High-fat diet increases circulating palmitic acid produced by gut Bacteroides thetaiotaomicron to promote thrombosis

Abstract

Circulating palmitic acid (PA) is generally considered to be provided from diets and endogenous synthesis and is adversely correlated with cardiovascular disease (CVD). It is unknown, however, if gut microbiota modulates circulating PA and potentiates CVD risk. Here we demonstrate that, in CVD patients, elevated circulating PA is accompanied with hypercoagulability and high gut Bacteroides thetaiotaomicron (BT) abundance. PA promotes coagulation by inhibiting a major endogenous anticoagulant activated protein C (APC) and enhancing platelet activation. Importantly, BT is capable of synthesizing PA, and high-fat diet amplifies gut BT colonization. Our findings show that BT transplantation elevates plasma PA and triggers hypercoagulation without alternating host lipogenesis. Hesperidin, a dietary flavonoid, inhibits PA-APC interaction to prevent hypercoagulation induced by PA or BT transplantation. Collectively, we reveal the promotion of high-fat diet on gut BT colonization that elevates circulating PA and CVD risk, suggesting an approach controlling CVD by targeting PA and BT.

Graphical abstract

Figure 1

Elevated level of PA is accompanied by hypercoagulability in plasma and high abundance of gut BT (A) A shortened plasma recalcification time (PRT) in patients with CVD indicated a tendency to hypercoagulation compared to the normal controls (normal controls, N = 37. patients with CVD D, N = 29). (B and C) Plasma metabolites were identified in plasma from the normal controls and patients with CVD by ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) analysis. The two groups exhibited a distinct separation trend in the ESI⁻ (B) and ESI⁺ (C) mode using partial least squares discriminant analysis (PLS-DA). (D) Plasma fatty acids with adjusted p values less than 0.05 and a fold change greater than 1.2-fold (±0.3785) are highlighted and labeled in volcano plot assay (red: up-regulated; blue: down-regulated; gray: no significant change). PA was elevated in plasma of patients with CVD compared to normal controls, which was selected as a metabolite of interest for validation (indicated by purple arrow). (E) The heatmap of the 23 identified fatty acids in plasma with a significant increase in PA. The colors from red to blue indicate the relative contents of fatty acids in the two groups (red: elevated; blue: decreased). (F) The area under the curve (AUC) of PA was 0.8015 by ROC curve analysis, indicating that PA is a potential biomarker for the diagnosis of CVD. (G) Plasma level of PA was quantified by liquid chromatography-mass spectrometry (LC-MS), which showed a significant elevation in patients with CVD compared to normal controls (normal controls, N = 37. patients with CVD, N = 29). (H) Spearman’s correlation between PRT and plasma PA level and a linear regression line with 95% confidence intervals, showing a negative correlation between them (r = −0.3694, p = 0.004). (I) Fecal samples from patients with CVD and normal controls were collected for 16S rRNA gene sequencing analysis. The principal co-ordinate analysis (PCoA) demonstrated a significant distinction between two groups. (J) The relative abundance of Bacteroides was most significantly higher in patients with CVD among the top 10 species with the highest abundance at the genus taxonomic level in each group, as demonstrated in a columnar cumulative plot of relative abundance. (K) The abundance of BT in feces, quantifying by copy number using quantitative real-time PCR (real-time qPCR), was significantly higher in patients with CVD (normal controls, N = 37. patients with CVD, N = 29). (L) A positive correlation was observed between fecal BT abundance and plasma PA level, as determined by Spearman analysis. A linear regression line with 95% confidence intervals was also plotted. The correlation coefficient is 0.4224, and p = 0.001. ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001 by unpaired t test (A, G, and K). See also Figure S1 and Tables S1–S3.

Figure 2

Gut BT can produce PA, which induces hypercoagulability (A) PA level in Bacteroides thetaiotaomicron (BT), Bacteroides fragilis (BF), Bacteroides ovatus (BO), Lactobacillus reuteri (LE), and Lactobacillus johnsonii (LJ) culture supernatant analyzed by LC-MS (n = 3). (B–D) The addition of PA to plasma shortened the PRT (B) and activated partial thromboplastin time (APTT) (C) but had no significant effect on prothrombin time (PT) (D). POA, MA, OA, and EA had no effect on PRT, APTT, and PT (n = 3). (E) Intravenous administration of PA promotes arterial thrombosis in the model of carotid artery thrombosis induced by 10% FeCl₃, whereas POA, MA, OA, and EA had no effect on arterial thrombosis. Representative images of carotid artery blood flow (top) by laser speckle perfusion imaging are shown, and the region of interest (orange rectangle) was placed in the carotid artery to quantify blood flow change. Blood flow change in the region of interest is shown (bottom) by using perfusion unit (n = 6). (F) Intravenous administration of PA promotes venous thrombosis in the model of venous thrombosis induced by ligating the inferior vena cava, whereas POA, MA, OA, and EA had no effect on venous thrombosis. The inferior vena cava was removed 24 h later for H&E staining to observe thrombosis (left), and the weight of venous thrombus in each group was counted (right) (n = 6). Scale bar, 100 μm. (G–I) PA inhibits mouse saphenous vein bleeding as evaluated by the average time to bleeding cessation (G), the time of first bleeding (H), and the number of times bleeding stopped (disruptions) during the observation period of 30 min (I). POA, MA, OA, and EA had no effect on saphenous vein bleeding (n = 6). (J and K) PA inhibits mouse tail bleeding, whereas POA, MA, OA, and EA do not affect tail bleeding (n = 6). (L and M) PA inhibits hepatectomy surgical bleeding, whereas POA, MA, OA, and EA had no effect on hepatectomy surgical bleeding. Mice underwent a surgical hepatectomy procedure to remove 5 mg of the left lower lobe of the liver. 20 min after incision closure, red blood cell (RBC) (L) and hemoglobin (M) rinsed by 5 mL saline from the lobe were measured (n = 6). ∗p < 0.05 by unpaired t test (A). ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 by one-way ANOVA (B–M). Lines indicate mean ± SEM. All comparisons were made relative to the normal control (NC) group. See also Figure S2.

Figure 3

PA interacts with and inhibits activated protein C and enhances platelet activation (A) PA inhibited the amidolysis of APC on the chromogen substrate S-2366, whereas POA, MA, OA, and EA had no effect on APC activity (n = 3). (B) The inhibitory constant of PA on APC (n = 3). (C) Decreased APC activities of plasma from patients with CVD (normal controls, N = 37. patients with CVD, N = 29). (D) Representative SDS-PAGE and quantification analysis of FVα heavy chain (FVα HC) released from 30 nM of FVα hydrolyzed by 60 nM APC mixed with 0, 12.5, 25, and 50 μM PA, respectively. Blots of APC and FVα light chain (FVα LC) were also shown (n = 3). (E) SPR analysis of the interaction between PA and APC. The association rate constant (K_a), dissociation rate constant (K_d), and equilibrium dissociation constant (K_D) values for the interaction between PA and APC were 42.75 M⁻¹ s⁻¹, 0.004051 s⁻¹, and 9.48 × 10⁻⁵ M, respectively. (F and G) Binding interaction between APC (10 μM) and PA (100 μM) was measured using isothermal titration calorimetry (ITC) at 25°C. The left panel depicts the calorimetric output during the injection of PA into the APC solution (F). The right panel depicts the binding isotherm (G). (H and I) After incubation of PA with platelet-rich plasma (PRP) (H) and washed platelet (I) for 10 min at 37°C, PA shows an induction of platelet aggregation, whereas POA, MA, OA, and EA had no effect on platelet aggregation (n = 3). (J and K) PRP (J) and washed platelets (K) were incubated with PA for 10 min at 37°C and stained with anti-human CD41 antibody and with antibodies against the P-selectin marker CD62P, fixed and analyzed by flow cytometer. The calculated activation ratio is the percent of activated platelets cells, which are positive for CD62P of total platelets (CD41-positive cells). Increased P-selectin expression in the presence of PA in both PRP and washed platelets (n = 3). ∗p < 0.05, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.001 by one-way ANOVA (A, C, D, G, H, I, J, and K). Lines indicate mean ± SEM. All comparisons were made relative to the normal control (NC) group. See also Figures S3 and S4.

Figure 4

PA promotes thrombin generation by inhibiting APC (A) Facilitation of FVa- and FXa-induced conversion of prothrombin to thrombin by PA (n = 3). (B) PA inhibited FVα hydrolytic inactivation mediated by APC. In the absence of APC, the activity of FVa in the prothrombinase assay was designated as 100% (n = 3). (C–G) PA promoted thrombin generation as evaluated by thrombin generation assays (TGA) in human plasma including the thrombin generation curve (C), the endogenous thrombin potential (ETP) (D), the peak of thrombin generation (peak) (E), the time to initiation of thrombin generation (lag time) (F), and the time to peak (tt Peak) (G). POA, MA, OA, and EA had no effect on thrombin generation (n = 3). (H–L) PA blocked the suppression of thrombomodulin (TM), which promotes the activation of protein C (PC) to APC, on thrombin generation as monitored by thrombomodulin thrombin generation assays (TM-TGA). TM was added to plasma to promote the activation of PC to APC. The thrombin generation curve (H), the endogenous thrombin potential (ETP) (I), the peak amount of thrombin (peak) (J), the timing of initiation (lag time) (K), and the time to peak (tt Peak) (L). POA, MA, OA, and EA had no effect on TM-thrombin generation (n = 3). ns: no significance; ∗p < 0.05, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.001 by one-way ANOVA (A–L). Lines indicate mean ± SEM.

Figure 5

BT transplantation increases circulating PA to result in hypercoagulability without altering alternating host endogenous lipogenesis (A) Bacterial copy number in the feces of normal control mice and bacteria-transplanted mice by qPCR (n = 6). NC, normal control; BT, Bacteroides thetaiotaomicron; BF, Bacteroides fragilis; BO, Bacteroides ovatus; LE, Lactobacillus reuteri; LJ, Lactobacillus johnsonii; EC, Escherichia coli. (B) BT transplantation increased circulating PA, while other bacteria had no effect on PA level (n = 6). (C) Heatmap of the fatty acid profiles in the blood from control and BT-transplanted mice (n = 12). (D–G) The expression of genes involved in PA transport and metabolism in liver and gut samples was not changed in mice with BT transplantation. (D) Fatty acid synthase (FAS). (E) Stearoyl-CoA desaturase 1 (SCD1). (F) Peroxisome proliferator-activated receptor γ (PPARγ). (G) Acetyl CoA carboxylase 1 (ACC1) (n = 6). (H and I) BT transplantation promoted coagulation in mice as evaluated by the PRT (H) and APTT (I) (n = 6). (J) BT transplantation inhibited plasma APC activity in mice (n = 6). (K) BT transplantation inhibited mouse tail bleeding (n = 6). (L and M) BT transplantation inhibited hepatectomy surgical bleeding as evaluated by measuring red blood cell (RBC) count (L) and hemoglobin (M) (n = 6). (N) BT transplantation exacerbated FeCl₃-induced aortic thrombosis. Representative images of carotid artery blood flow (top) in 10% FeCl₃-treated mice by laser speckle perfusion imaging. Relative blood flow was shown (bottom) by using perfusion unit (n = 6). (O) High-fat diet (HFD) facilitates gut colonization of BT in mice. BT copy number in the feces of BT-transplanted mice with normal control diet (BT-NCD) and high-fat diet (BT-HFD) by qPCR (n = 6). (P) HFD increased circulating PA in BT-transplanted mice (n = 6). (Q) HFD promoted the gene expressions involving in lipogenesis including fabD, fabG, fabF, and fabZ in gut BT (n = 6). ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 by unpaired t test (A–Q). Lines indicate mean ± SEM. See also Figures S5 and S6 and Table S4.

Figure 6

PA aggravates MI in a mouse cardiac ischemia-reperfusion injury model (A) PA exacerbates MI, as evidenced by elevated S-T segments on the electrocardiogram (ECG), whereas POA, MA, OA, and EA had no effect on myocardial infarction (n = 6). (B) Representative images of triphenyl-tetrazolium chloride (TTC)-stained coronal heart sections (left) and quantitative analysis of heart infarct volume (percentage of total volume) (right) from mice at 24 h after MI surgery. Myocardial infarctions appear white and were measured by planimetry (n = 6). (C–F) Plasma levels of MI biomarkers including cardiac troponin I (cTnI) (C), creatine kinase-MB isoenzymes (CK-MB) (D), creatine kinase (CK) (E), and lactate dehydrogenase (LDH) (F) in plasma in respective groups were determined by ELISA kit (n = 6). (G) Statistic for percentage of fibrotic area (% of total volume) (top) and representative images of heart sections stained with Masson's trichrome (bottom) (n = 6). Scale bar, 100 μm. (H) Representative images of heart sections stained with H&E (n = 6). Scale bar, 20 μm. ns, no significance; ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.001 by one-way ANOVA (B–H). Lines indicate mean ± SEM.

Figure 7

Hesperidin inhibits procoagulant effects induced by PA and BT transplantation (A) Inhibitory effect of hesperidin on PA-APC interaction by SPR analysis. He, hesperidin. (B) Hesperidin blocked PA’s inhibition on the hydrolysis of chromogenic substrate S-2366 by APC (n = 3). (C) Hesperidin blocked PA inhibition of FVα degradation by APC as shown by SDS-PAGE (left) and FVα heavy-chain (HC) quantification analysis (right). Blots of APC and FVα light chain (FVα LC) were also shown (n = 3). (D) Hesperidin blocked the inhibitory effect PA (50 μM) on FVα inactivation caused by APC (n = 3). (E) Hesperidin inhibited the conversion of prothrombin to thrombin promoted by PA (50 μM) (n = 3). (F and G) Hesperidin inhibited coagulation-promoting effects of PA (50 μM) as evaluated by measuring PRT (F) and APTT (G) (n = 3). (H and I) Intravenous hesperidin administration shows antithrombotic effect in 10% FeCl₃-induced aortic thrombosis model in normal (H) and BT-transplanted mice (I). After a single intravenous administration, mouse aortic thrombosis model was immediately induced by 10% FeCl₃. Representative laser speckle perfusion imaging (top) and relative blood flow were shown (bottom) by using perfusion unit (n = 3). (J and K) Oral hesperidin administration shows antithrombotic effect in 10% FeCl₃-induced aortic thrombosis model in normal (J) and BT-transplanted (K) mice. The mice were orally administered hesperidin (once daily) for 7 days and then subjected to the induction of aortic thrombosis by 10% FeCl₃. Representative laser speckle perfusion imaging (J: top; K: left) and relative blood flow were shown (J: bottom; K: right) by using perfusion unit (n = 6). ns, no significance; ∗p < 0.05, ∗∗n < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001, by one-way ANOVA (A–K). Lines indicate mean ± SEM. See also Figure S7.