Vitamin B12 (cobalamin) is a cobalt-containing modified tetrapyrrole that is an essential nutrient for higher animals. Its biosynthesis is restricted to certain bacteria and requires approximately 30 enzymatic steps for its complete de novo construction. Remarkably, two distinct biosynthetic pathways exist, which are termed the aerobic and anaerobic routes. The anaerobic pathway has yet to be fully characterized due to the inherent instability of its oxygen-sensitive intermediates. Bacillus megaterium, a bacterium previously used for the commercial production of cobalamin, has a complete anaerobic pathway and this organism is now being used to investigate the anaerobic B12 pathway through the application of recent advances in recombinant protein production. The present paper provides a summary of recent findings in the anaerobic pathway and future perspectives.
Of all the coenzymes, cofactors and prosthetic groups found in Nature, there is something beguiling and enigmatic about cobalamin, the biological form of vitamin B12. This may, in part, reflect the rich scientific history associated with its discovery, function and structure determination. For instance, Minot, Murphy and Whipple were awarded the Nobel Prize in Physiology or Medicine in 1934 for demonstrating that cobalamin was the anti-pernicious anaemia factor [1,2], a feat matched by Dorothy Hodgkin with her Nobel Prize in Chemistry in 1964 for using X-ray crystallography to determine its structure , whereas Woodward and Eschenmoser employed over 100 researchers in their quest to achieve the total chemical synthesis of adenosylcobalamin in 1973 [4,5]. The mystique surrounding cobalamin also revolves around a number of unique biological features including its structural complexity (it is chemically the most sophisticated of all cofactors and coenzymes), and the fact that it is made exclusively by certain prokaryotes. The biosynthesis of adenosylcobalamin requires approximately 30 enzyme-mediated steps proceeding via a number of highly labile, oxygen-sensitive, intermediates, making the synthesis challenging for eukaryotic cells . Yet the catalytic efficiency of cobalamin-dependent enzymes acts as a powerful evolutionary pressure to maintain their use within central metabolism and thus, although some systems, most notably higher plants, have opted out of the cobalamin world, others such as vertebrates have remained addicted to this micronutrient . This present review focuses on how some bacteria are able to synthesize adenosylcobalamin.
Cobalamin is a member of the modified tetrapyrrole family, whose other affiliates include molecules such as chlorophyll, haem, sirohaem and coenzyme F430. It is composed of a ring-contracted macrocycle into which is co-ordinated a Co2+ ion, where the reduced ring size allows for tighter metal co-ordination (Figure 1A). The framework is extensively decorated (methylations and amidations) and is termed a corrin ring. In cobalamin, the propionate side chain on ring D of the corrin ring is attached to a nucleotide loop that extends under the molecule. This nucleotide contains an unusual base called dimethylbenzimidazole, which provides a lower ligand for the central Co2+ ion. The upper ligand for the metal is formed through a metal–carbon bond with adenosine in adenosylcobalamin (coenzyme form) or with a methyl group in methylcobalamin (cofactor form). It is the unique properties of the cobalt–carbon bond that provide the catalytic power associated with the major cobalamin-dependent enzymes, promoting radical formation in coenzyme-mediated rearrangements and allowing for enhanced methyl transfer in cofactor-reliant methylations. In the present review, we describe how the corrin ring is assembled de novo.
Summary of the aerobic and anaerobic pathways
As with all modified tetrapyrroles, cobalamin is synthesized from uro'gen III (uroporphyrinogen III). Corrin ring synthesis then involves the addition of eight methyl groups, six amidations, cobalt insertion, decarboxylation and ring contraction, all of which generate cobyric acid. Remarkably, bacteria have evolved two related, although genetically distinct, pathways, which are referred to as the aerobic and anaerobic routes (Figure 1B). The major differences relate to the requirement for molecular oxygen and the timing of cobalt insertion. The better characterized aerobic pathway will be discussed first before a more detailed analysis of some recent advances of the anaerobic route. However, before we get to the detail, a quick note on nomenclature is required. The intermediates of the corrin pathway are generally referred to as precorrin-n, where n refers to the number of methyl groups that have been added to the periphery of the tetrapyrrole macrocycle. The oxidized version of these compounds are called factors, where factor II would be the oxidized form of precorrin-2 .
The aerobic pathway
The aerobic biosynthesis of the corrin macrocycle was solved through a collective effort involving a research group at Rhône-Poulenc (now Sanofi-Aventis), and the academic-based teams of Battersby at Cambridge and Scott at Texas A&M [6,9,10]. In brief, the aerobic cobalamin (cob) biosynthetic genes were identified and isolated from an industrial vitamin B12-producing strain of Pseudomonas denitrificans SC510 [11–13]. Then, using a range of biochemical techniques, the encoded Cob enzymes were characterized, each of the pathway intermediates were isolated, and the structure of the corresponding methyl esters were determined by NMR. Of particular interest is the process of ring contraction, which requires molecular oxygen. Here a mono-oxygenase (CobG) forms a hydroxylated γ-lactone intermediate (precorrin-3B) that undergoes a masked pinacol rearrangement during the contraction event in a reaction catalysed by CobJ and which extrudes the methylated C20 position (Figure 1C). This C2 fragment is subsequently lost as acetic acid. Cobalt is inserted only after the ring has been fully contracted (Figure 1D), at a comparatively late stage of the synthesis, using an ATP-dependent multi-enzyme complex (CobNST) that is similar to the magnesium chelatase of chlorophyll biosynthesis [14,15]. The requirement for molecular oxygen for corrin ring synthesis came as a surprise, not least as many of the precorrin intermediates are highly sensitive to oxygen. Moreover, it was also realized that many anaerobes were able to synthesize cobalamin and hence it was quickly appreciated that an alternative anaerobic pathway must also exist .
The anaerobic pathway
Indeed, isolation and sequencing of the cobalamin biosynthetic genes from anaerobic bacteria such as Propionibacterium shermanii (also known Propionibacterium freudenreichii), Salmonella enterica and Bacillus megaterium quickly confirmed this suspicion. These organisms were found to contain a highly similar, but genetically distinct, pathway whose individual enzymes were prefixed Cbi [for cobinamide (cbi) biosynthesis] [17,18]. A number of the genes associated with the anaerobic pathway are clearly similar in sequence to those of the aerobic pathway (~20–40% identical). However, a number of key differences were noted, such as the lack of a mono-oxygenase and a very different cobaltochelatase. These differences are discussed in context through the present mini-review, while an illustration of the predicted enzymatic steps in the anaerobic pathway is shown in Figure 2.
The anaerobic pathway
For the anaerobic biosynthesis of cobalamin, the first step is catalysed by the SUMT (S-adenosyl-L-methionine uroporphyrinogen III methyltransferase) enzyme, which methylates at C2 and C7 to convert uro'gen III to precorrin-2, using SAM (S-adenosyl-L-methionine) as a methyl donor. Structures of this enzyme in the presence and absence of uro'gen III and S-adenosyl-L-homocysteine have been determined, providing molecular detail on the mode of substrate binding within the class III methyltransferases. This class is largely associated with the peripheral methylation of the corrin framework , revealing some insight into the evolution of the pathway itself . Next, it would appear that precorrin-2 is oxidized to factor II by a precorrin-2 dehydrogenase (SirC) in an NAD+-dependent fashion. Factor II is preferred to precorrin-2 for metal insertion, probably due to the planar shape of the former. The chelatase that inserts cobalt into the isobacteriochlorin ring is either CbiX or CbiK, both of which share a common topology with the ferrochelatase associated with haem synthesis. In some organisms, CbiX is found as a very small protein, consisting of only approximately 100 amino acids. This smaller version of CbiX, termed CbiXS, aligns with both the N- and C-termini of CbiX and CbiK. It would thus appear that both CbiX and CbiK have arisen from a gene duplication and fusion event of a smaller protein .
High-resolution X-ray structures of CbiXS and CbiK in complex with their product, cobalt–factor II, have recently been characterized in detail, allowing the identification of the tetrapyrrole- and cobalt-binding sites  and provide a molecular overview of the mechanism of chelation. In this way, it is envisaged that factor II is bound within the active site in a distorted conformation, with the central cavity of the tetrapyrrole macrocycle directly below the cobalt-binding site. As the metal is inserted, two active-site histidine residues are able to facilitate the removal of two protons from the pyrrole nitrogens. The insertion of the metal induces greater planarity within the tetrapyrrole, encouraging it to be released from the active site.
After cobalt insertion into factor II by the cobaltochelatase, a SAM-dependent methylation at C20 is catalysed by CbiL to produce cobalt(II)–factor III. Despite reasonable similarity to CobI, the orthologous aerobic pathway enzyme CbiL does not methylate precorrin-2, providing further evidence of the genetic distinction between the aerobic and anaerobic pathways . CbiL has been studied from the thermophilic archaeon Methanobacterium thermoautotrophicus and has been shown to prefer the substrate cobalt–factor II to cobalt–precorrin-2, suggesting that the factor intermediates are favoured during the early stages of the anaerobic pathway . The enzyme cannot discriminate between metal oxidation states as it will methylate both cobalt(II)– and cobalt(III)–factor II. However, EPR data suggest that the enzymes provides a ligand for the metal-containing substrate, with a five-co-ordinate low-spin species observed . This would explain why CbiL only recognizes a metal-containing pathway intermediate.
The next step of the pathway is the ring-contraction event, catalysed by CbiH. This reaction involves a SAM-dependent methylation at C17 and mediates extrusion of the methylated C20 position between rings A–D, resulting in the formation of a δ-lactone ring. Unfortunately, little is known about the mechanism of the reaction or the oxidation state of the Co2+ ion. The anaerobic ring-contraction process has intrigued researchers for many years and still represents the major limitation in our understanding of the pathway. Only small quantities of the ring-contracted product, cobalt–precorrin-4, have ever been isolated. Moreover, cobalt–precorrin-4 is unstable and rapidly oxidizes to cobalt–factor IV, which is the form of the intermediate that has been analysed by NMR . It also remains unclear whether the ring-contraction enzyme, CbiH, can discriminate between cobalt–factor III or cobalt–precorrin-3, before its transformation into the product. However, at some point, the factor intermediates have to be converted back into the level of a hexahydroporphyrin and this is most likely to occur around the point of ring contraction. The formation of cobalt–precorrin-4 (and cobalt–factor IV) has only been reported in low yield and the inability to make large quantities of this intermediate has hindered the full elucidation of the pathway.
From cobalt–precorrin-4, the next step involves methylation at C11 in a reaction catalysed by CbiF to produce cobalt–precorrin-5A. This reaction was investigated by the use of an Escherichia coli lysate with the S. enterica CbiF overproduced . Unfortunately, the intermediate was extracted as an octamethylester, whereby the surrounding carboxylic acid side chains were esterified in methanol and concentrated sulfuric acid (95:5, v/v). The advantage of using esters is they are more stable and can be easily extracted into organic solvent. The problem, however, is that this method does not provide any information about the oxidation state of the Co2+ ion or macrocycle. It has also been suggested that the S. enterica CbiG may catalyse the opening of the δ-lactone ring to produce cobalt–precorrin-5B  and to release the extruded C2 fragment as acetaldehyde . CbiG is unique to the anaerobic pathway, with only weak similarity (in a short C-terminal sequence) to the uncharacterized CobE protein from the aerobic pathway. Although CobE is essential to vitamin B12 biosynthesis in P. denitrificans, no role has yet been assigned to this protein . The biosynthesis of the cobalt–precorrin-5 intermediates has only been achieved with crude cell extracts of recombinant bacteria containing elevated levels of the recombinant enzymes and therefore lacks the absolute certainty that is achieved when purified enzymes are used.
Another unique protein associated with the anaerobic pathway is CbiD. This protein was predicted to catalyse the C1 methylation of cobalt–precorrin-5B to cobalt–precorrin-6A . A crystal structure of the Archaeoglobus fulgidus CbiD has been released, although not published (PDB code 1SR8). The sequence and structure of CbiD bears no similarity to any other known enzyme. Normally, methyltransferases possess a SAM-binding domain that has a conserved GXGXG motif found at the N-terminus, but this trademark is absent from CbiD and is quite distinct from the other methyltransferases associated with cobalamin biosynthesis. Evidence that CbiD acts as the C1 methyltransferase comes from the observation that omission of CbiD from a recombinant strain producing all the enzymes for cobyrinic acid synthesis results in the synthesis of an analogue, C1-desmethyl-cobyrinic acid, that is lacking the C1 methyl group . However, direct evidence for the role of CbiD to act as the C1 methyltransferase has yet to be attained.
From cobalt–precorrin-5 to cobyrinic acid, the intermediates and reactions have yet to be characterized. Nonetheless, the enzymes CbiJ, CbiE, CbiT and CbiC share reasonable homology with enzymes from the aerobic pathway that have been characterized and shown to convert precorrin-6 into HBA (hydrogenobyrinic acid). In this respect, the transformation of cobalt–precorrin-6A into cobyrinic acid remains poorly resolved. Nevertheless, some research has been carried out on the subsequent amidation steps that oversee the amidation of the carboxylic acid side chains. Here, it has been shown that CbiA initially catalyses the amidation of the acetic acid side chains on rings A and B, whereas CbiP performs the remaining four amidation reactions [30,31].
Redox proteins in the anaerobic cobalamin pathway of B. megaterium
B. megaterium utilizes the anaerobic pathway and is a past industrial producer of vitamin B12 . Ironically, the organism can produce vitamin B12 when grown under either anaerobic or aerobic conditions, a feature that is not possible for other obligate or facultative anaerobes. Its anaerobic B12 repertoire is located within a 14 gene cobI operon [18,33]. The first three genes (cbiW, cbiH60 and cbiX) encoded by the cobI operon possess some unusual features for an anaerobic pathway. As described previously, the B. megaterium CbiX is a highly active cobaltochelatase and distinctively contains a single 4Fe–4S centre co-ordinated between a homodimer . The function of this redox centre remains unknown, but it is considered non-essential for activity . Interestingly, a small (~18 kDa) redox protein, CbiW (a putative thioredoxin), remains uncharacterized and is likely to co-ordinate another Fe–S cluster, but nonetheless its characterization has remained elusive owing to the inherent insolubility of the protein when produced heterologously in E. coli. Finally, the ring-contraction enzyme, CbiH60, is the most peculiar of these three proteins. Typical CbiH enzymes are 30 kDa in length, but the B. megaterium CbiH is double this size and is referred to as CbiH60. The extra length of CbiH60 is due to a C-terminal extension that is exclusive to Bacillus species. This C-terminus displays some low sequence similarity to the nitrite/sulfite reductase (NiR/SiR) family (NirA, NirB) and also has some similarity with the hydrogenase protein HydA. Both classes of protein share a 4Fe–4S cluster formed through four cysteine ligands. Sequence similarity between CbiH60 NiR/SiR-like domain, NirA, NirB, HydA and HyaA is centred on these cysteine ligands (Figure 3). Another distinct CbiH is also found in a number of archaeal species, such as M. thermoautotrophicus. Here, the CbiH protein has a shorter C-terminal extension that co-ordinates two 4Fe–4S centres . The role of these redox centres in cobalamin biosynthesis remains unknown; however, they could act as part of a redox system to ensure that the pathway intermediates are maintained at the correct oxidation state or alternatively be involved in some elaborate regulatory mechanism in the control of vitamin B12 biosynthesis.
CbiH60 and sequence similarity with NirA, NirB, HydA and HyaA
Conclusions and future perspectives
The anaerobic biosynthesis of vitamin B12 remains a major challenge in the elucidation of key biosynthetic metabolic pathways. However, the redox proteins associated with B. megaterium adds further interest and intrigue as attempts are made to finally expose the step-by-step synthesis of this remarkable molecule. This is likely to be hastened by recent advances in recombinant protein production in B. megaterium [36–38], where homologous protein production appears to give rise to larger quantities of stable cobalamin biosynthetic enzymes (S.J. Moore, R. Biedendieck and M.J. Warren, unpublished work). It is hoped that this advantage will allow a more detailed characterization of the individual steps of the anaerobic cobalamin pathway and allow the complete ex-vivo synthesis of cobyrinic acid using purified enzymes.
Frontiers in Biological Catalysis: Biochemical Society Annual Symposium No. 79 held at Robinson College, Cambridge, U.K., 10–12 January 2012. Organized and Edited by David Leys (Manchester, U.K.), Andrew Munro (Manchester, U.K.), Emma Raven (Leicester, U.K.) and Martin Warren (University of Kent, U.K.).
S.J.M. is supported by a Biotechnology and Biological Sciences Research Council (BBSRC) studentship.