Loss of O-GlcNAcylation in cardiac myocytes triggers the integrated stress response, contributing to heart failure

Abstract

Heart failure (HF) is a significant global health problem, affecting an estimated 64 million people worldwide. At the core of HF is the progressive dysfunction and irreversible loss of cardiac myocytes. O-GlcNAc transferase (OGT) is a conserved enzyme that catalyzes the addition of N-acetyl-glucosamine (GlcNAc) to serine or threonine residues of intracellular proteins. This dynamic protein modification, termed O-GlcNAcylation, has been implicated in nutrient sensing, metabolic regulation and stress adaptation. The integrated stress response (ISR) is a pathway that enables cells to rapidly respond to acute environmental changes and cell damage. During ISR, the translation factor eIF2α is phosphorylated, shutting down general translation but favoring the rapid production of stress-adaptive proteins. However, prolonged activation of the ISR can be detrimental to cells. In this study, we found that inhibiting OGT activates the GCN2/eIF2α/Atf4 signaling axis of the ISR. Activation of this pathway could be blocked by ISRIB, a small molecule that opposes the activity of phosphorylated eIF2α. Mice with inducible deletion of OGT in adult cardiomyocytes developed HF, and treatment with ISRIB significantly delayed the progression to HF. Our study reveals the regulatory impact of O-GlcNAcylation on the ISR and highlights a new potential strategy for alleviating HF.

Keywords: Atf4; GCN2; ISRIB; OGT; PERK; cardiomyocytes; cardiomyopathy; eIF2α; mTOR; translation initiation.

Figure 1

OGT inhibition in cardiac myocytes suppresses nascent protein synthesis and induces phosphorylation of the translational regulator eIF2α.A–B, primary neonatal rat ventricular myocytes (NRVMs) were incubated in serum-free medium for 24 h, then treated with Vehicle (0.1% DMSO), OGA inhibitor TMG (200 nM) or OGT inhibitor OSMI-1 (25 μM) for 6 h. Cells were exposed to phenylephrine (PE, 5 μM) in medium with L-azidohomoalanine (L-AHA, 25 μM) replacing L-methionine. After 4 h, cells were harvested and L-AHA reacted with biotin-alkyne. Streptavidin immunoblot band intensities were used to measure nascent protein synthesis. C, representative streptavidin blot of L-AHA incorporation into nascent proteins in NRVMs stimulated for four or 24 h with PE (5 μM) in the presence of Vehicle (0.1% DMSO), TMG or OSMI-1. D, schematic depicting eIF2α phosphorylation's effect on cap-independent translation and OGT inhibition's putative effect on eIF2α phosphorylation. E–F, Western blot analysis of phospho- and total eIF2α following OSMI-1 treatment (25 μM, 6 h). G–H, Western blot analysis and quantitation of eIF2α and S6 phosphorylation following OSMI-1 treatment (25 μM) for 1, 6, and 24 h. I, schematic illustrating the proposed mechanism: OSMI-1 inhibits OGT, leading to reduced protein O-GlcNAcylation. This reduction is associated with increased phosphorylation of eIF2α. Phosphorylated eIF2α selectively enhances the translation of Activating Transcription Factor 4 (ATF4). Elevated ATF4 protein levels then promote the transcription of ATF4 and C/EBP Homologous Protein (CHOP) genes, resulting in increased mRNA and protein levels of both ATF4 and CHOP. J–M, Western blot analysis of Atf4, CHOP, and O-GlcNAc following OSMI-1 treatment (25 μM, 6 h). N–S, qPCR analysis of mRNAs in DMSO- or OSMI-1-treated cells (25 μM, 24 h). Group sizes shown in respective bars. Statistical Analysis: Bars show mean ± standard error of the mean (SEM); dots denote individual biological replicates; numerals on bars indicate sample size (these conventions apply to all figures). Comparisons between vehicle (DMSO) and OSMI-1-treated samples in (F, K–S) used unpaired Student's t test. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Comparisons across treatment groups in B and H used one-way ANOVA followed by Tukey's post hoc test to identify differences between specific groups. Complete ANOVA statistics are reported in Table S4.

Figure 2

OGT inhibition activates mTOR phosphorylation and downstream signaling.A–D, NRVMs were exposed to OSMI-1 (25 μM, 6 h) and analyzed for phosphorylation of mTOR and its downstream targets S6 and 4EBP1 by Western blot. E–F, Western blot analysis and quantitation of phospho-mTOR and total mTOR following treatment with OSMI-1 (25 μM) for 1, 6, and 24 h. G–J, Western blot analysis of phospho-mTOR, total mTOR, phospho-S6, and total S6 in NRVM samples after treatment with Torin2 (1 μM). K–L, NRVMs were cultured in medium without serum for 24 h and then placed in L-AHA medium (DMEM depleted of L-methionine and L-cysteine and supplemented with 25 μM L-AHA) in groups containing the mTOR inhibitor Torin2 (1 μM) with or without OSMI-1 (25 μM). Six hours later, the cells were harvested and subjected to L-AHA-azide biotin-alkyne ‘click’ reaction, and the extent of L-AHA incorporation in each treatment group was assessed by streptavidin western blotting. M-N, Western blot analysis of phospho-eIF2α and total eIF2α after treatment for 6 h with OSMI-1 (25 μM), with or without Torin2 (1 μM). Bars represent means ± standard error. Comparisons between vehicle (DMSO) and OSMI-1-treated samples were performed using unpaired Student's t test. Comparisons among four groups were performed with one-way ANOVA followed by Tukey's post hoc test. ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S4.

Figure 3

ISRIB prevents the activation of downstream ISR effector Atf4 but does not rescue the deficit in nascent polypeptide synthesis caused by OGT inhibition.A, schematic depicting the proposed model: OSMI-1 inhibits OGT, leading to increased eIF2α phosphorylation and increases ATF4 protein levels through cap-independent translation. ISRIB, opposing the effects of phospho-eIF2α, is expected to reduce ATF4 translation and restore regular cap-dependent translation. B–D, Western blot analysis of phospho-eIF2α, total eIF2α, and ATF4 protein in NRVMs after treatment for 6 h with OSMI-1 (25 μM), with or without ISRIB (1 μM). E–F, qPCR analysis of Chop and Atf6 mRNAs in NRVMs treated for 24 h with or without OSMI-1 (25 μM) and with or without ISRIB (1 μM). G–H, Western blot and quantitation analysis of O-GlcNAc levels in NRVMs after treatment for 6 h with OSMI-1 (25 μM), with or without ISRIB (1 μM). I–J, NRVMs were cultured in medium without serum for 24 h and then placed in L-AHA medium (DMEM depleted of L-methionine and L-cysteine and supplemented with 25 μM L-AHA) in groups containing OSMI-1 (25 μM) and/or ISRIB (1 μM), with or without phenylephrine (PE, 5 μM). Six hours later, the cells were harvested and subjected to L-AHA-azide biotin-alkyne 'click' reaction, and the extent of L-AHA incorporation in each treatment group was assessed by streptavidin western blotting. Comparisons across groups were performed using one-way ANOVA with Tukey's post hoc test. ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S4.

Figure 4

Targeting upstream eIF2α kinases with small molecule inhibitors indicates the Eif2ak4/GCN2 is a key kinase responsible for eIF2α phosphorylation during OGT inhibition.A, schematic depicting the four eIF2α kinases: HRI (Heme-Regulated Inhibitor, EIF2AK1), PKR (Protein Kinase R, EIF2AK2), PERK (PKR-like ER Kinase, EIF2AK3), and GCN2 (General Control Non-derepressible 2, EIF2AK4), their respective small molecule inhibitors, and the potential impact of OGT inhibition by OSMI-1 in activating one or more of these kinases leading to increased eIF2α phosphorylation. B–E, Western blot analysis and quantitation of phospho-eIF2α, total eIF2α, ATF4, phospho-PERK, and total PERK in NRVMs treated with OSMI-1 (25 μM) with or without the GCN2 inhibitor GCN2iB (10 μM) for 6 h. F–I, Western blot analysis and quantitation of phospho-eIF2α, total eIF2α, ATF4, phospho-PERK, and total PERK in NRVMs treated with OSMI-1 (25 μM) with or without the PERK inhibitor GSK2606414 (10 μM) for 6 h. J–L, Western blot analysis and quantitation of phospho-eIF2α, total eIF2α, phospho-PERK, and total PERK in NRVMs treated with OSMI-1 (25 μM) with or without the PKR inhibitor PKR-IN (10 μM) for 6 h. M–N, Western blot analysis and quantitation of O-GlcNAc levels in NRVMs treated with OSMI-1 (25 μM) with or without the PKR inhibitor PKR-IN (10 μM) for 6 h. Comparisons across groups were performed with one-way ANOVA followed by Tukey's post hoc test. ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S4.

Figure 5

Knocking-down upstream eIF2α kinases with siRNA supports that EIF2AK4/GCN2 is a key kinase responsible for eIF2α phosphorylation, during OGT inhibition.A, schematic depicting the four eIF2α kinases: HRI (EIF2AK1), PKR (EIF2AK2), PERK (EIF2AK3), and GCN2 (EIF2AK4), their respective dsRNAs (Dicer substrate small interfering RNAs) used for knockdown, and the potential impact of OGT inhibition by OSMI-1 in activating one or more of these kinases leading to increased eIF2α phosphorylation. B, NRVMs were transfected with 20 nM non-targeting control (NTC) dsRNA, or dsRNA targeting rat HRI/EIF2AK1. 48 h later cells were processed for RNA isolation and mRNA quantification with qPCR. C–D, NTC and EIF2AK1 dsRNAs were transfected to NRVMs and 48 h later the cells were treated with or without OSMI-1 (25 μM) for 6 h. Samples were processed for Western blot to assess the levels of phospho- and total eIF2α. E–G, NRVMs were transfected with 20 nM NTC or dsRNA targeting rat PKR/EIF2AK2. 48 h after transfection/knockdown, the cells were incubated with or without OSMI-1 (25 μM) for 6 h, and samples were collected for Western blot to assess the levels of phospho-eIF2α, total eIF2α, and PKR. H–K, NRVMs were transfected with 20 nM NTC or dsRNA targeting rat PERK/EIF2AK3. After 48 h of transfection/knockdown, the cells were incubated with or without OSMI-1 (25 μM) for 6 h, and samples were collected for Western blot to assess the levels of phospho-eIF2α, total eIF2α, ATF4, phospho-PERK, and total PERK. L, NRVMs were transfected with NTC, or EIF2AK4/GCN2-targeting dsRNA. 48 h later cells were processed for RNA isolation and quantification of EIF2AK4/GCN2 mRNA with qPCR. M-P, NRVMs were transfected with 20 nM NTC or dsRNA targeting rat GCN2/EIF2AK4. After 48 h of transfection/knockdown, the cells were incubated with or without OSMI-1 (25 μM) for 6 h, and samples were collected for Western blot to assess the levels of phospho-eIF2α, total eIF2α, ATF4, phospho-GCN2, and total GCN2. Q–S, same as in M–P, but after 48 h of siRNA transfection the cells were exposed to Torin2 (1 μM) with or without OSMI-1 (25 μM) for 6 h, after which cells were harvested and protein extracted for western blots. Comparisons in B and L used unpaired Student’s t-tests. In all other panels, comparisons across groups were performed using two-way ANOVA with Tukey's post hoc test. ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S5.

Figure 6

Genetic ablation of OGT in cardiac myocytes in adulthood causes progressive heart dysfunction and failure that is associated with increased expression of eIF2α and GCN2 proteins.A, schematic representation of the genetic model for the selective knockout of O-GlcNAc transferase (OGT) in cardiac myocytes using an inducible Cre/loxP system. Cre recombinase is driven by the Myh6 promoter for cardiomyocyte-specific expression. Cre activity is tamoxifen-inducible due to the presence of two Mer domains (modified estrogen receptor) that promote nuclear localization of Cre upon binding to tamoxifen. Additionally, a tdTomato marker, inserted into the Rosa26 locus, was introduced by cross-breeding. The expression of tdTomato is driven by the ubiquitous CAG promoter (a hybrid promoter derived from the chicken β-actin promoter and the cytomegalovirus enhancer) but is preceded by a lox-stop-lox cassette that is removed only in the presence of Cre in cardiac myocytes. B, bright field and fluorescent confocal images from adult hearts negative or positive for Cre that were intraperitoneally injected with tamoxifen (20 mg/kg/day) for 4 days. The images show strong expression of tdTomato protein in cardiac myocytes of Cre-positive mice. Images were captured using a Nikon spinning disk confocal microscope (Nikon Ti-E) equipped with 5× and 10× objectives and a high-speed camera. tdTomato was excited with the 561 nm laser. Scale bars are 2 mm. C-G, Mice (5 male and one female per group, aged 12–13 weeks old) were injected with four daily intraperitoneal injections of tamoxifen (20 mg/kg/day), and hearts were collected 15 weeks later. Heart homogenates (20 μg/lane) were analyzed by Western blot for the expression of OGT, OGA, and overall O-GlcNAc levels. H, serial echocardiograms were obtained from flox-only control, Cre-only control mice, and OGT inducible cardiomyocyte specific knockout mice (OGT-icko) over the period of 15 weeks after tamoxifen injection. EF: Ejection Fraction, square symbols represent means ± standard deviation. I, gross morphological appearance of Control and OGT-icko hearts after 15 weeks of follow-up. J–L, heart and lung analysis in Control and OGT-icko hearts. Heart weight (HW), body weight (BW), tibia length (TL), and lung weight (LW) are measured. M-P, quantitative real-time PCR analysis of RNA extracted from Control and OGT icko hearts (15 weeks after tamoxifen injection) to assess OGT expression levels and markers of detrimental cardiac remodeling. Myh6: myosin heavy chain 6; Myh7: myosin heavy chain 7; Nppa: natriuretic peptide A. Q–S, heart homogenates (20 μg/lane) from Control and OGT-icko mice were analyzed by Western blot for the expression of O-GlcNAc transferase (OGT), eukaryotic initiation factor 2 alpha (eIF2α), and general control nonderepressible 2 (GCN2). Statistical Analysis: Bars represent means ± standard error. Comparisons between two groups were performed using an unpaired two-tailed Student’s t test. Comparisons across three groups (Panels J–L) were conducted using one-way ANOVA with Tukey's post hoc test. For the serial echocardiography analysis (panel H), a two-way ANOVA was used, followed by Tukey’s correction for multiple comparisons. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Tables S4 and S5.

Figure 7

Treatment with the ISR inhibitor ISRIB prevents the development of dilated cardiomyopathy in OGT-icko mice.A, timeline showing tamoxifen injection to produce Control and OGT-icko. The mice (male, aged 18 weeks old) were analyzed by echocardiography 2 weeks later and randomized to undergo implantation of osmotic pumps delivering ISRIB (0.5 mg/kg/day). Following the first 4-week period the ISRIB treatment continued for another 4 weeks with a second round of pump implantations. Serial echocardiograms were obtained before the first pump implantation and then every 2 weeks. B–F, each panel shows the results of ejection fraction (EF) at the beginning and then at 2, 4, 6, and 8 weeks of ISRIB infusion. G–K, Each panel shows the results of left ventricular internal diameter during diastole (LVID;d) at baseline, 2, 4, 6, and 8 weeks of ISRIB treatment. Four groups were used: Control with or without ISRIB and OGT-icko with or without ISRIB. The dashed line indicates the mean value for EF and LVID;d in control mice at the initial echo. L–N, Morphometric parameters of hearts from the four different groups at the end of the 8-weeks ISRIB experiment. Statistical Analysis: Bars represent means ± standard error. Comparisons across groups were performed using two-way ANOVA with Tukey's post hoc test. The number of mice per group is indicated in the bars ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S5.

Figure 8

ISRIB-mediated protection against OGT-cardiomyopathy associates with reduction in PERK and eIF2α protein levels and sustained increases in GCN2 and mTOR.A–D, Homogenates (20 μg/lane) were analyzed by Western blot for the expression of OGT, OGA, and overall O-GlcNAc levels. E–G, hearts were analyzed by Western blot for the expression of EIF2AKs PERK and GCN2. H–J, Western blot analysis of phospho- and total eIF2α, either as ratio or total eIF2α levels alone. K-M, Western blot analysis of phospho- and total mTOR in the same heart homogenates. Comparisons across three groups were performed using one-way ANOVA with Tukey's post hoc test. The number of mice per group is indicated in the bars ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Complete ANOVA statistics are reported in Table S4.

Figure 9

Working model summarizing the findings of the study and highlighting the relationship between O-GlcNAcylation and ISR activation. Lowering O-GlcNAcylation triggers the ISR via the GCN2/eIF2α signaling axis. This causes a reduction in general translation but accumulation of transcription factor Atf4. Activation of GCN2 by low O-GlcNAcylation requires mTOR. Amino acid levels are not substantially decreased in OGT-inhibited cardiomyocytes, except for glutamine. It is thus unlikely that a surge in uncharged tRNAs is activating GCN2 in this context. Concurrently with glutamine reduction, glutamate and aspartate levels increase, highlighting a potential metabolic switching in mitochondria. Lowering O-GlcNAcylation causes the downregulation of Gfpt1 mRNA which encodes the rate-limiting enzyme of the HBP Gfat1. Hearts with cardiomyocyte-specific knockout of OGT exhibit severe dysfunction (cardiomyopathy) over the course of 15 weeks of knockout that is significantly delayed by ISRIB, highlighting the importance of the ISR in the development of OGT cardiomyopathy.