Controlling the Regioselectivity of Fatty Acid Hydroxylation (C10) at α- and β-Position by CYP152A1 (P450Bsβ) Variants

Article abstract of DOI:10.1002/cctc.201901679

Regioselective hydroxylation on inactivated C?H bonds is among the dream reactions of organic chemists. Cytochrome P450 enzymes (CYPs) perform this reaction in general with high regio- and stereoselectivity (e. g. for steroids as substrates). Furthermore, enzyme engineering may allow to tune the properties of the enzyme. Regioselective hydroxylation of shorter or linear molecules (fatty acids), however, remains challenging even with this enzyme class, due to the high similarity of the substrate's backbone carbons and their conformational flexibility. CYPs hydroxylating fatty acids selectively in the chemically more distinct α- or ω- position are well described. In contrast, selective in-chain hydroxylation of fatty acids lacks precedence. The peroxygenase CYP152A1 (P450Bsβ) is a family member that displays fatty acid hydroxylation at both, the α- and β-position, with preference for the α-position. Herein we report the influence of hydrophobic active site residues on the hydroxylation pattern of this enzyme. By site directed mutagenesis and combination of the libraries, double and triple mutation variants were identified, which hydroxylated decanoic acid (C₁₀) with improved regio-selectivity in the β-position. Variants were identified with a 10-fold increase of the β-regioselectivity (expressed as α/β-ratio) compared to the wild type. In total 103 variants of CYP152A1 (P450Bsβ) were investigated.

Full text of DOI:10.1002/cctc.201901679

Full Papers
DOI: 10.1002/cctc.201901679
1
2
3
4
5
6
7
8
9
Controlling the Regioselectivity of Fatty Acid Hydroxylation
(C10) at α- and β-Position by CYP152A1 (P450Bsβ) Variants
Lucas Hammerer,[a] Michael Friess,[b] Jeyson Cerne,[b] Michael Fuchs,[b] Georg Steinkellner,[a]
Karl Gruber,[a, c] Koenraad Vanhessche,[d] Thomas Plocek,[d] Christoph K. Winkler,[a, b] and
Wolfgang Kroutil*[a, b]
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Regioselective hydroxylation on inactivated CÀ H bonds is
among the dream reactions of organic chemists. Cytochrome
P450 enzymes (CYPs) perform this reaction in general with high
regio- and stereoselectivity (e.g. for steroids as substrates).
Furthermore, enzyme engineering may allow to tune the
properties of the enzyme. Regioselective hydroxylation of
shorter or linear molecules (fatty acids), however, remains
challenging even with this enzyme class, due to the high
similarity of the substrate’s backbone carbons and their
conformational flexibility. CYPs hydroxylating fatty acids selec-
tively in the chemically more distinct α- or ω- position are well
described. In contrast, selective in-chain hydroxylation of fatty
acids lacks precedence. The peroxygenase CYP152A1 (P450Bsβ)
is a family member that displays fatty acid hydroxylation at
both, the α- and β-position, with preference for the α-position.
Herein we report the influence of hydrophobic active site
residues on the hydroxylation pattern of this enzyme. By site
directed mutagenesis and combination of the libraries, double
and triple mutation variants were identified, which hydroxy-
lated decanoic acid (C10) with improved regio-selectivity in the
β-position. Variants were identified with a 10-fold increase of
the β-regioselectivity (expressed as α/β-ratio) compared to the
wild type. In total 103 variants of CYP152A1 (P450Bsβ) were
investigated.
Introduction
oxygen dealkylation,[6] sulfoxidation,[7] anti-Markovnikov
oxidation[8] and in case of cytochrome c CÀ Si and CÀ B bond
formation.[9,10]
As the enzymatic hydroxylation reactions catalyzed by CYPs
occur with high regioselectivity, the enzymes may offer an
answer for chemistry’s demand for selective C-H activation
reactions. Importantly, the site of reaction may be altered by
mutagenesis.[11,12] Regioselective non-terminal hydroxylations of
fatty acids however remain challenging as the only selective
systems known attack the chemically distinct α- or ω-
positions.[13]
CYPs rely on molecular oxygen (O2) as well as the transfer of
two electrons from NADPH, to produce the reactive heme
intermediate (compound I).[14] This electron transfer is enabled
by either distinct electron transfer proteins or by a specific
domain attached to the heme domain like in case of self-
sufficient P450s such as BM3.[2] Due to the diversity of the
enzyme family, no universal electron transfer system is known
and the identification of a suitable electron transfer system for
specific CYPs remains challenging.[15] Alternatively, a few CYPs
are able to utilize H2O2 as oxygen and electron-donor such as
P450 peroxygenases from the CYP152 family.[16–21] In this case,
no electron transfer system is required and uncoupling
reactions are avoided to a large extend.[13]
In our search for cytochrome P450s for regioselective in-
chain hydroxylation of fatty acids we turned our attention to
the CYP152 family. These enzymes catalyze fatty acid hydrox-
ylation mainly but not exclusively at the α-position to the
carboxyl terminus. The binding of the fatty acid occurs via a
conserved active site arginine (Arg242 in CYP152A1) function-
ing as anchor for the substrates carboxy-group, which in turn
Cytochrome P450 enzymes (CYPs) are heme proteins, which are
found in many living organisms. CYPs are important detoxifica-
tion enzymes, as they turn apolar compounds more polar via
hydroxylation, which allows subsequent excretion.[1] Besides the
CÀ H oxidation,[2] P450s catalyze a vast variety of reactions such
as decarboxylation,[3] dehalogenation,[4] epoxidation,[5] amine/
[a] L. Hammerer, Dr. G. Steinkellner, Prof. K. Gruber, Dr. C. K. Winkler,
Prof. W. Kroutil
Austrian Centre of Industrial Biotechnology
c/o University of Graz
Heinrichstrasse 28, 8010 Graz (Austria)
E-mail: Wolfgang.Kroutil@uni-graz.at
[b] M. Friess, J. Cerne, Dr. M. Fuchs, Dr. C. K. Winkler, Prof. W. Kroutil
Institute of Chemistry
University of Graz
NAWI Graz, BioTechMed Graz
Heinrichstrasse 28, 8010 Graz (Austria)
[c] Prof. K. Gruber
Institute of Molecular Biosciences
University of Graz
Humboldtstrasse 50, 8010 Graz (Austria)
[d] Dr. K. Vanhessche, Dr. T. Plocek
Aroma Chemical Services International S.A
Route de St-Julien 184
CH-1228 Plan-les-Ouates, (Switzerland)
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cctc.201901679
This publication is part of a joint Special Collection with ChemBioChem on
“Excellence in Biocatalysis Research”. Please follow the link for more articles
in the collection.
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This
is an open access article under the terms of the Creative Commons Attri-
bution Non-Commercial NoDerivs License, which permits use and distribu-
tion in any medium, provided the original work is properly cited, the use is
non-commercial and no modifications or adaptations are made.
ChemCatChem 2019, 11, 1–9
1
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
facilitates the activation of H2O2 in
activation mechanism.[14,18,22]
a substrate assisted
1
2
CYP152B1 (P450Spα from Sphingomonas paucimobilis) pro-
duces exclusively the (S)-2-hydroxy fatty acids (e.g. for myristic
acid C14). The crystal structure of CYP152B1 was solved with
palmitic acid (C16) as substrate (PDB code: 3AWM) and revealed,
that only the α-carbon is accessible by the heme.[22,23] In
contrast, the orthologue CYP152A2 (P450Cla from Clostridium
acetobutyliticum) shows α- and β- hydroxylation on fatty acids
with preference for the α-position. A crystal structure of
CYP152A2 is not available yet.[24] CYP152MP from Methylobacte-
rium populi on the other hand displays hydroxylation on Cα to
Cɛ for the C12 fatty acid (71% preference for Cβ). Unfortunately,
the enzyme shows low activity and is poorly expressed.[25] OleT
(CYP152 L1) originates from Jeotgalicoccus species[3,26] and
performs predominantly oxidative decarboxylation[17] besides α-
and β-hydroxylation or desaturation if branched carboxylic
acids are used as substrates.[27]
CYP152A1 is a P450 peroxygenase found in Bacillus subtilis
(P450Bsβ). The enzyme shows similar reactivities as OleT, but its
α- and β-hydroxylation activity (e.g. with preference for the β-
position on C14) exceeds its activity for oxidative decarboxyla-
tion or desaturation (Scheme 1).[23,27,28]
The crystal structure was solved with C16 as substrate (PDB
code: 1IZO).[30] The mixed hydroxylation-pattern, high activity
and available crystal structure render CYP152A1 a suitable
starting point for mutagenesis studies to head for regioselective
β-hydroxylation, since no available biocatalyst allows exclusive
β-hydroxylation yet. Additionally, we were interested to
elucidate which residues influence the α- versus β-regioselectiv-
ity.
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
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 1. CYP152A1 structure (PDB: 1IZO) with hexadecenoic acid (C16)
bound as substrate (ball model). Arg242 is indicated as stick. The heme is
shown in red.[30]
formally moved one loop further (Asn239) in order to force
the substrate deeper into the active site. Alternatively, a
new arginine anchor was tested on the opposite side of the
native arginine in the active site (position 85) accompanied
by removing Arg242 (Figure 2).
b) Since CYP152MP catalyzes fatty acid-hydroxylation at vari-
ous in-chain positions (α-ɛ), differences in sequence be-
tween CYP152A1 and CYP152MP may help to understand
the altered regio-selectivity.[25] Consequently, based on a
sequence alignment of the two enzymes, deletions possibly
responsible for the altered reactivity were identified and
introduced into CYP152A1.
c) Positions within the active site and the hydrophobic
channel[30] were systematically searched via an alanine- (Ala)
and phenylalanine- (Phe) scan.
Subsequently, site directed mutagenesis of hotspots identi-
fied in strategies a)–c) was performed introducing other apolar
amino acids, such as valine (Val), leucine (Leu), isoleucine (Ile)
and tryptophan (Trp).
d) Combination of beneficial mutations.
In total, more than 100 variants (single, double and triple
mutations) were tested for their regioselectivity in hydroxylation
Results and Discussion
For shifting and understanding the regioselectivity in
CYP152A1, variants were created following different strategies:
a) The active site Arg242 (Figure 1) is binding the fatty acid
substrate via the carboxylate.[18] For shifting regioselectivity,
changing the position of this amino acid in the active site
was considered. Arg242 is located on one of the α-helices
sandwiching the heme (putative distal helix)[31] and was
Scheme 1. Fatty acid (decanoic acid, C10) hydroxylation by CYP152A1 variants using CamAB[17,29] as electron transfer system and FDH/formate for cofactor
recycling.
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
2
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
A possible reason might be that R242 is conserved in the
CYP152 family and is therefore is crucial for both, the substrate
binding and the activation of hydrogen peroxide.[33]
1
2
3
4
5
6
7
Strategy b): Deletion Variants Based on Sequence Alignment
with P450MP
8
9
According to literature CYP152MP shows high preference for
hydroxylation in β-position besides hydroxylation in α-, γ-, δ-
and ɛ-positions at low substrate concentration (0.5 mM of
dodecanoic acid). However, the enzyme expression was
insufficient for preparative applications.[25] Therefore the en-
zyme was used as template to identify mutations correlated to
the selectivity of the more active CYP102A1 (P450Bsβ) based on
a sequence alignment to CYP152MP. The alignment revealed
three single amino acid deletions in MP compared to P450Bsβ
(see Supporting Information). Hypothesizing that the deletions
may lead to a change of the enzyme’s structure, they were
stepwise introduced to CYP102A1 as single deletions, double
deletions and finally as triple deletion (L214-, S273-, I370-,
L214-/S273-, L214-/I370-, S273-/I370-, L214-/S273-/I370-). All
variants were expressed in soluble form, however, only the
S273- variant showed an active P450 signal. This variant
displayed the same regioselectivity for C10 (β/α=0.76) as the
wild type [wt (β/α=0.76)] (Table 1). Nevertheless, the S273-
variant was further evolved in strategy d).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 2. Active site of CYP152A1 with C16 bound as substrate and the amino
acid exchanges according to strategy a) indicated (Arg242Gln, Gln85Arg).
The heme is shown in yellow (Picture prepared with Pymol).
using decanoic acid (C10) as model substrate and the electron
transfer system CamAB with O2.
Finally, the variants showing preference for β-hydroxylation
were purified and tested with H2O2 as oxidant.
Decanoic acid was chosen as substrate for two reasons: (i) in
initial studies it was shown to be transformed in the mono-
oxygenase mode using CamAB as well as in the peroxygenase
mode using H2O2 as reagent. (ii) The potential product
myrmicacin (3-hydroxydecanoic acid) has a potential applica-
tion as herbicide.[32]
Strategy c): Alanine- and Phenylalanine-Scan of Active Site
Amino Acids
Finally to check the influence of all apolar amino acids shaping
the active site and the hydrophobic channel[30] an alanine-scan
was performed. Eleven amino acids (L41, L42, L70, V74, L78,
F79, V170, F173, F289, F292, L293) were identified from the
crystal structure (PDB code: 1IZO) and mutated to alanine
thereby in general increasing the cavity of the active site
(Figure 3).
All alanine variants were expressed in soluble form, showed
the P450 signal upon CO-titration and were active hydroxylat-
ing decanoic acid. The variants displayed varied ratios of α- and
β-hydroxylation (Table 1).
The L70A and F79A variants preferred α- over β-hydroxyla-
tion and showed a lower β/α ratio than the wt (L70A: β/α=
25:75; F79A: β/α=27:73; wt: β/α=44:56; entries 1, 4, 6). The
L70A variant resulted in >99% conversion, while the F79A
variant converted 74% of the substrate (C10).
In contrast, the L78A and F292A variants showed a clear
preference for β-hydroxylation (Table 1, entries 5, 7). L78A gave
the highest β/α hydroxylation ratio (β/α=66:33) of single
alanine variants with good conversion (61%). Variant F292A
reached completion with a β/α-OH ratio of 53:47.
Strategy a): Shifting the Active Site Arginine-Anchor
Arg242 was substituted with different polar amino acids while a
new anchor position (Arg) was introduced one turn upstream
on the same α-helix (variants N239R/R242T, N239R/R242S,
N239R/R242Q, N239R/R242N). Unfortunately, while sufficient
soluble expression of all variants was found, only the inactive
form of the enzyme (P420) was produced. Thus, no hydrox-
ylation activity was obtained.
Conversely, the introduction of the new arginine anchor at
position 85 and replacement the native Arg242 with polar
amino acids (variants Q85R/R242T, Q85R/R242S, Q85R/R242Q,
Q85R/R242N) gave soluble enzyme in the active form (peak at
450 nm). However, when tested with C10 for hydroxylation, no
conversion was obtained.
The L41A and L42A mutations led to high conversions (93%
and 99% respectively) with increased β/α-OH ratios (47:53)
compared to the wt (44:56). All other Ala scanning variants
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
3
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
Table 1. Hydroxylation of decanoic acid by CYP152A1 alanine- and phenylalanine-scanning variants, including tryptophan, valine, leucine and isoleucine
1
2
3
4
5
6
7
8
9
variants.[a]
Entry
Mutation
Conv.[b]
[%]
2a+2b[c]
[%]
α-OH[d]
[%]
β-OH[d]
[%]
1
2
3
4
5
6
7
8
wt
98
93
>99
97
61
74
98
99
95
64
30
>99
2
86
80
57
5
32
98
59
55
61
65
27
41
60
63
67
36
6
54
6
57
58
37
9
56
53
53
75
33
73
47
51
52
37
38
35
50
82
82
83
67
31
57
44
47
47
25
67
27
53
49
48
63
62
65
50
18
18
17
33
69
43
L41A
L42A
L70A
L78A
F79A
F292A
L41F
L42F
L78F
L78V
L78I
L78W
F79V
F79L
F79I
F79W
V170F
S273-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
9
10
11
12
13
14
15
16
17
18
19
13
63
[a] A complete list of all variants prepared as well as detailed conversions and standard deviations can be found in the Supporting Information. [b] Conversion
is based on substrate recovery. Reaction conditions: 10 mM C10, 5% EtOH, CFE (10 μM P450 enzyme), buffer (KPi 100 mM, pH 7.4), 0.05 U CamAB, 0.2 mM
°
NADH, �2 U FDH, 100 mM NH4-formate, 23–26 C, 20 h. [c] GC yield of 2a plus 2b, 2a,b were quantified via calibration using ω-hydroxydecanic acid as
internal standard; the difference to the conversion is due to further side products. [d] detected hydroxy acids normalized to a total of 100%. All reactions
were performed in triplicates.
counterintuitive, as the CYP152 family member CYP152B1
(P450Spα) which displays almost perfect α-selectivity, has a
phenylalanine at this position. This result is emphasized, as the
exactly same variant was already evaluated previously with
H2O2 as oxidant and a similar shift towards Cβ was found.[30] The
transformation using the L41F and L42F went nearly to
completion with a β/α-OH ratio of close to 1 (49:51 and 48:52,
entries 8 and 9). The L78F variant showed an increase in β/α-OH
ratio (63:37), but at the cost of activity (64% conversion).
In contrast to the phenylalanine-variants, mutations to
tryptophan instead of phenylalanine generally resulted in low
conversions (<37%) and were therefore not further inves-
tigated (see Supporting Information).
Figure 3. Active site of CYP152A1 (PDB code:1IZO) with C16 bound as
The results now obtained allowed to estimate the optimal
substrate. The apolar residues selected for mutagenesis (and Arg242) are
size of residues in certain positions to achieve preferentially β-
hydroxylation. The F79W variant, for example, showed a
conversion below 5%, while the F79A variant gave 74%
conversion with high preference to hydroxylate in α-position. In
order to fine-tune this position, F79 was mutated to valine,
isoleucine and leucine. The F79L variant matches the leucine at
the equivalent position in the α-selective CYP152B1 and was
already described to trigger a hydroxylation shift towards the α-
position (with H2O2 as oxygen source).[30] Accordingly, all three
variants showed β/α-OH ratios close to 0.20 (83% of α-OH),
which shows that this position is crucial for a shift to the α-
position. The highest conversion was obtained by F79V (86%),
followed by F79L (80%) and F79I (57%).
indicated in yellow. The heme is shown in red-brown.
showed either only low conversion or lower β/α-OH ratios than
the wt (see Supporting Information).
Since increasing the space in the active site resulted in
variants with a higher as well as variants with a lower β/α-
hydroxylation ratio than the wt, the effect of a reduction of the
size of the active site space was also investigated. Hence, apolar
amino acids were exchanged to phenylalanine (Phe-scanning)
and phenylalanine was exchanged to Trp.
Variants L41F, L42F, L78F, V170F, F79W, F173W, F289W were
expressed in soluble and active form. Introducing bulkier amino
acids generally resulted in higher β/α-OH ratios than the wt.
Variant V170F gave the highest β/α-OH ratio (69:31, entry 18),
however, at the expense of conversion (32%). This seems
Since increasing and decreasing the cavity-space at position
L78 (L78A, L78F) gave moderate conversions (~60%) but high
β/α-OH ratios (67:33 and 63:37 respectively), L78 was
exchanged to other hydrophobic amino acids such as valine,
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
4
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
isoleucine and tryptophan. Interestingly, the regio-selectivity of
these variants followed an almost size dependent trend: L78W
(50:50)<L78F (63:37)~L78I (65:35)~L78 V (62:38)<L78A
(67:33). The highest conversion was obtained with L78I (>
99%), while L78V gave 30% conversion and L78W only 2%
conversion.
In several cases, the variants produced higher amounts of β-
hydroxy acid in absolute values compared to the wt, but also
high amounts of the α-hydroxy acid due to a higher total
concentration of hydroxylated products which resulted finally in
still moderate/low β/α-OH ratio.
The residues L78, F79 and V170 are located very close to Cα
and Cβ of the bound fatty acid in the deposited structure (PDB:
1IZO) and it is therefore reasonable that amino acid exchanges
at these positions alter the stereospecificity. However, we would
not have been able to rationally predict the direction and/or
the magnitude of the shift in the β/α-ratio.
ratios were obtained as already mentioned with L78 A/F292A
(87:13), followed by variant L78 A/V170F (β/α-OH ratio 86:14,
45% conversion), L78I/V170F (β/α-OH ratio 81:19, 75% con-
version) and L41F/V170F (β/α-OH ratio 76:24, 68% conversion).
Next, variants that led to a β/α-OH ratio higher than the wt
and gave conversions >50% were combined to triple variants.
In total 19 triple variants were prepared, and all showed higher
β/α-OH ratios than the wt (see Table 2, entries 14–17 for
selected examples. The complete list can be found in the SI).
Variant L42F/L78 A/V170F gave the highest β/α-OH ratio of this
study (89:11, 50% conv., entry 16), followed by variant L42F/
L78I/V170F (β/α-OH ratio 83:17, 61% conv.). The best β-
hydroxylation ratio at >98% conversion was obtained by
variants L42A/L78I/F292A and L41F/L78I/F292A, with 68:32 and
72:28, respectively (Entries 17 and 14).
Beside the residues L78, F79 and V170 close to Cα and Cβ
and therefore influencing the hydroxylation pattern, three sites
influenced the α/β-ratio at least in combination with other
exchanges (i.e. L41, L42 and F292). These residues are located
at the entrance of the access tunnel leading to the heme
cofactor. Based on the structure in complex with palmitic acid,
the side chains of these residues should not be in direct contact
with the alkyl chain of a bound decanoic acid. Therefore, the
influence of amino acid exchanges at those sites are most likely
indirect, e.g. altering the access to the active site (thereby
possibly influencing the conversion) or inducing long-range
structural changes.
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
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Strategy d): Combination of Beneficial Mutations
In a next step the beneficial mutations were combined to
double variants. Interestingly, all except one of the 48 double
variants prepared showed higher β/α-OH ratios than the wt
(Table 2). The only exception was L78F/V170F which led to a
high amount of α-hydroxylation (β/α-OH ratio of 19:81,
entry 9). The highest β/α-OH ratio (87:13), with a conversion of
53% was achieved by the variant L78A/F292A (Entry 13).
Highest β/α-OH ratios paired with high conversions (>95%)
were achieved by variants L41F/L78I (73:27, entry 4), L42F/L78I
(69:31, entry 7), L42A/L78I (71:29, entry 5) and L78I/F292A
(66:34, entry 10). Moderate conversion (67% and 57%) were
obtained by L41A/V170F and L42A/L78F variants, which showed
a β/α-OH ratio of ~71:29 respectively. The highest β/α-OH
H2O2 as Oxygen-Source
The product distribution of CYP152 family members (α-OH and
β-OH) is strongly affected by the reaction conditions. Especially
Table 2. Hydroxylation of decanoic acid by CYP152A1 double and triple variants.[a]
Entry
Mutation
Conv.[b]
[%]
2a+2b[c]
[%]
α-OH[d]
[%]
β-OH[d]
[%]
1
2
3
4
wt
98
67
68
99
59
29
33
42
51
31
47
42
<2
59
42
9
56
29
24
27
29
29
31
29
81
34
19
14
13
28
17
11
32
44
71
76
73
71
71
69
71
19
66
81
86
87
72
83
89
68
L41A/V170F
L41F/V170F
L41F/L78I
5
6
L42A/L78I
L42A/L78F
>99
57
7
8
L42F/L78I
L42F/L78F
>99
85
9
L78F/V170F
L78I/F292A
L78I/V170F
L78A/V170F
L78A/F292A
L41F/L78I/F292A
L42F/L78I/V170F
L42F/L78A/V170F
L42A/L78I/F292A
14
96
75
45
53
98
61
50
10
11
12
13
14
15
16
17
9
46
30
12
53
>99
[a] A complete list including all prepared variants as well as detailed conversions and standard deviations can be found in the Supporting Information. [b] The
Conversion is based on substrate recovery. Reaction conditions: 10 mM C10, 5% EtOH, CFE (10 μM P450 enzyme), buffer (KPi 100 mM, pH 7.4), 0.05 U CamAB,
°
0.2 mM NADH, �2 U FDH, 100 mM NH4-formate, 23–26 C, 20 h [c] GC yield of 2a plus 2b, 2a,b were quantified via calibration using ω-hydroxydecanic acid
as internal standard; the difference to the conversion is due to further side products. [d] Detected hydroxy acids normalized to a total of 100%. All reactions
were performed in triplicates.
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
5
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
changing from electron transfer system to H2O2 was already
shown to change the distribution of hydroxylated products.[24]
Selected variants were purified and tested with H2O2 as
oxygen source. Omitting the electron transfer system CamAB by
supplying H2O2 altered both, the β/α-OH ratio and the
conversion (Table 3). Besides the wt all variants tested with
H2O2 showed a slight shift of the β/α-OH ratios towards α-
position is without precedence for regioselective hydroxylation
of fatty acids. Employing the alternative oxidant H2O2 the β/α-
OH ratios of all tested variants led to decreased ratio, never-
theless, the best variant with O2/CamAB was also the best
variant with H2O2 giving still 84% β-hydroxylation.
1
2
3
4
5
6
7
8
9
hydroxylation. For the variant L78A/F292A the conversion Experimental Section
increased 13% with H2O2 as oxygen source compared to O2 and
the electron transfer system CamAB. Also, for the variants L78I/
F292A, L42F/L78A/V170F, and L41F/L78I/F292A the use of H2O2
was beneficial for the conversion. For the wt the β/α-OH was
shifted in favor of hydroxylation in β-position (50:50, entry 1).
Hydroxy acid products were isolated from 20 ml reaction
mixtures and NMR spectra were recorded to proof the product
formation. To obtain the α-hydroxy acid the regioselective
CYP152A2 (P450Cla) was used, while for obtaining a 56:44
mixture of α- and β-hydroxy acids CYP152A1 (P450Bsβ wt) was
employed. A higher amount of β-hydroxy acid (62%) was
obtained with the P450Bsβ variant L78I/F292A. Spectra are
shown in the supplementary information.
General
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
GC-FID measurements were performed on an Agilent Technologies
7890A GC system equipped with an Agilent Technologies auto
sampler 7693. GC-MS measurements were performed on an Agilent
Technologies 7890A GC system equipped with an Agilent Tech-
nologies 5975C mass selective detector.
Decanoic acid, pyridine and N,O-bis(trimethylsilyl)trifluoroaceta-
mide (BSTFA) were bought from Sigma-Aldrich. H2O2 35% wt was
bought from Acros Organics, ω-hydroxydodecanoic acid was
bought from Aldrich-Chemistry and NADH from Roth. Formate
dehydrogenase (FDH, �0.4 U, evo-1.1.230.S) was supplied from
EvoCatal.
The Bsβ coding Gene was ordered from GeneArt and mutagenesis
primers were ordered from Eurofins Genomics.
Conclusions
Cloning of CYP152A1 Gene Into pET-28(a)+ Vector and Site
Directed Mutagenesis
In summary, we identified amino acid positions in the active
site and in the tunnel of CYP152A1 contributing to the
regioselectivity of the hydroxylation of decanoic acid. Whereas
mutation strategies on the carboxylate coordinating Arg242 or
based on a sequence alignment with an orthologue were not
successful, a systematic scan of active site residues led to
variants of interest. Thereby F79 was identified as a crucial
position to increase the α-hydroxylation activity, exchanges of
the neighboring L78 to apolar amino acids showed an increase
of the β/α-hydroxylation ratio compared to the wt, thus leading
to more β-hydroxylated product. The insights obtained from
single variants were bundled into several double and triple
variants with even higher β/α-OH ratios. The highest β/α-OH
ratio (89% β-OH) was obtained with the variant L42F/L78A/
V170F at 50% conversion, which represents a significant shift of
the hydroxylation pattern, since the wt led to preferentially the
α-hydroxy product (56%, only 44% β-OH). This selectivity for β-
The CYP152A1 pET-28(a)+ expression plasmid was prepared as
described (internal plasmid number pEG301).[17]
Site directed mutagenesis was performed with the QuikChange
mutagenesis kit. The mutagenic primers are summarized in the
Supporting Information.
Enzyme Expression and Purification
Overnight cultures (ONCs) of the transformants were prepared in
50 ml tubes [10 ml LB Medium+50 μg/mL kanamycin (KAN)]. The
°
tubes were incubated at 37 C and shaken (at 120 rpm) overnight.
ONCs were used to prepare glycerol stocks (500 μL culture+500 μL
°
30% glycerol stock) that were stored in the freezer (at À 20 C or
°
À 80 C). Plasmid mini-preps of the samples were prepared accord-
ing to the protocol of the QIAGEN plasmid Mini-prep kit.
Sequencing was performed by Microsynth.
Table 3. Hydroxylation of decanoic acid by selected CYP152A1 variants employing H2O2 as oxidant.[a]
Entry
Mutation
Conv.[b]
[%]
2a+2b[c]
[%]
α-OH[d]
[%]
β-OH[d]
[%]
1
2
3
4
5
6
7
8
wt
98
58
99
37
44
33
42
14
48
58
58
27
50
30
31
14
29
40
34
15
50
70
69
86
71
60
66
85
L41F/V170F
L42A/L78I
L78A/V170F
L78A/F292A
L78I/F292A
L41F/L78I/F292A
L42F/L78A/V170F
76
>99
>99
55
[a] A complete list including all prepared variants as well as detailed conversions and standard deviations can be found in the Supporting Information. [b] The
Conversion is based on substrate recovery. Reaction conditions: 10 mM C10, 5% EtOH, CFE (10 μM P450 enzyme), buffer (KPi 100 mM, pH 7.4), in total 20 mM
°
H2O2, 23–26 C, 20 h [c] GC yield of 2a plus 2b, 2a,b were quantified via calibration using ω-hydroxydecanic acid as internal standard; the difference to the
conversion is due to further side products. [d] Detected hydroxy acids normalized to a total of 100%. All reactions were performed in triplicates.
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
6
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
LB agar plates, and LB media were used for bacteria growth.
Shaking flasks (volume 250 mL and 1 L) were filled with medium
(100 mL or 330 mL respectively) and autoclaved. After cooling to
room temperature, kanamycin (50 μg/mL) and a trace element
solution[34] were added. The prepared medium was inoculated with
Determination of the CamAB Activity Determination
1
2
3
4
5
6
7
8
9
For activity determination, cells were resuspended in lysis buffer
(10 mL/g pellet). The suspension was ultrasonicated (Branson
Digital Sonifier) on ice [(30% amplitude, 2 sec pulse on, 4 sec pulse
off for 2 min)×2]. Cell debris was removed by centrifugation
°
the ONC (1 mL) and shaken at 37 C and 120 rpm. When the
cultures reached an OD600 between 0.6–0.8, they were cooled to
°
(14000 rpm, for 20 min, at 4 C) and the supernatant (CFE, soluble
fraction) was collected. To determine the activity of the electron
transfer system (CamAB) a photometric Cytochrome C assay was
performed.
°
25 C and 5-aminolevulinic acid was added (final concentration
0.5 mM; from a 1 mM stock). Cells were induced with IPTG (final
concentration: 0.1 mM). The flasks were shaken overnight at 25 C
and 120 rpm. After 16 hours the cells were harvested by centrifuga-
tion (8000 rpm, 20 min, 4 C). The cell pellets were washed with
phosphate buffer [pH 7.4, 100 mM (10 mL buffer per g pellet]. After
washing, the pellets were frozen (À 80 C) and stored until further
°
Reaction buffer (920 μL, 100 mM Kpi, pH 7.5) was mixed with
cytochrome C from bovine heart (18.4 mg/ml stock, 50 μL) and with
CFE of the respective electron transfer system (10 μL) in a cuvette.
The absorbance was measured at 550 nm (blank) on a photometer
(Agilent Technologies Cary 60 UV-VIS; software: CARYWinUV
Kinetics). To start the reduction on cytochrome C, NADH (20 μL)
was added to the mixture and reaction was followed 3 minutes at
550 nm. For the calculation, a molar extinction coefficient of
28 mMÀ 1 cmÀ 1 as used for Cytochrome C.[17]
°
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
°
use.
For purification, cells were resuspended in lysis buffer (10 mL/g). To
disrupt the cells the suspension was ultrasonicated (Branson Digital
Sonifier) on ice [(30% amplitude, 2 sec pulse on, 4 sec pulse off for
2 min)×2]. The cell debris was removed by centrifugation (10
°
000 rpm, for 20 min, at 4 C) and the supernatant (soluble fraction)
was purified via HisTrap (Qiagen) affinity chromatography. The
column (5 mL HisTrap) was washed with lysis buffer (25 mL). Next,
the sterile filtered supernatant fraction (0.45 μm Rotilabo® syringe
filter) was loaded onto the column. Afterwards the column was
washed with lysis buffer (25 mL). Enzymes were eluted with elution
buffer (up to 25 mL) and cleaning buffer was applied (25 mL) to
remove residual bound proteins. The column was washed with
Biotransformations Using CFE and the CamAB System
Reactions were carried out in glass vials (1 mL reaction volume) at
°
°
room temperature (23 C–26 C). Conversions of decanoic acid
(1.72 mg, 10 μmol) were performed employing cell free extracts
(CFE, 10 μM P450) in reaction buffer (100 mM KPi, pH 7.4). CamAB
(0.05 U), ammonium formate (100 mM), and formate dehydrogen-
ase (FDH, �2 U) were added subsequently. Afterwards, substrate
stock (50 μL, 10 mM final concentration) was added (prepared as
200 mM stock in EtOH) and the reaction was started by the addition
of NADH (20 μL from a 10 mM stock in buffer, 200 μM final
concentration). Reactions were stopped in analogy to the reactions
using H2O2 as oxygen donor (see below).
°
water (25 mL) and stored in a 20% EtOH solution at 4 C. Samples
and buffers were applied with a flow rate of 1 mL/min. Eluted
fractions were concentrated by centrifugation in VIVASPIN tubes
°
(20 mL; Membrane: 10000 MWCO PES; at 4 C at 4000 rpm). After
concentration, samples were desalted (PD-10 Desalting Columns,
°
GE Healthcare), eluted in the reaction buffer and stored at 4 C.
Description of buffers: lysis buffer: 300 mM KCl, 100 mM KPi
(pH 7.5), 20 mM imidazole; elution buffer: 300 mM KCl, 100 mM KPi
(pH 7.5), 300 mM imidazole; cleaning buffer: 300 mM KCl, 100 mM
KPi (pH 7.5), 500 mM imidazole.
Biotransformations Using Purified Enzyme and H2O2 as
Oxygen Donor
Reactions were carried out in glass vials (1 mL reaction volume) at
°
°
room temperature (23 C–26 C). Conversions of decanoic acid
(1.72 mg, 10 μmol) were performed employing purified enzyme
(10 μM) in reaction buffer (950 μL, 100 mM KPi, pH 7.4. Substrate
stock (50 μL) was added (prepared as 200 mM stock in EtOH) and
the reaction was started by the constant addition of 0.5 M H2O2
(5 μL/h until a total volume of 40 μl was added) with a kdScientific
pump, equipped with 1 ml Omnifix®-F syringes from BRAUN (ø
4.7 mm) and 100 Sterican® needles (ø 0.80 mm×120 mm). The
reaction was performed overnight and then stopped the next day
by the addition of HCl (100 μL, 5 M). Reaction products were
extracted with EtOAc (2×400 μL) containing internal standard (ω-
hydroxydodecanoic acid; 30 μM). The combined organic layers
were dried over Na2SO4. Samples (100 μL) were derivatized to their
respective trimethylsilyl (TMS) esters and ethers by addition of N,O-
bis(trimethylsilyl)trifluoroacetamide, containing 1% trimeth-
ylchlorosilane (BSTFA) solution mixed with pyridine (1:1, 200 μL
Determination of P450 Concentration Based on CO-Titration
To determine the concentration of active P450 enzyme the high
affinity to CO of the Fe(II) atom bound to the heme was exploited.
The soluble protein sample was diluted with reaction buffer (1:5 or
1:10) and flushed with CO. Afterwards the sample was blanked on
a GENESYS 10 scanning photometer (wavelength-range measured:
400–540 nm). Sodium dithionate (3 mg) was added, and the sample
was measured. The concentration of P450 enzyme was calculated
with a molar extinction coefficient of 91 mMÀ 1 cmÀ 1 [35]
.
Expression of the CamAB System
For the expression of CamAB the pACYCDuet/CamAB vector
construct was used as described (internal plasmid number
pEG304.[29] The final construct was cloned into E. coli C43 for
expression.
°
total) and incubation for 1 h (50 C, 600 rpm) and subsequently
analyzed on GC-FID and GC-MS.
ONCs were prepared in LB medium (5 mL), supplemented with the
respective antibiotics (chloramphenicol CAM, 25 μg/mL). Main
cultures were prepared in TB media (300 mL), inoculated with ONC
GC-FID: Carrier gas H2. Agilent HP-5 capillary column (30 m,
°
0.32 mm, 0.25 μm film); Temperature program: 170 C hold 3 min,
°
°
°
°
ramp 1: 10 C/min – 220 C – hold 1 min, ramp 2: 5 C/min – 280 C
°
(1 mL) and grown (37 C, 120 rpm) until an OD600 of 0.8 was
°
hold 1 min, postrun 300 C. Retention times: decanoic acid –
achieved. Cells were induced with IPTG (0.1 mM) and grown
3.2 min, α-hydroxydecanoic acid – 4.8 min, β-hydroxydecanoic acid
– 5.0 min, ω-hydroxydodecanoic acid (IS) – 8.9 min.
°
overnight (48 h, at 20 C). Afterwards, cells were harvested and
°
stored at À 80 C until further use.
GC-MS: Carrier Gas He. Agilent J&W HP-5 capillary column (30 m,
0,25 mm, 0.25 μm film); Temperature program (Achiral MSD): 100 C
°
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
7
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
°
°
[13] L. Hammerer, C. K. Winkler, W. Kroutil, Catal. Lett. 2018, 148, 787–812.
[14] R. Ramanan, K. D. Dubey, B. Wang, D. Mandal, S. Shaik, J. Am. Chem. Soc.
2016, 138, 6786–6797.
[15] L. Reisky, H. C. Büchsenschütz, J. Engel, T. Song, T. Schweder, J.-H.
Hehemann, U. T. Bornscheuer, Nat. Chem. Biol. 2018, 14, 342–344.
[16] P. C. Cirino, F. H. Arnold, Adv. Synth. Catal. 2002, 344, 932–937.
[17] A. Dennig, M. Kuhn, S. Tassoti, A. Thiessenhusen, S. Gilch, T. Bulter, T.
Haas, M. Hall, K. Faber, Angew. Chem. Int. Ed. 2015, 54, 8819–8822;
Angew. Chem. 2015, 127, 8943–8946.
[18] O. Shoji, Y. Watanabe, JBIC J. Biol. Inorg. Chem. 2014, 19, 529–539.
[19] Y. Wang, D. Lan, R. Durrani, F. Hollmann, Curr. Opin. Chem. Biol. 2017,
37, 1–9.
[20] A. W. Munro, K. J. McLean, J. L. Grant, T. M. Makris, Biochem. Soc. Trans.
2018, 46, 183–196.
[21] S. Gandomkar, A. Dennig, A. Dordic, L. Hammerer, M. Pickl, T. Haas, M.
Hall, K. Faber, Angew. Chem. Int. Ed. 2017, 57, 427–430.
[22] T. Fujishiro, O. Shoji, S. Nagano, H. Sugimoto, Y. Shiro, Y. Watanabe, J.
Biol. Chem. 2011, 286, 29941-29950, S29941/29941-S29941/29947.
[23] I. Matsunaga, A. Ueda, N. Fujiwara, T. Sumimoto, K. Ichihara, Lipids 1999,
34, 841–846.
[24] M. Girhard, S. Schuster, M. Dietrich, P. Durre, V. B. Urlacher, Biochem.
Biophys. Res. Commun. 2007, 362, 114–119.
[25] J. A. Amaya, C. D. Rutland, T. M. Makris, J. Inorg. Biochem. 2016, 158, 11–
hold 0.5 min, ramp 1: 10 C/min – 300 C; 0.442 bar, helium (0.7 mL/
1
2
3
4
5
6
7
8
9
°
min), injection temperature: 250 C, split ratio: 99/1, electron impact
ionization (70 eV), m/z 33–400 was scanned. Retention times:
decanoic acid – 7.9 min, α-hydroxydecanoic acid – 10.1 min, β-
hydroxydecanoic acid – 10.3 min, ω-hydroxydodecanoic acid (IS) –
13.9 min.
Acknowledgements
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
This study was financed by the Austrian FFG, BMWFJ, BMVIT, SFG,
Standortagentur Tirol and ZIT through the Austrian FFG-COMET-
Funding Program.
Conflict of Interest
The authors declare no conflict of interest.
16.
[26] J. L. Grant, C. H. Hsieh, T. M. Makris, J. Am. Chem. Soc. 2015, 137, 4940–
4943.
Keywords: Biocatalysis · Biotransformation · CÀ H · Cytochrome
P450 · Enzyme Engineering
[27] M. Pickl, S. Kurakin, F. G. Cantú Reinhard, P. Schmid, A. Pöcheim, C. K.
Winkler, W. Kroutil, S. P. de Visser, K. Faber, ACS Catal. 2019, 9, 565–577.
[28] O. Shoji, T. Fujishiro, H. Nakajima, M. Kim, S. Nagano, Y. Shiro, Y.
Watanabe, Angew. Chem. Int. Ed. 2007, 46, 3656–3659; Angew. Chem.
2007, 119, 3730–3733.
[29] A. Schallmey, G. den Besten, I. G. P. Teune, R. F. Kembaren, D. B. Janssen,
Appl. Microbial. Biotechnol. 2011, 89, 1475–1485.
[30] D.-S. Lee, A. Yamada, H. Sugimoto, I. Matsunaga, H. Ogura, K. Ichihara,
S.-I. Adachi, S.-Y. Park, Y. Shiro, J. Biol. Chem. 2003, 278, 9761–9767.
[31] I. Matsunaga, A. Ueda, T. Sumimoto, K. Ichihara, M. Ayata, H. Ogura,
Arch. Biochem. Biophys. 2001, 394, 45–53.
[32] H. Schildknecht, K. Koob, Angew. Chem. Int. Ed. 1971, 2, 124–125.
[33] I. Matsunaga, A. Ueda, T. Sumimoto, K. Ichihara, M. Ayata, H. Ogura,
Arch. Biochem. Biophys., 2001, 1, 45–53.
[1] H. Raunio, M. Kuusisto, R. O. Juvonen, O. T. Pentikäinen, Front.
Pharmacol. 2015, 6, 123–123.
[2] C. J. C. Whitehouse, S. G. Bell, L.-L. Wong, Chem. Soc. Rev. 2012, 41,
1218–1260.
[3] M. A. Rude, T. S. Baron, S. Brubaker, M. Alibhai, S. B. Del Cardayre, A.
Schirmer, Appl. Environ. Microbiol. 2011, 77, 1718–1727.
[4] H. Ahr, L. King, W. Nastainczyk, V. Ullrich, Biochem. Pharmacol. 1982, 31,
383–390.
[5] E. H. Oliw, J. Bylund, C. Herman, Lipids 1996, 31, 1003–1021.
[6] A. D. Rahimtula, P. J. O’Brien, Biochem. Biophys. Res. Commun. 1975, 62,
268–275.
[7] T. J. Volz, D. A. Rock, J. P. Jones, J. Am. Chem. Soc. 2002, 124, 9724–9725.
[8] S. C. Hammer, G. Kubik, E. Watkins, S. Huang, H. Minges, F. H. Arnold,
Science 2017, 358, 215–218.
[9] S. B. J. Kan, R. D. Lewis, K. Chen, F. H. Arnold, Science 2016, 354, 1048–
1051.
[34] J. Nazor, S. Dannenmann, R. O. Adjei, Y. B. Fordjour, I. T. Ghampson, M.
Blanusa, D. Roccatano, U. Schwaneberg, Protein Eng. Des. Sel. 2008, 21,
29–35.
[35] T. Omura, R. Sato, J. Biol. Chem., 1964, 7, 2379–2385.
[10] S. J. Kan, X. Huang, Y. Gumulya, K. Chen, F. H. Arnold, Nature 2017, 552,
132.
[11] S. Kille, F. E. Zilly, J. P. Acevedo, M. T. Reetz, Nat. Chem. 2011, 3, 738.
[12] F. H. Arnold, Angew. Chem. Int. Ed. 2018, 57, 4143–4148; Angew. Chem.
2018, 130, 4212–4218.
Manuscript received: September 4, 2019
Version of record online: ■■■, ■■■■
ChemCatChem 2019, 11, 1–9
www.chemcatchem.org
8
© 2019 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA

Full Papers
FULL PAPERS
1
2
3
4
5
6
7
8
L. Hammerer, M. Friess, J. Cerne,
Dr. M. Fuchs, Dr. G. Steinkellner,
Prof. K. Gruber, Dr. K. Vanhessche,
Dr. T. Plocek, Dr. C. K. Winkler,
Prof. W. Kroutil*
9
1 – 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Controlling the Regioselectivity of
Fatty Acid Hydroxylation (C10) at
α- and β-Position by CYP152A1
(P450Bsβ) Variants
Taking control of place. The regiose-
lectivity of the peroxygenase P450Bsβ
was shifted to increase the hydroxyla-
tion in either the α- or β-position.