HHS Public Access
`Author manuscript
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Published in final edited form as:
`Nat Prod Rep. 2018 August 15; 35(8): 792–837. doi:10.1039/c7np00067g.
`
`Recent Examples of α-Ketoglutarate-Dependent Mononuclear
`Non-Haem Iron Enzymes in Natural Product Biosyntheses
`
`Shu-Shan Gaoa,†, Nathchar Naowarojnab,†, Ronghai Chengb,†, Xueting Liub,c,*, and Pinghua
`Liub,*
`aState Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of
`Sciences, Beijing 100101, China
`
`bDepartment of Chemistry, Boston University, Boston, MA, 02215, USA
`
`cState Key Laboratory of Bioreactor Engineering, East China University of Science and
`Technology, Shanghai 200237, China
`
`Abstract
`α-Ketoglutarate (αKG, also known as 2-oxoglutarate)-dependent mononuclear non-haem iron
`(αKG-NHFe) enzymes catalyze a wide range of biochemical reactions, including hydroxylation,
`ring fragmentation, C-C bond cleavage, epimerization, desaturation, endoperoxidation and
`heterocycle formation. These enzymes utilize iron (II) as the metallo-cofactor and αKG as the co-
`substrate. Herein, we summarize several novel αKG-NHFe enzymes involved in natural product
`biosyntheses discovered in recent years, including halogenation reactions, amino acid
`modifications and tailoring reactions in the biosynthesis of terpenes, lipids, fatty acids and
`phosphonates. We also conducted a survey of the currently available structures of αKG-NHFe
`enzymes, in which αKG binds to the metallo-centre bidentately through either a proximal- or
`distal-type binding mode. Future structure–function and structure–reactivity relationship
`investigations will provide crucial information regarding how activities in this large class of
`enzymes have been fine-tuned in nature.
`
`Graphical Abstract
`Proximal- and distal-type αKG binding to the Fe(II) centre might play a crucial role in fine-tuning
`the catalysis of αKG-dependent non-haem iron enzymes.
`
`†These authors contributed equally
`9 Conflicts of Interest
`There are no conflicts to declare.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Inari Ex. 1017
`Inari Agric. v. Corteva Agriscience
`PGR2023-00022
`Page 00001
`
`

`

`Gao et al.
`
`Page 2
`
`1 Introduction
`
`In nature, α-ketoglutarate-dependent mononuclear non-haem iron (αKG-NHFe) enzymes
`catalyse a wide range of biochemical reactions. αKG-NHFe enzymes require iron (II) as a
`metallo-cofactor and αKG as a co-substrate.1–6 Throughout several decades, various aspects
`of this large enzyme superfamily have been summarised in many excellent reviews,
`including the crystal structures and mechanisms,6–13 reaction diversity and natural product
`biosynthetic pathways,1, 5, 6, 8, 14–24 mechanistic investigations using small molecular model
`systems6, 25 and their relevance to biological processes and human diseases.4–6, 26 Members
`of this superfamily are widely distributed across prokaryotes, eukaryotes and archaea.
`Among the pathways involving αKG-NHFe enzymes, those involved in antibiotic
`biosynthesis are some of the most extensively investigated areas, and several reviews have
`been devoted to penicillin, cephalosporin, cephamycin and clavam biosyntheses.27–30 Some
`pathways with pharmaceutical or agricultural relevance have also been summarised,
`including the biosynthesis of ethylene,31, 32 carnitine,33 collagen34 and coumarin.16 Some
`αKG-NHFe enzymes are known to be related to human diseases, including phytanoyl-CoA
`hydroxylase (PAHX) in Refsum disease,35 4-hydroxy-phenylpyruvate dioxygenase (4-
`HPPD) in tyrosinaemia type II and hawkinsinuria,36, 37 prolyl hydroxylase (P4H) in
`alcoholic liver cirrhosis38, 39 and lysyl hydroxylase (LH) in Ehlers–Danlos syndrome type
`VI.38 Knowledge gained from mechanistic characterisations of αKG-NHFe enzymes has
`been applied to guide inhibitor design and development, which was recently summarised by
`Schofield et al.26
`
`αKG-NHFe enzymes catalyse reactions as wide as those catalysed by haem-containing
`enzymes. Besides mechanistic investigations on enzymatic systems, studies on small
`molecular model systems are also one of the key sources for our mechanistic understanding
`of the catalytic processes catalysed by αKG-NHFe enzymes.6, 25 Based on the initial
`mechanistic proposal by Hanauske-Abel and Günzler,40 together with experimental and
`computational data accumulated over the last few decades, a generic mechanism for αKG-
`mediated oxygen activation was proposed involving Fe(IV)=O species as one of the key
`intermediates (Fig. 1A).6–12, 41 Starting from Fe(IV)=O species, hydroxylation is the most
`common type of reaction (e.g. hydroxylation of taurine 1 catalysed by TauD, Fig. 1B)
`catalysed by αKG-NHFe enzymes.42–47 In the absence of sulfate under aerobic conditions,
`Escherichia coli can utilize aliphatic sulfonates as sulfur sources. TauD and FMNH2-
`dependent SsuD are the key enzymes in this process.48, 49 TauD oxidises taurine 1 to 1-
`hydroxy-2-aminoethanesulfonic acid 2, which then spontaneously decomposes to
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00002
`
`

`

`Gao et al.
`
`Page 3
`
`aminoacetaldehyde 3 and sulfite (Fig. 1B).42–47 In addition to taurine 1, TuaD can oxidise
`taurine analogues, including pentanesulfonic acid, 3-(N-morpholino)propanesulfonic acid
`and 1,3-dioxo-2-isoindolineethanesulfonic.50 Several lines of evidence, including a
`combination of stopped-flow UV-visible absorption spectroscopy,45, 46, 51 EXAFS and
`Mössbauer spectroscopy,45, 46, 51–56 and isotope labelling,45, 57 support the involvement of
`an Fe(IV)=O species (A-6, Fig. 1A) as a kinetically competent intermediate. In
`hydroxylation reactions catalysed by αKG-NHFe enzymes, the reactive Fe(IV)=O species
`abstracts a hydrogen atom from the substrate to generate a substrate-based radical, and
`simultaneously, the Fe(IV)=O intermediate is reduced to the Fe(III)-OH species (A-6 →
`A-7, Fig. 1A). A subsequent hydroxyl radical rebound completes the substrate hydroxylation
`reaction and the αKG-NHFe enzyme returns to its initial Fe(II) state (A-7 → A-8, Fig. 1A).
`The proposed Fe(IV)=O intermediate (species A-6, Fig. 1A) was first trapped and
`characterised in TauD studies (Fig. 1B).45, 51 Since then, this key intermediate has been
`trapped and characterised in a few other enzymatic and model systems.54, 58–63 In addition
`to hydroxylation reactions, in natural product biosyntheses, many other types of reactions
`have also been attributed to αKG-NHFe enzymes, including desaturation,64 ring formation,
`65 ring expansion,66 halogenation,67 endoperoxidation68 and carbon skeleton
`rearrangements.69 αKG-NHFe enzymes also participate in modifications and repairs of
`macromolecules (DNA, RNA and proteins).6
`
`In 2013, Hangasky et al. reported a survey of 25 αKG-NHFe structures deposited in the
`protein data bank (PDB) shown as part of Table 1.9 αKG-NHFe enzymes possess a double-
`stranded β-helical fold (DSBH fold), which has also been called a cupin or jelly-roll fold.
`70, 71 The DSBH fold (Fig. 2A) was first observed in the crystallographic studies of
`isopenicillin N synthase (IPNS). Intriguingly, IPNS-catalysis did not require αKG as a co-
`substrate, and all four electrons required for catalysis were shown to be from its substrate.
`72–74 αKG-NHFe enzymes share a conserved His-X-Asp/Glu-Xn-His (2-His-1-carboxylate)
`motif, in which two His residues and a Glu or an Asp residue serve as the ligands to the iron
`centre.70, 75 Exceptions to this 2-His-1-carboxylate facial triad have also been reported, e.g.
`αKG-NHFe halogenases.76 In most αKG-NHFe enzymes, the metallo-centres and their
`ligands form octahedral complexes.1, 70, 71, 75 αKG coordinates to the NHFe centre
`bidentately using its C2 keto oxygen and C1 carboxylate as the ligands replacing two water
`molecules. In most reported structures, in addition to its bidentate interactions with the
`metallo-centre, αKG also interacts with a basic residue in the active site (e.g. Arg or Lys)
`through electrostatic interactions using its C5 carboxylate, which facilitates the positioning
`of αKG in the active sites.6
`
`Two different αKG binding modes have been observed upon inspection of the structure in
`PDB: the proximal and distal types.9, 71 In the proximal-type αKG binding mode (e.g.
`TauD•Fe(II)• αKG binary complex in Fig. 2B), the αKG C1 carboxylate coordinates to the
`Fe(II) centre at a position trans to the first histidine (His99), while the αKG C2 keto is at a
`position opposite to the acidic ligand Asp101 of the 2-His-1-carboxylate facial triad. In
`Table 1, we labelled this type of αKG binding mode as the proximal type. Upon substrate
`binding, the remaining water dissociates from the Fe(II) centre opening up the site for
`oxygen binding and activation (A-3, Fig. 1A). In a typical hydroxylation reaction, this
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00003
`
`

`

`Gao et al.
`
`Page 4
`
`oxygen binding site is adjacent to the substrate binding site.77 The resulting Fe(IV)=O
`species (A-6, Fig. 1A) points towards the substrate, allowing direct oxidation of the substrate
`by Fe(IV)=O species.
`
`Interestingly, in nearly ~50% of known αKG-NHFe structures (Table 1), αKG coordinates
`to the Fe(II) centre in a conformation different from the proximal type observed in TauD
`(Fig. 2B). For example, in the FtmOx1•Fe(II)• αKG binary complex shown in Fig. 2C, αKG
`displays a bidentate coordination to the Fe(II) centre and its C2 keto is at a position opposite
`to an acidic residue (Asp131 in Fig. 2C). However, in this structure, the C1 carboxylate of
`αKG coordinates to the Fe(II) centre at a position opposite to the distal histidine (His205) of
`the 2-His-1-carboxylate facial triad (Fig. 2C).68 This type of αKG binding mode is termed
`distal (Fig. 2C). Due to the change in αKG binding conformation relative to that of TauD,
`the remaining site for O2 binding and activation in FtmOx1 is not adjacent to the substrate
`binding pocket. Following the generic mechanistic model discussed in Fig. 1A for αKG-
`mediated oxygen activation, the Fe(IV)=O species formed (Fig. 2C) points away from the
`substrate binding site, making it inaccessible for subsequent chemical transformation.
`
`Two scenarios have been proposed to explain the catalytic processes in distal-type αKG-
`NHFe enzymes. In the first scenario, as exemplified by the endoperoxidation reaction
`catalyzed by FtmOx1 (Figs. 2C),68 αKG coordinates to the Fe(II) centre using the distal
`binding mode and the oxygen binding site is not adjacent to the substrate verruculogen.
`However, a tyrosine residue (Y224) is next to the oxygen binding site and is crucial to the
`endoperoxidation reaction (see Section 3.6.6).68 Alternatively, αKG can undergo a
`conformational switch from the distal to proximal mode, re-orienting the oxygen binding
`and activation site towards the substrate. In clavaminic acid synthase (CAS), two αKG
`binding modes have been observed experimentally (Figs 2D & 2E).78 In the absence of NO,
`the C1 carboxylate of αKG coordinates to the Fe(II) centre from a site opposite to that of the
`proximal histidine (His144, Fig. 2D). Interestingly, upon the introduction of NO to the
`CAS•Fe(II)• αKG•substrate complex (Fig. 2E), the C1 carboxylate is now positioned
`opposite to the distal histidine (His279) of the His-X-Asp/Glu-Xn-His facial triad. Two αKG
`binding modes (distal and proximal) have also been observed for AlkB,80 which catalyzes
`oxidative DNA demethylation. The presence of two αKG binding modes in both CAS and
`AlkB has led to the proposed αKG conformational switch (Fig. 2F), which is necessary to
`properly orient the Fe(IV)=O species towards the substrate for catalysis.71
`
`Early in 2017, two αKG binding modes were also observed for the ethylene-forming
`enzyme (EFE).79 In one structure, αKG displays a bidentate coordination to the Fe(II)
`centre in the distal-type binding mode.79, 81 Interestingly, in another EFE structure, αKG
`displays a monodentate coordination to the Fe(II) centre using its C5 carboxylate as the
`ligand (Fig. 2G).79 It is not yet known whether this binding mode is relevant to EFE
`catalysis.
`
`In natural product biosynthesis, reactions catalyzed by αKG-NHFe enzymes are widely used
`to either produce biosynthetic precursors or to modify the natural product skeletons after
`assembly. In this review, we summarize recent examples of αKG-NHFe enzymes involved
`in the modification of amino acids, and biosynthesis of terpenes, lipids and phosphorous-
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00004
`
`

`

`Gao et al.
`
`Page 5
`
`containing secondary metabolites. The materials covered here complement many of the
`recent reviews in this area, especially a recent book on αKG-dependent oxygenases.6
`Halogenases, which can install a halogen atom, Cl− or Br−, to an inactivated carbon centre
`are a subset of the αKG-NHFe enzyme superfamily and are covered in Section 2. Section 3
`is devoted to αKG-NHFe enzymes involved in amino acid modifications, and are prevalent
`in the biosynthesis of several types of natural products. Terpenes are one the largest classes
`of natural products.82 After assembly of their skeleton, extensive modifications are
`introduced to produce the final products. Some of the required tailoring reactions are
`catalyzed by αKG-NHFe enzymes (Section 4). Phosphonates are C-P bond-containing
`natural products with great pharmaceutical potential due to their structural similarity to
`phosphates and carboxylates.83 Phosphonate biosynthesis involves many novel reactions,
`some of which are mediated by αKG-NHFe enzymes and are described Section 5. In
`addition, Sections 6 and 7 are devoted to αKG-NHFe enzymes involved in lipid and fatty
`acid modifications and nucleoside antibiotics biosynthesis, respectively.
`
`In Table 1, we list all of the αKG-NHFe enzymes discussed in this review and an additional
`25 enzymes covered by Hangasky et al. in their structural analysis of αKG-NHFe enzymes
`previously deposited in the PDB.9 For enzymes listed in Table 1, when structural
`information was available, they have been classified as either the proximal or distal type to
`indicate their αKG binding mode. Interestingly, the number of proximal- and distal-type
`αKG-NHFe enzymes are approximately equal, indicating that both binding modes are
`common in nature. Thus far, most of the mechanistic information on αKG-NHFe enzymes is
`based on the characterization of proximal-type αKG-NHFe enzymes. Distal-type αKG-
`NHFe enzymes have to change the binding conformation of αKG to properly orient the
`Fe(IV)=O species for catalysis, and this conformational switch may be a critical mechanistic
`feature. In cases where a conformational switch is not employed, it is highly possible that
`nature explores the uniqueness of the distal-type binding mode to mediate novel chemical
`transformations (e.g. FtmOx1 catalysis, Fig. 2C). Future mechanistic characterizations of
`more proximal-type αKG-NHFe enzymes will provide answers to these questions.
`
`2 Halogenation
`
`In this section, we summarize some recent examples of halogenation reactions catalyzed by
`αKG-NHFe enzymes. Unlike flavin-dependent halogenases, which catalyze halogenation at
`aromatic or electron-rich carbons,20, 67, 118, 119 αKG-dependent halogenases perform much
`more challenging chemical transformations, catalyzing halogenation reactions at aliphatic
`carbons.120, 121 Most of the halogenases characterized to date act on substrates tethered to
`the phosphopantetheinyl arm of carrier proteins (Sections 2.1 and 2.2).6 In recent years,
`halogenases using stand-alone small molecules as the substrates have also been discovered
`(Section 2.3).
`
`Several crystal structures of αKG-dependent halogenases have been reported, including
`SyrB2, CytC3, CurA and WelO5 (Table 1). Interestingly, they do not have the typical 2-
`His-1-carboxylate “facial triad” as observed in other αKG-NHFe enzymes. Instead, the
`carboxylate ligand of the 2-His-1-carboxylate facial triad is replaced by a halide ligand,
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00005
`
`

`

`Gao et al.
`
`Page 6
`
`which together with some other active site interaction network modifications, is key to the
`selectivity of these halogenases.76, 95, 96, 113, 117
`
`2.1 Halogenation on carrier protein-tethered substrates
`
`SyrB2 catalyzes monochlorination of the methyl group of L-Thr-SyrB1 4 to 4-Cl-L-Thr-
`SyrB1 5, which is one of the steps in the biosynthesis of the phytotoxin syringomycin E 6
`(Fig. 3A), using SyrB1 as the carrier protein.113, 120, 122 Wild-type SyrB2 can also catalyze
`aliphatic nitration or azidation reactions.123 Drennan et al. reported the first structure of the
`αKG-NHFe halogenase SyrB2 and showed that its Fe(II) centre has 2-His and 1-chloride
`ligand, instead of the classic 2-His-1-carboxylate facial triad.113
`
`SyrB2 undergoes oxygen activation similar to other αKG-NHFe enzymes (Fig. 1A),
`generating an Fe(IV)=O species (Fig. 3B).113 The Fe(IV)=O species (B-4, Fig. 3B) then
`abstracts a hydrogen atom from the substrate to create a substrate-based radical (B-5, Fig.
`3B). Subsequently, the chlorine atom combines with the substrate-based radical, instead of
`going through a hydroxy-rebound to form a hydroxylation product, resulting in the
`formation of a chlorinated product (B-5 → B-1, Fig. 3B).60
`
`The biosynthesis of barbamide 11 (Fig. 4A) involves two αKG-NHFe halogenases, BarB1
`and BarB2, which work in tandem to trichlorinate the C5 methyl group of L-Leu-S-BarA 7.
`In these chlorination reactions, BarA is the carrier protein (Fig. 4A), and BarB2 chlorinates
`either L-Leu-S-BarA 7 or monochloro-Leu-S-BarA 8 to dichloro-leu-S-BarA 9.
`Interestingly, BarB1 can convert both mono- and di-chlorinated L-Leu-S-BarA (8 and 9) to
`(2S,4S)-5,5,5-trichloro-Leu-S-BarA 10 (Fig. 4A).124 CytC3 catalyzes the chlorination of
`aminobutyryl-S-CytC2 12 during biosynthesis of the Streptomycete antibiotic
`dichloroaminobutyrate 15, and functions in a similar manner to SyrB2, BarB1 and BarB2.
`Both γ-chloro- and γ, γ-dichloroaminobutyryl-S-CytC2 (13 and 14, Fig. 4B) are products
`of CytC3 catalysis.125
`
`In addition to the aforementioned cases where chlorinase is part of the non-ribosomal
`peptide synthetase (NRPS) machinery, chlorination has also been reported as a tailoring
`reaction for the biosynthesis of other natural products. HtcB is a fatty acyl halogenase
`involved in the biosynthesis of hectochlorin 20. The HtcB protein from Lyngbya majuscule
`contains three domains: an N-terminal catalytic αKG-NHFe halogenase domain, an acyl
`coenzyme A binding domain and an acyl carrier protein (ACP) domain.126 When ACP-
`tethered hexanoyl 16 was used as a substrate, 5-oxo- 17, 4-ene-5-chloro- 18 and 5,5-
`dichloro-haxanoyl-S-ACP 19 were all observed as products in HctB catalysis (Fig. 4C).128
`After this tailoring reaction, 5,5-dichloro-haxanoyl-S-ACP 19 was utilized as one of the
`building blocks for the biosynthesis of hectochlorin 20.
`
`In the above cases, amino acids tethered to the phosphopantetheinyl arm of carrier proteins
`or an acyl group tethered to an ACP domain were used as substrates by αKG-NHFe
`halogenases. KthP is an αKG-NHFe enzyme involved in the biosynthesis of kutzneride 2
`(23). For the chlorination step in the production of kutzneride 2 (23), a piperazyl functional
`group was tethered to the thiolation protein KtzC first (21, Fig. 4D), and KthP then stereo-
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00006
`
`

`

`Gao et al.
`
`Page 7
`
`and regio-selectively chlorinated the tethered piperazyl ring to generate (3S,5S)-5-
`chloropiperazate-S-KtzC 22 (Fig. 4D).127
`
`2.2 Halogenation-initiated formation of cyclopropanes
`
`Chlorination is employed by several natural product biosynthetic pathways to activate C-H
`bonds and the resulting halides are then used for the construction of other functional groups,
`e.g. cyclopropane. During biosynthesis of the Pseudomonas syringae phytotoxin coronatine
`28 (Fig. 5A), CmaB, an αKG-NHFe halogenase, chlorinates L-allo-isoleucine-S-CmaD 24
`to γ-chloro-L-allo-isoleucine-S-CmaD 25 (Fig. 5A). Subsequent intramolecular γ-
`elimination is mediated by a zinc-dependent enzyme, CmaC, resulting in the formation of
`coronamic acid-S-CmaD 26 (Fig. 5A). Coronamic acid (CMA, 27) is then released from
`CmaD and used as a building block for the biosynthesis of coronatine 28 (Fig. 5A).121
`
`An analogous pathway of coronatine biosynthesis is present in the biosynthesis of
`kutzneride 2 (23, Fig. 5B). In this process, KtzD, an αKG-NHFe enzyme, chlorinates the γ
`position of L-Ile-S-KtzC 29 tethered to a KtzC carrier protein to generate γ chloro-L-allo-
`Ile-S-KtzC 30. Subsequent cyclopropyl ring formation mediated by the flavoprotein KtzA
`affords (1S, 2R)-allo-CMA 31, which is in contrast to the zinc-dependent protein in the
`coronatine biosynthesis (Fig. 5B).129 It is worth mentioning that, the cyclopropyl group
`present in the (1S, 2R)-allo-CMA-S-KtzC intermediate 31 (Fig. 5B)129 is structurally
`distinct from the 2-(2-methylcyclopropyl)glycine moiety in kutznerides, e.g. kutzneride 2
`(23, Fig. 5B).130
`
`CurA and JamE are two polyketide synthase (PKS) megasynthases involved in the Lyngbya
`majuscule curacin 37131 and janmaicamide 39132 biosynthetic pathways, respectively (Fig.
`6A). An αKG-NHFe halogenase (Hal) domain is embedded in the CurA megasynthase, and
`plays a similar role to that of CmaB and KtzD in the construction of a cyclopropyl group.
`The Hal domain of CurA chlorinates (S)-3-hydroxy-3-methylglutaryl-ACP ((S)-HMG-ACP,
`32) to produce 33. Then, a dehydratase (ECH1) domain embedded in CurE catalyzes the
`dehydration of 33 to yield 3-methylglutaconyl-ACP 34. Subsequent decarboxylation
`mediated by a decarboxylase (ECH2) domain embedded in CurF affords an α, β-enoyl
`thioester, 3-methylcrotonyl-ACP 35. Finally, an enoyl reductase (ER) domain embedded in
`CurF catalyzes the cyclopropanation of 35 to 36, which then serves as a building block in the
`biosynthesis of curacin A 37 (Fig. 6A).133
`
`A similar biosynthetic route is present in the jamaicamide A pathway. Here, the halogenase
`domain embedded in JamE catalyzes the same chlorination step on 32 as that of CurA in
`curacin A biosynthesis (Fig. 6A). However, in contrast to the α, β-enoyl thioester-based
`product 35 from the CurF ECH2 domain, the JamJ ECH2 domain decarboxylates 34 to
`produce the β, γ-enoyl thioester intermediate 38 in the biosynthesis of jamaicamide 39 (Fig.
`6A).132, 134
`
`Several crystal structures of the CurA halogenase domain (Hal) in different ligand states
`have been reported.95 Two conformation states exist, namely an open and a closed state
`(Figs 6B & 6C), and the transition between the two states is triggered by αKG binding. A
`large lid covers the active site in the closed form, which is disordered in the open form.
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00007
`
`

`

`Gao et al.
`
`Page 8
`
`Upon αKG binding, CurA Hal undergoes a conformational change from the open to closed
`form, facilitating the substrate (S)-HMG-ACP binding (Figs 6B & 6C). Additionally, CurA
`Hal exhibits a high degree of substrate specificity, requiring both the C3 S-hydroxyl and C5
`carboxylate on (S)-HMG-ACP 32 for recognition. It has been suggested that the
`halogenation vs hydroxylation outcome in these halogenase reactions depends on the
`substrate positioning. The conformational switch triggered by αKG binding may allow the
`precise positioning of the substrate in the active site, thereby minimizing the competing
`hydroxylation reaction.95
`
`2.3 Halogenation on freestanding substrates
`In most early descriptions of αKG-NHFe halogenase reactions (sections 2.1 & 2.2), their
`substrates were covalently tethered to carrier proteins. These halogenases do not recognize
`stand-alone small molecule as a substrate. New information regarding their substrate
`specificity was discovered recently. WelO5, an enzyme involved in welwitindolinone
`biosynthesis, is the first reported αKG-NHFe chlorinase that utilizes a stand-alone small
`molecule as its substrate.117, 135, 136 WelO5 stereo-specifically chlorinates hapalindole-type
`molecules (40 → 41, Fig. 7A).117, 135, 136 In addition to chloride, WelO5 can use other
`halides, including bromide, as alternative halogenation agents.137
`
`Recently, the structure of the WelO5•αKG•substrate complex was reported (Fig. 7B).117 In
`this structure, after αKG and substrate binding, the vacant ligand site in the Fe(II) centre is
`directly facing the substrate. If this is the site for O2 binding and activation and no αKG
`rearrangement is involved, then the oxo group, instead of the chlorine group of the halo-oxo-
`iron(IV) intermediate, faces the substrate. To explain the chlorination activity, it was thus
`suggested that a switch in αKG conformation was required. A second-coordination shell
`residue, Ser189, was suggested to play a key role in controlling such an αKG
`conformational switch (Fig. 7B). This hypothesis is supported by results from the WelO5
`S189A mutant, which produces an equal amount of halogenation and hydroxylation
`products.117
`
`AmbO5 is another recent example of such an αKG-NHFe halogenase (Fig. 7A). AmbO5 has
`a high degree of substrate flexibility and catalyzes the chlorination of ambiguine (44, 46,
`50), fischerindole (40, 42) and hapalindole 52 alkaloids (Fig. 7A).138 Comparative analysis
`of AmbO5 and WelO5 revealed that a fragment of the C-terminal portion of WelO5 might be
`important for substrate selection and specificity.138 Indeed, replacing a fragment of 18
`residues in the WelO5 C-terminus with the corresponding AmbO5 sequence expanded the
`substrate scope of WelO5 catalysis.138
`
`Many halogenated natural products exhibit biological activities. For example,
`salinisporamide shows anti-cancer activity,139 while syringomycin functions as an anti-
`fungal agent.140 Halogenation is critical for the biological activities of these compounds, and
`as a result, developing new halogenation strategies continues to be a key area of interest.
`20, 67 In the last few decades, while many investigations have focused on the structural and
`mechanistic characterizations of these halogenases, some efforts have been devoted to
`developing new halogenases. Several groups have attempted the conversion of a hydroxylase
`to a halogenase and vice versa.113, 117 One of the successful examples is the WelO5 S189A
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00008
`
`

`

`Gao et al.
`
`Page 9
`
`mutant, which has exhibited a relaxed selectivity relative to wild-type WelO5, producing an
`equal proportion of hydroxyl and chlorine-modified products.117 Recently, Boal et al. also
`engineered an N-acyl amino acid hydroxylase, SadA, into a halogenase.141
`
`3 Amino Acid Modifications
`
`The presence of non-proteinogenic amino acids in alkaloids and non-ribosomal peptides is
`very common and, in many cases, these unnatural amino acids are supplied through
`dedicated biosynthetic pathways. Alternatively, after natural product skeleton assembly
`using 21 proteinogenic amino acids, additional structural diversity is then introduced
`through extensive modifications by tailoring enzymes. αKG-NHFe enzymes are one of the
`most common types of tailoring enzymes. These αKG-NHFe enzymes often show some
`degree of substrate promiscuity, readily incorporating substrate analogues into natural
`products through pathway engineering or by in vitro biocatalytic processes. For some
`valuable compounds, these αKG-NHFe enzyme-mediated biocatalytic transformations may
`have advantages relative to their synthetic organic pathways. In this section, some recent
`amino acid tailoring reactions are summarized.
`
`3.1 αKG-NHFe enzymes in carnitine biosynthesis
`L-Carnitine 59 plays a key role in fatty acid metabolism in all animals and in some
`prokaryotes. As a result, the enzymes involved in carnitine biosynthesis have been explored
`for therapeutic purposes.146, 147 Carnitine biosynthesis148–150 (Fig. 8A) begins with
`trimethylated lysine 55, and involves the following reactions: Nε-trimethyllysine
`hydroxylase (TMLH), 3-hydroxy-Nε-trimethyllysine aldolase (HTML aldolase), 4-N-
`trimethylaminobutyraldehyde dehydrogenase (TMABA dehydrogenase) and γ-
`butyrobetaine hydroxylase (BBOX). Both TMLH and BBOX are αKG-NHFe enzymes.
`TMLH catalyzes the stereo-selective conversion of (2S)-Nε-trimethyllysine 55 to (2S,3S)-3-
`hydroxy-Nε-trimethyllysine 56 (Fig. 8A),142, 151, 152 one of the key reactions in carnitine
`biosynthesis.148–150 HTML aldolase catalyzes the cleavage of 56 into 4-N-
`trimethylaminobutyraldehyde 57 and glycine using pyridoxal phosphate (PLP) as a cofactor.
`Then, TMABA dehydrogenase, an NAD+-dependent enzyme, oxidizes 4-N-
`trimethylaminobutyraldehyde 57 to γ-butyrobetaine (γ-BB) 58. The last step of this
`pathway is the BBOX-catalyzed hydroxylation of 58 to L-carnitine 59.
`
`TMLH has some degree of substrate flexibility and can accept several trimethyllysine
`analogues as alternative substrates (60–63, Fig. 8B), catalyzing the production of their
`corresponding hydroxylation products (64–67, Fig. 8B).143 TMLH’s substrate flexibility is
`reflected in at least two aspects: the chain length of the amino acid side-chain and the alkyl
`group on the lysine ε-nitrogen. Thus far, TMLH’s crystal structure has not been reported.
`However, the substrate promiscuity of TMLH implies that its active site pocket for the lysine
`side-chain binding is relatively flexible, and could be explored for the stereo-selective
`hydroxylation of substrate analogues for synthetic purposes. Recently, many αKG-NHFe
`enzymes involved in histone demethylation (e.g. the demethylation of methylated lysine
`residues) have been discovered and discussed in depth in a book edited by Schofield and
`
`Nat Prod Rep. Author manuscript; available in PMC 2019 August 15.
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`Author Manuscript
`
`PGR2023-00022 Page 00009
`
`

`

`Gao et al.
`
`Page 10
`
`Hausinger.6 Future structural work may also provide information on how this class of
`enzymes control regioselectivity.
`
`BBOX is an αKG-NHFe enzyme involved in the last step of L-carnitine biosynthesis,
`hydroxylating γ-butyrobetaine (γ-BB, 58) to L-carnitine 59 (Fig. 8A).153, 154 Because of the
`importance of L-carnitine in fatty acid metabolism, BBOX has been explored as a target to
`develop treatments for myocardial infarction.147 Interestingly, BBOX catalyzes the oxidation
`of its inhibitor 3-(2,2,2-trimethylhydrazinium)propionate (THP) 68 and produces multiple
`products, including 68g–68j, and 3-amino-4-(methylamino)butanoic acid (AMBA) 69 (Fig.
`8C).144 The oxidation of THP 68 involves N–N bond cleavage and C–C bond formation,
`which is likely achieved via a process related to a Stevens rearrangement (Stevens [1,2]-
`shift).155 The Fe(IV)=O intermediate abstracts a hydrogen atom from THP 68 to generate a
`radical intermediate 68a, which is followed by N–N bond cleavage and a [1,2]-shift to
`produce 68d (Fig. 8C). Two potential pathways (Pathways I and II) have been proposed for
`the subsequent steps (Fig. 8C). In pathway I, N-methyl hydroxylation followed by
`spontaneous decomposition of the hydroxylation product 68f affords the formaldehyde 68h.
`On the other hand, in pathway II, the methylene radical reacts with the imine 68b and the
`subsequent [1,5] H-shift results in a radical intermediate 68l. The hydroxyl radical rebounds
`to the intermediate 68l, generating the hydroxylation product 70 (acyclic aminal), which
`might be in equilibrium with its cyclic aminal form 71. Compound 70 is further converted to
`the final product AMBA 69 after eliminating the formaldehyde moiety (Fig. 8C).144
`
`The structure of BBOX in complex with an αKG analogue, N-oxalylglycine (NOG) and γ-
`BB has been reported (Fig. 8D).92 In this structure, the αKG analogue coordinates to the
`iron centre in the d