Regulation of Hypoxia-Inducible Factor 1α during Hypoxia by DAP5-Induced Translation of PHD2

Abstract

Death-associated protein 5 (DAP5) is an atypical isoform of the translation initiation scaffolds eukaryotic initiation factor 4GI (eIF4GI) and eIF4GII (eIF4GI/II), which recruit mRNAs to ribosomes in mammals. Unlike eIF4GI/II, DAP5 binds eIF2β, a subunit of the eIF2 complex that delivers methionyl-tRNA to ribosomes. We discovered that DAP5:eIF2β binding depends on specific stimuli, e.g., protein kinase C (PKC)-Raf-extracellular signal-regulated kinase 1/2 (ERK1/2) signals, and determines DAP5's influence on global and template-specific translation. DAP5 depletion caused an unanticipated surge of hypoxia-inducible factor 1α (HIF-1α), the transcription factor and master switch of the hypoxia response. Physiologically, the hypoxia response is tempered through HIF-1α hydroxylation by the oxygen-sensing prolyl hydroxylase-domain protein 2 (PHD2) and subsequent ubiquitination and degradation. We found that DAP5 regulates HIF-1α abundance through DAP5:eIF2β-dependent translation of PHD2. DAP5:eIF2-induced PHD2 translation occurred during hypoxia-associated protein synthesis repression, indicating a role as a safeguard to reverse HIF-1α accumulation and curb the hypoxic response.

Keywords: DAP5; HIF-1α; PHD2; eIF2β; hypoxia; translation initiation.

FIG 1

Phorbol ester stimulation promotes DAP5:eIF2β binding. (A to C) HEK293 cells expressing Flag-tagged DAP5, eIF4G1-Ct, DAP5(1-790), and DAP5(E862K) (A), Flag-DAP5 (B), or Flag-eIF2β (C) were induced with dox (16 h) before stimulation with TPA, as shown. Cell lysates were analyzed by immunoblotting or subjected to anti-Flag immunoprecipitation/immunoblotting. Data bars in panel A depict the mean ratio of eIF2β to Flag signal after 240 min of TPA from three assays. Experiments were normalized by setting the Flag-DAP5 (wt) ratio for each experiment to 1; error bars represent the SEM. *, significant P value (P < 0.05). (D) Schematic representation of DAP5 and eIF4GI(683-1600) domain arrangements, protein interactions, and major Ser/Thr phosphorylation sites. (E) HEK293 cells were dox induced for Flag-DAP5 expression (16 h), treated with TPA (240 min), harvested, and subjected to Flag-DAP5 IP. Immunoprecipitation-isolated complexes were digested with trypsin, followed by TiO₂ enrichment and LC-tandem MS (LC-MS/MS) analysis. Amino acid sequences of phosphopeptides identified by LC-MS/MS are shown; the highlighted amino acids before the asterisks indicate phosphorylated amino acids. The MASCOT ion score was calculated using the following equation: −10 log₁₀ (P), where P is defined as the absolute probability of the observed match being a random event (56). DAP5(T508) and DAP5(S902) were identified by this analysis by using a cutoff MASCOT score of 20. (F) Location of DAP5(T508)/(S902) relative to eIF4GI(S1186)/(S1597) phosphorylation sites, known to control TPA-induced association with MNK (22).

FIG 2

Phosphorylation of DAP5(S902) by CK2α is required for eIF2β binding. (A) S902/S1597 anchor consensus CK2α substrates at the DAP5/eIF4G1 C termini. (B) HEK293 cells expressing dox-inducible Flag-tagged DAP5, DAP5(S902A), or DAP5(S902D) were dox treated (16 h) before stimulation with TPA. Cell lysates were analyzed by immunoblotting or subjected to anti-Flag immunoprecipitation/immunoblotting. The ratio of eIF2β to Flag signal after TPA (4 h) was quantified across three repeat tests. HEK293 cells treated with TPA (4 h) were normalized to 1, with the SEM indicated. *, significant P value (P < 0.05). (C) Dox-inducible Flag-DAP5-expressing HEK293 cells were treated with dox (16 h) and TPA as shown. Lysates were tested as in panel B. All experiments were repeated three times with representative series as shown.

FIG 3

DAP5:eIF2β binding induced by MEK1-ERK1/2 signaling or during mitosis. (A) The top 24 predicted kinases for DAP5(T508) (http://www.phosphonet.ca; kinase predictor V2 scores). (B) Flag-DAP5(508T/A/E) immunoprecipitations show the specificity of an anti-p-T*PP antibody for T508. (C) HEK293 cells expressing Flag-DAP5 were dox induced (16 h) and pretreated with UO126, trametinib, BIRB 796, JNK inhibitor VIII, or DMSO(−) (2 h) prior to TPA stimulation. Cell lysates were analyzed by immunoblotting or anti-Flag immunoprecipitation/immunoblotting. (D) HEK293 cells expressing Flag-tagged DAP5, DAP5(T508A), or DAP5(T508E) were dox induced (16 h) before stimulation with TPA. Lysates were analyzed as in panels C and D. Data bars represent the average ratios of eIF2β to Flag immunoprecipitation signal at 4 h TPA, normalizing each experiment by setting the eIF2β/Flag ratio for DAP5(T508) to 1. (E) HEK293 cells with dox-inducible Flag-DAP5(T508E) were treated with dox (16 h), mock pretreated or pretreated with trametinib (2 h) prior to TPA stimulation and analyzed as in panel C. (F) HEK293 cells were treated with thymidine (4 mM; 16 h), followed by a medium change, and Ro3306, nocodazole, or DMSO(−), as shown. Dox was maintained (16 h) prior to harvest. Lysates were analyzed as in panels C to E. Data bars (right panels) represent mean ratio of p-T*PP/Flag and eIF2β/Flag, normalizing between experiments by setting the ratios for cells treated with thymidine block only to 1. (G) In vitro phosphorylation assay of Flag-DAP5 with recombinant JNK2 (control) or CDK1. Data bars represent the mean ratios of p-T*PP [p-DAP5(T508)] to Flag, normalizing between experiments by setting the −kinase/−ATP control to 1. All experiments were repeated three times; data bars represent the means from three tests; error bars denote the SEM. *, significant P value (P < 0.05).

FIG 4

DAP5 depletion attenuates global translation and paradoxically induces HIF-1α. (A) HeLa cell lines with dox-inducible eIF4GI or DAP5 depletion were treated with or without dox (96 h), stimulated with TPA as indicated, and treated with PMY (5 μM; 15 min) prior to harvesting. Puromycylation was stopped by washing cells with cold PBS and the addition of lysis buffer. (B) The results of mean quantification of the PMY/tubulin ratio from three experiments are shown, normalizing between experiments by setting the zero-time-point, no-dox value to 1. (C) HEK293 cells with dox-inducible DAP5 depletion combined with mock (−), wt DAP5, or DAP5(E862K) reconstitution were dox treated (96 h) prior to PMY. (D) Mean quantification of the PMY/tubulin ratio is shown for three experiments, normalizing by setting the no-dox ratio to 1. (E) HeLa cells with dox-inducible eIF4GI or DAP5 depletion were dox treated and TPA stimulated, and their lysates were subjected to immunoblotting as shown. (F) Mean HIF-1α, COX2, and MKP-2/tubulin ratios were quantified from three experiments, normalizing by setting the mock (eIF4GI)-depleted, no-dox ratio to 1. All assays were repeated three times, and a representative series for each panel is shown. Error bars denote the SEM. *, significant P value (P < 0.05).

FIG 5

DAP5 depletion induces HIF-1α by depressing PHD2 levels. (A) Schematic overview of posttranslational modifications and interactions of HIF-1α. (B) HeLa cells with dox-inducible DAP5 depletion were dox treated (96 h) and pretreated with CoCl₂ (4 h) before TPA stimulation. The cells were harvested for immunoblotting. The mean HIF-1α/tubulin ratio was quantified, normalizing between experiments by setting the no-dox, no-TPA control to 1. (C) HeLa cells with dox-inducible DAP5 depletion were dox treated (96 h) and lysed for immunoblotting with the antibodies shown. The mean PHD2/tubulin ratios were quantified, normalizing between experiments by setting the no-dox, no-TPA time point to 1. (D) HeLa cells with dox-inducible eIF4GI depletion were dox treated as shown and mock or TPA stimulated (240 min). The mean PHD2/tubulin ratio was quantified, normalizing by setting the 240-min TPA, no-dox ratio to 1 for each experiment. (E) The assay described for panel D was repeated with the addition of MG-132 pretreatment (5 μM; 1 h) prior to TPA stimulation. The mean OH-HIF-1α (P564)/HIF-1α ratio was quantified, normalizing by setting the no-dox value to 1. (F) HEK293 cells with dox-inducible knockdown/knock-in as described for Fig. 4C were dox treated as shown and analyzed by immunoblotting. The mean PHD2/tubulin ratios were quantified, normalizing by setting the no-dox values to 1. All data bars represent the means of three independent experiments; error bars display the SEM. *, significant P value (P < 0.05).

FIG 6

DAP5 is involved in PHD2 translation. (A and C) HeLa cells with dox-inducible DAP5 depletion were dox treated (96 h), and lysates were subjected to polysome profiling. (Top panel) A representative absorbance (260:280) profile of the fractionated gradient is shown in light blue with the peaks identified. Total RNA was purified from the fractions, RT-qPCR analyses for PHD2 (A) and GAPDH (C) mRNA were performed, and the percentages of the PHD2/GAPDH mRNA signal were determined for each fraction. This experiment was repeated twice and yielded similar results in each series; representative profiles are shown. Proteins from each fraction were TCA precipitated and analyzed by immunoblotting. (B) HeLa cells with dox-inducible DAP5 depletion were treated with dox (96 h) or left untreated. Total polyadenylated RNA was isolated from the cells and analyzed by RT-qPCR. The average of two assays with the uninduced controls was normalized to 1, and the SEM values are shown. Representative immunoblots from lysates are shown. (D) Schematic view of the ambiguously annotated 3′ and 5′ UTRs of PHD2. Alternative transcriptional start sites and polyadenylation signals (orange bars) are indicated (see the text for references).

FIG 7

DAP5 modulates AKT(T308) phosphorylation. (A) HEK293 cells with dox-inducible (endogenous) DAP5 depletion and either wild-type DAP5 or DAP5(E862K) reconstitution were treated with dox as shown (96 h). Lysates were subjected to immunoblotting with the indicated antibodies; the results of a representative test are shown. The p-AKT(T308)/AKT signal ratio from three independent series was quantified. The uninduced controls were normalized to 1. (B and C) Two malignant glioma cell lines (Du54 and U87) and one breast cancer cell line (Sum149) (B), as well as MEFs (C), were transiently transfected with siRNA targeting DAP5. Lysates were analyzed by immunoblotting. Mean HIF-1α/tubulin, PHD2/tubulin, and/or p-AKT(T308)/AKT1 signal ratios from three independent assays are shown, normalizing the control siRNA sample values to 1; error bars denote the SEM. HIF-1α (Du54 and Sum149 cells) and p-AKT (U87 cells) levels could not be quantified due to insufficient signal. *, significant P value (P < 0.05).

FIG 8

DAP5:eIF2β binding is induced during hypoxia and regulates the cellular hypoxia response through PHD2. (A) HeLa cells with dox-inducible DAP5 depletion, eIF4GI depletion, or DAP5 depletion, combined with PHD2 overexpression, were dox treated (96 h) as shown and placed in a chamber containing 0.5% O₂ for 6 or 24 h. Lysates were prepared and analyzed by immunoblotting. The mean HIF-1α/tubulin signal was quantified, normalizing between experiments by setting the (mock-depleted) HeLa cell 6-h value to 1. (B) U87 (top panels) and MEF (bottom panels) cells were transfected with DAP5 or control siRNA and cultured in 0.5% O₂. The mean HIF-1α/tubulin signal was quantified, normalizing between experiments by setting control siRNA samples to 1. (C) The HEK293 knockdown/knock-in cells described for Fig. 4C and 5F were treated with dox as shown and placed in 0.5% O₂. The mean HIF-1α/tubulin ratios were determined, normalizing by setting DAP5 depletion/DAP5 reconstitution samples to 1. (D) HEK293 cells with dox-inducible Flag-DAP5 expression were treated with dox (16 h); lysates were prepared and analyzed by immunoblotting or anti-Flag immunoprecipitation/immunoblotting as shown. The mean eIF2β/Flag ratio from immunoprecipitations and the mean PHD2/tubulin ratio from the input signal were quantified, normalizing by setting 0-h samples to 1. All data bars represent the averages from three independent tests; error bars indicate the SEM. *, significant P value (P < 0.05).