Unusual Fragmentation of β-Linked Peptides by ExD Tandem Mass Spectrometry
- First Online:
Ion-electron reaction based fragmentation methods (ExD) in tandem mass spectrometry (MS), such as electron capture dissociation (ECD) and electron transfer dissociation (ETD) represent a powerful tool for biological analysis. ExD methods have been used to differentiate the presence of the isoaspartate (isoAsp) from the aspartate (Asp) in peptides and proteins. IsoAsp is a β3-type amino acid that has an additional methylene group in the backbone, forming a Cα–Cβ bond within the polypeptide chain. Cleavage of this bond provides specific fragments that allow differentiation of the isomers. The presence of a Cα–Cβ bond within the backbone is unique to β-amino acids, suggesting a similar application of ExD toward the analysis of peptides containing other β-type amino acids. In the current study, ECD and ETD analysis of several β-amino acid containing peptides was performed. It was found that N–Cβ and Cα–Cβ bond cleavages were rare, providing few c and z type fragments, which was attributed to the instability of the Cβ radical. Instead, the electron capture resulted primarily in the formation of a and y fragments, representing an alternative fragmentation pathway, likely initiated by the electron capture at a backbone amide nitrogen protonation site within the β amino acid residues.
Key wordsIon molecule reactions Odd electron fragmentation Electron capture dissociation Electron transfer dissociation
Electron capture dissociation (ECD) [1, 2, 3] and other ion-electron reaction methods such as electron ionization/impact dissociation (EID) , and electron transfer dissociation (ETD)  (collectively known as ExD) belong to a group of odd-electron (OE) fragmentation techniques, which also include ultraviolet photodissociation (UVPD) , and metastable atom-activated dissociation (MAD) . These methods can generate fragments often unobservable in conventional slow-heating fragmentation techniques such as low energy collisionally activated dissociation (CAD)  and infrared multiphoton dissociation (IRMPD) . ECD and ETD are perhaps the most widely used among these OE fragmentation methods. In ECD of polypeptides, a low energy electron is captured by the multiply charged molecular ion either at a protonated backbone carbonyl site according to the Cornell mechanism [1, 10], or in a π orbital of an amide group in the presence of a remote charge according to the Utah-Washington mechanism , producing a charge reduced cation radical, with the subsequent backbone N–Cα bond cleavage leading to the formation of c and z• fragments. The radical on z• can induce further rearrangements within the molecule, a free radical cascade , producing additional backbone and side chain cleavages both in the proximity of and remote from the initial radical site . The fragmentation pattern generated by ECD may vary dramatically depending on the number and location of certain amino acids, modifications, and charges [14, 15, 16, 17, 18]. A number of experimental and computational studies have been carried out since ECD was first introduced [1,19], yet the mechanisms remain under discussion, probably because multiple competing pathways are involved [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]. Nevertheless, the capability of ECD to produce unique fragment ions not obtainable by conventional methods has led to a rapid optimization of ECD and the development of other ion-electron reaction based tandem MS techniques [4, 5, 21]. ExD has been broadly applied towards the structural analysis of various types of biomolecules, including proteins, oligosaccharides, oligonucleotides, and others [32, 33, 34, 35]. Further, ECD and ETD were found to be particularly useful for the characterization of post translational modifications (PTMs) in proteins as they cleave the backbone while preserving the labile groups and noncovalent interactions upon fragmentation [2, 36].
Deamidation of asparagine (Asn) and glutamine (Gln) residues and, similarly, isomerization of aspartic (Asp) and glutamic (Glu) acid residues, are two of the most common PTMs found in all proteins that have been studied recently. These PTMs accumulate with age in long lived proteins and are frequently associated with age related diseases, such as amyloid diseases and cataract formation in the eyes [37, 38, 39]. Both reactions are nonenzymatic and proceed spontaneously in physiologic conditions via formation of a succinimide or glutarimide ring intermediate, followed by rapid hydrolysis. Deamidation introduces an approximately 1 Da mass shift to the protein molecular mass (+0.984 Da) and can be easily identified using mass spectrometry (MS) . Isomerization does not change the molecular mass of the protein and thus cannot be identified as easily. However, the application of tandem MS methods, in particular, ExD techniques, has shown successful results: isomers of aspartic acid can be unambiguously identified using ECD, ETD, and EID methods [37, 41, 42, 43, 44, 45, 46, 47, 48]; likewise, identification of γ-glutamic acid by ECD seems to be promising .
Isoaspartic acid is a β-type amino acid that has one extra CH2 group in the polypeptide backbone, and one fewer on the side chain compared to Asp. ExD of isoAsp containing peptides generates additional signature fragment ions (c + 57 and z• −57) at the positions of the isomerized residues, allowing differentiation from the non-modified residues (the fragmentation scheme can be found in Supplemental Material, Figure S1.). These ions are formed by the Cα–Cβ backbone bond cleavage. In peptides consisting solely of α-amino acid residues, there is no such bond within the backbone. Fragments produced upon Cα–Cβ bond rupture are unique to β-amino acid residues and could be used to locate the position of the β-amino acid if observed. The result from the isoAsp experiment suggested a possible extension of the method to characterize peptides containing other β-type amino acids [50, 51, 52].
β-Peptides have been a subject of intense studies that investigated their structural, biological properties, and their interactions involved in peptide folding. β-Peptides were found to have richer conformational energy surface with more stable secondary structures . They also fold into helices or hairpin-type structures with larger variety than α-peptide secondary structures. β-Peptides are very stable against proteolytic degradation and other enzymes in human and various living organisms . These features provide great potential for β-peptides in biomedical application as proteolytically stable therapeutics. A fast and accurate MS based method would be of a great utility for analysis of β-peptide structure. In this study, the potential of the ExD based tandem MS methods to differentiate β-amino acid containing peptides from their α-analogues, as well as β2 from β3-type amino acids was investigated. In general, our results are in good agreement with the findings of recent ECD/ETD studies of small β-peptides  and ε-peptides . In the present study, in addition to simple model peptides such as Q06 and Substance P, a more complicated system of Puma BH3 peptide analogues of 26 amino residues was investigated. Furthermore, charge state dependence of the fragment appearance was studied. An alternative mechanism of ion-electron reaction induced dissociation of peptides within β-amino acid residues is discussed.
The Q06 β-peptide (β2Vβ2Aβ2Lβ3Vβ3Aβ3L) was kindly provided by Professor D. Seebach, and Dr. J. Gardiner, ETH Zurich, Switzerland. C-terminally amidated Substance P with two amino acids modified to β3-type amino acids (RPKPβQQFFGβLM) was custom synthesized by AnaSpec (San Jose, CA, USA). Puma BH3 pro-apoptotic protein analogue I (βEEQβWAREβIGAβQLRRβMADβDLNAβQYEβRR) and analogue II (βEEQWβAREβIGAQβLRRβMADDβLNAβQYERβR) were kindly provided by the group of Professor S. Gellman at the University of Wisconsin (Madison, WI, USA), where β indicates β3-type amino acids. Other reagents: non-modified Substance P, (2-aminoethyl)-trimethylammonium chloride hydrochloride (cholamine), triethylamine (TEA), and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (St. Louis, MO, USA); 2-(1 H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate (HBTU) from Novabiochem (La Jolla, CA, USA); and 1-hydroxybenzotriazole (HOBt) from AK Scientific, Inc. (Palo Alto, CA, USA).
2.2 Cholamine Reaction
Covalent attachment of cholamine to the Q06 β-peptide was performed as previously described . Briefly, 0.1 μmol of Q06 β-peptide was treated sequentially with 10 μL of 200 mM HOBt in DMSO, 200 μL of 100 mM cholamine in DMSO containing 200 mM TEA, and 10 μL of 200 mM HBTU in DMSO. The sample was left to react overnight at room temperature, and purified using a ZipTip C18 solid phase micropipette extraction column before MS analysis.
2.3 Mass Spectrometry
Most ECD experiments were performed on a custom built qQq-FTICR MS with a nanospray source and a 7 T actively shielded magnet [60, 61]. Samples were nanosprayed (50 nL/min, room temperature) at 5 μM concentration in 50:50 MeOH:H2O with 1% formic acid. Ions were isolated in the first quadrupole Q1, accumulated in the second quadrupole Q2, and transmitted into the ICR cell where they were irradiated with electrons emitted from an indirectly heated dispenser cathode (Heatwave; Watsonville, CA, USA) for ion fragmentation. The following ECD and EID parameters were employed: electron irradiation time 35–100 ms, cathode potential −0.2 to 1.2 V (ECD), −18 V (EID). Acquired spectra were zerofilled twice, internally calibrated, and analyzed manually using BUDA (Boston University Data Analysis, ver. 1.4, © 2000 by Peter B. O’Connor). ECD experiments of triply charged Substance P ions were performed on solariX FTICR instrument (Bruker Daltonics, Billerica, MA, USA) with 12 T actively shielded magnet. Electrospray was applied for enhanced production of triply charged ions.
ETD spectra with supplemental activation were acquired on an amaZon Ion Trap instrument (Bruker Daltonics, Billerica, MA, USA) using fluoranthene as the ETD reagent. Peptides were electrosprayed (2 μL/min, glass capillary temperature 220 °C) at 1 μM concentration in 50:50 MeOH:H2O with 1% formic acid using an Apollo II ion source. Data acquired on solariX and amaZon were analyzed using Bruker’s ESI Compass DataAnalysis 4.0 software.
3 Results and Discussion
3.1 Q06 β Peptide β2Vβ2Aβ2Lβ3Vβ3Aβ3L
Although the Q06 β-peptide only contains six β-amino acid residues, it can form helical secondary structures [55, 62]. This small and relatively simple peptide contains both β2 and β3 type amino acids, making it potentially an ideal system to study for the differentiation of β2 and β3 amino acids. However, only singly charged ions [M + H]+ were detected in all ESI/ECD experiments, either with the nanospray or with the electrospray ionization sources (Supplemental Figure S2a), which is likely due to the lack of basic amino acid residues within the peptide sequence. Since ECD and ETD are accompanied by charge neutralization upon the electron capture or transfer, they can not be performed on singly charged ions as the products would be neutral and undetectable. In this case, other fragmentation techniques could be applied that do produce fragments from singly charged ions such as CAD  or IRMPD . Expectedly, IRMPD of the Q06 β peptide resulted in b and y fragments with no information on the position of the modifications (Supplemental Figure S2b) . Additionally, electron ionization/impact dissociation (EID) can generate ECD type fragments from singly charged molecular ions , as well as Cα-Cβ cleavage for isoaspartic acid . However, only b and y fragments were detected in EID mass spectra of the Q06 β-peptide (Supplemental Figure S2c) . The lack of c and z• fragments could be due to the low fragmentation efficiency of EID in this particular experiment. Thus, the well established ECD method would seem to be a better approach, but it requires an increase in the number of charges on the peptide.
3.2 Charge Increase
Similar experiments of C-terminal charge tag attachment to α-peptides were done by Hunt’s group . Their result demonstrated increased formation of z ions. In the current study, no z fragments and only one ECD-type fragment, c5, was observed for the Q06 β-peptide. The electron capture at the quaternary ammonium provides an abundant loss of the neutral trimethylamine 59.0735 = N(CH3)3, leaving the radical on the C-terminal methylene group (Figure 1). This is expected to further induce radical initiated reactions and cleavages, such as the loss of ethylene. Although charge neutralization on the quaternary ammonium may not lead to backbone cleavages , electron capture at the protonated N-terminal amine of the doubly charged Q06 peptide should result in backbone fragmentations. Further, the recombination energy of the protonated N-terminal amino group is higher than that of the quaternary ammonium as calculated for the singly charged model cations by Jensen et al. [4.3 eV (MeNH3+) versus 3.1 eV (NMe4+)] , making it the preferred electron capture site. However, ECD type fragments were lacking in the ECD spectrum of the tagged Q06 peptide, which was dominated by b5and y fragments which could be merely the result of the residual vibrational excitation . Alternatively, the formation of y fragments could be facilitated by the backbone nitrogen protonation or a rearrangement that transfers a hydrogen atom to the backbone nitrogen [2, 15, 58]. Upon charge neutralization at the tag site and the subsequent tag loss, the excessive vibrational energy could induce the formation of the b5 ion, similar to that suggested by Cooper et al. in their study of b-ion formation in ECD . Nevertheless, it is hard to make definitive conclusions why this unusual fragmentation pattern was observed as two variables were introduced at the same time in the studied system: a cholamine tag and the incorporation of β-amino acids.
3.3 ETD of Q06 β Peptide
Electron transfer dissociation was further used to analyze the Q06 β peptide. As opposed to the first ESI experiment, a small peak corresponding to the doubly charged molecular ions, [MQ06 + 2H]2+, was observed, which could be due to the difference in the ionization sources between two instruments. In particular, the glass capillary in the electrospray source used in the ETD study was heated to 220 °C which could possibly increase the desolvation efficiency and production of higher charged ions compared to the unheated source used in the previous ECD experiment.
3.4 Modified Substance P
3.5 Puma BH3 Protein Analogues
As was noted earlier, the Puma BH3 peptides studied here have α-helix like conformations. It is possible that the N–Cβ bonds can be cleaved but protected from dissociation due to strong hydrogen bonding within the α-helix as was suggested previously  and consistent with the results from ECD mechanistic studies [24, 26]. Yet, there were many c and z• fragments present except for those missing within β-residues suggesting that the peptide was relatively unfolded, and thus, the hydrogen bonds should not interfere with the fragment separation. Furthermore, dissociation of the Cα–Cβ bond now seems to be an exception rather than the rule for β-amino acid residues as only one peak representing the Cα–Cβ cleavage of isoleucine was identified provided by the z193+•-C5H10 fragment ion (Figure 4). Indeed, according to Turecek and coworkers  who studied the β-alanine N-methyl amide model system, dissociation of the Cα–Cβ bond was slow and in competition with other dissociations from the most stable composition with a radical located on the C-terminal amide carbonyl. To conclude, the N-Cβ and Cα–Cβ backbone cleavages are still possible, but do not represent the dominant channel of fragmentation within β-amino acid residues.
3.6 Proposed Mechanism
The role of backbone nitrogen protonation in ECD has been previously discussed in the literature. In the original ECD study, the a and y ions were proposed to be formed from the backbone nitrogen protonated species . Backbone nitrogen protonation was also suggested to play a role in the formation of b ions in ECD. Theoretical investigation by the Uggerud group showed that electron capture by nitrogen protonated N-methyl-acetamide resulted in rapid amide bond dissociation and production of CH3CO and NH2CH2, corresponding to the b and y ions in peptides . Concomitantly, b ions can further lose CO to form a ions. In addition, b ions were present in ECD of peptides without basic amino acid residues . It was suggested that upon backbone nitrogen protonation, the peptide bond could be cleaved to form a b/y ion pair, and the subsequent intra-complex hydrogen atom transfer within this long-lived ion pair could lead to the formation of b and y• ions. Backbone amide nitrogen protonation was also invoked to explain the a, b, and b ion formation in nitrated peptides that were either acetylated at the N-terminus, or lacking basic amino acid residues . In the current study, however, neither b nor y ions observed were radicals. Thus, these b ions were most likely formed via the energetic fragmentation of vibrationally excited even-electron charged reduced species .
It should be noted that protonation on the backbone nitrogen is not thermodynamically favorable [75, 76, 77]. For instance, for the model system of N-methylacetamide, the proton affinity of the carbonyl oxygen was calculated to be ~60 kJ/mol higher than that of the backbone nitrogen . This is in agreement with the study of the dipeptide Lys-Gly, where the carbonyl oxygen protonated species was calculated to be ~45 kJ/mol more stable than the backbone nitrogen protonated species . Based on these results, peptide fragmentation via protonation of the backbone nitrogen would seem unlikely. However, theoretical investigations have usually been done on very small model systems, which often do not possess extensive intramolecular interactions such as hydrogen bonding and salt bridges that are expected to play a more important role in the fragmentation of larger peptides. Charge solvation by nearby backbone and side chain groups could appreciably change the relative stabilities of different protonated species. In β-peptides, the elongation of the backbone may make it better positioned (due to less steric hindrance by the adjacent side chain) for hydrogen bond formation. The addition of an extra methylene group in the backbone can also slightly increases the gas-phase basicity of the amide nitrogen, as the electron withdrawing carbonyl group is replaced by the electron donating alkyl group. Further, electron capture by the precursor ion could increase its internal energy considerably (by several eVs), and less favored protonation sites may become significantly populated as the system relaxes from the initial Rydberg state to low lying electronic states. In other words, the proton may initially reside on the carbonyl oxygen, but could migrate to the backbone amide nitrogen upon excitation. This argument is similar to the one used in the mobile proton model to explain the low energy CAD fragmentation behavior of peptide ions [75, 76, 77, 78], where it was proposed that the proton initially resides on the thermodynamically more stable sites, such as the lysine side chain or backbone oxygen, and later migrates to the less favorable sites, including the backbone amide nitrogen upon collisional activation to facilitate fragmentations. Finally, for the β-linked peptides studied here, electron capture at the protonated oxygen site cannot lead to “normal” c/z fragmentation due to the radical instability, which may further drive the migration of the proton to the less preferable backbone nitrogen site that could lead to the formation of a• and y ions upon electron capture. In general, the abundance of these unusual ECD fragments was fairly low, as expected from the low population of nitrogen protonated species (Fig. 5), but they were nonetheless present in competitive abundance when the primary ECD fragmentation channel was blocked within β-amino acid residues.
Further experiments are needed to test the proposed mechanism. In addition, basicity measurements and theoretical investigations specifically for β-amino acid containing peptides could provide a better understanding on how these a and y type fragments are formed.
Various peptides containing β-amino acid residues were analyzed in this study. Remarkably, N–Cβ bond cleavages were rare within the β-residues, and Cα–Cβ cleavages were seldom observed providing no evidence of the β-residues in spite of previous results of the isoaspartic acid. Furthermore, no distinct difference was found for the fragmentation within β2 versus β3-type amino acid residues. Meanwhile, a and y fragments were often produced at β-residues, particularly for the bigger peptides with α-helical like structures. The lack of z and c fragments and increased a and y fragment formation could imply the presence of β-residues in the peptide; however, this is a poor signature, because of the normal appearance of a and y fragments in ExD spectra of α-peptides, and various other reasons that can contribute to the disappearance of z and c fragments. Thus, currently, ExD methods cannot be used to reliably differentiate α- from β- or β2 from β3 type amino acids.
The introduction of one extra methylene group into the polypeptide chain destabilizes the Cβ radical formed by the N–Cβ bond rupture making this channel of fragmentation less favorable. Thus, the fragmentation occurs via alternative channels. The dominant products appear to be the a and y fragments, with the exception when the side chain of the β-residue can provide radical stabilization for the formation of the z and c fragments. It is suggested that appearance of such fragments may require protonation on the backbone amide nitrogen, which is further supported by the charge state-dependent study of modified Substance P peptide. The fragmentation mechanism for β-peptides has been proposed via backbone nitrogen protonation similarly to what was originally proposed for the ECD of α-peptides.
The minor ECD pathway of a and y fragment formations was little studied. Future studies of this fragmentation pathway would help with our understanding of the fragmentation of the β-amino acid containing peptides. Further ExD studies as well as computational studies and kinetic analysis specifically for the β-amino acids are needed for better characterization of β-peptides.
The authors acknowledge Sam Gellman, Lisa Johnson, Dieter Seebach, and James Gardiner for providing the samples of β-peptides for this study. They thank Brian Fray and Lloyd Smith for their cholamine reaction protocol. Catherine Costello is appreciated for the opportunity to use the AmaZon and SolariX instruments, Liang Han and Sandrine Bourgoin-Voillard for help with instrument use. This work was supported by the Warwick University Chemistry Department and Warwick Center for Analytical Science (EPSRC funded EP/F034210/1), and NIH/NCRR-P41 RR10888, NIH/NHLBI-N01HV28178, NIH/NIGMS-R01GM078293, S10 RR025082 grants.