eEF1A2 promotes PTEN-GSK3β-SCF complex-dependent degradation of Aurora kinase A and is inactivated in breast cancer

Abstract

The translation elongation factor eEF1A promotes protein synthesis. Its methylation by METTL13 increases its activity, supporting tumor growth. However, in some cancers, a high abundance of eEF1A isoforms is associated with a good prognosis. Here, we found that eEF1A2 exhibited oncogenic or tumor-suppressor functions depending on its interaction with METTL13 or the phosphatase PTEN, respectively. METTL13 and PTEN competed for interaction with eEF1A2 in the same structural domain. PTEN-bound eEF1A2 promoted the ubiquitination and degradation of the mitosis-promoting Aurora kinase A in the S and G2 phases of the cell cycle. eEF1A2 bridged the interactions between the SKP1-CUL1-FBXW7 (SCF) ubiquitin ligase complex, the kinase GSK3β, and Aurora-A, thereby facilitating the phosphorylation of Aurora-A in a degron site that was recognized by FBXW7. Genetic ablation of Eef1a2 or Pten in mice resulted in a greater abundance of Aurora-A and increased cell cycling in mammary tumors, which was corroborated in breast cancer tissues from patients. Reactivating this pathway using fimepinostat, which relieves inhibitory signaling directed at PTEN and increases FBXW7 expression, combined with inhibiting Aurora-A with alisertib, suppressed breast cancer cell proliferation in culture and tumor growth in vivo. The findings demonstrate a therapeutically exploitable, tumor-suppressive role for eEF1A2 in breast cancer.

Fig. 1.. eEF1A2 interacts with and promotes Aurora-A protein degradation.

(A) Aurora-A (top) and eEF1A2 (bottom) were immunoprecipitated from MCF-7 (left) and T47D (right) cell extracts and immunoblotted as indicated. (B) Western blotting for eEF1A2 and Aurora-A in a panel of human breast cancer cell lines. (C to E) MCF7 and MDA-MB-453 cells were transfected with non-targeting control or pooled siRNAs against eEF1A2 (C). SUM159 cells were transduced with CRISPR-Cas9 construct harboring sgRNA for non-targeting control or two distinct sgRNAs targeting eEF1A2 (D). MCF10A and MDA-MB-453 cells were transfected with eEF1A2-V5 (E). Quantification of Aurora-A relative to β-actin is shown means ± SD of three independent experiments (C-E, bottom). **P< 0.01, ***P< 0.001 by unpaired Student’s t-test. (F) MCF7 and T47D cells were transfected with non-targeting control or pooled siRNAs against CUL1. Lysates were immunoblotted for the indicated proteins. (G) MCF10A cells were transfected with eEF1A2-V5 and treated with the proteasome inhibitor MG132 for 4 hours. Cell extracts were immunoprecipitated for Aurora-A, followed by immunoblotting of CUL1, eEF1A2, GSK3β and Aurora-A as indicated. (H and I) Fractionation lysates of MCF7 cells were immunoprecipitated with the Aurora-A antibody, followed by immunoblotting for polyubiquitin (Lys⁴⁸; H), and CUL1, eEF1A2 and GSK3β (I). C = cytoplasmic fraction; N = nuclear fraction. All blots (A to I) are representative of three independent experiments.

Fig. 2.. eEF1A2 promotes Aurora-A degradation through SCF complex at S/G2 phase.

(A) Assessment of the correlation between eEF1A2 and Aurora-A during cell cycle progression. MCF10A, MCF12A and MCF7 cells were synchronized using double thymidine block (DTB 0 h), released from DTB for 4 hours (DTB 4 hours), subsequently treated with monastrol, MG132, and released in fresh media as indicated timepoints (30 and 60 min). Below each blot, protein expressions at each cell cycle stage from three independent experiments were immunoblotted and measured and shown as means ± SD. ^nsP>0.05 (not significant), *P< 0.05, ***P< 0.001 by unpaired Student’s t-test. (B) MCF7 cells were harvested at multiple time points during cell cycle progression as indicated in (A). Aurora-A and eEF1A2 from specific timepoints were immunoprecipitated followed by immunoblotting as indicated. All blots (A and B) are representative of three independent experiments.

Fig. 3.. Competitive interaction of eEF1A2 between PTEN and METTL13 regulates Aurora-A protein stability.

(A) eEF1A2 (left) and PTEN (right) were immunoprecipitated from MCF7 cell extracts, followed by immunoblotting as indicated. (B) Lysates of subcellular fractions of MCF7 cells were immunoprecipitated with the antibodies of eEF1A2 (left) and PTEN (right), followed by immunoblotting as indicated. C, cytoplasmic fraction; N, nuclear fraction. (C) MCF7 cells were transfected with non-targeting control or eEF1A2 siRNAs (left). MCF10A cells were transfected with eEF1A2 tagged V5 (right). Reaction products were immunoblotted as indicated. Quantification of Aurora-A relative to β-actin in blots from three independent experiments is shown as means ± SD. (D) In vitro binding assay of GST-eEF1A2 with PTEN (left) and METTL13 (right). (E) Immunoprecipitation of eEF1A2 with me2-eEF1A2 (Lys⁵⁵), METTL13, PTEN, and Aurora-A were compared between MCF7 cells expressing empty vector or Myc-METTL13. (F) eEF1A2 Lys⁵⁵me2 and Aurora-A expression in SUM159 cells upon deletion of PTEN by CRISPR/Cas9. Quantification of Aurora-A relative to β-actin in blots from three independent experiments is as shown means ± SD. (G) Protein lysates from subcellular fractionations of sgControl- or sgPTEN-transfected SUM159 cells were immunoblotted (top) and quantitated for Aurora-A relative to either α-tubulin or PARP, shown as means ± SD from three independent experiments (bottom). W = whole cell lysate; C = cytoplasmic fraction; N = nuclear fraction. All blots (A to G) are representative of three independent experiments. * P< 0.05, **P< 0.01 byunpaired Student’s t-test.

Fig. 4.. GSK3β interacts with and phosphorylates Aurora-A as a substrate for eEF1A2-mediated SCF complex.

(A) GSK3β phosphorylation of priming (top left) and autophosphorylation (top right) sites in Aurora-A. Consensus sites on Aurora-A sequences across human, Xenopus, and mouse were underlined. Numbers indicate the amino acid residues. Blots, middle: MCF7 cells were transfected with Aurora-A phospho-deficient or phospho-mimetic mutants, in which serine/threonine residues were replaced with alanine or aspartic acid, respectively. Protein stability of the Aurora-A triple-mutants for Ser²⁸³/Ser²⁸⁴/Thr²⁸⁸ or the Ser³⁴² mutants were analyzed after cells were treated with cycloheximide (CHX, 50 µg/ml) at various timepoints. Proteins were collected from three independent experiments and immunoblotted as shown, with quantification of Aurora-A relative to β-actin ratio as means ± SD (bottom). *P< 0.05, **P< 0.01, ***P< 0.001 by unpaired Student’s t-test. (B) In vitro kinase assay of GSK3β on Aurora-A at Ser²⁸³ and Ser²⁸⁴. GST-Aurora-A WT or phospho-deficient S283A/S284A was used as the substrate for GSK3β. Reaction mixtures were then subjected to SDS-PAGE, transferred to nitrocellulose membrane, stained with Ponceau S (bottom), and immunoblotted for phosphoserine antibody (top). (C) MCF7 cells were transfected with vector or with eEF1A2-V5. Cell lysates were subjected to immunoprecipitation for Aurora-A, followed by immunoblots as indicated. (D) Flag-tagged Aurora-A wild-type, phospho-mimetic and phospho-deficient mutants of GSK3β phosphorylation and autophosphorylation sites were transfected into MCF7 cells. Whole cell lysates at each cell cycle stage were subjected to immunoprecipitation for Aurora-A, followed by immunoblots as indicated. (E) MCF7 cells were synchronized using double thymidine block and release for 4 or 8 hours, as indicated. Whole cell lysates at each cell cycle stage were subjected to immunoprecipitation for Aurora-A, followed by immunoblots as indicated. Expo = exponential cell; DTB = double thymidine block and release. All blots (A to E) are representative of three independent experiments.

Figure 5.. Deficiency of Eef1a2 or Pten promotes Aurora-A protein stability in vivo.

(A) Western blot analyses of indicated proteins in normal wild-type (control) and Eef1a2^+/− brain tissues from 3-week-old mice. (B) Western blot analyses of indicated proteins in mammary glands from 8-week-old Pten^F/F (control) and MMTV-Cre/Pten^Δ/Δ mice. (C and D) Representative IHC images of sections of mammary glands from (C) 8-week-old virgin Pten^F/F (control) and MMTV-Cre/Pten^Δ/Δ mice (C), and of mammary glands from Pten wild-type mice and mammary tumors from Pten single-copy transposon-bearing (Pten^SBm2/+/Rosa26^SB11/+/Blm^m3/m3) mice (D). Images are representative of N = 2 to 4 mice per genotype. (E and F) Immunoblotting analysis (E) and IHC imaging (F) for the indicated proteins in normal mammary glands and mammary tumors from MMTV-rtTA/TRE-Aurora-A mice. Images are representative of 2 normal and 4 tumor-bearing mice, quantification of Aurora-A signal relative to β-actin from 3 independent blots is shown (A, B, and E; right). Scale bars (C, D, and F): 50 µm. All blots (A, B, and E) are representative of and data are means ± SD of three independent experiments. *P< 0.05, **P< 0.01 by unpaired Student’s t-test.

Figure 6.. Correlation of low eEF1A2 and active AKT with high Aurora-A expression in breast cancer and the effects of targeting the pathway in a xenograft.

(A to D) Representative images of sections derived from MDACC breast carcinoma cohort (A) and an independent cohort (C) immunohistochemically stained with antibodies to eEF1A2 and Aurora-A. Scale bars, 50 µm. Histograms (B and D) depict the staining distribution pattern of eEF1A2 and Aurora-A from breast tissue microarray samples collected at MD Anderson Cancer Center (B; N = 143) and an independent cohort (D; N = 57). P values were calculated using Chi-Square test. 0, no staining; 1, weak staining; 2, moderate staining; 3, strong staining. (E and F) In vivo tumor growth (E) and tumor volume scatter plot of day 21 (F) of PTEN-deleted SUM159 xenografts grown subcutaneously in SCID mice treated with vehicle (black, N = 10), alisertib alone (50 mg/kg/day given by oral gavage 4 times a week; red, N = 10), fimepinostat alone (30 mg/kg/day by oral gavage 4 times a week; blue, N = 8), alisertib + fimepinostat (purple, N = 8). Data are means ± SEM from 8–10 mice per group. **P < 0.01, ***P < 0.001 by unpaired Student’s t-test.

Fig. 7.. Summary model.

Schematic showing a mechanism of Aurora-A degradation promoted by eEF1A2 in the S and G2 phases of the cell cycle (left) and the consequences of its inactivation or deficiency (right) in cancer cells. Combination therapy of alisertib and fimepinostat reveals therapeutic vulnerabilities in cancer cells with abundant expression of Aurora-A. Nd8, neddylation; P, phosphorylation; Ub, ubiquitination.