Effects of Rosuvastatin on Nitric Oxide–Dependent Function in Aorta and Corpus Cavernosum of Diabetic Mice: Relationship to Cholesterol Biosynthesis Pathway Inhibition and Lipid Lowering

Male C57BL/6J mice from the University of Aberdeen breeding colony were aged 4–6 months on the day of experimentation. They had ad libitum access to standard laboratory diet and water. Experiments were performed in accordance with regulations specified by the U.K. Animal Procedures Act (1986) and the National Institutes of Health Principles of Laboratory Animal Care (1985 revised version).

Unless otherwise stated, all chemicals were obtained from Sigma (Poole, Dorset, U.K.). Diabetes was induced by intraperitoneal injection of streptozotocin (STZ) (AstraZeneca, Macclesfield, Cheshire, U.K.) dissolved in sterile saline at a dose of 125 mg/kg. After 48 h, diabetes was verified by the presence of hyperglycemia and glycosuria (Visidex II and Diastix; Ames, Slough, U.K.) in nonfasted mice. Diabetic mice were weighed daily and were rejected if body weight increased over 3 consecutive days or if blood glucose was <20 mmol/l.

Diabetes duration was 4 or 6 weeks, and both prevention and intervention studies were undertaken. The prevention study comprised a group of diabetic control mice untreated for 4 weeks; a second diabetic group received RSV (Crestor; AstraZeneca) treatment as an oral gavage (20 mg · kg⁻¹ · day⁻¹ in 0.5% Tween solution) after 48 h, following confirmation of STZ-induced diabetes. This dose was chosen based on the results of a dose-response study where it gave ∼95% correction of a sciatic nerve motor conduction velocity deficit in rats and on efficacy for inhibition of hepatic cholesterol biosynthesis in rodents (29,30). A further 4-week diabetic group received RSV (20 mg · kg⁻¹ · day⁻¹) and cotreatment with mevalonate (150 mg · kg⁻¹ · day⁻¹ in the drinking water). In addition, a group of nondiabetic mice received RSV treatment (20 mg · kg⁻¹ · day⁻¹) for 4 weeks. The intervention study used 6-week untreated diabetic controls and a second diabetic group that was given RSV (20 mg · kg⁻¹ · day⁻¹) for 2 weeks following 4 weeks of untreated diabetes.

On the day of experiments mice were anesthetized (5% halothane in air, with 0.1 ml 10% urethane in saline per 10 g body wt i.p.). The thoracic aorta was removed between the aortic arch and the diaphragm. The penis was excised at its base with removal of the glans penis and connective and adventitial tissues along the shaft. Blood was taken from the heart, and plasma was frozen for subsequent plasma glucose, total cholesterol, and total triglyceride determination by means of colorimetric spectophotometry using standard test kits (Sigma, Poole, Dorset, U.K.).

The thoracic aorta was cleared of connective tissue and two sections cut into 1-mm lengths were mounted as ring preparations, using 40-μm tungsten wire, into 5-ml organ baths in a small-vessel myograph for measurement of isometric tension. Aortas were bathed in carboxygenated (95% O₂:5% CO₂) modified Krebs-Ringer solution (144 NaCl, 5 KCl, 1.25 Ca²⁺, 1.1 MgSO₄, 25 NaHCO₃, 1.1 NaH₂PO₄ and 5.5 glucose: all in mmol/l) at 37°C (pH 7.35). Resting tension was maintained at 0.75 g. Tissues were left to equilibrate for 1 h, with frequent changing of bathing solution, before tissue viability was assessed for contractile responses and endothelial integrity, with 300 nmol/l phenylephrine followed by 100 nmol/l ACh. After a 45- to 60-min washout and recovery period, cumulative concentration response curves to phenylephrine were determined. This was followed by precontraction with phenylephrine at ∼80% of the maximal response to give a background tension against which to assess concentration response relationships for ACh and then sodium nitroprusside (SNP). In some experiments, concentration response curves were repeated after a 30-min incubation period with 10 μmol/l of the cyclooxygenase inhibitor flurbiprofen or the NO synthase inhibitor NG-nitro-l-arginine (l-NNA).

Individual corpus cavernosum were separated and mounted in 10-ml organ baths and bathed in Krebs-Ringer solution as for aorta. Tension was monitored by isometric transducers, and resting tension was set at 0.5 g. Tissues were left to equilibrate for 1 h before determination of cumulative concentration response curves for phenylephrine and for SNP against an ∼80% maximum phenylephrine precontraction. Tissues were washed and allowed to recover for 20–30 min between response curves. Noncumulative concentration response curves for ACh were established against individual 80% maximal phenylephrine precontractions.

Repetitive supramaximum transmural electrical field stimulation (EFS) of autonomic nerves was delivered via platinum wire electrodes placed on either side of the cavernosum (train duration 30 s, frequency 2–30 Hz, and pulse duration 5 ms, 90 mA). Frequency response curves were determined for noradrenergic vasoconstrictor nerves before incubation with atropine (1 μmol/l) and guanethidine (4 μmol/l) for 30 min to eliminate responses mediated by cholinergic and noradrenergic nerves, respectively. Following phenylephrine precontraction, EFS produced frequency-dependent relaxations mediated by NANC nerves. Some concentration or frequency response curves were repeated following a 30-min incubation period with 10 μmol/l flurbiprofen or l-NNA. At the end of the experiments tissues were lightly patted dry and weighed.

Data are expressed as means ± SE. They were subjected to Bartlett’s test for homogeneity of variances before ANOVA. Where significance was reached (P < 0.05), between-group differences were established using the Newman-Keuls multiple comparison test. Otherwise, data were analyzed by Kruskal-Wallis nonparametric ANOVA and Dunn’s multiple comparison test. Concentration response data were fitted by sigmoid curves using the least squares method to calculate EC₅₀. Whole-curve analyses comparisons were made using two-way ANOVA. For comparisons between two groups of data only, paired or unpaired two-tailed Student’s t tests were used. All calculations were made using a standard software package (Prism3; Graphpad, San Diego, CA).

Diabetic groups developed a more than threefold increase in plasma glucose concentrations (P < 0.001) and had 25–35% (P < 0.001) weight loss (Table 1); neither parameter was significantly altered by RSV or RSV with mevalonate cotreatment. Total plasma cholesterol was 5.0 ± 0.3 mmol/l (n = 18) in the control group. This was unaffected by 4-week diabetes (4.7 ± 0.4 mmol/l, n = 16) or RSV treatment (nondiabetic 4.5 ± 0.4 mmol/l, n = 14; diabetic 4.6 ± 0.5 mmol/l, n = 14). Triglyceride levels were unaffected by diabetes (control 1.13 ± 0.19 mmol/l, n = 13; 4-week diabetes 1.12 ± 0.24 mmol/l, n = 12), but were ∼40% reduced by RSV treatment (nondiabetic 0.64 ± 0.07 mmol/l, n = 12; 4-week diabetes 0.61 ± 0.21 mmol/l, n = 7), although this only reached statistical significance (P < 0.05) in the treated nondiabetic group.

Maximum EDR to ACh (Fig. 1), following phenylephrine precontraction, was similarly and significantly reduced by 4 and 6 weeks of diabetes compared with nondiabetic controls (56.9 ± 3.4, n = 15, and 58.1 ± 2.9, n = 10, vs. 76.2 ± 3.4%, n = 16; P < 0.001). RSV treatment completely prevented the diabetic deficit (76.5 ± 3.9%, n = 13, P < 0.01) (Fig. 1A). Mevalonate cotreatment at ∼79% attenuated (61.1 ± 5.1%, n = 11, P < 0.05) the effect of RSV alone. Maximum relaxation for RSV-treated nondiabetic rats (75.9 ± 3.8%, n = 11) did not differ from that of untreated controls. In the intervention group, RSV gave ∼57% reversal (68.4 ± 3.8%, n = 11) of the diabetic deficit to the extent that the relaxation did not differ significantly from the nondiabetic control value. Incubation with 10 μmol/l flurbiprofen did not alter ACh-mediated responses of control (n = 11) or 4-week diabetic (n = 9) mice (data not shown). In contrast, addition of l-NNA completely abolished ACh-mediated relaxations of control and 4-week diabetic aortas (data not shown). Diabetes tended to slightly reduce sensitivity to ACh, assessed by (−log)EC₅₀, although this did not reach statistical significance. Thus, all group (−log)EC₅₀ values were in the range of 7.32 ± 0.10 for the RSV-treated control group and 6.87 ± 0.13 for 4-week diabetic group.

Maximum endothelium-independent relaxation (∼95–99% for all groups) and sensitivity ([−log]EC₅₀, ∼7.9–8.0 for all groups) to the NO donor SNP following phenylephrine precontraction were unaltered by diabetes or by treatment (Table 2). However, in the presence of 10 μmol/l l-NNA, while there was no effect on maximum relaxation, there were small but significant increases in sensitivity to SNP in nondiabetic controls (P < 0.01), RSV-treated nondiabetic controls (P < 0.01), and prevention (P < 0.05) and intervention (P < 0.05) in RSV-treated diabetic groups. However, this effect was absent in aorta from 4- or 6-week diabetic mice. Cotreatment with mevalonate did not alter the l-NNA–mediated response (P < 0.01) for prevention in RSV-treated diabetic mice.

Maximum contraction to phenylephrine increased by >1.5-fold (P < 0.01) by 4 and 6 weeks of diabetes compared with nondiabetic controls (Table 2). Neither preventive RSV without or with mevalonate cotreatment nor intervention RSV treatment significantly altered the effects of diabetes on maximum contraction. RSV treatment of nondiabetic mice was also without effect. Similarly, sensitivity to phenylephrine, as measured by (−log) EC₅₀ values, increased by ∼0.7 log units after 4 and 6 weeks of diabetes (P < 0.001), and this was not significantly altered by treatment.

After preincubation with 10 μmol/l l-NNA, maximum contractions to phenylephrine of aorta (Table 2) from nondiabetic mice were ∼1.3-fold (P < 0.01) increased. In contrast, there was no significant effect of l-NNA on 4- and 6-week diabetic responses. RSV prevention (P < 0.01) and intervention (P < 0.05) treatments partially restored the augmented contractions in the presence of l-NNA. Cotreatment with mevalonate did not alter the augmented phenylephrine response (P < 0.05) of prevention RSV-treated diabetic mice. Similar to nontreated controls, RSV-treated nondiabetic aortas developed a ∼1.2-fold (P < 0.001) increase in maximum contraction with l-NNA. Both nondiabetic controls (P < 0.05) and 4-week diabetic (P < 0.01) but not 6-week diabetic aortas developed an increased sensitivity to phenylephrine in the presence of l-NNA. Similar trends were observed for the other groups, though statistical significance was not attained.

Tissue weights did not differ significantly between groups (7.0–7.5 mg, data not shown). Transmural EFS of corpus cavernosum caused frequency-dependent contractions that were abolished by incubation with 4 μmol/l guanethidine. Expressed relative to tissue weight, maximum contractions at 30 Hz were not significantly different between groups: tensions in mN/mg were in the range 0.059 ± 0.006 (prevention RSV-treated diabetic group, n = 13) to 0.078 ± 0.005 (RSV-treated nondiabetic group, n = 12).

EFS following phenylephrine precontraction in the presence of 1 μmol/l atropine and 4 μmol/l guanethidine produced frequency-dependent NANC relaxations (Fig. 2) that could be completely abolished by l-NNA (data not shown). Relaxations for diabetes of 4 and 6 weeks duration did not differ significantly and resulted in 25 and 33% reductions in maximum (20 Hz) relaxation compared with the nondiabetic control group (46.4 ± 3.4%, P < 0.01, and 40.8 ± 3.9%, P < 0.001, vs. 61.1 ± 3.3%), RSV treatment prevented 75% (57.3 ± 3.3%, P < 0.05) and reversed 71% (55.2 ± 3.6%, P < 0.05) of these diabetic deficits. Maximum NANC relaxation of RSV-treated nondiabetic mice (63.6 ± 1.7%) did not differ from that of the control group. Mevalonate cotreatment of RSV-treated diabetic mice gave a maximum relaxation value (45.6 ± 3.7%) that was not significantly different from that of the 4-week diabetic group but was reduced compared with RSV treatment alone (P < 0.05).

After phenylephrine precontraction, EDR in response to ACh (Fig. 3) after 4 and 6 weeks of diabetes revealed similar significant whole-curve deficits (P < 0.001) compared with controls. While RSV did not alter responses from nondiabetic tissues, treatment largely prevented (P < 0.001) the diabetic deficit, although relaxation across the concentration range remained slightly depressed compared with the nondiabetic control curve (P < 0.05). Cotreatment with mevalonate blocked the action of RSV, such that responses did not differ from those of diabetic controls. In addition, RSV partially reversed the diabetic curve deficit (P < 0.05), although responses remained lower than those of nondiabetic controls (P < 0.01).

Maximum endothelium-independent relaxation and sensitivity to SNP following phenylephrine precontraction of corpus cavernosum were unaltered by diabetes, RSV treatment, and RSV and mevalonate cotreatment. Maximum relaxations did not differ significantly between groups and were in the range 47.0 ± 4.0% (intervention RSV) to 55.1 ± 3.8% (nondiabetic control). Similarly, (−log) EC₅₀ values ranged from 6.08 ± 0.15 (intervention RSV) to 6.24 ± 0.09 (RSV-treated nondiabetic). In addition, l-NNA preincubation failed to alter SNP-mediated relaxations in any group (data not shown).

The maximum corpus cavernosum contractile responses and (−log) EC₅₀ values for phenylephrine were not altered by diabetes or treatment. Thus, maximum tensions for all groups ranged from 0.122 ± 0.007 (RSV-treated nondiabetic group) to 0.096 ± 0.007 mN/mg (preventive RSV-treated diabetes). Similarly, (−log) EC₅₀ values ranged from 6.07 ± 0.05 (nondiabetic control) to 6.27 ± 0.06 (preventive RSV-treated diabetes). l-NNA preincubation did not significantly alter corpus cavernosum maximum tension or (−log) EC₅₀ to phenylephrine.

Diabetes caused defective NO-mediated EDR to Ach stimulation in mouse aorta. This is the first report for STZ-induced diabetic mice and is in agreement with data from type 2 diabetic C57/BL/KsJ-db/db mice (8) and diabetic rats (2,6,7). Corpus cavernosum EDR was also impaired, in agreement with one report for diabetic mice (16) and several for rat, rabbit, and humans (15,17,18,31). EDR of mouse aorta and corpus cavernosum was unaffected by cyclo-oxygenase inhibition but was completely abolished by NO synthase blockade, as noted for rat aorta and corpus cavernosum (7,17). This suggests that in mouse aorta and corpus cavernosum, EDR is mediated by NO rather than prostanoids. Endothelium-independent relaxation to the NO donor SNP was unaffected by diabetes in both tissues, indicating that the smooth muscle cGMP mechanism was intact, in agreement with findings from diabetic nonmurine species (6–8,15–18). In diabetic patients, some studies have reported impaired endothelium-independent responses to NO donors (11,32,33), while other have not (12,18). The reason for this discrepancy is unclear but may reflect the severity and progression of vascular damage.

RSV treatment prevented and largely reversed the corpus cavernosum and aorta EDR deficits. This is the first report of statin effects on these parameters in experimental diabetes. Previous work showed that statins alter expression of several genes controlling vascular tone, inflammation, and coagulation in cultured human umbilical vein endothelial cells under normoglycemic conditions. Thus, mRNA for NO synthase-3 was increased, while endothelin-1 mRNA levels were reduced. In addition, mRNAs for interleukin-8, monocyte chemoattraction protein-1, and plasminogen activator inhibitor-1 were elevated, whereas thrombomodulin mRNA was reduced (34). Furthermore, statins protected cultured bovine aortic endothelial cells against adverse changes in NO synthase and endothelin-1 expression when challenged with oxidized LDL (35). In humans, statins reduce levels of circulating C-reactive peptide, an effect independent of lipid lowering (36), and decrease levels of soluble cell adhesion molecules (37). There are also complex statin effects on adhesion molecule expression by cultured endothelial cells (38). Thus, the prediction is that statins may protect vascular endothelium function in diabetes. However, a recent report failed to find any effect of atorvastatin on reduced forearm blood flow responses to endothelium-dependent and endothelium-independent agonists in type 2 diabetic patients (32). The reason for the discrepancy with the mouse data is unclear, but it may be that any statin effect on endothelial NO production in those patients was obscured by the blunted responses to NO donors also noted in that study, which does not occur in mice.

In patients with type 2 diabetes, plasma VLDL levels are typically higher, those of HDL are lower, and those of total and LDL cholesterol are about the same as in the nondiabetic population (20). However, LDL particles are smaller, denser, and more glycated and oxidized in diabetes, making them a risk factor for vascular disease. Together, LDL and VLDL triglycerides impair endothelium-dependent vasodilation by altering NO production (9,10,39,40). Under basal conditions, endothelial NO synthase-3 is targeted to cholesterol-rich membrane domains characterized by the presence of caveolin, which competes with calmodulin for NO synthase binding, perhaps explaining the inverse relationship between circulating cholesterol levels and NO activity (39,41).

In hypercholesterolemic patients, forearm blood flow responses to ACh infusion and NOS inhibition were reduced, suggesting decreased NO production (40). In addition, impaired EDR was found in vessels from hyperlipidemic-diabetic hamsters and rabbits (9,10) and hypertriglyceridemic rats (42). Although there may be indications of reduced NO production, enhanced inactivation of NO by reactive oxygen species is probably the most important mechanism by which NO-dependent endothelial function is blunted by diabetes (2). Indeed, antioxidants improve aorta and corpus cavernosum function of diabetic animals (2,4,7,9,16,17).

Total plasma cholesterol and triglyceride levels were unaltered by diabetes in mice, in agreement with results for several mouse strains (43). RSV had no effect on cholesterol levels, which are considerably lower in mice than humans, although there was a tendency toward lower triglycerides. While STZ-induced diabetic mice are not hyperlipidemic, they may be dyslipidemic; increased LDL and decreased HDL cholesterol has been reported in STZ-induced diabetic mice (44), though diabetes duration was twice that in the present study. In contrast to mice, 8-week STZ-induced diabetic rats have a 70% increase in total cholesterol and a sevenfold elevation of VLDL triglycerides (29). Very-high-dose RSV partially corrected the triglyceride elevation but was without effect on cholesterol. In these rats, treatment with RSV dose-dependently corrected nerve conduction velocity deficits with marked effects at doses that did not alter lipid levels. Thus, in both diabetic mice and rats, statin effects on neurovascular function can occur independently of their lipid-lowering action.

Increased endothelial NO production may account for the observed improvements in EDR, as RSV has been shown to elevate NO synthase-3 expression in mouse aorta without altering total plasma cholesterol and triglycerides (26–28). Statins also attenuate rolling, adherence, and transmigration of leukocytes (45), inhibit superoxide production, and reduce LDL oxidation (46,47) independent of lipid lowering.

The beneficial effects of RSV were markedly attenuated by cotreatment with mevalonate, the product of HMG-CoA reductase. Thus, mevalonate inhibited RSV action on aortic and cavernosal EDR and cavernosal NANC-mediated relaxation. This suggests that RSV effects are dependent upon cholesterol biosynthesis pathway inhibition, but independent of cholesterol lowering. Statins inhibit the synthesis of several metabolites downstream of mevalonate, including farnesyl and geranylgeranyl pyrophosphate. These cholesterol synthesis intermediates can isoprenylate proteins and have important roles in endothelial function (25–28). For example, RhoGTPases undergo geranylgeranyl modification, which is necessary for their membrane-associated activity. They negatively regulate NO synthase-3 expression in cultured human endothelial endothelial cells. Thus, by blocking Rho geranylgeranylation, statins can upregulate endothelial NO synthase-3 (48). However, investigations on such pleiotropic actions of statins have not been carried out in diabetic models or even under hyperglycemic conditions in tissue culture, and any change in the isoprenylation state of important regulatory proteins is unknown.

A notable finding was that RSV largely prevented and reversed the impaired NO-mediated NANC relaxation of corpus cavernosum smooth muscle. Statin effects on NANC nerves have not previously been reported. The data suggest that statins modulate the activation or expression of NO synthase in nitrergic nerves in diabetes. While the liver is the major target for cholesterol lowering by statins, HMG-CoA reductase is expressed and cholesterol is synthesized by most cells, including neural tissue (49). Thus, it is plausible that the mechanisms responsible for statin-induced improvements of nitergic nerve function in diabetes are direct and mechanistically similar to those for endothelial NO synthase-3 activation, involving altered isoprenylation and upregulation of NO synthase-1 activity. However, there are no published data on direct measurement of neuronal NO synthase-1 activity following statin therapy in the absence of diabetes, and an indirect vascular mechanism may also plausibly explain the present findings. As the nerve vascular supply is important for the development of diabetic neuropathy (2–4,29), the possibility that RSV improved endothelial function to an extent sufficient to protect major pelvic ganglion perfusion cannot be discounted—autonomic ganglion blood flow is halved by diabetes in rats (50). Such an effect would indirectly protect NANC nerve function. RSV vascular effects in brain are sufficient to protect against experimental ischemic stroke damage in nondiabetic mice (27).

RSV did not significantly alter the diabetes-induced increase in aorta responses to phenylephrine. This suggests that the mechanism responsible was not dependent on NO-linked dysfunction and remains to be established. Responses to exogenous phenylephrine or nerve-released norepinephrine were unaltered by diabetes or l-NNA in mouse corpus cavernosum, implying that the mechanism resulting in increased phenylephrine responsiveness in aorta is not present in penile tissue. However, l-NNA augmented the phenylephrine response of nondiabetic but not diabetic aorta, suggesting that basal NO release was diminished by diabetes. This is further supported by the increased sensitivity to SNP of nondiabetic but not diabetic aorta, observed following l-NNA. RSV promoted control-like responses to l-NNA, suggesting that the effect depended on improved NO bioavailability. There were no alterations of phenylephrine or SNP responses by l-NNA for corpus cavernosum. This suggests that under the in vitro experimental conditions, basal NO release is minimal in corpus cavernosum compared with aorta.

In summary, prevention and intervention RSV treatment improved aorta and corpus cavernosum NO-dependent function in diabetic mice; these benefits were largely attenuated by mevalonate cotreatment. RSV did not alter total plasma cholesterol in this model. Together the data suggest that statin effects on endothelial and nitrergic nerve dysfunction are independent of cholesterol lowering but act through inhibition of the cholesterol biosynthesis pathway. RSV may be suitable for further studies on diabetic neurovascular dysfunction, including clinical trials. The pleiotropic mechanisms of statin action suggest that the development of selective inhibitors of isoprenoid biosynthesis within the mevalonate-cholesterol pathway may potentially lead to further novel therapeutic approaches to diabetic vascular and nerve complications.

This work was supported in part by a PhD studentship (to M.R.N.) from Astra-Zeneca.