Changing Fates of the Substrate Radicals Generated in the Active Sites of the B₁₂-Dependent Radical SAM Enzymes OxsB and AlsB

Abstract

OxsB is a B₁₂-dependent radical SAM enzyme that catalyzes the oxidative ring contraction of 2′-deoxyadenosine 5′-phosphate to the dehydrogenated, oxetane containing precursor of oxetanocin A phosphate. AlsB is a homologue of OxsB that participates in a similar reaction during the biosynthesis of albucidin. Herein, OxsB and AlsB are shown to also catalyze radical mediated, stereoselective C2′-methylation of 2′-deoxyadenosine monophosphate. This reaction proceeds with inversion of configuration such that the resulting product also possesses a C2′ hydrogen atom available for abstraction. However, in contrast to methylation, subsequent rounds of catalysis result in C–C dehydrogenation of the newly added methyl group to yield a 2′-methylidene followed by radical addition of a 5′-deoxyadenosyl moiety to produce a heterodimer. These observations expand the scope of reactions catalyzed by B₁₂-dependent radical SAM enzymes and emphasize the susceptibility of radical intermediates to bifurcation along different reaction pathways even within the highly organized active site of an enzyme.

Graphical Abstract

INTRODUCTION

Radical S-adenosylmethionine (SAM) enzymes have been shown to participate in a wide variety of primary and secondary metabolic pathways.^1–6 A common mechanistic feature shared among these enzymes is the reduction of SAM (1) via single electron transfer from a reduced active site [Fe₄S₄]¹⁺ cluster. This induces the cleavage of SAM to l-methionine (Met) and a putative 5′-deoxyadenosyl radical (2) or its equivalent that initiates a radical mediated transformation of the nominal substrate (Figure 1).^7,8 Radical SAM enzymes have thus been recognized for the extremely broad range of chemical reactions they are capable of catalyzing given the relative ease that organic radicals are able to undergo bond cleavage, bond formation, and covalent bond rearrangements compared to their Lewis acid/base counterparts.

Figure 1.

Biosynthesis of oxetanocin A (5) and albucidin (7) each involves a B₁₂-dependent radical SAM enzyme, OxsB and AlsB, respectively, catalyzing oxidative ring contraction of 2′-dAMP (3) concomitant with the reduction of SAM to 5′-deoxyadenosine (5′-dA) and l-methionine. OxsB was also previously shown to catalyze methylation of 3 to afford a methylated derivative of 2′-dAMP.

A subgroup of radical SAM enzymes also binds cobalamin, which in most cases studied to date mediates net transfer of a methyl group from a second molecule of SAM to the substrate radical or anion via an intermediary methylcobalamin donor.^5,9–11 However, this class of enzymes does not appear to be strictly limited to catalyzing methylation reactions,¹² as proposed during the biosynthesis of ladderanes,¹³ hopanoids,¹⁴ bacteriochlorophyll,¹⁵ and herbicidins (Figure S1).¹⁶ Moreover, an increasing number of cobalamin-dependent radical SAM enzymes have been shown experimentally to catalyze reactions other than methylation in vitro. Examples include OxsB, which binds a single [Fe₄S₄] cluster in addition to cobalamin and operates together with its partner enzyme OxsA to catalyze the formation of intermediate (4) from 2′-deoxyadenosine monophosphate (2′-dAMP, 3) during the biosynthesis of oxetanocin A (5).¹⁷ Likewise, AlsB and AlsA mediate a similar oxidative cyclization as well as elimination of a one carbon fragment from 3 during the biosynthesis of albucidin (7, Figure 1).¹⁸ More recently, ThnL was found to catalyze C–S bond formation during the biosynthesis of carbapenems (Figure S1).¹⁹ Thus, the cobalamin-dependent radical SAM enzymes themselves may be just as mechanistically diverse as their parent group.

OxsB and AlsB themselves may also exhibit mechanistic flexibility. While recognized for catalyzing oxidative ring contractions of biosynthetic relevance, evidence has also been provided suggesting that these enzymes can support SAM-dependent methyltransferase activity as well. In particular, OxsB was previously reported to catalyze the methylation of 3 and its higher phosphates even though the precise regiochemistry of the methylation reaction was unknown.²⁰ This observation is consistent with a prior crystallographic investigation of OxsB where SAM was found to occupy two different binding configurations in crystallo.¹⁷ In the first configuration, the SAM sulfonium is located within 2.9 Å of the [Fe₄S₄] cluster and is thus poised for reductive cleavage. In the second configuration, however, the methylsulfonium is shifted 4.0 Å away from the cluster, placing it within 6.0 Å of the cobalt center where it could support methylation of the cobalamin. Such dual binding modes of SAM have been identified both structurally and spectroscopically in other B₁₂-dependent SAM methylases,^21,22 consistent with SAM serving as both a radical initiator and a methyl donor among many of these enzymes.

In an effort to define the ways in which the radical intermediates of OxsB can partition along different reaction pathways, the enzyme was characterized using structural analysis of its reaction products and its activity assessed with several substrate analogues and isotopologues. Follow-up experiments were also performed with AlsB. In these studies, OxsB was found to catalyze a C2′ methylation of 2′-dAMP where the net substitution of an H atom with a methyl group proceeds via inversion of configuration at the methylation site. Moreover, OxsB was also found to be responsible for a series of consecutive reactions beginning with C2′ methylation. In contrast to several other B₁₂-dependent radical SAM enzymes,^23–29 however, these reactions do not involve repeated methylation at either the same site or the newly introduced methyl substituent resulting in ethyl, iso-propyl, tert-butyl, or sec-butyl groups (Figure S2). Instead, OxsB catalyzes C–C dehydrogenation of the newly installed methyl group followed by radical addition of 5′-deoxyadenosine to the resulting methylidine moiety. Similar to OxsB, AlsB is also shown to catalyze an analogous sequence of reactions albeit under slightly different conditions. These observations expand the reaction scope of B₁₂-dependent radical SAM enzymes and lay the groundwork for distinguishing the factors that govern partitioning along different reaction pathways upon radical initiation in this class of enzymes.

RESULTS AND DISCUSSION

OxsB methylation activity.

OxsB catalyzes the oxidative ring contraction of 2′-dAMP (3) to 4 during the biosynthesis of oxetanocin A (5);¹⁷ however, it has also been shown to catalyze the methylation of 2′-dAMP as an apparent side reaction.²⁰ In contrast to ring contraction, methylation does not require the presence of OxsA.²⁰ Consistent with these previous observations, liquid chromatography–mass spectrometry (LCMS) analysis confirmed the formation of methylated 2′-dAMP (calcd. for C₁₁H₁₆N₅O₆P [M + H]⁺ m/z 346.0911) upon anaerobic incubation of a 30 μL solution containing 15 μM OxsB (reconstituted with iron and sulfide only, see the Supporting Information) with 0.5 mM SAM (1), 0.5 mM NADPH, 1 mM MgCl₂, 0.25 mM methyl viologen (MV), 1 mM dithiothreitol (DTT), 0.1 mM hydroxycobalamin (HOCbl), and 0.3 mM 2′-dAMP (3) in 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonate (HEPES) buffer at pH 7.5 for 15 h (Figure 2A, trace a). While controls indicated that MgCl₂ is not required for OxsB methylation activity, it was still included to remain consistent with previously published assay conditions where OxsB was assayed together with OxsA.³⁰

Figure 2.

Analysis of OxsB catalyzed methylation reaction. (A) Screening of required components for the production of methylated 2′-dAMP. Extracted ion chromatography (EIC) profiles extracted at m/z 346.0911, corresponding to the protonated adduct of methylated 2′-dAMP. (B) HPLC analysis of the OxsB methylation assay following treatment with CIP. The peak annotated as 2′-dA is the dephosphorylated substrate 3. (C) Mass profiles of species corresponding to peaks b, c, and d generated upon the use of SAM (black) or [Met-²H₈]-SAM (blue) with (i) 2′-dAMP and (ii) [2′−²H₂]-2′-dAMP. (D) Structures of species assigned to the products a, b, c, and d (see text).

The formation of methylation product remained discernible when either NADPH or DTT was excluded from the assay (Figure 2A, traces b and d); however, omission of both led to no observable methylation product (Figure 2A, trace i), suggesting that either can serve as the reductant to support catalysis. As expected, no methylation was observed whenever OxsB, SAM, MV, HO-Cbl, or 2′-dAMP was left out of the assay (Figure 2A, traces c and e–h). However, if OxsB was coexpressed with a cobalamin uptake system³¹ or OxsB reconstituted with HO-Cbl as well as iron and sulfide was used, additional cobalamin became unnecessary for the activity assay (Figure S3). These observations are consistent with the hypothesis that methylation requires both the radical SAM activity of OxsB as well as the cobalamin cofactor as is the case with ring contraction.¹⁷ MV is likely required to mediate single electron reduction of the [Fe₄S₄] cluster and possibly cobalamin as well by a two-electron reductant such as DTT or NADPH. The DTT/MV reducing system was used in the following experiments as the chromophore in NADPH produces additional chromatographic peaks that complicate the analysis. Unless specified otherwise, all subsequent assays employed OxsB reconstituted only with iron and sulfide before desalting by gel filtration to remove any residual reagents including DTT.

Identification of the OxsB Methylation Product.

In order to characterize the putative methylation product, a time course assay of OxsB activity with SAM, DTT, MV, MgCl₂, and HO-Cbl was conducted with 2′-dAMP as the substrate. The reaction mixture was aliquoted at 0, 2, and 4 h, worked up with calf intestinal alkaline phosphatase (CIP) and analyzed by high-performance liquid chromatography (HPLC) with UV absorbance detection at 260 nm. Five product peaks (a–e) were observed to form in a time-dependent manner following enzyme addition (Figure 2B). Peak e was only observed at trace levels precluding further analysis. Peak a was identified as 5′-deoxyadenosine (8), which is a byproduct formed upon reductive cleavage of SAM, based on coelution with the chemical standard. LCMS analysis (positive ion mode) yielded high-resolution m/z signals of 264.1076, 266.1229, and 513.1938 for peaks b, c, and d, respectively (Figure 2C). The MS of product c was suggestive of a protonated adduct of methylated 2′-dAMP after dephosphorylation (calcd. C₁₁H₁₅N₅O₃ [M + H]⁺ m/z 266.1248), whereas the mass of product b differed from that of c by 2 Da indicative of a dehydrogenated congener (calcd. C₁₁H₁₃N₅O₃ [M + H]⁺ m/z 264.1091). All three products (i.e., b, c, and d) were only observed following the CIP workup suggesting that each carries a phosphate group and are all likely derived from 2′-dAMP (Figure S4).

In order to determine the structures of the products giving rise to peaks b, c, and d, preliminary reactions were run with isotopologues of both 2′-dAMP and SAM (Figure 2C). When the reactions were run with SAM prepared from perdeuterated methionine (i.e., [Met-²H₈]-SAM), the mass of product c increased by 3 Da consistent with methyltransfer from SAM. If [2′−²H₂]-2′-dAMP was used as the substrate with natural abundance SAM, product c was observed with a 1 Da mass increase, implying loss of one C2′ hydrogen such that c may correspond to a dephosphorylated C2′-methylated derivative of 2′-dAMP (9, with undefined stereochemistry at C2′). In contrast, product b did not appear to retain either hydrogen atom from C2′ of the substrate as the use of [2′−²H₂]-2′-dAMP yielded b at natural abundance. Additionally, b increased in mass by only 2 Da when [Met-²H₈]-SAM was used with a natural abundance substrate (Figure 2C). Consequently, b was hypothesized to be the 2′-methylidene adduct of 2′-dAMP formed upon dehydrogenation of c (Figure 2D). Consistent with this hypothesis, b was found to coelute with a synthetic standard of 2′-methylidene-2′-deoxyadenosine (10, Figure S4). Product d also showed no change in mass when produced from [2′−²H₂]-2′-dAMP; however, the use of [Met-²H₈]-SAM rendered d with only a 1 Da increase in mass (Figure 2C). Products b and d differ in mass by 249 Da, suggesting the formation of d via the addition of a deoxyadenosyl moiety (251 Da) to b. While this moiety could originate in 2′-dAMP, this seems unlikely, because neither [2′R-²H]-2′-dAMP nor [2′S-²H]-2′-dAMP led to an increase in the mass of d and [3′−²H]-2′-dAMP only increased the mass of d by 1 Da (Figure S7).

On the other hand, replacing SAM with [Met-²H₈]-SAM or SAM derived from ATP with a labeled [3′, 4′, 5′, 5′−²H₄]-ribose (rib-[²H₄]-SAM) led to 1 and 4 Da increases in the mass of d, respectively (Figure S8). Therefore, product d appears to be an adduct formed between methylated 2′-dAMP and the adenosyl moiety of SAM. The phosphorylated form of d (denoted d-P) also exhibited the same changes in mass upon the use of the aforementioned isotopologues of SAM (Figure S8). Moreover, d-P can be isolated from the OxsB assay mixture by omitting workup with CIP. Due to the low production, only ¹H NMR spectra could be collected for d-P and its dephosphorylated form d. While most of the proton signals overlapped in the proton spectrum of d-P, they were clearly resolved in that of d. Nevertheless, a COSY cross peak between 2.52 and 6.19 ppm for d-P and 2.58 and 6.21 ppm for d was observed, consistent with a nonoxygenated carbon attached to an sp²-hybridized carbon (Figure S9). Based on these observations, the structures of d-P and d were tentatively assigned as 11-P and 11, respectively (Figure 2D).

OxsB Tolerates Substituents at C3′.

Since methylation appeared to involve regiospecific H atom abstraction from C2′, analogues of 2′-dAMP with different substituents at C3′ were investigated as alternative substrates. When OxsB was incubated for 15 h with SAM, DTT, MV, MgCl₂, and HO-Cbl under the above conditions with 0.2 mM substrate analogues possessing either a 3′-O-acetyl (12) or a 3′-Omethyl (13) substituent, LCMS analysis revealed enzymedependent formation of the corresponding methylated products, dehydrogenation products, and heterodimers. In contrast, the use of substrate analogues either having a 3′-F (14), 3′-epi-OH (15), or 3′-NH₂ (16) substituent (Figure 3B) only led to observable formation of the respective methylation products and dehydrogenation products.

Figure 3.

OxsB catalyzed sequential conversion of 3 to 11-P via methylation (3 → 18-P), dehydrogenation (18-P → 10-P), and adduct formation (10-P → 11-P). (A) Proposed reaction sequence of OxsB catalysis with 3 and 12 along with the chemical derivatization that allows for the conversion from 17-P to 18-P, the structures of which were assigned based on ¹H NMR and the NOE correlation between 2′-Me and H-8 in the dephosphorylated counterpart of the former. (B) Structures of other substrate analogues 14–16. (C) Time course HPLC analysis of the OxsB catalyzed reaction with (i) methylated product (18-P) or (ii) methylidene product (10-P) as the substrate followed by dephosphorylation.

Assay of the aforementioned substrate analogues with [Met-²H₈]-SAM resulted in 3 and 2 Da mass increases, respectively, in the corresponding peaks for the methylation and dehydrogenation products observed by LCMS (Figures S10 and S11). Formation of the putative heterodimer, however, was not observed with any of the substrate analogues 14–16. While this could be due to the enzyme not recognizing these analogues for the putative addition reaction, an alternative explanation is that the reaction involves the coupling of SAM with either the methyl or dehydrogenated product, which had not accumulated to a sufficient level to observe addition under the assay conditions. In any case, the observations imply that OxsB-catalyzed methylation and dehydrogenation tolerate structural variation at C3′ of the substrate.

Stereochemistry of OxsB Catalyzed Methylation.

The methylation product of 12 could be isolated in quantities sufficient for NMR characterization. Therefore, a large-scale incubation of OxsB with 12 was performed followed by HPLC separation. The collected product was treated with CIP to remove the 5′-phosphate, purified, and characterized by NMR. Based on NOE correlations observed between the 2′-Me and H-8 protons, the dephosphorylated methylation product of 12 was assigned as 17, in which C2′ is methylated with the S-configuration (Figure 3A). A similar stereoselectivity was expected for the other OxsB substrates including 2′-dAMP (3). Indeed, compound c was found to coelute with the hydrolysis product of 17 (i.e., 18) generated following treatment with NaOH to remove the 3′-O-acetyl group (Figure 3A and Figure S12). Moreover, c also coeluted with a standard of 18 prepared via hydrogenation of 10 using Pd/C under an H₂ atmosphere (18 was the only reduced product obtained). Therefore, the dephosphorylated methylation product of 2′-dAMP (i.e., peak c in Figure 2B) was assigned as 18 (i.e., 9 with S-stereochemistry at C2′), such that OxsB catalyzes methylation of both 2′-dAMP (3) and 3′-O-acetyl-2′-dAMP (12) with the same regio and stereochemistry.

OxsB Catalyzes Sequential Reactions.

It was hypothesized that the three observed products (i.e., 18-P, 10-P, and 11-P) are produced sequentially beginning with C2′-methylation of 2′-dAMP (3 → 18-P, Figure 3A). OxsB then catalyzes dehydrogenation of the C2′-CH₃ methyl to yield the methylidene 10-P rather than a second round of methylation as observed in other radical-mediated methyltransferases.^23–29 Although examples of radical SAM catalyzed dehydrogenation of alcohols, thiols, and amines are well-known,^32–38 biological C-CH₃ dehydrogenase activity has not been previously reported for a radical SAM enzyme. Nevertheless, NosL from nosiheptide biosynthesis catalyzes C–C desaturation of a substrate structural analogue,³⁹ and DarE from the darobactin maturation pathway has been identified to catalyze off-pathway dehydrogenation of the C_α–C_β bond of a tryptophan residue in its peptide substrate.^40,41 Once formed, 10-P could then serve as an acceptor for OxsB catalyzed addition of the 5′-deoxyadenosyl radical (2). This chemistry is effectively equivalent to that of MqnE⁴² and QueE⁴³ requiring ejection of one electron from the reduced radical adduct such that the overall addition (i.e., 10-P + SAM → 11-P + Met) is redox neutral.⁴⁴ As such, this chemistry differs from several other examples of radical SAM catalyzed C–C bond formation which are reductive in nature (i.e., alkene substrate + SAM + 2e⁻ → product + Met).⁴⁵

As an initial test of these hypotheses, OxsB was incubated with 18-P, 10-P, and 11-P individually. Compound 18-P could be prepared from isolated 17-P following large scale reaction of OxsB with 12 and deacetylation with 0.2 M NaOH at room temperature for 1 h (Figure 3A). OxsB was then incubated with 18-P in the presence of SAM, MgCl₂, MV, DTT, and HO-Cbl in 25 mM HEPES buffer at pH 7.5. Reaction aliquots (15 μL) were taken at 0, 2, 4.5, and 21 h, treated with CIP and deproteinized by acetonitrile precipitation before the supernatant was analyzed by HPLC. Under these conditions, 10 was predominantly observed at 2 h with the gradual appearance of 11 at later time points (Figure 3C). The accumulation of 5′-deoxyadenosine (8) was also observed in these reactions. These results are consistent with the hypothesis that OxsB catalyzes C–C dehydrogenation of 18-P to 10-P with SAM serving as the oxidant.

In order to determine whether 10-P is subsequently modified with 5′-deoxyadenosine to yield 11-P, the former was prepared by incubating synthetic 10 with ATP, MgCl₂, and adenosine kinase purified from Streptomyces miharaensis in 50 mM Tris·HCl at pH 8. When 10-P was assayed with OxsB and worked up with CIP, only 11 was observed to form (Figure 3B). Moreover, no 5′-deoxyadenosine formation was observed, which is consistent with 11 being the heterodimer of 10 and the 5′-deoxyadenosyl radical equivalent (2) resulting from reductive cleavage of SAM. When OxsB was assayed with 11-P, there was neither significant consumption of 11-P nor new peak production over 4.5 h (Figure S13), which implies that 11-P is the end product of this serial transformation. Therefore, OxsB catalyzes three sequential reactions beginning with C2′-methylation of 3 to yield 18-P followed by C–C dehydrogenation (18-P → 10-P) before C–C coupling with SAM to terminate the reaction sequence at the heterodimeric adduct 11-P.

Methylation is Mediated by MeCbl with Inversion of Configuration.

In order to investigate the sites of H atom abstraction during each step in the OxsB catalyzed reaction sequence, deuterium incorporation into 5′-deoxyadenosine (8) from the stereospecifically deuterated substrate isotopologues was evaluated. When either [2′R-²H]-2′-dAMP or [2′S-²H]-2′-dAMP was incubated with OxsB for 15 h as described above, the resulting 5′-deoxyadenosine showed 25% and 24% deuteration, respectively (Figure S14), suggesting both positions undergo H atom abstraction in competition with uncoupled reduction of SAM. Indeed, a parallel experiment using [2′−²H₂]-2′-dAMP as the substrate led to a significant increase in the observed deuteration of 5′-deoxyadenosine (41%, Figure S14); however, dideuteration of 5′-deoxyadenosine was not observed consistent with two separate H atom abstraction events. Moreover, deuterium loss in the methylation product was only observed with [2′R-²H]-2′-dAMP (Figure S5). OxsB catalyzed methylation thus proceeds with abstraction of the 2′-pro-R hydrogen atom followed by si-face methylation resulting in overall inversion of configuration (i.e., Walden inversion)⁴⁶ at C2′ (Figure 4A). Consequently, the C2′ H atom that was originally in the pro-S position of 3 migrates to the opposite face of the ribose ring in 18-P.

Figure 4.

(A) Proposed mechanism for OxsB catalyzed methylation of 2′-dAMP (3), showing the sources of methyl groups for the first and subsequent turnovers. (B) Mass profiles of 18 generated by OxsB reconstituted with [C²H₃]-Cbl at 2.5 and 6 h. Consistent with the regioselective H atom abstraction during methylation of 3, when [2′R-²H]-12 was assayed with OxsB for 15 h, 24% deuteration of 5′-deoxyadenosine was obtained with deuterium loss in the methylation product, and no such deuterium transfer was observed with [2′S-²H]-12 (Figure S16).

While the isotope labeling experiments indicated that the transferred methyl group originates in SAM, the immediate methyl donor may actually be methylcobalamin (19). In order to test this hypothesis, cobalamin-free OxsB was reconstituted with ferrous ammonium sulfate, sodium sulfide, and [C²H₃]-Cbl (D₃-19) such that additional cobalamin would not be required for OxsB catalyzed methylation as described above. SAM facilitates the regeneration of Me-Cbl during turnover;²⁰ therefore, reactions were conducted with a high concentration of OxsB (40 μM) over a relatively short time (2.5 h) in an effort to minimize multiple turnovers. Under these conditions, 79% of the methylation product (18) exhibited a [M + H]⁺ signal at m/z 269.1 after 2.5 h, which is 3 Da greater than that of unlabeled 18. Following a 6 h incubation, the fraction of labeled product decreased to 57% (Figure 4B). This result is consistent with Me-Cbl being the immediate methyl donor.

Consequently, methyltransfer from Me-Cbl is proposed to be radical mediated resulting in a cob(II)alamin (20) intermediary state with the substrate radical 22 as the methyl acceptor. Under the in vitro assay conditions, one electron reduction of 20 followed by methyl transfer from a second equivalent of SAM would release S-adenosylhomocysteine (SAH, 21) and regenerate Me-Cbl (19). Subsequent or concomitant reduction of the radical SAM [Fe₄S₄] cluster would then prime OxsB for another round of methylation (Figure 4A).

Dehydrogenation Involves H Atom Abstraction from C2′.

The transfer of both C2′ hydrogen atoms to two separate molecules of 5′-deoxyadenosine (see above) also implied that C–C dehydrogenation of 18-P to 10-P involves H atom abstraction from C2′ rather than the newly introduced methyl group. As a further test of this hypothesis, monodeuterated 18-P was isolated from the OxsB reaction with [2′S-²H]-2′-dAMP (see the Supporting Information), and a 50 μM solution was incubated with 20 μM OxsB for 1 h in an effort to minimize the uncoupled reduction of SAM and thus the accumulation of significant levels of unlabeled 5′-deoxyadenosine. Under these conditions, the resulting 5′-deoxyadenosine (8) showed 83% deuteration and formation of natural abundance 10-P, confirming that H atom abstraction takes place at C2′ during dehydrogenation of 18-P to 10-P (Figure S15). This experiment supported the intermediacy of radical 25 (see Figure 6) during the dehydrogenation reaction. Moreover, this result explains why both the pro-R and pro-S hydrogens of the substrate 3 appear separately in 5′-deoxyadenosine during prolonged reaction times. While only the pro-R hydrogen is transferred during methylation, the pro-S hydrogen in 3 is retained in the methylated product 18-P and only later transferred during the subsequent dehydrogenation reaction.

Figure 6.

(A) HPLC analysis of OxsB assay using the methylation product (18-P) as the substrate but different reducing systems. (B) Same as (A) with the methylidene product (10-P) as the substrate. (C) Proposed mechanism for the transformation of 18-P to 11-P. While the oxidation of intermediate 25 to 10-P is depicted stepwise, a concerted process involving proton coupled electron transfer may also be possible.

AlsB Catalyzes a Similar Reaction Sequence.

AlsB shares 30% sequence identity with OxsB, and recent characterization has established that the oxetane ring forming activity of AlsB also proceeds with abstraction of the 2′-pro-R hydrogen from 2′-dAMP (3).¹⁸ Given the similarity between the two enzymes, it was considered possible that AlsB may also catalyze the methylation of 2′-dAMP. To investigate this possibility, 15 μM AlsB, which was purified bound with cobalamin and further reconstituted with DTT, iron, and sulfide before gel filtration as described previously,¹⁸ was assayed with 0.3 mM 2′-dAMP, 0.5 mM SAM, 1.0 mM NADPH, and 0.25 mM MV in 25 mM HEPES buffer (pH 7.5), incubated at room temperature for 2 h and worked up with CIP. Under these conditions, only the formation of the oxetane related species 23, the structure of which is unknown and currently under investigation, and 24 was observed as previously reported¹⁸ with little to no significant production of 18, 10, or 11 (Figure 5A, trace 3).

Figure 5.

Comparison of AlsB and OxsB activity under different reduction conditions. (A) HPLC analysis comparing (1) the products of OxsB catalysis (i.e., 18, 10, and 11) with the assay of AlsB with 2′-dAMP (3) using (2) dithionite/MV or (3) NADPH/MV as the reducing system. While the structure of the product 23 remains unknown, 24 has been shown to be a mixture of two compounds as shown. The peak annotated as “Ade” is adenosine possibly resulting from the decomposition of NADPH. (B) HPLC analysis of the OxsB assay with 2′-dAMP (3) using 0.25 mM MV with 0.25 mM dithionite in the presence or absence of 1 mM DTT as the reducing system.

When the experiment was repeated, however, in the presence of MV with 0.25 mM dithionite used in place of 1.0 mM NADPH as the reductant, products consistent with 18, 10, and 11 were observed by HPLC after 2 h reaction (Figure 5A, traces 1 and 2). These assignments were based on equivalent HPLC retention times versus the corresponding standards as well as LCMS mass analysis. Under these conditions, neither products 23 nor 24 were observed.

It was noted that AlsB accumulated primarily the methylation product after workup with CIP (18, Figure 5A), whereas OxsB predominantly produced the heterodimer 11 (Figure 2B). This implied that the series of reactions catalyzed by the two enzymes are sensitive to the reductant used in the assay. When OxsB was incubated for 2 h with 0.25 mM dithionite and the remaining assay components including 3, MV, MgCl₂, SAM, HO-Cbl, and DTT followed by workup with CIP, 18 was formed as the dominant product (Figure 5B, trace 1). Omitting DTT did not change the predominance of 18, although production of both 18 and 5′-deoxyadenosine (8) was further decreased (Figure 5B, trace 2). These results suggested that dithionite may be sufficient for supporting the initial methylation reaction but less effective at facilitating the subsequent dehydrogenation and radical addition reactions compared to DTT.

Effect of Dithionite on OxsB Catalysis.

OxsB reconstituted with [Fe₄S₄] and HO-Cbl was also incubated for 2 h with either 0.1 mM 18-P or 10-P in the presence of 0.25 mM MV and either 1 mM DTT (Figure 6A,B, trace 1) or 1 mM DTT plus 0.25 mM dithionite (Figure 6A,B, trace 2) as the reducing system. Whereas the presence of DTT alone supported both dehydrogenation and adduct formation, the assays with dithionite failed to generate either product. The majority of B₁₂-dependent radical SAM enzymes characterized in vitro show little to no activity with dithionite,^23,47–50 which is known to interfere with the cobalamin cofactor.⁵¹ Nevertheless, there are known exceptions including PhpK and Fom3, of which the methylation activities were both initially characterized with dithionite,^52,53 although NADH/MV was later found to be a better reducing system for Fom3.⁵⁴ While an explanation for this behavior is uncertain, it may result from a competition between formation of methylated cobalamin versus poisoning of the cobalamin cofactor by dithionite leading to an overall reduction in activity.

CONCLUSIONS

OxsB and AlsB are known for their unique ability to catalyze contraction of the furanose ring in 2′-dAMP (3) to yield the highly unusual oxetane observed in the nucleosides oxetanocin A (5) and albucidin (7), respectively. In this work, however, these two enzymes are shown to also catalyze regio and stereoselective C-methylation reactions similar to other B₁₂-dependent radical SAM enzymes. This phenomenon is proposed to be induced by over-reduction of the active site cobalamin to the “supernucleophilic” cob(I)alamin state under in vitro assay conditions followed by methyltransfer from SAM to yield the bound MeCbl(III) cofactor (19). While ring contraction would normally be initiated via H atom abstraction from the 2′-pro-R position of 3 to yield the substrate radical 22,¹⁸ in the presence of bound MeCbl(III) a shunt reaction can take place resulting in radical mediated methyl transfer from 19 to 22 to form the methylation product 18-P (Figure 4A).

Methyl transfer proceeds with inversion of configuration consistent with the relative positioning of the radical SAM and cobalamin components within the OxsB active site.¹⁷ Similar arrangements have also been observed in two other structurally characterized B₁₂-dependent radical SAM methylases (i.e., Mmp10 and TokK).^21,27 However, exceptions are known where methylation catalyzed by cobalamin-dependent radical SAM enzymes occurs with the retention of configuration, such as GenK involved in the biosynthesis of gentamicin,^47,55 a glutamine C-methyltransferase during the maturation of coenzyme M reductase,²² and a SAM methylase Orf29 during the biosynthesis of 1-amino-2-methylcyclopropanecarboxylic acid.⁵⁶ The inversion of configuration during the OxsB and AlsB catalyzed C2′-methylations places the hydrogen atom that was originally in the 2′-pro-S position of the substrate 3 now in position for H atom abstraction in a subsequent round of catalysis. This resembles the repeated hydrogen atom abstraction seen for CysS, where the same carbon presents the H atom for abstraction during tert-butyl group formation,^24,26 as well as ThnK, TokK, and TmoD, where the added methyl group instead serves as the subsequent H atom donor.^{23,25,27–29}

In contrast to the consecutive methyltransferases, however, OxsB catalyzes the dehydrogenation of 18-P instead of either ring contraction or a second round of methylation (Figure 6C). The resulting product 10-P is no longer capable of donating a hydrogen from C2′; nevertheless, it can still facilitate reductive homolysis of SAM and subsequent radical addition to the methylidene moiety to yield the radical intermediate 27. As in the case of dehydrogenation, this intermediate does not undergo methylation but rather one electron oxidation to yield the heterodimeric product 11-P. It should be noted that dehydrogenation and C–C coupling do not require that 18-P and 10-P remain bound in the active site after the initial methylation reaction, and it is not clear what if any state is required of the cobalamin cofactor during these two later reactions.

These findings not only broaden the reaction scope of B₁₂-dependent radical SAM enzymes but also set the stage for further investigation of the factors that govern partitioning of the radical intermediates along different reaction pathways in this class of enzymes. Moreover, while the exact role played by cobalamin during the oxetane ring formation catalyzed by OxsB and AlsB is not yet clear, the observed methylation shunt is hypothesized to be related to the in vitro reducing environment, imprecisions of which may also help to explain the relatively poor in vitro activity reported for these enzymes. OxsB and AlsB thus stand as another example of just how sensitive enzymatic reactions involving radical intermediates can be to seemingly minor perturbations. As a consequence, small changes to the active site, substrate structure, or even surrounding environment can induce significant changes in reaction outcomes thereby serving as a cornerstone for the observed evolutionary diversification of radical SAM enzymes.