Volume 28, Issue 3, Pages 501-512 (November 2007) A Hypoxia-Controlled Cap-Dependent to Cap-Independent Translation Switch in Breast Cancer  Steve Braunstein, Ksenia Karpisheva, Carolina Pola, Judith Goldberg, Tsivia Hochman, Herman Yee, Joan Cangiarella, Rezina Arju, Silvia C. Formenti, Robert J. Schneider  Molecular Cell  Volume 28, Issue 3, Pages 501-512 (November 2007) DOI: 10.1016/j.molcel.2007.10.019 Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 1 Representative Levels of Protein Synthesis Factors in Breast Tumor Specimens Examined in This Study (A) IHC analysis of representative specimens. Normal breast epithelium, T1, and T3 (LABC) tumors are shown. Data are representative of patient specimens reported in Table S1, using three core punches per specimen. (B) Immunoblot analysis demonstrating specificity of antibody binding. Equal protein amounts of fresh normal breast epithelial tissue and posttreatment surgical resection of a T3 tumor are shown. (C) IHC of eIF4A levels. Representative normal breast epithelium and T3 tumor specimen are shown. Molecular Cell 2007 28, 501-512DOI: (10.1016/j.molcel.2007.10.019) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 2 Analysis of Global Protein Synthesis and VEGF-Specific mRNA Translation with Elevated 4E-BP1 and eIF4G Levels in Increasingly Transformed Cell Lines (A and B) Immunoblot comparison of immortalized MCF10A cells, transformed CRL1902 cells, and transformed BT474 cells for expression levels of identified factors. Cells were treated with rapamycin at 20 μg/ml for 4 hr. (C) Overall protein synthesis activity with overexpression of wild-type or mutant Δ4E-BP1. Global protein synthetic rates were determined by 35S-methionine labeling under normal or hypoxic conditions (0.5% O2 for 24 hr) (all p < 0.05, Student's t test, two sided). (D) Endogenous secreted VEGF protein levels under normoxic (N) and hypoxic (Hx) conditions in cell lines, determined by ELISA, normalized to equal levels of VEGF mRNA. (E) VEGF mRNA levels in BT474 cells under normoxic or 24 hr hypoxic (0.5% O2) conditions. mRNA levels were determined by RT-PCR. Results for other cell lines were similar (data not shown). (F) Gene silencing of eIF4GI. eIF4G was silenced in BT474 cells with an shRNA or a nonspecific control shRNA expressing lentivirus vector to levels found in MCF10A cells. Levels of eIF4G and eIF4A in MCF10A and BT474 cells were determined by immunoblot. (G) VEGF synthesis dependence on eIF4G. VEGF protein levels were determined by ELISA in BT474 cells, normalized to equal mRNA levels as shown in (E). (H) Overall protein synthesis levels with eIF4G reduction. Levels of protein synthesis were determined by 35S-methionine incorporation after eIF4G silencing under normoxic or 24 hr hypoxic conditions in vector control and 4E-BP1 overexpressing cells. Data in all studies were derived from at least three independent experiments. Standard errors of the mean were calculated from at least three independent experiments and all results achieve p < 0.05 (paired Student's t test). Molecular Cell 2007 28, 501-512DOI: (10.1016/j.molcel.2007.10.019) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 3 Overexpression of 4E-BP1 and eIF4G Confers a Cap-Dependent to IRES-Dependent mRNA Translation Switch (A–C) Bicistronic vectors containing a cap-dependent 5′NCR and cap-independent IRES for (A) VEGF, (B) HIF1α, or (C) EMCV were tested for cap-dependent and cap-independent translation by using a dual Luciferase reporter system under normoxic (N) and 24 hr hypoxic (Hx) (0.5% O2) conditions. Quantification of results, expressed as the rates of IRES/cap-mediated translation, were tabulated as shown on the right. All results are p < 0.05 controls versus experimental groups, determined by Student's t test, two sided. (D) Vector expressing the cap-dependent RFP and IRES-dependent GFP bicistronic reporter mRNA in hypoxic control and 4E-BP1 BT474 cells, visualized by confocal microscopy and quantified from ten independent fields of cells. (E) Integrity of cap-dependent initiation complexes under normoxia and increasing hypoxia with overexpression of 4E-BP1 protein. Complexes were purified by cap chromatography, and proteins detected by immunoblot analysis. Lower resolution SDS-PAGE was used to concentrate all phosphoforms of 4E-BP1 associated with eIF4E for better quantitation. Typical results are shown of representative studies from at least three independent experiments. (F) VEGF expression levels during normoxia (20% O2), 24 hr hypoxia at 5, or 0.5% O2 levels were determined by ELISA of secreted factor, normalized to equal VEGF mRNA levels, as described in the legend to Figure 2. Standard errors of the mean were calculated from at least three independent experiments and all results achieve p < 0.05 (paired Student's t test). Molecular Cell 2007 28, 501-512DOI: (10.1016/j.molcel.2007.10.019) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 4 Overexpression of 4E-BP1 Promotes Tumor Angiogenesis, Vascular Permeabilization, and VEGF Expression (A) Representative tumors grown in the CAM are shown, photographed in situ, corresponding to vector control, overexpression of vector alone, wild-type 4E-BP1, or the Δ4E-BP1 mutant (for BT474 cells only). (B) The vascular areas of tumors grown in CAMs from cell lines with or without overexpression of wild-type 4E-BP1 were determined as described in the Experimental Procedures. (C) Morphometric analysis of tumor vasculature in CAM studies, made visible by immunofluorescence staining with fluorescein isothiocyanate (FITC)-labeled Lycopersicon lectin (Vector Labs), followed by fluorescence photography. (D) Immunoblot analysis of VEGF-A protein levels and 4E-BP1 levels by high-resolution SDS-PAGE, obtained from equal amounts of the BT474 CAM tumor protein. (E) Tumor mass (weight) of CAM-derived tumors with increased expression of wild-type 4E-BP1 or Δ4E-BP1 mutant (in BT474 cells only). CAM tumor weights were quantified, and mean values with SEM shown, derived from at least three independent experiments. Standard errors of the mean were calculated from at least three independent experiments and all results achieve p < 0.05 (paired Student's t test). Molecular Cell 2007 28, 501-512DOI: (10.1016/j.molcel.2007.10.019) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 5 Effect of 4E-BP1, eIF4G, or eIF4E on Tumor Growth, Angiogenesis, and VEGF Expression (A) MDA-MB435 cells were stably reduced in expression of 4E-BP1, eIF4GI, or eIF4E by transformation with shRNA lentivirus vectors and implanted into 10-day-old CAMs, and tumors were grown for 8 days. Tumors were excised, and equal amounts of protein examined by SDS-PAGE immunoblot analysis as shown. (B) Top, representative tumors grown in the CAM from control and silenced MB435 cells were photographed in situ. Bottom, the tumor vasculature was made visible by immunofluorescence staining with FITC-labeled lectin as described in the legend to Figure 4. (C) Vascular density of tumors grown in CAMs shown in (B) was determined as described in the Experimental Procedures. (D) The level of VEGF expression secreted into CAMs was determined by assaying equal amounts of CAM extracts by ELISA. (E) Tumor growth was determined by comparing the excised weight and volume of tumors (mg × mm3), normalizing to the vector control tumors set at 100%. Data in all studies were derived from at least three independent experiments comprising five eggs per tumor cell line. (F) Vector control or HA-4E-BP1 overexpressing BT474 cells were implanted subcutaneously into the flank of nude mice with implantation of 60 day exogenous estrogen pellets; tumors were photographed in situ and excised at 21 days. IHC was carried out on paraffin-embedded sections. Typical results are shown. (G) BT474 tumor cell volumes were determined at 21 days from three independent trials of eight mice per group (p < 0.001). (H) BT474 control and 4E-BP1/eIF4G overexpressing tumors were grown in mice for 3 weeks while expressing the cap-dependent RFP/IRES-dependent GFP bicistronic reporter mRNA. Tumors were excised into OCT, and frozen sections were prepared and visualized by confocal microscopy (merged RFP/GFP image shown). Data were quantified from ten independent fields of cells and displayed as the ratio of GFP/RFP. (I) Tumor vasculature visualized by Texas red-dextran decoration of tumor vasculature endothelium. Mice were injected i.v. with stain and sacrificed 5 min later, and tumor vasculature was visualized in embedded sections by confocal microscopy and quantified. Arrows indicate blood vessels with costained erythrocytes. Standard errors of the mean were calculated from at least three independent experiments and all results achieve p < 0.05 (paired Student's t test). Molecular Cell 2007 28, 501-512DOI: (10.1016/j.molcel.2007.10.019) Copyright © 2007 Elsevier Inc. Terms and Conditions