An estimated 40,000 deaths will be attributed to breast cancer in 2016, underscoring the need for improved therapies. Evading cell death is a major hallmark of cancer, driving tumor progression and therapeutic resistance. To evade apoptosis, cancers use antiapoptotic Bcl-2 proteins to bind to and neutralize apoptotic activators, such as Bim. Investigation of antiapoptotic Bcl-2 family members in clinical breast cancer datasets revealed greater expression and more frequent gene amplification of MCL1 as compared with BCL2 or BCL2L1 (Bcl-xL) across three major molecular breast cancer subtypes, Luminal (A and B), HER2-enriched, and Basal-like. While Mcl-1 protein expression was elevated in estrogen receptor α (ERα)-positive and ERα-negative tumors as compared with normal breast, Mcl-1 staining was higher in ERα+ tumors. Targeted Mcl-1 blockade using RNAi increased caspase-mediated cell death in ERα+ breast cancer cells, resulting in sustained growth inhibition. In contrast, combined blockade of Bcl-2 and Bcl-xL only transiently induced apoptosis, as cells rapidly acclimated through Mcl-1 upregulation and enhanced Mcl-1 activity, as measured in situ using Mcl-1/Bim proximity ligation assays. Importantly, MCL1 gene expression levels correlated inversely with sensitivity to pharmacologic Bcl-2/Bcl-xL inhibition in luminal breast cancer cells, whereas no relationship was seen between the gene expression of BCL2 or BCL2L1 and sensitivity to Bcl-2/Bcl-xL inhibition. These results demonstrate that breast cancers rapidly deploy Mcl-1 to promote cell survival, particularly when challenged with blockade of other Bcl-2 family members, warranting the continued development of Mcl-1–selective inhibitors for targeted tumor cell killing.
Implications: Mcl-1 levels predict breast cancer response to inhibitors targeting other Bcl-2 family members, and demonstrate the key role played by Mcl-1 in resistance to this drug class. Mol Cancer Res; 15(3); 259–68. ©2016 AACR.
This article is featured in Highlights of This Issue, p. 235
The intrinsic apoptotic pathway is tightly regulated by Bcl-2 family members to support developmental processes and proper physiologic function (1). Apoptosis dysregulation often produces pathologic consequences, including cancer formation, progression, and therapeutic resistance (2). At the center of the intrinsic apoptotic pathway are the “effectors” Bax and Bak, which oligomerize at the outer mitochondrial membrane (OMM) by binding to “activators” (Bim, Bid, and Puma; ref. 3). Bak/Bax oligomerization promotes pore formation in the OMM, resulting in mitochondrial depolarization, disruption of oxidative phosphorylation, mitochondrial cytochrome-c release into the cytoplasm, and apoptosome activation, thus initiating caspase-dependent apoptosis (4–6). Antiapoptotic Bcl-2 family proteins (Bcl-2, Bcl-A1, Bcl-xL, Bcl-w, and Mcl-1) restrain the intrinsic apoptotic pathway by binding and sequestering effectors (7) and/or activators (8), thus favoring cell survival. Proapoptotic “sensitizers” (Bad, Hrk, and Noxa) bind and saturate antiapoptotic Bcl-2 proteins, thus favoring apoptosis (9–11). A delicate balance in the relative ratio of antiapoptotic Bcl-2 proteins to apoptotic activators or sensitizers is necessary for cell survival regulation. Cancers can exploit this pathway to evade apoptosis, often through increased levels of antiapoptotic Bcl-2 factors (12–14).
Nearly 250,000 new breast cancer cases will be diagnosed in the United States in 2016 (15). Despite advances in detection and treatment of breast cancers, it is estimated that up to 40,000 patients will die from breast cancer in the United States each year, often due to recurrent metastatic disease, highlighting the need for improved treatments that promote tumor cell killing. Several studies suggest that antiapoptotic Bcl-2 family proteins may be particularly attractive therapeutic targets in breast cancers. For example, Bcl-2 expression was observed in up to 70% of ERα+ breast cancers (16). While Bcl-xL remains less studied in primary breast tumors, Bcl-xL expression was increased in ductal carcinoma in situ (DCIS) compared with normal breast in some breast cancer subtypes (17). Interestingly, many breast cancers increase levels of Bcl-2 family proteins following treatment with tamoxifen (18), fulvestrant (19), and neoadjuvant chemotherapy (NAC; ref. 20). Bcl-2 and Bcl-xL levels reportedly predict poor response to taxanes (21), adriamycin (22), and the HER2 monoclonal antibody Herceptin (23). In addition, Bcl-2 family proteins are often upregulated in endocrine-resistant cancers (24, 25). These findings support an intense interest in research strategies to block the activity of Bcl-2 family proteins as a means to enhance tumor cell killing.
Pharmacologic inhibition of Bcl-2 family proteins has been achieved using compounds that bind to the BH3-domain binding pocket of Bcl-2 family members. These BH3 “mimetics” block the interaction of Bcl-2 family proteins with BH3-domain containing proapoptotic factors (i.e., Bcl-2 activators or sensitizers), including Bim (26). ABT-199/venetoclax, which specifically blocks Bcl-2 from interacting with BH3-domain–containing proapoptotic factors, is currently approved for clinical use in chronic lymphocytic leukemia (CLL; ref. 27), and is in clinical trial in several additional cancers, including breast cancers (28). ABT-263/navitoclax binds to and blocks the activity of Bcl-2 and Bcl-xL, and is showing efficacy in early-phase clinical trials in hematologic malignancies (29–31). Preclinical studies in luminal breast cancers show that ABT-263 (targeting Bcl-2 and Bcl-xL) or ABT-199 (targeting Bcl-2) may effectively increase tumor killing when cells are first “primed” with tamoxifen (32, 33).
These advances in Bcl-2 family targeting warrant a greater understanding of the molecular characteristics of breast cancers that might benefit from this treatment strategy. Examination of large clinical datasets identified expression and gene amplification of BCL2 and BCL2L1 (Bcl-xL). However, we found that MCL1 gene expression occurred more frequently in breast cancers than other Bcl-2 family members. Disruption of Mcl-1 activity increased caspase-activated apoptosis and impaired cell growth to a greater extent than combined disruption of Bcl-2 and Bcl-xL. Importantly, expression levels of MCL1 predicted sensitivity to ABT-263 in a panel of breast cancer cell lines, which may inform results in ongoing clinical trials, or guide patient selection for future trials.
Materials and Methods
Expression analysis of publicly available cancer cell line and breast cancer datasets
mRNA expressions of MCL1, BCL2, and BCL2L1 (Bcl-xL) were curated using cBio Portal (www.cbio.org) for cancer cell lines (CCLE) and breast tumor specimens (TCGA). Breast cancer specimens were stratified on the basis of PAM50 molecular markers (TCGA), and comparative genomic hybridization (CGH) analysis was used to observe alterations at the genetic level (amplifications). mRNA expressions of MCL1, BCL2, and BCL2L1 in breast cancer cell lines (CCLE) were correlated to the IC50 of ABT-263 as determined by the Sanger Institute (http://www.cancerrxgene.org/); data were fit to linear regression.
Cells and tumor tissue were homogenized in ice-cold lysis buffer [50 mmol/L Tris pH 7.4, 100 mmol/L NaF, 120 mmol/L NaCl, 0.5% NP-40, 100 μmol/L Na3VO4, 1× protease inhibitor cocktail (Roche), 0.5 μmol/L proteasome inhibitor (Santa Cruz Biotechnology)]. Proteins were resolved on 4%–12% SDS-PAGE gels and transferred to nitrocellulose membranes, which were blocked in 3% gelatin in TBS-T (Tris-buffered saline, 0.1% Tween-20), incubated in primary antibody [Mcl-1 S19, Bim, Bcl-2, Bcl-xL (Santa Cruz Biotechnology 1:500), β-actin, E-Cadherin (Cell Signaling Technology, 1:10,000)], secondary antibodies [rabbit, goat, mouse (Santa Cruz Biotechnology, 1:5,000–10,000)], and developed with ECL substrate (Thermo Scientific).
Proximity ligation assay
Cells cultured in 96-well plates were fixed with methanol, stained with the Duolink (Sigma) PLA protocol according to manufacturer's directions using Mcl-1 (Santa Cruz Biotechnology, 1:25) and Bim (Santa Cruz Biotechnology, 1:25) antibodies, counterstained with Hoechst, and scanned by ImageXpress Micro XL Automated Microscope. Proximity ligation assay (PLA) fluorescent puncta and Hoechst-stained cells were enumerated using ImageJ software.
Caspase activity assay
A total of 5 × 103 cells/well or 1 × 104 cells/well were seeded in 96-well plates in growth media and were treated with ABT-263 or DMSO for 4 to 48 hours. Caspase-Glo 3/7 Assay (Promega) was used according to the manufacturer's directions. Luminescence was measured on a Glomax Mutli+ Detection System (Promega) luminometer and was standardized to protein values.
Cell lines were purchased directly from ATCC (Homo sapiens ATCC CRL 2327; HTB-22; HTB-133; CRL-1500), and cultured in growth media (DMEM, 10% FBS, 1× antibiotics/antimycotics). Cells were transduced with lentiviral particles expressing three distinct shControl or shMCL1 sequences (Santa Cruz Biotechnology) and kept under constant puromycin selection (1 μg/mL, Life Technologies). For cell growth analyses, 2.5 × 103 cells/well [growth 3D Matrigel (BD Biosciences)] or 5 × 103 cells/well (growth monolayer) were seeded in a 96-well or 12-well plate, respectively. Media, antibiotic, and/or drug were changed every 3 days. For analysis, three-dimensional (3D) colonies were imaged after 14 days (Motic AE3, ProRes CapturePro v2.8.0) and enumerated using ImageJ software. Colonies in monolayer were stained with 0.01% w/v crystal violet (Sigma Life Sciences) and measured using ImageJ. Trypan blue–excluding cells were counted after seeding 5 × 104 cells/well in 12-well plates and treating with drug for 48 hours.
Statistical significance (P < 0.05) was determined using Student unpaired two-tailed t test or ANOVA with Bonferroni post hoc tests followed by Student unpaired two-tailed t test using GraphPad Prism5 software.
Mcl-1 is highly expressed in breast cancers
Antiapoptotic Bcl-2 family member transcripts were assessed in Cancer Cell Line Encyclopedia (CCLE) tumor cell line expression datasets (34). BCL2 transcripts were high in tumors of hematologic origin, but were relatively low in epithelial tumor cells, including breast, whereas BCL2L1 (Bcl-xL) transcripts were higher in tumors of epithelial origin (Supplementary Fig. S1). MCL1 levels were relatively high across several cancers of epithelial (lung, breast, ovary, pancreas, prostate, and stomach) and hematologic (B-cell lymphomas, myelomas) origin, and in melanomas (Fig. 1A). Focusing specifically on breast cancer, we assessed MCL1, BCL2, and BCL2L1 in cell lines stratified by PAM50 molecular subtypes (35–37). MCL1 was detected at higher levels than BCL2 and BCL2L1 across all breast cancer molecular subtypes (Fig. 1B). This observation was confirmed by Western blot analysis in a smaller panel of breast cancer lines, showing abundant Mcl-1 expression across most lines, and variable levels of Bcl-2 and Bcl-xL (Supplementary Fig. S2).
Next, we queried RNA-Seq data from The Cancer Genome Atlas (TCGA)-curated clinical breast cancer datasets for total MCL1, BCL2, and BCL2L1 transcript counts, finding more MCL1 in Luminal A (N = 333), Luminal B (N = 325), HER2-enriched (N = 150), and Basal-like (N = 211) samples than BCL2 and BCL2L1 (Fig. 1C). Basal-like tumors harbored highest MCL1 transcripts, followed by Luminal A, HER2-enriched, and finally Luminal B (Fig. 1D). CGH analysis of TCGA Luminal (A and B) breast cancers (N = 324) demonstrated that MCL1 gene amplifications were found in 21 of 324 tumor samples (7%), which was more frequent than amplifications in BCL2 (2/324) and BCL2L1 (1/324; Fig. 1E). Gene amplifications in MCL1 also occurred in HER2-enriched and Basal-like breast cancer specimens at rates higher than what was seen for other Bcl-2 family-encoding genes. Immunohistochemical analysis of clinical breast tumor specimens (N = 266) showed little Mcl-1 staining in normal breast epithelium, but substantial Mcl-1 upregulation in breast tumor specimens, with the highest expression seen in ERα+ tumors (Fig. 1F). These results suggest a role for Mcl-1 in breast cancer biology, motivating our continued investigation of Mcl-1 in ERα+ breast cancers.
Mcl-1 inhibition decreases luminal breast cancer tumor cell growth
We used MCL1 shRNA sequences (shMCL1; 2–3 sequences per cell line) in the luminal breast cancer cell lines HCC1428, MCF7, T47D, and ZR75-1 to knock down Mcl-1 expression. Western blot analysis of polyclonal puromycin-selected cells demonstrated decreased Mcl-1 protein expression in each cell line, with no change in Bcl-2 and Bcl-xL expression (Fig. 2A). PLA (38) showed a decreased association between Mcl-1 and Bim in cells expressing shMCL1 as compared with shControl nontargeting sequences (Fig. 2B). Mcl-1 knockdown increased caspase-3/7 activity in three of four cells lines (Fig. 2C), and decreased cell growth in three of four lines grown in monolayer (Fig. 2D) or in 3D Matrigel (Fig. 2E). These results confirm that specific Mcl-1 inhibition increases cell death in some, but not all ERα+ breast cancer cells.
ERα+ breast cancer cell lines have limited sensitivity to ABT-263
To determine how blockade of other Bcl-2 family members impacts the growth and survival of ERα+ breast cancer cells, we treated HCC1428, MCF7, T47D, and ZR75-1 with ABT-263, a compound that inhibits two Bcl-2 family members, Bcl-2 and Bcl-xL. Although each cell line showed increased caspase-3/7 activity at 4-hour exposure to ABT-263 (1.0 μmol/L), caspase activation was not sustained through 48 hours in three of the four cell lines (Fig. 3A), suggesting a rapid loss of sensitivity to ABT-263. Consistent with these findings, ABT-263 did not affect the total number of MCF7, T47D, or ZR75-1 cells grown in monolayer for 48 hours (Fig. 3B), or in 3D Matrigel for 14 days (Fig. 3C; Supplementary Fig. S3). PLA (38) confirmed that ABT-263 disrupted interaction of its target protein Bcl-2 with Bim (Fig. 3D and E), confirming on-target activity of ABT-263 at 4- and 24-hour treatment, despite the lack of caspase activation at distal time points in three of the four cells tested.
ABT-263 induces Mcl-1 expression and activity
Cells were treated with a time course of ABT-263 to determine whether rapid upregulation of Bcl-2 family proteins could desensitize cells to ABT-263. Although Bcl-xL expression remained relatively unchanged after 24-hour treatment with ABT-263 (Supplementary Fig. S4A), Bcl-2 expression was elevated at this time point. Despite an increase in Bcl-2 levels, Bcl-2/Bim interactions remain inhibited at 24-hour treatment (Fig. 3D and E), suggesting that elevated Bcl-2 expression does not enhance the ability for Bcl-2 to sequester Bim in the presence of ABT-263. Alternatively, Mcl-1 expression was promptly increased within 4- to 8-hour treatment with ABT-263 (Fig. 4A). This rapid upregulation was also sustained, as Mcl-1 levels were increased at 7-day treatment with ABT-263 in three of four cell lines (Fig. 4A and B). Increased Mcl-1/Bim interactions were observed in each cell line upon treatment with ABT-263 (Fig. 4C and D). While Mcl-1/Bim interaction increased in three of four cell lines at 4-hour treatment, these interactions decreased or returned to baseline at 24 hours in MCF7 and ZR75-1 cells, possibly due to decreased Bim expression at this time point (Supplementary Fig. S4B). In addition, the delayed upregulation of Mcl-1/Bim interactions in HCC1428 cells (24 hours) relative to the other three cell lines (4 hours) may explain the relatively increased sensitivity of HCC1428 cells to ABT-263 (as shown in Fig. 3), suggesting that complex interactions between Bcl-2 family proteins may determine sensitivity to ABT-263.
We assessed MCF7 xenografts treated in vivo with ABT-263 (20 mg/kg) for 16 days. Coprecipitation of Bim with Bcl-2 was seen in control-treated tumors, but was absent in MCF7 xenografts treated with ABT-263 (Fig. 4E), confirming on-target activity of ABT-263 within tumors. Increased Mcl-1 was seen in ABT-263–treated MCF7 whole tumor lysates (Fig. 4F), and PLA measured increased Mcl-1/Bim interactions in tumors treated with ABT-263 as compared with vehicle-treated tumors (Fig. 4G). Although decreased recovery of signal was achieved in formalin fixed paraffin embedded as compared with what was seen in methanol-fixed MCF7 cell cultures (comparing Fig. 4C to G), these results are consistent with the idea that tumors in vivo upregulate Mcl-1 activity in response to blockade of other Bcl-2 family members similar to what was seen in cell culture.
MCL1 expression correlates inversely with sensitivity to Bcl-2/Bcl-xL inhibition
Bcl-2 and/or Bcl-xL inhibition is currently under clinical investigation for the treatment of breast cancers. Given the high Mcl-1 expression levels in breast cancers, and the capacity for rapid Mcl-1 upregulation following Bcl-2/Bcl-xL using ABT-263, we assessed the relationship between MCL1 and sensitivity to ABT-263 across a panel of CCLE-curated ERα+ (n = 16), HER2-amplified (n = 9), and triple-negative (N = 18) breast cancer cell lines (using datasets published in refs. 34 and 39). MCL1 transcript levels correlated with the ABT-263 IC50 in ERα+ and HER2-amplified cells, but not in triple negative breast cancer (TNBC) cells (Fig. 5A). Interestingly, BCL2 and BCL2L1 transcript levels did not correlate directly or inversely with the ABT-263 IC50 in any cell type. These findings support the notion that Mcl-1 may be for a marker of de novo resistance of breast cancers to ABT-263. To test this idea directly, we overexpressed Mcl-1 in HCC1428, MCF7, and T47D cells (Fig. 5B). Although Bim interactions with Mcl-1 were upregulated by 20%–45% in cells transduced with the his-Mcl-1 adenovirus, we found ABT-263 more potently upregulated Mcl-1/Bim interactions in transduced cells (Fig. 5C), demonstrating that Mcl-1 acts as a sink for Bim upon ABT-263 treatment and that ABT-263 promotes maximal Mcl-1/Bim interactions. Consistent with the idea that Mcl-1 levels contribute to de novo ABT-263 resistance, we found that Mcl-1 overexpression substantially decreased caspase-3/7 activation in ABT-263-treated cells (Fig. 5D), further suggesting that ERα+ breast cancer cells use upregulation of Mcl-1 to evade cell death when challenged with ABT-263 (Fig. 5E).
Nearly 40,000 breast cancer–related deaths occur annually in the United States, often in the context of recurrent and/or therapeutically resistant disease (15). Additional molecularly targeted treatment strategies that effectively induce tumor cell killing could potentially decrease tumor recurrences, and would increase therapeutic options for patients with resistant disease. Tumors often rely on antiapoptotic Bcl-2 family proteins to evade tumor cell death. Given the clinical success of ABT-199 in CLL (27), and ongoing clinical investigations of ABT-263 and ABT-199 in breast cancers (28), we were motivated to understand which breast cancers express high levels of Bcl-2 family members, as this might indicate which breast cancers would benefit from Bcl-2 and/or Bcl-xL inhibition. We also investigated the Bcl-2 family member Mcl-1, given recently developed compounds that target Mcl-1 activity that have transitioned into early-phase clinical trials for hematologic malignancies (40). Data shown herein suggest that Mcl-1 may be a key survival factor across all breast cancer subtypes.
In previous studies, a dependency screening approach identified Mcl-1 as a key survival factor in TNBCs (41), consistent with our observations presented herein that among Bcl-2 family members, MCL1 mRNA expression was frequently higher than BCL2 and BCL2L1 mRNA in breast cancers, including TNBCs (Fig. 1C). At the protein level, immunohistochemical Mcl-1 staining of breast cancer tissue microarrays demonstrated profoundly elevated Mcl-1 in breast cancers as compared with normal breast epithelium (Fig. 1F). Furthermore, Mcl-1 levels were elevated in breast tissue containing ERα expression, suggesting that Mcl-1 may confer a selective survival advantage to breast cancer cells, particularly luminal breast cancer cells, but may also represent a vulnerability that can be exploited therapeutically. Mcl-1 knockdown in luminal breast cancer cells decreased Mcl-1 activity, as measured in situ by Mcl-1/Bim interactions (Fig. 2B). Importantly, Mcl-1 knockdown decreased tumor cell survival and tumor cell growth (Fig. 2C–E). These studies complement earlier studies from other groups indicating that Mcl-1 knockdown in some triple-negative (20, 42) and/or HER2-amplified breast cancer cells (43, 44) increases caspase activity. In contrast, combined Bcl-2 and Bcl-xL inhibition using ABT-263 had only an acute impact on caspase activation (Fig. 3A) and did not affect tumor cell growth (Fig. 3B and C). These findings are in agreement with studies from other groups demonstrating that ABT-263 was insufficient as a single agent to induce tumor cell killing in luminal breast cancer cells, but only once cells were primed with tamoxifen did Bcl-2/Bcl-xL or selective Bcl-2 inhibition with ABT-199 induce cell death (32).
Several studies show Mcl-1 production and stability is responsive to cellular cues, including inhibition of other Bcl-2 family members (45–49), making Mcl-1 ideally suited for rapidly evading therapeutically induced tumor cell death [reviewed in ref. 2], and emphasizing the value of Mcl-1 as a therapeutic target. Interestingly, we found that rapid and sustained Mcl-1 induction occurred in response to ABT-263 in luminal breast cancer cells (Fig. 4), although Mcl-1 depletion did not result uniformly in Bcl-2 or Bcl-xL upregulation in luminal breast cancer cells (Fig. 2A). This suggests that Mcl-1 may be a dominant antiapoptotic signal in luminal breast cancers. Previous studies in hematologic malignancies, colorectal cancers, and small-cell lung cancers show that Mcl-1 drives innate resistance to ABT-263, while Mcl-1 suppression could restore sensitivity to ABT-263 (45, 49, 50). Although this idea needs to be tested further in luminal breast cancers, it is possible that maximal tumor cell killing may only be achieved when Bcl-2, Bcl-xL, and Mcl-1 are inhibited, suggesting that the toxicities associated with targeting all three family members need to be explored. In addition, we show herein that MCL1 gene expression levels correlated inversely with sensitivity to ABT-263 (Fig. 5A), suggesting that MCL1 might be used as predictor of patient response to ABT-263 or ABT-199, a hypothesis that could be tested in ongoing and future clinical trials.
In summary, we find that MCL1 gene expression and amplification are frequent occurrences in breast cancers, and that genetic Mcl-1 inhibition increased apoptosis in luminal breast cancer cells, resulting in increased growth inhibition. In contrast, Bcl-2/Bcl-xL inhibition using ABT-263 did not sustain tumor cell killing or growth inhibition. Cells rapidly responded to ABT-263 by increasing Mcl-1 expression, but Mcl-1 depletion did not induce Bcl-2 or Bcl-xL upregulation. Together, these findings support a role for Mcl-1 in survival of breast cancer cells, warranting consideration of Mcl-1 in the design and interpretation of clinical trials investigating Bcl-2 family inhibitors.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M. Williams, R.S. Cook
Development of methodology: M. Williams, D. Elion, V. Sanchez, R.S. Cook
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Williams, L. Lee, D.J. Hicks, M.M. Joly, D. Elion, C. McKernan, M.V. Estrada
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Williams, D. Elion, B. Rahman, J.M. Balko, T. Stricker
Writing, review, and/or revision of the manuscript: M. Williams, M.M. Joly, J.M. Balko, T. Stricker, R.S. Cook
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Williams, D.J. Hicks, B. Rahman
Study supervision: R.S. Cook
We thank Dr. Stephen Fesik for critical review of this manuscript. We are grateful to Dr. Carlos Arteaga for thoughtful discussion of this manuscript and Joshua Bauer from the Vanderbilt High Throughput Screening Facility for his assistance with imaging and analysis of the Proximity Ligation Assay.
This work was supported by Specialized Program of Research Excellence (SPORE) grant NIHP50 CA098131 (Vanderbilt-Ingram Cancer Center), Cancer Center Support grant NIH P30 CA68485 (Vanderbilt-Ingram Cancer Center), NIH F31 CA195989-01 (M.M. Williams) and CTSA UL1TR000445 from the National Center for Advancing Translational Sciences. R.S. Cook received NIH award R01CA143126 and Susan G. Komen for the Cure grant KG100677. M.M. Joly received NRSA F31 predoctoral award CA186329-01. M. Williams received NRSA F31 predoctoral award CA195989-01 and an award from the Vanderbilt Institute for Clinical and Translational Research (CTSA UL1TR000445 from the National Center for Advancing Translational Sciences).
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.