Anal. Chem. 1990, 62, 121-124
121
Protocol for Liquid Chromatography/Mass Spectrometry of Glutathione Conjugates Using Postcolumn Solvent Modification Mark F. Bean,*J Sharon L. Pallante-Morell,2Deanne M. Dulik? and Catherine Fenselau2 Department of Pharmacology and Molecular Sciences, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205
A novel protocol for thermospray llquld chromatography/mass spectrometry (LC/MS) analysis of mixtures of glutathione conjugates Is reported. Solvent conditions for optimal hlghperformance llquld chromatography are not always the same as for optimal thermospray Ionization mass spectrometry. Lablie glutathione conjugates that give poor spectra In aqueous ammonium acetate yleid more intense molecular Ion signals wlth Increased percentages of acetonitrile. Dlrect injection thermospray ionlzatlon using 30-60 % acetonitrile in aqueous ammonium acetate produced protonated molecular ions for giutathlone conjugates of menadione, styrene oxlde, pentachiorophenyi methyl sulfone, chlorodlnltrobenzene, and chlorambucil. Slnce, the high percentages of organlc modifier needed for good molecular ion Intensity preclude chromatographic separation of these polar compounds, successful LC/MS was facilitated by postcolumn addition of organic modlflers to the mobile phase. This new methodology allowed excellent chromatographic separations and thermospray ionization mass spectra to be obtained for a mixture of haioalkane glutathione conjugates. Moreover, cleavage of the yglutamyl-cystelne amide bond of glutathione results In class-characterlstlc fragment Ions. Changes in the fragmentation pathways in spectra acqulred wlth and wlthout organlc modifiers shed light on the Importance of the desolvatlon process In obtaining good molecular Ion sensltlvlty in thermospray.
Reduced glutathione or (y-glutamylcysteiny1)glycine(GSH), the major non-protein thiol constituent of cells, plays a key role in detoxification of electrophilic xenobiotics (1). However, conversion of these conjugates to free thiols via the cysteine-8-lyase pathway may result in greater toxicity of compounds such as some halogenated alkanes (2). Also, recent reports (3) implicate glutathione conjugates in the development of resistance to antineoplastic agents. These chemically and thermally unstable polar metabolites require extensive derivatization for gas chromatography-mass spectrometric analysis ( 4 ) and do not produce molecular ions in direct chemical ionization (DCI) mass spectrometry (5). Characterization by fast atom bombardment (FAB) (6) or plasma desorption mass spectrometry (7) gives reliable molecular weight information but few diagnostic fragments and may be hampered by the presence of sample contaminants and salts. Low-energy collisionally induced dissociation analyses of glutathione conjugates ionized by FAB have also recently been presented (8-10). Enzymatic and chromatographic determinations of glutathione in biological samples have been reviewed (11). Present address: Department of Physical and Structural Chemistry, Smith Kline and French Research Laboratories, King of Prussia, PA 19406-0939. *Present address: Department of Chemistry, University of Maryland, Baltimore County, Baltimore, MD 21228. Present address: De artment of Drug Metabolism, Smith Kline and French Research Lagoretories, King of Prussia, PA 19406-0939.
Thermospray liquid chromatography/mass spectrometry (LC/MS) (12, 13) is an attractive technique for the analysis of glutathione conjugates for several reasons: (1)detection is independent of UV chromophoric moieties, (2)the analysis of complex mixtures of nonvolatile, structurally similar compounds is possible with minimal sample manipulation, and (3) it not only provides molecular weight information but also yields a number of fragment ions diagnostic of the presence of glutathione conjugates. Although isolated reports of thermospray mass spectrometry of glutathione conjugates have appeared in the literature (14-181, these studies failed to produce good data for the molecular ions (el% of the sample base peak intensity), and some did not involve the chromatographic potential of thermospray. The objective of this study was to develop a reliable and general protocol for the liquid chromatographic separation and mass spectrometric detection of glutathione conjugates in a mixture by the simplest possible means.
EXPERIMENTAL SECTION Reagents. HPLC grade acetonitrile, methanol, ethyl acetate, trifluoroacetic acid, and ammonium acetate were obtained from J. T. Baker Chemical Co. (Phillipsburg, NJ). Water was doubly distilled, deionized, filtered, and stored in high-density polypropylene plastic bottles. Reduced and oxidized glutathione ((7-glutamylcysteinyl)glycine), L-glutamine, L-glutamic acid, ethacrynic acid ([2,3-dichloro-4-(2-methylene-l-oxobutyl)phenoxylacetic acid), and l-chloro-2,4-dinitrobenzenewere obtained from Sigma Chemical Co. (St. Louis, MO). (S)-(-)-2Pyrrolidone-5-carboxylicacid was purchased from Aldrich Chemical Co. (Milwaukee, WI). Styrene oxide (racemic mixture) was purchased from MC/B (Gibbstown, NJ). The glutathione conjugates of trichloroethylene, hexafluoropropene, and 2chloro-1,1,2-trifluoroethylene were prepared by James Stevens, W. Alton Jones Cell Science Center, Lake Placid, NY. The glutathione conjugate of menadione (2-methyl-1,4naphthalenedione) was provided by Thomas Jones, University of Maryland School of Medicine, Baltimore, MD. Pentachlorophenyl methyl sulfone was provided by Vernon Feil, USDAMRLL, Fargo, ND, and chlorambucil (4-[bis(2-~hloroethyl)amino]benzenebutanoic acid) was provided by John Hilton, The Johns Hopkins University Oncology Center, Baltimore, MD. Preparation of Glutathione Conjugates, Glutathione conjugates of pentachlorophenyl methyl sulfone and chlorambucil were prepared enzymatically by using immobilized microsomal glutathione S-transferases as previously described (19). The ethacrynic acid, styrene conjugates of l-chloro-2,4-dinitrobenzene, oxide, and halogenated alkanes were synthesized under basic conditions (pH 10-12) according to a published procedure (20) and purified by solvent washes of the sample retained on octadecylsilyl silica (CIS)cartridges. Mass Spectrometry. Mass spectra were recorded on a MS80RF double-focusing mass spectrometer (Kratos Analytical, Manchester, UK). Three-seconds-per-decade scans were taken from m/z 900 to m/z 100 at a resolution of 1500. Data acquisition and processing were managed by the Kratos DS55 software on a Data General Nova 3 minicomputer with a Kratos fast preprocessor. Our version of the thermospray LC/MS interface, also from Kratos, did not have a probe temperature gradient controller, although the probe temperature was made to track mobile phase gradients manually without difficulty. Similarly, there was no
0003-2700/90/0362-0121$02.50/0 0 1990 American Chemical Society
122
ANALYTICAL CHEMISTRY, VOL. 62, NO. 2,JANUARY 15, 1990
filament power supply, repeller, or glow discharge electrode. The instrument was modified as follows: the low-voltage first beamcentering plate was moved 3 mm further away from the highvoltage source block; a heating tape around the vacuum manifold near the focus assembly maintained the ion optics at about 60 "C; a -90 "C refrigerated vacuum vapor trap (model RT409A, Savant Instruments, Hiksville, NY) was inserted in the solvent take-off line between the mass spectrometer and the liquid nitrogen cold finger. These alterations were effective in completely eliminating any high-voltage arcing in the source housing. Heating the focus assembly reduced the slight beam defocusing observable in solvent gradient runs. The extra vapor trap reduced liquid nitrogen consumption by two-thirds, prevented clogging of the glass cold finger inlet, protected the mechanical pump from corrosive vapors, and resulted in sublimation of all frozen solvents from the cold finger over to the refrigerated trap at the end of the day as the cold fmger warmed to room temperature. Pumping efficiency was not reduced by the extra trap and related plumbing. Calibration of the magnet scan was effected with 100-pL injections of dilute mixtures of poly(ethy1ene glycol) polymers in ammonium acetate and observation of the MNH4+ions. Liquid Chromatography. The principal solvent delivery system consisted of two Beckman 114M pumps and a Model 420 gradient controller (Beckman Instruments, Inc., Berkeley, CA). Precolumn hardware included a Rheodyne Model 7125 injector with 250-pL loop and Rheodyne three-way column switching valve (Rainin Instrument Co., Woburn, MA) and a 5-pm filter (Alltech Associates, Deerfield, IL). The following columns were used with appropriate 3-cm guard columns: Brownlee RP-8 Spheri-10 (25 cm X 0.46 cm i.d., 10 pm), Brownlee RP-18 (10 cm X 0.46 cm i.d., 5 pm), a 10 pm aminopropyl Shandon Hypersil APS (25 cm X 0.46 cm, 5 pm). A Brownlee CIS guard column was used in line in all cases. It served to dampen the probe temperature fluctuations observed in columnless direct injections caused by introduction of the sample in a solvent different from the operational mobile phase. The postcolumn solvent delivery system consisted of a Spectroflow 400 pump with special pulse dampener (Kratos Analytical, Manchester, UK) and a MCVT/ 100 micrometering T-valve (ScientificGlass Engineering Inc., Austin, TX) or a 10-pL Lee Micro Mixer (The Lee Co., Westbrook, CT).
RESULTS AND DISCUSSION Glutathione. Reduced glutathione was selected for preliminary optimization of experimental parameters for the thermospray LC/MS interface as well as the mass spectrometer. Initial results from direct HPLC injection via a C18guard column using an aqueous ammonium acetate mobile phase (0.1 M) were disappointing in that no quasi-molecular ions were observed (Figure la). Manipulating the probe and source temperatures or the buffer concentrations resulted in spectra that a t best had weak protonated molecular ions. Substitution of the buffer by 0.1 M trifluoroacetic acid or ammonium trifluoroacetate likewise resulted in no improvement. However, as the percentage of acetonitrile in the HPLC mobile phase was increased, and the probe temperature was decreased to compensate for the lower vaporization temperature of the mobile phase, we obtained excellent spectra of the reduced tripeptide (Figure Ib). In addition to an intense protonated molecular ion peak, the spectrum obtained in 30% acetonitrile (Figure 1b) differs from the one obtained without organic modifier (Figure la) in the mass of the base p e a k m / z 129 vs m / z 130. This difference is also observed in spectra of glutathione conjugates obtained with and without organic modifier. As has been noted previously (21),the most obvious path for formation of ions of mass 130 from glutathione involves temperaturedependent intramolecular cyclization of the y-glutamyl residue to yield protonated (mlz 130) and ammoniated (mlz 147) pyrrolidone carboxylic acid (pyroglutamic acid) and displacement of cysteinylglycine (protonated mass of 179). The prominent peak at m / z 129 in the spectrum acquired in the presence of acetonitrile differs from the pyrrolidone
130
a
lo07 I
150
200
250
300
350
m/z b
IMH)' 308
i
150
200
250
300
350
ll/Z
Figure 1. Thermospray ionization mass spectrum of glutathione, direct injection: (a) aqueous NH,OAc; (b) 30% acetonitrile in aqueous NH,OAc. carboxylic acid peak at m / z 130 by 1 dalton, requiring a difference in the number of nitrogens and suggesting that the peak is glutamine related rather than glutamic acid related. Although this probably arises from ammonolysis of the unusual, unhindered y-Glu-Cys bond followed by dehydration, the possibility of concurrent contributions from fragmentation of the molecular ion (C, + 2H like cleavages from the molecular ion and the dehydrated molecular ion) should not be excluded. Concurrent contributions to the peak at m / z 147 are also probable since protonated glutamine, ammoniated pyrrolidone carboxylic acid, and a fragment derived from cleavage of the N-C bond of the cysteinyl residue with retention of two additional hydrogens on the glutamyl portion are isobaric and since the observed peak has almost double the intensity in relation to m / z 179 when compared to the spectrum acquired without organic modifier. It seems apparent that organic solvents such as acetonitrile, methanol, or 2-propanol stabilize the glutathione conjugates against cleavage of the glutamic acid residue, resulting in intense molecular ions; the precise cause is not known. Increased concentrations of organic modifier have similarly been shown to minimize certain solvolytic reactions (22). Direct Injection of Glutathione Conjugates. The same mobile phase (30% acetonitrile in 0.1 M aqueous NH,OAc) that had been successful for the analysis of glutathione produced variable results for direct injections of glutathione conjugates through a guard column. The glutathione conjugate of a toxic metabolite of polystyrene monomer, styrene oxide, yielded good spectra showing MH' (16% of sample base peak intensity) and, due to NaOH contamination from the synthetic conjugation reaction, a smaller MNa+ (m/z 4501, Figure 2. The same protocol applied to the GSH conjugate of the diuretic drug ethacrynic acid yielded protonated molecular ions of low intensity and variable chlorine isotope peak distributions. Increasing the concentration of acetonitrile to 70% and the consequent decrease of the vaporizer tip temperature resulted in more consistent, more intense molecular ion patterns, and decreased the summed ion currents over the fragment ion region (Figure 3). High concentrations of acetonitrile also produced good spectra of the GSH conjugates
ANALYTICAL CHEMISTRY, VOL. 62, NO. 2, JANUARY 15, 1990
123
‘i‘
t, >loo! BO z v)
E 3
60 40
W
I-
!4
4
d
a
U
20
0
0 100
150
200
250
300
350
400
450
100
500
150 200
250
300
350
400
450
500
m/z
m/z
Flgwe 2. Thermospray lonizatbn mass spectrum of glutathbnyistyrene oxide, dlrect injection (50% acetonitrile in aqueous NH,OAc).
Flgure 5. Thermospray ionization mass spectrum of scan 64 of the chromatogram in Figure 4. LC/MS, postcolumn combined solvent was 27% acetonitrile In 1.5 M aqueous NH,OAc. TIWE NINI 5
0
10
1s
20
2s
30
100
> m c W
BO
60
o
W
zc
20 30 40 50 60 70 eo ACETONITRILE I N MOBILE PHASE ( X V/V)
io
Figure 3. Plot of ion intensities in the thermospray spectrum of giutathionylethacrynic acid as a function of the percentage of acetonitrile in the mobile phase.
4 -I W
40
a
20
0 0
t
50
100 150 200 250 300 350 400 450 500 550 SCAN I
Figure 8. Total ion LC/MS chromatogram of a mixture of achromophoric glutathione conjugates of halogenated alkanes (LC pumps, 0-20% MeCN In 0.2 M NH,OAc duing 20 min, 0.8 M m i n , CBcolumn; postcolumn pump, 100% MeCN, 0.4 mL/min): (1) glutathionyldichloroethylene, (2) glutathionyichlorotrifluoroethane,(3)unknown = ,
-
- + o
, 0
10
20
, .
.-: 30
40
50
f
4”, 60
,w 7C
80
SCAN C
Flgure 4. Total ion and selected ion chromatogram of a mixture of achromophoric glutathione conjugates of halogenated alkanes (40 % acetonitrile In 0.2 M NH,OAc, 0.8 mL/min, CBcolumn): (1) m / z 402, (2) m l z 424, (3) m l z 458, (4) total Ion current. of menadione, and chlorambucil; the spectra of pentachlorophenyl methyl sulfone and ethacrynic acid showed good relative intensity of the molecular ions ( 1 5 6 0 % of sample base peak) but isotopic ratios that varied from scan to scan indicating poor ion statistics. The conjugates of ethylene dichloride, chlorotrifluoroethane,and hexafluoropropane were less exigent, and produced intense molecular ions (15-100% of the sample base peak intensity) in 25-75% acetonitrile. LC/MS of GlutathioneConjugates. Although LC/MS provides much more structural information than HPLC with UV detection, it is especially useful for the analysis of compounds that absorb only below 215 nm where solvent gradients can produce sharply sloping UV absorption base-line traces. In this vein, attempts to separate and identify the components of a mixture of haloalkane glutathione conjugates using 30-70% acetonitrile in aqueous ammonium acetate and an octadecylsilyl silica column met with immediate frustration. Glutathione conjugates of even these highly nonpolar compounds were still too polar to be adequately retained on the column. Figure 4 illustrates the poor resolution of the three main components of the mixture as represented by the total
glutathionylpentafluoropropene, (4) glutathiinylhexafluoropropane,(5) unknown = giutathionylhexafluorobutene.
ion current and selected ion chromatograms of the protonated molecular ions expected for glutathionyldichloroethylene (402), glutathionylchlorotrifluoroethane (424), and glutathionylhexafluoropropane (458). In Figure 5 the three protonated molecular ion species are observed in the same scan. It is possible to recognize the three main components of the mixture even with negligible separation by following the changing ion intensities in the successive scans. Nevertheless, we were also interested in identifying minor contaminants in the mixture which were not discernible without better chromatographic separation. The use of a bonded octylsilyl silica or propylamino stationary phase instead of C18did not result in sufficient column retention. It was evident that the constrictions on the mobile phase necessary to produce consistent and intense quasimolecular ions negated the possibility of HPLC separations. Optimal mobile phase conditions for thermospray ionization were not the optimal conditions for chromatography. In order to free the chromatography from the constrictions of the mass spectrometer, it was necessary to insert a third, pulse-dampened, postcolumn pump, joined by a micrometering T valve to the probe inlet. It was now possible to perform the chromatography with a low percentage acetonitrile gradient and low flow rates in order to effect separation while adding acetonitrile postcolumn to achieve optimal spectra. The resulting ion chromatogram is illustrated for the separation of the mixture of haloalkane glutathione conjugates on a C8column in Figure 6. In addition to achieving base-line
124
ANALYTICAL CHEMISTRY, VOL. 62, NO. 2, JANUARY 15, 1990
ACKNOWLEDGMENT We thank Tom Jones, University of Maryland School of Medicine, and James Stevens, W. Alton Jones Cell Biology Center, Lake Placid, NY, for their contribution of several glutathione conjugates used in this study.
a
2 41 ' 60
0
LITERATURE CITED
129
Z40
I
,
-
205 ( 1) Moldeus, P.
(MH)+
0 100
150 200
250
300
350
400
450
500
(2) (3) (4)
b
(6)
n/z (5)
147 1007
I
(7) 295
(MH)
(8)
+
424
(9)
I
(10) 100
150 200 250
300 350
400
450
500
m/z
(1 1) (12) (13) (14)
Figure 7. Thermospray ionization mass spectrum of glutathionyldichloroethylene: (a) direct injection, 0.2 M aqueous NH,OAc; (b) LC/MS, postcolumn combined solvent was 27% acetonitrile in 1.5 M aqueous
(15)
NH,OAc.
(16)
separation of the principal components and excellent spectra, it was possible to identify two unknown fluorinated contaminants in the mixture. Peak 3 in Figure 6 can be assigned to a glutathionylpentafluoropropene(m/z 438) while peak 5 must be a glutathionylhexafluorobutene ( m / z 470). The improvement in quasi-molecular ion intensity using the new protocol over direct injection in aqueous NH40Ac is illustrated for glutathionyldichloroethylene in Figure 7. The spectral pattern in thermospray LC/MS of glutathione conjugates provides a series of three class-characteristic peaks: m/z 129, 147, and (MH - 129)+,which together indicate an N-terminal glutamic acid. Thus the two impurities in the haloalkane mixture were readily identifiable as glutathione conjugates using the combined LC/MS system.
(17) (18) (19) (20) (21) (22)
In Drug Metabolism, Mdecular Approaches and pharma
ceutical Implications; Siest. G., Ed.; Pergamon: Oxford, 1985; pp 69-73. Monks, T. J.; Lau, S. S. Drug Metab. Dispos. 1987, 15, 437-441. Arrick, E. A.; Nathan, C. F. Cancer Res. 1984, 4 4 , 4224-4232. Wolfe, D. E.; VandenHeuveC W. J. A.; Tyler, T. R.; Walker, R. W.; Koniuszy, F. R.; Gruber, V.; Arison, 8. H.; Rosegay, A,; Jacob, T. A,; Wolf, F. J. Drug Metab. Dispos. 1980, 8 , 131-138. Nelson, S. D.; Yaishnav, Y.; Kambora. H.; Balllle, T. A. Biomed. Mass Spectrom. 1981, 8 , 244-251. Dullk, D. M.; Fenselau, C.; Hllton, J. Biochem. Phannacol. 1988, 35, 3405-3409. Alal, M.; Dullk, D.; PallantsMorell, S. Presented at the 34th Annual Conference on Mass Spectrometry and Allled Topics, Cincinnati, OH, 1986; pp 1005-1006. Straub, K. M. I n Mass Spectromtry in Biomedical Research; Gaskell, S. J., Ed.; J. Wiley and Sons: Chichester, 1986; pp 115-134. Heeremans, C. E. M.; te Koppele. J. M.; Niessen, W. M. A.; van der Greef, J.; Brussee, J.; La Vos, G. F.; ten Noever de Brauw, M. C.; van der Greef, J. Biomed. Envkon. Mess Spectrom. 1968, 17, 181-186. Haroldsen, D. E.;Relily, M. H.; Hughes, H.; Gaskell. S. J.; Porter, C. J. Biomed. fnviron. Mass Spectrom. 1988, 615-621. Anderson, M. E. Methook Enzymol. 1985, 713, 548-555. Blakley, C. R.; Vestal, M. L. Anal. Chem. 1983, 55, 750-754. Vestal, M. L. I n Ion Formation from Organic So/&; Bennlnghoven, A., Ed.; Springer-Verlag: Berlin, 1983; pp 246-203. Nocerini, M. R.; Yost, G. S.; Carlson. J. R.; Liberato, D. T.; Breeze, R. G. Drug M t a b . Dispo~.1985, 73, 690-694. Betowski, L. D.; Korfmacher, W. A.; Lay, J. O., Jr.; Potter, D. W.; Hlnson, J. A. Biomed. Envkon. Mess Spectrom. 1987, 14, 705-709. Sasame, H. A.; Liberato, D. J.; Gillette, J. R. Drug Metab. Dispos. 1987, 75, 349-355. Parker, C. E.;Df~Witt,J. S.; Smith, R. W.; Gopinathan, M. E.; Hernandez, 0.;Tomer, K. E.; Vestal, C. H.; Sanders, J. M.; Bend, J. R. Biomed. Environ. Mass Spectrom. 1988, 623-634. Conchillo, A.; Casas, J.; Messguer, A,; Abian, J. Biomed. Environ. Mess Spectrom. 1988, 16, 339-344. Pallante, S. L.; Lisek, C. A.; Dullk, D. M.; Fenseiau, C. Drug Metab. Dispos. 1986, 14, 313-318. Ryan, A. J.; Bend, J. R. Drug Metab. Dispos. 1977, 5 , 363-367. Lagna, W. M.; Callery, P. S. Biomed. Environ. Mass Spectrom. 1985, 12, 699-703. Bean, M. F.; PallantaMorell, S. L.; Fenselau, C. Biomed. fnviron. Mass Spectrom. 1989, 18, 219-223.
RECEIVED for review May 15, 1989. Revised manuscript received October 2, 1989. Accepted October 12, 1989. The research described in this article was funded by grants from the National Institutes of Health (GM 21248 and RR02241) and from the National Science Foundation (DCB 85-09638).