If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Inserm, UMR-970, Paris Cardiovascular Research Center, PARCC, Paris, FranceINSERM, UMR1149, Centre de Recherche sur l'Inflammation, Paris, FranceUniversity Paris Diderot, Paris, FranceService d'Hépatologie, Centre de Référence des Maladies Vasculaires du Foie, DHU Unity, Pôle des Maladies de l'Appareil Digestif, Hôpital Beaujon, AP-HP, Clichy, France
Non-alcoholic fatty liver disease (NAFLD) and its complications are an expanding health problem associated with the metabolic syndrome. Liver sinusoidal endothelial cells (LSECs) are highly specialized endothelial cells localized at the interface between the blood derived from the gut and the adipose tissue on the one side, and other liver cells on the other side. In physiological conditions, LSECs are gatekeepers of liver homeostasis. LSECs display anti-inflammatory and anti-fibrogenic properties by preventing Kupffer cell and hepatic stellate cell activation and regulating intrahepatic vascular resistance and portal pressure. This review focusses on changes occurring in LSECs in NAFLD and on their consequences on NAFLD progression and complications. Capillarization, namely the loss of LSEC fenestrae, and LSEC dysfunction, namely the loss of the ability of LSECs to generate vasodilator agents in response to increased shear stress both occur early in NAFLD. These LSEC changes favour steatosis development and set the stage for NAFLD progression. At the stage of non-alcoholic steatohepatitis, altered LSECs release inflammatory mediators and contribute to the recruitment of inflammatory cells, thus promoting liver injury and inflammation. Altered LSECs also fail to maintain hepatic stellate cell quiescence and release fibrogenic mediators, including Hedgehog signalling molecules, promoting liver fibrosis. Liver angiogenesis is increased in NAFLD and contributes to liver inflammation and fibrosis, but also to hepatocellular carcinoma development. Thus, improving LSEC health appears to be a promising approach to prevent NAFLD progression and complications.
Non-alcoholic fatty liver disease (NAFLD) encompasses a spectrum of conditions including simple steatosis and non-alcoholic steatohepatitis (NASH), defined as the association of steatosis, hepatocellular damage, inflammation and varying degrees of fibrosis.
A recent modelling approach estimated that NAFLD cases in the United States will expand from 83 million in 2015, corresponding to about 25% of the population, to 100 million in 2030, corresponding to more than 33% of the population.
While simple steatosis is generally benign, NASH can progress to both cirrhosis and end-stage liver disease. NASH is currently a leading cause of liver disease among adults awaiting liver transplantation in Europe and in the United States and is projected to become the most common indication for liver transplantation in the next decade.
Understanding the mechanisms of NAFLD, and in particular how simple steatosis progresses to NASH and then to cirrhosis and/or liver cancer, is of the utmost importance.
The current view of the pathogenesis of NASH centres on the response of hepatocytes to insulin resistance and lipotoxicity. The immune system and hepatic stellate cell activation are regarded as secondary events.
The vascular endothelium, representing the interface between blood and other tissues of the body, is not only a physical barrier but is implicated in different physiological roles, such as haemostasis, metabolite transportation, inflammation, thrombosis, angiogenesis and vascular tone. The liver endothelium is mainly formed of liver sinusoidal endothelial cells (LSECs) which are highly specialized endothelial cells at the interface between blood derived from the visceral adipose tissue and the gut, on the one side, and hepatic stellate cells and hepatocytes, on the other side. LSECs have a unique phenotype in the human body as they lack a basement membrane and have a multitude of fenestrae organized into sieves, that regulate the transport of macromolecules, including lipids and lipoproteins, across the sinusoid.
This review will specifically focus on the role of LSECs in the pathophysiology of NAFLD and its complications.
LSECs and simple steatosis
Role of LSECs in lipid transfer in the normal liver
Dietary lipids present in the circulation have to be transported through the vascular endothelium to be metabolized by tissues. In physiological conditions, LSECs are major regulators of the bidirectional lipid exchange between the blood and the liver parenchyma. First, LSEC fenestrae allow for efficient transfer of lipoproteins, chylomicron remnants (small lipoproteins derived from chylomicrons generated by enterocytes from dietary lipids), and other macromolecules, from the sinusoidal blood to the space of Disse, where they are taken up by hepatocytes.
LSEC fenestrae form a selective barrier for lipids. Indeed, older studies using radiolabelled lipoproteins showed that larger lipoproteins do not cross LSEC fenestrae and remain in the lumen of the sinusoid.
Different fate in vivo of oxidatively modified low density lipoprotein and acetylated low density lipoprotein in rats. Recognition by various scavenger receptors on Kupffer and endothelial liver cells.
One of the most remarkable phenotypic changes is the loss of fenestrae, also called defenestration or sinusoidal capillarization, associated with the formation of a basement membrane on the abluminal surface of LSECs. Several independent groups reported that sinusoidal capillarization appears very early in NAFLD.
Triggers for sinusoidal capillarization are not fully identified, but we can speculate that excessive dietary macronutrients, including lipids, carbohydrates, and gut microbiota-derived products play a role.
Cogger and coworkers demonstrated in mice challenged with several diets varying in content of macronutrients and energy that LSEC porosity and fenestrae frequency are inversely correlated with dietary fat intake, while fenestrae diameter is inversely correlated with protein or carbohydrate intake.
In vitro studies suggested that defenestration occurs following excessive lipid exposure. For instance, exposure of human primary LSECs to oxidized low-density lipoprotein (ox-LDL) reduces the diameter and the porosity of the fenestrae.
The effect of FFA on fenestrae has also been tested in primary rat LSECs, but firm conclusions cannot be drawn as the authors did not test several concentrations of FFA but rather the presence vs. the absence of FFA, which does not adequately mimic in vivo conditions.
Moreover, it has been shown that a single injection of endotoxin in rats induces a decrease in both diameter and number of fenestrae suggesting that gut microbiota-derived products may contribute to LSEC capillarization, although caution is needed since the concentration of endotoxin used in that study was high.
In turn, capillarization favours liver steatosis (Fig. 1), as observed in mice deficient in plasmalemma vesicle-associated protein (PLVAP), an endothelial-specific integral membrane glycoprotein required for the formation of endothelial fenestrae.
These mice also have hyperlipoproteinemia and increased triglyceridemia due to the retention of chylomicron remnants in the blood. As mentioned above, Vegfb-/- mice also exhibit a reduction in the number of LSEC fenestrae and less uptake of labelled oleic acid due to capillarization, but steatosis was not evaluated.
A first hypothesis explaining this consequence of capillarization on steatosis could be that reduced LSEC permeability impairs the passage of hepatocyte-derived very low-density lipoprotein toward the sinusoidal lumen, thus inducing cholesterol and triglyceride retention in the liver. However, these lipoproteins may escape the liver through the lymphatic system.
As an alternative explanation, Herrnberger et al. proposed that chylomicron remnants originating from the blood, and required for synthesis of very low-density lipoprotein by hepatocytes, cannot reach hepatocytes due to LSEC capillarization; their absence in hepatocytes might then stimulate de novo hepatic lipid synthesis and induce steatosis as a compensatory mechanism.
However, there is no available data to ascertain this hypothesis. Similarly, Fraser and collaborators postulated that, following LSEC capillarization, chylomicron remnants and dietary cholesterol no longer cross the fenestrae to inhibit HMGCoA reductase, the rate limiting enzyme for hepatocyte cholesterol biosynthesis, consequently activating endogenous cholesterol synthesis in hepatocytes.
The dynamic part is due to liver endothelial dysfunction. Endothelial dysfunction is a pathological condition, common to all vascular beds, defined as the inability of blood vessels to dilate in response to increased blood flow. Endothelial dysfunction is generally indicated by the loss of nitric oxide bioavailability due to eNOS (also called NOS3) inhibition.
Isolated-perfused liver experiments performed in these animals showed augmented portal perfusion pressure and reduced vasodilatory response to acetylcholine, indicating liver endothelial dysfunction. These changes were observed in the absence of inflammation and fibrosis, suggesting that endothelial dysfunction is an early feature associated with steatosis in NAFLD.
Intrahepatic vascular resistance is increased even when steatosis is the only histological feature of NAFLD. This is due to the combination of a compression of sinusoids by fat-laden enlarged hepatocytes and of a dysfunction of LSECs due to reduced nitric oxide bioavailability.
Several mechanisms could account for this liver endothelial dysfunction associated with steatosis (Fig. 2). First, LSECs dysfunction can be induced by overabundance of lipids during steatosis. In vitro experiments showed that stimulation of human primary LSECs with ox-LDL downregulates eNOS expression through the ox-LDL receptor, LOX1.
In addition, exposure of primary LSECs to palmitic acid also attenuates nitric oxide bioavailability through peroxynitrite production by NOX1, a nitric oxide consuming enzyme highly expressed in LSECs of mice fed a high-fat diet.
Interestingly, V-PYRRO/NO – a diazeniumdiolate ion metabolized in the liver that spontaneously decomposes to nitric oxide with a very short half-life at physiological pH and that triggers cyclic guanosine 3′,5′-monophosphate (cGMP) synthesis – improves hepatic microcirculation in mice with steatosis induced by a high-fat diet.
Third, the gut microbiota also seems to contribute to liver endothelial dysfunction. Indeed, Garcia-Lezana and colleagues demonstrated that restoration of a healthy microbiota via faecal transplantation normalizes portal hypertension by improving intrahepatic vascular resistance and endothelial dysfunction in rats.
Interestingly, therapies augmenting nitric oxide availability in the liver ameliorate steatosis. The V-PYRRO/NO or the improvement of nitric oxide/cGMP signalling with the phosphodiesterase-5 inhibitor sildenafil protect against liver steatosis in mice fed a high-fat diet.
In animal models, liver expression of VEGF and CD105, an endothelial cell marker, increase after 3 days of methionine- and choline-deficient diet in obese and diabetic db/db transgenic mice and after 1 week of this diet in C57BL6/J mice, before NASH appears.
This oxidative stress in LSECs contributes to NASH. Indeed, mice with a global deficiency in NOX1, which is highly expressed in LSECs in NAFLD, had attenuated liver lesions when fed a high-fat diet, as shown by lower serum ALT level and lower hepatic cleaved caspase-3 expression.
Therefore, in NASH, ROS production takes place not only in hepatocytes, but also to some extent in LSECs, and seems to contribute to hepatocyte injury.
Anti-inflammatory role of LSECs at initial stages of NASH
Progression of simple steatosis to steatohepatitis is accompanied by adhesion of leukocytes to the sinusoidal endothelium followed by infiltration of leukocytes within liver parenchyma to form inflammatory foci.
In vitro, both human and murine LSECs exposed to FFA for a short period (16 hours) exhibit a downregulation of pro-inflammatory chemokines involved in monocyte and macrophage recruitment, through a MAPK dependent pathway.
LSEC capillarization and dysfunction are permissive for establishment of liver inflammation. Indeed, mice deficient in eNOS exhibit an accelerated hepatic inflammatory response, while improving nitric oxide/cGMP signalling with the phosphodiesterase-5 inhibitor sildenafil or with simvastatin prevents liver inflammation in rodents fed a high-fat diet.
During NALFD progression, LSECs then acquire a pro-inflammatory phenotype and functions (Fig. 3). LSECs pro-inflammatory phenotype during NASH is characterized by progressive overexpression of adhesion molecules including ICAM-1, VCAM-1 and VAP-1 (AOC3) at the surface of LSECs, as observed in mouse models of NASH.
Second, the release of inflammatory mediators by LSECs contributes to the inflammatory response by activating neighbouring Kupffer cells, and by favouring recruitment, adhesion and transmigration of blood leukocytes.
LSECs’ expression of ICAM-1, VCAM-1 and VAP-1 is crucial for these interactions since in vivo and in vitro studies showed reduced leukocyte adhesion to hepatic sinusoids when these receptors are blocked or not functional.
Moreover, inhibition of the VCAM-1 ligand, VLA-4 (or ITGA4), on monocytes using an anti-VLA-4 antibody inhibits adhesion and transendothelial migration of monocytes across LSECs – from wild-type mice fed a high-fat diet and from ob/ob obese mice – and improves liver inflammation.
Although the stimuli responsible for LSECs’ inflammatory phenotype and functions in NASH are not firmly identified, several mediators are potential candidates. This includes products derived from the visceral adipose tissue, such as ox-LDL, FFA and adipokines. Indeed, in vitro studies showed that stimulation of LSECs with ox-LDL and FFA (palmitate) activate NF-kB and TLR-4, respectively.
Moreover, circulating concentrations of several adipokines, including TNFα and IL-6, are increased in the portal vein in the context of metabolic syndrome, and may contribute to LSECs inflammatory phenotype.
To summarize, while LSECs play an anti-inflammatory role in the initial stages of NAFLD, a switch towards pro-inflammatory functions occurs during the course of NAFLD development, paving the way for NASH progression.
(Fig. 4). Indeed, several studies reported the formation of new vessels in the liver of patients with NASH. Moreover, serum VEGF and sVEGFR1 levels are higher in patients with steatosis and biopsy-proven NASH than in healthy individuals.
Second, hepatocyte-derived microvesicles link lipotoxicity with angiogenesis. Indeed, hepatocytes exposed in vitro to excessive amounts of saturated FFA, that mimics steatosis, release microvesicles with a pro-angiogenic activity.
Likewise, mice fed a methionine- and choline-deficient diet have high circulating levels of hepatocyte-derived microvesicles able to induce angiogenesis. Third, angiopoietin-2 is another mechanism of liver angiogenesis in NASH. Angiopoietins are key regulators of angiogenesis. Although angiopoietins-1 and 2 contribute to vascular stability and quiescence in physiological conditions, angiopoietin-2 promotes pathological angiogenesis in inflammatory conditions.
Lefere and coworkers recently showed that serum angiopoietin-2 levels are increased in patients with NASH and correlate with liver steatosis, inflammation and hepatocyte ballooning, but not with liver fibrosis.
Similar findings were observed with 2 murine models of NASH, namely mice fed a methionine- and choline-deficient diet and mice with neonatal injection of streptozotocin followed by 16 weeks of western diet.
In turn, angiogenesis promotes inflammation since various strategies of inhibition of angiogenesis all improve liver inflammation (Fig. 4). Coulon and colleagues showed in a mouse model of NASH that treatment with anti-VEGFR2 antibody improves liver vasculature and decreases liver inflammatory gene expression, both using preventive and therapeutic approaches.
Importantly, this effect of L1-10 therapy is at least partly mediated by an effect on LSECs since L1-10 treatment downregulates VCAM-1, ICAM-1 and MCP1 expression in liver endothelial cells isolated from mice fed a methionine- and choline-deficient diet.
To summarize, inflammation stimulates angiogenesis that in turn worsens inflammation, as shown by the anti-inflammatory effect of anti-angiogenic therapies (Fig. 4).
LSECs in NASH-related liver fibrosis
Liver fibrosis is defined as the excessive deposition of extracellular matrix in liver parenchyma. The main mechanism leading to liver fibrosis is a long-standing wound healing process caused by hepatocellular injury and inflammation and mediated by hepatic stellate cell activation.
Hepatic stellate cells are nonparenchymal cells close to LSECs, in the space of Disse, which store retinoids in physiological conditions and shift their phenotype to an activated myofibroblastic state during liver injury and inflammation, wherein they secrete large amounts of extracellular matrix compounds.
As detailed above, LSECs are major effectors of liver inflammation in NASH, and consequently also promote hepatic fibrosis. For example, LSECs overexpress VAP-1 during inflammation which, in addition to its pro-inflammatory functions in NASH, is directly involved in hepatic stellate cell activation.
LSEC capillarization and dysfunction precede liver fibrosis and are permissive for it, through the loss of the ability of LSECs to maintain quiescence of hepatic stellate cells.
Experiments performed using cultured LSECs and hepatic stellate cells highlighted the importance of cross-talk between these cells types in regulating each other’s phenotype. While healthy LSECs maintain hepatic stellate cell quiescence, capillarized LSECs lose this ability
The Hedgehog pathway regulates capillarization, as inhibition of Hedgehog signalling prevents capillarization and partially reverts the phenotype of LSECs from a dedifferentiated state to their differentiated state.
LSECs are thus Hedgehog-sensitive cells, but they are also Hedgehog producing cells. Similarly, quiescent hepatic stellate cells are Hedgehog-sensitive cells, while activated hepatic stellate cells become Hedgehog-producing cells, which are also able to release macrovesicles loaded with Hedgehog signalling molecules that interact with LSECs.
It is thus tempting to speculate that during NASH, Hedgehog ligands are released by hepatocytes and LSECs, thus activating LSECs themselves, as well as quiescent hepatic stellate cells, by autocrine and paracrine effects. Activated hepatic stellate cells can then secrete Hedgehog molecules, promoting LSEC capillarization which in turn favours hepatic stellate cell activation, promoting the fibrogenic process.
LSECs dysfunction promotes liver fibrosis
Endothelial dysfunction appears very early in the course of NAFLD and precedes fibrosis in animal models of NASH.
Several lines of evidence suggest that liver endothelial dysfunction contributes to liver fibrosis. First, in rats fed a high-fat diet, simvastatin increases liver eNOS expression and ameliorates liver fibrosis.
suggesting that additional LSEC-derived factors could be responsible for the reversion of activated hepatic stellate cells to quiescence.
To summarize, these data demonstrate that capillarization and LSEC dysfunction not only precede liver fibrosis, but also promote it (Fig. 5). In their differentiated state, LSECs are able to maintain hepatic stellate cell quiescence, making differentiated LSECs gatekeepers of fibrosis in NASH, as in other chronic liver diseases.
Liver endothelial-to-mesenchymal transition: a process promoting liver fibrosis?
Another important process that links endothelial cells to organ fibrosis is endothelial-to-mesenchymal transition, i.e. the mechanism by which endothelial cells convert into myofibroblasts and contribute to extracellular matrix deposition.
Capillarized LSECs secrete fibrogenic factors, such as TGF-β1, and extracellular matrix proteins, such as fibronectin and laminin, that may be considered as an endothelial-to-mesenchymal transition, as well as stimulating activation of neighbouring hepatic stellate cells.
In the liver disease field, only 1 study has described endothelial-to-mesenchymal transition in vivo, in patients with alcohol- or hepatitis C virus-related cirrhosis and in mice treated with carbon tetrachloride.
Liver angiogenesis correlates with the severity of liver fibrosis and promotes its development. Once established, fibrosis stimulates angiogenesis by increasing tissue hypoxia. Blocking pathologic angiogenesis prevents liver fibrosis.
Liver angiogenesis correlates with liver fibrosis in patients with NASH.
In turn, angiogenesis promotes liver fibrosis since several approaches inhibiting liver angiogenesis prevent NASH-related fibrosis (Fig. 4). First, in the study by Kitade and colleagues, neither angiogenesis nor fibrosis were observed in the absence of leptin signalling in a rat model of NASH.
Second, blocking the release of pro-angiogenic microvesicles from fat laden-hepatocytes or inhibiting their binding to target cells protects mice from steatohepatitis-induced pathologic angiogenesis and results in a reduction in liver fibrosis.
Third, Zhou and coworkers recently showed that a specific deletion of the physiological inhibitor of angiogenesis, prolyl-hydroxylase-2, in endothelial cells results in an overexpression of angiopoietin-2 and TGF‐β1 in the liver and promotes dietary-induced liver fibrosis in mice.
Whether this pro-fibrotic effect of endothelial prolyl-hydroxylase-2 deficiency is directly induced by promoting angiogenesis remains to be demonstrated. Fourth, 2 studies reported that inhibiting angiotensin-II receptor using telmisartan or candesartan inhibits liver angiogenesis and fibrosis in rats fed a choline-deficient, L-amino acid-defined diet.
Finally, Lefere and colleagues demonstrated that blocking angiogenesis by inhibiting the interaction between angiopoietin-2/Tie2 using the L1-10 peptibody improves liver fibrosis in preventive and therapeutic strategies in mice fed a methionine- and choline-deficient diet. Therapeutic application of L1-10 peptibody also prevents liver fibrosis in diabetic mice with NASH (streptozotocin and western diet model).
In most cases, HCC develops on a background of chronic liver disease (70–90% of all patients). The role of liver endothelial cells in HCC development, outside the NAFLD setting, has been reviewed elsewhere and is summarized in Fig. 6.
Interestingly, Laouirem and colleagues recently demonstrated that LSECs exposed to conditions mimicking NAFLD – namely high concentrations of glucose, insulin, or VEGFA – release FABP4. They also observed that FABP4 released by LSECs exerts pro-oncogenic effects, since it induces hepatocyte proliferation. In mice fed a high-fat diet, specific inhibition of FABP4 reduces HCC growth.
We can speculate that FABP4 from LSECs not only contributes to HCC progression but also to HCC development in a NAFLD setting (Fig. 7).
In NAFLD, angiogenesis is highly stimulated and promotes NAFLD-associated HCC, since various inhibitors of angiogenesis all prevent HCC development. First, leptin-mediated angiogenesis has been demonstrated to be involved in HCC development as neither angiogenesis nor HCC develop in the absence of leptin signalling in Zucker rats fed a choline-deficient, L-amino acid-defined diet.
Second, Yoshiji and colleagues showed that a conventional anti-angiogenic treatment with sorafenib inhibits the appearance of preneoplastic lesions in rats fed a choline-deficient, l-amino acid-defined diet.
Third, Tamaki and colleagues demonstrated that inhibition of angiotensin-II receptor with telmisartan inhibits HIF-1α activity and VEGF expression and prevents HCC development in the liver of rats fed a choline-deficient, L-amino acid-defined diet for 48 weeks.
NALFD associated angiogenesis promotes HCC. Blocking pathologic angiogenesis prevents HCC development and progression.
Gaps in knowledge and future directions
Even if our understanding of the role of LSECs in NAFLD has progressed over the last years, several aspects remain elusive. First, triggers responsible for LSEC alterations in NAFLD are mostly unknown. It has been suggested that mediators derived from the visceral adipose tissue and the gut are responsible, but this has not been convincingly established. Indeed, available in vitro studies considered each mediator individually and not in combination, as in vivo in the portal venous blood.
Second, mechanisms underlying endothelial changes in NAFLD, including capillarization, need to be defined which might provide new therapeutic targets for NAFLD. Third, the role of LSECs in NASH-related cirrhosis has not been specifically investigated. Whether LSEC function and phenotype differ in cirrhosis related to NASH from cirrhosis related to other causes remains to be assessed.
Fourth, although NAFLD is well recognized as favouring HCC development, we are still at a very early stage of understanding how LSEC changes might favour HCC development.
LSECs are gatekeepers of liver homeostasis in physiological conditions. In NAFLD, sinusoidal endothelial alterations, including capillarization and LSEC dysfunction, occur early in disease progression, at the stage of simple steatosis. These initial changes, triggered by lipotoxicity, adipokines, inflammation and gut microbiota-derived products are associated with a loss of the ability of LSECs to prevent liver inflammation and fibrosis associated with NASH. Indeed, altered LSECs fail to maintain Kupffer cells and hepatic stellate cells in a quiescent state. At the stage of NASH, altered LSECs contribute to liver angiogenesis, inflammation, fibrosis and HCC. Improving LSEC health has great therapeutic potential for NAFLD. The current challenge is the identification of strategies to specifically target LSECs in order to modulate their activity.
This work was supported by the “ Institut National de la Santé et de la Recherche Médicale ” (ATIP AVENIR), Paris Descartes University , the “Agence Nationale pour la Recherche” ( ANR-14-CE12-0011 , ANR-14-CE35-0022 , ANR-18-CE14-0006-01 ) and by the “ Association Française pour l’Etude du foie ” ( AFEF 2014 ). A.H. was supported by the “ Ministère de l’Enseignement Supérieur et de la Recherche ”.
Conflict of interest
The authors declare no conflicts of interest that pertain to this work.
Please refer to the accompanying ICMJE disclosure forms for further details.
A.H. and P-E.R wrote the manuscript.
The following are the Supplementary data to this article:
Different fate in vivo of oxidatively modified low density lipoprotein and acetylated low density lipoprotein in rats. Recognition by various scavenger receptors on Kupffer and endothelial liver cells.