Endothelial response to glucose: dysfunction, metabolism, and transport

Abstract

The endothelial cell response to glucose plays an important role in both health and disease. Endothelial glucose-induced dysfunction was first studied in diabetic animal models and in cells cultured in hyperglycemia. Four classical dysfunction pathways were identified, which were later shown to result from the common mechanism of mitochondrial superoxide overproduction. More recently, non-coding RNA, extracellular vesicles, and sodium-glucose cotransporter-2 inhibitors were shown to affect glucose-induced endothelial dysfunction. Endothelial cells also metabolize glucose for their own energetic needs. Research over the past decade highlighted how manipulation of endothelial glycolysis can be used to control angiogenesis and microvascular permeability in diseases such as cancer. Finally, endothelial cells transport glucose to the cells of the blood vessel wall and to the parenchymal tissue. Increasing evidence from the blood-brain barrier and peripheral vasculature suggests that endothelial cells regulate glucose transport through glucose transporters that move glucose from the apical to the basolateral side of the cell. Future studies of endothelial glucose response should begin to integrate dysfunction, metabolism and transport into experimental and computational approaches that also consider endothelial heterogeneity, metabolic diversity, and parenchymal tissue interactions.

Introduction

Due to their location between blood and tissue, endothelial cells have a unique relationship with glucose. Endothelial cells are exposed to varying blood glucose concentrations during pre- and post-prandial cycles. Endothelial cells take up glucose from the blood, which they then metabolize or transport to cells of the vascular wall and parenchymal tissue. Research has shown that glucose-induced endothelial dysfunction varies among different vessel types and in different vascular beds; that endothelial glucose metabolism is critical to cell phenotype and function; and that endothelial glucose transport is a regulated process. Indeed, endothelial glucose response is important in micro- and macro-vascular diseases ranging from atherosclerosis to cancer to Alzheimer’s disease [1,2].

In this review, I discuss current knowledge of endothelial dysfunction in altered glucose; how endothelial glucose metabolism contributes to cell function; and how endothelial glucose transport is regulated in health and disease (Figure 1).

Figure 1. Endothelial cells become dysfunctional in high glucose environments, which is connected to glucose metabolism and transport.

In normal glucose, endothelial cells take up glucose from the blood via glucose transporters (GLUT) and sodium-glucose cotransporters (SGLT). Glucose can then be metabolized, primarily via glycolysis. Pyruvate is shuttled into the mitochondria to be metabolized in the tricarboxylic acid (TCA) cycle or converted to lactate and transported out of the cell via monocarboxylate transporters (MCT). Alternatively, glucose can be transported to the vessel wall or parenchymal tissue paracellularly through cell–cell junctions or transcellularly via GLUTs and SGLTs at the cell abluminal surface. In high glucose, endothelial cells take up more glucose and increase glucose metabolism via glycolysis. Some of the excess glycolytic intermediate metabolites are shunted down glycolytic side branch pathways that contribute to endothelial dysfunction, including the polyol, pentose phosphate, hexosamine biosynthetic, and methylglyoxal pathways. Side branch pathway metabolism is further increased by superoxide overproduction from excess mitochondrial metabolism. Excess glucose also increases protein kinase C (PKC) activation, changes microRNA (miRNA) and long non-coding RNA (lncRNA) expression, and increases extracellular vesicle (EV) release while changing EV contents.

Glucose-induced endothelial dysfunction

Healthy endothelial cells maintain vascular homeostasis through control of permeability, inflammation, vascular tone, and injury repair [3]. However, when endothelial cells are exposed to high or low glucose [4], they become dysfunctional. Endothelial dysfunction is characterized by impaired endothelium-dependent vasodilation due to reduced nitric oxide (NO) bioavailability [5,6], dysregulated angiogenesis [7–9], as well as increased inflammatory adhesion molecules, permeability, and low-density lipoprotein oxidation [10–12]. Glucose-induced endothelial dysfunction is linked to enhanced glucose metabolism down glycolytic side branches; more recently, important roles for non-coding RNAs, extracellular vesicles, and glucose transporters have also been identified.

Classical pathways

Elevated extracellular glucose increases endothelial glucose uptake and metabolism through glycolysis and glycolytic side branches. The polyol pathway, which catalyzes glucose to sorbitol via aldose reductase, was the first glycolysis side branch identified as different in diabetes [13]. Polyol pathway reactions consume nicotinamide adenine dinucleotide phosphate (NADPH), potentially reducing the formation of the reactive oxygen species (ROS) scavenger reducer glutathione and thereby increasing oxidative stress. A recent study further suggested that the aldose reductase inhibitor fidarestat restored expression of the protective histone deacetylase sirtuin 1 (SIRT1) and reduced the detrimental effects of mammalian target of rapamycin (mTOR) to decrease ROS [14]. However, aldose reductase inhibitors only showed modest effects on diabetic vascular complications in human clinical trials [15], perhaps due to the weak inhibitory activity of clinical aldose reductase inhibitors [16].

A second classical pathway for glucose-induced endothelial dysfunction is advanced glycation end products (AGEs), which are formed by nonenzymatic glycation of protein amino groups. Excess methylglyoxal in hyperglycemia produces AGE, which damage endothelial cells by modifying intracellular proteins [17]; by modifying extracellular matrix proteins and their integrin interactions [18–20]; and by modifying plasma proteins, which then bind to cellular receptors of advanced glycation end products (RAGEs) [21,22]. Overexpression of glyoxalase, which detoxifies methylglyoxal, prevented vascular complications in diabetic animals [23]. While methylglyoxal increased inflammation and endothelial loss in diabetic mouse hearts, endothelial and cardiac function were both maintained when glyoxalase was overexpressed [24]. Recent studies further suggest that methylglyoxal-induced intracellular protein modifications activated the unfolded protein response and increased heat shock proteins in human aortic endothelial cells (HAEC), resulting in a pro-inflammatory and pro-thrombotic endothelial phenotype [25]. HAEC treated with methylglyoxal showed less proliferation, more cell death, and shorter network formation, which was mediated through ATP-sensitive potassium (K_ATP) channel dysfunction and mitogen-activated protein kinase (MAPK) pathways such as p38, c-Jun N-terminal kinase (JNK), and extracellular-signal-regulated kinase (ERK) [26].

Protein kinase C (PKC) activation is another classical pathway that mediates hyperglycemia-induced endothelial dysfunction [27,28]. Intracellular hyperglycemia increases the lipid second messenger diacyl glycerol (DAG), which activates many PKC family members [29]. PKC acts through varied pathways, including decreasing endothelial nitric oxide synthase (eNOS) and increasing NADPH oxidase to enhance oxidative stress [30]. Treatment of diabetic mice with a PKCβ inhibitor decreased atherosclerotic plaque formation and macrophage infiltration [31]. Diabetic therapies such as metformin and liraglutide prevented high glucose-induced changes in DAG, PKC, NADPH oxidase, and eventually oxidative stress [32]. More recently, high glucose inhibited human umbilical vein endothelial cell (HUVEC) expression of 8-oxoguanine glycosylase-1 (OGG1), a DNA repair protein. OGG1 overexpression reduced PKC activation and subsequent ROS production via NADPH oxidase [33].

A final classical pathway for glucose-induced endothelial dysfunction is protein O-GlcNAcylation, which is important in physiologic cell processes yet also contributes to vascular pathology. Protein O-GlcNAcylation increases in high glucose due to increased UDP-GlcNAc production via the hexosamine biosynthetic pathway (HBP) [34]. eNOS O-GlcNAcylation increased in diabetic animals, which inhibited eNOS phosphorylation and NO production [35]. In endothelial cells in vitro, elevated glucose and HBP metabolic rate increased eNOS O-GlcNAcylation and thereby decreased eNOS and NO bioavailability [36]. Negative O-GlcNAcylation effects on endothelial function were reversed by decreasing HBP metabolic rate via inhibition of glutamine–fructose-6-phosphate amidotransferase (GFAT), the rate-limiting enzyme of the HBP [35]. In 2015, Barnes et al. [37] reported increased O-GlcNAc in human pulmonary artery hypertension, a vascular disease associated with increased glucose metabolism. Endothelial cells extracted from forearm vein of people with diabetes also showed increased protein O-GlcNAcylation as compared with non-diabetic controls, and further high glucose culture of these cells reduced insulin-mediated eNOS phosphorylation [38].

All these different mechanisms may result from a single upstream hyperglycemia-induced process: excess mitochondrial superoxide production. When more glucose-derived pyruvate is oxidized in the tricarboxylic acid (TCA) cycle, more of the electron donors nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FADH₂) enter the mitochondrial electron transport chain. Mitochondrial membrane potential increases to its critical threshold. Electrons then back up to coenzyme Q, which donates electrons to oxygen to create superoxide. Mitochondrial superoxide can then activate other superoxide production pathways. Hyperglycemia increases mitochondrial membrane potential and ROS production in cultured primary arterial endothelial cells [39]. When membrane potential or electron transport chain were inhibited, or when manganese superoxide dismutase (MnSOD) was added, hyperglycemia no longer activated classical pathways or increased ROS [40–42].

Mitochondrial carbonic anhydrase inhibitors may reduce hyperglycemia-induced endothelial dysfunction by decreasing mitochondrial ROS production. Mitochondrial carbonic anhydrases (VA and VB) produce bicarbonate (HCO₃−) inside mitochondria, where it carboxylates pyruvate to oxaloacetate [43]. When these carbonic anhydrases are inhibited, pyruvate cannot be metabolized via the TCA cycle, which forces cells to metabolize glucose via glycolysis without producing superoxide [44]. The mitochondrial carbonic anhydrase inhibitor topiramate reduced respiratory rate and ROS production of mouse cerebral pericytes cultured in high glucose [45]. Topiramate additionally reduced oxidative stress and pericyte loss in streptozotocin-treated mice [44]. Amyloid aggregates in Alzheimer’s disease also induce mitochondrial dysfunction and oxidative stress [46]. Mitochondrial carbonic anhydrase inhibitors methazolamide and acetazolamide reduced mitochondrial membrane polarization, ROS, and apoptosis in immortalized human cerebral microvascular endothelial cells (hCMEC/D3) treated with Aβ peptide [47].

microRNAs and long non-coding RNAs

Diabetes and hyperglycemia affect endothelial microRNA (miRNA) and long non-coding RNA (lncRNA). miRNAs are small (~20 nucleotides) non-coding RNAs that digest specific mRNA using the RNA-induced silencing complex (RISC) [48]. Thirteen miRNAs were different between plasma samples of patients with diabetes vs. patients with normal glucose tolerance. miR-126, which is enriched in endothelial cells and important in both angiogenesis and vascular integrity, decreased in diabetic patient plasma, in hyperglycemic mice plasma, and in extracellular vesicles (EVs) released from HUVEC in high glucose [49]. A similar study found 10 additional miRNAs that were altered in human type II diabetes, a rat type II diabetic model, and HUVEC in high glucose. Of these, miR-26a-5p, miR-140-5p, and miR-29b-3p relate to endothelial apoptosis [50]. miR-29b-3p was later shown to interact with SIRT1 and decrease its expression to induce retinal microvascular endothelial cell apoptosis [51]. miR155, miR-146a, miR-200c, and miR-503 were up-regulated in diabetic nephropathy, diabetic arteries, and high glucose renal and muscle endothelial culture [52–54]. In contrast, protective miR-29, miR-106, miR-15b, and miR-16 were decreased in diabetes and high glucose HUVEC and retinal endothelial cell culture [55–57]. When these miRNAs were delivered or overexpressed, endothelial functions such as vasodilation were restored. Some miRNAs (e.g. miR-320) were further implicated in continued HUVEC dysfunction after euglycemia restoration (metabolic memory) [58].

LncRNAs have more than 200 nucleotides and are involved in transcriptional and post-transcriptional regulation as well as chromatin remodeling. A systematic screening for HUVEC lncRNAs showed that 100 lncRNAs were up-regulated and 186 were down-regulated by 24 h of high glucose [59]. Metastasis associated lung adenocarcinoma transcript 1 (MALAT1), a lncRNA inflammatory cytokine regulator known to increase in diabetic retina, increased in HUVEC exposed to high glucose and in diabetic mouse renal tissue. When MALAT1 was decreased via siRNA, glucose-induced ROS and inflammatory cytokines such as interleukin 6 (IL-6) and tumor necrosis factor-α (TNFα) decreased [60]. Several other lncRNAs, including retinal non-coding RNA 3 (RNCR3 or LINC00599), myocardial infarction associated transcript (MIAT), and maternally expressed 3 (MEG3) were up-regulated in diabetic human, mouse, and rat retinas as well as RF/6A retinal endothelial cells exposed to high glucose. Upon lncRNA knockdown, cell proliferation, migration, and tube formation decreased. RNCR3 and MIAT appear to be regulated by miR-185-5p and miR-150–5p, respectively [61–63]. These examples show pathophysiological interactions among different non-coding RNAs

Extracellular vesicles

EVs are particles delimited by a phospholipid bilayer that are naturally released from cells [64]. EVs have membrane proteins on the outside and contain genetic material (e.g. miRNA), lipids, and proteins such as adhesion molecules, membrane receptors, and cytokines on the inside. Circulating EVs are derived from endothelial and other blood cells, and their origin can be distinguished through membrane surface proteins [65]. EVs fall into three categories based on size and method of formation. Microparticles (0.1–1 μm diameter) are directly formed and released from the cell plasma membrane. Exosomes (less than 150 nm diameter) are generated within multivesicular endosomal compartments and released when the compartments fuse with the plasma membrane. Apoptotic bodies (50–5000 nm) are formed in the last stages of apoptosis, as the cell is disassembled. EVs facilitate cell–cell communication by directly interacting with target cell ligands to activate intracellular signaling or by transferring their contents to the target cell by fusion or endocytosis [66].

A recent meta-analysis demonstrated that circulating total microparticles, platelet-derived microparticles, and endothelium-derived microparticles were higher in type II diabetic patients than in non-diabetic controls [67]. Treatments such as pioglitazone decreased circulating endothelial microparticles in patients newly diagnosed with type II diabetes while metformin had a smaller effect [68]. HUVEC stimulated with high glucose in vitro released more EVs, and these vesicles decreased eNOS protein expression and endothelial-dependent relaxation while increasing pro-coagulant activity, ROS production, and caspase activity [69–71]. EV from high glucose-treated endothelial cells may communicate with neighboring cells, such as pericytes or monocytes, by releasing miRNA into the extracellular compartment [72,73]. However, conflicting reports show that high glucose does not change EV number or contents, perhaps because EV studies depend on endothelial cell and EV type as well as culture conditions.

SGLT2 inhibitors

Sodium-glucose cotransporters use energy from extracellular Na⁺ movement down its electrochemical gradient to facilitate glucose transport across the cell membrane. Sodium-glucose cotransporter-2 (SGLT2) is highly expressed in renal proximal tubules, where it mediates the reabsorption of ~90% of filtered glucose [74]. SGLT1, which has a higher glucose affinity, is expressed at lower levels downstream from SGLT2 in the kidney. SGLT1 is also expressed by epithelial cells in the small intestine, bile duct, pancreatic duct, and salivary glands [75]. While SGLT2 mRNA or protein has not been detected in untreated endothelial cells [76], recent studies suggest that stressors such as high glucose, hydrogen peroxide, and thrombin induce SGLT2 mRNA and protein expression, as well as SGLT2-mediated glucose entry [77,78].

SGLT2 inhibitors reduce glucose reabsorption in proximal renal tubules, facilitating urinary secretion of excess glucose and lowering blood glucose. In clinical trials, SGLT2 inhibitors such as empagliflozin and canagliflozin reduced cardiovascular disease, cardiovascular events, and total mortality [79,80]. Clinical data suggest this reduction in cardiovascular risk relates to improved endothelial function beyond glucose-lowering effects. Acute treatment of 16 Type II diabetic patients with dapagliflozin increased flow-mediated vasodilation and reduced pulse wave velocity, two gold-standard measures of in vivo endothelial function [81]. Eight weeks of empagliflozin treatment reduced pulse wave velocity in 40 normotensive Type I diabetic patients [82]. In a longer study, flow-mediated vasodilation increased in 40 Type II diabetic patients treated with metformin and dapagliflozin for 16 weeks as compared with 40 patients treated with metformin alone [83].

Pre-clinical studies in hyperglycemic diabetic animal models further confirm that SGLT2 inhibitors improve endothelial function [84]. C57BLKS/J-lepr^db/lepr^db diabetic mice fed dapagliflozin for 8 weeks or empagliflozin for 5 weeks had reduced arterial stiffness and improved response to the endothelial-dependent vasodilator acetylcholine [85,86]. Similarly, ipragliflozin (3 weeks) and empagliflozin (8 weeks) attenuated impaired endothelial-dependent vasodilation in streptozotocin-treated mice [87,88]. Empagliflozin also increased cardiac microvascular density and perfusion while improving microvascular barrier integrity in streptozotocin-treated mice, suggesting that SGLT2 inhibitors benefit both macro- and microvasculature [89].

SGLT2 inhibition additionally demonstrated direct impacts on endothelial function in vitro [90]. In porcine coronary artery endothelial cells cultured in high glucose, empagliflozin reduced glucose uptake and attenuated glucose-induced oxidative stress and reduced NO production [77]. HUVEC pre-treated with TNFα or high glucose decreased inflammatory adhesion molecule mRNA [91]. In human coronary artery endothelial cells treated with TNFα, empagliflozin, or dapagliflozin restored NO bioavailability perhaps via decreased intracellular ROS, since eNOS expression and signaling did not change [92]. Varied mechanisms are proposed for how SGLT2 inhibitors directly influence endothelial function, including reduced ROS [92,93], glycocalyx restoration [94], and decreased signaling pathway activation (e.g. PKC) [93]. Additional studies are essential to determining how SGLT2 inhibitors impact endothelial function.

Endothelial glucose metabolism

Endothelial cells primarily use glycolysis for ATP production despite ready access to oxygen, a process called aerobic glycolysis [95,96]. Glucose largely enters endothelial cells through members of the GLUT family of facilitative membrane transporters [97]. Glucose metabolism through glycolysis is then regulated by three rate-limiting enzymes: hexokinase, phosphofructokinase, and pyruvate kinase [98]. At the end of glycolysis, most pyruvate is converted to lactate to exit the cell, with less than 1% of glucose-derived pyruvate estimated to be shuttled into mitochondria for oxidative metabolism in the TCA cycle [96]. Instead, endothelial cells take up glutamine via the solute carrier family 1 member 5 (SLC1A5) transporter to supplement the TCA cycle [99]. Endothelial cells may preferentially use glycolysis over the TCA cycle to decrease ROS production, to reduce oxygen dependence so they can grow in hypoxic environments, or to rapidly produce ATP at the cell periphery.

Glucose plays an important role in angiogenesis, with recent studies showing that glycolysis is essential for endothelial cell migration whereas glutamine metabolism via the TCA cycle is essential for endothelial cell proliferation [100,101]. Endothelial cells rapidly increase glucose metabolism via glycolysis in response to pro-angiogenic signals such as growth factors [102,103]. In contrast, inhibition or silencing of the rate-limiting glycolytic enzyme phosphofructokinase-2/fructose-2,6-bisphosphatase-3 (PFKFB3) decreased vessel formation in vitro and in vivo [102,104–106]. Furthermore, when glycolysis was inhibited using 2-deoxy-d-glucose (2-DG), a structural glucose analog that is taken up by cells but cannot be metabolized via glycolysis, endothelial cells underwent ROS-triggered autophagy [107].

Glycolytic intermediates also impact endothelial function via metabolic pathways that branch off glycolysis [2]. The previously described polyol pathway metabolizes up to 33% of glucose when hexokinase is saturated in hyperglycemia. The pentose phosphate pathway (PPP) converts glucose 6-phosphate into the nucleotide synthesis precursor ribose 5-phosphate (R5P) [108] and is the primary generator of NADPH, which reduces oxidative stress [109,110]. Bovine aortic endothelial cells that overexpressed the PPP rate-limiting enzyme glucose 6-phosphate dehydrogenase (G6PD) decreased ROS and increased eNOS activity when exposed to hydrogen peroxide [111].

Further downstream in glycolysis, fructose-6-phosphate enters the HBP to produce UDP-N-acetylglucosamine (UDP-GlcNAc), the substrate for O-linked β-N-acetylglucosamine (O-GlcNAc) modification of protein serine/threonine residues [110,112]. Protein O-GlcNAcylation is important in vascular physiology and pathology [113,114]. UDP-GlcNAc production increases in a high glucose environment, leading to increased protein O-GlcNAcylation [34,115]. Aortic rings from high fat-fed mice with elevated O-GlcNAc had less vascular sprouting as compared with controls, which was abrogated by O-GlcNAc removal [116]. eNOS O-GlcNAcylation in diabetic animals decreased eNOS phosphorylation at serine 1177 and hence NO production [35].

Endothelial glucose metabolism is further linked to cell quiescence and activation by unidirectional laminar flow and oscillating disturbed flow, respectively. Prolonged exposure to unidirectional laminar flow (72 h) decreased HUVEC metabolism by increasing flow-sensitive transcription factor Kruppel-like factor 2 (KLF2) expression [117]. In contrast, disturbed flow increased glycolytic enzymes via ROS and HIF-1α, which increased glucose metabolism via glycolysis in HAEC [118]. Flow is also likely to regulate glycolytic side branch pathways, although this has not yet been explored.

Computational metabolic models enable predictions to be made across the cell metabolic network and are, therefore, a powerful supplement to experimental studies [119]. Stoichiometric models, which describe the metabolic steady state, do not require metabolic reaction kinetic parameters and can thus be constructed at genome scale [120]. The first large-scale stoichiometric metabolic models were developed for E. coli [121]. Since then, several genome-scale stoichiometric models were created for endothelial cells [122,123]. A model describing the metabolism of HUVEC tube formation was based on a generic genome-scale human metabolism model (Recon 1) that was pruned with endothelial protein mass spectrometry data [124]. More recently, an endothelial genome-scale model was developed to predict sepsis patient survival based on endothelial plasma metabolic markers [125]. This model pruned Recon 1 using transcriptomic data specific to HUVEC, microvascular endothelial cells, and pulmonary artery endothelial cells. Both models were then used to predict how endothelial glucose metabolism would shift with changes in metabolic enzyme activity.

Endothelial glucose transport

Endothelial cells transport glucose to the vascular wall and into the surrounding tissue. Glucose can cross the endothelium via paracellular transport between cells or via transcytosis across cells. However, endothelial cells in different vascular beds have significant structural differences that impact how glucose is transported. In this section, the blood-brain barrier (BBB) is, therefore, discussed separately from peripheral endothelial cells.

Transcellular glucose transporters include the previously described SGLT family as well as 14 members of the GLUT family (encoded by SLC2 genes). GLUTs 1–5 are the most highly studied and differ in their sugar specificity and kinetics, as well as in tissue localization and regulation. GLUTs 1, 3, and 4 have higher glucose affinity while GLUT2 has lower glucose affinity, and GLUT5 primarily transports fructose over glucose. GLUT1 is highly expressed in blood-tissue barriers, GLUT2 is primarily found in the abluminal membrane of the intestine and kidney epithelial cells, GLUT3 protein is largely restricted to brain, testes, and muscle, GLUT4 is the major glucose transporter of adipose tissue, skeletal muscle, and cardiac muscle, and GLUT5 is primarily expressed in the jejunum [126]. Most GLUTs are inhibited by cytochalasin B and phloretin.

Blood-brain barrier

The BBB is composed of brain microvascular endothelial cells (BMEC) supported by pericytes and astrocytes. BMEC are connected by adherens and tight junctions, which work together to reduce intercellular gaps to 1.4–1.8 nm and thereby make the BBB essentially impermeable to paracellular transport. BMEC also cannot form caveolae or transport molecules via caveolar-mediated transcytosis [127,128]. Therefore, the primary means of glucose transport in the BBB is via facilitative transporters.

The brain is highly dependent on glucose for its metabolic needs, and GLUT1 is widely considered the primary BMEC glucose transporter [129,130]. Immunogold cytochemistry revealed several times more GLUT1 at the abluminal BMEC plasma membrane than at the luminal plasma membrane [131]. This asymmetric GLUT1 distribution, as well as the lower glucose concentration in brain parenchyma (1–2 mM) as compared with blood (~5 mM), ensures rapid glucose transport into the brain [132]. The ratio of glucose metabolized vs. transported by BMEC is unknown, although early studies suggested that ~95% of 2-DG taken up by rat BMEC was not phosphorylated and was, therefore, available for transport [133].

Disrupted BBB glucose transport is likely important in disease. Cerebral glucose metabolism, measured by ¹⁸F-fluoro-2-deoxyglucose positron emission tomography, decreases in normal aging and in patients with Alzheimer’s disease even before amyloid and tau changes are detectable [134]. While reduced cerebral glucose metabolism has been considered a consequence of neurodegeneration-induced hypometabolism, recent studies suggest that reduced glucose transport by BMEC may be important in cognitive decline [135–137]. In humans, GLUT1 expression was reduced in Alzheimer’s patient brain capillaries post-mortem [138,139]. In mice, partial GLUT1 gene deletion decreased brain glucose uptake, reduced cerebral blood flow, disrupted the BBB, and led to neuronal dysfunctional and death [140,141].

Peripheral endothelial cells

Peripheral endothelial cells form a weaker barrier than BMEC; however, barrier tightness varies among peripheral vascular beds. Renal and pancreatic endothelial cells are fenestrated, enabling glucose to move freely across cells. Liver endothelial cells are discontinuous, which allows glucose to move freely via paracellular routes. And while aortic endothelial cells act as glucose barriers, corneal and retinal endothelial cells rapidly reach glucose equilibrium with the surrounding fluid, suggesting passive diffusion [142,143].

Detailed studies in coronary artery endothelial cells provide insight into glucose transport in this vascular bed. GLUT1 is still considered the primary endothelial glucose transporter in these cells, although GLUTs 1–5 and SGLT1 were observed by immunocytochemistry in en face preparations of rat coronary artery endothelial cells and by Western blot in cultured human coronary artery endothelial cells [144,145]. 2-NBDG, a fluorescent form of the glucose structural analog 2-DG, co-localized with glucose transporters, indicating that GLUTs and SGLTs are the primary means by which glucose enters these cells. In the intact artery, GLUTs were preferentially localized to the abluminal cell membrane, whereas SGLT1 was relatively evenly distributed. Glucose transporters were at or near cell–cell junctions in intact arteries [145]. In contrast, glucose transporters were more evenly distributed throughout the cell in human coronary artery endothelial cells cultured in vitro [144]. Thus, endothelial cells in culture may lose their polarity and therefore not be representative of in vivo glucose transporter distribution [146].

Once glucose enters the endothelial cell, some of it is phosphorylated by hexokinase into glucose-6-phosphate (G6P). Since G6P is not a GLUT influx or efflux substrate, G6P cannot exit the cell. Subcellular localization or vesicular compartmentalization may prevent glucose from phosphorylation and thereby enable glucose to be transported out of the cell. Alternatively, glucose may be incorporated into glycogen, which could be hydrolyzed near the basolateral membrane for transport [146]. Glucose efflux out of the endothelium is driven by abluminal membrane glucose transporter density and the glucose concentration gradient between cell and interstitial space. Microdialysis measurements in human muscle showed a gradient of up to 3 mM glucose between arterial glucose and muscle interstitial glucose after exercise or insulin stimulation [147,148]. These reduced interstitial glucose levels further confirm that the peripheral endothelium provides a significant glucose transport barrier.

Discussion

Our knowledge of endothelial glucose dysfunction, metabolism, and transport has advanced significantly, which enables us to better understand how metabolic changes contribute to disease. However, therapies designed to normalize glucose levels or decrease glucose-induced pathophysiology have not significantly reduced the disease burden. Additional mechanistic research into glucose effects on endothelial cells is needed to intelligently design therapies that target glucose-induced disease.

Quantitative measures of endothelial glucose dysfunction, metabolism, and transport are difficult to obtain. In vivo endothelial glucose metabolic and transport measurements are tricky since they are confounded by other metabolites and other cells and tissues. Yet simpler in vitro measurements also come with challenges, as endothelial cells lose their in vivo phenotype when cultured in vitro and small variations in culture conditions can lead to large variations in outputs. New techniques are needed to measure endothelial-specific glucose dysfunction, metabolism, and transport both in vivo and in vitro.

In most studies, endothelial glucose dysfunction, metabolism, or transport are considered in isolation. However, the endothelium dynamically integrates these glucose responses in health, and the disrupted interplay among the glucose responses may contribute to disease. For example, endothelial dysfunction but not metabolism or transport is studied in hyperglycemia, and endothelial metabolism and transport are studied independently but are tightly linked. Interdisciplinary studies that incorporate laboratory experiments and computational models of integrated endothelial glucose dysfunction, metabolism, and transport are essential to fully grasp the complexity of endothelial glucose responses.

Perspectives.

• Highlight the importance of the field: Endothelial glucose response is essential to a healthy vasculature and efficient nutrient transport throughout the body. Metabolic diseases such as diabetes lead to early and accelerated cardiovascular disease, and altered endothelial glucose metabolism and transport are important in cancer, pulmonary hypertension, and Alzheimer’s disease, among others.

• A summary of current thinking: Endothelial cells become dysfunctional in high glucose, and endothelial metabolism can be manipulated to change endothelial function. Therapies that target a single glucose response or pathway in the complex glucose processing network have not been highly effective in treating human disease.

• A comment on future directions: Future studies of endothelial glucose response should incorporate dysfunction, metabolism and transport into cell, animal, and human models that integrate experimental and computational approaches. These studies should also take into account endothelial heterogeneity, metabolic diversity, and parenchymal tissue interactions.

Abbreviations

2-DG

2-deoxy-d-glucose

AGE

advanced glycation end products

BBB

blood-brain barrier

BMEC

brain microvascular endothelial cells

DAG

diacyl glycerol

eNOS

endothelial nitric oxide synthase

EVs

extracellular vesicles

G6P

glucose-6-phosphate

HAEC

human aortic endothelial cells

HBP

hexosamine biosynthetic pathway

HUVEC

human umbilical vein endothelial cell

lncRNA

long non-coding RNA

MALAT1

metastasis associated lung adenocarcinoma transcript 1

MIAT

myocardial infarction associated transcript

NADPH

nicotinamide adenine dinucleotide phosphate

NO

nitric oxide

PKC

protein kinase C

PPP

pentose phosphate pathway

R5P

ribose 5-phosphate

ROS

reactive oxygen species

SGLT2

sodium-glucose cotransporter-

TCA

tricarboxylic acid

TNFα

tumor necrosis factor-α