Overexpression of ERBB2 and its neighboring genes on chromosome 17 occurs in approximately 25% of breast tumors and is associated with poor prognosis. While amplification of the 17q12-21 chromosomal region often correlates with an increase in the transcriptional rates of the locus, the molecular mechanisms and the factors involved in the coordinated expression of genes residing within the ERBB2 amplicon remain largely unknown. Here we demonstrate that estrogen-related receptor α (ERRα, NR3B1) and its coregulator PGC-1β are key effectors in this process. Using a mouse model of ERBB2-initiated mammary tumorigenesis, we first show that ablation of ERRα significantly delays ERBB2-induced tumor development and lowers the levels of amplicon transcripts. Chromosome 17q-wide binding site location analyses in human breast cancer cells show preferential recruitment of ERRα to DNA segments associated with the ERBB2 amplicon. Furthermore, ERRα directs the co-recruitment of the coactivator PGC-1β to segments in the 17q12 region and the recruitment of RNA polymerase II to the promoters of the ERBB2 and coamplified genes. ERRα and PGC-1β also participate in the de-repression of ERBB2 expression through competitive genomic cross-talk with estrogen receptor α (ERα) and, as a consequence, influence tamoxifen sensitivity in breast cancer cells. Taken together, our results suggest that ERRα and PGC-1β are key players in the etiology of malignant breast cancer by coordinating the transcriptional regulation of genes located in the 17q12 region, a process that also involves interference with the repressive function of ERα on ERBB2 expression. Cancer Res; 70(24); 10277–87. ©2010 AACR.
Breast cancer is a complex disease implicating distinct cell types and multiple signaling pathways that together engender a multiplicity of tumor subtypes (1). The molecular heterogeneity of breast tumor subtypes dictates their intrinsic response to specific therapeutic approaches and has therefore become an important aspect in the clinical management of the disease (2). Amplification of the 17q12–21 region leads to concomitant overexpression of ERBB2 and several coamplified genes and occurs in approximately 20% to 30% of breast tumors (3). The ERBB2-amplified breast cancer subtype is strongly associated with poor prognosis (4). Although overexpression of ERBB2 is usually linked to amplification of the 17q12 region, this region is not amplified in some ERBB2-positive breast tumors. However, in both instances, an increase in the transcriptional regulation rate of the locus is observed (5–8).
Loss of repression of ERBB2 has been linked to tamoxifen resistance in breast cancer cells (9, 10). It is believed that transcriptional repressors including FOXP3 (11), PAX2 (10), GATA4 (12), PEA3 (13), and MYB (14) act to quench the expression of ERBB2 in ERBB2-negative tumors. Notably, it has also been shown that estrogen receptor α (ERα, NR3A1), in cooperation with the transcription factor PAX2, can repress ERBB2 expression through binding to a cis-regulatory element in the presence of either 17β-estradiol (E2) or 4-hydroxy-tamoxifen (OHT; ref. 10). Indeed, ERα-positive tumors with the worst prognosis that do not respond to tamoxifen therapy often express high levels of ERBB2 (15–18).
A positive increase in the transcriptional control of the locus can contribute to relieve the transcriptional repression that takes place on ERBB2, and overexpression of ERBB2 is thought to drive the amplification of the 17q12 locus (19). Transcription factors such as AP-2 (8), YY1 (20), ETS (21), YB-1 (22), and EGR2 (23) have been shown to be recruited to the promoter of ERBB2 and play a role in its overexpression in breast cancer cells. Therefore, understanding the mechanisms governing the repressive and positive regulation of ERBB2 expression and its neighboring genes in the amplified locus is of considerable importance in identifying the transcription factors and molecular events involved in the establishment of ERBB2-positive tumors.
Although most studies on the 17q12 amplicon have been centered on the expression and activity of ERBB2 itself, it is now clear that coamplified genes not only contribute but could also be essential to the progression of ERBB2-positive breast tumors (24, 25). The minimal 17q12 amplicon includes genes that are involved in signal transduction (GRB7, PERLD1, PPP1R1B), transcription (MED1, IKZF3, NEUROD2, PNMT), cell migration and invasion (C17orf37, GRB7), inhibition of apoptosis (MED1), genomic instability (PERLD1), and tamoxifen resistance (STARD3, GRB7; refs. 26–28). Although transcriptional regulation of the coamplified genes in relation to the ERBB2 subtype is now considered an important aspect in the establishment of ERBB2-positive breast tumors, the molecular mechanisms associated with this phenomenon have yet to be investigated in detail.
The orphan nuclear receptor estrogen-related receptor α (ERRα, NR3B1) shares both structural and functional features with ERα (29). The expression of ERRα is inversely correlated to that of ERα but positively associates with that of ERBB2 and with poor prognosis in breast cancer (30, 31). A recent genome-wide binding sites location analysis in breast cancer cell lines intersected with gene expression data from breast tumors has shown that ERRα signaling contributes to known pathways linked to breast cancer progression, including those involving ERα and ERBB2 (32). Conversely, the transcriptional activity of ERRα can be modulated by the epidermal growth factor (EGF)/ERBB2 signaling pathway in breast tumors (33, 34). ERRα preferentially acts in concert with the coregulators peroxisome proliferator-activated receptor γ, coactivator-1α (PGC-1α), and PGC-1β, whose combined roles have been extensively studied in the context of the regulation of bioenergetic pathways (35, 36). Gene expression and binding sites location analyses in breast cancer cell lines have also demonstrated that ERα and ERRα display strict binding site specificity and maintain independent mechanisms of transcriptional activation, suggesting a prevailing ERα-independent role for ERRα in breast tumor development (32, 37). Nevertheless, a significant number of genes that are common targets of the 2 nuclear receptors are also important players in breast tumor development (32).
Here we show that ERRα is required for the full potential of ERBB2-driven mammary tumorigenesis in mice and that tumors lacking ERRα express lower levels of most amplicon genes. Chromosome-wide identification of regulatory regions occupied by ERRα and PGC-1β as well as gene expression data identify both factors as key regulators of the expression of ERBB2 and several coamplified genes in the 17q12 region in breast cancer cells. Finally, we demonstrate that an antagonistic interaction between ERα and the ERRα/PGC-1β complex at the ERBB2 locus is an important determinant of ERBB2 expression involved in the development of tamoxifen resistance in breast cancer cells.
Materials and Methods
Cell culture, reagents, and antibodies
MCF-7 and SKBr3 cells were cultured as described previously (38). Tam-R-MCF-7 cell line was derived as described previously (39). ERRα chromatin-immunoprecipitation (ChIP) assays were performed using an anti-hERRα polyclonal antibody raised and validated in our laboratory (32, 40). Other antibodies used were anti-RNA-Polymerase II (8WG16), anti-ERRα (Millipore), anti-PGC-1β, and anti-ERα (Santa Cruz Biotechnologies). siRNAs against ERα, ERRα, PGC-1β, and control (ON-Target-Plus siRNA pool) were obtained from Dharmacon.
Transgenic mouse model study
The derivation of the conditionally activated NeuNT (erbB2NT) has been described in detail (41). To generate mice that expressed erbB2NT in the mammary glands of animals carrying a null allele for Esrra, erbB2NT and MMTV-Cre mice were first bred with mice heterozygous for null alleles of Esrra (42). All mice were previously derived in a pure FVB genetic background. These mice were then bred to generate mice null for Esrra also carrying one copy of erbB2NT and one copy of MMTV-Cre (e.g., Ersra −/−/erbB2NT/MMTV-Cre). The resulting mice were examined for the presence of the excised recombinant erbB2NT allele through Southern blot analysis. The levels of amplification of the locus were determined by quantitative RT-PCR (qRT-PCR) using primers listed in Supplementary Table S1. Female mice were examined twice a week for mammary tumor development by palpation.
ChIP assays and ChIP-on-chip on chr.17q tiled arrays
ChIP was performed as described previously (38). Quantification of ChIP enrichment by real-time qRT-PCR was carried out using the LightCycler480 instrument (Roche). ChIP-on-chip was carried out on custom chr.17q Agilent tiled arrays (150-bp resolution). Chromatin was prepared from SKBr3 cells (for ERRα and PGC-1β ChIP-on-chip) or from MCF-7 cells exposed to 10 nmol/L E2 for 45 minutes (for ERα ChIP-on-chip). The primers used for standard ChIP are listed in Supplementary Table S1.
Computational motif discovery
De novo and known motif discovery was performed with MEME Suite (http://meme.sdsc.edu/meme4_4_0/intro.html). Motif discovery was also confirmed using the Genomatix Software Suite (http://www.genomatix.de/en/produkte/genomatix-software-suite.html).
siRNA for ERα and control were transfected in MCF-7 cells cultured in phenol-red-free DMEM media supplemented with hormone-deprived serum using the HyperFect reagent (Qiagen). Similarly, siRNAs for ERRα, PGC-1β, and control were transfected in SKBr3 cells using the HyperFect reagent for 48 to 60 hours.
mRNA from SKBr3 cells transfected with the ERRα or PGC-1β siRNA were reverse-transcribed into cDNA using Superscript (Invitrogen) and analyzed by qRT-PCR with SYBR-green–based RT-PCR (Roche). Alternatively, RNA was extracted from mice mammary gland tumors using the RNA tissue extraction kit (Qiagen) and reverse-transcribed using Superscript (Invitrogen). Primer pairs used for qRT-PCR are listed in Supplementary Table S2.
Proliferation assay assessed by 3H-thymidine incorporation
SKBr3 cells were transfected with the appropriate siRNA for 60 hours. Cells were incubated in the presence of 1 μCi 3H-thymidine for 4 hours prior to fixation and harvesting. Similarly, MCF-7 and Tam-R-MCF-7 cells were grown in phenol–red-free DMEM supplemented with hormone-deprived serum. Upon siRNA transfection, the media was supplemented with 100 nmol/L OHT (Sigma) or vehicle for 60 hours before harvesting and isotope counting.
Ablation of ERRα delays ERBB2-induced mammary gland tumorigenesis
To initiate our study on the functional relationship between ERRα and ERBB2 in mammary gland tumorigenesis, we first used the well-characterized Neu-NT knock-in transgenic mouse model of mammary tumorigenesis (41) to generate ERRα-deficient transgenic mice conditionally expressing the activated Neu under the control of the endogenous Erbb2 promoter. The choice of this model was guided by the previous observation that these mice develop focal mammary tumors with high frequency after a long latency period that bear amplified copies of the activated Erbb2 allele on chromosome 11 (41, 43). In addition, this model is ideal to study the impact of alteration in the gene regulatory machinery as expression of the Erbb2 locus remains under the control of endogenous regulatory elements. We observed that ablation of ERRα significantly delayed ERBB2-induced mammary gland tumorigenesis (P < 0.05; Fig. 1A), indicating that the presence of ERRα is required for optimal development of ERBB2-driven tumors. The ERRα-null mice lactate normally and the development of the mammary gland is not affected by the lack of ERRα expression (Fig. 1B). As expected, amplification of the Erbb2 locus was observed in the tumors derived from this model but the absence of ERRα did not significantly affect the level of amplification of the locus as compared with wild-type (Fig. 1C). To investigate the possible contribution of ERRα in the transcriptional regulation of Erbb2 and neighboring genes in the Erbb2 amplicon, we next assessed their levels of expression in the tumors. We found that the relative transcript levels of most of the amplicon genes tested, including Erbb2, were either significantly decreased or show a similar downward trend in tumors arising from ERRα knockout mice compared with the wild-type ones (Fig. 1D). Overall, these results suggest that ERRα might play a role in the development of ERBB2-driven mammary tumors through transcriptional regulation of Erbb2 and other genes located within the amplicon.
ERRα is recruited to specific sites at chr.17q12 and regulates the expression of ERBB2 and coamplified genes in human breast cancer cells
In order to investigate the role of ERRα in the transcriptional regulation of the genes located within the ERBB2 amplicon in human breast cancer cells, we first performed genomic location analysis of ERRα in the ERα-negative SKBr3 cell line using a custom tiled array covering the q-arm of human chromosome 17. ChIP-on-chip analysis of ERRα on the chr.17q region revealed 92 segments bound by ERRα and validated by standard ChIP analyses (Supplementary Table S3 and Fig. S1A). De novo DNA motif discovery confirmed the enrichment of the ERRE motif within the bound segments (Supplementary Fig. S1B). Remarkably, a significant enrichment of ERRα-bound segments is observed within the amplicon region (32 out of 92 segments) and 13 segments map to the minimal region steadily amplified in ERBB2-positive tumors (Fig. 2A, shaded region). ERRα binding events were observed in ERBB2 itself as well as in the transcriptional unit of genes that are consistently coamplified with ERBB2, such as PERLD1, C17orf37, and GRB7 (Fig. 2B).
To relate the binding profile of ERRα to the transcriptional regulation of target genes, we then monitored modulation in the recruitment of RNA-Pol II by ChIP to the promoters of genes located within the ERBB2 amplicon upon depletion of ERRα in SKBr3 cells. Indeed, a significant decrease in both ERRα and RNA-Pol II recruitment was observed at the promoter of CRKRS, a target gene at which ERRα binding occurs directly at the promoter (Fig. 3A). Furthermore, a decrease in RNA-Pol II recruitment was also observed at promoters of amplicon genes for which ERRα binding takes place far upstream of the transcriptional start sites or in intronic regions of genes such as ERBB2, GRB7, and PERLD1 (Fig. 3A). Depletion of ERRα had no effect on the occupancy of RNA-Pol II at the promoter of the gene, LZTS2, that is not a target of ERRα (Fig. 3B). These results demonstrate that the recruitment of ERRα at both near and distant sites from transcriptional start sites contributes to the recruitment of RNA-Pol II at the promoters of target genes. We next assessed the effect of ERRα depletion on the expression of amplicon genes in human breast cancer cells. In agreement with the observation made in ERBB2-induced tumors developed in the ERRα knockout mice, depletion of ERRα in SKBr3 cells leads to a significant decrease in the relative expression of most amplicon genes (Fig. 3C). These results indicate that ERRα positively regulates the expression of ERBB2 and numerous genes that coamplify with it on chr.17q12.
The coactivator PGC-1β is recruited to ERRα-bound segments in the chr.17q12 amplicon
We have previously shown that the coactivator PGC-1β can be corecruited with ERRα at specific genomic locations in SKBr3 cells (32). To assess whether PGC-1β contributes more broadly in the transcriptional regulation of amplicon genes, we performed PGC-1β ChIP-on-chip in SKBr3 cells hybridized on the chr.17q tiled array. The ChIP-on-chip experiment identified 73 segments significantly bound by PGC-1β on chr.17q (Supplementary Table S4 and Supplementary Fig. S2A) of which 24 were segments common to ERRα (Fig. 4A). Interestingly, PGC-1β recruitment was observed at sites shared with ERRα in regions located within key genes of the ERBB2 amplicon including ERBB2, PERLD1, GRB7, and NR1D1 (Supplementary Table S4 and Supplementary Fig. S2B). De novo DNA motif discovery identified the ERRE as the most enriched motif within the common segments shared by both factors (P = 7.8e-05), indicating that PGC-1β recruitment to the common segments occurs through ERRα (Fig. 4A and Supplementary Fig. S3). Indeed, depletion of ERRα in SKBr3 cells using specific siRNAs leads to a significant reduction in PGC-1β recruitment to the common sites located in chr.17q amplicon genes as assessed by standard ChIP (Fig. 4B). We further tested the effect of PGC-1β recruitment on the expression of the amplicon transcripts. As observed for ERRα, depletion of PGC-1β using specific siRNAs leads to a significant decrease in the relative levels of transcripts of several amplicon genes (Fig. 4C). Because ERBB2 and the coamplified genes are potent inducers of breast tumor growth, we next evaluated the impact of loss of ERRα and/or PGC-1β on cell proliferation. Specific knockdown of ERRα, PGC-1β, or of both factors, using specific siRNAs led to a significant decrease in SKBr3 cell proliferation as assayed by 3H-thymidine incorporation (Fig. 4D). Taken together, these results demonstrate that ERRα contributes to the corecruitment of PGC-1β to the chr.17q region and that both factors regulate the expression of amplicon genes and aggressive growth of SKBr3 breast cancer cells.
ERRα competes with ERα for recruitment to common segments in ERBB2 and GRB7 and contributes, along with PGC-1β, to tamoxifen resistance in MCF-7 cells
In ER-positive tumors that are tamoxifen-sensitive, the expression of ERBB2 is low due to transcriptional repression of the gene inflicted by various transcription factors, including ERα and the corepressor PAX2. It has been suggested that loss of this repression and consequential increase in ERBB2 expression and signaling contribute to acquired tamoxifen resistance (10). Overexpression of other coamplified genes such as GRB7 have also been implicated in resistance to endocrine therapy. We next explored the possibility that ERRα could interfere with ERα signaling in the regulation of specific genes within the ERBB2 amplicon and thus take part in the development of tamoxifen resistance. In order to assess a potential cross-talk at the genomic level, we performed an ERα ChIP-on-chip experiment in E2-treated MCF-7 cells using the same chr.17q arm tiled array (Supplementary Table S5). We identified 194 segments significantly bound by ERα, of which 10 overlapped with ERRα-bound loci and 5 overlapped with PGC-1β-bound loci (Supplementary Fig. S4). Notably, the overlapping ERα/ERRα segments include a site downstream of GRB7 as well as a site located in the first intron of ERBB2. Of particular interest, the intronic ERBB2 site was previously identified as bound by the ERα/PAX2 repressor complex and shown to play a role in tamoxifen sensitivity of MCF-7 cells (Fig. 5A).
We confirmed that treatment of MCF-7 cells with E2 leads to a downregulation of ERBB2 and GRB7 expression (Supplementary Fig. S5A). Examination of the bound sequences in ERBB2 and GRB7 revealed the presence of a mixed ERE/ERRE binding site as previously defined for shared ERα/ERRα binding sites (32), indicating that ERα and ERRα should compete for binding at these common sites. Using specific siRNAs and standard ChIP, we indeed observed that depletion of ERα from MCF-7 cells was accompanied by an increase in ERRα recruitment to the common sites that were otherwise not bound by ERRα in MCF-7 cells (Fig. 5B). Depletion of ERα had no effect on ERRα recruitment to ERRα-specific sites (Supplementary Fig. S5B). These results indicate that the concomitant presence of both nuclear receptors yields to competitive transcriptional regulation of ERBB2 and GRB7 gene expression.
We next asked whether this transcriptional cross-talk could play a role in endocrine resistance using a tamoxifen-resistant MCF-7 cell line generated in our laboratory (Tam-R-MCF-7). As expected, the expression of ERBB2 and GRB7 is increased in the Tam-R-MCF-7 cell line compared with the parental MCF-7 cells (Fig. 5C). We further show that depletion of either ERRα or PGC-1β (Supplementary Fig. S5C and D) leads to a significant decrease in ERBB2 and GRB7 transcripts in Tam-R-MCF-7 cells while having no effect on the expression of these genes in the parental MCF-7 cell line (Fig. 5C). We then assessed the effect of loss of ERRα or PGC-1β on the proliferation of parental MCF-7 and Tam-R-MCF-7 cells in the absence or presence of OHT. As shown in Figure 5D, the proliferation rate of MCF-7 cells is decreased upon OHT treatment whereas depletion of either ERRα or PGC-1β has no additional effect on the proliferation of these cells. In contrast, whereas OHT does not influence the proliferation rate of the Tam-R-MCF-7 cells, depletion of either ERRα or PGC-1β reinstates the antiproliferative effect of OHT in these cells. Taken together, these results suggest that both factors may play a role in the development of OHT resistance in breast cancer cells.
Amplification of the ERBB2 locus and neighboring loci on chromosome 17q12, which is accompanied by overexpression of the amplified genes, plays an important role in the development of tamoxifen resistance and a more aggressive breast cancer phenotype (44). The outcome of this work, derived from a functional genomic approach and validated in well-characterized in vivo and in vitro models of human breast cancer, demonstrates that the orphan nuclear receptor ERRα and its coregulator PGC-1β are key elements in the transcriptional regulation of genes located within the chr.17q amplicon, including ERBB2 itself. In addition, this study identifies molecular mechanisms through which these 2 factors can contribute to the development of tamoxifen resistance in breast cancer cells.
The observation that ERRα regulates the expression of several genes located within the ERBB2 amplicon denotes the importance of ERRα signaling in the progression of the ERBB2-driven breast tumor subtype. This is further emphasized by the observation that the presence of ERRα is required to observe the full oncogenic potential of ERBB2 in a well-established mouse model of human breast cancer (Fig. 1). Notably, we have shown that ERRα regulates the expression of most genes present in the minimal ERBB2 amplicon. The coamplified genes are mainly involved in signal transduction and transcription influencing various biological processes including cell migration, invasion, and survival as well as resistance to tamoxifen (9, 10, 24–28). It can therefore be envisioned that, by acting as a global transcriptional regulator of the ERBB2 amplicon, ERRα contributes to establish the ERBB2-positive tumor subtype which is characterized not only by increased ERRB2 signaling but also by various cellular processes controlled by coamplified genes. Mechanisms that trigger the amplification of the locus are not well understood but it has been observed that an increase in the transcription levels of ERBB2 often precedes locus amplification (5, 45). ERRα could inflict positive transcriptional pressure on the 17q12 region and thus participates actively in the establishment of ERBB2-positive aggressive tumors.
Our results are in agreement with previous observations demonstrating that expression of ERRα positively correlates with that of ERBB2 in breast tumors and that ERRα transcriptional activity is positively modulated by EGFR/ERBB2 signaling in breast cancer cells (30, 33, 34). Such interactions can be integrated in a model of positive feed-forward regulatory loops whereby all the processes involved further enhance the buildup of their own stimulus (46). In ERBB2-negative tumors, the expression of ERBB2 is maintained at low levels by various sequence-specific transcription factors such as Pax2, FOXP3, PEA3, GATA4, and MYB, as well as by ERα. It is thus likely that the positive feed-forward regulatory loop involving ERRα and PGC-1β favors escape from the negative regulation inflicted on ERBB2 by ERα during progression of breast cancer (Fig. 6). This model suggests a possible role for ERRα in mediating the transition of a subset of ER-positive luminal tumors toward the more aggressive ERBB2-expressing subtype.
The genomic convergence between ERRα and the ERBB2 amplicon described in this study is even more relevant considering the finding of competitive ERRα recruitment at a site in ERBB2 targeted by ERα/Pax2. This site has been shown to be involved in maintaining the repressive state of ERBB2 in ERα-positive endocrine-responsive tumors (10). We also observed competitive recruitment of ERRα/PGC-1β and ERα on a site proximal to GRB7. Interestingly, as it is the case for ERBB2,ERRα, and ERα have opposite transcriptional effects on the expression of GRB7 in breast cancer cells. The loss of negative regulation leads to overexpression of ERBB2 and ERBB2-coamplified genes such as GRB7 in luminal breast tumors and has been associated with a poor response to targeted endocrine therapy (47). The effect of ERRα/PGC-1β depletion on tamoxifen-mediated proliferation of Tam-R-MCF-7 cells suggests a mechanism whereby the transcriptional regulation of ERBB2 and coamplified genes by these factors can contribute to the establishment of acquired tamoxifen resistance. Other ERBB2-coamplified genes have been associated with the response to trastuzumab or anthracycline-based chemotherapies in single or combination therapies (48). It would therefore be of interest to assess how the transcriptional regulation of these genes by ERRα/PGC-1β affects the response to other targeted therapies.
Little is known about the expression and function of PGC-1β in breast cancer cells. Here we show that PGC-1β induces the expression of ERBB2 in breast cancer cells. In line with this observation, we have recently shown that ERBB2 signaling also induces the expression of PGC-1β in breast cancer cells (49). Therefore, together with ERRα, our results suggest that PGC-1β contributes to the establishment of the aggressive ERBB2-positive tumors through a positive feed-forward regulatory loop (Fig. 6).
In conclusion, this study clearly demonstrates that, in addition to its primary role in the control of cellular energy metabolic pathways in both normal and cancer cells (32, 35, 49), the ERRα/PGC-1β complex promotes the development of the ERBB2-positive tumor subtype and tamoxifen resistance in breast cancer via transcriptional control of the ERRB2 amplicon.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This work was supported by grants from the Canadian Institutes of Health Research (MOP-64275) and a Terry Fox Foundation Program Project Grant from the National Cancer Institute of Canada. G. Deblois and M.-C. Perry are recipients of studentships from the Fonds de la Recherche en Santé du Québec. G. Deblois is also supported by a predoctoral traineeship award (W81XWH-10-1-0489) from the U.S. Department of Defense Breast Cancer Research Program.
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.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received August 4, 2010.
- Revision received October 15, 2010.
- Accepted October 15, 2010.
- ©2010 American Association for Cancer Research.