Alkyl diols (1,5-pentanediol, 1,8-octanediol and 1,12-dodecanediol) and ether diols (diethylene glycol, triethylene glycol and tetraethylene glycol), terephthalic acid (TA) and 5-sulfoisophthalic acid (NaSIP) were purchased from Sigma-Aldrich (St. Louis, MO). Buffer components, bovine serum albumin (BSA) and methanol (HPLC grade) were also purchased from Sigma-Aldrich.

All standard chemicals used in this work were purchased from Sigma-Aldrich, Merck KGaA (Darmstadt, Germany) or Carl Roth (Karlsruhe, Germany), if not stated differently. Norleucine was obtained from IRIS Biotech GmbH (Marktredwitz, Germany). Enzymes for cloning and PCR were from Thermo Fisher Scientific (Waltham, MA). PCRs were performed using TaKaRa Ex Taq®High-Fidelity DNA polymerase (Clontech Laboratories, Inc., Mountain View, CA) or Dream Taq®DNA polymerase (Thermo Fisher Scientific). PCR primers were ordered from IDT Inc. (Coralville, IA) in standard desalted quality.

E. coli B834 (DE3) (E. coli B F– ompT hsdSB(rB– mB–) dcm+ gal met λ(DE3); Merck KGaA) was the host for the SPI experiment and E. coli DH5α (E. coli K-12 F– endA1 supE44 thi-1 recA1 relA1 gyrA96 phoA φ80lacZ∆M15 ∆(lacZYA-argF)U169 hsdR17 (rK– mK+) λ–; Thermo Fisher Scientific) was used for cloning experiments and plasmid propagation. Transformation of E. coli was carried out by electroporation as described by Seidman et al. (2001). pTTL was constructed according to Anderhuber et al. (2016). Briefly, the coding sequence of the lipase from T. thermohydrosulfuricus was PCR-amplified from synthetic DNA (IDT Inc.) using primers BPp244 and BPp245 (Anderhuber et al., 2016). The PCR fragment was inserted into pQE80L (Qiagen, Hilden, Germany) cut with EcoRI and HindIII by Gibson isothermal assembly (Gibson et al., 2009).

The SPI experiment was performed as described previously (Anderhuber et al., 2016). Briefly, M9 minimal medium (M9 MM) containing M9 salt buffer (48 mM Na₂HPO₄, 22 mM KH₂PO₄, 9 mM NaCl and 19 mM NH₄Cl) was used supplemented with 1 mM MgSO₄, 7 µM CaCl₂, 1 mg/L thiamine and trace elements (9 µM FeSO₄, 3.5 µM MnSO₄, 2.5 µM AlCl₃, 2 μM CoCl₂, 0.4 µM ZnSO₄, 0.5 µM Na₂MoO₄, 0.4 µM CuCl₂ and 0.5 µM H₃BO₄). 100 mg/L ampicillin was added for plasmid maintenance. 20 mM glucose was provided as the carbon source. For Met depletion at OD₆₀₀∼ 3, 3.5 g/L yeast extract (Carl Roth) were supplemented in the medium. Cells were grown in baffled shake flasks at 37°C with vigorous shaking. After depletion of Met, as indicated by growth arrest, the cultures were supplemented with 1 mM of either Met (cAA, parent TTL [Met]) or Nle (Nle, synthetic TTL [Nle]). Gene expression was induced by adding IPTG (VWR International, Vienna, Austria) to a final concentration of 0.5 mM and performed with vigorous shaking for 4 h at 30°C. Cells were harvested at 4,000 g for 20 min at 4°C.

Cell pellets were resuspended in 30 ml Ni-NTA Lysis Buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 7.4) and incubated for 30 min on ice. Cells were disrupted by sonication on ice (Branson Sonifier 250, Emerson Electric, St. Louis, MO). The sonication was performed for 6 min with the following settings: duty cycle: 70%, output: 7-8, sonication tip Φ∼1 cm. After centrifugation for 45 min at 40,000 x g and 4°C the lysates were filtered through 0.2 µm syringe filters and purified on Ni-NTA sepharose according to the manufacturer’s protocol (IBA GmbH, Goettingen, Germany). Finally, the buffer was exchanged for 100 mM Tris-HCl pH 7.0 with PD-10 columns (GE Healthcare, Chicago, IL).

The Bradford based Bio-Rad Protein Assay (Bio-Rad Laboratories GmbH, Munich, Germany) with bovine serum albumin as standard was used to determine protein concentrations of purified enzymes. The protein assay was performed according to the manufacturer’s instruction.

The protein solutions were desalted using Amicon Ultra 0.5 ml centrifugal filter units (Millipore, Billerica, MA). A final protein concentration of 30 pmol/µL was obtained with water containing 5% acetonitrile and 0.1% trifluoroacetic acid. The separation of possible protein variants was carried out on a capillary HPLC system (1200 Agilent, Agilent Technologies) using a PepSwift RP monolithic column (50 × 0.5 mm, Thermo Fisher Scientific) at a flow rate of 20 µL/min and a column temperature of 60°C. The gradient of solution A (water +0.05% TFA) and B (ACN +0.05% TFA) was performed as follows: 10% B for 5 min, 10%–100% B for 50 min, 100%–10% B for 1 min, 10% B for 15 min. The injection volume was 5 µL. The Thermo LTQ-FT mass spectrometer (Thermo Fisher Scientific) was operated with an ESI source in positive mode with the following settings: mass range: 300–2000 m/z, resolution 400000, 500 ms injection time, 1 microscan, source voltage 5 kV, capillary voltage 35 V, sheath gas flow 15. The protein mass spectra were deconvoluted by the Thermo Fisher Scientific software Protein Deconvolution 2.0, using the Xtract algorithm. The following main parameters were applied: charge carrier, H+; m/z range, minimal 800 to maximal 2000; minimal detected charge state, 4; s/n threshold, 5; relative abundance threshold, 20%. Trace amounts of unlabeled species might be present but fall below the detection limit of the mass spectrometry method (2–5%).

The structure of the TTL was modeled using the CATALOphore platform of Innophore GmbH (www.innophore.com) employing the program Yasara version 20.4.24.L.64 (www.yasara.org). The model is based on the structure of feruloyl esterase (Est1E) from Butyrivibrio proteoclasticus (PDB 2WTM) (Goldstone et al., 2010) as template, which shares 32% sequence identity and 50% sequence similarity with TTL. The active site cavity in the vicinity of the catalytic triad Ser-113, His-233 and Asp-203 was calculated using the program CavMan (Innophore GmbH, www.innophore.com) employing the LIGSITE algorithm (Hendlich et al., 1997). For the analysis of the hydrophobicity of the cavity, the hydrophobicity module of the program VASCo (Steinkellner et al., 2009) as implemented in CavMan was used.

The oligomeric model substrate bis-(benzoyloxyethyl) terephthalate (3PET) was synthesized as previously reported (Heumann et al., 2009). Other polyesters were synthesized in a two-step process and characterized as described by Haernvall et al. (2017b).

Model Substrates Polyester powders (10 mg/ml) were incubated in 1 ml of 100 mM Tris-HCl pH 7.0 and in the presence of 1 μM enzyme. The reaction mixture was shaken for 3 h at 70°C and 100 rpm (Infors HT Multitron, Infors AG, Bottmingen, Switzerland). In parallel, enzymes and polymers were incubated in pure buffer as blank reactions. All experiments were run in triplicates. Enzymes were precipitated by addition of ice-cold methanol (1:1 v/v), acidified with 0.1 M HCl to pH 4 and sedimented in a tabletop centrifuge (15 min, 0°C, 14,000 rpm; Hermle Labortechnik GmbH, Wehingen, Germany). Supernatants were used for HPLC analysis.

After hydrolysis, samples were analyzed by HPLC-UV on a system consisting of a Dionex UltiMate 181 3000 Pump (Dionex Cooperation, Sunnyvale, United States), a Dionex ASI-100 automated sample injector, a Dionex UltiMate 3000 column compartment and a Dionex UVD 340 U photodiode array detector. The hydrolysis products were separated by a reversed-phase column, [XTerra® RP18, (3.5 μm, 3.0 mm × 150 mm)] (Waters Corporation, Milford, United States) using a non-linear gradient where eluent A consisted of water, eluent B of methanol and eluent C of 0.01 N sulfuric acid. The separation was achieved by a non-linear gradient increased from 15% A to 40% A from 13 to 30 min, followed by an increase to 90% A during 5 min which was kept for 10 min to then be re-established to initial conditions within 1 min and equilibrated for 20 min. The injection volume was 5 μL, and the flow rate was 0.4 ml/min. The column compartment had a constant temperature of 40°C. The expected release products TA and NaSIP were detected via UV/VIS spectroscopy, and the release products were qualified and quantified based on calibration curves.

Enzyme adsorption on PET film was monitored as previously described (Ribitsch et al., 2013). Briefly, 0.5 × 1 cm² PET films were washed at 50°C in three consecutive steps of 30 min each with 1.5 ml each of 0.5% (v/v) Triton X-100, 100 mM Na₂CO₃ and deionized water. The washed membranes were incubated in 1 ml of 0.6 mg/ml enzyme solution for 2 h at 30°C. The films were washed twice by dipping into 1.5 ml TBST (25 mM Tris, 0.15 M NaCl, pH 7.6, 0.05% (v/v) Tween®-20) at room temperature. Then, the films were incubated with HisProbe (1:2500 dilution in TBST; SuperSignal West HisProbe Kit; Thermo Fisher Scientific, Waltham, MA) for 40 min at room temperature with shaking. The films were washed three times by dipping into 1.5 ml TBST as described above and were developed by incubating in 600 µL of SuperSignal West Pico Substrate Working Solution for 5 min. Chemiluminescence signals were detected using G:box Chemi HR16 and GeneSnap image acquisition software Version 7.05.01 (Syngene, Cambridge, United Kingdom) and quantified with Colorlite sph850 (Colorlite:Inovative color measurements; Germany).

The modelled structure of the lipase from Thermoanaerobacter thermohydrosulfuricus (TTL) exhibits a typical α/β-hydrolase fold with a central eight-stranded, mostly parallel β-sheet flanked by six α-helices (Figure 2). A catalytic triad is formed by the amino acid residues Ser-113, His-233 and Asp-203, with the serine being embedded in a GLSMGG sequence motif. The oxyanion hole is built up by the mainchain amide groups of Phe-37 and Met-114. The active site cavity is located at the upper end of the α/β-hydrolase core and delimited by a small cap domain consisting of residues 140 to 180. The TTL-sequence contains a total of 11 methionine residues, of which Met-114 and Met-142 are in close vicinity to the active site residue (Figure 2). It is conceivable that the replacement of those two methionine residues by norleucine modulates the catalytic activity of the enzyme most likely by altering its substrate binding properties. Most of the remaining methionine residues are part of the hydrophobic core of the protein (Figure 2). Amino acid replacements in this region could change the dynamic properties of the enzyme and its stability. Indeed, Hoesl et al. (2011) observed that the quantitative replacement of the Met residues in TTL resulted in a variant enzyme that was highly active without thermal activation while the parent enzyme containing exclusively Met residues was nearly inactive unless heat activated. Hoesl et al. hypothesized that the global replacement of Met by Nle would increase the hydrophobicity around the substrate binding site, which could enhance the accessibility of the substrate into TTL [Nle] in aqueous solution.

TTL was expressed in E. coli and purified from cleared cell lysates over a C-terminal 6xHisTag by affinity chromatography. Expression in the presence of the canonical Met yielded a defined band migrating at the 29.2 kDa molecular weight marker band as calculated for the theoretical molecular weight of TTL (Supplementary Figure S1). In contrast, the lipase expressed in the presence of Nle appeared as a blurred band at slightly lower molecular weight. In agreement with a previous report (Hoesl et al., 2011), the Nle variant exhibited accelerated electrophoretic mobility compared to the Met variant. The altered migration behavior was a first indication for the successful incorporation of Nle into the lipase. Typically, 20 mg/ml TTL [Met] and 10 mg/ml TTL [Nle)] were purified from 3 to 4 g of cell pellet obtained from cultures in shake flasks.

To analyze the incorporation efficiency of Nle, Met and Nle variants were subjected to intact protein mass analysis (Supplementary Figure S2). In the mass spectrum of TTL [Nle] the protein species with all 11 Met residues exchanged for Nle was the most prominent peak. A minor peak representing lipase with 10 Met residues exchanged for Nle was detected as well. This is in good agreement with previously published results (Hoesl et al., 2011).

The effect of the incorporating Nle into TTL on polymer degradation was assessed in a first step using the oligomeric model substrate 3PET. Interestingly, the global Met→Nle substitution in TTL had a positive impact on the hydrolysis as indicated by a ∼30 % higher amounts of hydrolysis products than to the parent enzyme (Figure 1A). Nevertheless, the pattern of the released molecules was similar for the two enzymes TTL (Met) and TTL (Nle). Benzoic acid (BA) was released in the highest concentration, followed by mono-(2-hydroxyethyl)terephthalic acid (MHET), terephthalic acid (TA) and hydroxyethylbenzoate (HEB), respectively. The total amount of released molecules for the hydrolysis of the oligomer model substrates was in the same range as previously reported for other enzymes, such as esterases from Clostridium botulinum (Perz et al., 2016a), a lipase from Thermomyces lanuginosus (Eberl et al., 2009), or cutinases from Fusarium solani pisi as well as from various Thermobifida species (Herrero Acero et al., 2011; Ribitsch et al., 2012b).

To further evaluate the impact of norleucine incorporation on polyester hydrolysis and to obtain a more detailed mechanistic insights into the hydrolysis behavior of TTL [Nle], a set of structurally different ionic phthalic acid polyesters were investigated. In a first step, TTL [Nle] was incubated with polymeric model substrates containing alkyl diols of different chain lengths, C5, C8, and C12 (Figure 1B). TTL [Nle] and the parent enzyme TTL [Met] were both active towards C5 while only a very low activity was detected towards C8 and C12 as indicated by the release of terephthalic acid. TTL [Nle] showed around 5% more activity towards C5 compared to TTL [Met]. However, enzymatic hydrolysis of polymeric substances is influenced by several parameters, such as water solubility of the substrates, crystallinity, molecular weight, glass transition temperature (Tg) and therefore difficult to predict (Chamas et al., 2020). As described recently (Haernvall et al., 2017b), polyesters C8 and C12 have lower water solubility, higher hydrophobicity and a higher crystallinity of 4 and 12%, respectively, than C5 with crystallinity below 1%, which may have influenced the hydrolysis. Low water solubility, high hydrophobicity (Okada et al., 1997) and crystallinity have previously been reported to have a negative impact on the enzymatic hydrolysis of polymers (Donelli et al., 2010). However, the glass transition temperature of the polymeric model substrates is decreasing with increasing chain length which would have been expected to have a positive impact on the enzymatic hydrolysis due to the higher flexibility of the polymer chains (Marten et al., 2003).

In a next step, TTL [Nle] and the parent enzyme TTL [Met] were incubated with the ether diol containing model substrates, EG2, EG3, EG4 (Figure 1C). Surprisingly, an up to 40% increased activity was found for TTL [Nle] towards all three polyesters (EG2, EG3, and EG4) when compared to the parent enzyme TTL [Met]. The highest activity was again detected on the shortest diol chain length, namely EG2, decreasing with increasing chain length. It can also be noted that an overall higher activity was detected for both enzymes towards all three ether diol containing substrates when compared to the alkyl diol analogs. This can be a consequence of the increased water solubility of the ether diol containing polymers and the increased hydrophilicity. It has previously been shown that increased water solubility and increased hydrophilicity is enhancing enzymatic hydrolysis. It has even been suggested as a parameter to tune polymeric biodegradation (Gigli et al., 2013).

In none of the cases, NaSIP was detected after hydrolysis, indicating that TTL [Met] and TTL [Nle] have a limited capacity to cleave ester bonds in close vicinity to the ionic monomer NaSIP. These results are in accordance with previously reported data for cutinase A and an arylesterase from Pseudomonas pseudoalcaligenes and a lipase from Pseudomonas pelagia (Haernvall et al., 2017a; Haernvall et al., 2017b).

We analyzed the adsorption of TTL [Met] and TTL [Nle] on PET films as described in the methods section. Adsorbed enzymes were detected by binding of horseradish peroxidase labeled HisProbe to the hexahistidine-fusion tag and chemiluminescence detection. TTL [Nle] adsorbed reproducibly better to the PET films than the parent protein (Figure 3). This result supports our notion that Nle improves the interaction of TTL with the polymer surface although a different polyester had been used. Both, TTL [Met] and TTL [Nle] survived well the incubation with PET because their esterase activities after the incubation, e.g., with PET, was not lower than before (Supplementary Information, Supplementary Figure S3). Our finding corroborates the extraordinary stability of TTL and its Nle reported previously (Hoesl et al., 2011).