Abstract
Enzymes are natures’ catalysts and will have a lasting impact on (organic) synthesis as they possess unchallenged regio- and stereo selectivity. On the downside, this high selectivity limits enzymes’ substrate range and hampers their universal application. Therefore, substrate scope expansion of enzyme families by either modification of known biocatalysts or identification of new members is a key challenge in enzyme-driven catalysis. Here, we present a streamlined approach to rationally select enzymes with proposed functionalities from the ever-increasing amount of available sequence data. In a case study on 4-phenol oxidoreductases, eight enzymes of the oxidase branch were selected from 292 sequences on basis of the properties of first shell residues of the catalytic pocket, guided by the computational tool A2CA. Correlations between these residues and enzyme activity yielded robust sequence-function relations, which were exploited by site-saturation mutagenesis. Application of a peroxidase-independent oxidase screening resulted in 16 active enzyme variants which were up to 90-times more active than respective wildtype enzymes and up to 6-times more active than the best performing natural variants. The results were supported by kinetic experiments and structural models. The newly introduced amino acids confirmed the correlation studies which overall highlights the successful logic of the presented approach.
Similar content being viewed by others
Introduction
The exponentially growing number of sequences in public databases reveals more and more of nature’s treasure chest1,2,3. At the same time, handling big data with ten-thousands of protein candidates becomes increasingly important to select suitable biocatalysts, as enzymatic approaches are at the forefront of synthetic applications. This development is reasoned by the excellent selectivity of enzymes and their ability to activate structural motifs, which were difficult, if not impossible, to target by classical chemical methods4,5,6. Therefore, efficient discovery and improvement of biocatalysts is an emerging topic7,8,9. Despite excellent tools for enzyme improvement, like directed evolution, a good and well-chosen starting point for a mutagenesis campaign saves resources and time10,11,12. Thus, computational approaches gain popularity for enzyme discovery7,13. Nowadays, many approaches, like sequence similarity networks (SSNs), allow the clustering of a vast sequence space14,15. But considering that single mutations can already change the substrate scope or the stability of enzymes dramatically, it becomes apparent that a more focused investigation of a few hundred sequences is beneficial to fetch more modest differences between the catalysts. Thus, the here-described approach considers the diversity of the first shell residues of the catalytic center, which requires a certain knowledge of the enzyme family or structural information as a starting point for residue selection. However, to great advancements in the field of structure prediction, structural models are nowadays remarkably easy to obtain, e.g., by using the AlphaFold2 algorithm16,17.
To demonstrate the applicability and the potential of this approach, the family of flavin-dependent 4-phenol oxidoreductases was chosen, which was recently investigated for its potential in the utilization of lignin-derived compounds18,19. The family is part of the widespread vanillyl alcohol oxidase/p-cresol methyl hydroxylase (VAO/PCMH) superfamily, among which the family of 4-phenol oxidoreductases distinguishes itself from other families as it is comprised of dehydrogenases and oxidases, which are further divided into bacterial and fungal enzymes20,21. This diversity within the family serves as an excellent model system as it allows for unique sequence features to be compared between all enzyme groups.
Within the 4-phenol oxidoreductase family, two fungal and seven bacterial sequences were described until today. The fungal vanillyl alcohol oxidases (VAOs) from Diplodia corticola and Penicillium simplicissimum share a similar substrate scope with differences for substitution patterns at the aromatic ring in o-position22,23,24. Of the bacterial enzymes, three oxidases and four dehydrogenases are described. The eugenol oxidases (EUGOs) from Rhodococcus jostii RHA1 and Nocardioides sp. YR527, and the 4-ethyl phenol oxidase (4EPO) from Gulosibacter chuangengensis represent the oxidase branch25,26,27, while the p-cresol methyl hydroxylase (PCMH) from Pseudomonas putida, the eugenol hydroxylase (EUGH) from Pseudomonas sp. OPS1, and the pinoresinol-α-hydroxylases (PRαHs) from Burkholderia sp. SG-MS1, as well as Pseudomonas sp. SG-MS2 are members of the dehydrogenase family28,29,30. While the oxidases use dioxygen as a terminal electron acceptor, the dehydrogenases are cytochrome c dependent31. All enzymes harbor a covalently bound flavin adenine dinucleotide (FAD) cofactor as a prosthetic group and accept phenolic substrates with varying substituents in p-and o-position to the hydroxy group. The size of the accepted substrate molecules ranges from small compounds like 4-cresol to the bulky tetrahydrofuran lignan pinoresinol. Overall, a broad reaction spectrum is observed, which includes hydroxylations in 4α- and 4ɣ-position, dehydrogenation, oxidative deamination, and cleavage of benzylic ethers (Fig. 1)23,25,32. Mechanistic studies for the VAO from P. simplicissimum (PsVAO) revealed that a hydride is transferred from the benzylic position of the substrate to the N5 atom of the FAD cofactor which results in the formation of a methide intermediate33. From this intermediate, either a proton is abstracted or water attacks as a nucleophile to yield the oxidized product. The catalytic cycle is closed with a two-electron transfer to the respective electron acceptor.
A hydride from the 4α-position is transferred to the FAD cofactor, which results in the formation of a quinone-like methide intermediate. In dependency of group X and the particular enzyme, different products are formed, which fall in one of the four shown categories (black boxes).
The high diversity in catalyzed reactions and substrate scope makes the enzyme family of 4-phenol oxidoreductases an interesting case study. Stereo- and regioselective oxidation is a cornerstone of (organic) synthesis, and phenolic compounds represent a common drug motive. Therefore, we decided to expand nature’s toolbox for these reactions while providing a streamlined approach for rational enzyme selection which includes a general-use software tool for sequence analysis (A2CA)34. Within this work, we demonstrated the capabilities of A2CA as a user-friendly, sequence-based alignment tool that allows for quick visualization and setting of selection criteria for efficient exploration of the natural sequence space. From the initial analysis, bacterial 4-phenol oxidases emerged as the most diverse branch of the family and were subsequently studied in detail. Within this exceptional versatile enzyme class, eight enzymes were selected by A2CA guidance and robust sequence-function relations were established by correlations of the residues’ diversity with the enzyme activity. In combination with an efficient oxidase screening assay, directed evolution of identified hot spot residues allowed us to expand the natural sequence space of 4-phenol oxidases towards substrates with non-natural substituents in o- and p-position.
Results
A2CA-guided enzyme selection based on function-specific clustering of the catalytic center
To streamline the analysis of the family of 4-phenol oxidoreductases, the first-shell amino acid residues of the catalytic pocket were grouped into five functional clusters according to their characteristics, which were derived from literature and geometric considerations (Fig. 2a). To account for residue movement, the EUGO from Rhodococcus jostii RHA1 (RjEUGO), was chosen as template, since five crystal structures were available35. Cluster properties resulted in the following four hypotheses regarding substrate binding: H1: Based on earlier mutagenesis studies, it can be speculated that the residues of the P-cluster are essential for substrate binding36. H2: Residues of the T-cluster and H-cluster likely interact with the substrates’ o-substituent(s). H3: The polar W-cluster probably interacts with the water nucleophile or polar groups of the substrate itself and, thus, is decisive for the reaction type. H4: The hydrophobic A-cluster likely restricts the size of the p-substituent.
a Functional cluster within the first-shell residues of the catalytic center in the crystal structure of RjEUGO (PDB: 5FXP). The FAD cofactor is shown in orange, while the substrate vanillyl alcohol (5) is shown in light red. The P-cluster is comprised of Y91, Y471, and R472 which are essential for the binding of the phenolate moiety of the substrate. The T-cluster is located at the substrate entrance tunnel, and residues G165, V166, and I427 may interact with an o-substituent of the bound substrate molecule. Likewise, G392 and H390 in the H-cluster are in contact with the other o-substituent. The substrate’s p-substituent interacts with the polar residues D151, Y169, R278, E378, and Q425 of the W-cluster on one side, while it is in contact with the non-polar residues M282, L381, and L438 of the A-cluster on the other side. V436 is located on top of the substrate. b Substrate orientation in the homology model of EUGO from Allonocardiopsis opalescens (AoEUGO) after 25 ns of simulation. The rotation relative to the 1,4-axis in RjEUGO is shown. c Substrate orientation in the crystal structure of 4EPO from Gulosibacter chungangensis (Gc4EPO, PDB: 7BPI).
Database searches resulted in 292 unique sequences, which clustered in three major clades containing subclades with characterized enzymes (Fig. 3a). Using the software tool A2CA, the first shell residues of the catalytic center were highlighted to display the natural cluster variability among the enzyme family (Fig. 3b). In agreement with H1, the P-cluster was found to be conserved for the all 4-phenol oxidoreductases. Regarding other clusters, large subclade-specific differences in diversity were observed. While PCMHs and EUGHs sequences contain little changes in the amino acid composition, diverse patterns were obtained for PRαHs, EUGOs, and 4EPOs, of which the bacterial 4-phenol oxidase branch (EUGOs and 4EPOs) was selected for detailed analysis. Next to the high diversity on the sequence level, oxidases require no co-substrate except readily available dioxygen and represent, therefore, excellent model systems.
a The phylogenetic tree clusters in three major clades (I–III). The respective subclades are named according to characterized enzymes. VAOs vanillyl alcohol oxidases, EUGOs eugenol oxidases, 4EPOs 4-ethylphenol oxidases, PCMHs p-cresol methyl hydroxylases, EUGHs eugenol hydroxylases, PRαH pinoresinol-α-hydroxylases, UE uncharacterized enzymes. Characterized enzymes (light red): VAOs from Diplodia corticola (XP_0201276121) and Penicillium simplicissimum (P56216), PCMH from Pseudomonas aeruginosa (WP_080263564) was not characterized but shares 91% sequence identity to the characterized enzyme from Pseudomonas putida (WP_032489501) which was filtered out by removal of redundancies, EUGH from Pseudomonas sp. OPS1 (AAM21269), PRαHs from Burkholderia sp. SG-MS1 (WP_168666336) and Pseudomonas sp. SG-MS2 (WP_105642528). Characterized enzymes used in this study (light orange): EUGOs from Rhodococcus jostii RHA1 (Q0SBK1) and Nocardioides sp. YR527 (WP_091045259) and 4EPO from Gulosibacter chungangensis (WP_158051316). Enzymes characterized in this study (green): 4-phenol oxidases from Geodermatophilus sabuli (WP_097207849), Geodermatophilaceae bacterium (WP_026846239), Streptomyces cavernae (WP_1283780151), Arthrobacter sp. (WP_0712138341) and Allonocardiopsis opalescens (WP_1062434191). b A2CA results for first shell residues in the catalytic center. The residues are grouped according to the clusters in Fig. 2. The indicated positions correspond to the crystal structure of the EUGO from R. jostii (PDB: 5FXP).
Based on the derived hypotheses (H2 to H4), the oxidases were studied by A2CA with regard to residue size in the A-, T-, and H-cluster and polarity in the W-cluster. The oxidases from Streptomyces cavernae (ScEUGO) and Geodermatophilus sabuli (GsEUGO) were identified with comparably small T-cluster residues (Fig. S1). On the other end of the spectrum, the enzyme from Gulosibacter chungangensis (Gc4EPO) was found to contain sterically demanding residues in the T-cluster, matching the recently described narrow substrate pocket26. As oxidases with mid-sized catalytic pockets, the enzymes from R. jostii (RjEUGO), Geodermatophilaceae bacterium (GbEUGO), and Norcadioides sp. (NspEUGO) were selected. The oxidases from Allonocardiopsis opalescens (AoEUGO) and Arthrobacter sp. UCD-GKA (AspEUGO) stood out in terms of polarity in the W- and H-cluster (Figs. S2 and S3). Further deviations from the consensus in the W-cluster were observed for Gc4EPO (Fig. S3), while no significant changes with respect to residue size or hydrophobicity were observed in the A-cluster (Fig. S4). In total, eight oxidases were selected for this study, of which five have not been described before (Table S1). The enzymes share a sequence identity of 76 to 50% (Table S2). All enzymes were designated as eugenol oxidases (EUGOs) with the exception of Gc4EPO for consistency with earlier studies. As RjEUGO was used as a template, residue numbering refers to this sequence if not stated otherwise.
All enzymes were successfully expressed in E. coli (Figs. S5–S12) and were found to have comparably physical characteristics (Table S3, Figs. S13 and S14). All eight oxidases were tested for their activity on 46 compounds to collect sufficient data for structure-function relations (Table S4). Product formation was validated by GC-MS measurements (Table S5).
Modulation of enzyme activity through substrate rotation by residue 392
Among the selected oxidases, residue 392 in the catalytic center was found to be remarkably variable (Fig. 3b): Five enzymes carry a Gly residue, while Gc4EPO contains a Phe, AspEUGO a Ser, and AoEUGO an Asp. Further, some oxidases from rhodococci harbor a Cys in this position but were not selected for this study due to their high similarity to RjEUGO. This naturally occurring diversity in residue size and polarity coincided with deviations from the expected substrate acceptance of these enzymes. Thus, we investigated the role of this position in detail. For Gc4EPO, a selectivity for non-methoxylated substrates would be expected due to the steric demand of Phe392. But no activity was found for chavicol (1) or 4-hydroxybenzyl alcohol (4), while an outstanding activity for 4-ethyl phenol (32) was observed (4.58 ± 0.18 s−1, Table S4). AoEUGO, harboring a sterically demanding but polar Asp392, was also found incapable of converting 1, while on the contrary, the highest activity for 4 was detected (2.6 ± 0.04 s−1). These drastic changes in substrate acceptance were found to be reasoned in a substrate rotation inside the catalytic pocket. In AoEUGO, the phenolate group of 1 was coordinated by residues 471 and 392, after 25 ns of simulation, instead of the canonical triad, resulting in unfavorable steric interactions of the p-allyl substituent with the W-cluster (Fig. 2b, S15 and S16). In contrast, increased polar interactions for the p-hydroxy substituent in 4 are beneficial for substrate turnover. In the crystal structure of Gc4EPO, interactions of the p-substituent with the W-cluster are reduced as the substrate is rotated in the other direction (Fig. 2c), which is likely caused by the steric effect of Phe392. The greater distance towards the polar cluster is in agreement with the enzyme’s low activities for reactions involving the addition of water and likely contributes to the observed favor for dehydrogenation reactions. The increased activity for these reactions becomes apparent for comparably high activities on 32 as well as vanillyl alcohol (5) and 3,4-dihydroxybenzyl alcohol (7, Table S4).
A substrate rotation could not be observed for AspEUGO as Ser392 is neither influential enough from a steric nor from a polarity aspect (Fig. S17). Nevertheless, steric effects of the residue were identified to increase the affinity for non-methoxylated benzyl alcohols in a comparative study with GbEUGO at optimal conditions for both enzymes (Figs. S18–S21). The KM value of 33 ± 2 µM for 4 was found to be 12 times lower than for GbEUGO (455 ± 50 µM), while similar kcat values around 18 ± 1 s-1 were determined at pH 9.5 (Fig. S22, Table S6). A similar picture is indicated for 3,4-dihydroxybenzyl alcohol (7), but interference of the substrate with the assay made the collection of reliable data difficult (Fig. S23). In contrast, no differences in KM were observed for the o-methoxylated substrates eugenol (2) or vanillyl alcohol (5, Figs. S24 and S25).
Sequence-function relations obtained from correlations between residue variety and enzyme activity
As the example of position 392 highlights the strong effect a single residue can have, the individual influence of all 17 residues of the first shell was investigated to verify the remaining hypotheses H2 to H4. The influence of the T- and H-cluster residues on the o-substitution pattern (H2) was investigated for 4-allyl phenols and 4-hydroxy benzyl alcohols (Fig. 4a). Therefore, the correlation between the logarithm of the activity and the residue size was calculated (Fig. 4b). Increasing activity with increasing residue size results in a positive correlation, while a negative correlation indicates the beneficial effect of small residues. For the T-cluster residues, an increasingly negative correlation with the number of o-methoxy groups is observed, which highlights that smaller residues are required for the acceptance of di-o-methoxylated substrates. For residue 392 of the H-cluster, no clear pattern was observed due to the before-mentioned alternative substrate binding modes, which overcompensate steric effects. Thus, it can be concluded that T-cluster residues have the highest contribution for the selectivity towards di-o-methoxylated substrates, while the repositioning effects of the H-cluster residue 392 is a key factor for the conversion of non-methoxylated substrates. In agreement with this, AoEUGO and Gc4EPO were identified as outliers for the conversion of chavicol (1, Fig. 4d), while all enzymes were active on eugenol (2), justifying the “EUGO” designation (Fig. 4e). In contrast, steric factors dominate the acceptance of 4-allyl-2,6-dimethoxy phenol (3), so that the influence of the T- and H-cluster is visible (Fig. 4f).
Compounds with p-allyl and -benzyl alcohol groups were compared to mitigate the influence of the p-position. a Observed activity on the indicated substrates by the color-coded oxidases of this study (compare Table S2). The error bars represent the standard deviation from a triplicate measurement. b Heatmaps for the correlation of the logarithmic enzyme activity with the change in the size of each individual residue among the selected oxidases. A negative correlation of the amino acid size and the activity means that for enzymes containing larger residues, lower activities were observed and vice versa. Conserved residues have no correlation. c Natural diversity of residues present at the indicated position of each oxidase. The oxidases are color-coded according to a, and the residues are grouped according to functional clusters in the catalytic center of RjEUGO (see Fig. 2a). d Plot of the combined amino acid volume of positions 166, 392, and 427 against the logarithm of the observed enzyme activity for chavicol (1). e Plot of the combined amino acid volume of positions 166, 392, and 427 against the logarithm of the observed enzyme activity for eugenol (2). f Plot of the combined amino acid volume of positions 166, 392, and 427 against the logarithm of the observed enzyme activity for 4-allyl-2,6-dimethoxy phenol (3). A correlation between residue size and activity can be seen.
Next to the number of o-substituents, the acceptance of different chemical groups in o-position could be attributed to the influence of T- and H-cluster residues for benzyl alcohol derivatives as model compounds (Fig. S26). For 3-bromo-4-hydroxybenzyl alcohol (8), the size and polarity of T-cluster residues were identified as determinants for selectivity, while for 3,4-dihydroxybenzyl alcohol (7), the size of the residues in positions 381 and 392 are most influential. This is supported by the dominating steric influence found for Ser392 in AspEUGO (see above). Overall, the halogen substituent seems to interact with the T-cluster, while the o-hydroxy and -methoxy groups rather interact with the H-cluster. In agreement with that, the methoxy group of vanillyl alcohol (5) is directed towards the H-cluster in the crystal structure of RjEUGO (Fig. 2a).
The W-cluster was proposed to have the largest influence on the reaction type (H3). To verify this hypothesis, the conversion of five mono-o-methoxylated substrates was compared (Fig. S27), for which different reactions are observed: alcohol oxidation (vanillyl alcohol, 5), deamination (vanillyl amine, 16), ether cleavage (vanillyl ethyl ether, 12), dehydrogenation or 4α-hydroxylation (4-ethylguiacol, 34), and 4ɣ-hydroxylation (eugenol, 2). A strong influence of the A-cluster and residue 392 were visible. Further, a strong polar effect in position 282 was observed for 12. Here, non-polar residues appeared to be beneficial in this position which is opposite to 16. Thus, ether cleavage and deamination reactions require different polar interactions from the A-cluster. For T-cluster residues, an increasing negative correlation was observed with regard to the chain length of the p-substituent. It can be speculated if increasing repulsion within the A-cluster pushes the substrate molecule towards the tunnel where smaller residues are beneficial to allow an alternative orientation. Along this line, similar patterns are observed for 2 and 12, which is likely caused by the chain length of the p-substituent.
As most selected enzymes are not active on substrates with a chain length larger than three atoms, it was difficult to select suitable substrates to determine factors restricting the activity with regard to the size of the p-substituents (H4). For p-alkyl substrates, Gc4EPO was found the most active enzyme, which strongly biases the analysis (Fig. S28). Diverse activity patterns were only observed for vanillyl butyl ether (13) and 4-cyclopentyl phenol (42). While for the latter no strong correlations were observed, negative correlations were found for several residues in the A-cluster for 13, supporting H4. Notably, the negative correlation for residue 392 in the H-cluster may indicate an orientation of the o-methoxy group towards this cluster.
Catalyst enhancement by site-saturation mutagenesis of hot spot residues
To further investigate H3 and H4, and to utilize the obtained structure-function relations to expand the substrate scope of 4-phenol oxidases, site-saturation mutagenesis was performed (Table S7). For this, a peroxidase-independent screening approach had to be developed (Figs. S29 and S30), as common peroxidases, like horse radish peroxidase, react with the phenolic substrates in a side reaction37,38. Building upon the initial hypotheses, three mutation aims were set: (i) Non-natural substituents in o-position, (ii) alteration of the W-cluster for improved deamination and ether cleavages, and (iii) sterically demanding groups in p-position. ScEUGO was selected as an initial starting point as the wide catalytic pocket was considered most tolerant for residue changes.
For target (i), screening was conducted for A166X and V427X libraries (ScEUGO numbering), resulting in five hits (Table S8). After follow-up investigations in crude extract (Figs. S31 and S32), ScEUGO V427Y was selected as the best candidate, for which a 4.6-fold increase in initial rate (0.36 ± 0.02 s−1) for 3-bromo-4-hydroxybenzyl alcohol (8, Table S9) was found, in comparison to the wildtype. Nevertheless, the performance of the variant was considered insufficient as the obtained activity was still only 16% of the activity of Gc4EPO wildtype on the same substrate (8). Therefore, Gc4EPO itself was chosen for site-saturation mutagenesis. A single hit was obtained from the screening of V166X and I432X libraries (Gc4EPO numbering). The obtained Gc4EPO V166D variant was found to be 6.3-times more active on 8 (14.7 ± 0.5 s−1) than wildtype enzyme (Fig. 5a). Moreover, the mutation restored the activity on 4-hydroxybenzyl alcohol (4), increasing the initial rate more than 90-fold (2.4 ± 0.1 s−1, Fig. 5b). Kinetic studies revealed that the increase in activity is KM-driven as the KM value of the variant towards 8 is about six times higher than for the wildtype (19 ± 1 µM vs. 134 ± 17 µM) (Fig. S33, Table S10). Comparably high KM values were observed for 4 (230 ± 63 µM) and 3,4-dihydroxybenzyl alcohol (9, 106 ± 33 µM, Figs. S34 and S35). Docking experiments and molecular dynamics simulations revealed that D166 interacts with the phenolate of the substrate molecule and the residues of the P-cluster, causing a slight rotation (Fig. S36). This altered binding mode may be responsible for the observed changes in KM values compared to the wild-type enzyme.
The respective improvement to the best-performing enzyme and the wild type is highlighted. The mutants were obtained after screening from a site-saturation mutagenesis library. Rates were determined by xylenol orange assay by applying 2 mM of the substrate in 50 mM potassium phosphate buffer at pH 7.5. The error bars represent the standard deviation from a triplicate measurement. a The performance of Gc4EPO could be 6.3-fold increased by a V166D mutation for 3-bromo-4-hydroxybenzyl alcohol (8), reaching the highest measured rate of 14.5 ± 0.5 s−1. b Activity was restored for Gc4EPO on 4-hydroxybenzyl alcohol (4) by the V166D mutation. The variant is more than 90 times faster than the wild type and only lags behind AoEUGO. c The deamination of vanillyl amine (16) was performed 2.2 times faster by ScEUGO Q425E compared to the best-performing natural enzyme variant, which represents a 3.6-fold improvement compared to the ScEUGO wildtype. d ScEUGO Q425E variant was 1.5-fold faster than the respective wildtype on 4-hydroxy-3,5-dimethoxybenzyl alcohol (6). e The ScEUGO V427I L282M double variant acts 14.3 times faster on 4-cyclopentyl phenol (42) and is 2.5-fold faster than the best-performing wild-type enzyme. The introduction of the second mutation increased the rate 1.3-fold. f On vanillyl butyl ether (13), the ScEUGO V427I variant is 4.3 times faster than the double variant. This represents a 12.3-fold improvement compared to the wildtype and is a 2.5-fold improvement to the best-performing wildtype enzyme.
To increase the sequence space in the W-cluster, ScEUGO E378T and ScEUGO E378Q variants were generated by QuikChange mutagenesis as the limited information obtained from activity correlation suggested that less polar residues (compared to Glu) might be beneficial for deamination reactions. While a strong reduction in 4ɣ-hydroxylation was observed for both variants, minor improvements for 4α-hydroxylation were detected compared to the wild type (Table S9). Both variants reached about 0.2 s−1 on 4-ethylguiacol (34), which represents the second highest rate after the outstanding Gc4EPO (2.4 ± 0.07 s−1). However, the activity for deamination on vanillyl amine (16) was reduced by about 50%. Thus, site-saturation mutagenesis was performed for position 425 in a second approach. ScEUGO Q425E was yielded as a single hit, which was found to be 2.2 times more active for the deamination of 16 than the best-performing wildtype enzymes, RjEUGO and AoEUGO, remarking a 3.6-fold improvement from the ScEUGO wildtype (Fig. 5c). In the structural model, the newly introduced carboxyl group of E425 interacts with the amine group of the substrate, leading to a favorable positioning of the benzylic hydrogens (Fig. S37). This good substrate fit of vanillyl amine is in agreement with an observed KM value of 114 ± 13 µM (Fig. S38). In addition to this, a 1.5-fold faster rate for 4-hydroxy-3,5-dimethoxybenzyl alcohol (6, 4.1 ± 0.6 s−1) was detected, compared to ScEUGO wildtype (Fig. 5d). On the downside, the activity for 4ɣ-hydroxylations was reduced by 90% for eugenol (2) and 70% for 4-allyl-2,6-dimethoxy phenol (3), while no activity was observed at all for 4α-hydroxylations (Table S9). It can be hypothesized that the carboxyl group of E425 interacts with the nucleophilic water, which inhibits the hydroxylation of the substrate molecule.
Before the A-cluster was targeted by mutagenesis according to aim (iii), the catalytic cavity of ScEUGO had to be narrowed down in a first attempt to accommodate substrates without o-substituents as substrates with sterically demanding p-substituents also contain no substituents in o-position. The demand for this strategy was highlighted by an initial screening round on these compounds targeting the A-cluster, which resulted in ScEUGO L381I as a single hit, for which no significant improvement compared to the wild type was detected (Fig. S39).
Since residues in the T-cluster were identified earlier as interaction sites for o-substituents (H2), the libraries A166X and V427X were screened, resulting in three hits (Table S8), of which ScEUGO V427I was identified as the most versatile variant (Fig. S40). Notably, the variant was found to have the fastest initial rate on eugenol (2) ever reported for an EUGO at neutral pH (7.2 ± 0.3 s−1). Regarding substrates with sterically demanding p-substituents, 4-cyclohexyl phenol (42) and vanillyl butyl ether (13), ScEUGO V427I converted both 1.9 and 2.5-times faster than the respective best-performing wildtype enzyme. Compared to ScEUGO wildtype, a 9.5- and 12.3-fold improvement was achieved, respectively (Fig. 5e, f). Thus, the variant was used as a starting point for the second site-saturation mutagenesis round to shape the A-cluster for larger p-substituents. From the correlation data, position 282 was suggested as a target since strong correlations were observed for 13 and 4-propyl phenol (37, Fig. S28). ScEUGO V427I L282M was obtained as a single hit from library screening, adding a 1.3-fold improvement in the activity on 42 (Table S9), which remarks a 2.5-fold improvement compared to RjEUGO as the best-performing wildtype enzyme and a 14.3-fold improvement compared to ScEUGO wildtype (Fig. 5e, f). A KM value of 74 ± 4 µM was determined, which indicates a good substrate fit (Fig. S41). Interestingly, the activity for 13 decreased in comparison to the single variant. Together with the fact that Met is not considerably smaller than Leu, it becomes clear that other effects are responsible for the increase in activity than steric factors. After docking and simulation experiments, the cyclopentyl ring is positioned similarly for the ScEUGO wildtype and the two variants (Fig. S42). Thus, the higher flexibility of the Met may rather allow for an easier planarization of the ring in the methide intermediate (Fig. S43).
Interestingly, ScEUGO V427I L282M reassembles the catalytic pocket of RjEUGO which was the most active wildtype enzyme for 4-cyclopentyl phenol (Fig. S44). Moreover, the catalytic pocket of the ScEUGO V427I variant is similar to the one from NspEUGO which is the most active wildtype enzyme on vanillyl butyl ether (Fig. S45). Thus, observed mutations mirror tendencies in the wild-type enzymes. This represents an important confirmation of to the correlations of residues in the catalytic center with activities for the wild-type enzymes. Further, this observation highlights that the catalytic ability for non-natural substrates is disclosed in the sequence space of the natural enzymes, which underlines the successful logic of the presented approach.
Discussion
In this work, a systematic investigation of the first shell residues in the catalytic center of flavin-dependent 4-phenol oxidoreductases of the VAO/PCMH superfamily was performed. The identification of functional clusters led to the methodical analysis of the enzyme family in the phylogenetic context, which resulted in the identification of bacterial 4-phenol oxidases as a comparably versatile and diverse enzyme family. Thus, this group was chosen as a model system to demonstrate the feasibility of our approach of rational selection of enzymes with novel characteristics based on the chemical properties of the before-defined amino acid clusters. This strategy was enabled by the computational tool A2CA presented in this work, as it connects information from the multiple sequence alignment with the respective phylogenetic data. In total, eight 4-phenol oxidases, of which five were uncharacterized so far, were selected to study the individual influence of every first shell residue on substrate acceptance. Correlations of the residues’ properties with the logarithm of the activity of the natural enzymes allowed the conclusion of sequence-function relations for the acceptance of substrates with variable numbers and types of residues in o-position as well as for the performed reaction type and the size of the residue in p-position. The correlation patterns were supported by kinetic data and structural models to validate hypotheses drawn from literature and propose new theories for previously not investigated residues, like e.g., position 378 (RjEUGO numbering). This fundamental understanding of the sequence–function relations in bacterial 4-phenol oxidases allowed the identification of hot spot residues for subsequent mutagenesis studies. To test hundreds of enzyme variants, a reliable and fast oxidase screening was developed that does not rely on the secondary reaction of a peroxidase, which would interfere with the phenolic substrates of the reaction itself. Site-saturation mutagenesis resulted in sixteen active protein variants and the successful expansion of the substrate scope toward compounds with halogen atoms in the o-position, sterically demanding groups in the p-position, and improved deamination reactions. Five of the obtained variants performed better than all-natural enzymes, while activity improvements of up to 90 times were achieved with respect to the respective wildtype enzyme. The newly introduced amino acids amplified tendencies observed for the natural enzymes and connect well to the correlation studies, which, overall, underlines the successful logic of the presented approach.
In conclusion, we present a time and resource efficient workflow to disclose the natural sequence space of any enzyme family and to leverage this knowledge to expand it towards non-natural substrates. We demonstrated the approach for the family of bacterial 4-phenol oxidases and hope that the overall concept will be expanded to enzyme families as well.
Methods
A2CA analysis
See Supplementary Note 1.
Heterologous protein production
See Supplementary Methods 1, Supplementary Table 1, and Supplementary Figs. 5–12.
Sequence analysis
See Supplementary Methods 2 and Supplementary Table 2.
Thermal stability and substrate conversion
See Supplementary Methods 3 and Supplementary Tables 3–5.
Homology modeling and molecular dynamics simulations
See Supplementary Methods 4.
Buffer optimization and pH screening
See Supplementary Methods 5.
Michaelis-Menten kinetics
See Supplementary Methods 6. Kinetic models used for fitting are provided in Supplementary Eqs. 1 to 5.
Correlation analysis
See Supplementary Methods 7. The formula used for calculation of the respective correlation coefficients is provided in Supplementary Eq. 6.
Oxidase screening
See Supplementary Note 2, Supplementary Methods 8, and Supplementary Figs. 29 and 30. Formulas used for normalization during the screening are provided in Supplementary Eqs. 7 to 9.
Site-directed mutagenesis and characterization of enzyme variants
See Supplementary Methods 8 and Supplementary Tables 7 and 8.
Data availability
Primary data for phylogenetic analysis and for sequence-activity correlations are provided in Supplementary Data 1 and Supplementary Data 2, respectively. Primary data for point diagrams and calibration rows are compiled in Supplementary Data 4. Raw data of the thermal shift assay is provided in Supplementary Data 5. Further data is included in the Supplementary Information and is available from the authors upon request.
References
The UniProt Consortium Ongoing and future developments at the Universal Protein Resource. Nucleic Acids Res. 39, D214–D219 (2011).
Kanehisa, M., Goto, S., Furumichi, M., Tanabe, M. & Hirakawa, M. KEGG for representation and analysis of molecular networks involving diseases and drugs. Nucleic Acids Res. 38, D355–D360 (2010).
Walter, M. C. et al. PEDANT covers all complete RefSeq genomes. Nucleic Acids Res. 37, D408–D411 (2009).
Reetz, M. T. Biocatalysis in organic chemistry and biotechnology: past, present, and future. J. Am. Chem. Soc. 135, 12480–12496 (2013).
Winkler, C. K., Schrittwieser, J. H. & Kroutil, W. Power of biocatalysis for organic synthesis. ACS Cent. Sci. 7, 55–71 (2021).
Wu, S., Snajdrova, R., Moore, J. C., Baldenius, K. & Bornscheuer, U. T. Biocatalysis: enzymatic synthesis for industrial applications. Angew. Chem. Int. Ed. Engl. 60, 88–119 (2021).
Mazurenko, S., Prokop, Z. & Damborsky, J. Machine learning in enzyme engineering. ACS Catal. 10, 1210–1223 (2020).
Otte, K. B. & Hauer, B. Enzyme engineering in the context of novel pathways and products. Curr. Opin. Biotechnol. 35, 16–22 (2015).
Chen, K. & Arnold, F. H. Engineering new catalytic activities in enzymes. Nat. Catal. 3, 203–213 (2020).
Qu, G., Li, A., Acevedo-Rocha, C. G., Sun, Z. & Reetz, M. T. The crucial role of methodology development in directed evolution of selective enzymes. Angew. Chem. Int. Ed. Engl. 59, 13204–13231 (2020).
Wang, Y. et al. Directed evolution: methodologies and applications. Chem. Rev. 121, 12384–12444 (2021).
Chica, R. A., Doucet, N. & Pelletier, J. N. Semi-rational approaches to engineering enzyme activity: combining the benefits of directed evolution and rational design. Curr. Opin. Biotechnol. 16, 378–384 (2005).
Robinson, S. L., Piel, J. & Sunagawa, S. A roadmap for metagenomic enzyme discovery. Nat. Prod. Rep. 38, 1994–2023 (2021).
CoppJ. N. et al. Chapter Twelve—exploring the sequence, function, and evolutionary space of protein superfamilies using sequence similarity networks and phylogenetic reconstructions. In Methods in Enzymology: New Approaches for Flavin Catalysis (ed B. A. Palfey) (Academic Press 2019), 620, pp. 315–347.
Gerlt, J. A. et al. Enzyme function initiative-enzyme similarity tool (EFI-EST): a web tool for generating protein sequence similarity networks. Biochim. Biophys. Acta 1854, 1019–1037 (2015).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Mirdita, M. et al. ColabFold: Making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).
Guo, Y. et al. Structure- and computational-aided engineering of an oxidase to produce isoeugenol from a lignin-derived compound. Nat. Commun. 13, 1–12 (2022).
Guo, Y. et al. One-pot biocatalytic synthesis of rac-syringaresinol from a Lignin-derived phenol. ACS Catal. https://doi.org/10.1021/acscatal.3c04399 (2023).
Gygli, G., de Vries, R. Pde & van Berkel, W. J. H. On the origin of vanillyl alcohol oxidases. Fungal Genet. Biol. 116, 24–32 (2018).
Ewing, T. A., Fraaije, M. W., Mattevi, A. & van Berkel, W. J. H. The VAO/PCMH flavoprotein family. Arch. Biochem. Biophys. 632, 104–117 (2017).
van den Heuvel, R. H. H., Fraaije, M. W., Mattevi, A., Laane, C. & van Berkel, W. J. H. Vanillyl-alcohol oxidase, a tasteful biocatalyst. J. Mol. Catal. B 11, 185–188 (2001).
Eggerichs, D. et al. Vanillyl alcohol oxidase from Diplodia corticola: residues Ala420 and Glu466 allow for efficient catalysis of syringyl derivatives. J. Biol. Chem. 299, 104898 (2023).
de Jong, E., van Berkel, W. J. H., van der Zwan, R. P. & de Bont, J. A. Purification and characterization of vanillyl-alcohol oxidase from Penicillium simplicissimum. A novel aromatic alcohol oxidase containing covalently bound FAD. Eur. J. Biochem. 208, 651–657 (1992).
Jin, J., Mazon, H., van den Heuvel, R. H. H., Janssen, D. B. & Fraaije, M. W. Discovery of a eugenol oxidase from Rhodococcus sp. strain RHA1. FEBS J. 274, 2311–2321 (2007).
Alvigini, L. et al. Discovery, biocatalytic exploration and structural analysis of a 4-ethylphenol oxidase from Gulosibacter chungangensis. Chembiochem 22, 3225–3233 (2021).
Eggerichs, D., Zilske, K. & Tischler, D. Large scale production of vanillin using an eugenol oxidase from Nocardioides sp. YR527. Mol. Catal. 546, 113277 (2023).
Keat, M. J. & Hopper, D. J. p-Cresol and 3,5-xylenol methylhydroxylases in Pseudomonas putida N.C.I.B. 9896. Biochem. J. 175, 649–658 (1978).
Brandt, K., Thewes, S., Overhage, J., Priefert, H. & Steinbüchel, A. Characterization of the eugenol hydroxylase genes (ehyA/ehyB) from the new eugenol-degrading Pseudomonas sp. strain OPS1. Appl. Microbiol. Biotechnol. 56, 724–730 (2001).
Shettigar, M. et al. Oxidative catabolism of (+)-pinoresinol is initiated by an unusual flavocytochrome encoded by translationally coupled genes within a cluster of (+)-pinoresinol-coinduced genes in Pseudomonas sp. Strain SG-MS2. Appl. Environ. Microbiol. https://doi.org/10.1128/AEM.00375-20 (2020).
Cunane, L. M., Chen, Z.-W., McIntire, W. S. & Mathews, F. S. p-Cresol methylhydroxylase: Alteration of the structure of the flavoprotein subunit upon its binding to the cytochrome subunit. Biochemistry 44, 2963–2973 (2005).
McIntire, W., Hopper, D. J. & Singer, T. P. p-Cresol methylhydroxylase. Assay and general properties. Biochem. J. 228, 325–335 (1985).
Fraaije, M. W., van den Heuvel, R. R. H., Roelofs, J. C. & van Berkel, W. J. H. Kinetic mechanism of vanillyl-alcohol oxidase with short-chain 4-alkylphenols. Eur. J. Biochem. 253, 712–719 (1998).
Eggerichs, D. & Tischler, D. Amino Acid Cluster Analysis (A2CA). Science Data Bank. https://doi.org/10.57760/sciencedb.09549 (2023).
Nguyen, Q.-T. et al. Biocatalytic properties and structural analysis of eugenol oxidase from Rhodococcus jostii RHA1: a versatile oxidative biocatalyst. Chembiochem 17, 1359–1366 (2016).
Ewing, T. A. et al. Two tyrosine residues, Tyr-108 and Tyr-503, are responsible for the deprotonation of phenolic substrates in vanillyl-alcohol oxidase. J. Biol. Chem. 292, 14668–14679 (2017).
Debon, A. et al. Ultrahigh-throughput screening enables efficient single-round oxidase remodelling. Nat. Catal. 2, 740–747 (2019).
Weiß, M. S., Pavlidis, I. V., Vickers, C., Höhne, M. & Bornscheuer, U. T. Glycine oxidase based high-throughput solid-phase assay for substrate profiling and directed evolution of (R)- and (S)-selective amine transaminases. Anal. Chem. 86, 11847–11853 (2014).
Acknowledgements
D. Eggerichs was funded by the Deutsche Bundesstiftung Umwelt (20019/625) and was supported by the German Research Council (DFG) within the framework of GRK 2341 (Microbial Substrate Conversion), which was awarded to D. Tischler. The authors thank Prof. M. Fraaije (Groningen, The Netherlands) for provision of the pBADNk vector encoding for RjEUGO.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
D. Eggerichs: Conceptualization, methodology, programming of A2CA, structural analysis, analysis of correlation data, data curation, writing—original draft. N. Weindorf: Calculation of homology models, autodocking, structural analysis, enzyme activity assays, writing. H.G. Weddeling: Site-saturation mutagenesis, development of oxidase screening, enzyme activity assays, Michaelis–Menten kinetics, structural analysis. I.M. Van der Linden: Enzyme characterization, Michaelis–Menten kinetics. D. Tischler: Conceptualization, methodology, supervision, funding acquisition, writing
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Eggerichs, D., Weindorf, N., Weddeling, H.G. et al. Substrate scope expansion of 4-phenol oxidases by rational enzyme selection and sequence-function relations. Commun Chem 7, 123 (2024). https://doi.org/10.1038/s42004-024-01207-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s42004-024-01207-1
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
