🧫 A team from Kobe University went digging in the sediment of a PFAS-polluted river in Osaka, looking for bacteria that could live there. They found seven strains. Three of them were also removing the chemical. The numbers are smaller than some headlines suggested, and whether the drop reflects actual breakdown is still unsettled.

What came out of the riverbed

The reasoning behind the experiment was blunt. If a synthetic chemical has been settling into a riverbed for decades, whatever is still alive down there has already made its peace with it. So take the mud, grow it in a medium spiked with PFOS and PFOA, and see what survives.

Associate professor Hideyuki Inui and colleagues at Kobe University's Biosignal Research Center collected sediment from a polluted river in Osaka Prefecture, identified by Kyodo News as the Samondo River. Seven strains grew. A 16S rRNA gene analysis sorted them into four genera: Rhodococcus, Priestia, Paenibacillus and Xanthobacter. Three of the seven lowered PFAS concentrations. The findings appeared online on June 19, 2026 in the Journal of Water Process Engineering, and the university announced them on July 15.

PFAS stands for per- and polyfluoroalkyl substances, a family of more than 10,000 synthetic compounds. Their defining feature is the carbon-fluorine bond, among the strongest in chemistry, which is exactly why they shrug off water, oil, heat and nearly every natural process that breaks other things down. Hence the nickname: forever chemicals.

17.8% and 97% describe different experiments

In laboratory medium, Xanthobacter sp. strain XaP reduced PFOS by up to 17.8%. Priestia sp. strain PmB reduced PFOA by up to 14.1%. Both the linear and the branched isomers of PFOS came down. Those are the university's own headline figures, and they are modest.

The team then took its best performer and dropped it into something messier: real leachate collected from an industrial waste landfill, the complicated liquid that drains out of a dump rather than a clean solution mixed in a flask. After a week, PFOA in that leachate fell 23.3%, Kyodo reported. For PFHxS, a shorter-chain relative of PFOS, removal reached 97%.

That 97% is the number that travelled. It is also the one most likely to be misread. It applies to a different compound in a different liquid than the 17.8% does. Summarising all this as "bacteria remove 97% of forever chemicals" collapses two experiments into a sentence that neither supports.

Breaking it down, or hanging onto it?

When they looked at how the chemicals and the cells interact, PFOS stuck to bacterial surfaces more strongly than PFOA did, and the gram-negative Xanthobacter proved especially good at holding on. Tracking the process over time pointed at extracellular polymeric substances, or EPS, the mesh of sugars and proteins that bacteria secrete around themselves.

Which leaves an uncomfortable possibility. If PFAS vanishes from the water because it has migrated onto the outside of a bacterium, nothing has been destroyed. It has been moved.

The team says so openly. Whether the drop reflects real breakdown or conversion into other PFAS still needs high-resolution mass spectrometry to establish. Their own list of open questions includes working out how much of the removal is adsorption versus degradation, and assessing the risk that captured PFAS gets released again later.

That last item is not hypothetical. Activated carbon also traps PFAS by adsorption, and Japan has already seen what happens when spent carbon is stored badly. A town of about 10,000 in Okayama Prefecture drank contaminated water for at least three years, and the source turned out to be bags of used filter media left sitting outdoors in the rain, as we covered earlier. Capture is half a solution until you know where the captured material ends up.

Burning it, or leaving it to the bacteria

Japan already has a destruction route that works. In June 2026, Kubota reported breaking down PFOS, PFOA and a substitute compound at rates above 99.999% in rotary surface melting furnaces running at 1,250 to 1,400°C, with evidence pointing to full mineralisation rather than partial breakage. We wrote about that result when it landed.

Against that, 17.8% at room temperature looks unimpressive. But the two are not competing for the same job. A melting furnace is a large fixed installation that needs enormous heat and needs the waste delivered to it. Bacteria are cheap, they work at ambient temperature, and in principle they can go where the contamination already sits. Inui told Kyodo that being able to collect and culture bacteria from the natural environment would bring cleanup costs down. The physical and chemical methods already in service all work: activated carbon, ion exchange resin, membrane separation. Cost is what limits how widely they get used.

Division of labour looks more plausible than replacement. Concentrated industrial waste goes to the furnace. Diffuse, low-level contamination spread across a riverbed is a different problem, and one a furnace was never going to reach.

The same family of bacteria, on the other side of the world

A study published in early 2025 reported that Labrys portucalensis F11, a strain isolated in Portugal, degraded roughly 90% of PFOS over 100 days, leaving defluorinated metabolites behind as evidence. Genuine bond-breaking, very slowly. That test used spiked laboratory media at concentrations the authors themselves note are far above anything in the environment. Kobe went into real leachate. The two are not being scored on the same course either.

L. portucalensis belongs to the family Xanthobacteraceae. So does Xanthobacter, the genus behind Kobe's strongest performer. Xanthobacter is in fact the family's type genus, the one it was named after.

Two separate teams on opposite sides of the planet converged on close relatives. That does not prove the family holds the answer. It does suggest the hunt for PFAS-active microbes is narrowing toward somewhere specific.

The result also lands in a year when the rules changed. Since April 2026, Japanese water utilities have been bound by a legally enforceable drinking water standard capping PFOS and PFOA at a combined 50 nanograms per litre, replacing a provisional target that carried no force. Regulation arrived first. The cleanup toolkit is still being assembled, and as of July 2026 these bacteria are at the stage of proving which mechanism they use, not being installed at a treatment plant.

Japan is measuring aggressively and cleaning up slowly. What does that sequence look like where you are? Has PFAS made it into your drinking water rules yet, or is it still something that surfaces mainly near a base or a factory?

参照