The Key Intermediate in Radical S-Adenosyl-methionine Catalysis: Caught in the Act.

The selective cleavage of strong chemical bonds is a central challenge in chemistry. A ubiquitous strategy employed both in nature and modern synthetic methods relies on the transient generation of discrete radical species whose high reactivity enables facile bond cleavage processes and the rapid development of functional diversity. Developing catalysts that effectively harness these radical species for precise bond (de)construction remains a leading challenge for synthetic chemists. Meanwhile, nature has mastered this task. Radical enzymes are universally deployed in biochemical processes and operate with unprecedented levels of selectivity and reactivity. Many of these natural catalysts are members of the aptly named radical SAM (rSAM) enzyme superfamily, which includes over 100 000 homologous enzymes.[1] A postulated intermediate central to rSAM reactivity is the 5′-deoxyadenosyl radical (5′dAdo·), generated via one electron reduction of S-adenosyl-methionine (SAM) by an essential Fe4S4 cofactor. Despite over 20 years of intensive investigation, this intermediate has evaded direct detection, leading many to wonder if this radical even exists in a discrete, free state. In this issue of ACS Central Science, Sayler and co-workers disclose the direct trapping and characterization of the elusive 5′dAdo· radical within the confines of the hydrogenase-cofactor maturase enzyme HydG.[2]

The importance of hydrogenase activity in microorganisms and the potential to exploit these enzymes for renewable energy production has attracted steadfast research interest from a large community of enzymologists, inorganic chemists, and biophysicists.[3] In particular, the highly complex, organometallic nature of the di-iron hydrogenase active site cofactor (termed the H Cluster, Figure A) has warranted investigations into the machinery responsible for its biosynthesis. HydG is one of two rSAM enzymes implicated in H cluster biosynthesis and the focus of the accompanying article. HydG serves two distinct functions: lyse tyrosine into the essential CO and CN– ligands and facilitate their installation onto a nascent Fe synthon, which is later matured into the complete H cluster.[4] Previous isotopic labeling studies demonstrated that the iron-bound CO and CN– ligands are derived from exogenous tyrosine, but mechanistic questions regarding how these diatomic ligands are generated remain unanswered. In line with other rSAM enzymes, 5′dAdo· is a key proposed intermediate thought to abstract a single hydrogen atom from its native substrate, tyrosine. However, when assay mixtures composed of HydG, SAM, and tyrosine are rapidly quenched, 5′dAdo· is not observed. Instead, a downstream intermediate of fragmented tyrosine (the 4-oxidobenzyl radical) is the only detectable organic radical, indicating that under these conditions free 5′dAdo· is, at best, a fleeting intermediate.[4]

Figure 1

(A) Native reaction catalyzed by HydG. Tyrosine lysis affords diatomic CO and CN– ligands that are components of the H cluster of di-iron hydrogenases. (B) Freeze quenching assay mixtures of HydG with the non-native substrates explored by Sayler[2] yield distinct EPR signals (right) that are assigned to the corresponding radical species. EPR signals are reproduced with permission from ref (2). Copyright 2019 American Chemical Society.

To address the outstanding question of how tyrosine fragments, the authors explored the reactivity of HydG with non-native substrate analogues lacking the amino group hypothesized to be the direct target of 5′dAdo· (Figure B). Replacement of tyrosine with 4-hydroxyphenylpropanoic acid (HPPA) or cis-p-coumaric acid in typical HydG reaction assays furnished new radical intermediates upon freeze quenching. In the case of HPPA, the relatively weak alkyl C2–H bond is rapidly cleaved by 5′dAdo· furnishing a remarkably stable HPPA·. Surprisingly, reactions employing cis-p-coumaric acid yield electron paramagnetic resonance (EPR) signals ascribable to the precursor 5′dAdo·. Apparently, the strong vinylic C–H bonds of cis-p-coumaric acid are not readily attacked by 5′dAdo· and enable the accumulation of this radical in the reaction mixture. To cement their assignment, the authors prepared a series of isotopically labeled 5′dAdo· analogues which further reveal the intimate electronic and geometric details of this fleeting radical species.

The work presented by the authors is the latest contribution to an extremely productive year of research directed toward the investigation of radical intermediates in rSAM enzymes. Particularly noteworthy are joint efforts by the Broderick and Hoffman laboratories[5,6] which have led to the discovery of a complementary, photolytic means of generating SAM-derived radical species. This research team demonstrated that cryogenic photolysis of SAM-bound mixtures of the pyruvate-formate lyase activating enzyme or HydG yields significant quantities of 5′dAdo· or methyl radicals (CH3·) respectively, resulting from photoinduced electron transfer from the Fe4S4 cofactor to SAM. Importantly, the spectroscopic features of 5′dAdo· generated via photoreduction or the in situ trapping described in this issue are essentially identical, safely establishing that this radical is indeed a viable intermediate in rSAM catalysis. The photoinduced generation of CH3· emphasizes that the one-electron reduction of SAM can, in principle, furnish one of three distinct radical intermediates. There are, however, very few known enzymes that circumvent the canonical generation of 5′dAdo· or its Fe-bound isomer.[7,8] Given the largely uncharacterized state of the rSAM superfamily,[1] it is plausible that new enzymes will be discovered that purposefully avoid 5′dAdo· generation in place of alternative radical intermediates attendant with orthogonal reactivity.

The accumulation of large quantities of the primary radical intermediate of rSAM enzymes in operando presents a unique opportunity to address basic questions in rSAM research. Perhaps the most surprising result disclosed by the authors is that 5′dAdo· is stable within HydG on the second time scale. Addressing how the protein environment stabilizes a primary alkyl radical to this extent will certainly be a topic of future discussion and experiments. Regardless, the stability of 5′dAdo· in HydG naturally opens the door to kinetic interrogation of this intermediate. If the native tyrosine substrate can displace cis-p-coumaric acid on a compatible time scale, single turnover stopped flow experiments could yield a wealth of mechanistic insight into key bond scission processes mediated by 5′dAdo·.

Now that the generation of free 5′dAdo· within HydG has been established, an assessment of the generality of this intermediate to other rSAM enzymes must be pursued. Is free 5′dAdo· a consensus intermediate or an exception to the rule?[8] The strategies described by Sayler and co-workers provide an instructive blueprint toward these future efforts.