Fabian
Schmitz
a,
Katja
Koschorreck
a,
Frank
Hollmann
b and
Vlada B.
Urlacher
*a
aInstitute of Biochemistry, Heinrich-Heine-University Düsseldorf, Universitätsstraße 1, 40225 Düsseldorf, Germany. E-mail: Vlada.Urlacher@uni-duesseldorf.de
bDepartment of Biotechnology, Delft University of Technology, Van der Maasweg 9, 2629 Hz Delft, The Netherlands
First published on 1st June 2023
Selective aromatic hydroxylation of substituted benzenes provides access to versatile phenolic synthons. Unspecific peroxygenases (UPOs) have been recognised as promising biocatalysts for synthetic chemistry. While UPOs accept diverse substrates and enable a broad range of oxygenation reactions, aromatic hydroxylation reactions catalysed by these enzymes have been rarely described. Here, we report on a UPO from Aspergillus brasiliensis (AbrUPO) heterologously expressed in Pichia pastoris at a concentration of 742 mg per litre that is able to catalyse aromatic hydroxylation of substituted benzenes. The preference of AbrUPO for aromatic or benzylic hydroxylation was found to depend on the number, chemical properties and length of existing ring substituents. While oxidation of ethylbenzene gave ring- and side-chain hydroxylation products at a 1:
1 ratio, increasing the chain-length of the alkyl substituent enhanced the preference for benzylic hydroxylation. With the para-disubstituted p-cymene as a substrate, the chemoselectivity of AbrUPO strongly shifted towards aromatic hydroxylation. All tested substituted phenols resulted in exclusive aromatic hydroxylation. The observed formation of low quantities of quinones was attributed to the inherent peroxidase activity, while further oxidation of benzylic alcohols to ketones was suggested to occur due to both peroxidase and peroxygenase activity of AbrUPO. ‘Overoxidation’ due to peroxidase activity could be completely avoided by adding ascorbic acid and shortening reaction time.
In this study, we identified, produced in recombinant Pichia pastoris (recently reclassified as Komagataella phaffii) and characterised an UPO from Aspergillus brasiliensis (AbrUPO). A set of substituted benzenes, phenols and other compounds were identified as substrates for AbrUPO. We showed that AbrUPO can mediate both benzylic and aromatic oxidation, and depending on ring substituents demonstrates a high chemoselectivity yielding either only ring- or only side-chain hydroxylation products.
Spectral properties of the purified enzyme were measured between 350–700 nm on a Lambda 35 spectrophotometer (Perkin Elmer, Waltham, USA). Peptide-N-amidase PNGase F (New England Biolabs, Frankfurt am Main, Germany) was used to deglycosylate 20 μg of purified AbrUPO under denaturing conditions according to the manufacturer's protocol.
UPO | Host organism | Accession number | Theoretical mol. weight [kDa] | Vol. activity [U l−1] |
---|---|---|---|---|
AbrUPO | Aspergillus brasiliensis | OJJ73116.1 | 29.27 | 93.0 ± 1.6 |
CmiUPO | Coprinellus micaceus | TEB27715.1 | 41.40 | 18.2 ± 1.8 |
GdiUPO | Gymnopilus dilepsis | PPR06026.1 | 40.99 | 2.6 ± 0.2 |
LspUPO | Leucoagaricus sp. | KXN81291.1 | 40.68 | 4.8 ± 0.1 |
PfiUPO | Pestalotiopsis fici | XP_007840602.1 | 27.80 | 18.0 ± 0.2 |
PabUPO | Psathyrella aberdarensis | RXW17550.1 | 41.52 | 5.5 ± 0.2 |
SstUPO | Sphaerobolus stellatus SS14 | KIJ32220.1 | 43.16 | 1.2 ± 0.1 |
All UPOs were expressed in P. pastoris, secreted into the culture medium and exhibited peroxidase activity with 2,2′-azinobis-(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) (Table 1). The UPO from Aspergillus brasiliensis (AbrUPO) that demonstrated the highest activity of 93 ± 1.6 U l−1 was chosen for further investigations. The enzyme was produced in a 7.5 l bioreactor via fed-batch cultivation of recombinant P. pastoris yielding a volumetric peroxidase activity of 23492 ± 192 U l−1 in a final volume of 4.6 l after 10 days (Fig. 1). Purified AbrUPO showed a specific activity of 31.7 ± 1.4 U mg−1 towards ABTS (ESI,† Table S2) which is comparable to other recombinant UPOs.28,29 Based on the specific activity of purified AbrUPO, a concentration of 742 mg per 1 litre of culture medium was calculated which is the highest expression level reported for a heterologously produced UPO in P. pastoris so far.30,31
Oxidised purified AbrUPO showed an absorption spectrum typical for heme-thiolate proteins with a Soret band at 421 nm, α-band at 571 nm, β-band at 540 nm and σ-band at 361 nm (ESI,† Fig. S1), similar to other UPOs.2,32–35AbrUPO belongs to the group of the so-called short UPOs.36 Accordingly, the theoretical MW of this protein is 29 kDa, however, the sequence contains 9 putative N-glycosylation sites. SDS-PAGE analysis revealed a strong band at around 70 kDa (ESI,† Fig. S1). The glycosylation degree of 55% is much higher than in other recombinant UPOs expressed in P. pastoris (ESI,† Table S3). The enzyme exhibited a T50 value of 52 °C and remained stable with over 90% of its initial activity after 60 min incubation at pH ranging from 3 to 8 (ESI,† Fig. S2), which is comparable to other UPOs.37,38
Substrate | Substrate depletion [%] | Product distribution [%] | |||||
---|---|---|---|---|---|---|---|
Benzylic oxidation | Aromatic oxidation | ||||||
a Verified by MS and reference substance. b Verified by MS and NIST20 database. c Traces are detectable after 15 min. | |||||||
1 |
![]() |
37 |
![]() |
![]() |
![]() |
![]() |
![]() |
2 |
![]() |
78 |
![]() |
![]() |
![]() |
![]() |
![]() |
3 |
![]() |
68 |
![]() |
![]() |
![]() |
![]() |
![]() |
4 |
![]() |
62 |
![]() |
![]() |
![]() |
![]() |
|
5 |
![]() |
64 |
![]() |
![]() |
![]() |
![]() |
![]() |
6 |
![]() |
74 |
![]() |
![]() |
![]() |
![]() |
|
7 |
![]() |
83 |
![]() |
![]() |
![]() |
![]() |
|
8 |
![]() |
94 |
![]() |
![]() |
|||
9 |
![]() |
97 |
![]() |
![]() |
|||
10 |
![]() |
>99 |
![]() |
![]() |
|||
11 |
![]() |
89 |
![]() |
![]() |
|||
12 |
![]() |
98 |
![]() |
![]() |
|||
13 |
![]() |
94 |
![]() |
![]() |
|||
14 |
![]() |
>99 |
![]() |
![]() |
m/z = 91![]() |
||
14c , 15 | |||||||
15 |
![]() |
18 |
![]() |
![]() |
|||
16 |
![]() |
93 |
![]() |
![]() |
![]() |
Benzylic hydroxylation of ethylbenzene 1 led to (S)-1-phenylethanol 1a (60%, ee = 18%, ESI,† Fig. S3) which was also prone to overoxidation yielding 7% acetophenone 1b. Substituted benzenes were also tested with PaDa-I, a variant of AaeUPO and one of the most studied UPOs so far, used in many UPO studies as a benchmark.39 PaDa-I led to high conversion of 1 (96%) (ESI,† Table S5), but no aromatic ring oxidation was observed. Interestingly, under the same reaction conditions PaDa-I induced a stronger further oxidation yielding 56% (R)-1-phenylethanol 1a (ee = >99%) and 44% acetophenone 1b (ESI,† Table S6).
Increasing the chain-length of the alkyl substituent to propylbenzene 2 and butylbenzene 3 increased both, the preference for side chain hydroxylation over ring hydroxylation as well as the stereoselectivity of benzylic hydroxylation (Table 1). After 180 min reaction with 3 only 5% of total product accounted to 2-butylbenzene-1,4-diol 3c. Notably, the stereoselectivity switched from (S) for 1-phenyl-1-propanol 2a (ee = 63%) to (R) for 1-phenyl-1-butanol 3a (ee = 99%, ESI,† Fig. S4 and S5) for which we are currently lacking a plausible explanation. Remarkably, oxidation of 1 by chloroperoxidase CfuCPO from Caldaromyces fumago gave (R)-1-phenylethanol (ee = 97%), while oxidation of 2 led to formation of (S)-1-phenyl-1-propanol (ee = 88%).40
PaDa-I demonstrated a high stereoselectivity during hydroxylation of 2 to benzylic alcohol 2a in (R)-configuration (ee = 99%, ESI,† Table S6). In reaction with 3 PaDa-I did not furnish benzylic alcohol 3a but catalysed aliphatic hydroxylation reactions at both adjacent positions of the alkyl chain (ESI,† Table S6). Conversion of 1 and 2 by PaDa-I (ESI,† Table S6) and by several other UPOs (AaeUPO, CglUPO, MroUPO, TteUPO and MthUPO) did not give any ring oxidation products and yielded only products of benzylic oxidation.41 This indicates a strong influence of the shape and amino acid composition of the substrate binding site in UPOs on substrate positioning and enzyme chemoselectivity.
Further activation of the benzylic C–H bond by methyl or ethyl substitution favoured side-chain hydroxylation in case of 4 and 5 (compared to 1) and shifted the chemoselectivity from aromatic ring hydroxylation towards benzylic hydroxylation as observed also with 2. Percentages of ring hydroxylation products with both substrates were very similar. Conversion of isobutylbenzene 6 was similar to butylbenzene 3 and led to very low ring hydroxylation (6%). However, a high ratio of benzylic alcohol 6a to ketone 6b was observed as in the reaction with propylbenzene 2. Aromatic hydroxylation was also observed to some extent in reactions of 3, 5 and 6 with PaDa-I (ESI,† Table S6).
Among mono-substituted benzenes only for 1 the mono-hydroxylated phenol product was detected after 180 min reaction with AbrUPO and identified as 2-ethylphenol 1e. The first OH-group was introduced at ortho-position, but not at para-position as could be expected. The second OH-group was introduced at para-position related to the firstly introduced OH-group. For 2–4 only diols were detectable after 180 min reaction at higher or lower concentrations which were in some cases further oxidised to the corresponding quinones (3–6%). The structures of the diols suggest that also in these cases the first OH-group was introduced at ortho- but not at para-position of the aromatic ring. The absence of hydroxylation at para-position, also favourable for aromatic substitution, can be explained by some steric restrictions present in the active site of AbrUPO.
With the para-disubstituted p-cymene 7AbrUPO demonstrated higher activity than with mono-substituted benzenes, and its chemoselectivity strongly shifted towards aromatic hydroxylation, which may partially be attributed to the electron-donating (+I) effect of the additional substituent (4vs.7). The mono-hydroxylated product was identified as carvacrol (5-isopropyl-2-methylphenol, 7d), which means that p-cymene was first hydroxylated at ortho-position related to the methyl group and not to the isopropyl group. The time course analysis of the AbrUPO-catalysed conversion of 7 (ESI,† Fig. S6) revealed that carvacrol 7d was consumed over time at the benefit of the diol product (7b). The ratio of di- to mono-hydroxylated products also increased with the amount of oxidant (H2O2) applied (ESI,† Fig. S7). PaDa-I-catalysed oxidation of 7 only yielded side-chain hydroxylation products (ESI,† Table S6). Finally, using phenolic starting materials 8–10 (Table 2) resulted in exclusive ring hydroxylation in para-position to the existing OH-group. Substrates 8–9 were converted with high activities (94–97% conversion) to furnish exclusively hydroquinone 4b (91–93%) which was partially oxidised to quinone 4c (7–9%). With thymol 10 as substrate conversion of >99% was achieved. Quinone ratio was higher (33%) than in reactions with 8 and 9. After 15 min reaction 10 was nearly completely converted to the diol 7b (data not shown).
Further oxidation of benzylic alcohols to the corresponding ketones may have the same origin, but may also occur due to the peroxygenase activity of this UPO. For instance, AaeUPO from A. aegerita has been shown to catalyse oxidation of 1-phenylethanol to acetophenone via peroxygenase activity.45 Alcohol dehydrogenation activity has been reported for cytochrome P450s and proposed to occur either via two subsequent H-abstractions from the carbon atom of the alcohol or via the gem-diol formation, which then undergoes dehydration.46 We observed that increasing the chain-length of the alkyl substituent from 1 to 3 increased the ratio of the benzylic alcohol to the corresponding ketone. If in case of 1 and 2 only minor amounts of the ketones 1b and 2b were found, with 3 the ratio has turned in favour of the ketone 3b, the main product (64%) of this reaction (Table 2). In order to shed more light on this aspect, the benzylic alcohols 1-phenylethanol 1a, 1-phenyl-1-propanol 2a, and 1-phenyl-1-butanol 3a were used as substrates of AbrUPO with and without ascorbic acid (Table 3). In presence of ascorbic acid, formation of ketones 1b, 2b, and 3b was observed, which indicates that oxidation of these alcohols occurs, at least to some extent, due to the peroxygenase activity of AbrUPO. The peroxygenase activity towards benzylic alcohols increased in the row from 1a to 3a and reached its maximum with 3a, which was converted to 80% to 3b in the presence of ascorbic acid. Without ascorbic acid, formation of ketones 1b and 2b was much higher, which allows us to assume that the contribution of the peroxidase activity towards 1a and 2a is higher compared to the peroxygenase activity. With 3a only a slight increase in formation of 3b was observed without ascorbic acid, indicating that oxidation of 3a can be mainly attributed to the peroxygenase activity of AbrUPO. These results further suggest that both the peroxidase and the peroxygenase activity are substrate dependent.
When we performed the reactions with ethylbenzene 1, propylbenzene 2 and butylbenzene 3 with and without ascorbic acid for only 15 min, the above described tendencies retained (Table 4). Quinones 1d, 2d and 3d were detected only in reactions without ascorbic acid, which confirms our previous suggestion regarding the formation of quinones due to the peroxidase activity of AbrUPO. More interesting is the ratio of benzylic alcohols to the corresponding ketones in these reactions. Lower ratios of the ketones 1b and 2b in the presence of ascorbic acid than in the absence are in line with our observation that the peroxygenase activity towards 1a and 2b is lower compared to the peroxidase activity of AbrUPO. The high ratio of 3a:
3b of 1
:
1 after addition of ascorbic acid is in good agreement with the observed high peroxygenase activity of AbrUPO towards the alcohol 3a.
When the reactions were performed for 180 min only a minor increase in conversion of 1–3 was observed (Table 2). Destabilisation or inactivation of AbrUPO seems to contribute only marginally to this stagnation since AbrUPO retained around 60% of its initial activity after 4 h at 30 °C (ESI,† Fig. S2). Another reason for that can be the generally low enzyme activity particularly towards 1. Finally, hydrogen peroxide might become limiting in course of the reaction when the peroxidase uses up hydrogen peroxide without increasing the product concentration due to reduction of generated radicals by ascorbic acid.
AbrUPO showed high activity for oxidation of sulphur-containing compounds 17–18 (ESI,† Table S4) as it was described for other UPOs.49 Sulfones were the main products under the investigated reaction conditions. Different terpenes and terpenoids were tested as substrates for AbrUPO as well (ESI,† Table S4). Among those, α-pinene 19 was the best substrate (conversion of 84%), but a product mixture was formed (ESI,† Fig. S28). Under the same conditions, conversion of verbenone 20 achieved 14%, while camphor (21) and valencene (22) were not oxidised at all. The bulky testosterone 23 was oxidised with 10% (ESI,† Fig. S32). The hydroxylated product could not be identified yet. Oxidation of testosterone by UPOs has only been described for CglUPO so far.50
C10–C13 fatty acids 24–27 were tested as substrates as well. Fatty acids were mainly hydroxylated at ω-1 but also at other positions (ESI,† Table S5). Interestingly, in the reactions with lauric acid 26 and tridecanoic acid 27 small amounts of lactones were formed (6–8%) (ESI,† Table S5). Other UPOs like AaeUPO and CciUPO from C. cinerea catalyzed the hydroxylation of lauric acid 26 predominantly at positions ω-1 and ω-2.35,51
In attempt to rationalise our observations we compared the active site of AbrUPO with the active sites of the long UPO PaDa-I and the two short UPOs, CglUPO and HspUPO from Hypoxylon sp. (Fig. 2).37,52 PaDa-I harbours a triad of phenylalanine residues (F69, F121 and F199) in close proximity to the heme group and two phenylalanine residues (F76 and F191) within the substrate access channel. In contrast, AbrUPO as well as HspUPO and CglUPO lack the triad of phenylalanines and possess only one or two phenylalanines close to the heme. Further, AbrUPO contains an additional glutamic acid residue at position 87, while the homologous positions in both other short UPOs are occupied by phenylalanine (F84 and F79, respectively) and in PaDa-I by alanine (A80). Less phenylalanine residues and their different location in the substrate binding site of AbrUPO compared to other UPOs might lead to an altered positioning of the aromatic substrates above the heme and facilitate aromatic oxidation. The influence of the active site amino acids in AbrUPO on chemoselectivity is currently under further investigation.
Compared to other reported UPOs, AbrUPO did not lead to a strong ‘overoxidation’. Our results indicate that further oxidation of benzylic alcohols to ketones occurs due to both peroxidase and peroxygenase activity of AbrUPO, both of which were substrate dependent. Further oxidation of hydroquinones to quinones was attributed to the inherent peroxidase activity of AbrUPO and could be completely avoided by adding ascorbic acid and shortening the reaction time. This makes AbrUPO not only an interesting biocatalyst for synthetic chemistry but also an attractive model for understanding the molecular factors governing the chemoselectivity of heme-thiolate enzymes and the starting enzyme for protein engineering studies.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3re00209h |
This journal is © The Royal Society of Chemistry 2023 |