Peptide Rearrangement during Quadrupole Ion Trap Fragmentation

Feb 15, 2003 - Mining a Tandem Mass Spectrometry Database To Determine the Trends and Global Factors Influencing Peptide Fragmentation. Eugene A. Kapp...
0 downloads 6 Views 323KB Size
Download PDF
Anal. Chem. 2003, 75, 1524-1535

Peptide Rearrangement during Quadrupole Ion Trap Fragmentation: Added Complexity to MS/MS Spectra Jesu´s Yagu 1 e, Alberto Paradela, Manuel Ramos, Samuel Ogueta, Anabel Marina, Fernando Barahona, Jose´ A. Lo´pez de Castro, and Jesu´s Va´zquez*

Centro de Biologı´a Molecular “Severo Ochoa”, Consejo Superior de Investigaciones Cientı´ficas, Facultad de Ciencias, Universidad Auto´ noma de Madrid, Cantoblanco, 28049, Madrid, Spain

The emergence of proteomics has placed great interest in the understanding of the mechanisms of MS/MS fragmentation of peptides under low-energy collisioninduced dissociation. In this work, we describe the presence of anomalous fragments, which correspond to neutral loss elimination of internal amino acids from ions of the b series in quadrupole ion trap MS/MS spectra from naturally occurring peptides. Internal amino acid elimination occurred preferentially with aliphatic amino acids. The phenomenon was more apparent when doubly charged precursors were fragmented and was inhibited when peptides were N-acetylated at the N-terminus. Fragmentation of isomeric peptides where some internal amino acids were relocated in N-terminal position produced MSn spectra indistinguishable from those of the original peptides, indicating that some b ions underwent a structural rearrangement process. Formation of anomalous fragments required a minimum activation time. Our data are consistent with a nucleophile attack of the N-terminal nitrogen over the electrophilic carbonyl carbon at one peptide bond, forming a cyclic b ion intermediate that, by reopening at preferential sites, exposes internal amino acids to the C-terminal side. Understanding of the mechanisms of MS/MS fragmentation of peptides under low-energy collision-induced dissociation has gained great interest in the last years. The emergence of proteomics has placed a high demand on the rapid identification of peptides by tandem mass spectrometry followed by database searching algorithms. Several programs, such as Sequest and Mascot, currently allow automated peptide identification by comparing their fragment spectra with “predicted” spectra generated from the sequences present in databases. However, a recent estimate suggests that a considerable proportion of tryptic peptides do not yield full sequence information through this approach,1,2 and it is well known that conventional dissociation pathways are insufficient to fully explain the patterns of peptide fragmentation. * Corresponding author. E-mail: [email protected]. (1) Simpson, R. J.; Connolly, L. M.; Eddes, J. S.; Pereira J. J.; Moritz, R. L.; Reid, G. E. Electrophoresis 2000, 21, 1707-1732. (2) Wee, S.; O’Hair, R. A. J.; McFadyen, W. D. Proceedings of the 50th ASMS Conference on Mass Spectrometry and Allied Topics; Orlando, FL, 2002; in press.

1524 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

In addition, the development of novel automated “de novo” sequencing algorithms capable of extracting partial or total sequence information from uninterpreted tandem mass spectra3 (Maroto et al., 2002) opens the possibility of applying these techniques to the analysis of proteins from species poorly represented in databases, a task currently done by manual interpretation of MS/MS spectra.4-6 These programs, which could also be very useful for the automated identification of posttranslational modifications, would greatly benefit from a more accurate prediction of peptide fragmentation. Peptide fragmentation is usually conceived in terms of unimolecular interactions between nucleophilic and electrophilic centers located in the backbone or in functional groups of side chains,7 which are particularly favorable when they involve the formation of five-membered-ring intermediates,8 although other cyclic intermediates are also possible.9 Peptide dissociation may be understood in the framework of the “mobile proton” model, which assumes that cleavage is “charge-directed” by protonation at the cleavage site; protonation at backbone sites produces the characteristic b- or y-type sequence ions, or both.10 Several fragmentation pathways are known that produce ions other than those explained by the formation of the conventional b and y series. They are usually observed when dissociations are induced by high-energy collision conditions. Other reactions that are well understood and are produced under low-energy collision conditions are those leading to the so-called internal fragments, which contain neither the N-terminus nor the C-terminus of the original peptide. They are typically produced by multiple collisions in rf-only quadrupole cells. Other reactions have been described that arise from the formation of cyclic intermediates with the (3) Maroto, F.; Scigelova, M.; Dufresne, C.; Va´zquez, J. Proceedings of the 50th ASMS Conference on Mass Spectrometry and Allied Topics; Orlando, FL, 2002; in press. (4) Pin ˜eiro, C.; Va´zquez, J.; Marina, A. I.; Barros-Vela´zquez, J.; Gallardo, J. M. Electrophoresis 2001, 22, 1545-1552. (5) Lo´pez, J. L.; Mosquera, E.; Fuentes, J.; Marina, A.; Va´zquez, J.; Alvarez, G. Mar. Ecol. Prog. Ser. 2001, 224, 149-156. (6) Lo´pez, J. L.; Marina, A.; Alvarez, G.; Va´zquez, J. Proteomics 2002, 2, 16581665. (7) O’Hair R. A. J. J. Mass Spectrom. 2000, 35, 1377-1381. (8) Schlosser, A.; Lehmann, W. D. J. Mass Spectrom. 2000, 35, 1382-1390. (9) Polce, M. J.; Ren, D.; Wesdemiotis, C. J. Mass Spectrom. 2000, 35, 13911398. (10) Wysocki, V. H.; Tsaprailis, G.; Smith L. L.; Breci, L. J. Mass Spectrom. 2000, 35, 1399-1406. 10.1021/ac026280d CCC: $25.00

© 2003 American Chemical Society Published on Web 02/15/2003

primary amine group on lysyl and ornithyl side chains, followed by ring opening and subsequent loss of internal amino acid residues in a triple quadrupole.11,12 Fragmentation of peptides inside an ion trap is particularly reproducible since it takes place under low-energy collision conditions, and the frequency of the excitation waveform is adjusted to produce resonance excitation on the precursor ion only. As a consequence, the fragments are not excited by the waveform used to excite the parent ion and, therefore, rarely fragment again. Thus, ion trap fragmentation needs little optimization of dissociation conditions. However, the longer experimental time frames available in the quadrupole ion trap enable reactions not seen in other instruments to be observed, such as intramolecular rearrangements of four-residue immonium ions.13 In this work, we analyze the generation of fragments that are not explained by the conventional sequence series, under lowenergy dissociation conditions in an ion trap. These anomalous fragments correspond to neutral losses of amino acids located in internal positions and were directly observed in the MS/MS spectra. Our results suggest that these fragments arise from a head-to-tail cyclization of doubly charged b ions followed by ring opening at different positions. These rearrangements contribute to increase the complexity of the spectra. Our results also suggest that lowering of the activation time tends to reduce the formation of anomalous fragments. EXPERIMENTAL SECTION Preparation of Peptides. All the peptides analyzed in this work are natural ligands eluted from immunoaffinity-purified MHC class I molecules obtained from HMy2-C1R cells transfected with B*2705 (GRIDKPILK, SRVKLILEY, QRKKAYADF),14 B*3909 (SRDKTIIMW),15 B*2703 (KRNGVIIAGY), or B*3905 (all the peptides with His in second position).16 A detailed protocol is described elsewhere17. Peptides showing anomalous fragment behavior were also synthesized by standard Fmoc chemistry in our Protein Chemistry facility. Mass Spectrometry. Peptides were analyzed by nanosprayquadrupole ion trap mass spectrometry using the conditions described previously,18,19 using either a LCQ Classic or a LCQ Deca XP machine (Thermo Finnigan, San Jose´, CA). For nanospray analysis, peptides were prepared from MHC I-bound peptide pools as described.17 Unless stated otherwise, the following conditions were used for peptide fragmentation: activation q parameter, 0.25; normalized collision energy (%), 35; activation (11) Tang, X.-J.; Thibault, P.; Boyd, R. K. Anal. Chem. 1993, 65, 2824-2834. (12) Tang, X.-J.; Boyd, R. K. Rapid Commun. Mass Spectrom. 1994, 8, 678686. (13) Vachet, R. W.; Bishop, B. M.; Erickson, B. W.; Glis, G. L. J. Am. Chem. Soc. 1997, 119, 5481-5488. (14) Ramos, M.; Paradela, A.; Va´zquez, M.; Marina, A.; Va´zquez, J.; Lo´pez de Castro, J. A. J. Biol. Chem. 2002, 277, 28749-28756. (15) Yagu ¨ e, J.; Ramos, M.; Va´zquez, J.; Marina, A.; Albar, J. P.; Lo´pez de Castro, J. A. Tissue Antigens 1999, 53, 227-236. (16) Yagu ¨ e, J.; Ramos, M.; Ogueta, S.; Va´zquez, J.; Lo´pez de Castro, J. A. Tissue Antigens 2000, 56, 385-391. (17) Paradela, A.; Garcı´a-Peydro, M.; Va´zquez, J.; Rognan, D.; Lopez de Castro, J. A. J. Immunol. 1998, 161, 5481-5490. (18) Marina, A.; Garcı´a, M. A.; Albar, J. P.; Yagu ¨ e, J.; Lo´pez de Castro, J. A.; Va´zquez, J. J. Mass Spectrom. 1999, 34, 17-27. (19) Ogueta, S.; Rogado, R.; Moreno, F.; Redondo, J. M.; Va´zquez, J. J. Mass Spectrom. 2000, 35, 556-565.

time, 30 ms; isolation width (m/z), 3.0. Helium pressure was not changed from that set during installation; typical vacuum inside the ion trap was ∼1 × 10-5 Torr. RESULTS In the course of studies performed in our laboratory, where naturally occurring peptides presented by different MHC class I molecules were analyzed by ion trap MS/MS,16,20-22 we observed that some of these peptides displayed a fragmentation behavior that could not be justified on the basis of the generation of the common N- or C-terminal series. Some examples are presented in Figures 1-3. All these MS/MS spectra were produced from doubly charged precursor species and yielded unidentified fragments with relatively high intensities (indicated by thick arrows in Figures 1, 2A, and 3A). We found that all the unexplained fragments could be interpreted by assuming the neutral loss of one or more internal amino acids from ions of the b series. For instance, in Figure 1A, which shows the MS/MS spectrum of peptide GRIDKPILK, the signal at m/z 683.3 differs from the b8+ ion by 210 Da, a difference that matches the neutral loss of amino acids P and I from b8+; lacking an adequate nomenclature, this peak was named [b8 - PI]+, and other fragments presumably produced by a similar mechanism were named accordingly. The peak at m/z 780.3, which also matched the b7+ fragment, corresponded to [b8 - I]+; according to the evidence that will be presented in this paper, we think that this peak contains contributions from both b7+ and [b8 - I]+. In the MS/MS spectrum from peptide KRNGVIIAGY (Figure 1B), the signals at m/z 739.3, 370.2, and 626.3 could be interpreted as [b8 - I]+, [b8 - I]2+, and [b8 II]+, respectively. Other peaks of lesser intensity, such as those at m/z 796.3, 725.3, 612.3, and 527.3, could also be assigned to [b9 - I]+, [b9 - IA]+, [b9 - IIA]+, and [b8 - VII]+. In Figure 2A, we found two unassigned peaks at m/z 680.1 and 662.1 in the MS/MS spectrum from peptide EHAGVISVL, the latter being one of the most intense peaks from this spectrum; these peaks matched the hypothetical fragments [b8 - I]+ and the neutral loss of water from this ion (which was denominated [b08 - I]+). A lower intensity peak at m/z 563.1 could also be assigned to [b08 - VI]+. Finally, fragments that correspond to internal elimination of G, D, A, and V from the ion b8 of peptide SHIGDAVVI, are shown in Figure 3A. We analyzed whether these anomalous fragments could be produced from the fragmentation of other precursors. When MS/ MS spectra of the corresponding singly charged [M + H]+ precursor ions of the same peptides were analyzed, the same unassigned peaks were found, but with a relatively lower intensity (data not shown). In clear contrast, these peaks were observed with much higher intensities in the MS3 spectra from the doubly charged b ions from which they appeared to derive by neutral loss of internal amino acids. Thus, fragmentation of ion b82+ from peptide EHAGVISVL (m/z 397.1) produced a spectrum containing abundant fragments deriving from neutral loss of internal amino acids (Figure 2B). Fragmentation of ion b82+ from peptide (20) Alvarez, I.; Martı´, M.; Va´zquez, J.; Camafeita, E.; Ogueta, S.; Lo´pez de Castro, J. A. J. Biol. Chem. 2001, 276, 48740-48747. (21) Yagu ¨e, J.; Marina, A.; Va´zquez, J.; Lo´pez de Castro, J. A. J. Biol. Chem. 2001, 276, 43699-43707. (22) Sesma, S. L.; Montserrat, V.; Lamas, J. R.; Marina, A.; Va´zquez, J.; Lo´pez de Castro, J. A. J. Biol. Chem. 2002, 277, 16744-16749.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1525

Figure 1. MS/MS spectrum of peptides GRIDKPILK (A) and KRNGVIIAGY (B), showing anomalous fragmentation patterns that correspond to neutral loss of internal amino acids from ions of the b series. In this and the following figures, the nomenclature of Roepstorff and Fohlman25 was used for the main and secondary series of fragment ions and that of Biemann26 for internal sequence ions; “/” and “0” superscripts denote neutral loss of one ammonia molecule and one water molecule, respectively. Anomalous fragments are labeled as explained in the text.

SHIGDAVVI (m/z 390.1) produced a spectrum whose most intense peak was [b8 - AV]+ (Figure 3B). Similar results were obtained from the fragmentation of b82+ from GRIDKPILK and KRNGVIIAGY (data not shown). In clear contrast, these unassigned signals were not observed when singly charged bions were used as precursors (not shown). These findings support the idea that internal amino acid elimination is produced from a doubly charged b-ion intermediate. To further explore this phenomenon, some of the anomalous ions were subjected to fragmentation. As shown in Figure 2C, almost all the fragments produced from fragmentation of [b08 I]+ corresponded to neutral loss of internal amino acids from b ions, suggesting that, after losing an internal Ile, additional internal 1526 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

adjacent amino acids are then sequentially lost. Similar behavior was observed when ion [b8 - AV]+ at m/z 609.1 from peptide SHIGDAVVI was fragmented (Figure 3C); as shown, besides AV, sequential elimination of D, G, and I, and also V and S, was observed. Interestingly, some fragments, such as [b8 - AVV S]+ in Figure 3C, seemed to originate from the neutral loss of nonconsecutive internal amino acids. Besides the four peptides described, we detected the presence of significant peaks that matched neutral loss elimination of internal amino acids in the MS/MS spectra from peptides THQDVHLGETL, QRKKAYADF, SRDKTIIMW, and SRVKLILEY (not shown). In total, 8 peptides displaying anomalous fragmentation behavior were detected among a population of 204 peptides

Figure 2. (A) MS/MS spectrum of peptide EHAGVISVL. (B, C) MS3 spectra of ions b82+ and [b08 - I]+ from the same peptide.

containing 8-13 residues, which correspond to a frequency of 4%; that is, 1 every 25 peptides had anomalous fragmentation. The majority of these peptides had a basic residue (Arg or His) and some others had Pro in second position. From this collection of cases, however, we were unable to deduce a common sequence pattern or motif, which could serve to predict the occurrence of this phenomenon. We could only conclude that internal elimination

was more frequent with amino acids containing aliphatic side chains, such as Ile, Leu, Val, and Ala. When the N-R-acetylated forms of peptides EHAGVISVL and SHIGDAVVI were subjected to fragmentation, no traces of neutral losses of internal amino acids from either the doubly protonated peptides or from the b82+ ions, nor the presence of fragments other than those expected by common fragmentations mechanisms Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1527

Figure 3. (A) MS/MS spectrum of peptide SHIGDAVVI. (B, C) MS3 spectra of ions b82+ and [b8 - AV]+ from the same peptide.

could be detected (not shown). Hence, the anomalous fragmentation required a free N-terminus. 1528 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

In another set of experiments, an Ala to Gly substitution in position 6 was performed in peptide SHIGDAVVI, and the

Figure 4. (A) MS/MS spectrum of peptide SHIGDGVVI. (B, C) MS3 spectra of ions b82+ and [b8 - GV]+ from the same peptide.

fragmentation of this new peptide was investigated. The fragmentation pattern of the doubly charged peptide SHIGDGVVI, except for a decrease in b6+ signal (probably due to a reduced cleavage yield at the Gly6-Val7 bond), was well conserved relative to the original peptide (compare Figure 3A with Figure 4A), although

the anomalous fragment at m/z 609, which in this case corresponded to [b8 - GV]+, was observed with lesser intensity. Fragmentation of the b82+ ion from the SHIGDGVVI peptide (m/z 383.2) produced a spectrum with a fragmentation pattern similar to that of the original peptide, predominating the internal losses Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1529

Figure 5. MS3 spectra of ion b82+ from peptides SVEHAGVIL (A) and VSHIGDAVI. (B).

of sequential amino acids (Figure 4B). In addition, a further fragmentation of the [b8 - GV]+ ion from the analogous peptide (m/z 609.0) produced a spectrum with essentially the same fragments as that of [b8 - AV]+ ion from SHIGDAVVI (Figure 4C). To explore whether a cyclization process may produce the observed phenomenon, we tried to generate fragments identical to those produced by neutral loss of internal amino acids, through conventional fragmentation pathways of analogous peptides where one or two internal amino acids are relocated in N-terminal position, just as would happen in a hypothetical cyclization phenomenon. We designed a synthetic peptide with the sequence SVEHAGVIL and studied the MS3 spectrum from its b82+ fragment. As shown in Figure 5A, this spectrum was essentially identical to that produced from the original peptide EHAGVISVL (Figure 2B). Similarly, fragmentation of b82+ ion from the rearranged peptide VSHIGDAVI (Figure 5B) produced a spectrum essentially indistinguishable from that derived from the original peptide SHIGDAVVI (Figure 3B). 1530 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

Consistently, the fragment spectra produced by the anomalous ion at m/z 662.1, corresponding to neutral loss of water and Ile from fragment b8+ from the original peptide EHAGVISVL (Figure 2C), was identical to that produced by fragment b07+ from the synthetic rearranged peptide (Figure 6A), and the spectra produced by the anomalous ion at m/z 609.1, corresponding to the neutral loss of AV from the original peptide SHIGDAVVI (Figure 3C), was identical to that produced by fragment b6+ from the rearranged peptide (Figure 6B). These results indicate that the doubly charged b8 fragments have the same structure in both the original and rearranged peptides and that the anomalous fragments derived from the original peptides by neutral loss of internal amino acids also have the same structure as those produced by conventional fragmentation of the rearranged peptides. These data strongly suggest the existence of a head-to-tail cyclization phenomenon in the doubly charged b8 precursors. Finally, we investigated whether the formation of anomalous fragments was influenced by the activation time, that is, the duration of the collision-induced fragmentation process inside the

Figure 6. (A) MS3 spectrum of ion b07+ from peptide SVEHAGVIL. (B) MS3 spectrum of ion b6+ from peptide VSHIGDAVI.

ion trap. For this purpose, we analyzed the fragmentation of peptide EHAGVISVL, comparing the intensity of the anomalous fragment [b08 - I]+ (m/z 662.1, Figure 2A) with that of fragments b6+, b7+, and b8+ (m/z 607.1, 694.1 and 793.1, respectively), as a function of the activation time. Using the minimum activation time allowed by our ion trap, 0.03 ms, the intensities of these three last ions were 20, 22, and 55%, respectively, of that of the precursor peak, while the peak corresponding to [b08 - I]+ was hardly detectable, remaining below 4%. Similar results were obtained with other anomalous fragments, indicating that this kind of fragmentation is negligible under these conditions. As shown in Figure 7A, upper panel, the relative intensity of the three ions of the b series was not significantly influenced by increasing the activation time in the range from 0.03 to 1000 ms, as expected for a collisioninduced fragmentation inside an ion trap, a process driven by resonance excitation waveforms that selectively affect the precursor ion and is little influenced in qualitative terms by collision

energy or fragmentation conditions. At low activation times, the relative intensity of [b08 - I]+ remained very low (Figure 7A, lower panel); however, when the activation time exceeded a critical value slightly below 1 ms, a sudden increase was detected. The intensity of [b08 - I]+ surpassed that of the most prominent fragments of the b series and remained essentially constant up to the highest activation time allowed by our ion trap (1000 ms, Figure 7A). Therefore, the formation of anomalous fragments required a minimum activation time. Figure 7B shows the fragment spectrum of the peptide at an activation time of 0.6 ms, around the critical point where rearrangement of b ions begins to take place. As shown, the fragment [b08 - I]+ was detected with a relatively low intensity, and no other anomalous fragment were noticeable. Although in these conditions the fragments were detected at lower intensities, and some peaks, such as that corresponding to b3+, could not be detected, the fragmentation of the peptide backbone was very similar to that Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1531

Figure 7. (A) Effect of the activation time on the generation of anomalous fragments from peptide EHAGVISVL. Each data set represents the abundance of ions in numerator relative to that of ions in denominator. The same collision energy was used in all experiments. (B) MS/MS spectrum from peptide EHAGVISVL using an activation time of 0.6 ms.

observed using higher activation times (compare with Figure 2A). This result suggests that the generation of anomalous fragments, even in the case of peptides prone to undergo this kind of rearrangements, may be maintained at low levels by an appropriate adjustment of the activation time. DISCUSSION Our data demonstrate that a certain proportion of peptides may produce anomalous fragmentation patterns under conventional dissociation conditions inside an ion trap, producing intense peaks that correspond to the neutral loss from b ions of amino acids located internally in the sequence. Our results also indicate that this phenomenon is inhibited by N-acetylation of the N-terminal nitrogen and is preferentially produced from doubly charged species, and the structure of the doubly charged precursors and the anomalous fragments is identical to those produced from isomeric peptides where the internal peptide sequence has been rearranged. Our data may be explained by assuming a head-totail cyclization of doubly charged b ions, followed by ring reopening at different sites and conventional fragmentation of the rearranged ions. A tentative mechanism is shown in Scheme 1. Formation of b ions is thought to take place by a nucleophilic attack of the 1532

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

carbonyl oxygen of the amino acid on the N-terminal side over the carbonyl carbon, which acts as electrophilic center due to protonation of the amide bond, producing a five-membered cyclic structure (protonated oxazolone),23 followed by a C-N bond cleavage without proton transfer (reaction I in Scheme 1). Ions of the b series have also been proposed to have a linear, open structure (acylium ion), which may be in equilibrium with the cyclic one (Scheme 1). This kind of nucleophile-electrophile interaction, although being the most common in low-energy fragmentation of peptides, is but one among a number of possible neighboring group interactions that may contribute to peptide ion fragmentation.7 Within this conceptual framework, it is also conceivable that the N-terminal nitrogen attacks the carbonyl carbon, forming a headto-tail cyclized b ion (reaction II). This cyclic intermediate may then reopen at different peptide bonds, producing linear b ions where internal residues may be relocated at the N- or C-termini. Conventional fragmentation mechanisms can then explain the neutral losses of these residues, or of a set of consecutive ones, that were originally located in internal positions. The nucleophilic (23) Arnott, D.; Kottmeier, D.; Yates, N.; Shabanovitz, J.; Hunt, D. Proceedings of the 42nd ASMS Conference on Mass Spectrometry and Allied Topics; Chicago, IL, 1994; p 470.

Scheme 1. Proposed Mechanism for the Formation of Cyclic b Ions and Amino Acid Rearrangementa

a

For further details see the text.

Scheme 2. Formation of Equivalent Structures from Different Precursors after Cyclization and Rearrangement of Internal Amino Acids

attack of the N-terminal nitrogen could also take place, after the bn+ ion is formed, over the carbon atom of the C-terminal acylium group, producing the isomeric cyclic intermediate (reaction III). According to this mechanism, the neutral loss elimination of Ile and other consecutive internal amino acids from the doubly charged peptide EHAGVISVL takes place through a b82+ intermediate (Scheme 2A). Cyclization of this species followed by ring reopening at the Ile-Ser bond rearranges Ile at the C-terminus, and the [b08 - I]+ ion is then produced after neutral water loss and conventional fragmentation at the last peptide bond. Cycliza-

tion of the doubly charged precursor followed by Leu elimination could also directly produce the cyclic b intermediate (Scheme 2A). Conventional fragmentation of the synthetic rearranged peptide SVEHAGVIL produces a doubly charged b8 intermediate with the same structure as that from the original peptide after cyclization and ring reopening (Scheme 2A). This scheme explains why the b07+ ion from the rearranged peptide is identical to the [b08 - I]+ ion from the original peptide. Similar argumentation explains the neutral loss of internal amino acids from peptide SHIGDAVVI, as well as the correspondence of fragments with those from the Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1533

rearranged peptide (Scheme 2B). This mechanism also explains why N-R-acetylated peptides failed to produce anomalous fragmentation. A similar cyclization mechanism has been proposed for lysyland ornithyl-containing peptides.11,12 In those works, cyclic species were proposed in which the carbonyl group at the C-terminus becomes bonded to the ω-amine group on the Lys and Orn side chains; subsequent ring opening and fragmentation at some preferential sites gave rise to mass-shifted b fragments, which corresponded to the neutral loss of amino acids originally located in internal positions.12 Although this mechanism is very similar to that reported in this work, these anomalous fragments were detected in a triple-quadrupole cell. In addition, we observed the phenomenon in peptides lacking Lys or Orn residues, and in our case, cyclization of doubly charged b ions appears to be produced by the N-terminal amine group and not by the side chains of these amino acids. This conclusion was supported by the lack of anomalous fragments from Nterminally acetylated peptides, as well as by the identity of MS3 spectra obtained from rearranged isomeric precursors. We also observed an unusually high intensity of the doubly charged b ions that presumably act as the cyclic intermediates, suggesting that cyclization confers on these ions an enhanced stability. The neutral loss of internal amino acids was preferentially observed from doubly charged precursors (peptides or b ions), suggesting that doubly charged species are more prone to undergo head-to-tail cyclization. It is also possible that the singly charged species form cyclic structures but do not fragment as readily as their double-charged counterparts. According to the mobile proton model, fragmentation of doubly charged species is more favorable due to the increased availability of mobile protons to transfer intramolecularly along the peptide backbone to enable charge-site-initiated dissociations.24 This would possibly explain why doubly charged cyclic structures are more prone than singly charged species to rearrangement by ring reopening and subsequent neutral loss of C-terminal amino acids. Intramolecular rearrangements of immonium (a-type) ions have also been observed in the product ion spectra of a number of peptides in a quadrupole ion trap.13 These arrangements produced anomalous fragments in the MS/MS spectrum when xenon was added to the helium gas used for fragmentation.13 These rearrangements have been proposed to take place by a proton transfer from the immonium nitrogen to the primary amine on the N-terminus, leading to a loss of NH3 and exposing an internal amino acid on the terminus of the new ion.13 In this case, the driving force for the specificity of the ring opening is most likely the fixed charge located on the immonium nitrogen,13 so that only one amino acid is rearranged to the C-terminal end and then subsequently lost. Our data, however, support the notion that the rearrangement observed during the course of our work occurs through a cyclic b-type ion, and therefore, more than one amino acid may be rearranged to the C-terminal end. In addition, the anomalous fragments described here were directly observed in the MS/MS spectrum under conventional dissociation collisions. (24) Tang, X.-J.; Boyd, R. K. Rapid Commun. Mass Spectrom. 1992, 6, 651657. (25) Roepstorff, P.; Fohlman, J. Biomed. Mass Spectrom. 1984, 11, 601. (26) Biemann, K. Methods Enzymol. 1990, 193, 886-887.

1534

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

Therefore, the phenomenon studied here is probably produced by a different mechanism. The internal energy gained by an ion during collision in the quadrupole ion trap is very low, and therefore, activation of ions to produce fragmentation involves hundreds to thousands of collisions with the gas over the activation time used,13 which in conventional ion traps is ∼30 ms. Our data indicate that for the rearrangement reaction to occur activation times in the millisecond range are needed. This is consistent with our proposition that the formation of an anomalous fragment takes place by a rearrangement reaction involving multiple steps (formation of cyclic structures followed by ring openings and subsequent fragmentation of amino acids relocated at the C-terminal end). Our data also indicate that, upon reaching the activation time needed for this process to occur, no alterations are observed in the fragmentation pattern. Finally, our results indicate that generation of anomalous fragments may be avoided, or at least minimized, by selecting a shorter activation time. Switching between short and long activation times during the same experiment may be used as a criterion to detect the generation of fragments produced inside an ion trap by this and probably other types of internal ion rearrangements. It has been proposed that the neutral fragments cleaved from the N-terminus upon formation of y ions (which would correspond to neutral b fragments) may adopt cyclic structures (reviewed in ref 9); formation of these structures would take place through a mechanism similar to that proposed in reaction II in Scheme 1, that is, interaction of the electrophilic C atom in the protonated peptide bond with a N-terminally located nitrogen.9 Energetic considerations have suggested that the cyclic structures of neutral fragments containing three or more residues may be more stable than the corresponding isomeric oxazolones.9 On the basis of gasphase basicities and proton affinities of small peptides containing glycine and alanine, it has been justified why the fragmentation pathways leading to generation of oxazolones (reaction I in Scheme 1) occur preferentially by the formation of protonated oxazolones (b ions) and neutral loss of shorter peptides, while the pathways leading to formation of cyclic structures (reaction II in Scheme 1), occur preferentially by the formation of neutral cyclic peptides and y ions. Thus, formation of charged cyclic peptides such as those proposed in Scheme 1 is an apparently unfavored process. However, these arguments have been applied to small peptides, and on increasing the size of the N-terminal neutral fragments, larger rings become comparably favorable.9 Therefore, it is conceivable that in peptides containing specific structures the cyclic intermediates may have a greater gas-phase basicity than the C-terminal fragments, taking the proton and giving rise to cyclic b ions. Our data do not presently allow discerning features in the primary sequences or structural determinants that correlate with the formation of cyclic intermediates and the neutral loss elimination of internal amino acids. As stated above, the number of peptides showing anomalous fragmentation patterns is too low to make conclusions on this point, except that aliphatic amino acids such as Ile, Leu, Ala, or Val are more prone to undergo internal elimination. In one case, we have found that a peptide having an Ala to Gly substitution maintained the propensity to form cyclic intermediates. However, these amino acids are very common in

peptides that do not show anomalous fragmentation, and therefore, it is likely that a number of different factors, such as gas-phase basicity, steric requirements associated with the cyclic structures, and competition with other reaction channels12 play a role in the process. CONCLUSIONS Fragmentation of peptides inside an ion trap is particularly reproducible since it takes place under low-energy collision conditions, and resonance excitation selectively affects the precursor ion. This is particularly adequate for fully automated, largescale identification and sequencing of peptides. In some cases, however, the longer experimental time frames available in the quadrupole ion trap enable the generation of fragments that do not correspond to the conventional N- or C-terminal series. In this work, we present evidence that b ions from some peptides may undergo a rearrangement process, which can be explained on the basis of a head-to-tail cyclization produced by the nucleophilic attack of the N-terminal nitrogen over the electrophilic carbonyl carbon at one peptide bond; the cyclic intermediate may then

reopen at different sites, exposing to the C-terminal side amino acids that were originally located in internal positions and that are then eliminated through conventional fragmentation pathways. Although it is not presently possible to deduce what the sequence determinants are that correlate with rearrangement of b ions, this phenomenon is strongly dependent on the time scale of collisional activation and may be minimized by using activation times below 1 ms. ACKNOWLEDGMENT We thank Dr. G. Drabner and Dr. M. -L. Hagmann for their helpful comments and suggestions. This work was supported by Grants SAF2000-0178 and CAM 08.5/0065.1/2001 to J.V. and Grants PM99-0098 and SAF99-005 to J.A.L.C., and by an institutional grant by Fundacio´n Ramo´n Areces to CBMSO.

Received for review November 4, 2002. Accepted January 22, 2003. AC026280D

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1535