Use of LC-MS/MS for the Open Detection of Steroid MetabolitesConjugated with Glucuronic AcidAndreu Fabregat,† Oscar J. Pozo,*,† Josep Marcos,†,‡ Jordi Segura,†,‡ and Rosa Ventura†,‡
†Bioanalysis Research Group, IMIM, Hospital del Mar, Doctor Aiguader 88, 08003 Barcelona, Spain‡Department of Experimental and Health Sciences, Universitat Pompeu Fabra, Doctor Aiguader 88, 08003 Barcelona, Spain
*S Supporting Information
ABSTRACT: In humans, conjugation with glucuronic acid is the mostimportant phase II metabolic reaction of steroidal compounds. Glucurono-conjugated metabolites have been conventionally studied by using β-glucuronidase enzymes to release the phase I metabolites. It is well-knownthat hydrolysis with β-glucuronidase presents some limitations that may resultin the underestimation of some conjugates. The aim of the present work was todevelop and to evaluate liquid chromatography-tandem mass spectrometry(LC-MS/MS) scan methods for the open detection of steroid glucuronides inurine samples. The mass spectrometric behavior of thirteen representativesteroid glucuronides, used as model compounds, was studied. Characteristicionization and collision induced dissociation behaviors were observeddepending on the steroid glucuronide structure. Neutral loss (NL of 176,194, 211, and 229 Da) and precursor ion (PI of m/z 141, 159, and 177, inpositive mode and m/z 75, 85, and 113, in negative mode) scan methods were evaluated. The NL scan method was chosen forthe open detection of glucuronoconjugated steroids due to its sensitivity and the structural information provided by this method.The application of the NL scan method to urine samples collected after testosterone (T) undecanoate administration revealedthe presence of two T metabolites which remain conjugated as glucuronides after an enzymatic hydrolysis of the urine. 3α,6β-Dihydroxy-5α-androstan-17-one (6β-hydroxyandrosterone) glucuronide and 3α,6β-dihydroxy-5β-androstan-17-one (6β-hydroxyetiocholanolone) glucuronide were established as the structures for these metabolites, by comparing the structure ofthe steroids released after chemical hydrolysis with reference materials. An increase of 50−300-fold of these metabolites after oraladministration of T undecanoate was observed, proving that their determination can be useful in the doping control field.Moreover, these results exemplify that significant information might be missed, unless direct methods for the determination ofsteroid glucuronides are employed.
In humans, endogenous compounds and xenobiotics suffer avariety of biotransformations which produce metabolites
with different chemical structures that may differ inphysiological, pharmacological, or toxicological propertiescompared to their parent compounds. In drug testing, thedetection of metabolites in biological samples is normally usedas an indication of the exposure to the parent drugs. For thesereasons, the identification and characterization of the structuresand properties of metabolites formed in vivo is important.Conjugation with glucuronic acid is one of the most
important phase II metabolic reactions that occurs during thebiotransformation of different endogenous compounds anddrugs of steroidal nature.1,2 Glucuronidation is catalyzed byuridine diphosphoglucuronosyl-transferases and uses uridine-5′-diphosphoglucuronic acid (UDP-glucuronic acid) as thecofactor. These enzymes catalyze the transfer of glucuronic acidfrom the cofactor UDP-glucuronic acid to the steroid structure,typically to an hydroxyl group.2 Glucuronidation is generallyconsidered a detoxification pathway because the glucurono-conjugates are less active and more water-soluble than theparent compounds, thus facilitating their elimination from the
body. However, for some compounds (17β-hydroxyestrogens,testosterone, and dihydrotestosterone), higher toxicity has beenreported for the glucuronide conjugates compared to thecorresponding parent compounds.3
Studies on phase II steroid metabolism have been tradition-ally performed using gas chromatography/mass spectrometry(GC/MS) or liquid chromatography-tandem mass spectrome-try (LC-MS/MS), after hydrolysis of the samples to release thephase I metabolites.1,2 Glucuronoconjugates have beenelucidated using hydrolysis procedures with enzymes with β-glucuronidase activity and, therefore, only conjugates withglucuronic acid hydrolyzable in the conditions employed havebeen systematically studied. It is well-known that the efficiencyof β-glucuronidase enzymes depends on the structure of theglucuronide.4 Moreover, different factors may affect thehydrolysis efficiency. The enzymatic hydrolysis of some steroidsmay be incomplete due to the presence of enzyme inhibitors in
Received: January 21, 2013Accepted: April 15, 2013Published: April 15, 2013
the urine matrix.5 Differences in the hydrolysis efficiency withdifferent β-glucuronidase preparations have been documentedfor steroids and other drugs.5−8 The incubation time,temperature, and pH conditions required to complete thehydrolysis of specific glucuronoconjugates depend on thecompound.4,9 The presence of contaminants in the prepara-tions may convert one steroid into another under some specificconditions.10,11 Due to these limitations, different glucurono-conjugated metabolites could be underestimated by themethods currently applied in steroid metabolism studies.Chemical hydrolysis approaches are also available to
hydrolyze glucuronoconjugates.11,12 However, they also presentsome limitations such as degradation of some analytes and anincrease in matrix interferences.Nowadays, LC-MS/MS offers the possibility to directly
analyze steroid conjugates without previous hydrolysis. Severalmethods have been described in the literature for thedetermination of the intact steroid conjugates in biologicalmatrixes.13−18 In addition, the use of LC-MS/MS has resultedin the identification of previously unreported conjugatedmetabolites for some exogenous and endogenous anabolicsteroids, such as some sulphates19−22 or conjugates withcysteine and N-acetylcysteine.23−25 The special characteristicsof triple quadrupole instruments, allowing for the detection ofcompounds with a common structure using neutral loss (NL)and precursor ion (PI) scan methods, play an important rolefor the detection and elucidation of new steroid metabolitesusing LC-MS/MS.LC-MS/MS behavior of glucuronoconjugated metabolites of
some xenobiotics has been described in the literature, and acommon fragmentation pattern has been revealed.16,17,26,27
Therefore, the open detection of the phase II metabolites inhuman urine can be realized by developing NL or PI scanmethods specific for glucuronide conjugatesThe purpose of this study was to develop LC-MS/MS scan
methods for the open detection of metabolites of steroidsconjugated with glucuronic acid. The ionization and collisioninduced dissociation behavior of thirteen model steroidglucuronides (Supporting Information S-1) was comprehen-sively evaluated, and NL and PI scan methods were developed.The usefulness of these methods for the study of steroidmetabolism was evaluated by studying the testosterone (T)metabolism.
■ EXPERIMENTAL SECTIONChemicals and Reagents. 4-Androsten-17β-ol-3-one-17-
glucuronide (TG), 5α-androstane-3β,17β-diol-17-glucuronide(5αββ-DiolG), 5α-androstane-3α,17β-diol-17-glucuronide(5ααβ-DiolG), 5β-androstan-3α-ol-17-one-3-glucuronide(EtG), and 5α-androstan-3α-ol-17-one-3-glucuronide (AndG)were obtained from Sigma-Aldrich (St Louis, MO, USA). 4-Androsten-17α-ol-3-one-17-glucuronide (EpiG), 5β-androst-1-ene-17β-ol-3-one-17-glucuronide (1-TG), 5α-estran-3α-ol-17-one-3-glucuronide (NorAndG), 5β-estran-3α-ol-17-one-3-glu-curonide (NorEtG), 3α,6β-dihydroxy-5α-androstan-17-one(6β-OH-And), 3α,6β-dihydroxy-5β-androstan-17-one (6β-OH-Et), 5β-androstan-7α,17α-dimethyl-3α17β-diol-3-glucuro-nide (BolasG), 3α,11β-dihydroxy-5α-androstan-17-one (11β-OH-And), and 3α,11β-dihydroxy-5β-androstan-17-one (11β-OH-Et) were obtained from NMI (Pymble, Australia). 4,6-Androstadien-17β-ol-3-one-17-glucuronide (6-TG), 5β-andro-stan-3α-ol-11,17-dione-3-glucuronide (11-KetoEtG), 3α,16β-dihydroxy-5α-androstan-17-one (16β-OH-And), and 3α,16β-
dihydroxy-5β-androstan-17-one (16β-OH-Et) were purchasedfrom Steraloids (Newport, USA). 5α-Androstan-17β-ol-3-one-17-glucuronide (DHTG) was obtained from fountain limited(Newport, USA). 5α-Androstan-3α-ol-17-one-β-glucuronide-2,2,4,4-2H4 (used as internal standard for LC-MS) was obtainedfrom Orpachem (Saint-Beuzire, France). The β-glucuronidasepreparation (from Escherichia coli type K12) was purchasedfrom Roche Diagnostics (Mannheim, Germany).Citric acid, sodium bicarbonate, urea, calcium chloride,
sodium chloride dihydrate, iron II sulfate heptahydrate,magnesium sulfate hepathydrate, sodium sulfate decahydrate,potassium dihydrogen phosphate, ammonium chloride, ana-lytical grade potassium carbonate, potassium hydroxide,disodium hydrogen phosphate, sodium hydrogen phosphate,tert-butyl-methyl ether, acetonitrile (LC gradient grade),methanol (LC gradient grade), formic acid, ammonium formate(LC/MS grade), and cyclohexane were obtained from Merck(Darmstadt, Germany). Lactic acid, creatinine, 2-mercaptoe-thanol, sodium metaperiodate, and urea were purchased fromSigma-Aldrich (St Louis, MO, USA).The derivatization reagent N-methyl-N-trimethylsilyl-trifluor-
oacetamide (MSTFA) was from Karl Bucher Chemische FabrikGmbH (Waldstetten, Germany). Diethyl ether was purchasedfrom Fisher scientific (Leicestershire, UK).Milli Q water was obtained using a Milli-Q purification
system (Millipore Iberica, Barcelona, Spain). The Sep-Pak VacRC C18 (500 mg) cartridges were purchased from Waters(Milford, Massachusetts, USA).
LC-MS/MS Instrumental Conditions. The study wascarried out using a triple quadrupole (XEVO TQMS) massspectrometer equipped with an orthogonal Z-spray-electrosprayionization source (ESI) interfaced to an Acquity UPLC systemfor the chromatographic separation (all from Waters Associates,Milford, Massachusetts, USA). Drying gas as well as nebulizinggas was nitrogen. The desolvation gas flow was set toapproximately 1200 L/h, and the cone gas flow was 50 L/h.A cone voltage of 25 V, and a capillary voltage of 3.0 kV wereused in both positive and negative ionization mode. Thenitrogen desolvation temperature was set to 450 °C, and thesource temperature was 120 °C.The LC separation was performed using an Acquity UPLC
BEH130 C18 column (2.1 × 100 mm i.d., 1.7 μm) (WatersAssociates), at a flow rate of 300 μL/min. Water and methanolboth with formic acid (0.01% v/v) and ammonium formate (1mM) were selected as mobile phase solvents. A gradientprogram was used; the percentage of organic solvent waslinearly changed as follows: 0 min, 30%; 6 min, 30%; 15 min,65%; 15.5 min, 95%; 17.5 min, 95%; 18 min, 30%; 20 min,30%.The study of the ionization of the steroid glucuronides was
performed by scanning the m/z range from 250 to 600 in bothpositive and negative modes. The fragmentation of each modelcompound was studied by using product ion scan methods inboth positive and negative modes at three collision energies(20, 30, and 40 eV).NL and PI scan methods tested for the open detection of
steroid glucuronides are summarized in Table 1. In order todetermine testosterone metabolites conjugated with glucuronicacid, a selected reaction monitoring (SRM) qualitative methodwas set up. The mass spectrometric conditions are described inTable 2.
GC/MS Instrumental Conditions. GC/MS analysis wascarried out on a 6890N gas chromatograph coupled with a
5975 mass spectrometer (Agilent Technologies, Palo Alto,California, USA). The steroids were separated on a HP-Ultra1cross-linked methyl-silicone column, 16.5 m × 0.2 mm i.d., filmthickness of 0.11 μm (J&W Scientific, Folsom, California,USA). Helium was used as the carrier gas at a constant pressureof 5 psi. A 2 μL aliquot of the final derivatized extract (seebelow) was injected into the system operated in splitless mode(valve opened at 2 min). The GC temperature was ramped asfollows: initial 180 °C, increased to 230 °C at 3 °C min−1,thereafter increased to 310 °C at 40 °C min−1, and held for 3min. The injector and transfer line were kept at 280 °C.Analytes were determined by scanning the m/z range between50 and 700.Artificial Urine Preparation. The artificial urine was
prepared on the basis of a protocol described elsewhere:28 0.1 gof lactic acid, 0.4 g of citric acid, 2.1 g of sodium bicarbonate, 10g of urea, 0.07 g of uric acid, 0.8 g of creatinine, 0.37 g ofcalcium chloride·2H2O, 5.2 g of sodium chloride, 0.0012 g ofiron II sulfate·7H2O, 0.49 g of magnesium sulfate·7H2O, 3.2 gof sodium sulfate·10H2O, 0.95 g of potassium dihyrogenphosphate, 1.2 g of dipotassium hydrogen phosphate, and 1.3 gof ammonium chloride were dissolved in 1 L of ultrapure water.The artificial urine presented a pH of 6.4 and a specific gravityof 1.0132.Sample Preparation. The urine samples were treated as
follows. After the addition of 50 μL of an internal standardsolution (androsterone-d4-glucuronide at 10 μg/mL), urinesamples (5 mL) were passed through a C18 cartridge,previously conditioned with 2 mL of methanol and 2 mL ofwater. The column was then washed with 2 mL of water, and
the analytes were eluted with 2 mL of methanol. The eluate wasevaporated to dryness under a nitrogen stream in a water bathat 50 °C. Then, the dry extract was reconstituted with 1 mL ofsodium phosphate buffer (1 M, pH 7), added with 30 μL of β-glucuronidase solution (type K12, E. coli), and hydrolyzed for2.5 h at 55 °C. After cooling to room temperature, 250 μL of a5% (w/v) potassium carbonate solution was added to thehydrolyzed sample to obtain a pH value of 9.5. Liquid−liquidextraction was performed twice by addition of 6 mL of tert-butyl methyl ether. Each extraction was then centrifugated(2200g, 5 min); the organic layers were discarded, and theaqueous phase was passed through a C18 column, previouslyconditioned with 2 mL of water and 2 mL of methanol. Thesample was washed with 2 mL of water, eluted with 2 mL ofmethanol, and evaporated to dryness. The residue wasdissolved in 150 μL of a mixture of water and acetonitrile(9:1, v/v), and 10 μL was directly injected into the LC-MS/MSsystem.For the target detection of the elucidated metabolites by the
SRM method (Table 2), a direct injection approach wasperformed. After the centrifugation (2200g, 5 min) of 300 μL ofurine, 150 μL was transferred into a LC vial and 10 μL of urinewas directly injected into the LC-MS/MS system.
LC Purification. In order to purify the target glucuronideconjugates, LC fractionation of urine extracts was conducted byemploying the gradient described in LC-MS/MS InstrumentalConditions. Fractions corresponding to the each targetglucuronide were isolated by collecting fractions of 50 scentered in the retention time of each glucuronide. In order toincrease the concentration of the analytes, this process wasperformed in quintuplicate, and the individual fractions werecombined, evaporated to dryness, and treated as described inChemical Hydrolysis of Steroid Glucuronides Resistant to β-Glucuronidase.
Chemical Hydrolysis of Steroid Glucuronides Resist-ant to β-Glucuronidase. For the hydrolysis of theglucuronides which remain in urine after β-glucuronidasetreatment, a chemical hydrolysis was performed on the basis ofa previous described procedure.12 The method was applied tothe fractions collected after LC purification. Briefly, the residueof the LC fraction was dissolved in 2 mL of an aqueous solutionof sodium metaperiodate (10% w/v), and the pH was adjustedto a value of 3 with an 1 M hydrochloric acid solution. Themixture was incubated at 55 °C for 40 min. Afterward, 1 mL of5 M sodium hydroxide solution was added, and the incubationcontinued for 15 min. After cooling, the mixture was extractedtwice with 6 mL of tert-butylmethyl-ether, and the organicphases were combined and evaporated. Finally, the dry residuewas derivatized with 50 μL of MSTFA/NH4I/2-mercaptoetha-nol (1000:2:6, v/w/v) for 1 h at 80 °C and analyzed by GC/MS.
Excretion Study Samples. Six healthy male volunteers(age, 27.2 ± 2.1 years; weight, 73.4 ± 4.0 kg; height, 1.75 ±0.03 m; mean ± standard deviation) were given a single oraldose of 120 mg of testosterone undecanoate (Androxon, three40-mg capsules; Organon). Ethical approval for the study wasgranted by the ethical committee of our institute (CEIC-IMASno. 94/467) and the Spanish Health Ministry (DGFPS no. 95/75). All of the subjects participating in the study gave theirwritten informed consent.Samples collected before T undecanoate administration and
at 0−4 h after administration were used for this study. Aliquotsof 50 mL of urine were frozen at −20 °C until analysis.
Table 1. Neutral Loss (NL) and Precursor Ion (PI) ScanMethods Developed for the Detection of Intact SteroidGlucuronides
Mass Spectrometric Behavior of Steroid Glucuro-nides. Ionization of Steroid Glucuronides. The massspectrometric behavior of steroid glucuronides was systemati-cally studied using thirteen representative compounds with theglucuronide group conjugated at different positions (Support-ing Information S-1). The same ionization pattern wasobserved for all the compounds in negative electrosprayionization with [M − H]− as the major ion (SupportingInformation S-2). Other minor ions were also observed for allcompounds, such as the [M − H + Na + HCOO]− ion. Inpositive mode, different behaviors were detected depending onthe structure, as it was previously reported for the freesteroids.29 Steroid glucuronides with a conjugated 3-ketofunction (e.g., TG) exhibited [M + H]+ as the major ion(Supporting Information S-2). In contrast, steroid glucuronideswith 17-keto function (e.g., EtG and AndG) or unconjugated 3-keto moiety (e.g., DHTG) showed the adduct [M + NH4]
+ asthe major ion (Supporting Information S-2). Other minor ionswere obtained for all compounds independently of its structure,like adducts [M + Na]+ and [M + K]+ (Supporting InformationS-2). Thus, the study of the ions formed in the ionization(either [M + H]+ or [M + NH4]
+) might be useful in order toextract structural information about the steroid molecule.Collision Induced Dissociation Behavior of Steroid
Glucuronides. Collision induced dissociation was studied inpositive and negative ionization modes (Supporting Informa-tion S-3). Characteristic fragmentation patterns were observedin both modes. In negative mode, ions at m/z 175, 157, 113, 85,and 75 coming from dehydrated glucuronide (gluc), [gluc −H]−, [gluc − H − H2O]
−, [gluc − H − H2O − CO2]−, [gluc −
H − H2O − CO2 − CO]−, and HOCH2CO2− (originated from
the glucuronide moiety), respectively, were observed in allcompounds. The [M − H − gluc]− ions were also observed inall compounds with the exception of AndG, EtG, and BolasG.The highest abundances were obtained for ions at m/z 113, 85,and 75 (relative abundances ranging from 40% to 100%),whereas ions at m/z 175 and 157 showed the lowest relativeabundances (Supporting Information S-3).In positive mode, the collision induced fragmentation pattern
of the steroid glucuronides was in agreement with the resultsobtained in the ionization analysis. In summary, steroidglucuronides with a conjugated 3-keto moiety, for which theformation of ion [M + H]+ was observed, showed preferentlythe ion coming from the neutral losses of 176 Dacorresponding to the dehydrated glucuronide. Besides,subsequent losses of water from [M + H − gluc]+ wereobserved to form [M + H − gluc − H2O]
+ and [M + H − gluc− 2H2O]
+. The most abundant ions observed for these steroidglucuronides were those generated after the loss of thedehydrated glucuronide moiety (Supporting Information S-3).Thus, the ions coming from the free steroid may contribute toextra structural information about the steroid conjugated withglucuronide moiety.On the other hand, for steroid glucuronides with a 17-keto
group or with an unconjugated 3-keto function, for whichionization resulted in the formation of [M + NH4]
+ as the most abundant ions. Finally, ionsat m/z 177, 159, and 141 coming from [gluc + H]+, [gluc + H− H2O]
+, and [gluc + H − 2H2O]+, respectively, were common
to all model compounds analyzed in positive mode (SupportingInformation S-3).
Open Screening Methods for AAS Conjugated withGlucuronic Acid. Several common NL related with theglucuronide structure were observed in the product ion scanspectra of the model compounds in both positive and negativeionization modes. However, due to the poor sensitivityobserved in negative mode for the [M − H − gluc]− ion(Supporting Information S-3), NL scan methods weredeveloped only in positive ionization mode (Table 1). Themost abundant product ions for steroid glucuronides with aconjugated 3-keto moiety were [M + H − gluc]+ and [M + H −gluc − H2O]
+ (Supporting Information S-3). Therefore, twoNLs of 176 and 194 Da were selected. In contrast, the mostabundant product ions observed for steroid glucuronides with a17-keto group or an unconjugated 3-keto moiety were [M +NH4 − NH3 − gluc − H2O]
+ and [M + NH4 − NH3 − gluc −2H2O]
+ (Supporting Information S-3). On the basis of thisfragmentation, two NL of 211 and 229 Da, corresponding tolosses of dehydrated glucuronic acid, water, and ammonia andto losses of dehydrated glucuronic acid, two molecules of water,and ammonia from the [M + NH4]
+ ion, respectively, wereselected.Several common product ions coming from the glucuronide
moiety were observed in the fragmentation of the modelcompounds in both positive and negative modes. In negativemode, all of the compounds showed product ions at m/z 113,85, and 75, and in positive mode, ions at m/z 177, 159, and 141were present for all steroid glucuronides. On the basis of theseobservations, two PI scan methods were developed (Table 1).The use of PI scan of m/z 113 has already been reported in theliterature, and it was successfully applied for the open detectionof new markers of 4-androstendione abuse.17
Evaluation of the Developed Methods. In order toevaluate the sensitivity and due to the endogenous character ofsome of the analytes, NL and PI scan methods were applied toartificial urine samples spiked with the steroid glucuronides atdifferent concentrations (10, 100, and 500 ng/mL). The limitsof detection (LOD), defined as the lowest concentrations witha value of the signal/noise ratio (S/N) of 3, obtained for eachmethod are summarized in Table 3.The PI scan in positive mode was shown to be more sensitive
than the PI scan in negative mode as summarized in Table 3.The lowest LOD were achieved when using the NL scanmethod, with overall values not greater than 30 ng/mL exceptfor BolasG at 100 ng/mL) (Table 3). This represents a bettersensitivity than those LOD reported for similar methodsdeveloped for the open detection of free steroids afterenzymatic hydrolysis of urine using PI scanning.30 Besides,the use of the NL scan method provides additional structuralinformation as explained before. Those steroid glucuronideswith a 3-keto conjugated moiety presented the NL of 176 and194 Da, and the rest of the compounds showed the NL of 211and 229 Da. However, AndG and EtG can be detected as the[M + Na]+ or [M + H − nH2O]
+ species (in addition to the [M+ NH4]
+ that gave the NL of 211 and 229) and, therefore, thesecompounds were observed in both NL scan methods (176 and194 Da ; 211 and 229 Da).The sensitivity of the PI and NL method was evaluated by
studying the TIC chromatogram. Using the TIC, the sensitivityof two peaks closely eluting depends on the relative abundanceamong them. Therefore, those analytes showing similarintensities (e.g 5ααβ-DiolG and NorAndG, RTs of 14.2 and
14.0 min, respectively) could be chromatographically resolved,and the LOD could be easily calculated. Nevertheless, the LODfor those compounds with very different sensitivities, e.g., TG(RT of 11.6 min) which is 4-fold more sensitive than theclosely eluting analytes 5αββ-DiolG and 11-KetoEtG (RTs of11.7 and 11.3 min, respectively), are more difficult to evaluate.
The peak observed for the TG hampers the detection of 5αββ-DiolG and 11-KetoEtG at the same concentration level.This problem can be minimized using NL methods. The fact
that NL methods are based on structural features other than theglucuronide moiety itself (see Open Screening Methods forAAS Conjugated with Glucuronic Acid) implies that closely
Figure 1. Results obtained after the application of the NL scan method to urine samples collected before (a, c, e, and g) and after TU administration(b, d, f, and h) at 176 Da (a and b); 194 Da (c and d); 211 Da (e and f); 229 (g and h).
eluting compounds with different structures can be detected byseparated NL functions. This is the case of TG where thepresence of the conjugated 3-keto moiety gives, predominantly,the NL of 176 and 194 Da, and the closely eluting compounds5αββ-DiolG and 11-KetoG were detected by the NL of 211and 294 Da due to the presence of 17-keto moiety (Table 3).Therefore, in contrast to the PI method, all three compoundscould be separately evaluated. Therefore, the NL method wasselected as the optimal method for the open detection ofsteroid metabolites conjugated with glucuronic acid.Application of Open Scan Methods to the Inves-
tigation of T Metabolism. Detection of T GlucuronideMetabolites. The developed NL scan methods were used toinvestigate the metabolism of T in humans. T is the mostdetected anabolic androgenic steroid in doping control tests,31
and its metabolism in humans has been extensively studied.1 Atpresent, its main metabolites are included in the steroid profileused in sports drug testing to indicate an exogenous Tadministration.32−34
The NL methods for the open detection of glucuronoconju-gated metabolites were applied to urine samples collectedbefore and after T undecanoate administration, previouslysubjected to hydrolysis with β-glucuronidase enzymes. There-fore, using this approach, only those metabolites conjugatedwith glucuronic acid but resistant to enzymatic hydrolysiswould be detected. As illustrated in Figure 1, the four acquiredNL signals showed an increase in two peaks (G1 and G2) afterT undecanoate administration. For the NL of 211 and 229 Da,these peaks had a precursor ion at m/z of 500 (Figure 1f,h)whereas, for the NL of 176 and 194 Da, the precursor ionswere, respectively, at m/z 505,465, and 447 (Figure 1b,d).Despite the different precursor ions observed for each NL,these peaks have the same retention time and peak shape.Therefore, they could be attributed to different adducts formedby G1 and G2.The full scan spectra were studied in order to obtain more
information about the formed adducts for both metabolites. G1and G2 showed similar full scan spectra (SupportingInformation S-4). The most abundant ion, which can beassociated with [M + NH4]
+, was at m/z 500. Ions formed aftersequential losses of water (m/z 465 and 447) from [M + H]+
were present. Additionally, other adducts were formed,including [M + Na]+ (m/z 505) and [M + K]+ (m/z 521).Only a small difference was observed between both spectra:whereas G1 showed the presence of [M + H]+ at m/z 483, thespectrum of G2 exhibited the ion at m/z 482 corresponding to[M + NH4 − H2O]
+ (Supporting Information S-4).The ions present in the full scan spectra are in agreement
with the different adducts exhibited for G1 and G2 in the NLmethods (Figure 1). Thus, the ion at m/z 500 obtained in theNL of 211 and 229 Da is the [M + NH4]
+ adduct which is ableto generate these neutral losses. According to the observationsin the NL for AndG and EtG, G1 and G2 can be detected as[M + Na]+ (m/z 505) or [M +H − nH2O]
+ (m/z 465, 447)producing NL of 176 and 194 Da, respectively. These weresimilar to that obtained for model compounds without aconjugated keto moiety (see Collision Induced DissociationBehavior of Steroid Glucuronides); data confirmed that thepeaks observed in the NL experiments (Figure 1) wereoriginated from the formation of the different adduct ions ofG1 and G2. In summary, the data suggested a molecular weightof 482 Da for both G1 and G2.
Characterization of T Glucuronide Metabolites. AlthoughG1 and G2 have a molecular weight of 482 Da, an ion at m/z500 is observed when analyzing by LC-MS/MS. The presenceof the ion at m/z 500 is related to the formation of the adduct[M + NH4]
+, typical for steroids with a keto moiety at C17, forexample, EtG and AndG15 which have a MW of 466 Da. Sincethis represents a 16 Da difference when compared to the MWof G1 and G2, hydroxy-androsterone or hydroxy-etiocholano-lone glucuronide can be postulated as feasible structures for G1and G2.In order to have more information about G1 and G2, the
product ion scans of m/z 500 were acquired. Similar spectrawere observed for both compounds (Supporting Information S-4). This fact was previously observed for stereoisomericanabolic steroids35−37 and, therefore, supports a stereoisomerrelation between G1 and G2.The product ion spectra showed ions coming from the losses
of water (m/z 465 and 447) and ions originated from the lossof ammonia and glucuronide moiety and the subsequent lossesof water (m/z 289, 271, 253). Analogous to the full scan results,G2 showed an ion at m/z 482 [M + NH4 − H2O]
+ whereas G1presented an ion at m/z 483 [M + NH4 − NH3]
+. The numberof losses of water is in agreement with the number of oxygenatoms present in the molecule for unconjugated steroids.38
Thus, the losses of water observed from the ion produced afterlosing ammonia and the glucuronide moiety suggested threeoxygen-containing functional groups for the unconjugatedsteroids of G1 and G2. This fact is in agreement with ahydroxy-androsterone or hydroxy-etiocholanolone structureproposed for G1 and G2An ion at m/z 81 was also present, and it was the base peak
at high collision energies (data not shown). This ion was notpreviously observed neither in the fragmentation of AAS liketestosterone, epitestosterone, androsterone, or etiocholanoloneconjugated with glucuronide nor in the fragmentation of freeAAS. Therefore, the formation of this ion seems to be related tothe extra hydroxyl group and the glucuronide.The comparison between the metabolite and a reference
material is necessary to confirm the chemical structure.Although several hydroxy-androsterone and hydroxy-etiochola-nolone standards were commercially available, their glucur-onide conjugates were not. Therefore, the steroids releasedafter hydrolyzing G1 and G2 were compared with the availablereference materials. To perform this comparison, G1 and G2were isolated by LC fractionation, followed by chemicalhydrolysis. Due to the low proton affinity expected for thereleased steroids, the use of GC/MS after derivatization wasneed. GC/MS full scan spectra and retention times for thesteroids released from G1 and G2 were compared with thoseobtained from commercially available compounds (6β-OH-Et,6β-OH-And, 16α-OH-Et, 16α-OH-And, 11β-OH-Et, and 11β-OH-And). As a result of this comparison, it was concluded that6β-OH-Et is the steroid released after chemical hydrolysis ofG1, and 6β-OH-And is released after chemical hydrolysis of G2(Supporting Information S-5).Nevertheless, G1 and G2 are conjugated with glucuronic acid
and both 6β-OH-Et and 6β-OH-And present two possible sitesfor glucuronoconjugation: the 3α and 6β hydroxyl groups.Therefore, a synthesis of the conjugated compounds would berequired in order to unequivocally confirm the completestructure of G1 and G2.
Hydrolysis Efficiency of T Glucuronide Metabolites. G1 andG2 (6β-OH-Et and 6β-OH-And glucuronides) were detected in
urine even after the enzymatic treatment, indicating that theseglucuronides are, at least partially, resistant to hydrolysis with β-glucuronidase. We conducted an evaluation of the resistance ofthese glucuronides to the action of the most commonly used β-glucuronidase. The percentages of the glucuronide metabolitesof 6β-OH-Et, 6β-OH-And, etiocholanolone, and androsteroneremaining in the urine after the treatment of urine sample withdifferent volumes of β-glucuronidase are summarized inSupporting Information S-6. After the addition of 30 μL of β-glucuronidase, almost no androsterone and etiocholanoloneglucuronides were detected, proving that they are almostcompletely hydrolyzed by the enzyme. In contrast, for G1 andG2, even after the addition of 500 μL of enzyme, 14% and 36%
of these glucuronides, respectively, were still present in theurine, indicating the incomplete cleavage of these compoundsunder conventional hydrolysis conditions.32−34,38,39
Analysis of T Glucuronide Metabolites after T Undeca-noate Administration. A SRM method (Table 2) for the directqualitative detection of G1 and G2 in urine was applied to theanalysis of urine samples collected from five volunteers beforeand 4 h after a single oral dose of T undecanoate. For thispurpose, urine samples before and after T undecanoateadministration were directly injected into the system asdescribed in Sample Preparation. The results showed theoccurrence of these metabolites in all preadministrationsamples (Figure 2a), confirming its endogenous origin. The
Figure 2. Chromatograms (LC-MS/MS) of the characteristic transitions of 6β-OH-Et and 6β-OH-And glucuronides (G1 and G2, respectively)obtained in urine extracts before administration (a) and after T undecanoate administration (b); (c) box plot diagrams of the signals obtained afteranalysis of samples before and after T undecanoate administration in six volunteers.
response obtained after T undecanoate administrationincreased by a factor of 100−300 for G1 and by a factor of50−200 for G2 (Figure 2b,c) in all the samples analyzed.6β-OH-Et and 6β-OH-And are currently evaluated as a part
of the steroid profile to indicate the exogenous administrationof endogenous steroids.32−34 However, contrary to our results,those investigations failed to detect the increase in the excretionof 6β-OH-Et and 6β-OH-And after the administration of T.Most likely, the reason for this discrepancy is the incompletehydrolysis of the conjugates under the conventional conditionsof analysis employed by those authors.34 In line with thisexplanation, Boccard et al.40 detected the increase of a putativehydroxy-androsterone or hydroxy-etiocholanolone glucuronideafter oral testosterone intake using an untargeted steroidomicapproach.Common routine analysis methods have been used to
calculate reference values for 6β-OH-Et and 6β-OH-And32.According to our hydrolysis experiments, these populationranges could be underestimated. In order to correctly evaluatethe reference ranges of these metabolites and its potential usefor the detection of an administration of exogenous T, thedirect quantitation of the intact steroid glucuronides, G1 andG2, would be required.
■ CONCLUSIONS
Steroid glucuronides show common ionization and collisioninduced dissociation behavior depending on the structure of thesteroid. On the basis of this common behavior, different NLand PI scan methods have been developed and evaluated forthe open detection of unknown steroid glucuronides. NL scanmethods were more sensitive than PI methods and providemore structural information. LOD of 30 ng/mL were obtainedusing the NL scan methods showing the sensitivity of themethod.The usefulness of the developed NL scan method has been
demonstrated by the detection T metabolites conjugated withglucuronic acid and resistant to hydrolysis with β-glucuroni-dase. Two glucuronides, G1 and G2, were detected. Thestructure of the steroids released after chemical hydrolysis ofG1 and G2 was confirmed as 3α,6β-dihydroxy-5α-androstan-17-one (6β-OH-And) and 3α,6β-dihydroxy-5β-androstan-17-one (6β-OH-Et), respectively. Due to their resistance tohydrolysis with β-glucuronidase, they have been under-estimated by the conventional conditions used for themeasurement of glucuronoconjugated metabolites in dopingcontrol. In contrast to previously published results, a significantincrease in G1 and G2 concentrations was observed after Tundecanoate administration when using LC-MS/MS for theirdirect detection. Additionally, the satisfactory sensitivity of theSRM method allowed for the detection of both G1 and G2after direct injection of the urine even at the basal levels.Therefore, the inclusion of these compounds in the steroidprofile can help in the detection of T misuse.In summary, the usefulness of LC-MS/MS for the open and
direct detection of steroid metabolites conjugated withglucuronic acid has been demonstrated. Our results regardingT metabolism also point out to the need of using this approachwhen investigating glucuronoconjugated metabolites in order tobypass the limitations of incomplete hydrolysis.
■ ASSOCIATED CONTENT*S Supporting InformationAdditional information as noted in text. This material isavailable free of charge via the Internet at http://pubs.acs.org.
■ ACKNOWLEDGMENTSGrants by World Antidoping Agency (project reference11A9RV), Ministerio de Economia y Competitividad (Gobiernode Espana) (project number DEP2012-35612), Generalitat deCatalunya (2009SGR00492 to the research team), and Institutode Salud Carlos III (OJP) are gratefully acknowledged. Thiswork has been carried out with the support of Consell Catala del’Esport. This work was supported by grants from Instituto deSalud Carlos III FEDER (CP10/00576).
■ REFERENCES(1) Schanzer, W. Clin. Chem. 1996, 42 (7), 1001−1020.(2) Van Eenoo, P.; Delbeke, F. T. J. Steroid Mol. Biol. 2006, 101 (4−5), 161−178.(3) Kuuranne, T. In Doping in sports; Thieme, D., Hemmersbach, P.,Eds.; Springer: Oslo, Norway, 2010; pp 66−73.(4) Mareck, U.; Geyer, H.; Opfermann, G.; Thevis, M.; Schanzer, W.J. Mass. Spectrom. 2008, 43, 877−891.(5) Graef, V.; Furuya, E.; Nishikaze, O. Clin. Chem. 1977, 23, 532−535.(6) Combie, J.; Blake, J. W.; Nugent, T. E.; Tobin, T. Clin. Chem.1982, 28, 83−86.(7) Romberg, R. W.; Lee, L. J. Anal. Toxicol. 1995, 19, 157−162.(8) Hackett, L. P.; Dusci, L. J.; Ilett, K. F.; Chiswell, G. M. Ther. DrugMonit. 2002, 24 (5), 652−657.(9) Gomes, R. L.; Meredith, W.; Snape, C. E.; Sephton, M. A. Anal.Bioanal. Chem. 2009, 393, 453−458.(10) Messeri, G.; Cugnetto, G.; Moneti, G.; Serio, M. J. SteroidBiochem. 1984, 20 (3), 793−796.(11) Gomes, R. L.; Meredith, W.; Snape, C. E.; Sephton, M. A. J.Pharm. Biomed. Anal. 2009, 49, 1133−1140.(12) Wade, A. P. Biochem. J. 1967, 103 (2), 19C−20C.(13) Bowers, L. D.; Sanaullah. J. Chromatogr., B 1996, 687, 61−68.(14) Hintikka, L.; Kuuranne, T.; Leinonen, A.; Thevis, M.; Schanzer,W.; Halket, J.; Cowan, D.; Grosse, J.; Hemmersbach, P.; Nielen, M.W.; Kostiainen, R. J. Mass. Spectrom. 2008, 43 (7), 965−973.(15) Pozo, O. J.; Van Enoo, P.; Van Thuyne, W.; Deventer, K.;Delbeke, F. T. J. Chromatogr., A 2008, 1183, 108−118.(16) Caron, P.; Audet-Walsh, E.; Lepine, J.; Belanger, A.;Guillemette, C. Anal. Chem. 2009, 81, 10143−10148.(17) Anizan, S.; Bichon, E.; Di Nardo, D.; Monteau, F.; Cesbron, N.;Antignac, J. P.; Le Bizec, B. Talanta 2011, 86, 186−194.(18) Bean, K. A.; Henion, J. D. J. Chromatogr., B: Sci. Appl. 1997, 690(1−2), 65−75.(19) Saudan, C.; Entenza, J. M.; Baume, M.; Mangin, P.; Saugy, M. J.Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2006, 844 (1), 168−174.(20) Gomez, C.; Pozo, O. J.; Geyer, H.; Marcos, J.; Thevis, M.;Schanzer, W.; Segura, J.; Ventura, R. J. Steroid Mol. Biol. 2012, 132 (3−5), 239−246.(21) Gomez, C.; Pozo, O. J.; Marcos, J.; Segura, J.; Ventura, R.Steroids 2013, 78, 44−52.(22) Strahm, E.; Kohler, I.; Rudaz, S.; Martel, S.; Carrupt, P. A.;Veuthey, J. L.; Saugy, M.; Saudan, C. J. Chromatogr., A 2008, 1196−1197, 153−160.
(23) Pozo, O. J.; Marcos, J.; Ventura, R.; Fabregat, A.; Segura, J. Anal.Bioanal. Chem. 2010, 398 (4), 1759−1770.(24) Fabregat, A.; Aristotelis, K.; Marcos, J.; Joglar, J.; Alfonso, I.;Segura, J.; Ventura, R.; Pozo, O. J. Steroids 2012, DOI: 10.1016/j.steroids.2012.11.017.(25) Pozo, O. J.; Gomez, C.; Marcos, J.; Segura, J.; Ventura, R. DrugTest. Anal. 2012, 4 (10), 786−797.(26) Gomez, C.; Segura, J.; Monfort, N.; Souminen, T.; Leinonen, A.;Vahermo, M.; Yli-Kauhaluoma, J.; Ventura, R. Anal. Bioanal. Chem.2010, 397 (7), 2903−2916.(27) Kuuranne, T.; Vahermo, M. J. Am. Soc. Mass. Spectrom. 2000, 11(8), 722−730.(28) Brooks, T.; Keevil, C. W. Lett. Appl. Microbiol. 1997, 24, 203−206.(29) Pozo, O. J.; Van Eenoo, P.; Deventer, K.; Delbeke, F. T. J. Mass.Spectrom. 2007, 42, 497−516.(30) Thevis, M.; Geyer, H.; Mareck, U.; Schanzer, W. J. Mass.Spectrom. 2005, 40 (7), 955−962.(31) Statistics released in 2010 at http://www.wada-ama.org/Documents/Resources/Testing-Figures/WADA_2010_Laboratory_Statistics_Report.pdf.(32) Van Renterghem, P.; Van Eenoo, P.; Van Thuyne, W.; Geirey,H.; Schanzer, W.; Delbeke, F. T. J. Chromatogr., B 2008, 876, 225−235.(33) Van Renterghem, P.; Van Eenoo, P.; Delbeke, F. T. Steroids2010, 75, 1047−1057.(34) Van Renterghem, P.; Van Eenoo, P.; Sottas, P. E.; Saugy, M.;Delbeke, F. T. Clin. Endocrinol. 2011, 75, 134−140.(35) Thevis, M.; Scanzer, W. Mass Spectrom. Rev. 2007, 26 (1), 79−107.(36) Guan, F.; Soma, L. R.; Luo, Y.; Uboh, C. E. J. Am. Soc. Mass.Spectrom. 2006, 17, 477−489.(37) Pozo, O. J.; Van Eenoo, P.; Deventer, K.; Grimalt, S.; Sancho, J.V.; Hernandez, F.; Delbeke, F. T. Rapid Commun. Mass Spectrom.2008, 22, 4009−4024.(38) Jimenez, C.; de la Torre, R.; Segura, J.; Ventura, R. RapidCommun. Mass Spectrom. 2006, 20, 858−864.(39) Delgadillo, M. A.; Garrostas, L.; Pozo, O. J.; Ventura, R.;Velasco, B.; Segura, J.; Marcos, J. J. Chromatogr., B 2012, 897, 85−89.(40) Boccard, J.; Badoud, F.; Grata, E.; Ouertani, S.; Hanafi, M.;Mazerolles, G.; Lanteri, P.; Veuthey, J.-L.; Saugy, M.; Rudaz, S. ForensicSci. Int. 2011, 213, 85−94.
