Formation of Nitroxyl and Hydroxyl Radical in Solutions of Sodium Trioxodinitrate

Despite its negative redox potential, nitroxyl (HNO) can trigger reactions of oxidation. Mechanistically, these reactions were suggested to occur with the intermediate formation of either hydroxyl radical (·OH) or peroxynitrite (ONOO–). In this work, we present further experimental evidence that HNO can generate ·OH. Sodium trioxodinitrate (Na2N2O3), a commonly used donor of HNO, oxidized phenol and Me2SO to benzene diols and ·CH3, respectively. The oxidation of Me2SO was O2-independent, suggesting that this process reflected neither the intermediate formation of ONOO– nor a redox cycling of transition metal ions that could initiate Fenton-like reactions. In solutions of phenol, Na2N2O3 yielded benzene-1,2-diol and benzene-1,4-diol at a ratio of 2:1, which is consistent with the generation of free ·OH. Ethanol and Me2SO, which are efficient scavengers of ·OH, impeded the hydroxylation of phenol. A mechanism for the hydrolysis of Na2N2O3 is proposed that includes dimerization of HNO to cis-hyponitrous acid (HO-N=N-OH) with a concomitant azo-type homolytic fission of the latter to N2 and ·OH. The HNO-dependent production of ·OH was with 1 order of magnitude higher at pH 6.0 than at pH 7.4. Hence, we hypothesized that HNO can exert selective toxicity to cells subjected to acidosis. In support of this thesis, Na2N2O3 was markedly more toxic to human fibroblasts and SK-N-SH neuroblastoma cells at pH 6.2 than at pH 7.4. Scavengers of ·OH impeded the cytotoxicity of Na2N2O3. These results suggest that the formation of HNO may be viewed as a toxicological event in tissues subjected to acidosis.

The biochemistry of nitroxyl (HNO) has attracted considerable interest in recent years. In cells, the biosynthesis of HNO is believed to proceed via reduction of NO ⅐ by superoxide dismutase (1) and cytochrome c (2), and reduction of S-nitrosoglutathione by low molecular weight and protein thiols (3)(4)(5). It has been suggested that HNO can affect the etiology of various pathophysiological conditions such as inflammation and neurodegenerative diseases, especially when H 2 O 2 and transition metal ions are present (6,7). Similar to NO ⅐ and NO ϩ , HNO is a potent inducer of the antioxidant protein heme oxygenase 1 (8), exhibits vasorelaxant properties (9), and modulates the activity of thiol-containing proteins, such as aldehyde dehydrogenase (10,11) and the N-methyl-D-aspartate receptor (12,13). In in vivo experiments, Paolocci et al. (14) observed that HNO exerts positive inotropic and lusitropic action, which unlike NO ⅐ and nitrates is independent and additive to ␤-adrenergic stimulation and increases the release of plasma calcitonin gene-related peptide; these results suggest that donors of HNO are potential prodrugs for the treatment of heart failure (14). At high doses, HNO has been shown to induce DNA singlestrand breakage (15,16) and a concentration-dependent cytotoxicity in murine thymocytes (16). This cytotoxicity was associated with activation of the nuclear nick sensor enzyme poly(ADP-ribose)polymerase, perturbation of the mitochondrial membrane potential, and an increased production of superoxide (16).
On a molecular level, there are several differences between the reactivity of NO ⅐ and HNO that may account for the distinct biological effects of the latter species: in contrast to NO ⅐ , HNO directly interacts with thiols (4,17), it preferentially binds to Fe III complexes (18,19), and acts as a hydroxylating agent (15,20,21). Recently, we have reported that HNO can generate ⅐ OH in a pH-dependent manner (20). Because of its high reactivity, ⅐ OH is one of the most toxic species that can be formed in biological systems. This free radical reacts with most cellular molecules at diffusion-controlled rates; thus it cannot diffuse from its site of generation further than the nearest molecules (22). Hence, we hypothesized that the pH-dependent generation of ⅐ OH from HNO can have toxicological significance, particularly because tissue acidification occurs under various pathological conditions, such as hypoxia, inflammation, and cancer (23)(24)(25). In the present work, we provide further experimental evidence that HNO generates ⅐ OH in an oxygen-independent manner. We also report that HNO exhibits a pH-dependent toxicity to normal human fibroblasts and SK-N-SH neuroblastoma cells that could be impeded by scavengers of radical species.

EXPERIMENTAL PROCEDURES
Reagents-All reagents used were purchased from Sigma. The solutions used in the experiments were prepared in deionized and Chelex 100-treated water or potassium phosphate buffer. Sodium trioxodini-trate was either purchased from Calbiochem, Inc. (La Jolla, CA) or synthesized as described in Ref. 26.
HPLC 1 Analysis-HPLC was performed with a Waters liquid chromatograph (Milford, MA). Separation was achieved with a C-18 reverse phase column (Microsorb, 4.6 mm ϫ 25 cm, 5 m, 100 A Rainin Instrument Co., Inc., Emeryville, CA). The mobile phase was saturated with helium and contained 10 mM lithium perchlorate and either water with 30% (v/v) methanol for analysis of phenol and benzene diols, or 70% methanol for analysis of N-tert-butyl-␣-phenylnitrone (PBN) derivatives. All HPLC analyses were conducted at a flow rate of 1 ml/min. Electrochemical detection of PBN-derived adducts was carried out at ϩ0.8 V with a LC-4C/CC5 amperometric system (Bioanalytical Systems, West Lafayette, IN) equipped with glassy carbon electrode and a Ag/AgCl reference electrode (27). Phenol and benzene diols were analyzed electrochemically at ϩ0.95 V.
Cell Experiments-Normal human fibroblasts or SK-N-SH neuroblastoma cells (800 -1000 cells per plate) were treated for 30 min at 37°C with Na 2 N 2 O 3 in 50 mM phosphate buffer (pH 6.2-7.4) containing 0.15 M NaCl and 0.2 mM CaCl 2 . Thereafter, the fluid was removed, the cells were covered with minimal essential medium containing 10% fetal bovine serum and incubated at 37°C for 4 h. In selected experiments, the incubation medium containing Na 2 N 2 O 3 also included ascorbic acid (5 mM) plus superoxide dismutase (300 units/ml), catalase (500 units/ ml), and EDTA (0.1 mM), or ␣-(4-pyridyl-1-oxide)-N-tert-butylnitrone (10 mM) or Me 2 SO (0.2 M). On completion of the incubations, the cell number was determined by the crystal violet method.
Anaerobic Experiments-To achieve anaerobic conditions, all solutions were placed in septum-caped vials and purged for 30 min with a stream of nitrogen. Thereafter, additions to the final reaction solution were made through the septum of the corresponding vial using a 0.10-ml gas-tight syringe.

RESULTS
EPR Analysis of the Hydrolysis of Na 2 N 2 O 3 -In model studies aimed at mimicking the biochemistry of HNO, sodium trioxodinitrate (Na 2 N 2 O 3 ; Angeli's salt) is often used as a donor of this species. Depending on the degree of protonation, the stability of this compound in aqueous solutions follows the se- (pK 1 ϭ 3.0 and pK 2 ϭ 9.35; Scheme 1) (28). 1c is relatively stable in alkaline solutions (pH Ͼ 10). However, the rate of decomposition of 1b within the pH interval 4 -8 is [H ϩ ]-independent and leads to the formation of HNO (k 1b ϭ 5.1 ϫ 10 Ϫ4 M Ϫ1 s Ϫ1 (29)). The latter species can dimerize to cis-hyponitrous acid (3a), which is unstable and decomposes to N 2 O and H 2 O. The decomposition of 3a is especially fast in aqueous solutions with pH 6 -13 (30,31), which most likely reflects shifts in the equilibrium between 3a and 3b in favor of the latter (Scheme 1). Bonner and Ravid (29) reported that the hydrolysis of Na 2 (O 15 NNO 2 ) at either pH 3.0 or 8.5 yields exclusively 15 N 2 O and NO 2 Ϫ , implying that this process follows the reaction sequence presented in Scheme 1 (29). However, these authors noted that the proportions of 15 N in the reaction products were different at pH 5.0, suggesting that some modification of pathway cannot be ruled out. At pH Ͻ 4, the decomposition rate of 1a increases with increasing acidity with production of NO ⅐ (28,29).
We recently reported that 1b can convert primary alcohols to aldehydes via the intermediate formation of ⅐ OH (20). In these experiments, the formation of ⅐ OH was characterized by EPR spin trapping analysis. However, we could not evaluate the absolute amounts of spin-trapped ⅐ OH as the resulting nitroxides are readily converted to EPR silent hydroxylamines under reductive conditions (32). Quantitative evaluation of the latter reaction is important because the production of ⅐ OH from HNO may have toxicological implications. Hence, we have carried out EPR/HPLC-UV/EC spin trapping experiments optimized for the quantitation of ⅐ OH under reductive conditions.
In the presence of Fe III and N-methyl-D-glucamine dithiocarbamate (MGD), the hydrolysis of 1b was paralleled by the appearance of the characteristic EPR spectrum of ⅐ ON-Fe II -MGD formed via the interaction of HNO and Fe III -MGD ( Fig.  1A; Scheme 2) (18). The formation of ⅐ ON-Fe II -MGD was H ϩindependent within the pH interval of 4.5 to 7.4 ( Fig. 1A), which is in agreement with previous findings (28) that the rate of 1b hydrolysis at these proton concentrations is constant. The substitution of Fe III -MGD with 5,5Ј-dimethyl-1-pyroline N-oxide (DMPO) resulted in the appearance of the typical EPR spectrum of DMPO/ ⅐ OH (6; Fig. 1B) (33), suggesting that the hydrolysis of 1b resulted in the formation of ⅐ OH. In contrast to the formation of HNO, the generation of 6 from 1b was strongly affected by the acidity of the reaction solutions. The latter implies that ⅐ OH was not directly derived from 1b but rather followed the release of HNO (Fig. 1, panel A compared with panel B ; Fig. 4D). For each experiment, a fresh stock solution of Na 2 N 2 O 3 was prepared in 10 mM NaOH).

SCHEME 1
The low stability of ⅐ OH-derived nitroxides (t1 ⁄2 ϳ30 s to 15 min (34,35)) is a limiting factor for quantification of ⅐ OH. To solve this experimental difficulty, we used an HPLC protocol for quantification of ⅐ OH that is based on the oxidation of Me 2 SO (32). The latter is oxidized by ⅐ OH to ⅐ CH 3 , which forms relatively stable nitroxides (t1 ⁄2 Ͼ 48 h) with PBN. The hydrolysis of 1b in the presence of Me 2 SO and PBN produced the typical EPR spectrum of PBN/ ⅐ CH 3 (4; Fig. 2 trace 2) (33). However, both 1b and HNO could act as reductants (36,37), suggesting that the EPR spectrum of 4 may not reflect the real amount of ⅐ OH and ⅐ CH 3 formed in this reaction system. In the presence of reductants, nitroxides can be readily reduced to the corresponding EPR silent hydroxylamines (32). In support of the latter assumption, the addition of K 3 [Fe(CN) 6 ] to an extract of a reaction solution consisting of 1b, Me 2 SO, and PBN resulted in a pronounced increase of the EPR signal of 4. This effect most likely reflected the oxidation of the EPR silent hydroxylamine 5 to 4 (Scheme 2). When the reaction solution was analyzed by HPLC with electrochemical detection, the predominant formation of 5 was observed (Fig. 3); an addition of K 3 [Fe(CN) 6 ] to the analyzed solutions resulted in the interconversion of the HPLC peaks reflecting the elution of 5 and 4, respectively. The identity of compounds 4 and 5 was confirmed by coinjections of authentic HPLC standards as described previously (32,38).
The formation of 4 and 5 in solutions of 1b, PBN, and Me 2 SO was well controlled and with an yield of 7.5% of the initial concentration of 1b (Fig. 4). The actual production of ⅐ OH in this reaction system, however, cannot be estimated as the efficiency of the ⅐ OH-dependent oxidation of Me 2 SO and the subsequent spin trapping of ⅐ CH 3 are undefined. Under anaerobic conditions, the reaction profile remained unchanged (Fig.  4A, open circles), which attests that the generation of ⅐ OH from HNO was O 2 -independent and reflected neither the intermediate formation of ONOO Ϫ (Scheme 4) (21, 39) nor the occurrence of Fenton-like reactions. Maximal production of 4 and 5 was observed within the pH interval of 4 to 6. At pH 6, the production of 4 and 5 was 1 order of magnitude higher than that at pH 7.4.
Hydroxylation of Phenol in Aqueous Solutions of Na 2 N 2 O 3 -Phenol is often used as a molecular probe to discriminate free ⅐ OH from other oxidizing species. For example, radiolytically generated ⅐ OH reacts with phenol via either abstraction of the hydrogen atom from its -OH function (k H ϭ 2.1 ϫ 10 9 M Ϫ1 s Ϫ1 ) or aryl addition (k Ar ϭ 6.6 ϫ 10 9 M Ϫ1 s Ϫ1 ) to give a phenol phenoxyl (10) or dihydroxycyclohexadienyl (8) radical, respectively (40,41). Under aerobic conditions, 8 interacts with O 2 to form benzene diols (9) and the superoxide anion radical; benzene-1,2-diol and benzene-1,4-diol, forming at a ratio of 2:1, are the main isomers generated in this reaction (41). In contrast, oxidation of phenol by Fenton-type reagents such as L x Fe II -OOH leads to the formation of benzene-1,2-diol and benzene-1,4-diol at a ratio that depends on the ligand of the corresponding iron complex and ranges from 11 to ϱ (41). It should be noted that 8 and 10 can follow several reaction pathways, implying that the formation of 9 can only be a qualitative marker for free ⅐ OH. For example, the hydroxylation of phenol by ⅐ OH in the absence of oxygen (or other electron acceptors, such as K 3 [Fe(CN) 6 ] and quinones (42)) predominantly leads to the formation of biphenyl diols (11 (41, 43)) (Scheme 3). Fig. 5 depicts the HPLC-EC profile of a reaction system consisting of 1b and phenol in 0.1 M phosphate buffer (pH ϭ 4.0). 1b caused a time-and dose-dependent hydroxylation of phenol to 9 with an overall yield of 80% (Fig. 6, A and B). Within the pH interval of 4 to 7 (⌬ pH ϭ 0.5), the ratio between benzene-1,2-diol and benzene-1,4-diol was 2.2 ϩ 0.24 (n ϭ 5; Thereafter, the reaction products were extracted with ethyl acetate (2 ϫ 6 ml), the extract was diluted 10 times in methanol and EPR spectra were recorded in the absence (spectrum 3) or presence of 1 mM K 3 [Fe(CN) 6 ] (spectrum 4; incubation time, 10 min). mean Ϯ S.E.; Fig. 6C), which further supports the notion for the generation of free ⅐ OH in solutions of 1b. In the absence of 1b, no formation of 9 was observed (Fig. 6A, open circles). The 1b-dependent hydroxylation of phenol proceeded with two pH optimums (Fig. 7). Because the pK of phenol is 9.98, the effects of H ϩ on the formation of 9 most likely reflected changes in the rates of reactions 8 3 9 and 8 3 10, respectively. Hence, pH variations in this reaction system may affect the formation of ⅐ OH and 9 to different extents. The production of 9 increased with increasing temperatures up to 35°C, whereas at higher temperatures a considerable autoxidation of the reaction products was observed (data not shown). The impeded formation of 9 under anaerobic conditions (Fig. 8A) was attributed to a shift in the equilibrium between 8 and 10, which ultimately resulted in a predominant formation of 11 (41). In contrast, the formation of 9 was markedly increased when the reaction solutions were purged with a stream of air. The latter effect was most likely because of a more efficient oxidation of 8 to 9 (40,(42)(43)(44). Desferrioxamine did not affect the formation of 9 to any significant extent, indicating that metal ions did not participate in the overall reaction mechanism. However, the hydroxylation of phenol was inhibited by scavengers of ⅐ OH such as ethanol and Me 2 SO (Fig. 8B) that are known to interact with this radical species at appreciable rates (k Me2SO ϭ 7 ϫ 10 9 M Ϫ1 s Ϫ1 and k EtOH ϭ 2.2 ϫ 10 9 M Ϫ1 s Ϫ1 , respectively (45,46)). In these experiments, the observed dose-dependent effects of Me 2 SO and ethanol on the 1b-induced hydroxylation of phenol is consistent with the competition of these compounds for free ⅐ OH (as has been previously noted, k Ar ϭ 6.6 ϫ 10 9 M Ϫ1 s Ϫ1 ).
Effects of pH on the Cytotoxicity of Na 2 N 2 O 3 -The pH-dependent production of ⅐ OH from Na 2 N 2 O 3 suggests that this compound may exert a H ϩ -amplified cytotoxicity. This possibility is interesting because acidification of tissues occurs under various pathological conditions, such as ischemia, inflammation, and cancer (24,47,48). For example, the intense metabolism of glucose to lactic acid leads to acidification in the microenvironment of tumor tissues. In actively glycolyzing tumors, the extracellular pH is in the range of 6.0 to 7.0 (reviewed in Ref. 24), whereas the extra-and intracellular milieu of most tissues has a pH of 7.4. Fig. 9 depicts the cytotoxic effect of Na 2 N 2 O 3 on normal human fibroblasts and SK-N-SH neuroblastoma cells. A marked cytotoxicity was observed at pH 6.2, as compared with pH 7.4, respectively. The toxic effect of Na 2 N 2 O 3 was pronounced several hours after the treatment of the cells. Delayed cell death induced by exposure to oxidative stress has been described in a number of other cell systems and has usually been attributed to apoptosis (49,50). The cytotoxicity of Na 2 N 2 O 3 was ⅐ OH-dependent as suggested by the protective effect of ascorbic acid, ␣-(4-pyridyl-1-oxide)-N-tert-butylnitrone, and Me 2 SO, which are efficient scavengers of ⅐ OH (Fig. 9, B and C) (32,34). No toxicity was observed when cells were treated with acidic buffers that did not contain Na 2 N 2 O 3 (data not shown).

DISCUSSION
The distinct cellular responses that HNO can trigger have made this species the subject of intense research. However, in contrast to NO ⅐ , for which there is a relatively well established mechanistic and derivative chemistry, our knowledge of HNO remains sparse. HNO is a weak acid whose pK value and redox potential have recently been corrected from 4.5 and Ϫ0.3 V to 11.4 and Ϫ0.8 V, respectively (37,51,52). Although HNO is a strong reductant, it has been established that this species triggers reactions of oxidation. Mechanistically, these reactions were suggested to occur with the intermediate formation of either ⅐ OH (15,20,21) or (an isomer of) (39) ONOO Ϫ (Scheme 4) (21,53). However, there are conflicting reports regarding the formation of ONOO Ϫ in acidic-to-neutral solutions of HNO. The potential of Na 2 N 2 O 3 , a commonly used donor of HNO, to generate ONOO Ϫ via the intermediate formation of singlet NO Ϫ ( 1 NO Ϫ ) was the focus of a series of studies. Upon relaxation of 1 NO Ϫ , a triplet state of NO Ϫ ( 3 NO Ϫ ) is formed that reacts with O 2 to form ONOO Ϫ (51,54). Various substrates of ONOO Ϫ were reported to undergo oxidation in neutral aqueous solutions of Na 2 N 2 O 3 with yields of the corresponding reaction products ranging from 1.9 to 65% (21,53). On the other hand, Donald et al. (55) reported that ONOO Ϫ and its product of decomposition, NO 3 Ϫ , are not formed in neutral solutions of Na 2 N 2 O 3 . In experiments aimed at assessing the Na 2 N 2 O 3 -dependent hydroxylation of phenol (pH 5-7.4), we could not observe the formation of 4-nitrophenol, which is the expected product of the interaction of phenol with ONOO Ϫ (56, 57). Shafirovich and Lymar (51) have recently pointed out that nitrogen (ϩ1), if formed in biological systems, must be present in the form of 1 HNO (pK a ϭ 11.4); direct addition of O 2 to 1 HNO that could yield ONOO Ϫ is spin forbidden and cannot be rapid, if it occurs at all (51). Measurable amounts of ONOO Ϫ , however, could be observed in alkaline solutions of Na 2 N 2 O 3 , which led to the cautionary notes that the use of aged stocks of this compound may result in contaminations with ONOO Ϫ (51,55).
The formation of ⅐ OH in solutions of Na 2 N 2 O 3 was first suggested by Hughes and Wimbledon (28), who observed that at pH Ͻ 3.0 H 2 N 2 O 3 (1a) decomposes via a free radical chain reaction that could be inhibited by ethanol (28). Because at pH Ͻ 3 H 2 N 2 O 3 generates HNO 2 and nitric oxide, the genera- tion of ⅐ OH was proposed to occur via the intermediate formation of 1d (Scheme 4) (28). Buchholz and Powell (58) suggested a similar mechanism for the formation of ⅐ OH in acidic solutions of 3a (3a 3 3c 3 ⅐ OH). However, direct detection of ⅐ OH in these reaction systems has not been presented.
Recently, Wink et al. (59) reported that 1b is toxic to Chinese hamster V79 lung fibroblast cells. At a molecular level, 1b exposure resulted in DNA double strand breaks in whole cells (59). This observation is in agreement with the studies of Ohshima et al. (15,60) who reported that HNO caused DNA strand breakage and base oxidation via an HNO-dependent generation of ⅐ OH. Scavengers of ⅐ OH, metal chelators, superoxide dismutase, and catalase impeded the 1b-dependent oxidation of DNA, indicating that superoxide anion radical, H 2 O 2 , and free ⅐ OH were involved in this process. These results led to the thesis that ⅐ OH could be generated via an HNO-dependent reduction of O 2 to superoxide anion radical (HNO ϩ O 2 3 NO ⅐ ϩ H ϩ ϩ O 2 . ) (15). However, the latter mechanism was not supported by the studies of Nelli et al. (61), who reported that in the presence of chelators of metal ions HNO does not undergo O 2 -dependent oxidation to NO ⅐ (Scheme 4). It should be pointed out that in solutions of Na 2 N 2 O 3 the occurrence of Fenton-like reactions via redox cycling of transition metal ions cannot be ruled out; recently, Al-Ajlouni and Gould (36) reported that HN 2 O 3Ϫ (1b) can directly reduce iron complexes (Scheme 4).
In this work, we present further experimental evidence that 1b can generate ⅐ OH. In aqueous solutions of 1b, phenol and Me 2 SO were oxidized to benzene diols and ⅐ CH 3 , respectively. Because the oxidation of Me 2 SO was O 2 -independent (Fig. 4A), it could be concluded that this process reflected neither the intermediate formation of ONOO Ϫ nor a redox cycling of transition metal ions that could initiate Fenton-like reactions. The treatment of aqueous solutions of phenol with 1b yielded benzene-1,2-diol and benzene-1,4-diol at a ratio of 2:1, which is consistent with the generation of free ⅐ OH (Fig. 6C). At pH 4, the overall production of benzene diols was 14% of the initial concentration of 1b (Fig. 7). However, the production of ⅐ OH could be higher, as it is unlikely that the hydroxylation of phenol proceeded quantitatively. The oxidation of phenol and Me 2 SO was optimal at pH 4 -5, implying that ⅐ OH was not directly released from 1b; the formation of HNO within the same pH interval was H ϩ -independent (Fig. 1, panel A compared with B, and Fig. 4D) (28,29). Furthermore, GSH and Fe III -MGD, which are efficient scavengers of HNO, markedly inhibited the hydroxylation of phenol (data not shown). These results allowed us to propose a reaction mechanism for the hydrolysis of 1b that includes dimerization of HNO to HO-NϭN-OH (3a) with a concomitant azo-type homolytic fission of the latter acid to N 2 and ⅐ OH (Scheme 4). A similar mechanism was established for the decomposition of dialkyl esters of 3a to RO ⅐ and N 2 (62). Within this mechanism, the effects of increasing concentrations of H ϩ on the formation of ⅐ OH can be explained with a shift in the equilibrium between 3a and 3b in favor of 3a (Scheme 1). The impeded production of ⅐ OH from 1b at pH Ͻ 4 most likely reflected the lack of formation of HNO; at pH Ͻ 4, 1b hydrolyzes to NO ⅐ (28).
With regard to the effects of pH on the ⅐ OH production from HNO, it is interesting to speculate that this phenomenon may have a toxicological significance. The HNO-dependent production of ⅐ OH is 1 order of magnitude higher at pH 6.0 than at pH 7.4. It could be generalized that metabolic hyperactivity or limited oxygen supply can cause a decrease of tissue pH. Acidosis is characteristic for such disease states as sepsis, arthritis, ischemia, and cancer. In these diseases, the intra-and/or extracellular pH of the affected tissues typically decreases from control values of 7.4 to ϳ6.0 (23,47,48,63,64). Hence, we hypothesized that HNO can exert selective toxicity to cells subjected to acidosis via a mechanism that includes an H ϩamplified generation of ⅐ OH. In support of this thesis, 1b was markedly more toxic to human fibroblasts at pH 6.2 than at pH 7.4. The cytotoxicity of 1b was ⅐ OH-dependent as suggested by the protective effect of scavengers of free radicals. In comparative experiments with human fibroblasts and SK-N-SH neuroblastoma cells, 1b exhibited a similar toxicological pattern, which leads to the conclusion that the extracellular pH was the predominant mediator of the toxic effect. 1b is a highly hydrophilic compound that has a partition coefficient for n-octanol/ water of CLogP ϭ Ϫ1.332 (estimated with ChemDraw; Cam-bridgeSoft.com, Inc., Cambridge, MA). Hence, in biological systems the 1b-dependent generation of HNO will predominantly occur in the extracellular milieu. Shafirovich and Lymar (51), however, proposed that the diffusibility and membrane permeability of HNO is similar to that of NO ⅐ (51), which implies that the intracellular interactions of HNO formed from 1b will also contribute to the toxicological effect of this compound. In an in vivo model of myocardial ischemia and reperfusion injury, Ma et al. (65) reported that 1b exhibited a marked necrotic effect, whereas S-nitrosoglutathione minimized muscle injury (65). The mechanisms responsible for these opposite effects are not well understood. It is tempting to speculate that the cytotoxicity of HNO released from 1b could be amplified by its conversion to ⅐ OH in ischemia-acidified tissue (64). However, further studies are needed to understand the effects of HNO in biological systems, as well as to advance the mechanistic algorithms for interpretation and prediction of the cytotoxicity of HNO-releasing prodrugs such as Na 2 N 2 O 3 , derivatives of nitrosobenzene (10), benzisothiazol (66), chloropropamide (67), and N-hydroxybenzene carboximidate (68). FIG. 9. Toxicity of Na 2 N 2 O 3 in normal human fibroblasts and SK-N-SH neuroblastoma cells. Normal human fibroblasts (panels A-C) or SK-N-SH neuroblastoma cells (panel C) were treated with Na 2 N 2 O 3 in 50 mM phosphate buffer (panels A and C, pH 6.2 and 7.4; panel B, pH 6.2). In selected experiments, the incubation medium containing Na 2 N 2 O 3 also included ascorbic acid (5 mM), ␣-(4-pyridyl-1-oxide)-N-tert-butylnitrone (POBN) (10 mM), or Me 2 SO (0.2 M; panel B). Each experimental point represents the mean of three experiments Ϯ S.E., expressed as a percentage of the cell count immediately following treatment. SCHEME 4