Vanillyl alcohol oxidase from Diplodia corticola: Residues Ala420 and Glu466 allow for efficient catalysis of syringyl derivatives

Vanillyl alcohol oxidases (VAOs) belong to the 4-phenol oxidases family and are found predominantly in lignin-degrading ascomycetes. Systematical investigation of the enzyme family at the sequence level resulted in discovery and characterization of the second recombinantly produced VAO member, DcVAO, from Diplodia corticola. Remarkably high activities for 2,6-substituted substrates like 4-allyl-2,6-dimethoxy-phenol (3.5 ± 0.02 U mg−1) or 4-(hydroxymethyl)-2,6-dimethoxyphenol (6.3 ± 0.5 U mg−1) were observed, which could be attributed to a Phe to Ala exchange in the catalytic center. In order to rationalize this rare substrate preference among VAOs, we resurrected and characterized three ancestral enzymes and performed mutagenesis analyses. The results indicate that a Cys/Glu exchange was required to retain activity for ɣ-hydroxylations and shifted the acceptance towards benzyl ethers (up to 4.0 ± 0.1 U mg−1). Our findings contribute to the understanding of the functionality of VAO enzyme group, and with DcVAO, we add a new enzyme to the repertoire of ether cleaving biocatalysts.

Vanillyl alcohol oxidases (VAOs) are a group of the 4-phenol oxidases family, which occur exclusively in ascomycetous fungi. These enzymes are hypothesized to be involved in the degradation of aromatic compounds from natural sources like lignin breakdown or amino acid transformation as the group of vao gene-containing ascomycetes prefers a saprophytic lifestyle or are found to be plant-pathogens and endophytes (1)(2)(3)(4). Phylogenetically, fungal VAOs are related to bacterial eugenol oxidases (EUGOs) and 4-ethyl phenol oxidases (4EPOs) (2,5). Until now, these three groups are characterized solely by a single member each where PsVAO from Penicillium simplicissimum, RjEUGO from Rhodococcus jostii RHA1, and Gc4EPO from Gulosibacter chungangensis represent the VAO, EUGO, and 4EPO groups, respectively (5)(6)(7). The putative VAO from Byssochlamys fulva V107 has been described as well; however, no sequence information is available for this enzyme (4).
Along with alcohol oxidation, dehydrogenation, hydroxylation, and deamination reactions, PsVAO also performs the cleavage of benzyl ethers which are indeed the proposed natural substrates from lignin degradation, as the expression of the corresponding encoding gene was shown to be induced by 4-hydroxy benzyl methyl ether (1,8). This hypothesis is corroborated by the subcellular localization of PsVAO in the peroxisomes through its C-terminal peroxisomal targeting signal (PTS1), as these organelles are involved in the oxidative metabolism (9). RjEUGO and Gc4EPO share similar reaction profiles, but substantial activity on ether substrates has been reported only for PsVAO so far ( Fig. 1) (5,10). This high substrate promiscuity made PsVAO an interesting candidate for biocatalytic applications like the production of vanillin from capsaicin or 2-methoxy-4-methylphenol and the stereoselective hydroxylation of 4-alkylphenol derivatives (11)(12)(13)(14). Based on detailed mechanistic studies and resolved crystal structures, some crucial active site residues have been identified (15,16). This allowed engineering of mutant enzymes displaying for example an inversion of the enantioselectivity for the hydroxylation of ethyl phenol (17,18).
The elucidation of the enzymatic mechanism revealed that in a first step, the covalently bound flavin adenine dinucleotide (FAD) cofactor is reduced by a hydride transferred from the phenolic substrate (11,19). As a consequence, a quinone methide intermediate is formed, which is a strong electrophile for the addition of a water molecule. Lastly, FAD is regenerated by molecular oxygen, yielding hydrogen peroxide as byproduct (2).
Despite the high promiscuity of PsVAO, little attention was drawn to other members of this enzyme family which is partially due to difficulties in expression and purification procedures in heterologous hosts, yielding dysfunctional enzymes trapped in an intermediate oxidation state of the covalent flavin cofactor (20). Yet, we decided to continue investigations on this enzyme family since the complexity of lignin degradation in dependency of the type of wood and the respective habitat of each fungus gives valid reason to expect larger functional diversity among other members than a single enzyme can show. We focused especially on ether cleavages as ethers in general, and the β-O-4 bond in particular, are the most abundant linkages in lignin (21). Besides, ether cleavage itself is challenging from a chemical perspective and often requires harsh methods; therefore, we figured a deeper look into nature's toolbox may reveal potential biocatalysts.
In this work, we systematically evaluated VAO sequences to obtain candidates with altered amino acid compositions in the catalytic center, compared to PsVAO, intending to find novel catalytic features. Based on this sequence analysis, we selected and successfully produced the VAO from Diplodia corticola (DcVAO). This ascomycete fungus is called the "bot canker of oak" and represents a plant-pathogen widespread in the eastern US and south-western Europe, responsible for infection of members of red and white oaks including the cork oak (Quercus suber) (22,23). As DcVAO contains a markedly different amino acid composition in the catalytic center compared to PsVAO, we figured it might be an interesting candidate to get insight into the overall enzyme family and their potential role in plant biomass degradation. Therefore, we studied the substrate scope and kinetic parameters of DcVAO and generated two point mutations to investigate the role of Glu466 in the catalytic center. Further, we performed an evolutionary analysis and resurrect the common ancestor of the whole clade of fungal VAOs plus some of the internal nodes. By this, we aimed to disclose the catalytic potential of the complete group.

Results
Genome mining and phylogenetic analysis allow identifying enzymes with altered amino acid compositions in the catalytic center The sequence of PsVAO was used to perform homology searches in fungal genomes resulting in about 200 sequences which cluster into five different clades (I to V)( Fig. 2A). For a prediction of altered characteristics, we inspected the key residues in the catalytic center of PsVAO in the multiple sequence alignment of the whole data set (Fig. 2B). Val185, Thr459, and Tyr187 (PsVAO numbering) restrict the catalytic pocket at the side of the of the substrate entry tunnel, while Tyr187 has been hypothesized to work as the gatekeeper for oxygen access (24). Asp170 is an essential residue for the substrate deprotonation in benzylic position and assists in covalent flavinylation (25,26). Thr457 and Cys470 are as well in close proximity to the benzylic position of the substrate molecule ( Fig. 2C) and probably modulate the overall reactivity. In the crystal structure of PsVAO (PDB: 2VAO), RjEUGO (5FXE), and Gc4EPO (7BPI), His422 (PsVAO numbering) is responsible for a covalent linkage of the FAD cofactor to the enzyme and restricts, together with Phe424, the catalytic pocket at the opposing side of the substrate entry tunnel (5,16,27). Notably, this Phe is replaced by a Gly residue in RjEUGO (27). With the exception of two enzymes, His422 is conserved among all the sequences analyzed. Further, the catalytic triad formed by Tyr108, Tyr503, and Arg504, which is responsible for the binding of the phenol group of the substrate, is also conserved among all VAOs (data not shown) (16,28). The same is true for Tyr187 and Asp170, which were both described as essential for the reaction mechanism. Differences in the family are rather found for secondary interactions with the substrate. Here, position 459 stands out as fairly variable. Thus, clade III and IV (PsVAO) are distinct from the other clades by the presence of a Thr in position 459 over predominantly Val in clade I and Cys in clade V. In clade II, Thr and Val are common, while this clade also differs from clade III and IV by the occurrence of Glu in position 457 over otherwise Thr. In clade V, Ala is mainly found at this site. Most important interaction points of substrates with groups ortho to the phenolic group are residues in position 424 and 185 (PsVAO numbering). Among the VAO group, the combination of Phe and Val is by far the most common. For position 424, exceptions are found in clade III and V. In clade III, only two VAOs, from D. corticola (DcVAO) and Macrophomina phaseolina (MpVAO) respectively, contain an Ala in position 424. In contrast, Ala, Gly, or Phe are found for enzymes of clade V with decreasing abundance. The introduction of another amino acid in position 424 often coincides with an amino acid change in position 470. These changes are found most of the time in clade V. In fact, DcVAO and MpVAO are the only two VAOs outside of clade V, which do not contain a Cys in position 470. In general, the combination of Ala424 and Glu470 is found predominantly, but Leu, Met, or in single cases Ser and Asn are also observed in position 470.
The combined occurrence of both amino acid variations raised the question how the catalytic properties of the respective enzymes are affected and which factors caused this unusual variation. Therefore, the habitats of selected fungi were included into the considerations. In this regard, D. corticola and M. phaseolina are both known plant pathogens with an overlap in habitat on oak trees (23,29  habitat was observed for Daldinia childiae, which is a member of clade V, and also contains the amino acid combination of Glu470 and Ala424 (according to the homologous positions in PsVAO). Thus, it is possible that the Phe/Ala exchange alters the catalytic pocket to accommodate for 2,6dimethoxylated substrates as they occur in syringyl-rich S-lignin found in hardwood. As we intended to test this hypothesis, DcVAO turned to our focus as D. corticola inhabits a narrow habitat of hardwood oaks. Further, the Cys/Glu exchange in position 470 alone is a remarkable change in the catalytic pocket (Fig. 2D), and we were interested if this residue influences the substrate scope of the enzymes. Interested in ether cleavages, we expected an altered performance as Glu470 could act as additional proton shuttle. Based on these assumptions, we selected DcVAO for further experimental characterization.
DcVAO can be produced heterologously in Escherichia coli as an FAD-saturated enzyme The vao gene from D. corticola was acquired as a DNA fragment via gene synthesis and was successfully cloned into the pET16bp vector using the Gibson assembly method. Difficulties in producing fungal VAOs in bacterial hosts were described before and we also experienced issues in that aspect. Therefore, we tested four different E. coli expression strains and confirmed the production of the enzyme by formation of vanillin and coniferyl alcohol in a whole-cell assay. E. coli NiCo21 (DE3) pKJE7 emerged as best suited host, which encodes for three additional chaperones on the pKJE7 plasmid to promote correct folding of the protein (30,31). Overexpression was performed at 4.5 l scale in a Braun Biotech bioreactor, and the obtained cells were used to provide a crude extract for subsequent protein purification. After His-tag affinity chromatography, a yellow enzyme was obtained with a yield of 9.1 mg per l broth. As reference, we also produced PsVAO as described earlier (32).
Spectral analysis of DcVAO showed three peaks at 273, 360, 441 nm, whereof the peak at 360 nm is characteristic for a covalent-bound flavin cofactor ( Fig. S1) (33). We obtained a ratio 280 to 441 nm of 9.4, which indicates full cofactor saturation. This was validated by comparison of the total protein content determined by Bradford analysis and the concentration for the FAD cofactor using the absorption at 441 nm (ε = 14,200 M −1 cm −1 ). Size-exclusion chromatography revealed the protein to occur in a dimeric and an octameric complex (Fig. S2). The same oligomerization state was described for PsVAO which was attributed to the presence of a so-called octamerization loop which is present for all VAOs but absent for exclusively dimeric EUGOs and 4EPOs (34). Our results strengthen this hypothesis. DcVAO was found to be stable up to a temperature of 50 C after an incubation of 2 h at the respective temperature (Fig. S3). Stability across a broad pH range from 6.5 to 9.5 ( Fig. S4) was observed. No activity could be determined at pH values of 4.5 and below, while at pH of 10.5, a remaining activity of 13% was found.
DcVAO accepts 2,6-dimethoxylated substrates and shows high activity for ether cleavages Enzyme activity for DcVAO and PsVAO were measured in an endpoint kinetic assay, determining the hydrogen peroxide concentration in the solution (xylenol orange assay) (10). Product formation was validated by GC-MS. In total, 38 substrates were tested and both enzymes were found to be active on about two thirds of them (DcVAO: 23, PsVAO: 24) ( Fig. 3 and Table S1). As described in literature, PsVAO was found to perform hydroxylations in αand ɣ-position, oxidation of benzyl alcohols, dehydrogenation, as well as oxidative ether cleavages. The same reaction pattern was also found for DcVAO (Table S2). Similar to PsVAO, we detected a preference for hydroxylations on substrates (Figs. S5-S7). DcVAO performed the hydroxylation of 1,2-bicyclic substrates and ethyl phenol derivatives, while for 4-cyclopentyl phenol (35), only dehydrogenation products were observed. Both enzymes performed rather poorly on deamination reactions, although PsVAO has been described to be capable of performing this type of reaction (14). Notable differences between the two enzymes were observed with regard to the substitution pattern ortho to the phenol moiety, as DcVAO was found to accept 2,6-dimethoxylated substrates. Activities of 3.5 ± 0.02 U mg −1 and 6.3 ± 0.5 U mg −1 on 4-allyl-2,6-dimethoxy-phenol (3) and 4-(hydroxymethyl)-2,6-dimethoxyphenol (6) were observed, respectively, while for PsVAO, only residual activities <0.15 U mg −1 were detected. Further, DcVAO achieved high conversions on vanillyl ether derivatives with activities up to 4.0 ± 0.1 U mg −1 for ethyl vanillyl ether (14). We further investigated this behavior by kinetic studies for both enzymes. The K M value for eugenol is similar for both enzymes, and a cooperative effect according to Hill was observed for both as well (Table 1 and Fig. S8). The cooperative effect is less pronounced for DcVAO (n = 2.6 ± 0.2) as for PsVAO (n = 4.8 ± 0.6), which cannot be fully explained yet as both enzymes were shown to have the same multimerization state. It is likely that the substrate inhibition observed for PsVAO at high concentrations (K i = 12.3 ± 4.5 mM) contributes to this effect. No inhibition was observed for DcVAO. Further, DcVAO has an about two-fold increased K M values for vanillyl alcohol and vanillyl ether derivatives and was found to be more active on vanillyl ethyl (14) and vanillyl butyl ether (15) (Figs. S9-S11). For the later, no activity was observed for PsVAO while DcVAO had a turnover frequency of 2.1 ± 0.1 s −1 (Fig. S12). On the other hand, PsVAO was found to be more active on vanillyl methyl ether (13).
The Michaelis-Menten kinetics for DcVAO and PsVAO showed that DcVAO has overall higher K M values for all substrates except for eugenol. Regarding the k cat , PsVAO has a higher turnover frequency on eugenol (2), vanillyl alcohol (5), and vanillyl methyl ether (13), while DcVAO is twice as active on vanillyl ethyl ether (14) and the only active enzyme on vanillyl butyl ether (15). This highlights a preference of DcVAO for longer chains compared to PsVAO.
As the highest activity of DcVAO was observed for vanillyl ethyl ether, also the highest total turnover number (>13,500 TTN) was reached for this substrate, while for vanillyl methyl Vanillyl alcohol oxidase from Diplodia corticola and butyl ether, a lower initial rate and a lower TTN was observed (>9200 and >7300 TTN, respectively, Fig. S13).

Structural analysis reveals key residues influencing the substrate selectivity
DcVAO was found to accept 2,6-dimethoxylated substrates and to be highly active on vanillyl ethyl derivatives. In order to investigate the molecular origins for these characteristics, we built a homology model for DcVAO based on the PsVAO structure (PDB 2VAO). In addition, a de novo model with AlphaFold2 was generated and compared to the previously obtained homology model and the crystal structure of PsVAO. The three structures aligned well, so that we decided to continue with the homology model for further calculations. Subsequently, a structure-guided analysis of the catalytic 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Table S1. B, substrate selectivity of DcVAO WT and variants compared to PsVAO WT. The reactions were performed with 50 nM of enzyme in potassium phosphate buffer at pH 7.5 and 25 C. Product formation was determined by the hydrogen peroxide production in a colorimetric assay. VAO, vanillyl alcohol oxidase. Vanillyl alcohol oxidase from Diplodia corticola center by docking several substrates to the catalytic cavity of DcVAO was performed and substrate interactions with the amino acids of the active side were evaluated. Although the majority of amino acids is the same for both enzymes (Fig. 2, C and D), notable differences were observed in the positioning of the substrate molecule: while in PsVAO, the methoxy group of 4-ethylguaiacol is directed towards Val185, all docked substrates are turned in DcVAO (Fig. S14) and for vanillyl ethyl ether, the single methoxy group is rather pointed in the direction of Ala420 (Fig. S15). As already pointed out in the phylogenetic analysis, most VAO sequences contain a Phe in position 424, while DcVAO contains an Ala instead (A420). This amino acid change widens the catalytic pocket and generates space for 2,6-dimethoxylated substrates (Figs. S16 and S17), which is in agreement with the higher K M values observed for DcVAO on 2-methoxylated substrates. It can be speculated, if this amino acid variation is related to a habitat containing high amounts of syringol (4-hydroxy-3,5dimethoxycinnamic alcohol), as it is the case for D. corticola or if other structural reasons are the determining factor. Nevertheless, the additional space in the catalytic pocket is likely the reason for single methoxylated substrates to adopt a different binding pose compared to PsVAO.
In position 466 (470 in PsVAO), DcVAO features a Glu over the more common Cys among other members of the family. This residue is located in close proximity to the para residue of the substrate. Docking of 4-allyl-2,6-dimethoxy-phenol (3) revealed that Glu466 is also likely to favor ɣ-hydroxylations as the O Glu -C ɣ distance was determined to be 3.7 Å (Fig. 4A). For 4-(hydroxymethyl)-2,6-dimethoxyphenol (6), a distance of 4.8 Å was found for the O Glu -O β and O Glu -C α interactions (Fig. 4B). Similar values were measured for the O Glu -C α distance (4.7 Å) in the DcVAO homology model docked with vanillyl ethyl ether (14), while the O Glu -O β distance was 0.6 Å shorter (4.1 Å) (Fig. 4C). In contrast, in the crystal structure of PsVAO (PDB 2VAO), a S-C α distance of 5.5 Å was measured (Fig. 4D). The positioning of Glu466 leaves the residue in a prime position to possibly act as a proton shuttle for hydroxylation reactions and may also facilitate ether cleavages.
To investigate this hypothesis, we replaced Glu466 by a Cys and a Leu in DcVAO. Cys is the most common amino acid for related VAOs and also present in PsVAO, while Leu occurs naturally in EUGOs and 4EPOs for which little to no activity for ether cleavages was reported (0.004 s −1 for RjEUGO) (7).
Variation of the residue in position 466 is probably a result from the repositioning of the substrate molecule DcVAO variants were generated by means of the Quik-Change method, and proteins were produced according to the same protocol as for the WT enzyme. In comparison, the variants were produced in lower yields (DcVAO E466C: 3 mg l −1 , DcVAO E466L: 1 mg l −1 ). The substrate scope of the variants was investigated by using the same endpoint kinetic measurements as described before. In total, 33 substrates were tested as we excluded five from the initial set since no conversion was observed neither for DcVAO or PsVAO. For DcVAO E466C and E466L, a significantly altered substrate profile was found: DcVAO E466C was found to retain 50 to 100% activity for the cleavage of vanillyl ethers where the percentage increased with the chain length (Fig. 3B). Unchanged conversion was observed for 5-hydroxyindane (37), 5,6,7,8-tetrahydro-2-naphtol (38), and 3,4-dihydroxybenzyl alcohol (7). For other benzyl alcohol derivatives, like vanillyl alcohol (5), the residual activity was 49%. A stronger reduction in activity was observed for hydroxylation reactions. 4-Ethylguaiacol (20) represents the only converted ethyl phenol derivative with a residual activity of 12%. Even lower activities were observed for the hydroxylation of allyl phenol derivatives resulting for this group in the overall strongest decrease.
For DcVAO E466L, an overall reduced activity was found as well. For p-allyl substrates, a more than ten-fold reduction was observed, while for p-benzyl alcohol derivatives, activities of 17% and lower were found compared to the WT enzyme. A residual activity of about 20% was found for vanillyl methyl (13) and ethyl ether (14), while no conversion could be detected for vanillyl butyl ether (15). The highest activity on ether substrates was observed for the ethyl derivative with 0.76 ± 0.05 U mg −1 . Notably, a comparably lower reduction was found for catechollike substrates. For 3,4-dihydroxybenzyl alcohol (7) and ethyl catechol (19), the residual activity of 90% and 61%, respectively, is rather high compared to other substrates. In case of methyl catechol (24), an activity of 0.83 ± 0.10 U mg −1 was observed, which represents an increase of 182% compared to the WT. As PsVAO is active on catechol-like substrates as well, it can be hypothesized that a hydrophobic interaction next to His422 (PsVAO numbering) might be beneficial for acceptance of these substrates. In case of DcVAO E466L, the respective Leu would perform this task, while Phe424 interacts with the substrate in PsVAO.
The tendency in decreasing activity from the WT over the E466C to the E466L variant highlights the importance of a polar interaction from residue 466 with the substrate in DcVAO. The strongest impact was found for ɣ-hydroxylations, which is in good agreement with the close O Glu -C ɣ distance. These results contrast with observations for PsVAO, as no impact on the activity on eugenol (2) or vanillyl alcohol (5) were observed in earlier studies with the C470L variant (10).
But as PsVAO acts on p-allyl phenols with up to 17.2 ± 0.2 U mg −1 for eugenol (2) and contains a Cys in the respective position, clearly other factors determine the substrate acceptance in DcVAO. The changed positioning of the substrate observed in the in silico analysis is the most likely explanation for this (Fig. S14). This raised the question about when and how the change in substrate orientation occurred. We therefore decided to resurrect three ancestral enzymes to attempt responding this.

Resurrection of ancestral VAOs helps in rationalizing substrate selectivity
To determine the potential substrate scope of the VAO family as well us to understand the sequence determinants of it, ancestral reconstruction was conducted. The common ancestors of the clades including PsVAO (AncPsVAO, PP : 0:94) and the one including DcVAO (AncDcVAO, PP : 0:94) were reconstructed along with the ancestor of the whole group (AncVAO, PP : 0:87) (Fig. 5A). Genes of ancestral enzymes were ordered as synthesized gene fragments, cloned, and  Table S3. 4EPO, 4-ethyl phenol oxidase; EUGO, eugenol oxidase; VAO, vanillyl alcohol oxidase. expressed in the same way as for the DcVAO gene. Enzyme yields were generally lower than DcVAO. AncVAO and AncPsVAO were produced with 2.3 and 2.2 mg per l broth, respectively, while 0.3 mg per l broth of AncDcVAO were obtained.
In total, 37 substrates were tested for conversion by the ancestral enzymes (Table S3). To better trace changes in the selectivity of each enzyme during evolution, the substrates were grouped into eight categories of chemically similar compounds. The percentage of converted substrates for each category gives the substrate profile of the enzyme (Fig. 5B). Ultimately, homology models were calculated for the ancestral proteins to correlate the substrate profile with changes in the amino acid composition.
Over the course of evolution, it seems plausible to speculate that the substrate profile of VAOs first diversified and later specialized as it is the case of modern enzymes. AncVAO shows a narrow selectivity (12 compounds) towards p-allyl phenols, benzyl alcohols, and amines, while ether and alkyl substrates were rarely accepted (Fig. 5B). The highest activities of 1 U mg −1 are observed for chavicol (1), eugenol (2), vanillyl alcohol (5), and 4-(1-hydroxyethyl)phenol (9). As this enzyme is the closest relative to EUGOs and 4EPOs, it is not surprising that the observed substrate profile is similar to that of RjEUGO, which also performs poorly in ether cleavage reactions and hydroxylations of cresol derivatives (10). This is also reflected in the structural model (Fig. S18A): Like RjEUGO, AncVAO harbors a Gly in position 424 and a Val in position 459 (PsVAO numbering). Docking experiments revealed that these residues cause the model substrate eugenol to adopt an alternative binding position which again is comparable to the binding mode observed in the crystal structure of RjEUGO (Fig. S18B). Thus, the positioning of the substrate might be the reason for the EUGO-like substrate scope of AncVAO. The fact that AncVAO contains a Cys in position 470 (PsVAO numbering) while in the structure of RjEUGO a Leu is found seems not to influence the substrate scope.
AncPsVAO and AncDcVAO have both a broader substrate scope than AncVAO (19 and 16 compounds, respectively) but do not accept many substrates in the respective categories where p-allyl phenols and benzyl alcohols are the exception. Both enzymes contain a Phe in position 424 (PsVAO numbering), and docked eugenol adopts a similar binding mode as observed for PsVAO (Figs. S19 and S20A). Further, AncPsVAO and AncDcVAO feature a Cys in position 470. With regard to the reaction profile, first tendencies are visible which carry over to the clades later on: AncPsVAO converts bicyclic compounds and such with longer alkyl chains. Also, the ether cleavage reaction is introduced in the reaction profile, which makes AncPsVAO the most promiscuous ancestral enzyme. PsVAO substrate profile contains 22 converted substrates where a clear selectivity towards longer chains in parapositions can be observed (Fig. 5, B and C). While 40 to 50% of p-alkyl substrates (chain length >1) were accepted by AncPsVAO, the acceptance increases to 60% for p-ethyl to 100% for p-alkyl substrates with a chain length >2 in PsVAO. Further, all cyclic substrates were accepted, which contain a steric demanding group in para-position as well. AncPsVAO is the only ancestor converting 4-hydroxy benzyl methyl ether, indicating that this selectivity was adopted for clade IV during evolution, thus strengthening the assumption that this compound might be the natural substrate for PsVAO (1). For p-allyl phenols, benzyl alcohols, and amines, the acceptance of PsVAO remained similar to AncPsVAO. Structure-wise, the catalytic center of AncPsVAO and PsVAO contain the same amino acid composition (Fig. S19), which is in agreement with the similar substrate profile. As a consequence, the higher activity observed for PsVAO might be due to different structural reasons.
For DcVAO, the same tendency for the chain length of p-alkyl substrates was observed as for PsVAO. But apart from this, the enzyme took a different path. DcVAO is rather versatile in ɣ-hydroxylations and the oxidation of benzyl-alcohol derivatives as the enzyme tolerates many residues in the ortho-positions of the substrate. This is partially reflected in clade III as AncDcVAO was found to accept a variety of benzyl alcohols with activities ranging between <0.1 to 0.31 U mg −1 . This includes 3-bromo-4-hydroxy benzyl alcohol (8, 0.16 ± 0.04 U mg −1 ), which was not converted by any other ancestral enzyme. Despite reasonable acceptance for p-allyl substrates, observed activities for AncDcVAO were below 0.1 U mg −1 for eugenol (2) and 4-allyl-2,6-dimethoxy-phenol (6). As AncDcVAO possesses a Cys in position 466 (DcVAO numbering), the enzyme is more comparable to the DcVAO E466C variant (Fig. S20). Thus, the low activity on allyl phenols is in agreement with the mutagenesis studies. Further, AncDcVAO is the most active ancestral enzyme on vanillyl ethyl ether (14) being the only one converting vanillyl butyl ether (15)(0.23 ± 0.16 and 0.29 ± 0.11 U mg −1 , respectively). This selectivity is in line with the substrate scope of wt DcVAO and E466C, which both accepted 100% of all benzyl ether compounds. A notable difference between AncDcVAO and DcVAO is the ability of the ancestral enzyme to perform deamination reactions. As this activity could be restored for DcVAO E466C on 4-(1-aminoethyl)phenol (10)(0.1 ± 0.03 U mg −1 ), it is tempting to attribute the loss of this reaction type to the introduction of Glu466 which would strongly interact with the amine group. Moreover, Glu466 seems to be mainly responsible for hydroxylation reactions in αand ɣ-position as AncDcVAO and DcVAO E466C were found to possess only residual activities on those substrates. But Glu466 is not solely responsible for the oxidation of alcohol or the cleavage of ether. For the later, the reducing effect in activity can even be offset by longer chain length in para-position as observed for butyl vanillyl ether (15). This is supported by in silico experiments as the distance between the reactive oxygen of Glu466 is closer to the ɣ-carbon atom than to the α-carbon atom of the substrate molecule. Another factor is the orientation of the substrate molecule in the catalytic pocket. Due to the introduction of Ala420, additional space is generated causing the substrates to rotate compared to PsVAO (Fig. S14). Thus, it is likely that the Cys/Glu exchange is necessary to retain hydroxylating activity after adoption towards 2,6-methoxylated substrates by the introduction of Ala420. This hypothesis is strengthened by the fact that both mutations occur only in combination. Nevertheless, the fact that AncDcVAO harbors a Phe in the respective position highlights that additional factors in protein folding may play a role as well.

Discussion
In this study, we describe the second member of the fungal VAO family, the enzyme from D. corticola, referred to as DcVAO. The enzyme was selected after phylogenetic analysis of the VAOs. Special attention was drawn to its amino acid composition in the catalytic center. DcVAO differs from the majority of VAO group members in the two positions Ala420 and Glu466 (DcVAO numbering). Ala420 widens the space in the catalytic pocket for substrates with moieties positioned ortho to the phenolic group. This became obvious comparing the substrate selectivity of DcVAO with the already described VAO from P. simplicissimum (PsVAO) containing a Phe in the respective position, as do most VAOs. DcVAO reached up to 6.3 U mg −1 on 2,6-substituted substrates like 4-(hydroxymethyl)-2,6-dimethoxyphenol (6), while PsVAO was barely active. Since the Ala exchange occurs naturally combined with the exchange of Cys in position 466, we further investigated this change in expectation of an improving effect on ether cleavages. Mutagenesis studies indicated that the introduced Glu466 in DcVAO is important for hydroxylation in αand ɣ-position, while ether cleavage and oxidation of benzylic alcohols was reduced but still possible in the E466C variant. The fact that PsVAO is active on these compounds, although containing a Cys in the same position, highlighted that other factors might be involved as well. In silico experiments revealed a changed orientation of the substrate molecule in the catalytic pocket of DcVAO compared to PsVAO, which is required to accommodate for 2,6-substituted compounds. As a consequence, Glu466 restores the original substrate scope of the enzyme. This observation was further supported by the analysis of the resurrected ancestors, on which AncDcVAO has a similar profile to that of DcVAO E466C variant. While AncDcVAO was found to have low activities in ɣ-hydroxylation reactions, the enzyme was able to perform deamination reactions. This reaction type was absent for DcVAO but could be restored in the DcVAO E466C variant.
It is tempting to attribute the wider catalytic pocket and the acceptance of 2,6-dimethoxylated substrates to the habitat of D. corticola as the fungus is a known pathogen to oaks. These trees contain hardwood, which is rich in 2,6-substituted syringyl lignin. Among the phylogenetic tree, some other examples are found for which the fungal habitat coincides with the amino acid composition in the catalytic center: M. phaseolina (clade III) and D. childiae (clade V) are both plant-pathogen infecting oak trees and contain the amino acid combination of Glu470 and A424, like DcVAO (29,35). Further, plant-pathogenic fungi like Sphaerulina musiva or Dothistroma septosporum with a habitat on softwood trees like conifers, pine, or poplar are found in clade I and harbor the amino acid combination of Cys470 and Phe424 which are the same as found in PsVAO (36,37). However, expression of these enzymes and subsequent characterization in vitro is required to validate this hypothesis drawn from the sequence data.
To summarize the knowledge obtained about the evolution of the VAO family, we can state that AncVAO shows a substrate profile more similar to that of the bacterial EUGO from R. jostii RHA1 and 4EPO from G. chungangensis as it could be expected from the evolutionary history. From there on, AncPsVAO evolved to accommodate substrates with large groups in para-position of the phenolic moiety like bicyclic compounds and such with longer alkyl chains. In contrast, AncDcVAO was found to have a high acceptance towards p-allyl, ether, and benzyl alcohol substrates, although the activity for ɣ-hydroxylations is rather poor. These features are in line with the high activity observed for DcVAO on these compounds where ɣ-hydroxylation was restored by the introduction of Glu466. Thus, we observe a varying degree of specialization of VAO enzymes over time, which highlights the catalytic versatility of this protein family.

Phylogenetic analysis and ancestral sequence reconstruction
Sequence data set was constructed by homology searches in BLASTp using as query PsVAO (Genbank accession P56216). From the raw data set, redundancies and none-fungal sequences were removed. As outgroups, the sequences from EUGO from R. jostii RHA1 (RjEUGO) and 4EPO from G. chungangensis (Gc4EPO) were included. A multiple sequence alignment was constructed by means of the ClustalW algorithm in the Mega 11 software (https://www.megasoftware.net/) (38). A phylogenetic tree was calculated from the alignment using neighbor joining algorithm in Mega 11 and EMBL online services (39,40).
Ancestral sequence reconstruction was performed in PAMLX v.4.9 as marginal reconstruction using CODEML (41). For this, a maximum likelihood phylogeny was inferred in RaxML v8.2.10, with 500 rapid bootstrapping employing an multiple sequence alignment constructed in MAFFT V7 (292 seqs, 652 sites) (42,43). Sequences were analyzed using an empirical substitution matrix, an empirical equilibrium amino acid frequencies, four gamma categories, and LG substitution matrix. The posterior probability distribution of the ancestral states at each site was analyzed at nodes corresponding to AncVAO, AncPsVAO, and AncDcVAO. The length of the ancestors (AncVAO and AncPsVAO, 571 amino acids and AncDcVAO, 570 amino acids) was treated by parsimony analyzing the presence/absence of gaps in the targeted nodes on the basis of the length of the derived sequences in each clade (44). The overall posterior probabilities ðppÞ for each targeted ancestor were determined as the mean value of the PPs per site. Sites were considered ambiguously reconstructed when the alternative states displayed PP >0.2.

Homology modeling and structural analysis
For structural analysis of DcVAO, a homology model calculated by YASARA and an AlphaFold2 structure were generated using the sequence from the NCBI database (Accession number XP_020127612) (45-47). As template for Vanillyl alcohol oxidase from Diplodia corticola homology modeling, the crystal structure of PsVAO (PDB 2VAO) was used as both enzymes share 62.5% sequence identity. The structures were inspected using PyMol (48). Autodocking was performed in YASARA using the VINA autodock software (https://vina.scripps.edu/) (49). The structures were refined by performing molecular dynamics simulation for 5 ns, using the YASARA2 force field, and creating a water-filled periodic simulation box 5 Å around the structures.

Gene synthesis and cloning
All sequences were codon optimized for E. coli and synthesized as a DNA fragment by Twist Bioscience. The genes were cloned blunt end into pJET2.1 vectors using the New England Biolabs kit. After sequencing, genes were cloned into the pET16 bp vector by means of the Gibson assembly (50). Primers are listed in Table S4. The success of the PCR was controlled by agarose gel electrophoresis. The Gibson assembly mix was digested with DpnI for 1.5 h at 37 C before being transformed in competent E. coli DH5a cells, which were plated on LB agar containing 100 μg ml −1 ampicillin. Clones were picked the next day and grown in a 5 ml LB medium (10 g l −1 tryptone, 10 g l −1 NaCl, 5 g l −1 yeast extract, 100 μg ml −1 ampicillin) overnight. Plasmids were isolated by the use of Macherey-Nagel NucleoSpin Plasmid Mini Kit, and the success of cloning was validated by sequencing.

Generation of protein variants
For the generation of DcVAO mutants, QuikChange mutagenesis was applied using the PrimeSTAR Max DNA polymerase from Takara Bio with 25 μl of total reaction volume, 100 ng of pET16 bp_DcVAO plasmid as template, and 10 pmol of the respective primer. Primers were designed for the respective mutation targets as fully and partially overlapping pairs (Table S5). In total, 30 PCR cycles were applied using an annealing temperature recommended for the primer. The success of the PCR was controlled by agarose gel electrophoresis. The PCR product was digested with DpnI for 1.5 h at 37 C before being transformed in competent E. coli DH5a cells, which were plated on LB agar containing 100 μg ml −1 ampicillin. Clones were picked the next day and grown in a 5 ml LB medium (10 g l −1 tryptone, 10 g l −1 NaCl, 5 g l −1 yeast extract, 100 μg ml −1 ampicillin) overnight. Plasmids were isolated by the use of Macherey-Nagel NucleoSpin Plasmid Mini Kit, and mutations were validated by sequencing.

Screening of the expression host
In an initial experiment, the plasmid encoding for DcVAO was transformed into E. coli Rosetta (DE3), E. coli SHuffle (DE3), E. coli NiCo21 (DE3) pKJE7, and E. coli BL21 (DE3). A 50 ml culture of autoinduction medium (12 g l −1 tryptone, 24 g l −1 yeast extract, 100 mM potassium phosphate buffer pH 7.0, 0.5 g l −1 glucose, 2 g l −1 lactose, 100 μg ml −1 ampicillin) was inoculated by a few colonies and grown at 37 C for 4 h before the temperature was reduced to 20 C for another 20 h of incubation. For cells harboring the pKJE7 plasmid, the lactose was replaced by 0.5 g l −1 arabinose and protein expression was induced by a final concentration of 1 mM IPTG after 4 h. The cells were harvested and washed once with 100 mM potassium phosphate buffer at pH 7.0 before a final OD 600 of 30 was established. Five hundred microliters of cell suspension was incubated for 2 h in 1.5 ml reaction tubes containing 2 mM substrate. GC-MS samples were prepared as described below to identify product formation.

Protein production and purification
From a 100 ml preculture in LB medium, 4.5 l of autoinduction medium (12 g l −1 tryptone, 24 g l −1 yeast extract, 100 mM potassium phosphate buffer pH 7.0, 0.5 g l −1 glucose, 0.5 g l −1 arabinose, 100 μg ml −1 ampicillin) in a 5 l fermenter were inoculated to a starting OD 600 of 0.05. The cultures were stirred at 300 rpm at 37 C for 4 h before the temperature was reduced to 20 C, and protein expression was induced by the addition of a final concentration of 1 mM IPTG upon reaching 20 C. The cultures were incubated for another 20 h. During the whole time, 2 l min −1 air were supplied. The final OD 600 ranged between 8 and 10. The cultures were harvested at 5000g for 20 min at 4 C. Cell pellets were washed in 100 mM potassium phosphate buffer pH 7.0 and centrifuged again. The cell pellets were either directly used for cell lysis or stored at −20 C for further use.
For lysis, cells were resuspended in buffer A (10 mM Tris-HCl pH 7.5, 500 mM NaCl) and lysed by sonication. The lysate was centrifuged at 20,000g for 1.5 h at 4 C. The supernatant was collected, filtered through a 0.2 μm pore, and applied to an Äkta Start FPLC system from GE Healthcare. Affinity chromatography was performed using a 5 ml Ni-NTA-HisTrap column from GE Healthcare. The protein was eluted at 100% buffer B (10 mM Tris-HCl pH 7.5, 500 mM NaCl, 500 mM imidazole) while before a washing step of 20% buffer B in buffer A was performed. During the whole process, the absorbance at 280 nm was monitored to collect proteincontaining fractions. The buffer was exchanged to 50 mM potassium phosphate buffer pH 7.5 by means of dialysis overnight at 8 C. Afterward, the dialyzed proteins were concentrated by ultrafiltration (Amicon 10 KDa cut-off) and stored with a final concentration of 60% glycerol at −20 C until further use.

Size-exclusion chromatography
Analytical size-exclusion chromatography was carried out using a Superdex 200 Increase 10/300 Gl column (Cytiva) in 50 mM sodium phosphate at pH 7.2 containing 150 mM sodium chloride. A constant flow rate was applied using a Knauer Azura FPLC system. 150 ml of a 25 mM protein solution was injected to the column. As reference, the Cytiva high molecular weight gel filtration calibration kit was used. Detection was performed at 280 nm.

Determination of protein concentration
The absorption of the covalently bound FAD cofactor was used to measure the amount of active protein in the solution.
Vanillyl alcohol oxidase from Diplodia corticola Measurement was carried out at 441 nm (ε = 14,200 M −1 cm −1 ) (6, 51). The total amount of protein was determined by Bradford assay using the RotiQuant reagent from Roth. Protein dilutions ranging from 10 to 1000 times were prepared in a volume of 200 μl containing a final concentration of 1x RotiQuant. The solution was incubated for 30 min in the dark before the absorption was measured at 595 nm. A standard curve with bovine serum albumin was used for protein quantification.

Xylenol orange assay for enzyme activity measurements
To obtain kinetic data of the enzymes, a 96-well plate-based assay was used to photometrically determine the production of hydrogen peroxide, which is a stochiometric byproduct of VAOs. A variation of this assay was applied before (10); however, the sample points were increased from two to three to improve data quality. During the assay, triplicates of 100 μl, containing a final concentration of 50 mM potassium phosphate buffer pH 7.5, a varying amount of substrate, and 50 nM of the respective enzymes were incubated and shaken at 25 C on a 96-well plate. The reactions were simultaneously started by the addition of enzymes. After 3, 6, and 9 min, 20 μl of each reaction were transferred onto a new 96-well plate, already containing 180 μl of detection solution (250 μM FeSO 4 , 25 mM H 2 SO 4 , 100 μM Xylenol-Orange). In the presence of hydrogen peroxide, the absorption of the dye shifts from 440 nm to 560 nm which was measured after incubation for 30 min. The amount of hydrogen peroxide was determined by a calibration curve, and the slope though all three measurement points was used to calculate the enzyme activity.

pH stability assay
The activity assay was performed similar to the xylenol orange assay but using Briton-Robinson buffer (40 mM acetic acid, 40 mM phosphoric acid, 40 mM boric acid) in 0.5 pH steps, while vanillyl methyl ether at a final concentration of 2 mM was used for all reactions. For pH stability, the enzyme stock solution was incubated in Briton-Robinson buffer for 2 h before the reaction described in the xylenol orange assay was performed.

Temperature stability assay
The activity assay was performed similar to the xylenol orange assay, but the enzyme aliquots were incubated at the respective temperature for 2 h before they were cooled on ice. Vanillyl methyl ether was used at a final concentration of 2 mM for all reactions.

Total turnover assay
The total turnover was determined by measurement of the reaction product vanillin at 350 nm. The reaction was performed in triplicates of 200 μl in a 96-well plate, containing a final concentration of 50 mM potassium phosphate buffer pH 7.5, 10 mM of substrate, and 5 nM of the respective enzyme. The plate was incubated and shaken at 30 C, and the absorption was measured in 15 min intervals. The turnover was calculated as the produced product molecules per active site.

GC-MS measurements
To identify the formed products of the converted substrates, triplicates of 400 μl in 1.5 ml tubes of each substrate were used, containing 2 mM of substrate, 50 mM of potassium phosphate buffer pH 7.5, and 500 nM of the respective enzyme. The reactions were shaken at 25 C and 750 rpm for 16 h. For sample preparation, 300 μl of the solution were extracted with the same volume of ethyl acetate by vortexing for 1 min. Phase separation was initiated by centrifugation for 5 min at 17,000g and room temperature. Afterward, the organic phase was dried over anhydrous MgSO 4 and 100 μl were used for analysis. On a Shimadzu Nexis GC-MS 2030, a 5 C min −1 gradient from 150 to 200 C was applied followed by a 5 min hold at 200 C using a FS-Supreme 5 ms column (length: 30 m, inner diameter: 0.25 mm, outer diameter: 0.36 mm). Compounds were identified by NIST 2017 library search.

Data availability
All data will be available upon request.