The endothelium, anatomically defined as the inner cellular lining of blood and lymphatic vessels, acts as a physiological barrier, which prevents direct contact between tissues and the circulation (9). It also plays a role in mediating angiogenesis, inflammation, thrombosis, vascular tone and energy homeostasis (8–12). Endothelial dysfunction is often defined as the reduced production or availability of nitric oxide (NO), the elevated release of vasoconstrictors such as endothelin-1 (13) and the upregulation of cell adhesion molecules, associated with enhanced leukocyte adherence, increased cell permeability, and platelet activation (14). Endothelial dysfunction is a well-established response to cardiovascular risk factors and precedes the development of atherosclerosis (15).

The endothelium is a dynamic organ, and is one of the first responders to metabolic and vascular changes (16), such as insulin level (17) and blood flow (18). Due to its intimate relationship with nearby cells, impaired endothelial barrier function, or imbalanced endothelial secretion of metabolites, results in dysregulated signaling pathways of peripheral tissues, which in turn contributes to the development of metabolic dysfunction and chronic tissue inflammation. Time-course experiments on diet-induced obese mice revealed the preceding role of endothelial dysfunction in the development of metabolic dysfunction and chronic inflammation in liver, muscle and AT (19). In obesity, dysfunctional endothelium displays a more pro-inflammatory and pro-fibrotic phenotype, resulting in increased vascular leakage and acceleration of tissue inflammation by recruitment and activation of immune cells (20).

In addition to acting as a physical barrier between local tissues and the circulation, the endothelium plays an important role in signaling. Since the discovery of endothelial-derived relaxing factor (EDRF), now identified as NO, the endothelial secretory profile has gained much attention (21, 22). There is a growing body of evidence of the presence of endothelial-derived small membrane vesicles, which are released to the extracellular space (23). These findings further establish the existence of an endothelial secretome and raise the question of its role in various diseases including obesity in CMS.

Obesity is a multisystem disorder driven by nutrient excess, leading to perturbation of glucose homeostasis, immune dysregulation, dyslipidaemia and disadvantageous alterations in adipocyte phenotype, which leads to a vicious cycle of cellular dysfunction contributing to CMS (15, 24). During nutrient excess, white AT (WAT) undergoes expansive remodeling, whereby white adipocytes adopt a hypertrophic/hyperplastic phenotype (25), accompanied by neovascularisation to support this remodeling (26, 27). In advanced obesity there is a mismatch between WAT expansion and vascularisation leading to inadequate perfusion (28), adipocyte ischaemia (28), inflammation (28–30), secretion of pro-inflammatory mediators (31) and the requirement to store lipids in tissues ill-equipped to deal with this challenge, all contributing to CMS (32).

AT is crucial in storing excess energy and maintaining thermal homeostasis (33). It is dynamically regulated by a wide range of physiological cues, including norepinephrine, insulin and acetylcholine (33). There are two main types of AT; WAT which is specialized for the storage of energy in the form of triglyceride (34) and thermogenic brown AT (BAT), which unlike WAT expresses uncoupling protein-1 (UCP-1) (35). UCP-1 uncouples cellular respiration from mitochondrial ATP synthesis, affording thermogenic AT the capacity to burn fat and generate heat (36). Recent studies indicate that at least two distinct types of thermogenic adipocyte exist in mammals: a pre-existing form established during development, termed classical BAT, and an inducible form, “beige” adipocytes. Beige adipocyte biogenesis can be stimulated by various environmental cues, such as chronic cold exposure, in a process frequently referred to as “beiging” of white fat (37, 38). The thermogenic capacity of beige adipocytes and the possibility of controlling their expression has stimulated substantial attention due to its potential application in the mitigation of obesity-related disease (39–42). Although extensive efforts have been made pharmacologically to activate AT thermogenesis, these attempts have been unsuccessful due to poor bioavailability and concerns over the side effects of the agents used (43). More importantly, a recent study reported that human brown adipocyte thermogenesis is actually mediated by β₂-adrenoceptors, instead of β₃-adrenoceptors as in rodents (44), explaining the lack of efficacy of β₃-adrenoceptor agonists to activate brown adipose thermogenesis or white adipose beiging. Thus, identifying alternative pathways to promote beige adipose biogenesis could lead to therapeutic interventions which reduce cardiovascular risk.

Alternative therapeutic goals also include reducing AT inflammation and fibrosis (45). Under normal physiology, inflammation is important in maintaining healthy adipogenesis and adipocyte function and helps regulate whole-body glucose tolerance (46). However, chronic adipocyte inflammation can link to metabolic defects such as insulin resistance (46–48). Current therapeutics targeting AT inflammation do not always present positive outcomes in human trials and potential risk of increased susceptibility to infections is yet to be explored (45). On the other hand, we still await clinical studies employing anti-fibrotic agents to determine the efficacy and safety of targeting WAT fibrosis to treat CMS.

AT is highly vascularized with a capillary network, allowing the reciprocal transport of nutrients, substrates and oxygen. During AT expansion, the capillary network also expands to meet the increasing demand for nutrients and oxygen (49). With particular interest in dissecting the communication between endothelial cells and adipocytes, here we review the role of endothelial-derived molecules in mediating metabolism and inflammation in the pathophysiology of CMS. We focus on the synthesis, release and the paracrine action of the endothelial secretome in AT and discuss how it may be exploited therapeutically in the future. Endothelial-derived molecules are grouped according to their physical properties.

Endothelial-derived NO has been traditionally viewed as a key modulator of cardiovascular function, and reduced bioavailability of NO has been associated with inflammation and vascular disease (10, 50). More recently, data suggests that NO also acts as a metabolic modulator, and reduced circulating NO levels have been observed before the development of peripheral insulin resistance in diet-induced obesity (12, 19). Under normal physiological conditions, endothelial nitric oxide synthase (eNOS, also known as NOS3) is active and generates NO transiently. The activity of eNOS is regulated by shear stress within vessels. The synthesis of NO can also be elevated upon stimulation by endogenous signaling molecules such as insulin, insulin-like growth factor 1 (IGF-1), acetylcholine and the pro-inflammatory molecule bradykinin (51–54). Within AT, eNOS-mediated NO generation can be upregulated by adipocyte-secreted adiponectin (55) and leptin (56). Apart from its well-known vasodilatory effect via activating the soluble guanylate cyclase (sGC)-generated cyclic guanosine monophosphate (cGMP) pathway in vascular smooth muscle, NO is thought to be important in regulating oxygen transport and mitochondrial respiration in other tissues (57). In contrast, in diseases such as CMS, eNOS activity is suppressed whereas inducible NOS (iNOS) in the endothelium is activated to generate potentially pathophysiological concentrations of NO, which reacts with superoxide to form peroxynitrite (58). Peroxynitrite is cytotoxic and is a known pro-atherosclerotic factor (58, 59) (reviewed in Peroxynitrite (ONOO^–)).

A range of factors, including elevated plasma levels of free fatty acids, pro-inflammatory cytokines such as tumor necrosis factor-alpha (TNFα), oxidized low-density lipoprotein (OxLDL) and asymmetric dimethylarginine, all common features of CMS, are known to inhibit eNOS activity (60–63). Under prolonged high fat diet, mice display profoundly reduced eNOS expression, specifically in AT suggesting diseased adiposity is coupled with a decline in eNOS activity (64). The WAT of diet-induced obese mice demonstrates a reduction in NO coupled with increased AT inflammation, characterized by increased gene expression of the pro-inflammatory TNFα (65). eNOS-deficient mice are insulin-resistant and hyperlipidaemic (66, 67), even on a chow diet, AT from eNOS-deficient mice displays a notably inflammatory profile, characterized by increased gene expression of TNFα, interleukin-6 (IL-6) and monocyte chemoattractant protein-1 (MCP-1) (65). Together these findings suggested endothelial-derived NO is insulin-sensitizing and anti-inflammatory in AT under normal physiological conditions. Promisingly, in eNOS-deficient mice, the addition of dietary nitrate at a dose just enough to restore physiological NO levels, is able to reverse the unfavorable metabolic sequelae, reducing both visceral fat deposition and circulating triglycerides (68). Restoring NO downstream signaling by administration of sildenafil also reduces AT inflammation in high fat diet-challenged mice (65). Complementary studies using mice overexpressing eNOS, demonstrate that eNOS over-expression offers protection against diet-induced obesity, with a more metabolically active AT, characterized by smaller adipocyte size and elevated mitochondrial biogenesis and metabolism (64). Taken together these murine studies show that eNOS activity is important in regulating AT function and inflammation.

Various therapeutic approaches have been employed to determine whether targeting NO can improve AT metabolism in rodents and humans. S-nitrosoacetyl penicillamine (SNAP) an NO donor has been shown to enhance mitochondrial biogenesis in primary murine brown adipocyte precursors and cultured 3T3-L1 white adipocytes, in a process mediated by cGMP (69). Nitrate, which can be utilized to produce NO under hypoxia, has been shown to have beneficial effects in AT in several studies. Nitrate can induce gene expression of beiging markers in cultured primary murine white adipocytes (70, 71), as well as inducing other favorable metabolic changes, including upregulation of mitochondrial respiration, increased uptake and oxidation of fatty acid and glucose (71, 72). Application of nitrite on palmitate-treated primary mouse white adipocytes also induces beiging (73). Although eNOS is not the only source of NO in mammals due to the presence of other isoforms of NOS and dietary nitrate intake, these findings support the hypothesis that eNOS-derived NO promotes adipocyte beiging and mitochondrial biogenesis.

eNOS-derived NO has also been well-characterized as an anti-inflammatory mediator. It inhibits platelet aggregation (74) and acts on the endothelium to suppress expression of adhesion molecules, thus suppressing leukocyte recruitment and accumulation (75). Impaired eNOS-NO signaling may contribute to AT inflammation in obesity, as characterized by increased leukocyte adhesion and platelet aggregation (76).

In rodent models, dietary nitrate resulted in beiging of white AT in rats (71) and more interestingly, in mice challenged with high fat diet, dietary nitrate supplementation reduced weight gain, improved glucose tolerance and attenuated AT inflammation (73, 77). Chronic dietary supplementation of L-arginine (the precursor of NO) reduced weight gain, circulating glucose and triglyceride levels in Zucker diabetic fatty rats (78). Human studies have also confirmed the benefits of targeting NO in CMS. In a randomized, double-blind, placebo-controlled clinical trial, 6 weeks of dietary L-arginine supplementation reduced central adiposity in women with obesity (79). In another clinical trial, 6 months of dietary L-arginine supplementation improved whole body insulin-sensitivity in individuals with obesity (80). However, conflicting results showing lack of efficacy of L-arginine treatment have also been reported in studies of patients with obesity and insulin resistance, raising the doubt of therapeutic potential of L-arginine (81, 82). A recent systematic review and meta-analysis of clinical trials suggested L-arginine supplementation resulted in reduced waist circumference but had no effect in reducing body weight or BMI (83). Future studies determining the appropriate dosage and length of the treatment are needed to improve treatment outcome using L-arginine.

Endothelin was the first potent endothelium-derived vasoconstrictor to be identified. ET-1 was isolated from the culture media of porcine aortic endothelial cells as a 21 amino acid peptide (84). Two other endothelins, ET-2 and ET-3, were later identified. The endothelin family is synthesized and released upon cleavage of its precursors by endothelin-converting enzymes (85). The synthesis of ET-1 can be upregulated by endogenous mediators such as thrombin (84), angiotensin II (86), insulin (87), lipoproteins (88, 89), and suppressed by NO (90) and prostacyclin (91). Secreted endothelin acts on endothelin receptors type A and B (ET_AR and ET_BR) and mediate glucose and lipid metabolism in the AT (92).

In in vitro experiments using mature 3T3-L1 adipocytes, ET-1 upregulated glycerol release in a dose-dependent manner (93). This effect was prevented by pre-treatment using an ET_AR antagonist BQ-610, but not an ET_BR antagonist BQ-788, suggesting ET-1-induced lipolysis was mediated by ET_AR (93). In time-course experiments the effect of ET-1 on glucose uptake was shown to be biphasic; in mature 3T3-L1 adipocytes, treatment for up to 6 h stimulated ET_AR-mediated glucose transporter expression and glucose uptake, whereas longer treatments suppressed glucose uptake (94). In cultured 3T3-L1 adipocytes, prolonged ET-1 treatment has also been shown to cause insulin resistance, in an ET_AR-mediated process (95, 96), which is consistent with another study where prolonged ET-1 stimulation enhanced lipolysis in differentiated subcutaneous white adipocytes, again mediated by ET_AR (97). Prolonged ET-1 stimulation also results in ET_BR-mediated suppression of insulin receptor substrate-1 (IRS-1) expression and insulin-stimulated anti-lipolytic responses in differentiated visceral adipocytes (98). Taken together, these studies demonstrate the detrimental effect that long-term exposure of ET-1 has on AT, which could contribute to the development of CMS in obesity.

From a conventional perspective, ET-1 is vasoconstrictive and pro-inflammatory. There is some evidence suggesting that ET-1 induces immune cell recruitment and activation (99, 100). ET-1 can stimulate monocytes to release IL-8 and MCP-1, which recruit neutrophils and monocytes, respectively (99). ET-1 can also stimulate degranulation of mast cells to release TNFα and VEGF (100). These in vitro findings suggest ET-1 can induce inflammation by activating circulating monocytes or tissue-resident mast cells, which may lead to AT inflammation and fibrosis in obesity (101).

In patients with obesity, ET-1 is upregulated in AT (97, 98). Elevated ET-1 activity has also been reported in patients with type II diabetes (102). Studies support a link between higher basal ET-1 signaling activity and greater adiposity, although whether ET-1 has causal role in excess adiposity is unknown. In support of ET-1 playing a causal role in the pathophysiology of CMS, patients with CMS have an elevated arterial ET-1 level which is reported to be associated with circulating triglyceride level (103). Enhanced ET-1 signaling is also observed in overweight individuals and is preserved in individuals with obesity suggesting heightened ET-1 signaling precedes the development of obesity and its associated complications in patients with CMS (104). In contrast, body mass index has been recently reported to be negatively correlated with ET-1 expression level in omental AT (105). Differences in these findings may be due to varying influence of other co-morbidities such as vascular diseases which were excluded in the former study (104) but not the latter (105), or the complex biphasic actions of ET-1 as described above.

Although whether AT ET-1 signaling plays a causal role in obesity remains inconclusive, preclinical studies suggest inhibiting ET-1 could be another possible therapeutic strategy for CMS. Mice with endothelial-specific overexpression of ET-1 fed with chow diet displayed impaired glucose intolerance and lower systemic energy expenditure (105). However, endothelial-specific ET-1 overexpression did not attenuate Western diet-induced WAT inflammation in mice (105). Pharmacological inhibition of systemic ET-1 signaling, using either the ET_AR-specific inhibitor atrasentan or the non-specific ET receptor inhibitor bosentan, lowered fasting blood glucose and enhanced adiponectin secretion in high fat diet-challenged mice (106). Atrasentan administration also lowered circulating triglycerides in high fat diet-challenged mice (106). The improved metabolic phenotype observed in high fat diet-challenged mice treated with of ET receptor blockers may be in part due to attenuated ET receptor-mediated perturbation in AT function (106). Besides, atrasentan administration also alleviated high fat diet-induced AT inflammation, as evidenced in reduced accumulation of immune cells and gene expression of pro-inflammatory cytokines such as TNFα (106). ET_AR-selective inhibition may be a promising pharmacological intervention to combat perturbed metabolism and elevated inflammation in AT in CMS.

Platelet-derived Growth Factor (PDGF)-CC is a protein mediator formed by two PDGF-C monomers (107). PDGF-CC is expressed in various cell types, which share a mesenchymal origin, including the endothelium (107, 108). It is produced and secreted in its full length and digested by extracellular proteases, such as plasmin, to release its active growth factor fragment (109). Endothelial release of PDGF-CC can be upregulated by VEGF (108). The growth factor fragment acts on the transmembrane PDGF receptor-alpha (PDGFRα) to mediate various cellular processes including proliferation (109).

A study using genetically modified mice and β3-selective adrenergic stimulation, using a β3 adrenoceptor agonist, CL-316243, demonstrated that the endothelium responds to adipocyte-secreted VEGF to secrete PDGF-CC, which then induces beiging in neighboring white adipocytes (108). Although increasing PDGF-CC signaling may be explored to develop adipocyte beiging therapeutics, it may also pose a risk of inducing WAT fibrosis. Between the two identified subpopulations of PDGFRα-positive adipocyte progenitors, high fat diet activates PDGFRα-positive adipocyte progenitors with high CD9 expression, which are more proliferative and pro-fibrotic in WAT than their low CD9-expressing counterparts (110). Notably, higher levels of PDGFRα-positive high CD9-expressing adipocyte progenitors were identified in patients with obesity and glucose intolerance or diabetes (110). As PDGFRα can also be activated by PDGF-AA in WAT (110), these findings suggest caution on specificity of adipocyte progenitor subset activation is required when designing therapeutics manipulating PDGFRα signaling to treat obesity and diabetes.

Heparanase is a heterodimeric endoglycosidase responsible for degrading cell surface or extracellular heparan sulfate proteoglycans (HSPGs) and is secreted by various mammalian cell types including endothelial cells (111). Heparanase is first synthesized as a 60 kDa protein, before undergoing lysosomal modification to give an active 50 kDa protein. Upon activation by cytokines, such as TNFα and IL-1beta (IL-1β) or OxLDL, endothelial cells secrete heparanase into the extracellular matrix (112). In contrast, vascular endothelial growth factor (VEGF) inhibits basal endothelial heparanase secretion (112).

Upon release into the extracellular matrix, heparanase breaks down matrix or membrane-bound HSPGs. HSPGs are heparan sulfate chains-containing glycoproteins, which function as structural proteins and cell surface receptors to mediate cellular signaling (113). HSPGs bind to a wide variety of molecules such as chemokines, growth factors and extracellular matrix proteins and enable a regulated release of such proteins upon their degradation by heparanase (111). Of interest to this review, endothelial-released heparanase mediates AT metabolism and inflammation. In terms of metabolism, endothelial cells activated by lysophosphatidylcholine, a product of lipoprotein lipase (LPL)-mediated lipolysis, secrete heparanase to modulate adipocyte secretion of LPL (114). LPL is taken up by endothelial cells and expressed at the luminal surface to hydrolyse circulating triglycerides, hence, the heparanase-HSPG signaling serves as a positive feedback loop to luminal LPL expression (114). In addition, as cell surface HSPG and HSPG-bound LPL mediate effective fatty acid uptake and storage in adipocytes (115, 116), depleting adipocyte HSPG and LPL may impair adipocyte lipid uptake and stimulate de novo fatty acid synthesis.

Regarding inflammation, endothelial cell-released heparanase is pro-inflammatory and pro-atherogenic. High levels of heparanase expression are found in atherosclerotic plagues in ApoE-knockout mice, suggesting heparanse may be involved in inflammation (112). As HSPGs govern immune cell recruitment and activation, and extracellular matrix composition (113, 117), elevation in heparanse-HSPG signaling in AT may drive AT inflammation and fibrosis in CMS.

Heparanase levels are significantly higher in both the plasma and the urine of humans with diabetes compared with healthy individuals (118). Higher heparanase gene expression was also found in the visceral AT of hyperglycaemic type 1 diabetic rats (119). Although the regulatory role in lipid metabolism and inflammation of heparanse-HSPG signaling has been extensively studied, the focus was mostly its effect in atherosclerosis and cancer, and not in CMS. Future work examining the AT-specific effect of the signaling pathway is needed to support the suggestion that endothelial-secreted heparanse may be involved in the progression of AT dysfunction in CMS.

In obesity, AT switches to a more pro-inflammatory profile, with elevated secretion of pro-inflammatory mediators such as tumor necrosis factor α (TNFα) and interleukin(IL)-6 (120–122). Alongside adipocytes, other cell types such as macrophages and endothelial cells are suggested as the primary source of pro-inflammatory mediators in the AT in obesity (120). Endothelial cells have been shown to secrete a diverse profile of pro-inflammatory mediators in response to adipocyte-derived signals (123). Twenty four-hour treatment using conditioned media from healthy human adipocytes stimulated sustained secretion of pro-inflammatory cytokines including TNFα and IL-6 from human umbilical endothelial cells (HUVECs) (123). Both TNFα and IL-6 have been extensively studied and reviewed as promoters of an inflammatory response, with local recruitment of immune cells, such as macrophages, which contribute to chronic inflammation, and may in turn lead to insulin resistance in AT (122). In addition to their pro-inflammatory actions, these endothelial-derived cytokines can also act directly on adipocytes to mediate various metabolic functions.

Reactive oxygen species (ROS) are highly reactive intermediates of oxygen produced by a range of cellular enzymes, such as NADPH oxidases (NOXs), complexes I and III of the mitochondrial respiratory chain, uncoupled eNOS or xanthine oxidases (XOs) (147). Most of the reactions mediated by these enzymes generate superoxide (O2-), which has a very short half-life and is converted to hydrogen peroxide (H₂O₂) and peroxynitrite (ONOO⁻) after reacting with water or NO, respectively. Under normal physiological conditions, intracellular ROS mediate cellular processes such as proliferation and differentiation (148). However, in diseased states, such as insulin resistance where eNOS is uncoupled and generates a large amount of superoxide, which can become stress signals, resulting in increased tissue or vascular inflammation and downregulated adipocyte function (147, 149, 150).

Prostaglandins are a family of lipid mediators formed from the conversion of arachidonic acids catalyzed by respective synthases (187). In AT, prostaglandin I2 (PGI₂) and E2 (PGE₂) are the two most abundant members of the family present (188). Studies have proposed the importance of the presence of both non-fat cells, predominantly endothelial cells, and adipocytes for optimal AT prostaglandin production and secretion (189, 190). Experiments using isolated endothelial cells from human adipose capillaries presented evidences that human AT endothelial cells are capable to secrete both PGI₂ and PGE₂ (191).

Upon activation by fatty acids, human AT endothelial cells (hATECs) increase expression of fatty acid transporters including FABP-4 to enhance fatty acid uptake from the circulation (221). hATECs then release fatty acids to adipocytes for storage (221). More importantly, endothelial cells release lipids to activate peroxisome proliferator-activated receptor gamma (PPARγ) signaling in adipocytes (221). Treatment using the lipid fraction of conditioned media from isolated hATECs was able to induce PPARγ activation in primary human adipocytes. PPARγ activation was enhanced further when the hATECs were pre-treated with oleic acid, a fatty acid (221). Human adipocytes are unable to produce PPARγ agonists but are responsive to PPARγ stimuli (221), therefore, these findings demonstrated that the AT endothelial cells are able to respond to circulating fatty acids, and secrete lipids, activating PPARγ signaling in adipocytes. As PPARγ is a key regulator of adipogenesis and adipocyte function, including mitochondrial respiration (222), future studies examining the identity of endothelial-derived PPARγ ligand and elucidating the mechanism by which the activated endothelial cells secrete such ligands may give rise to an alternative drug candidate to combat diseased adiposity in obesity.