Altering 2-Hydroxybiphenyl 3-Monooxygenase Regioselectivity by Protein Engineering for the Production of a New Antioxidant

Article abstract of DOI:10.1002/cbic.201700648

2-Hydroxybiphenyl 3-monooxygenase is a flavin-containing NADH-dependent aromatic hydroxylase that oxidizes a broad range of 2-substituted phenols. In order to modulate its activity and selectivity, several residues in the active site pocket were investigated by saturation mutagenesis. Variant M321A demonstrated altered regioselectivity by oxidizing 3-hydroxybiphenyl for the first time, thus enabling the production of a new antioxidant, 3,4-dihydroxybiphenyl, with similar ferric reducing capacity to the well-studied piceatannol. The crystal structure of M321A was determined (2.78 ?), and molecular docking of the 3-substituted phenol provided a rational explanation for the altered regioselectivity. Furthermore, HbpA was found to possess pro-S enantioselectivity towards the production of several chiral sulfoxides, whereas variant M321F exhibited improved enantioselectivity. Based on the biochemical characterization of several mutants, it was suggested that Trp97 stabilized the substrate in the active site, Met223 was involved in NADH entrance or binding to the active site, and Pro320 might facilitate FAD movement.

Full text of DOI:10.1002/cbic.201700648

Accepted Article
Title: Altering 2-hydroxybiphenyl 3-monooxygenase regioselectivity by
protein engineering for the production of a new antioxidant
Authors: Almog Bregman-Cohen, Batel Deri, Shiran Maimon, Yael
Pazy, and Ayelet Fishman
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemBioChem 10.1002/cbic.201700648
Link to VoR: http://dx.doi.org/10.1002/cbic.201700648
A Journal of
www.chembiochem.org

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
Altering 2-hydroxybiphenyl 3-monooxygenase regioselectivity by
protein engineering for the production of a new antioxidant
Almog Bregman-Cohen[a], Batel Deri[a], Shiran Maimon[a], Yael Pazy[b] and Ayelet Fishman* [a]
Abstract: 2-Hydroxybiphenyl 3-monooxygenase (HbpA) is a
flavin-containing NADH-dependent aromatic hydroxylase that
oxidizes a broad range of 2-substituted phenols. In order to
modulate its activity and selectivity, several residues in the active
site pocket were investigated by saturation mutagenesis. Variant
M321A demonstrated altered regioselectivity by oxidizing for the
first time 3-hydroxybiphenyl, thus enabling the production of a
new antioxidant, 3,4-dihydroxybiphenyl, with similar ferric
reducing capacity as the well-studied piceatannol. The crystal
structure of M321A was determined (2.78 Å) and molecular
have changed substrate reactivity by directed evolution,
identifying residues that are important for HbpA activity and
specificity. The double mutant D222V/V368A exhibited increased
activity towards indole, and could catalyze indole derivatives,
which wild type HbpA does not catalyze[4]. In addition, variant
I244V increased the activity towards 2HBP, 2-sec-butylphenol,
and guaiacol, while variant V368A/L417F resulted in increased
activity towards 2-tert-butylphenol, guaiacol and salicylaldehyde[5].
Recently, crystal structures of HbpA with FAD in the “in”
conformation (HbpA-IN) and with bound 2HBP and FAD in the
“out” conformation (HbpA-HBP) were determined enabling
structure-function studies[6]. It was suggested that Gly255 is part
of the substrate entrance cavity as substitution to the large
aromatic residue Phe, resulted in a significant decrease of
enzyme activity[6a]. His48 is located in the active site pocket of
HbpA and was found to be involved in the hydroxylation
mechanism by forming a polar interaction with the hydroxyl group
of the substrate[6]. Residue Asp117 was suggested to stabilize the
His48 conformation in order to enable proper substrate
orientation[6b]. Furthermore, Arg242 is involved in enabling the
movement of FAD, while the role of Trp225 is to properly orient
docking of the 3-substituted phenol provided
a rational
explanation for the altered regioselectivity. Furthermore, HbpA
was found to possess pro-S enantioselectivity towards the
production of several chiral sulfoxides, while variant M321F
exhibited improved enantioselectivity. Based on biochemical
characterization of several mutants, it was suggested that Trp97
stabilizes the substrate in the active site, Met223 is involved in
NADH entrance or binding to the active site, while Pro320 may
facilitate FAD movement.
the substrate when entering the active site[6a]
.
Introduction
To this end, there is still limited knowledge on important residues
in the active site which influence activity and selectivity and there
are no reports on the ability of HbpA or its variants to oxidize
substrates other than 2-substituted phenols (with the exception of
indole). It was our objective to expand the understanding of
structure-function correlations based on the HbpA crystal
structure with a bound substrate by semi-rational mutagenesis.
The new insights on the active site were used for identifying
variants with improved activity and altered selectivity for the
synthesis of products with potential industrial applications. HbpA
was found to selectively oxidize pro-chiral sulfides, while its
variant M321F exhibited elevated enantioselectivty. Furthermore,
it was demonstrated that by a single mutation of Met321 to Ala,
regioselectivity was altered enabling the hydroxylation of 3-
hydroxybiphenyl (3HBP) to 3,4-dihydroxybiphenyl (3,4-DHBP), a
newly described antioxidant.
2-Hydroxybiphenyl 3-monooxygenase (HbpA) (EC 1.14.13.44)
was first isolated from Pseudomonas azelaica HBP1, which can
grow on 2-hydroxybiphenyl (2HBP) as the sole carbon and energy
source[1] . HbpA is a single-component, flavin-containing and
NADH dependent aromatic hydroxylase, which belongs to group
A of Flavin-dependent monooxygenases[2]. It catalyzes the first
step in the degradation pathway of 2HBP forming 2,3-
dihydroxybiphenyl (2,3-DHBP) as the product, whereas molecular
oxygen is the source of the added oxygen in the hydroxyl group
of the product[1, 2b]. The enzyme is a homotetramer with a
molecular mass of 256 kDa, while each subunit contains one
molecule of non-covalently bound flavin adenine dinucleotide
(FAD)[2b]
.
HbpA has a broad substrate range and it catalyzes the
regioselective ortho-hydroxylation of a wide range of 2-substituted
phenols to the corresponding catechols[1, 3]. Two previous studies
Results and Discussion
[a]
[b]
A. Bregman-Cohen, B. Deri, S. Maimon, Prof. Dr. A. Fishman
Department of Biotechnology and Food Engineering
Technion-Israel Institute of Technology, Haifa, 3200003, Israel
E-mail: afishman@tx.technion.ac.il
Dr. Y. Pazy
Technion Center for Structural Biology
Library construction and screening
Recently, the crystal structure of HbpA with bound 2HBP was
determined[6a]. In addition, it was shown in several works
describing FAD-dependent monooxygenases that the
isoalloxazine ring of the FAD cofactor comprises two different
conformations with regard to the bound substrate[7]. The "out"
and "in" conformations refer to FAD facing outside enabling the
entrance of the substrate, and towards the binding pocket to
Lorry I. Lokey Center for Life Sciences and Engineering
Technion-Israel Institute of Technology, Haifa, 3200003, Israel

Supporting information for this article is given via a link at the end of
the document.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
allow hydroxylation, respectively (Figure 1). The "out"
conformation is stabilized by residues Trp293 and Arg242 while
as a result of substrate binding, FAD moves into the binding site
to adopt the "in" conformation[6]. The active site pocket of HbpA
is comprised of several hydrophobic residues, including Trp97,
Met223, Trp225 and Met321[6b]. Trp225 was suggested to
facilitate a proper substrate entrance into the binding pocket[6a]
.
Trp97 is located near the substrate in the opposite side of the
active site pocket and its role in the catalytic mechanism of HbpA
was not examined. Met223 and Met321 have not been
investigated, however it was shown in many proteins that
methionine can interact with aromatic amino acid residues to
stabilize protein structures and complexes through the Met
sulfur-aromatic binding motif[8]. As the native substrate of HbpA
consists of two aromatic compounds, these residues might be
involved in binding of the substrate in the active site.
Furthermore, the HbpA crystal structure revealed that Pro320 is
located close to the substrate hydroxyl moiety, however its
function in the active site pocket is unknown.
In order to expand the understanding of structure-function
correlations of HbpA and modulate its activity and selectivity,
studies focused on residues Trp97, Met223, Pro320 and Met321
(Figure 1). Site-directed saturation mutagenesis libraries were
constructed for these residues by the NNK codon degeneracy
approach. The libraries were comprised of about 30 colonies as
it was shown that for mutagenesis using NNK, a library of 22-35
variants is needed to ensure a 0.95 probability of discovering at
least one of the 2 or 3 top variants[9]. The activity of all variants
was examined in whole cells with 2HBP, by monitoring substrate
consumption and product formation using HPLC. In addition,
active variants showed formation of brown color, which indicated
on catechol formation[5]. This determination is based on the
assumption that if the activity on the native substrate was lost, a
new activity would probably not be obtained[10]. The variants that
retained HbpA activity on 2HBP, were sequenced and purified
for further characterization.
Figure 1. Active site of HbpA. Superposition of HbpA with bound 2HBP and
FAD in the “out” orientation (PDB code 5BRT, yellow[6a]) and HbpA with FAD in
the “in” orientation (PDB code 4CY8, blue[6b]). 2HBP is located above the C4 of
the isoalloxazine ring of FAD in the “in” conformation and forms a polar
interaction (black dashed lines) with His48 (grey sticks). The residues that were
subjected to mutagenesis (W97, M223, P320 and M321) are presented as
yellow sticks (PDB code 5BRT) or blue sticks (PDB code 4CY8). Asp117,
Arg242 and Trp225 were found in previous work to play a role in hydroxylation
mechanism (grey lines).
Kinetic parameters of HbpA variants
Kinetic studies of variants that maintained certain activity on the
native substrate, were performed to determine the specific activity
of NADH oxidation and product formation (Figure 2). In order to
understand the role of each residue in the catalytic mechanism of
HbpA, the steady state kinetic parameters were also measured
and the efficiency of each variant (kcat/Km) was calculated (Table
1).
Following the screening of library Trp97, two variants, W97A and
W97Y, exhibited low product formation. The enzymes were
purified and examination of their activity revealed that these
mutations caused significant decrease of specific activity and
efficiency towards 2HBP. W97A and W97Y demonstrated a 12-
fold and 19-fold decrease in specific activity, respectively, while
both variants exhibited a 5-fold decrease in hydroxylation
efficiency (Figure 2A). Variant W97Y exhibited a 5.5-fold increase
in Km value whereas W97A exhibited a 1.9-fold increase,
suggesting different reduced affinity towards the substrate (Table
1). The significant decrease in activity of both variant suggests
that Trp at this position is important for HbpA activity. In addition,
Trp97 was found to be conserved among flavin monooxygenases
(http://consurf.tau.ac.il/), however none of the structurally
homologous proteins contain Trp at this position[11]. Therefore, it
is proposed that the role of Trp97 is to stabilize 2HBP in the active
site pocket by forming π–π interactions hence supporting its
hydroxylation.
Table 1. Km and kcat values of engineered variants on 2HBP[a],[b]
Km
(µM)
kcat
kcat/Km
Variant
(sec-1)
(sec-1 µM-1)
WT
3.1 ± 0.3
2.26
0.48
0.46
1.97
0.73
1.83
4.54
1.24
2.79
3.94
2.75
3.01
0.73
0.08
0.03
0.30
0.26
0.08
0.06
0.10
0.18
2.2
W97A
W97Y
M223K
M223E
M223I
M223Q
M223A
M321L
M321F
M321V
M321A
6.0 ± 0.9
16.9 ± 2.6
6.5 ± 0.3
2.8 ± 0.3
23.6 ± 2.7
78.7 ± 16.9
12.3 ± 0.9
15.6 ± 1.9
1.79 ± 0.1
15.3 ± 1.1
46.2 ± 7.0
0.18
0.06
Met223 is located in front of the hydroxyl group of 2HBP and it
was suggested to stabilize the protein structure[8]. Replacement
of Met223 with Ala, Gln, Glu, Ile and Lys maintained a certain
degree of activity on 2HBP. Most variants exhibited a decrease in
specific activity and affinity. Replacement of Met to Glu resulted
in a significant decrease in the specific activity towards both
NADH and product formation (Figure 2B). Interestingly, M223E
[a] Determined by monitoring NADH consumption at 340 nm. All values were
[b]
corrected for the endogenous NADH oxidation.
The assays were
performed at 30°C in 50 mM phosphate buffer (pH 7.5) with 0.4 mM NADH
and 0.001 – 0.2 mM 2HBP.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
decreased the activity towards NADH by 3.2-fold, while its activity
towards 2HBP decreased by 2.8-fold resulting in only 10%
uncoupling activity. Considering that its affinity towards 2HBP was
not affected by this substitution (Table 1), decline in its activity
might be explained mostly by the decrease in its hydroxylation
efficiency. Variant M223Q illustrated a 25-fold increase in Km
value, confirming decreased affinity towards the substrate.
Furthermore, replacement of Met to Gln increased NADH
oxidation by 1.3-fold, but resulted in decrease of product
formation by 1.5-fold. These results indicate that Met223 is
involved, but not essential, for enzyme activity. Structural
comparison of HbpA with various homologous proteins revealed
that only phenol hydroxylase (PH) contains Met at this position,
Met277, but this residue is located in a different orientation with
respect to its substrate[11e]. Hence, it is concluded based on the
biochemical characterization that residue Met223 is involved in
NADH entrance or binding to the active site of HbpA.
monooxygenases and none of the homolog enzymes in this family
contain Met at this position in the active site[11]. Therefore,
biochemical results suggest that the amino acid at position 321
can affect electron transfer from NADH to the FAD cofactor in the
active site pocket of HbpA, while Met facilitates the most efficient
coupling of NADH consumption and product formation.
All mutations in library Pro320 resulted in activity loss, therefore it
appears that Pro at this position is crucial for the catalytic reaction.
The short distance (3.6-4.2 Å) of Pro320 from the FAD co-factor
may imply on its involvement in enabling the movement of FAD,
which is imperative for HbpA activity. Pro320 is conserved among
flavin monooxygenases. In aklavinone-11-hydroxylase (RdmE),
para-hydroxybenzoate hydroxylase (pHBH) and PH, this position
next to the FAD is occupied by Pro315, Pro293 and Pro364,
respectively (Figure 3)[11a, 11b, 11e]. Altogether, this highlights the
importance of proline at this position.
Numerous variants from library Met321 were active in the whole-
cell activity screen. The most active variants, M321L, M321F,
M321V and M321A, were purified for kinetic characterization.
Most mutations resulted in increase of specific activity towards
NADH with none or only small effect on product formation (Figure
2C). This improvement of specific activity towards NADH might be
a result of indirect involvement in NADH binding or oxidation by
inducing local conformational changes. Notably, all variants at
residue Met321 that exhibited relatively high hydroxylation activity
are occupied by hydrophobic amino acids, implying that amino
acids at position 321 are not involved in essential polar
interactions in the active site. Only replacement of Met to Phe
exhibited improvement in substrate affinity and enzyme efficiency
(Table 1). A reasonable explanation for this improvement may be
formation of π–π interactions between Phe and the substrate.
Figure 3. Residue P320 is
a conserved amino acid among flavin
monooxygenases. The active site of HbpA was superimposed with RdmE,
pHBH and PH (PDB codes 3IHG, 1IUV and 1PN0, respectively). P320 in HbpA
is shown in yellow sticks and the corresponding Pro residues in RdmE, pHBH
and PH are presented in pink, green and cyan sticks, respectively. The FAD in
the “out” conformation is presented in the same colors as the proline residue.
HbpA performs enantioselective oxidation of sulfides
Sulfoxides are chemical compounds that may possess chirality
and consequently may react differently in the asymmetric
environment of receptors and enzymes[12]. Chiral sulfoxides are
characterized by various biological activities such as flavor and
aroma precursors and antimicrobial properties[13]. However,
sulfoxide chirality may have a significant effect on the safety and
efficiency of many drug products and therefore require special
attention in the pharmaceutical industry[14]. Stereoselectively can
be obtained by the use of biocatalytic reactions utilizing pro-chiral
sulfides[15]. Several oxygenases were investigated for their ability
to perform this biotransformation, however it is still necessary to
Met321 is not
a
conserved residue among flavin
Figure 2. Specific activity of engineered variants on NADH consumption and 2,3-dihydroxybiphenyl (2,3-DHBP) production. All assays were performed at 30 °C
in 50mM phosphate buffer (pH 7.5) with 0.4 mM NADH and 0.2 mM 2HBP. NADH oxidation was determined at 340 nm. All values were corrected for the
endogenous NADH oxidation and have a standard error of ≤10%. 2,3-DHBP formation was determined by measuring product formation with reverse phase
HPLC. The values above the product formation columns represent the uncoupling activity which was calculated as the ratio of the NADH oxidation from the
product formation. Variants that retained HbpA activity on 2HBP by screening libraries of W97 (A), M223 (B) and M321 (C), were purified and their specific
activity was determined.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
increase the variety of enzymes that are available for the
Altering HbpA regioselectivity towards 3-hydroxybiphenyl
Phenolic compounds are secondary metabolites characterized by
various bioactivities, such as antioxidant, anticarcinogenic and
anti-inflammatory effects[19]. Structurally, phenolic compounds
can be defined as molecules with at least one aromatic ring,
bearing one or more hydroxyl groups and may have in addition
other substituents. Various biocatalytic processes for the
production of different phenolic antioxidants were reported in the
past decades, but the commercial availability of these
antioxidants is still limited[20]. HbpA ability to produce antioxidants
has never been reported. Given that phenolic antioxidants consist
of an aromatic ring with two hydroxyl groups at meta and para
position, we examined the activity of HbpA and all variants from
the saturation mutagenesis libraries that maintained activity on
the native substrate, on the isomers 3-hydroxybiphenyl and 4-
hydroxybiphenyl (3HBP and 4HBP, Figure 5). Only variant M321A
demonstrated altered regiospecificity by oxidizing 3HBP to 3,4-
dihydroxybiphenyl (3,4-DHBP), a reaction not performed by wild
type. Production of 3,4-DHBP was measured with HPLC and was
confirmed by GC-mass spectrometry. The specific activity of
variant M321A with 3HBP was 0.015 µmol min-1 mg protein-1,
which is 69-fold lower compared to its activity with 2HBP (Figure
2C). To uncover the basis for the new regioselectivity we
determined the crystal structure of variant M321A at 2.78 Å
resolution (Table S1, Figure S1, PDB 6EM0). The overall
structure of M321A with FAD in the “out” conformation was similar
to the structure of the wild type enzyme, except a slight movement
of Trp225, which also contributed to the enlargment of the binding
site pocket as the replacement of Met with Ala (Figure 6). This
structural comparison suggests that the regiospecificity of HbpA
was altered mostly due to the existence of Ala residue at position
321.
production of commercial chiral sulfoxides[16]
.
As the ability of HbpA to oxidize sulfides was never tested, its
activity on three alkyl aryl sulfides was firstly examined (Figure
4A). The enantiomeric excess (%ee) of wild type oxidation of me-
p-tolyl sulfide, thioanisole and Cl-thioanisole was found to be 46.1
+ 0.5% (proS), 45.7 + 1.3% and 50.9 + 0.3%, respectively (Figure
4B). Then, all variants from the saturation mutagenesis libraries
that retained HbpA activity on 2HBP, were further screened for
improved enantioselectivity on these three sulfides. Only variant
M321F exhibited elevated enantioselectivity on me-p-tolyl sulfide
and Cl-thioanisole compared with wild type HbpA (Figure 4B).
Apparently, the mutation of one residue in the active site pocket
is not enough to upgrade enantioselectivity of HbpA towards the
model substrates investigated, but it highlights the potential for
further improvement. No other member of group A of flavin-
dependent monooxygenases was reported to produce chiral
sulfoxides.
Interestingly,
several
Baeyer-Villiger
monooxygenases, which belong to group B, were reported to
synthesize chiral sulfoxides[17]. Phenylacetone monooxygenase,
4-hydroxyacetophenone monooxygenase and ethionamide
monooxygenase oxidized thioanisole with 43% ee (proR), ≥98%
(proS) and 33% (proS), respectively. In addition, phenylacetone
monooxygenase exhibited 11% ee (proR) while oxidizing of me-
p-tolyl sulfide. All reactions were done with purified enzymes and
the %ee was determined after 24 h[18]
.
Figure 5. The reaction used to examine HbpA regioselectivity. Wild type HbpA
and its engineered variants were screened for their ability to oxidize the para
and meta position of 3-hydroxybiphenyl (3HBP) and 4-hydroxybiphenyl (4HBP),
respectively, to form 3,4-dihydroxybiphenyl.
Next, molecular docking was performed using AutoDock Vina to
provide visualization of 3HBP in the active site pocket of variant
M321A (Figure 7)[21]. It was previously reported that the meta
position of 2HBP, in which the hydroxyl group is added, is oriented
in the active site above C4 of the isoalloxazine ring of FAD, where
molecular oxygen binds prior to hydroxylation[6a]. Therefore, the
criteria for the most probable docking model were low binding
energy as well as ∼5-6 Å distance between C4 of the
isoalloxazine ring of FAD and the para position of 3HBP. Initially,
the native substrate, 2HBP, was re-docked into the wild type
structure (PDB 5BRT). The best model presented high correlation
with the actual crystal structure that contains 2HBP, providing an
Figure 4. Chiral sulfoxide production by HbpA. (A) The model reaction used to
examine the ability of HbpA to oxidize sulfides. R=H for thioanisole, R=CH3 for
me-p-tolyl sulfide and R=Cl for p-chloro-thioanisole. (B) Enantioselectivity (%
proS) of HbpA and variant M321F. Enantioselectivity measurements were
obtained by HPLC using a chiral column.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
affinity of -7.4 kcal mol-1 (Figure S2). Similar docking parameters
were used for docking experiments involving variant M321A and
3HBP as the substrate. The docking model revealed that 3HBP is
located in the active site of M321A in a different conformation
compared to the native substrate (Figure 7A). The distance
between the target carbon of the substrate for hydroxylation and
C4 of the isoalloxazine ring of FAD was found to be 6.0 Å, which
is farther compared with 2HBP that exhibited a distance of 5.3 Å.
In addition, the hydroxyl group of 3HBP, which was shown to form
polar interaction with His48, is located in a distance of 3.0 Å
instead of 2.5 Å in the case of 2HBP. Replacement of Met321 with
the smaller Ala residue enlarged the active site pocket, enabling
the different orientation of 3HBP. Furthermore, replacement of
M321 with Phe exhibited small improvement in enantioselectivity
towards pro-chiral sulfides. Consequently, it seems that a variety
of amino acids at position 321 can be involved in HbpA activity
and selectivity and therefore further examination on residue 321
is required. The slight change in orientation of 3HBP, which is
farther from both the C4 of the isoalloxazine ring of FAD and
His48, may explain the 69-fold decrease in specific activity
towards product formation, compared to 2,3-DHBP.
Evaluating the antioxidant activity of 3,4-dihydroxybiphenyl
Information regarding the antioxidant capacity of 3,4-DHBP is not
available in the literature. Therefore, the FRAP assay, which
measures antioxidant potential of molecules through their ability
to reduce ferric iron to ferrous iron, was used[22]. Due to the
absence of a commercial standard of 3,4-DHBP, the compound
was obtained through the reaction of M321A on 3HBP. As a
control, a sample with only the substrate, 3HBP and NADH was
taken for analysis. The reaction mixture was extracted using ethyl
acetate with subsequent solvent evaporation and dissolution in
buffer. FRAP assay indicated that the control comprising only
3HBP had negligible absorbance (OD593 = 0.006±0.002), while
3,4-DHBP exhibited much higher absorbance (OD593
=
0.531±0.013), which indicates on antioxidant activity. The
antioxidant capacity of 3,4-DHBP was compared to the well-
studied antioxidant piceatannol. This catechol derivative of
resveratrol has a broad spectrum of health beneficial effects, such
as antioxidant effects, anti-inflammatory activities, anticancer and
cardioprotective properties and it can be found in berries, peanuts,
grapes, passion fruit seeds, wine and white tea[23]. Different
concentrations of piceatannol were analyzed by the FRAP
method. An amount of 0.34 mM 3,4-DHBP obtained from a
reaction of M321A with 1 mM 3HBP, had similar antioxidant
activity as 0.39 mM piceatannol. From this comparison it is
concluded that 3,4-DHBP is a strong antioxidant, at least based
on the FRAP assay.
Conclusions
The present work shows for the first time the ability to modulate
the regioselectivity of HbpA by one amino acid replacement.
Variant M321A was able to oxidize 3HBP to form 3,4-DHBP which
was shown to possess antioxidant capacity similar to the well-
studied antioxidant piceatannol. Furthermore, HbpA was
demonstrated to have mild pro-S enantioselectivity towards
different alky aryl sulfides and variant M321F showed a small level
of improvement compared to wild-type. Thus, residue Met321 is
a key determinant in orienting substrates in the active site.
Based on kinetic analysis of several mutants is was proposed that
Trp97 is involved in stabilizing 2HBP in the active site pocket
hence supporting its hydroxylation. Met223 was found to be
involved in NADH entrance or binding to the active site, while
Pro320, a conserved residue among flavin monooxygenases, is
suggested to facilitate movement of FAD. Met321 possibly
influences electron transfer from NADH to FAD, however, a
variety of amino acids at this position can modulate HbpA activity
and selectivity and therefore, further examination of this residue
is warranted. These new insights may allow tailoring biocatalysts
with improved activity and selectivity towards the synthesis of
chiral sulfoxides and antioxidants with potential industrial interest.
Figure 6. Structural comparison of variant M321A with wild type HbpA.
Superposition of M321A (magenta) with HbpA-HBP (cyan, PDB code 5BRT,
rmsd 0.203Å). The overall structure of variant M321A is similar to the structure
of the wild type, except a slight movement of Trp225 and the replacement of
Met to Ala at position 321 leading to enlargement of the active site pocket.
In order to understand why the wild type enzyme is not active on
3HBP, molecular docking of 3HBP was carried out with wild type
HbpA (PDB 5BRT). Several possible substrate conformations
were generated, however, in all of the models the angle between
the substrate and C4 of the isoalloxazine ring of the FAD did not
enable hydroxylation. Superposition of the wild type structure with
variant M321A and its docked 3HBP demonstrated that the
existence of Met at position 321 prevents 3HBP from being
located in the proper orientation in the active site of the wild type
enzyme (Figure 7B).
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
Figure 7. Molecular model of 3HBP in the active site of M321A predicted by AutoDock Vina. The energetically favorable conformation of 3HBP (magenta) was
visualized by PyMOL. The active site of variant M321A with FAD in the “out” conformation (gray) is superimposed onto HbpA-IN with FAD in blue (PDB code
4CY8) and HbpA-HBP (PDB code 5BRT, yellow). (A) The hydrogen bond between the hydroxyl group of the substrate and His48 is shown as cyan dashed lines
(2.5Å and 3.0Å for 2HBP and 3HBP, respectively), while the distance between the target carbon of the substrate for hydroxylation and C4 of the isoalloxazine
ring of FAD is shown as black dashed lines (5.3Å and 6.0Å for 2HBP and 3HBP, respectively). (B) Met321 from the structure of HbpA-HBP, which is shown in
yellow sticks, prevents the positioning of 3HBP in the active site of the wild type enzyme.
Site-directed saturation mutagenesis of HbpA and library
screening: Saturation mutagenesis libraries were generated
using high fidelity Pfu Plus DNA polymerase (Eurx Ltd.,Gdansk,
Poland) according to the manufacturer's instructions. Several
pairs of oligonucleotide primers were designed for this purpose,
Experimental Section
while the target amino acid positions were coded by NNK and
MNN (N represents A, T, G or C, K represents G or T and M
Reagents and materials: Nicotinamide adenine dinucleotide
(NADH), flavin adenine dinucleotide (FAD), 2,3-
represents
C or A). The various primers are listed in
dihydroxybiphenyl, 4-hydroxybiphenyl, thioanisole, methyl p-
tolyl sulfide (me-p-tolyl sulfide), 4-chlorophenyl methyl sulfide
(Cl-thioanisole), sodium dodecyl sulfate (SDS), disodium
phosphate, imidazole and kanamycin were purchased from
Sigma Aldrich Chemical Co. (Rehovot, Israel). 2-
Hydroxybiphenyl, monosodium phosphate and formic acid were
Supplementary Table S2. The variants were grown overnight at
37C in Luria–Bertani medium (5 ml) supplemented with
kanamycin (25µg ml−1). The cells were washed with 50 mM
sodium phosphate buffer pH 8, followed by resuspension with 50
mM sodium phosphate buffer pH 8 (5 ml) containing 1mM of
2HBP. The biotransformation was carried out for 24h at 30ºC
with shaking at 250rpm. Enzyme activity was determined by
analysis of 2,3-DHBP formation using HPLC.
purchased
from
Merck
(Darmstadt,
Germany).
3-
Hydroxybiphenyl was purchased from Acros Organics (NJ, USA).
Water, acetonitrile, methanol, ethanol, isopropanol (IPA),
hexane were purchased from J.T. Baker (PA, USA). Ethyl
acetate was purchased from Gadot (Haifa, Israel). HCl was
purchased from Daejung (Shiheung-city, South Korea). Yeast
extract and tryptone were purchased from Becton, Dickinson &
Company (MD, USA). NaCl was purchased from Bio-Lab
(Jerusalem, Israel).
Expression and purification: HbpA enzyme was purified from
E. coli BL21 (DE3; Novagen, Germany) harboring pET9a/hbpA
gene as previously reported[6a]. In order to ensure the existence
of FAD molecule in each enzyme monomer, samples of HbpA
were incubated overnight with FAD at a molar ratio of 1:1.
Residues of non-bound FAD were removed by dialysis against
20 mM sodium phosphate buffer, pH 7.5. Enzyme concentration
was determined spectrophotometrically based on a molecular
weight of 64.6 kDa and extinction coefficient of 79x103 cm-1 M-1.
Bacterial strains and plasmids: The sequence of the HbpA
gene from P. azelaica HBP1[1] was optimized for production in
Escherichia coli (Optimum Gene™) and synthesized by Gen
Script (Piscataway, NJ, USA). The optimized sequence was
cloned into the pET9a (Novagen, Germany) and a His-tag was
introduced to the N-terminus of the gene as previously
described[6a]. The vector was transformed into E. coli BL21 (DE3;
Novagen, Germany) for over-expression.
Analytical methods: Formation of 2,3-DHBP and 3,4-DHBP
were determined using high performance liquid chromatography
(HPLC) with a Dionex UltiMate 3000 (Thermo Scientific)
equipped with a Gemini 5 μ C18 110A column (5 μm, 4.6×250
mm; Phenomenex, Torrance, CA, USA). The results were
analyzed with Thermo Scientific Dionex Chromeleon software
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
version 7.1 SRI. The isocratic elution was performed with
acetonitrile: H2O 40:60 (v/v). The reaction assay and steady
state kinetics for 2-hydroxibiphenyl and NADH were performed
To define the grid size and location, AutoGrid function from
Autodock was implemented. In the case of wild type HbpA, the
final grid box was centered at 18, -48, and -15 with dimensions
of 10×10×10 and a spacing of 1Ǻ in each dimension. For variant
M321A, the final grid box was centered at 18, -20, and 15 with
dimensions of 10×10×12 and a spacing of 1Ǻ in each dimension.
The exhaustiveness was set at 200. Each run provided 9
possible substrate conformations. More than 10 individual runs
were made in order to receive reliable and comparable results.
The selected conformation had to fulfill two conditions: low
affinity energy (kcal mol-1) and a distance of 5-6Å between C4 of
the isoalloxazine ring of FAD and the target carbon of the
substrate for hydroxylation.
as previously described[6a]
.
The identification of the products 2,3-DHBP and 3,4-DHBP was
determined by GC/MS using a GC 6890N (Agilent Technologies,
Santa Clara, CA, USA) instrument equipped with a capillary
column (30 m x 0.32 mm x 0.25 µm), filled with HP-5 (5%-phenyl-
methylsiloxane) (Agilent Technologies, USA), and an HP-5975
mass spectra detector (Agilent Technologies, USA). The
temperature was programmed as follows: T1 = 70°C; dT/dt =
15°Cmin-1, T2 = 300°C; 18 min, split ratio 1 : 10. Under these
conditions, the retention times were: RT = 8.8 min for 2HBP, RT
= 10.2 min for 3HBP, RT = 10.4 min for 2,3-DHBP and RT = 12.0
min for 3,4-DHBP. Identification of these substances was done
by comparison with commercial standards and with a NIST
library.
Ferric reducing/antioxidant power (FRAP) assay: The
antioxidant capacity of 3HBP, 3,4-DHBP and piceatannol was
determined by FRAP assay (Benzie and Strain, 1996)[22]. FRAP
assay is based on the reduction of the Fe3+-TPTZ complex to the
ferrous form, which forms a blue complex (Fe2+-TPTZ) and
increases the absorption at 593 nm. Briefly, a fresh FRAP reagent
was prepared by mixing acetate buffer (300 mM, pH 3.6) with a
solution of 10 mM TPTZ in 40 mM HCl and 20 mM FeCl3 (10:1:1
volume). Sample solution (10 μl) was added to FRAP reagent
(200 μl) and the absorbance was measured at 593 nm after 4 min.
The reaction assay (1 ml) contained 50 mM phosphate buffer pH
7.5, 1 mM 3HBP, 6 μM protein and 2 mM NADH. The reaction
was stopped after 2 h by the extraction of 3HBP and 3,4-DHBP
with ethyl acetate (1:1 volume) in order to separate them from
other components of the reaction. Phase separation was
facilitated by centrifugation (13,400 x g for 2 min) using a Minispin
centrifuge (Eppendorph, Hamburg, Germany). The organic
supernatant was taken for solvent evaporation with nitrogen and
then resuspended (1:1 volume) in 50 mM phosphate buffer pH 7.5.
Concentrations of 3HBP and 3,4-DHBP in each sample were
determined by HPLC analysis. Different concentrations of the
well-known antioxidant piceatannol in the range of 0.1-1 mM were
used for calibration.
Activity on three pro-chiral sulfides: Wild type HbpA and
variants that retained native activity on 2HBP were further
screened for their enantioselectivity on thioanisole, me-p-tolyl and
Cl-thioanisole (Figure 4A). The reaction assay (1 ml) contained
50 mM phosphate buffer pH 7.5, 1 mM substrate, 1.5 μM protein
and 2 mM NADH. Reactions were stopped after 2 h by the
extraction of the substrates and products with ethyl acetate (1:1
volume). Phase separation was facilitated by centrifugation
(13,400 x g for 2 min) using a Minispin centrifuge (Eppendorph,
Hamburg, Germany). The organic supernatant was taken for
solvent evaporation with nitrogen and then resuspended (1:1
volume) in HPLC-grade ethanol due to the sensitivity of the chiral
column to organic solvents. The enantiomeric excess (%ee) of the
products was determined using HPLC 1100-series instrument
(Agilent Technologies,USA) equipped with LUX 5 µM Cellulose-2
chiral column (250 x 4.6 mm, Phenomenex, Torrance, CA, USA)
as described before[16a]. Detection was carried out with a UV-VIS
detector at 225 nm. The absolute configuration of all sulfoxides
was determined as previously described[16a]
.
Crystallization, data collection and structure determination:
HbpA variant was crystallized as previously described[6a]. X-ray
diffraction data was collected in-house using X-ray beam R-Axis
HTC (Rigaku, Japan). All data were indexed, integrated, scaled
and merged using Mosflm and Scala[24]. The structure of HbpA
variant M321A was solved by molecular replacement using
Phaser[25] and the coordinates of earlier determined HbpA
structure (PDB code 4Z2R). Refinement was performed using
Phenix[26] and Refmac5[27], coupled with rounds of manual model
building, real-space refinement and structure validation
Acknowledgements
A. B-C. is grateful to the Israeli Ministry of Science for their
support. The Russell Berrie Nanotechnology Institute at the
Technion, and the Technion Center of Structural Biology facility
of the Lorry I. Lokey Center for Life Sciences and Engineering are
kindly acknowledged for their support.
performed using COOT[28]
. Data collection, phasing and
refinement statistics are presented in Supplementary Table S1.
Keywords: Crystallography • 3,4-dihydroxybiphenyl • 2-
hydroxybiphenyl 3-monooxygenase (HbpA) • 3-hydroxybiphenyl
• saturation mutagenesis
Molecular docking: AutoDock Vina was used for docking
experiments (Scripps Research Institute, La Jolla, CA, USA)[21]
.
Wild type HbpA and variant M321A were prepared using
PyMol[29]. The FAD in "out" conformation was considered as part
of the enzyme and the receptor was prepared using AutoDock
tools package (Scripps Research Institute). The structure of the
substrates was built using Phenix[26]. Both 2HBP and 3HBP were
treated as flexible ligands by modifying their rotatable torsion,
while M321A and wild type HbpA were considered as a rigid
receptor.
References
[1]
[2]
H. P. Kohler, D. Kohler-Staub, D. D. Focht, Appl. Environ. Microbiol.
1988, 54, 2683-2688.
a) M. M. E. Huijbers, S. Montersino, A. H. Westphal, D. Tischler, W. J. H.
van Berkel, Arch. Biochem. Biophys. 2014, 544, 2-17; b) W. A. Suske,
M. Held, A. Schmid, T. Fleischmann, M. G. Wubbolts, H.-P. E. Kohler, J.
Biol. Chem. 1997, 272, 24257-24265.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
[3]
[4]
[5]
M. Held, W. Suske, A. Schmid, K. H. Engesser, H. P. E. Kohler, B.
Witholt, M. G. Wubbolts, J. Mol. Catal. B: Enzym. 1998, 5, 87-93.
A. Meyer, M. Würsten, A. Schmid, H. P. E. Kohler, B. Witholt, J. Biol.
Chem. 2002, 277, 34161-34167.
A. Meyer, A. Schmid, M. Held, A. H. Westphal, M. Röthlisberger, H. P.
E. Kohler, W. J. H. van Berkel, B. Witholt, J. Biol. Chem. 2002, 277, 5575-
5582.
[6]
[7]
a) M. Kanteev, A. Bregman-Cohen, B. Deri, A. Shahar, N. Adir, A.
Fishman, Biochim. Biophys. Acta, Proteins Proteomics 2015, 1854,
1906-1913; b) C. N. Jensen, T. Mielke, J. E. Farrugia, A. Frank, H. Man,
S. Hart, J. P. Turkenburg, G. Grogan, ChemBioChem 2015, 16, 968-976.
a) D. L. Gatti, B. A. Palfey, M. S. Lah, B. Entsch, V. Massey, D. P. Ballou,
M. L. Ludwig, Science 1994, 266, 110-114; b) B. Palfey, D. P. Ballou, V.
Massey, Biochemistry 1997, 36, 15713-15723.
[8]
[9]
C. C. Valley, A. Cembran, J. D. Perlmutter, A. K. Lewis, N. P. Labello, J.
Gao, J. N. Sachs, J. Biol. Chem. 2012, 287, 34979-34991.
Y. Nov, Appl. Environ. Microbiol. 2012, 78, 258-262.
[10] M. Soskine, D. S. Tawfik, Nat. Rev. Genet. 2010, 11, 572.
[11] a) D. L. Gatti, B. Entsch, D. P. Ballou, M. L. Ludwig, Biochemistry 1996,
35, 567-578; b) Y. Lindqvist, H. Koskiniemi, A. Jansson, T. Sandalova,
R. Schnell, Z. Liu, P. Mäntsälä, J. Niemi, G. Schneider, J. Mol. Biol. 2009,
393, 966-977; c) K. S. Ryan, A. R. Howard-Jones, M. J. Hamill, S. J.
Elliott, C. T. Walsh, C. L. Drennan, Proc. Nat. Acad. Sci. U.S.A. 2007,
104, 15311-15316; d) T. Hiromoto, S. Fujiwara, K. Hosokawa, H.
Yamaguchi, J. Mol. Biol. 2006, 364, 878-896; e) C. Enroth, Acta
Crystallogr., Sect D: Biol. Crystallogr. 2003, 59, 1597-1602.
[12] K. M. Rentsch, J. Biochem. Bioph. Methods 2002, 54, 1-9.
[13] R. Bentley, Chem. Soc. Rev. 2005, 34, 609-624.
[14] R. N. Patel, Coord. Chem. Rev. 2008, 252, 659-701.
[15] J. Legros, J. R. Dehli, C. Bolm, Adv. Synth. Catal. 2005, 347, 19-31.
[16] a) J. Shainsky, K. Bernath-Levin, S. Isaschar-Ovdat, F. Glaser, A.
Fishman, Protein Eng. Des. Sel. 2013, 26, 335-345; b) R. Feingersch, J.
Shainsky, T. K. Wood, A. Fishman, Appl. Environ. Microbiol. 2008, 74,
1555-1566; c) D. R. Boyd, N. D. Sharma, S. A. Haughey, M. A. Kennedy,
B. T. McMurray, G. N. Sheldrake, C. C. Allen, H. Dalton, K. Sproule, J.
Chem. Soc., Perkin Trans. 1 1998, 1929-1934; d) G. de Gonzalo, M. D.
Mihovilovic, M. W. Fraaije, ChemBioChem 2010, 11, 2208-2231.
[17] A. Rioz-Martínez, G. de Gonzalo, D. E. T. Pazmiño, M. W. Fraaije, V.
Gotor, European Journal of Organic Chemistry 2010, 6409-6416.
[18] G. de Gonzalo, G. Ottolina, F. Zambianchi, M. W. Fraaije, G. Carrea, J.
Mol. Catal. B: Enzym. 2006, 39, 91-97.
[19] a) C. J. Weng, G. C. Yen, Cancer Treat. Rev. 2012, 38, 76-87; b) N.
Balasundram, K. Sundram, S. Samman, Food Chem. 2006, 99, 191-203.
[20] a) Y. Achmon, A. Fishman, Appl. Microbiol. Biotechnol. 2015, 99, 1119-
1130; b) T. Furuya, K. Kino, Tetrahedron Lett. 2014, 55, 2853-2855.
[21] O. Trott, A. J. Olson, J. Comput. Chem. 2010, 31, 455-461.
[22] I. F. F. Benzie, J. J. Strain, Anal. Biochem. 1996, 239, 70-76.
[23] a) H. Piotrowska, M. Kucinska, M. Murias, Mutat. Res. Rev. Mutat. Res.
2012, 750, 60-82; b) M. A. Seyed, I. Jantan, S. N. A. Bukhari, K.
Vijayaraghavan, J. Agric. Food. Chem. 2016, 64, 725-737.
[24] A. Leslie, 1992, Joint CCP4 + ESF-EACMB Newsletter on Protein
Crystallography 26.
[25] A. McCoy, Acta Crystallogr., Sect D: Biol. Crystallogr. 2007, 63, 32-41.
[26] P. D. Adams, P. V. Afonine, G. Bunkoczi, V. B. Chen, I. W. Davis, N.
Echols, J. J. Headd, L. W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve,
A. J. McCoy, N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson,
J. S. Richardson, T. C. Terwilliger, P. H. Zwart, Acta Crystallogr., Sect D:
Biol. Crystallogr. 2010, 66, 213-221.
[27] a) G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr., Sect
D: Biol. Crystallogr. 1997, 53, 240-255; b) P. Skubak, G. N. Murshudov,
N. S. Pannu, Acta Crystallogr., Sect D: Biol. Crystallogr. 2004, 60, 2196-
2201.
[28] P. Emsley, K. Cowtan, Acta Crystallogr., Sect D: Biol. Crystallogr. 2004,
60, 2126-2132.
[29] W. L DeLano, 2002, The PyMOL Molecular Graphics System, DeLano
Scientific LLC, San Carlos, CA, USA.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/cbic.201700648
ChemBioChem
FULL PAPER
Entry for the Table of Contents
FULL PAPER
New regiospecificity
antioxidant. One single mutation,
M321A, changed 2-hydroxybiphenyl
3-monooxygenase

new
Almog Bregman-Cohen, Batel Deri, Shiran
Maimon, Dr. Yael Pazy and Prof. Dr. Ayelet
Fishman*
(HbpA)
regioselectivity from oxidizing 2-
substituted phenols to 3-substituted
phenols. The crystal structure
provided an explanation for the new
activity. The antioxidant produced,
3,4-dihydroxybiphenyl, had similar
ferric reducing capacity as the well-
studied piceatannol.
1 – 9
Altering 2-hydroxybiphenyl 3-
monooxygenase regioselectivity by
protein engineering for the
production of a new antioxidant
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.