Acute O-GlcNAcylation prevents inflammation-induced vascular dysfunction

Abstract

Acute increases in cellular protein O-linked N-acetyl-glucosamine (O-GlcNAc) modification (O-GlcNAcylation) have been shown to have protective effects in the heart and vasculature. We hypothesized that d-glucosamine (d-GlcN) and Thiamet-G, two agents that increase protein O-GlcNAcylation via different mechanisms, inhibit TNF-α-induced oxidative stress and vascular dysfunction by suppressing inducible nitric oxide (NO) synthase (iNOS) expression. Rat aortic rings were incubated for 3h at 37°C with d-GlcN or its osmotic control l-glucose (l-Glc) or with Thiamet-G or its vehicle control (H₂O) followed by the addition of TNF-α or vehicle (H₂O) for 21 h. After incubation, rings were mounted in a myograph to assess arterial reactivity. Twenty-four hours of incubation of aortic rings with TNF-α resulted in 1) a hypocontractility to 60 mM K⁺ solution and phenylephrine, 2) blunted endothelium-dependent relaxation responses to ACh and substance P, and 3) unaltered relaxing response to the Ca²⁺ ionophore A-23187 and the NO donor sodium nitroprusside compared with aortic rings cultured in the absence of TNF-α. d-GlcN and Thiamet-G pretreatment suppressed the TNF-α-induced hypocontractility and endothelial dysfunction. Total protein O-GlcNAc levels were significantly higher in aortic segments treated with d-GlcN or Thiamet-G compared with controls. Expression of iNOS protein was increased in TNF-α-treated rings, and this was attenuated by pretreatment with either d-GlcN or Thiamet-G. Dense immunostaining for nitrotyrosylated proteins was detected in the endothelium and media of the aortic wall, suggesting enhanced peroxynitrite production by iNOS. These findings demonstrate that acute increases in protein O-GlcNAcylation prevent TNF-α-induced vascular dysfunction, at least in part, via suppression of iNOS expression.

endothelial dysfunction is a hallmark for the onset and progression of a number of vascular pathologies related to systemic disorders such as atherosclerosis, hypertension, and diabetes. In arteries with dysfunctional endothelium, endothelium-dependent relaxation is usually impaired, favoring vasoconstriction and blood pressure elevation. The adaptive immune system plays a significant role in the development of endothelial dysfunction and the pathogenesis of hypertension (for a review, see Ref. 17). For example, activation of proinflammatory T lymphocytes can increase the production of TNF-α, and recombinant TNF-α has been shown to cause endothelial dysfunction in isolated and cultured arteries in the absence of blood cells (1, 22, 44, 50). TNF-α can increase the production of ROS, including superoxide anion and peroxynitrite, via upregulation of ROS-producing enzymes such as inducible nitric oxide (NO) synthase (iNOS) via the NF-κB transcription factor pathway (2, 14, 45). Enhanced NO scavenging by superoxide anion results in peroxynitrite formation and reduced NO bioavailability (9, 14). Peroxynitrite, formed by the reaction of NO with superoxide anion, reacts readily with proteins to form nitrotyrosine residues. Hence, an elevated nitrotyrosylated protein level is a marker for enhanced peroxynitrite formation and oxidative stress (21, 24, 37). The functional consequences of enhanced iNOS expression in the vasculature have also been well documented. Transfer of the iNOS gene into the rabbit carotid artery has been shown to inhibit both relaxation responses to ACh and contractile responses to phenylephrine (Phe) (16). Furthermore, administration of an antisense oligonucleotide to iNOS has been shown to prevent the impairment of vascular contraction induced by inflammatory stimuli, e.g., lipopolysaccharide (19).

Glucosamine (d-GlcN) is a hexosamine sugar that is metabolized by the hexosamine biosynthesis pathway and can stimulate O-linked N-acetyl-glucosamine (O-GlcNAc) modification (O-GlcNAcylation) of cytosolic and nuclear proteins (41). Levels of protein O-GlcNAcylation are regulated by O-GlcNAc transferase (OGT), which catalyzes the addition of a single UDP-GlcNAc moiety to the hydroxyl group of serine and threonine residues of a target protein, and β-N-acetylglucosaminidase (O-GlcNAcase or OGA), which catalyzes the hydrolytic cleavage of the O-linked sugar moiety from the protein. O-GlcNAcylation is a highly dynamic posttranslational modification that plays an important role in regulating nutrient- and stress-induced signal transduction pathways and many other biological processes (11, 18, 51). Increased O-GlcNAcylation by hyperglycemia has been implicated as a pathogenic contributor to insulin resistance and hence to the development of diabetes (32) and to augmented vascular contractility in response to endothelin-1 (26, 28) and DOCA-salt hypertension (27).

In contrast to these deleterious effects of increased O-GlcNAcylation, a growing body of evidence has demonstrated that acute and transient activation of O-GlcNAc levels is an endogenous stress response that is beneficial for cell survival, suggesting that short-term “fine balancing” of O-GlcNAc levels has protective and immunosuppressive effects on cellular function (4, 5, 11, 25). For example, in rodent models of trauma-hemorrhage, administration of d-GlcN during resuscitation resulted in increased protein O-GlcNAc levels that improved cardiac function and organ perfusion (48, 53). Our research group (46) recently demonstrated that treatment with either d-GlcN or the OGA inhibitor O-(2-acetamido-2-deoxy-d-glucopyranosylidene)amino-N-phenylcarbamate increased O-GlcNAcylated protein levels and inhibited infammatory responses in balloon-injured rat carotid arteries. Furthermore, d-GlcN has beneficial effects in experimental models of adjuvant and rheumatoid arthritis in rats (20) and in experimental encephalomyelitis in mice (53). In an in vivo rat model of endotoxin-induced inflammation, d-GlcN dose dependently suppressed iNOS expression in various organs and macrophages (33). Also, many in vitro studies (6, 7, 30) have shown that d-GlcN has anti-inflammatory properties.

Recently, we (47) have demonstrated that increased O-GlcNAcylation of NF-κB p65 induced by either d-GlcN or O-(2-acetamido-2-deoxy-d-glucopyranosylidene)amino-N-phenylcarbamate treatment inhibits TNF-α-induced NF-κB p65 signaling and inflammatory mediator expression in quiescent rat aortic smooth muscle cells. TNF-α has also been shown to induce the expression of iNOS protein via the NF-κB signaling pathway (2, 45). Based on the above findings, we tested the hypothesis that pretreatment with either d-GlcN or Thiamet-G, a recently developed selective and potent inhibitor of OGA (49), can inhibit TNF-α-induced vascular dysfunction in cultured rat aortic rings via suppression of oxidative stress markers, iNOS expression, and protein nitrotyrosylation.

MATERIALS AND METHODS

Animals.

Experiments were conducted in 12-wk-old male Sprague-Dawley rats (Charles River Breeding Laboratories). All procedures were approved by the Institutional Animal Care and Use Committee of the University of Alabama (Birmingham, AL) and were consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Isolation and culture of aortic rings.

Rats were killed by CO₂ inhalation, and the thoracic aorta was removed and placed in cold Krebs-Ringer buffer (KRB) with the following composition (in mM): 118.5 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, 25.0 NaHCO₃, and 5.5 d-glucose. The aorta was pinned down on a 90-mm glass petri dish coated with black Sylgard, soaked in cold KRB, carefully freed of adipose and connective tissue, and cut into four rings of 3 mm in length. Each ring was placed in a separate culture dish filled with 4 ml DMEM containing 120 U/ml penicillin, 120 μg/ml streptomycin, and 1% fetal calf serum (culture medium). The four culture dishes were subjected to four different treatments, which are described in Experimental protocols.

Myograph experiments.

After incubation, aortic rings were mounted between two stainless steel pins in a myograph chamber (Danish Myo Technology, Aarhus, Denmark) filled with 5 ml KRB solution, maintained at 37°C, and continuously aerated with 95% O₂-5% CO₂. Aortic rings were stretched to a passive force of 30 mN, equilibrated for 30 min, and depolarized with high-K⁺ KRB (K⁺ KHB; 60 mM KCl in KRB solution, replacing equimolar NaCl with KCl), thus generating a contraction, which was set to a 100% contraction level. After a 30-min washout period, a cumulative concentration-response curve (CRC) to Phe (0.001–10 μM) was performed. After a second 30-min washout period, aortic rings were contracted with a mixture of Phe (0.1–10 μM) and serotonin (0.1–3 μM) in such fashion that comparable contraction levels were achieved for all experimental groups. At a stable contraction, endothelium-dependent relaxations were performed by constructing a CRC to ACh (0.001–100 μM). In a subset of aortic rings, endothelium-dependent relaxations were assessed with the vasoactive peptide substance P (one single concentration of 1 μM) and the Ca²⁺ ionophore A-23187 (0.001–1 μM). Endothelium-independent relaxation was tested with the NO donor sodium nitroprusside (SNP; 0.001–10 μM). Aortic rings were incubated for 30 min with the NO synthase blocker N^ω-nitro-l-arginine methyl ester (l-NAME; 100 μM) to rule out any NO release induced by SNP.

Experimental protocols.

Pilot experiments were performed to optimize the duration of incubation and concentration of TNF-α needed to induce a significant impairment in ACh-induced endothelium-dependent relaxations. Based on these experiments, 24 h of incubation of aortic rings in culture medium did not significantly alter ACh-induced relaxations compared with freshly isolated and noncultured aortic rings. Three different concentrations of TNF-α (1, 10, and 100 ng/ml) were added to the aortic rings and incubated for 3, 6, or 21 h. We observed that 1 ng/ml TNF-α did not result in a significant rightward shift in the CRC to ACh at any incubation period, whereas 10 ng/ml TNF-α did result in a significant rightward shift in the CRC to ACh compared with vehicle (H₂O)-treated aortic rings that were incubated for 21 h but not for 3 and 6 h. Incubation of 100 ng/ml TNF-α for 21 h severely impaired endothelium-dependent ACh-induced relaxations. Hence, a concentration of 10 ng/ml TNF-α was chosen.

To study the potential protective effect of increased O-GlcNAcylation on TNF-α-induced endothelial dysfunction, we used two chemical agents known to increase overall cellular O-GlcNAcylation: d-GlcN and 1,2-dideoxy-2′-ethylamino-α-d-glucopyranoso-[2,1-d]-Δ2′-thiazoline (Thiamet-G), a potent OGA inhibitor. Aortic rings were pretreated for 3 h with either 5 mM d-GlcN or its osmotic control 5 mM l-Glc followed by an incubation with 10 ng/ml TNF-α or vehicle (H₂O; 4 μl) for an additional 21 h. Similarly, separate sets of rings were pretreated with Thiamet-G (0.1 μM) or vehicle (H₂O; 4 μl) followed by an incubation with 10 ng/ml TNF-α or vehicle for an additional 21 h. Thiamet-G was dissolved in DMSO at a stock solution of 10 mM and further diluted in H₂O to a concentration of 0.1 mM. Hence, the vehicle for Thiamet-G was H₂O.

Western blot analysis for O-GlcNAcylated proteins and iNOS.

Aortic segments (∼10 mm in length) were incubated in culture medium, treated with agents as described above, snap frozen in liquid nitrogen, and then stored at −80°C. Frozen segments were pulverized and lysed with T-PER lysis buffer in the presence of 1 mM PMSF, 1 mM sodium orthovanadate, and 1 μg/ml aprotinin, leupeptin, and pepstatin for 1 h on ice. After centrifugation, supernatants were collected, and the protein concentrations were determined by a Bio-Rad protein assay. Samples of 30 μg were separated by 10% SDS-PAGE and transferred by electroblotting onto a polyvinylidene diflouride membrane. Blots were incubated with a 1:5,000 dilution of anti-O-GlcNAc (CTD110.6, Covance) in 1% casein and PBS overnight at 4°C and then washed three times in PBS. Membranes were then incubated with a 1:10,000 dilution of horseradish peroxidase-conjugated goat anti-mouse IgM (Santa Cruz Biotechnology) in 1% casein and PBS for 1 h at room temperature. After a further wash in PBS, immunoblots were developed with enhanced chemiluminescence (Pierce), and the signal was recorded on X-ray film. Densitometric analysis was performed on the entire lane of each sample using Labworks Analysis Software (UVP), and the mean intensity was normalized to the control group. Protein loading was assessed by stripping the membranes and reprobing with anti-β-actin antibody (Sigma). In a similar fashion, immunoblots were incubated with a 1:1,000 dilution of anti-iNOS (M-19, Santa Cruz Biotechnology) in 1% casein and PBS overnight at 4°C and then washed three times with PBS. Membranes were then incubated with a 1:5,000 dilution of horseradish peroxidase-conjugated anti-rabbit IgG in 1% casein and PBS for 1 h at room temperature.

Immunohistochemical analysis.

To determine the level of nitrotyrosine expression, aortic rings pretreated (3 h) with either l-Glc, d-GlcN, or Thiamet-G followed by an incubation in the absence or presence of 10 ng/ml TNF-α for 21 h in DMEM (as described above) were fixed in 10% buffered formalin for 24 h, paraffin embedded, and sectioned for immunohistochemical analysis using a VECTASTAIN ABC kit (Vector Labs, Burlingame, CA). Sections were hydrated in PBS, and sites of endogenous peroxide were quenched in 3% H₂O₂ in absolute methanol (10 min at room temperature). Nonspecific background staining was blocked with 10% normal goat serum (1 h), and sections were incubated with the specific antibody against nitrotyrosine (Abcam, Cambridge, MA) diluted 1:300 in 0.5% BSA in PBS (16 h at 4°C). After an incubation in biotinylated secondary antibody (VECTASTAIN ABC kit, 30 min at room temperature), sections were treated with the enzyme conjugate (VECTASTAIN ABC kit, 30 min at room temperature) and placed in VECTOR NovaRED Peroxidase Substrate for 3 min at room temperature (with the development of the reaction product monitored under a microscope). Sections were then washed, counterstained with hematoxylin, and mounted. The chromogen produces a red reaction product at immunopositive sites, whereas cell nuclei stain blue to purple.

Drugs.

A-23187, ACh, d-GlcN, l-Glc, l-NAME, Phe, serotonin, SNP, and recombinant TNF-α were purchased from Sigma-Aldrich. Thiamet-G (SD Chemmolecules) was initially dissolved in DMSO and subsequently diluted in H₂O before use.

Data and statistical analysis.

Contractile responses to Phe are expressed as percentages of the maximal contractile response to K⁺ KRB. Relaxation responses are expressed as percentages of the maximal contractile response to Phe-serotonin. Individual CRCs were fitted to a nonlinear sigmoid regression curve (Graphpad Prism 5.0). Sensitivity (pEC₅₀) and maximal effect (E_max) are expressed as means ± SE. The statistical significance of effects and differences was determined using either one-way ANOVA (comparison of pEC₅₀ and E_max) or two-way ANOVA (comparison of CRCs). A Bonferroni post hoc test was used to compare multiple groups. P values of <0.05 were considered as statistically significant.

RESULTS

TNF-α (10 ng/ml) treatment resulted in a significant impairment in the contractile response to depolarizing solution (60 mM K⁺ KRB) in rat aortic rings that was unaffected by the addition of 5 mM l-Glc as an osmotic control. Maximal contractile force to 60 mM K⁺ KRB averaged 30.12 ± 1.76 mN in the absence of TNF-α and 23.47 ± 1.31 mN in the presence of TNF-α (P < 0.05). In contrast, pretreatment with 5 mM d-GlcN prevented the TNF-α-induced impairment in contractile response to the depolarizing solution. After d-GlcN treatment, maximal contractile force in response to 60 mM K⁺ was 34.76 ± 1.62 mN in the absence of TNF-α and 30.85 ± 2.04 mN in the presence of TNF-α. Similarly, TNF-α treatment resulted in the significant blunting of contractions to Phe (0.001–10 μM) in aortic rings that were pretreated with the osmotic control, 5 mM l-Glc (Fig. 1A). Sensitivity (pEC₅₀) to Phe averaged 6.95 ± 0.10 in the absence of TNF-α and 6.55 ± 0.19 in the presence of TNF-α (P < 0.05; Fig. 1A). Maximal contraction (expressed as a percentage of K⁺ KRB) to 10 μM Phe averaged 79 ± 5% in the absence of TNF-α and 45 ± 8% in the presence of TNF-α (P < 0.05; Fig. 1A). d-GlcN pretreatment prevented the TNF-α-induced reduction in sensitivity and maximal contraction in response to Phe (Fig. 1B). In d-GlcN treated rings, sensitivity to Phe averaged 6.95 ± 0.10 in the absence of TNF-α and 6.63 ± 0.11 in the presence of TNF-α (P = not significant; Fig. 1A), whereas maximal contractions to 10 μM Phe were comparable in the absence or presence of TNF-α (62 ± 6% vs. 65 ± 8%, respectively). Similarly, pretreatment with Thiamet-G (0.1 μM) prevented the TNF-α-induced impairment in contraction in response to Phe that was seen in rings pretreated with vehicle (Fig. 1, C and D).

Fig. 1.

Cumulative concentration-response curves to phenylephrine (Phe; 0.001–10 μM) in aortic rings pretreated for 3 h with 5 mM l-glucose (l-Glc; A), 5 mM d-glucosamine (d-GlcN; B), vehicle (H₂O; C), or 0.1 μM Thiamet-G (D) followed by the addition of either vehicle (−TNF-α) or 10 ng/ml TNF-α (+TNF-α) for 21 h. Data are expressed as means ± SE. *P < 0.05 via ANOVA.

TNF-α resulted in statistically significant impairment in ACh (0.001–100 μM)-induced endothelium-dependent relaxations in aortic rings pretreated with l-Glc (Fig. 2A). Sensitivity to ACh averaged 7.06 ± 0.04 in the absence of TNF-α and 6.44 ± 0.12 in the presence of TNF-α (P < 0.05). Maximal relaxation to 100 μM ACh averaged 89 ± 2% in the absence of TNF-α and 75 ± 6% in the presence of TNF-α (P < 0.05). The same significant rightward shift caused by TNF-α was observed in aortic rings pretreated with the enantiomer of l-Glc, 5 mM d-glucose (from 7.07 ± 0.5 to 6.61 ± 0.13, P < 0.05), or 5 mM d-Mannitol (from 7.16 ± 0.04 to 6.65 ± 0.13, P < 0.05). In contrast, d-GlcN suppressed the TNF-α-induced impairment in the relaxing responses to ACh (Fig. 2B). Sensitivity to ACh was not statistically significantly different in either the absence or presence of TNF-α in rat aortic rings treated with d-GlcN (pEC₅₀: 7.24 ± 0.07 vs. 6.97 ± 0.09, respectively). TNF-α resulted in significant reductions in sensitivity and maximal relaxation to ACh in vehicle (H₂O)-treated aortic rings compared with vehicle-treated rings in the absence TNF-α (P < 0.05; Fig. 2C). Sensitivity and maximal relaxation to 100 μM ACh averaged 7.42 ± 0.08 and 92 ± 3%, respectively, in the absence of TNF-α and 7.18 ± 0.07 and 75 ± 3%, respectively, in the presence of TNF-α. In contrast, in rings treated with Thiamet-G, similar relaxation responses to ACh were observed in the absence or presence of TNF-α (pEC₅₀: 7.28 ± 0.12 vs. 7.15 ± 0.06, respectively; Fig. 2D).

Fig. 2.

Cumulative concentration-response curves to ACh (0.001–100 μM) in Phe- and serotonin (5-HT)-contracted aortic rings pretreated for 3 h with 5 mM l-Glc (A), 5 mM d-GlcN (B), vehicle (H₂O; C), or 0.1 μM Thiamet-G (D) followed by the addition of either vehicle (−TNF-α) or 10 ng/ml TNF-α (+TNF-α) for 21 h. Data are expressed as means ± SE. *P < 0.05 via ANOVA.

To test whether the observed relaxation effects were specific for ACh, we examined the effects of the endothelium-dependent vasorelaxation peptide substance P and the endothelium-dependent Ca²⁺ ionophore A-23187. Substance P (1 μM) induced comparable relaxation responses in rings that were pretreated with either l-Glc or d-GlcN (32 ± 4% and 28 ± 4%, respectively; Fig. 3A). TNF-α blunted these responses in both groups, but the residual relaxation response was significantly greater in rings pretreated d-GlcN compared with the l-Glc osmotic control (14 ± 4% vs. 1 ± 1%, respectively, P < 0.05; Fig. 3A), suggesting that d-GlcN suppressed the TNF-α-induced endothelial dysfunction. The Ca²⁺ ionophore A-23187 (0.001–1 μM) resulted in relaxations at concentrations below 0.1 μM and contractions at concentrations above 0.1 μM (Fig. 3, B and C), and TNF-α did not impair responses to A-23187 in either l-Glc or d-GlcN-treated rings. Relaxation responses to the endothelium-independent NO donor SNP (0.0001–10 μM) were comparable for rings pretreated with l-Glc, d-GlcN, H₂O, or Thiamet-G in the absence or presence of TNF-α (data not shown).

Fig. 3.

A: relaxation response to a single concentration of substance P (SP; 1 μM) in Phe- and 5-HT-contracted aortic rings pretreated for 3 h with 5 mM l-Glc or 5 mM d-GlcN followed by the addition of either vehicle or 10 ng/ml TNF-α for 21 h. Data are expressed as means ± SE. *P < 0.05 vs. l-Glc or d-GlcN; #P < 0.05 vs. l-Glc + TNF-α. B and C: cumulative concentration-response curves to A-23187 (0.001–1 μM) in Phe- and 5-HT-contracted aortic rings pretreated for 3 h with 5 mM l-Glc (B) or 5 mM d-GlcN (C) followed by the addition of either vehicle (−TNF-α) or 10 ng/ml TNF-α (+TNF-α) for 21 h.

We hypothesized that the protective effects of d-GlcN and Thiamet-G on endothelial function were related to their ability to increase protein O-GlcNAcylation. Overall O-GlcNAcylated protein levels were significantly increased in aortic segments that had been incubated with d-GlcN and Thiamet-G for 24 h compared with their controls (l-Glc and H₂O, respectively; Fig. 4). No statistically significant differences in overall O-GlcNAcylated protein levels were observed between aortic segments pretreated with either l-Glc or H₂O (data not shown). TNF-α treatment had no effect on overall O-GlcNAc levels in any experimental group (Fig. 4).

Fig. 4.

Left: representative images of Western blots of overall cellular anti-O-linked N-acetyl-glucosamine (O-GlcNAc) levels in aortic segments pretreated for 3 h with 5 mM l-Glc or vehicle control (l-Glc-H₂O), 5 mM d-GlcN, or 0.1 μM Thiamet-G in the absence (−) or presence (+) of 10 ng/ml TNF-α for 21 h. Blots were probed with CTD 110.6 antibody and reprobed with antibody against β-actin. Right: the intensity of all bands per lane was measured, and the ratio of O-GlcNAc to β-actin was calculated. Data are expressed as means ± SE; numbers in parentheses are numbers of samples. *P < 0.05 vs. l-Glc-H₂O-treated aortic rings.

We next hypothesized that the increase in protein O-GlcNAcylation induced by d-GlcN and Thiamet-G would protect against TNF-α-induced expression of iNOS protein. Overnight incubation with TNF-α resulted in statistically significant upregulation of iNOS protein expression (normalized to β-actin) in segments pretreated with osmotic or vehicle control solutions (l-Glc or H₂O; Fig. 5). No statistically significant differences in iNOS protein levels were observed between aortic segments pretreated with either l-Glc or H₂O (data not shown). Pretreatment with both d-GlcN and Thiamet-G prevented the TNF-α-induced increase in iNOS expression (Fig. 5).

Fig. 5.

A and B: images of representative Western blots of inducible nitric oxide synthase (iNOS) in aortic segments pretreated for 3 h with 5 mM l-Glc or vehicle (Veh) control (l-Glc-H₂O), 5 mM d-GlcN (A and C), or 0.1 mM Thiamet-G (B and D) in the absence (−) or presence (+) of 10 ng/ml TNF-α for 21 h. Blots were reprobed with antibody against β-actin. C and D: the band intensity was measured, and the ratio of iNOS to β-actin was calculated. Data are expressed as means ± SE; numbers in parenthese are numbers of samples. *P < 0.05 vs. l-Glc-H₂O controls; #P < 0.05 vs. l-Glc-H₂O + TNF-α.

Finally, we tested whether the increased expression of iNOS resulted in elevated production of peroxynitrite. This free radical reacts readily with proteins to form nitrotyrosine residues. To test for increased oxidative stress, in the form of enhanced peroxynitrite generation, we performed immunohistochemical staining for nitrotyrosylated proteins on paraffin-embedded cross-sections of cultured rat aortic segments. Treatment with either vehicle (H₂O or l-Glc), d-GlcN, or Thiamet-G for 24 h in the absence of TNF-α did not result in detectible protein nitrotyrosylation in aortic rings (Fig. 6, left). Twenty-one hours of incubation of TNF-α after 3 h of pretreatment with vehicle resulted in an intense red staining in endothelial and smooth muscle cells, indicating nitrotyrosylated proteins. Nitrotyrosylated protein staining was markedly reduced in rat aortic rings pretreated with either d-GlcN or Thiamet-G (Fig. 6, right). Taken together, these findings support our hypothesis that TNF-α-induced vascular dysfunction is the result of increased oxidative stress and protein nitrotyrosylation caused by increased expression of iNOS protein and that d-GlcN and Thiamet-G suppress these deleterious effects, presumably via elevated protein O-GlcNAcylation.

Fig. 6.

Representative paraffin-embedded cross-sections of rat aortic rings pretreated for 3 h with 5 mM l-Glc or H₂O (Veh), 5 mM d-GlcN, or 0.1 μM Thiamet-G in the absence (left) or presence (right) of 10 ng/ml TNF-α for 21 h and then subjected to immunohistochemical staining for nitrotyrosine (red). Sections were counterstained with hematoxylin to stain nuclei (blue).

DISCUSSION

In the present study, we demonstrate that d-GlcN and Thiamet-G increase global protein O-GlcNAcylation and protect against TNF-α-induced impairment in contractility and endothelium-dependent relaxation in cultured rat aortic rings. Furthermore, we provide evidence that both d-GlcN and Thiamet-G, agents known to increase cellular O-GlcNAcylation via different mechanisms, protect against TNF-α-induced oxidative stress and vascular dysfunction by inhibiting iNOS expression and suppressing the nitrotyrosylation of proteins in the aortic endothelium and media.

Endothelial dysfunction is generally characterized by impaired endothelium-dependent relaxation due to lower NO bioavailability as a result of reduced endothelium-derived hyperpolarizations, increased production of contracting factors such as prostanoids, and increased levels of ROS (10, 15). TNF-α can increase the production of ROS, such as superoxide anion and peroxynitrite, via upregulation of ROS-producing enzymes, including iNOS via the NF-κB transcription factor pathway (2, 14, 45). TNF-α is known to mediate vascular collapse in sepsis resulting from bacterial endotoxin, and l-arginine-derived NO is believed to be the principal mediator of TNF-α-induced hypotension (23). Upregulation of iNOS protein expression has been shown to cause increased NO release during sepsis in vivo (40), and TNF-α-induced NO formation in the vasculature accounts for the loss of contractility observed during septic shock and antitumor therapy with cytokines (2, 22, 34).

In the present study, we demonstrated that isolated rat aortic rings pretreated with l-Glc followed by overnight exposure to recombinant TNF-α developed reduced contractile responses to depolarizing K⁺ solution. This finding is consistent with a study performed by Takahashi and colleagues (38), in which TNF-α delivered intravenously to Sprague-Dawley rats for 10 h resulted in hyporesponsiveness to depolarizing K⁺ solution in ex vivo aortic rings. We found that adrenergic receptor-mediated contractions induced by Phe were also reduced after TNF-α incubation in rings pretreated with osmotic or vehicle control solutions (l-Glc or H₂O), confirming other reports (16, 22, 50). ACh-induced endothelium-dependent relaxations were impaired in rings pretreated with l-Glc or H₂O and incubated overnight with TNF-α, consistent with results of other studies (16, 22, 42, 50) using cultured arteries.

TNF-α-induced endothelial dysfunction was also observed with the endothelium-dependent vasoactive peptide substance P but not by the nonreceptor-mediated endothelium-dependent Ca²⁺ ionophore A-23187. This is in agreement with a study by Kessler and colleagues (22) and suggests that the downstream intracellular Ca²⁺-calmodulin-dependent signal transduction pathway leading to vasorelaxation is not impaired by TNF-α. Relaxations to the NO donor SNP were comparable for rings pretreated with control solutions (l-Glc or H₂O) in the absence or presence of TNF-α, suggesting that the ability of smooth muscle cells to respond to NO was not impaired by TNF-α.

d-GlcN and Thiamet-G are known to increase cellular protein O-GlcNAcylation via different pathways. The hexosamine sugar d-GlcN increases O-GlcNAc levels by increasing flux through the hexosamine biosynthesis pathway (41), whereas Thiamet-G prevents the detachment of these O-linked sugar moieties from proteins by inhibiting OGA (49). The 5 mM concentration of d-GlcN used in this study has been shown in our laboratory to result in rapid and sustained elevations in protein O-GlcNAc levels in isolated rat aortic smooth muscle cells (47). Since Thiamet-G has an inhibitory constant of 21 nM for human OGA (49), and 25 nM Thiamet-G has been shown to increase overall O-GlcNAc protein levels in rat mesangial cells (13), we used a concentration of 0.1 μM for rat aortic rings. We showed that both d-GlcN and Thiamet-G increased overall O-GlcNAc levels in cultured aortic segments in both the absence and presence of TNF-α compared with control segments. Furthermore, we demonstrated that both d-GlcN and Thiamet-G prevented the TNF-α-induced impairment in endothelium-dependent ACh-induced relaxation, suggesting that the beneficial effects of d-GlcN and Thiamet-G on endothelial integrity were related to their capacity to increase cellular O-GlcNAc levels.

Much of the prior work that examined the effects of O-GlcNAcylation in the vasculature has focused on the detrimental effects of increased cellular O-GlcNAc levels on vascular function, with a particular emphasis on vascular complications of disease states such as diabetes and hypertension (8, 26, 27, 31, 32). In contrast, the findings of the present study and other recent studies (3, 5, 11, 29, 46) have shown that acute activation of pathways leading to increased O-GlcNAc levels is cardioprotective and vasoprotective. We postulate, therefore, that acute and transient activation of O-GlcNAc levels represents an endogenous stress response that is beneficial for cell survival. This is consistent with the notion that the short-term fine balancing of O-GlcNAc levels has protective and immunosuppressive effects on cellular function. This contrasts with the detrimental effects associated with chronic elevations in O-GlcNAc levels in conditions such as diabetes and some forms of hypertension.

TNF-α results in inflammatory gene transcription via the NF-κB transcription factor pathway (2, 14, 45). NF-κB activation is critical for the expression of a variety of genes involved in vascular inflammation, such as genes encoding for chemokines, adhesion molecules, and iNOS (2, 16, 35, 39, 40). For the activation of NF-κB, phosphorylation of the p65 subunit of NF-κB is critical. Recently, our laboratory (47) demonstrated a reciprocal relationship between O-GlcNAcylation and phosphorylation of the p65 subunit of NF-κB, such that increased NF-κB p65 O-GlcNAcylation inhibited TNF-α-induced expression of inflammatory mediators through inhibition of NFκB p65 signaling. Since TNF-α can induce the expression of iNOS protein in smooth muscle via the NF-κB signaling pathway, we hypothesized that protein O-GlcNAcylation induced by d-GlcN or Thiamet-G would inhibit NF-κB signaling and hence suppress iNOS expression. As anticipated, TNF-α induced a significant increase in the expression of iNOS protein in aortic rings that was completely inhibited by d-GlcN or Thiamet-G treatment.

Nitrotyrosine and iNOS are oxidative stress markers. Increased iNOS protein expression results in supraphysiological NO concentrations, and prolonged elevations in NO release result in oxidative stress due to increase peroxynitrite formed by the reaction of NO and superoxide anions (37). We demonstrated that TNF-α-induced upregulation of iNOS was associated with increased staining of nitrotyrosylated proteins in the aortic wall, providing evidence that increased oxidative stress and the resulting protein nitrotyrosylation, were responsible for the observed vascular dysfunction. While d-GlcN has been shown to reduce iNOS expression in an in vivo rat model of endotoxin-induced inflammation (33), isolated mouse macrophages (36), and human osteoarthritic chondrocytes (43), a link between O-GlcNAcylation and iNOS inhibition has never been postulated. Here, we demonstrate for the first time that TNF-α-induced iNOS expression in cultured rat aortic rings can be inhibited by agents that increase protein O-GlcNAcylation acutely.

A major limitation of this study is that we have not demonstrated a direct causal relationship between increased O-GlcNAcylation and inhibition of TNF-α-induced vascular dysfunction and iNOS expression. However, as described previously, earlier studies (16, 19) have documented that selective overexpression and deletion of iNOS gene expression have functionally significant effects in regulating vascular contractile and relaxing properties. The novelty of the present findings is that we have demonstrated important effects of increased protein O-GlcNAcylation on vascular function, iNOS expression, and oxidative stress. These ex vivo findings extend our previous demonstration that protein O-GlcNAc modification inhibits NF-κB signaling in cultured rat aortic smooth muscle cells by shifting the balance between O-phosphorylation and O-GlcNAcylation of NF-κB p65 toward O-GlcNAcylation (47). The earlier observations in isolated smooth muscle cells suggest a possible mechanism to explain our results (47). Another limitation of this study is the use of pharmacological agents to modulate protein O-GlcNAcylation. Future studies in our laboratory will use novel tissue-specific inducible transgenic mouse models with smooth muscle cell-selective deletion of the OGT gene and overexpression of a catalytically inactive OGA splice variant to provide a more rigorous test of our hypothesis. In these two transgenic mouse models, O-GlcNAc levels are expected to be decreased and increased, respectively, in a smooth muscle cell-specific manner.

GRANTS

This work was supported by National Institutes of Health Grants RO1-HL-087980 (to S. Oparil), HL-101192 and HL-079364 (to J. C. Chatham), and HL-080017 and HL-044195 (to Y. F. Chen) and by American Heart Association-Greater Southeast Affiliate Grants 0765398B (to D. Xing) and 10POST3180007 (to K. Gong). A Sponsorship by Pantarhei Bioscience (Zeist, The Netherlands), obtained by R. Hilgers, enabled the purchase of the myograph system (model 610M, Danish Myo Technology).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: R.H.P.H., Y.-F.C., J.C.C., and S.O. conception and design of research; R.H.P.H., D.X., and K.G. performed experiments; R.H.P.H., D.X., and Y.-F.C. analyzed data; R.H.P.H., D.X., J.C.C., and S.O. interpreted results of experiments; R.H.P.H., D.X., and S.O. prepared figures; R.H.P.H., J.C.C., and S.O. drafted manuscript; R.H.P.H., J.C.C., and S.O. edited and revised manuscript; R.H.P.H., D.X., Y.-F.C., J.C.C., and S.O. approved final version of manuscript.