Metabolic Regulation of Angiogenesis in Diabetes and Aging

Abstract

Impaired angiogenesis and endothelial dysfunction are hallmarks of diabetes and aging. Clinical efforts at promoting angiogenesis have largely focused on growth factor pathways, with mixed results. Recently, a new repertoire of endothelial intracellular molecules critical to endothelial metabolism has emerged as playing an important role in regulating angiogenesis. This review thus focuses on the emerging importance and therapeutic potential of these proteins and of endothelial bioenergetics in diabetes and aging.

The endothelium lines the blood vessel lumen, mediates blood flow homeostasis, and fulfills tissue metabolic demands by supplying nutrients and oxygen (26, 155). Most of the healthy endothelial cells (ECs) after adolescence are in quiescence, often for years, with new vessel generation (termed angiogenesis) occurring only in specific organs (e.g., endometrium during the luteal phase). ECs, however, display a significant plasticity to transform from a dormant to a highly angiogenic phenotype to (re)vascularize tissues when needed, such as during wound repair (133). Aging and many age-related ailments, such as diabetes (62, 158, 159, 161, 183, 204), muscle atrophy (4, 105, 170), and osteoporosis (102), are often accompanied by angiogenic failure or the inadequate ability of ECs to fulfill their normally required roles (termed EC dysfunction). This dysfunction results in tissue ischemia, giving rise to cardiovascular diseases or metabolic perturbations such as impaired glucose tolerance and excess lipid accumulation (lipotoxicity) (73, 75, 94, 123, 127, 161, 181).

Research for the past decades has uncovered critical pro- and anti-angiogenic growth factors, most notably including vascular endothelial growth factor (VEGF; also known as VEGFA) (26, 155). To rescue ischemic tissues, therapeutic angiogenesis with VEGF and other growth factors has been explored but has shown limited success (25, 160), underscoring our still insufficient understanding of the angiogenic process. Recently, it has become clear that EC metabolism contributes to the EC angiogenic phenotype and responsiveness to angiogenic growth factors, and thereby regulates angiogenesis (36, 49, 50, 67). In addition, many players in EC metabolism are also shown to be pivotal for age-related processes, including the control of stress response and longevity (70, 133, 154). In this review, in the first part, we will discuss a general view of EC metabolism and (sprouting) angiogenesis, and recent insights into how the former regulates the latter. This will set up a platform for the second part, in which we discuss how EC metabolism is altered in diabetes and age-related disorders, and potentially mediates the vascular dysfunction and angiopathy (angiogenesis impairment) associated with these conditions.

Metabolic Control of Angiogenesis

Unique Bioenergetics Regulation in ECs: High Dependence on Glycolysis

ECs are atypical nonmalignant cells that surprisingly depend on glycolysis to synthesize >80% of ATP (35, 37). Even under well-oxygenated conditions, the glycolysis rate in ECs is strikingly higher than in pericytes, smooth muscle cells, and any other nonvascular cell types thus far examined, and comparable to that of tumor cells, many of which use aerobic glycolysis for continuous growth and expansion (known as Warburg effect). Glycolysis is pivotal for the homeostasis of ECs, since near-total blockage of glycolysis [e.g., by 2-deoxyglucose (2-DG)] induces cell death (37, 49, 124, 177). Unlike in many other cell types, ATP generation in ECs relies minimally on fatty acid oxidation (FAO), glutamine oxidation, or glucose oxidation, i.e., on the mitochondrial respiratory chain (37, 152). Nevertheless, mitochondria do serve as a bioenergetic source for ECs under stressed conditions. Mitochondria also regulate cellular survival by mediating calcium handling and redox status by producing a moderate level of superoxide, and provide building blocks (acetyl-CoA) for cellular anabolic needs through the tricarboxylic acid (TCA) cycle (35, 45).

Within the vascular sprouts induced by angiogenic factors like VEGF, ECs upregulate glycolysis partly via increasing 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3), glucose transporter 1 (GLUT-1; mediates glucose transport across the plasma membrane), and lactate dehydrogenase-A (LDH-A; catalyzes the conversion of pyruvate and NADH into L-lactate and NAD⁺) (37, 147, 150, 217) (FIGURE 1). PFKFB3 most potently activates glycolysis through generating fructose-2,6-bisphosphate (F2,6P2), an allosteric activator of phosphofructokinase-1 (PFK-1) (198) (FIGURE 1). Knockdown of PFKFB3 and the consequent diminution of EC glycolysis in vitro decrease EC migration and proliferation, and blunt capillary sprouting from EC spheroids. Moreover, EC PFKFB3 gene ablation in vivo impedes vessel sprouting, branching, and outgrowth (37).

FIGURE 1.

Normal endothelial cell metabolism

A schematic of the principal pathways in healthy EC metabolism and the rate-limiting enzymes, carriers, and transporters (shown in blue). For simplification, not all metabolites or enzymes of the indicated pathways are shown. Glucose is transported via the constitutively expressed GLUT-1. ECs mainly depend on glycolysis to generate the needed energy to perform their functions. Mitochondria in ECs generate survival and angiogenic signals through ROS, and provide building blocks (such as acetyl-CoA) via TCA cycle for macromolecule synthesis. Glycolysis intermediates are shunted into side pathways for anti-oxidant defense and nucleotide synthesis (oxPPP and non-oxPPP) and glycosylation purposes (HBP). 3PG, 3-phosphogylcerate; Ac-CoA, acetyl-CoA; CPT1, carnitine palmitoyltransferase 1; DHAP, dihydroxyacetone phosphate; ETC, electron transport chain; F1,6P2, fructose 1,6-bisphosphate; F2,6P2, fructose 2,6-bisphosphate; F6P, fructose 6-phosphate; FA, fatty acid; FABP4, FA binding protein 4; FAO, FA oxidation; FAS, FA synthesis; FATP, FA transport protein; G3P, glyceraldehyde 3-phosphate; G6P, glucose 6-phosphate; G6PD, G6P dehydrogenase; GLS, glutaminase; Glc, glucose; Gln, glutamine; Glu, glutamate; GlucN6P, glucosamine 6-phosphate; GLUT-1, glucose transporter 1; HBP, hexosamine biosynthetic pathway; α-KG, α-ketoglutarate; Lac, lactate; LDH, lactate dehydrogenase; MCT, monocarboxylate transporter; NADPH, nicotinamide adenine dinucleotide phosphate; (Non-)oxPPP, (non-)oxidative branch of the pentose phosphate pathway; OAA, oxaloacetate; OXPHOS, oxidative phosphorylation; 6-PG, 6-phosphogluconate; 6-PGL, 6-phosphogluconolactone; PFK1, phosphofructokinase 1; PFKFB3, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3; Pyr, pyruvate; R5P, ribose 5-phosphate; ROS, reactive oxygen species; Ru5P, ribulose 5-phosphate; TCA, tricarboxylic acid cycle; TKT, transketolase; UDP-GlcNAc, uridine diphosphate N-acetylglucosamine.

Glycolysis also provides important metabolites that act as precursors for biosynthetic pathways via, for example, the pentose phosphate pathway (PPP) and hexosamine biosynthesis pathway (HBP) (FIGURE 1). The PPP harbors oxidative and non-oxidative branches. The rate of oxidative PPP is dictated by glucose-6-phosphate dehydrogenase (G6PD), the rate-limiting enzyme whose activity is in part determined by VEGF in ECs (144). The oxidative branch (oxPPP) of the PPP synthesizes NADPH from NADP⁺ and thus confers reducing power to regulate cellular redox state and biosynthetic potentials (164) (as demonstrated in tumor cells). Both oxidative and non-oxidative PPP branches generate ribose-5-phosphate (R5P), which is required for the biosynthesis of nucleotides (145) (as demonstrated in embryonic stem cells). Metabolites of PPP can also be converted back to the intermediates of glycolysis by transketolase (TKT) and transaldolase. HBP converts fructose-6-phosphate (F6P), a glycolysis intermediate, into glucosamine-6-phosphate (GlucN6P), which is then metabolized to uridine diphosphate N-acetylglucosamine (UDP-GlcNAc), an important substrate for glycosylation critical to numerous aspects of EC function (103) (FIGURE 1).

It may at first appear counterintuitive that ECs rely largely on glycolysis to synthesize ATP, in light of their continuous exposure to circulating oxygen and the 20-fold higher efficiency of oxidative phosphorylation (OXPHOS) to generate ATP from one glucose (115). Several possible reasons for this observation exist. First, the consumption of oxygen by ECs could decrease the abundance of transvascular transport of oxygen to surrounding tissue. Second, the low reliance of ECs on oxygen may allow them to more efficiently undergo sprouting and vascularization of hypoxic tissues. Third, exposure to high oxygen tension renders ECs vulnerable to oxidative stress, and lower OXPHOS and increased flux through oxPPP may serve to mitigate this stress. Fourth, although glycolysis is less efficient than OXPHOS, it is far more rapid, so that when glucose supply is not limiting, the rate of ATP production by glycolysis can exceed that of OXPHOS within minutes of growth factor stimulation (115, 152). Fifth, glycolytic enzymes are more compact and can be packaged in large actin-bound multi-enzyme glycolytic complexes, called metabolons, that likely form in lamellipodia and filipodia, the sites where rapid generation of ATP is most needed for rapid actin polymerization and depolymerization and angiogenesis to occur (37, 58, 88, 153, 162, 166, 219). Finally, the shunting of glycolysis intermediates to PPP and HBP would help sustain cell division and migration of sprouting ECs by providing biosynthetic power and redox homeostasis.

Mechanisms of Sprouting Angiogenesis and Vessel Formation: Interplay Between VEGF, Notch and Glycolysis

Upon tissue exposure to low oxygen tension, the key angiogenic factor VEGF (VEGFA) is produced and promotes vessel sprouting to restore tissue oxygenation (FIGURE 2, A AND B). When the vascular front reaches a VEGF gradient, VEGF binds to VEGF receptor (VEGFR) 2 on ECs and induces filopodia and lamellipodia. The EC that encounters the highest level of VEGF is destined to become a tip cell. VEGFC, a ligand of the VEGFR2 and VEGFR3, also activates tip cells (26). The tip cells migrate, take the lead at the vascular front, and coordinate with stalk cells that proliferate and elongate the newly developing branch (155). Vessels must match perfusion with tissue metabolic demands. This is accomplished by vessel remodeling and regression (pruning) (99). The blood flow initiation allows vascular connections to stabilize and keeps vessels quiescent and anti-thrombogenic, whereas abrogated flow causes endothelial retraction and apoptosis (26, 99, 155). Understanding the basic mechanism governing vessel formation is critical since, although excessive EC quiescence and lack of responsiveness to VEGF might underlie diabetic angiopathy, abnormally enhanced regression may cause vascular rarefaction, a hallmark of age-associated pathologies such as neurodegenerative diseases and muscle atrophy (sarcopenia), and a process poorly understood (4, 155). Neo-vessels must undergo maturation and attain stability to nurture tissue homeostasis. Maturation of nascent vessels entails the establishment of junctional barriers, incorporation of mural cells, and generation of extracellular matrices (basement membrane) to specialize the vessel cells to tissue-specific requirements and to form tightly aligned, long-lived ECs with cobblestone-like morphology (called phalanx cells) that allow appropriate tissue perfusion (FIGURE 2A) (2, 26, 155).

FIGURE 2.

Roles of Notch and glycolysis in tip-stalk cell regulation of sprouting angiogenesis

A: diagram showing principal steps of angiogenic vessel growth. B: model of tip-stalk cell specification, stalk-to-tip feedback loop, stalk-to-stalk synchronization, and tip-stalk inter-conversion, mediated by VEGF-Dll4-Notch signaling. Tip cells, by extending filopodia and lamellipodia, migrate toward the angiogenic factor VEGF, whereas stalk cells proliferate and elongate the emerging vessel sprout. VEGFA activation of VEGF receptor 2 (VEGFR2), and VEGFC activation of VEGFR3, respectively, upregulate Dll4 in tip cells. Dll4 in tip cells instructs neighboring ECs to become stalk cells by activating their Notch receptors, which release Notch intracellular domains (NICD) and induce VEGFR1 (and its soluble form sFlt1) that acts as a VEGF trap. NICD also represses VEGFR2 and VEGFR3 expression, thereby leading to the suppression of tip cell behavior. Notch signaling can induce Dll4 that signals back to tip cells to repress their migratory behaviors, forming a negative feedback loop that may underlie tip-stalk dynamic switching. Stalk cell Dll4 action to neighboring cells can be competitively antagonized and fine-tuned by Jagged (JAG), a Notch ligand highly expressed in stalk cells. Stalk cell Dll4 also activates Notch in a neighboring stalk cell, which could stabilize the stalk-stalk structure through synchronization of Notch activity. Conversely, reduced NICD activity and resultant increase in VEGFR2/Dll4 abundance in stalk cells can switch them to tip cells (inter-conversion), which then instruct the previous tip cells, through Dll4-Notch signaling, to assume a stalk cell phenotype. In all cases, tip cells display higher glycolysis rate than stalk cells. C: VEGF and Notch signaling regulate glycolysis. VEGF increases glycolytic flux through upregulation of GLUT1 mediated by Akt and through induction of PFKFB3, the glycolysis activating enzyme. In ECs in which Notch is active, PFKFB3 activity and glycolysis rate are reduced.

Tip and stalk cells are interchangeable EC phenotypes rather than determined cellular fates (87) (FIGURE 2B). VEGF-Notch signaling dictates tip vs. stalk cell specification, as has been reviewed in detail elsewhere (51, 155, 165). In essence, VEGFR2 activation increases the abundance of Notch ligand Delta-like 4 (Dll4) in tip cells, which binds to the membrane-anchored Notch expressed on the neighboring ECs (i.e., stalk cells) and triggers in these cells Notch-mediated downregulation of VEGFR2 and VEGFR3, and upregulation of VEGFR1 [also known as Fms-related tyrosine kinase 1 (Flt-1)] (FIGURE 2B). VEGFR1, existing as both membrane-bound and soluble secreted forms (the latter also known as sFlt-1), traps VEGF as a decoy. These processes render these adjacent ECs (stalk cells) more resistant to VEGF and less migratory, thus suppressing their ability to become another tip cell. Notch signaling in stalk cells also induces Dll4, which propagates Notch activation to neighboring (tip or stalk) cells (24). This propagation by stalk cell Dll4 of Notch signaling in neighboring cells is competitively intervened by Jagged, Notch ligands that are expressed in stalk cells (51). Fine tuning of Notch signaling within stalk cells occurs via cell-autonomous negative feedback mediated by Notch-regulated ankyrin repeat protein (Nrarp) (151) (FIGURE 2B). These feedback and feedforward regulations are thus thought to underlie fine-tuning of Notch signaling and drive the dynamic tip-stalk cell fate inter-conversions, as well as the stalk-stalk signal synchronization, at the vascular front (6, 87). Through continuous cell shuffling, the ECs that most quickly attain the highest VEGFR2-to-VEGFR1 ratio and Dll4 levels gain competence to become a tip cell (FIGURE 2B). Deregulated or sustained Notch signaling to various degrees causes cell cycle arrest (quiescence), senescence, apoptosis, or vascular regression (112, 113, 131, 202), which may lead to diabetic angiopathy defective in tip cell specification and migration, and aging-associated vascular rarefaction (see also FIGURE 3D).

FIGURE 3.

Endothelial angiogenic signaling by metabolic regulators

A: cross-coupling of pro-angiogenic (blue) and anti-angiogenic (red) metabolic regulators in ECs with reactive oxygen species (ROS) and bioenergetic machineries (glycolysis, OXPHOS). B: pro-angiogenic effects of EC fatty acid metabolism. FAO is coupled with dNTP synthesis and EC proliferation. C: Mechanisms for lactate, the glycolysis metabolite, to act as a pro-angiogenic signaling molecule. D: hypothetical mechanisms by which EC bioenergetic regulators mediate EC dysfunction associated with aging. 3-HIB, 3-hydroxyisobutyrate; AMPK, AMP-activated protein kinase; CPT1A, carnitine palmitoyltransferase 1A; FABP, fatty acid-binding protein; FATP, fatty acid transporter; NAD⁺, nicotinamide adenine dinucleotide; Nampt, nicotinamide phosphoribosyltransferase; PGC-1α, peroxisome proliferator activated receptor gamma coactivator 1 α; PHD2, prolyl hydroxylase domain protein 2. See text for other abbreviations.

ECs express wingless-type (Wnt) ligands and their frizzled (FZD) receptors, pathways known to stimulate EC proliferation. Notch activates Wnt signals in stalk cells, which accounts for the growth-promoting effects of Notch signaling on stalk cells, which would otherwise lead to EC quiescence (26, 151). Conversely, Wnt activates Notch pathway through transcriptional activation of Dll4 (31, 155).

VEGF augments glycolysis approximately twofold in ECs by increasing PFKFB3 expression, and does so before ECs begin to migrate. Conversely, pro-stalk Notch signaling induced by Dll4 represses glycolysis and provokes the opposite effect, suggesting that glycolysis rates are increased in tip cells compared with stalk cells (37) (FIGURE 2C). Stalk cells, however, still rely on glycolysis to exert their functions, because the active proliferation of stalk cells requires substantial generation of ATP and macro-molecules derived from the PPP and other side branches of the glycolytic pathway (199). Remarkably, forced expression of PFKFB3, despite leading to mild increase of glycolytic flux, overrides the stalk-inducing effect of Notch signal and prompts stalk cells to assume a tip cell phenotype, whereas PFKFB3 knockdown leads to opposite effects (37). Of note, forced induction or repression of PFKFB3 did not affect the gene expression signature for either tip or stalk cells. This indicates that PFKFB3 did not modulate tip vs. stalk cells via affecting tip/stalk cell gene expression. Rather, these results suggest that metabolism (e.g., PFKFB3-driven glycolysis) determines vessel branching independent of genetic signals.

The HBP pathway may also be relevant to the metabolism-angiogenesis link. Abolition of this pathway, among others, by 2-deoxyglucose (2-DG) halts angiogenesis, and, importantly, angiogenesis is rescued by mannose, the precursor for N-linked glycosylation (124). It is therefore conceivable that the HBP in ECs transmits nutrient status and remodels blood vessel supply by generating UDP-GlcNAc that activates key angiogenic receptors through posttranslational modifications. Indeed, VEGFR2 function depends on N-glycosylation (194). Also, it is noteworthy that Notch responsiveness to its ligands, Dll4 and Jagged1, requires its extracellular domain glycosylation (10).

Endothelial Metabolic Regulators: Coordination With Angiogenesis

Numerous model organisms have now revealed that longevity-promoting mutations often occur on genes that encode components of metabolic regulatory circuits (68, 95, 121). The findings, combined with developmental and postnatal studies in mice, have increasingly revealed that the mammalian orthologs of these longevity genes, although they may or may not prolong lifespan in higher organisms (81), do coordinate a plethora of cell functions that are commonly deteriorated in aging, such as metabolism, stress resistance, cell cycle progression, and programmed cell death, and thus critically contribute to the extension of disease-free aging (“health span”) (55, 76, 81, 133, 171, 178, 220). Burgeoning evidence has revealed that many of these key bioenergetics regulators also play crucial roles in angiogenesis, particularly in ischemic conditions. Key metabolic mediators include forkhead box “O” (FOXO) transcription factors (133, 142, 157, 212), nicotinamide phosphoribosyltransferase [Nampt; the rate-limiting enzyme in mammalian nicotinamide adenine dinucleotide (NAD⁺) biosynthesis] (14, 15, 206–209), NAD⁺-dependent protein deacetylases termed sirtuins (69, 70, 133, 154, 156), AMP-dependent kinase (AMPK) (128, 140, 141), liver kinase B1 [LKB1; also known as serine/threonine kinase 11 (STK11), an upstream kinase that activates AMPK] (134, 218), and PPARγ coactivator-1α (PGC-1α) (13, 173) (FIGURE 3). We review each of these below.

FOXOs.

FOXOs, transcription factors commonly harboring a highly conserved DNA binding domain (forkhead box) (133, 171, 197), are expressed in ECs (FOXO1, FOXO3a, and FOXO4 isoforms) (142, 157). The prototypes of FOXO target genes include those that are relevant to cell cycle progression (p21, p27), reactive oxygen species detoxification (MnSOD, coding for manganese-dependent superoxide dismutase), programmed cell death (FAS, coding for Fas cell surface death receptor; BIM, coding for Bcl2-interacting protein), and glucose metabolism (G6PC, coding for glucose-6-phosphatase) (119, 125, 129, 133, 171, 197), all of which are important in EC biology. The transcriptional activity of FOXOs is modified by a variety of posttranslational modifications (PTMs) that affect transcriptional capacity, DNA binding, subcellular localization and protein stability. Depending on PTMs and target genes, FOXOs can either activate or repress gene transcription. FOXOs are critical effectors of Akt, which phosphorylates and inactivates FOXOs by excluding them from nuclear localization. This event is required for PI3K-Akt to exert its potent pro-angiogenic effects downstream of VEGF (1, 96) (FIGURE 3A). Consistent with this role, FOXO1 overexpression abrogates EC migration and sprouting, whereas silencing of FOXO1 augments angiogenic capacity of ECs (157, 167, 182). Interestingly, global deletion of FOXO1 led to embryonic lethality by E10.5 due to a failure in forming vascular network (59, 82), perhaps because FOXO1, which is intensely expressed in a variety of embryonic vessels, also modulates tip stalk cell specification and balanced sprouting. FOXO3a and FOXO4 are not needed for embryonic vascular development, but they synergize with FOXO1 to suppress vascular growth in adult mice (28, 82), since inducible global triple knockout of the three isoforms in adult mice causes more widespread excessive proliferation of ECs and hemangioma formation compared with FOXO1 alone (142). Furthermore, deficiency of FOXO3a enhances recovery of vessel density in murine ischemic limbs (157). These studies indicate that ECs have high sensitivity to changes in FOXO pathways and that appropriate FOXO activity is essential to keep ECs quiescent. Very recently, Wilhelm et al. demonstrated a vascular-metabolic link, whereby FOXO1 serves as a master regulator of EC quiescence and attenuates metabolism of the cell by repressing glycolysis and mitochondrial respiration (212). Mechanistically, FOXO1 inhibits MYC, a nodal point that powerfully drives cell growth and anabolic metabolism via upregulation of glycolytic enzymes such as LDHA, LDHB (coding for lactate dehydrogenase A and B), ENO1 (coding for enolase 1), and PKM2 (coding for pyruvate kinase, muscle2 isoform). In ECs with forced expression of FOXO1, restoring MYC signaling rescued metabolic activity and branching behavior.

SIRT1, NO, Nampt.

FOXOs functions are also fine-tuned by reversible acetylation. Depending on the target promoter of FOXOs, acetylation can either enhance or blunt FOXO capability to transactivate gene expression (8, 22, 171, 196). Acetylation can also stabilize FOXO proteins by suppressing polyubiquitination and ensuing protein degradation. Conversely, SIRT1-mediated deacetylation of FOXOs alters the transcriptional capacity, subcellular localization, and protein stability of FOXOs (8, 22, 171, 196). Sirtuins, including SIRT1, are NAD⁺-dependent deacetylases that link metabolism to a broad spectrum of biological processes, ranging from oxidative stress responses and DNA repair to energy metabolism (55, 71, 76, 178). Sirtuin deacetylase activity requires NAD as substrate, yielding nicotinamide. A connection between sirtuins and longevity has been evident for over 10 years based on work done in model organisms such as worms and flies (21, 91, 116, 203) but debated in terms of their roles in mammalian anti-aging (81). What is certain is that sirtuins are crucial in implementing adaptive responses to broad cellular stresses that provoke tissue generation and age-related ailments (29, 32, 55, 71, 72, 76, 111, 116, 210), as demonstrated, for example, in the phenotypes of SIRT1-overexpressing mice that are resistant to cancer or diabetes (but do not live longer) (81). Genetic ablation of SIRT1 in mouse ECs is compatible with life but deteriorates revascularization in ischemic tissues, and abolishes EC branching and proliferation, leading to attenuated vascular density (156). This is partly due to de-repression of FOXO1 activity, since SIRT1 deacetylates EC FOXO1, thereby restricting its anti-angiogenic activity (156) (FIGURE 3A). Additionally, a recent report showed that SIRT1 directly targets Notch (69). By deacetylating NICD and thereby destabilizing it through proteasomal degradation, SIRT1 reduces the amplitude and duration of the Notch activity in a negative feedback loop (69). This promotes the vascularization of the nutrient-deprived tissues. In the absence of SIRT1 (mimicking nutrient abundance), ECs exhibit propensity to develop higher Notch signaling, resulting in a stalk cell-like phenotype with impaired vascular outgrowth (156). Thus FOXO1 and Notch, both fine-tuned by SIRT1 and thus NAD-dependent metabolism, cooperatively maintain EC quiescence and stalk cell phenotype.

Nitric oxide (NO) synthesized by endothelial NO synthase (eNOS) crucially mediates EC function and angiogenesis (174). The activity of eNOS is augmented when phosphorylated by Akt, downstream of VEGF and other factors. Importantly, SIRT1 deacetylates and activates eNOS, which may partly underlie pro-angiogenic signaling by SIRT1 (122). Conversely, SIRT1 is also activated by NO (139), suggesting a reciprocal regulatory loop. SIRT1 can also be activated by increases in cellular NAD⁺ level, which are maintained via a Nampt-mediated salvage pathway that recycles nicotinamide into NAD⁺. Hyperglycemia blocks SIRT1 from inhibiting FOXO1, causes premature senescence, and abrogates angiogenic capacity of ECs, all of which are rescued by forced expression of Nampt (15). Importantly, the Nampt/SIRT1 pro-angiogenic signal depends on an increase in glycolytic flux, again emphasizing the importance of bioenergetics in neovascularization (FIGURE 3A). These findings may link with the intriguing notion that glycolysis and its side branches can regulate cellular life span, in part by bypassing senescence induced by oxidative damage (43, 44, 98, 109).

PGC-1α.

PGC-1α is a transcriptional coactivator that powerfully exerts broad metabolic effects, including mitochondrial biogenesis, by co-activating numerous transcription factors including most of the nuclear hormone receptors (110, 168). EC PGC-1α cooperates with FOXO3a to detoxify ROS (135, 136, 195). It has been proposed that EC PGC-1α inhibits cell migration via ROS reduction and that the pro-migratory effect of NO is exerted by Akt-mediated inhibition of PGC-1α (13). In addition, our group recently demonstrated that EC PGC-1α suppresses EC migration in culture and ischemia-induced angiogenesis in vivo through activation of Notch signaling and inhibition of Akt/eNOS (173) (FIGURE 3A). Complete ablation of PGC-1α in ECs phenocopied FOXO1 or Dll4 deficiency (114, 157, 187) and caused VEGF-independent, aberrant augmentation of EC migration/capillary formation (60, 90, 173). Partial genetic ablation of EC PGC-1α, which normalized the incremental increase in PGC-1α caused by diabetes, led to functional recovery of tissue perfusion in ischemic legs (173). Thus PGC-1α appears to mediate the suppression of angiogenesis caused by diabetes.

AMPK, fatty acid-binding proteins, and transporters.

AMPK, stimulated by elevated AMP levels or LKB1, promotes angiogenesis by upregulating VEGFA (141) in skeletal muscles (FIGURE 3B). AMPK in ECs may additionally induce EC sprouting via its well-established ability to inactivate acetyl-CoA carboxylase (ACC), which synthesizes malonyl-CoA, an inhibitor of carnitine palmitoyltransferase (CPT) 1A and FAO (33, 56). Ablation of EC CPT1A, a rate-limiting enzyme of FAO, impairs de novo deoxyribonucleoside triphosphates (dNTPs) synthesis for DNA replication, resulting in sprouting failure owing to blunted proliferation of human and murine ECs (176). The mechanism by which CPT1A carries out this effect is not clear, since it cannot occur via FAO-mediated anaplerosis (175).

Other proteins that may coordinate EC metabolism with angiogenesis include fatty acid-binding protein 4 (FABP4) and VEGFB (FIGURE 3B). FABP4 is an intracellular lipid chaperone directly targeted by VEGF, as well as by Dll4-induced Notch signaling (78). FABP4 stimulates EC proliferation in vitro (52) and angiogenesis in vivo (63), thus also potentially bridging FA metabolism and angiogenesis (78). VEGFB, secreted by muscle cells, promotes endothelial lipid uptake and passage to peripheral tissues (cardiac and skeletal muscle) through the upregulation of fatty acid transporters FATP3 and FATP4 (74) [although the finding has been put into question (41)]. In agreement with findings that ectopic lipid deposition in muscle underlies the pathogenesis of insulin resistance, in vivo delivery of neutralizing VEGFB antibodies reduced endothelial-to-tissue lipid transport and restored insulin sensitivity and glucose tolerance in rodent models of diabetes (27, 75). In addition, both VEGFA and VEGFB stimulate AMPK in ECs, likely affecting metabolism, such as increasing FAO (163).

Metabolites of glycolysis and amino acid catabolism.

Finally, metabolites themselves can regulate EC metabolism, independently of their role as enzymatic substrate or product. For example, lactate, which is not significantly metabolized by ECs (201), at least under physiological extracellular glucose concentrations (101), can control angiogenesis by acting as a signaling molecule in an intracrine, autocrine, or paracrine fashion (147, 148) (FIGURE 3C). Extracellular lactate is transported into ECs via the mono-carboxylate transporter 1 (MCT1). Lactate competitively blocks the oxygen-sensing prolyl hydroxylase domain protein 2 (PHD2), and thereby activates hypoxia-inducible factor-1α (HIF-1α) as well as nuclear factor kappa-B (NF-κB) (a pseudo-hypoxic state). These lead to a secondary increase in VEGF, basic fibroblast growth factor (bFGF) (86, 184) and pro-angiogenic interleukin 8 (IL-8) (201), and activation of receptor tyrosine kinases: Axl, the angiopoietin receptor Tek (Tie2), and VEGFR2 (169). Lactate also engages the phosphoinositide 3-kinase (PI-3K)/Akt pathway, and reduced Akt and angiogenic activity of ECs caused by PFKFB3 knockdown (i.e., glycolysis inhibition) is rescued by lactate (214). Lactate also induces pro-angiogenic ROS in ECs (201). Recently, we have also described the modulation of FA uptake in ECs by 3-hydroxyisobutyrate (3-HIB), an intermediate of valine catabolism (89). Muscle cells catabolize valine, leading to paracrine secretion of 3-HIB, which in turn stimulates EC FA uptake and transport to underlying muscle tissue. Excess 3-HIB leads to excess accumulation of lipid intermediates in skeletal muscle and ensuing insulin resistance. The process depends on EC FATP3 and 4 (FIGURE 3B), akin to VEGFB as described above. This work links the regulation of amino acid catabolism with that of fatty acid catabolism; identifies a novel and little-studied bioactive metabolite; and provides an explanation for the epidemiological observation that valine and other branched chain amino acids are elevated in states of insulin resistance.

Mechanism of Diabetic and Age-Associated Angiopathy: Role of Cellular Redox and Bioenergetics

Age-Related Vasculopathies and Their Mechanistic Similarities to Diabetic Vascular Dysfunction

Endothelial dysfunction and impaired angiogenesis are hallmarks of aging and age-related vascular diseases, and are worsened in the setting of Types 1 and 2 diabetes (T1 and T2 DM) (4, 127, 161, 204). Diabetes, T2DM in particular, is a critical age-associated condition, with 26% of Americans ≥65 yr old having DM compared with 9% in the general population. There is an emerging notion that diabetic milieu could be permissive to the development of senescence, whereas cellular senescence may mediate progressive inflammation and underlie DM pathogenesis (143). In fact, aging, like DM, is a major risk factor for cardiovascular disease and is associated with decreased vascular density (18, 40, 205) akin to diabetic rarefaction (159, 183, 204) and the decline of angiogenic capacity (4, 5, 53, 102, 104, 215). These vasculopathies occur in multiple organs, like brain, muscle, kidney, bone, and heart, contributing to age-related pathologies such as dementia (5, 12), Alzheimer’s disease (18), sarcopenia (4, 105, 170, 205), and bone fracture (102), and exacerbate ischemic injury in aged population (53). As such, induction of angiogenesis (e.g., by exercise) has been proposed to ameliorate age-associated ailments (4, 12, 102, 105, 158). Mechanisms of endothelial dysfunction in aging and DM, therefore, seem to at least partly overlap. Given that the impact of aging on EC metabolism has not been as well-defined as that of hyperglycemia (FIGURE 4), application of the mechanisms underlying diabetic angiopathy may help in understanding the metabolic process of EC aging. Many preclinical studies of vascular aging have used hyperglycemia, insulin resistance, or diabetes to model cellular senescence (Table 1).

FIGURE 4.

Diabetic endothelial cell metabolism

In diabetic ECs, hyperglycemia increases production of ROS through a variety of intracellular sources, including NADPH oxidase, xanthine oxidase, uncoupled eNOS, and impaired pentose phosphate pathway (PPP). These result in slowed glycolytic flux. As a consequence, glycolytic intermediates are diverted into side-branch pathways such as the polyol pathway, glycation pathway, and dysregulated glycosylation, leading to further excessive ROS and advanced glycation end products (AGEs) production. 1,3BPG, 1,3-bisphosphoglycerate; 3DG, 3-deoxyglucosone; AGEs, advanced glycation end products; AR, aldose reductase; DAG, diacylglycerol; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; PARP1, polyADP-ribose polymerase 1; PKC, protein kinase C.

Table 1.

Reported changes of endothelial metabolic regulators and their downstream effects (as confirmed by rescue studies through gain or loss of function), in diabetes and aging-induced endothelial dysfunction

	Models	Expression/Activity	Changes in Downstream Signals	Changes in metabolism and redox	Downstream Vascular Phenotypes	Refs.
NAD+/NADH	HG-induced EC aging	Decreased	SIRT1 activity↓	Glycolysis↓	Angiogenesis↓	15
Nampt			FOXO1 activity↑	ROS↑	Premature senescence↑
	T2DM (db/db)	Decreased		eNOS dysfunction	EPC mobilization↓	209
SIRT1	HG or HF-induced	Decreased	p53 activity ↑	ROS ↑ (p66Shc↑, SOD2↓)		7, 30, 137, 191, 220, 223
					Angiogenesis↓
					Premature senescence↑
	EC aging;				Atherosclerosis
	ApoE−/− mice		FOXO1 activity ↑	eNOS dysfunction	Inflammation (Leucocyte attachment)↑
	EC aging	Decreased	p53 activity ↑	ROS ↑	Premature senescence↑	92, 138
				eNOS dysfunction
FOXOs(FOXO1, FOXO3a)	HG; T1 and T2 DM mice	Expression↑		ROS (ONOO-)↑	Angiogenesis↓	9, 83, 149, 189
					Apoptosis↑
		Nuclear localization↑		eNOS dysfunction	Inflammation (TNF-α)↑
	Human obesity;	Expression↑		eNOS dysfunction	Angiogenesis↓	93, 132, 193
					Insulin resistance (systemic) ↑
	HF-fed mice, LDLR−/− mice	FOXO phosphorylation↓			Atherosclerosis
PGC-1α	HG; T1 and T2 DM mice (HF, db/db)	Increased	Notch pathway↑	eNOS dysfunction	Angiogenesis↓	173
			Akt/eNOS↓

HG, high glucose; HF, high fat; T1 and T2 DM, Type 1 and Type 2 diabetes.

Although aging is a complex process involving telomere erosion, DNA damage, oxidative stress, or oncogenic activation, evidence suggests the relevance of bioenergetic perturbation, including mitochondrial dysfunction and attenuated glycolytic flux in aging. In fact, PGC-1α overexpression was reported to induce senescence in fibroblasts (213), whereas Notch signaling, the PGC-1α target in ECs, has been shown to promote EC senescence (112, 202). FIGURE 3D illustrates a hypothetical model in which metabolic regulators that induce EC quiescence (i.e., Notch, FOXO1, and PGC-1α) mediate vascular rarefaction and impaired angiogenesis in aging through inhibition of glycolysis, a bioenergetic change similar to that observed in hyperglycemia-exposed ECs, as detailed below (FIGURE 4).

Dysregulated Metabolism of Diabetic ECs

Various metabolic alterations have been suggested to underlie diabetic vascular dysfunctions, including elevated oxidative stress owing to production of multiple types of ROS, such as hydrogen peroxide and superoxide, and reactive nitrogen species (RNS) (19, 39, 50, 64, 65, 67, 161). Because ECs constitutively express a glucose transporter GLUT1, hyperglycemia in circulation proportionately induces high glucose in the cell (FIGURE 4). Hyperglycemia can produce ROS from both the cytosol and mitochondria. Increased superoxide anions, derived from xanthine or NADPH oxidases [the major ROS source in diabetic EC (16)], can directly react with NO to yield peroxynitrite (ONOO^–), which ultimately decreases the NO levels. Hyperglycemia also slows the PPP flux by blocking G6PD, thereby decreasing production of the major cellular reductant NADPH, and promotes oxidative stress (221). Cellular NADPH is needed for conversion of oxidized glutathione (GSSH) into reduced GSH, a critical ROS scavenger (106, 222). Therefore, depletion of NADPH levels contributes to ROS accumulation (66). Forced expression of G6PD mitigates oxidative stress in high glucose-treated ECs (106, 222).

A currently accepted “unifying hypothesis” posits that ROS and RNS activate the enzyme poly-ADP-ribose polymerase 1 (PARP1) through induction of DNA strand breaks (19, 65). The glycolytic enzyme GAPDH is ADP-ribosylated by PARP1 (47), which slows glycolytic flux and causes glycolytic intermediates to accumulate and be diverted into three glycolysis branch pathways that are shown to be augmented in ECs exposed to high glucose: 1) the polyol pathway, 2) the HBP, and 3) the glycation pathway. In preclinical studies, accentuated flow of all of these pathways by hyperglycemia was prevented by blockage of PARP1 with competitive inhibitors in aortic ECs (47). Excessive glucose is shunted into the polyol pathway and converted to sorbitol by aldose reductase (AR), the rate-limiting enzyme (118). This reaction additionally exhausts NADPH and further increases ROS. Sorbitol is then converted to fructose by sorbitol dehydrogenase, leading to the production of 3-deoxyglucosone, a highly reactive α-oxo-aldehyde that non-enzymatically generates toxic advanced glycation end-products (AGE) (211). AGEs reinforce vascular complications by cross linking the basement membrane components and altering the extracellular matrix. AGEs also bind to their receptor (RAGE) on ECs, promoting inflammation, leakage, and ROS production in the blood vessel (120). F6P overload also increases the HBP flux, which, as described above, is critical for protein glycosylation, but blunts normal angiogenic capacity under hyperglycemia. Increased O-glycosylation, for example, reduces eNOS activity (54). G3P and DHAP are diverted toward methylglyoxal pathway, further promoting ROS and AGE production. In addition, G3P and DHAP are used to synthesize diacylglycerol (DAG) that subsequently activates protein kinase C, a key driver of vascular abnormalities (34). Multienzyme glyoxalase system, of which glyoxalase-1 (Glo1) is rate-limiting, detoxifies methylglyoxal by converting it into pyruvate (192). Forced expression of Glo1 rescues hyperglycemia-induced angiogenesis defects in vitro, and transgenic overexpression of Glo1 in rats ameliorates vascular AGE formation and improves vasoreactivity (3, 17). Collectively, these observations suggest potential therapeutic avenues of targeting AR and Glo1 to treat diabetic patients.

Excessive glucose also activates arginase, which depletes the NO-precursor arginine and thereby uncouples eNOS, because whereas normally eNOS oxidizes arginine to form citrulline, in the absence of arginine eNOS instead produces superoxide anion in the presence of oxygen and NADPH (67, 126, 185). Other mechanisms that uncouple eNOS include decreased bioavailability of the cofactor tetrahydrobiopterin (BH4) (39), aberrant O-glycosylation via the HBP (54), and increased AGEs (186). As a result of NADPH exhaustion by polyol and glycation pathways, diabetic ECs display exaggerated production of ROS by NADPH, xanthine oxidases, and uncoupled eNOS, thereby assuming pro-inflammatory phenotypes with enhanced propensity to develop atherosclerosis (179). Remarkably, in addition to slowing glycolysis and diverting the glycolytic flux into side branches, and to uncoupling eNOS, hyperglycemia also leads to EC mitochondriopathy, which is characterized by aberrations in mitochondrial biogenesis and in mitophagy (the latter giving rise to damaged mitochondria accumulation), fragmented mitochondria, impaired function, and increased ROS production (190). This is partly exerted by ROS inhibiting cytochrome c oxidase in the mitochondrial electron transport chain (57). GAPDH inhibition and resultant shunting of glucose into glycolytic side branches were rescued by restricting ROS levels in mitochondria (19, 65, 130). Some studies, although not all, suggest that hyperglycemia shifts the normally glycolytic EC metabolism toward oxidative metabolism and increased mitochondrial respiration (48, 100, 130, 146).

Although these views suggest that EC mitochondrial ROS is elevated by hyperglycemia and exerts detrimental effects on EC function through the activation of PARP, blockade of GAPDH, and stalling of glycolysis, this scheme has been challenged by a concept termed “mitochondrial hormesis” (180). This concept proposes that production of mitochondrial superoxide can be an indicator of mitochondrial health and physiological oxidative phosphorylation. In response to excess hyperglycemic exposure, mitochondrial superoxide, oxidative phosphorylation, and mitochondrial ATP generation are reduced in several target tissues of diabetes complications. As a result, ROS generated from non-mitochondrial source(s) are increased, resulting in cell dysfunction. Thus the therapeutic restoration of mitochondrial superoxide [e.g., by activation of AMPK, an approach thought to induce autophagy and thereby restore bioenergetics machinery (43)] can be protective. It remains to be determined whether this concept applies to diabetic EC dysfunction. Further study is warranted to understand the exact effects of diabetes on EC mitochondrial superoxide generation and respiration.

Insulin resistance is a hallmark of T2DM and a significant condition leading to cardiovascular diseases, including atherosclerosis. The impact of insulin resistance on EC bioenergetics, however, is not sufficiently defined compared with that of hyperglycemia (179). In insulin-resistant rodent models [obese Zucker (fa/fa) rats and high-fat diet-fed mice], free fatty acid (FFA) released from adipose tissues induced profound production of ROS in ECs, which activated deleterious metabolic pathways diverted from glycolysis, similar to those seen in high glucose-exposed ECs, i.e., HBP, the glycation pathway, and PKC (46). Forced expression of PKCβ2 in the endothelium of apoE-null mice inhibited insulin-induced activation of PI-3K/Akt and eNOS (107). Similar to hyperglycemic stimulation, increased ROS by FFA seem to oxidize BH4 and provoke eNOS uncoupling. In sum, the weight of experimental evidence indicates that EC dysfunction in diabetes appears to be caused in large part by dysregulated feedback and feedforward loops between ROS/RNS and numerous metabolic pathways.

Roles of Nampt, SIRT1, FOXO1 and PGC-1α in Diabetes and Aging-Related Endothelial Dysfunction

Despite all of the preclinical evidence above, however, the precise roles of ROS in the development of vascular complications in diabetes and aging remain debated, largely due to disappointing outcomes of clinical trials with antioxidants (see below). Clearly, more integrative insights into vascular-metabolic interface are needed. It is thus noteworthy that impactful progress has been made recently in delineating the roles for Nampt, SIRT1, FOXO1, and PGC-1α in the etiology of diabetic and aging-associated vasculopathies, given that all of these nodal mediators broadly regulate EC survival, redox states, and bioenergetics (Table 1). Borradaile et al. showed that replicative aging of human ECs is associated with decreased NAD⁺-to-NADH ratio, Nampt, and SIRT1 levels (15). Importantly, Nampt overexpression increased glycolysis of high glucose-treated ECs, reduced ROS, and restored their tube-forming activities, again suggesting the relevance of glycolysis in EC angiogenic competence. The rescue effect of Nampt was dependent on SIRT1 activity and was blocked by constitutively active FOXO1 (15). The reduced NAD⁺ in high glucose-treated cells is proposed to be due to competitive consumption by PARP, which requires NAD⁺ to ADP-ribosylate substrates (39, 61, 179, 208). Several studies demonstrated SIRT1 downregulation in high glucose or H₂O₂-treated prematurely senescent ECs in culture, as well as in diabetic or atherosclerotic animal vessel ECs, and in circulating mononuclear cells from diabetic patients, associated with activation (acetylation) of p53 and FOXO1 and reduced levels of bioavailable NO (7, 30, 38, 92, 137, 138, 220, 223). Activation of EC SIRT1 with resveratrol or forced SIRT1 overexpression rescued the angiogenic capacity, and suppressed EC inflammation and atherogenesis (30, 92, 137, 138, 220, 223).

FOXO1 and FOXO3a activation (nuclear localization) also has been noted in high glucose-treated human aortic and rat brain/retinal microvascular ECs, mediating peroxynitrite formation, eNOS dysfunction, and apoptosis (9, 83, 189) (Table 1). In vivo studies further showed that FOXO1 mediates capillary EC regression in T1 and T2 DM rat retinas. The fatty acid palmitate and inflammatory cytokine TNF-α elevated EC FOXO1 in culture, and high-fat diet in mice increased FOXO1 protein in skeletal muscle capillaries by twofold. EC-specific depletion of FOXO1/3/4 prevented insulin resistance and rescued angiogenesis impairment in HFD-fed mice (132). Similarly, we recently demonstrated that PGC-1α is increased by approximately twofold in high glucose-treated ECs through ROS, in ECs from T1 and T2 DM mice, and in endothelial progenitor cells from T2DM patients (173). Forced expression of PGC-1α diminished EC migration and capillary formation through inhibition of Akt/eNOS pathway and activation of Notch signaling. EC-specific transgenic induction of PGC-1α by five- to sevenfold over the basal level (comparable to the increase in diabetic mouse ECs) in mice led to a condition mimicking diabetic endothelial dysfunction, whereas partial genetic ablation of EC PGC-1α markedly rescued wound healing and postischemic blood flow recovery in Type 1 and Type 2 diabetic mice.

Therapeutic Implications and Future Perspectives

Although traditionally regarded as a mere consequence of the state of a cell, bioenergetics (including glycolysis, mitochondrial OXPHOS, and oxidative stress) is now recognized also to dictate whether a cell proliferates, differentiates, or remains quiescent (36, 49, 50, 67). The discovery that ECs are highly glycolytic among other nonmalignant cell types and that increase in glycolytic flux promotes stalk cells to adopt a migratory tip cell phenotype has shed light to the crucial importance of bioenergetics in EC biology (37, 177). Glycolysis has thus emerged as a potential target for therapeutic angiogenesis to rescue conditions such as critical ischemia in diabetic heart and foot, where blood vessels have assumed phenotypes resistant to angiogenic signals such as VEGF (79, 108, 172). However, therapeutic induction of EC glycolysis may not be applicable to the prevention or treatment of atherosclerosis, since recent work has suggested that shear stress and Krüppel-like Factor 2 (KLF2) exert athero-protective effects in part through inhibition of glycolytic flux (42).

Molecules involved in EC fatty acid metabolism, such as VEGFB (74, 163), AMPK (128, 140), CPT1A (176), 3HIB (89), and FABP4 (52, 63, 78), may also provide novel viable targets to prevent or restore aging-associated vascular dysfunctions or to treat diabetic vasculopathies. The roles of many other metabolic pathways (for example, glutamine metabolism) also need to be investigated for deeper understanding of normal and pathological endothelial function. Importantly, since many of the findings discussed here in healthy and diseased ECs have been acquired only through in vitro/ex vivo studies, definitive in vivo validation is warranted before any clinical translation is considered. In addition, clarity on the metabolic alterations that occur in the ECs in diabetic/aged animals and patients need to be gained, for example by utilizing ¹³C metabolic flux tracer assays (20).

Although preclinical research on diabetic endothelial dysfunction has provided strong evidence supporting the central roles for ROS (19, 39, 50, 64, 65, 67, 161), the therapeutic potential of targeting PARP1 (47) and glycolysis-diverted pathways [AR (188, 200, 216), Glo1 (3, 17)], and the efficacy of using anti-oxidant agents [such as benfotiamine (11, 77)], the exact roles played by ROS in the etiology of diabetic vascular dysfunctions remain controversial. The mixed results seen in clinical trials using antioxidant therapies point to the importance of further studies and alternative approaches (80, 117). Accumulating knowledge about metabolic signaling nodes has presented unifying views that Notch, FOXOs, and PGC-1α share common capabilities to destine EC to quiescence (stalk or “phalanx” cell phenotypes), and that Nampt/NAD⁺/SIRT1 blocks these quiescence factors and triggers tip cell phenotypes to enhance and lead to angiogenesis (FIGURES 2 AND 3). Given the extensive signaling impacts of these metabolic regulators, pathological relevance (Table 1), as well as availability of compounds for intervention [selective FOXO inhibitor (93) and various SIRT1 activators (84, 85, 97), including metformin and resveratrol], this group of metabolic mediators could serve as potential therapeutic targets for diabetic angiopathies. Indeed, NAD for the treatment of diabetes and age-associated ailments is being tested in preclinical (208) and clinical settings.

From a therapeutic viewpoint, it is noteworthy that “mild” modulation of metabolism may suffice to confer significant alteration in vascular performance, without causing the toxic effects one might expect from complete blockade (such as EC death or uncontrolled hypervascularization). This is exemplified by the ability of 3PO, a PFKFB3 inhibitor, to block pathological angiogenesis [demonstrated in both non-tumor (177) and tumor (23) models] via inhibiting glycolysis by ~30% (177). Last, it is also important to understand in depth how variations in hyperglycemia-induced EC dysfunction occur across different vascular beds, as represented by so-called “diabetic paradox,” where there is a hyper-proliferation of the retinal vessels, whereas there is insufficient vascularization in ischemic limbs and myocardium (161). Although further details and the precise roles in vascular (dys)function remain to be defined, EC metabolism is an attractive avenue, with future therapeutic potential for the micro- and macrovasculopathies associated with aging and diabetes.