Introduction

Hyperglycaemia is a key risk factor for microvascular complications leading to excess morbidity and mortality in diabetic patients [1]. Microangiopathy, damage to the small blood vessels and capillaries, is a direct result of chronic hyperglycaemia and is the main cause of major complications such as retinopathy and nephropathy [2, 3]. Dysfunction of the vascular endothelium is considered to be an important factor in the initiation, progression and clinical outcome of diabetic microangiopathy [4]. One hypothesis as to how hyperglycaemia leads to endothelial dysfunction and, consequently, vascular complications is the formation of advanced glycation end-products (AGEs) [5].

AGEs are a heterogeneous family of non-enzymatically modified proteins, which are increased in patients with diabetes. In addition to the formation of AGEs by carbohydrates in the classical Maillard reaction, intracellular glucose-derived glycolytic intermediates, such as methylglyoxal (MGO), glyoxal (GO) and 3-deoxyglucosone (3-DG), also form AGEs [6]. Because these intracellular dicarbonyl compounds form many more glycated proteins and at a faster rate than do equimolar amounts of glucose, intracellular AGE formation is thought to play an important role in the link between AGEs and diabetic complications. Indeed, intracellular AGEs have been implicated in activating intracellular signalling pathways as well as in modifying the function of intracellular proteins, thereby contributing to diabetic vascular complications [7]. The very reactive oxo-aldehyde MGO has been identified as the major precursor in the formation of intracellular AGEs in endothelial cells [8]. MGO is formed mainly by fragmentation of the triose phosphates glyceraldehyde-3-phosphate and dihydroxy-acetone-phosphate and is, under physiological circumstances, efficiently detoxified to d-lactate by the glyoxalase system, in which glyoxalase-I (GLO-I) is the rate limiting enzyme [9]. MGO is known to have detrimental effects on cellular function and it is also evident that elevated levels of MGO are responsible for renal oxidative stress, as demonstrated in diabetic rats [10]. Most importantly, progression of hypertension and diabetic nephropathy in humans is also significantly related to increased levels of MGO [11, 12].

We previously showed in a rat model of diabetes that Glo1 overexpression detoxifies MGO and GO, and thereby decreases AGEs and markers of oxidative stress [13]. Furthermore, we recently demonstrated in an ex vivo model that high glucose levels and short-term diabetes-induced impairment of endothelium-dependent vasorelaxation in rat mesenteric arteries are mediated by intracellular MGO levels [14]. In addition, Glo1 overexpression protects against diabetes-induced retinal neuroglial and vasodegenerative pathology [15]. However, the involvement of the major AGE precursor MGO in long-term diabetes-induced endothelial dysfunction and, consequently, the development of renal microvascular complications remain largely unknown.

Therefore, the present study explored the effect of Glo1 overexpression on microvascular and renal function 24 weeks after streptozotocin (STZ)-induced diabetes in the rat. We hypothesised that elevated levels of MGO are involved in microvascular dysfunction and in the development of diabetic renal changes, and that Glo1 overexpression can prevent this.

Methods

Animal model

Heterozygous transgenic Glo1-overexpressing rats were crossed with wild-type Wistar rats (Nippon Seibutsu Zairyo Center, Saitama, Japan) to obtain enough Glo1-overexpressing progeny for the experiment [13]. Young adult wild-type and Glo1 transgenic male rats (10 weeks of age) were rendered diabetic for a period of 24 weeks by a single intravenous injection of STZ (Sigma-Aldrich, Zwijndrecht, the Netherlands; 45 mg/kg body weight). Only rats with fasting blood glucose levels above 15 mmol/l were included in the study (Contour test strips, Bayer Health Care, Leverkusen, Germany). Weight- and age-matched control wild-type rats were not injected. The rats were divided into three groups: wild-type control group (WtC; n = 9), wild-type STZ-induced diabetic group (WtD; n = 8) and Glo1-overexpressing STZ-induced diabetic group (TgD; n = 13).

The study protocols were approved by the Committee on Animal Experimentation and Welfare of Maastricht University.

Twenty-four weeks after the induction of diabetes, rats were killed and mesenteric arteries, kidneys, urine and EDTA blood were processed for further analysis.

Blood pressure and heart rate measurement

Under isoflurane (Baxter, Deerfield, IL, USA) anaesthesia a PE-10 catheter was inserted in the abdominal aorta via the femoral artery and connected to a pressure transducer (Miller Instruments, Houston, TX, USA). The pressure signal was digitally sampled at 2 kHz and mean arterial pressure (MAP) and heart rate were calculated over a 10–15 min time period after haemodynamic variables were stabilised.

Measurement of GLO-I activity

GLO-I activity was assayed by spectrophotometry (Synergy, BioTek, Winooski, VT, USA), by monitoring the increase in absorbance at 240 nm due to the formation of S-d-lactoylglutathione for 10 min at 37°C according to the method of McLellan and Thornalley [16].

Measurement of vascular function, morphology and gene expression

Vascular reactivity was measured as described earlier [14]. For further details see electronic supplementary materials [ESM] Methods.

On paraformaldehyde-fixed mesenteric arteries, media cross-sectional area, media thickness and lumen diameter were determined using Leica Qwin software (Leica, Groot Bijgaarden, Belgium) on 4 μm sections stained with Lawson’s solution (Boom, Meppel, the Netherlands), a classic elastin stain. To stain collagen, cross-sections were deparaffinised and incubated in phosphomolybdenic acid (0.2% wt/vol.) for 5 min, followed by incubation with Sirius red (Sigma-Aldrich). After washing with 0.1 mol/l HCl for 2 min, sections were dehydrated and protected with coverslips. Collagen content was calculated as the percentage collagen-positive area per total area.

After homogenisation of vascular tissue by the use of glass beads and a mini-bead beater, total cellular RNA was extracted using TRIzol isolation (Invitrogen, Life Technologies, Paisley, UK). Reverse transcription was performed with a Reverse Transcriptase kit from Invitrogen. Subsequently, quantitative PCR (qPCR) assays were performed using Quanta Universal PCR MasterMix on an ABI PRISM 7900 HT (Applied Biosystems, Foster City, CA, USA) and a primer of the target (i.e. vascular cell-adhesion molecule-1 [VCAM-1], intercellular adhesion molecule-1 [ICAM-1], endothelial nitric oxide synthase [eNOS]) or a primer of the housekeeping gene, β-actin, all purchased from Eurogentec (Liège, Belgium).

Immunohistochemical and histological examinations

Mesenteric arteries and kidneys were fixed with phosphate-buffered (pH 7.4) formaldehyde (4% vol./vol.) overnight at room temperature. Subsequently, arteries and kidneys were transferred to 70% vol./vol. ethanol, embedded in paraffin and processed for histological examination.

Anti-GLO-I (1:1,000; see ESM Methods), anti-N ε-(1-carboxymethyl)lysine (CML) (1:4,000; homemade) [17], anti-N δ-(5-hydro-5-methyl-4-imidazolon-2-yl)-l-ornithine (MG-H1) (1:100,000, generous gift from M. Brownlee), anti-eNOS (1:5,000; BD transduction laboratories, Breda, the Netherlands), anti-Wilms’ tumour-1 (WT-1) (undiluted; DAKO, Glostrup, Denmark), and anti-active-caspase-3 (1:200; Cell Signaling Technology, Danvers, MA, USA) were used as primary antibodies. After washing in PBS (pH 7.4), sections were incubated for 30 min with the appropriate biotin-labelled antibody (1:500) at room temperature and subsequently washed in PBS. After incubation with streptavidin–horseradish peroxidase (1:200, DAKO) for 60 min at room temperature, peroxidase was visualised with 3,3-diamino-benzidine-tetrahydrochloride/H2O2 (Sigma-Aldrich) for 3–5 min. Immunostaining was scored for anatomical localisation and its intensity or number of positive cells. For the intensity scoring, each positive vessel was given a score: 0, absent; 1, weak positivity; 2, moderate positivity; 3, strong positivity. Each multiplication score was then added and the sum was divided by the number of samples, resulting in an immunohistochemical score. All stainings were scored by two blinded observers.

The middle part of the right kidney (containing the papilla) was embedded in paraffin for examination. Two micron-thick sections were cut on a rotation microtome and stained with p-aminosalicylic acid and haematoxylin (DAKO). Total glomerulus volume was measured observer-blinded at a magnification of ×400 as previously described [18].

Creatinine measurement in plasma and urine

Urinary and plasma creatinine levels were measured by ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MSMS) (UPLC Acquity and Micromass Quattro Premier XE Tandem Mass Spectrometer; Waters, Milford, MA, USA) as described elsewhere [19]. Internal standard creatinine (methyl-d3) (CDN Isotopes, Quebec, Canada) was mixed with deproteinised plasma or diluted urine and subsequently analysed by UPLC-MSMS.

Biomarker detection

Rat kidney injury panel 1 was purchased from Meso Scale discovery (Rockville, MD, USA). This assay detects lipocalin-2, osteopontin and kidney-injury-molecule (KIM-1) in a sandwich immunoassay and uses a competitive assay format to detect albumin. Each 96-well plate had four carbon electrodes in the bottom of each well; each pre-coated with one of the four anti-kidney-injury marker antibodies of interest. The intra-assay variations of albumin, lipocalin-2, osteopontin and KIM-1 were 5.0%, 4.0%, 8.4% and 3.3%, respectively. Monocyte chemotactic protein-1 (MCP-1) was measured with MSD kit K151AYB-2 (Meso Scale) according to the manufacturer’s protocol (intra-assay variation was 3.3%). The markers ICAM-1 (Diaclone, Ann Arbor, MI, USA), IL-6 (Sanquin, Amsterdam, the Netherlands) and nephrin (Exocell, Philadelphia, PA, USA) were measured using standard ELISA procedures (according to the manufacturer’s protocol) and the intra-assay variations were 2.8%, 1.6% and 9.1% respectively.

Plasma nitrite and nitrate levels were determined using a commercially available colorimetric assay kit (Cayman Chemical, Ann Arbor, MI, USA) based on the Griess reaction. Before the assay, the plasma samples were ultrafiltered though a 30 kDa molecular mass cut-off filter (Ultrafree-MC centrifugal filter units; Millipore Corporation, Bedford, MA, USA) to reduce the background absorbance caused by haemoglobin.

Measurement of AGEs and oxo-aldehydes

The AGEs CML, N ε-(1-carboxyethyl)lysine (CEL) and MG-H1 were measured using UPLC-MSMS (Waters) as described earlier [20]. Oxo-aldehydes GO, MGO and 3-DG in plasma were assayed by derivatisation with o-phenylenediamine and UPLC-MSMS [21].

In vitro cell culture experiments

Human immortalised endothelial cells (ECRF-24) were cultured for Glo1 silencing experiments and conditionally immortalised mouse podocytes were cultured for apoptosis assays. For further details see ESM Methods.

Statistics

All values are expressed as mean ± SEM and p < 0.05 was considered statistically significant. For further details see ESM Methods.

Results

General physical and biochemical characteristics

Injection of the rats with STZ produced characteristic symptoms of diabetes, including hyperglycaemia, decreased body weight gain, hyperphagia, polydipsia and polyuria, without any differences between Glo1 genotypes (Table 1). Diabetes resulted in significant increased levels of GO, MGO and 3-DG. Glo1 overexpression decreased GO and MGO, but not 3-DG. The increased GLO-I activity in diabetes, and predominantly in the transgenic rats, was accompanied by significant increased levels of urinary d-lactate, the detoxified metabolite of MGO. In addition, 24 weeks of diabetes also significantly increased the levels of CML, CEL and MG-H1 in plasma. Both CEL and MG-H1 were significantly lower in the transgenic rats while CML was not altered (Table 1).

Table 1 General physical and biochemical characteristics of the rats after 24 weeks of diabetes
Full size table

Both MAP and heart rate were decreased in the wild-type diabetic rats compared with the control rats (Table 1). The decrease in blood pressure after 24 weeks of diabetes was significantly attenuated by Glo1 overexpression, although heart rate was not statistically influenced by the overexpression of Glo1.

Diabetes-induced AGE formation in rat mesenteric arteries is prevented by Glo1 overexpression

The mesenteric arteries of the Glo1-overexpressing rats showed an increase in GLO-I protein levels and GLO-I activity after 24 weeks of diabetes as compared with control and wild-type diabetic rats (Fig. 1a–d). The immunohistochemical stainings for CML (Fig. 1e–h) and MG-H1 (Fig. 1i–l) in rat mesenteric arteries were both significantly elevated in the wild-type diabetic group as compared with the control rats. Staining for MG-H1 was predominantly localised in the endothelial cells, while CML was also present in the smooth muscle cells and adventitia. Overexpression of Glo1 prevented this diabetes-induced elevation of both CML and MG-H1 in the diabetic mesenteric arteries.

Fig. 1
figure 1

Immunohistochemical staining of GLO-I (a–c) and the GLO-I enzyme activity (d) showed a functionally active overexpression of GLO-I in mesenteric arteries in GLO-I STZ rats (TgD) compared with wild-type control (WtC) and wild-type diabetic (WtD) rats (***p < 0.001). Units (U) in nmol/min. (e–h) Diabetes resulted in increased levels of CML staining (f) in isolated rat mesenteric arteries compared with control (e). Quantification of these stainings in arbitrary units (h) showed an increase in CML after 24 weeks of diabetes (*p < 0.05, WtD vs WtC), which could be normalised by Glo1 overexpression († p < 0.05, TgD vs WtD) (g). (i–l) Diabetes resulted in increased levels of MG-H1 staining in isolated rat mesenteric arteries (j) compared with control (i). Quantification of these stainings in arbitrary units (l) showed an increase in MG-H1 after 24 weeks of diabetes (*p < 0.05, WtD vs WtC), which was normalised by GLO-I overexpression († p < 0.05, TgD vs WtD) (k). Scale bars in (a) 100 μm in the main image and 50 μm in the insert

Full size image

Morphological analyses of the mesenteric arteries showed no differences in mesenteric lumen radius, media cross-sectional area or media thickness after 24 weeks of diabetes (see ESM Table 1). Analyses of the mesenteric arteries for collagen deposition also showed no significant differences in collagen content corrected for medial area between the different groups.

Diabetes-induced impairment of vasorelaxation in rat mesenteric arteries is reduced by Glo1 overexpression

To investigate the function of the resistance arteries in diabetes and the effect of Glo1 overexpression, we measured endothelium-dependent and -independent vasorelaxation in isolated second-order mesenteric arteries of the rats in a wire myograph. Maximum contraction did not differ between the groups after 24 weeks of diabetes (active wall tension for wild-type control, wild-type diabetic and Glo1-overexpressing diabetic rats was 3.6 ± 0.3, 3.7 ± 0.3 and 3.3 ± 0.3 N/m, respectively). Diabetes resulted in a decrease in total acetylcholine-induced, endothelium-dependent, nitric oxide (NO)-mediated and also NO-independent vasorelaxation after 24 weeks of diabetes (Fig. 2a–c). Sodium nitroprusside (SNP)-induced endothelium-independent vasorelaxation was not altered due to diabetes (Fig. 2d). Decreased endothelium-dependent reactivity of the mesenteric arteries of the diabetic rats was significantly improved by Glo1 overexpression (for potency and efficacy values see ESM Table 2).

Fig. 2
figure 2

White circles, control (WtC) rats; black circles, diabetic (WtD) rats; grey circles, GLO-I diabetic (TgD) rats; ACh, acetylcholine. Log10 transformed data are given on the x-axis. (a) Diabetes resulted in a statistically significant rightward shift of the dose–response curve to ACh (*p < 0.05, compared with WtC). Overexpression of Glo1 partially prevented this impairment († p < 0.05, TgD vs WtD). (b) The concentration–response curve of the vasorelaxation due to endothelium-derived NO was significantly shifted after 24 weeks of diabetes (*p < 0.05, compared with WtC). Overexpression of Glo1 resulted in a trend toward improvement, but was only borderline significant after 24 weeks of diabetes (p = 0.08). (c) The concentration–response curve of the vasorelaxation due to eNOS-independent relaxation was also significantly shifted after 24 weeks of diabetes (*p < 0.05, compared with WtC). Overexpression of Glo1 resulted in a significant improvement after 24 weeks of diabetes († p < 0.05, TgD vs WtD). (d). Endothelium-independent SNP-induced vasorelaxation was not altered after 24 weeks of diabetes

Full size image

Modulation of GLO-I activity regulates the expression of endothelial activation and inflammation markers

We next investigated the effect of GLO-I modulation on other markers of endothelial dysfunction. We found increased expression of the endothelial dysfunction markers VCAM-1 and ICAM-1 in the diabetic rat mesenteric arteries (p < 0.05). The increases in the mRNA levels of these markers were completely prevented by Glo1 overexpression (Fig. 3a, b). However, the increase in plasma ICAM-1 in the wild-type diabetic group (22.9 ± 1.4 vs 13.9 ± 0.6 ng/ml), could not be corrected by Glo1 overexpression (26.0 ± 1.8 ng/ml), indicating a local effect of GLO-I. Furthermore there were no significant effects of diabetes or Glo1 overexpression on Nos3 mRNA expression (Fig. 3c) or protein levels (Fig. 3e–h) in the mesenteric arteries. Nevertheless total plasma levels of the NO metabolites nitrite and nitrate (NOx) were significantly decreased in the wild-type diabetic rats, an effect which was prevented in the Glo1-overexpressing rats (Fig. 3d).

Fig. 3
figure 3

Both Vcam1 (a) and Icam1 (b) mRNA expression levels in the mesenteric arteries of wild-type diabetic rats (WtD) were elevated after 24 weeks of diabetes (*p < 0.05, compared with control [WtC]), and Glo1 overexpression (TgD) prevented this († p < 0.05, TgD vs WtD). The mRNA expression levels of eNOS (Nos3 mRNA) (c) were not altered by either diabetes or Glo1 overexpression; plasma levels of NO metabolites nitrite and nitrate (NOx) were significantly decreased by diabetes but not in Glo1-overexpressing rats (d). (e–h) Diabetes and Glo1 overexpression did not result in altered eNOS protein levels as shown by immunohistochemical staining (e, WtC; f, WtD; g, TgD rats) and its quantification (h). Scale bars in (e) 100 μm in the main image and 50 μm in the insert

Full size image

In line with the in vivo data, we observed that in cultured human endothelial cells downregulation of Glo1 resulted in the upregulation of several genes involved in pathways of vascular dysfunction and inflammation. Small-interfering RNA (siRNA) against Glo1 resulted in a decrease in Glo1 expression (Fig. 4a) and GLO-I activity (Fig. 4b) by more than 60%. This decrease in GLO-I activity resulted in dysfunction of the cultured endothelial cells, as indicated by a mild but significant decrease in endothelial cell viability (Fig. 4c), and increased expression of genes representing endothelial inflammation (Fig. 4d) and activation (Fig. 4e). Protein levels of MCP-1, IL-6 and ICAM-1 were also elevated in the supernatant fractions of the cultured endothelial cells and supported the results of the qPCR (Fig. 4f–h).

Fig. 4
figure 4

Transfection of cultured human endothelial cells with siRNA against GLO1 (siGLO-I) resulted in a significant (*p < 0.05, compared with scrambled siRNA [siScr]) decrease in GLO1 expression (a) and activity (b). This resulted in an almost 10% decrease in cell viability (*p < 0.05, compared with siScrambled), as measured by MTT assay (c). Furthermore several genes involved in inflammation (MCP1 [also known as CCL2], IL6 and TNFα) (d) and endothelial activation (VCAM1 and ICAM1) (e) were upregulated by siGLO-I. Upregulation of gene expression was confirmed by increased protein levels of MCP-1 (f), IL-6 (g) and ICAM-1 (h) in the cell supernatant fractions

Full size image

Markers of early renal damage are attenuated by Glo1 overexpression after 24 weeks of diabetes

Because endothelial dysfunction and inflammation are considered to be important factors in the initiation and progression of vascular complications, we next determined the effect of Glo1 overexpression on the development of diabetes-induced renal dysfunction. In our rat model, 24 weeks of diabetes resulted in nephromegaly, which tended to be decreased by Glo1 overexpression (Table 2). Detailed morphometric analyses of the wild-type diabetic kidneys showed a statistically significant increase in glomerular volume, which was not significant, of the diabetic kidneys from Glo1-overexpressing rats compared with the controls. Furthermore, the wild-type diabetic rats showed signs of renal hyperfiltration, as indicated by the increase in body-weight-corrected creatinine clearance.

Table 2 Renal characteristics of the rats after 24 weeks of diabetes
Full size table

Immunohistochemical analyses showed increased presence of CML and MG-H1 in the mesangial matrix and peritubular capillaries of the diabetic rats, which could not be prevented by Glo1 overexpression. This observation was also confirmed by UPLC-MSMS analysis of total kidney lysates (data not shown). Furthermore, extensive histological analysis of the kidney did not show any collagen deposition or fibrotic or inflammatory differences in glomerular, tubulointerstitial or capillary tissue between the three groups after 24 weeks of diabetes (ESM Fig. 1).

We next focused on early markers of nephropathy in the diabetic rats. Immunohistological staining of renal podocytes showed a decrease in number in the wild-type diabetic glomeruli (Fig. 5b) compared with controls (Fig. 5a). Glo1 overexpression significantly attenuated this decrease (Fig. 5c, d). Increased staining in the diabetic glomeruli (Fig. 5e, f) of the apoptotic marker active-caspase-3 indicates that this decrease in podocyte number is, at least partially, apoptosis-dependent and that this could be prevented by Glo1 overexpression (Fig. 5g, h). Subsequently, Glo1 overexpression also partially prevented the diabetes-induced increase in urinary nephrin (p = 0.22), another marker for podocyte loss (Fig. 5i).

Fig. 5
figure 5

Podocyte number and number of apoptotic cells were determined by WT-1 and active-caspase-3 staining, respectively. Compared with control rats (a), wild-type diabetic rats (b) showed significantly (**p < 0.01) fewer numbers of podocytes per glomerulus. Overexpression of Glo1 could attenuate this († p < 0.05 compared with wild-type diabetic, WtD) (c), as shown by the quantification (d). The number of apoptotic cells was significantly (*p < 0.05) increased after 24 weeks of diabetes (f) compared with controls (e). This increase was prevented by Glo1 overexpression († p < 0.05, compared with WtD) (g, h). Scale bar in (a) 100 μm. In line with this, the urinary podocyte marker nephrin was mildly elevated in the urine of the wild-type diabetic rats (i). In cultured podocytes, incubation with 1,000 μmol/l MGO for 48 h resulted in an increase of both annexin-V (AnnV)-positive and AnnV/propidium iodide (PI)-double-positive cells compared with control cells. (j) Representive FACS data of control (−MGO) and MGO-incubated cells (+MGO). (k) Quantification of the AnnV-positive cells (white bars), AnnV/PI-double-positive cells (grey bars) and the combined AnnV-positive population (black bars) (***p < 0.001 compared with control)

Full size image

In line with the in vivo data, we observed that in fully differentiated cultured podocytes incubation with 1,000 μmol/l MGO for 48 h resulted in an increase in apoptotic cells compared with control cells (Fig. 5j). Early and late apoptotic cells were quantified with annexin-V and propidium iodide FACS, respectively (Fig. 5k). These data indicate that the apoptosis-induced decrease in podocytes in the diabetic glomerulus can at least be partially caused by MGO cytotoxicity.

To address the effects of these diabetes-induced glomerular alterations on the filtration barrier function we next determined the presence of early markers of renal damage in the urine of the rats. Urinary albumin, lipocalin-2, osteopontin and KIM-1 were all elevated after 24 weeks of diabetes (Fig. 6a–d). This elevation was attenuated for three of the four early markers (albumin, osteopontin and KIM-1) in rats overexpressing Glo1. The composite score for albumin, lipocalin-2, osteopontin and KIM-1 was significantly increased in the wild-type diabetic group and Glo overexpression decreased this combination of renal injury markers (Fig. 6e).

Fig. 6
figure 6

Twenty-four weeks of diabetes increased the levels of albumin (a), lipocalin-2 (b), osteopontin (c) and KIM-1 (d) (all *p < 0.05 compared with wild-type control, WtC), and this elevation was significantly attenuated by Glo1 overexpression for osteopontin and KIM-1 († p < 0.05 compared with wild-type diabetes, WtD). Lipocalin-2 was still significantly elevated in the TgD animals compared with WtD (*p < 0.05), but the statistical significant diabetes-induced increase in albumin was lost due to Glo1 overexpression. After 24 weeks of diabetes the renal injury composite score (albumin, lipocalin-2, osteopontin and KIM-1) was significantly increased (***p < 0.001 compared with WtC), and could be partially decreased by Glo1 overexpression († p < 0.05 compared with WtD)

Full size image

Discussion

In this study we showed that diabetes-induced impairment of vascular function, as measured by an impaired endothelium-dependent relaxation and an increased expression of markers for endothelial dysfunction, in mesenteric arteries of diabetic rats is prevented by overexpression of Glo1. In line with this, downregulation of Glo1 in cultured human endothelial cells resulted in a pro-inflammatory and endothelial-activated profile. In addition, overexpression of Glo1 attenuated loss of podocytes in the glomerulus and also renal excretion of early markers of diabetic nephropathy. MGO-induced podocyte apoptosis and generation of AGEs in the vasculature seem to contribute to this pathogenesis. These data demonstrate a direct link between increased levels of AGEs, endothelial dysfunction and the development of renal changes in experimental diabetes.

Our results showed that the diabetes-induced elevation of the AGEs CML and MG-H1 in vascular tissue is prevented by overexpression of Glo1. This decrease in vascular tissue levels of MGO and GO-derived AGEs demonstrates the efficiency of Glo1 overexpression in the detoxification of both GO and MGO. This is in accordance with our previous study in which we demonstrated that Glo1 overexpression prevents elevated blood levels of GO and MGO, and thereby decreases plasma AGE levels after 12 weeks of diabetes [13]. In contrast, the CML levels in the plasma could not be significantly decreased by Glo1 overexpression after 24 weeks of diabetes, suggesting that an excessive formation of CML via additional pathways, like lipid peroxidation, possibly exceeds the detoxification rate in this prolonged state of diabetes.

We previously addressed the reaction of the mesenteric endothelium to acetylcholine after short-time incubations with MGO or high glucose and in diabetic rats 12 weeks after STZ injection [14]. In the current study we confirmed these data in mesenteric arteries after 24 weeks of diabetes and extended our observations by specifically addressing which pathways are involved in the diabetes-induced impairment of acetylcholine-dependent relaxation. Our approach revealed that the decreased endothelium-dependent relaxation is caused by impairment of both the eNOS/NO pathway and endothelium-derived hyperpolarising factors. In contrast with several studies suggesting a decrease in expression and activity of the eNOS enzyme [22, 23], our results show that its mRNA and protein levels were not altered by diabetes. Unfortunately, we did not address the phosphorylation state of eNOS, therefore we could not attribute the impairment of eNOS activity to decreased phosphorylation, as Dhar et al described [24]. However, we previously showed that MGO and MGO-arginine adducts do not directly inhibit the overall eNOS activity in vitro [25]. Furthermore the plasma levels of the NO metabolites nitrite and nitrate were significantly decreased in the wild-type diabetic rats, indicating a decrease in NO bioavailability rather than reduced NO production. Interestingly, mRNA levels of the genes encoding oxidative stress-sensing master switch nuclear factor (erythroid-derived 2)-like 2 (NRF-2), and NRF-2-responsive detoxifying genes Glo1, NADPH-oxidase-4 and sodium oxide dismutase-3, were increased in the wild-type diabetic rats (data not shown), and this oxidative stress response was absent in the GLO-I diabetic rats, suggesting that diabetes-induced quenching of NO by oxidative stress occurred.

Elevated levels of cellular adhesion molecules are associated with an increase in leucocyte adhesion and inflammatory activation [26]. Our results showed that Glo1 overexpression prevented the STZ-induced increase of both Vcam1 and Icam1 expression. In line with this observation, our experiments in cultured endothelial cells showed that GLO-I is an important regulator in the prevention of endothelial inflammation and activation. These results confirmed the link between markers of endothelial function and AGEs [27, 28].

Diabetes is characterised by hypertrophy of several target tissues including the kidneys and the vasculature [29, 30]. Consistent with these findings, we detected an increase in kidney weight and glomerular size after 24 weeks of diabetes. Overexpression of Glo1 revealed a trend of decreasing the renal hypertrophy, although not statistically significant. We did not see any differences in morphological lesions of the glomeruli even after 24 weeks of diabetes, which is in accordance with earlier studies in STZ rats. Hirose et al [31] reported that basal membrane thickness is the only variable of advanced diabetic glomerulopathy that increases significantly in STZ rats, albeit only after more than 6 months of diabetes. Bidani et al [32] speculated that blood pressure reduction in STZ rats is one of the reasons that the nephrotic damage observed in the STZ diabetes model is only modest. Indeed, in our model we also observed a decrease in blood pressure in the diabetic rats and, interestingly, Glo1 overexpression significantly blunted the diabetes-induced blood pressure depression.

In humans, there is convincing evidence that endothelial dysfunction is closely associated with the development of diabetic nephropathy [33, 34]. In type 1 diabetes, microalbuminuria and macroalbuminuria are accompanied by a variety of markers of endothelial dysfunction, such as increased plasma concentrations of VCAM-1, endothelin-1 and plasminogen activator inhibitor-1, and an impaired endothelium-dependent, NO-mediated vasorelaxation [35]. Consistent with this, we observed an increase in albuminuria in diabetes, which showed a tendency to be improved by Glo1 overexpression. In addition to albumin (which was not statistically significant), two other markers for nephropathy, KIM-1 and osteopontin, were significantly improved by Glo1 overexpression. Although the function of KIM-1 is unclear, its abundant tubular expression after damage points to a role in either tubular damage or repair [36]. Also osteopontin, a large phosphoglycoprotein adhesion molecule, is upregulated in diabetic nephropathy. A recent study by Nicholas et al suggests that osteopontin is a key profibrotic factor that contributes to the development of glomerulosclerosis and albuminuria in diabetes [37].

Diabetic nephrotic injuries of both the glomeruli and the tubulointerstitium are accompanied by increased protein excretion and consequently by a decline in renal function [38]. It is likely that diabetic nephropathy occurs as a result of an interplay between haemodynamic and metabolic factors [39]. Thus, endothelial dysfunction, as an early phenomenon, may explain the typical association between albuminuria and extra-renal complications. It may be speculated that endothelial dysfunction, as we observed in the rat mesenteric arteries, is also present in the intra-renal vasculature [40, 41], thereby resulting in increased pressure in the glomeruli, which consequently leads to glomerular dysfunction and urinary protein leakage. Our observations that Glo1 overexpression reduced diabetes-induced vascular and renal dysfunction, as well as data from other studies [30, 42, 43], support a role for glycation in these pathogeneses. In addition, a cohort study by Beisswenger et al showed that progression of diabetic nephropathy is also significantly related to elevation in dicarbonyl stress [11]. In this study elevated levels of MGO correlated with loss of podocytes, another mechanism preceding glomerular leakage [44]. This strengthens our observations that MGO induces podocyte apoptosis and that diabetes-induced podocyte loss can be prevented by Glo1 overexpression.

In summary, this study showed for the first time that overexpression of Glo1 reduces diabetes-induced endothelial dysfunction and attenuates early renal impairment in an animal model of diabetes. These observations suggest that inhibition of intracellular glycation prevents deterioration of vascular and renal function in experimental diabetes. A recent study with the angiotensin type 1 receptor blocker candesartan showed that GLO-I activity can be positively regulated by drug intervention [45]. Therefore activation of GLO-I may provide an approach to prevent endothelial dysfunction and vascular complications in diabetes.