Insulin signaling in vascular endothelium produces at least two types of discrete actions. First, insulin modifies endothelial homeostasis in arteries, thereby making the vascular wall less susceptible to atherosclerosis [1
]. Second, insulin may regulate its own delivery to skeletal muscle and other tissues [2
]. Whether this mechanism contributes significantly to systemic insulin sensitivity is not clear in spite of extensive investigation. In this issue, Kubota et al. (2011)
report that endothelial insulin signaling through insulin receptor substrate-2 (IRS-2), a docking protein relaying insulin receptor activation to intracellular signaling, contributes to transcapillary insulin transport in muscle and affects glucose tolerance in mice. These results suggest that endothelial cell function may be a therapeutic target for improving peripheral insulin sensitivity.
The rate of insulin delivery from the blood to the interstitial space is limited by transport across the capillary wall in tissues where endothelial cells form tight junctions. After binding to its receptors, insulin can be transported across cultured endothelial cell monolayers by transcytosis [4
], but little is known about intracellular insulin signaling in this process [5
]. It is also unclear whether transcytosis [2
] rather than passive diffusion at cell junctions [3
] is responsible for transendothelial transport of insulin in vivo
, which is important because transcytosis is more likely to be a regulated process which can be modified for therapeutic gain. Insulin resistance developing during high-fat feeding of mice can be detected earlier in endothelium than in other tissues [6
], suggesting that early reversal of endothelial insulin resistance could help prevent peripheral insulin resistance if there was a cause-and-effect relationship between the two.
Kubota et al. [7
] studied mice with obesity-associated insulin resistance induced by high-fat diet or the ob/ob
mutation, and found that IRS-1 and IRS-2 in endothelial cells were downregulated by 50 and 80%, respectively. Capillary blood volume increased 10 minutes after start of a euglycemic hyperinsulinemic clamp in control mice, followed by an increase in interstitial insulin concentrations after 60 minutes. However, in obese mice the increase in capillary blood volume in skeletal muscle was blunted at 10 minutes, and interstitial insulin concentrations were decreased at 60 minutes suggesting that insulin delivery from the blood to the interstitial space was delayed by transport across the capillary wall during physiological conditions, and further delayed in obesity-associated insulin resistance. To study whether insulin activates insulin signaling to facilitate its own transendothelial transport, the authors then created and studied mice with knockout of IRS-1 or IRS-2 in endothelial cells (ETIrs1KO and ETIrs2KO mice, respectively). ETIrs1KO mice were not different than their controls with respect to endothelial cell insulin signaling, whole-body insulin sensitivity, or glucose tolerance, but ETIrs2KO mice had similar defects in increases of capillary blood volume and interstitial insulin concentrations as mice with obesity-associated insulin resistance. In addition, ETIrs2KO mice had glucose intolerance relative to their controls. If a similar mechanism exists in humans, it would be a new target for insulin-sensitizing drugs.
Insulin-stimulated increases in blood flow or increased recruitment of perfused capillaries are both dependent on nitric oxide (NO) production and may augment insulin delivery to the interstitium. Because NO has been shown to mediate insulin-stimulated capillary recruitment, Kubota et al. (2011)
treated ETIrs2KO mice with beraprost sodium, a prostacyclin analogue which increases eNOS expression. In ETIrs2KO mice, this treatment increased eNOS protein levels in endothelial cells and completely normalized capillary recruitment, interstitial insulin concentrations, and whole-body insulin sensitivity. Beraprost sodium had no effect in eNOS knockout mice, making it unlikely that this effect was mediated by effects of prostacyclin directly on skeletal muscle perfusion.
Glucose intolerance in ETIrs2KO mice was mild, similar to that described for eNOS knockout mice [8
], and much less severe than in mice fed a high-fat diet. In the study by Kubota et al. [7
], plasma glucose levels measured 30 minutes after glucose administration were approximately 180 mg/dl higher in high-fat fed mice but only 40 mg/dl higher in ETIrs2KO mice, compared with controls. This was partly explained by a maintained ability in ETIrs2KO mice to reduce endogenous glucose production from the liver in response to insulin, an effect that was profoundly reduced in mice after high-fat feeding. This abnormality of whole-body glucose metabolism contrasted with a considerable impairment of glucose uptake in skeletal muscle during a euglycemic hyperinsulimic clamp. These experiments show that insulin signaling through IRS-2 in endothelial cells is important for insulin delivery to skeletal muscle interstitium and insulin-stimulated glucose uptake in muscle, but has little effect on whole-body glucose tolerance when insulin sensitivity in the liver is normal. Whether the contribution of endothelial insulin resistance to whole-body glucose intolerance is more pronounced when insulin signaling in the liver is impaired, as in human insulin resistance, remains to be determined.
The results of Kubota et al. are somewhat at odds with a previous study in mice with the insulin receptor deleted in endothelial cells [9
]. These animals were glucose tolerant except when challenged with diets with abnormal sodium content. However, when cross-bred with atherosclerosis-susceptible apolipoprotein E (apoE) null mice, atherosclerotic lesion area in the aorta and carotid artery was increased by up to 2.9-fold [10
]. Glucose and insulin tolerance, plasma lipids, and blood pressure were not different in the two groups, and the increased atherosclerosis susceptibility could not be recapitulated by transplantation of bone marrow from insulin receptor knockout mice [10
], showing that the phenotype was due to an isolated defect in endothelial cell insulin signaling. Therefore, loss of insulin action in endothelial cells in this model had a large impact on atherosclerosis development, but no effect on glucose tolerance [10
Kubota et al. speculates that selective impairment of the phosphatidylinositol 3-kinase (PI3K) pathway of insulin signaling, which is activated by IRS proteins, has a larger effect on insulin sensitivity in muscle than deletion of all insulin signaling [9
] because activation of the PI3K and Erk pathway of insulin signaling can have opposite hemodynamic effects. In addition to this question, it is unresolved whether the blunted increase in interstitial insulin concentrations during hyperinsulinemia in ETIrs2KO mice is due to a blunted increase in capillary surface area; or a defect in a mechanism by which transendothelial insulin transport is regulated by endothelial cell insulin signaling, with impaired capillary recruitment as an associated (or contributing) factor. Future challenges include determining whether the mechanisms impairing insulin signaling in obesity and type 2 diabetes are the same in endothelial cells as in classical insulin-sensitive tissues like muscle. Discoveries in this field may provide targets for augmenting endothelial insulin sensitivity which may not only improve muscle insulin sensitivity, but also help prevent long-term complications like atherosclerosis in people with the metabolic syndrome or diabetes.