Unexpected peaks in tandem mass spectra due to reaction of ...

Research Article

Received: 22 August 2014 Revised: 12 September 2014 Accepted: 15 September 2014 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2014, 28, 2645–2660

Unexpected peaks in tandem mass spectra due to reaction ofproduct ions with residual water in mass spectrometercollision cells

Pedatsur Neta*, Mahnaz Farahani†, Yamil Simón-Manso, Yuxue Liang, Xiaoyu Yangand Stephen E. SteinBiomolecular Measurement Division, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA

RATIONALE: Certain product ions in electrospray ionization tandem mass spectrometry are found to react with residualwater in the collision cell. This reaction often leads to the formation of ions that cannot be formed directly from theprecursor ions, and this complicates the mass spectra and may distort MRM (multiple reaction monitoring) results.METHODS: Various drugs, pesticides, metabolites, and other compounds were dissolved in acetonitrile/water/formicacid and studied by electrospray ionization mass spectrometry to record their MS2 and MSn spectra in several massspectrometers (QqQ, QTOF, IT, and Orbitrap HCD). Certain product ions were found to react with residual water incollision cells. The reaction was confirmed by MSn studies and the rate of reaction was determined in the IT instrumentusing zero collision energy and variable activation times.RESULTS: Examples of product ions reactingwithwater include phenyl and certain substituted phenyl cations, benzoyl-typecations formed from protonated folic acid and similar compounds by loss of the glutamate moiety, product ions formedfrom protonated cyclic siloxanes by loss of methane, product ions formed from organic phosphates, and certain negativeions. The reactions of product ions with residual water varied greatly in their rate constant and in the extent of reaction(due to isomerization).CONCLUSIONS: Various types of product ions react with residual water in mass spectrometer collision cells. As a result,tandem mass spectra may contain unexplained peaks and MRM results may be distorted by the occurrence of suchreactions. These often unavoidable reactions must be taken into account when annotating peaks in tandem mass spectraand when interpreting MRM results. Published in 2014. This article is a U.S. Government work and is in the publicdomain in the USA.

(wileyonlinelibrary.com) DOI: 10.1002/rcm.7055

In the course of expanding the NIST library of MS/MSspectra[1] for use in metabolomics, we recorded the spectraof many precursor ions using electrospray ionization inconjunction with several types of mass spectrometers. As forthe NIST/EPA/NIH EI Library, collected spectra aremanually evaluated[2] before they are added to the library.This quality control process often involves comparison ofspectra from different mass spectrometers and assignmentof each peak in the spectrum to a reasonable product ion.Peak assignment is further enhanced by the use ofinstruments with ppm-level mass accuracy. In recentmeasurements we noted the presence of unexpectedproduct ions arising from water adduction and undertooka more detailed study of this fragmentation process.Herein we report our findings on several groups of

* Correspondence to: P. Neta, Biomolecular MeasurementDivision, National Institute of Standards and Technology,Gaithersburg, MA 20899, USA.E-mail: [email protected]

† Present address: Office of Generic Drugs, Food and DrugsAdministration, Silver Spring, MD 20993, USA.

Published in 2014. This article is a U.S. Governmen


264

compounds, some of which are drugs, pesticides, ormetabolites, and some are simple compounds studied togain mechanistic insights.

Earlier mass spectrometric studies of protonated guanosineshowed unexpected peaks in the tandem mass spectra andrevealed that certain product ions react with water ormethanol in the collision cell.[3,4] Subsequent studies withseveral other protonated molecules also detected addition ofwater, alcohol, or acetonitrile to certain product ions.[5–8]

Similar water addition reactions were also reported forproduct ions from sodiated benzene dicarboxylate salts[9]

and some sodiated chalcones.[10] Some negatively chargedproduct ions were also found to attach a water molecule.For example, collision-induced dissociation of deprotonatedguanine led to formation of certain product ions which thenreacted with water[11] and the iodide ion formed fromiodobenzoate anion was detected as the water adduct.[12]

These findings provide possible explanations forunexpected product ions in tandem mass spectra and suggestcareful considerations in interpretation of multiple reactionmonitoring (MRM) results. For example, studies ofprotonated quinolone drugs raised questions about thevalidity of MRM results[13] which we then explained byfinding a reversible water loss and water addition reaction

t work and is in the public domain in the USA.

5

P. Neta et al.

2646

while the loss of CO2 was irreversible.[14] More recently, wefound that certain protonated aldehydes undergo loss of H2

to form ketene cations and that these cations can react withwater to produce the protonated carboxylic acids.[15]

Therefore, the MS2 spectrum of the protonated aldehydemay contain the peak of the corresponding protonatedcarboxylic acid and its fragmentation products, again raisingconcern in interpretation of MRM results.Water appears to be present in varying concentrations in

the collision cells of most electrospray ionization massspectrometers. Its reaction with product ions appeared morepronounced with increasing collision gas pressure.[15] It isalso possible that water is adsorbed and desorbed from metalsurfaces in the mass spectrometer. Experiments using D2Oinstead of H2O in the solvent mixture showed that the waterreacting with product ions does not originate in the solventused in the electrospray.[15] Since water is a common liquidchromatography (LC) fluid, its precise origin is notnecessarily clear. In any case, since attachment of water wasdetected in product ions of various compounds, weembarked on a search for guidelines to anticipate theoccurrence of such reactions. We studied folic acid and similarcompounds, which form aroyl cations and their wateradducts, other compounds forming aroyl cations, compoundsforming aryl cations, compounds forming cations withpositive charge on P, Si, or Sn, and some compounds formingnegative ions. The rate and extent of reaction of product ionswith water are also presented.

EXPERIMENTALa

The compounds were obtained from various sources forinclusion in the NIST MS/MS library. The latest release of thislibrary (2014) includes many of the MS2 and MSn spectradiscussed in this paper. The compounds were dissolved inacetonitrile/water/formic acid (50:50:0.1), and in some casesthe acetonitrile was replaced with methanol or propanol.For initial studies, electrospray ionization mass spectrometrywas carried out with a Micromass (Waters Corp., Milford,MA, USA) Quattro Micro triple quadrupole instrument(QqQ). First the mass spectra were recorded at different conevoltages to optimize the abundances of the precursor ions.Then, the precursor ion at the optimal cone voltage wasselected for fragmentation in the collision cell, with 0.21 Pa(1.6 mTorr) argon as collision gas, and the MS/MS spectrumwas recorded at 20 different collision voltages. The range ofcollision voltage spanned from near 0 V up to a value whereno precursor ion remained. Pseudo-MS3 spectra weremeasured by using a high cone voltage to produce thefragment ion in the cone region and this ion was then selectedfor MS/MS measurement as above. Spectra were acquired in’centroid’ mode, whereby signals within each individual timeinterval in a given spectrum were centered and integrated

aCertain commercial equipment, instruments, or materials areidentified in this document. Such identification does notimply recommendation or endorsement by the NationalInstitute of Standards and Technology, nor does it imply thatthe products identified are necessarily the best available forthe purpose.

wileyonlinelibrary.com/journal/rcmPublished in 2014. This article is a U.S. Governmen

by the instrument data system. Typically, m/z values werewithin 0.2m/z units of the theoretical m/z values throughoutthe m/z range of interest.

To examine the influence of type of collisional excitation,MS/MS spectra of the ions were also measured by ion-trap(IT) fragmentation (LTQ, Thermo Fisher Scientific, Waltham,MA, USA) with 0.35 mTorr helium as collision gas, usinganother beam-type collision cell (a quadrupole time-of-flightinstrument (model 6530; Agilent Technologies, Santa Clara,CA, USA) with 0.02 mTorr N2, and using HCD (higher-energy C-trap dissociation) in an orbital ion trap (OIT)instrument (Orbitrap Elite, Thermo Fisher Scientific,Waltham, MA, USA) with 0.5 Pa (3.75 mTorr) N2. All gasesused in the mass spectrometer collision cells were of ultrahighpurity grade (99.999%). Both MS2 and authentic MS3 and MS4

spectra were obtained with the LTQ, and this instrument wasalso used to determine the rate and extent of reaction of thevarious product ions with water as outlined in the secondsection of the results. While the QqQ and IT instrumentsprovide low-resolution spectra, the OIT instrument providesHCD spectra with high accuracy m/z values, which help inconfirming peak assignments. The accuracy of the m/z valuesis reflected in the annotation of the peaks from the variousinstruments; only the HCD spectra are annotated with foursignificant figures. The spectra presented in the figures arethe average of 20 to 100 individual spectra. Noise peaks areremoved if they appear in <20% of the individual spectra.Further details of quality control procedures are summarizedin a recent publication.[16]

RESULTS AND DISCUSSION

During measurement of MS/MS spectra of many compoundsfor inclusion in the NIST tandem mass spectral library, andwhile attempting to annotate all the product ion peaksobtained with high mass accuracy instruments, we cameacross a number of peaks that could be clearly ascribed toproducts of reaction with water. Representative compoundsexhibiting this behavior were studied in more detail forinclusion in this paper and spectra of other compounds areincluded in the MS/MS library. The spectral observationsare discussed below and the rates of reaction with water willbe discussed in the subsequent section.

Spectral observations

Folic acid and related compounds

Folic acid (1), its dihydro (2) and tetrahydro (3) derivatives,and the drugs methotrexate (5) and raltitrexed (6) (Table 1)have a glutamic acid residue attached to an aroyl group.Tandem mass spectra show that the [M+H]+ ions of thesecompounds undergo loss of the glutamic acid residue(147 Da) to form the product ion [M+H–147]+ (or [M+H–C5H9NO4]

+). Representative spectra obtained with folic andtetrahydrofolic acids are shown in Figs. 1 and 2. Thesespectra, as well as the spectra of other compounds listed inTable 1, show that, in addition to the [M+H–147]+ production, a peak corresponding to [M+H–129]+ is observed. Theabundance of this ion varies with collision energy and isdependent on instrument conditions. It is unclear how this



Table 1. Folic acid and related compounds, their precursor ions (p = [M+H]+), and main MS2 peaksa

Compound Structurem/z

[M+H]+m/z

[M+H–Glu]+m/z

[M+H–Glu+H2O]+

1 Folic acid 442 295 313

2 Dihydrofolicacid

444 297 315

3 Tetrahydrofolicacid

446 299 317

4 Methopterin 456 309 327

5 Methotrexate 455 308 326

6 Raltitrexed 459 312 330

7 N-(4-Aminobenzoyl)-L-glutamic acid

267 120 138

(Continues)

Reactions of product ions with water


Published in 2014. This article is a U.S. Government work and is in the public domain in the USA.

wileyonlinelibrary.com/journal/rcm

2647

Table 1. (Continued)


[M+H]+m/z

[M+H–Glu]+m/z

[M+H–Glu+H2O]+

8 Rimonabant 463 363 (M+H-C5H12N2) 381

aThe spectra were recorded with the QqQ, IT, and Orbitrap mass spectrometers. The latter instrument provides high-resolution spectra, which help confirm peak annotation. The m/z values in the table are the nominal values, observed in allthe instruments. Column 4 lists the m/z values of the precursor ion [M+H]+, column 5 lists the m/z values of the production formed by loss of the glutamate residue (or the C5H10NNH2 residue for entry #8), and the last column lists the m/zvalues of the product of reaction with water. Peaks observed at higher collision energies are not listed.

Figure 1. Electrospray ionization MS/MS spectra obtainedwith protonated folic acid: (a) high-resolution HCD spectrum,collision voltage 8 V; (b) low-resolution IT spectrum,normalized collision energy 35%. The precursor [M+H]+ ionat m/z 442 undergoes complete fragmentation in the iontrap and is not visible in spectrum (b); it is seen in theHCD spectrum (a) and its intensity varies with collisionvoltage. Loss of a glutamic acid residue from the precursorion yields the m/z 295 ion, which reacts with H2O to yieldthe m/z 313 ion.

Figure 2. Electrospray ionization MS/MS spectra obtainedwith protonated tetrahydrofolic acid: (a) high-resolutionHCD spectrum, collision voltage 13 V; (b) low-resolution ITspectrum, normalized collision energy 35%. The precursor[M+H]+ ion at m/z 446 undergoes complete fragmentationin the ion trap and is not visible in spectrum (b); it is seen inthe HCD spectrum (a) and its intensity varies with collisionvoltage. Loss of glutamic acid residue from the precursorion yields the m/z 299 ion, which reacts with H2O to yieldthe m/z 317 ion.

P. Neta et al.

2648

product ion can be formed directly from the precursor ion.Exact mass measurements show that the m/z differencebetween the two product ions corresponds to the massof a water molecule. A mechanism that can account for the[M+H–129]+ ion is the addition of H2O to the [M+H–147]+


ion. The simpler analogous compound 7, N-(4-aminobenzoyl)-L-glutamic acid, exhibits a similar behavior. These findingssuggest that the acylium ion formed by loss of the glutamicacid residue may react with residual water in the collision cellof the mass spectrometer, as observed for the acylium ions




produced from protonated 3-formylchromone.[15] We testedthis hypothesis by studying simpler acylium ions producedby losses of small molecules such as NH3 or amines.

Aroyl cations

The MS/MS spectra (Fig. 3) recorded with protonatedbenzamide (m/z 122) show loss of NH3 to produce thebenzoyl cation (m/z 105), but no reaction of this cation withH2O was detected. Instead, this ion undergoes loss of CO toproduce a phenyl cation (m/z 77), which then reacts withwater (to give the m/z 95 ion). Similar behavior was observedwith the benzoyl cation produced from methyl hippurate (10)and mebendazole (11), the 4-hydroxybenzoyl cation (m/z 121)from nifuroxazide (12), the 4-chlorobenzoyl cation (m/z 139)from moclobemide (13), and the 2-aminobenzoyl cation

Figure 3. Electrospray ionization MS/MS spectra obtainedwith protonated benzamide: (a) high-resolution HCDspectrum, collision voltage 21 V; (b) HCD spectrum of them/z 105 ion, collision voltage 15 V; and (c) low-resolution ITspectrum, normalized collision energy 35%. The precursor[M+H]+ ion has m/z 122. Ion trap CID (c) shows only lossof NH3 (m/z 105) and loss of CONH (m/z 79). Higher energyC-trap dissociation (a) shows also the secondary loss of COfrom the m/z 105 product ion to form the m/z 77 peak,corresponding to the phenyl cation, and addition of H2O tothis product gives the m/z 95 ion. The product ions of m/z 77and m/z 95 are more intense in the HCD MS3 spectrum (b).



264

(m/z 120) from 2-aminobenzamide (Table 2). In all these cases,the acylium ions do not react with water but undergo loss ofCO and the resulting phenyl cations do react with water.

Protonated N-methylphthalimide (15) (m/z 162) undergoesloss of CH2 =NH to form the m/z 133 ion, which is probablya benzoyl cation with an ortho-formyl group (OCHC6H4CO

+).This product undergoes two successive losses of CO groupsto form the phenyl cation which then reacts with H2O, areaction that is detected only in MS3 experiments.

Replacing benzene with pyridine introduces an additionalfactor in the fragmentation route of [M+H]+, i.e. whether theproton remains on the pyridine ring or is removed within thelost neutral molecule. Experiments with nicotinuric acid (16),nifenazone (17), and N,N-diethylisonicotinamide (18) showthat the pyridyl ion remaining after several neutral losses canbe observed in three forms: protonated pyridine (C5H5NH+,m/z 80), protonated pyridyl (C5H4NH+ or C5H5N

+, m/z 79),and the pyridyl cation (C5H4N

+, m/z 78). Only the latter reactswith water, but it is the least abundant of the three ions.Therefore, the reaction with water has negligible significancewith pyridine derivatives compared with benzene derivatives.

Comparison of simple aroyl cations with those from folic acidand derivatives

If the simple benzoyl and related cations in Table 2 do notreact with water, it is puzzling that the MS/MS spectra ofthe compounds in Table 1 show product ions correspondingto the benzoyl-type cations produced by loss of a glutamicacid residue as well as the products of addition of water tothese ions. It may be speculated that the latter product ionsare produced not by intermolecular reactions of the benzoylcations with residual water in the collision cell, but ratherby an intramolecular reaction, whereby a carboxyl group ofglutamic acid concomitantly inserts its OH group at the siteof the amide bond during its dissociation. This pathway issupported by the following findings. The MS2 spectrum ofprotonated folic acid (m/z 442) recorded in the IT massspectrometer shows formation of the [M+H–147]+ (m/z 295)as well as the [M+H–129]+ ion (m/z 313), the latter withabundances of ~8%. The same precursor ion also undergoesloss of water to produce the m/z 424 ion (with lowabundance). When the MS3 spectrum of the m/z 424 productions is recorded with the IT instrument, the major product isthe m/z 295 ion, while the m/z 313 ion was barely observable.Moreover, the MS3 spectrum of the m/z 295 ion recordedin the IT instrument shows no observable product ion withm/z 313. Parallel results were obtained with dihydrofolic acid.These findings suggest that the glutamic acid residue may beresponsible for the production of the [M+H–129]+ ion by anintramolecular mechanism, and, when the glutamyl residueundergoes loss of water, it can no longer participate in sucha process. This mechanism of formation of the [M+H–129]+

ion via intramolecular reaction with the glutamyl residueresembles the mechanism of formation of [bn-1 +H2O]+ ionsin the fragmentation of protonated peptides.[17,18] Onthe other hand, spectra recorded with the QqQ massspectrometer show somewhat different results. When them/z 295 product ion from folic acid is produced in the ionsource (by increasing ’cone’ voltage) and then selected intothe collision cell to obtain a pseudo-MS3 spectrum, anabundant product ion at m/z 313 is observed at low collision



9

Table 2. Compounds forming aroyl cations, their precursor ions (p= [M+H]+), and their MS2 peaks involving addition of watera


[M+H]+m/z

[ArCO]+ [Ar]+m/z[Ar]+

m/z[Ar +H2O]+

9 Benzamide 122 105 C6H5+ 77 95

10 Methyl hippurate 194 105 C6H5+ 77 95

11 Mebendazole 296 105 C6H5+ 77 95

12 Nifuroxazide 276 121 HOC6H4+ 93 111

13 Moclobemide 269 139 ClC6H4+ 111 129

14 2-Aminobenzamide 137 120 H2NC6H4+ 92 110

15 N-Methyl phthalimide 162 133 -> 105 C6H5+ 77 95

16 Nicotinuric acid 181106 C5H4N

+ 78 96107 C5H5N

+ 79 -108 C5H5NH+ 80 -

17 Nifenazone 309 106 C5H4N+ 78 96

107 C5H5N+ 79 -

(Continues)

P. Neta et al.

wileyonlinelibrary.com/journal/rcmPublished in 2014. This article is a U.S. Government work and is in the public domain in the USA.


2650

Figure 4. Electrospray ionization MS/MS spectra obtainedwith protonated rimonabant: (a) high-resolution HCDspectrum, collision voltage 32 V and (b) low-resolution ITspectrum, normalized collision energy 35%. The precursor[M+H]+ ion at m/z 463 undergoes complete fragmentationin the ion trap and is not visible in spectrum (b); it is seen inthe HCD spectrum (a) and its intensity varies with collisionvoltage. Loss of C5H12N2 from the precursor ion yields them/z 363 ion, which reacts with H2O to give the m/z 381 ion.



[M+H]+m/z

[ArCO]+ [Ar]+m/z[Ar]+

m/z[Ar +H2O]+

18 N,N-Diethyl isonicotinamide 179106 C5H4N

+ 78 96107 C5H5N

+ 79 -108 C5H5NH+ 80 -

aSee footnote in Table 1. Column 4 lists the m/z values of the precursor ion [M+H]+, column 5 lists the m/z values of the aroylcations produced by loss of various residues, column 6 gives the structures of the product ions formed after loss of CO fromthe aroyl cations, column 7 lists their m/z values, and the last column lists the m/z values of the product of reaction with water.Other peaks observed at higher collision energies are not listed.


energies, which is clearly due to reaction of the aroyl cation[M+H–147]+ with water in the collision cell. The sameresults were obtained with dihydrofolic acid. These findingsindicate that the intramolecular mechanism is not theonly pathway for producing the [M+H–129]+ ions fromthese compounds and that different mechanisms may beoperative under different conditions. In the IT instrumentthe precursor or product ions are collected in the ion trapand then allowed to undergo CID under 0.35 mTorr of He.In the QqQ instrument the ions are produced in the ion sourceand immediately selected into the collision cell under 1.6mTorr of Ar. Thus, in the latter case the ions have moreimmediate access (microseconds vs milliseconds) to higherconcentrations of water (due to higher pressure in the cell)and can attach a water molecule quickly. In the IT, however,the ions may rearrange to a non-reactive form before theyencounter water molecules.Another factor affecting the probability of reaction of the

acylium ions with water is the stability of this ion towardsrearrangement or fragmentation. While the simple benzoylcation (Table 2) undergoes loss of CO in preference to attachinga water molecule, the more complex acylium ions fromtetrahydrofolic acid (Fig. 2) or rimonabant (Fig. 4) (Table 1)are more stable and can attach a water molecule with highefficiency. Thus, the reaction of product ions with water takesplace in competition with other reactions of these ions and,therefore, observation of the water addition products in MS2

spectra depends on the balance among all the reactionsinvolved. In MS3 spectra, addition of water may be observedat low collision energies whereas at higher energies the ionsundergo fragmentation. The IT instrument (LTQ) allows us totrap product ions at zero collision energy and at extendedactivation times so that we can monitor their reaction withwater in the absence of CID. The relative rates for the wateraddition reactions will be discussed in section 2 of the results.

265

Aryl cations

The results in Table 2 suggest that simple benzoyl-type ionsdo not react with water to any appreciable extent that canbe observed in the tandem mass spectra but rather undergorapid loss of CO to form the corresponding aryl ions. Theselatter ions do react with water to form product ions with



significant abundance in the MS3 spectra. To extend thesestudies on aryl ions we examined other compounds whose[M+H]+ ions may fragment by various pathways to formaryl cations (not via decarbonylation of aroyl cations).

Protonated dapsone (19), sulfabenzamide (20), andsulfamethoxazole (21) (Table 3) produce the 4-aminophenylcation (m/z 92) which reacts with water to yield the m/z 110ion. Similarly, protonated p-toluenesulfonic acid (22) losesH2O and SO2 to form the 4-methylphenyl cation (m/z 91)which then yields the m/z 109 ion by reaction with water.



1

Table 3. Compounds forming aryl cations, their precursor ions (p = [M+H]+), and their MS2 peaks involving additionof watera


[M+H]+ Product ionm/z[Ar]+

m/z[Ar +H2O]+

19 Dapsone 249 p-H2NC6H4SO2H 92 110

20 Sulfabenzamide 277 p-C6H5CONHSO2H 92 110

21 Sulfamethoxazole 254 p-C4H6N2O3S 92 110

22 p-Toluenesulfonic acid 173 p-H2SO3 91 109

23 2,6-Xylidine 122p-NH3 105 -

p-NH3-C2H4 77 95

24 2-Aminopyridine 95 p-NH3 78 96

25 2-Aminopyrimidine 96 p-NH3 79 97

26 2-Chloroquinoline 164 p-HCl 128 146

27 Crimidine 172 p-HCl 136 154

28 6-Mercaptopurine 153 p-H2S 119 137

29 Trigonelline 138 p-HCO2H 92 110

(Continues)

P. Neta et al.



2652



[M+H]+ Product ionm/z[Ar]+

m/z[Ar +H2O]+

30 Flumequine 262 p-H2O-C3H6 202 220,238

31 Meclizine 391 p-C12H18N2-HCl 165 183

32 Azamethiphos 325 p-C4H8NO5PS 112 130

aSee footnote in Table 1. Column 4 lists the m/z values of the precursor ion (p = [M+H]+), column 5 shows the overall neutralloss to produce the aryl (or heteroaryl) cation from p, column 6 lists the m/z values of the aryl cations, and the last column liststhe m/z values of the product of reaction with water. Other peaks observed at various collision energies are not listed.


265

Protonated 2,6-xylidine (23) (m/z 122) loses NH3 and theproduct ion (m/z 105) barely reacts with water. However,this product ion loses C2H4 to form a phenyl cation (m/z 77),which reacts with water. Although most phenyl cationsbearing various substituents react with water, there is asignificant effect of the substituent on the rate and extent ofreaction, which will be discussed in section 2.Protonated 2-aminopyridine and 2-aminopyrimidine (Table 3)

undergo loss of NH3 and the product ions react with water.Protonated 2-chloroquinoline and crimidine (27) undergo lossof HCl and the product ions react with water. Protonated6-mercaptopurine undergoes loss of H2S to form the m/z 119ion, which reacts with H2O to produce the m/z 137 ion. Aparallel reaction is the loss of HCN from the ring, but this isnot followed by reaction with water. These heterocyclic productions, which react with water, are similar to the phenyl cation inthat one of the ring carbons is trivalent (after losing its bondingto the N, Cl, or S atom), and this is the driving force for theirreaction with water.Protonated trigonelline (29) undergoes loss of CO2 to

form the m/z 94 ion and parallel loss of HCO2H to producethe m/z 92 ion; only the latter ion reacts with water to formthe m/z 110 ion rapidly and completely. The m/z 94 ionproduced by loss of CO2 is essentially theN-methylpyridiniumion, which is not expected to attach a water molecule. To formthe m/z 92 ion, two hydrogens have to be removed along with(or in addition to) the CO2. Although protonated aliphaticcarboxylic acids are known to undergo loss of HCO2H, it isnot clear how the m/z 92 ion is formed from protonatedtrigonelline. Nevertheless, its reaction with water is efficientand the m/z 110 ion produced by this reaction is highlyabundant in the MS2 spectrum.



Protonated flumequine (30) undergoes loss of H2O andthen loss of C3H6 to produce the m/z 202 ion, which reactswith two water molecules in succession to form the m/z 220ion and then the m/z 238 ion. Protonated meclizine (31)undergoes loss of C12H18N2 (the 3-methylbenzylpyrazinemoiety) and then loss of HCl to form the m/z 165 production. This ion can react with water to produce the m/z 183ion. The reaction occurs with high abundance in MS4

experiments, although MS2 spectra show only lowabundance of the m/z 183 ion. Protonated azamethiphos (32)undergoes loss of C4H8NO5PS to form a chloropyridyl cation,m/z 112, which reacts with water to form the m/z 130 ion.

Phosphorus-centered ions

In Table 4 we summarize results for other types of productions that are found to react with water in the massspectrometer. The first group involves product ions with apositively charged phosphorus center. For example,protonated methylphosphonic acid (m/z 97) undergoes lossof water to produce the m/z 79 ion with a positively chargedphosphorus center. This product ion regains a water moleculeand reverts to the precursor ion, a reaction that is notapparent in MS2 spectra but is demonstrated in MS3

experiments. Protonated diethyl methylphosphonate (34)(m/z 153) produces the same m/z 79 ion by loss of C2H5OHand C2H4. The reaction of the m/z 79 ion with water toproduce the m/z 97 ion is clearly observable in this case, butthe results also show that some of the m/z 97 ions areproduced directly from the m/z 153 precursor by loss of twoC2H4 molecules. Protonated hexamethylphosphoramide (35)(m/z 180) undergoes loss of (CH3)2NH to form the m/z 135



3

Table 4. Various compounds, their precursor ions (p), and their MS2 peaks involving addition of watera

Compound StructurePrecursorion, m/z Product ion

Product ion,m/z

Production +H2O,

m/z

A P-centered p= [M+H]+

33 Methylphosphonic acid 97 p-H2O 79 97

34 Diethyl methylphosphonate 153 p-C4H10O 79 97

35 Hexamethyl-phosphoramide 180p-C2H7N 135 153

p-C4H12N2 92 110

36 Sulprophos 323p-C3H8S 247 265p-C5H12S 219 237,255

37 Fenamifos sulfoxide 320 p-C5H13N 233 251

B Si-centered p= [M+H]+

38 Octamethyl-cyclotetrasiloxane 297p-CH4 281 299p-2CH4 265 283p-3CH4 249 267

39 Decamethyl-cyclopentasiloxane 371 p-CH4 355 373p-C5H16Si 267 285

C Sn-centered p= [M+H�H2O]+

40 Cyhexatin 369 p-C6H10 287 305p-C12H20 205 223

(Continues)(Continues)

P. Neta et al.



2654


Compound StructurePrecursorion, m/z Product ion

Product ion,m/z

Production +H2O,

m/z

D Negative ions p= [M�H]�

41 Rhein 283 p-CO2 239 257

42 8-Carboxy-3-methylflavone 279 p-CO2 235 253

aSee footnote in Table 1. Column 4 lists the m/z values of the precursor ion, which is defined in the subheadings in rows A, B,C and D, column 5 shows the overall neutral losses to form the product ions from p, column 6 lists the m/z values of thoseproduct ions, and the last column lists the m/z values of the product of reaction with water. Other peaks observed atvarious collision energies are not listed.


265

ion and then loss of CH3N=CH2 to form them/z 92 ion. Theseproduct ions both react with water to produce the m/z 153and m/z 110 ions, respectively, but the latter appears with ahigher abundance. Protonated sulprophos (36) (m/z 323)undergoes loss of C3H7SH to form the m/z 247 ion and thenloss of C2H4 to form the m/z 219 ion, both of which react withwater. Protonated fenamifos sulfoxide (37) undergoes parallellosses of C2H4 and C3H7NH2 to produce the m/z 233 ion withpositively charged phosphorus, which then reacts with water.These are examples of compounds whose protonated ionsundergo losses of variousmolecules to yield product ionswitha positively charged P center, which then react with water.

Si- and Sn-centered ions

The second group in Table 4 includes protonated cyclicsiloxanes (38, 39). These ions lose a CH4 molecule and thenadditional CH4 or larger fragments. The product ions of thesefragmentations have a positively charged Si which reactswith water. The next example in Table 4 is cyhexatin (40).The [M+H]+ precursor ion was not observed in this casebecause it underwent rapid loss of H2O to form (C6H11)3Sn

+

(m/z 369). This ion loses two cyclohexene molecules insuccession to form (C6H11)2SnH

+ (m/z 287) and thenC6H11SnH2

+ (m/z 205). These product ions both react withwater to form the m/z 305 and m/z 223 ions, respectively.

Negative ions

While all the ions discussed above are positively charged, thelast two entries in Table 4 are the negative ions of rhein (41)and 8-carboxy-3-methylflavone (42), which lose CO2 and thenattach a water molecule. This type of reaction, however, is notobserved with simple aromatic carboxylic acids, nor withchromone- and coumarin-carboxylic acids. Further studies



are required to understand which structures of negative ionsformed by loss of CO2 from the carboxylate ion are capableof reacting with H2O. It appears that delocalization of thenegative charge to other sites within the molecule is necessaryto promote the reaction with water. In the case of positivelycharged product ions, however, it appears that localizationof the charge on C, P, Si, or Sn is necessary to drive thereaction with water, but if the charge is transferred to adifferent site (along with transfer of protons) the reaction withwater may not take place.

Rate of reaction with water

The LTQ ion-trap mass spectrometer permits us to obtainmultistage MS3 and MS4 spectra, which confirm observationswith the QqQ instrument, where the reaction with water wasobserved in pseudo-MS3 spectra of fragment ions producedin the ion source. The LTQ also permits us to follow thereaction with water by isolating the fragment ion in theabsence of collision energy and varying the activation timebetween 0.1 ms and 10 s. Such measurements show, forexample, the gradual and complete conversion of the phenylcation (m/z 77) into protonated phenol (m/z 95) in a time-dependent manner. The data points in Fig. 5(a) fit first-orderkinetics and give a rate constant of 1.2 s�1 for the reactionwith H2O. Although the reaction C6H5

+ + H2O→C6H5OH2+

is second order, the kinetic plot shows first-order behaviorbecause the concentration of water is in large excess overthe concentration of the ions, and, when the concentrationof water is increased, the observed rate constant increasesaccordingly.

While the phenyl cation is converted completely intoprotonated phenol, the reaction of other ions with watermay not proceed to completion. For example, the4-aminophenyl cation (m/z 92), produced from 19, 20, and



5

Figure 5. Reaction of product ions (solid circles) with waterto form the adduct ions (open circles) in the ion trap of theLTQ mass spectrometer as a function of activation time withrelative collision energy set at zero: (a) phenyl cation frombenzamide (Tables 2 and 5); (b) 4-aminophenyl cation fromdapsone (Tables 3 and 5), partial reaction because some ofthe ions are changed into an unreactive phenylaminium;and (c) the product ion from sulprophos (Tables 4 and 5)where the structure with a positive charge on P reacts withwater but another structure, probably with the charge on S,does not react.

P. Neta et al.

2656

21 (Table 3), reacts with water with the same rate constant,k =1.2 s�1, but the extent of reaction is only 25% (Fig. 5(b)).Another example (Fig. 5(c)) is the reaction with water of them/z 247 fragment ion produced from protonated sulprophos(36) (Table 4), which proceeds to 30% conversion with a rateconstant k=2.6 s�1. Since the LTQ activation time is limitedto 10 s, one cannot effectively follow reactions with rateconstants <0.1 s�1. Therefore, in some cases we may observe


a small extent of reaction of a product ion with water but maynot be able to determine whether the rate of reaction is slowor the reaction occurs only to a small extent (specific exampleswill be discussed).

The phenyl cation (m/z 77) reacts with water rapidly andis converted completely into protonated phenol (m/z 95)(Fig. 5(a)), but other fragment ions undergo only partialreaction with water (Figs. 5(b) and 5(c)). Partial reaction withwater may be due to an equilibrium state involving thereverse reaction in which the water adduct loses that watermolecule. Such an equilibrium is not likely in the presentLTQ experiments since the water adducts are found to bestable in the ion trap and will lose a water molecule onlywhen the collision energy is raised. It is likely that partialreaction indicates that the ion exists in two forms,distinguished by shifts of protons, and only one of theseforms reacts with water. For example, the 4-aminophenylcation (m/z 92) may exist 25% in this form (H2NC6H4

+) whichreacts rapidly with water, while 75% are converted into aform that does not react with water, such as phenylaminium(+HNC6H5). The rearrangement may occur rapidly to createan equilibrium mixture of products or it may occur during,and in competition with, the reaction with water.

In certain cases, the MS2 spectra obtained in collision cellinstruments exhibit substantial reaction of a product ion withwater but, when this product ion is collected in the ion trap,in an attempt to follow its reaction with water as a functionof activation time, very little reaction is observed. Theapparent discrepancy between such results may be due tothe different concentration of water in the differentinstruments and to the timing of the observations. In thecollision cell, the product ions are formed and can directlyreact with the residual water in the cell (withinmicroseconds). In the ion trap, however, those same productions are first trapped (for milliseconds) and then areexamined for their reaction with water; as a result, the ionsmay isomerize into a more stable unreactive form. Thissituation is pronounced with some folic acid derivatives(Table 1). The product ion formed from tetrahydrofolic acid(3), by loss of the glutamate residue, reacts with water almostcompletely, whereas the product ions formed by similarfragmentations from folic acid (1) and some other compounds(2, 4, 5, 6) in Table 1 react with water only to a very smallextent within the observable time range. This raises thequestion whether the reaction with water in the latter casesis very slow or the extent of reaction is low because ofisomerization of the ion (by proton transfer). When weexamine the results for folic acid (1) and methopterin (4) inmore detail, we realize that the reaction may be fast but onlya small fraction of the product ions undergo this reaction.The two types of behavior of the compounds in Table 1 areprobably related to the extent of conjugation between thepteridine heterocycle and the p-aminobenzoyl group. Whenthe pteridine is fully aromatic, the positive charge initiallylocated on the carbonyl group of the p-aminobenzoyl moietycan transfer to the pteridine so that only a small fractionof the ions retain the charge on the carbonyl and reactwith water. On the other hand, when the conjugationbetween the pteridine and p-aminobenzoyl is disrupted bysaturation of the double bonds (as in tetrahydrofolic acid)most of the positive charge remains near the carbonyl groupwhere the reaction with water can take place. In the case of



Table 5. Rate and extent of reaction of product ions with water

Compound Cation m/z k(+H2O), s�1 % conversion

9 Benzamide77 1.2 >9010 Methyl hippurate

12 Nifuroxazide 93 2.9 >90

19 Dapsone92 1.2 2520 Sulfabenzamide

21 Sulfamethoxazole

26 2-Chloroquinoline 128 2.5 >90

27 Crimidine 136 2 80

28 6-Mercaptopurine 119 0.9 30

29 Trigonelline 92 0.7 >90

30 Flumequine 202 0.7 >90

32 Azamethiphos 112 2.7 >90

(Continues)(Continues)



Published in 2014. This article is a U.S. Government work and is in the public domain in the USA.


2657


Compound Cation m/z k(+H2O), s�1 % conversion

33 Methylphosphonic acid 79 1.9 >9034 Diethyl methylphosphonate

36 Sulprophos 247 2.6 30

37 Fenamifos sulfoxide 233 1.6 >90

38 Octamethylcyclotetrasiloxane 281 2.7 >90

1 Folic acid 295 ~0.6 ~3

4 Methopterin 309 ~0.6 ~3

3 Tetrahydrofolic acid 299 0.17 >90

8 Rimonabant 363 1.9 >90

P. Neta et al.

2658

rimonabant (8), the aroyl cation reacts with water rapidlyand completely, indicating that the charge remains mainlyat the carbonyl group. The rate constants are summarizedin Table 5.


The reaction of product ions with water depends on theconcentration of water in the mass spectrometer. We haveshown previously that the abundance of the water additionproducts is dependent on the pressure of the collision gas




(since higher pressure permits higher water content).[15]

However, the concentration of water may vary over timeand may change when the mass spectrometer undergoespreventive maintenance service. Therefore, the rates ofreaction with water may vary and the values given in Table 5may not be reproducible in different instruments or at differenttimes. For example, the reaction of the phenyl cation (m/z 77)produced from benzamide to give the m/z 95 ion was found toproceed to near completion but the rate constant wask=1.2 s�1 in one measurement and k=1.5 s�1 when measuredseveral weeks later. In another example, the p-aminophenylcation (m/z 92) produced from dapsone (19) showed k=1.1 s�1

and several weeks later k=1.8 s�1; the extent of reaction was25% in the first case and 29% in the later measurement.Because of these variations, the rate constants summarized

in Table 5 should be assumed to have a wide margin ofuncertainty, probably around 50% when compared in thesame mass spectrometer under similar conditions, but higherthan that when compared in different instruments. Therefore,Table 5 is given only as a general guideline and it presents oneset of results which should be internally consistent to permitus to compare the behavior of different ions in their reactionwith water. It is clear from Table 5 that the phenyl cationreacts with water rapidly and completely, presumablyforming protonated phenol. Other cations with similarbehavior are those in which an aromatic or heterocyclic ringis missing one of the ring hydrogens and include4-hydroxyphenyl, 2-quinolinyl, and chloropyridyl. However,when the phenyl cation bears an amino group the rate ofreaction is similar but the extent of reaction is diminishedby a factor of four, probably due to partial rearrangement ofthe ion into an unreactive form, as discussed above. Productions with the positive charge on P or Si atoms also can reactwith water rapidly and completely. Comparison of the resultsobtained with sulprophos (36) and femamifos sulfoxide (37)suggests that the ion from 37 has a positive charge localizedon P (and undergoes complete reaction with water) but thepositive charge in the product ion derived from 36 is onlypartly localized on P and mostly localized on S, thusundergoing only partial reaction with water. And, finally,we note the difference between, on one hand, folic acid andmethopterin, where the pteridine ring is fully aromatic andthe reaction of the [M+H–Glu]+ ions occurs to a very smallextent, and, on the other hand, tetrahydrofolic acid, wherepartial saturation of that ring leads to complete reaction ofthe [M+H–Glu]+ ions with water.

265

CONCLUSIONS

Electrospray ionization MS/MS spectra of a variety ofprecursor ions show that certain product ions with a positivecharge on C, Si, P, Sn, or with a negative charge on C, canattach a water molecule in the collision cell of a massspectrometer to produce ions which cannot be formeddirectly from the precursor ions. Phenyl cations react withwater to form protonated phenols. This rule can be applied,in general, to cations constituted of aromatic or heterocyclicrings that are missing a hydrogen atom from one of the ringcarbons. However, if such ions undergo rearrangement, theymay be converted completely or partly into unreactive ions(see examples in Table 5). Whether aroyl cations react with



water depends on their detailed structure (compare Tables 1and 2). The simple benzoyl cation undergoes loss of COrather than reacts with water; only more stabilized aroylcations, which do not undergo decarbonylation, react withwater. However, these complex aroyl cations may rearrangeinto unreactive forms, and it is difficult to predict whichstructures will exhibit reaction with water. In the case ofproduct ions with a positive charge on P, Si, or Sn, it appearsthat if the charge is localized on these atoms the reaction withwater takes place. On the other hand, if a positive charge islocalized on N or S atoms the ions do not react with water.Therefore, it is advisable that any peak in the MS/MSspectrum, which is not assignable to a simple fragmentationproduct, be considered for the possibility of being a productof reaction of another ion with water. Mass accuracy in theppm range is critical for confirming such an assignment.

The abundance of the ions resulting from reaction withwater in the MS/MS spectra depends on the experimentalconditions, mostly as these conditions relate to theconcentration of water in the mass spectrometer, and alsovary with collision energy. The abundance may be as low asseveral percent of the base peak (as in Fig. 1(a)), sometimesbarely observable, or as high as becoming the most intensepeak in the MS2 spectrum (as in Fig. 4(a)). Small peaks inthe MS2 spectra, which are due to reaction with water, maybecome more abundant in MS3 spectra and may help confirmthe occurrence of the reaction with water. Formation of thewater addition product ions, and their subsequentdissociation in the collision cell, may lead to observation ofMS/MS spectra with unexpected product ion peaks andmay distort MRM results. We have outlined several types ofproduct ions that can react with water and we expect thatadditional types will be detected. Therefore, the possibilityof reaction of product ions with water should be taken intoaccount when interpreting tandem mass spectra and shouldbe carefully considered when using MRM measurements.The reaction with water can change the relative abundancesof product ions and can lead to formation of product ionswhich cannot be formed directly from the originalcompound. MRM measurements are more reliable if all thepeaks in the MS/MS spectrum of the precursor ion can beascribed to CID products or if an isotopically labeled internalstandard is used.

REFERENCES

[1] NIST MS/MS Library. 2014. Available: http://chemdata.nist.gov/mass-spc/msms-search/.

[2] P. Ausloos, C. L. Clifton, S. G. Lias, A. I. Mikaya, S. E. Stein,D. V. Tchekhovskoi, O. D. Sparkman, V. Zaikin, D. Zhu. Thecritical evaluation of a comprehensive mass spectral library.J. Am. Soc. Mass Spectrom. 1999, 10, 287.

[3] R. Tuytten, F. Lemiere, W. Van Dongen, E. L. Esmans,E. Witters, W. Herrebout, B. Van Der Weken, E. Dudley,R. P. Newton. Intriguing mass spectrometric behavior ofguanosine under low energy collision-induced dissociation:H2O adduct formation and gas-phase reactions in thecollision cell. J. Am. Soc. Mass Spectrom. 2005, 16, 1291.

[4] R. Tuytten, F. Lemiere, E. L. Esmans, W. A. Herrebout,B. J. van der Veken, E. Dudley, R. P. Newton, E. Witters.In-source CID of guanosine: gas phase ion-moleculereactions. J. Am. Soc. Mass Spectrom. 2006, 17, 1050.



9

http://chemdata.nist.gov/mass-spc/msms-search/

http://chemdata.nist.gov/mass-spc/msms-search/

P. Neta et al.

2660

[5] S. K. Lee, H. J. Kim, C. Jin, J. Lee. Formation of the unusual[M+H–11 Da]+ ion peak in the collision-induced dissociationmass spectrum of [M+H]+ ion of hydrochlorothiazide.J. Mass Spectrom. 2009, 44, 1538.

[6] X. Cao, Y. Yu, X. Ye, W. Mo. Solvation in gas-phasereactions of sulfonic groups containing ionic liquids inelectrospray ionization quadrupole ion trap massspectrometry. Eur. J. Mass Spectrom. 2009, 15, 409.

[7] C. P. Chander, G. Raju, G. Mathai, R. Srinivas, H. K. Gaikwad,R. Bantu, L. Nagarapu. Electrospray ionization tandem massspectrometry of 3-phenyl-N-(3-(4-phenylpiperazin-1-yl)-1H-pyrazole-5-carboxamide derivatives: unusual fragmentationinvolving loss of 11 u. Rapid Commun. Mass Spectrom. 2012,26, 207.

[8] P. Frycak, J. Jirkovsky, V. Ranc, P. Bednar, V. Havlicek,K. Lemr. Secondary processes in atmospheric pressurechemical ionization–ion trap mass spectrometry: a casestudy of orotic acid. J. Mass Spectrom. 2012, 47, 720.

[9] C.-C. Chan, F. U. Axe, M. Bolgar, A. B. Attygalle. Reactivityof gaseous sodiated ions derived from benzenedicarboxylate salts toward residual water in the collisioncell. J. Mass Spectrom. 2010, 45, 1130.

[10] Y.-Q.Wu, N.-W. Zhang, F. Li, K.-Z. Jiang. Deviant mass shift ofhydrated product ions from sodiated β-anilinodidrochalconesusing an ion-trap mass spectrometer. Eur. J. Mass Spectrom.2013, 19, 351.

[11] J. Sultan. Collision induced dissociation of deprotonatedguanine: fragmentation of pyrimidine ring and wateradduct formation. Int. J. Mass Spectrom. 2008, 273, 58.


[12] A. B. Attygalle, N. Kharbatia, J. Bialecki, J. Ruzicka,A. Svatos, E. J. Stauber. An unexpected ion-molecule adductin negative-ion collision-induced decomposition ion-trapmass spectra of halogenated benzoic acids. Rapid Commun.Mass Spectrom. 2006, 20, 2265.

[13] A. Kaufmann, P. Butcher, K. Maden, M. Widmer, K. Giles,D. Uria. Are liquid chromatography/electrospray tandemquadrupole fragmentation ratios unequivocal confirmationcriteria? Rapid Commun. Mass Spectrom. 2009, 23, 985.

[14] P. Neta, B. Godugu, Y. Liang, Y. Simón-Manso, X. Yang,S. E. Stein. Electrospray tandem quadrupole fragmentationof quinolone drugs and related ions. On the reversibility ofwater loss from protonated molecules. Rapid Commun. MassSpectrom. 2010, 24, 3271.

[15] P. Neta, Y. Simón-Manso, Y. Liang, S. E. Stein. Loss of H2 andCO from protonated aldehydes in electrospray ionizationmassspectrometry. Rapid Commun. Mass Spectrom. 2014, 28, 1871.

[16] X. Yang, P. Neta, S. E. Stein. Quality control for buildinglibraries from electrospray ionization tandem mass spectra.Anal. Chem. 2014, 86, 6393.

[17] G. C. Thorne, K. D. Ballard, S. J. Gaskell. Metastabledecomposition of peptide [M+H]+ ions via rearrangementinvolving loss of the C-terminal amino acid residue. J. Am.Soc. Mass Spectrom. 1990, 1, 249.

[18] Y.-M. She, O. Krokhin, V. Spicer, A. Loboda, G. Garland,W. Ens, K. G. Standing, J. B. Westmore. Formation of[bn-1 +H2O] ions by collisional activation of MALDI-formedpeptide [M+H]+ ions in a QqTOF mass spectrometer.J. Am. Soc. Mass Spectrom. 2007, 18, 1024.



Unexpected peaks in tandem mass spectra due to reaction of ...

Documents

Transcript of Unexpected peaks in tandem mass spectra due to reaction of ...