Polyethylene terephthalate (PET) and polybutylene adipate co-terephthalate (PBAT) were purchased from Macklin (Shanghai, China). Bis (2-hydroxyethyl) terephthalate (BHET) was purchased from Aladdin (Shanghai, China). Terephthalatebutanediol monoester (BT) was purchased from Aikon Biopharmaceutical R&D Co. Ltd (Jiangsu, China). Mono-(2-hydroxyethyl) terephthalic acid (MHET) was synthesized according to the protocol by Palm et al. (Palm et al., 2019). PET and PBAT semicrystalline films and microplastics samples were prepared according to the previously reported methods (Cui et al., 2021).

The full-length gene of jmPE13 and jmPE14 without the signal peptide sequence was codon-optimized (Supplementary Table S1), synthesized, and cloned into the pET-32b expression vector, by GENEWIZ (Suzhou, China). The mutants were generated by using the site-directed mutagenesis on wild-type jmPE13. The site-directed mutation of jmPE13 was carried out by whole-plasmid PCR using the primers listed in Table 1. The PCR products were digested with DpnI to remove the parent plasmid and purified with a PCR purification kit. The construct was transformed into E. coli BL21 (DE3) to express the protein. LB liquid medium containing ampicillin (100 μg/mL) was inoculated with a starter culture. Cultures were grown at 37°C until the OD₆₀₀ was approximately 0.6. Isopropyl-beta-D-thiogalactopyranoside (IPTG, 0.1 mM final concentration) was added to induce protein expression. Then they were incubated overnight at 16°C with shaking at 200 rpm. Cells were harvested by centrifugation for 10 min at 8,000 g at 4°C, and then suspended in 50 mM Tris-HCl (pH 8.0). After sonication, the cell suspension was centrifuged (4°C) at 12,000 g for 20 min, and the supernatant was subjected to nickel-chelating chromatography.

When using p-nitrophenol caprylate (pNP-C8) as substrate, the reaction system contained 50 mM Tris-HCl buffer at pH 8.0, 10 mM pNP-C8, and 0.2 mg/mL enzyme protein. The release of p-nitrophenol was recorded by measuring the absorbance at 405 nm in a Multiskan GO microplate reader at 37°C. When using PET or PBAT semicrystalline microplastics as substrate, the reaction system contained 50 mM Tris-HCl buffer at pH 8.0, 1 mg/mL microplastics, and 0.2 mg/mL enzyme protein. The mixture was incubated at 30°C with shaking at 1,000 rpm and the products were analyzed by high-performance liquid chromatography (HPLC). All experiments were performed three times and corrected for the subtraction of substrate self-decomposition, that is, buffer without enzyme protein as a control.

The effect of pH and temperature on the activity of the enzyme (0.2 mg/mL) was tested using pNP-C8 (10 mM) as substrate. The effect of pH on enzyme activity was determined by measuring the activity at a pH ranging from 4.5 to 9.2. 50 mM citrate-sodium citrate buffer was used for pH 4.0-6.6, 50 mM phosphate buffer was used for pH 6.6-7.8, and 50 mM Tris-HCl buffer was used for pH 7.8-9.2. The effect of temperature on enzyme activity was examined across the range of 25°C–65°C, in 50 mM Tris-HCl buffer at pH 8.0.

Thermal inactivation of the enzyme proteins was examined at 40°C. The purified proteins (1 mg/mL in 50 mM Tris-HCl buffer at pH 8.0) were incubated at 40°C and sampled at different intervals and then cooled on ice for 10min. Their residual enzyme activities were assayed at 37°C using pNP-C8 as substrate as described above.

Kinetic parameters of the enzymes for BHET were determined in 50 mM Tris-HCl buffer at pH 8.0 containing 0.03–2.4 mM of BHET and 0.2 mg/mL enzyme protein. The mixture was incubated at 37°C for 2 h and the products were analyzed by HPLC. The kinetic constants were obtained through nonlinear regression based on the Michaelis-Menten equation.

HPLC was used to analyze the products according to the methods described previously (Tournier et al., 2020) with some modifications. Briefly, 150 μL of the sample was mixed with 150 μL of methanol and 6.5 μL of 6 N HCl and filtered through a 0.22 μm filter before running HPLC. Measurement of the products was performed using an Agilent 1,260 Infinity II equipped with an InertSustain C18 column (4.6 × 250 mm, 5 μm) with a detection wavelength of 240 nm. The column oven was held at 25°C. The mobile phase was 1 mM H₂SO₄ in water with a gradient of methanol (30%–90%) at 1 mL·min⁻¹.

The post-consumer PET bottles were cut into 6 mm slices and incubated with 0.2 mg/mL of enzyme protein or protein-free 50 mM Tris-HCl buffer as treated and control groups, respectively. After incubating at 30°C for 7 days, the slices were dried in air and coated using gold. The surfaces of the slices were observed under SEM (Hitachi S-3400N) at different magnifications.

The protein sequences of the characterized PET-hydrolases were obtained from the UniProt database (https://www.uniprot.org/uniprotkb/). The amino acid sequences were conducted with multiple sequence alignment using the Clustal Omega web server (https://www.ebi.ac.uk/Tools/msa/clustalo/) (Sievers and Higgins, 2018). The results were rendered by ESPript 3.0 (Gouet et al., 1999). The neighbor-joining phylogenetic tree was created by MEGA-X (Kumar et al., 2018), and the figure was generated by the iTOL web server (https://itol.embl.de/) (Letunic and Bork, 2007).

The homology model structures of jmPE13 and jmPE14 were created by the SWISS-MODEL web server (https://swissmodel.expasy.org/) (Waterhouse et al., 2018) using the crystal structure of PET2 mutant (PDB entry: 7ECB) (Nakamura et al., 2021) as the template. Pymol software (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC, De Lano Scientific, San Carlos, CA, United States of America) was used to view the structure and generate figures.

AutoDock 4.1 (The Scripps Research Institute, La Jolla, CA, United States) (Morris et al., 2009) was used to predict the binding modes of jmPE13 and jmPE14 with BHET. AutoDockTools 1.5.6 was used to prepare the proteins and ligands for docking procedure. Kollman charges and polar hydrogens were added. AutoGrid was used to generate the grid maps. The grid dimensions were 60 points in each dimension separated by 0.375 Å. The files were generated in PDBQT format. For the ligand, random starting positions and orientations were used. The Genetic Algorithm was used with 2,500,000 energy evaluations and a population of 300 individuals; 100 runs were carried out.

The molecular dynamics (MD) simulations were performed using Gromacs v4.5.5 (Pronk et al., 2013), with the Gromacs 96 (54a7) force field. The model structures of jmPE13 and the mutants were solvated with a three-point water model in a cubic box. Na⁺ and Cl⁻ ions were added to neutralize the charges in the system. Then, a steepest descent energy minimization was performed, followed by a 100 ps NVT and a 100 ps NPT equilibration at 300 K, and 10 ns MD simulations were performed at 300 K.

We previously isolated a bacterium strain Pseudomonas sp. JM16B3 from aquaculture water. To investigate the plastic-degrading ability of JM16B3, we treated semicrystalline PET and PBAT films with the fermentation supernatant of this strain. SEM observation showed that after 72 h of treatment at 30℃, both kinds of plastic films were damaged to a certain extent (Supplementary Figure S1), suggesting that there may be extracellular enzymes with polyester degrading activity. To discover potential polyester hydrolases, we blasted the genome of JM16B3 with the protein sequence of IsPETase. Two proteins were found to share 50% and 51% sequence identity with IsPETase and were named jmPE13 and jmPE14, respectively.

There are 298 and 296 amino acids in jmPE13 and jmPE14, respectively, containing a signal peptide (amino acids 1-23) and a typical α/β hydrolase fold domain (Figure 1A). The protein sequences of jmPE13 and jmPE14 share 88% identity. Phylogenetic analysis showed that both jmPE13 and jmPE14 belong to type II PET-hydrolases (Figure 1B). Of the characterized enzymes, they showed the highest sequence identities to type IIa PET-hydrolases (53%–61%), while their protein sequence identities to type IIb and type I PET-hydrolases were 52%–53% and 46%–49%, respectively (Supplementary Table S2). Multiple sequence alignment showed that similar to the characterized type II PET-hydrolases, the loops after α2 and after β8 in jmPE13 and jmPE14 are longer than those of type I PET-hydrolases (Figure 2). JmPE13 and jmPE14 have the typical catalytic triads of the α/β hydrolase superfamily, which are S165-H243-D211 and S163-H241-D209, respectively.

To analyze their structural characteristics, three-dimensional structural models of jmPE13 and jmPE14 were constructed using the crystal structure of PET2 (Nakamura et al., 2021) as a template. The models showed an α/β hydrolase fold with a nine-stranded β-sheet at the center surrounded by seven α-helices (Figure 3A; Supplementary Figure S2). Two disulfide bonds were formed in their structures, one near the active center, which is typical for type II PET-hydrolases, and the other at the C-terminus. A shallow cleft is formed on the molecular surface near the catalytic center of jmPE13 and jmPE14, which may be the substrate binding pocket.

To predict the possible binding modes of jmPE13 and jmPE14 to polyester plastic substrates, we performed molecular docking of the model structures with the PET model substrate BHET. As shown in Figure 3B, BHET could be well accommodated in the substrate-binding cleft. In the jmPE13-BHET complex, twelve residues (G91, Y92, L93, P125, W164, S165, M166, Y189, A191, V213, H243, and F244) formed intermolecular interactions with the substrate through hydrogen bonds, van der Waals, and Pi-Alkyl (Figure 3C). In jmPE14-BHET complex, the residues involved in substate-binding included Y90, Q122, P123, W162, S163, M164, W187, V211, H241, and F242 (Supplementary Figure S2).

To characterize the enzyme activity and catalytic properties of jmPE13 and jmPE14, we recombinantly expressed these two proteins. The genes of jmPE13 and jmPE14 (without the signal peptide) were cloned into the pET-32b vector and transformed into E. coli BL21 (DE3) for expression of the proteins. After nickel affinity chromatography, purified proteins were obtained (Supplementary Figure S3). To determine their optimal catalytic conditions, we examined the effects of pH and temperature on the enzyme activities of jmPE13 and jmPE14. As shown in Figure 4A, B, their optimal pH was about 7.8–8.0, and the optimal temperature was about 34°C–37°C. In the range of pH 7.5-8.5 and temperature 30°C–40°C, both jmPE13 and jmPE14 can maintain more than 70% of the highest enzyme activity. To investigate the thermal stability of these two enzymes, the enzyme proteins were incubated at 40°C and the residual activities were examined. The thermal inactivation curves showed that jmPE13 could maintain more than 70% of the enzyme activity within 2 h of incubation, then decreased to 50% at about 2.5 h, and almost completely lost the enzyme activity after incubation for more than 6 h (Figure 4C). The thermal stability of jmPE14 was slightly lower than that of jmPE13. Although it maintained more than 80% activity within 1 h, it rapidly dropped to less than 50% after 1.5 h incubation.

To investigate the polyester hydrolyzing activity, purified proteins of jmPE13 and jmPE14 were incubated with the slices of post-consumer PET bottles at 30°C for 7 days. The surface morphology of the PET slices was observed by SEM. Compared with the buffer control group, the surface of the PET slices was significantly damaged after being treated with jmPE13 and jmPE14 (Figure 5A). We further tested the hydrolytic activity of jmPE13 and jmPE14 on PET and its derived oligomer BHET by HPLC. The results showed that after 5 h of incubation at 30°C, jmPE13 could completely convert BHET to MHET, while jmPE14 was slightly less active and a small amount of BHET remained (Figure 5B). For semicrystalline PET microplastics, the reaction was performed at 30°C for 90 h. JmPE13 and jmPE14 hydrolyzed PET to produce BHET and MHET (Figure 5C). Since some PET hydrolases have been reported to hydrolyze PBAT, we also examined the hydrolyzing activity of jmPE13 and jmPE14 on PBAT. The results showed that, after incubation at 30°C for 48 h, both enzymes could hydrolyze PBAT to produce BT (Figure 5D). These results indicate that jmPE13 and jmPE14 have hydrolytic activity on PET and PBAT polyesters.

Although jmPE13 and jmPE14 can hydrolyze polyester plastics, their hydrolysis activity and thermal stability are still very low (Figures 4, 7). To find sites for modification to improve its enzyme activity, we performed a structural alignment between jmPE13 and the known PET hydrolase mutant LCC^ICCG (PDB entry: 6THT) (Tournier et al., 2020). As shown in Figure 6A, a significant difference is that the α2 of jmPE13 is followed by a longer loop, while the corresponding position of LCC^ICCG is an α-helix, which is consistent with the results of multiple sequence alignment (Figure 2). We targeted this loop and designed four mutants referring to the structure of LCC^ICCG (Figure 6B). In mutants M1 (R146S/P149A/Y151R) and M4 (R146S), the selected sites were mutated to the corresponding residues of LCC^ICCG, and a truncated mutant M2 (ΔS143N144S145/P149A/Y151R) was constructed based on M1. In addition, we tried to introduce a disulfide bond to reinforce this loop (M3: S143C/I154C). All mutants except M1 were solubly expressed in E. coli. (Supplementary Figure S3).

We examined the enzyme activity of the mutants, but unfortunately, both M2 and M3 showed varying degrees of decreased hydrolytic activity for PET and PBAT compared to jmPE13 (Figures 7A, B). M3 was almost completely inactive on PET. However, M4 showed a significant increase in hydrolytic activity for both polyesters. Moreover, this increase in activity became more pronounced when the reaction time was prolonged. After the reaction at 30°C for 90 h, the activity of M4 against both PET and PBAT reached about 3 times that of jmPE13 (Figures 7C, D). The improvement of PET hydrolytic activity of M4 was mainly due to the increase of BHET products, while the production of MHET was comparable to that of jmPE13. The enzymatic kinetic parameters of jmPE13 and M4 on BHET were determined. The maximum reaction rate (V_max) of M4 (5.22 ± 0.17 μM/min) was higher than that of jmPE13 (4.09 ± 0.5 μM/min); however, the K _m value of M4 (0.58 ± 0.05 mM) for BHET was also higher than that of jmPE13 (0.39 ± 0.07 mM), indicating that the increased activity of the mutant was accompanied by a decrease in its affinity for this small substrate. To investigate the thermostability of the mutants, each enzyme protein was incubated at 40°C for a certain time and their residual activity against PBAT was examined (Figure 7E; Table 2). The results showed no significant change in the thermal stability of M2 and M3 relative to the wild-type, with a slight increase and decrease, respectively. However, the thermal stability of M4 has been significantly improved, and its thermal inactivation half-life at 40°C is about 1.5 h longer than that of jmPE13, which is about 1.5 times that of jmPE13.

To understand the molecular basis of the enhanced stability and enzymatic activity of M4 by the R146S single point mutation, we performed MD simulations of jmPE13 and M4. The results showed that this mutation resulted in a decrease in root mean square fluctuation (RMSF) of a loop near the catalytic center and a C-terminal loop, indicating a decrease in the flexibility of these two regions (Figure 8A). The overall rigidity of the molecule was also increased as indicated by the change in root mean square deviation (RMSD) (Figure 8B). We speculate that the increased activity of M4 may also be due to its improved stability, which makes its activity decline more slowly during the reaction period. This is also consistent with the phenomenon that the increase in activity is more pronounced with prolonged reaction time.