Critical Comparison of the Superoxide Dismutase-Like Activity of Carbon Anti-Oxidant Nanozymes by Direct Superoxide Consumption Kinetic Measurements

Abstract

The superoxide dismutase-like activity of poly(ethylene-glycolated) hydrophilic carbon clusters (PEG-HCCs), anthracite and bituminous graphene quantum dots (PEG-aGQDs and PEG-bGQDs, respectively), and two fullerene carbon nanozymes, tris malonyl-C₆₀ fullerene (C3) and polyhydroxylated-C₆₀ fullerene (C₆₀-OH_n), were compared using direct optical stopped-flow kinetic measurements, together with three native superoxide dismutases (SODs), CuZnSOD, MnSOD, and FeSOD at both pH 12.7 and 8.5. Computer modeling including both SOD catalytic steps and superoxide self-dismutation enabled the best choice of catalyst concentration with minimal contribution to the observed kinetic change from the substrate self-dismutation. Bi-exponential fitting to the kinetic data ranks the rate constant (M⁻¹ s⁻¹) in the order of: PEG-HCCs > CuZnSOD ≈ MnSOD ≈ PEG-aGQDs ≈ PEG-bGQDs > FeSOD >> C3 > C₆₀-OH_n at pH 12.7, and MnSOD > CuZnSOD ≈ PEG-HCCs > FeSOD > PEG-aGQDs ≈ PEG-bGQDs >> C3 ≈ C₆₀-OH_n at pH 8.5. Nonlinear regression of the kinetic model above yielded the same ranking as bi-exponential fit, but provided better mechanistic insight. The data obtained by freeze-quench EPR direct assay at pH 12.7 also yields the same ranking as stopped-flow data. This is a necessary assessment of a panel of proclaimed carbon nano SOD mimetics using the same two direct methods, revealing a dramatic, 3 – 4 order of magnitude difference in SOD activity between PEG-HCCs/PEG-GQDs from soluble fullerenes.

Graphical Abstract

Superoxide dismutase (SOD) plays fundamental roles in guarding against oxidative stress by transforming superoxide (SO) radical to neutral, less toxic hydrogen peroxide and molecular oxygen as shown in eq 1:¹

$$2{{\text{O}}_2}^{-}•+2{\text{H}}^{+}\to {\text{O}}_2+{\text{H}}_2{\text{O}}_2\left(k_{\text{s}},k_{-\text{s}}\right)$$ (1)

All three types of SODs are multimeric metalloenzymes (dimer to hexamer). The nickel SOD (NiSOD) is found in bacteria, and the other two types: iron- or manganese-containing SOD (FeSOD or MnSOD) and copper and zinc-containing SOD (CuZnSOD) are present in archaea, bacteria, and eukarya kingdoms.^1,2 In humans, MnSOD is exclusively located in the mitochondrial matrix, while CuZnSOD is primarily cytosolic.²

As an important defense against oxidative stress, defective or substrate-overwhelmed SOD leads to pathological conditions. For example, familial amyotrophic lateral sclerosis (FALS), which accounts for 10 – 15% cases of ALS, is linked to the mutations of CuZnSOD,³ and some mitochondrial dysfunction is due to lack of MnSOD.⁴ Conversely, many traumatic and pathological conditions generate high levels of SO that can overwhelm SOD activity and cause various deleterious health problems.⁵ Thus, there is great significance in developing effective antioxidants that have SO quenching activities like SOD. SODs, liposome-incorporated SODs, and metal complexes of SOD mimetics, especially the small manganese complexes have shown beneficial effects in reducing cell death, tissue injury and inflammation, and the pace of aging due to oxidative stress.^6–8 However, short shelf-life, high cost, and poor cellular permeation of SODs, and the metal-associated toxicity of the small metal complexes are constant concerns.⁸ The metal-free stable nitroxide radical, acting as a SOD mimetic via either a hydroxyl amine intermediate (reductive mode) or an oxoammonium cation intermediate (oxidative mode), shows good cellular penetration and minimal immunogenic and toxic responses, but suffers from too many side reactions, such as oxidation of semiquinone, reduction of metal ions, and the interruption of physiological radical chain reactions.⁸ In the last decade, several types of nano-antioxidants with enzyme-like activity acting as metal-free SOD mimetics have been developed from carbon-based nanoparticles including fullerene-like molecules,^9–15 carbon nanotubes (CNTs), and disintegrated single-walled CNTs, typified by hydrophilic carbon clusters (HCCs),^16,17 and most recently, graphene quantum dots.¹⁸ These materials can be classified as nanozymes, or nanoparticles with enzymatic properties.¹⁹ The HCCs are covalently solubilized with polyethylene glycol to form poly(ethylene-glycolated) hydrophilic carbon clusters (PEG-HCCs). While PEG-HCCs exhibit SO quenching activity that rivals CuZnSOD,¹⁷ it is unknown how the SO quenching activity of PEG-HCCs compares with other carbon nanoparticles including fullerene-based molecules and graphene quantum dots (GQDs), with characteristics consistent with carbon anti-oxidant nanozymes.^19,20 The key issue rests on the choice of assay method used to measure enzymatic activity.

CuZnSOD is a diffusion-limited, superefficient enzyme with a k_cat ~ 5 million s⁻¹;²¹ MnSOD and FeSOD are also very efficient, exhibiting k_cats of 25,000 – 40,000 s⁻¹.²² Different from other enzymes, whose specific activities are usually expressed by turnover numbers (or k_cat, s⁻¹), SOD activity is more commonly expressed as a 2^nd-order rate constant, with units of M⁻¹s⁻¹, in which the concentration component refers to [SOD], rather than [SO].² This reason for expressing SOD steady-state kinetics in this unorthodox way is because during the steady-state turnover of highly efficient SOD, shown by the two 2^nd-order reactions below (eq 2 and 3), it is the concentration of total SOD (reduced plus oxidized forms) but not that of the substrate, O₂⁻•, remains almost constant:^1,21–24

$${{\text{O}}_2}^{-}•+\text{SOD}/\text{nanozyme}\to {\text{O}}_2+{\text{SOD}}^{-}/{\text{nanozyme}}^{-}\left(k_{\text{r}},k_{-\text{r}}\right)$$ (2)

$${{\text{O}}_2}^{-}•+{\text{SOD}}^{-}/{\text{nanozyme}}^{-}+2{\text{H}}^{+}\to {\text{H}}_2{\text{O}}_2+\text{SOD}/\text{nanozyme}\left(k_{\text{o}},k_{-\text{o}}\right)$$ (3)

To possess similar high SO quenching activity as SOD, an SOD mimetic has to exhibit both high K_M, usually in mM range, and large V_max, as those of native SODs, and a redox potential falling between −0.18 and +0.91 V, the midpoint potential (E_m) values for oxidative and reductive SOD (or nanozyme) reactions (eq 2 and 3).

The biggest difficulty encountered in steady-state kinetic assays for SOD and its mimetics is the fast self-dismutation of SO in aqueous solution. Both O₂⁻• and HO₂ are unstable, showing a pH-dependent self-dismutation with a 2^nd-order rate constant peaking at pH 4.8, the pK_a of HO₂, as fast as 3 × 10⁷ M⁻¹s⁻¹. At pH higher than 4.8, the rate constant declines vs pH with a slope of −1, shown in eq 4:^25,26

$$\log k\left({\text{M}}^{-1}{\text{s}}^{-1}\right)=12.78-\text{pH}.$$ (4)

Due to the fast SO self-dismutation, the activities of SOD and SOD mimetics are often assayed using indirect coupled assay procedures consisting of a SO-generating reaction coupled to a colorimetric redox response.^23,27,28 Hypoxanthine/xanthine oxidase (HX/XO) is the most commonly used SO-generating system, and cytochrome c (cyt c) reduction, inhibition of either pyrogallol autoxidation, nitroblue tetrazolium reduction, epinephrine oxidation, or dihydroethidine oxidation by SO have been used as the coupled colorimetric redox responses. These coupled indirect assay methods are convenient and good for large-scale assays and automation, however, the amount of SO thus generated is low and the generation rate is usually slow compared to the efficiency of SOD or SOD mimetics and easily becomes the overall rate-limiting step in the assays. Moreover, the redox responses are usually nonspecific and subject to interference and side reactions.^23,27,28 Therefore, the kinetic data acquired by these indirect methods cannot give the intrinsic turnover rate and are often too complicated to resolve the underlying mechanism.

In contrast to the indirect coupled assay procedure above, direct assay methods enable real time measurements of SO disappearance through a temporally sufficient supply of SO for a steady-state assay. One such method is the in situ generation of SO from pulse radiolysis that allows for tracking the early stages of rapid SOD reactions. However, the amount of SO generated by pulse radiolysis is relatively low,^23,29,30 thus making it difficult to optically follow low levels of O₂⁻• in the reactions with many nanozymes, since many nanozymes absorb in the same range as SO (200 – 300 nm), and because SO has a low molar absorptivity constant in that range (< 2400 M⁻¹cm⁻¹).²⁵ On the other hand, electron paramagnetic resonance (EPR) can be used to monitor the EPR signal of SO, allows for a temporally sufficient supply of SO and is free of optical interference due to nanozymes. SO anion radical exhibits easily distinguishable EPR features of axial symmetry, and the EPR method has been used to successfully characterize the kinetics of [SO] in elucidating the reaction mechanism of alternative function as SO synthase of nitric oxide synthase.^31–33 Recently the steady-state kinetics of nanozyme PEG-HCCs and its simplified analog, perylene diimide,^17,34 were quantified using manual freeze or rapid-freeze quench combined with EPR (FQ-EPR or RFQ-EPR) to directly measure [O₂⁻•] as frozen samples at μM to mM concentrations.^31,32,35 However, FQ-EPR or RFQ-EPR requires a substantial amount of sample and is extremely time consuming in building a kinetic curve. Alternatively, stopped-flow kinetic measurements allow for mixing of sufficient SO with SO quenching catalysts to directly follow the UV absorbance of SO for its fast kinetic change,^22,36–38 and are therefore advantageous over FQ-EPR or RFQ-EPR.

In this study, we performed a comparative study of the kinetics of several carbon antioxidant nanozymes, including PEG-HCCs, poly(ethylene-glycolated) anthracite and bituminous graphene quantum dots (PEG-aGQDs and PEG-bGQDs, respectively), and two smaller water-soluble carbon nano-antioxidants, tris malonyl-C₆₀ fullerene (C3) and polyhydroxylated-C₆₀ fullerene (C₆₀-OH_n), in SO quenching using the stopped-flow method at two different pHs, and comparing to CuZnSOD, MnSOD, and FeSOD. Critical assessment was accompanied with computer modeling for the experimental data. These data are complemented with FQ-EPR data. We ranked the SO quenching activities of SODs and the carbon antioxidant nanozymes. The comparative study using direct assay methods revealed large difference in SO turnover of the various nanozymes.

RESULTS AND DISCUSSION

SO self-dismutation at different pHs

SO dismutates in aqueous environments, imposing a large obstacle for determining the kinetics of any catalyzed dismutation. The rate of SO self-dismutation has been shown to be pH-dependent in previous studies.²⁵ We first quantified such pH-dependent self-dismutation of SO using the stopped-flow method to establish the database for background corrections in the kinetic measurements in the presence of catalysts.

When SO in DMSO was mixed with aqueous buffers, its absorbance at 244 nm decayed with time and was followed at four different pH values, 7.5, 8.0, 8.5, and 12.7 (Figure 1). The 2^nd-order decay rate constants showed strong dependence on pH, exhibiting a linear dependence of log k on pH in eq 5,

$$\log k\left({\text{M}}^{-1}{\text{s}}^{-1}\right)=11.08-0.76\times \text{pH},$$ (5)

similar to that reported previously (eq 4).^25,26

Figure 1.

Time courses of SO self-dismutation at different pHs. The time courses of A₂₄₄ were followed after mixing 5 mM of KO₂ in DMSO with buffers at different pHs ([KO₂]_final = 192 μM): 12.7 (black), 8.5 (red), 8.0 (blue), and 7.5 (green). Data at times shorter than 5 ms are not shown due to contamination by mixing artifacts. Each time course is labelled with the 2^nd-order rate constants k in M⁻¹s⁻¹, obtained by fitting to eq 6 (see Experimental Methods section). Inset, log k vs pH and linear fit.³⁹

In Figure 1, in nearly neutral to slightly alkaline pH range (pH 7.5 to 8.5), SO dismutated rapidly; and the rate constants jumped 10 fold, from 3.3 × 10⁴ M⁻¹s⁻¹ to 3.4 × 10⁵ M⁻¹s⁻¹, with pH decrease of 1 unit, from 8.5 to pH 7.5. Conversely, SO self-dismutation slowed down more than 1000 times, k = 30 M⁻¹s⁻¹, when pH increased to 12.7. Therefore, the SO self-dismutation can essentially be ignored when the activities SODs or nanozymes were assayed at pH 12.7. In contrast, kinetic data of SODs and other catalysts acquired at physiological pH likely needs correction for background contributed by significant self-dismutation.

Activities of SODs and nanozymes at strong alkaline pH

In a previous study, the SO quenching activity of PEG-HCCs was analyzed at strong alkaline pH 12.7 using the FQ-EPR method.¹⁷ However, the FQ-EPR method is a steady-state assay based on a single reaction time point, making it difficult for obtaining whole kinetic curves of SODs and nanozymes. We therefore conducted stopped-flow assays for SODs and nanozymes at this pH (Figure 2) and compared their SO quenching activities (Table 1). Data analysis was rather straightforward due to the very small SO self-dismutation background (Figure 1). Since the time courses in the reactions of most SODs and nanozymes had barely any initial linear portions because of their significant SO consumption activities (see below), the data were analyzed by fitting to exponential decay function (eq 7, Experimental Methods section). The time courses of CuZnSOD and MnSOD generally exhibited monophasic exponential decays and those of nanozymes were typically biphasic. However, for convenient direct comparison of the rate constants of CuZnSOD and MnSOD with those of nanozymes, the time courses of CuZnSOD and MnSOD were also fit with the biphasic exponential function (eq 7), and the two rate constants (k₁’ and k₂’) of each SOD were very similar as shown in Table 1.

Figure 2.

SO quenching activities of SODs and nanozymes at pH 12.7. Black lines in each panel: time course of self dismutation of 192 μM KO₂. (A) Time courses of SO quenching by 20 nM SOD: CuZnSOD, MnSOD (blue), and FeSOD (green). (B) Time courses of SO quenching by nanozymes: 20 nM PEG-HCCs (red), 20 nM of PEG-aGQDs (green), 20 nM of PEG-bGQDs (blue), 2 μM C3 (yellow), and 30 μM C₆₀-OH_n (purple). The time courses with PEG-HCCs, PEG-aGQDs, and PEG-bGQDs were vertically adjusted to remove the background absorbance of these nanozymes. Due to the high concentrations of C3 and C₆₀-OH_n required in the reactions, their time courses showed large vertical backgrounds and were vertically shifted arbitrarily to bring the traces into y-axis scale range. Data at time shorter than 5 ms are not shown due to contamination by mixing artifacts.

Table 1.

Rate constants of SO dismutation by SODs and nanozymes at pH 12.7

Catalyst	biphasic ratea, s−1		[catalyst]	2nd-order rate constantb, M−1s−1		FQ-EPR kcat, s−1/KM, mM
	k1’	k2’		k1	k1
CuZnSOD	0.33 ± 0.005	0.26 ± 0.021	20 nM	1.7 (± 0.3) × 107	1.3 (± 0.1) × 107	4.4 × 104/2.7d
MnSOD	0.70 ± 0.003	0.68 ± 0.016	20 nM	3.5 (± 0.02) × 107	3.4 (± 0.08) × 107	1.8 × 105/10.2d
FeSODc	0.05 ± 0.005	0.043 ± 0.016	20 nM	2.5 (± 0.3) × 106	2.2 (± 0.8) × 106	N.D.e
PEG-HCCs	2.66 ± 0.11	1.16 ± 0.059	20 nM	1.3 (± 0.06) ×108	5.8 (± 0.3) × 107	2.0 × 105/0.75f
PEG-aGQDs	2.01 ± 0.130	0.10 ± 0.026	20 nM	1.0 (± 0.07) × 108	4.9 (± 1.3) × 106	2.9 × 105/6.5g
PEG-bGQDs	0.81 ± 0.063	0.36 ± 0.031	20 nM	4.1 (± 0.3) × 107	1.8 (± 0.2) × 107	1.0 × 105/32g
C3	0.057 ± 0.003	0.06 ± 0.005	2 μM	2.9 (± 0.2) × 104	3.0 (± 0.3) × 104	1.1 /11.5d
C60-OHn	0.046 ± 0.003	0.04 ± 0.004	30 μM	1.5 (± 0.1) × 103	1.3 (± 0.1) × 103	3.0/19d

^aexponential fit using eq 7

^bk = k’/[catalyst]

^cFeSOD did not show any significant SO consumption activity at this pH although its time course was fit well with eq 7

^dthis study; measured in the same way as in ref¹⁷ except in the freeze-trapping preparation of each EPR sample, SOD was diluted individually into each reaction mixture immediately before the reaction was started since both CuZnSOD and MnSOD gradually lost activity at pH 12.7; larger data scattering (n = 4) was thus observed in the SOD samples than in the nanozyme samples

^enot determined, FeSOD deactivated rapidly at this pH

^fref¹⁷

^gref¹⁸.

Although significantly away from their optimal pH 6 – 8, CuZnSOD and MnSOD still showed strong SO quenching activities at pH 12.7 (Figure 2A), and the two nearly identical rate constants for CuZnSOD and MnSOD were 1.7 × 10⁷/1.3 × 10⁷ M⁻¹s⁻¹ and 3.5 × 10⁷/3.4 × 10⁷ M⁻¹s⁻¹, respectively (Table 1). FeSOD, on the other hand, completely lost its SO quenching activity; actually some inhibition to the SO dismutation was observed in the presence of FeSOD, as SO A₂₄₄ decayed slower in the presence of FeSOD (Figure 2A). The reason for this phenomenon is unclear but likely a result of significant enzyme denaturation.

Our stopped-flow measurements demonstrated that PEG-HCCs had SO consumption activity stronger than those of SODs at pH 12.7 with rate constants 1.3 × 10⁸/5.8 × 10⁷ M⁻¹s⁻¹ (Table 1). These rate constants corresponded to k_cats of ~2.8 × 10⁵/1.2 × 10⁵ s⁻¹, with [KO₂] similar to those in our previous FQ-EPR study,¹⁷ quite similar to the value measured in that study (Table 1). Conversely, stopped-flow measurement revealed a small, ~2-fold difference, in PEG-HCCs’ biphasic rate constants, different from CuZnSOD or MnSOD.

The rate constant of the fast phase of PEG-aGQDs, 1.0 × 10⁸ M⁻¹s⁻¹ was the same as that of PEG-HCCs but its slow phase rate constant, 4.9 × 10⁶ M⁻¹s⁻¹ was 10 times slower than that of PEG-HCCs (Table 1). This significant kinetic difference suggests different SO consumption mechanisms in PEG-aGQDs and PEG-HCCs. Compared to those of PEG-aGQDs, PEG-bGQDs has a ~2.5× slower fast phase rate constant, 4.1 × 10⁷ M⁻¹s⁻¹, but a ~4 times faster slow phase rate constant, 1.8 × 10⁷ M⁻¹s⁻¹ (Table 1). The difference between PEG-aGQDs and PEG-bGQDs SO quenching kinetics highlighted the difference between the quantum dots made from different sources of coal. Overall, PEG-HCCs, PEG-aGQDs, and PEG-bGQDs showed significantly better SO quenching activities than CuZnSOD and MnSOD at strong alkaline pH.

Nanomaterials C3 and C₆₀-OH_n had only marginal SO quenching capability at pH 12.7 (Figure 2B). To assay their activities, concentrations of 100- and 1500-fold higher than those of SODs, PEG-HCCs or quantum dots, had to be used for C3 and C₆₀-OH_n, respectively (Figure 2B), obtaining rate constants of 2.9 × 10⁴/3.0 × 10⁴ M⁻¹s⁻¹ and 1.5 × 10³/1.3 × 10³ M⁻¹s⁻¹, respectively (Table 1). Overall C3 or C₆₀-OH_n have SO consumption activity 10³ – 10⁴ fold lower than those of SODs and other nanozymes (Table 1).

Although compared to FQ-EPR method, the stopped-flow method is significantly more efficient and less time and material consuming, it does suffer from the shortcomings that the workable [SO] range with optical detection was much lower than SODs’ or nanozymes’ K_Ms, varying from hundreds of μM to tens of mM. On the other hand, EPR has a much larger dynamic range in signal detection, and the workable [SO] is much closer or higher than K_Ms of nanozymes. Taking advantage of the slow self-dismutation of SO at high pH, we made efforts to measure the k_cats and K_Ms of CuZnSOD, MnSOD, C3, and C₆₀-OH_n using FQ-EPR method and present the measured parameters in Table 1 together with those of PEG-HCCs,¹⁸ PEG-aGQDs,¹⁸ and PEG-bGQDs,¹⁸ reported previously, to supplement our stopped-flow kinetic measurements. FeSOD inactivated rapidly at pH 12.7 and its activity could not be determined by this method. The FQ-EPR data indicated that PEG-aGQDs showed SO quenching efficiency similar to those of PEG-HCCs and MnSOD and better than those of PEG-bGQDs and CuZnSOD, corroborating the observations by stopped-flow (Table 1). FQ-EPR also showed that C3 and C₆₀-OH_n had very low SO quenching activities (Table 1). Among the nanozymes, PEG-HCCs had the smallest K_M, < 1 mM; K_Ms of other nanozymes were all significantly higher (Table 1).

Activities of SODs and nanozymes at pH 8.5

Although it was straightforward to determine the SO quenching activities of SODs and nanozymes at pH 12.7, this pH is far away from the physiological pH, where the actions of SODs and nanozymes are relevant in physiological functions. We therefore made efforts to analyze the SO quenching activities of SODs and nanozymes at a pH closer to that in physiological environments. However, measuring the kinetics of SODs and nanozymes at lower pH was much more challenging than at pH 12.7; the rapid SO self-dismutation imposed a huge background event (Figure 1) and had to be taken into account in experimental design and data analysis. To measure the activity of a SOD or a nanozyme, the time course of its SO quenching reaction must be reasonably separated from that of SO self-dismutation. We thus conducted stopped-flow measurements at pH 8.5, and through careful choice of the experimental conditions based on computer simulations (see below), we were able to achieve reasonable separation between the time courses of SO self-dismutation and its consumption by SODs or several nanozymes. At even lower pHs, such as 8.0 or 7.5, it became impractical to separate catalyzed SO consumption from the extremely fast SO self-dismutation (Figure 1).

Based on the SO dismutation rate at pH 8.5, 3.3 × 10⁴ M⁻¹s⁻¹ (Figure 1), we first conducted computer simulations to find optimal experimental conditions to achieve large separation between kinetics in the presence and absence of a catalyst. Computer simulations demonstrated that for the time course of a SO consumption reaction by a relatively weak SOD or nanozyme at 20 nM level, with a rate constant ≤ 1 × 10⁷ M⁻¹s⁻¹, to be clearly separated from that of SO self-dismutation, the [SO] needs to be lower than 10 μM (Figure 3A – B). However, the optical signal of SO is very small at this concentration, < 0.02. Although at even lower [SO], time courses of even less active catalysts are well separated from that of SO self-dismutation, the workable optical signal changes are too small (Figure 3B – E) and the time frames needed to follow the reactions are too long, >> 100 s (Figure 3C – E), to be practical for stopped-flow measurements. Another way of achieving time course separation is to increase the concentration of an SOD or a nanozyme (Figure 4). As shown in another computer simulation, when [catalyst] reacting with 100 μM SO reaches 100 nM, the time course of a relatively weak SOD or nanozyme, with a 1 × 10⁷ M⁻¹s⁻¹ rate constant, separates clearly from that of SO self-dismutation (Figure 4B). When [catalyst] is raised to 20 μM, the time course of a very weak catalyst, with a rate constant of 1 × 10⁵ M⁻¹s⁻¹, can be well separated from that of SO self-dismutation (Figure 4E). Based on these simulation assessments, we chose to measure the SO consumption activity of SODs and nanozymes with 100 μM SO, adjusting [catalyst] to achieve acceptable time course separation (Figure 5). However, SO stock was unstable during the experiments due to unclear reasons; decay up to ~30% was often observed (Figure 5B). This varying initial decay of SO stock could not be prevented by including dehydrating agent sodium sulfate and/or metal chelator diethylenetriaminepentaacetic acid (DETAPAC) (data not shown). Nonetheless, this limited fluctuation in [SO] did not affect measuring the rate constants as [SO] >> [catalyst] condition was always maintained except for C3 and C₆₀-OH_n. In the latter cases, this did not affect any our conclusions (see below).

Figure 3.

Computer simulation of SO dismutation with different [SO]. The time courses of SO optical signal were simulated using SCoP based on a kinetic model including simultaneously progressing irreversible self-dismutation (eq 1) and catalyzed dismutation (eq 2 and 3). The SO self-dismutation rate constant was set at measured value at pH 8.5, k_s = 3.3 × 10⁴ M⁻¹s⁻¹ (Figure 1); the two rate constants of a catalyzed SO dismutation were set identical k_r = k_o; [catalyst] was set at 20 nM; [SO] varied from 10 nM to 100 μM. The time course of self-dismutation is represented by black circles and those of catalyzed reactions by solid lines. Ranges of rate constants k_r/k_o varied from: 1 × 10⁵ M⁻¹s⁻¹, black; 1 × 10⁶ M⁻¹s⁻¹, red; 1 × 10⁷ M⁻¹s⁻¹, green;1 × 10⁸ M⁻¹s⁻¹, yellow; 1 × 10⁹ M⁻¹s⁻¹, blue. [O₂⁻•] is marked in each panel: 100 μM (A); 10 μM (B); 1 μM (C); 100 nM (D); 10 nM (E).

Figure 4.

Computer simulation of SO dismutation with different [catalyst]. The time courses of SO optical signal were simulated using SCoP based on a kinetic model including parallel irreversible self-dismutation (eq 1) and catalyzed dismutation (eq 2 and 3). The SO self-dismutation rate constant was set at measured value at pH 8.5, k_s = 3.3 × 10⁴ M⁻¹s⁻¹ (Figure 1); the rate constants of catalyzed SO dismutation were set at identical k_r = k_o; [SO] was set at 100 μM and [catalyst] varied from 20 nM to 20 μM. The time course of self-dismutation is represented by black circles and those of catalyzed reactions by solid lines. The [catalyst] used in simulations are marked in each panel: 20 nM (A) & (F); 100 nM (B); 500 nM (C); 2 μM (D); 20 μM (E).The varying rate constants used in each panel: 1 × 10⁵ M⁻¹s⁻¹, black; 1 × 10⁶ M⁻¹s⁻¹, red; 1 × 10⁷ M⁻¹s⁻¹, green;1 × 10⁸ M⁻¹s⁻¹, yellow; 1 × 10⁹ M⁻¹s⁻¹, blue. Panel F: biphasic exponential fits to the simulated time courses (same as in (A) represented with open circles) using the same color scheme as above. The fit rate constants are divided by [catalyst] = 20 nM to obtain the 2^nd-order rate constants: 3.7 × 10⁸/9.0 × 10⁷ M⁻¹s⁻¹ (black); 3.7 × 10⁸/9.0 × 10⁷ M⁻¹s⁻¹ (red); 3.8 × 10⁸/9.2 × 10⁷ M⁻¹s⁻¹ (green); 4.8 × 10⁸/1.4 × 10⁸ M⁻¹s⁻¹ (yellow); 2.2 × 10⁹/9.8 × 10⁸ M⁻¹s⁻¹ (blue).

Figure 5.

SO quenching activities of SODs and nanozymes at pH 8.5. Black circles in each panel: self dismutation of 192 μM KO₂. (A) SO consumption activities of SODs (circles): 40 nM CuZnSOD (red), 40 nM MnSOD (blue), and 200 nM FeSOD (green). (B) SO consumption activities of nanozymes (circles): 100 nM PEG-HCCs (red), 500 nM of PEG-aGQDs (green), 500 nM of PEG-bGQDs (blue), 13.5 μM C3 (yellow), and 30 μM C₆₀-OH_n (purple). The time courses with PEG-HCCs, PEG-aGQDs, and PEG-bGQDs were vertically adjusted to remove the background absorbance of these nanozymes. Due to the high concentrations of C3 and C₆₀-OH_n required in the reactions, their time courses showed large vertical backgrounds and were vertically shifted arbitrarily to bring the traces into y-axis scale range. Data at time shorter than 5 ms are not shown due to mixing artifacts. Black lines in both (A) and (B), fit by 2^nd-order function (eq 6) for SO self-dismutation; other lines are SCoP fits to the experimental data: (A) CuZnSOD (red), MnSOD (blue), and FeSOD (green), (B) PEG-HCCs (red), PEG-aGQDs (green), and PEG-bGQDs (blue).

The time courses of SODs were separated well from SO self-dismutation using 40 nM CuZnSOD or MnSOD and 200 nM FeSOD (Figure 5A). Similar to at pH 12.7, the time courses of CuZnSOD and MnSOD showed single exponential decays, although with much faster rates, a ~50-fold increase for CuZnSOD and a ~35-fold increase for MnSOD (Tables 1 and 2). For convenient comparison with nanozymes, the time courses of CuZnSOD and MnSOD were fit with biphasic exponential function (eq 7) and the rate constants obtained were 8.8 × 10⁸/6.3 × 10⁸ and 1.2 × 10⁹/1.2 × 10⁹ M⁻¹s⁻¹, respectively (Table 2). These values indicated these SOD reactions were diffusion-limited at pH 8.5, consistent with those reported before for these SODs although our data indicated that MnSOD was slightly more active than CuZnSOD at this pH, different from what were reported before. The activity of FeSOD was also much higher at pH 8.5 than at pH 12.7; its time course was also fit with two identical rate constants, 1.1 × 10⁸/1.1 × 10⁸ M⁻¹s⁻¹ (Table 2).

Table 2.

Rate constants of SO dismutation by SODs and nanozymes at pH 8.5.

catalyst	biphasic ratea, s−1		[catalyst]	2nd-order rate constantb, M−1s−1		RFQ-EPRc kcat, s−1
	k1’	k2’		k1	k2
CuZnSOD	35.3 ± 1.5	25.2 ± 1.8	40 nM	8.8 (± 0.4) × 108	6.3 (± 0.4) × 108	6.5 × 107
MnSOD	48.8 ± 2.0	48.3 ± 2.6	40 nM	1.2 (± 0.5) × 109	1.2 (± 0.07) × 109
FeSOD	22.0 ± 0.33	22.2 ± 0.4	200 nM	1.1 (± 0.1) × 108	1.1 (± 0.1) × 108
PEG-HCCs	30.9 ± 9.1	5.1 ± 0.4	50 nM	6.2 (± 1.8) ×108	1.0 (± 0.1) × 108	1.1 × 106
PEG-aGQDs	43.5 ± 2.8	17.5 ± 2.1	500 nM	9.0 (± 1.0) × 107	4.0 (± 1.0) × 107
PEG-bGQDs	43.7 ± 6.9	12.2 ± 1.5	500 nM	9.0 (± 2.0) × 107	3.0 (± 1.0) × 107
C3d	21.6 ± 3.7	3.0 ± 0.3	13.5 μM	1.6 (± 0.3) × 106	2.2 (± 0.2) × 105
C60-OHnd	28.1 ± 0.4	5.0 ± 0.05	30 μM	9.4 (± 0.7) × 105	1.7 (± 0.2) × 105

^aexponential fit using eq 7

^bk = k’/[catalyst]

^cmeasured at pH 7;¹⁷ calculated based on samples freeze-quenched at 20 ms

^dC3 and C₆₀-OH_n did not show any SO quenching activity at this pH although their time courses were fit well with eq 7; these values are only for comparison purpose.

To measure the activities of PEG-HCCs, PEG-aGQDs, and PEG-bGQDs, similar concentration to CuZnSOD and MnSOD was used for PEG-HCCs, but 10 times higher concentration of PEG-aGQDs and PEG-bGQDs were needed (Figure 5 and Table 2). In contrast to their stronger or similar activities compared to SODs at pH 12.7 (Table 1), PEG-HCCs, PEG-aGQDs, and PEG-bGQDs showed lower activities than CuZnSOD and MnSOD at pH 8.5 (Table 2), consistent with the results indicated by RFQ-EPR method at physiological pH.¹⁷ Moreover, compared to SODs, PEG-HCCs, PEG-aGQDs, and PEG-bGQDs exhibited significantly weaker activity pH-dependence; PEG-HCCs and PEG-bGQDs showed < 3.5-fold increase in SO consumption activities at pH 8.5 than at pH 12.7, while PEG-aGQDs showed an almost unchanged fast phase rate and a ~ 8-fold increase in slow phase rate (Tables 1 and 2). At pH 8.5, PEG-HCCs remained the strongest among these nanozymes, same as at pH 12.7 (Table 2), and PEG-aGQDs and PEG-bGQDs exhibited almost identical SO consumption activities at pH 8.5 (Table 2). The latter observation was different from that at pH 12.7, where PEG-aGQDs and PEG-bGQDs showed noticeably different SO consumption activities (Tables 1).

The two smaller size nanomaterials, C3 and C₆₀-OH_n, did not exhibit any SO quenching capability at pH 8.5, even at concentrations of 13.5 to 30 μM, respectively (Figure 5B and Table 2). In fact, at such high concentrations, C3 and C₆₀-OH_n appeared to inhibit SO dismutation, indicated by the slower decay rate of their reaction time courses compared to that of SO self-dismutation (Figure 5B). This is in contrast with their measurable, although low, SO consumption activities at pH 12.7 (Table 1). This pH-dependence of C3 and C₆₀-OH_n activities is different from those of PEG-HCCs, PEG-aGQDs, and PEG-bGQDs, suggesting different SO consumption mechanisms for the different types of nanozymes.

Although all the time courses could be fit satisfactorily with the standard biphasic exponential eq 7, such data analysis did not take the existing background of 2^nd-order SO self-dismutation into account, and the rate constants by fitting, k₁ and k₂ may not be equal to the true rate constants, k_r and k_o (eq 2 and 3). To examine how a parallel self-dismutation reaction may affect the values of rate constants obtained by biphasic exponential fitting, we simulated time courses of SO dismutation using SCoP, a general purpose simulation software package, based on a computer model including both SO self-dismutation (eq 1) and catalytic consumption by a SOD or nanozymes (eq 2 and 3), and then fit the simulated time courses with the biphasic exponential equation. The rate constants obtained by fitting to the SCoP-simulated time courses may be significantly different from the input values of the rate constants for simulation (Figure 4F). In fact, the rate constants obtained by fitting are generally larger than the input rate constants except for the fastest ones (Figure 4F). Moreover, the faster the rate constants are, the smaller is the discrepancy between the input and fitted values (Figure 4F). For example, the fitted values for the 2^nd-order rate constants are k₁/k₂ = 3.8 × 10⁸/9.2 × 10⁷ M⁻¹s⁻¹, respectively, for k_r/k_o = 1.0 × 10⁶ M⁻¹s⁻¹ used in simulation, a ~ 100 – 380 fold overestimate (Figure 4F), while for k_r/k_o = 1.0 × 10⁹ M⁻¹s⁻¹ in simulation, the fit values are 2.2 × 10⁹/9.8 × 10⁸ M⁻¹s⁻¹, only 2.2-fold overestimate for the fast phase and actually slight underestimate for the slow phase. If SO self-dismutation was not included in simulation (only eq 2 and 3), fitting to a simulated time course yielded rate constants equal to the input values (data not shown), highlighting how exponential fitting to experimental time courses may be misleading in obtaining the true values of the rate constants. Based on the indications from SCoP kinetic simulations, we analyzed the experimental data using non-linear regression function in SCoP software package against the computer model with the three parallel reactions, taking into account the observed substrate decay by both self-dismutation and catalysis of a SOD or nanozyme.

SCoP analysis of the experimental data at pH 8.5 revealed that the two redox reactions of each SOD (eq 2 and 3) had rather distinguishable rate constants, unlike the identical or very similar rate constants obtained by direct biphasic exponential fits (Table 3). The rate constants of PEG-HCCs, PEG-aGQDs, and PEG-bGQDs by SCoP analysis were also noticeably different from the corresponding biphasic exponential fits (Table 3). For each SOD, SCoP analysis revealed a faster rate constant for the SOD reduction reaction (eq 2, k_r) than that for the SOD⁻ oxidation reaction (eq 3, k_o) (Table 3). The largest difference between the two rate constants was those of CuZnSOD, ~ 15-fold. The opposite was observed for PEG-HCCs and PEG-bGQDs, SCoP analysis indicated that the oxidation of nanozyme was 2 – 3 times faster than their reduction rate constant. PEG-aGQDs had a ~ 2 times faster reduction rate constant than its oxidation rate constant. Despite the difference between the biphasic fit and SCoP analysis, both methods yielded the same conclusion that the SO quenching activity of PEG-HCCs, PEG-aGQDs, and PEG-bGQDs were weaker than those of SODs at pH 8.5, in contrast to the stronger activities of PEG-HCCs, PEG-aGQDs, and PEG-bGQDs than those of SODs at pH 12.7.

Table 3.

Rate constants of SO dismutation by SODs and nanozymes at pH 8.5 by computer modeling.

Catalyst	k1 a, M−1s−1	k2 b, M−1s−1
CuZnSOD	6.1 × 109	4.0 × 108
MnSOD	2.8 × 109	7.6 × 108
FeSOD	1.8 × 108	6.3 × 107
PEG-HCCs	5.8 × 107	1.4 × 109
PEG-aGQDs	1.5 × 108	3.0 × 107
PEG-bGQDs	4.4 × 107	1.2 × 108

^a2^nd-order constant for eq 2

^b2^nd-order constant for eq 3.

Much lower SOD-like activities of soluble fullerene derivatives, C3 and C₆₀-OH_n

Soluble fullerene derivatives have been proposed as efficient SOD mimetics.^{9,10,12–15,40–42} Detailed quantitative SOD-activity assays, both SO consumption by XO/HX and cyt c coupled assay, and production of O₂ (by Clark electrode) and H₂O₂ (by Amplex Red), were reported in the study by Ali et al.⁹ However, the concentration was 40 μM for cyt c⁺³ and 80 μM for C3, both in large excess to the substrate generation, which is a meager 1.1 μM/min from XO/HX system, conditions far from steady-state for a catalyst. Thus the estimated turnover number, 1025 s⁻¹ and K_M, 125 μM, could be sizably off. A similar problem was noticed that the SOD catalytic rate determined by pulse radiolysis is much larger than that using XO/HX system in generating SO substrate, as the former generates substrate much more efficiently.²⁹

Kahnt et al.¹² and Liu et al.¹⁰ evaluated the SOD-like activities of two water soluble dendrofullerenes and several monoadduct dendrofullerenes, respectively. The former study generated the substrate by pulse radiolysis and claimed an initial 10¹⁰ M⁻¹s⁻¹ rate upon addition of O₂⁻• to C₆₀ to form a C₆₀-O₂⁻• intermediate(s) for SOD-like catalysis, but this method is concerning for the formation of several other transient radical species generated by the radiolysis and the rapidscan data did not provide conclusive structure information to confirm any possible intermediate(s) or product(s),¹² prohibiting a definitive assessment of their SOD-like activity. The study by Liu et al.¹⁰ led to the conclusion that few of the monosubstituted dedrofullerenes showed SOD-like activities approaching the levels of MnSOD and FeSOD. They used equal-mixing stopped-flow with dentritic C₆₀ and KO₂ substrate both in DMSO containing 0.06% water. The critical “0.06%” water, or 33 mM, was claimed to be crucial in providing sufficient water to solvate SO anion, the valid form of SO for self-dismutation or reaction with SOD. However, KO₂ in DMSO/0.06% water is fairly stable and showed a 270 nm absorption peak, in contrast to the 245 nm of O₂⁻• in aqueous solution.²⁵ This substrate stock solution has no observable EPR signal as we routinely observed for the SO anion radical, even with water content increased to > 1% in the DMSO (data not shown). Furthermore, electron spin-echo envelop modulation studies concluded that each SO anion needs 4 water molecules⁴³ and each K⁺ cation needs 7 water molecules in the first layer of solvation.⁴⁴ Full solvation of both O₂⁻• and K⁺ from KO₂ requires more water in the secondary shell and even the bulk water to stabilize the solvated ions. It is clear that 0.06% water in DMSO cannot fulfill the claimed 2 mM O₂⁻• production that could be detected by either optical or EPR. It thus calls to question the SOD-like activity determined by Liu et al.¹⁰ We demonstrated in this study that water soluble fullerenes C3 and C₆₀-OH_n showed very poor activity compared with SODs, PEG-HCCs, and PEG-GQDs by 3 – 4 orders at two pH settings (k₁ and k₂ in Tables 1 and 2). This large difference was only revealed using direct assay methods, stopped-flow and FQ-EPR, in our comparative study.

What is the best method to use at physiological pH?

When the SOD assay was done at neutral or acidic pH, the self-dismutation reaction becomes so fast, as illustrated in Figure 1, that except for CuZnSOD and MnSOD, the other catalysts will suffer from a complicated kinetic data analysis due to the substantial contribution from the self-dismutation of SO. Even at pH 8.5, we had to use computer simulations to locate the optimal substrate and catalyst concentrations to achieve good rate separation between the self-dismutation and the catalyzed reactions to enable the measurements for all the samples using stopped-flow. If the unknown samples show activity near that of CuZnSOD or MnSOD, measurements at physiological pH can be achieved without significant contribution from background self-dismutation, but for samples that show much weaker activities including FeSOD, GQDs, and fullerenes, then raising the pH to slow the self-dismutation and using computer modeling to help data analysis is unavoidable. Nonlinear regression of the kinetic data to a specific mechanistic model including the substrate self-dismutation should provide the best mechanistic insights into the data measured near physiological pH.

The commonly used coupled indirect assays always have substrate generation as the rate-limiting step and cannot give the true intrinsic turnover numbers of the catalyst, especially on the ones that are very active. Even though these methods can be useful for convenient high-throughput screening, they can only provide a rough estimate on a comparative basis.

CONCLUSIONS

At pH 12.7, simple direct fit to the experimental data readily ranked the nanozymes compared to SODs. Our kinetic data ranks the rate constant (M⁻¹s⁻¹) in the order of: PEG-HCCs > CuZnSOD ≈ MnSOD ≈ PEG-aGQDs ≈ PEG-bGQDs > FeSOD >> C3 > C₆₀-OH_n at pH 12.7. The ranking of the data at pH 12.7 is corroborated by our FQ-EPR data, from which both V_max and K_M steady-state kinetic parameters were determined.

At near physiological pH 8.5, by carefully selecting the experimental conditions, the separation between time courses of SO self-dismutation and quenching reactions by SODs and nanozymes near physiological pH was achieved. SCoP computer modeling further revealed the kinetic difference between the 1^st and the 2^nd chemical steps of catalytic turnover in both metal-based SODs and carbon-based nanozymes.

At pH 8.5, nanozymes PEG-HCCs, PEG-aGQDs, and PEG-bGQDs have strong SO consumption activities, though slightly weaker than CuZnSOD or MnSOD. The overall ranking is MnSOD > CuZnSOD ≈ PEG-HCCs > FeSOD > PEG-aGQDs ≈ PEG-bGQDs >> C3 ≈ C₆₀-OH_n. Among these nanozymes PEG-HCCs have the highest activity while the activities of PEG-aGQDs and PEG-bGQDs are quite similar. The starting materials for preparing quantum dots seem to affect their SO quenching capability. C3 and C₆₀-OH_n show little SO quenching activity at pH 8.5 and weak activity at pH 12.7. These behaviors are quite different from those of SODs, PEG-HCCs, and GQDs. The vast separation of SOD activity between SODs, PEG-HCCs, GQDs, and fullerene-derivatives calls for future careful mechanistic investigation into these several categories of carbon antioxidant nanozymes to gain insights into the structure-function relationships.

EXPERIMENTAL METHODS

Materials

Potassium SO (KO₂), crown ether 18-Crown-6, anhydrous DMSO (99.7% purity, bovine CuZnSOD, MnSOD from Escherichia coli, and FeSOD from Escherichia coli were from Sigma-Aldrich (St. Louis, MO). Nano-antioxidants, including PEG-HCCs, PEG-aGQDs, PEG-bGQDs,¹⁸ C3 was provided and purified by the laboratories of Professors J. Tour and L. Wilson at Rice University (Houston, TX). C₆₀-OH_n was obtained from Millipore-Sigma.¹⁴ Molar concentrations of C3 and C₆₀-OH_n were calculated using M.W. of 1026 and 1100 (n = 22, based on mostly used in different studies), respectively; those for PEG-HCCs and GQDs were determined using the averaged molecular weight estimated from the specific fraction isolated by size-exclusion columns according to that detailed in previous publications.^17,18 PEG-HCCs and GQDs were assumed to have one active site per molecule, without considering the multiple-site model or the difference in surface area of the nanoparticles, to simplify activity comparison with SODs and fullerene-derivatives.

A stock solution of 100 mM KO₂ was freshly prepared as follows: 60 mg crown ether was fully dissolved in 1 ml anhydrous DMSO containing 5 – 10 pieces of ~1 mm silica gel desiccant by stirring for 15 min. KO₂ powder (7.1 mg) was then added to the DMSO/crown ether solvent and the mixture was stirred for another 20 – 30 min until KO₂ was totally dissolved.

Stopped-flow Measurements

The kinetic measurements were conducted with an Applied Photophysics DX-17MV stopped-flow machine (Leatherhead, UK) and the time courses of A₂₄₄ were followed. The SO dismutation reactions in the absence or presence of SODs or nanozymes were performed in unequal mixing mode to minimize the final [DMSO] in the reaction mixtures; SO in DMSO/crown ether with a desired concentration was placed in a 100 μl syringe and mixed with aqueous buffers, with or without SODs or nanozymes, in a 2.5 ml syringe. Reactions at pH 7.5, 8.0, and 8.5 were buffered with 50 mM KPi and reactions at pH 12.7 were conducted in 50 mM NaOH.

The data analysis was conducted using the program coming with the stopped-flow instrument. For SO self-dismutation (eq 1), the data were analyzed using a 2^nd-order rate function eq 6, ²⁵

$$A=\left\{A_0/\left(A_0\times k\times t+1\right)\right\}+A_{\text{f}},$$ (6)

where A is absorbance at 244 nm, A₀ the initial absorbance, and A_f the final absorbance. The 2^nd-order rate constant, k, was in units of O.D.⁻¹s⁻¹ and was converted to units of M⁻¹s⁻¹ by multiplying the value with the molar absorbance coefficient of SO.²⁵

For the reactions of SODs or nanozymes (eq 2 and 3), the time courses were analyzed using a bi-exponential decay function eq 7,

$$A=A_{\text{f}}+A_1{\text{e}}^{-k1\times t}+A_2{\text{e}}^{-k2\times t},$$ (7)

where k₁ and k₂ are the rate constants of the fast and slow phases, respectively, in units of s⁻¹. A₁ and A₂ are the amplitudes of the fast and slow phases, respectively. The rate constants were then divided by [SOD] or [nanozymes] to obtain the 2^nd-order rate constants.

The data in time shorter than 5 ms were typically contaminated with mixing artifacts in the unequal mixing mode and were excluded from data analysis.

All the nanozymes absorb stronger at 244 nm than native SODs, and their absorptions do not change during the reactions with superoxide, verified by the unchanged absorbance of the nanozymes during their reactions with stoichiometric levels of superoxide (data not shown). For reactions of superoxide with PEG-HCCs or GQDs, superoxide was totally consumed and the residual absorbance was attributed to the nanozyme (the H₂O₂ product had very low absorption at this wavelength). The same value of absorbance was then subtracted from the time courses for easier comparison with those of SODs.

Computer Modeling

SCoP (Simulation Resources Inc., Redlands, CA) was used to generate simulated data based on specified kinetic blocks, and also fit to the time courses of SO decay experimental data by non-linear regression. The computer model involved three parallel reactions, including SO self-dismutation (eq 1, rates k_s/k_−s) and catalyzed SO dismutation by SODs or nanozymes (eq 2, rates k_r/k_−r and eq 3, rates k_o/k_−o). The rate constant of SO self-dismutation⁴¹ was based on the 2^nd-order fit (eq 6) to the experimental data and was kept fixed in the computer modeling; the rate constants of SODs or nanozymes (k_r and k_o), in units of M⁻¹s⁻¹, were allowed to float. Moreover, all reactions were set irreversible (k_−s, k_−r, and k_−o = 0). In the reactions of a nanozyme, a parameter for extinction coefficient was also included to account for the extra optical absorbance by the nanozyme. Such extinction coefficient parameters were allowed to float since the information of accurate nanozyme concentrations was not available as the MW of each nanozyme distributes over a range. The SCoP fits were iterated until the error function reached minimum; the value of error function was calculated after each round based on the least squared errors to the experimental data.