Can Gut Bacteria Give You Belly Fat? Two of Them Just Did — to Mice.

Most obesity research still runs on BMI, which is height and weight and nothing else. It cannot tell the difference between a bodybuilder and a sedentary office worker, and — more importantly — it cannot tell you where the fat is.

That matters, because the fat you can pinch is not the fat that hurts you. Visceral fat is the deep stuff packed around the liver, pancreas and intestines. It behaves like an active organ, pumping out inflammatory signals and free fatty acids directly into the vein that feeds the liver. It is the fat that drives insulin resistance, fatty liver and heart disease — and you can carry a lot of it at a perfectly ordinary BMI.

So a team at Seoul National University analyzed 1,771 generally healthy Korean adults from a health-screening cohort, combining CT-measured visceral fat with gut-microbiome data — and then went a step beyond the usual cross-sectional microbiome study.

Two people can share the same BMI while carrying very different amounts of visceral fat. In this study, CT-measured visceral fat showed particularly consistent associations with gut-microbiome composition.

What they found

1. Visceral fat stood out among the obesity measures. CT-measured visceral fat showed more consistent and significant associations with microbial taxa than conventional measures such as BMI. That is a useful methodological result: where fat is stored may capture metabolic biology that body weight alone misses. It does not mean visceral fat was the strongest predictor of every microbiome feature.

2. They built a composite metabolic yardstick. Using principal-component analysis, the researchers combined nine correlated obesity-related clinical variables — including visceral fat, waist circumference, insulin and triglycerides. The first component, used as a composite obesity score, captured 50.4% of the variation across those nine measurements. That does not mean it predicted 50.4% of obesity; it means one statistical component summarized about half of the shared variation in the selected measurements.

3. They named names — on both sides. Several species lined up with higher visceral fat, triglycerides and insulin: Ruminococcus gnavus, Dorea longicatena, Blautia wexlerae and Clostridium glycyrrhizinilyticum. Others went the other way, sitting with leaner, healthier profiles — including Akkermansia muciniphila and the fibre-fermenting Ruminococcus bromii. Dorea longicatena had the clearest positive link to the obesity score. People with high visceral fat also had measurably less bacterial diversity and a distinguishably different community overall.

4. They found a possible metabolic clue — not yet a mechanism. In a 44-person shotgun-metagenomic subset, a microbial glycogen-synthesis pathway was associated with the composite obesity score (P = 0.049). Its association with visceral fat itself was weaker (P = 0.097). The authors propose that microbial carbohydrate metabolism may be connected to host fat accumulation, but the biochemical chain linking bacterial glycogen synthesis to human fat storage was not demonstrated.

5. Then they functionally tested candidate bacteria in mice. The researchers depleted the mice's resident microbiota with antibiotics, placed them on a high-fat diet, and gave three groups a daily oral dose for six weeks: control solution, D. longicatena, or Bifidobacterium adolescentis (14–15 mice per group). This moves beyond human correlation, although it does not establish that the same causal relationship operates in people.

6. Under these experimental conditions, both bacteria promoted greater fat accumulation. From around week 3–4, both bacterial groups gained significantly more weight than controls. Body-composition measurements showed increased fat mass without a significant change in lean mass, while adipose depots were heavier and serum leptin increased. The livers were also heavier, with higher expression of lipogenesis-related genes including Fasn, Acc and Gck. These findings support a biological effect in this antibiotic-depleted, high-fat-diet mouse model; they do not show that the same bacteria make humans gain weight.

In this antibiotic-depleted, high-fat-diet mouse model, the groups receiving either candidate bacterium accumulated more fat than controls. This is a functional animal experiment, not proof of the same effect in humans.

Now the part the headline can't fit

An important effect-size check: each tested clinical variable explained less than 1% of the variation in overall microbiome composition. This is not the same as saying gut bacteria explain less than 1% of obesity.

Under one percent — but be careful what that means. The reported R² values were below 0.01 for associations between clinical variables and overall microbiome composition. In other words, each clinical variable explained less than 1% of variation in the microbial community. That is not the same as saying that the microbiome explains less than 1% of visceral fat or obesity. The result instead tells us that these statistically significant community-level associations had small effect sizes.

The human half cannot tell you which way the arrow points. Everyone was sampled once. So "people with more visceral fat have more Dorea" is equally consistent with the bacteria promoting fat, the fat tissue reshaping the gut environment, or — most likely — a shared diet driving both. Cross-sectional data simply cannot separate these, which is precisely why the mouse work matters.

But those mice were not living your life. Their native gut bacteria were destroyed with a three-antibiotic cocktail, leaving an unnaturally empty ecosystem that a newcomer can colonise far more easily than a normal one. They were then fed a high-fat diet and given 100 million cells of a single species by gavage, every single day, for six weeks. That is a stress test of a biological mechanism, not a model of a person eating yoghurt. It shows the bacterium can push fat accumulation under permissive conditions. It does not show it does so in an intact human gut.

Food intake leaves an important question. Food intake was recorded weekly, but a clear between-group comparison is not reported in the main text. Without that information, we cannot tell how much of the additional weight gain reflected altered energy intake versus other metabolic effects.

The mechanism is two findings held next to each other. The authors are commendably direct about this: the bacterial glycogen-synthesis signal and the switched-on liver fat genes are, in their words, two independent associative findings. Nobody has traced the biochemical chain from a microbe storing starch to a liver cell building fat. The glycogen pathway result also rests on deep sequencing of just 44 people — 22 with high scores against 22 with low — which is a narrow base for the paper's central mechanistic idea.

And the awkward one: Bifidobacterium adolescentis. Bifidobacteria are commonly presented as “friendly” gut bacteria, yet the specific B. adolescentis strain tested here (KCTC 3216) promoted greater fat accumulation in this mouse model. That is not evidence that B. adolescentis as a species — much less commercial bifidobacterial probiotics — causes weight gain in humans. The published literature is conflicting, and the authors themselves state that its influence on obesity remains uncertain. Strain, host, diet and the surrounding microbial ecosystem can all matter.

And these were generally healthy, middle-aged to older Korean adults. The average age was 58 and the mean BMI was 23.8 kg/m², although individual BMI values varied widely. The cohort excluded people with metabolic syndrome. Whether the same microbial signatures appear in younger populations, people with obesity or metabolic disease, or populations eating different diets remains uncertain. The main cohort was also profiled using 16S rRNA sequencing, which has more limited taxonomic resolution than shotgun metagenomics.

This post is a plain-language summary of a published research paper and is not medical advice. Nothing here is a reason to start, stop or change a supplement or medication — talk to your own doctor about that.

Source
Kim N-E, Cho B, Kim Y, Kim G, Yie G-E, et al. (corresponding author Won S). Linking visceral fat accumulation to gut microbiota: key bacterial taxa and their roles in the glycogen synthesis pathway. Experimental & Molecular Medicine. 2026. doi.org/10.1038/s12276-026-01828-7
For the record — cross-sectional human cohort plus mouse validation. 1,771 healthy adults from a Seoul National University Hospital health-screening programme (Feb 2017–Nov 2018), mean age 58.1 ± 11.4 y, 45.2% women, mean BMI 23.8 ± 3.1 kg/m², mean CT-measured visceral fat area 126.6 ± 49.3 cm²; faecal 16S rRNA (V3–V4) sequencing, with shotgun metagenomics in a 44-person subset (22 high vs 22 low obesity score). A composite obesity score (PC1) from nine clinical variables explained 50.4% of variance. Species associated with visceral fat, triglycerides and insulin at BH P < 0.05 included Ruminococcus gnavus, Clostridium glycyrrhizinilyticum, Dorea longicatena and Blautia wexlerae (positive) and Blautia obeum, Ruminococcus lactaris, Intestinibacter bartlettii, Akkermansia muciniphila and Ruminococcus bromii (negative); D. longicatena vs obesity score β(TMAT) = 0.045, BH P < 0.001. Shannon diversity lower in the high-visceral-fat group (P < 0.05); beta-diversity separation P = 0.02 (PERMANOVA). GLYCOGENSYNTH-PWY correlated with PC1 (P = 0.049) and visceral fat (P = 0.097). All reported R² values were below 0.01. Four enterotypes identified (DMM clustering); enterotype 3 had the lowest metabolic risk profile, enterotype 4 the highest. Mouse study: C57BL/6 mice, antibiotic depletion (ampicillin, neomycin, vancomycin) for 1 week plus 1-week washout, then high-fat diet with daily oral gavage of 10⁸ CFU for 6 weeks; 14–15 mice per group (PBS, D. longicatena, B. adolescentis). Both bacterial groups showed significantly greater body-weight gain from week 3–4, increased fat mass with unchanged lean mass, heavier eWAT and iWAT, raised serum leptin (P < 0.05), heavier livers and upregulated hepatic Fasn, Acc and Gck (P < 0.05). Food intake was recorded weekly but no between-group comparison is reported in the text. Authors' stated limitations include the unexplained biochemical link between bacterial glycogen synthesis and hepatic lipogenesis (described as two independent associative findings), the healthy-only cohort, enterotypes being representative patterns rather than discrete entities, the very small R² values, and conflicting prior literature on B. adolescentis. Ethics: SNUH IRB C-1609-051-790; animal work GIST IACUC GIST-2023-060.