Discrimination of Isobaric Leucine and Isoleucine ... - ACS Publications

Jun 4, 2010 - By variation of the buffer gas pressure, ISD and “post-source decay” (PSD) dissociation channels can ..... Additional information as...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 5628–5635

Discrimination of Isobaric Leucine and Isoleucine Residues and Analysis of Post-Translational Modifications in Peptides by MALDI In-Source Decay Mass Spectrometry Combined with Collisional Cooling Jens Soltwisch and Klaus Dreisewerd* Institute of Medical Physics and Biophysics, Westfa¨lische Wilhelms-Universita¨t Mu¨nster, Robert-Koch-Strasse 31, 48149 Mu¨nster, Germany Collisional cooling employed in an orthogonal time-offlight mass spectrometer (o-TOF MS) stabilizes fragment ions that are generated by matrix-assisted laser desorption ionization in-source decay (MALDI ISD). By variation of the buffer gas pressure, ISD and “post-source decay” (PSD) dissociation channels can be switched on and off to some extent. Under high-pressure conditions, ISD type fragments of post-translationally modified (PTM) peptides are generated that contain even labile bound side groups; the examples of a phosphorylated and an O-glycosylated peptide are shown. At elevated laser fluences, d- and w-type fragment ions of peptides are detected as a result of high-energy side chain cleavage. This allows for differentiation of the isobaric amino acid residues leucine and isoleucine. Reduction of the cooling efficiency by lowering the buffer gas pressure results in the loss of the d-/w-type species, presumably in secondary metastable dissociation processes. This also enhances cleavage of the side groups from the PTM peptides and can be used to corroborate identification of the modification site. Furthermore, these measuring conditions generate small amounts of fragments containing sequence information about the cyclic part of a disulfide peptide by inducing symmetric and asymmetric cleavage of the intramolecular S-S bond. Ultraviolet matrix-assisted laser desorption ionization (UVMALDI1) is widely regarded as a “soft desorption” technique. Nevertheless, a fraction of the analyte ions generated by MALDI undergoes subsequent dissociation reactions.2 This generally produces two distinct types of fragment species. The accompanying processes have been named “in-source-” (ISD)3 and “postsource decay” (PSD),4 alluding to the time scales on which the * To whom correspondence should be addressed. E-mail: dreisew@ uni-muenster.de. (1) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299. (2) Dreisewerd, K. Chem. Rev. 2003, 103, 395. (3) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67, 3990. (4) Kaufmann, R.; Spengler, B.; Lu ¨ tzenkirchen, F. Rapid Commun. Mass Spectrom. 1993, 7, 902.

5628

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

Figure 1. Nomenclature of peptide fragment ions (Adopted and modified from ref 6. Copyright 2009 American Chemical Society; nomenclature according to ref 5).

fragmentation takes place, i.e., within or after leaving the MALDI ion source of time-of-flight mass spectrometers (TOF MS). PSD species arise from the unimolecular decay of thermally (vibrationally) excited precursor ions. This charge-driven process favors the cleavage of weaker bonds in the precursor ions. As a consequence, for peptides, predominantly b- and y-type fragments are generated, as well as a variety of internal fragment and immonium ions (nomenclature according to ref 5). Similar lowenergy fragment species are produced by low-energy collisioninduced dissociation (CID) tandem mass spectrometry; the latter constitutes the most widely used ion activation method in MS. Both fragmentation techniques provide high analytical sensitivities in the femtomole range of analyte but have two disadvantages. First, post-translational modifications (PTMs) (e.g., of phosphorylated and/or glycosylated peptides) are often not retained in the generated fragments because the labile-bound side groups are easily cleaved. Second, isobaric amino acid (AA) residues cannot be differentiated. This problem is particular relevant for the differentiation of leucine (Leu) and isoleucine (Ile) residues which together compose about 15% of the total AA content in proteins. Differentiation of Leu and Ile by tandem mass spectrometry is possible if diagnostic d- and/or w-type fragment ions (Figure 1) are generated. Their formation involves a side chain cleavage between the β- and γ-carbon of the alkyl residue, next to a cleavage of the N-CR bond of the peptide backbone, and thus requires (5) Papayannopoulos, I. A. Mass Spectrom. Rev. 1995, 14, 49. (6) Lottspeich, F., Engels, J. W., Simeon, A., Eds. Bioanalytik, 2nd ed.; Spektrum Akademischer Verlag: Heidelberg, Germany, 2006. 10.1021/ac1006014  2010 American Chemical Society Published on Web 06/04/2010

a localized high-energy excitation.7 In combination with MALDI or electrospray ionization (ESI), several high-energy dissociation techniques have been developed that potentially allow Leu/ Ile differentiation. Among these are high-energy (he) CID,8 hot electron capture dissociation (hECD),9 VUV-photodissociation,10 and excitation by interaction with metastable rare gas atoms.11 Of these activation techniques, heCID in combination with MALDI dual time-of-flight (TOF-TOF) instrumentation12,13 is currently the only one that is routinely employed. In contrast to MALDI PSD, MALDI ISD species are generated in a fast charge-remote dissociation process that is initiated by the transfer of a neutral hydrogen atom from an excited matrix molecule to, presumably, a carbonyl group of the peptide backbone, giving rise to a hypervalent radical.14,15 Subsequently, the neighboring N-CR bond is primarily cleaved, generating cand z-type ions; a-, y-, and to a lower extent, b-type ions are typically also found in MALDI ISD mass spectrometry. The ISD process resembles similarities to electron capture (ECD16) and electron transfer dissociation (ETD17),15 but because it is initiated by the transfer of a neutral hydrogen rather than a charged particle, it can be used to analyze singly protonated as well as negatively charged ions. Like ESI ECD/ETD, MALDI ISD mass spectrometry may also be utilized for the analysis of larger peptides or even small proteins for top-down protein sequencing.18-20 Eventually, d- or w-type fragments of peptides may also be generated by regular ECD using subthermal electrons,9 and more rarely by MALDI ISD21,25 However, in particular for the ISD case, their occurrence appears as difficult to predict. Therefore, in general, MALDI ISD is not considered as a method that allows discrimination between Leu and Ile.25 However, work by Stimson et al. indicated that the abundance of charge-remote fragment ions, including those arising from side chain cleavages, may be enhanced if particular “cool” MALDI matrixes are employed that are inducing a low level of vibrational excitation.22 MALDI ISD experiments are typically carried out on “axial-” TOF instruments that utilize high vacuum ion sources (p e 10-6 mbar). However, with these instruments, the mass accuracy for assignment of the fragment ions is typically only in the 100 (7) Johnson, R. S.; Martin, S. A.; Biemann, K.; Stults, J. T.; Watson, J. T. Anal. Chem. 1987, 59, 2621. (8) Medzihradszky, K. F.; Campbell, J. M.; Baldwin, M. A.; Falick, A. M.; Juhasz, P.; Vestal, M. A.; Burlingame, A. L. Anal. Chem. 2000, 72, 552. (9) Kjeldsen, F.; Haselmann, K. F.; Budnik, B. A.; Jensen, F.; Zubarev, R. A. Chem. Phys. Lett. 2002, 356, 201. (10) Zhang, L. Y.; Reilly, J. P. Anal. Chem. 2009, 81, 7829. (11) Berkout, V. D. Anal. Chem. 2006, 78, 3055. (12) Yergey, A. L.; Coorssen, J. R.; Backlund, P. S.; Blank, P. S.; Humphrey, G. A.; Zimmerberg, J.; Campbell, J. M.; Vestal, M. L. J. Am. Soc. Mass Spectrom. 2002, 13, 784. (13) Macht, M.; Asperger, A.; Deiniger, S. Rapid Commun. Mass Spectrom. 2004, 18, 2093. (14) Takayama, M. J. Am. Soc. Mass Spectrom. 2001, 12, 420. (15) Ko ¨cher, T.; Engstro ¨m, A.; Zubarev, R. A. Anal. Chem. 2005, 77, 172. (16) Zubarev, R. A.; Kelleher, N. L.; McLafferty, F. W. J. Am. Chem. Soc. 1998, 120, 3265. (17) Syka, J. E. P.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 9528. (18) Suckau, D.; Resemann, A. Anal. Chem. 2003, 75, 5817. (19) Suckau, D.; Resemann, A. J. Biomol. Techn. 2009, 20, 258. (20) Demeure, K.; Quinton, L.; Gabelica, V.; De Pauw, E. Anal. Chem. 2007, 79, 8678. (21) Marzilli, L.; Golden, T. R.; Cotter, R. J.; Woods, A. S. J. Am. Soc. Mass Spectrom. 2000, 11, 1000. (22) Stimson, E.; Truong, O.; Richter, W. J.; Waterfield, M. D.; Burlingame, A. L. Int. J. Mass Spectrom. Ion Processes 1997, 169-170, 231–240.

ppm range. Thus, structural assignment is difficult based solely on the m/z values of individual ion signals, in particular those of low intensity. This drawback can be attributed to the high laser fluences that are necessary to generate a sufficient ISD yield. In turn, these conditions produce wide kinetic energy spreads of the emitted ions.23 The problem is exacerbated by the metastable character of the ISD fragments, which may lead to subsequent fragmentation reactions.22,24 The latter factors and the general overlap between PSD and ISD processes generally also hampers the MALDI ISD TOF mass analysis of PTM peptides which are carrying labile side groups. Several studies have nevertheless demonstrated that the initial ISD fragments generally retain even labile-bound functional groups, like phosphate in phosphopeptides.25 Orthogonal (o-)TOF mass spectrometers are typically operated with buffer gases for collisional cooling and focusing. This configuration provides a laser power-independent mass accuracy for the analysis of both molecular and fragment ions26,27 and reduces the extent of metastable fragmentation. For modern instruments, the mass accuracy is in the low parts per million range. At optimal collisional cooling conditions, which are typically realized at a gas pressure in the MALDI ion source of about 1 mbar, thermal fragmentation channels are effectively suppressed.27–29 Conversely, a reduction of the source gas pressure can provide an easy means to obtain structural information by allowing vibrationally excited analyte ions to dissociate. These processes may be enhanced by increasing the ion extraction potential.27 Mostly, the occurrence of fragment species like these are also observed in MALDI PSD and low-energy CID experiments (e.g., y-type ions) and has been noted in works addressing these latter features.27,28 An interesting feature of the collisional cooling is that also PTM peptides carrying labile side groups are stabilized. By reduction of the buffer gas pressure, cleavage of the side group(s) may be induced and this way the modification characterized to some extent. So far such analyses appear to have focused on the loss of the functional groups from the molecular ion but not from backbone fragments.27 However, only the analysis of the latter would typically allow identification of the modification site. We have recently shown that typical ISD fragment species (for instance c-type ions of peptides) may also be utilized for structural analysis by UV-MALDI o-TOF mass spectrometry.29 Notably, the highest abundances of these species were obtained at optimal collisional cooling conditions, while the yields of the ISD ion species dropped significantly with reduced buffer gas pressure. In line with previous assumptions,30 these observations suggested that the ISD ions generally constitute internally excited species (23) Berkenkamp, S.; Menzel, C.; Hillenkamp, F.; Dreisewerd, K. J. Am. Soc. Mass Spectrom. 2002, 13, 209. (24) Brown, R. S.; Feng, J.; Reiber, D. C. Int. J. Mass Spectrom. 1997, 169/170, 1. (25) Hardouin, J. Mass Spectrom. Rev. 2007, 26, 672. (26) Krutchinsky, A. N.; Loboda, A. V.; Spicer, V. L.; Ens, W.; Standing, K. G. Rapid Commun. Mass Spectrom. 1998, 12, 508. (27) Loboda, A. V.; Ackloo, S.; Chernushevich, I. V. Rapid Commun. Mass Spectrom. 2003, 17, 2508. (28) Ackloo, S.; Loboda, A. Rapid Commun. Mass Spectrom. 2005, 19, 213. (29) Soltwisch, J.; Souady, J.; Berkenkamp, S.; Dreisewerd, K. Anal. Chem. 2009, 81, 2921. (30) Brown, R. S.; Feng, J. H.; Reiber, D. C. Int. J. Mass Spectrom. 1997, 169, 1.

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5629

that may readily fragment in secondary processes, unless these are rapidly de-excited. Therefore, it may be speculated that the occasional observation of d- and w-type fragments in ISD experiments may be associated with a high excitation of these species,22 such that they are produced but lost in subsequent dissociation processes. Intriguingly, these informative fragment species might then be preserved by rapidly de-exciting them. Here, we show that under optimal collisional cooling conditions, in combination with the application of elevated laser fluences, small percentages of diagnostic d- and w-type ISD ions of peptides appear to be “consistently” produced by MALDI ISD o-TOF mass spectrometry. We demonstrate the utility of this feature with the example of two pairs of peptides with exchanged Leu/Ile residues. In the second part of this communication, we show that at high gas pressure, labile post-translational modifications are also retained in the ISD fragments but that the side groups can be cleaved off by reducing the gas pressure. The examples of a phosphorylated and an O-glycosylated peptide are presented. Finally, we demonstrate that recording mass spectra at low and high buffer gas pressure also allowed obtaining information about the amino acid sequence in the cyclic part of a peptide cross-linked by an intramolecular disulfide bridge. EXPERIMENTAL SECTION Renin substrate (porcine, DRVYIHPFHLLVYS, MW 1759.04 u), fibrinopeptide A (human ADSGEGDFLAEGGGVR, MW 1536.57 u), Ala-Gly-[Arg8]-vasopressin (C3-C8; MW 1212.36 u), ubiquitin (human, MW 8564.84 u), 2,5-dihydroxybenzoic acid (DHB), 1.5-diaminonaphthalene, and 5-aminosalicylic acid were from Sigma-Aldrich (Steinheim, Germany), modified renin substrate (DRVYIHPFHLIVYS) and fibrinopeptide A (ADSGEGDFIAEGGGVR) were from GenScript (Piscataway, NJ), phosphorylated angiotensin II (DRV(pY)IHPF, MW 1126.17 u), and O-glycosylated erythropoietin (EPO, 117-131; EAISPPDAA-*SAAPLR, where *S is GalNAc-Ser; MW 1668.82 u) were from Protea Biosciences (Nimes, France). For the MALDI MS analysis, 0.5-1 µL of aqueous DHB matrix solution was mixed with equal volumes of aqueous 10-4 M peptide solution on the sample plate and allowed to dry. The employed prototype UV-MALDI o-TOF mass spectrometer has been described previously.29,31 An N2 laser (λ ) 337 nm; τ ) 3 ns) served for desorption/ionization. The focal spot size on the sample was ∼200 × 230 µm2. Laser fluences were adjusted to 900-1000 J/m2, approximately a factor of 2.5 above the ion detection threshold fluence. Mass spectra were acquired during 90 s at a laser pulse repetition rate of 30 Hz. Ions were generated in an N2 buffer gas atmosphere of adjustable pressure (between ∼0.05 and 2 mbar). Pressure values were read out with a capacitive gauge as described.29 For the analysis of the ISD fragments, a pressure of 1.2 mbar provided the best results. After passing a differentially pumped quadrupole ion guide region, ions were mass-separated using a TOF potential of 10 kV. All mass spectra were acquired in positive ion mode. MoverZ software (Genomic Solution, Ann (31) Dreisewerd, K.; Mu ¨ thing, J.; Rohlfing, A.; Meisen, I.; Vukelic´, Zˇ.; PeterKatalinic´, J.; Hillenkamp, F.; Berkenkamp, S. Anal. Chem. 2005, 77, 4098. (32) O’Connor, P. B.; Lin, C.; Cournoyer, J. J.; Pittman, J. L.; Belyayev, M.; Budnik, B. A. J. Am. Soc. Mass Spectrom. 2006, 17, 576.

5630

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

Arbor, MI) was used for data evaluation. With the application of internal calibration and the use of the signal of the matrix and that of the molecular analyte ion, a mass accuracy of better than 20 ppm was generally achieved. Peptide sequence data were obtained using Protein Prospector v 5.3.2 (University of California, San Francisco, CA). For all assigned ion signals, experimental and theoretical mass values matched within this accuracy range. Unless noted, all assigned species represent singly protonated molecules.

RESULTS AND DISCUSSION Differentiation of Leucine and Isoleucine Residues. MALDI o-TOF mass spectrometry allowed differentiation of Leu and Ile residues in peptides. As examples, mass spectra of renin substrate, fibrinopeptide A, and their Leu/Ile-substituted isomers, recorded at a nominal buffer gas pressure of 1.2 mbar of N2, are plotted in Figures 2a,b and 3a,b, respectively. Depending on the position of the basic arginine residue, a partial series of c-type (renin substrate; Figure 2) or z-type ions (fibrinopeptide A, Figure 3) are detected. In addition, a-, b-, and y-fragment series of comparable abundance complement the AA sequence assignments; da- and/or wa-type ions, diagnostic for Leu/Ile, upon displaying the characteristic mass difference of 14.02 u (Figure 1) are detected in all four peptides. In addition, db or wb ions are recorded for fragmentation of the Ile residue (Figures 2b and 3b). Side chain cleavage was also observed for Val, Ser, Glu, and Asp residues. Hydrogen rearrangement reactions32,33 are indicated for some residues and gave rise to a few satellite peaks [e.g., c10 + 1 next to c10 for the 9Ile-containing peptide (Figure 2b, bottom); interestingly, this doublet is not detected for the Leu residue (Figure 2a, bottom)]. Experiments with further test peptides (e.g., substance P and neurotensin) confirmed the general fragmentation patterns, in particular the occurrence of the d- and w-type fragments. Compared to the signals of the intact molecular ions [M + H]+, which are detected with an intensity of ∼15 000 counts (small insets in Figures 2a,b and 3), d- and/or w-type ions of the four test peptides are detected with counts of a few tens only (lower trace in Figures 2 and 3). The unspecific (homogeneous) chemical noise, as common for UV-MALDI o-TOF mass spectrometry, produces background signals with ion counts of about 5 in the m/z range of ∼1300 (as displayed in Figure 2a,b bottom) and about 7 at an m/z of 700 (Figure 3a,b bottom). Hence, the displayed da11 fragment of renin substrate is detected with a signal-to-noise (s/n) ratio of ∼4.5, while the wa8 fragment of fibrinopeptide A is exhibiting an s/n of ∼8. Despite the ostensively low s/n ratio of the diagnostic ion signals, they can still clearly be identified amidst the homogeneous chemical noise and be used specifically for the differentiation of the isobaric residues. For renin substrate and fibrinopeptide A, the current limit of detection for the d/w-fragments was about 10-20 pmol of analyte prepared for the MALDI ISD MS analysis. In comparison, the generation of d-/w-fragments from (33) Savitski, M. M.; Kjeldsen, F.; Nielsen, M. L.; Zubarev, R. A. J. Am. Soc. Mass Spectrom. 2007, 18, 113.

Figure 2. MALDI o-TOF mass spectra of (a) renin substrate (porcine, MW 1759.04 u) and (b) its (Leu/Ile)11 substituted isomer recorded at a buffer gas pressure of 1.2 mbar; the small insets show the full mass spectra including the molecular ion regions, the expanded regions at the bottom the m/z range from 1270 to 1370. Asterisks denote matrix cluster-derived ion species. (c) Mass spectrum of renin substrate acquired at the lowest accessible buffer gas pressure of 0.05 mbar. The assigned cnym - NH3 fragment species are formally identical to bnym species. However, experiments with varied buffer gas pressure suggest that these species are generated by secondary fragmentation of excited c-type fragments; d/w-type fragments are not detected at the gas pressure of 0.05 mbar.

similar peptides by MALDI heCID13,22 and nano-ESI hECD34 MS was previously reported from about 1 pmol of prepared analyte. Even at analyte concentrations that are producing only low s/n ratios close to the detection limit, the presence or strict absence of the diagnostic peaks, in addition to the information provided by the series of other more abundant c/z-, y-, a-, and b-type species, may provide decisive information. In the case of the four test compounds, determination of the complete AA sequence was possible. A MALDI o-TOF mass spectrum of renin substrate that was acquired at the minimal buffer gas pressure of ∼0.05 mbar is displayed in Figure 2c. In line with our previous findings,29 at this pressure the abundances of the cn- and zn-type ions are substantially reduced. Instead, a set of secondary fragment species of the type cnym - NH3 is now detected; d- and w-type ions were not detected at the low pressure setting for any of the tested peptides. The abundances of the secondary fragments arising from the d- and w-type ions are probably too low to allow their detection. (34) Kjeldsen, F.; Haselmann, K. F.; Sørensen, E. S.; Zubarev, R. A. Anal. Chem. 2003, 75, 1267.

The dependence of the overall ISD-type fragment ion yield on the gas pressure has been determined in detail in ref 29. However, because lower laser fluences (by about a factor of 1.5) were used in that work, d/w-type fragments were not produced/detected. The normalized yield-pressure dependences of the diagnostic da11ion of Iso/Leu substituted renin substrate and of the flanking c10-ion are plotted in Supplementary Figure 1 in the Supporting Information for comparison. Both ISD-type fragments display essentially the same yield-pressure dependence and reproduce those determined previously.29 A slightly faster decline of the d-fragment abundances with decreasing pressure may reflect a stronger internal excitation. With increasing pressure, fragment abundances first rise sharply before saturation of the ISD ion yield is reached at about 0.9 mbar of N2. Because matrix adduct cluster abundances (see below) tend to increase further with gas pressure, the optimal pressure setting for the MALDI ISD MS analysis is typically found just after reaching this saturation plateau. MALDI ISD mass spectra of the same samples recorded with an axial delayed extraction TOF mass spectrometer (Reflex III, Bruker Daltonik, Bremen, Germany) did not display any d/wfragment ions (data not shown). This instrument is operated with Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5631

Figure 3. MALDI o-TOF mass spectra of (a) fibrinopeptide A (human, MW 1536.57 u) and (b) its (Leu/Ile)9-substituted isomer recorded at a buffer gas pressure of 1.2 mbar. The small insets show the full mass spectra including the molecular ion regions, the expanded regions at the bottom the m/z range from 670 to 770. Asterisks denote matrix cluster derived ion species.

high vacuum of ∼1 × 10-7 mbar in the ion source. The lack of collisional cooling could explain why typically no d/w-fragments are found in MALDI ISD MS measurements utilizing this instrument type. 1,5-Diaminonaphthalene (DAN) and 5-amiosalicylic acid, two other ISD active matrixes,35 were also tested for MALDI ISD MS analysis of the test peptides. Under the current experimental conditions, neither matrix showed a significant advantage compared to DHB (data not shown). Employment of DAN for the analysis of very large peptides (e.g., ubiquitin; MW 8565 u) resulted in slightly enhanced abundances of ISD fragment ions (data not shown). However, a disadvantage of the DAN matrix is that it produces a pronounced background of matrix ion clusters under the elevated pressure conditions. Partially, these clusters are exhibiting m/z values of up to 1000 and above (data not shown). Ubiquitin was the largest peptide investigated so far, that for the prepared amount of 30 pmol, produced useful backbone fragment ion series. For example, ∼60% of the theoretical y-type fragments were detected in this case and by using the DHB matrix. Because of the generally lower intensities of individual fragment ion signals obtained from this large peptide, d/wfragments of Leu/Ile could only be identified for 2 out of 16 residues (data not shown). Analysis of Post-Translationally Modified Peptides. To examine whether the combination of MALDI ISD and collisional cooling could also be used to identify the modification sites in PTM peptides, we investigated phosphorylated angiotensin II (phosphorylation at tyrosine) and O-glycosylated [GalNAc-Ser]erythropoietin (EPO, 117-131). Mass spectra of the two compounds, recorded at 1.2 mbar of N2, are displayed in parts a and c of Figure 4, respectively. In line with the results of previous studies,27 in both cases the mass spectra are dominated by far by (35) Sakakura, M.; Takayama, M. J. Am. Soc. Mass Spectrom. 2010, 2, 979.

5632

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

the molecular ion signal [M + H]+ (not shown) while [M - X + H]+ signals (where X is either HPO3 or GalNAc) are detected with an intensity of about 2-3% of the [M + H]+ peak. A low intensity signal at m/z 1028.56 may moreover indicate partial loss of H3PO4 from the molecular phosphopeptide ion. On the basis of the mass values, these species can, however, not be differentiated from those which would result from a sequential loss of HPO3 and water. [M - H2O + H]+ ions are detected with intensities comprising about 1% of that of the molecular ion signal. A series of N-terminal ions, (a,b,c)3-7, identifies the AA sequence of the peptide except for the first two residues. At the high pressure setting, ions with masses below about 300 u become exceedingly difficult to differentiate from the chemical background. Matrix cluster ions also make out a large part of the unlabeled signals visible in the spectrum. This problem is at least mitigated by the fact that the matrix clusters are detected as distinct, well reproduced species and can for the most part thus be readily differentiated from the analyte signals (see discussion below); for improved clarity the matrix-derived ions are not labeled in Figure 4. Partial loss of neutral HPO3 is found from some of the a-, b-, and y-type fragment ions of the phosphopeptide (e.g., for the a5 species in Figure 4a) but was not observed for any of the c-, z-, or w-type fragments, an observation that is likely to reflect the different dissociation pathways. Given the already relatively low signal intensities of the ISD fragment ions, it can, however, not be excluded that a small percentage is producing the neutral loss, leading to ion species comprising even lower abundances, not discernible from the chemical background. Lowering the buffer gas pressure increases the extent of cleavage of HPO3 (and/or GalNAc) from the a-, b-, and y-type fragments; this technically feasible step (adjusting the pressure

Figure 4. MALDI o-TOF mass spectra of post-translationally modified peptides: (a,c) phosphorylated angiotensin II (MW, 1126.17 u), (b) glycosylated [GalNAc-Ser]-erythropoietin O-glycosylated erythropoietin (EPO, 117-131; MW 1668.82 u), recorded a buffer gas pressure of (a,c) 1.2 and (b) 0.05 mbar. Signals labeled with # could also represent ions produced by direct loss of phosphoric acid H3PO4 from the molecular ion; the [M + H - 15]+ ion species in part a presumably represents loss of neutral NH.35

takes a few seconds) can thus be used to corroborate the assignment of the modification site. Figure 4b shows the example of a mass spectrum which was recorded at the lowest accessible buffer gas pressure of ∼0.05 mbar. In line with our previous findings,29 this spectrum demonstrates a significant increase in the overall fragment ion yield because of the now active thermal (PSD) fragmentation channel. Cleavage of HPO3 from the c-/z-type fragments was at this low pressure also observed for some of the species but resulted in very low intensities of the corresponding ion signals, barely above the noise level. Since the matrix cluster background is substantially reduced at the low gas pressure, AA assignment can now be extended to include also smaller fragments like b2,3 and y2,3 ions. Together, the full AA sequence is covered. For the O-linked glycopeptide, a similar behavior is found (Figure 4c; only the high pressure data are shown). Because of the C-terminal position of the basic arginine, a series of N-terminal y-, z-, and w-ions can be identified in this case. Alike for the

phosphopeptide, partial loss of the glycan moiety is only observed for the PSD-type y-fragments. The absence of radical satellite peaks of the form Y0 ± 2H•, where Y0 ) M-GalNAc, as typically observed in the ECD tandem MS analysis of O-linked glycopeptides performed with a Fourier-transform ion cyclotron resonance (FTICR) instrument36 furthermore indicates that radical site induced elimination processes probably do not play a role in the cleavage of the glycan moiety. Together, these data suggest that post-translational modifications can be tracked by MALDI o-TOF ISD mass spectrometry and modification sites identified in a straightforward way. Analysis of a Cyclic Disulfide Peptide. Finally, to ask whether the approach may also provide advantageous features regarding the analysis of peptides that are containing an intramolecular disulfide bond,18,37,38 we investigated Ala-Gly-[Arg8]vasopressin. At high gas pressure, fragment ion species that were either containing the cross-linked C3-C8 part or representing the AA sequence outside the disulfide loop were detected with highest abundance (Figure 5a). However, a closer look suggests that various low abundant b-, c, y-, and w-type ions, arising from backbone cleavage and ring-opening, are also found [e.g., Y4 (S•) and c7(SH) + Na, where S• denotes a fragment containing a sulfur radical]. Future work will have to evaluate to which extent a general fragmentation scheme can be derived for the dissociation of disulfide-bridged peptides.38 Reducing the gas pressure to 0.05 mbar gave rise to additional fragments resulting from peptide bond cleavage in the cyclic part and, hence, may add valuable information about the AA sequence of the disulfide peptide (Figure 5b). Also, ion signals that are presumably representing internal fragments from the cyclic part of the peptide (e.g., QN, FQ, and FQN) are detected. Notably, both symmetric cleavages of the disulfide bond [e.g., b4(SH)] as well as asymmetric cleavage [e.g., b4R (see figure caption for nomenclature)] are furthermore observed, next to a multitude of fragment ion signals which have so far only partially been assigned. The tandem MS analysis of peptides that are cross-linked by a disulfide bridge generally provides a challenge because AA sequence information can often only partially be obtained from underivatized compounds; typically chemical reduction of the disulfide bonds is therefore required.37 The approach described here could constitute a complementary method for obtaining initial AA sequence information from underivatized peptides, even if currently only low s/n yields are obtained for most of the fragments reflecting the cyclic part and the dissociation pathways are not well understood. Matrix Cluster Ion Series. Mass spectra acquired at elevated buffer gas pressure contain a distinct series of matrixderived cluster ions (cf. Figure 5a for an example). Their abundance is generally even more enhanced in the negative ion mode (data not shown). Notably, in both ion modes these species are, however, only detected with more sizable abundance at elevated buffer gas pressure. While up to ∼m/z 700, DHB cluster ions are then generally of comparable or even (36) Mormann, M.; Paulsen, H.; Peter-Katalinic´, J. Eur. J. Mass Spectrom. 2005, 11, 497. (37) Fukuyama, Y.; Iwamoto, S.; Tanaka, K. J. Mass Spectrom. 2005, 41, 191. (38) Mormann, M.; Eble, J.; Schwoppe, C.; Mesters, R. M.; Berdel, W. E.; PeterKatalinic´, J.; Pohlentz, G. Anal. Bioanal. Chem. 2008, 392, 831.

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5633

Figure 5. MALDI o-TOF mass spectra of Ala-Gly-[Arg8]-vasopressin (C3-C8; MW 1212.36 u) recorded at a buffer gas pressure of (a) 1.4 and (b) 0.05 mbar. SH denotes a fragment species resulting from reductive cleavage of the disulfide bond, a S• fragment containing a sulfur radical; β denotes a fragment species that arises from asymmetric S-S cleavage, where the SSH moiety in retained in the fragment ion, while R ions lost it and lead to dihydroalanine; nomenclature according to ref 38. In part a, presumable identities of cluster ions of the DHB matrix are assigned; X is tentatively assigned to a neutral loss from DHB: HsC≡CsCO2H.

higher abundance (in the low m/z range) than the analytederived fragment ions (for the 20-30 pmol of peptide analyzed here), their signal intensities typically drop below noise level above m/z values of ∼800 (cf. Figure 5a). On the basis of the mass values, the most prominent matrix cluster ion series can be tentatively assigned to [(DHB - H2O)n + H]+ and [(DHB - H2O)n + X + Na]+, respectively, which differ by mass increments of 137.02 and 70.01 u, respectively (values are for the monoisotopic masses). On the basis of its mass, X is presumably a neutral loss from DHB giving rise to HsC≡CsCO2H (theoretical monoisotopic mass, 70.005 u). Similar cluster ion series, which were, however, not fully evaluated in that work, were described previously by Krutchinsky and Chait in a study performed at atmospheric pressure using a Paul ion trap as the mass analyzer.39 CONCLUSIONS MALDI ISD o-TOF mass spectrometry in combination with collisional cooling provides a few noteworthy features. From a mechanistic point of view, valuable insights into the energetics and fragmentation pathways of MALDI-generated ions can be 5634

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

obtained. However, the methodology may also be useful for structural identification of peptides and can, for example, allow differentiation of isobaric Leu/Ile residues via the generation of diagnostic d- and w-type ions. The observation of these highenergy fragments has not been consistently reported before in the context of MALDI ISD mass spectrometry and obviously requires a sufficient activation cross section using an ISD-active matrix and by applying high laser fluences. At the same time, the dissociation products need to be stabilized by rapid collisional cooling. MALDI ISD o-TOF MS performed at low and elevated buffer gas pressures may also form an interesting add-on for the analysis of post-translationally modified peptides by allowing identification of the modification site. The preliminary results also indicate the possibility to obtain structural information about the cyclic part of disulfide-bridged peptides from underivatized samples, though these mass spectra are particular complex and the dissociation pathways are so far not well understood. Current sensitivity limits for the analysis of standard peptides are in the 10 pmol range. Therefore, one of the major challenges for future developments will clearly be to increase the analytical sensitivity. Eventually, such an improvement may be achieved by identifying particular ISD-active matrix compounds or by adding hydrogen donors.40 Future studies will also have to reveal to which extent other types of mass analyzers (e.g., FTICR and Paul ion trap instruments) can be employed for a similar ISD analyses at optimal buffer gas pressure. In this regard, a study performed by Fuchser et al. with an FTICR mass spectrometer appears to show very promising results, even if this study seems to have focused on the identification of the classical c- and z-type ISD fragments.41 Two disadvantages of the current protocol are the need to purify more complex samples prior to the MS/MS analysis and the potential overlap of fragment and matrix ion signals. The first may be overcome by applying standard liquid chromatography protocols, as was suggested previously.27 Choosing matrixes that produce relatively low abundances of cluster ions and/or differential analysis after identification of matrix compounds could address the second problem. At least for the here studied DHB matrix, distinct cluster ion series could rapidly be identified. Naming the two main MALDI fragmentation channels as ISD and PSD was closely related to the historical development of these techniques, which was widely based on the use of axial TOF instruments in combination with high-vacuum ion sources. If alternative configurations are realized, e.g., by adapting atmospheric and intermediate pressure ion sources to quadrupole ion trap, FTICR, or o-TOF mass spectrometers, the terminology unfortunately appears as notably imprecise because of the different residence times of ions in the ion source. In view of the essentially elucidated dissociation pathways,15 one might therefore argue whether the use of a revised terminology (e.g., hydrogen transfer dissociation (HTD) in the ISD case, in analogy to ETD) would (39) Krutchinsky, A. N.; Chait, B. T. J. Am. Soc. Mass Spectrom. 2002, 13, 129. (40) Delvolve, A.; Woods, A. S. Anal. Chem. 2009, 81, 9585. (41) Fuchser, J.; Witt, M.; Macht, M. Annual Conference of the Association of Biomolecular Resource Facilities (ABRF), Memphis, TN, February 7-10, 2009; poster V61-S2.

not describe the involved process and hence the methodology better.

ACKNOWLEDGMENT We would like to thank Stefan Berkenkamp and Sequenom GmbH for providing use of the o-TOF instrument, Gottfried Pohlentz and Alfredo Jesus Iban ˜ éz for help with peptide identification, and Michael Mormann for helpful discussions. J.S. is grateful for a Ph.D. grant from the subsidy program

Innovative Medizinische Forschung (IMF) of the University Hospital Mu¨nster. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review March 5, 2010. Accepted May 18, 2010. AC1006014

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5635