Gasotransmitters in Vascular Complications of Diabetes

Diabetes is characterized by hyperglycemia and insulin resistance or deficiency. Diabetes is a top 10 cause of death worldwide; its prevalence is increasing and currently estimated to be 9% among adults (1,2). Both type 1 and type 2 diabetes are important risk factors for vascular diseases, with a two- to fourfold increased risk when compared with individuals without diabetes (3). These vascular complications are divided into microvascular (retinopathy, neuropathy, and nephropathy) and macrovascular (cerebrovascular, coronary artery, and peripheral arterial diseases) complications with respective clinical symptoms (Fig. 1). Although the pathophysiology of type 1 and type 2 diabetes is different, the proposed underlying mechanism leading to vascular complications seems to be similar and is thought to be related to endothelial dysfunction (4,5) and the associated formation of reactive oxygen species (ROS). Chronic hyperglycemia promotes multiple biochemical pathways to overproduce ROS, either through mitochondrial overproduction or through enzymatic responses to high glucose (6).

Diabetic retinopathy is the main cause of blindness in adults. The worldwide prevalence is approximately 35% in patients with diabetes (7). The essentials of diabetic retinopathy can be best characterized by the combination of increased vessel permeability and progressive vascular occlusion. Although the clinical diagnosis of retinopathy is still made by the changes in small (early) and larger (later) vessels, it has become clear that almost every cell type in the retina can be subject to damage by complex metabolic changes, induced by chronic hyperglycemia (8–11).

Polyneuropathy is defined as a diffuse and bilateral disturbance of functions or pathological changes in multiple peripheral nerves. Diabetic peripheral polyneuropathy is very frequent in the course of diabetes and even in prediabetes, affecting up to 50% of all patients with diabetes (12–14). However, although being frequent and severe, it is inadequately treated in most patients—77% of those with chronic painful peripheral neuropathy report persistent pain over 5 years (15). Experimental studies suggest the importance of neurovascular vasodilation in diabetic neuropathy (16); however, the mechanisms remain poorly understood, which may explain the current lack of adequate treatment in (diabetic) neuropathic pain.

Diabetic nephropathy (DN) is one of the leading causes of end-stage renal disease in the Western world, occurring in ∼30% of patients with type 1 and type 2 diabetes and accounting for about 40% of new cases of end-stage renal disease based on U.S. data (17). At the structural level, the glomeruli are often affected as evidenced by basement membrane thickening, mesangial lesions (Kimmelstiel-Wilson lesions), and nodular sclerosis. Clinically, DN is accompanied by proteinuria and chronic renal failure. In addition, arterioles are often affected. The mechanisms leading to renal changes include the metabolic defect, nonenzymatic glycation of proteins, and hemodynamic changes, such as hypertension leading to glomerular hypertrophy (18).

Macrovascular complications are characterized by the development of atherosclerosis in arteries throughout the body. Atherosclerosis results from a proinflammatory state starting with endothelial dysfunction and culminating in the narrowing of the arterial lumen as a result of atherosclerotic plaque formation. As opposed to stable plaques, vulnerable, nonstable plaques are prone to rupture, causing downstream ischemic events such as transient ischemic attack and stroke (Fig. 1).

Nitric oxide (NO) was first recognized as an endothelium-derived relaxing factor (19). It is endogenously formed from its substrate l-arginine by three different nitric oxide synthase (NOS) enzymes. Endothelial NOS (eNOS) is predominantly associated with vascular tone. Inducible NOS (iNOS), although also present in the vascular system, is mainly active in the immune system under conditions of oxidative stress. It functions as a promoter of inflammation. Neuronal NOS (nNOS), present in neurons and skeletal muscle cells, is important for neuronal cell-cell interactions (20). NO acts as a vasodilator and inhibits platelet aggregation and stabilizes atherosclerotic plaques (21). In humans, NO-dependent vasodilatation is impaired in patients with type 2 diabetes, and lower eNOS expression and reduced NO production are the suggested underlying cause (22,23). Blockade of NOS causes insulin resistance in a rat model, indicating that in this model loss of NO synthesis precedes type 2 diabetes (24). In different animal models for diabetes, lower bioavailability of NO is observed. Reduced NO production was found in spontaneous type 1 diabetic BioBreeding rats (25) as well as streptozotocin (STZ)-induced type 1 diabetes in male Sprague-Dawley rats (26). In mouse models of diet-induced obesity and type 2 diabetes, NO bioavailability is reduced, leading to endothelial dysfunction and impaired NO-mediated vasodilatation (27,28). In contrast to these protective effects of NO, iNOS-produced NO seems to play an important role in inducing nitrosative stress and inflammation, also in the course of diabetes. Thus, NO seems to play a dual role in the development and progression of diabetes as well as in the development of vascular dysfunction (29).

Effects of NO depletion and supplementation on the development of microvascular complications have been primarily studied in experimental models as summarized in Table 1 and discussed below.

In the retina, iNOS is sensitive to hyperglycemia and responsible for overproduction of NO (30,31). The resulting surplus of NO is either quenched by advanced glycation end products (AGEs) or leads, through the reaction with superoxide, to the formation of peroxynitrite with subsequent nitrosylation of proteins, lipids, and DNA. NO production is important in inflammatory signaling, and inflammation is thought to be important in incipient diabetic retinopathy (32). Increased reactive nitrogen species (RNS) has been observed in diabetic rat retinae and in vitro, and these changes were corrigible by aminoguanidine, an inhibitor of NO synthases (33). As an inhibitor of AGEs, aminoguanidine reduces vascular cell damage in several animal models (33,34). Zheng et al. (30) found that nitrosative stress was reduced in the retinae of iNOS^−/− mice, together with an inhibition of vasoregression and retinal thinning. However, the essential role of iNOS for the development of diabetic retinopathy seems not to be the case for other NOS isoforms, as deletion of eNOS exacerbates diabetic retinopathy (35). In STZ-induced type 1 diabetes, eNOS^−/− mice developed more severe retinopathy compared with wild-type diabetic control mice. The worsened phenotype in these eNOS^−/− mice was accompanied by increased iNOS expression, further suggesting an important role for iNOS in the development of diabetic retinopathy. However, eNOS^−/− mice suffer from higher blood pressure, so the worsened retinal phenotype can partly be explained by hypertensive injury. In essence, NO appears to have a dual role (i.e., protective and noxious effects) in the diabetic retina, as schematically shown in Fig. 2A.

Until now, NO was the best-characterized gasotransmitter contributing to nociception and pain. Its downstream targets within the peripheral nervous system (PNS) include cyclic guanosine monophosphate (cGMP) production by activation of soluble guanylyl cyclase (sGC) and phosphorylation of membrane receptors and channels by cGMP-dependent protein kinases (36)—mechanisms usually associated with increased nociception. Consistently, different members of the large family of transient receptor potential channels, several of which are known as nociceptive sensor molecules such as TRPV1 and TRPA1, are activated by NO via cysteine S-nitrosylation (37). In contrast, several mechanisms were identified that may induce antinociception and analgesia and increase efficacy of analgesic compounds. In the central nervous system, NO interacts with the descending inhibitory control mechanisms of nociception (38). In patients with type 2 diabetes suffering from painful diabetic neuropathy, treatment with the NO donors glyceryl trinitrate (39) and isosorbide dinitrate (40) significantly improved pain symptoms, indicative of the beneficial action of NO in diabetic neuropathy. Similar effects were observed when locally applying a NO-releasing cutaneous patch (41). These effects may, however, also be indirectly explained by variations in local microcirculation: transient changes in sciatic nerve microcirculation were observed in response to NO in animals with STZ-induced diabetes developing diabetic neuropathy (42). In STZ-induced diabetic rats, NOS activity is increased in primary sensory neurons (43). The potent oxidant peroxynitrite, a product of a superoxide anion radical reaction with NO, was suggested to play a role in the induction of peripheral diabetic neuropathy and neuropathic pain via induction of RNS (44), including protein nitrosylation, lipid peroxidation, DNA damage, and cell death (29). Hyperglycemia activates iNOS and therefore generally increased nitrosative stress in the PNS (44,45). Absence of iNOS reduced nitrosative stress in peripheral nerve fibers displaying normal nerve conduction velocities; diabetic neuropathy was also less severe in diabetic iNOS^−/− mice than in diabetic wild-type mice. Thus, diabetic neuropathy depends on nitrosative stress induced in axons and Schwann cells by NO produced from iNOS. In contrast, nNOS is required for maintaining PNS function and nerve fiber density and contributes to a lesser extent to the development of diabetic polyneuropathy (45). In summary, NO may play pivotal direct and indirect roles in the progression of diabetic neuropathy, presumably by impairing microcirculation in PNS at pathophysiological levels and contributing to oxidative stress and inflammation and tissue injury (29), as schematically shown in Fig. 2B.

There is still controversy regarding whether the generation of NO is enhanced or decreased in DN. In the early stages of DN, Chiarelli et al. (46) found significantly higher concentrations of NO end products (nitrite/nitrate) in the serum of DN patients with microalbuminuria compared with the serum of healthy individuals. However, an association as such does not imply causality per se. This excess of NO can indicate an upregulated inflammatory response by iNOS or a (protective) compensatory response against renal injury, mediated by eNOS. In experimental STZ-induced type 1 diabetes, renal NO production is decreased in the early phase of the disease (47). Deficiency of eNOS results in accelerated nephropathy in diabetic mice (48,49), also supporting a protective role for NO in DN (50). Supplementation of tetrahydrobiopterin (BH₄), a cofactor of NOS, reduced proteinuria and renal damage in type 2 diabetic rats (51). Taken together, NO production is clearly modulated in DN, and decrements in its expression point to a contributing role for this gasotransmitter in DN. Scavenging of ROS positively influences the redox status and may mechanistically underlie these findings. The modes of action of NO in the development of DN are schematically shown in Fig. 2C.

As discussed above, NO demonstrates both protective and damaging properties in the development of microvascular disease. The producing enzyme seems to play a major role in the contrasting actions of NO: eNOS- and nNOS-derived NO exerts the vast majority of their positive effects via upregulation of the production of cGMP by activation of sGC. However, iNOS-produced NO is involved in inflammatory signaling and is an important contributor to the development of diabetic angiopathy. In addition, the presence of ROS is important in the actions of NO. An excess of NO in the presence of abundant ROS (superoxide) production leads to the formation of peroxynitrite with subsequent nitrosylation of proteins, lipids, and DNA. There is also increasing evidence for harmful effects of NO in protein tyrosine nitration (52). Protein nitration is a posttranslational modification that takes place in the combined presence of oxidative stress and NO, which is the case in disease conditions such as diabetes (53). Given the data available, we conclude that NO plays a dual role in the progression and maintenance of diabetic microvascular complications, which is mostly driven by the expression of its producing enzymes (NOS) and the presence of ROS.

Carbon monoxide (CO), the second gasotransmitter, is produced by the different heme oxygenase (HO) enzymes as a product of heme metabolism (54). Heme is converted to biliverdin, iron, and CO by HO. Three different isoforms of HO exist; the inducible form, HO-1, and the constitutive isoforms, HO-2 and HO-3. HO-1 and HO-2 are physiologically active, whereas the role of HO-3 in human physiology remains unclear (55). CO has numerous physiological functions, including vasodilation and inhibition of platelet aggregation. In skeletal muscle biopsies and circulating leukocytes from patients with type 2 diabetes, mRNA expression of HO-1 was dramatically decreased compared with age-matched control subjects without diabetes (56,57). In STZ-induced type 1 diabetic rats, a decreased vasorelaxant function of CO was demonstrated, despite higher HO-1 expression levels (58). In Zucker diabetic fatty (ZDF) rats, CO production was decreased in aortic tissue compared with that in nondiabetic controls. Increasing HO-1 activity with cobalt protoporphyrin resulted in higher levels of CO, lower glucose levels, and increased insulin sensitivity (59). These data are in favor of reduced vascular risk in the presence of higher CO levels, which might be mediated via effects on insulin sensitivity (60). Taken together, reduced bioavailability of CO in the diabetic state is accompanied by insulin resistance and a reduction of endothelial health, indicating a potential role for the HO-1/CO pathway in the development of diabetes and its associated complications. The effects of CO depletion and supplementation in diabetic mice and rats on the development of microvascular complications are summarized in Table 2 and will be discussed below.

Oxidative stress in the diabetic retina promotes the activation of HOs (61). In the diabetic retina, HO-1 is predominantly found in glial cells, in particular in Müller cells, and to some extent in the microvasculature (62). In vitro, HO-1 overexpression protects retinal endothelial cells from high glucose and oxidative/nitrosative stress (63). In a STZ-induced type 1 diabetes model in rats, HO-1 upregulation by hemin resulted in protection against the development of diabetic retinopathy (64). This protection is reflected by the downregulation of p53, vascular endothelial growth factor (VEGF), and HIF-1α and a reduction of diabetes-induced apoptosis in retinal ganglion cells (RGCs). On the contrary, HO-1–derived CO is proangiogenic (65), and angiogenesis, causing increased retinal blood flow, is a key factor in the development of diabetic retinopathy (66). This implies that the proangiogenic effects of CO may actually aggravate diabetic retinopathy. The effects of CO in the diabetic retina are schematically shown in Fig. 2A.

In the case of diabetic neuropathy, CO acts as a pain-modulating second messenger within the nervous system (67). The activation of HO/CO signaling reduced symptoms of neuropathic pain, presumably by the activation of anti-inflammatory and antioxidant mechanisms (68). CO exerts antinociceptive effects and increases the anti-allodynic and antihyperalgesic efficacy of morphine in chronic inflammation and neuropathic pain (69)—the latter strictly dependant on NO produced by nNOS and iNOS. Furthermore, CO relieves neuropathic pain symptoms by reducing the expression of iNOS and nNOS as well as by reducing the activation of spinal microglia (70). Interestingly, the constitutive isoform HO-2 is coexpressed with NOS in the PNS and central nervous system (67), and CO, similar to NO, is also capable of activating the proalgesic cGMP protein kinase pathway (70). In fact, there is a close interaction between the CO and NO systems in the course of neuropathic pain, suggesting that they might act as cotransmitters in neuronal signaling transmission (67). In nociception, the more stable CO may set basal activity by tonic background stimulation and NO may transiently amplify nociceptive signaling. Substances increasing endogenous CO (e.g., CO-releasing molecules [CORMs] or HO inducers alone or in combination with analgesics) may be useful for the treatment of (diabetic) neuropathic pain. The effects of CO in diabetic neuropathy are depicted in Fig. 2B.

In the kidney, HO-1 and HO-2 are important in cytoprotection and serve as physiologic regulators of heme-dependent protein synthesis during which CO is produced. Inducers of the HO pathway (like hemin) are protective against renal inflammation and ameliorate DN in type 2 diabetic ZDF rats and STZ-induced type 1 diabetic rats (71–73). The antioxidant effect of HO-1 is believed to play a role in renal protection in diabetic rats (74). The opposite, deficiency of HO-2, results in higher superoxide anion levels and increased renal dysfunction after STZ-induced diabetes (75). Enhanced production of CO seems to be beneficial for the kidney in DN, suggesting possibilities for therapeutic intervention. The effects of CO in DN are shown in Fig. 2C.

Besides the beneficial effects of CO, high CO concentrations are toxic because of the high affinity of CO to bind heme proteins such as hemoglobin. Due to the high levels of CO bound to hemoglobin (forming carboxyhemoglobin), oxygen is not able to bind to that particular hemoglobin molecule, and disrupted oxygen transport develops. However, endogenous production of CO by HO enzymes obviously does not result in toxic levels. In contrast to NO, the exact working mechanism and molecular targets of CO are mostly unknown. Nevertheless, one of the known pathways is that CO is able to increase cGMP production by activation of sGC, albeit with lower affinity than NO. Moreover, CO is able to bind to complex IV (cytochrome c oxidase) of the mitochondrial electron transport chain and thereby regulate ROS production. In summary, CO is mainly protective in diabetic vascular disease via inhibition of ROS formation, via interaction with NO, and via the sGC/cGMP pathway.

Hydrogen sulfide (H₂S) is the third gasotransmitter and was recognized as such in the 1990s (76,77). It is endogenously produced by three different enzymes. The pyridoxal-5′-phosphate (PLP)–dependent enzymes cystathionine β-synthase (CBS) and cystathionine γ-lyase (CSE) are the two major H₂S-producing enzymes. The third H₂S-producing enzyme is 3-mercaptopyruvate sulfurtransferase (3MST). The main substrates for H₂S production are homocysteine and cysteine. 3MST produces H₂S from 3-mercaptopyruvate, which is produced by the enzymes cysteine aminotransferase and d-amino acid oxidase from l-cysteine and d-cysteine, respectively (78). H₂S is a physiologically active compound and is called endothelium-derived hyperpolarizing factor (79,80); it causes vasodilatation but also acts as scavenger of ROS and stimulating angiogenesis (81). Renal CSE and CBS expression and H₂S production are markedly lowered in spontaneous diabetic Ins2^Akita mice (82). In nonobese diabetic mice, another mouse model of type 1 diabetes, it was also shown that diabetic mice had lower H₂S levels compared with nondiabetic mice (83). In STZ-induced type 1 diabetic rats, H₂S levels were lower compared with age-matched nondiabetic rats (84). However, CSE deficiency delayed the onset of STZ-induced type 1 diabetes, and diabetes was accompanied by increased pancreatic H₂S production without changes in pancreatic CSE of CBS protein expression (85). Another player in the development of vascular complications is insulin resistance. Insulin resistance is affected by H₂S in a mouse model with high-fat diet–induced obesity. Interestingly, both the inhibition of H₂S production by dl-propargylglycine (PPG) and the treatment with slow-release H₂S donor GYY4137 improved insulin resistance in these mice (86). This unexpected beneficial effect of PPG could be explained by an upregulation of HO-1 resulting in higher CO levels, an effect of PPG that was recently described (87). In addition, this contradiction could be explained by the fact that PPG is an unspecific CSE inhibitor (based on its cofactor PLP), thereby potentially inhibiting other PLP-dependent enzymes as well (88). In humans, diabetes is associated with lower levels of H₂S. In a small group of patients with type 2 diabetes, plasma H₂S levels were reduced by 73% compared with those in healthy (lean) individuals (89). Interestingly, obesity is correlated with lower levels of H₂S compared with those in lean volunteers. Collectively, human and experimental diabetes are associated with reduced H₂S bioavailability, which might be related to increased cardiovascular risk as observed in subjects with diabetes. The effects of H₂S depletion and supplementation in diabetic mice and rats on the development of microvascular complications are summarized in Table 3 and will be discussed below.

H₂S has recently received attention in research on diabetic retinopathy as some H₂S-related changes are compatible with a significant role of H₂S in the development and propagation of diabetic retinopathy. Reduced H₂S-mediated cell protection supposedly plays a role in retinal diseases as CBS expression is found in various eye compartments, including the retina, suggesting that the trans-sulfuration pathway is present in the eye (90). Many CBS deficiency–related eye disorders are associated with increased homocysteine levels, and the retinae of CBS^+/− mice are characterized by RGC loss, which is mediated by mitochondrial dysfunction (91). It is thus conceivable that H₂S is neuroprotective and there is indeed experimental proof for protective properties of H₂S in the retina, as evidenced by a decreased RGC loss in H₂S-pretreated animals after retinal ischemia/reperfusion (I/R) (92). Si et al. (93) investigated the effect of H₂S in experimental retinopathy of STZ-induced type 1 diabetic rats. They reported beneficial effects on neuronal dysfunction (based on electroretinography) and retinal structure (i.e., inhibition of diabetes-induced retinal thickening and extracellular matrix proteins), while others clearly showed a link to improved endothelial function, such as tightened blood-retinal barrier and reduced vasoregression. H₂S is a known proangiogenic signaling molecule and can thereby also contribute to enhanced angiogenesis in the diabetic retina. In line with this, increased levels of H₂S were observed in vitreous body of patients with proliferative diabetic retinopathy compared with patients with rhegmatogenous retinal detachment (94). The effects of H₂S in the diabetic retina are schematically shown in Fig. 2A.

H₂S has mainly been reported to increase pain sensitivity via several proposed modes of action (95). These include sensitization of voltage-gated sodium and calcium channels (95–97) and/or suppression of potassium channels. Furthermore, the pronociceptive transient receptor potential channels TRPV1 and TRPA1 (98,99), as well as NMDA receptors, were suggested to be sensitized by H₂S. H₂S displayed pronociceptive actions in inflammatory pain, both in STZ-induced type 1 diabetes and nondiabetic control animals. Interestingly, when treated with antagonists of the H₂S-producing enzymes, pain reduction was much more pronounced in diabetic animals than it was in nondiabetic animals, indicative of an increased H₂S sensitivity of the nociceptive system in rats suffering from diabetes (100). Conversely, reduction of H₂S reduced the tactile allodynia developed in course of diabetes. The T-type voltage-gated calcium channel Ca_V3.2 is sensitized by H₂S, leading to increased pain sensitivity (101). An increased H₂S tissue content and hyperactivity of Ca_V3.2 were observed in chemotherapy-induced neuropathic hyperalgesia and pain that could be reversed by blocking H₂S production (96). Painful peripheral diabetic neuropathy is accompanied by an enhancement of Ca_V3.2 T-type calcium channels and neuronal excitability. Thus, Ca_V3.2 T-channels are thought to represent signal amplifiers in peripheral sensory neurons, contributing to hyperexcitability that ultimately leads to the development of pain in diabetic neuropathy (102). Conversely, antinociceptive effects also have been reported for H₂S. These effects were antagonized by the ATP-sensitive potassium (K_ATP) channel blocker glibenclamide and by NOS inhibition (103). Inhalation of H₂S reduced the development of neuropathic pain by reducing the resulting increase in interleukin-6 and chemokines, which was attributed to an inhibition of microglia activation in course of neuropathy (104). Furthermore, H₂S functions as a neuroprotective agent by enhancing the production of glutathione, a major intracellular antioxidant that scavenges mitochondrial ROS (105). Collectively, these data indicate that H₂S displays both pro- and antinociceptive actions in diabetic neuropathy. The effects of H₂S in diabetic neuropathy are depicted in Fig. 2B.

In DN patients with atherosclerosis who are on dialysis, lower plasma levels of H₂S were measured, which could indicate a loss of the supposed protective effects of H₂S in these patients (106). This might be caused by endothelial damage or downregulation via other pathways of the enzymes producing H₂S. Also, high urinary sulfate, as a proxy for H₂S, is significantly associated with a slower decline in glomerular filtration rate in patients with type 1 diabetes and DN (107). In the experimental setting, exogenous H₂S reduces blood pressure and prevents the progression of DN in spontaneously hypertensive rats (108). Renal protection via blood pressure reduction is also shown in angiotensin II–induced hypertension and proteinuria in rats (109). Other studies also suggest that H₂S is a key modulator in renal remodeling and that its actions can be affected by the matrix metalloproteinase 9, which is shown to modulate CBS and CSE (82). In STZ-induced type 1 diabetic mice, the administration of H₂S attenuated oxidative stress and inflammation, reduced mesangial cell proliferation, and inhibited the renin-angiotensin-aldosterone system (110). These data indicate that in DN H₂S has predominantly beneficial effects and is therefore a promising target for intervention. Thiosulfate might be the perfect H₂S donor as it is already in use in the clinic for patients with calcifylaxis with end-stage renal disease (111) and has been shown to be beneficial in hypertensive renal disease in rats (109). The effects of H₂S on the kidney in DN are schematically shown in Fig. 2C.

Knowledge on the working mechanisms of H₂S is continuously increasing. H₂S-regulated vasodilatation acts partly via activation of K_ATP channels, and a rather new hypothesis is the interference of H₂S with the cGMP pathway by inhibition of phosphodiesterase type 5 activity, a mode of action comparable with “natural sildenafil” (112). ROS production is decreased by H₂S through direct interference with the mitochondrial respiration chain. It binds to cytochrome c oxidase, thereby directly inhibiting the formation of ROS. Another important effect of H₂S in terms of diabetic angiopathy is angiogenesis. The VEGF receptor 2 is the natural target for H₂S to achieve its proangiogenic effect (113). In diabetic retinopathy, the development of new vessels reflects the severity. However, increased angiogenesis might also have some protective effects, e.g., angiogenesis of vaso nervorum in diabetic neuropathy. The effects of H₂S on different ion channels are mainly important in diabetic neuropathy. Its interference with, for instance, TRPV1, TRPA1, and Ca_V3.2 channels contributes to increased nociception. Taken together, H₂S exerts dual effects in diabetic angiopathy, positive effects via its vasodilatory actions, and unwanted detrimental effects via different ion channels and angiogenesis.

Gasotransmitters have been studied in diabetes-associated macrovascular disease and therapeutically used in clinical practice (NO only, CO and H₂S have not yet been used). The effects of gasotransmitters depletion and supplementation in human and experimental diabetes on the development of endothelial function and macrovascular disease are summarized in Table 4 and briefly discussed below. In 1992, NO donor sodium nitroprusside (SNP) was used to measure NO-dependent vasorelaxation in patients with type 1 diabetes. SNP-stimulated vasodilatation was decreased in patients without diabetes compared with subjects with diabetes, indicating a lower NO sensitivity (114). In patients with type 2 diabetes, the addition of NOS cofactor BH₄ resulted in improved forearm blood flow, an effect that was nullified by NOS-inhibitor N^G-monomethyl-l-arginine (l-NMMA) (115). In patients with type 2 diabetes and coronary artery disease, treatment with NO substrate l-arginine and NOS cofactor BH₄ protected against I/R endothelial dysfunction in the forearm vasculature (116). The important role of NO in the macrovasculature is also shown in animal models of diabetes. Lepr^db/db eNOS^−/−double knockout mice showed an aggravated vascular phenotype compared with diabetic Lepr^db/db or eNOS^−/−single knockouts, as evidenced by an increased aortic wall thickness and reduced re-endothelialization after arterial injury (117). In ApoE^−/− mice and mice with STZ-induced type 1 diabetes, treatment with bone marrow–derived mononuclear cells overexpressing eNOS resulted in reduced plaque progression and improved postischemic neovascularization, an effect that was completely inhibited by NOS inhibitor l-N^G-nitro-l-arginine methyl ester (l-NAME) (118).

Protective properties of CO in diabetes have been mainly investigated in STZ-induced type 1 diabetes in rats or mice. Exposure of the tail artery to CO ex vivo resulted in vasodilatation, an effect that was reduced in arteries of STZ-induced diabetic rats, indicating a reduced sensitivity for CO in diabetic animals (58). In a myocardial I/R model in HO-1^−/− diabetic mice, infarct size and mortality were dramatically worsened compared with wild-type (HO-1^+/+) diabetic mice, without affecting glucose levels (119). In diabetic rats, CORM-3 or HO-1 inducer cobalt protoporphyrin (CoPP) preserved endothelial function and vascular relaxation, an effect that was reversed by HO inhibitors chromium mesoporphyrin (CrMP) and tin mesoporphyrin (SnMP) (120–122). In a model of myocardial I/R injury, treatment with CO-releasing compound PEGylated carboxyhemoglobin bovine (PEG-COOH) drastically reduced infarct size and troponin levels in STZ-induced diabetic mice. In mice receiving PEG-COOH during reperfusion only, infarct size was reduced, suggesting CO as a potential therapeutic agent for patients after myocardial infarction (123). In STZ-induced diabetes in rats, induction of HO-1 with hemin, and treatment with CORM-2 to lesser extent, attenuated vascular damage and oxidative stress and improved vascular relaxation compared with nontreated rats (124).

H₂S as a therapeutic agent in diabetic vascular disease is evaluated in both mice and rats. In STZ-induced diabetic Sprague-Dawley rats, treatment with H₂S donor sodium hydrosulfide (NaHS) improved vascular relaxation and NO bioavailability. This indicates that H₂S is a potential therapeutic agent in diabetic vascular disease via cross talk with NO (26). Ex vivo administration of H₂S substrate l-cysteine also resulted in dose-dependent vasorelaxation in rat middle cerebral arteries, which was reduced in diabetic animals (125). The vasorelaxant effects of NaHS are reduced with the addition of K_ATP blocker glibenclamide, demonstrating that NaHS-induced vasorelaxation takes place via activation of K_ATP channels (126). Ex vivo overexpression of CSE improved vascular relaxation in hyperglycemic conditions and reduced ROS production, while CSE mRNA knockdown with small interfering RNA resulted in a more pronounced ROS production (127). Beneficial properties of H₂S have been shown in models for myocardial injury as well. The addition of NaHS in diabetic rats resulted in preserved cardiac function (128), reduced infarct size, reduced ROS and inflammatory parameters such as tumor necrosis factor-α and interleukin-10, and inhibited expression of adhesion molecules such as intracellular adhesion molecule 1 (129). In a model of myocardial I/R injury in diabetic Lepr^db/db mice, treatment with H₂S donor sodium sulfide (Na₂S), either before I/R or only during reperfusion, diminished infarct size, troponin levels, and ROS (130,131). Although studies on the role of gasotransmitters in the development of macrovascular disease are limited to endothelial dysfunction rather than atherosclerosis, we propose that gasotransmitters may also modulate atherogenesis via different mechanisms as schematically depicted in Fig. 3. Considering the data described above and summarized in Table 4, gasotransmitters seem to be promising targets for intervention in the course of diabetic macrovascular disease.

Although NO, CO, and H₂S have different molecular structures and routes of endogenous production, they do share various physiological properties such as the ability to bind to heme groups (132) and to promote vasorelaxation by stimulation the sGC/cGMP pathway. NO and CO stimulate cGMP production by targeting sGC, and H₂S by inhibiting phosphodiesterase type 5 activity (133). NO as well as CO and H₂S are direct ROS scavengers, partly by direct interaction with the mitochondrial respiratory chain. They all engage on the K_ATP channels to achieve this antioxidant effect, both in the vasculature and nervous system (134–136). In addition, they share several common intracellular pathways, such as nuclear factor-κB, nuclear factor-like 2, and mitogen-activated protein kinases, thereby exerting antiapoptotic, antioxidant, and anti-inflammatory effects (133). NO, CO, and H₂S inhibit the expression of intracellular adhesion molecule 1, vascular cell adhesion molecule 1, and E-selectin, thereby promoting endothelial health and integrity. Finally, all three gasotransmitters act as proangiogenic substances via the VEGF pathway (113,137). On the basis of these functional similarities, it is likely that gasotransmitters have mutual interactions. Such a relationship between CO and NO has been investigated intensively and is mainly mediated via the sGC/cGMP pathway. These effects include blood pressure regulation and inflammation (138). Cytoprotective effects of NO donors in endothelial cells were abolished in the presence of the HO inhibitor tin protoporphyrin (139). It has already been shown that H₂S exerts its effects via NO production, as H₂S promotes eNOS production and activity (140). Additionally, vasorelaxant effects of H₂S were diminished when aortic rings were pretreated with NOS inhibitor l-NAME (141). Reciprocally, NO donor SNP stimulates endogenous H₂S production via upregulation of CBS (142). Although CO and H₂S share a lot of functional characteristics, the mutual relationship between these two gasotransmitters has barely been studied. In one study of a myocardial I/R injury mouse model, HO-1 expression was upregulated 24 h after intravenous H₂S treatment, which was accompanied by a protection against I/R-induced damage (143).

In order to study the association of gasotransmitters and the development of diabetes-associated vascular complications, reliable and sensitive assays to measure NO, CO, and H₂S are indispensable. Various methodologies are used to measure the different gasotransmitters and these will be briefly described. First, measuring NO is quite a challenge because of its instability. There are different methods of measuring NO. Most commonly used is the relatively simple Griess method, which does not measure NO directly but rather its oxidated products nitrite and nitrate. However, nitrite and nitrate can be detected more precise by high-performance liquid chromatography (144). NO can also be directly measured using gas phase chemiluminescence, which involves the reaction of NO with ozone (O₃) to form excited nitrogen dioxide. During relaxation to (unexcited) nitrogen dioxide, a photon is released that is then detected by chemiluminescence. Using this method, NO release from different body fluids and tissues can be measured. In addition to the aforementioned methods, different fluorescent probes and electrodes are currently available, with the possibility to measure NO in fluids and tissues and intracellularly in cells in vitro (145).

CO levels also can be measured using different methods. The most commonly used and relatively simple way to measure CO is in the air using gas chromatography. In vivo, CO is generally measured in red blood cells by determining the percentage of carboxyhemoglobin relative to total hemoglobin concentration. Finally, some studies use [¹⁴C]Heme in vitro to measure endogenous ¹⁴CO production (146).

When considering H₂S as therapeutic target, reliable methods to determine H₂S levels in body fluids and tissues are needed. However, measurements of H₂S are difficult because of its volatility. For that reason, stable end products like sulfate or thiosulfate can be measured in serum or urine (147), although a few methodologies have been described to measure H₂S itself. The methylene blue assay is the most commonly used technique. It is based on the oxidative coupling of H₂S with two N,N-dimethyl-p-phenylenediamine molecules, forming the methylene blue dye that can be detected spectrophotometrically. However, this technique is extremely pH dependent and not very sensitive and reliable. A more sensitive method is based on monobromobimane in which two monobromobimane molecules form the stable sulfide dibimane in the presence of H₂S. Sulfide dibimane can be separated by reverse-phase chromatography and detected by a fluorescence detector. Now, fluorescent probes and sulfide selective electrodes are extensively used however, with different sensitivity and reliability (148). As yet, H₂S measurements are difficult, with variable reliability, thereby complicating studies on the role of H₂S and its use as therapeutic target in various diseases, including diabetes-associated vascular disease.

Patients with diabetes have a two- to fourfold increased risk for cardiovascular disease, and adequate treatment and preventive strategies are still lacking. As discussed in this Perspective, the different gasotransmitters appear to be important mediators in the development of diabetic angiopathy and therefore are potential targets for intervention. As aforementioned, NO-based interventions are already applied in humans and readily available. SNP is clinically used and acts as a direct NO donor by releasing NO from the ferrous ion center without the need for enzymatic action. The same is true for nitroglycerin and other organic nitrates, which are well established for their vasodilatory effects during angina. Organic nitrates act as NO donors by enzymatic or nonenzymatic breakdown of nitrates into nitrite and NO (149). Molsidomine and Linsidomine are registered in several European countries as antianginal drugs and act as vasodilators by the nonenzymatic release of NO. Finally, dietary products with high nitrate content can act as NO donors. The intake of beetroot juice lowered blood pressure significantly, an effect that was accompanied by higher levels of total urinary nitrite/nitrate (150). CO administration or CORMs are not in clinical use yet, albeit that some of the vascular protective effects of acetylsalicylic acid and statins are attributed to induction of HO-1. In human endothelial cells in vitro, a dose-dependent increase of HO-1 expression was seen after statin (151,152) or acetylsalicylic acid (153) treatment. However, this effect was not reproduced in human subjects as no differences in HO-1 expression were observed between patients treated with acetylsalicylic acid, statin, or placebo (154). The antioxidative actions of polyphenol resveratrol are also partly attributed to HO-1 upregulation as shown by increased HO-1 expression levels in STZ-induced type 1 diabetes in Sprague-Dawley rats (122). Although the HO-1–inducing effects of resveratrol have not yet been described in humans, this dietary supplement is readily available for human use. Similar to CO, H₂S is also not clinically used in humans yet, although intravenous Na₂S administration has been performed in a phase 1 safety study (155). This study revealed increased H₂S and thiosulfate levels after Na₂S administration, indicating that circulating H₂S levels can be achieved following parenteral administration. The H₂S metabolite thiosulfate can also act as a H₂S donor via enzymatic conversion by rhodanese (also known as thiosulfate sulfurtransferase) (156). Thiosulfate is used as a treatment for calcifylaxis in patients with end-stage renal disease (111) and has been described as a protective agent in hypertensive heart and renal disease in rats (109,157). Sulfhydrylated ACE inhibitors, such as zofenopril and captopril, show additional beneficial effects in different trials (158). Recently, it was demonstrated that the beneficial effects of sulfhydrylated ACE inhibitors are explained by H₂S release (159). Finally, H₂S is also generated by various species of sulfate-reducing bacteria in the gut. Germ-free mice showed significantly lower levels of H₂S (160), indicating that the addition of dietary sulfate or sulfur-containing amino acids can act as natural H₂S donors.

Various gasotransmitter-based strategies are currently being studied as potential strategy to treat vascular dysfunction. So far, these strategies have not been explored in the context of diabetes-associated vascular disease. Because of the toxicity of high concentrations of gasotransmitters, as well as their potential deleterious effects on the development of vascular disease (as discussed in this Perspective), prudence is called for when considering exogenous administration of gasotransmitters. However, gasotransmitter-based interventions are relatively safe, mainly because these gases are also produced endogenously and therefore are highly promising candidate therapeutics.

Acknowledgments. The authors would like to thank Amanda Gautier from Gautier Scientific Illustration for preparing the artwork.

Funding. This work was supported by grants from the Deutsche Forschungsgemeinschaft (IRTG 1874-1 DIAMICOM) (to H.-P.H., W.G., and J.-L.H.), the Dutch Kidney Foundation (NSN C08-2254 and IP13-114) (to H.v.G.), and the Graduate School of Medical Sciences of the University of Groningen (to J.C.v.d.B.)

Duality of Interest. No potential conflicts of interest relevant to this article were reported.