大家好，今天推送的文章是2024年9月发表在 ACS Catalysis 上的“A High-Throughput Screening Platform for Engineering Poly(ethylene Terephthalate) Hydrolases”，通讯作者是来自美国洛斯阿拉莫斯国家实验室的Hau B. Nguyen。
PET hydrolase hydrolyzes polyethylene terephthalate (PET), making waste plastics potentially recyclable in the bioindustry. Many of the current PET hydrolases have been engineered to improve catalytic activity and stability, but current screening methods have limitations in screening large libraries, including at high temperatures. The authors' team has developed a platform to simultaneously detect the protein solubility, thermostability, and activity of multiple PET hydrolase mutants by paired plate splitting GFP and model substrate screening. The authors' team applied the platform to improve the performance of the benchmark PET hydrolase leaf-branch compost cutinase (LCC) through directed evolution.
1
An integrated platform for engineered PET hydrolases
As shown in Figure 1, the authors developed a screening platform for directed evolution by random mutagenesis that can simultaneously and efficiently assess the PET hydrolase repertoire of ≥ 104 mutants in each evolutionary round, including protein expression and solubility, thermal stability close to 70 °C, and catalytic performance. The platform consists of four components: (1) the creation of a large random mutagenesis library by DNA shuffling, (2) the HT (≥104 mutants) colony-level co-screening assay for the activity, expression, and solubility of the enzyme variant on BHET model substrate agar plates, and (3) the validation of cell lysates (∼102 mutants) using amorphous PET films or high-crystallinity PET powder substrates to identify the PET selected from (2). Enzyme variants with higher activity on substrates, (4) using purified enzymes and a variety of PET substrates, including amorphous PET films, amorphous PET powders, and high-crystallinity PET powders, were used to characterize the final enzyme optimum (1 to 10 variants). Integration into the directed evolution platform enables the generation of random mutagenic libraries from the starting enzyme scaffold, screening at different selection pressures (e.g., elevated temperature and elevated substrate concentration), selection of improved enzyme variants, and validation of their improved properties and sequences (Figure 1D, F), with each round of evolution taking approximately 6-8 weeks. The improved enzyme variants are pooled and used as parents for further directed evolution, restarting the directed evolution cycle. After several rounds of directed evolution, the authors selected the final enzyme optimization protocol for quantitative analysis of aromatic monomer products using HPLC and detailed performance analysis by measuring PET hydrolysis in a pH-controlled bioreactor.
Figure 1
The first step in the platform is an HT co-screening assay, which assesses the activity and solubility of a large library of enzyme variants on model substrate agar plates at the colony level. Figure 2 details how the HT co-screening assay works, which involves two simultaneous steps: (1) assessment of enzyme solubility and concentration by splitting GFP complementation; (2) Activity is assessed by reaction with BHET as a model substrate.
Figure 2
The team of authors evaluated the catalytic performance of an improved enzyme variant (∼101-102) selected from the HT co-screening assay described above on amorphous PET film specimens or high-crystallinity PET powders (Figure 1c). The identified modified enzyme variants are then used as parents for the next round of directed evolution or, if the desired engineering objectives have been achieved, for expression, purification, and more thorough characterization using HPLC and pH-controlled bioreactors.
After the construction of the platform, the authors used the HT screening platform to perform leaf-branch compost cutinase (LCC) engineering, with the aim of demonstrating that the newly developed platform can improve the catalytic efficiency of the benchmark PET hydrolase LCC-ICCG, one of the most potent PET hydrolases. Directed evolution is initiated first using wild-type LCC (LCC-WT) as a starting template. DNA shuffling was used to create a library of DNA fragments containing random mutations, cloned into a C-terminal GFP11-tagged pET21b(+)-GFP11 screening vector, and transformed into E. coli. The enzyme libraries were then screened using HT co-screening assays, first coarse and then fine. Most of the enzyme variants screened out in the first round contained a single mutation, including P38L, V118I, and L159Q, which exhibited higher activity (larger transparent region) on BHET agar plates compared to LCC-WT, and these variants were then pooled together as parents for a second round of directed evolution. After two rounds of directed evolution, the variant LCC-F2 was obtained, with mutation sites of V118I, A149V, L159E, and V202I, which was more active than LCC-WT and had comparable enzyme solubility/concentration on BHET agar plates. Validation assays were performed using enzymes normalized to a final concentration of 0.1 μM in cell lysate, and the results showed that LCC-F2 was approximately 13% more active against highly crystalline PET powders after 6 hours of reaction compared to LCC-WT. However, this variant is about 23% less active on the same substrate compared to LCC-ICCG, so the authors proceeded with multiple rounds of directed evolution.
Previous reports have shown that the addition of disulfide bonds to LCC-WT (D238C−S283C) improves the thermal stability of the enzyme. In order to improve the performance of LCC-F2, the authors' team added disulfide bonds to the LCCF2 and LCC-P38L variants to construct the same disulfide bonds, and used LCC-ICCG as one of the starting scaffolds for shuffling in the next round of directed evolution. After the third round of evolution, a variant, LCC-F6, was selected, which contained mutations P38L, L117P, and A149V in addition to the LCC-ICCG mutations (Y127G, D238C, F243I, and S283C), as it exhibited activity comparable to LCC-ICCG on both amorphous PET film specimens and high-crystallinity PET powder substrates after 6 hours of reaction at 70 °C. This variant also has similar expression and solubility levels compared to LCC-ICCG. The authors' goal was to design a new LCC variant that would outperform the benchmark LCC-ICCG for hydrolysis on amorphous PET specimens, which are ubiquitous in laboratory-scale PET hydrolase tests. Using LCC-F6 as a template, a fourth round of directed evolution resulted in two optimal variants: LCC-B8 and LCC-C9. In addition to the mutations present in the LCC-ICCG parents (Y127G, D238C, F243I, and S283C), LCC-B8 contains mutations P38L, L117P, A149V, and S247L, and LCC-C9 contains mutations P38L, Y61C, M91I, L117P, A149V, and S247L. After 6 hours of reaction with amorphous PET film specimens at 70 °C, LCC-B8 was shown to release aromatic products 5.2-fold higher and the maximum rate was 5.6-fold higher compared to LCC-ICCG, while LCC-C9 was released approximately 11-fold higher aromatic products and 10.6-fold higher maximum rate. But these two variants showed lower activity against high-crystallinity PET powders, with approximately 16% and 22% reduction in aromatic products compared to LCC-ICCG, respectively. The two variants, LCC-B8 and LCC-C9, were pooled together and subjected to another round of directed evolution, and the fifth round of directed evolution yielded the best variant, LCC-LANL, which resulted in a 14.3-fold higher release of aromatic products and a 13.9-fold higher maximum rate after 6 hours of reaction with an amorphous PET film specimen at 70°C compared to LCC-ICCG. Therefore, after expression and purification from 1 L of cell culture, this variant, LCC-LANL, which contains nine mutations: P38L, Y61C, M91I, L117P, A149V, H218Y, Q224H, S247L, and T256I, was selected for final characterization, in addition to the mutations present in the LCC-ICCG scaffold (Y127G, D238C, F243I, and S283C).
2
Characterization and structural analysis of LCC-LANL
The authors compared the performance of the purified enzyme LCC-LANL with LCC-ICCG and found that the maximum catalytic rate of LCC-LANL was higher than that of the parent enzyme LCC-ICCG when hydrolyzing amorphous PET film specimens at 68 °C (Figure 3). The analysis was performed using UV absorbance and HPLC aromatic monomer quantification (Figure 3a), and the results were similar (Figure 3b-e). While the initial reaction was performed under conditions under which the highest activity of LCC-ICCG was reported, the authors chose to further evaluate LCC-LANL and LCC-ICCG at different temperatures and longer reaction times. Therefore, additional tests were performed on LCC-LANL and LCC-ICCG at 65 °C and 70 °C. Similar to the results at 68 °C, LCC-LANL was observed to exhibit a higher maximum rate than LCC-ICCG at 65 °C, with similar results between UV absorbance measurements and HPLC monomer quantification. However, at 70 °C, the maximum hydrolysis rate of LCC-LANL did not increase significantly compared to LCC-ICCG, and activity began to decrease after 8 hours of incubation. However, when using purified proteins, the product release and maximum rate increase of LCC-LANL versus LCCICCG is not as great as using crude cell lysates with the same enzyme concentration.
Figure 3
Although the maximum rate of LCC-LANL is significantly increased compared to LCC-ICCG, its activity appears to stabilize after 12 hours. This may be due to reduced thermal stability or product inhibition. A heat treatment of 80 °C for 1 h was implemented as the selection pressure when screening LCC-LANL variants to maintain the high thermal stability of LCC-ICCG and to demonstrate that LCC-ICCG does not experience product inhibition. However, in order to fully determine the extent and potential impact of the reduction in the thermal stability of LCC-LANL, the authors conducted further experiments, and differential scanning calorimetry analysis showed that the thermal unfolding of each enzyme was essentially an irreversible process, with LCC-LANL and LCC-ICCG exhibiting comparable energy impairment (Eact) as they unfolded from their native state to their denatured state. However, LCC-LANL can overcome this energy barrier at a given frequency at a temperature (Tact) about 10 °C lower than LCC-ICCG, so its kinetic stability is lower than that of the latter variant. Despite this property, the thermal stability of LCC-LANL was assessed using a split GFP complementation assay (Figure 2e), and it was observed that while LCC-LANL enzyme viability was reduced above 72°C, at lower temperatures, LCC-LANL exhibited similar thermal stability to LCC-ICCG, with both proteins essentially retaining 100% of their protein after heat treatment at 72°C. The authors then evaluated the performance of LCC-LANL on a variety of PET substrates that these enzymes may encounter in industrial applications.
The structure of LCCLILL was modeled using the Rosetta Macromolecule Modeling Suite. Most mutations obtained by directed evolution result in increased hydrophobicity of solvent exposure. In particular, the S247L mutation has been shown to increase the hydrophobic surface area near the site of enzyme activity (Fig. 5a, b). While this may destabilize the protein structure, it can be advantageous in the presence of a substrate that is predominantly hydrophobic. In addition, given the tight packing of H218 and K194, this mutation eliminates the strong charge-charge repulsion interaction between them. A 0.4 Å increase in distance between K194 and H218 was found compared to the original analytic structure (PDB: 8JMP) obtained after running the Rosetta Relax energy minimization program. The predicted configuration of K194 in LCC-LANL may also allow for better solubilization, which may be applicable to lysine, which is one of the most hydrophilic residues and does not participate in ion-pair interactions. Other results of structural refinement using Rosetta Relax showed that the remaining mutations were far from the active site. Among them, the M91I and T256I mutations may have a synergistic effect in establishing a more homogeneous hydrophobic core, albeit with several voids. These may lead to increased mobility of LCC-LANL enzymes at lower temperatures (65 °C) compared to 68 and 70 °C, resulting in improved catalytic efficiency. However, voids can also reduce thermal stability. The L117P mutation is predicted to be unstable because it directly removes the backbone hydrogen bond and creates a cavity in the protein core through a lever arm effect on the backbone. These may be valuable target sites that evolve in a continuous evolutionary cycle on the LCC-LANL template.
Figure 5
The authors then performed molecular docking of PET with LCC-ICCG and LCCCL, and flexible ligand docking simulations revealed several stable non-productive binding patterns of PET substrates near the LCC-ICCG active site. Of the 100 predicted optimal binding modes, 60 were productive, with substrate atoms within 4 Å of the catalytic residues S165, D210, and H242. It is important to note that all 100 predicted optimal binding patterns are valid for LCC-LANL. Rosetta scores and their constituent terms for these patterns and the set as a whole were statistically analyzed. The analysis showed that the difference in total Rosetta score (total_score) between the docking and free states of LCC-LANL was higher than that of LCC-ICCG. This suggests that the binding of LCC-LANL to PET is more stable than that of LCC-ICCG. Higher stability after binding to PET may manifest as a higher binding affinity, which may explain the higher catalytic efficiency observed in bioreactor experiments using amorphous PET film specimens. This may also explain the difference in the maximum rate of LCC-LANL and LCC-ICCG in cell lysates compared to purified protein samples, as there is more non-specific binding in cell lysates, which may result in LCC-LANL binding more frequently to PET film specimens than LCC-ICCG. However, the same trend was not observed in amorphous PET powders, which may be due to changes in surface structure and charge as they were milled, and increased crystallinity. A more detailed study of the interaction of LCC-LANL with this substrate type may provide valuable insights into why LCC-LANL has higher hydrolytic activity on amorphous films rather than powders.
Article Links:
https://doi.org/10.1021/acscatal.4c04321