Cloning and Sequencing of the Coenzyme B12-binding Domain of Isobutyryl-CoA Mutase from Streptomyces cinnamonensis, Reconstitution of Mutase Activity, and Characterization of the Recombinant Enzyme Produced inEscherichia coli *

Isobutyryl-CoA mutase (ICM) catalyzes the reversible, coenzyme B12-dependent rearrangement of isobutyryl-CoA to n-butyryl-CoA, which is similar to, but distinct from, that catalyzed by methylmalonyl-CoA mutase. ICM has been detected so far in a variety of aerobic and anaerobic bacteria, where it appears to play a key role in valine and fatty acid catabolism. ICM from Streptomyces cinnamonensisis composed of a large subunit (IcmA) of 62.5 kDa and a small subunit (IcmB) of 14.3 kDa. icmB encodes a protein of 136 residues with high sequence similarity to the cobalamin-binding domains of methylmalonyl-CoA mutase, glutamate mutase, methyleneglutarate mutase, and cobalamin-dependent methionine synthase, including a conserved DXHXXG cobalamin-binding motif. Using IcmA and IcmB produced separately in Escherichia coli, we show that IcmB is necessary and sufficient with IcmA and coenzyme B12 to afford the active ICM holoenzyme. The large subunit (IcmA) forms a tightly associated homodimer, whereas IcmB alone exists as a monomer. In the absence of coenzyme B12, the association between IcmA and IcmB is weak. The ICM holoenzyme appears to comprise an α2β2-heterotetramer with up to two molecules of bound coenzyme B12. The equilibrium constant for the ICM reaction at 30 °C is 1.7 in favor of isobutyryl-CoA, and the pH optimum is near 7.4. TheK m values for isobutyryl-CoA,n-butyryl-CoA, and coenzyme B12 determined with an equimolar ratio of IcmA and IcmB are 57 ± 13, 54 ± 12, and 12 ± 2 μm, respectively. AV max of 38 ± 3 units/mg IcmA and ak cat of 39 ± 3 s−1 were determined under saturating molar ratios of IcmB to IcmA.

The coenzyme B 12 -dependent isobutyryl-CoA mutase (ICM 1 ; butanoyl-CoA:2-methylpropanoyl-CoA mutase, EC 5.4.99.13) catalyzes the reversible rearrangement of isobutyryl-CoA to n-butyryl-CoA. Although closely related to the well known and widely distributed methylmalonyl-CoA mutase (MCM) reaction (1) (Fig. 1), MCM does not catalyze the rearrangement of isobutyryl-CoA to n-butyryl-CoA (2,3). ICM has been detected in several polyketide antibiotic-producing streptomycetes, where it appears to play a key role in valine and fatty acid catabolism as well as in the production of fatty acid-CoA thioester building blocks for polyketide antibiotic biosynthesis (4). In earlier work (5), purification of ICM from the monensin-producing microorganism Streptomyces cinnamonensis gave a protein (IcmA) of ϳ65 kDa whose gene was subsequently cloned and expressed in Escherichia coli. However, recombinant IcmA alone showed no ICM activity. Using IcmA with a His 6 tag attached to its N terminus, a second subunit of ICM with an apparent mass of Ϸ17 kDa as determined by SDS-polyacrylamide gel electrophoresis (denoted IcmB) was isolated by affinity chromatography from S. cinnamonensis, which gave ICM activity when combined with IcmA and coenzyme B 12 (5). In this work, we describe for the first time the cloning and sequencing of the gene encoding IcmB and the first characterization of ICM reconstituted from small and large subunits produced separately in E. coli.
The structure determination of the cobalamin-binding domain of MetH, a member of the methyltransferase family, revealed for the first time a protein-bound form of methylcobalamin, a vitamin B 12 derivative (10). The cobalamin was shown bound to the protein with a histidine residue providing an axial imidazole ligand to cobalt, replacing the DMB group appended to the corrin ring. This key histidine residue is found in the motif DXHXXG, which is conserved in some (but not all) of the coenzyme B 12 -dependent mutases (11). The crystal structure of MCM from P. shermanii revealed not only an active site, inaccessible to solvent, embedded along the axis of the (␤/␣) 8barrel in MutB (7,8), but also the coordination of cobalt in coenzyme B 12 by the histidine in the conserved DXHXXG motif within the C-terminal cobalamin-binding domain. In the case of ICM, however, the large subunit (IcmA) contains no contiguous cobalamin-binding domain, but instead requires a separate small subunit (IcmB) to bind coenzyme B 12 and to afford active mutase (5). This suggested that IcmB has taken on the role of a separate cobalamin-binding domain in ICM, a conclusion that is confirmed here by the high sequence similarity between IcmB and the cobalamin-binding domains of MutB, methyleneglutarate mutase, and MetH as well as the small subunit (MutS) of glutamate mutase.
Production of IcmA and Assay-IcmA was produced in E. coli BL21 pLysS[pET3a-icmA] as described earlier (5). The ICM assay was as described earlier (2). Briefly, n-butyryl-CoA (or isobutyryl-CoA; 280 M final concentration) was added to an assay mixture containing 50 M coenzyme B 12 , 50 mM potassium phosphate (pH 7.4), and IcmA and IcmB. The reaction was incubated in the dark at 30°C, typically until 10 -15% conversion of substrate to product, and then stopped by addition of 100 l of 2 M KOH containing 0.002% (v/v) n-valeric acid. After acidifying with 100 l of 15% (v/v) H 2 SO 4 and saturating with NaCl, the solution was extracted with EtOAc. The extract was analyzed directly by gas chromatography using a FFAP capillary column (10 m ϫ 0.53 mm; Hewlett-Packard Co.).
Purification of IcmB-IcmB was purified from both S. cinnamonensis and Streptomyces lividans TK64 as described earlier (5). The N-terminal amino acid sequences of both proteins were determined by the Edman method using an automated sequencer. The N-terminal sequences obtained were as follows: S. lividans, GVAAGPIRVVVAK-PGLDGHD; and S. cinnamonensis, GVAAGPIRVVVAKPGLDGHDRG-AKVIARAL.
Cloning icmB-General DNA manipulations were performed in E. coli (13) and Streptomyces (14) as described. Two oligonucleotides were designed based on the N-terminal protein sequence of IcmB (S ϭ G and C) (Oligo-1 and -2).
S. cinnamonensis genomic DNA digested with the restriction enzyme SalI was shown by Southern blotting to contain fragments of ϳ1.65 kb that hybridized to both oligonucleotide probes. DNA fragments in the size range 1.6 -1.8 kb were isolated from SalI-digested genomic DNA and ligated into SalI-cut pUC18 (15). A clone was isolated from this library using Oligo-1 as a probe. This clone (pOCI706) contained an Ϸ1.65-kb DNA insert, which was sequenced on both DNA strands by the dideoxy method using dye terminator chemistry (Perkin-Elmer). The GCG Version 8.1-unix software package (Genetics Computer Group, Madison, WI) (16) was used for sequence analysis. The sequence is shown in Fig. 2.
Expression of icmB-The 411-base pair icmB gene was amplified by polymerase chain reaction using the following oligonucleotides as primers and pOCI706 DNA as template: ICMBFOR, AGTCATCATATGGG-TGTGGCAGCCGGGCCGATCC; and ICMBBACK, ATCATGGATCCT-CAGACGGCCTGTCGCACGTTCC. The NdeI and BamHI restriction sites in the primers are underlined. After polymerase chain reaction, the product was digested with NdeI and BamHI and ligated between the corresponding restriction sites in pET3a (12). The resulting plasmid (pOCI711) was introduced into E. coli BL21(DE3) pLysS for production of IcmB.
Cells containing pOCI711 were grown in 2 ϫ 20 ml of LB medium containing ampicillin (50 g/ml) and chloramphenicol (34 g/ml) to A 600 nm Ϸ 0.7 at 30°C. The cells from this preculture were used to inoculate 8 ϫ 500 ml of 2YT medium (Difco) containing ampicillin and chloramphenicol. The cells were grown to A 600 nm Ϸ 0.6 with shaking at 200 rpm and 30°C. Protein expression was induced with isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.3 mM and shaken for an additional 4 h. The cells were harvested by centrifugation at 4°C and stored at Ϫ20°C.
The cell pellet from 1.5 liters of culture was sonicated in buffer A (50 mM Tris-HCl (pH 7.5), 5 mM EDTA, 1 mM benzamidine, 1 mM phenylmethanesulfonyl fluoride, 1 mM dithiothreitol, 0.1% (v/v) 2-mercaptoethanol, and 5% (v/v) glycerol). The cell-free extract (ϳ35 mg of protein) was applied to a Q-Sepharose column (1.5 ϫ 7 cm; Fast flow, Amersham Pharmacia Biotech) equilibrated with buffer B (50 mM Tris-HCl (pH 7.5) and 5 mM EDTA). The column was washed with 20 column volumes of buffer B and then eluted at a flow rate of 3 ml/min with a linear gradient of 0 -1 M KCl in buffer B over 100 ml. The fraction containing IcmB eluted in the range 100 -350 mM KCl as detected by an ICM activity assay. The ICM-active fractions were pooled (20 mg of protein) and dialyzed against buffer C (50 mM Tris-HCl (pH 7.5) and 0.5 mM EDTA) containing 1 mM dithiothreitol.
The dialyzed sample (18 mg of protein) was applied to a MonoQ column (HR 5/5, Amersham Pharmacia Biotech) equilibrated with buffer C. The column was washed with 50 column volumes of buffer C and then eluted at a flow rate of 1.5 ml/min using a 0 -1 M KCl gradient in buffer C over 100 ml. The fractions eluting between 90 and 200 mM KCl possessed the highest ICM activity and were pooled and dialyzed against buffer C (yield of 5 mg of protein).
The dialyzed sample (4.5 mg of protein) from above was rechromatographed on the MonoQ column under the same conditions as described above. IcmB eluted as a single peak (yield of 0.5 mg of protein) at 200 mM KCl and was homogeneous as determined by SDS-polyacrylamide gel electrophoresis (see Fig. 4). The protein was dialyzed against 50 mM potassium phosphate (pH 7.5) containing 1 mM dithiothreitol and stored at 4°C. IcmB can be stored in this way for several weeks without significant loss of activity.
Characterization of IcmA and IcmB-Native molecular masses were determined by gel filtration chromatography (Superose 12 HR 10/26 column, Amersham Pharmacia Biotech) in buffer D (50 mM potassium phosphate (pH 7.4), 150 mM KCl, and 0.5 mM EDTA). Standards for molecular mass calibration were blue dextran (2000 kDa), catalase (232 kDa), aldolase (158 kDa), bovine serum albumin (67 kDa), and ribonuclease A (13.6 kDa). The extinction coefficients of IcmA and IcmB at 280 nm were determined using the Beer-Lambert law and quantitative amino acid analysis for determination of protein concentrations: ⑀ 280 ϭ 122,000 Ϯ 12,000 for IcmA and ⑀ 280 ϭ 10,280 Ϯ 2100 for IcmB. Each measurement was carried out in triplicate, and the means Ϯ S.D. are given. The protein concentrations determined in this way were in good agreement with those from Bradford assays (17). The molecular mass of recombinant IcmB in deionized water was determined by mass spectrometry using an ion-spray source.
Equilibrium Constant Determination-The equilibrium constant for the ICM reaction at pH 7.4 and 30°C was obtained by determining the equilibrium concentrations of both n-butyryl-CoA and isobutyryl-CoA. Assays contained 25 nM IcmA, 25 nM IcmB, and 0.1 mM coenzyme B 12 in 3.2 ml of 50 mM potassium phosphate buffer (pH 7.4) equilibrated at 30°C. n-Butyryl-CoA or isobutyryl-CoA (140 M) was added to start the reaction; and periodically, samples were taken to determine the ratio of isobutyryl-CoA to n-butyryl-CoA. This experiment was then repeated using 10 different starting ratios of n-butyryl-CoA to isobutyryl-CoA (ratios in the range 0.4 -2.1) roughly in the range of the estimated equilibrium constant determined above. The equilibrium constant for the ICM reaction was obtained by plotting the change in [isobutyryl-CoA] versus the [isobutyryl-CoA]/[n-butyryl-CoA] ratio. This gave K eq ϭ 1.7 Ϯ 0.05 in favor of isobutyryl-CoA.
Kinetic Analyses-For determination of K m and V max values, initial velocities were determined using the gas chromatography assay (see above) in 50 mM potassium phosphate buffer (pH 7.4) at 30°C. For the measurement of K m values, IcmA and IcmB (each at 5 nM) were used in a 1:1 molar ratio. The conversion of substrate to product was followed by withdrawing samples from the assay mixture (typically five time points for each assay) up to a maximum of 10 -15% conversion, and these data were used to calculate initial rates. Typically, six different substrate  Table II).
The general form of the rate equation for a bireactant mechanism is represented by Equation

RESULTS
Sequence and Expression of icmB-The 1.65 kb of S. cinnamonensis genomic DNA, isolated from a library using oligonucleotides encoding the N terminus of IcmB, were sequenced and analyzed using CODONPREFERENCE in the GCG software (16). This revealed two complete ORFs (ORF1 and ORF2) and one (ORF3) with incomplete sequence data (Fig. 3). ORF1, starting at nucleotide 422 and ending at nucleotide 832, was identified as the icmB gene, encoding a protein of 136 residues with a mass of 14,333 Da. The putative ORF2, immediately upstream of ORF1, encodes a protein of only 55 residues and, like ORF3, lacks significant similarity to proteins in the EBI/ Swiss Protein Data Bank. The deduced amino acid sequence of IcmB shows significant identity to several cobalamin-dependent proteins in the data base, as shown in Table I.
To produce IcmB in E. coli, NdeI and BamHI sites were introduced at the 5Ј-and 3Ј-ends of icmB such that the NdeI site encodes an ATG start codon, and the gene was cloned in the expression vector pET3a (12). After introduction into E. coli BL21(DE3) pLysS, the production of active IcmB could be de-tected in cell extracts in an assay with IcmA and coenzyme B 12 . After isolation by standard methods (see "Experimental Procedures"), the protein was homogeneous as determined by SDSpolyacrylamide gel electrophoresis (Fig. 4), with an apparent mass of ϳ17 kDa, as observed in earlier work (5). An electrospray mass spectrum of IcmB gave a molecular ion with a mass of 14,203 Da, consistent with the deduced protein sequence minus methionine.
Characterization of ICM-Upon gel filtration chromatography, IcmB eluted with an apparent mass of 17 kDa, suggesting  Gel filtration of a 1:1 molar ratio of IcmA to IcmB led to the elution of the two subunits separately from the column at the expected apparent masses. When IcmA and IcmB were first incubated with coenzyme B 12 , a single major peak eluted from the gel filtration column with an apparent mass of Ϸ152 kDa. This peak showed a UV spectrum with max at 522 and 375 nm, characteristic of enzyme-bound coenzyme B 12 , and high ICM activity (3.3 mol/min/mg). As determined by SDS-polyacrylamide gel electrophoresis, this peak showed bands corresponding to both IcmA and IcmB and, as determined by densitometric scanning, in almost exactly a 1:1 molar ratio (Fig. 5). This indicates a likely ␣ 2 ␤ 2 -quaternary structure for native ICM with bound coenzyme B 12 . The pH optimum for the ICM reaction was ϳ7.4, and the activity diminished only slightly at pH 6.0 and 8.0. Kinetics-The steady-state kinetic properties of the enzyme were investigated. The two subunits and coenzyme B 12 are necessary and sufficient to observe ICM activity with n-butyryl-CoA or isobutyryl-CoA as substrate. The specific activity of the enzyme was seen to vary with the molar ratio of IcmA to IcmB in the assay. The activity approached saturation only after a severalfold molar excess of IcmB to IcmA was present (Fig. 6). The plot of activity against IcmB concentration appeared slightly sigmoidal, suggestive of some degree of cooperativity in formation of the holoenzyme from IcmA, IcmB, and coenzyme B 12 . Fitting the velocity data to Equation 2 gave a Hill coefficient of 1.34 Ϯ 0.02.
Under the standard assay conditions, as used during the earlier purification of IcmA (5), the specific activity of the enzyme in the presence of a 1:1 molar ratio of IcmA to IcmB was ϳ8.0 mol/min/mg. In the earlier work (5), after partial purification of IcmA from extracts of S. cinnamonensis, the specific activity was ϳ0.023 mol/min/mg, and after reconstitution of mutase from recombinant His 6 -IcmA and IcmB isolated from S. cinnamonensis, it was ϳ1.0 mol/min/mg.
For the determination of K m values, the initial rates were typically measured at six different substrate concentrations in the range 13.3-200 M, and for each fixed concentration of substrate, rates were typically determined at five different coenzyme B 12 concentrations between 4.8 and 200 M. A 1:1 molar ratio of IcmA to IcmB was used throughout. The data were fitted to Equation 1 by nonlinear least-squares regression using the computer program Leonora Version 1.0 (Fig. 7, A and  B). The apparent K m values determined were 54 Ϯ 12 M for n-butyryl-CoA, 57 Ϯ 13 M for isobutyryl-CoA, and 12 Ϯ 2 M for coenzyme B 12 . The V max was determined to be 38 Ϯ 5 units/mg IcmA from a Hanes plot (Fig. 7C) using varying concentrations of n-butyryl-CoA and saturating amounts of IcmB (15-fold molar excess compared with IcmA). Assuming one active site per IcmA monomer, k cat ϭ 39 Ϯ 3 s Ϫ1 at saturating concentrations of IcmB. DISCUSSION icmA and icmB are not adjacent in the chromosome of S. cinnamonensis. In contrast, mutA and mutB from this organism share overlapping start and stop codons (2), a device that may lead to transcriptional coupling and hence to the production of stoichiometric amounts of the two proteins. It is presently unclear how far apart icmA and icmB are in the S. cinnamonensis genome, although the regions extending over 8 kb upstream and 11 kb downstream of icmA, isolated in an earlier work (5), do not contain sequences similar to icmB. Two hyperthermophilic bacteria, Archeoglobus fulgidus (21) and Pyrococcus horikoshii (22), whose genomes have now been completely sequenced, each contain two ORFs of similar size and sequence to icmA and icmB, although they were reported as MCM-like sequences. mcmA1 and mcmA2 from A. fulgidus encode proteins of 548 and 144 residues and are located only 2.5 kb apart, whereas the large and small ICM-like genes in P. horikoshii encode proteins of 563 and 147 residues, but are located ϳ230 kb apart. It is interesting to speculate that, in both organisms, these MCM-like proteins might be components of ICM. However, this remains to be proven. The distinguishing features of ICM include an MCM-like large subunit of only ϳ65 kDa, with a separate small subunit of ϳ14 kDa providing the cobalaminbinding domain. An isolated report has appeared of a propionateinduced MCM from Euglena gracilis that contains two mutases with apparent molecular masses of 72,000 and 17,000 Da (23). Neither protein, however, has yet been purified to homogeneity. Although in MCM a single polypeptide chain comprises both the coenzyme B 12 -and substrate-binding domains, the glutamate mutase from clostridia consists of two subunits, MutE (which forms a tightly associated homodimer) and MutS (which is the coenzyme B 12 -binding domain). The active glutamate mutase holoenzyme appears to be an ␣ 2 ␤ 2 -heterotetramer that binds up to two molecules of coenzyme B 12 , as deduced here for ICM.
The analysis of the kinetics for ICM is complicated by the fact that IcmA and IcmB interact only weakly until coenzyme B 12 is added, and then IcmA binds to IcmB in a cooperative manner. This also complicates any attempts to measure the stoichiometry of bound coenzyme B 12 . As a result, the apparent kinetic parameters depend on the relative concentrations of the two subunits, as observed earlier in studies of glutamate mutase (24,25). Here, the substrate K m values with ICM were determined using equimolar ratios of IcmA and IcmB, and the V max was determined relative to the IcmA concentration at saturating concentrations of IcmB.
The assay used for the determination of kinetic constants is a discontinuous gas chromatography assay, which is not well suited for accurate initial velocity measurements. Notwithstanding this, in the determination of K m values, a nonlinear least-squares regression fitting of the initial velocities to a velocity equation for a bireactant system converged for both substrates. Double-reciprocal plots gave sets of lines that intersected on or close to the x axis (Fig. 7, A and B). The results are consistent with either a random or ordered sequential mechanism, involving the reversible formation of a ternary complex between enzyme, substrate, and coenzyme B 12 . The k cat value of 39 Ϯ 3 s Ϫ1 determined for ICM (Fig. 7C) assumes one active site per IcmA monomer subunit.
The apparent K m values for n-butyryl-CoA and isobutyryl-CoA are in the same range as the K m values reported elsewhere for methylmalonyl-CoA and succinyl-CoA with MCM (26), but are at least an order of magnitude lower than the apparent K m for glutamate with glutamate mutase, measured under comparable conditions (1:1 molar ratio of MutE to MutS) (Table II). It is interesting to note that MCM and ICM have similar substrate structures and most likely similar substrate-binding sites in their large subunits (see below), reflecting the similar K m values for their substrates. On the other hand, ICM and glutamate mutase share a similar mode of cobalamin binding involving two independent subunits and have similar K m values for coenzyme B 12 . In comparison, a significantly lower K m is found for coenzyme B 12 with MCM (Table II).
Early insights into the structures of cobalamin-dependent enzymes came from a primary sequence comparison between the C-terminal domains (now also known as ␣/␤-domains) of MetH and MCM (from both prokaryotic and eukaryotic organisms) with MutS of bacterial glutamate mutase (11). This revealed a region of highly conserved sequence comprising the motif DXHXXG (where X is any amino acid), which was invariant in all the proteins examined. After the determination of the crystal structure of the cobalamin-binding fragment of MetH, Drennen et al. (10) were able to define a sequence fingerprint for cobalamin binding that included D 757 XHXXG . . . S 804 XL . . . G 833 G. This fingerprint is also apparent in IcmB (Fig. 8).
A notable feature in the crystal structures of both the cobalamin-binding fragment of MetH from E. coli (10) and MCM from P. shermanii (7) is the replacement of the DMB ligand from cobalt of the coenzyme by the histidine residue in the DXHXXG motif. The protein brings the cobalamin cofactor to a base-off/His-on form and places the DMB group in a central hydrophobic cleft in the cobalamin-binding domain. The histidine participates in a hydrogen-bonding network comprising 3 residues: His 759 -Asp 757 -Ser 810 in MetH and His 610 -Asp 608 -Lys 604 in MCM. In the case of IcmB, the corresponding residues appear to be well conserved and are represented by His 20 -Asp 18 -Lys 14 , which most likely are also involved in binding coenzyme B 12 . The other residues in the fingerprint sequence appear to line the hydrophobic cleft, to interact with the dis -FIG. 7. A and B, double-reciprocal plots of 1/initial velocities (v) versus 1/substrate concentration for n-butyryl-CoA and isobutyryl-CoA, respectively, as substrate. The experimental data points are shown together with the best fit lines generated by the program Leonora after fitting to Equation 1. The coenzyme B 12 concentration was varied between 4.8 and 200 M or between 6.5 and 200 M, as indicated on each plot. All measurements were repeated in triplicate for determination of K m values. In C, a Hanes plot is shown with initial velocities (v) and n-butyryl-CoA concentration and with saturating coenzyme B 12 (200 M), as used to determine V max and k cat (see "Results").  (24) and MCM (1,26,28) determined elsewhere placed DMB group, and thereby to anchor the cobalamin molecule to the protein. These comparisons therefore suggest that the residues contacting the lower face of the corrin ring are conserved in IcmB and these other cobalamin-binding proteins. The structural conservation of the cobalamin-binding domain is readily apparent from the available crystal structures of MetH and MCM as well as the NMR solution structure of MutS (27). The tertiary structures of the cobalamin-binding domains in MCM and MetH are essentially superimposable (7). The core structure consists of a five-stranded twisted parallel ␤-sheet surrounded by five ␣-helixes. The NMR solution structure of the coenzyme B 12 -free form of MutS also revealed a similar tertiary structure (27), except for the first ␣-helix, which in solution appeared slightly disordered. The displacement of the DMB group of coenzyme B 12 upon binding to the cobalamin-binding domain appears to be a common feature of several (but not all) coenzyme B 12 -dependent mutases.
As shown in earlier work (5), the sequence of the (␤/␣) 8barrel in MutB comprising residues A1-A400 (7) is highly conserved in IcmA. This suggests that the triosephosphate isomerase barrel and much of the acyl-CoA-binding site identified in the crystal structure of MutB are also conserved in IcmA. The triosephosphate isomerase barrel uses a hole through its center to bind substrate, but also appears to open in the absence of ligand (8). Many of the residues that interact with CoA along the hole seem to be highly conserved in a sequence comparison with IcmA (Fig. 9). One residue, Tyr A89 in MutB, is conserved in all MCM sequences and is located near the bottom of the substrate-binding hole near to the interface with coenzyme B 12 . A Y89F mutant of MCM was prepared recently, and its structure was determined by crystallography (28). Although the mutant enzyme structure was essentially superimposable on the wild-type structure and the K m for succinyl-CoA was not significantly affected, the k cat of the mutant was 580-fold lower than that of the wild type. Hence, it was suggested that Tyr A89 plays a key role in the MCM reaction, although not as a site for a protein-based radical. In the sequence comparison with IcmA (5), this Tyr A89 residue corresponds to Phe 80 , so the hydrogen-bonding ability of a Tyr-OH is not needed at this site in ICM. The k cat values for ICM and wild-type MCM are very similar (Table II).
Very recently, a crystal structure of MCM with bound substrate revealed an interaction between Arg A207 and the carboxyl group of methylmalonyl-CoA (9). As far as the different substrate specificities of MCM and ICM are concerned, it is intriguing to note that this Arg A207 residue is replaced in IcmA by a glutamine, whereas many other residues around the active site are highly conserved between MutB and IcmA (Fig. 9). Presently, it seems reasonable to assume that the catalytic mechanisms of ICM and MCM will be largely similar or identical.
Acknowledgment-We thank Annelies Meier for expert technical assistance.