Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

An intriguing problem in internal medicine is the development of generalized angiopathy in the course of diabetes mellitus. This is associated with the occurrence of endothelial dysfunction (Oyama et al., 1986; Cameron and Cotter, 1992; Pieper and Peltier, 1995; Olbrich et al., 1996; Pieper et al., 1997). It is known that hyperglycemia can lead to changes in endothelial nitric-oxide production or release; short-term or subacute exposition to high D-glucose was shown to result in enhanced NO release (Graier et al., 1993, 1996), whereas chronic exposure during an entire cell culture passage leads to reduced NO-release (Olbrich et al., 1999) and diminished calcium signals (Salmeh and Dhein, 1998). It was hypothesized by these authors and by others (Pieper et al., 1997; Du et al., 2000; Nishikawa et al., 2000) that free radicals may be involved in the pathophysiology of endothelial dysfunction. Meanwhile, there is broad evidence supporting an important pathophysiological role for free radicals in diabetic vasculopathy (Spitaler and Graier, 2002). Thus, a central role for superoxide overproduction in the pathobiochemistry of the main pathways of hyperglycemia-related changes, i.e., activation of protein kinase C, accumulation of advanced glycation end products, and increased flux of glucose through the aldose reductase pathway, has been shown (Nishikawa et al., 2000). Recently, Rösen and colleagues (1998) showed that vitamin E can prevent reduction in endothelial NO release in diabetic rat heart. They postulated that during hyperglycemia the endothelium may be deprived with L-arginine or that an increased NO level may be inactivated by increased levels of free radicals (Rösen et al., 1998). In addition, Cinar and colleagues (2001) showed in a 3-month model of diabetes a protective effect of 1000 mg of vitamin E/kg of chow against endothelial dysfunction. Similarly, a protective effect was shown in a mouse model (Göcmen et al., 2000) or in a 2-month rat model (Keegan et al., 1995). In support of these studies, Kunisaki and colleagues (1995) showed in a rat model that diabetes led to enhanced protein kinase C translocation and increased diacylglycerol formation, which both could be prevented by a 2-week vitamin E treatment in retinal vascular endothelial cells. In another diabetes rat model, vitamin E attenuated, but did not completely prevent, diabetes-induced endothelial dysfunction (Karasu et al., 1997a). Two-month, dietary, vitamin E supplementation in the diabetic rat reduced lipid peroxide levels (Karasu et al., 1997b). It remained unclear, however, whether such positive effects of vitamin E might be attenuated in models with longer duration of the disease.

Although vitamin E treatment was shown to prevent from thromboxane overproduction in diabetic patients (Gisinger et al., 1988), which has been suggested to indicate a possible vasoprotective effect, and although improvement of endothelial function in diabetic patients by vitamin E has been found (Skyrme-Jones et al., 2000), others failed to demonstrate a positive preventive effect of additional vitamin E in diabetes mellitus (Nickander et al., 1994; Dagenais et al., 2001; Lonn, 2001; Lonn et al., 2001).

Since diabetes is a chronic disease and most animal studies investigated a considerably short duration of diabetes (from 1 to 3 months), we wanted to know whether vitamin E influences vascular function in a long-term diabetes rat model of a 7-month duration, which is approximately a quarter of the normal life span of the rat. Furthermore, we wanted to elucidate the effects of vitamin E in a subchronic cell culture model of hyperglycemia, previously established by our group (Salameh and Dhein, 1998), with a duration of an entire cell culture passage (and not only 24 h as often used). Thus, the aims of our study were to test whether, in an in vivo type I diabetes mellitus rat model, vitamin E deprivation might worsen the development of endothelial dysfunction and whether vitamin E supplementation (medium and high) might prevent it. Furthermore, it should be investigated whether in a subchronic cell culture model chronic hyperglycemia leads to reduced endothelial NO production and whether this can be prevented by vitamin E.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

In Vivo Study

All experiments were performed according to the ethical rules of the Council for International Organization of Medical Science and the German laws for animal welfare. We used a streptozotocin rat model of type I diabetes with a duration of diabetes of 7 months, as described (Dhein et al., 2000). Four experimental groups were investigated: 1) control animals without treatment (normal vitamin E alimentation, 75 mg/kg of chow) (n = 9), 2) diabetic animals without treatment (normal vitamin E alimentation, 75 mg/kg of chow) (n = 6), 3) diabetic animals receiving a vitamin E-enriched diet (1300 mg/kg of chow) (n = 8), and 4) diabetic animals receiving a vitamin E-deprived diet (0.55 mg/kg of chow) (n = 6). Vitamin E was supplied as -tocopherol. Vitamin E plasma levels were 10 ± 1 mg/l in animals receiving medium vitamin E alimentation, 19 ± 2 mg/l in animals receiving a vitamin E-enriched diet, and 2.2 ± 1 mg/l in rats receiving a vitamin E-deprived diet. For comparison normal rat diets are supplemented with vitamin E ranging from 30 to 200 mg/kg of chow (so that 75 mg/kg of chow resembles a normal rat diet) (Lehr et al., 1999).

For induction of diabetes mellitus, six-week-old male Wistar Kyoto rats (140 ± 20 g) were rendered diabetic with an i.p. injection of streptozotocin (60 mg/kg b.wt.), as described (Dhein et al., 2000). Two weeks after the induction of diabetes mellitus, animals were randomized to the treatments, i.e., no treatment, a vitamin E-enriched diet, or vitamin E-deprived diet. The animals did not receive an antidiabetic treatment.

Vascular Function. For functional measurements of smooth muscle and endothelial function, a mesenteric loop was isolated with the appertaining intestine (8 cm in length) according to the technique described earlier (Dhein et al., 1992, 2000; Olbrich et al., 1996). The mesenteric artery was cannulated and perfused with oxygenated Tyrode's solution (161.02 mM Na⁺, 5.36 mM K⁺, 1.8 mM Ca²⁺, 1.05 mM Mg²⁺, 146.86 mM Cl, 23.80 mM HCO₃, 0.42 mM H₂PO₄, and 10.00 mM glucose, pH adjusted to 7.4; gassed with 95% O₂ and 5% CO₂). An 8-cm loop of the small intestine was ligated, and all side branches of the mesenteric vessels were sealed by ligation so that an isolated mesenteric fold with the appertaining intestine, and the perfusing arterial network was prepared. This preparation was fixed to a perfusion system with a constant perfusion pressure of 70 cm of H₂O, which corresponds to the actual physiological perfusion pressure in the mesenteric artery in this model. Ten cannulas were inserted into the intestine to provide drainage. With the help of a microscope (Carl Zeiss GmbH, Jena, Germany) and a video camera (Sony, Tokyo, Japan), which was mounted behind the ocular of the microscope, the mesenteric vessels were displayed on a monitor (Sony). The total magnification was 240-fold. In the course of the experiments, pictures of the arteries were recorded. Vessel diameters were determined during the experiment directly on the screen and, after the experiments, re-evaluated in the digitalized pictures using a frame grabber board (Data Translation, Inc., Marlboro, MA) with JAVA software (Jandel Scientific, Erkrath, Germany). The vessel diameter was assessed by analyzing the first derivative of the gray level along a cross sectional line (orthogonal to the vessels longitudinal axis). The distance between the extremata corresponds to the vascular diameter. We classified microvessels according to the generation theory of Ley and colleagues (1986) as G1 vessels, which are the branch perfusing the isolated loop. The subsequent branches were classified as G2, G3, and G4 vessels, the latter being those vessels at the border between the mesenterium and gut. More details of the method are given by Olbrich et al. (1999).

Cell Culture Study

Cell Isolation and Culture. In previous investigations, we established a subchronic cell culture model of hyperglycemia-induced endothelial dysfunction (Salameh and Dhein, 1998) using porcine aortic endothelial cells exposed to hyperglycemia for an entire culture cell passage (4 days). Therefore, porcine aortic endothelial cells were isolated and cultured according to Rosenthal and Gotlieb (1990), as previously described (Salameh et al., 1997). Briefly, the endothelial cells were harvested from porcine thoracic aorta using 1 mg/ml dispase, seeded (100,000 cell/cm²) in plastic 9.6-cm² Petri dishes (Nalge Nunc International, Wiesbaden, Germany), and cultured with M199 at 37°C, saturated humidity and 5% CO₂. After reaching confluence, the cells were passaged and seeded again. Purity of the cell culture was tested by uptake of 1,1'dioctadecy-l3,3,3'33'-tetramethylindo-carbocyanine-acetylated low-density lipoprotein (Dil-Ac-LDL) (Voyta et al., 1984) and, for detecting contaminating smooth muscle cells, by staining of -smooth muscle actin. At the start of the third passage, the cells were submitted to the various treatments. The different experimental protocols were carried out with cells of the same cell line for intraindividual control (i.e., all cells were derived from the same aorta) at the moment when they were seeded for third passage.

Histological Studies. For H&E staining, endothelial cell monolayers were washed three times with phosphate-buffered saline (PBS) (containing 137 mM NaCl, 2.68 mM KCl, 8.1 mM Na₂HPO₄, 1.47 mM KH₂PO₄, buffered at pH 7.4) and fixed in paraformaldehyde solution (4% paraformaldehyde in PBS) for 30 min at room temperature. After we removed the fixative, the cells were stained in 1% hematoxylin and 0.5% eosin following classical protocols and embedded in Karion F (Merck, Darmstadt, Germany).

NADPH-Diaphorase Staining. The NADPH-diaphorase reaction was carried out according to Hope et al. (1991). In brief, confluent monolayers were fixed in 4% paraformaldehyde for 30 min at room temperature and incubated in the staining solution, containing 0.5 mM nitro blue tetrazolium, 1 mM -NADPH, 0.2% Triton X-100, 50 mM Tris, and 75 mM NaCl, buffered at pH 8.0, for 20 h at 37°C. Thereafter, the preparations were washed three times in PBS and embedded in Karion F.

Measurement of Nitric Oxide Release. To characterize endothelial function, we measured the NO release spectrophotometrically (UV-DU-7500; Beckmann Coulter, Inc., Munich, Germany) under basal conditions and after stimulation with ATP (1 mM) using the methemoglobin assay (Feelisch and Noack, 1987), based on the rapid oxidation of reduced methemoglobin (oxy-Hb, oxyhemoglobin, Fe²⁺) to methemoglobin (Met-Hb; Fe³⁺) by nitric oxide. The suitability and specificity of this assay has been demonstrated previously (Kelm et al., 1997). We monitored increasing amounts of methemoglobin versus oxyhemoglobin by means of the difference spectrum (Feelisch and Noack, 1987). The bioassay was calibrated as described previously (Feelisch and Noack, 1987; Kelm et al., 1988). We found an extinction coefficient e of 39 mM¹ cm¹, which is nearly identical to that described by Feelisch and Noack (1987). After reaching confluence, the porcine aortic endothelial cell were washed three times with HEPES buffer (composed of 145.0 mM NaCl, 5.0 mM KCl, 2.5 mM CaCl₂, 1.0 mM MgCl₂, 10.0 mM HEPES, and 5.0 mM D-glucose), at pH 7.4 and 37°C, preincubated with 4 ml of HEPES buffer for 20 min at 37°C, and supplemented with oxy-Hb-solution (4 µM). After an equilibration period of 50 min, 1 mM ATP was added, and subsequently, NO release was recorded for 40 min; the cycling time was 10 min, at 37°C, for each cell culture condition intraindividually. To obtain the actual formation of methemoglobin, representing the NO release by PAEC, we subtracted the spontaneously occurring formation of methemoglobin, determined from a cell-free incubation solution from the measurements.

Chemicals. All chemicals were obtained from Sigma-Aldrich (Deisenhofen, Germany) except for Dil-Ac-LDL, which was obtained from Paesel & Lorei (Frankfurt, Germany); dispase was obtained from Roche Molecular Biochemicals (Mannheim, Germany); nitro blue tetrazolium was purchased from Biomol (Hamburg, Germany). KCl were obtained from Merck, and ACh and heparin were supplied by Serva (Heidelberg, Germany). All cell culture media and fetal calf serum were obtained from Sigma-Aldrich (St. Louis, MO); the cell culture material was obtained from Nalge Nunc International. All chemicals were of analytic grade and were dissolved in bidistilled water if not stated otherwise.

Statistical Analysis

For statistical analysis, a two-factorial analysis of variance was performed. If analysis of variance indicated significant differences or significant interactions between disease and treatment, the data were further analyzed with a post hoc Tukey-high standard deviation test. For the statistical analysis, we used Systat for Windows software, version 5.02 (Systat, Evanston, IL). Differences were considered significant if p < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

In Vivo Study

Streptozotocin injection caused diabetes mellitus within 2 weeks and blood glucose levels of >18 mM in all diabetic groups (no differences between the three diabetic groups). In nondiabetic age-matched control animals, we found blood glucose levels ranging from 3.5 to 5.6 mM. Body weight was reduced in diabetic animals [228 ± 10 (d.m.), 237 ± 3 (d.m. + vitamin E), 199 ± 7 g (d.m.  vitamin E)] compared with nondiabetic age-matched controls (418 ± 8 g). Plasma vitamin E levels were 10 ± 1 mg/l (animals receiving 75 mg/kg of chow), 19 ± 2 mg/l (animals receiving a vitamin E-enriched diet), and 2.2 ± 1 mg/l (rats on a vitamin E-deprived diet) (see Table 1).

With regard to the vascular parameters, the initial diameters of the mesenterial microvessels under resting conditions were somewhat enlarged in diabetic animals (see Table 2). This was not influenced by the treatment.

View this table: [in this window] [in a new window]   TABLE 2 Initial diameter of mesenteric microvessels (generation G1 to G4) of nondaibetic control rats, diabetic rats, and diabetic rats (d.m.) receiving a high vitamin E diet (d.m. + Vit E), or receiving a vitamin E-deprived diet (d.m.  Vit E) under resting conditions given as the means ± S.E.M.

Regarding smooth muscle vascular function, we found a reduction in vessel diameter by 70 mM KCl between 44 and 29% in nondiabetics. In G1, G2, and G3 vessels, this was not significantly altered in diabetics. In G4 vessels, KCl-induced constriction was slightly attenuated in diabetics. Additional vitamin E treatment or vitamin E deficiency did not influence KCl contraction (see Fig. 1A).

View larger version (44K): [in this window] [in a new window]   Fig. 1.   Vascular reactions in nondiabetic and chronic (6 months) streptozotocin-induced diabetic (d.m.) rat mesenterial microvasculature with or without additional vitamin E treatment or with vitamin E deficiency. The figure shows the reactions of four branches of the mesenteric microcirculation (G1 to G4, initial diameters see Table 1). All values are given as the means ± S.E.M. of n = x experiments (for n see Table 1). Statistical differences to nondiabetics is marked by an asterisk; statistical difference versus diabetes mellitus is indicated by a # (p < 0.05). A, shows the effect of 70 mM KCl on vascular tone in G1 to G4 mesenteric microvessels in a percentage of initial diameter. B, gives the relaxation of the mesenteric microvessels in response to 1 µM GTN as a percentage of KCl contraction. C, the effect of 1 µM ACh is depicted as a percentage of KCl contraction. Note that in vitamin E-deficient diabetics, ACh no longer induced relaxation but resulted in a slight further contraction. D, shows the effect of 3 µM L-NG-nitro-arginine on vascular tone. The contractions evoked by 3 µM L-NG-nitro-arginine are normalized to the contractions elicited by KCl, and data are given as a percentage of KCl contraction.

GTN application in all vessels lead to significant vasodilation, which was diminished in diabetics. Additional treatment with vitamin E did not influence GTN-induced vasorelaxation. In vitamin E-deprived rats, however, we found significantly decreased GTN-induced relaxation compared with untreated diabetics (see Fig. 1B).

Acetylcholine induced vasorelaxation in all vessels in nondiabetic rats. In diabetic rats, this ACh-induced relaxation was significantly decreased, which was not influenced by vitamin E treatment. With vitamin E deficiency, however, we found complete abolition of ACh-induced relaxation. In contrast, ACh in these rats induced slight vasoconstriction (Fig. 1C).

Finally, we applied L-N^G-nitro-arginine (LNNA), which resulted in a significant vasoconstriction of all vessels. The LNNA-induced vasoconstriction reached 8 to 14% of the KCl-constriction. In diabetic rats, this LNNA-induced constriction was significantly reduced in G3 and G4 vessels. This was not influenced by vitamin E treatment. In rats with vitamin E deficiency, however, we found significant attenuation of LNNA-induced vasoconstriction in all vessels (Fig. 1D).

Cell Culture Study

NO Release. Cells reached confluence after 3.5 ± 0.5 days without differences between the groups. We found the typical difference spectrogram for Met-Hb versus oxy-Hb with an isobest at 412 ± 1 nm and maximum extinction at 402 ± 1 nm, as described by Feelisch and Noack (1987). Under basal condition using normal cells, there was a slow increase in extinction, as can be seen in Fig. 2A during the first 50 min, indicating increasing formation of Met-Hb and release of NO. Stimulation with 1 mM ATP led to a further increase in extinction. In cells that were grown under hyperglycemic conditions, however, basal formation of Met-Hb was significantly reduced (Figs. 2B and 3). ATP-stimulated formation of Met-Hb was also clearly diminished (Fig. 3). Quantitatively we found basal release of 80 ± 20 pMol · 1 Mio cell¹ · 10 min¹, which was significantly diminished to 40 ± 10 pMol · 1 Mio cell¹ · 10 min¹ (p < 0.05) (Fig. 3). ATP-stimulated NO-release was significantly reduced from 130 ± 20 (normal cells) to 92 ± 10 pMol · 1 Mio cell¹ · 10 min¹ in hyperglycemic cells (p < 0.05).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Difference spectrum of oxyhemoglobin versus methemoglobin obtained from cultured PAEC under normal (A) and hyperglycemic (B) conditions over a period of 90 min displaying the characteristic absorption maximum at 401 nm, with an isobestic point at 411 nm, a negative absorption maximum at 421 nm, and straight lines between 401 and 411 nm (±1.25 nm tolerance regarding the precision of the instrument). The reaction was carried out in HEPES-buffered solution in the presence of 4.0 µM/l oxyhemoglobin at pH 7.4 and 37°C, as described (see Materials and Methods). Note the increase in Met-Hb formation evident from increasing absorption with time and its attenuation in hyperglycemic cells (B).

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Formation of nitric oxide in normo- and hyperglycemic endothelial cells with or without additional vitamin E treatment under basal conditions (left) and after stimulation with 1 mM ATP (right). All values are given as the means ± S.E.M. of n = 6 experiments. Statistical differences to nondiabetics is marked by an asterisk; statistical difference versus diabetes mellitus is indicated by a # (p < 0.05). Note that vitamin E prevents from hyperglycemia-induced reduction in basal but not stimulated NO formation.

1

1

1

1

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Formation of nitric oxide in normo- and hyperglycemic (+15 mM D-glucose) endothelial cells and in cells treated with 15 mM L-glucose as an osmotic control under basal conditions (left) and after stimulation with 1 mM ATP (right). All values are given as the means ± S.E.M. of n = 6 experiments. Statistical differences to nondiabetics is marked by an asterisk,; statistical difference versus diabetes mellitus is indicated by a # (p < 0.05). Note that 15 mM L-glucose (in contrast to 15 mM D-glucose) does not reduce NO formation.

Histology. For further characterization of the glucose effects, we investigated the cultured cells histologically. In all cell lines investigated, >99% of the tested cells incorporated Dil-Ac-LDL, with no significant differences between the two experimental groups. The content of contaminating smooth muscle cells was <0.1%. The high D-glucose-treated cell lines did not show any promotion of smooth-muscle cell growth in culture. The three control groups (5 mM D-glucose without any treatment or with 20 mg/l vitamin E or with 15 mM L-glucose) exhibited no significant differences concerning their cell morphology. Moreover, the number of giant cells and the NADPH-diaphorase activity per cell were not different between the control groups.

View larger version (88K): [in this window] [in a new window]   Fig. 5.   Original photographs showing NADPH-diaphorase activity as assessed by histochemical nitro blue tetrazolium staining in normoglycemic (upper panel) and hyperglycemic (lower panel) endothelial cells (400× magnification). Formation of the blue formazan stain indicates NADPH-diaphorase activity. Note the typical positive dark staining in the perinuclear zone, which is diminished in the hyperglycaemic cells.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In Vivo Study. Taken together our data show that smooth muscle contractile function was not altered by diabetes mellitus, as evident from the nearly unchanged KCl-induced contractions. Relaxation to GTN and ACh were significantly attenuated, however. This diabetes-induced deficit in relaxation was not influenced by high vitamin E treatment but was further enhanced by vitamin E deficiency. It is well known that GTN-induced vasorelaxation depends on glutathione-dependent release of NO from S-nitrosothiol-derivatives of GTN. Since diabetes deprivation of glutathion occurs in vascular tissue (Kinalski et al., 2000), it is reasonable that in our study relaxation to GTN was attenuated in diabetic rats. In contrast, ACh-induced relaxation depends on release of NO from functional vascular endothelium. As in previous studies (Olbrich et al., 1996, 1999; Dhein et al., 2000), ACh-induced vasorelaxation was diminished in diabetics, indicating endothelial dysfunction. Interestingly, although this was not affected by high vitamin E treatment, vitamin E deficiency further diminished vasorelaxant response to GTN and completely abolished relaxation to ACh, with a reversal of the ACh-response so that vasoconstriction occurred. It should be mentioned that vitamin E deprivation for 4 to 12 months itself impairs endothelial relaxation, whereas smooth muscle responses are not affected (Rubino and Burnstock, 1994; Rubino et al., 1995; Ralevic et al., 1995). These investigators, however, observed only reduced vasorelaxation in vitamin E deprivation. Thus, both diabetes mellitus and vitamin E deprivation can lead to an impairment of endothelial function. The combination of diabetes and vitamin E deprivation, however, seems to be even more deleterious, leading to complete loss of endothelial function so that ACh elicits vasoconstriction. This is, to our best knowledge, the first article showing this dramatic change in the response to acetylcholine.

In Vitro Study. To investigate the results of the in vivo study in more detail, we exposed endothelial cells subchronically to hyperglycemia for an entire culture passage, which resulted in reduced NO release. This could partially be prevented by vitamin E; only the hyperglycemia-induced reduction in basal NO release was prevented by vitamin E, whereas reduction in stimulated NO release was not prevented.

What Are the Reasons for the Discrepant Findings with Vitamin E? There is some contradiction between the in vivo study and the in vitro results in our study that should be discussed. Although in the in vitro study vitamin E prevented hyperglycemia-induced impairment of NO-release, we did not observe influence of vitamin E on LNNA-response. On the one hand, there is a radical scavenging effect that might be present in both the in vivo and the in vitro situation. On the other hand, the vasoprotective effect of -tocopherol in vivo also comprises binding of the vitamin to the vitamin E binding protein, transportation with the lipoproteins, and preservation of unsaturated bonds in essential fatty acids such as -linolenic acid and eicosapentaenoic acid. Thus, in long-term in vivo situation, this more complex mechanism of action of vitamin E involving essential fatty acids might be affected in diabetes. Accordingly, a reduction in the eicosapentaenoic/arachidonic acid ratio was found in diabetic rats (Ikeda and Sugano, 1993). The metabolism in long-term diabetes probably is more complex than simple hyperglycemia, as may be indicated by the loss of body weight in these rats. Moreover, this difference between results obtained from cultured cells and in vivo results may indicate that diabetes mellitus means more than simply hyperglycemia. In addition, results from cultured porcine aortic endothelial cells cannot completely simulate the situation in the long-term in vivo mesentery artery of the diabetic rat. Another aspect is the duration of the disease or hyperglycemia. In a rat model, 7 months probably is a long duration. Thus, positive effects seen with vitamin E in a model with only a 2-week or 2- to 3-month duration (Keegan et al., 1995; Kunisaki et al., 1995; Karasu et al., 1997a,b) may be attenuated by the long duration of the disease.