Carbon–fluorine bond cleavage mediated by metalloenzymes

2.1. Histidine-ligated heme-dependent dehaloperoxidase (DHP)

Metalloenzyme-mediated defluorination is a rare catalytic transformation found across the biosynthesis of natural metabolites. However, multifunctional globin-like DHP protein is a well-characterized example of an enzyme capable of promoting such chemistry. DHP was originally discovered in Amphitrite ornata, a marine polychaete worm of the family Terebellidae.⁵⁰ In addition to binding and transporting molecular oxygen as hemoglobin, DHP has long been known to have multiple catalytic properties such as activity typical of an oxygenase, an oxidase, a peroxygenase, and even a peroxidase for dehalogenation.^51–53 Herein, we will focus on discussing the peroxidase activity of this enzyme.

DHP oxidizes trihalophenol (TXP) into its corresponding quinone by utilizing hydrogen peroxide (H₂O₂) as the cosubstrate, resulting in the elimination of the para-substituted halogen (F, Cl, Br, or I) on TXP in the form of a halide anion, as shown in Fig. 1A. In the resting state, DHP possesses a histidine-ligated b-type ferric heme (Fig. 1B), as well as an overall fold that is typical of the globin family.⁵⁴ The enzyme is predominantly dimeric in solution, and it crystallizes as a dimer in an asymmetric unit.³⁰ The crystal structures of DHP in complex with the substrate analogs or inhibitors exhibit diverse binding modes in the active site, indicative of a spacious substrate cavity.^55–58 The crystal structures of the enzyme-substrate (ES) complexes were determined by soaking them in a native substrate, namely, tribromophenol (TBP). However, the occupancy of TBP is extremely low in the determined ES complex structures (occupancy: ~10%; PDB entries: 4FH6 and 4FH7).⁵⁶ Although the structures of the ES complex are contentious due to their low occupancies, TBP has been shown to bind at the active site, with its hydroxyl group pointing toward the heme center, while the para-substituent to be eliminated points away (Fig. 1B). This structural information excludes the possibility of direct activation at the para position by a hemebased oxidant during the catalytic reaction.

The dehalogenation mechanism of DHP is presumed to involve two one-electron oxidation processes mediated by the high-valent heme iron oxidants, i.e., Compound I (Cpd I)- and Compound II (Cpd II)-like species, respectively.^59–62 Cpd I and Cpd II represent two reactive ferryl-oxo intermediates in the catalytic cycle of the thiolate-ligated cytochrome P450, where the former contains a porphyrin-based cation radical, which is absent in the latter. These high-valent species, or their equivalents with a histidine-ligated heme, are commonly proposed oxidants in the catalytic pathways of heme-containing enzymes.^63,64 Fig. 2 shows the proposed defluorination mechanism of DHP. Even though the binding order of TXP and H₂O₂ is undetermined,^55,65 it is believed that the conformational flexibility of a distal histidine, i.e., His55, plays a critical role in adopting an activated DHP.^66,67 In the activated form, His55 is in close contact with the heme center, as shown in Fig. 1B. Then, the protein promotes the heterolytic cleavage of the O-O bond by providing a proton to the ferric hydroperoxo, resulting in a histidine-ligated Cpd I-like species.^65,68 Such a species is chemically equivalent to the so-called Compound ES (Cpd ES), which is a Cpd I isoform with a ferryl heme and a nearby protein-based aromatic radical.^69,70 Subsequently, Cpd I or Cpd ES proceeds with the first one-electron oxidation of the substrate, yielding a dissociable organic radical on the substrate and a Cpd II-like species.^71,72 Then, the Cpd II-like species undergoes a second one-electron oxidation process on a resonance structure of the substrate radical, affording a phenoxy cation. The formation of the stable 2,6-difluoroquinone product proceeds through the elimination of fluoride, which occurs after the nucleophilic attack by a water molecule at the para-substituted carbon atom bearing the most cationic character.^57,71 The¹⁸O-isotope-labeling studies confirmed that the oxygen atom incorporated into the final product is derived from water rather than the peroxide, which is consistent with the binding orientation of the substrate (as revealed by the crystal structures).⁵⁷

Although DHP functions as a natural dehalogenase that oxidizes phenols with various halogen substituents, affinity and catalytic efficiency studies have shown that DHP favors brominated substrates under physiological conditions.^50,73 Fluorinated phenols, on the other hand, have weak binding affinities and limited turnovers.^50,58,73 However, the construction of an artificial DHP could be a promising method to develop competent biocatalysts that effectively cleave aromatic C-F bonds. Indeed, artificially created DHP activity has been achieved by mutating active-site residues in structurally related myoglobin.⁷⁴ Although the defluorination activity of the native protein is yet to be reported, the engineered enzyme demonstrates a 1000-fold increase in efficiency on a trichlorophenol substrate. Consequently, engineered DHP is a prime candidate for defluorination activity, which confirms that residue-level modification is an effective strategy to achieve engineered enzymatic C-F bond cleavage of harmful fluorinated compounds.

2.2. Histidine-ligated heme-dependent tyrosine hydroxylase

A new class of biocatalysts hydroxylating _L-tyrosine to _L-dihydroxy-phenylalanine (DOPA) has been found in the biosynthetic pathways of antitumor and/or antibacterial antibiotics.⁷⁵ While their protein structures remain to be determined, these tyrosine hydroxylases are heme-dependent proteins, hereafter referred to as heme TyrHs. A protein sequence analysis revealed that this type of enzymes comprised a b-type histidine-ligated heme.^76,77 The first of this class of enzyme, namely, LmbB2, was identified from the biosynthetic gene cluster of lincomycin.⁷⁸ In addition to LmbB2, other characterized heme TyrHs include the HrmE of hormaomycin,⁷⁹ Orf13 of anthramycin,⁷⁶ SibU of sibiromycin,⁸⁰ TomI of tomaymycin,⁸¹ and Por14 of poranthramycin.⁸² The amino acid sequence alignment of these TyrHs shows no similarity with any known conserved domains or conserved motifs.⁷⁶ Therefore, this group of proteins probably contains a novel structural fold, and determining the structural features of this enzyme family is imperative for mechanistic understandings. The heme-dependent TyrH is responsible for the first step in the construction of a pyrroline moiety in the biosynthetic pathways of the natural product by converting _L-tyrosine into DOPA through hydrogen-peroxide-mediated oxidation. Unlike P450, nonheme TyrH, and other aromatic hydroxylases,^83–87 the catalytic hemes in TyrHs are distinguished by an axial histidine ligand. Therefore, their mechanism of C-H bond activation and oxygen atom transfer cannot be simply extrapolated.³²

LmbB2 has been shown to be capable of hydroxylating a range of _L-tyrosine analogs with ring-deactivating meta-substituents such as 3-fluorotyrosine.³² Moreover, in addition to the expected C-H bond cleavage product, an additional product derived from the C-F bond cleavage has been observed (Fig. 3A). C-F bond cleavage requires two additional electrons for fluoride versus proton elimination; hence, the mechanism of C-F bond hydroxylation is likely to differ from native C-H bond hydroxylation. In the process of C-F bond cleavage, the consumption of additional peroxide and formation of oxygen were found, which suggests that the oxidation of peroxide into oxygen is presumably the source of two necessary electrons. Moreover, ¹⁸O-isotope-labeling experiments show that H₂O₂ is the oxygen source for hydroxylation in both C-H and C-F bond cleavage reactions. The detection of the double-scrambled product (with ¹⁸O atoms on both hydroxyl groups of DOPA) indicates a potential ketone or radical intermediate with broken aromaticity during catalysis. Collectively, a mechanism of C-F bond cleavage promoted by the heme-dependent TyrH was proposed (Fig. 4).

Substrate binding to the active site allows for H₂O₂ activation by the ferric heme, affording a ferric-bound hydroperoxide. It has been speculated that 3-fluorotyrosine can bind in the active site with two distinct conformations, with the fluorine pointing toward or away from the heme center.³² Fluorine pointing away from the heme results in the native hydroxylated product, while fluorine oriented toward the heme results in the defluorinated product. In the latter case, due to the strong electron-withdrawing property of the fluorine substituent, the covalently bound carbon (C3) is partially electropositive. Therefore, a ferric heme-bound hydroperoxide can perform a nucleophilic attack at C3, resulting in a Cpd I-like species. In contrast to alkyl systems, fluoride is an excellent leaving group in a nucleophilic aromatic substitution reaction.⁸⁸ It is eliminated from the ring to re-aromatize, yielding DOPA as the final product. The resulting Cpd I-like species can react with an additional equivalent of H₂O₂ to release oxygen and return to the resting state via catalase-like activity. Besides 3-fluorotyrosine, 3-chloro- and 3-iodotyrosine compounds have also been identified with dual reactivities. 3-Chloro-tyrosine has the best overall conversion (normal hydroxylation plus dehalogenation). At the same time, 3-fluorotyrosine exhibits the highest dehalogenation efficiency, despite the C-F bond being the most durable among all the carbon-halogen (C-X) bonds. The different distributions of hydroxylated products from two distinct reaction pathways are a cumulative outcome of the steric and inductive effects of these halogen substituents.

Cleaving an aromatic C-F bond is not a common feature among the histidine-ligated heme proteins because they typically function as globins or peroxidases to bind oxygen or transfer electrons. But a closer comparison between DHP and heme TyrH suggests that these two enzymes catalyze very different dehalogenation reactions (Table 1), even though they utilize the same biocatalytic cofactor and oxidant (H₂O₂). In the case of DHP, the para-substituted halogen is eliminated from the phenyl ring via a two-electron oxidation process, resulting in the elimination of the halide and formation of a quinone. The oxygen atom incorporated into the quinone is from a water molecule. In contrast, besides C-H bond cleavage, the heme TyrH promotes the elimination of the meta-substituted halogen via nonoxidative hydroxylation to generate catechol and halide. In addition, the order of dehalogenation reactivity of heme TyrH inversely reflects the size of the halogen atoms, whereas the opposite is true for DHP. Such contrary results indicate that the governing factor of catalysis is distinct between these two enzymes, which can be explained by their distinct key reactive species, i.e., Cpd I-like species for DHP and ferric-bound hydroperoxo for heme TyrH, respectively.

Table 1.

Comparison of the dehalogenation reactions catalyzed by two histidine-ligated heme enzymes, namely, DHP and heme-dependent TyrH

	DHP	Heme TyrH
Cofactor	Histidine-ligated heme	Histidine-ligated heme
Reaction scheme
Substrate	para-Substituted phenol	meta-Substituted phenol
Product	Oxidative C-X bond cleavage, quinone	Non-oxidative C-X cleavage, catechol
Source of oxygen	H2O	H2O2
Dehalogenation reactivity	Br > Cl > F	F > Cl > 1
Reactive species	Cpd l-like species	Ferric heme-bound hydroperoxo

Open in a new tab

4.1. Tetrahydrobiopterin-dependent aromatic amino acid hydroxylase

Similar to heme-dependent enzymes, nonheme iron enzymes perform a multitude of oxidation chemistries. Like heme TyrH, a nonheme iron-dependent tyrosine hydroxylase (nonheme TyrH) also catalyzes the conversion of _L-tyrosine to DOPA, and such a nonheme TyrH is biologically more significant than its heme counterpart. As a rate-limiting step in the biosynthesis of catecholamines, the reaction requires molecular oxygen and tetrahydrobiopterin (BH₄) as the cosubstrate (Fig. 6B).^87,111 One oxygen atom from molecular oxygen is inserted into the cosubstrate BH₄, generating 4a-hydroxytetrahydropterin (4a-OH-BH₄). The other oxygen atom is incorporated into the aromatic ring of tyrosine. The active enzyme is a tetramer, and each of the subunits contains a ferrous iron coordinated by two histidines and one glutamate as the cofactor, as shown in Fig. 6A.^112,113 In the resting state, water molecules comprise the remaining ligands in a distorted octahedral complex. In some instances, the enzyme is activated by BH₄ and oxygen, but without a bound _L-tyrosine; this enzyme can self-hydroxylate its phenylalanine residue, i.e., Phe300, into 3-hydroxyl-phenylalanine (3-OH-Phe300).¹¹³ This alternative activity is likely to be an enzyme self-protection mechanism, as initially found in TfdA¹¹⁴ as well as other nonheme iron enzymes, to prolong the enzyme activity during undesired and uncoupled oxidation.^115–118 The current understanding of the catalytic mechanism of nonheme TyrH is based on a variety of biochemical and spectroscopic studies.^119–122 The hydroxylation reaction can be divided into two half-reactions. The first half-reaction involves BH₄ hydroxylation to generate 4a-OH-BH₄ and a ferryl-oxo intermediate. The second half-reaction starts with a high-valent iron species oxidizing tyrosine through an electrophilic addition, resulting in an Fe(ii)-bound cationic intermediate.¹²³ Then, the proton is eliminated to liberate the final DOPA product.

When 3-fluorotyrosine is used as a substrate in place of _L-tyrosine, the hydroxylation reaction takes place only on C3 due to the small size of fluorine and consequently produces a defluorinated product, i.e., DOPA (Fig. 6C). Notably, unlike heme TyrH, the reaction yields only the defluorinated product without hydroxylation at the C5 position. This selectivity may indicate that the active site recognizes and interacts specifically with the C3 substituent such that the monosubstituted tyrosine analog has a single binding mode to the active site that only allows defluorination. Other 3-halogenated tyrosine analogs have the same outcome with dehalogenation efficiencies of F > Cl > Br.³³ However, the products derived from halogen atoms and the detailed dehalogenation mechanism remain unknown. The likelihood of the formation of a fluorine cation or a fluorine radical is extremely low in a biological system. If the fluorine substituent is eliminated as fluoride (as seen in most cases), the source of the additional electrons required relative to the C-H bond cleavage requires further investigation. The reaction stoichiometry of BH₄ in the 3-fluorotyrosine reaction may reveal this mystery. It is also unclear if the Fe(iv)-oxo species involved in the hydroxylation of the native C-H bond is responsible for C-F hydroxylation.

Interestingly, 4-halogenated phenylalanine can react with nonheme TyrH, too. With the exception of 4-fluorophenylalanine, which produces only tyrosine, challenging nonheme TyrH with either 4-chloro- or 4-bromophenylalanine generates multiple products as the outcome of the NIH shift and dual hydroxylation sites (C3 and C4). In the case of nonheme TyrH, the NIH shift represents the 1,2-migration of the halogen functional group at the site of substitution in aromatic hydroxylation reactions. In addition to tyrosine, reactions on 4-chloro- and 4-bromophenylalanine also yield 3-halo-tyrosine and 3-hydroxy-4-halo-phenylalanine as the products (Fig. 6D). Nonheme TyrH can hydroxylate at both C3 and C4 positions of 4-halogenated phenylalanine depending on the size of the substituent at C4. It is proposed that hydroxylation at C3 results in 3-hydroxy-4-halo-tyrosine through the reaction route proposed for the hydroxylation of _L-tyrosine. Hydroxylation at the C4 position yields an Fe(ii)-bound cationic intermediate, which can undergo either an NIH shift of the halogen substituent forming 3-halo-tyrosine or direct elimination of the halide yielding tyrosine.^33,124 Here, 4-fluorophenylalanine is used to illustrate the defluorination catalyzed by nonheme TyrH (Fig. 7). After the first half-reaction forming 4a-OH-BH₄ and ferryl-oxo intermediate, electrophilic substitution occurs only at C4 to generate an Fe(ii)-bound cationic intermediate. Without an NIH shift, fluoride atom is directly eliminated to create the re-aromatized product, i.e., tyrosine. For 3-chloro and 3-bromo-substituted phenylalanine analogs, the ratio of hydroxylation at C3 and C4 is nearly 1 : 1. In the case of C4 hydroxylation, a product with more NIH shift (3-halotyrosine) is observed as the size of the halogen atom increases (Br > Cl ≫ F).With halogen elimination, tyrosine is generated (F > Br, Cl). However, the fate of the halogen upon elimination is unresolved, although halide is the most plausible leaving agent of the reaction. There should be an electron source to provide the two necessary electrons and eventually generate halide as the final product. Unfortunately, neither such an electron donor nor a final halogen-containing product has been identified. Further investigation on this system is needed in order to gain a better mechanistic understanding.

The nonheme TyrH has two other siblings. The mononuclear, nonheme, iron-dependent aromatic amino acid hydroxylases, namely, phenylalanine hydroxylase (PheH) and tryptophan hydroxylase (TrpH), have also been well studied. All the three enzymes are pterin-dependent hydroxylases, and they share a significant degree of homology in their catalytic domains. Spectroscopic data have revealed that their catalytic pathways may proceed through a similar Fe(iv)-oxo intermediate.^125–127 The defluorination activity of PheH on 4-fluorophenylalanine to tyrosine has been reported in an earlier study.¹²⁸ The formation of fluoride was detected, and the oxidation of BH₄ was proposed to provide the required electrons, which is a feasible explanation for the questions raised in the TyrH system. Given the defluorination capacities of nonheme TyrH and PheH, it is reasonable to extrapolate that C-F bond cleavage is accessible by TrpH. Future studies regarding the activities and product distribution of fluorinated tryptophan analogs are anticipated.

Overall, the defluorination capacity of the nonheme iron enzyme TyrH is fairly interesting. This enzyme can cleave the C-F bond at different sites when facilitated by oxygen addition via electrophilic substitution. The cationic intermediate is highly destabilized due to fluorine substitution, which promotes re-aromatization to cleave the C-F bond. However, the nature of the fluorine elimination and potential electron source to balance the reaction require additional investigations.

4.2. Nonheme iron hydroxylase with 2-oxoglutarate as the cosubstrate

As discussed in nonheme TyrH that utilizes both BH₄ and tyrosine as organic substrates, several nonheme iron, oxygen-dependent enzymes split dioxygen into two substrates via stepwise incorporation. Such an oxygen-activation method is also employed by a well-studied enzyme superfamily, i.e., 2-oxoglutarate-dependent hydroxylases. These enzymes are also nonheme iron dioxygenases that insert two oxygen atoms into a 2-oxoglutarate (2OG, also known as α-ketoglutarate) molecule and a primary substrate. The catalytic Fe(ii) center is coordinated by a 2-His-1-carboxylate facial triad and three water molecules, and the carboxylate group normally serves as a monodentate ligand.^129,130 The cosubstrate 2OG ligates to the iron center in a bidentate fashion to replace two of the water molecules, which is followed by the binding of the primary substrate nearby to displace the remaining water.

Such a binding order triggers dioxygen ligation and activation, yielding ferric-ion-bound superoxide. The distal oxygen of the superoxide attacks the keto carbon of 2OG to form a bicyclic intermediate with a peroxide bridge. Heterolytic O-O bond cleavage affords an Fe(iv)-oxo species and promotes the oxidative decarboxylation of 2OG to release CO₂. The Fe(iv)-oxo species has been observed by spectroscopic methods.^131–134 After the oxygen transfer to the iron-bound 2OG, the succinate product remains bound to iron in a monodentate manner. The high redox power of Fe(iv)-oxo species allows hydrogen atom abstraction from the adjacent primary substrate, affording ferric hydroxo and a substrate radical.¹³⁵ The second oxygen is delivered by hydroxyl radical rebound to the substrate radical, forming a hydroxylated product. The dissociation of the hydroxylated product and succinate from the active site completes the entire catalytic cycle. It is noteworthy that the process of hydrogen atom abstraction and oxygen rebound in 2OG-dependent hydroxylases is similar to CYP-mediated hydroxylation. Such a typical catalytic mechanism mediated by 2OG-dependent hydroxylases is well understood.^136,137

If the site of hydroxylation has a fluorine substituent on the carbon, the hydroxylation can lead to defluorination after the oxidation of the substrate, as evident in prolyl hydroxylase.^138–140 Prolyl hydroxylase is also a 2OG-dependent hydroxylase, which catalyzes 3- or 4-hydroxylation on proline residues in diverse proteins.¹³⁷ Herein, our discussion will focus on prolyl-4-hydroxylase (P4H). As an important oxygen-sensing mechanism,^141–143 the oxygen-dependent proline hydroxylation on a hypoxia-inducible factor is an irreversible and stereoselective process, which produces (2S,4R)-4-hydroxyproline (Hyp) in the target protein (Fig. 8A). With the multitude of substrate P4H reveals the wide distribution and various biological functions of the Hyp-presented proteins. Thus far, different P4Hs have been found to alter the protein conformation, tune enzymatic stability and activity, prepare for further modifications, as well as promote protein-protein interactions.^144,145 Two biologically essential examples are hypoxia-inducible factor (HIF)- and collagen-related P4Hs; both are considered to be important therapeutic targets.^146–148 Cellular adaptation to hypoxia involves the expression of multiple genes regulated by HIFs. Under hypoxic conditions, low oxygen availability limits the hydroxylation activity of P4H on the prolyl residues of HIFs and therefore prevents signaling for the proteasomal degradation of HIFs.^148–150 Similar to all 2OG-dependent hydroxylases, P4H binds 2OG in a bidentate fashion, and the target proline residues are located nearby (Fig. 8B). In this particular structure, P4H is substituted with manganese and forms a complex with a fragment of HIF with Pro564 to be readily oxidized.¹⁵⁰

Collagen is a dominant protein of the extracellular matrix, which is composed of a three-stranded helix with high tensile strength.¹⁵¹ As a significant connective tissue in the human body, its stability is significantly increased by the formation of Hyp due to the establishment of water bridges and/or stereoelectronic effects.^{34,145,146,151} Studies have shown that the incorporation of (2S,4S)-4-fluoroproline (Flp) into collagen in addition to Hyp results in increased thermostability.^34,152 However, the incorporation of Flp cannot be guaranteed in the presence of P4H due to its defluorination activity (Fig. 8C).^138–140 After the formation of hydroxylated Flp through the typical 2OG-dependent hydroxylation mechanism, fluorine can be spontaneously eliminated from the hydroxylated product, affording (2S)-4-keto-proline (Kep) as the final product (Fig. 9). The elimination of fluoride rather than hydroxide occurs because fluoride is a much better leaving group. The steady-state kinetic parameters of an Flp-containing peptide are comparable to those of the natural substrate, which indicates that the rearrangement of the hydroxylated product to defluorination is not a rate-limiting step. The dehalogenation capacity of P4H has not been determined in halogens other than fluorine.

In addition to P4H, other 2OG-dependent hydroxylases that catalyze defluorination have also been reported, such as γ-butyrobetaine hydroxylase.¹⁵³ The defluorination process is similar to the P4H-mediated reaction, which forms the keto product via fluoride elimination. The use of fluorinated substrate analogs to examine the activity of more enzymes in the 2OG-dependent superfamily seems to be a good strategy for further mechanistic exploration and identification of detoxification activity.

5.1. Rieske nonheme iron dependent dioxygenase

Rieske oxygenases belong to another class of nonheme iron-containing enzyme family that catalyze a wide variety of biotransformations in diverse organisms by utilizing molecular oxygen.¹⁵⁴ The most prominent reaction catalyzed by Rieske oxygenase discussed in the literature is the cis-dihydroxylation of aromatic compounds, which was first identified in the biodegradation pathway of Pseudomonas putida.¹⁵⁵ Besides the same mononuclear iron-based catalytic center (2-His-1-carboxylate coordination) as tetrahydrobiopterin- and 2OG-dependent hydroxylases, Rieske oxygenases also bear an iron-sulfur cluster that is commonly found as a [2Fe-2S] cluster held by two cysteines and two histidines. These enzymes are typically trimeric in structure, and the iron-sulfur cluster mediates the transfer of electrons to the nonheme iron center across two different subunits, which are bridged by either a conserved aspartate or glutamate residue (Fig. 10A).

A cis-dihydroxylation reaction mediated by a Rieske oxygenase enzyme involves a two-electron reduction in the presence of oxygen and NADPH. A reductase extracts two electrons from NADPH to initiate the electron transfer, and the Rieske center shuttles the electrons to the catalytic iron center potentially through the aspartate/glutamate residue.¹⁵⁶ Rieske dioxygenases are capable of oxidizing four major types of aromatic compounds, namely, naphthalene, phenanthrene, benzoate, and toluene/biphenyl.¹⁵⁷ Certain Rieske oxygenases have been reported to exhibit defluorination activity when exposed to fluorinated substrates, such as toluene-1,2-dioxygenase and 2-halobenzoate-1,2-dioxygenase.^29,35 Herein, 2-halobenzoate-1,2-dioxygenase (2HD) is used as an example to illustrate the detailed mechanism of C-F bond cleavage promoted by Rieske dioxygenases. The 2HD enzyme controls the first step of biodegradation of 2-halobenzoate to generate a catechol by eliminating halide and CO₂ (Fig. 10B), providing the substrate for subsequent oxidative ring cleavage by intradiol or extradiol dioxygenase enzymes.^29,158,159

As shown in Fig. 11, the proposed catalytic cycle begins with the binding of an aromatic substrate to dispel a water molecule. The resulting five-coordinated ferrous center promotes oxygen ligation to generate a ferric-ion-bound superoxide intermediate, followed by reduction by the Rieske cluster and subsequent protonation to yield a ferric hydroperoxo species. The ferric hydroperoxo species has oxygen atoms chelated to the iron in a side-on manner, as revealed by the X-ray crystallographic data (Fig. 10A).^160,161 The structural study revealed that a solvent molecule nearby may serve as the proton source.¹⁶⁰ Homolytic O-O bond cleavage coupled with substrate oxidation yields a ferryl-oxo species and a substrate-based radical intermediate with a hydroxyl group added to the primary substrate. The high-valent iron complex further initiates the coupling of the second oxygen to the substrate radical, yielding a ferric alkoxy complex (Fig. 11).^154,162 It is not clear which carbon becomes hydroxylated first, either the α-carbon or the fluorinated carbon, which may rely on their relative distances to the peroxide oxygen atom. The transfer of the second electron from the Rieske cluster and protonation results in the formation of a dihydroxylated product and regeneration of the resting state of the enzyme. Before the product is liberated from the active site, the energetically favorable re-aromatization induces the elimination of CO₂ and fluoride from the ring (Fig. 11). The enzyme is promiscuous with very broad substrate selectivity. Catechol transformation has been observed in the reactions of 2-fluoro-, 2-chloro-, 2-bromo-, and 2-iodobenzoate with decreasing activity,²⁹ which may result from the combined effect of steric hindrance and leaving group efficiency in the process of aromatic dehalogenation.

Another Rieske enzyme capable of C-F cleavage is toluene-1,2-dioxygenase. It has been found that 3-fluoro-substituted benzene underwent defluorination when incubated with toluene-2,3-dioxygenase with the formation of fluoride (Fig. 10C).³⁵ Collectively, Rieske nonheme iron oxygenase adds another member to the metalloenzymes that promote defluorination through hydroxylation.

5.2. Nonheme iron thiol dioxygenase

Cysteamine dioxygenase (ADO) and cysteine dioxygenase (CDO) are the only two known mammalian thiol dioxygenases.¹⁶³ These enzymes share many common structural and functional features. They are nonheme iron-dependent enzymes that oxidize the thiol group of the substrate and insert both oxygen atoms of O₂ into the same sulfur atom, affording sulfinic acids. The thiol dioxygenases function as thiol regulators and oxygen sensors in mammalian cells.^164–166 Structurally, the active site of ADO/CDO is centered at a mononuclear, nonheme ferrous ion coordinated by three histidine residues along with a protein-derived, cysteine-tyrosine (Cys-Tyr) crosslinked cofactor located in the second coordination sphere.^36,167,168 As shown in Fig. 12A, the Cys-Tyr cofactor is covalently crosslinked through a thioether bond between the side chains of a cysteine residue and a tyrosine residue (Cys93 and Tyr157 in human CDO, and Cys220 and Tyr222 in human ADO, respectively). The presence of the Cys-Tyr cofactor increases the catalytic efficiency by stabilizing substrate binding and potential intermediates.^169–172 This post-translational modification has recently been studied through the genetic code expansion method to incorporate an unnatural amino acid, i.e., 3,5-difluorotyrosine (F₂-Tyr), to replace the native tyrosine of the Cys-Tyr cofactor.^36,37 This unnatural amino acid is introduced with the intention of disrupting the formation of a Cys-Tyr crosslink, since the C-F bond is more durable than the corresponding C-H bond, consequently increasing the difficulty of C-S bond formation. However, the enzyme-based oxidant is found to be sufficiently powerful, such that the added C-F bond cannot prevent the self-catalyzed formation of the Cys-Tyr cofactor. The C-F bond is cleaved in the engineered protein, resulting in a monofluorinated Cys-Tyr cofactor and a fluoride anion, as shown in Fig. 12B. This finding may provide a clue as to why some fluorinated compounds are toxic and carcinogenic, even though the C-F bond is thought to be inert. Iron-containing enzymes in humans could unexpectedly dismantle the fluorine-protected molecules, resulting in toxic metabolites before reaching their intended medical targets.

The formation of the Cys-Tyr cofactor requires the presence of dioxygen and the primary substrate in both human thiol dioxygenases. An uncrosslinked ternary complex crystal structure of engineered human CDO, bound with the primary substrate and nitric oxide (Fig. 12C), has provided mechanistic insights into the cofactor biogenesis of these thiol dioxygenases and the C-F bond cleavage mechanism. It has been proposed that a cysteinyl radical generated by the nonheme iron-bound oxidant attacks the tyrosine aromatic ring to initiate the crosslink formation and promote C-F bond cleavage in the case of a fluorinated tyrosine.¹⁷⁰

Fig. 13 shows the mechanism of C-F bond cleavage promoted by CDO. Once the enzyme-substrate complex is formed, oxygen is activated by the iron center to produce an iron-bound superoxide. Subsequent hydrogen atom abstraction from Cys93 by ferric superoxide leads to the formation of a cysteinyl radical and an iron-bound hydroperoxide. Then, F₂-Tyr157 is oxidized by the thiyl radical to form a transient difluoro-tyrosyl radical that is covalently bound to Cys93 through a thioether bond. In order to develop the crosslink, however, the subsequent steps of wild-type and monofluorinated Cys-Tyr cofactor formation are distinct because the elimination of a proton or a fluoride differs by two electrons. In the creation of the native Cys-Tyr cofactor, the O₂ bound to the nonheme iron ion is proposed to be reduced to H₂O₂.¹⁷⁰ While in the creation of monofluorinated Cys-Tyr cofactor, two electrons leave the system with the fluoride anion, and superoxide functions only as a facilitator without being consumed. After the attack by the thiyl radical, the hydroxyl group of the difluoro-tyrosyl radical is deprotonated, forming a semiquinone-like intermediate upon fluorine elimination. One resonance form abstracts a hydrogen atom from the iron-bound hydroperoxide to generate the final, monofluorinated crosslinked cofactor. The iron center returns to the ferric superoxide form to proceed with oxygenation of the substrate, i.e., cysteine, forming cysteine sulfinic acid (CSA) as the product. Similarly, the incorporation of 3,5-dichlorotyrosine can also generate a monochlorinated cofactor with C-Cl bond cleavage; however, the efficiency of cofactor biogenesis is significantly decreased.³⁷

A computational study on the cofactor biogenesis of C-F bond cleavage in F₂-Tyr157-incorporated CDO has been recently reported.¹⁷³ The results, in general, support the mechanism proposed based on the experimental observations shown in Fig. 13.^37,170 The computational work confirms the view that the cofactor biogenesis starts from a thiyl radical on Cys93 rather than Tyr157. The rate-limiting step is identified as the step of C-F bond dissociation with an energy barrier of 18.8 kcal mol⁻¹. An intriguing deviation is proposed in the computational study, according to which an F atom transfers from C3 to C4 in F₂-Tyr157 before the C-F bond cleavage takes place. The necessity of such F-atom migration remains unclear and requires experimental verification. This computational study also highlights the electrostatic influences of seven active-site residues on C-F bond cleavage. It is further suggested that the mutagenesis of Asp87 and Phe165 could improve the defluorination efficiency by lowering the energy barrier.¹⁷³

In addition to CDO and ADO, there are other thiol dioxygenases, such as 3-mercaptopropionic acid dioxygenase (MDO) and 2-mercaptosuccinic acid dioxygenase, which are equipped with a similar protein scaffold and iron center, i.e., a mononuclear iron coordinated by three histidines.^174–177 Although the Cys-Tyr cofactor is not always present in other thiol dioxygenases, such a crosslink could be generated by a specific mutation, e.g., G95C in MDO leads to the formation of an artificial Cys-Tyr cofactor.¹⁷⁸ It would be interesting to investigate whether MDO G95C and other thiol dioxygenases can cleave the aromatic C-F bonds from exogenous ligands or their engineered Cys-Tyr cofactors.