Excessive O-GlcNAcylation Causes Heart Failure and Sudden Death

Failing myocardium from model animals, and patients, is marked by increased protein O-GlcNAcylation (the attachment of O-linked β-N-acetylglucosamine [O-GlcNAc] moieties to cytoplasmic, nuclear, and mitochondrial proteins).¹ It is unknown whether excessive O-GlcNAcylation is a cause or consequence of cardiomyopathy. The hexosamine biosynthesis pathway is a metabolic sensor that uses glucose, amino acids, and fatty acids to synthesize uridine diphosphate N-acetylglucosamine (UDP-GlcNAc), the substrate for O-GlcNAc transferase (OGT). O-GlcNAc is cycled on and off proteins by the activity of 2 enzymes: OGT, which adds GlcNAc from UDP-GlcNAc to proteins; and O-GlcNAcase (OGA), which removes UDP-GlcNAc from proteins.² Dynamic changes in O-GlcNAc are important components of the stress response and appear essential given that constitutive OGT knockout is embryonically lethal.³ Furthermore, inducible loss of myocardial OGT in adult mice causes increased susceptibility to myocardial injury,² indicating that O-GlcNAcylation is necessary, and suggesting that elevated O-GlcNAcylation can be beneficial. Alternatively, excessive O-GlcNAcylation is also suspected to contribute to myocardial dysfunction in diabetes and hyperglycemia¹ and OGT inhibitors can reverse or prevent pathologic myocardial hypertrophy.¹ Recent work has highlighted the role of O-GlcNAcylation in cardiac injury through interaction with CaMKII (calcium/calmodulin-dependent protein kinase II) and Stim1 (stromal interaction molecule 1) and a beneficial role in heart failure through HDAC4 (histone deacetylase 4).^4–7 Major unresolved questions remain regarding the potential role of excessive O-GlcNAcylation in causing or contributing to cardiomyopathy.

The complexity of myocardial responses to pathologic stress, and lack of genetic tools to control O-GlcNAcylation levels in vivo, independent of glucose or pathologic stress, has limited understanding of the role of increased O-GlcNAcylation in cardiomyopathy. We developed novel mouse models to independently control O-GlcNAcylation levels in myocardium and directly test the hypothesis that excessive O-GlcNAcylation causes or contributes to cardiomyopathy. We report that transgenic myocardial OGT overexpression (OGT TG) causes increased O-GlcNAcylation, dilated cardiomyopathy, and premature death. In contrast, transgenic myocardial OGA overexpression (OGA TG) does not cause cardiomyopathy. However, OGA TG mice had reduced myocardial O-GlcNAcylation, and were protected against cardiomyopathy attributable to transverse aortic constriction (TAC) surgery, a model of acquired pathologic myocardial hypertrophy.^8,9 Interbreeding of OGT TG with OGA TG mice reduced cardiac O-GlcNAcylation toward wild-type (WT) levels, rescued dilated cardiomyopathy, and prevented premature death seen in the OGT TG mice. We identified reduced expression of genes and proteins important for oxidative phosphorylation and impaired energetics in OGT TG hearts; these patterns were restored to near WT levels in hearts from OGT TG×OGA TG interbred mice. Taken together, these data show that excessive O-GlcNAcylation is sufficient to cause severe cardiomyopathy, heart failure, and premature mortality. Our findings identify novel targets likely to explain, at least in part, the deleterious effects of excessive myocardial O-GlcNAcylation, and suggest that reducing myocardial O-GlcNAcylation could be a successful therapeutic approach for cardiomyopathy and heart failure.

Methods

The data that support the findings of this study are available from the corresponding author on reasonable request.

Animal Models

All animal studies were carried out in accordance with the guidelines of the Johns Hopkins Institutional Animal Care and Use Committee under protocol M017M290. No human studies were performed. Mice used in these studies were a mixture of male and female animals on a C57BL/6J background. Animals used in the majority of studies were 7 to 12 weeks of age unless otherwise noted in the figure legends.

OGT TG Mice

Human cDNA encoding the nucleocytoplasmic variant of the human OGT gene was fused with a C-terminal Myc epitope tag. The resulting construct was cloned into the pBS-αMHC-script-hGH vector for myocardial expression. Pronuclear injections of linearized DNA (digested with NotI) were performed in the Johns Hopkins Transgenic Mouse Core Facility and embryos implanted into pseudopregnant females to generate C57BL/6J F1 mice. Insertion of the transgene into the mouse genome was confirmed by polymerase chain reaction analysis (see the Data Supplement) using the forward primer 5′-GGA CTT CAC ATA GAA GCC TAG C-3′ and reverse primer 5′-CAC TGC GAA CAC AGT ACA AAT C-3′, producing a product of 500 base pairs.

OGA TG Mice

Human cDNA encoding the long form of MGEA5 (meningioma expressed antigen 5; OGA) was fused with an N-terminal hemagglutinin epitope tag. The resulting construct was cloned into the pBS-αMHC-script-hGH vector for myocardial expression. Pronuclear injections of linearized DNA (digested with NotI) were performed in the Johns Hopkins Transgenic Mouse Core Facility and embryos implanted into pseudopregnant females to generate C57BL/6J F1 mice. Insertion of the transgene into the mouse genome was confirmed by polymerase chain reaction analysis (supplement) using the forward primer 5′-TGGTCAGGATCTCTAGATTGGT-3′ and reverse primer 5′-TCATAAGTTGCTCAGCTTCCTC-3′, producing a product of 850 base pairs.

AC3-I TG and CaMKIIδ S280A Knock-In Mice

Experimental studies were performed on male and female mice with C57BL/6J background. C57BL/6J and mice lacking a functional nicotinamide adenine dinucleotide phosphate oxidase (p47^−/−) were purchased from The Jackson Laboratory. Our laboratory previously described the generation of AC3-I¹⁰ transgenic mice.

CaMKIIδ-S280A knock-in mice harboring a point mutation in the mouse CaMKIIδ gene to substitute serine 280 with alanine (S280A) were generated on a C57BL/6J background using CRISPR/Cas9 technology. A single-guide RNA (target sequence 5′-CTGTTGCCTCCATGATGCACAGG-3′) was designed to target CaMKIIδ. Synthetic single-stranded DNA for CRISPR-homology repair was designed to harbor mutations including S280A (TCC→GCG) and NsiI recognition site (ATGCAT). Genotyping of founder mice and generations of offspring was performed initially by both direct sequencing of polymerase chain reaction–amplified fragments and polymerase chain reaction genotyping from tail DNA with the following primers: forward, 5′-AGGAAATGCTTGCCAAAGTAGTG-3′; reverse, 5′-CCAGCACATACTGCCCTAGC-3′.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 8 software. Sample size and information about statistical tests are reported in the figure legends. Data are presented as mean±SEM. Pairwise comparisons were performed using a 2-tailed Student’s t-test. For experiments with more than 2 groups, data were analyzed by a 1-way analysis of variance followed by a Tukey post hoc multiple comparisons test. For the Kaplan-Meier survival analysis, data are represented as means±SEM; significance was determined using the log-rank (Mantel-Cox) test.

Procedures used for TAC; murine echocardiography; arrhythmia monitoring by electrocardiographic telemetry implant; Western blot; OGT and OGA activity assays; mitochondrial isolation; complex I, II, and IV activity assays; Seahorse mitochondrial bioenergetics measurements; and ventricular myocyte intracellular Ca²⁺ measurements were performed according to published methods^11–16 and are detailed in the Extended Methods in the Data Supplement.

Results

Myocardial-Targeted OGA Overexpression Protects Against Left Ventricular Hypertrophy and Heart Failure

Elevated myocardial O-GlcNAcylation has been reported in multiple models of cardiomyopathy.^1,17–21 We first investigated whether TAC, a validated model of pathologically increased left ventricular afterload,⁸ resulted in augmented myocardial O-GlcNAcylation (see Methods). We found robust elevation in total O-GlcNAcylation in hearts from C57BL/6J mice with TAC compared with sham-operated mice (Figure 1A and 1B). These results were consistent with previous reports of increased O-GlcNAcylation in hearts subjected to pathologic stress.^1,21 On the basis of these findings, we next investigated whether attenuation of O-GlcNAcylation during sustained cardiac stress could be beneficial.

• Download figure
• Download PowerPoint

To test the effect of reducing myocardial O-GlcNAcylation under conditions of pathologic stress, we developed OGA TG mice in which OGA expression was under control of the α-myosin heavy chain promoter (Figure Ia in the Data Supplement; see Methods).²² The OGA TG mice were born in normal Mendelian ratios. The O-GlcNAcylation levels in OGA TG hearts were lower compared with WT hearts at baseline (Figure 1C and 1D). The antibody used to detect OGA by Western blot preferentially recognizes human OGA such that the transgenic (human OGA) is overrepresented when compared with mouse OGA. Underpinning this observation, whereas human and mouse OGA homologs are 97% identical, the antibody used in this study was raised against the first 50 amino acids, in which the identity between these homologs is 78% identical. These observations preclude a quantitative comparison. A representative Western blot is shown comparing OGA levels in WT and OGA TG mice (Figure 1E) where OGA TG protein levels were loaded at 1/100th that of WT. Coomassie loading is shown in Figure Ib in the Data Supplement (upper panel) for blots assayed for OGA in WT and OGA TG animals. Consistent with the decreased O-GlcNAcylation measured in OGA TG mice (Figure 1D), we detected significant increases in the abundance of human OGA mRNA in OGA TG (Figure 1F) and in OGA activity (≈20-fold) in OGA TG heart lysates compared with WT heart lysates (Figure 1G). OGA and OGT expression are coupled by mechanisms that include changes in transcription and splicing,^23,24 so we examined the protein expression levels of OGT. We found increased OGT protein expression (Figure 1H and 1I; Coomassie loading is shown in Figure Ib in the Data Supplement [lower panel]), nonsignificant changes in OGT transcript levels (Figure 1J), and increased OGT activity (Figure 1K) in OGA TG lysates. Localization of OGA expression to the heart in OGA TG mice was confirmed by analyzing different tissue types (Figure Ic in the Data Supplement). The OGA TG mice had modestly but significantly increased heart weight/body weight ratios compared with WT littermate controls (Figure IIa in the Data Supplement), but no difference from WT mice in left ventricular function by echocardiography (Figure 1L and 1M), left ventricular dimensions (Figure 1N), or Nppa expression (Figure IIb in the Data Supplement), a marker of pathologic hypertrophy.²⁵ The OGA TG mice exhibited no premature mortality (Figure 1O). We interpreted these findings to suggest that myocardial OGA overexpression is effective at decreasing baseline O-GlcNAcylation but this change is well tolerated without deleterious effect on cardiac structure, function, or mortality.

Elevated cardiac O-GlcNAcylation is associated with pathologic hypertrophy and heart failure in animal models and is well described in patients with hypertension and aortic stenosis, conditions of increased left ventricular afterload.¹ On the basis of these associations, we next challenged OGA TG and WT littermate control mice with TAC or sham surgery (Figure 2A). The OGA TG mice showed significantly attenuated myocardial O-GlcNAcylation (Figure 2B and 2C) after sham and TAC surgery compared with WT mice after TAC surgery. Protein loading was similar between groups assessed for total O-GlcNAc levels, as analyzed by Coomassie total protein staining (Figure IIc in the Data Supplement). OGA protein levels were assessed with 1:1 protein loading comparing WT sham and WT TAC (25 µg of protein loaded per sample; Figure 2D and 2E) and with 1:1 protein loading comparing OGA TG sham and TAC (0.25 µg protein loaded per sample; Figure 2F). Coomassie is shown for OGA levels (Figure IId in the Data Supplement). OGT protein levels were increased in WT TAC mice compared with WT sham but not significantly different in OGA TG sham versus TAC animals (Figure 2G). Coomassie is shown for OGT levels (Figure IIe in the Data Supplement). OGA TG mice had less severe hypertrophy after TAC compared with WT mice after TAC (Figure 2H). OGA TG mice after TAC had no significant change in systolic function, unlike WT mice after TAC (Figure 2I). Both pathologic hypertrophy and heart failure after TAC have been associated with perturbation of mitochondrial energetics and increased O-GlcNAc is linked to changes in Complex I²⁶ so we assayed complex I function but found no significant difference between WT and OGA TG groups undergoing sham or TAC surgery (Figure 2J). Marker genes for hypertrophic myocardial remodeling, Nppa and Myh7, which were normal at baseline in OGA TG mice (Figure IIb and IIf in the Data Supplement), showed reduced expression (Figure 2K and 2L) compared with WT littermate controls after TAC. In contrast, left ventricular wall thickness and left ventricular ejection fraction were similar in both sham groups (Figure 2H and 2I). Taken together, we interpreted these results to support a view that excessive myocardial O-GlcNAcylation is associated with pathologic stress and that increased OGA expression can reduce excessive O-GlcNAcylation and protect against myocardial hypertrophy after TAC surgery.

• Download figure
• Download PowerPoint

Increased O-GlcNAcylation, Dilated Cardiomyopathy, Arrhythmias, and Premature Death in OGT TG Mice

Our findings up to this point showed that reducing O-GlcNAcylation may be beneficial to protect against pathologic hypertrophy attributable to TAC surgery. We next used an orthogonal approach to assess the role of increased O-GlcNAcylation in cardiomyopathy by developing an OGT TG mouse model. OGT TG mice were designed for myocardial-targeted OGT overexpression, using the α-myosin heavy chain promoter, similar to our OGA TG mice (Figure IIIa in the Data Supplement; see Methods). OGT TG and WT littermate mice were sacrificed at 8 weeks of age; OGT protein overexpression was exclusively localized to heart (Figure IIIb in the Data Supplement). Cardiac O-GlcNAcylation (Figure 3A and 3B) and OGT protein expression were increased compared with WT littermates (Figure 3C and 3D). As expected, OGT transcripts were increased in OGT TG mice compared with WT mice (Figure 3E). OGT activity in OGT TG hearts was significantly increased compared with WT hearts (Figure 3F). Cardiac OGA expression (Figure 3G and 3H) was not significantly increased over WT littermates. Coomassie protein loading controls are shown in Figure IIIc and IIId in the Data Supplement. Transcript expression of murine Oga was slightly increased in the OGT TG versus WT mice (Figure 3I) with OGA activity mildly increased in OGT TG compared with WT mice (Figure 3J). The OGT TG mice developed dilated cardiomyopathy (Figure 3K) with reduced left ventricular ejection fraction (Figure 3L) and increased left ventricular diameter (Figure 3M). Nppa transcript levels obtained from RNA sequencing were increased in OGT TG mice compared with OGA TG mice (Figure IIf in the Data Supplement). Left ventricular function began to decline after 6 weeks of age (Figure IIIe in the Data Supplement). By 8 weeks, OGT TG mice had significantly increased left ventricular mass, assessed by echocardiography (Figure IIIf in the Data Supplement) and by morphometric analysis (Figure IIIg in the Data Supplement). Echocardiography measurements demonstrated a dilated cardiomyopathy phenotype in OGT TG mice (Figure IIIh in the Data Supplement). Ages of mice used in these experiments are noted in the Figure legends. These findings illustrate that increased myocardial OGT with concomitant increase in O-GlcNAcylation were sufficient to cause progressive cardiomyopathy.

• Download figure
• Download PowerPoint

Sudden death is a major complication of many types of heart failure and OGT TG mice exhibited a striking pattern of premature mortality (Figure 3N). Arrhythmias are an important cause of premature mortality in heart failure.²⁷ To determine whether arrhythmias contributed to premature death in OGT TG mice, we surgically implanted electrocardiographic monitors in OGT TG and WT littermate mice (mice 20–22 weeks of age were used to determine cardiac electric activity preceding death). We detected bradycardia, spontaneous ventricular tachycardia, and ventricular fibrillation leading to death in OGT TG but not in WT control mice (Figure IVa in the Data Supplement, left panel). The OGT TG mice exhibited a higher arrhythmia score (see Methods¹¹; Figure IVa in the Data Supplement, right panel), confirming an increased arrhythmia burden in OGT TG compared with WT controls. These findings suggested that premature death in OGT TG mice was related, at least in part, to arrhythmias. An important proarrhythmic feature of failing heart muscle is increased intracellular Ca²⁺ leak from the sarcoplasmic reticulum to the cytoplasm.²⁸ Leak of Ca²⁺ from the intracellular Ca²⁺ stores can be detected as Ca²⁺ sparks through cytosolic Ca²⁺-activated fluorescent indicators.²⁹ We measured intracellular Ca²⁺ sparks (see Expanded Methods in the Data Supplement) in ventricular myocytes isolated from OGT TG and WT hearts and found increased Ca²⁺ spark frequency and size in the OGT TG ventricular myocytes compared with WT (Figure 4A through 4G). We next measured SERCA2A (sarcoplasmic endoplasmic reticulum Ca²⁺ ATPase) and PLN (phospholamban) expression in OGT TG and WT control hearts and found lower levels of SERCA2A and decreased phosphorylated PLN in OGT TG compared with WT hearts (Figure 4H and 4I). Excessive CaMKII activity can cause increased intracellular Ca²⁺ sparks,³⁰ cardiomyopathy,³¹ and arrhythmias.³¹ Furthermore, CaMKII may be activated by O-GlcNAcylation of S280, a residue at the intersection of the regulatory and catalytic domains.⁴ On the basis of these concepts, we initially hypothesized that CaMKII was a critical cardiomyopathic signal for OGT TG cardiomyopathy. To test this idea, we first interbred OGT TG mice with mice engineered with a knock-in replacement of S280 (S280A) on CaMKIIδ,³² the predominant CaMKII isoform in myocardium. S280A mice are born in Mendelian ratios, develop normally, and have heart morphometry and function indistinguishable from WT littermates.³² The OGT TG×S280A interbred mice had similar left ventricular dysfunction and dilation compared with OGT TG mice in a WT CaMKIIδ background (Figure IVb and IVc in the Data Supplement). We next considered the possibility that CaMKII was activated by excessive O-GlcNAcylation in OGT TG mice independent of S280. To test this idea, we interbred OGT TG mice with an established mouse model of myocardial CaMKII inhibition attributable to transgenic expression of a CaMKII inhibitory peptide AC3-I that inhibits all CaMKII isoforms.¹⁰ Mirroring the S280A phenotype, the interbred OGT×AC3-I mice exhibited severe left ventricular dysfunction and dilation (Figure IVb and IVc in the Data Supplement), similar to age-matched OGT TG mice (Figure 3l and 3M). Taken together, these findings did not support our initial hypothesis that CaMKII was an important contributor to cardiomyopathy in OGT TG mice.

• Download figure
• Download PowerPoint

Reduced Oxidative Phosphorylation in OGT TG Cardiomyopathy

O-GlcNAc can modify thousands of proteins,^23,24,26,33 many of which could potentially contribute to cardiomyopathy. We focused on mitochondrial proteins because excessive O-GlcNAcylation has been implicated in compromising oxidative phosphorylation²⁶ and failing myocardium is marked by depressed energetics.^34,35 We evaluated mitochondrial respiration and tested complexes I, II, and IV with a focus on complex I, on the basis of previous work highlighting complex I as a target for O-GlcNAcylation and altering mitochondrial respiration.^26,36,37 We evaluated the activity of complex I (Figure 4J and 4K) using blue native polyacrylamide gel electrophoresis³⁸ and normalized its activity to complex V protein expression (Figure 4L). We found marked reduction in complex I activity in OGT TG hearts compared with WT controls (Figure 4J and 4K). In contrast, the activity of complexes II and IV showed no significant change in activity between WT and OGT TG (Figure Va and Vb in the Data Supplement). These data show that OGT TG overexpression leads to increased O-GlcNAcylation and loss of complex I expression and activity, suggesting that OGT TG cardiomyopathy is attributable at least in part to depressed energetics.

No Evidence for Increased Myocardial Death or Deterioration of Intracellular Ca²⁺ Transients in OGT TG Hearts

Many types of acquired and genetic cardiomyopathies exhibit increased myocardial cell death³⁹ or depressed intracellular Ca²⁺ concentration transients.²⁸ We found no differences in myocardial cell death (Figure Vc in the Data Supplement) and only a small amount of fibrosis (Figure 4M). These findings suggested that the notable loss of myocardial performance in OGT TG hearts was not attributable to loss of myocardium or myocardial scarring. Heart muscle cells contract and relax under control of intracellular Ca²⁺ concentration transients, grading myofilament interactions to regulate muscle shortening.⁴⁰ We found that mechanically unloaded OGT TG ventricular myocytes had modest but significantly reduced resting (diastolic) cytosolic Ca²⁺ concentrations (Figure VIa and VIb in the Data Supplement) and faster decay of peak (systolic) cytosolic Ca²⁺ (Figure VIc in the Data Supplement) compared with WT controls. OGT TG and WT ventricular myocytes had similar intracellular Ca²⁺ transient amplitudes (Figure VId in the Data Supplement) and caffeine-releasable intracellular, sarcoplasmic reticulum, Ca²⁺ stores (Figure VIe in the Data Supplement). The findings suggested that, despite increased intracellular Ca²⁺ sparks (Figure 4A through 4G), systolic function in OGT TG was not impaired because of defective intracellular Ca²⁺ transients or paucity of sarcoplasmic reticulum Ca²⁺ reserve.

Rescue of Dilated Cardiomyopathy and Premature Mortality in OGT TG and OGA TG Interbred Mice

We interpreted our findings in the OGT TG mice to suggest that excessive O-GlcNAcylation levels were a direct cause of cardiomyopathy. However, we also considered that OGT overexpression could have unanticipated pathologic actions, independent of O-GlcNAcylation. To further test these possibilities, we interbred OGT TG and OGA TG mice. O-GlcNAcylation levels from OGT TG×OGA TG interbred hearts were similar to O-GlcNAcylation levels in WT hearts and significantly less than in OGT TG heart lysates (Figure 5A and 5B). Double transgenic mouse heart weights, adjusted for body weight, were less than OGT TG, and were not different from WT littermates (Figure 5C). Echocardiography at 6 to 8 weeks of age revealed significant improvement in left ventricular ejection fraction (Figure 5D and 5E) and left ventricular dilation (Figure 5F). Because of the robust expression of OGT in mice overexpressing the OGT transgene, signal intensity between WT and transgenic mice exceed the linear range of the imaging system. A representative Western blot is shown in Figure 5G comparing OGT levels in WT, OGT TG, OGA TG, and OGT×OGA TG mice, where OGT TG and OGT×OGA TG protein levels were loaded at 1/10th that of WT. Quantification is approximate and OGT expression is significantly elevated in the interbred mice. The rescue of O-GlcNAcylation levels and myocardial function in double transgenic mice occurred despite elevated levels of cardiac OGT in OGT×OGA TG mice (Figure 5G and 5H). OGT transcript levels in OGT×OGA TG mice were increased when compared with OGT TG and WT mice (Figure 5I). OGA protein expression loaded 1:1 in WT and OGT TG (Figure 5J and 5K) was not significantly different. OGA protein expression levels loaded 1:1 between OGA TG and OGT×OGA TG mice were not significantly different (Figure 5J and 5L). Transcript analysis is notable for increased murine and decreased human OGA transcripts in OGT×OGA TG mice compared with OGA TG hearts (Figure 5M). Protein loading assessed by Coomassie stain of heart lysates for OGT is shown in Figure 5N and for OGA in Figure 5O. We interpreted these data to suggest that OGT TG cardiomyopathy was attributable to significantly elevated O-GlcNAcylation levels rather than a nonspecific consequence of transgenic protein overexpression. OGT TG×OGA TG mice were protected from the increased premature death seen in OGT TG mice (Figure 5P). Similar to our experiments in OGT TG mice (Figure 4J through 4L), we evaluated mitochondrial function using blue native gel activity assays for complex I. We conducted mitochondrial functional studies in OGT TG isolated mitochondria before detectable reduction in heart function, using 6- to 8-week-old mice, in order to examine mitochondrial function and activity responses to OGT overexpression independent of overt cardiomyopathy. We found similar complex I activity in the interbred and WT mice (Figure 6A) normalized to complex V protein expression (Figure 6B). We then performed a complex I activity assay using spectrophotometric quantification and confirmed that complex I activity was decreased, by approximately half, in OGT TG compared with WT and recovered to WT levels in double transgenic mice (Figure 6C). We assessed oxygen consumption rates in isolated mitochondria in the 4 experimental groups (WT, OGT TG, OGA TG, OGT×OGA TG) using the Seahorse XF96 Analyzer and confirmed decreased complex I–linked respiration in OGT TG versus WT mice (Figure 6D). We did not detect differences between WT and OGA TG or between OGA TG and OGT×OGA TG mitochondria in the absence of adenosine diphosphate (state 2; Figure 6E). In contrast, we measured larger differences in state 3 respiration (ie, in the presence of adenosine diphosphate) among OGT TG, WT, and OGT×OGA TG cardiac mitochondria (Figure 6F). We did not detect a difference in the state 3 oxygen consumption rates between WT and OGT TG×OGA TG mitochondria, suggesting that complex I activity is equivalent in these groups. Taken together, these results support a concept that OGT TG cardiomyopathy is driven, at least in part, as a consequence of impaired metabolism and energetics attributable to loss of complex I expression and activity.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Transcriptional Reprogramming in OGT TG Hearts

Given the known role of O-GlcNAcylation modification in modulating transcriptional pathways²³ and the complexity of targets and pathways potentially under the influence of pathologic O-GlcNAcylation, we hypothesized that multiple gene programs were affected in OGT TG cardiomyopathy. We compared polyA transcriptomes by RNA sequencing using hearts from age- and sex-matched WT, OGT TG, OGA TG, and OGT TG×OGA TG interbred mice. Principal component analysis showed that each of these groups exhibited gene expression patterns that were more similar within than between groups (Figure 7A). To gain further insight into genes with the potential to drive O-GlcNAcylation cardiomyopathy, we focused on RNA sequencing data gene sets that were significantly altered in OGT TG compared with WT hearts and where these genes were repaired toward WT expression in the OGT×OGA TG interbred hearts. Gene and transcript expression levels (FPKM [fragments per kilobase of transcript per million] values) were computed with the tool Cuffdiff2 v.2.2.1 and imported into Partek GS v7.0, where we performed 2-tailed Student’s t-test analyses comparing gene expression changes (multiple-test false discovery rate–adjusted q value<0.05) between the OGT TG and WT group and the OGT×OGA TG group versus WT. We found that 2813 genes were significantly changed in the OGT TG versus WT hearts and 1798 individual genes showed significant expression differences between the WT and the OGT×OGA TG hearts. We applied QIAGEN Ingenuity Pathway Analysis and identified the top 5 significant canonical pathways and biological functions in OGT TG and OGT×OGA TG hearts using a regulation z score and an overlap P value (Fisher exact test; P<0.05). The most prominent functional pathways identified included inhibition of oxidative phosphorylation pathways (Figure 7B) in the OGT TG hearts compared with hearts from OGT×OGA interbred mice and activation of pathways involved in inflammation, inflammatory cell signaling, and transcription (Figure VIIa in the Data Supplement).

• Download figure
• Download PowerPoint

The prominent changes exhibited in oxidative phosphorylation pathways align with our data from blue native gels and complex I activity assays (Figure 6) that suggested impairment in mitochondrial energetics, with subsequent recovery in the interbred OGT TG×OGA TG mice. We compared the hallmark oxidative phosphorylation gene set defined by Liberzon et al²⁰ with genes identified by our RNA sequencing study and identified 177 genes out of 200 genes listed in this gene set. We found that the majority of these genes (highlighted in red) are downregulated in OGT TG hearts compared with WT hearts (Figure 7C) and less gene downregulation occurred in OGT×OGA TG compared with WT hearts (Figure 7D), suggesting that OGT TG hearts are deficient in expression of oxidative phosphorylation genes and OGT×OGA TG interbreeding recovered expression of critical oxidative phosphorylation–related genes toward WT levels. This observation was further supported by hierarchical clustering analysis (Figure VIIb in the Data Supplement) and scatter plot analysis (Figure 7E). These data appeared to confirm our experimental findings and support published work⁴¹ highlighting defective mitochondrial energetics in response to excessive O-GlcNAcylation.

Discussion

The association between increased O-GlcNAcylation with diverse forms of cardiac stress is well established.^42–44 However, to our knowledge, no direct causal relationship has been established between increased O-GlcNAcylation and cardiomyopathy. Mouse models with near elimination of myocardial OGT exhibited exaggerated responses to injury.² We interpret this important finding to indicate that some increase in O-GlcNAcylation, likely within an acute or subacute timeframe, is required for optimal cardiac responses to stress. Our results show therapeutic benefit from modest reduction of O-GlcNAcylation in OGA TG mice after TAC and severe cardiomyopathy, heart failure, and sudden death resulting from massive and chronic O-GlcNAcylation increases in OGT TG mice. Collectively, we interpret these data to strongly suggest that excessive myocardial O-GlcNAcylation contributes to cardiomyopathy. In contrast with our new data, most work supporting a connection between excessive O-GlcNAcylation and cardiomyopathy has focused on hyperglycemic conditions, including in models of type I diabetes.²⁶ Our study provides new evidence that exposure to chronic, excessive O-GlcNAcylation is sufficient to cause dilated cardiomyopathy and premature death, even in the absence of hyperglycemia, diabetes, or metabolic disease. This finding may be broadly important because excessive O-GlcNAcylation could add to other established cardiomyopathy mechanisms and because it suggests that reversal or prevention of excessive myocardial O-GlcNAcylation could be an innovative and effective therapeutic strategy for cardiomyopathy and heart failure.

We recognize that transgenic models may imperfectly represent the pathologic potential of pathways linked to protein overexpression and this caveat could apply to the OGT TG mouse. To our knowledge, all published data supporting a contribution of OGT overexpression to cardiomyopathy are from in vitro experiments performed using adenoviral expression of OGT on ventricular myocytes⁴⁵ or the inhibition of OGA using PUGNAc or Thiamet G.⁴⁶ No studies have examined the effect of increased OGT protein levels on the heart in vivo. However, the findings that OGT TG cardiomyopathy and excessive myocardial O-GlcNAcylation were rescued by interbreeding with OGA TG mice, resulting in very high levels of transgenic protein overexpression, strongly suggests that OGT TG cardiomyopathy was a specific consequence of excessive OGT enzymatic activity. Some investigators have proposed a nonenzymatic role for OGT, acting as a scaffold or chaperone,⁴⁷ and our results do not exclude this possibility. We do not yet know whether O-GlcNAc-–modified proteins are different in OGT TG hearts compared with WT hearts with increased O-GlcNAcylation attributable to pathologic stress. The OGA TG mice have mild myocardial hypertrophy but no other measured phenotypic differences compared with WT littermate controls. Our finding that OGA TG mouse hearts had less stress-induced O-GlcNAcylation and were resistant to TAC surgery suggests that transgenic overexpression of OGA is well tolerated and is capable of reversing pathologic O-GlcNAcylation.

Our RNA sequencing studies showed that a wide variety of transcripts were affected by increased O-GlcNAcylation through OGT overexpression. Expression of genes involved in metabolism was reordered toward WT levels in OGT TG×OGA TG hearts. The role of depressed energetics and deranged metabolism is a consistent finding in dilated cardiomyopathy in patients^19,35,39 and excessive O-GlcNAcylation is associated with defects in oxidative phosphorylation in part by actions in mitochondria.^{13,20,26,41,48} We found reduced expression of a number of genes encoding complex I proteins, loss of complex I proteins, and decreased complex I activity in OGT TG hearts. Complex I gene expression, complex I proteins, and complex I activity were remodeled toward WT levels in the OGT TG×OGA TG interbred hearts, suggesting that defective metabolism contributes to OGT TG cardiomyopathy.

In contrast to myocardial OGT overexpression, we found that myocardial OGA overexpression did not cause cardiomyopathy, and that OGA TG hearts were partly resistant to TAC-induced cardiomyopathy. These results suggested to us that reducing chronic O-GlcNAcylation elevation by OGA activation could be a novel therapeutic strategy for cardiomyopathy. The cardiomyopathy phenotypes in OGT TG hearts were notable for an absence of large amounts of fibrosis or myocardial death, features often associated with irreversible disease.⁴⁹ Thus, it may be that O-GlcNAcylation contributes to heart disease by mechanisms, including depressed energetics, that are reversible. An intriguing finding is enhanced arrhythmogenicity in the OGT TG mice with decreased levels of SERCA, phosphorylated PLN, and increased intracellular Ca²⁺ sparks, all features associated with failing hearts. It has been previously observed that prolonged periods of diabetes resulted in decreased SERCA protein levels.⁴⁶ Yokoe et al⁵⁰ have described inhibition of PLN phosphorylation by O-GlcNAcylation in a diabetic model of cardiomyopathy and our findings are complementary to these data. Heart failure of diverse etiologies is often comorbid with arrhythmia, wherein the regulation of SERCA plays a central role. Our findings highlight the relevance of developing O-GlcNAc-based therapies to serve as novel regulators of SERCA, with the potential to become potent heart failure and antiarrhythmic therapies. Given the public health consequences of cardiomyopathy and heart failure despite current treatments, the quest for improved therapeutic options remains an important goal. Our findings are suggestive that attenuation of excessive O-GlcNAcylation could improve cardiomyopathy, heart failure, and arrhythmias. These findings may serve as a springboard for the development of future therapies.

Acknowledgments

The authors thank Teresa Ruggle at the University of Iowa for help with graphic art.

Supplemental Materials

Expanded Methods

Data Supplement Figures I–VII

References 51–58

Footnotes