Diabetes is known to increase the risk of cardiovascular events and cause myocardial microvascular complications [60, 61]. Damage to vascular endothelial cells caused by oxidative stress is followed by a series of pathological changes that include vascular inflammation, vasoconstriction, thrombosis and atherosclerosis [62]. Diabetes and glycotoxicity promote coronary atherosclerosis by exacerbating endothelial dysfunction and increasing oxidative stress, blood lipids, and autonomic dysfunction [63].

Vascular inflammation is mediated in part by mitochondrial fission, which is a complex mechanism involving tumour necrosis factor-α (TNF-α), Drp-1, NF-κB, and vascular cell adhesion molecule-1 (VCAM-1) [64]. Primary cultures of rat aortic endothelial cells transduced with an adenovirus encoding a dominant-negative Drp1K38A mutant showed significant inhibition of TNF-α-induced NF-κB-driven promoter activity and VCAM-1 induction. These factors are responsible for chronic persistent inflammation and atherosclerosis [65]. SGLT2i were reported to improve endothelial function in mice with STZ-induced diabetes by inhibiting abnormal mitochondrial fission and endothelial inflammation [46]. The anti-inflammatory effect of empagliflozin on arterial endothelial cells has been linked to the activation of AMPK and inhibition of Drp1 by the phosphorylation of Ser-637 [66].

Macrophage infiltration and polarization towards the M1 phenotype are key events in the development of atherosclerosis [67]. Macrophages can polarize to two phenotypes: M1 and M2. M1 macrophages exacerbate the inflammatory response, and M2 macrophages are involved in the resolution of inflammation [68]. In a mouse model of diabetes with atherosclerosis and hypercholesterolemia, empagliflozin decreased the proliferation of plaque-resident macrophages and reduced the size of atherosclerotic plaques [69]. The mechanism involved reducing M1 phenotype polarization and increasing M2 polarization in response to SGLT2i [70].

SGLT2i have also been reported to delay endothelial cell senescence, and the effect may depend on angiotensin converting enzyme (ACE) activity and angiotensin type 1 receptor (AT1R). Angiotensin II induces endothelial senescence [71], and Khemais-Benkhiat et al.[72] reported that hyperglycaemia increased the protein expression of ACE and AT1R and increased β-galactosidase, a biomarker of cellular senescence[73], in porcine coronary endothelial cells. Empagliflozin reversed these changes in the presence of hyperglycaemia but did not affect ACE or AT1R in control cells in the absence of hyperglycaemia [72].

Hyperuricaemia is another independent risk factor for diabetes. Uric acid levels increase in the early stages of impaired glucose metabolism, and hyperuricaemia is associated with micro- and macrovascular complications of diabetes [74, 75]. Uric acid concentrations higher than physiological levels inhibit NO synthesis, reduce NO activity and induce NF-κB, leading to the induction of monocyte chemoattractant protein 1 and cyclooxygenase 2 (COX-2), which mediate inflammation and atherosclerosis [76]. Chino et al. [77] found that the SGLT2i luseogliflozin increased uric acid excretion and that GLUT9 isoform 2 was involved. SGLT2i increased the concentration of glucose in the proximal tubule, which caused GLUT9 isoform 2 and other transporters to reabsorb more glucose and excrete more uric acid. In the collecting ducts, a high concentration of glucose prevents GLUT9 isoform 2 from reabsorbing uric acid (Fig. 2).

SGLT2i can improve vasodilatory inflammation by decreasing uric acid levels. In proximal renal tubules, SGLT2i acts on SGLT2, decreasing the reabsorption of Na ions and glucose and increasing glucose concentrations in the lumen, which is exchanged with uric acid by GLUT9 isoform 2. This leads to increased uric acid exclusion. In the collecting duct, high levels of glucose inhibit this exchange and reduce the absorption of uric acid, thereby draining it. Lowering the concentration of uric acid in the blood helps reduce inflammation in the blood vessels

Studies have shown that flow-mediated dilation (FMD) is reduced in young T2DM patients relative to healthy individuals [78], and in coronary arteries, the maximal pharmacologic flow reserve is significantly lower in diabetes patients than in healthy individuals [79]. Dapagliflozin has been shown to improve FMD in patients with T2DM [80], which may depend on the inhibition of COX-2. The increase in ROS production may involve COX-2/prostaglandin E2 (PGE2)/E-type prostaglandin receptor 4 (EP4)/extracellular signal-regulated kinase 1/2 (ERK1/2)/NADPH oxidase isoform 4 (Nox4) signalling [81]. Increased ROS resulting in vasoconstriction involves the initiation of calcium flux and stimulating pathways leading to the sensitization of contractile elements to calcium [82]. Therefore, this kind of vasoconstriction can be inhibited by selective COX-2 inhibitors. SGLT2i inhibit COX-2 mRNA expression and vasoconstriction [83]. Vasodilation caused by selective COX-2 inhibitors was attenuated in genetically obese Zucker rats because the production of the vasodilator PGE2 promoted by COX-2 in the endothelium was attenuated [84]. COX-2 inhibition can thus improve vasodilation over a limited range (Fig. 3).

SGLT2i can improve vasodilatory functions through COX-2. In endothelial cells, SGLT2i inhibit the production of ROS-induced PGH2 by inhibiting COX-2 and reduce the production of PGE2 and TXA2 downstream of PGH2. In vascular smooth muscle, although PGE2 activates EP4 receptors to dilate blood vessels, TAX2 activates TP receptors to cause vasoconstriction. In addition, downstream EP4 can cascade into ERK1/2/NOX4 to produce ROS, which contributes to vasoconstriction and the activation of COX-2. In vascular smooth muscle, COX-2 activates and catalyses AA to produce PGH2, which constricts blood vessels

Voltage-dependent K+ (Kv) channels regulate membrane resting potential and vascular tone. The opening of smooth muscle Kv channels results in hyperpolarization and vasodilation [85]. Dapagliflozin activates Kv channels by directly activating protein kinase G (PKG) independent of guanylyl cyclase, resulting in endothelial-independent vascular smooth muscle relaxation [86]. This kind of PKG/Kv channel signalling is effective and feasible [87]. TNF-α levels are elevated in T2DM and may impair insulin signalling and lead to insulin resistance [88]. Uthman et al. [89] reported that dapagliflozin and empagliflozin inhibited TNFα-induced ROS generation in human coronary arterial endothelial cells, and the subsequent decrease in NO consumption was responsible for improved blood vessel dilatation. An increase in L-arginine synthesis may indirectly improve coronary flow reserves by increasing NO synthesis. L-arginine is a substrate for nitric oxide synthase (NOS), which converts arginine to NO. An increase in L-arginine synthesis in the kidneys and an increase in NO bioavailability in response to SGLT2i were reported to increase coronary flow velocity reserve in an ob/ob − / − mouse model [90]. SGLT2i may act directly on cardiomyocyte NHE1 to reduce cytosolic Na+. Empagliflozin and canagliflozin significantly decreased coronary perfusion pressure in isolated C57 mouse hearts under constant-flow conditions, which was consistent with the dilation of coronary vessels [91].

The microvascular complications of diabetes include a reduction in the density of arterioles on the surface of the heart and have also been described in animal models of diabetes [92, 93].

Some SGLT2i promote angiogenesis. For example, empagliflozin significantly reduced the loss of CD31 + microvessels and decreased the size of defects in zones of perfusion in diabetes model mice [46]. In the model, empagliflozin activated AMPK via an increased AMP/ATP ratio that then led to the failure of Drp1 recruitment to the mitochondria and weakened mitochondrial fission. The resulting reduction in ROS production alleviated cell senescence and decreased F-actin dissolution into G-actin, which contributed to cardiac microvascular endothelial cell migration and neovascularization.

Unfortunately, some SGLT2i inhibit angiogenesis. For example, canagliflozin inhibited the in vitro proliferation of human umbilical vein endothelial cells and the formation of blood vessels in allograft liver tumours [11]. Canagliflozin also inhibited angiogenesis in the lower limbs of mice with diabetes and lower limb ischaemia by inhibiting the secretion of vascular endothelial growth factor A by bone marrow-derived mesenchymal stem cells and reducing the proliferation and migration of mesenchymal stem cells [12].