Cardioprotective effects of semaglutide on isolated human ventricular myocardium

Abstract

Aims: Semaglutide, a glucagon-like peptide-1 (GLP-1) receptor agonist, has shown promising effects in reducing cardiovascular events in patients with obesity and heart failure (HF) with preserved ejection fraction (HFpEF) irrespective of concomitant diabetes. However, the exact mechanisms underlying its cardioprotective actions remain unclear. Our study investigates the direct effects of semaglutide on human cardiomyocytes, focusing on calcium (Ca) and sodium (Na) handling and its potential to improve myocardial contractility.

Methods and results: Human left ventricular cardiomyocytes were isolated from non-failing (NF) hearts, patients with aortic stenosis and a HFpEF-like phenotype (AS), and those with end-stage HF with reduced ejection fraction (HFrEF). Late Na current (I_Na), sarcoplasmic reticulum (SR) Ca leak, and contractility were assessed in isolated cardiomyocytes treated with semaglutide. CaMKII inhibitor autocamtide-2-related inhibitory peptide and GLP-1 receptor antagonist exendin 9-39 (Ex-9-39) were used to elucidate signalling pathways. Semaglutide reduced late I_Na in AS and HFrEF cardiomyocytes to levels comparable to NF. Additionally, semaglutide decreased diastolic SR Ca leak and improved systolic Ca transients and contractility in AS and HFrEF tissue. These effects were mediated through GLP-1 receptor agonism and were comparable to CaMKII inhibition. In multicellular preparations, semaglutide differentially improved myocardial contractility in AS and HFrEF in a dose-dependent manner.

Conclusion: Semaglutide directly modulates ion homeostasis in human cardiomyocytes, reducing proarrhythmic diastolic SR Ca leak and enhancing systolic function, which may explain its observed clinical benefits. These findings provide mechanistic insights into the cardioprotective effects of semaglutide and suggest its potential therapeutic use in HF.

Keywords: Calcium; Heart failure; Human cardiomyocytes; Semaglutide.

Figure 1

Semaglutide (Sema) reduces late sodium current (I_Na) and diastolic calcium (Ca) leak in human ventricular cardiomyocytes. (A) Original recordings (left panel) and mean data (right panel) of late I_Na in human ventricular cardiomyocytes from non‐failing (NF), aortic stenosis (AS), and heart failure with reduced ejection fraction (HFrEF) patients (n = patients). In cardiomyocytes from NF patients, no relevant late I_Na could be observed upon vehicle control (V), which was consequently unaltered by Sema. In contrast, significant late I_Na was observed in cardiomyocytes from patients with AS and a HFpEF‐like phenotype, which was reduced to NF levels already at 50 nmol/L Sema. This effect was comparable to CaMKII inhibition with 2 μmol/L autocamtide‐2 related inhibitory peptide (AIP). The inhibitory effect of Sema was abrogated by glucagon‐like peptide‐1 (GLP‐1) receptor antagonism with exendin‐9–39 (Ex‐9–39). Similarly, significant late I_Na was observed in cardiomyocytes from patients with end‐stage heart failure, which was inhibited by 50 nmol/L Sema. Ex‐9–39 also prevented the inhibitory effect of Sema on late I_Na in HFrEF cardiomyocytes. Data tested with a mixed‐effects model (overall *p = 0.0002). P‐values in figure represent Holm–Sidak post‐hoc p‐values. (B) Original recordings (left panel) and mean data (right panel) of late I_Na in murine cardiomyocytes from wild‐type C57BL6 mice upon H₂O₂ stress (n = mice). The H₂O₂‐induced enhancement of late I_Na was inhibited by 50, 100 and 300 nM Sema independent of the concentration. Data tested with a mixed‐effects model (overall *p = 0.0040). (C) Original confocal line scan recordings of Ca sparks (left panel) and mean data of Ca spark frequency (CaSpF, right panel) in human ventricular cardiomyocytes from NF, AS and HFrEF patients (n = patients). In NF cardiomyocytes, expectedly, no significant occurrence of Ca sparks was observed, and Sema did not affect Ca spark frequency. In contrast, significant occurrence of Ca sparks was observed in human AS and HFrEF cardiomyocytes, which were reduced to NF levels already by 50 nmol/L Sema, which was similar to the effect of CaMKII inhibition in HFrEF with 2 μmol/L AIP. In AS cardiomyocytes, no additional (or inverse) effect was observed upon 300 nmol/L Sema compared to 50 nmol/L. In contrast, GLP‐1 receptor antagonism with Ex‐9–39 abrogated the effect of simultaneous Sema exposure. Data tested with a mixed‐effects model (overall *p = 0.0039). (D) Original confocal line scan recordings of Ca sparks (left panel) and mean data of CaSpF (right panel). H₂O₂ stimulation was required to induce relevant sarcoplasmic reticulum Ca sparks in murine wild‐type cardiomyocytes and for a notable suppression of sarcoplasmic reticulum Ca sparks by Sema. Sema showed efficacy at 50, 100, and 300 nmol/L, but even at 300 nmol/L could be suppressed by GLP‐1 receptor antagonism with Ex‐9–39. Data tested using a mixed‐effects model (overall *p = 0.0027). (A–D) P‐values represent Holm–Sidak post‐hoc p‐values. Data shown as mean ± standard error of the mean per patient (A + C) or mouse (B + D).

Figure 2

Semaglutide (Sema) improves systolic calcium (Ca) transients dependent on glucagon‐like peptide‐1 receptor (GLP‐1‐R) expression levels. (A) Original recordings (left) and mean Ca transient amplitudes (CaTransAmpl) (right) of human ventricular cardiomyocytes from non‐failing (NF) patients. Expectedly, Sema did not show a significant positive inotropic effect. Data tested with a mixed‐effects model (overall p = NS). (B) Original recordings (left) and mean CaTransAmpl (right) of human ventricular cardiomyocytes from patients undergoing aortic valve replacement due to aortic stenosis (AS) and a heart failure with preserved ejection fraction‐like phenotype. In cardiomyocytes from AS, Sema improved CaTransAmpl. Data tested with a mixed‐effects model (overall *p = 0.0219). (C) Original recordings (left) and mean caffeine‐induced Ca amplitudes (right) as a marker of sarcoplasmic reticulum Ca content of human ventricular cardiomyocytes from AS. Sema improved sarcoplasmic reticulum Ca content comparably at 50 and 300 nmol/L. Data tested with a mixed‐effects model (overall *p = 0.0160). (D) Original recordings (left) and mean CaTransAmpl (right) of human ventricular cardiomyocytes from patients with heart failure with reduced ejection fraction (HFrEF). Sema did not improve CaTransAmpl at 50 nmol/L but was significant and effective at 300 nmol/L. Sema at 300 nmol/L showed comparable efficacy to CaMKII inhibition with 2 μmol/L autocamtide‐2‐related inhibitory peptide (AIP) (AIP showed borderline non‐significance in the post‐test, but a clear numerical effect). Data tested with a mixed‐effects model (overall *p = 0.0204). (E) Original recordings (upper panel) and mean CaTransAmpl (lower panel) of murine ventricular cardiomyocytes exposed H₂O₂‐stress ± Sema. Expectedly, H₂O₂ depressed CaTransAmpl. While numerical values differed between 50 and 300 nmol/L, all concentrations were effective and significant in ameliorating the pathological H₂O₂ stimulus. Data tested with a multifactor mixed‐effects model (overall *p for factor drug/intervention = 0.0152, *p for factor frequency = 0.001). For better readability, the full test results and post‐test values can be found in online supplementary Figure  S2A . (F) Western blot showing reduced expression of the GLP‐1‐R in ventricular myocardium from HFrEF compared to AS (tested with Student's t‐test). (A–F) Data shown as mean ± standard error of the mean per patient (A–D, F) or mouse (E). P‐values in figures (A–E) represent Holm–Sidak post‐hoc p‐values (shown if main test indicated significance). GAPDH, glyceraldehyde‐3‐phosphate‐dehydrogenase (housekeeping protein); V, vehicle control.

Figure 3

Semaglutide (Sema) improves contractility in human ventricular multicellular preparations. (A) Original recordings of contractions (force development, left) and normalized developed tension (right) of human left ventricular muscle slices from patients undergoing aortic valve replacement due to aortic stenosis (AS) and a heart failure with preserved ejection fraction‐like phenotype. Sema improved developed tension at 50, 100, and 300 nmol/L compared to vehicle control (V) (in separate slices from the same patients to exclude confounding rundown effects). Data tested using a multifactor mixed‐effects model (overall *p for factor drug/intervention = 0.0145, p for factor time = NS). Data shown as mean ± standard error of the mean (SEM) of individual slices (due to the normalization to baseline and comparison to corresponding vehicle slices from the same patients). (B) Original recordings of contractions (force development, left) and normalized developed tension (right) of human left ventricular muscle slices from patients with heart failure with reduced ejection fraction (HFrEF). Sema improved developed tension at 300 nmol/L. Data tested using a Mann–Whitney test. Data shown as mean ± SEM of individual slices (due to the normalization to baseline and comparison to corresponding vehicle slices from the same patients). (C) Original recordings of contractions (force development, left) and normalized developed tension (right) of isolated murine ventricular trabeculae exposed to H₂O₂ stress. H₂O₂ was expectedly negatively inotropic compared to the baseline measurement of the same trabeculae. Sema exerted a positive inotropic effect at 50, 100, and 300 nmol/L that was similar to CaMKII inhibition with 2 μmol/L autocamtide‐2‐related inhibitory peptide (AIP). Glucagon‐like peptide‐1 receptor antagonism with exendin‐9–39 (Ex‐9–39) abrogated the inotropic effect of Sema even at 300 nmol/L. Data shown as mean ± SEM per mouse. Data tested using a Kruskal–Wallis test (overall *p < 0.0001). (D) Original recordings of contractions (force development, left) and normalized developed tension (right) of isolated murine ventricular trabeculae exposed to vehicle control, Sema, or Sema + Ex‐9–39. Expectedly, without a pathological stimulus such as H₂O₂, no significant effect of Sema on developed tension was observed (and consequently no antagonization of such an effect by Ex‐9–39 could be detected). Data shown as mean ± SEM per mouse. Data tested using a Kruskal–Wallis test (overall p = NS). (A–D) P‐values represent Holm–Sidak (A + B) or Dunn (C + D) post‐hoc p‐values (shown if main test indicated significance).