A 786 base pair (bp) gene fragment encoding MtCut (GenBank accession: SJZ42839) without the N-terminal signal peptide was amplified from genomic DNA of M. thermotolerans DSM45154. Using seamless cloning (Transgen Biotech, Beijing, China), the gene fragment was cloned into the pET22b (+) vector (Novagen, Madison, WI, United States) between the NdeI and XhoI restriction sites, along with a C-terminal hexahistidine (6 × His) tag. The construct was verified by DNA sequencing. Recombinant protein was produced in Escherichia coli BL21 (DE3) (Novagen) grown in Luria Broth media containing 100 μg/ml ampicillin at 37°C until the optical density reached 0.6–0.8 at 600 nm. Protein expression was then induced by adding 0.5 mM isopropyl β-d-1-thiogalactopyranoside (Sangon, Shanghai, China) at 16°C for 16 h. Cells were harvested by centrifugation at 5,000 × g for 10 min, and then the pellet resuspended in 20 mM Tris-HCl (pH 8.0), 500 mM NaCl, and 5 mM imidazole. Cells were disrupted by sonication on ice, and cell debris removed by centrifugation at 10,000 × g for 30 min. The clarified lysate was loaded onto equilibrated Ni-NTA resin (Qiagen, Hilden, Germany). After rinsing in binding buffer (20 mM Tris-HCl (pH8.0), 500 mM NaCl, 5 mM imidazole) and washing buffer [20 mM Tris-HCl (pH8.0), 500 mM NaCl, 20 mM imidazole], recombinant protein was eluted using 20 mM Tris-HCl (pH8.0), 200 mM imidazole. The proteins were further purified by AKTA primer plus system with a HiPrep DEAE FF 16/10 column (GE Healthcare, Chicago, United States) over a linear 0–1.0 M NaCl gradient. Purified proteins were concentrated using an Amicon-Ultra-15 device (Molecular Weight Cut-off = 10 kDa, Merck-Millipore Co., United States). Protein purity was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and protein concentrations determined by absorbance at 280 nm using the molar extinction coefficient calculated from amino acid sequence composition (https://web.expasy.org/protparam/).

All reactions were performed in 96-well plates in 200 μl total volume. We tested para-nitrophenyl (pNP) esters (Aladdin, Shanghai, China) with chain lengths of C₂, C₄, C₆, C₈, C₁₀, C₁₂, C₁₄, C₁₆, and C₁₈ as cutinase type substrates. To 190 μl 20 mM Tris-HCl (pH8.0) buffer plus dissolved pNP esters, we added 10 μl purified MtCut (5 μg/ml). Reaction times ranged from 10–30 min at 30°C, and plates were measured in triplicate. The p-nitrophenol product was measured at 405 nm using an EnSight™ Multimode microplate reader (PerkinElmer Inc., Spokane, WA, United States). One unit of enzyme activity was defined as the amount of enzyme required to convert 1.0 μmol p-nitrophenol per min under standard conditions. Data, v _o (mM/min) versus [S]₀ (mM) were fitted to a Michaelis-Menten model to calculate kinetic parameters. To determine the optimal temperature and pH, reaction samples using p-NP butyrate as substrate were incubated in the temperature range 5–70°C, and pH range 7.0–9.5, respectively. Enzyme thermal and pH stability parameters were measured after a specified incubation period at a constant temperature and pH range. Residual activity was determined after 10 min incubation at 30°C. The effects of calcium ions and EDTA on enzyme properties were determined by adding calcium ions or EDTA at a final concentration of 1 mM. We measured the effect of final calcium ion concentration ranged from 0 to 500 mM on the cutinase activity.

We estimated the thermal characteristics and crystallinity of PET using differential scanning calorimetry (Netzsch DSC 214, Bavaria, Germany) with a heating rate of 10°C/min in a nitrogen environment. Crystallinity (θ) calculations were based on the following equation (Tournier et al., 2020): θ = ( Δ H f − Δ H c c ) Δ H f 100 % × 100 % Where ΔH _f is the melting enthalpy (J/g), ΔH _cc is the enthalpy of cold crystallization (J/g), ΔH _f 100% is the theoretical melting enthalpy of 100% PET crystallization with a value of 140.1 J/g. PET microparticle (KAI YUAN Plastication Technology Co., Dongguan, China) has a melting temperature (T _m) of 247.2°C, a crystallization temperature of 173.7°C, a glass transition temperature (T _g) of 79.1°C, a melting enthalpy (ΔH _f) of 59.78 J/g, and with no observed enthalpy of cold crystallization (ΔH _cc), thus the crystallinity percentage is 42.67%. PET low crystallinity (lc-PET) films (Goodfellow Ltd., Bad Nauheim, Germany) have a T _m of 247.5°C, a crystallization temperature of 173.8°C, a T _g at of 71.9°C, a ΔH _f of 37.86 J/g and a ΔH _cc of 23.4 J/g, thus the crystallinity percentage is 10.32%.

PET-hydrolytic activity was measured as previously described (Espino-Rammer et al., 2013; Chen et al., 2021; Kaabel et al., 2021). Briefly, 10 μg/ml purified enzyme was incubated with 4 mg high-crystallinity PET microparticles (42.67% crystallinity) in 20 mM Tris-HCl (pH 8.5), 500 mM NaCl, 10% (v/v) dimethyl sulfoxide, and 10 mM CaCl₂. Samples were incubated at 150 rpm, 40°C. The reaction was terminated by adding 10 μl 1 M HCl to a 2 ml reaction supernatant, and a 25 μl aliquot analyzed by high performance liquid chromatography (Agilent 1,200, CA, United States) equipped with an SB C-18 column (5 μm, 4.6 × 150 mm, Agilent). The mobile phase was 20 mM phosphoric acid containing a 10–100% methanol linear gradient flowing at 1 ml/min. Eluates were monitored at 240 nm and peak areas for BHET, MHET, and TPA determined based on known standards.

BHETase and MHETase assays were performed in a total 1,000 μl volume with BHET or MHET dissolved in 900 μl 20 mM Tris-HCl (pH8.0). The concentration gradient of MHET and BHET was stetted as a range from 0.05 to 1.0 mM. Enzymatic reaction was started by adding 100 ul of enzyme solution at 40°C in triplicate. The reaction time of MtCut (0.02 mg/ml), ICCG (0.05 mg/ml), and IsPETase (0.05 mg/ml) were 15, 120, and 120 min, respectively. Substrate reduction during the reaction was detected by high performance liquid chromatography. One unit of enzyme activity was defined as the amount of enzyme required to reduce 1.0 μmol substrate per min under standard conditions. Data, v _o (mM/min) versus [S]₀ (mM) were fitted to a Michaelis-Menten model to calculate kinetic parameters.

Primers (Supplementary Table S1) were designed with substituted codons at target sites to generate mutants using one-step site-directed mutagenesis (Zheng et al., 2004). PCR was performed using Phanta Master DNA polymerase (Vazyme, Nanjing, China) with the following parameters: 95°C for 10 min, followed by 32 cycles of 95°C for 30 s, 50°C for 30 s, and 72°C for 4 min and then 72°C for 10 min. PCR products were digested with DpnI (MBI Fermentas, Vilnius, Lithuania) to remove the methylated parent plasmid and then purified using a PCR purification kit (GenStar, Beijing, China). The linearized plasmid derivatives were transformed into E. coli XL1-Blue competent cells to derive mutant plasmids. After purification and identification, target plasmids were transformed into E. coli BL21 (DE3) cells for mutant enzymes production, and protein was expressed and purified as described above.

DSF studies were performed to assess the thermal stability of MtCut and ICCG by determining T _m values. White clear 96-well PCR plates were used, with wells containing 12.5 μl buffer A (20 mM Tris-HCl, pH8.0, 200 mM NaCl), 10 μl 1 mg/ml protein solution in buffer A, and 2.5 μl 50 × SYPRO Orange (Sigma-Aldrich, St Louis, United States) solution in ddH₂O, to a final volume of 25 μl. DSF studies were conducted using a Bio-Rad CFX96 real-time PCR system (Bio-Rad, Hercules, CA, United States), set on the fluorescence resonance energy transfer channel using 490 nm excitation and 580 nm emission filters. Samples were heated from 10 to 95°C at 0.05°C/s. Protein unfolding was monitored by detecting changes in SYPRO Orange fluorescence. T _m values were determined from the peaks of the first derivatives of the melting curve using CFX Manager software (Bio-Rad).

The MtCut protein structure was predicted using the ColabFold implementation of AlphaFold2 with default parameters (Jumper et al., 2021; Mirdita et al., 2021). The built MtCut structure was prepared for computational docking using the protein preparation wizard in MOE software (version 2019.0102, Chemical Computing Group, Montreal, Canada). Hydrogen atoms were added to the protein–ligand complex at pH 7. The hydrogen bond network and protein structure were further optimized to the overall lowest potential energy configuration using protonate 3D. The ligands were docked using rigid-fit method with the carbonyl oxygen of the ester bond constrained in the oxyanion hole formed by Ser178 and His 256). The top-ranked docking conformations based on the default scoring function of GOLD were selected for further investigation.

All MD simulations were performed using AMBER20 (Case et al., 2021). The AMBER FF19SB force field was applied and the SHAKE algorithm used to restrict all covalent bonds involving hydrogen atoms, with a time step of 2fs. The Particle Mesh Ewald method was used to treat long-range electrostatic interactions. For the solvated system, two steps minimization were performed before the heating step. The first 4,000 cycles of minimization were performed with all heavy atoms restrained with 50 kcal/(mol·Å2), whereas solvent molecules and hydrogen atoms were free to move. Then, non-restrained minimization was conducted using 2,000 cycles of steepest descent minimization and 2,000 cycles of conjugated gradient minimization. Afterwards, the whole system was heated from 0 to 308 K in 50 ps using Langevin dynamics at a constant volume, and then equilibrated for 400 ps at a constant pressure of 1 atm. A weak constraint of 10 kcal/(mol·Å2) was used to restrain all heavy atoms during heating steps. Periodic boundary dynamic simulations were conducted for the whole system using a constant composition, pressure, and temperature ensemble at a constant pressure of 1 atm and 308 K in the production step. In the production phase, a 100 ns simulation was conducted, after which the whole system was heated from 308 to 343 K in 100 ns using Langevin dynamics at a constant volume. Finally, 100 ns MD simulations were conducted at a constant pressure of 1 atm and 343 K. Trajectories were further analyzed using Cpptraj (Roe and Cheatham, 2013).

The amino acid sequence of the MtCut catalytic domain from M. theromotolerans DSM45154 shared 66% amino acid sequence identity with Cut190 from Saccharomonospora viridis (Kawai et al., 2014), 62% with Est119 from Thermobifida alba (Thumarat et al., 2012), 61% with TfCut1 from Thermobifida fusca, 57% with LCC from leaf-branch uncultured bacterium (Sulaiman et al., 2012), and 46% with IsPETase from I. sakaiensis (Han et al., 2017). These homologous enzymes are cutinase-like, with reported PET hydrolyzing activity. MtCut contained a conserved pentapeptide sequence motif (GHSMG), and a catalytic triad (Ser178-Asp224-His256) (Supplementary Figure S1). To examine MtCut catalytic function, a recombinant enzyme produced in E. coli was purified to homogeneity using a combined His-tag affinity and ion exchange chromatography approach (Figure 1A). The purified MtCut molecular mass agreed with the calculated 34 kDa value. Using pNP-C₄ as the cutinase model substrate, MtCut exhibited its highest activity at 35°C and pH8.0 (Figures 1B,C). MtCut also showed typical cold-adapting properties, with more than 30 and 50% of full activity retained at 5 and 15°C, respectively. In Addition, the enzyme was unstable under heating treatment, with only 20% activity detected at 30°C for 2 h (Figure 1D). Enzyme kinetics toward pNP acyl esters of different chain lengths were determined under standard conditions (Table 1). Purified MtCut was active toward all tested pNP esters and generated the highest k _cat value against pNP-C₈. Thus, MtCut was a cutinase-type enzyme rather than an esterase or true lipase. However, the affinity toward longer acyl chains (C₁₄-C₁₈) was higher than shorter chains (C₄-C₁₂). Overall, no significant difference in catalytic efficiency (k _cat/K _m) was found among C₂-C₆, C₁₄-C₁₈ substrates, respectively. Molecular docking was performed to analyze the underlying mechanism for different kinetic parameters. Interestingly, a good correlation between catalytic efficiency and ligand binding energy against MtCut was found (Supplementary Figure S2A). Closer examination of the docked structure does not show significant pose differences (Supplementary Figure S2B).

Enzyme thermostability and activity were enhanced by Ca²⁺; by adding 1 mM CaCl₂, the optimum temperature (for pNP-C₄ hydrolysis) of MtCut increased from 35 to 40°C, and also thermostability was considerably enhanced (Figures 1B,D). Notably, 60% activity was detected at 55°C upon Ca²⁺ addition, while only 30% activity was observed in the absence of Ca²⁺. Enzyme Ca²⁺-enhanced thermostability was confirmed by DSF, with the T _m determined as the minimal value of each melting peak (Figure 2A). The T _m value of MtCut was 33°C without Ca²⁺ but increased to 35, 39.5, and 41.5°C at Ca²⁺ concentrations of 10, 100, and 300 mM, respectively. MtCut activity towards pNP-C₄ increased with increasing Ca²⁺concentration, while the pNP-C₄ hydrolase activity was inhibited by 33% in the presence of 1 mM EDTA (Supplementary Figure S3). MtCut PET hydrolysis was also enhanced by Ca²⁺, as observed by pNP-C₄ hydrolase activity, but was maximum at 10–100 mM CaCl₂ and decreased at higher Ca²⁺ concentrations (Figure 2B). In addition, the optimum enzyme temperature for PET was 45°C (Supplementary Figure S4), which was higher than pNP-C₄. Similar Ca²⁺-activating behaviors toward pNP-esters and PET hydrolysis were previously reported for another cutinase, Cut190 from S. viridis (Kawai et al., 2014).

To analyze the PET-hydrolyzing activity of MtCut, microparticle (mp-PET) and lc-PET films with distinct morphology and crystallinity were selected, and hydrolytic performances were evaluated using released aromatic monomers (combined BHET, MHET, and TPA). Although the mp-PET possessed a higher crystallinity (42.67%) than the lc-PET films (10.32%), MtCut displayed more efficient degradation against mp-PET (Figure 3A), suggesting the PET hydrolysis was influenced more by surface-enzyme contact than crystallinity. Therefore, mp-PET was used as a standard substrate for further PET hydrolysis analyses. On closer inspection of released aromatic products, MtCut (10 μg/ml) hydrolyzed mp-PET to MHET and TPA, plus trace BHET amounts (Figure 3B). When the enzyme concentration was increased, the proportion of MHET decreased, with no MHET detected after a 500 μg/ml MtCut reaction over 72 h (Figure 3C). Thus, we speculated that MtCut exhibited higher hydrolytic activity against BHET than MHET, as confirmed by kinetic analyses on both compounds (Table 2). Although MtCut showed a higher affinity towards MHET (K _m = 0.75 mM) than BHET (K _m = 3.07 mM), the k _cat value for BHET (4.23 s⁻¹) was much higher than MHET (0.13 s⁻¹), leading to an overall 7-fold higher catalytic efficiency for BHET.

PET hydrolysis by MtCut at 40°C was compared with the well-characterized IsPETase and ICCG (variant of LCC with improved activity and thermostability) molecules. IsPETase and ICCG outperformed MtCut at the early stages in a 60 h-reaction in terms of total released aromatic compounds, whereas MtCut was better at later reaction stages (Figure 3D). Product profiles showed that the three enzymes not only differed in terms of reaction rates, but also in terms of product diversity. All enzymes generated MHET and TPA as main products. However, IsPETase and ICCG showed a higher MHET to TPA ratio, while MtCut released both products in equal quantities (Figure 4A). In terms of barely altered MHET levels at later stages and the relative slower rate of TPA production by IsPETase and ICCG, we speculated that MHET very likely inhibited the enzyme activity. When mp-PET hydrolysis reactions were performed in the presence of MHET, significant inhibitory effects were detected toward IsPETase and ICCG. In contrast, no MHET inhibition toward MtCut was observed (Figure 4B). In Addition, IsPETase and ICCG exhibited weak hydrolyzing activity against MHET at 40°C, while the catalytic efficiency of MtCut was 10-fold higher (Table 2). The catalytic efficiencies on MHET were found to be in accordance with binding energy by molecular docking analyses (Supplementary Figure S5). Thus, different MHET effects and activities generated distinct PET hydrolysis product profiles, and suggested MtCut was a novel PET hydrolase, uninhibited by MHET.

Using DSF, MtCut thermostability was compared with ICCG, one of the most stable PET hydrolases, using DSF (Figure 5A). One obvious distinction was observed between MtCut and ICCG and related to temperature adaptation; T _m values were 33 and 74.5°C, respectively. To elucidate the temperature adaptation mechanism of both homologous enzymes and identify key components to engineer thermostability in cutinase-like enzymes, MtCut and ICCG (PDB entry: 6THT) structures were analyzed. Since our attempts to obtain MtCut crystals failed, we built an atomic coordinate structure using AlphaFold2 (Figure 5B). MD simulations at different temperatures were performed on MtCut and ICCG enzymes to investigate global structure changes and localized flexibility of individual amino acid residues.

When the system temperature increased from 35 to 70°C in the 100 ns simulation, the recorded root mean square deviation (RMSD) of MtCut was divided into two segments and that of ICCG changed from chaotic zone to stable zone (Supplementary Figures S6A,B). Our structure cluster analysis showed that MtCut transformed from a single-group to a multi-group (Supplementary Figure S6C), indicating the protein structure had rapidly changed. As expected, ICCG adjusted to a single group with increasing temperature (Supplementary Figure S6D). Both computational and experimental results for ICCG were in agreement and showed the enzyme was more stable under higher temperatures. Root mean square fluctuation (RMSF) analysis of each residue was used to describe structural flexibility during the heating program. When compared with ICCG, two loops (residues 94–99 and 289–301) on the MtCut surface exhibited higher RMSF values (Figures 5C,D). Interestingly, the latter loop was reportedly involved in Ca²⁺ binding, and the two mutations D238C/S283C (corresponding to residues Asp252 and Ser301 on MtCut) allowed disulfide-bond formation and thermal stabilization of LCC (Tournier et al., 2020). Moreover, the RMSF of MtCut residues 249–262, which comprised the catalytic loop with the active site His256, were remarkably increased when the simulation temperature increased from 35 to 70°C. While the paired region of thermophilic ICCG was resistant to heat denaturation, less RMSF increases were observed. Thus, we propose the catalytic loop which undergoes conformational changes in response to environmental temperature likely controls temperature adaptation in cutinase-like PET hydrolases.