Discovery, biosynthesis, and structure elucidation of metabolites of a doping agent and a direct analogue, tetrahydrogestrinone and gestrinone, using human hepatocytes

Article abstract of DOI:10.1021/ac050150y

Tetrahydrogestrinone (18a-homo-pregna-4,9,11-trien-l7β-ol-3-one, THG) is an anabolic androgenic steroid sold to athletes as an undetectable performance enhancer. Being an unapproved substance, no legitimate in vivo human excretion studies could be performed to identify urinary markers of this doping agent. In vitro systems were used as an alternative approach to study the human metabolism of THG and the gestrinone analogue, which is a marketed drug. Incubations of both compounds in the presence of human hepatocytes led to formation of oxidative and glucuroconjugated metabolites. Microgram quantities of the major in vitro metabolites were biosynthesized using human hepatocytes, characterized by HPLC/ MS/MS, and their structures elucidated by NMR. Due to high structure similarity, both THG and gestrinone had an analogous in vitro metabolic pathway leading to successive addition of a hydroxyl and a β-glucuronic acid at C-18. This in vitro metabolite of gestrinone was consistent with a previously reported major but unknown human urinary metabolite. The structure of another metabolite of THG was proposed to be a glucuroconjugate of an oxidative product with a hydroxyl group most likely at C-16e. In vitro information reported therein could significantly impact the identification of new urinary markers of THG for doping control purposes.

Full text of DOI:10.1021/ac050150y

Anal. Chem. 2005, 77, 3164-3172
Discovery, Biosynthesis, and Structure Elucidation
of Metabolites of a Doping Agent and a Direct
Analogue, Tetrahydrogestrinone and Gestrinone,
Using Human Hepatocytes
Jean-Franc¸ ois Le´ vesque,* Erin Templeton, Laird Trimble, Carl Berthelette, and Nathalie Chauret†
Merck Frosst Centre for Therapeutic Research, P.O. Box 1005, Pointe-Claire/ Dorval, Que´bec, Canada H9R 4P8
Tetrahydrogestrinone (18a-homo-pregna-4,9,11-trien-
17â-ol-3-one, THG) is an anabolic androgenic steroid sold
to athletes as an undetectable performance enhancer.
Being an unapproved substance, no legitimate in vivo
human excretion studies could be performed to identify
urinary markers of this doping agent. In vitro systems
were used as an alternative approach to study the human
metabolism of THG and the gestrinone analogue, which
Figure 1. Chemical synthesis of THG.
is a marketed drug. Incubations of both compounds in
the presence of human hepatocytes led to formation of
oxidative and glucuroconjugated metabolites. Microgram
quantities of the major in vitro metabolites were biosyn-
thesized using human hepatocytes, characterized by HPLC/
MS/MS, and their structures elucidated by NMR. Due to
high structure similarity, both THG and gestrinone had
an analogous in vitro metabolic pathway leading to suc-
cessive addition of a hydroxyl and a â-glucuronic acid at
C-18. This in vitro metabolite of gestrinone was consistent
with a previously reported major but unknown human
urinary metabolite. The structure of another metabolite
of THG was proposed to be a glucuroconjugate of an
oxidative product with a hydroxyl group most likely at
C-16E. In vitro information reported therein could signifi-
cantly impact the identification of new urinary markers
of THG for doping control purposes.
agency (WADA).4 THG was shown to be a potent androgen and
progestin5 and was prohibited by doping control authorities.4 THG
is not approved as a drug or dietary supplement by the Food and
Drug Administration (FDA), and therefore, no legitimate in vivo
metabolism studies in human can be conducted for the identifica-
tion of potential urinary markers for doping control purposes.6
The only excretion study available was performed in one baboon
and led to identification of THG in urine. No information was
provided regarding the urinary excretion of THG metabolites in
the primate.1
It has been well documented that the pharmaceutical industry
uses in vitro systems such as hepatocytes to study the metabolism
of discovery compounds and ultimately postulate the main
biotransformations that could occur in vivo in humans.7 In vitro
metabolism studies offer an important advantage over clinical
studies in that they minimize safety issues by avoiding human in
vivo exposure to potentially harmful substances. It has been
previously reported that metabolites formed in human hepatocyte
incubations of banned AAS (e.g., androstenedione and norandro-
stenedione) correlated well with urinary metabolic profiles ob-
served in humans following per os administration.8 Furthermore,
in vitro metabolic systems can be used to produce sufficient
amount of metabolites for structure elucidation by nuclear
magnetic resonance (NMR).9-11 These can then be used as
In June 2003, the structure of a new anabolic androgenic
steroid (AAS) extracted from a spent syringe was elucidated as
tetrahydrogestrinone (18a-homo-pregna-4,9,11-trien-17â-ol-3-one,
THG; Figure 1) by the UCLA Olympic Analytical Laboratory.1
Allegedly designed by Balco laboratories in the United States,
THG was sold as an undetectable performance enhancing sub-
stance to elite amateur and professional athletes.2,3 It is a direct
analogue of gestrinone (Figure 1), a marketed drug for the
treatment of endometriosis and banned by the world antidoping
(4) World Anti-Doping Agency. The 2004 Prohibited List. International Standard.
World Anti-Doping Code. Montreal, Canada, 2004.
(5) Death, A. K.; McGrath, K. C.; Kazlauskas, R.; Handelsman, D. J. J. Clin.
Endocrinol. Metab. 2004, 89, 2498-2500.
* Corresponding author. E-mail:
jeanfrancois_levesque@merck.com.
Phone: 514 428-3726. Fax: 514 428-4900.
(6) FDA Consumer Mag. 2004, 38, 4.
(7) Ekins, S.; Ring, B. J.; Grace, J.; McRobie-Belle, D. J.; Wrighton, S. A. J
† Current address: Virochem Pharma, 275 boulevard Armand-Frappier, Laval,
Que´bec, Canada H7V 4A7.
Pharmacol. Toxicol. Methods. 2000, 44, 313-324.
(1) Catlin, D. H.; Sekera, M. H.; Ahrens, B. D.; Starcevic, B.; Chang, Y.-C.;
Hatton, C. K. Rapid Commun. Mass Spectrom. 2004, 18, 1245-1249.
(2) Knight, J. Nature 2003, 425, 752.
(8) Le´vesque, J.-F.; Gaudreault, M.; Houle, R.; Chauret, N. J. Chromatogr., B
2002, 780, 145-153.
(9) Watt, A. P.; Mortishire-Smith, R. J.; Gerhard, U.; Thomas, S. R. Curr. Opin.
(3) Handelsman, D. J. Sci. STKE 2004, 244, pe41.
Drug Discovery Dev. 2003, 6, 57-65.
3164 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005
10.1021/ac050150y CCC: $30.25 © 2005 American Chemical Society
Published on Web 04/02/2005

authentic standards for in vivo sample analysis and biological
activity testing. Such an approach has been used to identify new
in vitro metabolites of norandrostenedione,12 a precursor of
nortestosterone banned by the WADA.4
Hospitalier de l’Universite´ de Montre´al (CHUM), human liver
tissues were obtained from fully informed consenting patients (age
(gender): 64 (M), 72 (F), 72 (F), 56 (F), 72 (F)) undergoing partial
hepatectomies at a local hospital due to hepatic metastasis of
adenocarcinoma. Tissue samples were transported from the
operating room in ice cold University of Wisconsin solution, and
human hepatocytes were isolated as previously described.14,15 In
summary, a two-step collagenase perfusion was performed and
the viability of the cells was determined by tryptan blue uptake.
Preparations showing cell viability under 80% were discarded.
For small-scale incubations, 0.5 mL of 2 × 106 cells/mL in
Krebs-Henseleit buffer (pH 7.4, with 3 g/L Hepes added) (Sigma)
were preincubated for 20 min at 37 °C under 95%:5% O2/CO2 (BOC
Gases; Montreal, Canada) in a 48-well plate. A final concentration
of 50 µM gestrinone or THG (2.5 µL of a 10 mM stock solution
in acetonitrile) was added to each well and incubated at 37 °C
under 95%:5% O2/CO2 for 2 h. The incubations were quenched
by adding one volume of acetonitrile to each well. Control
incubations included quenched incubates spiked with the parent
and a blank incubation. The quenched samples were transferred
into Eppendorf tubes, vortexed, and centrifuged at 14 000 rpm
for 10 min.
For large-scale incubations, a protocol similar to that described
above was used except that a total volume of 75 mL of 2 × 106
cells/mL solution was divided into two 125-mL sterile plastic
Erlenmeyer flasks and 50 µM gestrinone or THG (dissolved in
acetonitrile as above) was added to each solution after a 20-min
preincubation period. After 3-h incubation, the solution was
transferred into separate tubes and frozen at -80°C pending
metabolite isolation.
(2) â-Glucuronidase Hydrolysis. Following solid-phase
extraction isolation of desired phase II metabolites using the
protocol described in under Solid-Phase Extration and methanol
evaporation, 500 µL of 125 mM phosphate buffer (pH 6.9) and
2000 units of â-glucuronidase were added to the dried fractions.
Control incubations were also conducted in the absence of
â-glucuronidase along with a blank sample. All samples were
incubated for 1 h at 50 °C in a water bath. Samples were quenched
with one volume of acetonitrile, vortexed, and centrifuged for 10
min at 14 000 rpm.
This paper reports the use of in vitro systems as an alternative
approach to in vivo clinical studies in humans, to determine the
metabolism of THG and gestrinone. Both doping agents were
incubated in the presence of human hepatocytes from five different
donors in order to evaluate their metabolic profiles. Subsequent
biosynthesis, isolation, and characterization by high-pressure
liquid chromatography (HPLC) coupled to photodiode array
(PDA), fluorescence, mass spectrometer (MS) or MS/MS detec-
tors, and NMR were done to elucidate the structure of the major
metabolites. To evaluate the in vivo relevance of these new
metabolites of THG, in vitro versus in vivo correlation was
assessed for gestrinone based on urinary metabolites previously
reported in the literature in humans following per os administra-
tion.13
EXPERIMENTAL SECTION
Reagents. Gestrinone was obtained from APIN Chemicals
(Oxon, U.K.). â-Glucuronidase from Escherichia coli type IX-A,
NADPH, and 99% absolute ethanol were purchased from Sigma
(St. Louis, MO). Lindlar catalyst (5% palladium on calcium
carbonate poisoned with lead) was purchased from Strem Chemi-
cals (Newburyport, MA). Formic acid and monobasic sodium
phosphate were obtained from Anachemia (Montreal, Canada),
and ammonium hydroxide was obtained from A&C American
Chemicals (Saint-Laurent, Canada). Methanol and acetonitrile
were HPLC grade and obtained from Fisher (Nepean, Canada).
THG Synthesis and Purification. THG was synthesized by
reduction of the alkyne portion of gestrinone to an alkane (Figure
1). To a solution of gestrinone (50 mg, 0.162 mmol) in EtOH (2
mL, 0.8 M) at room temperature was added Lindlar catalyst (5%
palladium on calcium carbonate poisoned with lead) (10 mg) and
a balloon of hydrogen (gas). The mixture was stirred at room
temperature for 30 min. The reaction was stopped, filtered through
Celite, and washed with 20 mL of ethanol. Purification was
achieved by preparative HPLC using a Waters 600 Multisolvent
delivery system with a Waters 486 tunable absorbance detector
(Milford, MA) and an XTerra C18 5 µm, 30 × 100 mm column
(Waters) using 45% acetonitrile in water at a flow rate of 12 mL/
min for 30 min. The injection volume was 250 µL, and the single-
wavelength UV detector was set at 345 nm. Fractions were
collected manually, analyzed by HPLC/PDA/fluorescence to
determine purity. Pure fractions were then pooled and evaporated
to dryness. Scale-up to 150 mg of gestrinone and 30 mg of Lindlar
catalyst was performed using conditions previously described.
Purity and structure of THG were confirmed by HPLC/PDA/MS
and NMR analysis.
(3) Human Recombinant P-450 Incubations. The incuba-
tions were performed using 50 pmol/mL human recombinant P450
enzyme (rP450) in 125 mM phosphate buffer (pH 7.4), with 1 mM
NADPH and 20 µM gestrinone or THG added, in a final volume
of 500 µL. The rP450 (Supersomes) used were purchased from
BD Gentest Corp. (Bedford, MA) and included the following:
1A1+OR, 1A2+OR, 1B1+OR, 2A6+OR+b5, 2B6+OR+b5,
2C8+OR, 2C9*1+OR+b5, 2C9*2+OR, 2C9*3+OR, 2C18+OR,
2C19+OR+b5,
2D6*1+OR,
2E1+OR+b5,
2J2+OR+b5,
In Vitro Metabolism. (1) Human Hepatocyte Isolation and
Incubation. As approved by the bioethic committee of the Centre
3A4+OR+b5, 3A5+OR, 3A7+OR+b5, 4A11+OR, 4F2+OR+b5,
4F3A+OR+b5, 4F3B+OR+b5, 4F12+OR+b5, and an insect cell
control. The samples were incubated in a water bath at 37 °C for
1 h. Samples were quenched with one volume of acetonitrile,
vortexed, and centrifuged for 10 min at 14 000 rpm.
(10) Chauret, N.; Yergey, J. A.; Brideau, C.; Friesen, R. W.; Mancini, J.; Riendeau,
D.; Silva, J.; Styhler, A.; Trimble, L. A.; Nicoll-Griffith, D. A. Bioorg. Med.
Chem. Lett. 2001, 11, 1059-1062.
(11) Nicoll-Griffith, D. A.; Yergey, J.; Trimble, L.; Williams, H.; Rasori, R.;
Zamboni, R. Drug Metab. Dispos. 1992, 20, 383-389.
(12) Le´vesque, J.-F.; Gaudreault, M.; Aubin, Y.; Chauret, N. Steroids 2005, 70,
305-317.
(14) Li, A. P.; Roque, M. A.; Beck, D. J.; Kaminski, D. L. J. Tissue Cult. Methods
1992, 14, 139-146.
(13) Kim, Y.; Lee, Y.; Kim, M.; Yim, Y.-H.; Lee, W. Rapid Commun. Mass Spectrom.
2000, 14, 1717-1726.
(15) Silva, J. M.; Day, S. H.; Nicoll-Griffith, D. A. Chem.-Biol. Interact. 1999,
121, 49-63.
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 3165

Metabolite Isolation. (1) Solid-Phase Extraction. The
protocol used here was adapted from a reported procedure for
selective isolation of conjugated and unconjugated metabolites
using anionic solid-phase extraction (SPE) cartridges.16 The
incubation mixtures were thawed at room temperature and diluted
1:1 with 0.1% formic acid(aq). Samples were then vortexed and
centrifuged for 10 min at 4300 rpm. The supernatant obtained was
loaded on a conditioned Oasis MAX 6 mL SPE cartridge (Waters)
and then washed once with 3 mL of 2% NH4OH(aq) and three times
with 3 mL of methanol to recover unconjugated metabolites. The
glucuroconjugated metabolites were eluted in six 3-mL fractions
of 2% formic acid in methanol. All washings and eluates were
collected for Analytical HPLC/PDA/Fluorescence Analysis. The
fractions containing metabolites of interest were pooled for further
purification by preparative HPLC.
(2) Preparative HPLC. Methanolic fractions obtained from
SPE isolation were evaporated to dryness and reconstituted in
425 µL of a 1:1 mixture of methanol and water. Isolation was
carried out using a Waters 2790 HPLC system and Waters 996
PDA detector. An injection volume of 100 µL, a flow rate of 1 mL/
min, and an XTerra MS C18 2.5 µm, 4.6 × 50 mm column (Waters)
were used. The mobile phase was composed of 0.1% formic acid
For the HPLC/PDA/MS/MS, a Waters 2790 chromatograph
and a Waters 996 PDA were coupled with a Micromass Quattro
LC mass spectrometer. The same parameters for the liquid
chromatography were used as for HPLC/PDA/fluorescence
analysis. The mass spectrometer was equipped with an electro-
spray ionization source and was operated in positive ion mode.
The cone voltage was set at 30 V, and daughter ion scans were
performed using collision energies of 10, 20, or 30 eV.
(3) NMR Analysis. All NMR spectra were acquired on a
Varian (Palo Alto, CA) Inova 600-MHz spectrometer equipped with
an inverse H/C/N 5-mm cryogenic probe. Gestrinone and THG
were dissolved in 160 µL of acetonitrile-d3 and placed in a standard
3-mm NMR tube. Metabolites were prepared by first dissolving
in 115 µL of DMSO-d6 (Cambridge Isotopes) to which was added
5 µL of D2O (Cambridge Isotopes). The solution was placed in a
3-mm Shigemi (Shigemi Inc.) NMR tube matched to DMSO-d6.
Proton and carbon assignments were made using a combination
of 1D-1H spectra and 2D gCOSY, NOESY, HSQC, and HMBC
experiments.
RESULTS AND DISCUSSION
Gestrinone, a commercially available compound, was used as
a starting material for the organic synthesis of THG. In the
presence of Lindlar catalyst, the ethynyl moiety of gestrinone was
reduced to an ethyl group (Figure 1). Due to the presence of three
conjugated double bonds and a ketone moiety, both gestrinone
and THG displayed a UV spectrum with λmax at 345 nm and could
be monitored by fluorescence detection at λexc 365 nm to λem 465
nm. UV and fluorescence spectra of THG are presented in Figure
S-1 (Supporting Information). When the reaction was allowed to
go to completion, THG was formed, as well as several overreduced
side products devoid of a fluorescence signal (λexc at 365 nm to
λem at 465 nm) with λmax at 308 or 248 nm instead of 345 nm. By
HPLC/MS, the corresponding [M + H]+ ions of these side
products were 6 and 8 u higher than that of gestrinone ([M +
H]+ at m/z 309), respectively. In addition, totally reduced
compounds with [M + H]+ at m/z 319 and showing no UV signal
were also formed. Hence, successive reduction of the ethynyl
group and of the alkene moieties at C-11, C-9, and C-4 of
gestrinone occurs under these conditions, leading to formation
of numerous undesirable overreduced products. Since these were
not resolved from THG by preparative HPLC, the reaction was
repeated under the same conditions but the reaction was stopped
once total consumption of gestrinone and a 1:1 ratio of dihy-
drogestrinone to THG had been achieved. No side products were
observed under these conditions, and THG was purified by
preparative HPLC. The reaction was scaled up to 150 mg of
gestrinone and 30 mg of Lindlar catalyst and yielded 30.1 mg of
THG (>99% pure). The structure of THG was confirmed by
HPLC/PDA/MS/MS and NMR analysis and results for THG, as
well as for gestrinone, are presented in Figures S-2 and S-3
(Supporting Information) and Table 1.
(eluent A) and methanol (eluent B). An isocratic elution,
(aq)
followed by a quick gradient, was used to separate the metabolites;
50% eluent B for 17 min, increase to 90% eluent B over 1 min, and
remain at 90% eluent B for 4 min. All peaks were collected
manually and analyzed by HPLC/PDA/fluorescence for purity
determination. Pure fractions were combined and evaporated to
dryness prior to metabolite characterization.
In Vitro Metabolite Characterization. (1) Analytical HPLC/
PDA/Fluorescence Analysis. An Agilent (Palo Alto, CA) 1100
Series liquid chromatograph equipped with a PDA and a fluores-
cence detector was used with a Zorbax RX-C18, 5 µm, 4.6 × 150
mm column (Agilent). The mobile phase was composed of 0.1%
formic acid (aq) (eluent A) and acetonitrile (eluent B). The gradient
elution program was from 10 to 60% eluent B over 17 min, to 90%
eluent B in 1 min, and then isocratic at 90% eluent B for 4 min.
Prior to HPLC analysis, supernatants obtained from in vitro
incubates were diluted 1:1 with water to decrease the organic
content to less than 25%. An injection volume of 50 µL, a flow rate
of 1 mL/min, and a column temperature of 40 °C were used. The
PDA detector was set to acquire from 210 to 400 nm. The
fluorescence detector was programmed with an excitation wave-
length at 365 nm and an emission wavelength at 465 nm. The
fluorescence detector was also programmed to perform a multiple-
wavelength emission scan (λexc at 365 nm; λem 390-500 nm, in
5-nm steps) or a multiple-wavelength excitation scan (λexc 240-
440 nm, in 5-nm steps; λem at 465 nm).
(2) HPLC/PDA/MS and HPLC/PDA/MS/MS Analysis.
A Waters 2790 chromatograph and a Waters 996 PDA were
coupled with a Micromass ZQ mass spectrometer (Manchester,
England). HPLC condition used were identical to those reported.
The mass spectrometer was equipped with an electrospray
ionization source and was operated in both positive and negative
ion modes with a cone voltage of 30 V. The masses scanned in
both ion modes were from 100 to 800 u.
The liver is known to be an important site of metabolism for
a large variety of xenobiotics. Considering its potential implication
in the in vivo metabolism of THG and gestrinone, human
hepatocytes from five different donors were isolated and incubated
in the presence of both compounds, separately. Quenched
incubations were analyzed by HPLC/PDA/fluorescence and
HPLC/PDA/MS, and metabolic profiles obtained for THG are
(16) Le´vesque, J.-F.; Laterreur, J.; Sorensen, D.; Chauret, N. Abstract from 229th
ACS National Meeting, San Diego, CA. March 2005.
3166 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

Table 1. 1H and 13C Chemical Shifts for THG, M1, and M2 as Well as for Gestrinone and Md
THG
THG - M1
1H (partial)
THG - M2
1H
gestrinone
1H
gestrinone - Md
position
1H
13C
13C
13C
1H
13C
1
2.89, 2.81
2.39
24.05
36.41
198.16
122.88
156.53
31.09
2.01
2.78, 2.68
2.30
2.85, 2.60
2.33, 2.28
198.5
24.05
36.75
2.90, 2.84
2.40
24.19
36.48
198.14
123.19
156.3
31.08
27.08
37.5
141.46
127.22
124.57
140.19
51.6
49.14
22.02
39.35
78.12
88.56
72.64
22.59
10.49
2.88, 2.64
2.34, 2.30
24.18
36.83
198.29
123.49
156.51
31.24
27.58
37.33
142.44
127.68
125.10
140.60
55.27
49.47
23.01
41.53
78.46
90.11
N/O
80.00
20.62
104.10
74.18
77.30
72.50
74.01
172.41
2
3
4
5.70
5.64
5.63
123.26
156.87
31.19
27.59
37.35
142.72
127.45
125.12
141.23
55.03
49.83
23.69
36.77
83.57
31.53
7.93
5.72
5.65
5
6
2.60
1.91, 1.25
2.59
2.49
1.80, 1.12
N/O
2.54
1.84, 1.09
2.72
2.61
1.92, 1.28
2.58
2.57
1.85, 1.08
2.73
7
8
37.52
142.26
126.57
124.34
141.51
50.91
49.02
22.25
33.39
83.47
30.19
7.01
9
10
11
12
13
14
15
16
17
20
21
18
18a
1′
2′
3′
4′
5′
6′
6.64
6.43
6.56
6.33
6.55
6.33
6.69
6.50
6.59
6.41
1.73
1.59, 1.50
2.00, 1.64
N/O
2.16, 1.33
3.80
1.60
1.85, 1.38
1.87, 1.74
2.02
1.73, 1.52
2.30, 2.10
1.98
1.90, 1.48
2.29, 2.13
1.60, 1.40
0.95
1.81, 1.47
1.02
1.47, 1.06
0.80
1.58, 1.16
0.85
2.74
1.72, 1.48
1.01
3.22
3.94
1.39
4.21
2.98
3.08
2.99
3.13
24.13
10.75
1.66, 1.28
0.89
4.03
80.52
21.13
103.74
74.19
77.11
72.45
74.1
1.38
4.12
4.27
2.96
2.91
3.10
3.10
3.05
3.02
3.12
3.17
172.64
Figure 2. Chromatograms of fresh human hepatocyte incubations (n ) 5 donors) of THG generated by HPLC/PDA (at λ 345 nm).
pictured at Figure 2. The percent THG remaining after a typical
2-h incubation ranged from 6 to 26% with an average of 13%. All
metabolites observed by UV and MS displayed λmax at ∼345 nm
and a fluorescence signal, indicating no reduction of the conju-
gated system of THG during the incubation. Oxidation and
glucuronidation were the main biotransformations observed based
on HPLC/MS analysis. In ESI positive ion mode, both M1 and
M2 displayed [M + H]+ ions at m/z 505, 192 u higher than that
of the parent drug, consistent with a combination of oxidation (+16
u) and glucuronidation (+176 u). In the case of M3, the [M +
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 3167

Figure 3. HPLC/MS/MS of M1 of THG generated in ESI+ ion mode at (A) 10, (B) 20, and (C) 30 eV collision energy (daughter ions of [M +
H]+ at m/z 505).
Figure 4. HPLC/MS/MS of M2 of THG generated in ESI+ ion mode at (A) 10, (B) 20, and (C) 30 eV collision energy (daughter ions of [M +
H]+ at m/z 505).
H]+ ion was 176 u higher than that of THG, suggesting the
presence of a glucuroconjugate, whereas for M4, the [M + H]+
ion at m/z 325 was indicative of a monooxidation. An unknown
metabolite, M5, was also observed in some incubations. No sulfate
adducts of the parent or oxidative metabolites were observed in
ESI- ion mode in any incubation analyzed. Interestingly, the
metabolic profiles of THG obtained using fresh human hepatocytes
from five different donors were very similar, leading overall to
the formation of M1 and M2 as the main metabolites. To
characterize and elucidate the structure of these two metabolites
by NMR, larger quantities were required.
A large-scale biosynthesis was performed using fresh human
hepatocytes from donor 4. Under the incubation conditions used
for the large scale (3 h) reaction, complete consumption of THG
was noticed, leading to formation of 18% M1 and 33% M2, based
on total peak area (λ at 345 nm). An anionic SPE cartridge was
used to selectively isolate glucuroconjugated from unconjugated
steroids. Then, M1 and M2 from the glucuroconjugated fraction
were isolated by preparative HPLC. Based on NMR analysis,
approximately 15 and 60 µg of M1 and M2 were isolated,
respectively.
Isolated THG metabolites were then characterized by HPLC/
MS/MS and NMR to elucidate their structures. HPLC/MS/MS
analysis was performed on M1 and M2 at collision energies of
10, 20, and 30 eV and daughter ions of [M + H]+ at m/z 505
were acquired (Figures 3 and 4). In general, M2 showed
significantly more fragmentation than M1 under the same MS/
MS conditions. At a collision energy of 10 eV, a parent ion at m/z
505 was largely predominant for M1 whereas for M2, the ion at
m/z 311 [M - glucuronide - H2O]+ was the main species present.
As the collision energy increased, different fragmentation patterns
were observed. For M1, the neutral loss of the glucuronic acid
(176 u) to m/z 329 was preponderant, even at 30 eV. In the case
of M2, sequential glucuronic acid and water losses was observed
at 20 eV (to m/z 311 and 293), and the formation of fragments at
m/z 267 and 239 was favored at higher collision energy (30 eV).
Based on their dissimilar fragmentation patterns observed by MS/
MS, it was speculated that the sites of hydrolylation or glu-
curonidation on M1 and M2 were different.
Complete proton and carbon NMR assignments were per-
formed on THG, M1, and M2 and are shown in Table 1. In the
case of THG, two triplets each integrating for three protons were
3168 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

Figure 5. Chromatograms of fresh human hepatocyte incubations (n ) 5 donors) of gestrinone generated by HPLC/PDA (at λ 345 nm).
observed, corresponding to the methyl protons at C-18a and C-21.
In the case of M2, one of the two triplets was replaced with a
doublet that coupled to a quartet integrating for one proton at 4.0
ppm, suggesting that an oxidation had occurred at either position
18 or 20. The triplet correlated to the readily assignable carbon
at position 17 in the HMBC experiment identifying carbon 18 as
the site of oxidation in M2. In addition to the expected signals
for the steroid, signals for the glucuronic acid were also observed.
The anomeric proton had a 3JHH coupling constant of 7 Hz, which
is consistent with it being in the â configuration. To determine
through which of the two possible hydroxyl groups the â-glu-
curonic acid was linked to the steroid, the HMBC spectrum was
examined. A strong correlation was observed between the ano-
meric proton at C-1′ of the â-glucuronic acid and C-18 of M2. A
similar interaction was also observed between the proton at C-18
and the C-1′ carbon of the glucuronide. This unambiguously
confirmed that the â-glucuronic acid was attached to M2 via its
newly introduced hydroxyl function at C-18 (structure shown in
Figure 7).
In the case of M1, the small amount of metabolite that was
isolated limited the NMR experiments that could be acquired. In
the proton spectrum, two triplets integrating for three protons
each, corresponding to the methyl groups, were observed. This
implied that the two ethyl groups of M1 were unmodified and
suggested that the oxidation occurred on one of the steroid rings.
Using the 1D 1H NMR spectrum and the 2D-gCOSY data, all
protons excluding those at C-14, C-15, and C-16 were determined
to be unchanged. Observation of an alcoholic methane proton at
3.8 ppm excludes C-14 as the site of hydroxylation. Although
partially obscured, a CHOH-CH2 proton coupling pattern was
present, indicating that the site of hydroxylation was either C-15
or C-16. Typical 1H NMR signals for a â-glucuronide were
observed in a 1:1 stoichiometric ratio with those of the steroid
core. However, insufficient material was isolated to successfully
acquire the HMBC spectrum that would allow the identification
of the glucuronide linkage site and confirm that the newly
introduced alcohol was at C-15 or C-16. It is well known for AAS
that C-16 is an important site of hydroxylation in human, whereas
oxidation at C-15 has never been reported in vivo in human.17 It
was therefore proposed that hydroxylation was likely to occur at
C-16ꢀ. The site of glucuronidation would be on either of the alcohol
moieties at C-16ꢀ or C-17â. A tentative structure is shown in Figure
7.
Gestrinone was incubated side by side with THG in human
hepatocytes from the same five donors in order to compare their
metabolic profiles (Figure 5). Similar profiles were obtained across
donors with, on average, 68% gestrinone remaining after the 2-h
incubation. Six potential metabolites were identified in the
supernatant of the incubations by HPLC/PDA/MS. Two metabo-
lites, Ma and Mf, remain unknown as their mass spectra did not
show diagnostic mass changes as compared to the parent
compound. Mb, Mc, and Md were all hydroxylated (+16 u) and
glucuroconjugated (+176 u) metabolites with a parent ion at m/z
501 in ESI+. The parent ion for Me was determined to be at m/z
485, which is in agreement with a glucuronide adduct of gestrino-
ne. As for THG, a large-scale biosynthesis was performed using
cells from donor 4, and solid-phase extraction procedures were
used to selectively isolate glucuroconjugated from unconjugated
(17) Schanzer, W. Clin. Chem. 1996, 42, 1001-1020.
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 3169

Figure 6. HPLC/MS/MS of the main metabolite of gestrinone (Md) generated in ESI+ ion mode at (A) 10, (B) 20, and (C) 30 eV collision
energy (daughter ions of [M + H]+ at m/z 501).
Figure 7. Proposed metabolic pathways leading to formation of main in vitro metabolites of THG in fresh human hepatocytes.
metabolites. All major metabolites were isolated by preparative
HPLC, although only the major metabolite (Md) gave sufficient
amount to allow NMR characterization.
a doublet integrating for three protons was located upfield at 1.4
ppm. There was also a signal integrating for a single proton at
3.9 ppm from a hydrogen located on a carbon linked to an oxygen.
This suggested that the site of hydroxylation was at C-18,
analogous to that seen in M2. A strong correlation was observed
between the anomeric proton at C-1′ of the â-glucuronic acid and
C-18 of Md. A similar interaction was also observed between the
proton at C-18 and the C-1′ carbon of the glucuronide. This
confirmed that the â-glucuronic acid was attached to Md via its
newly introduced hydroxyl function at C-18 (Figure 8).
M1, M2 (THG), and Md (gestrinone) are all oxidative me-
tabolites conjugated to a glucuronic acid. As depicted at Figures
7 and 8, both M2 and Md would have a similar metabolic pathway
with first an oxidation at C-18, followed by addition of a â-glu-
curonic acid on the new alcohol functionality at C-18. In the case
of M1, a combination of hydroxylation, most probably at C-16ꢀ,
and glucuronidation would be implicated in its formation.
To evaluate whether the oxidative intermediates leading to the
formation of glucuronide adducts were present in human hepa-
HPLC/MS/MS analysis of Md revealed a fragmentation
pattern similar to that seen for M2 of THG. HPLC/MS/MS
analysis was performed on Md at collision energies of 10, 20, and
30 eV, and daughter ions of [M + H]+ at m/z 501 were acquired
(Figure 6). Main ions observed were at m/z 501 [M + H]+, 307
[M - glucuronide - H2O]+, m/z 289 [M - glucuronide - 2H2O]+,
and 263. All major fragments of Md were 4 u lower than those of
M2 of THG, most likely due to the presence of an alkyne moiety
at C-20 and C-21. Based on these data, it was suggested that Md
could be a direct analogue of the M2 with hydroxyl and glu-
curonide moieties linked to C-18.
Both gestrinone and its metabolite (Md) were characterized
by NMR and chemical shifts (1H and 13C) are presented at Table
1. The 1D 1H NMR spectrum of gestrinone had one triplet
integrating for three protons that represented the protons at C-18a.
This triplet was no longer present in the 1D 1H NMR of Md, and
3170 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

ketone at C-3 and C-17, reduction of the C-4-5-ene function, and
phase II conjugation (glucuronidation and sulfation).17 Structures
of THG and gestrinone are unique due to the presence of four
conjugated double bonds in A, B, and C rings, leading to highly
planar molecules with 3D structures distinct from known AAS
(e.g., testosterone). Only trenbolone (estra-4,9,11-trien-17â-ol-3-
one; Figure S-5 Supporting Information), a veterinary product
banned by the WADA, shows structural similarities with its
conjugated system. However, trenbolone is devoid of ethyl or
ethynyl side chains at C-13 and C-17. When it was administered
to human, no metabolism was reported on the A, B, and C rings.19
Main metabolites excreted in urine were formed via isomerization
at the secondary C-17 (D ring) and glucuronidation. Steroids with
ethyl, ethylnyl, or both side chains attached to C-13 or C-17 but
with saturated B and C cycles have been studied in vivo in humans
(structures shown at Figure S-5 Supporting Information). As for
THG and gestrinone, some biotransformations observed involved
oxidations at C-16 (norgestrel)20 or on ethyl side chains (ethyl-
estrenol and norethandrolone)21,22 at the terminal carbon. How-
ever, it is well documented that secondary carbons, such as C-18
of our two analytes, are more prone to P450-mediated oxidation
than primary carbons, such as C-18a.23-25
When gestrinone was dosed orally in human, parent and
urinary metabolites were observed in the hydrolyzed fraction of
urine samples.13 These were identified as glucuronide adducts by
enzymatic hydrolysis using â-glucuronidase. Kim et al. have
characterized unconjugated forms of two unknown urinary me-
tabolites of gestrinone by HPLC/MS/MS using a protocol similar
to what was mentioned above. Reported mass spectra for one of
the in vivo metabolites were in good agreement with those
generated for the unconjugated form of Md (Md′). By HPLC/
MS in ESI positive ion mode, both the in vivo metabolite and Md′
have [M + H]+ at m/z 325 with a major fragment at m/z 263.
HPLC/MS/MS on the ion at m/z 263 led to similar fragmentation
patterns with major daughter ions at m/z 235, 217, 205, 167, and
153.13 Furthermore, the analysis of its TMS-enol TMS-ether
derivative by GC/MS, reported in the literature, led to formation
of fragments at m/z 423 ([M - 117 u]+.) and 117 (e.g., [CH(OSi-
(CH3)3)CH3]+.), characteristic for an ethyl side chain containing
a TMS-ether moiety.13 Based on these observations, the metabolite
of gestrinone identified in vitro in human hepatocyte incubations
(Md) could correspond to this unidentified urinary metabolite
reported in the literature following a per os dose of gestrinone in
humans. In addition, the glucuroconjugated form of gestrinone
observed by Kim et al. is consistent with Me formed in human
hepatocyte incubations of gestrinone. A co-injection of authentic
Figure 8. Proposed metabolic pathways leading to formation of
main in vitro metabolites of gestrinone in fresh human hepatocytes.
tocyte incubations, isolated metabolites were treated with â-glu-
curonidase to hydrolyze M1, M2, and Md to their unconjugated
forms (M1′, M2′, and Md′). None of the aglycon metabolites were
detected in the human hepatocyte incubations of THG and
gestrinone, suggesting a fast and efficient glucuronidation of
oxidative metabolites with hydroxyl moieties at C-18 or C16ꢀ in
human hepatocytes. By HPLC/MS/MS, the main fragments of
Md′ and M2′ (analogous metabolites with an alcohol at C-18) were
62 u lower than the parent [M + H]+ (from m/z 325 to 263 for
Md′ and from m/z 329 to 267 for M2′) (Figure S-4 Supporting
Information). Minimal fragmentation could be observed for these
unconjugated metabolites, as compared to their conjugated
analogues. It may be possible that the presence of glucuronide
adducts would affect their fragmentation pattern. Further work
would be needed to elucidate the mechanism implicated in the
formation of ions at m/z 263 or 267 for Md′ and M2′, respectively.
In the case of M1′, its fragmentation pattern was similar that of
M1, with the exception of the neutral glucuronic acid loss. In the
presence of â-glucuronidase enzyme, glucuroconjugates of THG
(M4) and gestrinone (Me) were converted back to parent
compounds. This confirmed that glucuronidation, most probably
on the hydroxyl at C-17â, would be another biotransformation
implicated in the in vitro metabolism of THG and gestrinone
(Figures 7 and 8).
P450s are known to play an important role in the metabolism
of many xenobiotics in human. Considering that a primary route
of metabolism of THG and gestrinone involved an oxidative
process, incubations in microsomes expressing mainly P450
enzymes were conducted. Formation of Md′ (for gestrinone), M1′,
and M2′ (for THG) were observed in the presence of rP450 3A4,
and to a lesser extent with rP450 3A5 and 3A7. Since P450 3A is
the most abundant subfamily in human liver, it is likely to be an
important player in the in vivo metabolism of THG and gestrino-
ne.18
(19) de Boer, D.; Gainza Bernal, M. E.; van Ooyen, R. D.; Maes, R. A. A. Biol.
Mass Spectrom. 1991, 20, 459-466.
(20) Sisenwine, S. F.; Kimmel, H. B.; Liu, A. L.; Ruelius, H. W. Drug Metab. Dispos.
1975, 3, 180-188.
(21) Schanzer, W.; Donike, M. Anal. Chim. Acta 1993, 275, 23-48.
(22) Ward, R. J.; Lawson, A. M.; Shackleton, C. H. J. Steroid Biochem. 1977, 8,
1057-1063.
(23) Ortiz de Montellano, P. R. In Handbook of drug metabolism; Woolf, T. F.,
Ed.; Marcel Dekker: New York, 1999; pp 109-130.
(24) Smith, D. A.; Jones, B. C.; Walker, D. K. Med. Res. Rev. 1996, 16, 243-
266.
(25) Leblanc, Y.; Roy, P.; Wang, Z.; Li, C. S.; Chauret, N.; Nicoll-Griffith, D. A.;
Silva, J. M.; Aubin, Y.; Yergey, J. A.; Chan, C. C.; Riendeau, D.; Brideau, C.;
Gordon, R.; Xu, L.; Webb, J.; Visco, D. M.; Prasit, P. Bioorg. Med. Chem.
Lett. 2002, 12, 3317-3320.
Major biotransformations reported for most AAS are usually
hydroxylation at C-2, C-6, or C-16, oxidoreduction of alcohol or
(18) Shimada, T.; Yamazaki, H.; Mimura, M.; Inri, Y.; Guengerich, F. P. J.
Pharmacol. Exp. Ther. 1994, 270, 414-423.
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 3171

standards of Md and Me and urine samples from the excretion
study would be needed to further confirm this hypothesis.
Doping control laboratories currently use GC/MS to evaluate
the presence of urinary markers of banned AAS. As a generic
method, TMS derivatization is conducted on isolated urinary
steroids prior to GC/MS analysis. Chemical derivatization meth-
ods reported in the literature were tested to synthese TMS-ether
TMS-enol gestrinone.13,26 None of them led to the formation of
the desired derivative based on GC/MS analysis. The main side
product formed had an M+. at m/z 448, 4 u lower than that
expected for TMS-ether TMS-enol gestrinone. Similar difficulties
were reported for the synthesis and analysis of TMS-enol TMS-
ether derivatives of THG and trenbolone, both containing a
conjugated system.1,19 The lack of stability encountered was most
probably due to chemical or thermal degradation during the
derivatization procedure or in the inlet of the GC/MS. This would
have an important impact on the detection of urinary markers of
THG in a doping control perspective. Therefore, detection of THG
metabolites would be more suitable by HPLC using a combination
of PDA, fluorescence, and MS or MS/MS detectors.
the major metabolites were biosynthesized, isolated, and charac-
terized by HPLC coupled to PDA, fluorescence, MS, and MS/
MS detectors and NMR for structure elucidation. Oxidation at C-18
and most likely at C-16ꢀ and glucuronidation were the main
biotransformations observed. In vitro metabolism information
generated on THG could have a significant impact on the
understanding and the prediction of in vivo metabolism of this
unapproved xenobiotic, considering that the main metabolite of
gestrinone identified in vitro was consistent with a major urinary
metabolite reported as unknown in human.
ACKNOWLEDGMENT
The authors thank Robert Houle for isolation of human
hepatocytes and suggestions regarding large-scale incubation
conditions, and Dr. Marc Bilodeau for the procurement of fresh
human tissues. We also acknowledge Dr. Deborah Nicoll-Griffith
for her support as well as helpful discussions during the prepara-
tion of the manuscript.
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
CONCLUSION
Using fresh human hepatocytes, the in vitro metabolism of
Received for review January 25, 2005. Accepted March 9,
2005.
THG and gestrinone was studied, and microgram quantities of
(26) Ayotte, C.; Goudreault, D.; Charlebois, A., J. Chromatogr., B 1996, 687,
3-25.
AC050150Y
3172 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005