Selective C-H Halogenation of Alkenes and Alkynes Using Flavin-Dependent Halogenases

Abstract

Single component flavin-dependent halogenases (FDHs) possess both flavin reductase and FDH activity in a single enzyme. We recently reported that the single component FDH AetF catalyzes site-selective bromination and iodination of a variety of aromatic substrates and enantioselective bromolactonization and iodoetherification of styrenes bearing pendant carboxylic acid or alcohol substituents. Given this inherent reactivity and selectivity, we explored the utility of AetF as catalyst for alkene and alkyne C-H halogenation. We find that AetF catalyzes halogenation of a range of 1,1-disubstituted styrenes, often with high stereoselectivity. Despite the utility of haloalkenes for cross-coupling and other applications, accessing these compounds in a stereoselective manner typically requires functional group interconversion processes, and selective halogenation of 1,1′-disubstituted olefins remains rare. We also establish that AetF and homologues of this enzyme can halogenate terminal alkynes. Mutagenesis studies and deuterium kinetic isotope effects are used to support a mechanistic proposal involving covalent catalysis for halogenation of unactivated alkynes by AetF homologues. These findings expand the scope of FDH catalysis and continue to show the unique utility of single component FDHs for biocatalysis.

Graphical Abstarct

The single component flavin-dependent halogenase AetF halogenates a range of 1,1-disubstituted styrenes, often with high stereoselectivity, and AetF and homologues of this enzyme also halogenate terminal alkynes. These findings expand the scope of FDH catalysis, and mutagenesis studies and deuterium kinetic isotope effects provide insight into the unique utility of single component FDHs for biocatalysis.

Halogen substituents can greatly influence the physical and biological properties of molecules.^[1–3] The need to install halogen substituents at different sites in diverse compounds has driven the development of methods capable of directly halogenating sp³−, sp²−, and sp-hybridized C-H bonds, and generating halogenated sp³ and sp² centers through halofunctionalization of alkenes and alkynes.^[4–8] Given that these reactions typically involve reagents that generate electrophilic or radical halogen intermediates, controlling their selectivity is often challenging.^[5,9] Enzymatic chlorination, bromination, and iodination also proceed by electrophilic or radical mechanisms, but the selectivity of these processes is controlled by the enzyme scaffolds that evolved to harness this reactivity.^[10–12] Enzymatic fluorination proceeds by nucleophilic substitution, so the selectivity of this process is substrate-controlled.^[13] Among the halogenases reported to date that catalyze chlorination, bromination, and iodination, flavin-dependent halogenases (FDHs) have proven particularly attractive for biocatalysis.^[14,15] The earliest characterized FDHs halogenate aromatic moieties in natural products (Figure 1A).^[16] While this reactivity dominates subsequently characterized FDHs, enol halogenation has also been reported,^[17,18] and a proposal that an FDH is responsible for alkynyl C-H halogenation in the natural product jamaicamide was recently confirmed.^[19] Many of these enzymes have since been shown to possess broad substrate scope,^[20,21] high site selectivity toward diverse synthetic substrates,^[22,23] and amenability to protein engineering efforts.^[24–27]

Figure 1.

A) Representative substrate scope of native FDH catalysis. B) FDH-catalyzed halocyclization of substituted alkenes. C) FDH-catalyzed alkene C-H halogenation.

FDHs catalyze these reactions by generating an electrophilic halogen species, likely hypohalous acid, within their active sites, where it is activated toward electrophilic attack through H-bonding to an active site residue, typically lysine.^[28–30] Given this mechanism, we anticipated that FDHs might be capable of catalyzing a far broader range of halogenation reactions than is observed in nature. We extended the aromatic substrate scope noted above^[20–23] to include substrates that undergo enantioselective desymmetrization^[31] and atroposelective halogenation.^[32] Moreover, we showed that these enzymes catalyze enantioselective halocyclization of olefins,^[33] demonstrating that they can interact with π systems distinct from those that they evolved to act on (Figure 1B).^[34–36] This reaction involves attack of the olefin by both the electrophilic halogen species generated by the FDH and a pendant nucleophile on the substrate.^[9] We hypothesized that in the absence of the substrate nucleophile, halosubstitution involving water as a nucleophile or perhaps olefin C-H halogenation could be possible using FDHs, in analogy to the corresponding halocyclization and native C-H halogenation reactions (Figure 1C). Despite the utility of haloalkenes for cross-coupling and other applications, accessing these compounds in a stereoselective manner typically requires functional group interconversion processes, and selective halogenation of 1,1’-disubstituted olefins remains rare.^[37–39] Direct synthesis of brominated styrenes through C-H functionalization has been reported, but 1,1’-disubstituted substrates were either not discussed,^[40] or lead to mixtures of cis- and trans-brominated products.^[41,42] Stereoselective C-H halogenation of alkenes would therefore represent a particularly useful target for FDH catalysis.

On the other hand, a significant complication to FDH catalysis involves the fact that most of these enzymes are two-component systems that require a flavin reductase to supply reduced flavin. The halogenase catalyzes the reaction of reduced flavin, O₂, and X⁻ to generate the HOX required for substrate halogenation.^[15] In this regard, single component FDHs like AetF,^[43,44] which possesses both flavin reductase and FDH activity in a single enzyme, are advantageous since one less enzyme is required. We recently reported that the single component FDH AetF catalyzes site-selective bromination and iodination of a variety of aromatic substrates, including relatively electron deficient compounds.^[36] We also showed that this enzyme catalyzes enantioselective bromolactonization and iodoetherification of styrenes bearing pendant carboxylic acid or alcohol substituents. Given this inherent reactivity and selectivity, which required extensive evolution to achieve using variants of the tryptophan halogenase RebH,^[14] we explored the utility of AetF as catalyst for alkene C-H halogenation. We find that AetF catalyzes halogenation of a range of 1,1-disubstituted styrenes, often with remarkably high stereoselectivity. We also establish that AetF and homologues of this enzyme can halogenate terminal alkynes, showing that this reactivity is not limited to FDHs in which it evolved.^[19] These findings expand the scope of FDH catalysis and continue to show the unique utility of single component FDHs for biocatalysis.

In preliminary experiments aimed at expanding the scope of AetF-catalyzed bromocyclization, we found that some substrates underwent monobromination without apparent cyclization. Characterization of these compounds indicated that they were undergoing C-H bromination, often providing a single product isomer in which the bromine substituent was trans to the p-methoxybenzene substituent on the substrates (Figure 2). Because halocyclization reactions catalyzed by RebH variants or AetF have required electron donating substituents to achieve high conversions,^[34–36] the activity of AetF toward a variety of 1,1-disubstituted p-methoxystyrene derivatives was examined.

Figure 2.

A) Substrate scope for AetF-catalyzed halogenation of alkenes. B) AetF-catalyzed halogenation of p-methoxyphenylacetylene. Conversion is defined as % area for the product peaks relative to the % area for the starting material + product peaks in GC/MS chromatograms of crude reactions conducted on a 75 μL scale. Isolated yields were obtained from reactions conducted on a 30–40 mL scale, and the selectivity of these reactions was determined by GC/MS or NMR spectroscopy. Differences between conversions and isolated yields represent the formation of uncharacterized by-products in some cases (see GC/MS chromatograms in the Supporting Information) and product loss during protein removal and the multiple extractions required for isolation.

Analysis of these reactions revealed that AetF halogenates a range of styrenes, and no trace of the corresponding halosubstitution products was observed. Substrates with relatively small substituents provided either low yields (1a) or selectivity (1b), but longer aliphatic side chains provided good-to-high conversions and remarkable selectivity even when they contained ketone or ester substituents (1c-e). Substituted benzene substituents were similarly well tolerated (1f-j). The p-methoxy substituent common to these substrates could be replaced with other electron rich aromatics (1k-n), albeit with compromised selectivity in the case of furan and thiophene derivatives 1k and 1l, but at least one electron rich aromatic group on the olefin was required for activity. Steady state kinetic analysis of reactions involving substrates 1b and 1c revealed that these reactions proceed with a k_cat of 12.4 and 13.1 min⁻¹, which compares favorably with the rates of aromatic halogenation and halocyclization involving single component FDHs (typically 0.4–3 min⁻¹).^[14,34] The K_M values for these reactions are 96.2 and 123.9 μM, which is high relative to reported values for aromatic halogenation catalyzed by conventional FDHs^[45,46] but low compared to halocyclization catalyzed by the same.^[34] Notably, some substrates that underwent selective bromination were also iodinated with high selectivity (1o and 1p). Finally, we examined the activity of AetF toward p-methoxyphenylacetylene, and this compound underwent both bromination and iodination in good-to-high conversion (Figure 2B).

With the goal of further expanding the scope of FDH-catalyzed alkene and alkyne halogenation, we next evaluated the activity of several AetF homologues. Searching the NCBI and Uniprot databases using the AetF sequence as a query sequence returned approximately 134 sequences, many of which were annotated as flavin-containing monooxygenases. To identify FDHs from these sequences, we targeted those containing a lysine residue at the same position of the sequence alignment as the catalytically essential lysine in AetF. Inspection of the alignment indicated that this residue is contained in the motif [TLIV]KXW[NPS], which was present in 34 unique sequences (Table S1, Figure S1). This set of enzymes contains only three previously reported enzymes, PhmJ, VatD, and JamD, which are involved in the biosynthesis of natural products containing halogenated alkene and alkyne moieties. A recently reported phylogenetic analysis of AetF starting from a much larger number of sequences arrived at a similar set of enzymes, suggesting that relatively few single component FDHs are present in sequence databases. The structures of these sequences predicted using AlphaFold were similar to that of AetF,^[47] and importantly, the putative active site lysine residue in these structures overlaid well with the catalytically essential^[36] K258 in AetF (Figure S2). The greatest variation in the structures was observed in the C-termini, and analysis of these sequences using TMHMM^[48] suggested that this region often contains predicted membrane-spanning regions (see the Supporting Information). Four representative sequences lacking such regions, including JamD, which natively halogenates the terminal alkyne moiety during the biosynthesis of jamaicamycin,^[19] were synthesized and expressed in E. coli. All four genes expressed well in E. coli and were purified by Ni-NTA affinity chromatography.

The activity of the purified enzymes was evaluated on a small panel of alkenes and alkynes aimed at establishing the influence of substrate structure on halogenation (Table 1). Consistent with our findings for halocyclization,^[34–36] an electron rich aromatic group is required for alkene C-H halogenation. Neither styrene nor 1-decene underwent halogenation, and the modest conversion of p-methoxy styrene compared to the larger substrates examined in Figure 2 suggests that additional structural features assist with substrate binding for productive halogenation. Broader reactivity was generally observed toward alkynes. For example, four of the five enzymes evaluated halogenated p-methoxyphenylacetylene, and three of the enzymes halogenated phenylacetylene, indicating that less electronic activation is required for alkyne halogenation. Indeed, even 1-decyne was halogenated by JamD and MCE9613948 (hereafter MCE9).

Table 1.

Halogenation of alkenes and alkynes catalyzed by AetF and homologues.

	Conversion (%)[a]
Substrate	HalA[b]	AetF	JamD	MBL4[c]	MCE2[c]	MCE9[c]
	241	1 ± 0.4	9 ± 2	0	0	17 ± 1
	135	0	0	0	0	0
	117[d]	0	0	0	0	0
	149	98 ± 0.2	96 ± 0.3	10 ± 2	0	95 ± 0.5
	135	79 ± 4	91 ± 0.6	0	0	85 ± 2
	107[d]	0	63 ± 1	0	0	47 ± 3

^[a]Conversions were determined from relative integration values of starting material and product in 70 μL analytical scale reactions conducted in triplicate.

^[b]Halenium affinity (HalA) values were calculated as previously reported.^[20]

^[c]Enzyme names are shortened forms of the full NCBI accession codes MBL4604125, MCE2448131, and MCE9613948.

^[d]Propene and propyne were used for HalA calculations in place of 1-decene and 1-decyne to avoid complications from different conformers.

A recent study on JamD^[19] also reported its activity on phenylacetylene and 1-decyne, though only trace activity was observed for AetF on the native alkyne-containing JamD substrate, jamaicamide B. Because JamD did not halogenate phenol or tryptophan, both of which are halogenated by AetF, or 1-dodecene or 5-decyne, the authors concluded that a unique property of JamD renders it chemoselective for terminal alkyne halogenation. Our finding that JamD catalyzes styrene halogenation more efficiently than AetF and that AetF catalyzes halogenation of phenylacetylene complicates this picture. Of the small panel of AetF homologues investigated, only JamD and MCE9 can halogenate 1-decyne, but substrate activation seems to play a significant role in the apparent chemoselectivity of these enzymes.

Substrate activation toward electrophilic attack by halogen reagents can be quantified using calculated halenium affinity (HalA) values,^[49] in this case involving the enthalpy associated with electrophilic attack of Br⁺ on alkynes or alkenes to form the corresponding cationic adducts. As a reference, the HalA value for bromination of the 7-position of indole, one of the sites that AetF halogenates on tryptophan, is 130 kcal/mol, so compounds with HalA values higher than this should be sufficiently activated to undergo halogenation by this enzyme. The HalA values associated with terminal C-H halogenation of styrene, phenylacetylene, and the p-methoxy derivatives of these compounds are all above this value (Table 1). The fact that styrene and phenylacetylene have identical HalA values but only the latter is halogenated by AetF, JamD, and MCE9 suggests that the linear geometry of alkynes may facilitate their halogenation by these enzymes. Interestingly, the HalA values for both propene and propyne, surrogates for 1-decene and 1-decyne, respectively, are well below 130 kcal/mol. While it is therefore not surprising that 1-decene is not halogenated, 1-decyne has the lowest HalA (107 kcal/mol) of any of the compounds evaluated; this value is also lower than any aromatic substrates halogenated by AetF or other FDHs.^[20,36] The fact that 1-decyne is nonetheless halogenated by JamD and MCE9 suggests that these enzymes can override the relatively unfavorable electronics of this substrate.

Mutagenesis was used to provide insight into the unique reactivity of JamD and MCE9 toward unactivated alkynes. Models of these enzymes were generated using AlphaFold^[50] and overlaid with the crystal structure of the AetF•Trp•FAD complex^[47] (Figure 3A). This analysis revealed several residues in the putative active sites of these enzymes that are conserved, including K258, T373, and E200 using AetF residue numbers. Mutating each of these residues in AetF or JamD to alanine led to variants with no activity on tryptophan, phenylacetylene, or 1-decyne, showing that they are essential for halogenase catalysis (Figure 3B). Notably, P202 and P429 in AetF are replaced by H202 and Y428 in JamD and MCE9. We found that introducing H202F into JamD eliminated its ability to halogenate phenylacetylene or 1-decyne, and that a lesser reduction was observed for Y428F, highlighting the importance of these residues for alkyne halogenation. The H202A variants of both JamD and MCE9 also gave reduced yields for halogenation of phenylacetylene and 1-decyne. While we had hoped that introducing P202H and P429Y into AetF might lead to variants with activity on 1-decyne, low activity was observed on all substrates, including tryptophan, so other features in JamD and MCE9 must be required to take advantage of the reactivity conferred by these residues in those scaffolds.

Figure 3.

A) Overlay of models of JamD (yellow) and MCE9 (purple) generated using AlphaFold^[50] with the crystal structure of the AetF•Trp•FAD complex (PDB ID 8CJE).^[47] B) Effects of E200A, H202A, and Y428F on JamD-catalyzed bromination of phenylacetylene and 1-decyne. Conversions determined from relative integration values of starting material and product in GC/MS or LC/MS chromatograms (see the Supporting Information). Alkyne reactions with measurable activity were conducted in triplicate; all other conversions are single measurements. See data file for complete data.

The large negative effects of E200A and H202F on the ability of JamD to halogenate 1-decyne led us to speculate that these residues might play unique roles in alkyne halogenation related to differences in intermediates involved in electrophilic alkyne, alkene, and arene halogenation. We presume that halogenation of alkenes catalyzed by AetF, JamD, and MCE9 proceeds through a mechanism analogous to that accepted for aromatic halogenation (Figure 4A).^[14,15] The benzylic carbocation intermediate in this mechanism would explain the need for electron donating groups on styrene substrates (Figure 2A), and further electrostatic stabilization of this intermediate by E200 would explain the importance of this residue. We proposed a similar step-wise mechanism for halolactonization of carboxylate-substituted styrenes with electron donating substituents catalyzed by RebH variants (Figure 4B).^[34] We hypothesized that the lower stereoselectivity observed for substrates lacking electron donating groups could result from a switch to a concerted mechanism that had been proposed^[51] for the same substrates using small molecule halogenating reagents.

Figure 4.

Proposed mechanisms for FDH-catalyzed reactions: A) arene and alkene C-H halogenation, B) halolactonization, and C) alkyne C-H halogenation (potential role of Y428 as a proton relay between E200 and H202 omitted for clarity).

In the context of alkyne halogenation catalyzed by AetF homologues, E200 could play a similar role as the pendant carboxylate in halocyclization reactions, taking the electrostatic interaction proposed in other FDHs^[28,29] to the extreme of covalent catalysis^[52] (Figure 4C). This intermediate would eliminate the need for formation of a vinyl cation during alkyne halogenation, which would be particularly high energy for simple alkynes such as 1-decyne (HalA ~107 kcal/mol), but perhaps feasible for activated alkynes like 4-methoxyphenylacetylene (HalA = 149 kcal/mol). Notably, JamD and MCE9, the only enzymes in this study that halogenate 1-decyne, possess H202, which could serve as the base required to deprotonate the covalent adduct, regenerate E200, and form the observed halogenated alkyne. H202 could also acidify the terminal vinyl C-H of the vinyl ester intermediate by H-bonding to the ester carbonyl, and, given its proximity to E200 and H202, Y428 could serve as a proton relay in this process to explain its effect on catalysis (not shown).

We previously established that halogenation of tryptophan by the FDH RebH proceeds with a negligible deuterium kinetic isotope effect (KIE),^[24] consistent with rate limiting electrophilic attack (Figure 4A). We suspected that the deprotonation step of the proposed covalent mechanism for 1-decyne halogenation could be at least partially rate limiting given the high pK_a of the intermediate vinyl C-H bond (Figure 4C). JamD- and AetF-catalyzed bromination reactions of H- and D-substituted p-methoxyphenylacetyelene and 1-decyne (JamD only) were therefore conducted in parallel, and deuterium KIEs were measured as ratios of initial rates (k_H/k_D). Bromination of p-methoxyphenylacetyelene proceeds with a k_H/k_D of 1.1 and 1.2 for JamD and AetF respectively, while the JamD-catalyzed reaction of 1-decyne exhibited a k_H/k_D of 1.9. The low k_H/k_D values obtained from the parallel p-methoxyphenylacetyelene reactions are consistent with an electrophilic mechanism analogous to those for electron rich arenes and alkenes (Figure 4A). On the other hand, the high value obtained for 1-decyne indicates that C-H cleavage occurs in the rate limiting step for halogenation of this substrate. A comparable k_H/k_D of 2.3 was reported for Au(I)-catalyzed bromination of phenylacetylene, which was proposed to involve rate-limiting deprotonation.^[53] While other roles for E200 and H202 can be envisioned in the context of canonical FDH catalysis (e.g. H202 orienting E200 for electrostatic stabilization), a model involving covalent catalysis nicely rationalizes the substrate preferences and deuterium KIEs for the enzymes evaluated in this study (Figure 4C). AetF acts through a conventional electrophilic FDH mechanism involving electrostatic stabilization by E200 that is sufficient for halogenation of activated alkenes and alkynes. JamD and MCE9 can act in a similar fashion, but H202 in these enzymes enables covalent catalysis for halogenation of less reactive alkynes, whose orthogonal π* orbitals offer greater flexibility for attack by E200 than the single π* orbital of alkene substrates.

In summary, this study shows that the single component FDH AetF can catalyze halogenation of 1,1-disubstituted styrenes with high stereoselectivity. While no examples of FDH-catalyzed alkene C-H halogenation have been reported, natural products with halogenated alkenes exist, so it will be interesting to see if this capability is exploited in nature or if these products are formed through more convoluted pathways. AetF also catalyzes halogenation of terminal alkynes, showing that this reactivity is not limited to FDHs in which it evolved.^[19] Homologues of AetF, including JamD and MCE9, provide higher yields for both alkene and alkyne halogenation and are uniquely capable of halogenating unactivated alkynes like 1-decyne. The calculated HalA value for 1-decyne is far lower than those for substrates that AetF can halogenate, suggesting that JamD and MCE9 may act by a unique mechanism that avoids high energy cationic intermediates. A substrate-dependent change in mechanism is also consistent with the observation of a significant deuterium KIE in the JamD-catalyzed bromination of 1-decyne but not JamD- or AetF-catalyzed bromination of p-methoxyphenylacetylene. JamD and MCE9 possess an active site histidine residue (H202) that is not present in AetF, and we propose that this residue enables covalent catalysis by an active site glutamate residue (E200). A multiple sequence alignment of the 34 AetF homologues examined in this study shows that the ExxH motif containing these residues is conserved in most other AetF homologues, suggesting that many of these enzymes may also possess activity on alkynes. Together, these findings expand the scope of FDH catalysis and continue to show the unique utility of single component FDHs for biocatalysis. Further studies will be required to validate our covalent catalysis hypothesis and to establish whether the ability of JamD and MCE9 to act on unactivated alkynes can be extended to other less activated FDH substrates.