Translational downregulation of Twist1 expression by antiproliferative gene, B-cell translocation gene 2, in the triple negative breast cancer cells

Abstract

Twist1, a key transcription factor regulating epithelial-mesenchymal transition and cancer metastasis, is highly expressed in invasive cancers in contrast to the loss of BTG2^/TIS21 expression. Based on our observation that forced expression of BTG2^/TIS21 downregulated Twist1 protein expression without altering mRNA level, we investigated molecular mechanisms of the BTG2^/TIS21-inhibited Twist1 translation in the triple negative breast cancer (TNBC) cells and in vivo BTG2^/TIS21-knockout (KO) mice and human breast cancer tissues. (1) C-terminal domain of Twist1 and Box B of BTG2^/TIS21 interacted with each other, which abrogated Twist1 activity. (2) BTG2^/TIS21 inhibited translational initiation by depleting eIF4E availability via inhibiting 4EBP1 phosphorylation. (3) Expression of BTG2^/TIS21 maintained p-eIF2α that downregulates initiation of protein translation, confirmed by eIF2α-AA mutant expression and BTG2^/TIS21 knockdown in MEF cells. (4) cDNA microarray analysis revealed significantly higher expression of initiation factors-eIF2A, eIF3A, and eIF4G2-in the BTG2^/TIS21-KO mouse than that in the wild type. (5) BTG2^/TIS21-inhibited translation initiation lead to the collapse of polysome formation and the huge peak of 80s monomer in the BTG2^/TIS21 expresser, but not in the control. (6) mRNAs and protein expressions of elongation factors were also downregulated by BTG2^/TIS21 expression in TNBC cells, but much higher in both TIS21-KO mice and lymph node-positive human breast cancers. (7) BTG2^/TIS21-mediated Twist1 loss was not due to the protein degradation by ubiquitination and autophagy activation. (8) Twist1 protein level was significantly higher in various organs of TIS21-KO mice compared with that in the control, indicating the in vivo role of BTG2^/TIS21 gene in the regulation of Twist1 protein level. Altogether, the present study support our hypothesis that BTG2^/TIS21 is a promising target to combat with metastatic cancers with high level of Twist1 without BTG2^/TIS21 expression.

Fig. 1. Expression of BTG2^/TIS21 accompanies with loss of Twist1 protein, not mRNA, level.

a When 293TN cells were transfected with v5-Twist1 for 48 h, its expression was significantly reduced in the dose-dependent manner of BTG2-HA. Anti-v5, anti-HA, and anti-α-tubulin antibodies were applied for immunoblot analysis. b To examine regulation of v5-Twist1 mRNA expression by BTG2-HA, real-time PCR analysis was performed at 48 h. L32 was used for loading control. Note the absence of significant difference in the Twist1 mRNA expression by BTG2-HA (c) MDA-MB-231 and MDA-MB-468 cells transduced with either Ad-LacZ (100 moi) or Ad-TIS21 (100 moi) for 48 h were subjected to immunoblot analysis and the significant loss of Twist1 expression along with ZEB and Snail was observed in the BTG2^/TIS21 expresser. d The TNBC cells were subjected to RT-PCR analysis and mRNAs levels were measured. Note absence of significant changes in the levels of Twist1, Zeb1, and Snail between the BTG2^/TIS21 and the LacZ expressers. Immunoblots (a, c, d) are representative of three independent experiments. Data are expressed as mean ± SD from the more than two independent experiments

Fig. 2. Knockdown of BTG2^/TIS21 expression recovers Twist1 activity.

a Regulation of Twist1 expression by BTG2^/TIS21 gene was re-examined via measuring its target gene expression by real-time PCR after forced expression of BTG2^/TIS21 gene in MDA-MB-231 cells. Note the reciprocal regulation of E-cadherin and N-cadherin expressions without significant change of Twist1 mRNA expression by BTG2^/TIS21 gene. GAPDH served as a control for amplification. b The same experiment was performed in MDA-MB-468 cells and similar regulation was observed in the BTG2^/TIS21 expresser. c To confirm the activity of BTG2^/TIS21 gene in the regulation of Twist1 target genes, MDA-MB-231 cells were transfected with siControl or siTIS21 (100 nM) for 24 h, and then transduced with either Ad-LacZ or Ad-TIS21 (100 moi). Total cellular RNAs were isolated 24 h later and subjected to real-time PCR analysis, and the expressions of E-cadherin, N-cadherin, and BTG2^/TIS21 were measured along with GAPDH as an internal control. p < 0.05 was considered as statistically significant. Note the significant changes of E-cadherin and N-cadherin expressions by the level of BTG2^TIS21 gene expression. d The same experiment was performed in MDA-MB-468 cells and the similar regulation of Twist1 target gene expression by BTG2^TIS21 was also found in the cells. Immunoblot analyses reveal the knockdown efficiency of siTIS21 in the MDA-MB-231 (e) and MDA-MB-468 (f) cells. Anti-HA and anti-α-tubulin antibodies were applied for the experiment (n = 2). All data are expressed as mean ± SD after two independent experiments

Fig. 3. C-terminal region of TWIST1 interacts with BTG2 box B domain.

a In vitro immunoprecipitation (IP) analysis was performed in 293TN cells with anti-HA antibody. Pulldown of BTG2-HA using 1.0 µg of anti-HA antibody showed its interaction with v5-Twist1, depending on its concentration. Normal IgG was employed as a negative control of IP assay. Immunoblotting is the representative of three independent experiments. b Fractionation of the cytoplasm and nucleus of 293TN cells transfected with v5-Twist1 and BTG2-HA was performed, and the interaction of Twist1 and BTG2 was further examined by IP with 1.0 µg of anti-HA antibody. The data suggest that interaction was mainly observed in the nuclear fraction than in the cytoplasm. Star indicates nonspecific band. c In vitro GST-pulldown assay. GST and GST-BTG2 beads were incubated with 293TN cell lysates transfected with v5-Twist1 at 4 °C overnight. The beads were pulled down and subjected to immunoblot analysis with anti-v5 antibody. Note the direct interaction of BTG2 with v5-Twist1. d IP analysis was performed in 293TN cells after transfection with Flag-Twist1 (1.0 μg) and v5-BTG2 (1.0 μg) genes. In vitro interaction of Flag-Twist1 and v5-BTG2 was able to be shown by IP with 1.0 µg of anti-Flag antibody. Immunoblotting is the representative of three independent experiments. e For mapping BTG2 domain bound to Twist1, 293TN cells were transfected with either BTG2-HA wild type or deletion mutants (ΔBox-A, ΔBox B) along with v5-Twist1 gene. IP analysis performed with anti-HA antibody found that overexpression of ΔBox-B mutant significantly reduced BTG2-Twist1 interaction, indicating that box B in BTG2-HA interacted with v5-Twist1 protein. *The reduced interaction between BTG2-HA and v5-Twist1. BTG2 with ΔBox-B also abrogated interaction with endogenous cNOT7. f IP analysis with anti-v5 antibody. Procedure was the same as described in the e. Note the star revealing significant loss of the interaction. g To investigate the domain in Twist1 interacting with BTG2-HA, 293TN cells were transfected with either Twist1 wild type (1–202) and its deletion constructs (1–121, 1–161) along with BTG2-HA. IP performed with anti-HA antibody showed that the wild-type Twist1, but not the deletion mutants, could interact with BTG2-HA. h In vitro interaction analysis by GST-pulldown assay. GST and GST-BTG2 beads were incubated with 293TN cell lysates transfected with either wild type or the deletion constructs of Twist1 gene at 4 °C overnight. When the beads were subjected to immunoblot analysis with anti-Flag antibody, the interaction of BTG2 with wild-type Twist1, but not deletion mutants, was observed. All data clearly indicate the in vitro interactions of C-terminal region of Twist1 with BTG2 box B domain. Blots (b, c, and e–g) are representative of two independent experiments

Fig. 4. BTG2^/TIS21-induced loss of TWIST1 expression is not due to protein degradation.

a MG132 treatment failed to protect the BTG2^/TIS21-induced Twist1 loss. MDA-MB-231 (3 × 10⁵/60 mm dish) cells were transduced with either Ad-LacZ (100 moi) or Ad-TIS21(100 moi) virus for 48 h and then treated with MG132 (10 µM) for various time points to block proteasome activity. Twist1 protein was accumulated by the treatment in the LacZ expresser, whereas it was failed in the BTG2^/TIS21 expresser as opposed to accumulation of BTG2^/TIS21 expression that served as a positive control. α-Tubulin served as a loading control. b BTG2^/TIS21-induced Twist1 loss cannot be protected by NH₄Cl treatment. MDA-MB-231 (3 × 10⁵ cells) cells transduced with either Ad-LacZ or Ad-TIS21 were treated with 10 mM of NH₄Cl to block lysosome activity before immunoblot analysis at the time points. Twist1 protein was accumulated in the LacZ expresser by the treatment; however, it was lost in the BTG2^/TIS21 expresser. NH₄Cl treatment also failed to alter the expression of p62. NH₄Cl increased LC3B, the autophagosome-associated lipid form of LC3, which accumulates if their lysosomal degradation is inhibited, which served as a positive control. α-Tubulin served as a loading control. c Ubiquitination analysis. To investigate whether BTG2^/TIS21-induced TWIST1 loss is regulated by ubiquitination or not, 293TN cells were transfected with either v5-Twist1 (1.0 μg) and/or BTG2-HA (1.0 μg) before IP analysis with anti-v5 antibody and immunoblot assay with anti-ubiquitin antibody. v5-Twist1 expression was slightly increased in the vector expresser by MG132 treatment (compare lane 4 vs. lane 3 in the v5-panel); however, it was rather decreased in the BTG2^/TIS21 expresser (compare lane 6 vs. lane 5 in the v5 panel). The interaction between v5-Twist1 and BTG2^/TIS21-HA was almost similar and there was no difference in the ubiquitination after MG132 treatment. d No alteration of autophagy signal in the BTG2^/TIS21 expresser vs. LacZ control. MDA-MB-231 cells transduced with either Ad-LacZ or Ad-TIS21 were subjected to immunoblot analysis, to examine p62 and LC3B conversion. α-Tubulin as loading control. Blots are a representative of two independent experiment

Fig. 5. BTG2^/TIS21-induced Twist1 loss is associated with the failure of polysome formation.

a Twist1 expression was significantly induced by knockdown of endogenous BTG2^/TIS21 expression. To confirm the effect of BTG2^/TIS21 on the Twist1 protein loss, TIS21^/BTG2 expression in the wild-type MEF was removed by transfection with siTIS21 (12.5~50 nM) for 2 days and then analyzed by immunoblotting with anti-Twist1 antibody. The induction of Twist1 expression appeared in the siTIS21-dependent manner, compared with that in the control. In the same cells, p-eIF2α level maintained in the siControl-transfected cells was significantly reduced by knockdown of BTG2^/TIS21 expression. Total eIF2α levels remained unaltered. α-Tubulin served as a loading control. RT-PCR analysis showed that Twist1 mRNA level was unchanged. TIS21 knockdown was analyzed by RT-PCR and GAPDH served as a control. b Wild-type MEF cells overexpressed with either eIF2α or eIF2α-A/A (S⁵¹A/S⁵¹A) mutant were transfected with siTIS21-RNAs (50 nM) and siControl-RNAs (50 nM), and the cells were subjected to immunoblot assay in 24 and 48 h of the transfection. Knockdown of BTG2^/TIS21 expression by siTIS21 significantly induced Twist1 expression compared with that in the siControl (lanes 2 and 3 vs. lane 1); however, Twist1 expression was consistently high in the MEF cells with eIF2α-A/A mutant expresser (lanes 5 and 6 vs. lane 4). The phosphorylation of eIF2α completely disappeared by the knockdown of BTG2^/TIS21 expression for 48 h, suggesting that BTG2^/TIS21 expression might maintain the p-eIF2α level that inhibits translation of Twist1. c, d Real-time qPCR analyses revealing mRNA expressions of BTG2^/TIS21 (c) and Twist1 (d) in the MEF cells transfected with wild-type and mutant eIF2α. Transfection of MEF with siTIS21 markedly reduced BTG2^/TIS21 expression; however, Twist1 transcription was not significantly changed by BTG2^/TIS21 knockdown. Relative expressions were presented based on that of GAPDH. e Polysome profiling assay; to assess a general translational control by BTG2^/TIS21 gene, TNBC cells transduced with either Ad-LacZ or Ad-TIS21 for 48 h were subjected to polysome profiling analysis. Polysome formation in the BTG2^/TIS21 expresser was collapsed after a huge peak of the 80S monomer, whereas the profile was well maintained in the LacZ control. Numbers in X axis (1–12) indicate fractions collected up to 500 μL per each tube. f Real-time PCR analysis; to analyze abundance of the Twist1 mRNAs in each fraction, real-time PCR analysis was performed. The amount of Twist1 mRNA was higher in the 2–5 fractions of the BTG2^/TIS21 expresser; however, it was much less in the 6–12 fractions compared with those in the LacZ expresser. In contrast, levels of GAPDH mRNA were similar between the LacZ and BTG2^/TIS21 expressers in each fraction. Blots are representative of three independent experiments. All data are expressed as mean ± SD after two independent experiments

Fig. 6. Downregulation of elongation factor expression by BTG2^/TIS21 gene in MDA-MB-231 cells.

a When the cells transduced with either Ad-LacZ or Ad-TIS21 were analyzed by real-time PCR, mRNA expressions of eEF1α, eEF1β2, eEF1γ, and eEF2 were significantly reduced in the BTG2^/TIS21 expresser than those in the LacZ control. GAPDH served as an internal control of transcription. All data are expressed as mean ± SD after three independent experiments. b To validate the activity of BTG2^/TIS21 gene regulating the transcription of elongation factors, knockdown of the exogenously expressed TIS21 gene was performed by transfection with either siControl or siTIS21 (100 nM) for 24 h after adenovirus transduction for 24 h. Real-time PCR analysis showed significant recovery of the BTG2^/TIS21-induced transcriptional repression by TIS21 knockdown in the cells. GAPDH served as an internal control. p < 0.05 was considered as statistically significant. All data are expressed as mean ± SD after two independent experiments

Fig. 7. In vivo activity of BTG2^/TIS21 gene in the TIS21-knockout mice and human breast cancers.

a To analyze in vivo regulation of Twist1 expression by BTG2^/TIS21 gene, five organs were extracted from each TIS21-KO (six heads) and the WT (six heads) female mice, and then subjected to immunoblot analyses with 40 µg/lane in each sample. When the ratios of Twist1/α-tubulin expression were examined by Image J software, the level was significantly higher in the TIS21-KO mice than that in the WT (n = 30 organs in each group). b Relapse-free survival vs. Twist1 expression was analyzed from open data in the HER2+ and Basal-type breast cancers. Higher expression of Twist1 significantly decreased the relapse free survival of HER2+ (p < 0.05) and Basal-type (p < 0.01) breast cancers. c Relapse-free survival vs. BTG2 expression was analyzed from open data in the HER2+ and Basal-type breast cancers. Higher expression of BTG2 significantly increased the relapse free survival of the HER2+ (p < 0.001) and Basal-type (p < 0.001) breast cancer patients

Fig. 8. Summary depicting the effects of BTG2^/TIS21 gene on the downregulation of Twist1 translation.

Reciprocal immunoprecipitation analysis confirmed the interaction of C-terminal of Twist1 to BTG2^/TIS21 box B domain, which had been screened by protein chip analysis. In the present assay, cNOT7 binding to BTG2^/TIS21 was employed as the interaction control. Adenoviral transfer of BTG2^/TIS21 gene into TNBC cells significantly reduced Twist1 protein, but not mRNA, expression and exhibited inhibition of Twist1 activity regulating E-cadherin and N-cadherin expressions. BTG2^/TIS21-mediated Twist1 protein loss was due to the failures of protein translation by inhibiting cap-dependent initiation via reduced mTORc1 activity and p-eIF2α maintenance, and by collapse in 80S polysome formation. That could lead to failure of protein translation at the initiation step. In addition, BTG2^/TIS21 reduced in vivo and in vitro expressions of eukaryotic elongation factor 1 (eEF1) isoforms and eEF2. All the mechanisms could be confirmed by cDNA microarray data, RT-qPCR, and immunoblotting analyses in the TNBC cells, BTG2^/TIS21-KO mice, and human breast cancer tissues. Taken altogether, BTG2^/TIS21 gene inhibited the initiation and elongation steps of Twist1 translation. Thus, cancer progression including EMT phenomenon can be reduced in the cells with BTG2^/TIS21 high expresser. In conclusion, the BTG2^/TIS21-mediated Twist1 protein loss exhibited better prognosis in the relapse-free survival of malignant breast cancers than that in the lower BTG2^/TIS21 and higher Twist1 expressers