lable at ScienceDirect

Analytica Chimica Acta 1011 (2018) 59e67

Contents lists avai

Analytica Chimica Acta

journal homepage: www.elsevier .com/locate/aca

Time-resolved method to distinguish protein/peptide oxidationduring electrospray ionization mass spectrometry

Jiying Pei a, b, Cheng-Chih Hsu c, Kefu Yu b, Yinghui Wang b, Guangming Huang a, *

a Department of Chemistry, School of Chemistry and Materials Science, University of Science and Technology of China (USTC), Hefei, 230026 PR Chinab School of Marine Sciences, Guangxi University, Nanning 530004, PR Chinac Department of Chemistry, National Taiwan University, Taipei 10617, Taiwan

h i g h l i g h t s

* Corresponding author.E-mail address: gmhuang@ustc.edu.cn (G. Huang)

https://doi.org/10.1016/j.aca.2018.01.0250003-2670/© 2018 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� EC- and CD-induced peptide/proteinoxidation during ESI-MS was sys-tematically investigated.

� A time-resolved method was pro-posed to distinguish the two kinds ofoxidation.

� EC- and CD-induced peptide/proteinoxidations were found to be closelyrelated with the experimentalparameters.

a r t i c l e i n f o

Article history:Received 19 October 2017Received in revised form5 January 2018Accepted 8 January 2018Available online 31 January 2018

Keywords:Protein/peptide oxidationCorona dischargeElectrochemistryTime-resolutionElectrospray ionizationMass spectrometry

a b s t r a c t

Electrospray ionization mass spectrometry (ESI-MS) is one of the most prevalent techniques used tomonitor protein/peptide oxidation induced by reactive oxygen species (ROSs). However, both coronadischarge (CD) and electrochemistry (EC) can also lead to protein/peptide oxidation during ESI. Becausethe two types of oxidation occur almost simultaneously, determining the extent to which the twopathways contribute to protein/peptide oxidation is difficult. Herein, a time-resolved method wasintroduced to identify and differentiate CD- and EC-induced oxidation. Using this approach, we separatedthe instantaneous CD-induced oxidation from the hysteretic EC-induced oxidation, and the effects of thespray voltage and flow rate of the ESI source on both oxidation types were investigated with a home-made ESI source. For angiotensin II analogue (b-DRVYVHPF-y), the dehydrogenation and oxygenationspecies were the detected EC-induced oxidation products, while the oxygenation species were the majorCD-induced oxidation products. This time-resolved approach was also applicable to a commercial HESIsource, in which both CD and EC were responsible for hemoglobin and cytochrome c oxidation withupstream grounding while CD dominated the oxidation without upstream grounding.

© 2018 Elsevier B.V. All rights reserved.

1. Introduction

Protein/peptide oxidative modification in vivo is closely

.

associated with various diseases and aging, for example, Sultanaet al. reported that elevated protein nitrotyrosine was found withAlzheimer's disease [1]. Exposure to reactive oxygen species (ROSs)is thought to play a critical role in the modification [2,3]. In vivoprotein/peptide oxidation modification can be characterized bycircular dichroism [4], fluorescence spectroscopy [5], gel electro-phoresis [6], UV-vis spectroscopy [4], dynamic light scattering

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e6760

(DLS) [4], nuclear magnetic resonance (NMR) [7] and mass spec-trometry (MS) [8]. Among them, MS is one commonly usedanalytical technique for its capacity of obtaining highly accurateintact masses and the three-dimensional structure of proteins [9].Electrospray ionization-mass spectrometry (ESI-MS) is preferredfor the operation under atmospheric pressure conditions [10e12].However, a major obstacle for the ESI-MS technique is that twomajor types of artificial protein/peptide oxidation occur during theESI process, i.e., electrochemistry (EC)- [13] and corona discharge(CD)-induced [14] oxidation, which can mislead investigations ofin vivo protein/peptide oxidation. Thus, knowing the properties ofthese two types of instrument-related oxidation is of greatimportance to accurately study in vivo protein/peptide oxidation.

An ESI source can be viewed as a controlled-current electrolyticcell [13,15]. When an analyte solution flows through the stainless-steel capillary during a typical ESI process, an oxidation reactionoccurs at the solution/electrode interface in the positive mode.Moreover, the oxidation ratios of the analytes are proportional tothe solution/electrode contact time. In addition to the EC-inducedoxidation, the CD in the gas phase can also result in protein/pep-tide oxidation during the ESI process [14,16,17]. Once the sprayvoltage in an ESI source exceeds a critical onset value, a plasmacontaining abundant ROSs is generated in the vicinity of the emittertip, and the plasma further induces various oxidation reactions(Fig. 1). Unexpected protein oxidations complicate spectruminterpretation, reduce sensitivity by splitting ion current, interferewith peak isolation for MS2 fragmentation, and alter the chemicalstate of the proteins. To decrease CD during ESI, organic solvent isgenerally added, which will change the conformation of proteins/peptides [18], and is unfavorable for native protein analysis[19e21]. On the other hand, protein oxidation during ESI can alsobe strengthened for protein footprinting investigation. Forexample, Downard et al. utilized ESI-based oxidation to investigateprotein folding/unfolding, protein interactions, impact of oxidationon protein aggregation, and the residue side chain solvent acces-sibility of proteins based on the oxidation level and rate [14].Therefore, to effectively avoid or take full advantage of these twotypes of oxidation, differentiation of them is necessary. However,since the CD- and EC-induced oxidations occur almost simulta-neously, their contributions to the ESI-based protein/peptideoxidation are unclear [22e25].

Fig. 1. Schematic diagram of the CD- and EC-induced protein/peptide oxidation duringESI-MS. The EC-induced oxidation occurs at the solution/electrode (conductor) inter-face, and the analyte is oxidized to Ox1 (labeled in red). Alternatively, the CD-inducedoxidation occurs at the tip of the spray emitter, and the analyte is oxidized to Ox2(labeled in blue). The oxidation extent of the EC-induced oxidation is proportional tothe solution/electrode contact time, while that of the CD-induced oxidation does notdepend on the spray time. (For interpretation of the references to colour in this figurelegend, the reader is referred to the Web version of this article.)

Kim et al. [26] and Liu et al. [24] observed unexpected protein/peptide oxidation, which was decreased by rearranging LCplumbing, in a conventional liquid chromatography-mass spec-trometry (LC-MS) configuration. The oxidations were expected tobe relatedwith the enhanced electrochemical reaction of ESI sourcethrough construction of an upstream grounding loop between theion source and the nearest upstream grounded metal element[25,27]. However, Morand et al. favored CD that was responsible forpeptide oxidation during ESI, because they observed the appear-ance of the modified ions with the observation of a faint blue huenear the tip of the electrospray needle [22]. Chen et al. [23]attributed Ab peptide oxidation to gradual corrosion of stainlesssteel electrospray emitters that could promote electrical discharge.They also thought that the increased emission current strength-ened the electrochemical reactions associated with Ab peptideoxidation. Use of redox buffers and reduction of electric field weresuggested to decrease the oxidation. Lars Konermann et al. sys-tematically determined the effects of EC and CD on protein oxida-tion [28]. Based on a visible inspection of the light emission to judgewhether CD occurred and the record of the voltage-current plotsunder different nebulizer gas conditions, it was concluded thatprotein oxidation in ESI is predominantly mediated by CD-generated ROSs. Additionally, an off-line electrolysis experimentwas conducted to illustrate that EC cannot account for proteinoxidation under typical ESI operating conditions. Although LarsKonermann helped to understand the protein oxidation mecha-nism in ESI, direct MS proof is still required to investigate the twotypes of simultaneous oxidation. Previously, we minimized the EC-induced oxidation for MS measurements of proteins using amodified ESI method; however, the CD-induced oxidation was stillobserved [29]. Considering that the EC-induced oxidation is pro-portional to the solution/electrode contact time and the CD-induced oxidation is not, we were inspired to design a time-resolved ESI approach to distinguish the CD- and EC-induced pro-tein/peptide oxidation.

The time-resolved method is based on the variation in the an-alyte oxidation extent with the spray time. To better isolate the twotypes of oxidation in space and time, a homemade ESI source wasused to investigate the CD- and EC-induced oxidation. Using thehomemade ESI source, the effects of the spray voltage, flow rate,and solvent on the CD- and EC-induced oxidation were separatelyinvestigated. Also, the CD extent was characterized by the lumi-nescence recorded by a photomultiplier. Then, protein oxidation ina commercial HESI source with and without upstream groundingwas investigated using the established method.

2. Experimental

2.1. Materials and reagents

HPLC-grade methanol (CH3OH) was purchased from HoneywellBurdick& Jackson Inc. (U.S.A.). Angiotensin II analogue (Ang II

0) and

KKTCAA were synthetized by Sangon Biotech (Shanghai SangonBiological Engineering Technology & Services Co. Ltd., Shanghai,China). Cytochrome c (Cyt c) and hemoglobin (Hb) were purchasedfrom Sangon Biotech. Melittin, reserpine, and formic acid (FA) wereobtained from Sigma-Aldrich Chemical Co. Ltd. (U.S.A.). NH4Ac wasobtained from Sinopharm Chemical Reagent Co. Ltd. (Beijing,China). All the reagents were used without any further purification.Distilled water (18.2MU) was produced by a Milli-Q system (Mil-lipore Inc., Bedford, MA, U.S.A.).

Unless noted otherwise, all peptide and protein solutions wereprepared in CH3OH, CH3OH/H2O (v/v, 1:1) or H2O with the additionof NH4Ac and FA.

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e67 61

2.2. Mass spectrometry

All MS experiments were performed using a Thermo LTQ orOrbitrap Exactive Plus mass spectrometers (Thermo Fisher Scien-tific, San Jose, CA, U.S.A.). A homemade or commercial ESI source(named HESI) was used throughout the experiments. For thehomemade ESI source, the solution was sprayed from a 50-60 cm-long fused silica capillary (i.d.100 mm, o.d. 365 mm) (Fig. S1), and thespray voltage (3e6 kV) was applied on the syringe needle (Type304 stainless steel, length 51mm, i.d. 0.41mm, o.d. 0.72mm). Toobtain a more stable spray, the capillary tip was etched using themethod introduced by Kelly [30]. The spray emitter of the HESIsource is metal material, on which the spray voltage was applieddirectly. The mass spectrometer conditions during the experimentswere set as follows: S lens voltage, 42% (positive mode); capillarytemperature, 275 �C; ion injection time,10ms; signal average, threemicroscans; automatic gain control (AGC), 3� 104 for LTQ Velos Proand 1� 106 for Orbitrap Exactive Plus.

2.3. Detection of light emission

Light emission during the ESI process was measured using aphotomultiplier integrated in a MPI-A electrogenerated chem-iluminescence analytical system (Xi'an Remex Analysis InstrumentCo., Ltd., China). The 5 cm polyimide coating on the front end of thefused silica capillary was burned off to create a detection window(close to the etched emitter). The capillary emitter was placedparallelly to the photomultiplier with the burned end right abovethe detector. A grounding metal electrode was placed opposite theemitter with the distance of approximately 0.5 cm. Sample solutionwas delivered through the built-in injection pump of the massspectrometer, and spray voltage was applied by the built-in highvoltage power supply of the mass spectrometer. Luminescent in-tensity was recorded by the included software of the MPI-A elec-trogenerated chemiluminescence analytical system, and theaverage values for three determinations were used as the lumi-nescent intensities.

3. Result and discussion

3.1. Observation of CD- and EC-induced peptide oxidation duringhomemade ESI source

First, we used Ang II’ (b-DRVYVHPF-y, MW¼ 1031.5189), apeptide that contains the readily oxidizable amino acid, tyrosine[31], as the model compound to differentiate the CD- and EC-induced oxidations. The homemade ESI setup in Fig. S1 was used,and an approximately 50 cm insulated fused silica capillary wasused to deliver the sample solution from an injection syringe,where the spray voltage was applied, to the spray tip. The Ang II0

solution was continuously measured for 8min with a spray voltageof 2.5 kV and a flow rate of 2 mLmin�1. The results showed that AngII0 was barely oxidized during the first 2min. After 2min, oxidationspecies with m/z of 515.8, 514.8, 524.8, 523.8 and 522.8 began toappear in the spectra (Fig. 2a-1, a-2). According to the high-resolution MS spectra, these compounds were identified as[M � 2H þ 2Hþ]2þ (measured mass (after calibration),Mmea. ¼ 515.75975, d ¼ (measured mass � theoretical mass)/theoretical mass ¼ 0.58 ppm), [M � 4H þ 2Hþ]2þ

(Mmea. ¼ 514.75255, d ¼ 1.75 ppm), [M þ O þ 2Hþ]2þ

(Mmea.¼ 524.76405, d¼�1.24 ppm), [M þ O - 2H þ 2Hþ]2þ

(Mmea.¼ 523.75645, d¼�0.86 ppm), and [M þ O - 4H þ 2Hþ]2þ

(Mmea.¼ 522.74835, d¼�1.44 ppm), respectively (Fig. S2). Thedehydrogenation and oxygenation sites of Ang II0 are thought to beon the tyrosine side chain based on the MS2 spectra [32,33]

(Fig. S3). When the oxidation ratio of Ang II’ (Iox/I,Iox¼ I514.8 þ I515.8 þ I522.8 þ I523.8 þ I524.8, I ¼ Iox þ I516.8) was fittedwith the spray time, the curve showed that the oxidation ratio wasconstant for the first 2min and then began to increase (Fig. 2a-3).When a flow rate of 1 mLmin�1 was applied, the onset time for theoxidation increasewas delayed to 4min (Fig. S4). The volume of the50 cm-long fused silica capillary with a radius of 50 mm (the radiusof the capillary used throughout the experiment was 50 mm) wascalculated to be approximately 4 mL. The time of 4min (or 2min)with a flow rate of 1 mLmin�1 (or 2 mLmin�1) corresponds to thetime required for the sample solution with electrode (the syringeneedle) contact to pass through the 50 cm fused silica capillary.Therefore, these oxidations should be attributed to EC. However,when a spray voltage of 3.5 kV was applied with a flow rate of2 mLmin�1, the oxidation products of [M þ O þ 2Hþ]2þ (m/z 524.8)were observed at the very beginning of the spray, and otheroxidation products, including [M � 2H þ 2Hþ]2þ (m/z 515.8) and[M þ O - 2H þ 2Hþ]2þ (m/z 523.8), were observed 2min later(Fig. 2b-1, b-2). Since the CD reaction accompanies the electrosprayprocess [34], the oxidation observed at 0min can be attributed toCD (Fig. 2b-3). It is interesting that though both CD and EC pro-duced [Mþ 2Oþ 2Hþ]2þ ions of Ang II0, the oxidation sites in aminoacid were different based on the MS2 spectra. While the oxidationsite of EC-induced [M þ 2O þ 2Hþ]2þ was b4 (Fig. S3), the possibleoxidation sites of CD-induced [M þ 2O þ 2Hþ]2þ included y1, y2,and y5 (Fig. S5). This difference might be due to that CD-inducedoxidation, which occurred in millisecond timescale, was moredrastic than EC-induced oxidation, leading more amino acids to beoxidized. The signal intensity of [M þ O þ 2Hþ]2þ (m/z 524.8)decreased after approximately 2 min. This can be attributed to theconsumption of Ang II0 by the EC-induced oxidation that occurs atapproximately 2 min, which would result in less CD-induced[M þ O þ 2Hþ]2þ. Additionally, the partial [M þ O þ 2Hþ]2þ wastransformed to [M þ O - 2H þ 2Hþ]2þ by EC. Based on the aboveresults, CD mainly leads to the formation of covalent þ16 Da ad-ducts of Ang II0, whereas EC can produce both dehydrogenation andoxygenation products.

However, CD- and EC-induced oxidation products were specificto compounds. When a peptide, b-KKTCAA-y, was tested, CD-induced oxidation products included [M � 2H þ Hþ]þ (m/z 619.3),[M þ 2O þ Hþ]þ (m/z 653.3), [M þ 2O - 2H þ Hþ]þ (m/z 651.3),[M þ 3O þ Hþ]þ (m/z 669.3), and [M þ 3O - 2H þ Hþ]þ (m/z 667.3),while EC-induced oxidation products included [M � 2H þ Hþ]þ (m/z 619.3), [M þ 2O þ Hþ]þ (m/z 653.3), [M þ 3O þ Hþ]þ (m/z 669.3),and [2M � 2H þ 2Hþ]2þ (m/z 620.3) (Fig. S6). Besides, EC and CDcould induce both the dehydrogenation and oxygenation productsof enkephalin (b-YGGFM-y) (Fig. S7), a peptide containing tworeadily oxidizable amino acids, tyrosine and methionine. The dif-ference of CD- and EC-induced oxidation products might be relatedto the compound's chemical properties. Bateman reported thatpeptides with such amino acids as methionine, aromatic aminoacids and leucine were more easily oxygenated by the electro-chemical reaction in capillary electrophoresis-nanoelectrospraymass spectrometry [35].

3.2. Characterization of CD- and EC-induced peptide oxidation viaadjustment of spray voltage, flow rate, and solvent composition

Using this time-resolved method, we investigated the effect ofthe spray voltage on the CD- and EC-induced oxidation. In ourdevice, the EC-induced oxidation was not observed in the first4.5 mL; thus, we could investigate the CD-induced oxidationwithout any influence from the EC-induced oxidation for the first4.5min at a flow rate of 1 mLmin�1. Here, a time of 4.5min corre-sponds to the time required for the sample solution in contact with

Fig. 2. Total ion chromatogram (TIC), selected ion chromatograms (SICs), mass spectra, and oxidation curves (Iox/I vs time) of Ang II’ (2.5 mgmL�1 in H2O with 5mM NH4Ac) afterundergoing CD- and EC-induced oxidation during ESI-MS with the homemade ESI source with a spray voltage of (A) 2.5 kV and (B) 3.5 kV. Flow rate¼ 2 mLmin�1. The massspectrummagnifications are for both the upper and lower panels and are applicable to the other figures in the text and in the Supporting Information. The species withm/z 519e521are discharge peaks.

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e6762

the electrode to pass through the approximately 57 cm-long fusedsilica capillary. The length of the fused silica capillary was variablefrom 50 cm to 60 cm during the experiment process, because theetched capillary was easily broken off at the tip. When the sprayvoltage was increased from 2.5 kV to 4.5 kV in a time interval of30 s, more severe Ang II0 oxidationwas observed, and the oxidationratio of the CD-induced oxidation remained relatively constant ateach step of the increased spray voltage (Fig. 3a and b). This is inagreement with previous investigations that noted that a higherspray voltage produced more ROSs and resulted in more severeanalyte oxidation [34]. To investigate the effect of the spray voltageon the EC-induced oxidation, the spray voltage was tuned before4.5min and set to 2.5 kV afterwards to avoid variations in the CD-induced oxidation. Thus, any change after 4.5min could be iden-tified as an EC-induced change. The EC-induced oxidation ratiogently increased after 4.5minwithout a change in the spray voltage(Fig. 3c, curve 1). When the spray voltage changed from 2.5 kV to4.5 kV once or twice in the first 4.5min, the CD-induced oxidationwas immediately observed, as indicated by a “rectangular bulge.”Additionally, the slopes of the oxidation curves increased after4.5min, compared with the oxidation curves slope withoutadjustment of spray voltage (Fig. 3c, labeled by the histograms).When the starting time to adjust the spray voltage was changed,the time when the slopes of the oxidation curves increased variedaccordingly, with the time difference being 4.5min as shown in

Fig. S8. These results indicated that a higher spray voltage wouldalso result in more severe EC oxidation, which was due to that moremolecules would be oxidized to compensate the spray current forthe elevated spray voltage. Since it took approximately 4.5min(depending on the volume of the fused silica capillary and the flowrate of the ESI source) for the sample solution with the electrodecontact to pass through the fused silica capillary, the increased ECoxidation would be observed with a time delay of 4.5min.

When the oxidation curves of 2 and 3, inwhich the spray voltagewas tuned once (2) and twice (3) in the first 4.5min, respectively,were subtracted from the control experiment of curve 1, in whichthe spray voltage was not tuned in the first 4.5min, the netoxidation ratios of the two rectangular voltages (2.5/ 4.5 kV,1min each) were obtained (Fig. 3d). The “rectangular bulge” rep-resenting the enhanced CD-induced oxidation and the “risingslope” representing the enhanced EC-induced oxidation wereobserved. The chromatograms and mass spectra corresponding tothe oxidation curve of 3 are shown in Fig. 3e and f. With theelevation of the spray voltage at 1e2 min and 3.5e4.5 min, thesignal intensity of [M þ O þ 2Hþ]2þ immediately increased,accompanying the signal decrease of Ang II’. The EC-inducedoxidation products, [M � 2H þ 2Hþ]2þ and [M þ O -2H þ 2Hþ]2þ, were observed 4.5 min later, and the production rateof [M þ O - 2H þ 2Hþ]2þ and [M þ O þ 2Hþ]2þ increased at5.5e6.5min and 8e9min. The above results indicated that the

Fig. 3. Effect of the spray voltage on (aeb) CD- and (cef) EC-induced Ang II’ (2.5 mgmL�1 in H2O with 5mM NH4Ac) oxidation with the homemade ESI source. Flowrate¼ 1 mLmin�1. The spray voltages were adjusted to 2.5 kV, 3.5 kV, and 4.5 kV during the investigation of the CD-induced Ang II0 oxidation. The operation was conducted in thefirst 3min before the solution with the EC oxidation was electrosprayed. The spray voltages were varied between 2.5 kV and 4.5 kV in the investigation of the EC-induced Ang II0

oxidation. (d) Subtracted curves of 1 by 2 and 3 in (c). (eef) The corresponding chromatograms and mass spectra for curve 3 in (c).

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e67 63

mass spectral patterns from the spray voltage adjustment could bea criterion to distinguish CD-from EC-induced peptide oxidation.

The effect of the flow rate on the CD- and EC-induced oxidationcould also be investigated using this time-resolved method. Toavoid interference from the EC-induced oxidation when investi-gating the effect of the flow rate on the CD-induced oxidation, thetime to adjust the flow rate was set to ensure that the solution incontact with the electrode was not sprayed during the process. Theflow rate was changed from 0.2 mLmin�1 to 5 mLmin�1 before3min. The results showed that Ang II0 was more severely oxidizedby CD at a smaller flow rate (Fig. 4a). This can be attributed to twoaspects: 1) slower solution flow rates lead to Ang II0 having longerresidence time at the capillary tip and thus result in higher

Fig. 4. Effect of the flow rate on (a) CD- and (b) EC-induced Ang II’ (2.5 mgmL�1, in H2O with 5CD- and EC-induced Ang II0 oxidation in individual runs with the homemade ESI source. Tincreased spray flow rate (from 0.5 mLmin�1 to 5 mLmin�1). The curves in the dotted box inwhich was adjusted beforehand. A spray voltage of 3 kV was used in (aec).

oxidation levels, as suggested byMaleknia et al. [14]; 2) slower flowrates result in poor stability of the electrospray [36,37] (Fig. S9) thatmay be responsible for the formation of ROSs and analyte oxidation.When the flow ratewas held constant to investigate its effect on theEC-induced oxidation, the time for the sample solution to passthrough the fused silica capillary was 8.76, 4.26, 1.71, and 0.8min,respectively, at flow rates of 0.5, 1, 2.5, and 5 mLmin�1, with thelength of the fused silica capillary being approximately 54 cm. Andthe extent of EC-induced oxidation was inversely proportional tothe flow rate (Fig. 4b). This can be explained by the theory that thequantity of electric charges (Q) in an electrolytic cell is proportionalto the contact time (t) between electrode and solution, according toeq (1) in which I is the current.

mMNH4Ac) oxidationwith the homemade ESI source. (c) Effect of the flow rate on thehe rectangular frames indicate the decreased CD-induced Ang II0 oxidation due to thedicate the decreased EC-induced Ang II0 oxidation due to the increased spray flow rate,

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e6764

Q¼ It (1)

With longer contact time between Ang II0 solution and the sy-ringe needle, more Ang II molecules were oxidized.

When the flow rate was adjusted from 0.5 mLmin�1 to5 mLmin�1 for one or two segments of 0.3e0.4min before the so-lution in contact with the electrode was sprayed, the curves inFig. 4c were obtained. With the elevated flow rate, one or two“rectangular concaves” appeared in the oxidation curves, indicatingthe decreased CD-induced oxidation. EC-induced oxidation wasobserved at <4.3min and 5.2min, indicating the shortened time forthe sample solution to pass through the fused silica capillary. Thereduction of the oxidation curve slopes labeled by the dashed boxesin curves 2 and 3 in Fig. 4c proved the decreased EC-induced oxi-dations at the higher flow rate. Therefore, the mass spectral pat-terns from the flow rate adjustment could also be a criterion todistinguish the CD-from the EC-induced peptide oxidation.

The solvent composition is known to affect the discharge extentduring the electrospray process [34], and the electrochemical be-haviors of an analyte in a traditional electrolytic cell are related tothe solvent composition [38,39]. Using the time-resolved method,we explored the effect of the solvent on the CD- and EC-inducedoxidation in ESI-MS. Ang II0 in three solvents (CH3OH, CH3OH/H2O(v/v, 1:1), and H2O) was separately electrosprayed for 10min. Themass spectra recorded in the first 4min (the length of the fusedsilica capillary was approximately 51 cm) were used to investigatethe effect of the solvent on the CD-induced oxidation and the massspectra recorded after 4min were used to investigate the EC-induced oxidation. As shown in Fig. 5a, Ang II0 in H2O was themost severely oxidized by CD among the three solvents because themost dramatic discharge was formed using H2O as the spray sol-vent. The discharge was characterized by the luminescence recor-ded by a photomultiplier (Fig. S10) [34,40]. However, the solventalso affects the products of the EC-induced oxidation. While Ang II0

in H2Owas oxidized to dehydrogenation and oxygenation products,Ang II0 in CH3OH was oxidized to Schiff-base adducts with m/z522.8 (Fig. 5b) [41,42]. All these products were identified in thehigh-resolution MS and MS2 spectra (Fig. S11). However, Ang II0 inCH3OH/H2O (v/v, 1:1) was barely oxidized, which may be due to thedifference using Gibbs energy for the electron-transfer reactionbetween the electrode and the different solvents [43,44]. Similarphenomena were observed for melittin (a peptide) (Fig. S12). Theabove results indicated that the solvent-dependent oxidation fea-tures could possibly serve to differentiate CD-from EC-inducedprotein/peptide oxidation.

Fig. 5. Effect of solvent on (a) CD- and (b) EC-induced Ang II’ (2.5 mgmL�1 with 5mM NH4AH2O (v/v, 1/1), and H2O, respectively, in the measurement of CD-induced oxidation. The spra

3.3. Differentiation of CD- and EC-induced protein oxidation duringcommercial ESI source with the time-resolved method

There is a 50-60 cm-long fused silica capillary insulating theelectrode where EC reaction occurs and the spray tip where CDreaction occurs, with the homemade ESI source. However, thesample solution is sprayed from a metal capillary directly, duringwhich the EC and CD reactions occur almost simultaneously, with acommercial HESI source. Therefore, we also tried the applicabilityof the time-resolved method to a commercial HESI configuration.Because the signal of Ang II0 aqueous solution was severely sup-pressed by CD with a commercial HESI source, the oxidation peaksof Ang II0 could hardly be observed. Thus, we chose Hb and Cyt c asthe model proteins. The oxidation products of both aHb and bHbare oxygenation species (Fig. S13), and Iox/I is defined as:

Iox; aHb ¼Xi¼23

i¼13

I½aHbþO� (2)

Iox; bHb ¼Xi¼23

i¼14

�I½bHbþO� þ I½bHbþ2O�

�(3)

IoxI¼ Iox

IHb þ Iox(4)

where i is the charge state. The definition of Iox/I for Cyt c is referredto the Supporting Information.

First, we subjected an Hb solution to continuous electrospray for4min at a flow rate of 2 mLmin�1 at different spray voltages withthe unmodified HESI source. The HESI source is equipped with ametal spray emitter (length 148mm, i.d. 100 mm) on which thespray voltagewas applied. The upstreammetal union, isolated withthe metal emitter by a 14mm-long PEEK tubing (i.d. 125 mm), wasgrounded to prevent people from high voltage attack (Fig. S14a).The sample solution was delivered through an injection syringe(the same injection syringe as that in homemade ESI source) by thebuild-in injection pump of the mass spectrometer. The oxidationextents of both the a and b chains of Hb increased with a constantspray voltage of 3 kV or 5.5 kV (Fig. 6a, left and middle figure).Because the b chain of Hb contains more easily oxidized aminoacids than the a chain (1 Met vs 1 Cys and 3 Met for bHb), it wasmore severely oxidized [28]. CD-induced oxidation that wasobserved at the very beginning of the spray was more severe at

c) oxidation. The spray voltages are 6 kV, 6 kV, and 4 kV for solvents of CH3OH, CH3OH/y voltages are 3 kV in all solvent systems in the measurements of EC-induced oxidation.

Fig. 6. Oxidation curves of the two chains of Hb [5 mgmL�1 in CH3OH/H2O (v/v, 1:1) with 10mM NH4Ac and 0.5% FA] under different spray voltage conditions in a commercial HESIsource (a) with upstream grounding and (b) without upstream grounding. Flow rate¼ 2 mLmin�1.

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e67 65

higher spray voltage. The increased oxidation of Hbwith spray timeshould be attributed to EC for the longer contact time between Hbsolution and the spray emitter (acted as the electrode here) whenthe sample solution passed through it. The oxidation ratios stabi-lized (46% for bHb and 35% for aHb at 3 kV; 65% for bHb and 51% foraHb at 5.5 kV) after approximate 1.5min. Though it took 0.58min(theoretically calculated) for sample solution to pass through thewhole metal emitter, the much longer time (1.5min, which wasinversely proportional to the flow rate as shown in Fig. S15) forstabilization of the oxidation ratio might be related with the deadvolume of the adjustable union used to fit the metal emitter (insetof Fig. S14a). When the spray voltage was increased from 3 kV to5.5 kV during the spray process (0.3e0.8min), the increased CD-induced oxidations were immediately superposed on the regularoxidation curves. Then, the oxidation ratios of both the a and b

chains decreased after the maximum (from 52% to 46% for bHb andfrom 39% to 36% for aHb) (Fig. 6a, right figure). This is due to thatthe fresh solution encountered a 3 kV spray voltage, which resultedin less EC-induced oxidation. Based on the above results, both CDand EC play a role in Hb oxidation in a conventional HESI sourcewithout any modification.

Upstream grounding constructs an external loop whichstrengthens the electrochemical reaction of an ESI source, based onwhich Konermann utilized it to enhance the signal intensities ofsome hardly ionizable species [27] and observed protein unfoldingdue to electrochemically induced pH changes [25]. Then, weinvestigated Hb oxidation in the HESI source after remove of theupstream grounding (Fig. S14b). The spray current decreased from52 mA to 1 mA when the metal union was removed from the up-stream grounding union holder, which also indicated the con-struction of an external loopwith themetal union on. The oxidationratios hardly changed during the spray process, but elevation ofspray voltage strengthened the oxidation immediately (Fig. 6b).

That means CD dominated Hb oxidation without upstreamgrounding. The same results were obtained for Cyt c with andwithout upstream grounding (Figs. S16eS17). The results are inaccordance with Konermann's that CD accounts for protein oxida-tion under typical ESI operating conditions (without upstreamgrounding) [28]. Liu et al. observed dramatic steroid sulfate andpeptide oxidation during a conventional LC-MS [24]. By artificiallygrounding between the analytical column and the spray emitter,the oxidation was markly decreased. Kim et al. also observed pro-tein oxidation in LC-MS, whichwas decreased by rearranging the LCplumbing [26]. All of these might be related with the upstreamgrounding loop that can strength the electrochemical reaction of anESI source.

Through adjusting experimental parameters, CD- and EC-induced protein/peptide oxidation can be efficiently distin-guished, and the oxidation ratios can be read from the oxidationcurves directly. The oxidations can be strengthened or reducedaccording to our requirements by adjusting the experimental pa-rameters that are influential to CD or EC. For example, we can in-crease CD-induced oxidation for proteomics investigations byelevating the spray voltage, decreasing the flow rate or using wateras solvent. Oppositely, we can decrease EC-induced oxidation forin vivo protein/peptide analysis in conventional LC-MS by removingthe unexpected upstream grounding loop.

4. Conclusion

In summary, a time-resolved method was developed to distin-guish CD-from EC-induced protein/peptide oxidation. Since the CDreaction occurs near the Taylor cone, an adjustment of the sprayvoltage or flow rate has an instant effect on the oxidation. However,the EC reaction is closely associated with the solution/electrodecontact. Variations in the spray voltage or flow rate have a

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e6766

hysteresis effect on the oxidation curve. The delay time depends onthe length of the fused silica capillary in the homemade ESI setup.Therefore, the CD- and EC-induced oxidation can be distinguishedby observing the mass spectral features that differ upon changingthe spray voltage or flow rate. In addition, the analyte oxidation isclosely related to the solvent composition. The spray solvent-dependent oxidation features can also be used to differentiateCD- and EC-induced protein/peptide oxidation, which will be moreimportant for native MS, in which no organic solvent is used andthe CD-induced oxidation is much more severe. Overall, this studycan potentially contribute to a better understanding of proteinoxidation during ESI-MS. The proportion of CD- and EC-inducedoxidation can be read from the oxidation curves, and the unex-pected EC-induced protein/peptide oxidation in HESI or LC-MS canbe reduced by rearranging the instrument configuration.

Acknowledgements

The authors are grateful for the financial support from the Na-tional Key Research and Development Program of China(2016YFA0201300-2), the National Natural Science Foundation ofChina (21665003, 21475121), the Guangxi Natural Science FundProject (No. 2016GXNSFBA380140), and the China PostdoctoralScience Foundation (2017M612858).

Appendix B. Supplementary data

Supplementary data related to this article can be found athttps://doi.org/10.1016/j.aca.2018.01.025.

References

[1] R. Sultana, M. Perluigi, D.A. Butterfield, Protein oxidation and lipid peroxida-tion in brain of subjects with Alzheimer's disease: insights into mechanism ofneuro degeneration from redox proteomics, Antioxidants Redox Signal. 8(2006) 2021e2037.

[2] S. Reeg, T. Grune, Protein oxidation in aging: does it play a role in agingprogression? Antioxidants Redox Signal. 23 (2015) 239e255.

[3] M.T. Lin, M.F. Beal, Mitochondrial dysfunction and oxidative stress in neuro-degenerative diseases, Nature 443 (2006) 787e795.

[4] P. Attri, N. Kumar, J.H. Park, D.K. Yadav, S. Choi, H.S. Uhm, I.T. Kim, E.H. Choi,W. Lee, Influence of reactive species on the modification of biomoleculesgenerated from the soft plasma, Sci. Rep. 5 (2015) 8221.

[5] A. van Dam, J.C.V. Schwarz, J. de Vos, M. Siebes, T. Sijen, T.G. van Leeuwen,M.C.G. Aalders, S.A.G. Lambrechts, Oxidation monitoring by fluorescencespectroscopy reveals the age of fingermarks, Angew. Chem. Int. Ed. 53 (2014)6272e6275.

[6] G.Z. Qin, X.H. Meng, Q. Wang, S.P. Tian, Oxidative damage of mitochondrialproteins contributes to fruit senescence: a redox proteomics analysis,J. Proteome Res. 8 (2009) 2449e2462.

[7] T. Senda, M. Senda, S. Kimura, T. Ishida, Redox control of protein conformationin flavoproteins, Antioxidants Redox Signal. 11 (2009) 1741e1766.

[8] I. Torres-Cuevas, J. Kuligowski, M. C�arcel, C. Ch�afer-Peric�as, M. Asensi,R. Solberg, E. Cubells, A. Nu~nez, O.D. Saugstad, M. Vento, J. Escobar, Protein-bound tyrosine oxidation, nitration and chlorination by-products assessed byultraperformance liquid chromatography coupled to tandem mass spec-trometry, Anal. Chim. Acta 913 (2016) 104e110.

[9] A. Sinz, Chemical cross-linking and mass spectrometry to map three-dimensional protein structures and protein-protein interactions, Mass Spec-trom. Rev. 25 (2006) 663e682.

[10] J.B. Fenn, M. Mann, C.K. Meng, S.F. Wong, C.M. Whitehouse, Electrosprayionization for mass-spectrometry of large biomolecules, Science 246 (1989)64e71.

[11] J. Chen, D.L. Rempel, B.C. Gau, M.L. Gross, Fast photochemical oxidation ofproteins and mass spectrometry follow submillisecond protein folding at theamino-acid level, J. Am. Chem. Soc. 134 (2012) 18724e18731.

[12] J.J. Mo, Q.R. Yan, C.K. So, T. Soden, M.J. Lewis, P. Hu, Understanding the impactof methionine oxidation on the biological functions of igg1 antibodies usinghydrogen/deuterium exchange mass spectrometry, Anal. Chem. 88 (2016)9495e9502.

[13] M. Abonnenc, L. Qiao, B.H. Liu, H.H. Girault, Electrochemical aspects of elec-trospray and laser desorption/ionization for mass spectrometry, Annu. Rev.Anal. Chem. 3 (2010) 231e254.

[14] S.D. Maleknia, K.M. Downard, Advances in radical probe mass spectrometryfor protein footprinting in chemical biology applications, Chem. Soc. Rev. 43

(2014) 3244e3258.[15] G.J. Van Berkel, F.M. Zhou, Characterization of an electrospray ion source as a

controlled-current electrolytic cell, Anal. Chem. 67 (1995) 2916e2923.[16] G.Z. Xu, M.R. Chance, Hydroxyl radical-mediated modification of proteins as

probes for structural proteomics, Chem. Rev. 107 (2007) 3514e3543.[17] L.W. Wang, M.R. Chance, Protein footprinting comes of age: mass spectrom-

etry for biophysical structure assessment, Mol. Cell. Proteomics 16 (2017)706e716.

[18] J.A. Loo, R.R.O. Loo, H.R. Udseth, C.G. Edmonds, R.D. Smith, Solvent-inducedconformational changes of polypeptides probed by electrospray-ionizationmass spectrometry, Rapid Commun, Mass Spectrom. 5 (1991) 101e105.

[19] H. Metwally, R.G. McAllister, V. Popa, L. Konermann, Mechanism of proteinsupercharging by sulfolane and m-nitrobenzyl alcohol: molecular dynamicssimulations of the electrospray process, Anal. Chem. 88 (2016) 5345e5354.

[20] C.A. Cassou, H.J. Sterling, A.C. Susa, E.R. Williams, Electrothermal super-charging in mass spectrometry and tandem mass spectrometry of nativeproteins, Anal. Chem. 85 (2013) 138e146.

[21] H.X. Wang, G. Yong, S.L. Brown, H.E. Lee, M.A. Zenaidee, C.T. Supuran,W.A. Donald, Supercharging protein ions in native mass spectrometry usingtheta capillary nanoelectrospray ionization mass spectrometry and cyclicalkylcarbonates, Anal. Chim. Acta 1003 (2018) 1e9.

[22] K. Morand, G. Talbo, M. Mann, Oxidation of peptides during electrosprayionization, Rapid Commun, Mass Spectrom. 7 (1993) 738e743.

[23] M.L. Chen, K.D. Cook, Oxidation artifacts in the electrospray mass spectrom-etry of Ab peptide, Anal. Chem. 79 (2007) 2031e2036.

[24] S. Liu, W.J. Griffiths, J. Sj€ovall, On-column electrochemical reactions accom-panying the electrospray process, Anal. Chem. 75 (2003) 1022e1030.

[25] L. Konermann, E.A. Silva, O.F. Sogbein, Electrochemically induced pH changesresulting in protein unfolding in the ion source of an electrospray massspectrometer, Anal. Chem. 73 (2001) 4836e4844.

[26] K. Kim, P.D. Compton, T.K. Toby, P.M. Thomas, J.T. Wilkins, R.K. Mutharasan,N.L. Kelleher, Reducing protein oxidation in low-flow electrospray enablesdeeper investigation of proteoforms by top down proteomics, EuPA OpenProteom. 8 (2015) 40e47.

[27] R.A. Ochran, L. Konermann, Effects of ground loop currents on signal in-tensities in electrospray mass spectrometry, J. Am. Soc. Mass Spectrom. 15(2004) 1748e1754.

[28] B.L. Boys, M.C. Kuprowski, J.J. No€el, L. Konermann, Protein oxidative modifi-cations during electrospray ionization: solution phase electrochemistry orcorona discharge-induced radical attack? Anal. Chem. 81 (2009) 4027e4034.

[29] J. Pei, X. Zhou, X. Wang, G. Huang, Alleviation of electrochemical oxidation forpeptides and proteins in electrospray ionization: obtaining more accuratemass spectra with induced high voltage, Anal. Chem. 87 (2015) 2727e2733.

[30] R.T. Kelly, J.S. Page, Q.Z. Luo, R.J. Moore, D.J. Orton, K.Q. Tang, R.D. Smith,Chemically etched open tubular and monolithic emitters for nanoelectrosprayionization mass spectrometry, Anal. Chem. 78 (2006) 7796e7801.

[31] K. Saito, A.W. Rutherford, H. Ishikita, Mechanism of proton-coupled quinonereduction in photosystem II, Proc. Natl. Acad. Sci. U. S. A 110 (2013) 954e959.

[32] L. Hammarstr€om, S. Styring, Proton-coupled electron transfer of tyrosines inphotosystem II and model systems for artificial photosynthesis: the role of aredox-active link between catalyst and photosensitizer, Energy Environ. Sci. 4(2011) 2379e2388.

[33] B.A. Barry, J. Chen, J. Keough, D. Jenson, A. Offenbacher, C. Pagba, Protoncoupled electron transfer and redox active tyrosines: structure and function ofthe tyrosyl radicals in ribonucleotide reductase and photosystem II, J. Phys.Chem. Lett. 3 (2012) 543e554.

[34] I.L. Kanev, A.Y. Mikheev, Y.M. Shlyapnikov, E.A. Shlyapnikova, T.Y. Morozova,V.N. Morozov, Are reactive oxygen species generated in electrospray at lowcurrents? Anal. Chem. 86 (2014) 1511e1517.

[35] K.P. Bateman, Electrochemical properties of capillary electrophoresis-nanoelectrospray mass spectrometry, J. Am. Soc. Mass Spectrom. 10 (1999)309e317.

[36] J.P.W. Stark, M.S. Alexander, K.L. Smith, Electrospray pulsation: a diagnostic tounderstand cone-jet stability and minimum flow, J. Appl. Phys. 115 (2014)184e186.

[37] P.J. McClory, K. Hakansson, Corona discharge suppression in negative ionmode nanoelectrospray ionization via trifluoroethanol addition, Anal. Chem.89 (2017) 10188e10193.

[38] A. Dey, F.E.J. Jenney, M.W.W. Adams, E. Babini, Y. Takahashi, K. Fukuyama,K.O. Hodgson, B. Hedman, E.I. Solomon, Solvent tuning of electrochemicalpotentials in the active sites of HiPIP versus ferredoxin, Science 318 (2007)1464e1468.

[39] X.J. Wang, L. Wang, J.J. Wang, T. Chen, Solvent effects on electrochemicalbehavior of poly(ferrocenylsilane) films, Electrochim. Acta 52 (2007)3941e3949.

[40] G.Y. Li, J.Y. Pei, Y. Yin, G.M. Huang, Direct sequencing of a disulfide-linkedpeptide with electrospray ionization tandem mass spectrometry, Analyst140 (2015) 2623e2627.

[41] B. Metz, G.F. Kersten, P. Hoogerhout, H.F. Brugghe, H.A. Timmermans, A. deJong, H. Meiring, J. ten Hove, W.E. Hennink, D.J. Crommelin, W. Jiskoot,Identification of formaldehyde-induced modifications in proteins: reactionswith model peptides, J. Biol. Chem. 279 (2004) 6235e6243.

[42] S. Cheng, Q. Wu, H.D. Dewald, H. Chen, Online monitoring of methanolelectro-oxidation reactions by ambient mass spectrometry, J. Am. Soc. MassSpectrom. 28 (2016) 1e8.

J. Pei et al. / Analytica Chimica Acta 1011 (2018) 59e67 67

[43] P.A. Zagrebin, R. Buchner, R.R. Nazmutdinov, G.A. Tsirlina, Dynamic solventeffects in electrochemical kinetics: indications for a switch of the relevantsolvent mode, J. Phys. Chem. B 114 (2010) 311e320.

[44] I.M. Aldous, L.J. Hardwick, Solvent-mediated control of the electrochemicaldischarge products of non-aqueous sodium-oxygen electrochemistry, Angew.Chem. Int. Ed. 55 (2016) 8254e8257.