INTRODUCTION

The primary cobamide synthesized by Salmonella typhimurium LT2 is Co-5,6-dimethylbenzimidazolyl-Co-adenosylcobamide (Ado-Cbl)¹ (Fig. 1). Cobamides containing compounds other than 5,6-dimethylbenzimidazole (Me₂Bza) as lower (Co) ligands have been isolated from other prokaryotes. The lower ligands fall within two classes: (i) derivatives or analogs of benzimidazole (Bza) (e.g. 5-hydroxybenzimidazole, 5-methylbenzimidazole, 5-methoxy-6-methylbenzimidazole, adenine) (3-5) and (ii) phenolics (e.g. p-cresol, phenol) (5, 6). It is unclear why some cobamide-producing prokaryotes have evolved to synthesize these alternative cobamides.

Cobamide-producing prokaryotes can be guided into synthesizing cobamides containing a lower ligand base different from the one usually found in the de novo synthesized cobamide by exogenously providing the alternative base to growing cells. Examples of this phenomenon termed "guided biosynthesis" have been documented (1, 7, 8).

Friedmann et al. (9) proposed that the cobalamin biosynthetic enzyme nicotinate-mononucleotide (NaMN):Me₂Bza phosphoribosyltransferase plays an important role in the incorporation of alternative lower ligand bases into cobamides in nature and during guided biosynthesis. This proposal was based on the observation that partially purified NaMN:Me₂Bza phosphoribosyltransferases from Propionibacterium shermanni and from Clostridium sticklandii were both shown to use a variety of base substrates in place of Me₂Bza in vitro (10, 11). It was proposed that this lack of base substrate specificity was likely responsible for the incorporation of alternative bases into cobamides in vivo. However, the incorporation of alternative NaMN:Me₂Bza phosphoribosyltransferase products generated in vitro into active cobamides in vivo was not demonstrated.

In this paper we report the purification and biochemical characterization of the NaMN:Me₂Bza phosphoribosyltransferase (CobT) from S. typhimurium. Previously we established a correlation between the presence of CobT and phosphoribosyltransferase activity using crude cell-free extracts containing increased levels of CobT protein (12). In this paper we demonstrate that homogeneous CobT is necessary and sufficient for the synthesis of N¹-(5-phospho--D-ribosyl)-5,6-dimethylbenzimidazole (-ribazole 5-phosphate) (Fig. 2) and that the enzyme phosphoribosylates a wide range of alternative base substrates in vitro. We also present evidence that the resulting ribotides are incorporated into active cobamides in vivo. These results strongly support that CobT is responsible for the incorporation of alternative lower ligands into cobamides in S. typhimurium. Additionally, this result may help explain the inability to isolate mutants blocked in the synthesis of Me₂Bza in this bacterium.

CobT-catalyzed synthesis of -ribazole 5-phosphate.

EXPERIMENTAL PROCEDURES

Strain Construction, Culture Media, and Growth Conditions

The two bacterial strains used in this work were derivatives of S. typhimurium strain LT2 and contained metE205 ara-9 mutations in the background. Strain JE2461 (cobA367::Tn10d(Tc)/pGP1-2 T7 rpo⁺ kan⁺ pJO27 bla⁺+ cobT⁺ (cloning vector: pT7-5)) (12) was used to increase the cellular level of CobT. Strain JE3762 (metE205 ara-9 cobB1206::MudJ DELcob291 strA1) was used in the biological activity assay system described below. DELcob291 spans the entire cobT gene (13).

The composition of complex and chemically defined culture media and the concentration of nutritional supplements and antibiotics have been reported (14, 15). Growth conditions for cobT overexpression were as follows. One-liter cultures of strain JE2461 were grown with shaking (180 orbits/min) in Luria-Bertani medium supplemented with ampicillin (60 µg/ml) and kanamycin (60 µg/ml) in 2-liter Erlenmeyer flasks at 30 °C for 3.5 h (about 3 × 10⁸ colony-forming units/ml; 55-70 Klett units). Overexpression of the CobT protein was induced by transferring the cultures to a 42 °C shaking water bath (200 rpm). After 30 min at 42 °C, the cultures were transferred to 37 °C and incubated with shaking (180 orbits/min) for 3 h.

Growth conditions for the biological activity assays were as follows. Cells grown overnight in complex medium were pelleted and washed three times with 14.5 mM sterile NaCl. Approximately 10⁸ cells were resuspended in 3 ml of molten 0.7% (w/v) agar and overlaid on one of the following: (i) Vogel-Bonner minimal medium containing glucose (11 mM) as carbon and energy source and cobinamide (20 nM); or (ii) No carbon Vogel-Bonner minimal medium containing 1,2-propanediol (50 mM) or ethanolamine (20 mM) as carbon and energy source, cobinamide (20 nM), and methionine (0.3 mM). Cbl (7 pmol) was spotted onto each plate to serve as the positive control; Me₂Bza, 4,5-dimethyl-1,2-phenylenediamine (Me₂Pda), Bza, histidine, adenine, guanine, or imidazole (150 nmol) was spotted and served as the negative control. Plates were incubated at 37 °C.

In Vivo Assessment of Cbl Biosynthesis

Cbl biosynthesis was assessed in vivo by demanding strain JE3762 to grow under conditions that required the synthesis and utilization of Cbl. Strain JE3762 was unable to synthesize Cbl from its precursors cobinamide and Me₂Bza but synthesized Cbl when provided with -ribazole 5-phosphate. Two in vivo biological activity assays with different sensitivities were employed.

Methionine synthase catalyzes the last step in methionine biosynthesis. The metE and metH genes of S. typhimurium LT2 encode different methionine synthase enzymes. The activity of MetH (EC 2.1.1.13) is Cbl-dependent, whereas that of MetE (E.C.2.1.1.14) is Cbl-independent (16). Strain JE3762 carries a metE mutation that forces the cells to synthesize methionine using the MetH enzyme. Therefore, derivatives of metE mutants unable to synthesize Cbl are phenotypically methionine or Cbl auxotrophs (17). Approximately 25 Cbl molecules/cell satisfy the methionine requirement of a growing cell (18).

Ethanolamine ammonia-lyase (EC 4.3.1.7), the first enzyme in ethanolamine catabolism in S. typhimurium LT2, requires Ado-Cbl to function (19). 1,2-Propanediol dehydratase (EC 4.2.1.28), which catalyzes the first step in 1,2-propanediol catabolism in this bacterium, is also an Ado-Cbl-dependent enzyme (20, 21). Therefore, strains that cannot synthesize Ado-Cbl are unable to utilize 1,2-propanediol or ethanolamine as carbon and energy sources. Approximately 500 Cbl molecules/cell are needed for continued growth on ethanolamine (18).

In Vitro Phosphoribosyltransferase Activity Assay

The in vitro phosphoribosyltransferase assay relies on changes in the chromatographic behavior of radiolabeled Me₂Bza as described elsewhere (12). CobT protein was diluted with glycine-NaOH buffer (50 mM, pH 10.0) containing bovine serum albumin (0.5 mg/ml) and sodium azide (0.03%, 4.6 mM). The final concentration of bovine serum albumin in the reaction mixture was 0.05 mg/ml. The reaction mixture volume was 20 µl and contained homogeneous CobT (0.012 µg), NaMN (100 nmol), [2-¹⁴C]Me₂Bza (2 nmol, 42.5 µCi/µmol), and glycine-NaOH buffer (2 µmol, pH 10.0) unless otherwise stated. Reactions were run for 10 min at 37 °C unless otherwise stated. A unit of CobT activity was defined as the amount of enzyme needed to synthesize 1 µmol of -ribazole 5-phosphate/min. This assay was used for different analyses as described below.

The reaction mixtures contained [2-¹⁴C]Me₂Bza (2.5 nmol), unlabeled Me₂Bza (20 nmol), and NaMN (20 nmol). The amount of enzyme assayed was always within the linear portion of the rate versus [enzyme] plot. The amounts of protein used for each purification step were: crude cell-free extract, 0.7 µg; phenyl-Sepharose-purified CobT protein, 0.19 µg; Cibracon Blue-purified CobT protein, 0.09 µg. Reactions were run in duplicate and averaged.

The reaction mixtures contained [2-¹⁴C]Me₂Bza (2.5 nmol), unlabeled Me₂Bza (20 nmol), and NaMN (20 nmol). Assays were performed at pH 9.7. For each optimization condition, the average of five independent reactions is reported. For optimal pH assays, the following buffers were used: phosphoric acid, pH 12.0; disodium hydrogen phosphate-NaOH, pH 11.0; glycine-NaOH, pH 10.0, pH 9.7; CHES, pH 9.0; HEPES, pH 8.5, pH 8.0; and MOPS, pH 7.0.

Quantification of -Ribazole 5-Phosphate

The product of the reaction was detected using a PhosphorImager 4451 (Molecular Dynamics, Sunnyvale, CA). After separation of product ([¹⁴C-2]-ribazole 5-phosphate) from reactant ([¹⁴C-2]Me₂Bza) by thin layer chromatography (see below), the thin layer chromatography sheet was exposed to a PhosphorImager screen for 20-25 h, and the percentage of total radioactivity in an assay corresponding to [¹⁴C-2]-ribazole 5-phosphate was determined. Background was determined from a no-enzyme control. The amount of radioactivity in a sample (7 µl) from the reaction mixture was assessed by scintillation counting to determine the total number of counts in the reaction mixture. The total number of counts in the assay corresponding to product was then determined.

Radiolabeled Me₂Bza was synthesized as described (12) with the following modifications. 20 µmol of [¹⁴C]formic acid (NEN Life Science Products; specific radioactivity, 55 mCi/mmol) and 40 µmol of Me₂Pda were used in a 100-ml final volume. Radiolabeled [¹⁴C-2]Me₂Bza (42.5 µCi/µmol) was stored in 100% methanol at 0 °C.

Biochemical Characterization of the Enzymatic Activity

The kinetic constants for NaMN were determined by holding [2-¹⁴C]Me₂Bza constant at a saturating concentration (2 nmol) while the NaMN concentration was varied. For the kinetic analysis of Me₂Bza, a saturating concentration of NaMN (100 nmol) was used while the Me₂Bza concentration was varied. Data from four independent trials were analyzed with the nonlinear regression data analysis program Enzfitter (Elsevier-Biosoft, Cambridge, United Kingdom).

Nicotinamide mononucleotide (NMN) was substituted for NaMN in reaction mixtures containing varying concentrations of NMN and a saturating concentration of [¹⁴C-2]Me₂Bza (2 nmol). Data from three independent trials were analyzed with the nonlinear regression data analysis program Enzfitter.

Ribose 5-phosphate (0.96 µmol, 160 nmol, and 16 nmol) was substituted for NaMN in reaction mixtures.

Bza, Me₂Pda, imidazole, histidine, adenine, or guanine (1 µmol each) was added to the reaction mixture in lieu of Me₂Bza. Because these compounds were not radioactive, formation of products was detected using a biological activity assay (see above). If the reaction mixture restored the ability of strain JE3762 to make a physiologically active cobamide, growth was observed. To increase the amount of product, assays were scaled up to 1 ml (50-fold) and contained 15 µg of CobT and 3 µmol of NaMN. Assays were run at pH 10.0, incubated for 1 h at 37 °C, heat inactivated at 90 °C for 15 min, centrifuged for 10 min (14,800 × g) in a Marathon 13K/M microcentrifuge (Fisher Scientific), and the supernatant concentrated under vacuum with a SpeedVac concentrator (Savant Instruments, Inc., Farmingdale, NY). Assays were resuspended in 50 µl of double-distilled water, and a sample (10 µl) was tested for biological activity. Identical assays were run without CobT to demonstrate the dependence of the biological activity on CobT.

Phosphoribosyltransferase reactions were performed using [8-¹⁴C] adenine in place of Me₂Bza. The 20-µl reactions used 100 nmol of NaMN and 0.18 µg of CobT. The assay was incubated at 37 °C for 60 min.

Chromatography and Spectroscopy

HPLC purification of the CobT reaction product was performed as described previously (12).

Native molecular mass analysis of CobT was performed using a BioSep-SEC-S2000 column (300 × 7.8 mm; Phenomenex, Reno, NV) equilibrated with 50 mM H₂PO₄, pH 7.0, at a rate of 1 ml/min. Protein elution was monitored at 280 nm. Protein molecular weight standards used included blue dextran, cytochrome c, carbonic anhydrase, alcohol dehydrogenase, and -amylase (Sigma). A 20-µl sample (4 mg/ml) was injected. All samples were run twice and averaged. The elution time for CobT was 7 min.

Thin layer chromatography analysis of the CobT reaction products was performed as described previously (12) with the following modifications. Silica gel plates (20 × 10 cm) were used; 1-cm-wide lanes separated by 0.5 cm were scored on the plates; plates were developed for approximately 45 min using CHCl₃:MeOH (3:2).

UV-visible spectroscopy and fast atom bombardment mass spectrometry of the CobT reaction product were performed as described (12).

Purification of CobT

All purification steps were performed at 4 °C. The buffer used throughout the purification was 50 mM Tris-HCl, pH 7.5 (at 4 °C). Additions to this buffer are stated below. The buffers used throughout the purification (except to resuspend the cells) contained ethylene glycol or glycerol as indicated to stabilize the protein.

A 2-liter culture of strain JE2461 was grown under conditions that overexpressed cobT. The resulting 3.4 g of wet paste was resuspended in 45 ml of buffer. After the addition of protease inhibitors (EDTA, 1 mM; phenylmethylsulfonyl fluoride, 0.5 mM), cells were broken by sonication (18 min, 50% duty cycle; setting 5) using a model 550 Sonic Dismembrator (Fisher Scientific). To minimize heat denaturation of protein during sonication, the extract was maintained at a temperature below 15 °C. Cell debris was removed by centrifuging at 44,000 × g for 1 h (Sorvall RC-5B refrigerated centrifuge, DuPont Instruments).

Finely ground Ultrapure^TM ammonium sulfate (Schwarz/Mann, ICN Biomedicals Inc., Cleveland, OH) was added to the extract to 10% saturation, incubated at 4 °C for 30 min, and centrifuged for 15 min at 10,000 × g to remove precipitates. Ethylene glycol was added to the supernatant (10% (v/v), i.e. 1.8 M). The extract was loaded onto a phenyl-Sepharose CL-4B (Sigma) column (2.5 × 6.0 cm, a 30-ml bed volume) equilibrated with buffer containing 1.8 M ethylene glycol and 10% saturation ammonium sulfate. The column was equilibrated and developed at a flow rate of 30 ml/h. After loading, the column was washed with 30 ml of the equilibrating buffer followed by a 150-ml linear gradient that simultaneously decreased the amount of ammonium sulfate from 10% saturation to zero and increased the ethylene glycol concentration from 1.8 to 8.9 M; the gradient was followed by a 60-ml wash with buffer containing 8.9 M ethylene glycol. CobT eluted from the column toward the end of the gradient. Fractions containing CobT were identified by SDS-PAGE and Coomassie Blue staining. CobT fractions were pooled and loaded directly onto the next column.

The sample was loaded onto a 1.5 × 14-cm Cibracon Blue 3GA (Sigma) column (25-ml bed volume) equilibrated with buffer containing 5.4 M ethylene glycol at a flow rate of 25 ml/h. After loading the sample, the column was washed with 50 ml of buffer containing 1.4 M glycerol and 200 mM NaCl; the column was developed with a 125-ml linear gradient to buffer containing 1 M NaCl and 1.4 M glycerol; the column was further washed with 50 ml of buffer containing 2 M NaCl and 1.4 M glycerol. CobT eluted within the gradient, with the peak of protein detected at 450 mM NaCl. A cross-section of the CobT-containing fractions revealed homogeneous CobT as judged from Coomassie Blue-stained SDS-PAGE gels. These homogeneous fractions were pooled, concentrated using a Centriprep-10 (Amicon, Inc., Beverly, MA), and dialyzed overnight against buffer containing 2.7 M glycerol. The enzyme was aliquoted and stored at 89 °C in a Revco freezer.

NH₂-terminal Sequencing

The NH₂-terminal sequence of purified CobT was determined at the Protein/Nucleic Acid Shared Facility at the Medical College of Wisconsin (Milwaukee, WI) using conventional Edman sequence chemistry. Sequencing was conducted with a Beckman/Porton LF 3000 gas phase sequenator. Data were reported using the Beckman system Gold software.

Other Procedures

Protein concentrations were determined by a modification of the turbidimetric method reported by Kunitz (22) using a standard curve generated using known quantities of bovine serum albumin. These values corresponded well to values obtained using the Bio-Rad Protein Assay Kit, which is based on the Bradford dye-binding procedure (23). SDS-PAGE was performed with 12% polyacrylamide gels (24) and stained with Coomassie Blue (25). SDS-PAGE standards included phosphorylase b, bovine serum albumin, ovalbumin, carbonic anhydrase, and trypsin inhibitor (Bio-Rad). The native isoelectric point (pI) of CobT was determined as reported (26). Conductivity measurements were performed using a model 35 YSI conductivity meter (Yellow Springs, OH). Densitometry scans were performed with a GS300 Transmittance/Reflectance Scanning Densitometer (Hoefer Scientific Instruments, San Francisco).

RESULTS

cobT Overexpression and Visualization of CobT

cobT was overexpressed using a construct where cobT was transcribed from a bacteriophage T7 promoter (27). The cloning of this construct was described previously (28). Based on densitometry scans of Coomassie-stained SDS-PAGE gels, CobT was judged to be approximately 12% of the total soluble protein in crude cell-free extract of the overexpressing strain (data not shown). The relative mobility of CobT in this system was determined previously (28).

Purification of CobT

The purification of CobT from crude cell-free extract is summarized in Table I. CobT specific activity in the crude extract (Table I) was determined 12 h after sonication and storage at 4 °C. This value (0.22 unit/mg), was 1.5-fold lower than the specific activity obtained immediately after sonication (0.33 unit/mg), suggesting inactivation or degradation of CobT. Interestingly, after the initial loss of activity, the specific activity did not decrease any further, even after 1 week of continued storage at 4 °C (data not shown). One explanation for these results may be that whatever was responsible for inactivating or degrading CobT also became inactivated or degraded.

Table I. CobT purification profile Total protein Total activity Specific activity Yield Purification mg µmol product/min units/mg protein % -fold Cell-free extract 256.5 55.8 0.2 100 0 Phenyl-Sepharose CL-4B 27.6 51.1 1.9 92 8.5 Cibracon Blue 3GA 5.3 24.0 4.5 43 20.7

The purification of CobT is shown in Fig. 3. CobT was 97% pure as judged by densitometry scans of Coomassie-stained 12% SDS-PAGE (lane C, Fig. 3).

NH₂-terminal Sequence

The NH₂-terminal sequence (21 amino acids) of homogeneous CobT was determined by automated Edman degradation. The amino acid sequence of homogeneous CobT was identical to that predicted from the DNA sequence confirming that the cobT product had been purified. Additionally, the sequence established that as predicted elsewhere (29), the translational start site was 30 bases downstream of the start site originally proposed (30).

In Vitro Requirements for CobT Phosphoribosyltransferase Activity

Previously, using crude cell-free extract containing increased levels of CobT, we demonstrated that CobT was the NaMN:Me₂Bza phosphoribosyltransferase responsible for the synthesis of -ribazole 5-phosphate, an intermediate in the assembly of the nucleotide loop of Cbl (Fig. 2) (12). Using homogeneous CobT, we repeated these experiments to investigate the exact requirements for this activity in vitro. After HPLC purification, the identity of the product was confirmed by UV-visible spectroscopy and mass spectral analysis to be -ribazole 5-phosphate (data not shown). In addition, the product was determined to be biologically active, i.e. it restored the ability of a cobT mutant strain (JE3762) to synthesize Cbl. Therefore, it was concluded that homogeneous CobT was necessary and sufficient to generate -ribazole 5-phosphate from NaMN and Me₂Bza in vitro.

Characterization of the CobT Activity

The optimal temperature for CobT activity was 45 °C (6.3 units/mg). CobT maintained 12% of its specific activity at 65 °C (0.8 unit/mg) and 45% at 30 °C (2.8 units/mg). Although maximum activity was obtained at 45 °C, activity assays were performed routinely at the more physiological temperature of 37 °C.

The optimal pH for CobT activity was pH 10.0 (6.5 units/mg). CobT maintained 28% of its specific activity at pH 8.0 (1.8 units/mg) and 2% at pH 11.0 (0.11 unit/mg).

CobT was preincubated for 10 min at several temperatures by itself, or with NaMN (20 nmol), or with Me₂Bza (20 nmol). CobT was then assayed for phosphoribosyltransferase activity at 37 °C. The specific activities were compared with the one observed when CobT was preincubated on ice. Preincubation at 37 °C with no additions resulted in a drop in specific activity (2.5 units/mg versus 5.4 units/mg, i.e. 46% of the specific activity of CobT kept on ice). However, CobT activity was stable to temperatures as high as 50 °C when preincubated with NaMN (4.5 units/mg). We concluded that NaMN has a stabilizing effect on CobT at increased temperatures. Me₂Bza did not increase CobT stability at any temperature (data not shown).

CobT activity was inhibited about 2.6 fold at increased Me₂Bza concentrations. CobT activity at 1.0 mM Me₂Bza was 5.9 units/mg compared with 15.1 units/mg at 0.1 mM Me₂Bza.

Other Properties

CobT activity was linear with respect to protein concentration up to 0.02 µg. CobT activity was linear with respect to time up to 12 min.

Based on the cobT nucleotide sequence, CobT was predicted to be 356 amino acids in length with a molecular mass of 36,560 Da. This value corresponded well to the position of homogeneous CobT on 12% SDS-PAGE gels (Fig. 3, lane C). Native molecular mass was determined by gel filtration chromatography to be about 68,000 Da, suggesting that native CobT was a dimer.

The experimentally determined isoelectric point revealed two bands on isoelectric focusing gels corresponding to a pI of 5.8 and 5.6. Both points differ from the predicted isoelectric point for denatured protein, 6.1. The molar extinction coefficient of CobT at 280 nm was predicted to be 9,970 using the Edelhoch method as described (31).

Kinetic Parameters

Based on data from four independent trials, the apparent K_m of CobT for NaMN was 680 µM, the V_max was 134 pmol of product/min, and the k_cat was 817 pmol of product/min/pmol of CobT dimer. The apparent K_m value (680 µM) was similar to the CobT homolog from C. sticklandii (700 µM) (11) and 8-fold higher than the homolog from Pseudomonas denitrificans (83 µM) (32).

The apparent K_m of CobT for Me₂Bza could not be determined because of the low specific activity of [2-¹⁴C]Me₂Bza. However, we were able to conclude that the apparent K_m was less than 10 µM (data not shown). The P. denitrificans homolog had a K_m for Me₂Bza of 16 nM (32). The K_m for Me₂Bza of the C. sticklandii homolog was not determined.

Substrate Specificity

NMN and ribose 5-phosphate were both tried in lieu of NaMN in phosphoribosyltransferase reactions. CobT used NMN as substrate in vitro; however, CobT phosphoribosyltransferase activity was inhibited at [NMN] > 40 mM. Because of this inhibition, kinetic parameters were derived from a rate curve that did not reach saturation. Based on data from three independent trials, the apparent K_m of CobT for NMN was 30 mM, 44-fold higher than the one measured for NaMN. This supports the idea that NaMN is the in vivo substrate for CobT. The V_max was 114 pmol of product/min, and the k_cat was 695 pmol of product/min/pmol of CobT dimer. The CobT utilization of NMN was in contrast to the C. sticklandii homolog, which did not utilize NMN as a substrate (11). Ribose 5-phosphate was not utilized by CobT.

Phosphoribosyltransferase activity assays were performed with a variety of bases in place of Me₂Bza including Bza, Me₂Pda, imidazole, histidine, adenine, and guanine. Formation of product was detected using biological activity assays. If the reaction mixture restored the ability of strain JE3762 to make an active cobamide, growth was observed. No attempt was made to quantitate the growth response. Representative bioassay results are shown in Figs. 4 and 5. Reaction mixtures containing Me₂Bza, Me₂Pda, or Bza resulted in a product that restored growth when assaying methionine synthase (Fig. 4), 1,2-propanediol dehydratase (data not shown), and ethanolamine ammonia-lyase (data not shown). However, imidazole, histidine, adenine, and guanine reaction mixtures restored growth when assaying ethanolamine ammonia-lyase (Fig. 5) and 1,2-propanediol dehydratase (data not shown) but not methionine synthase (data not shown). These data demonstrate CobT recognition of a variety of base substrates in vitro, generation of cobamides from the CobT reaction products in vivo, and utilization of the resulting cobamides by cobamide-dependent enzymes in vivo. The implications of these results are addressed under "Discussion."

CobT utilization of adenine was tested directly by performing phosphoribosyltransferase activity assays using [8-¹⁴C]adenine in place of [2-¹⁴C]Me₂Bza followed by separation of product from reactant by thin layer chromatography. Because the product, adenine ribose 5-phosphate, had been reported to be heat-sensitive (11), the assays were terminated either with acid or heat. Regardless of the method of assay termination, CobT utilized adenine (1 mM) as a substrate with a specific activity of 0.8 unit/mg, 15-fold less active than the positive control Me₂Bza (12 units/mg).

DISCUSSION

We have demonstrated CobT recognition of a variety of bases as substrates in vitro and the incorporation of these alternative products into active cobamides in vivo. We know that the resulting cobamides in the case of the histidine, adenine, guanine, and imidazole reactions are not transformed into Cbl because no growth response was observed when methionine synthase activity was demanded. This supports that S. typhimurium incorporated the CobT in vitro reaction products directly into a cobamide in vivo. These results provide strong support for the role of the NaMN:Me₂Bza phosphoribosyltransferase in the incorporation of alternative lower ligands into cobamides in nature and during guided biosynthesis. Additionally, these results demonstrate that methionine synthase, ethanolamine ammonia-lyase, and 1,2-propanediol dehydratase have different requirements for the identity of the lower ligand.

Methionine synthase was more stringent than ethanolamine ammonia-lyase or 1,2-propanediol dehydratase in its utilization of different cobamides in vivo. One explanation is that the alternative lower ligands may prevent the enzyme from binding the cobamide. The structure of the methionine synthase domain that binds Cbl has been solved (33). The structure indicates that Me₂Bza and the nucleotide loop swing away from the corrin ring and are bound by the enzyme in a pocket. To achieve this conformation, the coordination bond between Me₂Bza and cobalt is replaced by a coordination bond between a histidine residue (His⁷⁵⁹) of the enzyme and cobalt. Perhaps the lower ligands of the alternative cobamides generated in our experiments did not fit into the binding pocket or were locked into a base-on conformation. If the inability to bind was the reason for the lack of activity, we would predict that ethanolamine ammonia-lyase and 1,2-propanediol dehydratase bind cobamides differently than methionine synthase. The structure of the Cbl binding domains of ethanolamine ammonia-lyase and 1,2-propanediol dehydratase have not been determined to date.

However, our results suggest that methionine synthase can use alternative cobamides that contain lower ligands structurally similar to Me₂Bza, since the products of the Bza and Me₂Pda reactions resulted in the in vivo synthesis of cobamides utilized by this enzyme. We cannot rule out that Me₂Pda may have been converted to Me₂Bza in vivo. However, in the case of Bza, we know that S. typhimurium synthesizes a physiologically active cobamide with Bza as the lower ligand when supplied with exogenous Bza (1), supporting the argument that benzimidazolyl-cobamide was generated in this assay. This suggests that methionine synthase is able to utilize cobamide containing Bza as the lower ligand.

The ability of methionine synthase to utilize alternative cobamides is further supported by Eberhard et al. (34) who showed that a variety of cobamides containing different lower ligands supported growth of an Escherichia coli strain that required active cobamide for methionine synthase activity. However, in contrast to our results, they also saw growth with imidazolyl-cobamide. It has been reported previously that some prokaryotes when supplied with exogenous complete cobamides can replace the alternative lower ligands with their natural base (8, 9). Perhaps this explains the in vivo activity observed for imidazolyl-cobamide. Mervyn and Smith (7) also list examples of cobamides besides Cbl which support in vivo methionine synthase activity.

Identification of Me₂Bza biosynthetic genes and isolation of their products have been elusive. The only Me₂Bza auxotrophs isolated in S. typhimurium carry mutations in cobT. We have proposed previously that cobT mutants are Me₂Bza auxotrophs because a protein referred to as CobB substitutes for CobT activity when endogenous Me₂Bza levels are increased (12).² Although we have not ruled out that CobT may in addition to its phosphoribosyltransferase activity be involved in Me₂Bza biosynthesis, we predict that CobT is not solely responsible as suggested elsewhere (35). There are a number of reasons that may explain the difficulty with isolating mutations in Me₂Bza biosynthetic genes: (i) null alleles of Me₂Bza biosynthetic genes may be lethal or (ii) more than one Me₂Bza biosynthetic pathway could exist in S. typhimurium. Alternatively, based on the results presented herein, it is possible that in vivo more than one base can substitute for Me₂Bza. Therefore Me₂Bza auxotrophy would not be observed unless the synthesis of all potential lower ligands was eliminated.