Palmitic acid in type 2 diabetes mellitus promotes atherosclerotic plaque vulnerability via macrophage Dll4 signaling

Abstract

Patients with Type 2 Diabetes Mellitus are increasingly susceptible to atherosclerotic plaque vulnerability, leading to severe cardiovascular events. In this study, we demonstrate that elevated serum levels of palmitic acid, a type of saturated fatty acid, are significantly linked to this enhanced vulnerability in patients with Type 2 Diabetes Mellitus. Through a combination of human cohort studies and animal models, our research identifies a key mechanistic pathway: palmitic acid induces macrophage Delta-like ligand 4 signaling, which in turn triggers senescence in vascular smooth muscle cells. This process is critical for plaque instability due to reduced collagen synthesis and deposition. Importantly, our findings reveal that macrophage-specific knockout of Delta-like ligand 4 in atherosclerotic mice leads to reduced plaque burden and improved stability, highlighting the potential of targeting this pathway. These insights offer a promising direction for developing therapeutic strategies to mitigate cardiovascular risks in patients with Type 2 Diabetes Mellitus.

Fig. 1. Interplay between type 2 diabetes mellitus (T2DM), acute vascular events, and plaque vulnerability.

A Schematic representation of the cohort study derived from the National Inpatient Sample (NIS) database. B, C Logistic regression analysis depicting the relationship between T2DM and ACS (B), and T2DM and acute cerebral infarction (C) in the NIS database (2016-2018). Model 1 indicates the univariate regression analysis; Model 2 is adjusted for age, gender, and T2DM; Model 3 is further adjusted for additional confounders including obesity, hyperlipidemia, COPD (chronic obstructive pulmonary disease), hypertension, and atrial fibrillation. (Error bars, which extend from the center of the red square, represent the 95% confidence interval for odds ratio. The center for the error bars is the point estimate of the odds ratio for each variable. n = 21400282 subjects). D–I Comparative analyses of various plaque and vascular characteristics, including external elastic membrane (EEM) - cross-sectional area (CSA). (D, data presented as mean ± SD, P = 0.1678), lesion length (E, data presented as mean ± SD, P = 0.3074), fibrous cap thickness (F, data presented as mean ± SD, *P = 0.017), plaque burden (G, data presented as mean ± SD, ***P = 0.0002), minimum lumen area (H, data presented as mean ± SD, **P = 0.005), and lipid necrotic core area (I, data presented as mean ± SD, P = 0.6005), between non-T2DM and T2DM subjects. J Representative IVUS (intravascular ultrasound) images of a normal coronary artery, a vulnerable plaque, and a stable plaque. The IVUS catheter, blood flow, fibrous cap, and lipid necrotic core are indicated in brown, red, yellow, and blue, respectively. K Logistic regression analysis revealing the relationship between T2DM and plaque vulnerability in the cohort study. Model 1 indicates the univariate regression analysis; Model 2 is adjusted for age, gender, and T2DM; Model 3 is further adjusted for T2DM, age, gender, BMI, LDL, smoking, hypertension, and serum creatinine levels. (Error bars, which extend from the center of the red square, represent the 95% confidence interval for odds ratio. The center for the error bars is the point estimate of the odds ratio for each variable. n = 21400282 subjects. D, E Two-tailed unpaired Student t-test, F, G, H, I Two-tailed unpaired Student t-test with Welch’s correction). Source data are provided as a Source Data file.

Fig. 2. Unveiling divergent metabolomic patterns in T2DM versus non-T2DM subjects within the “Chronic Stable Coronary Atherosclerotic Disease Prospective Cohort” registry.

A Orthogonal partial least squares discriminant analysis (OPLS-DA) scatter plot visualizing metabolomic divergence between non-type 2 diabetes mellitus (non-T2DM) (depicted in blue) and T2DM (depicted in orange) human serum samples in the combined model (COMB). The X and Y axes represent the contributions of individual samples to the first two principal components (PC1 and PC2), respectively. B Cross-validation plot of the OPLS-DA model with 200 permutations, indicating robustness and absence of overfitting, as evidenced by intercepts of R2 = (0.0, 0.940) and Q2 = (0.0, 0.548). C A volcano plot delineating the pairwise comparisons of metabolites in T2DM relative to non-T2DM subjects. The vertical dashed lines depict the twofold abundance difference threshold, and the horizontal dashed line marks the P = 0.05 significance threshold. Metabolites exhibiting significant alterations are highlighted in orange (up-regulated) or green (down-regulated). D Venn diagram illustrating unique and shared metabolites between serum samples from non-T2DM and T2DM subjects in the cohort. E Depiction of the top 30 metabolites with the most significant differences based on the absolute values of log2 fold change. The bar graph represents the log2 fold change of each metabolite. F Annotation of metabolites using the Lipid Maps database, with text on the left and right sides denoting classes and categories, respectively. G Illustration of the frequency of significantly altered metabolites within each category as annotated by Lipid Maps. H Representation of the frequency of significantly altered metabolites within each class within the “Fatty Acyls” category. I Display of all significantly differentially expressed metabolites within the “fatty acids and conjugates” class based on the significance of their log2 fold change. The bars in the graph represent the log2 fold change values for each metabolite. Source data are provided as a Source Data file.

Fig. 3. Distinct metabolomic signatures in atherosclerotic mice complicated with diabetes.

8-week-old male C57BL/9 background ApoE^-/- mice and ApoE^-/- mice received STZ injections were fed with HFD for 20 weeks. A A flowchart of establishing mouse models for atherosclerosis (AS) and atherosclerosis complicated with diabetes (AS + DM). Oil Red O (ORO) staining of aortic root plaques (B) and aorta (C) in AS and AS + DM mice. Scale bars: 400 μm (B), 1 cm (C). Quantitation of plaque area in AS and AS + DM mice expressed as the ratio of ORO-positive area to the intra-external elastic membrane (EEM) area (D, n = 6 mice, data presented as mean ± SD, ***P < 0.001)), and ORO-positive area to total aortic area (E, n = 6 mice, data presented as mean ± sd, ***P < 0.001). F Orthogonal partial least squares discriminant analysis (OPLS-DA) scatter plot illustrating the distinct metabolomic profiles between AS (blue) and AS + DM (orange) mice serum samples in the combined model (COMB). The axes denote the contributions of individual samples to the first two principal components (PC1 and PC2). G Cross-validation plot of the OPLS-DA model, corroborated by 200 permutations. The intercepts of R2 = (0.0, 0.996) and Q2 = (0.0, 0.889) confirm the model’s robustness and lack of overfitting. H A volcano plot delineating the pairwise comparisons of metabolites in AS + DM relative to AS mice. The vertical dashed lines depict the twofold abundance difference threshold, and the horizontal dashed line marks the P = 0.05 significance threshold. Metabolites exhibiting significant alterations are highlighted in orange (up-regulated) or green (down-regulated). I Top 30 significantly changed metabolites by log2 fold change. J A Venn diagram visualizing unique and shared metabolites of AS + DM and AS mice. K Metabolite annotations are performed using the Lipid Maps database, with text on either side signifying classes or categories. Illustration of the frequency of significantly altered metabolites within each category (L) and class (M) under the “Fatty Acyls” category as annotated by Lipid Maps. N Display of all significantly altered metabolites within the “fatty acids and conjugates” class based on the significance of their log2 fold change. Source data are provided as a Source Data file.

Fig. 4. Association between palmitic acid and atherosclerotic plaque vulnerability in T2DM.

A Orthogonal partial least squares discriminant analysis (OPLS-DA) scatter plot demonstrates the metabolomic differences between non-T2DM and T2DM human serum samples. B Cross-validation plot for the OPLS-DA model reveals model robustness and absence of overfitting, as validated by the intercepts of R2 = (0.0, 0.891) and Q2 = (0.0, 0.367). C Volcano plot detailing the results of pairwise comparisons of metabolites in T2DM relative to non-T2DM subjects. D Venn diagram displays the overlap of 705 annotated metabolites shared between human and mouse models. E Display of the top 30 altered metabolites, ranked by their log2 fold change. F Metabolite annotation using the Lipid Maps database. G Visualization of significantly altered metabolites annotated by Lipid Maps database. H Venn diagram shows the 34 shared human-mouse annotated metabolites classified within the “Fatty acyls” category according to the Lipid Maps database. Depiction of the count of distinctively altered metabolites across categories (I) and within “Fatty acyls” category (J). K Visualization of significantly altered metabolites in the “Fatty acyls” category. L Comparison of serum palmitic acid (PA) concentration between non-T2DM and T2DM human subjects. (n = 55 subjects in non-T2DM and n = 46 subjects in T2DM, data presented as mean ± SD, two-tailed unpaired Student t-test with Welch’s correction, ***P < 0.001). M Two-tailed Pearson’s linear regression analysis shows a significant negative correlation (r = −0.262, P = 0.008) between serum PA concentration and coronary plaque fibrous cap thickness. The shaded area around the regression line is the error band, representing confidence interval. N Logistic regression analysis reveals the relationship between PA and plaque vulnerability in the cohort study. Model 1 indicates the univariate regression analysis; Model 2 is adjusted for age and gender; Model 3 is adjusted for age, gender, BMI, LDL, smoking, and hypertension. (The center for the error bars is the point estimate of the odds ratio for each variable. n = 101 subjects). O A Kaplan-Meier survival curve plots the 180-day MACE-free survival in Higher PA group versus Lower PA group. Source data are provided as a Source Data file.

Fig. 5. Mitigation of palmitic acid-induced atherosclerotic plaque vulnerability by macrophage Dll4 deletion.

8-week-old male C57BL/9 background ApoE^-/- mice, (Dll4^flox/flox; Lyz2-Cre^+/-; ApoE^-/-) mice were fed with standard chow, high-fat diet (HFD) or HFD containing 5% palmitic acid for 20 weeks. A Displayed are exemplary images of aortic root plaques stained with oil red O (ORO). The scale bar indicates a length of 200 μm (n = 3 mice, data presented as mean ± SD, left to right: ***P = 0.0007, ***P = 0.0007, *P = 0.0363, *P = 0.0274, ***P = 0.0005, *P = 0.0290, ***P = 0.0005). B Depicted are representative images of aortic root plaques subjected to Masson’s trichrome staining. The scale bar denotes a length of 200μm (n = 3 mice, data presented as mean ± SD, left to right ***P < 0.001, ***P = 0.0008, ***P < 0.001, ***P = 0.0001). C Presented are selected images of aortic root plaques stained with Sirius Red. The scale bar signifies a length of 200μm (n = 3 mice, data presented as mean ± sd, left to right ***P < 0.001, *P = 0.0373, *** P = 0.0005, **P = 0.0078). D Shown are sample images of aortic root plaques that have undergone double immunofluorescent staining for CD68 and Dll4. The scale bar represents a length of 200μm (n = 3 mice, data presented as mean ± sd, left to right ***P = 0.0002, **P = 0.0038, ***P < 0.001, ***P = 0.0002). E Exhibited are representative images of aortic root plaques that have undergone double immunofluorescent staining for α-smooth muscle actin (α-SMA) and notch intracellular domain 1 (NICD1). The scale bar equates to a length of 200μm (n = 3 mice, data presented as mean ± sd, left to right ***P < 0.001, *P = 0.0168, ***P < 0.001, *P = 0.0208). F Depiction of CellEvent Green Senescence staining of the aortic root. The scale bar represents 200μm. (n = 3 mice, data presented as mean ± sd, left to right ***P < 0.001, ***P = 0.0004, ***P < 0.001, **P = 0.0021). A–F: one-way ANOVA with the Tukey post hoc correction. Source data are provided as a Source Data file.

Fig. 6. The modulatory role of Dll4 silencing on Notch signaling-mediated VSMC senescence through the suppression of palmitic acid-induced macrophage TLR4 pathway activation.

A Representative images of immunofluorescent staining of CD68 and Dll4 in macrophages, with quantified expression of CD68 (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001) and Dll4 (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001, scale bar = 50 μm). B Western blots images and quantifications of relative levels of TLR4 (left to right *P = 0.0202, *P = 0.0298, P = 0.9882), FOXC2 (n left to right ***P < 0.001, ***P < 0.001, ns P = 0.9347), and Dll4 (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001), as well as the relative phosphorylation level of ERK (left to right ***P < 0.001, ***P < 0.001, ns P = 0.8364). C A schematic diagram illustrating contact and non-contact co-culturing models of macrophages and VSMCs. D Immunofluorescent images of Dll4 in macrophages and NICD1 in VSMCs. Accompanying quantifications illustrate the relative Dll4 expression (left to right ***P < 0.001, ***P < 0.001), and the nuclear translocation of NICD1 (ns P = 0.30945, ***P < 0.001) in VSMCs, scale bar = 50 μm. E Western blots images and quantifications of relative expressions of HES1 (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001), SIRT1 (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001), P21 (left to right ***P < 0.001, ***P < 0.001, ***P = 0.0003), and the nuclear translocation of NICD1 (left to right ***P < 0.001, ***P < 0.001, **P = 0.0032) in VSMCs. F Immunofluorescent staining of NICD1 (scale bar = 50μm), SA-β-gal staining (scale bar = 50 μm), and CellEvent Green staining (scale bar = 40 μm) in VSMCs. The corresponding quantifications stand for nuclear translocation of NICD1 by co-localization analysis with DAPI in VSMCs (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001), the extent of VSMC senescence by quantifying the percentage of SA-β-gal positive stained VSMCs (left to right ***P < 0.001, ***P < 0.001, *P = 0.0108) and mean fluorescent intensities of CellEvent Green staining (left to right ***P < 0.001, ***P < 0.001, ***P < 0.001). n = 6 independent replicates in all experiments, All data are presented as mean ± SD. A, B, E, F: one-way ANOVA with the Tukey post hoc correction, D: Two-tailed unpaired Student t-test and Two-tailed unpaired Student t-test with Welch’s correction. Source data are provided as a Source Data file.