Accessing a synthetic FeIIIMnIV core to model biological heterobimetallic active sites

Metalloproteins with dinuclear cores are known to bind and activate dioxygen, with a subclass of these proteins having active sites containing FeMn cofactors and activities ranging from long-range proton-coupled electron transfer (PCET) to post-translational peptide modification. While mechanistic studies propose that these metallocofactors access FeIIIMnIV intermediates, there is a dearth of related synthetic analogs. Herein, the first well-characterized synthetic FeIII–(μ-O)–MnIV complex is reported; this complex shows similar spectroscopic features as the catalytically competent FeIIIMnIV intermediate X found in Class Ic ribonucleotide reductase and demonstrates PCET function towards phenolic substrates. This complex is prepared from the oxidation of the isolable FeIII–(μ-O)–MnIII species, whose stepwise assembly is facilitated by a tripodal ligand containing phosphinic amido groups. Structural and spectroscopic studies found proton movement involving the FeIIIMnIII core, whereby the initial bridging hydroxido ligand is converted to an oxido ligand with concomitant protonation of one phosphinic amido group. This series of FeMn complexes allowed us to address factors that may dictate the preference of an active site for a heterobimetallic cofactor over one that is homobimetallic: comparisons of the redox properties of our FeMn complexes with those of the di-Fe analogs suggested that the relative thermodynamic ease of accessing an FeIIIMnIV core can play an important role in determining the metal ion composition when the key catalytic steps do not require an overly potent oxidant. Moreover, these complexes allowed us to demonstrate the effect of the hyperfine interaction from non-Fe nuclei on 57Fe Mössbauer spectra which is relevant to MnFe intermediates in proteins.


Introduction
6][17][18][19] The reproducible coordination of one Fe ion and one Mn ion in designated sites is unusual: Fe is more abundant than Mn under physiological conditions, 20,21 and the Irving-Williams series supports stronger binding of Fe II over Mn II , so an Fe II Fe II active site is thermodynamically more likely to form than an Fe II Mn II one. 22It has been proposed that the tertiary structures of proteins enforce the site-specic binding of these metal ions, 17 but the thermodynamic requirements for enzymatic functions (e.g., the relative accessibility of various oxidation levels of Fe vs. Mn) may also serve as a driving force for the metal selectivity, and these factors are underexplored.2][33][34][35] In this work, we describe a series of FeMn complexes in the same ligand architecture and explore their proton transfer, electron transfer, and proton-coupled electron transfer (PCET) properties.We examine a high-valent complex that contains an Fe III -(m-O)-Mn IV core that is relevant to key intermediates found for Class Ic RNR and R2lox.
We have previously reported the usage of the multifunctional ligand [poat] 3− (N,N ′ ,N ′′ -[nitrilotris-(ethane-2,1-diyl)]tris(P,Pdiphenylphosphinic amido)) [36][37][38][39][40] in which the C 3 trianionic framework can support a metal center up to the 4+ oxidation level in high spin states.Our studies have shown that the phosphinic amido groups can form intramolecular hydrogen bonds (H-bonds), 38 act as sites for proton storage, and be part of an auxiliary metal ion binding site, allowing us to construct discrete unsymmetrical bimetallic complexes.We have systematically varied the identity of the auxiliary metal center in a series of M/Fe dinuclear compounds (M = rst row transition metals, group II alkali earth metals) to probe their effects on the electronic structures of Fe IV ]O units 36 as well as magnetic and redox phenomena, 37 and prepared a series of di-Fe complexes with the general formulation [(TMTACN)Fe m -(m-O(H))-Fe n (H) poat] z+ (TMTACN = 1,4,7-trimethyl-1,4,7-triazacyclononane; m = II, III; n = II, III, IV; z = 0, 1, 2) that spans four oxidation states. 39We now report that replacing the Fe center in the [poat] 3− site with a Mn ion provides a new heterobimetallic system that can be oxidized to the Fe III Mn IV level and serve as a synthetic model for FeMn-containing enzymatic active sites (Scheme 1).

Preparative routes and structure
The starting Mn synthon, K[Mn II poat], was prepared by treating H 3 poat with three equivalents of KH and then subsequently adding Mn II (OAc) 2 (Scheme 2).The isolated K[Mn II poat] salt, in the presence of 18-crown-6 and [Fe II (TMTACN)(OTf) 2 ], was allowed to react with isopropyl 2-iodoxybenzoate at −35 °C to yield [(TMTACN)Fe III -(m-O)-Mn III poat](OTf) ([Fe III (O)Mn III poat] OTf); crystalline needles suitable for X-ray diffraction were obtained aer multiple rounds of recrystallization at room temperature.The molecular structure revealed an Fe-O-Mn core with Mn-O1 and Fe-O1 bond lengths of 1.767(3) and 1.802(3) Å, respectively, and an Fe/Mn distance of 3.205(5) Å (Fig. 1A and Table 1); these values are comparable with those of the di-Fe III analog. 39The Mn III site in the [poat] 3− framework adopts a trigonal bipyramidal geometry with an N 4 O primary coordination sphere comprising the N-atom donors of the [poat] 3− ligand and a bridging oxido ligand; the Fe III site in the Fe-TMTACN adduct is 6-coordinated with an N 3 O 3 primary coordination sphere comprising the TMTACN ligand, two Oatoms from the phosphinic amido groups from the [poat] 3− ligand, and the bridging oxido ligand.This heterobimetallic complex, with its three-atom-bridge motif, a dynamic primary coordination sphere, and the ability to form an intramolecular H-bond network, incorporates crucial elements observed in the FeMn active sites of Class Ic RNR and R2lox. 12,14,16,41he movement of protons is critical in many enzymatic processes, 4,[42][43][44][45][46] and so we explored the protonation/ deprotonation steps within [Fe III (O)Mn III poat] + .The addition of one equivalent of 2,6-lutidinium triate to [Fe III (O)Mn III- poat] + at −60 °C resulted in the immediate replacement of the l max = 795 nm feature with a new band at 750 nm (Scheme 1 and Fig. S1A †), which was determined from electron paramagnetic resonance (EPR) spectroscopy (see below) to be the protonated species [(TMTACN)Fe III -(m-OH)-Mn III poat] 2+ ([Fe III (OH)Mn III poat] 2+ ).This hydroxido-bridged complex was thermally unstable, even at −60 °C, and further converted to a species having a peak at l max = 648 nm aer one hour or upon warming to room temperature (Scheme 1 and Fig. S1B † N-atom of one phosphinic amido group with its P]O unit now coordinated to the Mn III center (Fig. 1B and Table 1).‡ The Mn-O1 and Fe-O1 bond lengths of 1.793(2) and 1.795(2) Å, respectively, and the Fe/Mn distance of 3.206(1) Å are similar to the measurements in the core of [Fe III (O)Mn III poat] + .The H-atom at the N-position could be identied using a difference-Fourier map and was also corroborated by a strong n(N-H) feature at 3260 cm −1 in the FT-IR spectrum (Fig. S2 †).The systematic blue-shiing of this optical transition upon protonation, as well as the intramolecular H + transfer, were also observed for the di-Fe series. 39gnetic properties of the Mn III Fe III complexes S-and X-band EPR spectroscopy (Fig. 2) revealed that each Fe III Mn III complex has a distinct six-line hyperne feature from its 55 Mn center near g = 2.Although the spectral differences between the complexes are subtle, they are reproducible and could be simulated with distinctly different parameters for each species.33,47 Simultaneous least-squares tting of the spectra for both frequencies provided the g-and A( 55 Mn)-tensors for the coupled system  (Table 2).A rotation between the g-and A-tensors was required to simulate both the S-and X-band spectra with the same parameter set.The g-values are all larger than 2 because the gvalues of the coupled S = 1/2 state are given by g c = 2 − (4/3)(g Mn − 2) where the g-values for the Mn III centers are all less than 2 assuming g Fe = 2.The deviations in the Mn III g-values from 2 are at most 0.03, indicating that the spin-orbit contributions to the A-tensor are small.The spin-dipolar contribution to the Atensor (A SD ) was obtained by subtracting the isotropic value (A iso ) = trace(A Mn )/3 from A Mn for the uncoupled S = 2 Mn III site.The uncoupled values were determined from A Mn = −3/4 A C Mn , where A C Mn is the magnetic hyperne tensor for coupled spin S = 1/2.Only the magnitude of the hyperne values can be determined from EPR spectroscopy, but the isotopic value is known to be negative.9][50] The values of A SD are close to those obtained from density functional theory (DFT) calculations for the complexes and are within the uncertainty of the small spin-orbit contributions.The g-and A SD -tensors for the Mn III sites are rhombic, owing to a distorted empty d z 2 orbital having mixtures with other d-orbitals as indicated by DFT.The distortion is inuenced by the Fe-O-Mn angle being substantially less than 180°.As we have previously described, a rhombic A SD tensor was observed for the isoelectronic d 4 Fe IV site of [Fe III (O)Fe IV poat] 2+ , in which the di-Fe oxido bridge core is also bent; moreover, these ndings contrast with those found for the related [Fe IV poat(O)] − complex that has an axial A SD tensor. 36,39he EPR features of [Fe III (OH)Mn III poat] 2+ were further examined at higher temperatures to demonstrate the presence of the hydroxido bridging ligand.Upon warming above 20 K, a new signal appeared with features at g = 5.8, 4.9, and 3.0, which was absent from both [Fe III -(m-O)-Mn III ] complexes (Fig. 3).These g-values are expected for an S = 3/2 spin system with E/D = 0.16.When the spectra were plotted as intensity × temperature, an intensity increase was found at higher temperature as the S = 3/2 excited state became populated.The temperature dependence of the S = 3/2 signal (Fig. 3, inset) was tted using the spin Hamiltonian JS Fe $S Mn with J = +40 cm −1 , a value indicative of an antiferromagnetically coupled dinuclear system with a hydroxido bridging ligand. 51DFT calculations of the [Fe III (OH)Mn III poat] 2+ complex also gave a J-value of   The black traces are S = 1/2 simulations using the parameters given in Table 2.The g = 2 position is indicated.
+40 cm −1 .The zero-eld values D for the sites cannot be determined but are ∼0.5 cm −1 for Fe III TACN complexes and ∼2 cm −1 for Mn III complexes in a trianionic ligand scaffold similar to [poat] 3− . 49These values are small relative to J = 40 cm −1 and have only a minor effect (only a few percent) on the observed A-tensor.
For [Fe III (O)Mn III poat] + and [Fe III (O)Mn III Hpoat] 2+ , the temperature dependence of the microwave power required to half-saturate the EPR signal (P 1/2 ) was measured over the temperature range 4-40 K (Fig. S3 †).A t of the data (Fig. S3 †) to an Orbach relaxation curve and DFT gave the experimental and computed J-values, respectively (Table 2).The values are greater than 100 cm −1 and are consistent with an oxido bridging ligand. 51These values are similar to those of the oxido bridged Fe III Mn III complexes with TACN and TMTACN ligands (J ∼ 130 cm −1 ) from magnetic susceptibility measurements. 31ariable-eld 57 Fe Mössbauer spectra and simulations of [Fe III (O)Mn III poat] + (Fig. 4) are indicative of an S = 5/2 Fe III center exchange-coupled antiferromagnetically to an S = 2 Mn III center with J > 100 cm −1 , thus producing a coupled S = 1/2 ground state.The Fe III site has an isomer shi (d) of 0.53 mm s −1 , quadrupole splitting (DE Q ) of −1.84 mm s −1 , and an isotropic A-tensor of −20 T. These parameters are indicative of a high-spin Fe III center. 7,47However, the experimental and simulated 57 Fe Mössbauer spectra did not match at low magnetic eld (45 mT, Fig. 4).The cause of this mismatch appears to result from the spin expectation of the Fe being reduced from its normal value to give an overall spectral line splitting that was smaller than expected.One possibility is that this lower spin expectation occurs for coupled metal spin systems with zero-eld splitting (D) comparable to the exchange coupling (J).However, this possibility would cause the simulations at higher elds to not match the data, which they clearly do (Fig. 4), and was therefore ruled out.
The origin of the effect was discovered to be from the hyperne interaction of the 55 Mn nuclear spin (I = 5/2) with the electronic system spin (S = 1/2).For Fe complexes, the energy of the hyperne interaction (S$A Fe $I Fe ) is approximately 0.001 cm −1 .For small magnetic elds produced by permanent  magnets (∼50 mT) that are commonly used in Mössbauer spectrometers, the energy splitting between the electronic spin states is ∼0.04 cm −1 , consequently, the electronic spin expectation at the Fe site is dominated by the magnetic eld interaction (m B S$g$B).The EPR analysis of the S = 1/2 signal in [Fe III (O)Mn III poat] + (Table 2) gave an isotropic hyperne constant for 55 Mn of A iso = 260 MHz (0.0087 cm −1 ), which is more than 10% of the electronic energy splitting.Thus, the nuclear spin of Mn site signicantly alters the electronic spin expectation observed at the Fe site.
To examine how this interaction affects the analysis of the Mössbauer data, we have incorporated the 55 Mn hyperne interaction (S$A Mn $I Mn ) into our simulation soware to allow diagonalization of the spin Hamiltonian with the electronic and Mn nuclear spin states.The new simulations of Mössbauer spectra collected at 7.5 and 45 mT show a dramatic change which closely matches the data (Fig. 5).The effect of this interaction is only observed at lower magnetic elds, while simulations at higher elds are unaffected by the Mn nuclear spin because the electronic spin at the Fe site is dominated by the magnetic eld interaction.

Preparation and properties of an Fe III -(m-O)-Mn IV complex
Dioxygen-activating FeMn enzymes such as RNR Class Ic and R2lox have been proposed to access higher oxidation levels (e.g., Fe III Mn IV and Fe IV Mn IV ) to achieve function; [6][7][8][9]12,14,18 we therefore used cyclic voltammetry to evaluate the redox properties of our Fe III Mn III complexes. The cyclc voltammogram of [Fe III (O) Mn III poat] + revealed a nearly reversible, one-electron redox event at +0.20 V vs. [FeCp 2 ] +/0 , which was assigned to the Fe III Mn IV/III process (Fig. 6A).This potential is signicantly more negative than the Fe III Fe IV/III potential of +0.55 V vs.  6B).While these features are poorly resolved, they resemble those reported for the catalytic Fe III Mn IV intermediate in RNR Ic. 12 This new species did not show an EPR signal, suggesting an integer spin complex (Fig. S4 †).This redox process was reversible: the addition of FeCp 2 to this new compound regenerated the optical and EPR features of [Fe III (O) Mn III poat] + and [FeCp 2 ] + , supporting the hypothesis that the reaction was an outer-sphere electron transfer.
Variable-eld 57 Fe Mössbauer spectra and simulations of the oxidized species showed that the features associated with [Fe III (O)Mn III poat] + are nearly absent (Fig. 7).The simulations were based on an S = 5/2 Fe III ion antiferromagnetically exchange-coupled to an S = 3/2 Mn IV ion with J > 100 cm −1 , producing a coupled system with an S = 1 state that is lowest in energy.The Fe III site has the parameters d = 0.50 mm s −1 , DE Q = Fig. 5 Mössbauer spectra (red traces) of [ 57 Fe III (O)Mn III poat] + in PrCN at 4.2 K and the magnetic fields listed.The simulations (black traces) are for S = 5/2 Fe III antiferromagnetically exchange-coupled to S = 2 Mn III with inclusion of the 55 Mn hyperfine interaction using the Atensor and rotation derived from EPR spectroscopy (Table 2).The dashed lines are without the 55 Mn hyperfine interaction.([Fe III (O)Mn IV poat] 2+ ).Although this species has an S = 1 spin state, no signals were observed in parallel-mode EPR spectra.A minor species, assigned to [Fe III (O)Mn III poat] + , was also present in the Mössbauer spectra (15%) and is most evident in the 45 mT spectrum it was included in all simulations.As observed in the spectra of [Fe III (O)Mn III poat] + , the Mössbauer spectrum of [Fe III (O) Mn IV poat] 2+ at 45 mT was affected by the Mn nuclear spin (Fig. 7, dashed line).The effect is minor owing to the small spin expectation for the spin S = 1 state at low eld (Fig. 7
Our reactivity studies additionally found that [Fe III (O)Mn IV- poat] 2+ did not react with compounds having relatively weak C-H bonds, such as 9,10-dihydroanthracene (72.9) and xanthene (70.2).This lack of reactivity was also found for the analogous [Fe III (O)Fe IV poat] 2+ species, which also reacts with similar phenols to rst produce [Fe III (OH)Fe III poat] 2+ . 39In both systems, the homolytic cleavage of the O-H bond of the phenolic substrate leads to the initial protonation of the oxido ligand, which is sterically hindered by the [poat] 3− ligand and methyl groups of the TMTACN ligand.We proposed that the PCET process involved for homolytic ArO-H bond cleavage by [Fe III (O)Fe IV- poat] 2+ favors the protonation of a site near to the metal center even though it was hindered (that is, rather than protonating the ligand to form [Fe III (O)Fe III Hpoat] 2+ ).A similar process appears to be operative for the Fe III Mn IV analog.However, [Fe III (O)Mn IV- poat] 2+ did not react with 2,6-tert-butyl-4-R-PhOH (R = -OMe (BDFE = 72.6 kcal mol −1 ), -t Bu (75.5), and -H (77.0), 52 substrates that are substantially more sterically hindered.Steric effects cannot be the only reason for this lack of reactivity because we have previously shown that the analogous [Fe III (O)Fe IV poat] 2+ did react with these substrates. 39We do not completely understand the lack of reactivity for [Fe III (O)Mn IV poat] 2+ and its apparent difference from that found for the Fe III Fe IV analog.With this said, the two bimetallic complexes have differing Fe III M IV /Fe III- M III redox potentials with [Fe III (O)Fe IV poat] 2+ being the stronger oxidant that may contribute to its increased reactivity. 39

Conclusions
In summary, the rigid [poat] 3− /TMTACN ligand scaffold allowed us to construct discrete FeMn complexes and examine their individual or coupled proton and electron transfer steps.We described the reactivity, spectroscopic character, and electrochemical properties of a high-valent compound that contains an Fe III -(m-O)-Mn IV core.We have previously shown that mononuclear Mn-O(H) complexes in trigonal symmetry have lower M IV/ III reduction potentials than their Fe counterparts, [55][56][57] but, to our best knowledge, this work presents the rst direct experimental comparison of the redox properties, and the resultant changes in the electronic structures, of related FeMn and di-Fe compounds at an oxidation state above 3+.Our studies found that the lower potentials of FeMn complexes compared to their di-Fe analog could be used to form a complex with an Fe III -(m-O)-Mn IV core.Within a biological context, a more thermodynamically accessible Mn IV ion over an Fe IV ion can be one important contributing factor leading to the selective binding of FeMn sites in Class Ic RNR and R2lox enzymes, especially when the key catalytic steps do not require an overly potent oxidant (such as di-Fe IV ) that may cause undesirable damage to the proteins.[Fe III (O)Mn IV poat] 2+ shares similar electronic, magnetic, and spin-exchange features as RNR R2 Ic-X, suggesting the former to be the rst wellcharacterized model for the biological Fe III Mn IV core.
The Mössbauer ndings on [Fe III (O)Mn IV poat] 2+ and [Fe III (O) Mn III poat] + revealed a mismatch between the data and simulations at low external magnetic elds that is caused by the hyperne interaction of the 55 Mn nuclear spin with the electronic system spin.This effect on the Mössbauer spectra of inorganic complexes is uncommon because it requires an electronic interaction between Fe and a second metal ion having a nuclear spin and large hyperne values.The possible inuence of the 55 Mn nuclear spin has been suggested in the Mössbauer spectra of an Mn IV Fe IV intermediate in Chlamydia trachomatis ribonucleotide reductase. 8Our ndings for [Fe III (O) Mn III poat] + and [Fe III (O)Mn IV poat] 2+ support this premise and demonstrate a new analysis to obtain quantitative agreement between Mössbauer data and simulations that incorporate the 55 Mn A-tensor and rotation derived from EPR spectroscopy.To our knowledge, this is the rst quantitative demonstration of the effect of the hyperne interaction from a non-Fe nucleus on the Mössbauer spectra of inorganic complexes.Our work also highlights the importance of accessibility during PCET processes: the initial formation of the [Fe III (OH)Mn III poat] 2+ species upon PCET with phenolic substrates, in which the proton transfers to the bridging oxido ligand and the electron reduces the Mn IV center, appears to be necessary for reactivity to occur.Within a growing body of literature and improved understanding of FeMn enzymes and their oxidative chemistry, the [(TMTACN)Fe m -(m-O(H))-Mn n (H)poat] z+ system serves as a useful structural, spectroscopic, and functional model to complement biochemical investigations.
Scheme 1 Individual or coupled proton and electron transfers in the FeMn complexes described in this work.

Fig. 1
Fig. 1 Thermal ellipsoid diagrams depicting the molecular structures of [Fe III (O)Mn III poat] + (A) and [Fe III (O)Mn III Hpoat] 2+ (B).Ellipsoids are drawn at the 50% probability level, and only the phosphinic amide H atom is shown for clarity.The triflate counterions are outer-sphere and are not interacting with the cation.
a d[M-N/O x ] denotes the displacement of the metal atom from the 3atom plane.N/O eq represented the plane formed by N2, N3, N4 or O2, N3, N4.N TMTACN represents the plane formed by N5, N6, N7. b Trigonality structural parameter, s 5 = (b − a)/60°.b is the largest bond angle observed, and a is the second largest bond angle observed.

Fig. 3 XFig. 4
Fig. 3 X-band (9.620 GHz) EPR spectra of [Fe III (OH)Mn III poat] 2+ at 3 mM in CH 2 Cl 2 at the measurement temperatures listed.The inset shows the temperature dependence of the S = 3/2 signal fitted with J = 40 cm −1 (JS Fe $S Mn ).The signal at g = 4.3 is from a minor Fe III impurity.
[FeCp 2 ] +/0 observed in the di-Fe analog, 39 suggesting that the redox change occurs at the Mn site.Treatment of [Fe III (O) Mn III poat] + with [FeCp(C 5 H 4 C(O)Me)]OTf or [N(p-C 6 H 4 Me) 3 ]OTf at −60 °C resulted in the loss of the diagnostic peak at l max = 795 nm and the concurrent appearance of shoulders at 420, 490, and 610 nm (Fig.

Fig. 7
Fig. 7 Mössbauer spectra (red traces) of [ 57 Fe III (O)Mn IV poat] 2+ in PrCN at 4.2 K and the magnetic fields listed.The simulations (black traces) are for an S = 5/2 Fe III ion antiferromagnetically exchange-coupled to S = 3/2 Mn IV ion (see text for parameters).The vertical arrows mark spectral features from a minor amount of [Fe III (O)Mn III poat] + .The dashed line in the 45 mT spectrum is without the55 Mn hyperfine interaction (see inset).

− 1 .
31 mm s −1 , and an isotropic A-tensor of −20 T. These values are typical of a high-spin Fe III center and close to those of [Fe III (O)Mn III poat] + , indicating that the Mn center has been oxidized.The new species was therefore formulated as [(TMTACN)Fe III -(m-O)-Mn IV poat] 2+

Table 2
Parameters derived from EPR spectroscopy and DFT (in parentheses) for the Fe III Mn III species a a A-values are in MHz.b Euler angles a, b, g indicate the rotation of A relative to g. c cm −1 (JS Fe $S Mn ).

Table 3
Spectroscopic comparison of [Fe III (O)Mn IV poat] 2+ and Fe/Mn RNR Complex S T l (nm, 3 M ) 57 Fe d/DE Q a A x,y,z ( 57 Fe) b J c