Why the 'colon cancer bacterium' is normal in your mouth

Ask someone to name a bacterium that lives in their mouth and you will usually get Streptococcus, or a shrug. Almost nobody says Fusobacterium. And yet if you scraped the plaque off one tooth tonight and put it under a microscope, you would very likely be looking at them: long, thin rods, tapered at both ends like a spindle. Fusus is Latin for spindle. That is the whole of the name, and it may be the most honest thing said about this group, because most of what follows in the press is either too kind or too frightening.

What Fusobacteriota is

Fusobacteriota is a phylum of Gram-negative, strictly anaerobic, mostly spindle-shaped bacteria whose best-known human members belong to the oral genus Fusobacterium.

A phylum is a very wide branch of the bacterial tree — the level above class, order, family and genus. So this is not one bug. It is a whole limb of life, and Fusobacterium nucleatum, the species that gets all the press, is a single twig on it. Keep that scale in mind; almost every confusion about this organism comes from mistaking the twig for the limb.

Where it lives in the body

Primarily, the mouth. Not the tongue alone and not the teeth alone, but the whole architecture: dental plaque, the gingival sulcus where gum meets tooth, the tonsils, the throat. When Dewhirst and colleagues built the first curated map of the human oral microbiome for the Journal of Bacteriology, they catalogued 619 oral taxa in thirteen phyla and separately analysed 36 043 16S rRNA gene clones, and Fusobacteria — as the name was then spelled — was one of the thirteen ².

Below the mouth the picture changes. In the lower gut of healthy people, Fusobacterium nucleatum is described as rare ⁹. It does turn up in the intestine, in the placenta, in the bloodstream, in abscesses — but as a traveller, not a resident. That distinction, resident here and visitor there, turns out to be the key to the entire story.

What it does for us

Two things, and I want to be precise about how strong each one is.

It holds the plaque community together. Dental plaque is not a smear; it assembles in a sequence, early colonisers first and late anaerobes last, and the physical distance between cells of different species governs what can grow at all ³. Fusobacterium nucleatum sits in the middle of that sequence and binds an unusually wide range of partners — it has been described as creating biological bridges that stabilise the multispecies dental biofilm, an ability traced in part to a galactose-sensitive adhesin called Fap2 ¹¹. A stable, ordered community is what a healthy mouth actually is. Take out the connector and you do not get a cleaner mouth; you get a different and less predictable one.

It appears to prime the gum's own antibiotic. In work published in Infection and Immunity, cell wall extract of F. nucleatum switched on human beta-defensin 2 — an antimicrobial peptide the gum manufactures itself — in cultured gingival epithelial cells within 2 to 4 hours. Cell wall extract from the periodontal pathogen Porphyromonas gingivalis did not. The same paper found beta-defensin 2 messenger RNA in 14 of 15 non-inflamed gum samples ⁴. The authors' own reading was that commensal organisms have a role in keeping the epithelial barrier intact.

Now the caveat, said out loud, because it matters more than the finding: neither of these is a health claim. Nobody has shown that more Fusobacterium makes a mouth healthier, and nobody sells it as a probiotic, for good reason. What the evidence supports is narrower and more interesting — this organism is a structural and signalling part of a normal mouth, not an invader that slipped in.

What it does against us

The same organism, in the wrong place, is genuinely dangerous.

In the mouth it is a long-recognised participant in periodontal disease. Beyond the mouth, a review in Current Opinion in Microbiology catalogues its association with adverse pregnancy outcomes, inflammatory bowel disease, appendicitis, cardiovascular disease, respiratory infections and Lemierre's syndrome — the throat infection that spreads into the jugular vein ¹⁰. In mice, a strain of F. nucleatum with a disrupted fap2 gene showed a 2-log reduction in colonisation of the placenta compared with the normal strain, which is a clue to how it reaches places it should never be ¹¹.

Then there is the colon. In 2012 two groups published in the same issue of Genome Research, independently and by different methods. Castellarin and colleagues found Fusobacterium over-represented in tumour tissue by RNA sequencing and confirmed it by quantitative PCR across 99 subjects, at p = 2.5 × 10⁻⁶ ⁵. Kostic and colleagues started from whole genome sequences of 9 tumour and normal pairs, confirmed the enrichment in 95 carcinoma-normal pairs, and saw the bacteria inside the tumours by fluorescence microscopy ⁶. When two labs reach the same place from opposite directions, the finding is usually real.

A year later came a mechanism. Rubinstein and colleagues showed that the FadA adhesin of F. nucleatum binds E-cadherin on human cells, activates beta-catenin signalling and stimulates the growth of colorectal cancer cells. The binding site on E-cadherin maps to an 11-amino-acid stretch, and a synthetic peptide copying that stretch abolished the effect. In colon tissue from patients with adenomas and adenocarcinomas, fadA gene levels ran more than 10 to 100 times higher than in normal individuals ⁷.

And in Science, Bullman and colleagues found the bacterium does not stay behind: it travelled with tumours into distant liver metastases, survived successive passages of human tumours through mice, and treatment of those mice with the antibiotic metronidazole reduced bacterial load, cancer cell proliferation and overall tumour growth ⁸.

That last result is the one most often misquoted. It was mice carrying human tumour grafts. It is a reason to run a trial, not a treatment anyone should act on.

The name — new and old

Until recently the rank of phylum was not covered by the bacterial naming code at all, which is why textbooks carried informal names for decades. That changed, and in 2021 Oren and Garrity published formal names and descriptions for forty-two phyla of prokaryotes at once ¹. Fusobacteriota is the current name. Fusobacteria is the older spelling and remains its synonym; NCBI Taxonomy lists both against the same entry, so a paper using either word is talking about the same organisms.

What the evidence does not say

Here is where most coverage of this phylum goes wrong.

In 2024 a team writing in Nature closed 135 F. nucleatum genomes — 80 strains from the mouths of people without cancer, and 55 strains cultured from tumours in 51 patients with colorectal cancer. What they found was a split. The subspecies animalis, long treated as a single thing, is in fact two clades, and only one of them dominates the tumour niche. The pattern held in tumour tissue from 116 patients and in stool metagenomes from 627 patients with colorectal cancer set against 619 healthy people ⁹.

So "Fusobacterium causes colon cancer" is too coarse to be true. The accurate sentence is longer and less satisfying: one sub-lineage, carrying particular genes for surviving the gut, is enriched in colorectal tumours, plausibly contributes to their growth, and makes up a small fraction of the Fusobacterium most people carry in their mouths.

There is no consumer test for it. There is no diet shown to shift it. And there is no evidence that attacking your own plaque harder — beyond ordinary brushing, flossing and a dentist — does anything except disturb a community that, in a healthy mouth, this phylum helps hold together ³. The takeaway is not fear of a bacterium. It is that where a microbe lives can matter as much as which microbe it is.