Enzymatic Late-Stage Oxidation of Lead Compounds with Solubilizing Biomimetic Docking/Protecting groups

Article abstract of DOI:10.1002/chem.201802331

Late-stage functionalization of lead compounds is of high interest in drug discovery since it offers an easy access to metabolites and derivatives of a lead compound without the need to redesign an often long multistep synthesis. Owing to their high degre

Full text of DOI:10.1002/chem.201802331

DOI: 10.1002/chem.201802331
Full Paper
&
Biocatalysis
Enzymatic Late-Stage Oxidation of Lead Compounds with
Solubilizing Biomimetic Docking/Protecting groups
Clare Vickers,[a] Gisela Backfisch,[b] Frank Oellien,[a] Isabel Piel,[a] and Udo E. W. Lange*[a]
Chem. Eur. J. 2018, 24, 1 – 13
1
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Abstract: Late-stage functionalization of lead compounds is
of high interest in drug discovery since it offers an easy
access to metabolites and derivatives of a lead compound
without the need to redesign an often long multistep syn-
thesis. Owing to their high degree of chemoselectivity, bio-
catalytic transformations, enzymatic oxidations in particular,
are potentially very powerful because they could allow the
synthesis of less lipophilic derivatives of a lead compound.
In the majority of cases, enzymatic oxidations have been
used in an empirical way as their regioselectivity is difficult
to predict. In this publication, the concept of using docking/
protecting groups in a biomimetic fashion was investigated,
which could help steer the regioselectivity of a P450BM3
-
mediated oxidation. A novel set of docking/protecting
groups was designed that can be cleaved under very mild
conditions and address the often problematic aqueous solu-
bility of the substrates. Vabicaserin was used as tool com-
pound containing typical groups such as basic, aliphatic,
and aromatic moieties. The results were rationalized with
the help of in silico docking and molecular dynamic studies.
Introduction
Late-stage oxidation (LSO) of complex molecular structures is
increasingly used in lead optimization of drug candidates,
where the aim is to improve the physicochemical properties of
a promising lead molecule, to introduce a synthetic handle for
further structural diversification, or to produce a specific me-
tabolite for toxicity or biological studies.[1] Whilst there are sev-
eral chemical methods which can be employed for LSO,[2] bio-
transformations using either whole cells or isolated enzymes
can offer a complementary method for LSO, often yielding al-
ternative oxidation products to those produced by chemical
means.[3] Additionally the ability to perform bio-LSO (LSO using
whole cells or isolated enzymes) under mild pHs at room tem-
perature, allows chemo-selective CÀH bond oxidation on
highly functionalized scaffolds.[4]
Chemists utilizing biotransformations for LSO of drug candi-
dates may come across various obstacles such as limited sub-
strate scope, poor substrate aqueous solubility, or unwanted
metabolism of reactive functionality (e.g., amines, thiols, or CÀ
H bonds alpha to heteroatoms).[5] As many biologically active
molecules contain amine functionalities, the reactivity of
amines can be particularly problematic when considering
these structures for bio-LSO modification as not only unstable
metabolites are formed, but amines and their metabolites can
inhibit P450 enzymes (see Figure 1).[6]
Figure 1. Possible oxidation of 18 and 28 amines by P450s and inhibitory
pathways of nucleophilic amines and their metabolites. Other inhibitory
complexes may also be possible with the metabolic intermediates preceding
the nitroso-complex (IV) but are not shown here.[7]
Whilst protein engineering can be used to address the
issues of substrate scope,[8] solubility[9] and regioselectivity,[8d,10]
this strategy is not often conveniently available in most organ-
ic chemistry laboratories. Commercially available kits of engi-
neered P450 mutants could address these issues for certain
substrates, however these represent a black box concerning
substrate scope as vendors do not provide information as to
the precise modifications made to the P450s, additionally a
lack of diversity within such panels can limit their generality
and wider application.[11] In this respect, more accessible sub-
strate engineering strategies have been reported which facili-
tate the reaction of unnatural substrates with various P450s,
and generally involve covalently attaching a chemical auxiliary
which can act as a docking group.[11b,12] If the binding interac-
tions between the enzyme and natural substrate are known
the incorporation of a specific recognition motif within the
docking group can mimic these interactions and thus lead to
improved enzymatic recognition and improved reactivity.[13]
Additionally, the docking group can function as a protecting
group for reactive functionality, the dual role giving rise to the
term “docking/protecting groups” or “d/p groups”,[14] which
[a] C. Vickers, F. Oellien, I. Piel, Dr. U. E. W. Lange
Neuroscience Discovery, Medicinal Chemistry
AbbVie (Deutschland) GmbH & Co. KG
Knollstrasse, D-67061 Ludwigshafen (Germany)
E-mail: udo.lange@abbvie.com
[b] G. Backfisch
Development Sciences, DMPK and Bioanalytical Research
AbbVie (Deutschland) GmbH & Co. KG
Knollstrasse, D-67061 Ludwigshafen (Germany)
Supporting information and the ORCID identification number(s) for the au-
thor(s) of this article can be found under:
https://doi.org/10.1002/chem.201802331.
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
2
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
serve to shift the chemo-/regioselectivity of the oxidation away
from reactive functionality by modifying one or more of the re-
gioselectivity determining factors such as CÀH bond reactivity,
accessibility, and proximity to the heme group[15] (Figure 2).
(see Figure 3).[11a,17,24] Thus, incorporation of a carboxylate
moiety into the d/p group which mimics that of the natural
substrate could therefore facilitate the docking of the non-nat-
ural substrate, potentially changing the site of oxidation with
Figure 2. Schematic of the docking/protecting group (d/p group) strategy
showing the three steps: (i) protection of the basic amine, (ii) biooxidation,
and (iii) deprotection yielding the final hydroxylated product.
Figure 3. Crystal structure P450BM3 complex with N-palmitoylglycine (4KPA),
showing interactions between the ligand’s carboxylate and the active site’s
Y51 and R47 (the carboxylate recognition motif).[16] The original substrate,
palmitic acid, was modified via amide formation with glycine which in-
creased both the conversion and aqueous solubility.[12c,17]
Additionally, in the case of nucleophilic amines, an electron
withdrawing protecting group would eliminate the propensity
of this functionality towards inhibition of P450 enzymes via
the pathways shown in Figure 1.[18,19]
respect to that observed with the original non-protected sub-
strate. Indeed, it has been reported that the modification of
palmitic acid via amide formation with glycine not only en-
hanced the aqueous solubility but also increased conversion
with P450BM3 by acting as a docking groups (Figure 3).[25,26] The
reverse of this d/p concept, using the P450PikC from Streptomy-
ces venezuelae, which has an amine rather than a carboxylate
recognition motif as in P450BM3, was used in conjunction with
various amine containing d/p groups for the biooxidation of
menthol and was successful in producing different metabolites
depending on the nature of the of the d/p group.[13] The ad-
vantage of developing a d/p group which can be used with
P450 systems containing a carboxylate recognition motif is
that it can be used in conjunction with the commercially avail-
able P450 kits based on P450BM3 and will provide a convenient
methodology which can be exploited in synthetic chemistry
laboratories.
Classical amine protecting groups such as Bn (benzyl), Bz
(benzoyl), BOC (tert-butoxycarbonyl), Cbz (carboxybenzyl) or
trifluoroacetyl groups[12b,20] have been reported to have also in-
creased reactivity of the protected substrates. This observation
could result from either increased hydrophobic interactions be-
tween the hydrophobic d/p group and the active site’s hydro-
phobic residues, or via polar interactions of the amide/carba-
mate moiety.[21] However, these protecting groups are not ideal
d/p groups for bio-LSO as not only do they decrease the aque-
ous solubility of the substrates but the conditions required for
deprotection may not be compatible with the oxidized prod-
ucts, for example, dehydration of hydroxylated products or hy-
drogenation of alkene products. Therefore, a d/p group which
can be modified to incorporate a functionality which enhances
aqueous solubility and also offers the opportunity to control
the regioselectivity of oxidation via specific and predictable in-
teractions with polar active site residues could be superior in
comparison to previously reported d/p groups. Considering
the latter aspect, the choice of biocatalyst used for the oxida-
tion will determine which functional groups should be incor-
porated for regioselectivity control.
As mentioned above the conditions required for deprotec-
tion must be mild enough to avoid dehydration of the hy-
droxylated product, therefore strongly acidic conditions should
be avoided. Considering the two issues of ease of deprotection
and versatility with respect to carboxylate incorporation, nitro-
phenylsulfonamides (Nosyl) groups are of interest as they can
be removed under mildly basic conditions with a thiolate via a
nucleophilic aromatic substitution (see Figure 4),[27] and also
offers four different sites for introduction of a carboxylate
moiety.
Of the known P450s which are suitable for use as biocata-
lysts, that is, are easily produced and stable enough to be
used synthetically, the well-studied C12–C16 fatty acid metab-
[22]
olizing P450 from Bacillus megaterium, CYP102A1, P450BM3
,
offers an attractive starting point for development of this strat-
egy as the natural substrate recognition motif is known. A ty-
rosine Y51,[23] and arginine, R47,[16] have been reported to facili-
tate binding of carboxylate containing substrates in P450BM3
Herein, we report the development of a series of novel car-
boxylated 2-Ns d/p groups (see Figure 5), based on the d/p
concept described above, which have the potential to enable
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
3
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
substrate-recognizing proteins, a fairly large lipophilic and
highly flexible binding site which exhibits ligand-induced con-
formational changes during substrate binding.[25] Binding
modes retrieved from docking experiments therefore heavily
vary and rely on the underlying X-ray protein conformation. In
recent years molecular dynamic simulations (MDS) have been
shown to provide a much deeper insight in the CYP structure-
function compared to standard docking approaches by taking
into account the conformational flexibility of CYPs as well as
the role of water. For example, while the metabolic regio- and
stereoselectivity of the P450BM3 substrate in Figure 3 cannot be
explained by the binding mode of the X-Ray structure alone,
the application of MDS on the crystal structure successfully
identified the relevant binding mode and predicted the correct
regioselective SOMs.[31] In addition, approaches combining
docking with MDS showed improved prediction capabilities
compared to simple static docking experiments.[29,32]
Figure 4. Nosyl deprotection conditions allowing facile isolation of the de-
protected amine from the mercaptan by-product: (i) solid supported thio-
late; (ii) fluorous thiolate; (iii) mercaptoacetic acid and base.
The experimentally observed SOMs using P450BM3-F87V mu-
tants, biomimetic d/p groups (A–E) and Vabicaserin (1) as
model substrate were structurally rationalized with the help of
a combined docking/MDS approach. The feasibility to guide
the design of d/p ligands applying this concept was also evalu-
ated.
Results and Discussion
Modified Nosyl groups: protection/deprotection conditions
An initial feasibility study regarding the ability of the modified
2-Ns protecting groups (d/p=A—E in Figure 5) to act as pro-
tecting groups, that is, facile conditions for protection and de-
protection, was carried out (see Figure 6 and 7). In most cases
Figure 5. Docking/protecting (d/p) groups used in this study in combination
with Vabicaserin (1) as the model substrate.
the synthesis of different hydroxylated compounds from a sub-
strate using the same or a set of P450 enzymes possessing a
carboxylate recognition motif. For a feasibility study Vabicaser-
in (1) was chosen as model substrate as the structure offers a
variety of CÀH bonds of differing reactivity for potential oxida-
tion (e.g., CÀH bonds alpha to amines in addition to aromatic,
benzylic, and non-activated aliphatic CÀH bonds).[28]
Figure 6. Nosylamide preparation. Reaction conditions: d/p=A–C and E
(i) Amine (1 equiv), CO2H-Ns-Cl (2 equiv), DIPEA (5 equiv), in 2% TPGS-750m,
10 min at room temperature;[33] d/p=D (ii) Amine (1 equiv), CO2Me-Ns-Cl
(2 equiv), DIPEA (5 equiv), in CH2Cl2, followed by ester hydrolysis with NaOH
(1m) in MeOH/MeCN.
A broad range of computational methods have successfully
been used to rationalize and predict substrate binding and
metabolism for various CYP (cytochrome P450) proteins in the
last decade.[29,30] Today, state-of-the-art computer-aided ap-
proaches are capable of predicting sites of metabolism (SOMs)
among the three highest-ranked atom positions in 70–90% of
all cases. However, computational structure-based approaches
like docking are facing several hurdles when applied to P450
proteins. The P450BM3 protein is designed to accommodate a
wide variety of substrates and has in contrast to many other
Figure 7. Nosyl group cleavage with N-acetylcysteine.
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
4
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
protection of the amine with the carboxylated Ns-Cl reagent
was directly achieved in one step. Only d/p group D required
the use of the corresponding ester and a two-step protocol.
The use of polymer bound thiols provides an attractive de-
protection method for small scale reactions owing to the
simple work-up consisting of filtration followed by evaporation
of solvent.
Modified Nosyl (2-Ns) groups: effect on bio-oxidation
conversion, number of metabolites formed and region of
oxidation
The series of carboxylated 2-Ns groups (d/p=A–E, in Figure 5)
were investigated for their effect on the biocatalytic oxidation
of Vabicaserin with P450BM3 and the P450BM3-F87V mutant, a mu-
tation which has been shown to broaden the substrate scope
to include bulkier drug-like substrates.[8f,11a,38]
Deprotection of protected Vabicaserin analogues (2–6) with
N-acetylcysteine under mildly basic conditions (Cs2CO3) in a
polar aprotic solvent (DMSO) at room temperature gave quan-
titative conversion to the free amine, the product being isolat-
ed in good to moderate yields (see Figure 7).[34,35] However, de-
protection of compound (2) required elevated temperatures
and resulted in a complex mixture of products.
The conversion of Vabicaserin as the free amine was very
low (ꢀ1%) with both the wild type and mutant enzyme, how-
ever once the amine was protected conversion increased with
both enzymes (see Figure 8). As expected the F87V mutant
gave higher conversion as a result of increased access to the
heme which was confirmed by the molecular dynamics simula-
tions that indicated in the P450BM3 wild type F87 had to be ro-
tated to open the heme site for the substrate.[39] The observed
general increase in conversion of the all protected analogues
tested compared to that of the free amine results from the
amine being rendered non-nucleophilic by the electron with-
drawing nature of the d/p groups and thus being unable to
participate in the inhibitory pathways outlined in Figure 1. Va-
bicaserin (1) caused a type-II shift in the UV spectrophotomet-
ric assay for the heme absorbance band indicating the forma-
tion of an inhibitory complex with the P450BM3-F87V mutant.[40]
Additionally, in general all the d/p groups (A–E) increase the
potential for formation enzyme-substrate complex stabilizing
interactions resulting in increased binding affinity which could
contribute to the increase in conversion. Aqueous solubility ap-
pears to be a prerequisite for conversion, with the poorly solu-
ble compounds (7, 8 and 9) being poorly converted. Interest-
ingly, whilst all five of the modified 2-Ns groups (d/p=A–E) in-
creased solubility of the protected compounds (2–6) compared
to Boc (7), Cbz (8) and 2-Ns (9), a significant increase in conver-
sion (>90-fold increase in conversion with the F87V mutant
compared with the conversion of the free amine) was only ob-
served in three out of the five protected Vabicaserin analogues
(2, 3 and 6), with the remaining two compounds (4 and 5)
having conversions comparable with that of the analogues
protected with the common protecting groups [Boc (7), Cbz
(8) and 2-Ns (9)]. This observation suggests that the modified
2-Ns groups (A–E) can only function as a d/p group when the
carboxylate recognition motif is positioned in the enzyme’s
active site in such a way that a CÀH bond of the protected
substrate is held close enough to the heme (c.f. MDS section
below).
Modified Nosyl (2-Ns) groups: effect on solubility
Aqueous substrate solubility is important in biocatalysis as
poor solubility can result in irreproducible screening results
and the need to develop special formulation techniques for
preparative reactions.[36]
The aqueous solubility of protected Vabicaserin analogues
(2–6 shown in Figure 5) under the conditions required for the
bio-oxidation reaction was compared with Vabicaserin protect-
ed with common electron withdrawing protecting groups
(Boc, Cbz and 2-Ns) and unprotected Vabicaserin (1) (see
Figure 8).
Figure 8. The effect of the protecting groups on conversion of Vabicaserin
(1) with P450BM3-WT and P450BM3-F87V. Conversion is based on substrate deple-
tion as measured in by LCMS using an internal standard. Measurements
were performed in triplicate.[37]
The positioning of the d/p group carboxylate moiety also
had an impact on the oxidation pattern, with the different pro-
tecting groups each having a different regioselectivity and
thus led to a different mixture of aliphatic, aromatic and other
oxidation products (illustrated in Figure 9).
The free amine was highly soluble (>600 mm), however the
solubility of the Boc (7), Cbz (8) and 2-Ns (9) protected Vabica-
serin derivatives was very low (<5 mm). The use of the car-
boxylated 2-Ns groups (d/p=A–E) led to a much increased
aqueous solubility with levels approaching that of the free
amine (>500 mm) corresponding to over a 100-fold increase in
solubility for all five derivatives.
The site/region of oxidation was determined by LC-MS/MS
analysis of the crude reaction mixtures. Monohydroxylation
(M+16) was favored over other possible oxidation products.
Interestingly, in the reactions of analogues 2 or 3 and 6, the fa-
vored region of oxidation was inverted, with oxidation at an
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
5
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 11. Preparative reaction with the F87V mutant and protected Vabica-
serin analogue 6. Four major M+16 products (10–13) were isolated from
the mixture after removal of the protecting group. Reaction conditions:
(i) Substrate (0.25 mm), P450BM3-F87V mutant (10 mgmLÀ1), Potassium phos-
phate buffer (pH 8, 100 mm), MeCN (1.25%), NADP+(1 mm), Glucose
(25 mm), Glucose Dehydrogenase (0.5 mgmLÀ1), at r.t. (ii) N-acetyl cysteine,
Cs2CO3, DMSO, r.t., yielding compounds 10, 11, 12 and 13 in 12%, 7%, 10%
and 8% yield, respectively over two steps.
products could be isolated and characterized to show that in
addition to the phenol product (10), the aliphatically hydroxy-
lated compounds 11, 12 and 13 were produced (Figure 11).
Figure 9. Composition of P450BM3-F87V oxidation reactions with protected Va-
bicaserin analogues (2–6). The region of oxidation (%AUC as determined by
LC-MS/MS) is dependent on the positioning of the protecting group carbox-
ylate moiety. The numbers in brackets refer to the number of metabolites
found within each regional classification (stereoisomers could not be sepa-
rated).
Computer-aided rationalization of d/p group-related regio-
selectivity
To rationalize the observed oxidation patterns of two of the
substrates, docking and molecular dynamic simulation studies
were initiated. In a first step, protected Vabicaserin analogues
3 and 6 were docked including two different protonation
states into two corresponding F87V mutants of the crystal
structure representations of P450BM3 (1FAG: X-ray of cyto-
chrome P450BM3 with fatty acid substrate palmitoleic acid,
4KPA: X-ray of cytochrome P450BM3 in complex with N-palmi-
toylglycine). In all cases the Vabicaserin moiety was orientated
towards the heme group in a reactive distance between 4.4
and 5.8 ꢁ from the iron atom while the carboxylated protec-
tion group was located next to the entrance of the substrate
binding site mimicking the carboxylate group of the endoge-
nous substrates. Compared to unprotected Vabicaserin (1) ad-
ditional hydrogen bonds and van der Waals interactions be-
tween the d/p group and the entrance of the binding pocket
increase the substrate affinity of 3 and 6 to the enzyme which
is also in line with the significantly higher conversion of 3 and
6.
aromatic CÀH bond being favored for 2 and 3, and aliphatic C-
H oxidation dominating the reaction of 6.
In order to investigate the factors affecting the observed re-
gioselectivity further, and determine the finite site of oxidation
by NMR, the reactions of protected compounds 3 and 6 with
P450BM3-F87V mutant were scaled up (see Figure 10 and 11). The
reaction of compound 3 was very selective for aromatic oxida-
tion at the most activated position in the aromatic ring of Vabi-
caserin (10 in Figure 10), whereas compound 6 gave a mixture
of four M+16 oxidation products of which the four major
All top poses of analogue 3 in the F87V mutant of 1FAG and
4KPA, respectively, revealed the same binding mode, orientat-
ing the aromatic ring of Vabicaserin in an almost orthogonal
position to the heme group positioning the terminal (para) CÀ
H bond to the iron ion, respectively.[41] In the 4KPAF87V mutant
the substrate was bound deeper into the active site leading to
a slightly different orientation of the protection group and to a
shorter distance between the terminal aromatic CÀH bond and
the iron of 4.2 ꢁ (Figure 12, green pose) compared to the C-Fe
distance of 5.7 ꢁ in 1FAGF87V mutant.[39] These results demon-
Figure 10. Preparative reaction with the F87V mutant and protected Vabica-
serin analogue 3. The product was purified after the protecting group was
removed. Reaction conditions: (i) Substrate (0.25 mm), P450BM3-F87V mutant
(10 mgmLÀ1), Potassium phosphate buffer (pH 8, 100 mm), MeCN (1.25%),
NADP+(1 mm), Glucose (25 mm), Glucose Dehydrogenase (0.5 mgmLÀ1), at
r.t. (ii) N-acetyl cysteine, Cs2CO3, DMSO, r.t., 20.8% yield over two steps.
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
6
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 14. Comparison of best docking pose of compound 6 (green) and its
Figure 12. Comparison of best docking pose of compound 3 (green) and
the refined MDS result (purple) on 4KPA P450BM3-F87V
corresponding MDS retrieved binding mode (purple) on 4KPA P450BM3-F87V
.
.
strate that automated docking was not only sufficient to pre-
dict the experimentally favored region of oxidation (Figure 9
and 10), but also identified the regioselective site of metaboli-
zation for this substrate.
Vabicaserin was positioned towards the prosthetic group or
the Vabicaserin ring system was oriented in parallel to the
heme.[42] Interestingly, these “fuzzy” in silico observations were
in alignment with the experimental results which showed a
much larger number (Figure 11) and less biased ratio (Figures 9
and 10) of the resulting oxidation products. However, the cor-
rect identification of the experimentally found SOMs
(Figure 11) could not be obtained from these binding modes.
Subsequent MDS was performed to analyze, if the right SOMs
and products could be identified after the simulation.[26]
During the subsequent 10 ns Molecluar Dynamics Simulation
(MDS) a significant improvement of the initial docking pose
was observed (Figure 12, magenta pose). While the Vabicaserin
part was only translated slightly (RMSD 1.26 ꢁ), without chang-
ing the orientation of the correctly identified aromatic CÀH
bond towards the heme iron, the formerly non-interacting pro-
tection group was re-orientated completely establishing a set
of new hydrogen bond interactions with the backbone amines
of S72, A74, L75, Y51 (water mediated) and A330. The distance
of the aromatic C-H atom only increased very slightly from
4.2 ꢁ to 4.6 ꢁ without weakening the regioselective prediction
power.
In contrast to compound 3, most of the docking poses of
analogue 6 showed significantly rearrangements during the
MDS. The substrate often moved deeper into the binding
pocket leading to decreased distances between the experi-
mentally found SOM and the heme. This is exemplarily de-
scribed below for the top pose[26] of analogue 6 on 1FAGF87V
and 4KPAF87V. The best docking pose of compound 6 on
1FAGF87V (Figure 13, green pose) exhibited a very similar bind-
ing mode like analogue 3. In addition the docking pose re-
vealed an interaction with Y51 because of the slightly longer
carboxylate chain showing that the d/p protection group is
mimicking the carboxylate of the endogenous substrate.[43] The
aromatic ring was positioned towards the heme (5.2 ꢁ be-
tween aromatic (para) CÀH bond and iron) which was in agree-
ment with the experimental finding that one out of four ob-
served oxidations took place on the aromatic ring system
(Figure 11, compound 10). Despite the similar docking pose
compound 3 only showed small rearrangements during the
simulation (Figure 12) ), MDS of analogue 6 on 1FAGF87X re-
vealed a significant rearrangement (Figure 13, magenta pose).
The Vabicaserin core rotated by 180 degrees positioning the
cyclopentane ring towards the prosthetic group and moving
the aromatic ring away from the heme. The final binding pose
(RMSD 4.39 ꢁ) revealed numerous direct and water mediated
hydrogen bonds between the protection group and the pro-
tein (Y51, R47, S72, A74, A330, S332). Even more interestingly it
placed the correct ring-bridging CÀH bond, which was the
major SOM in the P450BM3-F87V-mediated oxidation yielding
compound 11, to the heme iron with the shortest distance of
4.5 ꢁ. Surprisingly the MDS on the second-best pose, which
almost shared the same binding mode but only differed by
Automated docking of Vabicaserin analogue 6 on the 1FAG
(Figure 13) and 4KPA F87V structures (Figure 14) revealed a sig-
nificantly different docking behavior compared to the docking
results of analogue 3. While all top poses of analogue 3
showed a conserved binding mode, docking experiments with
analogue 6 resulted in many different binding modes covering
poses similar to analogue 3 (aromatic ring oriented towards
the heme) but also binding modes where the aliphatic part of
Figure 13. Best automatic docking solution of compound 6 on 1FAG
P450BM3-F87V (green) compared to the refined binding mode (purple) after a
10 ns MDS.
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
7
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
the protonation pattern of the Vabicaserin nitrogen (different
sides),[43] resulted in a completely different simulation trajecto-
ry. As observed for compound 3, the orientation of the aromat-
ic moiety of the Vabicaserin did not change during the simula-
tion and therefore supported the oxidation of the terminal aro-
matic (para) C-H and explaining the presence of compound
10.[44] This indicates that even small modifications can have a
large impact on the binding mode and SOM.
The best docking pose of compound 6 on 4KPAF87V showed
a completely different binding pose compared to those found
[45]
for compound 3 and compound 6 on 1FAGF87V
.
The aliphatic
cyclopentane ring was positioned towards the prosthetic
group revealing a short distance of 3.8 ꢁ between the terminal
CÀH bond and the iron (Figure 14, green pose). Also this ob-
servation was supported by the experimental findings showing
that the aliphatic ring system was the primary oxidation region
of compound 6 (Figures 9 and 11). The succeeding MDS led to
an improved binding mode showing many additional interac-
tions (R47, Y51, A74, S72, A330 and L437). More importantly
the Vabicaserin core rotated by 90 degrees (Figure 14, magen-
ta pose). While the core was aligned orthogonal to the heme
in the starting pose, the cyclopentane ring was now in parallel
to the prosthetic group facing the aliphatic CÀH bonds to-
wards the iron heme. Despite the correct rotation of the ring,
the methylene group that was oxidized leading to compound
13 was not the closest to the iron (distance to Fe 4.9 ꢁ). In-
stead the terminal methylene group of the cyclopentane ring
revealed the shortest distance to the heme (4.0 ꢁ).
In summary, the combined docking/MDS approach proved
to be more reliable to rationalize the biomimetic d/p-group re-
lated regioselectivity behavior compared to static docking sol-
utions alone. Especially for compound 6 substantial reorienta-
tions of the Vabicaserin moiety could be found allowing better
explanations of the experimental oxidation pattern. Neverthe-
less, it also exhibits a few limitations regarding its prospective
application in d/p-group and P450BM3 design.[46] The large size
of d/p-protected Vabicaserin analogues limited the number of
possible docking poses and also restricted the rearrangement
of the Vabicaserin core during the MDS. In addition, it could be
observed that very small structural differences between crystal
structures of P450 (1FAG and 4KPA) as well as on the ligand
(e.g., ionization pattern) can influence the binding pose and
MDS dramatically. Our MDS also did not consider the Fe-
bound catalytic oxygen which also can influence the prediction
power of our approach.
Figure 15. P450BM3 mutant library screening results: Bar chart depicts the re-
action composition of the selected P450BM3 mutant with protected Vabica-
serin analogues 3 and 6 determined by LC-MS/MS. The product distribution
is shown by the %AUC contribution the M+16 metabolites to the overall re-
action composition. Red arrows on the structures mark the possible site of
oxidation in the M+16 products (M1 and M2 for reaction with 3 and M1-4
for reaction with 6). The region of where the oxidation had occurred was de-
termined by the fragmentation pattern in the MS/MS spectrum (in cases
where an unambiguous structural assignment based on MS/MS data was
not feasible potential sites of oxidation are marked with multiple red
arrows). Reaction conditions: Substrate (0.25 mm), “selected P450BM3 mutant”
(10 mgmLÀ1), Potassium phosphate buffer (pH 8, 100 mm), MeCN (1.25%),
NADP+(1 mm), Glucose (25 mm), Glucose Dehydrogenase (0.5 mgmLÀ1), at
r.t.
Combining substrate engineering with protein engineering
In order to demonstrate how d/p groups A–E can facilitate the
use of bio-LSO in general synthetic chemistry laboratories, the
protected analogues of Vabicaserin 3 and 6 were screened for
oxidation with a panel of 23 commercially available P450BM3
mutants.[47,48]
screening results are included in the supporting information).
The composition of the reaction of the free amine and the se-
lected mutant was mostly composed of unconverted starting
material (owing to likely enzyme inhibition, c.f. Figure 1) and
no M+16 products were observed. In contrast, the corre-
sponding reactions of the protected Vabicaserin analogues 3
and 6 saw the reaction composition being dominated by mon-
ohydroxylated products (M+16, metabolites M1 and M2 for 3
and M1À4 for 6, c.f. Figure 15).
LC-MS/MS was used to quickly identify the region where the
oxidation had occurred and a P450BM3 mutant displaying good
conversion and giving access to a diverse range of metabolites
was selected from the screening panel (see Figure 15, the full
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
8
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
The combination of Vabicaserin and d/p group B allowed
access to an aromatic hydroxylated product, whereas the use
of d/p group E was gave aliphatic oxidation. The reactions
with the selected mutant were scaled up and the products
characterized by NMR (see Figure 16 and 17). The use of the
ducing either corresponding phenols or aliphatic alcohols. The
use of commercially available screening panels of P450BM3 mu-
tants in combination with this d/p group strategy offers medic-
inal chemists a readily available toolkit of protecting groups
and P450BM3 mutants which can be screened to find the best
combination for the compound of interest. In addition, the ap-
plication of high throughput LC-MS/MS to the analysis the oxi-
dation reactions can quickly identify if any products of interest
have been generated and thus guide the mutant-protecting
group selection for preparative reactions. Scale up and the
ease of d/p group cleavage with for example, N-acetyl cysteine
generated oxidation products in milligram quantities. No at-
tempts were made at this stage to optimize the biocatalytic
oxidation for increased turnovers as the activity we realized
was sufficient to generate enough material for structural char-
acterization and biological testing. Furthermore, rather than
evolving improved catalysts that would not be accessible to
the average chemist, we aimed to rely on commercially avail-
able enzymes and kept synthesis protocols amenable to stan-
dard organic chemistry laboratories. Moreover, the carboxy-Ns
d/p groups have been shown to enhance the substrate solubil-
ity in aqueous media and thus could be beneficial in combina-
tion with other biocatalytic reactions. Future work could be di-
rected towards optimization of d/p group interaction with the
recognition motif in the enzyme via further protein engineer-
ing thereby tailoring the combination to yield the oxidation
products of interest, for example, in P450BM3 Y51 and R47
could be shifted to neighbored positions and/or hydrophobic
residues could be targeted for mutagenesis to incorporate
more polar residues which could interact with the d/p nitro
group. Furthermore, modifying the size of hydrophobic active
site residues via mutagenesis could broaden both the sub-
strate scope and diversity of metabolites generated.
Figure 16. Scale up reactions with the selected P450BM3 mutant and Vabica-
serin protected with d/p group 2. Reaction conditions: Substrate (0.25 mm),
selected P450BM3 mutant (10 mgmLÀ1), Potassium phosphate buffer (pH 8,
100 mm), MeCN (1.25%), NADP+(1 mm), Glucose (25 mm), Glucose Dehydro-
genase (0.5 mgmLÀ1), at r.t. (ii) N-acetyl cysteine, Cs2CO3, DMSO, r.t., yielding
7% and 11.6% of compounds 10 and 14 respectively.
Figure 17. Scale up reactions with the selected P450BM3 mutant and Vabica-
serin protected with group 5. Reaction conditions: (i) Substrate (0.25 mm),
selected P450 mutant (10 mgmLÀ1), Potassium phosphate buffer (pH 8,
100 mm), MeCN (1.25%), NADP+(1 mm), Glucose (25 mm), Glucose Dehydro-
genase (0.5 mgmLÀ1), at r.t. (ii) N-acetyl cysteine, Cs2CO3, DMSO, r.t., to give
compound 12 in 8% isolated yield.
The combined docking/MDS approach allowed the rationali-
zation of the d/p group related regioselectivity and oxidation
pattern, but also showed limitations identifying the SOM with
the shortest distance to the heme in some cases. Using addi-
tional P450BM3 crystal structures as well as longer MD simula-
tion times might overcome these limitations.
two different d/p groups in combination with the single select-
ed P450BM3 mutant, allowed access to another hydroxylated
product 14 in addition to the previously obtained phenol 10
and alcohol 12, demonstrating the ability of the 2-Ns d/p
groups to increase product diversity. In addition, the combina-
tion of this d/p group methodology with other P450 panels
containing the necessary d/p recognition motif, broadens the
accessibility of this bio-LSO technology removing the need for
specialist laboratories/ training. Based on the above-mentioned
analysis the d/p groups reported here are expected to work
best with P450BM3 mutants containing a suitably positioned
d/p recognition motif.
Experimental Section
Materials
All commercially available reagents were used without further pu-
rification. Deionized water was used for preparation of all aqueous
solutions and for work-up of reactions. The enzymes P450BM3 and
mutant F87V were obtained as crude lyophilized lysates from
Almac Sciences Limited. The commercially available P450BM3 screen-
ing kit used in this work was obtained from Codexis. The sulfonyl
chlorides used in preparation of the sulfonylamides were obtained
from Enamine Ltd.
Conclusions
Preparative P450 oxidation of 3 with P450BM3-F87V
The combination of substrate engineering with carboxylated 2-
Ns amine d/p groups and protein engineering with point mu-
tations has been used successfully to increase conversion of
the model compound Vabicaserin to desirable monohydroxy-
lated (M+16) analogues and to alter the regioselectivity pro-
Oxidation: A solution of 3 (25 mg, 0.052 mmol) in a 1:1 mixture of
acetonitrile/100 mm potassium phosphate buffer at pH 8 (2.73 mL),
was added to 210 mL of 120 mm potassium phosphate buffer at
pH 8 containing the NADPH recycling system: 1.2 mm NADP+,
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
9
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
30 mm Glucose, 0.6 mgmLÀ1 Glucose dehydrogenase. This mixture
was then added to a baffled 1 L conical flask containing a solution
of P450BM3-F87V mutant (200 mg in 40 mL of 100 mm potassium
phosphate buffer at pH 8), and the reaction mixture shaken at
228C, 300 rpm for 2 hours. The reaction mixture was diluted with
acetonitrile (300 mL) and stirred for 1 hour at r.t. before the pH
was adjusted to pH 2 via addition of HCl (1m) before the organic
solvent was removed in vacuo and the aqueous mixture extracted
with EtOAc (3ꢂ200 mL), the combined organics were then dried
(Na2SO4), filtered and concentrated to give a dark yellow residue
(14 mg) which was used directly in the next step without further
purification.
P450 Structure Preparation and Ligand Preparation
Crystal structures of substrate bound P450BM3 were downloaded
from the Protein Data Bank (http://www.rcsb.org: code 1FGA and
4KPA). Structures of the corresponding F87V mutations were gen-
erated by converting Phe87 into Val using the Mutate Residue tool
in the Schrçdinger Suite (Ref: Schrçdinger Maestro Suite 2016-2,
Schrçdinger, LLC; New York, NY: 2016.). The substrate-enzyme
structures were processed with Schrçdingers Protein Preparation
Wizard. Bond orders were assigned and adjusted, hydrogen atoms
were added and water molecules beyond 5.0 ꢁ from the ligand
were deleted as well as water molecules with less than 3 hydrogen
bonds to non-waters. The protonation and tautomeric states of the
amino acids and all hydrogen bonds were adjusted at pH of 7.0. Fi-
nally, the enzyme-substrate complex was minimized with conver-
gence of heavy atoms to an RMSD of 0.3 ꢁ using an OPLS3.
Deprotection: To a suspension of cesium carbonate (117 mg,
0.36 mmol) and N-acetylcysteine (19.45 mg, 0.117 mmol) in DMSO
(0.25 mL) was added a solution of the oxidation product isolated in
the previous step (14 mg) in DMSO (0.25 mL) and the yellow mix-
ture stirred vigorously at r.t. for 12 hours before the reaction mix-
ture was diluted with water (10 mL) and extracted with EtOAc (3ꢂ
10 mL). The combined organics were then washed with water
(30 mL) and sat. NaHCO3 (30 mL) before being concentrated to a
dark orange oil (6 mg) which was purified via preparative SFC
(column: Waters Viridis 2-EP column; mobile phase gradient: 95%
CO2 and 5% modifier (1 min), then 50% modifier (3 min), followed
by a wash step. As modifier MeOH with 0.2 vol% of concentrated,
aqueous ammonia solution was used), to give the phenol 10 as a
colorless oil (1.5 mg, 20.8% yield over two steps).
The protected substrates were prepared by using the LigPrep
Wizard and the OPLS3 force field in the Schrçdinger suite. For
each substrate all ionization states were enumerated at a pH of
7.4.
P450 Docking Studies
The docking grid file was generated based on the prepared X-ray
structures using the Schrçdinger Suite. The bound fatty acid in the
crystal structure was used to define the centroid of the substrate
binding site. All prepared substrates were docked using Glide, the
SP scoring function and default values. The best 5 docked poses
for each ligand were selected based on the Glide Score.
Preparative P450 oxidation of 6 with P450BM3-F87V
Oxidation: A solution of 6 (53 mg, 0.112 mmol) in a 1:1 mixture of
acetonitrile/100 mm potassium phosphate buffer at pH 8 (6.2 mL),
was added to 413 mL of 120 mm potassium phosphate buffer at
pH 8 containing the NADPH recycling system: 1.2 mm NADP+,
30 mm Glucose, 0.6 mgmLÀ1 Glucose dehydrogenase. This mixture
was then added to a baffled 1 L conical flask containing a solution
of P450BM3-F87V mutant (400 mg in 85.7 mL of 100 mm potassium
phosphate buffer at pH 8), and the reaction mixture shaken at
228C, 300 rpm for 6 hours. The reaction mixture was diluted with
acetonitrile (500 mL) and stirred for 1 hour at r.t. before the pH
was adjusted to pH 3 via addition of HCl (1m) before the organic
solvent was removed in vacuo and the aqueous mixture extracted
with EtOAc (3ꢂ400 mL), the combined organics were then dried
(Na2SO4), filtered and concentrated to give a dark yellow residue
(67 mg) which was used directly in the next step without further
purification.
Molecular Dynamics Simulations
Molecular Dynamics Simulations were performed for all selected
docking poses using the Desmond package in combination with
the OPLS3 force field in the Schrçdinger suite. Each docked sub-
strate-enzyme result was prepared for MDS by using Schrçdingers
System Builder tool. The system was solvated using the SPC sol-
vent model and 13Na+ ions in an orthorhombic simulation box.
For each protected substrate a 10 ns simulation was performed at
300 K and 1.013 bar. The system was relaxed using default values
before simulation.
Acknowledgements
The support by Stephanie Heitz (repeating some of the enzy-
matic oxidation experiments and sample preparation), Christi-
an Mꢃller (SFC purification), Claudia Krack (NMR analysis),
Sylvia Hellwig, Susann Brixner (LC-MSMS sample preparation),
Silvia Krakow, Vera Bender-Schubert (solubility measurements),
Anita Wilhelm-Alkubaisi (LC-MSMS analytic) and Markus Huber
(UV spectra) is gratefully acknowledged.
Deprotection
To a suspension of cesium carbonate (493 mg, 1.512 mmol) and N-
acetylcysteine (112 mg, 0.687 mmol) in DMSO (0.2 mL) was added
a solution of the oxidation product isolated in the previous step
(67 mg) in DMSO (0.5 mL) and the yellow mixture stirred vigorously
at r.t. for 2 hours before the reaction mixture was diluted with
water (10 mL) and extracted with EtOAc (3ꢂ10 mL). The combined
organics were then washed with water (30 mL) and sat. NaHCO3
(30 mL) before being concentrated to an orange oil (30 mg) which
was purified via preparative SFC (column: Waters Viridis 2-EP
column, elution gradient: 95% CO2 and 5% modifier (5 min) then.
50% modifier (3 min), followed by a wash step. As modifier MeOH
with 0.2 vol% of concentrated, aqueous ammonia solution was
used) to give phenol 10 and alcohols 11, 12 and 13 as colorless
oils (4.0 mg, 11.9% yield; 2.3 mg, 6.9%; 3.4 mg, 10.1% and 2.6 mg,
7.7% yield, respectively over two steps).
Conflict of interest
The design, study conduct, and financial support for this re-
search were provided by AbbVie. AbbVie participated in the in-
terpretation of data, review, and approval of the publication.
G.B., F.O. and U.E.W.L. are employees of AbbVie. C.V. and I.P.
were employees of AbbVie at the time when they contributed
to the study.
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
10
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[17] A. Hegde, D. C. Haines, M. Bondlela, B. Chen, N. Schaffer, D. R. Tomchick,
M. Machius, H. Nguyen, P. K. Chowdhary, L. Stewart, C. Lopez, J. A. Pe-
terson, Biochemistry 2007, 46, 14010–14017.
[18] a) M. A. Correia, P. F. Hollenberg, Cytochrome P450, Structure, Mechanism,
and Biochemistry, 4th ed., Springer, Basel, 2015, pp. 177–259; b) K. L.
Hanson, B. M. VandenBrink, K. N. Babu, K. E. Allen, W. L. Nelson, K. L.
Kunze, Drug Metab. Dispos. 2010, 38, 963–972; c) C. Bensoussan, M. De-
laforge, D. Mansuy, Biochem. Pharmacol. 1995, 49, 591–602; d) H. Hirao,
N. M. Thellamurege, P. Chuanprasit, K. Xu, Int. J. Mol. Sci. 2013, 14,
24692–24705.
[19] The propensity of P450s to form inhibitor complexes shown in Figure 1
depends of the identity of the P450 enzyme, therefore screening a very
diverse library of P450s could identify cases where no inhibition occurs.
[20] a) Z. Li, H. J. Feiten, D. Chang, W. A. Duetz, J. B. van Beilen, B. Witholt, J.
Org. Chem. 2001, 66, 8424–8430; b) D. L. Chang, H. J. Feiten, B. Witholt,
Z. Li, Tetrahedron: Asymmetry 2002, 13, 2141–2147.
[21] R. A. Johnson, M. E. Herr, H. C. Murray, G. S. Fonken, J. Org. Chem. 1970,
35, 622–626.
[22] a) C.-H. Yun, K.-H. Kim, D.-H. Kim, H.-C. Jung, J.-G. Pan, Trends Biotechnol.
2007, 25, 289–298; b) S. S. Boddupalli, B. C. Pramanik, C. A. Slaughter,
R. W. Estabrook, J. A. Peterson, Arch. Biochem. Biophys. 1992, 292, 20–
28.
[23] H. Li, T. L. Poulos, Nat. Struct. Biol. 1997, 4, 140–146.
[24] L. A. Cowart, J. R. Falck, J. H. Capdevila, Arch. Biochem. Biophys. 2001,
387, 117–124.
Keywords: docking/protecting group · enzyme catalysis · late-
stage oxidation · molecular dynamics · P450BM3-mediated
oxidation
[1] a) K. P. Cusack, H. F. Koolman, U. E. Lange, H. M. Peltier, I. Piel, A. Vasude-
van, Bioorg. Med. Chem. Lett. 2013, 23, 5471–5483; b) K. Gottfried, U.
Klar, J. Platzek, L. Zorn, ChemMedChem 2015, 10, 1240–1248; c) A. Rent-
meister, F. H. Arnold, R. Fasan, Nat. Chem. Biol. 2009, 5, 26–28; d) V.
Rea, D. Falck, J. Kool, F. J. J. de Kanter, J. N. M. Commandeur, N. P. E. Ver-
meulen, W. M. A. Niessen, M. Honing, MedChemComm 2013, 4, 371–
377.
[2] M. S. Chen, M. C. White, Science 2007, 318, 783–787.
[3] Q. Michaudel, G. Journot, A. Regueiro-Ren, A. Goswami, Z. Guo, T. P.
Tully, L. Zou, R. O. Ramabhadran, K. N. Houk, P. S. Baran, Angew. Chem.
Int. Ed. 2014, 53, 12091–12096; Angew. Chem. 2014, 126, 12287–12292.
[4] D. Falck, F. R. Pirkolachachi, M. Giera, M. Honing, J. Kool, W. M. A. Nies-
sen, J. Pharm. Biomed. Anal. 2014, 88, 235–244.
[5] S. T. Seger, P. Rydberg, L. Olsen, Chem. Res. Toxicol. 2015, 28, 597–603.
[6] B. Testa, P. Jenner, Drug Metab. Rev. 1981, 12, 1–117.
[7] “Inhibition of cytochrome P450 enzymes”: M. A. Correia, P. F. Hollen-
berg, Cytochrome P450: Structure, Mechanism, and Biochemistry, 4th ed.
(Ed.: P. R. Ortiz de Montellano), Springer, Cham, Heidelberg, New York,
Dordrecht, London 2015, Chap. 5, pp. 177–259.
[25] K. D. Dubey, B. Wang, S. Shaik, J. Am. Chem. Soc. 2016, 138, 837–845.
[26] A detailed analysis of all docking poses and their corresponding MDS
results can be found in the Supplementary Material.
[8] a) J. C. Lewis, F. H. Arnold, Chimia 2009, 63, 309–312; b) C. H. Hsieh,
T. M. Makris, Biochem. Biophys. Res. Commun. 2016, 476, 462–466;
c) G. E. Tsotsou, A. Sideri, A. Goyal, G. Di Nardo, G. Gilardi, Chemistry
2012, 18, 3582–3588; d) J. Reinen, J. S. van Leeuwen, Y. Li, L. Sun,
P. D. J. Grootenhuis, C. J. Decker, J. Saunders, N. P. E. Vermeulen, J. N. M.
Commandeur, Drug Metab. Dispos. 2011, 39, 1568–1576; e) E. Gillam, M.
Hayes, Curr. Top. Med. Chem. 2013, 13, 2254–2280; f) G. Di Nardo, G. Gi-
lardi, Int. J. Mol. Sci. 2012, 13, 15901–15924.
[9] a) T. S. Wong, F. H. Arnold, U. Schwaneberg, Biotechnol. Bioeng. 2004, 85,
351–358; b) V. Stepankova, S. Bidmanova, T. Koudelakova, Z. Prokop, R.
Chaloupkova, J. Damborsky, ACS Catal. 2013, 3, 2823–2836; c) S.
Kumar, L. Sun, H. Liu, B. K. Muralidhara, J. R. Halpert, Protein Eng. Des.
Sel. 2006, 19, 547–554.
[27] a) R. De Marco, M. L. Di Gioia, A. Leggio, A. Liguori, M. C. Viscomi, Eur. J.
Org. Chem. 2009, 3795–3800; b) C. Christensen, R. P. Clausen, M. Begtr-
up, J. L. Kristensen, Tetrahedron Lett. 2004, 45, 7991–7993; c) W. Kurosa-
wa, T. Kan, T. Fukuyama in Preparation of Secondary Amines From Pri-
mary Amines Via 2-Nitrobenzenesulfonamides: N-(4-Methoxybenzyl)-3-
Phenylpropylamine, Wiley, Hoboken, 2003, pp. 186-186; d) T. Kan, T. Fu-
kuyama, Chem. Commun. 2004, 353–359; e) T. Fukuyama, C.-K. Jow, M.
Cheung, Tetrahedron Lett. 1995, 36, 6373–6374; f) P. G. M. Wuts, J. M.
Northuis, Tetrahedron Lett. 1998, 39, 3889–3890; g) S. Vꢄzina-Dawod, A.
Derson, E. Biron, Tetrahedron Lett. 2015, 56, 382–385.
[28] Z. Tong, A. Chandrasekaran, W. DeMaio, R. Espina, W. Lu, R. Jordan, J.
Scatina, Drug Metab. Dispos. 2010, 38, 2266–2277.
[29] J. Kirchmair, M. J. Williamson, J. D. Tyzack, L. Tan, P. J. Bond, A. Bender,
R. C. Glen, J. Chem. Inf. Model. 2012, 52, 617–648.
[30] A. Tarcsay, G. M. Keseru, Expert Opin. Drug Metab. Toxicol. 2011, 7, 299–
312.
[31] P. C. Nair, R. A. McKinnon, J. O. Miners, Drug Metab. Rev. 2016, 48, 434–
452.
[32] P. H. Keizers, C. de Graaf, F. J. de Kanter, C. Oostenbrink, K. A. Feenstra,
J. N. Commandeur, N. P. Vermeulen, J. Med. Chem. 2005, 48, 6117–6127.
[33] M. Parmentier, M. K. Wagner, K. Magra, F. Gallon, Org. Process Res. Dev.
2016, 20, 1104–1107.
[10] a) K. Zhang, B. M. Shafer, M. D. Demars, H. A. Stern, R. Fasan, J. Am.
Chem. Soc. 2012, 134, 18695–18704; b) P. Meinhold, M. W. Peters, A.
Hartwick, A. R. Hernandez, F. H. Arnold, Adv. Synth. Catal. 2006, 348,
763–772; c) A. Trefzer, V. Jungmann, I. Molnar, A. Botejue, D. Buckel, G.
Frey, D. S. Hill, M. Jorg, J. M. Ligon, D. Mason, D. Moore, J. P. Pachlatko,
T. H. Richardson, P. Spangenberg, M. A. Wall, R. Zirkle, J. T. Stege, Appl.
Environ. Microbiol. 2007, 73, 4317–4325; d) O. Lentz, A. Feenstra, T. Hab-
icher, B. Hauer, R. D. Schmid, V. B. Urlacher, ChemBioChem 2006, 7, 345–
350; e) S. Kumar, E. E. Scott, H. Liu, J. R. Halpert, J. Biol. Chem. 2003, 278,
17178–17184; f) V. Rea, S. Dragovic, J. S. Boerma, F. J. J. de Kanter,
N. P. E. Vermeulen, J. N. M. Commandeur, Drug Metab. Dispos. 2011, 39,
2411–2420.
[34] It is reported that N-Boc-cysteine could be used as an alternative to the
unpleasant smelling mercaptoacetic acid for the nucleophilic aromatic
substitution of the sulfonyl group of on a modified Nosyl group
(Figure 4).
[35] J. J. Lee, S. Ha, H. J. Kim, H. J. Ha, H. Y. Lee, K. J. Lee, ACS Chem. Biol.
2014, 9, 2883–2894.
[11] a) C. J. C. Whitehouse, S. G. Bell, L.-L. Wong, Chem. Soc. Rev. 2012, 41,
1218–1260; b) R. Fasan, ACS Catal. 2012, 2, 647–666.
[12] a) V. Polic, K. Auclair, Bioorg. Med. Chem. 2014, 22, 5547–5554; b) D.
Chang, H. J. Feiten, K. H. Engesser, J. B. van Beilen, B. Witholt, Z. Li, Org
Lett. 2002, 4, 1859–1862; c) D. C. Haines, D. R. Tomchick, M. Machius,
J. A. Peterson, Biochemistry 2001, 40, 13456–13465.
[36] M. Adamczak, S. H. Krishna, Food Technol. Biotechnol. 2004, 42, 251–
264.
[13] A. R. H. Narayan, G. Jimꢄnez-Osꢄs, P. Liu, S. Negretti, W. Zhao, M. M. Gil-
bert, R. O. Ramabhadran, Y.-F. Yang, L. R. Furan, Z. Li, L. M. Podust, J.
Montgomery, K. N. Houk, D. H. Sherman, Nat. Chem. 2015, 7, 653–660.
[14] a) A. de Raadt, H. Griengl, H. Weber, Chemistry 2001, 7, 27–31; b) M. R.
Sarkar, E. A. Hall, S. Dasgupta, S. G. Bell, ChemistrySelect 2016, 1, 6700–
6707; c) S. Li, M. R. Chaulagain, A. R. Knauff, L. M. Podust, J. Montgom-
ery, D. H. Sherman, Proc. Natl. Acad. Sci. USA 2009, 106, 18463–18468.
[15] a) T. W. Huang, J. Zaretzki, C. Bergeron, K. P. Bennett, C. M. Breneman, J.
Chem. Inf. Model. 2013, 53, 3352–3366; b) G. Cruciani, Drug Discovery
Today Technol. 2013, 10, e145–e146; c) G. Cruciani, M. Baroni, P. Bene-
detti, L. Goracci, C. G. Fortuna, Drug Discovery Today Technol. 2013, 10,
e155–e165; d) P. Rydberg, D. E. Gloriam, J. Zaretzki, C. Breneman, L.
Olsen, ACS Med. Chem. Lett. 2010, 1, 96–100.
[37] a) The solubility measurements were performed at 248C, pH 8 with 1%
MeCN as a co-solvent (the same conditions as the biocatalysis reac-
tions). Detection limits of the solubility assay: maximum >600 mm and
minimum <5 mm). b) Conditions for P450BM3-F78V mediated oxida-
tion: c.f. general procedure for P450 mutant screen in Supplementary
Information.
[38] a) E. Stjernschantz, B. M. A. van Vugt-Lussenburg, A. Bonifacio, S. B. A.
de Beer, G. van der Zwan, C. Gooijer, J. N. M. Commandeur, N. P. E. Ver-
meulen, C. Oostenbrink, Proteins Struct. Funct. Bioinf. 2008, 71, 336–
352; b) B. M. van Vugt-Lussenburg, E. Stjernschantz, J. Lastdrager, C.
Oostenbrink, N. P. Vermeulen, J. N. Commandeur, J. Med. Chem. 2007,
50, 455–461; c) C. F. Butler, C. Peet, A. E. Mason, M. W. Voice, D. Leys,
A. W. Munro, J. Biol. Chem. 2013, 288, 25387–25399.
[16] J. Catalano, K. Sadre-Bazzaz, G. A. Amodeo, L. Tong, A. McDermott, Bio-
chemistry 2013, 52, 6807–6815.
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
11
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
[39] For MDS discussion see supplementary material.
[40] J. B. Schenkman, Biochemistry 1970, 9, 2081–2091.
[41] C.f. Figure S1 in the Supporting Information.
[42] C.f. Figure S2 in the Supporting Information.
[43] C.f. Figure S2A in the Supporting Information.
is uncertain as the original carboxylate binding motif (R47, Y51) may
have been altered.
[48] The commercially available P450BM3 kit used in this case was obtained
from codexis and contained engineered mutants of P450BM3 in which
the identity of the mutations made to each mutant is not disclosed.
http://www.codexis.com.
[44] C.f. Figure S3 in the Supporting Information.
[45] C.f. Figure S2B in the Supporting Information.
[46] The enzymatic oxidation of d/p protected secondary amines of similar
size to Vabicaserin worked successfully using P450BM3-F87V. Unpub-
lished results.
[47] The exact nature of the mutations made to these commercially avail-
able P450BM3 variants is not disclosed by the vendor and therefore the
presence of suitable binding motifs for the d/p groups in the mutants
Manuscript received: May 9, 2018
Revised manuscript received: September 12, 2018
Accepted manuscript online: && &&, 0000
Version of record online: && &&, 0000
&
&
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
12
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
Novel solubilizing docking-protecting
groups enable the regioselective
&
Biocatalysis
C. Vickers, G. Backfisch, F. Oellien, I. Piel,
U. E. W. Lange*
P450BM3-mutant mediated late-stage oxi-
dation of lead compounds by binding
the substrate (orange) in a biomimetic
fashion (e.g., palmitic acid, green). Enzy-
matic oxidation, mild deprotection and
purification yields hydroxylated deriva-
tives that are difficult to obtain by clas-
sical chemical synthesis.
&& – &&
Enzymatic Late-Stage Oxidation of
Lead Compounds with Solubilizing
Biomimetic Docking/Protecting groups
Combining biocatalytic oxidation with protective-group chemistry enables one to
modulate the site of oxidation. Lead compounds protected with different solubilizing
biomimetic docking/protecting groups yield different hydroxylated derivatives with
the same P450BM3-mutant. Additional hydroxylation products are obtained when the
P450BM3-mutant is also varied. Derivatives of this type are difficult to obtain otherwise.
Additionally, molecular-dynamic simulations help to rationalize the observed oxidation
patterns. For more information, see the Full Paper by U. E. W. Lange et al. on page &
& ff.
Chem. Eur. J. 2018, 24, 1 – 13
www.chemeurj.org
13
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ