Radical SAM Enzymes: A [4Fe-4S] Cluster Cracks S-Adenosylmethionine

One cluster, one SAM, one radical: the unifying reaction

Radical SAM enzymes are defined by a single chemical event: a reduced [4Fe-4S]¹⁺ cluster donates one electron to S-adenosylmethionine (SAM), cleaving the bond between the sulfonium sulfur and the 5′-carbon of the adenosyl group. The products are L-methionine and the 5′-deoxyadenosyl radical (5′-dAdo•) — a primary carbon radical so reactive it will abstract a hydrogen atom from almost any C–H bond it is held next to, including bonds with dissociation energies near 100 kcal/mol. That single H-atom abstraction is the committed step; everything downstream is the enzyme steering the resulting substrate radical toward isomerization, elimination, sulfur insertion, ring formation, or complex rearrangement.

The superfamily was recognized bioinformatically by Heidi Sofia and coworkers in 2001 (Sofia, Chen, Hetzler, Reyes-Spindola & Miller, Nucleic Acids Research), who noticed that ~600 sequences from wildly different pathways — heme, biotin, lipoic acid, molybdopterin, and DNA-repair biosynthesis — all shared one telltale fingerprint. That fingerprint is the CX₃CX₂C motif: three cysteines spaced Cys-X-X-X-Cys-X-X-Cys, which together bind only three of the four irons of a [4Fe-4S] cube. The superfamily has since exploded; sequence-similarity network resources such as RadicalSAM.org now catalog well over 700,000 members, making it the largest enzyme superfamily yet described.

The genius of the design is a redox inversion. The cell uses a strong reductant (the low-potential cluster) to manufacture a strong oxidant (the carbon radical). As Perry Frey put it, SAM here acts as a "poor man's coenzyme B12" — accomplishing the same 5′-deoxyadenosyl-radical chemistry as expensive adenosylcobalamin, but from cheap, ubiquitous SAM and an iron-sulfur cluster that predates the oxygen atmosphere.

The site-differentiated cluster and the geometry of cleavage

An ordinary [4Fe-4S] ferredoxin cluster has all four irons capped by protein cysteinates. Radical SAM clusters are deliberately site-differentiated: three irons are held by the CX₃CX₂C cysteines, leaving one iron — the "unique" iron — with an open coordination sphere. SAM binds this unique iron as a bidentate N,O-chelate through its own amino group and carboxylate, positioning the sulfonium center within van der Waals distance of the cluster. Crystallography of pyruvate formate-lyase activating enzyme, MoaA, HydE, and lysine 2,3-aminomutase all confirm this same coordination geometry.

Mechanistically this chelation does two things. First, it locks SAM's sulfonium syn to the cluster so the S–C5′ bond points at the electron source, enforcing regioselective cleavage of S–C5′ (not the S–Cγ bond that would give a methyl radical, nor S–C(methionine)). Second, ligation raises the cluster's reduction potential window and couples electron transfer to bond breaking. The overall step is:

• [4Fe-4S]¹⁺ + SAM → [4Fe-4S]²⁺ + L-Met + 5′-dAdo•

Note the electron bookkeeping: the cluster is oxidized by one electron (1+ → 2+), the sulfonium sulfur S⁺ is reduced to neutral thioether sulfur in methionine, and the electron pair that formed the S–C5′ bond leaves entirely on carbon as the odd electron of the radical. This is formally a one-electron reductive cleavage, and it is thermodynamically uphill by tens of kcal/mol in isolation — the enzyme pays for it with binding energy and by immediately consuming 5′-dAdo• in the exergonic H-atom abstraction that follows.

The Ω intermediate: the radical is 'never free'

For two decades the 5′-deoxyadenosyl radical was drawn as a discrete, freely tumbling species. The picture changed in 2016, when Masaki Horitani, Joan Broderick, Brian Hoffman and coworkers (Science 2016) trapped and characterized an organometallic intermediate, dubbed Ω, in the pyruvate formate-lyase activating enzyme reaction. Using electron-nuclear double resonance (ENDOR) spectroscopy on ¹³C-labeled SAM, they showed a genuine Fe–C5′ σ-bond: the 5′-carbon is covalently attached to the unique iron of the cluster.

Under this refined mechanism, reductive cleavage of SAM does not release a bare radical into solution. Instead the nascent C5′ radical is captured by the unique iron to form Ω, and the true radical-generating step is the homolysis of that Fe–C5′ bond, which liberates 5′-dAdo• only at the instant it is needed, held rigidly against the substrate's target C–H. Broderick and Hoffman summarized the result with the memorable line that the "free radical is never free": the enzyme never lets a promiscuous carbon radical wander, which is exactly how it avoids off-pathway damage to itself and to the genome.

The regioselective S–C5′ cleavage itself has a satisfying geometric origin. It is the enforced binding geometry, not any special electronic distortion, that dictates which bond breaks: the bidentate N,O-chelation of SAM to the unique iron rigidly positions only the S–C5′ bond toward the reducing cluster, aiming the donated electron at C5′ rather than at the other carbons flanking the sulfonium. The clean result is that of the three C–S bonds available at the sulfonium, only S–C5′ is broken, every single turnover.

Worked example: lysine 2,3-aminomutase

Lysine 2,3-aminomutase (LAM/KAM) from Clostridium subterminale is the canonical case study, worked out largely by Perry Frey, Marcus Ballinger, and George Reed beginning in the late 1980s. It interconverts L-α-lysine and L-β-lysine — a formal 1,2-shift of the α-amino group from C2 to C3 — and it needs three partners: the [4Fe-4S] cluster, SAM, and a pyridoxal-5′-phosphate (PLP) cofactor bound to the substrate as an external aldimine.

The choreography is a textbook radical relay:

• Step 1: [4Fe-4S]¹⁺ reduces SAM → 5′-dAdo• + methionine.
• Step 2: 5′-dAdo• abstracts the C3 hydrogen of the PLP-lysine aldimine (the abstracted H ends up on the 5′-methyl of 5′-deoxyadenosine), generating a substrate radical.
• Step 3: the radical cyclizes through an azacyclopropylcarbinyl radical intermediate that migrates the C–N bond from C2 to C3.
• Step 4: the product radical reabstracts a hydrogen from 5′-deoxyadenosine, regenerating 5′-dAdo• and releasing β-lysyl-PLP.

The PLP is essential: by forming the aldimine it delocalizes the unpaired spin into the electron-poor pyridinium π-system (which acts as an electron sink), lowering the barrier for the otherwise difficult 1,2-amino migration. Frey's group nailed the key intermediates directly — EPR spectroscopy detected the PLP-stabilized substrate and product radicals and even the allylic 5′-dAdo•-derived species — providing some of the first spectroscopic proof that these enzymes really run through discrete carbon radicals rather than concerted two-electron chemistry.

Limits, subtleties, and where the debates still live

Several features separate a real radical SAM enzyme from a naïve cartoon. First, SAM is not always catalytic. In many members — the sulfur-inserting enzymes like biotin synthase (BioB) and lipoyl synthase (LipA) — SAM is a co-substrate consumed stoichiometrically (two SAM per two C–H bonds functionalized), and part of one of the enzyme's own iron-sulfur clusters is sacrificed as the sulfur donor: biotin synthase draws its sulfur from a sacrificial [2Fe-2S] cluster, while lipoyl synthase sacrifices an auxiliary [4Fe-4S] cluster. Either way the protein is self-limiting and must be reconstituted. Contrast this with lysine 2,3-aminomutase, where 5′-dAdo• is regenerated each cycle and SAM behaves as a true cofactor.

Second, many members carry auxiliary iron-sulfur clusters beyond the SAM-binding one — an added complication that structural work through the 2010s repeatedly uncovered (e.g., the SPASM and twitch domains that host [4Fe-4S] clusters for downstream electron transfer in enzymes making ribosomally synthesized natural products, RiPPs). These auxiliary clusters are the reason the field could not simply read function from the CX₃CX₂C motif alone.

Third, real subtleties remain genuinely debated. The precise nature of the electron-transfer/bond-cleavage coupling — whether S–C5′ homolysis is fully concerted with electron transfer (an inner-sphere, Ω-mediated event) or better described as sequential ET then cleavage — is still refined case by case; the Ω organometallic intermediate is well-established in PFL-AE and several others but has not been trapped in every family member. Likewise, whether Ω lies on the productive pathway in all radical SAM enzymes, or is sometimes an off-pathway resting/rescue state, remains an active question. And because 5′-dAdo• is so reactive, direct kinetic measurement of its 'free' concentration is essentially impossible — the 'never free' model is precisely what resolves that paradox.

Reach: from antibiotics to the molecular clock

Because a caged 5′-deoxyadenosyl radical can attack almost any C–H bond, radical SAM chemistry underwrites a staggering breadth of biosynthesis. Members build biotin (BioB), lipoic acid (LipA), heme (HemN, oxygen-independent coproporphyrinogen oxidase), molybdopterin (MoaA), thiamine, and F420; they activate the glycyl-radical enzymes of anaerobic metabolism (PFL-AE installs the catalytic glycyl radical on pyruvate formate-lyase); and they perform the most chemically audacious tailoring steps in complex natural products.

Notable feats include fosfomycin biosynthesis (the cobalamin-dependent radical SAM C-methyltransferase Fom3 installs the methyl group), the carbon-skeleton rearrangements of the herbicide phosphinothricin, the thiostrepton and other RiPP macrocyclizations, and — strikingly — the tRNA- and protein-modifying methylthiotransferases (MiaB installs a methylthio group on a tRNA nucleobase; RimO methylthiolates an aspartate residue of ribosomal protein S12) that install methylthio groups on nucleobases and proteins. The DNA-repair photolyase-adjacent enzyme spore photoproduct lyase even uses 5′-dAdo• to reverse UV damage in bacterial spores in the dark.

The evolutionary reading is that this superfamily is ancient. Its dependence on an oxygen-sensitive [4Fe-4S] cluster and its prevalence in anaerobes point to origins on the early, reducing Earth, before the Great Oxygenation Event forced life to invent oxygen-tolerant alternatives (many of which, like B12-dependent isomerases, do the same 5′-dAdo• chemistry by a different route). Twenty-plus years after Sofia's 2001 paper, radical SAM enzymes remain a frontier for both mechanistic enzymology and biocatalyst discovery: every genome sequenced adds thousands more of these clusters waiting to crack a sulfonium and let a caged radical fly.

Two routes to the 5′-deoxyadenosyl radical: radical SAM vs. adenosylcobalamin (coenzyme B12)

Feature	Radical SAM enzymes	Adenosylcobalamin (B12)
Radical source	S-adenosylmethionine + [4Fe-4S]¹⁺	5′-deoxyadenosylcobalamin (AdoCbl)
Bond cleaved	S–C5′ (reductive, one-electron)	Co–C5′ (homolytic, thermal)
Cofactor cost per turnover	1 SAM consumed (often stoichiometric)	AdoCbl regenerated (catalytic)
Radical carrier / cap	Unique Fe of the cluster (Ω intermediate)	Cob(II)alamin, Co(II)
Bond strength broken	SAM S–C ≈ 60 kcal/mol (lowered by binding)	Co–C ≈ 30 kcal/mol
Evolutionary age / scope	Ancient, anaerobic-adapted, >700k members	Later, O₂-tolerant B12-dependent isomerases