Anal. Chem. 1996,67,390-398
Mass Spectrometric Identification of Amino Acid Transformations during Oxidation of Peptides and Proteins: Modifications of Methionine and Tyrosine Swapan K. Chowdhury,* Jamshid Eshraghi,t Henry Wotfe, Diane Forde, Allan 0. Hlavac, and David Johnston Analytical Sciences Department, Sterling Winthrop Phamaceutical Research Division, 7250 South Collegeville Road, Collegeville, Pennsylvania 79426-0900
Iiquid chromabgraphy/electrosprayionization mass spectrometry (LC/ESI-MS), tandem mass spectrometry with on-lineliquid chromatography(LC/ESI-MS/MS)and highresolution mass spectrometrywith liquid secondary ionization (LSI-MS) were utilized to identify the modified amino acids in peptides and proteins formed during oxidation with performic acid. The procedure of protein oxidation was chosen to assist in protein unfolding by oxidizing the cystines to cysteic acids to allow for more complete proteolytic digestion and to create additional cleavage sites for endoproteinaseAsp-N. Investigation of the Asp-N peptide map of oxidized superoxide dismutase (SOD)by LC/ESI-MSrevealed that an proteolytic & w e n t of the protein was missing. In its place, two peptides with molecular weights 66 and 100higher than that calculated for the missing peptide were observed. To identify the modified amino acids in the unexpected peptides, a model peptide with some amino acid similarities (tyrosine, arginine, methionine, lysine) to the missing peptide was chosen and was subjected to similar oxidation and enzymatic digestion steps, conditions, and reactions. After oxidation and digestion, the model peptide (TU'; sequence, Ac-MDKVLNRY) showed three major peaks in LC/MS. The peptides in the three peaks were identitied as the unmodified peptide and two peptides whose molecular weights were 66 and 100 higher than that of TAP. The LC/ESI-MS/MS of these reaction products demonstrated that in the two modified peptides the N-terminal methionine has been transformed into methionine sulfone (molecularweight increase of 32),whereas m/z values of all the fragmentions containing C-terminal arginine and tyrosine were 34 and 68 higher than those in the unmodi6ed peptide. To establish whether the arginine or tyrosine or both were modified, another peptide neuromedin N (sequence, IUPYIL) was chosen that contains tyrosine but no arginine or methionine. The LC/MS analysis of the oxidized peptide again showed three peaks. The early-eluting peak corresponds to residual unmodified peptide and the molecular weight of the two later-elutingpeptides are respectively 34 and 68 higher than that of neuromedin N. The LC/ESI-MS/MS of the peptides confirmed that the tyrosine underwent transformations with an increase of molecular weight by 390 Analytical Chemistry, Vol. 67,No. 2, January 75, 7995
34 or 68. An accurate molecular weight measurement together with the determination of the atomic composition of the modified peptides showed the presence of one and two chloro substitutionsin tyrosine in the two later-eluting peptide peaks, respectively. This finding is in agreement with the detection of 3-and 3,5-dichlorotyrosinesin acid hydrolysis (Sanger, F.; Thompson, E. 0. P. Biochim. Biophys. Acta 1963,71,468).We believe that sodium chloride used as buffer in peptides and SOD solutions produced electrophiles in the oxidizing medium that caused electrophilic aromatic substitution in tyrosine. This observation was further supported by the identification of mono and dibromo peptides when chloride salts were replaced with bromides. Oxidation of proteins with performic acid is a well-established analytical procedure that has been used extensively' for cleaving disulfide bonds in proteins since its introduction by Sanger.2 The procedure has also been used in amino acid analysis of proteins. The composition and content of the different amino acids present in proteins is determined by dissociating proteins into individual amino acids by acid hydrol~sis.~ The direct quantitation of labile residues, such as methionine and cystine, is difficultbecause these sulfurcontaining amino acids are destroyed to different degrees by acid hydrolysis+ and therefore, these amino acid residues are routinely quantified by conversion to more stable species prior to hydrolysis. One approach to obtain more stable species is to convert cysteines and cystines to cysteic acids and methionine to methionine sulfone by oxidation with performic acid. This method converts both cysteines and cystines into one form, cysteic acids, allowing the total determination of this amino acid content of a protein. The disulfide bonds, if present, are broken and the two cysteines are converted into cysteic acids, which can cause the protein to unf0ld.5-~ The unfolding of the protein can be used to t Current address: Philadelphia College of Pharmacy and Sciences, 600 S. 43rd St, Philadelphia, PA 19104. (1) Hrs, C. H.W. Methods Enzymol. 1967,11, 197-199. (2)Sanger, F.Biochem. J. 1949,44, 126. (3) (a) Moore, S.;Spackmann, D. H.; Stein, W. H. Anal. Chem. 1958,30,1185. 6)Speckmann, D. H.; Stein, W. H.; and Moore, S. Anal. Chem. 1958,30, 1190. (4) Elkin, R G.; Griffith, J. E. J.--Assoc. Ofi Anal. Chem. 1985, 68, 11171121,and references therein. (5) (a) Schram, E.;Moore, S.; Bigwood, E. J. Biochem.]. 1954,57,33-37.(b) Moore, S.J. Bid. Chem. 1963,238,235-237.
0003-2700/95/0367-0390$9.00/0 0 1995 American Chemical Society
expose refractory cleavage sites on protease-resistant proteins, such as superoxide dismutase. The exposure of additional enzymatic or chemical cleavage sites following oxidation-induced unfolding of a protein enables production of smaller fragments, which can be characterized by mass spectrometric peptide mapping, tandem mass spectrometry, or a combination of both. The increase in molecular weight of a peptide or a protein due to oxidation of cysteines and cystines can be used to determine the number of such residues present and the number of those involved in disulfide bonds by mass spectrometry.8 The presence of cysteic acids is also shown to simplifymass spectrometry/mass spectrometry (MS/MS) spectra of peptides containing arginine at the C-terminus by promoting a given series of fragment ions,g which provide sequence information on the peptide. Despite the important utility of the performic acid oxidation procedure, the method has some shortcomings. Whereas the oxidation of methionine, cysteine, and cystine generates stable species, other amino acids, such as tryptophan, tyrosine, serine and threonine, were observed to undergo undesired side reactions.' Modification of tyrosines proceeds slowly under mild conditions, while oxidation of alkyl hydroxy amino acids requires drastic conditions. The actual nature of modifications of these amino acids is not well understood. We have used the performic acid oxidation procedure to convert the cysteine and cystines of bovine super oxide dismutase (SOD) into cysteic acids, creating new cleavage sites (cysteic acids) in addition to the aspartic acids recognized by the endoproteinaseAspN. AspN is an enzyme that cleaves proteins at the amino terminus of the acidic residue aspartic acid. Following the protein oxidation, digestion with AspN, and peptide mapping by liquid chromatography/electrospray ionization mass spectrometry (LC/ESI-MS),l0J we found that one of the expected fragments of the protein was missing. Instead, several peptides with unexpected molecular weights were observed. The present investigation was undertaken to (a) identifythe unknown peptides obtained in the peptide-mapping procedure following performic acid oxidation of SOD, (b) pinpoint the amino acids that have undergone changes, and (c) determine the nature of modifcations of those amino acids. In addition, two commercially available peptides, neuromedin N and TAP (simian virus tumor antigenic peptide), which have some amino acid similarities to the missing peptide were also investigated. Liquid chromatography coupled to eledrospray ionization mass spectrometry (LC/ESI-MS) tandem mass spectrometry with on line LC (LC/ESI-MS/MS),llb-eand liquid secondary ionization (6)Hirs, C. H. W.J Bid. Chem. 1956,219,611-621. (7) Moore, S.J. Bid. Chem. 1963,238,235-237. (8) Sun, Y.; Smith, D. L. Anal. Biochem. 1988,172,130-138. (9)Burlet, 0.; Yang, C.-Y.; Gaskell, S. J. J. Am. SOC.Mass Spectrom. 1992,3, 337-344. (10)Eshraghi, J.; Chowdhury, S. K Anal. Chem. 1993,65, 3528-3533. (11) (a) Whitehouse, C. M.; Dreyer, R N.; Yamashita, M.; Fenn, J. B.Anal. Chem. 1985, 57, 675-679. (b) Huang, E. C.; Henion, J. D. J. Am. Soc. Mass Spectrom. 1989,1,158-165.(c) Can, S.A; Hemling, M. E.; Bean, M. F.; Roberts, G. D. Anal. Chem. 1991,63,2802-2824.(d) Hunt, D.F.; Michel, H.; Dickinson, T. A; Shabanowitz, J.; Cox, A L.; Sakaguchi, IC;Appella, E.; Grey, H. M.; Sette, A Science 1992, 256, 1817-1820. (e) Hunt, D. F.; Henderson, R A; Shabanowitz, J.; Sakaguchi, K; Michel, H.; Sevilir, N.; Cox, A L.; Appella, E.; Engelhard, V. H. Science 1992,255, 1261-1266.
mass spectrometry (LS1-MS)l2 were utilized for the present studies. Finally, the elemental composition of the peptides prior to and after oxidation and enzymatic digestion was determined using the peak-matching technique in SI-MS at high resolution. EXPERIMENTALSECTION
(a) Materials. Trinuoroacetic acid (HPLC grade), acetonitrile (HPLC grade), sodium phosphate (monobasic, monohydrate), sodium phosphate (dibasic, heptahydrate), and sodium chloride (crystalline) were obtained from Baker Analyzed; 2-methoxyethano1 (glass distilled), 30% hydrogen peroxide (semiconductor grade), and formic acid (95-97%) were from Aldrich; Tris (hydroxymethyl)aminomethane was from International Biotechnologies, Inc.; and distilled water was prepared with Milli-Q distillation apparatus from Millipore System. Bovine superoxide dismutase13was obtained from DDI pharmaceuticals (Mountain View, CA); neuromedin N (sequence, KIPYIL) was from Sigma; simian virus tumor antigenic peptide TAP (sequence,Ac-MDKVLNRY) was from Bioscience, Inc.; and endoprotease AspN (sequencing grade) was from Boehringer Mannheim. (b) Protein Oxidation and Enzymatic Digestion. (i) Performic Acid Oxidation. Performic acid solution for protein oxidation was prepared by mixing 0.5 mL of 30% hydrogen peroxide with 9.5 mL of formic acid. The mixture was allowed to stand at room temperature for 2-2.5 h before use. A diluting buffer solution was prepared by dissolving 5.21 g of sodium phosphate (monobasic,monohydrate), 3.28 g of sodium phosphate (dibasic, heptahydrate), and 8.5 g of sodium chloride in 1000 mL of purified water. Sample solutions were made by dissolving the peptides (neuromedin-N and TAP) separately in the diluting buffer to obtain a concentration of 1mg/mL and 6 mg /mL SOD. Equal volumes (0.2 mL) of a sample solution and the performic acid solution were mixed in a microcentrifuge tube and were allowed to stand for 2 h in dark in an ice/salt bath at -10 to 0 "C. The reaction was stopped by freezing in a dry ice/2-propanol bath and then removing the acid in a vacuum centrifuge (Savant). After all the liquid has been removed, the dried material was resuspended in 0.4 mL of puritled water for either digestion with AspN or analysis by liquid chromatography (LC) or LC/MS. (ii) Digestion with Asp-N. Approximately 2 pg of the enzyme AspN was dissolved in 75 pL of distilled water in a microcentrifuge vial. A 125pL aliquot of 0.1 M Tris buffer @H adjusted to 7.5 by adding 0.1 N HC1) was added to the vial containing AspN, and the enzyme was dissolved thoroughly in the buffer. Oxidized sample (100 pL) was added to the enzyme solution and mixed gently using a vortex mixing apparatus. The digestion was allowed to continue for 6 h with continual mixing at 37 "C in the case of SOD and 4 h in the case of neuromedin-N and TAP. To avoid possible autodigestion of the enzyme, the peptides were digested for a shorter period than SOD because there is no aspartic acid or cysteic acid in these peptides. The digestion was stopped by freezing the solution at -20 "C until the material was subjected to further experiments. (iii) Scaleup Oxidation, Asp-N Digestion, and Isolation of Peptides. The same oxidation procedure was performed on a large scale so that oxidized peptides could be isolated for LSI-MS (12) (a) Barber, M.; Bordoli, R S.; Sedgwick, R D.; Tyler, A N.J. Chem. Soc.,
Chem. Commun. 1981,6,325.@) Aberth, W.; Straub, IC M.; Burlingame, A L.Anal. Chem. 1982.52,2029. (13) Steinman, H.M.; Vishweshwar, R; Naik, J. L.; Abemethy, J. L; Hill, R L. J. Biol. Chem. 1974,249,7326-7338.
Analytical Chemistry, Vol. 67, No. 2, January 75, 7995
391
determination and accurate molecular weight measurement. Largescale oxidation and AspN digestion was performed on neuromedin N only. A 25 mg sample of neuromedin N was dissolved in 25 mL of the diluting buffer and dispensed in 500 pL aliquots into each of 50 1.5 mL polypropylene centrifuge tubes. Freshly prepared performic acid (500 pL, 2.5 mL of 30%hydrogen peroxide diluted to 25 mL with formic acid) was added to each tube. Each tube was immediately capped, vortexed, incubated in an ice/salt bath for 3 h, then dried by centrifugal lyophilization, and stored at -20 "C until needed for further experiments. Oxidized peptide sample (12 mg) was dissolved in the HPLC solvent A (0.1%aqueous TFA) and injected onto a Vydac C18 (25 x 2.5 cm) HPLC column equilibrated with an 8020 mixture of solvents A and B (0.1%TFA in acetonitrile) respectively. The solvents were delivered by a Waters 840 gradient HPLC system (Waters, Milford, MA). The peptides were eluted using a linear gradient, and the two main peaks (see later) were collected and dried by centrifugal lyophilization. The elution gradient was as follows: time (min) init 22 62 67 72
flow rate (mL/min) 10 10 10 10 10
A / B ratio a020 8020 6535
0:lOO 0100
(c) Mass Spectrometry. (i) LC/ESI-MS and LC/ESI-MS/
MS. The electrosprayionization mass spectrometer,its coupling to an on-line liquid chromatograph, and its operation have been described in detaillo. Briefly, the HPLC system consists of a Waters 600 delivery apparatus (Waters) controlled by a Waters 600 MS system, a microcapillary column Puscia from LCpackings, San Francisco, CA; dimension, 320 pm i.d. x 15 cm; static phase, Vydac C-18), and a Spectroflow 783 UV detector equipped with a very low dead-volume cell. The HPLC pump was operated at flow rates between 300 and 600 pL/min prior to splitting the liquid flow with a ratio of 1:lOO using a commercial splitter (Accurate, LC-packings). The smaller fraction of the split passed through a 0.5 pL injection loop system to the microcapillary column, while the larger fraction went to a waste container. The microcapillary column output was fed through the UV detector into the electrospray ionization mass spectrometer through a 25 pm i.d. fused silica capillary tube. The signal from the UV detector, operated at 214 nm, was sent to the data acquisition computer @EC-station 5000/120) of the mass spectrometer, so that the UV trace and the total ion current obtained at the mass spectrometer detector could be simultaneously monitored.1° The LC effluents were ionized at the ESI source, and the ions were analyzed by a Finnigan TSQ-700 triplequadrupole mass spectrometer (Finnigan MAT, San Jose, CA). For LC/MS experiments, only the first of the three quadrupoles was used and data were recorded in the centroided mode of acquisition. The electrospray needle assembly consisted of three concentric tubes.1° The LC effluents passed through the innermost stainless steel tube, the sheath liquid (2-methoxyethanol at a flow rate of 3 pL/ min) through the second concentric tube, and the sheath gas (nitrogen, 60 mL/min at room temperature) through the outermost tube. Countercurrent dry nitrogen was used as a drying gas (30 mL/min at 270 "C) to assist the evaporation of solvent molecules from the electrosprayed microdroplets, resulting in 392 Analytical Chemistry, Vol. 67, No. 2, January 15, 7995
solvent-free peptide and protein ions for mass spectrometric determination. The flow rate of the LC effluent was 4 pL/min and that of the sheath liquid 3 pL/min, resulting in a total flow rate of 7 pL/min entering the ESI source. The electrospray needle assembly was operated at ground potential and the gold-plated cylindrical electrode at -3.6 to -3.8 kV. The distance between the needle and the transport capillary tube was 2.2 cm.l0 The LC gradient for LC/MS and LC/MS/MS experiments was as follows: time (min) init 5 30 120 150
flow &L/min)
%A
%B
4 4 4 4 4
100 100 55 30 100
0 0 45 70
0
The mobile phases were A = 0.1%aqueous trinuoroacetic acid (TFA) and B = 0.1%TFA in acetonitrile. For LC/MS/MS experiments, the LC and ESI conditions used were the same. The first of the three quadrupoles (Ql) was used to select the parent ion which underwent fragmentation due to collision with a target gas in the radio frequency-only quadrupole (Q2), and the fragment ions were analyzed using the third quadrupole (Q3). Argon was used as the target gas at a pressure of 3 mTorr for the neuromedm N and 2.7 mTorr for TAP. The collision offset voltages used in the Q2 were -15 V for TAP and -20 V for neuromedm N. These were the optimum conditions found for sufficient fragmentation of the two peptides. (ii) LSI-MS at Low Resolution and Peak Matching at High Resolution. LSI-MS experiments for isolated peptides before and after oxidation and digestion at both low and high resolution were performed on a doublefocusing high-resolution mass spectrometer, Kratos Concept lH, equipped with a Cs+ gun. The ion source was held at 8 kV relative to ground, and the gun was operated at 13-15 kV relative to the source potential. All the peptides were run in a matrix comprised of glycerol and thioglycerol in equal proportions containing 5%acetic acid. For simple LSI-MS experiments, the instrument was tuned at a resolution (10% valley definition) of 1000 and for exact mass measurements at 800010 000. The mass spectrometerwas calibrated using CsI, and for peak-matchmg experiments, ions of interest were bracketed using either protonated poly(ethy1ene glycol) ions or CsI cluster ions. Data were acquired and processed using Kratos Mach3 software. RESULTS AND DISCUSSION
LC/ESI-MS and LC/ESI-MS/MS Of Oxidized a d Asp-N Treated SOD. The liquid chromatograms of SOD, oxidized and digested with AspN, are shown in Figure 1. The top panel is the absorption output from the UV detector, and the bottom panel is a reconsbructed ion chromatogram @IC) detected by the mass spectrometer. The peaks in the RIC were detected about 1min after they have been recorded by the UV detector because the UV detector and the ESI source are connected in serieslO. The numbers on various peaks in Figure 1 represent the peptide number expected from the cleavage of oxidized SOD by AspN. The sequence, AspN cleavage sites, and the peptides generated from the oxidized SOD are given in Chart 1: The peptides are assigned as Ax, where x is the peptide number ranging from 1to 15. A1 is first peptide from the N-terminus, and A15 is the last peptide at the C-terminus. To simplify Figure 1,only the number is shown and not all the peptides have been assigned to the peaks,
Table I. Observed nJz Values of Peptide Ions and Observed and Calculated Molecular Weights of Peptides Obtained from the AspN Digestlon of Oxidized SOD (Figure l ) a
molecular weight
peptide fragments A1
A2 A3
Scan II
Figure 1. Liquid chromatograms of performic acid oxidized and Asp-N digested SOD. The top panel is the UV absorption profile, and the bottom panel is the reconstructed ion chromatogram (RIC) detected by the mass spectrometer following electrospray ionization of the LC effluents. The numbers on various peaks represent peptide numbers of SOD shown in Chart 1. Peaks designated * did not correspond to any expected peptide or combinations. The peptide 3' corresponds to the sequence DGPVQGTIHF that results from the cleavage at the N-terminus of glutamic acid of A3 (Chart 1). 2-methoxyethanol was used as a sheath liquid at a flow rate of 3 mUmin and was added to the LC effluents eluting at a flow rate of 4 pUmin, resulting in a total flow rate of 7 pUmin. The mass spectrometer was scanned from mlz 300 to 2000 in 2 s.
Chart 1. Sequence of Bovine SODi3 a A c A T K A V C * V L K G D G P V O G T I H F E A KG D T V V V T G S I T G L T E A1 A2 A3 A4
G
D H G F H V H O F G D N T O G C* T S A C P H F N P L S K K H G G P K A5 A6 A7
W P L G N V T A DKNGVAIV PPLISLSGEYSIIGRTMV A8 A9 A10 All A 12
VHEKP P B L G R G G N E E S T K T G N A G S R L A C * G V I G I A K A13
A14
A 15
Ax (x = 1-1 5 ) are expected peptide fragments from Asp-N digestion of the oxidized protein. C* represents cysteic acid (oxidized cystine). a
as several of them coeluted. However, the measured m/z values corresponding to various peaks obtained from the LC/MS experiment together with their calculated molecular weight are given in Table 1. In general, a good agreement between the observed and the calculated molecular weight is observed in Table 1. Examination of the data presented in Table 1 indicates that with the exception of the peptide fragment A12 (sequence; DPLISLSGEYSIIGKMWHEKP) all other peptides have been identified. However, the measured molecular weight of the peptides corresponding to the peaks designated * in Figure 1do not correspond to the peptide A12 or any other combinations of amino acid sequences of SOD. The mass spectrum derived from the peaks * in Figure 1is shown in Figure 2. Di- and triprotonated molecules from six peptides, designated al, az, a3, a,bl, and bz, have been identified. The molecular weights of these peptides derived from the observed m / z of the peptide ions are 2608.7, 2641.5, 2722.7, 2757.0, 2253.8, and 2287.9, respectively. The first two values are 66.7 and 99.5 u higher than that calculated ( m / z 2542.0) from the sequence of A12. It should be noted that A13 is a single amino acid peptide, aspartic acid (Chart l), and the failure to cleave by AspN before this amino acid residue will result in a peptide with the sequence D P L I S L S G E Y S I I G I E K P D (containing an additional D), which we designate A12+13. The molecular weight of this peptide is calculated to be 2657.0. The
A4 A5 A6 A7 A8 A9 A10 A1 1 A12 A13+14 A14 A15 (I
m/26
measd
calcdC
531.0 566.8 728.6(2+), 1456.6(1+) 1448.6 590.7 (2+), 1180.4(1+) 534.0 (weak response) 670.8(3+), 1006.3 (2+) 420.9(2+), 841.4(1+) 689.4 408.3(2+), 815.3(1+) 346.0
530 566 1455 1447 1179 533 2010 840 688 814 345
736.1(3+), 1103.0(2+) 697.0(3+), 1045.5(2+) 808.5
2205 2089 808
530 566 1455 1448 1181 533 2011 840 688 814 345 2541 2205 2090 807
The sequence and identification of the peptides are given in Chart
1. bThe m / z values are for sing1 protonated molecules, unless indicated in parenthesis. The c&ulated molecular weights are
monoisotopic values.
measured molecular weight of the peptides a3 and a4 are respectively 66 and 100 higher than that of A12+13. Additional measurements of the peptides isolated from this peak (*) using infusion ESI-MS also confirmed that the measured molecular weight of the peptides are 66 and 100 u higher than those expected for A12 and A12+13. The peptides designated bl and bz have been determined to have molecular weights that are also 66 and 100 u higher than the calculatedvalue for the sequence DPLISLSGEYSIIGRTMVVH (calculated molecular weight 2187.6). This peptide, designated A12', lacks the last three amino acids of A12 and formed due to cleavage at the N-terminus of a glutamic acid residue located near the C-terminus of A12. It is known that AspN cleaves at the N-terminus of glutamic acid residues in addition to aspartic acids, but at a slower rate. It is clear that peptides, A12, A12' and A12+13, have undergone modifications either during oxidation or digestion with AspN resulting in an increase of molecular weight of 66 or 100 u. In order to determine the nature of modification of these peptides, we have performed LC/MS/MS on the ions of A12 from the digest and MS/MS with infusion ESI on the isolated peptides corresponding to the peak * (Figure 1). Sufficient fragmentation of these large peptides was not obtained to positively identify the modified amino acids. We therefore used smaller peptides with some key amino acids that were suspected to undergo modifications. The targeted amino acids were tyrosine, arginine, methionine, histidine, and lysine. Lysine can be eliminated because A12' did not contain any lysine; however, it underwent modifications with the same molecular weight increase as A12 and A12+13. We identified a commercial peptide, TAP (sequence; Ac-MDKVLNRY), that contains four of the five amino acids with the exception of histidine. This peptide was subjected to performic acid oxidation and AspN digestion. The peptide before and after oxidation and digestion was investigated by LC/ESI-MS and LC/ ESI-MS/MS. LC/EsI MS and LC/ESI-MS/MS of TAP. The chromatograms of TAP prior to oxidation or digestion is shown in Figure Analytical Chemistry, Vol. 67, No. 2, January 15, 1995
393
Data:
+/989>1044
-
/306>430
I
I a,
= A12 t 68
a2 = A12
"1
t
a3 =A (12t13) t 68 a 4 = A(12t13) t 100
100
E+ 03
5.01
r4
II
40
20
800
900
1100
1000
1300
1200
1400
1' 10
mlz Figure 2. LC/ESI mass spectrum of the peak clusters designated * in Figure 1. 2+ and 3+ ions from six peptides, al, a2, a3, a,bl, and bz, were observed. Identification of these peptides is given in the text. 9 99:
iii
1
loo
I
I
Scan#-
500
1000
1500
2000
2500
Figure 3. Liquid chromatograms of an unoxidized peptide, TAP. Top panel is the absorption output from an UV detector, and the bottom panel is the RIC from the detector of the mass spectrometer. The inset in the bottom panel corresponds to the mass spectrum of the major peak shown in the figure. Experimental conditions are the same in Figure 1.
3. Both the W absorption profile and the RIC show an intense peak and some minor impurity peaks. The ESI mass spectrum corresponding to the intense peak is shown in the inset of the bottom panel of Figure 3. The molecular weight obtained (1080.0) from the mono and diprotonated molecules agrees closely with that calculated (1080.2) for TAP. The LC/ESI chromatograms of the oxidized and AspN digested TAP is shown in Figure 4. Three intense (designated, a, c, and d) and three weak (b, e, f) peaks were detected. The peptide in peak a corresponds to the starting material with a measured molecular weight of 1080. The mass spectra corresponding to the peaks c and d are shown in Figure 5. The molecular weight of the peptides determined from the measured m / z values in peaks c and d are respectively 1146.0 and 1180.3. These values are again 66 and 100 u higher than the calculated molecular weight of TAP. The measurements indicate that probably the amino acids that underwent modifications in TAP are the same as those that were modified in A12', A12, and A12+13 of SOD discussed above. We subjected the diprotonated molecules of the peaks (a, c, d) to MS/MS with on-line LC/ESI. LC/ESI-MS/MS spectra of the two peptides corresponding to m / z 394 Analytical Chemistry, Vol. 67, No. 2, January 15, 1995
-
i . . . , . . , ., . , , , . . . , , , . . , , , LEO0
Scan #
2000
2200
. . , . . . . , , . , , . . , .1 2400
Flgun, 4. Liquid chromatograms of the performic acid oxidized TAP. Top and bottom panels are respectively UV-absorption profile and RIC. Six peaks, designated a-f were observed. Experimental conditions are the same as in Figure 1.
574.2 (c) and 591.0 (d) are shown in Figure 6. The minor peaks, b, e, and f have not been investigated any further although their mass spectra were recorded. The majority of the fragment ions in Figures 6 and those obtained from the fragmentation of m/z 541 (data not shown) have been identifled, and their assignments are summarized in Chart 2. The m / z of the fragment ion containing the N-terminal methionine is 174 for TAP while those from moditied peptides, peaks c and d are both 206, Le., an increase of 32. The data demonstrate that the mass of methionine in peptides from peaks c and d has increased by 32 u compared to the unoxidized peptide. This increase corresponds to the oxidation of methionine to methionine sulfone, which is in agreement with previous rep o r t ~ . ~Examination ~~. of other fragment ions in Chart 2 indicates that all the fragment ions that contain the C-terminal arginine and tyrosine have their masses increased by 34 u for the peptide from peak c and 68 u for that from peak d. The differences in m / z values of the fragment ions demonstrate that the residues D, K, V, L, and N have not been modified in peptides from both c and d. For example, the d ~ e r e n c ebetween the C-terminal fragment (14) (a) Roepstorff, P.;Fohlman, J. Biomed. Mass Spectrom. 1984, 11, 601. (b)
Biemann, K Biomed. Enuiron. Muss Spectrom. 1988, 16,99-111.
Dda: t1201>246
-
Peak c 2i"'
100 7
'"1
598
I-!?"'698.9
b1(206.1)
Data: r t 2 B 0 > 3 l i - i99,:r5 1
1.03
(M+2H)2+
'i.
I
(M'+H)+
'r7.24
il36rl70
i
Peak d
1.1
*
MW = 1180 = tap + 100
(WG2H)
.os
7 1e
.34
lo BOo ]
Y
L 732.9
( MQH)
2oL
+
1181.9
20
I
llaC7
1131.3
I
800
1000
1200
m/z
Figure 5. Mass spectra corresponding to peaks c and d in Figure 4. Two ions, mono- and diprotonated, for each of the two peptides were obtained. The molecular weight of the peptide from peak c is 1146 and that from peak d is 1180, which are respectively 66 and 100 u higher than that calculated for the unmodified TAP.
ions (peak d) m / z 975.7 and 860.0 is 115.7. This value corresponds to an unmodiiied D; for those between m / z 860 and 732.9 the difference is 127.1, indicating no modification of K; and so forth. Together with the increase of 32 u for methionine oxidation and 34 and 68 u for the modification of R and/or Y, total increases of 66 and 100 u in peaks c and d respectively, can be accounted for. Thus, residues R and/or Y at the C-terminus have been transformed with an increase of mass of 34 u for the peptide from peak c and 68 u for that from peak d. The fragmentation information, however, was not sufficient to pinpoint which of the two amino acids was modified because ions due to fragmentation between R and Y were not observed. We, therefore, selected another peptide, neuromedin N (sequence; KIPYIL), which does not have a methionine or arginine but contains a tyrosine. LC/ESI-MS and LC/ESI-MS/MS of Neuromedin N. The UV absorption profile and LC/ESI chromatogram @IC) of neuromedin N prior to oxidation or digestion is shown in Figure 7. Only one peak is seen in both the UV trace and the RIC. The mass spectrum corresponding to the peak is shown in the inset of the bottom panel. The measured molecular weight of 745, deduced from the m / z of 746 for (M H)+ ion, agrees with that calculated (m/z 745.4) for neuromedin N. The UV absorption profile and LC/ESI mass chromatogram of the oxidized and digested neuromedin N are shown in Figure 8. The single peak observed in Figure 7 has changed to three peaks designated a, b, and c after oxidation and digestion. The mass spectra corresponding to the three peaks are shown in Figure 9. The measured mass of the peptide in peak a, 745.8 u, is in good
Figure 6. LC/ESI-MS/MS spectra of the diprotonated molecules, m/z 574 and 591 of peaks c (top panel) and d (bottom panel), respectively, in Figure 5. The progeny ions were selected with the first quadrupole and the fragment ions were analyzed by the third quadrupole.The third quadrupole was scanned from d z 5 0 to 1200. The collision energy was 20 eV, and collision gas pressure was 2.7 mTorr. The assignment of the fragment ions from the both spectra are given in Chart 2. Identification of fragment ions, yn and bn, corresponds to the published assignment pr0~edure.l~
Chart 2
tap
+ 100
M l D / K I V l L I N / R
Y
(m/z 591 j Peak d
+
agreement with that calculated for the unmodified neuromedin N, while those of the peaks b (779.3 u) and c (813.2 u) are respectively 34 u and 68 u higher (Figure 9). Note that both the modified peptides of TAP (Chart 2) also showed a mass increase of 34 and 68 u after the subtraction of 32 u for methionine oxidation. This observation indicates that the tyrosine and not the arginine underwent modifications in performic acid oxidation Analytical Chemistry, Vol. 67, No. 2, January 15, 1995
395
Data:
(.l,O.
. ,,,2 ,,,, 746.8
1
loo
(M+H)+
I
Peak a
r
so.
400
Scan #
500
600
700
BOO 1400
----C
Figure 7. Liquid chromatograms of an unoxidized peptide, neuromedin N. Top panel is the absorption output from an UV detector, and the bottom panel is the RIC from the detector of the mass spectrometer. The inset in the bottom panel corresponds to the mass spectrum of the peak shown in the figure. Experimental conditions are the same as in Figure 1.
1600
m/z Data:
+i6I)rSlS
~
/511r554
780.3
I
Peak b
(M'+H)+
M + 34
.1Et00 777
1561.6
I .
1cno
I200
I
1'61
L* 160
m/z Scan #
-
Figure 8. Liquid chromatograms of the performic acid oxidized neuromedin N. Top and bottom panels are respectively UV absorption profile and RIC. The single peak observed in Figure 7 is transformed into three peaks, a-c, following oxidation and digestion. Experimental conditions are the same as in Figure 1.
+
because neuromedin N does not contain an arginine. To confirm the above observation, the monoprotonated molecules, (M I-€)+, (M 34 €I)+, and (M 68 H)+(m/z 746.8,780.3, and 812.2, respectively), where M is neuromedin N, of peaks a, b, and c (Figure 9) were subjected to on-line LC/MS/ MS. The fragmentation spectra corresponding to the three peaks are shown in Figure 10. In addition to the ions from a single cleavage in the amide backbone, a number of ions from the internal cleavages (two cleavages) were also identified. The m/z values of the Gterminal fragments at proline containingfour amino acids (PYIL) are 505.1 (peak a), 539.1 (peak b), and 573.2 (peak c), demonstrating that the modificationsare in the four C-terminal residues (PYIL,) , while those for the C-terminal fragments containing only IL are the same in all three peptides ( m / z 245.2 beak a), 245.2 (peak b), and 244.7 (peak c)). Therefore, the modification was not present in the last two residues (IL) and must be between P and Y. Examination of the ions from the internal cleavages clearly establishes that it is tyrosine whose mass was increased by 34 u in peak b and 68 u in peak c. For example, the m/z of PY in (a) is 260.9, that in (b) is 295 (an increase of 34), and that in (c) is 328.9 (an increase of 64 u) . Similarly, the internal cleavage ions containing YI also shows the same increases. In both fragment series (py and YI) only Y is common. Redundant information was also obtained from the m/z value of the internal cleavage peak containing IP in (b), m/z 210.6, that eliminates P
+
+ +
s -
'O
26
+ +
396 Analytical Chemistry, Vol. 67, No. 2,January 15, 1995
m/z Flguro 9. Mass spectra corresponding to the three peaks a-c in Figure 8. Only singly protonated molecules of each were obtained. The molecular weight of the peptide from peak b is 779 and that from c is 813, which are respectively 34 and 68 u higher than that calculated for the unmodified neuromedin N (745).
as a possible candidate for modification. It is thus confirmed that tyrosine is the modified amino acid in peaks b and c. It should be noted that subsequent investigations of neuromedin N upon performic acid oxidation before and after AspN digestion indicated that the modifications of tyrosine take place during oxidation and no further change is observed as a result of digestion with AspN (data not shown). Although we have confirmed that methionine undergoes modification to methionine sulfone, the precise nature of tyrosine modification is not established from the molecular weight increase alone. One can assume from the data presented above that modification of tyrosine may be taking place in two steps and in each step a mass increase of 34 u takes place. To determine the structures of the modified tyrosines, we have isolated the modified
mu:
Peak a
p a6o.s
i
py
(YI-43) 245.2
2113
PYI 37,:l
505.1
Y2 5::
I
I
m/z LUMS/MS mlz 814 (M+68)
py' 110 9
I
y2
I
/
m/z Figure I O . LC/ESI-MS/MSspectra of monoprotonated molecules, m/z 747 (top panel), 781 (middle panel), and 814 (bottom panel), arising from peaks a-c, respectively, in Figure 8. The collision energy was 15 eV, and collision gas pressure was 3.0 mTorr.
peaks seen in the liquid chromatograms of oxidized neuromedin N (Figure 8) and determined their accurate molecular weight by a peak-matching technique using LSI-MS. The data are presented below. LSI-MS of Isolated, Oxidized Neuromedin N by a PeakMatching Technique. Neuromedin N was oxidized in large scale so that sufficient quantities of the oxidized products could be isolated by a preparative HPLC technique (see Experimental Section) for LSI-MS investigation. Three major peaks were observed upon HPLC analysis of the oxidized neuromedin N (data not shown). The isolated material from the three peaks (Figure 8) was studied by LSI-MS at both low and high resolution. Isotopic pattern of the peptide ions containing modzed tyrosine indicated the presence of one and two chlorine atoms in the two peptides of mass 779 (M 34) and 813 (M + 68), where M is
+
neuromedin N. Accurate mlz of the monoprotonated molecules of the two peptides were determined to be 780.442 94 (calculated m/z 780.442 67 for C38HaN708Cl) and 814.403 91 (calculated m/z 814.403 69 for C ~ ~ H ~ Z N ~by O peak-matching ~C~Z) technique at a resolution of -10,000. The measured mlz values agree with those calculated to 0.35 and 0.27 ppm, respectively. The excellent agreement observed between the measured and calculated accurate molecular weights confirms that the tyrosine underwent mono- and dichloro substitution at the aromatic ring during performic acid oxidation (Scheme 1). It should be noted that LCl ESI-MS data also indicated an isotope pattern corresponding to 1 and 2 chlorine atoms in the ion peaks containing modified tyrosines. Although the structure of the transformed tyrosine in performic acid oxidation has not been reported previously, halogenation of tyrosines during acid hydrolysis has been detected.15J6 Sanger and Thompson15observed the formation of k h l o r e , 3,5dichloro-, 3-bromo-, and 3,5dibromotyrosine when tyrosine was incubated for 16 h with 30% HzOz and residues of redistilled 6 N HCl at room temperature. The formation of bromotyrosines was attributed to the presence of trace amounts of brominated impurities in HCl.l5 In the present experiment, no bromotyrosines were observed. Origin of mosine Halogenation in Peptides and Proteins. In order to gain an understanding of the effect of the type and the amounts of different salts present in the oxidizing solution on the extent of tyrosine halogenation, experiments were carried out without any salt and with a number of different salts. The salts were added to the diluting buffer at different concentrations, and the oxidation products were analyzed by a reversed-phase HPLC method. The amount of each reaction product was determined from the corresponding area of the W absorption peak. The results obtained from 11 different experiments (data not shown) demonstrate that (i) trace levels of chloride in the water or the phosphate salts are sufficient to cause monochlorination of tyrosine up to 15%of the unreacted Neuromedin, (ii) when a chloride salt is added as a component of the buffer at a concentration of 72 mM no Neuromedin was left unreacted and the chromatograms were dominated by Neuromedin with monoand dichlorotyrosines, (iii) the increase of the concentration of chloride salts results in the increase of the amount of dichloro species relative to that of monochloro, and (iv) substitution of sodium chloride with potassium chloride results in the same overall level of monochlorotyrosine but with a slightly lower proportion of bischloro produd. When a chloride salt was present, the presence of a phosphate salt did not iduence the degree of chlorination. In a separate experiment (data not shown), the replacement of the chloride salt with a bromide salt resulted in the formation of mono- and dibromo tyrosines. (15) Sanger, F.;Thompson, E. 0. P. Biochim. Biophys. Acta 1963,71, 468471. (16) Munier, R]. Chromutog. 1958,I , 524.
Analytical Chemistry, Vol. 67, No. 2, January 15, 1995
397
The data presented above demonstrate that the reaction of performic acid with NaCl, used in the diluting buffer (see Experimental Section) or with the chloride impurities in the reagents used, have produced electrophiles that caused electrophilic aromatic ~ubstitutionl~ forming 3- and 3,5dichlorotyrosines.15 Because of the strong electron-donating nature (resonance effect) of the OH group located in position 4 of tyrosine, electrophilic aromatic substitutions are expected in positions 3 and 5.17 We therefore, conclude that the tyrosine transformations (17) Streitwieser,A, Jr.; Heathcock, C. H. Introduction to Organic Chemistry, Macmillan Publishing Co.: New York, 1981.
398 Analytical Chemistty, Vol. 67,No. 2,January 15, 1995
in performic acid oxidation resulted from the mono and dichloro substitution on the aromatic ring.15J7 ACKNOWLEWMENT We thank Dr. R Weinkam for his encouragement and support. Received for review September 14, 1994. November 1, 1994.@
Accepted
AC94Q92QY @
Abstract published in Advance ACS Abstracts, December 1,1994.
