Enhanced NF-κB Activity Impairs Vascular Function Through PARP-1–, SP-1–, and COX-2–Dependent Mechanisms in Type 2 Diabetes

All experiments were performed according to the American Guidelines for the Ethical Care of Animals and were approved by Tulane University Health Sciences Center Animal Care and Use Committee. Type 2 diabetic male mice (db⁻/db⁻) (8- to 10-week-old males) and their homologous controls were purchased from The Jackson Laboratory (Bar Harbor, ME), housed in groups of five mice, and maintained at a temperature of 23°C with 12-h light/dark cycles. Mice were fed on a solid standard diet (Na⁺ content 0.4%) and water. Mice were divided into six groups: 1) control mice infused with saline (n = 10), 2) control mice that received dehydroxymethylepoxyquinomicin (DHMEQ) (an NF-κB inhibitor, 6 mg/kg i.p. injection twice a week) for 4 weeks (n = 10), 3) control mice that received IKK-NBD peptide (an NF-κB inhibitor, 500 μg/kg/day daily i.p. injection) for 4 weeks (n = 10), 4) db⁻/db⁻ mice infused with saline (n = 10), 5) db⁻/db⁻ mice treated with DHMEQ (6 mg/kg i.p. injection twice per week) for 4 weeks (n = 10), and 6) db⁻/db⁻ mice treated with IKK-NBD peptide (500 μg/kg/day i.p. daily) for 4 weeks (n = 10). The body weight and blood glucose levels were recorded weekly during the experimental period. Blood glucose measurements were obtained from tail blood samples using a blood glucose meter (Prestige Smart System HDI; Home Diagnostic, Fort Lauderdale, FL) in all groups of mice after a 6-h fast as previously described (18). Systolic blood pressure was measured by the tail-cuff machine as previously described (19).

At the end of the treatment period, mice were anesthetized with isoflurane and blood samples were collected from carotid artery into containing heparin tubes. Then, tissues (heart and MRA) were harvested immediately, placed in physiological salt solution (PSS) solution (composition in millimoles per liter: NaCl 118, KCl 4.7, CaCl₂ 2.5, KH₂PO₄ 1.2, MgSO₄x7H₂O 1.2, NaHCO₃ 25, and glucose 11, pH = 7.4), and processed appropriately for further studies.

In another series of experiments, we used 8-week-old double knockout between db⁻/db⁻ and p-50NF-κB male mice (db⁻/db^{−p50NF-κB−/−}, n = 5) and between db⁻/db⁻ and PARP-1 male mice (db⁻/db^{−PARP-1−/−}, n = 5). The PARP-1 knockout mice were provided by A.B. The p50NF-κB knockout mice were purchased from The Jackson Laboratory. To generate double knockout, we bred heterozygote db⁻/db⁺ with PARP-1 or p50NF-κB knockout mice. Before the animals were killed, the body weight and blood glucose levels were measured. Then mice were anesthetized with isoflurane and then coronary arterioles (CAs), and MRAs were immediately harvested, placed in PSS solution, and processed appropriately for further studies.

To study the metabolic characterization of the mice on drugs and crossed with knockouts, we measured the insulin and cholesterol levels using the Insulin ELISA kit (Mercodia, Uppsala, Sweden) and cholesterol kit (Cayman Chemical Company, Ann Arbor, MI), respectively. We also measured markers of inflammation (interleukin-6 and tumor necrosis factor [TNF]-α [both Bioscience, San Diego, CA).

Diameter. The left anterior descending CAs and MRAs were isolated, cannulated with glass micropipettes, and perfused with PSS bubbled with a 95% O₂ plus 5% CO₂ gas mixture. The vessels were pressurized to 50 mmHg using pressure-servo control perfusion systems (Living Systems Instruments, St. Albans, VT) for a 40-min equilibration period. The vessel diameter was monitored by a video image analyzer as previously described (19,20).

Intraluminal pressure was increased from 25 to 100 mmHg in a stepwise manner to measure myogenic tone. At the end of the experiments, vessels were incubated with a calcium-free PSS to determine passive diameter. Myogenic tone is calculated as the percentage between active and passive diameter. For determination of the endothelium-dependent relaxation (EDR), pressurized arteries were preconstricted with thromboxane agonist (U-46619, 10⁻⁷ mol/L) and then cumulative concentrations (10⁻⁹ to 10⁻⁵ mol/L) of acetylcholine (ACh) were assessed. The NO synthase inhibitor NG-nitro-l-arginine methyl ester (L-NAME) (10⁻⁴ mol/L) was added to the superfusate for 30 min before the assessment of the vasodilator response to ACh.

Isometric tension recording. MRAs from control and diabetic mice were carefully cleaned of fat and connective tissue and then cut into rings (2 mm in length). MRAs were mounted in a small-vessel dual-chamber myograph for measurement of isometric tension. After a 30-min equilibration period in PSS solution bubbled with carbogen at 37°C and pH = 7.4, arteries were stretched to their optimal lumen diameter for active tension development. After a second 30-min equilibration period, the vessels were exposed to phenylephrine (PE) (10⁻⁵ mol/L) and the presence of functional endothelium was assessed by the ability of ACh (10⁻⁶ mol/L) to induce relaxation.

For determination of the role of endothelial NO synthase (eNOS) and COX-2 in the impaired EDR in diabetic mice, MRAs were incubated with NS 398 (10 μmol/L), a COX-2–selective inhibitor, for 1 h and then EDR was performed after precontraction with PE. For determination of the role of eNOS and NADPH oxidase in the impaired EDR in diabetic mice, MRAs were incubated with L-NAME (100 μmol/L) and apocynin (100 μmol/L) for 30 min and then EDR was performed after precontraction with PE.

The same protocol was used for double knockout db⁻/db⁻ and p50 NF-κB (db⁻/db^{−p50NF-κB−/−}) and db⁻/db⁻ and PARP-1 (db⁻/db^{−PARP-1−/−}) male mice. After precontraction with PE (10⁻⁵ mol/L) and the steady maximal contraction, cumulative dose-response curves were obtained for ACh (10⁻⁸ to 10⁻⁵ mol/L) and sodium nitroprusside (SNP) (10⁻⁸ to 10⁻⁵ mol/L) in the presence or absence of NS-398, L-NAME, and apocynin.

MRA and CAs from diabetic mice were carefully cleaned of fat and connective tissue and then cut into rings (2 mm in length). Arteries from MRA were mounted in a small-vessel dual-chamber myograph for measurement of isometric tension. Arteries were incubated with either p65NF-κB short hairpin RNA (shRNA) lentiviral particle (Santa Cruz Biotechnology, Santa Cruz, CA), to downregulate p65NF-κB expression, or AG1478 (LC Laboratories, Woburn, MA), epidermal growth factor receptor (EGFR) tyrosine kinase (EGFRtk) inhibitor for 4 h. After precontraction with PE (10⁻⁵ mol/L) and the steady maximal contraction, cumulative dose-response curves were obtained for ACh (10⁻⁸ to 10⁻⁵ mol/L) and SNP (10⁻⁸ to 10⁻⁵ mol/) in the presence or absence of L-NAME as described above.

Freshly isolated hearts MRAs from all groups were immediately frozen in liquid nitrogen and then homogenized in ice-cold lysis buffer as previously described (18–20). Western blot analysis was performed for phosphorylated and total eNOS (1:1,000 dilution; Cell Signaling, Boston, MA), phosphorylated and total p65NF-κB (1:1,000 dilution; Cell Signaling), cleaved and total PARP-1 (1:1,000 dilution; Cell Signaling), and COX-2 (1:500 dilution, Santa Cruz Biotechnology) using specific antibodies. Blots were stripped and then reprobed with either β-actin (1:2,000 dilution, Santa Cruz Biotechnology) or glyceraldehyde-3-phosphate dehydrogenase (1:2,000 dilution, Santa Cruz Biotechnology) antibodies to verify the equal loading among the samples.

Mouse coronary arterial endothelial cells (ECs) were purchased from Celprogen (San Pedro, CA), and they were cultured according to the manufacturer instructions using complete growth media. Cultures were incubated at 37°C in a humidified 5% CO₂ incubator, and the medium was changed every 2 days. Confluent cell monolayers were harvested by a brief incubation with 0.25% trypsin-EDTA solution (GIBCO; Invitrogen, Grand Island, NY) and subcultured to perform further experiments between passages 3 and 7.

ECs were transiently transfected with lipofectamine LTX and Plus transfection reagents (Invitrogen) in 80% confluent 12-well culture dishes, including 1 µg of the reporter plasmid pGL2-eNOSPromoter luciferase containing 1,621-bp fragment of human eNOS promoter and 0.5 µg of the reporter plasmid pRL-TK (Promega, Fitchburg, WI) in the presence or absence of 1 µg of the expression vector pCMV4-p65. The plasmids pGL2-eNOSPromoter and pCMV4-p65 were purchased from Addgene (cat. nos. 19297 and 21966, respectively). After 48 h of incubation with normal glucose (5 mmol/L) or high glucose (25 mmol/L), cells were lysed in lysis buffer (Promega), and luciferase and renilla activities were measured with the dual Luciferase Assay System (Promega) with a luminometer (Fluoroskan Ascent FL; Laboratory Systems, Kilsyth, Victoria, Australia). The renilla activity was used to normalize the transfection efficiency. In another set of experiments, the ECs were only transfected with 1 µg of the expression vector pCMV4-p65 and incubated with normal or high glucose to isolate RNA for further experiments.

Sp1 and eNOS mRNA levels were determined in heart tissues and ECs. Total RNA from heart was obtained using the RNeasy Fibrous Tissue Mini Kit (Quiagen, Valencia, CA) according to the manufacturer’s recommendations or using the TRI Reagent (Sigma-Aldrich, St. Louis, MO) for cells. A total of 1 μg DNase I–treated RNA was reverse transcribed into cDNA using the High Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA) with random hexamers in a 20-μL reaction. PCR was performed in duplicate for each sample using 1 μL cDNA as a template ×1 of TaqMan Universal PCR Master Mix (Applied Biosystems) and ×10 of Taqman Gene Expression Assays (Applied Biosystems) in a 20-μL reaction. Assays-on-Demand (Applied Biosystems) of TaqMan fluorescent real-time PCR primers and probes were used for Sp1 (Mm00489039_m1), Nos3 (Mm00435217_m1), and glyceraldehyde-3-phosphate dehydrogenase (Mm99999915_g1), which was used as endogenous control to normalize results. Quantitative RT-PCR was carried out in an ABI PRISM 7000 Sequence Detection System (Applied Biosystems) using the following conditions: 2 min at 50°C and 10 min at 95°C followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Relative mRNA levels were determined using the 2-∆∆C_t method. Results are expressed as the relative expression of mRNA in treated mice or transfected cells compared with untreated mice or control cells.

PI hydrochloride, ACh, L-NAME, apocynin, and NS398 were obtained from Sigma-Aldrich. IKK-NBD peptide was purchased from Enzo Lifesciences (Farmingdale, NY). The DHMEQ was synthesized by K.U. (Aichi Medical University, Nagakute, Japan).

Results are expressed as means ± SEM. Concentration response curves were analyzed using the GraphPad Prism 4.0 software (GraphPad, La Jolla, CA). One-way or two-way ANOVA was used to compare each parameter when appropriate. Comparisons between groups were performed with t tests when the ANOVA test was statistically significant. Values of P < 0.05 were considered significant. Differences between specified groups were analyzed using the Student t test (two tailed) for comparing two groups, with P < 0.05 considered statistically significant.

Blood glucose levels and body weight were elevated in diabetic mice (383.1 ± 28.8 mg/dL and 43.3 ± 0.6 g, respectively) compared with control mice (131.1 ± 1.8 mg/dL and 25.7 ± 0.5 g) but were not affected by the NF-κB inhibitors or in double knockout mice (db⁻/db^{−p50NF-κB−/−} and db⁻/db^{−PARP-1−/−}), indicating that enhanced NF-κB and PARP-1 activity are involved in type 2 diabetes–induced vascular complication (Table 1). Systolic blood pressure was similar in all groups (Table 1).

Insulin levels as well as cholesterol, interleukin-6, and TNF-α were increased in the diabetic mice compared with control mice (Figure Suppl. 1). Treatment with either DHMEQ or peptide had no effect on the control group. Additionally, NF-κB inhibitors reduced interleukin-6 levels and TNF-α in the diabetic group with no effect on insulin and cholesterol. Interestingly, all of these parameters were reduced in double knockout mice (db⁻/db^{−p50NF-κB−/−} and db⁻/db^{−PARP-1−/−}) (Supplementary Fig. 1).

Myogenic tone was significantly increased in the CA and MRA from diabetic mice compared with control mice and was normalized after NF-κB inhibition (Fig. 1A and B). The EDR was also impaired in CA and MRAs from diabetic mice and was rescued after NF-κB inhibition (Fig. 1C and D). To strengthen our data, we examined myogenic tone and the EDR in CAs and MRAs from db⁻/db^{− p50NF-κB−/−} and db⁻/db^{−PARP-1−/−} double knockout mice. It was observed that the myogenic tone and EDR in CAs and MRAs were also normalized in these two double knockout mice (Fig. 1E–H).

We did not observe any effect on myogenic tone or EDR in CAs or MRAs from control mice treated with DHMEQ and IKK-NBD peptide (Supplementary Fig. 2A–D). The inhibition of eNOS with L-NAME reduced the EDR in all groups of mice (Supplementary Fig. 2E and F).

To determine whether NF-κB is downstream of EGFRtk and that the acute downregulation of NF-κB improves coronary vascular function in diabetes, we incubated CAs from db⁻/db⁻ mice with the EGFRtk inhibitor (AG1478, 1 μmol/L) for 4 h. Our results show a significant improvement in EDR associated with a reduction in NF-κB activity (Figs. 1I and 2G).

Furthermore, in isolated coronary artery from db⁻/db⁻ mice transfected with p65NF-κB shRNA lentiviral particles for 4 h, there was a significant reduction in p65NF-κB expression and improvement in coronary EDR (Figs. 1J and 2K). As a control experiment, CA from db⁻/db⁻ mice transfected with scrambled shRNA did not show improved coronary EDR (Fig. 1J).

To confirm the effect of NF-κB inhibition, we performed Western blot analysis and found that NF-κB activity was elevated in db⁻/db⁻ mice compared with db⁻/db⁻ mice treated with DHMEQ and IKK-NBD peptide and compared with the double knockout db⁻/db^{− p50NF-κB−/−} and db⁻/db^{−PARP-1−/−} mice (Fig. 2A, D, and H). The eNOS phosphorylation at Ser 1177 and the expression were significantly reduced in CA from db⁻/db⁻ mice compared with controls, db⁻/db⁻ mice treated with DHMEQ and IKK-NBD peptide, and db⁻/db^{− p50NF-κB−/−} and db⁻/db^{−PARP-1−/−} mice (Fig. 2B, E, and I).

To study the effect of NF-κB inhibition on PARP-1 signaling, we performed Western blot in CA from all groups. Our data indicate that cleaved PARP-1 levels and total PARP-1 were increased in the diabetic group compared with diabetic mice treated with DHMEQ and IKK-NBD peptide and in double knockout db⁻/db^{− p50NF-κB−/−} and db⁻/db^{−PARP-1−/−} mice (Fig. 2C, F, and J).

Figure 3A shows the key representative curves made in the wire myograph to test the EDR. Our data demonstrated that EDR is impaired in MRAs in db⁻/db⁻ mice and the inhibition of NF-κB with DHMEQ and IKK-NBD peptide significantly improved the EDR (Fig. 3B and Supplementary Table 1).

We did not observe a change in EDR in MRA from control mice treated with DHMEQ and IKK-NBD peptide (Supplementary Fig. 3A). The inhibition of eNOS with L-NAME reduced EDR in all groups of mice (Supplementary Fig. 3B). To further elucidate the mechanism of vascular endothelial dysfunction in diabetic mice related to NF-κB, we studied the role of the COX-2 and oxidative stress pathways. Thus, the inhibition of NADPH oxidases by apocynin showed a significant improvement in MRA (Fig. 3C and Supplementary Table 1). In addition, the COX-2 inhibition by NS398 significantly improved the EDR in MRA (Fig. 3D and Supplementary Table 1). Western blot analysis revealed that in MRA from db⁻/db⁻ mice, eNOS activity was significantly reduced, while NF-κB and PARP-1 activity and COX-2 expression were significantly augmented (Fig. 4A–D). Interestingly, the treatment of db⁻/db⁻ mice with NF-κB inhibitors (DHMEQ and IKK-NBD peptide) increased eNOS activity and reduced NF-κB, PARP-1 activity, and COX-2 expression (Fig. 4A–D).

To strengthen our studies, we examined the EDR in MRA in double knockout mice (db⁻/db^{− p50NF-κB−/−}). Western blot analysis confirmed the absence of p50NF-κB in MRA in db⁻/db^{− p50NF-κB−/−} mice (Fig. 4G). Furthermore, the EDR was significantly improved in MRA isolated from db⁻/db^{− p50NF-κB−/−} mice associated with increased eNOS activity (Figs. 3E and 4E and Supplementary Table 1). To further determine the mechanism of NF-κB in vascular function, we studied the role of PARP-1 and COX-2 as downstream signaling for NF-κB. Thus, EDR was significantly improved in double knockout db⁻/db^{−PARP-1−/−} mice and arteries from db⁻/db⁻ mice treated with COX-2 inhibitor (Fig. 3E and F and Table 1). These data were also associated with reduction in cleaved PARP-1 and COX-2 expression in MRA from db⁻/db^{− p50NF-κB−/−} mice (Fig. 4I and K). Furthermore, Western blot analysis revealed that in MRA from db⁻/d^{− PARP-1−/−} mice, PARP-1 was absent and this was associated with reduced p65NF-κB phosphorylation or COX-2 expression or with an increase in eNOS phosphorylation (Fig. 4F, H, J, and L). Endothelium-independent relaxation in response to SNP in MRA was similar in control, db⁻/db⁻, and db⁻/db^{− p50NF-κB−/−} mice (Supplementary Fig. 4A).

Isolated MRAs from db⁻/db⁻ mice treated with EGFRtk inhibitor significantly improved the EDR (Fig. 3G and H and Supplementary Table 1). Western blot analysis confirmed the reduction in p65NF-κB phosphorylation in MRA (Fig. 4M) indicating that p65NF-κB is downstream of EGFRtk. The acute downregulation of p65NF-κB in isolated MRAs from db⁻/db⁻ mice transfected with p65NF-κB shRNA lentiviral particles significantly improved the EDR, which is inhibited by L-NAME (Fig. 3H and Supplementary Table 1). Western blot analysis confirmed the downregulation of p65NF-κB in MRA transfected with p65NF-κB shRNA lentiviral particles (Fig. 4N). Endothelium-independent relaxation in response to SNP in MRA showed no difference between controls, db⁻/db⁻, and db⁻/db⁻ transfected with lentivirus shRNA p65NF-κB (Supplementary Fig. 4B).

To elucidate the mechanism by which eNOS expression and activity were decreased in diabetes, we studied the Sp-1 expression known as downstream signaling of NF-κB. Thus, the mRNA levels of Sp-1 were significantly reduced in coronary artery isolated from db⁻/db⁻ mice compared with controls (Fig. 5A). Interestingly, db⁻/db⁻ mice treated with NF-κB inhibitor significantly increased Sp-1 expression in CAs (Fig. 5A). We did not see any change in Sp-1 expression in control mice treated with NF-κB inhibitor (Fig. 5A). In primary ECs isolated from CAs, transfected with a reporter plasmid containing the eNOS promoter, the stimulation of these cells with high glucose and/or overexpression of p65NF-κB significantly reduced Sp-1 expression (Fig. 5B). In addition, eNOS promoter activity was significantly reduced in ECs stimulated with high glucose and/or transfected with p65NF-κB (Fig. 5C). These results were associated with the reduction in eNOS expression (Fig. 5D).

Our study demonstrates that enhanced NF-κB activity is an important factor involved in macro- and microvascular dysfunction in type 2 diabetes. This evidence came from our ex vivo and in vitro experiments where vascular EDR and myogenic tone were significantly improved in type 2 diabetic mice treated with NF-κB inhibitors and double knockout mice (db⁻/db^{−p50NF-κB−/−} and db⁻/db^−PARP-1) and with the downregulation of NF-κB using shRNA lentiviral particles. We further demonstrated that in diabetes, NF-κB inhibition improves vascular function by PARP-1–, Sp-1–, and COX-2–dependent mechanisms. These results suggest that the NF-κB pathway could be an important target for a therapeutic approach to reverse diabetes-induced vascular dysfunction.

In diabetes, vascular complications are major causes for morbidity and mortality (21–24). Several mechanisms have been proposed such as hyperglycemia (25), altered COX-2 activity (26), oxidative stress (27), insulin resistance, and inflammation (28). Although previous studies established the relationship between NF-κB, diabetes, and cardiovascular diseases (29,30), the role and mechanism of NF-κB in diabetes-induced vascular dysfunction are still unknown.

Myogenic tone is a characteristic of resistance arteries, which plays an important role in the local regulation of vascular diameter change and tissue blood flow (31). Previous studies have investigated impaired small artery and arteriolar function in diabetic patients and have shown that vasomotor dysfunction affects both endothelial and smooth muscle cell–mediated regulatory mechanisms (32,33). In type 2 diabetic mice, we and other investigators have demonstrated an increase in myogenic tone (15,34,35). Interestingly, in the current study we observed that the augmented myogenic tone in type 2 diabetes is the result of enhanced NF-κB and PARP-1 activity. Thus, NF-κB and PARP-1 pathway inhibition, using pharmacologic and genetic approaches, reduced the potentiated myogenic tone in diabetic mice.

Vascular reactivity is also regulated by EDR mechanism. This mechanism is highly dependent not only on the bioavailability of NO in large arteries but also on the endothelium-dependent hyperpolarizing factor and prostaglandin metabolites in resistance arteries (36). Our results indicate that EDR is impaired in CA and MRA in type 2 diabetes, which is in agreement with previous studies (31,37,38). It has been shown that loss of endothelium-derived NO is associated with increases in NF-κB activity (39). In addition, it has been reported that cultured porcine coronary EC function is impaired with senescence, associated with enhanced NF-κB activity (40). In this study, we demonstrated that type 2 diabetes is associated with increased vascular NF-κB activity and impaired vascular function. Thus, the inhibition of NF-κB activity with different approaches (pharmacologic, double-knockout mouse models and downregulation of protein expression by NF-κB shRNA lentiviral particles) significantly improved EDR in CA and MRA. Importantly, the effect of NF-κB activity inhibition is not specific to one vascular bed indicating the critical importance of the NF-κB pathway in the global regulation of vasculature function in type 2 diabetes. It is also interesting to note that NF-κB inhibition did not reduce blood glucose and insulin levels or body weight, suggesting that enhanced NF-κB is a consequence of type 2 diabetes.

It is well-known that active NF-κB interacts with PARP-1 and translocates to the nucleus to target promoters (41). In addition, we previously reported that inhibition of PARP-1 activity improved vascular function in type 2 diabetes (15). In the current study, we found an increase in PARP-1 activity in arteries from type 2 diabetic mice. The inhibition of PARP-1 activity in db⁻/db⁻ mice, using genetic deletion, significantly improved vascular function. In addition, the inhibition of the NF-κB pathway reduced PARP-1 activity, indicating that NF-κB regulates vascular function in type 2 diabetes by a PARP-1–dependent mechanism. These results are supported by studies in the literature demonstrating the formation of a complex NF-κB–PARP-1 in the nucleus, which binds to DNA to modulate gene expression (38,42). This activity of this chemical complex is also supported by our experiments showing an improvement in vascular function in double knockout db⁻/db^{−PARP-1−/−} mice.

To further investigate the downstream mechanism of NF-κB and PARP-1 and because of the importance of the transcription factor Sp-1 in the regulation of eNOS promoter activity, we studied Sp-1 expression. Interestingly, we found that Sp-1 mRNA levels were significantly reduced and were rescued by the inhibition of NF-κB in CAs from type 2 diabetic mice. In addition, in vitro studies show that overexpression of p65NF-κB in transfected primary cultured coronary ECs or incubation with high glucose reduced eNOS promoter activity and eNOS and Sp-1 mRNA levels. Our results are in agreement with studies indicating that the transcription factor Sp-1 binds to the eNOS promoter (43). In addition, it has recently been demonstrated that p65NF-κB subunit interacts with Sp-1 and negatively regulates gene expression (44). Therefore, our results suggest that enhanced NF-κB signaling impairs vascular function by PARP-1– and Sp-1–dependent mechanisms.

To provide information about the mechanism of impaired vascular function caused by enhanced NF-κB activity in diabetes, we observed that vascular COX-2 expression was enhanced in db⁻/db⁻ mice. These data are consistent with studies showing the induction of COX-2 in type 2 diabetes (45,46).

The NF-κB is a ubiquitous family of transcription factors that also control the expression of genes involved in the inflammatory response, such as COX-2. Our studies showed that inhibition of NF-κB activity in db⁻/db⁻ mice reduced COX-2 expression indicating that COX-2 is downstream to NF-κB. In addition, the in vitro inhibition of COX-2 improves vascular function in db⁻/db⁻ mice. These data indicate that inhibition of NF-κB improved vascular function by a COX-2–dependent mechanism.

Although a previous population-based study showed that COX-2 inhibitors might lead to increase the risk of myocardial infarction in the general population (47), it has been shown that a post hoc analysis for different coxibs revealed a significant association with incident atrial fibrillation for etoricoxib but not for celecoxib. These results indicate that the issue is mostly related to the nature of COX-2 inhibitors (48).

In summary, our study demonstrates the importance of NF-κB in the regulation of vascular function in type 2 diabetes by PARP-1–, Sp-1–, and COX-2–dependent mechanisms (Supplementary Fig. 5). Therefore, NF-κB could be a potential target for a novel therapeutic strategy to reverse diabetes-induced vascular dysfunction.

Although there are studies involving NF-κB in diabetes, the proposed study is innovative, in our opinion, because our approach provided direct evidence that enhanced NF-κB activity causes vascular dysfunction in terms of myogenic tone and EDR by PARP-1–, Sp-1– and COX-2–dependent mechanisms in two different vascular beds. There is a paucity of studies that explore the role of the NF-κB pathway and the mechanism (PARP-1, Sp-1, and COX-2) by which type 2 diabetes impairs the regulation of arteries, which leads to heart disease.

Emerging evidence from experimental and clinical research indicates that an NF-κB pathway plays pivotal roles in cardiovascular diseases. Lack of such knowledge is a fundamental problem because without it, endothelial dysfunction that causes coronary artery disease will still present a high risk for myocardial infarction in diabetic patients. Thus, our ex vivo and in vitro data clearly indicate an enhanced NF-κB pathway associated with impaired EDR in three vascular beds. Importantly, the inhibition of NF-κB activity improves vascular EDR. Thus, in our study we demonstrated the role of enhanced NF-κB activity in the dysfunction of vascular EDR by PARP-1–, Sp-1–, and COX-2–dependent mechanism.

The authors acknowledge grant support from the National Institutes of Health (1R01HL095566 [principal investigator K.M.] and 5R01HL097111 [principal investigator M.T.]).

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

M.K., S.-K.C., and M.G. conducted experiments and wrote the manuscript. A.B. generated PARP-1 knockout mice. K.U. generated and provided DHMEQ. M.T. and S.B. participated in discussion and wrote the manuscript. K.M. was the principal investigator, designed the research, and wrote the manuscript. K.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis.