All animal experiments were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals published by the National Research Council (The National Academy Presses, eighth edition, 2011), pre-approved and performed under the regulations of the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (Approval #A2017-024C3 and #A2022-175C3). Wild-type (WT) mice (C57BL/6jjcl) were purchased from CLEA Japan, Inc. All experiments were performed on male mice. For the in vivo study, C57BL/6jjcl mice aged 9–11 weeks (20–25 g of body weight) underwent transverse aortic constriction (TAC) or sham surgery as previously described (17). TAC operation was performed using 7-0 polypropylene with a 27-gauge needle as an indicator under anesthesia and mechanical ventilation. C57BL/6jjcl mice were prepared separately for the isolation of cardiac fibroblasts (CFs).

TAC-operated mice were divided into three groups to receive valsartan (Val; 30 mg/kg body weight) (Tokyo Chemical Industry Co., Tokyo, Japan) or sacubitril/valsartan (SacVal; 60 mg/kg body weight) (Angene Chemical, London, UK) dissolved in corn oil or just corn oil (vehicle; Veh). The dosages of Val and SacVal were selected according to previous studies (18–20). Corn oil was also administered to the mice in the sham-operated mice. Oral administration was performed by gavage twice daily starting the day after TAC or sham surgery. Drug administration was continued until the heart extraction was performed 14 days after surgery.

Blood pressure in awake mice was evaluated using a tail-cuff plethysmograph (BP-98AL; Softron, Tokyo, Japan), following the manufacturer's instructions. The mice were examined by plethysmography before surgery and on days 1, 7, and 14 after surgery. Surface electrocardiography (ECG) and ultrasound echocardiography (UCG) were performed in a blinded manner on postoperative day 14 as described previously (21, 22). During the examination, mice were anesthetized with less than 1.0% of isoflurane. The body temperature of mice was monitored during the experiments and maintained at 37.0°C using a heating pad and a light bulb.

Programmed stimulation for inducing AF, in vivo, was conducted using a custom-made pacing electrode inserted into the murine esophagus under anesthesia with 1.5% isoflurane. Body temperature was monitored and maintained at 37.0°C as described above. A pacing bipolar catheter was inserted into the esophagus, just dorsal to the left atrium, where the smallest capture threshold was obtained. All programmed electrical stimulations were delivered at a voltage that was twice the capture threshold. Atrial stimulus was performed with programmed stimulation up to triple extra stimulus pacing. Extra-stimuli were delivered after 10 atrial beats (S1) with a pacing cycle length (PCL) of 80 ms. The first extra stimulus (S2) was initially set 40 ms after the last pacing stimulus of S1. The S2 stimulus was delivered at progressively shorter coupling intervals with 5 ms steps until the effective refractory period (ERP). A second extra stimulus (S3) was added 40 ms after S2 and scanned with a 5 ms decremental step until the ERP was reached. A third extra stimulus (S4) was introduced after S3 in the same way as S2 and S3.

AF was defined as rapid and irregular beats in which the P wave was not recognized on the ECG and lasted for more than one second. The AF duration was defined as the interval between the initiation of arrhythmic atrial beats and the onset of the first recovered sinus beat.

Optical mapping was performed in a blinded manner, with minor modifications, as described previously (23). The mice were intraperitoneally injected with unfractionated heparin (200 IU) and sacrificed by cervical dislocation. Hearts were rapidly excised and connected to a Langendorff perfusion circuit. Subsequently, the hearts were stained with 5.0 µM of di-4ANEPPS (Santa Cruz Biotechnology, Dallas, Texas, USA), followed by the administration of 250 µM of blebbistatin (Toronto Research Chemicals, Toronto, Canada). The fluorescence was recorded by Complementary Metal Oxide Semiconductor camera (MiCAM ULTIMA; Brainvision Inc., Tokyo, Japan) with a sampling rate of 0.2 ms and spatial resolution of 66 µm. An ex vivo electrophysiological study was performed using a custom-made pacing electrode placed on the right atrium. The pacing stimulations were delivered at a voltage twice the capture threshold. Constant pacing was delivered with a pacing interval of 120 ms. The pacing interval was decreased in 20 ms steps until the cessation of 1:1 conduction. Action potential duration (APD) was obtained at 90% repolarization (APD₉₀) for each image pixel. Conduction velocity (CV) was derived using an isochronal map as previously described (24). Action potential (AP) rise time was defined as the interval from the onset, which was at the highest second derivative of the AP, to peak AP deflection. The dependence of APD₉₀, CV, and AP rise time on the PCL was also evaluated, namely rate adaptation. Heterogeneity was assessed using the coefficient of variation (CoV) from a map [defined as the standard deviation divided by the median value inside a region of interest (ROI)]. In all the above analyses, the ROI was commonly set inside the left atrium (LA), excluding the border. Comparisons between the groups were performed using the median values in the ROI on each map.

The excised hearts were immersed in 10% buffered formalin and the paraffin sections were subjected to Masson's trichrome staining. The stained sections were photographed (DZ-710, Keyence, Osaka, Japan), and the fibrotic areas were quantified by using the ImageJ software (NIH). The proportion of fibrotic area was calculated as the ratio of the fibrotic area to the total cross-sectional area.

Total RNA was extracted from heart tissue using the RNeasy Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's protocol. cDNAs were synthesized using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, Massachusetts, USA) from 200 ng total RNA obtained from the heart tissue.

For RNA extraction from the cultured cells and cDNA synthesis, cell lysis and reverse transcription were conducted using the SuperPrep II Cell Lysis & RT Kit for qPCR (TOYOBO, Osaka, Japan) according to the manufacturer's instruction.

Quantitative polymerase chain reaction (qPCR) was performed using StepOnePlus (Thermo Fisher Scientific) with PowerSYBR Green Master Mix (Thermo Fisher Scientific) and custom-made primers. Expression levels were normalized to Gapdh. The primer sequences are listed in Supplementary Table S1.

Protein samples from murine atrial tissues were extracted using radioimmunoprecipitation assay buffer. Protein concentration was measured using a BCA protein Assay Reagent Kit (Thermo Fisher Scientific). Of the protein samples, 20 µg was loaded on a 4%–15% SDS-PAGE gel (Bio-Rad Laboratories, inc., Hercules, California, USA) for the separation and transferred to PVDF membranes with Trans-Blot Turbo system (Bio-Rad Laboratories, inc.). After blocking with Blocking One or Blocking One-P (Nacalai tesque, Kyoto, Japan), the membranes were incubated overnight at 4°C with the primary antibody, as described in Supplementary Table S2. HRP-labeled secondary antibodies (Supplementary Table S2) were used to detect specific bands. Chemiluminescent signals were detected using an iBright CL1500 (Thermo Fisher Scientific). The band intensity was quantified using ImageJ software and normalized to that of GAPDH.

Adult cardiac fibroblasts (CFs) were isolated and cultured as previously described with minor modifications (25). In brief, C57BL/6jjcl mice without surgery were sacrificed by cervical dislocation, and hearts were quickly excised, and CFs were isolated from whole murine hearts. The isolated CFs were resuspended with Dulbecco's Modified Eagle Medium/F12 (Nacalai tesque) with 10% FBS, followed by plating into 100 mm dishes and culture in a humidified chamber at 37°C with 5.0% CO₂. CFs were used for in vitro assays between passage 1 and 3. After CFs were seeded in 96-well plates (3.0 × 10⁴ cells per well) and cultured for 24 h, culture medium was replaced with serum-free medium. At the same time, a pre-treatment with Atrial Natriuretic Peptide (ANP) (Phoenix Pharmaceuticals, Inc., Burlingame, California, USA), B-type Natriuretic Peptide (BNP) (Bachem, Bubendorf, Switzerland) or C-type Natriuretic Peptide (CNP) (MedChemExpress, Monmouth Junction, New Jersey, USA) at 1.0 or 5.0 µM was initiated with or without antagonists of their cognate receptors; A71915 (Sigma Aldrich, Burlington, Vermont, USA) antagonizing natriuretic peptide receptor-A (NPR-A); P19 (Phoenix Pharmaceuticals, Inc.) antagonizing natriuretic peptide receptor-B (NPR-B); AP811 (R&D Systems, Minneapolis, Minnesota, USA) antagonizing natriuretic peptide receptor-C (NPR-C), at 1.0 µM each, as well as LBQ657 (sacubitrilat; Cayman Chemical, Ann Arbor, Michigan, USA) at 10 or 50 µM. Transforming growth factor β (TGFβ; PeproTech, Cranbury, New Jersey, USA) (1.0 ng/ml) was added 30 min after the application of NPs, followed by another incubation for 24 h. After incubation with the drugs, RNA was extracted from the cultured cells and reverse transcription was performed as described above. The mRNA expression levels of several fibrosis-related genes were assessed in vitro. LBQ657 was dissolved in dimethyl sulfoxide (DMSO), TGFβ in citric acid, and the other drugs in ultra-pure water. We established a vehicle control to which an equivalent dose of these solvents was added. The final concentrations of DMSO and citric acid were adjusted to be less than 0.1%.

All data are presented as mean ± standard error of the mean (SEM). Statistical analyses were performed using PRISM 9 software (GraphPad). Statistical differences were assessed using one-way or two-way ANOVA for multiple comparisons of individual means, followed by Tukey's post-hoc analysis. For the statistical comparison of AF inducibility, chi-square test adjusted using the Benjamini-Hochberg procedure was performed. P < 0.05 was considered statistically significant.

Perpetual ventricular pressure overload by TAC is known to induce not only ventricular remodeling but also atrial remodeling, which is associated with the high inducibility of AF (26, 27). To establish the AF mouse model, we performed TAC surgery on WT mice, and acute pressure overload with TAC successfully caused heart failure with cardiac hypertrophy and deteriorated ventricular function in two weeks (Figures 1A,B). Regarding hypertrophy at the gene level, enhanced Myh7 and reduced Myh6 expression were observed in the ventricles with pressure overload, which was restored in the SacVal-treated group (Supplementary Figure S1). UCG showed that SacVal ameliorated the increase in wall thickness, dilation of the left ventricular diastolic diameter, and a significant decline in both fractional shortening and ejection fraction in comparison with the Veh (Figures 1C,D). Blood pressure was measured on days 1, 7, and 14 days after TAC procedure (Supplementary Figure S2A). Veh-, Val-, and SacVal-treated mice showed similar changes in peripheral blood pressure (Supplementary Figure S2B). The heart rate of mice in awake on postoperative day 14 did not differ between the groups (Supplementary Figure S2C). Furthermore, we found that the increase in left atrial weight in response to TAC was alleviated by SacVal and Val (Figure 1B). We measured the expression of the mRNA encoding ANP, BNP, and CNP: Nppa, Nppb, and Nppc respectively. TAC mice showed a significant increase in the expression of Nppa in atria and Nppb in ventricles. In contrast, the expression of Nppc in atria significantly decreased when exposed to pressure overload. Administration of SacVal significantly reduced the expression of Nppb in the ventricles as much as Val did, compared to the Veh group. In contrast, moderate reduction in the SacVal-treated group compared with the Veh-treated group was seen in that of Nppa in atria, and little change between them was observed in that of Nppc in atria (Figure 1E). Regarding ANP and CNP expression at the protein level in atria, ANP displayed similar trends to the gene expression of Nppa, but CNP showed different trends from its gene expression, indicating that the tissue concentration of CNP in atria increased under heart failure and SacVal modified the levels (Supplementary Figures S3A,B). Alternatively, the decreased gene expression of the cognate receptors of the NPs (Npr1, Npr2, and Npr3) in atria was observed in the Veh-treated group, whereas the Npr2 and Npr3 expression was significantly recovered in the SacVal-treated group; while the incremental recovery in their expression was not significant in the Val-treated group (Supplementary Figure S3C). In addition, pressure overload significantly suppressed the expression of Mme (encoding NEP) in atria with respect to both gene and protein expression levels, and SacVal recovered their expression significantly, whereas Val did not show a significant recovery (Figures 1F,G). In short, the gene expression profiles of NPs, their cognate receptors, and Mme in heart tissue were significantly altered in heart failure, and then SacVal restored those alterations as well as Val.

To examine the influence of SacVal on atrial arrhythmogenicity, we assessed the inducibility of AF by programmed stimulation from the esophagus (Figure 2A). Surface ECG at baseline revealed that P-wave duration was increased in the Veh-treated group compared to that in the sham-treated group. Treatment with SacVal remarkably shortened the P-wave duration in comparison to the Veh-treated group, which was similar to that with Val (Figure 2B). Treatment with SacVal also tended to ameliorate the prolongation of QRS duration due to heart failure; however, these changes were not significant (Figure 2B). The programmed stimulation induced AF with a relatively high probability in the Veh-treated group and the occurrence of AF was significantly suppressed only in SacVal-treated group (40% vs. 4.5%), whereas the inducibility of AF in the Val-treated group was approximately 20% (Figure 2C). The duration of induced AF in the SacVal-treated group tended to be shorter than that in the Veh- or Val-treated groups (Figure 2D). These findings suggested that SacVal could reduce the susceptibility to AF in heart failure more than Val.

To investigate the potential effects of SacVal treatment on atrial electrophysiological properties, the optical mapping with high spatiotemporal resolution was conducted. We processed the data obtained from atrial fluorescence into an isochronal map and an APD map, which were subsequently converted into a CV map and an AP rise time map (Figures 3A,B). These electrophysiological alterations were further assessed using a decremental PCL by 20 ms down to 40 ms. Representative maps for each group at PCL 100 ms and 40 ms are shown in Figure 3C. When compared between groups, APD maps revealed that the prolonged APD₉₀ under heart failure was restored by both Val and SacVal significantly (Figure 3D). CV was reduced in the Veh-treated group, and SacVal-treatment reversed this reduction in CV, which was not observed in the Val-treated group (Figure 3E). This finding corresponds to the change in P-wave duration observed in the body-surface ECG. Similarly, the prolonged AP rise time observed in Veh-treated group got shortened significantly in SacVal-treated group as well as Val-treated group (Figure 3F). The difference in APD₉₀ between the groups was diminished at a PCL of 40 ms (Figure 3G), whereas the significant restoration of CV and AP rise time was consistently observed even with a PCL of 40 ms only in the SacVal-treated group compared to the Veh-treated group (Figures 3H,I). The restitution curves of APD₉₀, CV and AP rise time (Figures 4A–C) confirmed that SacVal improved those restitution in heart failure more effectively than Val. Taken together, SacVal might have a more favorable anti-arrhythmogenic effect than Val by preventing impaired atrial electrophysiological properties in heart failure.

To further investigate the electrophysiological effects of SacVal, we evaluated the spatial heterogeneity of the APD₉₀, CV, and AP rise time. The coefficient of variation (CoV) derived from each map was compared between the groups. The CoV of APD₉₀ tended to be slightly smaller in all groups, with no significant differences as the PCL shortened (Figure 5A). In contrast, the CoV of the CV showed a tendency to increase with a shorter PCL in all groups. Notably, the CoV of the CV in the Veh-treated group significantly increased at PCL 40 ms, which was alleviated in the SacVal-treated group (Figure 5B), suggesting that SacVal could suppress the atrial heterogeneity of the CV in heart failure. Similarly, SacVal tended to suppress the increase in CoV of AP rise time (Figure 5C).

Regarding the electrophysiological alterations in the APD₉₀, CV and AP rise time, we assessed the gene expression levels of ion channels, including connexins. Corresponding to the results of APD₉₀, the downregulated gene expression of several subtypes of potassium ion channels, including Kcnd2 and Kchh2 in the Veh-treated group was restored in the SacVal-treated group, while SacVal did not restore the downregulated expression of Kcnj2 (Supplementary Figure S4A). Additionally, Cacna1c and its encoding protein Cav1.2 did not show apparent differences between the groups (Supplementary Figures S4B,C). Meanwhile, despite the improvement of impaired atrial conduction by SacVal, only slight alterations in the expression of Scn5a, Gja1, Gja5, and their encoding proteins, Nav1.5, Cx43, Cx40, respectively were found, compared with Val (Supplementary Figures S4D,E, S5A–C); thus, these altered gene and encoding protein expressions could not explain the additional anti-arrhythmic effect of SacVal, and another mechanism could be contributing to it. Taken together, SacVal administration led to the less heterogeneity in the electrophysiological properties, which could contribute to improving the atrial arrhythmogenicity.

Atrial fibrosis was examined to assess the underlying mechanism of the atrial electrophysiological effects of SacVal, particularly conduction heterogeneity. Histological assessment with Masson's trichrome staining showed that the deposition of collagen fibers in the atria significantly increased with heart failure, and SacVal significantly reduced the volume of atrial fibrosis compared with Veh and even Val (Figures 6A,B). The expression levels of the fibrosis-related genes, such as Tgfb1, Col1a2 and Col3a1, in the SacVal-treated group were significantly suppressed to the same extent as those in the Val-treated group (Figure 6C). Similarly, the expressions of Tnf, Il6 and Il1b were also assessed and found to be alleviated by SacVal and Val (Supplementary Figure S6). Furthermore, to investigate the signaling pathways related to fibrosis, the protein expression level of SMAD3, which is phosphorylated to activate the signaling cascade, was examined. Corresponding to the histological findings, the phosphorylated SMAD3 at Ser423/425 (pSMAD3) was elevated in the atria of TAC mice, whereas SacVal and not Val tended to attenuate the expression of pSMAD3 (Figures 6D,E). In summary, SacVal manifested the anti-fibrotic alterations in atria with heart failure, which could be more effective than Val, and might contribute to the beneficial effect of SacVal on atrial conduction properties.

According to previous in vivo and in vitro studies, the effects of NPs are assumed to be enhanced by NEP inhibitors (28, 29); hence, we examined the functions of ANP, BNP, and CNP against fibrosis in vitro using isolated CFs (Figure 7A). BNP and CNP conferred significant suppression of the gene expression of Ccn2 and Acta2 which indicate activation of CFs, after the application of exogenous TGFβ, whereas ANP did not bring any significant anti-fibrotic alterations (Figure 7B). Subsequently, we evaluated the anti-fibrotic effect of the NPs by adding LBQ657, a metabolite and an active form of sacubitril. Intriguingly, the addition of LBQ657 to CNP, but not to BNP, was associated with a significant enhancement of the anti-fibrotic effect, suggesting that sacubitril could enhance the anti-fibrotic effect of CNP rather than BNP on CFs, at least in vitro (Figure 7C). Similar trends were observed for the expressions of Col1a2 (Supplementary Figure S7). To confirm that the effects of LBQ657 were attributed to the prevention of CNP degradation, we evaluated whether the additional effect of LBQ657 was antagonized by the NPR-B inhibitor, P19. We found that the existence of P19 inhibited the anti-fibrotic effects of CNP (Figure 7D). Furthermore, the anti-fibrotic effect of CNP was similarly diminished by the addition of the NPR-C inhibitor, AP811 (Figure 7E). These results indicate that the additional anti-fibrotic effect when SacVal was added was partly mediated by CNP via NPR-B and NPR-C.