Knockdown of the Transforming Growth Factor-β Type III Receptor Impairs Motility and Invasion of Metastatic Cancer Cells

Transforming growth factor-β (TGF-β) belongs to a family of growth factors known to regulate cell growth and motility (1). TGF-β1 and TGF-β3 can bind with high affinity to the TGF-β type II receptor (TβRII), which results in the recruitment of the type I receptor (TβRI). The constitutively active type II serine-threonine kinase phosphorylates and activates the type I receptor kinase, which results in recruitment and activation of Smads 2 and 3. Smad 4 is able to bind to activated Smad 2/3, and the entire complex translocates to the nucleus to activate downstream target genes (1, 2). In contrast, the type II receptor binds with low affinity to TGF-β2. The type III receptor (TβRIII) or betaglycan binds with high affinity to all three TGF-β isoforms and is required for presenting TGF-β2 to the type II receptor (2).

TGF-β is growth inhibitory in normal cells, but becomes growth promoting during tumorigenesis. A role for the types I and II receptors in tumorigenesis has been well established. It has been reported that the type III receptor is required for TGF-β2–mediated epithelial-to-mesenchymal transition (EMT) in the developing heart (3), which suggests that it may also play a role in stages of cancer progression where EMT occurs. Recent reports using overexpression of TβRIII in MDA-231 breast cancer cells and DU145 prostate cancer cells resulted in decreased tumor invasion, suggesting a tumor suppressor role for this receptor (4, 5). Additionally, a recent report using cDNA microarrays showed that decreased TβRIII expression correlated with higher tumor grade in a cohort of breast cancers (4). Our laboratory has recently published data supporting a tumor suppressor role for TβRIII in nontumorigenic NMuMG mammary epithelial cells that is dependent on the ability of TβRIII to modulate nuclear factor-κB (NF-κB) signaling (6). In contrast, an oncogenic role for TβRIII was suggested by studies in lymphoma, B-chronic lymphocytic leukemia, and certain ovarian tumors (7–9).

NF-κB is a family of homodimeric and heterodimeric transcription factors known to regulate cell growth and survival (10). NF-κB is constitutively active in many types of cancer, including breast (11), colon (11), prostate (12), and ovarian (13), suggesting the importance of this survival pathway in tumorigenesis (14). Several studies suggest a link between the NF-κB and TGF-β signaling pathways. Sovak and colleagues showed that TGF-β–mediated growth inhibition of MCF-7 cells requires a decrease in NF-κB signaling (15). Moreover, secreted fibroblast growth factor-5 (FGF-5) and TGF-β2 induce NF-κB activity, and this autoinduction may result in the constitutive activation of the NF-κB pathway observed in many tumors (16). Monocytes and megakaryocytes from patients with idiopathic myelofibrosis also exhibit constitutive NF-κB activity, which results in an increase in TGF-β1 production (17, 18). These data support a connection between the TGF-β and NF-κB pathways.

In the study reported here, we investigated the role of endogenous TβRIII during tumorigenesis by using a loss-of-function approach. We used short hairpin RNAs (shRNA) directed specifically against TβRIII to determine the role of endogenous TβRIII in breast cancer cells. We found that MDA-231 and polyomavirus middle T oncogene (PMTLuc) cells that have decreased TβRIII expression are less motile and invasive in vitro and in vivo and that this decrease in invasion is dependent on NF-κB signaling.

Cell culture, viral infections, and shRNA. NMuMG and MDA-231 cells were purchased from American Type Culture Collection. The 4T1 cell line was a gift from F. Miller (Karmanos Cancer Center), and the EMT-6 cell line was a gift from B. Teicher (Lilly Research Laboratories). The PMTLuc cells were derived from mammary tumors arising in the MMTV/PyVmT transgenic mice (19). MDA-231 cells were grown in improved modified Eagle's medium (IMEM, Cambrex) supplemented with 10% fetal bovine serum (FBS). 4T1, EMT-6, and PMTLuc cells were grown in DMEM (Cambrex) supplemented with 10% FBS. Phoenix-Ampho and 293A cells were grown in DMEM containing 10% FBS. TβRIII-null mouse embryonic fibroblasts were grown in DMEM (Cambrex) supplemented with 10% FBS. All cells were maintained in a humidified 5% CO₂ incubator at 37°C. Retroviruses expressing shRNAs specific to mouse or human TβRIII were generated as previously described (6). Retrovirus containing mouse or human sh-RIII were used to infect MDA-231 and PMTLuc cells in the presence of 4 μg/mL polybrene. Stably expressing cells were selected with 1 μg/mL puromycin. The pGEM-4Z rat TβRIII plasmid was a gift from Dr. Joan Massague (Memorial Sloan Kettering Cancer Center). Rat TβRIII was digested out of pGEM-4Z using EcoRI and cloned into the EcoRI site of the LZRS-MS-neo retroviral plasmid. 231-kd cells were infected with retroviruses containing rat TBRIII and selected using 600 μg/mL G418. An adenovirus containing constitutively active IKK2 was provided by Dr. Timothy Blackwell (Vanderbilt University). An adenovirus containing green fluorescent protein (GFP) was used as a control. 293A cells were infected with adenovirus to produce a concentrated stock of virions. For adenoviral infection, MDA-231 cells were plated to ∼70% confluency in 100-mm dishes. Adenovirus was added to the cells in 3 mL of serum-free media for 1 h, at which point 7 mL of media containing 10% FBS was added. Cells were allowed to grow for 48 h before being subjected to further treatment.

Antibodies and reagents. TGF-β1 and TGF-β2 were purchased from R&D Systems. Growth factor–reduced Matrigel was purchased from Clontech. Antibodies to Smad 2/3 were from Transduction Laboratories, P-Smad2 and IKK2 from Cell Signaling, and actin from Sigma.

¹²⁵I–TGF-β1 affinity cross-linking assays. ¹²⁵I-TGF-β1 affinity cross-linking assays were performed as described (20). Cross-linked samples were separated on a 3% to 12% SDS-PAGE gradient gel and visualized by autoradiography.

RNA isolation and quantitative PCR. Total RNA was extracted using the RNeasy Mini-kit (Qiagen) per the manufacturer's directions. RNA (5 μg) was reversed transcribed in a 100-μL reaction. Real-time PCR was carried out on 500 ng of cDNA using the iQ SYBR Green Supermix from Bio-Rad in a Bio-Rad iCycler iQ multicolor real-time PCR detection system. Primers were designed using the Universal Probe Library form Roche. Primer sequences were as follows: human TβRI (5′-GTTAAGGCCAAATATTCCAAACA and 5′-ATAATTTTAGCCATCACTCTCAAGG), human TβRII (5′-GATTTCATCTTCGGCTTGAAA and 5′-GCTCAGGAGGAATAGTGTGGA), human TβRIII (5′-GATTTCATCTTCGGCTTGAAA and 5′-GCTCAGGAGGAATAGTGTGGA), mouse TβRIII (5′-CCCCAGATGGTGTGGTTTAC and 5′-TGGCCAGCCACTGCTATC), and actin (5′-GGGGTGTTGAAGGTCTCAAA and 5′-AGAAAATCTGGCACCCC). A standard curve was generated by amplifying known concentration of 231-con cDNA using actin primers. All Ct values were equilibrated to the actin control. All reactions were performed in triplicate.

Cell growth and motility assays. Cells (1 × 10⁴ per well) were seeded in 12-well plates in medium containing 10% FBS. Cells were harvested every other day for the indicated number of days, and cell numbers in each well were measured using a Coulter Counter. Three-dimensional growth assays were carried out in growth factor–reduced Matrigel (BD Biosciences) as described (21). Phase-contrast pictures were taken using an Olympus CK40 microscope. Cells were dissolved from the Matrigel using Cell Recovery Solution (BD Biosciences), and their numbers were measured in a Coulter Counter. For motility assays, confluent sheets of cells were “wounded” by scraping with a pipette tip at time of treatment. Wound closure in the presence of the added ligand was assessed over time, as described previously (22). Transwell assays were performed using BD BioCoat growth factor– reduced Matrigel invasion chambers according to the manufacturer's protocol (BD Biosciences). Briefly, cells (2.5 × 10⁴ per well) were plated in serum-free medium in the upper chamber of an 8-μm pore Matrigel-coated transwell after Matrigel rehydration in DMEM for 2 h. Cells were incubated for 24 h in serum-free media ± 2 ng/mL TGF-β1 or TGF-β2. After 24 h, cells that had migrated to the underside of the transwell filters were fixed and stained using Diff-Quick Stain Set from Dade Behring AG. Cells in five random fields at 200× magnification were counted.

Immunoblotting assays. Cells were plated in 100-mm plates and allowed to grow overnight. Cells were then placed in serum-free media for 16 h, after which media containing either 2 ng/mL TGF-β1 or TGF-β2 was added for 6 h. Cells were lysed in NP40 lysis buffer [20 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1% NP40, 20 mmol/L NaF], as previously described (22). Protein concentrations were determined using the BCA Protein Assay Reagent (Pierce), and 50 μg of protein were separated by 9% SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were blocked in TBST containing 5% bovine serum albumin for 1 h and then incubated with primary antibody overnight at 4°C. This was followed by incubation with secondary antibody for 1 h at room temperature. Membranes were washed thrice in TBST, and bands were visualized using enhanced chemiluminescence (Amersham). Densitometric quantification was performed using the NIH software ImageJ.

Transcriptional assays. Cells were seeded in 60-mm plates and transfected with 5 μg of either 3TPLux or NF-κB-Luc and 0.5 μg pCMV-Renilla (Promega) using Superfect transfection reagent (Qiagen), according to the manufacturer's protocol. The next day, cells were divided equally into 48-well plates and incubated overnight in serum-free media, after which media containing either 2 ng/mL TGF-β1 or TGF-β2 were added for an additional 24 h. Firefly luciferase and Renilla reniformis luciferase activity was measured using the Dual Luciferase Reporter System (Promega) according to the manufacturer's published protocol in a Monolight 3010 luminometer (Analytical Luminescence Laboratory).

ELISA assays. TGF-β1 and TGF-β2 ELISAs were performed according to the manufacturer's protocol (R&D Systems), as described (23). All ELISA data were corrected for cell number and expressed as picogram per milliliter.

Electrophoretic mobility shift assays. Electrophoretic mobility shift assays (EMSA) were performed using the Promega Gel Shift Assay kit (Promega) according to the manufacturer's protocol, as previously described (6). Nuclear extracts were harvested, as described previously (24). Briefly, nuclear extracts (10 μg) were incubated with ³²P-labeled NF-κB oligonucleotides, separated by 6% SDS-PAGE, and visualized by autoradiography. Unlabeled NF-κB oligonucleotides (cold NF-κB) were used as a competitive inhibitor, and unlabeled Oct-1 oligonucleotides were used as a negative control.

Apoptosis assays. Apoptosis assays were performed as described (25). Briefly, cells were plated in 100-mm plates and allowed to grow overnight. The following day, cells were treated with either 2 ng/mL TGF-β1 or TGF-β2 in serum-free media for 72 h or 50 μmol/L etoposide in serum-containing media for 24 h. Adherent and floating cells were collected and fixed in 1% paraformaldehyde before being subjected to terminal deoxynucleotidyl transferase-mediated nick end labeling (TUNEL) analysis with the use of an apo-bromodeoxyuridine assay kit (Phoenix Flow Systems) according to the manufacturer's protocol. TUNEL-positive cells were quantitated in a FACSCalibur Flow Cytometer (BD Biosciences).

Xenograft studies. Cells (1 × 10⁶) were resuspended in 200 μL PBS and injected with a 22-gauge needle into the right inguinal mammary gland (#4) of anesthesized 6-wk-old athymic nude mice (Harlan Sprague-Dawley) and allowed to grow for 8 wk. All animals were kept at five mice per cage in microisolator units and provided with sterile water and chow according to Institutional Animal Care and Use Committee guidelines. Mammary fat pads, tumors, and lungs were collected, fixed in 10% formalin, and embedded in paraffin. Tumor volume in cubic millimeter was calculated by the formula: volume = width² × length / 2 (26).

High expression of TβRIII correlates with breast cancer cell invasiveness. We first examined endogenous TβRIII protein levels in cancer and nontumorigenic breast epithelial cell lines. NMuMG, 4T1, MDA-231, PMTLuc, and EMT-6 cells were affinity cross-linked with TGF-β1–¹²⁵I (Supplementary Fig. S1A and B) to examine TβRIII protein levels. Simultaneously, these cell lines were screened for their ability to form three-dimensional colonies in Matrigel (Supplementary Fig. S1C). The 4T1, EMT-6, MDA-231, and PMTLuc cell lines formed invasive structures in the Matrigel compared with the small acini formed by the nontumorigenic NMuMG cells. Our data indicate that high TβRIII expression correlates with the ability to form invasive structures in Matrigel and suggest that TβRIII correlates with metastatic properties of breast cancer cells.

Decreased TβRIII expression in MDA-231 cells impairs response to TGF-β. MDA-231 cells were infected with retrovirus containing shRNA specific to human TβRIII (231-kd) or mouse TβRIII (231-control). Individual clones were isolated through serial dilution and initially screened for TβRIII expression by semiquantitative reverse transcription–PCR (data not shown). Positive clones were confirmed by receptor affinity cross-linking with ¹²⁵I–TGF-β1. Affinity cross-linking assays showed reduced expression of TβRIII in the cells transduced with TβRIII shRNA (Fig. 1A). TβRIII expression was rescued in the 231-kd cells by infection with a retrovirus encoding rat TβRIII (231-kd&RIII). Decreased expression of TβRIII in the 231-kd cells was confirmed by quantitative PCR (QPCR; Fig. 1B). No significant change in TβRI or TβRII mRNA was detected. Knockdown of TβRIII partially impaired TGF-β–induced phosphorylation of Smad2 (Fig. 1C). Additionally, treatment with TGF-β2 resulted in prolonged phosphorylation of Smad2 in the 231-con cells, but not in the TβRIII knockdown cells (Fig. 1C). Consistent with these data, knockdown of TβRIII eliminated TGF-β–induced transcriptional activity as measured in cells transiently transfected with the 3TPLux reporter (Fig. 1D).

Loss of TβRIII results in decreased motility and invasion. TGF-β can have negative or positive effects on epithelial cell growth (27, 28). Thus, we next examined the effects of reduction of TβRIII on MDA-231 cell growth and motility. The 231-kd cells grew slower than control and parental cells in monolayer (Fig. 2A). This phenotype was abrogated by the reexpression of TβRIII in the 231-kd&RIII cells. Additionally, 231-kd cell migration in a wound closure assay was impaired compared with control cells (Fig. 2B). The addition of TGF-β ligands had no effect on the migration of the 231-kd cells. The ability to grow and invade through growth factor–reduced Matrigel is a surrogate indicator of the metastatic ability of cancer cells. The 231-con cells were invasive through Matrigel-coated transwells (Fig. 2C). This invasion was enhanced in the presence of TGF-β ligands. In contrast, the 231-kd cells were significantly less invasive (Fig. 2C) and were not responsive to added TGF-β2. The 231-con cells also formed invasive structures when grown in Matrigel, whereas the 231-kd cells were not invasive and only formed small acini (Fig. 2D). Invasive growth under these conditions was restored when a rat TβRIII expression vector was reintroduced into the 231-kd cells (Fig. 2D).

MDA-231 cells make their own TGF-β ligands and, as supported by previous reports (20), are under autocrine regulation by endogenous TGF-βs. ELISAs were used to determine if the changes in observed cell behavior could involve changes in the production of autocrine ligands. The 231-kd cells secrete significantly less TGF-β2 compared with control cells. Secreted levels of TGF-β2 were restored upon stable expression of rat TβRIII into the 231-kd cells (Supplementary Fig. S2). These data suggest that the changes in growth seen in the 231-kd cells may at least be partially mediated by a decrease in autocrine TGF-β2 secretion.

To extend these findings in MDA-231 cells to other metastatic lines, we knocked down expression of TβRIII in PMTLuc cells. These cells were derived from mammary tumors arising in the MMTV/PyVmT transgenic mice (19). PMTLuc-con and PMTLuc-kd cell lines were generated as described for the MDA-231 cells (Fig. 1). QPCR was used to confirm TβRIII knockdown in the PMTLuc-kd cells (Fig. 3A). Western blot analysis of p-Smad2 showed a muted response of PMTLuc-kd cells to TGF-β2 (Fig. 3B) In addition, the PMTLuc-kd cells grew slower in a two-dimensional growth assay (Supplementary Fig. S3A) and migrated less in a wound closure assay (Supplementary Fig. S3B) compared with the PMTLuc-con cells. Similar to the 231-kd cells, the PMTLuc-kd cells were markedly less invasive through Matrigel-coated transwells (Fig. 3C) and formed smaller colonies in three-dimensional Matrigel (Fig. 3D).

Loss of type III TGF-β receptor reduces NF-κB activity. NF-κB family transcription factors are known regulators of cell growth and survival (10). NF-κB activity has been associated with EMT and metastasis in breast cancer (29). Furthermore, autocrine TGF-β2 has been shown to activate NF-κB in cancer cells (14, 16, 30). Therefore, we used an NF-κB responsive promoter to examine NF-κB activity in the cells lacking TβRIII. We found that the 231-kd cells had significantly lower basal NF-κB activity compared with the 231-con cells (Fig. 4A). These data were confirmed using EMSA examining NF-κB–DNA binding (Fig. 4B). NF-κB–DNA binding was restored in the 231-kd&RIII cells reconstituted with rat TβRIII. Similarly, the PMTLuc-kd cells also had significantly lower basal NF-κB–DNA binding compared with the PMTLuc-con cells (Fig. 4C). To further confirm these results, we examined NF-κB activity in TβRIII-null, heterozygous, and wild-type mouse embryonic fibroblasts (MEF). Wild-type MEFs had higher basal NF-κB promoter activity, as well as TGF-β2–induced activity, compared with heterozygous and TβRIII-null MEFs (Fig. 4D). QPCR was used to confirm the level of TβRIII mRNA in the MEFs (Fig. 4D).

Loss of TβRIII expression sensitizes cells to apoptosis. Since the TβRIII knockdown cells grew slower and demonstrated decreased ability to form colonies in Matrigel, we performed TUNEL analysis to investigate the levels of apoptosis in these cells. In serum-free conditions, the 231-kd cells exhibited a higher basal level of apoptosis compared with the 231-con cells; the proportion of TUNEL-positive cells was reduced by the addition of TGF-β1, but not TGF-β2 (Fig. 5A), consistent with the effects seen on P-Smad2 (Fig. 1C). When serum-starved, the PMTLuc-kd cells also exhibited a six-fold higher percentage of apoptotic cells and were markedly more sensitive to etoposide-induced cell death compared with control cells (Fig. 5A). These data suggest that TβRIII is involved in cell survival under conditions of cell stress.

Up-regulation of NF-κB activity restores tumor cell invasiveness and reduces apoptosis. We next used an adenovirus containing constitutively active IKK2 (CA IKK2) to restore NF-κB activity in 231-kd cells. Transcriptional assays, using an NF-κB responsive promoter and Western blot analyses were performed to confirm the expression and activity of CA IKK2 (Fig. 5B). 231-kd cells transduced with CA IKK2 formed invasive structures in Matrigel (Fig. 5C), similar to the 231-con cells. Additionally, transduction of CA IKK2 into 231-kd cells decreased the level of apoptosis (Fig. 5D). These results suggest that TβRIII differentially regulates cell motility, invasion, and apoptosis of MDA-231 cells through the modulation of NF-κB activity.

Reduction in TβRIII in MDA-231 cells impairs tumorigenicity in vivo and lung metastases. MDA-231 cells form metastatic tumors when injected in athymic nude mice. Due to the decreased growth and motility seen in the 231-kd cells, we hypothesized that reduced expression of the type III receptor would impair tumorigenicity and metastases in vivo. Cells were injected s.c. into the #4 left inguinal mammary fat pad of nude mice. Mice were sacrificed 8 weeks after tumor cell injection, and their tumors were harvested for further analysis. All of the mice injected with 231-con cells developed palpable tumors, whereas only three of five mice injected with 231-kd cells did so (Fig. 6A). In addition, the tumors that formed in mice injected with the 231-kd cells were significantly smaller than those formed in mice injected with 231-con cells (Fig. 6B). None of the mice injected with 231-kd cells developed distant metastases, whereas three of five mice injected with 231-con cells showed metastases to the lung (Fig. 6C).

It has recently been suggested that the TGF-β type III receptor acts as a tumor suppressor. Several studies have shown that overexpression of TβRIII in breast, prostate, and ovarian cancer cells results in decreased motility and growth (4, 5, 31). Additionally, through the use of cDNA tissue arrays, it has been reported that TβRIII expression is lower in higher-grade breast cancers (4). Sharifi and colleagues found that TβRIII expression was decreased in malignant prostate tissue compared with benign prostate (32). In accordance with these data, we have recently reported that knockdown of the type III TGF-β receptor in nonmetastatic NMuMG cells resulted in an NF-κB–dependent decrease in E-cadherin and an increased ability of cells to form invasive structures in Matrigel and to form tumors in mice (6). In contrast, Woszczyk and colleagues found all three of the TGF-β receptors to have higher expression in high-grade lymphomas compared with low-grade lymphomas (7), suggesting a tumor promoting role for these receptors. A tumorigenic role for TβRIII was also suggested by studies in B-chronic lymphocytic leukemia and certain ovarian sex cord-stromal tumors (8, 9).

Due to this apparent contradiction and to the potential limitations of overexpression studies, we decided to investigate the role of TβRIII in breast cancer using a loss-of-function approach. An initial screen of five breast cancer cell lines suggested that high TβRIII levels correlate with an increase in metastatic ability. To confirm these data, we stably expressed an shRNA specific to TβRIII in two metastatic breast cancer cells lines, MDA-231 and PMTLuc cells (19). We found that down-regulation of TβRIII resulted in decreased cell growth and motility in vitro (Fig. 2 and Supplementary Fig. S3) and in vivo (Fig. 6) and an increase in apoptosis (Fig. 5). In addition, we showed that the decreased invasion in Matrigel and increased apoptosis of these cells was due to decreased NF-κB activity (Fig. 4). This is the first report that addresses a mechanism by which TβRIII modulates cell growth, motility, and apoptosis in tumorigenic breast cancer cells.

These data are counterintuitive to recent reports that suggest a tumor suppressive role for TβRIII. TGF-β signaling can be growth inhibitory or growth promoting, depending on the cellular context. The molecular actions behind this switch have yet to be clearly identified. It is possible that TβRIII may also play differing roles in different cell types, which may be dependent on activation or inhibition of other oncogenes or tumor suppressor genes. For example, oncogenic Ras has been shown to overcome the growth inhibitory effects of TGF-β (33). Furthermore, active Ras, in cooperation with Smad signaling, is required for metastasis (34). This may be relevant to our data because MDA-231 cells express mutant K-ras, and the PMTLuc cells were derived from tumors expressing the polyoma virus middle T antigen.

In this report, we present data demonstrating that TβRIII mediates its effects on cell growth, motility, and apoptosis through NF-κB. We have shown that MDA-231 and PMTLuc cells deficient in TβRIII had lower basal NF-κB signaling. Restoration of NF-κB signaling in the MDA-231 cells was able to restore invasion in Matrigel and decreases the basal level of apoptosis. TβRIII may also interact with other signaling pathways. Lu and colleagues reported that secretion of TGF-β2 and FGF-5 can induce NF-κB activity and that this autocrine induction results in the activation of NF-κB signaling found in many tumors (16). We find that TGF-β2 secretion by the 231-kd cells is decreased compared with the control cells (Supplementary Fig. S2). This decreased autocrine production of TGF-β2 may account for the decreased NF-κB activity seen in these cells.

The significant changes in the production of TGF-β1 and TGF-β2 by the 231-kd cells (Supplementary Fig. S2) may also account for the differences in growth and invasion found in these cells. We cannot rule out the possibility that the decreased cell growth and migration of the 231-kd cells is at least partially due to changes in autocrine ligand production after loss of TβRIII nor can we exclude additional roles of the type III receptor in the modulation of cell behavior and NF-κB signaling.

In summary, the data presented in this report suggest an oncogenic role for TβRIII in MDA-231 human breast cancer cells and PMTLuc mouse mammary tumor cells. Knockdown of TβRIII in these cells resulted in decreased growth and motility, both in vitro and in vivo. Furthermore, decreased TβRIII expression resulted in an inability to invade Matrigel that was dependent on NF-κB signaling.

C.L. Arteaga: commercial research grant, Merck, Lily, and Monogram Biosciences; expert testimony, BMS, InNexus, OSI Pharm, and AstraZeneca. The other authors disclosed no potential conflicts of interest.

Grant support: NIH R01 grant CA62212 (C.L. Arteaga), Breast Cancer Specialized Program of Research Excellence grant P50 CA98131, and Vanderbilt-Ingram Comprehensive Cancer Center Support grant P30 CA68485. T.L. Criswell was partially supported by National Cancer Institute grant T32 CA 09592.

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.