Modulation of CREB3L2-ATF4 heterodimerization via proteasome inhibition and HRI activation in Alzheimer's disease pathology

Abstract

Alzheimer's disease (AD) pathology includes transcriptional changes in the neurons, which are in part caused by the heterodimerization of two stress response transcription factors, CREB3L2 and ATF4. We investigated the role of proteasome inhibition and the eIF2α-kinase HRI in the formation of CREB3L2-ATF4 in neurons exposed to soluble oligomeric Aβ₄₂. While HRI activation increased ATF4 expression, it decreased CREB3L2 and CREB3L2-ATF4 levels. Proteasome inhibition, induced by Aβ₄₂, leads to increased levels of both transcription factors in the nucleus. These findings suggest that CREB3L2 levels are normally kept low due to rapid degradation, but proteasome inhibition in response to Aβ₄₂ disrupts this balance, increasing CREB3L2 and heterodimer levels. Activation of HRI not only reduced CREB3L2 and heterodimer levels but also prevented the transcriptional dysregulation of a CREB3L2-ATF4 target, SNX3. Our results suggest that manipulating the HRI pathway during proteasome inhibition could help restore normal gene expression in the context of AD-related protein accumulation.

Fig. 1. Proteasome inhibition induces formation of CREB3L2-ATF4 complexes in neurons.

A The 5’ uORF of CREB3L2 is conserved across various higher order vertebrates. B Induction of ER stress causes PERK-dependent CREB3L2 synthesis. Cortical neurons were cultured for 10 DIV and treated with thapsigargin (1 μM) and/or GSK2606414 (5 μM) for 5 h. Scrambled control or CREB3L2-targeting siRNA was applied for 36 h before treatments. n = 2 replicates; unrelated lanes crossed out. C-D Immunoblot analysis of nuclear ATF4 (C) and CREB3L2 (D) levels in cortical neurons treated with vehicle or Aβ₄₂ for 36 h. Means ± SEM of n = 4-5 independent biological replicates, normalized to HDAC1 expression; *p = 0.0153, **p = 0.0082, unpaired t-test. E PLA for CREB3L2-ATF4 in cortical neurons treated with Aβ₄₂ for 8 h. Means of means ± SEM of n = 5 independent biological replicates, 48-50 optical fields per condition. Number of puncta normalized to area of neuronal soma. Unpaired, two-tailed t-test. **p < 0.01. Scale bar, 5 µm. F-G Immunoblot analyses of ATF4 (F) and CREB3L2 (G) levels in nuclear lysates of vehicle (DMSO), thapsigargin (TG), or bortezomib (BTZ) treated cortical neurons. Means ± SEM of n = 7 independent biological replicates, normalized to HDAC1. One-way ANOVA followed by Dunnett’s multiple comparison test. ****p < 0.0001. H PLA of CREB3L2-ATF4 in cortical neurons treated with vehicle (DMSO), bortezomib (BTZ), or thapsigargin (TG) for 5 h. Means of means ± SEM of n = 4 independent biological replicates, 40 optical fields per condition. Number of puncta normalized to area of neuronal soma. One-way ANOVA followed Dunnett’s multiple comparison test. *p < 0.05. Scale bar, 5 µm. I Co-immunoprecipitation of CREB3L2 with anti-ATF4 antibodies or control IgG from lysates of cortical neurons treated with vehicle (DMSO), thapsigargin (TG), or bortezomib (BTZ) for 5 h. Means ± SEM of n = 5 independent biological replicates. One-way ANOVA followed by Dunnett’s multiple comparison test. **p < 0.001.

Fig. 2. Soluble oligomeric Aβ₄₂ causes reduced proteasome activity in neurons.

A Validation of the proteasome activity assay. Proteasome activity assays on lysates of cortical neurons treated with vehicle or Aβ₄₂ for 8 hours. Peptidase activity of the 20S proteasome was measured at 120 minutes using specific fluorogenic substrates (50 µM): Suc-LLVY-AMC (chymotrypsin-like; B), Z-LLE-AMC (caspase-like; C), or Z-ARR-AMC (trypsin-like; D). Means ± SEM of n = 7 (B) or 9 (C, D) independent biological replicates. Unpaired, two-tailed t-tests. *p < 0.05; **p < 0.01.

Fig. 3. Proteasome inhibition activates S2P-dependent cleavage and nuclear localization of CREB3L2.

A Subcellular distribution of a GFP-CREB3L2_S1P (full-length) fusion protein carrying a mutated S1P cleavage site with or without S2P co-expression in HEK293 cells. Scale bar, 25 μm. B Immunoblot analysis for full-length and cleaved (clv) GFP-CREB3L2 or GFP-CREB3L2_S1P in HEK293 cells expressing either S1P or S2P. C Immunoblot analysis of endogenous CREB3L2 processing in HEK293 cells treated with vehicle (DMSO), thapsigargin (TG), or bortezomib (BTZ) for 5 h; shown are n = 2 independent biological replicates. D Immunoblot analysis of HRP-tagged, full-length CREB3L2 processing in HEK293 cells treated with Aβ₄₂ in the presence of vehicle or nelfinavir (NF) for 2 h. E Quantitative immunofluorescence imaging of cortical neurons expressing full-length GFP-CREB3L2 incubated with Aβ₄₂ or vehicle (2 h). Cells were pre-treated for 30 minutes with NF or control before Aβ₄₂ stimulation. Nuclear-to-cytoplasmic GFP fluorescence ratios are presented as means ± SEM of 138-161 neurons per condition, obtained over n = 3 independent biological replicates. Holm-Šídák multiple unpaired t-tests, ****P-value < 0.000001. Scale bar, 25 μm. F Immunoblot analysis of nuclear CREB3L2 and ATF4 levels in cortical neurons treated with vehicle (DMSO) or bortezomib (BTZ) in the presence or absence of nelfinavir (NF). Means ± SEM of n = 6 independent biological replicates, normalized to HDAC1. Holm- Šídák multiple unpaired t-tests, ****p < 0.0001; *p < 0.05. G PLA of CREB3L2-ATF4 in cortical neurons treated with vehicle (DMSO) or bortezomib (BTZ) in the presence or absence of nelfinavir (NF). Means of means ± SEM of n = 6 independent biological replicates, 58-60 optical fields per condition. Number of puncta normalized to area of neuronal soma. Holm-Šídák multiple unpaired t-tests, *p < 0.05. H Immunoblot analysis of endogenous S2P levels in cortical neurons treated with vehicle (DMSO) or bortezomib (BTZ) obtained over n = 5 biological replicates, normalized to β-actin. Same membranes as analyzed in Fig. 1 D and F.

Fig. 4. Long-term HRI activation negatively regulates CREB3L2 levels.

A Immunoblot analysis of CREB3L2 and ATF4 protein levels in whole cell lysates of vehicle (DMSO)- or BtdCPU-treated cortical neurons. Means ± SEM of n = 5 independent biological replicates, normalized to β-actin. Unpaired, two-tailed t-test; *p < 0.05. Same membranes as analyzed in Fig. 1 D and F. B Immunoblot analysis of ATF4 levels in vehicle (DMSO) or BtdCPU-treated cortical neurons transduced with either shCtrl or shHRI. Means ± SEM of n = 8 independent biological replicates. Holm-Šídák multiple unpaired t-tests; *p < 0.05. C Immunoblot analysis of nuclear CREB3L2 levels in vehicle (DMSO)- or BtdCPU-treated cortical neurons. Means ± SEM of n = 4 biological replicates, normalized to HDAC1. Unpaired, two-tailed t-test; *p < 0.05. D Time course analysis of CREB3L2 expression in cortical neurons treated with DMSO or BtdCPU. Means ± SEM of n = 4 independent biological replicates. Holm-Šídák multiple unpaired t-tests; **p < 0.01.

Fig. 5. HRI activation interferes with CREB3L2-ATF4 signaling.

A Immunoblot analysis of CREBL2 and ATF4 in vehicle (DMSO) or bortezomib (BTZ)-treated cortical neurons transduced with either shCtrl or shHRI. Means ± SEM of n = 8 independent biological replicates. Two-way ANOVA; *p < 0.05, ***p < 0.001. ATF4 and CREB3L2 ratios of BTZ- to DMSO- treated samples in shCtrl compared to shHRI for each trial. B PLA of CREB3L2-ATF4 in cortical neurons treated with DMSO or BtdCPU and Aβ₄₂ or vehicle for 8 h. Means of means ± SEM of n = 3 independent biological replicates, 62–70 optical fields per condition for nuclear ATF4-CREB3L2 events. Holm- Šídák multiple unpaired t-tests, *p < 0.05. Scale bar, 5 µm. C Immunoblot analysis of CREB3L2, ATF4, and SNX3 in whole cell lysates of cortical neurons treated with vehicle or Aβ₄₂ in the presence of DMSO or BtdCPU. Means ± SEM of n = 5 independent biological replicates. Holm-Šídák multiple unpaired t-tests; *p < 0.05, **p < 0.01. SNX3 ratios of Aβ₄₂- over vehicle-exposed neurons compared between DMSO- and BtdCPU-treated. Paired t-test, two-tailed. *p < 0.05.