Different role of fatty acid metabolic disturbances in atherosclerosis versus abdominal aortic aneurysm.
Figure 1. Lipid metabolism differs between AS and AAA mouse models.
(A) Schematic illustrations of the animal experimental design.
(B) Representative images of macroscopic features and ORO staining of abdominal aortas.
(C) Volcano plots showing the lipidomic alterations in AS and AAA mice compared to controls (n=5 mice/group). Colored areas highlight lipids with a P < 0.05 and VIP > 1.
(D) Pie charts showing the composition of the altered lipid metabolites by lipid classes in AS and AAA mice compared to controls.
(E, F) Bar plots showing the degree of lipid metabolite changes in AS and AAA mice compared to controls. Red bars represent the top two most significantly altered lipid subclasses.
(G, H) Dot plots showing log2 fold changes in lipid subclasses in AS and AAA mice compared to controls, and the corresponding P values displayed as −log10 (P-value). Dot colors correspond to different lipid subclasses, and the dot size indicates significance. Each dot represents a lipid metabolite. Only lipids with P < 0.05 are shown.
(I) Volcano plots showing the magnitude and significance of the altered lipid metabolites between AS and AAA mice.
(J) Pie charts showing the composition of the altered lipid metabolites by lipid subclasses between AS and AAA mice.
(K) Bar plots showing the degree of lipid metabolite changes between AS and AAA mice.
(L) Dot plots showing log2 fold changes in lipid subclasses between AS and AAA mice, and the corresponding P values displayed as −log10 (P-value). Dot colors correspond to different lipid subclasses, and the dot size indicates significance. Each dot represents a lipid metabolite. Only lipids with P < 0.05 are shown.
Figure 2. Lipid metabolism differs between patients with AS and AAA.
(A) Volcano plots showing significant alterations in the lipidome of CAD and AAA groups compared to healthy control group (n=121 per group). Red dots indicate upregulated lipid metabolites (VIP >1 and P < 0.05).
(B) Pie charts showing the composition of the altered lipid metabolites by lipid classes in CAD and AAA groups compared to healthy control group.
(C, D) Bar plots showing the degree of lipid metabolite changes in CAD and AAA groups compared to the healthy control group. Red bars represent the top two most significantly altered lipid subclasses.
(E, F) Dot plots showing log2 fold changes in lipid subclasses in CAD and AAA groups compared to healthy control group, and the corresponding P values displayed as −log10 (P-value). Dot colors correspond to different lipid subclasses, and the dot size indicates significance. Each dot represents a lipid metabolite. Only lipids with P < 0.05 are shown.
(G) Volcano plots showing the magnitude and significance of the altered lipid metabolites between CAD and AAA patients.
(H) Pie charts showing the composition of the altered lipid metabolites by lipid classes between CAD and AAA patients.
(I) Bar plots showing the degree of lipid metabolite changes between CAD and AAA patients. Red bars represent upregulated lipid subclasses and blue bars represent downregulated lipid subclasses in CAD compared to AAA patients.
(J) Dot plots of log2 fold changes in lipid subclasses between CAD and AAA patients and the corresponding P values displayed as −log10 (P-value). Dot colors correspond to different lipid subclasses, and the dot size indicates significance. Each dot represents a lipid metabolite. Only lipids with P < 0.05 are shown.
(K) Scatter plots showing the top 5 upregulated lipids and top 5 downregulated lipids in CAD compared to AAA patients. Data were analyzed by Student's t-test.
(L) Comparison of differential lipid abundances between human and mouse sets.
Figure 3. FA metabolism is identified as the top altered lipid-related pathway contributing to the lipid metabolic differences between AS and AAA.
(A) PCA score plots of RNA-Seq data from mouse aortas among control, AS, and AAA groups (n=4 per group).
(B) Bar plots showing the number of DEGs in AS and AAA mice compared to controls. DEGs were defined as genes with P < 0.05 and FC ≥ 2.
(C, D) Bar plots showing the top 5 enriched GO biological process terms of the upregulated and downregulated DEGs in the (C) AS vs. control and (D) AAA vs. control comparisons.
(E) Volcano plots showing the magnitude and significance of gene expression changes between AS and AAA mice. DEGs were defined as P < 0.05 and FC ≥ 1.5.
(F, G) Bar plots showing the top 10 enriched GO biological process terms of the upregulated and downregulated DEGs between AS and AAA mice.
(H) Venn diagrams showing overlapping DEGs among the indicated groups.
(I) Bubble charts showing the top 5 pathways enriched for DEGs within the “fatty acid metabolism process” GO term.
(J) Heatmaps showing the expression patterns of DEGs within the “fatty acid metabolism process” GO term.
(K) The protein–protein interaction analysis showing the top-ranked genes.
(L) Top, Quantification of Acadm mRNA and ACADM protein expression measured by RT-qPCR and Western blot in mouse abdominal aortic segments (n=6 per group). Data were analyzed by one-way ANOVA test followed by Tukey's post-hoc test. Bottom, ACADM protein expression assessed by Western blot in mouse abdominal aortic segments.
Figure 4. LD∙ATTEC treatment reduces LDs and alleviates AS in mice.
(A–C) Representative images and quantifications of LDs induced by treatment with 100 µg/mL ox-LDL for 24h in BMDMs treated with the indicated compounds (n=6 independent experiments). Data were analyzed using one-way ANOVA, followed by Tukey's post-hoc test.
(D) Flow cytometry analysis of LDs stained with BODIPY 493/503 in BMDMs pretreated with 100 µg/mL ox-LDL for 24h and subsequently treated with the indicated compounds.
(E) Representative images of the ORO-stained whole aortas, and macroscopic features of wire injury-induced AS in abdominal aortas.
(F) Representative images of abdominal aortas visualized using MUI (top), ORO-stained crossed-sections of abdominal aortas (middle), crossed-sections stained with HE and Masson of abdominal aortas (bottom).
(G–I) Quantification of the ORO–positive areas in the whole aortas (percentage of whole aortas; n=10 per group) (G), lesion area volume of abdominal aorta sections (×104 μm2; n=10 per group) (H); and lipid deposition (percentage of lesion area; n=10 per group) (I). Data were analyzed by Student's t-test.
(J) Representative images of the carotid arteries and MUI visualizations. White arrows indicate thrombi.
(K) Representative photomicrographs of carotid frozen sections stained with ORO, HE, and Masson. White arrows indicate thrombi.
(L–O) Quantification of the incidence of plaque rupture (percentage; n=15 per group) (L); lesion area volume (×103 μm2; n=10 per group) (M); lipid deposition (percentage of lesion area; n=10 per group) (N); and collagen content (percentage of lesion area; n=10 per group) (O). Data were analyzed using Fisher's exact test (I) or Student's t-test (J–L).
(P) Representative images of the macroscopic features of AAA formation in the indicated groups.
(Q) The incidence of Ang II-induced AAA in the indicated groups. Data were analyzed using Fisher's exact test (n=19 per group).
(R, S) Quantification of the maximal diameter of abdominal aortas measured by a Digital Vernier Caliper and total aortic weight/BW in the indicated groups (n=19 per group). Data were analyzed by Mann-Whitney U test (R and S).
Abbreviations: AAA, abdominal aortic aneurysm; Ang II, angiotensin II; AS, atherosclerosis; BMDM, bone marrow-derived macrophage; BW, body weight; LDs, lipid droplets HE staining, hematoxylin and eosin staining; MUI, micro-ultrasound imaging; ORO, Oil Red O; oxLDL, oxidized low-density lipoprotein; Masson, Masson’s trichrome.
Figure 5. LD clearance by LD∙ATTEC treatment induces different transcriptomic changes in AS and AAA mice.
(A) Schematic overview of the animal experimental design.
(B, C) Volcano plots showing log2 fold change and P-value of individual genes in mouse aortas (n=4 mice per group).
(B) Comparison between the AS_Vehicle and AS_LD∙ATTEC groups.
(C) Comparison between the AAA_Vehicle and AAA_LD∙ATTEC groups.
(D) Venn diagrams showing the overlap of genes with fold change ≥ 2 among the indicated groups.
(E, F) Bar plots showing the top 5 enriched GO biological processes of upregulated and downregulated DEGs.
(E) Comparison between the AS_Vehicle and AS_LD∙ATTEC groups.
(F) Comparison between the AAA_Vehicle and AAA_LD∙ATTEC groups.
(G) Bubble plot showing the top 5 pathways enriched for genes within “Fatty acid metabolic process” GO term.
(H) Volcano plots depicting DEGs involved in fatty acid metabolism following LD∙ATTEC treatment.
(I) Top, Quantification of Acadm mRNA and ACADM protein expression measured by RT-qPCR and Western blot in abdominal aortic segments of AS mice treated with vehicle or LD∙ATTEC (n=6 per group). Data were analyzed by Student's t-test. Bottom, ACADM protein expression assessed by Western blot in mouse abdominal aortic segments.
(J) Top, Quantification of Acadm mRNA and ACADM protein expression measured by RT-qPCR and Western blot in abdominal aortic segments of AAA mice treated with vehicle or LD∙ATTEC (n=6 per group). Data were analyzed by Student's t-test. Bottom, ACADM protein expression assessed by Western blot in mouse abdominal aortic segments.
Abbreviations: Ang II, angiotensin II; AS, atherosclerosis; AAA, abdominal aortic aneurysm; GO, gene ontology.
Figure 6. Myeloid-specific Acadm overexpression represses LD formation and mitigates AS but not AAA in mice.
(A) BMDMs were infected with the lentiviral empty vectors or the Acadm-overexpression lentiviral vectors. The overexpression efficiency of Acadm expression was determined by RT-qPCR and Western blot analyses. Left, Quantification of Acadm mRNA expression in BMDMs measured by RT-qPCR (n=6 per group). Right, Representative images of Western blot for ACADM in BMDMs and quantification for Western blot analysis (n=6 per group). Data were analyzed by Student's t-test.
(B) BMDMs were infected and treated with ox-LDL (100ug/ml, 24h). Representative images and quantification of BODIPY493/503 staining of the LDs in the indicated groups. The LD number per cell and average LD size were analyzed (n=6 independent experiments).
(C) BMDMs were isolated from myeloid-specific Acadm overexpression mice. Left, Quantification of Acadm mRNA expression in BMDMs measured by RT-qPCR (n=6 per group). Right, Representative images of Western blot for ACADM in BMDMs and quantification for Western blot analysis (n=6 per group). Data were analyzed by Student's t-test.
(D) Representative images of the ORO-stained whole aortas.
(E) Quantification of the ORO–positive areas in the whole aortas (percentage of whole aortas; n=10 per group). Data were analyzed by Student's t-test.
(F) Representative images of histological staining (ORO staining, HE staining, and Masson staining) of crossed-sections of abdominal aortas in the indicated groups.
(G,H) lesion area volume of abdominal aorta sections (×104 μm2; n=10 per group) (G), and lipid deposition (percentage of lesion area; n=10 per group) (H). Data were analyzed by Student's t-test.
(I) Representative images of the macroscopic features of AAA formation in the indicated groups.
(J) The incidence of AngII-induced AAA in the indicated groups (n=20 per group). Data were analyzed using Fisher's exact test.
(K) Quantification of the maximal diameter of abdominal aortas by a Digital Vernier Caliper in the indicated groups (n=20 per group). Data were analyzed by Mann-Whitney U test.
(L, M) Representative Doppler ultrasound images and quantification of the maximal diameter of abdominal aortas in the indicated groups (n=14 per group). Data were analyzed by Mann-Whitney U test.
(N) Representative images of histological staining (ORO staining, HE staining, and Masson staining) of crossed-sections of abdominal aortas in indicated groups.
(O, P) Quantification of lipid deposition (percentage of lesion area, n=10 per group) (O) and collagen content (percentage of lesion area, n=10 per group) and (P). Data were analyzed by Student's t-test.
Figure 7. Acadm deficiency disrupts mitochondrial function via lipotoxic metabolite accumulation in BMDMs.
(A) Intracellular acetyl-CoA, free-CoA, and free fatty acid levels in BMDMs after Acadm knockdown (n=6 independent experiments). Data were analyzed by Student’s t-test.
(B) Summarized OCR tracings in BMDMs transduced with Acadm-knockdown or control lentivirus (n=6 independent experiments).
(C) Quantification of basal, maximal, and ATP-coupled OCRs in indicated groups (n=6 independent experiments). Data were analyzed by Student’s t-test.
(D) Representative images and quantitative analysis of mitochondrial membrane potential (MitoTracker Green, green) and mitochondrial ROS levels (MitoSOX, red) in control and Acadm-knockdown BMDMs (n=5 independent experiments). Data were analyzed by Student’s t-test.
(E) Summarized OCR tracings in BMDMs transduced with Acadm-knockdown or control lentivirus, followed by etomoxir treatment (n=6 independent experiments).
(F) Quantification of basal, maximal and ATP-coupled OCRs in indicated groups (n=6 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
(G) Representative images of mitochondrial membrane potential assessed by MitoTracker Green staining in control and shAcadm-transduced BMDMs treated with vehicle (DMSO) or etomoxir (2.5 μmol/L).
(H) Quantification of MitoTracker Green staining (n=5 independent experiments).
(I) Representative images of mitochondrial ROS levels assessed by MitoSOX staining in control and shAcadm BMDMs treated as in (G).
(J) Quantification of MitoSOX staining (n=5 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
Figure 8. The establishment of machine learning model utilizing lipidomic features differentiating CAD vs AAA.
(A) OR values of predictor variables and their 95% confidence intervals in the logistic model incorporating lipidomic features for CAD vs. AAA in the training cohort (n=169).
(B) Nomograms constructed based on the training cohort for CAD vs. AAA.
(C) Calibration curves of the logistic regression analysis in the training cohort for CAD vs. AAA. Calibration intercept = 0.001 (SE 0.288, P > 0.99); calibration slope = 1.004 (SE 0.170, P = 0.979). Joint test of intercept = 0 and slope = 1: χ² = 0.001 (df = 2, P > 0.99). No statistically significant deviation from ideal calibration was observed.
(D) ROC curve for four predictive models in the validation cohort (n=73) for CAD vs. AAA.
(E) Radar plots of evaluation metrics for machine learning predictive models for CAD vs. AAA.
(F) Boxplots of concentration of the identified key lipid metabolites with significant coefficients in the logistic regression analysis in the indicated group. Data were analyzed by Student's t-test.
Supplemental Figure 1. Differential associations of plasma lipids with two vascular diseases (AS vs. AAA) in mice.
(A) Correlation of aortic lesion area percentage with plasma levels of TC, LDL-C, and TAG in wire injury-induced AS (n=10-20 mice/group as indicated by the number of individual dots). The P values and correlation coefficients (ρ) were determined using Spearman's rank correlation analysis.
(B) Correlation of aortic lesion area percentage with plasma levels of TC, LDL-C, and TAG in HFD-induced AS (n=10-20 mice/group as indicated by the number of individual dots). The P values and correlation coefficients (ρ) were determined using Spearman's rank correlation analysis.
(C) Correlation of aortic diameters with plasma levels of TC, LDL-C, and TAG in Ang II-induced AAA (n=10–20 mice/group as indicated by the number of individual dots). The P values and correlation coefficients (ρ) were determined using Spearman's rank correlation analysis.
(D) Correlation of aortic diameters with plasma levels of TC, LDL-C, and TAG in CaPO4-induced AAA (n=10-20 mice/group as indicated by the number of individual dots). Correlation coefficients were calculated using both Spearman’s rank correlation (ρ) and Pearson correlation (r), and corresponding 95% confidence intervals are reported. The linear regression line represents the ordinary least squares (OLS) fit, with the shaded area indicating the 95% confidence interval of the regression line.
Supplemental Figure 3. Differential associations of baseline lipid traits and lipid metabolism-related proteins with future risk of CAD and AAA.
(A, B) Associations of metabolites with the risk of CAD and AAA. Metabolites outside the red circle were significantly associated with the risk of CAD and AAA (FDR-adjusted P values < 0.05).
(C) Stacked bar plots showing the number and composition of metabolites significantly associated with the risk of CAD and AAA.
(D) Plot showing the associations of routine lipids with the risk of CAD and AAA.
(E) Heat map showing differential associations pattern of metabolites with the risk of CAD and AAA from NMR-based metabolomics.
(F) Consistency and heterogeneity of metabolites associations with the risk of CAD and AAA. The hazard ratio of each metabolite is given with 95% CIs in gray vertical and horizontal error bars. Metabolites with P ≤0.01 for heterogeneity between associations with CAD and AAA are marked by red dots.
(G, H) Manhattan plots showing the differential associations pattern of proteins with the risk of CAD and AAA. Proteins above the horizontal dotted black line were significantly associated with the risk of CAD and AAA (FDR-adjusted P values < 0.05).
(I) Venn diagram illustrating the overlap of lipid metabolism-related proteins significantly associated with the risk of CAD and AAA.
(J, K) Enrichment analyses of lipid metabolism-related proteins associated with the risk of CAD and AAA, highlighting significant pathways identified through GO and Reactome.
Supplemental Figure 4. Propensity score matching and competing risk analysis of the UK Biobank data.
(A) Stacked bar plots showing the number and composition of metabolites significantly associated with the risk of CAD and AAA after PSM.
(B) Plot showing the associations of the routine lipid metabolites with the risk of CAD and AAA after PSM.
(C) Sub-distribution hazard ratios (sHR) for routine lipid metabolites in the association with the risk of CAD and AAA, using Fine-Gray competing risk models.
(D) Sensitivity analysis of heterogeneity in metabolite associations with incident CAD and AAA using a control-group sample-splitting approach. The control group was randomly split into two independent subsets for the Control vs. CAD and Control vs. AAA comparisons to avoid covariance arising from shared controls. Heterogeneity between βCAD and βAAA was evaluated using a Wald test. Hazard ratios per SD increment are shown with 95% confidence intervals as horizontal (AAA) and vertical (CAD) error bars. Metabolites with P ≤0.01 for heterogeneity between associations with CAD and AAA are marked by red dots.
Supplemental Figure 5. Metabolomics reveals disease-specific metabolic responses in AS and AAA mice.
(A) PCA score plots showing the metabolomics profiles of plasma samples from mice among control, AS and AAA groups (n=6 per group).
(B, C) Volcano plots showing differential metabolites identified in plasma in AS and AAA mice compared to controls. Differential metabolites were identified using P < 0.05 and VIP > 1.
(D, E) Pathway analysis of differentially expressed metabolites. Each bubble represents a metabolic pathway, with the size indicating the pathway impact and the color reflecting significance.
Supplemental Figure 6. Differential transcriptomic and metabolomic regulation in HFD-induced AS and CaPO4-induced AAA.
(A) Schematic illustration of the animal experimental design.
(B) Representative images of macroscopic features and ORO staining of abdominal aortas.
(C) Volcano plot showing the magnitude and significance of gene expression changes in the HFD-induced AS mice compared to controls (n= 3 mice in the HFD-induced AS group and n= 4 mice in the control group). DEGs were defined as P < 0.05 and FC ≥ 2.
(D, E) Selected categories identified from (D) GO analysis and (E) Reactome analysis of upregulated and downregulated genes in the HFD-induced AS mice compared to controls.
(F) Volcano plot showing the magnitude and significance of gene expression changes in the CaPO4-induced AAA mice compared to controls (n= 3 mice in CaPO4-induced AAA group and n= 4 mice in the control group).
(G, H) Selected categories identified from (G) GO analysis and (H) Reactome analysis of upregulated and downregulated genes in the CaPO4-induced AAA mice compared to controls. DEGs were defined as P < 0.05 and FC ≥ 2.
(I) PCA score plots showing metabolomic profiles of plasma samples from mice among control, AS and AAA groups (n=6 per group).
(J) Pathway analysis of differentially expressed metabolites. Each bubble represents a metabolic pathway, with size indicating pathway impact and color representing significance.
Supplemental Figure 7. A gene screening strategy based on GEO datasets.
(A) Venn diagram showing the overlap of DEGs and 4 co-expressed genes identified from the GSE57691 microarray dataset and mouse RNA-seq data.
(B) The mRNA expression of 4 candidate genes in abdominal aortas from WI-injury AS mice (n=6 per group). Data were analyzed by Student's t-test.
(C) Upper panel: Quantification of ACADM mRNA and ACADM protein expression measured by RT-qPCR and Western blot in human AS and non-AS segments (n=5 per group). Data were analyzed by Student's t-test. Lower panel: ACADM protein expression assessed by Western blot in human AS and non-AS segments.
(D) Upper panel: Quantification of ACADM mRNA and ACADM protein expression measured by RT-qPCR and Western blot in human AAA and non-AAA segments (n=5 per group). Data were analyzed by Student’s t-test. Lower panel: ACADM protein expression assessed by Western blot in human AAA and non-AAA segments.
(E) Representative immunofluorescence images of ACADM expression in human AS versus non-AS segments, co-stained with macrophage marker F4/80, smooth muscle cell marker αSMA, and DAPI.
(F) Representative immunofluorescence images of ACADM expression in human AAA versus non-AAA segments, co-stained with macrophage marker F4/80, smooth muscle cell marker αSMA, and DAPI.
(G, H) Acadm mRNA and ACADM protein expression measured by RT-qPCR and Western blot in isolated BMDMs stimulated with ox-LDL (100 µg/mL, 24 h; n=6 independent experiments). Data were analyzed using one-way ANOVA, followed by Tukey's post-hoc test.
Supplemental Figure 8. Medium-chain acylcarnitines are elevated in AS but not in AAA in both mouse and human samples.
(A) Heatmaps showing the relative abundance of acylcarnitine species in mouse aortic tissues.
(B) Quantification of representative medium-chain acylcarnitines (hexanoylcarnitine [C6], octanoylcarnitine [C8], and decanoylcarnitine [C10]) in mouse aortic tissues (n=5 per group). Data were analyzed using one-way ANOVA, followed by Tukey's post-hoc test.
(C) Heatmaps showing the relative abundance of acylcarnitine species in mouse plasma.
(D) Quantification of representative medium-chain acylcarnitines (hexanoylcarnitine [C6], octanoylcarnitine [C8], and decanoylcarnitine [C10]) in mouse plasma (n=5 per group). Data were analyzed using one-way ANOVA, followed by Tukey's post-hoc test.
(E) Heatmaps showing the relative abundance of acylcarnitine species in human plasma.
(F) Quantification of representative medium-chain acylcarnitines (hexanoylcarnitine [C6], octanoylcarnitine [C8], and decanoylcarnitine [C10]) in human plasma (n=5 per group). Data were analyzed using one-way ANOVA, followed by Tukey's post-hoc test.
Supplemental Figure 9. LD clearance by LD∙ATTEC is autophagy-dependent in BMDMs.
(A) Top: Representative images and quantification of LDs (stained with BODIPY 493/503; green), autophagosomes (LC3B; red), and nuclei (DAPI; blue) in BMDMs. LD∙ATTEC induced significant partial colocalization of LDs and autophagosomes. Bottom: Quantification of the percentage of LDs colocalized with autophagosomes (n=6 independent experiments). Data were analyzed using the Student's t-test.
(B) Similar analysis to (A), but performed on aortic segments from mice (AS_Vehicle and AS_LD∙ATTEC mice) 2 weeks after vehicle or LD∙ATTEC treatment (n=6 independent experiments). Data were analyzed using the Student's t-test.
(C) Representative images and quantification of BODIPY493/503 staining of LDs in BMDMs treated with an autophagy inhibitor for 24h following 24h stimulation with ox-LDL (100 µg/mL). The LD number per cell and average LD size were determined (n=6 independent experiments). Data were analyzed using two-way ANOVA, followed by Tukey's post-hoc test.
(D) Quantification of Atg5 mRNA expression assessed by RT-qPCR in BMDMs after transfection with SiCon and SiAtg5 (n=6 independent experiments). Data were analyzed by Student's t-test.
(E) Representative images and quantification of BODIPY493/503 staining of the LDs in the WT versus Atg5-knockout BMDMs treated with LD∙ATTEC for 24h following 24h stimulation with ox-LDL (100 µg/mL). The LD number per cell and average LD size were analyzed (n=6 independent experiments). Data were analyzed using two-way ANOVA, followed by Tukey's post-hoc test.
Abbreviations: BMDM, bone marrow-derived macrophage; LD, lipid droplet; n.s, no significance.
Supplemental Figure 10. Lipid droplet deposition, blood pressure and plasma lipids in AS and AAA mice treated with vehicle or LD∙ATTEC.
(A) Left, Representative photomicrographs of abdominal aortic frozen crossed-sections from wire injury-induced AS mice and the Ang II-induced AAA mice stained with ORO. Right, Quantification of the lipid deposition (percentage of lesion area; n=6). Data were analyzed by one-way ANOVA, followed by Tukey's post-hoc test.
(B) Systolic blood pressure measurements at 0, 7, 14, 21, and 28 days following treatment with either vehicle or LD∙ATTEC in AS and AAA mice (n=10 per group). Group differences over time were analyzed using generalized estimating equations (GEE).
(C-E) Plasma levels of TG, TC, and LDL‐C in AS and AAA mice after 4 weeks of treatment with vehicle or LD∙ATTEC (n=10 per group). Data were analyzed by two-way ANOVA followed by Tukey's multiple-comparison test.
Abbreviations: AS, atherosclerosis; AAA, abdominal aortic aneurysm; TG, triglycerides; TC, total cholesterol; LDL-C, low-density lipoprotein-cholesterol; ORO, Oil Red O.
Supplemental Figure 11. LD∙ATTEC treatment reduces lipid droplets but fails to rescue AAA-relevant phenotypes in mice.
(A) Representative images of abdominal aortas visualized by MUI using the B mode and histological staining (ORO, HE, and Masson) of cross-sections of aneurysms.
(B) Quantification of the maximal diameter of abdominal aortas measured by MUI using the B mode in the indicated groups (n=14 per group).
(C, D) Quantification of collagen contents (percentage of lesion area, n=7 per group) (D) and lipid deposition (percentage of lesion area, n=10 per group) (D). Data were analyzed by Mann-Whitney U test (B) or Student's t-test (C and D).
Abbreviations: Ang II, angiotensin II; AAA, abdominal aortic aneurysm; AS, atherosclerosis; HE, hematoxylin and eosin; Masson, Masson’s trichrome; MUI, micro-ultrasound imaging.
Supplemental Figure 12. LD∙ATTEC treatment leads to different metabolomic changes between AS and AAA mouse models.
(A) PCA score plots of the metabolomic profiles of plasma samples from mice in the indicated groups (n=8 mice per group for AAA_Vehicle and AAA_LD∙ATTEC; n=7 mice per group for AS_Vehicle and AS_LD∙ATTEC).
(B) Proportions of differential metabolites. Left panel: AS_Vehicle group versus AS_LD∙ATTEC group. Right panel: AAA_Vehicle group versus AAA_LD∙ATTEC group.
(C) Volcano plot showing differential metabolites between groups.
(D) Heatmap of top 20 differential metabolites in the indicated groups.
(E) Quantification of the acylcarnitine levels (e.g., L-Carnitine, O-Acetylcarnitine, Dodecanoylcarnitine, Propionylcarnitine, cis-5-Tetradecenoylcarnitine) in the plasma of the indicated groups. Data were analyzed by Student's t-test (normally distributed variables) or Mann-Whitney U test (non-normally distributed variables).
Supplemental Figure 13. LD∙ATTEC treatment markedly reduces lipid contents in AS mice.
(A) PCA of the metabolomic profiles between the AS_Vehicle and AS_LD∙ATTEC groups (n=5 mice per group).
(B) Volcano plots showing differential lipid metabolites between the AS_Vehicle and AS_LD∙ATTEC groups.
(C) Pie charts showing the composition of the altered lipid metabolites by lipid classes between the AS_Vehicle and AS_LD∙ATTEC groups.
(D) Bar plots showing the degree of lipid metabolite changes between the AS_Vehicle and AS_LD∙ATTEC groups. The red bar represents the most significantly decreased lipid subclasses in the AS_LD∙ATTEC compared to the AS_Vehicle group.
(E) Dot plots showing log2 fold changes in lipid subclasses between the AS_Vehicle and AS_LD∙ATTEC groups, and the corresponding P values displayed as −log10 (P-value). Dot colors correspond to different lipid subclasses, and the dot size indicates significance. Each dot represents a lipid metabolite. Only lipids with P < 0.05 are shown.
(F–I) Quantification of the TAG, DAG, CE, and FFA levels in the aortas from mice in the indicated groups. Data were analyzed by Student's t-test (for normally distributed variables) or Mann-Whitney U test (for non-normally distributed variables).
Supplemental Figure 14. Schematic diagram of the construction of conditional myeloid cell-specific Acadm overexpression mice (Acadmflox/floxLysm-Cre, AcadmTg).
Supplemental Figure 15. Clinical association between ACADM function and carotid plaque burden in the UK Biobank cohort.
(A) Schematic workflow for testing the association of ACADM predicted loss-of-function (pLoF) variants with atherosclerosis in the UK Biobank. pLoF variants in ACADM were identified from UK Biobank whole-exome sequencing data. Participants carrying ACADM pLoF variants were compared with noncarriers in association analyses to assess their relationship with plaque burden, measured by Carotid intima–media thickness (cIMT).
(B) Comparison of cIMT between carriers and noncarriers of pLoF variants in ACADM. Data were analyzed using linear regression adjusted for age, sex, assessment center, principal components (PC1–PC10), and genotype array.
(C) Effect estimates (β) with 95% confidence intervals (CI) and corresponding P-values.
Abbreviations: cIMT, carotid intima–media thickness; CI, confidence interval; pLoF, predicted loss of function.
Supplemental Figure 16. Acadm overexpression preserves mitochondrial homeostasis in BMDMs under ox-LDL stimulation.
(A) Intracellular acetyl-CoA, free-CoA, and free fatty acid levels in BMDMs with or without Acadm overexpression (n=6 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
(B) Summarized OCR tracings in BMDMs transduced with Acadm-overexpression or control lentivirus, followed by ox-LDL stimulation (100ug/ml, 24h) (n=6 independent experiments).
(C-E) Quantification of basal, maximal and ATP-coupled OCRs in indicated groups (n=6 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
(F, G) Representative images and quantitative analysis of mitochondrial membrane potential assessed by Mito Tracker Green staining in PBS- or ox-LDL–treated BMDMs with or without Acadm overexpression (n=5 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
(H, I) Representative images and quantitative analysis of mitochondrial ROS levels assessed by MitoSOX staining in PBS- or ox-LDL–treated BMDMs with or without Acadm overexpression (n=5 independent experiments). Data were analyzed by two-way ANOVA followed by Tukey's post-hoc test.
Supplemental Figure 17. Machine learning model utilizing lipidomic features for the diagnosis of CAD.
(A) OR values of predictor variables and their 95% confidence intervals in the logistic model incorporating lipidomic features for CAD vs. HC in the training cohort (n=169).
(B) Nomograms constructed based on the training cohort for CAD vs. HC.
(C) Calibration curves of the logistic regression analysis in the training cohort for CAD vs. HC. Calibration intercept = 0.001 (SE 0.306, P > 0.99); calibration slope = 1.006 (SE 0.182, P = 0.974). Joint test of intercept = 0 and slope = 1: χ² = 0.0001 (df = 2, P > 0.99). No statistically significant deviation from ideal calibration was observed.
(D) ROC curve for four predictive models in the validation cohort (n=73) for CAD vs. HC.
(E) Radar plot of evaluation metrics for machine learning diagnostic models for CAD vs. HC.
(F) Boxplots of concentration of the identified key lipid metabolites with significant coefficients in the logistic regression analysis. Data were analyzed by Student's t-test.
Supplemental Figure 18. Machine learning model utilizing lipidomic features for the diagnosis of AAA.
(A) OR values of predictor variables and their 95% confidence intervals in the logistic model incorporating lipidomic features for AAA vs. HC in the training cohort (n=169).
(B) Nomograms constructed based on the training cohort for AAA vs. HC.
(C) Calibration curves of the logistic regression analysis in the training cohort for AAA vs. HC. Calibration intercept < 0.001 (SE 0. 225, P > 0.99); calibration slope = 1.001 (SE 0.157, P = 0.995). Joint test of intercept = 0 and slope = 1: χ² < 0.001 (df = 2, P > 0.99). No statistically significant deviation from ideal calibration was observed.
(D) ROC curve for four predictive models in the validation cohort (n=73) for AAA vs. HC.
(E) Radar plot of evaluation metrics for machine learning diagnostic models for AAA vs. HC.
(F) Boxplots of concentration of the identified key lipid metabolites with significant coefficients in the logistic regression analysis. Data were analyzed by Student's t-test.