[Advances in Enzymology - and Related Areas of Molecular Biology] Advances in Enzymology and Related...

LYSINE 2,3-AMINOMUTASE AND THE MECHANISM OF THE INTERCONVERSION OF LYSINE AND P-LYSINE

By PERRY A. FREY and GEORGE H. REED, Institute for Enzyme Research, The Graduate School, and Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin-Madison, Madison, Wisconsin

C O N T E N T S

I.

11.

111. IV. V.

VI . VII.

VIII.

Introduction A. The Mechanistic Significance of Lysine 2.3-aminomutase B. The Metabolic Significance of Lysine 2,3-aminomutase Properties of Lysine 2,3-aminomutase A. Purification of the Enzyme with its Cofactors B. Molecular Eroperties of the Enzyme C. Activation of the Enzyme The Role of S-Adenosylmethionine in Hydrogen Transfer The Formation of Substrate Radicals as Intermediates The Role of Pyridoxal Phosphate in Amino Group Migration The Importance of Iron-Sulfur Clusters and Cobalt Adenosylcobalamin-dependent Aminomutases Adenosyl-cofactors as Evolutionary Predecessors of Adenosylcobalamin

I. Introduction

H. A. Barker and co-workers observed the enzymatic conversion of lysine into P-lysine in Clostridia in 1966 (1). They purified and described lysine 2,3-aminomutase, the enzyme that catalyzes reaction 1, in 1970 and showed that the rearrangement proceeds with conservation of hydrogen in the substrate; that is, no protons from the solvent are incorporated into nonexchangeable positions of lysine or P-lysine (2). Barker and co-workers described the molecular properties of lysine 2,3-aminomutase and reported that the enzyme

1

Advances in Enzymology and Related Areas of Molecular Biology, Volume 66 Edited by Alton Meister

Copyright © 1993 by John Wiley & Sons, Inc.

2 PERRY A. FREY AND GEORGE H. REED

(1)

is activated by S-adenosylmethionine, ferrous ions, and pyridoxal phosphate (PLP) (2).

A. THE MECHANISTIC SIGNIFICANCE OF LYSINE 2,3-AMINOMUTASE

The lysine 2,3-aminomutase reaction is interesting from a mechanistic standpoint because of the cofactor requirements and the mechanistic barrier to 1,Zamino group migrations. In reaction 1 there is cross-migration of an unactivated hydrogen and a neighboring group between adjacent carbon atoms. It is an example of one that follows the pattern of Eq. 2, in which X is the (Y amino group and H is one of the P-hydrogen atoms of lysine. These reactions normally require adenosylcobalamin as a cofactor, and, as originally established by Abeles and his associates, adenosylcobalamin mediates hydrogen transfer in these reactions (3). However, lysine 2,3-aminomutase is not activated by adenosylcobalamin and does not contain a BIZ derivative (2). Barker and his associates originally reported that in addition to being activated by ferrous ions, PLP, and S-adenosylmethionine, the purified enzyme contains iron and PLP (2, 4).

(2) I I I I

-cp-ca- * -cp-ca- I I I I

X H H X

Recent studies show that the enzyme requires four cofactors: iron-sulfur clusters, PLP, cobalt, and S-adenosylmethionine (5 ) . The adenosyl moiety of S-adenosylmethionine is now known to mediate the transfer of hydrogen between carbon-2 and carbon-3 of the substrate in a manner that appears to be analogous to the action of the adenosyi moiety of adenosylcobalamin (6, 7).

LYSINE 2.3-AMINOMUTASE 3

B. THE METABOLIC SIGNIFICANCE OF LYSINE 2,3-AMINOMUTASE

The importance of lysine aminomutases, including lysine 2,3-aminomutase, in Clostridial lysine metabolism is reviewed in this series by T. C. Stadtman (8). The amino groups in lysine are not properly placed to allow the molecule to be degraded easily to acetyl CoA. P-Lysine can be oxidized to a P-ketoacid, which can be activated to the CoA ester and cleaved in p-ketothiolase-fashion to acetyl CoA. Lysine 2,3-aminomutase, p-lysine mutase, and D-lysine 5,6- aminomutase play important roles in the isomerization of lysine into structures that are easily degraded to acetyl CoA for use as a cellular fuel and building block. These reactions allow lysine to serve as the carbon and nitrogen source for Clostridia. P-Lysine is also signifi- cant as a constituent of antibiotics that are produced by Strepto- myces and Norcardia (9- 14).

11. Properties of Lysine 2,3-Aminomutase

A. PURIFICATION OF LYSINE 2,3-AMINOMUTASE WITH ITS COFACTORS

In the original purification of lysine 2,3-aminomutase, Barker and co-workers reported that the enzyme contains about 3 atoms of iron and 1.7 molecules of pyridoxal phosphate per hexameric unit, it is activated by S-adenosylmethionine, and its activity is 5 to 7 inter- national units per mg of protein in their standard assay. The original purification has been repeated in this laboratory, and the activity and cofactor composition originally reported has been confirmed (5, 15). Improvements in the growth of Clostridium SB4, and in the purification of the enzyme have led to an increase in the specific activity of lysine 2,3-aminomutase to more than 30 units per mg protein (5, 16). The purified enzyme has also been found to contain inorganic sulfide and cobalt, as well as iron and PLP.

In early purification efforts in this laboratory, the major part of lysine 2,3-aminomutase activity was routinely lost from partially purified preparations whenever they were subjected to high resolution chromatographic procedures. This proved to be attributable largely to the separation of tightly bound substrate from the enzyme in the


course of chromatography. The inclusion of 0.1 mM lysine in purification buffers eliminated this problem (15).

Lysine 2,3-aminomutase purified in this laboratory is not more homogeneous than that described by Barker and co-workers, but it is more completely constituted with cofactors. The activity of the enzyme is maximal in crude extracts and throughout the purification procedure, when the bacteria are grown in a medium that is sup- plemented with 100 pM CoC12. Purified enzyme preparations gen- erally contain zinc and copper as well as cobalt and iron; however, the presence of zinc and copper is at the expense of cobalt and of enzymatic activity. The most active preparations contain a minimum of zinc and copper and up to 5 or more gram atoms of cobalt per mole of hexameric enzyme. The zinc and copper content of the enzyme is minimized, and the activity is maximized, by including 5 pM CoCl2 in the purification buffers ( 5 , 16).

Another purification measure that is required to attain maximal activity in the purified enzyme is the rigorous exclusion of oxygen. Enzyme purified under rigorously anaerobic conditions contains approximately 12 gram atoms of iron and of sulfide per mole of hexameric protein (5). The iron and sulfide content corresponds to six Fe2S2 clusters or to three Fe4S4 clusters; EPR data are incon- sistent with two-iron clusters but are consistent with the presence of four-iron clusters. In this laboratory, all purification steps are now carried out inside an anaerobic chamber. The purification buffers also contain 10 pM PLP and 0.1 mM lysine, which stabilize the activity, in addition to 5 pM CoCl2. Enzyme purified in the presence of both PLP and lysine contains up to 5.5 gram atoms of PLP per mole of hexameric enzyme (17).

B. MOLECULAR PROPERTIES OF THE ENZYME

Barker and his associates described lysine 2,3-aminomutase from Clostridium SB4 as a hexameric protein (MW 280-290K) composed of subunits with a molecular weight of 47,000 (2, 4). Recent experiments with lysine 2,3-aminomutase purified in this laboratory fur- ther supported this subunit composition (17). (a) Peptide mapping studies indicated that the subunits are chemically identical. Cleavage of the protein by cyanogen bromide gave nine major peptides, in agreement with the presence of eight methionine residues per sub-

LYSINE 2,3-AMINOMUTASE 5

unit. Lf the subunits differed significantly in amino acid sequence, up to twice this number of cyanogen bromide peptides would have been found. (b) Cross-linking experiments were consistent with a hexameric structure of the protein. Partially cross-linked samples of lysine 2,3-aminomutase give bands corresponding to monomers, dimers, tetramers, and hexamers upon polyacrylamide gel electro- phoresis in SDS. No trimers or pentamers could be observed, indicating that the hexameric structure is composed of three dimeric units.

The best preparations of lysine 2,3-aminomutase obtained to date contain, per hexameric unit, approximately 12 Fe, 12 S2-, 5.5 PLP, and 5 Co, as well as much smaller residues of Zn and Cu (5 , 16, 17). The specific enzyme activity increases with increasing amounts of Co. Increases in Co are at the expense of Cu and Zn, and increased specific activity is correlated with decreasing amounts of Zn and of Cu associated with the enzyme. It has so far not been possible to obtain a preparation that is completely free of Zn or Cu. The amount of Zn in the best preparations is near one gram atom per mole hexamer, and that of Cu is about 0.3 gram atom per mole hexamer. It remains possible that Zn is required for activity; however, the sim- plest interpretation of the available data is that Cu and Zn do not support enzymatic activity but that Co is required.

C . ACTIVATION OF THE ENZYME

The first preparations of purified lysine 2,3-aminomutase were essentially inactive, and all experiments in which the functions of this enzyme were probed had to be preceded by an activation protocol to convert the enzyme into an active form. Barker and co- workers activated the enzyme by incubating it anaerobically with dithionite and glutathione for three hours and then adding S-adenosylmethionine (2). The degree of activation was somewhat increased by substituting dihydrolipoate for glutathione and extending the activation to five hours (15).

Inasmuch as exposure of the enzyme to small amounts of oxygen inactivates it, one might suppose that lysine 2,3-aminomutase that had been rigorously protected from oxygen in the course of purification would be active. This is not true, however, because purification inside an anaerobic chamber does not prevent the enzyme


from losing activity. The enzyme must always be activated with S- adenosylmethionine. Purification under anaerobic conditions pro- tects the metal cofactors from being irreversibly degraded and allows maximal activity to be attained upon activation by S-adenosylmethionine. Although an enzyme that has been purified under strictly anaerobic conditions still requires activation by S-adenosylmethionine, it does not absolutely require the reductive incubation with glutathione or dihydrolipoate. The best preparations exhibit 50 to 75% of their maximum activities upon addition of S-adenosylmethionine with no preliminary reductive incubation (16). However, the reductive incubation leads to maximum activity.

111. The Role of S-Adenosylmethionine in Hydrogen Transfer

Early experiments showed that substrate hydrogen is conserved in the lysine 2,3-aminomutase reaction (2). Stereochemical analysis showed that hydrogen transfer and amino group migration proceed with inversion of configuration (18). The 3-pro-R hydrogen of lysine is transferred to the 2-pro-R position of P-lysine according to reaction 3. Amino group transfer is intramolecular.

D H + H~N-OO H NH: -

+ H~NV~\P/~OO - A t

%, 'H3N H H D

In one experiment, Aberhart and co-workers used lysine 2,3-aminomutase to convert a mixture of lysine and [3,3-ZHJysine to p- lysine by use of lysine 2,3-aminomutase. They analyzed the product for deuterium and reported that it consisted of monodeutero-P-lysine (18). This experiment showed that deuterium in the 3-pro-R position of [3,3-2H2]1ysine is transferred to unlabeled lysine in the production of two molecules of monodeutero-P-lysine, P-[2-2Hl]1ysine, and p- [3-2Hl]lysine. Therefore, although substrate hydrogen is conserved in the product, hydrogen transfer is intermolecular. Baraniak et a1 varied the ratio [3,3-2H2]lysine/lysine and analyzed the deuterium labeling in the samples of P-lysine produced from the mixtures (7).


Monodeutero-p-lysine and dideutero-p-lysine were detected in the products at all substrate ratios, and a plot of the ratio p-[’H2]lysine/ p-[ 2Hl]lysine versus [3 ,3-2Hz]lysine/lysine was a straight line with a positive intercept and a positive slope. The value of 0.11 for the intercept showed that intramolecular deuterium transfer takes place in about 11% of the turnovers as the ratio [3,3-2H2]lysine/lysine ap- proaches zero. Therefore, the conversion of lysine to P-lysine proceeds with both intramolecular and intermolecular hydrogen transfer.

Adenosylcobalamin-dependent rearrangements also characteris- tically proceed with conservation of hydrogen in the substrate and with both intermolecular and intramolecular hydrogen transfer (2, 19, 20). This results from the action of the adenosyl moiety of adenosylcobalamin in mediating hydrogen transfer, in which carbon- 5‘ accepts hydrogen from the substrate and transiently becomes the methyl group of 5’-deoxyadenosine (19, 21).

The adenosyl moiety of S-adenosylmethionine mediates hydrogen transfer in the lysine 2,3-aminomutase reaction in the manner of the adenosyl moiety of adenosylcobalamin in B 12-dependent rearrangements (6, 7). The experimental evidence supporting this role for S- adenosylmethionine is that hydrogen bonded to the 5’-methylene group of S-adenosylmethionine becomes part of the pool of substrate hydrogens that appear in the reaction product. Thus, when lysine 2,3-aminomutase is activated with S-[5’-3H2]adenosylmethionine and used to catalyze reaction 1, it incorporates tritium into both lysine and P-lysine according to reaction 4 (6, 7).

E.[5’-’H2]AdoMet E,AdoMet

Lysine . [3H]Lysine + P-L3H]Lysine

The following properties of lysine 2,3-aminomutase support the conclusion that the adenosyl moiety of S-adenosylmethionine mediates hydrogen transfer: 1) The enzyme incorporates tritium from S-[5’-3H2]adenosylmethionine into both lysine and p-lysine. 2) At chemical equilibrium in the foregoing experiment, the ratio of tritium in lysine and P-lysine is the same as the chemical ratio of lysine and P-lysine at equilibrium. 3) When the enzyme is activated with an


excess of S-[5’-3H2]adenosylmethionine, the amount of tritium incorporated into lysine and P-lysine depends upon the amount of enzyme used, rather than on the amount of the cofactor in the reaction mixture. 4) When a sample of the enzyme is partially activated with less than a stoichiometric amount of S-[5’-’H2ladenosyl- methionine, all of the tritium is transferred from the cofactor to lysine and P-lysine. These experiments indicate that the activation of lysine 2,3-aminomutase by S-adenosylmethionine establishes a pool of hydrogen atoms that participates in the hydrogen transfer process. This pool includes the 5’methylene hydrogen atoms of the adenosyl moiety in S-adenosylmethionine and either the 3-pro-R hydrogen atom of lysine or the 2-pro-R hydrogen atom of P-lysine. These experiments also indicated that the activation of lysine 2,3- aminomutase by S-adenosylmethionine is stoichiometric and, for practical purposes, irreversible; that is, free S-adenosylmethionine does not exchange with enzyme-bound S-adenosylmethionine at a rate that compares with that of the enzymatic conversion of lysine into P-lysine.

If the adenosyl moiety of S-adenosylmethionine mediates hydrogen transfer, as required by the foregoing experiments with S45’- 3H2]adenosylmethionine, the reverse transfer may also take place; that is, the 3-pro4 hydrogen of lysine or the 2-pro4 hydrogen of lysine may be incorporated into S-adenosylmethionine, at least into the enzyme bound cofactor. Experiments designed to detect the transfer of deuterium from 3-[ ’Hllysine to S-adenosylmethionine by lysine 2,3-aminomutase failed at first to verify such a process (22, 23). These experiments were carried out with S-adenosylmethionine in stoichiometric excess of the enzyme, and they implicitly depended upon the activation of lysine 2,3-aminomutase by free S-adenosylmethionine being reversible so that enzyme-bound S-adenosylmethionine would be in equilibrium with the free cofactor. The experiments also depended on S-adenosylmethionine bound to the enzyme being in equilibrium with the adenosyl moiety in its catalytically active chemical form. The former assumption proved to be incorrect, as shown by the observation that the amount of tritium mobilized into substrates from S-[5’-3H2]adenosylmethionine depended upon the amount of enzyme rather than on the amount of cofactor in the reaction mixture (6). In recent experiments, lysine 2,3-aminomutase was activated with a stoichiometrically compa-


rable amount of S-adenosylmethionine and used to convert [3- 3Hllysine into p-lysine. S-Adenosylmethionine was reisolated from the reaction mixture and found to contain tritium (24). These experiments confirmed the earlier conclusion that the adenosyl moiety of S-adenosylmethionine mediates hydrogen transfer in the lysine 2,3-aminomutase reaction.

IV. The Formation of Substrate Radicals as Intermediates

Inasmuch as the lysine 2,3-aminomutase reaction is formally similar to adenosylcobalamin-dependent rearrangements, and the hydrogen transfer mechanisms also appear to be similar, the question of whether organic radicals participate as intermediates naturally arises because organic radicals are intermediates in some or all of the adenosylcobalamin-dependent rearrangements. Organic radicals are generated in the course of the lysine 2,3-aminomutase reaction, as shown by the electron paramagnetic resonance (EPR) spectra in Fig. 1 (25) . The EPR spectra in Fig. 1 are those of four samples of the enzyme at 77 K that differ only in the presence or absence of lysine and S-adenosylmethionine, singly or together. Spectrum A is that of lysine 2,3-aminomutase that had been incubated with dihydrolipoate and PLP in preparation for activation by S-adenosylmethionine; no organic radical appears to be present. Spectrum B is an identical sample to which lysine had been added before freezing the solution for EPR analysis, and spectrum C is another identical sample to which S-adenosylmethionine had been added before freezing. Neither spectrum shows evidence of the presence of an organic radical. Spectrum D is of a sample to which both lysine and S- adenosylmethionine had been added before freezing, and this spectrum clearly shows an EPR signal centered at g = 2.003. The radical spectrum shows extensive structure corresponding to hyperfine cou- plings of the unpaired electron spin with neighboring nuclear spins.

The EPR signal of the organic radical in Fig. 1 is always observed in repetitions of the experiment, and it corresponds in signal intensity to between 9 and 10% of the active sites when the reaction solution is frozen in the steady-state after addition of lysine and before very much P-lysine is produced. In the course of the reaction, the intensity of the radical signal declines to an equilibrium value at a rate that corresponds closely to the rate of approach to chemical

g = 2.0028

i A

C

LI . . . . ~ . . . . l . . . . l . . . . I . . . . I . . . . l . . . . , . . . . , . . . . , . . ~

3160 3200 3240 3280 3320 Gauss

Figure 1. Electron paramagnetic resonance spectrum of organic radical intermediates in the lysine 2.3-aminomutase reaction.

The samples contained 31 )IM lysine 2,3-aminornutase (hexamer), which were frozen at 77 K within 90 s after the following treatments. (A) The EPR spectrum of the enzyme after reductive incubation in preparation for activation. (B) The enzyme reduced as in A plus 200 mM L-lysine. (C) The enzyme reduced as in A plus 1.2 mM S-adenosylmethionine. (D) The enzyme reduced as in A plus 200 mM L-lysine and 1.2 mM S-adenosylmethionine.

10

LYSINE 2,3-AMINOMUTASE 1 1

equilibrium between lysine and P-lysine. Moreover, the intensity of the signal at equilibrium is directly proportional to the amount of enzyme activity in the reaction mixture (25). Finally, the signal changes slightly in the course of the approach to equilibrium, showing that the signal is inhomogeneous at equilibrium. The signal at equilibrium includes the signal for a second radical that does not appear to be an intermediate.

The properties of the EPR signal at g = 2 indicate that organic radicals may be intermediates in the lysine 2,3-aminomutase reaction. Radical intermediates would appear only in the presence of the active enzyme and a substrate, which is the case in Fig. 1. The concentration of an intermediate will be determined by the concentration of the enzyme, as is the concentration of the radical in Fig. 1. The concentration of an intermediate will also be determined by the ratio of the concentrations of lysine and p-lysine. Under conditions in which the concentrations of lysine and P-lysine change, as they do in the course of approaching equilibrium, the concentration of an intermediate is likely to change owing to differences in kinetic parameters of steps connecting the intermediate to the substrate or product. Thus, in a simple isomerization of a substrate to a product through one intermediate X according to Scheme 1, the substrate and product are connected to X

Scheme I

through steps governed by different rate constants. The fraction of enzyme complexes that are in the form of X will depend on the relative values of the rate constants and the ratio of substrate to product. Within the first few seconds of mixing the substrate with the activated enzyme, little product is present and the concentration ratio of X to total enzyme, [X]/[ET], is determined largely by the concentration of the substrate and the values of k l , k2, k3 , and k4. At equilibrium, the concentration of the substrate is decreased and that of the product is increased because of the equilibrium constant (Keq = 6), and the ratio [X]/[ET] is governed by the values of all the rate constants and the ratio of substrate to product. Inasmuch as the product concentration is larger than the substrate concentra-


tion at equilibrium in the lysine 2,3-aminomutase reaction, the values of k s , k g , k7, and ks must be very important in determining the ratio [X]/[ET]. The change in signal intensity for the organic radical in lysine 2,3-aminomutase in the course of the approach to equilibrium is in accord with these expectations.

In the case of lysine 2,3-aminomutase, both the signal intensity and the signal shape change in the course of approaching equilibrium, which suggests that the radical signal is inhomogeneous at equilibrium and represents the presence of more than one organic radical. An inhomogeneous signal could arise from two radicals corresponding to two intermediates that are chemically connected by a reversible step. If two radical intermediates, XI and Xz, are on the reaction pathway, Scheme 2 can account for the change in the radical signal intensity in the course of

Scheme 2 ki ks ks k7 k9

E + S e E . S e X 1 e X z Y E . P S E k2 k4 kb kio + P

approaching equilibrium for the same reasons as in Scheme 1, except that the ratio of total radical to total enzyme is [XI + X~]/[ET]. If radical intermediates XI and X2 are at equilibrium with each other throughout the approach to equilibrium, the EPR signal will not change in shape with time. However, if two intermediates are not at equilibrium but are connected by one or more steps in a steady- state, the ratio of XI to X2 will change with the approach to equilibrium between the substrate and product. This appears to be the case because the EPR signal changes significantly in the course of the reaction, indicating the presence of two or more radicals at equilibrium (25). As explained in the following section, the radical signal that is observed in the steady-state immediately after mixing L-lysine with the activated enzyme is that of a single radical species; the second species appears later but does not seem to be an intermediate.

Initial experiments with isotopically-substituted samples of lysine show that the carbon skeleton of lysine hosts the steady-state radical. Resolution in the spectrum obtained with [3,3,4,4,5,5,6,6- 2H6]lysine as the substrate is substantially improved relative to that with lysine shown in Fig. I (25). The major splittings are, however, preserved in the spectrum with [3,3,4,4,5,5,6,6-2H6]lysine, and the


enhanced resolution arises from the abolition of longer range cou- plings to the radical center by protons at carbons 3 through 6 upon substitution of deuterium at those positions. The fact that the unpaired spin is detectably coupled to these protons indicates that the organic radical is located on the carbon skeleton of lysine. There is only a slight increase in spectral resolution of samples prepared in DzO, and this observation shows that the unpaired electron spin is very weakly coupled to exchangeable protons (25). Recent EPR measurements with [2-'3C]lysine, [2-*H]lysine, and [2-''NIlysine identify the steady-state radical-the major species in Fig. I-as a P-lysine imine radical with the unpaired electron on the a carbon. The EPR data show that the bond linking carbon-3 to nitrogen is approximately coplanar with the p-orbital on carbon-2 containing the unpaired electron, as shown in the following structure (26).

This conformation is fully compatible with the stereochemistry of the reaction.

V. The Role of Pyridoxal Phosphate in Amino Group Migration

The mechanistic importance of PLP in the lysine 2,3-aminomutase reaction and in other aminomutase reactions has been unclear for many years and remains uncertain to date. The catalytic function of PLP in nearly all known enzymatic and model nonenzymatic reactions is to stabilize carbanionic intermediates through its ability to delocalize electron pairs (27). (An exception to this rule is gly- cogen phosphorylase, in which PLP is required in a different capacity that does not entail imine formation with the substrate.) In the case of lysine 2,3-aminomutase, EPR measurements implicate substrate based organic radicals as intermediates. The transient formation of organic radicals does not rule out carbanionic interme-


diates if an electron transfer step is required in the mechanism; however, the amino acid carbanions that would be required for l ,2-amino group migrations include species that would not be stabilized by PLP.

A potential nonenzymatic model reaction for 1 ,2-amino group migration through the intermediate formation of organic radicals is based on the well-known cyclopropyl carbinyl radical rearrangement shown in reaction 5 (28). The reverse reaction proceeds with a rate

II - A L. -

constant of about lo8 s - ' . If imine radicals undergo the analogous process according to reaction 6, a route by which PLP could facil-

itate a 1,2-amino group migration would be available. Reaction of an amine with PLP to form an imine, followed by radical formation at the carbon atom adjacent to the imine nitrogen would allow the rearrangement in reaction 6 to take place.

A precedent for reaction 6 appears to be the conversion of the benzaldimine of a-methyl-P-bromoalanine ethyl ester into a mixture of the benzaldimine derivatives of a-methylalanine ethyl ester 1 and a-methyl-p-alanine ethyl ester 2 in the presence of a free radical chain initiation system, as illustrated in Fig. 2 (29). Under kinetic conditions allowing rearrangement, the formation of the rearranged product 2 in Fig. 2 is favored over the formation of product 1. This can be rationalized by the fact that the a radical in Fig. 3 is more stable than the P radical. Because of the reversible connection between the two radicals through an isomerization mechanism such as that of reaction 6, the more stable radical will be the predominant form at equilibrium and will be quenched to the corresponding prod-


CH3 I

N Br -CH*-F -C02Et

HC+

4 BusSnH 2, 2'-azobis(isobutyronitri1e) 1

/ a 3

I I

N

CH3-C-COzEt

HC('

1 2

Figure 2. A nonenzymatic model for an a,p-imino radical rearrangement. Abstraction of the bromine atom from the benzaldimine of 2-methyl-3-bromoal-

anine ethyl ester produces the 3-radical, which is either quenched or undergoes rearrangement to the 2-radical. Quenching of the 2-radical produces the rearrangement product (29). This reaction is thought to be a model for the action of PLP in facilitating a,p-imino rearrangements of radical intermediates in aminomutase reactions.


Figure 3. A hypothetical mechanism for the a,p-imino rearrangement of p-imino radicals.

This mechanism is analogous to the well documented cyclopropyl carbinyl radical rearrangement in which one of the ring atoms is nitrogen instead of carbon. It may be regarded as an azocyclopropyl carbinyl radical rearrangement.

uct. The greater stability of the a radical in Fig. 3 is due to the delocalization of the unpaired electron into the ester group. Under the conditions of the experiment, the ratio of 1 : 2 was found to be 1 : 13 (29).

The reaction in Fig. 2 can be a model for the isomerization of lysine and p-lysine if PLP forms an imine with the a-amino group of lysine at the active site of the enzyme, if the rearrangement is not impeded by the substitution of the pyridoxylidene group for the benzylidene group, and if the organic radical intermediates in the enzymatic reaction are structurally analogous to those in the model reaction. The intermediates in the putative model reaction are thought to be those in Fig. 3; however, the mechanism has not yet been definitively proven, although it appears to be reasonable.

There is little evidence for the direct participation of PLP in the mechanism of the lysine 2,3-aminomutase reaction. The enzyme is activated by PLP, and it is stabilized by the presence of PLP in purification buffers. However, all efforts to reduce PLP onto the enzyme or the substrate by treatment with NaBH4 or NaB3H4 have failed to afford convincing evidence of imine formation, either between PLP and an amino group at the active site of the enzyme or between PLP and lysine (30). If PLP forms an imine at the active site, it is either not reactive with NaBH4 or not accessible to the reagent. In the mechanism of Fig. 2 and 3, the imine is not protonated, and there appears to be no mechanistic advantage for a protonated imine relative to an unprotonated imine in this isomerization.


Unprotonated imines are much less reactive with NaBH4 than protonated imines. If a lysine-PLP imine is an intermediate in the lysine 2,3-aminomutase reaction, it may be unprotonated and therefore less reactive with NaBH, than the protonated imines of other PLP-dependent enzymes. Moreover, such a species might be shielded from reaction with external reagents by other interactions with the enzyme or/and a metallic cofactor.

The EPR measurements in D20 described above indicate that the migrating nitrogen atom in the radical intermediate is not bonded to more than on< hydrogen atom. The EPR data are, therefore, most compatible with an imine nitrogen in the radical, and this constitutes the only direct evidence for PLP being linked as an irnine with the migrating nitrogen atom.

VI. The Importance of Iron-Sulfur Clusters and Cobalt

Very little is known about the roles of the iron-sulfur clusters and cobalt in the lysine 2,3-aminomutase reaction. The presence of iron in the enzyme and the activating effect of added iron have been known since 1970, but inorganic sulfide and cobalt have only recently been discovered as constituents of the active enzyme (5) .

The evidence for the presence of iron-sulfur complexes as constituents of this enzyme consists of a) the presence of both iron and sulfide in approximate stoichiometric equivalence in the purified enzyme, b) the long wavelength visible absorption spectrum of the enzyme, and c) the fact that the purified enzyme exhibits an EPR signal at 10 K that corresponds to an iron-sulfur cluster ( 5 , 31). The EPR signal is complex and not yet characterized fully (31). The EPR signal is abolished by the reductive incubation that is required to attain maximum activation by S-adenosylmethionine. When inter- preted in the context of the vast literature on the properties of iron- sulfur clusters, the experimental facts leave little doubt that iron- sulfur clusters are associated with the enzyme.

The evidence for the presence of cobalt consists of a) all samples of the enzyme contain cobalt, and b) the purified enzyme exhibits an EPR spectrum at 4 K that is characteristic of high spin Co(I1) (5 ) . The spectrum of Co(I1) in unactivated lysine 2,3-aminomutase is shown in Fig. 4.

The mechanistic importance of iron-sulfur clusters and cobalt in

18 PERRY A, FREY AND GEORGE H. REED

I 1 I I I

765 965 1165

I 1 1 1 I i

1000 3000 gauss

5000

Figure 4. EPR spectrum of high-spin cobalt(I1) bound to lysine 2,3-aminomutase at 4.3 K. The inset is an expansion of the g = 7 region, showing the Co2+ hyperfine structure.

the action of lysine 2,3-aminomutase is a subject of speculation and experimental exploration. The only currently available evidence for their importance as cofactors is that the activity of the enzyme depends on the presence of both iron-sulfur clusters and cobalt (5 ) . Although no experiment directly implicates either of these cofactors in catalysis, it seems likely that one or both of them participate in the reaction; one may be required for activation of the enzyme by S-adenosylmethionine and the other for an essential catalytic function of the activated enzyme.


A key process in the activation of lysine 2,3-aminomutase appears to be the chemical reaction of a reduced form of the enzyme with S-adenosylmethionine. Reaction of the enzyme with a thiol such as glutathione or dihydrolipoate places it in a chemical state in which it can react with the adenosyl moiety of S-adenosylmethionine reversibly. The chemical nature of this process allows the 5’-methylene group to participate in the transfer of substrate hydrogen between the a and @ carbon atoms of the substrate in a manner that is analogous to the function of 5’-deoxyadenosine-5’yl-hereafter referred to as the 5’-deoxyadenosyl radical-in adenosylcobalamin- dependent rearrangements. The 5’-deoxadenosyl moiety in the catalytically active enzyme is most likely cleaved from methionine and bonded to one of the metallic cofactors. The properties of the adenosyl cofactor at the active site may include the capacity to generate the 5’-deoxyadenosyl radical reversibly, according to the scheme in Fig. 5. This radical could initiate the rearrangement by abstracting the 3-pro-R hydrogen atom from lysine, or the 2-pro-R hydrogen from @-lysine. The resulting substrate radical could undergo the reversible isomerization shown in Fig. 6 if lysine is bound as an a-imine to PLP. The structure of the substrate radical intermediate that can be observed in EPR experiments, the spectrum of which is shown in Fig. 1 , corresponds to intermediate 5 in Fig. 6.

Cleavage of the 5 ’-deoxyadenosyl moiety from S-adenosylmethionine appears to be mandated by the fact that both hydrogen atoms bonded to carbon-5’ are subject to be transferred to lysine and @-lysine in reaction 4. The most reasonable rationale for the participation of both hydrogen atoms of the 5’ methylene group in the hydrogen transfer pool is that the 5’-deoxyadenosyl moiety becomes detached from methionine by a process that allows free ro- tation of the methylene group about the bond linking it to carbon- 4‘ in its catalytically functional form in Fig 5 and 6 . Thus, tritium in both stereotopic positions of carbon-5’ in [5‘-3H2]adenosylmethionine is subject to transfer to lysine and @-lysine through po- sitional isotope exchange in the intermediate 5’-deoxyadenosine, as shown in Fig. 6 .

Experimental evidence for the cleavage of S-adenosylmethionine is provided by the fact that it is cleaved to 5’-deoxyadenosine and methionine in the course of the reaction (15). Upon mixing S-aden-


AdoMet 1/

Figure 5 . Hypothetical steps in the activation of lysine 2,3-aminomutase. The activation process begins with the incubation of lysine 2,3-aminomutase in

the presence of a reducing system. This prepares the enzyme for activation by S- adenosylmethionine. This incubation is required to attain maximum activity; however, enzyme that has been purified under rigorously anaerobic conditions exhibits 50 to 70 percent of maximum activity without preliminary incubation. Addition of S- adenosylmethionine then leads to the active enzyme. The 5’-deoxyadenosyl moiety is probably transferred to a metal cofactor, which allows it to form the 5’-deoxyadenosyl radical reversibly.

0 H

zC-R

HH

I€

+H

~N(C

HH

,),-C

, ,C

OO

-

&

HC

"N H

3C-R

H

H3C

-R

H3C

-R

3 4

5

Fig

ure

6.

A p

ropo

sal f

or th

e m

echa

nism

of

the

lysi

ne 2

,3-a

min

omut

ase

reac

tion

. L

ysin

e is

bou

nd t

o at

the

act

ive

site

as

its

a-al

dim

ine

with

PL

P.

The

S'-d

eoxy

aden

osyl

rad

ical

abs

trac

ts t

he 3

-pro

-R h

ydro

gen

of l

ysin

e to

for

m a

p-r

adic

al im

ine.

An

azoc

yclo

prop

yl c

arbi

nyl r

adic

al r

earr

ange

men

t lea

ds t

o th

e a-

imin

o ra

dica

l, w

hich

abs

trac

ts

a hy

drog

en a

tom

fro

m 5

'-deo

xyad

enos

ine

to f

orm

the

P-l

ysin

e-P

LP

ald

imin

e an

d 5'

-deo

xyad

enos

ine.


osylmethionine with the enzyme and lysine, a fraction of the S- adenosylmethionine that is bound to lysine 2,3-aminomutase is cleaved to 5’-deoxyadenosine and methionine. This supports the mechanism in Fig. 6, in which 5’-deoXyadenosine is a catalytic intermediate and methionine is retained at the active site.

The chemical structure of the catalytically functional adenosyl cofactor at the active site of lysine 2,3-aminomutase is not known. In S-adenosylmethionine, the bond linking the adenosyl moiety to the sulfur atom in methionine is strong and probably cannot be cleaved homolytically through noncovalent binding interactions with the active site. However, S-adenosylmethionine is an alkylating agent and can be expected to react with a nucleophile to transfer the adenosyl moiety. At the active site of lysine 2,3-aminomutase, this would lead to the reversible transformation of S-adenosylmethionine into methionine and an adenosyl species that reversibly forms the 5’-deoxyadenosyl radical. In Fig. 5, this process is illustrated as including the transfer of the adenosyl moiety to one of the metallic cofactors. The metallic cofactors are attractive as recipients of the adenosyl moiety, in part because of their presence and sup- posed importance in catalysis, but more importantly because a transfer of the adenosyl moiety to any nucleophilic group of an amino acid side chain in a protein could not potentiate the formation of a 5’-deoxyadenosyl radical.

The weak cobalt-carbon bond in adenosylcobalamin [Co(III)- CH2--] readily undergoes homolytic scission to Co(I1) and the 5‘- deoxyadenosyl radical (32, 33). This process constitutes a chemical precedent for the capacity of metallic cofactors, especially species of adenosyl-Co(III), to potentiate the reversible formation of the 5’- deoxyadenosyl radical. However, lysine 2,3-aminomutase does not contain adenosylcobalamin or any cobamide (2, 5) so that the activation by S-adenosylmethionine cannot lead to the formation of an adenosyl-BI2 species.

The fact that the enzyme contains Co(I1) compels us to consider whether the catalytically functional adenosyl cofactor might be an adenosyl-Co(I1I) species with a weak cobalt-carbon bond similar to that in adenosylcobalamin, but with cobalt ligands other than those provided by a cobamide. There is currently no evidence for the formation of such a species associated with this enzyme. How- ever, the detection of an adenosyl-Co(II1) species is difficult be-


cause of the fact that it could not be observed by EPR spectroscopy, and, with ligands other than those provided by a cobamide, detection by UV or visible absorption spectroscopy would be difficult. The difficulties would be compounded if such a species were to exist only at a low steady-state level. An alternative function of Co(I1) in this reaction is that it might participate in binding the substrate or in activating the iron-sulfur cluster. EPR data show that Co(I1) is not integrated into the iron-sulfur cluster.

The function of the iron-sulfur cluster is as mysterious as that of Co(I1). The most obvious hypothesis is that the cluster somehow participates with S-adenosylmethionine in the formation of an adenosyl-Co(II1) species that may serve as the source of the 5’- deoxyadenosyl radical in Fig. 4 and 5. Carbon-cobalt bonds can be formed by either of two routes, a) the alkylation of Co(1) and b) the coupling of an organic radical with Co(I1). These processes are described by reactions 7 and 8 and are well-known reactions of vitamin BI2 in two of its oxidation states, B12s [Co(I)] and [Co(II)].

\ / \THZR / Co(1) + Br-CH,R + ,tyII) + Br- (7)

’I\

S-adenosylmethionine is an alkylating agent and could react in an alkylation similar to reaction 7, in which the alkylating center is carbon-5’ of the adenosyl moiety and the leaving group is methionine. In order for the activation of lysine 2,3-aminomutase to include a process analogous to reaction 7, a nucleophilic Co(1) species would have to be generated in the course of the activation by S-adenosylmethionine. Available EPR data do not indicate that the observ- able high spin Co(I1) is reduced by the conditions employed in the activation protocol. However, the data do not exclude the possibility that some of the cobalt associated with the enzyme may be unobservable in EPR experiments. Co(1) has a d8 electronic configura-


tion, and the geometry of several Co(1) complexes renders them diamagnetic. Paramagnetic forms of Co(1) would have an even spin and would likely escape detection by EPR. The possibility of the formation of Co(1) is difficult to exclude on the basis of negative evidence; however, the reduction potentials of known species of Co(I1) are very low (<-0.5 V ) , lower than that provided by the activation conditions.

One way to circumvent the generation of Co(1) in the activation might be to generate the 5 '-deoxyadenosyl radical from S-adenosylmethionine and allow it to react with Co(I1) according to reaction 8. The 5'-deoxyadenosyl radical cannot be generated by homolytic cleavage of the bond linking carbon-5' with sulfur in S-adenosylmethionine under the conditions of enzymatic experiments; however, the alkylation potential of S-adenosylmethionine might be brought into play to transfer the adenosyl moiety to a reduced iron- sulfur cluster. Much of the negative charge associated with iron- sulfur clusters is concentrated on the sulfur ligands, and most of the additional charge associated with a reduced cluster resides on sulfur (34). Sulfur anions are highly reactive nucleophiles that might well undergo alkylation by a molecule such as S-adenosylmethionine. (Even the neutral sulfur in a sulfide such as methionine can be alkylated, as in the alkylation of methionine by ATP to form S- adenosylmethionine.) The hypothetical alkylation of a bridging sulfur in a reduced cluster (Fed&) by the adenosyl moiety of S- adenosylmethionine is shown in Fig. 7. An adenosyl-FeS cluster, such as that formed by initial alkylation of a sulfur ligand, might hold the adenosyl moiety weakly enough to allow the 5 ' -deoxyadenosyl radical to be formed, as shown in Fig. 7. Formation of the radical would be accompanied by a one electron oxidation of the cluster. There are no well-established chemical precedents for the reactions in Fig. 7, which must therefore be regarded as speculative. These reactions provide a reasonable, if hypothetical, scenario that can be subjected to experimental tests, both in search of nonenzymatic model reactions and in chemical and spectroscopic analysis of enzymatic intermediates.

The hypothetical mechanism in Fig. 7 could facilitate the formation of an adenosyl-Co(II1) species by generating the 5'-deoxyadenosyl radical for coupling with Co(1I). Alternatively, the alkylated cluster could be the immediate precursor of the 5'-de-


Figure 7. A hypothetical reaction of S-adenosylmethionine with a reduced iron- sulfur cluster.

Alkylation of a sulfide in a reduced iron-sulfur cluster, shown here as Fe4S4, by carbon-5' of S-adenosylmethionine would form an adenosyl-cluster. Such a species might be capable of reversible dissociation to the 5'-deoxyadenosyl radical and Fe& oxidized by one electron.


oxyadenosyl radicals that are required for the hydrogen abstraction steps in the rearrangement mechanism of Fig. 6. The EPR data in Fig. 1 show that, in the absence of substrate, the concentration of the 5’-deoxyadenosyl radical is either below the limit of detection or the radical is strongly exchange-coupled to another paramagnetic center such that the pair is EPR silent.

Available evidence does not allow us to conclude that the mechanism in Fig. 7 leads to the formation of the 5’-deoxyadenosyl radical nor that an adenosyl-Co(II1) species is an intermediate. An adenosyl-Co(II1) species would be diamagnetic, and the EPR signal for high-spin Co(I1) is not lost upon activation of lysine 2,3-aminomutase; therefore, available EPR data do not provide evidence for the transformation of Co(I1) into an adenosyl-Co(1II) complex. The available information cannot exclude an adenosyl-Co(II1) as an unobservable but important species, but there is at the present time no evidence for it.

VII. Adenosylcobalamin-dependent Aminomutases

Other well-characterized aminomutases are P-lysine mutase, D-lysine mutase, and ornithine mutase, which catalyze reactions 9 through 11, respectively. The purification, properties, and cofactor requirements of these enzymes have been reviewed by Stadtman (8) and Baker and Stadtman (35). These enzymes differ from lysine 2,3- aminomutase. All three are multi-subunit enzymes composed of un- like subunits. They are adenosylcobalamin-dependent enzymes that lack iron-sulfur clusters and are not activated by S-adenosylmethionine. PLP Is the one cofactor that is required by all aminomutases. Despite the differences in cofactors, the chemical mechanisms by which the aminomutases catalyze the a,&cross-migrations of hydrogen and an amino group may be similar.

NH3+ P-Lysine mutase NHj+ NH:

+H2N-oo- - - uoo- L-P-Lysine ~-eryrhro-3,5-Diarninohexanoate


NH; o-Lysine mutase - woo- +HzN

NHT NH: D-Lysine 2.5-Diaminohexanoate

Y NH: NH?

D-Ornithine 2.5-Diaminovalerate

The P-Lysine mutase and D-lysine mutase reactions (9 and 10) are both catalyzed by a single protein complex purified from Clos- tridium kluyverii (35). The complex consists of two proteins desig- nated El and Ez, in which component El contains adenosylcobalamin and PLP and carries out the aminomutase reactions. Component E2 stabilizes the activity of El and also enhances the activity brought about by an activation procedure. Component El appears to have two active sites, presumably one for catalyzing each of the two reactions, 9 and 10. Monovalent cations, thiol regents, and ATP also optimize and regulate the two activities of the complex (35).

Adenosylcobalamin mediates hydrogen transfer by these mutases through the intermediate formation of 5’-deoxyadenosine. The evidence for this is that tritium in [5’-3Hz]adenosylcobalamin is transferred to both the substrates and products of the reactions (36-38). This is the classical role of adenosylcobalamin in all of its enzymatic reactions (3), and, as explained above, it is the role of S-adenosylmethionine in the lysine 2,3-aminomutase reaction. The most probable mechanistic rationale for the action of adenosylcobalamin is that the enzyme-cofactor binding interactions induce the reversible formation of Blzr and the 5’-deoxyadenosyl radical. The 5’-deoxy-


adenosyl radical initiates the reaction by abstracting a hydrogen atom from carbon-5 of p-lysine or D-lysine to form 5’-deoxyadenosine and a Iysyl radical. The lysyl radical undergoes a rearrangement to a product-related radical, probably by a mechanism analogous to that in Fig. 6, and the product radical abstracts a hydrogen from carbon-5 ’ of 5’-deoxyadenosine. The experimental evidence for this mechanism in the p-lysine and D-lysine mutase reactions consists of the function of adenosylcobalamin in mediating hydrogen transfer and the fact that amino transfer is intramolecular (39). Al- though no direct evidence for the formation of Bizr or intermediate organic radicals has been presented, much evidence for such intermediates is available for other adenosylcobalamin-dependent reactions (3,39,40). While there is no reason to expect the mechanisms of the p-lysine and D-lysine mutase reactions to diverge from this pattern, the details of the mechanism remain to be elucidated.

An important question regarding all aminomutase reactions is what is the role of PLP. According to the mechanism presented in Fig. 6 for lysine 2,3-aminomutase, PLP forms an aldimine with the migrating amino group of the substrate, a @-hydrogen atom is ab- stracted by the 5’-deoxyadenosyl radical, and the (3-imino radical undergoes the rearrangement through the intermediate formation of an azocyclopropyl carbinyl radical. The formation of the aldimine between PLP and the migrating amino group is clearly essential to this mechanism, yet little evidence for this intermediate is available for the lysine 2,3-aminomutase reaction. However, in the case of the D-lysine mutase reaction, there is good evidence for the formation of an aldimine between PLP and the migrating amino group. The enzyme catalyzes a very slow PLP-dependent exchange of protons between water and carbond of D-lysine and carbon-5 of 2,5- diaminohexanoate (42). Although these exchange reactions are slower than the overall reaction, they nevertheless implicate aldimine formation as the most probable mechanism through which PLP can potentiate hydrogen exchange specifically at these positions. Inasmuch as these exchanges are known to be side reactions, rather than steps in the overall reaction, the fact that they are slow does not detract from their significance as indicators of the formation of aldimines between the migrating amino group and PLP.

The mechanism of the ornithine mutase reaction appears to be similar to that of the other adenosylcobalamin and PLP-dependent


aminomutases (35). The enzyme consists of large and small subunits, the larger of which contains adenosylcobalamin. Original reports of the presence of p-leucine and adenosylcobalamin-dependent leucine 2,3-aminomutases in bacteria, plants, and animals (43-45) could not be confirmed in careful studies of human and rat tissues (46). The reason for the discrepancies remain unknown, but state-of-the-art bioanalytical methods fail to provide any evidence for the presence of this enzyme in human and rat tissues. Possible problems with the bioanalytical detection of p-leucine raise questions about the exis- tence of p-leucine aminomutases in bacteria and plants.

The question of the role of PLP in aminomutase reactions is fur- ther complicated by the absence of a requirement for PLP in the ethanolamine deaminase reaction (Eq. 12) (47). Ethanolamine de-

aminase is an adenosylcobalamin-dependent enzyme that catalyzes the deamination of ethanolamine to acetaldehyde by a mechanism that is thought to be analogous to that by which diol dehydrase catalyzes the dehydration of 1,2-propanediol (reaction 13). The

No H3C-CH-CH2 ---* H3C-CHZ-C + H20 (13)

H I I \

OH OH

mechanism entails the initial abstraction of a hydrogen atom from carbon-1 of the substrate by a radical associated with the enzyme- adenosylcobalamin complex in a process that is initiated by the 5 ’ - deoxyadenosyl radical. The substrate radical then undergoes the transformations shown in Fig. 8, in which ammonia is eliminated from carbon-2 of the substrate radical to form an intermediate oxycation radical. Ammonia then adds to carbon-1 of the oxycation radical to form a radical related to the rearrangement product, and this radical abstracts a hydrogen from carbon-5’ of 5’-deoxyadenosine to regenerate the enzyme-adenosylcobalamin complex and to


+H3N OH

Figure 8. The mechanism of the ethanolamine deaminase reaction.

form the ammonia adduct of acetaldehyde as the initial product. The enzyme catalyzes the deamination of the carbinolamine to form acetaldehyde and ammonia. Considerable evidence supports the formation of the carbinolamine adduct of ammonia with acetaldehyde as an enzyme-bound intermediate (48). Thus, the ethanolamine deaminase reaction can be regarded as analogous to the other aminomutase reactions because it entails the a$-cross-migration of a hydrogen and an amino group in the substrate.

Despite the apparent analogy between the aminomutase and the ethanolamine deaminase reactions, there is a fundamental difference that may be the chemical basis for the PLP requirement in the aminomutase reactions. The elimination of ammonia to form the oxycation radical in Fig. 8 can be expected to be more facile than the


corresponding elimination that would have to take place in the aminomutase reactions. The substrates of the aminomutase reactions lack hydroxyl groups as substituents adjacent to the migrating amino groups; therefore, the elimination of ammonia from the initial radical intermediate in the aminomutase reactions would produce cation radicals rather than oxycation radicals, such as that in the ethanolamine deaminase reaction. The positive charge should be stabilized to a greater degree by delocalization in an oxycation radical than in a cation radical, as illustrated below in a valence bond formulation

of the resonance forms. Therefore, the oxycation radical in the ethanolamine deaminase reaction should be much more stable-more correctly, far less unstable-than the corresponding cation radicals that would have to be formed if the aminomutase reactions pro- ceeded by the mechanism in Fig. 8. Apparently, cation radical formation does not take place in the aminomutase reactions; it is not obvious how it could be catalyzed by an enzyme, and the sponta- neous reaction apparently does not take place rapidly. Thus, the high activation barrier to cation radical formation may prevent the aminomutase reactions from proceeding by mechanisms analogous to that of ethanolamine deaminase.

PLP can provide the alternative rearrangement mechanism shown in Fig. 6 for the p-imino radical intermediates of the aminomutase reactions. This is a new mechanistic role for PLP, which has been known for over thirty years as a coenzyme that stabilizes carbanionic intermediates through the delocalization of the anion within imine intermediates formed between PLP and amino acids (27). In the aminomutase reactions, PLP appears to facilitate the a,p-imino rearrangement of a p-imino radical, probably through the interme-


diate formation of an aziridine radical-an azocyclopropyl carbinyl radical-from the substrate-PLP imino radical.

VIII. Adenosyl-cofactors as Evolutionary Predecessors of Adenos ylcobalamin

Adenosylcobalamin is highly evolved to provide the right balance of chemical properties that allow it to exist as a stable molecule in aqueous and aerobic solutions, while carrying within it the potential to undergo a homolytic scission of the cobalt-carbon bond in the active sites of enzymes. Adenosylcobalamin is a complex molecule that is built up in more than twenty biosynthetic steps. It is difficult to imagine how such a complex molecule would have appeared through incremental evolution in response to the survival value, to a primitive organism, of one or two radical reactions. In incremental evolution, a molecule that affords survival value can appear sud- denly and give its host a competitive advantage, or it can appear in a few metabolic steps within related organisms that compete for survival. However, when the biosynthesis of such a molecule requires many steps, as does the biosynthesis of adenosylcobalamin, it is difficult to imagine how related organisms would evolve the ten to twenty of the more than twenty enzymes required to produce adenosylcobalamin without losing competitive advantage. The biosynthesis of such a molecule in many steps would quickly become misdirected. Of course cells carrying misdirected biosynthetic pathways might survive, but it seems unlikely that even those carrying most of the numerous steps in the biosynthesis of adenosylcobalamin could have any survival advantage over those not carrying these activities.

The most obvious scenario that would allow a complex molecule such as adenosylcobalamin to evolve would be for the basic radical chemistry to be available to primitive organisms through the exis- tence of cofactors that were more primitive than adenosylcobalamin. Such cofactors could potentiate the survival of particular organisms in certain environments by providing radical-based chemical reactions that were not available to other organisms in the same environments. These early cofactors could be refined over evolutionary time to more idealized forms. It seems likely that some of the early cobalt-carbon cofactors were biosynthetic precursors of adenosyl-


cobalamin, or closely related molecules. However, the first cofactors that served to generate organic radicals may not have been related to cobalamin. Thus, primitive cofactors probably afforded organic radical chemistry as a part of the chemical context from which competing organisms could draw in the evolution of metabolic pathways.

The nature of the hypothetical, primitive cofactors is not known. However, the hypothesis that such cofactors existed is supported by the survival of the adenosyl-cofactor associated with lysine 2,3- aminomutase. While the chemical structure of this cofactor is not yet known, it almost certainly exists. In view of the fact that this adenosyl-cofactor has survived evolution to the present day, we do not wish to claim that it is primitive or inferior to adenosylcobalamin in catalyzing the isomerization of lysine to p-lysine under the anaerobic conditions in which Clostridiu grow. However, it does not survive aerobic conditions, and it is labile even under anaerobic conditions in the active site of lysine 2,3-aminomutase; it must be maintained through regeneration from S-adenosylmethionine (15). Thus, it could not compete with adenosylcobalamin as a general cofactor with the capacity to provide radical chemistry at enzymatic active sites in aerobic organisms, and it could not circulate in the bloodstreams of higher animals. This or analogous cofactors may have been evolutionary predecessors to adenosylcobalamin and may have provided the chemical context in which adenosylcobalamin could evolve as a more perfect coenzyme.

IX. Activation of Other Radical Enzymes by S- Adenosylmethionine

Cofactors similar to the adenosyl-cofactor of lysine 2,3-aminomutase may activate two other bacterial enzymes that are currently being intensively investigated. Pyruvate formate lyase catalyzes reaction 14 in E. coli. Pyruvate formate lyase must be activated by an

0 0 II + CoASH +

It + HCOO- A

H3C SCoA A

H3C COO-

(14)


activating enzyme; in the activation process an enzymatic radical is created according to the scheme in Fig. 9 (49). The activating enzyme is itself activated by S-adenosylmethionine. Evidence indicates that the pyruvate formate lyase reaction proceeds through a radical mechanism (50, 51). Little is known about the activating enzyme, but, inasmuch as it cleaves S-adenosylmethionine to methionine and a nucleoside that seems to be 5'-deoxyadenosine (49), it too appears to contain an adenosyl-cofactor. This cofactor probably generates the 5'-deoxyadenosyl radical that abstracts a hydrogen from an amino acid of pyruvate formate lyase to produce the enzymatic radical. The amino acid radical is now known to be a glycyl radical (49a).

The ribonucleotide reductase in E. coli grown under aerobic conditions contains a tyrosyl radical that participates in catalysis of reaction 15 (52, 53). The tyrosyl radical is generated by an oxidative

' H HO

process requiring the participation of dioxygen and a p-0x0-diiron complex associated with the enzyme. This ribonucleotide reductase cannot function in E. coli grown anaerobically because the production of the tyrosyl radical requires dioxygen. Ribonucleotide reduc- tases in anaerobic bacteria are often adenosylcobalamin-dependent, and the 5'-deoxyadenosyl radical plays the same role in initiating catalysis as does the tyrosyl radical in the enzyme from E. coli and animals. When E. coli are grown anaerobically, they contain a different ribonucleotide reductase, one that catalyzes the reduction of ribonucleotide triphosphate and that is not B 12-dependent (54).

The anaerobic E. coli ribonucleotide reductase is activated by an

LYSINE 2,3-AMTNOMUTASE 35

PFL Activating enzyme (inactive) 7 t

AdoMet

PFL Activating enzyme (activ

(inactive)

. I I

(active)

Figure 9. Activation of pyruvate formate lyase (PFL). The glycine radical in the activated enzyme has recently been characterized EPR (49a).

enzyme that requires S-adenosylmethionine; the process appears to be analogous to the activation of pyruvate formate lyase (54). It is very likely that the activating enzyme, which is activated by S- adenosylmethionine, generates an amino acid radical at the active center of the anaerobic E. coli ribonucleotide reductase. This process is likely to entail the cleavage of S-adenosylmethionine to a different adenosyl-cofactor that would serve as the source of a 5 ' - deoxyadenosyl radical, which would abstract a hydrogen atom from an amino acid in the active site of the reductase. Such an activation process would allow the anaerobic reductase to utilize an amino acid radical to catalyze the reduction of ribonucleotides by the same mechanism as that of the aerobic reductase.

Lactyl CoA dehydratase from Clostridium propionicum catalyzes reaction 16, in which water is eliminated @,a to the ester carbonyl

0 II

OH 0 I II

H~C--CH--C-SCOA ---* H*C=CH--C-SCoA + Hz0

(16)

instead of through the conventional a,@ elimination. The @ hydrogen atom of lactyl CoA is not acidic enough to allow this elimination to proceed by conventional acid-base catalysis. Two proteins, El and EZ, are required to catalyze this reaction, and Ez contains two iron- sulfur clusters as well as riboflavin and flavin mononucleotide (55, 56). Radical mechanisms have been proposed for reaction 16. It

36 PERRY A, FREY AND GEORGE H. REED

should be interesting to learn in future investigations how the iron- sulfur clusters in lactyl CoA dehydratase and in lysine 2,3-aminomutase participate in radical mechanisms.

Acknowledgments

K.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

The authors are grateful to the National Institutes of Health for the support of their research on lysine 2,3-aminomutase through Grant DK28607 (PAF) and Grant GM35752 (GHR). The research described herein as emanating from our laboratories was performed in collaboration with M. L. Moss, R. M. Petrovich, F. J. Ruzicka,

B. Song, and M. D. Ballinger.

References

Costilow, R. N., Rochovansky, 0. M., and Barker, H. A., Identification of p- lysine as an intermediate in lysine fermentation, J . Biol. Chem., 241,1573 (1966). Chirpich, T. P., Zappia, V., Costilow, R. N., and Barker, H. A., Lysine 2,3- aminomutase. Purification and properties of a pyridoxal phosphate and S- adenosylmethionine-activated enzyme, J. Biol. Chem., 245, 1778 (1970). Abeles, R. H. and Dolphin, D., The vitamin BIZ coenzyme, Acc. Chem. Res., 9 , 114 (1976). Zappia, V. and Barker, H. A., Studies on lysine 2,3-aminomutase. Subunit structure and sulfhydryl groups, Biochim. Biophys. Acta, 207, 505 (1970). Petrovich, R. M., Ruzicka, F. J., Reed, G. H., and Frey, P. A., Metal cofactors of lysine 2,3-aminomutase, J . Biol. Chem., 266, 7656 (1991). Moss, M. and Frey, P. A., The role of S-adenosylmethionine in the lysine 2,3- aminomutase reaction, J . Biol Chem., 262, 14859 (1987). Baraniak, J., Moss, M. L., and Frey, P. A., Lysine 2,3-aminomutase. Support for a mechanism of hydrogen transfer involving S-adenosylmethionine, J. Biol. Chem., 264, 1357 (1989). Stadtman, T. C., Lysine metabolism by Clostridia, in Advances in Enzymology and Related Areas of Molecular Biology, Vol. 38, Meister, A., Ed. Wiley, New York, p. 413 (1973). French, J. C., Bartz, Q. R., and Dion, H. W., Myomycin, a new antibiotic, J. Antibiot. (Japan), 26, 272 (1973). Haskell, T. H., Fusari, S. A., Frohardt, R. P., and Bartz, Q. R., The chemistry of viomycin, J . Am. Chem. Soc., 74, 599 (1952). McGahren, W. J., Hardy, B. A., Morton, G. O., Lovell, F. M., Perkinson, N. A., Hargreaves, R. T., Borders, D. B., and Ellestad, G. A., (p-Lysyl- oxy)myoinositol guanidine glycoside antibiotics, J. Org. Chern., 46, 792 (1981).


12.

13.

14.

15.

16. 17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Could, S. J., Martinkus, K. J. , and Tam, C.-H., Biosynthesis of streptothricin F. 1. Observing the interaction of primary and secondary metabolism with [1,2- I3C2]acetate, J . Am. Chem. Soc., 103, 2871, (1981). Gould, S. J., Martinkus, K. J., andTann, C.-H., Studies ofnitrogen metabolism using 13C NMR spectroscopy. 2. Incorporation of L-[guanidino- ‘3C,1SN2]arginine and 2-[guanidino-I3C ,2-”N]arginine into streptothricin F, J . Am. Chem. SOC., 103, 4639 (1981). Could, S. J . and Thiruvengadam, T. K., Studies of nitrogen metabolism using I3C NMR spectroscopy. 3. Synthesis of ~~-[3-”C,2-”N]lysine and its incorporation into streptothricin F, J . Am. Chem. SOC., 103, 6752 (1981). Moss, M. L. and Frey, P. A., Activation of lysine 2,3-aminomutase by S-adenosylmethionine, J . Biol. Chem., 265, 18112 (1990). R. B. Petrovich and P. A. Frey, Manuscript in preparation. Song, K. B. and Frey, P. A,, Molecular properties of lysine 2,3-aminomutase, J. B i d . Chem., 266, 7651 (1991). Aberhart, D. J., Gould, S. J., Lin, H.-J., Thiruvengadam, T. K., and Weiller, B. H., Stereochemistry of lysine 2,3-aminomutase isolated from Clostridium subterminale SB4, J . Am. Chem. SOC., 105, 5461 (1983). Frey, P. A., Essenberg, M. K., and Abeles, R. H., Studies on the mechanism of hydrogen transfer in the cobamide coenzyme-dependent dioldehydrase reaction, J . Biol. Chem., 242, 5369 (1967). Miller, W. W. and Richards, J. H., Mechanism of action of coenzyme BIZ. Hydrogen transfer in the isomerization of methylmalonyl coenzyme A to suc- cinyl coenzyme A, J . Am. Chem. SOC., 91, 1498 (1969). Frey, P. A. and Abeles, R. H., The role of the BIZ coenzyme in the conversion of 1 ,2-propanediol to propionaldehyde, J. Biol. Chem., 241, 2732 (1966). Aberhart, D. J., Studies on the mechanism of lysine 2,3-aminomutase, J. Chem. SOC. Perkin Trans., 1, 343 (1988). Aberhart, D. J. and Cotting, J. A., Mechanistic studies on lysine 2,3-aminomutase: Carbon-13-deuterium crossover experiments, J. Chem. SOC. Perkin Trans., I , 2119 (1988). Kilgore, J. L. and Aberhart, D. J., Lysine 2,3-aminomutase: role of S-adenosyl- L-methionine in the mechanism. Demonstration of tritium transfer from (2RS, 3RS)-[3-3H]lysine to S-adenosyl-L-methionine, J . Chem. SOC. Perkin Trans., 1 , 79 (1991). Ballinger, M. D., Reed, G. H., and Frey, P. A., An organic radical in the lysine 2,3-aminomutase reaction, Biochemistry, 3Z, 949 (1992). Ballinger, M. D., Reed, C. H., and Frey, P. A., The structure of a substrate radical intermediate in the reaction of lysine 2,3-aminomutase, Biochemistry, In press. Metzler, D. E., Ikawa, M., and Snell, E. E., A general mechanism for vitamin B6-catalyzed reactions, J . Am. Chem. SOC., 76, 648 (1954). Griller, D. and lngold, K. U., Free radical clocks, Acc. Chem. Res., 13, 317 ( 1980).


29. Han, 0. and Frey, P. A., Chemical model for the pyridoxal 5'-phosphate-dependent lysine aminomutases, J . Am. Chem. SOC., 112,8982 (1990).

30. Song, K. B. and Frey, P. A., Unpublished experiments. 31. Petrovich, R. M., Ruzicka, F. J., Reed, G. H., and Frey, P. A., Characterization

of iron-sulfur clusters in lysine 2,3-aminomutase by electron paramagnetic resonance spectroscopy, Biochemistry, In press.

32. Kim, S. H., Chen, H.-L., Feilchenfeld, N., and Halpern, I . , Thermal decom- position and cobalt-carbon dissociation energies of organocobalamin: neopen- tyl-, (cyclopentylmethy1)-, (cyclohexylmethy1)-, (tetrahydrofufury1)-, and ((tet- rahydro-2H-pyryl)methyl)cobalamin, J . Am. Chem. SOC., 110, 3120 (1988).

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

Hay, B. P. and Finke, R. G., Thermolysis of the Co-C bond in adenosylcorrins. 3. Quantification of the axial base effect in adenosylcobalamin by the synthesis and thermol ysis of axial base-free adenosylcobinamide. Insights into the ener- getics of enzyme-assisted cobalt-carbon bond homolysis, J . Am. Chem. SOC., 109, 8012 (1987). Noodleman, L., Norman, J. G., Jr., Osborne, J. H., Aizman, A., and Case, D. A., Models for ferredoxins: electronic structures of iron-sulfur clusters with one, two, and four iron atoms, J . Am. Chem. Soc., 107, 3418 (1985). Baker, J . J. and Stadtman, T. C., Amino mutases in BIZ: Biochemistry and Medicine, Vol. 2, Dolphin, D., Ed., Wiley, New York, p. 203 (1982). Retey, J., Kunz, F., Stadtman, T. C., and Arigoni, D., Zum mechanismus der P-lysin mutase-reaktion, Experienria, 25, 801 (1969). Morley, C. G. D. and Stadtman, T. C., Studies on the fermentation of D+ lysine. On the hydrogen shift catalyzed by the Blz coenzyme-dependent D-a- lysine mutase, Biochemistry, 10, 2325 (1971). Retey, J., Kunz, F., Arigoni, D., and Stadtman, T. C., Zur kenntnis der p- Lysin-Mutase Reaktion: Mechanismus und sterischer verlauf, Helv. Chim. Actu, 61, 2989 (1978). Bray, R. C. and Stadtman, T. C., Anaerobic degradation of lysine. 111. "N studies on the conversion of lysine to 3,5-diaminohexanoate, J . Biol. Chem., 243, 381 (1968). Golding, B. H., Mechanisms of Action of the BIZ Coenzyme: Theory and Models, in B I Z : Chemistry, Vol. 1, Dolphin D., Ed., Wiley, New York, p. 543 (1982). Frey, P. A., Importance of organic radicals in enzymatic cleavage of unactivated C-H bonds, Chem. Rev., 90, 1343 (1990). Morley, C. G. D. and Stadtman, T. C., The role of pyridoxal phosphate in the BIZ coenzyme-dependent D-a-lysine mutase reaction, Biochemistry, I I , 600 (1972). Poston, J. M., Leucine 2,3-aminornutase: a cobalamin-dependent enzyme present in bean seedlings, Science, 195, 301 (1977). Poston, J. M., Coenzyme B12-dependent enzymes in potatoes: leucine 2,3-aminomutase and methylmalonyl-CoA mutase, Phytochemistry, 17, 401 (1978).


45. Poston, J. M. and Hemmings, B. A., Cobalamins and cobalamin-dependent enzymes in Candida uiilis, J Bacieriol., 140, 1013 (1979).

46. Stabler, S., Lindenbaum, J., and Allen, R. H., Failure to detect p-leucine in human blood or leucine 2,3-aminomutase in rat liver using capillary gas chromatography-mass spectrometry, J. Biol. Chem., 263, 5581 (1988).

47. Kaplan, B. H. and Stadtman, E. R., Ethanolamine deaminase, acobamide coenzyme-dependent enzyme. 11. Physical and chemical properties and interactions with cobamides and ethanolamine, J. Biol. Chem., 243, 1794 (1968).

48. Carty, T. J., Babior, B. M., and Abeles, R. H., The mechanism of action of ethanolamine ammonia-lyase, a B ,*-dependent enzyme. IX. Interaction of the enzyme-coenzyme complex with reaction products, J. Biol. Chem., 246, 6313 (1971). Knappe, J., Neugebauer, F. A., Blaschkowski, H. P., and Ganzler, M., Post- translational activation introduces a free radical into pyruvate formate-lyase, Proc. Natl. Acad. Sci. USA, 81, 1332 (1984).

49a. Volker Wagner, A. F., Frey, M., Neugebauer, F. A., Schiifer, W., and Knappe,

49.

50.

51.

52.

53.

54.

55.

56.

J., The free radical in pyruvate formate lyase is located on glycine-734, Proc. Natl. Acad. Sci. USA, 89, 996 (1992). Unkrig, V., Neugebauer, F. A,, and Knappe, J., The free radical of pyruvate formate lyase. Characterization by EPR spectroscopy and involvement in catalysis as studied by the substrate analog hypophosphite, Eur. J. Biochem., 184, 723 (1989). Plaga, W., Frank, R., and Knappe, J., Catalytic-site mapping of pyruvate formate lyase. Hypophosphite reaction on the acetyl-enzyme intermediate affords carbon-phosphorus bond synthesis (1-hydroxyethyl phosphonate), Eur. J. Biochem., 178, 445 (1988). Reichard, P., Interactions between deoxyribonucleotides and DNA synthesis, Ann. Rev. Biochem., 57, 349 (1988). Stubbe, J., Protein radical involvement in biological catalysis, Ann. Rev. Biochem., 58, 257 (1989). Eliasson, R., Fontecave, M., Jornvall, H., Krook, M., Pontis, E., and Reichard, P., The anaerobic ribonucleotide triphosphate reductase from Escherichia coli requires S-adenosylmethionine as a cofactor, Proc. Nail. Acad. Sci. USA, 87, 3314 (1990). Kuchta, R. D. and Abeles, R. H., Lactate reduction in Clostridiumpropionicum. Purification and properties of lactyl-CoA dehydratase, J . Biol. Chem., 260,13 181 (1985). Kuchta, R. D., Hanson, G. R., Holmquist, B., and Abeles, R. H., Fe-S centers in lactyl-CoA dehydratase, Biochemistry, 25, 7301 (1986).

[Advances in Enzymology - and Related Areas of Molecular Biology] Advances in Enzymology and Related...

Documents

Transcript of [Advances in Enzymology - and Related Areas of Molecular Biology] Advances in Enzymology and Related...