Written by: Zhen LiKyle B LaPennaNatalie D GehredXiaoman YuW H Wilson TangJake E DoironHuijing XiaJingshu ChenIan H DriverFrank B SachseNaoto MuraokaAntonia KatsoudaParaskevas ZampasAmelia G HaydelHeather QuiriarteTimothy D AllertonAlexia ZagourasJennifer WilcoxTatiana GromovaYueqin ZhengAndreas PapapetropoulosSanjiv J ShahTraci T GoodchildMartin B JensenThomas E SharpThomas M VondriskaDavid J Lefer
A detailed description of the materials used for the studies included in the current article can be found in the Major Resources Table in the Supplemental Material. Data and analytic methods will be made available from the first author (zli@cpu.edu.cn) or the corresponding author (david.lefer@cshs.org) on reasonable request.Plasma samples were collected from either ambulatory patients with HFpEF prospectively enrolled in the outpatient clinic (with symptoms of heart failure and echocardiogram showing left ventricular ejection fraction ≥50%) or age- and sex-matched control subjects recruited from the community without existing cardiovascular diseases (confirmed by echocardiography, pulmonary function testing, and cardiac biomarkers) under protocols approved by the institutional review board (No. 06–805 and No. 10–727, respectively) at the Cleveland Clinic. All participants were provided with written informed consent. Baseline characteristics of all participants were described previously.39 Briefly, a total of 96 patients (controls, n=48; HFpEF, n=48) with similar age, sex, and race distribution were included. Significant differences were observed in multiple clinical parameters, with patients with HFpEF displaying higher NTproBNP (N-terminal pro-B-type natriuretic peptide), body mass index, prevalence of hypertension, coronary artery disease, and atrial fibrillation, along with greater use of ACE inhibitors and β-blockers.39Both male Wistar Kyoto rats (WKY) and ZSF1 obese rats (Charles River Laboratories, Wilmington, MA) were used in the present study. ZSF1 obese rats spontaneously developed severe cardiometabolic HFpEF as previously characterized.36,37,39–44 Due to the drastic phenotypic difference between WKY and ZSF1 obese strains, randomization was not feasible. Both ZSF1 obese rats and WKY controls were evaluated at 2 time points (14 and 26 weeks of age). Group sizes of 6 to 8 animals per group were estimated through power analysis. With a significance level of 5% and a power of 80%, assuming the mean left ventricular end-diastolic pressure for WKY control rats is 7±9 mm Hg as suggested by preliminary data, this sample size was sufficient to detect a minimum 1.8- to 2-fold increase in left ventricular end-diastolic pressure, which was designated as one of the primary end points for validating the successful establishment of the HFpEF model. Rats were housed at an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)-accredited animal facility of Cedars-Sinai Medical Center or Louisiana State University Health Sciences Center (LSUHSC) in a temperature-controlled and 12-hour light/dark cycle.
Starting at the age of 9 weeks, male C57BL/6 N (Charles River Laboratories, Wilmington, MA) were randomized into 2 groups, and treated with either L-NG-nitro arginine methyl ester in drinking water (0.5 g/L) and a high-fat diet to induce HFpEF or normal drinking water and normal diet for 10 weeks.19,45 Both groups were evaluated at 2 time points (5- and 10-weeks following initiation of 2-hit). Group sizes of 9 to 11 animals per group were estimated through power analysis. With a significance level of 5% and a power of 80%, assuming the mean left ventricular end-diastolic pressure for control mice is 6±9 mm Hg as suggested by preliminary data, this sample size was sufficient to detect a minimum 1.8- to 2-fold increase in mean left ventricular end-diastolic pressure, which was designated as one of the primary end points for validating the successful establishment of the HFpEF model. Mice were housed at an AALAC-accredited animal facility of Cedars-Sinai Medical Center or LSUHSC in a temperature-controlled and 12-hour light/dark cycle.
Only male rats and mice were used in the current article, as female patients with HFpEF are more likely to display lower body mass index as compared with their male counterparts,46 which differs from the cardiometabolic phenotypes of the ZSF1 obese rat and 2-hit mouse models of HFpEF. Additionally, previous studies have highlighted the cardiovascular protective effects of estrogen in female sex47,48; thereby, studies using ovariectomized or postmenopausal aging female rodents are warranted in the future to fully investigate the dynamic controls of protein s-nitrosylation/nitrosative stress in HFpEF for both sexes.
All studies were Cedars-Sinai Medical Center or the LSUHSC Institutional Animal Care and Use Committee approved and received care according to AALAC guidelines.
Both plasma and tissue homogenate nitrite quantification were performed using a high-performance liquid chromatography system (ENO30 Analyzer; Amuza Inc, San Diego) specifically designed for nitrite determination as described previously.19,49–51 Briefly, biological fluid samples, such as plasma or tissue homogenates, were subjected to methanol precipitation to remove lipids and proteins. Then the supernatants were analyzed for nitrite content. Varying concentrations of pure sodium nitrite solutions between 0.039 to 5 μmol/L were used to generate the standard curve for the quantification of the nitrite concentration in the samples.
The levels of total RxNO (nitrosothiols in the plasma and RxNO in the tissue homogenates) were quantified using an ion chromatography device coupled with gas-phase chemiluminescence (nCLD88; Echo Physics, Duernten, Switzerland) as previously described.12 Briefly, biological fluid samples, such as plasma or tissue homogenates, were incubated with 2.0% mercuric chloride, followed by acidified sulfanilamide, then the reaction mixture was subjected to a reductive denitrozation reaction by iodine–iodide. The denitration reaction detaches the nitroso group from the RxNO, which then is released as gas-phase NO with a one-to-one ratio. Finally, the amount of gas-phase NO was analyzed through the nCLD88 system.
Treadmill exercise capacity of ZSF1 obese or WKY rats, as well as control or HFpEF mice, were measured using an IITC Life Science 800 Series rodent-specific treadmill (IITC Life Science, Woodland Hills, CA). Rats were first allowed to acclimate to the treadmill environment for a period of 5 minutes. Following acclimation, rats were carried through a warm-up phase of initially 6 meters per minute with gradual increase to 12 meters per minute during a 4-minute ramp-up time for a total of 5 minutes. The animals were then running at a rate of 12 meters per minute with no ramp incline until a state of exhaustion. Both duration and calculated distance were plotted along with work (kg×m).
At each time point, animals were anesthetized using isoflurane at 3% to 4% for induction and 2 to 2.5% for maintenance during the procedure. The right common carotid was cannulated with a high-fidelity pressure catheter (SP200 Pressure Control Unit, Transonic, NY) which measured the systemic pressures at systole and diastole. The pressure catheter was then introduced into the left ventricle of the heart. Left ventricular end diastolic pressures and ventricular relaxation time constant were measured as previously described.52,53
The levels of 3-nitrotyrosine (3-NT) were measured with commercially available ELISA assay kits in plasma samples from control or patients with HFpEF (No. STA-305; Cell Biolabs, San Diego, CA), and from control or HFpEF rodent models (No. NBP2–66363; Novus Biologicals, Toronto, Canada). Circulating cyclic guanosine monophosphate (cGMP) levels in plasma samples from control or ZSF1 obese rats were measured with a commercially available ELISA kit (No. 581021, Cayman Chemical, Ann Arbor, MI).
Left ventricle of the heart tissue was collected from WKY and ZSF1 obese rats at 14 or 26 weeks of age and immediately snap-frozen in liquid nitrogen. Heart nuclei were isolated using a lysis buffer consisting of 0.25 mol/L sucrose, 10 mmol/L Tris-HCl pH 7.5, 25 mmol/L KCl, 5 mmol/L MgCl2, 45 μmol/L actinomycin D, supplemented with 1X protease inhibitor (G6521, Promega), 0.4 U/μL RNasin ribonuclease inhibitor (N2515, Promega), 0.2 U/μL SuperaseIn (AM2694, Thermo Fisher). Briefly, heart tissue samples were minced into smaller pieces with scissors in a 1 mL lysis buffer. The minced tissue was then homogenized in a Dounce homogenizer on ice with 10 strokes of pestle A, followed by 10 strokes of pestle B. The homogenates were then filtered through a 40 μm cell strainer and centrifuged at 400g for 5 minutes at 4 °C. The nuclei pellet was resuspended in 2% BSA in PBS supplemented with protector RNase inhibitor (03335402001; Sigma-Aldrich, Burlington, MA) at 0.2 U/μL. Heart nuclei were stained with Sytox red (S34859; Thermo Fisher, Waltham, MA) and sorted and purified through fluorescence-activated cell sorting. Eight thousand to ten thousand nuclei from each rat heart were processed using a 10× Genomics microfluidics chip to generate barcoded Gel Bead-In Emulsions according to manufacturers’ protocols. Indexed single-cell libraries were then created according to 10× Genomics specifications (Chromium Next GEM Single Cell 5′ v2.1-Dual Index Libraries). Samples were multiplexed and sequenced in pairs on an Illumina Novaseq X (Illumina, San Diego, CA). The sequenced data were processed into expression matrices with the cell ranger single-cell software 9.0.0 obtained at the following website (https://www.10xgenomics.com/support/software/cell-ranger/latest/release-notes/cr-release-notes). FASTQ files were then obtained from the base-call files from Novaseq X sequencer and subsequently aligned to the rat genome NCBI (National Center for Biotechnology Information) Rnor6.0, with a read length of 26 bp for cell barcode and unique molecule identifier (read 1), 8 bp i7 index read (sample barcode), and 98 bp for actual RNA read (read 2). Each rat sample yielded ≈300 mol/L reads.
For targeted analysis, Anndata54 and Scanpy55 were used to load and preprocess h5ad files for each sample for import into Seurat. Scrublet56 was used to identify and remove putative doublets before AnnData objects were concatenated by week. Barcodes, features, matrix files, and metadata were extracted into a folder for import into R with the Seurat Read10× function.57 The 14-week and 26-week Seurat objects were created with CreateSeuratObject. Mitochondrial gene percentage was calculated for each cell in each Seurat object before removing cells containing >5% mitochondrial reads. Mitochondrial genes and genes with <10 reads across all cells in each object were also removed. Lastly, any cells with <500 reads were removed. The 14- and 26-week Seurat objects were merged by time point, and layers joined. Subsets of cardiomyocyte compartments were selected based on celltypeleiden0.6 column. Samples were split by week and log-normalized with NormalizeData. FindVariableFeatures was used to identify the top 2000 variable features per week, and the features were centered and scaled with ScaleData. Principal Component Analysis (RunPCA function) was run on the split objects with default parameters before layers were integrated back together (IntegrateLayers function) with method=RPCA (reciprocal PCA) Integration. A shared nearest-neighbor graph was created with FindNeighbors (using 15 RPCA dimensions) and clustered with FindClusters (with a resolution of 0.5). Clusters were assigned cell types based on known marker genes. To create the heatmaps, the integrated Seurat object was subset into separate endothelial, myocyte, fibroblast, and pericyte objects. Seurat FindMarkers function with the Wilcoxon rank-sum test option was used to identify differentially expressed genes in ZSF1 obese cells in each cell type, with a Bonferroni-corrected P value cutoff of 0.05. Results were intersected with a list of genes involved in NO signaling and regulation of protein s-nitrosylation. Then the average Log2fold-change was plotted with ggplot2 geom_tile function.58
s-nitrosylated proteins were labeled and isolated using a Pierce S-nitrosylation Labeling Kit (Thermo Scientific 90105) following the instructions from the manufacturer with modifications. Briefly, 100 μg of proteins isolated from either the 26-week-old WKY or ZSF1 obese rat heart tissue was resuspended and lysed in 4 volumes of HENS buffer. For positive and negative controls, samples were optionally incubated with 200 μmol/L s-nitrosoglutathione or reduced glutathione, respectively, for 30 minutes at room temperature, and excess reagents were removed using Zeba Spin Desalting Columns equilibrated with HENS buffer. Protein concentration was determined using the BCA assay, and equal amounts of protein (100 μg per condition, 1–2 mg/mL) were adjusted to 100 μL in HENS buffer. To block free thiols, 2 μL of 1 mol/L MMTS (20 mmol/L final) was added, followed by vigorous vortexing and incubation for 30 minutes at room temperature. Proteins were precipitated by adding 6 volumes of prechilled (−20 °C) acetone and incubating at −20 °C for ≥1 hour. After centrifugation (10 000g, 10 minutes, 4 °C), supernatants were discarded and pellets air-dried for 10 minutes. Then, pellets were resuspended in 100 μL HENS buffer and labeled with 1 μL iodoTMT (iodoacetyl tandem mass tag) reagent and 2 μL of 1 mol/L sodium ascorbate. Then the labeled s-nitrosylated proteins were isolated using immunoprecipitation technique with anti-TMT antibody (Thermo Scientific 90075) and subjected to downstream proteomic analysis as previously described.59 Additionally, normalization and differential analysis were conducted on bioladder.cn platform, whereas pathway enrichment analysis were performed on metascape. org gene annotation and analysis platform.
mRNA expression profiles were assayed from heart, liver, and kidney tissue. Following RNA extraction using TRIzol reagent (Invitrogen, 15596026), incubation and chloroform (FisherScientific, AA32614k2)–mediated phase separation, cDNA was synthesized using iScript cDNA Synthesis Kit (Bio-Rad, 1708890). Subsequently, quantitative real time PCR was performed to probe for the mRNA expression using iTaq Universal SYBR Green Supermix (Bio-Rad, 1725122) through CFX Duet Real-Time PCR system (Bio-Rad, 12016265) following manufacturer’s instructions (30 seconds of polymerase activation and DNA denaturation at 95 °C, followed by 40 cycles of 5 second denaturation at 95 °C and 30 second annealing/extension at 60 °C).
For eNOS (endothelial nitric oxide synthase; No. 4453320; Thermo Fisher, Waltham, MA), iNOS (No. 4331382; Thermo Fisher, Waltham, MA), nNOS (No. 4351372; Thermo Fisher, Waltham, MA), and housekeeping 18S ribosomal RNA (No. 4333760; Thermo Fisher, Waltham, MA), commercially available probes were used. Primer sequence for other tested genes (HBb [hemoglobin subunit-β], Trx2, and GSNOR) genes are listed below. 2^delta-deltaCT values corrected to 18S rRNA were used for data analysis.
Tissue were lyophilized with mortar and pestle and then homogenized in lysis buffer 150 mmol/L NaCl (Calbiochem, 7760), 1% NP-40 (Sigma-Aldrich, 74385), 0.5% Na-deoxycholate (AppliChem, A1531,0025), 0.1% SDS (PanReac AppliChem, A2572), 50 mmol/L Tris-HCl, pH=7,4 (Sigma-Aldrich, T1503), 2 mmol/L EDTA (Merck, 4005) supplemented with a cocktail of protease (PI, Roche, 5892970001), and phosphatase inhibitors (PhoI, Roche, 4906837001). Lysates were centrifuged (13.000 rpm, 15 minutes, 4 °C), and the protein concentration in the supernatants was quantified using the DC protein assay (BIO-RAD, 5000116). Concentration was normalized before Western blot analysis. Samples were separated on 10% or 12% SDS–PAGE and transferred to a nitrocellulose membrane (Macherey-Nagel; Düren, Germany), after Laemmli buffer containing 4% SDS, 10% β-mercaptoethanol (Sigma-Aldrich, M6250), 20% glycerol (Melford, GI345), 0.004% blue bromophenol (AppliChem, A2331,0025), and 0.125M Tris-HCl was added. The membranes were blocked (5% milk [PanReac AppliChem, A0830]) and probed with the following antibodies: anti-β-tubulin (Abcam, ab15568), anti-GAPDH (Proteintech, 10494–1-AP or Cell Signaling 2118S), anti-eNOS (Cell Signaling, 32027s), anti-peNOSs1177 (Cell Signaling, 9571), anti-pVASP (Cell Signaling, 3114S), anti-HBb (Abcam, ab214049), anti-Trx2 (Abcam, ab185544), anti-GSNOR (Proteintech, 11051–1-AP), and anti-iNOS (Proteintech, 18985–1-AP). Immunoblots were next processed with anti-rabbit secondary antibody (Merck, AP132P; MedChemExpress, HY-P8001; Jackson ImmunoResearch, 111–005-003) and visualized using the HRP substrate. In a subset of Western blot experiments, membranes were incubated with No-Stain Protein Labeling Reagent (Invitrogen, Waltham, MA) to visualize total protein as a loading control. Quantification and normalization of Western blots were performed using ImageJ software (NIH Image, National Institutes of Health). In the subset that used total protein as a loading control, target proteins were normalized to the signal from the entire lane; however, only a portion of the membrane around the same molecular weight as the target proteins was shown due to limited space allowed.
Oxidoreductase activity of Trx in tissue were measured using a commercially available fluorometric activity assay kit (No. 500228; Cayman Chemical, Ann Arbor, MI) following the instructions by the manufacturer in combination with a Synergy LX multimode microplate reader (Agilent, Santa Clara, CA).
GSNOR activity in tissue was measured as previously described.34,60 Briefly, tissue samples were homogenized in a buffer of 50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1 mmol/L EDTA, 0.1% Triton, and 1:100 protease inhibitor cocktail (Sigma Aldrich, P8340). The samples were then centrifuged at 10 000 RCF for 10 minutes at 4 °C and brought to a concentration of 0.1 mg/mL in a reaction buffer of 20 mmol/L Tris-HCl, 0.5 mmol/L EDTA. Samples were then added 75 μmol/L NADH with or without 100 μmol/L of GSNO. Samples were measured using fluorescence at 350 nm excitation/460 nm emission. Every minute for 1 hour, where the rate is determined by NADH consumption with and without the GSNO substrate.
Thoracic aortas of the rats were removed for vascular reactivity testing at sacrifice as previously described.51,52 The rings were first equilibrated in Krebs-Henseleit solution and provided a tension of 0.5 g for 60 minutes to reach equilibration. They were then pretreated with phenylephrine for maximal constriction and then followed with challenges of titrated acetylcholine and subsequently sodium nitroprusside, and measured relaxation as compared with phenylephrine maximal contraction.
HL-1 cardiac muscle cell line was purchased (Sigma Aldrich, SCC065) and maintain as previously described.61 All cells were cultured in DMEM (Sigma Aldrich, D5030) supplemented with 10% fetal bovine serum (Sigma Aldrich, F0392), 100 U/mL penicillin (Sigma Aldrich, P3032), and 100 U/mL streptomycin (Sigma Aldrich, S9137), 100 μmol/L norepinephrine (Sigma Aldrich, A7257), and 2 mmol/L glutamine (Sigma Aldrich, G8540). Cells were then plated and divided into the following groups: control, HBb overexpression, HBb overexpression+DETA NONOate (5 μmol/L; Cayman Chemical); control, Trx2 overexpression, Trx2 overexpression+DETA NONOate; control, HBb overexpression, and HBb overexpression+N6022 (50 μmol/L; MedchemExpress, NJ). pCMV6 vector carrying HBb (Origene, MR227259) or Trx2 (Origene, MR201355) overexpression were transfected into HL-1 cells with Lipofectamine 3000 (Sigma) according to the manufacturer’s instructions and incubated 48 hours for optimal protein overexpression. Following an additional 24 hours of treatment, cellular RxNO content were analyzed using an ion chromatography device coupled with gas-phase chemiluminescence (nCLD88, Eco Physics, Duernten, Switzerland) as previously described.12
Experimental data in this study were analyzed by personnel blinded to the group assignments. Except for data generated through snRNA sequencing and proteomic analysis, statistical analysis was conducted using the Prism 10 software (GraphPad, San Diego, California) and presented as mean±SEM. Normal distribution was tested by the D’Agostino and Pearson test. Differences in data among 2 groups were compared using an unpaired Student t test or a Mann-Whitney U test if the data did not follow a normal distribution or violated the assumption of equal variance. For multiple group comparison over time, 2-way ANOVA was used. For a data set involving multiple factors, such as timepoints or concentrations, 2-way ANOVA analysis followed by a Bonferroni multiple comparison test was used. A P value of <0.05 was considered statistically significant. The presented data may have different numbers of animals per group as only a subset of the rats or mice from each group were used for certain experiments. Additionally, no animal/data point was excluded unless a technical failure occurred in procedures such as invasive hemodynamics measurement, sample collection (ie, plasma volume), and treadmill running (lack of voluntary participation) or was deemed an outlier through outlier tests performed using the ROUT (Robust Outlier detection Test) method developed by Prism 10.
RESULTS
To evaluate NO bioavailability as well as protein s-nitrosylation, and nitrosative stress, we measured plasma nitrite, nitrosothiols, and 3-NT in 48 patients with HFpEF and 48 control subjects. Nitrite is an established biomarker of biologically active NO, whereas nitrosothiols and 3-NT are biomarkers for protein s-nitrosylation and nitrosative stress, respectively.62–64 We did not observe any significant change in circulating nitrite levels in patients with HFpEF (Figure 1A). Despite a lack of increased NO bioavailability, circulating nitrosothiol levels were significantly elevated (P=0.031) in patients with HFpEF, indicating increased protein nitrosylation (Figure 1B). Importantly, 3-NT, a well-recognized biomarker for nitrosative stress, was increased by >2-fold in HFpEF patients as compared with control subjects (Figure S1). These data provide strong evidence for pathological protein s-nitrosylation and nitrosative stress in patients with HFpEF despite relatively normal levels of circulating NO.
Male WKY control and ZSF1 obese rats were monitored for a period of 12 weeks starting at the 14 weeks of age to evaluate the progression of HFpEF in ZSF1 obese rats. At 14 and 26 weeks of age, body weight, echocardiographic parameters, and treadmill exercise performance were measured (Figure S2). A subset of rats was assessed for LV and systemic hemodynamics and aortic vascular reactivity, followed by sacrifice for other molecular determinations. An illustrative protocol is depicted in Figure 1C.
Throughout the 12-week study, ZSF1 obese rats displayed markedly higher body weight as compared with WKY control rats (Figure S2A), accompanied by significantly impaired glucose handling ability and reduced insulin sensitivity (data not shown), as depicted previously.40,42 Invasive hemodynamic measurements revealed that ZSF1 obese rats exhibited severe hypertension with significantly elevated systolic and diastolic blood pressures throughout the study (Figure S2B; P=0.0006 at 14 weeks and P=0.02 at 26 weeks of age for systolic BP; P=0.0006 at 14 weeks and P=0.02 at 26 weeks of age for diastolic BP). Importantly, we observed significant elevations in LV end-diastolic pressure (Figure S2C; P=0.001 at 14 weeks and P=0.001 at 26 weeks of age) and Tau (Figure S2D; P=0.001 at 14 weeks and P=0.002 at 26 weeks of age) in ZSF1 obese rats, indicating severe cardiac diastolic dysfunction, one of the hallmarks of HFpEF. Severe exercise intolerance was also observed in ZSF1 obese rats as measured in treadmill running distance during exercise (Figure S2E; P=0.0006 at 14 weeks and P=0.0012 at 26 weeks of age).
Endothelial function and eNOS–derived NO signaling have long been considered crucial for cardiovascular hemostasis, and reduced NO signaling has been linked to diastolic dysfunction.65 To assess the extent of dysregulation in NO signaling, we first measured the eNOS-dependent vascular function in isolated, thoracic aortic ring segments. Throughout the 12-week study, thoracic aortic rings from ZSF1 obese rats displayed significantly blunted responses to the endothelial-dependent vasodilator, acetylcholine (P<0.05 or 0.01 at multiple concentrations) as compared with the WKY control group (Figure 1D and 1E). In contrast, vascular reactivity to the direct nitro-vasodilator, sodium nitroprusside, remained unchanged. Endothelial dysfunction and attenuated NO bioavailability were further corroborated by markedly reduced nitrite levels in the circulation and myocardial tissue (Figure 1F and 1G; P=0.007 at 14 weeks and P=0.001 at 26 weeks of age for plasma nitrite; P=0.007 at 14 weeks and P=0.06 at 26 weeks of age for cardiac nitrite). Further, we observed no difference between ZSF1 rats versus WKY control rats in total cardiac eNOS expression but a significant reduction (P=0.004) in phosphorylation on the serine 1177 residues of cardiac eNOS, an eNOS activation site, at 26 weeks of age, confirming the disruption in eNOS activity (Figure 1H). Downstream effectors of eNOS activation, circulating cGMP and PKG-mediated pVASP (phosphorylation of vasodilator-stimulated phosphoprotein) were also significantly reduced (P=0.037 for cGMP; P=0.0006 for pVASP) in ZSF1 obese rats (Figure 1I and 1J). Taken together, these results indicate profound endothelial dysfunction and eNOS dysregulation accompanied by significant reductions in NO bioavailability. Additionally, we measured the mRNA levels of eNOS, iNOS, and nNOS isoforms. There was no difference observed in eNOS and nNOS mRNA levels throughout the study (Figure 1K and 1L). Interestingly, iNOS mRNA levels were reduced in ZSF1 rats throughout the study, although its protein expression was reduced at 14 weeks of age, but remained unchanged at 26 weeks of age (Figure 1M; Figure S3A and S3B).
To evaluate the extent of nitrosative stress in ZSF1 obese rats, we measured nitrosothiols in the plasma as well as RxNO in the heart. Additionally, we also measured circulating 3-NT given its value as a well-recognized nitrosative stress biomarker. Starting at 14 weeks of age ZSF1 obese HFpEF rats exhibited age-dependent increases in both the nitrosothiols and 3-NT levels compared with WKY control rats (Figures 2A; Figure S2F). In contrast, there was no difference in cardiac RxNO levels at the early stage of HFpEF development (ie, 14 weeks of age); however, at 26 weeks of age, ZSF1 obese rats displayed significantly elevated levels of RxNO in the heart, indicating a significant increase in myocardial protein s-nitrosylation (Figure 2B; P=0.002). Given the systemic nature of HFpEF pathology, we next assessed total RxNO levels in the liver, kidney, and lung—organs frequently implicated in HFpEF-associated comorbidities and complications. Similar to the heart, hepatic, renal, and lung RxNO levels were elevated during the HFpEF manifestation in ZSF1 obese rat model (Figure 2C through 2E).
To elucidate the molecular landscape of protein s-nitrosylation in cardiometabolic HFpEF, we performed proteomic profiling of s-nitrosylated proteins in cardiac tissue from 26-week-old WKY rats and HFpEF ZSF1 obese rats. A total of 261 s-nitrosylated proteins were identified in WKY hearts, although a markedly higher number, 473 s-nitrosylated proteins, were detected in ZSF1 obese hearts (Figure 2F). Notably, 271 proteins were uniquely s-nitrosylated in ZSF1 obese hearts, compared with only 9 unique proteins in WKY hearts (Figure 2G). Despite 252 proteins being commonly s-nitrosylated across both groups, principal coordinate analysis demonstrated a clear segregation between the s-nitrosylated proteomes of ZSF1 obese and WKY hearts, indicating a marked global shift in nitrosylation patterns in the HFpEF context (Figure 2H). Further differential analysis revealed that 336 proteins exhibited significantly increased (P<0.05) s-nitrosylation in ZSF1 obese hearts, 100 remained unchanged (P≥0.05), and 37 showed significantly reduced (P<0.05) s-nitrosylation compared with WKY controls (Figure 2I). Pathway enrichment analysis of the differentially s-nitrosylated proteins highlighted the top 20 most enriched pathways, prominently featuring key metabolic processes such as carboxylic acid metabolism, carbon metabolism, branched-chain amino acid catabolism, sulfur metabolism, the tricarboxylic acid cycle, and pyruvate metabolism (Figure 2J). In addition to metabolic pathways, enrichment was also observed in pathways related to cellular stress responses, mitochondrial protein quality control, and AMPK (AMP-activated protein kinase) signaling, underscoring a multifaceted disruption in metabolic and stress-adaptive networks in HFpEF hearts. A list of all s-nitrosylated proteins identified is included in Table S1.
Next, we performed snRNA Seq to acquire molecular insights into the development of elevated cardiac nitrosative stress despite the reduction in NO bioavailability and signaling. Various cell types were evaluated using snRNA Seq, including cardiomyocyte, fibroblast, endothelial, and pericytes identified as major cardiac resident cell types (Figure 3A). A substantial amount of differentially expressed genes were also identified either over time (Figure 3B) or between different genotypes (Figure 3C). Targeted analysis revealed changes in the expression of a variety of genes related to NO generation and signaling in 4 cardiac resident cell types. No change was observed in eNOS or nNOS expression, and iNOS was significantly downregulated (P<0.05) in myocardial endothelial cells in early HFpEF (ie, 14 weeks of age; Figure 3D and 3E). Notably, expression of HBb, a well-characterized transnitrosylase,66,67 was markedly elevated in all 4 resident cardiac cell types at both the early and late stages of HFpEF (Figure 3D and 3E). In contrast, expression of common denitrosylases, such as thioredoxins, GSNOR, AKR1a1, remained relatively stable as the only change observed was a mildly increased Trx2 expression in cardiomyocytes in 14-week-old ZSF1 obese rats (Figure 3D and 3E). Importantly, we confirmed that our observations made in ZSF1 obese rat model of cardiometabolic HFpEF were true in human HFpEF patients.68 Targeted analysis of the transcriptome of cardiac biopsies from patients with HFpEF68 demonstrated a marked upregulation of HBb when compared with either control or HFrEF (Figure 3F). Taken together, our data reveal a crucial imbalance in transnitrosylase and denitrosylase expression, particularly in HBb expression, suggesting that this imbalance may be the primary culprit responsible for elevated cardiac protein s-nitrosylation and subsequent nitrosative stress.
To further examine this hypothesis, we performed real time PCR, Western blot, and activity assay experiments to measure HBb, Trx2, and GSNOR in cardiac tissue homogenates. Associated with substantial formation of cardiac interstitial and perivascular fibrosis (Figure 4A and 4B), and consistent with the snRNA Seq analysis, striking increases in cardiac HBb mRNA and protein levels were observed in ZSF1 obese rats at both 14 and 26 weeks of age (Figure 4C and 4D). mRNA and protein levels of Trx2 remained unaltered in ZSF1 obese hearts at 14 weeks of age but significantly increased (P=0.05 and 0.0286, respectively) at 26 weeks of age; however, Trx enzyme activity was downregulated in ZSF1 obese hearts at both 14 and 26 weeks of age (Figure 4E and 4F). Similarly, despite no change in mRNA and protein levels, cardiac GSNOR activity was significantly reduced (P=0.0411) in the ZSF1 obese heart at 26 weeks of age as compared with WKY control (Figure 4G and 4H). Interestingly, HBb, Trx2, and GSNOR were all among the s-nitrosylated proteins that were upregulated in ZSF1 obese hearts as revealed by proteomic analysis (Table S1).
To further investigate the effects of modulating HBb, Trx2, and GSNOR expression/activity on cellular nitrosative status, we performed complementary experiments using HL-1 cardiomyocytes. Overexpression of HBb alone led to a significant increase (P=0.03) in intracellular RxNO levels, which was further amplified by cotreatment with DETA NONOate, a potent NO donor (Figure 4I). In contrast, Trx2 overexpression alone did not significantly alter RxNO levels; however, in the presence of DETA NONOate, a modest but significant increase (P=0.003) in RxNO was observed, though substantially lower than that induced by HBb overexpression (Figure 4J). Furthermore, pharmacological inhibition of GSNOR using N6022 in HBb-overexpressing cells resulted in an additional elevation of RxNO levels (Figure 4K), supporting the role of GSNOR in regulating denitrosylation under conditions of heightened nitrosative stress.
Given that HFpEF comorbidities (ie, hypertension, obesity, and diabetes) impact multiple organ systems and the pathophysiology of HFpEF involves organs beyond the heart and circulation, we evaluated the involvement of transnitrosylase and denitrosylase (ie, HBb, Trx2, and GSNOR) in the liver and kidney in ZSF1 obese rats. Associated with extensive hepatic steatosis found in ZSF1 obese liver (Figure 5A and 5B), we observed markedly elevated hepatic HBb mRNA and protein levels throughout the 12-week study (Figure 5C and 5D). Hepatic Trx2 mRNA level was downregulated at 14 weeks of age, but protein expression and enzymatic activity was unaffected. Although at 26 weeks of age, hepatic Trx2 protein increased but the enzymatic activity was reduced in ZSF1 obese rats (Figure 5E and 5F). Conversely, GSNOR mRNA, protein, and enzymatic activity were all reduced in ZSF1 obese hepatic tissue at both 14 and 26 weeks of age (Figure 5G and 5H).
In association with the age-dependent formation of tubulointerstitial and perivascular fibrosis in ZSF1 obese kidneys (Figure 6A and 6B), a profound increase in HBb mRNA and protein levels was observed in ZSF1 obese kidneys at both 14 and 26 weeks of age, consistent with those demonstrated in the cardiac and hepatic tissue (Figure 6C and 6D). Renal Trx2 mRNA was reduced in ZSF1 obese rats at 26 but not 14 weeks of age; however, renal Trx2 protein was increased significantly (P=0.0286) at 26- but not 14-week-old ZSF1 obese rats. In contrast, renal Trx2 enzymatic activity was reduced in both 14- and 26-week-old ZSF1 obese HFpEF rats (Figure 6E and 6F). Conversely, renal GSNOR mRNA and protein levels were reduced at both 14 and 26 (significantly, P=0.0286) weeks of age, whereas GSNOR activity was reduced only in 26-week-old ZSF1 obese rats (Figure 6G and 6H).
To further examine the dysregulated dynamic of NO and RxNO, and the role of transnitrosylase and denitrosylase in the setting of HFpEF, we performed additional experiments in another well-established rodent model of cardiometabolic HFpEF, the 2-hit murine model induced by L-NG-nitro arginine methyl ester and high-fat diet treatment (Figure 7A).19,44,45,69 Mice treated with L-NG-nitro arginine methyl ester and high fat diet displayed significantly lowered plasma (P=0.03) and cardiac tissue (P=0.0006) levels of nitrite, indicating a pathological reduction in NO bioavailability (Figure 7B and 7C), similar to observations in the ZSF1 obese rat. Interestingly, plasma 3-NT and nitrosothiols, were significantly elevated (P=0.006 or <0.0001, respectively) in the 2-hit mouse model of HFpEF (Figure 7D and 7E). Following 10 weeks of 2-hit, RxNO levels were also markedly elevated in the heart and various peripheral organs, including the liver, kidney, and lung (Figure 7F), highlighting the paradoxical systemic imbalance of NO bioavailability and excessive nitrosative stress. Correspondingly, HBb protein expression was upregulated in the heart, liver, and kidney of 2-hit HFpEF mice (Figure 7G through 7J). Conversely, GSNOR protein expression was not altered in the heart and kidney, although significantly increased (P=0.0159) in the liver of the 2-hit HFpEF mice (Figure 7K through 7N). Despite unchanged protein levels, the enzymatic activity of GSNOR was reduced in the heart, liver, kidney, and lung of the 2-hit HFpEF mice when compared with the control (Figure 7O).
DISCUSSION
HFpEF is a heterogenous, multiorgan disease associated with high morbidity and mortality.1,4,70 The prevalence of HFpEF has increased significantly in recent years and has now eclipsed HFrEF in terms of new HF diagnoses. There are multiple classes of effective therapeutic treatments accessible for patients with HFrEF, yet these treatments are largely ineffective for treating HFpEF.71 Thus, despite similar symptoms, there are likely very distinct underlying cellular and molecular pathological mechanisms that distinguish HFrEF from HFpEF.71 Elucidation of these divergent mechanisms is necessary to identify effective therapeutic targets to aid in the development of effective therapies for the treatment of HFpEF.
Given the well-established role of NO signaling in the regulation of cardiovascular health and the numerous reports demonstrating reduced NO bioavailability in CV diseases, NO-based therapies have been tested in HFpEF.15–18 However, this therapeutic approach has failed to result in meaningful clinical efficacy in patients with HFpEF. Clinical studies using nitrite, nitrate, or organic nitrates to replenish NO bioavailability, or sGC stimulators and phosphodiesterase inhibitors to amplify NO-mediated cellular signaling have produced neutral results, despite the plethora of evidence suggesting that depleted NO bioavailability as a hallmark feature of HFpEF.10 However, it is possible that increased levels of superoxide radical anions that occur in HFpEF conjugates with the supplemental NO to form peroxynitrite, significantly suppressing the therapeutic action of NO-based therapy. Indeed, we have previously reported that NO-based therapy (ie, sodium nitrite), when combined with a powerful antioxidant and peroxynitrite scavenger, hydralazine, was highly effective in reducing the severity of HFpEF in the 2-hit mouse model.45 Such an approach warrants further exploration in both basic and clinical research settings.
Additionally, recent preclinical studies have suggested that excessive nitrosative stress and accumulation of RxNO, such as s-nitrosylated proteins, play a causative role in the pathogenesis and progression of HFpEF.19,20 Schiattarella et al19 described an increase in iNOS levels coupled with observation of increased cardiac protein s-nitrosylation (in particular cardiac IRE1α [serine/threonine-protein kinase/endoribonuclease inositol-requiring enzyme 1α]), which in turn, participated in the pathogenesis of HFpEF in the 2-hit mouse model of HFpEF. Moreover, using a different animal model, the salty drinking water/unilateral nephrectomy/aldosterone rat model of HFpEF, Yoon et al20 demonstrated that upregulation of nNOS, instead of iNOS, was responsible for increased s-nitrosylation of histone deacetylase 2, which in turn, induced cardiac diastolic dysfunction. These landmark studies established the possible link between nitrosative stress as well as pathological accumulation of RxNO in HFpEF, providing an alternative explanation for the neutral results produced by NO-based therapies for the treatment of HFpEF. In the present study, we sought to confirm the previous findings of elevated NO production by NOS and excessive nitrosative stress in the setting of HFpEF. Our initial studies in patients with HFpEF revealed for the first time that plasma nitrosothiols and 3-NT are significantly elevated in patients with HFpEF as compared with controls, which is indicative of excessive nitrosative stress in agreement with previous reports19,20; however, we did not observe significant changes in circulating nitrite, a well-established biomarker and sink of systemic NO bioavailability. These findings support the potential utility of circulating nitrosothiols and 3-NT as diagnostic biomarkers for HFpEF, pending validation in large and diverse patient cohorts. In contrast, circulating nitrite appears to have limited biomarker value due to its susceptibility to dietary and lifestyle-related fluctuations which can be challenging to control in human studies. Subsequently, we evaluated nitrite and RxNO bioavailability in 2 well-established rodent models of cardiometabolic HFpEF, the ZSF1 obese rat model and the 2-hit mouse model.19,38,41 In concert with our observations in clinical samples, ZSF1 obese rats displayed significantly increased levels of s-nitrosylated proteins in both the circulation and other organs during the progression of HFpEF. Increased RxNO levels observed in the heart as well as peripheral organs, such as the liver, kidney, and the lung (mouse), suggest that nitrosative stress is a systemic pathological component of HFpEF. However, in both ZSF1 obese rat and 2-hit mouse model, we observed profound reductions in nitrite levels in the circulation, cardiac and other tissues (data not shown). Together with the observations of impaired endothelial-dependent vasodilation in response to acetylcholine treatment, reduced phosphorylated eNOS (serine 1177), decreased circulating cGMP, and lower cardiac pVASP, these findings conclusively demonstrate that NO bioavailability and signaling are severely impaired in these models of HFpEF. Taken together, these data identified a paradoxical dysregulation of NO–RxNO dynamic characterized by impaired NO production and signaling coupled with excessive circulating and tissue protein s-nitrosylation. Notably, such phenomenon is contradictory to what was observed in the context of other cardiovascular diseases, including myocardial infarction and HFrEF, in which NO availability and signaling is similarly reduced while overall protein s-nitrosylation is largely insufficient.12,34,72–78 Our findings suggest that increased production of NO may not be the cause of nitrosative stress in HFpEF, but an alternative cellular process (ie, trans nitrosylation and denitrosylation processes) are responsible for the excess accumulation of RxNO/nitrosative stress.
Indeed, the levels of these RxNO are also delicately regulated by the transnitrosylases, enzymes that transfer NO onto the cysteine residues of proteins, and by the denitrosylases, enzymes that remove the NO moiety from the cysteine residues.24,25,31,79,80 Similar to other posttranslational protein modifications, such as phosphorylation and sulfhydration, hyper- or hypo- s-nitrosylation of specific proteins resulted in altered function, which could lead to a wide range of effects that impact disease progression.24,30,31 To better understand the spectrum of such elevations in RxNO levels in HFpEF hearts, we performed iodoTMT-based purification and subsequent proteomic analysis in 26-week-old WKY and ZSF1 obese hearts to identify cardiac s-nitrosylated proteins. Our data provide compelling evidence for a significant alteration in the s-nitrosylation landscape in the context of HFpEF. The marked increase in the number of s-nitrosylated proteins in ZSF1 obese hearts compared with WKY controls highlights the profound dysregulation of protein nitrosylation in HFpEF. The identification of 271 uniquely s-nitrosylated proteins in the ZSF1 obese hearts suggests that these proteins may play critical roles in the pathophysiology of HFpEF. The global shift in the s-nitrosylated proteome, as revealed by principal coordinate analysis, further underscores the complexity of the nitrosylation modifications in the disease context. Notably, the enriched metabolic pathways in differentially s-nitrosylated proteins, including carboxylic acid metabolism, branched-chain amino acid catabolism, and the tricarboxylic acid cycle, align well with the HFpEF phenotype, where metabolic derangements are heightened. Branched-chain amino acid catabolic pathways and sulfur metabolism have been previously reported to be linked to HFpEF severity.39,81 Our data revealed that s-nitrosylation of key proteins in these pathways may produce functional consequences that influence HFpEF disease progression and further emphasize their importance in disease progression. Additionally, pathways related to cellular stress responses and mitochondrial protein quality control suggest that nitrosative stress exacerbates mitochondrial dysfunction and metabolic abnormalities in HFpEF. These findings highlight the critical role of s-nitrosylation as a key posttranslational modification contributing to the pathogenesis of HFpEF. Future experiments are needed to verify the functional consequences of s-nitrosylation on these proteins and further elucidate their roles in HFpEF progression.
Given the elevation of Rx(s)NO and 3-NT levels in multiple organs of animal models and in patients with HFpEF, our findings indicate that excessive protein s-nitrosylation likely contributes to HFpEF disease progression. However, the roles of transnitrosylase and denitrosylase, and their intricate interplay with bioactive NO, have not been investigated in the context of HFpEF. Using snRNA seq in an unbiased manner, we unexpectedly discovered that hemoglobin (subunit beta), a protein that is normally expressed mainly in erythrocytes, was shown to be markedly upregulated in all 4 major resident cardiac cell types: cardiomyocytes, endothelial cells, fibroblasts, and pericytes. Corroborating the snRNA seq analysis, we also observed striking upregulation of HBb in mRNA and protein levels in cardiac, hepatic, and renal tissues. To determine whether HBb was upregulated in human HFpEF hearts, we examined the data from a landmark study by Hahn et al investigating the myocardial gene expression signature in human HFpEF and found that cardiac biopsies from patients with HFpEF displayed an over 256-fold increase in HBb when compared with control subjects, and an over 64-fold increase when compared with patients with HFrEF.69 Notably, iNOS expression was also significantly upregulated in patients with HFpEF; however, the magnitude of this increase was substantially lower than that observed in HBb.19,69 It is well-established that hemoglobin is not normally expressed in appreciable levels in the myocardium and other peripheral organs.66,72,82–85 We are aware of only a few studies describing elevated hemoglobin expression in nonerythrocytes, such as cardiac and hepatic tissues, triggered by a pro-oxidative and inflammatory cellular environment, features that are commonly observed in organs affected by HFpEF.86–88 Importantly, cell culture-based experiments confirmed that HBb upregulation alone is sufficient to increase overall cellular RxNO content, which can be further magnified with the presence of exogenous NO (ie, NO donor). A previous study reported that HBb exerts SNO synthase in addition to transnitrosylase activity, which may account for the abrupt increase of de novo s-nitrosylation.89 Taken together, our findings provide novel insights into the potential deleterious roles of hemoglobin in cardiometabolic HFpEF. Specifically, we proposed that elevated HBb across multiple organs substantially enhanced overall trans-nitrosylation processes in the heart, liver, and kidney, thereby contributing to the excessive RxNO observed in these key organs and driving the pathogenesis and progression of cardiometabolic HFpEF. Interestingly, although previous studies have attributed transnitrosylase activity to tetrameric hemoglobin,66,72,90 our findings demonstrate for the first time that HBb monomer alone is sufficient to mediate protein trans-nitrosylation.
Other key factors responsible for the dynamic control of protein s-nitrosylation include Trx and GSNOR, which are among the most ubiquitously expressed and well-characterized denitrosylases and are responsible for denitrosylating a wide range of protein targets.23,28,79,91,92 Trx, primarily recognized for antioxidant properties, is a bifunctional nitrosylase as it has recently been shown to either de-nitrosylate or trans-nitrosylate a wide range of protein targets in mammalian cells in a context-dependent manner.26,91,92 snRNA seq analysis revealed that both Trx isoforms (Trx 1 in the cytosol and Trx 2 in the mitochondria) remained relatively stable, except a mild increase in Trx2 in ZSF1 obese cardiomyocytes at 14 weeks of age (snRNA seq analysis showed Trx2 to be the predominant isoform in the heart). As a result, we further examined Trx2 mRNA, protein expression, and enzymatic activity levels in ZSF1 obese rats. Significantly reduced Trx enzymatic activity as an oxidoreductase was observed in the cardiac, hepatic, and renal tissues, despite unaltered or increased protein expression. As a previous study reported, when oxidized, Trx loses its oxidoreductase activity but exhibits an increased propensity to function as a transnitrosylase.26 We hypothesize that during the pathogenesis of cardiometabolic HFpEF, substantial oxidative stress suppresses the oxidoreductase activity of Trx, as supported by our data, potentially leading to enhanced transnitrosylase activity. Additional cell-culture-based experiment showed that Trx2 overexpression in HL-1 cells alone does not alter overall cellular RxNO levels, whereas together with treatment of NO donor, Trx2 overexpression led to a modest but significant increase of cellular RxNO, but not to the extent of HBb overexpression. Taken together, these data suggest that despite altered expression and activity of Trx2, HBb may be the more predominant transnitrosylase that drives the pathological accumulation of RxNO in the pathogenesis and progression of cardiometabolic HFpEF.
Conversely, GSNOR is a well-established denitrosylase, and previous studies of GSNOR had focused on the therapeutic benefits of GSNOR inhibition. It has been shown that reduced GSNOR activity exerts beneficial effects in animal models of hypertension, atherosclerosis, and myocardial ischemia/reperfusion injury by promoting angiogenesis, inhibiting apoptosis, and reducing inflammation and oxidative stress.24,28,75–78 However, recently emerged evidence suggests otherwise: deficiency in GSNOR resulted in excessive nitrosative stress, leading to cellular senescence, improper mitochondrial turnover, hepatic insulin resistance in diabetes and obesity, aberrant placental s-nitrosylation and preeclampsia, as well as cardiac hypertrophy and dysfunction.32,93–95 In the present study, we also observed reduced GSNOR protein expression and enzymatic activity in an age-dependent manner, which were associated with increased RxNO/nitrosative stress levels in multiple organs, including the heart, liver, kidney, and lung. Cell culture-based experiment revealed that in the context of HBb overexpression, GSNOR inhibition by N6022 resulted in further elevation of cellular RxNO content, informing possible therapeutic benefits of GSNOR activation instead of inhibition in the context of cardiometabolic HFpEF.
Taken together, we revealed for the first time that circulating markers of protein s-nitrosylation and nitrosative stress were increased in patients with HFpEF, and that the equilibrium between endogenous NO production and signaling (ie, NOS expression, activity, and downstream effectors) and enzymatic modulation of s-nitrosylation by transnitrosylase and denitrosylases pathways (eg, HBb expression, Trx2, and GSNOR expression and activity) is disrupted in the pathogenesis of cardiometabolic HFpEF. We also demonstrate for the first time that such deranged protein s-nitrosylation and nitrosative stress are present systemically (heart, liver, kidney, and lung) in cardiometabolic HFpEF in 2 well-characterized rodent models. Our findings are the first evidence that dysregulated transnitrosylase or denitrosylases, instead of overproduction of NO by NOS, contribute to the pathological systemic accumulation of RxNO during the pathogenesis and progression of HFpEF, potentially leading to comorbidities in peripheral organs, including the liver and kidney. Our findings also support the emerging paradigm that HFpEF is a systemic syndrome rather than a disease condition that is restricted to the heart. Restoration of the balance between transnitrosylase and denitrosylase expression and activity represents a new therapeutic approach to combat the heightened nitrosative stress seen in cardiometabolic HFpEF patients and may prove beneficial in conjunction with therapies that enhance physiological NO production and signaling.
The current study used the ZSF1 obese rat and the 2-hit mouse models of HFpEF, both of which are well-established models that replicate key features of cardiometabolic HFpEF, including obesity, hypertension, and diastolic dysfunction. However, these models represent only a subset of the heterogeneous HFpEF patient population. As such, our findings may not fully capture the spectrum of molecular alterations, particularly in the trans- and de-nitrosylation machinery (eg, HBb, Trx2, GSNOR), that may be present in other HFpEF phenogroups, such as inflammation-driven or aging-related HFpEF. Development of additional animal models that more fully reflect the diverse causes of HFpEF will be essential for advancing mechanistic understanding and therapeutic development.
Moreover, although our proteomic analysis identified numerous proteins that are hyper-nitrosylated in the HFpEF heart, the functional consequences of these modifications remain to be determined. Additionally, future studies using targeted loss-of-function approaches such as tissue- or cell-type–specific knockout of HBb in the heart, liver, or kidney are warranted to directly test the mechanistic role of HBb-mediated trans-nitrosylation in HFpEF pathogenesis and multiorgan involvement.
Lastly, our study included only male rats and mice, in part due to reports that female patients with HFpEF more frequently present with lower body mass index, which differs from the cardiometabolic phenotypes modeled here. Importantly, previous studies have also demonstrated sex-specific differences in GSNOR expression and activity.34,35 Taken together, these key differences suggest that our findings in the current study may not directly translate to females. Therefore, future research incorporating female HFpEF models, including post-menopausal or ovariectomized animals, will be critical to determine whether the dysregulation of trans-/de-nitrosylation dynamics exhibits sex-specific patterns or responses.