- We started mass-producing plastic in the 1950s. In 2016, Japanese researchers found a bacterium, Ideonella sakaiensis, living on a PET bottle and using it as food.
- It does this with two enzymes: PETase cuts the plastic into a fragment called MHET, and MHETase cuts that into two small molecules the cell can eat.
- The leading explanation for how it evolved so fast is exaptation: bacteria already had enzymes for a natural plastic-like polymer (cutin, the wax on leaves), and a few mutations retooled them for PET.
- It is real and useful for recycling PET in a factory. It is not going to eat the plastic in the ocean. Those are two very different claims and most headlines blur them.
- There's an interactive timeline further down - 1907 to 2025, click any event.
This started with a dumb question. I was reading something about recycling and wondered whether plastic counts as organic chemistry or inorganic. It's organic - plastics are long carbon chains, same branch of chemistry as the molecules in your body. Fine. But that led me to the next question: if plastic is just carbon chains, does anything eat it?
Turns out yes. And the story is stranger than I expected, because plastic in the modern sense has only existed for about 70 years, and evolution isn't supposed to move that fast. I went down the rabbit hole for a day. Here's what I found, with the sources, because this is the kind of topic where half the internet is confidently wrong.
The bottle that was being eaten
In 2016 a team led by Kohei Oda and Kenji Miyamoto published a paper in Science with a very literal title: "A bacterium that degrades and assimilates poly(ethylene terephthalate)". They had collected sludge and debris from outside a PET bottle recycling plant in Sakai, Japan, screening for anything that could live on plastic.
They found it. Ideonella sakaiensis strain 201-F6 was growing on a piece of PET film and using it as its main source of carbon and energy - the plastic itself, not some label glue on top of it. Under an electron microscope the cells were stuck to the film by thin appendages, little tendrils, which are thought to pipe the bacterium's enzymes straight onto the polymer surface.
It's slow. A thin, low-crystallinity PET film took roughly six weeks to break down at 30 °C. But it was real digestion - in one experiment about three-quarters of the film's carbon ended up as CO₂, which means the bacterium wasn't just scratching the surface, it was fully metabolising the thing. PET, the plastic in your water bottle and a lot of your clothes, was food.
How it actually breaks the plastic apart
PET is a polyester. It's a long chain built from two small molecules - terephthalic acid and ethylene glycol - joined by ester bonds. If you want to eat the chain, you need to cut those bonds. I. sakaiensis does it in two steps, with two enzymes working as a relay.
PETase is the first enzyme. It attacks the surface of the plastic and clips the ester bonds, releasing a fragment called MHET (mono-2-hydroxyethyl terephthalate). Mechanically it's an ordinary serine hydrolase - the same Ser-His-Asp "catalytic triad" trick that a lot of enzymes use to cut bonds. Nothing exotic in the chemistry. What's special is that it happens to work on a man-made polymer.
MHETase is the second enzyme. It takes that MHET fragment and splits it into the two original building blocks: terephthalic acid and ethylene glycol. Now they're small, soluble molecules, and the cell metabolises them for growth. The two enzymes together take a solid brick of plastic all the way down to monomers. That's the whole trick.
The neat consequence: if you can recover pure terephthalic acid and ethylene glycol, you can build brand-new, virgin-quality PET from them. Break the bottle down to Lego bricks, snap a new bottle together. That's the entire pitch for enzymatic recycling, and it's why industry got interested fast.
The part that bugged me: 70 years is not a lot of time
Here's what sent me deeper. PET was patented in the 1940s and only became a mass-market material from the 1950s onwards. Bottles came even later. So this bacterium developed a two-enzyme system for a substrate that has existed for a human lifetime, tops. Evolution usually works on timescales that make that look absurd.
The answer isn't that bacteria are magic. It's that they mostly didn't build anything new. The leading explanation is exaptation - an existing tool getting repurposed.
1. They already had a "natural plastic" enzyme
Plants coat their leaves in cutin, a waxy polyester made of fatty acids linked by ester bonds. It's a natural polymer that's chemically not that far from PET. Plants have used it for over 400 million years, and for just as long, bacteria and fungi have made enzymes called cutinases to break it down and get at the plant.
A cutinase already cuts ester bonds in a polyester. PET is an ester-linked polyester. So the "new" ability was never zero to one. It was a cutinase that could weakly chew on PET as a side effect, and then a handful of mutations tuned it to do it better. PETase turns out to be structurally very close to known cutinases - it's basically a cutinase that got good at the wrong substrate.
2. Plastic became a food source, so there was pressure to get better at it
Once plastic was everywhere - landfills, rivers, recycling plants, and the floating debris in the ocean - it turned into an abundant pile of carbon sitting there unused. In 2013 researchers named the microbial communities that colonise ocean plastic the "plastisphere". In a spot where plastic is the main thing around, any microbe that can extract even a little energy from it has an edge. That's ordinary natural selection, just pointed at a brand-new resource we created by accident.
3. Bacteria evolve fast, and they share genes sideways
Bacteria divide in minutes, live in populations of billions, and mutate constantly, so they explore a lot of variations quickly. They also swap genes directly through horizontal gene transfer - a useful gene doesn't have to be reinvented, it can spread across a community. A 2021 Chalmers study mined ocean and soil samples and found over 30,000 candidate plastic-degrading enzymes, and, tellingly, the number of these genes in a location tracked the local level of plastic pollution. The more plastic, the more degrading genes. That's the fingerprint of adaptation actually happening.
One honest caveat, because I saw a lot of overclaiming here. There's a fair argument about whether this counts as dramatic "new" evolution or mostly the discovery of capacity that was already lurking. The reasonable read is: the raw ability (a cutinase touching PET) pre-existed, and the mutations that sharpened it for PET are real, recent evolution. Both things are true. Anyone telling you bacteria "learned" or "taught themselves" to eat plastic is putting a brain on an enzyme. There's no brain. There's a fast selection process running on molecules that were already close.
The whole story, on a timeline
I built this to keep the dates straight, because the news coverage tends to mash them together. Filter by the kind of event, and click any entry to expand it. Every one links to a source.
No events in that category.
What it can do, and what it can't
This is where I have to be a killjoy, because the gap between the science and the headlines is wide.
What's real: as a way to recycle PET in a controlled setting, this works and it's improving fast. Since 2016 the enzymes have been massively upgraded. In 2018 a Portsmouth and NREL team solved PETase's structure and, while poking at it, accidentally made a better version. In 2020 the French company Carbios engineered a cutinase that broke down 90% of PET in about 10 hours. In 2022 a UT Austin lab used machine learning to design FAST-PETase, which chewed through untreated PET from 51 different products in a week, at under 50 °C. Carbios is now building a 50,000-tonne-a-year plant in France. That's a genuine industrial process arriving.
What's not real, at least not yet:
- It's PET-only. PETase and its cousins work on PET. The other big plastics - polyethylene bags, polypropylene, PVC, polystyrene - are plain carbon-carbon chains with no ester bonds to cut. Polyurethane has more breakable bonds but still no workhorse enzyme on PETase's level. A real solution needs several more breakthroughs like PETase, for materials that are harder to attack.
- It needs heat and clean input. The fast enzymes want 50-72 °C and reasonably pure, low-crystallinity plastic. That's a factory reactor, not a beach. Dyes, coatings and mixed waste gum up the works.
- It will not clean the ocean. This is the big one. The 2025 discovery that ocean bacteria across most of the planet carry PET-degrading genes is fascinating, but the natural rate is glacial. Those microbes are not out there dissolving the garbage patch. Pollution goes in far faster than any bug takes it out.
The honest framing is the one the Guardian used in a good 2023 long-read: this is a promising end-of-life recycling technology for one type of plastic, not a cleanup for the mess already out there. Both can be worth getting excited about. They're just not the same thing, and conflating them is how you end up with people assuming the problem is solved.
So, how many people actually know this?
That was my original question, the one that made me want to write this up. The answer is: kind of. "There's bacteria that eat plastic" floats around as a vague hopeful eco-fact. The 2016 discovery was big news - Science, the BBC, the Guardian, a pile of Reddit threads. But the actual mechanism, the name Ideonella sakaiensis, the honest limits - that stays niche. Which is a shame, because the real story is better than the headline. We accidentally ran a 70-year experiment in evolution, dumped a new food source across the entire planet, and watched microbes that were already halfway there close the gap. That's worth understanding properly, hype stripped out.
I'm not a biologist. I'm a developer who got curious and read the papers, and I've linked every real claim above so you can check my work. If you know this field and I got something wrong, tell me - I'd rather fix it than be confidently wrong on the internet.