Glutamate and 2-Methyleneglutarate Mutase From Microbial Curiosities to Paradigms for Coenzyme B,,-dependent Enzymes Wolfgang Buckel La bo ra to rium fur Mikrobiolog ie Fac h be reic h Bio log ie Ph ilipps- Un ive rsita t D-35032 Marb urg Germany Bernard T. Golding Department of Chemistry Bedson Building The University of Newcastle upon Tyne Newcastle upon Tyne UK NE7 7RU Dedicated to H. A. Barker in his ninetieth year 1 Introduction In the late 1950s H. Albert Barker discovered a light-sensitive yellow-orange cofactor for the carbon-skeleton rearrangement of glutamate to 3-methylaspartate in Clostridium tetanomorphum a strict anaerobic bacterium fermenting glutamate to ammonia CO acetate butyrate and H,.I The enzyme catalysing this process was isolated named glutamate mutase (EC 5.4.99.1),and shown to act specifically on (S)-glutamate which equilibrated with (2S,3S)-3-methylaspartate (Scheme 1).* Barker proved that the cofactor was (S)-glutamate (2S,3S)-3-methylaspartate Scheme 1 Interconversion of (5’)-glutamate with (2S,3S)-3-methylaspartate catalysed by coenzyme B ,,-dependent glutamate mutase related to vitamin B,,,actually pseudo-vitamin BI2,by showing that treatment with an excess of cyanide gave the characteristic reddish purple dicyanocobalamin (see Box concerning cobamide nomen- clature). Subsequently dark-red crystals of the coenzyme form of vitamin B I ,were isolated from Propionibacteriutn sherrnanii and subjected to X-ray analysis which yielded the structure shown in the Box.’ The presence of a cobalt-carbon a-bond in coenzyme B (adenosylcobalamin) was an unexpected feature that would ulti- mately prove to be of crucial importance for the biological action of the coenzyme. A decade later Thressa C. Stadtman identified 2-methyleneglu- tarate mutase (EC 5.4.99.4) in Clostridium barkeri which fer- ments nicotinate via 2-methyleneglutarate to ammonia CO acetate and propionate! During this degradation the enzyme catal- yses the interconversion of 2-methyleneglutarate with (R)-3-methylitaconate (Scheme 2) and requires coenzyme B as ~~~~~~~ ~~ ~ ~ ~ ~ Wovgang Buckel studied chemistry in Munich and then moved to bioclzernistry working under F. LynenlH. Eggerer for his PhD and postdoctoral studies in which chiral acetates were developed with J. W.Cornforth. He spent 1970171 with H. A. Barker at Berkeley which stim- ulated his interests in energy metabolism and radical reac-tions in anaerobic bacteria. He has been professor of microbi- ology at Marburg since 1987 where he also cycles up and down the hills. w -o2,+R40*-02cH& K = 0.07 Me HR 2-meth yleneg lutarate 3-methylitaconate Scheme 2 Interconversion of 2-methyleneglutardte and (R)-.?-methylita- conate catal ysed by coenzyme B ,,-dependent 2-methylenegulatarate mutase cofactor. Hence the reaction catalysed by 2-methyleneglutarate mutase is very similar to that of glutamate mutase. Both enzymes remained as microbial curiosities until cloning and over-expres- sion of their genes in Esclzerichiu coli enabled the production of relatively large amounts of homogeneous ap~-proteins.~-~ Today there are three distinct classes of molecular rearrangements known to be catalysed by an enzyme in partnership with coenzyme B (see Table 1). In these reactions a migrating group X and a hydro- gen atom exchange places on adjacent carbon atoms. The reac- tions differ according to the nature of the migrating group X and the substituent Y (H or OH) at the adjacent carbon from which the hydrogen atom is abstracted. In this review we focus on the mech- anisms of the reactions catalysed by glutamate and 2-methyleneg- lutarate mutase. We aim to stimulate the reader to learn more about the fascinating chemistry of the radicals postulated as intermedi- ates in these arrangements. 2 Mechanism of Action -Initial Conclusions Early studies of coenzyme B ,,-dependent enzymatic reactions demonstrated that the essence of coenzyme function was to be found in the 5’-methylene group of the 5’-deoxyadenosyl residue bound to cobalt. It was shown by isotopic labelling that the migrating hydrogen became attached to this group leading to the supposition that 5 ’-deoxyadenosine is an intermediate. Bernard Golding studied chemistrv at Manchester. where he worked with Rod Rickards (15 u Ph D student. Pos tdoctoral work in the mid-5ixtier with Albert Esclzenmoser on the s wi-thesis of vitamin B led to investigations of the mode cf action of B coerizvmes. He i profes.wr of orguriic chemistry and currentlv head of depart-ment at Newcastle wherr he also has research group study-ing carcinogenesis arid unti-cancer drug design. arid (I bicvcle to evade the city’s trafic. 329 CHEMICAL SOCIETY REVIEWS 1996 the higher stability of the enzyme from the latter organism Box B Nomenclature Glutamate mutase is composed of two components E a dimer (E 1b Me cc:" In this review coenzyme B refers to the substance adenosyl- cobalamin (AdoCbl) in which R = 5'-deoxyadenosyl in the above structure By definition all cobalamins contain 5,6-dimethylbenzimidazole Coenzyme B is also a cobamide (i e any B ,,derivative as shown with a heterocyclic base connected to the lower ribose) In pseudo-vitamin B ,,the base is adenine connected to ribose at N-7 and ligated to cobalt at N-9 In dicyanocobalamin the 5,6-dimethylbenzimidazolehas been displaced by cyano but remains connected to ring D through the ribose-phosphate-propanolamine however the adenosyl has been replaced by a cyano group HO OH Y-' 5'-deoxyadenosyl= H& I NH2 Homolytic cleavage of the Co-C u-bond generates cob( r1)alamin and the 5'-deoxyadenosyl radical which should be reactive enough to abstract a hydrogen atom even from an unactivated position (e g methyl group) of a substrate molecule SH (see Scheme 3) The substrate-derived radical S* rearranges to a product-related radical P- which is quenched by 5'-deoxyadeno- sine with regeneration of the 5'-deoxyadenosyl radical and for- mation of product PH lo The mode of conversion of S-into Pa has been the subject of controversy At one extreme it has been pos- tulated that the rearrangement proceeds via organic radicals with cob(~)alamin as a mere spectator lo At the other extreme the cob(r1)alamin may conduct the rearrangement through organocorrinoid intermediates I I In Sections 3-7 experimental evidence pertaining to the mechanistic questions is presented leading to decisions about the mode of action of glutamate and 2- methyleneglutarate mutase It is concluded that the most likely mechanisms involve fragmentation of S. to the alkene acrylate and a carbon-centred radical (X-,i e 2-glycinyl radical for gluta- mate mutase and 2-acrylylyl radical for 2-methyleneglutarate mutase) which recombine to give P- perhaps with the assistance of cob(I1)alamin In the conclusion to this review the extension of this mechanism to other coenzyme B ,,-dependent reactions (see Table 1 ) is considered Are glutamate and 2-methyleneglutarate mutase paradigms for coenzyme B I ,-dependent enzymes? 3 Enzymology Glutamate mutase was first isolated from C tetunomorphum2 and more recently from the related C cochlearium l2 The enzymes are very similar in their properties the only significant difference being m = 107 600) and S a monomer (a,14700) The genes coding for the polypeptide chains cr and E have been cloned from both organisms in E coli they were designated as mut genes in C tetanomorph~m~-~and as glm genes in C cochleurium In both organisms the genes are clustered in the same order mutS-mutL-mutE-bma6 and glmS-glmL-glmE-bma? respectively The S-and E-genes code for the corresponding glutamate mutase components whereas the L-genes possibly code for proteins that act as molecu- lar chaperones but are not required for functional expression of the E- and S-genes in E coli The fourth gene of both clusters bma codes for the consecutive enzyme in the glutamate fermentation pathway P-methylaspartase 1(2S,3S)-3-methylaspartateammonia lyase EC 4 3 1 21 I? The deduced amino acid sequences of MutE and GlmE show 90% identity to each other but no significant simi- larity to any other known protein In contrast MutS and GlmS which are only 82% identical to each other share significant amino acid sequence similarities to domains of the cobalamin-dependent enzymes 2-methyleneglutarate mutase (see below) methyl- malonyl-CoA mutases from several microorganisms and mammals including humans (EC 5 4 99 2) as well as methionine synthase from E coli (MetH EC 2 1 1 13) The separate overexpression of the glnzE-and glmS-genes in E coli followed by simple two-step purifications led to homogenous components E and S containing not a trace of a cobamide Addition of an excess of coenzyme B to a mixture of both components immediately led to an active enzyme Activity was not only observed with mixtures of the components from the same organism but also with MutE + GlmS and GlmE + MutS which were active in the presence of the coenzyme (U Leutbecher and W Buckel unpublished) Upon gel filtration the active enzyme composed of GlmE and S eluted as a complex ~p,,which contained 1 0 coen- zyme Hence the coenzyme glues both components together This is not the case however with MutE and S which even in the pres- ence of the coenzyme separate on a gel filtration column Interestingly GlmS is able to bind the coenzyme alone albeit in sub-stoichiometric amounts (0 5 molhnol) whereas GlmE contains no trace of coenzyme after incubation with an excess of adenosyl- cobalamin followed by gel filtration The binding of the coenzyme to GlmS alone is consistent with the amino acid sequence similari- ties of MutS and GlmS with the cobalamin binding domains of other enzymes (n b the binding of the coenzyme to MutS is probably too weak to be observable by gel filtration) It should be mentioned that all these binding experiments were performed with the commer- cially available adenosylcobalamin (coenzyme B rather than with the natural coenzyme which in the case of C tetanomorphum was identified as the corresponding derivative of pseudo-vitamin B I Before the recombinant rnutases became available it was erroneously assumed by all workers in this field that the coenzyme binds to component E rather than to S Now it is obvious that the former purifications of component E from the original clostridia gave a mixture of the colourless apoprotein as well as an inactive E,U complex containing aquocobamide and cob(rr)amide most probably derivatives of pseudo-vitamin B Only GlmE was recently obtained from its native organism C cochlearium as a homogeneous highly active (specific activity 11 s I) colourless apoprotein devoid of any GlmS l4 2-Methyleneglutarate mutase was purified from C barkeri in the dark (n b the Co-C bond of adenosylcobalamin is sensitive towards light) as an apparent homotetramer (a4,267 000) containing up to two adenosylcobalamin and 0 1-0 2 cob(1i)alamin l6 In contrast to glutamate mutase the enzyme required no additional adenosyl- cobalamin for activity owing to its content of this coenzyme Upon treatment with 8 mol dm urea followed by dialysis against buffer the enzyme was completely inactivated but had lost only about half of the cobalamins Addition of adenosylcobalamin restored the activity almost completely Hence the enzyme contains at least four different cobalamin species which are distinguished by their content of 5'-deoxyadenine and by their binding to the enzyme It can be concluded that only those coenzyme B 12 molecules that are reversibly attached to the enzyme yield activity GLUTAMATE AND 2-METHYLENEGLUTARATE MUTASE-W BUCKEL AND B TGOLDlNG 33 1 Table 1 Coenzyme B,,-dependent enzymes (a b and Y are variable substituents; X IS the migrating group -OH NHT or a carbon- centred group) 7 v a-c-c-b a-c-C-bI I I I YH YH n.b. For class I1 enzymes the species on the right-hand side of the above equilibrium loses HX with the formation of bCH,CYa With ribonucleotide reductase the product is the species on the rhs of the equilibrium with X =H Substituent a b X Y Ref Class I carbon skeleton mutuses 1 Glutamate mutase EC 5 4 99 1 2 2 Methyleneglutarate mutase EC 5 4 994 4 Isobutyryl-CoA mutase EC 5 499 3 Methylmalonyl-CoA mutase EC 5 4 992 coy co C02 Me H H H H 2-Glycinyl 2-Acry late Formyl-CoA Formyl-CoA H H H H 2,9,2 1 8,1520 32 39 Class I1 elirninases 5 Propanediol dehydratase EC 4 2 1 28 H H ,CH CF OH OH 42 6 Glycerol dehydratase EC 4 2 130 H CH,OH OH OH 42 7 Ethanolamine ammonia-lyase EC 4 3 1 7 H H CH NH OH 43 8 Ri bonucleottde-triphosphatereductase c 4‘of c 1’of OH OH 28 ECI 1742 ri bonucleotide ri bonucleotide Class 111 aminornutuses 9 P-Lysine-5,6-aminomutase EC 5 4 3 3 4-(3-Aminobutyrate) H N H H 44 10 D-Ornithine-45 aminomutase EC 5 4 3 4 3-~-Alanine H NH H 44 4-( D-2-Ami nobutyrate) coenzyme Q2 cob(ii)alamin (Ado-Cl+ =5’-deoxyadenosyl) H H radical P H Ad0Ado4 H Ado4 H H H H Scheme 3 Pathway for coenzyme B ,,-dependent enzymic reactions illustrated with 2-rnethyleneglutarate as substrate SH and (R)-3-rnethylitaconate as product PH Part (a)shows conversion of SH to the substrate derived radical S. part (b) shows conversion of S-into the product-related radical P- and hence product PH (Ado-CH =5‘-deoxyadenosyl) CHEMICAL SOCIETY REVIEWS 1996 The gene rngrrz encoding the single polypeptide a of which 2- methyleneglutarate mutase is composed has been cloned sequenced and overexpressed in E cofi By comparison with other cobalamin-dependent enzymes the C-terminus (cu 100 amino acids) of the deduced amino acid sequence was identified as the coenzyme-binding domain (see above) In a manner analogous to the genes coding for glutamate mutase,rngm is followed by the gene mil coding for the consecutive enzyme in the pathway of nicotinate fermentation 3-methylitaconate A-isomerase (EC 5 3 3 6) Like p-methylaspartase this enzyme eliminates the methine hydrogen from its substrate The homogenous overproduced apo-2-methyl- eneglutarate mutase contained no trace of a cobalamin consistent with the inability of E colr to synthesise these compounds But on addition of adenosylcobalamin holo-2-methyleneglutarate mutase was immediately obtained with a specific activity twice as high as that of the enzyme purified from C hurkeri (B Beatrix 0 Zelder F Kroll and W Buckel unpublished) Furthermore the reconsti- tuted enzyme also contained no cob(1r)alamin (see above) which was erroneously suggested to be required for activity Is The apo- enzyme was shown to bind one coenzyme B,* per two polypeptides suggesting a similar structure to glutamate mutase from C cochleanurn since the size of the a-polypeptide (66 800) equals about that of E + r~ (68 500) Hence in 2-methyleneglutarate mutase the part of the enzyme containing the active site corre- sponding to subunit E of glutamate mutase apparently is fused together with the coenzyme-binding domain 4 Cryptic Substrate Stereochemistry The diastereotopic methylene protons at C-4 of glutamate and the enantiotopic methylene protons at C-4 of 2-methyleneglutarate are expected to be distinguished by the respective enzymes Isotopic labelling has revealed that H,y,is removed from glutamate,I7 whilst HRe is removed from 2-methyleneglutarate Considering the absolute configurations of the substrate and product molecules it was deduced that both enzymes cause an inversion of configuration Ado% at C-4 during the sequence of hydrogen abstraction and group migration The stereochemical data described are summarised in Scheme 4 To determine the stereochemistry of formation of the methyl groups of (2S,3S)-3-methylaspartate and (/?)-3-methylitaconate from their respective precursors all three hydrogen isotopes have been applied (R)-2-Oxo[ 3-2H I ,3-3H lglutarate was prepared by heating 2-oxoglutarate in D,O and incubating the resulting 2-0x013- 2H,jglutarate with isocitrate dehydrogenase in tritiated water After conversion into (2S,3S>-13-,H I ,3-3H)glutamate the labelled amino acid was fermented with C tetunornorphurn to give labelled butyrate from which chiral acetate was obtained applying two con- secutive Schmidt degradations The acetate contained cu 90% of the original tritium but was racemic (2S,3S)-[3-2H .3- 3HjGlutamate prepared by introducing the hydrogen isotopes in the reverse order also gave racemic acetate l9 This result supports the postulated intermediacy (see Section 2) of a methylene radical cor- responding in structure to 3-methylaspartate (cf Scheme 5) However the experiment should be repeated with purified gluta- mate mutase to exclude the possibility that the racemisation is caused by another enzyme in the multistep fermentation pathway A similar approach to that described for glutamate was used with 2- methyleneglutarate and again there was an apparent racemisation (G Hartrampf P Sanchez J W Cornforth and W Buckel unpub- lished results) In an approach intended to probe for the intermediacy of a cyclo- propylcarbinyl radical in the 2-methyleneglutarate mutase reaction (E)-2-(rnethylet~e-~HI )methyleneglutarate was synthesised and shown to equilibrate with its (a-isomer on exposure to the enzyme 2o The significance of this result is discussed in Section 8 5 Kinetic Properties Both glutamate and 2-methyleneglutarate mutase are highly spe- cific for their respective substrates Despite a wide ranging search no other substrates have been discovered * l5 23 The reason for this H AdoA ,,H H AdoAH Scheme 4 Stereochemistry of the 2 methyleneglutarate reaction (a)and glutamate mutase reaction (6) (a)HKUIS abstracted from C 4 of 2 methyleneglu tarate the abstracted H mixes with the 5’ methylene hydrogens of adenosylcobalamin the acrylate residue migrates to this C-4 with inversion of config uration (b)HStis abstracted from C 4 of (S) glutamate the abstracted H mixes with the 5’ methylene hydrogens of adenosylcobalamin the glycinyl residue migrates to this C 4 with inversion of configuration GLUTAMATE AND 2-METHYLENEGLUTARATE MUTASE-W BUCKEL AND B TGOLDING li Scheme 5 Equilibration of 2 methyleneglutarate and (R)3 methylitaconate and their corresponding radicals either rva a cyclopropylcarbinyl radical (path (I) or by fragmentation to acrylate and the 2 acrylate radical (path 6) unusual specificity could lie in the unique mechanism whereby three carbon centres are implicated in the molecular rearrange- ment Not even fluoro- or methyl-substitution of glutamate permits sub- strate behaviour although some such derivatives are inhibitory (see Table 2) l6 21-23 Of the diastereoisomeric 4-fluoroglutamates only the (2S,4S)-isomer is inhibitory presumably because the fluoro substituent replaces the hydrogen that is normally abstracted the (2S,4R)-isomer does not interact with the enzyme rac-2-Methylglutarate does not inhibit glutamate mutase whereas 2- methyleneglutarate is an active inhibitor presumably by occupying the glutamate binding site in a manner that cannot be matched by either enantiomer of 2-methylglutarate The inhibition of glutamate mutase by 2-methyleneglutarate was taken as an additional argu- ment for the intermediacy of an imino derivative in the glutamate mutase reaction I2 However the recent discovery (see Section 8) that glutamate mutase is inhibited synergistically by acrylate and glycine2l may explain this result if the key structural feature of 2- methyleneglutarate is the presence of an acrylate moiety that occu- pies the acrylate binding site Both enantiomers of 3-methylitaconate are inactive with glutamate mutase It was origi- nally reportedI that (S)-3-methylitaconate inhibits glutamate mutase but this is now known to be due to inhibition of the auxil- iary enzyme j3-methylaspartase a component of an assay system for glutamate mutase Using (S)-12,3,3,4,4-2H,]glutamate as a substrate? a kinetic isotope effect V,lV = 7 was observed whereas K = 2 4 mmol dm remained similar to that observed with the unlabelled amino acid (1 5 mmol dm 7 By applying regiospecifically labelled glu- tamates it was shown that only a hydrogen at C-4 most likely H$ which is abstracted by the 5'-deoxyadenosyl radical contributes much to the rate-limiting step 23 The isotope effect of the transfer of a tritium from the coenzyme to the substrate i Y from 5'-deoxyadenosine to the product-related radical was estimated as 13 5-18 from which a deuterium isotope effect of 6-7 4 was cal- culated In this experiment It was shown that the hydrogen was removed from the coenzyme at a rate comparable to that of its appearance in the product 3-methylaspartate This important obser- vation excludes the intermediacy of a protein-based radical in the catalytic turnover with the coenzyme acting merely as a radical ini-tiator 24 Already 25 years ago Kung and Stadtman reported a series of inhibitors for 2-methyleneglutarate mutase,2* among which ita- conate proved to be the most effective Ih An intriguing result of the earlier work was the purported inhibitory action of tram-1-methyl-cyclopropane- 1,2-dicarboxylate with the corresponding c u-isomer being less active these compounds were regarded as analogues of a presumed intermediate Using a continuous optical assay with homogenous 3-methylitaconate A-isomerase as an auxiliary enzyme it was recently shown that none of the four stereoisomers Table 2 Substrates and inhibitors of glutamate and 2-methyleneglutarate mutase rlJ' Enzyme Compound Effect Km or K,lmmol dm Ref Glutamate mutase (S) Glutamate (2R,3RS) 3 Fluoroglutamate (2S,4S) 4-F1uoroglutamate 2 Methyleneglutarate Glycine + acrylate (2S,3S) 3 Methylaspartate Substrate Competitive inhibition Competitive inhibition Competitive inhibition Inhi bition Substrate 15 06 0 07 04 CYI 5 each 05 12 12 12 12 12 21 2 Methyleneglutarate mutase 2 Methyleneglutarate (R)-3Methylitaconate Itaconate Mesaconate (methylfumarate) Succinate Acrylate Substrate Substrate Com peti tive inhi bi tion Competitive inhibition Competitive inhibition Inhibition 37 < 07 >1 >I ca 1 -10 (see text) 16 16 22 22 21 Glutamate mutase was not inhibited by (R) glutamate (R) or (S) 3 methylitaconate 4 mmol drn '(2s4R) 4 fluoroglutamate 2 methyl 3 methyl z1 4 methyl 23 N methylglutamate (10 mmol dm each) 20 mmol dm glycine 20 mmol dm acrylate 10 mmol drn '(S) aspartate ' 2 Methyleneglutarate mutase was not inhibited by 10 mmol dm (S) glutamate 15 mmol dm (RS)2 methylglutarate 20 mmol dm of all four stereoisomers of I methylcyclopropane 1 2 dicarboxylate Not determined 334 of 1-methylcyclopropane- 1,2-dicarboxylate was able to inhibit sig- nificantly 2-methyleneglutarate mutase The very recent discov- ery of the inhibitory power of the simple compound acrylate leads to an entirely new mechanistic proposal which will be discussed below The plot of the reciprocal initial velocity as a function of the acrylate concentration (Dixon plot) fitted better to a quadratic equa- tion than to a linear one This agrees well with the requirement that two acrylate molecules mimic intermediates in the 2-methylene- glutarate mutase reaction A remarkable feature of coenzyme B I ,-dependent enzymes is their apparent ability to handle safely free radicals which would be highly reactive if detached from the protein 25 The enzymes are not perfect however because they are slowly destroyed in the presence of substrate a process which is accelerated by exposure to air9 Prolonged anaerobic incubation in the dark ( I5 h 37 "C) yields the 'inactive complexes' of glutamate and 2 methyleneglutarate mutase in which the enzyme-bound coenzyme Biz has been con- verted into aquocobalamin and cob(~~)alamin (Section 6) The pres- ence of these inactive complexes in the native clostridia shows that this also happens in vivu I4-l6 It would be of interest to see whether these complexes are repaired degraded or simply washed out by the growing bacteria Remarkably the sensitivity of the coenzyme in 2-methyleneglutarate mutase towards light decreases during catalysis This experiment shows that the Co-C bond of the coenzyme the final target of light with h < 600 nm has already been cleaved during substrate turnover 6 Electron Paramagnetic Resonance Studies Electron paramagnetic resonance spectroscopy (EPR) has proved to be a powerful tool for gaining insights into the structure and func- tion of glutamate and 2-methyleneglutarate mutase According to the mechanism described in Section 2 EPR signals of cob(r1)alamin (gr,ca 2 3) and of an organic radical (g 2 00) should be observable during catalysis The first EPR spectra of the three carbon-skeleton rearranging mutases which were published in 1992,14 Is 27 clearly showed however only one 'catalytic' signal around g ca 2 1 This was generated by addition of the corresponding substrate to the EPR-silent mixture of enzyme and coenzyme Owing to the lack of sufficient amounts of enzymes and to the presence of inactive cob(1r)alaminin some preparations (see Section 3) the signals were of low resolution and therefore difficult to interpret The subsequent introduction of molecular biology into coenzyme B research led to the availability of large amounts of enzymes yielding intense EPR spectra with excellent resolution Thus addition of gluta mate to a mixture of component E with a twofold molar excess of component S (1 0-20 mg proteidsample) and a tenfold molar excess of coenzyme B I ,gave the expected signal in the g ca 2 1 region with an eightfold hyperfine splitting of the g line centred at 1 985 Comparison of this 'catalytic' spectrum with that of cob(~~)alaminrevealed similarities and differences The signal of cob(1I)alamin with g ca 2 3 was shifted to g ca 2 1 whereas the coupling constant (A = 106S G) of the eightfold hyperfine splitting of the g line was reduced to 50 G The characteristic threefold superhyperline splittings of each of the eight g lines due to coupling with the I4N nucleus of the axial base (I = 1) were not resolved The spectrum of the catalytic species was interpreted in terms of a tight coupling of the unpaired electron of cob(1I)alamin with that of a carbon-centred radical Recently the spectrum has been almost perfectly simulated by using parameters similar to those applied for the interaction of a thiyl radical with cob(i~)alamin in ribonucleo- side triphosphate reductase from LactobacilluJ leichmanni 28 The simulation of the spectrum of glutamate mutase revealed that Co" and an organic radical are coupled together by an isotropic exchange coupling which is at least 10 GHz This means that the electrons are interacting either directly via orbital overlap or indi- rectly via a super-exchanget mechanism In addition the electrons are interacting via a zero field splitting term of about 300 MHz A rough estimate of the distance between the two species gives -6 A (G Gerfen personal communication) The very similar and highly resolved spectrum obtained with 2-methyleneglutarate rnutase,' t Medinted hq orbital.. ofnnim acid re\tdue< or the w lvent CHEMICAL SOCIETY REVIEWS 1996 can be explained in the same way Interestingly the inhibitors of glutamate mutase (2S,4S)-fl uorogl utamate and 2-methylenegl u-tarate induced spectra similar but not identical to those induced by glutamate Double integration yielded spin concentrations up to ]SO% as compared to the concentration of component E These high values support the idea of a biradical being responsible for the EPR spectrum In contrast (S)-glutamate induced a spectrum with only 50% spin concentration Freeze-stopped experiments showed that glutamate induced the full EPR spectrum within less than 25 ms whereas (2S,4S)-fluoroglutamate required more than 5 s 12 Interestingly the combination of glycine plus acrylate but not the single compounds induced an EPR spectrum with glutamate mutase showing the characteristic signal at g ca 2 1 This is con- sistent with the synergistic inhibition of the enzyme by both sub- stances 21 Likewise addition of acrylate to 2-methyleneglutarate mutase afforded an EPR spectrum similar to that obtained with 2- methyleneglutarate2I or the competitive inhibitor itaconate l6 In summary the EPR spectra suggest that during catalysis the sub- strate-derived radical closely interacts with Co" Although a direct coordination of the radical by ColI should result in radical pairing and give rise to an EPR-inactive species acrylate itself could coor- dinate to Co" with the 2-glycinyl or 2-acrylate radical located nearby (see Figure 1) The nature of the axial base coordinated to the cobalt of co- enzyme B was also revealed by EPR spectroscopy p-Cresolylcobamide in the Co" state does not show the characteristic threefold superhyperline splitting of the g lines due to the absence of an axial nitrogen base (base off) Upon binding to a methyl- transferase however this splitting occurs indicating coordination to a i4N-containing ligand The latter was identified as histidine by incorporation of (imidazole-isN)histidineinto the protein Now a twofold splitting was observed due the spin 1 = 1/2 of the I5N nucleus 29 X-Ray crystallography of the coenzyme B binding domain of the methionine synthase (MetH) from E cofi directly showed the coordination of the conserved histidine residue (see section 3) to the cobalt atom 30 Mixing of unlabelled component E and glutamate with the completely isN-labelled component S yielded an EPR spectrum of the catalytic species with sharper g lines By using a histidine-requiring mutant of E cofi a I5N-labelled component S was prepared in which only the histidines remained unlabelled Its EPR spectrum could not be distinguished from that of the completely unlabelled component S indicating the coordination of a histidine to the cobalt Upon formation of cob(r1)alamin by prolonged incubation of the completely labelled component S with unlabelled component E and glutamate the typical twofold superhyperfine splitting of the g lines was observed demonstrating that a histidine also coordinates to the inactive species 31 The conserved histidine residue 359 of 2-meth- yleneglutarate mutase which was identified by sequence alignment with methionine synthase? * 30 was converted into a glutamine residue by site-directed mutagenesis The resulting mutant was completely inactive since it was not able to bind coenzyme B Furthermore wild type 2-methyleneglutarate mutase is also active when combined with adenosyl-p-cresolylcobamide,demonstrating the low importance of the axial base of the coenzyme for biological activity (E Stupperich F Kroll and W Buckel unpublished) H adoTH H &7 Figure 1 Postulated intermediate state in the glutamate mutase reaction showing an acrylate cob(ii)alamm complex GLUTAMATE AND 2-METHYLENEGLUTARATE MUTASE-W BUCKEL AND B TGOLDING (la-3a) matching in structure the species proposed as intermedi- ates in the 2-methyleneglutarate reaction (see Scheme 6),gave pn- manly di-tert-butyl 2-methyleneglutarate presumably via the corresponding free radicals The bromides were also reacted with glutamate glutamate dehydrogenm NADH + NH4+1 '"C cobaloxime(I) which gave the alkylcobaloxime lb from bromides la and 3a and the alkylcobaloxime 2b from bromide 2a Alkylcobaloxime 2b did not readily rearrange into alkyl-cobaloxime lb It was therefore proposed that this lack of reactiv- ity of organocobalt species 2b compared to the corresponding free H T D radicals supports the mechanism of Scheme 6 path a (see also loenzymes'2' Scheme 3 and Section 8) Murakami and his coworkers have described attempts to model glutamate mutase by preparing an organocorrinoid bearing an alkyl group derived from 3-methylaspartate and photolysing this mater- ial in a micellar matrix to give glutamate in low yield 35 This is the only model system to achieve the conversion of 3-methylaspartate into glutamate but further studies are needed to elucidate the reac tion pathway 8 Mechanism of Action -Decision For 2-methyleneglutarate mutase model studies (see Section 7) supported a mechanism in which the substrate-derived and product- related radicals are interconverted via an intermediate cyclopropyl- carbinyl radical (see Scheme 6 path a) A similar mechanism was however impossible for glutamate mutase because of the lack of suitable .rr-bond with which the radical centres in S. and P. could interact The proposal7 that such a n-bond could be generated by formation of an imine from the amino group of glutamate and a car- bony1 function within the protein cannot be sustained (see Section 3) Furthermore a mechanism whereby the 2-glycinyl moiety migrates via a bridged transition state can also be excluded because of the predicted high energy of such a species M However the dis- covery of synergistic inhibition of glutamate mutase by glycine and acrylate gave the first experimental support for a fragmen-tation-recombination mechanism for this enzyme (Scheme 7a) already proposed many years ago 36 37 The similarities between 2 methyleneglutarate and glutamate mutase with respect to the reac- tions catalysed enzymes cofactor and EPR data point to a commonality of mechanism We have therefore proposed frag- mentation-recombination mechanisms for both of these enzymes in which acrylate is a common intermediate 21 Such a mechanism for 2-methyleneglutarate mutase arose during our studies of the four isomers of 1 -methylcyclopropane- 1,2-dicarboxyIates as potential inhibitors of 2-methyleneglutarate mutase The surprising failure of any of these compounds to inhibit 2-methyleneglutarate mutase and especially the inactivity of the (R,/?)-isomer led us to re-evalu- ate the long-held mechanistic hypothesis of Scheme 6 path a The startling discovery that 2-methyleneglutarate mutase is inhibited by acrylate,2' with a square dependence on acrylate concentration sup- ports the mechanism of Scheme 6 path b for this enzyme (see also Scheme 76) The stereochemical features of the glutamate and 2-methylene- glutarate mutase reactions (see Section 4) are fully explicable by the mechanisms of Scheme 6 path 6 and Scheme 7 The stereo chemistry of the initial hydrogen abstraction will be governed by the precise positioning of the adenosyl radical with respect to the protein-bound substrate The fact that H is removed from gluta- mate whilst HReis removed from 2-methyleneglutarate is not sur- prising even though the active sites of the enzymes may be similar Thus a 120" rotation about the C(3)-C(4) bond causes a lateral movement of the carboxylate of only ca 2 8 and serves to present either H or HRU to the adenosyl radical Small differences in TD butyrate (2s 3S)-J-methylaspartate (2S)glutamate 2 x Schmidt dagradatmnI T 4 steps V 2-methykneglutarate mutaseHi&'" c-coenzymeBl2D T(3R)-3-methylitaconate 2-mthylenaglutarate Scheme 6 Syntheses of (2S,3S) [3 2H,3 3H]glutamate and (3s)2 methyl enel 3 'H.3 ?Hjglutarate their conversion to chiral methyl labelled (2S,3s)3 methylaspartate and (3R) 3 methylitaconate respectively and the degradation to chiral acetates The conversion of glutamate to butyrate was performed with growing cells of C tetunomorphurn Recently the coordination of a histidine nitrogen to cobalt within the enzymes catalysing carbon-skeleton rearrangements has been confirmed by the crystal structure of methylmalonyl-CoA mutase from P shermanii revealing an extraordinary long Co-N distance of 2 53 A The observed 0 32 A difference from the corresponding Co-dimethylbenzimidazole bond length in free adenosylcobalamin (2 21 A)is thought to facilitate homolysis of the Co-C bond 32 It would be of interest to measure the Co-N distance in methionine synthase which catalyses heterolytic cleavage of the Co-methyl bond 7 Model Studies Possible mechanisms for the equilibration of 2-methyleneglu- tarate with 3-methylitaconate catalysed by 2-methyleneglutarate mutase are shown in Scheme 6 33 The key intermediate in path a is a cyclopropylcarbinyl radical which by cleavage of its C( 1)-C(2) bond connects with the substrate-derived radical whilst cleavage of the C( 1)-C(3) bond leads to the product-related radical (cf Scheme 3) There is ample precedent for these processes in non-enzymatic chemistry Thus conversion of the cyclopropylcarbinyl radical to the but-3-enyl radical is one of the fastest unimolecular reactions known (k = los s I at 298 K) whilst the reverse reaction is also relatively fast (k = lo3s I at 298 K) Many examples of these types of interconversion have been descnbed in which the butenyl or cyclopropylcarbinyl system bears alkyl aryl and/or ester substituents 34 It has been shown33 that treatment with triphenyltin hydnde of each of the bromides H.. Co+Bu tB?73 "pX t Bu@C la X=Br 2a X=Br 3a X=Br 1b X = Co(dmgHhpy 2b X = Co(dmgHkpy 3b X = Co(dmgHkpy CHEMICAL SOCIETY REVIEWS 19% 11 li H.. H 11 H a b Scheme 7 Proposed reaction pathways for coenzyme B ,,-dependent reactions CoA mutasej (from ref. 21). protein structure especially with respect to the positions of car- boxylate-binding functions could suffice to bring about this alter- ation. Fragmentation of the substrate-derived radical requires a specific conformation in which its C(2)-C(3) bond is nearly parallel to the p orbital at the radical centre (see Scheme 7). This leads to acrylate and a 2-glycinyl radical from glutamate and acrylate and the 2- acrylate radical from 2-methyleneglutarate. Addition of the 2- glycinyl radical to the Re face at C-2 of acrylate leads to a product-related radical of correct stereochemistry [i.e. that corre- sponding to (2S,3S)-3-methylaspartate1. Provided that addition of the 2-acrylate radical occurs to the Si face of acrylate a product- related radical of correct stereochemistry is also generated [ i.e. that corresponding to (R)-3-methylitaconate 1. Both processes lead to the observed inversion of configuration at the centre from which hydro- gen is abstracted and to which a group migrates. Throughout the processes described the migrating group remains in contact with a particular face of the acrylate molecule. The observed equilibration of (E)-(methylene-2H,)2-methyleneglutarate with its (2)-isomer catalysed by 2-methyleneglutarate mutaseZ0 (see Section 4 and Scheme 6) can be explained by noting the linearity of the 2-acrylate radical and postulating a time dependent rotation of this radical either about the C( I)-C(2) bond or the C( I)-C(Z)-C(3) axis that is slower than substrate turnover (see Scheme 6 path h).This result however is also consistent with the intermediacy of a cyclopropyl-carbinyl radical (Scheme 6 path a).The fragmentation-recombi- nation route discussed for the Class I enzymes glutamate and 2-methyleneglutarate mutase (see Table 1) can be immediately ;*2-H$co H CoAS 0 H H CoAS-( 0 C I (a)glutamate mutase; (b)2-methyleneglutarate mutase; (c) methylmalonyl-applied to methylmalonyl-CoA mutase (Scheme 7c) with acrylate and the 2-formyl-CoA radical as intermediates. In support of this proposal recent studies in Karlsruhe have shown that methyl- malonyl-CoA mutase is synergistically inhibited by acrylate and formyl-CoA (A. Abend and J. Rktey personal communication). The mechanism shown in Scheme 7ccontains a stereochemical subtlety. In contrast to the glutamate and 2-methyleneglutarate reactions which both proceed with inversion at C-4 of substrate the transfor- mation of methylmalonyl-CoA to succinyl-CoA takes place with retention of configuration at the corresponding carbon centre. This observation can be explained if the acrylate exists in two confor- mations (see Scheme 7c) which interconvert by rotation about their C( l)-C(2)-bond. This enables 'the error in the cryptic stereochem- istry of methylmalonyl-CoA mutase'17.38 to be understood. In addi-tion to the expected migration of H at C-3 of succinyl CoA to the methyl group of methylmalonyl-CoA the 'exchangeable hydrogen' at C-2 of methylmalonyl-CoA migrated to C-3 of succinyl-CoA. To elucidate this result a 1,2-hydrogen shift in the intermediate suc- cinyl-CoA radical was invoked but it can be better explained if there is an occasional removal of H,! from C-3 leading directly to the correct acrylate conformation for further elaboration to (R)-methylmalonyl-CoA.2' Recently it has been shown that isobutyryl- CoA muta~e,3~ to which the fragmentation mechanism may also be applied proceeds with retention of c~nfiguration.~~ The stereo- chemical 'error' made by this enzyme which has to handle propene according to the fragmentation-recombination mechanism is even more pronounced than that observed with methylmalonyl-CoA mutase. GLUTAMATE AND 2-METHY LENEGLUTARATE MUTASE- W BUCKEL AND B TGOLDING 9 Concluding Comments Remarkably the coenzyme B ,,-dependent eliminases and amino- mutases (Table 1 ) have coenzyme B ,,-independent counterparts which catalyse essentially identical reactions The existence of several coenzyme B ,,-independent ribonucleotide reductases is well established ** In addition a coenzyme B ,,-independent but iron- containing diol dehydratase has been reported The iron-sulfur dependent lysine-2,3-aminomutase also uses the 5’-deoxyadenosinyl radical but this reagent is derived from S-adeno- sylmethionine (SAM the “poor man’s coenzyme B,,”) 41 On the other hand no coenzyme B ,,-independent counterpart to the carbon- skeleton mutases (Class I ,Table 1) has been discovered yet This may due to the fact that only the Class I enzymes require Corifor coordi- nation of the acrylate in order to enable the addition of the radical fragment to the a-carbon of this intermediate,leading to the branched products Hence in these enzymes Coilmight act not as a mere spec- tator but as a conductor of the catalysis In contrast the Class I1 and 111 enzymes apparently require coenzyme B ,,only as a generator of 5 ’-deoxyadenosyl radicals The intermediacy of such a hypothetical Coil-acrylate 7r-complex (Figure 1) in the catalysis of the carbon- skeleton mutases causing the enhancement of the reactivity of the a-carbon of acrylate remains however to be established Soon after his discovery of glutamate mutase H A Barker wrote ‘the precise role of the coenzyme in the interconversion of gluta- mate and P-methylaspartate is not yet known’ I Nearly 40 years later one may begin to understand Acknowledgements This work was supported by grants from the Engineering and Physical Sciences Research Council Commission of European Communities Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie The authors wish to thank their coworkers Birgitta Beatrix Harald Bothe Gerd Broker Chris Edwards. Friedrich Kroll and Oskar Zelder for their enthusiasm in performing the latest experiments leading to new insights into the mechanism of action of coenzyme B 10 References I H A Barker. H Weissbach and R D Smyth Proc Narf Acad Sci USA 1958,44.1093 2 R L Switzer In ‘B,2’. ed D Dolphin Wiley New York 1982 vol 2. p 289 3 P G Lehnert and D C Hodgkin. Nature. 1961,192.937 4 H F Kung,S Cederbaum.L Tsai andT C Stadtman,Proc Narl Acad Sci USA 1970.65,978 5 E N G Marsh and D E Holloway FEBS Letr 1992,310 167 6 D E Holloway and E N G Marsh J Biol Chem . 1994,269,20425 7 M Brecht J Kellermann and A Pluckthun. FEBS Lett 1993,319,84 8 B Beatrix 0 Zelder D Linder and W Buckel Eur J Biochem 1994 221,101 9 0 Zelder B Beatrix U Leutbecher and W Buckel Eur J Biochem 1994,266,577 10 B T Golding Chem Br .1990,26,950,and references therein 11 E I Ochiai in ‘Metal Ions in Biological Systems’ ed H Sigel and A Sigel Marcel Dekker New York 1994,vol 30 p 255 12 U Leutbecher R Bocher D Linder and W Buckel Eur J Biochem 1 992,205,759 13 H A Barker R D Smyth R M Wilson and H Weissbach J Biol Chem 1959,234,320 14 U Leutbecher S P J Albracht and W Buckel. FEBS Lett 1992.307 144 15 C Michel S P J Albracht and W Buckel Eur J Biochem ,1992,205 767 16 0 Zelder and W Buckel Biol Chem Hoppe Seyler 1993,374,85 17 J Retey and J A Robinson ‘Stereospecificity in Organic Chemistry and Enzymology’ Verlag Chemie Wernheim 1982,and references therein 18 G Hartrampf and W Buckel Eur J Biochem 1986,156,301 19 G Hartrampf and W Buckel FEBS Lett 1984,171,73 20 C Edwards B TGolding F Kroll B Beatrix. G Broker and W Buckel J Am Chem Soc 1996,118,4192 21 B Beatrix 0 Zelder F Kroll. G Orlygssoti B TGolding and W Buckel Angew Chetn Int Ed Engl 1995,34,2398.and references therein 22 H F Kung and T C Stadtman,J Bid Chem 1971,246,3378 23 B Hartzoulakis and D Gani Proc Ind Acad Sci (Chem Sci 1 1994 106,1165 24 E N G Marsh Biochemistrv 1995,34,7542 25 J Retey,Angew Chem 1990,102,373 26 C Michel and W Buckel FEBS Left 1991,281,108 27 Y Zhao,P Suchand J Retey,Angew Chem 1992,104,212 28 S Lrcht G J Gerfen and J Stubbe Science 1996,271,477,and refer ences therein 29 E Stupperich H J Eisinger and S P J Albracht Eur J Biochem 1990,193,105 30 C L Drennan S Huang J T Drummond R G Mathews and M L Ludwid Science 1994,266 1669 31 0 Zelder B Beatrix F Kroll and W Buckel FEBS Lett 1995,369 252 32 F Mancia. N H Keep A Nakagawa. P F Leadlay S McSweeney. B Rasmussen P Bosecke,O Diat and P R Evans.Structure 1996.4.339 and references therein 33 S Ashwell A G Davies B TGolding R Hay Motherwell and S Mwesigye Kibende J Chem Soc Chem Commun 1989,1483 34 D C Nonhebel Chem Soc Rev 1993,22,347 35 Y Murakami Y Hisaeda X M Song and T Ohno J Chem Soc Perkin Trans 2 1992,1527 36 B T Golding and L Radom J Am Chem SOC 1976,98,6331 37 B T Golding in ‘B,?’ ed D Dolphin Wiley. New York 1982,vol I p 543 38 W E Hull,M Michenfelderand J Retey,Eur J Biochem ,1988,173,191 39 B S Moore R Eisenberg C Weber A Bridges D Nanz and J A Rob1nson.J Am Chem Soc 1995.117 11285 40 M G N Hartmanrs in ‘Metal Ions in Biology’ ed H Sigel and A Sigel Marcel Dekker New York 1994. vol 30. ch 7 p 201 41 P A Frey. FASEB J . 1993.7,662 42 T TorayaandS Fukui,in ‘Blz’,ed D Dolphin,Wrley,New York 1982 vol 2,p 233 43 B M Barbior in ‘B,2’ ed D Dolphin Wiley New York 1982,vol 2 p 263 44 J J Baker and T C Stadtman in ‘BIZ’ ed D Dolphin Wrley New York. 1982.~012,p 203