BD2-selective BRD4 inhibitor ZX29 mitigates myocardial ischemia–reperfusion injury via BRD4–YY1–NCOA4 axis disruption and ferroptosis inhibition

Abstract

Myocardial ischemia/reperfusion injury (MI/RI) remains a major barrier to ischemic heart disease treatment, as reperfusion therapies often exacerbate cardiac damage. Bromodomain-containing protein 4 (BRD4) is a key epigenetic target in cardiovascular diseases, but pan-BRD4 inhibitors cause severe off-target toxicity by disrupting basal bromodomain 1 (BD1) function, highlighting the need for BD2-selective agents. This study aimed to develop a BD2-selective BRD4 inhibitor and clarify its mechanism. ZX29 was developed via virtual screening of 100,000 compounds and structure-based optimization, exhibiting potent BRD4 binding with an IC₅₀ of 0.27 ± 0.03 μmol/L and K_d of 1.68 μmol/L, outperforming clinical agent RVX-208. ZX29 specifically binds BRD4 BD2 via H437/V439 residues. In murine MI/RI models, oral ZX29 significantly improved cardiac systolic and diastolic function, reduced myocardial infarct size, and mitigated lipid peroxidation. Mechanistically, ZX29 disrupted the BRD4–Yin Yang 1 (YY1) interaction and inhibited the nuclear receptor coactivator 4 (NCOA4)–ferritin heavy chain 1 (FTH1) degradation axis, suppressing lysosomal ferrous iron release and ferroptosis. Moreover, ZX29 preserved mitochondrial function by maintaining mitochondrial membrane potential and boosting ATP production. Ultimately, this study validates selective BRD4 BD2 inhibition as a safe and effective therapeutic strategy for MI/RI and establishes ZX29 as a viable clinical candidate.

Graphical abstract

The novel BRD4 BD2 inhibitor ZX29 disrupts BRD4–YY1 interaction, thereby downregulating NCOA4, suppressing NCOA4-mediated ferritinophagy/FTH1 degradation and myocardial ferroptosis, and preserving mitochondrial homeostasis.

1. Introduction

The bromodomain and extra-terminal (BET) protein family comprises four members: BRD2, BRD3, BRD4, and BRDT¹. Among these, BRD4 has emerged as a potential therapeutic target for a spectrum of diseases, including cancer, inflammation, fibrosis, cardiovascular diseases (CVDs), and viral infections2, 3, 4, 5, 6, 7. BRD4 is highly enriched in super-enhancers—regulatory genomic regions that drive the expression of critical pathogenic factors in CVDs, such as tumor necrosis factor-α (TNF-α) and alpha-smooth muscle actin (α-SMA)^8,9. Targeting BET family proteins holds considerable potential for the management of CVDs.

To date, multiple BRD4 inhibitors have been developed. For example, JQ1 and OTX015 bind to BRD4's bromodomain by mimicking acetylated lysine; purine derivatives (CPI-0610) exhibit high BRD4 selectivity; and naphthyridinopyridine compounds (TEN-010) possess favorable oral bioavailability10, 11, 12. However, pan-BRD4 inhibitors simultaneously block two distinct functions of BRD4: the basal function of BD1, which sustains cell proliferation, and the disease-specific functions of BD2, which regulate inflammation, fibrosis, and cellular stress13, 14, 15. These inhibitors have exhibited target-related toxicity, non-specific therapeutic effects, and suboptimal patient tolerance in preclinical studies and early-phase clinical trials, confining drugs such as JQ1 and OTX015 to clinical trials exclusively for cancer, and none have been approved for CVD treatment to date16, 17, 18, 19. In contrast, BD2-selective inhibitors can precisely target disease-related pathways without disrupting the basal physiological activities of normal cells, endowing them with substantial research and clinical promise20, 21, 22.

The development of BD2-selective inhibitors is based on the sequence and structural differences between the BD1 and BD2 domains, particularly variations in the amino acid composition of the ZA loop and BC loop²³. In recent years, significant progress has been made in developing inhibitors that specifically act on either the BD1 or BD2 domain of BRD4^13,24. For instance, apabetalone (RVX-208) and ABBV-744 exhibit high selectivity for BRD4 BD2; among these, RVX-208 is the most extensively studied BD2-specific inhibitor and has progressed to phase III clinical trials for CVD treatment²⁵. Nevertheless, there remains an unmet need to develop highly specific BRD4 inhibitors possessing distinct physicochemical, pharmacokinetic, and pharmacodynamic properties for CVD therapy.

Ischemic heart disease (IHD) remains the leading cause of sudden cardiac death26, 27, 28. To effectively restore blood supply to ischemic myocardium, beneficial coronary revascularization strategies—including thrombolytic therapy, angioplasty, coronary artery bypass grafting (CABG), and percutaneous coronary intervention (PCI)—can efficiently achieve myocardial reperfusion^29,30. However, the sudden restoration of oxygen supply may subsequently induce severe ischemia/reperfusion (I/R) injury and trigger fatal arrhythmias^31,32. Lethal reperfusion injury represents a major limitation to the efficacy of current reperfusion therapies; previous studies have demonstrated that reperfusion leads to more frequent ventricular fibrillation than coronary occlusion itself³³. Thus, there remains an unmet clinical need to develop safe and effective therapeutic agents for IHD patients, which can enhance cardiac tolerance to MI/RI.

Researchers have conducted preclinical studies to evaluate the potential cardioprotective effects of BRD4 inhibitors. Early studies demonstrated that JQ1 attenuated cardiac fibrosis and improved cardiac function in streptozotocin (STZ)-induced diabetic mouse models. Separately, JQ1 was shown to reverse transverse aortic constriction (TAC)-induced pathological cardiac remodeling^34,35. The novel quinolinone-based BRD4 inhibitor C-34 effectively suppressed fibroblast activation in vitro and mitigated cardiac fibrosis in vivo, which was associated with reduced BRD4 expression¹⁹. In patients with acute coronary syndrome (ACS) and type 2 diabetes mellitus (T2DM), apabetalone treatment showed a trend toward reducing major adverse cardiovascular events (MACE). Additionally, phase III clinical trials evaluating RVX-208's efficacy in heart failure with reduced ejection fraction (HFrEF) and heart failure with preserved ejection fraction (HFpEF) are currently underway³⁶.

Despite the well-documented cardioprotective benefits of BRD4 inhibitors mentioned above, their effects on MI/RI remain insufficiently investigated. Therefore, the development of more potent BRD4 inhibitors with favorable druggability has become an urgent need. Such inhibitors not only expand the chemical structural diversity of BRD4 inhibitors and fill the gap in existing compound libraries but also provide a novel research avenue for in-depth exploration of BRD4's functional mechanisms in the pathogenesis and progression of cardiovascular diseases.

In the present study, high-throughput virtual screening was performed against BRD4, which ultimately identified a potent compound 25 with an IC₅₀ value of 6.95 μmol/L. Subsequent structure-based optimization and biological evaluation led to the generation of a novel series of coumarin-6-carboxamide derivatives as BRD4 inhibitors. Eventually, ZX29 was developed, which exhibited significant efficacy against myocardial ischemia–reperfusion injury. Mechanistically, we found that ZX29 directly engages in extensive interactions with the BD2 binding pocket of long-form BRD4, thereby inhibiting the activation of the NCOA4–FTH1 pathway mediated by the BRD4–YY1 interaction. This inhibition suppresses ferroptosis in cardiomyocytes and alleviates ferroptosis-induced mitochondrial dysfunction. Collectively, our data provide a new insight into the therapeutic strategy of BD2-specific BRD4 inhibitors for myocardial ischemic diseases.

2. Materials and methods

2.1. Materials

Primary antibodies against BRD4 (Cat#13440), GPX4 (Cat#52455), and YY1 (Cat#46395) were obtained from Cell Signaling Technology (MA, USA). Antibodies against Bcl2 (Cat#sc-7382) and BAX (Cat#sc-7480) were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). β-MHC (Cat#M8421) was obtained from Sigma–Aldrich (Saint Louis, USA). GAPDH (Cat#60004-1-Ig), Actin (Cat#20536-1-AP), SMA (Cat#14395-1-AP), PTGS2 (Cat#27308-1-AP), NCOA4 (Cat#83394-4-RR), FTH1 (Cat#11682-1-AP), ACSL4 (Cat#22401-1-AP), Fibronectin (Cat#15613-1-AP) antibodies were purchased from Proteintech Group (Wuhan, China). The antibody against ANF was purchased from Cloud Clone Corporation (Wuhan, China).

ZX29 was synthesized and purified to >99% purity in our laboratory. RVX-208 and Fer-1 were purchased from TargetMol (Boston, USA). Ang II was purchased from Sigma–Aldrich (Shanghai, China). DuoLink in situ proximity ligation assay (PLA) Detection Kit was purchased from Sigma–Aldrich (Saint Louis, USA). Rhodaminephalloidin was purchased from FuShen (Shanghai, China). The Mouse Troponin T, cardiac muscle (cTnT) ELISA Kit and mouse creatine kinase MB isoenzyme (CK-MB) ELISA kit, mouse N-terminal pro-brain natriuretic peptide (NT-proBNP) ELISA kit were obtained from CUSABIO (Wuhan, China). BRD4 (BD2) Binding assay kit was obtained from Revvity (MA, USA). 2,7-Dichlorodihydrofluorescein diacetate and Lipofectamine 3000 Transfection Reagent were supplied by Thermo Fisher Scientific (USA). The Cell Counting Kit-8 (CCK-8) was purchased from Beyotime Biotechnology (Beijing, China). TMRE dye was purchased from Abcam (Cambridge, UK). ERtracker dye and mitotracker dye were purchased from Invitrogen (USA). Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), and phosphate-buffered saline (PBS) were from Gibco (NY, USA). TUNEL staining kit was from Vazyme (Shanghai, China). pcDNA-4mtD3cpv was purchased from Semrock (Rochester, NY). The Cell Counting Kit-8 (CCK-8) was purchased from Beyotime Biotechnology (Beijing, China). The pan-BET inhibitor JQ1 was purchased from TargetMol (TargetMol, USA, T2110). ELISA kits for mouse TNF-α and IL-1β, as well as assay kits for total cholesterol (TC), triglycerides (TG), alanine aminotransferase (ALT), creatinine, and hydroxyproline were all purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China) and Elabscience (Wuhan, China). SYPRO Orange dye was purchased from (Invitrogen, USA).

2.2. Clinical samples

This study was approved by the Human Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University (Ethics approval No [2015] 169; Approval date: 2015-11-24), and followed the principles for the use of human tissues stated in the Declaration of Helsinki. All patients provided written informed consent before participating in the study. We collected cardiac tissue samples from patients who were clinically diagnosed with myocardial infarction, developed decompensated heart failure after undergoing percutaneous coronary intervention (PCI), and eventually died. Control heart samples were obtained from prospective multi-organ donors who did not exhibit cardiovascular pathology but could not be transplanted for technical reasons. These tissue samples were used for immunohistochemistry, immunofluorescence, and immunoprecipitation experiments.

2.3. Animals

SPF-grade C57BL/6 mice (8–10 weeks old, 23–28 g) were acquired from the Laboratory Animal Center of Sun Yat-sen University (Guangzhou, China). All animal experiments in this research were sanctioned by the Research Ethics Committee of Sun Yat-sen University and were in line with the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996). The mice thus obtained were kept in temperature-regulated cages with a 12-h light-dark cycle and had unrestricted access to water and a standard chow diet. At the conclusion of the experiment, all mice were humanely euthanized by intraperitoneal injection of sodium pentobarbital (150 mg/kg, Merck). The experimental protocols were executed in accordance with China's Animal Welfare Legislation, and were examined and approved by the Sun Yat-sen University Committee on the Ethics of Laboratory Animal Care and Use (Ethics approval No. SYSU-IACUC-2022-000485; Approval date: 2022-02-28).

Upon arrival at the experimental facility, mice were acclimatized for one week prior to experimentation. They were randomly assigned to five groups: control group, model group, low-dose ZX29 group (5 mg/kg/day), high-dose ZX29 group (20 mg/kg/day), and RVX-208 group (positive control, 20 mg/kg/day). ZX29 and RVX-208 powders were dissolved in a solution consisting of 10% DMSO and 90% corn oil. Drugs were administered once daily via oral gavage, initiated 4 days prior to MI surgery, continued on the day of surgery, and maintained until 1 day post-surgery. Sham and MI/RI groups received equal volumes of the vehicle solution via the same route.

2.4. Animal model of MI/RI

For the prophylactic regimen, drugs were administered once daily via oral gavage (p.o.), starting 4 days prior to MI surgery and continuing uninterrupted through the day of reperfusion until sacrifice on the following day. After oral administration of ZX29, the mice were anesthetized with 2% isoflurane. The heart was exposed by a left thoracic incision. Myocardial infarction was produced by placing a 7.0 silk suture slipknot around the left anterior descending coronary artery. After 45 min of ischemia, the slipknot was released to allow reperfusion for 24 h or 2 weeks. Sham-operated mice underwent left thoracotomy only. For the therapeutic regimen, oral administration of ZX29 or JQ1 was initiated immediately at the onset of reperfusion (0 h), with a second dose administered 12 h later to achieve rapid steady-state coverage. For long-term evaluations, this was subsequently transitioned to a once-daily maintenance dose.

To ensure the consistency and reliability of the model, surface lead II electrocardiogram (ECG) was monitored throughout the procedure. Successful myocardial ischemia was defined by specific criteria: (1) visual confirmation of the left ventricular anterior wall turning pale immediately after ligation; and (2) significant ST-segment elevation on the ECG. Successful reperfusion was confirmed by: (1) the return of a bright red color to the ischemic myocardium; and (2) the resolution of ST-segment elevation. Mice that failed to meet these criteria or died during the surgery and 24-h reperfusion period were excluded from the study. Mortality rates were recorded for quality control: no mortality occurred in the Sham group. The MI/RI model group had a mortality rate of approximately 20%. In the treatment groups, mortality was 10% for ZX29 and RVX-208, and 20% for JQ1.

2.5. Cell culture conditions

Rat cardiomyocyte cell line (H9c2), human cardiomyocyte cell line (AC16), and mouse cardiomyocyte cell line (HL-1) were obtained from the American Type Culture Collection (ATCC). H9c2 and AC16 cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. HL-1 cells were cultured in Claycomb medium supplemented with 10% FBS, 0.1 mmol/L norepinephrine, 2 mmol/L l-glutamine, and 1% penicillin/streptomycin. All cell lines were maintained in a humidified incubator at 37 °C with 5% CO₂.

2.6. Differential scanning fluorimetry (DSF)

To evaluate the BD2-domain selectivity of ZX29, differential scanning fluorimetry (DSF) assays were performed using recombinant BRD4-BD1 and BRD4-BD2 proteins, adapting the standardized protocol established by Niesen et al. Briefly, purified BRD4-BD1 or BRD4-BD2 proteins were diluted to a final concentration of 2 μmol/L in the assay buffer (10 mmol/L HEPES, 150 mmol/L NaCl, pH 7.4). The protein solutions were incubated with SYPRO Orange dye (Invitrogen) at a final concentration of 5 × , and ZX29 was added at a final saturating concentration of 10 μmol/L. An equivalent volume of DMSO (final concentration 0.1% v/v) served as the vehicle control. The reaction mixtures (20 μL total volume per well) were loaded into a 96-well PCR plate, sealed, and centrifuged. The thermal melting experiments were performed using a real-time PCR system, with the temperature continuously ramped from 20 to 90 °C at a steady rate of 1 °C/min. The melting temperature (T_m) was determined by plotting the raw fluorescence intensity against temperature and fitting the data to a Boltzmann sigmoidal curve using non-linear regression. The T_m was defined as the inflection point of this melting curve.

2.7. Western blot

Total proteins from cardiac tissues or cultured cells were extracted using RIPA lysis buffer supplemented with protease and phosphatase inhibitor cocktails. Protein concentrations were determined using a BCA Protein Assay Kit (Beyotime, China). Equal amounts of protein samples (30–50 μg) were separated by 10% or 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred onto PVDF membranes. The membranes were blocked with 5% non-fat milk in TBST for 1 h at room temperature, followed by overnight incubation at 4 °C with the respective primary antibodies. After washing with TBST three times, the membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system and quantified using ImageJ software.

2.8. Dual-luciferase reporter assay

To functionally validate the specific YY1 binding motif on the Ncoa4 promoter, a targeted dual-luciferase reporter assay was performed. The wild-type Ncoa4 promoter fragment containing the core YY1 motif (TCCATC) and a mutant fragment (Ncoa4-MUT, in which the core motif was mutated to TGGTAC) were cloned into the pGL3-basic vector to generate Ncoa4-WT-Luc and Ncoa4-MUT-Luc plasmids, respectively. Cells in 24-well plates were co-transfected with 500 ng of the reporter plasmid (Ncoa4-WT-Luc or Ncoa4-MUT-Luc), 50 nmol/L of siRNA (si-NC or si-YY1), and 50 ng of pRL-TK Renilla control plasmid using Lipofectamine 3000. At 48 h post-transfection, cells were lysed using 1 × passive lysis buffer. Firefly and Renilla luciferase activities were sequentially quantified using the Dual-Luciferase Reporter Assay System. The relative luciferase activity was calculated as the ratio of Firefly to Renilla luminescence to normalize for transfection efficiency.

2.9. Chromatin immunoprecipitation (ChIP) and qPCR

Cardiomyocytes were cross-linked with 1% formaldehyde and quenched with glycine. Chromatin was fragmented to 150–900 bp via micrococcal nuclease digestion and mild sonication. The sheared chromatin was incubated overnight at 4 °C with antibodies against BRD4, YY1, or normal rabbit IgG, followed by capture with Protein G magnetic beads. After sequential washes with low- and high-salt buffers, the DNA was eluted, de-cross-linked with proteinase K at 65 °C, and purified. The enriched DNA was analyzed by qPCR using SYBR Green Master Mix and specific primers for the Ncoa4 promoter.

2.10. In vivo BRD4 knockdown model

To establish a strictly cardiomyocyte-specific BRD4 knockdown model, we utilized adeno-associated virus serotype 9 (AAV9), which exhibits a strong natural tropism for cardiac tissue. Mice were injected via the tail vein with AAV9 carrying short hairpin RNA targeting Brd4 or a scrambled control sequence. To confirm exclusive expression within cardiomyocytes and prevent off-target knockdown in cardiac fibroblasts or other organs, the transcription of the shRNA cassettes was driven strictly by the cardiac-specific troponin T (cTnT) promoter. Following the injection, mice were housed under standard conditions for a 3-week viral incubation period to achieve optimal and stable BRD4 knockdown in the myocardium. To determine whether BRD4 is essential for the cardioprotective effects of ZX29, we subjected these mice to the identical combined prophylactic and therapeutic dosing regimen used in our previous wild-type animal experiments. Specifically, starting exactly 21 days post-AAV injection, mice received a once-daily oral administration of high-dose ZX29 (20 mg/kg/day) or an equal volume of vehicle. This once-daily dosing was initiated 4 days prior to the MI/RI surgery, continued on the day of surgery, and maintained for 2 consecutive days following the onset of reperfusion.

2.11. Biochemical and hematological analysis

At the end of the treatment periods, blood samples were collected from mice via the retro-orbital plexus under deep anesthesia. For hematological analysis, whole blood was collected into EDTA-coated tubes. For serum biochemical analysis, whole blood without anticoagulants was allowed to clot at room temperature and subsequently centrifuged at 1500×g for 15 min at 4 °C. Serum levels of alanine aminotransferase (ALT), creatinine, total cholesterol (TC), and triglycerides (TG) were quantified using commercial enzymatic assay kits (Nanjing Jiancheng, China). Furthermore, serum concentrations of cardiac troponin T (cTnT), creatine kinase-MB (CK-MB), and NT-proBNP were determined using specific mouse ELISA kits according to the manufacturer's protocols (Elabscience, Wuhan, China). To quantify myocardial fibrosis, a biochemical hydroxyproline assay was performed. Infarcted myocardial tissues were weighed, homogenized, and subjected to alkaline hydrolysis. The hydroxyproline content in the tissue homogenates was measured using a colorimetric assay kit following the manufacturer's instructions.

2.12. Isolation and culture of neonatal mouse cardiomyocytes (NMCMs)

Fresh myocardial tissues from neonatal C57BL/6J mice (1–2 days old) were harvested and rinsed three times with pre-chilled PBS to remove residual blood. The tissues were minced into fragments of approximately 1 mm³, followed by the addition of a digestive solution containing 0.1% collagenase II and 0.05% trypsin at a ratio of 10 mL per gram of tissue. The mixture was subjected to digestion with gentle agitation at 37 °C for 20 min. After digestion, the sample was centrifuged at 300×g for 5 min, and the supernatant was discarded. The resulting pellet was resuspended in DMEM supplemented with 10% FBS, and the suspension was filtered through a 70 μm cell strainer to eliminate tissue debris. The filtrate was collected and centrifuged again at 300×g for 5 min; the pellet was then resuspended in fresh medium. The cell suspension was seeded into culture dishes and incubated at 37 °C with 5% CO₂ for 2 h. Non-adherent cells, which contained NMCMs, were collected thereafter. Trypan blue staining was performed for cell counting, and the cell concentration was adjusted to 1 × 10⁶ cells/mL for subsequent experiments. Aseptic conditions were maintained throughout the entire procedure, and all reagents were pre-cooled to 4 °C to minimize cell damage.

2.13. Hypoxia/reoxygenation (H/R)

RVX-208 and ZX29 (synthesized in-house, purity >99% by HPLC) were dissolved in dimethyl sulfoxide to prepare 10 mmol/L stock solutions, which were aliquoted and stored at −20 °C. For in vitro experiments, stock solutions were diluted with serum-free DMEM/F12 medium to final working concentrations: 1 μmol/L RVX-208, 0.5 μmol/L ZX29, and 1 μmol/L ZX29. The final concentration of DMSO in all groups was ≤0.1% (v/v), which had no detectable cytotoxicity on NMCMs. When NMCMs reached 80% confluence, the medium was replaced with serum-free DMEM/F12, and cells were divided into five groups: DMSO, H/R, H/R + ZX29^low, H/R + ZX29^Hi, H/R + RVX-208. NMCMs were incubated with 1 μmol/L RVX-208, or 0.5 μmol/L ZX29 (low) or 1 μmol/L ZX29 (Hi) for 2 h. After pretreatment, NMCMs were exposed to hypoxia by transferring the plates to a three-gas incubator (Thermo Scientific, Cat. No. 3111) maintained at 37 °C with a gas mixture of 1% O₂, 5% CO₂, and 94% N₂ for 4 h. For reoxygenation, the medium was replaced with fresh serum-free DMEM/F12 (pre-warmed to 37 °C), and cells were returned to a standard incubator (37 °C, 5% CO₂, 95% air) for 6 h. At the end of reoxygenation, cells and culture supernatants were collected for subsequent molecular and functional assays.

2.14. Cellular thermal shift assay

After transfection with BRD4 overexpression or mutant plasmids, HEK293T cells were treated with 40 μmol/L ZX29 or vehicle for 5 h. Subsequently, cells were collected by trypsinization and resuspended in PBS. The resuspended cells were divided into separate tubes, heated at gradient temperatures for 5 min, and subjected to six freeze-thaw cycles in liquid nitrogen. Supernatants were obtained by centrifugation at 13,000×g for 15 min at 4 °C, mixed with loading buffer, and boiled for 5 min. Western blot was performed to detect BRD4 protein levels.

2.15. Compound-conjugated magnetic bead immunoprecipitation assay

To validate the ZX29–BRD4 interaction, ZX29 was covalently conjugated to TurboBeads carboxy nanoparticles via esterification using standard EDC/NHS coupling chemistry. For the in vitro pull-down assay, whole-cell lysates from H9c2, AC16, and HL-1 cells were prepared using a mild non-denaturing lysis buffer (50 mmol/L Tris–HCl pH 7.4, 150 mmol/L NaCl, 0.5% NP-40, supplemented with protease inhibitor cocktails). The pre-cleared cell lysates (∼1 mg of total protein) or purified recombinant BRD4 proteins (WT or the H437E/V439K mutant) were incubated with the ZX29-conjugated magnetic nanoparticles overnight at 4 °C under continuous gentle rotation. To confirm binding specificity within the BD2 pocket, four parallel experimental conditions were evaluated: (1) a beads control using unconjugated nanoparticles to assess background binding; (2) an experimental group using ZX29-conjugated beads incubated with WT BRD4; (3) a competition assay where WT BRD4 was pre-incubated with a molar excess of free ZX29 competitor prior to the addition of conjugated beads; and (4) a mutation assay utilizing the BD2-binding deficient BRD4-H437E/V439K protein. After stringent washing with high-salt buffer (300 mmol/L NaCl), the enriched proteins were eluted by boiling in 1 × SDS loading buffer and analyzed by Western blot.

2.16. Homogeneous time-resolved fluorescence (HTRF) assay

8 μL ZX29 compound and 4 μL GST-BRD4 (10–50 nmol/L) were added to microplate wells, shaken at 200×g for 1 min (Eppendorf North America, 5810R, Hauppauge, NY, USA), and incubated at room temperature for 30 min. Subsequently 4 μL biotin-H4 (20–100 nmol/L) was added, shaken at 300 rpm for 1 min, and incubated at room temperature for 1 h to form BRD4-biotin-H4 complexes. Anti-GST-cryptate and streptavidin-d2 were mixed 1:1 in buffer; 4 μL of this mixture was added, shaken at 300 rpm for 1 min. The plate was sealed, protected from light, and incubated at room temperature for 3 h. Fluorescence signal was detected by PerkinElmer Envision microplate reader equipped with 320 nm excitation and 620 and 665 nm emission. The results were analyzed by SPSS 20 and GraphPad Prism nine software.

2.17. Construction of stable BRD4-KO cell lines

The design of BRD4-KO cell lines was carried out with the aid of the website crispr.mit.edu, and they were established through the employment of CRISPR/Cas9 gene-editing technology. The target sequences are presented in Supporting Information Tables S1–S3. Recombinant LentiCRISPR-sgBrd4, along with plasmids VSV-G and psPAX2, was transfected into HL-1 cells. This transfection led to the generation of lentiviral particles. These particles were then utilized to infect HL-1 cells for the purpose of gene knockout. Post-infection, the cells were subjected to selection using puromycin and were sorted into 96-well dishes. After a period of 2 weeks, the surviving clones were picked out, expanded, and further selected based on the BRD4 expression level determined by Western blot, aiming to confirm the successful knockout of BRD4.

2.18. Echocardiographic analysis

Mice were subjected to transthoracic echocardiography (VisualSonics VeVo 2100 Imaging System) for assessment of cardiac structure and function after 24 h reperfusion. Echocardiography was performed, and M-mode tracings were recorded.

2.19. Oxygen consumption measurements

Oxygen consumption rate (OCR) was measured using a Seahorse XF analyzer according to the manufacturer's protocol. 1 h before the assay, the culture medium was replaced with Seahorse XF DMEM supplemented with 5 mmol/L glucose, 1 mmol/L pyruvate, and 10 mmol/L glutamine. After incubation at 37 °C for 1 h in a CO₂-free incubator, the oxygen consumption rates were determined in basal conditions, as well as in the presence of oligomycin (1.5 μmol/L), FCCP (1 μmol/L), and Rotenone/Antimycin A (0.5 μmol/L). Subsequently, the oxygen consumption rates were normalized against the total cell number, which was quantified by Hoechst 33,342 staining at the conclusion of the Seahorse experiment.

2.20. Evans blue/TTC staining

After 24 h reperfusion, 1.5% Evans blue (Sigma–Aldrich, E2129) was injected into the aorta. The heart was removed and frozen at −80 °C immediately. Subsequently, the heart was horizontally sectioned into 5–6 slices, and incubated with 1% 2, 3, 5-triphenyl tetrazolium chloride (TTC, Beyotime, China) for 15 min at 37 °C. The left ventricular (LV) area, area at risk (AAR), and infarct area (IA) of each slice were quantified using ImageJ software.

2.21. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining

Apoptosis in both cultured cardiomyocytes and myocardial tissues was assessed using a TUNEL Staining Kit according to the manufacturer's instructions. For myocardial tissues, mouse hearts were perfused with ice-cold phosphate-buffered saline and fixed with 4% paraformaldehyde, embedded in paraffin, and coronally sectioned (6 μm thick). Three to five sections from each heart were deparaffinized in xylene, rehydrated through a graded ethanol series, and treated with proteinase K for antigen permeabilization. Subsequently, the sections were subjected to the infarct area TUNEL staining assay. For in vitro cultured cells: NMCMs subjected to various treatments were washed with cold PBS, fixed with 4% paraformaldehyde for 15 min at room temperature, and permeabilized with 0.1% Triton X-100 in PBS for 5 min. The cells were then washed and incubated with the TUNEL reaction mixture in a humidified dark chamber at 37 °C for 1 h. Following the TUNEL reaction, both tissue sections and cell coverslips were counterstained with DAPI for 5 min to visualize the nuclei. Fluorescent images were acquired using a confocal laser scanning microscope. The apoptotic rate was quantified by calculating the percentage of TUNEL-positive cells relative to the total number of DAPI-stained nuclei using ImageJ software.

2.22. Surface plasmon resonance measurement (SPR)

Purified recombinant wild-type BRD4 and the BD2-binding deficient BRD4-H437E/V439K mutant proteins were utilized as ligands and immobilized onto a gold nanoparticle sensor chip via capture-coupling. Subsequently, varying gradient concentrations of the tested compounds were injected into the flow chamber in filtered PBS buffer at a steady flow rate of 20 μL/min. The compounds were passed over the sensor surface for 5 min to monitor the protein–ligand association phase. Following each injection, the dissociation phase was recorded, and the sensor chip surface was regenerated by injecting a mild hydrochloric acid solution. The resulting binding sensorgrams were analyzed using the Biacore 8K Evaluation Software according to the manufacturer's guidelines. The kinetic parameters, including the association rate constant (k_a), the dissociation rate constant (k_d), and the equilibrium dissociation constant (K_d = k_d/k_a), were calculated by fitting the data to a 1:1 Langmuir binding model, with adjustments made for wavelength shifts associated with protein concentration variations.

2.23. Immunoprecipitation coupled with mass spectrometry (IP–MS)

Cells were lysed in total lysis buffer (10 mmol/L HEPES, pH 7.4, 150 mmol/L NaCl, 1 mmol/L EGTA, 1% NP-40, and protease inhibitor cocktail). 3 mg of lysates were immunoprecipitated with anti-BRD4 antibody overnight and then incubated with protein A/G beads for 4 h. Bound proteins were eluted by 0.2 mol/L glycine (pH 2.5). Proteins from each sample were precipitated with 30% trichloroacetic acid (v/v), washed with ice-cold acetone, air-dried, trypsinized, and analyzed by tandem mass spectrometry. Digested peptides were reconstituted with 0.1% formic acid and analyzed by LC–MS/MS with automatic inclusion of isotopic C13 incorporated ions. DTASelect was used to generate PSMs search results with a max FDR of 5%, achieving a protein FDR <1% when at least two peptides per protein were assigned. BRD4-interacting protein candidates enriched in mitochondria were analyzed by Ingenuity IPA software.

2.24. In situ proximity ligation assay (PLA)

In situ PLA was performed using the Duolink® In Situ Red Starter Kit Mouse/Rabbit (Sigma–Aldrich, USA) according to the manufacturer's protocol. Briefly, fixed cells or tissue slides were blocked and then incubated overnight at 4 °C with the following primary antibody pairs: rabbit anti-BRD4 (1:500) and mouse anti-YY1 (1:100), or rabbit anti-BRD4 (1:500) and mouse anti-H3K27ac (1:200). After washing, the samples were incubated with PLA probes (anti-rabbit PLUS and anti-mouse MINUS) for 1 h at 37 °C. Subsequent ligation and rolling-circle amplification (RCA) were carried out using the provided ligase and polymerase to generate concatemeric products. These products were hybridized with fluorescently labeled oligonucleotides, appearing as distinct red spots. Nuclei were counterstained with DAPI, and signals were visualized using a confocal laser scanning microscope (Fluoview FV1000, Olympus).

2.25. Immunofluorescence

Human or mouse heart sections (6 μm thick) were deparaffinized in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed in Tris-EDTA buffer (pH 9.0) at 95 °C for 30 min. To block non-specific binding and permeabilize the tissue, sections were incubated with PBS containing 10% goat serum and 0.2% Triton X-100 for 1 h at 37 °C. The sections were then incubated overnight at 4 °C with primary antibodies against BRD4 (1:200), GPX4 (1:500), ACSL4 (1:500), or FTH1 (1:500). After three washes with PBS, the slides were incubated with corresponding Alexa Fluor-conjugated secondary antibodies (1:500) for 2 h at room temperature in the dark. Nuclei were counterstained with DAPI for 5 min. Images were acquired using a confocal laser scanning microscope (Zeiss, Germany) and analyzed using ZEN software.

2.26. GST pull-down assay

pGEX-GST-BRD4 was transformed into Rosetta (DE3) cells and induced for expression by isopropylthio-β-d-galactoside. GST-BRD4 fusion protein was added to lysate of mouse heart tissue and mixed at 4 °C overnight. GST protein was used as a negative control. After incubating the mixture with GST-beads (Beyotime Institute of Biotechnology, Nantong, China) for another 2 h at 4 °C, beads were harvested by centrifugation and washed four times with 0.3% NP-40 buffer. Samples were boiled in SDS loading buffer and subjected to Western blot analysis. The beads were washed 3 times with washing buffer and then denatured by the addition of 2 × loading buffer, boiled 5 min and analysed by Western blot.

2.27. Measurement of caspase-3 activity

Caspase-3 activity was determined using a fluorometric kit from BD Biosciences. Briefly, heart tissue or cultured cardiomyocytes were lysed on ice and the supernatant was obtained. Reactions were carried out in assay buffer with 10 mmol/L DTT and 50 μg proteins. The fluorescence emission of AFC was measured by Spectra Max-Plus microplate spectrophotometer (excitation at 400 nm and emission at 505 nm). Caspase-3 activity was presented as nmol AFC/h/mg protein.

2.28. Statistical analysis

Statistical analyses were performed using GraphPad Prism 9.0. Multiple group comparisons were analyzed by one-way ANOVA followed by Tukey's post hoc test. Differences between two groups were assessed by unpaired two-tailed Student's t-test. Data are presented as mean ± standard error of mean (SEM), where n represents the number of independent experiments or the number of animals. A P value <0.05 was considered statistically significant.

3. Results

3.1. Screening for potent candidates of BRD4 inhibitors

In our previous study, BRD4 was identified as a potential pathogenic target in myocardial diseases, prompting high-throughput screening for small-molecule BRD4 inhibitors³⁷. Virtual screening was performed via MOE-based molecular docking using approximately 100,000 small molecules from the ChemDiv database, with five independent conformations generated per molecule. Molecules were ranked by the London dG score of their optimal conformation, yielding the top 2000 candidates (Fig. 1A). These 2000 molecules then underwent ADMET filtering to exclude compounds with pharmacokinetic/toxicological liabilities, resulting in 500 molecules. Hierarchical clustering analysis was further conducted on these 500 molecules: structural diversity was evaluated via Tanimoto similarity coefficients and binding mode uniqueness via conformation superimposition in the BRD4 active site. This combined analysis of structural diversity and binding modes yielded 100 candidate inhibitors with distinct chemical scaffolds. Later, we synthesized these candidate compounds and detected their binding signals with BRD4 at a concentration of 5 μmol/L through single-concentration and multi-concentration Surface Plasmon Resonance (SPR) screening techniques (Fig. 1B). A total of 20 hit compounds were identified (Table 1), with compound 25 exhibiting a dissociation constant (K_d) of 13.2 μmol/L for BRD4. To investigate how benzene ring substituents affect potency, derivatives 26–44 were synthesized and evaluated against BRD4 via a homogeneous time-resolved fluorescence (HTRF) assay. As summarized in Table 1, the structure–activity relationship (SAR) at the R1 position was highly substituent-sensitive. Substituent electronic properties modulated molecular electron density: ethynyl groups in compounds 29/30 acted as electron-withdrawing groups (EWGs), the amino group in compound 32 as an electron-donating group (EDG), and the ester group in compound 33 showed dual EWG (carbonyl) and EDG (alkoxy) characteristics. Notably, none of these R1 substitutions improved activity relative to compound 25. Synthesis of derivatives 34–41 revealed that large heterocyclic groups at R1 impaired activity. In contrast, compound 42—with a simplified structure and direct linkage to the coumarin core (vs. the extended alkyl chain in 25)—exhibited an IC₅₀ of 2.1 μmol/L. Further optimization of compound 42 (via aromatic ring substituent and spatial arrangement modifications) yielded ZX29 (molecule 44), which displayed potent BRD4 inhibition (IC₅₀ = 0.27 ± 0.03 μmol/L), corresponding to an approximately 26-fold potency enhancement compared with compound 25 (Fig. 1C). NMCMs treated with 5 μmol/L of the tested compounds for 24 h maintained >80% viability, indicating low cytotoxicity (Fig. 1D). Ultimately, ZX29 was selected as the lead candidate for subsequent cellular and in vivo experiments.

Figure 1.

Discovery of ZX29, a novel BRD4 inhibitor. (A) Graphic illustration of the screening of compounds. (B) The binding signal values of compounds to the recombinant BRD4 protein at a concentration of 5 μmol/L. (C) Discovery of coumarin-6-carboxamide derivatives as BRD4 inhibitors. (D) NMCMs were treated with indicated compounds (5 μmol/L) for 24 h, respectively. Cell viability was measured by CCK-8 assay. The data are presented as mean ± SEM, n = 3.

Table 1.

The structure–activity relationships of 3-substituted coumarin derivatives 25–44.

Compd.	R1	BRD4 IC50 (μmol/L)
25		6.95 ± 0.42
26		5.65 ± 0.12
27		9.78 ± 0.18
28		8.57 ± 0.38
29		16.85 ± 0.55
30		12.23 ± 0.64
31		12.81 ± 0.38
32		6.64 ± 0.08
33		4.29 ± 0.47
34		10.22 ± 0.40
35		10.43 ± 0.08
36		13.26 ± 0.31
37		12.89 ± 0.48
38		15.27 ± 0.23
39		11.81 ± 0.18
40		18.81 ± 0.25
41		19.31 ± 0.09
42		2.10 ± 0.19
43		3.41 ± 0.12
44 (ZX29)		0.27 ± 0.03

The IC₅₀ was calculated from HTRF assay. Data are expressed as the means ± SEM, n = 3.

3.2. Chemical synthesis of ZX29

The synthesis of novel 3-(4-hydroxy-3,5-dimethylphenyl)-6,8-dimethoxy-2H-chromen-2-one was illustrated in Fig. 2. The chemical structure of ZX29 was confirmed by ¹H NMR and ¹³C NMR spectroscopy (Fig. 2, Supporting Information Figs. S1 and S2). The intermediate 2 was firstly obtained through substitution reaction of commercially available 3,5-dimethoxybenzaldehyde and NBS in 95% yield. The intermediate 4 was obtained through addition and reduction reactions of commercially available 2,6-dimethylphenol and glyoxylic acid in 87% yield. Next, intermediates 2 and 4 were reacted by Perkin reaction to obtain compound 5 in 74% yields. Subsequently, under N₂ atmosphere, the desired compound 6 was obtained through CuI-catalyzed cyclization of intermediate 5 in 45% yields.

Figure 2.

Reagents and conditions: (a) NBS, CH3CN, 0 °C, 6 h; (b) Glyoxylic acid, NaOH, H₂O, 50 °C 12 h; (c) SnCl₂‧2H₂O, HCl, 80 °C 16 h; (d) Ac₂O, Et₃N, 120 °C 12 h; (e) CuI, K₂CO3, 1,10-phenanthroline, DMF, N₂, 130 °C 12 h.

3.3. ZX29 directly targets the BD2 domain of BRD4

To explore the binding mode of ZX29 to BRD4, molecular docking was performed using MOE2019. Simulations showed ZX29 localized to the ligand-binding pocket of BRD4 bromodomain 2 (BD2) (Fig. 3A and B). BRD4 BD2 mediates epigenetic regulation via specific binding to acetylated lysine (Kac) residues, with the WPF motif (critical for ligand binding) forming a hydrophobic platform via key amino acid side chains. Notably, the coumarin core's aromatic ring in ZX29 established dual H–π interactions with Val439—this multi-site synergy presumably enhances BD2 selectivity (avoiding non-specific BD1–Asp144 binding) and binding stability. Additionally, the R1-position phenyl group of ZX29 formed extensive hydrophobic interactions with BRD4 residues Val439, His437, and Leu385 (Fig. 3C), supporting ZX29's BRD4-binding potential. Molecular dynamics simulations confirmed ZX29's stable retention in the BRD4 BD2 pocket (Fig. 3D). SPR assays revealed a dose-dependent ZX29–BRD4 interaction with a K_d of 1.68 μmol/L, which is lower than that of the reported BRD4 inhibitor RVX-208 (K_d = 6.21 μmol/L) (Fig. 3E and F). MicroScale Thermophoresis (MST) further validated this interaction: increasing ZX29 concentrations altered BRD4 thermophoresis, yielding a K_d of 1.6 ± 0.3 μmol/L (Fig. 3G). To establish the domain selectivity profile of ZX29, Differential Scanning Fluorimetry (DSF) was performed against both the BD1 and BD2 domains of BRD4. Remarkably, ZX29 induced robust thermal stabilization of the BRD4-BD2 domain (ΔT_m = 8 °C), whereas the thermal unfolding profile of the highly conserved BRD4-BD1 domain remained entirely unaffected (ΔT_m = 0 °C) (Supporting Information Fig. S3). These data confirm that ZX29 possesses exceptional selectivity for the BD2 domain. To validate the critical binding residues predicted by our molecular docking, a BRD4-H437E/V439K mutant was generated via site-directed mutagenesis. Notably, SPR analysis demonstrated that the introduction of these mutations almost completely abrogated the binding affinity of ZX29 (Supporting Information Fig. S4). Consistent with these in vitro kinetic findings, Cellular thermal shift assay (CETSA) in HEK293T cells showed that ZX29 significantly enhanced the thermal stability of exogenous wild-type BRD4, but failed to stabilize the BRD4-H437E/V439K mutant (Fig. 3H–J). Co-immunoprecipitation (Co-IP) with carboxyl-coated nanoparticles further verified the interaction: in H9C2, AC16, and HL-1 cell lysates, ZX29-conjugated nanoparticles efficiently pulled down BRD4 (Fig. 3K). To validate the binding specificity, in vitro competition and mutation pull-down assays were performed. While ZX29-conjugated beads demonstrated strong affinity for wild-type BRD4, this interaction was completely abrogated by either pre-incubation with a free ZX29 competitor or the introduction of H437E/V439K mutations within the BD2 pocket, alongside a clean background in the unconjugated beads control (Fig. 3L). Collectively, these data establish that ZX29 specifically and competitively binds to the BD2 domain of BRD4.

Figure 3.

ZX29 binds BRD4. (A) The chemical structure of ZX29. (B) Docking of ZX29 and BRD4 (PDB ID: 6C7Q). (C) The 2D image reveals that ZX29 interacts with several amino acids of BRD4 (H437, V439). Blue dotted line signifies a hydrogen bond. (D) The molecular dynamics simulation for ZX29–BRD4 complexes. (E) SPR analysis of the binding affinity of ZX29 to BRD4 protein .(F) SPR analysis of the binding affinity of RVX-208 to BRD4 protein. (G) The K_d value of BRD4 with ZX29 was detected by MST. (H–J) CETSA presented the thermal stability of BRD4-WT, and BRD4-H437E/V439K plasmid mutant proteins after treatment with ZX29 (40 μmol/L) for 5 h. (K) Protein lysates from H9C2, AC-16, and HL-1 cells (lanes 1–3, respectively) were incubated overnight with the compound ZX29 conjugated to magnetic beads or with magnetic beads alone lanes 4–6, respectively) and were analyzed by Western blot. (L) Validation of the specific binding between ZX29 and the BRD4 BD2 pocket via an in vitro competitive pull-down assay. Recombinant WT-BRD4 or the mutant BRD4-H437E/V439K protein was incubated with either unconjugated beads (Beads) or ZX29-conjugated beads (Beads-ZX29). In the competition group, WT BRD4 was pre-incubated with a molar excess of free ZX29 competitor prior to the addition of Beads-ZX29. The data are presented as mean ± SEM, n = 3.

3.4. ZX29 exerts anti-apoptotic effects and disrupts BRD4 chromatin-tethering in cardiomyocytes

BRD4 is a key epigenetic driver of cardiovascular pathogenesis, but the functional role of BRD4 in MI/RI, and the therapeutic potential of BD2-selective inhibition, remain incompletely defined. We first assessed BRD4 expression in human myocardial tissues and found that BRD4 protein levels were significantly upregulated in failing hearts from post-MI reperfusion patients compared with healthy donor hearts (Fig. 4A and C). In healthy myocardium, BRD4 showed weak, diffuse nuclear localization with sparse foci, whereas failing hearts exhibited robust BRD4 upregulation and enhanced nuclear foci formation (Fig. 4B). The baseline clinical characteristics of the enrolled subjects are detailed in Supporting Information Table S4. Given the dose-limiting toxicities of pan-BRD4 inhibitors, we focused on developing a BD2-selective BRD4 inhibitor, with the clinical agent RVX-208 as a positive control.

Figure 4.

ZX29 reduces cardiomyocyte death by inhibiting ferroptosis. (A) BRD4 immunodensity was assessed in the left ventricle of normal individuals and patients with HF (15 fields per coverslip, n = 4, scale bar = 200 μm). (B) Fluorescence detection of BRD4 in sections of the left ventricle from normal individuals and patients with HF (15 fields per coverslip, n = 4, scale bar = 20 μm). (C) The protein levels of BRD4 in the left ventricle from normal individuals and patients with HF. (D) Cell death was measured by TUNEL assays in NMCMs treated with ZX29 (low dose: 0.5 μmol/L; high dose: 1 μmol/L) and RVX-208 (1 μmol/L) followed by H/R treatment. TUNEL (green) represents cell death (scale bar = 50 μm). (E, G) Flow-cytometric analysis was used to determine the percentage of apoptotic cells (n = 3). (F) Cell viability detected by the CCK-8 assay at different treatment for NMCMs (n = 3) (H, I) Immunohistofluorescence staining using antibodies against BRD4 (green), merged images displaying nuclei are also presented (15 cells/group, n = 3, scale bar = 20 μm). (J) Protein levels of BRD4 in NMCMs treated with ZX29 (low dose: 0.5 μmol/L; high dose: 1 μmol/L) and RVX-208 (1 μmol/L) followed by hypoxia/reoxygenation (H/R) treatment. (K) NMCMs were stained with anti-BRD4 and anti-H3K27ac. In situ PLA was performed (scale bar = 5 μm). (L) The mRNA levels of Brd4 in cells were measured by qRT-PCR (n = 4). (M) The mRNA levels of Rela in NMCMs treated with ZX29 and RVX-208 were measured by qRT-PCR (n = 4). (N) Cell death was measured by TUNEL assays in NMCMs transfected with BRD4-Flag (scale bar = 50 μm). (O) Cell viability was detected by the CCK-8 assay at different treatments for NMCMs (n = 4). (P) The protein levels of BRD4, BAX, and Bcl2 in the ZX29-treated NMCMs were detected by Western blot (n = 3). (Q) The mRNA levels of Rela in NMCMs transfected with BRD4-Flag were measured by qRT-PCR (n = 4). (R) Cell death was measured by TUNEL assays in BRD4-Flag and BRD4-H437E/V439K treated NMCMs (scale bar = 50 μm). The data are presented as mean ± SEM.

We next evaluated the anti-apoptotic effects of ZX29 in H/R-injured NMCMs. TUNEL staining, CCK-8 assays, and flow cytometry consistently showed that ZX29 dose-dependently suppressed H/R-induced cardiomyocyte apoptosis, with 1 μmol/L ZX29 exhibiting greater potency than an equivalent dose of RVX-208 (Fig. 4D–G). To dissect the underlying mechanism, we first assessed the effects of ZX29 on BRD4 expression and activity. Neither ZX29 nor RVX-208 altered total BRD4 protein or Brd4 mRNA levels (Fig. 4J and L, Supporting Information Figs. S5 and S6), indicating these inhibitors do not regulate BRD4 expression. However, ZX29 significantly reduced the number of BRD4 nuclear foci—a marker of chromatin-tethered transcriptional activity—in H/R-exposed NMCMs (Fig. 4H and I), suggesting ZX29 disrupts BRD4's chromatin binding rather than its expression.

In support of this, PLA assays revealed that ZX29 dose-dependently inhibited the interaction between BRD4 and the active chromatin mark H3K27ac (Fig. 4K), and suppressed transcription of the BRD4 target gene Rela (Fig. 4M). We then validated the on-target specificity of ZX29 via rescue experiments. Overexpression of wild-type BRD4 completely reversed the anti-apoptotic effects of ZX29 and restored Rela transcription (Fig. 4N–Q). Furthermore, mutation of the ZX29-binding BD2 domain of BRD4 fully abrogated the anti-apoptotic activity of ZX29, whereas mutation of the BD1 domain had no effect (Fig. 4R). Taken together, these data establish that ZX29 exerts its anti-apoptotic effects via selective, on-target inhibition of the BRD4 BD2 domain, which disrupts BRD4's chromatin-tethered transcriptional regulatory function in cardiomyocytes.

3.5. ZX29 attenuates MI/RI and improves myocardial function in vivo

Dose-escalation studies established 20 mg/kg/day as the optimal cardioprotective dose, providing a wide preclinical safety margin. The experimental timeline for the MI/RI model and prophylactic dosing regimen is depicted in Fig. 5A. ZX29 dose-dependently reduced myocardial infarct size (Fig. 5B), with the infarct area/left ventricular (IA/LV) ratio significantly decreased; notably, high-dose ZX29 exhibited superior efficacy over the clinical BD2 inhibitor RVX-208 at the equivalent dose (Fig. 5C). Echocardiographic assessment demonstrated that ZX29 preserved both systolic and diastolic functions, as reflected by increased left ventricular ejection fraction (LVEF) and fractional shortening (LVFS), reduced LV end-systolic and end-diastolic diameters (LVESD and LVEDD) (Fig. 5F–J), and a restored mitral E/A ratio (Fig. 5R).

Figure 5.

ZX29 significantly alleviates cardiac injury caused by MI/RI. (A) Schematic illustration of ZX29 and RVX-208 administration in mice. ZX29 and RVX-208 were dissolved in a solvent consisting of 10% dimethyl sulfoxide (DMSO) and 90% corn oil, and administered via oral gavage at the indicated dose and frequency. C57BL/6J mice were given ZX29 (low dose: 10 mg/kg/day; high dose: 20 mg/kg/day) and RVX-208 (20 mg/kg/day) then subjected to 45 min occlusion of the left anterior descending coronary artery followed by 24 h reperfusion. (B, C) Myocardial infarct size in mice subjected to cardiac ischemia and 24 h reperfusion (n = 3). Representative TTC staining results are shown. (D) Myocardial caspase-3 activity in mice was measured. (E) Quantitative analysis of serum MDA (n = 6). (F–J) Echocardiographic measurements of LVEF, LVFS, LVESD, and LVEDD were performed (n = 8). (K–N) Serum levels of NT-proBNP, CK-MB, LDH and cTnT in experimental animals (n = 6). (O) Representative images of TUNEL-stained cardiac tissue and quantification of TUNEL-positive cells (scale bar = 50 μm). (P) The expression of BRD4, BAX and Bcl2 in heart was detected and quantified (n = 4). (Q) Representative images of immunohistochemistry staining for TNF-α, IL-1β, and CD68 in mice. (R) Representative echocardiography images for assessment of diastolic function using PWD in mice. The data are presented as mean ± SEM.

At the biochemical level, ZX29 dose-dependently lowered myocardial caspase-3 activity, serum malondialdehyde (MDA), and the circulating cardiac injury markers cTnT, NT-proBNP, CK-MB, and LDH (Fig. 5D, E, and K–N). TUNEL assays confirmed a marked reduction in apoptotic cardiomyocytes in the infarct border zone (Fig. 5O), which was consistent with an increased Bcl-2/Bax ratio, while total BRD4 protein levels remained unaltered (Fig. 5P, Supporting Information Fig. S7 and S13). Mechanistically, ZX29 did not alter total BRD4 protein levels (Fig. S7B and Fig. S13) but reduced the number of nuclear BRD4 foci and suppressed Rela transcription (Supporting Information Figs. S8–S10), leading to a pronounced anti-inflammatory phenotype. Immunohistochemistry further confirmed attenuation of MI/RI-induced elevation of TNF-α, IL-1β, and IL-6, alongside reduced infiltration of CD68⁺ macrophages (Fig. 5Q andSupporting Information Figs. S11–S13).

Notably, in a therapeutic regimen initiated at the onset of reperfusion, ZX29 consistently reduced infarct size and preserved cardiac function, outperforming the pan-BET inhibitor JQ1 in both parameters (Supporting Information Fig. S14). Importantly, unlike RVX-208, ZX29 did not affect serum lipid levels, indicating that its cardioprotective effects are independent of systemic lipid modulation (Supporting Information Fig. S15). The obligatory role of BRD4 was confirmed via AAV9-cTnT-shBRD4-mediated cardiomyocyte-specific knockdown, which completely abolished the protective actions of ZX29 (Supporting Information Fig. S16). Safety profiling revealed an excellent tolerability profile. No significant hepatotoxicity, nephrotoxicity, or hematological toxicity was detected at therapeutic doses. Critically, unlike the pan-BET inhibitor JQ1, which caused severe thrombocytopenia, ZX29 fully preserved platelet counts alongside other hematological parameters (Supporting Information Figs. S17–S19). Moreover, ZX29 mitigated MI/RI-induced leukocytosis, further corroborating its anti-inflammatory activity.

3.6. Effect of ZX29 on pathological myocardial remodeling induced by MI/RI

MI/RI-induced cardiac remodeling is a critical pathological process, as adverse remodeling drives the progression to HF. To determine whether ZX29 ameliorates myocardial remodeling via targeting and inhibiting BRD4, we evaluated fibrosis and hypertrophy indices in mouse myocardium at 2 weeks post-I/R. Masson's trichrome staining showed that both ZX29 and RVX-208 significantly reduced collagen deposition (Fig. 6B). To biochemically corroborate these histological findings, we quantified myocardial hydroxyproline content, a gold-standard marker for tissue fibrosis. Consistent with the morphological observations, ZX29 dose-dependently attenuated the MI/RI-induced accumulation of hydroxyproline in the infarcted hearts (Supporting Information Fig. S20). In contrast, wheat germ agglutinin (WGA) staining revealed that neither ZX29 nor RVX-208 improved pathological myocardial hypertrophy (Fig. 6A and C). Compared with sham-operated mice, MI/R mice exhibited significantly upregulated expression of Col3a1, Ctgf, α-Sma, and fibronectin; ZX29 attenuated these changes in a dose-dependent manner (Fig. 6D–F and I). Additionally, ZX29 administration downregulated the mRNA expression of Tgfb1 and its downstream effector Smad3-suggesting ZX29-mediated BRD4 inhibition restricts TGF-β1 signaling transduction (Fig. 6G and H). Western blot analysis confirmed ZX29 had no effect on the levels of myocardial hypertrophy markers (β-MHC, ANF) in myocardial tissue (Fig. 6J). Given that cardiac fibroblast migration and invasion contribute to post-MI/R myocardial fibrosis, we established a cardiac fibrosis model using Ang II. CCK-8, scratch, and Transwell assays demonstrated ZX29 dose-dependently inhibited the proliferation, migration, and invasion of neonatal mouse cardiac fibroblasts (NMCFs) (Fig. 6K–N and Supporting Information Fig. S21). Western blot analysis further showed ZX29 dose-dependently suppressed Ang II-induced upregulation of the fibrosis markers fibronectin and α-SMA in NMCFs (Supporting Information Fig. S22), while having no impact on the expression of hypertrophy markers (Supporting Information Fig. S23). Consistently, immunofluorescence staining revealed BRD4 overexpression reversed ZX29-mediated inhibition of fibronectin expression in NMCFs (Fig. 6O and P). Rhodamine–Phalloidin staining of F-actin revealed that Ang II induced a significant increase in NMCM surface area, whereas neither ZX29 nor RVX-208 reduced this enlargement (Fig. 6Q and R), corroborating the in vivo findings that ZX29 does not attenuate pathological cardiomyocyte hypertrophy.

Figure 6.

ZX29 alleviates cardiac fibrosis without affecting cardiac hypertrophy. (A) Representative whole-heart cross-sections and corresponding magnified views of the infarct border zones stained with hematoxylin and eosin (scale bar = 200 μm) and Masson's trichrome at 2 weeks post-MI/RI. Heart sections stained with Masson's trichrome to detect fibrosis (scale bar = 200 μm). WGA staining was examined (scale bar = 100 μm). (B, C) Quantification of cardiac fibrosis and myocyte cross-sectional area in different groups of mice (n = 6) (I) The protein levels of Fibronectin and SMA in the cardiac tissue of ZX29 treated mice were detected by Western blot (n = 4). (D–H) The mRNA levels of Col3a1, Ctgf, Sma, Tgfb1 and Smad3 in ZX29 treated cardiac tissue were measured by qPCR (n = 4). (E) The protein levels of Fibronectin and SMA in the si-ATAD3A treated NMCFs were detected by Western blot (n = 3). (F) WGA staining (scale bar = 50 μm). (J) The protein levels of ANF and β-MHC in the cardiac tissue of ZX29 treated mice were detected by Western blot (n = 4). (K, L) Representative images of NMCFs wound healing at 0 and 24 h. Quantitative analysis of migration rate in each group (n = 4). (M, N) Representative images of Transwell assay showing invaded cells stained with crystal violet (scale bar = 100 μm; n = 4). (O, P) Neonatal mouse cardiac fibroblasts (NMCFs) were immunostained for fibronectin (scale bar = 20 μm; n = 4). (Q, R) The surface area of NMCMs was measured (scale bar = 20 μm; n = 4). Data are represented as mean ± SEM.

3.7. ZX29 inhibits MI/RI-induced cardiac ferroptosis

While the in vivo phenotypic benefits of ZX29 were evident, the downstream molecular executioners driving this cardioprotection remained to be defined. To delineate the molecular mechanism of ZX29-mediated cardioprotection, we performed RNA sequencing on myocardial tissues from Sham, MI/RI, and MI/RI + ZX29 (20 mg/kg) mice (Fig. 7A). Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis of differentially expressed genes (DEGs) identified the ferroptosis pathway as the most significantly enriched pathway modulated by ZX29, with treatment markedly reversing MI/RI-induced dysregulation of ferroptosis-related genes (Fig. 7B). Gene set enrichment analysis (GSEA) further validated that MI/RI induced a robust pro-ferroptotic gene signature, which was significantly blunted by ZX29, pointing to ferroptosis inhibition as a central mediator of ZX29's protective effects (Fig. 7C).

Figure 7.

ZX29 ameliorates MI/RI-induced ferroptosis in heart tissue. (A) RNA-seq analysis reveals the DEGs in murine hearts from three groups: sham (n = 3), MI/RI (n = 3), and MI/RI + ZX29 (20 mg/kg, i. g.; n = 3), (B) Heat map representing the ferroptosis-related genes detected in RNA-seq analysis of murine heart tissue. Gene expressions were normalized with row Z-score. (C) Gene set enrichment analysis (GSEA) of modulated genes in ZX29 (20 mg/kg)-treated versus MI/RI murine hearts. (D) Cardiac MDA was measured in murine hearts (n = 4). (E–G) The mRNA levels of Ptgs2, Ncoa4 and Slc7a11 in ZX29 treated cardiac tissue were measured by qPCR (n = 6). (H–J) The protein levels of NCOA4 and FTH1 in heart tissue were detected by Western blot (n = 4). (K–N) Representative images and quantification of cofluorescent immunohistochemistry staining for cTnT with GPX4, ACSL4, and FTH1 in mice (n = 6). (O, P) Representative images and % area of Prussian blue staining with DAB enhancement (scale bar = 100 μm) (O, Q) Representative images and quantification of fluorescent immunohistochemistry staining for DHE (scale bar = 20 μm). (R) Cardiac GSH levels were measured (n = 6). Statistical significance was calculated using one-way ANOVA with Tukey's multiple comparisons test. Data are represented as mean ± SEM.

We next validated these transcriptomic findings at the molecular and functional levels. ZX29 dose-dependently corrected the MI/RI-induced dysregulation of canonical ferroptosis marker genes (Ptgs2, Slc7a11, and Ncoa4) in the injured myocardium (Fig. 7E–G). Consistent with these transcriptional changes, immunofluorescence and Western blot revealed that MI/RI upregulated the pro-ferroptotic protein acyl-CoA synthetase long-chain family member 4 (ACSL4) and downregulated the ferroptosis suppressors GPX4 and FTH1, effects that were fully reversed by ZX29 in a dose-dependent manner (Fig. 7H–N). Functionally, biochemical quantification showed that ZX29 dose-dependently mitigated MI/RI-induced myocardial lipid peroxidation, as evidenced by significantly reduced MDA content (Fig. 7D). Consistently, dihydroethidium (DHE) and Prussian blue staining revealed that ZX29 dose-dependently abrogated MI/RI-induced elevations in myocardial reactive oxygen species (ROS) and total iron levels (Fig. 7O–Q). In addition, ZX29 significantly restored the depleted myocardial reduced glutathione (GSH) content caused by MI/RI injury (Fig. 7R). Collectively, these data demonstrate that ZX29 inhibits cardiac ferroptosis in the murine MI/RI model.

3.8. ZX29 inhibits the NCOA4–FTH1 pathway via BRD4 to attenuate cardiomyocyte ferroptosis

NCOA4-mediated ferritinophagy and subsequent FTH1 degradation are core drivers of ferroptosis. Given our in vivo finding that ZX29-induced BRD4 inhibition inversely regulates NCOA4 and FTH1 expression, we first defined the functional link between BRD4 and the NCOA4–FTH1 axis in cardiomyocyte ferroptosis.

We examined the dynamic expression of BRD4, NCOA4, and FTH1 in NMCMs over the time course of H/R injury. Western blot revealed that BRD4 and NCOA4 were upregulated in a synchronized time-dependent manner, while FTH1 was progressively downregulated, pointing to a potential regulatory role of BRD4 in the NCOA4–FTH1 axis (Fig. 8A). Gain- and loss-of-function assays further validated that BRD4 is a positive regulator of cardiomyocyte ferroptosis: the ferroptosis inhibitor Fer-1 partially reversed BRD4 overexpression-induced cardiomyocyte apoptosis, while BRD4 knockdown significantly enhanced cell viability and reduced LDH release (Fig. 8B–D).

Figure 8.

ZX29's suppressive effect on the NCOA4–FTH1 pathway. (A) Western blot analysis of BRD4, NCOA4, and FTH1 protein levels in NMCMs at different time points (0, 3, 6, 12, and 24 h) after ischemia. Quantitative analysis of relative protein levels normalized to GAPDH (n = 3). (B, C) Cell viability assessed by CCK-8 assay in NMCMs under indicated conditions (EV, BRD4-Flag, siNC, siBRD4, with/without H/R and Fer-1 treatment) (n = 4). (D) LDH content in murine hearts was measured (n = 4). (E) The protein levels of GPX4, NCOA4 and FTH1 in ZX29 treated NMCMs were measured by Western blot (n = 4). (F) Flow cytometry analysis of lipid reactive oxygen species in NMCMs using C11-BODIPY staining under indicated conditions (DMSO, H/R, H/R + ZX29 at different doses; n = 3). (G) qPCR analysis of Ptgs2 mRNA level (n = 3). (H, I) Biochemical assays for MDA and GSH levels in NMCMs under indicated conditions (DMSO, H/R, H/R + ZX29). (J, K, Q) Representative fluorescent images and quantitative analysis of cytoplasmic Fe²⁺ stained with FerroOrange (scale bar = 20 μm). (N, L) Representative fluorescent images and quantitative analysis of ROS stained with DHE (scale bar = 20 μm). (M) Quantification of cellular count in NMCMs with indicated treatments (n = 4). (O) Flow cytometry analysis of lipid reactive oxygen species in cells using C11-BODIPY staining under indicated conditions (DMSO, H/R, H/R + ZX29 at different doses) (n = 3). (P) Quantitative analysis of MDA activity in NMCMs (n = 6). (R) The protein levels of NCOA4 and FTH1 in ZX29 treated NMCMs were measured by Western blot (n = 4). (S) The GSH/GSSG ratio was detected in NMCMs (n = 4). (T) Quantitative analysis of cytoplasmic Fe²⁺ stained with FerroOrange (n = 4). Data are represented as mean ± SEM.

We next assessed the effect of ZX29 on ferroptosis in H/R-injured NMCMs. Consistent with our in vivo findings, ZX29 dose-dependently corrected H/R-induced dysregulation of canonical ferroptosis markers, including normalized Ptgs2 expression, restored levels of the ferroptosis suppressors FTH1 and GPX4, and downregulated the ferritinophagy mediator NCOA4 (Fig. 8E and G). ZX29 also mitigated H/R-induced intracellular Fe²⁺ overload, ROS production, lipid peroxidation, and MDA accumulation, with quantitative cell counting confirming a dose-dependent decrease in ferroptotic cell number (Fig. 8F and H–M), while replenishing depleted intracellular GSH levels (Fig. 8N). These data confirmed that ZX29 protects cardiomyocytes from H/R-induced ferroptosis in vitro.

We validated that the anti-ferroptotic effect of ZX29 is strictly dependent on BRD4 inhibition. Overexpression of Brd4 in NMCMs completely abrogated the protective effects of ZX29 on ferroptosis, and reversed ZX29-mediated suppression of the NCOA4–FTH1 pathway (Fig. 8O–T). Collectively, these results establish that ZX29 attenuates cardiomyocyte ferroptosis via BRD4-dependent inhibition of the NCOA4–FTH1 axis.

3.9. ZX29 alleviates MI/RI-induced ferroptosis by inhibiting BRD4–YY1 interaction

We performed fluorescence recovery after photobleaching (FRAP) assays to assess the effect of ZX29 on BRD4 chromatin-binding dynamics in NMCMs. ZX29 dose-dependently accelerated the fluorescence recovery of BRD4-GFP after photobleaching, indicating that ZX29 promotes BRD4 dissociation from chromatin and impairs its recruitment of downstream transcriptional regulators (Fig. 9A). Given that ZX29 accelerates BRD4 dissociation from chromatin, we hypothesized that it disrupts specific BRD4-transcription factor complexes. Indeed, our unbiased IP–MS interactome screening revealed a profound remodeling of the BRD4 interactome upon ZX29 treatment. Among the identified BRD4-interacting proteins, YY1 was prioritized for further investigation based on a dual rationale. First, the BRD4–YY1 interaction exhibited the most profound sensitivity to ZX29, showing the most dramatic dissociation upon treatment (Supporting Information Fig. S24). Second, bioinformatics analysis via the JASPAR database predicted that YY1 uniquely possesses high-affinity binding sites within the promoter region of Ncoa4, our core ferroptotic target (Fig. 9J)—a direct regulatory link absent in other top candidate interactors. Thus, we hypothesized that ZX29 disrupts the BRD4–YY1 axis to regulate Ncoa4 transcription. Alongside YY1, we also detected ZX29-induced reductions in BRD4 binding to other known transcriptional regulators including P-TEFb, c-Myc, E2F1, and SMAD3, which were not further pursued given the lack of a direct link to ferroptosis regulation. Analysis of public single-cell RNA sequencing (scRNA-seq) datasets of human myocardium showed that BRD4 and YY1 share nearly identical cellular distribution patterns, with predominant co-expression in cardiomyocytes and cardiac fibroblasts (Fig. 9B). Correlation analysis using the GEPIA 2.0 database further confirmed a strong positive correlation between BRD4 and YY1 expression in human left ventricular samples (Fig. 9C).

Figure 9.

ZX29 suppresses the BRD4–YY1 interaction in myocardial ischemia–reperfusion. (A) FRAP analysis of BRD4-GFP in NMCMs under indicated conditions. Representative images show fluorescence recovery at 0 s (pre-bleach), immediately after bleaching, and 120 s post-bleach (scale bar = 10 μm; n = 4). (B) Public single-cell RNA-Seq datasets show the expression of BRD4 and YY1 by cell type in human heart tissue via single-cell RNA sequencing. (C) Correlation analysis of BRD4 and YY1 expression in human heart tissue via GEPIA2.0. The scatter plot shows a positive correlation (R = 0.85, P = 1.5e-57). (D) Representative images and quantitative analysis of in situ PLA detecting the BRD4–YY1 interaction in myocardial sections from mice subjected to MI/RI with the indicated treatments. The white rectangular boxes denote the regions displayed at higher magnification in the bottom row. Data are presented as mean ± SEM (n = 4). (E) Co-IP analysis evaluating the endogenous interaction between BRD4 and YY1 in clinical samples of healthy and failing human hearts. Data are presented as mean ± SEM (n = 3). (F) Representative fluorescence images of YY1 staining with BRD4 (scale bar = 20 μm; n = 3). (G) IP analysis of BRD4 and YY1 interaction in myocardial tissues. Left, representative Western blot bands; right, quantitative analysis of relative band density (n = 3). (H) Identification of the interaction between YY1 and BRD4 by GST pull-down. GST alone and GST fusion proteins were affinity-purified before pull-down assay. (I) NCOA4 immunodensity was assessed in the left ventricle of normal individuals and patients with HF (15 fields per coverslip, n = 4, scale bar = 200 μm). (J) The JASPAR website predicted YY1 to have binding sites with the NCOA4 promoter sequence. ChIP assay assessing the binding between YY1 and NCOA4 (n = 3). (K) ChIP–qPCR analysis validating the physical binding of YY1 to the predicted sites within the NCOA4 promoter in AC16. Normal rabbit IgG was utilized as a negative isotype control (n = 3). (L) Dual-luciferase reporter assay in cells co-transfected with wild-type (NCOA4-WT-Luc) or mutant (NCOA4-MUT-Luc) promoter reporter plasmids alongside siNC or siYY1 (n = 3). (M) Western blot analysis of YY1, NCOA4, and BRD4 protein levels in NMCMs treated with ZX29 (n = 3). (N) Flow cytometry analysis of lipid reactive oxygen species in cells using C11-BODIPY staining under indicated conditions (n = 3). (O) Left, IP assay detecting FTH1 interaction with NCOA4 under indicated conditions; right, Quantitative analysis of FTH1 band density normalized to IgG. (P) Representative fluorescent images and quantitative analysis of cytoplasmic Fe²⁺ stained with FerroOrange (scale bar = 20 μm; n = 4). (Q) Quantification of cellular count in NMCMs with indicated treatments (n = 4). Data are represented as mean ± SEM.

We next validated the BRD4–YY1 interaction and its modulation by ZX29 both in vitro and in vivo. In situ proximity ligation assay (PLA) on mouse myocardial sections showed that MI/RI markedly enhanced the BRD4–YY1 interaction, which was dose-dependently inhibited by ZX29 (Fig. 9D). Co-IP and immunofluorescence co-staining confirmed that the BRD4–YY1 interaction was significantly upregulated in failing human and mouse hearts, and that ZX29 inhibited this interaction via direct binding to the BRD4 BD2 domain—an effect reversed by BRD4 overexpression (Fig. 9E–G, Supporting Information Figs. S25 and S26). Direct physical binding between BRD4 and YY1 was further confirmed by computational docking (Z-DOCK 3.0.2; Supporting Information Fig. S27) and in vitro GST pull-down assays (Fig. 9H). In line with these regulatory findings, immunohistochemical quantification demonstrated that MI/RI-induced upregulation of NCOA4 protein in the myocardium was dose-dependently attenuated by ZX29 treatment (Fig. 9I).

To delineate the functional link between the BRD4–YY1 interaction and ferroptosis, we investigated whether Yy1 transcriptionally regulates the key ferritinophagy mediator Ncoa4. ChIP–qPCR confirmed that Yy1 specifically and exclusively bound to the first predicted core motif (TCCATC) in the Ncoa4 promoter (Fig. 9K). Dual-luciferase reporter assays further validated that Yy1 directly drives Ncoa4 transcription via this motif: Yy1 knockdown significantly suppressed wild-type Ncoa4 promoter activity, while mutation of the core binding motif completely abolished promoter activity (Fig. 9L). Consistently, Yy1 overexpression or knockdown bidirectionally modulated Ncoa4 mRNA and protein levels in NMCMs (Supporting Information Fig. S28). Critically, ZX29 inhibited H/R-induced Ncoa4 upregulation in cells expressing wild-type BRD4 or a BRD4-BD1 mutant, but this effect was completely abrogated by mutation of the ZX29-binding BD2 domain (H437E/V439K) (Fig. 9M).

We confirmed that ZX29 inhibits ferritinophagy and ferroptosis via the BRD4–YY1–NCOA4 axis. Co-IP assays showed that ZX29 disrupted the NCOA4–FTH1 interaction in the ischemic myocardium (Supporting Information Fig. S29). Consistently, Co-IP results in H/R-challenged NMCMs further verified that ZX29 dose-dependently weakened the intracellular NCOA4–FTH1 protein interaction (Fig. 9O), and confocal microscopy revealed that ZX29 reduced H/R-induced lysosomal Fe²⁺ accumulation and FTH1–lysosome colocalization in NMCMs (Supporting Information Figs. S30 and S31). Rescue experiments further confirmed that Ncoa4 overexpression completely reversed the anti-ferroptotic effects of ZX29: it abrogated ZX29-mediated suppression of lipid ROS production (Fig. 9N), intracellular Fe²⁺ overload (Fig. 9P), and ferroptotic cell death (Fig. 9Q), as well as its effects on Ptgs2 expression, GSH/GSSG ratio, and lipid peroxidation (Supporting Information Fig. S32). Taken together, these data demonstrate that ZX29 disrupts the BRD4–YY1 interaction to suppress NCOA4-mediated ferritinophagy, thereby inhibiting cardiomyocyte ferroptosis during MI/RI.

3.10. ZX29 preserves mitochondrial structural and functional homeostasis via BRD4 inhibition

Having established that ZX29 disrupts the BRD4–YY1–NCOA4 axis to suppress ferritinophagy, we next investigated the downstream subcellular consequences. Because mitochondria are the primary execution sites of ferroptosis—where iron overload drives catastrophic lipid peroxidation and metabolic collapse—we hypothesized that the ultimate cardioprotective efficacy of ZX29 relies on preserving mitochondrial homeostasis. We first evaluated mitochondrial dynamics and bioenergetics in vitro. In H/R-treated NMCMs, ZX29 dose-dependently restrained mitochondrial iron accumulation and localized lipid peroxidation (Fig. 10B and C). Consequently, ZX29 prevented H/R-induced mitochondrial fragmentation, maintaining a highly interconnected mitochondrial network (Fig. 10A). This structural protection directly translated to bioenergetic rescue; Seahorse extracellular flux analysis revealed that ZX29 significantly restored maximal respiratory capacity and ATP production in injured cardiomyocytes (Fig. 10D). Consistent with the suppression of ferroptotic damage, ZX29 also reversed the H/R-induced depletion of mitochondrial DNA (mtDNA) copy number, stabilized the mitochondrial membrane potential (MMP), and quenched mitochondrial ROS (mtROS) generation (Fig. 10E and G–I).

Figure 10.

ZX29 restores cardiac mitochondrial homeostasis via inhibiting BRD4. (A) Representative images of mitochondria of NMCMs (scale bar = 10 μm). Quantification of fragmented mitochondria for the indicated conditions (n = 4). (B, C) Representative fluorescence images of mitochondrial LPs using MitoPeDPP and quantitative analysis of fluorescence intensity to determine levels of LPs in NMCMs (green, MitoPeDPP; red, MitoBright LT Deep Red; scale bar = 20 μm; n = 4). (D) Seahorse assay for the OCR in HL-1 cells. Cells were treated with ZX29. Calculations of ATP production, maximal respiration are shown. (E) mtDNA replication was assessed by measuring the extension of D-Loop (n = 4). (F) Representative TEM images of mitochondrial ultrastructure in murine myocardial tissues from the indicated groups (scale bar = 1.0 μm; n = 3). (G–I) Representative fluorescence images and the quantitative analysis of JC-1 staining with mtSOX Deep Red staining (scale bar = 20 μm; n = 4). (J) Mitochondrial structure of the heart assayed after 24 h reperfusion (scale bar = 2 μm; n = 3). (K) Electron confocal microscopy of mitochondria in NMCMs with BRD4-Flag and ZX29 (scale bar = 5 μm; n = 3). (L) Recordings of mitochondrial membrane potential, measured by TMRE (n = 3). (M) Recordings of mitochondrial membrane structure integrity, measured by Nonyl Acridine Orange (NAO) (n = 3). Data are represented as mean ± SEM.

To provide definitive in vivo evidence linking our anti-ferroptotic mechanism to organelle preservation, we evaluated the microarchitecture of the murine ischemic myocardium using transmission electron microscopy (TEM). In the vehicle-treated MI/RI model, cardiomyocytes exhibited canonical morphological hallmarks of ferroptosis, characterized by profound mitochondrial swelling, extensive outer membrane rupture, and the severe dissolution of cristae. Strikingly, ZX29 treatment conferred a robust, dose-dependent protection against this ultrastructural collapse. As directly visualized in Fig. 10F, high-dose ZX29 effectively reversed the ischemia-induced architectural damage, maintaining intact outer mitochondrial membranes and well-organized, dense cristae arrays that closely resembled the healthy myocardium. Quantitative analysis of mitochondrial injury scores further confirmed the dose-dependent protective effect of ZX29 against mitochondrial ultrastructural damage in ischemic cardiac tissue (Fig. 10J).

Ultimately, we confirmed that these profound mitochondrial protective effects are strictly dependent on on-target BRD4 inhibition. Overexpression of BRD4 in NMCMs completely abrogated the ability of ZX29 to preserve the MMP, suppress mtROS production, and maintain mitochondrial network integrity (Fig. 10K). Furthermore, flow cytometry with TMRE and nonyl acridine orange (NAO) staining verified that ZX29-mediated rescue of MMP loss and cardiolipin peroxidation was entirely reversed in a BRD4-dependent manner (Fig. 10L and M).

4. Discussion

IHD remains one of the leading causes of morbidity and mortality globally, imposing a substantial burden on public health systems³⁸. Reperfusion therapies such as PCI are the gold standard for restoring blood flow to ischemic myocardium in MI patients³⁹. While drugs like the β1-adrenergic receptor antagonist metoprolol have shown promise in reducing infarct size when administered pre-reperfusion, supporting evidence primarily stems from small-scale, single-center trials with limited generalizability⁴⁰. To address this gap, we herein report the development of ZX29—a novel coumarin derivative with selective binding to the BD2 of BRD4—and demonstrate its dose-dependent protective effects against MI/RI-induced cardiac injury.

4.1. Insights into BRD4 inhibitors

BRD4 plays a pivotal role in cardiac development, cardiomyocyte hypertrophy, myocardial fibrosis, and inflammation. Our single-cell RNA sequencing analysis identified fibroblasts and cardiomyocytes as the major BRD4-expressing cells in the heart. Further, in an Ang II-induced myocardial hypertrophy model, we observed that BRD4 upregulation correlated positively with fibrosis severity—a finding consistent with growing evidence that BRD4 inhibitors alleviate fibrosis and myocardial hypertrophy.

However, pan-BRD4 inhibitors like JQ1 have significant limitations: their short half-lives—with JQ1 showing 0.9 h post-intravenous injection—and off-target toxicities such as bone marrow suppression and gastrointestinal irritation overlap with BD1-selective agents, restricting use in vulnerable populations like elderly heart failure patients and those with anemia^41,42. In contrast, BD2-selective inhibitors offer favorable pharmacokinetics and reduced hepatorenal burden^43,44. For example, the clinical BD2 inhibitor RVX-208 exhibits a prolonged half-life of 12–16 h, improves cardiac function and enhances endothelial function in phase II trials⁴⁵. Aligning with this, our study confirms ZX29's safety via pathological analysis of liver/kidney tissues and demonstrates its dose-dependent inhibition of CD68 signaling in mouse hearts, validating ZX29's anti-inflammatory efficacy in MI/RI.

4.2. ZX29 mitigates myocardial fibrosis

Myocardial fibrosis, a hallmark of progressive cardiac dysfunction, is classified into reactive (hypertension/pressure overload-induced) and replacement (post-MI) subtypes⁴⁶. Our data show ZX29 inhibits both subtypes, with superior efficacy to RVX-208. Notably, neither ZX29 nor RVX-208 affects hypertrophy progression—likely due to their shared binding to BRD4 BD2, which may not overlap with the BD1-dependent pathways driving hypertrophy. Molecular docking further confirms ZX29's specificity: it forms extensive interactions with the hydrophobic pocket of BD2, and mutating key hydrophobic residues (H437/V439) to hydrophilic variants abrogates ZX29's regulation of BRD4 downstream pathways and its anti-apoptotic effects in cardiomyocytes. This underscores BD2 as a critical target for fibrosis, while highlighting the need for subtype-specific BRD4 inhibition to avoid unintended effects.

4.3. BRD4 BD2 inhibition alleviates MI/RI by suppressing ferroptosis

Ferroptosis—an iron-dependent, non-apoptotic cell death modality—is tightly linked to MI/RI pathogenesis^28,47. Under physiological conditions, the antioxidant system maintains redox homeostasis, thereby preventing ferroptosis; however, MI/RI disrupts this balance. Ischemia increases iron release and impairs iron transport, whereas reperfusion triggers ROS bursts to exacerbate lipid peroxidation, ultimately inducing ferroptosis in cardiomyocytes via STING-mediated autophagic degradation of GPX4⁴⁸.

Emerging evidence has implicated BRD4 in ferroptosis regulation, yet its function shows striking tissue specificity: in most cancers, BRD4 suppresses ferroptosis by modulating FSP1 and SCD1 signaling, while BRD4 silencing alleviates neuronal ferroptosis in cerebral hemorrhage by repressing ATF3 expression49, 50, 51. Notably, investigations into BRD4's role in myocardial ferroptosis remain scarce, and our study aims to address this crucial knowledge gap. Here, we show that BRD4 BD2 inhibition mediated by ZX29 modulates ferroptosis-associated proteins: it reduces the expression of ACSL4 (a pivotal mediator of exogenous ferroptosis) and elevates levels of the ferroptosis suppressors GPX4 and FTH1. Concomitantly, ZX29 mitigates MI/RI-induced increases in ROS, MDA, and iron ions in both myocardial tissue and circulation.

NCOA4-dependent ferritinophagy is a key mediator of endogenous ferroptosis⁴⁷. Under physiological conditions, NCOA4 mobilizes stored iron to maintain iron homeostasis, but its pathological overexpression enhances ferritinophagy, promoting labile Fe²⁺ release, accelerating Fenton reactions, and driving lipid peroxidation⁵². In failing human hearts, we observed NCOA4 upregulation; in MI/RI models, BRD4 and NCOA4 exhibited progressive upregulation post-reperfusion with concurrent FTH1 downregulation. Our data confirm that ZX29 suppresses this axis. RNA-seq analysis showed that ZX29 downregulated NCOA4, while in vivo and in vitro experiments demonstrated that ZX29 reverses FTH1/NCOA4 dysregulation and lowers lysosomal Fe²⁺ release—an effect we further validated using Fe²⁺–lysosome colocalization tracking. Critically, NCOA4 overexpression abrogated ZX29's cardioprotective effects, while rescue experiments confirmed ZX29 acts through direct BD2 binding. Together with our observation that ZX29 mediates ACSL4 downregulation in cardiac tissues, these findings identify ZX29 as a dual inhibitor of endogenous (NCOA4–FTH1-dependent) and exogenous (ACSL4-mediated) ferroptosis.

4.4. ZX29 preserves mitochondrial function by sparing BRD4's physiological roles

Mitochondrial function and ferroptosis are tightly linked^53,54. Cardiac tissue relies heavily on mitochondrial metabolism to maintain contractile function, and notably, mitochondria confer ferroptosis resistance via three core mechanisms: fatty acid β-oxidation, GPX4 activation, and lipid peroxidation suppression⁵⁵. Growing evidence shows that mitochondria play multifaceted roles in both the initiation and execution of ferroptosis⁵⁶. Under conditions of mitochondrial impairment or iron overload, increased electron leakage from the electron transport chain (ETC) boosts ROS production, which in turn induces lipid peroxidation⁵⁷. Subsequent exacerbation of ferroptosis triggers distinct ultrastructural alterations in mitochondria⁵⁸. These changes are characterized by mitochondrial fragmentation, enhanced mitochondrial membrane permeability, crumpling of mitochondrial membranes, and a marked reduction in the number and integrity of mitochondrial cristae—all of which severely disrupt mitochondrial homeostasis.

Mitochondrial homeostasis is tightly coupled to iron metabolism^59,60. Moderate intracellular iron accumulation has been shown to upregulate the expression level of PGC-1α, which in turn promotes mitochondrial biogenesis—a key adaptive response that enhances the cell's capacity to counteract oxidative stress⁶¹. In contrast, excessive iron overload disrupts the delicate mitochondrial redox balance, and this equilibrium is frequently perturbed under pathological conditions⁶². A well-characterized example is doxorubicin (DOX)-induced cardiomyopathy: DOX has strong tropism for accumulating in cardiomyocyte mitochondria, raising mitochondrial iron levels markedly and subsequently surging intracellular ROS production⁶³. This mitochondrial iron overload forms a harmful positive feedback loop amplifying ROS generation, ultimately causing severe mitochondrial dysfunction and progressive cardiomyocyte death.

Under stress conditions, NCOA4-mediated ferritinophagy has been identified as a critical regulatory pathway that actively drives mitochondrial iron accumulation⁶⁴. This process, in turn, exacerbates the overproduction of mitochondrial ROS, further compromising mitochondrial function. In the present study, our experimental data demonstrated that treatment with ZX29 effectively blocked the NCOA4–FTH1 signaling pathway. Functionally, this inhibitory effect not only significantly preserved the structural homeostasis of mitochondria but also resulted in a marked enhancement in mitochondrial energy production, as evidenced by improved ATP synthesis efficiency. Given that only mutation of the BD2-binding site abrogated the inhibitory effect of ZX29 on the interaction between NCOA4 and FTH1, we propose that the selective inhibition of BRD4's BD2 domain is the primary mechanism underlying ZX29-mediated mitochondrial protection.

As an epigenetic reader, BRD4 controls the nuclear-encoded mitochondrial transcriptome to regulate cardiac function. Loss of BRD4 in cardiomyocytes leads to significant downregulation of nuclear-encoded mitochondrial genes, impaired activities of mitochondrial ETC and tricarboxylic acid cycle (TCA) enzymes, resulting in insufficient myocardial energy supply and ultimately systolic dysfunction. Genetic disruption of cardiomyocyte BRD4 in the mature adult heart can trigger systolic dysfunction⁶⁵. These findings indicate that global BRD4 inhibition or knockout is not an appropriate strategy for treating CVDs. For instance, the potent BET inhibitory effect of second-generation BET inhibitors (such as IBET-151) causes adverse cardiac side effects, among which mitochondrial disorder and dysfunction are major contributing factors⁶⁶.

In our study, ZX29 exerts multifaceted mitochondrial protection. This may be attributed to ZX29's unique property of not disrupting the BRD4–GATA4 axis and bromodomain-independent BRD4 functions—two elements critical for maintaining physiological mitochondrial homeostasis—while simultaneously abrogating BRD4's pro-ferroptotic effects. These observations validate our hypothesis that selective BD2 inhibition offers superior safety profiles and therapeutic prospects.

4.5. BRD4–YY1 interaction: A novel transcriptional axis mediating myocardial ferroptosis

The mechanisms by which BRD4 regulates gene transcription are complex and diverse, encompassing binding to super-enhancers, direct interaction with transcription factors, involvement in chromatin remodeling, and modulation of transcription elongation⁸. As a key component of super-enhancers, BRD4 relies on its binding to super-enhancer regions to recruit transcriptional regulatory complexes, thereby exerting transcriptional activation effects on target genes. Notably, this binding exhibits cell-type specificity, as BRD4 binds to distinct super-enhancer regions across different cell types, thereby allowing it to orchestrate unique gene expression profiles.

In cardiovascular research, BRD4, as a core epigenetic regulator, has been validated by numerous studies to regulate downstream pathways through direct binding to specific transcription factors. In adult cardiomyocytes, BRD4 forms a bromodomain-independent complex with the cardiac-specific transcription factor GATA4; this complex directly binds to the regulatory regions of nuclear-encoded mitochondrial genes (Ppargc1a and Ppargc1b), synergistically activating pathways related to mitochondrial biogenesis⁶⁷. In cardiac fibroblasts, BRD4 directly interacts with the TEA domain transcription factor TEAD1 and promotes β-catenin nuclear translocation, thereby initiating the expression of fibrotic genes including Col1A1 and α-SMA⁶⁸.

In both cellular and cardiac tissue models, we observed that MI/RI led to a significant increase in the number of nuclear BRD4 foci—a morphological manifestation of BRD4's specific binding to transcription factors. This increase was markedly inhibited by ZX29 treatment. Furthermore, we investigated BRD4's binding affinity to nuclear targets using FRAP assays. After ZX29 treatment, the rate and extent of BRD4-GFP fluorescence recovery were significantly enhanced, directly reflecting an increased proportion of free BRD4. Since BD2 is the critical domain of BRD4 for recognizing and anchoring to chromatin, we hypothesize that ZX29 impairs BRD4's ability to bind to chromatin or transcription factors and regulate downstream gene expression by occupying the BD2 domain.

Through IP–MS screening, we identified YY1 as a BRD4-interacting protein whose binding to BRD4 is disrupted by ZX29. We confirmed that YY1 directly drives Ncoa4 transcription and found that enhanced BRD4–YY1 interaction correlates with elevated NCOA4 expression in failing human hearts, validating the clinical relevance of this axis. This is the first report linking BRD4–YY1 signaling to myocardial ischemia–reperfusion injury, as prior studies of this interaction have been restricted to oncology. Collectively, our findings define the BRD4–YY1–NCOA4 axis as a novel therapeutic target for MI/RI.

A key limitation of this study is the absence of genome-wide multi-omics to map the full scope of ZX29-mediated epigenetic regulation. Epigenetic modulators exert inherent pleiotropic effects on the chromatin landscape, so future work with ATAC-seq and ChIP-seq/CUT&Tag will be critical to fully characterize the global chromatin changes and pan-transcriptomic pharmacological profile of ZX29 in ischemic myocardium.

The translational potential of ZX29 is further supported by its well-defined preclinical dosing regimen and administration schedule. For dose selection, the 20 mg/kg in vivo dose was determined from our preliminary dose-escalation study, in which we found that the cardioprotective efficacy of ZX29 reached a plateau at this dose, with no further enhancement of myocardial salvage or cardiac function improvement observed at higher concentrations. Meanwhile, histological and serum biochemical assessments confirmed no detectable hepatorenal or hematological toxicities at this dose, supporting a wide preclinical safety margin. This dose is also translationally aligned with the clinical BD2-selective inhibitor RVX-208. Based on body surface area normalization, the 20 mg/kg murine dose of ZX29 converts to a human equivalent dose of approximately 1.6 mg/kg/day, which is nearly identical to the clinically recommended dose of RVX-208 (100 mg b. i.d., ∼2.86 mg/kg/day), and is well below the 10 mg/kg no-observed-adverse-effect level (NOAEL) of RVX-208 reported in preclinical rodent studies69, 70, 71.

For the 4-day prophylactic administration window, this design is specifically tailored to the unique pharmacodynamics of epigenetic BET inhibitors. Unlike acute GPCR antagonists that act immediately upon receptor binding, BET inhibitors require a sufficient lead time to remodel the chromatin landscape, including the BRD4–YY1–NCOA4 axis identified in this study. This pretreatment effectively establishes a ferroptosis-resistant molecular phenotype in cardiomyocytes to preempt reperfusion injury, while the short peri-operative exposure window avoids the cumulative systemic toxicities associated with long-term BET inhibitor administration.

5. Conclusions

This study validates BD2-selective BRD4 inhibition as a safe and effective strategy for MI/RI by targeting three interconnected pathological processes including fibrosis, ferroptosis, and mitochondrial dysfunction. ZX29 offers greater efficacy than RVX-208, a favorable safety profile, and potential oral bioavailability—advantages that underscore its promise as a clinical candidate. Future work should focus on optimizing ZX29's pharmacokinetics and exploring its synergistic effects with standard reperfusion therapies. Ultimately, our findings highlight the potential of BD2-selective inhibitors to address the unmet clinical need for MI/RI treatment.

Author contributions

Zhongkai Wu, Yong Zou, Peiqing Liu were responsible for the experimental design and supervision of this project. Zeyu Li performed most of the experiments and was the main writer of the manuscript, Ou Hu contributed to animal experiments. Zeyu Li contributed to the data analysis and manuscript writing. All authors made important suggestions regarding the manuscript.

Conflicts of interest

The authors declare no conflicts of interest.

Appendix A. Supplementary information

The following is the Supporting Information to this article.