Credit: Juan Gaertner/Science Photo Library / Getty Images

In June, French biotech Carbios presented its biorecycled bottle to the world. Made from polyethylene terephthalate (PET) plastic, the transparent flask contains French beauty brand L’Occitane’s almond oil shower gel. But this is not just any plastic. It is not made from crude oil or natural gas. This PET is made from biorecycled plastic, digested from used bottles with microbial enzymes.

Plastic-eating enzymes are just one example of the means by which microbes can be harnessed to eat pollutants—from the ‘forever chemicals’ per- and polyfluoroalkyl substances (PFAS) to cyanide and petroleum1. Governments and biotechs are investing in bioremediation, as the science of using living organisms to remove pollutants is known and proven, with the market expected to grow by $8.29 billion between 2023 and 2028. Bioremediation is “affordable, doable and uncontroversial,” says Jillian Banfield, a microbiome researcher at the University of California, Berkeley.

But while isolated microbial enzymes are being commercialized for bioremediation, other potential approaches, such as adding live or genetically modified microbes to contaminated water or soil, pose technological challenges or may be blocked by regulatory barriers.

Evolve to eat

The burgeoning field of plastic biorecycling was given a boost in 2016 when a paper in Science described a new bacterial species, Ideonella sakaiensis 201-F62. This microbe, found growing on bottles in a Japanese plastic recycling facility, contained the first described PETase enzyme, which depolymerizes PET into mono(2-hydroxyethyl) terephthalate (MHET). A second enzyme in the same bacterium then cleaves MHET into its constituent monomers: terephthalic acid and ethylene glycol. These are, conveniently, the key ingredients to make more PET.

Bacteria evolve rapidly and so will find a way to adapt to their environment and to eat whatever carbon sources are available, including plastic, explains Ronan McCarthy, an engineer at Brunel University in the UK. It’s likely that I. sakaiensis 201-F6 evolved its plastic-eating enzymes from ferulic acid esterases, which are useful for digesting through plant walls.

“There's a reason microorganisms inhabit almost every site on the planet,” says McCarthy admiringly, who adds that bacteria are powerful tools “to tackle some of the more challenging environmental pollutants in a very sustainable and eco-friendly way.”

Ideonella sakaiensis 201-F6, as the first bacterium found to use plastic as a food source, sparked a wave of interest in bioremediation. Until then, scientists’ focus had been on cutinases—lipolytic/esterolytic enzymes that hydrolyze cutin (a component of the plant cuticle). Cutinases allow fungi to digest their way through the plant skin, allowing spore adhesion and release, but some can also digest plastic, albeit slowly. Plastic degradation represents an “enzymatic kind of promiscuity” for these species, says McCarthy, as naturally occurring cutin has carbon–carbon bonds similar to those of manufactured plastic.

For Carbios, the secret sauce is plastic-eating microbial enzymes, which it supercharges using genetic engineering to speed up the naturally slow process. To find the best plastic-eating enzyme for its biorecycling plant, the biotech screened five fungal and bacterial enzymes (including from I. sakaiensis 201-F6) to see which was fastest at depolymerizing PET. The winner was the unimaginatively named ‘leaf-branch compost cutinase’—discovered by another Japanese team in 2012 (ref. 3) and so called because it was found during a metagenomic screen of leaf and branch compost collected from a Japanese park.

“This enzyme was the best,” says Alain Marty, CSO of Carbios, “but was not very good,” and so he and his team set out to improve it through genetic engineering4. Marty directed the team to target 11 amino acids in active sites for mutagenesis, generating 209 variants. They then used microfluidic screening to test the mutants, searching for those with increased depolymerization activity or improved thermal stability. They hit the jackpot, with the best of the mutants able to depolymerize 90% of PET in under 10 h. Crucially for Carbios’ business model, the products of this reaction could then be used for plastic manufacturing of new bottle-grade PET.

The company announced this year that it will churn out biorecycled plastic for two ‘grandes dames’ of French cosmetics, L’Oréal and L’Occitane, as well as for clothing manufacturers, who call PET by its common name: polyester. The plant’s location in the French commune of Longlaville is conveniently close to the borders of Germany, Luxembourg and Belgium, and it will soon be biorecycling 50,000 tonnes of plastic waste per year (much of it coming across the nearby borders), equivalent to 2 billion bottles. “It is a good location to find waste,” says Marty.

Carbios was founded by VC fund Truffle Capital (Truffle CEO Philippe Pouletty is chair of the Carbios board) and has received grants of €54 million from French national and regional governments, as well as investment from L’Oréal’s BOLD (Business Opportunities for L'Oreal Development) fund, L’Occitane and tire company Michelin. The company raised €114 from the issuance of 3 million new shares in May 2021 to help cover the €230 million cost of the new plant. As well as selling the raw recycled ingredients for plastic, Carbios aims to sell its enzymes and license the technology to other companies, says Marty.

Government interest in plastic biorecycling is high. French president Emmanuel Macron praised Carbios’ new plant, which is due to open by the end of 2025, in a statement for the ground-breaking ceremony in April: “At a time when governments are negotiating an international treaty against plastic pollution in Ottawa, the groundbreaking of Carbios’ biorecycling plant is particularly significant.” In August, Carbios announced that it would partner with Northampton-based FCC Environment, a recycling and waste management company, to build a similar plant in the UK.

Boosting biofilms

Researchers can harness bacteria for bioremediation without growing them through metagenomics, in which all the genes in a sample of water or soil are sequenced—allowing new genes (and their encoded proteins) to be discovered. Metagenomics is a useful tool for finding new enzymes, explains McCarthy, whose lab has found plastic-eating enzymes, most of which are “not very efficient,” he says. Rather than mutate the enzyme itself to increase its activity, as Carbios is doing, McCarthy has found a different way to boost activity, by coaxing the bacteria into a biofilm. The advantage of a biofilm, he explains, is that it will form a layer on the plastic (rather than floating free), increasing the surface area5 exposed to the enzyme. “You're trapping the enzyme around the actual plastic,” he says, describing research his lab published in October 2023.

McCarthy’s idea of using biofilms came from a chance encounter on a beach. As he was pushing his then-6-month-old son in his buggy, he saw that “a huge amount of plastic had been washed up.” A shiny film on the bottles alerted the microbiologist to what looked like a bacterial biofilm. McCarthy reasoned that the bacteria were “using the plastic to survive,” so he took the plastic waste back to his lab, cultured the biofilm in ‘synthetic seawater’ and successfully enriched for plastic-eating bacteria6, finding strains of Pseudomonas stutzeri—a species of bacteria that had not previously been known to digest any plastic, including the expanded polystyrene (EPS) plastic from which the bottles were made.

McCarthy sequenced the bacteria’s genes and found two putative esterases and one putative MHETase, which are likely responsible for the plastic digestion and which could be commercialized. McCarthy is also looking to license pathways that control biofilm formation. The UK government is “putting a lot of funding” into bioengineering approaches to tackle pollutants such as plastic, he says, including a £13 million Environmental Biotechnology Innovation Centre hosted at Cranfield University that includes nine other UK universities, including Brunel. The new center has water treatment plants that can be used for experimentation, says McCarthy, allowing new microbes and their activities to be tested “at scale.” Much of the global funding for bioremediation comes from governments, partly because of public pressure to tackle plastic pollution but also because some government bodies, such as the US Department of Defense, own contaminated sites.

McCarthy’s approach uses live bacteria, but many bioremediation biotechs opt for purified enzymes rather than whole microorganisms, in part because of stringent regulations for releasing live organisms into the environment. Engineered bacteria would be even more challenging to release because they are classified as genetically modified organisms (GMOs), whereas isolated enzymes from those bacteria would not be. Unlike the politically polarizing GMOs, enzymes (which are found in most homes as an ingredient in laundry detergent) are an “uncontroversial technology,” says McCarthy, and so ripe for commercialization.

Another advantage of using enzymes is that their genes can be cloned from large metagenomics databases, including genes from new, unknown and unculturable bacteria. This massively increases the pool of potential enzymes, as “the vast majority of microorganisms are not cultivatable in pure cultures,” says Banfield, who has discovered dozens of new bacteria and bacterial genes using metagenomics.

Plastic-degrading enzymes could be ubiquitous in the future, according to McCarthy, just as millions of households use laundry detergent with stain-removing enzymes. “If you can put the solution in everybody's home, then it really has transformative potential,” he says.

Synthetic solutions

The best place to find pollutant-eating microbes is at the polluted site itself, whether that be a plastic recycling plant or a cyanide-rich gold mine.

Pisa, Italy-based DnD Biotech samples microbial communities from contaminated sites to find bacteria that can feed on the pollutant. In a pilot project, the biotech sampled soil contaminated with petroleum hydrocarbons in the Persian Gulf region, finding species of Pseudomonas, Bacillus and Acinetobacter that can degrade the petrol, according to a company statement. Once isolated, its scientists cultured these petrol-eating bacteria and then reintroduced them into the field site, boosting remediation. DnD Biotech is also investigating bioremediation of PFAS and is involved in the MIBIREM (toolbox for microbiome-based remediation) project, a public–private partnership to bioremediate the 324,000 severely contaminated sites across Europe, including mines, landfills and petrol stations. Funded by the European Union, MIBIREM’s approach comes with advantages of low cost and minimal regulatory challenges, as the bacteria are native to the local soil or water and are not GMOs.

But searching for naturally occurring microbes that digest pollutants is “a big bottleneck,” says Peyman Salehian, CEO of Singapore-based Allozymes. His company take a different approach: using synthetic biology.

Salehian describes how his team engineer new enzymes. They start with machine learning to predict protein sequences that might fulfil a certain role, such as digesting a pollutant. The gene that codes for this synthetic protein is then engineered into bacteria and the resultant proteins are functionally screened against the intended target. Using this approach, Allozymes screens more than 20 million enzyme variants per day, according to Salehian, using a proprietary droplet-based technology.

Allozymes scientists are engineering enzymes for the cosmetics industry and to digest agricultural waste, but another focus is bioremediation, in part thanks to a chance meeting Salehian had with scientists from the Congo. These Congolese scientists described to Salehian how parts of their country have been blighted by gold mining and its environmental costs, due to the toxic chemicals (including cyanide and mercury) that miners use to extract the precious metal from its ore. This inspired Allozymes’ scientists to screen synthetic enzymes predicted to catalyze the separation of gold from its ore. The team could then use these enzymes in a column to replace cyanidation as well as to detect gold in the soil, so that mining can be more targeted, although this approach is still being tested by company scientists. Another focus for Allozymes, says Salehian, is creating synthetic enzymes that can convert used lithium-ion batteries into ‘black mass’—a valuable commodity that contains lithium, manganese, cobalt and nickel.

Allozymes has received £15 million in Series A funding in May 2024, as well as investment from the government of Singapore. Company revenue comes from three streams: designing enzymes for other companies, which can then go on to commercialize them; licensing its own bio-solutions to companies; and offering access to its growing enzyme library to others. Allozymes’ daily screening of millions of synthetic enzymes has given it the “largest enzyme data library in the world,” says Salehian, leading the Singapore Economic Development Board to dub it “the Google of enzymes.” Its library matches DNA sequence and protein structure with functional data—useful for designing bioremediation strategies and as a rich data source for machine learning.

An ocean of opportunities

While Allozymes creates new enzymes, other scientists continue to discover them in the wild. Bioinformatician Rob Finn leads BlueRemediomics, a European consortium to screen metagenomics data from the deep blue sea. Finn’s research starts with the massive sequence databases held at EMBL’s European Bioinformatics Institute (EMBL-EBI) in Hinxton, UK, including sequences taken from the sea and ocean, such as those from the Tara Oceans expedition7, which traveled to 210 sites across every major oceanic region in the world and discovered more than 40 million genes along the way. Finn’s team has produced the MGnify resource, which he describes as “one of the biggest databases of proteins in the world,” with more than two and a half billion protein sequences.

Finn’s team starts with the amino acid sequence of a known plastic-degrading enzyme, such as a cutinase, belonging to a class of enzyme called α/β-hydrolases. They then search the MGnify protein database to find sequences that are similar to those of cutinases and have been found in the ocean. By comparing the 3D models of these sequences to those in the AlphaFold Protein Structure Database, developed by Google DeepMind and hosted at EMBL-EBI, the researchers can see if the sequences are predicted to fold like the known plastic eaters.

Finn’s initial bioinformatics screen yielded about 10,000 sequences, which his collaborators on the BlueRemediomics project then refined using machine learning algorithms to select for key structural features, with a few tens expressed in vitro to check if they can digest plastic. Believing that nature can always be improved upon, the team not only test the proteins expressed from naturally occurring genes but also use a low-fidelity polymerase chain reaction to generate imperfect copies of each gene; they quickly found that some of the imperfect copies were better at eating plastic. BlueRemediomics researchers can carry out “a million of those assays in a week,” says Finn, and have identified new plastic-eating enzymes, including at least one new enzyme that worked at a lower temperature—a priority for the bioremediation field (these findings have yet to be published in a peer-reviewed journal).

As well as searching for plastic eaters, BlueRemediomics researchers look for microbial enzymes that can digest bone (a major waste product from aquaculture), pesticides, antibiotics and hormones, all of which are increasingly found in water. As part of this project, in 2023 Finn and his colleagues released the freely available MGnify Genomes resource of over 300,000 microbial genomes8, including many genes from bacteria that have never been cultured.

Once a pollution-eating enzyme has been discovered, the end product must be carefully checked, cautions Finn, as the enzyme could “just produce another derivative of the pollutant that is just as bad or even worse.” Some bioremediation processes involve multiple enzymes in the same bacteria, or even different species of bacteria working together, and so isolated enzymes might be less effective than the whole microbe.

BlueRemediomics is funded by the European Commission and is collaborating with several companies to commercialize its discoveries, including aquaculture companies and an agri-tech company.

The potential of pellets

Compared to traditional chemical remediation, “bioremediation is typically low cost, less maintenance and less efficient,” says Susie Dai, an environmental hazard researcher at Texas A&M University in the United States, who also leads public health surveillance for the Iowa State Hygienic Laboratory. Dai has screened hundreds of fungi9 for their ability to remove microplastics from water. These small particles of plastic are shed from bottles and clothes and have been linked to some negative health effects10 in humans and animals.

Dai found three promising microbial species, including two newly discovered strains of white rot fungi, that can quickly absorb microplastics (as well as the even smaller nanoplastics) made from polystyrene and polymethyl methacrylate—both of which are commonly found in water. Unlike cutinase-containing fungi, white rot fungi do not digest the plastic but store it in internal pellets. Pelleting “is a low-cost method,” says Dai, as the fungi could be easily added to water treatment plants and then removed after their meal of plastic.

Not-so-forever chemicals

Plastic pollution is a priority for governments and funders, but other pollutants might have more immediate impacts on human health. Of these, some of the most concerning for public health are the more than 9,000 varieties of PFAS. These are found in a wide range of household items (including non-stick frying pans), are toxic at trace levels and are known as ‘forever chemicals’ owing to their long life and stability, due to their strong carbon–fluorine bonds. PFAS can leach into water sources and so need to be removed via concentration using expensive means such as reverse osmosis or ion-exchange resins, followed by burning.

Researchers are racing to find microbes that can consume PFAS and either digest it or pellet it for later removal. Although PFAS-eating microbes have been found, bioremediation has been too slow to be practical. So, Dai’s team hit on a different approach, but not one using microbes. They instead turned to a type of organism used in bioremediation for many years: plants11.

Scientists regularly use plants to clean waterways and soil, including planting sunflowers at Chernobyl to extract radioactive cesium and strontium. Dai and her team take a different approach, using a ‘nano-framework’ of artificial plants and natural fungi. They engineered the artificial plants from low-cost cellulose and lignin, which form an amphiphilic environment that attracts and concentrates PFAS. Their artificial plant also conveniently provides a home for bacteria and white rot fungi, which can digest the pollutants. Dai’s bioengineered processing factory could provide a low-cost route to removing PFAS from water—now a priority for US water companies since the US Environmental Protection Agency (EPA) announced in April 2024 legally enforceable maximum levels for six different PFAS in drinking water.

Plants are well equipped to deal with contaminants that bacteria are unable to digest, such as antibiotics. Alongside pesticides and hormones, antibiotics are considered “emerging contaminants,” says Byong-Hun Jeon, an environmental engineer at Hanyang University in Seoul, South Korea, who uses plants and microalgae to decontaminate water. Antibiotics are naturally produced by microorganisms to kill bacteria, so bacteria themselves are not a useful tool to remove them, says Jeon.

Levels of antibiotics in water are often below the threshold set by regulatory agencies, such as the EPA, but low concentrations may still influence the environment, including fish, according to Jeon. His research has identified plants and microalgae that can naturally remove antibiotics and other emerging contaminants from water, and so can be planted in wetlands to gradually remove the pollutants.

Like Dai’s team, Jeon focuses on networks of organisms: plants, algae and bacteria working together in a natural bioremediation factory. Plants metabolize the antibiotics and then the constituent parts are degraded by bacteria, says Jeon. Natural plants and microalgae are useful bioremediators, partly because of cost but also because of the relative lack of regulatory hurdles. Like McCarthy, Jeon was inspired by nature—in his case the algal blooms seen in polluted rivers in Seoul, which he reasoned must be feeding on the pollutants. “Algae in the natural environment might be helping the removal or reduction or degradation of emerging contaminants,” says Jeon.

Dye is another emerging contaminant in need of removal, especially in India, where decades of industrial textile production have contaminated many water sources. The Indian textile industry is worth 4% of GDP and employs millions of workers, so bioremediation strategies are a priority for the Indian government, which commissioned a report on the topic in 2019.

Today, remediation of textile dyes is a required step before effluents can be disposed of, usually into rivers and seas. Chlorella species of microalgae can be used for dye removal12 because of their azoreductase enzymes, which turn the dye into a harmless biomass. Jeon’s research showed that the microalga Chlamydomonas mexicana can decolorize and biotransform three dyes: Red HE8B, Reactive Green 27 and Acid Blue 2913. Bioremediation was boosted further when the microalgae was combined with activated sludge—a culture of microbes generated during the treatment of sewage or other organic waste. Activated sludge is widely used for wastewater treatment around the world, but Jeon and other environmental engineers are experimenting with mixing the sludge with specific microbes to boost its effects. With the global textile industry projected to be worth over $750 billion by 2027, such dye-eating microbes could be big money if commercialized.

Genetically engineered soil

The future of bioremediation might lie in editing microbial communities directly in a contaminated site, so that they can remove a pollutant. “Editing communities, I think, will take place in the environment one of these days,” says Banfield, who expects environmental trials of gene editing to take place within 5 years—although regulatory hurdles remain: “Genome editing of microbial communities is a big ask and we are currently lacking the regulatory infrastructure and guidelines,” she says.

Some bacterial species will perhaps never be cultured outside their natural environment, as they are obligate symbionts, relying on their community neighbors to grow. Banfield worked with CRISPR discoverer Jennifer Doudna to design a technique to engineer these unculturable bacteria using a transposon-based CRISPR–Cas gene editing tool that they call DART (for DNA-editing All-in-one RNA-guided CRISPR–Cas Transposase)14.

Their new approach introduces a randomly integrating mariner transposon through conjugation, electroporation or natural DNA transformation; this provides a footprint for shotgun metagenomics sequencing. Newly identified bacteria can then be edited using a CRISPR-Cas Tn7 transposase. The approach is at proof of principle for now, but it could be used to engineer bacteria to remove contaminants from soil.

Beyond the regulatory challenges, more research is needed into microbial communities and the techniques to edit them, says Banfield, as microbiome-editing technology “has a long way to go to be effective enough to have an outcome that causes real change.”