Altered bioavailability of epoxyeicosatrienoic acids is associated with conduit artery endothelial dysfunction in type 2 diabetic patients

This study was performed in a total of 74 subjects. Because the presence of hypertension may affect the evaluation of EETs pathway [7, 10], this cohort was divided in 4 groups as follow: healthy subjects, essential hypertensive subjects, type 2 diabetic subjects without hypertension and type 2 diabetic subjects with hypertension (Table 1). Subjects who smoked more than 5 cigarettes per day, with cardiac and/or cerebrovascular ischemic vascular disease, heart failure or impaired renal function (estimated glomerular filtration rate < 60 mL/min/1.73 m²) were excluded from the study. Patients with type 2 diabetes must not have a HbA1c > 9.4 mmol/L (7.5%), obtained with lifestyle management and standard hypoglycaemic agents except insulin.

Radial artery endothelium-dependent flow-mediated dilatation was assessed using the previously validated hand skin heating technique, from 34 to 44 °C, allowing to induce a progressive and sustained increase in blood flow [7, 10, 11]. Radial artery endothelium-independent dilatation, corresponding to the ability of smooth muscle cells to relax in response to exogenous NO, was assessed then using 0.3 mg sublingual glyceryl trinitrate (GTN). See the Additional file 1 for further details.

A 4-F catheter was inserted into the forearm cephalic vein, when accessible, allowing blood sampling in the venous return at 34 and 44 °C for the quantification of the endothelial factors that have shown to be involved in flow-mediated dilatation [7]. Quantification of plasma 8,9-EET, 11,12-EET, 14,15-EET and corresponding DHET regioisomers was performed by a recently validated method based on lipid extraction and saponification followed by liquid chromatography coupled to tandem mass spectrometry analysis (LC–MS/MS) [10]. Total EETs + DHETs levels and ratio of the preferential substrate of sEH 14,15-EET to its metabolite 14,15-DHET was used as an index of EETs production and sEH activity respectively [10]. In addition, the levels of epoxy fatty acids synthesized by CYPs from linoleic acid and from the omega-3 docosahexaenoic acid (DHA): 9,10- and 12,13-epoxyoctadecenoic (9,10-EpOME and 12,13-EpOME), 19,20-epoxydocosapentaenoic acid (19,20-EpDPA) and their respective diols produced by sEH, 9,10-, 12,13-dihydroxyoctadecenoic acid (9,10-DiHOME and 12,13-DiHOME) and 19,20-dihydroxydocosapentaenoic acid (19,20-DiHDPA), were quantified by LC–MS/MS. Quantification of the plasma level of the NO metabolite nitrite was performed by a tri-iodide/ozone-based chemiluminescence assay [7]. Additional blood sampling was performed at 34 °C to quantify the whole blood level of reactive oxygen species (ROS) by electron paramagnetic resonance spectroscopy (Miniscope MS-200, Magnettech) [7]. In addition, peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation in order to determine sEH protein expression and activity using an ultrasensitive nanobody polymeric horseradish peroxidase-based immunoassay and a [³H]-trans-1,3-diphenylpropene oxide-based radioactivity assay respectively [12, 13]. Results were normalized to total protein concentration in PBMCs. In addition, mRNA expression levels of sEH and of the main CYPs involved in the vascular production of EETs CYP2C9 and CYP2C19 were determined in PBMCs by quantitative RT-PCR (primers are shown in Additional file 1: Table S1). Absolute abundance of mRNA relative to beta2-microglobulin housekeeping gene expression was calculated based on a cDNA standard curve. See Additional file 1 for further details.

Hyperglycemic and hyperinsulinemic euglycemic clamps were performed in 8 subjects selected from the healthy group (mean [SD] age: 46 ± 5 years, 6 male subjects) and explored on two separate occasions after a 12 h overnight fasting. Briefly, a venous catheter was inserted in a large vein at the antecubital fossa of the dominant arm, allowing a variable infusion rate of a 20% glucose solution alone or combined with continuous regular insulin administration (Actrapid^® 100 UI/mL, Novo Nordisk: 250, 200, 150, 100 mU/m² of body surface area [BSA] per min for 2 min each, and then 80 mU/m² of BSA per min) to achieve and maintain either a target steady-state capillary glucose concentration of 11 mmol/L (200 mg/dL) for the hyperglycemic clamp or 5.5 mmol/L (100 mg/dL) for the hyperinsulinemic clamp, without change in glucose infusion rate for at least 30 min before assessing radial artery endothelium-dependent and -independent dilatation. A control procedure was also performed with saline infusion. Blood sampling were performed before beginning the clamp, at glycemic steady-state before hand skin heating and at the end of hand skin heating to determine the evolution of nitrite, EETs and DHETs bioavailability.

Analyses were performed using NCSS software (version 07.1.14) and R software (version 3.5.1). The Shapiro–Wilk test was used to assess normality of data.

Differences between healthy, hypertensive, type 2 diabetic and hypertensive type 2 diabetic groups were assessed by Fisher’s exact test for categorical variables and by one-way ANOVA or nonparametric Kruskall–Wallis test for continuous variables, followed in case of significance, by Tukey’s test or Dunn’s test for multiple comparisons.

Significant predictors for radial artery flow-mediated dilatation (percent change in radial artery diameter from 34 to 44 °C) were assessed using both backward and forward stepwise regression based upon Akaike Information Criterion (AIC) using a null model without predictors and a full model including diabetes, hypertension, sex, age, smoking status, mean arterial blood pressure, BMI, LDL cholesterol level, triglyceride-glucose (TyG) index (log [fasting triglycerides (mg/dL) × fasting glucose (mg/dL)/2]), creatininemia, baseline radial diameter, variation in shear stress and GTN-induced dilatation as predictors. Similar analyses were performed for shear stress variation during heating and GTN-induced dilatation.

Pearson correlation analyses were performed to determine linear relationships between biological and functional variables of interest.

The hyperglycemic clamp increased plasma glucose and insulin concentrations while the hyperinsulinemic clamp increased insulin levels to a higher level, without significantly affecting glucose level (Additional file 1: Figure S3a, b).

Compared to saline infusion, the basal radial artery diameter was increased by both clamps at steady-state before hand skin heating but statistical significance was only reached with the hyperinsulinemic clamp, without change in mean wall shear stress and in systemic hemodynamics (Additional file 1: Table S4). Radial artery flow-mediated dilatation in response to heating (Fig. 5a) was reduced by both clamps, without change in mean wall shear stress variation (Fig. 5b) nor in endothelium-independent dilatation to GTN (Fig. 5c).

The plasma nitrite levels were not significantly modified by both clamps at steady-state before hand skin heating (Fig. 5d). Nitrite levels increased during hand skin heating in presence of saline, was not modified during the hyperglycemic clamp but increased compared to baseline conditions during the hyperinsulinemic clamp (Fig. 5d). In addition, the hyperinsulinemic clamp increased the plasma levels of EET regioisomers (Additional file 1: Figure S4a) without significant change in the levels of DHET regioisomers (Additional file 1: Figure S4b), resulting in an increase in total EETs + DHETs (Fig. 5e). In addition, the 14,15-DHET-to-14,15-EET ratio was decreased during the hyperinsulinemic clamp (Additional file 1: Figure S5). A trend for an increase in the levels of EET regioisomers was also observed during the hyperglycemic clamp but there was no change in the 14,15-DHET-to-14,15-EET ratio (Fig. 5e; Additional file 1: Figures S4 and S5). Heating induced an increase in the plasma levels of EETs and DHETs in presence of saline but not during both clamps and even, EETs and DHETs became similar to baseline conditions (Fig. 5e; Additional file 1: Figure S4).

In this study, we carefully selected type 2 diabetic patients treated according to recent recommendations for optimization of glycemic control together with management of associated risk factors, mainly LDL cholesterol and blood pressure. To evaluate the proper impact of type 2 diabetes independently from the rise in blood pressure, we compared the functional and biological data of healthy, essential hypertensive, type 2 diabetic and hypertensive type 2 diabetic subjects.

In this context, we demonstrated that type 2 diabetic patients either in absence or in presence of hypertension display a decreased sustained radial artery flow-mediated in response to heating, taking into consideration cardiovascular risk factors, the flow-dependent stimulus and endothelium-independent dilatation to GTN. These results demonstrate that well-controlled type 2 diabetic patients still exhibit an endothelial dysfunction of peripheral conduit arteries, as previously shown using more transient increase in blood flow induced by post-ischemic hyperemia [1, 2, 14‐16]. In addition, a decreased endothelium-independent dilatation in response to exogenous NO was also observed in type 2 diabetic patients but only in absence of hypertension, suggesting that antihypertensive agents may contribute to prevent smooth muscle cell dysfunction in type 2 diabetes. To note, the decrease in the flow-mediated dilatation of patients with treated hypertension alone appeared less marked but was statistically significant when compared to healthy subjects, taking into consideration cardiovascular risk factors. In fact, decreasing blood pressure per se allows to normalize endothelial function in less than half of the patients and some antihypertensive medications, in particular blockers of the renin-angiotensin system, have direct protective effects on vascular function [7, 17, 18].

Regarding the NO pathway, the basal nitrite level was elevated but did not further increase during the sustained endothelial stimulation in the two groups of diabetic patients. This may be related to the increase in oxidative stress in diabetic patients illustrated by the elevation in the whole blood levels of ROS. Indeed, previous work suggested that hyperglycemia may lead to an increased expression of endothelial NO-synthase to compensate for the enhanced inactivation of NO by superoxide anions and promoting increased peroxynitrite formation [19, 20]. A parallel increase in the expression of inducible NO-synthase related to the pro-inflammatory phenotype associated with type 2 diabetes may exist [20]. Part of the NO synthesized in excess is not degraded but converted to nitrite explaining the increased basal level [19]. However, the absence of increase in nitrite level during heating demonstrate an impairment in stimulated NO bioavailability at the level of conduit arteries in type 2 diabetes. As supported by correlations analyses, the NO/ROS imbalance may contribute to the observed reduction in endothelium-dependent dilatation of conduit arteries in type 2 diabetic patients. Accordingly, a decrease in NO bioavailability was previously reported in forearm resistance arteries using functional approaches based on the local infusion of the NO-synthase inhibitor L-NMMA during endothelial stimulation [21, 22]. In contrast, patients with treated essential hypertension still had an increase in NO level during heating and no change in ROS levels. Thus, while non-treated hypertensive patients have a reduced NO bioavailability [7], this maintained response in subjects mainly treated with blockers of the renin-angiotensin system may contribute to partially preserve the endothelial vasomotor function.

Furthermore, our results demonstrate for the first time in humans that the endothelial dysfunction of conduit arteries in type 2 diabetes is associated with an alteration in the EETs pathway. In fact, according with previous clinical studies [7, 23, 24], we cannot detect a significant difference between groups for the baseline plasma levels of EET and DHET regioisomers but this may be due to the limited number of patients in the hypertensive and diabetic groups. However, we can demonstrate that, in contrast to what is observed in the healthy subjects, EETs did not increase during the endothelial stimulation in hypertensive patients, as previously shown in both treated and non-treated subjects [7, 10], and they even decreased in the 2 groups of diabetic patients. Although the magnitude of the increase in EETs in healthy subjects appears modest, we observed using functional inhibition of CYPs that a similar magnitude of the variation in EETs during heating can contribute up to one-third of the endothelial-dependent dilatation in response to the sustained flow increase [7]. Thus, because plasma levels of DHETs and thus EETs + DHETs also decreased during heating in type 2 diabetic patients, we can speculate that the production of EETs cannot increase in response to shear stress contributing, as suggested from correlation analyses, to the endothelial vasomotor dysfunction. In fact, increasing blood flow while keeping the same EET production than during baseline flow conditions has a diluting effect; this is probably the main explanation for the decrease in local concentrations of EETs and DHETs in diabetic patients. This alteration in EETs production may be related to a reduction in CYPs expression levels, as previously observed at the renal level in obese and in salt-sensitive Dahl rats and notably contributing to the development of high blood pressure [4, 25, 26]. Unfortunately, in accordance with previous results [27], we were not able to detect in PBMCs the expression of the main CYPs involved in the vascular production of EETs to confirm this hypothesis. In addition to the putative decrease in EETs production, our results show the presence of an increased conversion of EETs to DHETs by sEH. Indeed, we noticed an increased 14,15-DHET-to-14,15-EET ratio in plasma and directly demonstrated an increased sEH expression and activity in PBMCs of type 2 diabetic patients compared to healthy subjects. Although sEH measurements can only be performed non invasively in vivo in PBMCs, they probably reflect the evolution of sEH at the vascular level, which probably plays an important role in metabolizing the EETs released during the endothelial stimulation, preventing their relaxing effect on vascular smooth cells. This hypothesis is supported by the inverse correlations between sEH activity with EETs variation during heating and the magnitude of flow-mediated dilatation. This increased sEH expression/activity is consistent with most of the results obtained in animal models of obesity-induced insulin resistance [27‐30]. Of note, the percentage of patients that were on blockers of the renin-angiotensin system was similar in the hypertensive and hypertensive type 2 diabetic groups and thus, these agents may have contributed to prevent the upregulation of sEH expression induced by angiotensin II in essential hypertensive subjects but not in type 2 diabetic patients [31].

In contrast, no change in the plasma levels of other epoxy fatty acids and diols derived from linoleic acid or omega-3 were observed, suggesting that their altered metabolism does not contribute to the endothelial dysfunction in our population of well-controlled type 2 diabetic patients. These results extent recent works showing that that the levels of omega-3 and linoleic acid appear no altered when glycemic control is achieved and that omega-3 supplementation is ineffective to improve the metabolic and inflammatory status of patients with well-controlled type 2 diabetes [32, 33]. Finally, as previously suggested [23, 34], we cannot exclude that an increase in the levels of the vasoconstrictor 20-hydroxyeicosatetraenoic (20-HETE) synthesized from arachidonic acid by CYPs hydroxylases contributes to the endothelial dysfunction of essential hypertensive and type 2 diabetic patients.

Going further, we assessed the role of hyperglycemia and hyperinsulinemia in the development of endothelial dysfunction using clamp studies in healthy subjects. Both clamps slightly induced an increase in radial artery diameter, without significant change in mean wall shear stress, but this increase appears stronger with the euglycemic hyperinsulinemic clamp reaching statistical significance compared to basal conditions. This arterial vasodilating effect was probably related to the potentiation of endothelial factor release by insulin, although the increase in nitrite levels was not significant. Previous results showed that acute hyperglycemia or hyperinsulinemia dilates peripheral resistance arteries by a NO-dependent mechanism involving endothelial NO-synthase phosphorylation [35, 36], and our results suggest that this mechanism may be less efficient during more sustained hyperinsulinemia. In contrast, we observed that the levels of EET and DHET regioisomers were increased at steady-state during the euglycemic hyperinsulinemic clamps. To our knowledge, this is the first study showing directly in humans that insulin promotes the production of EETs and that this mechanism may contribute to the vasodilator effect and potentiate insulin sensitivity [9]. Moreover, we observed a marked decrease in the 14,15-DHET-to-14,15-EET ratio, suggesting that insulin may also decrease EETs degradation. Whether the transitional tyrosine phosphorylation of sEH, suggested to modulate its activity [37, 38], is involved in this effect of insulin remains to be determined.

Despite this basal vasodilator effects and as generally observed using post-ischemic hyperemia [39‐44], hyperglycemia and hyperinsulinemia reduced the endothelium-dependent flow-mediated dilatation of peripheral conduit arteries but the mechanisms involved appear somewhat different. As suggested from the absence of increase in nitrite levels during heating, hyperglycemia may reduce local NO bioavailability during endothelial stimulation. This may be related to the increase in oxidative stress, as suggested from the reversion of endothelial dysfunction induced by hyperglycemia obtained with antioxidants [41, 42]. In addition, we did not observe an increase in EETs during the hyperglycemic clamp suggesting altered EETs production. For the euglycemic hyperinsulinemic clamp, the impact on stimulated NO availability appears less marked because nitrite levels finally increased in response to heating but only when compared to baseline conditions before the beginning of the clamp. However, EETs levels did not increase and even slightly decreased in response to heating. Thus, the difference between the basal conditions before the beginning of the clamp was no more statistically significant, demonstrating that a marked alteration in stimulated EETs production cannot compensate for the diluting effect of the increase in blood flow. The mechanism of the discrepancy between the impact of hyperinsulinemia at baseline and during flow stimulation on EETs production remain to be elucidated but insulin may promote the release of EETs stored in phospholipids pools while altering the de novo production of EETs by CYPs, needed for flow-mediated dilatation [7, 45].