Atherosclerosis is a multifactorial disease associated with many risk factors, such as smoking, type 2 diabetes mellitus, obesity, hypertension and metabolic syndrome. Lipid-driven chronic inflammation is a typical feature of atherosclerosis that is initiated by LDL accumulation in the arterial wall and its modification by oxidative stress-mediated mechanisms1,2,3,4. LDL accumulation results in the recruitment of macrophages, dendritic cells and lymphocytes. The absence of tissue repair or a defect in the resolution of inflammation leads to atherosclerotic lesion progression, characterized by the formation of a central necrotic core and the presence of inflammatory cells5. Persistent inflammatory stimuli owing to continuous accumulation of LDL in the arterial intima, coupled with impaired efferocytosis and an inflammatory plaque phenotype leads to plaque instability and eventual thrombosis5. The central role of inflammation in the atherosclerotic process is supported by interventional studies performed in the past decade showing that the risk of cardiovascular disease can be attenuated with the use of anti-inflammatory therapies6,7. A more detailed definition of the mechanisms underlying inflammation-related cardiovascular disorders such as atherothrombosis could help identify novel therapeutic approaches to preventing atherosclerotic progression.

The gut microbiota comprises trillions of microbes, including bacteria, archaea and viruses, that colonize the entire gut from the stomach to the small intestine and, to a major extent, the colon8. Gut dysbiosis is an alteration of the gut microbiome characterized by a reduction in microbial diversity, with predominance of bacteria such as Bacteroidetes and Enterococcaceae and a reduction in Bifidobacterium and Firmicutes9. A growing body of evidence indicates that gut dysbiosis is implicated in the atherothrombotic process via increased translocation of viable bacteria or bacterial products such as lipopolysaccharides (LPS) and trimethylamine-N-oxide (TMAO) into the systemic circulation10.

Translocation of LPS into the systemic circulation has negative effects on body homeostasis, shifting the host immune defence to a pro-inflammatory state, which has been shown to predispose to atherosclerosis11,12 (Box 1). LPS is a pathogen-associated molecular pattern, and is recognized by innate immune system receptors, such as Toll-like receptors (TLRs), a family of highly conserved, membrane pattern-recognition receptors that are implicated in atherosclerotic and thrombotic processes13,14,15. For example, TLR2, TLR4 and TLR9 are involved in platelet activation and thrombus formation16,17.

Together, these data indicate the existence of a gut–systemic circulation axis, which might contribute to atherothrombosis via a sequence of events including increased LPS translocation from the gut to the portal circulation and impaired degradation of LPS in the liver. The LPS can then reach the systemic circulation to induce arterial inflammation and ultimately thrombus growth via binding to TLR4 (Fig. 1). In this Review, we describe the mechanisms underlying LPS translocation from the gut to the systemic circulation and the ensuing low-grade endotoxaemia, the effect of low-grade endotoxaemia on the atherosclerotic and thrombotic processes, the clinical settings associated with low-grade endotoxaemia and cardiovascular complications, and the potential therapeutic tools to improve gut permeability and eventually eliminate low-grade endotoxaemia.

Fig. 1: Interplay between low-grade endotoxaemia and vascular disease.
figure 1

Lipopolysaccharide (LPS) can translocate into the systemic circulation as a consequence of downregulation of intestinal adhesion proteins induced by gut dysbiosis. LPS is metabolized by liver cell enzymes and excreted through the bile. However, if degradation and biliary excretion are impaired, LPS can reach the systemic circulation. In the arteries, LPS can bind to Toll-like receptor 4 (TLR4) in leukocytes, endothelial cells and platelets. TLR4 activation leads to the recruitment of the TIRAP–MyD88 complex and downstream signalling, eliciting an inflammatory response that induces plaque instability (which can lead to rupture and erosion), leukocyte activation with the formation of neutrophil extracellular traps (NETs), clotting activation and thrombus formation.

Low-grade endotoxaemia and gut permeability

Low-grade endotoxaemia

According to studies performed in healthy individuals, low-grade endotoxaemia is defined as circulating levels of LPS of >20 ng/ml15,18,19. Compared to patients with sepsis, low-grade endotoxaemia is characterized by at least twofold lower blood LPS levels13,20 and absence of a specific clinical picture. Low-grade endotoxaemia can occur as a consequence of gut microbiota dysbiosis and disruption of intestinal barrier function21.

LPS is a glycolipid component of the outer membrane of Gram-negative bacteria and is composed of carbohydrates and a lipid A portion. LPS levels in the peripheral circulation usually increase after food intake, coinciding with elevation in the levels of apolipoprotein B48, a protein synthetized by intestinal cells to transport chylomicrons in the peripheral circulation22. LPS is embedded in newly synthesized chylomicrons and passes across the intestinal barriers into the lymphatic system and then into the bloodstream23,24. In the bloodstream, LPS is transported by the LPS-binding protein, a specific 60-kDa acute-phase response glycoprotein, and cleared from the blood circulation by plasma lipoproteins including HDL, LDL and VLDL25,26. HDL is the main lipoprotein implicated in LPS transport and clearance by hepatic bile, and the HDL–LPS interaction directly protects against the toxic effect of LPS, as shown in vitro and in vivo27,28,29. In normal settings, LPS is transported to the liver, where it undergoes degradation by specific liver enzymes (such as acyloxyacyl hydroxylase and alkaline phosphatase) or excretion into the bile via scavenger receptors30. The inability of liver cells to completely metabolize or excrete LPS into the bile might have consequences not only in the establishment of low-grade endotoxaemia but also in inducing liver damage, such as non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH)31. The pro-inflammatory damage induced by LPS occurs via binding of its lipid moiety lipid A to TLR4 (ref.32), which binds LPS via the membrane-bound co-receptor CD14 (ref.33). LPS binding to TLR4 leads to the recruitment of the adaptor protein myeloid differentiation primary response protein 88 (MyD88) to the cytoplasmic domain of TLR4 (ref.34), resulting in activation of the transcription factor NF-κB35.

Gut barrier and permeability

The human intestine has several lines of defence that protect against the translocation of microorganisms or microbial products into the bloodstream. This multilayer barrier constitutes the largest interface between the external environment and the host. The first line of defence consists of mucus that separates the gut microbiota from a single layer of different types of epithelial cells that forms the intestinal epithelial barrier36,37. The continuous intercellular barrier of intestinal epithelial cells tightly regulates the absorption, secretion and transport of water, ions and organic molecules across the epithelium38. The intestinal epithelial barrier consists of a physical barrier formed by the apical plasma membrane of enterocytes, which are held together by tight junction proteins, adherens junction proteins (cadherins and catenins), gap junction proteins (connexins) and desmosomes (desmoglein and desmocollins)39 (Fig. 2). Tight junction proteins are the most apical junctional complex regulating gut permeability and paracellular diffusion and, therefore, are crucial for maintaining cell-to-cell adhesion and gut barrier health40. The tight junction complex includes transmembrane proteins, such as claudins, occludin, tricellulin and junctional adhesion molecules, and intracellular scaffold proteins such as the zonula occludens proteins ZO1, ZO2 and ZO3 that bind the transmembrane structure to the cytoskeleton40,41 (Fig. 2). The gut–vascular barrier, an additional cellular barrier, is situated below the epithelial barrier and is also involved in controlling the translocation of microorganisms into the portal vein42,43.

Fig. 2: Mechanisms of gut permeability-mediated low-grade endotoxaemia.
figure 2

The gut epithelial barrier consists of the apical plasma membrane of enterocytes, held together by tight junction proteins (claudin and occludin) and adherens junction proteins (E-cadherin and catenin), as well as the zonula occludens proteins ZO1 and ZO2, which are adaptor proteins necessary for the structural and regulatory functions of tight junctions. a | Upregulation of junctional proteins can be induced by microbiota metabolites including polyphenols, indole and indole derivatives, short-chain fatty acids (SCFAs) and polyamines. b | Downregulation of junctional proteins is mediated by: lipopolysaccharides (LPS) through binding to Toll-like receptor 4 (TLR4); by zonulin, a protein that activates the EGF receptor (EGFR) through transactivation of the proteinase-activated receptor 2 (PAR2), thereby inducing protein kinase C (PKC) phosphorylation; and by pro-inflammatory cytokines, including IL-1β, interferon-γ (IFNγ) and tumour necrosis factor (TNF). JAM, junctional adhesion molecules.

Gut barrier permeability is regulated by intrinsic and extrinsic mechanisms in the intestinal epithelial cells, and can be altered by exogenous factors, such as excessive alcohol intake and non-steroidal anti-inflammatory drugs, or endogenous factors, such as inflammation linked to systemic disease40. The gut microbiota has a protective role in the digestive mucosa by participating in the maintenance of the physiological integrity of tight junction proteins44. The gut microbiota generates a variety of metabolites from dietary products that have important effects on gut barrier function and immune responses45. Microbiota metabolites with a protective role for the gut barrier include short-chain fatty acids (SCFA), indole and indole derivatives (which are generated from tryptophan microbial digestion40), bile acid metabolites, polyamines and polyphenols (Fig. 2). Moreover, gut dysbiosis is a prerequisite for the modification of gut barrier functionality and the translocation of microorganisms or microbial products into the systemic circulation40. For example, in animal models of diabetes and obesity, treatment with a large spectrum of antibiotics reduced the levels of LPS in the systemic circulation coincidentally with a reduction in gut permeability via upregulation of tight junction proteins such as ZO1 and occludin20.

Diet is an important factor affecting gut permeability. In rats fed a high-fat diet, changes in gut microbiota are characterized by an imbalance in the ratio of Gram-negative to Gram-positive bacteria that might predispose to alterations in gut barrier defence21. The relationship between a high-fat diet, metabolic disease and blood LPS levels was first outlined by Cani and colleagues13,20. They demonstrated in mouse models that a 4-week high-fat diet chronically increased the plasma LPS concentration two to five times, corresponding to one to two orders of magnitude lower than the levels attained with infections46. The high-fat diet also affected gut barrier integrity by reducing the number of Bifidobacteria20, which have been shown to protect intestinal barrier integrity via upregulation of tight junction proteins such as ZO1 and occludin47. Finally, Cani and colleagues showed that chronic, experimental metabolic endotoxaemia induced by LPS infusion triggered the development of obesity, diabetes and liver insulin resistance, mimicking the negative effects elicited by a high-fat diet alone20.

Ageing is another important factor in gut barrier dysfunction, which in turn can promote low-grade systemic inflammation, a characteristic observed in older individuals48. A reduction in gut microbiota diversity and an imbalance between opportunistic and commensal bacteria (with increased numbers of opportunistic bacteria such as Enterobacteriaceae, Clostridium perfringens and Clostridium difficile, and decreased numbers of commensal bacteria such as Bacteroides, Bifidobacteria and Lactobacilli) have been reported in older individuals49,50. Age-related gut dysbiosis negatively influences gut barrier permeability and is responsible for translocation of microbial products into the systemic circulation, with ensuing systemic inflammation and ultimately impaired bacteria-killing immune responses51.

Together, these data indicate that diet-induced gut microbiota alterations are the primum movens of a sequence of events including dysbiosis-mediated increase in gut permeability, in which LPS has a key role in the disassembly of intestinal adhesion proteins (Fig. 2). LPS binding to TLR4 in intestinal cells begets an inflammatory process that ultimately downregulates the levels of tight junction proteins and favours the translocation of LPS into the systemic circulation. Several studies in experimental models using TLR4 inhibitors or Tlr4-knockout animals have demonstrated a crucial role for the LPS–TLR4 pathway in modulating gut barrier integrity52,53,54,55. Nevertheless, the intestinal barrier includes factors that can protect from LPS-induced damage. For example, intestinal alkaline phosphatase (IAP) removes one of the two phosphate groups of the LPS lipid A moiety, resulting in LPS degradation to monophosphoryl LPS, which can still bind to TLR4 but acts as antagonist56. In animals fed a high-fat diet, IAP overexpression improved intestinal function by maintaining the integrity of intestinal mucosa, thereby reducing LPS translocation into the systemic circulation and lipid accumulation in the liver, resulting in the attenuation of atherosclerotic plaque burden57. Biosynthesis of HDL subspecies, such as HDL3, by intestinal cells is another tool to counteract LPS pro-inflammatory activity, given that HDL3 can sequester LPS in the portal vein, thereby preventing its binding to TLR4 on liver macrophages and protecting against liver damage58.

Gut permeability and cardiovascular risk

Emerging evidence indicates that gut dysbiosis is a risk factor for cardiovascular disease in mice with type 1 diabetes mellitus59, obesity60 or hypertension10. Studies to assess gut permeability in patients at risk or with coronary heart disease have mostly been conducted by measuring serum levels of zonulin, which is an indirect marker of gut permeability. Zonulin is a 47-kDa protein released by epithelial cells of the small intestine after stimulation by gliadin or gut dysbiosis61. In interstitial epithelial cells, the signalling pathway downstream of zonulin leads to protein kinase C phosphorylation, which triggers the disassembly of tight junction proteins such as ZO1 (ref.62). Increased serum levels of zonulin together with elevated levels of LPS have been detected in individuals with type 2 diabetes, obesity63,64,65 or acute or chronic cardiovascular disease15 compared with the levels in healthy individuals. Increased gut permeability and low-grade LPS endotoxaemia related to gut dysbiosis have also been detected in patients with non-septic pneumonia19,66,67 or in the acute phase of myocardial infarction, as well as in experimental models of intestinal anoxia40,68 and systemic inflammation-associated overproduction of pro-inflammatory cytokines, such as interferon-γ and tumour necrosis factor (TNF)67 (Fig. 2).

Of note, given that analysis of serum zonulin levels is an indirect measure of gut permeability, other assays should be used to assess changes in gut permeability in humans. In this regard, analysis of d-lactate might be an alternative approach because an increase in d-lactate levels in the blood is considered to be a marker of increased intestinal permeability secondary to bacterial infection or experimentally induced gut injury69,70. However, measurement of urinary excretion of dextrose and mannitol after oral ingestion is likely to be a better approach for analysing gut burrier dysfunction because dextrose and mannitol are absorbed through the paracellular and transcellular pathways, respectively71, but such an analysis is cumbersome and requires specific expertise72. Nevertheless, the relationship between gut permeability and low-grade endotoxaemia in patients at risk of or with cardiovascular disease should be further investigated.

LPS and atherosclerosis

Atherogenesis is a complex process that is thought to be initiated by LDL crossing the arterial wall, where LDL is trapped in the subendothelial layer and undergoes oxidative changes that lead to the formation of oxidized LDL (oxLDL)73. OxLDL elicits recruitment of inflammatory cells, production of inflammatory cytokines and a shift in endothelial cells to a pro-atherogenic phenotype, all of which perpetuate the inflammation in the arterial intima74. The relevance of oxidative stress in atherogenesis has been supported by evidence of impaired macrophage uptake of oxLDL in patients undergoing carotid endarterectomy who received vitamin E (which has antioxidant properties) and injected with native radiolabelled LDL75.

Experimental studies have suggested that gut dysbiosis is involved in the aforementioned mechanisms of atherosclerosis by promoting the development of metabolic diseases that favour arterial inflammation, such as dyslipidaemia, obesity and hypertension10,76. Pathogenic gut microbiota are more frequent in patients with symptomatic atherosclerosis than in those with asymptomatic atherosclerosis, and are predictive of the risk of coronary heart disease76,77,78. Further support for the connection between gut microbiota and cardiovascular disease has been provided by studies in germ-free mouse models that have shown a causal link between gut dysbiosis, hypertension, vascular dysfunction, systemic inflammation and atherothrombosis10,51,79.

In this context, data on the role of LPS in atherosclerosis might provide insights into the mechanisms linking gut microbiota to atherosclerosis and its complications. LPS is a pro-atherogenic molecule through its pro-oxidant properties, which are mediated by the activation of NOX2 (ref.80), the catalytic core of NADPH oxidase, among the most important cellular producers of reactive oxygen species (ROS). LPS concentrations similar to those found in the plasma of individuals without sepsis amplified platelet responses in vitro to common agonists via TLR4-mediated, NOX2-derived ROS formation, which is a mechanism favouring LDL oxidation81,82,83. A significant correlation between the presence of low-grade endotoxaemia and elevated circulating levels of oxLDL has been found in patients with impaired fasting glucose84, which provides indirect evidence for the pro-oxidant properties of LPS.

Support for the putative role of LPS in atherosclerosis has been provided by immunohistochemistry analysis of carotid atherosclerotic plaques from patients undergoing endarterectomy, which revealed the presence of LPS adjacent to plaque macrophages with high TLR4 levels85. By contrast, LPS was not detected in atherosclerosis-free thyroid arteries from the same patients85. In humans, LPS in circulation is mostly bound to lipoproteins (80–97%)86,87, with the highest concentration in LDL (35.7%) and the lowest in VLDL (13.9%)88. However, VLDL particles carry a higher number of LPS molecules88. The observation that circulating LPS is transported by pro-atherogenic lipoproteins, such as VLDL and LDL, might be relevant in the pathogenesis of atherosclerosis. LPS could enter the arterial wall bound to these pro-atherogenic lipoproteins, which would favour LDL oxidation89 and thereby contribute to propagation of arterial inflammation. A further contributor to the pro-atherogenic process is LPS-binding protein-mediated LPS transfer from HDL to LDL26 (Fig. 3).

Fig. 3: Mechanisms of LPS-mediated atherosclerosis.
figure 3

a | LDL can cross into the arterial wall and undergo oxidation in the subendothelial space, leading to the formation of oxidized LDL (oxLDL). OxLDL is taken up by macrophages, inducing foam cell formation and inflammatory cytokine production. b | Lipopolysaccharides (LPS) can cross into the arterial wall, either together with LPS binding protein (LBP) or by LBP-mediated LPS transfer from HDL to LDL particles. LPS binds to Toll-like receptor 4 (TLR4) in several cell types, leading to phosphorylation of Toll–interleukin-1 receptor domain-containing adaptor protein (TIRAP) and recruitment of the myeloid differentiation primary response protein 88 (MyD88) to the cytoplasmic domain of TLR4. Downstream signalling induces the activation of the transcription factor nuclear factor-κB (NF-κB), which increases the production of pro-inflammatory cytokines, such as IL-8 and tumour necrosis factor (TNF); oxidative stress via upregulation of NADPH oxidase 2 (NOX2)-derived reactive oxygen species, which further promotes LDL oxidation; and destabilization of the atherosclerotic plaque via activation of the arachidonic acid pathway and biosynthesis of leukotriene B4 (LTB4), which attract leukocytes to the atherosclerotic lesion. O2, superoxide.

Experiments in animal models have been conducted to substantiate the role of LPS as a trigger of atherosclerosis. One study found that a single LPS infusion induced a profound systemic inflammatory response without changes to the atherosclerotic plaque90. However, other experiments have consistently shown an association between LPS infusion and arterial injury91,92. Daily intravenous or intraperitoneal LPS infusions in animals accelerated atherosclerosis in the aorta, together with increased production of pro-inflammatory cytokines, such as IL-8 and TNF, and autoantibodies against oxLDL, and increased accumulation of activated lymphocytes and deposition of IgG and IgM in the arterial intima91,92. In isolated human saphenous vein samples, LPS at concentrations as low as 0.1 ng/ml increased the production of ROS and chemotactic cytokines, such as IL-8 and CCL2, via interaction with TLR4 (ref.93). These changes were mitigated by administration of statins, which are known to have antioxidant properties beyond their LDL-lowering effect94. The relevance of the LPS–TLR4 axis in atherogenesis is further supported by studies in mice. In atherosclerosis-prone hypercholesterolaemic Apoe−/− mice, LPS-induced TLR4 activation increased neointima formation, whereas lack of either the adaptor molecule MyD88 or its upstream receptor TLR4 reduced atherosclerotic burden95,96. Tlr4−/−Ldlr−/− mice fed a carbohydrate-rich diet or a control diet had markedly reduced aortic atherosclerotic lesion areas compared with Ldlr−/− mice97.

In accordance with these experiments, the analysis of human atherosclerotic plaques revealed TLR4 overexpression in several cells types, including macrophages, vascular smooth muscle cells and dendritic cells14,98. However, it is unclear whether TLR4 overexpression is mediated by LPS or other TLR4 ligands, such as oxLDL, cleaved fibrinogen or heparin sulfate proteoglycan. Furthermore, studies in patients with the Asp299Gly variant in the TLR4 region on chromosome 9, which is associated with impaired TLR4 signalling, have provided inconclusive results14.

LPS has also been shown to destabilize atherosclerotic lesions, which renders the atherosclerotic plaque more vulnerable to rupture or erosion. Intratracheal injection of LPS in Apoe−/− mice fed a Western diet99 induced progression from stable to unstable phenotypes in aortic arch atherosclerotic plaques; histological features of plaque vulnerability included reduced cap thickness, leukocyte infiltration, increased necrotic core and thrombus formation at the edges of atherosclerotic plaque99. Consistent with these findings, a marked increase in atherosclerotic plaque size was detected in hypercholesterolaemic mice after a single LPS infusion mimicking the acute endotoxaemia detectable during infections100. Plaque instability was caused by acute inflammation of the arterial wall via leukocyte infiltration and the ensuing formation of neutrophil extracellular traps (NETs)100. A potential mechanism for LPS-induced plaque instability is activation in atherosclerotic lesions of the arachidonic acid pathway that leads to biosynthesis of leukotrienes, such as leukotriene B4 (LTB4), which are potent attractants of leukocytes101. In Apoe−/− mice, injection of 1.5 mg/kg of LPS each day for 5 days resulted in overproduction of LTB4 in atherosclerotic plaques, infiltration of leukocytes into the subluminal area and plaque destabilization, as indicated by increased collagen digestion, increased necrotic core size and reduced cap thickness101. The relevance of LTB4 in leukocyte-mediated atherosclerotic plaque progression was indicated by attenuation of arterial inflammation in mice with a deficiency in polyunsaturated fatty acid 5-lipooxygenase, the enzyme that converts arachidonic acid to LTB4 (ref.101) (Fig. 3).

LPS and thrombosis

Studies have shown that gut microbiota might be implicated in the atherothrombotic process and suggested production of TMAO as a potential mechanism79,102. LPS translocation into the systemic circulation might be another pathway.

The relationship between low-grade endotoxaemia and thrombosis was first hypothesized based on observations in patients with advanced liver disease, who have an increased risk of bleeding and thrombotic complications in the portal and peripheral circulation103,104,105. Low-grade endotoxaemia was detected in both vascular trees in these patients and was associated with increased biosynthesis and activity of tissue factor in monocytes106, a glycoprotein that converts factor X to factor Xa after binding to factor VIIa107. Stimulation of human monocytes with LPS in vitro resulted in increased thrombin generation, an effect that was blunted by treatment with the antioxidant vitamin E108. In endothelial cells in vitro, LPS addition caused a shift to a prothrombotic phenotype characterized by overexpression of tissue factor and overproduction of thrombin-activatable fibrinolysis inhibitor (also known as carboxypeptidase B2) and plasminogen activator inhibitor 1 (refs.109,110). Overexpression of tissue factor by LPS-treated mouse endothelial cells in vitro was related to LPS binding to TLR4 because tissue factor activity and mRNA were significantly higher in wild-type endothelial cells than in Tlr4−/− endothelial cells111. Under the same experimental conditions, LPS-induced tissue factor activation was inhibited in wild-type endothelial cells by pre-treatment with an anti-TLR4 antibody111. Further evidence of the prothrombotic effects mediated by LPS comes from an in vitro study showing that stimulation of human endothelial cells with LPS induced the release of prothrombotic molecules, such as von Willebrand factor and factor VIII, via formation and secretion of Weibel–Palade bodies112 (Fig. 4). This phenomenon was blunted by pretreating the cells with a TLR4 inhibitor.

Fig. 4: Mechanisms of LPS-mediated thrombosis.
figure 4

Low-grade endotoxaemia caused by lipopolysaccharide (LPS) promotes thrombus formation at the site of atherosclerotic plaque rupture or erosion by binding to Toll-like receptor 4 (TLR4) in different cell types, including endothelial cells, monocytes, neutrophils and platelets. a | In endothelial cells, TLR4 activation induces the release of von Willebrand factor (vWf) and factor VIII (FVIII) via formation and secretion of Weibel–Palade bodies (WPb) and the upregulation of tissue factor (TF) secretion, which converts factor X (FX) to activated factor X (FXa) to generate thrombin. b | In monocytes, TLR4 induces upregulation of TF release. c | In neutrophils, TLR4 signalling triggers the formation of neutrophil extracellular traps (NETs) and cathepsin G-mediated platelet activation. d | In platelets, LPS-mediated TLR4 signalling leads to platelet activation via NADPH oxidase 2 (NOX2)-mediated oxidative stress. FVIIa, activated factor VII; FIX, factor IX; FIXa, activated FIX; MyD88, myeloid differentiation primary response protein 88; NF-κB, nuclear factor-κB; PAR4, proteinase-activated receptor 4; O2, superoxide; TIRAP, Toll–interleukin-1 receptor domain-containing adaptor protein.

Further support for a role of the LPS–TLR4 axis in thrombosis has been provided by immunohistochemistry analysis of coronary thrombi from patients with myocardial infarction15. The analysis revealed the presence of LPS within the thrombus alongside the overexpression in leukocytes of TLR4 and cathepsin G, a protein favouring leukocyte–platelet binding and activation113,114. The interplay between LPS and leukocytes might also contribute to thrombosis via release of NETs (Fig. 4), which comprise DNA and histones and are released upon neutrophil stimulation by pattern-recognition receptor activation or chemokines115. The release of NETs requires ROS formation and calcium mobilization into the neutrophil, which in turn activate protein–arginine deaminase type 4 (PAD4) to deaminate arginine residues on histones116. LPS stimulates NET formation in vitro in a dose-dependent manner117 and requires binding to TLR4 and increased NOX-generated ROS, as indicated by the observation that the TLR4 inhibitor TAK242 and the NOX inhibitor diphenyleneiodonium suppress LPS-mediated NETosis118.

To explore whether the prothrombotic effect of LPS could be recapitulated in vivo, we developed a mouse model of low-grade endotoxaemia that involved intraperitoneal LPS injection (0.5 mg/kg) at a concentration that corresponded with the LPS level detected in human thrombus (40 pg/ml)15. In LPS-treated animals, thrombus growth was accelerated and was associated with increased levels of systemic biomarkers of platelet activation15. Both changes were inhibited by co-administration of a TLR4 inhibitor, reinforcing the hypothesis that TLR4 has a pivotal role in the prothrombotic effect of LPS15. This finding is consistent with the significant correlation between platelet TLR4 upregulation and low-grade endotoxaemia detected in patients with coronary thrombosis119 (Fig. 4).

Other components of the TLR family are also implicated in the thrombotic process mediated by gut microbiota. For instance, TLR2 recognizes and is activated by lipoprotein components of Gram-positive or Gram-negative bacteria and has direct pro-thrombotic effects98. Germ-free or TLR2-deficient mice had reduced thrombus growth after carotid artery injury compared with controls, and this effect was counteracted by intestinal colonization with microbiota120. However, the interplay, if any, between LPS and TLR2 in this context remains to be established.

LPS and cardiovascular events

Circulating LPS levels have been assessed in the general population121,122, in patients at risk of cardiovascular events18,123 and in patients with metabolic diseases31,65,124,125, clinically overt cardiovascular disease (such as atrial fibrillation126,127 or myocardial infarction15,68) or acute infections19,66,128. The association between circulating LPS levels and atherosclerotic burden and the clinical sequelae of atherosclerosis have been assessed in cross-sectional studies and prospective analyses (Table 1).

Table 1 Low-grade endotoxaemia in various clinical settings

Wiedermann and colleagues were among the first to investigate the effect of low-grade endotoxaemia on the risk of atherosclerosis in a study that included 516 individuals with risk factors for atherosclerotic disease18. After 5 years of follow-up, individuals with LPS >50 pg/ml at baseline had a threefold increased risk of incident carotid artery atherosclerosis compared with individuals with LPS levels <50 pg/ml. These findings were supported and extended by a study that included 2,568 individuals without previous cardiovascular disease121. After 10 years of follow-up, baseline serum levels of LPS-binding protein were significantly associated with incident cardiovascular disease, after adjusting for conventional risk factors for cardiovascular disease121. This finding was corroborated by a study that analysed circulating LPS levels in 7,927 individuals from a population-based chronic disease risk factor survey123. The study found a significant association between the LPS to HDL cholesterol ratio and cardiovascular events after a follow-up of 10 years123. Evidence for a relationship between endotoxaemia and thrombosis was also provided by an analysis of the genetic profile linked to endotoxaemia and its association with thrombosis in participants of three Finnish cohorts122. The study identified single‐nucleotide polymorphisms at five genetic loci that were associated with serum endotoxin activity; several of the single‐nucleotide polymorphisms were associated with expression of nearby genes affecting the clotting system. Interestingly, a significant association between the genetic risk score of endotoxaemia and venous thromboembolism was also identified122.

Low-grade endotoxaemia has also been investigated in patients with stable or unstable cardiovascular disease in cross-sectional and prospective studies15,68. Elevated circulating levels of LPS and d-lactate, a marker of gut permeability, have been reported in patients in the early phase of acute myocardial infarction, suggesting that coronary ischaemia is a trigger for gut barrier dysfunction68. This hypothesis was corroborated by a study in an animal model of coronary ischaemia, in which coronary ischaemia was associated with increased gut permeability and downregulation of tight junction proteins68. Furthermore, in patients with myocardial infarction, a significant association was observed between circulating LPS levels detected during the acute phase of the disease and major adverse cardiovascular events after 3 years of follow-up68. Furthermore, elevated circulating LPS levels in the peripheral and coronary circulation have been observed in patients with myocardial infarction compared with the levels in patients with stable cardiovascular disease and healthy individuals15. Of note, circulating LPS levels correlated with the serum levels of zonulin and several markers of inflammation, such as IL-1β and TNF, reinforcing the hypothesis that coronary ischaemia is associated with intestinal barrier dysfunction15.

A growing body of evidence indicates that low-grade endotoxaemia occurs in patients with metabolic disease, such as type 2 diabetes, obesity, NAFLD and NASH31,65,124 (Box 2). A systematic review of studies in patients with diabetes and obesity found a significant association between circulating LPS levels and triglyceride and total cholesterol levels, fasting glycaemia, insulinaemia, HbA1c levels and C-reactive protein levels124. The elevation in circulating LPS levels was more marked in patients with macroalbuminuria and was reduced in patients taking hypoglycaemic drugs, and insulin therapy exacerbated endotoxaemia compared with other antidiabetic drugs124. In a prospective study involving 3,781 patients with type 1 diabetes, the number of antibiotic purchases, a surrogate of bacterial infection, as well as high LPS activity were independently associated with incident cardiovascular events during a follow-up of 15 years125.

Metabolic endotoxaemia can also have a negative effect on liver cells, in which LPS can localize and elicit liver damage. Higher serum LPS levels and LPS hepatocyte localization have been detected in patients with NAFLD or NASH and in mouse models of these conditions compared with the levels in controls31. Immunohistochemistry analysis revealed a potential causal interplay between the liver steatosis and fibrosis and the TLR4 overexpression in macrophages and platelets31. Of note, a significant association was observed between platelet TLR4 levels and circulating LPS levels, suggesting a role for LPS in increasing platelet activation. This finding might provide insights on the role of platelets in liver inflammation as well as in cardiovascular events in patients with liver disease129,130.

Atrial fibrillation is another clinical setting in which the role of LPS as a trigger of arrhythmia or cardiovascular events has been explored. In patients with atrial fibrillation, a progressive increase in circulating LPS levels has been observed with increasing age, with high LPS levels found in middle-aged and older patients with atrial fibrillation127. A potential cause–effect relationship between endotoxaemia and atrial fibrillation was investigated in experimental models of atrial fibrillation, which demonstrated a role for gut dysbiosis in promoting atrial fibrillation in old rats and a reduction in atrial fibrillation susceptibility in old rats that received faecal transplantation from young animals127. Overexpression of TLR4 and activation of the NLRP3 inflammasome in the atria were key factors driving atrial fibrosis and increased susceptibility to atrial fibrillation127. The pro-inflammatory activity of LPS might also be relevant for the clinical outcomes of atrial fibrillation, as evidenced by a higher incidence of major adverse cardiovascular events during a follow-up of 3 years in patients with blood LPS levels >100 pg/ml than in patients with LPS <100 pg/ml126.

Low-grade endotoxaemia has also been detected in the acute phase of infection in patients with community-acquired pneumonia or with coronavirus disease 2019 (COVID-19), even in the absence of overt sepsis19,66. Of note, both community-acquired pneumonia and COVID-19 can be complicated by venous and/or arterial thrombosis during the hospital stay, with patients with COVID-19 having an incidence double than that seen in patients with community-acquired pneumonia128. In patients with COVID-19, circulating LPS levels correlated with serum zonulin levels and independently predicted the risk of venous and arterial thrombosis19.

Therapeutic implications

Potential therapeutic approaches to improving gut permeability and low-grade endotoxaemia have been proposed, but the clinical relevance of these options needs to be determined (Fig. 5). The Mediterranean diet, which is rich in fibre, is likely to be one of the best approaches to improving gut permeability and eventually eliminating low-grade endotoxaemia. Indeed, high adherence to the Mediterranean diet has been associated with beneficial microbiome-related metabolomic profiles, an increase in plasma SCFA levels and a reduction in circulating LPS levels compared with low adherence to the Mediterranean diet126,131,132. Administration of extra virgin olive oil, a key component of the Mediterranean diet, reduced endotoxaemia and serum zonulin levels and improved the metabolic profile of healthy individuals and patients with impaired fasting glucose133. These findings have been corroborated by an interventional study that found decreased circulating levels of LPS in patients with metabolic syndrome who received a fibre-rich diet for 75 days134.

Fig. 5: Potential therapeutic strategies to reduce low-grade endotoxaemia.
figure 5

Gut dysbiosis downregulates junction proteins, which leads to increased gut permeability and increased translocation of lipopolysaccharide (LPS) into the systemic circulation. Dietary and pharmacological interventions might improve gut barrier dysfunction by upregulating the levels of tight junction proteins and the zonula occludens protein ZO1 and thereby lower circulating LPS levels. Potential interventions include a Mediterranean diet, probiotics (live bacteria or microbial metabolites such as urolithin A), prebiotics (fermentable fibres), interventions that increase glucagon-like peptide 1 (GLP1) and GLP2 levels, and statins.

Prebiotics, which are plant-derived fibres with beneficial effects on gut microbiota, gut permeability and low-grade endotoxaemia, is another potential strategy to reduce low-grade endotoxaemia. Indeed, mice fed a diet supplemented with a fermentable fibre (oligofructose) showed improved gut barrier function and reduced circulating LPS levels compared with control mice135. Specific mechanisms that might account for the improved gut permeability included upregulation of ZO1 and occludin. An increase in glucagon-like peptide 1 (GLP1) and GLP2 levels is another intriguing and potential beneficial effect mediated by prebiotics. Indeed, mice fed a prebiotic-rich diet had an upregulation of GLP1 and GLP2 in the portal circulation compared with control mice, with a significant correlation between high GLP2 levels in the portal blood and improved gut barrier function135. This finding was recapitulated in another study in mice showing that GLP2 administration improved gut permeability, upregulated ZO1 and occludin levels in epithelial cells, lowered circulating LPS levels and reduced liver steatosis135. All these changes were associated with a reduction in systemic inflammation and oxidative stress135. However, the effect of these molecules on atherothrombosis remains to be established. Of note, administration of inulin, a prebiotic polysaccharide, in Apoe−/− mice slowed atherosclerosis progression and reduced liver steatosis136.

Another potential nutrition-related strategy to improve gut barrier permeability is the high intake of fish oils to increase the ratio of n-3 polyunsaturated fatty acids (n-3 PUFAs) to n-6 PUFAs137. However, the role of n-3 PUFAs in the modulation of gut permeability needs to be further clarified.

Other approaches to improving gut permeability include the use of probiotics, which are live bacteria, such as Lactobacillus plantarum or Akkermansia muciniphila, or microbial metabolites such as urolithin A or oxyberberine40. For example, urolithin A administration in mice improved gut barrier function via upregulation of claudin 4, occludin and ZO1 levels in intestinal epithelial cells, whereas oxyberberine administration downregulated the expression of Tlr4 and the production of pro-inflammatory cytokines such as Il-1β, IL-6 and TNF in colonic tissue40. Similar protective effects on barrier function have been achieved in vitro by incubation of epithelial cells with SCFAs, such as propionate, butyrate and acetate, that stem from dietary fibre and indigestible carbohydrate fermentation by commensal microbiota such as Bifidobacterium, Bacteroides, Enterobacter, Faecalibacterium and Roseburia40. In addition, administration of a diet rich in SCFAs to patients with metabolic syndrome resulted in upregulation of epithelial adhesion proteins40,138. However, the implementation of these strategies is so far limited to patients with inflammatory bowel disease40. Therefore, further studies are needed to assess the effect on gut permeability and low-grade endotoxaemia in patients at risk of or with cardiovascular disease.

Statins are lipid-lowering drugs that effectively reduce atherosclerosis and its complications139. Evidence indicates that statins have beneficial effects on gut barrier function by modulating gut dysbiosis or upregulating the levels of intestinal adhesion proteins140,141. Furthermore, in vitro experiments have shown that atorvastatin inhibits the pro-inflammatory effects caused by low-grade endotoxemia93. However, whether statins can counteract the effects of low-grade endotoxaemia in humans and the potential implication of these effects on the development of cardiovascular disease remain to be established.


Experimental data support the association between low-grade endotoxaemia, atherosclerosis and thrombosis, and indicate that gut dysbiosis-induced changes in intestinal permeability are a key step for LPS translocation into the systemic circulation. So far, different experimental assays for measuring LPS activity or concentration have been used for clinical purposes (Box 3), but no consensus exists about the methodology that better reflects LPS biological activity in humans142,143. Therefore, an internationally validated methodology for measuring blood LPS is mandatory to confirm and substantiate the relationship between low-grade endotoxaemia and cardiovascular events. In addition, targeting dysbiosis-induced gut barrier dysfunction to lower LPS translocation into the systemic circulation or impair LPS pro-inflammatory activity should be an important objective to establish whether a cause–effect relationship between low-grade endotoxaemia and cardiovascular events does exist.

In conclusion, the coexistence of low-grade endotoxaemia and atherosclerotic cardiovascular disease in patients provides further insight into the inflammatory processes underlying atherosclerosis. Gut permeability-related low-grade endotoxaemia might represent a novel tool to explore further the role of inflammation in the pathogenesis of atherothrombosis.