1 Introduction

A standard procedure for tandem mass spectrometry (MS/MS)-based peptide identification is matching experimental data obtained by collision-induced dissociation (CID) to theoretical spectra using bioinformatics software. The software is based on genomic information and assumes even cleavages occurring along the peptide backbone [1–3]. B and y ions generated by cleavage of amide bonds in protonated peptides, as well as a ions formed by loss of CO from b ions, are typical fragments. Many factors affect peptide fragmentation patterns: peptide sequence, presence and position of particular amino acids, charge, and even instrumental parameter settings [4–11]. In fact, some proteins in biological samples undergo post-translational modifications (PTMs), producing characteristic product ions [12–15]. Therefore, a clear understanding of the dissociation of peptide ions in gas phase is necessary; this will benefit de-novo sequencing and unambiguous protein identification in proteomics.

After great effort by the worldwide gas-phase ion chemistry community, the ‘mobile proton’ and ‘pathways in competition’ models have been established, which can explain the cleavages observed in protonated peptides [4–11]. Furthermore, fine details of structure, mechanism of formation, and activity of various fragment ions (such as b and a ions) have been extensively studied [7, 16, 17]. It is generally accepted that most common b ions are formed by nucleophilic attack of carbonyl oxygen on the adjacent amide carbon, generating an oxazolone structure in the C-terminus. However, amino acid side chains can competitively attack the same carbon to form isomeric b ions. For example, attack by the aspartic acid (Asp) side chain carboxylate forms a b ion with an anhydride structure; this ion loses CO2 and CO (72 Da) upon further CID fragmentation [18, 19].

Many studies have focused on cleavage of tryptic peptides attributable to the numerous applications of trypsin in proteomics [2–5]. Tryptic peptides contain an arginine (R) or lysine (K) residue in their C-terminus. However, peptides with an R residue at the N-terminus can also be produced via non-tryptic or tryptic cleavages from proteins containing KR, RR, and RP sequences [20, 21]. Compared with tryptic peptides, less attention is paid to the fragmentation pathways of these non-tryptic peptides. Recently, Paizs and co-workers demonstrated that the pathways leading to b2 and b2 + H2O fragments of the peptide RGD were different from those of common tryptic peptides, involving a salt-bridge stabilized configuration and an anhydride intermediate attributable to the R residue in the N-terminus of RGD [22].

In this work, an N-terminal deuterohemin-containing hexapeptide, DhHP-6, was chosen as a model N-terminal charged peptide. The structure is shown in Figure 1a. This peptide was initially designed as a short peptide mimetic of cytochrome c (Cyt c) and shows biological activities similar to microperoxidase-11, generated from hydrolytic cleavage of Cyt c [23]. The iron oxidation state has been determined unequivocally as Fe(III). This peptide can be singly charged through electrospray ionization (ESI) without protonation, consistent with previous reports and our results [24–26]. A series of novel bn-44 ions (rather than bn-28) were detected after CID using an ion trap mass spectrometer. High resolution mass spectra (HRMS) showed that the elimination of 44 Da could result from the loss of CO2 (theoretical mass: 43.9898) or C2H4O (acetaldehyde; 44.0262). Because these fragment ions are seldom generated by common protonated peptides, the unusual fragmentation pathways implied piqued our interest.

Figure 1
figure 1

(a) Structures of N-terminal deuterohemin or deuteroporphyrin containing hexapeptides (DhHP-6 or DpHP-6). DhHP-6 molecular formula: C59H76FeN14O12 + Cl–; exact cation mass: 1228.5. (b) Sequences of DhHP-6 derived peptides and deuterohemin-peptide complexes (note: A* is β-Ala)

Full size image

Ternary metal complexes [Mn+(L)m–P](n-m)+ (M: transition metal ion; L: suitable auxiliary ligand; P: neutral peptide or amino acid) have a similar structure to DhHP-6. These complexes produce radical cations upon CID [27–29]. In the MS/MS mass spectrum of Cyt c, along with a and y ions, c and z radical ions were detected [30]. Since these radical ions are commonly observed from electron capture dissociation (ECD), Breuker and McLafferty termed this unusual phenomenon ‘Native ECD’ and used these radical fragment ions to characterize the noncovalent interactions of Cyt c [30]. Our question is: Does the formation of bn-44 ions from DhHP-6 involve a radical reaction in the gas phase? If not, which fragmentation mechanism generates these bn-44 ions? To answer these questions, a series of peptides were designed, as shown in Figure 1b. In addition, we calculated the proton affinity of deuteroporphyrin with DFT and compared its PA value with that of Arg and Lys. Studies of these peptides and peptide-deuterohemin mixtures demonstrated that the gas-phase dissociation of DhHP-6 was greatly influenced by deuterohemin and that an N-terminal charge on the peptide rather than a radical reaction leads to formation of bn-44 ions.

2 Experimental

2.1 Materials and Sample Preparation

All peptides, including isotope labeled peptides, used in this work (purity ≥ 90%) were synthesized using a conventional solid-phase method by Changchun BCHT Co., Ltd. Methanol (chromatographic grade) and formic acid (analytical grade) were purchased from Tedia Company Inc. (Cincinnati, OH, USA). Alanine-1-13C and Fmoc-chloride were purchased from Sigma Chemical (St. Louis MO, USA) to generate Fmoc-alanine-1-13C as described previously [31]. Each peptide was dissolved in Milli-Q water to make a stock solution of 2 mmol/L and diluted to a final concentration of 30 μmol/L with methanol/water/formic acid at a volume ratio of 49.5:49.5:1, for direct ESI-MS analysis.

A noncovalent complex of [deuterohemin (FeIII) (hexapeptide)]+ was prepared by mixing 100 μL of peptide stock solution and 100 μL of saturated deuterohemin solution (partially soluble in methanol, insoluble in water) and diluted with 800 μL methanol before ESI-MS analysis [28].

2.2 Mass Spectrometry

An Agilent G6300 ESI mass spectrometer (Agilent Technologies, Palo Alto, California, USA) in positive ion mode was used for MSn experiments involving DhHP-6. Samples were infused at 180 μL/h and determined in triplicate on 3 different day. The optimized instrumental conditions were as follows: spray voltage, 4000 V; capillary exit voltage, 150 V; skimmer voltage 50 V; dry gas, N2, 325°C (5.00 L/min); nebulizer gas, N2 (15 psi). For CID MS/MS experiments: isolation selection, standard; fragmentation cutoff, manually set; isolation width, 3.0 m/z; fragmentation Ampl, 2.0 V for singly charged ions and 1.0 V for doubly and triply charged ions; 0.65 V for MS/MS/MS performance.

MicroTOF-Q II (Bruker Daltonics, Bremen, Germany) was used as a high resolution mass spectrometer to distinguish the loss of CO2 (43.9898) from C2H4O (44.0262). Instrumental parameters were as follows: capillary voltage, 4000 V; dry gas, N2, 200°C; nebulizer gas, N2, 0.4 bar; isolation width 3.0 m/z (varied from 1.0 to 5.0 to examine potential isobaric contamination); ISCID energy, 10 eV; collision energy, 60 eV for singly charged precursor ions and 35 eV for doubly and triply charged ions. In addition, mass calibration was performed using sodium formate solution before sample injection into the ESI source.

2.3 Determination of Proton Affinity

The proton affinity (PA) of deuteroporphyrin was calculated at the same level as Lys and Arg using density functional theory (DFT), as reported by Paizs et al. [32]. Ab initio calculations were carried out to find the global minimum for deuteroporphyrin in neutral and protonated form. Molecular dynamics (MD) simulations were performed using Discovery Studio (DS) 2.5 [33] with a CHARMm force field [34]. The structures derived in this way were further optimized at the HF/3-21G, B3LYP/6-31G(d), and finally B3LYP/6-31 + G(d,p) levels. Total energies calculated at B3LYP/6-31 + G(d,p) level were corrected for zero-point vibration energies (ZPE) at the same level. The PA value of deuteroporphyrin was obtained as the difference between the ZPE-corrected DFT total energies of the protonated and neutral forms given in kcal mol–1. All four imine nitrogen atoms in deuteroporphyrin were checked as protonation sites. All quantum chemistry calculations were performed with Gaussian 09 [35].

3 Results and Discussion

3.1 Fragmentation Patterns of DhHP-6 Ions are Affected by Labile Protons and Fixed Charge

The peptide DhHP-6 produces three precursor ions, m/z 1228.5(1+), 614.8(2+), and 410.2(3+), shown in Supplementary Figure S1 (Supporting Information). Figure 2 shows that each precursor ion gives different fragmentation patterns upon CID using an ion trap mass spectrometer.

Figure 2
figure 2

MS/MS spectra of each precursor ion (a) M+, (b) [M+ + H]2+, and (c) [M+ + 2H]3+ using the ion trap mass spectrometer (left) and corresponding fragmentation pathways of precursor ions to produce b and y ions (right)

Full size image

The MS/MS spectrum of the singly charged ion presents a series of b ions as shown in Figure 2a. The peak at m/z 563.2 corresponds to positively charged deuterohemin. Note that the singly charged ion has no labile proton because the charge is carried by the N-terminal Fe(III). This phenomenon has been reported previously [24–26]. For peptide ions without a labile proton, it has been suggested that a proton can dissociate from the carboxylic acid group and attach to the amide nitrogen, initiating cleavage of the amide bond through a salt-bridge intermediate (Figure 2a, pathway A). This may explain the fragmentation of the singly charged ion [21, 22].

Our interest is in the detection of a series of abundant bn-44 ions (n: 2, 3, 4). A common bn ion usually features a protonated oxazolone in its C-terminus that dissociates to produce an ions by loss of CO (28 Da) [7, 36, 37]. Some amino acid side chains can also be involved in competitive fragmentation reactions, resulting in b ions, as the “Asp effect” mentioned above [18, 19]. Therefore, it is conceivable that the generation of bn-44 ions involves either a specific b ion or a specific fragmentation pathway. Upon further dissociation of singly charged bn ions (n: 2, 3, 4) only bn-44 ions were detected in the MS3 spectra (Supplementary Figure S2, Supporting Information), confirming that bn-44 ions were generated by dissociation of the corresponding bn ions.

These bn-44 ions were also observed in the MS/MS spectrum of the doubly charged DhHP-6 ion. This extra proton can migrate to various sites upon excitation, resulting in cleavage of amide bonds and production of a series of complementary b/y sequencing ions (Figure 2b, pathway B) [5–7].

No bn-44 ions were detected in the MS/MS spectrum of the triply charged DhHP-6 ion (Figure 2c). However, selective cleavage occurred at a His residue, leading to the generation of abundant [b2 + H]2+ ions. This histidine effect has been extensively studied because it results in the loss of sequence information [38]. It has been suggested that a proton moves from imidazole to the nitrogen of the C-terminal adjacent amide bond, promoting nucleophilic attack by the imidazole nitrogen atom on the carbon of the protonated amide. This process leads to the generation of a bicyclic ion as shown in Figure 2c, pathway C [38, 39]. Supplementary Figure S2b (Supporting Information) shows the MS3 spectrum of the [b2 + H]2+ ion. The [a2 + H]2+ ion was a unique product of [b2 + H]2+ through the loss of CO. This suggests differences between [b2 + H]2+ and singly charged b2 ions in either structure or fragmentation pathways.

Comparing the product ions generated by DhHP-6 ions of different charge, we can conclude that the number of labile protons influences the fragmentation pathways of the precursor ions. In particular, a series of novel bn-44 fragments were detected in the MS/MS spectra of both singly and doubly charged ions. The questions remain, however: What is the cause of the neutral loss of 44 Da, and where does it come from? To address these questions, isotope labeling and high resolution mass spectrometry was employed to study a series of DhHP-6-derived peptides and deuterohemine–peptide complexes.

3.2 Differentiation of C2H4O and CO2 by High Resolution Mass Spectrometry

The ESI-Q-TOF instrument was used as a high resolution mass spectrometer to identify the neutral loss. Figure 3 gives the MS/MS spectrum of the singly charged ion generated by the ESI-Q-TOF instrument. The MS/MS spectra of the doubly and triply charged ions are presented in Supplementary Figure S3 (Supporting Information). To estimate potential isobaric contamination due to the 3.0 m/z isolation width, MS/MS spectra of the m/z 614.0 fragment with uni-resolution were performed (Supplementary Figure S4, Supporting Information).

Figure 3
figure 3

ESI-Q-TOF MS/MS spectra of the singly charged DhHP-6 ion. (a) Full scan spectrum (left) and differences between experimental and theoretical masses of the neutral loss fragments from bn: H2O, CO, CH3CHO, and CO2 (right); extended MS/MS spectra of close to (b) the b2 +, (c) the b3 +, and (d) the b4 + ions

Full size image

Comparing Figures 2, 3, and Supplementary Figure S3, Supporting Information, different fragmentation patterns are observed for the same ions using different mass spectrometers [4, 5, 40]. In the ESI-Q/TOF mass spectra, abundant bn-H2O, bn-CO, and bn-44 ions (n: 2, 3, 4) were obtained from the singly charged ion (Figure 3); the doubly charged DhHP-6 ion produced only b type ions (Supplementary Figure S3a, Supporting Information), in contrast to the b/y ions observed using the ion trap (Figure 2). These b ions showed different neutral losses: the singly charged bn ions lost predominantly 44 Da (n: 2–4) fragments, whereas the doubly charged b ions lost CO (28 Da) and H2O (18 Da). Note that two possible neutral losses of 44 Da can be unambiguously distinguished: CO2 (43.9898) from the oxazolone or C2H4O (44.0262) from the Thr side chain. As shown in Figure 3a (right), the neutral loss of 43.993 Da from the b2 ion could be assigned as CO2 with a mass error of 0.0032 Da, and of 44.025 Da from the b4 ions to C2H4O with a mass error of 0.0012 Da. However, the loss of 44.017 Da from the b3 ion is intermediate between the masses 44.0262 (C2H4O) and 43.9898 (CO2), which suggests a mixture of two molecules. This was investigated using isotope labeling.

3.3 Investigation of bn-44 Ions Using Derivatives and Isotope Labeling

To further investigate the formation of bn-44 ions, DhHP-6 was modified and isotopically labeled. First, considering that the N-terminal carboxylate of DhHP-6 may undergo decarboxylation, resulting in the loss of 44 Da (CO2) [18, 27], NH2-DhHP-6 was obtained by amidation of the N-terminal carboxylate of DhHP-6. Thus, there is one mass difference between DhHP-6 and NH2-DhHP-6. However, decarboxylation is not apparent from the product ions generated from singly charged NH2-DhHP-6, which differ by only 1 Da from the corresponding products of DhHP-6 (Supplementary Figure S5, Supporting Information).

To determine whether the 44 Da fragment was lost from the third Thr residue and to identify any effect of Thr on the formation of bn-44 ions, the Thr residue in DhHP-6 was replaced with Ala. In addition, the carboxyl carbon was labeled with 13C to investigate whether 13CO2 became apparent in the oxazolone structure of the b3 ion. As expected, the b2-CO2 ion was still abundant, in contrast with the less abundant b3-45 (13CO2) ion; no bn-C2H4O (n: 3, 4) or bn-CO2 (n: 4) ions were detected (Figure 4). Thus, it can be concluded that bn ions (n: 2, 3) formed close to the N-terminus of DhHP-6 prefer CO2 loss and that formation of bn-C2H4O (n: 3, 4) may indeed be related to the Thr side chain. The absence of b3-C2H4O and the presence of the less abundant b3-45 (13CO2) suggest that the loss of 44.017 Da from b3 reflects both CO2 and C2H4O.

Figure 4
figure 4

Q-TOF mass spectra: (a) full scan MS/MS spectrum of the singly charged DhHP-6-(A-1-13C) ion; (b) MS/MS spectrum from m/z 750 to 850

Full size image

Previous reports showed that protonated peptides containing Thr or Ser residues lose H2O preferentially upon CID [7, 41–43]. CO2 loss often involved a carboxylate, especially for peptide radical cations [18, 27]. The mechanism of the loss of the 44 Da in this case is investigated further as described below.

3.4 Factors Affecting the Formation of bn-CO2 and bn-C2H4O

We have noted that DhHP-6 has similar composition to the [metalIII(salen)(peptide)] complex, except for a covalent connection instead of a noncovalent interaction between metal-ligand and peptide. The [CuII(ligand)(peptide)] complex gave various product ions, including both radical and neutral losses from amino acid side chains upon CID because both radical-driven redox reactions and charge-directed cleavages occurred [27–29, 44].

To investigate if the fragmentation of DhHP-6 ions involves radical reactions, deuterohemin(FeIII) was mixed with the hexapeptide and subjected to CID. Because the hexapeptide YGGFLR has often been used to evaluate oxidation by metal-ligand complexes [27, 28], it was mixed with the deuterohemin. Figure 5a shows the MS/MS spectrum of the complex [deuterohemin (FeIII)(YGGFLR)]+. The radical ion YGGFLR+● (m/z: 711) was observed, as were [deuterohemin(FeIII)]+ and [YGGFLR + H]+ (m/z: 712). However, when YGGFLR was replaced with AHTVEK-NH2, no radical cation P+● was detected, as shown in Figure 5b. This suggests that a metalloporphyrin can serve as an alternative to the metal-ligand complex in the gas-phase redox reaction. However, it is evident that a great barrier exists for electron transfer from AHTVEK-NH2 to Fe(III) in deuterohemin. Thus, radical-driven fragmentation of DhHP-6 is not feasible.

Figure 5
figure 5

MS/MS spectra of (a) [deuterohemin(FeIII) (YGGFLR)]+ and (b) [deuterohemin(FeIII) (AHTVEK-NH2)]+ using the Q-TOF mass spectrometer

Full size image

To investigate the effect of the N-terminal Fe(III) on fragmentation patterns of DhHP-6, DpHP-6 was obtained from DhHP-6 by removing Fe(III) from the N-terminal deuterohemin. Unlike DhHP-6, DpHP-6 was normally charged due to protonation. Similar product ions were observed in the MS/MS spectrum of DpHP-6 as for DhHP-6, except for a 53 Da shift (Supplementary Figure S6, Supporting Information). The singly charged DpHP-6 ion also produced a series of b-type fragments, and neutral losses from the bn (n: 2, 3, 4) ions of DpHP-6 were the same as from DhHP-6. No y ions were detected, despite the fact that the basic residue Lys was located in the C-terminus of DpHP-6. “This suggests that the N-terminal deuteroporphyrin has a very strong proton affinity (PA), sequestering the added proton. This was confirmed by DFT calculations, which revealed a PA value for deuteroporphyrin of 248.2 kcal mol–1, much higher than that of Lys (237.3 kcal mol–1) and slightly lower than that of Arg (253.3 kcal mol–1). These results suggest that the N-terminal charge plays a crucial role in the loss of the 44 Da neutral fragments through a charge-remote fragmentation pathway.

To confirm the effect of a fixed charge on the formation of bn-44 ions, an Arg residue was used in place of the N-terminal deuterohemin of DhHP-6 [10, 18]. The MS/MS spectrum of singly charged RAHTVEK-NH2 was compared with that of singly charged AHTVEK-NH2. The bn-44 ions (n: 4–5), corresponding to bn-C2H4O ions, were generated exclusively by RAHTVEK-NH2; no bn-CO2 ions were detected among accurate m/z values (Figure 6). This suggests that deuterohemin plays an important role in the formation of the bn-CO2 ions and that the formation of bn-C2H4O ions (n: 3–5) is directly related to the Thr residue and N-terminal fixed charge.

Figure 6
figure 6

MS/MS spectra of the singly charged (a) RAHTVEK-NH2 and (b) AHTVEK-NH2. Inset: extended MS/MS spectra of RAHTVEK-NH2 in m/z range of (c) 315–370; (d) 415–470; and (e) 510–570

Full size image

3.5 Proposed Mechanism for the Elimination of CO2 and CH3CHO

Dehydration of Thr side chains initiated by a labile proton has been widely reported [7, 41–43]. However, there is no labile proton in the singly charged bn ions of DhHP-6, DpHP-6, and RAHTVEK-NH2: the proton is sequestered in the N-terminus. In addition, it is evident that CH3CHO is eliminated from the Thr side chain without the involvement of a labile proton. A charge-remote elimination reaction involving a six-membered ring and H rearrangement is suggested, corresponding to the formation of bn-CH3CHO as shown in Scheme 1b. This charge-remote fragmentation pathway, resulting in CH3CHO loss, is similar to the loss of methane sulfenic acid (CH3SHO, 64 Da) from methionine sulfoxide-containing peptide ions in the absence of labile protons (Scheme 1c) [15].

Scheme 1
scheme 1

Proposed mechanism for the generation of (a) bn-CO2 and (b) bn-CH3CHO ions, and (c) the loss of CH3SHO

Full size image

CO2 loss from the non-protonated oxazolone ring of DhHP-6 b2 ion may involve a 2H-azirine structure (Scheme 1a). This structure has been widely reported [45]. It was proposed by Van Stipdonk et al. to explain fragmentation of metal complexes with positively charged peptides upon CID using an ion-trap mass spectrometer [46]. Aromatic groups (phenyl, chlorophenyl, etc.) have been reported to promote the synthesis of 2H-azirine compounds [45, 47]. We speculate that porphyrin or imidazole may play a similar role in the stabilization of this structure because of the similar π-system.

4 Conclusions

This work provides significant information on the effect of the deuterohemin group in the N-terminus on the dissociation patterns of DhHP-6 upon CID process. These gas-phase reactions are also influenced by charge states, the identity and position of the amino acids, and even the instruments used. Our conclusion is based mainly on results obtained using a Q-TOF instrument. A number of bn-44 ions were detected. Two different neutral fragments [CH3CHO (44.0262 Da) and CO2 (43.9898 Da)] have been unambiguously identified by accurate mass measurements, and the formation pathways have been proposed. The experimental and theoretical results gave evidence that both deuterohemin and deuteroporphyrin can function as a charge carrier, similar to the N-terminal R residue, leading to the loss of CH3CHO from Thr through a charge-remote reaction. This novel fragmentation of Thr in the absence of labile protons provides featured additional information, which could be used for promoting the identification of peptide sequence. The porphyrin may show a neighbor effect, resulting in loss of CO2 from the non-protonated oxazolone nearby. This could lead to the generation of a novel 2H-azirine structure. This is consistent with prior results in the synthesis of 2H-azirines: aromatic groups are known to increase the stability of aziridine and azirine rings. Because one-third of natural proteins are metalloproteins, the present work using a model peptide presents important fragmentation information. In addition, we demonstrate that the deuterohemin group can induce formation of radical peptide ions by gas-phase redox reactions when mixed with a suitable peptide. This result suggests that it may be possible to develop and apply deuterohemin to produce radical fragment ions in CID without the use of ECD and ETD, providing structural information and, therefore, facilitating protein identification. Much more research on dissociation of metalloproteins is required; this research may be of value in manual de-novo sequencing for identification of metalloproteins.