CobD, a Novel Enzyme with L -Threonine- O -3-phosphate Decarboxylase Activity, Is Responsible for the Synthesis of ( R )-1-Amino-2-propanol O -2-Phosphate, a Proposed New Intermediate in Cobalamin Biosynthesis in Salmonella typhimurium LT2*

The cobD gene of Salmonella typhimurium LT2 has been cloned, sequenced, and overexpressed. The overexpressed protein had a molecular mass of ; 40 kDa, in agreement with the mass predicted by the deduced amino acid sequence (40.8 kDa). Computer analysis of the deduced amino acid sequence of CobD identified a consensus pyridoxal phosphate-binding motif. The role of CobD in cobalamin biosynthesis in this bacterium has been established. CobD was shown to decarboxylate L - threonine O -3-phosphate to yield ( R )-1-amino-2-propa-nol O -2-phosphate. We propose that the latter is a substrate in the reaction catalyzed by the CbiB enzyme proposed to be responsible for the conversion of adeno-sylcobyric acid to adenosylcobinamide and that the product of the reaction is adenosylcobinamide phosphate, not adenosylcobinamide as previously thought. The implications of these findings are discussed in light of the demonstrated kinase activity of the CobU enzyme (O’Toole, G. A., and Escalante-Semerena, J. C. (1995) J. Biol. Chem. 270, 23560–23569) responsible for the conversion of adenosylcobinamide to adenosylcobinamide phosphate. These findings shed light on the strategy used by this bacterium for the assimilation of exogenous unphosphorylated cobinamide from its

The biochemistry of adenosylcobalamin biosynthesis has been studied for over 4 decades, with an accelerated pace of progress being accomplished in the last 15 years due to the application of genetic and recombinant DNA approaches (reviewed in Refs. [1][2][3][4][5]. Several procaryotes have been used as model systems to study cobalamin biosynthesis. The best studied ones are the strictly respiring Pseudomonas denitrificans, the facultative anaerobe Salmonella typhimurium LT2, the aerotolerant anaerobe Propionibacterium freundenreichii (shermanii), and the strict anaerobe Eubacterium limosum. There are, however, key differences between the pathways leading to the synthesis of the corrin ring in these organisms. Most notable is the time of cobalt insertion into the macrocycle (6 -11). While cobalt insertion in P. denitrificans occurs late in the pathway (12,13), cobalt appears to be inserted very early in the synthesis of the corrin ring in S. typhimurium and P. freundenreichii (8,10,11). The majority of the reactions of the corrin ring biosynthetic pathway in P. denitrificans has been firmly established (2), whereas the identities of most of the intermediates of this pathway in S. typhimurium and P. freundenreichii have not been elucidated.
One important unanswered question about the synthesis of the corrin ring regards the metabolic origin of the (R)-1-amino-2-propanol moiety linking the macrocycle to the nucleotide (Fig. 1). Early studies performed on Streptomyces griseus demonstrated that in this bacterium, label from [ 15 N]L-Thr was incorporated into the (R)-AP 1 moiety of Cbl. However, evidence for direct decarboxylation of L-Thr was not obtained (14). Later, a series of studies by  investigated the nonenzymatic decarboxylation of L-Thr. These authors showed that L-Thr decarboxylation occurred optimally at pH 8 when diaquocobyric acid was present in a reaction mixture containing Tris-Cl buffer and a reductant (e.g. glutathione, ␤-mercaptoethanol, or borohydride) (15,16). This work led to a model for interactions between diaquocobyric acid and L-Thr. Although the nature of the interaction between these two compounds was not established, it was demonstrated that for the interaction to occur, the cobalt ion in the ring had to be in its Co(II) oxidation state (17). While these studies presented a thought-provoking mechanism for the synthesis of (R)-AP and its incorporation into cobinamide, they failed to demonstrate that direct decarboxylation of L-Thr leads to (R)-AP synthesis.
Recent work in S. typhimurium identified the cobD gene as one whose gene product was required for the synthesis of (R)-AP in this bacterium (21). Analysis of cobD mutants argued against the hypothesis that the threonine dehydrogenase pathway was the one responsible for the synthesis of (R)-AP in this bacterium.
In this paper, we present the cloning, sequencing, and overexpression of the cobD gene of S. typhimurium and present in vitro evidence showing that the CobD protein directly decarboxylates L-threonine O-3-phosphate to yield (R)-1-amino-2propanol O-2-phosphate in the absence of corrinoid; the enzyme showed stereospecificity for the L-isomer.
On the basis of this finding, we propose that the end product of the corrin ring biosynthetic pathway in this bacterium, and probably others, is AdoCbi-P, not AdoCbi as previously thought. To our knowledge, this is the first report of an L-threonine-O-3-phosphate decarboxylase enzyme.

Bacterial Strains
All strains used in this study were derivatives of S. typhimurium LT2, unless noted, and their genotypes are listed in Table I. Tn10d(Cm) refers to the transposition-defective element described by Elliott and Roth (22).

Culture Media, Chemicals, and Growth Conditions
The E minimal medium of Vogel and Bonner (23) was routinely supplemented with 11 mM glucose as the carbon/energy source. Difco Nutrient Broth™ (8 g/liter) with NaCl (5 g/liter) added was used as the rich medium. LB broth was used for experiments involving plasmid manipulation and for protein overexpression. Difco Granulated Agar™ (15 g/liter) was added for solid support. When present in the medium, (CN) 2 Cby and dicyanocobinamide were at 45 nM, and (R)-AP was at 1.3 mM. Concentrations of antibiotics were as described elsewhere (24). All chemicals were purchased from Sigma, except (CN) 2 Cby, which was a gift from F. Blanche (Rhône-Poulenc Rorer S. A., Vitry-sur-Seine Cedex, France); L-and D-lactaldehydes were synthesized as described (25). Cell density of cultures was monitored with a Klett photoelectric colorimeter (Manostat Corp., New York, NY) or with a Spectronic 20D spectrophotometer (Milton Roy Co., Rochester, NY).

In Vivo Assessment of Cbl Biosynthesis
Cbl biosynthesis was assessed in vivo by demanding synthesis of methionine via the Cbl-dependent methionine synthase, MetH (5-methyltetrahydrofolate-homocysteine methyltransferase, EC 2.1.1.13) (26,27). cobD mutant JE2216 carried a mutation in metE encoding the Cbl-independent methionine synthase, MetE (5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase, EC 2.1.1.14); hence, growth of the strain depended on the availability of either methionine or Cbl. As described elsewhere (21), Cbl biosynthesis in cobD mutants was restored when the medium was supplemented with (CN) 2 Cby and (R)-AP under aerobic growth conditions or by (R)-AP alone under anaerobic growth conditions. The need for (CN) 2 Cby under anaerobic conditions is bypassed because S. typhimurium synthesizes the corrin ring de novo when its environment is devoid of oxygen (27,28).
Complementation Studies-Both plasmids pCOBD2 and pCOBD6 and the corresponding vector-only controls pSU38 and pT7-7 were transformed into cobD mutant strain JE2216 and its recombinationdeficient derivative JE4097. Transformants were tested for their ability to grow aerobically on minimal medium supplemented with glucose, (CN) 2 Cby, and ampicillin (pT7-7 derivatives) or kanamycin (pSU38 derivatives). Strains that grew under these conditions were assessed as positive for complementation of the cobD phenotype and therefore carried a plasmid with a functional cobD gene.

Recombinant DNA Techniques
Plasmid Isolation-Plasmid DNA was isolated from cultures using a QIAprep Spin Plasmid™ kit (QIAGEN Inc., Chatsworth, CA) and transformed into strains made competent using a standard calcium chloride treatment (32). Restriction endonucleases, T4 DNA ligase, and calf intestinal alkaline phosphatase were purchased from Promega (Madison, WI) and used in accordance with the manufacturer's specifications. PCR products were isolated from Tris borate/EDTA-containing 1% agarose gels using the QIAquick™ gel extraction kit (QIAGEN Inc.).
DNA Sequencing-The complete DNA sequence of cobD was generated in part using the dideoxy method with the Sequenase ® Version 2.0 kit (U. S. Biochemical Corp.) and by nonradioactive sequencing at the University of Wisconsin Biotechnology Center. The sequence presented is the result of both strands being sequenced at least three times in their entirety. DNA sequence was analyzed using the software programs DNA Strider™, BLAST (33), MacTargsearch, and ProSite (34).
Subcloning of cobD-Two oligonucleotide primers were used to amplify the Ϫ125 to ϩ1148 region of cobD of pCOBC4 using PCR methodology. Each primer contained a 3Ј-region complementary to pCOBC4 followed by a 5Ј-noncomplementary end that generated either an EcoRI (primer 1) or a BamHI (primer 2) restriction endonuclease site. Amplification between the two primers was performed using Vent R ® polymerase (New England Biolabs Inc., Beverly, MA) in a Temp-Tronic Thermocycler (Barnstead Thermolyne, Dubuque, IA). Reaction conditions were as follows: denaturation at 94°C for 2 min, annealing at 50°C for 1.5 min, and extension at 72°C for 1 min. The primers used were as follows: primer 1 (5Ј-CCCGAATTCCGCCATGACGCACCAGCCACAA-TCGC-3Ј), which hybridized to the Ϫ125 to Ϫ100 region upstream of cobD and generated an EcoRI restriction endonuclease site; and primer 2 (5Ј-CCCGGATCCTGGAAAAAGGCGAATCCTGCGACGAT-3Ј), which hybridized to the ϩ1123 to ϩ1148 region downstream of cobD and generated a BamHI restriction endonuclease site. These primers were used to generate a 1291-bp fragment, which was gel-purified, digested with EcoRI and BamHI, and ligated into the intermediate copy number vector pSU38 (35) to generate plasmid pCOBD2.
Overexpression and Visualization of CobD-Two oligonucleotide primers were used to amplify the Ϫ16 to ϩ1148 region of cobD of plasmid pCOBC4 by PCR. One primer (primer 2) contained a 3Ј-region complementary to plasmid pCOBC4 followed by a 5Ј-noncomplementary end that generated a BamHI restriction endonuclease site. The second primer (primer 3) was designed to introduce an NdeI restriction Novel Decarboxylase Encoded by the cobD gene of Salmonella endonuclease site immediately 5Ј to the translation initiation site of cobD. Amplification between the two primers was performed using Vent R ® (exo Ϫ ) polymerase (New England Biolabs Inc.). Reaction conditions were as follows: denaturation at 94°C for 2 min, annealing at 50°C for 1.5 min, and extension at 72°C for 1 min. The primers used were as follows: primer 2 and primer 3 (5Ј-TTTTGGCTGGAGGCATAT-GGCTTTATTCAACACCG-3Ј), which hybridized to the Ϫ16 to ϩ19 region of cobD and generated an NdeI restriction endonuclease site. These primers generated a 1173-bp fragment, which was purified, digested with NdeI and BamHI, and ligated into the T7 overexpression vector pT7-7 (36). The resulting plasmid (pCOBD6) and pT7-7 were transformed into Escherichia coli strain BL21/DE3, generating strains JE4094 and JE4096, respectively. These strains contained the T7 RNA polymerase in a -lysogen under control of an IPTG-inducible promoter.
CobD overexpression was performed as follows. A 0.1% (v/v) inoculum was added to LB broth containing ampicillin (100 g/ml) and grown at 30°C with shaking to ϳ70 Klett units. IPTG was added to 400 M, and incubation was continued for ϳ12 h. A 100-l sample of culture was pelleted, and 100 l of 2 ϫ sample buffer (37) was added and heated at 100°C for 5 min. Cell-free extracts were cleared after spinning at 12,000 ϫ g for 5 min in a Sorvall ® SS34 rotor RC5B low speed ultracentrifuge (DuPont).

Biochemical Techniques
In Vitro Assays for CobD Activity-Cell-free extracts used in the in vitro activity assays were obtained from 1-liter cultures grown in LB broth containing ampicillin (100 g/ml). Cultures were shaken at 30°C in 2-liter Erlenmeyer flasks. After induction, cells were pelleted, resuspended in 20 ml of 50 mM PIPES (pH 6.8), and broken by passing twice through a French pressure cell at 8958 kilopascals. Membrane and soluble fractions were separated by centrifugation at 40,000 ϫ g for 1.5 h. The supernatant was dialyzed at 4°C against 50 mM PIPES and 10 mM dithiothreitol (pH 6.8) and kept under N 2 with 10 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride added. All reactions were performed in 1.5-ml Eppendorf tubes (Fisher).
General Assay and Derivatization Conditions-Both aminotransferase and decarboxylase assays were started by the addition of ϳ100 g of protein. The reaction mixtures were incubated at 37°C for either 30 or 90 min as indicated. The reactions were terminated, and proteins were precipitated by incubation at 100°C for 10 min. Precipitated proteins were removed by a 1-min centrifugation using a Marathon™ 13K/M microcentrifuge (Fisher). Samples were filtered with a Spin-X ® filter (Corning Costar Corp., Cambridge, MA) to remove any debris prior to HPLC analysis.
A 5-l sample was derivatized with oPA in the presence of ␤-mercaptoethanol to generate highly fluorescent thio-substituted isoindole derivatives of primary amines as described previously (38). A 10-l sample of this mixture (ϳ1 nmol of substrate) was analyzed via HPLC as described below.
Alkaline Phosphatase Assays-Alkaline phosphatase treatments of cobD reaction mixtures were performed as follows. A 20-l sample of the reaction mixture was diluted to 28 l with alkaline phosphatase buffer (50 mM Tris-HCl (pH 9.3), 1 mM MgCl 2 , 100 M ZnCl 2 , and 1 mM spermidine) containing 2 units of calf intestinal alkaline phosphatase. The mixture was incubated at 37°C for 30 min. The reaction was terminated, and proteins were precipitated by incubation at 100°C for 10 min and removed by centrifugation. The sample was then filtered with a Spin-X ® filter to remove debris. Primary amines in the mixture (5-l sample) were derivatized, and 15 l of this mixture (ϳ1 nmol of substrate) was analyzed by HPLC as described above. DL-Thr-P was used as a standard for this procedure, which, after treatment and derivatization, eluted with a retention time of 28 min (same as L-Thr).
Protein Analysis-Proteins present in soluble cell-free extracts were resolved on 12% SDS-polyacrylamide gels using the Laemmli system (39) and visualized by Coomassie Blue staining. Protein concentration was estimated by the method of Kunitz (40).

Nucleotide Sequence of cobD and Predicted
Amino Acid Sequence of CobD-Previously reported work from our laboratory identified 211 bp of the 5Ј-end of the cobD gene (41). We determined the remaining sequence of cobD using plasmid pCOBC4 (cobC ϩ cobD ϩ ). Analysis of the sequence revealed an open reading frame of 1095 bp with an OPAL (TGA) stop codon (Fig. 2). The sequence predicted a polypeptide of 364 amino acids with a molecular mass of 40,810 Da.
A putative promoter was identified using the MacTargsearch program. The Ϫ10 region of the cobD promoter was identified 23 bp 5Ј to the putative translation start site. The Ϫ10 hexamer (TAAAAT) had a five out of six match with the consensus sequence, whereas the Ϫ35 hexamer (TTGCGC) had only a three out of six match. A putative ribosome-binding site sequence (Shine-Dalgarno sequence; GGAGG) was located 4 bp upstream of the initiation codon. This sequence analysis suggested that cobD and cobC were divergently transcribed and that only 1 bp separated the translation initiation codons of the two genes (Fig. 2). If the putative promoter and Shine-Dalgarno sequences of cobD were correct, they reside within the open reading frame of cobC. The predicted CobD polypeptide contained a consensus PLPbinding motif (SLTKFYAIPGLRLG) and showed striking homology to the histidinol-phosphate transaminases (EC 2.6.1.9) and, to a lesser extent, to the aspartate transaminases (EC 2.6.1.1) and tyrosine transaminases (EC 2.6.1.5) of many organisms. CobD was most similar to HisH, the histidinol-phosphate transaminase from Bacillus subtilis (GenBank™ accession number M15409), sharing 28% identity and 68% similarity at the amino acid level. Additionally, CobD is ϳ28% identical and 67% similar to the CobC protein of P. denitrificans (Gen-Bank™ accession number M32223), thought to be involved in (R)-AP synthesis (11).
Subcloning of cobD-Two oligonucleotide primers were used to amplify the cobD region of pCOBC4 using PCR methodology. Primers 1 and 2 contained a 3Ј-region complementary to pCOBC4 and a 5Ј-noncomplementary end that generated either an EcoRI or a BamHI restriction enzyme site. These primers were used to amplify the Ϫ125 to ϩ1148 region of pCOBC4. This fragment contained the entire cobD gene plus some flanking sequences. The 1291-bp fragment was digested with EcoRI and BamHI and ligated into the intermediate copy number vector pSU38 to generate pCOBD2. This plasmid complemented CobD function, i.e. restored cobalamin biosynthesis from (CN) 2 Cby without the addition of (R)-AP in cobD mutants. cobD was resequenced from pCOBD2 to ensure that no mutations were introduced during amplification.
Overexpression of CobD-Two oligonucleotide primers (primers 2 and 3) were used to amplify the cobD region of pCOBC4 using PCR. Primer 2 contained a 3Ј-region complementary to pCOBC4 and a 5Ј-noncomplementary end that generated a BamHI restriction enzyme site. Primer 3 was designed to introduce an NdeI restriction site immediately 5Ј to the translation start site of cobD. The NdeI restriction site was required to align cobD with the efficient ribosome-binding site of the T7 overexpression vector pT7-7. Primers 2 and 3 were used to amplify the Ϫ16 to ϩ1148 region of cobD. The 1173-bp fragment generated was digested with NdeI and BamHI and ligated into pT7-7 digested with the same enzymes to generate plasmid pCOBD6. cobD was resequenced from pCOBD6 to ensure that no mutations were introduced during amplification. Plasmid pCOBD6 and the control overexpression vector (pT7-7) were transformed into E. coli strain BL21/DE3 to generate strains JE4094 (pCOBD6) and JE4096 (pT7-7), which were used in overexpression experiments. Following induction, proteins in the crude cell-free extracts were resolved by SDSpolyacrylamide gel electrophoresis and visualized by Coomassie Blue staining. As shown in Fig. 3, cell-free extracts of strain JE4094 grown in the presence of IPTG (lane B) contained an extra protein band of ϳ40 kDa compared with the strain containing the expression vector (pT7-7) lacking cobD (lane D). Cell-free extracts of strain JE4094 grown without IPTG (lane A) contained drastically reduced amounts of the ϳ40-kDa protein. The molecular mass of the overexpressed protein correlated well with the predicted molecular mass of 40.8 kDa for CobD.
CobD Does Not Have Lactaldehyde Aminotransferase Activity-Computer analysis of cobD sequence data suggested CobD was a PLP-dependent aminotransferase that catalyzed the reaction responsible for the synthesis of (R)-AP. We investigated this possibility and hypothesized that lactaldehyde was one of the substrates for CobD. To investigate whether cobD encoded lactaldehyde aminotransferase, activity assays were performed with crude cell-free extracts of JE4094 (pCOBD6) and JE4096 (pT7-7) after induction of T7 RNA polymerase. The assays were performed by providing either D-or L-lactaldehyde as the acceptor molecule and one L-amino acid as the amino donor. All 20 common L-amino acids were tested as possible donors. After incubation with CobD-enriched cell-free extract, primary amine-containing compounds were analyzed by reverse-phase HPLC after derivatization with oPA. The derivatized compounds were detected with a fluorescence detector as described under "Experimental Procedures." These assays did not yield any primary amine compounds except for the amino acid included in the reaction mixture as a potential substrate (data not shown).
CobD Has L-Threonine-O-3-phosphate Decarboxylase Activity-To investigate whether cobD encoded DL-threonine-O-3phosphate decarboxylase, activity assays were performed on crude extracts of JE4094 (pCOBD6) and JE4096 (pT7-7) after induction of T7 RNA polymerase. The assays were performed by providing DL-Thr-P as substrate. After incubation with CobD-enriched cell-free extract, primary amine-containing compounds were analyzed by reverse-phase HPLC after derivatization with oPA as described above.
Assay mixtures containing cell-free extract of strain JE4094 were found to contain DL-Thr-P (elution time, 23 min) and one additional compound (elution time, 30 min) that was absent from mixtures containing cell-free extract of the control strain JE4096 (data not shown). Extended incubation (1.5 h) resulted in roughly equivalent amounts of substrate and product, suggesting that only one stereoisomer of DL-Thr-P was converted to product.
To determine which stereoisomer was the substrate for CobD, assays were performed with L-threonine O-3-phosphate, whose stereochemistry at the ␤-carbon is the same as that of (R)-AP. Assay mixtures containing cell-free extract of strain JE4094 (pCOBD6) converted more than half of L-Thr-P to product (elution time, 30 min) after a 30-min incubation (Fig. 4B). Extended incubation (1.5 h) resulted in complete conversion to product (Fig. 4C), indicating that L-Thr-P was the substrate for CobD. Control assay mixtures containing cell-free extract of strain JE4096 (pT7-7) did not contain detectable levels of product (Fig. 4A).

The Product of the CobD-catalyzed Reaction Is (R)-1-Amino-2-propanol O-2-Phosphate-The PLP-dependent decarboxylation of L-threonine O-3-phosphate predicted the formation of either (R)-1-amino-2-propanol or (R)-1-amino-2-propanol O-2-
phosphate depending on the reaction mechanism. The product of the reaction had a different elution time (30 min) compared with authentic (R)-AP (35 min), suggesting that the phosphate group was retained in the product. To test for the presence of the phosphate in the product, reaction mixtures were subjected to treatment with alkaline phosphatase. As shown in Fig. 5, this treatment changed the elution time of the product from 30 to 35 min, which corresponded with that of authentic (R)-AP. Alkaline phosphatase treatment of reaction mixtures incubated for only 30 min (i.e. incomplete) contained an additional compound whose elution time (28 min) was in good agreement with that of authentic L-Thr (data not shown). DISCUSSION We discuss below several important contributions to the field of cobalamin biosynthesis made by the work presented in this paper.
Role of CobD in Cobalamin Biosynthesis in S. typhimurium-CobD is the first enzyme reported to be able to decarboxylate L-threonine O-3-phosphate to yield (R)-1-amino-2propanol O-2-phosphate. In vitro data indicate that the CobD enzyme has stereospecificity for L-Thr-P and is unable to decarboxylate the D-isomer of this compound. Although it was not investigated in this work, the existence of a consensus PLPbinding motif within cobD makes it likely that this is a member of the PLP-dependent decarboxylases. The fact that the phosphate group in the substrate is retained in the product raises important mechanistic questions that will be best addressed when homogeneous protein becomes available.
The meaning of the homology of CobD to class I PLP-dependent aminotransferases and the lack of homology of this enzyme to other PLP-dependent decarboxylases suggests that CobD may be a member of a new family of decarboxylases that evolved from a common ancestor of class I PLP-dependent aminotransferase enzymes.
Understanding the Role of CobD Changes Our View of the de Novo Corrin Ring Biosynthetic Pathway-On the basis of our data, we propose the model shown in Fig. 6 for the last step of de novo corrin ring biosynthesis in S. typhimurium, a reaction currently thought to be catalyzed by the CbiB protein. In this model, L-Thr is phosphorylated by an uncharacterized kinase to yield L-Thr-P, which is decarboxylated by CobD to yield (R)-1- amino-2-propanol-O-2-phosphate. The latter is proposed to be the true cosubstrate for the CbiB enzyme. Thus, we propose that the end product of the de novo pathway for corrin ring biosynthesis is AdoCbi-P, not AdoCbi as previously thought.
There is evidence that several proteins, including one homologous to CobD, may be involved in the attachment of (R)-AP in P. denitrificans. It is interesting that Rémy et al. (42) found an unexpectedly high K m (ϳ20 mM) for (R)-AP in an in vitro system that generated AdoCbi from (R)-AP and 5Ј-deoxyadenosylcobyric acid. This high K m for (R)-AP suggests that the correct substrate for this reaction may be (R)-1-amino-2-propanol-O-2-phosphate in this organism as well.
Timing of Decarboxylation-It is not clear what the timing of the decarboxylation is. It is possible that L-Thr-P is the substrate for CbiB and that carboxylated AdoCbi-P is the substrate for CobD. At this point, no evidence is available to rule out this possibility, although we feel that it is an unlikely scenario given the putative involvement of PLP in the reaction. If the mechanism of catalysis involves the formation of an imine (Schiff base) between PLP and the ␣-amino group of L-Thr-P prior to decarboxylation, such a bond would not be possible if the ␣-amino group were derivatizing the propionyl substituent of ring D of the macrocycle (Fig. 1). If, however, decarboxylation occurs after amidation of the propionyl substituent, then the interaction of PLP with the secondary amine at C-␣ would yield an enamine. Further insights into this problem require the biochemical analysis of the CbiB protein.
Explanations for the Observed Phenotypes of cobD Mutants-Since adenosylcobalamin biosynthesis in cobD mutants is restored by exogenously supplied (R)-AP (21), it is assumed that this compound must be phosphorylated before it can be used as substrate by the CbiB enzyme (Fig. 6). Alternatively, CbiB may be able to catalyze the synthesis of AdoCbi from 5Ј-deoxyadenosylcobyric acid if an excess of unphosphorylated (R)-AP is available. If CbiB cannot use (R)-AP as substrate, we predict the existence of a kinase enzyme to convert (R)-AP into (R)-1amino-2-propanol O-2-phosphate (Fig. 6). Again, insights into this problem must await the biochemical analysis of the CbiB

protein.
Implications of the Synthesis of AdoCbi-P on Our View of the Late Steps of the Cobalamin Biosynthetic Pathway (i.e. the Nucleotide Loop Assembly Pathway)-Our proposal that AdoCbi-P is the end product of corrin ring biosynthesis raises an important question regarding the role of the kinase activity of the CobU enzyme in the synthesis of AdoCbi-GDP from AdoCbi via an AdoCbi-P intermediate (43). To reconcile the results presented herein with the documented kinase activity of CobU, we hypothesize that under anaerobic growth conditions (e.g. in the gut) when S. typhimurium can synthesize the corrin ring de novo, the kinase activity of CobU is not required for the assembly of the nucleotide loop because the product of the CbiB reaction is AdoCbi-P. However, the kinase activity of CobU would become relevant under aerobic growth conditions (e.g. free-living) when unphosphorylated cobinamide may be present in the environment. A prediction from this hypothesis is that mutant CobU enzymes lacking only kinase activity should display a nucleotide loop assembly phenotype under aerobic growth conditions, but should be able to synthesize adenosylcobalamin de novo under anaerobic growth conditions. It is possible, however, that (R)-1-amino-2-propanol O-2phosphate may be dephosphorylated before, during, or after attachment to 5Ј-deoxyadenosylcobyric acid. In this scenario, the end product would be AdoCbi, and the kinase activity of CobU would be required for synthesis of the nucleotide loop regardless of growth conditions. These ideas are currently under investigation.
Implications of the Organization of the cobD and cobC Genes-Analysis of nucleotide sequence data raises important questions regarding regulation of expression of cobD and its neighbor, cobC. The fact that cobD is separated by only 1 base pair from the neighboring, divergently transcribed cobC gene places the putative regulatory region for cobD within the cobC coding sequence. Similarly, the putative cobC promoter appears to be located within the cobD coding sequence. The effect of this organization on the transcription of both genes is unclear and is currently under investigation. If this organization were correct, we predict that insertions of transposable elements proximal to the 5Ј-end of either cobD or cobC should affect both gene functions, resulting in strains displaying additional phenotypes to those expected for cobC or cobD mutants with lesions outside the overlapping regions.
Comparison of the cobC/cobD Region of S. typhimurium with E. coli-The E. coli phpB gene (GenBank™ accession number U23163) showed homology to cobC. When comparing the two genes, we found that cobC contained an additional 93 bp at the 5Ј-end. Over the region that phpB shares homology with cobC, The remaining 93 bp of the 5Ј-end of cobC showed no identity to the region upstream of phpB. There was no identity to all but the extreme 3Ј-end of cobD. A, the 155-bp region immediately 3Ј to cobC shared 79% identity with the region 3Ј to phpB. B, phpB was 68% identical to the 3Ј-end of cobC. PhpB was 68% identical and 85% similar to this region of CobC at the protein level. C, the first 93 bp of cobC did not share homology with the phpB region of E. coli. D, the last 25 bp of cobD and 103 bp 3Ј to cobD shared 74% identity with the 3Ј-end of orfUU. *, this 22-bp region between phpB and orfUU was not homologous to any portion of the cobC/cobD region; q, these two regions are 53-bp inverted repeats. The repeat 3Ј to cobC was not homologous to E. coli sequence 3Ј to phpB. they were 68% identical (Fig. 7). At the protein level, they were 68% identical and 85% similar. Although E. coli has homologs to many S. typhimurium genes involved in the later steps of Cbl biosynthesis, we did not identify a homolog to cobD. Since cobC and cobD are adjacent in S. typhimurium, we analyzed the phpB region of E. coli further (GenBank™ accession number AE000168 U00096) (44). We found that E. coli contained regions homologous to sequences 3Ј to both cobC and cobD. Interestingly, the last 25 bp of cobD and 103 bp immediately 3Ј to cobD shared 74% sequence identity with the 3Ј-end of the hypothetical orfUU gene of E. coli. The function of OrfUU is not known, but appears to be essential for growth. We do not have additional sequence 3Ј to cobD to determine whether S. typhimurium possesses a homolog to orfUU.
It appears as though the 93 bp of the 5Ј-end of cobC and the 1070 bp 5Ј of cobD (1164 bp total) were either gained by S. typhimurium or deleted from E. coli. In E. coli, phpB and orfUU are separated by an intergenic region of only 22 bp that showed no homology to either the 5Ј-end of cobC or the 3Ј-end of cobD.
Further analysis of the S. typhimurium cobC/cobD region identified two 53-bp regions of identical inverted repeats. One of these regions spanned from bp ϩ883 to ϩ935 of the cobD sequence. The second region was located 158 -210 bp 3Ј to cobC. This second region is 3Ј to a region with 79% identity to E. coli; however, the homology stopped abruptly at the beginning of the repeat. Because this repeat spans all the sequence 3Ј to cobC that we have so far obtained, the size of the repeat may be larger than 53 bp. The evolutionary significance of these inverted repeats is unclear and is currently under investigation.