Fragmentation of Protonated Thioether Conjugates of Acrolein Using Low Collision Energies

Christa H. Oberth and A. Daniel Jones Facility for Advanced Instrumentation, University of California, Davis, California, USA

The protonated mercapturic acid conjugate of acrolein, S-(3-oxopropyI)-N-acetyl-L-cysteine (I), undergoes facile retro-Michael loss of acrolein in the gas phase. To determine whether extensive loss of acrolein would impede structural characterization of acrolein-peptide ad ducts, fragmentation reactions of a series of conjugates, formed by l,4-Michael addition of acrolein to peptides and cysteine derivatives, were investigated at collision cell potentials up to - 50 V using a triple quadrupole mass spectrometer. Differences in fragmentation dynamics suggest protonation at the sulfur of the N-acetylcysteine conjugate 1 facilitates retro-Michael elimination of acrolein with a low activation energy relative to other fragmentations. Analogous fragmentation was eliminated after borohydride reduction of the aldehyde to an alcohol. Retro-Michael fragmentation was not significant for acrolein conjugates of glutathione derivatives, suggesting that proton sequestration occurs in peptides with multiple amide linkages even when the peptide does not contain a basic amino group. An unexpected outcome of these experiments was the observation of a facile gas-phase cleavage of peptides on the N-terminal side of S-(3-oxopropylkysteine residues. Such fragmentation behavior may prove useful for locating cysteine residues in peptides. (J Am Soc Mass Spectrom 1997, 8, 727-736) © 1997 American Society for Mass Spectrometry

ovalent modification of proteins and DNA by gates from biological samples [3,4, 13-17]. Using FAB reactive electrophiles such as xenobiotic metab­ or newer ionization techniques coupled with tandem Colites can lead to cancer or cell death [1]. Protein mass spectrometry (MS/MS), researchers determined sulfhydryl groups are especially vulnerable to electro­ structures of thioether conjugates extracted from bio­ phiIic attack and form thioether conjugates whose pres­ logical matrices [3-8, 13-22]. In these MS/MS experi­ ence indicates mechanisms of metabolic activation or ments, structural information was obtained from colli­ detoxification. Knowledge of protein targets and sites of sion induced dissociation (CIO). In fact, Murphy et al. covalent attachment can be used to identify reactive [20] concluded that fragmentation patterns of thioether metabolites, and these modified proteins can serve as conjugates have much in common, but that universal biomarkers of exposure to xenobiotics in clinical and recognition of class characteristic ions has been uncer­ epidemiological studies [1-8]. One important class of tain . xenobiotic-protein adducts forms via lA-Michael addi­ More detailed information regarding activation en­ tion of a,,B-unsaturated carbonyls to nucleophilic amino ergies for specific fragmentation reactions can be ob­ acid residues. Many a,,B-unsaturated carbonyls includ­ tained by generating energy resolved product ion spec­ ing acrylic monomers, combustion by-products or me­ tra. Energy resolved product ion spectra are acquired tabolites such as acrolein [1], and endogenous com­ by making incremental changes in collision cell poten­ pounds such as 4-hydroxynonenal react with protein tial and determining the dependence of product ion nucleophilic side chains via Michael addition [9-11]. abundances on the translational energy of precursor Mass spectrometry has emerged as a powerful ana­ ions [23, 24]. At the lowest collision energies, only lytical tool for the characterization of post-translation­ fragmentation reactions with the lowest activation en­ ally modified peptides and proteins because this ap­ ergies are accessible. Incremental increases in collision proach can identify the modifying group and sites of energy provide access to reaction pathways leading to modification [12]. Earlier studies demonstrated the util­ more and different product ions [23-28]. Normalized ity of fast atom bombardment mass spectrometry product ion abundances can be plotted to show their (FAB/MS) for identification of specific thioether conju- dependence upon collision cell potential. These graphs are termed energy resolved dissociation curves, and they assist efforts to optimize CIO experiments aimed at Address reprint requests to Dr. A. Daniel Jones, Facility for Adv anced structure determination or specific detection. Subtle Instrumen tation, University of California, Davis, CA 95616. E-mail: [email protected] differences in fragmentation dynamics are easily stud-

© 1997 American Society for Mass Spectrometry. Published by Elsevier Science Inc. Received July 15, 1996 1044-0305/ 97/51 7.00 Revised March 12, 1997 PI! 51044-0305(97)00032-9 Accepted March 13, 1997 728 OBERTH AND JONES J Am Soc Mass Spectrom 1997, 8, 727-736

Experimental

I. R, = -CH,CH,CHO } Materials II. R, = .CH,CII,CH,OH { III. R, = .CH,CII,CII=NOH Caution: Acrolein is a potent and toxic lachrymator and should be handled ill a fume hood to minimize exposure. Sodium borohydride is a toxicflammable solid and should be Y2 Z2 Y, handled carefully.

IV. R, = -CH,CH,CHO } N-Acety1cysteine, glutathione, and oxidized gluta­ -c~t~r~-,~ V. R, = ·CH,CII,CH,OIl thione (GSSG) were obtained from Sigma (S1. Louis, .. { VI. R, = -CH,CH,CH=NOH SR,~ MO). Acrolein, sodium borohydride (NaBH4), hydrox­ ~ a, RI( ylamine hydrochloride, trifluoroacetic acid (TFA), and Sephadex QAE-A-50, were obtained from Aldrich (Mil­ V2 Z2 Y, waukee, WI). Acetic anhydride, acetonitrile, and dithio­ JVII. R. = ·H } threitol were from Fisher (Pittsburgh, PA). Bond-Elut )-~d~jdL-=" \ VIII. R, = .CH,CH,CHO CBA solid phase extraction cartridges were from Ana­ SRx~ Iytichem (Varian, Harbor City, CA). All water used in ,------T- the following procedures was obtained from a Barn­ Rx a, steadlThermolyne (Dubuque, IA) Nanopure water fil­ tration system.

Synthesis g h Syntheses of conjugates were conducted by adding Figure 1. Structures of thioether conjugates used in this study. acrolein to solutions of various thiols, flushing the headspace of the reaction vials to remove oxygen, and ied by using tandem quadrupole instruments where keeping reaction vials wrapped in aluminum foil to ion-molecule collisions are low energy processes [23-25, prevent photochemical degradation. Reaction progress 27-331. In such experiments, most fragmentation is was monitored by using free zone capillary electro­ charge directed as fragmentation barriers are lowered phoresis (CE) with spectrophotometric detection at 214 by a nearby positive charge. nm. When reactions were complete, products were Electrospray and FAB ionization of peptides yield placed under a stream of nitrogen to remove unreacted protonated molecules in which the proton can be coor­ acrolein and acetonitrile, and the remaining moisture dinated by hydrogen bonding along several sites of the was removed by lyophilization. Synthetic thioether peptide backbone [29-35]. Because thioether conjugates conjugates in this study were stored under nitrogen at of N-acety1cysteine have fewer sites for proton coordi­ -30 °C. nation, their fragmentation behavior can be different than larger peptide thioether conjugates. N-Acetylcysteine Conjugates. S-(3-0xopropy1)-N-acetyl­ The principal aim of this study has been to determine L-cysteine (Compound I) was synthesized by dissolv­ relationships between structural features and fragmen­ ing 20 mg of N-acety1cysteine (0.12 mmoles) in one mL tation dynamics in protonated acrolein thioether conju­ of a 1:1 (v.v) acetonitrile:water solution. 20 ILL of gates by using low energy CID tandem mass spectrom­ acrolein (0.30 mmo1) was added by micropipette. 1H etry experiments. This has been accomplished by NMR results are in agreement with spectra published generating energy resolved product ion spectra [23-28] earlier [37, 40] (in d3-acetonitrile): 1.93 ppm (s, acetyl for a series of structurally related peptides that vary in Me, 3 H), 2.67 (tt, -5-CH2--CH 2-CH = 0, C-2, 2 H), 2.77 size and in the number of basic functional groups. (dt, -5-CH2--CH2-CH = 0, C-1, 2 H), 2.95 (dq, Cys-f3, 2 Acrolein thioether conjugates were chosen as model H), 4.55 (dt, Cys-«, 1 H), 6.89 (d, amide, 1 H), 9.65 (t, compounds for Michael-thioether conjugates. Acrolein -5-CH2--CH2--CH = 0,1 H). ESI MS results: m i z 220 is an important industrial chemical [1,361, a metabolite [M + Hl". of the cancer chemotherapeutic agent cyclophospha­ Conversion of 5.2 mg of I (0.024 mmol) to S-(3­ mide, [36-38] and an unwanted side-product of com­ hydroxypropy1)-N-acetyl-L-eysteine (Compound II) was bustion and cigarette smoke [38, 39]. Compounds for accomplished by dissolving I in 2 mL 95% ethanol! this study are acrolein conjugates of N-acetyl-L-cys­ water followed by addition of 1 mg NaBH4• The mix­ teine, glutathione, and N-acetylglutathione. Derivatives ture was sonicated for 20 minutes in an ultrasound of the aldehyde groups were prepared from these bath. An additional 1 mg of NaBH4 was added and the conjugates to assess whether common derivatization of mixture sonicated for an additional 10 min. Unreacted the aldehyde group would alter fragmentation behav­ NaBH4 was quenched by dropwise addition of 0.1 N ior. Structures of thioether conjugates used in this study HCI (until gas evolution stopped, pH 4.0). Purification are displayed in Figure 1. of II was performed by using a Sephadex QAE-A-50 J Am Soc Mass Spectrom 1997, 8,727-736 FRAGMENTATION OF PROTONATED ACROLEIN THIOETHER CONJUGATES 729 anion exchange column. The column was pretreated 2 H), 2.86 (m, Cys-{3, 2 H), 3.64, (t, - 5-CH2-CH2-eH 2­ with successive washes with absolute ethanol and wa­ OH, C-3, 2 H), 3.90 (t, Glu-a, 1 H), 3.98 (s, Gly-a, 2 H), ter. The quenched reaction product was loaded on the 4.56 (m, Cys-o, 1 H). EST MS: m / z 366 [M + H] + . anion exchange column and the column was washed To synthesize the oxime S-O-N-hydroxyiminopro­ with water to elute residual sodium ion. Compound II pyl)glutathione (Compound VI), 6.0 mg of IV (0.017 was eluted from the column with 0.1% trifluoroacetic mmol) was derivatized with 16.2 mg (0.23 mmol) hy­ acid in water. This eluted fraction was then lyophilized. droxylamine hydrochloride in 0.5 M ammonium acetate The 1H NMR spectrum agreed with a published spec­ pH 7.0. IH NMR (in D20 ): 2.13 ppm (q, Glu-{3, 2H), 2.49 trum [37] and showed no detectable starting material. (d, Glu-')', 2H), 2.64 (q, -5-CH2-eH2-eH = NOH NMR spectrum (in D20 ): 1.91 ppm (s, acetyl Me, 3 H), (trans), C-2, 2 H), 2.76 (m, -5-CH2-eH2-eH = NOH, 1.82 (m, -5-CH2-e!::h-CH2-0H, C-2, 2 H), 2.65 (t, c-i. 2H), 2.85 (q, -5-CH2-eH2-eH = NOH (cis), C-2, 2 -5-CH2-CH2-0H, C-l , 2 H), 29.95 (d of q, Cys-{3, 2 H), H), 3.05 (m, Cys-{3, 2 H), 3.74 [t (obscur ed by Gly-a), 3.67 (t, -5-C!:h-CH2-CH2-0H, C-3, 2 H), 4.36 (d of d, Glu-a, 1 H], 3.76 (s, Gly-a, 2 H), 4.59 (t, Cys-o, 1 H), 6.85 Cys-o, I H). EST MS results: m/ z 222 [M + H]+. (t, - 5-CH2-eH2-eH = NOH (cis), C-3, 1 H), 7.50 (t , Tosynthesize5-(3-N-hydroxyimmopropyl)-N-acety 1­ -5-CH2-eH2-eH = NOH (trans) C-3, IH). ESI MS: L-cysteine (Compound III, a mixture of two geometric m/z 379 [M + H]+. isomers), 4.9 mg of Compound I (0.022 mmoles) was dissolved in 1.0 mL of 0.2 M ammonium carbonate (pH N-Acetylglutathione Conjugates. N-Acetylglutathione (Com­ 8). 15 mg of solid hydroxylamine hydrochloride (0.22 pound VII) was synthesized by dissolving 16 mg (0.026 mmoles) was added to the above solution and mixed by mmoles) oxidized glutathione (GSSG) in 200 p.Lof acetoni­ using a vortex mixer. IH NMR (in D20 ): 2.05 ppm (s, trile:water (SO/SO v /v) followed by addition of 200 p.L (2.1 acetyl Me, 3 H), 2.51 (q, -5-CH2-eH2-eH = NOH mmoles) aceticanhydride. The mixture was held at 70°C for (trans), C-2, 2 H), 2.68 (t, -5-CH2-eH2-eH = NOH 18 hours, After cooling, volatile byproducts were removed (cis), C-2, 2 H), 2.77 (m, - 5-C!::h-eH2-CH = NOH, C-l , under a stream of nitrogen. Reduction of the disulfide bond 2 H), 3.0 (m, Cys-{3, 2 H), 4.4 (t of d, Cys-a, 1 H), 6.91 (t, of acetylated GSSG was accomplished by dissolving the -5-CH2-eH2-eH = NOH(cis), C-3, IH), 7.51 (t, glassy product in 1 mL of water followed by addition of 7.8 - 5-CH2-eH2-e!:! = NOH(trans), C-3, 1 H). EST MS mg (0.05 mmoles) dithiothreitol. The reduction was allowed results: m / z 235 [M + H] +. to proceed at 25 °C for four hours. Dithiothreitol was removed using a Sephadex QAE-A-50 anion exchange Glutathione Conjugates. To synthesize S-(3-oxopropyD­ column pretreated with successive washes with 100% glutathione (Compound IV), 10 p.L of acrolein in ace­ ethanol and water. The reaction products were loaded tonitrile solution (0.69 M) was added to 1 mL aqueous on the column, washed with 7 mL of water followed by solution containing 20 mg (0.065 mmol) of glutathione. elution of the product with 7 mL of acetonitrile contain­ Three additional 10 p.L aliquots of acrolein/ acetonitrile ing 0.1% trifluoroacetic acid . The eluted product was were added at successive one hour intervals. When the lyophilized. Confirmation of VII was performed by EST reaction was complete, the product was placed under a MS: m/z 350 [M + H]+. To synthesize S-(3-oxopro­ stream of nitrogen to remove unreacted acrolein and pyl)-N-acetylglutathione (Compound VIII) VII was acetonitrile, then frozen and lyophilized. IH NMR spec­ dissolved in 1 mL of water and 0.5 mL of this solution tra agree with published results [37] (in D20 ): 2.14 ppm was mixed with 4.6 mL (0.098 mrnol) of acrolein. The (q. Glu-{3, 2 H), 2.52 (dt, Clu-v, 2 H), 2.63 (t, -5-CH2-eH reaction mixture was left at 25 °C for 4 h and lyophi­ 2-CH = 0 , C-2, 2 H), 2.83 [t (obscured by overlapping lized . EST mass spectra confirmed formation of VIII: mat 2.881, -5-CH2-eH2-eH = 0 , C-l, 2 H), 2.88 (m, m/ z 406 [M + H]+. Cys-{3, 2 H), 3.79 (t, Glu-a, 1 H), 3.94 (s, Gly-a, 1 H), 4.58 (rn, Cys-o, 1 H), 9.65 (t, - 5-C!::h-eH2-eH = 0 , 1 H). ESI Instrumentation MS results: m / z 364 [M + H]+. Reduction of IV to S-(3-hydroxypropyDglutathione IH NMR spectra were recorded on a General Electric (Compound V) was accomplished by addition of 1 mg 0-300 (Fremont, CA) spectrometer (7 Tesla) operating NaBH4 to a solution of 6.0 mg of IV (0.017 mmol) in 1 at room temperature. IH chemical shifts are expr essed mL of 90% ethanol in water (v [ v) . The mixture was as ppm downfield from tetrameth ylsilane. All mass sonicated for 30 minutes in an ultrasound bath to spectra were acquired using a YG Quattro-BQ triple ensure mixing. An additional 1 mg of ~aBH 4 was quadrupole mass spectrometer (YG Biotech, Altrin­ added and sonicated for another 10 min. Unreacted cham, UK) connected to a microflow pump (model NaBH 4 was quenched as described above. The pH of p.LC-500, Isco Inc., Lincoln, NE). The solvent system (1:1 the mixture was raised to pH 9 with 1 M ammonium acetonitrile:water v:v) was deliver ed to the ionization carbonate and 1 M sodium hydroxide to ensure reten­ source at a flow rate of 5 p.L/min. Loop injections of 10 tion of V on the anion-exchange column for purification p.L were made by using analytes dissolved in a 1:1 as described for II. IH NMR (in D20 ): 1.79 (m, acetonitrile:water solution (approximate final concen­ -5-CH2-eH2-eH2-0H, C-2, 2 H), 2.17 (q, Glu-{3, 2 H), tration, 30 pmol/p.L) . The source temperature was held 2.54 (d, Glu-v, 2 H), 2.63 (t, -5-CH2-CH2-eH2-OH, c-r. at 65 °C, and the capillary voltage was set at +3.5 kY. 730 OBERTH AND JONES J A m Soc Mass Spectrom 1997, 8, 727-736

Table 1. Partial list of product ions observed at collision cell potential of - 50 V Product ion m/z ('Yo rel ative abundance}"

Loss of Loss of CzHzO CzHsNO Precursor ion b, y, Zz bz yz Acetyl (kete ne) (acetamide) 9 h SRx Rx N-AcCysteine NA e NA NA NA NA 43 (100) 122 (95) 105 (55) 76 (90) 59 (67) NA NA I NANA NA NA NA 43 (17) 178 (33) 161 (64) 132 (27) 115 (22) 89 (100) 57 (7) II NA NA NA NA NA 43 (19) 180 (43) 163 (100) 134 (18) 117 (54) 91 (78) 59 (12) III NA NA NA NA NA 43 (15) 194 (7) 176 (10) 147 (10) 130 (48)C 105 (55) 72 (93) Glutathio ne 130 (30) 76 (100)b 162 (27) 233 (7) 179 (30) NA NA NA 76 (100) 59 (6) NA NA IV 130 (60) 76 (23) NOe 289 (6) NO NA NA NA 132 (65) 115(75) 89 (40) 57 (16) V 130 (60) 76 (15) 220 (55) 291 (6) 237 (10) NA NA NA 134 (30) 117 (95) 91 (18) 59 (7) VI 130 (61)d 76 (52) 233 (17) 304 (12) NO NA NA NA 147 (13) 130 (61) 104 (100) 72 (40) VII 172 (35) 76 (32)b NO 275 (10) 179 (21) 43 (2) 307 (10) 291 (3) 76 (32) NO NA NA VIII 172 (27) 76 (7) NO 331 (11) NO NO NO NO 132 (40) 115 (5) 89 (12) NO aAbunda nce relative to base peak (excluding prec ursor) = 100. bProduct ion m/z 76 from GSH and VII parent may consist of mixt ures of isobars: y, and g. cProd uct ion m/z 130 from III may also include iso bar du e to loss of SRx to form protonated dehyd roalan ine. dproduct ion m/z 130 from VI may cons ist of an isobaric pair: b, and h. "Abbrev iatio ns: N-AcCyst ein e: N-acety l-L-cystei ne; NA. not applicable; NO. not detecte d « 2% base peak abunda nce).

The resolution of the first quadrupole was adjus ted to nom enclature for peptide cleavages as modified by give a peak w id th at half-he ight of about 0.75 Da. For Biemann [42]. The sing le lett er nomenclature used for tandem mass spectro metry experime nts, argon was proposed internal fragments has been described by introduced into the collision cell at a measured pressure Haroldsen et al. [13J and Ballard et al. [17J. No me ncla­ 1.0 X 10- 3 mbar. The resoluti on of the second qu adru­ ture for product ion s arising from cleavage at the sulfur pole was adjusted to give a peak width at half-height of ar e based on conventions prop osed by Deterding et al. 1.5 Da, This setting allowe d sufficient ion beam tran s­ [15], w here R x d enotes the xenobio tic moiety. mission to ensure sig nal-to-noise ratio adequa te for spectrum interpretation. Ion kin etic ene rgies th rou gh the first quadrupole were set to 2 eV. Results and Discussion Individual scans were averaged using Mass lynx" To facilit ate comparisons of fragm ents derived from the multichannel acquisition (MCA) mode. Product ion compo unds used in this study, a partial list of product spectra were obtained at 20 s/scan to ensure reliable ions generated by using a collision cell potential of -50 quantitative measurem ents of weak sig na ls whe re ion V is provided in Table 1. statistics limited precision of peak height measure­ ments. The collision cell potentials (negative) used to generate energy resolved product ion spectra we re 0, 5, N-Acetylcysteine Conjugates 10,20, 35, and 50 V. Preliminary high energy cm expe rime nts (E lab = 8000 eV, not shown) on the acrolein conjugate of N-acetyl­ Energy Resolved Dissociation Curves cysteine revea led the predominance of retro-Michael loss of acrolein from protonated I. In the current study, Energy resolved dissociation curves were prod uced by energy resolved product ion spectra of N-acetylcysteine plotting normalized product ion intensi ties against col­ we re obtain ed (not shown) to establish the relative ease lision cell po tential. Norm alized relative intensities of of fragme ntation of the bonds in N-acetylcysteine with­ specific product ions we re calculated by usin g the out influence of a modifying group. Spe ctr a of product formula: ion s d erived from protona ted N-acetylcysteine were obtained under the same collisiona l activation condi­ tions used through out the se experiments. The most prevalent fragmentation route for N-acetylcysteine was formati on of proton ated cysteine (m / z 122) via loss of

w here ij is the intensity of a selected product ion and ketene (C2H20 ). Ano the r product generated at all col­

2: ij + k ··· is the sum ma tion of all product ion intensities lision potentials was observed at m / z 76, which is with relative in tensity of 2% or high er . attributed to successive losses of ketene and CO 2 pro­ ducing frag me nt g (see Figure 1 for description). The Nomenclature acetyl ion (m / z 43) was observed at higher collision potentials. Peptide fragmenta tion nomenclature used throughout Representative product sp ectra from CID of proto­ this repor t is based on the Roep storff an d Fohlman [41] nated I are displ ayed in Figure 2. The d ominant frag- 50 100 150 200 100

Figure 2. CID product spectra of comp ound I at collision cell potenti als ranging from 0 to - 50 V. mentation pathway of I at lower collision energies potential. The most abundant product at the lowest involves bond cleavage between the sulfur and the potential again arises via cleavage of a sulfur-carbon acrolein moiety, yielding protonated N-acetylcysteine bond, but in the case of II this pathway involves loss of tm I z 164) via retro-Michael reaction (Figur e 3). At RxSH and formation of protonated N-acetyldehydroala­ higher collision energies, more products were gener­ nine (mlz 130) instead of loss of R, which was not ated including acetyl (mI z 43) and m I z 178 (loss of observed. CID spectra obtained at the three lowest ketene) analogous to the behavior of unmodified N­ potentials show two additional fragments in similar acetylcysteine. Fragmentation between the sulfur and yields: loss of acetamide (ml z 163) and loss of ketene the cysteine skeleton yields the protonated fragment (m I z 180). This indicates that under these conditions II SRx tm t z 89) which is the most abundant product at fragments via multiple pathways involve similar acti­ collision cell potentials greater than 30 V. Other frag­ vation barriers, but the activation energies of these ments formed under more energetic conditions are reactions are greater than for retro-Michael fragmenta­ protonated cysteine (mlz 122) and mt z 161 (loss of tion of I. At higher collision potentials, fragmentation of acetamide). Energy resolved dissociation curves for II yields SR x (m I z 91). Appearance of these fragments protonated I (Figure 4a) illustrate the ease of formation indicate that sulfur is an important cleavage point for I of the retro-Michael product which dominates the prod­ and II, but conversion of the aldehyde to the alcohol uct ion spectrum at low collision potentials (Ejab < 20 resulted in a significant increase in the activation energy eV). leading to loss of Rx . Reduction of the aldehyde to an alcohol, yielding II, The fragmentation behavior of oxime derivative III leads to marked changes in fragmentation dynamics is similar to I (Figure 4c). At low collision cell voltages, (Figure 4b). In contrast to I, dissociation of protonated the predominant fragmentation pathway yields the II yields no single dominant peak at any collision cell retro-Michael product, protonated N-acetylcysteine at 732 OBERTH AND JON ES J Am Soc Mass Spectrom 1997,8,727-736

collision cell potential, the predominant fragment ob­ served is Y2' corresponding to loss of the v-glutamyl group at m/z 179. This fragmentation route has been well documented in earlier studies [3, 4, 13, 15, 17, 20] and several investigations have taken advantage of this CompoundI. fragmentation by developing scans for neutral losses of Figure 3. Proposed mechani sm for retro-Michael fragm entation 129 Da which improve specificity for detection of glu­ of compound I. tathione conjugates.[3-8, 13, 17] At high collision cell voltages (50 V), the most abundant fragment results from cleavage at the cystei­ m / z 164. Similarities in fragmentation between I and III nyl-glycyl amide linkage to yield Yv protonated glycine point to a key role played by the Sp2 hybridization of the at m/z 76. Other prominent fragments are: b , the aldehyde or oxime groups in stabilizing the retro­ I glutamyl acylium ion (m/z 130); arH20 (m/z 84); Z2' Michael transition state. the Cys-Gly fragment minus the amino group of the

cysteine (m / z 162); and b2, the '}'-Glu-Cys fragment Glutathione Conjugates (m / z 233). At high collision cell potentials, the entire complement of band y ions is observed suggesting Energy resolved product ion spectra of glutathione sufficient proton mobility to facilitate fragmentation were obtained to establish the effect of the modifying along the peptide backbone. group on fragmentation of the peptide backbone (data Under all collision conditions, the protonated gluta­ not shown). These spectra were obtained under the thione adduct (IV) did not yield a detectable Y2 frag­ same collisional activation conditions used for all glu­ ment, but instead gave a dominant Y2-H20 fragment tathione adducts in this study, and several key frag­ (m/z 217, Figure 5) at lower collision energies. The ments are listed in Table 1. At all but the highest transient formation of Y2 cannot be ruled out, but was not observed at any cell potential or at lower collision 0.8 4a gas pressures. 07 --0---- 12289 Because the Y2 fragment is not observed, a scan for 06 -131 --0- 161 neutral loss of 129 Da may not be effective in screening ~ 05 -- 164 for acrolein thioether conjugates of glutathione in com­ ~O4 ----- 178 plex matrices. Other intramolecular Schiff base product S 03 ions formed from product ion g have been reported by 0.2 0.1 Haroldsen et al., [l3] and were observed in these 0.0 experiments as well. 0 to 20 30 40 50 At higher collision cell voltages the bI fragment of IV is abundant, suggesting protonation occurs near the v-glutamyl-cysteinyl amide linkage, but proton transfer 0.8 4b does not lead to formation of Y2 fragments. Such 0.7 - mlz 91 --0-- mlz 117 0.6 reactions have been reported in fragmentation of gluta­ - mlz130 0.5 --0- mlz 163 thione and other glutathione thioether conjugates. [3,4, ~ -- mlzl80 04 13, 17, 20] With the exception of the Y2 product, a full § OJ complement of band Y fragments is observed at a 0.2 collision potential of 50 V. 0.1 As discussed earli er, I retro-Michael elimination of 0.0 acrolein predominates at low collision cell voltages. 0 to 20 30 40 50 This fragmentation process is not observed in IV, and

only one fragment, SRx at m / z 89, can be attributed to 0.8 4c cysteinyl-C-S bond cleavage. The SRx fragment only 0.7 appears at higher collision cell voltages. - mlzl04 0.6 --0-- mlz 122 Energy resolved dissociation curves for IV are - mJzl30 ~ 05 --0- mJzl64 shown in Figure 6a. These plots highlight the predom­ ~ -- mJz192 __ 0.4 -- mJzl94 inant formation of Y2-H20. At low collision energies, all 0.3 other fragmentation routes combined contribute less than ten percent of total product ions. At collision 0.2 1 ----=~~=---=E;:::::::;2::::~=:::=~ 0.1 ,.. potentials of 20 V and higher, other fragmentation 0.0 g:;;==i~*==:;::::::~=::::;:::=;::===$;~'T""'"=:=~ o to W ~ 40 50 routes become energetically accessible, but the route Collision Cell Potential (V) that leads to formation of Y2-H20 remains the most favorable. Figure 4. Energy resolved dissociation curves of selected ions of N-acetylcy steine analogs: (a) Compound I; (b) Compound II; (c) As anticipated based on the behavior of II, reduction Compound III. of IV to S-(3-hydroxypropyl)glutathione (V) results in J Am Soc Mass Spectrom 1997, B, 727-736 FRAGMENTATION OF PROTONATED ACROLEIN THIOETHER CONJUGATES 733

+ 1.0 6a 0.9 0.8 ---{}- 89 84 0 7 - 115 c;; --0-- 132 'S 0.6 E-o -- 162 0.5 -- 217 S 0.4 289 0.3 0.2 0.1 0.0 0 10 20 30 40 50

1.0 6b 0.9 --{}--91 84 0.8 - 117 .. 0.7 --0-- 130 ~ 0.6 --_ 220134 0.5 - 291 S -+- 237 0.4 0.3 0.2 0.1 0.0 0 to 20 30 40 50

1.0 6c 0.9 08 0.7 --a--130 104 ] 0.6 _ 147 .s 0.5 --0-- 217 --.- 232 - 0.4 ---Q- 304 - 0.3 0.2 Figure 5. Proposed structure of Y2-HZO fragment observed for 0.1 compounds IV, VI, and VIII. 0.0 0 10 20 30 40 50 Collision Cell Potential (V) product spectra with more fragments (data not shown). Figure 6. Energy resolved dissociation curves for selected prod­ At a collision cell potential of zero volts, Y2 (ml z 237) uct ions derived from CID of protonated glutathione analog and Zz (m / z 220) fragments appear with low fractional precursors: (a) Compound IV; (b) Compound V; (c) Compound abundances. At all lower collision cell potentials tested, VI. the Zz fragment dominates the spectra, but at higher collision potentials the internal fragment h (m / z 118) and b] become most prevalent. The internal fragment g collision potentials tested that produce minor amounts (m I z 134) is also present at higher collision cell volt­ of Y2 and Z2 fragments. Desp ite their low activation ages. An intramolecular Schiff base is not anticipated energy barriers, these two fragmentation routes are not from decomposition of protonated V as it does in the the most favored routes at higher collision potentials. In

yz-H20 fragment from IV, because the aldehydic func­ fact, the Y2 peak is diminished at collision cell potentials tionality of the acrolein conjugate is not present. As a greater than 20 V, perhaps due in part to subsequent result, no fragments indicating Schiff base formation are fragmentation of the yz ion. present. Whe n IV is derivatized with hydroxylami ne hydro­ Product spectra from protonated V show that frag­ chloride to give the acrolein-oxime product VI (data not ments resulting from cleavage at the propanol-C-S bond sho wn), the fragment representing cleavage at the Cys­

are minor although cleavage of the Cys-C-S bond gives C-S bond (SRXI m/z 104) dominates at higher collision

a more abundant fragment SRx at m/ z 91. This frag­ cell potentials. This same fragment is prominent in the mentation pathway was observed, however, for all product ion spectrum of III taken at a collision voltage thioether conjugates in this study at a collision potential of 35 V. A minor fragment, R, at m l z 72 also is present of 50 V. at higher collision cell voltages, which is due to cleav­ Energy resolved product ion spectra of V show the age of acrolein-oxime-C-S bond. This fragment is entire complement of b and y ions and substantiate our analogous to the R, peak at m/ z 56 seen in the product earlier finding that reduction of the acrolein moiety spectra of IV. produces more fragments at any given collision poten­ On e notable feature in the fragm entation of proto­ tial tested. Energy resolved dissociation curves of V nated VI is the appearance of an abundant product at (Figure 6b) are different from those observed for IV m/ z 217. The low appearance energy of this peak because there are two fragmentation routes at the low (Figure 6c) is analogous to that seen for m / z 217 for IV, 734 OBERTH AND JONES J Am Soc Mass Spectrom 1997, 8, 727-736

which would involve displacement of hydroxylamine 1.0 7a in a manner analogous to displacement of water from 0.9 76 -iJ- 83 IV. Formation of this product suggests both oximes and 0.8 - 130 aldehydes are reactive toward neighboring amide nitro­ ~ 0.7 - 172 179 gens, and their presence may facilitate cleavage of these 0.6 amide bonds. The energy resolved dissociation curves ] - 275 .s 0.5 (Figure 6c) for VI are similar in appearance to those - ;::; 0.4 obtained for IV. Both have accessible fragmentation pathways with low activation energy barriers. These 0.3 results combined with the fragmentation behavior of III 0.2 suggest that derivatization of Michael thioether adducts 0.1 of aldehydes or ketones to form oximes is unlikely to 10 20 30 40 50 yield more useful fragmentation than can be obtained for the underivatized adduct. Fragments at m/z 84, 130, 162, and 179 are observed in the product spectra of both IV and VI. Product 1.0 7b spectra of V also include fragments at m / z 84 and 130. 89 0.9 All of these fragments arise from the glutathione por­ --iJ- 115 tion of the molecule and therefore do not provide 0.8 132 ~ 172 information regarding the nature of the modifying 0.7 - 217 group. -; 0.6 - 331 2 0.5 - S N-Acetylglutathione Conjugates 0.4 0.3 Energy resolved product ion spectra of N-acetylgluta­ thione (not shown) exhibit marked differences from 0.2 glutathione. At the lowest collision cell potentials 0.1 tested, several fragmentation pathways have compara­ 40 50 ble activation energies resulting in multiple product ions . These spectra suggest that the precursor ion is Collision Cell Potential (V) heterogeneous with respect to site of protonation, be­ Figure 7. Energy resolved diss ociation curves of selected prod­ cause several peaks are present including the entire uct ions derived from cm of protonated N-acetylglutathione complement of band y ions. analogs: (a) Compound VII; (b) Compound VIII . Energy resolved dissociation curves for compound VII are shown in Figure 7a. A notable feature is the abundance of fragments at lower collision potentials. exclusively b2 (m / z 331). At a collision potential of 5 V, Acylation at the N-terminus eliminates the basic amino a second fragment representing Y2-H20 (m/ z 217) group, thus protonation becomes more de localized than accounts for about 30% of products. At higher poten­ for glutathione. tials the g fragment (m / z 132) is most abundant, while As expected, the fragmentation behavior of N-acetyl­ the rest of the fragments share similar relative intensi­ S-(3-oxopropyl)-glutathione (VIII) shares many charac­ ties. teristics observed for VII as well as IV. As was the case In low energy CIO experiments, most fragmenta­ in product spectra of IV, the N-acetylated analog VIII tions are charge directed [26, 28, 31, 32, 34]. Charge­ does not yield Y2' but Y2-H20 (m/z 217), again with a remote fragmentation usually involves high activation low activation energy. Based on the ease of formation of energies [31, 32, 35, 43-45]. It is expected to play a Y2-H20, it is proposed that acrolein may serve as a minor role in our experiments, because center-of-mass volatile derivatizing agent for cysteine residues capable collision energies exceeding 10 eV are often required for of inducing low activation energy fragmentation on the charge-remote fragmentations to become dominant [43, N-terminal side of cysteine groups in CIO experiments. 45]. Such reactions cannot be ruled out because center­ Retro-Michael fragmentation of VII would yield a of-mass collision energies in the experiments described fragment with m / z 350. A small peak at this mass was herein range between 4-8 eV, and these experiments observed, but only at the highest collision potential were conducted under multiple collision conditions. tested. These results indicate that even in the absence of lf the fragmentation processes with the lowest acti­ a basic amino group, protonation at the thioether sulfur vation energies are charge directed, the dominant retro­ was not significant when multiple amide bonds were Michael fragmentation for I and III suggests protona­ present. tion occurs at the sulfur for these N-acetylcysteine Energy resolved dissociation curves for VIII are derivatives. Furthermore, II fragments primarily at the shown in Figure 7b. At a collision potential of 0 V, VIII sulfur at low collision potentials, only the fragmenta­ cleaves at the cysteinyl-glycyl amide bond to give tion is loss of SRx (m/z 130). While it is widely believed J Am Soc Mass Spectrom 1997, 8, 727-736 FRAGMENTATION OF PROTONATED ACROLEIN THIOETHER CONJUGATES 735

that the peptide backbone is an important site of Reduction of the aldehydic carbonyl of thioether protonation [29, 30, 32, 34, 35, 46], the differences in conjugates formed from Cl:,f3-unsaturated aldehydes or behavior of the compounds investigated herein can be ketones is an effective method of chemical modification attributed to several factors. In aqueous solution, pro­ that prevents retro-Michael fragmentation in N-acetyl­ tonated peptide N-terminal amino groups are stabilized cysteine Michael adducts. Conversion of the aldehyde by solvation [29, 46]. In the gas phase, charge delocal­ group to an oxime had only minor effects upon frag­ ization can be accomplished by internal solvation of the mentation. charge by amide groups [29,34,47]. The proton affinity Acrolein adducts of glutathione (IV) and N-acetyl­ of formamide is 198.8 kcal/mol [47], which approxi­ glutathione (VIII) did not give sign ificant Y2 fragments mates the intrinsic proton affinity of a single amide under any conditions, but both yielded abundant Y2 ­ group. Additional amide groups, such as those present H20 fragments via a low activation energy pathway. in the glutathione conjugates, enhance the proton affin­ Strategies that rely on specific neutral losses based upon ity, as illustrated by the proton affinities of dipeptides elimination of the glutamyl group should be conducted Cly-Cly and Ala-Ala, which are 208 and 210 kcal/mol, with this in mind. Such specific cleavages on the N­ respectively [48]. Proton affinities increase with each terminal side of acrolein adducts of cysteine residues additional amide bond with a maximum effect reached might be exploited to localize cysteine sulfhydryls in at three amide linkages [48]. In addition, coordination peptides. of a proton with up to three proton acceptors is possible (albeit with diminishing bond strength per coordinate) [46,47]. Since the N-acetylcysteine derivatives I, II, and Acknowledgments III have only one amide group, the proton is less likely The authors acknowledge the technical assistance of Wanda Smith to be localized at an amide group, and the thioether of the UC Davis NMR Facility for generating NMR spectra. The group is more able to serve as a site of protonation. NMR spectrometer was pu rchased with funds from NSF grant no. Proton affinities of formamide [47], propionaldehyde DIR90-16484. This work has been supported by NIEHS Superfund Basic Research Program (grant no. 2P42 ES-04699) and the NIEHS [49], and diethylsulfide [49] are 198.8, 193.6, and 205.6 Center for Environmental Health Science at DC Davis (grant no. kcal/mol, respectively. Proton affinity of the sulfur in 1P30 ES-05707). This work was presented in part at the 43rd acrolein conjugates may be enhanced by coordination of Annual ASMS Conference on Mass Spectrometry and Allied the proton with the acrolein aldehyde and the thioether Topics, held in Atlanta, GA, May 21-25, 1995. sulfur, with a bridging proton forming a six-membered ring. Differences in fragmentation behavior between References N-acetylcysteine derivatives and glutathione deriva­ tives can be explained by the ranking of basicities 1. van Welie, R. T. H.; van Dijck, R. G. J. M.; Vermeulen, N. P. E.; (single amide < thioether < multiple amides) and van Sittert,N. J. Crit. Rev. Toxieo/. 1992,22,271-306. charge-directed fragmentation. 2. Nelson, E. Crit. Rev. Toxieol. 1992, 22, 371-389. 3. Pearson, P. G.; Threadgill. M. D.; Howald, W. N.; Baillie, T. A. Biomed. Environ. Mass Spectrom . 1988, 16,51-56. Conclusions 4. Straub, K M. In Mass Spectrometry in Biomedical Research; Gaskell, S. J., Ed.; Wiley: New York, 1986; pp 115-134. Retro-Michael fragmentation is the lowest activation 5. Davis , M. R.; Baillie, T. A. J. Mass Spectrom. 1995, 30, 57-68. energy fragmentation pathway for protonated acrolein 6. Davis, M. R.; Kassahun, K.; [ocheirn, C. M.; Brandt, K M.; conjugates of N-acetylcysteine, and the retro-Michael Baillie, T. A. Chem. Res. Toxicol. 1993, 6, 376-383. product dominates low energy CID spectra of these 7. Kassahun, K; Hu, P.; Grillo, M. P.; Davis, M. R.; [in, L.; Baillie, compounds even under multiple collision conditions. T. A. Chem. Bio!. Interact. 1994, 90, 253-275. 8. [in, L.; Davis, M. R.; Kharasch, E. D.; Doss, G. A.; Baillie, T. A. The low barriers to these reactions are attributed to Chem. Res. Toxicol. 1996, 9, 555- 561. protonation at thioether sulfur and stabilization of the 9. Bruenner, B. A.; Jones, A. D.; Germ an, J. B. Chem. Res. Toxicol. retro-Michael transition state owing to the electronic 1995, 8, 552-559. configuration of these molecules. Similar behavior can 10. Esterbauer, H.; Zollner, H.; Scholz, N. Z. Naturjorsch. 1975, 30c, be expected for other Michael addition conjugates of 466-473. N-acetylcysteine. 11. Bolgar, M.S. ; Gaskell, S. J. Anal. Chem. 1996, 68, 2325-2 330. 12. Kaur, 5.; Hall, S. c, Burlingame, A. L. Prog. cu«. Bio!. Res. In contrast, retro-Michael fragmentation was not a 1991, 372,107-11 7. competitive reaction for glutathione and N-acetylgluta­ 13. Harold sen, P. E.; Reilly, M. H.; Hu ghes, H.; Gaskell, S. J. thione conjugates of acrolein. For these compounds, the Biomed.Emnron. Mass Spectrom, 1988, 15, 615-621. proton is more effectively sequestered by the amide 14. Pearson, P. G.; Slatter, J. G.; Rashed, M. S.; Han, D.-H.; Grillo, groups. Spectra of glutathione and N-acetylglutathione M. P.; Baillie, T. A. Biochem. Biophys. Res. Commull. 1990, 166, conjugates show a complement of band y type ions to 245-250. illustrate this point. Retro-Michael fragmentation of the 15. Deterd ing, L.J.; Srinivas, P.; Mahmood, N. A.; Burka, L. T.; Tomer, K B. Anal. Biochem. 1989, 183, 94-107. N-acetylglutathione adduct VIII was minimal, suggest­ 16. Ackerman, B. L.; Watson, J. T.; Newton, J. F.; Hook, J. B.; ing that the peptide backbone efficiently sequesters the Braselton, W. E., Jr. Biomed. Mass Spectrom. 1984,11,502-511. proton and diminish protonation of thioether sulfur, 17. Ballard , K; Raftery, M. J.;[aeschke, H.; Gaskell, S. J. J. Am. Soc. even in the absence of a basic N-terminal amino group. Mass Specirom . 1991, 2, 55-68. 736 OBERTH AND JONES J Am Soc Mass Spectrom 1997, 8, 727-736

18. Baillie, T. A.; Pearson, P. G.; Rashed, M. S.; Slatter, J. G.; 32. Burlet, 0.; Yang, c.-Y.; Gaskell, S. J. J. Am. Soc.Mass Spectrom . Howald, W. N. In Biological Mass Spectrometry; Burlingame, 1992, 3, 337-344. A 1.; McCloskey, J. A, Eds.; Elsevier: Amsterdam, 1990; pp 33. Poulter, 1.; Taylor, 1. C. E. Int. J. Mass Spectrom. Ion Proc. 1989, 567-578. 91, 183-197. 19. Pearson, P. G.; Howald, W. N.; Nelson, S. D. Anal. Chem. 1990, 34. Tang, X.-J.;Boyd, R. K. RapidCommun. Mass Spectrom . 1992,6, 62, 1827-1836. 651-657. 20. Murphy, C. M.; Fenselau, C: Gutierrez, P. 1. J. Am. Soc. Mass 35. Cox, K. A.; Gaskell, S. J.; Morris, M.; Whiting, A J. Am. Soc. Spectrom. 1992, 3, 815-822. Mass Spectrom . 1996, 7, 522-531. 21. Lay, J. 0., Jr.; Potter, D. W.; Hinson, J. A Biomed. Environ. Mass 36. Cohen, S. M.; Garland, E. M.; St. John, M.; Okamura, T.; Smith, Spectrom. 1987, 14, 517-521. R. A. Cancer Res. 1992, 52, 3577-3581. 22. Heeremans, C. E. M.; te Koppele, M.; Niessen, W. M. A; van J. 37. Ramu, K.; Fraiser, 1. H.; Mamiya, B.; Ahmed, T.; Kehrer, J. P. der Greef, Brussee, La Vos, G. F.; ten Neover de Brauw, J.; J.; Chem. Res. Toxicol. 1995,8, 515-524 . M. C. Biomed. Environ. Mass Spectrom. 1988, 17, 181-186. 38. Patel, J. M.; Block, E. R. Toxicol. Appl.Pharmacal. 1993, 122,46-53. 23. McLuckey, S.; Glish, G. 1.; Cooks, R. G. Int. J. Mass Spectrom. 39. Mahut, B.; Delacourt, c, de Blik, J.;Mani, T. M.; Scheinmann, Ion Phys. 1981,39,219-230. P. Chest 1993, 104, 1286-1287. 24. Fetterolf, D. D.; Yost, R. A Int. J.Mass Spectrom. IonPhys. 1982, 40. Hashmi, M.; Vamvakas, S.; Anders,M.W. Chem. Res. Toxieo/. 44,37-50. 1992, 5, 360-365. 25. Nacson, S.; Harrison, A G. Int. J. Mass Spectrom. Ion Proc. 1985, 63, 325-337. 41. Roepstorff, P.; Fohlman, J. Biomed . Environ . Mass Spectrom. 1984, 11, 601. 26. McLuckey, S. A; Cooks, R. G. In Tandem Mass Spectrometry; McLafferty, F. W., Ed.; Wiley: New York, 1983; pp 303-320. 42. Biemann, K. Biomed. Environ . Mass Specirom. 1988, 16, 99-111. 27. Dawson, P. H.; Douglas, D. J. In Tandem Mass Spectrometry; 43. Adams, J. Mass Spectrom. Rev. 1990, 9, 141-186. McLafferty, F. W., Ed.; Wiley: New York, 1983; pp 125-148. 44. Alexander, A J.; Thilbault, P.; Boyd , R. K. RapidCommun.Mass 28. Hayes, R. N.; Gross, M. 1.; McCloskey, J. A., Eds.; In Methods Spectrom. 1989, 3, 30. in Enzymology, Vol. 193; Academic: New York, 1990; pp 45. Feistner, G. J.; Hsieh, 1. 1. J. Am. Soc. Mass Spectrom . 1995, 6, 237-263. 836-846. 29. Hunt, D. F.; Yates, J. R., III; Shabanowitz, J.; Winston, S.; 46. Yeh, R. W.; Grimley, J. M.; Bursey, M. M. BioI. Mass Spectrom. Hauer, C. R. Proc. Nat!. Acad. Sci. USA 1986, 83, 6233-6237. 1991,20, 443-450. 30. Tang, X.-J.; Thibault, P.; Boyd, R. K. Anal. Chem. 1993, 65, 47. Meot-Ner (Mautner), M. J. Am. Chem. Soc. 1984, 14, 278-283. 2824-2834. 48. WU, J.; Lebrilla, C. B.J. Am. Soc. Mass Spectrom . 1995,6,91-101. 31. Burlet, 0.; Orkiszewski, R. S.; Ballard, K. D.; Gaskell, S. J. Rapid 49. Harrison, A G. Chemical Ionization Mass Spectrometry, 2nd ed.; Commun. Mass Spectrom . 1992, 6, 658-662. CRC: Boca Raton, 1993.