Organic nitrates such as nitroglycerin (glyceryl trinitrate; GTN) are widely used in the therapy of cardiovascular diseases such as stable and unstable angina (Abrams, 1995). The anti-ischemic effects of organic nitrates are largely caused by venous and coronary artery dilation as well as the improvement of collateral blood flow, which all decrease myocardial oxygen consumption. Their use, however, is limited because of the rapid development of tolerance and cross-tolerance characterized by decreased sensitivity of the vasculature to the organic nitrates and to endothelium-dependent vasodilators, respectively (Mangione and Glasser, 1994; Warnholtz et al., 2002). It is thought that organic nitrates induce vasorelaxation by generating nitric oxide (NO) or a related S-nitrosothiol. Both enzymatic and nonenzymatic mechanisms of organic nitrate metabolism/bioactivation have been proposed. The list of candidate enzymes includes glutathione S-transferases (Lau et al., 1992), the cytochrome P450 system (Schroder, 1992), xanthine oxidoreductase (O'Byrne et al., 2000), and mitochondrial aldehyde dehydrogenase (ALDH-2) (Chen et al., 2002). The active metabolite(s) of GTN activates the target enzyme soluble guanylyl cyclase (sGC), increasing tissue levels of the second messenger cGMP. cGMP in turn activates a cGMP-dependent protein kinase (cGK-I), which mediates vasorelaxation by phosphorylating proteins that regulate intracellular Ca²⁺ levels (McDonald and Murad, 1995; Ignarro, 2002). An impairment of the NO signaling pathway by increased formation of reactive oxygen species (ROS) (Munzel et al., 1995) as well as an impaired biotransformation of organic nitrates may contribute to the development of tolerance and cross-tolerance. ALDH-2 has recently been identified as a GTN-metabolizing enzyme that is subjected to an oxidative mechanism-based inactivation (Chen et al., 2002; Sydow et al., 2004). These results provide the missing link between tolerance and cross-tolerance, the oxidative stress concept, and a decreased GTN biotransformation. However, the ALDH-2 concept was criticized in a recent study (DiFabio et al., 2003), because an appreciable number of Asiatic people possess a mutation of ALDH-2, which causes a loss of function, but yet show no signs of inherited tolerance to GTN. This apparent discrepancy may be explained if one recalls that this mutation affects the cofactor binding (Sladek, 2003), thereby probably leaving the ALDH-2 esterase and GTN reductase activity intact.

Several organic nitrates have been reported to induce tolerance. Examples include isosorbide-5-mononitrate (ISMN), isosorbide dinitrate (ISDN), and GTN (Abrams, 1987). In contrast, pentaerythritol tetranitrate (PETN) treatment has been described to cause less tolerance (Fink and Bassenge, 1997; Jurt et al., 2001) and induce less side effects, such as headaches (Pfaffenrath et al., 1998). In animal studies using dogs or rabbits, it was shown that chronic therapy with PETN causes no tolerance and less oxidative stress and has beneficial effects on the progression of atherosclerosis in hyperlipidemic rabbits (Mullenheim et al., 2001; Schwemmer and Bassenge, 2003). These effects may be explained by the property of PETN to induce ferritin and heme oxygenase-1, as shown in cultured endothelial cells (Oberle et al., 1999, 2003). Increased expression of ferritin could decrease the tissue levels of free iron and thereby reduce the formation of free radicals induced by a Fenton-type reaction. The protective effects of increased heme oxygenase-1 expression has been shown to be secondary to increased degradation of porphyrins to form bilirubin, a potent antioxidant (Yamaguchi et al., 1996), as well as carbon monoxide, which is a weak stimulator of soluble guanylyl cyclase and inhibits platelet aggregation (Brune and Ullrich, 1987).

With the present study, we sought to determine whether or not the biotransformation of GTN by ALDH-2 represents a common pathway of organic nitrate bioactivation and whether or not all organic nitrates induce oxidative stress during bioactivation, irrespective of their molecular structure. Therefore, ISMN, ISDN, GTN, and PETN and its metabolites PETriN, PEDN, and PEMN were tested for their effects on mitochondrial ROS formation and ALDH-2 activity. Moreover, we assessed the effect of the ALDH-2 inhibitor benomyl on the vasodilator potency of these different organic nitrates in isolated, endothelium-intact rat aortic rings and observed a marked difference between PETN and GTN on one hand and the other nitrates on the other hand. Our present findings help to establish a unifying concept for the molecular mechanism underlying nitrate tolerance and cross-tolerance. Thus, in vivo GTN tolerance primarily impairs the high-potency pathway and is caused by oxidative inactivation of ALDH-2.

Materials and Methods

Materials. GTN was purchased as an ethanolic stock solution (102 g/l) from UNIKEM (Copenhagen, Denmark). For organ chamber tension studies, GTN was used from a Nitrolingual infusion solution (1 mg/ml) from G. Pohl-Boskamp (Hohenlockstedt, Germany). PETN [with 80% (w/w) lactose], PETriN, PEDN, and PEMN were kindly donated by ALPHARMA-ISIS GmbH & Co. KG (Langenfeld, Germany). ISDN [with 60% (w/w) lactose] was obtained from Alexis Biochemicals (San Diego, CA), ISMN was obtained from Acros Organics (Fairlawn, NJ), and Ellman's reagent [5,5′-dithio-bis(2-nitrobenzoic acid); DTNB] and PEG-SOD were obtained from Sigma-Aldrich (Steinheim, Germany). PETN and PETriN are barely soluble in aqueous solutions and were used from stock solutions in dimethyl sulfoxide. For better comparison, all organic nitrates were used from stock solutions in dimethyl sulfoxide. 1,2- and 1,3-GDN were purchased from Cerilliant (Round Rock, TX). Peroxynitrite [oxoperoxonitrate (1-)] was synthesized according to the quenched-flow method as described previously (Saha et al., 1998). Benomyl [methyl-1-(butylcarbamoyl)-2-benzimidazole carbamate] was purchased from Chem Service, Inc. (West Chester, PA); daidzin (7-glucoside of 4′,7-dihydroxy-isoflavone) was purchased from INDOFINE Chemical Company, Inc. (Hillsborough, NJ); 3-morpholino-sydnonimine (Sin-1) was purchased from Calbiochem (San Diego, CA); and L-012 was purchased from Wako Pure Chemical Industries (Tokyo, Japan). All other chemicals were of analytical grade and obtained from Sigma-Aldrich, Fluka (Buchs, Switzerland), or Merck (Darmstadt, Germany).

Animal Model, in Vivo Nitrate Tolerance. Wistar rats treated with ethanol (sham) or nitroglycerin (3 days, 0.48 μmol/h) were studied as described recently (Munzel et al., 2000). All animals were treated in accordance with the Declaration of Helsinki and with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the U.S. National Institutes of Health and was granted by the Ethics Committee of the University Hospital Eppendorf.

Isometric Tension Studies. Vasodilator responses to organic nitrates and other vasodilators were assessed with isolated aortic rings mounted for isometric tension recordings in organ chambers, as described previously (Munzel et al., 1995). In selected experiments, vessels from rats were incubated with benomyl, an irreversible inhibitor of ALDH (10 μM). In vitro tolerance development as a result of PETN treatment was assessed by ex vivo incubation of rat vessels with 100 μM PETN for 30 min at 37°C in Krebs-HEPES buffer (5.78 g/l NaCl, 0.35 g/l KCl, 0.37 g/l CaCl₂, 0.30 g/l MgSO₄, 2.1 g/l NaHCO₃, 0.14 g/l K₂HPO₄, 5.21 g/l HEPES, and 2.0 g/l d-glucose) followed by a 1-h washout phase and subsequent recording of concentration-response curves with PETN. Cross-tolerance of PETN to GTN was assessed by the measurement of vasodilator responses to PETN with aortic rings pre-exposed to GTN in vitro (100 μM for 30 min at 37°C) or in vivo (3 days of chronic infusion).

Activity Assay for cGK-I and sGC. Aortic segments (1 cm) from GTN- and sham-treated rats were frozen and homogenized in liquid nitrogen. Immunoblotting was performed with a mouse monoclonal antibody (16C2) specific for the phosphorylation state of VASP at serine 239 (P-VASP), as described previously (Mulsch et al., 2001; Schulz et al., 2002). The effect of benomyl (10 μM) on VASP phosphorylation was tested in the absence or presence of either acetylcholine (0.1 μM), sodium nitroprusside (0.1 μM), PETN (0.05 μM), GTN (0.1 μM), ISDN (10 μM), or ISMN (1 mM). For PETN (0.05 μM), GTN (0.1 μM), and ISDN (100 μM), the activation of sGC was determined by the formation of cGMP, which was measured by a nonradioactive assay kit from Amersham Biosciences Inc. (Freiburg, Germany). The effect of ALDH-2 inhibition on the organic nitrate-dependent stimulation of sGC was determined by the addition or exclusion of benomyl (10 μM).

ALDH-2 Dehydrogenase and Esterase Activity in Isolated Rat Heart Mitochondria. The activity of ALDH in isolated mitochondria was determined by measuring the conversion of benzaldehyde to benzoic acid. Rat heart mitochondria were prepared according to a previously published method (Raha et al., 2000). Hearts from sham-treated rats were homogenized in HEPES buffer (50 mM HEPES, 70 mM sucrose, 220 mM mannitol, 1 mM EGTA, and 0.033 mM bovine serum albumin) and centrifuged at 1500g (10 min at 4°C) and 2000g for 5 min (the pellets were discarded). The supernatant was then centrifuged at 20,000g for 20 min, and the pellet was resuspended in 1 ml of HEPES buffer. The latter step was repeated, and the pellet was resuspended in 1 ml of Tris buffer (10 mM Tris, 340 mM sucrose, 100 mM KCl, and 1 mM EDTA). The mitochondrial fraction (approximately 5-10 mg/ml total protein) was kept on ice and diluted to approximately 1 mg/ml protein in 0.25 ml of PBS and preincubated for 10 min at room temperature in the presence or absence of the ALDH-2 inhibitors benomyl (1 μM) or chloral hydrate (10 μM) as well as peroxynitrite (500 μM), hydrogen peroxide (10 mM), Ellman's reagent (5 mM), or organic nitrates (50-5000 μM, except PETN, because of its limited solubility). For the measurement of ALDH-2 dehydrogenase activity, benzaldehyde (200 μM) was added, and the samples were incubated for another 30 min at 37°C. ALDH activity in cytosolic fractions was measured in the 20,000g supernatant of heart homogenates (2 mg/ml total protein) upon incubation with 200 μM NAD⁺ and 200 μM benzaldehyde for 30 min at 37°C. To determine the oxidative inactivation of ALDH-2 by Sin-1-derived peroxynitrite, isolated mitochondria (1.6 mg/ml in PBS) were lysed by sonication (to provide direct access for peroxynitrite and PEG-SOD) and incubated for 90 min at 37°C in the presence or absence of Sin-1 (0.1 or 1 mM), PEG-SOD (500 U/ml), or dithiothreitol (DTT) (1 mM). Then 1 mM NAD⁺ and 0.5 mM benzaldehyde were added to the samples and further incubated for 30 min at room temperature. For the measurement of ALDH-2 esterase activity, methylbenzoate (1 mM) was added, and the samples were incubated for another 30 min at 37°C. Samples were sonicated and centrifuged at 20,000g (4°C) for 20 min, and the supernatant was purified by size-exclusion centrifugation through a Microcon YM-10 filter device from Millipore Corporation (Billerica, MA). Each sample (200 μl) was subjected to HPLC analysis. The system consisted of an LKB pump, a Knaur UV/Vis detector, and a C18-Nucleosil 125 × 4 100-3 reversed phase column from Macherey-Nagel (Düren, Germany). The mobile phase contained acetonitrile (35% v/v) in 50 mM citric acid buffer (65% v/v), pH 2.2. The substrate and its products were isocratically eluted at a flow rate of 0.8 ml/min, detected at 275 nm, and quantified using internal and external standards (benzoic acid, benzaldehyde, and methylbenzoate). The typical retention times were 4.2, 6.2, and 12.6 min, respectively.

Measurement of Reactive Oxygen Species Production from Isolated Heart Mitochondria. Mitochondrial stock solutions were diluted to final total protein concentrations of approximately 0.1 mg/ml in 0.5 ml of PBS. The dye L-012 (100 μM) was used as described previously (Daiber et al., 2004) to quantify ROS after the addition of the complex II substrate succinate (final concentration, 4 mM). Chemiluminescence (CL) was monitored over 5 min using a Lumat LB9507 from Berthold Technologies (Bad Wildbad, Germany), and the signal at 5 min was expressed as counts per minute. ROS production was quantified in mitochondria from control rats exposed to increasing concentrations of organic nitrates (50-5000 μM, except PETN because of its limited solubility) in vitro.

Measurement of GTN Biotransformation in RAW 264.7 Macrophages by GC-MS Analysis. RAW 264.7 macrophages were cultured to confluence in 75-cm² flasks in Dulbecco's modified Eagle's medium from Biochrom (Berlin, Germany) supplemented with 10% fetal calf serum, 1% penicillin/streptomycin, 4 mM glutamine, 1 mM pyruvate, and 1.5 mg/ml carbonate. Cells (50 × 10⁶) were incubated in 10 ml of Hanks' buffered salt solution containing 10% fetal calf serum for 30 min with 1 to 10 μM GTN. To determine the effect of ALDH-2 inhibition on GTN biotransformation, the cells were incubated with 1 μM GTN in the presence of either benomyl (50 μM) or daidzin (50 μM). The culture medium was then discarded, and the cells were scraped from the culture bottle in 1 ml of AgNO₃ (20 mM) solution (Chen et al., 2002) and frozen at -20°C. After thawing, the cells were sonicated, and GTN as well as its metabolites 1,2- and 1,3-GDN were extracted with 2 ml of ethyl acetate (25% v/v)/pentane (75% v/v) as described previously (Gerardin et al., 1982). The organic phase was centrifuged for 10 min at 1600g and evaporated using a gentle nitrogen stream. The remaining sample was dissolved in acetonitrile. An aliquot of each sample (20 μl) was injected into a GC-MS unit GC Star 3400 CX, MS Saturn 3 from Varian, Inc. (Palo Alto, CA). GTN and its metabolites were separated on an Optima delta 6 (0.25-μm film, 30-m length, 0.25-mm diameter) gas chromatography column from Macherey-Nagel. The nitrates were detected at the mass of the nitro group (m/z = 46) using the electron ionization mode. The temperatures were 150°C for the injector, 280°C for the transfer line, and 170°C for the ion trap. The temperature program for the column was 80°C for 4 min, 150°C at 5.75 min, 166°C at 13.75 min, and 300°C at 16.43 to 18.43 min. The helium flow was 2 ml/min, and typical retention times were 11.8 min for 1,2-GDN, 12.2 min for 1,3-GDN, and 13.2 min for GTN.

Statistical Analysis. Results are expressed as mean ± S.E.M. One-way analysis of variance (Bonferroni or Dunn's method) was used for comparisons of vascular responses, L-012 chemiluminescence, cGK-I activity, cGMP responses, and ALDH-2 dehydrogenase and esterase activity. The EC₅₀ value for each experiment was obtained by log-transformation.

Results

Vasodilator Responses. In endothelium-intact rat aorta, the vasodilator potency of the different organic nitrates roughly correlated with the number of -ONO₂ groups present in the molecule. PETN and GTN exhibited the highest potency (EC₅₀, 8.5 and 22 nM, respectively; Fig. 1A and Table 1), followed by PETriN (EC₅₀, 0.2 μM), ISDN (EC₅₀, 5 μM), and ISMN (EC₅₀, 170 μM). The vasodilator potency of PEDN (EC₅₀, 25 μM) was lower than that of PETriN, and PEMN exhibited the lowest potency (EC₅₀, 1 mM) (Fig. 1B and Table 1). PETN and GTN showed potency similar to the biological inducer of relaxation ACh (EC₅₀, 35 nM).

The role of ALDH-2 in the bioactivation of the organic nitrates was assessed by the addition of benomyl (10 μM) to the organ bath solution. Benomyl induced a marked rightward shift of the concentration-response curves of PETN, PETriN, and GTN (Fig. 1, A and B), indicating that ALDH-2 is involved in the bioactivation of these nitrates. In contrast, benomyl did not significantly affect the vasodilator potency of ISDN and ISMN (Fig. 1A, Table 1), thereby excluding the involvement of ALDH-2 in the bioactivation process of these two nitrates. Likewise, benomyl did not affect the vasodilator potency of the higher-denitrated PETN metabolites PEDN and PEMN (Fig. 1B). Therefore, bioactivation of PEDN and PEMN, similar to ISDN and ISMN, presumably involves another, ALDH-2-independent pathway(s). The vasodilator response to the endothelium-dependent agonist ACh was not affected by benomyl (Fig. 1B and Table 1).

In additional experiments with higher concentrations of benomyl (100 μM), we ascertained that the concentration of benomyl used (10 μM) was maximally efficacious with regard to ALDH-2 inhibition (data not shown). We thereby excluded the possibility that higher concentrations of organic nitrates surmounted inhibition by benomyl (as observed by the rightward shift of the concentration-response curve) because of the incomplete inhibition of ALDH-2. We cannot entirely exclude that the carbamoylating compound benomyl inhibits other targets in addition to ALDH-2, but Chen et al. (2002) showed that acetaldehyde, a natural substrate of ALDH-2, also inhibits GTN-dependent relaxation, probably by acting as a competitive inhibitor for GTN biotransformation (Chen et al., 2002). In a recent study, we made a similar observation with the specific ALDH-2 inhibitor daidzin, further supporting an important role for ALDH-2 in the bioactivation of GTN (Sydow et al., 2004).

To assess the influence of so-called in vitro tolerance on the vasodilator responsiveness of PETN, isolated rat aortic rings were exposed for 30 min to high concentrations of PETN (100 μM) or GTN (100 μM), followed by a 1-h washout period, before the assessment of vasodilator responses. Pretreatment with PETN caused a small but significant rightward shift of the PETN concentration-response curve (EC₅₀ increased from 8.5 to 27 nM; Fig. 1C), indicating a weak degree of tolerance/tachyphylaxis. In contrast, in vitro treatment with high concentrations of GTN induced a larger rightward shift of the PETN vasodilator response curve (EC₅₀ increased from 8.5 to 85 nM). This effect was even more pronounced when using aortic rings from in vivo GTN-treated rats (EC₅₀ increased to 182 nM; Fig. 1C and Table 1). These results indicate that PETN induces a weak in vitro autotolerance (tachyphylaxis) and GTN induces a pronounced in vitro and in vivo cross-tolerance to PETN.

Effect of ALDH-2 Inhibition on Activation of cGK-I and sGC by Organic Nitrates, SNP, and ACh. cGK-I activity was assessed by the P-VASP detected by means of Western blotting. The 48-kDa protein VASP (on blots appearing as a 45- and 50-kDa double band) is a prominent and ubiquitous cGK-I substrate (Eigenthaler et al., 1993) and can be taken as a monitor for cGK-I activity in intact cells and tissues (Mulsch et al., 2001). ACh, SNP, PETN, GTN, ISDN, and ISMN caused a concentration-dependent increase in P-VASP levels (Fig. 2A). The bar graph shows summarized densitometric data of P-VASP levels from at least three independent experiments with aortae from different animals, demonstrating a significant stimulation of cGK-I activity by all vasodilators used. The stimulatory effect of the organic nitrates on cGK-I activity increased with an increasing number of -ONO₂ groups (PETN > GTN > ISDN > ISMN); therefore, these results are in accordance with those obtained by isometric tension recordings. Coincubation with the ALDH-2 inhibitor benomyl (10 μM) significantly decreased the P-VASP levels in PETN- and GTN-stimulated vessels almost to control levels, whereas with all other organic nitrates, SNP and ACh VASP phosphorylation was not significantly altered (Fig. 2A). These results corroborate our vasodilator experiments showing that ALDH-2 specifically bioactivates PETN and GTN but not the lower-nitrated organic nitrates. The failure of benomyl to affect ACh- and SNP-dependent VASP phosphorylation proves that this ALDH-2 inhibitor does not interfere with the endothelium-dependent and -independent NO signaling pathway (i.e., NO synthase activity and sGC activation).

The activation of sGC by different organic nitrates was detected by measuring cGMP levels in rat aortic tissue with a cGMP-specific enzyme immunoassay. Treatment of isolated blood vessels with GTN (0.1 μM) or PETN (0.05 μM) caused a marked 2.5- to 5-fold increase in cGMP levels, which was almost reduced to control levels in the presence of benomyl (10 μM; Fig. 2B). In contrast, the 5-fold increase in cGMP induced by ISDN (100 μM) was slightly, though not significantly, reduced by the addition of benomyl (Fig. 2B). These results support our conclusion that GTN and PETN requires ALDH-2-dependent bioactivation to elicit a cGMP response, whereas ISDN-induced cGMP is largely independent of this bioactivation pathway. It should be noted that the high- and low-potency organic nitrates were employed at different concentrations to achieve similar pronounced effects with all organic nitrates tested.

Mitochondrial Reactive Oxygen Species Formation. The formation of mitochondrial ROS was detected by a CL-based assay using the probe L-012. Isolated rat heart mitochondria were incubated in vitro with high concentrations of organic nitrates (50 μM-5 mM) in the presence of L-012 (100 μM), and CL was recorded for 5 min (see Materials and Methods). ISMN had no marked effect on ROS formation even with the highest concentration applied (5 mM), whereas ISDN at 5 mM showed a significant 2-fold increase in the CL signal (Fig. 3A). With GTN, a marked concentration-dependent increase in ROS formation was achieved, resulting in a near 7-fold increase above control at 5 mM GTN (Fig. 3A). PETN (500 μM) induced only a 3-fold increase in CL and thus was less efficacious than 500 μM GTN (5-fold increase). With PETN and its metabolites, the capability to elicit mitochondrial ROS formation decreased with the loss of -ONO₂ groups (PETN > PETriN > PEDN ≥ PEMN) (Fig. 4A).

Inhibition of ALDH-2 Dehydrogenase Activity by Organic Nitrates. The activity of ALDH-2 in isolated rat heart mitochondria was measured by the conversion of benzaldehyde to benzoic acid, as described previously (Sydow et al., 2004). ISMN, even at a high concentration (5 mM), did not alter the activity of ALDH-2 (Fig. 3B). ISDN induced a clear concentration-dependent decrease in ALDH-2 activity, a maximum of 50% at 5 mM, whereas GTN, at intermediate concentrations (500 μM), already completely inhibited the enzyme (the remaining signal corresponds to background activity) (Fig. 3B). In contrast, PETN was a less potent inhibitor of ALDH-2 activity (Fig. 3B). ALDH-2 activity was not significantly affected by PEDN and PEMN treatment (Fig. 4B). We were surprised to find that PETriN and PETN seemed to be equally effective in inhibiting ALDH-2 dehydrogenase activity. In a comparison of Figs. 3, A and B, and 4, A and B, it becomes evident that the potency of the different organic nitrates to inhibit ALDH-2 dehydrogenase activity is mirrored by their potency to elicit mitochondrial ROS formation. These findings therefore demonstrate a direct correlation between organic nitrate-dependent ROS formation and ALDH-2 dehydrogenase inhibition. However, one should note that cytosolic ALDH activity was also inhibited upon in vivo treatment with GTN (60.8 ± 4.5 μM in control cytosol versus 23.1 ± 0.5 μM in cytosol from tolerant animals, n = 3; MV± S.E.M.).

Inhibition of ALDH-2 Dehydrogenase Activity by Sin-1-Derived Peroxynitrite. To demonstrate the sensitivity of ALDH-2 dehydrogenase activity toward oxidative inactivation, we assessed the effect of the peroxynitrite generator Sin-1 on ALDH-2 activity. Sin-1 releases peroxynitrite constantly over a time period of 90 min (at pH 7.4 and 37°C) to generate low steady-state concentrations of peroxynitrite (Daiber et al., 2002). In mitochondria from control animals, the presence of Sin-1 (1 mM) caused a 60% loss of ALDH-2 activity (15.7 ± 2.1 versus 6.1 ± 1.1 μM benzoic acid/30 min, n = 3; MV ± S.E.M.). Sin-1 at a concentration of 100 μM still caused 30% inhibition of ALDH-2 (15.7 ± 2.1 versus 9.8 ± 1.7 μM benzoic acid/30 min, n = 3; MV ± S.E.M.). The Sin-1 (1 mM)-dependent decrease in ALDH-2 activity was partially prevented in the presence of PEG-SOD (500 U/ml, 6.1 ± 1.1 versus 10.4 ± 0.8 μM benzoic acid/30 min, n = 3; MV ± S.E.M.) and completely prevented by the addition of DTT (1 mM, 6.1 ± 1.1 versus 14.8 ± 1.1 μM benzoic acid/30 min, n = 3; MV ± S.E.M.). DTT (1 mM) was also able to restore ALDH-2 activity in isolated mitochondria from in vivo GTN-treated, tolerant animals (10.1 ± 1.2 versus 17.1 ± 1.1 μM benzoic acid/30 min, n = 3; MV ± S.E.M.) to the level of control.

Inhibition of ALDH-2 Esterase Activity by Organic Nitrates. Esterase activity in isolated rat heart mitochondria was measured by the conversion of methylbenzoate to benzoic acid. Because several esterases are present in mitochondria, special precautions had to be taken to specifically determine the contribution of ALDH-2 to total esterase activity. The horizontal line 1 in Fig. 5A (right axis) defines the level of total esterase activity in control mitochondria, whereas line 2 defines the oxidation-resistant esterase activity. The esterase activity of ALDH-2 involves redox-sensitive sulfhydryl groups and should be inhibited by oxidants (Senior and Tsai, 1990; Tsai and Senior, 1991). Therefore, this activity is represented by the difference between lines 1 and 2. ALDH-2 esterase activity was almost completely inhibited by thiol-oxidants such as peroxynitrite (500 μM) and hydrogen peroxide (10 mM; Fig. 5A). It is interesting that GTN applied in vitro and in vivo also inhibited ALDH-2 esterase activity to a comparable extent (Fig. 5A). In contrast, PETN, ISDN, and ISMN did not significantly affect esterase activity (Fig. 5A). The mixed disulfide-forming reagent DTNB caused complete inhibition of total esterase activity, whereas the ALDH inhibitors benomyl (1 μM) and chloral hydrate (10 μM) inhibited only the oxidant-sensitive esterase activity (Fig. 5B), supporting the conclusion that ALDH-2 esterase activity accounts for the oxidant-sensitive esterase. PETN and its metabolites did not alter esterase activity (Fig. 5B).

Effect of ALDH-2 Inhibition on GTN Biotransformation in RAW 264.7 Macrophages. According to a previous study (Chen et al., 2002), RAW 264.7 macrophages contain appreciable amounts of ALDH-2. To confirm that ALDH-2 accounts for GTN biotransformation, we measured the formation of 1,2- and 1,3-GDN from GTN by these macrophages by means of GC-MS analysis. As shown in Fig. 6, the ratio of 1,2-GDN/1,3-GDN increased 4-fold with decreasing GTN concentrations (from 10 to 1 μM). This finding indicates that the pattern of biotransformation is highly sensitive to the GTN concentration used and is in accordance with a substrate-based inactivation of ALDH-2 by excess GTN. Experiments with the ALDH-2 inhibitors benomyl and daidzin (50 μM each) were performed in the presence of 1 μM GTN. Both inhibitors significantly decreased the 1,2-GDN/1,3-GDN ratio to approximately 1 (Fig. 6).

Discussion

We (Sydow et al., 2004) and others (Chen et al., 2002) have shown that ALDH-2 significantly biotransforms GTN in vitro and in vivo and that inhibition of this enzyme markedly decreased the vasodilator potency of GTN. In the present study, we sought to determine whether ALDH-2 is also required for the bioactivation of other commonly used organic nitrates.

We now show that ALDH-2 inhibition by benomyl decreases the vasodilator potency of PETN and PETriN in rat aorta but not that of ISDN, PEDN, ISMN, and PEMN (Fig. 1, Table 1). These findings were corroborated by our observation that only GTN- and PETN-induced activation of both cGK-I (assessed by P-VASP formation) and sGC (assessed by cGMP formation) were decreased by ALDH-2 inhibitors, whereas activation of these enzymes by the other organic nitrates were not affected by ALDH-2 inhibition (Fig. 2, A and B). Moreover, we show that GTN, PETN, and PETriN exhibit the highest potency to induce mitochondrial ROS formation (Figs. 3A and 4A) and inhibit ALDH-2 dehydrogenase activity (Figs. 3B and 4B). In general, the vasodilator potency, dependence on ALDH-2-catalyzed bioactivation, stimulation of mitochondrial ROS, and inhibition of ALDH-2 dehydrogenase activity increase with the number of -ONO₂ groups present in the organic nitrate molecule. A remarkable deviation from this general behavior was the finding that PETN and PETriN did not inhibit ALDH-2 esterase activity, whereas GTN did (Fig. 5).

The Requirement of ALDH-2 for cGMP-Dependent Vasodilator Activity of Organic Nitrates Is Confined to Nitrates with High Vasodilator Potency. The vasodilator potency of the different organic nitrates tested here roughly correlated with the number of nitrate groups present in the molecule. It decreased in the order PETN > GTN > PET-riN > ISDN > PEDN > ISMN > PEMN (Table 1). This order is in accordance with previous studies in the same and in other vessel types (Parker et al., 1975). ALDH-2 inhibition by benomyl specifically affected the vasodilator responses to GTN, PETN, and PETriN, but not to mono- and dinitrates (Table 1), indicating a specific interference with bioactivation of higher-nitrated organic nitrates. Benomyl did not influence ACh-induced relaxations; i.e., it did not interfere with the endothelium-derived NO signaling pathway. In nitrate-tolerant vessels, benomyl was unable to further decrease vasodilator responses to GTN (Sydow et al., 2004) (Table 1), suggesting that both benomyl and nitrate tolerance interfere with the same bioactivation step specifically required for GTN/PETN/PETriN. The results obtained by vasorelaxation experiments were corroborated by the measurement of cGK-I and sGC activation. GTN- and PETN-dependent activation of cGK-I and sGC, which resulted in increased levels of P-VASP and cGMP, respectively, could be suppressed to control levels by benomyl (Fig. 2). In contrast, VASP phosphorylation and cGMP formation induced by all other organic nitrates showed no significant alteration upon the addition of benomyl. Moreover, neither basal nor ACh- or SNP-stimulated P-VASP levels were affected by benomyl, thereby excluding an unspecific impairment of NO/cGMP signaling by this compound.

Two Independent Pathways Accounting for Vasodilator Activity of Organic Nitrates. In many vascular beds, the vasodilator-response curve to GTN consists of a high- and a low-potency component (Ahlner et al., 1986), suggesting that bioactivation may not be uniform for a given nitrate and between different nitrates. Superoxide-generating sGC inhibitors such as LY83583 (6-anilino-5,8-quino-linedione) and methylene blue (Malta, 1989), as well as the G_i-inhibitory pertussis toxin (Ahlner et al., 1988), primarily abolish the high-potency component, which can be overcome by higher concentrations of GTN. In contrast, the much less potent ISDN and ISMN exhibit a monophasic dose-response curve. Our previous (Sydow et al., 2004) and present findings showing that the maximal relaxation to GTN, PETN, and PETriN is not affected by benomyl supports the hypothesis of at least two pathways of bioactivation for these organic nitrates. The high-potency pathway may now be identified as the ALDH-2-dependent bioactivation step.

Organic Nitrate-Induced Oxidative Stress and Inhibition of ALDH-2 Dehydrogenase and Esterase Activity. Oxidative stress plays an important role in the development of nitrate tolerance and cross-tolerance (Hink et al., 2003). Superoxide and peroxynitrite were identified as the reactive species formed in tolerant vessels, and peroxynitrite scavengers such as uric acid and ebselen normalized endothelial dysfunction in tolerant vessels. Here we present evidence that at least a part of the organic nitrate-induced oxidative stress may originate from mitochondrial nitrate metabolism. Using L-012-dependent chemiluminescence (Daiber et al., 2004), we detected peroxynitrite and superoxide in isolated rat heart mitochondria upon in vitro administration of organic nitrates. It is interesting that the potency of organic nitrates to generate mitochondrial superoxide and peroxynitrite, as well as to inhibit ALDH-2 dehydrogenase activity, increased with their vasodilator potency. This is not contradictory, because ALDH-2 contains three cysteine residues in the catalytic center (Fig. 7), rendering the dehydrogenase activity highly sensitive toward oxidative inactivation (Senior and Tsai, 1990; Tsai and Senior, 1991). In addition to its dehydrogenase activity, ALDH-2 also exhibits esterase activity (Scheme 1), which has been proposed to be essential for the bioactivation of GTN (Chen et al., 2002). This activity also involves oxidation-sensitive cysteine residues that are different from the dehydrogenase activity (Tsai and Senior, 1991). As depicted in Scheme 2, the bioactivation of GTN by ALDH-2 could lead to oxidative inactivation of the enzyme, presumably by the formation of disulfide and/or sulfenic and sulfinic acid. Indeed, 30 to 40% of total (DTNB-sensitive) mitochondrial esterase activity was sensitive to inhibition by oxidants such as peroxynitrite and hydrogen peroxide and inhibited to a similar extent by benomyl and chloral hydrate. Therefore, this part of ALDH-2 esterase activity is expected to be essential for the bioactivation of GTN and PETN and impaired by nitrate tolerance. Indeed, we could demonstrate that GTN in vivo and in vitro treatment inhibited the mitochondrial esterase activity to a similar extent. In contrast, all other nitrates did not affect esterase activity. In accordance with the oxidative stress concept, previous reports have shown that antioxidants such as ascorbic acid positively influence the development of nitrate tolerance (Daniel and Nawarskas, 2000; Schwemmer and Bassenge, 2003) and that oxidatively inactivated ALDH-2 activity can be restored by the addition of DTT (Chen et al., 2002; Sydow et al., 2004).

Another characteristic feature of the high-potency bioactivation pathway is the rapid desensitization (tachyphylaxis) by GTN concentrations >1 μM (Chen et al., 2002). At the molecular level, tachyphylaxis can now be related to autoinhibition of GTN bioactivation by mitochondrial ALDH-2, because in RAW 264.7 macrophages GTN concentrations >1 μM decreased the formation of 1,2-GDN, which is associated with vasorelaxation and sGC activation (Brien et al., 1988). Furthermore, the formation of the 1,2-GDN metabolite was inhibited by the ALDH inhibitors benomyl and daidzin (Fig. 6).

Comparison of GTN versus PETN and PETriN. Although the highly potent nitrates share a common bioactivation pathway, we noted some remarkable differences between GTN and PETN/PETriN. First, PETN and PETriN elicited significantly less mitochondrial superoxide formation at similar concentrations compared with GTN. Second, PETN and PETriN caused less inactivation of ALDH-2 dehydrogenase activity compared with GTN. Moreover, in contrast to GTN, both PETN and PETriN did not inhibit mitochondrial esterase activity. These results may explain the previous findings that PETN lacks not only in vivo tolerance (Mullenheim et al., 2001; Schwemmer and Bassenge, 2003), but also, as shown here, in vitro tolerance. In contrast, PETN bioactivity was reduced by cross-tolerance after treatment with GTN in vivo and in vitro, indicating that both nitrates share a common bioactivation pathway. In addition, preliminary results of an in vivo PETN study demonstrate that high-dose PETN in vivo treatment neither causes significant tolerance versus PETN-dependent relaxation nor significant cross-tolerance versus GTN-dependent relaxation (M. Coldewey, A. Daiber, A. Seeling, D. Stalleicken, J. Lehmann, A. Mülsch, and T. Münzel, unpublished observations) (data not shown).

Mechanism of sGC Activation by Organic Nitrates. The present ALDH-2 concept does not provide the molecular link between mitochondrial nitrite formation and sGC activation leading to vasorelaxation (see Scheme 2). Nitrite may activate sGC via intermediate formation of HNO₂ and disproportionation into NO₂ and NO. However, because of the low pK of HNO₂, this requires at least 1000-fold higher concentrations of nitrite at physiological intracellular pH than provided by vasorelaxing concentrations of GTN, or a pH around 4. Indeed, we could not detect the release of NO from isolated blood vessels during GTN (<1 μM)-induced relaxation (Kleschyov et al., 2003). Therefore, free diffusible NO may not account for sGC activation by low concentrations of GTN. Nevertheless, we may not completely rule out the possibility that our NO EPR detection approach fails to detect minute amounts of NO generated from GTN and PETN within discrete intracellular compartments (e.g., mitochondria). However, even with this possibility admitted, our hypothesis of the existence of two different bioactivation mechanisms accounting for organic nitrates with high and low vasodilator potency is not disproved. Moreover, our observations result in another attractive hypothesis: that the vasodilator potency of organic nitrates correlates with their ability to induce mitochondrial reactive oxygen and nitrogen species formation (such as peroxynitrite). At low steady-state concentrations, peroxynitrite has been reported to cause vasodilation (Villa et al., 1994). This implies that GTN- and PETN-dependent relaxation at concentrations <1 μM may be induced by mitochondrial peroxynitrite formation.

In contrast, sGC activation by higher concentrations of nitrates seems to be mediated by free NO, because we could detect the GTN- and ISDN-derived NO at concentrations exceeding 10 μM. This ALDH-2-independent pathway is apparently shared by the other low-potency organic nitrates. Benomyl no longer consistently inhibits vasorelaxation at these high concentrations of GTN.

In summary, we provide further evidence for the hypothesis that at least two different pathways exist for the bioactivation of organic nitrates (Ahlner et al., 1986; Kleschyov et al., 2003): The high-affinity pathway metabolizes organic nitrates with high vasorelaxant potency (GTN, PETN, and PETriN), depends on ALDH-2 esterase activity, and is only operative in the submicromolar concentration range; it does not generate measurable quantities of free nitric oxide and is subject to tachyphylaxis and tolerance by either long-lasting in vivo exposure to low concentrations of these organic nitrates or short-term in vitro exposure to concentrations exceeding 1 μM GTN. The low-affinity pathway is operative at higher concentrations of the highly potent organic nitrates and for the organic nitrates with lower vasorelaxant potency (ISDN, ISMN, PEDN, and PEMN). This pathway is independent of ALDH-2, yields measurable quantities of free NO, and is less prone to tolerance development.