Studies of the CobA-Type ATP:Co(I)rrinoid Adenosyltransferase Enzyme of Methanosarcina mazei Strain Gö1

Abstract

Although methanogenic archaea use B₁₂ extensively as a methyl carrier for methanogenesis, little is known about B₁₂ metabolism in these prokaryotes or any other archaea. To improve our understanding of how B₁₂ metabolism differs between bacteria and archaea, the gene encoding the ATP:co(I)rrinoid adenosyltransferase in Methanosarcina mazei strain Gö1 (open reading frame MM3138, referred to as cobA_Mm here) was cloned and used to restore coenzyme B₁₂ synthesis in a Salmonella enterica strain lacking the housekeeping CobA enzyme. cobA_Mm protein was purified and its initial biochemical analysis performed. In vitro, the activity is enhanced 2.5-fold by the addition of Ca²⁺ ions, but the activity was not enhanced by Mg²⁺ and, unlike the S. enterica CobA enzyme, it was >50% inhibited by Mn²⁺. The CobA_Mm enzyme had a K_m^ATP of 3 μM and a K_m^HOCbl of 1 μM. Unlike the S. enterica enzyme, CobA_Mm used cobalamin (Cbl) as a substrate better than cobinamide (Cbi; a Cbl precursor); the β phosphate of ATP was required for binding to the enzyme. A striking difference between CobA_Se and CobA_Mm was the use of ADP as a substrate by CobA_Mm, suggesting an important role for the γ phosphate of ATP in binding. The results from ³¹P-nuclear magnetic resonance spectroscopy experiments showed that triphosphate (PPP_i) is the reaction by-product; no cleavage of PPP_i was observed, and the enzyme was only slightly inhibited by pyrophosphate (PP_i). The data suggested substantial variations in ATP binding and probably corrinoid binding between CobA_Se and CobA_Mm enzymes.

Bioinformatics analyses of archaeal genome sequences readily identify putative orthologs of B₁₂ biosynthetic genes found in Salmonella enterica (28, 44). However, positive identification of gene function requires in vivo and in vitro analyses. To facilitate the latter, we have taken advantage of our extensive collection of S. enterica strains lacking specific functions needed for coenzyme B₁₂ synthesis. We have successfully used this approach to identify archaeal nonorthologous replacements of bacterial B₁₂ biosynthetic functions needed for the assembly of the nucleotide loop of coenzyme B₁₂ (25, 44, 47, 48, 50, 51) and for the transport of corrinoids across the membrane (49).

In these studies we turned our attention to the archaeal enzyme responsible for the conversion of vitamin B₁₂ to its coenzymic form, adenosyl-B₁₂ (AdoB₁₂, AdoCbl). Not all archaeal genomes contain orthologs of the S. enterica cobA gene, which in this bacterium and in Escherichia coli encodes the housekeeping enzyme responsible for attaching the adenosyl moiety from ATP to the cobalt ion of the corrin ring (13, 24, 41). In S. enterica, the cobA gene is constitutively expressed, and the activity of the CobA_Se enzyme is required for de novo biosynthesis of the corrin ring (13) and for salvaging complete and incomplete corrinoids from the environment (39-41). In other words, the broad specificity of the CobA_Se enzyme allows S. enterica to salvage a broad spectrum of corrinoids from its environment and to solve the need for corrinoid adenosylation during de novo corrin ring biosynthesis.

The genome of the methanogenic archaeon Methanosarcina mazei strain Gö1 contains a putative cobA ortholog (ORF MM3138), which we hypothesized might play physiological roles similar to those of the CobA_Se enzyme.

MATERIALS AND METHODS

Plasmid construction.

The M. mazei cobA⁺ gene (locus tag MM3138) was PCR amplified from Methanosarcina mazei strain Gö1 genomic DNA by using the primers “M.mazei CobA pT7-7 fwd” (5′-TACGGGCCATATGGCCGGTGGCATGGTATAC-3′) and “M.mazei CobA pT7-7 rev” (5′-GGATCCGTGATCAATATTCAAGTCCTTCCCTTGCAG-3′). The PCR fragment was digested with NdeI/BamHI and ligated into cloning vector pT7-7 (42), yielding plasmid pMmaCOBA4.

Analysis of growth behavior.

Strains and plasmids used in the present study are listed in Table 1. The ability of the M. mazei cobA⁺ gene to complement a Salmonella cobA strain was assessed in 96-well plates (Becton/Dickinson). Strains were grown in lysogenic broth (LB) (4) (2 ml) overnight cultures containing the appropriate drug at 37°C. Plates contained no-carbon E (NCE) minimal medium (3), trace minerals (1), MgSO₄ (1 mM), and NH₄Cl (30 mM). When cells were grown on glycerol as the sole carbon and energy source, the medium was supplemented with glycerol (30 mM), 5,6-dimethylbenzimidazole in dimethyl sulfoxide (300 μM), and dicyanocobinamide [(CN)₂Cbi, 200 nM]. When cells were grown on ethanolamine, the medium was supplemented with ethanolamine hydrochloride (30 mM, pH 7), methionine (0.5 mM), and hydroxocobalamin (HOCbl, 200 nM). Samples (10 μl) of overnight LB cultures containing the appropriate drug were used to inoculate each well containing 190 μl of fresh medium. Plates were incubated in a BioTek plate reader at 37°C with maximum aeration.

TABLE 1.

Strains and plasmids

Strain or plasmida	Genotype	Source or referenceb
Strains
E. coli JE3892	BL21(λDE3) F−dcm ompT hsdS(rB− mB−) gal λ(DE3)	New England Biolabs
S. enterica
TR6583	metE205 ara-9 cobA+	K. Sanderson via J. Roth
JE1293*	cobA366::Tn10d(cat+)
JE7180*	cobA366::Tn10d(cat+) eut1141 (ΔeutT)
JE7954*	cobA366::Tn10d(cat+) eut1141 (ΔeutT)/pMmaCOBA4	This study
JE7955*	cobA366::Tn10d(cat+) eut1141 (ΔeutT)/pT7-7	This study
Plasmids
pMmaCOBA4	bla+M. mazei strain Gö1 PT7cobAMm	This study
pT7-7	bla+	42

^a*, Derivative of strain TR6583.

^bUnless otherwise indicated, all strains are part of the laboratory collection.

Isolation of the CobA_Mm protein.

Plasmid pMmaCOBA4 was transformed into E. coli BL21(λDE3) for overexpression (17). Fresh transformants were grown overnight in 10 ml of LB plus ampicillin (100 μg/ml). The latter cultures were used to inoculate three 1.5-liter flasks containing LB plus ampicillin. Cells were shaken at 37°C until late log phase, when gene overexpression was induced with 0.5 mM IPTG (isopropyl-β-d-thiogalactopyranoside), and the flasks were shifted to 25°C. Cells were harvested the next day by centrifugation at 4°C for 20 min at 5,000 × g in a Beckman Coulter Avanti-J-20 XPI centrifuge fitted with a JLA 8.1000 rotor. The cell paste was resuspended in 40 ml of 50 mM glycine buffer (pH 9.5) containing 5 mM dithiothreitol (DTT), and 1 mM phenylmethylsulfonyl fluoride. Cell slurry was passed twice through a French pressure cell at 1,250 kPa to ensure breakage. Cell extracts were centrifuged at 4°C for 1 h at 43,667 × g in a Beckman Coulter Avanti-J-25I centrifuge fitted with a JA 25.50 rotor. The supernatant was decanted and treated with a few crystals of DNase for 20 min at room temperature and dialyzed twice for 30 min against 750 ml of 50 mM glycine buffer (pH 9.5) containing 5 mM DTT. After treatment, cell extracts were halved and purified independently.

Native CobA_Mm protein was purified by using an ÄKTA Explorer fast-protein liquid chromatograph (Pharmacia). First, each half of the treated cell extract was loaded onto two 5-ml HiTrap Q Fast Flow columns (Amersham Biosciences) connected in a series. After a 20-ml wash with buffer A1 (50 mM glycine [pH 9.5], 5 mM DTT), CobA_Mm protein was eluted in a linear gradient to 30% buffer B1 (50 mM glycine [pH 9.5], 1 M NaCl, 5 mM DTT) over 200 ml (0 to 300 mM NaCl). Fractions containing CobA_Mm were pooled and dialyzed overnight against 50 mM glycine (pH 9.5), 4 M NaCl, and 5 mM DTT. After dialysis, each sample was applied to two 5-ml HiTrap Phenyl High Performance columns (Amersham Biosciences) in a series. After a 20-ml wash with buffer A2 (50 mM glycine [pH 9.5], 4 M NaCl, 5 mM DTT), CobA_Mm was eluted with a linear gradient to 100% buffer B2 (50 mM glycine [pH 9.5], 4 M ethylene glycol, 5 mM DTT) over 100 ml. Fractions containing CobA_Mm were pooled and dialyzed against buffer A1. The sample was concentrated to 10 ml by using an Amicon Centricon with a YM10 membrane (cutoff = 10 kDa). Each sample was purified by using an 8-ml Source 15Q column (Amersham Biosciences) After a 16-ml wash with buffer A1, CobA_Mm was eluted with a linear gradient to 15% B1 over 80 ml (0 to 150 mM NaCl). The purity of CobA_Mm protein was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (23), followed by Coomassie blue staining (32). Fractions enriched for CobA_Mm protein were concentrated by using an Amicon Centricon device with a YM10 membrane and dialyzed against 10 mM glycine buffer (pH 9.5). Glycerol was added to dialyzed CobA_Mm protein to a final concentration of 10% (vol/vol), and samples were stored at −80°C until used.

Adenosyltransferase assays.

Assays were performed in sealed quartz cuvettes flushed with oxygen-free N₂. Reactions contained 0.2 M Tris-Cl buffer (pH 8, 37°C), 50 μM HOCbl, 500 μM ATP, 500 μM CaCl₂, 2.5 mM Ti(III)citrate reductant (26), and 2.5 μM CobA_Mm protein. The reaction rate was monitored by determining the decrease in absorbance at 388 nm (9, 20). For flavodoxin-dependent assays, Ti(III)citrate was replaced by FldAH₆ protein (C terminus H₆-tagged flavodoxin, 2 μM), Fpr protein [ferredoxin(flavodoxin):NADPH oxidoreductase, 2 μM], and NADPH (500 μM) (16). The amount of adenosylcobalamin (AdoCbl) synthesized after 30 min was monitored by determining the decrease in absorbance at 525 nm after photolysis for 10 min on ice as described previously (41).

³¹P-NMR spectroscopy.

A 10-ml corrinoid adenosylation reaction mixture containing Ti³⁺ citrate (3 mM), ATP (100 μM), HOCbl (100 μM), CaCl₂ (500 μM), and CobA_Mm protein (5 μM) in 0.2 M Tris-Cl buffer (pH 8, 37°C) was incubated for 2 h at 37°C. After incubation, EGTA was added to 20 mM, reactions were concentrated in a SpeedVac overnight at room temperature, and 100% D₂O was added to 6%. Orthophosphate, pyrophosphate, and triphosphate standards were added to 100 μM when indicated. The data were collected at the regional nuclear magnetic resonance (NMR) facility at Madison at the University of Wisconsin-Madison.

RESULTS

The protein encoded by M. mazei Gö1 open reading frame (ORF) MM3138 functions as an ATP:co(I)rrinoid adenosyltransferase in vivo.

The predicted amino acid sequence of the M. mazei Gö1 MM3138 protein (referred to herein as CobA_Mm) was 31% identical to the S. enterica CobA protein (Fig. 1). This suggested that the MmCobA amino acid sequence could be the ATP:co(I)rrinoid adenosyltransferase of this methanogenic archaeon. The presence of a CobA-type, short ATP-binding P-loop motif (2) (Fig. 1) strengthened this hypothesis. Noteworthy differences between the primary amino acid sequences of these proteins were the absence in CobA_Mm of residues 4 to 25 of CobA_Se. A deletion of these residues in CobA_Se severely impairs the interactions between CobA_Se and flavodoxin A (8). Three other shorter stretches of amino acids in CobA_Se were missing in CobA_Mm, while a 15-residue stretch present in CobA_Mm was absent in CobA_Se (Fig. 1). We speculate that, based on these differences, the three-dimensional structures of these proteins may differ significantly. Efforts to solve the three-dimensional crystal structure of CobA_Mm are under way.

FIG. 1.

Alignment of CobA from S. enterica with archaeal homologs. The primary amino acid sequence of CobA_Se is 31% identical and 52% similar to that of CobA_Mm. The dark-shaded box indicates the P-loop ATP-binding motif, while the light-gray box indicates regions of conservation. Asterisks denote absolutely conserved amino acid residues. Sen, Salmonella enterica; Mma, Methanosarcina mazei strain Gö1; Mac, Methanosarcina acetivorans strain C2A; Has, Halobacterium salinarum strain NRC-1; Pae, Pyrobaculum aerophilum; Pab, Pyrococcus abyssi; Pfu, Pyrococcus furiosus; Pho, Pyrococcus horikoshii; Tac, Thermoplasma acidophilum; Tvo, Thermoplasma volcanium. Alignment was generated by using DNA* software package v.1.66 (DNASTAR, Inc.) without adjustments.

To demonstrate the function of CobA_Mm in vivo, ORF MM3138 was amplified from M. mazei strain Gö1 DNA, and the amplification product was cloned into overexpression vector pT7-7 (42); the resulting plasmid was named pMmaCOBA4 (Table 1). Plasmid pMmaCOBA4 was transformed into S. enterica metE cobA eutT strain JE7180, and the ability of the resulting strain (JE7954) to grow under conditions that required low or high levels of AdoCbl was assessed. A control strain was constructed by transforming the cloning vector pT7-7 into strain JE7180, yielding strain JE7955.

Low Cbl requirement test.

Low Cbl levels (1 nM) satisfy the methionine requirement of an S. enterica metE strain during growth on glucose or glycerol (45, 49). Strains lacking Cbl-independent methionine synthase (MetE) methyltransferase activity methylate homocysteine via the Cbl-dependent methionine synthase (MetH) enzyme (18). During aerobic growth on glycerol, a culture of strain JE7954 (metE cobA eutT/pMmacobA⁺) reached the same cell density as a culture of strain TR6583 (metE cobA⁺ eutT⁺) (Fig. 2A, solid symbols), but the rate of growth of strain TR6583 (cobA⁺) was faster by a factor of 2. The positive effect of CobA_Mm on the conversion of cobinamide to AdoCbl was substantial compared to the growth behavior of control strains (Fig. 2A, open symbols).

FIG. 2.

cobA⁺_Mm compensates for the lack of cobA_Se under conditions that require low levels of AdoCbl for growth. (A) Growth behavior of S. enterica strains grown in minimal medium containing glycerol as sole carbon and energy source. (B) Growth in minimal medium containing ethanolamine as sole carbon and energy source. Derivatives of strain JE7180 [cobA366::Tn10d(cat⁺) eut1141 (ΔeutT)] carrying plasmids pMmaCOBA4 cobA⁺_Mm (strain JE7954) or pT7-7 (VOC, strain JE7955) were used to investigate CobA_Mm function in vivo. Strains TR6583 (cobA⁺ eut⁺) and JE1293 [cobA366::Tn10d(cat⁺)] were used as controls. VOC, vector-only control.

High Cbl requirement test.

To grow on ethanolamine as a sole carbon and energy source, S. enterica must synthesize AdoCbl before the EutR regulatory protein (29) can activate transcription of the ethanolamine utilization (eut) operon (9, 34) that encodes functions needed to catabolize ethanolamine (6, 7, 21, 30, 31, 36, 37). Growth of S. enterica strain JE7954 [cobA eut (eutT)/pMmaCOBA4 cobA⁺_Mm] on ethanolamine requires high levels of AdoCbl (≥150 nM [6]). As shown in Fig. 2B, the CobA_Mm enzyme did not compensate for the lack of CobA_Se enzyme during growth of strain JE7954 on ethanolamine as sole carbon and energy source. There are several possible explanations for why the CobA_Mm enzyme did not support growth on ethanolamine as efficiently as the CobA_Se enzyme did. For example, CobA_Mm may not have been expressed at a high enough level to support wild-type growth, it may not have been stably maintained, or its interactions with flavodoxin may have been suboptimal. At present, no data favoring any of these possibilities is available, however. Nevertheless, the data presented in Fig. 2A support the conclusion that the CobA_Mm protein has activity that can attach the adenosyl upper ligand to the corrin ring of cobinamide in vivo, allowing its conversion to AdoCbl and the use of the latter in methionine synthesis.

Initial biochemical characterization of CobA_Mm.

CobA_Mm enzyme was purified from E. coli. We obtained 8 mg of CobA_Mm protein per liter of culture grown under the conditions described above; CobA_Mm was calculated to be 88% homogeneous (Fig. 3). In vitro corrinoid adenosyltransferase assays with CobA_Mm protein were performed by using published protocols for assaying CobA_Se activity (9).

FIG. 3.

CobA_Mm protein purity. CobA_Mm protein was isolated as described in Materials and Methods. After Coomassie blue staining, protein purity was quantified by using a Fotodyne.

Reductant and pH_opt.

Potassium borohydride (KBH₄) was tested as a reducing agent in lieu of Ti³⁺ citrate but, for unknown reasons, it was detrimental to the activity of the enzyme. CobA_Mm activity was assayed at various pH values. Under conditions where Mg²⁺ ion concentration was saturating for CobA_Se, the highest rate of AdoCbl synthesis occurred at pH 8 (e.g., at pH 7.5 the specific activity is 0.75, at pH 8 the specific activity is 1.02, and at pH 8.5 specific activity is 0.94 nmol min⁻¹ mg⁻¹).

Cation requirement.

We tested the effect of metal ions on CobA_Mm activity under conditions that used Ti³⁺ citrate as a reductant at pH 7. Ca²⁺ ions stimulated CobA_Mm activity 2.5-fold over activities measured in reaction mixtures where metal salts were not added. Co²⁺, Ni²⁺, and Zn²⁺ ions stimulated the CobA_Mm activity 1.75-, 1.5-, and 1.25-fold, respectively. Surprisingly, and unlike CobA_Se, CobA_Mm activity was substantially inhibited by Mn²⁺ ions (60% reduction) relative to the no-additions control, and Mg²⁺ ions did not significantly improve enzyme activity (Fig. 4). In light of the above results, Ca²⁺ ions were present at 0.5 mM in all subsequent assays.

FIG. 4.

CobA_Mm enzyme activity is enhanced by the addition of Ca²⁺ ions. Various metal salts were tested for their effect on CobA_Mm activity in vitro. Salts were added to the reaction mixture containing 0.2 M Tris-Cl buffer (pH 8, 37°C), CobA_Mm (2 μm), ATP (100 mM), HOCbl (50 μM), and Ti(III) citrate (2.5 mM) as a reductant.

Kinetic parameters and corrinoid substrate preference.

We used the optimized conditions described above to gain insights into the affinity of CobA_Mm for its substrates and the velocity of the reaction. From the data shown in Fig. 5 we calculated a K_m^ATP of 3 μM when Cbl was held at 0.05 mM. When the ATP was held at 0.5 mM and the concentration of Cbl was varied, we calculated a K_m^Cbl of 0.9 μM and a V_max of 3 nmol min⁻¹.

FIG. 5.

Kinetic analysis of the CobA_Mm-catalyzed reaction. (A) HOCbl was held constant at 0.05 mM, while the concentration of ATP was varied. V_max was determined to be 3 nmol of AdoCbl formed min⁻¹ mg⁻¹; the apparent K_m^ATP was calculated at 3 μM. (B) The ATP concentration was held constant at 0.50 mM, while the HOCbl concentration was varied. V_max was 3 nmol of AdoCbl formed min⁻¹ mg⁻¹, and the apparent K_m^HOCbl was calculated to be 1 μM.

While the kinetic analyses were performed with HOCbl as substrate, the uncertain physiological role of CobA_Mm in vivo prompted us to test the ability of the enzyme to use Cbi as substrate since it lacks the lower 5,6-dimethylbenzimidazole ligand. Under substrate saturation conditions, the specific activities when Cbl or Cbi was used as a substrate was 1.56 or 0.90 nmol min⁻¹ mg⁻¹, respectively. Therefore, under the conditions of the in vitro assay, CobA_Mm preferred Cbl over Cbi as a substrate.

Nucleoside triphosphate substrate.

The activity of CobA_Mm as a function of different nucleoside triphosphates was determined. The data in Fig. 6A were plotted relative to the activity of the enzyme when ATP was the substrate. Unlike CobA_Se, the base of nucleoside substrates tested had a significant negative effect on CobA_Mm activity. The most dramatic effect was obtained with CTP. Although CTP is a better substrate than ATP for Coba_Se (15), it was a poor substrate for CobA_Mm. A second difference between CobA_Se and CobA_Mm was observed when deoxynucleoside triphosphates (dNTPs) were used. dNTPs are not substrates for CobA_Se (15), but CobA_Mm used dNTPs as substrates, and its activity with dCTP was equivalent to the activity measured with ATP. The most striking difference between CobA_Se and CobA_Mm was observed with ADP. Although ADP is not a substrate for CobA_Se (15), CobA_Mm retained 37% of its activity when ADP substituted for ATP. This result suggested that the γ phosphate was likely the primary phosphate for ATP coordination.

FIG. 6.

CobA_Mm protein interacts with the 2′-OH and the γ phosphate of ATP but is not inhibited by PPP_i. (A) Specificity of the CobA_Mm enzyme for its nucleoside triphosphate substrate. When added, ADP was present in the reaction mixture at 500 μM. The reaction rate for a mixture containing ATP was 2.3 nmol of AdoCbl min⁻¹ ± 0.1. Ado, adenosine. (B) When added to the reaction mixture, PPP_i, PP_i, or P_i was present at 1 mM. The reaction rate for a mixture with no additions was 1.8 nmol of AdoCbl min⁻¹.

Triphosphate is the by-product of the CobA_Mm reaction.

In an effort to identify the phosphate by-product of the CobA_Mm reaction, we tested the sensitivity of the enzyme to triphosphate (PPP_i), pyrophosphate (PP_i), or orthophosphate (P_i) (all at 1 mM, Fig. 6B). CobA_Mm was not inhibited by the presence of PPP_i and in fact was slightly stimulated (Fig. 6B). This stimulatory effect was not due to a simple increase in ionic strength; the addition of up to 10 mM NaCl to the assays had no effect on the rate of catalysis (data not shown). PP_i was a weak inhibitor (28% decrease in specific activity), and P_i had no significant effect. CobA_Se is strongly inhibited by PPP_i and has been shown to release PPP_i as a reaction by-product (15). In light of this information, we hypothesized that CobA_Mm probably cleaved PPP_i to PP_i and P_i as part of the catalytic cycle. We addressed this possibility using ³¹P-NMR spectroscopy (Fig. 7). The ³¹P-NMR spectrum showed an accumulation of PPP_i in the reaction mixture, no increase in P_i signal, and no detectable PP_i peak. In Fig. 7A, the signal proximal to Cbl was not identified. We speculate that it was likely the result of ion coordination by the phosphate moiety in the nucleotide loop of Cbl. In Fig. 7C and D, the doublet proximal to the α-PPP_i signal was not identified and may arise from ion interactions with PPP_i. On the basis of this information we concluded that CobA_Mm does not cleave PPP_i to PP_i and P_i. The absence of inhibition by PPP_i and only weak inhibition by PP_i may reflect the importance of purine ring coordination in substrate binding to the enzyme active site.

FIG. 7.

³¹P-NMR analysis of CobA_Mm reaction by-products. Strong signals for the β phosphate of ATP (centered at −21.51 ppm) and PPP_i (centered at −22.39 ppm) were observed after incubation of CobA_Mm with its substrates. No signal was detected from PP_i at −7.41 ppm. (A) Complete reaction mixture; (B) ATP standard; (C) PPP_i standard; (D) mixture of ATP, PPP_i, PP_i, P_i, and HOCbl.

DISCUSSION

We presented here experimental evidence supporting the annotation of an archaeal gene encoding a protein with CobA-type ATP:co(I)rrinoid adenosyltransferase activity. To reflect these findings, we propose that ORF MM3138 from the methanogenic archaeon Methanosarcina mazei strain Gö1 be hereafter referred to as cobA.

Potential differences between archaeal and bacterial CobA enzymes.

Although CobA_Mm and CobA_Se may have evolved from a common ancestor (Fig. 1), there could be significant differences between the mechanisms of catalysis used by these enzymes. Proper positioning of the ATP substrate in the active site of CobA_Se is critical to catalysis. However, biochemical data suggest that CobA_Se and CobA_Mm may not bind ATP the same way. For example, the absolute requirement of CobA_Se for the 2′-OH of the ribose in ATP is not shared by CobA_Mm, suggesting that CobA_Mm presents the target (5′-C) for the nucleophilic attack by Co(I) by different means. The role of the base of NTPs is also significantly different between the archaeal and bacterial enzymes. Although CobA_Mm uses ITP, GTP, CTP, and UTP as substrates, CobA_Se can use them more efficiently. In fact, CobA_Se prefers CTP and UTP over ATP (15). It is therefore likely that CobA_Mm interacts with the purine ring more than CobA_Se, which only forms a single H-bond between the carbonyl oxygen of Asn37 and the amino group of adenine (2). Therefore, in CobA_Mm, positioning of the 5′-C may be mediated by interactions with the γ phosphate and the purine ring rather than by interactions with the α and β phosphates and the 2′-OH as in CobA_Se. This difference in ATP binding was unexpected considering the relatedness of the two enzymes.

CobA_Se and CobA_Mm may interact differently with its corrinoid substrate. Many of these differences may be due to the absence of an N-terminal helix in CobA_Mm. Residues 1 to 23 of CobA_Se form an N-terminal α helix that closes over the occupied active site (2). CobA_Mm has only 10 residues before the ATP-binding P-loop motif (Fig. 1). A plausible explanation would be that the higher apparent K_m of CobA_Mm for Cbl and the sluggishness of the enzyme may be attributable to the absence of the N-terminal helix. It is important to note, however, that an N-terminal truncation of Coba_Se results in a more active enzyme, as long as the substrates are present at saturating levels (8), suggesting that an enzyme lacking the N-terminal helix undergoes faster substrate binding and product release. The V_max for CobA_Mm is also ca. 17% of that of CobA_Se (3 versus 18 nmol of AdoCbl min⁻¹ mg⁻¹). This decreased CobA_Mm activity may be a consequence of the overexpression/purification procedures or suboptimal in vitro assay conditions, or it may represent a true difference in enzyme catalysis. Additional work is required to determine the reasons for this important difference.

The marked difference in the ability between CobA_Mm and Coba_Se in the use ADP as a substrate suggests significant variations in how these enzymes bind ATP. The three-dimensional structure of CobA_Se complexed with MgATP and Cbl shows the Mg²⁺ ion octahedrally coordinated by two nonbridging phosphate oxygens from the α- and β-phosphates, the hydroxyl group of Thr42, a carboxylate oxygen of Glu128 and two water molecules. This coordination is similar to that observed in other enzymes (e.g., F₁ATPase), with some differences. The three-dimensional structure of CobA_Mm complexed with substrates is needed to better understand how CobA_Mm binds ATP, its cation, and its corrinoid substrate. Efforts to obtain a three-dimensional structure of CobA_Mm are currently under way.

CobA_Mm may be a Ca²⁺-dependent ATP:co(I)rrinoid adenosyltransferase.

ATP-binding enzymes use divalent cations to coordinate the triphosphate and partially shield the negative charge. The same is true for cobalamin adenosyltransferases. It is known that CobA_Se prefers Mn²⁺ but will use Mg²⁺ ions in vitro (40). A second class of ATP:co(I)balamin adenosyltransferase found in S. enterica is encoded by the pduO gene of the 1,2-propanediol utilization (pdu) operon. The PduO enzyme prefers Mg²⁺ as a cation (19, 20). We were very surprised to see that Ca²⁺ ions stimulated activity 2.5-fold, but Mg²⁺ ions did not stimulate activity, and Mn²⁺ ions inhibited the enzyme (Fig. 4). In these assays, CobA_Mm was not treated with chelating agents, so the presence of any accessory bound cations cannot be ruled out. However, the 2.5-fold stimulation of enzyme activity was specific for Ca²⁺. There are only a few examples in the literature of CaATP binding enzymes, e.g., the human heat shock protein 70 (hHsp70) (27, 35), β-actin (10), and synapsin I (14, 46). It is important to note that hHsp70 and β-actin have been crystallized complexed with MgATP and seem to require a divalent cation for ATP binding, whereas synapsin I specifically requires CaATP for enzyme activity (5, 14). Although it is possible that CobA_Mm requires Ca²⁺ ions for structural integrity and not ATP binding, a possibility exists that CobA_Mm may be a unique CaATP:co(I)rrinoid adenosyltransferase.

Can Salmonella FldA serve as the electron shuttle protein for CobA_Mm?

In S. enterica, the physiological electron transfer protein for the CobA enzyme is flavodoxin (FldA) (16). Although the Methanosarcina genomes contain flavodoxin homologs, corrinoid adenosyltransferases in these organisms may use one of several ferredoxins or other unknown means to generate co(I)rrinoid substrates for CobA_Mm. The inability of CobA_Mm to support the growth of S. enterica on ethanolamine may be a reflection of poor coupling between S. enterica FldA and CobA_Mm proteins. If SenFldA is indeed the reductant for CobA_Mm, their coupling may be insufficient to support growth on ethanolamine. However, even if poor, such coupling satisfies the Cbl requirement of the cell for methionine synthesis (Fig. 2A). In vitro assays using possible reductants from M. mazei are needed to address the question of co(II)rrin reduction in archaea.

Why does Methanosarcina need an adenosyltransferase?

Methanogenic archaea use pseudo-B₁₂, 5-hydroxybenzimidazolylcobamide, or cobalamin as a methyl donor (11, 12, 22, 33, 38), and sequenced genomes contain homologs of B₁₂-dependent ribonucleotide reductases (43). At present, it is unclear what role the CobA_Mm enzyme plays in vivo. It is possible that, like CobA_Se, CobA_Mm is needed to attach the adenosyl upper ligand to an intermediate of the corrin ring biosynthetic branch of the pathway, to adenosylate corrinoids from its environment that may or may not contain a lower ligand, or both (47, 50). Very little is known about B₁₂-dependent metabolism in archaea. Biochemical data suggest salvaging of incomplete corrinoids and/or de novo biosynthesis proceeds through an adenosylated intermediate (47). However, details of the de novo biosynthetic pathway, transport, and precursor salvaging in archaea remain to be elucidated.