Buffer components (K₂HPO₄ and KH₂PO₄, Trizma base, and NaCl), bovine serum albumin (BSA), and p-nitrophenol butyrate (p-NPB) were purchased from Sigma-Aldrich (USA). Amorphous polyethylene terephthalate (PET) and crystalline PET (crystallinity >50%, size<300 µm) were purchased from Goodfellow. Polybutylene adipate terephthalate (PBAT) was purchased from BASF, and polylactic acid (PLA) and mixed waste containing PET (MW) were provided by AIMPLAS. Lysogeny broth (LB), kanamycin, and isopropyl-β-D-thiogalactoside (IPTG) were purchased from Sigma-Aldrich (Germany), while Escherichia coli BL21-Gold (DE3) competent cells were purchased from Agilent.

Chemicals for fermentation media and semisynthetic medium were purchased from Sigma-Aldrich (Germany) where not otherwise specified.

Polyester films were sized to 0.5 cm × 1 cm, washed in Triton X-100 (5 g/L) and sodium carbonate (100 mM), and rinsed with ultrapure water. Additionally, the same polymers were cryomilled with two cycles of 1:30 min each with a frequency of 30 s⁻¹, with a 1 min at 5 s⁻¹ cooling step in between. Powders were afterward homogenized with a sieve of 500 µm pore size. Crystalline PET was already available as a powder with <300 µm size.

The protein concentrations of both the HiC and eluted Thb fractions were estimated through a Bradford-based Bio-Rad Protein Assay (Bio-Rad Laboratories GmbH, Munich, Germany) using bovine serum albumin (BSA) as the standard. The activity of HiC was measured in the final buffer used for polyester hydrolysis (1M KPO), while Thb activity was characterized in different buffers and molarities. The substrate for all the assays was p-nitrophenyl butyrate (p-NPB), and the volumes in use were kept constant for all the samples: the enzyme was diluted accordingly into the buffer of analysis (at set pH and molarity). The volume of each sample was 20 μL, placed in triplicate in a 96-well plate. A 200-μL aliquot of p-NPB containing solution (obtained diluting the substrate in the buffer of interest) was added to the samples and to a triplicate set of blanks (buffer only), and the increase of absorbance was recorded for 5 min at 405 nm on a plate reader (Tecan infinite M200, Tecan Austria GmbH). The same measurement was repeated after dilution in pre-heated buffers (ranging from 50°C to 60°C–70°C). The results were expressed in units (U) per mL, where U corresponds to the amount of enzyme that catalyzes 1 µmol of substrate per minute. Thermostability was also evaluated through activity assessment for 72 h (with time points at 0 h, 6 h, 24 h, 48 h, and 72 h) to evaluate the loss of activity in a real incubation set-up, in the buffers in which the activity was measured as the highest. The values obtained through the p-NPB assay after each time point were measured and compared to the assessed initial one (indicated as 0 h) and assumed as 100%. These measurements were performed on enzyme fractions incubated at 30°C, 50°C, and 70°C.

Polyester powders were incubated in 2-mL Eppendorf tubes with a total reaction volume of 2 mL, where Thb or HiC was diluted to 1 µM final concentration. The films were sized to be 1 cm × 0.5 cm, while 10 mg of each powdered polymer was incubated in the same buffer volume. The buffer in use was 1 M KPO pH 8. The Eppendorf tubes were placed horizontally in an orbital shaker incubator at 70°C and 150 rpm shaking. Hydrolysis of the powder polymers was performed also with Phl7, expressed and produced in E. coli as described in (Fohler et al., 2024).

After incubation of the polyester samples, they were washed in the three subsequent steps as summarized for the preparation prior to hydrolysis to remove any possible attached residuals. The polymers were dried and weighed to compare with the original weight and to calculate their mass loss, which was reported as a percentage. The released monomers were quantified by using high-performance liquid chromatography (HPLC) in two different settings. An Agilent Technologies 1260 Infinity instrument was used in combination with a C18 reversed-phase column. Terephthalic acid (TPA) was detected via a UV lamp, and the method used was based on a gradient of methanol and 0.1% formic acid (flow 0.85 mL/min), with an injection volume of 10 µL. The samples were diluted in methanol and acidified with 30 µL 6 N HCl before being centrifuged at 4°C (15 min and 14,000 rpm) and filtered through 0.2-µm polyamide filters into HPLC vials. All other monomers were detected through their refractive index (using a Transgenomic IC SEP-ION-300 and a mobile phase of 0.01 N H₂SO₄, with a flow rate of 0.325 mL/min, 45°C). The Carrez protocol was used to remove proteins through the addition of potassium hexacyanoferrate (II) trihydrate and zinc sulfate heptahydrate reagents. The samples were centrifuged for 30 min at 4°C and 14,000 rpm before filtering into HPLC vials through 0.2-µm polyamide filters. Calibration curves were prepared with serial dilution of all monomers (individually either in buffer or ultrapure water according to each monomer detection protocol) and used to estimate their concentrations in the hydrolysates.

A homology model of Thb was created with the Swiss model server (Waterhouse et al., 2018) in the automated mode. The crystal structure of PHL7, entry 8BRA in the protein databank, with a sequence identity of 90%, was used as a template; see Supplementary Figure S1 in the supporting material for the sequence alignment. The obtained model showed a QMEANDisco score of 0.87 (Studer et al., 2020). An Alphafold 3 (Abramson et al., 2024) model for Thb was also reported. The resulting model was compared to the crystal structure of HiC, with PDB code 4OYY, by aligning the catalytic triads of both models.

Affinity-based purification by an N-terminal Strep-tag yielded a pure protein with a molecular weight of 35 kDa according to SDS-PAGE analysis (Supplementary Figure S2 (ESI)). While the small-scale culture (total volume 200 mL in shake flasks) gave 2.64 mg_enzyme/L_culture or alternatively 0.27 mg_enzyme/g_biomass, this yield was significantly improved when the culture was upscaled to the 2 L bioreactor with overall better control of the parameters and optimized media component (Supplementary Tables S1–S3). The final purified enzyme was, in fact, 33 mg_enzyme/L_culture or 0.84 mg_enzyme/g_biomass from the bioreactor cell biomass, proving that the upscale was not only possible but also allowed higher final yields.

The specific activity on p-NPB as a substrate (1 M KPO buffer, pH 8, T 70°C) was 54.1 U/mg. The activity based on the same model substrate (p-NPB) was also compared in various incubation conditions regarding buffer type (Tris HCl and KPO), molarity, temperature, and pH. The most suitable environment proved to be KPO, for which the highest activity of 152.5 U/mg was found on 100 mM KPO pH 8 and 60°C (Supplementary Figure S3, ESI). The better results for KPO than Tris HCl confirmed the previous finding of a putative inhibitory effect of Tris HCl on enzyme activity in favor of the adoption of KPO, whose phosphate ions were expected to stabilize the enzymes (Schmidt et al., 2016; Gamerith et al., 2017; Weinberger et al., 2017).

Thermostability at specific temperatures and conditions was tested in the next step. In agreement with previous results (Daniel and Danson, 2013; Schmidt et al., 2016; Ismail et al., 2024), enhanced stability was seen at the highest KPO molarity of 1M (50°C: 10.8 U/mg; 60°C: 48.7 U/mg; 70°C: 54.1 U/mg), while a reported large stabilization effect of Ca²⁺ (Sonnendecker et al., 2022) could not be confirmed when CaCl₂ was added to a final concentration of 2 mM.

In addition, the p-NPB assay was also carried out under the incubation conditions used for polymer hydrolysis (1 µM concentration in the presence of PET stripe (1 cm × 0.5 cm) with time points at 6 h, 24 h, 48 h, and 72 h (Supplementary Figure S4A). Additionally, 30°C and 50°C were considered to monitor the effect of temperature on the long-term stability (Supplementary Figures S4B, 4C). Thb and HiC followed a similar trend regarding stability. After 6 h of reaction, the activity on p-NPB decreased by more than 50%, while at the end of the incubation (72 h), the residual activity was 20% for Thb and <5% for HiC. Lower temperatures also caused significant activity loss (Supplementary Figures S4B, 4C). In particular, while the stability of HiC at 50°C was retained >10% for the whole experiment (owing to various additives and stabilizers in the commercial formulation), Thb at the same temperature was barely active at 48 h, maintaining a low basal activity until 72 h when compared with the initial one. These results confirmed the destabilization of the enzyme in buffer at high temperature for longer incubation time, which is a common response of enzymes to heat. However, a sharp decrease in the residual activity in the reaction mixture may also indicate partial adsorption of the enzyme on the present PET. Moreover, it must be taken into account that the activity assay on p-NPB was carried out on an aliquot of the soluble fraction of the enzyme, which only estimated the actual available free enzyme and could be affected by several parameters throughout the duration of the incubation. Therefore, in the next step, the activity of the polymer was assessed via weight loss and released monomer.

The specificity of Thb on a range of aromatic or aliphatic polyesters was compared to the H. insolens cutinase, which has been well studied for polyester hydrolysis. Previously (Marten et al., 2005; Gigli et al., 2019), a higher susceptibility to enzymatic hydrolysis was reported for aliphatic polyesters than the aromatic polyester. Weight loss and soluble molecules released from the polyesters were quantified to assess whether Thb followed this hypothesis.

The overlay of the HiC structure with the one from Thb displayed significant differences between the two enzymes, offering a structural interpretation of their different specificity on aromatic and aliphatic polyesters. Strikingly, the HiC catalytic triad (Ser105, Asp160, and His173) aligned almost perfectly with that from Thb (Figure 3A). The Thb ɑ/β-hydrolase fold, in which the catalytic triad Ser131-His208-Asp177 is located, is conserved in other PET hydrolyzing enzymes. S131 provides the nucleophile site, while the acid D177 stabilizes the base H208 (Han et al., 2017) (Figure 3B). The Alphafold 3 model was virtually identical to the homology model, giving a root-mean-square deviation after an alignment of 0.16 Å (Supplementary Figure S15).

Serine hydrolases generally possess a cleft that determines substrate specificity. A surface representation is displayed in Figures 3C, D, where aromatic groups (Phe, Tyr, or Trp) are shown in magenta. Thb (Figure 3C) displays a shallow but wide cleft with a larger number of aromatic groups than HiC (Figure 3D). A wide and shallow but aromatic cleft can explain the preference for aromatic substrates of Thb, while the more pronounced cleft of HiC could preferentially accommodate less bulky, aliphatic substrates.

The co-crystallization of PHL7 with TPA was already carried out and demonstrates the interaction between the enzyme active site and its substrate. I179, W156, D177, H209, A131, F63, and M132 are responsible for the direct interaction of the enzyme with TPA (Richter et al., 2023). These are well-conserved in the Thb structure; therefore, the hydrolytic activity is not disrupted by the slight divergence between the two homologous proteins.