The experimental animals used in this study were clean-grade Sprague–Dawley male rats, 8 weeks old, weighing 200–250 g, and purchased from Henan Skebes Biotechnology Co., Ltd. [Production Licence Number: SCXK (YU) 2020–0005]. These rats were housed in the experimental animal facility of the Anhui Province Key Laboratory of Meridian Viscera Correlationship, where the environmental conditions were maintained at a temperature of 24 ± 2 °C, relative humidity of 50–60%, and a 12 h light–dark cycle. During the housing period, the animals had free access to food and water. All experiments were conducted after the rats had undergone a 1-week adaptation period. Our work adhered to “the Guide for the Care and Use of Laboratory Animals” issued by the Chinese National Institutes of Health and was approved by the Animal Care and Use Committee of Anhui University of Chinese Medicine, with approval number AHUCM-rats-2024145.

Our work comprises four independent experiments, and the detailed grouping information is summarized in Supplementary Table S1. In experiment I, 24 rats were randomly divided into the following three groups: (1) the Sham group (n = 6), rats underwent sham CHF model surgery only. (2) The CHF group (n = 6), rats underwent CHF modeling surgery only. (3) The EA group (n = 6), rats were subjected to EA intervention following the establishment of the CHF model. (4) the Sham EA group (n = 6), rats were subjected to sham EA intervention following the establishment of the CHF model. Additionally, to verify the neural connections between the heart, the Shenmen (HT7) acupoint, and the PVN, the tracer virus was injected into the heart (n = 3) and the acupoint (n = 3). To ensure the precise location of the PVN, cholera toxin subunit B (CTB) was injected into the PVN for tracing (n = 3).

In experiment II, we performed lesioning of the PVN region, and 24 rats were randomly assigned to the following three groups: (1) the CHF + Saline group (n = 6), saline was injected into the PVN of the CHF rats. (2) The CHF + Saline+EA group (n = 6), following the injection of saline into the PVN of the CHF rats, EA intervention was performed. (3) The CHF + KA group (n = 6), kainic acid (KA) was injected into the PVN of the CHF rats. (4) The CHF + KA + EA group (n = 6), following the injection of KA into the PVN of the CHF rats, EA intervention was performed.

In Experiment III, we modulated CRH neurons in the PVN using the inhibitory virus, and 24 rats were randomly assigned to the following four groups: (1) The Sham+mCherry+CNO group (n = 6): The PVN of sham rats was injected with the control virus (rAAV-EF1α-DIO-mCherry, expressing only fluorescent protein). Following viral expression, clozapine-N-oxide (CNO) was intraperitoneally injected. This group served as a vector control to exclude off-target effects. (2) The CHF + mCherry+CNO group (n = 6): The PVN of CHF rats was injected with the control virus. Following viral expression, CNO was intraperitoneally injected. (3) The CHF + hM4Di + CNO group (n = 6): The PVN of CHF rats was injected with the inhibitory virus (rAAV-EF1α-DIO-hM4Di-mCherry). Following viral expression, CNO was intraperitoneally injected to activate the virus and inhibit neuronal activity. (4) The CHF + hM4Di + Saline group (n = 6): The PVN of CHF rats was injected with the inhibitory virus. Following viral expression, saline was intraperitoneally injected.

In Experiment IV, CRH neurons in the PVN were modulated using the excitatory virus, and EA intervention was performed. 24 rats were randomly divided into the following four groups: (1) The CHF + mCherry+CNO group (n = 6): Similar to Experiment III, CHF rats received the control virus followed by CNO injection. (2) The CHF + mCherry+CNO + EA group (n = 6): The control virus was injected into the PVN of CHF rats. Following viral expression, CNO was administered via intraperitoneal injection and combined with EA intervention. (3) The CHF + hM3Dq + CNO + EA group (n = 6): The excitatory virus (rAAV-EF1α-DIO-hM3Dq-mCherry) was injected into the PVN of CHF rats. Following viral expression, CNO was intraperitoneally injected to activate the virus and excite neuronal activity during EA intervention. (4) The CHF + hM3Dq + Saline+EA group (n = 6): The excitatory virus was injected into the PVN of CHF rats. Following viral expression, saline was administered via intraperitoneal injection combined with EA intervention.

The sample size was determined based on the “Resource Equation Method” to strictly adhere to animal welfare ethics (Reduction) while ensuring statistical validity (Charan and Kantharia, 2013). The study comprised four independent experimental series. For four experiments (4 groups each, n = 6), E was 20. These values fall within the optimal range (10–20) recommended for animal research. This sample size strategy is also consistent with established protocols in previous studies on EA for cardiovascular diseases (Wu et al., 2025; Zhou et al., 2025). Furthermore, post-hoc power analyses indicated that the statistical power (1-β) for key outcome measures exceeded 0.90 across all major comparisons, confirming the adequacy of the sample size.

The CHF rat model was established using the left anterior descending (LAD) artery ligation method (Litwin et al., 1994). Rats were fasted for 12 h prior to surgery but allowed access to water. General anesthesia was induced using 3% isoflurane (R510-22-10, RWD Life Science Co., Ltd.), and once unconscious, the rats were secured to the operating table. Anesthesia was maintained with 1.5–2% isoflurane. During anesthesia, standard lead electrocardiograms (ECG) were recorded using the Powerlab Standard Limb II lead system (ML118, AD Instruments International Trading Co., Ltd.). For surgical procedures, the hair on the left anterior chest was shaved and the area disinfected. A vertical incision was made on the left anterior chest perpendicular to the ribs, and the pectoralis major and minor muscles were bluntly dissected. A curved hemostatic forceps was inserted into the thoracic cavity at the fourth intercostal space, and the intercostal space was dilated using a chest expander to extrude the heart. The LAD branch of the coronary artery was ligated using 7–0 non-absorbable suture material at a position 2–3 mm below the left atrial appendage and a needle insertion depth of 0.5 mm to induce ischemia. After ligation, quickly reposition the heart into the thoracic cavity and expel any air from the thoracic cavity. ST-segment elevation (≥0.2 mV) on the ECG is used as the criterion for successful ligation (Figure 1A). Subsequently, layered sutures were performed using 4–0 absorbable sutures, and rats were given intraperitoneal injections of penicillin (200,000 U/mL, 0.5 mL/d) for three consecutive days to prevent infection. Four weeks (28 days) postoperatively, left ventricular ejection fraction (LVEF) was assessed via echocardiography. The model was considered successful if the LVEF value was ≤ 45% (Yuan et al., 2018; Figure 1B). The sham group underwent a single needle puncture at the corresponding site after thoracotomy without ligating the LAD coronary artery. Postoperatively, rats were monitored until recovery from anesthesia and administered penicillin via intraperitoneal injection to prevent infection and Carprofen (5 mg/kg, s.c.) to alleviate pain and distress daily for three consecutive days. Rats with abnormal ECG preoperatively, those that died during the experimental period, and those with unsuccessful modeling were excluded from the study. To maintain a consistent sample size, any excluded animals were replaced by additional rats that underwent the same procedures and met the inclusion criteria.

Based on published techniques for locating acupoints in experimental animals and comparisons with primate models (Shu et al., 2024), our study selected the Shenmen (HT7) point on the Hand-Shaoyin Heart Meridian for EA intervention. This point is situated at the wrist, on the palmar side of the transverse wrist crease, at the ulnar end, within the radial notch of the flexor carpi ulnaris tendon (Figure 1C). Prior to EA intervention, the bilateral Shenmen point regions in rats were disinfected according to standard protocol. Disposable sterile acupuncture needles (0.25 × 25 mm, Suzhou Tianxie Acupuncture Instrument Co., Ltd.) were inserted perpendicularly into the point to a depth of 2–3 mm. Following insertion, the HT7 point was connected to the negative electrode of the EA device (HANS-200A, Nanjing Jisheng Medical Technology Co., Ltd.), while the area 1 mm lateral to the point was connected to the positive electrode. The intervention employed a continuous current wave with a frequency of 2 Hz and an amperage of 1 mA, with stimulation intensity calibrated to elicit slight limb tremors (Figure 1D).

For the Sham EA group, a minimal acupuncture protocol widely adopted in previous studies was employed to strictly control for non-specific effects such as cutaneous stimulation (Wan et al., 2025; Xu et al., 2026; Guo et al., 2026). In this group, sterile needles were inserted shallowly into the subcutaneous tissue to a depth of 0.5 to 1 mm at the same acupoint without connecting to the electrical stimulator or delivering any current.

All EA and sham EA interventions were performed under isoflurane-induced anesthesia, administered once daily for 30 min over seven consecutive days. The control group, not receiving EA treatment, underwent anesthesia for the same duration.

Under isoflurane anesthesia (induction at 3%, maintenance at 1%), rats were secured in a stereotaxic apparatus. The ear bars were symmetrically positioned to ensure that the skull was horizontally aligned. The adequacy of fixation was verified by the alignment of the nose, the absence of head movement, and the skull’s horizontal orientation. After the scalp was shaved and disinfected, a midline incision was performed, and the dura mater was removed to expose Bregma. A burr hole was drilled at the coordinates corresponding to the PVN (relative to Bregma: anteroposterior [AP] ± 1.92 mm, mediolateral [ML] ± 0.35 mm, dorsoventral [DV] ± 7.6 mm), as per the stereotaxic coordinates from a rat brain atlas.

In accordance with the experimental grouping described previously, a microinjector (R480, RWD Life Science Co., Ltd.) was used to inject 30 nL of the following solutions into the PVN at a rate of 10 nL/min: CTB (CTB-555, Wuhan Shumi Brain Science and Technology Co., Ltd.), KA (K0250, Shanghai Sigma Audley Trading Co., Ltd.), saline (ST341, Beyotime Biotechnology Co., Ltd.), inhibitory virus [the 1:1 mixture of rAAV-CRH-CRE-WPRE-hGH polyA and rAAV-EF1α-DIO-hM4D(Gi)-mCherry-WPREs, Wuhan Shumi Brain Science and Technology Co., Ltd.], excitatory virus [the 1:1 mixture of rAAV-CRH-CRE-WPRE-hGH polyA and rAAV-EF1α-DIO-hM3D(Gq)-mCherry-WPREs, Wuhan Shumi Brain Science and Technology Co., Ltd.], and control virus (rAAV-EF1α-DIO-mCherry-WPRE-hGH, Wuhan Shumi Brain Science and Technology Co., Ltd.). After injection, the microinjection needle was left in place for 10 min, and the incision was closed with sterile sutures. Postoperatively, rats were monitored until recovery from anesthesia and administered penicillin via intraperitoneal injection to prevent infection and Carprofen (5 mg/kg, s.c.) to alleviate pain and distress daily for three consecutive days. After data collection from each rat, histological identification was performed by perfusion and sectioning to observe the injection site and confirm its accuracy, and data from inaccurate sites were excluded.

We employed pseudorabies virus (PRV) and herpes simplex virus (HSV) as trans-synaptic tracers to map the multi-synaptic neural circuits connecting the periphery to the central nervous system. To verify the neural connection between the heart and the PVN, we performed a microinjection of a tracing virus into the rat heart. Rats were anesthetized (induction at 3%, maintenance at 1%) and secured on an animal operating table, followed by tracheal intubation and connection to the small animal ventilator (R415, RWD Life Science Co., Ltd.). Once the physiological state of the rat stabilized, the chest was prepared and disinfected. A thoracotomy was then performed, and the sternum was retracted using a rib spreader to expose the heart. A microinjection of PRV tracing virus (PRV-CAG-mRFP, Wuhan Shumi Brain Science and Technology Co., Ltd.) was administered into the left ventricular wall at three distinct sites, with a volume of 1 μL per site and a rate of 100 nL/s. After the injection, the thoracic incision was sutured and disinfected.

To verify the neural connection between the HT7 acupoint and the PVN, we performed a microinjection of a tracing virus into the HT7 acupoint region. Rats were anesthetized (induction at 3%, maintenance at 1%) and secured on an animal operating table, and the area surrounding the HT7 acupoint was depilated, prepared, and disinfected. The skin was then incised to facilitate needle insertion. A microinjection of HSV tracing virus (HSV-EGFP, Wuhan Shumi Brain Science and Technology Co., Ltd.) was administered at three distinct sites 2–3 mm beneath the skin of the acupoint, with a volume of 1 μL per site and a rate of 100 nL/min. After the injection, the incision was sutured and disinfected. Postoperatively, rats were monitored until recovery from anesthesia and administered penicillin via intraperitoneal injection to prevent infection and Carprofen (5 mg/kg, s.c.) to alleviate pain and distress daily for three consecutive days.

Five days after the microinjection of the tracing virus, when the rats exhibited significant abnormal conditions, they were euthanized by continuous anesthesia with 5% isoflurane. The brain tissue was then extracted and observed under the microscope (magnification: × 10; DP72, Olympus Optical Co., Ltd.) to examine the viral tracing in the PVN region.

The ECG was recorded from rats using the Powerlab Standard Limb II lead system (ML118, AD Instruments International Trading Co., Ltd.). Rats were anesthetised (induction at 3%, maintenance at 1%) and secured to an animal restraint table. Needle electrodes were inserted subcutaneously into the right forelimb and left hindlimb, and intramuscularly into the right hindlimb. ECG recording commenced once stable waveforms were obtained. Each animal was recorded for 30 min. Following completion, electrodes were removed, rat skin cleaned, and equipment switched off. Heart rate variability (HRV) was measured using the HRV module in LabChart 8 software. Quantitative analysis was performed in the low-frequency (LF, 0.2–0.75 Hz), high-frequency (HF, 0.75–2.5 Hz), and LF/HF ratio of the power spectrum.

Rats were anesthetized with isoflurane (induction at 3%, maintenance at 1%) and secured in a supine position on the animal platform. Chest hair was removed using a depilatory agent, and a coupling agent was applied. Cardiac function was then assessed using a digital ultrasound system (VINNO6 Lab, Feiyino Technology Co., Ltd.) equipped with an 18 MHz echo transducer. Then used M-mode recording to measure LVEF and left ventricular short-axis shortening (LVFS). For each rat, three technical replicate echocardiographic measurements were performed, and the mean value was calculated.

Neuronal discharge in the PVN brain region of rats was recorded using electrophysiological techniques. Rats were anesthetized (induction at 3%, maintenance at 1%) and secured in a stereotaxic apparatus using the same method as in the brain stereotaxic surgery. The scalp hair was shaved, disinfected, and a midline incision was made to remove the dura mater and expose Bregma. As described previously, a cranial hole was drilled above the PVN brain region after locating it. Subsequently, an eight-channel (2 × 4) microelectrode array (designed, Kewa Suzhou Medical Technology Co., Ltd.) was implanted. The electrode was slowly advanced to the target brain region at a rate of 5 μm/s using electric propulsion. Once stable neural activity was observed, recording began and lasted for 400 s. After recording, the incision was sutured with sterile sutures, and the rats were given intraperitoneal injections of penicillin for three consecutive days to prevent infection. The OmniPlex multichannel acquisition system (version 1.20.0, Beijing Plexon Technology Co., Ltd.) was used to record neuronal discharge (filtered at 150–8000 Hz, sampling rate 40 kHz) and field potentials (filtered at 0.7–400 Hz, sampling rate 1 kHz). After data acquisition, Offline Sorter software (version 4.7.2, Beijing Plexon Technology Co., Ltd.) was first used to remove typical high-amplitude interference signals using the waveform cross-correlation method. Subsequently, NeuroExplorer software (version 5, Beijing Plexon Technology Co., Ltd.) was employed to analyze neuronal signals. To strictly define functional activity and exclude background noise, neurons were classified as “active” only if they exhibited a spontaneous mean firing rate greater than 2 Hz and a stable signal-to-noise ratio exceeding 3:1 throughout the recording period (Quiroga et al., 2004; Luther and Tasker, 2000; Buzsáki, 2004). Based on these inclusion criteria, spike discharge raster diagrams, autocorrelograms, and spectrograms were generated to compare the frequency and characteristics of neuronal signals among the groups. Postoperatively, rats were monitored until recovery from anesthesia and administered penicillin via intraperitoneal injection to prevent infection and Carprofen (5 mg/kg, s.c.) to alleviate pain and distress daily for three consecutive days. Finally, histological identification was performed. After tissue fixation and sectioning, the location was observed to confirm the accuracy of the electrode implantation site.

All experiments in this work were conducted in accordance with the principle of minimizing animal suffering. After the intervention, rats were subjected to tissue collection. Blood was first collected via intraperitoneal puncture, followed by the extraction of the heart and brain tissues. This sequence prevented blood loss and coagulation due to the removal of the heart and brain, thereby ensuring the quality and quantity of the blood samples. Prior to blood collection, rats were anesthetized (induction at 3%, maintenance at 1%), and blood was drawn from the abdominal aorta. Subsequently, rats were euthanized by continuous induction with 5% isoflurane until cessation of heartbeat and respiration, confirming death. The heart and brain tissues were then harvested. Brain tissues were immediately rinsed in phosphate buffered saline (PBS) to remove surface blood and post-fixed by immersion in 4% paraformaldehyde at 4 °C for 24–48 h. Throughout the blood collection and tissue extraction procedures, strict aseptic techniques were adhered to in order to prevent sample contamination. The samples were stored according to experimental requirements for subsequent analyses.

After blood collection from the abdominal aorta of rats, the blood was placed vertically in a refrigerator at 4 °C and allowed to coagulate naturally for 10–20 min. Subsequently, the blood was centrifuged in a high-speed refrigerated centrifuge (4 °C, 3000 revolutions per min, 15 min). The supernatant obtained from centrifugation was prepared as standard samples. Standard samples were prepared, spiked, and incubated in strict accordance with the manufacturer’s instructions for the Enzyme-linked immunosorbent assay (ELISA) kit (AF03205-B, AF03182-B, Hunan Aifang Biotechnology Co., Ltd.). Absorbance values were measured using an ELISA instruction, and the serum levels of N-terminal pro-brain natriuretic peptide (NT-proBNP) and norepinephrine (NE) in each sample group were calculated based on the standard curve generated.

After blood collection, the diaphragm and pericardium were incised, and the heart was carefully extracted using curved forceps. The heart was then rinsed in pre-cooled 0.9% sodium chloride solution at 4 °C. A transverse incision was made 5 mm above the apex of the heart. Hematoxylin–eosin (HE) staining is employed to assess pathological changes in myocardial tissue. The procedure for HE staining is as follows: Initially, paraffin-embedded tissue sections are placed on a staining rack and subjected to deparaffinization by sequential immersion in xylene I, II, and III for 3 min each, followed by rehydration through a graded series of ethanol solutions (absolute ethanol I for 2 min, absolute ethanol II for 2 min, 95% ethanol for 1 min, 90% ethanol for 1 min, and 80% ethanol for 1 min). After rehydration, the sections are briefly drained on a staining rack for 30 s. The sections are then stained with hematoxylin for 10 min, followed by a gentle rinse in tap water to remove excess stain. Subsequently, the sections are differentiated in a 1% hydrochloric acid-ethanol solution for 6 s, drained for 4 s, and then soaked in tap water for 10 min. For counterstaining, the sections are immersed in eosin for 2.5 min, followed by a 30 s drain on the staining rack. Dehydration is performed through a graded ethanol series (80% ethanol for 10 s, 90% ethanol for 10 s, 95% ethanol for 2 min, absolute ethanol I for 3 min, absolute ethanol II for 3 min, and absolute ethanol III for 3 min). The sections are then cleared in xylene (xylene I, II, and III for 3 min each). Finally, the sections are mounted with neutral resin, taking care to avoid the formation of air bubbles, and allowed to dry at room temperature for 2–3 d before examination under a microscope (magnification: × 20; DP72, Olympus Optical Co., Ltd.).

Masson staining was employed to evaluate the condition of myocardial fibrosis. The procedure was executed with precision using a Masson staining kit, as described herein. Initially, tissue sections were placed on a staining rack and subjected to deparaffinization by sequential immersion in xylene I, II, and III for 3 min each, followed by a rehydration series through graded ethanol solutions (95, 70, and 30%) and distilled water for 2 min each. Subsequently, the sections underwent a warm water rinse at 30–40 °C for 30–60 s twice, ensuring the removal of all droplets except for the samples. The staining process involved hematoxylin staining for 60 s, followed by a tap water rinse for 30 s, and then Masson’s trichrome staining for 30–60 s with another tap water rinse. Differentiation was performed in a 6–8% phosphoric acid solution for 6–8 min, adjusted based on the appearance of collagen fibers under a microscope (magnification: × 20; DP72, Olympus Optical Co., Ltd.) until they turned pink. Counterstaining with light green was applied for 5 min, followed by a final tap water rinse. Dehydration was achieved through a series of ethanol solutions and xylene, culminating in mounting with neutral resin. To quantify collagen deposition within the myocardial tissue, the collagen volume fraction (CVF) was determined using ImageJ software (version 6.0).

To verify the accuracy of the stereotaxic injection and the extent of the excitotoxic lesions, brain sections containing the PVN were processed for Nissl staining. Briefly, after the brain tissues were harvested and fixed in 4% paraformaldehyde as described above, they were dehydrated and sectioned at a thickness of 30 μm. The sections were mounted on gelatin-coated slides, air-dried, and rehydrated. They were then stained with 0.1% cresyl violet solution (Sigma-Aldrich, St. Louis, MO, United States) for 10–15 min at room temperature. Following staining, the sections were differentiated in 95% ethanol, dehydrated through a graded series of alcohols (70, 90, and 100%), cleared in xylene, and cover slipped with neutral gum. The stained sections were examined under a light microscope (Olympus, Tokyo, Japan) to assess neuronal survival. Lesioned areas were identified by the loss of Nissl bodies, neuronal shrinkage (pyknosis), and the absence of normal neuronal cytoarchitecture compared to the intact surrounding regions.

Fixed brain tissues were dehydrated in graded sucrose solutions (15 and 30%) and sectioned into 30-μm coronal slices using a cryostat. The rat brain slices were immersed in PBS and washed three times, each for 5 min. Subsequently, the brain slices were placed in a blocking buffer solution containing 0.5% Triton X-100 and 3% bovine serum albumin (BSA) for 1 h at room temperature.

For c-Fos expression analysis, the slices were incubated with the primary antibody (2250S, Saixintong Shanghai Biological Reagents Co., Ltd.) at a 1:500 dilution overnight at 4 °C. For the co-localization analysis of virus and CRH neurons, the slices were incubated with anti-CRH antibody (10944-1-AP, Wuhan Sanying Biotechnology Co., Ltd.) overnight at 4 °C. On the following day, the slices were rinsed three times with PBS, each for 5 min. After rinsing, the slices were immersed in a buffer solution containing 0.5% Triton X-100 and 3% BSA. Corresponding secondary antibodies (111–545-003, Shanghai Youningwei Biotechnology Co., Ltd.) were added at a 1:500 dilution, and the mixture was incubated at room temperature in the dark for 2 h. Following this, the slices were rinsed three times with PBS, mounted with an anti-fluorescence quenching solution containing 4′,6-diamidino-2-phenylindole (DAPI, abs9235-25 mL, Aibixin Shanghai Biotechnology Co., Ltd.), and stained. After staining, three images were randomly selected for photography. Images were captured using a panoramic tissue section imaging system (magnification: × 20; PanoBrain, Meca Scientific, PB-ST100, Shanghai BOYE Biotechnology Co., Ltd.). The number of c-Fos-positive neurons was analyzed using ImageJ software (version 6.0). For co-localization verification, the endogenous fluorescence of mCherry and the immunofluorescence of CRH were observed, and the percentage of mCherry-positive neurons co-expressing CRH was calculated.

All data were analyzed using GraphPad Prism (version 8.0) and presented as the mean ± standard deviation (S. D.). Normality was assessed via the Shapiro–Wilk test, while homogeneity of variance was evaluated using the Brown-Forsythe test. For comparisons between two groups, data were initially tested for normality. Those satisfying normality criteria were analyzed using either the Student’s T-test or Paired T-test. Data that failed to meet normality assumptions were subjected to the Mann–Whitney U test or Wilcoxon signed-rank test. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was employed if normality and homogeneity assumptions were met, followed by the Holm–Šidák (or Bonferroni) post-hoc test to correct for multiple comparisons. If assumptions were not met, the Kruskal-Wallis H test was used, followed by Dunn’s post hoc test. Correlations were assessed using Pearson’s correlation coefficient. To control for the inflated type I error rate due to multiple testing in correlation analyses, p-values were adjusted using the Benjamini-Hochberg False Discovery Rate (FDR) method (or Bonferroni correction). Statistical significance was set at p < 0.05.

To assess the effects of EA on CHF, the model was established via LAD ligation followed by EA or Sham EA treatment. Systematic evaluations including histology, echocardiography, and biochemical assays were conducted (Figure 2A).

Histological analysis showed that the CHF group exhibited significant cardiomyocyte edema, inflammation, and severe fibrosis. EA treatment markedly improved myocardial structure and reduced fibrosis, whereas the Sham EA group retained severe pathological changes and high CVF levels comparable to the CHF group (Figures 2B,C).

Functionally, EA treatment significantly reversed the reduced LVEF and LVFS observed in CHF rats while the Sham EA group failed to improve cardiac function (Figures 2D,E). Consistently, EA lowered the elevated serum NT-proBNP, NE levels, and HRV LF/HF ratio which suggested reduced cardiac burden and restored autonomic balance. In contrast, the Sham EA group showed no significant improvement in these biochemical and autonomic parameters compared to the CHF group (Figures 2F–H).

Overall, EA exerted significant cardioprotective effects. Given the lack of therapeutic efficacy in the Sham EA group, we reasoned that the underlying central neural mechanisms would remain consistent with the CHF baseline. Therefore, the Sham EA group was excluded from subsequent invasive mechanistic analyses to focus on the specific mechanisms of effective EA treatment.

Following the administration of PRV, a transneuronal tracer known for its ability to map motor efferent pathways, into the heart and HSV, a tool used for tracing sensory afferent pathways, into the HT7 acupoint, we observed corresponding viral fluorescence in the PVN of the rat brain. The presence of PRV in the PVN indicates a motor efferent pathway from the PVN to the heart, suggesting that the PVN can modulate cardiac function (Figure 3A). Conversely, the detection of HSV in the PVN signifies the sensory afferent pathway from the HT7 acupoint, indicating that sensory information from HT7 is transmitted to the PVN (Figure 3B). Together, these findings confirm the existence of neural circuits connecting the heart, HT7 acupoint, and PVN, thereby establishing a functional link between acupuncture stimulation at HT7 and cardiac regulation via the PVN. These viral tracing results qualitatively confirm the structural existence of the neural circuit connecting the heart, PVN, and HT7. The specific functional involvement of CRH neurons within this circuit was subsequently investigated using cell-type specific chemogenetics in the following experiments.

Immunofluorescence staining was used to detect c-Fos-positive neurons in the PVN (Figure 3C). Results showed significant upregulation in the CHF group compared with the sham group, which was downregulated by EA (Figures 3D,E). This indicated that EA inhibits the expression of c-Fos in the PVN.

The location of the PVN was confirmed by administering the CTB injection (Figure 4A). Electrophysiological recordings showed that PVN neuronal activity frequency was higher in the CHF group than in the sham group and lower in the EA group than in the CHF group (Figures 4B–D). Spectrograms analysis revealed higher spectral energy in the CHF group than in the sham group and lower in the EA group than in the CHF group (Figure 4E). Autocorrelation analysis indicated two active neurons in the sham and CHF groups, with four neurons showing enhanced activity after EA at the Shenmen acupoint (Figure 4E).

To elucidate the underlying mechanisms, we performed a series of correlation analyses. The LF/HF ratio was positively correlated with CVF, NT-proBNP, and NE, and negatively correlated with LVEF and LVFS (Figures 5A–E). The expression of c-Fos was positively associated with CVF, NT-proBNP, NE, and the LF/HF ratio, and negatively associated with LVEF and LVFS (Figures 5F–K). The mean firing rate was positively correlated with CVF, NT-proBNP, NE, the LF/HF ratio, and c-Fos expression, and negatively correlated with LVEF and LVFS (Figures 5L-R).

Collectively, these findings delineate a strong statistical association among the variables: in CHF, PVN neuronal activation (marked by c-Fos) is positively correlated with autonomic imbalance (elevated LF/HF ratio) and deteriorated cardiac function. This suggests that the therapeutic effects of EA may be linked to its modulation of this axis.

To explore the role of the PVN in the effects of EA on CHF, we induced PVN lesions with KA and compared outcomes among the CHF + Saline, CHF + Saline+EA, CHF + KA, and CHF + KA + EA groups (Figures 6A,B). Prior to functional assessment, the accuracy and specificity of the lesions were confirmed via Nissl staining. The PVN region in KA-treated rats exhibited significant neuronal loss, nuclear shrinkage, and necrosis, whereas the surrounding brain tissues remained structurally intact (Figure 6B).

Regarding the impact of the lesion itself, the CHF + KA group exhibited reduced sympathetic activity and slightly improved cardiac function compared to the CHF + Saline group, indicating that ablating the hyperactive PVN offers partial cardioprotection. However, the CHF + Saline+EA group demonstrated the most significant improvements in myocardial structure, reduced fibrosis, and enhanced cardiac function (increased LVEF and LVFS) compared to the CHF + Saline group.

Notably, when EA was applied to lesioned rats, the CHF + KA + EA group showed further improvements compared to the CHF + KA group, suggesting EA exerts residual effects via alternative pathways. Nevertheless, compared to the potent effects in the CHF + Saline+EA group, the CHF + KA + EA group exhibited significantly attenuated cardioprotection, characterized by lower LVEF and LVFS, and higher NT-proBNP and NE levels (Figures 6C–H). HRV analysis confirmed these findings, showing the lowest LF/HF ratio in the CHF + Saline+EA group, whereas the CHF + KA + EA group showed a significantly higher ratio (Figure 6I). These findings underscore the critical role of the PVN in mediating the maximal cardioprotective effects of EA in CHF.

To elucidate the role of CRH neurons in the beneficial effects of EA on CHF, we injected adeno-associated virus (AAV) vectors into the PVN and modulated the activity of CRH neurons using the chemogenetic tool CNO (Figures 7A,B). First, we rigorously validated the specificity and efficacy of our chemogenetic system. Specifically, in the CHF + hM4Di + Saline group, immunofluorescence staining revealed that 92.86% of mCherry-expressing neurons co-localized with endogenous CRH, confirming that the virus specifically targeted CRH neurons with minimal off-target infection (Figure 7C). Subsequently, to verify the functional efficacy of inhibition in vivo, we analyzed c-Fos expression. As shown in Figure 7D, CNO administration significantly decreased the percentage of mCherry+ neurons co-expressing c-Fos in the CHF + hM4Di + CNO group compared to the CHF + hM4Di + Saline group, confirming the effective suppression of neuronal activity. Following this validation, we assessed the cardiac outcomes. The CHF + mCherry+CNO group showed severe myocardial damage and increased fibrosis, while the CHF + hM4Di + CNO group exhibited improved structure and reduced fibrosis (Figures 7E,F). Echocardiography revealed reduced LVEF and LVFS in the CHF + mCherry+CNO group, which improved with hM4Di + CNO treatment (Figures 7G,H). Serum NE and NT-proBNP levels decreased with hM4Di + CNO, indicating reduced sympathetic activity (Figures 7I,J). HRV analysis showed improved autonomic balance in the CHF + hM4Di + CNO group (Figure 7K). These findings suggest that inhibiting CRH neurons can ameliorate CHF, similar to EA effects.

The chemogenetic strategy was employed to activate CRH neurons and assess their impact on rats with CHF undergoing EA treatment (Figures 8A,B). We verified the efficacy of chemogenetic activation. In the CHF+hM3Dq+CNO+EA group, CNO administration significantly increased the percentage of mCherry+ neurons co-expressing c-Fos compared to the CHF+hM3Dq+Saline+EA group (Figure 8C), confirming that hM3Dq-mediated activation successfully reactivated PVN neurons and counteracted the inhibitory effect of EA. The CHF+mCherry+CNO+EA group showed improved myocardial structure and reduced fibrosis compared to the CHF+mCherry+CNO group, indicating the beneficial effects of EA (Figures 8D–G). However, the CHF+hM3Dq+CNO+EA group exhibited worsened cardiac pathology, suggesting that CRH neuron activation counteracted benefits of EA. Echocardiography confirmed enhanced cardiac function in the CHF+mCherry+CNO+EA group, which was diminished in the CHF+hM3Dq+CNO+EA group. Serum NE and NT-proBNP levels were reduced in the CHF+mCherry+CNO+EA group but increased in the CHF+hM3Dq+CNO+EA group (Figures 8H,I). HRV analysis showed improved autonomic balance in the CHF+mCherry+CNO+EA group and worsened balance in the CHF+hM3Dq+CNO+EA group (Figure 8J). These results indicate that CRH neuron activation diminishes cardioprotective effects of EA in CHF rats.