Heart and Circulatory Physiology

Insulin resistance does not diminish eNOS expression, phosphorylation, or binding to HSP-90

David Fulton, M. Brennan Harris, Bruce E. Kemp, Richard C. Venema, Mario B. Marrero, David W. Stepp


Previously, using an animal model of syndrome X, the obese Zucker rat (OZR), we documented impaired endothelium-dependent vasodilation. The aim of this study was to determine whether reduced expression or altered posttranslational regulation of endothelial nitric oxide synthase (eNOS) underlies the vascular dysfunction in OZR rats. There was no significant difference in the relative abundance of eNOS in hearts, aortas, or skeletal muscle between lean Zucker rats (LZR) and OZR regardless of age. There was no difference in eNOS mRNA levels, as determined by real-time PCR, between LZR and OZR. The inability of insulin resistance to modulate eNOS expression was also documented in two additional in vivo models, the ob/ob mouse and the fructose-fed rat, and in vitro via adenoviral expression of protein tyrosine phosphatase 1B in endothelial cells. We next investigated whether changes in the acute posttranslational regulation of eNOS occurs with insulin resistance. Phosphorylation of eNOS at S632 (human S633) and T494 was not different between LZR and OZR; however, phosphorylation of S1176 was significantly enhanced in OZR. Phosphorylation of S1176 was not different in the ob/ob mouse or in fructose-fed rats. The association of heat shock protein 90 with eNOS, a key regulatory step controlling nitric oxide and aberrant O2 production, was not different between OZR and LZR. Taken together, these results suggest that changes in eNOS expression or posttranslation regulation do not underlie the vascular dysfunction seen with insulin resistance and that other mechanisms, such as altered localization, reduced availability of cofactors, substrates, and the elevated production of O2, may be responsible.

  • syndrome X
  • Zucker
  • obesity

insulin resistance, a prediabetic condition associated with obesity, is increasing at an alarming rate in Western cultures (26, 30). Diabetes increases the incidence of many vasculopathies, including atherosclerosis, microvascular disease, and poor wound healing (44, 47). Of even greater concern is the mounting evidence that insulin-resistant states, independent of frank diabetes, also promote vascular dysfunction (43, 52). Thus the effect of insulin resistance on vascular signaling mechanisms is an area of increasing scrutiny.

Insulin-resistant states and diabetes are associated with reduced endothelium-dependent relaxation, and this dysfunction of the endothelium is highly correlated with future cardiovascular events (1, 14, 34, 49). In the obese Zucker rat (OZR), we previously showed that endothelium-dependent relaxation is impaired (14); however, the mechanisms underlying reduced synthesis or action of nitric oxide (NO) remain to be established.

Endothelium-dependent relaxation is mediated by the synthesis and subsequent diffusion of NO from the endothelium to the underlying smooth muscle (21, 35). Of the three NO synthases (NOS), the endothelial isoform (eNOS) is uniquely positioned within the vascular endothelium and alone is responsible for endothelium-dependent NO-mediated relaxation (20). The amount of NO produced in response to a given stimulus is tightly regulated by transcriptional and posttranscriptional events. Aside from chronic changes in gene expression, the activity of eNOS is acutely regulated by posttranslational modifications, including protein-protein interactions and phosphorylation (16). In endothelial cells, eNOS is dynamically bound to the molecular chaperone heat shock protein (HSP) 90, caveolin-1, and dynamin-2 and is phosphorylated on serine, threonine, and tyrosine residues in a stimulus-dependent manner (11, 16).

The relation between insulin and NO production has been examined in several models, many of which demonstrate that insulin promotes NO-mediated vasodilation and increases the expression of eNOS in cultured cells (6, 28, 39, 46). This led to the expectation that insulin-resistant states would be associated with reduced eNOS expression, and, indeed, this has been shown in some studies (28). A limitation of previous studies, however, has been that eNOS expression has been evaluated by mRNA levels (Northern blot and PCR). Furthermore, changes in gene expression can only chronically influence the amount of NO produced by endothelial cells, and the impact of the insulin-resistant state on the acute posttranslation regulation of eNOS has yet to be studied.

The goal of the present study was to test the hypothesis that eNOS signaling, as assayed by eNOS expression levels, phosphorylation, and interaction with HSP-90, is reduced in insulin-resistant states. Three insulin-resistant animal models were studied: the OZR, the fructose-fed rat, and the ob/ob mouse. These models exhibit insulin resistance in its most common context (obesity and leptin resistance in OZR), in the absence of obesity (fructose-fed rat), and in the absence of leptin resistance (ob/ob mouse). The expression levels of three NOS isoforms were measured: inducible NOS (iNOS), eNOS, and neuronal NOS (nNOS). Three vascular compartments were analyzed: heart, aorta, and hindlimb musculature. The acute regulation of eNOS was monitored through association with HSP-90 and protein phosphorylation of T494, S632, and S1176. Finally, an in vitro model of insulin resistance was induced in endothelial cells via adenoviral expression of protein tyrosine phosphatase (PTP) 1B (51). Regardless of the model, eNOS expression, phosphorylation, or binding to HSP-90 was undiminished. Collateral changes in iNOS or nNOS were also not observed.

Taken together, these data provide compelling evidence that aberrations in the expression, protein-protein interactions, and phosphorylation of eNOS are not a major defect in insulin-resistant states. These findings suggest that frank diabetes, with the complete loss of insulin, may be required to reduce NOS expression and that the impairments of NO-mediated responses in insulin-resistant states may reflect conditions of altered NO bioavailability or alterations in the target pathways for NO.


Animal models.

All animals were used in accordance with the guidelines for animal use of the Institutional Animal Care and Use Committee of the Medical College of Georgia, an American Association of Laboratory Animal Care-accredited animal facility. Lean Zucker rats (LZR) and OZR were obtained from Harlan (Indianapolis, IN) at 6 wk of age and maintained on standard chow ad libitum until they were used for study. OZR (fa/fa) possess a missense mutation in the leptin receptor (G269P) (48) that renders the receptor dysfunctional (5). The fa is an autosomal recessive locus, and heterozygotes (LZR) do not express the fa phenotype. On the day of death, each animal was sedated with isoflurane anesthesia. A catheter was inserted into the jugular vein, and a glucose tolerance test was performed by measuring the clearance of a 60-mg glucose bolus over 30 min. Plasma samples were acquired for plasma cholesterol, triglyceride, and insulin. Animals were then killed with an overdose of pentobarbital sodium anesthesia (75 mg/kg), and the heart, aorta, and a portion of hindlimb skeletal muscle were quickly removed and snap frozen in liquid nitrogen. To reduce the confounding influence of obesity in leptin-resistant animals, young, minimally obese Zucker rats were also studied. Insulin resistance independent of obesity was modeled by feeding a 66% fructose diet (Harlan Teklad) to LZR for 8 wk (33). Insulin resistance independent of leptin resistance was modeled using ob/ob mice (Jackson Laboratories) and tissue samples (a generous gift from Mark Hamrick, Medical College of Georgia).

Plasma analysis.

Plasma glucose was measured with a standard glucometer (Precision Xtra) and expressed as milligrams per deciliter. Plasma total cholesterol and triglycerides were measured by individual kits (WakoUSA, Richmond, VA). Rat plasma insulin was assayed by rat-specific enzyme immunoassay (Alpco, Windham, NH). Plasma NOx (NO2 + NO3) levels were determined by NO-specific chemiluminescence, as described previously (15). Briefly, plasma proteins were precipitated with ethanol, and NO3 was quantitatively reduced to NO2 with nitrate reductase. NO2 was then measured from plasma samples and endothelial cell media via conversion to NO in an NO analyzer (model 280i, Ionics Instruments, Boulder, CO).

Determination of endothelium-dependent vasodilation.

Endothelium-dependent vasodilation was assessed using the isolated gracilis artery preparation, as described previously (14). Briefly, gracilis arteries were isolated from the hindlimb of LZR and OZR, mounted between two glass pipettes, and pressurized at 60 mmHg. Vessels generated 15–20% baseline tone, which was further increased with 10−6 M phenylephrine to a final diameter that was 50% of passive diameter. Endothelium-dependent vasodilation to acetylcholine is ∼100% sensitive to nitro-l-arginine methyl ester in this preparation (14) and was used as an index of NO-mediated dilation. Serial doses of acetylcholine from 10−7 to 3 × 10−5 M were applied to the bath, and diameter was recorded during steady-state conditions. Data are expressed as percent increase in preconstricted diameter.

Western blotting.

Blood vessels and tissue samples were snap frozen in liquid nitrogen immediately after extraction and pulverized as described previously. Proteins were solubilized in a lysis buffer containing 1% Nonidet P-40, 0.1% SDS, 0.1% deoxycholate, 50 mM Tris·HCl, 0.1 mM EGTA, 0.1 mM EDTA, 5 mM sodium fluoride, 1 mM sodium pyrophosphate, 1 mM sodium vanadate, 1 mmol 4-(2-aminoethyl)-benzenesulfonyl fluoride, and a protease inhibitor cocktail tablet (Roche Diagnostics, Mannheim, Germany), pH 7.5. Insoluble protein was removed by centrifugation, and protein content in the supernatants was quantified using the Lowry method with bovine serum albumin as the standard. For detection of phosphorylated eNOS species, eNOS was affinity purified from cell lysates using 2′,5′-ADP-Sepharose (Pharmacia). Proteins were size fractionated by SDS-PAGE and immunoblotted with antibodies to eNOS (BD Biosciences); phosphorylated eNOS (S1179; Zymed laboratories); T497 (Upstate); S635 (Bruce Kemp, Melbourne, Australia); iNOS, nNOS, insulin receptor (IR), PTP-1B, and HSP-90 (BD Biosciences); phosphorylated Akt (Cell Signaling); and phosphorylated IR (Biosource). The intensities of the bands corresponding to the proteins of interest were measured using densitometry.

To assess interactions between eNOS and HSP-90, cell lysates were precleared with protein A/G-agarose (sc-2003, Santa Cruz Biotechnology). Anti-eNOS (N30020, Transduction Laboratories) was incubated overnight at 4°C. Protein A/G-agarose was subsequently added, and samples were incubated for an additional 3 h at 4°C. Immunoprecipitated proteins bound to the agarose beads were washed twice, and the samples were boiled in SDS sample buffer to elute the proteins from the beads. Agarose beads were pelleted by centrifugation, and protein supernatants were used for immunoblotting as described above.

Cell culture.

Bovine aortic endothelial cells (BAECs, passage 2–4) were cultured in medium 199 containing 10% (vol/vol) iron-supplemented FCS, 5% (vol/vol) FCS, 100 IU/ml penicillin, and 100 μg/ml streptomycin. NO release from endothelial cells was determined as previously described (15).

Real-time PCR.

Real-time RT-PCR was performed using the SmartCycler system (Cepheid). Plasmids containing fragments of rat eNOS (accession no. AF085195), HSP-90 (S45392), and Tie2 (AF030423) were generated via RT-PCR of total RNA from rat heart as described below and cloned into pcDNA3 (Invitrogen). Primers were as follows: 5′-agcccgggacttcatcaatcag-3′ (sense) and 5′-gccccaaacaccagctcactctc-3′ (antisense) for eNOS, 5′-aggcccaaaacccactcca-3′ (sense) and 5′-ggcatcctcatcaccctccag-3′ (antisense) for HSP-90, and 5′-cgtgctattggcgtttctgatt-3′ (sense) and 5′-gattgtttttggccttcctgttta-3′ (antisense) for Tie2. Total RNA (500 ng) was reversed transcribed (Superscript II, Invitrogen), and 10% of the reaction mix was combined with Omnimix Taq reagent (Cepheid), primers, and SYBR-green intercalating fluorescent dye. The PCR cycle was as follows: denaturation at 94°C for 15 s, annealing at 55°C for 30 s, and extension at 72°C for 10 s, with optics turned on. The SmartCycler system quantifies the degree of fluorescence generated by the accumulating SYBR-green with each cycle of PCR. The cycle threshold (Ct) was the number of PCR cycles required for the fluorescent signal to rise above background. Standard curves were generated from predetermined concentrations of control plasmids, and comparative assessment of Ct from samples run in parallel was used to determine the relative abundance of each gene.

RNA isolation.

Frozen tissue was pulverized in a mortar and pestle cooled in liquid nitrogen. The resulting homogenate, still frozen, was immersed in TRIzol reagent (Invitrogen) and homogenized with a Polytron homogenizer. RNA was isolated by 100% isopropanol precipitation and washed twice with 75% ethanol. Quantification of RNA was performed by spectrophotometry, and purity and integrity of 16S and 28S ribosomal RNA was assessed by minigel electrophoresis. RNA (500 ng) was reverse transcribed with a commercial RT kit (Superscript II), and resultant cDNA was used for real-time RT-PCR or generation of plasmids containing eNOS, HSP-90, and Tie2 fragments, as described above.

Generation of PTP-1B adenovirus.

The gene encoding PTP-1B was isolated from human umbilical vein endothelial cells by RT-PCR. Briefly, total RNA was extracted from human umbilical vein endothelial cells using TRIzol. cDNA was synthesized from oligo(dT)-primed total RNA (1 μg) using Superscript II. The 1.3-kb gene for PTP-1B was amplified using primers [accession no. M31724; 5′ atggagatggaaaaggagttcg-3′ (sense) and 5′-ctatgtgttgctgttgaacagg-3′(antisense)] and subcloned into the adenoviral shuttle vector pAdtrack (19). Subsequent clones were verified for integrity via bidirectional DNA sequencing and protein expression. Replication-deficient adenoviruses expressing PTP-1B or green fluorescent protein (GFP), under the control of the cytomegalovirus (CMV) promoter, were generated using the pAdTrack-CMV vector and AdEasy system. Viruses were amplified in HEK-293 cells, purified using CsCl, titered using a cytopathic effect assay, and stored in PBS containing 5% sucrose and 2 mM MgCl. BAECs were infected with PTP-1B and control (GFP) adenoviruses at multiplicity of infection of 50, and 24–72 h later, cells were assayed for insulin resistance and eNOS protein expression.

Chemical reagents.

All buffer reagents, enzymes, and glucose were acquired from Sigma.


Values are means ± SE. Statistical differences were analyzed using Student's t-test or ANOVA followed by Bonferroni's multiple comparison test. Statistical significance was considered when P < 0.05.


Metabolic parameters.

The baseline characteristics of young and adult LZR and OZR and lean Zucker rats fed a fructose diet are summarized in Table 1. Insulin-resistant rats (young and adult OZR and fructose-fed LZR) show typical patterns of hyperinsulinemia and dsylipidemia associated with insulin resistance. Fructose-fed rats displayed these changes with no difference in body weight compared with LZR. Young OZR were also hyperinsulinemic at a significantly lower level of obesity (44 and 68% in young and adult rats, respectively). Finally, fasting levels of blood glucose were comparable in all rats, documenting insulin resistance, but not frank hyperglycemia, in the models chosen.

View this table:
Table 1.

Baseline characteristics of young and adult LZR and OZR

Endothelium-dependent relaxation and plasma NOx.

Assessment of endothelium-dependent dilation is shown in Fig. 1A. As reported previously (14), acetylcholine caused dose-dependent increases in vascular diameter in gracilis from LZR, reaching a maximal effect of 84 ± 4%. In contrast, acetylcholine-induced dilation was blunted in gracilis arteries from OZR, reaching only 43 ± 7% response (P < 0.05, n = 6). This finding confirms the existence of impaired NO-mediated vasodilation in OZR. As a corollary, we determined plasma NOx as an index of in vivo NO production. Plasma levels of NO2 were significantly less in OZR than in LZR (Fig 1B). NO3 and total NOx species were reduced in OZR vs. LZR but did not achieve statistical significance.

Fig. 1.

Impaired endothelium-dependent vasodilation and reduced plasma NO2 in obese Zucker rats (OZR). A: pressurized gracilis arteries (60 mmHg) from lean Zucker rats (LZR) and OZR were preconstricted with 10−6 M phenylephrine, and endothelium-dependent vasodilation to 10−7–3 × 10−5 M ACh was obtained. Data are expressed as percent increase in preconstricted diameter (n = 5). *P < 0.05. B: NO2, NO3, and NO2 + NO3 (NOx) in plasma from LZR and OZR (n = 12). ns, Not significant.

eNOS protein expression.

Expression of eNOS protein is documented in Fig. 2. Western blot analysis of eNOS protein from the hearts of young (6–8 wk old, Fig. 2A) and adult (10–25 wk old, Fig. 2B) Zucker rats shows no difference in the levels of expression between obese and lean rats. Consistent with these findings, no significant difference was observed in two other vascular tissues: aorta (Fig. 2C) and skeletal muscle (Fig. 2D). We next determined the level of eNOS expression in two other models of insulin resistance: eNOS was affinity purified from the hearts of fructose-fed rats and C57BL/6J (control) and obese (ob/ob) mice. The relative expression of eNOS was not significantly different between ob/ob mice (Fig. 3A) and control (LZR) and fructose-fed rats (Fig. 3B). These models, in which insulin resistance is independent of leptin resistance (in both cases) and obesity (in the former case), were without differences in the level of eNOS protein expression.

Fig. 2.

Expression of endothelial nitric oxide (NO) synthase (eNOS) is unchanged in vasculature of OZR. eNOS was affinity purified from heart and skeletal muscle of LZR and OZR and, together with aortic cell lysates, was immunblotted for total eNOS. Densitometric values obtained for individual eNOS bands are presented along with representative immunoblots. A and B: eNOS expression levels in hearts of young (A; 6–8 wk old, n = 4) and adult (B; 10–25 wk old, n = 4) OZR and LZR. C and D: aortic (C) and skeletal muscle (D) eNOS in LZR and OZR (n = 4).

Fig. 3.

eNOS expression is not diminished in other in vivo models of insulin resistance. eNOS was affinity purified from hearts of ob/ob mice (A; n = 4) and fructose-fed rats (B; n = 4) and immunoblotted for total eNOS.

eNOS mRNA expression.

Expression of eNOS mRNA is documented in Fig. 4. Total RNA was extracted from the hearts of LZR and OZR, and the integrity of ribosomal 28S/18S is shown in Fig. 4A. Optimal PCR primers were designed for rat eNOS, and the sensitivity and selectivity of these probes are shown in Fig. 4B, with the amplification of as little as 1 pg of eNOS cDNA. Quantitative assessment of eNOS mRNA levels, as determined by real-time PCR, reveals no significant difference between LZR and OZR (Fig. 4C). These findings are consistent, whether normalized to the abundant and ubiquitous HSP-90 or to the endothelial cell-specific Tie2 gene.

Fig. 4.

eNOS mRNA levels are not different between LZR and OZR. A: integrity of ribosomal RNA from total RNA extracted from hearts of LZR and OZR. B: primers designed to amplify rat eNOS were specific and highly sensitive. C: quantitative comparison of eNOS mRNA levels in hearts of LZR and OZR. Level of eNOS mRNA was not significantly different between groups whether normalized to heat shock protein 90 (HSP-90, n = 8) or endothelial cell-specific Tie2 (data not shown). There was no significant difference in expression of Tie2 between OZR and LZR. Ct, cycle threshold.

Expression of other NOS isoforms.

In addition to eNOS, NO is also produced by iNOS and nNOS. The relative levels of expression of iNOS and nNOS were not significantly different in cardiac tissue between LZR and OZR (Fig. 5, A and B). Furthermore, expression of HSP-90, which we employed as a housekeeping gene, was not different between LZR and OZR (Fig. 5C).

Fig. 5.

Cardiac expression of neuronal and inducible NOS isoforms (nNOS and iNOS, respectively) is unchanged in OZR. NOS was affinity purified from hearts of lean LZR and OZR and immunoblotted with antibodies that recognize rat iNOS (A; n = 4), nNOS (B; n = 4), or the abundantly expressed HSP-90 (C; n = 4)

Insulin resistance in endothelial cells in vitro.

In a fourth model of insulin resistance, we used adenovirus-mediated gene delivery to endothelial cells in culture to overexpress PTP-1B. This model excludes the possible influence of hemodynamic forces, hyperglycemia, and dyslipidemia on eNOS gene expression. PTP-1B specifically dephosphorylates and, thus, reduces the activity of the insulin receptor and is believed to be upregulated in states of insulin resistance (4). Confluent BAECs were transduced with adenoviruses encoding GFP (control) or PTP-1B. Cells were allowed to recover for 72 h and then immunoblotted for expression of total eNOS. The insulin receptor was dephosphorylated in endothelial cells transduced with the PTP-1B adenovirus vs. control (GFP) cells (Fig. 6A). In addition, in response to insulin, the activation of Akt, as indicated by phosphorylation of S473, was abolished in cells overexpressing PTP-1B (Fig. 6B). Having demonstrated an inability of these cells to respond to insulin, we next looked at the expression of eNOS. In endothelial cells treated with insulin for 72 h, eNOS expression levels were not significantly different between GFP- and PTP-1B-overexpressing cells (Fig. 6A). To determine the functional relevance of insulin resistance in endothelial cells, we measured the release of NO via its stable end product in aqueous media, NO2. In endothelial cells transduced with PTP-1B, the ability of insulin to acutely stimulate NO release was abolished relative to control (GFP-transduced) cells (Fig. 6C). Overall capacity to produce NO was unaffected, inasmuch as the ability of ionomycin to directly stimulate NO was equivalent in both populations of cells (Fig. 6C, right).

Fig. 6.

No change in eNOS expression in an in vitro model of insulin resistance. Endothelial cells were transduced with adenoviruses (multiplicity of infection = 50) expressing green fluorescent protein (GFP, viral control) or protein tyrosine phosphatase (PTP) 1B, which specifically dephosphorylates the insulin receptor (IR). A: PTP-1B overexpression dephosphorylates the insulin receptor but does not reduce eNOS protein expression. B: insulin (100 nM) and vascular endothelial growth factor (VEGF, 50 ng/ml, 5 min) stimulate phosphorylation of the downstream effector Akt in GFP-treated endothelial cells, but insulin is unable to activate Akt in PTP-1B-overexpressing cells. Con, control. C: bovine aortic endothelial cells transduced (multiplicity of infection = 50) with adenovirus (Ad) encoding GFP or PTP-1B were stimulated with insulin (100 nM) or ionomycin (1 μM). Ability of insulin to liberate NO is abolished in PTP-1B-overexpressing cells, whereas ionomycin generates equivalent levels of NO (n = 6). *P < 0.05.

Posttranslational modification of eNOS.

The activity of eNOS is highly regulated by posttranslational modifications, including phosphorylation and protein-protein interactions. The phosphorylation of eNOS at human T495 [T495(hum)] is reported to inhibit enzyme activity and increase O2 formation. In contrast, phosphorylation of S1177(hum) and S633(hum) is associated with increased enzyme activity. We next examined whether the level of eNOS phosphorylation at T494, S632, and S1176 was different in OZR and LZR (Fig. 7). The level of eNOS phosphorylation at S1176 normalized to total eNOS is significantly increased in hearts of OZR vs. LZR (Fig. 7A). In contrast, in the other two models of insulin resistance, the fructose-fed rat and the ob/ob mouse, there was no significant difference in S1176 phosphorylation (Fig. 8). The degree of phosphorylation at the other regulatory sites, T494 and S632, was not significantly different between groups (Fig. 7, B and C).

Fig. 7.

Posttranslational phosphorylation of eNOS is not reduced in OZR. eNOS was affinity purified from hearts of LZR and OZR and immunoblotted with phosphorylation state-specific antibodies to S1176 (equivalent to human S1177), S632 (human S633), and T494 (human T495). A: level of S1176 phosphorylation is significantly higher in OZR than in LZR (n = 7). Degree of T494 (B; n = 4) and S632 (C; n = 4) phosphorylation was not different between groups.

Fig. 8.

Phosphorylation of S1176 is unchanged in ob/ob mice and fructose-fed rats. eNOS was affinity purified from hearts of ob/ob mice (A; n = 4) and fructose-fed rats (B; n = 4) and immunoblotted with antibodies to phosphorylated S1176 (human S1177).

The protein-protein interaction of HSP-90 with eNOS acutely regulates eNOS activity and can facilitate phosphorylation of S1177(hum) (13, 17). Reduced binding of HSP-90 to eNOS can produce an “uncoupled” enzyme, whereby the diminished synthesis of NO is exaggerated by an increased production of O2 (37). To determine the extent of HSP-90 association, eNOS was immunoprecipitated from the hearts of LZR and OZR, and the amount of bound HSP-90 was assessed by immunoblot. There is no significant difference between LZR and OZR in the amount of HSP-90 bound to eNOS (Fig. 9).

Fig. 9.

Protein-protein interactions between eNOS and HSP-90 are unchanged in OZR and LZR. eNOS was immunoprecipitated (IP) from hearts of LZR and OZR and immunoblotted for eNOS and bound HSP-90 (n = 3).


Insulin-resistant states represent an emerging risk factor in the development of vascular disease. Because many of the vasculopathies accelerated by insulin resistance and diabetes center around NO, we hypothesized that induction of insulin resistance would abrogate the expression or regulation of enzymes that generate NO. To test this hypothesis, we have employed four different models of insulin resistance, in vivo and in vitro. In OZR, endothelium-dependent vasodilation is significantly impaired and plasma levels of NO metabolites are reduced. These data suggest that the ability of eNOS to deliver biologically active NO to target tissues is compromised in OZR. However, eNOS protein levels are not significantly different in young vs. adult animals or across several tissues, including heart, aorta, and skeletal muscle. In addition, no significant differences in eNOS expression were found in other models of insulin resistance, namely, ob/ob mice, fructose-fed rats, and PTP-1B-overexpressing endothelial cells vs. their respective controls. Our present study is the first to examine whether insulin resistance influences the posttranslational regulation of eNOS. In OZR, the phosphorylation states of T494 [T495(hum)] and S632 [S633(hum)] are not significantly different, whereas the phosphorylation of S1176 [S1177(hum)] is slightly increased. The protein-protein interaction between eNOS and HSP-90 was also unaffected by insulin resistance. Thus it appears that impaired endothelium-dependent relaxation or NO signaling in states of insulin resistance is not due to deficits in the synthesis of NO. It is therefore likely that other factors, such as the generation of O2-derived free radicals, changes in localization, or constraining concentrations of cofactors or substrates, are contributing to the dysfunction.

The study of insulin resistance is complicated by the fact that, in the clinical setting, the condition rarely presents alone but, most commonly, occurs in the context of obesity and the metabolic syndrome X. These coincident risk factors may obscure the effects of insulin resistance as an independent variable on vascular function. In this study, we have addressed this issue by comparing four different strategies to invoke insulin resistance. We first used the traditional OZR model, which develops insulin resistance secondary to leptin resistance and obesity (3, 24). This model most closely describes the insulin resistance that manifests in patient populations. A common criticism of this model is that the competing effects of obesity and hypertension in these animals obscure the direct consequences of insulin resistance. To address this, we also used the fructose-fed rat model, which develops insulin resistance secondary to elevated sugar consumption, allowing the examination of insulin resistance without the potentially confounding effects of obesity (25). Third, we employed the ob/ob mouse (22), which presents with obesity and insulin resistance, but not the leptin resistance in OZR. A final model of insulin resistance was generated in vitro in endothelial cells and excludes the in vivo complications of metabolic, hormonal, and hemodynamic forces. In our present study, none of these animals or cells displayed reductions in eNOS expression or regulation. This finding provides strong evidence that eNOS is resistant to detrimental effects of insulin resistance, regardless of how insulin resistance is presented.

An interdependent relation exists between insulin and eNOS signaling. eNOS-derived NO has been proposed to facilitate glucose uptake into liver and skeletal muscle, a phenomenon emphasized by the insulin-resistant phenotype of mice with homozygous deletion of eNOS (7, 41). Conversely, insulin has been proposed as a major regulator of eNOS protein expression. In animal models of insulin resistance and specific deletion of endothelial insulin receptors, reduced expression of eNOS mRNA has been reported (28, 50). Exogenous insulin has also been shown to increase eNOS expression (9, 28). However, in our present study, we find that insulin resistance does not decrease eNOS expression. Although the majority of studies reporting a decrease have investigated only eNOS mRNA levels (28, 50), the differences between our study and others showing reduced levels of eNOS protein are not readily explained. In the present study, we have used four different models of insulin resistance and three different vascular beds and have measured mRNA and protein levels without significant differences. More recent studies in ob/ob mice, OZR, and fructose-fed rats support our findings and have not observed a reduction in eNOS protein or mRNA expression (27, 34, 40). Another possibility is that the local concentration of eNOS may be altered. A recent study using diabetic rats showed no difference in the level of eNOS protein and mRNA isolated from total hearts; however, reduced levels of eNOS were found in select vessels via immunohistochemistry (10).

Reduced endothelium-dependent relaxation is a hallmark of insulin resistance; yet very little is known about the consequences of insulin resistance in the acute, posttranslational regulation of eNOS. In addition to calcium-calmodulin, eNOS activity is influenced by its subcellular localization, phosphorylation state, and protein-protein interactions (11, 16). In isolated activity assays, the attenuated function of eNOS extracted from the tissues of fructose-fed animals hints at compromised posttranslational regulation (31, 42). However, such assays do not account for changes in subcellular localization or the intracellular milieu, i.e., reduced levels of cofactors or substrates. Phosphorylation of specific sites on eNOS has been proposed to acutely regulate eNOS activity positively and negatively. Phosphorylation of bovine eNOS has been reported on S116, T497, S617, S635, and S1179 (−2 aa in the human gene and −3 aa in the rat and mouse) (11). Additionally, rat eNOS contains a substitution at T113, equivalent to S114 on the human gene. The phosphorylation of S1176 [S1177(hum)] occurs in a time frame consistent with NO release and increases the calcium sensitivity and specific activity of the enzyme. Phosphorylation of this residue is stimulated by a variety of agonists, including shear stress, and is dependent on the correct subcellular targeting of the enzyme to peripheral aspects of the Golgi and plasma membrane (16). In OZR, we observed an increase in S1176 [S1177(hum)] phosphorylation vs. control. The exact mechanisms underlying the increase in phosphorylation are not known. However, no significant increase in phosphorylation was observed in the two other animal models of insulin resistance, suggesting that other perturbations in OZR mediate the increase in phosphorylation. In addition, other studies have shown that high glucose does not affect eNOS phosphorylation at this residue (38) or results in decreased phosphorylation via O-linked glycosylation (8). We next addressed whether insulin resistance influences phosphorylation of eNOS at other residues. Phosphorylation of T495 reduces eNOS activity through reduced calmodulin association and increased O2 production (12, 18, 29). S633 phosphorylation of eNOS increases eNOS activity in cultured cells and in enzyme activity assays (2, 32). There was no significant difference in the level of eNOS PO4 at T494 [T495(hum)] or S632 [S633(hum)] in OZR vs. lean controls. Taken together, there is no evidence to suggest that alterations in eNOS phosphorylation mediate the reduced activity of eNOS observed in OZR.

The reduced association of eNOS with HSP-90 underlies the impaired vascular responses in hypercholesterolemia (36) and results in an uncoupled eNOS that produces excess O2 in preference to NO (37). Little is known about the impact of insulin resistance on the balance of HSP-90-eNOS interactions. Recently, Stalker et al. (45) described reduced association of HSP-90 and eNOS in a rat model of hyperglycemia. However, in our studies, we did not observe a decrease in HSP-90 binding, suggesting that insulin resistance in itself may be insufficient and that frank diabetes may be required to disrupt the eNOS-HSP-90 complex.

Having found no perturbations in the expression or posttranslational regulation of eNOS, we were surprised that insulin resistance could, under the right conditions, reduce NO release. In isolated endothelial cells that overexpress PTP-1B, which selectively negates insulin signaling, the ability of insulin to produce NO was abolished (Fig. 6C). However, in response to other stimuli, the overall capacity to generate NO was not affected. These data demonstrate that the state of insulin resistance in endothelial cells per se does not necessarily reduce the capacity of these cells to produce NO. In vivo, the overall contribution of insulin to NO release is not known, but it is likely to be overshadowed by other factors, such as shear stress, bradykinin, and other humoral agents. In plasma samples from OZR, there was a significant reduction in the level of NO2 relative to plasma samples from LZR, suggesting reduced in vivo synthesis of NO in OZR. However, the ability of NO2 to track in vivo NO production is controversial, and NO3, because of its longer half-life, is reputedly a better indicator of in vivo NO production (53). The level of NO3 in plasma from OZR was reduced but not significantly different from that in plasma from LZR. One possible explanation for these findings is that the level of NO3 in OZR may be artificially elevated because of excessive food consumption (23).

On the basis of the evidence presented in these studies, impaired endothelium-dependent relaxation in OZR cannot be attributed to decreased expression or impaired posttranslational modification of eNOS. However, we do observe reduced NO2 in plasma samples of OZR, suggesting that the in vivo generation of NO is compromised. Previous studies have shown that antioxidants can partially restore endothelium-dependent relaxation in the Zucker rat (14), suggesting that additional mechanisms are responsible for the vascular dysfunction. The source of the excessive intravascular production of O2 in the OZR is not known. However, in light of our present data showing no change in HSP-90 association or phosphorylation of T494 [T495(hum)], it is unlikely to derive from “uncoupled” eNOS. Additional mechanisms underlying vascular dysfunction include reduced levels of tetrahydrobiopterin or upregulation of NADPH oxidases (NOXs). Future studies to identify the source of intravascular O2 may provide an avenue for therapeutic intervention and, eventually, improvement in vascular function in diabetes and insulin resistance.


This work was supported by National Heart, Lung, and Blood Institute Grants R01 HL-074279-01 (D. Fulton) and R01 HL-067303 (D. W. Stepp) and American Heart Association Grants SDG 0330196N (D. Fulton) and SDG 0030370Z (D. W. Stepp).


We thank James Mintz and Julie Campbell for technical assistance.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract