Resistin-like molecule γ attacks cardiomyocyte membranes and promotes ventricular tachycardia

Abstract

Ventricular tachycardia disrupts the heart's coordinated pump function, leading to sudden cardiac death. Neutrophils, which are recruited in high numbers to the ischemic myocardium, promote these arrhythmias. Comparing neutrophils with macrophages, we found that resistin-like molecule γ (Retnlg or RELMγ) was the most differentially expressed gene in mouse infarcts. RELMγ is part of a pore-forming protein family that defends the host against bacteria by perforating their membranes. In mice with acute infarcts, leukocyte-specific Retnlg deletion reduced ventricular tachycardia. RELMγ elicited membrane defects that allowed cell exclusion dyes to enter the cardiomyocyte interior and also caused delayed afterdepolarizations and later cardiomyocyte death, both of which are strong arrhythmogenic triggers. Human resistin likewise attacked membranes of liposomes and mammalian cells. We describe how misdirected innate immune defense produces membrane leaks and ventricular arrhythmia.

Fig. 1.. Mouse Retnlg and human RETN are upregulated in neutrophils after MI.

(A) ScRNA-seq was previously performed (20) in three mice 24 hours after MI and reanalyzed here. Retnlg expression labeled (red) in cellular subsets (PMN, neutrophils). (B) Fold change of the most upregulated genes in neutrophils compared to macrophages/monocytes. (C) Retnlg expression in neutrophils (PMN), monocytes (Mono), and macrophages (Mac) from the infarct area 24 hours after MI (n=3 mice; FDR < 2.2e-16). (D) Retnlg expression relative to Gapdh measured by quantitative PCR in circulating neutrophils (PMN) and monocytes at baseline (n=7 mice), 5 hours (n=3 mice), and 24 hours (n=4 mice) after MI; one-way ANOVA, Tukey’s post-hoc test. (E) Retnlg expression relative to Gapdh determined by quantitative PCR in the left ventricle (LV; n=5 mice), macrophages (Mac; n=8 mice), monocytes (Mono; n=8 mice), and neutrophils (PMN; n=8 mice) 5 hours after MI. (F) Immunofluorescent staining of neutrophils (Ly6G), RELMγ, and nuclei (DAPI) in the femurs of C57Bl6 mice; arrows indicate co-localization of Ly6G and RELMγ, scale bar 20 μm. (G) Immunofluorescent staining of Ly6G⁺ neutrophils, RELMγ and nuclei (DAPI) in the heart 24 hours after MI; arrows indicate co-localization of Ly6G and RELMγ, scale bar 20 μm. (H) Spatial transcriptomics of mouse heart cross-section for ischemic zone (IZ) score and Retnlg 24 hours after ischemic injury. (I) Pearson correlation for Retnlg and IZ gene set score co-localizations. Each dot represents a spatial transcriptomics spot. (J) SnRNA-seq and spatial transcriptomics were previously performed in infarcted human hearts and controls (25) and reanalyzed here. Spatial plot of the ischemic zone after MI and control tissue of a human heart. (K) RETN positive dots per mm² in ischemic zone (IZ, n=9) and control tissue (Ctrl, n=4) of human hearts, two-sided unpaired t-test. Data are mean±SD.

Fig. 2.. RELMγ deficiency reduces arrhythmia.

(A) Experimental outline: Bone marrow donors were either C57Bl/6 Retnlg^+/+ or Retnlg^−/− mice. Transplant recipients were C57Bl/6 wild-type mice (n=10 Retnlg^+/+, n=10 Retnlg^−/−) that subsequently underwent the STORM protocol (↓K⁺, hypokalemic diet; MI, myocardial infarction). (B) ECG from Retnlg^+/+ BMT control and Retnlg^−/− BMT, scale bar 500 ms. (C) Ventricular tachycardia (VT) burden evaluated 24 hours after MI (n=9 Retnlg^+/+ BMT, n=10 Retnlg^−/− BMT), Mann-Whitney U test. (D) Arrhythmia 24 hours after MI (n=9 Retnlg^+/+ BMT, n=10 Retnlg^−/− BMT), Mann-Whitney U test. (E) VT burden expressed as cardiac cycles in VT per hour. (F) Experimental outline: Ly6G^CreRetnlg^fl/fl mice and Retnlg^fl/fl control mice underwent the STORM protocol of hypokalemia and MI. (G) Ventricular tachycardia burden 24 hours after MI (n=5 Retnlg^fl/fl, n=6 Ly6G^CreRetnlg^fl/fl), Mann-Whitney U test. (H) Arrhythmia 24 hours after MI (n=5 Retnlg^fl/fl, n=6 Ly6G^CreRetnlg^fl/fl), Mann-Whitney U test. (I) Stimulated action potential recordings (whole-cell current-clamp) of mouse cardiomyocytes after incubation with RELMγ or vehicle for 30 min (DAD, delayed afterdepolarization). (J) Percentage of arrhythmic cells with DADs (control, 8 cardiomyocytes from 5 mice; RELMγ, 8 cardiomyocytes from 3 mice), Fisher's exact test. (K) DAD incidence per minute (control, 8 cardiomyocytes from 5 mice; RELMγ, 8 cardiomyocytes from 3 mice), Mann–Whitney test. Data are mean±SD.

Fig. 3.. RELMγ attacks lipid membranes.

(A) Electron microscopy of liposomes in the presence of vehicle or recombinant RELMγ; asterisks indicate pore, scale bar 50 nm. (B) Pore diameter in liposomes after exposure to RELMγ. (C) Electron microscopy of liposomes in the presence of either vehicle or human RETN, scale bar 50 nm. (D) Pore diameter after exposure to RETN. (E) Calcein-loaded liposomes treated with RELMγ. Dye efflux is expressed as percentage of maximal release after triton. Lower panel shows dose-response. Results average at least 3 independent experiments. (F) Calcein-loaded liposomes were treated with different concentrations of RETN. Lower panel shows concentration curves. Results average 3 independent experiments. (G) Confocal microscopy of a mouse cardiomyocyte; arrows indicate co-localization of phosphatidylserine staining (apotracker) and RELMγ, scale bar 10 μm. (H) Confocal microscopy of a mouse cardiomyocyte, showing phosphatidylserine (apotracker), RELMγ, and dextran; circles indicate co-localization of phosphatidylserine, RELMγ, and dextran cell entry. Numbers indicate areas that are (1) proximal, (2) intermediate and (3) distant to the dextran entry, scale bar 20 μm. (I) Target to baseline ratio (TBR) of fluorescence intensity indicating dextran influx in area 1-3 as well as control cardiomyocytes (no dextran influx), n=4 cells/group, two-way ANOVA for repeated measurements. (J) MINFLUX image of RELMγ, cyan dots indicate RELMγ molecules (on-events), black outline indicates convex hull area, look-up table reflects distance to nearest center of on-event (0 nm to 75 nm), scale bar 200 nm. (K) Carpet area in super-resolution imaging (n=19 areas in 19 cardiomyocytes). (L) Scanning electron microscopy of a mouse cardiomyocyte showing ‘carpet-like’ deposits (white arrows), scale bar 2 μm. Data are mean±SD.

Fig. 4.. RELMγ and resistin promote mammalian cell death.

(A) Apoptosis assay of H9C2 cells exposed to 10 μM RELMγ or vehicle control in acidic pH (see methods) using propidium iodide (PE) and apotracker green (FITC). (B) Live cell fraction by flow cytometry, n=5 replicates per group, two-sided unpaired t-test. (C) Apoptosis assay of EXPI293 cells exposed to RELMγ or vehicle control in acidic pH, using propidium iodide (PE) and apotracker green (FITC). (D) Live cell fraction by flow cytometry, n=10 replicates per group, two-sided unpaired t-test. (E) Apoptosis assay with 10 μM human RETN (n=5 replicates) or vehicle (n=4 replicates), live cell fraction by flow cytometry, two-sided unpaired t-test. (F) Cell death assay measuring relative fluorescent units (RFU) of CellTox bound to primary isolated cardiomyocytes incubated with 0.5 μM Staurosporine with and without 1 μM RELMγ over 300 minutes, 5 replicates per group, two-way ANOVA for repeated measurements. (G) Immunofluorescence staining of infarct region, 5 hours after MI, in C57BL/6 Retnlg^+/+ BMT and Retnlg^−/− BMT mice; staining of cardiomyocytes (troponin), apoptotic cells (TUNEL), and nuclei (DAPI); arrows indicate apoptotic cells, scale bar 50 μm. (H) Analysis of TUNEL⁺ cells (n=6 Retnlg^+/+ BMT, n=7 Retnlg⁻/⁻ BMT). Ten fields of view per MI were analyzed, two-sided unpaired t-test. (I) Immunofluorescence staining of infarcted region in Retnlg^+/+ BMT and Retnlg^−/− BMT mice 24 hours after ischemia reperfusion injury; staining of cardiomyocytes (troponin), apoptotic cells (TUNEL), and nuclei (DAPI); arrows indicate apoptotic cells, scale bar 50 μm. (J) Analysis of TUNEL⁺ cells (n=7 Retnlg^+/+ BMT, n=8 Retnlg^−/− BMT), three fields of view per mouse were analyzed, two-sided unpaired t-test. Data are mean±SD.

Fig. 5.. RELMγ in ischemic stroke.

(A) ScRNA-seq of the brain was previously performed in 3 mice 24 hours after stroke (45) and re-analyzed here for Retnlg expression in brain leukocyte clusters (PMN, neutrophils). (B) Retnlg expression in brain after stroke in neutrophils (PMN) and other cell clusters (Mono/Mac, monocytes/macrophages; Micro, microglia; NK, natural killer cells; PMN, neutrophils; BC, B-cells; IL, innate lymphoid cell; TC, T cells). (C) Fold change of Retnlg by quantitative PCR in whole blood 24 hours after stroke (n=6 mice) compared to naive mice (n=4), two-sided unpaired t-test. (D) Triphenyltetrazolium chloride staining of brain after middle cerebral artery occlusion in Retnlg^−/− (n=11) and Retnlg^+/+ mice (n=10). (E) Quantification of stroke size, two-sided unpaired t-test. (F) Apoptosis assay of Neuro2a cells exposed to 10 μM RELMγ or vehicle control in acidic pH, using propidium iodide (PE) and apotracker green (FITC. (G) Live cell fraction by flow cytometry, 5 replicates per group, two-sided unpaired t-test. (H) Confocal microscopy of a Neuro2a cell in pH 5, incubated with fluorescently labeled RELMγ and (I) fluorescently labeled dextran, scale bar 5 μm. Magnification box: scale bar 1 μm. Data are mean±SD.