A vitamin B12 transporter in Mycobacterium tuberculosis

Vitamin B12-dependent enzymes function in core biochemical pathways in Mycobacterium tuberculosis, an obligate pathogen whose metabolism in vivo is poorly understood. Although M. tuberculosis can access vitamin B12 in vitro, it is uncertain whether the organism is able to scavenge B12 during host infection. This question is crucial to predictions of metabolic function, but its resolution is complicated by the absence in the M. tuberculosis genome of a direct homologue of BtuFCD, the only bacterial B12 transport system described to date. We applied genome-wide transposon mutagenesis to identify M. tuberculosis mutants defective in their ability to use exogenous B12. A small proportion of these mapped to Rv1314c, identifying the putative PduO-type ATP : co(I)rrinoid adenosyltransferase as essential for B12 assimilation. Most notably, however, insertions in Rv1819c dominated the mutant pool, revealing an unexpected function in B12 acquisition for an ATP-binding cassette (ABC)-type protein previously investigated as the mycobacterial BacA homologue. Moreover, targeted deletion of Rv1819c eliminated the ability of M. tuberculosis to transport B12 and related corrinoids in vitro. Our results establish an alternative to the canonical BtuCD-type system for B12 uptake in M. tuberculosis, and elucidate a role in B12 metabolism for an ABC protein implicated in chronic mycobacterial infection.


Summary
Vitamin B 12 -dependent enzymes function in core biochemical pathways in Mycobacterium tuberculosis, an obligate pathogen whose metabolism in vivo is poorly understood. Although M. tuberculosis can access vitamin B 12 in vitro, it is uncertain whether the organism is able to scavenge B 12 during host infection. This question is crucial to predictions of metabolic function, but its resolution is complicated by the absence in the M. tuberculosis genome of a direct homologue of BtuFCD, the only bacterial B 12 transport system described to date. We applied genome-wide transposon mutagenesis to identify M. tuberculosis mutants defective in their ability to use exogenous B 12 . A small proportion of these mapped to Rv1314c, identifying the putative PduO-type ATP : co(I)rrinoid adenosyltransferase as essential for B 12 assimilation. Most notably, however, insertions in Rv1819c dominated the mutant pool, revealing an unexpected function in B 12 acquisition for an ATP-binding cassette (ABC)-type protein previously investigated as the mycobacterial BacA homologue. Moreover, targeted deletion of Rv1819c eliminated the ability of M. tuberculosis to transport B 12 and related corrinoids in vitro. Our results establish an alternative to the canonical BtuCD-type system for B 12 uptake in M. tuberculosis, and elucidate a role in B 12 metabolism for an ABC protein implicated in chronic mycobacterial infection.

Introduction
The genome of Mycobacterium tuberculosis, obligate human pathogen and causative agent of tuberculosis, encodes three B 12 -dependent enzymes. Previous work in our laboratory has established that both the methylmalonyl-coenzyme A (CoA) mutase, MutAB [1], and the metH-encoded methionine synthase [2] are functional, and require B 12 for activity. Mycobacterium tuberculosis also possesses a predicted pathway for B 12 biosynthesis [3], but appears not to produce the cofactor in vitro [1,2] or in macrophages [4]. Nevertheless, the bacillus can use exogenous vitamin B 12 and encodes a B 12 -responsive riboswitch that suppresses transcription of the alternative, B 12 -independent methionine synthase, metE, in B 12 -replete conditions [2]. These observations imply a role for the cofactor & 2013 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution in M. tuberculosis pathogenesis. However, it is uncertain whether B 12 is available during infection, and which mycobacterial genes are required for its uptake and assimilation.
Vitamin B 12 and B 12 derivatives are members of the cobalamin group of corrinoid macrocycles [5]. Cobalamins are structurally complex, comprising a defining tetrapyrrole framework with a centrally chelated cobalt ion held in place by a lower axial base, dimethylbenzimidazole and an upper ligand that determines the cofactor form (figure 1). The cyano group in vitamin B 12 (cyanocobalamin, CNCbl) must be replaced by deoxyadenosine and methyl ligands, respectively, during conversion to the biologically active cofactors: adenosylcobalamin (AdoCbl or coenzyme B 12 ), which is required by methylmalonyl-CoA mutase, and methylcobalamin (MeCbl), which serves as an intermediary in the synthesis of methionine from homocysteine and methyltetrahydrofolate [6]. The reactivity of B 12 cofactors derives from the cobalt-coordinated organic ligands [7] and, together with the size of the cobalamin core, underlies the need for multi-component systems to mediate controlled translocation and delivery of B 12 across the cell membrane to its target enzyme [8].
Although bioinformatic analyses have predicted alternative vitamin transporters [9], BtuCD-BtuF remains the only confirmed bacterial B 12 transport system identified to date [10]. The Escherichia coli model is the best characterized: a high-affinity corrinoid transporter, BtuB, operates with the TonB-ExbBD complex to traffic B 12 across the outer membrane into the periplasm [11] where it is captured by the btuF-encoded B 12 -binding protein and delivered to the ATPbinding cassette (ABC) importer, BtuCD, which spans the cytoplasmic membrane [12]. Mycobacterium tuberculosis is characterized by a notoriously complex cell envelope comprising a cytoplasmic membrane and an external cell wall [13]. However, despite its demonstrated ability to use exogenous B 12 [2,4], the proteins involved in mycobacterial B 12 transport and assimilation are unknown: M. tuberculosis is included in the small number of B 12 -using bacteria that lack a candidate BtuFCD-type B 12 transport system [3,9,14] as well as an identifiable homologue of TonB [15].
In this study, we used random mutagenesis to identify genes whose disruption abrogated the ability of M. tuberculosis to use exogenous vitamin B 12 in vitro. Our results establish an essential role in B 12 uptake for Rv1819c, a predicted ABC protein implicated in chronic infection in vivo [16], thereby revealing an alternative to the well-characterized BtuCD system for B 12 transport.

Construction of transposon mutant library
A library of transposon (Tn) mutants was constructed in M. tuberculosis H37Rv DmetH, using the MycoMarT7 phage as described [17]. For the primary screen, transductants were plated across multiple 7H10 plates containing 20 mg ml 21 kan and 10 mg ml 21 CNCbl at a density of 20 000 colony forming units (CFU) per plate. The secondary screen was performed in duplicate in microtitre plate format and, for each Tn mutant, comprised four parallel wells containing 0.1 per cent propionate plus 20 mg ml 21 kan as base medium in each well: the first well constituted a growth control and contained only the base medium; in well 2, 10 mg ml 21 CNCbl was added to the base medium; in well 3, the base medium was supplemented with 0.1 mM 3NP; and in well 4, 0.1 mM 3NP and 10 mg ml 21 CNCbl were added.

Identification of transposon insertion sites
A combination of Tn-linker [18] and rescue cloning [19] strategies was applied to identify Tn insertion sites using the oligonucleotides in the electronic supplementary material, table S1.

Construction of mutant strains of Mycobacterium tuberculosis
Mycobacterium tuberculosis mutants were constructed using suicide plasmids described in electronic supplementary material, table S1. Genetic complementation used tweety-based vectors [20]. rsob.royalsocietypublishing.org Open Biol 3: 120175 3.5. DNA sequencing Mycobacterium tuberculosis genomic DNA was sequenced using an Illumina GenomeAnalyzer II, as described previously [21].

Homology modelling
The initial detection of crystal structures related to Rv1819c was performed using HHsearch [22] and COMA [23]. The Rv1819c model was then generated using a previously described iterative approach [24,25]. Briefly, both the set of structural templates and corresponding alignments were refined until the resulting model stopped improving and the visual inspection revealed no significant flaws.

A forward genetic screen identifies B 12 uptake mutants
We showed previously that deletion of the B 12 -dependent methionine synthase, MetH, renders M. tuberculosis sensitive to vitamin B 12 during growth on solid medium [2]. This phenotype depends on the function of a B 12 riboswitch that is located immediately upstream of metE, the gene encoding an alternative, B 12 -independent methionine synthase in M. tuberculosis. In wild-type M. tuberculosis, exogenous B 12 suppresses transcription of metE by binding to the riboswitch [2], possibly ensuring efficient B 12 -dependent methionine synthesis by MetH. In the metH deletion mutant, however, riboswitch-mediated suppression of metE in response to B 12 effectively results in the complete shutdown of methionine synthase activity, thereby eliminating production of an essential amino acid and so inhibiting bacillary growth [2]. This effect is most profoundly manifest on solid medium, where exposure to 10 mg ml 21 CNCbl results in a 3log 10fold reduction in viable CFU of DmetH knockout mutants [2]. Here, we exploited the observed B 12 sensitivity of metH mutants in a genetic screen designed to elucidate a potential B 12 transport system in M. tuberculosis (figure 2). To this end, we constructed an unmarked metH knockout of the laboratory strain, M. tuberculosis H37RvJO [21] (electronic supplementary material, figure S1a) and confirmed that it phenocopied the previously described hygromycin (hyg)marked DmetH (BB) deletion mutant [2] during growth on B 12 -containing solid medium (see the electronic supplementary material, figure S1b). The unmarked DmetH knockout was used as background strain in which to generate a Tn mutant library using the MycoMarT7 phage [17] that carries a kan resistance marker and inserts randomly at TA dinucleotides [19]. In the primary screen, the library of insertion mutants was plated on solid medium containing kan and CNCbl to enable the identification of genes whose disruption alleviated the growth defect of the metH mutant (figure 2a). In total, 612 individual clones were isolated, each of which was picked and regrown in standard liquid medium; of these, 35 grew poorly or not at all and were eliminated, leaving 577 'B 12 -resistant' insertion mutants for further analysis. Previously, in characterizing the DmetH (BB) mutant, we noted the high frequency at which suppressor mutants arose spontaneously on B 12 -containing solid medium, with singlenucleotide polymorphisms (SNPs) in the B 12 riboswitch located upstream of metE accounting for approximately 10-20 per cent of these [2]. In the current screen, we used dual selection on kan and CNCbl in order to limit the potentially confounding effects of spontaneous riboswitch mutations: according to these criteria, growth on CNCbl plus kan would require successful transduction with the kan-resistant Tn as well as disruption- rsob.royalsocietypublishing.org Open Biol 3: 120175 spontaneous or Tn-mediated-of B 12 -dependent growth inhibition. Nevertheless, we predicted that a significant proportion of B 12 -resistant mutants might contain Tn insertions in the riboswitch motif. So, in order to minimize the impact of disruptions to the B 12 riboswitch, we applied a secondary screen (figure 2a) to determine the capacity of the insertion mutants to assimilate exogenous CNCbl for growth in liquid medium containing propionate in the presence of 3NP, an inhibitor of the key methylcitrate cycle enzyme, isocitrate lyase [26]. Two prior observations informed the design of this screen: (i) the inhibitory effect of genetic (DprpDC) or chemical (3NP) abrogation of methylcitrate cycle enzymes during growth in liquid medium containing propionate can be alleviated by supplementing the culture with CNCbl, thereby enabling M. tuberculosis to use propionate as a carbon source via the methylmalonyl pathway that includes the B 12 -dependent methylmalonyl-CoA mutase, MutAB [1]; (ii) for reasons that are not clear, B 12 -mediated growth inhibition is less effective in liquid versus solid medium-that is, the DmetH mutant can grow in B 12 -supplemented liquid medium (see the electronic supplementary material, figure S1c). The secondary screen therefore assessed the ability of all 577 Tn mutants to use exogenous CNCbl for growth in liquid medium containing propionate in the presence of 3NP (figure 2a). The majority of Tn mutants (n ¼ 493) phenocopied the parental DmetH strain in this assay, and were eliminated as candidate B 12 uptake mutants. In contrast, the remaining 84 Tn mutants were unable to grow in well 4, suggesting impaired ability to use exogenous B 12 for methylmalonyl pathwaydependent propionate catabolism. To verify these results, 43 of the 84 mutants were selected at random for phenotypic confirmation of disrupted B 12 uptake in batch culture (data not shown) and on B 12 -containing solid medium (see the electronic supplementary material, figure S2a). ). This result strongly suggested a role in B 12 uptake for a predicted ABC transport protein previously identified as the putative M. tuberculosis homologue of BacA [16,27], a protein of cryptic function implicated in chronic infection in multiple host-pathogen models [25].  3). By contrast, the complemented derivative carrying a full-length copy of Rv1819c at the attB site, referred to as DbacA::pKLMt5 in the original study [16], was able to use B 12 for growth (see the electronic supplementary material, figure S3a). Similarly, integration of full-length Rv1819c at attB restored the B 12 -sensitive phenotype of a randomly selected DmetH Rv1819c::Tn mutant during growth on solid medium supplemented with CNCbl (see the electronic supplementary material, figure S3b), and reversed the inability of the same mutant to use B 12 for growth in propionate-containing liquid medium supplemented with 3NP ( figure 3), confirming the essentiality of Rv1819c in this assay. It was noticeable in the propionate utilization experiment (see the electronic supplementary material, figure S3a) that the DbacA::hyg mutant started to replicate after two to three weeks of apparent growth arrest, possibly indicating the emergence of suppressor mutants. To circumvent this complication, we deleted the prpDC locus [28] in this strain, thereby negating the need to use 3NP to eliminate methylcitrate pathway function [1]. In contrast to the single prpDC deletion mutant, the double DbacA::hyg DprpDC knockout exhibited no growth at all in propionate over the 28-day time course (see the electronic supplementary material, figure S3c), even when supplemented with CNCbl, strongly suggesting that Rv1819c is required for the assimilation of exogenous B 12 to enable methylmalonyl-CoA pathway function.  We reported previously that SNPs in the metE-associated B 12 riboswitch accounted for 10-20 per cent of all B 12 -resistant mutants isolated after plating the DmetH (BB) knockout on medium containing CNCbl, whereas the remaining B 12 -resistance mutations were unknown [2]. To investigate the possibility that mutations in Rv1819c might account for B 12 resistance in those clones lacking riboswitch mutations, we plated the DmetH (BB) strain on medium containing CNCbl and sequenced the riboswitch region and Rv1819c locus in 10 spontaneous B 12 -resistant mutants. Consistent with previous results [2], two isolates carried independent mutations in the highly conserved B12-box motif within the metE riboswitch [29], namely C ! T transversions at positions 2155 and 2163 relative to the metE start codon, respectively. Notably, four other B 12 -resistant mutants had wild-type riboswitch sequences, but contained non-synonymous SNPs in Rv1819c  (see the electronic supplementary material, table S2), supporting the inferred role of Rv1819c in B 12 uptake. To eliminate the possibility that an additional, unidentified mutation (or mutations) might account for the observed phenotype, we sequenced the genome of a representative Rv1819c point mutant, SP09 (see the electronic supplementary material,  table S2). The parental, B 12 -sensitive strain, DmetH (BB), was differentiated from the laboratory strain, H37RvJO [21], only in the targeted deletion of metH sequence. Moreover, the Rv1819c mutation constituted the sole polymorphism separating SP09 from its DmetH (BB) parent and, importantly, complementation with wild-type Rv1819c at the attB locus restored B 12 sensitivity to both SP09 and SP18 (see the electronic supplementary material, figure S4).

Disruption of
In the primary Tn screen (figure 2a), 'B 12 -resistant' mutants had been selected on kan and CNCbl in order to limit the potentially confounding effects of spontaneous riboswitch mutations. To verify the utility of this approach, we analysed the insertion sites in a random selection of 20 of the 493 DmetH Tn mutants subsequently eliminated in the secondary screen owing to their inability to use exogenous B 12 for growth in propionate. All 20 mutants contained insertions in the B 12 riboswitch region directly upstream of metE (data not shown), confirming that disrupted riboswitch function represents a major mechanism for loss of B 12 regulation in strains which carry an intact Rv1819c gene.

Rv1819c is essential for corrinoid transport in Mycobacterium tuberculosis
Mycobacterium tuberculosis is predicted to encode a complete pathway for B 12 biosynthesis, including enzymes required for the conversion of the B 12 precursor, cobinamide, to AdoCbl through the addition of dimethylbenzimidazole and deoxyadenosine ligands [3]. The E. coli corrinoid transporter, BtuFCD, mediates uptake of cobinamide as well as CNCbl and AdoCbl [30], suggesting that Rv1819c might fulfil a corresponding role in M. tuberculosis. In support of this idea, cobinamide-provided as the dicyanide salt, (CN) 2 Cbi-was unable to complement the growth defect of DmetH Rv1819c::Tn mutants in propionate in the presence of 3NP, mimicking similar observations with AdoCbl and CNCbl ( figure 3). Insertions in Rv1819c also alleviated the growth inhibitory effect of AdoCbl, CNCbl and (CN) 2 Cbi on the metH knockout mutant on solid medium, a phenotype that was reversed upon complementation with wild-type Rv1819c (see the electronic supplementary material, figure S5). In combination, these results confirmed the essentiality of Rv1819c for corrinoid transport in M. tuberculosis.

Impaired vitamin B 12 uptake in spontaneous bleomycin-resistant Rv1819c mutants
Domenech et al. [16] showed that deletion of Rv1819c decreased the susceptibility of M. tuberculosis to the glycopeptide antibiotic, bleomycin, a phenotype commonly associated with BacA function [31][32][33]. We determined the minimum inhibitory concentration (

Rv1819c encodes an ATP-binding cassette-type transporter
Rv1819c was previously included in a group of 'BacA-related' proteins identified on the basis of their similarity to the highly conserved BacA and SbmA proteins of Sinorhizobium and E. coli, respectively [27]. Unlike BacA/SbmA orthologues, however, which are predicted to require an interaction with a separate cytoplasmic protein for function, Rv1819c encodes both transmembrane (TMD) and nucleotide-binding (NBD) domains of an ABC transport protein on a single polypeptide. Sequence similarity analyses using only the TMD located M. tuberculosis Rv1819c in a cluster distinct from BacA/SbmA (see the electronic supplementary material, figure S8a). Moreover, these analyses indicated that Rv1819c was more closely related to ABC proteins other than BacA/SbmA in both E. coli and Sinorhizobium, namely YddA [34] and ExsE [35], respectively. The Rv1819c NBD similarly identified YddA rsob.royalsocietypublishing.org Open Biol 3: 120175 and ExsE as close homologues in an equivalent similarity search (see the electronic supplementary material, figure S8b), together with the recently described human ABC-type B 12 transporter, ABCD4 [36].
We built a homology model of Rv1819c based on the crystal structures of two polyspecific ABC exporters, Staphylococcus aureus Sav1866 [37] and Salmonella typhimurium MsbA [38]. Consistent with known ABC protein architecture [39], Rv1819c is predicted to form a homodimer (figure 5), with each subunit comprising an N-terminal TMD fused to a highly conserved NBD that features all the motifs characteristic of functional ABC transporters (see the electronic supplementary material, figure S9a). Unlike Sav1866 and MsbA, though, the TMD domain of Rv1819c possesses an extra N-terminal region which is predicted to contain an additional transmembrane helix (see the electronic supplementary material, figure S9b). Proteomic analyses in the closely related M. bovis BCG suggest that this region is present in the mature protein [40], and therefore is not a signal peptide. However, in the absence of a close structural template containing seven transmembrane helices, we omitted the first 65 N-terminal residues in building the Rv1819c model. The predicted structure nevertheless provides a useful framework for the interpretation of experimental data. Notably, all three SNPs which resulted in substituted amino acids in the spontaneous B 12 -resistant and Bleo R mutants (see the electronic supplementary material, table S2) affect residues located in conserved regions of Rv1819c ( figure 5). While the structural consequences of the P349T and G411D mutations require further investigation, L442S affects a conserved position in the putative nucleotidebinding pocket formed by two interacting ABC domains. In Sav1866, the corresponding residue, Ile356, makes a van der Waals contact with the sugar moiety of the bound ADP [37], and so supports the inferred association between a distorted pocket and crippled protein function.

A PduO-type adenosyltransferase is required for assimilation of vitamin B 12
CNCbl must be adenosylated to generate the active cofactor, AdoCbl [41] (figure 6a). The genome of M. tuberculosis is predicted to encode both CobO (Rv2849c) and PduO (Rv1314c) ATP:co(I)rrinoid adenosyltransferases [3], non-homologous enzymes that catalyse this reaction in other bacteria [42,43]. It was notable, therefore, that six putative B 12 uptake mutants (figure 2) contained Tn insertions in pduO (see the electronic supplementary material, figure S2d), because this suggested that an impaired ability to convert exogenous CNCbl to the cofactor form could confer B 12 resistance in the primary screen, as well as eliminate the ability of M. tuberculosis to use B 12 for growth in propionate. To confirm the role of PduOdependent adenosylation in these phenotypes, we evaluated the abilities of the DmetH pduO::Tn mutants to assimilate different corrinoids for growth in propionate in the presence of 3NP (figure 6b). The mutants were unable to use either cyano form, CNCbl or (CN) 2 Cbi, both of which are adenosylated in the biosynthetic pathway to AdoCbl (figure 6a). In contrast, supplementation with AdoCbl itself restored growth in this assay (figure 6b), suggesting bypass of PduO function. In combination, these observations implicate PduO as sole active adenosyltransferase in M. tuberculosis during growth in vitro.

Discussion
Our results identify Rv1819c as sole corrinoid transporter in M. tuberculosis under standard in vitro conditions and, moreover, establish the capacity of the organism to scavenge rsob.royalsocietypublishing.org Open Biol 3: 120175 corrinoids. The association of Rv1819c with B 12 uptake is unexpected, particularly given previous studies suggesting that Rv1819c might function in ATP-dependent peptide transport [16,44]. Moreover, the properties enabling polyspecific translocation of compounds such as bleomycin, B 12 and antimicrobial peptides which lack obvious structural similarity remain unclear [45]. In Gram-positive organisms, ABC-mediated importers function together with a high-affinity substrate-binding protein (SBP) that is anchored to the extracytoplasmic membrane [46]. Although M. tuberculosis possesses in excess of 30 ABC transporters, as well as 15 putative SBPs [47], we failed to identify a candidate B 12 -binding protein, raising the possibility that Rv1819c-mediated B 12 import occurs in the absence of a specific SBP or that multiple proteins fulfil this role [48]. In most bacteria, the components of the ABC transporters involved in the uptake of ferric siderophores, haem and vitamin B 12 are closely related [49]. However, our screen identified Rv1819c as sole transport candidate, excluding the possibility that other ABC proteins might perform overlapping functions in mycobacterial B 12 transport, at least under in vitro conditions. Instead, in associating Rv1819c with B 12 uptake, our results add to the expanding complement of atypical mycobacterial nutrient acquisition systems. For example, a novel pathway was recently elucidated that enables the scavenging of haem [50]-a tetrapyrrole which, like B 12 , is derived from d-aminolaevulinic acid via a uroporphyrinogen III intermediate [5].
In that system, uptake is mediated by the combined activity of a haem-binding protein and the MmpL family members MmpL3 and MmpL11-predicted RND-type efflux pumps which have been associated with multiple cellular functions [51]. In addition to haem import, recent evidence suggests that MmpL3 fulfils an essential role in exporting trehalose monomycolate across the cell membrane for incorporation into cell wall mycolic acids [52,53], and it has also been implicated in the susceptibility of M. tuberculosis to diverse small molecules [54,55]. It is tempting, therefore, to consider the analogy with Rv1819c-itself a predicted export protein that has now been implicated in the uptake of antimicrobial peptides [16,44] and vitamin B 12 , and might also play a role in cell wall biogenesis [16]. Rv1819c has been extensively investigated as M. tuberculosis BacA [16,44]. Unlike BacA/SbmA orthologues, however, deletion of Rv1819c does not render M. tuberculosis hypersusceptible to other antimicrobial drugs and cell disrupting agents [16]. Moreover, our structural model of M. tuberculosis Rv1819c predicts an ABC transporter comprising both TMD and NBD within a single polypeptide. This distinguishes the mycobacterial protein from BacA proteins in Brucella and other intracellular pathogens [32] that contain the TMD only and, importantly, is supported by sequence analyses that situate Rv1819c in a separate cluster from the BacA subfamily even when based on TMD sequence alone. The M. tuberculosis protein also differs from BacA proteins in its potential role in pathogenesis. While the essentiality of the mycobacterial protein for the maintenance of chronic infection in vivo [16] is reminiscent of BacA-like phenotype, closer inspection of the comparative in vivo infection dynamics of different 'bacA' mutants suggests divergent function: for example, in contrast to the Brucella and Sinorhizobium deletion mutants [27,32], the M. tuberculosis Rv1819c knockout is not impaired in its ability to establish an infection [16]. It is tempting, therefore, to consider the virulence defect of the Rv1819c deletion mutant in the light of recent studies describing the accumulation during chronic infection of cholesterol-rich lipid bodies inside foamy macrophages and their infecting bacilli [56,57]. That is, Rv1819c might function to ensure adequate supply of host-derived corrinoids for the B 12 -dependent utilization of propionate derived from cholesterol catabolism, a possibility that requires further investigation.
Although designed to detect a putative vitamin B 12 transporter, our screen also established the essentiality of the PduO-type adenosyltransferase for the assimilation of exogenous corrinoids. The M. tuberculosis genome contains both pduO and cobO adenosyltransferases; therefore, the inferred inactivity of the alternative enzyme in vitro might indicate functional adaptation of CobO to de novo B 12 biosynthesis [3], or to specific environmental conditions, including anaerobiosis [41]. Intracellular trafficking of B 12 in humans requires the sequential activity of multiple proteins which fulfil dual roles as molecular chaperones and in the enzymatic modification of the cofactor [58]. For example, MMACHC catalyses the reductive decyanation of CNCbl [59] while mediating LMBD1-dependent [60] transfer from the lysosome into the cytoplasm. Recent evidence further suggests that this process is facilitated by the interaction of LMBD1 with ABCD4 [36]-an ABC transporter and homologue of Rv1819c (see the electronic supplementary material,  rsob.royalsocietypublishing.org Open Biol 3: 120175 figure S8). In a subsequent step, the ATP:corrinoid adenosyltransferase attaches the axial ligand and ensures delivery of the resulting AdoCbl cofactor across the mitochondrial membrane to its target enzyme, methylmalonyl-CoA mutase [61]. Given that Tn-mediated disruption of pduO alleviated the B 12 sensitivity of the metH mutant in the primary screen, it is tempting to speculate that PduO might function not merely in enzymatic conversion of exogenous corrinoids, but also in delivery of the active cofactor into the cytoplasm. Our current model for the translocation of B 12 across the mycobacterial cell wall into the cytoplasm therefore proposes the sequential activity of the ABC transporter, Rv1819c and the PduO-type adenosyltransferase ( figure 6a).
Our Tn screen also identified four low-frequency insertions associated with compromised B 12 uptake (figure 2b). Targeted sequencing of Rv1819c and the metE riboswitch in these strains excluded spontaneous mutations as the underlying cause of the observed B 12 phenotypes. A single mutant carried an insertion in rpfB, which encodes a resuscitation-promoting factor. To explore this result further, we retested DmetH rpfB::Tn in parallel with an rpfB deletion mutant of H37Rv, constructed previously [62]. Although the Tn mutant was not able to use exogenous CNCbl for growth in propionate-containing medium, the DrpfB knockout strain phenocopied wild-type H37Rv in this assay (data not shown), thereby excluding a role for RpfB in B 12 uptake. It is possible that rpfB::Tn possesses an additional, unidentified polymorphism, affecting B 12 assimilation; alternatively, polar effects on the downstream gene, ksgA, encoding dimethyladenosine transferase, might contribute to the observed phenotype [63], a possibility under investigation. Two additional Tn insertions mapped to mymA, encoding a putative flavin-dependent monooxygenase. The predicted role of MymA in the maintenance of cell wall ultrastructure [64] suggests that compromised B 12 uptake in these mutants might be non-specific; however, this requires further investigation, and is complicated by the fact that mymA is the first gene in a seven-gene operon [65]. We also isolated a mutA::Tn mutant, whose inability to use propionate for growth in B 12 -containing medium is consistent with impaired methylmalonyl-CoA mutase function. The basis for the B 12 resistance of this mutant in the primary screen is unclear, however, and probably also the result of an additional spontaneous mutation. The final Tn insertion mapped to Rv2927c, a gene which previous saturation mutagenesis studies have predicted as essential for growth of M. tuberculosis in vitro [66,67]. Although the function of Rv2927c is unknown, it has been proposed to operate as part of the cell division machinery [68]. It seems probable that, like the mymA::Tn mutants, the failure of Rv2927c::Tn to assimilate B 12 is non-specific. However, given the inferred requirement for PduO-dependent adenosylation in the assimilation of exogenous B 12 , the prediction that Rv2927c might function in de novo adenosine nucleotide biosynthesis [69] is intriguing, and the subject of current investigation.