The “rigid region” identified by joint analysis of all TPL structuresTPL monomer consists of a large domain and a small domain20. The catalytic active sites are situated within the crevice formed between the large and small domains. During catalysis, the small domain undergoes movement towards the large domain, defined as an open-closed state transition20 (Fig. 1A). This conformational change is crucial for enzyme activity. To further characterize this conformational change, we analyzed 88 monomer conformers from 22 PDB entities using a joint structure analysis approach. We discovered that the open and closed conformers separate along the second dimension, indicating that the second eigenvector bears the key conformational features distinguishing these two states (Fig. 1B). The results showed that there are four ‘’hinge regions’’ that undergo movement towards a ‘’rigid region’’. These ‘’hinge regions’’ are Pro21-Ser51, Cys351-Tyr371, Arg377-Thr402 and His430-Ile456, while the ‘’rigid region’’ is almost equivalent to the large domain, consisting of residues Gly52-Phe346 (Fig. 1C and D). Additionally, we analyzed the relative mobility of all residues and found that ‘’hinge regions’’ exhibit greater flexibility (Fig. 1D).Fig. 1(A) Projection of TPL monomer structures onto the second and third modes. Each point represents a monomer conformer, colored according to its PDB ID. The open and closed TPL monomers primarily differ in their composition of the second eigenvector. (B) Heatmap of structural differences between TPL monomer structures. We used the superimpose module in Biopython to align Cα traces of one TPL monomer onto another and calculate the r.m.s.d. values. The r.m.s.d. for each pair of structures is indicated with a linear color gradient, as shown in the color bar on the right. Annotations on left of the heatmap show the results of hierarchical clustering, demonstrating all TPL monomers can be divided into two states. (C) Monomer structure of tyrosine phenol-lyase. The closed conformation is shown in slate, and the open conformation is in limon. Five mutations are shown in sticks, the ‘’hinge region’’ is represented as cartoon and the ‘’rigid region’’ is shown as tube. (D) The second left eigenvector U2 of TPL monomer structure dataset. The lower panel shows the second left singular vector bears the conformational features that distinguishes the open and closed states of TPL monomers. The hinge regions undergo relative movements about the rigid region, which undergoes insignificantly positional changes, suggesting it remains stable during conformational transition. The upper panel plots the average cumulative r.m.s.d. value of each residue, indicating its relative mobility.Conformational flexibility plays different roles in enzyme activity. In “hinge regions” that transmit conformational changes, flexibility is crucial for enzyme activity. However, for regions that move as a rigid body, flexibility may be detrimental to activity. B-factor analysis alone could not discern these regions. Therefore, we employed a structure-based approach to analyze the conformational flexibility of TPL and identified the “rigid region” that may undergo collective movement. We hypothesized that stabilizing this region would facilitate the conformation transmission and improve the enzyme activity.In silico screening for stabilizing mutationsOur design strategy focuses on reducing the computational and experimental costs in engineering. Towards this objective, we utilized the phylogenetic sequence information of TPL. Initially, we generated a multiple sequences alignment and calculated a position-specific substitution matrix (PSSM)21, which reflects the log-likelihood of observing any of the 20 amino acids at each position. We defined the allowed mutations for computational scanning at positions distant from the active site (above 8Å) based on a favorable PSSM score7. Subsequently, we employed the Rosetta application Cartesian_ddg to calculate the energy difference between the wild-type and the mutant (∆∆G). This calculation was carried out in two rounds. In the first round, we individually modeled each mutation against a TPL monomer. Mutations with a stabilizing effect (∆∆G < −1.0) underwent further assessment in the second round. The catalytic active site is located at the interface between two subunits of the catalytic dimer. Additionally, many PLP-dependent enzymes, such as glutamic acid decarboxylase22, DOPA decarboxylase23, and aminolevulinic acid synthase24, which belong to the same family as TPL, function in dimeric form25. Thus, we hypothesized that the dimer represents the fundamental functional unit of TPL. To exclude the mutations that would disrupt the dimeric form of TPL, simultaneous mutations were introduced at corresponding positions in both subunits of the dimer. Ultimately, our analysis identified 162 single mutations (∆∆G < −1.0) with the potential to stabilize the dimeric form of TPL (Fig. 2A).Fig. 2(A) Stability prediction of mutations modeled on catalytic dimer of tyrosine phenol-lyase. The heatmap illustrates the calculated ΔΔG values, indicating stabilizing (red) and destabilizing (blue) variants. The mutants excluded from calculation are represented in white. Hinge regions are highlighted in red. (B) Spatial arrangement of hotspots within the catalytic dimer of tyrosine phenol-lyase. The figure shows 26 hotspots as sticks (PDB ID: 6mo3), with hinge regions highlighted in red and chains colored in blue and green.We ranked these mutations based on the calculated ∆∆G values, from low to high, and randomly selected five out of the top ten mutations for activity measurement (Table 1, Table S4, Fig. S6, Fig. S7 and Fig. S8). Among them, two mutations exhibited slightly improved activity and were located within the ‘’rigid region’’. In contrast, the R397L mutation in the ‘’hinge region’’ exhibited lower activity (Fig. 1C). We also observed a substantial reduction in activity with the A5L mutation at the N-terminus (Fig. S6, Fig. S7 and Fig. S8), which is consistent with previously reported inactive mutation T15A .Table 1 The activity of mutants with single mutation.Scanning for the beneficial combinatorial mutationsWe identified 26 positions with multiple stabilizing mutations (∆∆G < -3.0, Table 2), which were recognized as hotspots. Additionally, we excluded certain hotspots (2 and 10) located in the N-terminus of TPL, as mutations in this region consistently disrupted the enzyme activity. Observing the spatial proximity among some of hotspots, we hypothesized that combining nearby stabilizing mutations would further enhance stability and activity. Consequently, we clustered 16 of these hotspots into seven groups, where the hotspots within each group were spatially adjacent (Table 3; Fig. 2B).Table 2 Hotspots of combinatorial mutations strategy.Table 3 Seven groups of 16 hotspots.Groups 1 and 3 each consisted of three hotspots, while the remaining groups comprised two hotspots each. Subsequently, we calculated the ∆∆G values for all combinations of mutations within these seven groups (Table 4). Since Groups 1 and 3 both exhibited three hot spots and the lowest calculated ∆∆G values, we compared these groups and found that the variants in Group 3 had a lower average energy. Consequently, we first tested all variants from Group 3 and observed a correlation between activity and the calculated ∆∆G values. Based on these findings, we subsequently tested only the variant with the lowest ∆∆G value from Group 1.Table 4 ∆∆G values of combinatorial mutants.Experimental results revealed that the five mutants exhibited elevated activity levels, as detailed in Table S3 and Fig. S8. Notably, the mutant A206S/E202A/R201Y demonstrated the highest activity, reaching 3.37 ± 0.49 µmol/mg·min. Following closely, the mutant A206N/E202A/R201H exhibited the second-highest activity level, with a value of 3.01 ± 1.25 µmol/mg·min. Importantly, all six amino acid sites (G149, N151, A153, R201, E202, and A206) within these high-activity mutants were situated away from the catalytic site, residing on the surface of TPL. Furthermore, these amino acid sites were in the flexible region (Fig. 1D), aligning with prior reports suggesting that engineering the flexible region with high mobility profile is advantageous for enhancing enzyme activity9.
