Emerging roles of vascular endothelium in metabolic homeostasis

Abstract

Our understanding of the role of the vascular endothelium has evolved over the past two decades, with the recognition that it is a dynamically regulated organ and that it plays a nodal role in a variety of physiological and pathological processes. Endothelial cells (ECs) are not only a barrier between the circulation and peripheral tissues, but also actively regulate vascular tone, blood flow and platelet function. Dysregulation of ECs contributes to pathological conditions such as vascular inflammation, atherosclerosis, hypertension, cardiomyopathy, retinopathy, neuropathy and cancer. The close anatomical relationship between vascular endothelium and highly vascularized metabolic organs/tissues suggests that the crosstalk between ECs and these organs is vital for both vascular and metabolic homeostasis. Numerous reports support that hyperlipidemia, hyperglycemia and other metabolic stresses result in endothelial dysfunction and vascular complications. However, how ECs may regulate metabolic homeostasis remains poorly understood. Emerging data suggest that the vascular endothelium plays an unexpected role in the regulation of metabolic homeostasis and that endothelial dysregulation directly contributes to the development of metabolic disorders. Here, we review recent studies about the pivotal role of ECs in glucose and lipid homeostasis. In particular, we introduce the concept that the endothelium adjusts its barrier function to control the trans-endothelial transport of fatty acids, lipoproteins, lipoprotein lipases, glucose and insulin. In addition, we summarize reports that ECs communicate with metabolic cells through EC-secreted factors and we discuss how endothelial dysregulation contributes directly to the development of obesity, insulin resistance, dyslipidemia, diabetes, cognitive defects and fatty liver disease.

Vascular endothelium is well-situated to regulate metabolic homeostasis

Blood vessels are broadly divided into two categories, macrovasculature and microvasculature¹. The macrovasculature is composed of large vessels, such as arteries and veins, and is responsible for carrying blood towards or away from organs. The microvasculature--small arterioles, capillaries and venules--primarily participates in the regulation of local blood perfusion and metabolic exchange between circulation and peripheral tissues^{1, 2}. In most organs, small arterioles (diameter 10–100 μm) are well innervated and highly responsive to sympathetic vasoconstriction. They play a major role in tuning the systemic vascular resistance and blood flow volume, ultimately influencing downstream capillary fluid exchange. Capillaries and post-capillary venules are main sites for fluid and macromolecular exchange. Due to the low shear stress force and selective expression of endothelial adhesion molecules, venules are also important sites for leukocyte adhesion^3–5.

The vascular endothelium is a thin layer of simple squamous cells lining the interior surface of the entire cardiovascular system, from the largest arteries and veins to the smallest capillaries. It forms a critical boundary between the circulating blood and underlying tissues. In the larger vessels, such as veins and arteries, the vessel wall consists of endothelial cells (ECs), smooth muscle cells and elastic fibers. In the capillaries, however, ECs constitute the entire blood vessel wall. They create a large surface area with relatively high permeability for the exchange of gases, fluids, electrolytes and macromolecules. There are three types of capillaries: continuous, fenestrated and discontinuous capillaries. Continuous capillaries are low permeability conduits located in skeletal muscle, heart, adipose tissue, lung, skin and the central nervous system, where intercellular gaps are tight and the basement membrane is continuous for ECs. Fenestrated capillaries are located at the choroid plexus of brain and endocrine organs where intercellular clefts exist between neighboring ECs but the basement membrane surrounding ECs is continuous. In liver, bone marrow and spleen, discontinuous capillaries (sinusoids) display gaps in both endothelial layer and basement membrane, therefore providing the highest permeability.

ECs serve as “first responders” in the modulation of vascular tone, permeability and interaction with circulating macromolecules. Activation of ECs is an upstream event necessary to expand the vascular network through the processes of vasculogenesis, angiogenesis and arteriogenesis. The close anatomical association between metabolic tissues/cells and ECs, especially microvascular ECs that play a critical role in metabolic exchange, suggests that they may possess potential functional interdependence. Dysregulation of ECs contributes to many cardiovascular, circulatory and blood diseases in response to metabolic stress and vascular injury^6–9. However, how ECs regulate metabolic homeostasis remains poorly understood. A growing body of evidence suggests that ECs are also actively involved in the regulation of metabolic processes. Dysregulation of ECs aggravates insulin resistance, adipose inflammation, diabetes, non-alcoholic fatty liver disease and other metabolic conditions. In this review, we will discuss about the emerging roles of ECs in the control of metabolic homeostasis and how endothelial dysregulation contributes to the development of metabolic disorders.

Metabolic processes involved in the regulation of energy homeostasis

Energy homeostasis is a balance of energy inflow (food intake) and energy expenditure, involving a variety of biological processes such as nutrient adsorption and transport, substrate metabolism, energy storage and clearance and appetite control. Energy provision (adenosine triphosphate, ATP) required for daily activities is mainly provided within the body by two types of macromolecules-carbohydrates and lipids. Major organs or tissues that are involved in the regulation of energy metabolism include the brain, gastrointestinal tract, liver, adipose tissue, pancreas and muscle. Tightly regulated cellular crosstalk in and between these organs or tissues plays a central role in maintaining energy homeostasis in vital organs and individual health. Excess calorie intake, stationary lifestyle, and genetic predisposition lead to disruption of energy balance, contributing to development of metabolic diseases such as obesity, metabolic syndrome, diabetes mellitus, insulin resistance and non-alcoholic fatty liver diseases.

Dietary carbohydrates and lipids are digested and processed into glucose and fatty acids (FAs). When glucose perfuses pancreatic endocrine tissue, it increases intracellular ATP/ADP in pancreatic beta cells and promotes insulin secretion¹⁰. Insulin in the circulation in turn activates insulin receptor-mediated signaling pathways in skeletal muscle, heart and adipose tissue, resulting in increased GLUT4-mediated glucose uptake^{11, 12}. Additional glucose is stored in skeletal muscle and liver in the form of glycogen^13–16. Glycogen breakdown in skeletal muscle meets the energy demands of muscle during exercise. In circumstances such as prolonged, low-intensity exercise, FA oxidation can also provide ATP to muscle cells¹⁷. In adipocytes, glucose is used to produce glycerol, along with free FAs (FFAs) delivered from the liver for TG synthesis. The liver acts as a hub to metabolically coordinate with various tissues, including skeletal and cardiac muscle and adipose tissue. Under feeding conditions when the liver is saturated with glycogen (~5% of liver mass), glucose is converted into FFAs, which are further esterified into TG, or amino acids in hepatocytes. Under fasted conditions or increased physical activity, glucagon secreted from pancreatic alpha cells activates glycogenolysis and gluconeogenesis signaling pathways in the liver to increase glucose production¹⁸. FFAs stored as TG in adipose tissue can also be mobilized through lipolysis and used by peripheral tissues, which is stimulated by glucagon and adrenaline and inhibited by insulin¹⁹. In addition, hormones released from peripheral tissues are integrated into the brain^{20, 21}, providing signals for appetite control by neurotransmitters and modulators such as serotonin (reviewed by Yeo GS et al²²). For the past several decades, it has been well recognized that the crosstalk between different metabolic organs/tissues plays an essential role in the regulation of metabolic homeostasis^{23, 24}. The signaling network responsible for intercellular communication includes the hormonal and nervous communication. However, the underlying molecular pathogenesis of the communications between organs/tissues during the development of metabolic diseases remains to be explored. This review will discuss recent data supporting that the vascular endothelium plays a crucial role in these communications between metabolic organs/tissues.

Endothelial responses to metabolic dyshomeostasis

Although the microvascular endothelium is critical for metabolic exchange, the detailed regulatory mechanisms of trans-endothelial transport of FAs, lipoproteins and metabolic hormones and regulators remain incompletely understood. The following sections highlight recent advances that describe how ECs directly control the trans-endothelial transport of these macromolecules by regulating endothelial barrier function. We particularly focus on molecular pathways that regulate crucial transporters and receptors of these transcytosis machineries. Since ECs actively communicate with surrounding cells in metabolic tissues by secreting metabolic factors/regulators, we also summarize recent reports about these endothelial-derived metabolic regulators (nitric oxide, extracellular matrix proteins, hormones, growth factors and enzymes). Last, ECs also regulate metabolic tissues indirectly through its own activation, proliferation and differentiation. We will discuss these processes briefly here since they have been reviewed elsewhere.

1. Endothelial barrier function in metabolic regulation

Vascular endothelium is anatomically interposed between the vascular space and peripheral tissues and therefore acts as a barrier regulating the exchange of macromolecules. However, a significant role of the vascular endothelium in control of nutrients (FAs, glucose, amino acids) and metabolic regulators (i.e., hormones and enzymes) across endothelial barrier for their delivery to peripheral tissues has not been well appreciated. Endothelial permeability is regulated by two distinct pathways—the transcellular pathway through which solutes are actively transported across the endothelial layer, or the paracellular pathway by which solutes passively move through the intercellular space between neighboring ECs. Recent studies unveil specific transporters, receptors and signaling pathways required for the transcellular transport of a variety of macromolecules such as FAs, lipoproteins, glucose, hormones (i.e. insulin, leptin) and enzymes (i.e. lipoprotein lipase) in ECs (Fig. 1). In this section, we review these transcytosis machineries for maintaining metabolic homeostasis.

Figure 1.

Trans-endothelial transport of FAs, lipoprotein, glucose and insulin.

1.1. Transport of FAs and lipoproteins

Most dietary lipids consist of long-chain FAs (LCFAs; ≥12 carbons), whereas FAs with short (<6 carbons) and medium-chains (with 6~12 carbons) are less frequently found in the diet and do not need active transport across cell membranes. It was once thought that FAs could enter cells by diffusion until the first putative FFA transporter (mAspAT/FABP_PM) was discovered in rat aortic ECs in 1991²⁵. Since then, many different types of FA transport proteins including fatty acid translocase (FAT/CD36), FATPs (FATP1–6) and intracellular fatty acid binding proteins (FABPs) have been discovered.

In the liver, a fenestrated sinusoidal endothelial layer allows direct access of circulating FFAs to the space of Disse. However, ECs lining fat-utilizing peripheral tissues such as cardiac and red skeletal muscle form a continuous and nonfenestrated layer, which raises the need for an efficient trans-endothelial transport system for FAs. In ECs, FATP3, FATP4, FABP3–5, FABP_PM and CD36 have been detected^26–31. Hagberg et al³² demonstrated that co-expression of FATP3 and FATP4 synergistically increases lipid accumulation in ECs, suggesting that FATP3 and FATP4 are required for LCFA uptake by ECs. VEGF-B secreted by neighboring cardiomyocytes induces FATP3 and FATP4 expression in ECs. Mice lacking VEGF-B exhibit decreased LCFA uptake and lipid deposition in muscle, but gain more weight due to the increased lipid accumulation in white adipose tissue (WAT)^{32, 33}. More interestingly, administration of anti-VEGF-B antibody 2H10 improves peripheral insulin sensitivity and decreases lipid accumulation in high-fat diet (HFD) fed rat or db/db diabetes mice. These studies for the first time demonstrate that targeting the barrier function of the vascular endothelium can be an effective strategy for treating insulin resistance and type 2 diabetes mellitus (T2DM). However, the translational path of inhibiting VEGF-B to treat dysglycemia requires further evaluation since VEGF-B has salutary effects on coronary arteriogenesis and is neuroprotective^34–36. Another paracrine stimulus for FA trans-endothelial transport is 3-hydroxy-isobutyrate (3-HIB), a catabolic intermediate of the branch-chained amino acid (BCAA)-valine and generated by skeletal muscle cells. Increased circulating levels of BCAA are implicated in the progression of diabetes in animals and patients³⁷. Elevated 3-HIB levels are also observed in skeletal muscle and serum of db/db mice or diabetic patients³⁸. Knockdown of HIBC hydrolase, which blocks the generation of 3-HIB, in myotubes almost completely inhibits FA uptake in ECs³⁸. In addition, knockdown of FATP3 and FATP4 reduces FA uptake in response to 3-HIB, suggesting that 3-HIB promotes FA transport across ECs through FATP3 and FATP4³⁸. However, how 3-HIB regulates the activity of FATP3 and FATP4 remains to be further explored.

FABP4 and FABP5 are exclusively expressed in ECs of capillaries and venules of cardiac and skeletal muscles, but not in arterioles, arteries or aorta³⁹. In addition, their expression is detectable in the capillary ECs of brown and white adipose tissue, but not in liver, brain and lung. In FABP4/5 double knockout mice, FA uptake in the heart and skeletal muscle is significantly decreased, whereas glucose uptake is dramatically elevated to compensate for the limited supply of FAs³⁹. Consistent with these findings, reduced FA uptake is observed in heart or liver-specific FABP knockout mouse models⁴⁰. However, about 60% of FA uptake remains in mouse hearts lacking both FABP4/5, indicating that additional trans-endothelial FA transport systems operate independently of FABP4/5. Nevertheless, the pivotal role of FABP4 in FA uptake and transfer through the endothelial layer is further supported by recent studies with apelin, EC-specific apelin receptor (APLNR), EC-specific FoxO1, and apelin; EC-specific FoxO1 double knockout mouse models⁴¹. In adult metabolic tissues, APLNR is expressed specifically in ECs. Apelin/APLNR triggers endothelial FoxO1 inactivation and blocks FABP4 expression. In vitro FA uptake and transfer studies indicate that the disruption of apelin signaling pathways increases FA transfer across microvascular ECs, which can be abrogated by the FABP4 inhibitor BMS309403. Consistent with these observations, increased FA accumulation occurs in the heart, skeletal muscle, liver and brown adipose tissue in apelin knockout mice⁴¹. EC-specific FoxO1 depletion could ‘rescue’ FA accumulation and increase glucose clearance in apelin; EC-specific FoxO1 double knockout mice. FAs are also accumulated in metabolic tissues of EC-specific APLNR knockout mice. Treatment with BMS309403 alleviates FA accumulation and thereby improves glucose utilization. Apelin also inhibits oleic acid-induced vascular hyperpermeability by stabilizing VE-cadherin-based adherens junctions of ECs⁴². In response to HFD feeding, apelin-deficient mice display an obese phenotype, whereas Apelin transgenic mice have reduced body mass and subcutaneous adipose tissue⁴². Collectively, these observations indicate that apelin signaling within ECs regulates metabolic homeostasis likely through multiple molecular mechanisms such as FA transcytosis and EC dysfunction.

CD36, also known as fatty acid translocase (FAT), was initially cloned as an adipocyte membrane protein that facilitates FA uptake⁴³. Cell culture studies demonstrate that CD36 also promotes the transport of FA across endothelial layer⁴⁴. CD36-null mice display a dramatic decrease of FA uptake and utilization in heart, skeletal muscle and adipose tissue^{45, 46}. In contrast, mice with CD36 overexpression in muscle exhibit increased FA oxidation in response to contraction⁴⁷. Cardiac-restricted overexpression of PPARα results in myocyte lipid accumulation and cardiac dysfunction, that can be rescued by the depletion of CD36 or heart-specific lipoprotein lipase (LPL)^{48, 49}. These data suggest that CD36-mediated FA uptake is rate-limiting for FA utilization in these metabolic tissues. Paradoxically, mice with depletion of CD36 have increased FA uptake in the liver and developed hepatic steatosis⁵⁰. The opposing effects of CD36 depletion on FA uptake by muscle, adipose tissue and liver still remain a mystery. It is likely that differential permeability of ECs in the liver (which are highly permeable) and muscle and adipose tissues (which are less permeable) contributes to this discrepancy. An EC-specific CD36 knockout mouse model will be required to dissect the role of CD36 in trans-endothelial transport of FAs and elucidate the role of this endothelial mechanism for the overall FA uptake in metabolic tissues.

PPARγ is a master transcriptional regulator of lipid metabolism, tuning gene expression involved in the release, transport and storage of FAs, such as LPL and CD36 (reviewed by Lehrke M et al⁵¹). The specific depletion of PPARγ in ECs leads to decreased expression of FA-handling proteins, including CD36 and FABP4^{52, 53}. Mice with PPARγ depletion in ECs exhibit increased serum levels of VLDL and FFAs. Uptake of FA analogues is significantly reduced in heart, red skeletal muscle and adipose tissues of these mice⁵², and lipid accumulation is reduced in adipose tissue and skeletal muscle, but not liver⁵³. These data suggest that PPARγ also plays a crucial role in FA transport across the nonfenestrated endothelium in skeletal muscle and adipose tissue. PPARγ agonists (TZDs) are popular drugs for treating diabetes. However, substantial clinical and preclinical experiences have indicated that TZDs lead to increased cardiovascular risk⁵⁴. PPARγ-induced FA accumulation and cardiac dysfunction may be responsible for this side effect, which could be tested in future studies by blocking FA transport with inhibitors of FABP4 or CD36.

The endothelial outer membrane surface is the location where most intravascular lipoprotein metabolism takes place. Chylomicrons are converted metabolically to chylomicron remnants and VLDL to LDL. Hydrolysis of TG-rich lipoproteins generates FAs, which are transported across the endothelium through multiple mechanisms as we discussed above. However, it is still not clear whether intact TG can be transported across the endothelium. It is known that lipoproteins are transported across the endothelium via coordinated receptor-mediated endocytosis and exocytosis. For example, LDL, acetylated LDL, oxidized LDL and VLDL can all be taken up by ECs through LDL receptors, the scavenge receptor and VLDL receptors⁵⁵. LDL is cleared mainly by the liver where tissue parenchymal cells have direct contact with blood. However, tissues containing non-fenestrated ECs are responsible for ~15–35% of LDL metabolism⁵⁶. LDLR is expressed on the apical side of microvascular ECs and promotes LDL transcytosis across the blood brain barrier⁵⁷. CD36 is a key scavenger receptor that is required for internalization of oxidized LDL in ECs. It can be induced by LDLR-related protein 1 (LRP1) through increasing PPARγ transcriptional activity⁵⁸. Depletion of LRP1 in ECs results in decreased oxidized LDL uptake. Mice with LRP1 depletion display improved lipid homeostasis. SR-BI, ATP binding cassette transporter G1 (ABCG1) and A1 are the main transporters for HDL. Recent studies have demonstrated that SR-BI and ABCG1 are involved in HDL/Apo-AI uptake and transcytosis in aortic ECs⁵⁹. Knockdown of SR-BI blocks HDL internalization in human brain microvascular ECs⁶⁰. However, the physiological role of SR-BI in endothelial transcytosis of HDL requires further characterization since SR-BI knockout does not alter the transfer of HDL-derived lipids into ECs of mouse liver⁶¹. Interestingly, SR-BI and ABCA1, but not ABCG1, are expressed in lymphatic ECs, a key player responsible for transporting cholesterol from subcutaneous tissues back to plasma and liver during reverse cholesterol transport. Lim et al⁶² performed studies suggesting that SR-BI but not ABCA1 is responsible for HDL uptake in lymphatic ECs. The transport of HDL to lymph nodes by lymphatic vessels is impaired in SR-BI knockout mice, resulting in cholesterol accumulation in the skin, strongly suggesting that lymphatic ECs play a critical role in HDL transport across endothelial layer and reverse cholesterol transport.

Currently the understanding of lipoprotein transport across the endothelium is still very limited, mainly because of the lack of good tools to visualize lipid transport directly. Fluorescently labeled lipids have been widely used to image lipids in live cells. However, the modification of lipids (i.e. BODIPY) might change the biochemical properties of the tagged lipids. Recently progress has been made to directly visualize HDL uptake in ECs in vitro and in vivo⁶³. The further development of high-resolution imaging techniques for lipids in cells and tissues will surely expedite our understanding of the molecular machinery for lipid transport across the endothelial layer of blood vessels.

1.2. Transport of LPL

A group of lipases, including lipoprotein lipase (LPL), hepatic lipase (HL) and endothelial lipase (EL), are required for lipoprotein metabolism. They bind to heparin and are anchored to the cellular surface to hydrolyze TG and phospholipids within circulating lipoproteins. These lipases are expressed differentially in many tissues and have unique preferences for lipoprotein substrates⁶⁴. EL mainly hydrolyzes phospholipids and is involved in HDL metabolism. HL hydrolyses both phospholipid and TG and is required for the metabolism of HDL, apolipoprotein (apo) B and LDL. LPL is the main lipase for TG and is responsible for the hydrolysis of VLDL and chylomicrons. LPL, secreted by adipocytes and myocytes, needs to be transported across the EC layer into capillary lumen for TG hydrolysis. A glycosylphosphatidyl-inositol-anchored protein called GPIHBP1 can bind LPL and is required for vesicle-mediated trans-endothelial transport of LPL and lipolysis of chylomicrons^{65, 66}. PPARγ is the transcription factor required for GPIHBP1 induction in ECs. Mice with PPARγ knockout display decreased expression of GPIHBP1 in ECs, inhibited LPL activity and TG and FA metabolism⁵³. A deficiency of GPIHBP1, due to missense mutations or blocked by its endogenous autoantibodies, prevents LPL from reaching the capillary lumen, resulting in low LPL plasma level and severe hypertriglyceridemia in patients (chylomicronemia, defined as a TG level >11.3 mM or 1000 mg/dL)^67–74. Cell culture studies indicate that this GPIHBP1-dependent LPL movement across endothelial layer is a caveolin-1 independent process. However, additional studies are required for better understanding of the molecular mechanisms by which GPIHBP1 facilitates the transfer of LPL across endothelial layer. Interestingly, recent data suggest that GPIHBP1 can also stabilize LPL activation by blocking angiopoietin-like protein 4-catalyzed unfolding of LPL’s hydrolase domain⁷⁵.

1.3. Transport of glucose

Glucose transport is regulated by adjusting local blood flow and trans-endothelial access, allowing the eventual delivery of glucose to metabolic cells. Vasodilation is induced by insulin through nitric oxide-dependent mechanisms in skeletal muscle, which contributes to both insulin sensitivity and responsiveness in non-obese humans^{76, 77}. This endothelium-dependent vasodilation is impaired by elevated plasma FFAs⁷⁸. Surprisingly, recent evidence indicates that trans-endothelial transport of glucose is a dynamically regulated process. Glucose uptake occurs primarily by facilitated diffusion, an energy-independent process that uses glucose transporters to move glucose across the cellular membrane⁷⁹. In ECs, the main glucose transporter is GLUT1. It has been generally accepted that the endothelium does not play a rate-limiting role in glucose metabolism, due to the incapability of ECs to downregulate the rate of glucose transport when exposed to high glucose levels⁸⁰. However, this concept has been challenged by recent studies. Prolonged exposure of ECs to high glucose decreases the rate of glucose transport in brain and heart ECs as a consequence of decreased GLUT1 expression^{81, 82}. Patients with genetic GLUT1 deficiency syndrome display abnormalities attributed to decreased trans-endothelial glucose delivery to the brain^{83, 84}. GLUT1 is a HIF1α-responsive gene in ECs⁸⁵. HIF1α depletion in ECs decreases glucose uptake and overexpressed GLUT1 can inhibit this decrease. Importantly, Endothelial-specific HIF1α knockout mice display decreased CSF/blood glucose ratios⁸⁵, which is a diagnostic criterion for clinical GLUT1 deficiency syndrome. These observations raise the intriguing possibility that the endothelium plays a critical role in regulating systemic glucose metabolism by controlling glucose transport across the endothelial barrier. Further investigation of the role of GLUT1 in ECs with emerging genetic tools will be vitally important for understanding the precise contributions of endothelial glucose transport in metabolic processes.

In insulin-sensitive tissues, glucose uptake is stimulated via insulin-dependent activation of its receptor (IR) and insulin receptor substrates (IRS), leading to translocation of glucose transporters⁸⁶. In ECs, insulin signaling activates two major pathways, phosphoinositide 3 kinase (PI3K)-Akt and RAS-mitogen activated protein kinase, to regulate vasoreactivity (reviewed by Kubota T et al⁸⁷). In vitro studies have shown divergent results about whether insulin signaling regulates glucose transport and metabolism in microvascular ECs. Insulin signaling does not regulate glucose transport in human micro- and macrovascular endothelial cells⁸⁸ and bovine brain and retinal ECs^{89, 90}, whereas glucose transport and glycogen synthesis are increased by insulin in bovine ECs isolated from adipose tissue or retinas or rabbit ECs^91–93. Recent in vivo studies demonstrate that mice with endothelial depletion of IR display no changes in fed and fasting blood glucose levels⁹⁴, which suggests that glucose uptake of ECs is unlikely regulated by insulin signaling.

1.4. Transport of hormones

Multiple pathways have been identified for the transport of hormones across the endothelial barrier. Compared to transcytosis, paracellular transport is a more passive mechanism for hormone transport. During inflammation, this passage is increased specifically in venules by opening junctions, which can be inhibited by hormones such as ghrelin⁹⁵. Adiponectin can also block endothelial hyperpermeability induced by angiotensin II or TNFα⁹⁶. In addition, studies indicate that transcytosis is required for trans-endothelial transport of hormones. For example, leptin is transported via transcytosis across capillary ECs of gastric intestinal tract or blood brain barrier^{97, 98}. Insulin is one of the best-studied hormones that is transported across the endothelial barrier through transcytosis, which we will discuss here as an example of how ECs regulate the transcytosis of hormones.

Trans-endothelial transport of insulin is a rate-limiting step in insulin-induced glucose uptake by skeletal muscle⁸⁷. Increased insulin action kinetics are observed in liver compared with skeletal muscle, suggesting that the fenestrated endothelium of liver capillaries allows free passage of insulin while continuous capillaries of skeletal muscle need an insulin transport mechanism. In ECs, insulin binds avidly to IR and also, at high concentrations, to insulin-like growth factor 1 receptor (IGF1R). Receptor-mediated transport of insulin is supported by in vitro studies indicating that transport of insulin across ECs is inhibited by blocking antibodies to IR or IGF1R⁹⁹. Earlier studies with endothelial specific knockout mice of IR or IGF1R, generated by Tie2-Cre-mediated depletion, failed to show differences in plasma insulin levels and glucose or insulin tolerance tests^{94, 100, 101}. However, recent studies using different genetic approaches have begun to unveil a pivotal role for receptor-mediated insulin transport in systemic glucose homeostasis. In an endothelial IR knockout mouse model generated by Cdh5-Cre-mediated depletion, the depletion of IR in ECs inhibits the delivery of insulin to target tissues where continuous endothelium barriers exist, such as skeletal muscle, brown fat, brain¹⁰². This inhibition results in impairment in glucose tolerance, brain-controlled appetite responses, and mild obesity. Intracellular insulin signaling mediators including PI3K and eNOS are required for insulin transport across ECs, indicated by their inhibitor studies¹⁰³. EC-specific knockout mice of IRS-2 display similar metabolic defects including glucose intolerance and impaired insulin action on target tissues¹⁰⁴.

Fcγ receptor IIB (FcγRIIB), the C-reactive protein (CRP) receptor, has recently been demonstrated to block endothelial insulin transport by antagonizing eNOS activity¹⁰⁵. Mice with endothelial-specific FcγRIIB depletion display improved glucose homeostasis in response to CRP. Nitric oxide can promote insulin uptake and across through ECs via nitrosylating PTP1B, an inhibitory phosphatase of insulin signaling¹⁰⁶. In addition, both caveolae- and clathrin-coated vesicles have been suggested to mediate the uptake and transcytosis of insulin across endothelial layer^{107, 108}. However, it is still not clear how these two pathways are regulated and whether they are regulated differentially in the vessel beds of different tissues. Although insulin transport across endothelial layer is tightly regulated through receptor-dependent processes, at very high levels insulin transport may occur via non-saturable processes such as passive diffusion^{109, 110}. However, it is unclear how this happens, and how much this independent diffusion contributes to the global regulation of insulin and glucose metabolism. Although the trans-endothelial transport of insulin remains to be further studied, available evidence has underscored the role of endothelial insulin signaling in the maintenance of whole body metabolic homeostasis. Targeting endothelial insulin transport could potentially become a therapeutic target for treating insulin resistance and T2DM.

Recent observations, especially with genetic approaches (Table 1), suggest that ECs play a critical role in controlling the transcytosis of FAs, lipoprotein, glucose and hormones such as insulin. However, our understanding is still very preliminary and a lot of questions remain to be answered. For example, is intact TG transcytosed through ECs? How do FABP, FATP and CD36 coordinate together for FA transport? Does the transport of these macromolecules take place bidirectionally? In addition, amino acids such as arginine and citrulline play an important role in metabolic homeostasis. The transport of amino acids across the endothelial monolayer also requires complicated transport systems, which has been reviewed elsewhere (i.e. Mann GE et al¹¹¹) and will not be discussed in this review. Yet we also know that paracellular transport likely contributes to the increased transport of macromolecules in response to vascular inflammation associated diabetes¹¹². In addition, the crosstalk of transcellular and paracellular routes exists and consequently alters endothelial barrier function¹¹³. The exact roles of paracellular transport of macromolecules in metabolic processes are still poorly understood and merit further investigation.

Table 1.

A summary of the genetic approaches used for evaluating the roles of trans-endothelial transport of macromolecules in metabolic homeostasis.

Cargo	Model System	Signaling Effects and Metabolic Phenotype	Ref.
FA	Mice of Vegfb−/−, Nrp1iEC−/− or db/db;Vegfb−/−; HFD-fed rat with 2H10 (Vegf-b neuAb) injection	FATP3/4 induction in ECs, decreased LCFA uptake and lipid deposition in muscle, ameliorated insulin resistance and T2DM	Hagberg et al.(2010, 2012)32, 33
	Knockdown of Hibch or Hibadh in myotubes or skeletal muscle	Changes in 3-HIB metabolism, FA uptake in ECs, TG plasma levels	Jang et al. (2016)38
	Db/db mice or diabetic human subjects	Increased 3-HIB levels	Jang et al. (2016)38
	Mice of apelin−/−, AplnriEC−/−, Foxo1iEC−/− and apelin−/−; Foxo1iEC−/−, BMS309403 (FABP4 Inhibitor)	FoxO1-dependent FABP4 induction in ECs, FA uptake and trans-endothelial transport, FA accumulation in metabolic tissues; glucose tolerance	Hwangbo et al. (2017)41
	HFD-fed apelin−/− or apelin transgenic mice	FoxO1-dependent FABP4 induction, Obese, EC dysfunction	Sawane et al. (2013)42
	HFD-fed PparγEC−/− mice	Decreased CD36 and FABP4 induction, increased VLDL and FFA levels, reduced lipid accumulation, improved glucose responses	Kanda et al.(2009)53, Goto et al. (2013)52
LDL	Ldlr−/− mice	LDLR-dependent transcytosis of LDL across BBB, CNS drug delivery	Molino et al. (2017)57
oxLDL	HFD-fed Lrp1EC−/− mice	Decreased expression of CD36, oxLDL uptake by ECs, improved lipid and glucose responses	Mao et al. (2017)58
HDL	SR-BI−/− mice	Compromised transport of HDL through lymphatic vessels, accumulation of cholesterol and HDL in plasma	Lim et al. (2013)62
LPL	Gpihbp1−/− mice	TG transport along heart capillaries, severe chylomicronemia	Beigneux et al.(2007)65
	Human loss-of-function mutations of GPIHBP1, GPIHBP1 autoantibodies in patients with chylomicronemia	Familial chylomicronemia, low plasma level of LPL	Beigneux et al.(2009,2015,2017)67–69 Olivecrona et al.(2010)71 Plengpanich et al.(2014)72 Rios et al.(2012)73 Surendran et al.(2012)74 Buonuomo et al.(2015)70
Glucose	HIF1αEC−/− mice	Decreased GLUT1 induction, glucose uptake and CSF/blood glucose ratios	Huang et al. (2012)85
Leptin	Normal and STZ-diabetic mice	Glucose and insulin increase leptin across BBB of normal but not STZ-diabetic mice	Kastin AJ et al. (2001)98
Insulin	IREC−/− mice	Delayed insulin signaling in muscle, fat and brain, hyperinsulinemia, impaired glucose tolerance, exacerbated HFD-induced insulin resistance	Konishi et al. (2017)102
	Mice of Irs2EC−/−, Irs1/2EC−/−, HFD-fed mice with beraprost sodium (BPS) injection	Reduced insulin delivery and glucose uptake in skeletal muscle, BPS induces eNOS and improves insulin delivery and glucose uptake in skeletal muscle of HFD-fed mice	Kubota T et al. (2011)104
	Mice of FcγRIIBEC−/−, FcγRIIBEC−/−;TG-CRP, eNOS (S1176D);TG-CRP	The CRP/FcγRIIB axis inhibits eNOS activity. Depleted FcγRIIB or eNOS constitutive activation (S1176D knock-in) improves CRP-induced glucose intolerance and insulin resistance.	Tanigaki K et al. (2016)105

iEC, inducible EC; STZ, streptozotocin; CSF, cerebrospinal fluid; TG, transgenic.

2. Endothelial cell-secreted factors regulating metabolic signaling

ECs play a crucial role in survival, angiogenesis and inflammation by secreting regulatory cytokines. Recent studies, including numerous genetic analyses (Table 2), indicate that ECs can also secrete factors that act as paracrine or endocrine regulators for metabolic cells/tissues and are actively involved in the maintenance of metabolic homeostasis. We group these EC-secreted factors into four different categories based on their biochemical characteristics and cellular functionality: nitric oxide (NO), insulinotropic factors, growth factors and enzymes.

Table 2.

A summary of representative genetic approaches used for evaluating the roles of EC-secreted factors in metabolic homeostasis.

Secreted Factor	Model System	Signaling Effect and Metabolic Phenotype	Ref.
NO	Foxo1/3/4EC−/− mice	Constitutive insulin signaling and NO production in ECs and increased tyrosine nitration of hepatic IR, eNOS inhibition (L-NAME) inhibits HFD-induced insulin resistance in Foxo1/3/4EC−/− mice.	Tsuchiya et al. (2013)114
	Ship2EC−/− mice	Decreased NO bioavailability and increased insulin resistance in fat and muscle	Watt et al. (2017)115
	eNOS gene variants	Metabolic syndrome, T2DM	González-Sánchez et al. (2007)116, Monti et al.(2003)117, Rittig et al.(2008)118
	eNOS−/− mice	Hyperinsulinemia, hyperlipidemia, insulin resistance,	Duplain et al.(2001)119 Cook et al.(2003, 2004)120, 121
	e/nNOS−/− mice	Insulin resistance	Shankar et al.(2000)122
	e/n/iNOS−/− mice	Visceral obesity, hypertriglyceridemia, glucose intolerance	Nakata et al.(2008)123
ECM proteins	Vegfaepi−/− mice	Few islet ECs, loss of intraislet basement membrane, less insulin secretion	Nikolova et al.(2006)124
	NOD mice	Degradation of islet basement membrane at onset of insulitis and diabetes	Irving-Rodgers et al.(2008)125
TSP-1	TSP-1−/− mice, islet transplant-ation from WT/TSP-1−/− mice	Decreased insulin secretion and glucose tolerance, TSP-1 derived from ECs is important for insulin secretion	Olerud et al.(2011)126
HGF	CCl4-induced mouse liver injury model, injected with VEGF agonist (Flt/KDR selective)	VEGF/Flt1 induces HGF and reduces liver damage	LeCouter et al.(2003)127
PDGF-CC	Vegfr2iEC−/− and Pdgf-c−/−mice, Adenoviral PDGF-C	Impaired WAT-beige transition, adenoviral PDGF-C improves glucose tolerance in HFD-obese mice.	Seki et al.(2016)128
EL	EL−/− or EL transgenic mice	HDL cholesterol level is negatively associated with EL expression level.	Ishida et al. (2003)129, Ma et al.(2003)130
	EL−/− mice	Decreased FAO and ATP accumulation in heart, more severe systolic dysfunction and LV dilation	Nakajima et al. (2013)131
	EL gene variants	Strong association with HDL level	Ma et al.(2003)130, Edmondson et al. (2009)132, Singaraja et al. (2013)133

iEC, inducible EC; epi, epithelial; NOD, non-obese diabetic; FAO, fatty acid oxidation

2.1. Nitric oxide

NO is an essential determinant of cardiovascular homeostasis through its role in modulating vasomotor tone, maintaining integrity of the vasculature and protection from cellular injury. NO is predominantly synthesized by endothelial NO synthase (eNOS) in ECs. eNOS can be activated by fluid shear stress and numerous other vasoactive agonists through protein phosphorylation, protein-protein interactions, Ca²⁺ signaling and subcellular translocation of proteins. Interestingly, circulating lipoproteins such as HDL can also activate eNOS via a SR-BI-mediated caveolae signaling complex, whereas oxidized LDL inhibits eNOS by disrupting caveolae structure^{134, 135}. In liver sinusoidal ECs, insulin significantly increases eNOS activity and NO production¹¹⁴. ECs with triple knockout of the three FoxO proteins have increased insulin-stimulated NO production, which in turn promotes tyrosine nitration of hepatic IR^{114, 136}. Mice lacking FoxO proteins develop hepatic insulin resistance in response to high-fat diet, which can be inhibited by eNOS inhibitor L-NAME¹¹⁴. However, insulin sensitivity in skeletal muscle and adipose tissue is not affected in EC-specific FoxO knockout mice. Endothelial Shc homology 2–containing inositol phosphatase 2 (SHIP2) is another regulator of NO generation in response to insulin and shear stress^{115, 137}. Interestingly, EC-specific SHIP2 knockout mice have impaired NO bioavailability and vasodilation, and develop insulin resistance in skeletal muscle and adipose tissue¹¹⁵. In addition, NO also increases food intake by regulating leptin and serotonergic signaling pathways in the brain^{138, 139}. In the pancreas, NO affects both exocrine and endocrine functions¹⁴⁰, although the physiologic role of NO in insulin secretion in beta islets remains incompletely understood (summarized by Broniowska et al¹⁴¹). eNOS has been suggested to be involved in obesity and metabolism through multiple processes, such as promoting fat oxidation, hepatic insulin sensitizing substance and adipocyte browning, decreasing gluconeogenesis etc. (Sansbury BE et al¹⁴⁰). Genetic studies indicate a strong association of eNOS mutations with insulin resistance, diabetes, hypertriglyceridemia and HDL decrease^116–118. However, studies with eNOS, eNOS/nNOS or eNOS/nNOS/iNOS triple knockout mice suggest a protective role of NO in obesity, insulin resistance, hyperlipidemia, hypertension and other metabolic diseases^119–123. NO exerts multifunctional roles through cGMP-dependent protein kinase-mediated protein phosphorylation, S-nitrosylation of cysteine residues or nitration of tyrosine residues. Therefore, the identification of specific mediators for NO in hepatocytes, beta islets or neuronal cells may provide safer therapeutic tools for inhibiting hepatic insulin resistance, diabetes and other metabolic disorders.

2.2. Insulinotropic factors

Pancreatic islets in the adult are one of the most vascularized organs, containing fenestrated capillaries that constitute 8~10% of the islet volume and are organized into a glomerular-like network. While the high vascular density and endothelial function are maintained by VEGF secreted from beta cells^142–144, recent studies indicate that EC-secreted factors and extracellular matrix proteins also play an important role in regulating insulin secretion from beta cells.

Extracellular matrix (ECM) proteins are the non-cellular component within all tissues and organs. Islet ECM contains classical basement membrane components such as collagens, laminins, heparin sulfate proteoglycans and hyaluronan^{125, 145}. Intra-islet ECs synthesize ECM components to preserve the function and integrity of blood vessels. Interestingly, beta cells also depend on the ECM components from ECs¹²⁴. These ECM proteins not only provide structural support for beta cells, they also promote beta cell growth, differentiation, survival, migration and potentiate insulin secretion. For example, collagen IV, fibronectin and heparin sulfate are crucial for beta cell migration, growth, survival and maturation^{124, 146}. Collagen IV can also potentiate insulin secretion by interacting with α₁β₁ integrins on beta cells. Laminins and fibronectin was reported to promote insulin gene expression and beta cell proliferation^{124, 146}.

Thrombospondin-1 (TSP-1) is an endogenous inhibitor of angiogenesis and tumor growth¹⁴⁷. In pancreatic islets, TSP-1 is primarily expressed in islet ECs¹²⁶. TSP-1-deficient mice demonstrate glucose intolerance and a defect in glucose-stimulated insulin secretion¹²⁶. The islet transplantation experiments indicate that TSP-1 derived from ECs is important for beta cell function¹²⁶. Although the underlying mechanisms are not well understood, recent studies suggest that TSP-1 promotes beta cell function by activating the TGFβ−1 pathway and protecting beta cells from lipotoxicity by activating PERK and NRF2^{148, 149}. Endothelin-1 (ET-1), a potent vasoconstrictor, is expressed in both ECs and islet endocrine cells¹⁵⁰. In vitro studies with inhibitors indicate that ET-1 can stimulate insulin secretion directly through its receptor ET(A) and the PKC signaling pathway^{151, 152}.

2.3. Growth factors

A number of growth factors are secreted from ECs within metabolic organ/tissues. Among them include hepatocyte growth factor (HGF), which can be secreted from islet ECs¹⁵³. HGF can stimulate beta cell proliferation and insulin production¹⁵⁴. Beta cell-specific depletion of HGF receptor c-met results in impaired insulin secretion and glucose homeostasis^{155, 156}. The secretion of HGF from islet ECs can be stimulated by VEGF-A and insulin. Islet mass is rather stable during adult life. However, it increases in response to obesity, insulin resistance or pregnancy. Interestingly, vascular growth is closely associated with beta cell proliferation during pregnancy, which is likely due to the induction of insulin and VEGF in beta cells in response to placental lactogens and prolactin¹⁵³. Insulin and VEGF can stimulate HGF production in islet ECs through paracrine effects, which further enhances beta cell proliferation. The crosstalk between islet beta cells and ECs forms a feed forward signaling loop to coordinate the growth of beta cells and vasculature. Similarly, this crosstalk also exist for hepatocytes and liver sinusoidal ECs. In the liver sinusoidal ECs, HGF can be also induced by VEGF-A that is released from hepatocytes¹⁵⁷. The secreted HGF further promotes hepatocyte proliferation¹²⁷.

Connective tissue growth factor (CTGF/CCN2) is another growth factor secreted from islet ECs¹⁵⁸. CTGF null embryos display defects in islet morphogenesis, beta cell proliferation and decreased insulin (+) cells¹⁵⁹. Mice with EC-specific depletion of CTGF display impaired beta cell proliferation, while overexpressed CTGF enhances beta cell proliferation^160–162. In addition, CTGF stimulates beta cell maturation, regeneration and modulates responses upon streptozotocin-induced diabetes. In adipose tissue, VEGF stimulates the production PDGF-CC via VEGFR2 activation in ECs¹²⁸. Mice with depleted endothelial VEGFR2, knockout Pdgf-c gene or pharmacological blockade of PDGFR-β demonstrate impaired WAT-beige transition upon cold or adrenergic activation. These observations raise the provocative possibility that PDGF-CC secreted from ECs promotes WAT browning. This model proposes that the local crosstalk between metabolic cells and ECs, a critical process for metabolic homeostasis, is mediated by paracrine factors such as growth factors. As more EC-derived growth factors are identified to participate in the crosstalk of ECs and their neighboring metabolic cells, they provide new targets for the treatment of diabetes and other metabolic disorders.

2.4. The special case of endothelial lipase

Endothelial lipase (EL) shares substantial homology with LPL and HL and is the only known lipase expressed predominantly by ECs. EL is also secreted by smooth muscle cells, cardiomyocytes and macrophages but far less abundantly. In 1999, two laboratories independently cloned the EL gene from cultured ECs^{163, 164}. In vitro studies demonstrate that EL can hydrolyze phospholipids in chylomicrons, very low density lipoprotein, intermediate density lipoprotein, LDL and HDL. However, opposite results were reported regarding the changes of LDL-C concentrations in EL knockout mice^{129, 130}. The majority of studies in cultured cells and animals support that the lipase activity of EL is more specific to HDL than other lipoproteins¹⁶⁵. However, genetic studies show that few SNPs of EL gene are related to changes in plasma HDL-C concentrations. Conflicting data have also been obtained from studies with human patients regarding the association of EL and HDL concentrations^{132, 133, 166, 167}. In ECs, EL can be induced by inflammatory cytokines (i.e. IL-1β and TNFα) and biomechanical forces^{168, 169}. Human plasma EL concentrations are strongly associated with inflammatory markers CRP and interleukin-6 (IL-6)¹⁷⁰. Mice with EL depletion display severe heart failure due to decreased cardiac uptake of FAs¹³¹, suggesting the significant role of EL in FA metabolism. However, the specific roles of endothelial or cardiac EL remain to be further characterized.

Emerging data suggest that ECs communicate with neighboring cells within the microenvironment of metabolic organs through secreting regulatory factors (Fig. 2). Dysregulation of secretory pathways of ECs may play vital roles in the progression of metabolic disorders. Better understanding of these EC-specific secretory pathways may provide easy-to-access therapeutic tools for treating diabetes, pancreatic disorders and cardiomyopathy.

Figure 2.

Metabolic regulation by endothelial cell-secreted factors.

3. Endothelial dedifferentiation in response to metabolic dyshomeostasis

Endothelial differentiation from pluripotent progenitor cells is a crucial process for embryonic development¹⁷¹. Liver sinusoidal ECs experience a differentiation process during the gestational stage that is tightly regulated by EC-derived VEGF and hepatoblast-derived humoral factors¹⁷². Following differentiation, liver sinusoidal endothelium becomes highly permeable due to the formation of fenestrae between liver sinusoidal ECs (LSECs) and the disappearance of cellular basement membrane. In pathologic and in vitro conditions, matured LSECs are dedifferentiated, demonstrating the loss of fenestrae and synthesis of basement membrane¹⁷³. This process is called capillarization and has been recognized as an early event in chronic liver injury in animal models and patients^{174, 175}. In cirrhotic liver, LSEC capillarization is induced, which in turn activates hepatic stellate cells and macrophages, leading to liver fibrosis¹⁷⁶. Differentiated LSECs are considered gatekeepers of fibrosis and progression from a simple steatosis to the early stage of nonalcoholic steatohepatitis. In rat models, the restoration of LSEC differentiation results in regression of liver fibrosis, suggesting that differentiation of LSECs may play a role in fibrosis regression¹⁷⁷. However, more human studies will be needed to support this hypothesis. Pseudocapillarization is similar to capillarization in morphology, but refers to phenotype changes of LSECs in aged animals and humans^{178, 179}. Fenestrae structures of LSECs allow passage of lipoproteins and chylomicron remnants. Reduced fenestrae numbers during pseudocapillarization impairs lipoprotein metabolism, contributing to dyslipidemia and atherosclerosis associated with old age^{180, 181}.

4. Endothelial dysfunction in response to metabolic events

Metabolic organs such as liver, islet, adipose tissue, cardiac and skeletal muscle contain extensive capillary networks. Endothelial dysfunction is observed during the development of metabolic syndrome, diabetes and insulin resistance¹⁸². NO is a major player in endothelial function. NO regulates organ blood flow and systemic blood pressure by acting as a potent vasodilator. NO bioavailability is diminished in obese and diabetic animals and in human patients^{183, 184}. The expression and activity of eNOS are decreased and uncoupling of eNOS is induced in metabolic diseases (reviewed by Sansbury BE et al¹⁴⁰), which may contribute to the disruption of metabolic homeostasis. Studies also suggest that endothelial dysfunction may contribute to the dysregulation of metabolic homeostasis by modulating vessel tone, NO bioavailability and inflammation.

In skeletal muscle, insulin stimulates vasodilatory effects through IR-dependent eNOS activation in ECs¹⁸⁵. This vasodilatory effect of insulin is inhibited by inflammatory factors, FFAs, oxidative stress and adipokines^186–190. Insulin, as well as other hormones such as GLP-1 and adiponectin, carbohydrate-containing meals, and exercise can all increase microvascular perfusion in skeletal muscle even in the absence of total blood flow increase, which precedes and correlates with the rise in muscle glucose uptake (reviewed by Sansbury BE et al¹⁴⁰). The increase in microvascular blood volume of skeletal muscle and heart is blunted in obesity and insulin resistance. In the heart, the inhibition of endothelium-derived NO alters mitochondrial oxygen consumption and substrate utilization by the heart^{191, 192}.

In islets, blood flow is impaired by the depletion of IRS-2 in ECs¹⁹³. However, treatment with the angiotensin-converting enzyme inhibitor enalapril maleate, which increases islet blood flow, can restore insulin secretion by islet cells. In addition, the treatment of islets with conditioned media from dysfunctional islet ECs decreases glucose-stimulated insulin secretion and total insulin content¹⁹⁴. These data suggest that ECs are critical for glucose uptake in muscle and insulin secretion from islets by regulating blood flow. Vasoactive drugs have been shown to enhance islet and muscle blood flow and promote glucose homeostasis¹⁹⁵. Further translational studies with these vasoactive drugs could identify therapeutic applications in islet disorders.

In adipose tissue, endothelial dysfunction is found in obese mice and human subjects exhibiting inflammatory responses¹⁹⁶. Sun et al recently show that miRNA-181b, a microRNA that inhibits endothelial inflammatory responses, is repressed in diabetic ECs¹⁹⁷. Interestingly, the systemic delivery of miR-181b can reduce inflammation and EC dysfunction in white adipose tissue and improve glucose tolerance and insulin sensitivity in diet-induced obese mice. These observations suggest that reagents that inhibit endothelial dysfunction might be effective for treating obesity-induced insulin resistance.

LSECs are the main source of NO in the normal liver. Fluid shear stress-dependent eNOS activation promotes NO production¹⁹⁸. Flow-responsive transcription factor Kruppel-like factor 2 (KLF2) is responsible for the induction of NO and downregulation of vasoconstrictive protein endothelin-1¹⁹⁹. Endothelial dysfunction occurs early in chronic liver disease, precedes fibrosis and inflammation and persists in severe cirrhosis^{200, 201}. In cirrhotic liver, capillarization of LSECs is associated with endothelial dysfunction, leading to increased intrahepatic vascular resistance. Statins improve LSEC function, leading to improvements in liver fibrosis and portal pressure decrease²⁰². The roles of endothelial dysfunction in chronic liver diseases have been summarized in many reviews^{173, 203}.

5. Crosstalk between metabolic events and endothelial-dependent vascular growth

Endothelial-dependent processes of vascular growth, including vasculogenesis and angiogenesis, play a key role in cardiovascular development during embryogenesis. Angiogenesis, a remodeling process of an established capillary network, is also required for pathological conditions such as wound healing. Interestingly, emerging data suggest that endothelial growth is also involved in regulating metabolic homeostasis. For example, a positive correlation exists between skeletal muscle capillarity and insulin sensitivity^{204, 205}. Skeletal microvasculature recruitment is recognized as an important determinant of muscle glucose uptake and insulin sensitivity²⁰⁶. Endothelial-specific depletion of FoxO1/3/4 results in increased skeletal muscle microvascular density²⁰⁷. The expansion of the capillary network leads to increased insulin transport across the capillary endothelium, which may partially explains why EC-specific FoxO depletion protects mice from HFD-induced insulin resistance. In the endocrine pancreas, angiogenesis plays an important role in maintaining islet structure and function through regulation of local blood flow. The survival and optimal engraftment of transplanted islets in type 1 diabetes is dependent on the vascular density and blood flow of islet microvessels²⁰⁸. Pro-angiogenic factors (i.e. VEGF, GLP-1) play crucial roles in angiogenesis after islet transplantation^{209, 210}.

In adipose tissue, angiogenesis is well recognized during adipose tissue expansion (reviewed by Cao Y^{211, 212}). The crosstalk of adipocytes and ECs is a key event during the coordination of vessel growth and fat expansion. A group of angiogenic factors are secreted from adipocytes to stimulate new vessel formation, such as VEGF, FGF-2, HGF, TGF-β, leptin, resistin and visfatin^{211, 212}. In addition, angiogenesis Inhibitors are shown to block fat mass expansion in mice^213–215.

Hepatic angiogenesis is closely linked with liver fibrosis²¹⁶. A vicious circle between fibrosis and pathological angiogenesis is likely to occur, where parenchymal hypoxia activates HIF-dependent liver revascularization and wound healing processes in the liver. However, pathologic angiogenesis provides immature and highly permeable vessels that fail to rescue liver hypoxia, which further promotes the development of liver fibrosis²¹⁷. Consonant with this model, liver-specific knockout of HIF1α reduces liver fibrosis²¹⁸. In addition, anti-angiogenic therapy is highly effective in reducing liver fibrogenic progression (summarized by Bocca C. et al²¹⁶).

Taken together, these studies suggest that targeting angiogenesis can be a potential therapeutic intervention for obesity through blocking adipose tissue expansion or for chronic liver disease through reducing liver fibrosis. However, concerns exist since angiogenic blockers have detrimental effects on skeletal muscle insulin and glucose delivery, beta cell function and diabetes-related wound healing process. Dissecting specific angiogenic pathways for individual metabolic processes is required for safer drug targeting strategies.

Summary and perspectives

Our understanding of ECs, which were initially viewed as a thin layer of inactive cells lining in the inner surface of the circulatory system, has progressed significantly. It has been well recognized that ECs play active roles in a variety of pathological and physiological processes. However, most research has focused on the regulatory roles of endothelial dysfunction in atherosclerosis, coronary artery disease, stroke, hypertension, angiogenesis-related diseases and diabetes-associated vascular complications. Based on the emerging data we have discussed in this review, we can begin to appreciate that the crosstalk between ECs and metabolic tissues/cells is bidirectional. The vascular endothelium possesses underappreciated roles in maintaining systemic metabolic homeostasis through controlling the trans-endothelial transport and secretion of metabolic regulators, or adjusting its own functional status such as differentiation, dysfunction and growth, which indirectly affect the function of metabolic cells/tissues. Dysregulation of endothelial processes results in metabolic disturbance, which in turn exacerbates endothelial and vascular dysfunction and likely acts as a feedback loop for the development of metabolic disorders. Better understanding of these dynamically regulated processes may open up new avenues of the therapeutic approaches for metabolic disorders based on manipulation of these endothelial specific signaling events.

This review focuses on recent progress regarding the endothelial regulation of metabolic processes at least partially through local interplay of microvascular ECs and metabolic cells within their residing organs. Accumulated in vivo evidence gathered from EC-specific knockout mouse models reveals a causal relationship of malfunctional endothelial actions in the disruption of systemic metabolic homeostasis. We speculate that microvascular ECs are the main endothelial cells responsible for crosstalk with metabolic cells/tissues, which is essential for both vascular and metabolic homeostasis. However, our knowledge of these functional roles of microvascular endothelium remains limited. Technical hurdles to the study of microvascular endothelial cells in vivo still exist, such as the difficulty to perform quantitative analysis and direct in vivo imaging on microvasular endothelium. In addition, the specific roles of microvascular vs macrovascular dysregulation in the development of metabolic disorders needs to be further characterized.

During the study of the crosstalk between microvascular ECs and metabolic cells/tissues, one important issue we need be aware of is the phenotypic heterogeneity of ECs. Difference exists between arteries, veins and capillaries, as well as within a variety of vessel beds of specific organs^{219, 220}. In addition, endothelial heterogeneity might even exist between distinct regions of the same organ. For example, ~84% of vascular receptors of the islet ECs are unique to the islet and are not detected in ECs in the surrounding exocrine pancreatic tissue²²¹. It makes islet ECs unique in response to glucose-dependent regulation of nitric oxide and other angiogenic or angiostatic factors such as VEGF, PDGF, endostatin etc (reviewed by Olsson R et al²⁰⁸). It raises needs to identify endothelial factors specific for microvascular ECs (arteriole ECs, capillary ECs or venule ECs) in different metabolic organs such as skeletal muscle, adipose tissue, pancreas and liver. In addition, cell culture studies needs to be performed with microvascular ECs isolated specifically from the designated vessels beds. Given that the in vivo microenvironment is an important determinant for metabolic phenotype, cell culture studies can be only used as supportive evidence and needs to be confirmed with in vivo studies. These studies will help us better understand the unique roles of microvascular ECs within individual metabolic organs. More importantly, it will provide new insights into designing safer therapeutic strategies and specific drug delivery. In addition, macrovascular function might also impact metabolic processes through regulating vascular tone, endocrine pathways or other unknown mechanisms. It suggests the possible involvement of the interaction of ECs with other cells such as smooth muscle cells, which might become another interesting research area for understanding the etiology of metabolic disorders.

It is generally recognized that ECs rely mainly on glycolysis as an energy source during their quiescent and active status such as sprouting angiogenesis^{222, 223}. Metabolic stress such as hyperglycemia can inhibit endothelial glycolytic flux, and glycolytic intermediates are diverted into polyol pathway, hexosamine biosynthesis pathway and methyglyoxal production²²⁴. It remains unclear how endothelial metabolism is coordinated with whole body metabolic homeostasis. Further studies can be directed to test whether specific connections exist between these two processes. If specific dysregulation of endothelial metabolic signaling pathways contributes to certain aspects of disrupted whole body glucose or lipid homeostasis, targeting endothelial metabolic process could become another potential therapeutic targeting strategy for treating metabolic disorders.

Non-Standard Abbreviations and Acronyms

3-HIB

3-hydroxy-isobutyrate

ABCA1

adenosine triphosphate binding cassette transporter A1

ABCG1

adenosine triphosphate binding cassette transporter G1

APLNR

apelin receptor

Apo

apolipoprotein

ATP

adenosine triphosphate

BCAA

branched-chain amino acid

BODIPY

boron dipyrromethene

CRP

C-reactive protein

CTGF

connective tissue growth factor

EC

endothelial cell

ECM

extracellular matrix

EL

endothelial lipase

eNOS

endothelial nitric oxide synthase

ET-1

endothelin-1

FA

fatty acid

FABP

fatty acid binding protein

FATP

fatty acid transport protein

FcγRIIB

Fcγ receptor IIB

FFA

free fatty acid

FoxO

Forkhead box O

GLP-1

glucagon-like peptide 1

GLUT

glucose transporter

GPIHBP1

glycosylphosphatidylinositol anchored high density lipoprotein binding protein 1

HDL

high density lipoprotein

HFD

high-fat diet

HGF

hepatocyte growth factor

HIF1

hypoxia-inducible factor 1

IGF1R

insulin-like growth factor 1 receptor

IL-6

interleukin-6

IR

insulin receptor

IRS

insulin receptor substrate

LCFA

long-chain fatty acid

LDL

low density lipoprotein

LDLR

low density lipoprotein receptor

HL

hepatic lipase

LPL

lipoprotein lipase

LRP1

low density lipoprotein receptor-related protein 1

LSEC

liver sinusoidal endothelial cell

NO

nitric oxide

NRF2

nuclear factor erythroid 2–related factor 2

PDGF

platelet-derived growth factor

PERK

protein kinase RNA-like endoplasmic reticulum kinase

PI3K

phosphoinositide 3 kinase

PPAR

peroxisome proliferator-activated receptor

PTP1B

protein tyrosine phosphatase 1B

SHIP2

Shc homology 2–containing inositol phosphatase 2

SR-BI

scavenger receptor class B type I

T2DM

type 2 diabetes mellitus

TNF

tumor necrosis factor

TSP-1

thrombospondin-1

TZD

thiazolidinedione

VEGF

vascular endothelial growth factor

VLDL

very low density lipoprotein

WAT

white adipose tissue