Disulfiram Reduces Atherosclerosis and Enhances Efferocytosis, Autophagy, and Atheroprotective Gut Microbiota in Hyperlipidemic Mice

Abstract

Background

Pyroptosis executor GsdmD (gasdermin D) promotes atherosclerosis in mice and humans. Disulfiram was recently shown to potently inhibit GsdmD, but the in vivo efficacy and mechanism of disulfiram's antiatherosclerotic activity is yet to be explored.

Methods and Results

We used human/mouse macrophages, endothelial cells, and smooth muscle cells and a hyperlipidemic mouse model of atherosclerosis to determine disulfiram antiatherosclerotic efficacy and mechanism. The effects of disulfiram on several atheroprotective pathways such as autophagy, efferocytosis, phagocytosis, and gut microbiota were determined. Atomic force microscopy was used to determine the effects of disulfiram on the biophysical properties of the plasma membrane of macrophages. Disulfiram‐fed hyperlipidemic apolipoprotein E^−/− mice showed significantly reduced interleukin‐1β release upon in vivo Nlrp3 (NLR family pyrin domain containing 3) inflammasome activation. Disulfiram‐fed mice showed smaller atherosclerotic lesions (~27% and 29% reduction in males and females, respectively) and necrotic core areas (~50% and 46% reduction in males and females, respectively). Disulfiram induced autophagy in macrophages, smooth muscle cells, endothelial cells, hepatocytes/liver, and atherosclerotic plaques. Disulfiram modulated other atheroprotective pathways (eg, efferocytosis, phagocytosis) and gut microbiota. Disulfiram‐treated macrophages showed enhanced phagocytosis/efferocytosis, with the mechanism being a marked increase in cell‐surface expression of efferocytic receptor MerTK. Atomic force microscopy analysis revealed altered biophysical properties of disulfiram‐treated macrophages, showing increased order‐state of plasma membrane and increased adhesion strength. Furthermore, 16sRNA sequencing of disulfiram‐fed hyperlipidemic mice showed highly significant enrichment in atheroprotective gut microbiota Akkermansia and a reduction in atherogenic Romboutsia species.

Conclusions

Taken together, our data show that disulfiram can simultaneously modulate several atheroprotective pathways in a GsdmD‐dependent as well as GsdmD‐independent manner.

Nonstandard Abbreviations and Acronyms

ASC

apoptosis‐associated speck‐like protein containing a caspase recruitment domain

GsdmD

gasdermin D

NETs

neutrophil extracellular traps

Nlrp3

NLR family pyrin domain containing 3

Research Perspective.

What Is New?

• We show that blocking GsdmD (gasdermin D) in mice with the Food and Drug Administration‐approved antialcoholism drug disulfiram effectively reduces inflammasome activity and development of aortic root atherosclerotic plaques.

• Disulfiram induces several atheroprotective pathways in vitro and in vivo, such as autophagy, efferocytosis, and enrichment of atheroprotective gut microbiota.

What Question Should Be Addressed Next?

• Because disulfiram induced autophagy in myeloid, endothelial, and smooth muscle cells, the tissue‐specific effect of disulfiram on atherosclerosis should be assessed, and as disulfiram is upstream of pyroptotic cell death and IL‐1β (interleukin‐1 beta) release, its anticardiovascular disease effects should be compared with canakinumab (IL‐1β antibody).

Atherosclerosis, a sterile and chronic inflammatory disease, is the major cause of cardiovascular disease (CVD)‐related mortalities. ¹ , ² The complex nature of atherosclerosis is highlighted by the involvement of multiple pathways such as cholesterol efflux, Nlrp3 (NLR family pyrin domain containing 3)/AIM2 (absent in melanoma 2) inflammasome, neutrophil extracellular traps (NETosis), autophagy, efferocytosis, and gut microbiota. ¹ , ³ , ⁴ , ⁵ , ⁶ Cumulative defects or dysregulation of these pathways over time promotes the progression of atherosclerosis. Thus, tackling CVD with cholesterol‐lowering therapeutics, such as statins, alone may not be enough to reverse the trend of increasing prevalence of CVD. New CVD therapeutics that can be used in conjunction with cholesterol‐lowering therapies are needed to target multiple pathways simultaneously. Human trials showed that IL (interleukin)‐1β might serve as 1 of the targets for adjuvant therapy for CVD. ⁷ , ⁸ , ⁹ IL‐1β acts as a local vascular and as a systemic contributor in the pathogenesis of CVD. ¹ , ⁸ , ⁹ , ¹⁰ The generation of biologically active IL‐1β in atherosclerotic plaque area depends on activation of the Nlrp3 inflammasome. ¹¹ , ¹² , ¹³ More recently, the AIM2 inflammasome was also identified as an age‐related risk factor for CVD, ¹⁴ pointing toward a sustained role of inflammasomes in exacerbating atherosclerosis in humans. Activation of inflammasomes leads to cleavage of pore‐forming protein GsdmD (gasdermin D) in atherosclerotic plaques, allowing local as well as systemic release of proinflammatory cytokines, which further exacerbates disease via recruitment of new immune cells to plaque areas. ¹⁵ , ¹⁶ , ¹⁷ Recent studies highlight the atherogenic role of GsdmD, ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ with GsdmD mRNA being upregulated in peripheral blood monocytes of patients with coronary artery disease, and the presence of cleaved N‐terminal‐GsdmD fragment in mouse and human atherosclerotic plaques, ¹⁵ , ¹⁶ , ¹⁷ but the mechanism of atherogenic activity of GsdmD is still not fully resolved. GsdmD can be cleaved via capase‐1/Nlrp3 dependent canonical inflammasome pathway, or via a caspase‐11 (in mice) or caspase‐4/5 (in humans) dependent noncanonical inflammasome pathway. The N‐terminal‐GsdmD fragment generated via a noncanonical pathway can in turn activate the canonical inflammasome in a positive feedback loop. ¹⁹ GsdmD thus serves as a common executor of both canonical and noncanonical pyroptosis. ²⁰ , ²¹ , ²² , ²³

Targeting GsdmD seems to be a better option than targeting IL‐1β alone, as GsdmD mediates highly inflammatory pyroptosis, which is upstream of IL‐1β release. This leads to a large increase in danger‐associated molecular patterns molecules that can further amplify inflammation in atherosclerotic plaques. Blocking GsdmD can reduce the levels of not only IL‐1β but also of IL‐18 and other danger‐associated molecular patterns. Furthermore, studies have shown the role of GsdmD in the formation of neutrophils extracellular traps (NETs), ³ , ²⁴ , ²⁵ thus blocking GsdmD can also reduce atherogenic NETosis. In contrast to GsdmD/IL‐1β atherogenic role, autophagy and efferocytosis serve as major atheroprotective pathways. Defective autophagy was observed in advanced human and mouse atherosclerotic lesions, and upregulation of autophagy was shown to protect against CVD. ⁴ , ²⁶ , ²⁷ , ²⁸ , ²⁹ Efficient efferocytosis of damaged cholesterol‐laden foam cells is 1 of the strategies used by the host immune system to promote the regression of atherosclerotic plaques and prevent CVD. ⁵ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ In addition to intracellular atheroprotective pathways, the gut microbiota and associated metabolites can modulate the host's risk of developing CVD, independent of traditional risk factors. ³⁵ One striking example is the gut microbiota‐mediated conversion of dietary choline into atherogenic trimethyl amine oxide, which promotes atherosclerosis in humans. ⁶ , ³⁶ , ³⁷ , ³⁸ Thus, the next generation of statin‐adjuvant CVD therapeutics are expected to target inflammasomes, GsdmD, NETosis autophagy, efferocytosis, and gut microbiota.

The US Food and Drug Administration‐approved drug disulfiram was shown to potently inhibit GsdmD pore formation and block formation of NETs, ²⁵ , ³⁹ but its antiatherosclerotic effect and mechanism are not clear. Here, we report that disulfiram reduces the atherosclerotic burden and necrotic core area in plaques in a hyperlipidemic mouse model of atherosclerosis. Disulfiram causes plasma membrane remodeling and induces antiatherosclerotic pathways such as autophagy and efferocytosis. Furthermore, disulfiram promotes enrichment of atheroprotective gut microbiota and reduction in atherogenic bacterial species in hyperlipidemic mouse model of atherosclerosis.

METHODS

The authors declare that all supporting data are available within the article and its online supplementary files.

Cell Culture

Cell culture and treatment conditions for RAW‐ASC, RAW‐ASC‐GsdmD^−/−, THP‐1, THP‐ASC‐GFP, Jurkat, HepG2, human aortic endothelial cells and mouse aortic smooth muscle cells are described in Data S1. Bone marrow–derived macrophages were isolated and cultured as described earlier. ¹⁵

In Vitro and In Vivo Inflammasome Assembly

In vivo Nlrp3 inflammasome assembly, in vitro Nlrp3 inflammasome assembly, and IL‐1β release assays were performed as described earlier. ¹⁵ Detailed methods are described in Data S1.

Viability Assay

Viability of cells was measured by Live/Dead Fixable Blue Stain kit, with detailed method in Data S1.

Cytotoxicity Assay

Toxicity of disulfiram on cells was measured by the CyQuant LDH Cytotoxicity Assay kit (ThermoFisher), with detailed method in Data S1.

Mice and Diets

Animal experiments were performed in accordance with approved protocols from the Cleveland State University and the Cleveland Clinic Institutional Animal Care and Use Committees. The C57BL6J‐wild type mice were purchased from the Jackson Laboratories and the C57BL/6J‐GsdmD^−/− mice were generated earlier ⁴⁰ and kindly provided by Dr Russell Vance (University of California, Berkeley). C57BL/6J‐apoE^−/− (apolipoprotein E) mice were purchased from Jackson Laboratories and bred in house. Mice were maintained in a temperature‐controlled facility with the standard 12‐hour light/dark cycle.

Atherosclerosis Studies

To promote hyperlipidemia, mice were weaned onto a Western‐type diet (WTD, TD.88137) containing 42% calories from fat as described by others. ⁴¹ , ⁴² , ⁴³ Disulfiram (Sigma; #PHR 1690) was milled into WTD at 100 mg/kg per day by Envigo, as described previously. ⁴⁴

Atherosclerotic Lesion and Necrotic Core Microscopy and Quantification

Mice were euthanized by CO₂ asphyxiation 15 weeks post diet start. Whole blood was collected from the retro‐orbital plexus via heparinized capillaries and mixed with EDTA. The mice were then perfused with 10 mL PBS followed by harvesting. Some of the hearts were directly embedded into optimal cutting temperature solution (OCT) (fresh sections) or fixed in 10% phosphate buffered formalin first and embedded into OCT (fixed sections). Hearts were sectioned into 5 μm sections (fresh hearts) or 10 μm sections (fixed hearts) using Leica CM3050 S Cryostat. Quantitative assessment of necrotic cores was determined by acellular regions in plaques, and quantitative assessment of atherosclerotic lesions was performed oil Red O and hematoxylin as described previously, ⁴⁵ , ⁴⁶ and using the Olympus CX43 Microscope and cellSens software.

Autophagy Assay

RAW‐ASC, RAW‐ASC‐GsdmD^−/−, RAW Difluo mLC3, HepG2, THP‐1 macrophages, human aortic endothelial cells, and primary mouse aortic smooth muscle cells were treated ±disulfiram ranging from 0.5 to 10 μmol/L, or ±25 μmol/L rapamycin (Invivogen; # tlrl‐rap), or ±20 μmol/L chloroquine (Sigma # C6628) for 2 hours. Autophagy markers were probed by microscopy or western blot analysis. Detailed methods are described in Data S1.

Western Blotting

Western blot analysis was performed on protein extracts±various treatment conditions. Detailed methods are described in Data S1.

Indirect Immunofluorescence

THP‐1 macrophages were treated with indicated dose of disulfiram and probed for MerTK expression. Detailed methods are described in Data S1.

Efferocytosis and Phagocytosis Assay

THP‐1 macrophages were grown in chamber slides (ibidi) and treated ±disulfiram. Jurkat cells (ATCC) were labeled with 1 μmol/L calcein for 1 hour and washed with PBS 3 times. Calcein‐labeled cells were treated with 1 μmol/L of staurosporine for 16 to 20 hours to induce apoptosis. After 16 hours of treatment, apoptosis was confirmed by looking at the membrane blebbing of the cells under confocal microscope. Apoptotic Jurkat cells were incubated with THP‐1 macrophages in the ratio 1:1 for 4 hours. After 4 hours, cells were washed with 3 times with PBS for 5 minutes each. Cells were fixed using 3.7% paraformaldehyde for 30 minutes, washed with PBS, followed by imaging by confocal microscopy. For phagocytosis assay, the THP‐1 macrophages were incubated with Latex beads‐rabbit‐IgG‐FITC complex (Cayman Chemical # 500290) with a final dilution of 1:200 at 37 °C for 1 hour. Cells were washed with 3 times with PBS for 5 minutes. Cells were fixed using 3.7% paraformaldehyde for 30 minutes, washed with PBS, followed by imaging by confocal microscopy using FITC and DIC channel. Efferocytosis and phagocytosis was determined by measuring mean fluorescence intensity of greater than 30 cells per treatment group.

Atomic Force Microscopy

THP‐1 macrophages treated with ±5 μmol/L disulfiram for 2 hours were subjected to atomic force microscopy analysis as previously described. ⁴⁷ In brief, a tipless cantilever glued with a 4.5 μm bead was submerged into cell culture where the approach/retraction velocity was 5 μm/s, exerting an indented force of 2 nN onto cells. Young's moduli (modulus of elasticity or stiffness) of the cells were determined using a Hertz model, ⁴⁸ whereas adhesion, defined as the force required to separate cell surface and cantilever tip, was calculated directly from the force‐indentation curves as detailed elsewhere. ⁴⁹ , ⁵⁰

Immunofluorescence of Aortic Sinus

Fresh hearts embedded in OCT were cryosectioned (5 μm) using Leica CM3050 S Cryostat. Sections were fixed in 4% phosphate‐buffered formalin for 10 minutes at room temperature. After PBS wash, sections were incubated with BLOXALL Blocking Solution (Vector Labs; SP‐6000) for 10 minutes, washed, and then blocked with Animal‐Free Blocker and Diluent (Vector Labs; SP‐5035) for 1 hour at room temperature to block nonspecific binding. Sections were then incubated overnight with ±primary antibody at 1:50 dilution, washed, and then incubated with 1:150 dilution of Alexa‐488 goat antirabbit secondary (Invitrogen; #A11034) for 1 hour at room temperature. Slides were then washed and mounted with mounting media containing DAPI (Molecular Probes; #36964) and examined using the Nikon Eclipse Ti confocal and Nikon NIS Elements Imaging Software version 4.13.

Gut Microbiota Sequencing

The fresh fecal samples were collected from mice using sterile forceps and then microbial DNA was extracted using the DNeasy PowerSoil ProKit (Qiagen; # 47016), following the manufacturer's instructions. Using methods explained earlier 16S rRNA gene amplicon sequencing and bioinformatics analysis were performed. ⁵¹ , ⁵² , ⁵³ Briefly, raw 16S amplicon sequence and metadata were demultiplexed using split_libraries_fastq.py script implemented in QIIME2. ⁵⁴ Individual fastq files without nonbiological nucleotides were processed using divisive amplicon denoising algorithm pipeline. ⁵⁵ The output of the dada2 pipeline (feature table of amplicon sequence variants) was processed for alpha and beta diversity analysis using phyloseq ⁵⁶ and microbiomeSeq packages in R.

Statistical Analysis

Statistics were performed using GraphPad Prism 9. Data are represented as mean±SD or mean±SEM for in vitro and in vivo studies, with at least replicates performed for each experiment. Two‐tailed t tests were performed for comparisons between 2 groups with normally distributed (parametric) data, or by 2‐tailed Mann–Whitney U test for nonparametric data. For microbiome data, differential abundance analysis was performed using the random‐forest algorithm, implemented in the DAtest package. We assessed the statistical significance (P<0.05) throughout, and whenever necessary, we adjusted P values for multiple comparisons according to the Benjamini and Hochberg method to control false discovery rate. ⁵⁷ Linear regression (parametric test), and Wilcoxon (nonparametric) test were performed on genera and amplicon sequence variants abundances against metadata variables using their base functions in R (version 4.1.2). ⁵⁸

RESULTS

Disulfiram Inhibits In Vitro and In Vivo Nlrp3 Inflammasome Activity

GsdmD, the final executor of inflammasome activity, promotes the progression of atherosclerosis, with cleaved N‐terminal GsdmD found in humans and mice atherosclerotic plaques. ¹⁵ , ¹⁶ , ¹⁷ Disulfiram was shown to block GsdmD‐mediated pyroptosis and lipopolysaccharide‐induced septic death in mice. ³⁹ First, we confirmed that disulfiram blocks GsdmD activity in macrophages and mice. Wild type and GsdmD^−/− bone marrow‐derived macrophages were primed with lipopolysaccharide and then stimulated by ATP to induce Nlrp3 inflammasome. Disulfiram inhibited IL‐1β release in wild type bone marrow‐derived macrophages, with no additive effects in GsdmD^−/− bone marrow‐derived macrophages (Figure 1A). For in vivo efficacy, wild type and GsdmD^−/− mice (fed with WTD±100 mg/kg per day disulfiram) were induced for inflammasome activity via lipopolysaccharide+ATP injections as described in our earlier work. ¹⁵ Disulfiram‐fed mice showed ~80% reduction in IL‐1β release in peritoneal lavage (P<0.0001), with no additive effects observed in GsdmD^−/− mice (Figure 1B). Disulfiram treatment also reduced plasma IL‐1β levels (~70% decrease) in WTD‐fed wild type mice, whereas as expected the GsdmD^−/− mice showed minimal levels of serum IL‐1β regardless of treatment (Figure 1C). As shown in our earlier work, ¹⁵ injection of lipopolysaccharide or ATP alone does not induce IL‐1β release in the plasma (Figure 1C). We extended these studies in the RAW‐ASC cell line, which unlike the parental RAW264.7 cells, express ASC and are inducible for inflammasome assembly and activity. disulfiram did not affect viability and was not cytotoxic to these cells at the tested concentrations of 5 and 10 μmol/L (Figure 1D and 1E). As shown in Figure 1F, disulfiram reduced IL‐1β release in a dose‐dependent and GsdmD‐dependent manner. Disulfiram treatment did not block cleavage of procaspase‐1, or IL‐1β into its mature form, or cleavage of GsdmD into pore‐forming N‐terminal fragment (Figure S1A through S1C). These data are consistent with a previously published study showing that disulfiram inhibits GsdmD pore‐forming activity and release of IL‐1β, without inhibiting cleavage of GsdmD or IL‐1β. ³⁹ As expected, the cleaved form of GsdmD was detected only in lipopolysaccharide+ATP treated cells and not in GsdmD^−/− cells (Figure S1D). Furthermore, using human monocytes (THP‐ASC‐GFP) and fluorescent microscopy for ASC puncta formation, we directly show that disulfiram did not block Nlrp3 inflammasome assembly (Figure S1E). As expected, no ASC puncta formation was observed without lipopolysaccharide+ nigericin treatment (Figure S1E). Taken together, these data indicate that disulfiram inhibits in vitro and in vivo IL‐1β release in a GsdmD‐dependent fashion.

Figure 1. Disulfiram reduces IL‐1β secretion in vitro and in vivo in a GsdmD‐dependent manner.

A, ELISA of bone marrow‐derived macrophages derived from C57BL/6J‐WT and C57BL/6J‐GsdmD^−/− mice ±disulfiram treatment±Nlrp3 inflammasome induction by LPS+ATP; n=5 mice/group. ELISA of peritoneal lavage (B) and ELISA of plasma (C) from C57BL/6J‐WT and C57BL/6J‐GsdmD^−/− mice fed with WTD±disulfiram and treated with LPS+ATP injections to activate the inflammasome; n=5 mice/group. D, Live/dead fixable blue dead cell assay for cell viability of RAW‐apoptosis‐associated speck‐like protein containing a caspase recruitment domain (ASC) cells treated with 5 and 10 μmol/L disulfiram for 2 and 6 hours; n=4. E, CyQuant LDH cytotoxicity assay for toxicity of RAW‐ASC cells treated with 5 and 10 μmol/L disulfiram for 0 and 6 hours. Positive control is maximum LDH release (M‐LDH); n=3. F, ELISA for IL‐1β from WT RAW‐ASC and RAW‐ASC‐GsdmD^−/− cells treated with ±5 and 10 μmol/L disulfiram and primed with 1 μg/mL LPS for 4 hours, followed by stimulation with 1 mmol/L ATP for 20 minutes. Values displayed as mean±SD, significance was determined by 1‐way ANOVA with different letters indicating significant difference of P<0.05, groups sharing same letters indicates that there is no significant difference between the groups. DSF indicates disulfiram; GsdmD, gasdermin D; IL‐1β, interleukin‐1 beta; LA, LPS+ATP; LDH, lactate dehydrogenase; LPS, lipopolysaccharide; nd, not detectable; WT, wild type; and WTD, Western‐type diet.

Disulfiram Reduces Atherosclerosis in Hyperlipidemic Mice

Previous studies demonstrated that global knockout of GsdmD resulted in smaller atherosclerotic lesions in hyperlipidemic mice. ¹⁵ , ¹⁷ To determine if disulfiram can reduce atherosclerosis, apoE^−/− mice were fed an atherogenic WTD or WTD containing 100 mg/kg per day disulfiram for 15 weeks, as shown in the schematic diagram (Figure S2A). Disulfiram‐fed male mice showed a small but significant increase in food intake (~10% increase) versus WTD‐fed males (P=0.035), whereas no differences were found in female mice (P=0.909; Figure S2B). There were no significant changes in body weight in disulfiram‐fed mice in both sexes (P=0.154 for males, P=0.094 for females; Figure S2C), nor were there any significant changes in liver weight in disulfiram‐fed versus WTD‐fed control mice (P=0.681 for males, 0.269 for females; Figure S2D). The disulfiram‐fed mice showed a small but significant increase in plasma cholesterol (~11% increase in males with P=0.012, and ~13% increase in females with P=0.009; Figure S2E). Plasma was fractionated using fast protein liquid chromatography and the peaks showed no differences in distribution of apo B particles (very low‐density lipoprotein, low‐density lipoprotein [LDL], intermediate‐density lipoprotein, or high‐density lipoprotein cholesterol P=0.731 for males, and P=0.682 for females; Figure S2F). We also determined if disulfiram altered complete blood counts in these hyperlipidemic mice. Red blood cells showed no difference in males fed with disulfiram versus WTD only, but there was a small but significant decrease in number of red blood cells in females (~16% decrease with P=0.004; Figure S2G). No significant differences were found in the number of platelets, white blood cells, eosinophils, monocytes, lymphocytes, or neutrophils in disulfiram‐treated versus control mice (Figure S2H through S2M).

The progression of atherosclerosis was determined by the oil red O staining of aortic root sections from WTD‐fed versus disulfiram‐fed mice. As shown in Figure 2A and 2B, atherosclerotic lesions were significantly smaller with 26.7% and 29.4% reduction in males and females, respectively (P=0.008 for males, and P=0.005 for females). As shown in Figure 2A and 2C, necrotic cores were also significantly decreased in disulfiram versus WTD‐fed mice, with 49.85% and 45.63% reduction in males and females, respectively (P=0.0006 for males and P=0.0208 for females). Necrotic cores are also defined as areas lacking extracellular matrix determined by total loss of collagen. The disulfiram‐mediated reduction in necrotic core area was also confirmed by collagen staining using picrosirius red stain (Figure S3). These data indicate that disulfiram reduces atherosclerosis independent of cholesterol levels and without altering host immune cell profile.

Figure 2. Disulfiram reduces atherosclerotic lesions and necrotic cores in hyperlipidemic apoE^−/− mice.

A, Oil Red O staining of atherosclerotic lesions in aortic sinus root sections (10 μm) from mice fed with WTD±disulfiram; n=9–13 mice per group and per sex. Black arrows indicate necrotic cores. B, Quantification of atherosclerotic lesions. C, Quantification of necrotic core in atherosclerotic lesions. Values displayed as mean±SD; P values determined by t test. apoE indicates apolipoprotein E; DSF, disulfiram; and WTD, Western‐type diet.

Disulfiram Induces Autophagy In Vitro and In Vivo

Autophagy plays an atheroprotective role, but is defective in advanced atherosclerotic lesions and can serve as a target for antiatherosclerotic therapeutic. ⁴ , ²⁶ , ²⁷ , ²⁹ To determine if disulfiram can induce autophagy, we treated RAW‐Difluo mLC3 reporter cells (Invivogen) with disulfiram or rapamycin (positive control). RAW‐Difluo mLC3 cells express the RFP::GFP::LC3 fusion protein where RFP (acid‐stable) and GFP (acid‐sensitive) show autophagosome formation and colocalization of GFP and RFP is signal for increased autophagic flux. disulfiram treatment increased the number and colocalization of LC3‐GFP and LC3‐RFP puncta versus nontreated control cells (Figure 3A and 3B). These data indicate the fusion of autophagosomes with lysosomes and ~7‐fold increased autophagy flux in disulfiram‐treated cells. We confirmed disulfiram‐induced autophagy in RAW‐ASC cells, by probing for autophagy markers p62 and LC3‐II by western blotting. The adaptor protein p62 binds to unfolded proteins and damaged membranes to chaperone them to the autophagosome for degradation, thus p62 levels are reduced during autophagy induction. As shown in Figure 3C and 3D and Figure S4A through S4C, disulfiram‐treated cells showed ~0.5‐fold in p62 levels (P=0.005), whereas the levels of LC3‐II were increased by ~2.6‐fold (P=0.002). To ensure that increased LC3‐II is not due to autophagy blockage, disulfiram‐induced autophagy was confirmed by treating RAW‐ASC wild type with chloroquine, an autophagy blocker, and disulfiram+chloroquine. As shown in Figure S4D and S4E, there was an additive accumulation of LC3‐II in disulfiram+chloroquine treated cells, indicating increased autophagic flux. Disulfiram‐induced autophagy was also confirmed in human THP‐1 macrophages, where LC3‐II levels were increased upon disulfiram treatment (Figure S4F and S4G). As disulfiram is a potent inhibitor of GsdmD, we determined if disulfiram effects on autophagy were GsdmD‐dependent. Interestingly, the autophagy induction by disulfiram was independent of GsdmD, as LC3‐II levels remained high in disulfiram‐treated GsdmD^−/− cells (Figure 3E; Figure S4H through S4J). In addition to macrophages, endothelial cells and smooth muscle cells play a crucial role in atherosclerosis. ⁵⁹ Thus, we tested if disulfiram induces autophagy in these cells. Human aortic endothelial cells and primary mouse aortic smooth muscle cells were treated with disulfiram, and as shown in Figure 3F through 3I and Figure S4K and S4L, disulfiram increased autophagy by ~2.5‐fold in HAECs (P<0.0001) and by ~1.6‐fold in MASMCs (P=0.0063), as determined by increased LC3‐II expression. Next, we determined if disulfiram could induce autophagy in vivo. The aortic root sections containing atherosclerotic lesions from mice fed with WTD±disulfiram were probed with anti‐mouse specific anti‐LC3 antibody. As shown in Figure 4A and 4B, expression of LC3 was significantly higher (~2‐fold increase) in aortic root atherosclerotic lesions from disulfiram‐fed versus WTD‐fed control mice (P<0.0001). No differences were found in WTD versus WTD+disulfiram samples with control antibody (Figure S5A). To determine if disulfiram induced autophagy in other tissues, liver homogenates from mice fed with WTD±disulfiram were probed for LC3‐II expression. As shown in Figure 4C and 4D and Figure S5B through S5E, expression of LC3‐II was significantly higher in the liver of disulfiram‐fed versus WTD‐fed control mice (~3‐fold increase for males, P=0.015 and ~1.3‐fold increase for females, P=0.041). Autophagy is also known to reduce lipid droplets in the liver. ⁶⁰ , ⁶¹ Thus, to determine if autophagy induction in the liver alters the accumulation of lipid‐droplets, the liver sections from mice fed with WTD±disulfiram were stained with hematoxylin‐eosin stain. The WTD‐fed mice showed an extensive presence of lipid droplets in liver, while disulfiram‐fed mice had a lower accumulation of lipid droplets (Figure 4E). To further confirm the effect of disulfiram on autophagy in liver cells, we treated human hepatocytes (HepG2 cells) with various doses of disulfiram and found higher LC3‐II expression, with ~2.2‐fold increase in 2 μmol/L disulfiram versus control, ~5‐fold increase in 5 μmol/L disulfiram versus control, and ~5.2‐fold increase in 10 μmol/L disulfiram versus control, with P<0.01 and P<0.001 (Figure 4F and 4G; Figure S5F and S5G). We also probed for LDL receptor and Cyp7a1 (the rate‐limiting enzyme for converting cholesterol to bile acids). As shown in Figure 4F and 4G and Figure S5G through S5I, disulfiram increased LDL receptor in a dose‐dependent manner, with ~1.3‐fold increase in 2 μmol/L disulfiram versus control, ~1.35‐fold increase in 5 μmol/L disulfiram versus control, and ~1.5‐fold increase in 10 μmol/L disulfiram versus control, with P<0.05 and P<0.01. Disulfiram increased Cyp7a1 in a dose‐dependent manner, with ~1.3‐fold increase in 2 μmol/L disulfiram versus control, ~1.2‐fold increase in 5 μmol/L disulfiram versus control, and ~1.5‐fold increase in 10 μmol/L disulfiram versus control, with P<0.05 and P<0.01.

Figure 3. Disulfiram induces autophagy in macrophages independent of GsdmD.

RAW‐Difluo mLC3, RAW‐apoptosis‐associated speck‐like protein containing a caspase recruitment domain (ASC), and RAW‐ASC‐GsdmD^−/−, human aortic endothelial cells, and mouse aortic smooth muscle cells cells were treated ±5 μmol/L disulfiram for 2 h. A, Confocal microscopy of autophagy reporter cell line. RAW‐Difluo mLC3 ±disulfiram and ±positive control rapamycin. B, Percentage of cells showing colocalization of GFP‐LC3 and RFP‐LC3 puncta in RAW‐Difluo mLC3 ±disulfiram±rapamycin treatment, n=100 cells/group. Values are displayed as mean±SD, with P value determined by 1‐way ANOVA, with different letters indicating a significant difference of P<0.05. Western blot (C) and quantification (D) of autophagy markers p62 and LC3‐II in RAW‐ASC cells, n=3, P values determined by t test. E, Western blot of autophagy markers LC3‐II in RAW‐ASC vs RAW‐ASC‐GsdmD^−/− cells ±5 μmol/L or 10 μmol/L disulfiram±positive control rapamycin. Western blot (F) and quantification (G) of autophagy marker LC3‐II in HAECs, n=3, P values determined by t test. Western blot (H) and quantification (I) of autophagy marker and LC3‐II in MASMCs, n≥3. Western blot quantification values displayed as mean±SD, P values determined by t test. a.u. indicates arbitrary units; Ctrl, control; DSF, disulfiram; GsdmD, gasdermin D; HAECs, human aortic endothelial cells; KO, knock out; MASMCs, mouse aortic smooth muscle cells; MWM, molecular weight marker; and Rapa, rapamycin.

Figure 4. Disulfiram induces autophagy in atherosclerotic plaques/liver tissue of hyperlipidemic apoE^−/− mice and in cultured hepatocytes.

A, Expression of LC3 in atherosclerotic lesions by indirect immunofluorescence microscopy using LC3 primary and Alexa488‐labeled secondary antibody. B, Quantification of mean fluorescence intensity, n=50 fields; values displayed as mean±SD, P value determined by t test. Western blot (C) and quantification (D) of LC3‐II in liver homogenate of mice fed with WTD ± disulfiram; upper panel shows males and lower panel shows females; n=4. Values displayed at mean±SD, P values determined by t test. E, Representative hematoxylin‐eosin staining of liver sections from hyperlipidemic mice fed with WTD±disulfiram for 15 weeks. Western blot (F) and quantification (G) of LC3‐II, LDLR, and Cyp7a1 in HepG2 cells treated with various doses of disulfiram vs untreated cells, n=3; values displayed as mean±SD, P value significance against control determined by t test, *P<0.05, **P<0.01, ***P<0.001. apoE indicates apolipoprotein E; a.u., arbitrary units; DSF, disulfiram; GsdmD, gasdermin D; LDLR, low‐density lipoprotein receptor; and WTD, Western‐type diet.

Disulfiram Increases Efferocytosis and Phagocytosis in Macrophages

The effect of disulfiram on atheroprotective pathways such as phagocytosis ⁶² and efferocytosis ⁵ , ³³ was determined. Phagocytosis in disulfiram‐treated macrophages was probed by incubating cells with IgG‐FITC coated latex beads. As shown in Figure 5A and 5B and Figure S6A, disulfiram‐treated THP‐1 macrophages showed a ~1.7‐fold increase in phagocytic activity versus control (P=0.005), as determined by the uptake of FITC‐coated latex beads. Disulfiram‐treated THP‐1 macrophages also showed markedly enhanced efferocytosis (~2.1‐fold increase) versus control (P=0.015), as determined by the uptake of calcein‐labeled apoptotic Jurkat cells by macrophages (Figure 5C and 5D; Figure S6B). To decipher the mechanism of enhanced efferocytosis, effect of disulfiram on the expression of efferocytic receptor MerTK was determined. As shown in Figure 5E, disulfiram‐treated THP‐1 macrophages showed markedly enhanced cell‐surface expression of MerTK. As phagocytosis and efferocytosis are affected by the biophysical properties of the macrophage cell membrane, ⁴⁷ , ⁶³ , ⁶⁴ we measured the elasticity and adhesion of disulfiram‐treated versus control cells by atomic force microscopy. As shown in Figure 5F, disulfiram treatment increased membrane rigidity by ~1.4‐fold (ordered state) of THP‐1 macrophages, denoted by a significant increase in Young's Modulus of elasticity (P=0.0007). Additionally, atomic force microscopy showed an ~1.3‐fold increase in cell adhesion force in disulfiram‐treated macrophages versus nontreated controls (Figure 5G; P<0.0001).

Figure 5. Disulfiram enhances efferocytic and phagocytic activity of macrophages.

THP‐1 macrophages pretreated with ±5 μmol/L disulfiram for 2 hours, were incubated with either IgG‐FITC‐coated latex beads for 1 hour (phagocytosis assay) or calcein‐labeled staurosporine‐induced apoptotic Jurkat cells for 4 hours (efferocytosis assay). Microscopy (A) and quantification (B) of phagocytosis of FITC‐labeled latex beads, with n≥20 cells per treatment group; values displayed as mean±SEM, with P values determined by t test. Microscopy (C) and quantification (D) of efferocytosis of calcein‐labeled apoptotic Jurkat cells, with n≥20 cells per treatment group; values displayed as mean±SEM, with P values determined by t test. E, Indirect immunofluorescence of MerTK in THP‐1 macrophages ±5 μmol/L disulfiram for 2 hours, using antihuman MerTK primary antibody and Alexa488‐labeled goat antirabbit secondary antibody. Cells were imaged by confocal microscopy; black scale bar is 25 μm and white scale bar is 10 μm. Young's modulus of elasticity by atomic force microscopy (F) and adhesion strength by atomic force microscopy (G); 30 cells analyzed per group, with P values determined by t test; values displayed as mean±SEM. a.u. indicates arbitrary units; and DSF, disulfiram.

Disulfiram‐Treated Hyperlipidemic Mice Showed an Altered Gut Microbiota Profile

The gut microbiome plays a pivotal role in the progression of atherosclerosis. ⁶ , ³⁶ , ⁶⁵ , ⁶⁶ disulfiram has been shown to have antimicrobial effects. ²⁵ , ⁶⁷ Thus, we determined if disulfiram could modulate the gut microbiome of hyperlipidemic apoE^−/− mice. The 16s ribosomal RNA (16s rRNA)‐based quantitative polymerase chain reaction sequencing was performed on the fresh fecal samples from apoE^−/− mice fed with WTD±disulfiram diet for 15 weeks. The alpha and beta diversity in microbial population across each group was analyzed. Alpha diversity is a measure of microbiome diversity/complexity in each sample, whereas beta diversity is a measure of similarity or dissimilarity between groups. As shown in Figure 6, the gut microbiome profile was significantly different between the disulfiram‐fed versus WTD‐fed control mice, with alterations seen in both alpha (P<0.05) and beta (P<0.05) diversity (Figure 6A and 6B). The disulfiram‐fed mice showed marked reduction in levels of Romboutsia, Lactococcus, and Blautia bacteria genera compared with that of WTD‐fed mice (P<0.05) and highly significant enrichment in Turicibacter, Bidifobacterium, Lactobacillus, and Akkermansia (P<0.05 for all species; Figure 6C and 6D).

Figure 6. Disulfiram modulates gut microbiota in hyperlipidemic apoE^−/− mice.

The gut microbiota profile of apoE^−/− mice fed with WTD±disulfiram for 15 weeks using 16sRNA‐based quantitative polymerase chain reaction sequencing; n=5 for WTD and n=8 for disulfiram. A and B, Shannon plots showing alpha diversity of gut microbiota in WTD‐fed vs disulfiram‐fed mice. C, Graph showing the relative abundance of bacterial genera that are significantly altered in WTD‐fed vs disulfiram‐fed mice; P<0.05. D, Relative abundance of various bacterial species in WTD‐fed vs disulfiram‐fed mice. apoE indicates apolipoprotein E; DSF, disulfiram; and WTD, western‐type diet.

Collectively, these data indicate that disulfiram modulates multiple atheroprotective pathways and its effect on atherosclerosis may be mediated via GsdmD‐dependent as well as GsdmD‐independent manner.

DISCUSSION

LDL cholesterol lowering therapeutics provide irrefutable benefits for treating CVD, but there is still a 50% to 70% residual risk for the major adverse coronary event even in high‐dose statin‐treated subjects. Thus, adjuvant therapies targeting pathways other than lowering LDL cholesterol are needed. The CANTOS (Canakinumab Anti‐Inflammatory Thrombosis Outcome Study) trial, using IL‐1β antibody in humans, showed the feasibility of adjuvant therapies that target inflammation, independent of lipid levels. But in 2018, the Food and Drug Administration declined to approve canakinumab (Ilaris) for cardiovascular risk reduction on the strength of data. Thus, a drug that targets multiple pathways simultaneously may serve as an adjuvant therapeutic for chronic and complex diseases such as atherosclerosis. Blocking pyroptosis, rather than letting cells rupture and targeting only released IL‐1β, may be more beneficial as other inflammatory molecules such as IL‐18 and Hmgb1 (High‐mobility group box protein 1) can also be blocked. Moreover, earlier work from our lab and others showed that blocking pyroptosis shifts the balance toward apoptotic cell death. ¹⁵ , ¹⁷ Furthermore, a previous study showed that disulfiram can also induce endoplasmic reticulum stress and unfolded protein response‐mediated apoptosis. ⁶⁸ Apoptosis is beneficial in dampening atherosclerosis and cells undergoing apoptosis can be efficiently cleared via phagocytosis and efferocytosis. ³⁰ In addition, inducing autophagy in plaques may result in the degradation of lipid droplets and the generation of free cholesterol that can be effluxed from arterial foam cells via the apoA1‐ABCA1 pathway. ⁶⁹ , ⁷⁰ , ⁷¹ In addition to targeting cellular pathways, modulation of gut microbiota to promote the enrichment of atheroprotective and depletion of atherogenic bacterial species could be another strategy to slow down atherosclerosis progression. ³⁶ , ⁷²

Given the recent findings showing the role of GsdmD in atherosclerosis, ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ we tested the antiatherosclerotic activity and effect of GsdmD inhibitor, disulfiram, on atherosclerosis‐related pathways. disulfiram, also known by its trade name Antabuse, has been approved by the Food and Drug Administration and has been in use for decades for treating chronic alcoholism, as an inhibitor of alcohol metabolizing enzyme aldehyde dehydrogenase. disulfiram effects seem to be broader than targeting aldehyde dehydrogenase alone, as highlighted by several clinical trials that are underway for repurposing disulfiram for various diseases (https://clinicaltrials.gov/search?intr=Disulfiram), but its efficacy as an antiatherosclerosis agent is not yet explored. We found that disulfiram reduced atherosclerosis progression in a hyperlipidemic apoE^−/− mouse model and modulated various pathways involved in atherosclerosis.

Activation of the Nlrp3 inflammasome in advanced human atherosclerotic plaques results in GsdmD cleavage and further disease progression with thinning of the fibrous cap and increase in the necrotic core area in plaques. ¹¹ , ¹³ In agreement with a previous study, ³⁹ disulfiram‐treated macrophages and disulfiram‐fed hyperlipidemic mice showed reduced IL‐1β release upon Nlrp3 inflammasome assembly (Figure 1), indicating a reduction in pore formation. disulfiram does not block Nlrp3 inflammasome directly, nor does it block cleavage of GsdmD, but rather it inhibits the terminal event of membrane pore formation and IL‐1β release. ³⁹ Thus disulfiram‐treated cells are expected to retain their ability to destroy intracellular pathogens. disulfiram, thus may serve as better therapy for atherosclerosis than Nlrp3 inhibitors, as Nlrp3 is essential for countering several pathogens. ⁷³ Using a hyperlipidemic apoE^−/− mouse model of atherosclerosis, we found that disulfiram reduced the size of atherosclerotic lesions (Figure 2). The effect of atherosclerosis is not attributable to feeding behavior, as food intake was not compromised as the disulfiram‐fed male mice ate slightly more food versus WTD‐fed control (Figure S2). The disulfiram effects on atherosclerosis were independent of plasma cholesterol levels as there was a small increase in plasma cholesterol in disulfiram‐fed mice (Figure S2). Disulfiram‐fed mice did not show any significant changes in plasma levels of platelets, white blood cells, eosinophils, monocytes, lymphocytes, and neutrophils (Figure S2). These data indicate that disulfiram effects on atherosclerosis are unlikely due to major changes in the blood cholesterol or immune cell population. It is more probable that the effects are due to changes in the intraplaque immune cell profile, with expected lower inflammation and blunted recruitment of monocytes in the plaque area in disulfiram versus WTD‐fed control mice. In addition to apoptotic cell death, the vast majority of cells in advanced lesions undergo necrotic cell death, ⁷⁴ leading to further amplification in inflammation. The role of the necrotic core in the pathogenesis of atherosclerosis is not fully deciphered, but it is widely believed to stimulate inflammatory pathways and reduce plaque stability. ⁷⁴ Disulfiram‐fed mice showed reduced necrotic core within atherosclerotic plaques (Figure 2), indicating that disulfiram can also modulate necrosis in plaques. disulfiram may reduce necrosis via reducing intraplaque IL‐1β levels or by enhancing the phagocytic and efferocytic activity of macrophages. Both phagocytosis and efferocytosis are essential for blocking the progression of atherosclerosis. ³¹ , ⁶² Efferocytosis is regulated by interaction between exposed phosphatidylserine on apoptotic cells and phosphatidylserine receptors (MerTK or AxlTK) on macrophages. ⁵ , ³³ disulfiram enhanced expression of MerTK (Figure 5), which may result in efficient clearance of necrotic cells from atherosclerotic plaques. Disulfiram also upregulated autophagy, a major atheroprotective pathway. Disulfiram‐induced autophagy in macrophages, endothelial cells, smooth muscle cells, and disulfiram‐fed mice also showed induced autophagy in atherosclerotic plaques (Figures 3 and 4). The liver homogenate of disulfiram‐fed mice also showed increased LC3‐II expression versus WTD‐fed control mice. Similar results were found in hepatocytes (Figure 4), indicating that disulfiram effects on autophagy induction may be global, and not limited to arterial macrophages. The liver sections from our disulfiram‐fed mice also showed fewer lipid droplets versus WTD‐fed control mice (Figure 4). One of the mechanisms for this effect could be disulfiram‐induced lipophagy in the liver. However, the effect of increased lipid degradation in the liver did not get translated into lower plasma cholesterol levels and we believe that this is due to the inability of apoE^−/− mice to clear cholesterol from systemic circulation. In addition to testing disulfiram effects on cellular atheroprotective pathways, we also profiled the gut microbiome of disulfiram‐fed mice. The rationale was that recent studies showed cross‐talk between GsdmD and gut microbiota, ⁷⁵ , ⁷⁶ and GsdmD is functionally deactivated by disulfiram treatment. ³⁹ The disulfiram‐fed hyperlipidemic mice showed markedly altered gut microbiota profile, with a reduction in levels of Romboutsia, Lactococcus, and Blautia bacteria, whereas bacteria species such as Turicibacter, Bidifobacterium, Lactobacillus, and Akkermansia were significantly enriched compared with WTD‐fed controls (Figure 6). Previous studies have shown the atherogenic role of Romboutsia ⁶⁶ , ⁷⁷ and the atheroprotective role of Akkermansia ⁶⁶ , ⁷² , ⁷⁷ Bifidobacterium, ⁷⁷ , ⁷⁸ Lactobacillus ⁷⁸ and Turicibacter ⁷⁹ in the hyperlipidemic apoE^−/− mouse model of atherosclerosis.

Conclusions

Based on our data, disulfiram can serve as an adjuvant therapy for treating atherosclerosis and CVD. The disulfiram effects on atherosclerosis seem to be GsdmD‐dependent (IL‐1β release) as well as GsdmD‐independent (autophagy). Thus, further work is required to understand the pleiotropic effects of disulfiram on atherosclerosis‐related pathways and gut microbiota. For example, microbial transplantation using germ‐free mice and bone marrow transplantations are needed to decipher the contribution of gut microbiota and myeloid cells on disulfiram‐mediated modulation of atherosclerosis. Furthermore, knockouts of Atg7 or MerTK in a GsdmD^−/− background can provide evidence for the role of disulfiram‐induced autophagy and efferocytosis in atherosclerosis. An elegant study showed that disulfiram can also prevent body weight gain in mice fed with an obesogenic diet and reversed diet‐induced obesity. ⁴⁴ Obesity is a well‐known risk factor for accelerated atherosclerosis and increased rates of cardiovascular death. ⁸⁰ , ⁸¹ Thus, disulfiram can be repurposed as an adjuvant therapy for chronic inflammatory metabolic diseases such as atherosclerosis and obesity. Further work is required to delineate the specific contributions of various pathways toward anti‐CVD properties of disulfiram.

Sources of Funding

This work is supported by the National Institutes of Health‐National Heart, Lung, and Blood Institute R01‐HL148158, American Heart Association Transformational Project award 23TPA1063910, and Cleveland State University start‐up funds to Kailash Gulshan. Prabhakara R. Nagareddy is supported by the National Institutes of Health (HL156856, HL137799) and American Heart Association (TPA97002). Chandrasekhar R. Kothapalli is supported by the National Science Foundation (1337859). Olga Cherepanova is supported by the National Institutes of Health (HL150193). Jonathan D. Smithis supported by National Institutes of Health (HL156499).

Disclosures

None.

Supporting information

Data S1

Figures S1–S6