Controlling Non-Native Cobalamin Reactivity and Catalysis in the Transcription Factor CarH

Abstract

Vitamin B₁₂ derivatives catalyze a wide range of organic transformations, but B₁₂-dependent enzymes are underutilized in biocatalysis relative to other metalloenzymes. In this study, we engineered a variant of the transcription factor CarH, called CarH*, that catalyzes styrene C–H alkylation with improved yields (2–6.5-fold) and selectivity relative to cobalamin. While the native function of CarH involves transcription regulation via adenosylcobalamin (AdoCbl) Co(III)–carbon bond cleavage and β-hydride elimination to generate 4′,5′-didehydroadenosine, CarH*-catalyzed styrene alkylation proceeds via non-native oxidative addition and olefin addition coupled with a native-like β-hydride elimination. Mechanistic studies on this reaction echo findings from earlier studies on AdoCbl homolysis to suggest that CarH* selectivity results from its ability to impart a cage effect on radical intermediates. These findings lay the groundwork for the development of B₁₂-dependent enzymes as catalysts for non-native transformations.

Graphical Abstract

INTRODUCTION

Derivatives of vitamin B₁₂ comprise a family of cobalt corrin complexes (cobalamins) with different axial ligands that confer unique reactivity to these complexes in their roles as enzyme cofactors (Figure 1A).¹ B₁₂-dependent enzymes include adenosylcobalamin (AdoCbl)-dependent isomerases,² methyl-cobalamin (MeCbl)-dependent methyltransferases,³ and dehalogenases.⁴ Within these classes, variation of the axial ligand distal to the substrate binding pocket is also observed. The dimethylbenzimidazole (dmb) ligand of cobalamins can be bound to the cobalt center (dmb-on), as in class II B₁₂-dependent isomerases, or unbound (dmb-off), as in class I B₁₂-dependent isomerases.² In the latter case, a distal histidine residue from the protein scaffold can bind the cobalt center in a dmb-off/His-on fashion.⁵ These structural variations and the unique interactions between cobalamins and their respective enzyme scaffolds enable these naturally occurring organometallic complexes to engage substrates via both radical and polar chemistry.⁶

Figure 1.

(A) Structures of representative cobalamins. (B) Comparison of the intermolecular C–H abstraction observed for AdoCbl-dependent isomerases² and a proposed radical-based β-hydride elimination mechanism for CarH (Ad = adenine).¹⁶ (C) Potential CarH-catalyzed styrene C–H functionalization involving β-hydride elimination.

When removed from their natural protein scaffolds, cobalamins can catalyze an even broader range of C–C bond-forming reactions than those observed in nature.⁷ Moreover, Co(III)–alkyl complexes analogous to AdoCbl and MeCbl have been proposed as intermediates in reductive dehalogenation,⁸ alkyl halide and alkene coupling reactions,⁹ and a number of transformations involving alkenyl C–H functionalization.^10–12 These reactions proceed via a highly nucleophilic Co(I) intermediate¹³ that undergoes oxidative addition with a variety of organic electrophiles to generate Co(III)–alkyl complexes.¹⁴ These complexes can react via radical coupling or olefin addition pathways to generate different organic products. Unlike enzymatic transformations, however, product mixtures are typically obtained due to homolysis of the Co(III)–alkyl bond and subsequent free radical chemistry, and very few of these reactions can be achieved with catalyst-controlled selectivity.⁷ We were therefore drawn to the potential non-native reactivity of cobalamin-containing proteins and the control that protein scaffolds could exert over the reactivity of cobalamin–alkyl intermediates.

Homolysis of the Co(III)–carbon bond of the AdoCbl cofactor in B₁₂-dependent isomerases (Figure 1B) forms the basis of a variety of C-H functionalization reactions that proceed via C–H abstraction by the 5′-deoxyadenosyl radical (Ado^•).² We reasoned that similar active-site localization of a Co(III)–alkyl complex generated from the reaction of the Co(I) form [cob(I)alamin] of a suitable cobalamin-containing protein with simple organic electrophiles could be used to confer selectivity to a variety of cobalt-catalyzed transformations. The light-dependent transcription factor CarH¹⁵ piqued our interest as a candidate enzyme due to its proposed mechanism involving β-hydride elimination of an AdoCbl intermediate (Figure 1B).¹⁶ If an analogous intermediate could be accessed via initial alkylation of Co(I) with a simple alkyl electrophile lacking β-hydrogen substituents,¹⁴ reaction of the resulting Co(III)–alkyl with a reagent capable of generating an intermediate with β-hydride substituents could enable olefin formation¹⁷ and catalyst turnover.

In considering reactions that could test this hypothesis, we were drawn to a study by Gryko involving β-alkylation of styrenes using alkyl diazoacetates catalyzed by the hexamethyl ester of CNCbl.¹⁰ This reaction was proposed to proceed via radical addition of a Co(III)–carboxymethyl species to styrene, followed by β-hydride elimination (Figure 1C), the latter mirroring a proposed radical pathway for native CarH reactivity¹⁶ (a polar mechanism was also proposed).^16,18 An undesired alkane byproduct also forms under the reaction conditions (~2:1) and was proposed to result from free-radical side reactions. Herein, we show that an engineered variant of CarH corrects this issue and increases styrene alkylation yields relative to cobalamin by stabilizing Co(III)–alkyl species generated from different organic electrophiles to enable selective reaction of these species with styrenes. This non-native reactivity suggests that cobalamin-containing proteins could serve as catalysts for a wide range of Co-catalyzed transformations.

RESULTS

CarH is a B₁₂-dependent transcription factor that exploits a remarkable exaptation of AdoCbl Co(III)–carbon bond cleavage^18,19 to enable light-activated gene expression,¹⁵ rather than the more common radical rearrangement processes catalyzed by AdoCbl-dependent enzymes (Figure 1B).² In its AdoCbl-bound form, CarH forms a tetramer that binds DNA to repress transcription of carotenoid biosynthesis genes that protect organisms from light damage.²⁰ Upon exposure to light, the Co(III)–carbon bond of AdoCbl most likely undergoes homolytic cleavage¹⁹ (although heterolytic cleavage has also been proposed¹⁸), leading to loss of 4′,5′-anhydroadenosine¹⁶ and formation of bis-His cobalamin,¹⁸ which causes a conformational change in the enzyme that favors monomer formation, DNA dissociation, and transcription activation. The structure of CarH comprises a C-terminal domain that binds AdoCbl in a dmb-off/His-on fashion (H177), a four-helix cap above the AdoCbl cofactor, and a N-terminal DNA-binding domain.²⁰

We reasoned that apo-CarH could be reconstituted with hydroxocobalamin¹⁸ which would be protonated to give aquocobalamin at pH 7.5²¹ so that the vacant Ado-binding site could be used for substrate binding and that dmb-off/His-on cobalamin binding could allow for modulation of reactivity via distal axial ligand variation. We first examined CarH from Thermus thermophilus (TtCarH), but its low solubility complicated characterization. This issue was addressed by fusing a N-terminal His₆-MBP tag to the enzyme, which we also presumed would disrupt potential tetramer formation. A histidine residue (H132) near the Ado binding pocket was mutated to glycine to preclude its binding to the Co center of cobalamin¹⁸ and potential reactions with added electrophiles. Apo MBP-TtCarH H132G (hereafter CarH*) was expressed in Escherichia coli (20 mg/L), purified by Ni-NTA chromatography, and reconstituted with a threefold excess of hydroxocobalamin. The enzyme was purified by size-exclusion chromatography to provide the desired cobalamin-bound form of CarH* as a monomer in good yield (84%).

The feasibility of reducing CarH* to its cob(I)alamin form and alkylating this species with different electrophiles was next evaluated. Several chemical reductants can be used to generate cob(I)alamin,¹⁴ but most of these are heterogeneous or absorb UV–vis light at wavelengths that overlap with diagnostic absorptions for the different oxidation states of cobalamin. We therefore turned to UV–vis absorption spectroelectrochemistry to generate the cob(I)alamin form of CarH* and monitor its reaction with ethyl diazoacetate (EDA, Figure 2A), ethyl chloroacetate (ECA), and p-methoxystyrene. At a potential of −1.5 V, reduction of CarH* occurred, as evidenced by the appearance of a peak at 388 nm, consistent with cob(I)alamin (ii), and concomitant loss of a peak at 357 nm, associated with aquocob(III)alamin (i) (Figure 2B).²² Addition of either EDA or ECA to ii led to rapid (<15 s) formation of cob(II)alamin (iii), as indicated by a peak at 475 nm (Figure 2C,D). A gradual disappearance of cob(II)alamin and appearance of a second cob(III)alamin species, assigned as v (presumably via the intermediacy of iv),²² then occurred before reduction of the sample back to cob(I)alamin. The absorbances associated with v were red-shifted by roughly 18 nm and broadened relative to those associated with i, consistent with previously reported spectra of cobalamin complexes with increasingly less electronegative axial ligands²³ (from oxygen to carbon in the current case). It should also be noted that cob(II)alamin could also be formed from reduction of protons in water by ii,²⁴ and v could form via two-electron oxidative addition of ii, so the relevance of the observed cob(II)alamin peaks to the formation of v cannot be definitively established from these data. No significant buildup of peaks consistent with cob(III)alamin was observed upon addition of styrene (Figure 2E). Similar spectra were observed for the reaction of ECA with cob(I)alamin at −1.2 V, but the rates of the reactions were much faster, presumably due to improved access of cobalamin to the electrode in the absence of the CarH* protein components (Supporting Information Figure S1C). Interestingly, a peak corresponding to a Co(III) species following addition of EDA to cob(I)alamin was not red-shifted as it was for CarH*, suggesting that intermediate v does not build up for cobalamin in solution as it does within CarH* (Supporting Information Figure S1B).

Figure 2.

(A) Scheme for electrochemical reduction and reaction of CarH* with EDA. (B) UV–vis absorbance spectra for the cob(III)-, cob(II)-, and cob(I)alamin forms of CarH* (22 μM) generated following electrolysis for the specified times. Arrows indicate direction of absorbance change for different species. (C–E) UV–vis absorbance spectra for CarH* (22 μM) following addition of 22 μM EDA (C), ECA (D), or styrene (E).

Encouraged by the observation of a cob(III)alamin species consistent with ethyl carboxymethyl intermediate v envisioned in Figure 1, we next sought to establish whether subsequent reaction of this species would occur in the presence of styrene. Toward this end, p-methoxystyrene and EDA were incubated with 4.4 mol % of CarH* and excess Ti(III) citrate, which have been used to access Co(I) in a variety of Co-catalyzed transformations.⁷ Alkene 1a formed in 57% yield, constituting 13 turnovers under optimized reaction conditions (Table 1, entry 1). A small amount of alkane 2a also formed (1a:2a = 19:1). A significantly lower yield of 1a (22%) and lower selectivity for 1a over 2a was obtained using hydroxocobalamin (1a:2a = 4:1, Table 1, entry 2). No 1a or 2a was observed for two other B₁₂-dependent enzymes, methionine synthase and human methylmalonyl-CoA mutase (MetH and hMUT, respectively, Table 1, entries 3 and 4),⁶ construction of which is described in the Supporting Information. When reduced to their cob(II)alamin forms using Na₂S₂O₄ in the presence of EDA and p-methoxystyrene, these enzymes gave trace amounts of cyclopropanation product 3a,²⁵ but CarH* showed no detectable cyclopropanation activity (Table S2). Together, these results highlight the unique capability of CarH* to catalyze styrene C–H functionalization.

Table 1.

Styrene Alkylation Catalyzed by Hydroxocobalamin and Different B₁₂-Dependent Enzymes

entry	catalyst	1a (%)a	2a (%)	3a (%)
1	HOCbl	22(±4)	5(±1)	0
2	CarH*	57(±5)	3(±1)	0
3	MetH	0	0	0
4	hMUT	0	0	0

^aYields determined by GC/MS analysis of crude reaction mixtures relative to 3,3′,5,5′-tetramethylbenzidine internal standard. Yields and standard deviations for reactions conducted in triplicate are provided. Reactions were conducted in glass vials covered with aluminum foil to exclude light.

Several additional styrenes and diazo substrates were next examined to evaluate the substrate specificity of CarH*-catalyzed styrenyl C–H functionalization (Table 2). CarH* variants containing active-site mutations were also evaluated (Figures S25 and S26), but only the W131F variant provided substantially higher conversions in some cases (Table 2, entries 3 and 6, Figure S26). A modestly good conversion was obtained for electron-rich styrenes, although excellent conversion was observed for an acetamido-substituted styrene (Table 2, entry 4), and the enzyme exhibited a preference for relatively small styrenes and diazoacetates and diazoacetamides. No activity was observed for donor–acceptor or acceptor–acceptor carbene precursors, again highlighting the requirement for relatively small substrates, although phenyl diazoacetate and diethyl diazoacetamide were both tolerated. These results indicate that the protein scaffold plays a role in regulating substrate access to the cobalamin cofactor in CarH* and that the enzyme does not simply serve as a radical initiator as cobalamin does in the absence of the protein scaffold.

Table 2.

Substrate Scope of CarH*-Catalyzed Styrene Alkylation

				yield (%)a
entry	1	R1	R2	HoCbl	CarH*
1	a	OMe	OEt	22(±4)	57(±5)
2	b	OEt	OEt	11	22
3	c b	NMe2	OEt	20	49
4	d	NHAc	OEt	15(±2)	97(±4)
5	e	Me	OEt	10	25
6	f b	H	OEt	8	29
7	g	OMe	OPh	13	32
8	h	OMe	NEt2	14	68

^aYields determined by GC/MS analysis of crude reaction mixtures relative to 3,3′,5,5′,-tetramethylbenzidine internal standard.

^bReactions were performed using CarH* W131F. Yields and standard deviations for selected reactions conducted in triplicate are provided. Reactions were conducted in glass vials covered with aluminum foil to exclude light.

To gain further insight into the relative reactivity of cob(I)alamin within CarH* versus in solution, the rates for reactions of p-methoxystyrene (1) and β,β-d₂-p-methoxystyrene (β,β-d₂-1) with EDA in the presence of added CarH* or hydroxocobalamin were next measured in H₂O (Table 3, entries 1 and 3). Due to poor substrate solubility, initial rates (v_i) could only be acquired within a narrow concentration range that prohibited Michaelis–Menten kinetic analysis. Comparing v_i using a 0.5 mM substrate, however, showed that CarH* provides ~5-fold rate acceleration relative to hydroxocobalamin, consistent with the higher yields obtained from the enzyme under optimized conditions (Table 2). Normal deuterium kinetic isotope effects (KIEs) were observed for styrene alkylation catalyzed by CarH* and hydroxocobalamin (e.g., k_H/k_D = 1.85 and 2.15, respectively, Table 3, entry 1). This finding indicates that C–H cleavage occurs in the rate-limiting step of the reaction, as would be expected if β-hydride elimination was at least partially rate-limiting.

Table 3.

Deuterium Kinetic Isotope Effects and Solvent Isotope Effects for Styrene Alkylation Reactions Catalyzed by Hydroxocobalamin and CarH*

				initial rate (μM·min−1)			kH/kD		kH2O/k50% D2O
entry	styrene	diazo	solvent	CarH*	HOCbl	kCarH*/kHOCbl	CarH*	HOCbl	CarH*	HOCbl
1	1	EDA	H2O	8.87 (0.43)	1.66 (0.08)	5.34 (0.37)	1.85 (0.15)	2.15 (0.15)	0.86 (0.07)	1.67 (0.11)
2	1	EDA	50% D2O/H2O	10.27 (0.63)	0.99 (0.05)	10.37 (0.82)	1.63 (0.11)	1.86 (0.19)
3	β,β-d2-1	EDA	H2O	4.77 (0.28)	0.77 (0.04)	6.19 (0.50)			0.75 (0.07)	1.45 (0.15)
4	β,β-d2-1	EDA	50% D2O/H2O	6.29 (0.25)	0.53 (0.05)	11.86 (1.86)

The rates for the analogous reactions of 1 and β,β-d₂-1 in 50% D₂O/H₂O were also examined to probe the susceptibility of styrene alkylation to solvent isotope effects (SIEs).²⁶ A normal SIE is observed for hydroxocobalamin (k_H2O/k_50%D2O >1.4), while an inverse SIE is observed for CarH* (k_H2O/k_50%D2O < 0.9) (Table 3, entries 1/2 and 3/4). To identify sites in intermediates involved in styrene alkylation that undergo H/D exchange during the reaction and that might therefore play a role in the observed SIEs, gas chromatography/mass spectrometry (GC/MS) was used to analyze products from reactions conducted in 50% D₂O/H₂O. While incorporation of a single deuterium into intermediate v was expected based on the mechanism shown in Figure 2A, a mixture of d₁- and d₂-products was obtained. Only slightly higher deuterium incorporation was observed for the CarH*-catalyzed reaction (78% conversion to d₁- and d₂-products vs 66% for hydroxocobalamin), suggesting that this process was not responsible for the observed SIE. No H/D exchange was observed in the reaction of d₈-styrene in H₂O, indicating that H/D exchange occurs at the α-carbon of intermediate v, presumably due to acidification by its electrophilic cob(III)-alamin center.

Further differences in cob(I)alamin catalysis within CarH* versus in solution were revealed by monitoring methoxystyrene consumption in reactions of 1a conducted with and without EDA at low conversions (Figure S21). When the CarH*-catalyzed reaction had proceeded to give a 22% yield of 1a, 24% of the total methoxystyrene had been consumed. In the absence of EDA, however, 43% of styrene was consumed, and polymethoxystyrene oligomers were observed by GPC (Supporting Information Figure S2).⁹ During the same period of time, reactions catalyzed by hydroxocobalamin gave only 3% yield of 1a and 36% of methoxystyrene consumption in the presence of EDA and 52% consumption of methoxystyrene in the absence of EDA.

Finally, we investigated the effects of TEMPO added on CarH* and hydroxocobalamin-catalyzed styrene alkylation.²⁷ As expected for a reaction involving free-radical intermediates, the yields for the hydroxocobalamin-catalyzed reactions were reduced from 18% to 13% to 7% in the presence of 1 and 5 equiv of TEMPO, respectively (Figure S18). On the other hand, CarH*-catalyzed reactions were not affected by up to 5 equiv of TEMPO. No conversion was observed for either catalyst in the presence of 20 equiv of TEMPO, and the TEMPO adduct of the ethyl carboxymethyl radical was observed by GC/MS analysis of these reaction mixtures (Figure S18).²⁸ Both d₁- and d₂-TEMPO adducts were obtained for analogous reactions conducted in 50% D₂O/H₂O, consistent with the isotopic exchange at the α-carbon noted above.

DISCUSSION

Non-native enzyme catalysis brings the exquisite selectivity of enzymes to bear on chemical reactions that did not happen to emerge in nature. This capability can be revealed by exposing an enzyme to substrates, reagents, or conditions that are distinct from those associated with its native activity but that enable mechanistically feasible pathways involving its active-site residues and cofactors.²⁹ Particularly illustrative examples of this capability involve the use of heme enzymes as selective catalysts for carbene and nitrene insertion reactions.^30,31 While cytochromes P450 are best known for catalyzing oxene insertions, non-native carbene and nitrene insertion reactions are catalyzed by several heme enzymes with dramatically improved rates and unprecedented selectivity relative to the analogous reactions catalyzed by heme or other iron–porphyrin complexes.

The state of catalysis using vitamin B₁₂ derivatives is similar to that employing heme and related complexes. Cobalamins catalyze a broad range of reactions, including reductive olefin coupling, cross-coupling, dehalogenation, and cyclopropanation,⁷ but achieving catalyst control over the selectivity of these reactions is challenging since the substituents of the cobalamin corrin ring that could interact with substrates are distal to its cobalt center (Figure 1A). B₁₂-dependent proteins solve this problem by embedding cobalamin cofactors in a chiral protein pocket that regulates substrate access to and reactivity with the cobalt center.¹ As mentioned above, studies on the mechanism of CarH, which involves Co(III)–carbon bond homolysis of its AdoCbl cofactor^18,20 to release 4′,5′-anhydroadenosine (Figure 1B),¹⁶ suggested that it might be uniquely suited for non-native catalysis involving an analogous β-hydride elimination (Figure 1C). This hypothesis was validated by successful catalysis of styrene alkylation using EDA in the presence of CarH*.

Several findings from this effort suggest that CarH* and related enzymes could serve as useful starting points for further exploration of non-native cobalamin catalysis. First, the desired activity was observed for CarH but not for the extensively studied B₁₂-dependent enzymes hMUT or MetH.⁶ While the poor solubility of CarH initially limited the utility of this finding, this issue was addressed via fusion to MBP to generate CarH*, providing a means to explore the activity of CarH* and perhaps homologous proteins. In addition, observation of the cob(I)alamin form of CarH* using UV–vis spectroelectrochemistry and conversion of this species to others consistent with cob(III)alamin-alkyl complexes upon addition of either EDA or ECA (Figure 2C,D) indicate that this protein can be used to access the non-native Co(I) oxidation state and chemical transformations associated with it.⁷

Most notably, however, the mechanistic studies outlined above indicate that the control exhibited by CarH over the reactivity of the AdoCbl cofactor¹⁶ extends to non-native reactions. The improved styrene alkylation yields and chemoselectivity of cob(I)alamin catalysis within CarH* versus in solution highlight this potential. Improved chemoselectivity is illustrated by both improved production of the desired alkylated styrenes over the corresponding alkane side products under the optimized reaction conditions and, in the presence of added TEMPO, production of alkene over ethyl carboxymethyl–TEMPO adducts. Interestingly, Finke reported that ethylene glycol-induced caging of a [Ado^{• •}Cbi] pair uniquely enables β-hydride elimination to generate 4′,5′-didehydroadenosine during homolysis of AdoCbi derivatives (Cbi = cobinamide). Ado^• could, however, be intercepted under strong cage conditions using high concentrations of TEMPO to yield an Ado–TEMPO adduct.²⁷ Drennan suggested that CarH could exert a similar caging effect on Ado^• to facilitate β-hydride elimination.¹⁶ The TEMPO trapping studies in the current study suggest that the same phenomenon results in caging of the ethyl carboxymethyl radical that would form upon homolysis of ethyl carboxymethyl–Cbl intermediate v (Figure 2A) to enable radical addition to styrene and subsequent β-hydride elimination in preference to cage escape and radical polymerization observed for hydroxocobalamin. Despite this cage effect, high TEMPO concentrations could be used to intercept the α-carboxymethyl radical to generate the corresponding TEMPO adduct of this compound in analogy to the AdoCbl system studied by Finke.²⁷ While Sension and Penner-Hahn have provided evidence that native CarH activity involves a concerted β-hydride elimination to generate 4’,5’-anhydroadenosine from a triplet metal-to-ligand charge transfer state accessed following photoexcitation,¹⁹ our reactions were conducted in the absence of light, so we presume that this pathway is not operative.

Docking simulations were conducted to show how the CarH* active site might accommodate ethyl carboxymethyl–Cbl intermediate v and p-methoxystyrene. The dark-state, AdoCbl form of CarH was selected for modeling since its Ado ligand is a better mimic of the non-native ethyl carboxymethyl ligand of v than the His132 ligand that binds to Cbl in the light state.²⁰ Docking a geometry-optimized structure of ethyl carboxymethyl–Cbl (v, Figure 2A) into apo-CarH* (Figure 3A) using RosettaLigand shows how the active site might accommodate this alkyl fragment within a narrow binding cleft (Figure 3B). Docking p-methoxystyrene into this structure shows how sterically undemanding styrenes could access this cleft in an orientation that places the styrene β-position approximately 3 Å from the α-carbon of v to enable the observed chemoselectivity. The predicted binding poses (Supporting Information Figure S3) are consistent with the observation that α-substituted diazoacetates are not tolerated but phenyldiazoacetate and N,N-dimethyldiazoacetamide are tolerated since both project their substituents out of the active site. Likewise, styrene binding within the active site explains why bulky styrene substituents are not tolerated and why the even bulkier TEMPO reacts less readily with v to generate the ethyl carboxymethyl adducts that form readily in the presence of hydroxocobalmin. This structure also illustrates how mutation of W131 would be expected to play a role in modulating CarH* reactivity due to its position directly above the putative styrene-binding pocket.

Figure 3.

(A) Geometry-optimized structure of ethyl carboxymethyl–Cbl complex v (Figure 2A) docked into a structure of CarH (PDB ID:5C8A).²⁰ (B) Closeup of active site showing narrow, hydrophobic substrate-binding channel between two helices of the 4-helix cap of CarH (van der Waals surface rendered).

In addition to these reactivity differences, the observed isotope effects also point to the involvement of the CarH* scaffold in modulating cobalamin reactivity. Similar deuterium KIEs are observed for reactions conducted with hydroxocobalamin and CarH*, suggesting that the rate-limiting step of both reactions is the same. The different SIEs and the fact that an inverse SIE is observed for the CarH*-catalyzed reaction in particular suggest that more subtle changes in non-rate-limiting steps may be responsible for the observed selectivity of CarH*.³² Control reactions conducted in 9% glycerol/H₂O provide similar yields at low substrate conversion to those conducted in pure water (Figure S24), indicating that the increased viscosity of D₂O relative to H₂O is not responsible for the inverse SIE of CarH*.³² Inverse SIEs for enzyme-catalyzed reactions are most often attributed to displacement of metal-bound waters or H/D exchange involving active-site cysteine residues.³² CarH does not possess the latter, and cob(I)alamin would not be expected to bind water.²¹ Moreover, similar levels of deuterium incorporation into products generated from reactions catalyzed by hydroxocobalamin and CarH* indicate that deuterium exchange involving the substrate is not responsible for the SIE. Inverse SIEs have also been attributed to differences in ligand binding affinity for different conformational states of an enzyme.³³ Given that the four-helix cap that sits on top of the cobalamin cofactor in CarH (Figure 3) is known to undergo large conformational changes,²⁰ we tentatively suggest that differences in the affinity in those conformation states for cob(III)alamin–alkyl intermediates could explain the observed inverse SIEs and account for the improved selectivity of CarH* relative to cobalamin in solution. Quantitative kinetic analysis and proton inventory experiments with a suitable model system will be required to better clarify these possibilities.^32,33

Putting these findings together with the original mechanistic hypothesis of Gryko,¹⁰ we propose that CarH*-catalyzed styrene alkylation proceeds via the mechanism shown in Figure 4. Catalysis is initiated by reduction of aquocobalamin to cob(I)alamin (ii), which undergoes oxidative addition with EDA to generate intermediate v. Addition of styrene to this complex, presumably via homolysis and recombination of organic radical and cob(II)alamin intermediates,¹⁷ would give a second cob(III)alamin–alkyl complex. β-Hydride elimination from this species would form the observed alkene products and hydridocobalamin, which would undergo deprotonation to reform the cob(I)alamin form of CarH*. In the absence of CarH*, homolytic dissociation of v to generate cob(II)alamin and the ethyl carboxymethyl radical can occur. While this species could add back to cob(II)alamin to re-enter the catalytic cycle as suggested by Gryko,¹⁰ polymerization dominates under the reaction conditions used in the current study.

Figure 4.

Mechanism of styrene alkylation (adapted from ref 10) accounting for differences in selectivity between cob(I)alamin in CarH* and in solution.

CONCLUSIONS

B₁₂-dependent enzymes are underutilized in biocatalysis relative to other metalloenzymes. Recent examples exploring the substrate scope of native B₁₂-dependent enzyme catalysis^4,34 and the broad range of reactions catalyzed by cobalamin cofactors⁷ suggest that suitable systems could find broader application in biocatalysis. This study establishes that a B₁₂ protein can catalyze a transformation that diverges substantially from reactivity associated with its native function. While CarH is a transcription regulator that acts via AdoCbl homolysis and β-hydride elimination to generate 4’,5’-didehydroadenosine,^15,16,18,20 CarH* catalyzes styrene C–H alkylation via non-native oxidative addition and olefin addition processes coupled with a native-like β-hydride elimination. Mechanistic studies on this reaction echo findings from earlier studies of AdoCbl homolysis in solution under strong cage conditions and in CarH itself to suggest that CarH* can enable non-native radical chemistry with improved selectivity relative to the cobalamin.^16,27 These findings suggest that engineered CarH* variants and perhaps related B₁₂-dependent enzymes hold great promise as catalysts for a wide range of transformations.