Exploring the pH dependence of an improved PETase

Abstract

Enzymatic recycling of plastic and especially of polyethylene terephthalate (PET) has shown great potential to reduce its negative impact on our society. PET hydrolases (PETases) have been optimized using rational design and machine learning, but the mechanistic details of the PET depolymerization process remain unclear. Belonging to the carboxylic-ester hydrolase family with a canonical Ser-His-Asp catalytic triad, their observed alkaline pH optimum is generally thought to be related to the protonation state of the catalytic His. Here, we explore this aspect in the context of LCC^ICCG, an optimized PETase, derived from the leaf-branch compost cutinase enzyme. We use NMR to identify the dominant tautomeric structure of the six histidines. Five show surprisingly low pKa values below 4.0, whereas the catalytic H242 in the active enzyme displays a pKa value that varies from 4.9 to 4.7 when temperatures increase from 30°C to 50°C. Whereas the hydrolytic activity of the enzyme toward a soluble substrate can be modeled by the corresponding protonation/deprotonation curve, an important discrepancy is found when the substrate is the solid plastic. This opens the way to further mechanistic understanding of the PETase activity and underscores the importance of studying the enzyme at the liquid-solid interface.

Figure 1

Catalytic triad of the LCC^ICCG PETase. Composed of Asp210, His242, and Ser165, the three residues are essential for the hydrolase activity. The red arrow represents schematically the attack of the serine hydroxyl oxygen on the PET ester bond. To see this figure in color, go online.

Figure 2

Connecting the ¹H^ε1-¹³C^ε1 resonances to the H-N imidazole signals in LCC^ICCG-S165A. (A) HMQC spectrum with the histidine ¹H^ε1-¹³C^ε1 correlations. (B) Magnetization flow in the original (39) (orange) and the modified (blue) pulse sequence. Both τ and π tautomers are shown for clarity. (C) Overlay of the 2D ¹H^N-¹³C planes extracted at the ¹⁵N^ε2 frequency of H112, H164, H191, and H291 (dark blue) and the N^δ1 frequency of H242 (light blue). The inserts show the line shape in the ¹³C dimension, with a singlet for all ¹³C^ε1 nuclei, a doublet for the ¹³C^δ2 signals of the four τ tautomers, and a triplet for the ¹³C^γ of H242. (D) Overlay of (red) a ¹H^N-¹⁵N TROSY spectrum acquired with 16 scans and (gray) a best-TROSY (43) spectrum acquired with 128 scans. To see this figure in color, go online.

Figure 3

Identification of the tautomeric state of the histidines in LCC^ICCG-S165A using long-range scalar couplings. (A) ¹H^ε1-¹⁵N HMQC spectrum showing the signals of protonated (160–176 ppm, top part of the panel) and non-protonated (240–260 ppm, bottom part of the panel) ¹⁵N nuclei. For imidazole rings in the τ tautomeric state such as H218, we observe a strong H^ε1/N^δ1 and a weaker H^ε1/N^ε2 correlation (connected by a blue line). For H242 in the π tautomer state (red line), both ¹H^ε1 and ¹H^δ2 protons connect to the ¹⁵N^ε2 resonance (insert, red arrows), whereas only the H^ε1 connects to the protonated N^δ1 (insert, green arrow). (B) Assigned ¹H^ε1-¹³C^ε1 HMQC spectrum. To see this figure in color, go online.

Figure 4

pH titration of the different imidazole H^ε1-C^ε1 resonances in LCC^ICCG-S165A. (A) Long-range ¹H^ε1-¹⁵N spectra showing the expected upfield shift that accompanies protonation of H242. (Insert) For the crosspeaks characterizing the imidazole rings of H112 and H191, we start observing an upfield shift when the pH reaches 4.5. The resonance of H164, on the contrary, shifts downfield upon decreasing the pH, which cannot correspond to an increase in its own protonation level. (B) ¹H^ε1-¹³C^ε1 HMQC spectra acquired as a function of pH. Only the resonance of the catalytic H242 shifts appreciably in both dimensions. (C) ¹H^ε1 (blue) and ¹³C^ε1 (red) chemical shift values of the catalytic histidine (H242) plotted as a function of pH. The dots represent experimental data, whereas the solid lines correspond to the best fit to the Henderson-Hasselbalch equation leading to a single pKa value of 4.79 ± 0.04. For a statistical analysis of the curve fit, see Fig. S8 in the supporting material. To see this figure in color, go online.

Figure 5

Comparison of the inactive S165A (blue) and active S165 (red) LCC^ICCG enzymes. (A) Structural comparison of the catalytic triads in the crystal structures of both enzymes. (B) ¹H^ε1-¹³C^ε1 spectra of both enzymes showing the downfield shift of the H242 crosspeak in the active enzyme. To see this figure in color, go online.

Figure 6

LCC^ICCG catalyzed hydrolysis of soluble and insoluble species as a function of pH. (A) Zoomed view on the H242 ¹H^ε1-¹³C^ε1 resonance at 30°C as a function of pH. (B) H242 ¹H^ε1 and ¹³C^ε1 chemical shift values as a function of pH. The dots represent the experimental data. The best fit toward the Henderson-Hasselbalch equation (line) gives a pKa value of 4.90 ± 0.05. (C) pH-dependent rates of hydrolysis of BHET. Circles and triangles represent experimental data points of two independent series. The dashed line corresponds to the H242 protonation state based on the Henderson-Hasselbalch equation with the pKa fixed to the experimentally determined value of 4.90. (D) Zoomed view on the H242 ¹H^ε1-¹³C^ε1 resonance at 50°C as function of pH. (E) ¹H^ε1 and ¹³C^ε1 chemical shift curves of H242 as a function of pH at 50°C. The dots represent the experimental values, and the line displays the best fit toward the Henderson-Hasselbalch equation, yielding a pKa value of 4.70 ± 0.05. (F) Hydrolysis of PET powder as function of pH. Dots represent experimental initial velocities as a function of pH, with error bars from the triplicate assay. The protonation/deprotonation curve of H242 at 50°C is modeled as a dashed line with a pKa value of 4.70. To see this figure in color, go online.