Preparative-Scale Production of Testosterone Metabolites by Human Liver Cytochrome P450 Enzyme 3A4

Article abstract of DOI:10.1002/adsc.202000251

Just like the drugs themselves, their metabolites have to be evaluated to succeed in a drug development and approval process. It is therefore essential to be able to predict drug metabolism and to synthesise sufficient metabolite quantities for further pharmacological testing. This study evaluates the possibility of using in vitro biotransformations to solve both these challenges in the case of testosterone as a representative component for steroids. The application of cells of Pichia pastoris with expressed membrane-associated human liver cytochrome P450 enzyme (P450) 3A4 in two cycles of a preparative-scale bioreactor experiment enabled the isolation of the common metabolites 6β-hydroxytestosterone and 6β-hydroxyandrostenedione on a 100 mg scale. Side-product formation caused by enzymes intrinsic to P. pastoris was reduced. In addition more polar testosterone metabolites formed by a P450 3A4-catalysed bioconversion, than the known mono-hydroxylated ones, are reported and 6-dehydro-15β-hydroxytestosterone as well as the di-hydroxylated steroids 6β,16β-dihydroxytestosterone, 6β,17β-dihydroxy-4-androstene-3,16-dione and 6β,12β-dihydroxyandrostenedione were isolated and verified by NMR analysis. Their respective biological significance remains to be investigated. Whole-cell P450 catalysts expressed in P. pastoris qualify as a tool for the preparative-scale synthesis of human metabolites. Biotransformation processes in combination with standard chemical procedures allow the isolation and characterisation even of minor drug metabolite products. (Figure presented.).

Full text of DOI:10.1002/adsc.202000251

Title: Preparative-Scale Production of Testosterone Metabolites by
Human Liver Cytochrome P450 3A4
Authors: Nico Fessner, Matic Srdic, Hansjörg Weber, Christian
Schmid, David Schoenauer, Ulrich Schwaneberg, and Anton
Glieder
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Adv. Synth. Catal. 10.1002/adsc.202000251
Link to VoR: https://doi.org/10.1002/adsc.202000251

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
FULL PAPER
DOI: 10.1002/adsc.202000251
Preparative-Scale Production of Testosterone Metabolites by
Human Liver Cytochrome P450 Enzyme 3A4
Nico D. Fessner,a Matic Srdič,b,c Hansjörg Weber,d Christian Schmid,a,e David
Schönauer,b Ulrich Schwaneberg,f and Anton Gliedera*
a
Institute of Molecular Biotechnology, Graz University of Technology, NAWI Graz, Petersgasse 14/3, Austria
Fax: (+43)-316-873-9302; phone: (+43)-316-873-4074; e-mail: a.glieder@tugraz.at
SeSaM-Biotech GmbH, Aachen, Germany
Bisy GmbH, Hofstaetten, Austria
Institute of Organic Chemistry, Graz University of Technology, NAWI Graz, Austria
Austrian Centre of Industrial Biotechnology (ACIB), Graz, Austria
Institute of Biotechnology, RWTH Aachen University, Aachen, Germany
b
c
d
e
f
Received: ((will be filled in by the editorial staff))
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201######.((Please
delete if not appropriate))
Abstract. Just like the drugs themselves, their metabolites 3A4-catalysed bioconversion, than the known mono-
have to be evaluated to succeed in a drug development and hydroxylated ones, are reported and 6-dehydro-15β-
approval process. It is therefore essential to be able to predict hydroxytestosterone as well as the di-hydroxylated steroids
drug metabolism and to synthesise sufficient metabolite 6β,16β-dihydroxytestosterone,
quantities for further pharmacological testing. This study androstene-3,16-dione
6β,17β-dihydroxy-4-
6β,12β-
and
evaluates the possibility of using in vitro biotransformations dihydroxyandrostenedione were isolated and verified by
to solve both these challenges in the case of testosterone as a NMR analysis. Their respective biological significance
representative component for steroids. The application of remains to be investigated. Whole-cell P450 catalysts
cells of Pichia pastoris with expressed membrane-associated expressed in P. pastoris qualify as a tool for the
human liver cytochrome P450 enzyme (P450) 3A4 in two preparative-scale synthesis of human metabolites.
cycles of a preparative-scale bioreactor experiment enabled Biotransformation processes in combination with standard
the isolation of the common metabolites 6β- chemical
procedures
allow
the
isolation
and
hydroxytestosterone and 6β-hydroxyandrostenedione on a characterisation even of minor drug metabolite products.
100 mg scale. Side-product formation caused by enzymes
intrinsic to P. pastoris was reduced. In addition more polar
testosterone metabolites formed by a P450
Keywords: human drug metabolites; cytochrome P450
3A4, Preparative-scale synthesis; steroids; whole cell
biotransformation
reticulum
containing
cytochrome
P450
Introduction
monooxygenases (P450s)
– or computational
predictions,[10,11] and there are also novel concepts
emerging such as “organs-on-chips”.[12] However,
none of these approaches can provide sufficient
metabolite quantities. While conventional methods of
chemical synthesis could produce the required
amounts of such materials, it is often difficult and
time-demanding to functionalise structurally complex
drugs with specific regioselectivity. For these reasons
the potential of biocatalysis, operating in a synthetic
late-stage fashion,[13] was investigated by many
studies for the preparative-scale production of human
metabolites. There are several examples of successful
applications of recombinant human P450s,[14–16] and
for some specific products also microbial P450s were
deemed successful at providing typical human drug
metabolites at such a scale.[17–25] However, in many
Poor pharmacokinetics or toxicity caused a major
percentage of drugs to fail approval at a late stage of
drug development processes,[1] although clinical trials
are extremely costly and time-consuming.[2] Recently,
the FDA acknowledged that individual drug
metabolites might have a different pharmacological
or chemical activity compared to the parent drug, and
each must now be investigated separately to assess a
drug’s safety.[3] Consequently, an efficient, quick and
authentic identification as well as preparative
production of metabolites is highly important to the
pharmaceutical industry.[4]
Traditional methods to elaborate
a
drug’s
metabolic profile include animal models,[5–7] liver
microsomes[8,9] – vesicles of fragmented endoplasmic
1
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
cases microbial P450s resulted in different product
profiles and desired metabolites were only obtained
after time intensive enzyme engineering.
P. pastoris was the host organism of choice
because of the observed advantages in a comparative
expression study for human liver P450 2D6/CPR
catalysts in different standard expression hosts. P.
pastoris was identified to be superior to Escherichia
coli, Saccharomyces cerevisiae and Yarrowia
lipolytica.[45] Furthermore, also compared to other
yeasts convincing characteristics of P. pastoris are
high cell density achievable in a cheap growth
medium,[46] excellent capacity for native-like post-
translational modifications of eukaryotic proteins,[47]
tolerance for membrane protein production such as
for P450 3A4,[48,49] and strongly regulated promoters
that enable controlled growth on two individual
carbon sources (glycerol/glucose and methanol).[50,51]
These features allow a highly productive cell
cultivation and efficient bioconversion under
controlled conditions in a bioreactor.[46,49] First
successful expression of active human P450 3A4 was
reported in 2013.[52] Here we demonstrated the almost
complete conversion of about one gram of
testosterone in a 1.3 litre bioreactor system,
validating the potential of this system for further up-
scaling.
The body’s mechanism for clearance of chemicals
like drugs, typically consists of two enzymatic phases
with different significance.[26,27] In phase I the
compound is made more polar via hydrolytic
conversions or oxidative functionalisation, while
phase II consists of a conjugation step attaching polar
units like peptides, acids or sugar moieties to the
newly installed or liberated functional group. More
than 90% of the phase I enzymatic drug degradation
reactions are caused by human liver P450s.[28] Among
those, P450 3A4 is the key player in human
xenobiotic clearance. It is the most abundant enzyme
in this group,[29,30] and, due to its versatility,
responsible for the degradation of more than 50% of
approved drugs including testosterone (Scheme
1A).[28,31] Therefore, P450 3A4 in particular offers
itself as a representative and meaningful model and
tool for the study of enzymatic drug metabolism by
P450 enzymes. The goal was to employ the hepatic
function of this key enzyme for the preparative
synthesis and identification of the respective
metabolites to generate new data for future drug
evaluations and models for drug metabolite
predictions.
Testosterone
A)
B)
Drug
Liver P450 3A4 expressed in recombinant E. coli
was described to hydroxylate testosterone (1) at four
different positions, namely 6β, 2β, 1β and 15β in
descending order of rates.[32] Thus, 6β-
hydroxytestosterone (2) is typically used as the
determinant of the overall testosterone bioconversion
efficiency (Scheme 2A).[33] However, oxidation of 1
at positions 2α, 6α, 7α, 11β, 16α and 17 (forming the
ketone androstenedione (3)) was also observed when
using isolated human liver microsomes.[32,34] In fact,
upon applying 1 to hepatic rat microsomes Pfeiffer
and Metzler could detect at least 17 metabolites by
HPLC analysis, some of which included di-
hydroxylated derivatives of 1 and 3.[35] Although rats
are not a reliable model to predict drug metabolism in
humans,[36,37] it was still hypothesised that the
conversion of 1 in the human liver may produce a
similarly high number of metabolic products. Many
enzymes can produce metabolites; the related human
liver P450 2D6 for example also converts steroids,
albeit poorly.[38] Still, in that case a further
presumption was that P450 3A4 itself could be
mainly responsible for such a large metabolic product
spectrum due to its prevalence in the testosterone
metabolism and its extraordinary active site
promiscuity,[39,40] cooperativity[41] and multiple
substrate binding sites.[42,43]
P. pastoris
3A4
3A4
enzyme
3A4
3A4
3A4
3A4
P450
3A4
3A4
3A4
3A4
6b-OHT
OH
OH
OH
P450
3A4
P450
3A4
whole-cell
biotransformation
phase I
functionalisation
O
O
O
OH
& other
metabolites
phase II
conjugation
purification
enzyme
enzyme
O
O
OH
OH
-OOC
4
3
2
1
0
ppm
O
X
isolated 6b-OHT
HO
Scheme 1. Major routes of in vivo human drug metabolism
are compared to an in vitro imitation approach. A)
Administered testosterone drug (red) is metabolised in the
liver by P450 3A4 (green) and conjugation enzymes
(orange) in two phases to yield an O-glucuronide (pink)
derivative for excretion. B) The same testosterone drug is
added to cells of P. pastoris expressing recombinant P450
3A4 to simulate human phase I metabolism in vitro, and
allow isolation of metabolites.
In this study cells of Komagataella phaffii (Pichia
pastoris) containing the main wild-type P450 3A4 at
high expression level were used to simulate the
human metabolism of testosterone and to synthesise
its well-characterised metabolites such as 6β-
hydroxytestosterone at a 100 mg scale (Scheme
1B).[44]
2
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
A)
Testosterone (1)
6b-Hydroxytestosterone (2)
2 (28.7 min)
OH
OH
P450 3A4
4 (32.8 min)
H
H
1
200
100
0
2
15
H
H
H
H
O
O
6
OH
P. pastoris
P. pastoris
B)
(46.6 min)
O
O
1
3 (48.9 min)
H
H
P450 3A4
H
H
H
H
O
O
20
30
40
50
OH
min
Androstenedione (3)
6b-Hydroxyandrostenedione (4)
Scheme 2. Illustration of the major reactions observed in
this study employing whole-cell catalysts. A) Cytochrome
P450 3A4 catalysed the conversion of testosterone (1) to
6β-hydroxytestosterone (2); minor hydroxylation positions
of 1 reported in the literature for this enzyme are indicated
in blue. B) Competing redox enzymes intrinsic to P.
pastoris can oxidise 1 to furnish 4-androstene-3,17-dione
(3); Overexpressed P450 3A4 also accepted the latter as a
Figure 1. The HPLC analysis of metabolites of 1 obtained
by the bioconversion of 1 with P450 3A4 (2 mM 1, 1 mL
total volume, cell concentration = 200 OD600, 225 rpm,
22h, 30 °C,).
substrate
to
yield
the
corresponding
6β-
Formation of ketones 3 and 4 occurred, although
P450 3A4 is not known to oxidise 1 at position 17. In
fact, this oxidation happened regardless of the
presence of the expressed P450 as shown by the
corresponding negative control experiment using
wild-type P. pastoris cells, indicating the existence of
an intrinsic oxidase or dehydrogenase in P. pastoris,
competing with P450 3A4 for the substrate. After 24
hours more than 50% of 1 had been converted to 3
(Figure 2). This side reaction was noticed to be
reversible by subjecting 3 as the sole substrate, which
yielded an equilibrium between the two components
as a consequence of the redox state of the host cells
(Figure S1). The prevalence of low amounts of these
additional oxidation products compared to the control
reaction employing cells expressing no P450 3A4
indicated high efficiency of the expressed human
P450 enzyme.
Furthermore, these results suggested the
responsibility of a dehydrogenase for this reaction
rather than an oxidase. The same effect had been
noticed in the related yeasts S. cerevisiae and
Schizosaccharomyces pombe (S. pombe)[54] and the
intrinsic glucose-6-phosphate dehydrogenase was
suggested as a possible candidate for steroid
interconversion in S. cerevisiae.[55] In P. pastoris
related intrinsic alcohol dehydrogenases (ADHs)
might be responsible for the oxidation of 1 to 3 over
time. However, many dehydrogenases can be found
in the organism’s genome as dehydrogenases play a
role in different biochemical pathways.[56] For
example, three ADHs of P. pastoris that catalyse the
reversible oxidation of alcohols with simultaneous
reduction of the cofactor NAD(P)+ to NAD(P)H had
been described in more detail.[57] The identification of
the one(s) specifically accounting for the observed
steroid conversion remains to be investigated.
hydroxyandrostenedione 4, confirming observations of
previous studies.[31]
Results and Discussion
The analysis of steroids is symptomatically
challenging because of the small polarity changes
induced upon functionalisation of the large
hydrophobic scaffolds.[53] The development of a high-
resolution separation method was therefore key to the
successful analysis and isolation of steroidal
metabolites, which is why the proven HPLC
conditions of Pfeiffer and Metzler[35] were adapted.
Commercial human P450 3A4/hCPR expressing P.
pastoris cells (stored as frozen cells at -80 °C) were
diluted to obtain standardised whole-cell catalyst
stock solutions of a cell concentration of OD600 = 200,
which was used for all experiments. Figure 1
displays the metabolite profile obtained from a 1 mL-
scale test experiment using 2.0 mM 1 after a 22 hours
biotransformation. The high number of peaks with a
wide Rf spread suggested the formation of various
products with rather different polarity. Peaks eluting
at 46.5, 28.7 and 48.8 min could be assigned to
compounds 1, 2 and 3, respectively, with the aid of
authentic reference materials. The peak of compound
4 was deduced from corresponding results shown in
Figure 3. The other peaks around 2 and 4 could only
partly be assigned during the course of this study.
3
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
1
600
400
200
0
4
1 (46.4 min)
1h
600
400
200
0
3
2
20h
8h
4.5 mM
4h
20h
8h
4h
4h
2.5 mM
0.5 mM
3 (48.8 min)
20h
8h
4h
24h
30
40
min
50
46
49
47 min 48
Figure 3. HPLC traces of the bioconversion of 1 at
different concentrations (0.5, 2.5 and 4.5 mM shown in red,
blue and green, respectively) by P450 3A4 in shake flasks
(30 mL, 130 rpm, 24h, 28 °C, OD600 = 200). Samples of
each biotransformation were taken at three different time
points in between 4 and 20 hours.
Figure 2. HPLC monitoring of a bioconversion of 1 using
the empty vector control strain lacking P450 3A4
expression, which shows oxidation to 3 by P. pastoris’
intrinsic enzymes (2.5 mM 1, 30 mL, OD600 = 200, 130
rpm, 24h, 28 °C). After 24h, the concentration of ketone 3
exceeded that of substrate 1.
Having established the optimal reaction time, it was
attempted to further minimise the side-product
formation by either changing the availability of the
ADHs or their co-factors within the cells. The former
can be achieved by addressing the ADH expression
levels, which depend on the cell’s metabolic state.
Usually strains of P. pastoris are grown on glucose or
glycerol as carbon source, and the recombinant
expression of the desired protein is subsequently
induced by the addition of methanol in MutS-type
strains (Methanol utilisation Slow), only then
activating the responsible, tightly regulated
promoter.[58] A change in metabolism will also trigger
or suppress the expression of ADHs.[59] Following
methanol induction, cells were thus again exposed to
either glucose or glycerol for 3 hours before the
biotransformation. In another attempt to enhance the
availability of reduced cofactors, methanol was added
to the biotransformation not only to limit the
availability of NAD(P)+, but also to supply a
substrate competing with 1 for the ADH active site.
In addition, high NAD(P)H concentrations supply
sufficient electrons to P450 3A4 via the reductase,
which often represents the rate-limiting step.[60]
Because high methanol concentrations are rather
lethal to P. pastoris,[61] three different concentrations
from 0.5 to 3% were tested. The results of both these
strategies are presented in Figure 4.
Yet, due to efficient overexpression of the
recombinant enzyme, the presented yeast model does
not lead to an unrealistic metabolite profile for human
testosterone metabolism in the liver as ketone
formation also happens when using human liver
microsomes as mentioned before.[34] However, the
side-product formation negatively affected the
preparative-scale metabolite synthesis, as the yield of
individual components was reduced, and their
subsequent separation became more complicated.
Since 3 is likewise accepted by P450 3A4 as a
substrate (Scheme 2B),[31] this leads to a much more
complex product mixture, essentially composed of
duplicate sets of complementary metabolites when all
derivatives of 1 were produced in the equivalent
ketone form.
Controlling the reaction progress thus became
crucial to maximise product yield by anticipating the
time, at which the rate of oxidation of 2 to 4 was
higher than the rate of its formation, in order to stop
the conversion beforehand. As indicated best in
Figure 3 (upper traces, 4.5 mM 1), concentrations of
1 decreased over time while those of 3 increased, and
likewise the peaks of the desired product 2
diminished, while those of 4 rose correspondingly.
Because peaks of 2 at 8 hours reaction time were
slightly lower than at 4 hours both for runs with 2.5
and 4.5 mM 1, the optimum point in time had already
passed in between. With 0.5 mM of 1, one hour was
sufficient to fully exhaust the substrate (data not
shown). In comparison, a concentration of 4.5 mM 1
was too high for the biocatalyst to be fully used-up
within 20 hours, indicating decay in enzyme activity
over time or potential substrate/product inhibition at
higher concentrations. Consequently, for subsequent
biotransformations 2.5 mM of substrate 1 was chosen.
4
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
Conversion of 1
Conversion to monohydroxylated 1
Therefore, it seems like both strategies for increasing
product selectivity were successful individually as
well as in combination, and suppression of ADH
oxidation could enhance selectivity from 54% for
untreated conditions to a peak value of 79% (glucose
and methanol). A study demonstrated how the co-
expression of the human 17β-hydroxysteroid
dehydrogenase type 3 could further suppress the side-
product formation in P. pastoris.[55] Another option
would be the generation of knock-out strains lacking
ADH genes via knock-out, eg. using recently
established CRISPR/Cas9 technology.[62]
Ref%
percentage of
of
100
90
80
70
60
50
40
30
20
10
0
60%
79%
60%
77%
58%
75%
71%
54%
With reaction conditions optimised, a scale-up
biotransformation (BT1) from a several 10 mL scale
in 2.5 L cultivation flasks to 1.3 litres in a bioreactor
under controlled conditions was performed next
(Figure 5). Almost 1 gram of 1 was used with
implementation of 1% methanol addition.
6133 7343 7444 6638 7457 7559 6445 6448
-
L
-
H
H
H
C
H
H
-
L
O
O
O
O
O
O
e
e
e
e
e
G
,
G
,
M
M
M
M
M
h
h
3
3
%
%
%
%
%
5
1
3
1
1
.
0
L
C
L
O
G
G
h
h
3
3
Figure 4. Conversion of 1 (black) to the respective fraction
of mono-hydroxylated derivatives of 1 (red) are shown for
biotransformations with cells in microwell plates under
standard conditions (2.5 mM testosterone, 0.4 mL, 310
rpm, 17h, 28 °C, OD600 = 200) without (---) and with the
addition of methanol (0.5, 1 and 3% MeOH), or with cells
pre-treated by 3 hours cultivation in glycerol (3h GOL) or
glucose (3h GLC) and a combination of cultivation and
methanol addition (3h GOL, 1% MeOH; 3h GLC, 1%
MeOH). For a comparison, the calculated Ref% indicates
the percentage of formed conversion to mono-
hydroxylated products relative to the total uptake of 1.
biotransformation
wash down
400
1
2
3
4
200
0
15 min
30 min
1h
2h
3h
4h
5h
7h
8h
Zone:
A
B
C
D
10
20
30
40
50
60
min
Figure 5. HPLC monitoring of the scale-up
biotransformation (BT1) in a bioreactor under controlled
conditions is shown (2.5 mM of 1 = 959 mg, 1.3 L, 400
rpm, 8h, 28 °C, OD600 = 200, pH 7.0, air flow = 5.0 L/min).
All
biotransformations
were
intentionally
incubated for 17 hours, i.e. beyond the optimal
reaction time, to intensify a potential reduction effect
on the 17-ketone formation. “Ref%” was calculated
for better quantification and facilitated comparison of
the individual approaches. For the total conversion of
1 (black bars) all cumulated conversions caused by
P450 3A4 were considered, while disregarding ADH
involvement; thus 4 was included since hydroxylation
must have preceded the oxidation, but not 3. The total
conversion data also acknowledge only three mono-
hydroxylated derivatives of 1 leaving many extra
peaks unaccounted (Figure S2). A significant
difference could be observed across the different
approaches relative to untreated cells. Addition of
methanol alone increased the total conversion of 1
and the fraction of mono-hydroxylated products as
seen by a slight increase in Ref%. Upon pre-
incubation with glucose, the conversion increased to
a similar level as for methanol addition, while
glycerol pre-treatment had almost no effect.
Interestingly though, less oxidation to ketones
occurred in both cases as represented by significantly
higher Ref% values suggesting that the ADHs present
in P. pastoris cells that were tuned to carbon source
metabolism cause less steroid oxidation. In
combination with methanol addition both cases
After 8 hours (black trace) a conversion of 85% and a
mono-hydroxylation selectivity factor of 62% had
been achieved, similar to the results in shake flasks
(Figure 3), but significantly better than by 96-well
plate cultivation (Figure 4). Enzyme stability and
oxygen-transfer rates were considered to be some of
the major limitations of P450s. Quite likely, the
oxygen requirements of P450s had been met more
accurately by the greater oxygen supply in the
bioreactor.[63] A high cell density needs careful
adaptation to match the enzyme’s oxygen demand,
and thus the full potential of a cell density at OD600
=
200 might not have been fully exploited.[64] On the
other hand, the known rather poor stability of P450
3A4[30] may account for the observed shrinking in
conversion rate from 21% (Ref% = 87%) during the
first 15 minutes (15 min, orange) to 45% (81%) after
one hour (1h, turquoise) and 73% (70%) after five
hours (5h, green) of reaction time despite the
application of whole-cell biocatalysts, which supports
membrane-bound enzymes with the supply of
cofactors and electrons.[60,65]
generally experienced
a
slight further boost.
5
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
The recovered cells from the first batch were used for
a second cycle of biotransformation (BT2) in the
bioreactor under identical conditions in order to test
the durability of the catalyst system and the amount
of testosterone substrate that can be metabolised. In
fact, within 9 hours (9h, blue) of the second cycle
another 66% of 1 gram of the substrate was converted
(Figure 6) with a Ref% of 58% that dramatically
dropped afterwards. This is an extraordinary
performance of a human P450 considering their poor
stability and short lifetime usually described in
literature.[66]
HPLC samples by mass detector indeed strongly
supported such hypothesis (Figure S3). This
experiment also indicated the presence of five, not
just four mono-hydroxylated derivatives of 1.
A zoomed-in perspective of zones A and B from
Figure 6 was not sufficient to analyse the different
compounds in detail (Figure S4). Another HPLC
method was developed specifically for compounds
eluting in zones A and B and revealed a complex
picture of more than ten different peaks as displayed
in Figure 7.
biotransformation
wash down
9h
400
2
37h
1
4
3
48h
88h
0h
2h
9h
200
11h
37h
48h
65h
A
B
105h
0
Zone:
A
B
C
D
5
10
15
min
20
25
10
20
30
min
40
50
Figure 7. An illustration of the HPLC traces of zones A
and B of BT2. A new HPLC method was used for better
separation of the more polar metabolites.
Figure 6. HPLC monitoring for the second cycle of the
scale-up biotransformation (BT2) in a bioreactor under
controlled conditions (2.5 mM of 1 = 959 mg, 1.3 L, 400
rpm, 8h, 28 °C, OD600 = 200, pH 7.0, air flow = 5.0 L/min)
is shown.
For isolation and product characterisation, the
extracted metabolite mixtures from runs BT1 and
BT2 were separately pre-purified by manual column
chromatography in order to simplify the product
isolation by preparative HPLC. However, further
HPLC purification was obsolete because clean
fraction of 2 and 4 could be collected, which
furnished 108 mg of 2 from the mixture of the first
bioreactor cycle (yield = 11.3%, productivity[67] [g /
g] using dry cell weight = 0.16%) and 87 mg of 4
from the second (9.1%, 0.13%). Higher masses were
reported for human metabolites of other drugs,[68] but
to the best of the author’s knowledge the quantities
isolated here represent the highest reported in
literature for these particular metabolites in
comparison to other preparative studies.[16,69,70]
Additionally, a refinement or repetition of the
purification procedure holds the potential of
increasing the yield further.
From both mixtures of the two cycles of
biotransformations executed in the bioreactor, two
sets of two samples of about 10 mg each could also
be isolated by column chromatography. They
contained more polar compounds than 2 or 4
seemingly pure as determined by TLC analysis. NMR
and HPLC analysis, however, clearly indicated the
need for further purification, which was done by
collecting the fractions manually using an analytical
HPLC instrument as illustrated in Figure S5. At a
Between 17 (not shown) and 37 hours the peaks of
compounds 1 and 2 had vanished completely.
Nevertheless, the biotransformation was continued
for a total of 105 hours to ensure all derivatives of 1
had been dehydrogenated to their respective ketone
equivalents. The reduced number of peaks simplified
the HPLC analysis and revealed only those peaks
belonging to derivatives of 3. The controlled reactor
conditions, efficient performance of the biocatalyst,
large amount of substrate and frequent sampling
provided a more comprehensive picture about the
kinetics of substrate consumption and sequence of
metabolite formation. Apparently, all metabolite
peaks in Figures 5 and 6 clustered into four distinct
zones A to D according to their polarity and their
elution times (marked in blue). Because zone D
comprises the untreated, non-hydroxylated substrates
1 and 3 and zone C the mono-hydroxylated
metabolites 2 and 4, it seemed logical to assume di-
hydroxylated metabolites in zone B and even tri-
hydroxylated ones in zone A. The extent the
regiochemistry of hydroxylation affects the polarity
of the molecule decreases with rising degree of
hydroxylation matching the smaller zone widths
towards shorter elution times. Analysis of the same
6
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
wavelength of 270 nm the metabolites had only about
10% absorbance enabling a more precise isolation of
specific peaks than a preparative HPLC would have
achieved (Figures S6-S9). Compounds 1β- (5) and
15β-hydroxytestosterone (6) could be isolated from
BT1,[71,72] which confirmed the results published by
Guengerich et al.[32] In addition, 6-dehydro-15β-
hydroxytestosterone (7) could be uncovered as the
minor component of a mixture with 6. More polar
compounds of BT1 were only present as mixtures and
with quantities of less than 1 mg each, making an
identification by NMR spectroscopy impossible. The
same was true with many of the isolated fractions of
samples of BT2. However, for the first time the NMR
spectra of four isolated peaks confirmed the
formation of the di-hydroxylated compounds 6β,16β-
dihydroxytestosterone
androstene-3,16-dione
(8)
(9)
6β,17β-dihydroxy-4-
and 6β,12β-
dihydroxyandrostenedione (10). Compounds 7, 8 and
10 were also confirmed by high-resolution mass
spectrometry, while the spectrum for 9 was
inconclusive due to the presence of several other
compounds. Their elution times and their structures
are displayed in Figure 8. Compound 5 eluted
between the peaks of 2 and 4. Most likely, the peak
eluting just after 4 is 2β-hydroxytestosterone, which
was not isolated in this study. Consequently, the rate
of formation of the mono-hydroxylated metabolites
reported by Guengerich et al.[32] could not be
confirmed here, but looks rather like 6β > 15β ≈ 2β >
1β.
Figure 8. Display of the elution times of the newly
discovered
metabolites
6-dehydro-15β-
hydroxytestosterone (7), 6β,16β-dihydroxytestosterone (8),
6β,17β-dihydroxy-4-androstene-3,16-dione (9) and 6β,12β-
dihydroxyandrostenedione (10) relative to the already
known 15β-hydroxytestosterone (6).
Although the discovered metabolites 7 – 10 occupy
a noteworthy area of the HPLC profile at least in this
study, their existence remained unknown until now.
Having found evidence for the presence of di-
hydroxylated metabolites of P450 3A4, the peaks
around those of 8 and 10 will likely be either of the
same sort or of other oxidised species like 9 or 7.
Given the fact that many NMR spectra indicated the
presence of compound mixtures, although some of
these were derived from just one HPLC peak, there
must be even a larger amount of individual di-
hydroxylated compounds formed than represented by
the number of peaks.
Providing more stable catalysts in combination
with highly efficient (co-)expression levels of these
enzymes could therefore generate
a
different
metabolic profile. Here, high stability was achieved
in form of whole-cell biocatalysts using robust yeast
chassis, which also allows cost effective scale up and
independence of NADPH addition or regeneration.
Alternatively ancestral sequence reconstruction of the
P450 3 family was also found to enhance stability of
the enzyme itself without a loss in activity, albeit
slightly changing its regioselectivity.[75]
Remarkably,
poly-hydroxylated
testosterone
products have barely been considered in combination
with microsomal liver enzymes. Database searches
yielded only few articles that suggested the presence
of poly-hydroxylated testosterone metabolites since
1968.[35,73,74] Perhaps this might be due to the
chronically bad stabilities and thus short lifetimes of
recombinant human P450s causing a rapid decline in
enzyme activity after a few hours in the most
commonly used microbial host organisms.[45]
Another reason might be that such profile is
beyond the normal expectations. Not realising the
complexity of the human liver P450s and their
metabolic spectrum generated, which is still not
completely understood nowadays,[31] many studies
focused on the major metabolite 2, which is accessed
most easily and promises good results more quickly.
Furthermore, the hydroxylation of 1 is not a
prerequisite for entering phase II metabolism because
the molecule already has an alcohol functional group
attached (Scheme 1). The majority of phase I steps
rather involve 17-oxidation, reductions of the A ring
or the 3-position making consecutive hydroxylation
steps less predictable.[76] Nevertheless, just like
compound 5 was discovered in 2004 as a novel
metabolite[32] and in-vitro experiments pointed
towards some physiological potential,[77] the
7
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
existence of such poly-hydroxylated products might
as well have pharmacological relevance, which is yet
to be investigated.
Wet cell wt.
used
100 g
200 g
Substrate
used
1
288 mg
959 mg
Hydroxylated testosterone products may be already
minor metabolites in the human liver,[76] though it is
still desirable and necessary to identify and synthesise
such minor metabolites in sufficient quantities.[78]
Clearly, efficient biocatalysis enables access to
them.[79] Owed to its high cell density, P. pastoris is
frequently advertised by literature as excellently
suitable for its application in large-scale bioreactor
(25 mL, 40 mM) (3.33 mL, 1 M)
Product
isolated
2
59 mg
108 mg
11.3%
0.16%
Yield of 2 (%)
20.5%
Productivity[67] n.d. (no dry cell
wt. data
available)
experiments[49,80–83]
. Human P450 3A4 enzyme
catalyst preparations based on P. pastoris[84–86] are
commercially available from several companies,
claiming also multiple cycles of biotransformation.[70]
However, no peer-reviewed example of the
bioreactor-scale application of human P450 enzymes
produced by P. pastoris followed by several cycles of
whole-cell biotransformations and description of
experimental procedure for such scalable human
P450-catalysed biotransformations had been
published so far.
Alternatively, E. coli expressing microbial
monooxygenases was used to make some individual
metabolites of testosterone with excellent
yields.[21,79,91,92] Those enzymes are soluble, can be
easily tuned via directed evolution, are often self-
sufficient with high coupling efficiencies and
turnover numbers, and have a more narrow substrate
tolerance than the human liver P450s. However, often
non-human metabolites are formed as main products
and long lasting enzyme engineering was necessary
to change the substrate and product selectivity,
making it inconvenient human drug metabolites of
new drug candidates.
Reliable expression levels are difficult to obtain for
membrane-bound proteins and hardly give
representable information about the catalytic
efficiency of an enzyme that is so dependent on the
electron-transfer from the reductase to the heme
domain.[60,87,88] The volumes used for cultivation and
the subsequent biotransformation outcome provide
Incubation of 2 with the empty plasmid cells as the
negative control showed no conversion, other than
the alcohol oxidation ruling out an influence of the
yeast’s ADHs. Therefore it seems like 2 and 4 are
also acceptable substrates of P450 3A4 as already
implied by the study of Pfeiffer and Metzler, who
identified their di-hydroxylated metabolites by adding
mono-hydroxylated ones to the liver microsomes.[35]
The presence of a 6β-hydroxyl seems to change the
enzyme’s hydroxylation regioselectivity relative to 1
because positions 16 and 12 were observed as the
second hydroxylation site. These results are in line
with those by Guengerich et al.,[35] whose results
suggested a major selectivity difference of P450 3A4
upon the small change in the substrate structure 1 to
better comparability: While in
a
bioreactor
experiment using whole-cell E. coli[16] human P450
3A4 catalysts, Vail et al. needed more than 3 L of the
E. coli culture to perform their conversion of 1 and to
isolate 59 mg of 2 in 20.5% yield. In this study the
reaction employed less than 0,5 L of the yeast culture
broth to obtain 108 mg of the same metabolite (Table
1). In addition due to their size P. pastoris cells are
easy to remove from the reaction broth. Yields of
other metabolites than 2 were not discussed in the E.
coli
based
biotransformation.
Furthermore,
confirmation of the identity of their product was only
provided using LC-MS.
dihydrotestosterone with
a
reduced Δ4,5-bond.
While P450 3A4 was also successfully expressed
in other yeasts,[89,90] no preparative biotransformation
with testosterone was performed with those catalysts.
On the other hand, using S. pombe excellent human
P450 expression and biocatalysis was exemplified by
Drꢀgan et al., but no direct comparison is possible as
a different human P450 converting another drug was
reported for their preparative synthesis.[68]
However, the here discovered α,β,γ,δ-unsaturated
compound 7 was formed either after sequential or
step-wise hydroxylation i.e. Δ6,7-elimination after the
first or second hydroxylation at positions 6 and 15.
This means that position 15 is either a conserved
target of P450 3A4 with 2 as the substrate or the
regioselectivity was not affected by the structural
changes of double unsaturation. It seems also
plausible that just like the 17β- and 16β-hydroxyl
oxidation, the Δ4,5-bond reduction was caused by the
ADHs, and compounds 7, 9 and 10 are therefore no
natural metabolites generated by P450 3A4
biocatalysis alone. But it seems reasonable to assume
that the 17β-alcohol equivalent of compound 10 will
be such metabolite. Interestingly, P450 3A4 seems to
have a stereoselective preference for the β-face of the
steroidal scaffold in agreement with the observations
of Guengerich et al.[93] Not only all mono-, but also
the newly identified di-hydroxylations of 1 occurred
on the β-face.
Table 1: Comparison of the cultivation and
biotransformation parameters between this study and
a previous publication of Vail et al.[16]
Vail et al.[16]
BT1, this
study
Cultivation
Volume
10 L
5 L
Wet cell wt.
produced
~ 300 g
~ 2000 g
8
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
80 °C. NMR spectra were recorded with a Varian/Agilent
Inova 500 MHz NMR spectrometer equipped with an
indirect detection probe 5 mm.
Conclusion
Human liver enzyme P450 3A4 apparently diversifies
the steroidal scaffold of 1 in a late-stage fashion to a
larger extent than previously thought.[32] Not only
four mono-hydroxylated derivatives at positions 1β,
2β, 6β and 15β are formed,[13] but also several di-
hydroxylated and perhaps even at least one tri-
hydroxylated metabolite. In this work so far the
positions 16β and 12β could be clearly identified as
accessible sites for a second hydroxylation by P450
3A4. However, the data presented suggest the
presence of a vast number of additional poly-
hydroxylated steroids opening up the opportunity for
many interesting future discoveries with the help of
an improved HPLC separation method combined
with highly advanced analytical instrumentation used
by Guengerich et al.[32]
(A) Reaction tube biotransformation: HPLC
profile analysis (Figure 1)
Test tubes with screw caps (20x150 mm) were used. A cell
concentration OD600 of 200 was generated by resuspending
cells in 100 mM phosphate buffer (pH 7.4). The
biotransformation was started by adding 25 μL of 100 mM
testosterone in DMSO to 975 μL of the cell solution and
the reaction mixture was incubated at 28 °C and 225 rpm.
The biotransformation was stopped after 22 hours by
adding 1 mL of a 1:1 mixture (v/v) of acetonitrile/methanol.
The resulting mixture was vortexed, centrifuged at top
speed and the supernatant taken for HPLC analysis.
Compounds were separated via a reverse-phase column
Zorbax SB-C18 (21.2 mm i.d. x 25 cm). Water containing
0.1% acetic acid (A) and acetonitrile (B) were used for
elution at 25 °C in the following ratios: 0 min: A/B 75/25;
50 min: A/B 0/100; 52 min: A/B 75/25; 60 min: A/B 75/25.
This study displays how enzymatic metabolite
production can balance the industrial production
requirements of time, quantity and profile
authenticity.[63] The constraints of efficiency, stability
and scalability of recombinant human P450s often
reported in literature[14,66] could be successfully
bypassed using P. pastoris-based whole-cell
biocatalysts at efficient expression levels. Such robust
tool enabled the synthesis of new human metabolites
at a preparative scale for the first time. The yeast’s
protein production features allow for easy
transferability of the results to further scale-up
strategies and potential industrial application[80]
because the optimal conditions (pH, oxygen content,
temperature, nutrient supply) for biomass growth,
P450 enzyme expression and substrate conversion
can differ dramatically among individual hosts and
chemical reactions.[4,63] Hence, easy adaption to a
bioreactor is just as important as strategies for
optimisation strategies.[66] Moreover, the application
of surprisingly simple standard chemical procedures
for reaction work-up, product purification and
(B) 2.5 L shake flask biotransformations:
Kinetic study (Figures 2 and 3)
Cells were resuspended in 30 mL of phosphate buffer until
an OD600 of 200 was obtained and the broth was filled into
a sterile 2.5 L shake flask, 0.5, 2.5 or 4.5 mM of 100 mM
testosterone in DMSO or 2.5 mM of 100 mM
androstenedione in DMSO was added. For the negative
control, wild-type cells of P. pastoris were handled as the
other samples with
a
2.5 mM testosterone end
concentration. The flasks were shaken at 130 rpm for 24
hours. Samples (1 mL) were taken after 4, 8, 20 and 24
hours, and treated as well as analysed as described above.
(C) 96-well plate biotransformations: Change
in metabolism (Figure 4)
Cells were resuspended in 40 mL of 100 mM phosphate
buffer (pH 7.4) until an OD600 of 200 was obtained and the
broth was filled into a sterile 250 mL shake flask. Then
1030 μL of 20% glucose, 340 μL of 60% glycerol were
added to obtain a final concentration of 0.5%. Control cells
were left untreated. The cells were shaken for 3 hours at
140 rpm before they were used for biotransformation. The
cell broth was added into a 96-deep-well-plate (390 μL
each) and the reaction was started with the addition of 10
μL of 100 mM testosterone in DMSO. Methanol was
added into the respective wells to obtain a final methanol
concentration of 0.5, 1 or 3%. For each variation of the cell
conditions, 14 repeats were tested. The plates were
incubated at 28 °C at a speed of 320 rpm in a tilted
orientation on the shaker to ensure maximal oxygen
availability. The addition of 1 mL of a 1:1 mixture (v/v) of
acetonitrile/methanol stopped the reaction. The plate was
then centrifuged at 16,100 g for 10 min and the supernatant
was transferred into 96-well GreinerV plates for HPLC
analysis. Separation was carried out via a Kinetex C18
(100 Å; 50 x 4.6 mm; 2.6 μm) reverse-phase column. A
positive electrospray ionisation mode was selected for the
mass spectrometer. Water containing 0.1% acetic acid (A)
and acetonitrile (B) were used for elution at 25 °C in the
following ratios: 0 min: A/B 80/20; 1 min: A/B 80/20; 3
min: A/B 0/100; 5 min: A/B 0/100; 5.01 min: A/B 80/20; 6
min: A/B 80/20.
analytical
identification
should
make
the
implementation of this approach worthwhile for
chemists and industry alike.
Experimental Section
All solvents and chemicals were purchased from Sigma-
Aldrich/Merck (Steinheim/Darmstadt, Germany), VWR
International (Fontenay-sous-Bois, France), Carl Roth
GmbH (Karlsruhe, Germany) or Fisher Scientific
(Loughborough, UK) in best available purity and were
used as received without further purification. HPLC tubes
were bought from Macherey-Nagel (Düren, Germany) and
the corresponding caps and inserts from Bruckner
Analysentechnik (Linz, Austria). In experiments (A) and
(B) an Agilent Technologies 1100 Series HPLC was used,
for experiments (C) and (D) an Agilent Technologies 1200
Series HPLC system coupled with a G1956B mass
selective detector (MSD) and an Agilent Technologies
1100 Series HPLC system (D) were employed,
respectively. The bioreactor Biostat Cplus from Sartorius
BBI Systems was used for experiment (D). The cells of P.
pastoris with expressed P450 3A4 were obtained from
Bisy GmbH (Hofstaetten, Austria). OD measurements
were executed with an Eppendorf BioPhotometer plus.
They had been cultivated, then stored as frozen pellets at -
(D) Preparative-scale biotransformation in a
bioreactor (Figures 5, 6, 7 and 8)
The obtained cells (ca. 200 g wet corresponding to 68 g
dry cell weight) were resuspended in 100 mM phosphate
buffer (pH 7.4) and filled up to 1.3 L to obtain an OD600 of
200. The cell broth was filled into the previously sterilised
bioreactor, stirred at 400 rpm and kept at pH 7.4 with 1M
solutions of potassium hydroxide and phosphorous acid.
9
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
The biotransformation was started with the addition of 13.3
mL of methanol and 3330 μL of a 1 M solution of
testosterone (959 mg) in DMSO. Samples were taken
regularly during the course of the reaction and
simultaneously analysed using HPLC for end point
determination. The biotransformation was stopped after 8
hours by centrifugation of the collected reaction broth and
washing the cells twice with phosphate buffer. After the
final centrifugation the cell pellet was suspended again in
1.3 L of phosphate buffer, filled into the same bioreactor
and the second cycle of biotransformation was carried out
analogously to the first one. Samples were taken with
increasing time intervals and the biotransformation was
stopped after 105 hours as before. Both aqueous reaction
broths were worked up individually by liquid-liquid
extraction washing the aqueous phase several times with
ethyl acetate (3x 150 mL). The resulting organic layer was
dried over MgSO4, concentrated by rotary evaporation and
loaded onto the column for chromatographic purification
(50% EtOAc in hexane). 6β-hydroxytestosterone (108 mg,
11.3% yield, 0.16% productivity[67] [g / g] using dry cell
weight for the catalyst) was isolated as a white crystalline
solid, and 6β-hydroxyandrostenedione (87 mg, 9.1%,
0.13%) as an off-white solid. Mixtures of other mono- or
dihydroxylated metabolites were also collected with
masses up to 10 mg. These mixtures were further purified
with the help of an Agilent Technologies 1100 Series
HPLC system adapted for manual preparative collection.
Separation was carried out via a reverse-phase Purospher
Star RP-18e (5.0 μm; 250 x 4 mm) column at 35 °C and a
flow rate of 1 mL/min, and water containing 0.1% acetic
acid (A) and acetonitrile (B) were used as the eluents in a
steadily increasing gradient. Then methanol (C) was mixed
in to wash the column thoroughly. Mono-hydroxylated
metabolites were purified using the following method with
the ratios: 0 min: A/B 80/20; 31 min: A/B 75/25; 31.01
min: B/C: 60/40; 34.00 min: B/C 60/40; 34.01 min: A/B
80/20; 36 min: A/B 80/20. Di-hydroxylated metabolites
were purified using the method with the ratios: 0 min: A/B
85/15; 33.50 min: A/B 80/20; 33.51 min: B/C: 60/40;
36.00 min: B/C 60/40; 36.01 min: A/B 85/15; 38 min: A/B
85/15. The aqueous HPLC solvents were removed under a
stream of nitrogen. The compounds 1β- (5) and 15β-
200.5 (C-3), 168.3 (C-5), 126.4 (C-4), 72.6 (C-6), 53.7 (C-
9), 50.9 (C-14), 47.7 (C-13), 38.1 (C-10), 37.3 (C-7), 37.1
(C-1), 35.8 (C-12), 34.2 (C-2), 31.3 (C-16), 29.5 (C-8),
21.8 (C-15), 20.3 (C-11), 19.6 (C-19), 13.8 (C-18)
1β-hydroxytestosterone (5, C19H28O3, white crystalline
1
solid, 1 mg, 0.1%): H NMR (500 MHz, CDCl3): δ = 5.79
(1H, s, 4-H), 4.04 (1H, dd, J = 7.6, 7.0 Hz, 1α-H), 3.67 –
3.62 (1H, m, 17-H), 2.54 (2H, d, J = 7.8 Hz, 2α-, 2β-H),
2.49 (1H, dddd, J = 14.8, 13.8, 5.3, 1.3 Hz, 6β-H), 2.33
(1H, dddd, 14.2, 4.4, 2.6, 0.4 Hz, 6α-H), 2.11-2.03 (1H, m,
11α-H), 2.02-1.97 (1H, m, 16α-H), 1.91 – 1.85 (2H, m, 7β-,
12β-H), 1.69 – 1.57 (3H, m, 8-, 11β-, 15α-H), 1.49 – 1.41
(2H, m, 16β-H, 1β-OH), 1.33 – 1.29 (1H, m, 15β-H), 1.25
(3H, s, 19-H), 1.16 – 1.08 (2H, m, 9-, 12α-H), 1.06 – 0.94
(2H, m, 7α-, 14-H), 0.80 (3H, s, 18-H).
In the COSY spectrum the carbinol proton (δ 4.04 ppm) of
interest was found to couple with protons in the region of
2.54 ppm corresponding to those of positions 2 or 6, and
with a proton of the doublet at 1.46 ppm, which should
correspond to the newly introduced hydroxyl group. No
coupling to 8-H at δ 1.69 ppm was observed as previously
described by Guengerich et al.[80,81] Additionally, the
HSQC spectrum revealed the carbinol proton (δ 4.02 ppm)
to be attached to the carbon at 74.0 (C-1) ppm. With the
carbon shifts in hand, the HMBC spectrum confirmed the
hydroxylation at C-1 due to the coupling of the 18-Hs to C-
10, -9, -5 and the carbinol carbon -1. The NOESY
spectrum indicated interactions between the carbinol
proton and protons at 2.53 (2α- and 2β-H), 1.69 ppm (11β-
H), 1.47 (1β-OH) and 1.11 (9-H). The correlation with
11β-H and the lack of it with the H-19 protons proved the
hydroxylation having occurred at 1β.
15β-hydroxytestosterone (6, C19H28O3, white solid, 1 mg,
0.1%): 1H NMR (500 MHz, CDCl3): δ = 5.75 (1H, s, 4-H),
4.24 – 4.20 (1H, m, 15α-H), 3.59 (1H, dd, J = 14.0, 8.7 Hz,
17α-H), 2.66 – 2.60 (1H, m, 16α-H), 2.47 (1H, ddd, J =
15.6, 15.3, 5.0 Hz, 6β-H), 2.43 (1H, dd, J = 14.1, 5.0 Hz,
2β-H), 2.37 – 2.28 (2H, m, 2α-, 6α-H), 2.13 – 2.08 (1H, m,
7β-H), 2.07 – 2.03 (1H, m, 1β-H), 2.00 (1H, ddd, J = 11.4,
10.8, 2.95, 8-H), 1.87 – 1.83 (1H, m, 12β-H), 1.72 (1H,
ddd, J = 14.2, 14.1, 4.3 Hz, 1α-H), 1.63 – 1.57 (2H, m,
11α-, 16β-H), 1.47 (1H, ddd, J = 13.2, 13.1, 3.9 Hz, 11β-
H), 1.35 (1H, d, 15β-OH), 1.24 (3H, s, 19-H), 1.16 – 1.04
(2H, m, 7α-, 12α-H), 1.07 (3H, s, 18-H), 1.00 (1H, ddd, J =
12.3, 11.3, 4.0, Hz, 9-H), 0.85 (1H, dd, J = 11.3, 5.6 Hz,
14-H). 13C NMR (75 MHz, CDCl3): δ = 199.8 (C-3), 171.3
(C-5), 124.0 (C-4), 81.2 (C-17), 69.2 (C-15), 55.3 (C-14),
54.4 (C-9), 42.4 (C-13), 38.9 (C-10), 38.0 (C-12), 35.9 (C-
1), 34.0 (C-2), 32.8 (C-6), 31.6 (C-8), 31.2 (C-7), 20.7 (C-
11), 17.5 (C-19), 13.9 (C-18).
hydroxytestosterone
(6),
6-dehydro-15β-
hydroxytestosterone (7) and, 6β,16β-dihydroxytestosterone
(8), 6β,17β-dihydroxy-4-androstene-3,16-dione (9) and
6β,12β-dihydroxyandrostenedione (10) could be identified.
Compounds 5, 6, 7 and 8 were obtained as white/off-white
solids, the appearance of 9 and 10 was hardly definable
because of too little quantities. For the same reason other
isolated compounds, which were also present as mixtures,
could not be elucidated.
6β-hydroxytestosterone (2, C19H28O3, white crystalline
1
solid, 108 mg, 11%): H NMR (300 MHz, CDCl3): δ =
5.80 (1H, s, 4-H), 4.34 (1H, m, 6α-H), 3.65 (1H, dd, J =
10.3, 8.1 Hz, 17α-H), 2.50 (1H, ddd, J = 14.9, 4.5, 2.0 Hz,
2β-H), 2.40 (1H, ddd, J = 15.6, 4.3, 2.2, 2α-H), 2.02 (4H,
m, 16α-H, 1α-H, 7β-H, 8-H), 1.88 (1H, ddd, J = 12.2, 4.1,
3.0 Hz, 12β-H), 1.70 (1H, dd, J = 14.2, 4.2 Hz, 1β-H), 1.61
6-dehydro-15β-hydroxytestosterone (7, C19H26O3, white
1
solid, 3 mg, 0.3%): H NMR (500 MHz, CDCl3): δ = 6.30
(1H, d, J = 9.86 Hz, 7-H), 6.16 (1H, dd, J = 9.86, 2.72 Hz,
6-H), 5.69 (1H, s, 3-H), 4.41 – 4.37 (1H, m, 15α-H), 3.65 –
3.60 (1H, m, 17-H), 2.70 – 2.66 (1H, m, 16α-H), 2.56 (1H,
(1H, m, 15α-H), 1.57 (1H, m, 11α-H), 1.49 (1H, m, 11β-H), dd, J = 14.23, 5.41 Hz, 2β-H), 2.51 (1H, dd, J = 5.34, 1.53
1.45 (1H, m, 16β-H), 1.40 (1H, m, 15β-H), 1.38 (3H, s, 19-
CH3), 1.21 (1H, m, 7α-H), 1.09 (1H, dd, J = 12.8, 4.3 Hz,
Hz, 2α-H), 1.17 (3H, s, 19-C), 1.12 (3H, s, 18-H). HRMS
(TOF-EI+) m/z: calcd. for C19H28O4 302.1882, found
12α-H), 0.98 (1H, m, 14-H), 0.90 (1H, m, 9-H), 0.81 (3H, s, 302.1864.
18-CH3). 13C NMR (75 MHz, CDCl3): δ = 200.5 (C-3),
168.5 (C-5), 126.5 (C-4), 81.8 (C-17), 73.1 (C-6), 53.9 (C-
This compound was isolated as the minor component in a
9), 50.6 (C-14), 43.1 (C-13), 41.1 (C-10), 38.2 (C-7), 37.3
(C-1), 36.6 (C-12), 34.4 (C-2), 30.6 (C-16), 29.9 (C-8),
23.4 (C-15), 20.7 (C-11), 19.7 (C-19), 11.2 (C-18).
mixture with 15β-hydroxytestosterone as reflected by
HPLC, NMR and HRMS analysis. The peaks at 6.30 and
6.16 ppm implied the presence of another alkene group and
the roof effect observed between two shifts indicates a
6β-hydroxyandrostenedione (4, C19H26O3, off-white solid, strong second-order coupling effect. This was supported by
87 mg, 9%): 1H NMR (300 MHz, CDCl3): δ = 5.77 (1H, s,
4-H), 4.35 (1H, s, 6α-H), 2.49-2.36 (3H, m, 2α-H, 2β-H,
16β-H), 2.16-1.96 (5H, m, 1β-H, 7β-H, 8-H, 15α-H, 16α-
H), 1.84 (1H, ddd, J = 12.4, 3.7, 2.7 Hz, 12β-H), 1.71-1.59
(3H, m, 1α-H, 11α-H, 15β-H), 1.51 (2H, ddd, J = 14.04,
13.3, 3.5 Hz, 11β-H), 1.36 (3H, s, 19-CH3), 1.29-1.19 (3H,
m, 7α-H, 12α-H, 14-H), 1.01-0.95 (1H, m, 9-H), 0.90 (3H,
s, 18-CH3). 13C NMR (75 MHz, CDCl3): δ = 220.1 (C-17),
their strong correlation in the NOESY spectrum. In the
HMBC spectrum the 18-Hs at 1.13 ppm coupled with
carbons at 80.9 (C-17), 52.8 (C-14), 43.8 (C-13) and 37.7
(C-12) ppm, and the 19-Hs at 1.19 ppm with carbons at
163.4 (C-5), 51.1 (C-9), 36.8 (C-10) and 32.6 (C-1) ppm
ruling out the known α,β-unsaturated compound 1-
dehydrotestosterone. The interaction of the proton at 6.16
ppm with the H-4 (5.69 ppm) in the NOESY spectrum
10
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
therefore revealed the alkene to be between positions 6 and
7. Here, a correlation between the other alkene proton
(6.30 ppm) and the carbinol proton (4.39 ppm) also
strongly pointed towards the hydroxylation to have
occurred at position 15, on the β-face of the steroid due to
the lack of a coupling to 18-H. Indeed, the carbinol proton
(4.40 ppm) was attached to a carbon with a shift of 69.0
ppm in the HSQC spectrum, comparing well with the shifts
of position 15 of 15β-hydroxytestosterone. Additionally,
the carbinol proton (4.40 ppm) coupled to a proton at 2.69
ppm in the COSY spectrum, which in turn was found to
correlate with 17-H (3.62 ppm), fitting well to 15α-H and
16α-H, respectively.
however, quite likely this is 17β-OH as it appears as a high
1
doublet signal in the H spectrum integrating to only 0.6
protons. In contrast to the first carbinol proton H-6α of this
compound or 17α-H of any other testosterone derivative
determined here, no evidence for any other interaction of
17α-H was noticeable in the COSY spectrum. The same
was observed in the NOESY spectrum, where 17α-H only
coupled with two protons at 2.09 (12α-H) and 1.54 (14-H).
The lack of coupling to neighbouring protons at position
16 strongly pointed towards their absence. In addition, C-
17 (85.5 ppm) must have experienced a deshielding effect
compared to its usual shift of about 81 ppm supporting the
presence of a vicinal ketone functional group at position 16.
6β,16β-dihydroxytestosterone (8, C19H28O4, off-white
6β,12β-dihydroxyandrostenedione (10, C19H26O4, <1 mg,
1
1
solid, 1 mg, 0.1%): H NMR (500 MHz, CDCl3): δ = 5.82
<0.1%): H NMR (500 MHz, CDCl3): δ = 5.85 (1H, s, 4-
(1H, s, 4-H), 4.36 (1H, br, 6α-H), 4.22 – 4.17 (1H, m, 16α-
H), 4.42 – 4.40 (1H, m, 6α-H), 3.80 (1H, dd, J = 11.5, 4.6
Hz, 12α-H), 3.50 – 3.48 (1H, m, 17-H), 1.63 (1H, br, 6β-
H), 3.40 (1H, dd, J = 8.86, 7.98 Hz, 17-H), 1.40 (3H, s, 19-
H), 0.94 (1H, ddd, J = 11.8, 11.2, 3.9 Hz, 9-H), 0.90 (3H, s, OH), 1.42 (3H, s, 19-H), 1.08 (1H, ddd, J = 12.5, 11.5, 3.9
18-H), 0.83 (1H, ddd, J = 13.2, 10.7, 7.1, 14-H). HRMS
(TOF-EI+) m/z: calcd. for C19H28O4 320.1988, found
320.1975.
Hz, 9-H), 1.03 (3H, s, 18-H). HRMS (TOF-EI+) m/z: calcd.
for C19H26O4 318.1831, found 318.1828.
In the proton spectrum two carbinol protons were found at
4.41 and 3.80 ppm, which are attached to carbons at 72.7
and 72.4 ppm, respectively. For this compound the HMBC
spectrum was very informative. The coupling of the 18-Hs
to a carbon at 222 ppm indicated that this compound had to
be an androstenedione derivative with a ketone functional
group at C-17. In addition to this interaction, the 18-Hs
(1.05 ppm) couple with three other carbons at 48.7, 51.5
and 72.3 ppm corresponding to C-13, -14, and -12,
respectively. Therefore, one of the hydroxylations occurred
at C-12. The NOESY spectrum revealed that this carbinol
proton at C-12 should be in the axial position because a
correlations to 9-H (1.08 ppm), 14-H (1.27 ppm) and 11α-
H (1.81 ppm) was visible. The other hydroxylation pattern
(4.41, 72.7 ppm) compared well with the shifts of 6β-
hydroxyandrostenedione. In the NOESY spectrum a strong
interaction to the 4-H proton (5.85 ppm) was the most
convincing. Two interactions with other protons at 2.15
and 1.29 ppm (7α-H and 7β-H), as well as one to a proton
at 1.63 ppm in both the NOESY and COSY spectra left no
room for doubt. The latter shift was assigned to the 6β-OH
as it is a broad singlet in the proton spectrum with no other
interactions in the COSY, just as for 6β,16β-
The first carbinol proton at 4.36 ppm was attached to a
carbon at 73.0 ppm as determined by HSQC. Therefore it
did not only have the same proton and carbon shifts as the
6α-H of 2, but was also found to couple with protons of
peaks at 2.04 (7β-H) and 1.23 (7α-H) ppm in the COSY
spectrum. A slight coupling to a proton at 1.58 ppm was
also visible in the COSY spectrum, which is likely the 6β-
OH as it showed no other coupling and appeared as a broad
singlet in the proton spectrum. In addition, the first
carbinol proton (4.36 ppm) showed a clear correlation to 4-
H in the HMBC and NOESY spectrum. The latter
spectrum also revealed hydroxylation to have occurred at
the β-face of testosterone by the lack of any correlation
with the protons at position 19. The second carbinol proton
at 4.19 ppm was attached to a carbon at 69.9 ppm as
determined by HSQC. A strong correlation with 17-H in
the COSY spectrum clearly indicated the other
hydroxylation to have occurred at C-16. The presence of
the same interaction in the NOESY spectrum in
combination with the lack of a correlation with the C-18
methyl group suggested the hydroxyl group being in the
equatorial position.
dihydroxytestosterone
and
6β,17β-dihydroxy-4-
androstene-3,16-dione.
6β,17β-dihydroxy-4-androstene-3,16-dione (9, C19H26O4,
<1 mg, <0.1%): 5.85 (1H, s, 4-H), 4.40 (1H, br, 6α-H),
3.78 (1H, br, 17α-H), 2.62 (1H, d, J = 3.22 Hz, 17β-OH),
2.58 – 2.51 (1H, m, 2β-H), 2.45 – 2.36 (2H, m, 2α-, 15β-
H), 1.62 (1H, br, 6β-OH), 1.42 (3H, s, 19-H), 1.15 – 1.10
(1H, m, 9-H), 0.81 (3H, s, 18-H)
Acknowledgements
This project has received funding from the European Union‘s
Horizon 2020 research and innovation programme, OXYtrain
MSCA-ITN, under grant agreement No 722390. The authors
would like to thank Prof. Breinbauer for access to his chemistry
facilities and material. The authors are also grateful to Dr. G.
Strohmeier and Prof. M. Winkler for their support with the HPLC
instruments.
The interaction between the first carbinol proton at 4.39
ppm and H-4 (5.96 ppm) in the NOESY spectrum pointed
towards the hydroxylation to have occurred at position 6.
In the HSQC spectrum the carbinol proton is attached to a
carbon with a shift of 72.3 ppm fitting the usual 6α-proton
shifts. It was also found to couple with protons at 1.98 (7α-
H) and 1.36 (7β-H) ppm as well as to one proton at 1.62
ppm likely to be the 6β-OH in the COSY spectrum. The
latter coupling is visible with the same shift and peak
shape as in the proton spectra of 6β,16β-
Author contributions
dihydroxytestosterone
and
6β,12β-
N.D.F., M.S. and A.G. devised the study concept.
N.D.F. conducted all experiments. M.S. and C.S.
provided material. N.D.F. and M.S. developed
methods. N.D.F. and H.W. performed data
acquisition and analysis. C.S., D.S., U.S and A.G.
jointly supervised. N.D.F. wrote the original
manuscript. N.D.F. and A.G. reviewed and edited the
manuscript. A.G. administrated the project. D.S., U.S.
and A.G. acquired the funding. All authors critically
reviewed and approved the manuscript.
dihydroxyandrostenedione. In the HMBC spectrum a
proton at 1.95 ppm was found to interact with a carbon at
215.0 ppm indicating the presence of a second ketone like
in derivatives of androstenedione. However, the 18-Hs
showed coupling to C-17 at 85.5 ppm as well as to other
carbons in a range from 35 to 45 ppm. Consequently, the
ketone had to be at a different position than 17. The 19-Hs
coupled to C-5 at 158.7 ppm and to other carbons in a
range from 30 to 53 ppm. The shifts of the A-ring in
general were the same as for 6β-hydroxytestosterone. The
peak shape of 9-H is quite distinct and could be identified
easily in the proton spectrum. The proton interacted with
carbons below 60 ppm. Therefore only positions 15 and 16
qualified having the ketone functional group attached. In
the COSY 17α-H correlated with a proton at 2.58 ppm,
11
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
References
[20]
D. Schmitz, S. Janocha, F. M. Kiss, R. Bernhardt,
Biochim. Biophys. Acta, Proteins Proteomics 2018,
1866, 11–22.
[1]
[2]
[3]
H. van de Waterbeemd, E. Gifford, Nat. Rev. Drug
Discov. 2003, 2, 192–204.
[21] X. Ren, J. A. Yorke, E. Taylor, T. Zhang, W. Zhou,
L. L. Wong, Chem. - Eur. J. 2015, 21, 15039–
15047.
S. Piantadosi, Clinical Trials: A Methodologic
Perspective, Wiley & Sons Inc., Hoboken, 2017.
FDA, Guidance for Industry: Safety Testing of
Drug Metabolites, U.S. Department Of Health And
Human Services, Food And Drug Administration,
Center For Drug Evaluation And Research
(CDER), 2016.
[22]
G. Di Nardo, G. Gilardi, Int. J. Mol. Sci. 2012, 13,
15901–15924.
[23]
M. Poraj-Kobielska, M. Kinne, R. Ullrich, K.
Scheibner, G. Kayser, K. E. Hammel, M.
Hofrichter, Biochem. Pharmacol. 2011, 82, 789–
796.
[4]
C. Martinez, S. Rupashinghe, Curr. Top. Med.
Chem. 2013, 13, 1470–1490.
[24]
[25]
C. D. Murphy, Biotechnol. Lett. 2015, 37, 19–28.
[5]
[6]
D. S. Wishart, Drugs R D 2008, 9, 307–322.
R. Weis, M. Winkler, M. Schittmayer, S.
Kambourakis, M. Vink, J. David Rozzell, A.
Glieder, Adv. Synth. Catal. 2009, 351, 2140–2146.
L. F. Shyur, N. S. Yang, Curr. Opin. Chem. Biol.
2008, 12, 66–71.
[7]
O. Fiehn, D. Robertson, J. Griffin, M. van der
Werf, B. Nikolau, N. Morrison, L. W. Sumner, R.
Goodacre, N. W. Hardy, C. Taylor, J. Fostel, B.
Kristal, R. Kaddurah-Daouk, P. Mendes, B. van
Ommen, J. C. Lindon, S. A. Sansone,
[26]
[27]
O. Ghisalba, M. Kittelmann, in: Mod. Biooxidation
Enzym. React. Appl. (Eds.: R. D. Schmid, V.
Urlacher), Wiley-VCH, Weinheim, 2007, pp. 211–
232.
T. Testa, in: Pract. Med. Chem. (Ed.: C. G.
Wermuth), Elsevier Ltd, Amsterdam, 2008, pp.
655–673.
Metabolomics 2007, 3, 175–178.
[8]
L. Di, E. H. Kerns, Curr. Opin. Chem. Biol. 2003,
7, 402–408.
[28]
[29]
S. Rendic, F. P. Guengerich, Chem. Res. Toxicol.
2015, 28, 38–42.
[9]
S. Asha, M. Vidyavathi, Appl. Biochem.
Biotechnol. 2010, 160, 1699–1722.
M. Miyazaki, K. Nakamura, Y. Fujita, F. P.
Guengerich, R. Horiuchi, K. Yamamoto, Drug
Metab. Dispos. 2008, 36, 2287–2291.
[10]
[11]
S. R. Kazmi, R. Jun, M. S. Yu, C. Jung, D. Na,
Comput. Biol. Med. 2019, 106, 54–64.
J. Kirchmair, A. H. Göller, D. Lang, J. Kunze, B.
Testa, I. D. Wilson, R. C. Glen, G. Schneider, Nat.
Rev. Drug Discov. 2015, 14, 387–404.
[30]
[31]
[32]
P. Anzenbacher, E. Anzenbacherová, Cell. Mol.
Life Sci. 2001, 58, 737–747.
F. P. Guengerich, Chem. Res. Toxicol. 2017, 30, 2–
12.
[12]
[13]
[14]
E. W. Esch, A. Bahinski, D. Huh, Nat. Rev. Drug
Discov. 2015, 14, 248–260.
J. A. Krauser, M. Voehler, L.-H. Tseng, A.-B.
Schefer, M. Godejohann, F. P. Guengerich, Eur. J.
Biochem. 2004, 271, 3962–3969.
N. D. Fessner, ChemCatChem 2019, 11, 2226–
2242.
M. Winkler, M. Geier, S. P. Hanlon, B. Nidetzky,
A. Glieder, Angew. Chemie Int. Ed. 2018, 57,
13406–13423.
[33]
[34]
J. A. Krauser, F. P. Guengerich, J. Biol. Chem.
2005, 280, 19496–19506.
H. C. Man, P. L. Skipper, J. S. Wishnok, S. R.
Tannenbaum, Drug Metab. Dispos. 2005, 33, 714–
718.
[15]
[16]
[17]
K. Schroer, M. Kittelmann, S. Lütz, Biotechnol.
Bioeng. 2010, 106, 699–706.
R. B. Vail, M. J. Homann, I. Hanna, A. Zaks, J.
Ind. Microbiol. Biotechnol. 2005, 32, 67–74.
[35]
[36]
E. Pfeiffer, M. Metzler, Arch. Toxicol. 2004, 78,
369–377.
J. M. Caswell, M. O’Neill, S. J. C. Taylor, T. S.
Moody, Curr. Opin. Chem. Biol. 2013, 17, 271–
275.
X. Cao, S. T. Gibbs, L. Fang, H. A. Miller, C. P.
Landowski, H. C. Shin, H. Lennernas, Y. Zhong, G.
L. Amidon, L. X. Yu, D. Sun, Pharm. Res. 2006,
23, 1675–1686.
[18]
[19]
H. M. Girvan, A. W. Munro, Curr. Opin. Chem.
Biol. 2016, 31, 136–145.
[37]
[38]
R. Zuber, E. Anzenbacherová, P. Anzenbacher, J.
Cell. Mol. Med. 2002, 6, 189–198.
A. Worsch, F. K. Eggimann, M. Girhard, C. J. von
Bühler, F. Tieves, R. Czaja, A. Vogel, C. Grumaz,
K. Sohn, S. Lütz, M. Kittelmann, V. B. Urlacher,
Biotechnol. Bioeng. 2018, 115, 2156–2166.
M. Geier, A. Braun, P. Fladischer, P. Stepniak, F.
Rudroff, C. Hametner, M. D. Mihovilovic, A.
Glieder, FEBS J. 2013, 280, 3094–3108.
12
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
[39]
[40]
M. Paloncꢁovꢂ, V. Navrꢂtilovꢂ, K. Berka, A. Laio,
M. Otyepka, J Chem Theory Comput 2016, 12,
2101–2109.
[60]
E. O’Reilly, V. Köhler, S. L. Flitsch, N. J. Turner,
Chem. Commun. 2011, 47, 2490–2501.
[61] W. Zhang, M. A. Bevins, B. A. Plantz, L. A. Smith,
M. Ekroos, T. Sjögren, Proc. Natl. Acad. Sci. U. S.
A. 2006, 103, 13682–13687.
M. M. Meagher, Biotechnol. Bioeng. 2000, 70, 1–8.
[62]
[63]
[64]
[65]
[66]
[67]
[68]
A. Weninger, A. M. Hatzl, C. Schmid, T. Vogl, A.
Glieder, J. Biotechnol. 2016, 235, 139–149.
[41] J. C. Hackett, J. Biol. Chem. 2018, 293, 4037–4046.
[42]
Y. Kapelyukh, M. J. I. Paine, J. D. Maréchal, M. J.
Sutcliffe, C. R. Wolf, G. C. K. Roberts, Drug
Metab. Dispos. 2008, 36, 2136–2144.
M. T. Lundemo, J. M. Woodley, Appl. Microbiol.
Biotechnol. 2015, 99, 2465–2483.
W. A. Duetz, J. B. Van Beilen, B. Witholt, Curr.
Opin. Biotechnol. 2001, 12, 419–425.
[43]
I. G. Denisov, P. J. Mak, Y. V. Grinkova, D.
Bastien, G. Bérubé, S. G. Sligar, J. R. Kincaid, J.
Inorg. Biochem. 2016, 158, 77–85.
R. A. Sheldon, D. Brady, Chem. Commun. 2018,
54, 6088–6104.
[44]
[45]
[46]
[47]
[48]
L. C. Wienkers, T. G. Heath, Nat. Rev. Drug
Discov. 2005, 4, 825–833.
V. B. Urlacher, M. Girhard, Trends Biotechnol.
2019, 37, 882–897.
M. Geier, A. Braun, A. Emmerstorfer, H. Pichler,
A. Glieder, Biotechnol. J. 2012, 7, 1346–1358.
R. Kratzer, J. M. Woodley, B. Nidetzky,
Biotechnol. Adv. 2015, 33, 1641–1652.
G. P. L. Cereghino, J. L. Cereghino, C. Ilgen, J. M.
Cregg, Curr. Opin. Biotechnol. 2002, 13, 329–332.
C. A. Drǎgan, F. T. Peters, P. Bour, A. E.
Schwaninger, S. M. Schaan, I. Neunzig, M.
Widjaja, J. Zapp, T. Kraemer, H. H. Maurer, M.
Bureik, Appl. Biochem. Biotechnol. 2011, 163,
965–980.
V. Puxbaum, D. Mattanovich, B. Gasser, Appl.
Microbiol. Biotechnol. 2015, 99, 2925–2938.
A. Emmerstorfer, T. Wriessnegger, M. Hirz, H.
Pichler, Appl. Microbiol. Biotechnol. 2014, 98,
7671–7698.
[69]
[70]
Y. Kanamori, K. Fujita, K. Nakayama, T.
Kamataki, H. Kawai, Drug Metab. Pharmacokinet.
2003, 18, 42–47.
[49]
B. Byrne, Curr. Opin. Struct. Biol. 2015, 32, 9–17.
M. Subramanian, S. Das, E. Martinez, D. .
Callewaert, in Rapid Identif. Prod. Metab. Using
CypExpressTM 2D6, 3A4, 2C9, 19th Annual
Meeting Of The International Society For The
Study Of Xenobiotics, San Francisco, 2014.
[50] V. Juturu, J. C. Wu, ChemBioChem 2018, 19, 7–21.
[51]
M. Ahmad, M. Hirz, H. Pichler, H. Schwab, Appl.
Microbiol. Biotechnol. 2014, 98, 5301–5317.
[52] M. Geier, C. Schmid, A. Glieder, Chim. Oggi 2013,
31, 24–27.
[71]
D. N. Kirk, H. C. Toms, C. Douglas, K. A. White,
K. E. Smith, S. Latif, R. W. P. Hubbard, J. Chem.
Soc. Perkin Trans 2 1990, 2, 1567–1594.
[53]
H. L. J. Makin, J. W. Honour, C. H. L. Shackleton,
W. J. Griffith, in: Steroid Anal. (Eds.: H. L. J.
Makin, D. B. Gower), Springer Science &
[72]
[73]
[74]
[75]
I. Černꢁ, J. Fajkoš, V. Pouzar, Steroids 1996, 61,
58–64.
Business Media, Dordrecht, 2010, pp. 163–784.
[54]
T. Pajič, M. Vitas, D. Žigon, A. Pavko, S. L. Kelly,
R. Komel, Yeast 1999, 15, 639–645.
J. J. Sheets, R. W. Estabrook, Biochemistry 1985,
24, 6591–6597.
[55] M. Shao, X. Zhang, Z. Rao, M. Xu, T. Yang, H. Li,
Z. Xu, S. Yang, Green Chem. 2016, 18, 1774–
1784.
B. P. Lisboa, J.-A. Gustafsson, J. Sjövall, Eur. J.
Biochem. 1968, 4, 496–505.
Y. Gumulya, J.-M. Baek, S.-J. Wun, R. E. S.
Thomson, K. L. Harris, D. J. B. Hunter, J. B. Y. H.
Behrendorff, J. Kulig, S. Zheng, X. Wu, J. E. Stok,
J. J. De Voss, G. Schenk, U. Jurva, S. Andersson,
E. M. Isin, M. Bodén, L. Guddat, E. M. J. Gillam,
Nat. Catal. 2018, 1, 878–888.
[56]
R. A. Canuto, Dehydrogenases, InTech, London,
2012.
[57]
H. Zhang, Q. Li, L. Wang, Y. Chen, Biochim.
Biophys. Acta, Gen. Subj. 2018, 1862, 1199–1208.
[58] F. W. Krainer, C. Dietzsch, T. Hajek, C. Herwig, O.
Spadiut, A. Glieder, Microb. Cell Fact. 2012, 22,
1–14.
[76]
[77]
O. J. Pozo, J. Marcos, R. Ventura, A. Fabregat, J.
Segura, Anal. Bioanal. Chem. 2010, 398, 1759–
1770.
[59]
T. Vogl, L. Sturmberger, T. Kickenweiz, R.
Wasmayer, C. Schmid, A. M. Hatzl, M. A.
Gerstmann, J. Pitzer, M. Wagner, G. G. Thallinger,
et al., ACS Synth. Biol. 2016, 5, 172–186.
C. Jo Corbin, S. M. Mapes, J. Marcos, C. H.
Shackleton, D. Morrow, S. Safe, T. Wise, J. Joe
Ford, A. J. Conley, Endocrinology 2004, 145,
2157–2164.
13
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
[78]
D. J. B. Hunter, J. B. Y. H. Behrendorff, W. A.
[87]
[88]
Y. Farooq, G. C. K. Roberts, Biochem. J. 2010,
432, 485–493.
Johnston, P. Y. Hayes, W. Huang, B. Bonn, M. A.
Hayes, J. J. De Voss, E. M. J. Gillam, Metab. Eng.
2011, 13, 682–693.
V. R. Dodhia, C. Sassone, A. Fantuzzi, G. Di
Nardo, S. J. Sadeghi, G. Gilardi, Electrochem.
Commun. 2008, 10, 1744–1747.
[79]
[80]
S. Kille, F. E. Zilly, J. P. Acevedo, M. T. Reetz,
Nat. Chem. 2011, 3, 738–743.
[89]
[90]
[91]
A. Braun, M. Geier, B. Bühler, A. Schmid, S.
Mauersberger, A. Glieder, Microb. Cell Fact. 2012,
11, 106.
V. Looser, B. Bruhlmann, F. Bumbak, C. Stenger,
M. Costa, A. Camattari, D. Fotiadis, K. Kovar,
Biotechnol. Adv. 2015, 33, 1177–1193.
B. Palabiyik, S. Karaer, N. Arda, S. Erturk Toker,
G. Temizkan, S. Kelly, A. Topal Sarikaya,
Biologia (Bratisl). 2008, 63, 450–454.
[81]
Z. Yang, Z. Zhang, Biotechnol. Adv. 2018, 36,
182–195.
[82] A. Schmideder, S. Hensler, M. Lang, A. Stratmann,
U. Giesecke, D. Weuster-Botz, Process Biochem.
2016, 51, 177–184.
H. Venkataraman, S. B. A. de Beer, L. A. H. van
Bergen, N. van Essen, D. P. Geerke, N. P. E.
Vermeulen, J. N. M. Commandeur, ChemBioChem
2012, 13, 520–523.
[83]
C. W. Theron, J. Berrios, F. Delvigne, P. Fickers,
Appl. Microbiol. Biotechnol. 2018, 102, 63–80.
[92]
[93]
D. Zehentgruber, F. Hannemann, S. Bleif, R.
Bernhardt, S. Lütz, ChemBioChem 2010, 11, 713–
721.
[84]
BioVision, “EZCyp,” can be found under
https://www.biovision.com/ezcyptm-active-
human-cytochrome-p450-3a4.html, 2020.
Q. Cheng, C. D. Sohl, F. K. Yoshimoto, F. P.
Guengerich, J. Biol. Chem. 2012, 287, 29554–
29567.
[85]
[86]
S. Aldrich, “CYPExpress,” can be found under
https://www.sigmaaldrich.com/catalog/product/sig
ma/mtoxce3a4?lang=de&region=AT, 2020.
Bisy, “Biocatalysts,” can be found under
https://www.bisy.at, 2020.
14
This article is protected by copyright. All rights reserved.

10.1002/adsc.202000251
Advanced Synthesis & Catalysis
FULL PAPER
Preparative-Scale Production of Testosterone
Metabolites by Human Liver Cytochrome P450
Enzyme 3A4
OH
OH
Pichia
pastoris
H
H
CyPto4ch5ro0me
H
H
H
H
O
O
Adv. Synth. Catal. Year, Volume, Page – Page
= mono-hydroxylation products
= di-hydroxylation products
= oxidation
4x
4x
Metabolites
Testosterone
yields =
<0.1 - 11%
Nico D. Fessner, Matic Srdič, Hansjörg Weber,
Christian Schmid, David Schönauer, Ulrich
Schwaneberg, and Anton Glieder*
15
This article is protected by copyright. All rights reserved.