Mechanisms of O₂ Activation by Mononuclear Non-Heme Iron Enzymes

Abstract

Two major subclasses of mononuclear non-heme ferrous enzymes use two electron donating organic cofactors (alpha-ketoglutarate or pterin) to activate O₂ to form Fe^IV=O intermediates that further react with their substrates through hydrogen atom abstraction or electrophilic aromatic substitution. New spectroscopic methodologies have been developed enabling the study of the active sites in these enzymes and their oxygen intermediates. Coupled to electronic structure calculations the results of these spectroscopies provide fundamental insight into mechanism; this Prospective summarizes the results of these studies in elucidating the mechanism of dioxygen activation to form the Fe^IV=O intermediate and the geometric and electronic structure of this intermediate that enables its high reactivity and selectivity in product formation.

Graphical Abstract

Introduction

Iron enzymes play a major role in O₂ utilization by nature.^1–3 Their diverse range of functions include antibiotic, ^4,5 neurotransmitter^6,7 and natural product biosynthesis, ⁸ hypoxia regulation, ⁹ DNA repair, ¹⁰ bioremediation ^11,12 and drug metabolism ^13,14 and are related to a number of genetic diseases.^15,16 On a molecular level, they perform H-atom abstraction (HAA) leading to hydroxylation,^17,18 halogenation,⁸ desaturation,¹⁹ ring closure²⁰ and expansion²¹ and electrophilic attack on aromatics (electrophilic aromatic substitution, EAS)²² and on heteroatoms.^23,24 O₂ activating Fe enzymes have either heme or non-heme active sites, where the latter can be mononuclear or binuclear. Here we focus on the mononuclear non-heme iron (NHFe) enzymes. These are divided into two classes based on whether or not they utilize a 2 e⁻ organic cofactor (either alpha ketoglutarate (αkg) or tetrahydrobiopterin (pterin)).²⁵ In both classes, the majority of the enzymes utilize a facial triad (2 histidines, 1 carboxylate) ligand set,²⁶ however in several enzymes, the carboxylate is replaced by a third histidine²⁷ and in the halogenases, it is replaced by a halide.²⁸ NHFe enzymes have been challenging to study as they lack the intense spectroscopic features of the porphyrin ring characteristic of heme enzymes. While both non-heme and heme Fe enzymes are accessible by Mössbauer spectroscopy, this technique mainly probes redox and spin state. It does not provide detailed geometric and electronic structure insight to complement crystallography of the ferrous (Fe^II) sites and fully define intermediates, and thus the catalytic mechanism. Our focus has been to develop variable temperature variable field (VTVH) magnetic circular dichroism (MCD) spectroscopy to study the ferrous sites and their interactions with cosubstrates,²⁹ nuclear resonance vibrational spectroscopy (NRVS) to define the geometric structures of intermediates^30–32 and MCD on these intermediates to define their electronic structure and frontier molecular orbitals (FMOs) for reactivity.³³ These spectroscopic methods are presented in detail in references 29–33 and here we focus on their application to provide chemical insights into non-heme iron enzyme reaction mechanisms.

From Table 1, there are three subclasses of Fe^II facial triad enzymes that do not use organic cofactors: those having non-redox active, cysteine containing substrates (isopenicillin N synthase (IPNS),³⁴ cysteine dioxygenase (CDO),²⁷ persulfide dioxygenase (ETHE1)),³⁵ those having 1 e⁻ donating Fe₂S₂ Rieske centers (Rieske dioxygenases (RDO))¹¹ and those having redox active catecholate substrates (the extradiol dioxygenases (EDO)).³⁶ Recent studies have utilized the above-mentioned combination of spectroscopic methods coupled to density functional theory (DFT) calculations to evaluate O₂ activation by all three subclasses.^37–39 Investigation into the oxygen-activated intermediates in these subclasses yielded two possible intermediates, an Fe^III-hydroperoxo and an Fe^III-superoxide. While Fe^III-hydroperoxo species can be trapped and spectroscopically and structurally defined in the RDO^40,41 and EDO, ⁴² the barriers for their attack on their corresponding substrates are too high to be catalytically relevant.^37,38 Alternatively, Fe^III-superoxide intermediates in all three subclasses have significantly lower reaction barriers that are consistent with observed catalytic rates.^37–39 In the RDO, O₂-activation to form an Fe^III-O₂⁻ is endergonic but the reaction is driven by proton coupled electron transfer from the Rieske center following superoxo attack on the aromatic ring of the substrate (which is bound in the protein pocket oriented above the O₂ binding site).^38,43 In the two other subclasses, the cysteine containing³⁹ and catecholate³⁷ substrates coordinate the Fe(II) as strong donors, lowering its reduction potential such that Fe^III-superoxide formation is exergonic and has been observed both in IPNS⁴⁴ and in an EDO with a less reactive substrate.⁴² From spectroscopy supported calculations, the lower energy barriers for substrate attack in all three subclasses derive from the overlap of the highest occupied molecular orbital (HOMO) of the substrate with the frontier molecular orbital (FMO) of the Fe^III-superoxo, which is the unoccupied π* orbital of the superoxide (Figure 1). This FMO is highly anisotropic with respect to the FeO₂ plane and must be oriented by the ligation on the Fe for effective electrophilic attack on the substrate.^37–39

Table 1:

O₂ activating mononuclear non-heme Fe enzyme classes.

Figure 1:

Fe^III-superoxo π* FMO (i.e. the LUMO) for electrophilic attack on substrate.

Again, from Table 1, there are two subclasses of NHFe enzymes that use organic cofactors: the αkg and the pterin dependent enzymes. (Note from Scheme 1 that the αkg and pterin are, in fact, cosubstrates as one atom of the O₂ is initially incorporated into the oxidized product). These enzymes use an Fe^II facial triad site plus the 2e⁻ donating organic cofactor/cosubstrate to reduce O₂ by four electrons to generate a high spin S = 2 Fe^IV=O intermediate^25,45,46 that goes on to perform a wide range of reactions (Scheme 1).

Scheme 1:

Reaction mechanisms for cofactor (αkg and pterin) dependent non-heme iron enzymes. Adapted from [Iyer, S. R. et al. (2021) Direct coordination of pterin to Fe^II enables neurotransmitter biosynthesis in the pterin-dependent hydroxylases. Proceedings of the National Academy of Sciences 118.] Copyright [2021].

The αkg dependent Fe^II enzyme reactions⁴⁷ involve HAA followed by hydroxylation, halogenation, desaturation or the related ring closure or expansion, important in a variety of functions, including antibiotic biosynthesis,^4,5 hypoxia regulation⁹ and DNA repair.¹⁰ In the pterin dependent enzymes, the Fe^IV=O reaction involves EAS leading to aromatic amino acid hydroxylation such as in neurotransmitter biosynthesis.⁴⁸ In this Prospective, we focus on the recent advances made in understanding: 1) the mechanism of O₂ activation by the cofactor dependent NHFe enzymes to form the Fe^IV=O intermediate, and 2) the factors that enable the Fe^IV=O reactivity and selectivity, derived from the application of spectroscopies, kinetics and electronic structure calculations.

1. O₂ Activation Mechanism

1.1. Sequential vs. Concerted

MCD spectroscopy has proven to be a powerful method to study ferrous active sites and their interactions with cosubstrates. In particular, MCD enables the study of d-d transitions, i.e., ligand field (LF6) transitions that are too weak to observe in the absorption spectrum of metalloproteins, but directly probe the geometric structure of the high spin ferrous active sites.⁴⁹ The application of MCD spectroscopy to a range of different NHFe^II enzymes has led to the general mechanistic strategy⁵⁰ illustrated in Scheme 2 and demonstrated by the representative MCD data on the enzyme FIH⁵¹ (factor inhibiting HIF (hypoxia inducible factor)) in Figure 2A. The resting enzyme has two LF transitions in the 10,000 cm⁻¹ region split by ^~2000 cm⁻¹ indicative of a six coordinate (6C) Fe^II site (see references 48 and 49 by for spectroscopic details).

Scheme 2:

General mechanistic strategy for cofactor dependent NHFe enzymes.

Figure 2:

Near IR MCD spectra of Fe^II sites in FIH (A) and DAOCS (B) for resting enzyme (top), substrate bound (right second down), akg bound (left middle and right third down) and both akg and substrate bound (bottom). Ferrous ligand field (LF) transitions and charge transfer (CT) transitions. Figure adapted from [Iyer, S. R. et al. (2018) O₂ activation by non-heme Fe^II α-ketoglutarate-dependent enzyme variants: Elucidating the role of the facial triad carboxylate in FIH. Journal of the American Chemical Society 140, 11777–11783.] and [Goudarzi, S. et al. (2020) Evaluation of a concerted vs. sequential oxygen activation mechanism in α-ketoglutarate—dependent nonheme ferrous enzymes. Proceedings of the National Academy of Sciences 117, 5152–5159]. Copyright [2018] and [2020] respectively.

Binding substrate to the protein pocket minimally perturbs these transitions (not included in Fig 2A) showing that the substrate does not directly coordinate to the Fe^II. Alternatively, when αkg is bound new transitions appear at higher energy (Fig 2A middle).⁵² These are charge transfer (CT) transitions and demonstrate that the αkg is bound to the Fe^II (bidentate from their analysis⁵² and from crystal structures⁵³). Importantly, from Fig 2A, the Fe^II site still has two LF transitions in the 10,000 cm⁻¹ region split by 2,000 cm⁻¹ indicating that the Fe^II site remains 6C when αkg is bound. Thus, while all four electrons required for O₂ reduction to form the Fe^IV=O intermediate are available (two from the Fe^II and two from the αkg), the Fe^II is coordinatively saturated and unreactive with O₂. However, when the substrate binds to the protein pocket near the αkg bound Fe^II center, the LF transitions show large changes (Fig 2A bottom), indicating that the Fe^II site is now five coordinate (5C) and can bind and react with O₂ in the presence of substrate.

However, an alternative sequential mechanism⁵⁴ has been proposed based on crystallography and calculations, initially on the enzyme Deacetoxycephalosporin-C synthase (DAOCS) and then applied to human DNA oxidative demethylase ABH2⁵⁵ and (S)-2-hydroxypropyl-phosphonic acid epoxidase.⁵⁶ This proposal involves αkg binding to the Fe^II active site, reaction with O₂ to generate a ferrous-peroxysuccinate intermediate and upon subsequent substrate binding, its conversion to an Fe^IV=O that reacts to form product. MCD studies on Fe^II-DAOCS have defined a behavior of the Fe^II active site that enabled the experimental evaluation of a possible sequential mechanism.⁵⁷ As shown in Fig 2B, the resting Fe^II site is 6C and only minimally perturbed by substrate (penicillin-G) binding in the vicinity of the Fe^II. Importantly, when αkg alone is bound (demonstrated by the CT transitions in the higher energy region in Fig 2B the third spectrum down), in contrast to the behavior of other αkg dependent NHFe enzymes, as in Fig 2A, the LF transitions of DAOCS show a major change (including the negative feature at 11,600 cm⁻¹), indicating that a component (^~45%) of the Fe^II sites has become 5C. Thus this site has become coordinatively unsaturated and since αkg is bound, it can now react with O₂ in the absence of substrate. Note from Fig 2B bottom, that substrate addition to the αkg bound DAOCS Fe^II site further changes the MCD spectrum indicating that more of the Fe^II site (^~60%) become 5C in this ternary complex.

In order to evaluate the concerted versus sequential reaction mechanism in DAOCS, stopped flow absorption and rapid freeze quench (RFQ) Mössbauer and electron paramagnetic resonance (EPR) studies were pursued on the O₂ reaction of αkg bound Fe^II-DAOCS both in absence and presence of substrate. The results of these studies presented in work by Goudarzi et al.⁵⁷ are summarized in Scheme 3. In the absence of substrate (Scheme 3A), a succinate bound Fe^III-OH species rapidly forms. Its subsequent reaction with penicillin G resulted in only the unproductive one electron oxidation of the substrate. Thus, in the absence of prebound substrate, the reactive O₂ intermediate has rapidly dissipated, and a sequential mechanism is not operative. Alternatively, when the O₂ reaction is performed with both the αkg and substrate simultaneously bound (Scheme 3B), Fe^IV=O formation is observed that rapidly reacts with the bound substrate to form an Fe^III-OH/succinate and substrate radical that concertedly decay with the productive formation of the ring expanded cephalosporin product. Note that the ring expansion rearranges the radical away from the Fe^III-OH precluding rebound hydroxylation and resulting in desaturation.

Scheme 3:

Fe^II-DAOCS/αkg reaction with O₂ in the absence (A) and presence (B) of bound substrate (penicillin G). Adapted from [Goudarzi, S. et al. (2020) Evaluation of a concerted vs. sequential oxygen activation mechanism in α-ketoglutarate—dependent nonheme ferrous enzymes. Proceedings of the National Academy of Sciences 117, 5152–5159.] Copyright [2020].

Thus, the general mechanistic strategy where both cofactor and substrate are bound to generate a 5C Fe^II site for O₂ activation is an effective gating mechanism utilized by NHFe enzymes for coupled catalysis (Scheme 2).

1.2. Cofactor Coordination and Mechanism of Fe^IV=O Formation

As presented in the Introduction, for both the αkg and pterin dependent NHFe enzymes, the reactions of the substrate and cofactor bound Fe^II sites have been shown by Mössbauer spectroscopy to form Fe^IV=O intermediates.^25,45,46 For the pterin dependent enzyme tryptophan hydroxylase (TPH), we have recently trapped a precursor intermediate that experimentally defines a common mechanism for both subclasses of cofactor dependent NHFe enzymes in Fe^IV=O formation.⁵⁸ From the inset in Figure 3A bottom, addition of substrate to Fe^II-TPH preloaded with the pterin cofactor results in changes in the LF region of the MCD spectrum, indicating a 6C to 5C conversion, paralleling the results summarized above for most αkg dependent enzymes (Fig 2A).

Figure 3:

Substrate binding (red) to pterin-bound (blue) Fe^II tryptophan hydroxylase (TPH). A. Abs (top) and MCD (bottom, near-IR MCD inset). B. Resonance Raman spectra (excitation at 334.5 nm). Adapted from [Iyer, S. R. et al. (2021) Direct coordination of pterin to Fe^II enables neurotransmitter biosynthesis in the pterin-dependent hydroxylases. Proceedings of the National Academy of Sciences 118.] Copyright [2021].

Importantly, when the substrate is also bound (Fig 3A red), a new CT feature appears at 330 nm (30,000 cm⁻¹) in the absorption spectrum that is intense and derivative shaped in MCD, indicating that it is associated with a paramagnetic center. Thus, this is a CT to the Fe^II, and from the resonance Raman (rR) data in Fig 3B, it is from the pterin with its carbonyl bound to the Fe^II (the C=O stretch is at 1601 cm⁻¹, decreased from 1695 cm⁻¹ by coordination). This 330 nm CT requires that the pterin is bound to the Fe^II in this ternary complex. Previously, based on crystallography,⁵⁹ it had been generally thought that the pterin does not bind to the Fe^II which, as presented below, would have resulted in too high of a barrier for the O₂ activation reaction.

From Figure 4A, reaction of the substrate plus pterin bound 5C Fe^II site with O₂ eliminates the 330 nm CT and initiates the growth of a new CT at 442 nm. This 442 nm intermediate has a Mössbauer spectrum (Fig 4C) indicating that it has a high spin Fe^II center, while the rR (Fig 4D) and other spectroscopic data indicate that this intermediate is a peroxy-oxidized pterin with both the peroxide and the carbonyl of the pterin bound to the Fe^II. Specifically, the rR shows the 1570 cm⁻¹ ν(C=O) is again shifted down in energy due to coordination, and the ¹⁸O₂ isotope perturbation is distributed over the stretches and bonds of the ring formed by this bidetate peroxy-pterin coordination. The 442 nm peroxy intermediate decays with a solvent kinetic isotope effect of about four (Fig 2B), indicating a proton-induced O-O cleavage to produce hydroxybiopterin and the subsequent Fe^IV=O intermediate. This is the first time a precursor to the Fe^IV=O intermediate has been spectroscopically defined in either cofactor dependent subclass and elucidates the mechanism of Fe^IV=O formation.

Figure 4:

Reaction of substrate and pterin bound Fe^II-TPH with O₂. A. Kinetics of first 175 ms of the reaction. B. Formation and decay kinetics at 442 nm in H₂O and D₂O. C. Mössbauer spectrum of 442 nm intermediate. D. Resonance Raman spectrum of 442 nm intermediate (excitation at 457.9 nm). Adapted from [Iyer, S. R. et al. (2021) Direct coordination of pterin to Fe^II enables neurotransmitter biosynthesis in the pterin-dependent hydroxylases. Proceedings of the National Academy of Sciences 118.] Copyright [2021].

As shown in Figure 5A, O₂ binding to the open coordination position of the Fe^II results in some Fe^III-superoxo character. Attack of the distal O of the nascent superoxo on the C4α position of the pterin results in the doubly oxidized peroxy pterin bound to an Fe^II (the 442 nm intermediate). Thus, the pterin needs to donate one electron to the superoxo and a second electron to the Fe to form this peroxy-ferrous intermediate.

Figure 5:

Reaction Coordinate for Fe^II-pterin TPH. A. Effect of pterin binding to Fe^II on reaction energetics (red to green). B. Schematics of electronic structure contributions of pterin carbonyl coordination on the energetics in A. Red arrow indicates e-donation from Fe^II to O₂, blue arrow indicates e- donation from pterin to O₂; the green arrow indicates donation of a second e- from pterin to Fe (to compensate for the Fe donation to the O₂). The green electron donation is facilitated by pterin coordination to the Fe (depicted on the right). Adapted from [Iyer, S. R. et al. (2021) Direct coordination of pterin to Fe^II enables neurotransmitter biosynthesis in the pterin-dependent hydroxylases. Proceedings of the National Academy of Sciences 118.] Copyright [2021].

As shown in Fig 5B right, pterin coordination to the Fe enables this second electron to transfer directly to the Fe (green arrow) and lowers the barrier for reactivity by 14.7 kcal/mol (Fig 5A green vs. red). Since the rate determining step for the reaction is product release with a barrier of 16 kcal/mol,^60,61 if the pterin were not bound to the Fe, the 21 kcal/mol calculated barrier would result in 10⁴ slower turnover, thus enzymatic dysregulation that would impact the metabolic pathways necessary for proper brain function.⁶² Thus, there is a common mechanism utilized by both the αkg and pterin dependent NHFe enzymes to form the Fe^IV=O intermediate. Concerted substrate plus cofactor binding to the active site pocket opens a coordination position on the Fe^II for O₂-activation. For both subclasses, the cofactor binds directly to the Fe enabling the electron transfer to the FeO₂ required for Fe^II-peroxy-oxidized cofactor formation.

2. Fe^IV=O Structure and Reactivity

Proceeding from the Fe^II-peroxy-oxidized cofactor now observed by spectroscopy in the pterin dependent enzymes⁵⁸ and predicted by crystallography⁶³ for the αkg dependent enzymes, protonation of the distal O or direct O-O cleavage leads to the hydroxybiopterin or bound succinate, respectively, and the Fe^IV=O S = 2 intermediate. In the αkg enzymes, the Fe^IV=O performs HAA followed by rebound hydroxylation,^17,18 halogenation⁸ (in halogenases) or desaturation¹⁹ and related reactions. In the pterin dependent enzymes, the Fe^IV=O performs EAS on aromatic amino acid substrates. M. Bollinger, C. Krebs and colleagues provided high concentration, high quality samples of the Fe^IV=O S = 2 intermediates in the hydroxylating TauD (degradation of sulfonic acids) and the halogenating SyrB2. These samples enabled geometric structural determination by NRVS ^64,65 and, combined with parallel studies on the structurally defined S = 2 Fe^IV=O model of L. Que and colleagues,⁶⁶ electronic structure determination by VTVH MCD.^33,67

NRVS uses a synchrotron to scan the ⁵⁷Fe nuclear Mössbauer transition with a focus on the vibrational sidebands at higher energy,⁶⁸ where the NRVS intensity reflects Fe motion in a normal mode at the observed energy. The NRVS spectrum of the Fe^IV=O intermediate in TauD is given in Figure 6; that of SyrB2 is very similar.

Figure 6:

NRVS spectrum of the Fe^IV=O intermediate of TauD. Adapted from [Srnec, M. et al. (2020) Nuclear Resonance Vibrational Spectroscopic Definition of the Facial Triad Fe^IV=O Intermediate in Taurine Dioxygenase: Evaluation of Structural Contributions to Hydrogen Atom Abstraction. Journal of the American Chemical Society 142, 18886–18896.] Copyright [2020].

The features in region 3 are the His N-Fe stretches, the intense feature in region 2 is the trans axial bend of the Fe^IV=O, and the two peaks in region 1 are carboxylate O-Fe stretches. To contribute to this high energy region 1, the two carboxylates (one from the facial triad and one from the succinate formed by decarboxylation) must both be coordinated monodentate, as bidentate carboxylates have longer Fe-O bonds resulting in lower energy stretches shifted into region 2. Thus, both the carboxylate of the facial traid and that of the succinate are monodentate and the Fe^IV=O is 5C. Simulations of the NRVS data with a series of possible 6C and 5C structures showed that the Fe=O intermediate in TauD (and in SyrB2) is trigonal bipyramidal (TBP) with the oxo along the C₃ axis.⁶⁴ A schematic of this structure is shown in Figure 7A.

Figure 7:

Fe^IV=O structure and FMOs. A. Schematic of Fe^IV=O intermediate with possible substrate orientations along (σ) and perpendicular (π) to the Fe-oxo bond. B. σ (black) and π (red) FMOs enabling HAA along (σ) and perpendicular to (π) the Fe-oxo bond. PE indicates promotion energy required to reach the dπ FMO. Adapted from [Srnec, M. et al. (2020) Nuclear Resonance Vibrational Spectroscopic Definition of the Facial Triad Fe^IV=O Intermediate in Taurine Dioxygenase: Evaluation of Structural Contributions to Hydrogen Atom Abstraction. Journal of the American Chemical Society 142, 18886–18896.] Copyright [2020].

Due to the short Fe-O bond (1.62 Å) of this TBP Fe^IV=O S = 2 site, the LF results in the dz²/dσ orbital at highest energy and unoccupied (Fe^IV is d⁴), thus this is the FMO for the HAA reaction and, based on orbital overlap, this would occur along the Fe-O bond (Fig 7B, black). However, VTVH MCD spectra of the Fe^IV=O intermediate in SyrB2⁶⁷ and of the TBP Fe^IV=O Que model³³ , Figure 8A (note these spectra are very similar further supporting the TBP enzyme intermediate structure from NRVS), show a low-lying electronic transition near 12,000 cm⁻¹ (red double arrow). This transition is very weak in absorption but intense in MCD and from this intensity ratio, its derivative shape, its temperature dependence, and its vibronic structure in the Fe=O stretch vibration, this is the dπ* to dσ* LF transition. As shown in Fig 7B red, this dπ* to dσ* transition creates a hole in the dπ* orbital, an FMO activated for attack on the substrate perpendicular to the Fe-O bond. From calculations⁶⁴ correlated to the MCD data, this dπ* FMO has 80% oxo (i.e., oxyl) character and from Fig 8B, rapidly comes down in energy as the Fe-O bond is elongated to reach the TS for HAA (1.82 Å).

Figure 8:

Experimental insight into dπ* FMO. A. Abs (top) and MCD (bottom) definition of Fe^IV=O FMOs in SyrB2 (left) and its TMG₃tren model complex (right). Red double arrows indicate dπ* to dσ* LF excitation at an Fe=O bond length of 1.62 Å. B. Dependence of the dπ* to dσ* excitation energy (the promotion energy (PE) in Fig. 7B) on Fe=O bond length (1.82 Å at the TS). Insets show the σ and π FMOS at an Fe=O distance of 1.62 Å. Adapted from [Srnec, M. et al. (2020) Nuclear Resonance Vibrational Spectroscopic Definition of the Facial Triad Fe^IV=O Intermediate in Taurine Dioxygenase: Evaluation of Structural Contributions to Hydrogen Atom Abstraction. Journal of the American Chemical Society 142, 18886–18896.] and [Srnec, M. et al. (2016) Electronic Structure of the Ferryl Intermediate in the alpha-Ketoglutarate Dependent Non-Heme Iron Halogenase SyrB2: Contributions to H Atom Abstraction Reactivity. Journal of the American Chemical Society 138, 5110–5122.] Copyright [2020] and [2016] respectively.

Thus, the promotion energy (PE) to reach this dπ* FMO in Fig 7B is only 7 kcal/mol, and, due to its high oxyl character (inset in Fig 8B, right), the TS is earlier in C-H elongation (i.e., better overlap) resulting in less distortion energy (relative to σ attack along the Fe^IV=O bond). The lower distortion energy compensates for the promotion energy and both the π and σ FMOs in Fig 7B are found to have comparable HAA reactivity.⁶⁴ However, due to its orientation towards the substrate the dσ* FMO can only enable rebound hydroxylation (assuming the substrate radical does not rearrange as in DAOCS (Scheme 3B) or decay⁶⁹). Alternatively, the π FMO enables selectivity in rebound, thus far elucidated for the halogenases, as summarized below.

For SyrB2, reaction coordinate calculations⁶⁵ correlated to the NRVS data lead to an Fe^IV=O oriented perpendicular to the C-H bond of the substrate, L-threonine (Fig 7A, π substrate orientation). From this orientation, HAA results in an HS Fe^III-OH with the substrate radical oriented above the plane containing both the Fe^III-OH and the Fe^III-Cl bonds (Fig 9A left). From this orientation, while rebound hydroxylation is 12 kcal/mol more favorable (Figure 9A, red), the halogenation and hydroxylation are found to have very similar barriers (Fig 9A, halogenation (green) favored).⁷⁰ While the thermodynamics of hydroxylation is found to lower the barrier by 4 kcal/mol, the halogenation rebound has a 4.5 kcal/mol lower (thermodynamically corrected) intrinsic barrier.

Figure 9:

Rebound halogenation vs hydroxylation. A. After HAA by SyrB2 the resultant Fe^III-OH has the substrate radical oriented above the HO-Fe^III-Cl plane. Energetics of rebound hydroxylation in red and rebound halogenation in green. B. FMOs for rebound, dπ* of HO-Fe^III is 4.5 kcal/mol higher than dπ* of Cl-Fe^III in energy. Adapted from [Srnec, M. et al. (2017) Frontier Molecular Orbital Contributions to Chlorination versus Hydroxylation Selectivity in the Non-Heme Iron Halogenase SyrB2. Journal of the American Chemical Society 139, 2396–2407.] Copyright [2017].

This reflects the lower energy of the halide dπ* orbital relative to the hydroxo dπ* orbital of the Fe^III (Fig 9B right, from their difference in LF), where the rebound mechanism involves two asynchronous transfers: electron transfer from the substrate radical to the appropriate dπ FMO on the Fe followed by late ligand (Cl⁻ or OH⁻) transfer to the substrate. Thus, the thermodynamic and intrinsic barrier contributions to the rebound barrier cancel and the product of rebound is determined by the relative overlap of the substrate radical with the two dπ* FMOs (Fig 9B), and thus varies with the positioning of the substrate as observed experimentally.⁷¹ This emphasizes the important role of the second coordination sphere of the protein in positioning the substrate to control the selectivity.

3. Concluding Comments

These spectroscopic and experimentally calibrated DFT insights into the reactivity of cofactor-dependent non-heme iron enzymes sets the stage for fundamental insight into the reaction mechanisms of these and related enzymes. For the αkg dependent enzymes, we need to evaluate whether the π and σ FMO concepts extend to differences in selectivity in desaturation vs hydroxylation. For the pterin dependent enzymes, an outstanding issue is whether the π FMO enables reaction with the substrate that must be oriented perpendicular to the Fe=O bond formed by O-O cleavage of the Fe^II-peroxy-pterin intermediate. Further, in comparing the two enzyme subclasses, it is important to determine whether the Fe^IV=O intermediate in the pterin enzymes is selectively activated for EAS based on its ligation differences relative to the Fe^IV=O of the αkg dependent enzymes (no bound succinate and bidentate coordination of the facial triad carboxylate). Finally, studies are underway to determine the nature of Fe^IV-oxo activation in the binuclear NHFe intermediate Q in soluble methane monooxygenase and in Compound I and II in heme enzymes and how these correlate to the mononuclear NHFe enzyme catalysis summarized above.