Reassessment of the Late Steps of Coenzyme B12 Synthesis in Salmonella enterica: Evidence that Dephosphorylation of Adenosylcobalamin-5′-Phosphate by the CobC Phosphatase Is the Last Step of the Pathway

ABSTRACT We report that cobC strains of Salmonella enterica serovar Typhimurium are impaired in the ability to salvage cobyric acid (Cby), a de novo corrin ring biosynthetic intermediate, under aerobic growth conditions. In vivo and in vitro evidence support the conclusion that this new phenotype of cobC strains is due to the inability of serovar Typhimurium to dephosphorylate adenosylcobalamin-5′-phosphate (AdoCbl-5′-P), the product of the condensation of α-ribazole-5′-phosphate (α-RP) and adenosylcobinamide-GDP by the AdoCbl-5′-P synthase (CobS, EC 2.7.8.26) enzyme. Increased flux through the 5,6-dimethylbenzimidazole and cobinamide (Cbi) activation branches of the nucleotide loop assembly pathway in cobC strains restored AdoCbl-5′-P synthesis from Cby in a cobC strain. The rate of the CobS-catalyzed reaction was at least 2 orders of magnitude higher with α-RP than with α-ribazole as substrate. On the basis of the data reported herein, we conclude that removal of the phosphoryl group from AdoCbl-5′-P is the last step in AdoCbl biosynthesis in serovar Typhimurium and that the reaction is catalyzed by the AdoCbl-5′-P phosphatase (CobC) enzyme. Explanations for the correction of the Cby salvaging phenotype are discussed.

We took genetic and biochemical approaches to further investigating the timing of phosphate removal in serovar Typhimurium during the assembly of the nucleotide loop. This analysis was aided by the uncovering of a new phenotype of serovar Typhimurium cobC strains, which we argue provides the first in vivo support for the preference of the AdoCbl-5Ј-P synthase (CobS) enzyme for ␣-RP. We conclude that ␣-RP is the physiological substrate for CobS, that CobC catalyzes the last step of the AdoCbl biosynthetic pathway, and that removal of the 5Ј-O-P from AdoCbl-5Ј-P by CobC is the last step of the pathway.
Corrinoid extractions. Two liters of minimal NCE medium supplemented with glycerol, MgSO 4 , trace minerals, CNCby, and DMB was inoculated with 20 ml of an overnight cell culture (2 ϫ 10 9 CFU/ml) of strain JE2718 (⌬cobC1151). Cultures were grown at 37°C in an orbital shaker at 3 cps for 24 h. Cells were harvested at 8,983 ϫ g for 15 min with a JLA-8.1000 rotor in an Avanti J-20 XPI Beckman/Coulter refrigerated centrifuge; cell paste (3.8 g) was stored at Ϫ20°C until being used. Cell paste was suspended in methanol (1:2, wt/vol) and incubated at 55°C with gentle shaking for 1.5 h. The cell suspension was cleared by centrifugation at 40,000 ϫ g for 2 h in a Beckman/Coulter J25-I refrigerated centrifuge. The supernatant was dried under vacuum with a SpeedVac concentrator (Thermo Savant), and dry material was suspended in 5 ml of distilled water, the pH was verified to be neutral, and it was filtered through a 0.2-m syringe filter (Nalgene). Corrinoids were purified from the sample with an Alltech 900-mg C 8 Maxi-Clean cartridge (Alltech) equilibrated with double-distilled water (ddH 2 O). The cartridge was washed with 5 ml of ddH 2 O and eluted with 2 ml of methanol. Eluted samples were dried under vacuum as described above, suspended in 250 l of ddH 2 O, and stored at 4°C until being used.
RP-HPLC. Corrinoids present in the samples were converted to their cyano form by the addition of KCN (1 mol; final reaction volume, 150 l), followed by irradiation with a 60-W incandescent light at a distance of 6 cm on ice for 15 min. Samples were filtered with Spin-X centrifuge filters (Corning). An efficient reverse-phase high-performance liquid chromatography (RP-HPLC) method for the resolution of cyanocorrinoids has been reported (5). We used a Waters HPLC system equipped with an Alltima (Alltech) HP C 18 HL 5 column (150 by 4.6 mm) equilibrated and developed at a flow rate of 1 ml min Ϫ1 . The column was equilibrated with a 70% A-30% B buffer system (see below). A 25-min linear gradient was applied until the composition of the buffer system was 30% A-70% B. The solvents used were as follows: buffer A, 150 mM potassium phosphate buffer (pH 8.0) containing 10 mM KCN; buffer B, 100% methanol. Corrinoid elution from the column was detected with a Waters photodiode array detector. Authentic CNCby and CNCbl were used as standards.
Mass spectrometry. HPLC-purified compounds were dried under vacuum, suspended in 1 ml of ddH 2 O, loaded onto an Alltech C 8 Maxi-Clean cartridge, and processed as described above; corrinoids were eluted with 1 ml of 100% methanol. Samples were dried under vacuum. Dried samples were suspended in 50 l of ddH 2 O and filtered on 0.22 Spin-X columns (Corning). Mass spectrometry analysis of the material was performed at the Mass Spectrometry Facility of the University of Wisconsin-Madison Biotechnology Center. Mass spectra were obtained with an MDS Sciex 4800 matrix-assisted laser desorption ionization-time-of-flight (MALDI-TOF) mass spectrometer.
(ii) CobS-enriched cell extracts. Ten milliliters of lysogenic broth containing ampicillin (100 g/ml) was inoculated with a fresh transformant of Escherichia coli strain C41(DE3) carrying plasmid pCOBS5 (cobS ϩ ) and incubated overnight at 37°C. Two liters of lysogenic broth supplemented with ampicillin (100 g/ml) in a 4-liter flask were inoculated with 10 ml of the overnight culture and incubated at 37°C with continuous shaking (200 rpm). Isopropyl-␤-D-thiogalactopyranoside was added to the cell culture at a final concentration of 0.4 mM when it reached a cell density at 650 nm of ϳ0.6. Cultures were incubated at 37°C for 3 h after induction and harvested by centrifugation as described for corrinoid extractions, and the cell paste (4.5 g, wet weight) was stored at Ϫ20°C until being used. The same procedure was performed with E. coli strain C41(DE3) carrying the empty pET15b plasmid as negative control.
Cell paste was resuspended in 30 ml of 0.1 M Tris-Cl buffer (pH 7.9, 24°C) containing 500 l of protease inhibitor cocktail for His-tagged proteins (Sigma). Cells were broken with a French pressure cell operating at 8 ϫ 10 4 kPa. Cell extracts were obtained after centrifugation at 4°C at 5,000 ϫ g for 15 min. Cell membranes were obtained from a high speed spin at 75,000 ϫ g for 90 min with a JA-25.50 rotor in an Avanti J-25I Beckman/Coulter refrigerated centrifuge. Membranes were resuspended in 10 ml of 0.1 M Tris-Cl buffer (pH 7.9, 24°C) with a glass homogenizer and were solubilized by the addition of DHPC (1,2diheptanoyl-sn-glycero-3-phosphocholine) to a final concentration of 15 mM. DHPC was added slowly to avoid denaturation of proteins. The detergentcontaining CobS-enriched extract was incubated on ice for 30 min and centrifuged at 4°C at 75,000 ϫ g for 30 min. Glycerol (10% [vol/vol], final concentration) was added to the soluble membrane extract, which was flash frozen in liquid N 2 prior to storage at Ϫ80°C. Protein concentration was determined with a Bradford Bio-Rad kit (7).
(iii) Synthesis of CobS substrates. Protocols for the synthesis and purification of AdoCbi-GDP and ␣-RP have been described (35,37). In this work, we used an Alltima HP C 18 AQ 5 column (150 by 4.6 mm) (Alltech) and a Waters 600 HPLC system to resolve products from reagents. ␣-R was derived from ␣-RP by treatment with shrimp alkaline phosphatase (1 U/l; Promega) followed by purification with a C 18 SepPak cartridge (Waters). Prior to use, the C 18 cartridge was equilibrated with ddH 2 O, washed with 10 ml of ddH 2 O after sample application, and eluted with 2 ml of 100% methanol. ␣-R was dried under vacuum as described above, suspended in 150 l of ddH 2 O, and stored at Ϫ20°C until being used.
(v) Bioassays for detection of CobS activity. Strain JE8248 (⌬cobS) was used as indicator strain. Cells of an overnight NB culture were washed twice with sterile saline. Two hundred microliters of culture was added to 4 ml of molten 0.7% (wt/vol) agar and overlaid on E medium (19) supplemented with glucose. In vitro reactions were diluted 1:10 (vol/vol) with ddH 2 O, and 1 l was spotted onto the overlay. CNCbl was applied as a positive control. Growth was assed after overnight incubation at 37°C; the diameter of the zone of growth is reported in centimeters.
(vi) RP-HPLC analysis of cobalamin synthase (CobS) reaction products. Reaction products were separated with a System Gold HPLC system (Beckman/ Coulter) equipped with an Alltech HP C 18 5 column (150 by 4.6 mm) (Alltech). System II for corrinoid separation method was employed as described previously (5). For quantitative purposes, a standard curve for CNCbl was constructed. The lower limit of detection of CNCbl was 5 pmol, with an upper limit of 2,000 pmol (r 2 ϭ 0.9999). A third of the reaction mixture volume (50 l) was injected, and the total amount of product was calculated by multiplying the amount of CNCbl detected by a factor of 3.

RESULTS
Serovar Typhimurium cobC strains cannot salvage cobyric acid (Cby). During the course of studies aimed at identifying the gene encoding the enzyme responsible for the phosphorylation of L-Thr and AP, we uncovered a new phenotype for cobC strains of serovar Typhimurium. We used Tn10d(tet ϩ ) transposon mutagenesis to isolate derivatives of strain JE2216 [cobD1302::Tn10d(cat ϩ )] that failed to salvage Cby when provided with AP but grew well when provided with Cbi (Fig. 1). To demonstrate that a single Tn10d(tet ϩ ) element caused the observed inability to make AdoCbl from Cby and AP, P22 phage was grown on the original mutant strain and the phage lysate was used as donor to transduce strain JE2216 to tetracycline resistance; the reconstructed strain failed to salvage Cby in the presence of AP. Sequencing of the DNA flanking the Tn10d(tet ϩ ) element showed that the transposon was inserted in the cobC gene (data not shown). The inability of the reconstructed strains to convert Cby and AP to AdoCbl was 100% cotransducible with a Tn10d(tet ϩ ) element located within the cobC gene (40). This growth defect was also observed with a ⌬cobC strain (JE2217). Although the extent of the deletion in strain JE2217 has not been established, it does not affect the expression of the adjacent cobD gene (data not shown).
To confirm this new phenotype of cobC strains, we tested a previously isolated cobD cobC strain (JE4724) from our laboratory strain collection. Indeed, strain JE4724 failed to convert Cby and AP to Cbl but efficiently converted cobinamide (Cbi) to Cbl. This new phenotype of a cobC strain suggested a block in de novo corrin ring biosynthesis caused by the apparent inactivation of the kinase responsible for the phosphorylation of L-Thr or AP. As predicted by the pathway shown in Fig. 1, the absence of AP-P would prevent conversion of Cby to Cbi-P by the Cbi-P synthase (CbiB) enzyme.
Supplements that bypass the need for CobC function during Cbl synthesis from Cby and AP in a cobC strain. (i) Addition of L-Thr-P to the medium. Given the structural similarity between L-Thr and AP (Fig. 1), we investigated whether the lack of CobC would affect the synthesis of L-Thr-P or AP-P from L-Thr or AP, respectively. Because AP-P was not commercially available but L-Thr-P was, all subsequent studies were performed with L-Thr-P. We assessed the responsiveness of serovar Typhimurium strain JE2718 (⌬cobC1151) to L-Thr-P during growth in minimal medium containing glucose and CNCby, with or without DMB; a cobC ϩ strain (TR6583) was used as positive control. L-Thr-P restored the conversion of Cby to AdoCbl by a cobC strain ( Fig. 2A). This result was consistent with a block in the phosphorylation of L-Thr in a cobC strain. Whether the effect was direct or indirect remained unclear. Surprisingly, addition of DMB in lieu of L-Thr-P to the medium also restored Cby salvaging in the cobC strain to the same degree as L-Thr-P did ( Fig. 2A). When added together, DMB and L-Thr-P further improved the ability of the cobC strain to salvage Cby ( Fig. 2A). Similar but much less pronounced effects were observed with the control cobC ϩ strain (Fig. 2B). Because the addition of DMB restored AdoCbl synthesis from Cby, it was unlikely that the CobC enzyme was involved in L-Thr or AP phosphorylation. This idea was supported by the inability of CobC to phosphorylate L-Thr or AP in vitro, regardless of the phosphate donor used in the assay (data not shown).
(ii) Increased levels of AdoCbi-P synthase (CbiB) enzyme restore Cby salvaging in cobC strains. We further explored the possibility that Cby salvaging in cobC strains was due not to a block in L-Thr or AP phosphorylation but to a defect farther VOL. 189, 2007 REASSESSMENT OF THE LATE STEPS OF B 12 SYNTHESIS down the pathway, perhaps related to the activity of the AdoCbl-5Ј-P synthase (CobS) enzyme. To test this possibility, we increased the level of AdoCbi-P synthase (CbiB) enzyme in the cell by placing the cbiB ϩ gene on a plasmid with a promoter that was no longer controlled by oxygen (23). We assessed growth of the cobC strain carrying plasmid pCBIB4 (cbiB ϩ ) in minimal medium with glycerol and Cby, with or without DMB (Fig. 3). We note that residual expression of cbiB ϩ from the plasmid used in this experiment was previously reported to restore AdoCbl synthesis in a cbiB strain (61). Growth of strain JE7777 (⌬cobC/pCBIB4 cbiB ϩ ) on medium containing Cby (without DMB supplementation) was greatly improved compared to the growth of control strains (Fig. 3A).
Increasing the level of the CbiB enzyme had the same effect on Cby salvaging as the addition of L-Thr-P or DMB to the culture medium (Fig. 2). When DMB was added to the medium, growth of the control strains (Fig. 3B) was strongly stimulated, and the combination of DMB and higher levels of CbiB enzyme resulted in growth of the cobC strain that was indistinguishable from that of the cobC ϩ strain (Fig. 3B). Growth of the ⌬cobC strain did not improve when the level of AdoCbl-5Ј-P synthase (CobS) enzyme was increased. On the contrary, increased levels of CobS resulted in a drastic Cby salvaging deficiency in the cobC strain (Fig. 4). Growth of a cobC strain carrying plasmid pCOBS8 (cobS ϩ ) was assessed in minimal medium containing glycerol and Cby, with or without DMB supplementation. Growth of strain JE9469 (⌬cobC/ pCOBS8 cobS ϩ ) in medium lacking DMB was significantly reduced compared to control strains (Fig. 4A). Addition of DMB to the medium stimulated growth of strain JE9469 (Fig.  4B), but growth was not as robust as that of the control strain (Fig. 4B). We note that increased levels of CobS enzyme did not have a negative effect on the growth of a cobC strain when Cbl was present in the medium, suggesting that high levels of CobS somehow prevent synthesis of AdoCbl from Cby in this strain (data not shown).
A cobC strain accumulates AdoCbl-P. We used RP-HPLC and MS analysis (Fig. 5 and 6) to analyze the corrinoid content in a serovar Typhimurium ⌬cobC strain (JE2718) grown in minimal NCE-glycerol medium supplemented with Cby and DMB. Under these conditions, we detected the accumulation of substantial amounts of Cbl-P and minor amounts of AdoCbl (Fig. 5). The identities of Cbl-P (12.5 min) and Cbl (16.3 min) were confirmed by their UV-visible spectra (not shown), by their masses (Fig. 6), and by the shift in retention time after treatment with alkaline phosphatase (Fig. 5). Upon dephosphorylation, the Cbl-P retention time shifted from 12.5 min to 16.3 min.
␣-RP is the preferred substrate of the AdoCbl-5-P synthase (CobS) enzyme. A physiological explanation for the results presented in Fig. 2 and 3 can be gleaned from Fig. 1. One possible explanation was that ␣-RP, not ␣-R (as shown in Fig.  1), was the preferred substrate of CobS. Results shown in Fig.  7 support this idea. To investigate this possibility, we used bioassays to analyze the activity of CobS as a function of ␣-RP or ␣-R. Data shown in Fig. 7 (inset) reflect the amount of complete corrinoids synthesized by CobS when the enzyme was incubated with saturating levels of AdoCbi-GDP and equimolar amounts of ␣-RP or ␣-R under the same conditions. The presence of Cbl in the reaction mixture was detected by a bioassay that employed a Cbl auxotroph as the indicator strain (JE8248 ⌬cobS). We interpreted the results from these experiments to mean that ␣-RP was a better substrate for CobS than ␣-R ( Fig. 7A and B), hence the better growth response of the indicator strain. We performed RP-HPLC to get a quantitative assessment of CobS activity as a function of ␣-RP or ␣-R. Under the same assay conditions, CobS converted 2,500 Ϯ 70 pmol of ␣-RP into AdoCbl-P in 30 s (Fig. 7), while CobS converted only 20 Ϯ 5 pmol of ␣-R into Cbl in 30 s (Fig. 7). Together, these results indicate that ␣-RP was the preferred substrate of CobS.

DISCUSSION
We report here the first in vivo evidence that the last step of the AdoCbl biosynthetic pathway in serovar Typhimurium is the dephosphorylation of AdoCbl-5Ј-P and that the CobC enzyme catalyzes this reaction. Our in vivo evidence is strongly supported by results from in vitro experiments. This new knowledge changes our current view of the late steps of AdoCbl biosynthesis in several ways. First, it does not support the idea that the condensation of AdoCbi-GDP and ␣-R by the cobalamin synthase (CobS) enzyme is the terminal step of the pathway. Second, it brings into question whether ␣-R is an intermediate of the pathway, as previously suggested (40,44,57). Although these conclusions directly apply to coenzyme B 12 synthesis in serovar Typhimurium, their validity remains to be shown in other B 12 -producing prokaryotes.
Why are cobC strains unable to salvage Cby? The block of Cby salvaging in cobC strains is a complex phenotype that we explain by modifying the sequence of reactions presented in Fig. 1 to that presented in Fig. 8. Two key modifications are noted. First, ␣-RP, not ␣-R, is the cosubstrate for the AdoCbl-5Ј-P synthase (CobS) enzyme, and second, CobC is the AdoCbl-P phosphatase that converts AdoCbl-P to AdoCbl, the end product of the pathway. Dephosphorylation of AdoCbl-P by CobC is not an unprecedented idea. In fact, evidence that CobC dephosphorylates AdoCbl-P in vitro was previously reported, but the physiological significance of this activity was not explored (35).
This new model explains all the data reported in this paper. First, a cobC strain cannot salvage Cby because it makes AdoCbl-P, which cannot be dephosphorylated in the absence of CobC; hence, AdoCbl-P cannot be used by Cbl-dependent methionine synthase. Cby salvaging is restored in a cobC strain when either L-Thr-P or DMB is added to the medium or the level of cobinamide-P synthase (CbiB) enzyme is increased. We propose that Cby salvaging is restored under these conditions because the amount of AdoCbl-5Ј-P made by CobS is higher and the putative nonspecific phosphatase that dephosphorylates AdoCbl-5Ј-P requires higher levels of it before it can be used as substrate. There is precedent for such a scenario in AdoCbl biosynthesis in serovar Typhimurium (54). CobS can synthesize more AdoCbl-5Ј-P when it is saturated with either AdoCbi-GDP (CbiB, L-Thr-P effect) or ␣-RP (DMB effect). Steps for the synthesis of AdoCbi-GDP and ␣-RP remain as in Fig. 1.

Other important conclusions. (i) CobC does not phosphorylate L-Thr or AP.
Even though the inability of a cobC strain to salvage Cby is corrected when L-Thr-P is present in the medium, one should not conclude that the CobC protein has L-Thr kinase activity. If a cobC strain were truly blocked in L-Thr phosphorylation, AdoCbl synthesis would be correctible only by the addition of L-Thr-P. The suggestion that CobC phosphorylates L-Thr is further weakened by the correction of the Cby phenotype by the addition of DMB or the increase in the level of AdoCbi-P synthase (CbiB) enzyme. Instead, we propose that all the above conditions that restore Cby salvaging in cobC strains do so by increasing the level of AdoCbi-GDP or the level of ␣-RP, with the concomitant increase in AdoCbl-5Ј-P synthesized by CobS. The identity of the L-Thr kinase enzyme involved in AdoCbl synthesis remains unknown.
(ii) On the previously reported growth behavior of cobC strains. Previous studies of CobC by our laboratory showed that under highly aerobic growth conditions, a serovar Typhimurium cobC strain required DMB to grow, but under anaerobic conditions the cobC strain grew well without the addition of DMB (40). While work presented here provides plausible explanations for the effect of DMB on the aerobic growth of cobC strains, it does not provide insights into why DMB supplementation is not needed during anaerobic growth. We hypothesize that since DMB is not synthesized by serovar Typhimurium under anaerobic conditions (33), the incorporation of adenine or methyl-adenine into B 12 may be more efficient, hence eliminating the need for exogenous DMB.
(iii) In wild-type serovar Typhimurium, the conversion of AdoCby to AdoCbi-P by the CbiB enzyme is limited under aerobic conditions. This is not unexpected, given that under aerobic growth conditions, expression of the cbi genes is low but not off (23). The low level of CbiB made by serovar Typhimurium under aerobic conditions is clearly sufficient to salvage Cby as long as the CobC enzyme is functional, probably reflecting on the differences in affinity of CobC and the putative nonspecific phosphatase acting on AdoCbl-P. Efforts to identify the putative nonspecific AdoCbl-P phosphatase are ongoing.