Nanoparticle-based itaconate treatment recapitulates low-cholesterol/low-fat diet-induced atherosclerotic plaque resolution

Abstract

Current pharmacologic treatments for atherosclerosis do not completely protect patients; additional protection can be achieved by dietary modifications, such as a low-cholesterol/low-fat diet (LCLFD), that mediate plaque stabilization and inflammation reduction. However, this lifestyle modification can be challenging for patients. Unfortunately, incomplete understanding of the underlying mechanisms has thwarted efforts to mimic the protective effects of a LCLFD. Here, we report that the tricarboxylic acid cycle intermediate itaconate (ITA), produced by plaque macrophages, is key to diet-induced plaque resolution. ITA is produced by immunoresponsive gene 1 (IRG1), which we observe is highly elevated in myeloid cells of vulnerable plaques and absent from early or stable plaques in mice and humans. We additionally report development of an ITA-conjugated lipid nanoparticle that accumulates in plaque and bone marrow myeloid cells, epigenetically reduces inflammation via H3K27ac deacetylation, and reproduces the therapeutic effects of LCLFD-induced plaque resolution in multiple atherosclerosis models.

Keywords: ApoE(−/−) mice; CP: Immunology; TCA cycle; atherosclerosis; cholesterol; itaconate; nanoparticle; plaque resolution.

Figure 1.. Systemic metrics of atherosclerosis resolution in dietary cessation CPC model

(A) Illustration of the CPC experimental system. (B) Plasma cholesterol and triglyceride levels after CPC. Data are presented as mean ± SD. (C) Liver cholesterol and triglyceride levels after CPC normalized on dry tissue weight. Data are presented as mean ± SD. (D) Plasma lipoprotein subclass analysis after CPC. Lipoproteins (vLDL, LDL, IDL, HDL) were separated using fast protein liquid chromatography with in-line analysis of cholesterol (top) and triglyceride (bottom) content in the effluent. Data are presented as mean ± SD. (E) Whole-liver mRNA transcript counting using NanoString. The data are shown as a heatmap with a color scale anchored to the minimum and maximum expression values for each gene. (F) ELISA analysis of IL-1β in plasma after CPC. Data are presented as mean ± SD. (G) Flow-cytometric analysis of Ly6C⁺ and LyC⁻ monocytes subsets in blood, spleen and bone marrow after CPC. Data are presented as mean ± SD. (H) Flow-cytometric analysis of subpopulations of Ly6C⁺ and Ly6C⁻ monocytes in blood. The data are presented in the same manner as in (E). Data were analyzed and p values were obtained by one-way ANOVA with Tukey’s multiple comparisons test or pairwise t test with Holm post hoc correction. n =5 (Ctrl), 6 (Prog), and 8 (Cess) animals in (A)–(F) and n = 5–7 (Ctrl), 6–9 (Prog), and 7–10 (Cess) animals in (G) and (H).

Figure 2.. Cessation reduces atherosclerosis and plaque inflammation through glycolytic metabolism activation

(A) Representative photomicrographs of H&E- and oil-red-O-stained sections of aortic root from CPC. Left: quantification of atherosclerosis burden using analysis of luminal occlusion in consecutive tissue sections starting from aortic sinus (AS). Right: quantification of oil red O staining in the adjacent tissue sections. Data are presented as mean ± SD. Scale bar, 200 μm. (B) Necrotic core analysis using H&E-stained sections from (A). Data are presented as mean ± SD. (C) Representative immunofluorescence images of Mac-3- and αSMA-stained aortic root tissue sections and corresponding quantitative ratio analysis (right). Counterstain is DAPI (blue). Data are presented as mean ± SD. Scale bar, 200 μm. (D) NanoString analysis of whole aortas for Cd11b and S100A9. Data are presented as mean ± SD. (E) Schematic of transcriptomic analysis of aortas. (F) Principal component (PC) analysis in bulk RNA-seq from whole aortas of CPC mice. (G) Gene ontology (GO) analysis comparing Cess and Prog RNA-seq datasets. Most represented downregulated GO terms in cellular component sub-ontology are shown. (H) Heatmap displaying fold change (log₁₀) of mRNA expression of specific genes as indicated, measured using RNA-seq and normalized vs. Ctrl samples in whole aortas from CPC. Filled circles indicate p < 0.05. (I) Schematic showing SIRM experiments in cultured aorta organoids and the analysis of uniformly labeled [¹³C]glucose (U-¹³C) and lactate using HSQC-NMR spectroscopy. Data are presented as mean ± SD. (J) Diagram hypothesizing a metabolic shift and fuel utilization in Cess vs. Prog in CPC. Filled circles represent ¹³C-labeled metabolites derived from [¹³C]glucose, and open circles depict ¹²C metabolites originating from other sources. The diagram summarizes the findings from SIRM (see also Figure S2G). Data were analyzed and p values were obtained using pairwise t-test with Holm post hoc correction. n = 3–5 (Ctrl), 4–7 (Prog), and 5–8 (Cess) animals.

Figure 3.. IRG1 and itaconate expression in mouse and human lesions

(A) Relative levels of the metabolites of the TCA cycle in whole aortas from CPC. White/black square on bottom left denotes not detected. (B) SIRM analysis of itaconate isotopologs in CPC aortas from (A). Bottom: scheme illustrating the production of itaconate. Data are presented as mean ± SD. p values were obtained from Student’s t-test. (C) qPCR analysis of the levels of Irg1 transcripts in whole aortas from separate CPC cohorts. N.d., not detected. Data are presented as mean ± SD. p values were obtained from Student’s t-test. (D) Immunofluorescence micrographs depicting positive staining for IRG1 and CD68 in BCA lesions from Prog mice with DAPI counterstain. Scale bars, 100 and 20 μm. (E and F) Histologic assessment with H&E and Movat pentachrome stains and immunofluorescence analysis of IRG1 and CD68 expression in human (F) stable vs. (E) vulnerable thin-cap fibroatheroma (TCFA) plaques from mid-left anterior descending artery (MLAD) with DAPI counterstain. Quantification of co-localization between IRG1 and CD68 is included in (F). Scale bars, 1 mm and 100 μm. (G) BMDMs were treated with free cholesterol for 48 h in the presence or absence of acyl-coenzyme A cholesterol acyltransferase inhibitor (ACAT-i) followed by qPCR mRNA analysis of the expression of Irg1 and Il6 as indicated. Data are presented as mean ± SD. p values were obtained from ANOVA with Tukey post hoc. (H) Immunoblot of IRG1 knockdown in iBMDMs using shRNA. (I) iBMDMs with the IRG1 knockdown (IRG1 KD) or control cells were treated with cholesterol as in (G), and the levels of IkBα and iNOS were determined in the cell lysates using immunoblotting. Staining with Coomassie brilliant blue (CBB) was used as total protein loading control.

Figure 4.. ITA-LNP is an engineered nanometabolite that raises intracellular itaconate levels, exhibits robust anti-inflammatory and metabolic effects in vitro and in vivo, and accumulates in plaque and bone marrow

(A) Chemical structure of itaconate pro-drug and schematic of self-assembled ITA-LNP. (B) Immunoblot for inflammasome assays in Ctrl/ITA-LNP-treated BMDMs. (C) Seahorse analysis of BMDMs pre-treated with Ctrl/ITA-LNP with LPS stimulation. Data are presented as mean ± SD. (D) Immunoblotting for various metabolic targets of itaconate in Ctrl/ITA-LNP-treated BMDM lysate. p values were obtained from Student’s t-test. (E) SIRM experiments in BMDMs cultured with U-¹³C-labeled glucose and treated with Ctrl/ITA-LNP for 24 h. [U-¹³C]glucose and [U-¹³C]lactate levels in supernatants and lysates were measured with HSQC-NMR spectroscopy and normalized to the total lysate protein content. Bottom: IC-MS analysis of itaconate, succinate, and 1,3-bisphosphoglyceric acid (1,3-BPG) isotopologs in the same experiments. Data are presented as mean ± SD. p values were obtained from Student’s t-test. (F) Fluorescence imaging of whole aortas excised 24 h post bolus injection of Atto647-labeled ITA-LNP in Ldlr^−/− mice fed with HCHFD and corresponding oil red O staining of the same aortas. (G) Flow-cytometric analysis of Ctrl/ITA-LNP-targeted cells in blood and plaque post bolus injection in Ldlr^−/− animals fed with HCHFD for 12 weeks. Data are presented as mean ± SD. (H) Atto647-labeled ITA-LNP were bolus injected intravenously in high-fat-fed Ldlr^−/− mice (20 weeks on HCHFD), and the organs were excised following extensive perfusion at 6 h post injection. Mice injected with saline served as a control. Bottom: quantification of Atto647 content normalized to total protein content of the respective organs. n = 2 animals per group; statistical analysis was not performed.

Figure 5.. ITA-LNP accumulates in plaque and alleviates inflammation

(A) Timeline of efficacy testing experiments in Ldlr^−/− mice. (B) En face and histologic analysis of aortic root after oil red O staining from Ctrl/ITA-LNP-treated mice from A. Scale bars, 2 mm and 300 μm. (C and D) Quantification of (C) necrotic core and (D) plaque burden in oil-red-O-stained brachiocephalic artery (BCA) samples from (A). Data are presented as mean ± SD. p values were obtained from Student’s t-test. (E) Quantification of aortic root plaque burden in oil-red-O-stained consecutive tissue sections from (A). Data are presented as mean ± SD. p values were obtained from ANOVA with Tukey’s post hoc test. (F) Plasma lipids, IL-1β levels, and liver lipids as indicated from (A). Data are presented as mean ± SD. p values were obtained from ANOVA with Tukey’s post hoc test. (G) Metrics of plaque stability: macrophages and smooth muscle actin staining in BCA samples from Ctrl/ITA-LNP-treated mice with corresponding quantitative ratio analysis (right). Data are presented as mean ± SD. Scale bar, 200 mm. p values were obtained from Student’s t-test. (H) Metrics of IL-1β signaling: IL-6 and phospho-IRAK in BCA samples from the same mice. Right: quantification with target levels normalized to the total area. n = 6–15 animals per group. Data are presented as mean ± SD. p values were obtained from Student’s t-test. Scale bar, 200 μm.

Figure 6.. ITA-LNP targets inflammatory, epigenetic, and plaque vulnerability pathways in atherosclerotic lesions

(A) Schematic of in vivo cell biotinylation using ITA-LNP incorporating FSL-biotin. (B) In vitro proof-of-concept studies demonstrating biotin label stability in BMDMs treated with ITA-LNP-FSL-biotin on day 1. Surface biotin presence was visualized with FITC-streptavidin. (C) Schematic of experimental workflow for in vivo labeling and magnetic isolation of ITA-LNP-targeted cells in plaque and blood of Ldlr^−/− Cess animals simultaneous with ITA-LNP therapy. (D and E) mRNA from Ctrl/ITA-LNP-targeted cells in plaque from (C) was subjected to bulk RNA-seq and data analysis of representative hierarchical clusters of gene ontology (GO biological process). p values were derived from Benjamini-Hochberg multiple testing with False Discovery Rate (FDR) < 0.05. (F) GO analysis of Ctrl/ITA-LNP-targeted cells in blood. (G) Schematic of tandem stenosis to form unstable plaque (TS). A ligature was placed in the right carotid artery (RCA). (H) Characterization of the TS model. I, oil red O lipid staining; II, gross pathology; III, Prussian blue iron staining in RCA sections. Scale bars, 10 μm and 50 μm. (I) Experimental workflow scheme for ITA-LNP testing in TS model. (J) CD45⁺ cells from RCA segments were subjected to scRNA-seq. Uniform manifold approximation and projection (UMAP) clustering analysis identified distinct cell types. (K) GO analysis of the most prevalent clusters from (J). (L) Heatmap indicating differentially expressed genes in cluster 1 (macrophages).

Figure 7.. ITA-LNP “reprograms” bone marrow to immunomodulatory phenotype through impaired histone acetylation

(A) Ctrl/ITA-LNP were injected into wild-type mice followed by neutrophil isolation from bone marrow. Neutrophils were cultured and stimulated by LPS or TNF-α in vitro. (B) Cytokines in supernatants from (A) as determined by Luminex (n = 7). Filled circles indicate p < 0.001. (C) Principal component (PC) analysis after bulk RNA-seq of mRNA obtained from (A). (D) GO analysis of Ctrl/ITA-LNP treated neutrophils with and without LPS stimulation. (E) ChIP-seq heatmaps showing acetylation of H3K27 near the transcription start site (TSS). The data were obtained from whole bone marrow cells in separate experiments conducted as shown in (A) but without stimulation in vitro. (F) Immunoblot probing for H3K27ac and pan-acetylated lysine (Kac) in lysates of cultured BMDMs treated with Ctrl/ITA-LNP. (G) Profiles of H3K27ac ChIP-seq occupancy and ATAC-seq in loci of the indicated genes from flow-sorted ITA-LNP-targeted bone marrow cells. See text and Figure S7A for identity of these cells. (H) WikiPathways GO analysis of gene set enrichments for downregulated genes in ITA-LNP vs. Ctrl-LNP groups from ATAC-seq datasets. (I) RNA-seq counts of histone acetylation regulating genes differentially expressed in whole bone marrow cells in separate experiments conducted as shown in (A) but without stimulation. Data are presented as mean ± SD. The p values were obtained from ANOVA with Tukey post hoc test.