Breast cancer cell mesenchymal transition and metastasis directed by DAP5/eIF3d-mediated selective mRNA translation

Abstract

Cancer cell plasticity enables cell survival in harsh physiological environments and fate transitions such as the epithelial-to-mesenchymal transition (EMT) that underlies invasion and metastasis. Using genome-wide transcriptomic and translatomic studies, an alternate mechanism of cap-dependent mRNA translation by the DAP5/eIF3d complex is shown to be essential for metastasis, EMT, and tumor directed angiogenesis. DAP5/eIF3d carries out selective translation of mRNAs encoding EMT transcription factors and regulators, cell migration integrins, metalloproteinases, and cell survival and angiogenesis factors. DAP5 is overexpressed in metastatic human breast cancers associated with poor metastasis-free survival. In human and murine breast cancer animal models, DAP5 is not required for primary tumor growth but is essential for EMT, cell migration, invasion, metastasis, angiogenesis, and resistance to anoikis. Thus, cancer cell mRNA translation involves two cap-dependent mRNA translation mechanisms, eIF4E/mTORC1 and DAP5/eIF3d. These findings highlight a surprising level of plasticity in mRNA translation during cancer progression and metastasis.

Keywords: CP: Cancer; CP: Molecular biology; DAP5; breast cancer; epithelial-to-mesenchymal transition; metastasis; protein synthesis; translation initiation.

Figure 1.. Increased eIF4G2/DAP5 mRNA and protein expression are strongly correlated with metastasis and reduced survival in triple-negative breast cancer patients

(A) TCGA cohort for metastasis-free survival for ER⁻/PR⁻/Her2⁻ (triple-negative) breast cancer (TNBC) patients. Kaplan-Meier estimates derived from mRNA expression data on the basis of high and low expression levels of eIF4G2 mRNA (top) and eIF3d (bottom). Univariate analysis was performed using the log rank test. Cox proportional hazard regression models were also used. (B) Representative IHC staining showing score for DAP5 protein levels of TNBC primary tumors that did not metastasize during 8-year follow up, and primary tumor and recurrent matched metastases that recurred within 8 years. See also Tables S1 and S2 for IHC scoring summaries, statistical analysis, patient demographics, and tumor characteristics. Patient demographics are representative of first clinical presentation for the overall U.S. TNBC population.

Figure 2.. DAP5 expression promotes breast cancer cell migration, EMT, and resistance to anoikis

(A) Representative immunoblots of 3 independent studies of ITG proteins in Nsi and DAP5-silenced MB-231 cells. Dox induction of shRNAs for 4 days followed by immunoblot analysis of equal protein amounts. (B) Representative immunoblots of 3 independent studies of EMT biomarker protein levels in Nsi and DAP5-silenced 4T1 and MB-231 cells carried out as in (A). (C) Representative immunoblots of 3 independent studies of eIF4E and eIF2α proteins in Nsi and DAP5-silenced MB-231 cells carried out as in (A). (D) Representative immunoblots of 3 independent studies to test requirement for eIF3d cap binding activity for DAP5-dependent mRNAs in MB-231 cells. Cells were small interfering RNA (siRNA) silenced for eIF3d for 1 day and transfected with vectors expressing WT, α5, or α11 cap binding mutants of eIF3d for 2 days, and equal protein amounts analyzed using immunoblot. (E) Matrigel Transwell invasion assays performed with 4T1 and MB-231 cells silenced for 24 h with Nsi or DAP5 Dox-inducible shRNAs. Mean number of invading cells/field with SEM from 3 replicates for each cell line. See also Figure S4. (F) Cell migration wound healing assays performed with 4T1 and MB-231 cells silenced for 24 h with Nsi or DAP5 Dox-inducible shRNAs. Time 0 represents 100% wound separation of the cell layer. Mean of percentage of migrated cell surface areas (percentage closure) with SEM from 3 replicates for each cell line. See also Figure S4. (G) Induction of apoptosis by cell detachment (anoikis) determined in 4T1 and MB-231 cells silenced for 24 and 48 h for Nsi, DAP5, or eIF4GI, comparing cells maintained on adherent or ultra- low-adherence plates. Percentage cell apoptosis determined by annexin V-fluorescein isothiocyanate (FITC) staining and flow cytometry, quantified using FlowJo software. Mean with SEM of 3 independent studies per cell line. (H) Representative immunoblot of 3 independent studies of DAP5, eIF4G1, eIF3d, and control β-actin protein levels in Nsi or DAP5-silenced 4T1 and MB-231 cells, silenced for 2 days with Dox-inducible shRNAs. Data are represented as mean ± SEM. n.s., not significant. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 by unpaired parametric two-tailed t test.

Figure 3.. Silencing eIF4G2 mRNA in human MB-231 cell tumors does not impair primary tumor growth but strongly blocks metastasis

(A) NOD/SCID/γ mice orthotopically injected with 5 × 10⁶ MB-231 cells stably transformed to express Firefly luciferase and red fluorescent protein (RFP), Dox-inducible Nsi or shDAP5 RNAs. Silencing initiated by Dox addition to drinking water 28 days after tumor cell implantation, tumors ~100–150 mm³ in size. At 49 days, mice were sacrificed, tumors and lungs collected. (B) Growth of tumors was recorded every 7 days by quantitative caliper measurement. n = 8 or 9 mice per group repeated twice. Volumes of non-silenced control compared with DAP5-silenced tumors was not statistically significant. (C) Primary tumors weighed at excision 49 days. Non-silenced control compared with DAP5-silenced tumors not statistically significant. n = 7–9 mice/group. Representative of 2 trials. (D) Equal protein amounts of whole tumor lysates at 49 days post-excision, subjected to immunoblot analysis for DAP5 and control actin levels. n = 4 tumors (of 8 or 9 mice) per arm were evaluated. (E) Representative Firefly bioluminescent images from 8 or 9 mice per group (3 shown), silenced for Nsi or DAP5 as in (A) immediately prior to sacrifice at 49 days. Repeated 3 times. (F) Lungs excised, subjected to bioluminescent imaging for RFP. Representative images of 8 or 9 mice per group (4 shown) of mice silenced for Nsi or DAP5 as in (A). Repeated twice. (G) Quantification of metastatic burden in lungs of mice silenced for Nsi or DAP5 as in (A). Total RFP fluorescent flux quantified (photons [ps]/sec), representative of total lung tumor burden, at 49 days. n = 8 or 9 mice per group. Repeated twice. (H) Representative H&E-stained images of lungs of 8 or 9 mice per group harvested at 49 days. Arrows indicate dark staining tumor masses. (I) Quantification of number of lung metastases (mets) per field at 1003 magnification of H&E-stained lungs as shown in (H). A minimum of 5 fields per lung were quantified from 9 mice/group, mean plus SEM. (J) DAP5-silenced tumors have reduced levels of EMT proteins. Immunoblot of equal protein amounts from control Nsi and shDAP5-silenced tumors harvested at 49 days. Representative results of 6 tumors are shown. Data are mean plus SEM. Statistical significance was assessed by two-way ANOVA with Dunnett post-ANOVA test determination for analysis of repeated measures. n.s., not significant. ***p < 0.001. See also Figures S2 and S3.

Figure 4.. Silencing eIF4G2 mRNA in murine 4T1 cell tumors does not impair primary tumor growth but strongly blocks metastasis

(A) Schema of animal study. Balb/c mice subcutaneously injected in the flank with 5 × 10⁵ 4T1 cells stably transformed to express Firefly luciferase and red fluorescent protein (RFP), Dox-inducible Nsi or shDAP5 RNAs. Silencing initiated 8 days after tumor cell implantation, tumors at 75–100 mm³ in size, by Dox addition to drinking water. At 20 days, mice were sacrificed and tumors and lungs collected. (B) Equal protein amounts of whole tumor lysates at 20 days post-excision subjected to immunoblot analysis for DAP5 protein and control actin levels. n = 4 tumors per condition were evaluated. (C) mRNA extracted from tumors removed at 20 days and equal RNA amounts used to determine eIF4G2 mRNA levels normalized to Gapdh mRNA. n = 3 tumors per condition; mean and SEM are shown. (D) Growth of tumors recorded every 2–4 days by quantitative caliper measurement. n = 8 or 9 mice per group repeated twice. Volumes of Nsi control compared with DAP5-silenced tumors not statistically significant. (E) Representative Firefly bioluminescent images of 8 mice per group (4 shown), silenced for Nsi or DAP5 as in (A), immediately prior to sacrifice at 20 days. Repeated 3 times. (F) Lungs excised from mice and subjected to fluorescence imaging for RFP. Representative images of 8 or 9 mice per group (4 shown) of mice silenced for Nsi or DAP5 as in (A). Repeated twice. (G) Quantification of metastatic burden in lungs of mice silenced for Nsi or DAP5 as in (A). Total RFP fluorescence flux quantified (photons [ps]/sec), representative of total lung tumor burden harvested at 20 days. n = 7 or 8 mice per group. Repeated twice. (H) Quantification of eIF4G2 mRNA levels in Nsi and DAP5-silenced metastases in lungs of mice. n = 3 per group. Lungs excised, tumor cells dispersed, isolated by RFP FACS, subjected to qRT-PCR for eIF4G2 mRNA normalized to Gapdh mRNA. (I) Representative H&E-stained images of lungs of 8 mice/group harvested from mice bearing Nsi or shDAP5 tumors. Arrows indicate dark staining tumor masses. (J) Quantification of number of lung metastases (mets) per field at 100× magnification of H&E-stained lungs as shown in (I). A minimum of 5 fields per lung quantified, mean plus SEM of n = 8 mice per condition. Data are mean plus SEM. Statistical significance was assessed by two-tailed Student’s t test for unpaired experimental values (C and H), or two-way ANOVA with Dunnett post-ANOVA test determination for analysis of repeated measures with (D, G, and J). n.s., not significant. ***p < 0.001. See also Figure S4.

Figure 5.. DAP5 promotes breast cancer cell peri-tumoral stromal invasion and tumor angiogenesis and reduced metastatic disease survival

(A) MB-231 cells (1 × 10⁷) expressing RFP and Dox-inducible control Nsi or shRNA to DAP5 implanted in mammary fat pad of NOD/SCID/γ mice, Dox induced at 28 days, when tumors 100–150 mm³ in size, maintained for 20 days, tumors with stroma excised, embedded, sectioned for immunofluorescence microscopy for RFP (red) and DAPI (blue) stained nuclei. Representative images of RFP expressing MB-231 cells beyond the periphery of tumor, two different tumors analyzed, 3 fields each. Scale bar, 200 μm. Arrows indicate direction of tumor cell stromal migration away from tumor capsule. (B) Quantitation of images in (A). Quantitation determined from 2 independent tumors, 3 fields each (n = 6), using ImageJ software. (C) Tumors obtained as in (A) from non-silenced Nsi control and DAP5-silenced tumor specimens sectioned, IHC stained for CD31 to highlight tumor vasculature. Representative sections shown from 5 different tumors. Scale bar, 1,000 μm. (D) Quantitation of CD31 stained MB-231 cells from images in (C). Quantitation determined from 3 different tumors, 3 fields each. (E) Schema of animal study for survival determination with DAP5 tumor-specific silencing. Animals subcutaneously injected in flank with 1 × 10⁵ 4T1-Nsi or 4T1-shDAP5 cells. At 12 days, Dox added to the drinking water to induce eIF4G2 mRNA silencing. (F) Tumor growth recorded every 2–4 days by quantitative caliper measurement until 22 days as control mice began to die. n = 8–10 mice per group, repeated twice. (G) Survival of mice with 4T1-Nsi or 4T1-shDAP5 tumors silenced starting at 13 days. n = 15 mice/group. Animals either died or were sacrificed when terminally moribund. Statistical significance was determined using unpaired non-parametric log rank Mantel-Cox test. Data are mean with SEM. Statistical significance (B, D, and F) assessed using two-tailed Student’s t test.

Figure 6.. Silencing eIF4G2 mRNA in established metastases strongly reduces metastatic burden

(A) Schema of animal study. Balb/c mice subcutaneously injected in the flank with 1 × 10⁷ 4T1 cells stably transformed to express Firefly luciferase and RFP, Dox-inducible Nsi or shDAP5 RNAs. Tumors aseptically excised at 12 days, ~100–125 mm³ in size, Dox added to drinking water at 13 days to express shRNAs. At 23 days, mice were sacrificed, and lungs were collected. (B) Primary tumors weighed at excision at 12 days, weight of non-silenced control compared with DAP5-silenced tumors. n = 10–13 mice/group. (C) Growth of tumors recorded 6 and 12 days prior to surgical excision of tumors, by quantitative caliper measurement. n = 8 mice per group repeated twice. (D) Lungs excised from 2 mice per group at 12 days prior to shRNA induction and subjected to fluorescent imaging for RFP. Images show qualitatively similar metastatic tumor burden in lungs of mice prior to Nsi or DAP5 silencing. (E) Lungs excised from mice at 23 days, subjected to fluorescent imaging for RFP. Representative images of 8 mice per group (4 shown) for mice silenced for Nsi or DAP5 as in (A). n = 8 mice/group. Repeated twice. (F) Quantification of metastatic burden in lungs at 23 days of mice silenced for Nsi or DAP5 as in (A). Total RFP fluorescence flux quantified (photons [ps]/sec), representative of total lung tumor burden. n = 8–10 mice per group. Repeated twice. (G) Quantification of number of lung metastases (mets) per field at 100× magnification of H&E-stained lungs at 31 days. A minimum of 5 fields per lung quantified, mean plus SEM. n = 11 or 12 mice per condition. (H) Quantification of eIF4G2 mRNA levels in Nsi and DAP5-silenced metastases in lungs of mice at 31 days. n = 3 per group. Lungs excised, tumor cells dispersed, isolated by FACS gating on RFP and subjected to qRT-PCR for eIF4G2 mRNA normalized to Gapdh mRNA. Data are mean plus SEM. Statistical significance by two-way ANOVA with Dunnett post-ANOVA test determination for analysis of repeated measures. n.s., not significant. **p < 0.01 and ***p < 0.001.

Figure 7.. Overexpression of DAP5 promotes increased metastatic colonization

(A) Representative immunoblot of select translation factor proteins in equal amounts of lysates from less transformed parental MB-231 cells, and two increasingly more transformed and more metastatic variants, LM0 and ML2 cells. (B) Representative immunoblot of 3 independent studies of parental MB-231 cells and parental MB-231 cells stably transfected with DAP5 cDNA (MB-231 OE cells). (C) Representative light field microscopic images of trypan blue stained parental MB-231 cells and MB-231 OE cells. n = 3 independent plating of cells, images representative from 12 different fields chosen at random. (D) Matrigel Transwell invasion assays performed with parental MB-231 cells or MB-231 OE cells, carried out as in Figure 2E. Results represent the mean of invading cells/field with SEM from 3 independent studies. Statistical analysis by unpaired two-tailed t test. (E) Cell migration wound healing assay. performed with parental MB-231 cells and MB-231 OE cells carried out as in Figure 2F. Statistical analysis by unpaired two-tailed t test. (F) Mice were injected in the retro-orbital (RO) sinus with 10³ parental MB-231 cells or MB-231 OE cells, both expressing Firefly luciferase. Ten days later representative Firefly bioluminescent images were obtained from 3–5 mice per group (3 shown). Repeated 2 times. (G) Lungs excised at 10 days post-RO injection of mice described in (E), subjected to bioluminescent imaging for Firefly luciferase, repeated twice. (H) Quantification of metastatic burden in lungs of mice from (E). Total luciferase fluorescent flux quantified (photons [ps]/sec). *p < 0.05 by unpaired two-tailed t-test.