Bacteria make vitamin B12 through one of the longest biosynthetic pathways in nature: roughly 30 enzymatic steps that convert aminolevulinic acid into adenosylcobalamin, coenzyme B12, and all their intermediate forms. Two distinct routes exist. One needs oxygen, one forbids it. Only certain bacteria and archaea run them, which is why essentially every B12 molecule in a human body, a multivitamin, or a fortified food began inside a prokaryotic cell. The pathway explains why B12 is the largest and most complex vitamin, and why the world's supply depends on fermentation tanks rather than soil or sunshine.
The coenzyme framework: a metal ion wrapped in a ring
Vitamin B12 is not a single molecule but a family named cobalamins. Every member centers on a corrin ring, a shrunk-down relative of heme in which four pyrrole-like rings connect by direct bonds and hold a single cobalt atom. This corrin construction is the pathway's organizing problem. Nothing so intricate assembles in one enzyme, so cells build it in strict sequence, and each intermediate passes down an assembly line. The defining feature is the lower axial ligand: a nucleotide loop that folds back and tucks its dimethylbenzimidazole base beneath the cobalt. Attaching that loop, adenosine triphosphate linkage included, is the final commitment step. The upper face carries the working group: an adenosyl radical in coenzyme B12, used by enzymes that rearrange carbon skeletons, or a methyl group in methylcobalamin, used in methylation chemistry. The difference between the two is just the upper ligand, yet cells route the same core through separate tailoring reactions. The cobalt's oxidation state also shifts during assembly. When cobalt sits at the reduced Co(I) state, it rivals the strongest reductants in biology; that reactivity is why the pathway must be performed in protected, enzyme-bound compartments rather than freely in the cytoplasm.
Two assembly lines: oxygen-requiring and oxygen-forbidden
The biosynthesis splits into two mutually exclusive routes. The aerobic pathway runs in organisms such as Rhodobacter and Pseudomonas and recruits oxygen for ring contractions that forge the corrin skeleton. The anaerobic route runs in Salmonella, Bacillus, and Propionibacterium; its enzymes are famously oxygen-sensitive, and its ring-contraction chemistry instead relies on reactions mediated by S-adenosylmethionine radicals and cobalt chelation timing. Why keep two versions? The aerobic route appeared later; the anaerobic route reflects the early Earth and stays viable only in low-oxygen niches, which is why gut anaerobes that make B12-related compounds fail to produce true cobalamin: they cannot sustain the oxidative chemistry the complete pathway requires. This bifurcation is the pathway's central fork. A cell's oxygen environment, not its species name, dictates which gene set is useful. The two routes converge only at adenosylcobinamide, a late intermediate that both lineages can modify into finished coenzyme forms. A plain-language version: the same final product can be reached by two different factory layouts, one that uses oxygen as a tool and one that must avoid it. Switching layouts mid-production is impossible, so the organism's habitat is written into its genome.
How intermediates travel between enzymes and why it matters
The pathway is not a chain of free-floating chemicals. Intermediates are channeled, delivered from one active site to the next by dedicated carrier proteins and by a remarkable B12-binding protein that helps insert cobalt at the right step. Two enzymes are so complex they contain their own internal tunnels and partner proteins; one, the cobamide synthase, is among the most elaborate proteins in bacteria, arranged as an internal scaffold that positions a moving substrate at successive modification stations. The engineering answer to why: each step generates a fragile or reactive species, and diffusion would waste it. The cost of failure appears just downstream. Hydrolysis of a single phosphate off a late intermediate yields a dead-end compound; one premature oxidation of cobalt locks the ring in a useless form. Because repair is not possible, the cell discards whole intermediates and remethylates fresh ones. This turnover is one reason industrial yields per liter depend on fine tuning of genes controlling the availability of aminolevulinic acid, cobalt, and the pathway's rate-limiting enzymes. The point below is often missed: the pathway is not limited by the final assembly steps but by how much starting material the cell can spare from other demands. That is why cobalt added carefully, but not excessively, be a real bottleneck regulator in fermentation design.
Why bacteria make B12: the chemical jobs the coenzyme does
Bacteria do not produce B12 for humans; they produce it because their own metabolism requires it. The coenzyme's dominant job is radical chemistry, the kind that lets a cell rearrange carbon bonds in ways ordinary acid-base chemistry cannot. Dehydratases and mutases use the adenosyl radical to shuffle hydroxyl or amino groups between carbons in amino acids and sugars. In some fermenting organisms, these reactions allow growth on glycerol or propanediol; in others, B12-dependent enzymes process the carbon skeleton of amino acid breakdown. The methylcobalamin form transfers methyl groups in methionine synthesis, and that same chemistry in methanogenic archaea is central to their energy metabolism. So B12 is a prosthetic group: the cell invests in building it because its enzymes need it. This is why the pathway is regulated, and why it is expensive: the reducing power consumed to put cobalt at the right oxidation state, and the large gene cluster, are worth only when the cell can turn a profit from the reactions B12 enables. The corollary for human supply: no crop plant requires B12 for its own growth, so no plant accumulates it. Every vegetarian and vegan food gets its B12 only from microbial contamination or supplementation.
Quantified context: how much, how fast, how supplied
Several numbers fix the scale. These bacteria are among the most generous producers of any vitamin. Propionibacterium strains used in industrial fermentation were selected for titers in the tens of milligrams per liter range, though published yields vary by strain and process design; the point is that L-cobalamin production is measured in milligrams, not micrograms, per liter. The human adult requirement of 2.4 micrograms per day (by food labeling convention) sets the supply question in perspective. The discrepancies matter: values in literature are stated as grams per liter in some fermentation trials and as milligrams per liter in others, and unit conversions between them are a common source of confusion. Dietary-recommendation thresholds for older adults come from the same generous intake target, roughly 2.4 micrograms daily, rather than from a derived deficiency threshold. Put differently: one liter of good fermentation broth can hold enough B12 to supply several thousand daily requirements; one kilogram of a leafy green will not. That is the entire reason fortified foods and supplements exist. The next section examines where the practical numbers turn ambiguous.
Edge cases and the strict reality of the narrow producer list
Most claims that bacteria make B12 should be treated skeptically at the strain level. Many gut organisms, including lactic acid bacteria and several Bacteroides, make related cobamides, but often not the AdoCbl form humans require; some cells secrete what they make, others do not, and the amount colonizing the colon is produced too far downstream to be absorbed by the ileum, where the intrinsic factor receptor operates. The consequence is direct. A food labeled "fermented" has not earned a B12 claim; the label should name the organism and, ideally, show a measured value. There are other caveats that professionals know well. The anaerobic route tolerates low oxygen but degrades under normal aeration, so fermentation must use carefully staged oxygen control and often two-phase processes. High cobalt is not automatically good; excessive metal is toxic to the bacterium itself, and regulatory limits on cobalt-based colorants and residues apply to finished products. For a consumer, none of this is visible; below are the red flags.
Fermented foods reported to supply B12: verify the strain and the measured label value, not the recipe. "B12-producing" gut microbiome claims: the ileum, not the colon, absorbs B12, so strains in the colon can matter less than marketing implies. Unspecified "cobalt-based" supplements: confirm the finished product is labeled as cyanocobalamin or methylcobalamin, not merely as containing cobalt. Bulk fermentation B12 sold without a titer: unit inconsistencies (grams vs. milligrams, as discussed above) make unverified labels unreliable. Dietary claims on plant foods: treat them as unverified unless the food is fortified and labeled.
Misconceptions and one counterintuitive detail
The most persistent myth is that bacteria make B12 inside the human gut and that this feeds you. It is a myth in two parts: most human-gut bacteria do not make genuine cobalamin, and for those that do, production in the colon occurs where almost no absorption happens. The second myth is that B12 is a plant vitamin; no plant makes it, and the trace amounts in some algae come from bacterial contamination and are largely inactive. The counterintuitive detail is that the B12 in your supplements has usually not been through a plant, an animal, or a soil mill. It comes from a fermenter. Some professional reviews also treat the aerobic route as the only "real" pathway. In fact, the oxygen-forbidden route has been engineered into gut-friendly organisms and into industrial strains precisely because it avoids the oxidation byproducts that complicate processing. The third misconception, widespread in eco-labels, is that "cobalt-free" formulations are safer. The metal is not contamination; it is the functional atom. When cobalt is absent, the pathway stalls at the last steps and yields incomplete cobamides without the human-active upper ligand. The label "cobalt-containing" is a feature, not a defect.
The commercial milestone for synthetic B12 needs care. Many timelines cite a historical claim of the first chemical synthesis during the 1970s. The date as a fixed fact is not securely established here, so treat it as traditional narrative rather than a verifiable timestamp. The operative fact is safer to state: researchers succeeded in chemically building a B12 molecule in the laboratory, an achievement so complex that total chemical synthesis never became the commercial route. Industrial production stayed microbial.
The time course of a fermentation run
| Time point | What happens in the fermenter |
|---|---|
| 0 hours | Inoculation; grown starter culture added to a cobalt-containing medium |
| First growth phase | Cells multiply, produce aminolevulinic acid-derived precursors; oxygen is supplied for aerobic routes |
| Transition | Culture switches to low-oxygen or non-growth-optimized conditions; corrin ring assembly deepens |
| Hours later | Nucleotide loop attached; cobinamide branches converted toward adenosylcobalamin |
| Late phase | Product accumulates largely inside cells; the pathway's final tailoring enzymes work |
| Harvest | Cells collected; extraction and purification release the cobalamin family |
| Purification | Chromatographic steps separate cyanocobalamin or methylcobalamin, with quality testing |
| Finished | Purity judged against compendial tests; small deviations fail the batch |
Exact duration, titer, and yield are proprietary and vary by strain, so the table gives sequence, not clock time. The operational rule holds across processes: the final tailoring step can only convert what upstream steps have produced, and when cobalt or aminolevulinic acid supply lags, the whole run's yield collapses. A batch that misses one stage cannot be rescued by adding product later.
Practical actions for anyone who evaluates B12 claims
A few concrete steps separate verified products from plausible ones. If ingested B12 and the 2.4-microgram daily label value (the US reference standard) are in conflict, trust the labeled measured amount, not the marketing. If a food is not fortified, demand a stated measured B12 value per serving, and reject a claim rounded up from a "produces B12" organism without a number, since the dosage from incidental trace contamination is unpredictable. For those who avoid animal products, choose a supplement or fortified product with a declared amount per serving, and revisit it when product or recipe changes. In manufacturing evaluation, tie yield to the process: verify which pathway the strain uses, confirm oxygen and cobalt control, and check the label since fermentation units should be stated consistently, not as unitless titers. If a product's claim cites a scientific citation, the exact study and measured values should be checkable; if not, the claim has no measurable basis.
The practical set ends at one threshold question: are you covered against the expected intake target? B12-related blood tests are for clinicians to interpret, not a self-triage tool; the actionable consumer step is keeping a verified, labeled intake source, not estimating from "bacteria are present" labels. Recheck dietary adequacy annually for restrictive diets.
Limitations: what this pathway explains and what it cannot
The science above explains the supply, not any individual's medical status, and this article makes no determination about whether any particular food or supplement works for a given person. The pathway's main open questions are real. Fermentation yields depend on proprietary strains, so broad claims about "most B12" should be treated as direction, not measurement. Estimates of which gut bacteria make accessible B12 vary by method, and amounts recovered from stool do not certify absorbed amounts. Whether the anaerobic pathway's products ever match aerobic yields at industrial scale is a current question, and published answers are sparse. Most consumer claims need a measured label and a clinician's judgment, not a demonstration in the abstract. The lesson of the microbial factory is the limit of our numbers: we can map nearly every one of the roughly 30 enzymatic steps and still not set a single threshold for your body from a science article alone.