RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 6,257-264 (1992)

Evidence for a Lysine-specific Fragmentation in Fast-atom Bombardment Mass Spectra of Peptides Bradley L. Ackermann,* Robert J. Barbuch, John E. Coutant, John L. Krstenanskyt and Thomas J. Owen Marion Merrell Dow Research Institute, 2110 East Galbraith Road, PO Box 156300 Cincinnati, OH 45215-6300. USA SPONSOR REFEREE: Dr Simon J. Gaskell, Center for Experimental Therapeutics. Baylor College of Medicine, Houston, TX 77030, USA

A fragmentation process observed for peptides that contain lysine, or other amino acids which possess a free amino group on their sidechain, is reported. The ions generated by this process are found 16Da below the acylium-type B ions that result from fragmentation at the C-terminal side of lysine or other amine-containing residues in fast-atom bombardment (FAB) mass spectra. These ions, which are referred to as (B - 16) ions, permit differentiation between the isobaric amino acids lysine and glutamine in peptide mass spectra. High resolution measurements indicate that (B - 16) ions differ in composition from the corresponding B ions by the removal of one oxygen atom. Formation is believed to occur through a cyclization process initiated by nucleophilic attack by the free amino group of the lysine sidechain at the carbon of the acylium ion (B ion). A similar process initiated directly from the protonated peptide may also occur. Analogous cyclization processes are restricted for glutamine because this residue is comparatively less nucleophilic than lysine (i.e., amide vs amine). Although (B - 16) ions have been detected under high energy collisionally induced dissociation, they are formed less readily than by FAB mass spectrometry. A mechanism consistent with this observation as well as other experimental evidence is presented to account for the formation of (B - 16) ions.

The success or failure of peptide sequence analysis using fast-atom bombardment (FAB) mass spectrometry is directly related to the fragmentations observed. As a result, considerable attention has been given to understanding how peptides fragment during FAB. To date, a number of fragmentation routes have been reported.’-‘ A majority of the fragments observed arise from cleavage at, or adjacent to, the relatively labile amide bonds which connect the individual amino acids. Peptide sequence analysis is facilitated because most fragment ions retain either the N- or C-terminus in their structure. This leads to the production of fragment-ion series which differ in mass according to the residue weights of the consecutive amino acids in the sequence. These common fragmentation schemes have been widely used in peptide sequencing and are designated by a standard nomenclature first proposed by Roepstorff and Fohlman’ and later modified by Biemann.’ The ability to sequence peptides entirely by mass spectrometry has been demonstrated.”-” The general strategy involves enzymatic degradation of a larger peptide or protein into a series of smaller peptides which are amenable to sequence analysis by FAB in conjunction with collisionally induced dissociation (CID). A fundamental limitation to this approach, however, is the inability to guarantee fragmentation at each amide bond, particularly as the molecular weight of the peptide exceeds 2500 Da.I2 Consequently, considerable effort has been directed at finding ways to increase the efficiency of forming peptide fragment ions Author to whom correspondence should be addressed. ‘Current address: Institute of Bio-Organic Chemistry, Syntex Research, Palo Alto, CA 94303, USA.

095 1-4198/92/040257-08 $05.00

01992 by John Wiley & Sons, i Ltd

-in the mass spectrometer. Reported alternatives to conventional CID include photon-induced dissociation (PID),I3.l4 and surface-induced dissociation (SID). l 5 Other approaches, such as the fragmentation of multiply charged ions formed by electrospray l h and the use of ion-trap detection (ITD) schemes,I7 confer greater efficiency upon the overall CID process. Another problem encountered in peptide sequencing is the inability to distinguish isobaric amino acids using the dominant fragment-ion series of the ptptide backbone. In this regard, the observation of sidechainspecific fragmentation has proven crucial to sequencing unknown peptides by mass spectrometry. The most important example is the identification of D and W ions by Johnson and coworkers.7 These investigators demonstrated that under high-energy conditions it is possible to observe fragmentation of the C13-Cy bond on the sidechains of many amino acids. In the case of Leu vs Ile, fragmentation of this bond reveals the branching pattern and, hence, the isomeric identity. A similar problem is encountered with the isobaric amino acids lysine and glutamine (residue weight 128Da) which also cannot be distinguished by the normal fragment-ion series. Typically, these residues are differentiated by selective derivatization of lysinelx (which requires a separate analysis), or by enzymatic specificity if the peptides were formed by tryptic digestion. Although not widely used, other reported methods involve the selective derivatization of glutamine’”.”’ and the observation of perchlorate adducts when lysine-containing peptides are run in a FAB matrix containing perchloric acid.” Over the course of analysing the fragmentation patterns of several peptides prepared by solid-phase peptide synthesis, a fragmentation process unique to lysine Receioed 12 February 1992 Accepted 12 February I992

258

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

I C';

.

xsl

' 0°1

I

'347

751 50

000

1100

1200

1300

11

m'z Figure 1. FAB mass spectrum of substance P, H-Arg-Pro-Lys-Pro-GIn-Gln-Phe-Phe-Gly-Leu-Mct-NH* in glycerol thioglycerol

+

(and other residues containing a free amino group) was discovered. The resulting fragments appear 16 Da below the acylium-type B ions which result from cleavage at the C-terminal side of lysine or other residues containing a free amine. These products, referred to as (B - 16) ions, have been observed both by FAB and under high-energy CID conditions. In this Communication, the characteristics of (B - 16) ions are discussed and a mechanism is proposed to account for their formation.

EXPERIMENTAL Peptide synthesis The peptides used in this study were synthesized on a model 430A peptide synthesizer (Applied Biosystems, Foster City, CA, USA) using standard solid-phase methods. Substance P was purchased from Sigma Chemical Co. (St Louis, MO, USA). Peptides synthesized in-house were purified by reverse phase highperformance liquid chromatography (HPLC) prior to FAB mass spectrometric analysis.

Mass spectrometry All data were recorded on a ZAB2-SE double focusing mass spectrometer (VG Analytical Ltd. Manchester,UK) equipped with a cesium ion gun. A cesium anode potential of +25 kV and a source acceleration voltage of +8 kV were used for all experiments. Peptides were either applied as solids or from solution

(acetonitrile + water) to the FAB matrix on the probe tip. The maximum quantity used when the samples were applied neat was estimated to be 0.2 mg. Matrices used included 1:l (w/w) glycerol thioglycerol, glycerol, and rn-nitrobenzyl alcohol (MNBA). Mass calibration was performed using cesium iodide as the reference compound. For linked-scanning experiments, a parent-ion attenuation of about 70% was used with helium as the target gas. Mass-analysed ion kinetic energy (MIKE) spectra, however, were acquired for metastable ions without collision gas. Accurate mass determination was performed at a resolution of 10 000 (10% valley definition) using a voltage scan to cover the mass region of interest. Both known peptide fragments and dominant matrix peaks were used as mass references. High resolution and tandem mass spectrometric (MS/MS) data were collected as continuum data and integrated by the multi-channel analyser mode of the VG data system.

+

RESULTS AND DISCUSSION An example of (B - 16) ion formation is shown in Fig. 1 which displays the FAB mass spectrum of substance P. The ion at m/z 366 is assigned as (B, - 16) in extension of the standard nomenclature7 and emphasizing the structural relationship (as discussed below) to the corresponding B3 ion (rn/z382) formed from cleavage after lysine. An equally important observation is that the analogous fragmentation was not observed for either glutamine residue. The corresponding ions for Gln residues in the 5 and 6 positions would appear at m / z 591 and 719, respectively.

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

A second example of (B - 16) ion formation appears in Fig. 2 which contains the FAB mass spectrum of a 13 amino acid model a-helical peptide (MAP). This particular peptide contains two lysine residues in addition to a succinyl group at the N-terminus. The FAB mass spectrum indicates that two (B - 16) ions were formed: (B, - 16) ( m / z 568) and (BIl- 16) ( m / z 1350). Both of these ions were also detected by high energy CID. A portion of the linked scan at constant BIE for the [ M + H ] + ion of MAP (mlz 1738) is shown in Fig. 3. The peaks at mlz 1350 and 1366 are assigned as (B,,- 16) and B , , , respectively. In order to understand how (B - 16) ions are formed, high-resolution measurements were performed. In all instances, it was concluded that (B-16) ions differ from the corresponding B ions by one oxygen atom. For substance P, the (B3- 16) ion was found to have a m / z value of 366.2621. This value differs by +1.1 ppm from the composition predicted for the loss of oxygen from the B3 ion (i.e., C17H32N702). Other differences of 16 Da from the B, ion, such as NH2and CH4,were also considered. However, the m/z values predicted for these compositions differed from the observed value by -66 and -100 ppm, respectively. In an attempt to define the origin of (B - 16) ions, several experiments were undertaken using both MIKE and linked scanning. Two potential precursors, B and C , were considered in addition to [M + HI+. MIKE spectra were recorded for several lysine-containing B ions and in all cases a loss of 1 6 D a was observed. However, in most cases intense losses of 17 and 18 Da

50

v;’ ClA

were also detected and in some instances a loss of 16 Da was recorded from non-lysine-containing B ions. At the same time, no evidence was obtained under MIKE examination for the formation of (B - 16) ions from any lysine-containing C ion examined. The possibility that (B - 16) ions are products of B ions is consistent with the observation that (B - 16) ions are often present for N-terminally blocked peptides since blocked peptides are known to favor the production of B ions”. As a complement to the above experiments, parent-ion scans were attempted for various (B - 16) ions using the method of linked scanning at constant B 2 / E . Unfortunately, (B - 16) ions could not be positively confirmed as products of B ions. A lack of sensitivity may have contributed to this result. In accordance with the above data, a mechanism was proposed to account for (B - 16) ion formation. The general scheme is illustrated in Scheme 1 for the lysine homologue ornithine. Although B ions are thought to be precursors of (B-16) ions (pathway A), other possible mechanistic routes originating directly from the protonated peptide (pathways B and C) cannot be dismissed from the data at hand. In all instances, (B - 16) ion formation is initiated by nucleophilic attack by the free amino group on the ornithine sidechain. Attack may occur either at the carbon atom of the acylium type B ion (pathway A) or at the carbonyl carbon of the adjacent peptide bond (pathways B and C). When attack begins from the B ion, a cyclic, ahydroxy immonium species is produced. This resonance-stabilized intermediate is also common to

86 697

&+ti)+ 1730

1001

259

I

50

Figure 2. FAB mass spectrum of MAP, succinyl-Leu-Leu-Glu-Lys-Leu-Leu-Glu-Trp-Leu-Leu-Lys-Glu-Leu-Leu-NH, , in glycerol + thioglycerol.

260

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES B1 3

100-

?

HOOC-C-

L- L

-

E

-

K

-

L

-

L

-

E

-

-

W

L

-

K

- E - L - L - NH2

80-

60-

40-

20-

0-

I

I

300

200

I

500

400

100

TIME x 10

mlz Figure 3. Linked scan at constant BIE of [M + HI+ ions, rnlz 1738, of MAP.

pathway B. In this latter case, cyclization within the protonated peptide is followed by the elimination of the C-terminal portion of the peptide as a neutral amine to arrive at the common 6-membered intermediate. Once the cyclic immonium species is formed, a displacement of a hydroxyl radical by a hydrogen radical leads to the formation of the cyclic resonance-stabilized (B - 16) ion. Pathway C represents a third possible route to the (B - 16) ion. In a manner similar to pathway B, nucleophilic attack occurs within the protonated peptide. However, in this case, water is eliminated to form a

Pathwav A \ -YNH'

'n

cyclic immonium species. The C-terminal portion of the peptide is then displaced by a hydrogen radical to account for formation of the resonance-stabilized (B - 16) ion. Other possible precursors for (B-16) ions were considered as the mechanism of Scheme 1 was formulated. Certainly, a more favorable process than radical displacement would be to lose a water molecule from a (B +2) ion. However, this possibility was dismissed because (B + 2) ions have not been observed for any of the peptides exhibiting (B - 16) ions. Further, the (B + 2) ion is difficult to arrive at mechanistically.

I

PathwayB

I

\

/+ -

-H -NYH

/QTFxQ\H I

(8-16)ion

Scheme 1. Proposed mechanism to account for (B - 16) ion formation

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

26 1

~

Table 1. Extent of (B - 16) ion formation as a function of sidechain length or nucleophilicity

- 16)

0 P-G-LYS-G-D-F-OH

769

417

--(CH2)4-NH2

7

5.0

3.5

+D-C-Orn-G-D-F-OH

755

403

-(CHZ),-NH,

6

9.4

2.0

Slze

4

&D-G-Gln-CDF-OH

769

417

-(CH~)Z--CO-NH~

6

(SIN)

(y)

MW

(B4

Sidcchain

Ring

Sequence

51

1.9

0.4

0.4

Abbreviations: lysine (Lys), ornithine (Orn), diaminobutyric acid (Dab), diarninopropionic acid (Dap), glutamine (Gln). For S/N estimation, N taken as the average of 20 adjacent, nonanalyte background peaks.

(B - 16) ion formation by the direct loss of an oxygen atom was also dismissed for mechanistic as well as thermodynamic considerations. Based on the proposed mechanism, one would predict that ring size would have an influence on the likelihood of (B-16) ion formation. To test this hypothesis, a series of hexapeptides were synthesized differing only in the length of their amino-containing sidechains (Table 1). In addition, a glutamine analogue was prepared. The predicted ring size for each of the respective (B - 16) ions is listed in Table 1. Each of the analogues was run by FAB mass spectrometry in glycerol thioglycerol at the 15 pg level. A comparison of the mass spectra obtained appears in Fig. 4 (a-e). Each mass spectrum represents an average of 5 consecutive scans taken over the region of optimum desorption of the analyte. Visual inspection of the low-mass portion of each mass spectrum enables the following order to be defined for ease of (B-16) ion formation: O m > Dab > Lys > Dap. This conclusion is supported by the signal-to-noise (S/N) ratios calculated from the individual mass spectra which are listed in Table 1. As anticipated, the extent of (B - 16) ion formation paralleled ring stability: 6-membered > 5-membered > 7membered > 4-membered. The fact that no (B - 16) ion was observed for the diaminopropionic acid (Dap) analogue is in agreement with the proposed mechanism since it would necessitate formation of a less favorable 4-membered ring. The proposed mechanism also accounts for the observed selection for lysine over glutamine in the ability to form (B - 16) ions. Although it is conceivable that glutamine could form a cyclic species analogous to the (B - 16) ion, its formation should be limited by the comparatively low nucleophilicity of the amide moiety. The FAB mass spectrum for the glutamine analogue is shown in Fig. 4(e). In this spectrum, the predicted (B - 16) ion ( m / z 417) was not readily distinguished above background (S/N 1.9). As an alternative indication of the feasibility of (B-16) ion formation, the ratio in intensity of

+

(B4- 16) to B4 also appears in Table 1. Although the value for the lysine analogue is higher than expected, this ratio again demonstrates the dependence of ring size on (B - 16) ion formation. Further, the inability of glutamine to form (B - 16) ions was confirmed as the ratio for the Gln analogue is not greater than that observed for the Dap analogue which requires formation of a 4-membered ring. While it is acknowledged that the indices of (B - 16) ion formation listed in Table 1 cannot be interpreted quantitatively, the fact that both show a similar trend supports the proposed mechanism which predicts the observed dependence of ring size on (B - 16) ion formation and explains the inability of glutamine to‘ form (B - 16) ions. An important feature of the proposed mechanism is that it invokes a radical displacement step for formation of the (B-16) ion. While the source of the hydrogen radical was not indicated in Scheme 1, a likely place is from the FAB matrix. In a recent paper by Reynolds et al. , the ability of certain matrices to scavenge radicals was dem~nstrated.’~ These authors cited the radical scavenging ability of rn-nitrobenzyl alcohol (MNBA) over other matrices such as glycerol to account for a reduction in beam-induced damage that occurs in the FAB mass spectra of dyes. The ability of MNBA to scavenge radicals was therefore used to test the proposed mechanism for (B-16) ion formation. FAB mass spectra were recorded for the ornithine analogue in Table 1 both in glycerol and MNBA. A comparison of the results appears in Fig. 5 which displays the portion of the mass spectrum surrounding the (B - 16) ion in each case. Although each spectrum contains evidence for cleavage at Lys (i.e., B4 rn/z 419), the (B4- 16) ion at mlz 403 was only readily observed in glycerol (Fig. 5(a)). In MNBA a small peak was apparent, but had a signal-to-noise ratio of less than two (Fig. 5(b)). The data shown were acquired by dissolving the solid peptide in each matrix. However, the results were reproduced using a dimethylsulfoxide (DMSO) solution of the peptide at the 20 pg level. In this latter experiment, very little fragmentation was observed using MNBA and no (B4- 16) ion was pres-

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

262

gram quantities by FAB mass spectrometry (data not shown). Although our experience in examining CID spectra for lysine-containing peptides is limited at this point, there have been a number of instances where (B - 16) ions readily seen by FAB mass spectrometry were not detected in the CID spectra recorded on the [M H]+ ion using linked scanning at constant BIE. The poor efficiency in forming (B - 16) ions under CID is attributed to two factors. The first is related to the need for hydrogen radicals to effect (B-16) ion formation as outlined by the mechanism of Scheme 1. In the ion source during FAB there exists an abundant supply of hydrogen radicals; either in the matrix itself or in the selvedge (high-pressure) region at the matrix/ vacuum interface. In contrast, (B - 16) ion formation

ent. The above data lend support to the proposed mechanism because if hydrogen radicals are scavenged more readily by MNBA, this could account for why (B - 16) ions were difficult to observe in this matrix. A key observation about (B - 16) ions is their apparent difficulty in undergoing formation under high energy CID conditions as compared to conventional FAB mass spectrometry. This finding is in contrast to the formation of other reported sidechain fragments, most notably D and W ions, which are readily detected in peptide CID spectra. The MAP peptide discussed earlier (Figs2 and 3) is an example of this behavior. Whereas larger quantities had to be applied to permit detection of the (B5- 16) and (B,]- 16) ions by CID, it is possible to detect these same ions using submicro-

+

, 4 J ' & i - O r n l G -419 D-F-OH

a

756 [M4+

c.

lBr-161

I'

'I:

403

T

1

50

C: 436

493

[M+Na]+ 778

792

350

400

450

500

550

600

650

700

750

800

SO8

I

350

400

450

500

600

550

650

700

350

800

750

400

450

500

mlz

550 d Z

600

650

700

750

800

433

d D G - G l n f G -D-F-OH

[M+H]+

751

523 490

I

25

350

400

450

500

550

600

650

700

750

800

d z

Figure 4. FAB mass spectra acquired for a series of four synthetic hexapeptides differing only in the length of their amino-containing sidechains (SCC Table I ) . For comparison. Fig. 4e contains the FAB mass spectrum of the corrcsponding glutaminc analogue.

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

263

436.4

100 C’;

75.

50. 4033 401.3

64

392 3 41 9.4

I 2 5.

433.3

436.4 B4 B4

419.3

I I I

I C,’

391.3

184-16(

I/

403.3

I 4iO

mlz Figure 5. Comparison of FAB mass spectra obtained in (a) glycerol and (b) MNBA for the ornithine-containing hexapcptidc of Fig. 4b. Only the region of the (B, - 16) ion is shown in each case.

in the first field-free region requires hydrogen radicals to be supplied through an intra-ionic hydrogen transfer during CID. This situation is far more limiting than for conventional FAB and may be the rate-limiting step to (B - 16) ion formation under CID. It should be pointed out, however, that such a process is not without precedent as demonstrated by the mechanistic work of Mueller et af. on the origin of Y” ions under CID.I4 A second factor working against the formation of (B - 16) ions is competition by other routes of fragmentation. The most notable examples are the losses of water and ammonia from B ions which are more pronounced under CID than FAB. This is especially true for lowenergy CID, as evidenced by the work of Hunt et af. who routinely use these ions to support sequence assignments for unknown peptides. Unfortunately, because the intensity of (B - 16) ions is small under CID, their detection in realistic examples of peptide sequencing may be difficult and could explain why they have not been reported previously. Although the difficulty in forming (B - 16) ions by C I D can be explained by the proposed mechanism, this finding ultimately limits the scope of application for

’‘

(B - 16) ions. Nonetheless, the data presented show that this route of fragmentation is capable of differentiating lysine and glutamine in FAB mass spectra. In this context, we have found (B-16) ions to be a convenient and routinely used aid in the interpretation of FAB mass spectra acquired for synthetic peptides.

REFERENCES I . D. H. Williams, C. V. Bradley, S. Santikarn and G . Bojescn, Biochem. J . , 201, 105 (1982). 2. R. S. Johnson, S. A. Martin and K. Bicmann. I n ( . J . Mu.ss Spectrom. Ion Processes, 86, 137 (1988). 3. R. S. Johnson, S. A. Martin, K. Bicmann, J . T. Stults and J . T. Watson. Anal. Chem., 59. 2621 (1987). 4. D. Rcnncr and G . Spitellcr, Biomed. Enoiron. Muss Spectroni., 15, 75 (1988). 5. G . C. Thornc. K. D. Ballard and S. J . Gaskcll, J . Am. Soc. Ma.ss Spectrom., 1 , 249 (1990). 6 . S. Naylor and G . Moneti. Biomed. Enoiron. Mass Spec~rom..18. 405 (1989). 7 . P. Rocpstorff and J . Fohlman, Biomed. Mass Spectrom., I I , 601 ( 1984). 8. K. Biemann. Biomed. Enoiron. Mass Spectrom., 16. 99 (1988). 9. R. S. Johnson and K. Bicmann, Biochemistry, 26. 1209 (1987).

264

EVIDENCE FOR A LYSINE-SPECIFIC FRAGMENTATION IN FAB MASS SPECTRA OF PEPTIDES

10. G. Gaede and K. L. Rinehart, Biol. Chem. Hoppe-Seyler, 371, 345 (1990). 11. G. Gabrielides, A. L. McCormack, D. F. Hunt and S. Christakos, Biochemistry, 30, 656 (1991). 12. E. R. Williams, J. J . P. Furlong and F. W. McLafferty, J . Am. SOC. Mass Specrrom. 1 , 288 (1990). 13. C. B. Lebrilla, D. T.-S. Wang, T. J . Mizoguchi and R. T. McIver, Jr., J . A m . Chem. SOC., 111, 8593 (1989). 14. S. A. Martin, J . A. Hill, C. Kittrell and K. Biemann, J . Am. SOC. Mass Spectrom., 1, 107 (1990). 15. M. E. Bier, J. C. Schwartz, K. L. Schey and R. G. Cooks, In!. J . Mass Spectrom. Ion Processes, 103, 1 (1990). 16. R. D. Smith, J. A . Loo, C. J. Barinaga, C. G. Edrnonds and H. R. Udseth, J . A m . SOC.Mass Spectrom., 1, 53 (1990). 17. R. E. Kaiser, Jr., R. G. Cooks, J. E. P. Syka and G. C. Stafford, Jr., Rapid Commun. Mass Spectrom., 4, 30 (1990).

18. D. F. Hunt, J. Shabanowitz, J. R. Yates, P. R. Griffin and N. 2. Zhu, in Mass Spectrometry of Biological Materials, ed. by C. N. McEwen and B. S. Larsen, p. 169. Marcel Dekker, New York (1990). 19. W. Kausler, K. Schneider and G. Spiteller, Biomed. Enuiron. Mass Spectrom., 17, 15 (1988). 20. J . C. Nutkins and D. H. Williams, J . Chem. SOC., Chem. Commun., 825 (1990). 21. A. Malorni, G. Marino and A. Milone, Biomed. Enuiron. Mass Spectrom., 13, 477 (1986). 22. P. Roepstorff, P. Hojrup and J. Moiler, Biomed. Mass Spectrom., 12, 181 (1985). 23. J. D. Reynolds and K. D. Cook, J . Am. SOC. Mass Spectrom., 1 , 149 (1990). 24. D. R. Mueller, M. Eckersley and W. J. Richter, Org. Mass Speclrom., 23, 217 (1988).