Pyridoxal-5′-phosphate as the catalyst for radical isomerization in reactions of PLP-dependent aminomutases

Biochimica et Biophysica Acta 1814 (2011) 1548–1557

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b b a p a p

Review

Pyridoxal-5′-phosphate as the catalyst for radical isomerization in reactions of PLP-dependent aminomutases☆ Perry A. Frey ⁎, George H. Reed Department of Biochemistry, University of Wisconsin-Madison, 1710 University Avenue, Madison, WI 53726, USA

a r t i c l e

i n f o

Article history: Received 30 November 2010 Received in revised form 8 March 2011 Accepted 9 March 2011 Available online 22 March 2011 Keywords: Pyridoxal-5'-phosphate S-Adenosyl-L-methionine Adenosylcobalamin Lysine 2,3-aminomutase Glutamate 2,3-aminomutase Lysine 5,6-aminomutase Radical isomerization

a b s t r a c t PLP catalyzes the 1,2 shifts of amino groups in free radical-intermediates at the active sites of amino acid aminomutases. Free radical forms of the substrates are created upon H atom abstractions carried out by the 5′deoxyadenosyl radical. In most of these enzymes, the 5′-deoxyadenosyl radical is generated by an iron–sulfur cluster-mediated reductive cleavage of S-adenosyl-(S)-methionine. However, in lysine 5,6-aminomutase and ornithine 4,5-aminomutase, the radical is generated by homolytic cleavage of the cobalt–carbon bond of adenosylcobalamin. The imine linkages in the initial radical forms of the external aldimines undergo radical addition to form azacyclopropylcarbinyl radicals as central intermediates in the catalytic cycles. In the case of lysine 2,3-aminomutase, the multistep catalytic mechanism is corroborated by direct spectroscopic observation and characterization of a product radical trapped during steady-state turnover. Analogues of the substrate-related radical having substituents adjacent to the radical center to stabilize the unpaired electron are also observed and characterized spectroscopically. A functional allylic analogue of the 5′deoxyadenosyl radical is observed spectroscopically. A high-resolution crystal structure fully supports the mechanistic proposals. Evidence for a similar free radical mediated amino group transfer in the adenosylcobalamin-dependent lysine 5,6-aminomutase is provided by spectroscopic detection and characterization of radicals generated from the 4-thia analogues of the lysine substrates. This article is part of a Special Issue entitled: Pyridoxal Phospate Enzymology. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In the most widely-known, classical polar reaction mechanisms of PLP, amino groups of substrates react with PLP to form aldiminium intermediates, commonly known as Schiff bases, and the aldiminium group facilitates the formation of a carbanion in the substrate portion of the adduct. Carbanion formation arises either from the dissociation of a proton from the α-C―H bond or by the elimination of CO2 from the α-C―CO− 2 group. The driving force for proton dissociation or decarboxylation is the resonance delocalization of the carbanionic negative charge facilitated by the strategic presence of the aldiminium group between PLP and the substrate (Scheme 1). The carbanion undergoes the further reactions most typical of PLP-dependent enzymes, including racemization, transamination, β-elimination, βdecarboxylation, γ-elimination, and β-replacement reactions of amino acids and related molecules. These reactions are the subjects of most of the articles in this volume.

This review concerns the nonpolar, free radical reaction mechanisms of PLP-dependent amino acid aminomutases. These enzymes catalyze reversible transfers of amino groups between adjacent carbon atoms by free radical mechanisms [1,2]. The free radical intermediates lend these enzymes a unique place among PLPdependent enzymes. The currently-known PLP-dependent aminomutases are lysine 2,3-aminomutase, arginine 2,3-aminomutase, lysine 5,6-aminomutase (aka β-lysine mutase), ornithine 4,5-aminomutase, and glutamate 2,3-aminomutase. This review describes the molecular properties of these enzymes, their cofactor requirements, structures insofar as they are known, and the mechanism of intramolecular amino group transfer. The role of pyridoxamine 5′-phosphate (PMP) in one-electron steps of ascarylose biosynthesis [3] is covered in another article in this issue. 2. Lysine 2,3-aminomutase (LAM) 2.1. Discovery

☆ This article is part of a Special Issue entitled: Pyridoxal Phospate Enzymology. ⁎ Corresponding author. Tel.: +1 608 2620055; fax: +1 608 265 2904. E-mail address: [email protected] (P.A. Frey). 1570-9639/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2011.03.005

LAM in Clostridium subterminale SB4 and other anaerobic bacteria was the first aminomutase to be discovered. The enzyme was discovered and purified during investigations of lysine metabolism

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

1549

of EF-P. Thus, E. coli LAM (YjeK) seems to carry out the aminomutation required to generate the β-lysyl group of the spermidine moiety in posttranslationally modified Lys34. YjeK catalyzes aminomutation of free (S)-lysine, but it is not yet known whether it would work more efficiently on (S)-lysyl-tRNALys or on the terminal Lys in Nε-Lys-Lys34 of post-translationally modified EF-P. The extensive amino acid sequence homology between YjeK in E. coli and LAM in Cl. subterminale, and the lengths of the sequences—415 amino acids in Cl. subterminale LAM and 315 amino acids in YjeK, with no gaps—do not suggest the presence of specialized domains in YjeK that would presumably be necessary to bind a ribonucleic acid or a protein substrate [1]. Position 34 in EF-P corresponds to the location of the post-transcriptional residue hypusine [Nε-(4-amino-2-hydroxybutyl)-lysine] in eukaryotic translation initiation factor 5A (elF5A) [12,13]. 2.3. Role of PLP in the reaction mechanism Scheme 1. Generation of resonance-stabilized carbanions at C2 of external PLP-aldimines.

in Clostridia [4]. This new enzyme catalyzed the interconversion of (S)-lysine and (S)-β-lysine according to equation 1.

As originally described, the activity of the purified enzyme absolutely requires S-adenosyl-L-methionine (SAM), a strong reductant such as dithionite, and strictly anaerobic conditions. As originally prepared, LAM contains PLP and iron, and the activity is modestly increased by the addition of PLP or ferrous iron. The reaction of Clostridial LAM proceeds with the stereochemistry shown in equation 1, wherein the 3-pro-R hydrogen in (S)-lysine migrates to the 2-pro-R position in (S)-β-lysine, accompanied by the cross migration of the amino group from C2 to C3 [5]. The reaction proceeds with overall inversion of configuration at C2 and C3. Unlike PLP-dependent enzymes that involve carbanionic intermediates, hydrogen transfer by LAM does not involve exchange of C2(H) or of C3(H)in the substrate with solvent protons [4]. 2.2. Biological significance of LAM The classical LAM is often found in anaerobic bacteria that utilize (S)-lysine for growth as a source of nitrogen or/and carbon. In biosynthetic contexts, LAM and arginine 2,3-aminomutase produce (S)-β-lysine and β-arginine, respectively, for inclusion into a group of antibiotics that contain β-aminoacyl substituents. β-Lysyl groups are substituents in streptothricin F and capreomycin IB [6,7]. Myomycin has a β-lysyl dipeptide as the β-aminoacyl substituent [8]. Blastcidin S contains a β-arginyl group [9]. Some bacterial LAMs produce β-lysine having the (R)-configuration as opposed to the (S)-configuration of the more classical LAMs. For example, Escherichia coli contain the gene yjeK, which encodes a variant of the classical LAM. This yjeK gene product, YjeK has the same cofactor requirements as the classical LAMs, but the amino group transfer has a different stereochemical outcome. YjeK from E. coli transforms (S)-lysine into (R)-β-lysine at a rate that is about 0.1% that of aminomutation catalyzed by LAM from Cl. subterminale [10]. YjeK is postulated to be involved in the post-translational modification of the strictly-conserved bacterial elongation factor P (EF-P), which contains a spermidine residue at amino acid position 34, [11]. Genetic evidence implicates two genes in post-translational modification of Lys34, yjeA and yjeK. YjeA appears to be a lysyl-tRNA synthetase, and it is postulated to function with YjeK to link the (R)-βlysyl residue to the ε-amino group of Lys34 within the primary transcript

A viable chemical mechanism involving a carbanion-stabilizing role of PLP in the interconversion of (S)-lysine and (S)-β-lysine by LAM cannot be written. Moreover, the overall reaction pattern is that of adenosylcobalamin-dependent reactions, in which a hydrogen atom and a functional group or carbon fragment exchange positions between adjacent carbons [2]. LAM does not contain or require a cobalamin coenzyme; however, the adenosyl moiety of SAM mediates hydrogen transfer in the reaction of LAM, in strict analogy with the well-established function of the adenosyl moiety in adenosylcobalamin [14,15]. Moreover, 5′-deoxyadenosine derived from SAM is present as an intermediate during the course of the reaction [16]. 5′Deoxyadenosine is also the hydrogen transferring intermediate in adenosylcobamamin-dependent enzymatic reactions [2]. These facts inspire the chemical mechanism in Fig. 1, in which PLP reacts initially in the fashion typical of PLP-enzymes to form the external aldimine with (S)-lysine [14]. However, in a nontraditional role, PLP facilitates a radical isomerization step and unlocks the 2,3-nitrogen migration. In this mechanism, the adenosyl moiety of SAM functions exactly as the adenosyl moiety in adenosylcobalamin in mediating hydrogen transfer. The mechanism postulates a reductive cleavage of SAM to generate the 5′-deoxyadenosyl-radical (5′-deoxyadenosine-5′-yl), which abstracts the 3-pro-R hydrogen from the lysyl-side chain to form the PLP-2-aldiminolysine-3-yl-radical 1 and 5′-deoxyadenosine. The substrate-related radical 1 undergoes rearrangement to the productrelated PLP-3-aldiminolysine-2-yl-radical 3, by way of the intermediate formation of PLP-azacyclopropylcarbinyl radical 2. Radical 2 arises through reversible radical addition to the imine group of the external aldimine, which involves unpairing the π-electrons of the imine group. One of the π-electrons forms the third σ-bond in the cyclopropyl ring by pairing with the unpaired electron in the lysyl side chain. This leaves the remaining π-electron from the imine at C4′ of PLP, in perfect position to be stabilized by resonance delocalization throughout the pyridine ring. While this role of PLP in aminomutation is non-traditional, PLP is chemically very well suited for this function. The the π-electron system of the pyridine ring can readily stabilize radical intermediate 2 in Fig. 1 by delocalizing the unpaired electron. The imine groups in either of the external PLP-aldiminolysine-yl radicals are also well suited to accept radical addition to form the PLP-azacyclopropylcarbinyl radical 2 in Fig. 1. Chemical precedent for the isomerization of radical intermediate 2 is available in the direct EPR spectroscopic observation of the parent radicals in equation 2 [17].

In a nonenzymatic counterpart of the PLP-dependent aminomutase reactions, benzaldehyde plays the role of PLP [18]. In a general

1550

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

Fig. 1. A chemical mechanism for the reaction of LAM. In this mechanism, (S)-Lysine is bound as the external PLP-aldimine by transaldimination with the internal Lys337-aldimine of PLP. SAM is assumed to be cleaved to generate the 5′-deoxyadenosyl radical in a reversible process, and the resultant radical abstracts the 3-pro-R hydrogen from the lysyl side chain to generate the substrate-related radical 1 and 5′-deoxyadenosine, which remains bound to the enzyme. Radical 1 undergoes isomerization to the azacyclopropylcarbinyl radical 2 and then to the (S)-β-lysine-related radical 3. Hydrogen abstraction from 5′-deoxyadenosine generates the external PLP-aldimine of (S)-β-lysine and regenerates SAM. Transaldimination with Lys337 releases the product and regenerates the internal Lys337-aldimine of PLP.

method for alkyl radical formation, the trimethyltin radical abstracts bromine from alkyl bromides to form alkyl radicals. In an application of this method, the trimethyltin radical initiates the radical isomerization of 2-methyl-3-bromo-benzaldiminoalanine ethyl ester, as illustrated in Fig. 2. The resultant C3-radical generated from 2methyl-3-bromo-benzaldiminoalanine ethyl ester suffers two fates. Direct quenching by trimethyltin hydride, the precursor of the

trimethyltin radical, leads to 2-methyl-benzaldiminoalanine ethyl ester. Alternatively, isomerization of the initial C3-radical to the C2radical in Fig. 2, followed by quenching with trimethyltin hydride, leads to 2-methyl-benzaldimino-β-alanine ethyl ester. Under the reported conditions, product analysis reveals a 13:1 ratio of β-alanyl:alanyl products [18]. This ratio corresponds to the values of equilibrium constants for the reactions of LAM and glutamate 2,3-aminomutase

Fig. 2. A chemical counterpart of the radical isomerization in Fig. 1. Heating (Δ) azobisisobutyronitrile (AIBN) with trimethyltin hydride generates trimethyltin radicals, which abstract bromine from 2-methyl-3-bromo-benzaldiminoalanine ethyl ester to generate an analogue of radical 1 in Fig. 1 [18]. The radical undergoes isomerization to the corresponding analogue of radical 3 in Fig. 1. Quenching of the radical mixture by tributyltin hydride generates the product mixture corresponding to the 1,2-migration of the imino group.

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

[19,20]. The formation of the dominant β-alanyl product validates the chemical propensity of radicals such as 1 and 3 in Fig. 1 to undergo the proposed isomerization. The direct observation of ring-opening in an azacyclopropylcarbinyl radical [17] implicates radical 2 in Fig. 1 as the most likely intermediate. The possibility that radical 2 is a transition state as opposed to an intermediate has been discussed as an alternative mechanism [21]. This type of mechanism has been explored for the presumed 1,2-migration of the aminium moiety of the ethanolamine radical in the adenosylcobalamin-dependent ethanolamine ammonia-lyase [22,23]. A role for PLP in such a direct mechanism for LAM is not clear. DFT calculations of energy profiles of models of the reaction of LAM indicate that the pyridine ring lends significant stability to the azacyclopropylcarbinyl moiety [24]. The spectroscopic evidence for discrete azacyclopropylcarbinyl radicals [17] reinforces the notion that these species are sufficiently long-lived to be intermediates. 2.4. Identification and characterization of radical intermediates In the validation of multistep chemical reaction mechanisms, the discrete intermediates are identified and rigorously characterized as the first order of procedure. In the reaction of LAM one radical intermediate has been observed by freeze-quench EPR in the steady state of the reaction. Isotope-edited EPR spectroscopic experiments led directly to the structure of radical 3 in Fig. 1, the product-related radical, as the predominant radical intermediate [25–27]. Thus, effects on the EPR signal upon substitution of (S)-[2-2H]lysine or (S)-[2-13C]lysine for (S)-lysine showed that the unpaired electron resided primarily in a p-orbital at C2 of the lysyl side chain. Analysis of spectra resulting from the use of perdeutero-(S)-lysine, (S)-[3,3,4,4,5,5,6,6-2H8]lysine, (S)-[α-15N]lysine and (S)-[α-15N,2-2H]lysine showed that the hyperfine structure of the signal could be accounted for by the splittings brought about by the α-hydrogen, the β-nitrogen, and the β-hydrogen in radical 3 of Fig. 1. The results allowed the conformation of the side chain in radical 3 to be assigned [26]. PLP was implicated in the structure of radical 3 by the results of 2H-ESEEM (electron spin echo envelope modulation) spectroscopy employing [4′-2H]PLP as the coenzyme. The magnitude of the dipolar hyperfine coupling between the unpaired electron at C2 of lysine and deuterium at C4′ of PLP corresponded to a distance of b 3.5 Å between the electron and nuclear spins [27]. Rapidmix-freeze-quench EPR experiments and lineshape analysis showed radical 3 to be kinetically competent to be an intermediate in catalysis [28]. The fact that in radical 3 there is stabilization of the unpaired electron through delocalization into the π-system of the adjacent carboxylate group accounts for the stability of this radical relative to the other radical in Fig. 1. This delocalization mode was confirmed by the observation of 13C hyperfine splitting in EPR spectra obtained employing (S)-[1-13C]lysine [29]. The unpaired electrons in radical 1 and the 5′-deoxyadenosyl radical would not be delocalized by resonance and so would be too high in energy for these intermediates to be sufficiently populated in the steady-state to be detected by EPR. The unpaired electron in radical 2 should be delocalized into the pyridine ring of PLP, and this would be a stabilizing effect. However, the azacyclopropyl ring in radical 2 incorporates bond angle strain, which destabilizes it, and this radical has never been observed by EPR. Evidence supporting radical 1 and the 5′-deoxyadenosyl radical as intermediates is available through the observation of EPR spectra of chemically related species in the reactions of closely related analogs of (S)-lysine and SAM. 4-Thia-(S)-lysine is a substrate for LAM, and the one radical intermediate that is observable by EPR in the steady-state is the 4-thia-analogue of radical 1 in Fig. 1, the chemical structure of which is shown in Fig. 3 [30,31]. The aminomutation product in the reaction of 4-thia-(S)-lysine is chemically labile and decomposes to βmercaptoethylamine, ammonia, and formylacetate (malonate semi-

1551

Fig. 3. Structures of analogues of the substrate-related radical 1 in Fig. 1. At the left is the structure of the 4-thia-analogue of radical 1 in Fig. 1, which is generated by employing 4-thia-(S)-lysine as a substrate in place of (S)-lysine. At the right is the allylic analogue of radical 1 generated by employing trans-4,5-dehydro-(S)-lysine as a suicide inactivator in place of the substrate (S)-lysine.

aldehyde). However, isotope edited EPR experiments analogous to those employed in the characterization of radical 3 in the reaction of (S)-lysine definitively identify the 4-thia-analogue of radical 1 as an intermediate. The 4-thia-group stabilizes the unpaired electron at C3 through delocalization by electron exchange with the nonbonding electrons in the p-orbitals of sulfur [32]. The slow forward reaction of the 4-thia-analogue of radical 1 and its relative stability allow it to accumulate and be observed by EPR spectroscopy. 4,5-Dehydro-(S)-lysine also reacts with LAM but as a suicide inactivator rather than as a true substrate [33]. The double bond in the side chain of 4,5-dehydro-(S)-lysine is in position to stabilize an unpaired electron at C3 through delocalization as an allylic radical. The molecule reacts with LAM through the first two chemical steps in Fig. 1 as if it were a substrate. However, upon hydrogen transfer from C3 of the suicide inactivator to the 5′-deoxyadenosyl radical, the 4,5dehydro analogue of the substrate related radical 1 is formed, and the reaction comes to rest. The structure of the 4,5-dehydro-analogue of radical 1, shown in Fig. 3, is an allylic radical, and it is too stable to react further in either the forward or reverse direction. Consequently, 4,5-dehydro-(S)-lysine is a suicide inactivator of LAM. Nonetheless, the formation of the allylic analogue of radical 1 in Fig. 1 supports the mechanism. The unpaired electron in the 5′-deoxyadenosyl radical in Fig. 1 resides, without resonance delocalization, on C5′ and so is too high in energy to be populated at an EPR-detectable concentration in the steadystate. However, S-3′,4′-anhydro-5′-deoxyadenosyl-(S)-methionine (anSAM) functions as a true coenzyme in place of SAM [34,35]. anSAM functions at a much slower rate than SAM but does not inactivate LAM. EPR spectra of reaction mixtures freeze-quenched in the steady-state show a single radical, which by isotope-editing is clearly the 3′,4′anhydro-analogue of the 5′-deoxyadenosyl radical. The chemical structures of anSAM and the 3′,4′-anhydro-5′-deoxyadenosyl radical are shown in Fig. 4. The analogue radical is kinetically competent as an intermediate in the anSAM-activated reaction of LAM. With the identification of radical 3 as an intermediate, and of chemical analogues of radical 1 and the 5′-deoxyadenosyl radical in reactions of 4-thia-(S)-lysine and anSAM with LAM, the chemical mechanism in Fig. 1 is corroborated. The identification of Lys337 as the point of linkage between PLP and LAM in the internal aldimine is based on the biochemical identification of the homologous residue in LAM from Bacillus subtilis [36]. Reduction of the internal PLP-aldimine in LAM from B. subtilis by NaBH4 inactivates the enzyme. Based on the

Fig. 4. Chemical formulas of anSAM and the 3′,4′-anhydro-5′-deoxyadenosyl radical.

1552

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

analysis of peptide fragments by mass spectrometry, Lys346 is the residue linked to PLP. Based on the amino acid sequence of LAM from Cl. subterminale [37] aligned with that for the enzyme from B. subtilis, Lys346 in B. subtilis corresponds to Lys337 in Cl. subterminale. 2.5. Radical initiation by SAM Details of the identification of SAM in radical initiation for the reaction of LAM are beyond the scope of this review. SAM undergoes reversible reductive cleavage to the 5′-deoxyadenosyl radical and methionine by the mechanism shown in Fig. 5. LAM contains a fouriron, four-sulfide cluster that is labile to molecular oxygen and that must be reduced to the state [4Fe–4S]1+ for the enzyme to display activity [38,39]. For these reasons, LAM is extremely sensitive to oxidation by air and is active under anaerobic conditions. SAM is directly ligated through the amino and carboxylate groups of the methionyl moiety to one of the four irons in the cluster [40]. Seadenosyl-seleno-(S)-methionine (SeSAM) is nearly as effective as SAM in activating LAM. When SeSAM activates the enzyme, reductive cleavage results in selenium becoming directly ligated to iron in the iron–sulfur cluster [41]. These facts lead directly to the mechanism in Fig. 5 for the formation of the 5′-deoxyadenosyl radical to initiate the chemical mechanism in Fig. 1. During the 1980s and 1990s, four other non-methyltransferases were discovered to require SAM and anaerobic conditions. These SAM-dependent enzymes were the activating enzymes for pyruvate formate lyase and class III ribonucleotide reductase, biotin synthase, and lipoyl synthase. These enzymes and LAM were found to share the cysteine motif CxxxCxxC associated with iron–sulfur centers having just three cysteine-ligands from the protein. At the turn of the 21st century, a search of bacterial genomes uncovered nearly 600 proteins with this motif [42]. This group of proteins was named the Radical SAM superfamily. By 2008, the superfamily had grown to more than 2800 members [1]. The members of this superfamily have been found to be required in many diverse biological functions, including metabolism; DNA repair; biosyntheses of DNA, vitamins, antibiotics, pyrroloquinoline quinone, quinoprotein cofactors and metal centers in enzymes; and post-translational chemical modifications of enzymes and nucleic acids—all in all, more than 40 distinct biological processes for Radical SAM enzymes have been uncovered by 2008, and the number is growing [1]. The detailed chemistry in most of these radical-based processes has not as yet been uncovered. 2.6. Molecular structure of LAM The crystal structure of LAM with SAM and (S)-lysine bound at the active site and with the [4Fe–4S] center in the (2+) state is illustrated in Fig. 6 [43]. The functional unit appears to be a domain-swapped dimer (Fig. 6A). The global structure of each subunit contains a 3/4-βbarrel, which is recognized as typical but not universal among Radical SAM enzymes, and the active site is housed within the barrel. The PLP binding site with contacts about N1 of the pyridine ring are shown in Fig. 6B. This site is unusual among PLP-dependent enzymes.

The structure shows a fixed water molecule hydrogen bonded to N1 and held in place by three main chain hydrogen bonds. It is very likely that N1 is neutral and unprotonated. It appears that LAM does not make use of enhanced stabilization for radical 3 that is predicted to accompany protonation of N1 in the pyridine ring [24]. Fig. 6C shows molecular details in the binding of the PLP-lysine aldimine, SAM, and the [4Fe–4S] center. In the external aldimine Lys337 is released and does not appear to interact directly with the lysyl side chain. Lysine is held in place by its aldimine linkage to PLP, by an ionic interaction between the α-carboxylate and Arg134, and by ionic interactions between the ε-amino group and both Asp293 and Asp330. Carbon-3 of the lysyl side chain is 3.8 Å from C5′-of the adenosyl moiety in SAM, in excellent position for hydrogen abstraction upon reductive cleavage of SAM. Finally, the coordination of SAM through the amino and carboxylate groups of the methionyl moiety to the unique iron in the [4Fe–4S] center is confirmed by the structure. The structure elegantly supports all other data regarding the mechanistic functions of the cofactors of LAM, including the reductive cleavage of SAM by the mechanism in Fig. 5. 2.7. Other radical SAM aminomutases A gene encoding a homologue of LAM, while lacking Asp293 and Asp330 for binding the ε-amino group, appears in several bacterial genomes. This gene encodes an asparagine and a lysine residue in place of Asp293 and Asp330 of LAM. Expression of this gene from Cl. difficile in E. coli produces a protein similar to LAM except that it efficiently functions as a glutamate 2,3-aminomutase [20]. Presumably the substitution of asparagine and lysine for Asp293 and Asp330 in LAM allows the Cl. difficile homologue to bind (S)-glutamate and function by the same PLP and SAM-dependent mechanism as LAM. A radical intermediate appears at the steady-state, and isotope-edited EPR spectroscopy indicates that it is the glutamyl analogue of radical 3 in Fig. 1. Streptomyces griseochromogenes produce Blasticidin S, which contains β-arginine as a β-aminoacyl substituent. The gene cluster for the biosynthesis of this antibiotic includes a gene encoding a homologue LAM [9]. The enzyme encoded has been assigned as arginine 2,3-aminomutase. 3. Lysine 5,6-aminomutase 3.1. Discovery In their studies of lysine metabolism in Clostridium sticklandii, Dr. Thressa Stadtman and her associates discovered a PLP and adenosylcobalamin-dependent enzyme that catalyzed the interconversion of (S)-β-lysine and 3,5-diaminohexanoate [44–47]. For a time this enzyme was known as β-lysine mutase. This group also discovered (R)-lysine mutase that produced 2,5-diaminohexanoate and functioned less efficiently on (S)-lysine. It subsequently became clear that β-lysine mutase and (R)-lysine mutase were the same enzyme, now known as lysine 5,6-aminomutase. The purified enzyme was characterized as a cobalamin protein composed of distinct subunits designated α and β.

Fig. 5. A chemical mechanism for the reductive cleavage of SAM.

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

1553

Fig. 6. The molecular structure of LAM. Part A. A ribbon diagram of the global x-ray crystallographic structure of Cl. Subterminale LAM [43]. Part B. The enzyme-PLP contacts in the structure of LAM with (S)-lysine at the active site, showing the fixed water molecule hydrogen bonded to N1 of PLP. Part C. The active site showing the structural relationships among the external PLP-lysyl aldimine, SAM, and the [4Fe–4S]-cluster. Adapted from Figs. 2, 3, and 4 of reference [43] and reproduced with permission of the Proceedings of the National Academy of Sciences.

Research in this field was hampered by the presence of degraded forms of cobalamin tightly bound to the enzyme purified from Clostridia. This problem was overcome by cloning the genes encoding the α- and βsubunits in Cl. sticklandii and co-expressing them in E. coli, which does not carry out de novo biosynthesis of cobalamins [48]. The nearly identical genes were also cloned from Porphyromonas gingivalis and similarly expressed [49]. Purified from E. coli, the enzyme contained PLP but, as expected, no cobalamins. Upon activation with adenosylcobalamin the protein displayed enhanced activity relative to that purified from Clostridia. Of the three known substrates for the Porphyromonas enzyme, the best one proved to be (R)-lysine (kcat/Km =6.3 ×102 M− 1 s− 1),

followed by (S)-β-lysine (kcat/Km =4.8× 102 M− 1 s− 1) and (S)-lysine (kcat/Km = 0.43 M− 1 s− 1) [50]. 3.2. Mechanistic characterization The working hypothesis for the mechanism of the 5,6-aminomutase reaction is similar to that for LAM, substituting adenosylcobalamin for SAM as the source of the 5′-deoxyadenosyl radical and the ε-amino group of the substrate forming the external aldimine with PLP and undergoing migration between C6 and C5, as shown in Fig. 7. Reaction of (S)-β-lysine produces (3S,5S)-diaminohxanoate [51]. Biochemical

1554

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

Fig. 7. A chemical mechanism for the reaction of lysine 5,6-aminomutase. In this mechanism, (R)-Lysine is bound as the external PLP-aldimine by transaldimination with the internal Lys144β-aldimine of PLP. Adenosylcobalamin is assumed to be cleaved to cob(II)alamin [Cbl(II)] and the 5′-deoxyadenosyl radical in a reversible process, and the resultant radical abstracts a hydrogen from C5 of the lysyl side chain to generate the substrate-related radical 4, and 5′-deoxyadenosine and Cbl(II) remain bound to the enzyme. Radical 4 undergoes isomerization to the azacyclopropylcarbinyl radical 5 and then to the product-related radical 6. Hydrogen abstraction from 5′-deoxyadenosine generates the external PLP-aldimine of 2,5-diaminohexanoate and regenerates adenosylcobalamin. Transaldimination with Lys144β releases the product and regenerates the internal aldimine.

analysis revealed a novel PLP binding site, with PLP bound to Lys144β [49]. Participation of the 5′-deoxyadenosyl moiety of adenosylcobalamin in mediating hydrogen transfer is known from the early work of Rétey and coworkers [52]. The radical isomerization is well precedented in the reaction of LAM. Moreover, high-level ab initio theoretical calculations show that other mechanisms would require much higher energy barriers between steps [24]. Until very recently, however, no direct experimental evidence supporting this mechanism has been obtained. Unlike LAM and most adenosylcobalamin-dependent enzymes, no paramagnetic species, neither free radicals, such as 4, 5, or 6, nor cob(II)alamin shown in Fig. 7 could be detected by EPR spectroscopy in the steady-state. Nor could the characteristic absorption spectrum of cob(II)alamin be observed in the steady-state. Consideration of the chemical nature of the radicals in Fig. 7 can account for the absence of spectroscopic evidence for their presence in the reaction. Radicals 4 and 6 are primary and secondary carbon-based radicals having no adjacent functional groups that could stabilize the unpaired electrons by delocalization. Having only relatively weak stabilization from hyperconjugation with β-substituents, these radicals lack sufficient stability for their concentrations to rise to levels detectable by EPR spectroscopy under conditions compatible with enzymatic function. The azacyclopropylcarbinyl radical 5 allows for delocalization of the unpaired electron into the pyridine ring of PLP; however, like the corresponding radical 2 in the reaction of LAM, steric strain in the azacyclopropyl group elevates the free energy of this species to a level that it is not sufficiently populated in the steady state to be detected by spectroscopy. While these considerations provide a logical explanation of the absence of detectable radical intermediates in the mechanism of Fig. 7, they do not provide support for the mechanism. Evidence for the substrate related intermediate 4 in Fig. 7 is available in recent experiments with substrate analogues, the (S)- and (R)enantiomers of 4-thialysine [53,54]. Both are reversible suicide inhibitors, but not irreversible inactivators, of lysine 5,6-aminomutase.

Both analogues induce the transformation of adenosylcobalamin into cob(II)alamin, as detected by absorption spectrophotometry. Both compounds also induce the formation of 4-thialysine-based free radicals at the active site, as detected by EPR spectroscopy. Neither analogue undergoes complete turnover into detectable products. Instead, both react to form 4-thia-analogues of radical 4 in Fig. 7, and these radicals are diverted from turnover by reversible isomerization into dead-end radicals that are stabilized by the captodative effect. The chemical transformations are shown in Fig. 8. Just as the 4-thia group stabilizes the unpaired electron at C3 in the reaction of LAM, the substrate-related radical in Fig. 1, it also stabilizes the unpaired electron at C5 in the reaction of lysine 5,6-aminomutase. This too is the substrate-related radical for the action of lysine 5,6aminomutase by the mechanism in Fig. 8. In the case of lysine 5,6aminomutase, the active site is known to catalyze the exchange of C6 (H) with solvent protons [46] in a process that must be catalyzed by a base at the active site. This base might be Lys144β, but in any case its presence opens the possibility of proton-abstraction of C6(H) from the initially-formed radical in Fig. 8, followed by transfer to C4′ of PLP. This isomerization would further stabilize the radical by delocalizing the unpaired electron further to include C6 and Nε, as shown in Fig. 8. In this way, captodative stabilization of the unpaired electron is brought into play, with Nε serving as the electron-accepting captoentity, and 4S serving as the electron-donating dative-entity. The result being greater stabilization than in the initial, substrate-related radical stabilized only by the dative-effect of 4S. Radical formation by the enantiomers of 4-thialysine to the initial radicals with the unpaired electron at C5–4S supports the mechanism in Fig. 8 for the reaction of lysine 5,6-aminomutase. Further isomerization of the initial radical to the captodative species in Fig. 8 is verified by the observation of nuclear hyperfine β-coupling between PLP-C4′(H) and the unpaired electron [54], as well as by UV/Vis spectrophotometry and ab initio calculations on the relative stabilities of candidate radicals [53].

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

1555

zation. Binding of the phosphate group appears very strong, with two hydrogen bonded ionic bonds and three or four additional hydrogen bonds to main chain N–H groups. In a surprising feature of the structure, the PLP-Lys144β internal aldimine lies about 25 Å away from the cobalamin macrocycle. Presumably, upon formation of the external PLP-lysine aldimine the structure undergoes a conformational change that brings the lysyl side chain into contact with the 5′-deoxyadenosyl radical and stimulates cleavage of the cobalt–carbon bond. A subunit rotational model appears to allow for such a conformational change [55]. 3.4. Suicide inactivation

Fig. 8. Radical reactions of 4-thialysine with lysine 5,6-aminomutase. Both 4-thia-(R)-lysine and 4-thia-(S)-lysine react with the complex of PLP, adenosylcobalamin, and lysine 5,6-aminomutase to generate cob(II)alamin, 5′-deoxyadenosine, and 4-thialysine-based free radicals [53,54]. The reactions proceed to equilibria leading to about 80% inhibition of lysine 5,6-aminomutase at saturating 4-thialysine. In analogy to the reactions of (S)- and (R)-lysine, 4-thia-(R)-lysine reacts more rapidly than the (S)-enantiomer. Isotope-edited EPR spectroscopy indicates that the first radical formed within a few seconds is analogous to the substrate-related radical 4 in Fig. 7 [53]. Isotope-edited EPR and spectrophotometric analysis indicates that the isomerization of the initial radical over a period of ten minutes does not lead to analogues of any of the other radicals in Fig. 7. All evidence points to isomerization through proton transfer from C6 of 4-thialysine to C4′ of PLP. This results in delocalization of the unpaired electron among C5, C6, Nε, and 4S of 4-thialysine.

The arrangement of PLP and adenosylcobalamin binding sites in the internal aldimine is presented as a locking mechanism preventing adventitious cleavage of the cobalt–carbon bond, with the potential complications of exposing the enzyme to the destrictive effects of radical side reactions in the absence of the substrate. In this hypothesis, cobalt– carbon cleavage occurs only when the substrate is present to form the external aldimine and release the cross-link to Lys144β, allowing the conformational change. Then, upon cleavage of the cobalt–carbon bond the 5′-deoxyadenosyl radical would be in position to contact the lysyl side chain and initiate the mechanism by hydrogen abstraction from C5 as proposed in Fig. 8. While the foregoing hypothesis might be correct, the model does not protect the enzyme from suicide inactivation. In fact, upon reaction with any of the substrates lysine 5,6-aminomutase rapidly loses activity. Unlike the widely studied suicide inactivation of PLP-dependent enzymes by substrate analogues, this process is induced by the presumed natural substrates. Substrate-induced inactivation leads to the formation of cob(III)alamin, with no sign of cob(II)alamin as an intermediate, even under strictly anaerobic conditions [56]. The only products detected in the inactive enzyme are cob(III)alamin, 5′deoxyadenosine, and the substrate and product, 2,5-diaminohexanoate. However, conducting the reaction in 3H2O produces [3H]lysine and 2,5[3H]diaminohexanoate. Substrate-induced inactivation is explained as resulting from electron transfer from cob(II)alamin to radicals 4 and/or 6 in the mechanism of Fig. 7. Electron transfer explains the formation of cob(III)alamin, and quenching of the resultant lysyl carbanions by protonation explains the observation of tritiated substrate and product in 3H2O. Interestingly, electron transfer from cob(II)alamin to the 4thialysyl radicals in Fig. 8 does not occur. It appears that the 4-thialysyl radicals are not strong enough oxidizing agents to accept an electron from cob(II)alamin [53]. 3.5. PLP-independent aminomutases

3.3. Molecular structure A crystal structure of a complex of PLP and adenosylcobalamin with lysine 5,6-aminomutase is shown in Fig. 9 [55]. The structure shows the α-subunit as a beta- or TIM-barrel with PLP embedded and linked to the β-subunit through the aldimine bond to Lys144β. The β-subunit incorporates a Rossmann fold hosting the 5,6-dimethylbenzimidazolyl tail of cobalamin, with the macrocycle lying at the surface within a cleft between the subunits. The buried 5,6-dimethylbenzimidazolyl tail sustains the base-off binding mode for adenosylcobalamin assigned by EPR [48]. The structure shows adenosylcobalamin cleaved to cobalamin and 5′-deoxyadenosine. The weak cobalt–carbon bonds generally do not survive x-ray crystallographic analysis, so that the adenosyl moiety generally appears as 5′-deoxyadenosine. Interesting features of the PLP-binding site include the contact of PLP-N1 with Ser238α and the ionic and hydrogen bonded contacts of the phosphate group with Arg268α and Arg184α, as well as several main chain hydrogen bonds. The contact with Ser128α is undoubtedly nonionic, as is the N1 contact in LAM. This reinforces the idea that a pyridinium-type electron sink is not required for radical isomeri-

Tyrosine 2,3-aminomutase and phenylalanine 2,3- aminomutase do not require PLP, SAM or adenosylcobalamin [57,58]. These enzymes catalyze aminomutation by a polar, elimination/addition mechanism, in which a β-proton and the α-amino group are eliminated from the substrate to form ammonia and p-hydroxycinnamate or cinnamate, respectively, as intermediates. Ammonia is then added back to the β-carbon in the double bond of p-hydroxycinnamate or cinnamate, respectively, to form β-tyrosine or β-phenylalanine. The PLP-dependent radical mechanism is avoided in these cases because of the polar reactivities of the β-hydrogens in side chains of tyrosine and phenylalanine. The β-hydrogens in these amino acids are benzylic and therefore weakly acidic and subject to removal as protons in polar elimination reactions. Elimination of ammonia from tyrosine or phenylalanine is chemically similar or identical to the elimination of ammonia by phenylalanine ammonia-lyase and histidine ammonia-lyase. These latter enzymes employ the amino acid-derived cofactor 4-methylideneimidazole-5-one (MIO) moiety within their amino acid sequences [59]. This cofactor facilitates the elimination of ammonia. Tyrosine 2,3-aminomutase and

1556

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557

Fig. 9. X-ray structure of lysine 5,6-aminomutase Part A is a ribbon diagram of the overall structure of the α/β dimeric unit. The α-subunit is a β-barrel (TIM barrel) embedding PLP in its binding site. The β-subunit includes a Rossmann fold embedding the dimethylbenzimidazolyl tail of cobalamin. The cobalamin macrocycle lies at the surface of the β-subunit. Part B is a space-filling close-up, showing the spatial relationships of PLP, cobalamin, and 5′-deoxyadenosine. Part C shows a stereodiagram of the contacts between PLP, residues in the α-subunit, and the crosslink to Lys144 of the β-subunit. Adapted from Figs. 2 and 3 of reference [55] and reproduced with permission of the Proceedings of the National Academy of Sciences.

phenylalanine 2,3-aminomutase have the same MIO-cofactor in their structures to catalyze ammonia release, but they also catalyze the readdition of ammonia to form the β-amino acids. References [1] P.A. Frey, A.D. Hegeman, F.J. Ruzicka, The radical SAM superfamily, Crit. Rev. Biochem. Mol. Biol. 43 (2008) 63–88. [2] P.A. Frey, Cobalamin coenzymes in enzymology, in: L. Mander, H.-W. Liu (Eds.), Comprehensive Natural ProductsII Chemistry and Biology, 7, Elsevier: Oxford, 2010, pp. 501–546. [3] C.-W.T. Chang, D.A. Johnson, V. Bandarian, H. Zhou, R. LoBrutto, G.H. Reed, H.-W. Liu, Characterization of a unique coenzyme B6 radical in the ascarylose biosynthetic pathway, J. Am. Chem. Soc. 122 (2000) 4239–4240. [4] T.P. Chirpich, V. Zappia, R.N. Costilow, H.A. Barker, Lysine 2,3-aminomutase: purification and properties of a pyridoxal phosphate and S-adenosylmethionineactivated enzyme, J. Biol. Chem. 245 (1970) 1778–1789. [5] D.J. Aberhart, S.J. Gould, H.J. Lin, T.K. Thiruvengadam, B.H. Weiller, J. Am. Chem. Soc. 105 (1983) 5461–5470. [6] T.K. Thiruvengadam, S.J. Gould, D.J. Aberhart, H.-J. Lin, Biosynthesis of Streptothricin F. 5. Formation of β-lysine by Streptomyces L-1689-23, J. Am. Chem. Soc. 105 (1983) 5470–5476. [7] D.E. DeMong, R.M. Williams, Asymmetric synthesis of (2S,3R)-capreomycidine and the total synthesis of capreomycin IB, J. Am. Chem. Soc. 125 (2003) 8561–8565; G. Agnihotri, H.-W. Liu, PLP and PMP radicals: a new paradigm in coenzyme B6 chemistry, Bioorg. Chem. 29 (2001) 234–257. [8] J. Davies, M. Cannon, M.B. Mauer, Myomycin: mode of action and mechanism of resistance, J. Antibiot. (Tokyo) 41 (1988) 366–372. [9] M.C. Cone, X. Yin, L.L. Grochowski, M.R. Parker, T.M. Zabriskie, The Blasticidin S biosynthesis gene cluster from Streptomyces griseochromogenes: sequence analysis, organization, and initial characterization, Chembiochem 4 (2003) 821–828. [10] E. Behshad, F.J. Ruzicka, S.O. Mansoorabadi, D. Chen, G.H. Reed, P.A. Frey, Enantiomeric free radicals and enzymatic control of stereochemistry in a radical mechanism: the case of lysine 2,3-aminomutases, Biochemistry 45 (2006) 12639–12646. [11] M. Bailly, V. de Crécy-Lagard, Predicting the pathway involved in posttranslational modification of elongation factor P in a subset of bacterial species, Biol. Direct 5 (2010) 3. [12] H.L. Cooper, M.H. Park, J.E. Folk, B. Safer, R. Braverman, Identification of the hypusine-containing protein hy+ as translation initiation factor eIF-4D, Proc. Natl Acad. Sci. U.S.A. 80 (1983) 1854–1857. [13] M.H. Park, The post-translational synthesis of a polyamine-derived amino acid, hypusine, in the eukaryotic translation initiation factor 5A (eIF5A), J. Biochem. 139 (2006) 161–169.

[14] M. Moss, P.A. Frey, The role of S-adenosylmethionine in the lysine 2,3aminomutase reaction, J. Biol. Chem. 262 (1987) 14859–14862. [15] J. Baraniak, M.L. Moss, P.A. Frey, Lysine 2,3-aminomutase. Support for a mechanism of hydrogen transfer involving S-adenosylmethionine, J. Biol. Chem. 264 (1989) 1357–1360. [16] M.L. Moss, P.A. Frey, Activation of lysine 2,3-aminomutase by S-adenosylmethionine, J. Biol. Chem. 265 (1990) 18112–18115. [17] W.C. Danen, C.T. West, An electron spin resonance investigation of the aziridylcarbinyl and related free radicals, J. Am. Chem. Soc. 96 (1974) 2447–2453. [18] O. Han, P.A. Frey, A chemical model for the pyridoxal-5-phosphate dependent lysine aminomutases, J. Am. Chem. Soc. 112 (1990) 8982–8983. [19] D. Chen, J. Tanem, P.A. Frey, Basis for the equilibrium constant in the interconversion of L-lysine and L-beta-lysine by lysine 2,3-aminomutase, Biochim. Biophys. Acta 1774 (2007) 297–302. [20] F.J. Ruzicka, P.A. Frey, Glutamate 2,3-aminomutase: a new member of the radical SAM superfamily of enzymes, Biochim. Biophys. Acta 1774 (2007) 286–296. [21] G. Agnihotri, H.-W. Liu, PLP and PMP radicals: a new paradigm in coenzyme B6 chemistry, Bioorg. Chem. 29 (2001) 234–257. [22] B.T. Golding, L. Radom, On the mechanism of action of adenosylcobalamin, J. Am. Chem. Soc. 98 (1976) 6331–6338. [23] S.D. Wetmore, D.M. Smith, J.T. Bennett, L. Radom, Understanding the mechanism of action of B12-dependent ethanolamine ammonia-lyase: synergistic interactions at play, J. Am. Chem. Soc. 124 (2002) 14054–14065. [24] S.D. Wetmore, D.M. Smith, L. Radom, Enzyme catalysis of 1,2-amino shifts: the cooperative action of B6, B12, and aminomutases, J. Am. Chem. Soc. 123 (2001) 8678–8689. [25] M.D. Ballinger, G.H. Reed, P.A. Frey, An organic radical in the lysine 2,3aminomutase reaction, Biochemistry 31 (1992) 949–953. [26] M.D. Ballinger, P.A. Frey, G.H. Reed, Structure of a substrate radical intermediate in the reaction of lysine 2,3-aminomutase, Biochemistry 31 (1992) 10782–10789. [27] M.D. Ballinger, P.A. Frey, G.H. Reed, R. LoBrutto, Pulsed electron paramagnetic resonance studies of the lysine 2,3-aminomutase substrate radical: evidence for participation of pyridoxal 5′-phosphate in a radical rearrangement, Biochemistry 34 (1995) 10086–10093. [28] C.H. Chang, M.D. Ballinger, G.H. Reed, P.A. Frey, Lysine 2,3-aminomutase: rapid mixfreeze-quench electron paramagnetic resonance studies establishing the kinetic competence of a substrate-based radical intermediate, Biochemistry 35 (1996) 11081–11084. [29] P.A. Frey, M.D. Ballinger, G.H. Reed, S-Adenosylmethionine: a ‘poor man's coenzyme B12’ in the reaction of lysine 2,3-aminomutase, Biochem. Soc. Trans. 26 (1998) 304–310. [30] W. Wu, K.W. Lieder, G.H. Reed, P.A. Frey, Observation of a second substrate radical intermediate in the reaction of lysine 2,3-aminomutase: a radical centered on the beta-carbon of the alternative substrate 4-thia-L-lysine, Biochemistry 34 (1995) 10532–10537.

P.A. Frey, G.H. Reed / Biochimica et Biophysica Acta 1814 (2011) 1548–1557 [31] J. Miller, V. Bandarian, G.H. Reed, P.A. Frey, Inhibition of lysine 2,3-aminomutase by the alternative substrate 4-thialysine and characterization of the 4-thialysyl radical intermediate, Arch. Biochem. Biophys. 387 (2001) 281–288. [32] N.C. Baird, The three-electron bond, J. Chem. Ed. 54 (1977) 291–293. [33] W. Wu, S. Booker, K.W. Lieder, V. Bandarian, G.H. Reed, P.A. Frey, Lysine 2,3aminomutase and trans-4,5-dehydrolysine: characterization of an allylic analogue of a substrate-based radical in the catalytic mechanism, Biochemistry 39 (2000) 9561–9570. [34] O.Th. Magnusson, G.H. Reed, P.A. Frey, Spectroscopic evidence for the participation of an allylic analogue of the 5′-deoxyadenosyl radical in the reaction of lysine 2,3-aminomutase, J. Am. Chem. Soc. 121 (1999) 9764–9765. [35] O.Th. Magnusson, G.H. Reed, P.A. Frey, Characterization of an allylic analogue of the 5′-deoxyadenosyl radical: an intermediate in the reaction of lysine 2,3aminomutase, Biochemistry 40 (2001) 7773–7782. [36] D. Chen, P.A. Frey, Identification of lysine as a functionally important residue for pyridoxal 5-phosphate binding and catalysis in lysine 2,3-aminomutase from Bacillus subtilis, Biochemistry 40 (2001) 596–602. [37] F.J. Ruzicka, K.W. Lieder, P.A. Frey, Lysine 2,3-aminomutase from Clostridium subterminale SB4: mass spectral characterization of cyanogen bromide-treated peptides and cloning, sequencing, and expression of the gene kamA in Escherichia coli, J. Bacteriol. 182 (2000) 469–476. [38] R.M. Petrovich, F.J. Ruzicka, G.H. Reed, P.A. Frey, Characterization of iron–sulfur clusters in lysine 2,3-aminomutase by electron paramagnetic resonance spectroscopy, Biochemistry 31 (1992) 10774–10781. [39] K.W. Lieder, S. Booker, F.J. Ruzicka, H. Beinert, G.H. Reed, P.A. Frey, SAdenosylmethionine-dependent reduction of lysine 2,3-aminomutase and observation of the catalytically functional iron–sulfur centers by electron paramagnetic resonance, Biochemistry 37 (1998) 2578–2585. [40] D. Chen, C. Walsby, B.M. Hoffman, P.A. Frey, Coordination and mechanism of reversible cleavage of S-adenosylmethionine by the [4De–4S] cemter om; usome 2,3-aminomutase, J. Am. Chem. Soc. 125 (2003) 11788–11789. [41] N.J. Cosper, S.J. Booker, F. Ruzicka, P.A. Frey, R.A. Scott, Direct FeS cluster involvement in generation of a radical in lysine 2,3-aminomutase, Biochemistry 39 (2000) 15668–15673. [42] H.J. Sofia, G. Chen, B.G. Hetzler, J.F. Reyes-Spindola, N.E. Miller, Radical SAM, a novel protein superfamily linking unresolved steps in familiar biosynthetic pathways with radical mechanisms: functional characterization using new analysis and information visualization methods, Nucleic Acids Res. 29 (2001) 1097–1106. [43] B.W. Lepore, F.J. Ruzicka, P.A. Frey, D. Ringe, The x-ray crystal structure of lysine2,3-aminomutase from Clostridium subterminale, Proc. Natl. Acad. Sci. USA 102 (2005) 13819–13824. [44] C.G. Morley, T.C. Stadtman, Studies on the fermentation of D-alpha-lysine. Purification and properties of an adenosine triphosphate regulated B12coenzyme-dependent D-alpha-lysine mutase complex from Clostridium sticklandii, Biochemistry 9 (1970) 4890–4900.

1557

[45] C.G. Morley, T.C. Stadtman, Studies on the fermentation of D-alpha-lysine. On the hydrogen shift catalyzed by the B12 coenzyme dependent D-alpha-lysine mutase, Biochemistry 10 (1971) 2325–2329. [46] C.D. Morley, T.C. Stadtman, The role of pyridoxal phosphate in the B12 coenzymedependent D-α-lysine mutase reaction, Biochemistry 11 (1972) 600–605. [47] J.J. Baker, C. van der Drift, T.C. Stadtman, Purification and properties of lysine mutase, a pyridoxal phosphate and B12 coenzyme dependent enzyme, Biochemistry 12 (1973) 1054–1063. [48] C.H. Chang, P.A. Frey, Cloning sequencing, heterologous expression, purification, and characterization of adenosylcobalamin-dependent D-lysine 5,6-aminomutase from Clostridium sticklandii, J. Biol. Chem. 275 (2000) 106–114. [49] K.H. Tang, A. Harms, P.A. Frey, Identification of a novel pyridoxal 5′-phosphate binding site in adenosylcobalamin-dependent lysine 5,6-aminomutase from Porphyromonas gingivalis, Biochemistry 41 (2002) 8767–8776. [50] K.H. Tang, A.D. Casarez, W. Wu, P.A. Frey, Kinetic and biochemical analysis of the mechanism of action of lysine 5,6-aminomutase, Arch. Biochem. Biophys. 418 (2003) 49–54. [51] F. Kunz, J. Rétey, D. Arigoni, L. Tsai, T.C. Stadtman, Die absolut Konfiguration der 3,4-Diaminohexansaure aus der β-Lysin-Mutase-Reaktion, Helv. Chim. Acta 61 (1978) 1139–1145. [52] J. Retey, F. Kunz, D. Arigoni, T.C. Stadtman, Zur Kenntnis der β-Lysin-MutaseReaktion: Mechanismus und sterischer Verlauf, Helv. Chim. Acta 61 (1978) 2989–2998. [53] K.H. Tang, S.O. Mansoorabadi, G.H. Reed, P.A. Frey, Radical triplets and suicide inhibition in reactions of 4-thia-D and 4-thia-L-lysine with lysine 5,6-aminomutase, Biochemistry 48 (2009) 8151–8160. [54] A.N. Maity, C.P. Hsieh, M.H. Huang, Y.H. Chen, K.H. Tang, E. Behshad, P.A. Frey, S.C. Ke, Evidence for conformotional movement and radical mechanism in the reaction of 4-thia-L-lysine with lysine 5,6-aminomutase, J. Phys. Chem. B 113 (2009) 12161–23263. [55] F. Berkovitch, E. Behshad, K.H. Tang, E.A. Enns, P.A. Frey, C.L. Drennan, A locking mechanism preventing radical damage in the absence of substrate, as revealed by the x-ray structure of lysine 5,6-aminomutase, Proc. Natl Acad. Sci. U.S.A. 101 (2004) 15870–15875. [56] K.H. Tang, C.H. Chang, P.A. Frey, Electron transfer in the substrate-dependent suicide inactivation of lysine 5,6-aminomutase, Biochemistry 40 (2001) 5190–5199. [57] S.D. Christenson, W. Liu, M.D. Toney, B. Shen, A novel 4-methylideneimidazole-5one-containing tyrosine aminomutase in enediyne antitumor antibiotic C-1027 biosynthesis, J. Am. Chem. Soc. 125 (2003) 6062–6063. [58] K.D. Walker, K. Klettke, T. Akiyama, R. Croteau, Cloning, heterologous expression, and characterization of a phenylalanine aminomutase involved in Taxol biosynthesis, J. Biol. Chem. 279 (2004) 53947–53954. [59] B. Langer, M. Langer, J. Rétey, Methylidene imidazolone (MIO) from histidine and phenylalanine ammonia-lyase, Adv. Prot. Chem. 58 (2001) 175–214.