The vicious cycle between transglutaminase 2 and reactive oxygen species in hyperglycemic memory–induced endothelial dysfunction

Abstract

Clinical trials suggested that the vascular system can remember episodes of poor glycemic control through a phenomenon known as hyperglycemic memory (HGM). HGM is associated with long-term diabetic vascular complications in type 1 and type 2 diabetes, although the molecular mechanism of that association is not clearly understood. We hypothesized that transglutaminase 2 (TGase2) and intracellular reactive oxygen species (ROS) play a key role in HGM-induced vascular dysfunction. We found that hyperglycemia induced persistent oxidative stress, expression of inflammatory adhesion molecules, and apoptosis in the aortic endothelium of HGM mice whose blood glucose levels had been normalized by insulin supplementation. TGase2 activation and ROS generation were in a vicious cycle in the aortic endothelium of HGM mice and also in human aortic endothelial cells after glucose normalization, which played a key role in the sustained expression of inflammatory adhesion molecules and apoptosis. Our findings suggest that the TGase2-ROS vicious cycle plays an important role in HGM-induced endothelial dysfunction.—Lee, J.-Y., Lee, Y.-J., Jeon, H.-Y., Han, E.-T., Park, W. S., Hong, S.-H., Kim, Y.-M., Ha, K.-S. The vicious cycle between transglutaminase 2 and reactive oxygen species in hyperglycemic memory–induced endothelial dysfunction.

Diabetes mellitus is a group of metabolic disorders characterized by chronic hyperglycemia. Hyperglycemia is associated with progressive damage to the endothelium of blood vessels, resulting in vascular complications and diabetes-related morbidity and mortality (1, 2). Hyperglycemia and subsequent metabolic changes develop into microvascular complications in the retina, renal glomerulus, and peripheral nerves, which contribute to blindness, end-stage renal disease, and neuropathies (1, 3). Diabetes is also related to macrovascular diseases of the arteries that supply blood to the heart and brain, including cardiovascular diseases such as myocardial infarction and cerebrovascular disease manifesting as stroke (2, 4, 5). Hyperglycemia-induced overproduction of mitochondrial superoxide and the resulting vascular damage cause diabetic microvascular and macrovascular complications through various pathogenic mechanisms, such as increased polyol pathway flux, increased intracellular advanced glycation end product (AGE) formation, PKC activation, and elevated hexosamine biosynthesis (1, 5, 6). Those pathogenic pathways activate inflammation and long-lasting epigenetic changes, leading to the hyperglycemic memory (HGM), or metabolic memory, phenomenon (1, 7).

Clinical trials demonstrated that blood glucose normalization did not prevent diabetic complications caused by HGM (2). HGM, described as persistent hyperglycemic stress after normoglycemia (7), was initially observed by Engerman and Kern (8), who found that retinopathy in diabetic dogs was not effectively improved by good glycemic control. The HGM phenomenon has been reported in clinical trials undertaken to evaluate the impact of intensive glucose therapy on the vascular complications of type 1 and type 2 diabetes (2). The Diabetes Control and Complications Trial–Epidemiology of Diabetes Interventions and Complications study demonstrated that intensive glycemic control can reduce microvascular complications and cardiovascular disease in patients with type 1 diabetes, but poor glycemic control can lead to the development of diabetic complications even many years after glycemic control is achieved (9, 10). The UK Prospective Diabetes Study was undertaken to evaluate the effect of tight glycemic control on the development of diabetic complications in patients with type 2 diabetes (11). In that study, intensive glycemic control reduced microvascular complications compared with conventional treatment. Furthermore, a 10-yr follow-up study demonstrated that the intensive treatment resulted in persistent reduction of microvascular complications and long-term reduction of the risks of myocardial infarction and all-cause mortality (12, 13). The findings of those large clinical trials demonstrate the benefits of tight glycemic control but also support the hypothesis that episodes of poor glycemic control can lead to long-term diabetic complications.

The underlying mechanisms of HGM have been investigated with the aim to improve the persistent vascular complications associated with hyperglycemic stress. Hyperglycemia-induced superoxide production by the mitochondrial electron transport chain is an upstream event in HGM development that inhibits glyceraldehyde 3-phosphate dehydrogenase (GAPDH) through the action of poly(ADP-ribose) polymerase, resulting in vascular damage through several pathogenic pathways (1, 6, 14). Hyperglycemia induced the persistent up-regulation of the pro-oxidant enzymes PKC-βΙΙ and NADPH oxidase in the retinas of diabetic rats supplemented with insulin (15). In the aortas of insulin-supplemented diabetic mice, PKC-βII induced continuous activation of the adaptor protein p66^Shc, which was associated with apoptosis, sustained generation of reactive oxygen species (ROS), and reduction of NO bioavailability (16). Sustained ROS generation up-regulates the expression of NF-κB subunit p65, accompanied by Su(var), enhancer-of-zeste, trithorax (Set)7-mediated monomethylation of histone H3 at lysine 4 (H3K4me1) (7, 17, 18). Set7 is involved in chromatin remodeling and persistent expression of inflammatory genes in response to hyperglycemia (18). Set7 knockdown prevents transient glucose-induced up-regulation of p65 and sustained expression of NF-κB–dependent monocyte chemoattractant protein 1 and VCAM-1 (19). Epigenetic changes and long-lasting inflammation are implicated in the persistent vascular damage caused by hyperglycemia (7). Although several pathways leading to HGM have been reported, we are still far from understanding the mechanisms responsible for HGM-related diabetic vascular complications.

In this study, we demonstrate that a molecular mechanism involving transglutaminase 2 (TGase2) is responsible for HGM. TGase2 is a member of the transglutaminase (TGase) family and catalyzes Ca²⁺-dependent cross-linking reactions through the transamidation of glutamine and lysine residues of substrate proteins (20). TGase2 is implicated in cardiovascular and neurodegenerative diseases and celiac disease (21–23) and is also associated with diabetic vasculopathy and retinopathy (24, 25). We hypothesized that TGase2 and intracellular ROS play a key role in HGM-induced endothelial dysfunction and apoptosis. We found that TGase2 activation and ROS generation were in a vicious cycle in the aortic endothelium of HGM mice and also in human aortic endothelial cells after glucose normalization. The vicious cycle played a key role in the sustained expression of inflammatory adhesion molecules and endothelial apoptosis. Our findings suggest that TGase2 activation and ROS generation might be important for understanding the progression of diabetic vascular complications caused by HGM.

MATERIALS AND METHODS

Cell culture

Human aortic endothelial cells were purchased from PromoCell (Heidelberg, Germany). Cells from passages 5 to 7 were used in experiments. Cells were grown on 2% gelatin-coated plates in M199 medium supplemented with 20% fetal bovine serum, 3 ng/ml basic fibroblast growth factor, 5 U/ml heparin, 100 U/ml penicillin, and 100 mg/ml streptomycin in a humidified 5% CO₂ incubator. For experiments, endothelial cells were incubated for 12 h in low-serum medium supplemented with 2% fetal bovine serum and antibiotics and then subjected to 1 of 3 glucose treatments: 5.5 mM d-glucose for 6 d (normal glucose), 30 mM d-glucose for 6 d (high glucose), or 30 mM d-glucose for 3 d followed by 5.5 mM d-glucose for 3 d (HGM). Treatment with 30 mM mannitol was used as an osmotic control.

Measurement of intracellular ROS levels

Intracellular ROS levels were measured using 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA; Thermo Fisher Scientific, Waltham, MA, USA) as previously described in Bhatt et al. (26). Cells were analyzed by confocal microscopy (K1-Fluo; Nanoscope Systems, Daejeon, Korea) immediately after H₂DCFDA labeling. Single-cell fluorescence intensities were determined for 30 cells per experiment. Intracellular ROS levels were determined by comparing the fluorescence intensities of treated cells with those of control cells (fold difference).

Measurement of active caspase-3–positive cells

Cells were fixed, permeabilized, and incubated with a blocking solution of 2% bovine serum albumin in 20 mM Tris (pH 7.6), 138 mM NaCl, and 0.1% Tween 20 for 30 min. The cells were then sequentially incubated with polyclonal active caspase-3 antibody (1:400; Cell Signaling Technology, Danvers, MA, USA), Alexa 546–conjugated goat anti-rabbit IgG in blocking solution, and 1 μg/ml DAPI (MilliporeSigma, Burlington, MA, USA). The stained samples were observed using confocal microscopy (K1-Fluo).

Measurement of apoptosis

TUNEL was performed using an APO-BrdU TUNEL Assay Kit (BD Biosciences, San Jose, CA, USA), as previously described in Bhatt et al. (25). Briefly, cells were fixed for 20 min with 1% (w/v) paraformaldehyde in PBS, followed by treatment for 30 min with 70% (v/v) ethanol on ice. The fixed cells were incubated for 1 h at 37°C with a DNA-labeling solution containing terminal deoxynucleotidyl transferase and 5-bromo-2-deoxyuridine in reaction buffer. The cells were then incubated with FITC-labeled 5-bromo-2-deoxyuridine antibody for 30 min and with 1 μg/ml DAPI for 10 min. The mounted cells were visualized using confocal microscopy (K1-Fluo).

Measurement of in situ TGase activity

In situ TGase transamidating activity was determined by confocal microscopy, as previously described in Lee et al. (27). Briefly, cells were incubated with 1 mM 5-(biotinamido)pentylamine at 37°C for 1 h and then fixed and permeabilized. After incubation with a blocking solution for 30 min, the cells were treated with FITC-conjugated streptavidin (1:200, v/v) in blocking solution for 1 h. Fluorescence intensities of single stained cells were determined using confocal microscopy (K1-Fluo).

Transfection with TGase2, p53, and p66^Shc small interfering RNAs

Cells were transfected with 100 nM TGase2 small interfering RNA (siRNA), p53 siRNA, or control siRNA (Santa Cruz Biotechnology, Dallas, TX, USA) using siLentFect lipid reagent (Bio-Rad, Hercules, CA, USA) according to the manufacturer’s instructions (24). Cells were also transfected with 100 nM p66^Shc siRNA (p66^Shc 5′-UGAGUCUCUGUCAUCGCUG[dT][dT]-3′ and p66^Shc 5′-AUGAGUCUCUGUCAUCGCU[dT][dT]-3′) from Bioneer (Daejeon, Korea). The expression levels of TGase2, p53, and p66^Shc were analyzed by Western blot using polyclonal TGase2 antibody (1:3000; Thermo Fisher Scientific), monoclonal p53 antibody (1:2000; Santa Cruz Biotechnology), and monoclonal Shc antibody (1:1000; Santa Cruz Biotechnology), respectively.

Western blot analysis

Proteins were extracted from cells using lysis buffer containing 50 mM Tris (pH 7.5), 1 mM EDTA, 150 mM sodium chloride, 1% Triton X-100, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 0.1 mM phenylmethanesulfonylfluoride, 25 mM β-glycerophosphate, and 2 mM sodium orthovanadate. Protein extracts were resolved by SDS-PAGE and transferred to PVDF membranes. The membranes were incubated with monoclonal antibodies against H3K4me1 and histone H3 (Abcam, Cambridge, United Kingdom), intercellular adhesion molecule 1 (ICAM-1) and VCAM-1 (Santa Cruz Biotechnology), phosphorylated Shc (phospho S36; Abcam), or β-actin (Cell Signaling Technology), followed by incubation with horseradish peroxidase–conjugated secondary antibodies. Protein bands were visualized using a ChemiDoc (Bio-Rad).

Generation of diabetic mice

Six-week-old male C57BL/6 mice were obtained from Daehan Biolink (DBL; Eumseong, South Korea). TGM2^−/− mice (C57BL/6), generated by disruption of exons 5 and 6 of the TGase2 gene (TGM2) by homologous recombination (28), were kindly provided by Dr. Soo-Youl Kim (National Cancer Center, Goyang, South Korea). The mice were maintained under pathogen-free conditions in a temperature-controlled room with a 12-h light/dark cycle. Diabetes was induced by a single daily intraperitoneal injection for 5 consecutive days of streptozotocin (50 mg/kg body weight; MilliporeSigma) freshly prepared in 100 mM citrate buffer (pH 4.5). Mice with fasting blood glucose concentrations ≥19 mM, polyuria, and glucosuria were considered diabetic. All animal experiments conformed to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD, USA) and were approved by the Institutional Animal Care and Use Ethics Committee of Kangwon National University.

Supplementation of mice with C-peptide and insulin using osmotic pumps

Eight weeks after the first streptozotocin injection, the mice were anesthetized with 3% isoflurane and implanted with an Alzet Mini-Osmotic Pump 2004 (Durect, Cupertino, CA, USA) (29), which provided supplementation with human C-peptide (CP; Peptron, Daejeon, South Korea), human recombinant insulin (MilliporeSigma), or both for 4 wk. One group of diabetic mice was supplemented with human recombinant insulin at a delivery rate of 58.4 pmol/min/kg to mimic HGM. To assess the effect of CP treatment on HGM, a different group of diabetic mice (HGM + CP) was supplemented with a mixture of human insulin and CP, at delivery rates of 58.4 and 35 pmol/min per kilogram, respectively. Diabetic and control mice also underwent sham operations. The vital statistics of each mouse were recorded throughout the experiment.

Supplementation of mice with cystamine by oral intake

Eight weeks after the first streptozotocin injection, cystamine (Cys; 225 mg/kg/d) was orally administered via drinking water for 4 wk. Assuming a water consumption rate of 4 ml/d/mouse (25 g), water bottles containing 1120 mg/L Cys in tap water were used. Control mice received regular tap water. Water consumption was checked every day in the morning.

Measurement of TGase activity in the endothelium of mouse aortas

In vivo TGase activity in the endothelium of mouse aortas was determined by confocal microscopy as previously described in ref. 25. Aortas from control, diabetic, insulin-supplemented diabetic (HGM), insulin- and Cys-supplemented diabetic (HGM + Cys), and insulin- and CP-supplemented diabetic (HGM + CP) mice were dissected and cut longitudinally to expose the endothelium for immediate staining (n = 8 mice/group). Aortic segments were incubated for 1 h at 37°C with 1 mM 5-(biotinamido)pentylamine in serum-free M199 medium, fixed for 10 min with ice-cold acetone, and probed for 1 h with FITC-conjugated streptavidin (1:200, v/v). The aortic segments were mounted en face on glass slides and observed by confocal microscopy (K1-Fluo).

Measurement of ROS generation and apoptosis in the endothelium of mouse aortas

ROS levels in the endothelium of mouse aortas were determined using H₂DCFDA and dihydroethidium (30). Briefly, aortic segments were incubated at 37°C either with 10 μM H₂DCFDA in serum-free M199 medium for 10 min or with 5 μM dihydroethidium (Thermo Fisher Scientific) in serum-free M199 medium in PBS for 30 min. The stained aortic segments were then observed using confocal microscopy (FV-300; Olympus, Tokyo, Japan).

Aortic endothelial apoptosis was measured by TUNEL assay as previously described in ref. 25. Briefly, aortic segments were fixed by sequential incubation with 1% (w/v) paraformaldehyde in PBS for 15 min and with 70% (v/v) ethanol on ice for 30 min. The fixed tissues (n = 8/group) were stained using an APO-BrdU TUNEL assay kit and observed using confocal microscopy (K1-Fluo).

Immunofluorescence

Mouse aortas were fixed overnight with 4% paraformaldehyde at 4°C and washed 3 times with PBS. Immunofluorescence staining was performed using either polyclonal antibodies against ICAM-1 and VCAM-1 (Santa Cruz Biotechnology) or monoclonal antibodies against phosphorylated Shc (phospho S36; Abcam) and vascular endothelial cadherin (VE-cadherin; BD Biosciences), followed by probing for 2 h with Alexa 546–conjugated goat anti-rabbit IgG or Alexa 647–conjugated goat anti-rat IgG, respectively. The stained tissues were mounted en face on glass slides and observed by confocal microscopy (K1-Fluo).

Statistical analysis

Data processing was performed using Origin 6.1 software (OriginLab, Northampton, MA, USA). Data were expressed as means ± sd of 3 to 8 independent experiments. Statistical significance was determined using an unpaired Student’s t tests. Values of P < 0.05 were considered statistically significant.

RESULTS

Persistence of oxidative stress, inflammation, and apoptosis in the aortic endothelium of HGM mice

We investigated whether endothelial dysfunction induced by hyperglycemia persisted after blood glucose normalization in the aortic endothelium of HGM mice that were supplemented for 4 wk with human recombinant insulin using osmotic pumps (Fig. 1A). Compared with nondiabetic controls, diabetic mice showed increased food and water consumption, decreased body weight, and hyperglycemia (Fig. 1B, C). Blood glucose normalization led to the normalization of those parameters in HGM mice, with partial recovery of body weight. To investigate the persistence of HGM-induced endothelial dysfunction, aortic segments of mice were longitudinally cut, and the endothelium was stained to visualize oxidative stress, inflammation, and apoptosis. The aortic endothelia were also stained for VE-cadherin along with nuclear counterstaining using DAPI (Fig. 1D). Hyperglycemia stimulated ROS generation in the aortic endothelia of diabetic mice, and glucose normalization by insulin supplementation had no effect on the ROS generation in HGM mice (Fig. 1E–H). The expression of the inflammatory adhesion molecules ICAM-1 and VCAM-1 was elevated in the aortic endothelia of the diabetic mice, and the elevated expression was not altered by glucose normalization in HGM mice (Fig. 5A–D). Furthermore, hyperglycemia-induced endothelial apoptosis in the mouse aortas, assessed by measurement of active caspase-3–positive cells and TUNEL staining, persisted after glucose normalization (Fig. 6A–D). Those results suggest that oxidative stress, inflammation, and apoptosis are involved in persistent hyperglycemic stress in the aortic endothelium of HGM mice.

Figure 1.

Persistence of oxidative stress induced by HGM in the aortic endothelium of diabetic (DM, diabetes mellitus) mice supplemented with insulin. Eight weeks after streptozotocin (STZ) injection, diabetic C57BL/6 mice were subcutaneously implanted with osmotic pumps that supplied human recombinant insulin for 4 wk (HGM). Aortic segments were cut longitudinally and used for the visualization of ROS generation, inflammation, and apoptosis in the endothelium by confocal microscopy. A) Scheme for generating DM and HGM mice. B, C) Body weight (B) and blood glucose levels (C) were monitored weekly (n = 10). D) The endothelium of mouse aortas was visualized with staining for VE-cadherin (red) and nuclear counterstaining with DAPI (blue). E–H) ROS generation in the endothelium of mouse aortas was visualized using dihydroethidium (DHE) (E) [quantitative analysis using fluorescence intensity of DHE (F)] and H₂DCFDA (G) [quantitative analysis using fluorescence intensity of H₂DCFDA (H)] in the aortic endothelium (n = 6). Scale bars, 50 μm. Results are expressed as means ± sd of 6 or 10 independent experiments. ***P < 0.001.

Figure 5.

Absence of HGM-induced expression of inflammatory adhesion molecules in the aortic endothelium of diabetic (DM, diabetes mellitus) TGM2^−/− mice and in human aortic endothelial cells treated with TGase2 siRNA or ROS scavengers. A–D) DM C57BL/6 (TGM2^+/+, gray bars) and TGase2-null (TGM2^−/−, blue bars) mice were supplemented for 4 wk with insulin (HGM). Expression of ICAM-1 was visualized in the aortic endothelium (A) and quantitatively analyzed using the fluorescence intensities in the aortic endothelium (n = 8) (B). Expression of VCAM-1 was visualized in the aortic endothelium (C) and quantitatively analyzed using the fluorescence intensities in the aortic endothelium (n = 8) (D). E–G) Human aortic endothelial cells were treated for 6 d with normal glucose (NG), high glucose (HG), or high glucose followed by normal glucose (HGM) in the presence of 100 nM control (Ctrl) or human TGase2 siRNA, 1 nM NAC, or 0.5 mM Trolox. Expression of ICAM-1 and VCAM-1 was analyzed by Western blot (E), and expressions of ICAM-1 (F) and VCAM-1 (G) were quantified by densitometry (n = 3). NS, nonsignificant. Scale bar, 50 μm. Results are expressed as means ± sd of 3 or 8 independent experiments. **P < 0.01, ***P < 0.001.

Figure 6.

Absence of HGM-induced endothelial apoptosis in the aortic endothelium of diabetic (DM, diabetes mellitus) TGM2^−/− mice and in human aortic endothelial cells treated with TGase2 siRNA or ROS scavengers. A–D) DM C57BL/6 (TGM2^+/+, gray bars) and TGase2-null (TGM2^−/−, blue bars) mice were supplemented for 4 wk with insulin (HGM). Active caspase-3–positive cells (red) were visualized with nuclear counterstaining with DAPI (blue) in the aortic endothelium (A) and quantitatively analyzed by confocal microscopy (n = 8) (B). Apoptotic cells in the aortic endothelium (green, indicated by arrows) were stained by TUNEL with nuclear counterstaining with DAPI (blue) and visualized by confocal microscopy (C) and quantitatively analyzed by confocal microscopy (n = 8) (D). E, F) Human aortic endothelial cells were treated for 6 d with normal glucose (NG), high glucose (HG), or high glucose followed by normal glucose (HGM) in the presence of 100 nM control (Ctrl) or human TGase2 siRNA, 1 nM NAC, or 0.5 mM Trolox. The percentage of active caspase-3–positive cells was determined by confocal microscopy (n = 6) (E). Apoptotic cells were analyzed by TUNEL staining (n = 6) (F). NS, nonsignificant. Scale bar, 100 μm. Results are expressed as means ± sd of 6 or 8 independent experiments. ***P < 0.001.

The vicious cycle between TGase2 activation and ROS generation in the aortic endothelium of HGM mice

We tested whether TGase2 plays a key role in the molecular mechanism of endothelial dysfunction in HGM mice. In vivo TGase activity was highly elevated in diabetic mice compared with that in control mice, and the hyperglycemia-induced TGase activation was sustained in HGM mice after blood glucose normalization (Fig. 2A, B). Oral administration of the TGase inhibitor Cys attenuated the HGM-induced TGase activation in HGM + Cys mice. In contrast to that in wild-type (TGM2^+/+) diabetic or HGM mice, TGase activity was not elevated in TGase2-null (TGM2^−/−) diabetic or HGM mice (Fig. 3A), demonstrating that TGase2 mostly contributed to the hyperglycemia-induced, sustained TGase activation in the aortic endothelium.

Figure 2.

The vicious cycle between TGase2 and ROS in the aortic endothelium of HGM mice. Diabetic (DM, diabetes mellitus) C57BL/6 mice were subcutaneously implanted with osmotic pumps that supplied human recombinant insulin (HGM) or insulin and CP (HGM + CP) for 4 wk. Another group of DM mice was supplemented for 4 wk with insulin by osmotic pump and with Cys by oral intake (HGM + Cys). TGase activity and ROS generation in the aortic endothelium of the mice were visualized and quantitatively analyzed by confocal microscopy. A, B) HGM-induced TGase activation in the aortic endothelium was inhibited by Cys, CP, or ex vivo treatment with NAC or Trolox. TGase activity was visualized by confocal microscopy (A) and quantitatively analyzed using the fluorescence intensities [n = 8 except for NAC and Trolox (n = 6)] (B). C, D) Cys and CP inhibit hyperglycemia-induced ROS generation in the aortic endothelium of HGM mice. ROS generation was visualized using H₂DCFDA (C); ROS levels were quantitatively analyzed using fluorescence intensities (n = 8) (D). Scale bars, 50 μm. Results are expressed as means ± sd of 6 or 8 independent experiments. ***P < 0.001.

Figure 3.

Absence of HGM-induced oxidative stress in the aortic endothelium of diabetic (DM, diabetes mellitus) TGM2^−/− mice. After supplementing diabetic C57BL/6 (TGM2^+/+, gray bars) and TGase2-null (TGM2^−/−, blue bars) mice for 4 wk with insulin (HGM), TGase activity and ROS generation in the endothelium of the mouse aortas were visualized and quantitatively analyzed by confocal microscopy. A) In vivo TGase activity was quantitatively analyzed using the fluorescence intensities in the aortic endothelium (n = 6). B–E) ROS generation was visualized using H₂DCFDA (B) [quantitative analysis using fluorescence intensity of H₂DCFDA (C)] and dihydroethidium (D) [quantitative analysis using fluorescence intensity of dihydroethidium (E)] in the aortic endothelium (n = 6). NS, nonsignificant. Scale bars, 50 μm. Results are expressed as means ± sd of 6 independent experiments. ***P < 0.001.

Aortic endothelial ROS levels were elevated in diabetic and HGM mice compared with those in control mice, and the elevation was inhibited by Cys administration in HGM + Cys mice (Fig. 2C, D), indicating the important role of TGase in HGM-induced ROS generation. In contrast to those in diabetic or HGM wild-type (TGM2^+/+) mice, the ROS levels were not elevated in diabetic or HGM TGM2^−/− mice (Fig. 3B–E), demonstrating the essential role of TGase2 in the sustained, hyperglycemia-induced ROS generation in the aortic endothelium of HGM mice.

We next treated aortic segments of HGM mice with the ROS scavengers N-acetylcysteine (NAC) and Trolox to study the role of ROS generation in the HGM-induced TGase2 activation. NAC and Trolox each prevented the sustained HGM-induced TGase2 activation in the aortic endothelium of HGM mice (Fig. 2A, B), demonstrating the essential role of ROS generation in the HGM-induced TGase2 activation. Furthermore, we supplemented HGM mice with CP for 4 wk to study the role of ROS generation in the HGM-induced TGase2 activation, because CP inhibits ROS generation in the aortic endothelium of diabetic mice (26). The CP supplementation prevented ROS generation and TGase2 activation in the aortic endothelium of the HGM + CP mice (Fig. 2). Our results demonstrate that a vicious cycle between TGase2 activation and ROS generation occurs in the aortic endothelium of HGM mice.

The vicious cycle between TGase2 activation and ROS generation in human aortic endothelial cells

To study the potential role of the vicious cycle between TGase2 activation and ROS generation in human aortic endothelial dysfunction, we investigated HGM-induced TGase2 activation and ROS generation in human aortic endothelial cells. Exposure to high glucose concentration increased in situ TGase activity, which persisted after glucose normalization (Fig. 4A, B), demonstrating that TGase is involved in HGM in human endothelial cells. The sustained TGase activation was inhibited in human endothelial cells that were treated with the ROS scavengers NAC or Trolox during glucose normalization, demonstrating that intracellular ROS levels are associated with persistent TGase activation (Fig. 4B). Treatment of the endothelial cells with the TGase inhibitors monodansylcadaverine and Cys prevented sustained ROS generation (Fig. 4C), suggesting that TGase activation and ROS generation were in a cycle.

Figure 4.

The vicious cycle between TGase2 and ROS in HGM in human aortic endothelial cells. Human aortic endothelial cells (n = 6) were treated for 6 d with normal glucose (NG) or high glucose (HG) or with high glucose for 3 d followed by normal glucose for 3 d (HGM) in the presence of 1 mM NAC or 0.5 μM Trolox and 20 μM monodansylcadaverine (MDC) or 50 μM Cys. Treatment with 30 mM mannitol (MN) was used as an osmotic control. In situ TGase activity and intracellular ROS levels were determined by confocal microscopy. A) Representative images of in situ TGase activity. Scale bar, 50 μm. B) Inhibitory effects of ROS scavengers on HGM-induced TGase activation. C) Inhibitory effects of TGase inhibitors on HGM-induced intracellular ROS generation. D–F) Endothelial cells were transfected with 100 nM control (Ctrl) or human TGase2 siRNA and subjected to HGM treatment. Expression of TGase2 protein was analyzed by Western blot (D). TGase2 siRNA prevented HGM-induced TGase activation (E) and ROS generation (F) (n = 6). Results are expressed as means ± sd of 6 independent experiments. ***P < 0.001.

We next investigated the contribution of TGase2 to the sustained TGase activation by transfecting endothelial cells with human TGase2-specific siRNA. TGase2 siRNA suppressed TGase2 protein expression and TGase activation, whereas control siRNA did not (Fig. 4D, E). Continued ROS generation after glucose normalization was prevented by TGase2 siRNA but not by control siRNA (Fig. 4F). Those results demonstrate that transient exposure to high glucose concentration induces TGase2 activation and ROS generation that persists after glucose normalization in human aortic endothelial cells.

The role of TGase2 and ROS in aortic endothelial inflammation and apoptosis

We used TGM2^−/− mice to investigate the role of TGase2 in aortic endothelial inflammation in HGM mice. The expression of the inflammatory adhesion molecules ICAM-1 and VCAM-1 was elevated in the aortic endothelium of diabetic and HGM wild-type (TGM2^+/+) mice (Fig. 5A–D). By contrast, there was no detectable increase in the expression of ICAM-1 and VCAM-1 in the aortic endothelium of diabetic or HGM TGM2^−/− mice (Fig. 5A–D).

We used TGase2 siRNA and the ROS scavengers NAC and Trolox to study the role of TGase2 and ROS in the sustained, HGM-induced inflammation of human aortic endothelial cells. Exposure to high glucose concentration increased the expression of the inflammatory adhesion molecules ICAM-1 and VCAM-1, which persisted after glucose normalization (Fig. 5E–G). TGase2 siRNA prevented the sustained expression of ICAM-1 and VCAM-1, whereas control siRNA did not. The ROS scavengers also inhibited the HGM-induced, persistent expression of ICAM-1 and VCAM-1 (Fig. 5E–G).

We next studied the function of TGase2 and ROS in aortic endothelial apoptosis in HGM mice. The percentage of active caspase-3–positive cells increased in the aortic endothelium of diabetic or HGM wild-type mice, but not in that of diabetic or HGM TGM2^−/− mice (Fig. 6A, B). The number of TUNEL-positive cells also increased in the aortic endothelium of the diabetic or HGM wild-type mice, but not in that of diabetic or HGM TGM2^−/− mice (Fig. 6C, D). Furthermore, TGase2 siRNA inhibited the HGM-induced apoptosis of human aortic endothelial cells, as assessed by measurement of active caspase-3–positive cells and TUNEL staining (Fig. 6E, F). The apoptosis induced by HGM was also inhibited by the ROS scavengers NAC and Trolox (Fig. 6E, F). Overall, our results suggest that the TGase2-ROS vicious cycle plays an important role in the aortic endothelial inflammation and apoptosis induced by HGM.

The roles of p53 and p66^Shc in the TGase2-ROS vicious cycle

We used human p53-specific siRNA to examine the function of p53 in the persistent TGase2 activation and ROS generation in human aortic endothelial cells, because p53 overexpression is associated with persistent ROS generation (31). High-glucose treatment increased the expression of p53, and the elevated expression was sustained after glucose normalization (Fig. 7A). Transfection of endothelial cells with human p53 siRNA, which blocked the expression of p53, significantly suppressed TGase2 expression (Fig. 7A). By contrast, transfection of the cells with TGase2-specific siRNA had no effect on HGM-induced p53 expression (unpublished results). The p53 siRNA also inhibited the persistent TGase2 activation and intracellular ROS generation induced by HGM treatment, whereas the control siRNA had no effect (Fig. 7B, C). Our findings suggest that p53 plays an important role as an upstream regulator of the TGase2-ROS vicious cycle in human aortic endothelial cells.

Figure 7.

The roles of p53 and p66^Shc in the TGase2-ROS vicious cycle. A–F) Human aortic endothelial cells were treated for 6 d with normal glucose (NG), high glucose (HG), or high glucose followed by normal glucose (HGM) in the presence of 100 nM control siRNA (Ctrl), human p53 siRNA, human p66^Shc siRNA, or human TGase2 siRNA. Phosphorylation of p66^Shc at Ser36 and expression of p53, TGase2, p66^Shc, and β-actin were analyzed by Western blot; human p53 siRNA suppressed TGase2 expression (A). In situ TGase activity (B) and intracellular and mitochondrial ROS levels (C), with HGM-induced elevation, were measured by confocal microscopy (n = 6), showing the effects of human p53 and p66^Shc siRNAs. Human TGase2 siRNA had no effect on HGM-induced p66^Shc expression but inhibited HGM-induced p66^Shc phosphorylation (p-p66^Shc) at Ser36, as shown by Western blot analysis (D) and densitometry quantification of p66^Shc expression (E) and phosphorylation of p66^Shc at Ser36 (n = 3) (F). G, H) Diabetic (DM, diabetes mellitus) C57BL/6 (TGM2^+/+, gray bars) and TGase2-null (TGM2^−/−, blue bars) mice were supplemented for 4 wk with insulin (HGM), and p66^Shc phosphorylation at Ser36 in the aortic endothelium was visualized (scale bar, 50 μm) (G) and quantitatively analyzed by confocal microscopy (n = 8) (H). NS, nonsignificant. **P < 0.01, ***P < 0.001.

We next transfected aortic endothelial cells with human TGase2-specfic siRNAs to study the role of TGase2 in p66^Shc activation, because p66^Shc is associated with sustained ROS generation (16). High-glucose treatment increased p66^Shc expression and p66^Shc-activating phosphorylation at the Ser36 residue, and those changes persisted after glucose normalization (Fig. 7D–F). Transfection with TGase2 siRNA suppressed the HGM-induced p66^Shc phosphorylation but did not affect the HGM-induced p66^Shc expression (Fig. 7D–F). Those results suggest that TGase2 is involved in p66^Shc activation in HGM.

We further investigated the role of TGase2 as the upstream activator of p66^Shc in HGM TGM2^−/− mice. The phosphorylation of p66^Shc was highly elevated in the aortic endothelium of diabetic wild-type mice compared with that in control mice, and the elevation persisted after glucose normalization (Fig. 7G, H). By contrast, the p66^Shc phosphorylation was not elevated in diabetic or HGM TGM2^−/− mice (Fig. 7G, H). Those results indicate that TGase2 plays an important role in the persistent p66^Shc activation induced by hyperglycemia in the aortic endothelium of HGM mice.

We next studied the role of p66^Shc in the persistent TGase2 activation in HGM by transfecting aortic endothelial cells with human p66^Shc-specific siRNA. The p66^Shc siRNA blocked p66^Shc expression (unpublished results) and prevented persistent TGase2 activation in HGM-treated aortic endothelial cells (Fig. 7B). The p66^Shc-specific siRNA also inhibited the sustained elevation of intracellular ROS levels (Fig. 7C), suggesting that p66^Shc is involved in the persistent TGase2 activation and ROS generation in HGM. Taken together, our results suggest that p66^Shc plays a role as an intermediator of the vicious cycle between TGase2 activation and ROS generation in HGM (Fig. 8).

Figure 8.

Schematic model depicting the HGM-induced persistence of endothelial inflammation and apoptosis maintained through TGase2 activation and ROS generation. p-p66^Shc, phosphorylated p66^Shc.

DISCUSSION

Chronic hyperglycemia develops into microvascular and macrovascular complications and leads to diabetes-associated morbidity and mortality (4). The Diabetes Control and Complications Trial–Epidemiology of Diabetes Interventions and Complications and UK Prospective Diabetes Study epidemiologic studies demonstrated that poor glycemic control can be remembered by the vascular system and that HGM can lead to long-term vascular complications after glycemic control is achieved in patients with type 1 or type 2 diabetes (2). It has been a challenge to understand the underlying mechanisms of HGM in order to improve diabetic complications associated with persistent hyperglycemic stress. In this study, we present a new molecular mechanism of HGM that is important for HGM-induced endothelial dysfunction. Hyperglycemia induced sustained TGase2 activation in the aortic endothelium of HGM mice and also in human aortic endothelial cells after glucose normalization, and the sustained TGase2 activation was in a vicious cycle with ROS generation. The vicious cycle regulated the sustained expression of the inflammatory adhesion molecules ICAM-1 and VCAM-1 as well as apoptosis in the aortic endothelium of HGM mice and among human aortic endothelial cells (Fig. 8). The regulation of the inflammatory adhesion molecules might involve Set7, because Set7 is associated with ROS-mediated up-regulation of NF-κB activation and persistent vascular inflammation (18, 19). Thus, TGase2 activation and ROS generation together make up a key mechanism of HGM and HGM-associated endothelial dysfunction, which suggests that they can be targeted to treat diabetic vascular complications.

TGase2 is implicated in diabetic vasculopathy and retinopathy caused by endothelial apoptosis and vascular leakage (24, 25). CP prevents hyperglycemia-induced endothelial apoptosis by inhibiting ROS-mediated TGase2 activation in the aorta, heart, and renal cortex of diabetic mice (25). In the retina of diabetic mice, TGase2 plays an essential role in hyperglycemia-induced vascular leakage by stimulating stress fiber formation and VE-cadherin disruption in endothelial cells (24). Although TGase2 is important in hyperglycemia-induced vascular dysfunction, its role in HGM is not known. In this study, we demonstrated the key role of TGase2 in HGM-associated endothelial dysfunction. Transient exposure to high glucose concentration caused TGase2 activation that persisted after glucose normalization, and silencing of TGase2 expression suppressed the sustained apoptosis and expression of ICAM-1 and VCAM-1 in human aortic endothelial cells. We found sustained elevation of TGase2 activity in the aortic endothelium of HGM mice whose blood glucose levels had been normalized by insulin supplementation. Furthermore, increases in apoptosis and the expression of inflammatory adhesion molecules were not detectable in the aortas of TGM2^−/− HGM mice. Thus, TGase2 is likely the key enzyme involved in vascular dysfunction induced by hyperglycemia and HGM, leading to long-term diabetic vascular complications.

The mechanism that keeps TGase2 activation in a vicious cycle with sustained ROS generation in diabetes is not clear, but the cycle might be explained by TGase2 regulation of GAPDH. TGase2 inhibits GAPDH by catalyzing cross-linking between lysine residues and polyglutamine repeats (32, 33). The inhibition of GAPDH causes vascular damage via a number of pathogenic pathways, including PKC activation and increased methylglyoxal formation, resulting in intracellular AGE formation (1, 3). PKC-βII activates NADPH oxidase and p66^Shc, which are associated with sustained ROS generation (16). Increased AGE levels also produce intracellular ROS by binding to the AGE receptors (3). Thus, TGase2 may cause sustained ROS generation by inhibiting GAPDH. Another mechanism might be TGase2 interaction with the apoptosis regulator, B-cell lymphoma 2–associated X (Bax). The BH-3 domain of TGase2 induces Bax to undergo a conformational change and translocate to the mitochondria, leading to the release of cytochrome c (34), which is associated with ROS generation (35). Taken together, those findings suggest that activated TGase2 can induce persistent ROS generation by inhibiting GAPDH, causing the translocation of Bax to the mitochondria, or both, although other potential mechanisms are not excluded.

It has been reported that persistent expression of transcription factor p53 is associated with sustained ROS generation (7, 29). The mitochondrial adaptor protein p66^Shc is an important mediator of persistent vascular hyperglycemic stress effects such as sustained ROS generation, elevation of caspase-3 activity, and apoptosis (7, 16). Therefore, we investigated the potential roles of p53 and p66^Shc in the TGase2-ROS vicious cycle. We found that the silencing of p53 suppressed TGase2 expression, but silencing of TGase2 had no effect on p53 expression. The continued TGase activation and ROS generation were prevented by p53 siRNA. Those results suggest that p53 plays a role as an upstream regulator of the vicious cycle. We then investigated the association of p66^Shc with the TGase2-ROS vicious cycle. TGase2 siRNA suppressed the sustained phosphorylation of p66^Shc at Ser36 in human aortic endothelial cells. The p66^Shc-activating phosphorylation was also elevated in HGM wild-type mice, but not in HGM TGM2^−/− mice, demonstrating the important role of TGase2 in hyperglycemia-induced p66^Shc activation in HGM. Interestingly, silencing of p66^Shc inhibited the sustained TGase2 activation and ROS generation induced by transient exposure to high glucose concentration in aortic endothelial cells, elucidating the regulation of TGase2 by p66^Shc. Thus, it is likely that there is a vicious cycle between ROS generation and sustained activation of TGase2 and p66^Shc in HGM, which is regulated by p53.

It is now plausible that HGM is the underlying mechanism of long-term vascular complications caused by hyperglycemia. We found that TGase2 activation and ROS generation were responsible for HGM-associated endothelial dysfunction. Therefore, TGase2 inhibitors, including peptides as well as amine and thiol compounds, might be candidates to improve microvascular and macrovascular diabetic complications. CP is secreted from pancreatic β-cells and inhibits hyperglycemia-induced endothelial apoptosis through ROS-mediated TGase2 activation in the aorta, heart, and renal cortex of diabetic mice (25). CP also prevents hyperglycemia-induced vascular leakage and metastasis of melanoma cells by inhibiting TGase2 in the lungs of diabetic mice (30). TGase2 plays an essential role in hyperglycemia-induced vascular leakage in the retinas of diabetic mice (24), which was prevented by CP, the benzodiazepine anesthetic midazolam, or cysteamine derived from coenzyme A degradation (27, 36, 37). In the present study, oral administration of Cys inhibited hyperglycemia-induced TGase2 activation and ROS generation in the aortas of HGM mice. These results suggest that TGase2 inhibitors might potentially be used as drugs to prevent HGM-associated diabetic complications.

In summary, we found that the vicious cycle between TGase2 activation and ROS generation is a key underlying mechanism of HGM. The new mechanism is associated with sustained expression of inflammatory adhesion molecules and increased apoptosis in the aortic endothelium of HGM mice with normalized blood glucose levels and in human aortic endothelial cells after glucose normalization. The TGase2-ROS vicious cycle is important in vascular dysfunction induced by HGM and therefore represents a potential therapeutic target for the treatment of diabetic vascular complications.

Glossary

AGE

advanced glycation end product

Bax

B-cell lymphoma 2–associated X

CP

C-peptide

Cys

cystamine

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

H2DCFDA

2′,7′-dichlorodihydrofluorescein diacetate

H3K4me1

monomethylation of histone H3 at lysine 4

HGM

hyperglycemic memory

ICAM-1

intercellular adhesion molecule 1

NAC

N-acetylcysteine

ROS

reactive oxygen species

Set

Su(var), enhancer-of-zeste, trithorax

siRNA

small interfering RNA

TGase

transglutaminase

TGase2

transglutaminase 2

TGM2

TGase2

VE-cadherin

vascular endothelial cadherin

AUTHOR CONTRIBUTIONS

J.-Y. Lee and Y.-J. Lee designed and performed experiments, analyzed data, and wrote the manuscript; H.-Y. Jeon, E.-T. Han, W. S. Park, S.-H. Hong, and Y.-M. Kim designed experiments and analyzed data; K.-S. Ha conceptualized the study, designed experiments, analyzed and interpreted data, and wrote the manuscript; and all authors approved the final version of the manuscript.