Role of human 3α-hydroxysteroid dehydrogenase isoforms (AKR1C1-AKR1C3) in the extrahepatic metabolism of the steroidal aromatase inactivator Formestane

Article abstract of DOI:10.1016/j.jsbmb.2019.105527

The clinical use of the steroidal aromatase inhibitor Formestane (4-hydroxandrostenedione, 4-OHA) in the treatment of advanced ER+ breast cancer has been discontinued, and therefore, interest in this remarkable drug has vanished. As a C-19 sterol, 4-OHA can undergo extensive intracellular metabolism depending on the expression of specific enzymes in the corresponding cells. We used the metabolites 4β-hydroxyandrosterone, 4β-hydroxyepiandrosterone and its 17β-reduced derivative as standards for the proof of catalytic activity present in the cell culture medium and expressed by the isolated enzymes. All of the aldo-keto reductases AKR1C1, AKR1C2, AKR1C3 and AKR1C4 catalysed the reduction of the 3-keto-group and the Δ^4,5 double bond of 4-OHA at the same time. Molecular docking experiments using microscale thermophoresis and the examination of the kinetic behaviour of the isolated enzymes with the substrate 4-OHA proved that AKR1C3 had the highest affinity for the substrate, whereas AKR1C1 was the most efficient enzyme. Both enzymes (AKR1C1and AKR1C3) are highly expressed in adipose tissue and lungs, exhibiting 3β-HSD activity. The possibility that 4-OHA generates biologically active derivatives such as the androgen 4-hydroxytestosterone or some 17β-hydroxy derivatives of the 5α-reduced metabolites may reawaken interest in Formestane, provided that a suitable method of administration can be developed, avoiding oral or intramuscular depot-injection administration.

Full text of DOI:10.1016/j.jsbmb.2019.105527

Journal Pre-proof
Role of human 3␣-hydroxysteroid dehydrogenase isoforms
(AKR1C1-AKR1C3) in the extrahepatic metabolism of the steroidal
aromatase inactivator Formestane
Runlan Wan, Xi Kong, Youzhe Yang, Siwen Tao, Youyou Chen,
Alexander Tobias Teichmann, Frank Heinrich Wieland
PII:
S0960-0760(19)30246-8
DOI:
https://doi.org/10.1016/j.jsbmb.2019.105527
SBMB 105527
Reference:
To appear in:
Journal of Steroid Biochemistry and Molecular Biology
Received Date:
Revised Date:
Accepted Date:
30 April 2019
31 October 2019
3 November 2019
Please cite this article as: Wan R, Kong X, Yang Y, Tao S, Chen Y, Teichmann AT, Wieland FH,
Role of human 3␣-hydroxysteroid dehydrogenase isoforms (AKR1C1-AKR1C3) in the
extrahepatic metabolism of the steroidal aromatase inactivator Formestane, Journal of Steroid
Biochemistry and Molecular Biology (2019), doi: https://doi.org/10.1016/j.jsbmb.2019.105527
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
© 2019 Published by Elsevier.

Role of human 3α-hydroxysteroid dehydrogenase isoforms (AKR1C1-AKR1C3) in the
extrahepatic metabolism of the steroidal aromatase inactivator Formestane
Runlan Wanabc, Xi Konga, YouzheYanga, Siwen Taod, Youyou Chene, Alexander Tobias Teichmann a*
Frank Heinrich Wielanda* FHwieland@gmx.de
,
a Sichuan Provincial Center for Gynaecology and Breast Diseases, The Affiliated Hospital of Southwest Medical University, Luzhou 646000, China
b
Nuclear Medicine and Molecular Imaging Key Laboratory of Sichuan Province, The Affiliated Hospital of Southwest Medical University, Luzhou
646000, China
c Key Laboratory of the Ministry of Education for Medical Electrophysiology, Southwest Medical University, Luzhou, 646000, China
d School of Pharmacy, Southwest Medical University, Luzhou 646000,China
e Department of Pharmacy, The Affiliated Hospital of Southwest Medical University, Luzhou 646000, China
Highlights


During its extrahepatic metabolism 4-OHA is reduced by aldo-keto reductases (AKR1C1-
AKR1C3) with 3α-, 3β-, and 17β-hydroxysteroid dehydrogenase activities.
The OH-group attached to C4 of steroids inhibits the reduction of the 4, 5 double bond by
SRD5A1 and SRD5A2. It is responsible for the ability of AKR1C isoforms to catalyse
conversion of 4-OHA to metabolites without a 4, 5 double bond.

Steroidal aromatase inactivators can be metabolized in the tumour and surrounding tissues,
leading to different steroids with different biological anti-tumour properties.
Abstract:
The clinical use of the steroidal aromatase inhibitor Formestane (4-hydroxandrostenedione, 4-OHA) in the
treatment of advanced ER+ breast cancer has been discontinued, and therefore, interest in this remarkable
drug has vanished. As a C-19 sterol, 4-OHA can undergo extensive intracellular metabolism depending on
the expression of specific enzymes in the corresponding cells. We used the metabolites 4-
hydroxyandrosterone, 4-hydroxyepiandrosterone and its 17-reduced derivative as standards for the proof
of catalytic activity present in the cell culture medium and expressed by the isolated enzymes. All of the
aldo-keto reductases AKR1C1, AKR1C2, AKR1C3 and AKR1C4 catalysed the reduction of the 3-keto-
group and the 4,5 double bond of 4-OHA at the same time. Molecular docking experiments using
microscale thermophoresis and the examination of the kinetic behaviour of the isolated enzymes with the
substrate 4-OHA proved that AKR1C3 had the highest affinity for the substrate, whereas AKR1C1 was the
most efficient enzyme. Both enzymes (AKR1C1and AKR1C3) are highly expressed in adipose tissue and
lungs, exhibiting 3-HSD activity. The possibility that 4-OHA generates biologically active derivatives
such as the androgen 4-hydroxytestosterone or some 17-hydroxy derivatives of the 5-reduced
metabolites may reawaken interest in Formestane, provided that a suitable method of administration can be
developed, avoiding oral or intramuscular depot-injection administration.
1

Keywords: Formestane, Aldo-keto reductase, 3α-HSD, 5α-reductase, Steroid metabolism, Breast cancer
1. Introduction
The steroidal aromatase inhibitor Formestane (4-hydroxandrostenedione, 4-OHA) often shows
antiproliferative effects on oestrogen receptor positive (ER+) breast cancers when oestrogen deprivation
caused by nonsteroidal aromatase inhibitors such as letrozole or anastrozole fails [1-2]. We have observed
that daily topical application of 4-OHA to a female ER+ breast cancer patient can cause a considerable
reduction of the size of the tumor within 4 to 6 weeks. This beneficial effect is independent of the
prevailing plasma level of oestradiol and is not due to local inhibition of aromatase alone [3].The
aromatase-independent mode of action most likely is based on biological effects of metabolites of 4-OHA.
Therefore, we wanted to characterize the cellular metabolism of 4-OHA and evaluate which enzyme is
primarily involved.
4-OHA is a 4,5 keto-steroid. The first and rate-limiting step in the metabolism of such androgens is
the reduction of the double bond, catalysed by either 5α- or 5β-reductase. The majority of metabolites
present in the urine after oral administration of 4-OHA are 5-reduced. Therefore, we wanted to determine
whether 4-OHA is a substrate for 5-reductase. Since the metabolism of 4-OHA in the male rhesus
monkey proceeds twice as fast as that of its counterpart 4-androstenedione (AD) [4], we anticipated that the
obstacle of 5-reductase is circumvented. To prove this, we expressed recombinant SRD5A1 and SRD5A2
(two distinct isoforms of 5-reductase) in COS-1 cells and examined the ability of these enzymes to
catalyse 5-reduction of Formestane. Bypassing 5-reductase could lead to the formation of metabolites of
4-OHA much faster and in many different tissues that do not express 5-reductase [5]. Our results showed
that 5-reductase did not contribute to the metabolism of 4-OHA.
Since four human aldo-keto reductases (AKR1C1-AKR1C4, AKR1Cs) are well known to play an
important role in the intracellular metabolism of steroids, often leading to new molecules exhibiting new
biological effects (intracrinology) [6], we used transiently expressed AKR1Cs as well as their isolated
counterparts in order to investigate the activities of the corresponding enzymes in the metabolic pathways
of 4-OHA and its metabolite 4-hydroxytestosterone (4-OHT), which binds with high affinity to the
androgen receptor[37]. In ER+ breast cancers, the metabolites of 4-OHA could have antiproliferative and
pro-apoptotic properties.
2. Materials and Methods
2.1 Chemicals and solvents
4-OHA was purchased from Thinker Chemical (Hangzhou, China). Dutasteride was bought from
Aladdin (Shanghai, China). 4-hydroxytestosterone (4-OHT) was supplied by Chiracon GmbH company
(06-10-1-19P2954), produced and purified as described previously [7]. AD, dihydrotestosterone (DHT),
androsterone (ADT) and 3β-androstanediol (3β-diol) were obtained from Yuan Ye (Shanghai, China). 5α-
androstanedione (5α-A) was bought from BioBioPha (Kunming, China). Epiandrosterone (EPI) was
purchased from Target Molecule Corp. Methyltestosterone (MT) was obtained from Meilunbio (Dalian,
China). Flufenamic acid, phenolphthalein, and N-methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA)
2

were purchased from Sigma. Testosterone (T) was obtained from Cambridge Isotope Laboratories, Inc.
NADPH-Na4 and NADP-Na2 were from BioFROXX. Other reagents and solvents such as t-butyl methyl
ether (MTBE) and dimethylsulfoxide (DMSO) were bought from Merck (Darmstadt, Germany). All other
reagents were of American Chemical Society grade or better.
2.1.1 Synthesis of 3β,4β-dihydroxy-5α-androstan-17-one (M2)
5α-androst-3-en-17-one (675 mg, 2.5 mmol, 1.0 eq.) was dissolved into 40 ml acetic acid, followed by the
addition of 1.27 g I2 (5.0 mmol, 2.0 eq.) and 1.1 g Cu(OAc)2 (5.5 mmol, 2.2 eq.). The mixture was stirred
and refluxed for about 7 hours and, according to the TLC plate, the starting material was consumed. After
cooling down to room temperature, 150 ml of a saturated sodium bisulfite solution were added to the
mixture and stirred for another 10 min, and the suspension was filtered through a celite pad. The mixture
was four times extracted with DCM (50 ml). The organic phase was combined and dried by MgSO4. Then
the solution was removed under reduced pressure to give the crude product. This was recrystallized from
acetonitrile and dried in a vacuum oven to yield the fine product M2 as white solid powder. The yield was
560 mg, 73.7%.1H NMR (600 MHz, CD3OD) 0.77 (3H, s, CH3-18), 1.23 (3H, S, CH3-19),3.51 (1H, m, H-
3), 3.73 (1H, d, H-4). 1H NMR data was consistent with related reference [8].
2.1.2. Synthesis of 3α,4β-dihydroxy-5α-androstan-17-one (M3)
5α-androst-3-en-17-one (675 mg, 2.5 mmol) was dissolved into 20 ml of 90% formic acid, followed by the
addition of 30% hydrogen peroxide (0.5 ml, 5 mmol) at room temperature. The solution was stirred for
about 2 hours. Thereafter 5α-androst-3-en-17-one was consumed completely according to the TLC plate.
The mixture was diluted with methanol (100 ml) and then basified with 10% aq. sodium hydroxide (pH
10). The mixture was stirred for about 30 min, and was then neutralized with 10% aq. HCl. The solution
was extracted three times with dichloromethane and the organic part was combined and washed with st. aq.
NaHCO3, filtered and dried with MgSO4. The solution was removed under reduced pressure, the crude
product was purified by Si-gel chromatography to give the fine compound M3 700.6 mg, yield89.6%.1H
NMR (600 MHz, CD3OD) 0.88 (3H, s, CH3-18), 1.08 (3H, S, CH3-19), 3.53 (1H, s, H-3), 3.79 (1H, d, H-
4). 1H NMR data was consistent with related reference [9].
2.1.3. Synthesis of 5α-androstane-3β,4β,17β-triol (Triol)
5α-androst-3-en-17β-yl acetate (791.2 mg, 2.5 mmol, 1.0 eq.) was dissolved into 40 ml acetic acid,
followed by the addition of 1.27 g I2 (5.0 mmol, 2.0 eq.) and 1.1 g Cu(OAc)2 (5.5 mmol, 2.2 eq.). The
mixture was stirred and refluxed for about 7 hours and the starting material was consumed over according
to the TLC plate. After cooling down to room temperature the suspension was filtered through a celite pad.
150 ml of a saturated sodium bisulfite solution were added to the mixture and stirred for another 10 min.
The mixture was extracted by DCM 50 ml for four times. The organic phase was combined and dried by
MgSO4, then the solution was removed under reduced pressure to give crude intermediate. The
intermediate was dissolved in 20 ml EtOH and treated with 5 ml 15% aq. NaOH solution for about 3 hours.
When the intermediate was consumed 40 ml water were added and the pH value was adjusted to 5 with 5%
aq. HCl. The mixture was extracted with ethyl acetate 30 ml for four times. The organic phase was
combined and dried by MgSO4, the solvent was removed under reduced pressure to give oil residue. The
residue was purified by Si-gel chromatography to give desired target Triol as white solid powder 346.1 mg,
3

yield 44.9%. 1H NMR (600 MHz, CD3OD) 0.74 (3H, s, CH3-18), 1.04 (3H, S, CH3-19), 3.47-3.51 (1H, m,
H-3), 3.54-3.59 (1H, t, J = 18, 12 Hz, H-17), 3.67 (1H, t, H-4). 1H NMR data was consistent with related
reference [10].
2.1.4 Synthesis of 5α-androstane-3α,17β-diol (3α-diol)
o
5α-androstane-3α-ol-17-one (871.32 mg, 3 mmol) was dissolved into 10 ml methanol and cooled to 0 C,
sodium borohydride (170.42 mg, 4.5 mmol) was added slowly. The mixture was stirred for a while and
allowed to reach room temperature until 5α-androstane-3α-ol-17-one was consumed completely. Water
(100 ml) was added to the mixture and the mixture was extracted with ethyl acetate (30 ml) for three times,
the organic part was combined and dried with MgSO4, after filtration, the solvent was removed under
reduced pressure to give the crude product; Then the crude product was recrystallized from methanol to
give the desired fine 3α-diol as white solid powder 694.4 mg, yield 79.1%.1H NMR (600 MHz, CD3OD)
0.73 (3H, s, 18-CH3), 0.79 (3H, s, 19-CH3), 3.61-3.65 (1H, t, H-17), 4.04 (1H, t, H-3).1H NMR data was
consistent with related references [11-13].
2.2 Transient transfection of recombinant SRD5A1, SRD5A2 and AKR1Cs in COS-1 cells
Coding sequences of SRD5A1, SRD5A2 and AKR1Cs were synthesized by Sangon Biotech according
to GenBank database (SRD5A1, AK313070.1; SRD5A2, NM_000348.3; AKR1C1, NM_001353.5;
AKR1C2, NM_001354.5; AKR1C3, NM_003739.5; and AKR1C4, NM_001818.3). All of these genes were
constructed in the pcDNA3.1 (+) vector, forming recombinant plasmids pcDNA3.1-SRD5A1, pcDNA3.1-
SRD5A2, pcDNA3.1-AKR1C1, pcDNA3.1-AKR1C2, pcDNA3.1-AKR1C3, and pcDNA3.1-AKR1C4,
respectively.
Since COS-1 cells (monkey kidney fibroblast cells) do not express or have low expression of
AKR1Cs, SRD5A1 and SRD5A2 [14-15], it is considered to be a useful model for studying the function of
these enzymes. COS-1-cells were purchased from American Type Culture Collection (Manassas, VA),
plated in six-well plates (2.5×105 cells/well), and maintained at 37°C and 5% CO2 in DMEM (GIBCO-
11995065) supplemented with 100 U/m penicillin, 100 μg/ml streptomycin, 2 mm L-glutamine, and 10%
FBS. After incubation for 24h, COS-1 cells were transfected with 1μg of recombinant plasmid DNA per
well. Cells were transfected using Lipofectamine™ 3000 Transfection Reagent (Thermo L3000001)
according to the manufacturer’s protocol, and all controls were transfected with the pcDNA3.1 (+) vector.
The medium was changed 6 h after transfection. In each of these constructs, enzyme expression is driven
by the pCMV promoter. The cells were harvested 48h after transfection and analysed by western blot and
qPCR. If the enzymes were overexpressed, they could be incubated with substrates or with/without
inhibitors.
2.3 Cloning, expression and purification of recombinant AKR1Cs in Escherichia coli（E. coli）
Similarly, we constructed the synthesized coding sequences of the AKR1Cs into the pET-28a vector
with an N-terminal His-tag, forming recombinant plasmids pET28a-AKR1C1, pET28a-AKR1C2, pET28a-
AKR1C3, and pET28a-AKR1C4. After transformation with these recombinant plasmids, E. coli
BL21(DE3) host cells overexpressed recombinant AKR1Cs. The recombinant proteins were purified on a
Ni-NTA agarose resin column. The purity of the enzymes was analysed by SDS-PAGE with Coomassie
4

blue staining, and the protein concentration was determined by an Enhanced BCA Protein Assay Kit
(Beyotime P0010S).
2.4 Quantitative real-time PCR (qPCR) analyses of recombinant SRD5A1, SRD5A2 and AKR1Cs in
COS-1 cells
After transfection for 48h, total RNA was extracted from the transfected and mock-transfected COS-1
cells with TRIzol (Life Technologies, Inc., Gaithersburg, MD). The first strand cDNA was synthesised
according to the PrimeScript™ RT reagent Kit with gDNA Eraser (TaKaRa RR047A). QPCR amplification
with TB Green™ IIPremix Ex Taq™ polymerase (TaKaRa RR820A) was performed according to the
manufacturer’s protocols. β-tubulin(GenBank accession number: XM_007973204.1) was also amplified
from the same samples to serve as an internal control. Primers used to detect the expression of recombinant
SRD5A1, SRD5A2 and AKR1Cs are listed in Supplementary Table 1.
2.5 Western blot analyses of recombinant SRD5A1, SRD5A2 and AKR1Cs
After transfection for 48h, total protein was extracted from the lysate of the transfected and mock-
transfected COS-1 cells using cell lysis buffer without protein denaturant (Beyotime P0013J) supplemented
with protease inhibitors of 1% PMSF and cocktail. After purification from E.coli, the recombinant AKR1Cs
enzymes were diluted for western blot analyses.
Protein concentration was determined by the Enhanced BCA Protein Assay Kit (Beyotime P0010S).
Equal amounts of proteins (40μg) were loaded and separated on 10% SDS-PAGE, and transferred to PVDF
membranes. The membranes were incubated at 4°C overnight with primary antibodies at a dilution in T-
BST containing 5% non-fat dry milk of 1 μg/ml for SRD5A1 (polyclonal antibody, SAB1401372, Sigma-
Aldrich), 1:2,000 for SRD5A2 (monoclonal antibody, Abcam, ab124877), 1:1,000 for AKR1C1 and
AKR1C2 (polyclonal antibody, CST #13035), 0.5 μg/ml for AKR1C3(monoclonal antibody, R&D system,
MAB-7678-SP), and 1:2,000 for AKR1C4 (monoclonal antibody, R&D system, AF6957), repectively, and
then for 2h at room temperature with the appropriate anti-IgG horseradish peroxidase (HRP)-conjugated
secondary antibody at a dilution of 1:5,000. 1:1,0000 for, β-tubulin was used as the loading control,
quantified with a dilution of 1:1,0000 of HRP-Conjugate mouse monoclonal antibody (Zen-bio, 700608 ).
2.6 Substrate conversion assay by recombinant SRD5A1, SRD5A2 and AKR1Cs in COS-1 cells
After transfecting the cells for 48h, 4-OHA, 4-OHT, AD, T and DHT were added to the medium as
substrates dissolved in DMSO at a final concentration of 10 μM with or without 10 μM of the inhibitor and
were incubated for 3.5 h. 2 mL of the medium were taken out and added the inner standard MT, and then
the sample was extracted after addition of 1.5 mL of K2CO3/KHCO3-solution (20%, 1:1, pH 9.6) with 4
mL of MTBE. The organic layer was separated and evaporated to dryness by nitrogen gas at 40 °C. After
derivatization by TMSF, GC-MSMS analysis was performed as described below.
2.7 Substrate conversion assay by purified recombinant AKR1Cs
The reaction system was 200 μL, with 100 mM sodium phosphate buffer (pH 7.4), 5 mM ammonium
sulfate, 0.2 mM NADPH-Na4, 24 μg of enzyme protein of AKR1Cs, and 50 μM of the substrate with or
without 100 μM of the inhibitor dissolved in DMSO. The reaction mixtures were maintained at 37°C for
3.5 h and then terminated and extracted by MTBE as the above shown. After derivatization by TMSF, GC-
5

MSMS analysis was performed as described below.
2.8 Microscale thermophoresis binding assay
The binding affinities of AKR1Cs and 4-OHA were measured using the microscale thermophoresis
(MST) method with a NanoTemper monolith NT.115 (US, California). The purified AKR1Cs from the E.
coli were fluorescently labelled by the Monolith NTTM Protein Labeling Kit (Germany, Munich). 10 mM
fluorescently labeled AKR1Cs proteins and 16 serial dilutions of test ligands (ranging from 62.5mM to
0.0153 mM) were respectively co-incubated for 5 min in assay buffer (10 mM KH2PO4, 10 mM K2HPO4,
pH 7.4, 0.2 mM NADPH-Na4). Then, the samples were loaded into the NanoTemper glass capillaries and
microthermophoresis was carried out at a light emitting diode (LED) power of 10% and a MST power of
80%. Quantitative binding parameters are obtained by using the serial dilution. By plotting Fnorm against
the logarithm of the different concentrations of the dilution series, a sigmoidal binding curve was obtained.
This binding curve can directly be fitted with the nonlinear solution of the law of mass action, with the
dissociation constant Kd values, calculated via NanoTemper software from duplicate reads of triplicate
experiments.
2.9 Determination of enzyme kinetics
The kinetic constants of reduction were determined in 200 μL systems containing 100 mM sodium
phosphate buffer (pH 7.4), with 5 mM ammonium sulfate, 200 μM NADPH-Na4, and 12 μg AKR1Cs
enzymes, respectively. The concentration of the substrates ranged between 2 and 400 μM. All of the
enzymatic reactions were followed at 37°C for 30 min. The initial velocities for the reduction of substrates
by AKR1Cs were determined by measuring the change of the emission of NADP+ fluorescence derivative
stimulated by NaOH at 460 nm with excitation at 360 nm after the reactions were terminated by HCl and
the left NADPH was removed simultaneously [16]. The detection method of the kinetic constants of
oxidation was the same with reduction except for changing NADPH-Na4 to NADP-Na2 and calculating the
reduction of NADP+. Km and Vmax were determined by the endpoint assay.
2.10 Identification of metabolites by GC-MSMS
Samples (2 mL cell culture medium, 200 μL purified enzyme reaction system) were added 10 ng MT
as an internal standard substance, adjusted to pH 9.0 with the addition of carbonate buffer (pH 9.6). The
samples were extracted by liquid-liquid extraction with MTBE (4 mL); the mixture was shaken
mechanically for 10 min and subsequently centrifuged (1500 g) for 5 min [17]. The organic layer was
transferred to a fresh glass tube and evaporated to dryness under nitrogen gas at 40 °C. The dry residue was
derivatized with 50 µL of MSTFA/NH4I/dithioerythritol (1000:2:4, v/w/w), which is equivalent to a
mixture of MSTFA/trimethyl iodo silane (TMS) and heated for 20 min at 60 °C. Trimethylsilyl (TMS)-enol
and TMS-ethers were formed [8].
The GC-MSMS analyses were performed on an Agilent 7890A gas chromatograph coupled to an
Agilent 7000 triple quadrupole mass spectrometer (Agilent Technologies SpA, Cernuscosul Naviglio,
Milano, Italy). The injector line operated at 280 °C and the transfer line at 300 °C. The initial temperature
of 177 °C was increased at 3 °C/min to 245 °C and then increased at 17 °C/min to 320 °C and kept for 3.2
min at the final temperature. The injection volume was 2 μL in split mode (1:3) [17-18]. Helium was used
6

as the carrier gas (1.5mL/min).The products were qualified by their retention times under specific MRM
mode [8, 18] as described in supplementary Table 2, and the ion chromatograms are shown in
supplementary Fig. 1. They were quantified by standard curves from continuous contents of standard
substances.
2.11 Statistical analysis
All experiments were performed in triplicates and the results are presented as the means ± SEM.
Statistical significance was determined by Student’s t-test, and Vmax and Km values were calculated
according to the Michaelis-Menten equation and a nonlinear regression curve, using GraphPad Prism
version 5 (GraphPad Software, San Diego, CA, USA). Histograms of two dimensions with metabolite
content ratio were drawn by GALLERY, an open source software from https://gallery.echartsjs.com.
3 Results
3.1 Cloning, expression and purification of enzymes
After the recombinant plasmids pcDNA3.1-SRD5A1, pcDNA3.1-SRD5A2, pcDNA3.1-AKR1C1,
pcDNA3.1-AKR1C2, pcDNA3.1-AKR1C3, and pcDNA3.1-AKR1C4 were transiently transfected into
COS-1 cells, qPCR results showed that the SDR5A- and AKR1C- enzymes were overexpressed, compared
with endogenous highly-expressed β-tubulin (Fig.1A). Furthermore, the proteins of the AKR1C enzymes
and the SRD5A enzymes were successfully transiently overexpressed in COS-1 cells, according to the
western blot result (Fig.1C upper and middle panel). Therefore, from both transcription- and protein levels
of the enzymes, it was concluded that this cell system could be used for identification of the metabolites of
4-OHA and 4-OHT originating from reactions catalysed by SRD5A1, SRD5A2 and AKR1Cs.
After the recombinant plasmids pET28a-AKR1C1, pET28a-AKR1C2, pET28a-AKR1C3, and
pET28a-AKR1C4 were transfected into E. coli BL21 (DE3), each AKR1Cs’ elution, flow-through and the
soluble supernatant of the lysate were used for SDS-PAGE analyses. All four enzymes were present in the
supernatant exhibiting a molecular weight of 37KD and were eluted with 90% purity (Fig. 1B).
Furthermore, the western blot also showed that the recombinant AKR1C proteins were expressed in COS-1
and in E.coli (Fig. 1C lower panel). Therefore, all of the AKR1C proteins from E. coli could be used for
catalytic reactions, molecular docking and kinetic analyses.
3.2 Analyses of the reaction products obtained in COS-1 cells after transfection with SRD5A1 and
SRD5A2
The incubation of AD and T with the medium of the COS-1 cells transfected with SRD5A1 and
SRD5A2 led to the conversion of AD into 5α-A, some DHT, and a little ADT, EPI, 3-diol and 3-adiol,
and the conversion of T mostly into of DHT, and a little 3-diol and 3-diol (Fig. 2C, D). The conversions
were inhibited by Dutasteride. Therefore, SRD5A1 and SRD5A2 were successfully temporarily expressed
with 5α-reductase activity in COS-1 cells. The set of experiments also showed that there were some
endogenous 3-HSD, 3-HSD and 17-HSD activities in COS-1 cells since some 3-, 3- and 17-
hydroxy products such as ADT, EPI, DHT, 3-diol and 3-adiol appeared. There was a little endogenous
5α-reductase activity, since small amounts of 5-reduced metabolites of AD and T were detected in mock-
7

transfected cells. However, compared with over-expressed recombinant 5α-reductases, this did not
influence the results obtained after incubation of 4-OHA and 4-OHT with 5α-reductase.
Neither 4-OHA nor 4-OHT were substrates for the recombinant 5α-reductase, and corresponding
metabolites could also not be detected (Fig. 2A, B), which agreed well with previous reports [19]. This
suggested that other intracellular sterol metabolizing enzymes are able to reduce the 4,5 double-bond of a
C-19 steroid with a hydroxyl group attached to C4, a configuration obviously impeding 5α-reduction as
judged by the usual method. Indeed almost all urinary metabolites of 4-OHA or 4-OHT after oral
administration show 5-reduction with concomitant or subsequent 3- or 3-reduction [8]. The obvious
choices were the human cytosolic aldo-keto reductases AKR1C1, AKR1C2, AKR1C3 and AKR1C4.
In contrast to oral administration, intravenous administration to postmenopausal breast cancer patient
leads to the appearance of the glucuronide of 4-OHA and of some minor metabolites, but a considerable
part of the urinary metabolites are the sulphates of M2 (4-hydroxy-epiandrosterone) and M3 (4-hydroxy-
androsterone) [20]. “Triol” is the 17-hydroxy derivative of M2, most likely formed within the cells.
3.3 Binding preference of 4-OHA and 4-OHT to the different AKR1Cs
The above results led to the assumption that other sterol metabolizing enzymes are able to reduce the
4,5 double-bond. We therefore performed docking experiments of AKR1Cs to 4-OHA and 4-OHT by
MST to get Kd values, directly reflecting their interaction.
Both 4-OHA and 4-OHT bound to all AKR1Cs yielding different Kd values (Fig.3). AKR1C3 had the
lowest Kd at 6.7 nM, followed by AKR1C1 and AKR1C4, and AKRIC2 had the highest Kd at 47.1 nM
(Fig.3 A-D). Hence, the binding affinity of 4-OHA with AKR1C3 was the highest. The affinity of
AKR1C2 for the steroid was considerably lower. Also to 4-OHT, AKR1C3 had the lowest Kd at 14.8nM.
AKR1C2 had the highest Kd at 57.4 nM. 4-OHT had somewhat lower binding affinities to all AKR1Cs
than 4-OHA (Fig.3 E-H), which suggested that the presence of a keto-group at C17 instead of a -OH-
group favours binding to the AKR1Cs.
3.4 Reaction product analysed after incubation with recombinant AKR1Cs from COS-1cells
To determine which of the major AKR1C enzymes is involved in the metabolism of 4-OHA and 4-
OHT, the assessment of the directionality of AKR1Cs in a cellular environment was investigated. The
metabolisms of AD, T, and DHT served as controls. When mock transfected COS-1 cells were incubated
with AD, it was converted to T, incubation with T gave rise to AD, 3α-diol and 3β-diol; and incubation
with DHT led to the appearance of 3α-diol and 3β-diol, apparently due to endogenous 3α-HSD, 3β-HSD
and 17β-HSD activities (Fig. 4C-E). However, the endogenous HSD activities did not affect the
examination of the functions of the transfected AKR1Cs since these had much higher activities than the
endogenous HSDs. For example, DHT is easily reduced by 3alpha/beta-HSD and in mock-transfected cells
60% of DHT is converted into 3alpha-diol and 3beta-diol. In AKR1Cs-transfected cells, however, at least
80% of DHT is reduced. Most efficient were AKR1C2 and AKR1C4-transfected cells. Nearly 100% of
DHT is reduced to the diols (Fig.4D).
The metabolism of AD in COS-1 cells transfected with the different AKR1Cs showed that only
AKR1C3 had obvious reductive 17beta-HSD activity (Fig. 4C). The metabolism of DHT showed that all
8

AKR1Cs had both 3α- and 3β-HSD activities, AKR1C2 and AKR1C4 had a higher 3α-HSD activity,
whereas AKR1C1 and AKR1C3 had a higher 3β-HSD activity (Fig. 4D). There was some slight oxidation
of T by AKR1C1 and AKR1C3 to AD. That of 4-OHT to 4-OHA was considerably more pronounced. In
the medium of COS-1 cells transfected with AKR1C2 or AKR1C4, diol formation was observed (and most
likely due to endogenous enzyme activity within the cells?) (Fig. 4E). The above results are consistent with
several related studies [21-23], suggesting that our recombinant enzyme reaction systems in COS-1 cells
were suitable to study the metabolism of 4-OHA and 4-OHT.
Since after intravenous injection of a small amount of 14C-labelled 4-OHA, the major metabolites
found in the urine of postmenopausal breast cancer patients are reduced at C3 and preferentially conjugated
with a sulphate group (M2 and M3 exist exclusively as sulphates) [20], we put emphasis on the ability of
the enzymes to catalyse the 3-reduction that would lead to M2. It is obvious that the extrahepatic
formation of this metabolite could be catalysed by AKR1C1 and AKR1C3 and not by AKR1C2 and
AKR1C4 (Fig. 4A). AKR1C4, which is expressed in the liver only, but not AKR1C2, had a higher capacity
to metabolize 4-OHA and the product was mainly M3, the 3-hydroxy-isomer of M2. Triol was most
likely the result of the interaction of M2 with AKR1C3. Our results demonstrated that all AKR1Cs could
reduce the 4-5 double bond since M2 and M3 are 5-androstanes. This corresponds to the urinary steroid
profile after oral ingestion of 4-OHA [8].
Furthermore, all AKR1Cs had 17-HSD activity and converted 4-OHA to 4-OHT, which was most
obvious for AKR1C3 and AKR1C4. The 17-HSD activity also led to Triol, the presumable metabolite of
4-OHA and 4-OHT. Overall, 4-OHA was metabolized by all AKR1Cs, and the rank order of metabolic
activity towards 4-OHA was AKR1C4>AKR1C3>AKR1C1>AKR1C2. All AKR1Cs catalysed the
conversion of 4-OHT to 4-OHA, M3 and Triol, demonstrating 17-oxidase, 4-5 double bond reductase, and
3/-HSD activities of these enzymes (Fig. 4B).
3.5 Analyses of the reaction products of 4-OHA and 4-OHT after incubation with recombinant
AKR1Cs purified from E. coli
For consolidating and extending the study of the role of AKR1Cs in the metabolism of 4-OHA and 4-
OHT, we further incubated the isolated recombinant AKR1Cs from E.coli with NADPH and the substrates
under investigation in order to detect and quantify the metabolites by GC-MSMS.
For AD, all isolated AKR1Cs reduced AD to T presenting 17-keto reductase activity. AKR1C3 was
most active as a 17-ketosteroid reductase followed by AKR1C1and then AKR1C2, whereas AKR1C4 only
weakly reduced AD to T (Fig. 5C). For DHT, all four enzymes exhibited both 3/-keto-reductase
activities to various degrees. The major sources of 3-diol were AKR1C1, AKR1C3, and AKR1C4, and
that of 3-diol was AKR1C2 (Fig.5D). In this set-up, isolated AKR1C4 preferentially catalysed the 3-
reduction of DHT. Interestingly, inhibition by phenolphthalein changed the stereoselectivity back to the
normal 3-reduction. For T, AKR1Cs had no catalytic function and even no oxidative 17-HSD activity
(Fig.5E). These results indicated that the purified recombinant AKR1Cs showed own intrinsic enzyme
activities and the reactions of the androgens AD, T to ADT or EPI could only be catalysed by all AKR1Cs
after classical 5-reduction.
9

For 4-OHA, some AKR1Cs behaved as 17-HSD or as 3-HSD, but all AKR1Cs converted 4-OHA to
M3, possibly reducing the keto-group at C4, thereby eliminating the double bond to varying extent.
AKR1C3 was most active as a 17-ketosteroid reductase and was inhibited by flufenamic acid. AKR1C4
showed higher activities of 4-keto-reductase, 3-HSD, and 3-HSD, producing the largest amount of M3
and M2, which was strongly inhibited by phenolphthalein. The rank order of possible metabolic activities
towards 4-OHA was AKR1C3>AKR1C4>AKR1C1>AKR1C2 (Fig.5A). This was in stark contrast to the
conditions found for the metabolism of DHT. In contrast to the SRD5A1-catalysed metabolism of AD in
the rhesus monkey the metabolism of the 4-hydroxy derivative apparently has no such bottle neck and
proceeds twice as fast [4].
Like testosterone 4-OHT is not a good substrate for AKR1C1 and AKR1C2. In contrast to T 4-OHT
undergoes some AKR1C catalysed metabolism. Both AKR1C3 and AKR1C4 had some oxidative 17-HSD
activity, and AKR1C4 has additional obvious 4-5 double bond reductase and 3-keto reductase activities,
making a good production of M3 (Fig. 5B). AKR1C3 predominantly catalysed the reduction at C17 of AD
as a first step, whereas it catalysed the first step in the metabolism of 4-OHT as oxidative 17-HSD, which
we have not investigated before. Apparently, AKR1C4 also catalysed the way of 4-OHT via a special 17-
hydroxy pathway to 3-adiol, a pathway whereby the 17 OH-group is preserved. This has been described
for the metabolism of testosterone [20]. AKR1C4 showed obvious 4-5 double bond reductase and 3-keto
reductase activities, producing M3 and some Triol, and it was almost entirely inhibited by phenolphthalein.
Flufenamic acid inhibited the 3α-HSD and 3β-HSD activities of all AKR1Cs.
3.6 Steady-state kinetic analyses
The determination of enzyme kinetic parameters such as Km, Vmax, Kcat and Kcat/Km for the
reduction of 4-OHA, 4-OHT and their controls (DHT, AD, and T) showed that the parameters for a given
sterol were different depending on the enzymes (Table 1). T was not metabolized by all AKR1Cs, AD was
only reduced by AKR1C3, and DHT was reduced by all AKR1Cs. In addition, oxidation of androsterone
(ADT) by AKR1Cs was determined, indicating that ADT was oxidized by AKR1C1, AKR1C2 and
AKR1C4, and was not oxidized by AKR1C3 (Supplementary Table 3). Taken together, the characteristics
of their affinity and reaction efficiency were nearly consistent with reference 21, confirming that these
enzymes had the desired activity and that a functioning reaction system was available.
For
4-OHA,
the
rank
order
of
Km
for
the
different
AKR1Cs
was
AKR1C3<AKR1C1<AKR1C4<AKR1C2, suggesting that 4-OHA had the highest affinity to AKR1C3 then
to AKR1C1 and AKR1C4, and exhibits the lowest affinity to AKR1C2. However, judged from Vmax and
Kcat values, AKR1C2 possesses the highest catalytic rate and the lowest is found with AKR1C3.The
Kcat/Km value represents the catalytic efficiency of an enzyme. Its values for 4-OHA were
AKR1C1>AKR1C3>AKR1C4>AKR1C2, which means that the catalytic efficiency of AKR1C1 is higher
than that of AKR1C3. These results are consistent with the binding preferences of 4-OHA to AKR1Cs
established by MST. 4-OHT has a high affinity for AKR1C4 and AKR1C1, then comes AKR1C2, and
there was no affinity of 4-OHT shown for AKR1C3, which was similar to the reaction efficiency. The
affinities of the AKR1Cs for 4-OHT were lower than for 4-OHA, and the reaction efficiency of the AKR1C
enzymes in the metabolism of 4-OHA was much higher than that of 4-OHT.
10

Discussion
The steroids 4-OHA and 4-OHT are selective inhibitors of aromatase [25] not only inhibiting the
enzyme but inactivating it (mechanism-based inhibitors). 4-OHA has been used successfully in metastatic
breast cancer after failure of tamoxifen [26]. The clinical usefulness of transdermal 4-OHT in the metastatic
setting is currently under investigation [27]. Because extensive hepatic metabolism and conjugation of 4-
OHA after oral administration lead to a low bioavailability, an injectable drug was developed, where a
single dose contained 250 mg of lyophilized 4-OHA and a fluid for reconstitution before intramuscular
injection (Lentaroni.m. Depot). This depot injection was given every two weeks.
In this study, we demonstrated that 4-OHA and 4-OHT were not the substrates for SRD5A1 and
SRD5A2 transiently expressed in COS-1 cells, and furthermore that the reduction of their 4,5 double bond
was catalysed by transiently expressed human cytosolic aldo-keto reductases AKR1C1-AKR1C4. AKR1C1
and AKR1C3 were the most active enzymes in the extrahepatic metabolism of 4-OHA.
After oral ingestion by male volunteers the majority of urinary metabolites were 5-reduced and
glucuronidated [8]. Of the urinary metabolites after oral ingestion only M2 is exclusively sulphated [8].
The hepatic metabolism of 4-OHA and 4-OHT involves all four AKR1C enzymes, which are expressed in
the liver to a similar extent [22]. AKR1C4 is only expressed in the liver. After an intravenous pulse dose of
1 mg of 14C-labelled 4-OHA in postmenopausal breast cancer patients who had previously received
fortnightly five i.m doses (250 mg) of Lentaron-Depot, urinary radioactivity showed an almost exclusive
association with two major sulphated metabolites and the glucuronide of 4-OHA. The sulphated
metabolites are M2 and M3, appearing in the urine in a ratio of 7:3 [20]. The predominance of sulphated
metabolites and the preponderance of the 3-OH metabolite came as a surprise. Our results contribute to
the explanation of this phenomenon by showing that AKR1C1 exhibits preferential 3-reduction and
AKR1C3 is not a very active 3- or 3-HSD. Both enzymes are responsible for the extrahepatic
metabolism of 4-OHA. The fact that there are not many glucuronides found in the urine of these patients
except the glucuronide of the unchanged drug indicates that after i.v. injection the metabolism of the drug
occurs to a considerable part extrahepatically. This most likely is also the case for the drug reaching the
bloodstream after the depot injections used in the clinic. Since 4-OHA undergoes some metabolism in
extrahepatic tissues and also in the tumour, which gives rise to active substances, this may explain its
effectiveness in cases where previous oestrogen deprivation by nonsteroidal aromatase inhibitors has failed.
The hydroxyl group at C4 apparently prevents the activity of 5-reductases but allows the elimination
of the 4,5 double bond. The responsible AKR1C enzyme probably first reduces the 4-keto group that is a
short-lived constituent of the keto-enol tautomerism and then reduces the keto-group at C3. In the
metabolism of 4-OHA, the hydroxyl group at C4 persists.
There are two other 4-OH-steroids closely related to 4-OHA and 4-OHT. These are the anabolic substances
oxymesterone (4-OHT with a 17 methyl-group) and oxabolone (4-OHT without C19). There are
differences in the metabolism of these two anabolic androgenic steroids and that of 4-OHA. Oxymesterone
gives rise to 5-reduced metabolites with a 4-keto group [17], and in oxabolone the 3-keto group may
persist after reduction of the double bond [26]. The mechanism of 5-reduction of 4-hydroxy steroids by
cytosolic aldo-keto reductases are still obscure.
11

The possibility to bypass 5-reductase apparently increases the speed of the metabolism of 4-OHA
and 4-OHT. In the male rhesus monkey, the metabolic clearance rate (MCR) of intravenously injected AD
is only 40% of that of injected 4-OHA [5]. The MCR of AD in postmenopausal women is 1200 L/24 h [30]
and that of 4-OHA is approximately 6000 L/24 h [19]. The greater MCR of 4-OHA is probably due to the
fact that AKR1C1 and AKR1C3 have a wider tissue distribution than 5-reductase. The most important
tissues in this respect are probably adipose tissue and the lung. AKR1C3 is highly expressed in lung, breast
and prostate gland [22] as well as in adipose tissue [30]. The activity of 5-reductase in the lung is very
low [31]. In subcutaneous adipose tissue, the expression of AKR1C3 is 30 times higher than that of 5-
reductases and the expression of AKR1C1 is 70 times higher [30]. This is an important aspect for
transdermal delivery of 4-OHA to the blood and the tumour.
After injection of DHT in rats, the majority of the radioactivity can already be found after only three
minutes in the 3-diol fraction. This quick metabolism is independent of the liver or kidney [32].The likely
organ of DHT-metabolism in the rat is the lung because 98% of the blood returning from the body to the
right atrium is bound to pass through the capillaries of the lung and every substance present in the venous
blood leaving the liver also has to pass through the lung. It has been shown that 100 mg of homogenized
3
lung tissue from adult male rats consumes 90% of the H-DHT. The capacity of the lung of the rat
sufficiently explains the rapid disappearance of DHT from the blood [33]. Intravenously or parenterally
administered substances can be metabolized by the lung, provided the necessary enzymes are expressed in
this organ. AKR1C1 and AKR1C3 are highly expressed in this organ [22]. The lung may play an important
role in the metabolism of 4-OHA.
In postmenopausal women, adipose tissue is a major source of oestrogens in the blood [34-35].
Nonsteroidal aromatase inhibitors such as anastrozole or letrozole can lower systemic levels of E2 by 90%.
[36], whereas the effect of 4-OHA is much weaker. Nevertheless, Formestane can induce disease
stabilization in postmenopausal advanced breast cancer patients progressing on letrozole, anastrozole, and
aminoglutethimide. This effect is completely independent of the E2 level, which usually increases after
withdrawal of nonsteroidal aromatase inhibitors [3]. The inability of 4-OHA to lower systemic levels of E2
sufficiently in comparison with nonsteroidal aromatase inhibitors may be due to the rapid change of
structure of the aromatase-inactivator 4-OHA in adipose tissue. The clinical effectivity of 4-OHA may be
due in a considerable part to the biological effects of locally formed metabolites of 4-OHA.
The most obvious of such steroids is 4-OHT, which has a very high affinity to the androgen receptor
(AR) [37]. In the past, the androgen testosterone-propionate (TP) has been successfully used in the
treatment of advanced breast cancer [38]. In cultures of ER+ breast cancer cells expressing the AR,
androgenic substances can inhibit the proliferation of the cancer cells [39]. AKR1C3 is highly expressed in
all types of breast cancer [40], even in cells representative of triple negative breast cancer [41].
The two major sulphated metabolites of intravenously administered 4-OHA are M2 and M3. They are
products of the metabolism of 4-OHA catalysed by the AKR1C enzymes. Since the 3 reduction prevails,
the most likely active candidate enzymes are AKR1C1 and, according to our results, AKR1C3. Both end
products (M2 and M3) can be reduced at C17 by AKR1C3 and become molecules with affinities to nuclear
receptors such as AR, ER and ER, etc. It is conceivable that 4-OHT is not the only biologically active
12

metabolite of 4-OHA. A major candidate is Triol (3,4,17-trihydroxy 5-androstane), a proven in vitro
metabolite. Together with 4-OHT this and other not yet evaluated metabolites may contribute to the clinical
efficacy of Lentaron-Depot even when oestrogen deprivation cannot inhibit the progression of advanced
breast cancer.
Conclusion
All four AKRICs (AKR1C1-AKR1C4) act as reductases for 4-OHA in both COS-1 cells and in
reaction systems with purified enzymes. They exert 3α-, 3β-, and 17β-hydroxysteroid dehydrogenase
activities and even a newly described catalytic function, 4-5 double bond reduction.
Acknowledgements
This work was supported by start-up funding (Grant numbers 16169, 16248) from the Affiliated
Hospital of Southwestern Medical University, Luzhou grants (Grant numbers 02-00040055, 02-00040108),
and a Southwestern Medical University grant (Grant number2018-ZRZD-004). We express heartfelt thanks
to Sheng Yang from the China Anti-Doping Agency, Tianzhu Zang from the University of Pennsylvania,
and Yuanji Gao from the Sichuan Normal University for their technical advice on the identification of
metabolites and to Iqra from the Sichuan Provincial Center for Gynaecology and Breast Diseases for
serious revision of the manuscript.
Declarations of interest: none.
References
[1] P. Carlini, A. Frassoldati, S. De Marco, A. Casali, E.M. Ruggeri, M. Nardi, P. PapaldoP, A. Fabi, F. Paoloni, F. Cognetti.
Formestane, a steroidal aromatase inhibitor after failure of non-steroidal aromatase inhibitors (anastrozole and letrozole):is a
clinical benefit still achievable? Ann Oncol. 11(2001) 1539-1543.
[2] P. Carlini, G. Ferretti, S. Di Cosimo, E. Collela, R. Tonachella, A. Romiti, S. Tomao, A. Frassoldati, P. Papaldo, A. Fabi,
E.M. Ruggeri, F. Cognetti. Is there a benefit by the sequence anastrozole-Formestane for postmenopausal metastatic breast
cancer women? J Steroid BiochemMol Biol. 86 (2003)107-109.
[3] N. Zilembo, E. Bajetta , E. Bichisao, A. Martinetti, I. La Torre, P. Bidoli, R. Longarini R, T. Portale, E. Seregni, E.
Bombardieri E. The estrogen suppression after sequential treatment with Formestane in advanced breast cancer patients.
Biomed Pharmacother.58 (2004) 255-259.
[4] A.M. Brodie, L.P. Romanoff, K.I. Williams. Metabolism of the aromatase inhibitor 4-hydroxy-4-androstene-3,17-dione
by male rhesus monkeys. J Steroid Biochem. 8 (1981) 693-696.
[5] T.M. Penning, M. Chen, Y. Jin. Promiscuity and diversity in 3-ketosteroid reductases. J Steroid BiochemMol Biol. 151
(2015) 93-101.
[6] F. Labrie. At the cutting edge intracrinology. Mol Cell Endocrinol.78 (1991) C113-118.
[7] L.H. Knox, E. Blossey, H. Carpio, L. Cervantes, P. Crabbé, E. Velarde, J.A. Edwards. Steroids.CCLXXVIII.1 Reductions
of 19-Substituted Androst-4-en-3-ones and Related Compounds. J. Org. Chem.30 (1965) 2198-2205.
[8] M. Kohler, M.K. Parr, G. Opfermann, M. Thevis, N. Schlörer, F.J Marner, W. Schänzer.Metabolism of 4-
hydroxyandrostenedione and 4- hydroxy-testosterone: Mass spectrometric identification of urinary metabolites. Steroids.3
(2007) 278-286.
[9] E.J. Tavares da Silva, M.L. Sá e Melo, A.S. Campos Neves. Novel approach to the synthesis of the aromatase inhibitor 4-
hydroxyandrost-4-ene-3,17-dione (4-OHA). J. Chem. Soc., Perkin Trans. 1(1996) 1649-1650.
13

[10] M.M. Cruz Silva, S. Riva, M. Luisa Sá e Melo. Regioselective enzymatic acylation of vicinal diols of
steroids.Tetrahedron61(2005) 3065-3073.
[11] A. Świzdor, A. Panek, N. Milecka-Tronina. Microbial Baeyer–Villiger oxidation of 5α-steroids using Beauveriabassiana.
A stereochemical requirement for the 11α-hydroxylation and the lactonization pathway.Steroids82 (2014) 44-52.
[12] N. Kennedy and T. Cohen. The stereoselective reductions of ketones to the most thermodynamically stable alcohols
using lithium and hydrated salts of common transition metals. J Org Chem. 80 (2015) 8134-8141
[13] I. Gavrilovic, K. Mitchell, A.D. Brailsford, D. A. Cowan, A.T. Kicman, R.J. Ansell. A molecularly imprinted receptor
for separation of testosterone and epitestosterone, based on a steroidal cross-linker. Steroids 76 (2011) 478-483 ]
[14] T. Laniꢀnik-Riꢁner, H.K. Lin, D. M. Peehl, S. Steckelbroeck, D.R. Bauman, T.M. Penning. Human Type 3 3α-
Hydroxysteroid Dehydrogenase (Aldo-Keto Reductase 1C2) and Androgen Metabolism in Prostate Cells, Endocrinology,
144 (2003) 2922-2932.
[15] L.N. Thomas, R.C. Douglas, J.P. Vessey, R. Gupta, D. Fontaine, R.W. Norman, I.M. Thompson, DA. Troyer, RS.
Rittmaster, CB. Lazier. 5alpha-reductase type 1 immunostaining is enhanced in some prostate cancers compared with benign
prostatic hyperplasia epithelium, J Urol.170 (2003) 2019-25.
[16] S.P. Dufrane, W.J. Malaisse, A. Sener. A micromethod for the assay of aldose reductase, its application to pancreatic
islets. Biochem Med. 32 (1984) 99-105.
[17] S. Yang, J. Lu, Y. Xu, X. Wang. New oxymesterone metabolites in human by gas chromatography-tandem mass
spectrometry and their application for doping control. Drug Test Anal. 7 (2016) 633-643.
[18] A.T.Cawley, G.J.Trout, R.Kazlauskas, A.V.George. The detection of androstenedione abuse in sport: a mass
spectrometry strategy to identify the 4-hydroxyandrostenedione metabolite. Rapid Commun.Mass Spectrom. 22(2008) 4147-
4157.
[19] H.L. Holland, S. Dore, W. Xu, F.M. Brown. Formation of 5 alpha steroids by biotransformation involving the 5 alpha-
reductase activity of Penicillium decumbens.Steroids. 59 (1994) 642-647.
[20] P.E. Lønning, J. Geisler, D.C. Johannessen, H.P. Gschwind, F. Waldmeier, W. Schneider, B. Galli, T. Winkler, W.
Blum, H.P. Kriemler, W.R. Miller, J.W. Faigle. Pharmacokinetics and metabolism of Formestane in breast cancer patients. J
Steroid Biochem Mol Biol. 77 (2001) 39-47.
[21] S. Steckelbroeck, Y. Jin, S. Gopishetty, B. Oyesanmi, T.M. Penning. Human cytosolic 3alpha-hydroxysteroid
dehydrogenases of the aldo-keto reductase superfamily display significant 3beta-hydroxysteroid dehydrogenase activity:
implications for steroid hormone metabolism and action. J Biol Chem. 279 (2004) 10784-10795.
[22] T.M. Penning, M.E. Burczynski, J.M. Jez, C.F. Hung, H.K. Lin, H. Ma, M. Moore, N. Palackal, K. Ratnam. Human
3alpha-hydroxysteroid dehydrogenase isoforms (AKR1C1-AKR1C4) of the aldo-keto reductase superfamily: functional
plasticity and tissue distribution reveals roles in the inactivation and formation of male and female sex hormones. Biochem J.
351 (2000) 67-77.
[23] T.M. Penning, Y. Jin, S. Steckelbroeck, T. Laniꢀnik Riꢁner, M. Lewis: Structure-function of human 3 alpha-
hydroxysteroid dehydrogenases: genes and proteins. Mol Cell Endocrinol. 215 (2004) 63-72.
[24] E.E Baulieu, P. Mauvais-Jarvis. Studies on testosterone metabolism. I.Conversion of testosterone-17-alpha-3H to 5-
alpha- and 5-beta-androstane-3-alpha,17beta-Diol-17-alpha-3H: A new "17-beta-hydroxyl pathway". J Biol Chem. 239
(1964)69-77.
[25] A.M. Brodie, W.C Schwarzel, H.J. Brodie. Studies on the mechanism of estrogen biosynthesis in the rat ovary-I.J
Steroid Biochem. 10 (1976)787-793.
[26] L.R Wiseman, D. McTavish. Formestane. A review of its pharmacodynamic and pharmacokinetic properties and
therapeutic potential in the management of breastcancer and prostatic cancer.Drugs. 45 (1993) 66-84.
[27] M. Zweifel, B. Thürlimann, S. Riniker, P. Weder, R. von Moos, O. Pagani, M Bigler, K.M. Rothgiesser,C. Pilop,
H.Hawle, C. Brauchli, C. Tapia, W. Schoenfeld, C. Sessa; Swiss Group for Clinical Cancer Research (SAKK). Phase I trial
of the androgen receptor modulator CR1447 in breast cancer patients. Endocr Connect. 7 (2017) 549-556.
14

[28] M.V. Barbarulo, F. Buiarelli, M. Ciardi, A. Giarrusso, A. Giarrusso, R. Cartoni. Capillary GC-MS investigation of the
metabolism and excretion of oxabolone in man J High Resolut Chromatogr 8 (1995)705-708.
[29] C. Longcope, W. Jaffee, G. Griffing. Metabolic clearance rates of androgens and oestrogens in ageing women.
Maturitas. 4 (1980) 283-290.
[30] K. Blouin, A. Veilleux, V. Luu-The, A. Tchernof. Androgen metabolism in adipose tissue: recent advances. Mol Cell
Endocrinol.301(2009) 97-103.
[31] L. Milewich, P.S. Parker, P.C. MacDonald. Testosterone metabolism by human lung tissue. J Steroid Biochem. 1 (1978)
29-32.
[32] V. L. Gay. Ineffectiveness of DHT treatment in producing increased serum DHT in orchidectomized rats. Evidence for
rapid in viva metabolism of DHT to androstanediol. Fed.Proc. 34 (1975) 303.
[33] L.W. Kao, M.J. Crosswell, J. Weisz. Steroid metabolism by the lung: conversion of dihydrotestosterone to 5 alpha-
androstan-3 alpha, 17 beta-diol by rat lung tissue in vitro. Endocrinology. 104 (1979)664-666.
[34] D.L. Hemsell, J.M. Grodin, P.F Brenner, P.K. Siiteri, P.C. MacDonald. Plasma precursors of estrogen. II. Correlation of
the extent of conversion of plasma androstenedione to estrone with age. J ClinEndocrinolMetab. 38 (1974)476-479.
[35] S.E. Bulun, E.R. Simpson. Competitive reverse transcription-polymerase chain reaction analysis indicates that levels of
aromatase cytochrome P450 transcripts in adipose tissue of buttocks, thighs, and abdomen of women increase with advancing
age. J Clin Endocrinol Metab. 78 (1994) 428-432.
[36] R.C. Coombes. Aromatase inhibitors and their use in the adjuvant setting. Recent Results Cancer Res. 152 (1998)277-84.
[37] J.H. Davies, R.J. Shearer, M.G. Rowlands, G.K. Poon, J. Houghton, M. Jarman, M. Dowsett. Effects of 4-
hydroxyandrost-4-ene-3,17-dione and its metabolites on 5alpha-reductase activity and the androgen receptor. J Enzyme
Inhib. 6 (1992)141-147.
[38] I.S. Goldenberg. Testosterone propionate therapy in breast cancer. JAMA.188 (1964) 1069-1072.
[39] S.N. Birrell, J.M. Bentel, T.E. Hickey, C. Ricciardelli, M.A. Weger, D.J. Horsfall, W.D.Tilley. Androgens induce
divergent proliferative responses in human breast cancer cell lines. J Steroid BiochemMol Biol. 52 (1995) 459-467.
[40] K.M. McNamara, H. Sasano. The role of 17βHSDs in breast tissue and breast cancers.Mol Cell Endocrinol. 18
(2018)30314-30319.
[41] K.M. McNamara, T. Yoda, A.M. Nurani, Y. Shibahara, Y. Miki, L. Wang, Y. Nakamura, K. Suzuki, Y. Yang, E. Abe,
H. Hirakawa, T. Suzuki, N. Nemoto, M. Miyashita, K. Tamaki, T. Ishida, K.A. Brown, N. Ohuchi, H. Sasano. Androgenic
pathways in the progression of triple-negative breast carcinoma: a comparison between aggressive and non-aggressive
subtypes. Breast Cancer Res Treat. 145 (2014) 281-293.
15

Figure Legends
Fig.1 Transfection into COS-1 cells and successful expression of SRD5A1, SRD5A2 and AKR1Cs. A:
Relative expression levels of mRNA of recombinant SRD5A1 and AKR1Cs in transfected COS-1 cells. β-
Tubulin was set as reference gene. The data is expressed as a histogram containing the means ± SEM. of
three independent experiments (n=3 each group).* P<0.05; ** P<0.01. B: SDS-PAGE analyses of
recombinant AKR1Cs after purification from the lysates of E. coli BL21(DE3) .C: Western blot analyses
of recombinant SRD5A1 and SRD5A2 and of AKR1Cs. pcDNA3.1 served as negative control, and β-
Tubulin was used as internal reference.
Fig.2 Recovery of metabolites from the medium of COS-1 cells transfected with SRD5A1 and
SRD5A2. After transfection of COS-1 cells for 48h with pcDNA3.1-SRD5A1 or pcDNA3.1-SRD5A2 or
only pcDNA3.1, 10μM of 4-OHA (A), 4-OHT (B), AD (C) and T (D) were incubated to the medium.
This was done in the absence (N-In) or the presence (In) of 10 μM of the inhibitor dutasteride for 3.5 h.
The data represents mean values of three different experiments. 4-OHA: 4-hydroxandrostenedione; 4-
OHT: 4-hydroxytestosterone; M2: 3β,4β-dihydroxy-5α-androstan-17-one; M3: 3α,4β-dihydroxy-5α-
androstan-17-one; Triol: 5α-androstane-3β,4β,17β-triol; AD: 4-androstenedione; T: Testosterone; ADT:
androsterone; EPI: Epiandrosterone; DHT: dihydrotestosterone; 3α-diol: 3α-androstanediol; 3β-diol: 3β-
androstanediol; 5α-A: 5α-androstanedione.
16

Fig.3 Binding affinities of 4-OHA and 4-OHT to AKR1Cs measured by microscale thermophoresis.
The Kd values for the ligands (4-OHA and 4-OHT) and the receptors (AKR1Cs) were calculated using the
mass action equation via NanoTemper software from duplicate reads of triplicate experiments. A-D:
binding affinity between 4-OHA and AKR1C1, AKR1C2, AKR1C3, and AKR1C4,respectively. E-H:
binding affinity between 4-OHT and AKR1C1, AKR1C2, AKR1C3, and AKR1C4,respectively. 4-OHA: 4-
hydroxandrostenedione; 4-OHT: 4-hydroxytestosterone.
Fig.4 Recovery and identification of metabolites from the culture-medium of COS-1 cells after
transfection with the four aldo-keto reductases. After transfection of COS-1 cells for 48h with
pcDNA3.1-AKR1Cs or pcDNA3.1, 10μM of 4-OHA (A), 4-OHT (B), AD (C), DHT (D) and T (E)were
17

incubated to the medium for 3.5h. This was done in the absence (N-In) or in the presence (In) of 100μM
flufenamic acid, the inhibitor of AKR1C1-AKR1C3, and in case of AKR1C4 in the absence or presence of
100μM of the inhibitor phenolphthalein. The data represents the mean values of three different
experiments. 4-OHA: 4-hydroxandrostenedione; 4-OHT: 4-hydroxytestosterone; M2: 3β,4β-dihydroxy-5α-
androstan-17-one; M3: 3α,4β-dihydroxy-5α-androstan-17-one; Triol: 5α-androstane-3β,4β,17β-triol; AD:
4-androstenedione; T: Testosterone; ADT: androsterone; EPI: Epiandrosterone; DHT: dihydrotestosterone;
3α-diol: 3α-androstanediol; 3β-diol: 3β-androstanediol; 5α-A: 5α-androstanedione.
Fig.5 Activities of purified aldo-keto reductases to 4-OHA, 4-OHT, AD and other natural androgens.
50M of 4-OHA (A), 4-OHT (B), AD (C), DHT (D) and T (E) were incubated for 3.5 h in the presence
18

(In) or absence (N-In) of 100 μM flufenamic acid, the inhibitor of AKR1C1-AKR1C3 or of 100 μM of the
inhibitor of AKR1C4 phenolphtalein. The reaction system was 200 μL at pH 7.4 containing 200 μM
NADPH and 24 μg of purified AKR1C-enzymes. The results represent the mean values of three different
experiments. 4-OHA: 4-hydroxandrostenedione; 4-OHT: 4-hydroxytestosterone; M2: 3β,4β-dihydroxy-5α-
androstan-17-one; M3: 3α,4β-dihydroxy-5α-androstan-17-one; Triol: 5α-androstane-3β,4β,17β-triol; AD:
4-androstenedione; T: Testosterone; ADT: androsterone; EPI: Epiandrosterone; DHT: dihydrotestosterone;
3α-diol: 3α-androstanediol; 3β-diol: 3β-androstanediol; 5α-A: 5α-androstanedione.
Table 1 Steady-state kinetic parameters for the four alde-keto reductases obtained from reduction of
4-OHA, 4-OHT, AD, DHT and T.
19

Kcat/Km(min-1 mM-1)
Enzyme SubstrateKm(μM)Vmax(nmol/min/mg) Kcat(min-1)
·
4-OHA 8.43
4-OHT 6.03
39.75
12.64
N.D.a
37.18
N.D.
1.47
0.47
N.D.
1.38
N.D.
174.47
77.56
N.D.
AKR1C1 AD
N.D.
16.00
N.D.
DHT
T
85.98
N.D.
4-OHA 33.62
4-OHT 26.87
52.27
41.08
N.D.
28.33
N.D.
1.93
1.52
N.D.
1.05
1.93
57.53
56.57
N.D.
AD
DHT
T
N.D.
10.29
N.D.
AKR1C2
101.87
57.53
4-OHA 2.30
4-OHT N.D.
8.41
N.D.
6.34
0.31
N.D.
0.23
0.53
N.D.
135.29
N.D.
AKR1C3 AD
2.12
11.65
N.D.
110.65
45.23
N.D.
DHT
T
14.24
N.D.
4-OHA 9.77
4-OHT 6.22
27.7
16.41
N.D.
42.74
N.D.
1.02
0.61
N.D.
1.58
N.D.
104.90
97.62
N.D.
AKR1C4 AD
N.D.
8.84
N.D.
DHT
T
178.89
N.D.
a: N.D., Not Detectable. 4-OHA: 4-hydroxandrostenedione; 4-OHT: 4-hydroxytestosterone; AD: 4-
androstenedione; DHT: dihydrotestosterone; T: Testosterone.
20

21