Prevention of Fibrosis and Pathological Cardiac Remodeling by Salinomycin

Abstract

RATIONALE:

Cardiomyopathy is characterized by the deposition of extracellular matrix by activated resident cardiac fibroblasts called myofibroblasts. There are currently no therapeutic approaches to blunt the development of pathological fibrosis and ventricle chamber stiffening that ultimately leads to heart failure.

OBJECTIVE:

We undertook a high-throughput screen to identify small molecule inhibitors of myofibroblast activation that might limit the progression of heart failure. We evaluated the therapeutic efficacy of the polyether ionophore salinomycin in patient-derived cardiac fibroblasts and preclinical mouse models of ischemic and nonischemic heart failure.

METHODS AND RESULTS:

Here, we demonstrate that salinomycin displays potent anti-fibrotic activity in cardiac fibroblasts obtained from heart failure patients. In preclinical studies, salinomycin prevents cardiac fibrosis and functional decline in mouse models of ischemic and nonischemic heart disease. Remarkably, interventional treatment with salinomycin attenuates preestablished pathological cardiac remodeling secondary to hypertension and limits scar expansion when administered after a severe myocardial infarction. Mechanistically, salinomycin inhibits cardiac fibroblast activation by preventing p38/MAPK (mitogen activated protein kinase) and Rho signaling. Salinomycin also promotes cardiomyocyte survival and improves coronary vessel density, suggesting that cardioprotection conferred by salinomycin occurs via the integration of multiple mechanisms in multiple relevant cardiac cell types.

CONCLUSIONS:

These data establish salinomycin as an antifibrotic agent that targets multiple cardioprotection pathways, thereby holding promise for the treatment of heart failure patients.

GRAPHIC ABSTRACT:

A graphic abstract is available for this article.

Repair of tissue injury requires the deposition of an ECM (extracellular matrix)-rich scar by activated resident fibroblasts referred to as myofibroblasts.^1,2 In heart disease or following an ischemic event such as myocardial infarction (MI), fibroblast activation is initially an adaptive attempt to preserve cardiac structure and function. However, excessive myofibroblast activation and fibrosis invariably develop in chronic heart disease of multiple causes. Ultimately, pathological cardiac scarring alters the biomechanical properties that support normal ventricle contractility and disrupts electrical and mechanical coupling of cardiomyocytes leading to diastolic dysfunction and an increased risk of arrhythmia—a primary cause of death in heart failure (HF).

ECM deposition by myofibroblasts is stimulated by many biomechanical cues, including mechanical tension, proreparative inflammatory signaling, Wnt/β-catenin signaling, canonical and noncanonical TGF (transforming growth factor)-β1 signaling, and Rho/ROCK (Rho-kinase) signaling.^3–5 Recent studies have utilized fibroblast-specific deletion strategies in mice to demonstrate the role of TGFBR (transforming growth factor beta receptor) 1/2, Smad2/3, and p38α (Mapk14) in myofibroblast activation, pathological fibrosis, and cardiac dysfunction.^5–8 Additional studies from our group and others have revealed Rho-dependent transcriptional changes also underlie pathological organ fibrosis.^3,8–10 Activated fibroblasts may also act as sentinels of tissue injury via paracrine mechanisms to modulate the reparative process. Many of the factors secreted by myofibroblasts are cytokines, chemokines, and matrix-remodeling factors—a complicated mixture that has both protective and deleterious effects during cardiac remodeling.¹¹

Despite a growing appreciation of the pathways underlying myofibroblast activation, pharmacological strategies to blunt the development of fibrosis remain elusive.¹² In the current study, we performed a high-throughput screen (HTS) of the 2300 compound MicroSource Spectrum library to identify compounds that blocked the activation of a myofibroblast reporter in cultured cells. This screen revealed polyether ionophores, particularly salinomycin, among the most potent inhibitors of myofibroblast activation, consistent with the results of a previous HTS to identify antifibrotic TGF-β inhibitors.¹³ The previous reports revealed salinomycin inhibited myofibroblast formation in human orbital fibroblasts and p38 signaling and EMT (epithelial to mesenchymal transition) in retinal pigment epithelial cells.^13,14 Given the identification of salinomycin among the most potent antifibrotic compounds in 2 independent screens, we were encouraged to test the efficacy of salinomycin in preclinical models of pathological cardiac fibrosis.

Salinomycin treatment resulted in near-complete inhibition of cardiac remodeling when treatment began at the time of insult in both ischemic (MI) and nonischemic (AngII [angiotensin II] infusion) disease models. Mechanistically, salinomycin works in multiple ways—directly inhibiting p38/MAPK (mitogen activated protein kinase)-dependent myofibroblast activation, altering the fibroblast secretome, and directly inhibiting cardiomyocyte hypertrophy by suppressing ERK (extracellular signal-regulated kinase) signaling and p65 nuclear accumulation in response to pathological stimuli. Importantly, salinomycin attenuated preestablished AngII-dependent remodeling when given as an interventional drug, and prevented infarct expansion and further functional decline when given 3 days after MI. In the MI model, we observed a significant degree of myocardial preservation. Thus, our study reveals salinomycin mediates protection from pathological remodeling that is not limited to its initially discovered antifibrotic properties and is an attractive candidate for further development for clinical use in the setting of HF.

METHODS

This study adheres to the Transparency and Openness Promotion guidelines. The data that support the findings in this study are available from the corresponding author upon reasonable request.

A detailed description of all experiments can be found in the Materials and Methods in the Data Supplement.

Study Approval

All animal experiments were approved by the University Committee on Animal Resources at the University of Rochester. Human samples were obtained following informed consent in accordance with an approved Institutional Review Board Protocol at the University of Rochester Medical Center.

Human Cardiac Fibroblast Isolation

HF cardiac fibroblasts (HFCF) were isolated from left ventricular (LV) tissue obtained from patients undergoing ventricle assist device implantation.

Mouse Models

AngII-infused and MI models were employed in 12-week old C57BL/6J male mice to assess nonischemic and ischemic cardiac remodeling, respectively.

Cardiac Physiology

Echocardiographic analysis was performed using a Vevo2100 ultrasound (VisualSonics, Toronto, Canada) and a linear-array 40 MHz transducer (MS-550D).

Gene Expression

Gene expression was evaluated using iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) and a CFX Connect Real-Time Polymerase Chain Reaction Detection System (Bio-Rad). Relative mRNA levels were normalized to ribosomal 18s and calculated using the 2^−ΔΔCt method. Primer sequences are found in Table I in the Data Supplement.

Histology, Morphometric Image Analysis, and Protein Measurements

Detailed Western blot, immunofluorescence, and image analysis procedures are found in the Expanded Methods. Representative images were chosen as the replicate closest to the statistical mean of the data range. Antibodies and dilutions are found in Table II in the Data Supplement.

High-Throughput Screen for Antifibrotic Compounds

A stable NIH-3T3 cell line harboring an ACTA2-luciferase myofibroblasts reporter was used to screen the SPECTRUM small molecule library of 2400 compound (MicroSource) pinned at 5 μmol/L using a Janus robotic liquid handling system (Perkin Elmer, Norwalk, CT) and a FlexDrop plate filler (Perkin Elmer, Norwalk, CT).

Cytokine Arrays

A total of 100 μL of sample was prepared by combining equal amounts of serum obtained 14 days after osmotic mini-pump implantation from saline-treated (n=3) and AngII-treated (n=5) mice. For HFCF conditioned media, a total of 100 μL of sample per treatment condition was prepared, with 25 μL of the sample coming from each distinct patient-derived line (n=4). Detection of cytokines was performed according to the manufacturer’s instructions (R&D Systems).

Statistical Analysis.

Data are reported as mean±SEM and statistics are performed using Prism 8.4 (GraphPad). Comparisons between 2 groups were performed using unpaired, 2-tailed t tests with Welch correction for unequal standard deviations. Comparisons between >2 groups were performed using 1- or 2-way ANOVA with appropriate corrections (Brown-Forsythe and Welch for 1-way ANOVA, Geisser-Greenhouse for 2-way ANOVA, and repeated-measures ANOVA as appropriate). All measures taken over a gradient employed repeated-measures ANOVA. Post hoc comparisons were made using the Dunnett T3 test (for 1-way ANOVA), Tukey test (2-way and repeated-measures ANOVA where ≥3 groups were compared), or Bonferroni test (2-way and repeated-measures ANOVA where 2 groups were compared). In all cases, P<0.05 is considered significant. Figure legends include denotation of significant comparisons, and for sake of space, a complete list of tests and P values for each comparison in each experiment may be found in Table III in the Data Supplement.

RESULTS

HTS Identifies Salinomycin as a Potent anti-fibrotic drug.

To identify small molecules that prevent myofibroblast activation and fibrosis, we developed an HTS using a luciferase reporter driven by the promoter and first intron of ACTA2, a well-characterized marker of activated myofibroblasts.⁹ NIH-3T3 fibroblasts were stably transfected with the ACTA2-luciferase reporter, and 3 clones generated a Z’ score of ≥0.80 when comparing TGF-β1 treatment to vehicle, and clone 14 displayed the most robust dynamic range (160 000 luminescence units; Figure IA and IB in the Data Supplement). We confirmed TGF-β1 dependent induction of endogenous Acta2 protein and RNA by Western blot and quantitative real-time polymerase chain reaction for clone 14 (Figure IC and ID in the Data Supplement).

ACTA2-luciferase clone 14 was used to screen the MicroSource Spectrum library—a diverse drug collection comprised of ≈2320 natural/botanical compounds and Food and Drug Administration–approved compounds (Figure 1A). Cells were cultured in 384-well format with vehicle, TGF-β1 (5 ng/mL), or TGF-β1 in combination with a compound from the library using a robotic liquid handling system (≈5 μmol/L), in duplicate. After 16 hours, luminescence and viability (propidium iodide) were measured; we identified 44 antifibrotic and 24 profibrotic hits that altered ACTA2-luciferase activity by more than 50% while maintaining <10% variability (Figure 1B and Figure IIA in the Data Supplement). Fifty-five antifibrotic and 56 profibrotic drugs altered reporter activity by >50% while inducing <10% cell death (Figure 1C and Figure IIB in the Data Supplement). Of the low cell death compounds, 29 antifibrotic drugs also displayed low cell death in primary human CF (Figure 1D and Figure IIC and IID in the Data Supplement). Three of the 29 antifibrotic hits are polyether ionophores (narasin, salinomycin, and monensin A; Figure IIB and IID in the Data Supplement), which each inhibited TGF-β1 dependent ACTA2-luciferase induction in the high nanomolar range (Figure 1E). Salinomycin was evaluated for the remainder of this study.

Figure 1. High-throughput screen reveals antifibrotic compounds.

. A, Schematic of high-throughput screening strategy of the SPECTRUM Collection using an NIH-3T3 cell line stably expressing an ACTA2-luciferase myofibroblast reporter. B, ACTA2-luciferase alteration (%) plotted against variability between duplicate wells (%). C, ACTA2-luciferase (luc) alteration (%) plotted against cell death (%). D, Cell death in human heart failure (HF) patient–derived cardiac fibroblasts (CF) and mouse NIH-3T3. E, Three potassium ionophores (salinomycin, narasin, monensin A) were evaluated for dose-responsive suppression of TGF (transforming growth factor)-β1 (5 ng/mL)-dependent ACTA2-luciferase. D-actinomycin was used as a negative control. F, Human CF were treated with TGF-β1 (5 ng/mL)±salinomycin (SNC; 1 μmol/L). ACTA2 and POSTN (periostin) expression was evaluated by quantitative real-time polymerase chain reaction. n=3 experiments. 1=P<0.05 relative to control. G, ACTA2 protein expression (red) was evaluated in human CF in response to TGF-β1 (5 ng/mL)±SNC (1 μmol/L) treatment by immunofluorescence. Phalloidin (green) labels actin in all fibroblasts and DAPI (blue) labels nuclei. Scale bar=25 μm. H, Quantification of percent ACTA2+ cells in (G), normalized to control. n=9 patients were assayed in triplicate. 1=P<0.05 compared with vehicle; 2=P<0.05 compared with TGF-β1/AngII (angiotensin II); 3=P<0.05 compared with TGF-β1/AngII/SNC. I. Collagen production was determined by hydroxyproline incorporation in human HF cardiac fibroblasts subjected to the indicated treatments. n=9 individual patients. 1=P<0.05 compared with saline-saline; 2=P<0.05 compared with TGF-β1/AngII/saline; 3=P<0.05 compared with saline-SNC. J Proteins were detected in the conditioned media isolated from human HF cultures of the indicated treatments by cytokine array dot blot. Each dot blot reflects the combination of conditioned media from 3 independent patient cultures. ACTA2 indicates smooth muscle alpha actin; AU, arbitrary units; CArG, CC-A rich-GG box; CXCL, C-X-C motif ligand; DAPI, 4’,6-diamidino-2-phenylindole; LIF, LIF interleukin 6 family cytokine; P, patient; SRE, SMAD responsive element; TSP, thrombospondin; T/SNC, TGF-β and salinomycin treatment; and VEGF, vascular endothelial growth factor.

Salinomycin did not impact CF proliferation at baseline, but partially inhibited proliferation induced by TGF-β1/AngII treatment (Figure IIE in the Data Supplement). Importantly, salinomycin reduced the expression of endogenous myofibroblast markers at the gene and protein level in CF isolated from HF patients (HFCF, Figure 1F through 1H) and suppressed collagen production as assessed by hydroxyproline incorporation (Figure 1I). Salinomycin also normalized the secretory profile as demonstrated by enhanced LIF (LIF interleukin 6 family cytokine) and attenuated TSP (thrombospondin)-1 in conditioned media (Figure 1J).¹⁵ This was accompanied by increased secretion of angiogenic factors such as CXCL (C-X-C motif chemokine ligand) 12 and VEGF (vascular endothelial growth factor).¹⁶

Salinomycin Prevents AngII-Induced Pathological Cardiac Fibrosis

We next evaluated the therapeutic efficacy of salinomycin as an antifibrotic in a preclinical mouse model of nonischemic pathological cardiac remodeling. Twelve-week-old male C57Bl/6J mice were administered saline or AngII (1000 ng/[kg·min]) via osmotic pumps and randomly assigned to either saline or salinomycin treatment starting at the time AngII was initiated (salinomycin preventative regimen, called SNCp). Since salinomycin is rapidly metabolized by the liver via CYP3A4 (cytochrome P450 family 3 subfamily A member 4) in mice, we used multiple daily intraperitoneal injections of salinomycin as opposed to a single large bolus (1 mg/kg 3× per day).¹⁷ Experimental design and timing of cardiac physiology and systemic blood pressure measurements are indicated in Figure 2A.

Figure 2. Salinomycin (SNC) prevents AngII (angiotensin II)-induced cardiac remodeling.

A, Experimental timeline for AngII infusion (1 mg/[kg·min]) and preventative SNC treatment (1 mg/kg 3× per day) in mice. Serial echocardiography and blood pressure (BP) time points are indicated. B and C. Echocardiographic assessment of Left ventricle (LV) ejection fraction (B) and fractional shortening (C). n=6 (control, saline), 6 (AngII, saline), 9 (control, SNCp), 9 (AngII, SNCp). 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to saline/saline condition at a given time point; 3=P<0.05 relative to saline/SNC condition at a given time point; 4=P<0.05 relative to AngII/SNC condition at a given time point. D, LV sections isolated at 28 d were stained with PicroSirius Red. Scale bar=100 μm. E and F. Quantification of fibrosis from (D) in perivascular (E) and interstitial (F) regions of 3 nonoverlapping fields of view per heart. n=3 (control, saline), 3 (AngII, saline), 5 (control, SNCp), 5 (AngII, SNCp) hearts. 1=P<0.05 relative to saline/saline; 2=P<0.05 relative to AngII/saline; 3=P<0.05 relative to saline/SNC; 4=P<0.05 relative to AngII/SNC. G, Dot blot was performed to evaluate the level of indicated proteins in the serum of mice 14 d into the AngII/SNCp regimen. Each dot blot is the result of serum pooled from 3 mice (AngII/saline), or 5 mice (AngII/SNCp). H and I, Fibroblasts isolated from hearts of indicated treatment at 28 d were subjected to quantitative real-time polymerase chain reaction for genes encoding (H) myofibroblast markers and (I) ECM (extracellular matrix) molecules. n=3 (control, saline), 3 (AngII, saline at 28 d), 4 (control, SNCp), 4 (AngII, SNCp). 1=P<0.05 relative to saline/saline; 2=P<0.05 relative to AngII/saline; 3=P<0.05 relative to saline/SNC; 4=P<0.05 relative to AngII/SNC. J, LV sections were stained for DAPI (nuclei), IB4 (isolectin-B4; vasculature), and Postn (periostin) (activated myofibroblasts). WGA (wheat germ agglutinin), lower) revealed cardiac structure. Scale bar=100 μm. K, Quantification of (J) in 3 nonoverlapping fields of view from the LV from each animal were measured. n=3 (control, saline), 3 (AngII, saline at 28 d), 5 (control, SNCp), 5 (AngII, SNCp). 1=P<0.05 relative to saline/saline; 2=P<0.05 relative to AngII/saline; 3=P<0.05 relative to saline/SNC; 4=P<0.05 relative to AngII/SNC. DAPI indicates 4’,6-diamidino-2-phenylindole; Echo, echocardiography; MMP, matrix metallopeptidase; Sal, saline; and SNCp, salinomycin preventative regimen.

Serial echocardiography revealed that, although mice treated with SNCp alone had normal cardiac physiology, SNCp prevented AngII-induced cardiac dysfunction (Figure IIIA through IIID in the Data Supplement). AngII-infused mice treated with saline are characterized by a persistent decrease in LV ejection fraction beginning at 14 days and a decrease in LV fractional shortening at 7 days, which were both completely normalized by SNCp treatment (Figure 2B and 2C). Of note, SNCp did not block the elevation in systolic blood pressure observed in AngII-infused mice nor did it alter the resting heart rate (Figure IIIE and IIIF in the Data Supplement).

To assess whether improved cardiac physiology is accompanied by reduced cardiac fibrosis, sections from mouse hearts were stained with PicroSirius Red. Collagen fibers were visualized using brightfield imaging (Figure 2D) or darkfield imaging with circularly polarized light, which takes advantage of the birefringent properties of collagen fibers, with thin fibers appearing green and more established (thicker) fibers generating a red signal (Figure IVA in the Data Supplement). SNCp treatment led to a significant reduction in perivascular and interstitial fibrosis (Figure 2D through 2F and Figure IVA and IVB in the Data Supplement). This SNCp-dependent reduction in fibrosis was accompanied by a decrease in the serum levels of MMP (matrix metallopeptidase)-2 and MMP-9, which are matrix-remodeling enzymes associated with fibrotic disease progression (Figure 2G).

Fibroblasts isolated from AngII-infused mice also expressed significantly higher levels of genes encoding myofibroblast markers and ECM proteins (Figure 2H and 2I) and exhibited lower levels of quiescent markers, including Tcf21 and Pdgfra (Figure 2H and Figure VA in the Data Supplement),¹⁸ changes that were normalized by SNCp treatment. Gene expression changes indicative of reduced fibroblast activation were confirmed by immunostaining of heart sections with an antibody against Postn (periostin), which was significantly reduced upon SNCp treatment (Figure 2J and 2K). AngII infusion also promoted a CF gene signature indicative of accelerated cell cycle, including alterations in pro-mitotic cell cycle genes (Ccna2, Cdk1), and a cell cycle inhibitor (Cdkn1a), which were partially normalized by SNCp (Figure VB in the Data Supplement). Likewise, immunostaining revealed a reduction in AngII-dependent accumulation of Ki67⁺ nuclei upon SNCp treatment (Figure VC in the Data Supplement), suggesting salinomycin may also inhibit pathological CF proliferation. Of note, expression of senescence-associated secretory phenotype (SASP) and matrifibrocyte markers were also normalized by SNCp administration, furthering the conclusion that salinomycin counteracts the pathological CF phenotype (Figure VIA through VIC in the Data Supplement). Taken together, we conclude that salinomycin blocks cardiac fibrosis in vivo at least partially via direct effects on CF.

Salinomycin Limits Preexisting AngII-Induced Cardiac Fibrosis

We next assessed the efficacy of salinomycin as an interventional drug (interventional salinomycin regimen, called SNCi) by subjecting mice to an AngII-infusion regimen for 7 days before randomly assigning AngII-infused mice to salinomycin or vehicle regimens (1 mg/kg, 3× daily) for an additional 7 days (Figure 3A). The initial 7 days of AngII-infusion induced a significant decline in cardiac function based upon LV ejection fraction and fractional shortening measurements (Figure 3B and 3C), an increase in fibrosis (Figure VIIA and VIIB in the Data Supplement), an increase in cardiomyocyte hypertrophy (Figure VIIC through VIIH), increased Ki67 positivity (Figure VIII and VIIJ in the Data Supplement), and increased Postn staining predominantly in the perivascular region (Figure VIIK and VIIL in the Data Supplement). Importantly, cardiac function recovered in mice that subsequently underwent SNCi treatment from day 7 to 14 - LV ejection fraction was indistinguishable from baseline following 7 days of SNCi treatment (Figure 3B and 3C). Restoration of cardiac function was accompanied by a significant reduction in perivascular fibrosis (Figure 3D through 3F and Figure VIIIA through VIIIC in the Data Supplement) and significantly fewer DAPI⁺ (4’,6-diamidino-2-phenylindole)/vimentin⁺ cells (Figure 3G and 3H). SNCi also reduced Ki67 positivity (Figure 3I and 3J) and Postn accumulation (Figure 3K and 3L), indicating salinomycin inhibits cardiac fibrosis and pathological remodeling even when given as an interventional regimen.

Figure 3. Salinomycin (SNC) attenuates pre-established AngII (angiotensin II)-induced cardiac hypertrophy and fibrosis.

A, Experimental timeline indicating duration of AngII infusion (1 mg/[kg·min]), the interventional SNC regimen (SNCi) start-point (1 mg/kg 3× per day), and echocardiography timepoints. B and C, Left ventricular (LV) ejection fraction (B) and fractional shortening (C) were measured by echocardiography at the indicated timepoints. n=4 (saline), 9 (SNCi). 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to AngII/saline at a given time point. D. LV sections were stained with PicroSirius Red to detect collagen. Scale bar=100 μm. E and F. Quantification of (D) in the perivascular (E) and interstitial (F) regions. n=4 (saline), 9 (SNCi). 1=P<0.05 relative to AngII/saline. G, DAPI⁺/vimentin⁺ cells in LV sections were detected by immunofluorescence. Scale bar=100 μm. H, Quantification of (G) in 3 nonoverlapping fields of view per heart. n=4 (saline), 9 (SNCi). 1=P<0.05 relative to AngII/saline. I, LV sections were stained for DAPI (nuclei) and Ki67 (proliferating cells) WGA (wheat germ agglutinin; lower) revealed cardiac structure. Scale bar=100 μm. J, Quantification of (I) in 3 nonoverlapping fields of view from each animal. n=4 (AngII/saline), 9 (AngII/SNCi). 1=P<0.05 relative to AngII/saline. K, LV sections were stained for DAPI (nuclei), IB4 (isolectin-B4; vasculature), and Postn (periostin) (activated myofibroblasts). WGA (lower panel) revealed cardiac structure. Scale bar=100 μm. L, Quantification of (K) in 3 nonoverlapping fields of view from each animal. n=4 (AngII/saline), 9 (AngII/SNCi). 1=P<0.05 relative to AngII/saline. DAPI indicates 4’,6-diamidino-2-phenylindole; and echo, echocardiography.

Salinomycin Inhibits p38- and Rho-Mediated Signaling in CF

Because salinomycin was identified as an antifibrotic using an HTS and was previously reported to suppress EMT in lung fibroblasts and retinal pigment epithelial cells via p38 inhibition,^13,14 we next asked whether p38 inhibition also underlies the therapeutic efficacy of salinomycin in vivo in the heart. We observed a striking accumulation of phospho-p38 in the nucleus of perivascular and interstitial cells in cardiac sections from mice treated with AngII (Figure 4A), similar to a recent report that p38 localizes to the nucleus of proliferative cancer cells.¹⁹ In contrast, nuclear phospho-p38 was rarely observed in SNCp- or SNCi-treated mice, although some cytoplasmic phospho-p38 remained, particularly in cardiomyocytes. We next examined the impact of salinomycin on TGF-β1- and AngII-dependent fibroblast activation (Figure IXA and IXB in the Data Supplement).²⁰ Salinomycin significantly blunted the acute phosphorylation of p38 by TGF-β1 (Figure 4B and 4C), but did not affect Smad2/3 phosphorylation (Figure 4B and 4D), similar to what is reported in human orbital fibroblasts.^5,13 Immunofluorescence detection of phospho-p38 in cultured neonatal CF confirmed that p38 activity is stimulated by TGF-β1 and AngII (Figure IXC in the Data Supplement); interestingly, we again observed accumulation of phospho-p38 in the nucleus of TGF-β1/AngII-treated CF cultures, and this accumulation shifted to a more cytoplasmic distribution with salinomycin.

Figure 4. Salinomycin (SNC) impedes p38/Rho-dependent myofibroblast activation in vivo and in vitro.

A, Representative images of LV sections that were stained with an antibody against phospho-p38, DAPI, and WGA (wheat germ agglutinin). Nuclear accumulation of phospho-p38 upon AngII (angiotensin II) treatment was blocked by SNCp and SNCi. Scale bar=50 μm. B, cardiac fibroblasts were pretreated with SNC (1 μmol/L) or vehicle before treatment with TGF (transforming growth factor)-β1 (5 ng/mL)/AngII (1 μmol/L). Total and phosphorylated p38 and Smad2/3 were detected by Western blot. C, Quantification of phospho-p38 levels from (B). D, Quantification of phospho-Smad2 levels shown in (B). For (C) and (D), n=4 per condition. 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to the TGF-β1/AngII/saline condition at a given time point. E, Cardiac fibroblasts obtained from five independent heart failure patients were cultured with TGF-β1 (5 ng/mL)/AngII (1 μmol/L)±SNC (1 μmol/L) or Y-27632 (5 μmol/L). Activated Rho was immunoprecipitated with the Rho-binding domain of rhotekin followed by detection of Rho by Western blot in the immunoprecipitate (act) or total cell lysate (tot). ACTA2 and GAPDH levels were evaluated in total cell lysates. F, Quantification of (E). n=3 per condition per patient. 1=P<0.05 relative to vehicle; 2=P<0.05 relative to TGF-β1/AngII/saline; 3=P<0.05 relative to TGF-β1/AngII/Y27632; 4=P<0.05 relative to TGF-β1/AngII/SNC. ACTA2 indicates alpha smooth muscle actin; act, active; DAPI, 4’,6-diamidino-2-phenylindole; p-p38, phoshp-p38; pSmad, phospho-Smad; SNCp or i, salinomycin preventative (p) or interventional (i) regimen; and tot, total.

We also evaluated Rho activation in HFCF by immunoprecipitating active Rho using the Rho-binding domain of rhotekin. We found that salinomycin pretreatment significantly inhibited Rho activation (5/5 isolates) and the induction of ACTA2 expression (3/5 isolates) in response to TGF-β1/AngII (Figure 4E and 4F). In contrast, Y27632, a ROCK inhibitor that does not impede Rho activation but prevents signaling downstream of active Rho, inhibited the induction of ACTA2 in 2/5 isolates. Taken together, our HTS screen identified a potent antifibrotic compound that ameliorates myofibroblast activation through both noncanonical TGF-β1/p38 and Rho-dependent signaling pathways.

Salinomycin Inhibits Hypertrophic Remodeling

Because SNCp blocked AngII-dependent fibrosis and deterioration in LV function, we hypothesized that SNCp might limit cardiac hypertrophy induced by AngII infusion. Although the LV mass of mice treated with salinomycin alone was indistinguishable from saline-infused control mice, SNCp prevented AngII-induced cardiac hypertrophy as assessed by echocardiography (Figure 5A). Consistent with this observation, the increase in heart weight to tibia length ratio observed in AngII-infused animals was blocked by SNCp (Figure 5B). AngII-induced hypertrophy was confirmed at the cellular level by measuring trans-verse cardiomyocyte cross-sectional area and longitudinal cardiomyocyte length-width ratio in sections stained with WGA (wheat germ agglutinin), revealing that AngII-induced cardiomyocyte growth was prevented by SNCp treatment (Figure 5C and 5D and Figure XA through XD in the Data Supplement). The induction of the fetal gene program by AngII, indicative of hypertrophic remodeling, was also blocked by SNCp (Figure XE in the Data Supplement).

Figure 5. Salinomycin (SNC) ameliorates AngII (angiotensin II)-induced concentric hypertrophy.

A, Left ventricular (LV) mass was estimated by echocardiography in the 28-d preventative experiment. n=6 (control, saline), 6 (AngII, saline), 9 (control, SNCp), 9 (AngII, SNCp). 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to saline/saline condition at a given time point; 3=P<0.05 relative to saline/SNC condition at a given time point; 4=P<0.05 relative to AngII/SNC condition at a given time point. B, Cardiac hypertrophy was evaluated by measuring the heart weight to tibia length ratio (HW/TL). n=6 (control, saline), 6 (AngII, saline), 9 (control, SNCp), 9 (AngII, SNCp). 1=P<0.05 relative to saline/saline; 2=P<0.05 relative to AngII/saline; 3=P<0.05 relative to saline/SNC; 4=P<0.05 relative to AngII/SNC. C, LV sections were stained with WGA (wheat germ agglutinin) to visualize myocyte cross-sectional area (CSA). Scale bar=100 μmol/L. D, Quantification of myocyte CSA from (C). At least 250 cardiomyocytes (CM) per animal were measured in 3 nonoverlapping fields of view. n=3 (control, saline), 3 (AngII, saline), 5 (control, SNCp), 5 (AngII, SNCp). 1=P<0.05 relative to saline/saline; 2=P<0.05 relative to AngII/saline; 3=P<0.05 relative to saline/SNC; 4=P<0.05 relative to AngII/SNC. E, LV mass was estimated by echocardiography in the 14-d interventional experiment. n=4 (saline), 9 (SNCi). 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to AngII/saline at a given time point. F, LV sections were stained with WGA to visualize myocyte CSA. Scale bar=100 μm. G, Quantification of CSA in (F) from at least 250 total myocytes across no fewer than 3 nonoverlapping fields of view. n=4 (saline), 9 (SNCi). H, Cardiac hypertrophy was evaluated by measuring the HW/TL. n=4 (AngII, saline), 9 (AngII, SNCi). I, Serum-starved neonatal CM were pretreated with saline or SNC (5 μmol/L) before TGF (transforming growth factor)-β1/AngII treatment. After 24 h cells were fixed and stained for actinin 1. J, Quantification of normalized actinin 1⁺ CM area in (I). n=9 wells per condition. 1=P<0.05 relative to 0.5% serum control; 2=P<0.05 relative to TGF-β1/AngII; 3=P<0.05 relative to TGF-β1/AngII/SNC. K. Total and phosphorylated ERK (extracellular signal-regulated kinase) 1/2 were detected by Western blot. L. Quantification of phospho-ERK1/2 (pERK) levels shown in (K). n=4 experiments. 1=P<0.05 relative to time 0 for a given condition; 2=P<0.05 relative to TGF-β1/AngII at a given time point. DAPI indicates 4’,6-diamidino-2-phenylindole; Hist, histology; SNCp and i, salinomycin preventative (p) or interventional (i) regimen; and T/A, TGF-β and AngII treatment.

Importantly, the increase in LV mass observed by echocardiography after 7 days of AngII treatment was partially normalized by SNCi between days 7 and 14 (Figure 5E). Likewise, a trend toward smaller cardiomyocyte cross-sectional area (Figure 5F and 5G) and a significant normalization in cardiomyocyte length-width ratio (Figure XF through XI in the Data Supplement) was also apparent after only 7 days of interventional SNCi treatment, corresponding to a trend in decrease in heart weight to tibia length ratio (Figure 5H). Normalization of chamber dimensions in both SNCp- and SNCi-injected mice was also observed by echocardiography (Figure XI in the Data Supplement), indicating salinomycin not only prevents maladaptive cardiac remodeling in response to AngII infusion but remarkably also normalizes preestablished hypertrophic remodeling when used as an interventional drug.

To evaluate whether these findings might represent a direct effect on cardiomyocytes, neonatal cardiomyocytes were harvested from C57Bl6/J mice and serum-starved in the presence or absence of salinomycin, then stimulated with TGF-β1/AngII to induce hypertrophy. TGF-β1/AngII induces a significant degree of hypertrophy in isolated cardiomyocytes as assessed by quantifying the area of Actn1 (alpha actinin)⁺ cells, which is significantly reduced by salinomycin treatment (Figure 5I and 5J), suggesting salinomycin may in part act directly on cardiomyocytes. Indeed, immunoblot revealed that TGF-β1/AngII -dependent ERK1/2 phosphorylation (a marker for concentric hypertrophy in cardiomyocytes) is suppressed by salinomycin (Figure 5K and 5L).²¹ We also found that TGF-β1/AngII-dependent nuclear accumulation of p65, which induces NFATc1 (nuclear factor of activated T cells 1)-mediated hypertrophy,²² was significant reduction by salinomycin pretreatment (Figure XIIA and XIIB in the Data Supplement). To rule out the possibility that salinomycin treatment might lead to cardiomyocyte apoptosis, we also stained representative sections with TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) to visualize double-stranded DNA breaks, observing negligible apoptosis at 28 days as a result of SNCp treatment (Figure XIIC and XIID in the Data Supplement). Taken together, these results indicate that salinomycin modulates the cardiomyocyte response to hypertrophic stimuli.

Salinomycin Promotes Expression of Angiogenic Factors In Vivo.

We showed that salinomycin potently blocked the AngII-dependent induction of genes encoding proinflammatory cytokines associated with pathological remodeling and SASP (Figure VIB in the Data Supplement). This inhibition was accompanied by an increase in serum levels of certain angiogenic and protective cytokines such as Ccn4 (cellular communication network factor 4, or Wisp1), Cxcl5, and Cxcl13^23–25 (Figure XIIIA in the Data Supplement). Furthermore, salinomycin ameliorated the AngII-induced increase in the serum levels of inflammatory cytokines, such as Ccl (C-C motif chemokine ligand) 12 and Ccl22, that have been shown to have deleterious effects on the heart,^26,27 suggesting that salinomycin promotes an antifibrotic, proangiogenic secretion profile. To assess potential effects of this altered secretory profile, we stained sections with an ERG (ETS transcription factor ERG) antibody and IB4 (isolectin-B4) to mark vascular endothelial cells (EC), revealing a reduction in ERG⁺ nuclei, which was partially restored by SNCp treatment (Figure XIIIB and XIIIC in the Data Supplement). SNCi treatment also increased the number of ERG⁺ nuclei (Figure XIIID and XIIIE in the Data Supplement), suggesting salinomycin might also increase the density of the vascular network via angiogenic factors in nonischemic remodeling.

Salinomycin Dramatically Improves Outcomes in Ischemic HF

We next evaluated whether salinomycin could blunt fibrotic scar formation and cardiac dysfunction following MI induced by surgical ligation of the left anterior descending coronary artery. As shown in Figure 6A, mice were randomly assigned to (1) a saline-treated control; (2) an interventional regimen initiated 3 days post-MI after the peak of inflammation (SNCi); or (3) a preventative regimen initiated the day before MI (SNCp). Infarction was confirmed by acute S-T elevation based on ECG measurements (Figure XIVA in the Data Supplement). Salinomycin treatment was associated with improvement in 14-day survival (Figure XIVB in the Data Supplement). SNCp treatment improved LV ejection fraction and completely blocked the increase in systolic volume observed at 3 and 14 days post-MI (Figure 6B and 6C and Figure XIVC through XIVE in the Data Supplement). SNCi treatment also blunted the increase in systolic volume occurring between 3 and 14 days post-MI relative to saline treatment, leading to a partial recovery in LV ejection fraction.

Figure 6. Salinomycin (SNC) limits expansion of left ventricular (LV) scarring and cardiac dilation in a mouse model of myocardial infarction (MI).

A, Experimental timeline for MI surgeries, start points of interventional (i) and preventative (p) SNC regimens (1 mg/kg 3× per day), and timepoints for echocardiography and sample isolation. B and C. LV ejection fraction (B) and systolic volume (C) were calculated at indicated timepoints by echocardiography. n=8 (saline), 9 (SNCi), 9 (SNCp). 1=P<0.05 relative to 0 d for a given condition; 2=P<0.05 relative to 3 d at a given condition, 3=P<0.05 relative to MI-saline at a given time point, 4=P<0.05 relative to MI-SNCi at a given time point. D. Serial sections of hearts isolated from the indicated treatment groups 14 d post-MI were stained with Picrosirius Red to label scar (magenta). Scale bar=1 μm. E, Quantification of scar area from 10 levels from the apex to the ligature shown in (D). n=8 (saline), 9 (SNCi), 9 (SNCp). 1=P<0.05 relative to MI-saline, 2=P<0.05 relative to SNCi. F, Picrosirius Red stained LV sections in the remote zone were visualized in polarized light. Scale bar=100 μm. G, Quantification of (F) n=8 (saline), 9 (SNCi), 9 (SNCp). 1=P<0.05 relative to sham, 2=P<0.05 relative to MI-saline, 3=P<0.05 relative to MI-SNCi. H, LV sections from the 14-d MI experiment were stained for DAPI (nuclei), IB4 (isolectin-B4; vasculature), and Postn (periostin) (activated myofibroblasts). WGA (wheat germ agglutinin; lower) revealed cardiac structure. Scale bar=100 μm. I. Quantification of (H) in 3 nonoverlapping fields of view from the LV from each animal was measured. n=8 (MI/saline), 9 (MI/SNCi), 9 (MI/SNCp). 1=P<0.05 relative to MI-saline, 2=P<0.05 relative to SNCi. echo indicates echocardiography; histo, histology; Postn, periostin; and SNCp and i, salinomycin preventative (p) or interventional (i) regimen.

PicroSirius Red staining revealed a profound reduction in scar area upon SNCp treatment, whereas SNCi treatment led to a partial reduction in scar (Figure 6D and 6E). Imaging of PicroSirius Red staining in circularly polarized light revealed a significant reduction in remote zone fibrosis even in SNCi-treated animals compared with saline-treated animals, indicating fibrosis expansion into healthy tissue is blocked starting at the initiation of salinomycin administration (Figure 6F and 6G). Reduced fibrosis was accompanied by a significant reduction in vimentin⁺ mesenchymal cells in the remote zone of SNCpand SNCi-treated mice 14 days post-MI, compared with saline-treated mice (Figure XIVF through XIVI in the Data Supplement). We also found that SNCi/SNCp treatment reduced the abundance of Postn+ cells (Figure 6H and 6I) and Ki67 positivity (Figure XIVJ and XIVK in the Data Supplement), confirming salinomycin blocks fibroblast activation in the MI model.

We next assessed the effects of salinomycin treatment on the ischemic myocardium itself by staining for 2, 3, 5-triphenyltetrazolium chloride. Mice were given saline or salinomycin for 3 days before MI and hearts were harvested 24 hours post-MI for 2, 3, 5-triphenyltetrazolium chloride staining; saline-treated animals exhibited nearly twice as much acute myocardial death (white tissue) than salinomycin-treated animals (Figure 7A and 7B). Likewise, TUNEL staining revealed that SNCi and SNCp treatment reduced the number of apoptotic cells present at 14 days post-MI (Figure 7C and 7D). On a molecular level, salinomycin facilitated phosphorylation of Akt even upon TGF-β1/AngII treatment (Figure 7E and 7F). Interestingly, we also observed improved vascularity in SNCp- and SNCi-treated mice as indicated by the density of ERG⁺/isolectin-B4 positive cells in the ischemic and remote zones (Figure XVA through XVD in the Data Supplement). These results suggest that salinomycin may limit infarct expansion in part by promoting myocardial survival and improving vascularity in the border zone. Taken together, our study establishes salinomycin as an antifibrotic compound that improves cardiac physiology following ischemic and nonischemic cardiac insult.

Figure 7. Impact of salinomycin (SNC) on postmyocardial infarction (MI) ischemia and cell death.

A, Mice were pretreated with SNC 3 d before MI. Hearts were isolated 24 h after MI for 2, 3, 5-triphenyltetrazolium chloride (TTC) staining to differentiate intact myocardium (red) from infarcted myocardium (white). B, Quantification of A. n=5 (saline), 6 (SNC). 1=P<0.05 relative to saline. C, Left ventricular sections were stained with TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) and DAPI to visualize apoptotic nuclei. Scale bar=100 μm. D, Quantification of TUNEL⁺ nuclei in (C) normalized to total nuclei. n=8 (saline), 9 (SNCi), 9 (SNCp). 1=P<0.05 relative to MI-saline, 2=P<0.05 relative to MI-SNCi. E, Neonatal cardiomyocytes were treated with SNC followed by stimulation with TGF (transforming growth factor)-β1/AngII (angiotensin II). Phospho- and total Akt were detected by Western blot. F, Quantification of phospho-Akt levels shown in (l). n=4 experiments. 1=P<0.05 relative to the 0 time point for a given condition; 2=P<0.05 relative to TGF-β1/AngII at a given time point. BZ indicates border zone; DAPI, 4’,6-diamidino-2-phenylindole; pAkt, phospho-Akt; SNCp and i, salinomycin preventative (p) or interventional (i) regimen; and RZ, remote zone.

DISCUSSION

Here, we report the findings of an HTS that identified salinomycin as an effective anti-fibrotic drug. We report positive outcomes of preventative and interventional salinomycin treatment regimens in preclinical mouse models of nonischemic and ischemic heart disease. Importantly, salinomycin ameliorates preestablished fibrosis and concentric hypertrophy in an AngII infusion mouse model of hypertension-induced cardiac remodeling and prevented the progression of fibrotic scarring and cardiac dilation following MI. This protection was traced to direct effects on both CF and cardiomyocytes and potentially through limiting SASP-dependent inflammation and promoting angiogenesis. Thus, salinomycin emerges as an attractive candidate for clinical development.

Fibroblasts, the primary source of ECM deposition in the heart, display phenotypic plasticity in response to pathophysiological signaling. The gene expression changes in CF upon treatment with salinomycin indicate a less proliferative and less activated phenotype. Here, we find that salinomycin interferes with stress-dependent Rho-ROCK signaling, ultimately ameliorating ECM deposition and the development of pathological fibrosis. Salinomycin also blocks acute p38/MAPK activation in response to TGF-β1/AngII stimulation of CF in vitro, consistent with a previous report.¹³ p38/MAPK activation is transduced by MKK (mitogen-activated protein kinase kinase) 3/6 and has been implicated in EC inflammatory signaling and pathological HF progression.^5,28 Of note, we observe robust accumulation of phosphorylated p38 in the nucleus of pathological CF both in vitro and in vivo, which is prevented by salinomycin treatment. Nuclear accumulation of phospho-p38 in cancer cells is associated with inflammatory signaling¹⁹ and inhibition of inflammatory and profibrotic gene transcription with BET (bromodomain and extra-terminal) bromodomain inhibition shows promise in preclinical models of HF.²⁹ Based on our findings, it is likely that salinomycin prevents maladaptive cardiac remodeling at least partially via inhibition of stress-responsive proinflammatory and profibrotic gene networks in CF.

The magnitude of cardioprotection afforded by salinomycin following MI indicates a more complex mechanism than simply blocking p38/MAPK-dependent myofibroblast activation. Preventative treatment with salinomycin resulted in near-complete protection from functional deterioration, and interventional treatment beginning 3 days post-MI prevented scar expansion and further loss of function. We found in the nonischemic models and in HFCF that salinomycin treatment altered the expression and secretion of cytokines that may impact the inflammatory response, an important aspect of post-MI remodeling. We also found the infarct zone to be more highly vascularized with ERG⁺ ECs in salinomycin-treated animals. Although this may simply result from a scar that is more easily penetrated by ECs, it may also reflect a more angiogenic milieu in salinomycin-treated animals; indeed, we detected proangiogenic and reparative cytokines at higher levels in the serum of mice treated with salinomycin. Salinomycin treatment may also suppress EMT and maintain EC identity, as was previously reported in other cell types.¹⁴

Salinomycin has a history of preclinical studies as a cancer therapeutic, particularly for overcoming multidrug resistance of cancer stem cells.³⁰ Salinomycin interferes with Wnt signaling, impedes the development of stem cell–like properties, and induces cell death via p53- and caspase-independent pathways.^31–34 Although salinomycin reduced proliferation in vitro and in vivo, salinomycin did not cause apoptosis in any cell type; conversely, it appears to have a strong antiapoptotic effect in cardiomyocytes in the ischemic mouse model. As dysregulation in p53 is also associated with senescence, it is interesting to speculate that salinomycin may directly interfere with SASP-related fibrotic and inflammatory programs.^35,36 Indeed, CF reportedly transition into a specialized senescent-like state called a matrifibrocyte after MI.^36,37 In vivo, salinomycin alters the expression of many SASP-related genes in a manner consistent with recently developed senomodulatory agents that prevent age-associated disorders^38,39; further studies are certainly warranted to investigate whether salinomycin may impact SASP in pathological cardiac fibrosis.

To this end, one senescence-associated gene that was normalized by salinomycin treatment was Pgc1a, a transcriptional coactivator that is primarily implicated in metabolism and mitochondrial reactive oxygen species detoxification. Cancer literature has shown roles for salinomycin in altering autophagy (and specifically mitophagy) as well as mitochondrial polarity.⁴⁰ Thus, future studies should focus on the potential that salinomycin blocks senescence and mitochondrial dysfunction, preventing the DNA damage responses that accelerate cardiac dysfunction.

Taken together, our study provides compelling evidence that salinomycin prevents cardiac fibrosis in ischemic and nonischemic heart disease. Importantly, salinomycin has an extremely high affinity for plasma proteins, a low predicted threshold for drug-drug interactions with other CYP3A4 targets, and a high clearance rate in mice and humans.¹⁷ Salinomycin appears to have limited side effects, particularly if delivered continuously at a low dose. The observation that narasin and monensin A also block myofibroblast activation indicates salinomycin-related polyether ionophore analogs should be carefully evaluated as antifibrotic agents for use in conditions associated with unrestrained scarring such as HF.

Novelty and Significance.

What Is Known?

• Heart disease is invariably associated with fibrosis, which stiffens the heart and can lead to diastolic dysfunction.

• Resident cardiac fibroblasts deposit extracellular matrix that develops into a fibrotic scar.

• There are only limited therapeutic strategies to prevent cardiac fibroblast activation and cardiac fibrosis.

What New Information Does This Article Contribute?

• The polyether ionophore salinomycin prevents fibroblast activation and the secretion of extracellular matrix by blocking Rho-kinase and p38/MAPK (mitogen activated protein kinase)-dependent signaling in cardiac fibroblasts.

• Salinomycin prevents scar expansion and improves cardiac function in a mouse model of major myocardial infarction.

• Salinomycin reverses preestablished hypertension-induced fibrosis and decline in cardiac function in AngII (angiotensin II)-infused mice.

Heart disease is invariably associated with the deposition of ECM (extracellular matrix) by resident cardiac fibroblasts. Excessive or inappropriate deposition of ECM leads to the formation of a fibrotic scar that impedes cardiac contraction and relaxation, and can initiate lethal arrhythmias. This study establishes the polyether ionophore salinomycin as a potent anti-fibrotic small molecule that prevents fibroblast activation and cardiac fibrosis. Salinomycin acts by inhibiting TGF (transforming growth factor)-β1 dependent p38/MAPK and Rho-kinase signaling in cardiac fibroblasts. The anti-fibrotic activity of salinomycin prevents scar formation and functional decline in preclinical mouse models of ischemic and nonischemic pathological cardiac remodeling. Importantly, interventional administration of salinomycin reverses preexisting AngII-dependent ventricular remodeling and blocks further scar expansion and functional decline when given 3 days after a major myocardial infarction. This study suggests that polyether ionophore analogs related to salinomycin should be further investigated for use in conditions associated with unrestrained scarring such as heart failure.

Nonstandard Abbreviations and Acronyms:

AngII

angiotensin II

EC

endothelial cell

ECM

extracellular matrix

ERK

extracellular signal-regulated kinase

HF

heart failure

HFCF

heart failure cardiac fibroblasts

HTS

high-throughput screen

IB4

isolectin-B4

LV

left ventricle

MI

myocardial infarction

ROCK

Rho-kinase

SASP

senescence-associated secretory phenotype

TSP

thrombospondin

VEGF

vascular endothelial growth factor

WGA

wheat germ agglutinin