Preserved endothelial function in human obesity in the absence of insulin resistance

Background Insulin resistance (IR) is frequently associated with endothelial dysfunction and has been proposed to play a major role in cardiovascular disease (CVD). On the other hand, obesity has long been related to IR and increased CVD. However it is not known if IR is a necessary condition for endothelial dysfunction in human obesity, allowing for preserved endothelial function in obese people when absent. Therefore, the purpose of the study was to assess the relationship between IR and endothelial dysfunction in human obesity and the mechanisms involved. Methods Twenty non-insulin resistant morbid obese (NIR-MO), 32 insulin resistant morbid obese (IR-MO), and 12 healthy subjects were included. Serum concentrations of glucose, insulin, interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), resistin and adiponectin were determined. IR was evaluated by HOMA-index. Endothelium-dependent relaxation to bradykinin (BK) in mesenteric microvessels was assessed in wire myograph. Results Serum IL-6, and TNF-α levels were elevated only in IR-MO patients while resistin was elevated and adiponectin reduced in all MO individuals. Mesenteric arteries from IR-MO, but not from NIR-MO subjects displayed blunted relaxation to BK. Vasodilatation was improved in IR-MO arteries by the superoxide scavenger, superoxide dismutase (SOD) or the mitochondrial-targeted SOD mimetic, mito-TEMPO. NADPH oxidase inhibitors (apocynin and VAS2870) and the nitric oxide synthase (NOS) cofactor, tetrahydrobiopterin failed to modify BK-induced vasodilatations. Superoxide generation was higher in vessels from IR-MO subjects and reduced by mito-TEMPO. Blockade of TNF-α with infliximab, but not inhibition of inducible NOS or cyclooxygenase, improved endothelial relaxation and decreased superoxide formation. Conclusions Endothelial dysfunction is observed in human morbid obesity only when insulin resistance is present. Mechanisms involved include augmented mitochondrial superoxide generation, and increased systemic inflammation mediated by TNF-α. These findings may explain the different vascular risk of healthy vs unhealthy obesity.


Background
Obesity represents one of the main health problems in modern societies, reaching epidemic proportions in developed countries [1]. Jointly to other consequences on health, obesity is one of the leading risk factors for cardiovascular disease (CVD). Endothelial dysfunction precedes in several years the clinical manifestations of CVD being a key event in its development [2]. A growing body of evidence reveals the presence of an altered endothelial vascular function manifested by a reduction in the endothelium-dependent vasodilatations in both animal models of obesity [3] and human obese subjects [4,5]. Frequently observed in obese patients, insulin resistance appears among the different pathogenic mechanisms leading to endothelial dysfunction. In fact, an impairment of endothelial function associated to insulin resistance has been documented in different vascular districts [3,4,6,7].
Although the high prevalence of both insulin resistance and endothelial dysfunction in obese people strongly suggests that these two usual features in the obese phenotype are causally linked, there is no available data in the literature directly addressing the implication of insulin resistance in the pathogenesis of endothelial dysfunction in human obesity. In fact studies showing impaired endothelial function in obese people have been carried out mainly by comparing endothelial function in vessels from obese subjects (without establishing differences between those who are sensitive or resistant to insulin action) to vessels from healthy adult subjects with normal weight (with preserved endothelial function). This is a key issue since clinical evidence suggests that there is a significant proportion (~30%) of obese individuals that are free from insulin resistance and does not manifest the panoply of cardiovascular diseases (CVD) usually observed in obese people [8]. This raises the hypothesis of a preservation of endothelial function and, consequently, vascular disease, when insulin resistance is lacking. Taken together, all these observations point to insulin-resistance as a major factor to explain the difference in CVD between both types of obese people.
Inflammation and reduced bioavailability of nitric oxide (NO) due in part to increased oxidative stress seem to play an outstanding role in the endothelial dysfunction associated to obesity [9,10]. Low-grade chronic vascular inflammation has been suggested to be a key event in the pathogenesis of endothelial dysfunction and CVD. Increased plasma concentrations of interleukin-6 (IL-6) and other inflammatory cytokines have been associated with the presence of endothelial dysfunction in obese patients [11]. In addition, tumor necrosis factoralpha (TNF-α) and IL-6 levels have been shown to be independent predictors of coronary endothelial function [12]. Furthermore, inflammation and oxidative stress are intimately related processes. In fact, inflammatory cytokines reduce NO-mediated responses by increasing reactive oxygen species (ROS) production that results in decreased NO availability [13]. However, once again, no data is available in human vessels concerning the role of inflammatory cytokines or the contribution of oxidative stress to endothelial dysfunction in obese subjects with respect to the presence or absence of insulin resistance.
Although strong evidences support the relationship between insulin resistance and obesity as well as between insulin resistance and endothelial dysfunction, studies analyzing the specific role of insulin resistance in endothelial dysfunction associated to obesity are definitely lacking. Therefore, the aim of the present study was to determine the role of insulin resistance on endothelial function in obesity as a marker of early vascular damage that could provide a potential explanation for the different vascular outcomes in these two largely recognized subtypes of human morbid obesity: healthy and unhealthy. With this purpose we evaluated endothelial-dependent vasodilation in mesenteric microvessels from morbid obese individuals, with or without insulin-resistance, and the influence of inflammation and vascular ROS generation on these responses.

Study population
The present study included 52 morbid obese (MO) subjects with a body mass index (BMI) ≥ 40 kg/m 2 aged between 25 and 65 years, who underwent bariatric surgery in the Hospital Universitario de Getafe with mixed techniques combining Roux-en-Y gastric bypass, vertical sleeve gastrectomy and adjustable gastric banding. Subjects with history or clinical evidence of CVD (congestive heart failure; cardiac and/or cerebrovascular ischemic disease) were excluded, but not those with cardiovascular risk factors like hypertension, type 2 diabetes, or dyslipidemia. Subjects were characterized as metabolic syndrome (MetS) if they met three of the following National Cholesterol Education Program Adult Treatment Panel III criteria as modified by the American Diabetes Association.
Insulin resistance was estimated by calculating the validated score for Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) [14]. MO subjects were distributed into two groups depending on the HOMA-IR: non-insulin resistant (NIR; HOMA IR< 3.8; n=20) and insulin resistant (IR; HOMA-IR≥ 3.8; n=32). Cut-off point was previously defined for a Spanish population corresponding to the 90th percentile HOMA-IR value in non-diabetic adult subjects [15].
Twelve healthy non-obese subjects were recruited, as control group, among patients undergoing laparoscopic surgery procedures in our hospital (hiatus hernia repair, achalasia or cholecystectomies). Exclusion criteria for this control group were: BMI≥ 30 kg/m 2 ; current or former smokers; hypertension (≥140/90 mmHg); diabetes; total cholesterol >200 mg/dl; history or clinical evidence of cardiovascular disease and any other condition that might interfere with the progress of the study. Written informed consent was obtained from the subjects who participated in the study. The study was approved by both the Ethics Committee and Research Committee of the Hospital Universitario de Getafe.

Biochemical measurements
Blood samples were collected for measurement of serum fasting glucose, glycosylated hemoglobin (HbA 1C ), serum fasting insulin, serum lipid profile and C-reactive protein (CRP) in the Laboratory of Clinical Analysis of our hospital. HOMA-IR was calculated as described by Mathews et al. [16].
Circulating levels of TNFα, IL-6, adiponectin (R&D systems) and resistin (Bender MedSystems) were determined in serum by ELISA, following the manufacturer instructions, in the Research Unit Laboratory. All samples were assessed in duplicate.

Measurement of vascular reactivity in isolated vessels
Small arteries (200-500 μm, approximately 2 mm length) were isolated from visceral fat obtained from MO or control subjects during laparoscopic surgery. The arteries were cleaned, and mounted as ring preparations on small vessel wire myographs, as described elsewhere [17,18]. Resting tension for the experiments was calculated to obtain an internal circumference equivalent to 90% of the tension of the vessels when relaxed in situ under a transmural pressure of 100 mmHg, using the Myo-Norm-4 program (Cibertec, Madrid, Spain). The vessels were contracted with K + 125 mmol/l and the segments failing to produce a tension equivalent to a pressure of 100 mmHg were rejected. After a washout period, the arteries were contracted again with 25 mmol/l K + , which produced approximately 80% of the maximum response. When the contraction reached a plateau, the endothelium-dependent relaxation was assessed by adding increasing concentrations of bradykinin (BK; 10 nmol/l to 3 μmol/l). Vasodilatory effects of insulin (0.01 nmol/l to 3 μmol/l) were also evaluated. Responses to sodium nitroprusside (SNP; 1 nmol/l to 300 μmol/l) were used to test non endothelium-dependent relaxant responses.

Detection of vascular superoxide anion generation in microvessel segments
Formalin-fixed, paraffin-embedded transverse sections (5 μm thick) were mounted on polylysine-coated glass slides. The in situ production of superoxide anion was measured by means of the fluorescent dye dihydroethidium (DHE) as described previously [18]. Briefly, mesenteric arterial sections were incubated for 90 min at 37°C with the fluorescent probe DHE (4 μmol/l; Calbiochem, Darmstadt, Germany). In the presence of superoxide anion, DHE is oxidized to ethidium that yields bright red fluorescence. After washing with PBS plus 0.1% Triton X-100, sections were mounted and visualized by fluorescence microscopy (Olympus BX51, Japan). The percentage of nuclei showing positive red signal with respect to total nuclei in arterial wall was determined with imaging software (McBiophotonics Image J, NIH, Bethesda, Maryland).

Statistical analysis
Categorical variables were analyzed by the Chi-square test. Results from numerical variables were expressed as mean±SEM. The number of subjects (N) and the number of vascular segments (n) used for relaxation curves are indicated in each graph. Complete concentrationresponse curves to BK or to SNP in vessels from the participants were compared by a two-factors analysis of variance (ANOVA). Comparison of pD 2 , (which is defined as the negative logarithm of the concentration of BK required to obtain 50% of maximal relaxation) for each subject was statistically analyzed by a one-factor ANOVA followed by a Student-Newmann-Keuls test. This same analysis was also used to compare serum cytokine determinations and DHE staining. Student t test was used in all other comparisons. Pearson′s correlation coefficient was used to assess correlation among variables. Models of lineal regression were constructed, with the pD 2 considered as the dependent variable. In all cases, a probability value of less than 5% was considered significant. Data were analyzed using version 20.0 of SPSS software (SPSS Corp, Chigaco, IL).

Results
Insulin resistant morbid obese (IR-MO) subjects when compared to non insulin-resistant morbid obese (NIR-MO) group showed significantly larger waist circumference and higher levels of fasting insulin, HOMA-IR, HbA 1C and CRP, while HDL cholesterol levels were lower (Table 1). In addition, the percentage of MetS was significantly higher in IR-MO versus NIR-MO subjects. Comorbid conditions and treatments received by the patients at the time of surgery are shown in Table 1.
Regarding biomarkers of inflammation and adipose tissue activity, TNF-α and IL-6 levels were increased in IR-MO patients when compared to control subjects, while there were no differences between NIR-MO and control subjects (Table 2). Finally, the levels of adiponectin were reduced and those of resistin increased, in both IR-MO and NIR-MO patients ( Table 2).

Endothelial dysfunction is associated with insulin resistance in morbid obese subjects
Small mesenteric arteries from NIR-MO did not show significant differences in the BK-evoked vasodilation when compared to control non-obese subjects ( Figure 1A). By opposite, a significant reduction in the endotheliummediated vasodilation induced by BK was observed in mesenteric microvessels obtained from IR-MO group when compared to those from control or NIR-MO subjects ( Figure 1A). To avoid a potential bias due to the comparison of the vascular responses instead of comparing the response in the subjects, we confirmed these results assessing the differences between the pD 2 values for BK averaged from all the vessels from each subject by group (control, NIR-MO and IR-MO). This approach yielded very similar results to that observed when we analysed the response of the vessels. pD 2 for BK were 7.23±0.20 (N=12), 7.27±0.11 (N=20) and 6.69±0.14 (N=32) for control, NIR-MO and IR-MO, respectively (p < 0.01 IR-MO vs. NIR-MO or control subjects by one-factor ANOVA followed by Student-Newmann-Keuls).
Accordingly, the in vitro vasorelaxant response to BK in microarteries derived from MO subjects decreased as the HOMA-IR index increased with a clear correlation between the HOMA-IR index and the BK pD 2 values ( Figure 1B). Furthermore, among all factor studied (Age, BMI, waist circumference, fasting glucose, HbA 1C , total cholesterol (TC), HDL and LDL-cholesterol, TC/HDL ratio, triglycerides, CRP and HOMA-IR), only HOMA-IR (p= 0.017) was associated to BK pD 2 values.
The addition of insulin (0.01 nmol/l to 3 μmol/l) to the organ bath evoked a very weak relaxation in isolated mesenteric arteries derived from obese and control subjects and no differences were observed among those subjects (E max : 16.0%±10.5 vs 11.2%±3.0 and vs 12.5%±3.2 for control, NIR-MO and IR-MO subjects, respectively; p=n.s.).  Oxidative stress contributes to endothelial dysfunction in IR-MO subjects The contribution of superoxide anions to endothelial dysfunction in IR-MO subjects was confirmed. Preincubation of the microarteries from these subjects with the superoxide scavenger, SOD (100 U/ml), improved the endothelial-dependent relaxation to BK ( Figure 1C), while it did not modify the responses to BK in vessels from NIR-MO subjects.
Consistently, superoxide production, assessed by DHE fluorescence, was clearly appreciated in vascular wall sections from IR-MO subjects, while the fluorescence signal was very weak or absent in vessels from NIR-MO and control non-obese subjects ( Figure 3A). Quantification of staining revealed a significant increase in superoxide production in arteries from IR-MO with respect to the other two groups ( Figure 3B). In fact, superoxide generation negatively correlated with pD 2 for BK in mesenteric microvessels ( Figure 3C). Furthermore, the improvement of endothelium-dependent vasodilation with mito-TEMPO was accompanied by a significant reduction of superoxide generation in arterial wall from IR-MO subjects ( Figure 3A-B).

Discussion
The main finding of the present study is that endothelial dysfunction is present only in vessels from obese people with insulin resistance, suggesting that the association between morbid obesity and endothelial dysfunction is linked to the presence of insulin resistance. Such alteration is mediated by increased mitochondrial oxidative stress and by systemic pro-inflammatory cytokines (mainly TNF-α). The presence of other cardiovascular risk factors commonly found in obesity does not seem to explain the impact of IR on endothelial function in MO subjects, since no significant differences were observed between both groups of obese subjects. In addition, only insulin resistance (HOMA-IR) remained associated with endothelial function in obese subjects after adjustment.
Furthermore, endothelial dysfunction associated to insulin resistance in obese subjects does not seem to rely on a defective vasodilatory action of insulin since, in contrast to BK-induced responses, insulin-induced vasodilatory responses were not different between MO subjects with or without IR.
To our knowledge this is the first study describing this phenomenon in human beings, although recent results obtained in the microcirculation of obese mice support our hypothesis [19]. Other previous studies had suggested a relationship between endothelial function and insulin, HOMA-IR and triglycerides in obese subjects [20] but without establishing a clear causal relationship since the design of the study did not allow for comparing people with insulin-resistance versus people without insulin-resistance. Supporting our findings there is another study showing that IR and not BMI is the major determinant in the impairment of flow mediated dilation in morbid obese subjects [21]. In this sense, an improvement of the endothelial function after weight reduction in obese patients correlates with reduction in HOMA index rather than with reduction in body weight [22].
Although not reaching statistical significance, a larger percentage of IR-MO subjects were taking hypoglycemiants, being the vast majority metformin and pioglitazone. It is unlikely that this fact could have influence on the impaired endothelial dysfunction in IR-MO group since these agents have been reported to even improve endothelial vasodilation [23]. Even more, to discard a possible bias in the results due to the inclusion of twelve patients with the diagnosis of type 2 diabetes in the group of IR-MO patients, we made additional analysis of the data after excluding diabetic patients. The differences between the groups (NIR-MO and IR-MO) remained without changes. Furthermore, when diabetic type 2 patients were excluded, the HOMA-IR still significantly correlated with poorer BK-induced vasodilation, therefore reinforcing the hypothesis that views insulin resistance as a main determinant of endothelial dysfunction in obese subjects independently of the presence of diabetes. In addition, the percentage of metabolic syndrome is more frequent in IR-MO when compared to NIR-MO subjects, supporting the correct classification of the patients.
Several pathophysiological mechanisms are possibly involved in the interactions among obesity, insulin resistance and endothelial dysfunction [24]. In this sense, microarteries from IR-MO subjects generate higher amounts of intravascular ROS than those from NIR-MO or control subjects, as demonstrated by DHE staining. In fact, the superoxide scavenger, SOD, caused a significant improvement of endothelium-dependent relaxation only in mesenteric arteries from IR-MO subjects, suggesting that increased superoxide production contributes to endothelial dysfunction in IR-MO vessels by reducing NO availability. The involvement of superoxide excess in endothelial dysfunction in obesity is supported by demonstration of increased vascular superoxide generation in obese subjects [25] and in animal models of obesity [26] and by the potentiating effect of superoxide scavengers on ACh-induced relaxations in aortic and mesenteric arteries from obese animals [27,28] and in mesenteric microarteries from obese patients [25]. Once again, these studies did not establish differences based on the presence or absence of insulin resistance. Present results suggest that increased superoxide production and subsequent NO inactivation is associated to IR rather than to obesity itself, since elevated vascular superoxide generation and improving vasodilation in response to SOD are not detected in obese subjects without IR, who do not present altered endothelial vasodilation although they have a BMI similar to those with IR.
Several sources of ROS could contribute to the increased superoxide production in the vasculature of IR-MO subjects. In contrast to that observed in the vasculature of obese subjects in other studies [25], we did not find evidence for the possible implication of NADPH oxidase since its inhibitors, apocynin and VAS2870, did not improve endothelial vasodilation in mesenteric microarteries from IR-MO subjects. NOS uncoupling can also act as an endothelial source of superoxide. Reduced availability of its cofactor, BH 4 , in pro-oxidant conditions has been proposed as a mechanism responsible for eNOS uncoupling [29]. The failure of BH 4 to modify vasodilatations suggests that this does not account for the reduced NO availability associated to IR in MO subjects. The improvement of endothelial vasodilation caused by the mitochondria-targeted superoxide scavenger, mito-TEMPO, in vessels from IR-MO subjects suggests that the origin of the impaired NOmediated responses in IR associated with obesity is an increased mitochondrial superoxide production that limits NO availability. Mito-TEMPO has been demonstrated to specifically scavenge mitochondrial superoxide in endothelial cells and vascular tissue [30,31]. Mitochondria is considered the major intracellular site of superoxide production generated from electron leaks in the mitochondrial electron transport system [32]. When metabolic alterations lead to inefficient nutrient oxidation, mitochondria increases the production of superoxide and other ROS [32]. There is substantial evidence proposing the importance of these mitochondrial ROS in the pathophysiology of the vascular damage linked to diabetes and insulin resistance [33]. In fact, clinical evidence has associated insulin resistance with mitochondrial dysfunction [34,35]. Although it is difficult to ascertain whether defects in mitochondria occur before or after the onset of insulin resistance [33], it has been demonstrated that induction of mitochondrial superoxide production results in insulin resistant adipocytes [36].
On the other hand, the elevated levels of CRP and the pro-inflammatory cytokines, TNF-α and IL-6, reflect the presence of low-grade inflammation in obese subjects with IR that could contribute to endothelial dysfunction. In this sense, IL-6 and TNF-α have been demonstrated to impair endothelial dilation [37,38]. Furthermore, the effects produced by infliximab on endothelial vasodilatation in obese subjects, only when IR is present, suggest a causal role of inflammatory cytokines, TNF-α at least, in the endothelial impairment exhibited by the vessels from these subjects. Previous evidences showing that blockade of TNF-α with infliximab results in the recovery of impaired endothelial vasodilation in omental vessels from obese subjects [25] support this hypothesis. CRP levels did not show any correlation with pD2, although we cannot definitively exclude its participation.
The impact of systemic inflammatory cytokines in endothelial dysfunction does not seem to be mediated by up-regulation of conventional local inflammatory pathways, since both inhibitors of iNOS and COX did not improve endothelial relaxation. Deleterious effects of pro-inflammatory cytokines on endothelial function in obese patients could be also mediated by oxidative stress generation since these effects can be reversed by superoxide scavengers [25,39]. Furthermore, exposure to TNF-α stimulates mitochondrial superoxide production in mouse adipocytes [36] and human retinal endothelial cells [40]. This is consistent with the reduction of superoxide production observed in arteries from IR-MO subjects after the treatment with infliximab. Thus, systemic inflammation and mitochondrial superoxide production could be two closely related or even interdependent mechanisms mediating endothelial dysfunction in insulin resistance associated with obesity.
Obese subjects exhibit increased circulating levels of pro-inflammatory resistin and decreased levels of the anti-inflammatory cytokine, adiponectin, irrespective of the state of IR, which is in agreement with previous studies done in obese patients [41]. Then, increased adiposity is associated with adipose tissue inflammation and dysregulation of adipokines but this condition is related to obesity per se independently of the insulin resistance state, and, therefore, it does not seem to be determinant for the development of endothelial dysfunction.
We recognize that one limitation of the study is that vascular function has been evaluated in subjects with severe obesity so caution is required when any extrapolation from the present is made to overweight or nonsevere obese subjects. Another potential limitation is that the present findings have been achieved based on an in vitro approach where endothelial function has been assessed in isolated microarteries segments. Therefore further in vivo studies (carotid artery ultrasound and/or endo-peripheral arterial tone) are needed to confirm our findings. However, a recent study showing in vivo a strong dependence of endothelial dysfunction on insulin resistance in obese individuals supports our findings [21].

Conclusions
The present results reveal the determinant role of insulin resistance in the impairment of endothelium-dependent vasodilation in morbid obese patients. Endothelial dysfunction is produced by means of a reduced NO availability as a result of increased vascular superoxide generation by mitochondria. Moreover, the inflammatory mediator TNF-α contributes to the impaired vasodilation associated to IR in obese subjects. These findings may help to explain the link between obesity, IR and vascular dysfunction and provide targets for specific intervention aimed at delaying or preventing the appearance of vascular disease.