Highly efficient conversion of superoxide to oxygen using hydrophilic carbon clusters

Significance

Mechanistic studies of nontoxic hydrophilic carbon cluster nanoparticles show that they are able to accomplish the direct conversion of superoxide to dioxygen and hydrogen peroxide. This is accomplished faster than in most single-active-site enzymes, and it is precisely what dioxygen-deficient tissue needs in the face of injury where reactive oxygen species, particularly superoxide, overwhelm the natural enzymes required to remove superoxide. We confirm here that the hydrophilic carbon clusters are unreactive toward nitric oxide radical, which is a potent vasodilator that also has an important role in neurotransmission and cytoprotection. The mechanistic results help to explain the preclinical efficacy of these carbon nanoparticles in mitigating the deleterious effects of superoxide on traumatized tissue.

Abstract

Many diseases are associated with oxidative stress, which occurs when the production of reactive oxygen species (ROS) overwhelms the scavenging ability of an organism. Here, we evaluated the carbon nanoparticle antioxidant properties of poly(ethylene glycolated) hydrophilic carbon clusters (PEG-HCCs) by electron paramagnetic resonance (EPR) spectroscopy, oxygen electrode, and spectrophotometric assays. These carbon nanoparticles have 1 equivalent of stable radical and showed superoxide (O₂^•−) dismutase-like properties yet were inert to nitric oxide (NO^•) as well as peroxynitrite (ONOO⁻). Thus, PEG-HCCs can act as selective antioxidants that do not require regeneration by enzymes. Our steady-state kinetic assay using KO₂ and direct freeze-trap EPR to follow its decay removed the rate-limiting substrate provision, thus enabling determination of the remarkable intrinsic turnover numbers of O₂^•− to O₂ by PEG-HCCs at >20,000 s⁻¹. The major products of this catalytic turnover are O₂ and H₂O₂, making the PEG-HCCs a biomimetic superoxide dismutase.

Reactive oxygen species (ROS), such as superoxide (O₂^•−), hydrogen peroxide (H₂O₂), organic peroxides, and hydroxyl radical (^•OH), are a consequence of aerobic metabolism (1, 2). These ROS are necessary for the signaling pathways in biological processes (3, 4) such as cell migration, circadian rhythm, stem cell proliferation, and neurogenesis (5). In healthy systems, ROS are efficiently regulated by the defensive enzymes superoxide dismutase (SOD) and catalase, and by antioxidants such as glutathione, vitamin A, ascorbic acid, uric acid, hydroquinones, and vitamin E (6). When the production of ROS overwhelms the scavenging ability of the defense system, oxidative stress occurs, causing dysfunctions in cell metabolism (7–16).

In addition to ROS, reactive nitrogen species (RNS) such as nitric oxide (NO^•), nitrogen dioxide, and dinitrogen trioxide can be found in all organisms. NO^• can act as an oxidizing or reducing agent depending on the environment (17), is more stable than other radicals (half-life 4–15 s) (18), and is synthesized in small amounts in vivo (17–22). NO^• is a potent vasodilator and has an important role in neurotransmission and cytoprotection (17, 18, 22, 23). Owing to its biological importance and the low concentration found normally in vivo, it is often important to avoid alteration of NO^• levels in biological systems to prevent aggravation of acute pathologies including ischemia and reperfusion.

One way to treat these detrimental pathologies is to supply antioxidant molecules or particles that renormalize the disturbed oxidative condition. We recently developed a biocompatible carbon nanoparticle, the poly(ethylene glycolated) hydrophilic carbon cluster (PEG-HCC), which has shown ability to scavenge oxyradicals and protect against oxyradical damage in rodent models and thus far has demonstrated no in vivo toxicity in laboratory rodents (24–27). The carbon cores of PEG-HCCs are ∼3 nm wide and range from 30 to 40 nm long. Based on these data, we estimate that there are 2,000–5,000 sp² carbon atoms on each PEG-HCC core. We have demonstrated the efficacy of PEG-HCCs for normalizing in vivo O₂^•− in models of traumatic brain injury with concomitant hypotension. Simultaneously, we observed normalization in NO^• levels in these experiments (26, 27). A better understanding of these materials is necessary to potentially translate these therapeutic findings to the clinic.

In the present work, we evaluated antioxidant properties of PEG-HCCs. Using spin-trap EPR spectroscopy, we demonstrate that PEG-HCCs scavenge O₂^•− with high efficiency. X-ray photoelectron spectroscopy (XPS) indicates that covalent addition of ROS to the PEG-HCCs is not responsible for the observed activity. Direct measurement of O₂^•− concentration using freeze-trap EPR demonstrates that PEG-HCCs behave as catalysts, and measurements made with a Clark oxygen electrode during the reaction reveal that the rate of production of O₂ is above that expected due to self-dismutation of O₂^•− in water. An equivalent amount of H₂O₂ is also simultaneously produced. Finally, selectivity for ROS is confirmed using a hemoglobin and a pyrogallol red assay; PEG-HCCs are unreactive to both NO^• and ONOO⁻. These results clarify the fundamental processes involved in the previously observed in vivo protection against oxygen damage (26, 27).

Results and Discussion

PEG-HCCs Scavenge ^•OH and O₂^•−.

The scavenging capacity of PEG-HCCs was evaluated by EPR with the spin trap 5-(diethoxyphosphoryl)-5-methylpyrrole-N-oxide (DEPMPO). DEPMPO produces relatively stable paramagnetic adducts upon reaction with ^•OH (3, 4, 28, 29) and O₂^•− (6, 29, 30) at room temperature (Fig. S1). We hypothesized that PEG-HCCs would scavenge O₂^•−, resulting in a decreased EPR signal of the spin-adduct. EPR amplitudes of the DEPMPO-OOH and DEPMPO-OH adducts were lowest in the presence of PEG-HCCs (Fig. 1 A and B, respectively). O₂^•− was generated from KO₂ in DMSO/crown ether (31) and the ^•OH radicals were generated in situ by the Fenton reaction (32). Simulations of the DEPMPO-OH and DEPMPO-OOH spectra confirmed the experimental identity of the spin-adducts (Fig. S1 and Table S1). Control experiments indicated that neither PEG nor PEG-HCCs destabilize the DEPMPO-OOH adduct. Note that the capture of ^•OH by PEG-HCCs was also demonstrated, but the mechanistic relevance is unclear because its generation from H₂O₂ and metal would likely be far from the PEG-HCC reaction centers. Hence, kinetic data on ^•OH are not highlighted here.

Fig. 1.

Effect of PBS, PEG, and PEG-HCCs on O₂^•− and ^•OH radicals. (A) EPR spectra obtained from the O₂^•− system or DEPMPO-OOH adduct at pH 7.4 and room temperature. Spectra were recorded after 70 s of the KO₂ addition. The PEG-HCCs spectrum was corrected by subtracting the signals of the PEG-HCCs alone. (B) EPR spectra obtained from the ^•OH system or DEPMPO-OH. Spectra were recorded after 90 s of the H₂O₂ addition. The adduct stability was followed for 30 min. No correction was necessary for the ^•OH scavenging experiments.

Further control experiments led to the detection of an intrinsic PEG-HCC radical possessing a symmetric spectrum centered at g = 2.0015 with an overall line width of 6 G and a half-saturation power at 0.74 mW at 115 K. We determined that for every 1 mol of PEG-HCCs there was 1 mol of unpaired electrons. This value was obtained based on five different batches of PEG-HCCs, including a batch that was 3 mo old, indicating that the nanoparticle-based radicals are very stable. Moreover, they are resistant to reductants such as ascorbate and dithionite and even inert to O₂ and NO^•. Varying the pH from 2 to 12 did not alter the EPR lineshape of the radical. In contrast, only a very small intrinsic radical was found for two known antioxidant fullerene derivatives, C₃ (33) and C₆₀-serinol (34), at comparable concentration. This apparent correlation implies that this intrinsic radical may be linked to the activity of neutralizing O₂^•− and warrants further study.

To confirm the O₂^•− specificity of our spin-trap assay and to estimate PEG-HCC activity, the experiments were repeated in the presence of SOD1 (Cu/Zn-SOD), an enzyme that decomposes O₂^•− to H₂O₂ and O₂ (35–37). As expected, dismutation of O₂^•− was manifested by a decrease in the intensity of the 8-hyperfine line pattern of the adduct (Fig. 2). Comparison of the spectra clearly shows that PEG-HCCs have a quenching effect comparable to or better than that of 10 U/mL of SOD1. Based on the EPR amplitude of the DEPMPO spin trap, we calculated the antioxidant trapping (6) of O₂^•− by PEG-HCCs as 98% (Table S2).

Fig. 2.

(A) Uncorrected and (B) corrected EPR spectra of the samples treated with SOD vs. PEG-HCCs. The dismutation of the O₂^•− radicals is being catalyzed by SOD1 causes a signal drop. (I) SOD1 0.01 U/mL. (II) SOD1 0.1 U/mL. (III) SOD1 1 U/mL. (IV) SOD1 10 U/mL. (V) PEG-HCCs (0.07 mg/mL or 170.1 nM, which is the same concentration used in Fig. 1A). The spectrum in B, V was corrected by subtracting the EPR signal of the starting PEG-HCCs.

Reaction of PEG-HCCs with O₂^•− Is Catalytic.

Two mechanisms were considered for the observed O₂^•− antioxidant activity: (i) radical annihilation owing to the covalent bond formation between the radical and the graphitic domains of the PEG-HCCs and/or (ii) transformation of O₂^•− to O₂ by the PEG-HCCs. Because XPS only detected a slight oxygen increase (<10%, Fig. S2) in the PEG-HCCs after KO₂ treatment, covalent oxygen addition to the PEG-HCCs cannot be the main mechanism.

To test the transformation hypothesis, we established a manual freeze-trap EPR steady-state kinetic assay for O₂^•− consumption using KO₂ to provide excess O₂^•− and therefore shift the rate-limiting step to the intrinsic capability of PEG-HCCs in turning over O₂^•−. This approach helped us to avoid the disadvantages of commonly used spin-trap EPR methods, which suffer from unfavorable trapping efficiencies and the loss of direct structural and kinetic information (17).

The typical EPR spectrum of 15-s freeze-trapped O₂^•− is characterized by the axial symmetry of its three principle g values (17, 38), and as Fig. 3 shows, the O₂^•− EPR signal decreased in the presence of PEG-HCCs. The rate of second-order self-dismutation of O₂^•− is very pH-sensitive and decreases exponentially from its pK_a (4.8) to pH 11 owing to increased charge repulsion between substrate molecules (39). Efforts to circumvent this at pH 8 by reducing [KO₂] to as low as 0.1 mM failed, because we could not freeze-trap any EPR-detectable radical. Therefore, to achieve sufficient and reliable concentrations of O₂^•− in solution, the quenching experiments are best carried out in 50 mM NaOH (pH 13).

Fig. 3.

Quenching of O₂^•− from KO₂ at varying pH. (A) EPR spectrum of O₂^•− in 50 mM NaOH in the presence and absence of PEG-HCCs. PEG-HCCs are able to quench O₂^•−. (B) Quenching of O₂^•− at varying pH values; 20 mM KO₂ in DMSO was dissolved in different 50 mM buffers.

As expected, the total spin concentration increased with the amount of added KO₂ (Fig. 4A). Consumption of O₂^•− was then calculated by subtracting these values from the control lacking PEG-HCCs and recalculated as turnover numbers (moles of consumed O₂^•− per moles of PEG-HCCs per second) with an average reaction time of 15 s (Fig. 4B). Our data showed saturation behavior of [O₂^•−]. The maximum O₂^•− turnover amounted to a dramatic ∼2.9 million (1.28 nM PEG-HCCs) and 1.1 million (6.4 nM PEG-HCCs) mol/15 s/mol, or 197,000 s⁻¹ and 73,000 s⁻¹, respectively. In a puzzling twist, lower concentration of PEG-HCCs showed higher turnover efficiency. This could suggest that at higher concentrations there is more nanoparticle aggregation, lowering the overall turnover number. Although further detailed explanation for the [PEG-HCCs] dependency is not possible at this point, the extraordinary capacity of PEG-HCCs in quenching O₂^•− is underscored.

Fig. 4.

KO₂ experiments in 50 mM NaOH. (A) O₂^•− total spin concentration based on double integration of obtained EPR spectra. For details, see Methods. (B) The turnover number (moles of superoxide per mole of PEG-HCCs per second) was calculated by subtracting the amount of remaining O₂^•− from the amount of O₂^•− in the control, dividing by the amount of PEG-HCCs and then by 15 s for the reaction time (the average time required for manually freeze-trapping each EPR sample).

Our experiments suggest that PEG-HCCs behave as catalysts because the molar ratio of O₂^•− consumed to PEG-HCCs is far beyond the number of active sites, presuming the active sites are C atoms involved in sp² bonding on the PEG-HCCs in quantities determined previously (24, 25). Identical quenching experiments carried out using C₃ and C₆₀-serinol showed little catalytic activity in quenching of O₂^•− (Fig. S3). Although the fullerenes have conjugated cores similar to those of the PEG-HCCs, there are clear structural differences. PEG-HCCs possess larger conjugated domains and bear a stoichiometric number of unpaired stable electrons, which could lower the activation energy for electron removal from O₂^•− to form O₂. Second, the C₆₀ fullerenes have a highly curved structure, destabilizing a radical that prefers a planar conformation, whereas the PEG-HCCs are primarily planar and not tubular because there is no remaining radial breathing mode in their Raman spectra (24). It should be noted that although C₃ has been previously reported to have catalytic activity in the turnover of O₂^•− (40), the overall rate-limiting step in those experiments was substrate availability. By contrast, in our study, materials were tested in the presence of excess substrate and show that the activity of <10 nM PEG-HCCs is several orders higher than that of 20 µM C₃. Whether the presence of the intrinsic radical in the PEG-HCCs bears on the activity differences found here is unclear.

PEG-HCCs Are SOD Mimetics.

Using a Clark-type oxygen electrode, we found that 4 nM PEG-HCCs substantially increased the rate of O₂ production but had no effect on the total amount of product (Fig. 5A). Subtracting the background O₂ formation rate (due to self-dismutation of O₂^•−) resulted in a rate solely owing to the activity of PEG-HCCs, which showed typical Michaelis–Menten saturation kinetics (Fig. 5B); the O₂^•− turnover by PEG-HCCs in this experiment was estimated at 22,000 s⁻¹. This value is in the same range as that obtained from the EPR experiments and is closer to the 73,000 s⁻¹ obtained with 6.4 nM PEG-HCCs. This >20,000 s⁻¹ catalytic turnover number is higher than most single-active-site enzymes, suggesting that a PEG-HCC possesses multiple active sites.

Fig. 5.

Simultaneous O₂ and H₂O₂ production in the presence and absence of PEG-HCCs. (A) PEG-HCCs (4 nM) enhance the turnover of O₂^•− to O₂ but the total amount of O₂ generated does not change. (B) PEG-HCCs (4 nM) exhibited Michaelis–Menten kinetics with apparent K_m = 750 µM and V_max = 88 µM/s. The estimated turnover of O₂^•− by PEG-HCCs is 22,000 s⁻¹. (C) The amount of H₂O₂ produced in the presence and absence of PEG-HCCs is identical. (D) The ratio of [O₂] to [H₂O₂] produced by PEG-HCCs is 1:1.

Having confirmed O₂ generation, our attention turned to H₂O₂, the other potential major product of O₂^•− transformation. We investigated the production of H₂O₂ at either 15 s or 30 min of O₂^•− transformation in the presence and absence of PEG-HCCs in 50 mM NaOH. O₂ production was first measured for each sample at pH 13, after which the pH was lowered to ∼8 by a strong buffer and HRP was added along with Amplex Red. The reaction of Amplex Red and H₂O₂ in the presence of HRP produces the highly colored resofurin, which can be measured spectrophotometrically. Comparison of H₂O₂ production in the presence and absence of PEG-HCCs revealed the total amount of H₂O₂ to be the same (Fig. 5C). We consistently obtained a 1:1 molar correlation between O₂ and H₂O₂ produced in the same sample (Fig. 5D).

We then assessed the formation and stoichiometry of OH⁻ by measuring the pH change caused by the addition of KO₂ or KOH to 20 mM Hepes (pH 7.2) in the presence and absence of PEG-HCCs. The pH increase caused by added KO₂ was very close to that of the KOH control, and the presence of PEG-HCCs had no effect (Fig. S4). This outcome indicates that both self-dismutation and turnover by PEG-HCCs follow the same mechanism leading to OH⁻ formation, and the stoichiometry between O₂^•− and OH⁻ is 1:1. Together, the O₂, H₂O₂, and OH⁻ results suggest that PEG-HCCs catalyze O₂^•− conversion as a dismutation process:

$${\mathrm{PEG-HCC}}^{•}+{\text{O}}_2^{{•}^{-}}\to {\mathrm{PEG-HCC}}^{-}+{\text{O}}_2$$ [1]

$${\mathrm{PEG-HCC}}^{-}+{\text{O}}_2^{{•}^{-}}+2\hspace{0.17em}{\text{H}}_2\text{O}\to {\mathrm{PEG-HCC}}^{•}+{\text{H}}_2{\text{O}}_2+2\hspace{0.17em}{\mathrm{HO}}^{-}.$$ [2]

Although there are product similarities to the model proposed by Ali et al. (40) for the C₆₀ derivative C₃, our proposed mechanism is quite different. PEG-HCC can initially act as an electron acceptor (Eq. 1) toward O₂^•− to form a highly delocalized, hence nonbasic, anion on the conjugated carbon framework, followed by acting as an electron donor to a second molecule of O₂^•− (Eq. 2) with rapid capture of two protons from water to complete the catalytic cycle. If the initial reaction is not dependent on the intrinsic radical domain of the PEG-HCC, then the first step affords PEG-HCC^•−, which returns to the neutral PEG-HCC in the second step. This latter motif permits the PEG-HCC to use multiple reaction centers, therefore better explaining the remarkably high turnover numbers that were recorded for conversion of O₂^•− to O₂. PEG-HCC anion stabilization can occur through the extended conjugated domain and through multiple neighboring carbonyl enolate-like interactions.

The similar pH shift observed with PEG-HCCs and the KOH control indicates that the abstraction of two protons from water to release OH⁻ (Eq. 2) proceeds to completion. Although it may seem that there is a shortage of protons at pH 13, there are in fact >50 M water molecules, an abundant source. Moreover, the water molecules that provide the two protons in Eq. 2 are likely to be hydrogen-bonded with the transiently formed O₂²⁻ to facilitate proton transfer. Indeed, electron spin echo envelope modulation studies carried out by Narayana et al. (41) on frozen O₂^•− samples prepared in 5 M NaOH found that each O₂^•− is hydrogen-bonded to four polarized water molecules ready for proton transfer. The extremely high pK_a values for the two deprotonation steps of H₂O₂ to O₂²⁻ (12.7 and 25, respectively) (42) provide further driving force for rapidly establishing a new equilibrium after proton transfer. Importantly, this suggests that the PEG-HCC–catalyzed O₂^•− turnover can occur by similar mechanisms whether at pH 13 (50 mM NaOH) or at physiological pH.

To support this prediction, we compared the activity of PEG-HCCs with SOD at physiological pH by two methods: (i) competition with cytochrome c in trapping O₂^•− released in situ from the reaction of ferrous endothelial nitric oxide synthase oxygenase domain (eNOS_ox) with oxygen under uncoupling conditions (43, 44) and (ii) direct rapid-freeze quench (RFQ) EPR of O₂^•− produced from KO₂.

Both Cu/Zn SOD and PEG-HCCs showed dose-dependent competition with cytochrome c, yielding EC₅₀ values of 2.5 nM and 254 nM, respectively (Fig. 6A). Although this seems to indicate that the catalytic turnover of O₂^•− by PEG-HCCs is two orders of magnitude less efficient, PEG-HCCs, but not SOD, also decreased the rate of cytochrome c reduction by approximately one order of magnitude (Fig. S5), indicating a complicated mechanism when using eNOS_ox to generate O₂^•−.

Fig. 6.

Comparison of O₂^•− quenching activity of SOD and PEG-HCCs at physiological pH. (A) EC₅₀ values of 2.5 nM for Cu/Zn SOD and of 254 nM for PEG-HCCs were determined using anaerobic stopped flow as a percentage of inhibition of reduction of cytochrome c³⁺. Concentration of eNOS_ox and cytochrome c was 10 µM and 15 µM, respectively. (B) KO₂ (2.2 mM) decayed to 0.52 ± 0.05, 0.39 ± 0.04, and 0.31 ± 0.02 mM upon self-dismutation, SOD-catalyzed dismutation, and PEG-HCCs–catalyzed dismutation, respectively; 20 nM each of SOD and PEG-HCCs was used. The error bars are SEM from four repeats. (C) The IC₅₀ of the PEG-HCCs, as determined by the SOD-inhibitable reduction of ferricytochrome c, was 0.20 ± 0.01 mg/mL or 486 ± 24 nM.

To avoid this problem, we used KO₂ to produce a continuous supply of O₂^•−. Direct presentation of O₂^•− as KO₂ in the RFQ EPR kinetic measurements (Fig. 6B) showed that after a 20-ms reaction 2.2 mM KO₂ decayed to 0.52 ± 0.05, 0.39 ± 0.04, and 0.31 ± 0.02 mM (± SEM, n = 4) upon self-dismutation, SOD-catalyzed dismutation, and PEG-HCC–catalyzed dismutation, respectively. We thus observed 0.13 mM and 0.21 mM extra O₂^•− consumption upon addition of SOD and PEG-HCCs, corresponding to turnover numbers of 0.65 × 10⁶ s⁻¹ and 1.05 × 10⁶ s⁻¹, respectively. On a molar basis, PEG-HCCs are as efficient at turning over O₂^•− as Cu/Zn SOD. Because the residual O₂^•− after 20-ms reaction was still at millimolar levels, steady-state kinetic requirements were met, and substrate provision was never a limiting factor. Using the results from the control sample, the second-order self-dismutation rate constant can be calculated as

$$\left(\frac{1}{0.52}-\frac{1}{2.2}\right)\times 9.25=k\times 0.02\text{s},$$

where 9.25 is the dilution factor after unequal mixing with buffer in our rapid-freeze apparatus. k is calculated to be 6.8 × 10⁵ M⁻¹⋅s⁻¹, matching the theoretical rate constant for self-dismutation at pH 7.7 (45), indicating that our RFQ system operation was indeed optimal. The turnover number obtained at pH 7 is much higher than that obtained by manual mixing at pH 13, and this could be due to a real pH-dependent rate-limiting step. However, the pH 13 experiment employs a linear time-dependence treatment for a 15-s reaction that might deviate substantially from the exponential kinetics anticipated from PEG-HCC catalysis and thus could lead to an underestimation of the activity.

We also evaluated the antioxidant activity of PEG-HCCs by exposing them to O₂^•− generated during the turnover of hypoxanthine–xanthine oxidase (HX/XO), a system in which the rate-limiting step is the release of O₂^•−, rather than in the antioxidant’s intrinsic turnover efficiency. Here also, PEG-HCCs behaved as effective antioxidants, achieving inhibition equivalent to half of the positive control (IC₅₀), which was measured in the presence of a large excess of SOD. The IC₅₀ of the PEG-HCCs was 0.20 ± 0.01 mg/mL or 486 ± 24 nM (Fig. 6C), higher than we observed for the eNOS_ox system. Given the excellent in vitro and in vivo efficacy of PEG-HCCs (26, 27), the reduced activity in the presence of cytochrome c does not seem to have a significant effect clinically.

PEG-HCCs Are Inert to NO^• and ONOO⁻.

Our previous in vitro endothelial culture and in vivo work on traumatic brain injury/hypotension models indicated no direct reaction between PEG-HCCs and NO^• (27). Here we studied the antioxidant activity of PEG-HCCs against NO^• using a hemoglobin assay to confirm the previous finding. The heme iron oxidation in oxyhemoglobin (HbO₂) by NO^• is a fast (∼10⁸ M⁻1⋅s⁻¹) (17), quantitative, and irreversible reaction in which metHb and nitrate ion are produced (46, 47). In general, NO^• can be indirectly determined by monitoring the production of metHb, estimated by the difference in absorbance at 401 and 411 nm with a difference extinction coefficient of 38 mM⁻¹⋅cm⁻¹ (20, 21). If PEG-HCCs and HbO₂ react with the NO^• radical at comparable rates, the PEG-HCCs would prevent the formation of metHb, resulting in a smaller absorbance difference. In the first experiments, HbO₂ and the PEG-HCCs were mixed and the reaction was initiated by addition of NO^•. The conversion of HbO₂ to metHb is shown in Fig. S6. PEG-HCCs had no effect on this conversion, indicating that neither the absorbance of HbO₂ nor metHb was affected by the presence of the PEG-HCCs at these concentrations. The difference between the control and the PEG-HCCs is only ∼5%, which is less than the experimental error of ∼10%. Hence, it can be concluded that either the PEG-HCCs are not quenching the NO^• radicals or that the reaction rate between the NO^• and the PEG-HCCs is much slower than the NO^• and HbO₂ reaction. To compare the rates of the reaction of PEG-HCCs with HbO₂ and PEG-HCCs with NO^•, sequential stopped flow experiments were carried out (48). Either buffer (control) or PEG-HCCs were incubated with NO^• in a 1:1 ratio for 20 ms, 1 s, or 1 min, and this solution was then mixed with HbO₂ in a 1:1 ratio. There was no interaction between the PEG-HCCs and NO^• even after 1 min of preincubation. Furthermore, NO^• did not decay or decompose under these conditions. These results corroborate the in vitro and in vivo studies of the PEG-HCCs where no NO^• reaction was observed.

Because NO^• reacts rapidly with O₂^•− to form ONOO⁻, the interaction between ONOO⁻ and PEG-HCCs was also studied using the ONOO⁻-induced quenching of the dye pyrogallol red (49, 50); if PEG-HCCs reacted with ONOO⁻, then quenching of the dye would be inhibited. Ascorbic acid, caffeic acid, and Trolox all inhibited the quenching reaction in a manner consistent with reported results (48), but PEG-HCCs had no effect (Fig. S7). Bearing in mind that NO^• is constantly produced in vivo and is freely diffusible, production of ONOO⁻ is more likely to occur in regions with a high local concentration of O₂^•−. As was previously demonstrated, PEG-HCCs efficiently scavenge O₂^•−, and this upstream scavenging effect will likely also decrease the amount of ONOO⁻ produced in vivo.

In summary, we have demonstrated that PEG-HCCs quench O₂^•− catalytically. Nanomolar concentrations of PEG-HCCs are able to rapidly scavenge micromolar to millimolar concentrations of O₂^•−. The turnover number of PEG-HCCs rivals that of Cu/Zn SOD at physiological pH. In addition, PEG-HCCs were inert to NO^•, thereby potentially improving vascular homeostasis (51). Mechanistic studies based on our new saturation steady-state kinetic measurements enabled accurate determination of the turnover of O₂^•− by the PEG-HCCs. The proposed mechanism, derived from EPR and oxygen electrode experiments at pH 13, also applies at physiological pH. Taken together, these results demonstrate the efficacy of a carbon nanoparticle-based SOD mimetic.

Methods

For materials, see Supporting Information.

Spin-Trap O₂^•− Scavenging Assay.

To a 2-mL microfuge tube was added potassium phosphate buffer (PBS, 50 mM, supplemented with 0.1 mM EDTA and 200 U/mL catalase); DEPMPO (20 mM); either PEG-HCCs (170 nM, Fig. S8 for MW estimation, or 0.07 mg/mL), PEG (14 µM or 0.07 mg/mL), or water; then KO₂ dissolved in DMSO (∼3.0 mM). KO₂/DMSO was prepared following a modified reported procedure (31). Briefly, a 0.1 M stock solution of KO₂ in DMSO/crown ether was prepared according to the following procedure: 71.1 mg of KO₂ and 600 mg of crown ether were dissolved in 10 mL of DMSO. The mixture was stirred for 45–60 min until a clear pale yellow solution was obtained. The stock solution, stored at −70 °C, was stable for several weeks with no apparent decomposition.

During EPR experiments, time 0 was the time at which KO₂ was added to the aqueous reaction mixture. Subsequently, the mixture was transferred to a capillary tube, sealed with Critoseal, and placed in the EPR instrument. Owing to the time required for sample transfer to the capillary tube and then to the EPR cavity, the earliest that the EPR spectrum could be recorded was at 70 s. EPR measurements were recorded in a Bruker EMX spectrometer using the following parameters: center field, 3,315 G; sweep width, 200 G; microwave frequency, 9.3 GHz; microwave power, 40 mW; receiver gain, 6.32 × 10⁵; modulation frequency, 100 kHz; modulation amplitude, 0.50 G; signal channel conversion, 163.8 ms; time constant, 327.7 ms; and sweep time, 167.8 s.

Direct EPR Detection of Radicals.

PEG-HCCs were frozen in 5-mm EPR tubes using liquid nitrogen. The EPR spectra of the intrinsic radicals and a CuSO₄ (1 mM) standard solution were recorded at 115 K using the following parameters: 3,310 G as center field, 2,000 G sweep width, microwave frequency 9.3 GHz, microwave power 1 mW, receiver gain 7.1 × 10³, modulation frequency 100 kHz, modulation amplitude 1.0 G. Because the copper standard and the PEG-HCC solutions were recorded under the same conditions, no corrections were necessary. The copper standard solution was supplemented with EDTA (10 mM) and NaClO₄ (100 mM) to make a homogeneous Cu-EDTA tetrahedral copper complex and uniform structure when frozen.

RFQ EPR Kinetic Measurements for O₂^•−.

Rapid-freeze quench EPR experiments were conducted using a System 1000 chemical/freeze quench apparatus (Update Instrument) with a Model 1019 syringe ram and a Model 715 controller. The ram velocity was 1.25 cm/s using a 0.008-inch nozzle to discharge the reaction solution. A low-temperature (125–130 K) isopentane bath was used for prechilling the assembly of a collecting funnel and EPR tube also filled with isopentane. Three syringes (two 2.1 mL, one 0.5 mL), two mixers, and two-push mixing syringe programs were used. In the control experiment, a 20 mM KO₂/DMSO solution was loaded into the 0.5-mL syringe and 0.1 M phosphate buffer, pH 7.7, was loaded in the 2.1-mL syringes. The first push mixed the solutions from the 2.1 mL syringes in mixer 1 while simultaneously filling the tubing to mixer 2. The second push mixed the buffer solution from mixer 1 with KO₂ from the 0.5-mL syringe in mixer 2. The final mixed sample was collected at 20 ms by rapid freeze quenching in prechilled isopentane. This mixing program achieves 9.25-fold dilution of KO₂ and < 9% DMSO in the final reaction mixture. In the experiments with SOD or PEG-HCCs, one of the 2.1-mL syringes was loaded with 20 nM SOD or PEG-HCCs in phosphate buffer. Reactions were conducted at room temperature (23 °C). Details of RFQ-EPR procedures have been described previously (52, 53). A packing factor of 0.45 was used for spin quantification of RFQ-EPR samples.

Steady-State Consumption of Superoxide.

PEG-HCCs (1.28 nM and 6.4 nM) were mixed with increasing amounts of KO₂ dissolved in DMSO/18-crown-6 (5, 10, 15, 20, 25, or 50 µL) in medium [either water, N-Tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid (TAPS), N-cyclohexyl-2-aminoethanesulfonic acid (CHES), N-cyclohexyl-3-aminopropanesulfonic acid (CAPS), or NaOH] for >15 s and then frozen in liquid N₂ to stop the reaction and preserve the remaining superoxide radical. EPR spectra were then recorded. To account for background dismutation of O₂^•−, a sample lacking PEG-HCCs was measured and its EPR subtracted from sample spectra to obtain the amount of KO₂ decay catalyzed (or neutralized) by the PEG-HCCs. For comparative study, the assay was repeated using C₃ or C₆₀-serinol instead of PEG-HCCs.

Steady-State Oxygen Formation.

Production of O₂ during the reaction of O₂^•− with PEG-HCCs was assayed polarographically at 24 °C with a YSI Model 5331 electrode (with a standard membrane) and a YSI Model 53 monitor. The reaction mixture (3.0 mL) contained 50 mM NaOH, 0.5 mM DETAPAC, and 4 nM PEG-HCCs. The reaction was started by addition of different amounts of KO₂. Activity of O₂ formation during the reaction was calculated from the initial slope.

Superoxide Turnover by Cytochrome c Reduction Using Burst Production of Superoxide from eNOS_ox.

Recombinant human eNOS_ox was expressed in Escherichia coli [tetrahydrobiopterin (BH₄) deficient] and purified as described previously (17, 44). Ferrous anaerobic eNOS_ox (10 µM) in 0.1 M phosphate buffer with 0.1 M NaCl and 10% (vol/vol) glycerol (pH 7.7) was prepared by anaerobic titration with a minimal amount of dithionite and then mixed with oxygenated buffer containing 15 µM of ferric cytochrome c alone (as a control) or with different amounts of PEG-HCCs or SOD in an anaerobic stopped flow (SX-18MV; Applied Photophysics) (17). The amount of superoxide released by eNOS_ox was determined by the cytochrome c reduction monitored at 550 nm (21 mM⁻¹⋅cm⁻¹) (54).

Cytochrome c Assay for O₂^•− from HX/XO.

The O₂^•− scavenging efficiency was determined using the method described by Quick et al. (55). The assay was performed in a 96-well plate with the following conditions (four replicates): (i) a water blank (325 μL per well), (ii) solution without PEG-HCCs, (iii) solution with PEG-HCCs, and (iv) solution containing SOD (90 μL; 10,200 U/mL in PBS). After the addition of the HX/XO, a BIO-TEK Powerwave XS spectrophotometer (Molecular Devices) was used to read the plate at 550 nm every 45 s for 8 min. The reaction rate (OD per minute) was determined and then plotted as a function of the cytochrome c concentration (ε_550nm = 21 mM⁻¹⋅cm⁻¹) against the log of the concentration of the test compound. IC₅₀ was determined using a nonlinear regression analysis (GraphPad Prism software, version 5.0).

Detection of H₂O₂ by Amplex Red Assay.

Five microliters of KO₂ from a 0.1 M stock solution (prepared as previously described) was added to 95 µL of 5 mM NaOH and incubated for 15 s or 30 min in the presence and absence PEG-HCCs (5 µL of 0.9 mg/mL PEG-HCCs in 100 µL of 5 mM NaOH). Two microliters of the reaction mixture was added to 1 mL of 100 mM TAPS buffer (pH 8.2) containing 40 µM Amplex Red and 10 U/mL HRP. The optical spectrum of resofurin, the product of the reaction of Amplex Red with hydrogen peroxide, was measured using a diode array UV-visible spectrophotometer (8453; Hewlett Packard). The concentration of H₂O₂ was calculated using an extinction coefficient of 58 mM⁻¹⋅cm⁻¹ at 570 nm.

pH Measurement of OH⁻ from KO₂.

The pH change caused by KO₂, or KOH as a control, in the presence or absence of 0.6 µM PEG-HCCs, was measured by a pH meter (VWR Scientific with Orion 910500 combination pH electrode) after the addition of 1, 2, or 3 mM KO₂ or KOH to 5 mL of 20 mM Hepes buffer (pH 7.2). The stock solution of KO₂ was prepared as in the spin-trap O₂^•− scavenging assay.