Effect of Redox Potential on Diiron-Mediated Disproportionation of Hydrogen Peroxide

Heme and nonheme dimanganese catalases are widely distributed in living organisms to participate in antioxidant defenses that protect biological systems from oxidative stress. The key step in these processes is the disproportionation of H2O2 to O2 and water, which can be interpreted via two different mechanisms, namely via the formation of high-valent oxoiron(IV) and peroxodimanganese(III) or diiron(III) intermediates. In order to better understand the mechanism of this important process, we have chosen such synthetic model compounds that can be used to map the nature of the catalytically active species and the factors influencing their activities. Our previously reported μ-1,2-peroxo-diiron(III)-containing biomimics are good candidates, as both proposed reactive intermediates (FeIVO and FeIII2(μ-O2)) can be derived from them. Based on this, we have investigated and compared five heterobidentate-ligand-containing model systems including the previously reported and fully characterized [FeII(L1−4)3]2+ (L1 = 2-(2′-pyridyl)-1H-benzimidazole, L2 = 2-(2′-pyridyl)-N-methyl-benzimidazole, L3 = 2-(4-thiazolyl)-1H-benzimidazole and L4 = 2-(4′-methyl-2′-pyridyl)-1H-benzimidazole) and the novel [FeII(L5)3]2+ (L5 = 2-(1H-1,2,4-triazol-3-yl)-pyridine) precursor complexes with their spectroscopically characterized μ-1,2-peroxo-diiron(III) intermediates. Based on the reaction kinetic measurements and previous computational studies, it can be said that the disproportionation reaction of H2O2 can be interpreted through the formation of an electrophilic oxoiron(IV) intermediate that can be derived from the homolysis of the O–O bond of the forming μ-1,2-peroxo-diiron(III) complexes. We also found that the disproportionation rate of the H2O2 shows a linear correlation with the FeIII/FeII redox potential (in the range of 804 mV-1039 mV vs. SCE) of the catalysts controlled by the modification of the ligand environment. Furthermore, it is important to note that the two most active catalysts with L3 and L5 ligands have a high-spin electronic configuration.


Introduction
Iron and manganese are the most important candidates for both oxygen activation and environmental sustainability. Both metal ions have a variety of oxidation states that can be easily tuned in terms of Lewis acidity and hardness/softness, which can lead to fine-tuning of redox systems. Catalases (including hydrogen peroxide oxidoreductase) are important antioxidants necessary for life in oxygen-metabolizing cells to regulate H 2 O 2 concentration by accelerating the disproportionation of a toxic oxygen metabolite, hydrogen peroxide, forming dioxygen and water [1][2][3][4][5]. H 2 O 2 as reactive oxygen species (ROS) can easily be formed enzymatically by oxidases and non-enzymatically as a byproduct of respiration or autoxidation of cellular components. Most physio-pathological situations such as neurodegenerative diseases (Parkinson's disease and Alzheimer's disease), chronic inflammatory diseases, chronic obstructive pulmonary diseases (COPD), chronic kidney diseases (CKD), metabolic diseases (diabetes) or cancers can be associated with oxidative stress resulting from H 2 O 2 [6][7][8][9][10][11][12][13][14]. Hydrogen peroxide exerts its toxic effects in several Harmful effects of H2O2 on biological processes such as aging, cancer and neurodegenerative diseases have been confirmed by numerous studies. These findings led to the development of synthetic catalase mimetics as therapeutic agents that may be effective against oxidative stress in human diseases such as cancer, Alzheimer's disease, aging and inflammatory and heart diseases. Among them, manganese-based mimics are the most widely investigated, mainly because of their low toxicity (Mn is not prone to generate HO• radical in Fenton type reactions); however, there are several iron-and copper-based systems too. A number of mechanistic and spectroscopic studies have been reported on the catalase activity of non-porphyrin mono-and dinuclear complexes with various ligands including Schiff bases, corroles, polyamines and polypyridyl derivatives. These papers mainly focused on the comparison of mono and dinuclear copper, manganese and iron complexes and the effect of various factors such as ligand donor sites, endogenous acid/base groups and steric parameters [28][29][30][31][32][33][34][35][36][37].

Scheme 2.
Structure of the ligands and their complexes involved in this study.

Results and Discussions
The UV-Vis spectra of complexes 1, 2 and 4 in acetonitrile showed two major features, a band around 380-550 nm and a stronger band around 270-370 nm (Figure 1a). These bands are typical for pyridine-based ligands containing low-spin iron(II) complexes and can be assigned to metal-to-ligand charge transfer (MLCT) and ligand-centered π to π* transitions, respectively [39,[48][49][50][51]. In contrast to the low-spin complexes 1, 2 and 4, in solution, 3 and 5 show no absorption in the visible region, which is consistent with the latter complexes' high-spin electronic configuration [46]. A general bathochromic effect was observed as the number of aromatic rings increased in the ligand framework (see L 1 , L 2 and L 4 compared with L 5 ). This phenomenon can be explained by an important delocalization of π electrons promoted by the linear annelation. Spectroscopic and redox data for complexes 1 to 5 are summarized in Table 1.  [40]. 3 [46].

Results and Discussions
The UV-Vis spectra of complexes 1, 2 and 4 in acetonitrile showed two major features, a band around 380-550 nm and a stronger band around 270-370 nm (Figure 1a). These bands are typical for pyridine-based ligands containing low-spin iron(II) complexes and can be assigned to metal-to-ligand charge transfer (MLCT) and ligand-centered π to π* transitions, respectively [39,[48][49][50][51]. In contrast to the low-spin complexes 1, 2 and 4, in solution, 3 and 5 show no absorption in the visible region, which is consistent with the latter complexes' high-spin electronic configuration [46]. A general bathochromic effect was observed as the number of aromatic rings increased in the ligand framework (see L 1 , L 2 and L 4 compared with L 5 ). This phenomenon can be explained by an important delocalization of π electrons promoted by the linear annelation. Spectroscopic and redox data for complexes 1 to 5 are summarized in Table 1. of Fe III d-orbitals and making oxidation less facile. Almost the same effect was observed in the case of the N-methylated benzimidazole arm of 2 (55 mV). A linear correlation was found between the energy of the metal-to-ligand charge transfer (MLCT) band and the half-wave potentials, indicating that the substitutions performed on the ligand greatly influenced the electron density of the metal center and thereby its redox properties ( Figure  2a). These data can also serve as a basis to investigate possible correlations between the half-wave potentials of the catalysts and their catalytic activities.
(a) (b) To date, around 40 peroxo-diiron(III) complexes are known as possible biomimics of diiron enzymes. Similarly to our previously reported [Fe III 2(µ-O2)(L 1−3 )4(CH3CN)] 4+ (1 P -3 P ) species, complex 4 P and 5 P can be generated in acetonitrile with H2O2. These species, similarly to 1 P -3 P (λmax = 720 (ε = 1360 M −1 cm −1 ); λmax = 685 (ε = 1400 M −1 cm −1 ); and λmax = 705 (ε = 1200 M −1 cm −1 ), respectively) [39,46], show characteristic absorption bands at λmax = 710 (ε = 1360 M −1 cm −1 ) and 695 nm (ε = 1200 M −1 cm −1 ), which can be assigned to the charge transfer between Fe III and the O2 2− ligand ( Figure 2B) [42]. The half-lives (t1/2) for 1 P and 4 P generated by four equivalents of H2O2 at 298 K are 390 s and 500 s, respectively, demonstrating that 4 P is more stable than 1 P . A similar trend can be observed based on their self-decay data (ksd = 1.955 × 10 −3 s −1 for 1 P and 1.177 × 10 −3 s −1 for 4 P ) under identical conditions at 298 K (Figure 3a,b) [40]. The most reactive species is 5 P ; however, we could not determine its t1/2 and ksd data because its steady-state concentration in the presence of four equivalents of H2O2 is below 10% at 298 K.   Figure 1b). The Fe III /Fe II redox couples for 3 and 5 are at considerably higher potentials than for 2 and 4. The electrochemical data confirm that in compounds 3 and 5 the iron centers are stronger Lewis acids than those in 2 and 4, which can be explained with the electron withdrawing nature of the thiazolyl and triazolyl groups compared with pyridyl and benzimidazole, respectively. Introducing an electron-donating methyl substituent at the 5-position of the pyridine rings (4) shifted the half-way potential to about 56.5 mV more negative than that of 1, strengthening the electron density of Fe III d-orbitals and making oxidation less facile. Almost the same effect was observed in the case of the N-methylated benzimidazole arm of 2 (55 mV). A linear correlation was found between the energy of the metal-to-ligand charge transfer (MLCT) band and the half-wave potentials, indicating that the substitutions performed on the ligand greatly influenced the electron density of the metal center and thereby its redox properties ( Figure 2a). These data can also serve as a basis to investigate possible correlations between the half-wave potentials of the catalysts and their catalytic activities.
To date, around 40 peroxo-diiron(III) complexes are known as possible biomimics of diiron enzymes. Similarly to our previously reported [Fe III 2 (µ-O 2 )(L 1−3 ) 4 (CH 3 CN)] 4+ (1 P -3 P ) species, complex 4 P and 5 P can be generated in acetonitrile with H 2 O 2 . These species, similarly to 1 P -3 P (λ max = 720 (ε = 1360 M −1 cm −1 ); λ max = 685 (ε = 1400 M −1 cm −1 ); and λ max = 705 (ε = 1200 M −1 cm −1 ), respectively) [39,46], show characteristic absorption bands at λ max = 710 (ε = 1360 M −1 cm −1 ) and 695 nm (ε = 1200 M −1 cm −1 ), which can be assigned to the charge transfer between Fe III and the O 2 2− ligand ( Figure 2B) [42]. The half-lives (t 1/2 ) for 1 P and 4 P generated by four equivalents of H 2 O 2 at 298 K are 390 s and 500 s, respectively, demonstrating that 4 P is more stable than 1 P . A similar trend can be observed based on their self-decay data (k sd = 1.955 × 10 −3 s −1 for 1 P and 1.177 × 10 It was also established that by adding additional H2O2 (four equivalents), 1 P and 4 P can be regenerated 2-3 times after decomposition (Figure 4a,b) with a relatively high yield (>50%). A similar result was observed in the case of 5 P but with much higher H2O2 ( Figure  5a) and a much lower steady-state concentration of 5 P . Many fewer cycles and yields can be achieved with 3 P (Figure 5b). It was also established that by adding additional H2O2 (four equivalents), 1 P and 4 P can be regenerated 2-3 times after decomposition (Figure 4a,b) with a relatively high yield (>50%). A similar result was observed in the case of 5 P but with much higher H2O2 ( Figure  5a) and a much lower steady-state concentration of 5 P . Many fewer cycles and yields can be achieved with 3 P (Figure 5b). It was also established that by adding additional H 2 O 2 (four equivalents), 1 P and 4 P can be regenerated 2-3 times after decomposition (Figure 4a,b) with a relatively high yield (>50%). A similar result was observed in the case of 5 P but with much higher H 2 O 2 ( Figure 5a) and a much lower steady-state concentration of 5 P . Many fewer cycles and yields can be achieved with 3 P (Figure 5b).
To compare the reactivity of catalysts 1-5 with H 2 O 2 , the reactions were carried out under the same conditions (  [45]. In contrast, the lowest activity was observed in the case of complex 4 with Yield = 6.8%, TON = 20.4 and TOF = 27 h −1 under identical conditions. Much more favorable values were observed in all cases at lower substrate (H 2 O 2 ) concentrations, which can be explained by the water content (Figure 7a,b). As we noticed before (Figures 4 and 5), the amount of the peroxo complexes decreases significantly with the increase in water, which suggests that the 1 P -5 P intermediates play a key role in the catalytic cycles.  To compare the reactivity of catalysts 1-5 with H2O2, the reactions were carried out under the same conditions ( [1][2][3][4][5]: [H2O2] = 1: 300 at 298 K), and the amount of dioxygen evolved during the disproportionation of H2O2 into H2O and O2 was monitored using a gas volumetric method ( Figure 6). Under this condition, the highest catalytic activity can be observed by the use of complex 5, based on the calculated yield (of dioxygen), TON (TON = turnover number = mol H2O2/mol catalyst) and TOF (turnover frequency = mol H2O2/mol catalyst/h calculated as 3600 × (Vin/2/[1-5]0) values (Table 2 and Figure 7). These data for 5 are Yield = 48%, TON = 145 and TOF = 1065 h −1 . The TOF values for 5 are significantly higher than those observed for previously reported Fe II (IndH) systems, [45]. In contrast, the lowest activity was observed in the case of complex 4 with Yield = 6.8%, TON   peroxo complexes decreases significantly with the increase in water, which suggests that the 1 P -5 P intermediates play a key role in the catalytic cycles.     peroxo complexes decreases significantly with the increase in water, which suggests that the 1 P -5 P intermediates play a key role in the catalytic cycles.     (Table 2).
To prove this assumption, we followed the formation and decomposition of peroxo intermediates (rise and fall of their visible chromophores at λ max = 720 (Figure 8a (Figure 12a)), as well as the appearance of evolving dioxygen during the disproportionation reaction of H 2 O 2 , using parallel UV-Vis and gas volume measurement methods (Figures 8b, 9b, 10b, 11b and 12b).
Based on the above measurements, the profile of the catalytic reactions can be divided into two main steps, namely a short lag phase with the formation and accumulation of peroxo intermediates, followed by their decomposition with linear dioxygen formation.
To prove this assumption, we followed the formation and decomposition of peroxo intermediates (rise and fall of their visible chromophores at λmax = 720 (Figure 8a (Figure 12a)), as well as the appearance of evolving dioxygen during the disproportionation reaction of H2O2, using parallel UV-Vis and gas volume measurement methods (Figures 8b, 9b, 10b, 11b and 12b). Based on the above measurements, the profile of the catalytic reactions can be divided into two main steps, namely a short lag phase with the formation and accumulation of peroxo intermediates, followed by their decomposition with linear dioxygen formation.  To prove this assumption, we followed the formation and decomposition of peroxo intermediates (rise and fall of their visible chromophores at λmax = 720 (Figure 8a (Figure 12a)), as well as the appearance of evolving dioxygen during the disproportionation reaction of H2O2, using parallel UV-Vis and gas volume measurement methods (Figures 8b, 9b, 10b, 11b and 12b). Based on the above measurements, the profile of the catalytic reactions can be divided into two main steps, namely a short lag phase with the formation and accumulation of peroxo intermediates, followed by their decomposition with linear dioxygen formation.  In order to determine the rate dependence on the various reactants, disproportionation runs were performed at different substrate and catalyst concentrations (Table 3). At constant [H 2 O 2 ] 0 , the initial rate of H 2 O 2 disproportionation varies linearly with the in situ-formed [catalyst 3 P , 4 P or 5 P ] 0.5 , meaning that all reactions are half-order in catalyst and suggesting a dissociation process via homolytic cleavage of the O-O bond (Figure 13a), similarly to the previously published 1 and 2 H 2 O 2 systems [40]. However, at low H 2 O 2 concentrations, the reactions are first-order in peroxide concentration as shown in Figure 13b   In the next step, the effect of ligand modification on the reaction rate was investigated by introducing an electron-donating (-CH 3 ) substituent into the benzimidazole (L 2 , 2) and phenyl ring (L 4 , 4), replacing a pyridyl arm with a thiazolyl ((L 3 , 3) and a benzimidazole arm with triazole (L 5 , 5). We found that complexes 3 and 5 containing the thiazolyl and triazole side chains are the most effective oxidants with the fastest rates: k cat = 11.3 × 10 −3 M −1/2 s −1 and 96.1 × 10 −3 M −1/2 s −1 , respectively. On the other hand, complexes 1 and 4, with unsubstituted and 4-methylpyridine arms, are the less efficient oxidants with k cat = 2.81 × 10 −3 M −1/2 s −1 and 2.16 × 10 −3 M −1/2 s −1 , respectively. These results provide clear evidence that the ligand environment of the metal center influences the redox potential and the catalase activity of the complexes. Table 1 and Figure 14a show that the redox potential (E 1/2 ) of the complexes and their activity increase with the increase in the electron-withdrawing capacity of the substituent/side-chain. A kinetic isotope effect (KIE) of 1.36 was observed for the decay of 4 P when the experiments were carried out in the presence of added H 2 O (k = 3.68 × 10 −3 s −1 ) or D 2 O (k = 2.70 × 10 −3 s −1 ) (Figure 14b). These results indicate the role of the water during the diiron-peroxo-mediated disproportionation reaction. disproportionation reaction of H2O2 were carried in MeCN at 298 K under pseudo-first-order conditions with an excess of H2O2 ( [3][4][5]0: [H2O2]0 = 0.5-2.5 mM: 150-300 mM) using an initial rate method monitoring the increase in the evolved dioxygen by gas volumetric measurement. The estimated initial rates (under same conditions) according to the reaction rate order are Vi = 1.51 × 10 −5 M s −1 , Vi = 1.86 × 10 −5 M s −1 , Vi = 3.40 × 10 −5 M s −1 , Vi = 5.65 × 10 −5 M s −1 and Vi = 59.2 × 10 −5 M s −1 , respectively to compounds 4, 1, 2, 3 and 5, respectively, at 298 K. Based on the observed initial rates, compound 5 has significantly higher reactivity than 4.
In order to determine the rate dependence on the various reactants, disproportionation runs were performed at different substrate and catalyst concentrations (Table 3). At constant [H2O2]0, the initial rate of H2O2 disproportionation varies linearly with the in situ-formed [catalyst 3 P , 4 P or 5 P ] 0.5 , meaning that all reactions are half-order in catalyst and suggesting a dissociation process via homolytic cleavage of the O─O bond (Figure 13a), similarly to the previously published 1 and 2 H2O2 systems [40]. However, at low H2O2 concentrations, the reactions are first-order in peroxide concentration as shown in Figure 13b    In the next step, the effect of ligand modification on the reaction rate was investigated by introducing an electron-donating (-CH3) substituent into the benzimidazole (L 2 , 2) and phenyl ring (L 4 , 4), replacing a pyridyl arm with a thiazolyl ((L 3 , 3) and a benzimidazole arm with triazole (L 5 , 5). We found that complexes 3 and 5 containing the thiazolyl and triazole side chains are the most effective oxidants with the fastest rates: kcat = 11.

Gas Volumetric Measurements
The catalase-like activity was investigated through the evolution of dioxygen. The reactions were carried out in a 30 cm 3 reactor under air at 25 • C. In a typical experiment, 4 × 10 −5 mol of [Fe II (L 3 ) 3 ](CF 3 SO 3 ) 2 , [Fe II (L 4 ) 3 ](CF 3 SO 3 ) 2 and [Fe II (L 5 ) 3 ](CF 3 SO 3 ) 2 (3/4/5) was dissolved in 20 cm 3 CH 3 CN and the reactor was closed with a rubber septum. The flask was connected to a graduated burette filled with oil. The right amount of H 2 O 2 was added to the reactor through the septum and the reaction was followed by the amount of the dioxygen evolved. At constant [H 2 O 2 ] 0 , the reactions exhibit 1 2 -order kinetics on catalysts (in situ formed 1 P -5 P ) and the 1 2 -order rate constants (k 1 ) were obtained from the slope of the plot of V i (after the leg phase) vs. [1 P -5 P ]. At fixed [1 P -5 P ], reaction rates showed a good linear dependence with [H 2 O 2 ] 0 , affording first-order rate constants (k 1 '). The catalytic constants were obtained from either k 1 /[H 2 O 2 ] 0 or k 1 '/[1 P -5 P ] 1/2 .

UV-Vis Spectroscopic Measurements
The reactive species [Fe III 2 (µ-1,2-O 2 )(L 1−5 ) 4 ] (1 P -5 P ) were generated by adding 150-300 equivalent H 2 O 2 to the solutions of (1-5) in a 1 cm quartz UV cuvette at 25 • C. The total volume was 2 cm 3 (in CH 3 CN). The formation and the decay of (1 P -5 P ) was monitored with a UV-Vis spectrophotometer at 720, 710, 700, 710 and 695 nm, respectively. In a typical experiment, 4 × 10 −5 mol of (1-5) was dissolved in 2 cm 3 CH 3 CN and the right amount of H 2 O 2 was added to the cuvette. Reactions were run at least in triplicate, and the data reported are the average of the reactions.

Conclusions
The reactivity of three iron(II) complexes (3, 4 and 5) with heterobidentate ligands compared with the previously reported 1 and 2 have been investigated in the disproportionation reaction of H 2 O 2 as possible biomimics of catalase enzymes. It can be concluded that the formation of the corresponding µ-1,2-peroxo-diiron(III) complexes (1 P -5 P ) and their catalase-like activity can be clearly verified in the case of all three precursor complexes (1)(2)(3)(4)(5). Similar to our previous kinetic results on the 1 P (2 P )-containing system, we have got a half-order for 3 P , 4 P and 5 P , which further confirms the dissociative mechanism, including the homolytic O-O cleavage and the formation of mononuclear oxoiron(IV) species. The formation of oxoiron(IV) can be described by a pre-equilibrium process, which is shifted in the direction of the starting peroxo complex. Its presence in a steady-state concentration is also supported by previous theoretical calculations [40]. During fine-tuning of the catalyst, we found clear evidence that the electron-deficient sites (3 and 5) are significantly more reactive, the higher the redox potentials of the Fe III /Fe II redox couple, the higher the catalase-like activity. Furthermore, it is important to note that the two most active catalysts with L 3 and L 5 ligands have a high-spin electronic configuration. This is in good agreement with the electrophilic nature of the proposed oxoiron(IV) species during the catalytic cycle. Based on the obtained results, the proposed route may represent a new alternative to the mechanism of dinuclear catalase systems (Scheme 3).

Conclusions
The reactivity of three iron(II) complexes (3, 4 and 5) with heterobidentate ligands compared with the previously reported 1 and 2 have been investigated in the disproportionation reaction of H2O2 as possible biomimics of catalase enzymes. It can be concluded that the formation of the corresponding µ-1,2-peroxo-diiron(III) complexes (1 P -5 P ) and their catalase-like activity can be clearly verified in the case of all three precursor complexes (1)(2)(3)(4)(5). Similar to our previous kinetic results on the 1 P (2 P )-containing system, we have got a half-order for 3 P , 4 P and 5 P , which further confirms the dissociative mechanism, including the homolytic O-O cleavage and the formation of mononuclear oxoiron(IV) species. The formation of oxoiron(IV) can be described by a pre-equilibrium process, which is shifted in the direction of the starting peroxo complex. Its presence in a steady-state concentration is also supported by previous theoretical calculations [40]. During fine-tuning of the catalyst, we found clear evidence that the electron-deficient sites (3 and 5) are significantly more reactive, the higher the redox potentials of the Fe III /Fe II redox couple, the higher the catalase-like activity. Furthermore, it is important to note that the two most active catalysts with L 3 and L 5 ligands have a high-spin electronic configuration. This is in good agreement with the electrophilic nature of the proposed oxoiron(IV) species during the catalytic cycle. Based on the obtained results, the proposed route may represent a new alternative to the mechanism of dinuclear catalase systems (Scheme 3).