Spiroiminodihydantoin and Guanidinohydantoin Are the Dominant

Paige L. McKibbin , Aaron M. Fleming , Mohammad Atif Towheed , Bennett .... Sarah Delaney, William L. Neeley, James C. Delaney, and John M. Essigmann...
0 downloads 0 Views 221KB Size
Download PDF
1510

Chem. Res. Toxicol. 2004, 17, 1510-1519

Spiroiminodihydantoin and Guanidinohydantoin Are the Dominant Products of 8-Oxoguanosine Oxidation at Low Fluxes of Peroxynitrite: Mechanistic Studies with 18O Jacquin C. Niles,† John S. Wishnok,† and Steven R. Tannenbaum*,†,‡ Biological Engineering Division and Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Room 56-738A, Cambridge, Massachusetts 02139 Received January 21, 2004

Peroxynitrite-mediated oxidation of 8-oxoguanosine results in the formation of two product classes distinguished by the source of their incorporated oxygen atoms. The first product class consists of dehydroguanidinohydantoin (DGh), N-nitro-dehydroguanidinohydantoin (NO2-DGh), and 2,4,6-trioxo[1,3,5]triazinane-1-carboxamidine (CAC) with peroxynitrite as the exogenous O atom source, and the second includes spiroiminodihydantoin (Sp), guanidinohydantoin (Gh), and 4-hydroxy-2,5-dioxo-imidazolidine-4-carboxylic acid (HICA), with water serving as the exogenous O atom source. The first product class forms exclusively at high peroxynitrite fluxes, while the second forms exclusively at limiting peroxynitrite fluxes. At intermediate peroxynitrite fluxes, both sets of products are formed. At high fluxes, DGh was the major reaction product, and after several of the peroxynitrite-derived radicals were eliminated as the exogenous O atom source, the peroxynitrite anion emerged as the most likely candidate. On the other hand, at lower fluxes, either Gh or Sp was the major product, depending on the pH of the reaction mixture. At low and high pH, respectively, Gh and Sp were the major products, and the plot of pH vs ratio of Sp/(Sp + Gh) had an inflection at pH 5.8. Interestingly, the pH dependence for oxidation of 8-oxoGuo with CoCl2 and KHSO5 was identical to that for oxidation by peroxynitrite, indicating that the phenomenon arises due to characteristics of an 8-oxoGuoderived rather than an oxidant-derived intermediate, since these two systems generate different reactive species. On the basis of these findings, a model in which 8-oxoGuo is oxidized to the bisimine intermediate, 1 is proposed. At high peroxynitrite fluxes, the reaction of 1 with ONOOpredominates over the reaction with H2O, leading exclusively to DGh, NO2-DGh, and CAC, while at limiting peroxynitrite concentrations, the reaction with H2O dominates, and Gh and Sp are formed exclusively. At intermediate peroxynitrite fluxes, the relative kinetics of the reaction between 1 and ONOO- or H2O are such that both product classes are formed. To explain the pH-dependent Gh and Sp yields, we propose that 5 has a pKa ∼ 5.8 and that the differential reactivity of the protonated and deprotonated form of 5 leads to its partitioning into Gh and Sp, respectively.

Introduction Peroxynitrite is a very reactive compound of biological interest derived from the diffusion-limited reaction between nitric oxide and superoxide (1). Peroxynitrite chemistry is complex, with pH and CO2 being two very important modifiers. The peroxynitrite anion (ONOO-) is stable, but the conjugate acid, peroxynitrous acid (pKa ) 6.8), rapidly decomposes (t1/2 ∼ 1 s at 37 ° C, pH 7) (2), yielding •OH and •NO2 free radicals in 28% yield (3-5). This proton-catalyzed pathway is the dominant decomposition mechanism in the absence of CO2. In the presence of CO2, however, ONOO- undergoes a rapid direct reaction with CO2 (k ) 3-5 × 104 M-1 s-1) (6, 7) to give the short-lived intermediate ONOOCO2-, which decomposes into CO3•- and •NO2 free radicals in 33% yield (8-10). The rapid reaction between peroxynitrite and CO2 ensures that in the absence of substrates capable of competing with CO2, it will decompose almost exclusively * To whom correspondence should be addressed. † Biological Engineering Division. ‡ Department of Chemistry.

via the CO2-catalyzed pathway (96-98%) under simulated physiologic conditions ([CO2] ∼ 1-1.3 mM), while the proton-catalyzed pathway will account for the remaining 2-4%. Peroxynitrite is reactive toward a variety of biologically relevant targets, including thiols (11, 12), antioxidants (12, 13), proteins (14-19), lipids (20), and, of particular interest to our group, DNA (21-25). In the latter case, damage to guanine is the predominant chemistry (26, 27), and several guanine-derived oxidation products, namely, 8-oxoG (23), 8-nitroG (28), nitroIm (29), Iz (26), and spiroiminodihydantoin (Sp)1 (26, 30), have been identified. We have also been investigating the further reactivity of 8-oxoG with peroxynitrite to test the hypothesis that 8-oxoG might be an important cellular target for peroxynitrite and other oxidants, leading possibly to toxic 1 Abbreviations: The 8-oxoGuo products listed are 2,3,5-tri-O-acetylβ-D-erythro-pentofuranosyl derivatives and are denoted as follows. DGh, dehydroguanidinohydantoin; NO2-DGh, N-nitro-dehydroguanidinohydantoin, CAC, 2,4,6-trioxo[1,3,5]triazinane-1-carboxamidine; Sp, spiroiminodihydantoin; Gh, guanidinohydantoin; HICA, 4-hydroxy-2,5dioxo-imidazolidine-4-carboxylic acid.

10.1021/tx0400048 CCC: $27.50 © 2004 American Chemical Society Published on Web 10/14/2004

Spiroiminodihydantoin and Guanidinohydantoin

and/or mutagenic products. Several products have been identified, including dehydroguanidinohydantoin (DGh) and N-nitro-dehydroguanidinohydantoin (NO2-DGh), both of which hydrolyze via parabanic acid to oxaluric acid (OA), and 2,4,6-trioxo[1,3,5]triazinane-1-carboxamidine (CAC), which hydrolyzes to cyanuric acid (CA) (31, 32). While DGh was the major product in the absence of reductant, in the presence of thiols, Sp became the major product (30). In these earlier studies, authentic peroxynitrite was introduced as a bolus to the nucleoside solution, generating high instantaneous peroxynitrite concentrations, thus raising the question of whether bolus addition is a suitable model for peroxynitrite chemistry in vivo. To address this issue, several models aimed at simulating constant low peroxynitrite fluxes have emerged and have proven helpful in elucidating peroxynitrite reactivity with the biologically relevant target tyrosine, which is nitrated in certain pathophysiologic conditions (33-35). In bolus addition studies in vitro, tyrosine yields 3-nitrotyrosine (3-NO2-Tyr) as a major product (9-17%) and 3,3′-dityrosine (3,3′-di-Tyr) as a minor product (0.5-5%) (36, 37). When peroxynitrite is infused, the yields of 3-NO2-Tyr and 3,3′-di-Tyr are each 10% (37). Furthermore, at the lowest peroxynitrite or peroxynitrite-derived radical fluxes generated using either xanthine oxidase and a NONOate (38, 39), pulse radiolysis (40), SOTS and NONOate (37), or 3-morpholinosydnonimine (SIN-1) (36, 37), it is clear that 3,3′-di-Tyr yields increase sufficiently (4-12%) that this compound becomes the major product at the expense of 3-NO2-Tyr (0.1-0.3%). This shift occurs because, at very low peroxynitrite fluxes, recombination of two Tyr• radicals efficiently competes with the reaction between Tyr• and •NO2 (37, 39, 40). Recognizing that changing peroxynitrite fluxes can dramatically alter product profiles by impacting the relative significance of competing reactions, we wanted to investigate whether the same is true for the reaction between peroxynitrite and 8-oxoGuo. To do this, we have probed the overall mechanism by characterizing the incorporation of 18O when the reaction was carried out using 18O-labeled peroxynitrite, H2O, O2, and CO2 and have measured how the yield of various classes of products defined by the source of their 18O atom(s) changed as a function of the peroxynitrite flux.

Materials and Methods 18O 2

General. and H218O (95 atom % 18O) were purchased from Isotec (Miamisburg, OH) and Aldrich (Milwaukee, WI), respectively. All solvents used were HPLC grade. Instrumentation. UV/vis measurements were made with an HP8452 Diode Array Spectrophotometer (Hewlett-Packard, Palo Alto, CA). HPLC was performed on an HP 1100 binary pumping system equipped with either a 1090 or an 1100 Diode Array Detector (Hewlett-Packard). Electrospray ionization mass spectrometry (ESI-MS) experiments were carried out using either an HP 5989B (Hewlett-Packard) or an LC-MSD-Trap-SL (Agilent, Palo Alto, CA) operated in either positive or negative ion modes. GC-MS experiments were done on an HP5989A mass spectrometer. Quantitative data are reported primarily as ratios and are based on integration of chromatographic peak areas. Peroxynitrite Synthesis. Peroxynitrite was prepared either by ozonolysis of an alkaline solution of sodium azide (Fisher) solution (41) or by the reaction of isoamyl nitrite (Aldrich) with hydrogen peroxide (Fisher) at pH 12-14 (42). In the latter case, excess hydrogen peroxide was removed by passing the peroxynitrite solution through a 7.5 cm × 1.5 cm manganese oxide

Chem. Res. Toxicol., Vol. 17, No. 11, 2004 1511 (Aldrich) column. 18O-labeled peroxynitrite was prepared by reacting 53 µmol of H218O2 (ICON, Summit, NJ) in 1 M HCl with 53 µmol of NaNO2 as previously described (43). The reaction was quenched by rapidly adding 35 µL of 5 M NaOH, and unreacted hydrogen peroxide was removed as above. Peroxynitrite concentrations were determined by making dilutions in 0.1 M NaOH and measuring the absorbance at λ ) 302 nm ( ) 1670 M-1 cm-1) (43). Analytical HPLC. A 250 mm × 4.6 mm, 5 µm Columbus C18 column (Phenomenex, Torrence, CA) was used with 50 mM aqueous ammonium acetate (A) and acetonitrile (B) as mobile phases. The elution conditions consisted of a 10 min isocratic wash with 5% B for 10 min followed by a gradient to 40% B over 20 min. An isocratic wash at 40% B for 5 min was followed by a gradient down to 5% B over 5 min. The flow rate was 1.0 mL/min, and the products were monitored simultaneously at 230 and 252 nm. Fractions containing the products were collected and analyzed by ESI-MS either directly or after lyophilizing and reconstituting in ∼50 µL of double-distilled water. Reactions with Infused Authentic Peroxynitrite. Authentic peroxynitrite was infused using a Harvard syringe pump (model 22) and 0.25 mm i.d. PEEK tubing into a vigorously stirred buffered nucleoside solution (200 µM) to achieve fluxes between 0.67 and 13.3 µM/s. The products in the final reaction mixtures were quantitated directly by HPLC, as described above. Isotope incorporation studies were carried out by infusing peroxynitrite (10 µL at µM/s over 10 min) into a vigorously stirred solution containing 2′,3′,5′-tri-O-Ac-8-oxoGuo (247 µM; 6 µL from a 7 mM stock), 150 mM KH2PO4, pH 7.5 buffer (100 µL), NaHCO3 (24 mM; 4 µL from a 1 M stock), and H218O (50 µL; 95 atom %). The final calculated isotopic purity of H218O was ∼28 atom %. The total reaction volume of 170 µL was the minimum that gave reproducible product yields in our setup. Products were analyzed by ESI-LC-MS and examined for 18O incorporation. For those primary products that incorporated label, exchange with bulk solvent after formation was ruled out by diluting the reaction mixture into unlabeled buffer and demonstrating that there was no change in the isotope enrichment level. Additionally, we determined that the starting nucleoside does not exchange with bulk solvent under our reaction conditions used, eliminating this as a possibility for label incorporation into products. Reactions with SIN-1. 2′,3′,5′-Tri-O-Ac-8-oxoGuo (200 µM) was treated with SIN-1 (0.1-0.5 mM) in 150 mM KH2PO4, 25 mM NaHCO3, and pH 7.5 buffer. SIN-1 stocks were prepared in double-distilled water and used within 5 min. Reaction mixtures were incubated at 37 °C, and aliquots were taken at regular time points and after 4 h for HPLC and/or LC-ESI-MS analysis. For isotope incorporation studies, 150 mM KH2PO4, pH 7.5 (150 µL), was dried in vacuo, and the residue was dissolved in H218O (50 µL). 2′,3′,5′-Tri-O-Ac-8-oxoGuo (2 µL), NaHCO3 (2 µL), and SIN-1 (2 µL) were added to achieve concentrations of 200 µM, 35 mM, and 1.2 mM, respectively, and the mixture was incubated at 37 °C for 1.5 h. The final calculated 18O content was ∼85 atom %. Products from this reaction were analyzed by LC-MS using the MSD-Trap in positive ion mode. Reactions with KHSO5/CoCl2. 2′,3′,5′-Tri-O-acetyl-8oxoGuo (0.2 µmol), KHSO5 (0.29 µmol), and CoCl2 (3.3 nmol) were incubated at ambient temperature in 1 mL of buffered solution for 2 h prior to analysis. Buffers were 150 mM KH2PO4, at pH 4.5, 6.6, or 7.2. The pH 7.2 buffer contained either 0 or 25 mM NaHCO3. Reaction of 2′,3′,5′-Tri-O-acetyl-8-oxoGuo with Peroxynitrite in the Presence of 18O2, H218O, and C18O2. Reactions done under 18O2 atmosphere were carried out in argon-purged 2 mL screw cap vials (Agilent) fitted with rubber septa. The buffer (150 mM potassium phosphate, 25 mM sodium bicarbonate, pH 7.6) was degassed by sparging with argon and then allowed to equilibrate over several hours with 18O2 that had been introduced into the headspace. The nucleoside and peroxynitrite

1512

Chem. Res. Toxicol., Vol. 17, No. 11, 2004

Niles et al.

Figure 1. Changes in the yield of Sp, DGh, and overall oxidation of 8-oxoGuo at varying peroxynitrite fluxes at pH 7.2-4 are summarized. The “other products” were not measured directly and were determined instead using the equation: [other products] ) [8-oxoGuo]oxidized - [Sp + DGh]. stock solutions were also thoroughly degassed in order to reduce the concentration of dissolved 16O2. The reaction of 2′,3′,5′-triO-acetyl-8-oxoGuo (200 µM) in 18O2-equilibrated buffer (1 mL) with peroxynitrite (1.5 mM) was initiated by adding the latter to the vigorously vortexed nucleoside solution. All solutions were introduced into the reaction vial through the septum using gastight syringes. The hydrolytically unstable DGh and NO2DGh were reduced with sodium borohydride to guanidinohydantoin (Gh) and NO2-Gh prior to ESI-MS, while CAC was directly analyzed. For reactions carried out in the presence of H218O and C18O2, 50 µL of 150 mM potassium phosphate, 25 mM NaHCO3, and pH 7.2 buffer in H16O2 were dried in vacuo and the residue was then taken up in 50 µL of H218O. The CO2 content was restored to ∼1-1.2 mM by adding 2.5 µL of 0.5 M NaHCO3 in H216O. Carbonic anhydrase (1 µM, 1.5 µL) was added to ensure rapid equilibration between H218O and C18O2, after verifying that it had no impact on the ratio or identity of the products formed. Incorporation of 18O label into CO2 was ascertained by analyzing the headspace gas by GC-MS in split mode using a PoraPlot Q column. The nucleoside (in this study, 3′,5′-di-O-acetyl-8-oxodG was used instead of 2′,3′,5′-tri-O-acetyl-8-oxoGuo, but the results hold true for both substrates since the relevant chemistry occurs on the nucleobase rather than on the sugar) (0.52 mM, 7 µL) was added to this mixture prior to the addition, with vigorous vortexing of peroxynitrite (2.5 mM, 2.7 µL). The final reaction mixture was ∼77 atom % (calculated) in H218O. A 5 µL aliquot was subjected to LC-MS analysis. For the LC, mobile phases were aqueous 50 mM ammonium acetate (A) and acetonitrile (B), respectively, and a 150 mm × 1 mm, 5 µm Columbus C18 column (Phenomenex) was used at a flow rate of 65 µL/min. The solvent was programmed from 5 to 30% B over 30 min then to 40% B over 10 min before returning to 5% B over 5 min. For the LC-MS analyses, an 80/20 2-propanol/water, 2% ammonium hydroxide solution was used as sheath liquid (65 µL/min) and spectra were obtained in negative ion mode.

Results Previously, we have shown that when 8-oxoGuo reacts with peroxynitrite added as a bolus, the major immediate product is DGh, with NO2-DGh and CAC being formed in smaller amounts (31, 32). In the present study, we wanted to investigate whether the extent of 8-oxoGuo oxidation and the resulting product profile change as a function of instantaneous peroxynitrite concentrations present during reaction. To achieve lower solution per-

oxynitrite fluxes, we (i) infused authentic peroxynitrite at varying rates, while delivering the same total amount, and (ii) generated peroxynitrite in situ using SIN-1. Impact of Various Delivery Modes on 8-OxoGuo Oxidation. In these studies, 2′,3′,5′-tri-O-Ac-8-oxoGuo was prepared and used as a model substrate for 8-oxoGuo chemistry, as previously described (32). Overall, oxidation of this substrate varied depending on the mode of peroxynitrite delivery into the reaction mixture. In bolus addition experiments, nucleoside consumption was 17 mol % of the total delivered peroxynitrite. This was the same both in the presence and in the absence of CO2 in the reaction buffer. By infusion, the oxidation yield increased slightly to 19 mol % for an infusion rate of 0.67 µM/s. Last, in SIN-1 experiments, the oxidation yield was 25 mol %. Therefore, it appears that decreasing peroxynitrite fluxes in reactions with 8-oxoGuo leads to an increase in the overall oxidation yield, suggesting that oxidation was more efficient at lower peroxynitrite concentrations. Impact of Delivery Mode on Product Profile. 1. Peroxynitrite Infusion Reactions. During infusion reactions, peroxynitrite was delivered into pH 7.2-7.4 buffered 2′,3′,5′-tri-O-Ac-8-oxoGuo solutions containing CO2 over 1, 10, and 20 min, producing instantaneous fluxes between 13.3 and 0.67 µM/s. As shown in Figure 1, with infusion, Sp becomes the major product. As the infusion rate decreased, the absolute Sp yield increased, along with increasing overall consumption of 2′,3′,5′-triO-Ac-8-oxoGuo. Thus, because the absolute yield of DGh remained constant, this compound accounted for a smaller fraction of the total observed products. Figure 2 shows that DGh yield normalized to the amount of 8-oxoGuo reacting increases with peroxynitrite fluxes up to 2.5 µM/s after which further increases in oxidant flux lead to no increase in DGh yield. Conversely, normalized Sp yields decrease with increasing peroxynitrite flux, and similar to DGh, no change in Sp yield occurs at oxidant fluxes above 2.5 µM/s. The levels of the other bolus addition products decreased to levels below our limits of quantitation. However, a new product not detected in bolus addition experiments was found in infusion reactions; this has been identified as 4-hydroxy-2,5-dioxo-

Spiroiminodihydantoin and Guanidinohydantoin

Figure 2. There is an inverse dependence of Sp and DGh yields on peroxynitrite flux, with DGh and Sp as the major products at high and low fluxes, respectively. Table 1. Summary of Source and Number of Incorporated 18O Label in the Various Peroxynitrite-Induced Oxidation and Nitration Productsa exogenous O atom source compound CAC DGh NO2-DGh Spb Gh HICA

O2

x (1)

H2O

x (1) x (1) x (3)

ONOOx (2) x (1) x (3)

a The numbers in parentheses indicate the numbers of exogenous O atoms incorporated from the respective sources. b Sp incorporates only one exogenous O atom. When peroxynitrite is added as a bolus in the presence of a thiol, O2 provides this atom (30). During infusion of authentic peroxynitrite or in situ generation, H2O is the source.

imidazolidine-4-carboxylic acid (HICA) by mass spectrometry (44). Last, Gh was also formed during infusion reactions, especially at lower pH (vide infra). 2. In Situ Peroxyxnitrite Formation. When SIN-1 was used to generate peroxynitrite in situ, Sp was the major product, accounting for greater than 90% of the consumed 2′,3′,5′-tri-O-Ac-8-oxoGuo. HICA and Gh were also identified. However, the immediate bolus addition products DGh, NO2-DGh, and CAC were not detected. Because the SIN-1 reactions required extended incubation periods, it was possible that these compounds degraded into their hydrolytic products, namely, OA and CA. However, we were unable to detect these products by HPLC-UV and LC-MS, leading us to conclude that the major bolus addition products were not formed under these conditions. Exploring the Mechanism of Formation of the Various Products. 1. Bolus Addition Specific Products. The dramatic differences in the product profiles of the bolus addition, infusion, and SIN-1 reactions led us to investigate the mechanism(s) by which these products were formed. Specifically, the source of exogenous O atoms incorporated into the various products was explored using 18O-labeled O2, H2O, and peroxynitrite, and the results are summarized in Table 1. The bolus addition products DGh, NO2-DGh, and CAC incorporated 18O label exclusively from peroxynitrite. DGh incorporated a single 18 O label. This was determined by negative ion MS analysis of the stable Gh, obtained by sodium borohydride

Chem. Res. Toxicol., Vol. 17, No. 11, 2004 1513

reduction of DGh. The product isolated from reactions with unlabeled and labeled peroxynitrite gave [M - H]ions at m/z 414 and 416, respectively, consistent with the presence of a single 18O label in this product. Similarly, NO2-DGh, was analyzed as its stable reduced derivative, NO2-Gh. Using unlabeled peroxynitrite, the [M - H]- ion was observed at m/z 459, while with 18 O-labeled peroxynitrite, ions were observed at m/z 461 (118O), 463 (218O), and 465 (318O). This suggested that up to three 18O atoms were incorporated into NO2-DGh, although, given its structural similarity to DGh, only one 18 O atom at the C4 position was expected. We hypothesized that two additional 18O atoms were being introduced via the nitro group, since, with the exception of the O atom at C4, all of the other oxygen atoms in NO2DGh are derived from the unlabeled parent nucleoside. To test this hypothesis, 3-NO2-Tyr from the reaction of tyrosine with unlabeled and 18O-labeled peroxynitrite was analyzed. Using unlabeled peroxynitrite, the 3-NO2-Tyr [M - H]- ion was observed at m/z 225, while using labeled peroxynitrite, ions were detected at m/z 225 (no 18 O), 227 (one 18O), and 229 (two 18O). This finding confirmed that the nitro groups derived from our 18Olabeled peroxynitrite contributed up to two 18O labels, and it supports our hypothesis that two of the 18O atoms in NO2-DGh are those of the nitro group. Last, CAC was shown to incorporate two peroxynitritederived 18O atoms. Using unlabeled peroxynitrite, ions for CAC were observed at m/z 428 [M - H]- and 386 [M - CH2N2]-, but with labeled peroxynitrite, ions were observed at m/z 432 and 390 [M - CH2N2]-, indicating the presence of two 18O atoms in the CAC from the latter experiment. Given the complex chemistry of peroxynitrite, it is important to distinguish whether the incorporated O atoms were derived from a related radical, namely, •OH, CO3•-, and •NO2, or directly from ONOO-. Experiments were performed in 25 mM NaHCO3 (∼1 mM CO2) containing phosphate buffer, where it is highly unlikely that •OH is the predominant proximal source of the incorporated O atom, since, under these conditions, >95% of the added peroxynitrite will degrade via its reaction with CO2 with minimal concomitant •OH formation. A contribution from •OH chemistry in the formation of DGh, NO2-DGh, and CAC cannot be excluded, however, and this is further addressed in the Discussion. Next, we explored whether O- transfer from the CO3•to a reaction intermediate could explain the observed results. Indeed, O- atom transfer from CO3•- to •NO2 has been reported (45) and may represent a pathway by which peroxynitrite decomposes to NO3- and CO2 (9). Therefore, we prepared 18O-labeled CO2 by equilibrating 150 mM KH2PO4 and 25 mM NaHCO3, pH 7.2-7.4, in H218O in the presence of carbonic anhydrase. At equilibrium, 18O incorporation into CO2 as determined by GCMS analysis of headspace gas was found to be 10, 33, and 57%, respectively, of unlabeled, monolabeled, and dilabeled product. Because two of three O atoms in CO3•are derived from CO2 and one from peroxynitrite, the above label distribution would lead to a product 16O/18O ratio ∼ 1.7 on an incorporated O atom basis, assuming no significant isotope effect. However, under these conditions, none of the products were labeled. This result was not unexpected, since, based on our experiments using labeled peroxynitrite and unlabeled CO2, the incorporation of 18O reflected the isotopic purity of the ONOO-,

1514

Chem. Res. Toxicol., Vol. 17, No. 11, 2004

with no dilution from unlabeled CO2. Taken together, these results indicate that O- atom transfer from CO3•is not a major mechanism leading to 18O incorporation into the reaction products. Furthermore, because •NO2 is not known to serve as an O- atom donor, •NO2-derived O atoms are expected only in the nitration product. Thus, by ruling out •OH, •NO2, and CO3•-, this leaves ONOOas the likely O atom donor, and this is expanded upon in the Discussion. 2. Infusion and SIN-1 Specific Products. Previously, we have shown that Sp formed during bolus addition reactions in the presence of a thiol incorporated an O atom derived from molecular oxygen (30). However, in the present studies, we have determined that Sp formed during peroxynitrite infusion and SIN-1 reactions incorporated a water-derived O atom. This conclusion is based upon finding that Sp obtained from reactions in H216O and H218O, respectively, gave rise to [M + H]+ ions at m/z 442 and 444, respectively. Sp, therefore, is formed via two distinctly different mechanisms depending upon the reaction conditions. Additionally, we found that Gh from experiments carried out in H216O and H218O gave rise to [M + H]+ ions at m/z 416 and 418, respectively, indicating that this product, similarly to Sp formed during infusion conditions, incorporated a single 18O atom from H2O. For both compounds, this finding was independent of pH. The absolute Sp and Gh yields varied significantly over the pH range 4-8 and were inversely related to each other, with Sp and Gh as the major products at high and low pH, respectively (Figure 3 A). DGh quantitation in these experiments was problematic, since it readily decomposed during the time required for analysis into OA, which has a very low extinction coefficient, making it difficult to reliably quantitate by HPLC with UV detection. At pH e 5.8, Gh was the major product, while at pH g 5.8, Sp was the major product. To distinguish between general and specific acid/base mechanisms, reactions were carried out with infused peroxynitrite in 150 mM sodium acetate buffer at pH 4 and 7.4. At pH 4, the Gh yield was identical to that obtained in 150 mM KH2PO4, and similarly, the Sp yield was the same in the two buffer systems at pH 7.4. Previously, using CoCl2 and KHSO5 to oxidize 2′,3′,5′tri-O-Ac-8-oxoGuo, pH-dependent Sp and Gh yields were observed (46). Therefore, we carried out 2′,3′,5′-tri-O-Ac8-oxoGuo oxidation reactions using CoCl2 and KHSO5 over the pH range 4-8 in 150 mM KH2PO4 buffer. Virtually the same patterns observed for peroxynitritemediated oxidation were seen using the CoCl2 and KHSO5 oxidation system (Figure 3B). With respect to Sp and Gh, these findings suggested that (i) there is a common intermediate that partitions in a pH-dependent manner to give Sp and Gh; (ii) the same intermediate(s) is formed using CoCl2/KHSO5; and (iii) a specific acid/ base, rather than a general acid/base mechanism, is implicated in the partitioning phenomenon at the nucleoside level.

Discussion In this study, it has been demonstrated that the product profile changes dramatically depending on the rate at which peroxynitrite is delivered into buffered 8-oxoGuo solutions. A main finding is that at physiologic pH, Sp is the major product at the relatively low flux

Niles et al.

Figure 3. pH-dependent yields of Gh and Sp for oxidation of 8-oxoGuo by peroxynitrite (A) and CoCl2/KHSO5 (B).

rates generated by infusion of authentic peroxynitrite and in situ production from SIN-1. This contrasts with our previous finding that DGh is the major product arising from bolus peroxynitrite addition. Using 18O-labeled O2, H2O, peroxynitrite, and CO2, we have shown unambiguously that the major bolus addition products (DGh, NO2DGh, and CAC) contain peroxynitrite-derived O atoms, and the major infusion or SIN-1 products (Sp and Gh) and the relatively minor product [HICA (44)] incorporated H2O-derived O atoms, thereby giving significant insights into the overall reaction mechanism. Mechanistically, we propose that oxidation of 8-oxoGuo occurs via two sequential and rapid one-electron oxidations to yield first the 8-oxoGuo radical and then the electrophilic intermediate, 1 (Scheme 1). Thermodynamically, oxidation of 8-oxoGuo, which has an electrochemically observed oxidation peak at 0.58 V (47, 48) and a midpoint potential, Eo ) 0.76 V (49) relative to NHE, by the peroxynitrite-derived •OH [Eo ) 1.9 V (50)], CO3•- [Eo ) 1.5-1.59 V (51, 52)], and •NO2 [Eo ) 1.04 V (50)] radicals, is favorable. Previously, the latter two have been shown to produce the 8-oxoGuo radical with bimolecular rate constants of 7.9 × 108 (53) and 5 × 106 M-1 s-1, respectively (54). While a rate constant has not been reported for the reaction between •OH and 8-oxoGuo, it is likely to be similar to or faster than that previously determined for dG, namely, 7.8 × 109 M-1 s-1 (55). While it is not entirely clear whether the second oxidation step

Spiroiminodihydantoin and Guanidinohydantoin

Chem. Res. Toxicol., Vol. 17, No. 11, 2004 1515

Scheme 1. Mechanism Proposed for Product Formation during Peroxynitrite-Induced Oxidation of 8-OxoGuo

precedes or succeeds subsequent reactions, we have assumed the former. In our proposed mechanism, the fate of 1 determines which products are formed. At high peroxynitrite fluxes, e.g., those occurring during bolus addition of authentic peroxynitrite, we propose that the peroxynitrite anion, ONOO-, acts as a nucleophile and attacks 1 at C5 to give 2. Precedent for ONOO- acting as a nucleophile is provided by its rapid reaction with CO2 (k ) 3-5 × 104 M-1 s-1) (6, 7) and ketones (k e 1.6 × 104 M-1 s-1) (56). Homolysis of the expectedly weak O-O bond (57) in 2 produces the alkoxyl radical, 3, which rearranges with cleavage of the C5-C6 bond and elimination of either CO or CO2 to give 4, a guanidinylidene radical species. The final products DGh and NO2-DGh are produced when 4 either abstracts an H atom or traps •NO2, respectively. CAC formation is more complicated and appears to occur via two distinct pathways, one through the further oxidation of DGh (31) and the other through a hydrolytically unstable product (unpublished results). In both cases, the CAC produced incorporates two peroxynitritederived O atoms. In proposing the importance of ONOO- as a nucleophile during 8-oxoGuo oxidation, several known peroxynitrite-derived radicals have been ruled out as being responsible for the incorporation of peroxynitrite-derived O atoms in the reaction products. We have demonstrated directly that CO3•--derived O atoms are not incorporated into any of the reaction products and that •NO2 can only account for O atom incorporation in the nitration product, NO2-DGh. The arguments against a significant direct role for •OH are based on (i) the relative kinetics of proton vs CO2-catalyzed peroxynitrite decomposition; (ii) the total number of electrons lost during oxidation of 8-oxoGuo to

final products; and (iii) the high reactivity of •OH radical toward nucleosides. The first-order rate constant for proton-catalyzed peroxynitrite decomposition at 25 °C at pH 7.4 is 1 s-1 (2), as compared with 30-50 s-1 for the CO2-catalyzed pathway at [CO2] ) 1 mM (approximately 25 mM NaHCO3, as in our experiments) and taking into account a bimolecular rate constant ∼3-5 × 104 M-1 s-1 for the reaction of ONOO- with CO2. Thus, 96-98% of the peroxynitrite is expected to decay via the CO2catalyzed pathway and only 2-4% via the intermediacy of ONOOH in the proton-catalyzed pathway. The established stoichiometry for peroxynitrite decomposition via these two pathways indicates that per mole of peroxynitrite, 67 (CO2-catalyzed pathway) and 56% (protoncatalyzed pathway) one-electron oxidizing equivalents are available to react with substrate (3-5, 8-10). Therefore, of the 2-4% peroxynitrite that decays via the protoncatalyzed pathway, only 1-2% one-electron oxidizing equivalents are available to react with substrates. Overall, formation of DGh and NO2-DGh involves a fourelectron oxidation, and CAC involves a six-electron oxidation of 8-oxoGuo, or a two-electron and four-electron oxidation, respectively, from 1. Given that the DGh and NO2-DGh combined yield was ∼6% per mole of peroxynitrite, at most, 8-16% of this can be generated exclusively through the •OH/•NO2 radical-mediated reaction. Assuming generously that 1 is generated by oxidizing equivalents from the CO2-catalyzed decomposition pathway, the expected proton-catalyzed derived oxidizing equivalents can only account for up to 32% of the combined DGh and NO2-DGh yields. Importantly, assuming that the reaction of •OH with 8-oxoGuo has a similar rate constant as that for Guo, •OH will compete with CO3•- for reaction with 8-oxoGuo, and thus, a

1516

Chem. Res. Toxicol., Vol. 17, No. 11, 2004

significant amount of the •OH produced will be consumed in this reaction. Altogether, these considerations emphasize that an •OH-dependent pathway alone cannot fully account for the incorporation of ONOO--derived O atoms into the above products. By the same analysis, ONOOH (maximum 2-4% yield) alone cannot quantitatively account for the observed O atom incorporation in DGh and NO2-DGh. Thus, while our experiments do not allow us to rule out or distinguish a small contribution from these two pathways (ONOOH vs •OH), having eliminated the other peroxynitrite-derived species, ONOO- itself remains as the most likely source of the peroxynitritederived O atoms in DGh and NO2-DGh. Indeed, precedent exists for direct reaction of peroxo compounds with reactive intermediates produced during Guo and 8oxoGuo oxidation by MnTmPyP/KHSO5 (58). In those studies, incorporation of a KHSO5-derived O atom into DGh was inferred after ruling out O2 and H2O as sources. Our results do not exclude alternative pathways to DGh formation, such as reaction of the 8-oxoGuo radical with superoxide (59), but emphasizes that multiple routes to the same final products exist depending upon the oxidation system employed. Our studies, in which ONOO- has been directly identified as the exogenous O atom source for DGh, are consistent with the above findings and further strengthens the precedent. Having strong evidence for the direct involvement of ONOO- in the formation of DGh and NO2-DGh, we reasoned that if it is serving as a nucleophile, then reducing its instantaneous concentration in the reaction mixture should allow other nucleophiles to compete for reactive intermediates. When authentic peroxynitrite was infused, in addition to the bolus addition products, the major product at neutral pH was Sp, while that at acidic pH was Gh. Furthermore, when peroxynitrite was generated in situ using SIN-1, Sp and Gh were again observed in a pH-dependent fashion, but importantly, the products containing peroxynitrite-derived O atoms were not formed. We have shown unambiguously that Sp and Gh incorporate H2O-derived O atoms and thus must arise from the reaction of water with an electrophilic intermediate, which we have proposed is 1 (Scheme 1), as previously suggested (60, 61). In explaining the pH-dependent Gh and Sp yields, we have proposed that at low pH, H2O reacts with 1 at C5 to give 5, which is then attacked at C6 by a second H2O to give 6. This reaction is similar to that proposed for the uricase-mediated oxidation of uric acid to allantoin, which are analogous to 8-oxoG and Gh, respectively (62). Rearrangement of 6 with cleavage of the N1-C6 bond results in 7, which then undergoes a 1,2-carboxylate shift to give 8, a reaction supported by results from two independent groups studying the oxidation of uric acid (62, 63). Decarboxylation of 8 gives the final product Gh. Formation of HICA has been described and is proposed to occur via the C4-hydration of 5 to give 9, which subsequently undergoes ring opening and hydrolysis to give HICA via the intermediate 10 (44). Given the complexity of the overall peroxynitritemediated oxidation of 8-oxoGuo with multiple reactive species and intermediates, defining the source of the exogenous O atom in Sp and Gh was helpful in proposing the simplified model in Scheme 1. First, it established that ONOO- and H2O compete for reactive intermediates, and in our model, 1 is proposed as the intermediate reactive toward both nucleophiles. Second, observing the same pattern of isotope incorporation using CoCl2 and

Niles et al.

Figure 4. pH-dependent ratios Sp/(Sp + Gh) for 8-oxoGuo oxidation by peroxynitrite and CoCl2/KHSO5. The solid and dashed lines are from the fitted Henderson-Hasselbach equation; each shows an inflection point at pH 5.8.

KHSO5 as the oxidizing system along with identical pHdependent product yields indicated that common 8oxoGuo-derived intermediates, rather than the specific oxidant, dictated these patterns. Indeed, with the CoCl2 and KHSO5 system, SO4•- is the proximal one-electron oxidant (64) and is poised thermodynamically (Eo ) 2.43 V) (50) and kinetically to mediate purine oxidation chemistry (65). Furthermore, in studies of 8-oxoGuo oxidation by singlet oxygen (61) and riboflavin-mediated photosensitization (59), 5 has been proposed as a key intermediate, and similar pH-dependent Gh and Sp yields to those in our studies were observed. In our proposed mechanism, ONOO- and H2O compete for the same intermediate, and thus, the combined DGh/ NO2-DGh/CAC and Sp/Gh/HICA yields reflect the total flux through 2 and 5, respectively. In our scheme, knowing that Sp and Gh arise via incorporation of H2Oderived O atoms and that their pH-dependent formation is due to 8-oxoGuo-derived intermediates provided insight into the partitioning of 5 into Sp and Gh. Plots of Sp/(Sp + Gh) vs pH show inflections for both ONOO- and CoCl2/ KHSO5 at pH 5.8 (Figure 4), which can thus be interpreted as the pKa of 5. Therefore, when 5 is protonated, electrophilicity at C6 is enhanced, favoring attack by H2O, and the C5-C6 bond is rendered relatively electron deficient, inhibiting its migration. However, upon deprotonation, attack by H2O at C6 is disfavored, and C5-C6 bond migration occurs more readily, a major driving force for this reaction being formation of the stable dihydantoin, Sp. We have assumed that a single intermediate 1 is capable of giving rise to both Sp and Gh. While we have been unable to isolate 1, 5-methoxy-8-oxoGuo can be isolated and its hydrolysis products identified (44). Under acidic conditions, 4-OMe-Gh, a Gh analogue, is the primary product while at neutral pH Sp is formed, presumably via exchange of the C5-OMe group with H2O to give 1. Thus, it seems reasonable to propose that 1 is sufficient to explain the pH-dependent formation of Sp and Gh. However, we cannot exclude the possibilities that (i) these products are formed via reaction of H2O with two distinct intermediates that have different pHdependent reaction rates and (ii) a single intermediate has two distinct protonation sites, with the status of each influencing Sp and Gh formation, respectively. Unfortunately, the pH-dependent flux through 2 could not be meaningfully carried out since (i) multiple peroxynitrite-derived species exist depending on buffer pH and

Spiroiminodihydantoin and Guanidinohydantoin

CO2 content; (ii) it is difficult to maintain defined solution CO2 concentrations at different pH values; and (iii) the relevant reacting species cannot always be distinguished based on product identity and incorporation of peroxynitrite-derived O atoms. Thus, while it is possible at neutral pH to ensure that peroxynitrite decays almost entirely via reaction of ONOO- with CO2 and/or 8-oxoGuo-derived reactive intermediates with minimal •OH formation, at lower pH, significant contributions from •OH and HOONO result. Indeed, we have observed DGh formation at pH 4.5 during bolus addition and infusion experiments done in the absence of added CO2 (unpublished results), where ONOO- concentration is