Labrys portucalensis: a newly described PFAS degrading strain

We started the conceptualization of PFAway by getting to know the state-of-the-art of microbial/enzymatic PFAS degradation from the literature and previous iGEM projects. It was clear that no PFAS catabolism or efficient enzymatic degradation had been described yet.

In fact, enzymes were often identified as active on mono/difluorinated substrates (Wackett and Robinson, 2024), but long-chain PFAS, such as PFOS and PFOA harboring 17 and 15 fluorine atoms, respectively, are too complex substrates for these enzymes. Nonetheless, several articles reported microbial PFAS degradation (Smorada et al., 2024). We identified several strains that had the potential for efficient PFAS degradation, as described in Table 1.

Table 1: Comparison of PFAS degradation by different bacterial strains

In the majority of bacterial strains investigated, PFAS degradation is fast and efficient. However, products of degradation have not been clearly identified, and when detected, they primarily correspond to fluoride ions and long-chain intermediates (like PFHpA for Pseudomonas plecoglossicida 2.4-D). This lack of knowledge limits the interest of such strains as the fate of PFAS remains unresolved, which is not suitable for clean water release.

In contrast, the products of degradation of PFOS by Labrys portucalensis have been characterized, and the authors who identified the strain proposed a degradation pathway resulting in PRPrA (clearly identified) and TFA (suspected) (see supplementary Figure S1). The extent of chain shortening observed in this organism is particularly significant, as it opens the possibility of coupling its metabolism with complementary systems capable of defluorinating the resulting short-chain PFAS.

+ Figure S1

In addition, L. portucalensis was grown in conditions that can easily be scalable (aerobic, with only PFOS as carbon source, in M9 minimal medium). Despite a long degradation time (Table 1), L. portucalensis seems to be the best strain to consider for long-chain PFAS degradation in contaminated water.

We therefore found the starting point of our project: L. portucalensis, a bacterial strain capable of degrading a long-chain PFAS into short-chain PFAS, as shown in Figure 1.

Characterization of L. portucalensis

1- PFOS degradation

We designed an assay to measure the PFOS degradation capacity of L. portucalensis. We used smaller volumes than in Wijayahena et al, 2025 and monitored PFOS degradation over time using LC-MS/MS (measurement performed by the ENVT, Argoul et al., 2025). We attempted to reduce the degradation time of PFOS by starting at two different cell densities (OD₆₀₀): 0.1 (as in Wijayahena et al, 2025) and 1.

For technical details on the PFOS degradation tests and on safety when working with PFOS, please go to the Protocols page and to the Safety page.

2- Strain characterization

The growth rate of L. portucalensis on glucose was evaluated in M9 minimal media. We also assessed the toxicity of TFA, PFPrA and fluoride ions (using NaF) by monitoring growth curves in the presence of these compounds. The experimental data were used to feed our mathematical model (see our Model page), while providing new insights on the properties of the strain.

For technical details on the toxicity assays, please go to the Protocols page.

We then had to find a way of defluorinating the obtained short-chain PFAS (Figure 1).

Directed evolution for improving DeHa2

Directed (or laboratory) evolution aims at accelerating natural evolutionary processes of proteins towards a user-defined goal. Mutagenesis techniques are employed to generate libraries of genes, which are then screened in order to select the most valuable variants. Several mutagenesis/screening rounds are performed, enabling to direct the evolution toward the desired protein activity.

In our case, the activity of DeHa2 on short-chain PFAS needs to be improved. We decided to apply directed evolution methods to generate DeHa2 variants and select for the ones with the highest activity on short-chain PFAS, starting with TFA and PFPrA as degradation products of PFOS. We decided to apply orthogonal replication (ORep) for generating variants of the DeHa2 gene (dehH2) coupled to a fluorescence-based droplet microfluidic method to select the best-performing variants. The whole directed evolution process of DeHa2 is represented in Figure 3.

Orthogonal replication (ORep)

This section focuses on explaining why we chose ORep for improving the activity of DeHa2 on short-chain PFAS, and how we designed the strain creation. For an in-detail description of this technique, please refer to Figure 3A.

We initially thought about classical methods for genetic diversification of DeHa2. Supplementary table S2 summarizes the properties of the ones we considered.

We first considered in vitro methods such as epPCR and epRCA. These are highly labor-intensive as they require cloning, expression, and assay of each enzymatic variant. This is why we shifted our focus to in vivo approaches, where continuous improvement of strains is possible since the variants are expressed directly by the host. In this context, we identified a recent study by Tian et al. (2024) describing orthogonal replication (ORep).

ORep is an in vivo hypermutagenesis system that uses intracellular resources to efficiently generate a library of user-defined gene variants without mutating the native genome nor interfering with the host replication machinery. Initially developed in the yeast <Saccharomyces Cerevisiae (Ravikumar et al., 2014) and later in Bacillus thuringiensis (Tian et al., 2023), ORep has recently been implemented in E. coli (Tian et al., 2024).

This technique has several advantages:
• High variant generation rate
• Fast screening without isolating or purifying the gene library
• No risk of deleterious chromosomal mutations

Establishing orthogonal replication in E. coli is a challenging task. We contacted the authors of the original paper to share with us some materials and tips but did not get an answer. Mindful of the broad potential of an orthogonal replication system, we nonetheless decided to implement it fully on our own and to facilitate its wide adoption by future iGEM teams by making all materials and protocols fully available.

We therefore have designed a kit for future users, to facilitate the cloning of their gene of interest in the linear replicon, as part of a directed evolution method. For more details, see our Contribution page.

Design of the ORep system

1- Creation of the replication operon

In Tian et al., 2024, E. coli was engineered to express a synthetic operon composed of four genes involved in the replication of phage PRD1 (Figure 5).

Figure 5: Structure of the synthetic replication operon under the control of IPTG inducible Ptac promoter. Pass the mouse or click on the different components to see their characteristics.

TP DNAP SSB DSB

Terminal protein gene: expresses the terminal protein that binds to ITR sequences, therefore functioning as origins for replication for PRD1 DNAP. PRD1 DNA polymerase gene: encodes a DNA polymerase derived from phage PRD1, for which terminal protein (TP) bound to inverted terminal repeats (ITR) naturally serve as replication origins in the phage genome.This DNAP was mutated by the authors to make it error-prone. Single-stranded DNA-binding protein gene: early-expressed gene in phage PRD1, hypothesized to be necessary for PRD1 replication. Double-stranded DNA-binding protein gene: early-expressed gene in phage PRD1, hypothesized to be necessary to build to replication.

We assembled this operon from three DNA fragments (gBlocks) following the cloning strategy illustrated in Figure 6.

With the intention to increase the mutation rate of our target gene, we chose to use a DNAP bearing two mutations which, when introduced individually, increase the polymerase mutagenic capacity. The mutation rates of the single-mutant DNAPs are provided in Table 2. The mutation rate of the double-mutant DNAP was not characterized by Tian et al.

Table 2: Mutation rates on the linear replicon of the DNAP variants. Rates are expressed in substitutions per base pair per generation (Tian et al., 2024)

The next step was the implementation of the operon in E. coli DH10B for expression.

2- Implementation of the replication operon in E. coli DH10B

The two options in hand were to insert the operon in the E. coli chromosome or to introduce it in a single-copy plasmid with which E. coli will be transformed. Discussing with Denis Jallet, we learned that chromosomic integration in the SS9 site of E. coli DH10B was commonly performed in his lab. We therefore decided to attempt chromosomic insertion of the replication operon.

	Denis Jallet is Researcher at the Toulouse Biotechnology Institute (TBI), specialized in Microbiology. He shared with us his protocols and materials for chromosomic insertion of genes in the E. coli genome, using the non-essential chromosomic site SS9.

Following Denis’ advice, we designed plasmids to implement the chromosomal insertion in the SS9 site, as illustrated in Figure 7.

Unfortunately the chromosomal insertion approach did not lead to the expected results and we set out to clone the operon in a single-copy plasmid. Our cloning strategy for implementing the replication operon on a single-copy plasmid is represented in Figure 8.

You can find out about the related DBTL cycles concerning these strategies in the Engineering page.

3- Creation of the linear replicon

Tian and coworkers proved that expression of the orthogonal replication operon led to the mutagenic replication of a linear replicon consisting of a double stranded linear DNA flanked with ITRs containing a user-defined gene (here, dehH2) and a kanamycine resistance gene (Kan^R). Figure 9 illustrates our cloning strategy for the linear replicon.

In this way, the ORep system is established: the synthetic replication operon enables the mutagenic replication of the linear replicon containing the DeHa2 gene (dehH2) and generates a library of variants. Our next task was to design a screening technique for the selection of the best variants.

Fluorescence-activated droplet sorting for the selection of DeHa2 variants

The activity of efficient DeHa2 variants generated by ORep will lead to the release of fluoride ions during short-chain PFAS degradation. We therefore chose to develop a screening technique based on the quantitative detection of the released fluoride ions using a fluorescence-based biosensor.

We first opted for the fluoride-sensitive riboswitch crcB (Baker et al., 2012), originally characterized from the Bacillus cereus crcB gene, which was shown to activate the expression of a reporter gene in response to fluoride. Inspired by the work of team SDU Denmark 2023, we aimed to adapt this strategy to control sfGFP expression, unsuccessfully. Therefore, we set out to implement another fluoride biosensor, named FluorMango. The engineering cycles for these strategies are further discussed in the Engineering page.

FluorMango is an RNA aptamer allowing for the direct quantification of fluoride by fluorescence (Husser et al., 2023). The structure and principle of this aptamer is presented in Figure 10.

When combined with the directed evolution cycle of DeHa2 (Figure 3), FluorMango serves as a reporter of the activity of the different DeHa2 variants expressed in E. coli ORep cells: the more efficient the degradation of short-chain PFAS, the higher the concentration of fluoride ions released in the external medium, and the higher the fluorescence intensity of FluorMango.

Importantly, because FluorMango cannot permeate the cell membrane, the ORep cells are individually encapsulated into microfluidic droplets to create a linkage between the gene variants and the phenotypic output, here the fluorescence intensity of FluorMango. Microfluidic droplet sorting is then employed to selectively isolate droplets exhibiting the highest fluorescence, thereby enriching for the most active variants. These enriched populations are subjected to iterative rounds of screening to progressively enhance catalytic performance. An overview of the directed evolution workflow is depicted in Figure 3B.

Proof of concept of the FluorMango screening method

1- Production of FluorMango

We chose to produce the FluorMango aptamer by annealing DNA primers followed by transcription with T7 RNA polymerase (Figure 11).

2- Assessing FluorMango’s sensitivity

According to Husser et al., 2023, the FluorMango aptamer shows a linear response to fluoride concentrations within a range of 150 µM to 1 mM, with a limit of detection of 68 µM and a limit of quantification of 90 µM. Therefore, we tested the response of the FluorMango aptamer in microplates across a fluoride concentration range of 0.1 mM to 3 mM, using NaF to generate the desired F^- concentrations.

To find out about the results of FluorMango tests in microplates, please go to our Best Measurement page.

3- Assessing FluorMango’s specificity

According to Husser et al., 2023, the FluorMango aptamer displays a remarkable specificity for fluoride as no fluorescence was observed in the presence of other halide ions. To ensure that the fluorescence detected is only due to the presence of fluoride ions under our conditions, we performed other negative controls with the aptamer in microplates described in Table 3.

Table 3: Negative controls performed with the FluorMango aptamer in microplate conditions..

For more information about the results of FluorMango tests in microplates, please go to our Best Measurement page.

4- Monitoring Defluorination Activity Using FluorMango

In contrast to Husser et al., 2023, we used 2-fluoropropionic acid instead of fluoroacetate as a monofluorinated substrate of DeHa2, due to its lower toxicity. According to Dodge et al., 2024, the Delftia acidovorans defluorinase DeHa2 efficiently degrades 2-fluoropropionic acid, with a specific activity of 1.7 ± 0.2 µmol·min^-1·mg^-1 as shown in Table S1. We therefore measured the real-time production of fluoride ions from 2-fluoropropionic acid by DeHa2 expressed in E. coli DH5α and induced with IPTG.

To find out about the results of FluorMango assays in microplates, please go to our Best Measurement page.

5- Enrichment of Fluoride-containing droplets with microfluidics

We designed a mock library experiment, whereby samples containing either FluorMango without NaF (negative control) or FluorMango with NaF were incubated for 1 hour. Then, droplets were generated from each sample, mixed to produce a population containing 20% of NaF-loaded droplets, and subjected to microfluidic sorting. Widefield fluorescence microscopy was used to observe droplet stability and sorting efficiency.

To learn more about the results of microfluidics assays, please visit our Best Measurement page. All protocols are accessible on our Protocols page.

Upon discovery of an active dehalogenase for the degradation of short-chain PFAS, a large amount of fluoride ions, beyond the toxicity threshold, will be released within the bacterium, here Pseudomonas putida. Consequently, the bacterial resistance to fluoride must be improved.

Improving P. putida’s response to fluoride stress

Pseudomonas putida KT2440 expressing the evolved DeHa2 will be exposed to large amounts of fluoride ions. We first thought that the natural resistance of P. putida could support fluoride resistance, but our modeling approach proved the contrary. Therefore, its tolerance to fluoride stress must be enhanced. You can find out about this engineering cycle on the Engineering page.

Defluorination of monofluorinated compounds by a recombinant DeHa in P. putida releases up to 50 mM of F^- in the extracellular medium (Dodge et al., 2024). The minimal inhibitory concentration (MIC) of fluoride ions for P. putida KT2440 in M9 minimal medium being 75 mM (Calero et al., 2022), we understand the importance of enhancing the fluoride stress resistance capacity of our chassis, especially if it comes to degrade multifluorinated compounds with the improved DeHa2.

We chose P. putida as a chassis for the DeHa2 expression because it is a widely studied genus for broad stress tolerance tests, harboring a particularly efficient redox metabolism (Martínez-García and de Lorenzo, 2024). Moreover, genes expressed in response to fluoride stress were identified for strain KT2440. fluC, a gene coding for a fluoride specific transporter, was identified as the main feature expressed during fluoride exposure (Calero et al., 2022).

FluC proteins are dimeric transmembrane transporters, highly specific for F^- anions and playing a major role in fluoride stress resistance (Stockbridge and Wackett, 2024). Their efflux rate stands at 10⁶ ions/s (Khusnutdinova et al., 2023), and the deletion of fluC in P. putida results in a highly F^--sensitive strain with a MIC of 0.5 mM of F^- in M9 minimal medium (~100 fold lower than for the WT KT2440 strain) (Calero et al., 2022).

Since the literature indicated a strong link between FluC’s expression and P. putida’s sensitivity to F^- ions, and as fluC seemed to be under fluoride sensitive epigenetic control, we decided to constitutively overexpress this transporter as an attempt to enhance the resistance of P. putida to F^- ions.

This enabled us to achieve an end-to-end degradation process of PFAS, as represented in Figure 12:

Overexpressing FluC in P. putida KT2440

1- Creation of pSEVA438-fluC

We constructed a plasmid named pSEVA438-fluC that contained the fluC gene directly amplified from the genome of P. putida KT2440, under the control of a m-toluic acid inducible promoter (P_m). The cloning strategy of pSEVA438-fluC is depicted in Figure 13.

2- Testing of P. putida WT against P. putida pSEVA438-fluC

P. putida KT2440 was then transformed with pSEVA438-fluC. Overexpression of fluC was induced in the modified strain without any fluoride sensitive upstream regulation mechanism.

MIC assays and growth curves analysis were conducted using varying concentrations of NaF in microplates for both KT2440 WT and KT2440-pSEVA438-fluC strains. We also conducted toxicity assays for TFA and PFPrA on KT2440-pSEVA438-fluC for further characterization of the strain.

For more details about the protocols of toxicity assays, please visit our Protocols page.

Entrepreneurship design

Entrepreneurship involves translating our PFAS remediation solution into real-world applications by integrating it into water treatment systems. We therefore needed to think of an industrial approach to effectively combine our two strains L. portucalensis and P. putida (Figure 12).

Our initial idea was to grow the two strains as a co-culture in a bioreactor for the degradation of PFAS contained in contaminated water. However, this approach presented several difficulties:

1- Even though the strains can be cultivated in the same medium and growth conditions, they show very different growth rates, respectively 0.24 h^-1 and 0.55 h^-1 for L. portucalensis (estimated based on Wijayahena et al, 2025) and P. putida (estimated in the lab and confirmed by Calero et al., 2022). This means that P. putida would overgrow L. portucalensis, which is prohibitive for the sequential degradation of long- and short-chain PFAS.

2- Direct co-cultivation in the treated water would complicate downstream separation and risk GMO release, which is strictly legislated in France.

To overcome these challenges, we contacted companies specializing in bioremediation and strain industrialization. Our first contact was with YpHen: a French company developing bioremediation solutions for soil decontamination.

	We expressed our concerns to Carmen Mirabelli, researcher at YpHen. After several discussions, she introduced us to a method YpHen is developing to treat pollution in industrial contexts: immobilizing bacteria in biodegradable beads.

Encapsulating the two strains individually in beads would solve both problems linked with culture in a bioreactor. It allows the bacteria to work efficiently without the competition issues inherent to co-culture, and, depending on their formulation, the beads can prevent the release of bacteria into water, addressing GMO-related concerns as well. Based on these considerations, we decided to collaborate with Carmen and Yphen to produce prototype beads.

Assessing the efficiency of beads for encapsulation of our bacterial strains

With the help of Carmen, we designed experiments to test the first prototypes, focusing on PFAS adsorption, bacterial viability over time within the beads, and potential bacterial release. For more details about prototype beads testing, please visit our Protocols page.

1- Production of beads

This part of the work was carried out in Grenoble at Yphen’s facilities, where the team has the most expertise in bead production. During the process, we tested several bead formulations, varying the activated carbon content at 10% and 20% to evaluate its effect on PFAS adsorption. Beads without activated carbon were also produced to determine if any other components of the bead might contribute to PFAS adsorption.

2- PFAS adsorption on beads

Following bead production, we conducted experiments to test the adsorption of PFAS (here PFOA) on beads containing 10% and 20% activated carbon, as well as on beads without activated carbon. These experiments are essential to inform the optimization of bead formulations for the next stages of development.

3- Strain viability in the beads

For the beads to effectively degrade PFAS, it is crucial that living bacteria remain active throughout the treatment period, which can be estimated using our model (see our Model page). We performed viability tests over 15 days to confirm that the bacteria remain active over this period. If bacteria survive for longer durations, the beads could be reused, reducing both the ecological footprint of the solution and its operational costs.

4- Bacterial release in water

Biosafety was a major focus, as highlighted by our human practice survey (see HP page). It was essential to ensure that the bacteria, especially the genetically modified P. putida, could not be released from the beads into the treated water.

	Raphaëlle Aubert is a French Data journalist at Le Monde. While describing our project, her first question was if engineered bacteria would be released in water after treatment. This highlighted the concern about public apprehension for GMO containment.

To evaluate this, we conducted experiments measuring bacteria release in demineralized water. Future experiments will test beads under conditions closer to real-world water systems. Measures to minimize GMO release have already been planned, mainly through improvements in bead formulation and production processes.