Negative Ion Tandem Mass Spectrometry for the Detection of

Zeqi Huang, Mehran Yazdanian, Mark A. Ator, Lisa D. Aimone, Kelli Zeigler, ..... J. Antoine, Christopher E. P. Goldring, Alvin J. L. Chia, Neil R...
1 downloads 0 Views 202KB Size
Download PDF
630

Chem. Res. Toxicol. 2005, 18, 630-638

Articles Negative Ion Tandem Mass Spectrometry for the Detection of Glutathione Conjugates Christine M. Dieckhaus,* Carmen L. Ferna´ndez-Metzler, Richard King, Paul H. Krolikowski, and Thomas A. Baillie Department of Drug Metabolism, WP75-100, Merck Research Laboratories, P.O. Box 4, Sumneytown Pike, West Point, Pennsylvania 19486 Received September 16, 2004

Characterization of S-linked conjugates of the endogenous tripeptide glutathione (γ-glutamylcysteinylglycine, GSH) represents a valuable indirect approach for the identification of chemically reactive, electrophilic intermediates formed during the metabolism of both foreign compounds and endogenous substances. In most cases, GSH adducts generated in vitro or excreted in the bile of animals are detected by the use of liquid chromatography-tandem mass spectrometry (LC-MS/MS), employing survey scans based on characteristic fragmentations of this class of conjugates. However, a limitation of current LC-MS/MS approaches, which typically employ electrospray ionization with analysis of positive ions, is that no single survey scan exhibits broad utility in the detection of unknown GSH adducts, since different structural classes of conjugate (aromatic, benzylic, aliphatic, thioester, etc.) behave differently upon collision-induced dissociation (CID) of the respective [M + H]+ parent ions. In the present study, we evaluated MS/MS in the negative ion mode as an alternative approach and report herein that the spectra obtained by CID of the [M - H]- ions of a number of representative GSH adducts, as well as GSH itself, are dominated by fragments originating from the glutathionyl moiety of the tripeptide. In particular, the anion at m/z 272, corresponding nominally to deprotonated γ-glutamyl-dehydroalanyl-glycine, was abundant in the negative ion spectra of free GSH and all GSH conjugates examined, suggesting that scanning for precursors of this ion may provide a generally applicable technique for the detection of adducts of unknown structure. The utility of this novel detection strategy was demonstrated in a series of in vitro and in vivo experiments where compounds known to undergo metabolic activation were examined for their propensity to form conjugates with GSH. In all cases, scanning for precursors of m/z 272 in the negative ion mode revealed the presence of the expected adducts and in some instances revealed additional conjugates that had not been reported previously. Positive ion MS/MS, on the other hand, was more useful than the corresponding negative ion scans in providing information on the molecular structure of GSH conjugates.

Introduction Chemically reactive, electrophilic metabolites have been implicated as mediators of the toxic response to a variety of drugs, environmental pollutants and endogenous products of oxidative stress, and are believed to act through covalent modification of tissue nucleophiles (1, 2). In many cases, these reactive intermediates are “soft” electrophiles, which may be captured, at least in part, by the endogenous tripeptide glutathione (γ-glutamylcysteinylglycine, GSH).1 Similar GSH “trapping” experiments can be performed in vitro, where test compounds are incubated with tissue preparations fortified with GSH and the cofactors required for metabolism. The resulting GSH conjugates, which are excreted into * Address correspondence to this author at Merck Research Laboratories, Department of Drug Metabolism, WP75-100, P.O. Box 4, Sumneytown Pike, West Point, PA 19486. E-mail: [email protected].

bile, are significant not only because they generally represent products of detoxification, but because they also provide valuable indirect information on the structures of the reactive metabolites from which they were derived. This knowledge is particularly important in mechanistic studies of metabolic activation, as well as in support of drug discovery and lead optimization efforts in the pharmaceutical industry (3). Hence, there is a need for sensitive and specific analytical methodology with which to detect and identify GSH adducts in biological samples, and in recent years this has been accomplished mainly 1 Abbreviations: 4-HNE, (()-4-Hydroxy-(2E)-nonenal; DACA-GSH, γ-glutamyl-S-{1-[4-(dimethylamino)phenyl]-3-hydroxypropyl}cysteinylglycine; D-SG, diclofenac-S-acyl-GSH; TMP-GSH, γ-glutamylS-[2,6-diamino-5-(3,4,5-trimethoxybenzyl)-pyrimidin-4-yl]cysteinylglycine; 4-HNE-GSH, γ-glutamyl-S-[2-hydroxy-1-(2-hydroxyethyl) heptyl]cysteinylglycine; HMBC, 1H-13C 2D gradient heteronuclear multiple bond correlation; LC-MS/MS, liquid chromatographytandem mass spectrometry; CID, collision-induced dissociation.

10.1021/tx049741u CCC: $30.25 © 2005 American Chemical Society Published on Web 03/09/2005

Negative Ion MS for Detection of GSH Conjugates

by approaches based on liquid chromatography - tandem mass spectrometry (LC-MS/MS) employing electrospray ionization and analysis of positive ions (3, 4). Traditionally, the most widely used MS/MS survey scan has been constant neutral loss scanning for 129 Da, which is based on loss from the [M + H]+ ion of the elements of pyroglutamic acid, derived from the glutamate residue of the conjugate (4). However, it has become apparent that not all classes of GSH adduct afford this neutral loss upon CID; aliphatic and benzylic thioether conjugates may eliminate the elements of GSH (307 Da) as a neutral, and/or yield the GSH2+ product ion (m/z 308), while thioester conjugates typically fragment by loss of glutamic acid (147 Da) from the [M + H]+ species (4, 5). Such conjugates may escape detection by 129 Da constant neutral loss scanning. Moreover, GSH adducts frequently yield doubly charged [M + 2H]2+ ions under positive ion electrospray conditions, CID of which leads to a family of singly protonated product ions, rather than the elimination of neutral species. In an attempt to develop a more broadly applicable MS/MS survey scan for the detection of unknown GSH conjugates belonging to different structural classes, we explored the fragmentation behavior of a number of representative examples under negative ion conditions. Unlike the situation in the positive ion mode, the negative ion MS/MS spectra of all of the conjugates examined proved to be closely similar to one another, and mirrored the negative ion spectrum of GSH itself. These findings suggest that the use of negative ions may be of value in the development of unbiased survey scans for GSH adducts in complex biological matrixes.

Experimental Procedures Chemicals, Reagents, and Instrumentation. All reagents were purchased from Sigma (St. Louis, MO) or Aldrich (Milwaukee, WI) and were of the highest grade available unless otherwise noted. HPLC-grade solvents were purchased from Fisher-Scientific (Pittsburgh, PA). (()-4-Hydroxy-2E-nonenal (4-HNE) was purchased from Caymen Chemical (Ann Arbor, MI). Rat liver microsomes were purchased from Xenotech (R1000, Kansas City, KS). The preparative HPLC system consisted of a Waters 2700 Sample Manager, Waters 600 Quaternary Pumps, Waters 2996 Photodiode Array Detector and a Waters Fraction Collector III controlled by Waters Empower Software (Milford, MA). LC-MS/MS analysis was conducted on a Sciex API 3000 mass spectrometer (AB/MDS Sciex, Toronto) equipped with an electrospray ionization source, a HTC PAL Autosampler (Leap Technologies, NC) and Perkin-Elmer Series 200 Micro Pumps (Perkin-Elmer, Norwalk, CT). The Sciex API 3000 instrument parameters were optimized and set as follows: 0.2 Da step size, 5 ms pause, 2 s scan rate or 50 ms dwell, NEB ) 8, CUR ) 12, IS ) ( 4000, TEMP ) 450, EP ) ( 10, CAD ) 4 and CXP ) ( 15. Additional instrument settings are listed in Table 1. GSH fragmentation pathways were elucidated utilizing MSn on a classic LCQ (FinniganMAT, San Jose, CA) equipped with an electrospray ionization source and PerkinElmer Series 200 Micro Pumps (Perkin-Elmer, Norwalk, CT). The LCQ was run in negative ion mode and the source settings were as follows: spray voltage ) 5 kV; sheath gas flow rate ) 90; auxiliary gas flow rate ) 5; capillary voltage ) -46 V, capillary temperature ) 200 °C. All NMR experiments were performed at 25 °C on a 500 MHz spectrometer (Inova, Varian Inc., Palo Alto, CA) equipped with a 5 mm HCN Coldprobe (Varian Inc.). Synthesis. 1. TMP-GSH (γ-Glutamyl-S-[2,6-diamino-5(3,4,5-trimethoxybenzyl)pyrimidin-4-yl]cysteinylglycine). TMP-GSH was prepared according to the published

Chem. Res. Toxicol., Vol. 18, No. 4, 2005 631 Table 1. The AB/MDS Sciex API 3000 Settings for the Respective Scans Q1 (m/z)

Q3 (m/z) DPa FPb CEc

scan

polarity

positive ion full scan product ion, singly charged product ion, doubly charged precursor of m/z 130 neutral loss of 307 neutral loss of 129 negative ion full scan product ion, singly charged precursor of m/z 272

+

75-800

75-800

30

200

+

[M + H]+

75-800

30

200

30

+

[M + 2H]2+ 75-800

30

200

18

+

150-600

130

35

150

18

+

350-800

43-493

35

150

30

+

350-800

221-671

35

120

30

-

75-800

75-800 -30 -200

-

[M - H]-

75-800 -30 -200 -30

-

300-800

272

-35 -100 -30

a DP, declustering potential. b FP, focusing potential. c CE, collision energy.

procedure with minor modification (6). Briefly, trimethoprim (100 mg, 1 equiv) was stirred in a 100 mL round-bottom flask with 15 mL of acetonitrile and 30 mL of phosphate buffer (0.1 M, pH 6) was added. Trimethoprim dissolved with stirring at room temperature. An aqueous solution of commercially available NaOCl (∼1 equiv) then was added with vigorous mixing, and the trimethoprim solution became pale yellow. After 5-10 s, GSH (522 mg, ∼5 equiv) in phosphate buffer (15 mL, 0.5 M, pH 8) was added to the reaction mixture. The resulting solution turned orange immediately and was stirred for an additional 30 min at room temperature. The solution was concentrated overnight under vacuum and reconstituted in 25% aqueous CH3CN. A portion of this sample was purified using preparative HPLC, which was carried out using a Luna C18(2) 250 × 21.2 mm column (Phenomenex, Torrance, CA) and a mobile phase consisting of 0.1% formic acid in water (solvent A) and 0.1% formic acid in methanol (solvent B). The HPLC flow rate was 15 mL/min. The product of interest was isolated using a gradient that began with 90% A for the first 5 min, decreased to 40% A over the next 20 min, then was held at 40% A for 3 min before returning to 90% A over 1 min. The system was equilibrated for 5 min before the next injection. An automated fraction collector was triggered to collect regions of the effluent based on the voltage output of a photodiode array detector with the output signal set at λ 254 nm. Under these conditions, the TMP-GSH conjugate eluted at 17.5 min. The fraction containing this product was dried under vacuum and stored in a desiccator at room temperature. Analysis by 1H NMR provided a spectrum in full agreement with that described by Lai, et al.(6), confirming that GSH conjugation had occurred at the 4-position of the 2,6-diaminopyrimidine ring. 2. DACA-GSH (γ-Glutamyl-S-{1-[4-(dimethylamino) phenyl]-3-hydroxypropyl}cysteinylglycine). 4-(Dimethylamino)cinnamaldehyde (10 mg, 1 equiv) was dissolved in 10 mL of methanol with stirring and treated with a solution of GSH (∼1.3 equiv) in phosphate buffer (0.1 M, pH 8). The resulting mixture was stirred overnight at room temperature. The following morning, excess NaBH4 was added (∼5 equiv) and the solution was stirred further for 1 h. The reaction then was quenched with 20% acetic acid (aqueous) and dried overnight under vacuum. The solid was reconstituted in 5 mL of 10% aqueous CH3CN for purification by preparative HPLC. The preparative HPLC was conducted essentially as described above, but with the following changes. The gradient started at 90% A for the first 5 min, decreased to 50% A over 35 min, then was held at 50% A for 3 min before returning to 90% A over 1 min and equilibrated for 6 min. The automated fraction collector was triggered to collect regions of the effluent based on a voltage output from the photodiode array detector with the output signal

632

Chem. Res. Toxicol., Vol. 18, No. 4, 2005

Dieckhaus et al.

Figure 1. Negative product ion spectra of GSH and model GSH adducts. Panel A shows the product ions obtained by CID of the [M - H]- species at m/z 306. Panel B shows the product ion spectrum of the representative benzylic GSH adduct, DACA-GSH, whose [M - H]- appears at m/z 483. Panel C shows the product ion spectrum of the model aromatic thioether GSH adduct, TMP-GSH, whose [M - H]- appears at m/z 594. Panel D shows the product ion spectrum of the thioester GSH adduct, D-SG, whose [M - H]- appears at m/z 583. Panel E shows the product ion spectrum of the representative aliphatic thioether compound GSH adduct, 4-HNE-GSH, whose [M - H]- appears at m/z 464. set at λ 220 nm. This led to the separation of four GSH adducts with retention times of 11.2, 11.7, 18.3 and 18.7 min. The GSH adduct with the 11.2 min retention time represented the desired 1,4-Michael addition product which was used in subsequent studies. The collected fraction was dried under vacuum and stored in a desiccator at room temperature. NMR data confirmed that GSH addition had occurred at carbon 1′ of DACA-GSH (labeling scheme shown in Figure 1). Thus, the methine (position 1′) hydrogen and carbon resonated at δH 4.02 and δC 46.1 ppm, respectively, while HMBC1 correlations are seen from H-c′ to C-1′ and from H-1′ to C-c′ placing the point of GSH conjugation at carbon 1′. The proton and carbon chemical shift assignments for this conjugate are summarized in Table 2. 3. D-SG (Diclofenac-S-acyl-GSH). D-SG was prepared via a procedure similar to that published by Grillo et al. (5). To 25 mL THF was added 500 mg (1 equiv) diclofenac sodium salt followed by 150 µL ethylchloroformate (1 equiv). The suspension was stirred for 15 min at room temperature, at which time 500 mg GSH (1 equiv) in 10 mL H2O was added with continuous stirring for 30 min. The pH of the solution was approximately 4-5. The THF was removed by rotary evaporation and the remaining aqueous solution was washed three times with diethyl ether to remove unreacted diclofenac. The aqueous layer

was further purified using preparative HPLC as described above, with the exception that the D-SG was purified isocratically at 20% A and 80% B. The automated fraction collector was triggered to collect regions of the effluent based on the voltage output from the photodiode array detector with the output signal set at λ 280 nm. Under these conditions, D-SG eluted at 7.2 min. The collected fraction was dried under vacuum and stored in a desiccator at room temperature. NMR analysis indicated that GSH addition had occurred at carbon 1′′ of diclofenac-GSH (labeling scheme shown in Figure 1). Thus, the hydrogens and carbon of the methlyene group at position 2′′ resonated at δH 4.06 and δC 46.4 ppm, respectively, while HMBC correlations were observed from H-c′ to C-1′′ and from H-2′′ to C-1′′ confirming that GSH conjugation has occurred at carbon 1′′. The proton and carbon shift assignments are summarized in Table 3. 4. 4-HNE-GSH (γ-Glutamyl-S-[2-hydroxy-1-(2-hydroxyethyl)heptyl]cysteinylglycine). 4-HNE was obtained as a 1 mg/mL solution in ethanol and was diluted to 1 mM with ethanol. To 800 µL of potassium phosphate buffer (0.1M, pH 7.4) was added 100 µL of 1 mM 4-HNE plus 100 µL of 10 mM GSH. The reaction was left for 20 min at room temperature without stirring at which time ∼5 mg of sodium

Negative Ion MS for Detection of GSH Conjugates Table 2. NMR Assignments for DACA-GSH Adduct positiona

δCb

3′ 2′ 1′ 1 2, 6 3, 5 4 1′′, 2′′ a′ b′ c′ d′ e′ f′ g′ h′ i′ j′

59.1 38.1 46.1 143.0 129.9 120.2 142.8 45.8 43.4 53.3 32.8 54.3 26.6 31.7 174.1 174.7 172.1 176.1

δHb (multiplicities,c J (Hz)) 3.36, 3.51 (m) 1.93, 2.06 (m) 4.02 (t, 8.0) 7.46 (d, 8.7) 7.38 (d, 8.7) 3.12 (s) 3.62 (m) 3.95 (dd, 5.0, 8.9) 2.66, 2.86 (dd, 5.0, 8.9, 14.4) 3.62 (m) 2.00 (m) 2.35 (t, 7.7)

a Labeling scheme as shown in Figure 1. b Chemical shifts in ppm. c Notations: m, multiplet; t, triplet; d, doublet; s, singlet; dd, doublet of doublets.

Table 3. NMR Assignments for D-SG Adduct positiona

δCb

1 2, 6 3, 5 4 1′ 2′ 3′ 4′ 5′ 6′ 1′′ 2′′ a′ b′ c′ d′ e′ f′ g′ h′ i′ j′

137.8 131.8 129.6 126.8 122.9 143.7 115.9 128.5 120.9 131.8 198.2 46.4 41.9 52.5 31.1 53.5 27.4 31.7 170.8 172.5 170.8 171.3

δHb (multiplicities,c J (Hz))

7.54 (d, 8.4) 7.25 (t, 8.4) 6.22 (d, 7.9) 7.07 (t, 7.6) 6.84 (t, 7.5) 7.22 (d, 7.5) 4.06 (s) 3.66 (m) 4.41 (m) 2.99, 3.41 (dd, 4.7, 9.6, 13.4) 3.28 (t, 6.1) 1.81, 1.93 (m) 2.26, 2.35 (m)

a

Labeling scheme as shown in Figure 1. b Chemical shifts in ppm. c Notations: m, multiplet; t, triplet; d, doublet; s, singlet; dd, doublet of doublets.

borohydride was added to reduce the aldehyde and thereby stop the reaction. The reaction mixture was then neutralized with 200 µL of 20% acetic acid and the products were used without further purification. The reaction between 4-HNE and GSH has been well characterized and is known to favor the 1,4-Michael addition products (1). The products appear to be a mixture of diastereomers that separate into four closely eluting peaks using the LC/MS conditions described below. The reaction was assumed to be complete when making subsequent dilutions. NMR Spectroscopy. Samples were dissolved in 160 µL of 2H O and transferred to 3 mm NMR tubes. 1H 1D, TOCSY 2 (1H-1H 2D total correlation spectroscopy with 80 ms mixing time), ROESY (1H-1H 2D rotating frame Overhauser enhancement spectroscopy with 300 ms mixing time), gHMQC (1H-13C 2D gradient heteronuclear multiple quantum coherence, optimized for 1JCH ) 140 Hz) and gHMBC (1H-13C 2D gradient heteronuclear multiple bond correlation, optimized for nJ CH ) 8 Hz) data sets were used to elucidate the structure of DACA-GSH. Tandem Mass Spectrometry of GSH. An aqueous solution of reduced GSH (50 µM) was infused (10 µL/min), together with mobile phase (50% aqueous CH3CN: 50% 0.1% formic acid in

Chem. Res. Toxicol., Vol. 18, No. 4, 2005 633 water, 200 µL/min), directly into the mass spectrometer. The following anions were isolated and subjected to CID: MS2 m/z 306 ([M-H]-); MS3 m/z 306 f 272; MS4 m/z 306 f 272 f 254. For all MSn experiments, the isolation width was set at 3 Da and the collision energy at 25. LC-MS Analysis of Synthetic GSH Conjugates. TMP-GSH, DACA-GSH, D-SG and 4-HNE-GSH were analyzed in positive and negative ion mode using the AB/MDS Sciex API 3000 instrument. For each sample, 100 µL of a 1 µM standard solution in water was loaded onto a Zorbax 5 µm SB-C18 4.6 × 250 mm column (Phenomenex, Torrance, CA), and the analytes were eluted with a gradient flow at 1 mL/min. The flow was split and 20% was directed into the mass spectrometer. The flow was diverted to waste for the first 5 min of each run. Mobile phase A consisted of 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in CH3CN. The gradient began with 100% A isocratic for 7 min before decreasing to 15% A over 38 min, returning to 100% A over 0.1 min and finally equilibrated at 100% A for 5 min. Microsomal Incubations and LC-MS Analysis. Stock solutions (5 mM) of acetaminophen, diclofenac and troglitazone in 50% aqueous CH3CN were prepared. Incubations (500 µL) containing 1 mg/mL rat liver microsomal protein, 50 µM test compound, 6 mM MgCl2, 1 mM NADP+, 10 mM D-glucose-6phosphate, D-glucose-6-phosphate dehydrogenase (EC1.1.1.49 from Torula Yeast Sigma G-8289, 7 units/mL) and 10 mM GSH in phosphate buffer (0.1M, pH 7.4) were incubated in a shaking water bath at 37 °C for 60 min. The control incubation contained 0.5% CH3CN. The reactions were stopped by the addition of an equal volume of CH3CN and centrifuged at 2000 x g for 5 min to pellet the protein. The supernatant was diluted further with an equal volume of water before LC-MS analysis. For each sample, 100 µL aliquots were loaded onto a Zorbax 5 µm SB-C18 4.6 × 250 mm column and analyzed as described above. Bile Sample and LC-MS Analysis. A fasted male Sprague-Dawley rat that had been cannulated at the common bile duct was dosed intraperitoneally with troglitazone at 100 mg/kg (100 mg/mL in DMSO). Bile was collected over dry ice for 0-8 h following drug administration and stored at -70 °C until analyzed. The bile sample was diluted with water 4-fold and an aliquot (50 µL) was loaded directly onto a Zorbax 5 µm SB-C18 4.6 × 250 mm column and analyzed by LC-MS as described above.

Results Characterization of GSH and Synthetic GSH Conjugates by Tandem Mass Spectrometry. 1. GSH. In the full scan positive ion mode, GSH yields a singly charged ion [M + H]+ ion at m/z 308 which, upon CID, affords fragments at m/z 76, 84, 130, 162, 179 and 233, the origins of which have been described in detail (4). In the negative ion mode, GSH yields as a singly charged [M - H]- species at m/z 306. As shown in Figure 1a, CID of this anion affords fragments at m/z 128, 143, 160, 179, 210, 254 and 272. The origins of these product ions were explored using MSn on a ThermoFinnigan LCQ mass spectrometer and a proposed fragmentation pathway is depicted in Scheme 1. First generation product ions from the [M - H]- species include m/z 160 and 272, which likely result from elimination of the elements of glutamine and H2S, respectively. The ion m/z 272, in turn, undergoes fragmentation to afford m/z 143 and 128 as second generation product ions, which likely result from cleavage of the γGlu-Cys amide bond, and m/z 254, which likely arises from the elimination of water. The latter species undergoes fragmentation to yield third generation product ions at m/z 210, which results from elimination of the elements of CO2, m/z 74, corresponding to deprotonated glycine, and m/z 179, whose formation is attributed

634

Chem. Res. Toxicol., Vol. 18, No. 4, 2005

Dieckhaus et al.

Scheme 1. Proposed Fragmentation of the [M - H]- Ion of GSH (m/z 306) Following CID

to loss of the glycine residue with accompanying hydrogen rearrangement to form the cyclic anion depicted in Scheme 1. Precedence for the latter mechanism has been reported previously (7). 2. DACA-GSH. This compound was synthesized, as a representative benzylic GSH conjugate, from the R,β-unsaturated aldehyde, 4-(dimethylamino)cinnamaldehyde. This R,β-unsaturated aldehyde was selected because it has structural similarity to the reactive metabolite, 2-phenylpropenal, implicated in felbamate idiosyncratic toxicities (8). In full scan positive ion mode, DACA-GSH forms both the singly charged [M + H]+ ion at m/z 485 and its doubly protonated counterpart at m/z 243. The product ion spectrum obtained by CID of the [M + H]+ species is typical of that of a benzylic GSH conjugate in that it exhibits a major product ion at m/z 178 through loss of the elements of GSH as a neutral (307 Da). Indeed, none of the product ions observed in this CID spectrum corresponded to the glutathionyl moiety of fragments thereof, and therefore the only MS/MS scan mode that would be potentially useful to detect the singly charged DACA-GSH parent ion would be the neutral loss of 307 Da. When the doubly charged [M + 2H]2+ species of DACA-GSH was selected for CID, product ions were observed at m/z 76, 84, 130, 162, and 179, which derive from the GSH moiety, and at m/z 160 and 178, which derive from the DACA moiety. Thus, CID of doubly charged DACA-GSH, in contrast to the [M + H]+ species, yields a family of singly charged product ions representing both the xenobiotic and tripeptide elements of the adduct structure. Neutral losses are not observed upon CID of the doubly charged parent

ion. A suitable survey scan for the latter might, therefore, be precursors of the abundant pyroglutamate fragment at m/z 130 (which also is evident in the CID spectrum of the singly charged parent). In full scan negative ion mode, DACA-GSH yields an abundant molecular anion [M - H]- at m/z 483, CID of which affords product ions at m/z 128, 143, 179, 210, 254 and 272 (Figure 1b). Note that these fragments are the same as those observed in the corresponding CID spectrum of GSH itself (Figure 1a), indicating that fragmentation of DACA-GSH in the negative ion mode results exclusively in cleavage of bonds within the glutathionyl moiety. This result can be explained by fragmentation of the conjugate being directed by negative charge localization on one of the carboxylate groups of the GSH residue, a feature that should be observed with most GSH adducts under negative ion MS/MS conditions. Consequently, scanning for precursors of negative ions such as m/z 254 or 272 should represent a potentially useful approach for the unbiased detection of GSH conjugates in biological samples. 3. TMP-GSH. This compound served as a model aromatic thioether conjugate with GSH. In the full scan positive ion mode, TMP-GSH also yielded both a singly charged [M + H]+ ion at m/z 596 and its doubly charged counterpart [M + 2H]2+ at m/z 299. When the [M + H]+ species was subjected to CID, fragment ions were formed with m/z 155, 181, 323 and 467, all of which were derived from the TMP residue of the conjugate. Pathways indicative of the presence of the GSH moiety led to the neutral losses of 129 Da (pyroglutamic acid) and 273 Da (γ-glutamyl-dehydroalanyl-glycine), to give the product

Negative Ion MS for Detection of GSH Conjugates

ions at m/z 467 and 323, respectively. As a result, neutral loss scanning for 129 or 273 Da in the positive ion mode should serve as a suitable survey scan for the [M + H]+ ion of TMP-GSH and related thioether GSH adducts. Fragmentation of the doubly charged ion, [M + 2H]2+ at m/z 299, afforded singly charged product ions at m/z 76, 84 and 130, derived from the glutathionyl moiety, and m/z 323 and 467, which derived from the TMP portion of the adduct. This was similar to the behavior exhibited by DACA-GSH, and again suggested that a suitable survey scan for doubly charged GSH conjugates in the positive ion mode would be a m/z 130 precursor ion scan. In the full scan negative ion mode, TMP-GSH yielded a prominent singly charged [M - H]- ion at m/z 594 with no doubly charged species observed. Fragmentation of m/z 594 by CID yielded products at m/z 128, 143, 179, 210, 254 and 272 (Figure 1c), which once again were reflective of the glutathionyl moiety, as noted earlier (Figure 1a and 1b). These results support the conclusion that CID of the deprotonated molecular anions of GSH adducts affords product ions which derive from the GSH moiety, and that precursor ion scanning for parents of m/z 254 or 272 should represent an effective means of detecting these adducts in complex mixtures. 4. D-SG. The full scan positive ion spectrum of D-SG, a model thioester-linked GSH conjugate, has been reported previously (5). In the positive ion mode, D-SG, in contrast to DACA-GSH and TMP-GSH, did not form a doubly charged ion, but afforded a prominent singly charged [M + H]+ species at m/z 585 (35Cl) and 587 (37Cl). The product ion spectrum of m/z 585 included fragments at m/z 162, 179, 215, 233, 250, 278, 292, 308, 438, 456 and 510. The most abundant product ion was generated by the loss of glutamic acid (147 Da) at m/z 438. Consequently, neutral loss scanning (147 Da) may be useful in detecting thioester-linked GSH conjugates when analyzed by positive ion electrospray MS/MS. The product ion generated by the neutral loss of 129 Da was observed as a minor product at m/z 456. In the full scan negative ion mode, D-SG formed singly charged [M - H]- ions at m/z 583 (35Cl) and 585 (37Cl). CID of the negative ion m/z 583 afforded fragments at m/z 128, 143, 179, 210, 254 and 272 (Figure 1d), which corresponded to the same glutathionyl-derived species observed in each of the negative ion MS/MS spectra discussed above (Figure 1a-1c). 5. 4-HNE-GSH. The full scan positive ion spectrum of 4-HNE-GSH, a representative aliphatic GSH conjugate, forms only the singly charged [M + H]+ ion at m/z 466. A doubly charged ion at m/z 233 is not observed. The product ion spectrum obtained by CID of the [M + H]+ yields product ions at m/z 141, 162, 177, 179, 216, 302, 319, 373, 391, and minor ions at m/z 159 and 337. Of these product ions, several indicate the presence of a GSH adduct. The product ions at m/z 162 and 179 are product ions common to GSH. Therefore a precursor scan for one of these ions should detect the presence of the 4-HNE-GSH adduct. The product ions with m/z 159, 319, 337 and 391 relate to the GSH-specific neutral losses of 307 Da (GSH), 147 Da (glutamic acid), 129 Da (pyroglutamic acid), 75 Da (glycine), respectively. As a result, neutral loss scans for any one of these differences should detect the 4-HNE-GSH adduct. In the full scan negative ion mode, 4-HNE-GSH formed a singly charged [M - H]- ion at m/z 464. Negative ion CID of m/z 464 afforded fragments at

Chem. Res. Toxicol., Vol. 18, No. 4, 2005 635

m/z 128, 143, 179, 210, 254 and 272 (Figure 1e), which correspond to the same glutathionyl-derived species observed in each of the negative ion MS/MS spectra discussed above (Figure 1a-1d). Therefore, all classes of GSH adduct examined in this survey should be amenable to detection by MS/MS using precursor ion scanning for these ions, including m/z 254 and 272. Evaluation of Alternative Survey Scans for Detecting Model GSH Adducts. Based upon the mass spectrometric characteristics of the model GSH conjugates outlined above, four potential MS/MS scan modes were selected for evaluation as potential survey scans for adducts of unknown structure. These were (i) neutral loss of 307 Da (positive ion mode), (ii) neutral loss of 129 Da (positive ion mode), (iii) precursors of m/z 130 (positive ion mode), and (iv) precursors of m/z 272 (negative ion mode). A mixture of DACA-GSH, TMP-GSH, D-SG and 4-HNE-GSH was analyzed by LC-MS/MS using each of these scan modes, and the resulting ion current chromatograms are depicted in Figure 2. As anticipated on the basis of their respective CID spectra, each of the conjugates was detected using positive ion mode, although no single scan mode revealed all four analytes (Figures 2a-c). In contrast, scanning for precursors of m/z 272 in the negative ion mode led to the successful detection of all of the GSH conjugates in a single run (Figure 2d). Use of Negative Ion Precursor Scan of m/z 272 for the Detection of GSH Adducts in Vitro. To extend these studies to an in vitro biological system capable of metabolic activation of test compounds, the following substrates were incubated with a rat liver microsomal preparation containing an NADPH-regenerating system and fortified with GSH (10 mM): acetaminophen, diclofenac and troglitazone. The negative ion precursor scan of m/z 272 detected the GSH adducts that have been reported previously (9-13). As shown in Figure 3, acetaminophen yielded two GSH conjugates ([M - H]- ions at m/z 455), consistent with the aromatic thioether conjugates that derive from cytochrome P-450-mediated oxidation reactions (10, 11). The positive ion product ion spectra for the two acetaminophen-GSH adducts are consistent with the 3′-GSH and 2′-GSH previously reported (11). Most notable, the product ion spectra do not contain the product ions m/z 177 or 306, which were observed for the ipso-adduct between GSH and N-acetylp-benzoquinone imine (11). The major troglitazone GSH conjugate that is detected corresponds to the direct addition of GSH to the troglitazone molecule, with an [M - H]- ion at m/z 745 (12-14). Diclofenac afforded two isomeric conjugates with [M - H]- ions at m/z 615 (35Cl-containing species), consistent with the aromatic thioether adducts known to be formed by 4- and 5-hydroxylation of diclofenac, followed by two-electron oxidation to the corresponding quinoneimine intermediates (9). (The acyl-GSH derivative, D-SG, employed above as a model compound, was not formed in these microsomal incubations). The control incubation shows that the peaks do not arise from other components in the incubation. The earliest eluting peak in each of the ion chromatograms is oxidized glutathione [M - H]- ion with m/z 611. Collectively, the results of these model studies demonstrate the potential of precursor ion scanning (parents of m/z 272 in the negative ion mode) for the detection of GSH conjugates formed through metabolic activation processes in vitro.

636

Chem. Res. Toxicol., Vol. 18, No. 4, 2005

Dieckhaus et al.

Figure 2. LC-MS/MS analysis of the model GSH adducts, DACA-GSH, TMP-GSH, D-SG and 4-HNE-GSH using the positive ion neutral loss of 307 Da scan (panel A), the positive ion neutral loss of 129 Da scan (panel B), the positive ion precursor of m/z 130 scan (panel C) and the negative ion precursor of m/z 272 scan (panel D). Panel A shows that the neutral loss of 307 Da scan identifies the benzylic thioether GSH adduct, DACA-GSH, and the aliphatic thioether GSH adduct, 4-HNE-GSH. Panel B shows that the neutral loss of 129 Da scan identifies the aromatic thioether GSH conjugate, TMP-DACA, the thioester GSH conjugate, D-SG and the aliphatic thioether GSH adduct, 4-HNE-GSH. Panel C shows that the positive ion precursor scan of m/z 130 scan identifies the doubly charged ion of the benzylic and aromatic thioether conjugates. Panel D shows the negative ion precursor of m/z 272 scan identifies all four model GSH adducts.

Evaluation of Alternative Survey Scans for the Detection of GSH Adducts of Troglitazone in Vivo. To determine the potential utility of the negative ion precursors of m/z 272 scan for detecting GSH adducts formed in vivo, untreated bile from a rat dosed with troglitazone was analyzed by LC-MS/MS. The results are shown in Figure 4a, and demonstrate that the negative ion survey scan detects two troglitazone GSH adducts, one with an [M - H]- ion at m/z 745 consistent with the direct addition of GSH to troglitazone (12-14), and another previously unidentified adduct with an [M - H]- ion at m/z 731. While the structure of the latter conjugate has not been determined fully, control bile data, coupled with positive ion MS/MS information, indicate that the ion is indeed a GSH adduct that is likely derived from troglitazone. In positive ion mode the unexpected GSH adduct with m/z 733, yields the daughter ions m/z 658, 604, 484 and 458. For comparison, the same bile sample was analyzed in the positive ion mode, scanning for the neutral loss of 129 Da and, as depicted in Figure 4b, the same two glutathione adducts were detected. The principal difference between the scan modes is the selectivity of GSH adduct detection, which was markedly superior in the negative ion survey scan. Therefore, based upon these preliminary results, it would appear that scanning for precursors of m/z 272 in the negative ion mode offers promise for the detection of unknown GSH adducts present in bile samples.

Discussion The use of positive ion electrospray MS/MS methods, suffers from a number of limitations in the detection of GSH conjugates due to the behavior of the MH+ ions of these conjugates under CID conditions. As a consequence of qualitative differences between the fragmentation behavior of different structural classes of GSH adducts, it has not been possible to devise a generally applicable survey scan for the detection of GSH conjugates of unknown structure (although the 129 Da constant neutral loss scan has proven useful in many cases). This deficiency of positive ion analysis is compounded further by the propensity of many GSH adducts to form doubly charged parent species ([M + 2H]2+), which typically do not fragment upon CID by way of neutral losses, but rather yield pairs of singly protonated species (e.g., m/z, 76, 84, 130, 162 and 179) that reflect components of the tripeptide moiety. Although it might be anticipated that precursor ion scanning for one of these singly charged species could represent a useful approach for the detection of the parent [M +2H]2+ species, the preferred charge state of an unknown GSH adduct cannot be predicted a priori, thereby limiting the applicability of this approach as a survey scan. In the present investigation, we have demonstrated that negative ion MS/MS shows promise in overcoming the above limitations since CID of the parent [M - H]ions of all major classes of GSH conjugates afford a

Negative Ion MS for Detection of GSH Conjugates

Chem. Res. Toxicol., Vol. 18, No. 4, 2005 637

Figure 3. LC-MS/MS analysis of rat liver microsomal incubations fortified with GSH and containing the following substrates: acetaminophen (panel A), diclofenac (panel B), troglitazone (panel C) or no compound control (panel D). GSH adducts were detected by means of scanning for precursors of m/z 272 in the negative ion mode. The early eluting peak at 16 min is oxidized glutathione [M - H]- with m/z ) 611.

Figure 4. LC-MS/MS analysis of bile from a rat dosed with troglitazone using the negative ion precursor of m/z 272 scan (panel A) or the positive ion neutral loss of 129 Da scan (panel B). The arrows show the troglitazone GSH adducts and the peaks marked with an asterisk are detected LC peaks unrelated to troglitazone. The difference between the total and extracted ion chromatograms indicate that the negative ion precursor of m/z 272 offers enhanced selectivity.

common series of product anions that derive from the glutathionyl moiety, Figure 1. The MS/MS spectra of model GSH adducts examined under negative ion conditions are qualitatively similar to that of free GSH which, as in the case of underivatized cysteinyl-containing peptides, eliminates the elements of H2S upon CID to

generate the anion at m/z 272 (15). This same m/z 272 ion was employed successfully in our studies to detect (through precursor ion scans) xenobiotic conjugates formed both in vitro and in vivo, Figures 3 and 4. In addition, since multiply charged anions typically are not observed in the negative ion spectra of GSH adducts, the population of ions in the (single) charge state of interest is enhanced, thereby potentially increasing sensitivity of detection. For these reasons, it would appear that scanning for precursors of m/z 272, or any GSH fragment ion with m/z 128, 143, 160, 179, 210 or 254, in the negative ion mode is a versatile survey scan for the detection of GSH conjugates of unknown structures. It should be noted, however, that the MS/MS spectra obtained by CID of the [M - H]- ions of GSH conjugates are dominated by fragments of the tripeptide moiety, and few structurally informative ions are available from which to infer details of the corresponding xenobiotic residue. In this regard, positive ion MS/MS is likely to prove more valuable for structure elucidation work once an unknown conjugate has been detected. Therefore, an attractive analytical strategy would be to combine precursor ion scanning of m/z 272 in the negative ion mode as a survey technique for adduct detection, with product ion scanning of the MH+ species under positive ion conditions for structure elucidation purposes. The technical capabilities required for this approach (polarity switching “on the fly” and data-dependent CID) are available on the current generation of commercial LC-MS/MS systems, and studies to examine the broad utility of such enhanced mass spectrometric methodology for studies with GSH conjugates currently are underway in our laboratories.

638

Chem. Res. Toxicol., Vol. 18, No. 4, 2005

Dieckhaus et al.

References (1) Esterbauer, H., Schaur, R. J., and Zollner, H. (1991) Chemistry and biochemistry of 4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radicals Biol. Med. 11, 81-128. (2) Uetrecht, J. P. (1999) New concepts in immunology relevant to idiosyncratic drug reactions: the “Danger Hypothesis” and innate immune system. Chem. Res. Toxicol. 12, 387-395. (3) Samuel, K., Yin, W., Stearns, R. A., Tang, Y. S., Chaudhary, A. G., Jewell, J. P., Lanza, T., Lin, L., Hagmann, W. K., Evans, D. C., and Kumar, S. (2003) Addressing the metabolic activation potential of new leads in drug discovery: a case study using ion trap mass spectrometry and tritium labeling techniques. J. Mass Spectrom. 38, 211-221. (4) Baillie, T. A., and Davis, M. R. (1993) Mass spectrometry in the analysis of glutathione conjugates. Biol. Mass Spectrom. 22, 319325. (5) Grillo, M. P., Hua, F., Knutson, C. G., Ware, J. A., and Li, C. (2003) Mechanistic studies on the bioactivation of diclofenac: identification of D-SG in vitro in incubations with rat and human hepatocytes. Chem. Res. Toxicol. 16, 1410-1417. (6) Lai, W. G., Nasir, Z., and Uetrecht, J. P. (1999) Metabolism of trimethoprim to a reactive iminoquinone methide by activated human neutrophils and hepatic microsomes. J. Pharmacol. Exp. Ther. 291, 292-299. (7) Harrison, A. G. (2001) Sequence-specific fragmentation of deprotonated peptides containing H or alkyl side chains. J. Am. Soc. Mass Spectrom. 12, 1-13. (8) Dieckhaus, C. M., Thomspon, C. D., Roller, S. G., and Macdonald, T. L. (2002) Mechanism of idiosyncratic drug rections: the case of felbamate. Chem. Biol. Interact. 142, 99-117. (9) Tang, W., Stearns, R. A., Bandiera, S. M., Zhang, Y., Raab, C., Braun, M., Dean, D., Pang, J., Leung, K., Doss, G., Strauss, J. R., Kwei, G. Y., Rushmore, T. H., Chiu, S. L., and Baillie, T. A.

(10)

(11)

(12)

(13)

(14)

(15)

(1999) Studies on cytochrome P-450-mediated bioactivation of diclofenac in rats and human hepatocytes: identification of glutathione conjugated metabolites. Drug Metab. Dispos. 27, 365372. Van de Straat, R., De Vries, J., De Boer, H. J. R., Vromans, R. M., and Vermeulen, N. P. E. (1987) Relationships between paracetamol binding to and its oxidation by two cytochrome P-450 isozymes - a proton nuclear magnetic resonance and spectrophotometric study. Xenobiotica 17, 1-9. Chen, W., Shockcor, J. P., Tonge, R., Hunter, A., Gartner, C., and Nelson, S. D. (1999) Protein and Nonprotein Cysteinyl Thiol Modification by N-Acetyl-p-benzoquinone Imine via a Novel Ipso Adduct. Biochemistry 38, 8159-8166. Kassahun, K., Pearson, P. G., Tang, W., McIntosh, I., Leung, K., Elmore, C., Dean, D., Wang, R., Doss, G., and Baillie, T. A. (2001) Studies on the metabolism of troglitazone to reactive intermediates in vitro and in vivo. Evidence for novel biotransformation pathways involving quinone methide formation and thiazolidinedione ring scission. Chem. Res. Toxicol. 14, 62-70. Tettey, J. N., Maggs, J. L., Rapeport, W. G., Pirohamed, M., and Parks, B. K. (2001) Enzyme-induction dependent bioactivation of troglitazone and troglitazone quinone in vivo Chem. Res. Toxicol. 14, 965-974. He, K., Rasmy, E. T., Pool, W. F., Reily, M. D., Reed, J. E., Bridges, A. J., and Woolf, T. E. (2004) Metabolic activation of troglitazone: identification of a reactive metabolite and mechanisms involved. Drug Metab. Dispos. 32, 639-646. Bilusich, D., Brinkworth, C. S., McAnoy, A. M., and Bowie, J. H. (2003) The fragmentation of [M - H]- anions derived from underivatised peptides. The side-chain loss of H2S from CYS. A joint experimental and theoretical study. Rapid Commun. Mass Spectrom. 17, 2488-2494.

TX049741U