C─H Bond Cleavage by Bioinspired Non-Heme Metal Complexes

Abstract

The functionalization of C─H bonds is one of the most challenging transformations in synthetic chemistry. In biology, these processes are well known and are achieved with a variety of metalloenzymes, many of which contain a single metal center within their active sites. The most well studied are those with Fe centers and the emerging experimental data show that high valent Fe–oxido species are the intermediates responsible for cleaving the C─H bond. This Forum describes the state of this field with an emphasis on non-heme Fe enzymes and current experimental results that provide insights into the properties that make these species capable of C─H bond cleavage. These parameters are also briefly considered in regard to Mn-oxido complexes and Cu-containing metalloenzymes. Synthetic Fe–oxido complexes are discussed to highlight their utility as spectroscopic and mechanistic probes, and reagents for C─H bond functionalization. Avenues for future research are also examined.

Synopsis Graphical Abstract

The cleavage of unactivated C─H bonds is a challenging transformation in synthetic chemistry yet is readily accomplished in biology by metalloenzymes. A sub-class of these enzymes contain active sites with single Fe centers that generate high valent Fe–oxido intermediates whose function is to cleave C─H bonds. This Forum describes the current state of this field from spectroscopic, mechanistic, and synthetic perspectives. Discussions on relevant Mn and Cu systems are also included.

Introduction

Functional group transformations are the heart of organic synthesis and many a student has been taught these concepts as a means of learning organic chemistry. For all the many types of reactions that have been developed, the most prevalent functional group, the C─H bond, is often overlooked. In fact, we usually neglect even showing these bonds and for the longest time, few practitioners even considered it a functional group. There are many good reasons for this neglect, the foremost being the large bond dissociation free energies associated with most C─H bonds – it is just too thermodynamically difficult to selectively cleave these bonds to afford targeted products.¹ However, the situation is different in biology where C─H bonds can be selectively and efficiently functionalized. This difference rests on a cadre of enzymes that have evolved to perform these difficult reactions and their usefulness in biosynthetic processes is impressive.² Although we recognize that there have been important advances in the development of synthetic reagents to accomplish these types of reactions, none compares to those that exist in nature. Understanding how enzymes accomplish C─H bond activation from both structure-function and mechanistic considerations remains an ongoing and challenging endeavor; yet these pursuits impact a wide area of science that encompass biochemistry, health, and synthetic chemistry. To the organic chemist, for instance, the ability to selectively convert a C─H bond at the late stage of a multi-step preparation would be a glad addition to any synthetic toolbox.

The aim of this Forum is to describe one of the main classes of enzymes that performs biological C─H bond functionalization, namely metalloenzymes that contain cofactors in which a non-heme, mononuclear Fe or Cu center has a central role. These metalloenzymes have been acknowledged to be potent at oxidizing organic substrates, yet we still are learning the details of how they manipulate C─H bonds. The focus will be on results over the last 5-10 years and covers both protein chemistry and synthetic complexes which have provided a wealth of information. The caveat is that this field is quite extensive with many outstanding contributions – this review is thus not meant to be comprehensive in content but rather, a selective Forum on some of the key accomplishments and controversies that still exist.

Cleavage of C─H Bonds by Metal Complexes: Comparisons.

Before going forward, it is instructive to briefly compare the two general mechanisms that transition metal complexes utilize to cleave C─H bonds. Within the organometallic field, it has been well documented that several 2^nd and 3^rd row transition metal complexes can cleave C─H bonds.^3-6 This type of process is illustrated using Bergman's classic IrCp complexes in which an oxidative addition reaction occurs to produce an (alkyl)Ir-hydride complex (Scheme 1A).^7,8 This two-electron process derives from the relatively low-valent Ir(I) complex that cleaves the C─H bond. Metal–oxido complexes function in a different manner, in which the metal center usually has a high formal oxidation state (greater than 3+).⁹ Gray has determined that the bonding between the metal center and oxido ligand contains substantial multiple bond character, resulting in an oxido ligand being electrophilic.¹⁰ The M–oxido unit is thus poised to cleave C─H bonds which typically occurs through a homolytic process to produce a hydrogen atom (H-atom) and a carbon-centered radical (Scheme 1B). The H-atom is transferred to the metal–oxido species with the oxido ligand being protonated to produce a hydroxido ligand and the metal center being reduced by one electron. Because the electron and proton end up on different atoms, this process is not actually an H-atom transfer but rather a proton-coupled electron transfer.^11-13 More recently, it has been described as a concerted proton-electron transfer process to emphasize that both the electron and proton are transferred together at the transition state.^1,14,15 We will return to this concept later in this Forum.

Scheme 1.

Cleavage of C─H bond by metal complexes using an organometallic approach (A) such as an IrCp-based precursor, or a metal-oxido (or hydroxido) approach (B).

The functionalization of C─H bonds comes after cleavage has occurred. To illustrate this point, we will use Fe=O and Fe–OH species because they are operative within proteins. High valent Fe=O intermediates are usually formed after the binding and activation of O₂.^16-22 For heme systems (e.g. cytochrome P450s), these intermediates are formally Fe^V=O (denoted, compound I) but spectroscopic studies conclusively show that it is an Fe^IV=O unit coupled to a ligand radical (Scheme 2).²³ After C─H bond cleavage, an Fe^IV–OH species is produced (denoted, compound II) that recombines with the substrate carbon radical to form an Fe^III–O(H)CR₃ species in a process that is referred as a rebound step.^24-26 It is the interplay between these two Fe species that drives biological C─H cleavage, which we will highlight in the ensuing sections.

Scheme 2.

Catalytic mechanism of aliphatic hydroxylation by cytochrome P450 enzymes.

Non-Heme Fe^IV=O Species.

Metalloproteins.

Bollinger and Krebs first reported the spectroscopic characterization of an enzymatic non-heme Fe^IV-oxido species in taurine dioxygenase (TauD), part of the superfamily of α-keto glutarate (α-KG) enzymes that has over a thousand members that catalyze a diverse set of reactions.^27-29 The active site of these enzymes in its resting state consists of a ferrous ion bound by two histidine residues and a carboxylate (aspartate or glutamate) in a facial coordination, known as the “facial triad”, with the remaining coordination sites occupied by 3 water molecules.^30-33 Water coordination is commonly found in resting states of non-heme proteins and serves important functions: 1) they stabilize the more reduced states of the metal centers and 2) they can be easily replaced by other ligands such as cofactors, small molecules that need to be activated, and substrates. All these roles are illustrated in α-KG enzymes and are represented in the consensus mechanism of action (Scheme 3).²⁸ Upon binding of the cofactor α-KG to the Fe(II) center, two water molecules are displaced; docking of the taurine substrate occurs proximal to the active site which causes the last aqua ligand to dissociate. The vacant coordination site allows for dioxygen to bind to the iron center, but the exact coordination of this species has not been experimentally confirmed. A reasonable proposal is for dioxygen to coordinate end-on,³⁴ leaving the distal O-atom to attack the α-keto carbon of the α-KG ligand, which leads to the dissociation of CO₂ which is the driving force to generate the reactive Fe^IV=O species (denoted intermediate J).³⁵ The Fe^IV=O species is then proposed to follow a similar pathway as proposed for P450s by hydroxylating taurine in a “rebound” mechanism, in which it first homolytically cleaves a C─H bond in taurine to form Fe^III–OH and the carbon-radical.^25,36 Recombination then occurs between the Fe^III–OH species and the resultant carbon radical to yield an alcohol. Although this mechanistic proposal is similar to that for P450s, there are important differences in that intermediate J is formally at the 4+ oxidation level, whereas compound I is more oxidized with the additional oxidizing equivalent stored on the ligands.²³ This difference also includes the rebound step which, for these non-heme oxygenases, is from an Fe^III–OH species. These differences could impact function: intermediate J decays at a rate of 13 s⁻¹ and has an observed kinetic isotope effect (KIE) of k_H/k_D ~ 50, both of which are different from those in heme monooxygenases For instance, a thermophilic P450 (CYP119) compound I decays at a rate of 220 s⁻¹ in the presence of 20 μM dodecanoic acid and a KIE value can range from 1 – 18 for P450 enzyme-catalyzed C─H bond activation.^23,37-39

Scheme 3.

Catalytic mechanism of aliphatic hydroxylation by TauD.

The key discovery in this field was the detection of intermediate J by Bollinger and Krebs using rapid spectroscopic methods.^35,38,40 They first employed stopped-flow UY-vis spectroscopy to observe the appearance of an absorbance feature at λ_max = 318 nm. Knowing the time scale for detection of intermediate J, they then used freeze-quench methods coupled with ⁵⁷Fe-Mössbauer spectroscopy to identify the oxidation level of the Fe center. Intermediate J has an experimentally determined isomer shift (δ) of 0.31 mm/s and quadrupole splitting (ΔE_Q) of −0.88 mm/s;³⁵ however, these commonly reported parameters are often not enough to conclusively determine the oxidation and spin states. For example, the parameters for intermediate J are not that different from the δ = 0.32 mm/s and ΔE_Q = −1.54 mm/s obtained for a synthetic Fe^III–oxido complex ([Fe^IIIH₃buea(O)]²⁻) that will be discussed in a later section.^41,42 However, for intermediate J there is additional evidence to support the assignment of an Fe^IV–oxido species: it lacks perpendicular-mode EPR signals and variable-field Mössbauer studies indicate it is an integer spin system. A subsequent report by Neese, Bollinger, and Krebs provided evidence to indicate it has an S = 2 spin ground state.⁴³ In particular, they examined the A-tensor obtained from Mössbauer measurements (A_x and A_y) and computations (A_z) to determine if there is an anisotropic component that arises from the spin-dipolar contributions, which cause one tensor component to be approximately twice the magnitude of the other two. For intermediate J these A-values are (A^SD = (+5.4, +5.4, −11.9) T), indicative of an S = 2 ground spin state with an unoccupied dz² orbital (Note: the A^SD tensor for an S = 1 Fe^IV=O complex is similar but the signs on the A-values are reversed). These magnetic studies were followed by corroborative reports from other spectroscopic measurements on intermediate J. Extended X-ray absorption fine structure (EXAFS) spectroscopy revealed a short Fe–O distance of 1.62 Å which supports the assignment of an Fe^IV=O unit,⁴⁴ and resonance Raman studies found an Fe–O vibrational feature at 821 cm⁻¹ that shifts to 787 cm⁻¹ upon substitution with ¹⁸O₂.⁴⁵ This shift of 34 cm⁻¹ is consistent with that predicted using a harmonic Fe–O oscillator model.

Since the first report of intermediate J in TauD, many other α-KG/Fe-dependent enzymes have been investigated and found to form an Fe^IV–oxido species including intermediates that catalyze reactions other than substrate hydroxylation.^27,46 One example is halogenation, as observed in SyrB2, which is a non-heme Fe halogenase.^47,48 The active site of SyrB2 differs from that of TauD by replacing the aspartate amino residue with a halide anion in the primary coordination sphere of the Fe center. An Fe^IV–oxido species still forms and functions to abstract an H-atom from the substrate, but instead of “rebound” of the hydroxido ligand, the halide recombines with the carbon radical to form a C─X bond (Scheme 4). Bollinger and coworkers have demonstrated experimentally that substrate orientation is crucial in the chemoselectivity.^49-51 By positioning the C─H bond of the substrate further away from the oxido moiety but closer to the chloro ligand, the rate of C─H cleavage decreases, but the hydroxylation pathway is suppressed, and the halogen-rebound is preferred. Other reactions catalyzed by these non-heme enzymes include substrate desaturation,^52,53 epoxidation,⁵⁴ and decarbonylation.⁵⁵

Scheme 4.

Catalytic mechanism of aliphatic halogenation by SyrB2.

Synthetic Mononuclear Fe–Oxido Analogs: The Beginning.

The trapping of non-heme Fe^IV–oxido species within proteins inspired the preparation of synthetic analogs, yet stabilizing these reactive species in abiotic systems was challenging. There was one example of a stable Fe–oxido complex that preceded its detection in proteins, but it was an Fe^III–oxido species. This complex, [Fe^IIIH₃buea(O)]²⁻ ([H₃buea]³⁻ = tris[(N'-tertbutylureaylato)-N-ethyl]aminato, Figure 1A) was generated from the activation of dioxygen by an Fe^II precursor, and spectroscopic and structural characterizations confirmed its formulation that includes a rather long Fe–O bond length of 1.816(3) Å.⁴¹ Prior to the discovery of [Fe^IIIH₃buea(O)]²⁻ all Fe^III–oxido units were incorporated into bridging structures with other metal ions, though most are with other Fe^III centers.^56,57 These species arise because of the strong thermodynamic driving force to form Fe^III–(μ-O)–Fe^III units, which are the basic unit in rust. Formation of multinuclear products is prevented in [Fe^IIIH₃buea(O)]²⁻ because of the [H₃buea]³⁻ ligand which controls both the primary and secondary coordination spheres.^58,59 Specifically, experimental and computational studies have directly linked the stability of this complex to the three intramolecular hydrogen bonds (H-bonds) involving the oxido ligand and the NH groups of [H₃buea]³⁻.⁶⁰ This complex is rare, and H-bonds appear to be needed for isolation as the only other two that are fully characterized in the condensed phase, which are reported by Fout (Figure 1B) and Agapie (Figure 1C), also have an intramolecular H-bonding network around the Fe^III–oxido unit.^61,62

Figure 1.

Terminal Fe^III–oxido complexes: [Fe^IIIH₃buea(O)]²⁻ (A), [N(afa^Cy)₃Fe^III(O)]⁺ (B), and [LFe₃O(PzNHtBu)₃Fe(O)]⁺ (C).

S=1 Fe^IV–Oxido Complexes.

The detection of a synthetic Fe^IV–oxido complex was prompted by Wieghardt's observation of a new Fe^IV species produced from the ozonolysis of [(cyclam-acetato)Fe^III(OTf)]⁺ (cyclam = 1,4,8, 11-tetraazacyclotetradecane).⁶³ The magnetic properties obtained by Mössbauer spectroscopy, such as a low isomer shift and an anisotropic A^SD tensor, were comparable to those of compound II in horseradish peroxidase and can best describe an S = 1 Fe^IV center. However, the low yielding generation of this species precluded further structural and spectroscopic characterizations. The breakthrough in preparation of synthetic Fe^IV–oxido complexes came from the laboratories of Nam and Que who reported the first crystallographically characterized Fe^IV-oxido complex in [Fe^IV(O)(TMC)(MeCN)]²⁺ (TMC = tetramethylcyclam, Figure 2A) and later followed that with the structure of [Fe^IV(O)(N4Py)]²⁺ (N4Py = N,N-bis(2-pyridylmethyl)-N-bis(2-pyridyl)methylamine, Figure 2B).^64,65 The structural analyzes of these complexes showed relatively short Fe─O bond lengths of 1.646(3) and 1.636(3) Å that are indicative of Fe^IV–oxido species. There have been numerous spectroscopic and computational studies on these complexes that corroborate their assignments as Fe^IV species, but they both have spin ground states of S = 1 which differ from those found in proteins.^20,28,66 Nevertheless, these complexes serve as important contributions to the field and there is now a library of over 100 examples of S = 1 Fe^IV–oxido complexes in similar ligand frameworks.^67-76 They have been particularly useful in delineating properties of electronic structure and demonstrating how optical spectroscopy is a reliable spectroscopic handle for Fe^IV–oxido complexes. Signature features of synthetic mononuclear Fe^IV–oxido species are weak bands around 700 – 900 nm that are assigned to a d-d transition.^66,67 The energy of the d-d transition is sensitive to the primary coordination sphere around the Fe^IV–oxido unit, both to the equatorial ligands within the N-ligand frameworks,⁶⁷ or the exogenous ligand trans to the oxido ligand.^77,78

Figure 2.

The first reported S = 1 Fe^IV-oxido complexes: [Fe^IV(O)(TMC)(MeCN²⁺ (A) and [Fe^IV(O)(N4Py)]²⁺ (B).

Biologically Relevant S = 2 Fe^IV–Oxido Complexes.

For all the success of Fe^IV–oxido complexes discussed above, their spin states differ from those found in proteins – this difference is important because electronic structure is often correlated with reactivity.^79-81 Alternative design concepts were thus needed; one approach is to change the coordination geometry around the Fe(IV) center. The rationale for this approach is rooted in the fundamental concept of bonding within a M–oxido unit. Most Fe^IV–oxido complexes have tetragonal symmetry, which favors an S = 1 spin state because of the arrangement of the d orbitals (Figure 3). These insights have been known for over 50 years and come from the seminal work of Gray who first used molecular orbital theory to describe the bonding in M–oxido complexes.^82,83 Gray's work was only for complexes in tetragonal symmetry and formed the basis for the Oxo-Wall concept.^9,10 Although many have made claims of making complexes that "broke" this wall, most were incorrect or lack sufficient experimental evidence, making the topic still controversial.^84-86 Within the context of Fe^IV–oxido complexes, a change in the coordination geometry to lower symmetry could afford high spin species. As first pointed out by Mayer and Thorn, C₃-symmetric M–oxido complexes are good candidates to stabilize high spin species because this symmetry mandates that the manifold of d orbitals contains two doubly degenerate E states (Figure 3).⁸⁷

Figure 3.

Qualitative d-orbital splitting diagrams for Fe^IV–oxido complexes with tetragonal (strong- and weak-field), trigonal, and tetrahedral symmetries.

This concept proved successful in preparing high spin Fe^IV–oxido species and there is now a small group of well-characterized trigonal bipyramidal Fe^IV–oxido complexes with S = 2 spin ground states. Members of this group use tripodal ligands to enforce the C₃-symmetry and include the following complexes, [Fe^IV(O)(TMG₃tren)]²⁺, [Fe^IVH₃buea(O)]⁻, [Fe^IV(O)(tpa^Ph)]⁻, and [Fe^IVpoat(O)]⁻ (Figure 4), where TMG₃tren = 1,1,1-tris{2-[N²-(1,1,3,3-tetramethylguanidino)]ethyl}amine, tpa^Ph = tris(5-phenylpyrrol-2-ylmethyl)amine), and [poat]³⁻ = N,N’,N’-(nitrilotris(ethane-2,1-diyl))tris(P,P-diphenylphosphinic amido)).^88-91 While Mössbauer spectroscopy confirmed the high-spin configuration for these systems by the anisotropic character of the A^SD-tensor (see above), an S = 2 signal was also observed for some of these complexes using parallel-mode EPR spectroscopy to independently verify the spin state.^42,89,90,92 Additionally, [Fe^IV(O)(TMG₃tren)]²⁺ and [Fe^IVH₃buea(O)]⁻ were crystallographically characterized, with Fe─O bond lengths of 1.661(2) and 1.680(1) Å, respectively.^89,93 From the works of Solomon who performed in-depth investigation into the electronic structure of [Fe^IV(O)(TMG₃tren)]²⁺, we know the frontier molecular orbitals to be π*(xz, yz), which leads to the highly reactive nature of the complex in comparison to less reactive S = 1 Fe^IV–oxido complexes.⁹⁴ In [Fe^IVH₃buea(O)]⁻, the frontier molecular orbitals are dxy and dx²-y² which are non-bonding with respect to the Fe–oxido unit, which is a consequence of the strong ligand field provided by the deprotonated urea donors within the trigonal plane.⁴²

Figure 4.

Trigonal Fe^IV–oxido complexes with S = 2 spin ground states: [Fe^IV(O)TMG₃tren]²⁺ (A), [Fe^IVH₃buea(O)]⁻ (B), [Fe^IV(O)tpa^Ph]⁻ (C), and [Fe^IVpoat(O)]⁻ (D).

The generation of synthetic Fe^IV–oxido complexes allowed for further exploration into the bonding within the Fe^IV–oxido unit. Within a broader context, there has been much debate on whether Mⁿ–oxido units, or its valence tautomer with M^n-1–O · (M^n-1–oxyl) units, are the active species during catalysis.⁹⁵ Discussions on the relevance of M^n-1–O · species are driven by computational studies but there is little experimental support for these assignments, especially for complexes with Fe and Mn centers.^96-98 The parallel-mode EPR signal of [Fe^IVH₃buea(O)]⁻ gave us the opportunity to experimentally address this question within an Fe^IV–oxido unit. The experiment was accomplished by measuring the broadening of the EPR signal at g = 8 of [Fe^IVH₃buea(¹⁷O)]⁻ (where I = 5/2 for an ¹⁷O nucleus).⁴² Substantial spin polarization was found on the oxido ligand as indicated by a ρ_p = 0.56 spins, where ρ_p is the spin population of the p orbitals on the oxido ligand. We have interpreted this large value of spin polarization on the oxido ligand to indicate significant covalency of the Fe–oxido bond, which is consistent with the assignment of an Fe^IV center from Mössbauer data (see above). This type of experiment has not yet been carried out on other non-heme Fe^IV–oxido systems, but DFT reported values of ${\rho }_P^{DFT}$ of 0.63 for [Fe^IV(O)TMG₃tren]²⁺ and 0.6 for intermediate J agree with our experimental findings.^43,88 Finally, we point out that if [Fe^IVH₃buea(O)]⁻ were to contain an oxyl radical, an oxygen-17 hyperfine interaction of nearly one full spin would have been observed on the oxido ligand (ρ_p ≈ 1).

The above examples of S = 2 Fe^IV–oxido species utilized tripodal ligands to promote C₃ symmetry and high spin iron centers — an alternative approach to their preparation is to weaken the ligand fields in complexes with tetragonal geometry (Figure 3). The work of Que highlights this approach: he has studied the tuning of spin-state on Fe(II) and Fe(III) complexes in various TPA-derived ligands (TPA = tris(2-methylpyridine)amine) and discovered incorporation of bulky substituents at the 6-position of the pyridine donors lengthens the Fe–N_py bond and promotes high spin electron configuration.⁹⁹ This observation led to a comparison of the tetragonal [Fe^IV(O)(L)(MeCN)]²⁺ complexes (L = TPA (tris(2-pyridylmethyl)amine), QBPA ((2-quinolylmethyl)bis(2-pyridylmethyl)amine), and TQA (tris(2-quinolylmethyl)amine), Figure 5), in which the pyridines in TPA were systematically replaced by the bulkier quinolines (QBPA contains 1 quinoline and 2 pyridines, TQA contains 3 quinolines).^100-102 While [Fe^IV(O)(TPA)(MeCN)]²⁺ and [Fe^IV(O)(QBPA)(MeCN)]²⁺ adopt the standard low-spin configuration, [Fe^IV(O)(TQA)(MeCN)]²⁺ is S = 2. Importantly, this study showed that a high-spin Fe^IV-oxido unit can exist in a tetragonal geometry within an appropriate ligand field.

Figure 5.

Systematic weakening of ligand field by incorporation of steric bulk in [Fe^IV(O)(TPA)(MeCN)]²⁺ (A), [Fe^IV(O)(QBPA)(MeCN)]²⁺ (B), and [Fe^IV(O)(TQA)(MeCN)]²⁺(C).

Role of Non-Covalent Interactions.

The [Fe^IVH₃buea(O)]⁻ (Figure 4B) complex also has a key structural difference when compared to most other Fe^IV–oxido complexes because of its intramolecular H-bonding network. Similar to the molecular structure of its Fe^III–oxido analog, [Fe^IVH₃buea(O)]⁻ has the potential to have three H-bonds to the Fe^IV–oxido unit. While H-bonds are prevalent within the active sites of many metalloproteins, there is still limited experimental data on how they affect the physical and chemical properties of metallocofactors. Insights into their effects for Fe^IV–oxido complexes came from a comparative study of [Fe^IVH₃buea(O)]⁻ and [Fe^IVpoat(O)]⁻ (Figure 4D), a complex with a tripodal ligand with phosphinic amido groups.⁹¹ Compared to the structure of [Fe^IVH₃buea(O)]⁻ the [poat]³⁻ ligand produced a similar ligand field and coordination geometry to the Fe center, but [Fe^IVpoat(O)]⁻ lacked the ability to form intramolecular H-bonds within the secondary coordination sphere. The experimental findings showed a significant weakening of the Fe=O bond in [Fe^IVH₃buea(O)]⁻ which was directly correlated to the presence of intramolecular H-bonds with the oxido ligand. Moreover, these studies showed how H-bonds affected the overall electronic structures in which the energies of the π*(xz, yz) orbitals are particularly sensitive. In the absence of H-bonding groups, [Fe^IVpoat(O)]⁻ displays similar spectroscopic characteristics and instability as [Fe^IV(O)(TMG₃tren)]²⁺.^88,94

The effects of electrostatic interactions have also been explored with Fe^IV–oxido complexes. In many cases, the reactivity of the Fe^IV–oxido complexes was enhanced upon addition of a Lewis acid which usually is a redox-inactive metal ion (e.g. Ca²⁺ and Sc³⁺). One notable example is the work of Fukuzumi and Nam who showed that the electron transfer rate in [Fe^IV(O)N4Py]²⁺ is accelerated by a factor of 10⁸ in the presence of redox-inactive metal ions (Figure 6A).¹⁰³ We have found that the rate of O₂ activation by [Fe^IIMST]⁻, a complex with a sulfonamido tripodal ligand ([MST]³⁻ = N,N’,N”-[2,2’,2”-nitrilotris(ethane-2,1-diyl)]tris(2,4,6-trimethylbenzenesulfonamido)), is correlated with the Lewis acidity of group 2 metal ions used to produce heterobimetallic complexes (Figure 6B).¹⁰⁴ These studies, and several others, demonstrate the influential and functional roles Lewis acids play in the chemistry of Fe complexes. However, an understanding of how Lewis acids interact with the Fe=O complexes to induce these functional changes is still lacking. In particular, there is little experimental evidence demonstrating that redox-inactive metal ions bond directly to the Fe^IV=O unit, which many have claimed is the principal reason for the changes in function.^103,105,106 Two cautionary examples are worth a mention. Following their work on electron transfer, Fukuzumi and Nam reported the molecular structure from XRD methods of a related heterobimetallic complex with an Fe^IV–(μ-O)–Sc^III core (Figure 6C), which was reasonably argued as the possible intermediate that gave rise to the increase in electron transfer rates.¹⁰⁷ However, a subsequent computational study by Swart suggested that this structure was incorrectly assigned and should be formulated as an Fe^III–(μ-O)–Sc^III species.¹⁰⁸ Que has repeated the experimental work and his findings agreed with those from the computational studies.¹⁰⁹ Another possibility that is rarely discussed in the literature is that in the presence of (adventitious) water, the redox-inactive metal ions can function as simple Brønsted acids, [M(OH₂)_m]ⁿ⁺, that protonate the Fe^IV–oxido complexes rather than directly bind. This point was recently illustrated in a study reported by Jackson who examined the effects of Sc³⁺ and Al³⁺ ions on the reactivity of [Mn^III(OH)(dpaq)]⁺ ([dpaq]⁻ = 2-(bis(pyridine-2-ylmethyl))amino-N-quinolin-8-ylacetamidate, Scheme 5).¹¹⁰ While addition of trivalent metal ions to the Mn^III–OH complex showed no significant spectroscopic perturbations, the resulting species demonstrated greatly enhanced reactivity towards hydrocarbons. These findings were comparable with the spectroscopic and reactivity properties of [Mn^III(OH₂)(dpaq)]²⁺,¹¹¹ which suggest that addition of Sc³⁺/Al³⁺ ions results in protonation of the Mn^III–OH complex, not metal binding.

Figure 6.

Effects of electrostatic interactions on metal-oxido and -hydroxido complexes. Addition of LAⁿ⁺ accelerates the rate of electron transfer for [Fe^IV(O)(N4Py)]²⁺ by up to 10⁸-fold (A). [(15c5)⊃LA^II-(μ-OH)-Fe^IIIMST]⁺ (LA = Ca, Sr) (B). The previously characterized [(TMC)Fe^IV–O–Sc^III(OTf)₄(OH)] was reformulated as [(TMC)Fe^III–O–Sc^III(OTf)₄(OH₂)] (C).

Scheme 5.

Addition of Sc^III or Al^III ions to [Mn^III(OH)(dpaq)]⁺ resulted in the protonation of the complex.

We have approached this problem differently by using an auxiliary binding site within the secondary coordination sphere of our [Fe^IVpoat(O)]⁻ complex to coordinate a redox-inactive metal ion. The goal was to pinpoint where the redox-inactive metal ion was coordinated and to examine whether it had any effects on the properties of the Fe^IV–oxido unit. The [Fe^IVpoat(O)]⁻ complex was designed for Lewis acids to bind in close proximity to the Fe^IV–oxido unit and our spectroscopic findings are consistent with a direct interaction between a redox-inactive metal ion and the oxido ligand (Figure 7).⁹¹ Thus, we were able to measure the changes in electronic and structural properties of the Fe^IV=O unit caused by the binding of Lewis acids, and our experimental and computational studies showed that the Lewis acid adducts modulate the strength of the π-bond within the Fe^IV=O unit. We could systematically tune the magnitude of the changes through the binding of Mg²⁺, Ca²⁺, Sr²⁺, Ba²⁺ ions to [Fe^IVpoat(O)]⁻. For instance, the electronic properties of [Fe^IVpoat(O)---Mg^II]⁺ varied significantly from those of [Fe^IVpoat(O)]⁻, but were identical to [Fe^IVH₃buea(O)]⁻ which used H-bonds to regulate the secondary coordination sphere. These studies allowed for the quantitative comparison of the effect of different types of non-covalent interactions to the Fe^IV–oxido moiety and illustrated that they can indeed have a strong impact on the properties of metal-oxido complexes.

Figure 7.

Binding and electrostatic effect of redox-inactive Group 2 metal ions on [Fe^IVpoat(O)]⁻.

Structural characterization of the series of [Fe^IVpoat(O)---LA^II]⁺ complexes was unsuccessful due to their thermo-instability; however, we recently prepared and solved the molecular structure of the analogous Fe^III-(μ-OH)–Mg^II species that provided useful structural insights. Addition of NMe₄OAc to the starting synthon K[Fe^IIIpoat(OH)] in CH₂Cl₂ resulted in precipitation of the insoluble KOAc. The reaction mixture was then filtered and treated with [Mg^II(TMTACN)(OTf)₂] (TMTACN = 1,4,7-trimethyl-1,4,7-triazacyclononane) to produce the desired [(TMTACN)Mg^II–(μ-OH)–Fe^IIIpoat]OTf complex (Figure 8A). Multiple recrystallizations provided single crystals suitable for X-ray diffraction that revealed a discrete bimetallic [Fe^III–(μ-OH)–Mg^II] core (Figure 8B). The Fe^III center adopts a trigonal bipyramidal geometry, comprising of the [poat]³⁻ ligand and an hydroxido ligand; the Mg^II site retains the tridentate TMTACN capping ligand, and binds to the [Fe^IIIpoat(OH)]⁻ complex via two O-atom donors from the phosphinic amido groups of [poat]³⁻ and the bridging hydroxido ligand, as proposed in the series of [Fe^IVpoat(O)---LA^II]⁺ complexes. Notably, binding of the Mg^II adduct is consistent with a Mg1–O1 distance of 1.983(3) Å. The remaining phosphinic amido arm that does not coordinate to the Mg^II ion participates in intramolecular H-bonding with the bridging hydroxido ligand, with an O1⋯O2 distance of 2.661(3) Å.

Figure 8.

Preparative route to [(TMTACN)Mg^II–(μ-OH)–Fe^IIIpoat]OTf from K[Fe^IIIpoat(OH)] (A). Thermal ellipsoid diagram depicting the molecular structure of [(TMTACN)Mg^II–(μ-OH)–Fe^IIIpoat]⁺ determined by X-ray diffraction (B). Ellipsoidsc are shown at 50% probability level, and only the hydroxido H atom is shown for clarity. Selected bond distances (Å) and angles (deg): Fe1–O1, 1.892(2); Fe1–N1, 2.236(3); Fe1–N_eq,avg, 2.012(3); O1⋯O2, 2.661(3); Mg1–O1, 1.983(3); Mg1–O3, 2.015(3); Mg1–O4, 2.037(3); Mg1–N_TMTACN,avg, 2.257(3); O1–Fe1–N1, 175.74(11); Fe1–O1–Mg1, 120.09(12).

Reactivity

C─H Bond Cleavage.

There have been many studies on the reactivity of synthetic Fe^IV–oxido complexes with organic substrates to model the C─H bond cleavage step found in non-heme monooxygenases. We have summarized some of those findings in Table 1 which also includes the rate of TauD. Fe^IV–oxido complexes with S = 2 spin ground states were calculated to be more reactive towards C─H bonds than their S = 1 counterparts.^79-81 However, no significant trend can be found when comparing experimentally obtained reactivity data of complexes with different ground spin states. The S = 1 [Fe^IV(O)TMC(MeCN)]²⁺ displays sluggish reactivity towards 9,10-dihydroanthracene (DHA,¹¹² BDE_C─H = 78 kcal mol⁻¹) and 1,4-cyclohexadiene (CHD,¹¹³ BDE_C─H = 77 kcal mol⁻¹),^88,114 while the similarly thermostable S = 1 [Fe^IV(O)N4Py]²⁺ demonstrated comparable, if not better, substrate reactivity than the S = 2 [Fe^IV(O)TMG₃tren]²⁺, [Fe^IVH₃buea(O)]⁻, and [Fe^IV(O)tpa^Ph]⁻complexes.^88,90,115 The diminished or absent substrate reactivity of the latter three complexes could be attributed to the steric bulk provided by the ligand frameworks. On the other hand, the S = 2 [Fe^IV(O)(TQA)(MeCN)]²⁺ displayed extraordinary reactivity towards DHA and cyclohexane (BDE_C─H = 99.3 kcal mol⁻¹),¹¹⁶ which is comparable with taurine oxidation by TauD-J after correction for the temperature difference.³⁷ It is obvious that more mechanistic work is needed to understand the connections between spin states and C─H bond cleavage.

Table 1.

Reactivity of synthetic and enzymatic Fe^IV–oxido complexes with C-H substrates.

Spin state	Complex	Substrate Reactivitya	References
1	[FeIV(O)TMC(MeCN)]2+	DHA (0.016) CHD (0.018)	88,114
	[FeIV(O)TPA*(MeCN)]2+b	DHA (0.042)	117
	[FeIV(O)N4Py]2+	DHA (2) CHD (1.3)	88
2	[FeIV(O)TMG3tren]2+	DHA (0.090) CHD (1.2)	88
	[FeIVH3buea(O)]−	No reactivity	115
	[FeIV(O)tpaPh]−	CHD (1.4)	90
	[FeIV(O)(TQA)(MeCN)]2+	Cyclohexane (0.37, −40 °C) DHA (200, −80 °C)	102
	TauD-J	Taurine (kobs = 13 s−1, 5 °C, aqueous)	37

^aUnless specified, rate constants (k₂, M⁻¹ s⁻¹) were measured in MeCN at −30 °C.

^bTPA* = tris(4-methoxy-3,5-dimethylpyridyl-2-methyl)amine.

Rebound Step and Fe^III–OH Species.

The proposed mechanisms for hydroxylation of substrates include a rebound step whereby a transient carbon radical combines with an Fe─OH species to produce a new C─OH bond (Schemes 2, 3).^25,36 Synthetic Fe^III–OR species that mimic the rebound step for the α-KG dependent, non-heme Fe proteins have been developed. Goldberg first reported the synthesis of an Fe^III–OCH₃ complex, [Fe^III(N3PyO^2Ph)(OCH₃)]⁺, ((N3PyO^2Ph)⁻ = 2-(((bis(6-phenylpyridin-2-yl)methyl)(pyridin-2-ylmethyl)amino)methyl)phenolate) and its reactivity towards trityl radical (Ph₃C•) to produce [Fe^II(N3PyO^2Ph)(Solv)]⁺ and Ph₃OCH₃.¹¹⁸ Further mechanistic studies showed a concerted rebound process, with no initial charge transfer.¹¹⁹ Goldberg and Fout then independently demonstrated different Fe^III–OH complexes that readily react with trityl radical to produce quantitative Ar₃COH (Scheme 6A, B).^120,121 Goldberg also investigated the selectivity of hydroxido vs. halogen rebound using a library of non-heme Fe complexes with the [BNPA^ph2O]⁻ (2-(bis((6-(neopentylamino)pyridin-2-yl)methyl)amino)-1,1-diphenylethan-1-olate) ligand framework: Fe^III(OH)(X) and Fe^III(X)₂ (X = Br, Cl).¹²² He showed that reactivity of Fe^III(OH)(X) with trityl radical produced near-quantitative Fe^II(X) and Ar₃COH, widi no Ar₃CX detected (Scheme 6C). Moreover, the Fe^III(X)₂ complex does react with trityl radical to produce near-stoichiometric Fe^II(X) and Ar₃CX (Scheme 6D). The preference of Fe^III(OH)(X) towards hydroxylation is tentatively attributed to a thermodynamic factor, for which OH-rebound is calculated to be more favorable than halogenation by > 15 kcal/mol.^123,124 Factors of selective halogenation by halogenase enzymes such as SyrB2 remain to be investigated.

Scheme 6.

Synthetic non-heme Fe(III) complexes demonstrate rebound reactivity of hydroxido and halide ligands with trityl radical substrates.

The work of Goldberg and Fout illustrated that a rebound step is chemically possible; however, Shaik and Nam have questioned whether this step is mechanistically viable in all systems.¹²⁵ They offered an alternate mechanistic possibility in which the substrate radical dissociates and reacts with another Fe^IV=O complex to produce an Fe^III–OCR species. They support this premise with theoretical calculations that predicted dissociation of the substrate radical that is formed after C─H bond cleavage is more favorable than Fe^III–OH rebound. Moreover, they offer experimental evidence using synthetic complexes. This idea is intriguing and offers a reasonable explanation for observations found with synthetic Fe^IV=O complexes. What is less clear is whether this proposal is relevant within the active site of non-heme metalloenzymes where the local environments around the Fe centers are more controlled than in synthetic systems and likely prevent dissociation of substrate radicals.

Rebound Step and Fe^IV–OH Species.

There is strong evidence for this step from studies using P450s with the key intermediate being compound II which is an Fe^IV–OH species — however, the role that this type of high valent metal-hydroxido species in other enzymatic systems remains controversial. Green has found that compound II has a pK_a(OH) of 12, which is unusually high for comparable metal–hydroxido species which have estimated pK_a(OH) values of less than 5.^26,126 He has argued that the difference is because compound II has an axially coordinated cysteine thiolate which is trans to the hydroxido ligand and elevates the pK_a(OH) value.

Other heme proteins with axial ligands containing N-atom donors produce intermediates that should not be protonated under normal conditions based on known thermodynamic information.^127-131 The difficulties in producing Fe^IV–OH species can be understood from the simple equilibrium expression (eq 1) which favors the products. In general, terminal oxido ligands are electrophiles and are weakly basic, which makes protonation thermodynamically unattainable.^10,126,132 For P450s, and other heme proteins with an axially coordinated thiolate, the strongly covalent Fe─S bond weakens the bonds within the Fe^IV=O unit, shifting the equilibrium to favor protonation. Other biologically relevant axial ligands do not appear to form Fe─X bonds that are sufficient to change the basicity of the oxido ligand; hence, to our knowledge there are no other biological examples, in both heme or non-heme proteins, where there is definitive experimental proof of an Fe^IV–OH species.

$$[{\text{Fe}}^{\text{IV}}-\text{OH}{]}^{+}\rightleftarrows \stackrel{K_a}{{\text{Fe}}^{\text{IV}}}=\mathrm{O}+{\mathrm{H}}^{+}$$ (1)

Our knowledge of non-heme Fe^IV–OH species is still incomplete, and we question whether they can be generated in a biologically relevant context. We make these points recognizing that there are a considerable number of mechanistic proposals that include Fe^IV–OH species as essential intermediates.^133-135 Intuitively Fe^IV–OH species could be considered reasonable intermediates, but there is little experimental evidence to support their formation and, furthermore, there is no thermodynamic evidence (e.g., K_a values in eq 1) to validate their stability. Synthetic verification of Fe^IV–OH species has also been difficult to achieve. Efforts to model compound II led Goldberg to prepare and structurally characterize a formally Fe^IV–OH species in a corrole framework, [Fe(OH)(ttppc)] (H₃ttppc = tris(2,4,6-triphenyl)-phenyl corrole, Figure 9A).¹³⁶ Its ability to react with a carbon radical substrate to form a new C─OH bond makes it a suitable synthetic model for compound II. However, the oxidation assignments of the Fe center and corrole ligand are unclear, and the species can only be accurately interpreted as one oxidizing equivalent above Fe^III–OH. Although the Mössbauer spectrum of [Fe(OH)(ttppc)] was reported, further variable-field experiments, as well as X-ray absorption spectroscopy, are required to elucidate the oxidation state of the Fe center.

Figure 9.

Synthetic attempts for Fe^IV-hydroxido complexes: [Fe(OH)(ttppc)] (A), [(TAML)Fe^IV-OH_n]^n-2 (B, n = 1 or 2), and [Fe^IVH₄buea(O)] (C).

Hendrich has attempted to generate and study an Fe^IV–OH species by subjecting the complex [(TAML)Fe^III–OH₂]⁻ ([TAML]⁴⁻ = [Me₄C₂(NCOCMe₂NCO)₂CMe₂⁴⁻) to chemical oxidation or electrolysis.¹³⁷ A new species was formed, and its EPR and Mössbauer parameters are consistent with a high-spin Fe(IV) center. However, the authors cannot definitively assign the Fe^IV species to be [(TAML)Fe^IV–OH]⁻ or [(TAML)Fe^IV–OH₂] (Figure 9B). No Fe–X vibrational features > 600 cm⁻¹ were observed for this species, and the features observed between 447–475 cm⁻¹ were proposed to be Fe–O vibrations, weakened due to protonation of the oxido ligand, and hydrogen-bonding with the bulk aqueous solvent. However, these features are comparable or weaker in energy than those of a series of Fe^III–OH complexes we have developed (477–594 cm⁻¹), which does not support the assignment for an Fe^IV–OH species.¹³⁸ Additionally, electrochemical experiments on [(TAML)Fe^III–OH₂]⁻ revealed a Nernstian one-electron redox event, which suggests oxidation does not result in geometric rearrangement or ligand modification, posing the possibility that an Fe^IV–OH₂ species could be produced.

We have approached this problem with experiments to protonate [Fe^IVH₃buea(O)]⁻ (Figure 4B): a thermodynamic study provided an estimated pK_a of ~11 for its conjugate acid, [Fe^IVH₃buea(O)(H)], which is a similar value to that found for compound II in P450s.²⁶ Treating [Fe^IVH₃buea(O)]⁻ with acids produced a new complex with distinct optical and magnetic properties.¹³⁹ This complex reverted back to [Fe^IVH₃buea(O)]⁻ upon the addition of base. Moreover, this protonated species could be prepared from treatment of [Fe^IVH₃buea(O)]⁻ with an oxidant which is consistent with initial formation of a formally Fe^V=O species that further reacts through a similar mechanism as proposed for P450s to produce a species similar to compound II. However, the vibrational and structural properties do not support the formation of Fe^IV–OH species and computational studies suggest that the [H₃buea]³⁻ ligand is the site of the protonation (and not the oxido ligand, Figure 9C). These results demonstrate that care should be taken in assigning the protonation state of an oxido ligand in high valent Fe complexes, especially when other sites of protonation are present such as a basic site within a protein active site or synthetic ligand.

Use in Chemical Synthesis.

Because of the utility of Fe-oxido as the active oxidant in nature, much attention has turned to the environmentally green Fe/O₂ (or H₂O₂, bleach) system for chemical processes. While the identity of the catalytically competent Fe species in many instances is not fully elucidated, this has led to the discovery of many new chemical transformations. Costas and White have independently made important contributions using molecular iron precursors in tetradentate N-based ligand frameworks, but most work is based on the N₂Py₂ scaffold, in which two pyridines are attached via a diamine linker.^140-142 They have developed various catalytic processes that include those for hydroxylation, dihydroxylation, and epoxidation and have modified these catalysts to improve site- and chemo-selectivities. They have shown that increasing the rigidity of the diamine backbone,^143-145 as well as protecting sites that are prone to oxidation on the ligand (e.g. by deuteration of the benzylic hydrogens),^146,147 can improve the robustness of the catalysts. Costas has further found that modulating the electronic and steric properties of the pyridine arms can have significant effect on chemoselectivity. For instance, installation of more electron donating groups at the γ-pyridine position increases the preference and yield for olefin epoxidation (Figure 10A),¹⁴⁸ while alkyl substitution at the α-position promotes the selectivity for olefin syn-dihydroxylation (Figure 10B).¹⁴⁹ Using a more supramolecular approach, it was found that installation of directing groups in the catalyst/substrate through covalent or non-covalent interactions can improve regioselectivity. The work of Costas illustrated this concept when he incorporated benzocrown-6-ethers onto [Fe^II(PDP)]²⁺ complexes that served as receptors for alkyl substrates with protonated primary amines (Figure 10C).¹⁵⁰ Upon treatment with H₂O₂, the complex can selectively perform C─H hydroxylations at the C8 and C9 positions of the alkyl chain; this regioselectivity is attributed to controlling the orientation of the substates via docking the R-NH₄⁺ within the internal crown ether. The docking of the alkyl substrates to promote specificity is reminiscent of how proteins increase selectivity through positioning of a C─H bond near reactive Fe-oxido centers by orienting substrates within an active site using H-bonds.^151,152 White has demonstrated the utility of late-stage site- and stereo-selective C-H bond functionalization of a variety of natural products using the catalyst [Fe^II(PDP)]²⁺ and [Fe^II(CF₃-PDP)]²⁺; the oxidized derivatives are compounds of pharmacological interests.¹⁴² For instance, selective hydroxylation of artemisinin can be achieved by [Fe^II(PDP)]²⁺ with comparable yields to the native P450 enzyme (Figure 10D).¹⁵³ White has also explored functionalization of unnatural amino acids using the same catalysts to provide new building blocks for oligopeptides and proteins.¹⁵⁴ To showcase a different type of reactivity of Fe–oxido complexes we turn to the work of Collins who has demonstrated the viability of using Fe catalysts in waste treatment industry (Figure 10E).^155,156 Using Fe-TAML (tetraamido macrocyclic ligands) complexes as catalysts and O₂/H₂O₂/NaOCl as the oxidant, he has been able to degrade explosives (e.g. trinitrotoluene, trinitrobenzene¹⁵⁷), endocrine disruptors (e.g. bisphenol A¹⁵⁸, nitrophenols¹⁵⁹), and other pollutants (e.g. commercial dyes^160,161, metaldehyde¹⁶²) and demonstrated that these systems can be adapted to an industrial setting. The identity of the competent oxidants for all these processes has often been proposed to be a high-valent Fe-oxido species; however, there is still little experimental evidence to verify these claims, leaving much to investigate.

Figure 10.

Representative work in Fe molecular catalysis in chemical synthesis. Electronic and steric modifications in the N₄-based ligands modulate chemoselectivity (A & B). Incorporation of supramolecular directing groups improve regioselectivity of C─H bond activation (C). Late-stage transformations of various natural products and their derivatives (D). Use of Fe-TAML catalysts in environmentally beneficial catalysis, such as waste treatment (E).

Mechanistic Considerations

Cleavage of C─H bonds by Metal-Oxido Species.

Within this Forum we have highlighted the utility of metal-oxido species for the cleavage of C─H bonds and we now address ideas on why they are able perform this function. There are two mechanistic camps that have guided our thinking on this process, one that emphasizes ground state thermodynamics,^1,12,163 while the other suggests that the radical character of the oxido ligand drives the reactivity. The attributes of both mechanistic proposals of C─H bond cleavage have been reviewed extensively and we will not duplicate those discussions here. We highlight the thermodynamic approach and how it can be applied to biological and synthetic C─H bond cleavage. This approach is rooted in physical organic chemistry^112,164,165 and adapted by Mayer for the cleavage of X─H bonds (X = C, N, O).^1,12 The central tenet is that the reactivity of a Mⁿ=O center can be evaluated by comparing the bond dissociation free energy (BDFE) of the C─H bond to be cleaved to that of the O─H bond formed in the resulting M^n-1–OH species. We note that the literature interchanges BDFE and BDE values with the older literature often using BDEs which is purely enthalpic.¹ The merits of using BDFEs values have been discussed by Mayer and the field has generally adopted its usage. In this Forums article, the terminology used by the authors of the articles to which we refer is used to most accurately report the data. Thermodynamic square schemes (Figure 11) are useful to illustrate the various connections between proton transfer (pK_a electron transfer (E_1/2), and BDFE. Square schemes also provide a convenient way to represent the three limiting mechanistic pathways for C─H bond cleavage: 1) stepwise proton transfer-electron transfer (PT-ET); 2) stepwise electron transfer-proton transfer (ET-PT); and 3) concerted proton electron transfer (CPET).

Figure 11.

Thermodynamic scheme for a metal-oxido complex.

The literature indicates that CPET pathways are the most common for reactions involving Fe^IV=O complexes and substrates with a C─H bond. This conclusion comes from numerous studies with synthetic Fe^IV=O species that show strong correlations between the log (k₂) and the BDFE_C─H of the substrates (where k₂ is the second-order rate constant of a reaction).^{1,15,76,90,115,166-171} Linear free energy relationships of this type follow from the Bell-Evans-Poyanyi (BEP) principle^172,173 which predicts that experimentally obtained rate constants increase with either a decrease in the BDFE_C─H of the substrate or the increase in energy released from forming the MO─H bond (that is, BDFE_O─H). These BDFE_O─H values, however, are missing from the literature – although there are reports for some systems, there are still too few examples that are derived from experiments. However, BDFE_O─H values can be determined from the E_1/2 value of the M=O species and the pK_a value of its conjugate acid, that is M–OH (eq 2), where the coefficients are derived from fundamental thermodynamic relationships and C_G is a constant that is dependent on the reference electrode and solvent.¹ The experimental difficulties are usually in determination of the pK_a value because these metal-hydroxido complexes are difficult to prepare (see above for Fe^IV–OH). Nevertheless, eq 2 illustrates the close relationship with reduction potential and pK_a values in determining the BDFE_O─H, which influences the overall driving force for the reactivity of a M=O species with C─H bonds.

$${\text{BDFE}}_{\mathrm{O}-\mathrm{H}}({\text{kcal mol}}^{\text{-1}})=(23.06{\text{kcal mol}}^{\text{-1}}{\mathrm{V}}^{\text{-1}})E{}^{{}_{°}}(\mathrm{V})+(1.37{\text{kcal mol}}^{\text{-1}})\mathrm{p}{K}_{\mathrm{a}}+{\mathrm{C}}_{\mathrm{G}}({\text{kcal mol}}^{\text{-1}})$$ (2)

The cleavage of C─H bonds is an oxidative process and chemical intuition would suggest that a stronger oxidant (i.e., M=O species) would increase the efficiency of a given reaction. However, highly positive potentials have the disadvantage of poor selectivity and/or irreversible destruction of proteins or ligands. Eq 2 shows that this condition can be circumvented – in fact, a less positive potential and a large pK_a value (more basic M=O species) can produce large BDFE_O─H values and reactive M=O species. This relationship can be shown graphically for the cleavage of the C─H bond in methane (Figure 12).¹⁷⁴ For a M=O species with a relatively low pK_a value of 5, the corresponding positive potential of 1.75 V vs NHE is large enough to oxidize the nearby aromatic side chains within the active site of a protein. When the pK_a value is raised to 17.5, the potential is lowered to ~1 V vs NHE which will inhibit off-path and destructive oxidations. Therefore, the basicity of the M=O unit is an important parameter that regulates reactivity toward C─H bonds.

Figure 12.

Relationship between reduction potential and pK_a for generic metal-oxido species in the homolytic cleavage of a C─H bond in methane with BDFE_C─H = 102 kcal/mol from eq 2. The vertical lines indicate regions of relatively high (blue) and low (red) potentials.

Basic M–Oxido Species.

This role of the basicity of the M=O unit has been studied in both proteins and synthetic complexes. Green was the first to show the relevance of this concept in the high valent Fe–oxido intermediate of cytochrome P450s (compound I). He demonstrated experimentally and computationally that the reduced intermediate of compound I is sufficiently basic to promote the abstraction of H-atoms from strong C─H bonds at relatively low one-electron reduction potentials. We have studied how the basicity of the metal-oxido unit in synthetic complexes can be used to regulate the cleavage of C─H bonds. In one study, we examined the reactivity of [Mn^IIIH₃buea(O)]²⁻ with the common test substrate 9,10-dihydroanthracene (DHA), and found clean conversion to anthracene and [Mn^IIH₃buea(OH)]²⁻, indicating C─H bond cleavage had occurred (Scheme 7).¹⁷⁵ However, the one-electron reduction potential for [Mn^IIIH₃buea(O)]²⁻ is quite negative and we were only able to estimate its value at less than −2.0 V versus [Fe^III/IICp₂]^+/0.¹³ This potential is in the range found for strong reductants such as cobaltocene and just based on its electrochemical potential should not be able to cleave C─H bonds. However, we measured the pK_a(OH) value of 28 in DMSO for [Mn^IIIH₃buea(OH)]⁻, the conjugate acid of the Mn^III–oxido complex.¹⁷⁶ This value indicates that the Mn^III–oxido unit in [Mn^IIIH₃buea(O)]²⁻ (its conjugate base) is sufficiently basic to compensate for the negative potential to promote reactivity.

Scheme 7.

Reactivity of [Mn^IIIH₃buea(O)]²⁻ towards 9,10-dihydroanthracene.

The idea that the basicity of a M–oxido unit is an important parameter suggests that it can be tuned to regulate reactivity toward C─H bonds. We tested this premise by developing a system that permitted us to systematically modulate the basicity of the Mn^III–oxido unit and determine whether these changes correlated with the rate of C─H bond cleavage. We redesigned [H₃buea]³⁻ into a series of new hybrid ligands [H₃bpuea-R]³⁻ (R = OMe, H, F, Cl, CF₃) in which one of the tripodal arms contains a para-substituted phenylurea unit, and we prepared the corresponding series of [Mn^IIIH₃bpuea-R(O)]²⁻ complexes (Figure 13).¹⁷⁷ These Mn^III–oxido complexes differed only in their H-bond involving the phenylurea groups – our experimental data showed that the changes in H-bond donor strength modulated by the R-group regulated the basicity of each complex. Moreover, we found a correlation between the pK_a(Mn^IIIOH) values and the second order rate constants for the reaction of the [Mn^IIIH₃bpuea-R(O)]²⁻ complex with DHA. Additional studies on these Mn^III–oxido complexes did not show any correlation between log(k₂) and the BDFE_C─H of DHA which would be characteristic of a synchronous CPET. However, our data also did not support a two-step process for rate-limiting PT followed by ET.

Figure 13.

The intramolecular H-bond donor strength in a series of [Mn^IIIH₃bpuea-R(O)]²⁻ complexes is modulated remotely by the R-group, which affects the basicity of the complex and the reaction rate towards 9,10-dihydroanthracene. The experimentally determined second order rate constant k₂ was corrected to k to account for 4 equivalents of C─H bonds that can possibly be cleaved in DHA.

A detailed description of the mechanism for this reaction is still not known, but it appears that it does not follow the BEP principle and a conventional CPET in which the proton and electron move together at the transition states (a synchronous process). We hypothesize that this process is still CPET but involves an asynchronous transition state in which PT precedes ET.^15,178 Similar asynchronous CPET (Figure 14A) has been suggested to occur in other systems and appears to be a new mechanistic description of C─H bond cleavage by metal–oxido species. The experimental work of Anderson on a unique Co^III–oxido complex provides strong evidence for a PT-dominated asynchronous CPET mechanism (Figure 14B),^179,180 as does work by Tolman on a Cu^III–O₂CAr complex (Figure 14C).¹⁸¹ The link between these complexes and [Mn^IIIH₃buea(O)]²⁻ is the high basicity of the metal–X unit (X = oxido, hydroxido, carboxylato) which seems to be characteristic of this type of mechanism. Moreover, computational studies by Srnec on Fe^IV=O complexes provide a theoretical framework to evaluate asynchronous processes and introduced an asynchronicity parameter to assess the contributions of PT and ET within transition states.¹⁸² However, additional experimental evidence is needed to evaluate whether this parameter will be useful in designing new reagents and understanding their mechanisms. It is obvious that the field would benefit from additional methods to analyze data to indicate which of the possible mechanisms is operative for a given system. This type of analysis has implications beyond garnering fundamental insights as it could establish a practical advantage for developing complexes toward chemical synthesis. Rather than using complexes that rely on BDFE_C─H to induce selectivity, M–oxido complexes that are sufficiently basic could lead to reagents with enhanced selectivity that favor targets based on acidity of the C─H bonds.

Figure 14.

Square scheme for asynchronous CPET (red arrows, (A)), Anderson’s Co^III-oxido (B), Tolman’s Cu^III-O₂CAr (C). Scheme in A is adapted from Anderson.¹⁷⁷

Some Remaining Challenges

Throughout this Forum article we presented topics on M–oxido and –hydroxido chemistry with an eye to highlight important accomplishments and point out where additional experiments might be needed. We end with two short discussions on areas that are not as complete, are somewhat controversial, have importance in chemistry and biology, and from our perspective are ripe for additional studies.

The Relevance of Fe^V=O in Biology and Beyond.

We have addressed many important aspects of Fe^IV-oxido and - hydroxido cofactors in metalloenzymes and synthetic analogues, but at this point one may wonder if higher valent analogous are known and useful. In heme systems there is no evidence that higher valent Fe systems are operative. The formation of the key oxidized intermediate occurs from the heterolytic cleavage of an Fe^III–OOH species to produce water and compound I, which is not an Fe^V=O species, but rather an Fe^IV=O unit coupled to a ligand-based radical (Scheme 2).²³ There is consensus supported from both computational and experimental studies for this assignment. These finding underline an important feature of active sites in metalloenzymes: instead of incorporating weakly-donating amino acid-based ligands or other scaffolds to support a thermodynamically unfavorable, high-valent metal ion (e.g. Fe^V), these enzymes are equipped to store additional oxidative equivalents on a redox-active organic moiety.^23,183

For non-heme systems the role of an Fe^V-oxido in both proteins and synthetic systems is controversial and the subject of continued debate.^184,185 Mechanistic proposals for the Rieske oxygenases provide an illustrative example of where there are concerns. Early mechanistic thoughts centered on the possible similarities between the O₂ activation pathways of these non-heme Fe monooxygenases and cytochrome P450s: O₂ binding was thought to result in heterolytic cleavage of an O─OH bond to generate a new Fe intermediate that is commonly proposed as an Fe^V(O)(OH) species (Scheme 8A).^133,186,187 However, there is still no experimental data to support the formation of an Fe^V-oxido species within Rieske proteins or for that matter, any metalloprotein. For Rieske proteins, recent investigations have supported alternative mechanisms that avoid the formation of high valent Fe intermediates. Solomon and Lipscomb investigated the native O₂ reaction in the Rieske enzyme benzoate 1,2-dioxygenase (BZDO), and determined in the native O₂ reaction an Fe^III-superoxo species reacts directly with the benzoate substrate without accessing higher valence states (Scheme 8B).¹⁸⁸ Additional experiments using a peroxide shunt route found that an Fe^III–OOH species is produced which also leads to product formation. This study again emphasizes nature’s ability to circumvent the need to access the high oxidation states by employing other mechanistic pathways: not only is it thermodynamically challenging to achieve an Fe^V species, but a highly reactive Fe^V cofactor could also lead to nonproductive pathways that can cause irreversible damage to the protein.

Scheme 8.

The proposed (A) and experimentally verified (B) mechanisms of benzoate 1,2-dioxygenase, a Rieske enzyme. An Fe^V intermediate was not observed.

The implication of Fe^V intermediates in non-heme enzymes has inspired attempts to prepare similar species within synthetic systems.^189-193 However, unlike the Fe^IV=O analogs, preparative difficulties in accessing Fe^V state, increased instability, and lack of structural data of proposed Fe^V complexes make the assignment of iron oxidation state challenging. Moreover, there is a general lack of identifiable spectroscopic features for non-heme Fe^V–oxido species that makes unambiguous assignments difficult. Two spectroscopic handles for Fe^V valence states have been proposed: 1) the Griffith-Taylor model is used to predict the g factors of S = 1/2 Fe^III centers, in which a large g-anisotropy is expected (g_max – g_min > 0.8), while an S = 1/2 EPR feature with a small g-anisotropy has been observed and proposed for low-spin Fe^V complexes;^192,194-196 and 2) anisotropy in the ⁵⁷Fe A-tensor is proposed to be a spectroscopic marker to identify 3d electron configuration and to assign Fe^V centers.¹⁸⁹ Recently, Ye and Neese performed detailed computational studies into several proposed Fe^V=O complexes, and concluded the x/y anisotropy of the A-matrix to be an unreliable spectroscopic handle for the assignment of the Fe^V state.¹⁹⁷ In this study, most of these reported Fe^V complexes are reformulated to have dominant Fe^IV character coupled with a ligand radical which is not unlike what occurs in cytochrome P450s. They only assigned [Fe^V(O)(TAML)]⁻ as a genuine Fe^V=O species, and stabilization of the high-valent Fe center is attributed to the tetraanionic ligand field.¹⁵⁵ The shortening of the Fe–O bond (1.58 Å) as well as a significantly more negative isomer shift (−0.42 mm/s) in [Fe^V(O)(TAML)]⁻ distinguish it from Fe^IV=O and other proposed Fe^V complexes.¹⁹⁰ Generation of an Fe^V-oxido species has also been proposed in the gas phase using helium tagging infrared photodissociation (IRPD) spectroscopy, a powerful technique used to measure IR spectra of mass-selected ions.¹⁹⁸ However, magnetic and redox properties of these species, which have proven essential for determining the oxidation level of Fe=O complexes, are not currently available in the gas phase to corroborate these claims.

For the above description it is apparent that the quest to prepare and characterize Fe^V–oxido complexes is still ongoing. Regardless if Fe^V=O species turn out to be relevant in biology, developing a complete understanding of their properties and reactivities is important in determining their use in chemical synthesis.^184,185 In many ways, the progress in Fe^V=O chemistry is reminiscent of where the field was 20 years ago in the pursuit of Fe^IV=O complexes – breakthroughs were made only after detailed investigations into their molecular/electronic properties which were coupled to vibrational data. The key lesson learned from these studies is that one technique alone cannot be used to make definitive assignments of oxidation state and structure, and computational findings need to be supported by the appropriate experimental results. We still have much to learn before consensus on Fe^V=O systems match that achieved for heme and non-heme Fe^IV=O complexes.

Single Site Cu Centers for C─H Bond Functionalization.

Research into the structure and function of metalloenzymes with single Cu centers has increased in the last 10 years, being spurred on by studies of particulate monooxygenases (pMMOs) and lytic polysaccharide monooxygenases (LPMOs). pMMOs have been known for some time^199-204 and have been reviewed previously^205-208 – our focus will be on LPMOs in which the involvement of Cu and O₂ has only been confirmed within the last dozen years.²⁰⁹ These metalloenzymes are commonly isolated from bacteria and fungi within compost heaps²¹⁰ and promote the degradation of biopolymers such as cellulose and chitin via oxidative cleavage of polysaccharide chains at either the C1 or C4 positions of the glucose unit (Scheme 9).^211,212

Scheme 9.

Substrate transformation performed by LPMOs on cellulose.

From an applied perspective, LPMOs have attracted much attention because of their role in enhancing the conversion of biomass to biofuels.^211,213-215 From a fundamental perspective, this transformation has intriguing mechanistic implications: it can be performed via activation of dioxygen in conjunction with exogenous electron and proton sources or with hydrogen peroxide. Dioxygen was reported to be involved in catalysis in 2010.²⁰⁹ Later in 2017, it was proposed that hydrogen peroxide could form in situ from dioxygen and then lose water resulting in a Cu^II-oxyl intermediate which abstracts a hydrogen atom from substrate and finally hydroxylates the polysaccharide.²¹⁶ Now evidence suggests that H₂O₂ can be used to initiate catalysis anaerobically.^217,218

The active site, which lies along the surface of the protein (Figure 15A), consists of a single Cu-center coordinated to three nitrogen atoms from two histidine side chains and the N-terminus amine (denoted the histidine brace, Figures 15B and C).^{211,212,215,219,220} While the histidine brace is highly conserved across LPMOs with some exception,²²¹ there are significant structural variations in the proteins from different types of organisms.²²² For example, the N-terminal histidine in fungal LPMOs, with exception in one organism,²²³ undergoes post-translational modification to insert a methyl group (Figure 15B).^212,223 This alteration has been shown to provide protection for the enzyme in environments required for function and reactive oxygen species produced in catalysis.²²³ Fungal LPMOs also have a tyrosine residue in the axial position ~2.7 Å away from the Cu ion (Figure 15B).²²⁰ This tyrosine is present in some bacterial LPMOs; however, many have a phenylalanine (Phe) occupying this position (Figure 15C).^224,225 Although it is generally agreed upon that the resting form of the enzyme has a Cu^II center, photoreduction often plagues crystallographic data which has made it difficult to interpret the metrical data. For example, photoreduction caused lengthening of the distances between water molecules and the metal in the structure of the bacterial LMPO (denoted AA10) from 2.6 and 3.2 Å, which is more consistent with a Cu^I center.^226,227 However, insight into substrate interaction with the protein has also been provided by X-ray diffraction studies. Crystallographic data has shown a cellotriose (G3) substrate docked over the face of the active site via H-bonding to water molecules and nearby amino acid residues (Figure 15D). Upon substrate docking, the axial water molecule is displaced and the Cu⋯O (Tyr164) distance decreases from 2.7 to 2.5 Å.²²⁰

Figure 15.

Crystallographic data for LPMOs. Surface representation of AA9 showing the protein in purple and the active site in pink (A), and the molecular structures of a fungal LPMO (AA9, PDB: 5ACG) (B), a bacterial LPMO (AA10, PDB: 6RW7) (C), and an AA9 in the presence of substrate analog (G3, pink, PDB: 5ACJ) (D). Secondary coordination sphere interactions are represented by dashed lines. Cu(II) is shown as a cyan sphere and Cu(I) is shown as a brass sphere.

The catalytic transformation performed by LPMOs requires dioxygen, two electrons, and two protons; however, the mechanism(s) that govern this reactivity is still debated.²²⁸ We will not discuss the merits of the various proposals in this Forum other than to mention that several possible reactive intermediates are proposed from computational studies that include Cu^II—O₂^•−, Cu^II—OOH, Cu^II—O^•, or a Cu^III—OH species.^229-236 At this juncture, there is little experimental evidence to support that these species are formed during turnover and it is obvious that more work is needed to validate which are relevant. It has also been found that the enzyme can function with H₂O₂ under anaerobic conditions.²¹⁶ The direct biological relevance of this discovery has sparked numerous reports. In one study, Marletta compared O₂ and H₂O₂ reactivity with the Cu^I intermediate in fungal LMPOs (denoted AA9s) and found that the reaction with O₂ gave regioselective substrate oxidation while the H₂O₂ reaction gave a faster rate with nonspecific reactivity consistent with formation of hydroxyl radicals.²³⁷ In a different investigation, Solomon found the catalytic rate was greatly enhanced (1000-fold) when H₂O₂ was used instead of O₂ in the reaction with a Cu^I AA9 LPMO which agrees with the faster rate observed for H₂O₂ by Marletta. However, in contrast to the report by Marletta, Solomon found the product profile of reactions for an AA9 LPMO with O₂ and H₂O₂ to be indistinguishable.²¹⁷ Väljamäe also observed an increased rate for oxidation of chitin by a bacterial LPMO when the transformation was driven by H₂O₂ rather than O₂.²²⁵

There have been several reported synthetic complexes that have attempted to model various species associated with the proposed mechanisms of LPMOs that include a variety of Cu-hydroxido, Cu-superoxido, and Cu-hydroperoxido complexes.^238,239 The possibility of Cu^III–OH being the active species of the LPMO catalytic cycle is reinforced by the synthetic analogue developed by Tolman which, when generated at −80 °C, can activate C─H bonds upwards of 90 kcal/mol (Figure 16A).²³⁸ Itoh has developed a Cu^II complex, Cu^II(LH₃)(tfa)₂, that incorporates the tridentate ligand (N-(2-(1H-imidazol-4-yl)-benzyl)histamine) that accurately models the histidine brace (Figure 16B).²³⁹ This complex has twist angles close to those found in active sites of LPMOs giving spectroscopic parameters similar to those of the enzyme and a quasi-reversible Cu^II/I redox couple within the range reported for the enzymes. Moreover, this model complex was able to catalytically cleave the glycosidic bond in 4-nitrophenyl β-D-glucopyranoside (PNPG) with a turnover number (TON) of 58 after 24 hours.

Figure 16.

Synthetic Cu^III–OH complex by Tolman (A), and synthetic Cu^II complex by Itoh with a histidine brace-type ligand (B).

Most of the above mechanistic and synthetic studies have focused on what occurs directly at the Cu center, yet it is plausible that effects within the secondary coordination sphere could contribute to catalysis.²²⁸ Marletta has examined the role of proximal amino acid residues on stability of a possible Cu^II–OOH intermediate.²²⁸ Using a combination of activity assays, site directed mutagenesis, and EPR spectroscopy, he proposed that residues H161 and Q167 within the secondary coordination sphere form H-bonds with the distal O-atom of putative Cu–OO and Cu–OOH intermediates (Figure 17) – these H-bonds assist in the subsequent O–O cleavage from the Cu-hydroxoperoxido species. The proximity of a redox-active tyrosine residue within 3 Å of the Cu-center in fungal LPMOs has also been proposed as possibly being involved catalysis. Walton has suggested that a Cu^II-tyrosyl radical species forms as a product when H₂O₂ reacts with the Cu^I intermediate.²²⁰ While the role of this intermediate is currently unknown, Walton proposes it could be forming in lieu of a Cu^III–OH or Cu^II–oxyl to act as the competent oxidant. However, work from Solomon states that this tyrosyl species is a minor product that is unlikely to be involved in catalysis because the tyrosine residue is not conserved across all LPMOs. His studies indicate formation of tryptophanyl and tyrosyl radicals, however, further study suggested they were generated only in small quantities during catalysis.²¹⁷ The small amounts of these radical species generated was suggested to indicated that oxidation of these amino acid residues is a minor, off-path route.

Figure 17.

Proposed CuII-OOH showing H-bonds to the distal O-atom involving H161 and Q167.

Vistas

This Forum provided an overview on Fe^IV=O, related Fe–OH, and Cu complexes as a means to understand C─H bond functionalization. The field has made great strides in recent years but there is much space to explore until we have a full appreciation for how these complexes can be applied to chemical problems. The presented topics were described with an eye to critique issues and point out where additional experiments might be beneficial. We have learned that regardless of the systems or method, few synthetic Fe=O complexes can compete with the selectivities and efficiencies of their biological counterparts. We have outlined selective key achievements by Costas and White, who explored increasing chemo- and regioselectivity of C─H bond functionalization, as well as by Collins, whose Fe(TAML) systems have been demonstrated to react efficiently towards the degradation of pollutants reminiscent of P450 enzymes, and have been adapted for green chemical and waste treatment industries (Figure 10). One obvious difference between the synthetic and biological systems is the extent to which the microenvironments around metal center(s) are controlled.¹⁹ Microenvironments are defined as the volume of space that surrounds a metal complex which includes the secondary coordination sphere. In metalloproteins, microenvironments around metallocofactors are highly regulated by the protein host and there is sufficient data to show that this regulation has a substantial impact on function. Our description of the possible role of secondary coordination sphere amino acid residues in the catalytic function of LPMOs offers one example of these effects. Another relevant example is to compare the non-heme Fe monooxygenase discussed above to their related halogenase counterparts. In halogenases such as SyrB2, the organic substrate is positioned in a designated orientation with respect to the metal cofactor in the active site such that, in spite of kinetic deficiencies (slower C─H activation) and thermodynamic considerations (OH-rebound is computed to be more favorable than halide-rebound by > 15 kcal/mol),^49,123,124 the halide-rebound pathway is preferred. Replicating this control of the microenvironment in synthetic systems is challenging: to-date only several Fe(O(H))(halide) complexes have been reported, and they demonstrate little selectivity towards halogenation without regulation in the secondary coordination sphere in bulk organic solvent.^122,240

Several advances in ligand design have produced synthetic systems that model some aspects found within active sites – these studies have increased our understanding of the parallels in the structures of synthetic constructs to architectural aspects found in metalloprotein active sites. However, microenvironments within synthetic systems have not yet achieved the regulatory control found within metalloproteins. Therefore, one of the next challenges is to increase the complexity of the microenvironments surrounding metal centers and to more accurately simulate the intricacies found within protein active sites; yet an increase in complexity comes with significant synthetic difficulties. Furthermore, the added complexity needs to accommodate interactions with water molecules that have been shown to be essential in regulating function within metalloproteins, a feature that is incompatible with most synthetic systems. In order to achieve these goals, scientists have turned to artificial metalloproteins (ArMs) in which the benefits of synthetic inorganic chemistry are leveraged with the power of protein hosts.^241-244 The utility of this approach can be found in the pioneering work of Arnold on directed evolution of cytochrome P450s that has spawned a wave of highly functional ArMs.^245-247 Other notable methods include de novo-designed proteins,^248,249 metal templated interface redesign assembly of proteins,^250,251 and inserting artificial metallocofactors into reconstituted/repurposed metalloproteins.^252-255

One additional method for generating ArMs is to selectively insert a synthetic metallocofactor into a non-metalloprotein host. Examples of ArMs that use this method are those prepared with biotin-streptavidin (Sav) technology.^256,257 First introduced by Whitesides²⁵⁸ and refined by Ward,^259-261 this method has been used to generate ArMs for a variety of different transformations. Within the context of this Forum, we describe two recent reports of non-heme Fe artificial proteins. Ward has engineered new ArMs made from biotinylated Fe(TAML) complexes and Sav that show enhanced reactivity compared to the free Fe complexes.²⁶² By systematically modulating the positioning of the Fe(TAML) complex within the protein pocket, they showed increased enantioselectivity in benzylic C─H bond activation, demonstrating the microenvironment of the active site plays an important role in dictating the transition state of the reaction. Additionally, mutations at key amino acid positions served to modulate both the primary and secondary coordination spheres of the metal complex, which allowed for structural characterization of the metalloproteins, and enhanced enantioselectivity and turnover capability towards substrate oxidation were also observed. We have used this approach to model the active site of non-heme Fe monooxygenase by engineering an Fe artificial protein with a facial triad active site.²⁶³ We biotinylated the common tridentate ligand dpa (di-(2-pyridylmethyl)amine) and prepared the corresponding Fe^II complex. Sav was modified by placing a glutamate at position 112 (S112E) and the resulting ArM had an Fe site with a facial triad composed of two N-atom donors from the dpa ligand and one O-atom donor from the carboxylate side chain from the glutamate with remaining ligands being water molecules. X-ray diffraction studies confirmed the structure and show a κ¹-coordination of the carboxylate that is identical to that found in the native monooxygenases. Moreover, the carboxylate forms an intramolecular H-bond with one of the aqua ligands which also matches what is found in natural proteins. These examples illustrate that biologically relevant Fe sites can be designed within protein hosts and thus offer another way to model both primary and secondary coordination sphere effects to produce functional systems.