Comparison of Nonheme Manganese- and Iron-Containing Flavone Synthase Mimics

Heme and nonheme-type flavone synthase enzymes, FS I and FS II are responsible for the synthesis of flavones, which play an important role in various biological processes, and have a wide range of biomedicinal properties including antitumor, antimalarial, and antioxidant activities. To get more insight into the mechanism of this curious enzyme reaction, nonheme structural and functional models were carried out by the use of mononuclear iron, [FeII(CDA-BPA*)]2+ (6) [CDA-BPA = N,N,N’,N’-tetrakis-(2-pyridylmethyl)-cyclohexanediamine], [FeII(CDA-BQA*)]2+ (5) [CDA-BQA = N,N,N’,N’-tetrakis-(2-quinolilmethyl)-cyclohexanediamine], [FeII(Bn-TPEN)(CH3CN)]2+ (3) [Bn-TPEN = N-benzyl-N,N’,N’-tris(2-pyridylmethyl)-1,2-diaminoethane], [FeIV(O)(Bn-TPEN)]2+ (9), and manganese, [MnII(N4Py*)(CH3CN)]2+ (2) [N4Py* = N,N-bis(2-pyridylmethyl)-1,2-di(2-pyridyl)ethylamine)], [MnII(Bn-TPEN)(CH3CN)]2+ (4) complexes as catalysts, where the possible reactive intermediates, high-valent FeIV(O) and MnIV(O) are known and well characterised. The results of the catalytic and stoichiometric reactions showed that the ligand framework and the nature of the metal cofactor significantly influenced the reactivity of the catalyst and its intermediate. Comparing the reactions of [FeIV(O)(Bn-TPEN)]2+ (9) and [MnIV(O)(Bn-TPEN)]2+ (10) towards flavanone under the same conditions, a 3.5-fold difference in reaction rate was observed in favor of iron, and this value is three orders of magnitude higher than was observed for the previously published [FeIV(O)(N2Py2Q*)]2+ [N,N-bis(2-quinolylmethyl)-1,2-di(2-pyridyl)ethylamine] species.

The primary aspect of ligand selection, as a continuation of our previous work, was to increase the reactivity of the catalyst and its intermediates towards flavanone [46].  (9)) nonheme iron(II) and oxoiron(IV) complexes, were chosen. Since nonheme oxomanganese (IV) complexes have proven to be versatile oxidants [40], in addition to iron-containing models we also aimed to elucidate the role of the metal cofactor through the comparison of well-defined iron-and manganese-containing systems. Previously reported [Mn II (N4Py*)(CH 3 (10) as possible intermediates in the oxidation reactions were chosen for these measurements [39,40]. In this work, catalytic oxidation of flavanone was performed with 2, 3, 4, 5, and 6. Catalytic oxidation of ethylbenzene was performed with 5 and 6, and stoichiometric oxidation reactions were performed with 7, 8, 9, 10, and 11. N,N,N',N'-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) is a well-known metal chelator. TPEN complexes are often used as Zn(II) and Cd(II) indicators, and in this case, substituting the pyridine for quinoline results in enhancement of fluorescence intensity and use of these ligands as fluorescent probes [47,48]. Fe(II) complexes of the TPEN group of ligands have interesting electronical properties, where conformational changes are linked to different spin-state interconversion processes [41]. Due to the interesting redox behaviour of these Fe(II) complexes they have been studied as superoxide dismutase mimics [49] and for their reactivity towards hydrogen peroxide [50].
In this work, a single-crystal structure was obtained for the complex [Fe II (±CDA-BQA)](CF 3 SO 3 ) 2 (5). The complex was prepared with the racemic version of ligand CDA-BQA*. In the CSD database, there are 29 structures of Fe(II) complexes of these types of ligands, with four pyridyl or quinoline groups connected by an ethylenediamine or cyclohexanediamine linker [51,52]. The only reported Fe(II) complex with a cyclohexanediamine linker is [Fe II (CDA-BPA*)](ClO 4 ) 2 (6) (CSD refcode YAMXAL) [41] The geometry of the newly synthesised [Fe II (CDA-BQA*)](CF 3 SO 3 ) 2 (5) (Figure 1) and [Fe II (CDA-BPA*)](ClO 4 ) 2 (6) is compared in Figure 2 and Table 1. Both complexes are prepared with racemic ligands, however, 5 crystallised as a racemate, while 6 has spontaneously resolved into its optical isomers, containing only the (R,R) enantiomer. While complex 6 has a regular octahedral geometry, the Fe-N bonds in 5 are elongated, forming a pentagonal bipyramidal geometry with an equatorial vacancy, as determined using the program FindGeo [53]. The reason for this is likely the steric crowding of the quinoline groups in 5. The significantly longer Fe-N bond lengths (>2.2 Å) are in agreement with a high spin Fe(II) centre in 5. The UV-Vis spectrum of 5 in acetonitrile is dominated by the intense π-π* band at 307 nm (=12,800 M −1 cm −1 ), and an additional broad feature of low intensity between 330 and 460 nm (λ max = 356 with ε = 2080 M −1 cm −1 ) can be assigned to an MLCT transition. The weak intensity is consistent with the high spin state of the iron(II) centre [46].

Catalytic Oxidation Reactions
The catalytic activities of the three ferrous and two manganese complexes [Fe II (Bn-TPEN)(CH 3 (4) were studied in the oxidation of flavanone, utilising mCPBA (meta-chloroperoxybenzoic acid) as the co-oxidant. The oxidation reactions were carried out under standard catalytic conditions (5:100:500 ratio for the catalyst:substrate:oxidant) in acetonitrile at room temperature (Table 2 and Figure 3). Similarly to our previous results, it took less than 10 min to have about 30-36% and 19-24% yields (based on the substrate) for the iron(II)-and manganese(II)-catalysed reactions, respectively. Much lower yields (0.36 and 2.8%) were observed for the Fe(ClO 4 ) 2 and Mn(ClO 4 ) 2 salts, respectively. Both the iron(II) and manganese(II)-catalysed oxidations of flavanone produced flavones (F) as a major product in all cases in addition to two minor products, namely 2-hydroxy-2-phenyl-chroman-4-one (A, 2-hydroxy-flavanone) and its open tautomeric form 1-(2-hydroxy-phenyl)-3-phenyl-propane-1,3-dione (D). It was found earlier [46] that complex 1 together with mCPBA oxidizes flavanone, and a turnover number (TON) of 6.93 for F, 0.11 for D and 0.02 for A was obtained with an overall yield of 9.42%. Its manganese analogue produced an overall yield of 18.8%, TON for F = 3.58, TON for D = 0.19 and TON for A =~0.01.  It is important to mention that the amount of 1,3-dione (D) can be increased from 0.92% (TON = 0.19) to 4.2% (TON = 0.92) in the presence of H 2 O, suggesting an equilibrium step during the flavone formation ( Figure 4B). However, when 2,6-di-tert-butylphenol (DTBP) was added significantly less flavanone was converted, suggesting a free-radical type mechanism as a parallel process of the metal-based oxidation ( Figure 4A). Significantly higher yields were observed for 3 (  Recently, chiral N4Py-type and L-proline derived aminopyridine containing oxoiron(IV) complexes were reported which could perform enantioselective oxidation of various substrates such as thioanisole, hydrocarbons, alkenes and substituted cyclohexanones [31,33]. To the best of our knowledge, these are the first examples of chiral nonheme oxoiron(IV) complexes tested in asymmetric C-H hydroxylation reactions. Since flavanone is a chiral molecule including benzylic C-H bonds, oxidative kinetic resolution (OKR) of racemic flavanones can in principle also be performed with a chiral iron catalyst and oxoiron(IV) intermediates. The oxidation of ethylbenzene that can be used as a model compound of flavanone, by chiral [Fe IV (O)(N4Py*)] 2+ (7) species showed moderate enantioselectivities up to 33% ee as a result of a non-rebound mechanism including the epimerization of the long-lived alkyl radical before the rebound step [36]. To increase the enantioselectivity, we also examined the enantioselectivity of the [Fe II (CDA-BPA*)] 2+ (6) catalyst and the in situ generated [Fe IV (O)(CDA-BPA * )] 2+ (11) intermediate from 6 and PhIO in the asymmetric C-H hydroxylation of ethylbenzene utilizing TBHP and PhIO as co-oxidants ( Figure 5 and Table 3). In these probe reactions, moderate enantioselective hydroxylation could be obtained in all cases, so the studies on oxidative kinetic resolution of racemic flavanones were discarded.

Stoichiometric Oxidation Reactions
We  (Table 4), where the most reactive complex reacts 80-fold faster than the slowest one [46]. The stability and reactivity of the complexes are also greatly influenced by the number of nitrogen-donor atoms and their hybrid state. PhIO in acetonitrile at room temperature results in a green species (11), characterised by a characteristic absorption band at 740 nm (ε ≈ 450 M −1 cm −1 ), which can be assigned as a ligand-field (d-d) transition on the low-spin (S = 1) Fe(IV) centre. This species is much less stable (t 1/2 ≈ 2 h at 25 • C) than that was found for 9 (t 1/2 ≈ 6 h at 25 • C). Their decay can be significantly enhanced by the addition of flavanone. Table 4. Rate constants and activation parameters for the oxidation of flavanone with oxometal(IV) complexes.
[ The detailed kinetic measurements were carried out in CH 3 CN and CH 3 CN/CF 3 CH 2 OH (TFE) = 1:1 solutions at 10-25 • C, and the decay of the oxoiron(IV) species 9 and 11 during the flavanone oxidation was followed as a decrease in absorbance at~740 nm ( Figure 6 and Tables S2-S5). The yields of flavone were around~80% for both complexes. The reaction rates in the presence of 10-50 times excess of substrate obeyed pseudo-first-order kinetics, and the pseudo-first-order rate constants (k obs ) were directly proportional to the concentration of flavanone, from which the reaction rate constants (k 2 ) are 0.68 ± 0.027 M −1 s −1 and 0.97 ± 0.04 M −1 s −1 for 9 and 11 at 10 • C, respectively ( Figure 7 and Table 4). These values are three orders of magnitude higher than those observed for the previously published [Fe IV (O)(N2Py2Q*)] 2+ species [46], showing clearly that the ligand framework significantly influenced the reactivity of the oxoiron(IV) species.   (7), which is consistent with our catalytic results. Based on the temperature dependence of the reactivity of 9 (with ∆H ‡ = 28 ± 2 kJ mol −1 , ∆S ‡ = −150 ± 8 J mol −1 K −1 , ∆G ‡ = 72.7 kJ mol −1 ), the value of -T∆S ‡ determined was bigger than ∆H ‡ , indicating an entropy-controlled reaction, contrary to the previously reported enthalpy-controlled reactions with N4Py-type ligands. As a result of a compensation effect increasing activation, enthalpies are offset by increasingly positive entropies yielding ∆H ‡ = 114 kJ mol −1 at the intercept ( Figure 8A). The experimentally determined difference between ∆G ‡ values is 20 kJ mol −1 , which is significant and consistent with the observed reaction rate order ( Figure 8B).   (Table 4 and Figure 7)). The difference in reaction rates and yields of products (~80% flavone for 9 and~40% flavone for 10 based on the complex concentration) can be explained by a different mechanism based on the literature data. While in the case of oxoiron(IV) complexes the reactions occur mostly via an oxygen-rebound mechanism [56], in the case of manganese the process involving C-H activation can be described mainly by a non-rebound mechanism with a smaller reaction rate (Scheme 3) [40].

Materials and Methods
Reactions were carried out in ordinary glassware and chemicals were used as purchased from commercial suppliers without further purification. GC analyses were performed on an Agilent 6850 gas chromatograph equipped with a flame ionization detector and a 30 m SUPELCO BETA DEX225 (CHIRASIL-L-VAL) (Sigma-Aldrich, Budapest, Hungary) column. ESI-MS samples were analysed using a triple quadruple Micromass Quattro spectrometer (Waters, Milford, MA, USA) and an HPLC-MS system (Agilent Technologies 1200, Budapest, Hungary) coupled with a 6410 Triple-Quadrupole mass spectrometer, operating in a positive ESI mode. Synthesis of the ligand was carried out in a microwave reactor (CEM Discover), (CEM Inc, Scottsdale, AZ, USA) monitored by TLC on aluminium oxide 60 F 254 neutral plates and detected with a UV lamp (254 nm). NMR spectra were obtained on a Bruker Avance 300 (Bruker Biospin AG, Fällanden, Switzerland) or 600 spectrometers, operating at 300 or 600 MHz for 1 H and 75 or 150 MHz for 13 C. The spectra are recorded at room temperature. Chemical shifts, δ (ppm), indicate a downfield shift from the residual solvent signal (δ H : 1.94 ppm, δ C : 118.26 ppm for CD 3  Synthesis of ligands CDA-BPA* and CDA-BQA*. The synthesis was performed according to a modified previously reported procedure [47]. The amine (1 eq.), K 2 CO 3 (12 eq.), 2-(chloromethyl)pyridine hydrochloride or 2-(chloromethyl)quinoline hydrochloride (4 eq.) and KI (1 eq.) were suspended in 50 mL of acetonitrile. The reaction mixture was heated in a microwave reactor (50 W, reflux) for 1 h. The solvent was evaporated in a vacuum, the residue suspended in ethyl acetate and washed three times with brine and saturated NaHCO 3 , the organic layer dried over anhydrous sodium sulphate, filtered and evaporated in a vacuum. The crude ligand was purified by automated flash chromatography on a pre-packed neutral alumina column (48 g (Table S1).
Catalytic oxidation of ethylbenzene was carried out under thermostatic (273 K) conditions. In a typical reaction, 500 µL of TBHP (diluted from 70% TBHP) solution in CH 3 CN was delivered by a syringe pump in air to a stirred solution (1 mL) of 6 catalysts, and ethylbenzene inside a vial. The final concentrations were 2 mM catalyst 6, 50 mM (100, 200 mM) oxidant, and 500 mM substrate. The mixture was stirred for 15 min. The products were identified by GC analysis, and their yields were determined by comparison with authentic compounds using bromobenzene (25.00 × 10 −3 M) as an internal standard in the reactions.

Conclusions
In conclusion, we previously found that N4Py-based iron(II) complexes capable of carrying out 2,3-desaturation of flavanone via 2-hydroxyflavanone intermediate formation can act as a functional flavone synthase model. As a continuity of this study, efforts have been made to enhance the catalytic activity by the use of TPEN-type ligands and investigate the role and effect of the metal cofactor through manganese and iron complexes with the same ligand framework.  (7), which is consistent with our catalytic results, and shows that addition of cyclohexanediamine as the chiral element successfully lead to increase in the catalytic activity. Although we have reported the first example of efficient flavanone oxidation by iron and manganese complexes, detailed studies are underway to elucidate the mechanisms of flavone formation.

Conflicts of Interest:
The authors declare no conflict of interest.