Abstract Background Lifestyle and metabolic diseases influence the severity and pathogenesis of cardiovascular disease through numerous mechanisms, including regulation via posttranslational modifications. A specific posttranslational modification, the addition of O‐linked β‐N acetylglucosamine (O‐GlcNAcylation), has been implicated in molecular mechanisms of both physiological and pathologic adaptations. The current study aimed to test the hypothesis that in cardiomyocytes, sustained protein O‐GlcNAcylation contributes to cardiac adaptations, and its progression to pathophysiology. Methods and Results Using a naturally occurring dominant‐negative O‐GlcNAcase (dnOGA) inducible cardiomyocyte‐specific overexpression transgenic mouse model, we induced dnOGA in 8‐ to 10‐week‐old mouse hearts. We examined the effects of 2‐week and 24‐week dnOGA overexpression, which progressed to a 1.8‐fold increase in protein O‐GlcNAcylation. Two‐week increases in protein O‐GlcNAc levels did not alter heart weight or function; however, 24‐week increases in protein O‐GlcNAcylation led to cardiac hypertrophy, mitochondrial dysfunction, fibrosis, and diastolic dysfunction. Interestingly, systolic function was maintained in 24‐week dnOGA overexpression, despite several changes in gene expression associated with cardiovascular disease. Specifically, mRNA‐sequencing analysis revealed several gene signatures, including reduction of mitochondrial oxidative phosphorylation, fatty acid, and glucose metabolism pathways, and antioxidant response pathways after 24‐week dnOGA overexpression. Conclusions This study indicates that moderate increases in cardiomyocyte protein O‐GlcNAcylation leads to a differential response with an initial reduction of metabolic pathways (2‐week), which leads to cardiac remodeling (24‐week). Moreover, the mouse model showed evidence of diastolic dysfunction consistent with a heart failure with preserved ejection fraction. These findings provide insight into the adaptive versus maladaptive responses to increased O‐GlcNAcylation in heart. Keywords: cardiac remodeling, diabetic cardiomyopathy, metabolism, mitochondria, protein O‐GlcNAcylation Subject Categories: Basic Science Research, Metabolism __________________________________________________________________ Nonstandard Abbreviations and Acronyms CaMKII calcium/calmodulin‐dependent protein kinase II CoA coenzyme A CS citrate synthase dnOGA dominant‐negative O‐GlcNAcase dnOGAh dominant‐negative O‐GlcNAcase overexpression HFpEF heart failure with preserved ejection fraction OGA O‐GlcNAcase O‐GlcNAc O‐linked β‐N‐acetylglucosamine OGT O‐GlcNAc transferase OXPHOS oxidative phosphorylation PDK pyruvate dehydrogenase kinase Clinical Perspective. What Is New? * We show that short‐term moderate increases in cardiomyocyte‐specific protein O‐GlcNAc levels is sufficient to initiate transcriptional reprogramming of gene expression consistent with changes seen in hypertrophy and heart failure. * Furthermore, long‐term, moderate increases in cardiac protein O‐GlcNAc levels led to cardiac hypertrophy, cardiac fibrosis, cellular remodeling, and diastolic dysfunction with maintained systolic function. * This study found that a sustained increase in cardiomyocyte protein O‐GlcNAc levels like that seen in diabetes, hypertrophy, or heart failure is sufficient to alter mRNA expression patterns for genes involved in metabolism, fibrosis, and antioxidant pathways, reducing mitochondrial oxidative phosphorylation activity and enhancing pathologic remodeling. What Are the Clinical Implications? * Protein O‐GlcNAc levels are increased in hearts with diabetes, hypertrophy, and heart failure; however, whether this plays a causal role remains unclear. * We examined cardiomyocyte‐specific effects of elevating protein O‐GlcNAcylation in hearts with short‐term (2‐week) and long‐term (24‐week) induction of a dominant‐negative O‐GlcNAcase mouse model, which allowed us to focus on O‐GlcNAcylation while ruling out the effects of O‐GlcNAc transferase overexpression or O‐GlcNAcase deletion. * The moderate sustained protein O‐GlcNAcylation leads to cardiac remodeling and dysfunction, providing novel insight into controversies about the adaptive versus maladaptive impact of O‐GlcNAcylation and its potential as a therapeutic target. Cardiovascular disease, and specifically heart failure (HF), is the leading cause of death in the United States[64] ^1 and worldwide.[65] ^2 HF associated with the development of cardiac hypertrophy is characterized by a maladaptation of energy substrate utilization.[66] ^3 , [67]^4 The perturbed cardiac metabolism can change ATP status as well as metabolic intermediates, which in turn can regulate cellular signaling pathways such as protein acetylation[68] ^5 and O‐linked N‐acetylglucosamine (O‐GlcNAcylation).[69] ^6 This second modification is distinct from other protein glycosylation changes that are added to proteins in the secretory pathway. O‐GlcNAcylation is regulated by 2 proteins, O‐GlcNAc transferase (OGT) and O‐GlcNAcase (OGA; also known as MGEA5).[70] ^6 Previous studies identified that OGT knockout caused embryonic lethality,[71] ^7 cardiomyocyte‐specific OGT knockout caused postnatal cardiac dysregulation,[72] ^8 and OGA knockout caused perinatal death in mice.[73] ^9 Given the essential nature of the proteins regulating this modification for cell viability, protein O‐GlcNAcylation is clearly required for normal cellular function. However, several reports have shown that the context, degree, and duration of changes in protein O‐GlcNAcylation all contribute to its regulatory role in cellular homeostasis. For example, in diabetes, which leads to decreased metabolic flexibility in the heart, it has been reported that increased cardiac protein O‐GlcNAcylation is associated with contractile and metabolic dysfunction in both rodent models and human heart tissue.[74] ^6 In both animal and human studies, diabetes is associated with higher levels of protein O‐GlcNAc modifying enzymes; OGT,[75] ^10 OGA,[76] ^11 glutamine:fructose‐6‐phosphate aminotransferase,[77] ^12 and expression of these genes. We have previously shown that in diabetes mouse models, there is an elevated level of cardiac protein O‐GlcNAcylation and impaired cardiac function with disrupted cardiac energy substrate utilization.[78] ^13 Of note, Prakoso et al[79] ^14 showed that diabetes‐induced O‐GlcNAcylation and cardiac dysfunction is associated with increased PI3K‐Akt signaling that was attenuated by reducing O‐GlcNAc levels by increasing OGA levels. Increased O‐GlcNAc has also been observed in cardiac hypertrophy and is associated with contractile dysfunction, while removing the excess O‐GlcNAc via overexpression of OGA reverses this adverse remodeling.[80] ^14 , [81]^15 In nondiabetic HF, increased O‐GlcNAcylation has been linked to similar outcomes. For example, Umapathi et al[82] ^15 reported that cardiac‐specific OGA expression also attenuated the adverse myocardial remodeling associated with pressure overload, whereas cardiac OGT overexpression and chronic increases in O‐GlcNAc alone resulted in a dilated cardiomyopathy, increased ventricular arrhythmia, and premature death. In contrast to the previous examples, a cardiac protective role in acute protein O‐GlcNAcylation has been shown in ischemia/reperfusion injury.[83] ^16 , [84]^17 Studies in cardiac stem cells reported that under ischemia/reperfusion conditions, Thiamet‐G treatment, an inhibitor of OGA, was cardioprotective, supporting a role for increased O‐GlcNAc levels.[85] ^18 Thus, while protein O‐GlcNAcylation is essential for cell viability and the maintenance of cellular signaling and is cardioprotective in response to acute stresses, under chronic stress conditions increased protein O‐GlcNAcylation is associated with pathologic cardiac adverse remodeling. The studies by Prakoso et al[86] ^14 and Umapathi et al[87] ^15 clearly showed that the overexpression of OGT and consequently increased O‐GlcNAc are sufficient to induce adverse cardiac remodeling. However, while the primary function of OGT is to catalyze O‐GlcNAcylation, it also has a protease activity and is responsible for cleavage and activation of host cell factor 1[88] ^19 as well as binding to Tet methylcytosine dioxygenase 1 to regulate chromatin structure.[89] ^20 These effects of OGT overexpression in the heart could be mediated in part by increased activities of its other functions. Therefore, in the current study, we used an inducible cardiomyocyte‐specific dnOGA overexpression (dnOGAh) mouse model that results in an increase in protein O‐GlcNAcylation levels. The dnOGA variant is naturally occurring and was identified in the spontaneous diabetic Goto–Kakizaki rat.[90] ^21 The variant is catalytically inactive and is associated with increased protein O‐GlcNAcylation.[91] ^22 We have previously shown that cardiomyocyte‐specific induction of the dnOGAh variant increases cardiac protein O‐GlcNAcylation to levels like those seen in diabetes rodent models.[92] ^13 Using this model, we examined the consequences of short‐term (2 weeks) and long‐term (24 weeks) elevation of cardiomyocyte‐specific protein O‐GlcNAcylation in adult mouse heart, within a range consistent with increases in O‐GlcNAc associated with pathologic stresses.[93] ^13 We observed that 2‐week increases in O‐GlcNAc levels resulted in a mixture of adaptive and maladaptive responses. These changes intensified over 24 weeks of elevated cardiac O‐GlcNAc levels, resulting in cardiac hypertrophy, increased extracellular matrix signaling, changes in the cardiac transcriptome with targets specific to antioxidant mechanisms, a decline in mitochondrial oxidative phosphorylation (OXPHOS) activity, increased cardiac fibrosis, and diastolic dysfunction. Thus, increased protein O‐GlcNAcylation may at first be adaptive but, when sustained, progresses to induce a moderate pathologic remodeling. Methods Data Availability Detailed experimental material and reagents are described in Data [94]S1. The data that support the findings of this study are available from the corresponding author upon reasonable request. Raw and processed files for RNA sequencing are available on the National Center for Biotechnology Information Gene Expression Omnibus database ([95]GSE211656). Animal Models and Experimental Design Mouse studies were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham (Animal Project Number: 20252). All animal procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, Eighth Edition, and we used the Animal Research: Reporting of In Vivo Experiments reporting guidelines.[96] ^23 At study termination, mice were anesthetized by 2% isoflurane inhalation in an induction chamber and euthanized by cervical dislocation followed by heart excision. The tissue collection was conducted between 6 and 8 am to standardize any circadian effects, and all samples were flash frozen in liquid nitrogen and stored at −80 °C until analysis. To generate dnOGAh mice, the previously described, tetracycline‐responsive element, enhanced green fluorescent protein, rat OGA splice variant mouse[97] ^22 was bred to the previously described mouse α‐myosin heavy chain promoter–driven codon optimized reverse tetracycline transactivator mouse.[98] ^24 We previously published the initial confirmation of inducible and cardiomyocyte‐specific expression of dnOGA for enhanced protein O‐GlcNAcylation.[99] ^13 For all studies, 8‐ to‐10‐week‐old Friend Virus B (FVB) background double transgenic mice (dnOGA and α‐myosin heavy chain promoter–driven codon optimized reverse tetracycline transactivator) were the experimental group (dnOGAh) and α‐myosin heavy chain promoter–driven codon optimized reverse tetracycline transactivator single transgenic mice were used as controls. Both male and female mice were used for a subset of studies, for example, echocardiography, heart hypertrophy, and histology. For the mRNA sequencing study and mitochondria OXPHOS complex studies, male mice were used. The mice were housed at 22 °C on 12/12‐hour light/dark cycle with ad libitum access to food and water. The transgene was induced by injection with 100 μg doxycycline (MilliporeSigma, Burlington, MA) in 0.9% NaCl solution and switched to 1 g/kg doxycycline‐supplemented NIH31 diet (Envigo, Madison, WI) for short‐term (2 weeks) and long‐term (24 weeks) groups. The control mice were also injected with doxycycline and fed doxycycline chow to control for indirect effects of doxycycline.[100] ^25 Echocardiography and Left Ventricular Catheterization Echocardiography assessment was conducted at the specified time points as previously described.[101] ^26 Briefly, mice were anesthetized by 2% isoflurane inhalation in an induction chamber and then transferred to a 37 °C platform with 1.0% to 1.5% isoflurane to maintain ≈400 beats/min, and images were acquired in the left lateral decubitus position with a 12‐mHz or MX400 (20–46 mHz, center transmit 30‐MHz) probe (VisualSonics) and analyzed using the Vevo 2100 or 3100 program, respectively. Interventricular septal thickness, left ventricular internal dimensions, and left ventricular posterior wall thicknesses at diastole and systole were measured using M‐mode images. Fractional shortening and ejection fraction were calculated as previously described.[102] ^27 Spectral Doppler was used to determine transmitral early and atrial wave peak velocities with the ratio of early‐to‐atrial calculated. Left ventricular hemodynamic measurements were conducted with a temperature calibrated 1.4‐Fr micromanometer‐tipped catheter (Millar Instruments, Houston, TX) as previously described.[103] ^28 Briefly, the catheter was inserted through right carotid artery of mice under 2% isoflurane anesthetized and analyzed using LabChart version 8.1 for MAC (ADInstruments, Sydney, Australia). Western Blotting Protein analysis was performed using Western blotting as described previously.[104] ^13 Briefly, total left ventricular tissue was homogenized in RIPA buffer (150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, 50 mmol/L Tris‐pH 7.5, 1% NP‐40, 1% sodium deoxycholate, 1 mmol/L Na[3]VO[4]) with Halt protease/phosphatase inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA), 10 μmol/L Thiamet‐G (SML0244; MilliporeSigma), and 10 μmol/L PMSF (36 978; ThermoScientific). Protein lysates (20–30 μg) were loaded and separated using 8 or 12% SDS polyacrylamide gels, separated and transferred to 0.45 μm immobilon‐FL polyvinylidene difluoride membrane (IPFL00010; MilliporeSigma). The membranes were immunoblotted with primary antibodies for OGA (14711‐1‐AP, ProteinTech), OGT (O6264; MilliporeSigma), GAPDH (Ab8245; Abcam, Cambridge, UK), RL2 (NB300‐524; Novus Biologicals, Centennial, CO), NADPH quinone oxidoreductase 1 (Ab34173; Abcam), catalase (219 010; EMD Millipore), HO‐1 (Ab13248; Abcam), citrate synthase (CS; [105]Ab129095; Abcam), succinate dehydrogenase B (Ab14714; Abcam), cytochrome c oxidase IV (Ab14744; Abcam), phospho‐pyruvate dehydrogenase E (PDHE) 1α^S293 (AP1062; MilliporeSigma), total PDHE1α ([106]Ab110334; Abcam), pyruvate dehydrogenase kinase (PDK) 1 (ADI‐KAP‐PK112; Enzo Life Science, Farmingdale, NY), PDK2 (sc‐100 534; Santa Cruz Biotechnology, Dallas, TX), PDK4 ([107]Ab214938; Abcam), calcium‐calmodulin–dependent protein kinase IIα (Upstate05‐032; MilliporeSigma), calcium‐calmodulin–dependent protein kinase IIδ (PA1‐21407; ThermoScientific), and β‐actin (GTX629630; GeneTex, Irvine, CA) in 1X casein blocking buffer (B6429; MilliporeSigma). Proteins were detected using fluorescence conjugated secondary antibodies: goat anti‐rabbit IgG (A21109; ThermoScientific) or donkey anti‐mouse IgG (936–32 212; Li‐Cor Biosciences, Lincoln, NE) and the membranes scanned using an Odyssey CLx infrared imaging system (9120; Li‐Cor Biosciences). The protein O‐GlcNAc modification is detected by primary antibody of CTD110.6 (UAB Epitope Recognition and Immunoreagent Core) and by secondary goat anti‐mouse IgM–horseradish peroxidase (401 225; Calbiochem) with SuperSignal West Pico PLUS Chemiluminescent Substrate (34 580; ThermoScientific). For the CTD110.6 antibody competition assay, samples were loaded and transferred to the membrane under the same conditions and then divided into 2 blots. One blot was incubated with CTD110.6 antibody alone and the other blot was incubated with CTD110.6 plus 100 mmol/L of GlcNAc overnight at 4 °C. The signal was captured by Amersham Imager 600 (GE Health Care, Chicago, IL). The images were analyzed using Image Studio Lite (Li‐Cor Biosciences) for most of the blots and ImageJ (National Institutes of Health) for CTD110.6 blots. We tested 2 commonly used O‐GlcNAc antibodies (ie, RL2 and CTD110.6) using our Western blot conditions because we found a broader range of clearer bands with CTD110.6 versus RL2 (Figure [108]S1A and Figure [109]1A). We also confirmed that CTD110.6 is O‐GlcNAc specific through GlcNAc competition assay (Figure [110]S1B). Figure 1. dnOGAh mice exhibited elevated cardiac protein O‐GlcNAc levels and higher levels of the proteins that regulate O‐GlcNAc modifications. Figure 1 [111]Open in a new tab A, Western blot analysis of left ventricular whole cell extract for protein O‐GlcNAc modification (CTD110.6) and protein levels of dnOGA (upper band; eGFP fusion), WT OGA (lower band), and OGT at 2‐week and 24‐week induction. B through E, Densitometric analysis of immunoblots. B, Left side: whole lane protein O‐GlcNAc quantification (25–250‐kDa protein size). B, Right side: upper size of protein O‐GlcNAc quantification (50–250‐kDa protein size). C, WT OGA protein quantification. D, WT OGA+dnOGA combined quantification. E, OGT quantification. All quantitative data are mean±SEM (n≥6); Statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (*P<0.05, **P<0.01, ***P<0.005, or ****P<0.001). Con indicates control; dnOGA, dominant‐negative O‐GlcNAcase; eGFP, enhanced green fluorescent protein; OGA, O‐GlcNAcase; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; and WT, wild‐type. Trichrome Staining Left ventricular tissue was fixed in 10% formaldehyde solution (F8775, MilliporeSigma) and paraffin embedded. Sections (4 μm thickness) were deparaffinized in xylene and dehydrated through 100%, 90%, and 80% ethanol (3 803 686; Leica, Buffalo Grove, IL). The sections were stained with Masson Trichrome (HT15; MilliporeSigma) following the manufacturer's instructions. The fibrosis region was confirmed by light microscopy (Nikon Eclipse E200 with NIS element software) and quantified with ImageJ software. We developed an unbiased and automated ImageJ macro function, which is available at [112]https://github.com/scurfman/ImageJ‐ColorThreshold. RNA Sequencing Analysis Left ventricular tissue was harvested from dnOGAh or control male mice following either 2 weeks or 24 weeks of doxycycline treatment (n≥3) for transcriptomic mRNA sequencing analysis. RNA isolation was performed using the RNeasy Fibrous Tissue Mini Kit (Qiagen Inc., Hilden, Germany) following the manufacturer's protocol. Isolated RNA was analyzed to ensure RNA quality, with RNA integrity number >7. Next‐generation RNA sequencing was then performed using Illumina HiSeq2000 for single‐read 75 bp sequencing at the Heflin Genomics Core at the University of Alabama at Birmingham. Adapters and low‐quality (Phred score<20) sequences were trimmed using Trim Galore (0.6.6), a wrapper tool around CutAdapt version 2.6. Trimmed reads were aligned to the mouse genome (GRCm39) via STAR version 2.7.3a. R software version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) was used to perform downstream analysis and data visualization. DESeq2 (1.30.1)[113] ^29 was employed within R to perform differential gene expression on the basis of a negative binomial generalized linear model. First, the count data were prefiltered to keep only rows that have at least 10 reads in total. DESeq2 then estimates the size factor or the normalizing factor for different read depths for each gene. It also performs gene‐wise dispersion estimates via maximum likelihood, followed by adjustment with empirical Bayes method. Finally, DESeq2 provides normalized count data for each contrast (changes in dnOGAh mice relative to controls at 2‐week and 24‐week doxycycline treatment) by negative binomial generalized linear model fitting. Quantification and significance of differential gene expression were determined using the Wald test that provides quantile‐normalized read counts along with log[2] fold change for each contrast, P values, and Benjamini–Hochberg adjusted P values. Heat map visualization and hierarchical clustering were performed with Ward's minimum squared variance algorithm, and dendrograms were generated by Euclidean distance via pheatmap version 1.0.12) using regularized logarithm transformed read counts. An unsupervised principal component analysis plot was generated using the scatterplot3d package. Using Horn's parallel analysis in PCAtools R packages verson 2.6.0, we indentified the first 3 principal components to be the optimum number of principal components to explain the variance in the data. EnrichR version 3.0 R interface package was used to perform and import pathway enrichment analysis data from the web‐based tool Enrichr.[114] ^30 Gene‐Concept Network plots were generated using clusterProfiler [115]^31 version 3.18.1 and ReactomePA [116]^32 version 1.34.0 R packages. To control for dnOGA induction, we assessed OGA expression levels in all samples. We found that in all but 2 dnOGAh samples, OGA transcript was >20 000 compared with control mice, which were all <5000 (Figure [117]S2A). The 2 dnOGAh mice that were nonresponders showed a distinctly different gene expression pattern (Figure [118]S2B) and were excluded from all further analyses. Raw and processed files for RNA sequencing are available on the National Center for Biotechnology Information Gene Expression Omnibus database ([119]GSE211656). All coding in this study is available at the GitHub repository: [120]https://github.com/bakshisayan/Chronic.dnOGAh.M.RNAseq_Ha.Bakshi.W ende_2021. Mitochondria OXPHOS Complex, CS, and Lactate Dehydrogenase Activity Analysis The mitochondrial electron transport activities were measured from frozen mouse heart tissue using the Seahorse XFe96 analyzer (Agilent, Santa Clara, CA) following the method previously described.[121] ^33 Briefly, frozen tissues were pulverized in liquid nitrogen and homogenized in MAS buffer (70 mmol/L sucrose, 220 mmol/L mannitol, 5 mmol/L KH[2]PO[4], 5 mmol/L MgCl[2], 1 mmol/L EGTA, 2 mmol/L HEPES at pH 7.4). Heart tissues were additionally homogenized with a glass Dounce homogenizer for 20 strokes. The homogenates were centrifuged at 1000g for 10 minutes at 4 °C, and 1 μg of heart mitochondria and homogenates were loaded into a Seahorse XFe96 microplate and then centrifuged at 2000g for 20 minutes at 4 °C without brake. Alamethicin (10 μg/mL final) and cytochrome c (10 μmol/L final) were then added to the MAS solution. The substrate injection solution was as follows: for complex I, NADH (1 mmol/L) followed by rotenone (1 μmol/L); complex II succinate (5 mmol/L)+rotenone (1 μmol/L) followed by antimycin A (10 μmol/L); complex III duroquinol (0.5 mmol/L) followed by antimycin A (10 μmol/L); and complex IV TMPD (0.5 mmol/L)+ascorbic acid (1 mmol/L) followed by azide (20 mmol/L). The XF analyzer running protocol was set as 3 minutes mix, 2 minutes wait, and 3 minutes measure cycle, 2 times each, repeated before and after port injection. CS and lactate dehydrogenase assays were measured spectrophotometrically as previously described.[122] ^34 , [123]^35 For CS activity, 20 mmol/L oxaloacetate, 10 mmol/L acetyl‐coenzyme A (CoA), and 5,5‐dithiobis (2,4‐nitrobenzoic acid) were added with the heart homogenate (10 μg) in 0.1% Triton X‐100 and 100 mmol/L Tris (pH 8.0) buffer, then measured the product of 20 mmol/L 5,5‐dithiobis (2,4‐nitrobenzoic acid)–‐CoA at 412 nm for kinetic absorbance change for nmol/min per mg protein at 37 °C. For lactate dehydrogenase activities, 0.3 mmol/L NADH and 10 mmol/L pyruvate were added in heart homogenate (10 μg) in PBS (pH 7.4) buffer, then measured NADH oxidation kinetics at 340 nm and calculated the activity as nmol/min per mg protein at 37 °C. Mitochondrial DNA Quantification DNA was isolated from the left ventricle using the DNeasy Blood and Tissue Kit (Qiagen, Germantown, MD) following the manufacturer's protocol. The mitochondrial DNA quantification was measured by real‐time quantitative polymerase chain reaction as described previously.[124] ^36 Mitochondrial DNA was detected by mt forward (5′‐CCCCAGCCATAACACAGTATCAAAC‐3′) and mt reverse (5′‐GCCCAAAGAATCAGAACAGATGC‐3′), and genomic DNA was detected by 18S forward 5′‐AAACGGCTACCACATCCAAG‐3′ and 18S reverse 5′‐CAATTACAGGGCCTCGAAAG‐3′ and amplified using SSoAdvanced Universal SYBR Green Supermix (172–5274; Bio‐Rad Laboratories, Hercules, CA) by the CFX96 Touch Real‐Time PCR Detection system (Bio‐Rad Laboratories). Polymerase chain reaction conditions were as follows: 94 °C for 2 minutes, followed by 40 cycles of denaturation at 94 °C for 15 seconds, annealing and extension at 60 °C for 1 minute. Relative mtDNA copy number was presented by the 2^−ΔΔCT method using 18S as a reference gene. Statistical Analysis Unless noted, 2‐way ANOVA followed by Tukey's multiple comparisons test were performed using Prism 9.2 for Mac OS X (GraphPad Software, San Diego, CA). In the case of smaller sample sizes (n=3) for left ventricular catheterization experiment, Asymptotic K‐Sample Fisher‐Pitman Permutation Testing was performed in R version 4.3.1, using the oneway_test function in the coin package version 1.4–2; this was followed by post hoc analysis with pairwise tests with the Benjamini–Hochberg procedure to control the false discovery rate using the pairwisePermutationTest in the rcompanion package version 2.4.30 and unpaired Student t test (2‐tailed) using Prism. All the results are presented as mean±SEM. Statistical significance was set at P<0.05. Results Cardiac‐Specific dnOGA Overexpression (dnOGAh) and Protein O‐GlcNAc Levels We have previously shown that in these mice dnOGA is expressed only in the hearts of double transgenic mice when doxycycline is present.[125] ^13 Cardiomyocyte dnOGA overexpression significantly increased cardiac total protein O‐GlcNAcylation (Figure [126]1A and [127]1B) with larger changes in proteins over 50 kDa at 1.32‐fold at 2 weeks of induction and 1.95‐fold increase at 24 weeks (Figure [128]1B). Like previous studies showing compensatory induction in OGA expression by Thiamet‐G, an OGA inhibitor,[129] ^15 , [130]^37 our model showed an induction of wild‐type OGA (1.58‐fold 2 weeks and 1.61‐fold 24 weeks; Figure [131]1A, [132]1C, and [133]1D). OGT levels were not altered at 2 weeks but were modestly increased 1.35‐fold at 24 weeks in dnOGAh versus control hearts (Figure [134]1A and [135]1E). Prolonged Protein O‐GlcNAc Levels Induce Cardiac Hypertrophy and Diastolic Dysfunction The increase in O‐GlcNAc levels via the induction of dnOGA in cardiomyocytes was associated with a progressive cardiac hypertrophy at 24 weeks, independent of sex, when compared with age‐matched controls (Figure [136]2A through [137]2F). These changes in mass were observed in both the left ventricle (Figure [138]2B and [139]2E) and right ventricle (Figure [140]2C and [141]2F). Notably, there were several significant changes in the expression of cardiac hypertrophic remodeling genes both at 2 weeks and at 24 weeks with induction of Nppa and Nppb (Figure [142]2G and [143]2H), suppression of Atp2a2 and Myh6 (Figure [144]2I and [145]2J), but no significant change in Myh7 (Figure [146]2K). Of these known HF markers, Nppa was further increased at 24 weeks in dnOGAh hearts (Figure [147]2G). Despite the cardiac hypertrophy at 24 weeks and changes in gene expression at both the 2‐week and 24‐week time points, dnOGA overexpression was not associated with a decline in systolic contractile function (Figure [148]2L and Table [149]S1). However, consistent with the cardiac hypertrophy, several indices of cardiac chamber dilation and wall thickening were significantly different at 24 weeks in both male and female dnOGAh mice compared with age‐matched controls (Figure [150]2M and Table [151]S1). Specifically, left ventricular internal dimension in both diastole and systole was larger only in male dnOGAh mice. In contrast, interventricular septum systolic and left ventricular posterior wall diastolic thicknesses were larger only in female dnOGAh mice (Table [152]S1). Even though systolic function was unaffected, long‐term induction of dnOGA showed impaired diastolic function as indicated by increased early‐to‐atrial ratio, and this was more pronounced in male mice, due to a reduced A‐wave (Figure [153]2N and Table [154]S1). Furthermore, in vivo catheterization measurement further confirmed elevated left ventricular end‐diastolic pressure (Figure [155]2O) and time constant tau (Figure [156]2P) in male mice at the 24‐week time point. Figure 2. Long‐term (24‐week) induction of cardiac protein O‐GlcNAcylation increased cardiac hypertrophy and cardiac remodeling and heart failure with preserved ejection fraction. Figure 2 [157]Open in a new tab Male mice ratios of heart weight to tibia length (TL) (A), left ventricle (LV) plus septum to TL (B), and right ventricle (RV) to TL (C), at 2‐week and 24‐week induction of dnOGAh compared with age‐matched controls (Con) (n≥10). Female mice ratios of heart weight to TL (D), LV plus septum to TL, and RV to TL (F) at 2‐week and 24‐week induction of dnOGAh compared with age‐matched controls (n≥10). Cardiac hypertrophy markers of mRNA normalized counts (G through K) (n=3 or 5) in mouse left ventricular tissue. Natriuretic peptide A (Nppa; ANP) (G), natriuretic peptide B (Nppb; BNP) (H), Atp2a2 (SERCA2A) (I), myosin heavy chain 6 (Myh6) (J), and Myh7 (K). Representative echocardiographic analysis in male mice (n≥5) and female mice (n≥8) at 24‐week induction of dnOGAh. Cardiac systolic function is presented as ejection fraction (L) from M‐mode left ventricular traces. Cardiac hypertrophy and remodeling are presented as left ventricular internal dimension at diastole (LVID;d) (M). Transmitral Doppler flow diastolic analysis of early to late (E/A) ratio (N). Additional echocardiogram parameters are listed in Table [158]S1. Left ventricular catheterization for in vivo cardiac hemodynamic function analysis in 24‐week induction of cardiac protein O‐GlcNAcylation mice (O and P). Left ventricular end‐diastolic pressure (LVEDP; O) and isovolumic relaxation constant τ (P). Statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (A through N), and unpaired Student t test (2‐tailed; O and P) (*P<0.05, **P<0.01, ***P<0.005 or ****P<0.001). All calculated and analysis data are mean±SEM. Con indicates controls; dnOGA, dominant‐negative O‐GlcNAcase; E/A, early to late ratio; EF, ejection fraction; LV, left ventricle; LVEDP; LV end‐diastolic pressure; LVID;d, left ventricular internal dimension at diastole; Myh, myosin heavy chain; Nppa, Natriuretic peptide A; Nppb, natriuretic peptide B; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; RV, right ventricle; Tau, isovolumic relaxation constant τ; and TL, tibia length. Cardiac Protein O‐GlcNAcylation and Left Ventricular Fibrosis Pathologic cardiac hypertrophy is associated with extracellular matrix accumulation and cardiac fibrosis; therefore, we evaluated morphometrics and histologic fibrosis in 2‐week and 24‐week dnOGAh groups. While there was no significant fibrosis at 2 weeks, 24‐week induction of dnOGA transgene showed Masson Trichrome–stained fibrosis in interstitial and perivascular regions in both male and female mice (Figure [159]3A through [160]3C and Figure [161]S3A through S3C). The mRNA expression levels of the matrix molecule secretion gene, Col1a2 was significantly increased only at 24 weeks in dnOGAh mice, whereas Col1a1 was unchanged (Figure [162]3D). Additionally, matrix metalloproteinase 2 (Mmp2), which degrades extracellular matrix proteins and releases cellular signaling factors that bind to the extracellular matrix, was also significantly increased at 24 weeks (Figure [163]3E). Another regulator of fibrosis, cellular communication network factor 2 (Ccn2 [164]^38 ) was also markedly increased at 24 weeks in dnOGAh mouse hearts (Figure [165]3F). Upstream regulatory mechanisms of extracellular matrix molecular pathways,[166] ^39 including transforming growth factor‐β2 and ‐β3, but not ‐β1, were also increased only at 24 weeks compared with age‐matched controls and 2‐week dnOGAh hearts (Figure [167]3G). Interestingly, a target downstream of transforming growth factor‐β signaling, and known early fibrotic signaling, periostin (Postn),[168] ^40 was significantly induced at both 2 weeks and 24 weeks in dnOGAh hearts (Figure [169]3H). To determine whether the left ventricular tissue changes were associated with a fibroblastic phenotype, we examined the expression of 3 specific fibroblast markers including cardiac muscle troponin T2, cardiac type (Tnnt2), actin alpha 2, smooth muscle and myofibroblast (Acta2), and cardiac cell development transcription factor 21 (Tcf21). In contrast to extracellular matrix and fibrosis gene induction at 24‐week dnOGAh induction, none of the fibroblast markers were changed at either time point (Figure [170]3I through [171]3K). Figure 3. Sustained protein O‐GlcNAcylation induces a progressive induction of extracellular matrix and fibrosis gene expression. Figure 3 [172]Open in a new tab Histologic examination of cardiac tissue using hematoxylin and eosin staining (A) and Masson Trichrome staining (B and C) in male mice. Data are mean±SEM (n≥4), and 6–8 left ventricular regions of each sample were taken for quantification. Extracellular matrix markers and fibrosis markers (D through H) were induced but cardiac muscle markers (I through K) were not changed in 24‐week dnOGAh hearts compared with age‐matched controls. mRNA sequencing normalized counts of collagen family (D), matrix metalloproteinase 2 (Mmp2) (E), cellular communication network factor 2 (Ccn2) (F), transforming growth factor‐β (Tgfb) family (G), periostin (Postn) (H), troponin T type 2 (Tnnt2) (I), Actin alpha (Acta; α‐smooth muscle actin [αSMA]) 2 (J), and transcription factor 21 (Tcf21) (K). Data are mean±SEM (n=3 or 5); statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (*P<0.05, ***P<0.005, or ****P<0.001). Acta indicates actin alpha; Ccn2, cellular communication network factor 2; Con, control; dnOGA, dominant‐negative O‐GlcNAcase; LV, left ventricle; Mmp2, metalloproteinase 2; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; Postn, periostin; Tcf21, and transcription factor 21; Tgfb, transforming growth factor beta family; and Tnnt2, troponin T type 2. Temporal Changes in Cardiac mRNA Expression in Response to Elevated Protein O‐GlcNAcylation To obtain further insight into the transcriptome regulatory mechanisms impacted by increased protein O‐GlcNAcylation and its duration, we conducted pathway enrichment analysis of our cardiac transcriptomic data in male mice. Heat map and hierarchical clustering of differentially expressed genes for both 2 weeks and 24 weeks of dnOGAh induction revealed distinct clustering with 2‐week and 24‐week segregating by genotype (Figure [173]4A). The separation between the groups was more evident by unsupervised principal component analysis with the 2 control groups clustering more tightly than the 2 dnOGAh groups (Figure [174]4B). Upon further analysis of the regulated transcripts, we found that a total of 2672 genes are distinctly regulated (±1.5‐fold change, Q<0.1) in 2‐week, and 1293 genes are independently regulated in 24‐week dnOGAh mouse hearts compared with age‐matched controls. One thousand seventy‐five genes are regulated in the same direction (either induced 573 or suppressed 502) at both time points of transgene induction with a modest 23 genes showing inverse regulation by the time point of dnOGA induction (Figure [175]4C and Table [176]S2). Using these differentially expressed genes, pathway analysis against the gene ontology (GO) biological process database found various enriched pathways (Table [177]S3). For example, at 2 weeks, dnOGAh hearts showed induction of genes involved in translation (GO:0006412) and mRNA splicing (GO:0000398) pathways; mitochondrial respiratory chain complex assembly (GO:0033108); cytokine‐mediated signaling (GO:0019221); and inflammatory response (GO:0006954) (Figure [178]4D). Also, at the 2‐week time point, dnOGAh hearts showed suppression of genes involved in cardiac muscle cell action potential (GO:0086001) and cardiac muscle tissue development (GO:0048738) (Figure [179]4D). Several genes were induced at both the 2‐week and 24‐week time points and enriched in pathways for filament sliding (GO:0033275 and GO:0033275) (Figure [180]4E). Genes consistently suppressed by dnOGA transgene induction at either time point enriched for pathways including the branched‐chain amino acid metabolic process (GO:0009081), acyl‐CoA metabolic process (GO:0006637), and hexose biosynthetic process (GO:0019319) (Figure [181]4E). Consistent with some of the candidate analyses of cellular remodeling (Figure [182]3), at 24 weeks dnOGAh hearts had gene induction consistent with extracellular matrix and collagen fibril organization pathways (GO:0030198; and GO: 0030199) (Figure [183]4F). Importantly, when we expanded this analysis to the Reactome analysis we found both component of extracellular matrix organization and collagen formation pathways were significantly changed (Figure [184]4G). The pathways most markedly downregulated in 24‐week dnOGAh mice were those associated with pathologic cardiac metabolic remodeling including fatty acid β‐oxidation pathways (GO:0006635; GO:0033539; and GO:0019395) (Figure [185]4F). Figure 4. Transcriptomic analysis of 2‐week vs 24‐week induction in the dnOGAh mouse heart relative to age‐matched controls. Figure 4 [186]Open in a new tab A, Heat map visualization and hierarchical clustering of all mouse genes generated from the sequencing library (aggregated into 100 kmeans clusters; Con‐2 weeks, blue; dnOGAh‐2 weeks, black; Con‐24 weeks, green; dnOGAh‐24 weeks, red). Rows correspond to individual RNA transcripts. B, Unsupervised principal component analysis using regularized logarithm‐transformed read count data of all 4 groups (2‐week Con, blue; 2‐week dnOGAh, black; 24‐week Con, green; 24‐week dnOGAh, red). The plot denotes the variance explained by first 3 principal components (PC1, PC2, PC3). C, Venn diagram comparing differential gene expression of dnOGAh mice relative to controls at 2 weeks and 24 weeks. The specific gene transcripts are listed in Table [187]S2. (D through F) Pathway enrichment analysis of differentially expressed genes (DEGs) against the Reactome database for top pathways only at 2 weeks (D), at both time points (E), only at 24 weeks (F). All enriched pathways and specific regulated genes for each pathway are listed in Table [188]S3. G, Gene‐Concept Network plot of genes listed for upregulation at 24‐week induction. BCAA indicates branched‐chain amino acid; CoA, coenzyme A; Con control; dnOGAh, dominant‐negative O‐GlcNAcase overexpression; and TOR, target of rapamycin. Chronic Protein O‐GlcNAcylation and Mitochondrial Parameters Previous studies, including our own, have shown a link to increased protein O‐GlcNAcylation and decreased mitochondrial function.[189] ^13 , [190]^26 , [191]^41 Therefore, we assessed mitochondrial electron transport activity. Mitochondrial complex I and II activities were suppressed only in the 24‐week dnOGAh group (Figure [192]5A and [193]5B). Complex III and IV were not changed at either time point of transgene induction (Figure [194]5C and [195]5D). None of the 4 complex activities measured were changed at 2 weeks (Figure [196]5A through [197]5D). CS activity, which is often used as an index of mitochondria content was also significantly reduced only in 24‐week dnOGAh hearts (Figure [198]5E). In contrast with CS and complex I and II activities, the activity of the cytosolic enzyme, lactate dehydrogenase was not affected by dnOGA transgene induction at either time point (Figure [199]5F). To determine whether the changes in activities were associated with decreased protein levels, we measured protein levels of CS and several mitochondrial OXPHOS complexes (Figure [200]5G through [201]5J). CS protein levels showed similar expression levels compared with age‐matched controls, though trending down in 24‐week dnOGAh hearts (P=0.054; Figure [202]5H). Nuclear DNA‐encoded succinate dehydrogenase B and cytochrome c oxidase IV showed decreased protein expression levels at 24 weeks (Figure [203]5I and [204]5J). Despite the defects of mitochondrial function above, mtDNA levels were not changed in dnOGAh mice (Figure [205]5K). Although mtDNA levels were not changed, mRNA expression levels of mtDNA encoded OXPHOS complex and nuclear‐DNA encoded mitochondrial OXPHOS complex genes varied significantly (Figure [206]S4). More specifically, nuclear DNA–encoded OXPHOS genes were markedly increased after 2‐week transgene induction and substantially decreased with age‐matched controls at 24‐week transgene induction (Figure [207]S4B). The transcriptional coactivator and regulator of mitochondrial oxidative capacity and volume, proliferator‐activated receptor gamma coactivator 1 alpha, was significantly decreased as early as 2‐week transgene induction, proceeding the decline in several of its known downstream targets (Figure [208]5L). However, another mitochondrial regulator, Nrf1, was not regulated in dnOGAh hearts at either time point (Figure [209]5M). Interestingly, the mitochondrial fusion gene, Mfn2, was also downregulated in dnOGAh hearts at both time points (Figure [210]5N); however, the mitochondrial fission gene, Fis1, was unchanged (Figure [211]5O). Figure 5. Prolonged protein O‐GlcNAcylation decreases mitochondrial electron transport complex activity and protein expression levels. Figure 5 [212]Open in a new tab A through D, Mitochondrial electron transport complex I–IV activities in left ventricular tissue homogenates using Seahorse XF analyzer. E, Citrate synthase (CS) activity and (F) lactate dehydrogenase activity of left ventricular tissue homogenate. Western blot analysis of mitochondrial proteins (G) and densitometric analysis (H through J). H, CS, succinate dehydrogenase B (SDHB; I), and cytochrome c oxidase IV (COXIV; J). All oxygen consumption rate (OCR) data and GAPDH normalized relative protein expression level are mean±SEM (n=4–6). K, Mitochondrial DNA (cytochrome oxidase subunit I) quantified and normalized to 18S genomic DNA quantification. Relative expression level calculated by 2^−ΔΔCT from real‐time quantitative polymerase chain reaction Ct value. Data presented as mean±SEM (n=6). L through O, mRNA‐sequencing normalized counts of mitochondrial biogenesis markers, peroxisome proliferator‐activated receptor gamma coactivator 1‐alpha (Ppargc1a; PGC‐1α; L), nuclear respiratory factor (Nrf1; M), mitofusin2 (Mfn2; N), and fission1 (Fis1; O). Data presented as mean±SEM (n=3 or 5); statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (*P<0.05, **P<0.01, or ****P<0.001). Con indicates controls; COXIV, cytochrome c oxidase IV; CS, Citrate synthase; dnOGA, dominant‐negative O‐GlcNAcase; Fis1, fission1; LDH, lactate dehydrogenase; LV, left ventricular; Mfn2, mitofusin2; mtDNA, mitochondrial DNA; Nrf1, nuclear respiratory factor; OCR, oxygen consumption rate; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; Ppargc1a, peroxisome proliferator‐activated receptor gamma coactivator 1‐alpha; and SDHB, succinate dehydrogenase B. Protein O‐GlcNAcylation Suppressed Energy Metabolism Pathways in Heart Defects in glucose metabolism are also associated with pathologic cardiac hypertrophy; therefore, we examined glucose metabolic regulation. Although expression of the primary cardiac glucose transporter (Slc2a4; glucose transporter type 4) level was maintained (Figure [213]6A), several glycolytic enzyme encoding genes (hexokinase [Hk2], glucophospho isomerase [Gpi], phosphofructokinase [Pfkm], and enolase [Eno3]) were significantly reduced in both 2‐week and 24‐week dnOGAh hearts (Figure [214]6A). UDP‐GlcNAc is synthesized via the hexosamine biosynthetic pathway and is required for protein O‐GlcNAcylation; therefore, to determine whether the dnOGAh model alters the hexosamine biosynthetic pathway, we examined several genes in this pathway. Glutamine‐fructose 6‐phosphate transaminase (Gfpt1), and UDP‐N‐acetylglucosamine pyrophosphorylase (Uap1) expression levels were not changed in dnOGAh mice (Figure [215]6B). Figure 6. dnOGAh changes glucose and fatty acid metabolism pathways. Figure 6 [216]Open in a new tab RNA normalized counts of (A) glycolysis pathway enzymes, solute carrier family 2 member 4 (Slc2a4; also known as glucose transporter type 4 [GLUT4]), hexokinase 2 (Hk2), glucose‐6‐phosphate isomerase (Gpi), phosphofructokinase, muscle (Pfkm), and enolase3 (Eno3). B, Hexosamine biosynthetic pathway (HBP) genes, glutamine‐fructose‐6‐phosphate transaminase 1 (Gfpt1; GFAT1), and UDP‐N‐acetylglucosamine pyrophosphorylase1 (Uap1). C, Pyruvate dehydrogenase kinase (PDK) family genes. D, Western blot analysis of PDK family protein levels and phospho‐pyruvate dehydrogenase (PDHE) 1α^Ser293, total PDHE1α (E) densitometry analysis. F, Fatty acid oxidation pathway genes, Cd36, carnitine palmitoyltransferase 2 (Cpt2), acyl‐CoA dehydrogenase long chain (Acadl; LCAD), malonyl‐CoA decarboxylase (Mlycd; MCAD) and patatin‐like phospholipase containing 2 (Pnpla2; ATGL). All normalized counts data (n=3 or 5) and normalized relative protein expression data (n=6 or 7) are mean±SEM; statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (*P<0.05, **P<0.01, ***P<0.005, or ****P<0.001). Acadl indicates Acyl‐CoA dehydrogenase long chain; Con, controls; Cpt2, carnitine palmitoyltransferase 2; dnOGA, dominant‐negative O‐GlcNAcase; Eno3, enolase3; Gfpt1, glutamine‐fructose‐6‐phosphate transaminase 1; Gpi, glucose‐6‐phosphate isomerase; Hk2, hexokinase 2; Mlycd, malonyl‐CoA decarboxylase; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; PDHE1α, dehydrogenase E1 subunit alpha; PDK, pyruvate dehydrogenase kinase; Pfkm, phosphorfructokinase, muscle; Pnpla2, patatin like phospholipase containing 2; Slc2a4, solute carrier family 2 member 4; and Uap1, UDP‐N‐acetylglucosamine pyrophosphorylase1. Because of the robust changes in glycolytic genes, we also looked at the regulation of the key regulator between glycolysis and mitochondrial glucose oxidation the pyruvate dehydrogenase complex. Pyruvate dehydrogenase complex is inhibited by phosphorylation, via the PDKs, decreases in the latter are often considered markers of a failing heart.[217] ^42 All 3 cardiac expressed isoforms trended down (Figure [218]6C). However, only Pdk1 and Pdk2 mRNA levels were significantly reduced in both time points by 2‐way ANOVA, with Pdk4 P=0.072 at both time points. However, we only detected a significant reduction in PDK1 protein levels (Figure [219]6D and [220]6E). Consistent with lower PDK levels, the phosphorylation of pyruvate dehydrogenase 1α was also reduced at both time points in dnOGAh hearts (Figure [221]6D and [222]6E). For further elucidation of energy metabolism downstream of pyruvate dehydrogenase complex, the tricarboxylic acid cycle genes were also analyzed. In both the 2‐week and 24‐week dnOGAh groups, there was a robust suppression of tricarboxylic acid cycle gene expression (Figure [223]S5 in the Data Supplement). Consistent with the lower CS activity at 24 weeks (Figure [224]5E), its gene expression was also lower (Figure [225]S5B). To determine if fatty acid oxidation utilization could compensate for this reduced glucose metabolism, we also examined several fatty acid uptake and oxidation genes. Surprisingly, we found suppression of fatty acid uptake gene (Cd36), mitochondrial fatty acyl‐CoA transferase (Cpt2), β‐oxidation genes (Acadl [acyl‐CoA dehydrogenase long chain] and Mlycd [malonyl‐CoA decarboxylase]), and lipolysis gene (Pnpla2) in both the 2‐week and 24‐week dnOGAh mouse hearts (Figure [226]6F). Protein O‐GlcNAcylation and Redox Mechanisms and Calcium Signaling HF‐associated decreases in mitochondrial OXPHOS activity are often linked to increases in reactive oxygen species generation. To determine if this was the case here, we examined mitochondrial antioxidant pathway genes and found glutathione peroxidase (Gpx) 1 and Gpx3 increased at both time points (Figure [227]7A and [228]7B). Conversely, superoxide dismutase (Sod)2 expression was reduced, but only in 24‐week dnOGAh hearts (Figure [229]7C). We further assessed Nrf2, the master regulator of redox pathways, also a known target of O‐GlcNAcylation[230] ^37 ; however, it was unchanged at the RNA level (Figure [231]7D). Interestingly, a downstream target of nuclear factor erythroid 2‐related factor 2 signaling, Nqo1 mRNA and protein expression was suppressed in 24‐week dnOGAh (Figure [232]7E, [233]7H, and [234]7I). In contrast, another downstream target of the nuclear factor erythroid 2‐related factor 2 signaling pathway, Hmox1 mRNA expression was induced at 2 weeks but not at 24 weeks (Figure [235]7F). Catalase, a third nuclear factor erythroid 2‐related factor 2 signaling target, showed no significant changes at either time point (Figure [236]7G, [237]7H, and [238]7I). A final component of redox signaling we found significantly regulated is NADPH oxidase (Nox) 4, which is known to produce reactive oxygen species and has been implicated in cardiac hypertrophy.[239] ^43 , [240]^44 Nox4 was robustly induced following dnOGA transgene induction but only at 24 weeks (Figure [241]7J). Figure 7. dnOGAh alters antioxidant pathway, NADPH oxidase (NOX) 4 signaling, and intracellular calcium signaling regulators. Figure 7 [242]Open in a new tab RNA analysis from mouse LV tissues. Oxidative stress response pathway genes (A through G). A, Glutathione peroxidase (Gpx1), Gpx3 (B), Sod2 (MnSOD, C), nuclear factor erythroid 2‐related factor 2 (Nfe2l2; nuclear factor erythroid 2‐related factor 2 [NRF2], D), NADPH quinone dehydrogenase 1 (Nqo1, E), heme oxygenase 1 (Hmox; HO‐1, F), and catalase (Cat, G). H and I, Western blot analysis of NRF2‐ARE pathway proteins and densitometric analysis. NADPH oxidase 4 (Nox4), and intracellular calcium signaling and its target pathway (J through P). J, Nox4, Ca^2+/calmodulin‐dependent protein kinase II d (Camk2d; K), ryanodine receptor 2 (Ryr2; L), stromal interaction molecule 1 (Stim1; M), inositol 1,4,5‐triphosphate receptor type 1 (Itpr1; IP3R1; N), voltage‐dependent anion channel 1 (Vdac1, O), and activating transcription factor 4 (Atf4; P). All RNA‐sequencing normalized counts data are mean±SEM (n=3 or 5), and densitometric analysis data are mean±SEM (n=6). Statistical significance was based on 2‐way ANOVA followed by Tukey's multiple comparisons test (*P<0.05, **P<0.01, ***P<0.005, or ****P<0.001). Atf4 indicates activating transcription factor 4; Camk2d, Ca^2+/calmodulin‐dependent protein kinase II d; Cat, catalase; Con, controls; dnOGA, dominant‐negative O‐GlcNAcase; Gpx, Glutathione peroxidase; Hmox1, heme oxygenase 1; Itpr1, inositol 1,4,5‐triphosphate receptor type 1; Nfe2l2, nuclear factor erythroid 2‐related factor 2; Nox4, NAD(P)H oxidase 4; Nqo1, NAD(P)H quinone dehydrogenase 1; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; Ryr2, ryanodine receptor 2; Sod2 superoxide dismutase 2; Stim1, stromal interaction molecule 1; and Vdac1voltage‐dependent anion channel 1. O‐GlcNAcylation has also been shown to affect cardiac function by regulation of calcium signaling and handling pathways. Specifically, calcium/calmodulin‐dependent protein kinase II (CaMKII) is known as a direct target of protein O‐GlcNAcylation.[243] ^45 , [244]^46 Although Camk2d mRNA levels were not changed (Figure [245]7K) regulation of the protein by O‐GlcNAcylation sensitizes ryanodine receptors (Ryr) to endoplasmic/sarcoplasmic reticulum Ca^2+ release and regulates NADPH oxidase.[246] ^45 Consistent with this additional mechanism of regulation, we found suppression of Ryr2 at both time points (Figure [247]7L). Another important cellular calcium signaling stromal interaction molecule (Stim1), also a known target of protein O‐GlcNAcylation,[248] ^47 had no changes in transcript levels (Figure [249]7M). However, inositol 1,4,5‐triphosphate receptor type3 (Itpr3) was downregulated in 2‐week dnOGAh, but not in 24‐week dnOGA transgene induction (Figure [250]7N). In contrast, voltage‐dependent anion channel (Vdac1) was suppressed only at 24 weeks (Figure [251]7O). We further validated calcium signaling disruption via Atf4 expression, which is regulated by calcium signaling–mediated endoplasmic reticulum stress,[252] ^48 and found it was induced at both time points (Figure [253]7P). Together, these findings suggest multiple levels of regulation of reactive oxygen species and calcium signaling via O‐GlcNAcylation that change in distinct ways depending on duration of the increase in O‐GlcNAc levels. Discussion Increases in cardiac protein O‐GlcNAcylation have been implicated in cardiac remodeling associated with diabetes and hypertrophy. We have shown that increased glucose availability suppressed ketone body metabolism in diabetic cardiomyocytes,[254] ^13 via protein O‐GlcNAcylation of 3‐oxoacid CoA‐transferase 1 and specificity protein 1.[255] ^26 We have previously shown that the cardiac‐specific inducible overexpression of dnOGA results in an increase in O‐GlcNAc levels like that seen in diabetes.[256] ^13 This dnOGA model also overcomes limitations associated with OGT overexpression models since OGT has functions as a protease and a scaffold in addition to its glycosyltransferase activity, which cannot be ruled out as potential contributing factors in the adverse effects of OGT overexpression in the heart. Glutamine:fructose‐6‐phosphate aminotransferase overexpression is also another mechanism for increasing O‐GlcNAc levels, but this is achieved by increasing UDP‐GlcNAc availability.[257] ^49 In addition to O‐GlcNAcylation, UDP‐GlcNAc is also the major substrate for all other protein O‐glycosylation, and thus changes in other glycoproteins cannot be ruled out as contributing factors. Also, in contrast to an inducible OGA‐knockout model,[258] ^50 our inducible dnOGA expression does not completely ablate OGA activity as indicated by the moderate increase in overall O‐GlcNAc levels, which is also reflected in the relatively minor changes seen in OGT protein expression in the heart following dnOGA induction. Although previous reports have shown that dnOGA increased protein O‐GlcNAcylation in intact cells, the molecular and biochemical mechanisms remain to be elucidated. One possibility is that dnOGA would outcompete endogenous wild‐type OGA for O‐GlcNAcylated proteins resulting in an apparent inhibition of endogenous OGA activity. Clearly, additional in vitro studies are needed to evaluate the interactions between OGA and dnOGA and how these alter overall hexosaminidase activity. In addition, overexpression of dnOGA increased the protein expression of wild‐type OGA and OGT (Figure [259]1). Tan et al[260] ^51 reported that cellular O‐GlcNAc level can change OGT and OGA splice variant and expression level, which remains to be determined in the current model. An important advantage of all inducible models is that the onset and duration of induction are easily controlled; therefore, the goal of this study was to examine the consequences of inducing dnOGA expression and the resulting increase in protein O‐GlcNAcylation, in adult mice in the short term (2 weeks) and long term (24 weeks) in the absence of any other metabolic or physiologic stressors. We found that 2 weeks of moderately enhanced O‐GlcNAcylation was sufficient to induce changes in the expression of genes linked to HF, metabolism, and Ca^2+ regulation. After 24 weeks of induction of the dnOGA transgene, there was evidence of cardiac hypertrophy, with preserved systolic function combined with impaired diastolic dysfunction. This was also associated with an increase in fibrosis, transcriptional changes in the extracellular matrix, calcium signaling, and antioxidant pathways, as well as changes in the transcriptional regulation and activities of mitochondrial electron transport machinery. Modulation of protein O‐GlcNAcylation in the heart has been highlighted as a potential therapeutic strategy in diabetic cardiomyopathy, cardiac hypertrophy, and ischemic HF[261] ^52 ; indeed, reductions in O‐GlcNAc levels in diabetic cardiomyopathy via OGA overexpression improved cardiac function.[262] ^14 While O‐GlcNAc levels are increased in cardiac hypertrophy controversy remains as to the direct role of O‐GlcNAc in adverse versus protective cardiac remodeling. For example, Zhu et al[263] ^53 reported that OGT deletion and the accompanying decrease in O‐GlcNAc resulted in worse outcomes in a model of pressure overload hypertrophy and concluded that the increase in O‐GlcNAc promotes compensated cardiac function. On the other hand, Gelinas et al[264] ^54 showed that reducing O‐GlcNAc levels blocked cardiomyocyte hypertrophy, and Umapathi et al[265] ^15 reported that OGA overexpression attenuated the response to pressure overload hypertrophy while preventing the increase in O‐GlcNAc levels. Umapathi et al also showed that hyper O‐GlcNAcylation modulated by OGT overexpression causes cardiac hypertrophy, loss of cardiac function, and sudden cardiac death. While our dnOGAh mouse model resulted in cardiac hypertrophy in both male and female mice by 24 weeks, in contrast to OGT overexpression models, there was no death, and the mice maintained cardiac systolic function (Figure [266]2L and [267]2M). Interestingly, 24 weeks of dnOGA transgene induction showed increased early‐to‐atrial ratio by mitral Doppler echocardiography and elevated left ventricular end‐diastolic pressure and tau by in vivo hemodynamic measurement characteristic of diastolic dysfunction, which is similar to diabetic cardiomyopathy (Figure [268]2N and [269]2O; Table [270]S1). Altogether, our model of prolonged protein O‐GlcNAcylation in cardiac tissue showed cardiac hypertrophy and diastolic dysfunction with preserved ejection fraction. Heart failure with preserved ejection fraction (HFpEF) is a clinical syndrome that is estimated to include >50% of HF cases. More importantly, the specific mechanisms and effective treatment of HFpEF have not yet been elucidated. Our model showed for the first time that sustained increases in protein O‐GlcNAcylation leads to HFpEF‐like phenotype independent of systemic hyperglycemia. Although we found a, HFpEF‐like phenotype in the 24‐week dnOGA group, the role of O‐GlcNAc in models of HFpEF, such as the high‐fat diet+L‐NAME (N[G]‐Nitro‐L‐arginine Methyl Ester) model,[271] ^55 requires further examination. Cardiac fibrosis is a common pathologic marker of HF that is characterized by fibroblast activation and excessive accumulation of extracellular matrix.[272] ^56 In the setting of cardiac hypertrophy, fibroblasts respond to changes in mechanical loading leading to activated myofibroblasts, which increases extracellular matrix remodeling. Cardiac fibrosis is classified as a replacement (of dying cardiomyocytes), interstitial, and perivascular fibrosis. Our dnOGAh model showed a fibrosis pattern that occurred in both perivascular and interstitial regions (Figure [273]3A through [274]3C). Although endothelial‐to‐mesenchymal transition, fibroblast differentiation, and infiltration of inflammatory cells are known mediators of cardiac remodeling, we are unable to elucidate the precise fibrosis development pathways in the dnOGA model. However, it is noteworthy that this cardiomyocyte‐specific increase in protein O‐GlcNAcylation was sufficient to induce cardiac fibrosis with transcriptomic molecular mechanisms of fibrotic signaling and extracellular matrix markers, in the absence of markers of fibroblast activation (Figure [275]3). Future studies are warranted to define the specific mechanisms by which an increase in cardiomyocyte O‐GlcNAcylation triggers adverse fibrotic remodeling of the heart. In addition to diastolic dysfunction and fibrosis, the current study also revealed that mitochondrial complex activity was decreased 24 weeks after dnOGA transgene induction (Figure [276]5). In addition, there was an activation of redox pathways including a robust induction of NADPH oxidase 4 expression only seen at 24 weeks (Figure [277]7J). Prior studies have shown that NADPH oxidase 4 mediates oxidative stress generating H[2]O[2].[278] ^57 NADPH oxidase 4 also regulates various cellular signaling pathways shown to be regulated in the current study including transforming growth factor‐β,[279] ^58 , [280]^59 mitochondrial OXPHOS complex I activity,[281] ^60 and nuclear factor erythroid 2‐related factor 2.[282] ^61 Surprisingly, even though we observed both reduced OXPHOS complex I and II activity as well as reduced citrate synthase activity in the dnOGAh mice, mtDNA levels were not changed suggesting that mitochondrial mass is unchanged (Figure [283]5K). Future studies are required to determine the mechanism for this disconnect. In addition to changes in redox signaling genes above, CaMKII is well known protein O‐GlcNAcylation target, mediates NOX signaling, and is a major mediator of heart disease.[284] ^45 , [285]^46 Although transcriptome CaMk2α and δ levels were not changed in dnOGAh mouse heart, CaMKII α and δ protein expression levels were induced in dnOGAh mouse heart (Figure [286]S6). Unfortunately, we were not successful in measuring CaMKIIα and δ phosphorylation. The increase in CaMKII protein in the absence of changes at the transcriptional level, could indicate an increase in protein stability. Protein O‐GlcNAcylation is known to regulate protein stability and degradation; however, the mechanism underlying the increase in CaMKII protein in the dnOGA heart remains to be elucidated. While cardiac hypertrophy and increased fibrosis were only observed in the 24‐week dnOGA group, it is noteworthy that some key transcriptional changes were evident after only 2‐week induction. For example, Nppa and Nppb were elevated at both 2 and 24 weeks of dnOGA induction, whereas Atp2a2 and Myh6 were significantly decreased at both time points (Figure [287]2). While most of the transcriptional regulators of fibrosis examined were only elevated in the 24‐week dnOGAh group, Postn, an essential component of extracellular matrix remodeling, was increased as early as 2 weeks. In addition, many of the transcriptional factors involved in regulating glucose and fatty acid metabolism were also decreased in both the 2‐week and 24‐week groups (Figure [288]6) as was Ryr2 (Figure [289]7). It has been previously reported that OGT overexpression is sufficient to induce adverse cardiac remodeling seen in both diabetes and hypertrophy; however, as noted earlier, it remained unclear whether this was a direct consequence of increased O‐GlcNAcylation or other functions of OGT. Here, we have shown that a modest increase in cardiac O‐GlcNAc levels via induction of dnOGA is sufficient to induce cardiac hypertrophy by 24 weeks that was associated with fibrosis and impaired diastolic dysfunction with preserved ejection fraction in both male and female mice. In addition, changes in gene expression were observed as early as 2 weeks after dnOGA induction including changes associated with HF and cardiac remodeling, even though there were no signs of cardiac remodeling at that time point (Figure [290]8). Nevertheless, several questions remain, such as how induction of dnOGA changes the O‐GlcNAc proteome and how those changes shift between 2‐week and 24‐week induction. Moreover, in the present study, the transcriptome analysis and mitochondrial OXPHOS function are limited to male dnOGAh mice, although female dnOGAh mice also exhibited a strong hypertrophic phenotype and different pattern of cardiac hypertrophy factors measured by echocardiogram. In addition, we do not know whether dnOGA hearts would develop overt systolic dysfunction and HF if the duration of induction were extended beyond 24 weeks. Nevertheless, we have shown that induction of dnOGA results in a modest increase in O‐GlcNAc levels (<2‐fold), like that seen in cardiac hypertrophy and diabetes, which is sufficient to induce cardiac remodeling that mimics some aspects of HFpEF. Future studies will enable us to identify the molecular mechanisms underlying this O‐GlcNAc–induced remodeling and help better understand the physiological and pathophysiological roles of cardiomyocyte protein O‐GlcNAcylation. Figure 8. Schematic of cardiomyocyte‐specific dnOGA overexpression mouse model. Figure 8 [291]Open in a new tab Graphic representation of cardiomyocyte‐specific dnOGAh overexpression on cellular protein O‐GlcNAcylation (upper left panel). The advantage of this mouse model is that it allows us to rule out the adverse effects of either OGT overexpression or OGA knockout. We found that cardiomyocyte specific protein O‐GlcNAcylation showed differential response with short‐term (2‐week) and long‐term (24‐week) induction associated with heart failure with maintained systolic function (lower panels), that progresses toward hypertrophy, fibrosis, and oxidative stress at the later time point. dnOGA indicates dominant negative OGA; OGA, O‐GlcNAcase; O‐GlcNAc, O‐linked β‐N‐acetylglucosamine; and OGT, O‐GlcNAc transferase. Thus, in conclusion, we have shown that induction of protein O‐GlcNAcylation using an inducible cardiomyocyte‐specific dnOGA mouse model resulted in cardiac hypertrophy, cardiac fibrosis, diastolic contractile dysfunction, reduced mitochondrial metabolic function, changed gene expression of several pathways including those involved in energy metabolism, and reduced oxidative stress signaling. We also highlight the importance of duration of enhanced protein O‐GlcNAcylation. Specifically, that short‐term induction was sufficient to alter HF‐linked gene expression, which, when sustained for long‐term induction, progressed to more severe cardiac hypertrophy and remodeling. While further molecular mechanisms of these changes need to be elucidated, our model supports the concept that sustained moderate increases in cardiac‐specific protein O‐GlcNAcylation is sufficient to induce molecular signaling changes and cardiac dysfunction in the absence of other changes in metabolic substrate delivery. Therefore, we propose that this increase in protein O‐GlcNAcylation could be a potential therapeutic target for multiple cardiac diseases. Sources of Funding This work was supported by National Institutes of Health (R21HL152354 to J.C. Chatham and A.R. Wende; R01HL133011 to A.R. Wende; F31HL154571 to L.A. Potter; T32GM008361 and T32HD071866 to S.F. Chang; R01HL118067 and R03AG042860 to N.‐S. Rajasekaran; P30AG050886 to J. Zhang; R56AG060959 to J.C. Chatham and J. Zhang; and S10OD032422 to UAB Heflin Center for Genomic Science Core Laboratories), and American Heart Association (Postdoctoral Fellowship 21POST834132 to C.‐M. Ha; 22POST909324 to S. Sunny; and Predoctoral Fellowship 23PRE1022560 to S. Bakshi). Disclosures None. Supporting information Data S1 Tables S1–S3 Figures S1–S6 [292]Click here for additional data file.^ (2.4MB, pdf) Acknowledgments