulation on cell proliferation and apoptosis during in vivo carcinogenesis have been previously studied independently. Indeed, loss of p53 activity disrupts apoptosis and accelerates the appearance of tumors (34), and increases cell proliferation levels (35), whereas deregulated ERα increases cell proliferation and the prevalence of DH/DCIS (20). Studies in mouse models have shown that loss of p53 has a different effect in the susceptibility of mammary tumor development depending on the strain. C57BL/6 p53^+/− mice are relatively resistant to mammary tumor development compared with BALB/cJ (36).

The objective of this study was to use genetically engineered mouse models to investigate the effect of the combination of deregulated ERα and p53 haploinsufficiency and compare this to each factor alone in age-dependent mammary preneoplasia development, effect on cell proliferation and apoptosis, expression of regulatory proteins including ERα, and parity protection. Each lesion independently and in combination increased age-dependent development of mammary preneoplasia. Molecular studies revealed that only p53 heterozygosity affected AKT activation and p27 expression, whereas only the combination of deregulated ERα and p53 haploinsufficiency increased extracellular signal-regulated kinase 1/2 (ERK1/2) and c-Src activation in association with decreased expression levels of ERα. Parity protected p53 heterozygous mice from developing mammary gland preneoplasia.

Mice carrying a transgene composed of the mouse mammary tumor virus (MMTV)–long terminal repeat linked to sequences encoding the tetracycline-responsive reverse transactivator (rtTA) for “tet-on” gene regulation (37) and a transgene composed of the tetracycline operator (tet-op) promoter linked to sequences encoding murine ERα [MMTV-rtTA/tet-op-ERα (CERM); ref. 38] mice were bred to mice carrying a homozygous p53 mutation (The Jackson Laboratory; ref. 39) to generate experimental CERM, p53^+/−, and compound CERM/p53^+/− mice. Mice were maintained on a C57BL/6 genetic background, and nontransgenic wild-type (WT) littermates obtained from breedings were used as controls. All mice were maintained and euthanized in accordance with institutional and federal guidelines approved by the Georgetown University Animal Care and Use Committee, and complete necropsies were performed at the specified euthanasia time points. MMTV-rtTA, tet-op-ERα transgenes, and normal and disrupted p53 alleles were identified using tail samples (Transnetyx). Mice were maintained on a special diet containing 200 mg of doxycycline per kilogram of food (Bio-Serv) during prenatal life until euthanized to induce transgene expression. Cohorts of nulliparous mice were euthanized at 4, 8, or 12 months of age to test the relationship between age and mammary preneoplasia development [4- and 8-mo time points: WT (n = 10–12), p53^+/− (n = 10), CERM (n = 10–13), and CERM/p53^+/− (n = 10); 12-mo time point: WT (n = 25), p53^+/− (n = 25), CERM (n = 25), and CERM/p53^+/− (n = 25)]. Mortality and morbidity were compared between female nulliparous mice [WT (n = 59), p53^+/− (n = 64), CERM (n = 63), and CERM/p53^+/− (n = 50)], but no statistically significant differences were found. Overall survival to age 12 months was 93% for WT, 84% for p53^+/−, 90% for CERM, and 88% for CERM/p53^+/−. Causes of premature death included euthanasia due to prolapsed anus (WT: n = 1; p53^+/−: n = 1), skin infection (p53^+/−: n = 3; CERM: n = 4), lymphoma (p53^+/−: n = 1), failure to thrive (WT: n = 1; p53^+/−: n = 1; CERM: n = 1; CERM/p53^+/−: n = 2), or found dead (WT: n = 2; p53^+/−: n = 4; CERM: n = 1; CERM/p53^+/−: n = 3). Cohorts of parous mice were mated at 2 months of age, underwent multiple pregnancies (≥3), nursed their pups for 21 days, and were euthanized at ≥10 months of age to test the effects of parity on mammary preneoplasia development [WT (n = 9; mean age, 14.0 ± 1.0 mo), p53^+/− (n = 7; mean age, 13.2 ± 0.9 mo), CERM (n = 9; mean age, 11.4 ± 0.7 mo), and CERM/p53^+/− (n = 8; mean age, 12.3 ± 0.5 mo)].

One inguinal mammary gland from each animal was dissected for whole-mount preparation (40). Whole mounts were examined under ×0.5 and ×4 magnification to evaluate the presence or absence of hyperplastic alveolar nodules (HAN). The other inguinal mammary gland was dissected and fixed in 10% buffered formalin overnight at 4°C and embedded in paraffin using standard techniques. Sections (5 μm) were cut and stained with H&E staining and evaluated for the presence or absence of DH and DCIS under a ×60 magnification. DH/DCIS was defined as mammary ductal epithelium consisting of at least four epithelial cell layers that extended into and either partially or totally obliterated the lumen. Digital photographs were taken using a Nikon Eclipse E800M microscope with Nikon DMX1200 software (Nikon Instruments, Inc.).

Unstained 5-μm tissue inguinal mammary gland sections were used for the detection of ERα, progesterone receptor (PgR), Src, and phospho-Src (p-Src) using the Vectastain ABC kit (Vector Laboratories, Inc.). Detection of Ki67 was accomplished using the Mouse On Mouse (M.O.M) Peroxidase kit (Vector Laboratories). In situ detection of apoptotic cell nuclei was performed using the ApopTag Plus Peroxidase In Situ Apoptosis kit (Millipore). Sections were deparaffinized, rehydrated, and boiled for 20 minutes in citrate buffer (pH 6.0) for antigen retrieval, and endogenous peroxidase activity was inactivated by 10-minute incubation in 3% hydrogen peroxide at room temperature. Tissues were blocked with the appropriate IgG-blocking reagent (10 min), exposed to the appropriate primary antibody (1 h) and to biotinylated anti-rabbit or anti-mouse IgG secondary antibody (30 min), exposed to ABC reagent (30 min), stained with 3,3′-diaminobenzidine chromogen (5 min; Dako), and counterstained with hematoxylin. Primary antibodies included rabbit polyclonal ERα antibody (1:500 dilution; Santa Cruz Biotechnology, Inc.), rabbit polyclonal PgR antibody (1:250 dilution; Santa Cruz Biotechnology), rabbit monoclonal Src (36D10) antibody (1:800 dilution; Cell Signaling Technology), rabbit polyclonal p-Src (Tyr⁴¹⁶) antibody (1:25 dilution; Cell Signaling Technology), and mouse monoclonal Ki67 antibody RTU-Ki67-MM1 (1:100 dilution; Novocastra). Proliferative index (PI) and apoptotic index (AI) were calculated as percentage of mammary epithelial cells showing stained nuclei in the total of at least 1,000 cells per section. The percentage of mammary epithelial cells showing nuclear-localized ERα and PgR were calculated by counting 1,000 cells per mouse. p-Src immunohistochemistry was scored using a combination of the percentage of mammary epithelial cells showing staining and stain intensity as read by two independent observers: 1 = <33% cells stained at low intensity; 2 = between 33% and 66% of cells stained at low intensity; 3 = >66% of cells stained at low intensity; 4 = >66% of cells stained at high intensity. Five sections were randomly selected from each experimental and control group for staining. Negative control slides, in which primary antibody was omitted, were analyzed in parallel. In the absence of primary antibody, no nuclear-specific staining was observed for ERα, PgR, and Ki67 and no cell-specific staining for p-Src and Src.

Dissected thoracic mammary gland was homogenized in radioimmunoprecipitation assay buffer (Cell Signaling Technology) containing 1 mmol/L phenylmethylsulfonyl fluoride plus protease and phosphatase inhibitors (Roche Diagnostics). After plunging the lysates with a syringe needle followed by centrifugation, supernatant protein concentration was measured using the BCA Protein Assay kit (Pierce). Protein samples (60 μg per sample) were resolved by using 4% to 12% gradient SDS-PAGE (Invitrogen) and transferred onto nitrocellulose membranes (Hybond ECL, GE Healthcare Bio-Sciences). Unless noted otherwise, all primary antibodies were from Cell Signaling Technology and used at a concentration of 1:1,000. Primary antibodies included phospho-p44/42 mitogen-activated protein kinase (MAPK; E10), p44/42 MAPK, phospho-AKT (D9E), AKT (C67E7), and p27 Kip1. p53 (FL-393) and actin (I-19) were obtained from Santa Cruz Biotechnology. Secondary antibodies included rabbit anti-goat and anti-mouse (Santa Cruz Biotechnology) and anti-rabbit (GE Healthcare Bio-Sciences). Four independent samples were randomly selected from each experimental and control group for Western blot analysis.

Total RNA was isolated by using Trizol reagent (Invitrogen) from thoracic mammary gland tissue snap frozen at the time of necropsy, quantified on a spectrophotometer, and 2 μg of total RNA were used to prepare cDNA by a reverse transcriptase reaction. Semiquantitative reverse transcription-PCR (RT-PCR) was performed by using REDTaq DNA polymerase (Sigma-Aldrich). Cycle numbers were tested for each primer combination to determine the cycle number when the reaction became saturated. The following primers were used for amplification: tet-op-ERα, CCACACCAGCCACCACCTTC (forward) and CCACTTCAGCACATTCCTTA (reverse); ERα, GACCAGATGGTCAGTGCCTT (forward) and GACCAGATGGTCAGTGCCTT (reverse); PgR b, GGTCCCCCTTGCTTGCA (forward) and CAGGACCGAGGAAAAAGCAG (reverse); PgR a+b, GGTGGGCCTTCCTAACGAG (forward) and GACCACATCAGGCTCAATGCT (reverse); p53, GGGACAGCCAAGTCTGTTATG (forward) and GGAGTCTTCCAGTGTGATGAT (reverse); actin, ATCGTGGGCCGCCCTAGGCA (forward) and TGGCCTTAGGGTTCAGAGGG (reverse). PCR gels were visualized by using the UV transilluminator FBTI-816 (Fisher Scientific), and images were captured using Kodak EDAS 290 (Kodak) and analyzed using Kodak 1D LE 3.6 and Scion Image (Scion Corp.) software. Four independent samples were randomly selected from each experimental and control group for RNA analysis.

Statistical differences among groups were analyzed with Fisher's exact for HAN and DH/DCIS prevalence; t test for ERα, PgR, Ki67, and ApopTag levels; and Mann-Whitney for p-Src scoring using GraphPad Prism version 4.03 for Windows (GraphPad Software). Mean data are presented as mean ± SE. Significance was assigned at P ≤ 0.05.

To test the relationship between age and mammary preneoplasia development secondary to p53 deficiency and/or deregulated ERα expression, HAN (Fig. 1A) and DH/DCIS (Fig. 1B) prevalence was compared in mammary glands from nulliparous mice ages 4, 8, and 12 months. At 4 months of age, none of the mice had developed HANs, but by 8 months of age, at least 20% of the mice from all three experimental genotypes showed HANs (Fig. 1C). By 12 months of age, HAN prevalence was significantly increased in CERM/p53^+/− mice compared with the 4- and 8-month time points [71% versus 0% (P ≤ 0.0001) and 71% versus 20% (P ≤ 0.008), respectively]. By 12 months of age, HAN prevalence was significantly increased in p53^+/− compared with the 4-month time point (35% versus 0%; P = 0.04), with a more modest increase compared with the 8-month time point (35% versus 20%). By 12 months of age, HAN prevalence increased to 32% in CERM mice compared with 30% at 8 months and 0% at 4 months. No WT mice showed HANs at either 4 or 8 months, but by 12 months of age, HAN prevalence increased to 8%. DH/DCIS appeared earlier and, at 4 months of age, was found in all three experimental groups CERM/p53^+/− (50%), CERM (23%), and p53^+/− (40%) but not in WT mice (Fig. 1C). Prevalence rates remained approximately the same at 8 months of age in all genotypes. Like HAN prevalence, by 12 months of age, DH/DCIS prevalence was increased in CERM/p53^+/− mice compared with 8 months of age (60% versus 40%). WT mice showed a 4% DH/DCIS prevalence rate at 12 months of age. In summary, whereas all mice showed an increase in HAN and DH/DCIS prevalence with age, the combination of deregulated ERα and p53 deficiency showed the highest prevalence and most significant increases in age-related preneoplasia development compared with either deregulated ERα or p53 deficiency alone.

To test if changes in the rates of either cell proliferation or apoptotic cell death were correlated with the higher prevalence of preneoplasia in the CERM/p53^+/− mice, PI and AI were compared between genotypes at 12 months of age (Fig. 2A–C). The combination of deregulated ERα and p53 deficiency showed the highest PI (41 ± 1%) compared with 24 ± 1%, 23 ± 3%, and 0.4 ± 0.3, respectively, in CERM, p53^+/−, and WT mice (all P ≤ 0.0004). The AI was lowest in CERM/p53^+/− (0.06 ± 0.04%) mice, but all three experimental genotypes (CERM: 0.19 ± 0.08%; p53^+/−: 0.12 ± 0.06%) were significantly lower than WT mice (0.66 ± 0.06%; all P ≤ 0.0009). In summary, whereas all three experimental genotypes showed abnormal apoptotic and proliferation rates, the mice with both deregulated ERα and p53 deficiency showed the most abnormal parameters.

Experiments were performed to test if changes in proliferation- or apoptosis-related genes were associated with the altered proliferation and apoptotic rates. At 12 months of age, all three experimental genotypes showed increased levels of ERK1/2 activation; however, only mice with p53 heterozygosity (CERM/p53^+/− and p53^+/−) showed increased AKT activation and decreased p27 expression (Fig. 2D). Experiments were also performed to compare levels of p53, ERα, and PgR, as an ERα downstream gene target, in the experimental and WT mice. Reduced levels of p53 expression at both protein and RNA levels were confirmed in p53 heterozygous mice (Figs. 2D and 3C). The mean percentage of mammary epithelial cells showing nuclear-localized ERα was significantly increased in CERM (18.30 ± 0.08%) compared with p53^+/− (8.8 ± 0.2%) and WT mice (8.3 ± 0.9%; both P ≤ 0.0001; Fig. 3A and B). Whereas the mean percentage of nuclear-localized ERα was significantly decreased in CERM/p53^+/− (6.9 ± 0.3%) compared with CERM mice (P ≤ 0.0001), nuclear-localized PgR was increased in both CERM/p53^+/− (16.5 ± 1.8%) and CERM (13.9 ± 0.4%) compared with p53^+/− (9.1 ± 0.1%) and control (9.7 ± 1.0%) mice (all P ≤ 0.02; Fig. 3A and B). The DH/DCIS lesions observed in CERM/p53^+/− stained positive for nuclear-localized ERα. To evaluate if the decrease in nuclear-localized ERα expression occurred at the protein level in the CERM/p53^+/− mice or was associated with decreased RNA levels, RNA from mammary gland tissues was analyzed using semiquantitative RT-PCR. Equivalent levels of transgene-specific ERα were found in CERM and CERM/p53^+/− mice, and levels of ERα, total PgR, and isoform b PgR RNA were not significantly different between genotypes (Fig. 3C). To determine if the decrease in ERα protein in the compound mice was associated with higher p-Src levels, relative levels of p-Src and Src were scored in the different genotypes. p-Src scores were significantly higher in CERM/p53^+/− (3.8 ± 0.3) compared with CERM (1.8 ± 0.3), p53^+/− (1.8 ± 0.5), and WT (1.3 ± 0.3) mice (Fig. 3A; P ≤ 0.05). In summary, the changes in proliferation and apoptotic rates were accompanied by expression changes in apoptosis- and proliferation-related proteins, with changes in p53, phospho-AKT, and p27 associated with p53 heterozygosity. Decreased expression levels of ERα in CERM/p53^+/− mice were associated with increased levels of p-Src expression.

To test if parity affected disease progression, cohorts of CERM/p53^+/−, CERM, p53^+/−, and WT mice were continuously mated beginning at 2 months of age and HAN and DH/DCIS prevalence was compared after serial pregnancies (Fig. 4). Parous p53^+/− mice showed a significant absence of HAN development following serial pregnancy that was not seen in CERM mice (P < 0.03) or other genotype. DH/DCIS prevalence showed a similar trend with the CERM/p53^+/− mice showing no DH/DCIS compared with a 60% prevalence in nulliparous mice (P < 0.004). Parity did not alter the mean percentage of mammary epithelial cells showing nuclear-localized ERα. Like nulliparous mice, ERα expression was significantly increased in parous CERM (13.1 ± 0.6%) compared with CERM/p53^+/− (6.0 ± 0.6%), p53^+/− (6.1 ± 0.6%), and WT (6.7 ± 0.2%) mice (all P ≤ 0.0002; Fig. 4C). In summary, parity was associated with a reduction in preneoplasia only in cohorts where p53 haploinsufficiency played a role.

This in vivo study showed that deregulated ERα and p53 have independent and collaborating roles in the genesis of mammary preneoplasia. Although both deregulated estrogen signaling pathways and loss of p53 function have been previously implicated in breast cancer development, this study showed that the two lesions have additive effects in promoting abnormal mammary epithelial cell growth. Instructive aspects of the study were the focus on p53 heterozygosity compared with complete loss of p53, comparison of two types of proliferative lesions (DH/DCIS and HANs), and investigation of the effect of both age and parity on prevalence of preneoplasia. The fact that a decrease but not complete loss of p53 was sufficient to induce the development of preneoplasia is compatible with the increased risk of breast cancer found in women with Li-Fraumeni syndrome (16, 17). The fact that the prevalence of preneoplasia increased with age is consistent with the increased incidence of breast cancer with age in women (1) and suggests that duration of a molecular lesion and/or cellular changes associated with aging are additional risk factors for preneoplasia development.

Breast epithelial cell homeostasis requires the balance of cell proliferation and apoptosis (41). In this study, both deregulated ERα and p53 heterozygosity independently and in combination altered this balance. The changes were associated with increased activation of ERK1/2. Changes in AKT activation were limited to either the p53 heterozygous mice or the combination. In vitro, estrogen signaling has been associated with activation of both ERK1/2 and AKT (42). These results indicate that in vivo loss of p53 function is more potent than deregulated ERα in inducing AKT activation. Activation of ERK1/2 and AKT is frequently found in different types of transformed cells and may influence chemotherapeutic drug resistance (43).

The cell cycle inhibitor p27 has been referred to as a candidate tumor suppressor (44), and the presence of functional p53 correlates with p27 expression in human breast cancer (45). p27 expression has been documented in human DCIS, but its clinicopathologic significance remains unclear (46). Reduced p27 levels due to protein degradation occur in half of carcinomas (including breast) and correlate with aggressive, high-grade tumors and poor prognosis (47). Low nuclear p27 levels are correlated with c-Src activation in human breast cancers, and in vitro studies have shown that c-Src activation increases p27 phosphorylation, mediating p27 proteolysis and driving cell proliferation (48). Our results showed decreased levels of p27 protein in the p53-deficient mice.

In the normal premenopausal breast, ERα-positive cells comprise 7% of the total epithelial cell population (49) similar to what we observed in WT mice here (8.3%). As reported previously, the mice with deregulated ERα in this study showed an ∼2-fold increase in mammary epithelial cells with nuclear-localized ERα (20). However, the CERM/p53^+/− mice showed a significant decrease in the percentage of mammary epithelial cells with nuclear-localized ERα independent from any detectable changes in RNA expression. Both transcriptional and posttranscriptional mechanisms regulate ERα protein expression levels. Studies show that c-Src cooperates with estrogen to activate ERα proteolysis and that increased c-Src activation leads to a reduced ERα half-life and increased ERα-driven transcription (32). The highest levels of the activated p-Src (Tyr⁴¹⁶) were found with the combination of deregulated ERα and p53 deficiency, whereas expression of the ERα downstream gene PgR was maintained. It is possible that p-Src plays a role in the observed reduction in ERα protein expression in this genotype. Mdm2 and p53 also have been implicated in the regulation of ERα protein degradation (50). Mdm2 overexpression in the absence of p53 can promote estrogen ligand-independent degradation of ERα. It is also possible that the reduction of p53 expression increased availability of Mdm2, leading to increased turnover of ERα.

Studies have shown that pregnancy early in reproductive life confers a 50% reduction in lifetime breast cancer risk (22) and that parity is protective in the BALB/c-Trp53^+/− mice (26). We evaluated the effect of pregnancy on the development of mammary gland preneoplasia in four genotypes of mice on a C57BL/6 background. A noticeable decrease in preneoplasia in p53-deficient mice but not in mice with deregulated ERα alone or WT mice was found, suggesting a possible protective effect of pregnancy in mice with disease initiated by loss of p53 function. This protection may be due to increased activation of p53 signaling through pregnancy (25) that compensated for its reduced expression levels.

As our knowledge of molecular biomarkers increases, risk assessment may include possible “interacting biomarkers” along with specific clinical, genetic, environmental, and pathologic factors. Our data showed that deregulation of ERα in the mammary gland and reduced p53 expression are independent but additive genetic determinants of increased risk that can be affected by both age and parity and suggest that these factors could function together as an “interacting biomarker.” These mouse models could be used to test possible preventive strategies that could mimic protective hormonal changes that occur during pregnancy with follow-up at the molecular level to identify critical signaling events that might serve as markers of response in women. The fact that preneoplasia was not prevented by parity in either the mice with deregulated ERα alone or the WT mice suggests that not all risk factors associated with increased breast cancer risk can be reduced by pregnancy. This is consistent with the appearance of human breast cancer in both parous and nonparous women.

No potential conflicts of interest were disclosed.

These studies were conducted in part using the Lombardi Comprehensive Cancer Center Histopathology and Tissue Shared Resource, and the Animal Shared Resource Core Facilities. We thank Massod Rahimi for scientific discussions and technical assistance.

Grant Support: National Cancer Institute, NIH grant 1R01CA112176 (P.A. Furth); National Cancer Institute “Research Supplements to Promote Diversity in Health-Related Research” (E.S. Díaz-Cruz); and Susan G. Komen for the Cure Postdoctoral Fellowship grant KG080359 (E.S. Díaz-Cruz).

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.