The sequences of known VAOs were used as a template for a Blastp search¹ (Altschul et al., 1990). Retrieved results were combined and duplicates removed. Sequences of known bacterial eugenol oxidases, the bacterial counterparts of fungal VAOs, were added to the sequences as an outgroup. A multiple sequence alignment (MSA) in MEGA 11 was performed using the ClustalW algorithm (Tamura et al., 2021; Thompson et al., 1994). The MSA was then used to construct a phylogenetic tree using the maximum likelihood method. Based on known residues that form the active site of VAOs, A²CA (Eggerichs et al., 2024) was then used to compare the constitution of the active site throughout the protein family. Interesting candidates were chosen accordingly.

Genes encoding for the selected VAOs were obtained as Escherichia coli codon-optimized synthetic gene fragments from Twist Biosciences (South San Francisco, CA, USA). Cloning into the pET28a expression vector for E. coli was done using Gibson method (Gibson et al., 2009). Primers for the pET28a backbone and the respective gene fragments with compatible overhangs were designed accordingly for insertion between the NdeI and NotI restriction sites (see Supplementary material, section 2) and used for amplification by PCR. The PCR was performed using the PrimeSTAR Max 2X Master Mix by Takara Bio (Saint-Germain-en-Laye, France). Initial denaturation of templates was done at 98 °C for 2 min, followed by 30 cycles of 20 s denaturation at 98 °C, annealing for 15 s at 60 °C, elongation for 10 s per 1 kb at 72 °C, and a final elongation for 5 min at 72 °C. Successful amplification of the fragments was confirmed via agarose gel electrophoresis and afterwards, a DpnI digestion was performed. The obtained fragments were then used for Gibson assemblies, using the HiFi-Assembly Mix and protocol from NEB (Frankfurt am Main, Germany), followed by heat shock transformation into E. coli DH5α and selection on LB-agar plates containing 35 mg/L kanamycin. Colony-PCR and Sanger-sequencing using T7 and T7term primers were used to identify successful colonies. Plasmids of positive hits were isolated and transformed into E. coli NiCo21(DE3) for heterologous protein production.

Production of DcVAO was done as described previously (Eggerichs et al., 2023a). The other VAOs were produced by heterologous protein production in E. coli NiCo21(DE3). For this, precultures were grown overnight at 37 °C. Precultures were then used to inoculate 1 L cultures in TB medium containing 35 mg/L kanamycin. The cells were grown at 37 °C, 120 rpm in a New Brunswick Innova 42R shaker (Hamburg, Germany) until an OD₆₀₀ of 1.0. Then, the temperature was reduced to 20 °C and IPTG was added to a final concentration of 250 μM. Incubation was continued overnight. Cells were then harvested via centrifugation at 4,000×g using a Beckman Coulter Avanti JXN-26 (Brea, United Kingdom) using a JLA-8.1000 rotor. The sedimented cells were then washed with 30 mL 50 mM potassium phosphate pH 7.5 and centrifuged again. The washed cells were stored at −20 °C for further use.

Approximately 4.5 g wet cells were mixed with 30 mL of 50 mM potassium phosphate pH 7.5, 500 mM NaCl (Buffer A) and then placed on ice until thawed. The cells were then lysed on ice using a Bandolin SonoPlus (Berlin, Germany) sonificator equipped with a MS 72 probe. The total cell lysis procedure involved 30 cycles of 10 s sonication and 10 s of resting, at an amplitude of 30%. The obtained lysates were centrifuged for 45 min at 50,000×g using a Beckman Coulter Anvanti JXN-26 (Brea, United Kingdom) and a JA-25.50 rotor. The obtained supernatant was filtered using a 0.2 μm filter for protein purification using Ni-NTA affinity chromatography using a Cytiva ÄKTA start FPLC (Marlborough, MA, USA) and a 1 mL His-Trap HP column. After washing, the lysate was loaded at a flow rate of 2 mL/min. The column was then washed with 10 column volumes of Buffer A, followed by washing with 10% Buffer B (50 mM potassium phosphate pH 7.5, 500 mM NaCl, 500 mM imidazole), followed by elution of the target protein with 100% Buffer B. The obtained proteins were subjected to buffer exchange using a Cytiva PD-10 column to 50 mM potassium phosphate pH 7.5. UV/Vis spectra were recorded using an Agilent Cary 60 spectrophotometer, and proteins were diluted with glycerol to a final concentration of 50% glycerol for storage at −20 °C. Determination of the protein concentration was done based on the covalently bound cofactor as described (Jin et al., 2007), yielding the amount of FAD-loaded protein.

Screening of the substrate scopes of the new VAOs was done using a xylenol orange-based assay, utilizing the stochiometric production of hydrogen peroxide, allowing for substrate-independent quantification, as it was described earlier for PsVAO, EUGO and variants thereof (Ewing et al., 2018). Reactions contained 2 mM of the tested substrates in 50 mM potassium phosphate pH 7.5 and 50 nM of the respective enzyme, with a total volume of 100 μL. End-point measurements in a 96-well format were performed at 3, 6, and 9 min by mixing 20 μL of the reaction mixtures with 180 μL of xylenol orange solution (625 mM H₂SO₄, 69.8 mg/L FeSO₄, 100 μM xylenol orange) and quantified by recording the absorbance at 560 nm using a Tecan Infinite Pro plate reader (Männedorf, Switzerland). Enzyme activities were calculated according to the corresponding slopes, hydrogen peroxide standards and the amount of used VAO. For determination of the pH-dependence of PvVAO in deamination reactions, the XO-assay was adjusted accordingly.

For the identification of the products formed by the VAOs, reactions were performed in 400 μL 50 mM potassium phosphate pH 7.5 with 2 mM of the respective substrates and 1 μM of the respective VAOs. Reactions in 1.5 mL tubes were incubated overnight at 25 °C while shaking at 750 rpm on an LLG Thermix Pro 2 (Meckenheim, Germany). Reactions were stopped by the addition of 400 μL of ethyl acetate, followed by rapid mixing for 60 s. After phase separation, the organic phase was transferred into a new 1.5 mL tube and dried by the addition of MgSO₄. Afterwards, the samples were analyzed via GC–MS using a Shimadzu QP2020 NX (Kyoto, Japan) and a CS-Chromatographie FS Supreme 5 ms 60 m column. The method consisted of a split temperature of 300 °C, a temperature gradient from 50 to 250 °C at 10 °C/min, column flow of 1.3 mL/min and EI of 1.25 keV. Identification of substrates and products was done via comparison of mass spectra to the NIST-2017 library.

For the generation of the phylogenetic tree, sequences of already known VAOs were used to perform a BLASTp search for fungal homologs (Benen et al., 1998; Eggerichs et al., 2023a; Zhang et al., 2024). In addition, putative VAOs that could not be produced as active proteins in Komagataella phaffii were also included (Gygli and van Berkel, 2017). The results were combined, duplicates removed, and three eugenol oxidases were included as an outgroup (Alvigini et al., 2021; Eggerichs et al., 2025; Jin et al., 2007; Weindorf et al., 2025). After construction of a phylogenetic tree and performing a multiple sequence alignment, and based on knowledge of the constitution of the active site, A²CA was then used to gain a deeper look into the diversity of the active site (Altschul et al., 1990; Eggerichs et al., 2024; Tamura et al., 2021; Thompson et al., 1994). As described earlier, the active site of VAOs and EUGOs can be divided into five clusters, based on their proposed interactions with either the substrate or the covalently bound FAD cofactor (Eggerichs et al., 2024). As a reference, the sequence of PsVAO was used (NCBI: P56216.1). For PsVAO, the A-cluster consists of four residues (L316, W413, V469, and C470) for coordination of the p-substituent of phenolic substrates, the H-cluster consisting of three residues (H61, H422, and L423) for covalent binding of the FAD cofactor, the P-cluster consisting of five residues (Y108, F424, I468, Y503, and R504) for coordination of the phenolic moiety of the substrates, the T-cluster consisting of three residues (G184, V185, and T459) forming the gate between the substrate tunnel and the active site, and the W-cluster consisting of six residues (D170, T188, R312, D409, E410, and T457) involved directly in catalysis as well as in binding and activating water molecules in hydroxylation reactions. For further investigation of the VAOs, focus was put on the diversity of their respective active sites as it was assumed that these would be a promising starting point to identify candidates with differences in the active site and possible changes in their substrate scopes. Figure 1A shows the active site of PsVAO in its entirety and highlights the residues of each cluster with regards to their positioning toward the substrate and FAD cofactor. Figure 1B shows the output of A²CA, linking the sequence motif of the active site for each member of the VAOs to its position in the phylogenetic tree, emphasizing inter-clade conservation of certain residues and conservation across the family, which is not directly obtained through a phylogenetic tree and a MSA alone, thus adding a layer of information on top of these.

Based on the phylogenetic tree alone, the outgroup of the bacterial enzymes can clearly be differentiated from the VAOs. The VAOs can be divided into two subfamilies, the first one spanning over six clades (C1–C6) and the second one containing only clade C7. The division into two fungal subfamilies was reasoned by C1–C6 featuring the characteristic octamerization loop, which has been shown to be responsible for the formation of tetramers of dimers for PsVAO (Ewing et al., 2016), while C7 lacks this loop, which is also the case for the bacterial counterparts (see Supplementary Figures S1, S2). Here, it is also worth noting that the constitution of this loop is also rather conserved within the respective clades.

The first subfamily of the VAOs contains the already described PsVAO (C4), DcVAO (C6), AaVAO (C4), AtVAO (C4), SaVAO (C2) and AmVAO (C1). For C3 and C5 and for the second subfamily (C7), no enzymes have been characterized yet. Focusing a bit more onto the constitution of the active sites, key features for the clades can be observed. Clade C1 differs from the other VAOs in the H-, T-, and W-cluster, where in the H-cluster H61 (PsVAO-numbering), proposed to be involved in autocatalytic covalent flavinylation (Fraaije et al., 2003; Fraaije et al., 2000), is replaced by a phenylalanine or tryptophan. In the T-cluster of this clade, T459 is mostly replaced by a valine or other aliphatic residue, like it is seen for some of the bacterial counterparts, while in the W-cluster, D408 is mostly replaced by a threonine. Clade C2 is closer to the other VAOs, with the change in the T-cluster of T459 to a valine or other aliphatic residue as mentioned for clade C1 is the only major difference. Concerning the amino acid residues that restrict the o-substituents of the phenolic substrates of VAOs (P- and T-cluster), clades C1–C5 are more restricted than clade C6 due to the presence of F424 (PsVAO-numbering). Except for having F424 in the P cluster replaced by a less sterically demanding alanine, the constitution of clade C6 also differs from the other clades by having in the A cluster C470 exchanged by a glutamate, introducing an acid function in the active site, by having H61 in the H cluster replaced by a phenylalanine and L468 exchanged by a sterically less demanding isoleucine or valine, and by having D408 in the W cluster replaced by the larger glutamate.

The second subfamily (clade C7) clearly differs from the rest of the VAOs and is most close to clade C6 from the first subfamily (Figure 1). In the A-cluster of C7, the glutamate found in C6 is mostly retained, while also harboring a tyrosine and the commonly found cysteine among the other VAOs. Further, L316 is mostly replaced by a glutamine, introducing a more hydrophilic group into the A-cluster. The H-cluster is also closest to clade C6 however, H61 can be replaced by a methionine. The P-cluster of C7 is also similar to clade C6 however, F424 can be replaced by a glycine instead of an alanine, which is also seen for the bacterial counterparts. The T-cluster of C7 differs by having T459 mostly exchanged by a cysteine. The W-cluster of C7 also differs by either having D408 replaced by a glutamate, as found in clade C6, or by an arginine, histidine or tyrosine. Overall, the second subfamily shows a higher diversity than the first subfamily. Intrigued by this diversity, five VAOs, AiVAO (NCBI: XP_025378284.1), ExVAO (NCBI: XP_013314277.1), HiVAO (NCBI: XP_024742623.1), PvVAO (NCBI: XP_028486711.1), and TpVAO (NCBI: XP_033679937.1) from the second subfamily were selected for further investigation (see Supplementary material, sequences in section 2, active sites Supplementary Figure S3).

Of the five selected VAOs from the second subfamily, four could be produced as soluble and FAD-loaded proteins in E. coli NiCo21(DE3) using TB medium, while TpVAO was produced as a cofactor-less protein where constitution with FAD was not possible. The others could be produced as FAD-loaded proteins in sufficient yields for investigation, without further optimization (see Supplementary material). It is worth noting that PvVAO presented itself as an orange protein at the beginning of the protein purification and underwent a color change from orange to yellow during the first washing step to remove unbound proteins from the used column.

The xylenol orange-based screening assay (Eggerichs et al., 2024; Ewing et al., 2018) was performed to assess the activity of the four newly produced VAOs with a set of substrates that cover the reactions that can be catalyzed by VAOs (Figure 2). Overall, the new VAOs were active on most of the tested substrates but showed lower specific activities (k_obs values) than the already described PsVAO and DcVAO.

A typical reaction of VAOs is the γ-hydroxylation of 4-allylphenols (see Figure 2A). PsVAO efficiently converted eugenol (1) (17.2 s⁻¹) and chavicol (3) (9.3 s⁻¹), while DcVAO converted next to 1 (9.9 s⁻¹) and 3 (9.2 s⁻¹) also 4-allyl-2,6-dimethoxyphenol (2) (3.5 s⁻¹). ExVAO did not show any activity toward these substrates. AiVAO, HiVAO and PvVAO performed these γ-hydroxylation reactions, but the k_obs values for 1 (0.3 s⁻¹, 1.4 s⁻¹, and 2.3 s⁻¹), 2 (0 s⁻¹, 2.0 s⁻¹, and 2.5 s⁻¹), and 3 (0.6 s⁻¹, 0.4 s⁻¹, and 0.3 s⁻¹) were significantly lower than with PsVAO and DcVAO.

A similar picture was observed for the VAO-mediated oxidation of 4-hydroxybenzyl alcohol (4), 3,4-dihydroxybenzyl alcohol (5), vanillyl alcohol (6), 4-hydroxy-3-bromobenzyl alcohol (7), 4-(1-ethanol)phenol (8) and 4-hydroxy-3,5-dimethoxybenzyl alcohol (9) (see Figure 2B). PsVAO oxidized 5 (2.0 s⁻¹), 6 (3.1 s⁻¹), and 8 (0.2 s⁻¹), DcVAO oxidized 4 (1.2 s⁻¹), 5 (0.5 s⁻¹), 6 (3.4 s⁻¹), 8 (3.5 s⁻¹), and 9 (6.3 s⁻¹), while the new VAOs were in general less active with these compounds. AiVAO was active with 5 (1.7 s⁻¹), 6 (1.0 s⁻¹), and 8 (2.5 s⁻¹), ExVAO with 5 (0.9 s⁻¹), 6 (0.3 s⁻¹), and 9 (0.4 s⁻¹), HiVAO with 4 (0.2 s⁻¹), 5 (0.2 s⁻¹), 6 (0.5 s⁻¹), 8 (0.3 s⁻¹), and 9 (1.2 s⁻¹), and PvVAO was only active with 6 (0.3 s⁻¹) and 9 (0.1 s⁻¹).

For α-hydroxylation (or dehydrogenation) reactions with 4-alkylphenols, a different picture emerged for the reactions with 4-ethylphenol (10), 4-ethylcatechol (11), 4-ethylguaiacol (12), 4-n-propylphenol (13), 4-n-butylphenol (14), 4-isopropylphenol (15) and 4-(1-methylpropyl)phenol (16) (see Figure 2C). PsVAO was active with 10 (3.8 s⁻¹), 11 (1.2 s⁻¹), 12 (7.4 s⁻¹), 13 (4.9 s⁻¹), 14 (1.5 s⁻¹), 15 (1.5 s⁻¹), and 16 (0.4 s⁻¹). For DcVAO, similar results as for PsVAO were obtained with 10 (2.0 s⁻¹), 11 (0.8 s⁻¹), 12 (7.6 s⁻¹), 13 (2.0 s⁻¹), 14 (0.7 s⁻¹), 15 (0.4 s⁻¹), and 16 (0.2 s⁻¹). ExVAO was not able to perform any reaction with 4-alkylphenols, while AiVAO was only active with 15 (0.4 s⁻¹). HiVAO was active with 10 (0.2 s⁻¹), 11 (0.3 s⁻¹), 12 (1.5 s⁻¹), 13 (0.5 s⁻¹), 14 (0.5 s⁻¹), 15 (0.4 s⁻¹), and 16 (1.2 s⁻¹), and PvVAO was active with 11 (0.5 s⁻¹), 12 (0.5 s⁻¹), 13 (1.3 s⁻¹), 14 (4.9 s⁻¹), 15 (2.0 s⁻¹), and 16 (0.4 s⁻¹). Overall, three out of four new VAOs were active with 4-alkylphenols but they generally did not perform as good as PsVAO and DcVAO. PvVAO was not active with 4-ethylphenol but it is worth noting that with sterically demanding p-alkylphenols like 14, 15, and 16, the enzyme performed better than PsVAO and DcVAO, thereby differing from the overall observed substrate scopes.

For the conversion of other substrates of VAOs, phenolic ethers like 4-(methoxymethyl)phenol (17), vanillyl ethyl ether (18), vanillyl butyl ether (19), and phenolic amines like vanillyl amine (20) and 4-(1-aminoethyl)phenol (21) were tested (see Figure 2D). For the phenolic ethers, PsVAO showed activity with 17 (3.0 s⁻¹), 18 (1.8 s⁻¹) and to a lesser extent with 19 (0.1 s⁻¹), while DcVAO performed good on 17 (0.6 s⁻¹), 18 (4.1 s⁻¹) as well as 19 (1.5 s⁻¹). In comparison, none of the new VAOs showed any activity toward any of these ethers, highlighting again a difference between the first and the second subfamily, as this standard reaction of VAOs is not performed by members of the second subfamily. For the phenolic amines, PsVAO and DcVAO did not show activity with 20 under the tested reaction conditions at pH 7.5, while PsVAO showed some activity with 21 (0.2 s⁻¹). Of the new VAOs, only PvVAO showed activity with phenolic amines however, decent for 20 (1.2 s⁻¹) and 21 (1.3 s⁻¹), and presenting itself as a good candidate for this type of reaction.

Identification of the respective products of the new VAOs was done via GC–MS (see Table 1). The oxidation of 4-hydroxybenzyl alcohols and γ-hydroxylation of 4-allylphenols yielded the expected products that have been described for PsVAO (Fraaije et al., 1995) and DcVAO (Eggerichs et al., 2023a). Notable differences were observed in the reactions with 4-alkylphenols. As previously reported, PsVAO mainly produced 1-(4′-hydroxyphenyl)alcohols from short-chain 4-alkylphenols and 1-(4′-hydoxyphenyl)alkenes from more bulky 4-alkylphenols (van den Heuvel et al., 1998, 2001; van den Heuvel et al., 2000b). It was also reported for PsVAO that the low yield of (4′-hydoxyphenyl)alkanones was due to the poor further oxidation of the initially formed (S)-1-(4′-hydroxyphenyl)alcohols, the preferred enantiomeric products of PsVAO (Drijfhout et al., 1998; van den Heuvel et al., 2000a). Here we found that PvVAO successfully catalyzed the oxidation of the initial 1-(4′-hydroxyphenyl)alcohol products to the corresponding ketones, suggesting a different stereospecificity. Interestingly, α-hydroxylation did occur in the reactions of PvVAO with the 4-sec-alkylphenols 15 and 16, resulting in the corresponding tertiary alcohols, suggesting a better access of water to the active site than with PsVAO and also DcVAO (van den Heuvel et al., 1998; van den Heuvel et al., 2000b).

Because PvVAO was found in the initial screening to be rather active with phenolic amines at pH 7.5, the pH-dependency of PvVAO toward 20 and 21 was tested in a pH range from 6.0 to 10.0 using 50 mM potassium phosphate and 50 mM glycine buffer. Britton-Robinson was omitted, as PvVAO did not show any activity using this buffer system. This might be due to an interference of buffer components such as barbital with the applied xylenol orange assay. However, with the glycine buffer system, both substrates 20 and 21 were converted by PvVAO in a pH-dependent manner as described for PsVAO. As seen in Figure 3, both substrates showed an increase in relative activity above pH 8.5, with maximal activity at pH 9.5, which is followed by a steep decline in activity at pH 10.0.