Antibodies and reagents

Primary antibodies against AP-2α (#3208, dilution: 1:1000), vWF (#65707, dilution: 1:1000), and GAPDH (#5173, dilution: 1:1000) were obtained from Cell Signaling Technology (Boston, MA, USA). Primary GTPCH1 antibody (#ab307507, dilution: 1:1000) was from Abcam. Primary phosphorylated AP-2α at serine 219 antibody was generated by Genscript Company (dilution: 1:1000) as we described previously²⁹. HRP-conjugated Affinipure Goat Anti-Rabbit IgG (H + L) (SA00001-1, dilution: 1:5000) were from Proteintech. Lovastatin (#438186), pravastatin (#1554206), atorvastatin (#524403), angintensin II (#05-23-0101), dihydroethidium (#309800), diaminofluorescein (#D224), and oxidized low-density lipoprotein (#AB3230) were purchased from Sigma-Aldrich Company (USA). High fat diet was purchased from Research Diet (D12492). Commercial kits for determinations of glucose, cholesterol, and triglyceride, LDL-C, and HDL-C were purchased from Jian-Cheng Bioengineering Institute (Nanjing, China). All drug concentrations are expressed as working concentrations in the buffer.

Generation of AP-2α
^flox/flox mice

Targeting vectors to generate AP-2α^flox/flox mice were constructed using a bacterial artificial chromosome (BAC, B6Ng01-349O17) recombineering (see Supplementary Fig. 8A for details). Bruce4 ES cells⁴¹ were electroporated with the targeting vector and positive clones identified by PCR and southern blotting were injected into blastocysts from C57BL/6-Tyrc-2J mice. The resulting male chimaeric mice were bred to female C57BL/6-Tyrc-2J mice to obtain germline transmission. The FRT-Neo-FRT cassette was removed by crossing with Flp deleter mice, then the mice were backcrossed at least three additional times onto a C57BL/6 background.

BAC modification

A homologous recombination-proficient E. coli strain (DY380) was used for the BAC recombineering⁴². The 1^st step of the process is the homologous recombination of the BAC using the modification cassette I. It occurs by crossing over between the homology arms and the genome in the BAC. In this case, recombination results in the incorporation of the modification cassette sequences into the genome to yield the modified BAC1 (mBAC1). The 2^nd step of the process is the homologus recombination of the mBAC1 using the modification cassette II. It occurs by a second homologous recombination event that occurs within the mBAC1. Recombination yields the precisely modified BAC2 (mBAC2), with the modification cassette II inserted at the correct position in the BAC. This modified BAC carries the loxP-3’ region for introduction to the HAC vector. In brief, overnight cultures containing the BAC were grown from single colonies, diluted 10-fold in LB medium, and grown to an optical density at 600 nm of 0.4–0.6 at 32 °C. Fifty milliliter cultures were then induced for the expression of recombineering factors by shifting the cells to 42 °C for 15 min followed by chilling on ice for 10 min. Cells were then centrifuged for 5 min at 5500 X g at 4 °C and washed with 10 mL of ice-cold 1 mM HEPES 2 times. Cells were then resuspended in 100 μL of ice-cold 1 mM HEPES and electroporated. Cell transformation was performed by electroporation of 1 μg linear DNA into 100 μL of ice-cold competent cells in cuvettes (0.1 cm) using a Bio-Rad gene pulser set at 1.75 kV, 25 μF with a pulse controller set at 200 ohms. One milliliter of SOC medium was added after electroporation. Cells were incubated at 32 °C for 1 h with shaking and spread on appropriate selective agar media.

Animals and induction of a ‘two-hit’ model of HFpEF

Male WT mice were obtained from Beijing Huafukang Company (Beijing, China). The CDH5-Cre^ERT2 transgenic mouse exhibits tissue-specific expression of an inducible Cre-ERT2 fusion protein, enabling tamoxifen-induced Cre recombinase activity in vascular endothelial cells⁴³, and were obtained from Taconic Biosciences, Inc (Model #13073). AP-2α^flox/flox/CDH5-Cre^ERT2 mouse was generated by crossing AP-2α^flox/flox mice with CDH5-Cre^ERT2 mice. Male CDH5-Cre^ERT2/AP-2α^flox/flox mouse was injected with tamoxifen (2 mg/mouse) for five consecutive days to induce endothelium-specific AP-2α gene knockout. Male CDH5-CRE^ERT2 mouse injected with tamoxifen serves as a control because of endothelial CreERT2 toxicity^44,45.

To induce HFpEF as the comorbidities of hypertension and obesity²⁴, mice received HFD administration plus AngII infusion at a rate of 0.2 mg/kg per day for 12 weeks using a miniosmotic pump (Alzet), as reported previously with minor modifications^46,47,48. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal protocol was reviewed and approved by the Animal Care and Use Committee, Qilu Hospital of Shandong University.

Conventional echocardiography (ECG) and doppler imaging

Transthoracic ECG was performed using a VisualSonics Vevo 2100 system equipped with an MS400 transducer (Visual Sonics). LVEF and other indices of systolic function were obtained from short-axis M-mode scans at the midventricular level, as indicated by the presence of papillary muscles, in conscious, gently restrained mice. Anesthesia was induced by 5% isoflurane and confirmed by a lack of response to firm pressure on one of the hindpaws. During echocardiogram acquisition under body-temperature-controlled conditions, isofluorane was reduced to 1.0–1.5% and adjusted to maintain a heart rate in the range of 400–500P beats per min.

To assess diastolic function, E, and E′were measured. After measurement of the parasternal short-axis view, the sonographic probe was tilted 45 degrees to visualize the parasternal 4-chamber view. Mitral E waves were recorded with pulse-wave mode at the mitral valve opening. Mitral annulus movement, also known as E′ wave, was assessed from the medial mitral valve annulus with tissue velocity image mode.

Parameters collected include: A, peak Doppler blood inflow velocity across mitral valve during late diastole; mitral DT, mitral early filling deceleration time; E, peak Doppler blood inflow velocity across mitral valve during early diastole; E′, peak tissue Doppler of myocardial relaxation velocity at mitral valve annulus during early diastole; HR, heart rate; IVRT, isovolumic relaxation time; IVS,d, end-diastolic interventricular septal wall thickness; LVEF, left ventricular ejection fraction; LVID,d, left ventricular internal diastolic diameter; LVID,s, left ventricular internal systolic diameter; LVPW,d, left ventricular end-diastolic posterior wall; LVFS, left ventricular fractional shortening.

Blood pressure measurement

Blood pressure was determined by a left carotid catheter method at the end of animal experiments⁴⁹. Mice were anesthetized with a ketamine and xylazine mixture and placed under warm light (37 °C). A catheter was inserted into the left common carotid artery, with the aid of a dissecting microscope, to measure arterial blood pressure. For catheter insertion, the left common carotid artery was carefully exposed via a 0.5–1.0 cm midline incision in the ventral neck region. The tip of the artery toward the head was ligated with a suture, and the tip toward the heart was occluded with a microclip. A small cut was then made in the vessel wall using microscissors. A 60 -cm catheter (PE10) containing a sterile 10% heparin-90% saline solution was inserted into the artery a distance of 0.65 cm toward the thorax. The arterial clip was removed, and the catheter was tied in place. Blood was directed to a pressure transducer through the catheter to obtain computerized blood pressure measurements (AD instruments). The mice were allowed to recover and the systolic and diastolic blood pressures and heart rate were monitored for at least 30 min in conscious states.

Exercise exhaustion test

After three days of acclimatization to treadmill exercise, an exhaustion test was performed in the experimental groups of mice. Mice ran uphill (20°) on the treadmill (Columbus Instruments) starting at a warm-up speed of 5 m/min for 4 minutes after which speed was increased to 14 m/min for 2 minutes. Every subsequent 2 min, the speed was increased by 2 m/min until the mouse was exhausted. Exhaustion was defined as the inability of the mouse to return to running within 10 s of direct contact with an electric-stimulus grid. Running time was measured and running distance was calculated.

Generations of AAV9 and infection to mice

The adeno-associated virus 9 (AAV9) construction compassing cDNA (AAV9-TIE-cDNA) was generated by according to the manufacturers’ recommendations from Shanghai Genechem Co., Ltd. (Shanghai, China). The endothelial cell specific promoter is “pAAV-TIEp-EGFP-MCS-3Flag-SV40 PolyA”. Viruses were packaged and amplified in HEK293A cells and purified using CsCl₂ banding followed by dialysis against 10 mM Tris-buffered saline with 10% glycerol. Titering was performed on HEK293 cells using the adeno-X Rapid Titer kit (BD Biosciences Clontech, PaloAlto, CA, USA) according to the manufacturer’s instructions.

Animal experimental protocols

In the first part of the animal study (Supplementary Fig. 5A), male WT mice received lovastatin administration (4 mg/kg/day) one week prior to HFD plus AngII treatments for another 12 weeks. In the second part of the animal study (Supplementary Fig. 8C), a male AP-2α^flox/flox/CDH5-Cre-ERT2 mouse was injected with tamoxifen (2 mg/mouse) or vehicle for five consecutive days to induce endothelium-specific AP-2α gene knockout. Then mice received lovastatin administration (4 mg/kg/day) one week prior to HFD plus AngII treatments for another 12 weeks. In the third part of the animal study (Supplementary Fig. 10A), male WT mice were infected with AAV9 expressing negative control (NC) shRNA or circRNA-RBCK1 shRNA via tail vein injection. One week later, mice were treated with lovastatin (4 mg/kg/day) one week prior to HFD plus AngII treatments for another 12 weeks. An injection of AAV9 was repeated once in 4 weeks. For in vivo infection, AAV9 were injected via tail vein in 100 µl of PBS containing 1 ×10¹¹ IFUs of loaded virus per mouse. The concentration of DNA was 10 mg/kg. Before mice sacrificed, conventional ECG and Doppler imaging were assessed.

Hematoxylin-eosin staining

Slides with section were placed in a metal staining rack and immersed in the filtered Harris Hematoxylin for 10 s. Then, the sections were incubated in EOSIN stain for 30 s. Dehydration was performed in ascending alcohol solutions (50%, 70%, 80%, 95% X 2, 100% X 2) followed by clearance with xylene (3-4 X) in Columbia staining jars. The slides were mount using Permount (xylene based).

Masson’s trichrome staining

The heart was fixed with 4% paraformaldehyde overnight at room temperature, dehydrated sequentially through ethanol, butyl alcohol, and embedded in paraffin. Hearts were serially cut from the apex to the base perpendicular to the long axis. Transverse sections (4 μm) were stained with Masson’s trichrome. Images were acquired on Panoramic MIDI (3D HISTECH Inc., Hungary).

Quantitative PCR

Total RNA was isolated using a TRIzol-based (Invitrogen) RNA isolation protocol. RNA was quantified by Nanodrop (Agilent Technologies), and RNA and miRNA quality were verified using an Agilent 2100 Bioanalyzer (Agilent Technologies). Samples required 260/280 ratios of more than 1.8, and sample RNA integrity numbers of more than 9 for inclusion. RNA was reverse transcribed using the TaqMan microRNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s instructions. Quantification of circRNA and mRNA was performed using an ABI PRISM7500 system, and miRNA concentrations were determined using an ABI PRISM7900 system (Applied Biosystems, Carlsbad, CA, USA). Before calculation using the ^ΔΔCt method, the levels of GAPDH were used to normalize the relative expression levels of circRNA and mRNA, and the levels of small nuclear U6 were used to normalize the miRNA expression levels. Primers used for quantitative PCR were listed in Supplementary Table 7.

Mutagenesis of miR-133a binding sites in circRNA-RBCK1

These reporter constructs were generated in two steps. First, a coding-region fragment containing the miR-133a binding sites was generated by PCR and cloned into the pMIR luciferase vector (Ambion) using SpeI and MluI cloning sites. Next, site-directed mutagenesis was performed, introducing three mutations into the binding site’s seed sequence in the 3′-UTR of circRNA-RBCK1. Subsequently, a DNA fragment containing the 3′-UTR with the mutant of miR-133 binding site (MT1, MT2, or MT3) was generated by PCR and cloned into pMIR vector, this time using MluI and HindIII sites. Again, site-directed mutagenesis was used to change all three binding sites within the seed sequence (MT1/2/3). All constructs were sequenced to confirm their identity.

Plasmid transfection into HEK293 and reporter assays

The sequences were synthesized and constructed into the pmirGLO luciferase plasmid vector, and promoter sequences were integrated and inserted into the GV272 vector by Genechem company (Shanghai, China). The plasmid constructs were co-transfected in HEK293 cells with the pCMV β-gal plasmid and 50 nM each of chemically synthesized miRNA oligonucleotides (Applied Biosystems) by using lipofectamine 2000 (Invitrogen). Cells were harvested 48 h after transfection, and luciferase and β-galactosidase activities were measured. The luciferase assay was conducted using a dual luciferase reporter assay (Vazyme, China) according to the manufacturer’s instructions.

Cell cultures and virus infection

Primary HUVECs, human EECs, human CAECs, human MCECs, and human AECs were obtained from Clonetics Inc. (Walkersville, MD, USA). Cells were grown in endothelial basal medium supplemented with 2% fetal bovine serum and penicillin (100 u/ml), and streptomycin (100 µg/ml). Cultured cells were used between passages 3 and 8. All cells were incubated in a humidified atmosphere of 5%CO₂ + 95% air at 37 °C. When 70–80% confluent, the cells were treated with different agents. For HEK293 cells, cells were cultured in M200 medium supplemented with 2% fetal bovine serum and penicillin (100 u/ml), and streptomycin (100 µg/ml). For lentivirus infection, cells were infected with lentivirus expressing AP-2α shRNA or circRNA-RBCK1 shRNA from Shanghai Genechem Co., Ltd. (Shanghai, China) overnight in antibiotics-free medium supplemented with 2% FBS. The target sequence of CircRNA-RBCK1 shRNA is TCTTGCAGCAGTGGGTGATTG. The target sequence of AP-2α shRNA is TCCCAGATCAAACTGTAATTA. These targets were designed by VectorBuilder Inc. The cells were then washed and incubated in fresh medium for an additional 12 h before experiments.

Isolations of myocardial capillary endothelial cells, endocardium endothelial cells, coronary arterial endothelial cells, aortic endothelial cells from mice

For isolations of mouse aortic endothelial cells and mouse endocardium endothelial cells, heart and aorta were rapidly excised from mice (Supplementary Fig. 7A). Then, left ventricle and the aorta were injected with PBS containing 0.1% collagenase for 15 min at 37 °C. After centrifugation at 250 g for 10 min, the cell pellets were collected for RNA extraction immediately.

For isolation of mouse myocardial capillary endothelial cells, the distal left anterior descending coronary artery was cannulated, and the epicardial and endocardial surfaces were removed (Supplementary Fig. 7B). The remaining myocardial tissue was digested in PBS containing 0.1% collagenase at 37 °C for 30 min, with gentle rotation. Digested tissue was pelleted at 250 g and resuspended in DMEM containing 20% BSA (Sigma) (w/v), then myelin fraction was separated by centrifugation at 1000 g for 10 min. The cell pellet was resuspended and filtered through a 70 µM nylon mesh and collected following centrifugation at 250 g. The cell pellets were collected for RNA extraction immediately.

For isolation of mouse coronary arterial endothelial cells, the remained heart tissues by cutting the distal left anterior descending coronary artery were injected with PBS containing 0.1% collagenase for 15 min at 37 °C through the aortic root (Supplementary Fig. 7B). After centrifugation at 250 g for 10 min, the cell pellets were collected for RNA extraction immediately.

RNA extraction and sequencing data analysis

Total RNA from 6 samples was extracted. We have generally utilized 100 ng of RNA for library construction for MeRIP-circRNA sequencing. Briefly, the mRNA with polyA in the total RNA was enriched by Oligo-dT magnetic beads. The intact mRNA was then fragmented using an ultrasound machine. The segmented RNA was divided into two parts. One part was added to an m6A-capturing antibody to enrich the mRNA fragments containing m6A methylation (MeRIP-seq), and the other part was used as an Input to directly construct a conventional transcriptome sequencing library (circRNA-seq). The conventional sequencing library was constructed according to the transcriptome library construction process. Illumina Hiseq X Ten was used for high-throughput sequencing of the library. The circBase database and Circ2Traits were used to annotate the identified circRNA. Then, DESeq2 software (v1.14.1) was used for data standardization and differentially expressed circRNA screening (log2FC ≥ 1.5, p-value ≤ 0.05).

Biotinylated RNA pull-down assay

The biotinylated-circRNA-RBCK1 probe was incubated with C-1 magnetic beads (Life Technologies, Carlsbad, CA, USA) to generate probe-coated beads, then incubated with sonicated HEK293 cells at 4 °C overnight, followed by eluted and quantitative RT-PCR. For miR-133a pulled down circRNA-RBCK1, HEK293 cells with circRNA-RBCK1 overexpression were transfected with biotinylated miR-33a mimics or mutant using Lipofectamine 2000. The cells were harvested, lysed, sonicated, and incubated with C-1 magnetic beads (Life Technologies, Carlsbad, CA, USA), followed by quantitative RT-PCR.

Chromatin-immunoprecipitation assay for the binding of AP-2α and RBCK1 gene promoter

The binding between AP-2α and RBCK1 gene promoter was performed using a bioinformatic analysis (http://genexplain.com/transfac). According to the scores, ChIP assays were performed by using a ChIP-IT kit (Upstate, 17-295), according to the manufacturer’s protocol. 1 × 10⁶ cells were seeded on a 10 cm dish. Proteins were cross-linked to DNA by adding formaldehyde directly to the culture medium at a final concentration of 1% and incubating for 10 min at 37 °C. The cells were harvested in SDS lysis buffer and added protease inhibitors. Cell lysates were sonicated to shear DNA to lengths between 200 and 1000 bp. Sheared chromatin was precleared with protein G beads prior to incubation overnight at 4 °C with 4 µg of AP-2α antibody or IgG antibody. Purified, immunoprecipitated chromatin fragments from IP samples were subjected to PCR. PCR products were subjected to agarose gel electrophoresis and stained with ethidium bromide. The primer sequences were as follow: forward, 5’-attcatgtgcaaacggggc-3’, and reverse, 5’-aggcgacccggaggtagcatt-3’.

Electrophoretic mobility shift assay for AP-2α activity

Subcellular fractions were prepared using NE-PER Nuclear and Cytoplasmic Extract kit (Cat78833) from PIERCE. EMSA were performed following the commercial kits. AP-2α kit (AY1002) is from Panomics Company.

RNA fluorescence in situ hybridization

The RNA fluorescence in situ hybridization assay was performed by using a FISH kit (RiboBio, Guangzhou, China) according to the manufacturer’s guidelines. Cy3-labeled circRNA-RBCK1 probes and Dig-labeled locked nucleic acid miR-133a probes (Ribo-Bio, Guangzhou, China) were measured by the FISH kit, followed by visualized with a confocal microscopy.

Immunofluorescence analysis

Sections were deparaffinized, rehydrated, and blocked with 5% normal serum. Incubate tissues with primary antibody for 1 h at room temperature or overnight at 4 ^oC. After washing, incubate with fluorescence-conjugated secondary antibody for 45 minutes. Digital images were captured under a fluorescence microscopy. Quantitative analysis was performed by calculating fluorescence intensity using Alpha Ease FC software (version 4.0 Alpha Innotech).

Western blotting

Tissues were homogenized on ice in cell-lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na₂EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na₃VO₄, 1 µg/ml leupeptin) and 1 mM PMSF. Cell was lysated with cell-lysis buffer. The protein content was assayed by BCA protein assay reagent (Pierce, USA). 20 µg proteins were loaded to SDS-PAGE and then transferred to membrane. Membrane was incubated with a 1:1000 dilution of primary antibody, followed by a 1:2000 dilution of horseradish peroxidase- conjugated secondary antibody. Protein bands were visualized by ECL (GE Healthcare). The intensity (area X density) of the individual bands on Western blots was measured by densitometry (model GS-700, Imaging Densitometer; Bio-Rad). The background was subtracted from the calculated area.

Detection of ROS

Cells were incubated with DHE (10 µM) for 30 min, homogenized, and subjected to methanol extraction. HPLC was performed using a C-18 column (mobile phase: gradient of acetonitrile and 0.1% trifluoroacetic acid) to separate and quantify oxyethidium (product of DHE and O₂^–) and ethidium (a product of DHE auto-oxidation). ROS level was determined by conversion of DHE into oxyethidine. To measure ROS production in the artery in situ, fresh frozen sections of MCA were isolated from mice, and were stained with 10 μM DHE for 30 min, rinsed, and observed by fluorescent microscopy. Results were quantified using BIOQUANT Image software.

Detection of intracellular NO

NO production in culture cells was detected using the fluorescent probe. Briefly, before the end of treatment, 10 µM DAF was added to the medium and incubated for 30 min at 37 °C, then washing with PBS twice. The DAF fluorescent intensity was recorded by fluorescent reader at the wave of excitation (485 nm) and emission (545 nm).

Measurement of BH4

Homogenates of aorta or cell lysates were suspended in distilled water containing 5 mM dithioerythrol, centrifuged at 12,000 g at 4 °C for 10 min, and then subjected to oxidation in acid or base. To 100 μl aliquot of supernatant, 20 μl of 0.5 M HCl and 0.05 M iodine were added for acidic oxidation, and 20 μl of 0.5 M NaOH plus 0.05 M iodine were added for basic oxidation. After incubation for 1 h in the dark at room temperature, 20 μl HCl was added to the basic oxidation only. All mixtures received 20 μl of 0.1 M ascorbic acid for the reduction of excess iodine. Samples were then centrifuged for 10 min at 12000 g at 4 °C. Biopterin concentrations were determined by HPLC with a PR-C18 column. Elution was at a rate of 1.0 ml/min of 50 mM potassium phosphate buffer, pH 3.0. Fluorescence was detected with an excitation at 350 nm and emission at 440 nm. BH4 concentrations were calculated as the difference in results from oxidation in acid and base.

Measurements of blood glucose, cholesterol, and triglyceride

Blood glucose, homocysteine, cholesterol, triglyceride, LDL-C, and HDL-C were assayed by using commercial kits as recommend by the protocol.

Statistical analysis

All quantitative results were expressed as mean ± SD. The normal distribution of data was tested by the Kolmogorov-Smirnov test before statistical comparisons, and the normality/equal variance was tested to determine whether ANOVA was appropriate. A one-way ANOVA followed by Tukey’s HSD test, Scheffe test or Dunnett test were used to multiple comparisons between two groups. Statistical analysis was conducted using IBM SPSS statistics 20.0 (IBM Corp., Armonk, NY, USA). P < 0.05 was considered significant. GraphPad Prism version 8 (GraphPad Software, San Diego, CA, USA, www.graphpad.com) was used to make figures.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.