The recently identified ‘‘radical-SAM’’61superfamily comprises a diverse group of enzymes which at first glance bear little resemblance to aconitase and the related dehydratases. However, as we shall see, the radical-SAM enzymes, like aconitase, appear to use a unique iron site in a [4Fe–4S]
cluster to coordinate substrate. A common feature of the members of the radical-SAM super- family is the presence in the amino acid sequence of a characteristic cluster-binding motif (CX3CX2C) containing only three cysteines, and evidence suggests that these three cysteines alone are responsible for cluster binding.62 The presence of only three cysteines in the cluster- binding motif is reminiscent of aconitase and related enzymes in that it suggests the presence of a unique, noncysteinyl-coordinated iron in a [4Fe–4S] cluster, although the putative noncysteinyl ligand has not been identified for any of the enzymes in this family.
The radical-SAM superfamily is so-called because its members catalyze radical reactions and utilize S-adenosylmethionine (SAM or AdoMet) as a required cofactor or co-substrate.61 These enzymes also contain catalytically essential iron–sulfur clusters.62The best-studied members of this family, including lysine aminomutase, pyruvate formate-lyase activating enzyme, anaerobic ribo- nucleotide reductase, lipoate synthase, and biotin synthase, were being investigated long before the superfamily was named, and have provided important clues to the catalytic mechanism and the role of the iron–sulfur cluster and AdoMet.62 Additional members of this family are being identified and isolated at a fairly rapid pace. In fact, the bioinformatics approach that was used to identify the superfamily has revealed hundreds of other putative members throughout the phylo- genetic kindgom, from simple unicellular organisms to humans.61
8.27.3.1.1 Lysine 2,3-aminomutase
Lysine 2,3-aminomutase, known alternatively as LAM or KAM, catalyzes a rearrangement reaction of lysine, the interconversion of L-lysine and L--lysine (Figure 8A). The reaction, which is catalytic in AdoMet, is directly analogous to rearrangement reactions catalyzed by coenzyme B12-dependent enzymes,63–66and in fact in both the B12 and the AdoMet-dependent enzymes the key initial step in catalysis is the abstraction of a hydrogen atom to yield a substrate radical intermediate.67,68 In the case of the B12-dependent enzymes, the hydrogen atom is abstracted by a B12-derived 50-deoxyadenosyl radical intermediate generated via homolytic CoC bond cleavage. In the case of the AdoMet-dependent lysine aminomutase, the hydrogen atom is abstracted by an AdoMet-derived 50-deoxyadenosyl radical. Such a radical mechanism was originally proposed by Moss and Frey,69and later supported experimentally by the observa- tion of a kinetically competent lysine radical,70–72 as well as by the spectroscopic observation of an allylic analogue of the putative 50-deoxyadenosyl radical intermediate.73,74 This intriguing parallel between the B12 and the AdoMet-dependent radical enzymes, the intermediacy of a 50-deoxyadenosyl radical, extends to other members of the radical-SAM superfamily as well. In addition to the requirement for an iron–sulfur cluster and AdoMet, LAM also requires pyridoxal 50-phosphate, which forms an adduct with the substrate during turnover.
8.27.3.1.2 Pyruvate formate-lyase activating enzyme
Pyruvate formate-lyase activating enzyme (PFL-AE) was identified early on as an iron-dependent protein required to activate pyruvate formate-lyase (PFL) for its essential function in anaerobic glucose metabolism in bacteria, the conversion of pyruvate and coenzyme-A (CoA) to formate and acetyl-CoA.75–77The activation of PFL for this central function involves the generation of a stable and catalytically essential glycyl radical,76 which is generated under anaerobic conditions by PFL-AE (Figure 8B) and is quenched once conditions become increasingly oxic (PFL is not needed under aerobic conditions) by the PFL deactivating enzyme.78 The activation of PFL by PFL-AE, unlike the LAM reaction, involves thestoichiometricconsumption of AdoMet, which is converted to methionine and 50-deoxyadenosine by reductive cleavage of the sulfur–carbon bond of AdoMet.76 Isotopic labeling studies showed early on that the hydrogen atom abstracted from the glycyl residue of PFL ended up at the 50-carbon of 50-deoxyadenosine,79 thereby implicating
an intermediate adenosyl radical in catalysis. The nature of the iron requirement for PFL-AE remained a mystery until PFL-AE was identified as an iron–sulfur protein.80
8.27.3.1.3 Anaerobic ribonucleotide reductase activating enzyme
The anaerobic ribonucleotide reductase (aRNR), like pyruvate formate-lyase, is a bacterial enzyme that operates only under anaerobic conditions. Like PFL, aRNR must be activated for catalysis, which involves the generation of a stable glycyl radical on the enzyme (Figure 8B).81,82 The aRNR activating enzyme, once considered to be merely the2subunits of an 22holoen- zyme,83 is now known to function catalytically with respect to the aRNR (i.e., the2compon- ent).84 However, the activase (2) is tightly associated with the reductase (2), unlike the activating enzyme of PFL-AE, which is not tightly associated with PFL. The aRNR activating enzyme has been shown to contain an iron–sulfur cluster and to utilize AdoMet in catalysis, converting it stoichiometrically to methionine and 50-deoxyadenosine.85,86
8.27.3.1.4 Biotin synthase and lipoate synthase
The biotin and lipoate synthases catalyze similar reactions, the insertion of sulfur into unactivated CH bonds to generate essential cofactors (Figures 8C and 8D). The substrate in the case of biotin synthase is dethiobiotin, with a single sulfur inserted into two CH bonds to generate the tetrahydrothiophene ring of biotin. In the case of lipoate synthase, two atoms of sulfur are inserted, one each into the CH bonds at positions 6 and 8 of octanoic acid to produce dihydrolipoate, which is typically isolated in the oxidized form shown inFigure 8D. (The actual
NH3+
COO– H H
+H3N H
NH3+
COO– NH3+ H
H H
HN NH O
H H
A B
C D
COOH H N N H
O
COOH H N N H S O
COOH
N N H
O O
N NH
O O
N NH
O O
N NH
O
DNA backbone DNA backbone O
HN NH O
H
COOH S
S
E
Figure 8 Reactions catalyzed by the radical-SAM superfamily. A. Lysine 2,3-aminomutase. B. Activating enzymes. C. Biotin synthase. D. Lipoic acid synthase. E. Spore photoproduct lyase.
substrate of lipoate synthase is not free octanoic acid, but octanoate bound to acyl-carrier protein (octanoyl-ACP).87) Both biotin synthase88 and lipoate synthase89,90 have been shown to contain iron–sulfur clusters, and both require AdoMet for catalysis.87,91 In the case of biotin synthase, there is evidence for hydrogen atom transfer from substrate into deoxyadenosine, thereby impli- cating an intermediate 50-deoxyadenosyl in the catalytic mechanism.92The deoxyadenosyl radical presumably abstracts a hydrogen atom from substrate to generate a substrate radical intermedi- ate, to which sulfur is added. Evidence points to an iron–sulfur cluster being the source of the added sulfur, as is discussed further inSection 8.27.3.4.
8.27.3.1.5 Spore photoproduct lyase
Spore photoproduct lyase (SPL) catalyzes the repair of UV-induced DNA damage in Bacillus, and possibly other spore-forming microorganisms (Figure 8E).93–95It has been shown to contain an iron–sulfur cluster and to utilize AdoMet in catalysis.96,97Like lysine aminomutase but unlike the other radical-SAM enzymes discussed above, SPL appears to use AdoMet catalytically;
i.e., AdoMet is not consumed during substrate turnover.98 Evidence for a radical mechanism of DNA repair has been obtained, as repair of damaged DNA labeled at C-6 of thymine results in specific label transfer into AdoMet, suggesting that C-6 H atom abstraction by a 50-deoxyadeno- syl radical intermediate is the initial step in DNA repair.98
8.27.3.2 Properties of the Iron–Sulfur Clusters
As stated previously, a common feature among the radical-SAM enzymes is the presence of a distinctive three-cysteine cluster-binding motif,62 as was also seen in aconitase. This conserved cluster-binding motif leads to similarities in the cluster properties among the members of this family. As will be discussed in more detail below, the [4Fe–4S] cluster is now known to be the catalytically relevant cluster. However, a distinctive feature of the clusters in these enzymes is their lability; thus, the literature on these enzymes provides evidence for [2Fe–2S] clusters, cuboidal and linear [3Fe–4S] clusters, and [4Fe–4S] clusters in a variety of oxidation states.62 This cluster lability, though somewhat confusing in the early literature on these enzymes, is reminiscent of aconitase, in which all these cluster forms were also observed. However even among the radical- SAM enzymes the degree of cluster lability is quite variable.62
Pyruvate formate-lyase activating enzyme is the member of the radical-SAM family whose cluster properties are most similar to those of aconitase. The cluster in pyruvate formate-lyase activating enzyme is quite labile, and in fact until 1997 it was not known that the enzyme contained an iron–sulfur cluster, as all preparations to that time had been done aerobically, under which conditions the cluster falls apart.75It was initially reported that PFL-AE contained a mixture of [2Fe–2S] and [4Fe–4S] clusters,80 and subsequent reconstitution studies of the apo enzyme provided evidence for a[4Fe–4S] cluster.99 Further studies showed that anaerobic isol- ation resulted in purification of a form of PFL-AE that contained primarily [3Fe–4S]þclusters, which upon reduction converted to [4Fe–4S] clusters.100,101This reductive cluster conversion from [3Fe–4S]þto [4Fe–4S]2þclusterseven in the absence of added ironwas remarkably reminiscent of aconitase (see Section 8.27.2.2), and suggested a labile cluster site. Adding to the similarity to aconitase, Mo¨ssbauer spectroscopy provided evidence for a linear [3Fe–4S] cluster in PFL-AE isolated under appropriate conditions.101Therefore all of the cluster forms previously identified in aconitase were also found in PFL-AE, and like aconitase it appeared to be relatively simple to interconvert between these cluster forms.101
The other members of the radical-SAM family have shown some, but not all, of the cluster properties observed for PFL-AE. For example, evidence for both [2Fe–2S]2þ and [4Fe–4S]2þ clusters have been reported for both biotin synthase102–104 and lipoate synthase.89,90,104 In fact, the [2Fe–2S] cluster in biotin synthase is quite stable, and thus biotin synthase can be purified aerobically with the [2Fe–2S] cluster intact,88and then reconstituted anaerobically to generate the [4Fe–4S] cluster.102–105 No significant amount of [3Fe–4S]þ cluster has been observed in biotin synthase, although lipoate synthase appears to contain some [3Fe–4S]þ depending on isolation conditions.87
It had been previously proposed that the reductive cluster conversions (in the absence of added iron) in the radical-SAM enzymes occurred by cluster ‘‘cannibalization,’’ as had also been
proposed for aconitase. That is, reduction results in labilization of the metal–ligand bonds, essentially releasing iron and sulfide which can then reassemble into the thermodynamically favored cluster form under reducing conditions. Direct evidence for such release and reabsorption of iron and sulfide during reductive cluster conversions was provided for biotin synthase,106 supporting this general mechanism of cluster conversion in the radical-SAM enzymes.
In contrast to these examples is lysine aminomutase, in which the only significant cluster form observed is the [4Fe–4S] form, although a [3Fe–4S] cluster can be generated upon oxidation.107 However, no [2Fe–2S] cluster has been reported for this enzyme. LAM is also the only member of this family in which a [4Fe–4S]3þcluster has been observed.107The iron–sulfur cluster in spore photoproduct lyase has not been characterized, although some evidence for [4Fe–4S] and [3Fe–4S]
forms has been obtained.97
In summary, the common CX2CX3C cluster-binding motif found in the radical-SAM enzymes confers some common properties to the clusters in these enzymes, including cluster lability.
However, the details of the lability and the precise cluster forms observed vary from enzyme to enzyme.
8.27.3.3 Involvement of the Clusters in Radical Catalysis
The variety of clusters observed in these enzymes, and the cluster lability, led to difficulties in determining unequivocally the active cluster form and oxidation state. Fontecave and co-workers showed that, in the absence of the aRNR (also known as the2domain), the aRNR-AE (or2 domain) in the [4Fe–4S]þ state reacts with AdoMet to generate methionine concomitant with cluster oxidation, with a stoichiometry of 2–3 methionines per cluster oxidized.85Although this is not the physiologically relevant reaction of radical generation on aRNR, it does demonstrate the ability of the [4Fe–4S]þ-aRNR-AE to reductively cleave AdoMet. Later studies on aRNR con- firmed the stoichiometry of approximately two methionines produced per [4Fe-4S]þ oxidized, although for aRNR-AE in the presence of the aRNR only one methionine, along with 0.5–0.9 glycyl radicals, was produced per [4Fe–4S]þoxidized.86
Evidence for the [4Fe–4S]þcluster as the active form of lysine aminomutase was obtained by Frey and co-workers, who showed by a combination of EPR spectroscopy and enzyme assays that the [4Fe–4S]þ-LAM generated in the presence of AdoMet was catalytically active.108Unlike aRNR-AE, however, LAM catalyzes a reversible reductive cleavage of AdoMet, and thus methionine production and cluster oxidation could not be monitored as evidence of turnover. It is of interest to note that in the case of LAM, the presence of AdoMet facilitates reduction to the [4Fe–4S]þstate; very little [4Fe–4S]þcluster is produced by the reduction of LAM with dithionite in the absence of AdoMet, while the presence of AdoMet or its analogue S-adenosylhomocysteine dramatically increases the quantity of [4Fe–4S]þproduced.108It is not clear whether the presence of AdoMet affects the redox potential of the cluster or whether some other effect, such as accessibility of the cluster by the reductant, is at work.
The most direct demonstration of the involvement of the [4Fe–4S]þcluster in radical catalysis by the radical-SAM enzymes was obtained in the case of PFL-AE.109PFL-AE was reduced from the [4Fe–4S]2þto the [4Fe–4S]þstate using photoreduction in the presence of 5-deazariboflavin.
EPR samples of the enzyme in the presence of AdoMet alone show increasing amounts of [4Fe–4S]þ with increasing times of illumination. (Unlike the aRNR-AE, PFL-AE does not reductively cleave AdoMet at an appreciable rate in the absence of the other substrate, PFL.) Parallel samples to which PFL had been added (addition was performed in the dark to eliminate the exogenous reductant) showed increasing quantities of a glycyl radical EPR signal with increasing illumination time. Spin quantitation of the EPR signals from the parallel PFL-AE/AdoMet and PFL-AE/AdoMet/PFL samples demonstrated a 1:1 correspondence between the amount of [4Fe–4S]þ in the former to the amount of glycyl radical in the latter.109 Furthermore, the [4Fe–4S]þ EPR signal disappears upon addition of PFL and generation of the glycyl radical, thereby suggesting that the [4Fe–4S]þcluster is oxidized to [4Fe–4S]2þ. It was proposed that the [4Fe–4S]þ cluster provided the electron necessary for reductive cleavage of AdoMet to generate methionine and the putative 50-deoxyadenosyl radical intermediate.109
Together, the results described for aRNR-AE, LAM, and PFL-AE point to the [4Fe–4S]þ cluster as the catalytically active cluster, and they point to a role for this cluster in providing the electron necessary for reductive cleavage of AdoMet, either reversibly (as in the case of LAM) or irreversibly (as in the cases of the two activating enzymes), as illustrated inFigure 9.62Therefore,
although many of the cluster properties of the radical-SAM enzymes are similar to those of aconitase, the precise role of the cluster in catalysis is not. In aconitase, the [4Fe–4S]2þ cluster serves as a Lewis acid, binding and activating substrate for the dehydratase reaction, but not acting in a redox role. In contrast, the cluster in the radical-SAM enzymes is a redox-active cluster, with the reduced [4Fe–4S]þstate being catalytically active. The oxidized [4Fe–4S]2þstate is produced either as an intermediate (e.g., LAM) or a product (e.g., aRNR-AE and PFL-AE) as AdoMet is used catalytically or stoichiometry, respectively.
8.27.3.4 Interaction of S-Adenosylmethionine with the Clusters
The involvement of a[4Fe–4S]þcluster and its role in reductive cleavage of AdoMet led to the question of precisely howthe [4Fe–4S]þcluster was involved in catalysis. Based on the evidence described in the previous section, it was conceivable that the cluster served as a remote electron- transfer center, reducing AdoMet via long-range electron transfer to generate an AdoMet radical which subsequently underwent reductive CS bond cleavage. Alternatively, the cluster might interact directly with AdoMet to mediate this unusual radical generation reaction. A number of pieces of evidence had hinted at the possibility that AdoMet interacted directly with the cluster, including the observation of dramatic changes in the EPR signal line shape of the [4Fe–4S]þupon addition of AdoMet110,111 and increased ability to reduce the cluster in the presence of Ado- Met.108
Additional evidence for a close association between AdoMet and the [4Fe–4S] cluster in the radical-SAM enzymes came from selenium K-edge X-ray absorption studies of lysine amino- mutase in the presence of the cleaved cofactor S-adenosyl-L-selenomethionine (Se-AdoMet).112 Selenium EXAFS of the LAM/Se-AdoMet complex itself did not show a close contact to an iron of the cluster. However, in the presence of DTT and the substrate analogtrans-4,5-dehydrolysine, the cofactor was cleaved to deoxyadenosine and selenomethionine, and a close (2.7 A˚) contact to an iron of the [4Fe–4S] cluster was observed in the selenium EXAFS.112This led the authors to propose a close association between AdoMet and the [4Fe–4S] cluster of LAM, with the sulfonium sulfur situated near to the presumed unique iron site of the cluster (Figure 10). Substrate binding would then bring AdoMet closer to the iron–sulfur cluster, allowing electron transfer from the cluster and ultimately sulfur–carbon bond cleavage, which would leave methionine in close contact with the unique iron site (Figure 10).
Electron-nuclear double resonance (ENDOR) studies of PFL-AE complexed to specifically isotopically labeled AdoMets has revealed the details of the interaction between AdoMet and the cluster in this enzyme.111–113Deuterium ENDOR spectraof PFL-AE in the [4Fe–4S]þstate complexed with methyl-D2-AdoMet showed a pair of peaks centered at the deuteron Larmor frequency and split by the hyperfine coupling to the spin of the cluster.111Examination of the field- dependence of the coupling showed that it was dipolar in nature, and gave an estimation of the
S Fe S Fe
S Fe Fe
S RS
RS
SR X
[4Fe–4S]+
–OOC NH3+
S O H A H H3C
HO OH
–OOC NH3+
S CH3
S Fe S Fe
S Fe Fe
S RS
RS
SR X
[4Fe–4S]2+
O H A H
HO OH
O H A H
HO OH H
+ SH + S•
Figure 9 A generalized reaction scheme for the radical-SAM enzymes. The [4Fe–4S]þ provides the electron necessary for the reductive cleavage of AdoMet to generate the intermediate adenosyl radical.
The adenosyl radical abstracts a hydrogen atom from substrate (SH) to initiate the radical reaction.
distance from the nearest deuteron to the closest iron of the cluster of approximately 3.0–3.8 A˚.111 Fully consistent results were obtained from 13C-ENDOR studies of [4Fe-4S]þ/PFL-AE in the presence of AdoMet labeled at the methyl carbon with 13C.111 In this case the estimated distance to the nearest iron of the cluster was 4–5 A˚, and the coupling could best be modeled through a combination of through-space and through-bond contributions. The presence of isotropic through-bond contributions to the 13C coupling requires that there be some orbital overlap, in order to provide a pathway for delocalization of unpaired spin density from the cluster. Due primarily to electrostatic considerations, it was proposed that the orbital overlap occurs via a close association of the AdoMet sulfonium with one of the3-bridging sulfides of the cluster (Figure 11).111
The results just described probed the interaction of AdoMet with the catalytically active [4Fe–4S]þ cluster of PFL-AE. Interaction of AdoMet with the oxidized [4Fe–4S]2þ cluster cannot be probed directly using ENDOR spectrsocopy, since the [4Fe–4S]2þcluster is diamagnetic. In order to probe the interaction with the 2þcluster, therefore, PFL-AE in the [4Fe–4S]2þstate was mixed
S Fe S Fe
S Fe Fe
S RS
RS
SR X
–OOC NH3+
Se O A CH3 OH
OH
[4Fe–4S]+ [4Fe–4S]+
S Fe S Fe
S Fe Fe
S RS
RS
SR X
–OOC NH3+
Se CH3
[4Fe–4S]2+ [4Fe–4S]2+
O A
OH OH
S Fe S Fe
S Fe Fe
S RS
RS
SR X
–OOC NH3+
Se O A CH3 OH
OH
S Fe S Fe
S Fe Fe
S RS
RS
SR X
–OOC NH3+
Se O A CH3 OH
OH substrate
2.7 Å
Figure 10 Interaction of AdoMet with the iron sulfur cluster in LAM to generate the intermediate deoxyadenosyl radical. Adapted from ref. 112.
S Fe S Fe
S Fe Fe
S RS
RS
O N H2 S C Ado
O
H H H 4-5 Å 3–3.8 Å
Figure 11 The interaction of AdoMet with the [4Fe–4S] cluster of PFL-AE as deduced from2H,13C,17O, and15N ENDOR measurements and54Fe Mo¨ssbauer studies.