Skip to Main Content
Skip Nav Destination

Regioselective halogenation of organic compounds using non-toxic halide salts and oxygen in aqueous media at ambient temperatures – a tough task for conventional chemical synthesis. However, thousands of halogenated natural products are formed by biosynthesis under such conditions without the need for reagents like molecular chlorine or bromine. Nature has evolved a variety of different enzymes able to halogenate their substrates by using Cl, Br, I or even F ions. In many cases, these reactions are highly regioselective and may address even electronically disfavoured positions. Unsurprisingly, the interest in halogenases continuously increases in the context of green chemistry to use them as biocatalysts for chemical synthesis. This review focusses on the class of flavin-dependent halogenases and their applications, while insights on their structure and mechanism are being discussed. Recent progress in utilising these enzymes for incorporation of halogen substituents in complex natural products by enzyme catalysis or fermentation is highlighted. Modification of the produced organohalogen compounds, for example in chemoenzymatic one-pot reaction cascades are described. Since the narrow substrate scope of flavin-dependent halogenases is still a crucial bottleneck, the current progress to obtain optimised and tailor-made biocatalysts by directed evolution of this enzyme class is highlighted.

The incorporation of halogen substituents into organic scaffolds is an important and valuable reaction in organic synthesis. Halogenated compounds form important intermediates in chemical,1  agrochemical,2  and medicinal3  research, since the carbon–halogen bond can be transformed chemically by e.g. nucleophilic substitution or metal-catalysed cross-coupling reactions. These reactions offer the unique opportunity to form complex structures, for analytical purposes4  up to industrial scale synthesis.5  In addition to the synthetic utility of halogenated intermediates, the incorporated halogen may dramatically affect the biological properties of an active pharmaceutical ingredient (API) or an agrochemical. Nowadays, up to 60% of all pharmaceuticals and agrochemicals on the market are halogenated.6–8  Halogenation exerts a major impact on the pharmacokinetic properties of an API by enhancing its metabolic stability toward cytochrome P450-catalysed oxidations.9  In addition, the biological activity of an API is in many cases directly linked to the halogen substituent. For instance, the absence of the chloro substituent in the natural antimitotic agent cryptophycin results in a drop of biological activity by a factor of 8.10 

However, the introduction of halogen substituents by conventional chemical approaches still suffers from several drawbacks. Especially the regioselective formation of carbon–halogen bonds in mechanistically less favoured positions remains a challenge, leading in many cases to product mixtures. Besides that, chemical halogenation is an environmentally hazardous process using molecular halogen, frequently in combination with Lewis acid catalysts, under harsh reaction conditions. Often, long multistep reactions starting from small halogenated aryls are necessary.11–17  As an indirect halogenation approach, metal-catalysed C2/C7-diboronation–C2-protodeboronation of 3-alkylindoles like tryptophan or tryptamine allow efficient access to valuable C7-boronoindoles that permit further C7 derivatisation.18  Nevertheless, only a single position at the indole ring can be addressed regioselectively using this approach, revealing the limits of current chemocatalytic methods to functionalise indole moieties.

Nature has evolved several enzymatic strategies to form halogenated natural products under much milder, environmentally friendly conditions. Halogenases can be divided into three major enzyme classes:19  fluorinases,20–22  haloperoxidases23  and O2-dependent halogenases. The latter can be divided into two subclasses depending on their reaction type, namely non-heme iron α-ketoglutarate-dependent halogenases24,25  and flavin-dependent halogenases, which are highlighted in this review. Today, more than 5000 naturally occurring organohalogen compounds are known26  including metabolites with diverse biological activities like hormones (e.g. thyroxine27 ), antibiotics like e.g. chloramphenicol28  and chlorotetracycline,29  and cytostatics (e.g. cryptophycin10,30 ). Many halogenated natural compounds originate from marine environments where chloride and bromide salts are highly abundant.26  Unlike chemical halogenations that usually require molecular halogen, biocatalytic halogenations proceed at ambient temperature in aqueous media at neutral pH and require non-toxic halide salts like NaCl or NaBr and O2 as the oxidant. Unsurprisingly, the interest in enzymatic halogenation increases continuously in the context of biocatalytic applications.

Among the halogenating enzymes, the class of flavin-dependent halogenases provides promising methodology for biocatalytic arene halogenation. The halogenated metabolites are formed with high regioselectivity that is in contrast to the relatively unspecific haloperoxidases.31,32  Investigations on the biosynthesis of 7-chlorotetracycline in 1995 led to the identification of a protein responsible for the introduction of the chloro substituent into the antibiotic with no sequence similarity to haloperoxidases.33  Based on these results, it was assumed that the identified gene encodes for a novel class of halogenating enzymes. Finally, five years later in 2000, the first characterisation of a flavin-dependent halogenase, PrnA from Pseudomonas fluorescens, was reported.34  This enzyme is part of the biosynthetic pathway of the halogenated antibiotic pyrrolnitrin, chlorinating l-tryptophan regioselectively at the 7-position of the indole ring. A large set of putative tryptophan halogenases has been discovered by sequence homology and genome analysis since then. Flavin-dependent enzymes require reduced flavin (FADH2) besides O2 and halide salts. Enzymes like PrnC,31  PltA,35,36  AoiQ,37  Rdc2,38  CtcP29  also accept other substrates than tryptophan (see Fig. 1). The bifunctional hybrid enzyme AoiQ, which is involved in the biosynthesis of several diaporthin derivatives, is particularly interesting. It consists of two functional domains, one for O-methylation and one for halogenation.37  VhaA that is involved in vancomycin biosynthesis acts on a phenyl ether in a carrier-bound hexapeptide sequence.39  Hence, the substrate scope is not only limited to indole moieties, but halogenation also takes place on substrates containing pyrrole,31  phenol,40–42  quinones,43  alkenes and alkynes44  as well as alkyl groups.28,37,45  These halogenations, however, take place mainly in the frame of non-ribosomal peptide synthesis on a carrier-bound amino acid.46  Noteworthy, MibH from the actinomycete Microbispora sp. 107891 is a novel type of flavin dependent halogenase as it catalyses the chlorination of Trp while incorporated in a ribosomally synthesised lanthipeptide precursor.47  The major challenge is to provide experimental evidence for most of the halogenases identified on the genome level due to lacking knowledge about the natural substrate.48  For example, VirX1, an iodinating flavin-dependent halogenase was discovered through a bioinformatics based search using data from all fully biochemically and structurally characterized flavin-dependent halogenases. Its substrate flexibility has been evaluated against a 400-member library in vitro. However, its natural substrate remains elusive so far.49 

Figure 1

Reactions catalysed by flavin-dependent halogenases. In all cases, the enzymes require the presence of an additional flavin reductase supplying FADH2.

Figure 1

Reactions catalysed by flavin-dependent halogenases. In all cases, the enzymes require the presence of an additional flavin reductase supplying FADH2.

Close modal

The tryptophan 7-halogenases PrnA and RebH were the first flavin-dependent halogenases to be characterised kinetically34,50  and structurally51,52  making them prototypic flavin-dependent halogenases. Their structural features are described here as a reference. The roughly 530 amino acids (60 kDa) fold into a single domain consisting of two subdomains (see Fig. 2). A box-shaped subdomain contains the FAD binding site and is highly conserved among flavin-dependent halogenases. The smaller subdomain, which is discontinuous in the sequence, is pyramid-shaped in tryptophan halogenases, but poorly conserved among other flavin-dependent halogenases (see Fig. 3). The substrate is bound at the interface of the two subdomains making the smaller subdomain an important determinant of substrate specificity (see Fig. 2A). All tryptophan halogenases of known structure form very similar dimers in the crystal. Structures of a single tryptophan halogenase in different states (e.g. apo, flavin-bound, substrate-bound) did not reveal any larger relative movements of the protomers in the dimer or of the subdomains in the monomer. Functionally important conformational changes take place in two loop regions. The substrate binding loop forms a lid that closes the active site when tryptophan is bound, but is flexible in the absence of substrate. The FAD binding loop is flexible or adopts an open conformation in the cofactor-free form and closes upon binding of FAD.53,54  In the dimer, the FAD binding loops lie at the outside of the dimer, while the substrate binding loops are close to the dimer interface. The dimer contact in tryptophan halogenases involves the poorly conserved smaller subdomain. Flavin-dependent halogenases with different substrate specificity often are monomeric or form different dimer arrangements.40,42,55  VirX1 is the only trimeric flavin-dependent halogenase.49  Structural peculiarities of other flavin-dependent halogenases are a unique zinc-binding motif in MalA56  and the covalent attachment of FAD in CmlS.28 

Figure 2

Structure of the tryptophan halogenase RebH (PDB ID 2oa1A). (A) The overall structure of flavin-dependent halogenases is made up of two subdomains. The ‘box’ subdomain (dark green) is conserved amongst all flavin-dependent halogenases, whilst the smaller subdomain (light green; ‘pyramid’ in Trp-Hals) is variable. The substrate (Trp, shown as sticks) is bound between the ‘box’ and ‘pyramid’ subdomain and coordinated with the involvement of a substrate binding loop (dotted circle). FAD (sticks) is bound by the FAD binding loop (dotted oval), and a chloride ion (green sphere) is bound in proximity to the isoalloxazine ring of the FAD. (B) The halide binding site consists of two backbone nitrogen atoms (light blue circles) that coordinate the halide (T359 and G360 in RebH; the respective distances are marked by pink dotted lines). The halide is 9.6 Å from the C7 of tryptophan which is the halogenation position (light blue dotted line).

Figure 2

Structure of the tryptophan halogenase RebH (PDB ID 2oa1A). (A) The overall structure of flavin-dependent halogenases is made up of two subdomains. The ‘box’ subdomain (dark green) is conserved amongst all flavin-dependent halogenases, whilst the smaller subdomain (light green; ‘pyramid’ in Trp-Hals) is variable. The substrate (Trp, shown as sticks) is bound between the ‘box’ and ‘pyramid’ subdomain and coordinated with the involvement of a substrate binding loop (dotted circle). FAD (sticks) is bound by the FAD binding loop (dotted oval), and a chloride ion (green sphere) is bound in proximity to the isoalloxazine ring of the FAD. (B) The halide binding site consists of two backbone nitrogen atoms (light blue circles) that coordinate the halide (T359 and G360 in RebH; the respective distances are marked by pink dotted lines). The halide is 9.6 Å from the C7 of tryptophan which is the halogenation position (light blue dotted line).

Close modal
Figure 3

Structures of selected flavin-dependent halogenases. The conserved FAD-binding subdomain is shown in gray, the variable smaller subdomain in blue. The orientation is the same as in Fig. 2. The top row shows tryptophan/indole halogenases sharing a pyramid-shaped small subdomain. The tryptophan 6-halogenases Thal (PBD ID 6h44) and Th-Hal (PDB ID 5lv9) represent the C7 clade (including PrnA, RebH and BorH) and the C5 clade (including PyrH, SttH, Tar14 and SatH), respectively. BrvH (PBD ID 6frl) has short substrate binding loops in the pyramid and represents halogenases with a wide substrate binding site including MibH and VirX. The middle row shows flavin-dependent halogenases that accept free substrates other than tryptophan or indole. Their C-terminal small subdomain is structurally distinct from the pyramid of tryptophan halogenases. MalA’ (PDB ID 5wgr) and PltM (PDB ID 6bza) share some structural features in their helical C-terminal region. CmlS (PDB ID 3i3l) is sometimes grouped as variant B but its arch-like small subdomain distinguishes it from other variant B enzymes suggesting that it may accept a free substrate. The bottom row shows variant B flavin-dependent halogenases that act on carrier protein-bound substrates. The C-terminal region of CndH (PDB ID 3e1t) is mostly disordered in the crystal so that the small subdomain is almost missing, leaving the active site lysine freely accessible. PDB entry 3nix is a functionally uncharacterized protein from Cytophaga hutchinsonii with high structural similarity to CndH. PltA (PDB ID 5dbj) is a verified variant B halogenase, in which a helical C-terminal small subdomain blocks access to the active site. PltA has high structural similarity to the variant B halogenases Bmp2 and Mpy16.

Figure 3

Structures of selected flavin-dependent halogenases. The conserved FAD-binding subdomain is shown in gray, the variable smaller subdomain in blue. The orientation is the same as in Fig. 2. The top row shows tryptophan/indole halogenases sharing a pyramid-shaped small subdomain. The tryptophan 6-halogenases Thal (PBD ID 6h44) and Th-Hal (PDB ID 5lv9) represent the C7 clade (including PrnA, RebH and BorH) and the C5 clade (including PyrH, SttH, Tar14 and SatH), respectively. BrvH (PBD ID 6frl) has short substrate binding loops in the pyramid and represents halogenases with a wide substrate binding site including MibH and VirX. The middle row shows flavin-dependent halogenases that accept free substrates other than tryptophan or indole. Their C-terminal small subdomain is structurally distinct from the pyramid of tryptophan halogenases. MalA’ (PDB ID 5wgr) and PltM (PDB ID 6bza) share some structural features in their helical C-terminal region. CmlS (PDB ID 3i3l) is sometimes grouped as variant B but its arch-like small subdomain distinguishes it from other variant B enzymes suggesting that it may accept a free substrate. The bottom row shows variant B flavin-dependent halogenases that act on carrier protein-bound substrates. The C-terminal region of CndH (PDB ID 3e1t) is mostly disordered in the crystal so that the small subdomain is almost missing, leaving the active site lysine freely accessible. PDB entry 3nix is a functionally uncharacterized protein from Cytophaga hutchinsonii with high structural similarity to CndH. PltA (PDB ID 5dbj) is a verified variant B halogenase, in which a helical C-terminal small subdomain blocks access to the active site. PltA has high structural similarity to the variant B halogenases Bmp2 and Mpy16.

Close modal

Halogenation in flavin-dependent halogenases follows an electrophilic aromatic substitution mechanism. Whereas most of the mechanism is undisputed, some details are still controversial.

The enzyme depends on its reduced cofactor FADH2 for the halogenation to take place.34  FADH2 is bound at a canonical binding site and reacts with molecular oxygen, leading to the formation of a FAD hydroperoxide species (see Fig. 4).57  A conserved halide binding site is found near the isoalloxazine moiety of the FAD (see Fig. 2B). A halide ion bound in that position attacks the hydroperoxide, generating a hypohalous acid (HOX). This hypohalous acid needs to travel to the substrate binding site which is ∼10 Å from the halide binding site. The substrate binding site contains a lysine residue which is essential for the enzyme's activity. It either coordinates the hypohalous acid51  or reacts with it, resulting in a haloamine which then performs the electrophilic aromatic substitution (see Fig. 4).52  The relative position of this lysine towards the substrate determines the regioselectivity.51,53,58 

Figure 4

Proposed mechanism of the enzymatic halogenation of tryptophan by RebH, based on investigations by Dong et al.,51  Flecks et al.69  and Yeh et al.52  FADH2 is generated by a flavin reductase. In the FAD binding site, the oxidation of FADH2 by molecular oxygen leads to the formation of a flavin peroxide intermediate that is subsequently attacked by a halide ion (X), generating a hypohalous acid (HOX) which then migrates to the tryptophan binding site. It is unclear whether the halogenating species is a lysine N-haloamine (yellow) or HOX (green). An electrophilic attack leads to the formation of a Wheland intermediate that is subsequently deprotonated, generating the halotryptophan.

Figure 4

Proposed mechanism of the enzymatic halogenation of tryptophan by RebH, based on investigations by Dong et al.,51  Flecks et al.69  and Yeh et al.52  FADH2 is generated by a flavin reductase. In the FAD binding site, the oxidation of FADH2 by molecular oxygen leads to the formation of a flavin peroxide intermediate that is subsequently attacked by a halide ion (X), generating a hypohalous acid (HOX) which then migrates to the tryptophan binding site. It is unclear whether the halogenating species is a lysine N-haloamine (yellow) or HOX (green). An electrophilic attack leads to the formation of a Wheland intermediate that is subsequently deprotonated, generating the halotryptophan.

Close modal

When the first flavin-dependent halogenases were described, their similarity to flavin-dependent monooxygenases led to the expectation that the reaction mechanism would be largely similar. Both enzymes contain the GxGxxG motif that is involved in the cofactor binding.59  Additionally, both were observed to depend on FADH2.34  In flavin-dependent monooxygenases, the reaction mechanism had been inferred from kinetic and structural studies.60  FADH2 reacts with molecular oxygen to form a flavin(C4a)-hydroperoxide which directly oxidises the substrate that is bound in proximity. The same peroxo intermediate was proven in flavin-dependent halogenases57  and suggested to form an epoxide34  or hydroxy61  moiety at the indole ring, followed by a nucleophilic attack of the chloride anion to release 7-chlorotryptophan after subsequent dehydration and re-aromatisation. An alternative hypothesis was the direct attack of a chloride ion at the flavin hydroperoxide to form a flavin hypochlorite intermediate that would attack the indole ring in an electrophilic aromatic substitution reaction.50  The utilisation of freely diffusing hypohalous acids as the halogenating agent had been explicitly ruled out because this process would lack selectivity.50  The structure elucidation of the flavin-dependent C7 tryptophan halogenases PrnA51  and RebH52,62  shed new light on enzymatic halogenation. The actual substrate binding site in PrnA and RebH as well as all other flavin-dependent halogenases is ∼10 Å distant from the FAD binding site (see Fig. 2B). Before, they had been assumed to be adjacent like in flavin-dependent monooxygenases.50,63  Interestingly, flavin-dependent halogenases contain a second consensus sequence motif (WxWxIP) which distinguishes them from the monooxygenases. The motif has been postulated to block substrate access to the bound FADH2, preventing monooxygenase activity.51  In addition, the PrnA structure contains a chloride ion bound to the re face of the flavin. This binding site is also present in other halogenases (e.g. PyrH,53  PltA55  and SttH64 ). One possible explanation for substrate halogenation was that the enzyme undergoes a conformational rearrangement that allows for a closer proximity between flavin and substrate binding site. The structures do not suggest such a large motion, and molecular dynamics simulations also do not show any indication of such a transition.65  Thus, reaction mechanisms based on epoxide formation by direct flavin peroxide oxidation or an electrophilic attack of flavin-hypochlorite were ruled out. In the case of flavin-dependent halogenases, halide binding and the subsequent reaction with the flavin hydroperoxide thus probably lead to the generation of a hypochlorous acid in a nucleophilic attack of the chloride ion on the flavin hydroperoxide. The HOCl is prevented from diffusing into the solvent by the closed active site,51  and is assumed to migrate through the enzyme to the substrate binding site.

While the generation of the HOX and its diffusion to the active site is basically undisputed, part of the halogenation mechanism remains elusive. Formally, a halenium ion (X+) is transferred from the HOX to the substrate. The transfer results in a σ complex (Wheland intermediate) which re-aromatises by the release of a proton. Two hypotheses have been formulated for the reaction leading to the Wheland intermediate, both of which could not be refuted up to now. Dong et al. suggested that the HOX is positioned by a lysine side chain at the entrance to the active site (K79 in PrnA).51  The positioning via hydrogen bonding might activate HOX by increasing its electrophilicity. The importance of the lysine residue was confirmed by mutation to alanine that caused a complete loss of activity in various halogenases.42,51,52,66–68  Additionally, a comparison between tryptophan halogenases that halogenate at C7,51  C5,53  and C658  position, respectively, of the indole moiety proved the regioselectivity to be based on proximity of the respective carbon atom to the lysine-εNH2. Yeh et al. postulated a different role of the lysine.52  They incubated RebH with FADH2, O2 and Na36Cl in the absence of substrate and subsequently removed the FAD and halide salts using a desalting column. 63 h later, upon addition of the substrate to the protein, they still observed formation of 36Cl-labeled chlorotryptophan and explained this by the formation of a lysine-εNH-chloramine. HOCl is known to rapidly react with the εNH2 of lysine; therefore, a chloramine was postulated to be the stable halogenating agent.52  However, formation of the lysine-εNH-chloramine could not be proven and quantum mechanical calculations led to the conclusion that the chloramine was a weaker halogenating agent than HOCl.69  Thus, the lysine chloramine might only act as a reservoir for the generation of HOCl (see Fig. 4). QM/MM calculations have been performed for halogenation reactions involving HOCl in PrnA70  and the lysine chloramine in MalA’.56  In both cases, the authors concluded that their calculations could explain the observed reaction outcome, but without a focus on which hypothesis was more likely.

In addition to the proximity to the catalytic lysine that is caused by the substrate orientation, the activation of a C(sp2)-H bond is important to explain which sites are halogenated. In nonenzymatic halogenation, regioselectivity is governed by the stability of the Wheland complex. The HalA scale (halenium affinity)71  quantifies this activation and is able to predict whether it is sufficient for halogenation in electrophilic aromatic substitution.16,71  In general, electronic activation is based on the ability of a potential halenium ion (X+) acceptor (substrate) to stabilise the free ion and is thus reduced when electron withdrawing groups are present in the substrate.

In tryptophan halogenases, the active site contains a conserved glutamate residue in addition to the lysine. Mutation of E346 in PrnA to glutamine decreased the catalytic activity by two orders of magnitude,51  and replacement by aspartate generated an inactive enzyme.69  The function of the glutamate is controversial. It was suggested to deprotonate the Wheland intermediate that is formed after the electrophilic attack,51  but, no kinetic isotope effect was observed in l-tryptophan halogenation,72  leading to the conclusion that deprotonation was not the rate-limiting step. Thus, halogenation probably does not rely on a general base. Another possible role of the glutamate is the coordination and possibly activation of the HOX.69 

The catalytic glutamate is absent in both phenol and pyrrole halogenases. While pyrrole is more electronically activated than tryptophan, phenol is not.73  This is in accordance with HalA calculations.16  Thus, an additional activation is probably unnecessary in pyrrole halogenases, but may be required in phenol halogenases. For PHBH, a flavin-dependent monooxygenase acting on a phenolic substrate, activation was shown to include a deprotonation step, generating a phenolate anion.60  Indeed, many phenol halogenases require substrates with a hydroxyl group ortho to the halogenation site. Rdc2 is unable to halogenate most indole compounds, although these are more activated than the phenol compounds. This might also be explained by the lack of the deprotonation step16  as indole compounds are less activated than phenolates. Supporting the hypothesis that the deprotonation step is necessary for the electrophilic aromatic substitution to take place, the D325A mutation led to abolished activity in RadH.74  According to a homology model, this residue is close to the substrate. However, the difference in the substrate binding sites should also be considered as a possible basis of the lack of halogenation activity.

As two-component flavin-dependent monooxygenases, flavin-dependent halogenases rely on a separate NAD(P)H-dependent flavin reductase to regenerate FADH2.34  A flavin reductase and a flavin-dependent halogenase may be encoded in the same biosynthetic gene cluster.75,76  Nevertheless, flavin-dependent halogenase and flavin reductase do not necessarily form a cognate pair, as FADH2 regeneration in vitro can be achieved by a flavin reductase from another species.34  FADH2 can also be regenerated by an organometallic complex,61  by NADH mimics77  or photochemically.78  In some two-component flavin-dependent monooxygenases, flavin transfer between reductase and monooxygenase can be achieved by free diffusion.79  This is also the case for PrnA,61  but may be different for other flavin-dependent halogenases. Photoreduction in PyrH was demonstrated to act on bound instead of free flavin.78  In CmlS, the isoalloxazine ring is covalently bound to the protein implying that FAD might be reduced within the flavin-dependent halogenase. However, a structural comparison with CndH revealed a potential conformational change that would allow the cofactor to exit the FAD binding cleft while remaining bound to CmlS.28  Structural studies of several halogenases suggest a negative coupling between flavin and substrate binding.42,52–54,62  Binding of substrate caused the isoalloxazine ring of flavin to either adopt an unusual binding pose or to completely move out of the binding cleft, potentially facilitating the regeneration of FADH2. With respect to flavin regeneration, Bmp5 is an exceptional flavin-dependent halogenase. Its sequence lacks similarity to canonical flavin-dependent halogenases but shares homology with single-component flavin-dependent monooxygenases. Like them, purified Bmp5 does not require a separate flavin reductase to regenerate FADH2in vitro but directly utilises NADPH.80 

While both chlorination and bromination have been regularly observed for flavin-dependent halogenases,41,52,56,59,67,68,81–87  iodination has been reported only for a few flavin-dependent halogenases42,49  and fluorination cannot be catalysed by these enzymes. The oxidation of fluoride ions cannot be achieved with hydrogen peroxide or oxygen, and halogenation in S-adenosyl-l-methionine (SAM) fluorinases proceeds via a nucleophilic pathway.88  As fluoride is highly electronegative, its hydration in aqueous solutions additionally requires the enzymes involved in fluorination to remove water molecules from the ion.89  Thus, flavin-dependent halogenases catalysing fluorination are not expected to be found.

Assessments of the halide specificity in different halogenases initially suggested a preference of chlorination over bromination. PyrH showed only 75% brominating compared to chlorinating activity when yield of 5-halotryptophan after 90 minutes was evaluated.59  The kcat of RebH is 1.4 min−1 for chlorination and 0.4 min−1 for bromination of tryptophan,50  and the specific bromination activity of SgcC3, a flavin-dependent halogenase that acts on a carrier protein-tethered substrate, is two-fold less than chlorination activity under similar reaction conditions.41  Similar results were observed for XszenFHal, where bromination led to 64% conversion of tryptophan within 24 h, while chlorination proceeded fully within that timeframe.86  In the activity assay for MalA, a halogenase that performs the dichlorination of premalbrancheamide, chlorination reactions were incubated for 20 minutes, whereas bromination reactions were performed overnight.56  In contrast to the dichlorination, only monobromination with a C9 regioselectivity (8 : 1 ratio of C9 to C8 halogenation) was observed, and the bromination yields were lower.

Early attempts to explain halide specificity were based mostly on the ionic radius of the halide, with a note that other factors such as electronegativity, basicity, nucleophilicity and the standard redox potential of the HOX/X couple needed to be taken into account, and more experimental data was needed to explain the specificity.90 

Halide preference was even more striking when competitive assays between chloride and bromide were performed. When pyrrolyl-S-PltL was converted to a halogenated species, >90% of the substrate was converted within 1 hour when only chloride was present compared to 10–20% in the presence of both chloride and bromide.36  Thus, the presence of bromide seemed to inhibit chlorination. A similar behaviour was observed for KrmI, a halogenase from the metagenome of Theonella swinhoei WA, a marine sponge.83  Competitive 5-hydroxytryptophan halogenation assays were performed with a total halide concentration of 50 mm, testing varying ratios of NaCl and NaBr. Unlike in most tryptophan halogenases, a preference for bromination over chlorination was observed; however, an increase in bromide concentration also resulted in a decrease in substrate conversion.

Other halogenases were characterised that also show a preference for bromination. The first ones were Bmp2 and Bmp5, a pyrrole and phenol halogenase from marine bacteria.80,91  The enzymes belong to a gene cluster coding for the pathways for formation of tetrabromopyrrole, a coral larval settlement cue, and pentabromopseudilin, a microbial antibiotic, respectively. Bmp5 catalyses the bromination and decarboxylation of 4-hydroxybenzoic acid. While it does not chlorinate, in vivo iodinating activity was observed. Bmp2 catalyses the tri- or tetrabromination of pyrrolyl-S-Bmp1 and does not chlorinate or iodinate. Mpy16, a structurally homologous pyrrole halogenase, dichlorinates its substrate. At equimolar or higher bromide concentrations, it also brominates its substrate, generating mixed monobrominated-monochlorinated products and dibrominated products.84  A comparison of the halide and substrate binding region between Bmp2 and pyrrolyl-S-ACP halogenases (including Mpy16) that catalyse one or two halogenations revealed three sites where otherwise conserved residues were exchanged in Bmp2. Thus, a Y302S/F306V/A345W triple mutant of Bmp2 was generated based on the corresponding residues in Mpy16 and observed to catalyse only monobromination.91  A structural analysis revealed no different positioning of the bound FAD cofactor or the catalytic lysine. Interestingly, the Tyr302 is part of the putative halide binding site, but despite the mutation to Ser, the protein retained its specificity for bromide.

The authors of these studies concluded that the halide specificity could not be merely explained by steric factors, but that other factors such as the redox potential of the flavin isoalloxazine ring in the surrounding needed to be considered.91  Another important effect is the enthalpic penalty associated with the desolvation of the halide ion, the reason why there are no flavin-dependent fluorinases, and why some halogenases might be unable to utilise chloride. They stressed that this “molecular filter”84  was necessary as chloride concentration in seawater is ∼600-fold higher as compared to bromide. Thus, brominated natural products needed to result from enzymes that were highly selective. Chlorinases, on the other hand, do not need to exert such a selectivity filter as chloride is much more abundant.

Four other halogenases (BrvH from Brevundimonas BAL3 and three enzymes from Xanthomonas campestris pv. campestris B100 (Xcc)) were characterised to preferentially (BrvH) or exclusively (Xcc halogenases) brominate. BrvH was identified by bioinformatic analysis from a marine metagenome sample.92  Characterisation revealed bromination activity towards different indoles. Tryptophan was not converted, and chlorination activity was negligible. The same behaviour was observed for the Xcc halogenases that were identified from the Xanthomonas genome based on the flavin-dependent halogenase sequence motifs GxGxxG and WxWxIP.93  The four halogenases show minimal differences in their respective substrate spectra.

What distinguishes these proteins from Bmp2 and Bmp5 is that the pathway they belong to is unknown. Thus, their native substrate cannot be inferred. The relevance of this detail was brought up in a family-wide activity profiling study.94  87 genome-mined halogenases were screened for activity towards 62 substrates and both chlorination and bromination yields after 18 h were evaluated. The authors noted that bromination was much more widespread than chlorination. Intriguingly, every halogenase that was active on at least one substrate had brominase activity, whereas only 16% of the enzymes chlorinated. The halogenation activities were clustered and one cluster (AC3) contained low-activity enzymes that only brominated the substrates that were electronically most activated. Additionally, the authors observed that RebH, which originally catalyses tryptophan chlorination, brominated a greater range of non-native substrates than it chlorinated and, based on the yield, preferred bromination over chlorination of l-tryptophan. Lewis and co-workers thus postulated that their findings, as well as the observations on BrvH and the Xcc halogenases, indicate that a preference for bromination is common in flavin-dependent halogenases and does not imply that the native substrate is being brominated.

Unlike chlorination and bromination, iodination activity is very rare in flavin-dependent halogenases. It had been shown early on in vivo for CrpH in plants and Bmp5 heterologously expressed in E. coli.80,95  However, in a competitive assay, iodide was shown to inhibit chlorination in RebH.52  While in presence of 15 mM iodide, the kobs of the chlorination was decreased by a factor of more than two, no activity was observed in presence of 50 mM iodide. This could be explained if binding of iodide is possible, but the formation of hypohalous acid is impeded, or the transit to the active site cannot be achieved. While a lack of iodination in most flavin-dependent halogenases was observed, competitive assays were not performed.38,41,82,83,86,93 

In vitro iodination was only published last year for the viral halogenase VirX1 and the bacterial enzyme PltM.42,49  The native substrate of VirX1 is unknown. The enzyme was thus screened against a 400-member substrate library, and kinetics were obtained for 14 selected substrates. kcat was higher for iodination compared to bromination for all but one substrate. In competition assays, the halide preference obtained was I>Br>Cl, corresponding to the decreasing oxidation potential of the halide.49  The bacterial halogenase PltM, which catalyses mono- and dichlorination of phloroglucinol, a transcriptional regulator in pyoluteorin biosynthesis,96  showed a very similar halide preference. In competitive assays with a 1 : 1 ratio of two halides and the natural substrate, iodination was preferred over bromination, and bromination was preferred over chlorination based on isolated yield.42  When only one halide was present, both for chlorination and iodination, mono- and dihalogenated products could be observed, but no dibrominated product was formed. Using a 10-fold excess of chloride over iodide, mixed products could be observed, which was not the case when using bromide and iodide. Under optimised reaction conditions, kinetics were analysed for three substrates. Unfortunately, kinetics could not be analysed quantitatively for iodination due to gradual enzyme precipitation in the presence of iodide. The halide preference seemed to depend upon the substrate.

Whilst the absence of fluorination in flavin-dependent halogenases can be readily explained by its electronegativity and the resulting enthalpic penalty of desolvation and the rare iodination may be partially due to the halide size, exclusive bromination is harder to explain. Hence, in order to completely understand halide selectivity in flavin-dependent halogenases, more experimental insight is needed.

Currently, flavin-dependent halogenases are divided into three main groups: Indole (mostly tryptophan) halogenases (e.g. PrnA,34  RebH,50  PyrH59 ), phenol halogenases (e.g. ChlA,66  Rdc2,97  PltM42 ) and pyrrole halogenases (e.g. PrnC,98  PltA,36  Bmp280 ).

Another classification distinguishes variant A halogenases that act on a free substrate and variant B halogenases that require a carrier-bound substrate and are often part of non-ribosomal peptide synthetases (NRPS).40  Most known variant B flavin-dependent halogenases are phenol or pyrrole halogenases (e.g. CndH,40  PltA,36  Bmp2,80  and HalY99 ). CmdE that halogenates tryptophan on C2 during chondramide biosynthesis is most likely a variant B tryptophan halogenase.40,100  In addition, a few aliphatic flavin-dependent halogenases have been described. Whereas CmlS most probably acts on an activated aliphatic carbon,28  the fungal flavoenzyme AoiQ is a bifunctional halogenase/methylase acting on an unactivated aliphatic carbon.37  This goes against the assumption that all flavin-dependent halogenases share the need for an electronically activated substrate.101  Classification of flavin-dependent halogenases is not always straightforward, e.g. if the native substrate is unknown. Moreover, distinction between variants A and B may not be possible solely based on sequence similarity28  and a flavin-dependent halogenase encoded in an NRPS gene cluster may still act on a free substrate.56  Several recent reviews comprehensively cover the substrate specificity of currently known flavin-dependent halogenases.102,103  Here, we focus on flavin-dependent halogenases for which structures are available.

Most structural information on substrate binding and regioselectivity derives from tryptophan halogenases. In the absence of substrate, the active site is solvent-accessible, whereas bound tryptophan is completely enclosed by the protein.51–53,58,64,67,68,85  Tryptophan is precisely positioned mostly by π stacking of its indole moiety and polar contacts of its amino and carboxy group with protein side chains. The seminal structure of PrnA revealed that the εNH2 group of the catalytic lysine is located close to the chlorine of bound l-7-chlorotryptophan leading Naismith and van Pée to suggest that this positioning determines regioselectivity.51  The structure of the tryptophan 5-halogenase PyrH with C5 positioned closest to the catalytic lysine corroborated this hypothesis.53  The π stacking region of PyrH is similar to that of PrnA and RebH, but the residues of the substrate binding loop that form polar contacts to the tryptophan backbone are unrelated. The indole moiety is flipped by roughly 180° compared to tryptophan 7-halogenases. Recent structures of the tryptophan 6-halogenases Thal and BorH further confirmed the importance of positioning the halogenated carbon relative to the εNH2 group of the catalytic lysine.58,68  As in PyrH, the indole moiety is flipped in Thal and BorH relative to PrnA and RebH. Tryptophan 6-halogenases fall into two clades (see Fig. 3). Sequence identity of Thal and BorH is highest with tryptophan 7-halogenases, while all other currently known tryptophan 6-halogenases cluster with the tryptophan 5-halogenases PyrH, SpmH104  and XszenFHal.86  Structures of SttH,64  Th-Hal67  and Tar14,85  three tryptophan 6-halogenases of the C5 clade, all lack substrate and the tryptophan binding pose could not be modelled unequivocally. Sequence comparison of SttH and PyrH revealed three residues in the substrate binding loop that influence regioselectivity through polar contacts with the amino and carboxy groups of tryptophan.64  Mutagenesis of SatH, the most recent tryptophan 6-halogenase of the C5 clade, highlighted two further residues important for regioselectivity.87  In contrast to residues of the substrate binding loop, these two residues are conserved between both clades of tryptophan 6-halogenases and contact the indole moiety.

MibH is an unusual tryptophan halogenase involved in the biosynthesis of a lantibiotic, a ribosomally synthesised and posttranslationally modified antimicrobial peptide.47  Whereas other tryptophan-halogenases halogenate free tryptophan, MibH chlorinates l-tryptophan in the 5 position only when it is part of the cognate peptide. The MibH active site is a solvent-exposed crevice and its surface accessible area is larger and more hydrophobic than in tryptophan halogenases acting on free tryptophan, supporting its preference for a peptide substrate.47  Ortega et al. performed docking of the substrate to MibH resulting in poses where the C5 of indole is closest to the catalytic lysine. The distance of 8.5 Å to the εNH2 is substantially longer than in other tryptophan halogenases and the authors envision this distance to be reduced upon substrate binding.

Recently, two flavin-dependent halogenases were identified based on their clustering with tryptophan halogenases in multiple sequence alignments. Their native substrate and the biosynthetic pathway in which they participate are unknown. Both BrvH and VirX1 do not halogenate free tryptophan but halogenate indole in vitro.49,92  BrvH was identified from a marine metagenome and converts indole to 3-bromoindole. Its structure revealed a substrate binding site that is open to the solvent instead of being covered by a loop (see Fig. 3).92  VirX1 derives from a virus infecting oceanic cyanobacteria and regioselectively halogenates diverse substrates. It preferably forms aryl iodide species.49  From a library containing 400 compounds, the conversion of indole was highest resulting in 95% 3-iodoindole. The VirX1 structure revealed an unusual trimer but the protomers share high structural similarity with other tryptophan halogenases.49  As in MibH and BrvH, the loops corresponding to the substrate binding loops of classic tryptophan halogenases from the C7 or the C5 clade are very short and, therefore, cannot act as a lid to close off the substrate binding site. The substrate binding site of VirX1 is even wider than that of BrvH, potentially explaining the broad substrate scope.49  Genomic analysis of Xanthomonas campestris pathovar campestris B100 (Xcc) revealed three flavin-dependent halogenases related to tryptophan halogenases. In vitro, they also halogenate various indole derivatives with highest yields for indole.93  The Xcc flavin-dependent halogenases share enzymatic and structural features with BrvH and VirX1. They halogenate C3 of indole and the structure of Xcc4156 features an open substrate binding site with short loops that cannot form a lid upon substrate binding.93,105  As a most likely non-native substrate, indole presumably forms fewer specific contacts with these flavin-dependent halogenases. Due to the wide and open substrate binding pockets, indole or its derivatives might not be positioned as uniquely and precisely by BrvH, VirX1 and Xcc flavin-dependent halogenases as tryptophan is in classic tryptophan halogenases. This assumption is supported by the low specific activity of BrvH towards indole92  and by the fact that many attempts to crystallise VirX1 with various substrates failed.49  Moreover, C3 is the most reactive position of indole for electrophilic aromatic substitution.106  An analysis of fourteen reaction products of VirX1 showed that halogenation was often mediated at the chemically most reactive position.49 

Variant A halogenases that accept free substrates hold particularly high promise of being applied as biocatalysts in synthetic chemistry. As various tryptophan halogenases with different regioselectivities have been published, enzymes accepting free substrates other than tryptophan or indole derivatives are particularly interesting. Although several such enzymes have been identified, structures are known for only two of them, namely MalA’56  and PltM.42  CmlS was also suggested to act on a free substrate,28  but will be discussed with variant B flavin-dependent halogenases below. Both MalA’ and PltM can dihalogenate their substrates, whereas formation of 6,7-dichloro-tryptophan during the synthesis of kutzneride is mediated by the successive action of the tryptophan 7-halogenase KtzQ and the tryptophan 6-halogenase KtzR that halogenates 7-chlorotryptophan much more efficiently than tryptophan.107 

For the fungal halogenase RadH, structural information was derived from homology modelling.74  RadH and Rdc216  are two closely related flavin-dependent halogenases involved in the production of radicicol in different fungal species. RadH regioselectively halogenates a range of bioactive aromatic scaffolds. A homology model of RadH based on the structure of the flavoprotein geranylgeranyl reductase showed three domains, namely flavin-binding, catalytic, and C-terminal.74  A large cavity in the catalytic domain was identified as the likely substrate binding site. Mutation of a lysine close to this site abolishes enzymatic activity. This lysine is around 4 Å away from the halogenated C6 of radicicol docked into the putative active site.74  The distance is consistent with the current model of regioselective halogenation lending credibility to both the homology model and the substrate docking. The model also showed aromatic residues potentially involved in π stacking with the aryl moiety of the substrate and polar residues within H-bonding distance of the substrate. Mutation of these residues to Ala significantly reduced activity.74 

The flavin-dependent halogenases MalA and the almost identical MalA’ from two different fungal strains perform the final steps in the biosynthesis of the dichlorinated indole alkaloid malbrancheamide.56  Although MalA is encoded in a gene cluster containing an NRPS, it catalyses the iterative dichlorination and monobromination of premalbrancheamide as a free substrate. The Michaelis–Menten kinetics led to the conclusion that MalA is equally selective for both sites of the indole ring.56  Structures of MalA’ were determined in complex with three different substrates, namely the non-halogenated premalbrancheamide, malbrancheamide B monochlorinated on C9 (corresponding to C6 of indole) and isomalbrancheamide B monochlorinated on C8 (corresponding to C5 of indole). The overall structure is similar to bacterial flavin-dependent halogenases but features a unique C-terminal Zn2+-binding motif (see Fig. 3).56  The substrate binding site is a large cavity that can accommodate the pentacyclic substrates. The three natural substrates have an almost identical binding pose suggesting that the substrate does not need to re-orient for the second chlorination.56  In the crystal structure, the distance between either of the two halogenated carbon atoms and the εNH2 group of the catalytic lysine is similar but substantially longer than the distance between catalytic lysine and halogenated carbon atom in tryptophan halogenases. According to molecular dynamics (MD) simulations of MalA’, the Cl atom of a modelled lysine chloramine is positioned more closely to the substrate. Again, the distances to C8 and C9 of the substrate are similar, leading the authors to conclude that Cl-Lys108 is preorganised to chlorinate both positions.56  A flexible region adjacent to the Zn2+-binding motif adopted two main conformations during the MD simulations. These conformations might represent the open and closed state of a substrate channel lid.56 

The bacterial flavin-dependent halogenase PltM catalyses mono- and dichlorination of phloroglucinol.96  PltM has a wide substrate profile catalysing the halogenation of various phenolic compounds and aniline derivatives.42  It is also active on larger molecules including FDA-approved drugs and dietary natural products. The PltM structure comprises the box-shaped FAD-binding region conserved among flavin-dependent halogenases and a unique helical C-terminal region with the substrate phloroglucinol bound at the interface between both regions (see Fig. 3).42  The substrate binding site is large enough to accommodate dihalogenated, but not trihalogenated phloroglucinol.42  Three out of four crystallographically independent chains contain bound substrate, but all in slightly different positions suggesting that positioning of phloroglucinol in PltM is less specific than that of tryptophan in tryptophan halogenases. The shortest distance of a halogenated carbon to the εNH2 of the catalytic lysine is comparable to that in tryptophan halogenases again confirming the importance of this distance. As phloroglucinol is symmetric, regioselectivity is not relevant in this case.

CndH halogenates a phenol group in a most likely carrier protein-bound intermediate of chondrochloren biosynthesis. Its structure revealed that CndH basically lacks the small subdomain (see Fig. 3).40  In the absence of substrate, the C-terminal region is partially unstructured in the crystal providing free access to the catalytic lysine. A large hydrophobic surface surrounding the active site in the flavin-binding subdomain forms a potential binding site for the putative carrier protein.40  A sequence alignment of the then known flavin-dependent halogenases together with functional data led Schulz, Müller and co-workers to suggest that flavin-dependent halogenases should be subdivided into variant A using small substrates like free amino acids and variant B accepting carrier-bound substrates.40 

The chloramphenicol halogenase CmlS is essential for the formation of the dichloroacetyl group, but its exact substrate remains unknown.28  CmlS has higher sequence identity to CndH (variant B) than to variant A tryptophan halogenases and chloramphenicol biosynthesis involves carrier protein-bound intermediates. However, CmlS has a well-structured C-terminal region blocking access to the active site (see Fig. 3).28  Without a major conformational change, the substrate would need to enter a tunnel leading to the catalytic lysine. Based on its structure, Jia and co-workers suggested CmlS to belong to variant A. In accordance, CmlS might prefer a free small molecule, e.g. a simple acyl group or the corresponding CoA thioester.28 

PltA dichlorinates a pyrrole moiety tethered to a NRPS thiolation domain during biosynthesis of the antifungal compound pyoluteorin.36  PltA forms dimers, but in contrast to tryptophan halogenase dimers, the FAD binding loop is part of the dimer interface and the substrate binding sites point to the outside. Each PltA protomer consists of the conserved FAD binding region and a unique C-terminal helical region that blocks access to the substrate binding site (see Fig. 3).55  Tsodikov and co-workers suggested that interactions with the substrate possibly could trigger a conformational change displacing the C-terminal region to open access to the halogenation site.55  The PltA structure clearly shows that variant B flavin-dependent halogenases do not necessarily contain a mostly disordered C-terminal region or a freely accessible catalytic lysine.

The flavin-dependent halogenase Bmp2 catalyses all four halogenations during the biosynthesis of tetrabromopyrrole.91  As other flavin-dependent pyrrole halogenases, Bmp2 operates on a pyrrole bound to a carrier protein. However, tetrabromination is unprecedented as all previously described pyrrole halogenases catalyse one, two, or three halogenations. The fourth bromination requires the joint action of two enzymes, the thioesterase domain of Bmp1 and Bmp2, resulting in offloading of the pyrrole and a decarboxylation concomitant with bromination.91  Structures of Bmp2 and the related pyrrole dihalogenase Mpy16 revealed key residues involved in controlling the degree of halogenation. Three amino acids lining the putative pyrrole binding site that are strictly conserved among all variant B pyrrolyl halogenases catalysing one or two halogenations of the pyrrole ring lack conservation in Bmp2. Mutating these residues to their conserved counterparts resulted in a Bmp2 variant that could catalyse only a single bromination of the pyrrole ring.91  These results demonstrate that side chains of residues lining the substrate binding site specify which pyrrole positions are accessible to halogenation, as has previously been demonstrated for tryptophan halogenases. Thus, the active sites of otherwise highly homologous flavin-dependent halogenases are uniquely evolved to afford distinct product profiles.91 

From a chemist's view, the regioselective introduction of halogen substituents using only halide salts and molecular oxygen at ambient temperatures is a highly attractive reaction, which makes flavin-dependent halogenases a promising tool for biocatalytic applications. All in all, there are several different approaches to use their biocatalytic abilities for synthetic purposes. Isolated halogenases may be used for biotransformations of natural or unnatural substrates in combination with cofactor regeneration systems. Metabolically engineered C. glutamicum strains with incorporated halogenase genes give access to halogenated l-tryptophan via fermentation.108  Narrow substrate scopes can be expanded using directed evolution techniques. In addition, novel halogenated products can be produced in combinatorial biosynthetic cascades. In this approach, genes encoding halogenases are introduced into organisms to provide halogenated intermediates.

Initial attempts to use flavin-dependent halogenases for biocatalytic halogenations suffered from low in vitro activities of all investigated enzymes. The halogenases characterised so far have relatively low kcat values, e.g. 1.0–1.4 min−1 for the tryptophan 7-halogenase RebH,109,110  0.1–6.8 min−1 for the tryptophan 7-halogenase PrnA,51,111  2.8 min−1 for the tryptophan 6-halogenase Thal81  and 0.5–3.6 min−1 for the tryptophan 5-halogenase PyrH.53,59  The total turnover numbers (TTN) are generally below 200.110,112  In 2001, the substrate scope of PrnA was investigated with tryptophan and several indole derivatives.113  Surprisingly, all accepted substrates are being halogenated at the electronically most preferred position C2 or C3, except for tryptophan where halogenation occurs solely at the C7 position. Accepted substrates range from N-methyltryptamine, 4-methyltryptamine, 5- and 3-methylindole to even phenylpyrrole derivatives. N-acetyltryptophan, gramine and indoles with carboxy or aldehyde moieties are not accepted as substrates by PrnA.

Several years later, in 2013, the substrate scope of the tryptophan 7-halogenase RebH came into focus.112  In comparison to the closely related tryptophan 7-halogenase PrnA mentioned above, RebH accepts several non-natural substrates with high regioselectivity even for electronically unfavoured positions. For example, halogenation of tryptamine, N-methyltryptamine, tryptophol and 2-methyltryptamine at the C7 position have been demonstrated. In addition, even substituted napthalene derivatives, e.g. 2-naphthylamine, or tricyclic tryptolines that are quite different from the natural substrate tryptophan are halogenated by RebH. Since PrnA and RebH are both tryptophan 7-halogenases with a high sequence identity of 55%,112  the difference in the substrate range and the preferred ability of RebH to halogenate at the deactivated C7 position is intriguing. Analysis of the binding pocket of both enzymes reveals only four different amino acids of 24 residues in 5 Å distance to the substrate.112 

An array of substituted indoles, obtained using a tryptophan synthase114  was used as substrates for biocatalytic halogenation by RebH.110  Interestingly, in many cases RebH is able to override the directing effects of several substituents already present at the aromatic ring and still halogenates at its preferred 7-position, e.g. to form 7-chloro-5-hydroxytryptophan or 7-chloro-5-fluorotryptophan (see Fig. 5).

Figure 5

Halogenated products formed by the tryptophan 7-halogenase wild type enzymes PrnA and RebH. Albeit their striking sequence similarity, PrnA halogenates unnatural substrates at the electronically preferred 2- or 3-position,113  whereas RebH mainly halogenates at its preferred 7-position. In addition, RebH accepts substrates quite distant to tryptophan, e.g. 2-naphtylamine or tricyclic triptoline.112  Even directing effects of substituents already present at the indole ring are overridden, forming synthetically challenging products like 5-hydroxy-7-chlorotryptophan.110 

Figure 5

Halogenated products formed by the tryptophan 7-halogenase wild type enzymes PrnA and RebH. Albeit their striking sequence similarity, PrnA halogenates unnatural substrates at the electronically preferred 2- or 3-position,113  whereas RebH mainly halogenates at its preferred 7-position. In addition, RebH accepts substrates quite distant to tryptophan, e.g. 2-naphtylamine or tricyclic triptoline.112  Even directing effects of substituents already present at the indole ring are overridden, forming synthetically challenging products like 5-hydroxy-7-chlorotryptophan.110 

Close modal

In 2015, the Micklefield group showed the successful chlorination of the non-indolic molecules kynurenine, anthranilamide and to a low extent anthranilic acid by wild-type PrnA and PyrH, which could be further improved via structure guided mutagenesis.115  Andorfer et al. examined up to 93 substrates in a preparative-scale comprehensive study to analyse the scope and specificity of flavin-dependent halogenases. They were able to show that the substrate scope of flavin-dependent halogenases is less strict than previously assumed.16 

As summarised above, wild-type flavin-dependent halogenases suffer from a narrow substrate scope, thereby limiting their biocatalytic application. Since the structures of many flavin-dependent halogenases are known and the mechanism has been largely elucidated (see Chapter 3), attempts to expand the substrate scope and regioselectivity of halogenases using rational design and site-directed mutagenesis modifying the substrate binding pocket were conducted.

After genetic incorporation of RebH into Catharanthus roseus,116  substantial amounts of accumulated 7-chlorotryptophan in the transgenic plant hinted towards a bottleneck in the biosynthesis of halogenated monoterpene indole alkaloids. Therefore, Glenn and co-workers focused on a series of 17 mutations in the active site of RebH that are in proximity of the carboxylate of tryptophan to redirect its specificity towards tryptamine, a direct downstream precursor of alkaloid biosynthesis.117  The resulting RebH Y455W mutant shows a 30-fold higher accumulation of chlorinated tryptamine in presence of tryptophan, indicating a higher selectivity towards the decarboxylated substrate and thereby bypassing the initial bottleneck. Based on the crystal structure, the narrower space in the redesigned active site due to the replacement of phenylalanine with tryptophan seems to exclude its natural substrate, whereas the decarboxylated tryptamine still fits into the pocket.117  Additionally, some approaches to widen the binding pocket by rational design have been made. The single amino acid replacement F103A in PrnA resulted in a change of its natural regioselectivity from the 7-position to a mixture of 5- and 7-halogenated tryptophan.111  The smaller alanine residue apparently allows the adoption of a different conformation of the substrate inside the active pocket, leading to products halogenated at the 5-position.

Shepherd et al. attempted to switch the regioselectivity of the tryptophan 6-halogenase SttH towards C5 halogenation. They compared the active site residues of SttH to the natural tryptophan 5-halogenase PyrH and generated a triple mutant of SttH, which showed almost the same enzymatic activity as the wild type but with less regioselectivity, producing a 32% : 68% mixture of the regioisomers l-5- and l-6-chlorotryptophan.64  Similar results, a 50% : 50% mixture of l-5- and l-6-chlorotryptophan were obtained by Kim and co-workers with the quintuple mutant SatH-GI-FET.87 

Moritzer et al. mutated five residues in the active site of the tryptophan 6-halogenase Thal by rational design to the corresponding residues in RebH, which gave a flavin-dependent tryptophan 7-halogenase. The mutant Thal-RebH5 shows a shift in its regioselectivity from C6 (Thal) to 95% C7 (RebH).58 

Whereas the rational approach based on identification of potential amino acids in the active site and their exchange by site-directed mutagenesis has significantly contributed to the understanding of protein structure and enzyme function, the development of optimised, tailor-made biocatalysts nowadays mainly relies on random or semi-rational approaches with selection-driven directed evolution strategies.118  This technique involves the formation of random libraries consisting of thousands of mutants, e.g. by error-prone PCR (epPCR), iterative saturation mutagenesis119  or DNA shuffling120  with subsequent selection of those enzyme variants possessing the desired characteristics, e.g. enhanced activity, enantioselectivity or broadened substrate scope. Regarding tryptophan halogenases, the major bottleneck for directed evolution is the high-throughput screening assay to identify the desired enzyme variant.

First attempts to optimise the tryptophan 7-halogenase RebH by directed evolution were based on epPCR followed by HPLC analysis to quantify halogenation activity of the resulting clones in 96-well plates.121  Although HPLC-based screenings suffer from comparatively low throughput and high data analysis efforts, Poor and co-workers screened 3381 enzyme variants in total for the best turnover after a heat treatment. Over three rounds of evolution, they were able to obtain a significantly more thermostable RebH variant. In each round, beneficial mutations were combined and these variants were used as a starting point for the next round of directed evolution. The final variant contained eight mutations, leading to an increase in protein melting temperature (TM) by 18 °C. The resulting protein is useful as a robust precursor for further optimisation121  and was further evolved to site-selectively halogenate much larger molecules compared to tryptophan.122  In four additional rounds of directed evolution, the size of the accepted substrate was gradually increased in a so-called substrate walking approach from tryptophan over tricyclic tryptoline towards very bulky molecules like carvedilol. This non-selective beta blocker has a molecular mass twice as high as the native substrate tryptophan, which makes a rational explanation of the binding mode difficult (see Fig. 6A).122 

Figure 6

Halogenase screening assays suitable for the analysis of a large number of enzyme variants obtained during directed evolution. Coupling of 3-aminophenylboronic acid to the enzymatically formed aryl halide by a Suzuki–Miyaura reaction leads to significantly altered spectral properties of the biaryl. Hence, a sensitive versatile fluorescence-based halogenase assay based on this fluorogenic Suzuki–Miyaura derivatisation allows the tracing of brominated tryptophan in lysate in a high-throughput format with high specificity, suitable for tryptophan 5-, 6- and 7-halogenases (B, right sequence).4  The enzymatic oxidation of catechols to their corresponding ortho-quinones by horseradish peroxidase forms a Michael acceptor in situ, that can directly form coloured adducts with (halogenated) arylamines in a Michael addition reaction. The formed products show significantly different absorption maxima that allow the quantification of the enzymatic halogenation of arylamines via colorimetric analysis (B, middle sequence).124  In contrast to both methods described before, monitoring the reaction process by HPLC is the most versatile method, however with a relatively low throughput combined with high efforts for data analysis (B, left sequence).121  Recent investigations on the directed evolution of RebH using an HPLC-based enzyme assay yielded variants with an expanded substrate scope. The evolved enzyme enables the halogenation of gradually larger molecules that were quite different to tryptophan over several rounds of evolution (A).122 

Figure 6

Halogenase screening assays suitable for the analysis of a large number of enzyme variants obtained during directed evolution. Coupling of 3-aminophenylboronic acid to the enzymatically formed aryl halide by a Suzuki–Miyaura reaction leads to significantly altered spectral properties of the biaryl. Hence, a sensitive versatile fluorescence-based halogenase assay based on this fluorogenic Suzuki–Miyaura derivatisation allows the tracing of brominated tryptophan in lysate in a high-throughput format with high specificity, suitable for tryptophan 5-, 6- and 7-halogenases (B, right sequence).4  The enzymatic oxidation of catechols to their corresponding ortho-quinones by horseradish peroxidase forms a Michael acceptor in situ, that can directly form coloured adducts with (halogenated) arylamines in a Michael addition reaction. The formed products show significantly different absorption maxima that allow the quantification of the enzymatic halogenation of arylamines via colorimetric analysis (B, middle sequence).124  In contrast to both methods described before, monitoring the reaction process by HPLC is the most versatile method, however with a relatively low throughput combined with high efforts for data analysis (B, left sequence).121  Recent investigations on the directed evolution of RebH using an HPLC-based enzyme assay yielded variants with an expanded substrate scope. The evolved enzyme enables the halogenation of gradually larger molecules that were quite different to tryptophan over several rounds of evolution (A).122 

Close modal

Belsare et al. demonstrated the utility of a targeted combinatorial codon mutagenesis (CCM) method on three different enzymes, of which one was the tryptophan 7-halogenase RebH.123  By mutating 25 residues comprising the FAD-binding pocket of RebH variant 0K (RebH-E461K),16  the catalytic importance of this residues was evaluated. Screening of more than 1000 mutants by UPLC finally revealed a RebH variant 1K (R231K) with a 1.7-fold increased conversion of 2-aminobenzoic acid.123  Although these results demonstrate the power of directed evolution for optimisation of tryptophan halogenases, there is still a need for a facile high-throughput screening assay apart from HPLC analysis to minimise the screening effort. These screens need to deal with large numbers of variants and preferentially rely on the formation of colorimetric or fluorometric compounds to obtain an easy and automatic data analysis.118  Regarding tryptophan halogenases, the major issue is to distinguish selectively between tryptophan and its halogenated derivative without separation by HPLC (see Fig. 6B, left sequence). In parallel to the directed evolution approaches mentioned above, a colorimetric high-throughput assay for arylamine halogenation was developed, also suitable for enzymatic halogenation.124  The oxidation of catechols to their corresponding ortho-quinones by horseradish peroxidase forms a Michael acceptor in situ that can directly form a coloured adduct with arylamines in a Michael addition reaction. When adducts formed from halogenated naphthylamine are compared with adducts from non-halogenated parent molecules, both compounds differ significantly in their spectral properties, making this an easy method to quantify enzymatic halogenation (see Fig. 6B, middle sequence). Since RebH accepts 2-napthylamine as a substrate, this assay might be useful for the optimisation of RebH activity towards that substrate. However, it is worth mentioning that due to the involvement of a Michael addition reaction, only arylamines as substrates are possible, preventing this assay from being used for substrate scope expansion. In addition, its applicability in 96-well plate screening assays is also questionable since numerous arylamines are present in bacterial lysate, e.g. the DNA bases.

A different approach to develop a high-throughput screening assay for halogenases was envisaged by Schnepel et al. A robust protocol for the Suzuki–Miyaura cross-coupling of brominated tryptophan with an arylboronic acid in E. coli lysate was established, suitable for the application in directed evolution in 96-well plates.4  The coupling of 3-aminophenyl boronic acid to brominated tryptophan leads to a fluorophore that can be used for direct quantification of halogenase activity in lysate in a 96-well plate reader. This assay can probably be expanded to indoles and other aryl halides as well, paving the way for efficient screening of large libraries in directed evolution (see Fig. 6B, right sequence).4 

After closer examination of the rebeccamycin biosynthesis, the Sewald group took advantage of the natural high selectivity of the l-amino acid oxidase RebO, an enzyme that oxidises l-7-halotryptophans in the next biosynthetic step after halogenation by RebH. RebO shows high selectivity (10-fold) for l-7-halotryptophans in comparison to its precursor l-tryptophan. This strong preference was exploited in a colorimetric RebO-coupled halogenase assay that indirectly quantifies l-7-halotryptophans in crude lysate via the stoichiometric formation of H2O2 in combination with a peroxidase catalysed reaction.125 

As the separation of regioisomers by HPLC can be challenging due to similar chemical properties, the Lewis group developed a mass-spectrometry based screening assay to identify RebH mutants with altered regioselectivity. Because of its synthetic availability, C7-deuterated tryptamine was used as substrate in the biotransformation. Mutants with different regioselectivity towards C7-deuterated tryptamine could be identified via high-throughput MALDI-ToF-MS owing to the mass difference of Δm/z=1 caused by the deuterium that remains in the molecule once a position other than the C–D site is being halogenated.72 

Beside the substrate-, enantio-, and regioselectivity, the stability of halogenases came into focus of enzyme engineering. The high-throughput fluorescence assay established by Schnepel et al. enabled them to identify a thermostable Thal double mutant (Thal-GR) from random mutagenesis with a TM=57 °C (TM (Thal-WT)=47 °C).4  Based on these findings, Minges et al. identified the mutation S359G as crucial for the increased thermostability, although the mutation is located in the active site.126  By rational combination of this mutation with directed evolution and site-saturation mutagenesis, a Thal triple mutant (Thal-GLV) with highly improved thermostability (TM=71 °C) and specific activity at 25 °C compared to the wild type, was obtained. In addition, the mutation K374W turned out to increase the overall halogenation efficiency at 25 °C significantly.126 

In the course of developing novel sustainable processes, traditional synthetic routes are often combined with, or even in some cases replaced by, biocatalytic approaches. Regarding halogenases, the high regioselectivity in combination with very mild reaction conditions and their compatibility with many unprotected functional groups make these enzymes highly interesting for integration into conventional chemistry. Investigations to use halogenases for chemical purposes started with mainly analytical scale reactions exploring the substrate scope.113  The most challenging hurdle to upscale halogenase reactions is their low operational stability in combination with low specific activities which makes preparative applications elusive.34,50,52,69,81,127  Lewis and co-workers attempted to use the halogenase RebH for regioselective tryptophan chlorination on a 100 mg scale.112  The coexpression of two chaperones and the fusion of the flavin reductase RebF to maltose binding protein (MBP) increased the recombinant expression level of RebH and RebF in E. coli nearly 10-fold, enabling the production of both enzymes in sufficient scale for preparative applications. Concomitant cofactor regeneration was achieved using a glucose dehydrogenase that oxidises glucose to gluconolactone to generate NADH from NAD+ and MBP-RebF that consumes NADH for the formation of FADH2 from FAD.112  By applying E. coli lysate containing overexpressed RebH for biocatalytic halogenation, this setup yielded nearly 100 mg 7-chlorotryptophan, although a large volume of E. coli culture was necessary since the enzyme is not stable under in vitro reaction conditions for more than 24 hours.112 

In 2015, Frese et al. found a straightforward solution to tackle the low operational stability of halogenases. Cross-linked enzyme aggregates (CLEAs) are formed by ammonium sulfate precipitation of the enzymes followed by glutaraldehyde cross-linking. This immobilisation of the halogenase RebH together with all necessary auxiliary enzymes needed for concomitant cofactor regeneration, i.e. the flavin reductase PrnF from Pseudomonas fluorescens and an alcohol dehydrogenase from Rhodococcus Sp., leads to a solid and recyclable biocatalyst that is stable for several days under reaction conditions.128  The CLEA approach improved the operational stability and led to quantitative formation of 7-bromotryptophan up to a multi-gram scale within 8 days with an easy purification strategy based on simple desalting by C18 silica filtration.128 

Cascade reactions are of great interest for chemists since they can be implemented in classic organic synthesis, as well as in enzymatic reactions. From an economic and ecological perspective, one-pot reaction cascades avoid unnecessary purification steps and afford big advantages in terms of time, solvent and waste savings.129  Hence, green processes like chemoenzymatic reaction cascades have become popular targets in academic and industrial research efforts during the past years.130  Chemoenzymatic one-pot reactions may also combine the stereo- and regioselectivity of enzymes with the high and unique reactivity of chemical catalysts.131,132  Flavin-dependent halogenases can be used in tailor-made synthetic pathways supplying halogenated compounds.133  On the other hand, halogen substituents are an excellent handle for late-stage diversifications.130,134–139 

Beside using overproduced and isolated halogenases, several workgroups also focused on the genetic incorporation of tryptophan halogenases into the complete metabolism of plants or bacteria to form new complex halogenated compounds.140 

The combinatorial reconstitution of the entire biosynthetic pathway of rebeccamycin in Streptomyces albus using genes isolated from the rebeccamycin-producing microorganism Lechevalieria aerocolonigenes as well as genes originating from the staurosporine-producing microbe Streptomyces longisporoflavus leads to numerous new antitumor indolocarbazole alkaloids. In total, 13 genes were assembled in 22 combinations leading to 32 different natural products. The co-expression of the tryptophan 5- or tryptophan 6-halogenase genes pyrH and thal, respectively, resulted in new indolocarbazole derivatives halogenated at the 5- or 6-position.141 

In most of the mentioned metabolic engineering approaches, the major issue is the mandatory acceptance of halogenated unnatural precursors in all following enzymatic steps. This problem results in the accumulation of halogenated intermediates instead of the halogenated final natural products in some cases. Seibold et al. observed the production of a halogenated phenylpyrrol intermediate during the biosynthesis of pyrrolnitrin in Pseudomonas aureofaciens when the original tryptophan 7-halogenase PrnA was replaced by the tryptophan 6-halogenase Thal. The biosynthesis stops at that intermediate and no pyrrolnitrin analog halogenated at the 6-position was observed.81 

As indole moieties in natural products are commonly formed from tryptophan as a precursor, the incorporation of tryptophan halogenases into their biosynthetic pathway provides the opportunity to rationally design novel complex natural products with altered bioactivity that are otherwise only accessible by laborious total synthesis. The genes encoding the tryptophan 7-halogenase RebH or the tryptophan 5-halogenase PyrH, respectively, were integrated in the medicinal plant Catharanthus roseus (Madagascar periwinkle), forming regioselectively chlorinated tryptophan that is incorporated in a predictable manner into complex monoterpene indole alkaloids.116  The halogen substituent in these modified alkaloids could also be addressed specifically after extraction in biorthogonal cross-coupling reactions on an analytical scale to form a set of different aryl and heteroaryl analogues (see Fig. 7A).142  In a similar approach, prnA was integrated into Streptomyces coeruleorubidus resulting in efficient production of chlorinated uridyl peptide antibiotic pacidamycin, in situ. After solid-phase extraction, the halogenated peptide could be diversified by Suzuki–Miyaura cross-coupling reactions (see Fig. 7B).143  Later on, Sharma et al. developed a so-called ‘cross-coupling medium’ containing sodium bromide for halogenation, potassium nitrate as sole nitrogen source and glycerol as carbon source for E. coli or Streptomyces coelicolor, respectively.144  This growth medium does not contain amino acids or glucose which impair Pd-catalysed cross-couplings.144  The authors screened different catalytic systems finding Pd(OAc)2 in combination with guanidyl ligands like 2-dimethylamino-4,6-dihydroxypyrimidine (DMADHP) or tetramethylguanidine (TMG) to have a sufficient catalytic activity at 37 °C. By changing the pacidamycin production strain to S. coelicolor, they were able to perform in vivo bromination of pacidamycin and subsequent Suzuki–Miyaura diversification in presence of living S. coelicolor cells (see Fig. 7C).144 

Figure 7

Integration of RebH into the monoterpene alkaloid biosynthesis yields chlorinated indole alkaloids116  that also can be diversified using Suzuki–Miyaura couplings (A).142  The incorporation of the tryptophan 7-halogenase gene prnA into Streptomyces coeruleorubidus leads to the formation of a chlorinated pacidamycin derivative that also can be modified via Pd-mediated cross-coupling reactions (B).143  Using a special ‘cross-coupling medium’ and a tetramethylguanidyl ligand, a Suzuki–Miyaura cross-coupling with in vivo brominated pacidamycin was possible in presence of living Streptomyces coelicolor cells (blue box) (C). Only little impact on cell viability was observed.144 

Figure 7

Integration of RebH into the monoterpene alkaloid biosynthesis yields chlorinated indole alkaloids116  that also can be diversified using Suzuki–Miyaura couplings (A).142  The incorporation of the tryptophan 7-halogenase gene prnA into Streptomyces coeruleorubidus leads to the formation of a chlorinated pacidamycin derivative that also can be modified via Pd-mediated cross-coupling reactions (B).143  Using a special ‘cross-coupling medium’ and a tetramethylguanidyl ligand, a Suzuki–Miyaura cross-coupling with in vivo brominated pacidamycin was possible in presence of living Streptomyces coelicolor cells (blue box) (C). Only little impact on cell viability was observed.144 

Close modal

Since the production of dyes by biosynthetic approaches is of great interest and introduction of halogens can increase spectroscopic and antibacterial properties, different attempts to incorporate flavin-dependent halogenases into the biosynthesis of dyes have been reported. By integrating the genes of the tryptophan 7-halogenase PrnA together with an engineered cytochrome P450 monooxygenase (CYP102G4) that hydroxylates the C3 position of halogenated indoles into E. coli, the in vivo one-pot biosynthesis of 7,7′-dichloroindigo by an enzymatic cascade was possible (see Fig. 8). Feeding halogenated indoles to E. coli cells expressing CYP102G4 led to a broad scope of halogenated indigo dyes with antibacterial properties.145  From an economic point of view, a one-pot synthesis of the halogenated compound starting with non-halogenated reactants is more attractive than an indirect biotransformation using halogenated indole precursor feedstocks. Therefore, investigations involving other halogenases than PrnA in combination with the oxygenase CYP102G4 are of great interest.

Figure 8

One-pot in vivo biosynthesis of 7,7′-dichloroindigo dye using the flavin-dependent tryptophan halogenase PrnA and the cytochrome P450 enzyme CYP102G4 in E. coli.145 

Figure 8

One-pot in vivo biosynthesis of 7,7′-dichloroindigo dye using the flavin-dependent tryptophan halogenase PrnA and the cytochrome P450 enzyme CYP102G4 in E. coli.145 

Close modal

The functionality of flavin-dependent halogenases in Nicotiana benthamiana (tobacco) plants indicated that incorporation of halogenases into plants for dye production might be possible.146  New-to-nature halogenated indicans, precursors in the indigo biosynthesis in plants, could be synthesised by a tailor-made biosynthetic enzyme cascade in tobacco. Therefore, genes encoding a microbial flavin-dependent halogenase (PyrH (C5), SttH (C6), RebH (C7)), a bacterial tryptophanase TnaA and an engineered mutant of the human cytochrome P450 monooxygenase 2A6 (2A6mut; L241C/N297Q) were infiltrated into the indican biosynthesis of tobacco plants. The glycosylation was performed by an endogenous tobacco glucosyltransferase yielding the desired halogenated metabolites (see Fig. 9).147 

Figure 9

Tailor-made biosynthesis of halogenated indicans in Nicotiana benthamiana (tobacco). The glucosyltransferase enzyme is endogeneous to the host plant. Tryptophan halogenases are PyrH from S. rugosporus, SttH from S. toxitricini and RebH from L. aerocolonigenes. The Tryptophanase TnaA originates from E. coli. 2A6mut is a L241C/N297Q mutant of human cytochrome P450 monooxygenase 2A6.147 

Figure 9

Tailor-made biosynthesis of halogenated indicans in Nicotiana benthamiana (tobacco). The glucosyltransferase enzyme is endogeneous to the host plant. Tryptophan halogenases are PyrH from S. rugosporus, SttH from S. toxitricini and RebH from L. aerocolonigenes. The Tryptophanase TnaA originates from E. coli. 2A6mut is a L241C/N297Q mutant of human cytochrome P450 monooxygenase 2A6.147 

Close modal

Furthermore, the halogenation of coumarins, which are known as commercial fluorescent dyes or therapeutic agents,148–150  is a nice example of the versatility of halogenases. Menon et al. engineered a de novo biosynthetic pathway to synthesise the halogenated “non-natural” 8-chloro-7-hydroxycoumarin in E. coli (see Fig. 10).74  The incorporation of four different genes was necessary to achieve this goal. The ammonia lyase TAL from Sacharothrix espanaensis catalyses the elimination of ammonia from tyrosine in vivo giving p-coumaric acid, which is then converted by the incorporated coumaryl-CoA ligase 4CL from Streptomyces coelicolor. Incubation of feruloyl-CoA 6′-hydroxylyase F6’H from Ipomoea batatas yields 7-hydroxycoumarin, which can then be halogenated in vivo by the phenolic flavin-dependent halogenase RadH variant D465E/T501S.74  This tailor-made enzymatic cascade yielded up to 1.1 mg L−1 of 8-chloro-7-hydroxycoumarin under fermentation conditions, giving an example for the production of halogenated coumarin from glucose in E. coli.74 

Figure 10

Engineered pathway using the flavin-dependent phenolic halogenase RadH for in vivo biosynthesis of “non-natural” 8-chloro-7-hydroxycoumarin in E. coli.74 

Figure 10

Engineered pathway using the flavin-dependent phenolic halogenase RadH for in vivo biosynthesis of “non-natural” 8-chloro-7-hydroxycoumarin in E. coli.74 

Close modal

Since Corynebacterium glutamicum, the work horse for fermentative amino acid production is unable to use tryptophan as a nitrogen source, it is a promising candidate for in vivo production of halotryptophans.151  Wendisch and co-workers investigated its capability for a fermentative production of halogenated tryptophan from halide salts, sugars and ammonia. Therefore, the genes of the flavin-dependent tryptophan-7-halogenase RebH and the corresponding flavin reductase RebF were introduced into a recombinant strain of C. glutamicum (see Fig. 11).152  Very recently, Veldmann et al. were able to utilise this metabolically engineered C. glutamicum to produce l-7-bromotryptophan in a fermentative process, yielding up to 1.2 g L−1.108 

Figure 11

Engineered metabolic pathway for the overproduction of tryptophan and the halogenation of tryptophan in a fermentative process using C. glutamicum. The anthranilic synthase is endogenously overexpressed, whereas the aldolase AroG (Mycolicibacterium smegmatis), tryptophan halogenase RebH, the flavin reductase RebF (both Lechevalieria aerocolonigenes) and the anthranilate phosphoribosyltransferase (E. coli) are expressed heterologously.108 

Figure 11

Engineered metabolic pathway for the overproduction of tryptophan and the halogenation of tryptophan in a fermentative process using C. glutamicum. The anthranilic synthase is endogenously overexpressed, whereas the aldolase AroG (Mycolicibacterium smegmatis), tryptophan halogenase RebH, the flavin reductase RebF (both Lechevalieria aerocolonigenes) and the anthranilate phosphoribosyltransferase (E. coli) are expressed heterologously.108 

Close modal

As halogen substituents are excellent handles for late-stage diversifications,142,143,153  Pd-catalysed cross-couplings have come into focus. Since enzymatic halogenation is highly regioselective with the formation of only one desired regioisomer, preparative chemoenzymatic one-pot approaches combining enzymatic halogenation and cross-coupling have recently attracted attention as a powerful synthetic tool to selectively discriminate between two similar, unactivated C–H positions for modification. However, combination of chemo- and biocatalysis often is challenging because of incompatibilities. Latham et al. determined the effect of cofactors and enzymes on the Pd-catalysed cross-couplings.154  They observed a deleterious effect of the cofactors but a major reduction of reactivity due to the presence of proteins.154  These may coordinate the transition metal and potentially inhibit its catalytic activity. Therefore, a compartmentalisation or immobilisation of at least one catalyst seems indispensable.154  To achieve this, the reaction mixture containing a purified halogenase (PyrH, RadH, RebH or SttH) and the flavin reductase Fre was filtered using a 10 kDa filter to remove the enzymes, followed by a subsequent Suzuki–Miyaura cross-coupling in crude filtrate. This approach afforded good yields in a two-step one-pot reaction (see Fig. 12).

Figure 12

Chemoenzymatic cascade reaction by Latham et al. After enzymatic halogenation using the halogenase RadH and flavin reductase Fre, the crude product was filtered using a 10 kDa size exclusion membrane. Subsequent Suzuki–Miyaura cross-coupling using the water soluble trisodium 3,3′,3-phosphinetriyltribenzenesulfonate (TPPTS) as ligand was performed in crude filtrate without further purification.154 

Figure 12

Chemoenzymatic cascade reaction by Latham et al. After enzymatic halogenation using the halogenase RadH and flavin reductase Fre, the crude product was filtered using a 10 kDa size exclusion membrane. Subsequent Suzuki–Miyaura cross-coupling using the water soluble trisodium 3,3′,3-phosphinetriyltribenzenesulfonate (TPPTS) as ligand was performed in crude filtrate without further purification.154 

Close modal

By using the CLEA methodology, the aryl halide solution could be directly used for subsequent Suzuki–Miyaura cross-coupling reactions in a two-step one-pot approach. The immobilised biocatalyst was removed by centrifugation (see Fig. 13).154  In addition, purification can be facilitated by an additional Nα-Boc protection as a third step in a three-step, one-pot approach.155  Finally, a single extraction step leads to an array of several 5-, 6- or 7-(hetero)-aryl-substituted tryptophan derivatives readily usable for peptide or peptidomimetic synthesis (see Fig. 14).155 

Figure 13

The preparative halogenation of tryptophan was achieved by immobilisation of the tryptophan halogenase (PyrH, RebH, or SttH) with all auxiliary enzymes (flavin reductase Fre, alcohol dehydrogenase ADH) necessary for concomitant cofactor regeneration as cross-linked aggregates (CLEAs). The biocatalyst was removed by centrifugation and Pd-mediated Suzuki–Miyaura cross-coupling was performed in the crude supernatant.154 

Figure 13

The preparative halogenation of tryptophan was achieved by immobilisation of the tryptophan halogenase (PyrH, RebH, or SttH) with all auxiliary enzymes (flavin reductase Fre, alcohol dehydrogenase ADH) necessary for concomitant cofactor regeneration as cross-linked aggregates (CLEAs). The biocatalyst was removed by centrifugation and Pd-mediated Suzuki–Miyaura cross-coupling was performed in the crude supernatant.154 

Close modal
Figure 14

Biocatalytic application of flavin-dependent halogenases combined with a chemocatalytic cross-coupling reaction. Depending on the immobilised halogenase, tryptophan was brominated regioselectively at the 5, 6, or 7 position, respectively. In a subsequent Suzuki–Miyaura cross coupling reaction, a set of (hetero)aryl-tryptophan derivatives could be synthesised in a one-pot approach.155 

Figure 14

Biocatalytic application of flavin-dependent halogenases combined with a chemocatalytic cross-coupling reaction. Depending on the immobilised halogenase, tryptophan was brominated regioselectively at the 5, 6, or 7 position, respectively. In a subsequent Suzuki–Miyaura cross coupling reaction, a set of (hetero)aryl-tryptophan derivatives could be synthesised in a one-pot approach.155 

Close modal

Additionally a compartmentalisation of both catalytic processes is also possible e.g. by using a polydimethylsiloxane (PDMS) membrane. Due to its impermeability for ionic species, it is capable of compartmentalising the cross-coupling and biocatalysis (see Fig. 15).154  However, tryptophan as an amino acid cannot pass the membrane. Hence, this method is limited to non-ionic indole derivatives like tryptophol as substrates for halogenases. Moreover, the Pd-catalyst needs high temperatures which makes a simultaneous combination of biocatalysis and cross-coupling impossible. A parallel cross-coupling would allow higher substrate concentrations since product inhibition of the enzymes would be prevented. Halotryptophan would be converted directly to the cross-coupling product that most likely will compete to a much lower extent with tryptophan for the halogenase binding pocket. At the same time, the equilibrium is shifted in favour of the product. Therefore, more reactive catalysts and a compartmentalisation element capable of holding back both catalysts but being permeable for tryptophan would be desirable for direct synthesis of arylated amino acids usable in peptide synthesis.

Figure 15

Consecutive two-step one-pot synthesis of arylated tryptophol combining biocatalytic bromination CLEAs and Suzuki–Miyaura coupling with compartmentalisation by a PDMS-thimble. Halogenation is done in phosphate buffer outside the PDMS-thimble at room temperature overnight. Suzuki–Miyaura cross-coupling starts inside the thimble by heating the mixture to 80 °C for additional 24 h.154 

Figure 15

Consecutive two-step one-pot synthesis of arylated tryptophol combining biocatalytic bromination CLEAs and Suzuki–Miyaura coupling with compartmentalisation by a PDMS-thimble. Halogenation is done in phosphate buffer outside the PDMS-thimble at room temperature overnight. Suzuki–Miyaura cross-coupling starts inside the thimble by heating the mixture to 80 °C for additional 24 h.154 

Close modal

Besides Suzuki–Miyaura cross-couplings, other Pd-mediated cross-couplings have been established in chemoenzymatic reaction cascades. Durak et al. performed a Buchwald–Hartwig amination with unprotected enzymatically halogenated tryptoline in crude halogenation-extracts without prior purification (see Fig. 16).156  Gruß et al. were the first ones combining flavin-dependent tryptophan halogenases with a subsequent Mizoroki–Heck coupling in a three-step one-pot reaction, receiving Boc-protected fluorogenic tryptophan derivatives (see Fig. 17),157  followed by Pubill-Ulldemolins and Sharma et al. using a Mizoroki–Heck reaction for the modification of complex natural compounds such as bromo-pacidamycin and barretin.158 

Figure 16

Chemoenzymatic reaction cascade with purified RebH. After removing cross-coupling-hindering enzymes by filtration over Celite and a single extraction step without further purification, a subsequent Buchwald–Hartwig amination was performed in dioxan.156 

Figure 16

Chemoenzymatic reaction cascade with purified RebH. After removing cross-coupling-hindering enzymes by filtration over Celite and a single extraction step without further purification, a subsequent Buchwald–Hartwig amination was performed in dioxan.156 

Close modal
Figure 17

Biocatalytic halogenation was performed utilising CLEAs containing the tryptophan 7-halogenase RebH. The solid biocatalyst was removed followed by subsequent Mizoroki–Heck diversification without prior isolation of the brominated tryptophan. After full conversion, Boc2O and an additional base were added to the mixture and the product was isolated by a simple extraction giving good yields after a three-step one-pot synthesis.157 

Figure 17

Biocatalytic halogenation was performed utilising CLEAs containing the tryptophan 7-halogenase RebH. The solid biocatalyst was removed followed by subsequent Mizoroki–Heck diversification without prior isolation of the brominated tryptophan. After full conversion, Boc2O and an additional base were added to the mixture and the product was isolated by a simple extraction giving good yields after a three-step one-pot synthesis.157 

Close modal

Besides combining transition-metal catalysed reactions with enzymatic halogenation in reaction cascades, classical chemical reactions can also be considered. Schnepel et al. combined an enzymatic cascade of the tryptophan halogenases PyrH and RebH with the stereoselective l-tryptophan oxidase RebO and a simple reduction by ammonia borane for a dynamic stereoinversion to yield brominated d-tryptophans (see Fig. 18). RebO stereoselectively oxidises l-5- and 7-bromotryptophan to the corresponding α-imino acid, which is in turn reduced nonselectively by ammonia borane to the racemic amino acid. From the d/l-mixture, only the l-enantiomer is continuously converted by the biocatalyst, leading to an accumulation of the desired d-bromotryptophan in good yields and high enantiopurity (94–98% ee).125 

Figure 18

Enzymatic one-pot cascade using the specific l-amino acid oxidase RebO for dynamic stereoinversion of previous halogenated l-bromotryptophans in preparative scales.125 

Figure 18

Enzymatic one-pot cascade using the specific l-amino acid oxidase RebO for dynamic stereoinversion of previous halogenated l-bromotryptophans in preparative scales.125 

Close modal

Optimisation of enzyme stability and activity are important goals in directed evolution, and further breakthroughs are expected in the future of biocatalytic halogenation. Recent developments already show the potential of halogenases for preparative halogenation up to a multi-gram scale. Expansion of the substrate scope by means of directed evolution also guided halogenases to perform regioselective halogenation on very large unnatural substrates. New methods for high-throughput screenings allow the even faster and more convenient adaptation of halogenases to specific substrates to obtain optimised and tailor-made biocatalysts. Combined with chemocatalytic modification of the halogenated compounds, flavin-dependent halogenases are currently moving from interesting low-turnover enzymes to striking biocatalysts suitable for preparative bioorganic synthesis.

To fully exploit their potential, insight from structural and mechanistical studies is needed. The understanding of most of the general reaction mechanism helped pave the way for rational enzyme design, resulting e.g. in switches in regioselectivity. Other questions such as the basis for halide specificity and substrate preference, especially of variant B halogenases, still require more investigation.

1.
F.
Diederich
and
P. J.
Stang
,
Metal-catalyzed Cross-coupling Reactions
,
Wiley-VCH
,
New York
,
1998
2.
Jeschke
 
P.
ChemBioChem
2004
, vol. 
5
 pg. 
571
 
3.
Wilcken
 
R.
Zimmermann
 
M. O.
Lange
 
A.
Joerger
 
A. C.
Boeckler
 
F. M.
J. Med. Chem.
2013
, vol. 
56
 pg. 
1363
 
4.
Schnepel
 
C.
Minges
 
H.
Frese
 
M.
Sewald
 
N.
Angew. Chem., Int. Ed.
2016
, vol. 
55
 pg. 
14159
 
5.
Jacks
 
T. E.
Belmont
 
D. T.
Briggs
 
C. A.
Horne
 
N. M.
Kanter
 
G. D.
Karrick
 
G. L.
Krikke
 
J. J.
McCabe
 
R. J.
Mustakis
 
J. G.
Nanninga
 
T. N.
Risedorph
 
G. S.
Seamans
 
R. E.
Skeean
 
R.
Winkle
 
D. D.
Zennie
 
T. M.
Org. Process Res. Dev.
2004
, vol. 
8
 pg. 
201
 
6.
Gribble
 
G. W.
Mar. Drugs
2015
, vol. 
13
 pg. 
4044
 
7.
Hernandes
 
M. Z.
Cavalcanti
 
S. M. T.
Moreira
 
D. R. M.
de Azevedo Jr.
 
W. F.
Leite
 
A. C. L.
Curr. Drug Targets
2010
, vol. 
11
 pg. 
303
 
8.
Isanbor
 
C.
O'Hagan
 
D.
J. Fluorine Chem.
2006
, vol. 
127
 pg. 
303
 
9.
Park
 
B. K.
Kitteringham
 
N. R.
O'Neill
 
P. M.
Annu. Rev. Pharmacol.
2001
, vol. 
41
 pg. 
443
 
10.
Weiss
 
C.
Sammet
 
B.
Sewald
 
N.
Nat. Prod. Rep.
2013
, vol. 
30
 pg. 
924
 
11.
Teng
 
X.
Degterev
 
A.
Jagtap
 
P.
Xing
 
X.
Choi
 
S.
Denu
 
R.
Yuan
 
J.
Cuny
 
G. D.
Bioorg. Med. Chem. Lett.
2005
, vol. 
15
 pg. 
5039
 
12.
Berthelot
 
A.
Piguel
 
S.
Le Dour
 
G.
Vidal
 
J.
J. Org. Chem.
2003
, vol. 
68
 pg. 
9835
 
13.
Mentzel
 
U. V.
Tanner
 
D.
Tønder
 
J. E.
J. Org. Chem.
2006
, vol. 
71
 pg. 
5807
 
14.
van Pée
 
K.-H.
Salcher
 
O.
Lingens
 
F.
Liebigs Ann. Chem.
1981
, vol. 
1981
 pg. 
233
 
15.
Andorfer
 
M. C.
Lewis
 
J. C.
Annu. Rev. Biochem.
2018
, vol. 
87
 pg. 
159
 
16.
Andorfer
 
M. C.
Grob
 
J. E.
Hajdin
 
C. E.
Chael
 
J. R.
Siuti
 
P.
Lilly
 
J.
Tan
 
K. L.
Lewis
 
J. C.
ACS Catal.
2017
, vol. 
7
 pg. 
1897
 
17.
Bartoccini
 
F.
Fanini
 
F.
Retini
 
M.
Piersanti
 
G.
Tetrahedron Lett.
2020
, vol. 
61
 pg. 
151923
 
18.
Loach
 
R. P.
Fenton
 
O. S.
Amaike
 
K.
Siegel
 
D. S.
Ozkal
 
E.
Movassaghi
 
M.
J. Org. Chem.
2014
, vol. 
79
 pg. 
11254
 
19.
Hillwig
 
M. L.
Liu
 
X.
Nat. Chem. Biol.
2014
, vol. 
10
 pg. 
921
 
20.
Lowe
 
P. T.
Cobb
 
S. L.
O'Hagan
 
D.
Org. Biomol. Chem.
2019
, vol. 
17
 pg. 
7493
 
21.
O'Hagan
 
D.
Deng
 
H.
Chem. Rev.
2015
, vol. 
115
 pg. 
634
 
22.
Feng
 
X.
Bello
 
D.
Lowe
 
P. T.
Clark
 
J.
O'Hagan
 
D.
Chem. Sci.
2019
, vol. 
10
 pg. 
9501
 
23.
Höfler
 
G. T.
But
 
A.
Hollmann
 
F.
Org. Biomol. Chem.
2019
, vol. 
17
 pg. 
9267
 
24.
Voss
 
M.
Honda Malca
 
S.
Buller
 
R.
Chem. – Eur. J.
2020
, vol. 
26
 pg. 
7336
 
25.
Duewel
 
S.
Schmermund
 
L.
Faber
 
T.
Harms
 
K.
Srinivasan
 
V.
Meggers
 
E.
Hoebenreich
 
S.
ACS Catal.
2020
, vol. 
10
 pg. 
1272
 
26.
G. W.
Gribble
in
The Alkaloids: Chemistry and Biology
, ed. H.-J. Knölker,
Academic Press
,
2012
27.
Dedieu
 
A.
Gaillard
 
J.-C.
Pourcher
 
T.
Darrouzet
 
E.
Armengaud
 
J.
J. Biol. Chem.
2011
, vol. 
286
 pg. 
259
 
28.
Podzelinska
 
K.
Latimer
 
R.
Bhattacharya
 
A.
Vining
 
L. C.
Zechel
 
D. L.
Jia
 
Z.
J. Mol. Biol.
2010
, vol. 
397
 pg. 
316
 
29.
Zhu
 
T.
Cheng
 
X.
Liu
 
Y.
Deng
 
Z.
You
 
D.
Metab. Eng.
2013
, vol. 
19
 pg. 
69
 
30.
Weiss
 
C.
Figueras
 
E.
Borbely
 
A. N.
Sewald
 
N.
J. Pept. Sci.
2017
, vol. 
23
 pg. 
514
 
31.
Kirner
 
S.
Hammer
 
P. E.
Hill
 
D. S.
Altmann
 
A.
Fischer
 
I.
Weislo
 
L. J.
Lanahan
 
M.
van Pée
 
K. H.
Ligon
 
J. M.
J. Bacteriol.
1998
, vol. 
180
 pg. 
1939
 
32.
Hohaus
 
K.
Altmann
 
A.
Burd
 
W.
Fischer
 
I.
Hammer
 
P. E.
Hill
 
D. S.
Ligon
 
J. M.
van Pée
 
K.-H.
Angew. Chem., Int. Ed.
1997
, vol. 
36
 pg. 
2012
 
33.
Dairi
 
T.
Nakano
 
T.
Aisaka
 
K.
Katsumata
 
R.
Hasegawa
 
M.
Biosci., Biotechnol., Biochem.
1995
, vol. 
59
 pg. 
1099
 
34.
Keller
 
S.
Wage
 
T.
Hohaus
 
K.
Hölzer
 
M.
Eichhorn
 
E.
van Pée
 
K.-H.
Angew. Chem., Int. Ed.
2000
, vol. 
39
 pg. 
2300
 
35.
Nowak-Thompson
 
B.
Chaney
 
N.
Wing
 
J. S.
Gould
 
S. J.
Loper
 
J. E.
J. Bacteriol.
1999
, vol. 
181
 pg. 
2166
 
36.
Dorrestein
 
P. C.
Yeh
 
E.
Garneau-Tsodikova
 
S.
Kelleher
 
N. L.
Walsh
 
C. T.
Proc. Natl. Acad. Sci. U. S. A.
2005
, vol. 
102
 pg. 
13843
 
37.
Chankhamjon
 
P.
Tsunematsu
 
Y.
Ishida-Ito
 
M.
Sasa
 
Y.
Meyer
 
F.
Boettger-Schmidt
 
D.
Urbansky
 
B.
Menzel
 
K.-D.
Scherlach
 
K.
Watanabe
 
K.
Hertweck
 
C.
Angew. Chem., Int. Ed.
2016
, vol. 
55
 pg. 
11955
 
38.
Zeng
 
J.
Zhan
 
J.
ChemBioChem
2010
, vol. 
11
 pg. 
2119
 
39.
Schmartz
 
P. C.
Zerbe
 
K.
Abou-Hadeed
 
K.
Robinson
 
J. A.
Org. Biomol. Chem.
2014
, vol. 
12
 pg. 
5574
 
40.
Buedenbender
 
S.
Rachid
 
S.
Müller
 
R.
Schulz
 
G. E.
J. Mol. Biol.
2009
, vol. 
385
 pg. 
520
 
41.
Lin
 
S.
van Lanen
 
S. G.
Shen
 
B.
J. Am. Chem. Soc.
2007
, vol. 
129
 pg. 
12432
 
42.
Mori
 
S.
Pang
 
A. H.
Thamban Chandrika
 
N.
Garneau-Tsodikova
 
S.
Tsodikov
 
O. V.
Nat. Commun.
2019
, vol. 
10
 pg. 
1255
 
43.
Winter
 
J. M.
Moffitt
 
M. C.
Zazopoulos
 
E.
McAlpine
 
J. B.
Dorrestein
 
P. C.
Moore
 
B. S.
J. Biol. Chem.
2007
, vol. 
282
 pg. 
16362
 
44.
Edwards
 
D. J.
Marquez
 
B. L.
Nogle
 
L. M.
McPhail
 
K.
Goeger
 
D. E.
Roberts
 
M. A.
Gerwick
 
W. H.
Chem. Biol.
2004
, vol. 
11
 pg. 
817
 
45.
Piraee
 
M.
White
 
R. L.
Vining
 
L. C.
Microbiology
2004
, vol. 
150
 pg. 
85
 
46.
Kittilä
 
T.
Kittel
 
C.
Tailhades
 
J.
Butz
 
D.
Schoppet
 
M.
Büttner
 
A.
Goode
 
R. J. A.
Schittenhelm
 
R. B.
van Pee
 
K.-H.
Süssmuth
 
R. D.
Wohlleben
 
W.
Cryle
 
M. J.
Stegmann
 
E.
Chem. Sci.
2017
, vol. 
8
 pg. 
5992
 
47.
Ortega
 
M. A.
Cogan
 
D. P.
Mukherjee
 
S.
Garg
 
N.
Li
 
B.
Thibodeaux
 
G. N.
Maffioli
 
S. I.
Donadio
 
S.
Sosio
 
M.
Escano
 
J.
Smith
 
L.
Nair
 
S. K.
van der Donk
 
W. A.
ACS Chem. Biol.
2017
, vol. 
12
 pg. 
548
 
48.
van Pée
 
K.-H.
Patallo
 
E. P.
Appl. Microbiol. Biotechnol.
2006
, vol. 
70
 pg. 
631
 
49.
Gkotsi
 
D. S.
Ludewig
 
H.
Sharma
 
S. V.
Connolly
 
J. A.
Dhaliwal
 
J.
Wang
 
Y.
Unsworth
 
W. P.
Taylor
 
R. J. K.
McLachlan
 
M. M. W.
Shanahan
 
S.
Naismith
 
J. H.
Goss
 
R. J. M.
Nat. Chem.
2019
, vol. 
11
 pg. 
1091
 
50.
Yeh
 
E.
Garneau
 
S.
Walsh
 
C. T.
Proc. Natl. Acad. Sci. U. S. A.
2005
, vol. 
102
 pg. 
3960
 
51.
Dong
 
C.
Flecks
 
S.
Unversucht
 
S.
Haupt
 
C.
van Pée
 
K.-H.
Naismith
 
J. H.
Science
2005
, vol. 
309
 pg. 
2216
 
52.
Yeh
 
E.
Blasiak
 
L. C.
Koglin
 
A.
Drennan
 
C. L.
Walsh
 
C. T.
Biochemistry
2007
, vol. 
46
 pg. 
1284
 
53.
Zhu
 
X.
de Laurentis
 
W.
Leang
 
K.
Herrmann
 
J.
Ihlefeld
 
K.
van Pée
 
K.-H.
Naismith
 
J. H.
J. Mol. Biol.
2009
, vol. 
391
 pg. 
74
 
54.
Moritzer
 
A.-C.
Niemann
 
H. H.
Protein Sci.
2019
, vol. 
28
 pg. 
2112
 
55.
Pang
 
A. H.
Garneau-Tsodikova
 
S.
Tsodikov
 
O. V.
J. Struct. Biol.
2015
, vol. 
192
 pg. 
349
 
56.
Fraley
 
A. E.
Garcia-Borràs
 
M.
Tripathi
 
A.
Khare
 
D.
Mercado-Marin
 
E. V.
Tran
 
H.
Dan
 
Q.
Webb
 
G. P.
Watts
 
K. R.
Crews
 
P.
Sarpong
 
R.
Williams
 
R. M.
Smith
 
J. L.
Houk
 
K. N.
Sherman
 
D. H.
J. Am. Chem. Soc.
2017
, vol. 
139
 pg. 
12060
 
57.
Yeh
 
E.
Cole
 
L. J.
Barr
 
E. W.
Bollinger
 
J. M.
Ballou
 
D. P.
Walsh
 
C. T.
Biochemistry
2006
, vol. 
45
 pg. 
7904
 
58.
Moritzer
 
A.-C.
Minges
 
H.
Prior
 
T.
Frese
 
M.
Sewald
 
N.
Niemann
 
H. H.
J. Biol. Chem.
2019
, vol. 
294
 pg. 
2529
 
59.
Zehner
 
S.
Kotzsch
 
A.
Bister
 
B.
Süssmuth
 
R. D.
Méndez
 
C.
Salas
 
J. A.
van Pée
 
K.-H.
Chem. Biol.
2005
, vol. 
12
 pg. 
445
 
60.
Entsch
 
B.
van Berkel
 
W. J. H.
FASEB J.
1995
, vol. 
9
 pg. 
476
 
61.
Unversucht
 
S.
Hollmann
 
F.
Schmid
 
A.
van Pée
 
K.-H.
Adv. Synth. Catal.
2005
, vol. 
347
 pg. 
1163
 
62.
Bitto
 
E.
Huang
 
Y.
Bingman
 
C. A.
Singh
 
S.
Thorson
 
J. S.
Phillips
 
G. N.
Proteins
2008
, vol. 
70
 pg. 
289
 
63.
van Pée
 
K.-H.
Annu. Rev. Microbiol.
1996
, vol. 
50
 pg. 
375
 
64.
Shepherd
 
S. A.
Menon
 
B. R. K.
Fisk
 
H.
Struck
 
A.-W.
Levy
 
C.
Leys
 
D.
Micklefield
 
J.
ChemBioChem
2016
, vol. 
17
 pg. 
821
 
65.
Ainsley
 
J.
Mulholland
 
A. J.
Black
 
G. W.
Sparagano
 
O.
Christov
 
C. Z.
Karabencheva-Christova
 
T. G.
ACS Omega
2018
, vol. 
3
 pg. 
4847
 
66.
Neumann
 
C. S.
Walsh
 
C. T.
Kay
 
R. R.
Proc. Natl. Acad. Sci. U. S. A.
2010
, vol. 
107
 pg. 
5798
 
67.
Menon
 
B. R. K.
Latham
 
J.
Dunstan
 
M. S.
Brandenburger
 
E.
Klemstein
 
U.
Leys
 
D.
Karthikeyan
 
C.
Greaney
 
M. F.
Shepherd
 
S. A.
Micklefield
 
J.
Org. Biomol. Chem.
2016
, vol. 
14
 pg. 
9354
 
68.
Lingkon
 
K.
Bellizzi
 
J. J.
ChemBioChem
2020
, vol. 
21
 pg. 
1121
 
69.
Flecks
 
S.
Patallo
 
E. P.
Zhu
 
X.
Ernyei
 
A. J.
Seifert
 
G.
Schneider
 
A.
Dong
 
C.
Naismith
 
J. H.
van Pée
 
K.-H.
Angew. Chem., Int. Ed.
2008
, vol. 
47
 pg. 
9533
 
70.
Karabencheva-Christova
 
T. G.
Torras
 
J.
Mulholland
 
A. J.
Lodola
 
A.
Christov
 
C. Z.
Sci. Rep.
2017
, vol. 
7
 pg. 
17395
 
71.
Ashtekar
 
K. D.
Marzijarani
 
N. S.
Jaganathan
 
A.
Holmes
 
D.
Jackson
 
J. E.
Borhan
 
B.
J. Am. Chem. Soc.
2014
, vol. 
136
 pg. 
13355
 
72.
Andorfer
 
M. C.
Park
 
H. J.
Vergara-Coll
 
J.
Lewis
 
J. C.
Chem. Sci.
2016
, vol. 
7
 pg. 
3720
 
73.
Pratihar
 
S.
Roy
 
S.
J. Org. Chem.
2010
, vol. 
75
 pg. 
4957
 
74.
Menon
 
B. R. K.
Brandenburger
 
E.
Sharif
 
H. H.
Klemstein
 
U.
Shepherd
 
S. A.
Greaney
 
M. F.
Micklefield
 
J.
Angew. Chem., Int. Ed.
2017
, vol. 
56
 pg. 
11841
 
75.
Sánchez
 
C.
Butovich
 
I. A.
Braña
 
A. F.
Rohr
 
J.
Méndez
 
C.
Salas
 
J. A.
Chem. Biol.
2002
, vol. 
9
 pg. 
519
 
76.
Lee
 
J.-K.
Zhao
 
H.
J. Bacteriol.
2007
, vol. 
189
 pg. 
8556
 
77.
Ismail
 
M.
Schroeder
 
L.
Frese
 
M.
Kottke
 
T.
Hollmann
 
F.
Paul
 
C. E.
Sewald
 
N.
ACS Catal.
2019
, vol. 
9
 pg. 
1389
 
78.
Schroeder
 
L.
Frese
 
M.
Müller
 
C.
Sewald
 
N.
Kottke
 
T.
ChemCatChem
2018
, vol. 
10
 pg. 
3336
 
79.
Sucharitakul
 
J.
Tinikul
 
R.
Chaiyen
 
P.
Arch. Biochem. Biophys.
2014
, vol. 
555–556
 pg. 
33
 
80.
Agarwal
 
V.
Gamal
 
E. A. A.
Yamanaka
 
K.
Poth
 
D.
Kersten
 
R. D.
Schorn
 
M.
Allen
 
E. E.
Moore
 
B. S.
Nat. Chem. Biol.
2014
, vol. 
10
 pg. 
640
 
81.
Seibold
 
C.
Schnerr
 
H.
Rumpf
 
J.
Kunzendorf
 
A.
Hatscher
 
C.
Wage
 
T.
Ernyei
 
A. J.
Dong
 
C.
Naismith
 
J. H.
van Pée
 
K.-H.
Biocatal. Biotransform.
2006
, vol. 
24
 pg. 
401
 
82.
Zeng
 
J.
Zhan
 
J.
Biotechnol. Lett.
2011
, vol. 
33
 pg. 
1607
 
83.
Smith
 
D. R. M.
Uria
 
A. R.
Helfrich
 
E. J. N.
Milbredt
 
D.
van Pée
 
K.-H.
Piel
 
J.
Goss
 
R. J. M.
ACS Chem. Biol.
2017
, vol. 
12
 pg. 
1281
 
84.
Thapa
 
H. R.
Lail
 
A. J.
Garg
 
N.
Agarwal
 
V.
Methods Enzymol.
2018
, vol. 
604
 pg. 
333
 
85.
Luhavaya
 
H.
Sigrist
 
R.
Chekan
 
J. R.
McKinnie
 
S. M. K.
Moore
 
B. S.
Angew. Chem., Int. Ed.
2019
, vol. 
58
 pg. 
8394
 
86.
Domergue
 
J.
Erdmann
 
D.
Fossey-Jouenne
 
A.
Petit
 
J.-L.
Debard
 
A.
de Berardinis
 
V.
Vergne-Vaxelaire
 
C.
Zaparucha
 
A.
AMB Express
2019
, vol. 
9
 pg. 
175
 
87.
Lee
 
J.
Kim
 
J.
Kim
 
H.
Kim
 
E.-J.
Jeong
 
H.-J.
Choi
 
K.-Y.
Kim
 
B.-G.
ChemBioChem
2020
, vol. 
21
 pg. 
1446
 
88.
Timmins
 
A.
de Visser
 
S. P.
Catalysts
2018
, vol. 
8
 pg. 
314
 
89.
Zhu
 
X.
Robinson
 
D. A.
McEwan
 
A. R.
O'Hagan
 
D.
Naismith
 
J. H.
J. Am. Chem. Soc.
2007
, vol. 
129
 pg. 
14597
 
90.
Blasiak
 
L. C.
Drennan
 
C. L.
Acc. Chem. Res.
2009
, vol. 
42
 pg. 
147
 
91.
El Gamal
 
A.
Agarwal
 
V.
Diethelm
 
S.
Rahman
 
I.
Schorn
 
M. A.
Sneed
 
J. M.
Louie
 
G. V.
Whalen
 
K. E.
Mincer
 
T. J.
Noel
 
J. P.
Paul
 
V. J.
Moore
 
B. S.
Proc. Natl. Acad. Sci. U. S. A.
2016
, vol. 
113
 pg. 
3797
 
92.
Neubauer
 
P. R.
Widmann
 
C.
Wibberg
 
D.
Schröder
 
L.
Frese
 
M.
Kottke
 
T.
Kalinowski
 
J.
Niemann
 
H. H.
Sewald
 
N.
PLoS One
2018
, vol. 
13
 pg. 
e0196797
 
93.
Ismail
 
M.
Frese
 
M.
Patschkowski
 
T.
Ortseifen
 
V.
Niehaus
 
K.
Sewald
 
N.
Adv. Synth. Catal.
2019
, vol. 
361
 pg. 
2475
 
94.
Fisher
 
B. F.
Snodgrass
 
H. M.
Jones
 
K. A.
Andorfer
 
M. C.
Lewis
 
J. C.
ACS Cent. Sci.
2019
, vol. 
5
 pg. 
1844
 
95.
Magarvey
 
N. A.
Beck
 
Z. Q.
Golakoti
 
T.
Ding
 
Y.
Huber
 
U.
Hemscheidt
 
T. K.
Abelson
 
D.
Moore
 
R. E.
Sherman
 
D. H.
ACS Chem. Biol.
2006
, vol. 
1
 pg. 
766
 
96.
Yan
 
Q.
Philmus
 
B.
Chang
 
J. H.
Loper
 
J. E.
eLife
2017
, vol. 
6
 pg. 
1
 
97.
Zeng
 
J.
Lytle
 
A. K.
Gage
 
D.
Johnson
 
S. J.
Zhan
 
J.
Bioorg. Med. Chem. Lett.
2013
, vol. 
23
 pg. 
1001
 
98.
Hammer
 
P. E.
Hill
 
D. S.
Lam
 
S. T.
van Pée
 
K. H.
Ligon
 
J. M.
Appl. Environ. Microbiol.
1997
, vol. 
63
 pg. 
2147
 
99.
Zhao
 
Y.
Yan
 
B.
Yang
 
T.
Jiang
 
J.
Wei
 
H.
Zhu
 
X.
Acta Crystallogr., Sect. F: Struct. Biol. Commun.
2015
, vol. 
71
 pg. 
972
 
100.
Rachid
 
S.
Krug
 
D.
Kunze
 
B.
Kochems
 
I.
Scharfe
 
M.
Zabriskie
 
T. M.
Blöcker
 
H.
Müller
 
R.
Chem. Biol.
2006
, vol. 
13
 pg. 
667
 
101.
van Pee
 
K.-H.
Curr. Org. Chem.
2012
, vol. 
16
 pg. 
2583
 
102.
Büchler
 
J.
Papadopoulou
 
A.
Buller
 
R.
Catalysts
2019
, vol. 
9
 pg. 
1030
 
103.
Fejzagić
 
A. V.
Gebauer
 
J.
Huwa
 
N.
Classen
 
T.
Molecules
2019
, vol. 
24
 pg. 
4008
 
104.
Liu
 
Z.
Ma
 
L.
Zhang
 
L.
Zhang
 
W.
Zhu
 
Y.
Chen
 
Y.
Zhang
 
W.
Zhang
 
C.
Org. Biomol. Chem.
2019
, vol. 
17
 pg. 
1053
 
105.
Widmann
 
C.
Ismail
 
M.
Sewald
 
N.
Niemann
 
H. H.
Acta Crystallogr., Sect. D: Struct. Biol.
2020
, vol. 
76
 pg. 
687
 
106.
James
 
P. N.
Snyder
 
H. R.
Org. Synth.
1959
, vol. 
39
 pg. 
30
 
107.
Heemstra
 
J. R.
Walsh
 
C. T.
J. Am. Chem. Soc.
2008
, vol. 
130
 pg. 
14024
 
108.
Veldmann
 
K. H.
Dachwitz
 
S.
Risse
 
J. M.
Lee
 
J.-H.
Sewald
 
N.
Wendisch
 
V. F.
Front. Bioeng. Biotechnol.
2019
, vol. 
7
 pg. 
219
 
109.
Martinie
 
R. J.
Livada
 
J.
Chang
 
W.-C.
Green
 
M. T.
Krebs
 
C.
Bollinger
 
J. M.
Silakov
 
A.
J. Am. Chem. Soc.
2015
, vol. 
137
 pg. 
6912
 
110.
Frese
 
M.
Guzowska
 
P. H.
Voß
 
H.
Sewald
 
N.
ChemCatChem
2014
, vol. 
5
 pg. 
1270
 
111.
Lang
 
A.
Polnick
 
S.
Nicke
 
T.
William
 
P.
Patallo
 
E. P.
Naismith
 
J. H.
van Pée
 
K.-H.
Angew. Chem., Int. Ed.
2011
, vol. 
50
 pg. 
2951
 
112.
Payne
 
J. T.
Andorfer
 
M. C.
Lewis
 
J. C.
Angew. Chem., Int. Ed.
2013
, vol. 
52
 pg. 
5271
 
113.
Hölzer
 
M.
Burd
 
W.
Reißig
 
H.-U.
van Pée
 
K.-H.
Adv. Synth. Catal.
2001
, vol. 
343
 pg. 
591
 
114.
Goss
 
R. J. M.
Newill
 
P. L. A.
Chem. Commun.
2006
pg. 
4924
 
115.
Shepherd
 
S. A.
Karthikeyan
 
C.
Latham
 
J.
Struck
 
A.-W.
Thompson
 
M. L.
Menon
 
B. R. K.
Styles
 
M. Q.
Levy
 
C.
Leys
 
D.
Micklefield
 
J.
Chem. Sci.
2015
, vol. 
6
 pg. 
3454
 
116.
Runguphan
 
W.
Qu
 
X.
O'Connor
 
S. E.
Nature
2010
, vol. 
468
 pg. 
461
 
117.
Glenn
 
W. S.
Nims
 
E.
O'Connor
 
S. E.
J. Am. Chem. Soc.
2011
, vol. 
133
 pg. 
19346
 
118.
Turner
 
N. J.
Trends Biotechnol.
2003
, vol. 
21
 pg. 
474
 
119.
Reetz
 
M. T.
Bocola
 
M.
Carballeira
 
J. D.
Zha
 
D.
Vogel
 
A.
Angew. Chem., Int. Ed.
2005
, vol. 
44
 pg. 
4192
 
120.
Stemmer
 
W. P.
Nature
1994
, vol. 
370
 pg. 
389
 
121.
Poor
 
C. B.
Andorfer
 
M. C.
Lewis
 
J. C.
ChemBioChem
2014
, vol. 
15
 pg. 
1286
 
122.
Payne
 
J. T.
Poor
 
C. B.
Lewis
 
J. C.
Angew. Chem., Int. Ed.
2015
, vol. 
54
 pg. 
4226
 
123.
Belsare
 
K. D.
Andorfer
 
M. C.
Cardenas
 
F. S.
Chael
 
J. R.
Park
 
H. J.
Lewis
 
J. C.
ACS Synth. Biol.
2017
, vol. 
6
 pg. 
416
 
124.
Hosford
 
J.
Shepherd
 
S. A.
Micklefield
 
J.
Wong
 
L. S.
Chem. – Eur. J.
2014
, vol. 
20
 pg. 
16759
 
125.
Schnepel
 
C.
Kemker
 
I.
Sewald
 
N.
ACS Catal.
2019
, vol. 
9
 pg. 
1149
 
126.
Minges
 
H.
Schnepel
 
C.
Böttcher
 
D.
Weiß
 
M. S.
Sproß
 
J.
Bornscheuer
 
U. T.
Sewald
 
N.
ChemCatChem
2020
, vol. 
12
 pg. 
818
 
127.
Muffler
 
K.
Kuetchou Ngnigha
 
A. R.
Ulber
 
R.
Chem. Ing. Tech.
2010
, vol. 
82
 pg. 
121
 
128.
Frese
 
M.
Sewald
 
N.
Angew. Chem., Int. Ed.
2015
, vol. 
54
 pg. 
298
 
129.
Gröger
 
H.
Hummel
 
W.
Curr. Opin. Chem. Biol.
2014
, vol. 
19
 pg. 
171
 
130.
Schnepel
 
C.
Sewald
 
N.
Chem. – Eur. J.
2017
, vol. 
23
 pg. 
12064
 
131.
Burda
 
E.
Hummel
 
W.
Gröger
 
H.
Angew. Chem., Int. Ed.
2008
, vol. 
47
 pg. 
9551
 
132.
Bornscheuer
 
U. T.
Z. Naturforsch. C
2019
, vol. 
74
 pg. 
61
 
133.
Cantillo
 
D.
Kappe
 
C. O.
React. Chem. Eng.
2017
, vol. 
2
 pg. 
7
 
134.
Latham
 
J.
Brandenburger
 
E.
Shepherd
 
S. A.
Menon
 
B. R. K.
Micklefield
 
J.
Chem. Rev.
2018
, vol. 
118
 pg. 
232
 
135.
Gruß
 
H.
Sewald
 
N.
Chem. – Eur. J.
2020
, vol. 
26
 pg. 
5328
 
136.
García-Pindado
 
J.
Willemse
 
T.
Goss
 
R.
Maes
 
B. U. W.
Giralt
 
E.
Ballet
 
S.
Teixidó
 
M.
Biopolymers
2018
, vol. 
109
 pg. 
e23112
 
137.
Ghabraie
 
E.
Kemker
 
I.
Tonali
 
N.
Ismail
 
M.
Dodero
 
V. I.
Sewald
 
N.
Chem. – Eur. J.
2020
, vol. 
26
 pg. 
12036
 
138.
Kemker
 
I.
Feiner
 
R. C.
Müller
 
K. M.
Sewald
 
N.
ChemBioChem
2020
, vol. 
21
 pg. 
496
 
139.
Kemker
 
I.
Schnepel
 
C.
Schröder
 
D. C.
Marion
 
A.
Sewald
 
N.
J. Med. Chem.
2019
, vol. 
62
 pg. 
7417
 
140.
van Pée
 
K.-H.
ChemBioChem
2011
, vol. 
12
 pg. 
681
 
141.
Sánchez
 
C.
Zhu
 
L.
Braña
 
A. F.
Salas
 
A. P.
Rohr
 
J.
Méndez
 
C.
Salas
 
J. A.
Proc. Natl. Acad. Sci. U. S. A.
2005
, vol. 
102
 pg. 
461
 
142.
Runguphan
 
W.
O'Connor
 
S. E.
Org. Lett.
2013
, vol. 
15
 pg. 
2850
 
143.
Deb Roy
 
A.
Grüschow
 
S.
Cairns
 
N.
Goss
 
R. J. M.
J. Am. Chem. Soc.
2010
, vol. 
132
 pg. 
12243
 
144.
Sharma
 
S. V.
Tong
 
X.
Pubill-Ulldemolins
 
C.
Cartmell
 
C.
Bogosyan
 
E. J. A.
Rackham
 
E. J.
Marelli
 
E.
Hamed
 
R. B.
Goss
 
R. J. M.
Nat. Commun.
2017
, vol. 
8
 pg. 
229
 
145.
Namgung
 
S.
Park
 
H. A.
Kim
 
J.
Lee
 
P.-G.
Kim
 
B.-G.
Yang
 
Y.-H.
Choi
 
K.-Y.
Dyes Pigm.
2019
, vol. 
162
 pg. 
80
 
146.
Fräbel
 
S.
Krischke
 
M.
Staniek
 
A.
Warzecha
 
H.
Biotechnol. J.
2016
, vol. 
11
 pg. 
1586
 
147.
Fräbel
 
S.
Wagner
 
B.
Krischke
 
M.
Schmidts
 
V.
Thiele
 
C. M.
Staniek
 
A.
Warzecha
 
H.
Metab. Eng.
2018
, vol. 
46
 pg. 
20
 
148.
Estévez-Braun
 
A.
González
 
A. G.
Nat. Prod. Rep.
1997
, vol. 
14
 pg. 
465
 
149.
Grover
 
J.
Jachak
 
S. M.
RSC Adv.
2015
, vol. 
5
 pg. 
38892
 
150.
Calcio Gaudino
 
E.
Tagliapietra
 
S.
Martina
 
K.
Palmisano
 
G.
Cravotto
 
G.
RSC Adv.
2016
, vol. 
6
 pg. 
46394
 
151.
Wendisch
 
V. F.
Metab. Eng.
2020
, vol. 
58
 pg. 
17
 
152.
Veldmann
 
K. H.
Minges
 
H.
Sewald
 
N.
Lee
 
J.-H.
Wendisch
 
V. F.
J. Biotechnol.
2019
, vol. 
291
 pg. 
7
 
153.
Corr
 
M. J.
Sharma
 
S. V.
Pubill-Ulldemolins
 
C.
Bown
 
R. T.
Poirot
 
P.
Smith
 
D. R. M.
Cartmell
 
C.
Abou
 
A.
Fayad
 , 
Goss
 
R. J. M.
Chem. Sci.
2017
, vol. 
8
 pg. 
2039
 
154.
Latham
 
J.
Henry
 
J.-M.
Sharif
 
H. H.
Menon
 
B. R. K.
Shepherd
 
S. A.
Greaney
 
M. F.
Micklefield
 
J.
Nat. Commun.
2016
, vol. 
7
 pg. 
11873
 
155.
Frese
 
M.
Schnepel
 
C.
Minges
 
H.
Voß
 
H.
Feiner
 
R.
Sewald
 
N.
ChemCatChem
2016
, vol. 
8
 pg. 
1799
 
156.
Durak
 
L. J.
Payne
 
J. T.
Lewis
 
J. C.
ACS Catal.
2016
, vol. 
6
 pg. 
1451
 
157.
Gruß
 
H.
Belu
 
C.
Bernhard
 
L. M.
Merschel
 
A.
Sewald
 
N.
Chem. – Eur. J.
2019
, vol. 
25
 pg. 
5880
 
158.
Pubill-Ulldemolins
 
C.
Sharma
 
S. V.
Cartmell
 
C.
Zhao
 
J.
Cárdenas
 
P.
Goss
 
R. J. M.
Chem. – Eur. J.
2019
, vol. 
25
 pg. 
10866
 
Close Modal

or Create an Account

Close Modal
Close Modal