Electron Transfer Dissociation (ETD) of Peptides Containing Intrachain Disulfide Bonds
The fragmentation chemistry of peptides containing intrachain disulfide bonds was investigated under electron transfer dissociation (ETD) conditions. Fragments within the cyclic region of the peptide backbone due to intrachain disulfide bond formation were observed, including: c (odd electron), z (even electron), c-33 Da, z + 33 Da, c + 32 Da, and z–32 Da types of ions. The presence of these ions indicated cleavages both at the disulfide bond and the N–Cα backbone from a single electron transfer event. Mechanistic studies supported a mechanism whereby the N–Cα bond was cleaved first, and radical-driven reactions caused cleavage at either an S–S bond or an S–C bond within cysteinyl residues. Direct ETD at the disulfide linkage was also observed, correlating with signature loss of 33 Da (SH) from the charge-reduced peptide ions. Initial ETD cleavage at the disulfide bond was found to be promoted amongst peptides ions of lower charge states, while backbone fragmentation was more abundant for higher charge states. The capability of inducing both backbone and disulfide bond cleavages from ETD could be particularly useful for sequencing peptides containing intact intrachain disulfide bonds. ETD of the 13 peptides studied herein all showed substantial sequence coverage, accounting for 75%–100% of possible backbone fragmentation.
Key wordsDisulfide bond Electron transfer dissociation Post-translational modification Peptide sequencing Mass spectrometry
Disulfide bond formation is a frequently occurring post-translational modification (PTM), critical to protein folding and structural stability . Characterization of proteins containing disulfide bonds remains challenging for tandem mass spectrometry based on collision-induced dissociation (CID). The difficulty is caused by the fact that disulfide bonds rarely dissociate in the presence of mobile protons . Therefore, two amide bonds must be cleaved to generate one sequence fragment. Due to the random nature of cleaving two amide bonds, CID spectra of disulfide-linked peptides are typically difficult to interpret and provide limited information for the backbone region within the disulfide loop. By far, the most widely-used approach for characterizing disulfide-containing proteins involves multi-enzyme digestion and disulfide bond reduction/alkylation prior to mass spectrometric analysis . With the disulfide bond already cleaved, rich sequence information can often be obtained from CID of disulfide-reduced peptides. Sophisticated strategies, including partial disulfide bond reduction or chemical modification combined with tandem mass spectrometry, are needed in order to elucidate disulfide connecting patterns . Generally, the whole process of identification and characterization of proteins containing disulfide bonds is time and sample consuming.
Forgoing solution reduction allows the opportunity of characterizing the connecting patterns of disulfide bonds. Many efforts have been directed toward developing gas-phase dissociation methods to characterize intact disulfide bonds within peptides or proteins. Preferential cleavage of disulfide bonds over peptide backbone bonds under CID conditions has been reported for deprotonated peptides [5, 6] and metal cationized peptides [2, 7–9]. Unique fragments associated with disulfide bond cleavages have been used to identify the presence of disulfide bonds within unknown proteins [9, 10]. Disulfide bonds can be selectively excited and cleaved by photons at 157 nm  and at 266 nm . When collisional activation is applied after photon dissociation, backbone fragmentation takes place, providing rich sequence information. Electron-based dissociation techniques, such as electron capture dissociation (ECD) [13, 14] and electron detachment dissociation (EDD) , have shown preferential cleavage of disulfide bonds over backbone (N–Cα) fragmentation. Electron transfer dissociation (ETD) is an analogue of ECD, conducted on electron dynamic ion trap instruments within the frame of ion/ion reactions . The dissociation chemistry of ETD is similar to ECD in many aspects, including competitive dissociation at disulfide bonds within peptide or protein ions [17, 18]. Recent reports have demonstrated that rich structural information can be obtained by using a multiple fragmentation approach involving ETD and CID, allowing for determination of complicated disulfide linkage patterns within recombinant therapeutic proteins [19, 20].
Given the increasing use of ECD and ETD for studying peptides or proteins containing disulfide bonds, it is important to understand the underlying principles. Several mechanisms have been put forward to account for disulfide bond cleavage in ECD or ETD. The “Coulomb assisted dissociation model” suggests the possibility of direct electron attachment at a disulfide bond, assisted by positive charges in close proximity [21, 22]. Alternatively, an electron can be first attached to a charged site and then transferred to the disulfide bond, either through-space or through-bond, to induce cleavage . For these mechanisms, a subsequent proton transfer is needed to form a thiyl radical (−S•) and a sulfhydryl (−SH) group at the cleavage site. Peptides containing interchain disulfide bonds have been more frequently studied due to the ease of detecting disulfide bond cleavage. So far, very limited experimental data have been collected on ECD or ETD of peptides containing intrachain disulfide bonds [24, 25].
Peptide and protein samples were purchased from AnaSpec (San Jose, CA,, USA) and Sigma-Aldrich (St. Louis, MO, USA) and used without purification. Table 1 lists each peptide with its name and single letter sequence. Working solutions of each peptide were prepared to a final concentration of 10 μM in 50/49/1 (vol/vol/vol) methanol/water/acetic acid solution for positive mode nanoelectrospray ionization (nanoESI), using a home-built source. All mass spectra were collected on a Velos LTQ mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) equipped with ETD capability. The lens conditions of the instrument were optimized for peptide ion and ETD reagent signals. ETD reaction time was around 100 ms. Subsequent CID was performed on selected product ions. Supplementary activation during ETD experiments was also employed to dissociate ~90 % of the first generation odd-electron charge reduced peptide ions, with values ranging from 15–25 (arbitrary units). Data shown in this study were typically an average of 50 scans.
3 Results and Discussion
3.1 General ETD Phenomena
3.2 Formation of c, z Ions
CID data of the doubly-charged c7 (m/z 433.2) of peptide 3 are shown in Figure 2b. The detection of neutral losses, such as −33 Da (SH) and −46 Da (CH2S), are highly suggestive of the presence of a thiyl radical. Several b and y ions are observed, indicating a typical open-chain structure. Interestingly, the y ions all have masses 1 Da higher than expected (based on the c ion structure proposed in Scheme 1), in concert with their complementary b ions forming with masses 1 Da lower than expected (y6 + 1/b1-1, y5 + 1/b2-1). This observation indicates hydrogen transfer from the first two amino acid residues to the thiyl radical. In fact, experimental and theoretical studies have shown that intramolecular hydrogen transfer to the thiyl radical is a facile process within polypeptides and cysteine ions [30–33].
CID of the c7 + 1 (m/z 856.4, sequence: CLPTRHM) of peptide 2 (Figure 2c) clearly suggests an open-chain structure, with b and y ions appearing exactly at their predicted masses. The c + 1 ions are likely formed via the addition of a hydrogen to the thiyl radical at the cysteine residue, rather than at any other place along the sequence. Figure 2c also shows a high abundance of fragments for the loss of 17 Da (NH3) from parent ions, as well as from b/y fragments to form b * /y * ions. The abundant ammonia loss is generally observed from CID of c ions derived from ETD of linear peptides  or CID of N-terminus amidated peptides .
An alternative pathway of forming c and z ions within the cyclic region begins with an initial cleavage at the disulfide bond, followed by N–Cα bond dissociation. If ETD cleaves the disulfide bond first, a thiyl radical (−S•) and sulfhydryl group (−SH) should be formed at the cleavage site . An α-carbon radical can be generated from radical migration , the rearrangement of which along the peptide backbone can induce N–Cα cleavage as observed in ECD of circular peptide systems . If these processes were at work in this study, complementary z + 1 and c – 1 ions would be formed. The fact that no z + 1 ions or c −1 ions were observed in ETD spectra for any of the peptides examined in this study suggests that this mechanism is not likely to be a dominant process. In addition, collisional activation of peptide thiyl radical ions showed that the radical-initiated backbone cleavage (forming c – 1 and z + 1 ions) was sensitive to amino acid composition and only counted for a small fraction among a variety of fragmentation channels . These results further suggest that even if N–Cα bond cleavage were to occur after disulfide bond cleavage, its contribution to the observed backbone fragments should be quite small.
3.3 Formation of c – 33/z + 33 and c + 32/z—32 Ion Pairs
With the observation of these abundant losses of 65 Da in mind, it may be predicted that a competitive mechanism for the formation of z – 32 and c – 33 ions also exists. This mechanism involves a perthiyl radical (−SS∙) abstracting the α-carbon hydrogen within its own cysteine residue, resulting in a β-cleavage that liberates ∙SSH, leaving the remaining piece modified to dehydroalanine. However, the formation of z + 33 and c + 32 ions can only be resulted from initial backbone cleavage followed by fragmentation at the C–S bond. The mechanisms proposed in Schemes 1 and 2 share a common feature, in that the N–Cα bond is the initial cleavage site in ETD and the subsequent radical-driven reactions cause either S–S or C–S bond dissociation within cysteinyl residues. This pattern of events would yield all the ions present in high abundance in ETD spectra, accounting for the observation that N–Cα and disulfide bond cleavage can arise from a single electron transfer event.
3.4 Neutral Loss of 33 Da
Neutral loss of 33 Da (•SH) from charge-reduced parent ions is widely observed for all peptides studied, although its relative abundance varies. The most probable pathway accounting for this loss involves cleavage first at the disulfide bond, forming a transient thiyl radical (−S•), and subsequent elimination of •SH to form dehydroalanine. The 33 Da loss ions should have an open chain structure sharing the same sequence as the original peptide ions, with one cysteine residue being modified to dehydroalanine and the other to a sulfhydryl group. The CID data of ([M + 2H]•+ – 33 Da) derived from ETD of doubly-protonated peptide 1 is shown in the Supporting Information (Figure S-1) and the observed fragments are consistent with the predicted ion structure. Unlike the c/z, c – 33/z + 33 and c + 32/z – 32 ion pairs, which are formed from initial cleavage at N–Cα bond followed by radical driven reactions, the 33 Da loss should only be resulted from direct ETD at the disulfide bond. Therefore, the relative intensity of this ion can be used to monitor the competition between disulfide bond cleavage and backbone N–Cα cleavage in ETD of peptides containing intrachain disulfide bonds, as evidenced below in the “Effect of Charge State” section.
3.5 Charge-Reduced Peptide Ions
Charge-reduced peptide ions are commonly observed in ETD experiments, including both even-electron and odd-electron species. The charge-reduced, even-electron ions ([M + (n-1)H](n-1)+) are proposed to be formed either as a consequence of ejecting a hydrogen atom  or by competitive proton transfer reactions . The radical peptide ions ([M + nH](n-1)•+), also referred to as electron transfer without dissociation (ETnoD) product, can be complexes consisting of non-covalently bonded c/z fragments  or intact peptide ions with significantly weakened N–Cα bonds after electron transfer . The ETnoD ions were abundantly observed for most of the peptides containing intrachain disulfide bonds herein. Note that due to the cyclic nature of the peptides, these ETnoD ions should still be in one piece after ETD cleavage at either the disulfide bond or an N–Cα bond at the peptide backbone. The different structural isomers of the ETnoD products may explain the observation of several different fragmentation behaviors, as discussed below.
3.6 Effect of Charge State
3.7 Implications to Peptide Sequencing
A common feature shared among all the peptides investigated in this study is that the entire or the majority of the peptide backbone is confined within the ring structure, due to the formation of an intrachain disulfide bond. Obtaining sequence information for this type of peptide is very challenging with the conventional CID method. This is because two bonds need to be cleaved to form one sequence ion, which is not always accessible from low energy CID conditions. ETD, as discussed above, can induce both N–Cα backbone cleavage and disulfide bond cleavage in a single electron transfer event, allowing the observation of sequence ions within the cyclic region. Table S-1 in the Supporting Information summarizes the observed major fragments resulting from ETD for 13 peptides containing intrachain disulfide bonds. The sites of fragmentation are indicated along the peptide sequence, based on the observed c/z ions. Rich sequence information can be obtained from all peptides. The extent of cleavage corresponds to 75%–100% of possible backbone fragmentation, which is highly desirable from the perspective of peptide sequencing. The detected c – 33, z + 33, c + 32, and z – 32 ions are listed in separate columns in Table S-1. These types of ions were observed in many cases and were typically more prevalent in ETD of lower charge states of peptide ions. Including these types of ions in the generation of in-silico peptide fragmentation should be beneficial to the identification of disulfide-linked peptides. In addition, the presence of these ions is unique to fragments formed within the cyclic backbone region confined by a disulfide bond, which can be potentially used to confirm the location of the disulfide bond.
Electron transfer dissociation (ETD) was shown to induce both disulfide bond cleavage and backbone fragmentation from a single electron transfer event, allowing for sequence information to be obtained from a region of peptide backbone cyclized by intrachain disulfide bonds. In addition to the c/z ions typically observed in ETD, new types of fragments corresponding to complementary ion pairs of c – 33/z + 33 and c + 32/z – 32 were identified. Mechanistic studies suggested that the formation of these ions involved initial ETD at N–Cα backbone followed by radical-driven reactions at cysteinyl residues. A competing fragmentation channel associated with ETD at the disulfide bond was detected, with a signature loss of 33 Da (SH) from the charge-reduced peptide ions. Similar to reports on ETD of linear peptides, reduced fragmentation was observed for ETD of the lower charge states of peptide ions. However, for the 13 peptides investigated herein, extensive backbone fragmentation was typically observed from ETD, suggesting that ETD could be very useful in providing sequence information for peptides containing intact intrachain disulfide bonds. Given the frequent appearance of c – 33/z + 33 and c + 32/z – 32 ion pairs in ETD, including these ions in sequencing should allow enhanced identification of proteins containing intrachain disulfide bonds.
S.R.C. thanks the support from the Purdue Ross Fellowship. X.M. acknowledges the financial aid by the China Scholarship Council (CSC) to support his research at Purdue University. The authors also acknowledge Professor Mary Wirth for use of the LTQ Velos ETD mass spectrometer.