Acetonitrile covalent adduct chemical ionization tandem ... · Acetonitrile covalent adduct chemical ionization tandem mass spectrometry of non‐methylene‐interrupted pentaene

Research Article

Received: 28 February 2011 Revised: 18 April 2011 Accepted: 19 April 2011 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2011, 25, 1933–1941

Acetonitrile covalent adduct chemical ionization tandem massspectrometry of non‐methylene‐interrupted pentaene fatty acidmethyl esters

Susana P. Alves1,2, Cynthia Tyburczy3, Peter Lawrence3, Rui J. B. Bessa1,4 andJ. Thomas Brenna3*1INRB – Instituto Nacional dos Recursos Biológicos, Unidade de Produção Animal, Fonte‐Boa, 2005‐048 Vale de Santarém,Portugal2REQUIMTE, ICBAS, Instituto de Ciências Biomédicas de Abel Salazar, Universidade do Porto, Rua Padre Armando Quintas,4485‐661 Vairão, Portugal3Division of Nutritional Sciences, Cornell University, Savage Hall, Ithaca, NY 14853, USA4Faculdade de Medicina Veterinária, Universidade Técnica de Lisboa, CIISA, Av. da Universidade Técnica, 1300‐477 Lisboa,Portugal

Acetonitrile covalent adduct chemical ionization tandem mass spectrometry (CACIMS/MS) has shown to be anefficient method for the identification of double‐bond position in homoallylic, conjugated and several polyene non‐methylene‐interrupted (NMI) fatty acid methyl esters. However, it has not been thoroughly evaluated for NMIhighly unsaturated fatty acids (HUFA) with more than four double bonds. Docosahexaenoic acid (DHA)‐rich singlecell oil (DHASCO®; Martek Biosciences, Corp.) was partially hydrogenated (partially hydrogenated DHASCO;PHDO) producing ten novel 22:5 and 22:6 HUFA isomers. In single‐stage MS, the ratio of [M+54]+/[M+54‐32]+ for the22:5 and 22:6 isomers indicated the presence of homoallylic or partially conjugated double‐bond systems. TheCACIMS/MS spectra revealed six 22:5 isomers with diagnostic ions corresponding to the homoallylic 22:5n‐6 and22:5n‐3 isomers, and four distinct NMI 22:5 isomers. Diagnostic ions for four 22:6 isomers were identical to thenative DHA illustrating that CACIMS/MS is sensitive to double‐bond position but not geometry. Three gaschromatography (GC) peaks for partially conjugated 22:6 isomers were also detected and clearly distinguishablefrom homoallylic 22:6 isomers, but their CACIMS/MS spectra did not yield prominent ions indicative of double‐bond position, possibly due to co‐elution. Overall, CACIMS/MS was effective for determining double‐bondposition in NMI 22:5 isomers. Further investigations are warranted to evaluate and determine fragmentationpatterns for partially conjugated and NMI 22:6 HUFA. Copyright © 2011 John Wiley & Sons, Ltd.

(wileyonlinelibrary.com) DOI: 10.1002/rcm.5065

The interest in the production of eicosapentaenoic acid (EPA;20:5n‐3) and docosahexaenoic acid (DHA; 22:6n‐3) from bothfish and microalgae sources has increased in the last fewyears due to their benefits to human health. DHA is essentialfor growth and functional development of the brain in infantsand is required for maintenance of normal brain function inadults.[1]

Partial hydrogenation is a commonly used method forconverting dietary unsaturated oils into semi‐solid fats thatare less expensive and more stable than animal fats.[2]

Hydrogenation of oils can be carried out using homogeneousor heterogeneous catalysts.[3] The first homogeneous catalytichydrogenation of unsaturated organic compounds wasachieved using (PPh3)3RhCl (Wilkinson’s catalyst).[4,5] This

* Correspondence to: J. T. Brenna, Division of NutritionalSciences, Cornell University, Savage Hall, Ithaca, NY 14853,USA.E‐mail: [email protected]

Rapid Commun. Mass Spectrom. 2011, 25, 1933–1941

19

catalyst has also been used to study the isomerization ofdouble bonds in unsaturated fatty acids (FA) using deuterium‐labeled FA, because it incorporates nearly a stoichiometricnumber of deuterium atoms during the reduction of olefinicand acetylenic compounds.[6–9]

The catalytic partial hydrogenation of fish and microalgaeoils invariably generates various non‐methylene‐interrupted(NMI) highly unsaturated fatty acids (HUFA; ≥4 doublebonds), some geometrical isomers of HUFA, and a few trans‐FA isomers.[3] The influence of these NMI or trans‐HUFA inthe diets of humans and implications for health outcomesand chronic disease risk remain unknown.

Studies of the formation and identification of geometricalisomers of HUFA originating from the partial hydrogenationor deodorization of EPA‐ and DHA‐rich oils are limited.Fournier et al. reported a method to analyze and quantifygeometrical isomers of EPA and DHA formed during thedeodorization of fish oil.[10,11] Their methodology is based onfractionation by silver‐ion thin layer chromatography andsilver‐ion high‐performance liquid chromatography, andidentifications were achieved only with the use of pure or

Copyright © 2011 John Wiley & Sons, Ltd.

33

S. P. Alves et al.

1934

synthesized standards. Mjos has extensively studied theresolution and elution patterns of trans‐isomers of EPA andDHA using neat standards or isomerized samples preparedby catalytic isomerization.[12,13] Furthermore, he developed amethod based on two‐dimensional FA retention indices (2D‐FARI) for a suitable identification of the trans geometry ofDHA isomers;[14] however, this method only separatesisomers into groups according to the number of trans doublebonds. These studies do not report a complete identificationof EPA‐ or DHA‐derived isomers and lack in the identifica-tion of NMI pentaene and hexaene fatty acid methyl esters(FAME) derived from EPA or DHA.4,4‐Dimethyloxazoline (DMOX) derivatives are exten-

sively used for the identification of double‐bond position inFA.[15–19] Destaillats et al. used DMOX derivatives for theidentification of FAmetabolites of rumelenic acid (cis‐9,trans‐11,cis‐15 18:3) and cis‐9,trans‐13,cis‐15 18:3 in the livers ofrats.[20] Picolinyl derivatives followed by mass spectralanalysis have also been used for the identification of double‐bond position of FAME.[21–23] However, these derivatizationmethods involve chemical treatment of the sample withthe potential for contamination or incomplete reaction.Due to their higher molecular weight, these derivativespromote changes in chromatographic properties thatmake it difficult to establish a correspondence withFAME chromatograms.Recent developments in mass spectrometry and chromato-

graphy provide new possibilities for the identification ofdouble‐bond position in lipids.[24–26] In addition to collision‐induced dissociation (CID) using low or high energy,[26–28]

dilithiated adduct ions of unsaturated FA obtained byMSn,[29–31] or ozone‐based methods (ozone electrosprayionization, OzESI‐MS, and ozone‐induced dissociation,OzID),[32–35] covalent adduct chemical ionization tandemmass spectrometry (CACIMS/MS) with acetonitrile has beensuccessfully used for the identification of double‐bondposition in methylene‐interrupted (i.e. homoallylic),[36–38]

conjugated,[39,40] and a few polyene NMI FAME.[41] Withthis technique, a product of acetonitrile self‐reaction, (1‐methyleneimino)‐1‐ethenylium (CH2=C=N+=CH2), gener-ated under CACIMS conditions, reacts with the analytedouble bond to yield molecular ions with 54 mass unitsabove the parent analyte. Under CACIMS/MS conditions,isolation and collisional activation of [M+54]+ ions yieldsdiagnostic fragments that enable unambiguous double‐bondpositional assignment for a wide range of FAME. Thistechnique has not been evaluated with many NMI HUFA.Therefore, we partially hydrogenated a DHA‐rich microalgaeoil as a source of NMI HUFA, and studied the CACIMS andCACIMS/MS spectra to determine patterns of fragmentationfor NMI HUFA.

EXPERIMENTAL

Chemicals

Wilkinson’s catalyst (tris(triphenylphosphine)rhodium(I)chlo-ride) was obtained from Sigma‐Aldrich (St. Louis, MO, USA)and DHA single cell oil (DHASCO®) fromMartek Biosciences,Corp. (Columbia, MD, USA). All solvents were obtained fromAldrich Chemical Co. (Milwaukee, WI, USA).

wileyonlinelibrary.com/journal/rcm Copyright © 2011 John Wil

Partial hydrogenation

DHASCO® (5mg) was partially hydrogenated by using40mg of Wilkinson’s catalyst and 2mL of benzene/ethanol(1:1) under hydrogen at atmospheric pressure. The reactionwas allowed to run for 15min at room temperature. The flaskwas opened and the solvent was removed by evaporation.[9]

FAME from partially hydrogenated DHASCO® (PHDO)were prepared by methylation with 1% sulfuric acid inmethanol.[42]

Instrumentation

Structural identification of FAME was performed on a VarianStar 3400CX gas chromatograph equipped with a 1078 split/splitless injector and coupled to a Varian Saturn 2000 ion trap(Varian Inc., Walnut Creek, CA). A BPX70 capillary column(60m×0.32mm×0.25 µm; SGE Inc., Austin, TX, USA) wasused. The column temperature and injector parameters forboth CACIMS and CACIMS/MS analysis of HUFA were asfollows: Injector temperature was maintained at 250 °C insplitless mode with a purge at 0.85min after injection; initialcolumn temperaturewas 150 °C ramped up to 200 °C at 20 °C/min, then increased to 225 °C at 1 °C/min, held for 2min, andthen increased to 255 °C at 15 °C/min for 10min. Helium wasused as carrier gas at a flow rate of 1mL/min.

Ion trap parameters were as follows: trap temperature,150 °C; manifold temperature, 45 °C; transfer line tempera-ture, 220 °C; axial modulation amplitude, 4.0V. The excitationstorage level varied fromm/z 93.8 to 94.7 and typical resonantexcitation amplitudes varied from 0.50 to 0.56V. Additionalmass spectrometer parameters have been described in detailpreviously.[36]

Quantitative analysiswas performedbyGC‐flame ionizationdetection (FID) using a HP 5890 Series II system (HewlettPackard, Palo Alto, CA, USA) equipped with a BPX70 fused‐silica column (25m×0.22mm i.d. × 0.25µm film; SGE Inc.,Austin, TX,USA). GCanalyseswere performed in triplicate andan equal weight FAMEmixture was used to calculate responsefactors. Integration of FAME peaks was performed usingGalaxie Workstation Chromatography (Varian, Inc., WalnutCreek, CA, USA). The column temperature and injectorparameters were as follows: Injector temperature was main-tained at 250 °C in splitless mode; initial column temperaturewas 150 °C rampedup to 200 °C at 20 °C/min, then increased to225 °C at 1 °C/min, held for 2min, and then increased to 255 °Cat 15 °C/min for 10min. Helium was used as carrier gas at aflow rate of 1mL/min.

RESULTS AND DISCUSSION

Partial hydrogenation of DHASCO®

The Wilkinson’s catalyst was an efficient reagent for partialhydrogenation of DHASCO®. A high amount of catalyst wasnecessary to ensure a rapid reaction, and minimize oxidationof the DHA. Reaction was performed under hydrogen atatmospheric pressure. It was observed that if a slightlypositive pressure was used, the reaction proceeded quickly tocomplete hydrogenation of unsaturated double bonds andthe production of the saturated 22:0.

ey & Sons, Ltd. Rapid Commun. Mass Spectrom. 2011, 25, 1933–1941

Table 1. Distribution of 22:5, 22:6 and partial sums (% oftotal FA) in DHASCO® and PHDO

FAMEPeak

numbera DHASCO® PHDO

22:5 isomers22:5 n‐6 n.d. 0.4922:5 cis‐4,7,10,13,19

11 n.d. 0.46

22:5 cis‐4,7,10,16,18

12 n.d. 0.36

22:5 cis‐4,7,13,16,19

13b n.d. 0.60

22:5 cis‐4,10,13,16,19

13b n.d.

22:5 n‐3 0.65 0.3222:6 isomers22:6 cis/trans‐4,7,10,13,16,19

13b n.d. 0.08

22:6 cis/trans‐4,7,10,13,16,19

14 n.d. 0.08

22:6 cis/trans‐4,7,10,13,16,19

15 n.d. 0.16

22:6 cis/trans‐4,7,10,13,16,19

16 n.d. 0.21

22:6 n‐3 46 3.722:6 partiallyconjugated

17 n.d. 0.10

22:6 partiallyconjugated

18 n.d. 0.13

22:6 partiallyconjugated

19 n.d. 0.13

Partial sumsc

SFA 30 49MUFA 22 43PUFA 1.1 1.2HUFA 47 7.1Trans‐FA 0.04 2.1

Abbreviations: SFA, saturated fatty acids; MUFA, mono-unsaturated fatty acids; PUFA, polyunsaturated fatty acids;HUFA, highly unsaturated fatty acids; Trans‐FA, trans fattyacids; n.d., not detected.aPeak numbers correspond to peaks in Fig. 1.bThese peaks co‐elute under the GC‐FID conditions.cSFA = sum of 12:0, 14:0, 16:0, iso‐18:0, 18:0, 20:0, 22:0 and24:0. MUFA= sum of 14:1c9, 16:1t9, 16:1c9, 18:1t, 18:1c9, 18:1c,20:1 isomers and 24:1c15. PUFA = sum of 18:2n‐6, 18:3n‐3,20:3n‐6, 22:2 and 22:3. HUFA = sum of total 22:4 isomers,total 22:5 isomers and total 22:6 isomers; Trans‐FA = sum of16:1t9, 18:1t, 22:1t20, 22:6 cis/trans isomers, 22:6 partiallyconjugated.

121110 min

22:0

22: 6

n-3

22:5

n-3

22:5

n-6

24:0

12 3 4 5 6 7 8 9 10

1112

1314

15

16 171819

PHDO

DHASCO®

Inte

nsity

(µV

)

Figure 1. Partial GC‐FID chromatograms of the long‐chainFAME region of the DHASCO® and PHDO. Tentative peakidentification: (1) cis‐4 22:1 and cis‐7 22:1; (2) cis‐10 22:1; (3) cis‐13 22:1; (4) 22:2; (5) trans‐20 22:1; (6) cis‐17 22:1; (7) 22:3; (8) 22:4;(9) 22:4; (10) 22:4; (11) cis‐4,cis‐7,cis‐10,cis‐13,cis‐19 22:5; (12) cis‐4,cis‐7,cis‐10,cis‐16,cis‐19 22:5; (13) cis‐4,cis‐7,cis‐13,cis‐16,cis‐1922:5 and cis‐4,cis‐10,cis‐13,cis‐16,cis‐19 22:5 and 4,7,10,13,16,19‐22:6; (14) 4,7,10,13,16,19‐22:6; (15) 4,7,10,13,16,19‐22:6; (16)4,7,10,13,16,19‐22:6; (17) 22:6; (18) 22:6; (19) 22:6.

CACIMS/MS for identification of HUFA

193

Table 1 shows the distribution of FA prior to and afterpartial hydrogenation of DHASCO®, which containedinitially 46% of total FA as DHA. The partial hydrogena-tion of DHASCO® produced 26 FA that were not initiallypresent in the oil. Figure 1 shows that 17 of these were22:4, 22:5 and 22:6 HUFA isomers. Despite the formation of17 new HUFA, the percentage of total HUFA in the PHDOdecreased by 85% and the percentage of saturated FA

Copyright © 2011Rapid Commun. Mass Spectrom. 2011, 25, 1933–1941

increased by 60%. Hydrogenation conditions also pro-moted the isomerization of unsaturated FA, and this wasverified by the increase of trans‐FA in the PHDO. More-over, it is widely recognized that catalytic partial hydro-genation of triacylglycerols invariably generates sometrans‐isomers.[3]

Identification of 22:5 isomers

Partial hydrogenation of DHASCO® produced five novel22:5 isomers that were not originally present in the oil. Oneisomer was determined to be 22:5n‐6 originating from thehydrogenation of a double bond at C19‐C20 on the DHAmolecule. All isomers eluted before 22:5n‐3, which wasdetected in the untreated DHASCO® oil (Fig. 1; peaks 11–13, 22:5n‐6). It is well known that the n‐6 series elutebefore the n‐3 series on the cyanopropylsilicone stationaryphase;[43] however, as far as we know there are noprevious reports showing the elution profile of NMI 22:5isomers.

Figure 2 presents a typical CACIMS spectrum of a 22:5FAME. The CACIMS spectra of all 22:5 isomers showedthe base peak ion [M+54]+ and the common ions, [MH]+,[MH–32]+, and [M+54–32]+ at m/z 345, 313 and 366,corresponding to the protonated molecule, and losses ofmethanol from the protonated molecule and from theadduct.[37] These characteristic ions are not representativeof double‐bond position; however, the intensity ratios of[M+54]+/[M+54–32]+ have previously been shown tobe related to double‐bond geometry in 18:2 and 18:3FAME.[40,41] For 18:3 FAME, homoallylic isomers show aratio >8 while partially conjugated and fully conjugatedisomers show intensity ratios ranging from 2–3 and <1.0,respectively.[41] In Table 2, the ratio of [M+54]+/[M+54–32]+

was 66 and 43 for the homoallylic 22:5 isomers (I, VI), andranged from 35–46 for the other 22:5 isomers (II–V). Thesimilarity in ratios between the homoallylic 22:5 FAMEand the other 22:5 isomers indicates that the novel 22:5isomers have a missing internal double bond rather than apartially conjugated double‐bond arrangement.[41] Thus, wereport that these four isomers (Table 2, II–V) produced fromthe partial hydrogenation of DHASCO are NMI 22:5isomers.

wileyonlinelibrary.com/journal/rcmJohn Wiley & Sons, Ltd.

5

S. P. Alves et al.

1936

Figure 3 shows a partial chromatogram presenting the orderof elution of the 22:5 isomers in PHDO, and their correspondingCACIMS/MS spectra. These spectra were obtained by colli-sionally activated dissociation of the [M+54]+ ions.Peak I (Fig. 3 – spectrum I) was identified as 22:5n‐6, a

common homoallylic FAME, with cis double bonds at carbons4, 7, 10, 13, 16. The spectrum presents two principal diagnostic

Table 2. CACIMS [M+54]+/[M+54–32]+ ratios for par-tially conjugated and homoallylic 22:5 and 22:6 isomers inPHDO

FAME [M+54]+/ [M+54‐32]+

22:5a

4c,7c,10c,13c,16c; n‐6 (I) 664c,7c,10c,13c,19c (II) 464c,7c,10c,16c,19c (III) 354c,7c,13c,16c,19c (IV) 404c,10c,13c,16c,19c (V) 437c,10c,13c,16c,19c; n‐3 (VI) 4322:6Homoallylic4c,7c,10c,13c,16c,19c; n‐3 (A)b 494,7,10,13,16,19 (B) 874,7,10,13,16,19 (C) 554,7,10,13,16,19 (D) 714,7,10,13,16,19 (E) 72Partially conjugatedPeak fc 15Peak g 27Peak h 19aLabels I–VI correspond to peaks in Fig. 3.bLabels A–E correspond to peaks in Fig. 4 andmass spectrain Fig. 5.cLabels f–h correspond to peaks in Fig. 4 and mass spectrain Fig. 6.

100 150 200 250 300 350 400 m/z0

25

50

75

100

81121

175

203

235

263

313

345

366

398[M+54]+

[MH]+

[MH-32]+

[M+54-32]+

Rel

ativ

e ab

unda

nce

(%)

Figure 2. CACIMS spectrum of 22:5n‐3 showing the basepeak [M+54]+ and the common ions [MH]+, [MH–32]+ and[M+54–32]+.


ions at m/z 270 and 286, corresponding to the ω diagnostic ion(fragment containing the methyl end of the native FA), and theα diagnostic ion (fragment containing the carboxyl group).Identical fragmentations for the 22:5n‐6 have been previouslypublished.[37] For peak II (Fig. 3 – spectrum II), the ionappearing at m/z 270 is also the ω diagnostic ion and at m/z246 is the α diagnostic ion. These ions suggest the presence offour homoallylic double bonds and an isolated double bond atcarbon‐19. The strong ion atm/z 324 corresponds to the doublebond at carbon‐4.

Diagnostic ions for peak III (Fig. 3 – spectrum III) indicatesthe presence of two separate homoallylic systems, onecontaining three double bonds at positions 4, 7, 10, and asecond homoallylic system with two double bonds at positions16, 19. The predominant cleavage patterns show fragmentationbetween C6–C7andC8–C9, identical to previously reports forhomoallylic triene systems.[36] The ω diagnostic ion wasobserved at m/z 270, and the α diagnostic ion at m/z 206. Thefragments atm/z 148 and 328 are consistentwith the presence ofdouble bonds at carbon 16, 19. The fragment at m/z 148corresponds to cleavage vinylic to the double bond betweenC15–C16.

Peak IV (Fig. 3 – spectrum IV) appears to result fromhydrogenation of the double bond at carbon‐10. It has ahomoallylic system at carbons 13, 16, 19, which produced twodiagnostic ions characteristic of a methylene‐interrupted triene.The ω diagnostic ion was observed at m/z 148, and the αdiagnostic ion at m/z 328; these ions originated from cleavagesvinylic to the center double bond. The minor ion atm/z 270 wasalso present, suggesting cleavage between C6–C7. Peak IVpartially co‐eluted with peak V, another product of partialhydrogenation of DHA (Fig. 3), although the CACIMS/MSspectra of these two isomers was sufficiently resolved to enableinterpretation.

The acetonitrile CACIMS/MS spectrum of peak V (Fig. 3– spectrum V) indicates the presence of a tetramethylene‐interrupted double‐bond system near the methyl end ofthe hydrocarbon chain. The ω diagnostic ion at m/z 188and the α diagnostic ion at m/z 328 correspond to cleavageinternal to the double‐bond system at sites allylic to thefirst double bond from each end. Consistent with spectraI–IV, the ion at m/z 324 corresponded to a double bond atcarbon‐4.

The last 22:5 isomer to be identified was the 22:5n‐3which was originally present in the DHASCO®. SpectrumVI in Fig. 3 shows the diagnostic ions for the 22:5n‐3 isomer,the ω diagnostic ion was observed at m/z 228 and the αdiagnostic ion at m/z 328. Identical fragmentations havebeen published elsewhere.[37] The CACIMS/MS techniqueenabled facile identification of the 22:5 isomers, each arisingfrom the hydrogenation of one double bond on the nativeDHA molecule without positional rearrangement of theremaining double bonds. The retention times and elutionpattern strongly suggest that all 22:5 isomers have doublebonds with cis configurations. Further, these data suggestthat the elution order for 22:5 NMI isomers is specific tolocation of the missing internal double bond. Isomers with amissing double bond near the terminal methyl end of theFAME elute before FAME with missing internal doublebonds near the carboxyl end. This conclusion is consistentwith the elution order for homoallylic FAME, with n‐6isomers eluting prior to n‐3 isomers.[43]


O

O

188 270

328 366

399

328275

135188

270

αω

22: 5 (4,10,13,16,19)

228328

366

399

m/z

αω

22: 5 n- 3 (7,10,13,16,19)

148

206 270328

366

399

ωα

150 200 250 300 350 m/z

22: 5 (4,7,10,16,19)

148188

270 328 366

399

αω

22: 5 (4,7,13,16,19)

100 150 200 250 300 350 m/z

0

270286 366

399

αω

22: 5 n- 6 (4,7,10,13,16)

246270

324342

366

399

ωα

22: 5 (4,7,10,13,19)I II

III IV

V VI

I

II III IV V VI

GC 22:5 region

CACI-MS/MS mass spectra

0.0

2.5

5.0

7.5

Inte

nsity

(kC

ount

s)

15.5 16.0 16.5 17.0 min

Rel

ativ

e ab

unda

nce

(%)

25

286233

217270

O

O

100 150 200 250 300 350 m/z

0

Rel

ativ

e ab

unda

nce

(%)

25

342

O

O246193

217270

O

O

217270

206153

148

328324

324

0

Rel

ativ

e ab

unda

nce

(%)

25

150 200 250 300 350 m/z

O

O

95148

328275

188324

270

0

Rel

ativ

e ab

unda

nce

(%)

25

150 200 250 300 350 m/z0

Rel

ativ

e ab

unda

nce

(%)

25

150 200 250 300 350

O

O328275

175228

0

Rel

ativ

e ab

unda

nce

(%)

25

324 324

324

324

324

288

288

271 271

289

271

95

275 271

217

135

324

271 217 235

Figure 3. Total ion current GC/MS chromatogram of 22:5 isomers in PHDO and theirCACIMS/MS spectra obtained upon collisionally deactivated dissociation of the [M+54]+

ion at m/z 398. (I) 22:5n‐6; (II) cis‐4,cis‐7,cis‐10,cis‐13,cis‐19 22:5; (III) cis‐4,cis‐7,cis‐10,cis‐16,cis‐19 22:5; (IV) cis‐4,cis‐7,cis‐13,cis‐16,cis‐19 22:5 partially coeluting with (V) cis‐4,cis‐10,cis‐13,cis‐16,cis‐19 22:5; (VI) 22:5n‐3. Spectra show diagnostic ions, labeled with α and ω.


wileyonlinelibrary.com/journal/rcmCopyright © 2011 John Wiley & Sons, Ltd.Rapid Commun. Mass Spectrom. 2011, 25, 1933–1941

1937

S. P. Alves et al.

1938

Identification of 22:6 isomers

Figure 4 presents two groups of 22:6 isomers that wereproduced by partial hydrogenation of DHASCO. The firstgroup was labeled with capital letters (A–E) and the secondgroup with lower‐case letters (f–h). As shown in Table 2, thefirst group appears to represent geometrical isomers of 22:6n‐3while the second group represents partially conjugated 22:6isomers. The higher retention times of peaks f–g in thecyanopropylpolysiloxane stationary phase compared withpeaks A–E also support the hypothesis of the partiallyconjugated structure.Further evidence of the double‐bond structure is found in ion

intensity ratios. In a previous investigation with 18:3 isomers,homoallylic FAME exhibited higher CACIMS [M+54]+/[M+54–32]+ intensity ratios compared with partially conjugatedFAME.[41] The values presented in Table 2 show that 22:6homoallylic FAME have ratios ranging from 49 to 87, whereaspeaks f–h have ratios ranging from 15 to 27, indicating thatthese 22:6 isomers are partially conjugated. Figure 4 shows theCACIMS spectra of a homoallylic 22:6 isomer and a partiallyconjugated 22:6 isomer. The two mass spectra show similarstrong ions corresponding to the base ion [M+54]+ and theadditional expected ions, [M+54–32]+, [MH–32]+ and [MH]+ atm/z 396, 364, 311 and 343, respectively.The CACIMS/MS spectrum for the native 22:6n‐3,

obtained upon collisional dissociation of the [M+54]+ ion atm/z 396, produced two diagnostic ions resulting from

100

100 1500

81

135

i

Rel

ativ

e ab

unda

nce

(%)

16 17 18 19

BC

A

D E

f g

0.0

2.5

5.0

7.5

10.0

Inte

nsity

(kC

ount

s)

Figure 4. Total ion current GC/MS chromatogrgeometric isomers of DHA (peak A); peaks f totypical CACIMS spectra: (i) homoallylic and (ii) pbase peak [M+54]+ and the common ions [MH]+


cleavage internal to the double‐bond system at sites allylicto the outer double bonds (Fig. 5 – spectrum A). Thus, the αdiagnostic ion was observed at m/z 326 and the ω diagnosticion was observed at m/z 268. The base peak corresponds tothe protonated [M+54]+ ion at m/z 397, which is commonlyobserved during CACIMS/MS for HUFA. Peaks B, C, D andE (Fig. 4) correspond to geometric isomers of the 22:6n‐3,each presenting double bonds at carbons 4, 7, 10, 13, 16, 19,but with different cis/trans configurations. The CACIMS/MSspectra enabled the identification of double‐bond position bythe production of the characteristic diagnostic ions at m/z 268and 326. All CACIMS/MS spectra in Fig. 5 are equivalent to22:6n‐3, showing that geometrical isomers yield substantiallyindistinguishable spectra. The intensities of ions in theCACIMS/MS spectra of conjugated linoleic acid (CLA)[40]

isomers were demonstrated to be a determinate of double‐bond geometry; however, in this case geometry cannot beestablished via these rules.

Studies of the isomerization of DHA using high tempera-tures showed that several mono‐trans‐DHA isomers areproduced,[12] and in samples heated to high temperaturesome di‐trans‐isomers were also produced in minor amounts.Other studies indicate that mainly mono‐ and di‐trans‐isomers of DHA are formed during deodorization of fishoils.[44] Our hydrogenation was conducted at low tempera-tures consistent with formation of exclusively mono‐trans‐DHA isomers (peaks B, C, D, E; Fig. 4). From analysis oftrienes, it is known that trans geometry in the terminal n‐3

200 250 300 350 400 m/z

161

201269 293

311343

396[M+54]

[MH]

[MH-32][M+54-32]

364

Homoallylic

20 min

h 100 150 200 250 300 350 400 m/z0

100

95135 161

215

269 293311

343

396[M+54]

[MH]

[MH-32][M+54-32]

364

Partially Conjugatedii

Rel

ativ

e ab

unda

nce

(%)

am of 22:6 isomers in PHDO. Peaks B to E,h, partially conjugated isomers of DHA andartially conjugated 22:6 isomers, showing the, [MH–32]+ and [M+54–32]+.


200 228268

286

326342

364

397

0

30

αω

O

O326273

215268

22:6 n-3 (4,7,10,13,16,19)A

100 150 200 250 300 350 m/zR

elat

ive

abun

danc

e (%

)

200228 268

286326

342364

397

B

100 150 200 250 300 350 m/z0

30

Rel

ativ

e ab

unda

nce

(%)

200 228268

286326

342

364

397

C

100 150 200 250 300 350 m/z0

30

Rel

ativ

e ab

unda

nce

(%)

200 228 268286

326342

364

397

D

100 150 200 250 300 350 m/z0

30

Rel

ativ

e ab

unda

nce

(%)

200228 268

286326

342364

397

E

100 150 200 250 300 350 m/z0

30

Rel

ativ

e ab

unda

nce

(%)

Figure 5. CACIMS/MS spectra obtained upon collisionally deactivated dissociation ofthe [M+54]+ ion for geometric isomers of DHA: (A) spectrum of 22:6n‐3, DHA and (B–E)spectra of geometric isomers of DHA. All show principal diagnostic ions at m/z 268 (ωdiagnostic ion) and 326 (α diagnostic ion).


193

and n‐6 positions leads to a substantial reduction in theretention time, while trans geometry in the other positionshas little effect or may give slightly higher retention times.[12]

Additionally, Fournier et al. described that the mono‐trans,cis‐4,cis‐7,cis‐10,cis‐13,cis‐16,trans‐19 22:6 (trans‐19 DHA) isthe first mono‐trans to elute in a cyanopropylpolysiloxanestationary phase (CP‐Sil 88 100‐m long).[10] Thus, wespeculate that peak B in Fig. 4 might be cis‐4,cis‐7,cis‐10,cis‐13,cis‐16,trans‐19 22:6, and peak C might be cis‐4,cis‐7,cis‐10,cis‐13,trans‐16,cis‐19 22:6. However, for correct iden-tification of 22:6 geometric isomers further studies arerequired.


Partial hydrogenation of 22:6n‐3 may in theory produce upto 12 positional isomers that originate from the isomerizationof one cis double bond to form a trans double bond. Usingpreviously reported rules for the fragmentation of homo-allylic and conjugated double‐bond systems in CACIMS/MS,[38,39] theoretical structures for these 12 positional isomerswere generated and evaluated for expected fragments thatwould be produced in CACIMS/MS. For each theoreticalstructure, prominent ions were determined and comparedwith the mass spectra for the three partially conjugated 22:6isomers presented in Figs. 6(f)–6(h). The spectra did not showprominent peaks as expected for the theorized structures.


9

0

214230

242

256

268286

326

364

397

f

150 200 250 300 350 m/z

30

Rel

ativ

e ab

unda

nce

(%)

298312

190 134

200228

268

300

322

342

364

397

g

100 150 200 250 300 350 m/z0

30

Rel

ativ

e ab

unda

nce

(%)

254

240

188

202228

254 300

322 342

364

397

282188

h

100 150 200 250 300 350 m/z0

30

314

214

200 268

Rel

ativ

e ab

unda

nce

(%)

Figure 6. CACIMS/MS spectra for partially conjugated isomers of DHA identified as peaks f–h in Fig. 4.

S. P. Alves et al.

1940

This may indicate that novel rules apply to the fragmentationof partially conjugated or NMI 22:6 positional isomers.However, in Fig. 4, peaks f–h are relatively small and wecannot rule out the potential for co‐elution of one or more22:6 positional and geometrical isomers, leading to super-imposed mass spectra that are largely un‐interpretable. Thus,based on the data available, we cannot definitively assigndouble‐bond positions to the three partially conjugated 22:6positional isomers.

CONCLUSIONS

Acetonitrile CACIMS/MS was applied for the identificationof several geometrical and positional isomers of HUFA withmore than four double bonds. These isomers were producedby the partial hydrogenation of a DHA‐rich microalgae oil.The single‐stage mass spectra provided structural informa-tion related to the presence of homoallylic and partiallyconjugated double‐bond systems. CACIMS/MS was effectivein the characterization of double‐bond position for the fournovel NMI 22:5 isomers, and yielded virtually indistinguish-able mass spectra for the four geometrical isomers of DHA.Acetonitrile CACI‐MS/MS fragments are not specific to thegeometry of double bonds of 22:6 isomers. The CACIMS/MSspectra of the positional isomers of DHA did not showdiagnostic ions corresponding to the previously reportedrules for the fragmentation of homoallylic and conjugateddouble‐bond systems, possibly reflecting chromatographicoverlap. Further work is necessary to evaluate and determinerules for the fragmentation of partially conjugated and NMI22:6 positional isomers.

AcknowledgementsS.P. Alves acknowledges the Fundação para a Ciência e aTecnologia (FCT), Ministério da Ciência, Tecnologia e EnsinoSuperior, Portugal, for financial support and PhD grant(SFRH/BD/37793/2007). This project was supported in partby Award T32DK007158 from the National Institute ofDiabetes and Digestive and Kidney Diseases of the NationalInstitutes of Health. The content is solely the responsibility ofthe authors and does not necessarily represent the officialviews of the National Institutes of Health.


REFERENCES

[1] L. A. Horrocks, Y. K. Yeo. Pharmacol. Res. 1999, 40, 211.[2] S. M. Teegala, W. C. Willett, D. Mozaffarian. J. AOAC Int.

2009, 92, 1250.[3] A. J. Dijkstra. Eur. J. Lipid Sci. Technol. 2006, 108, 249.[4] D. Evans, J. A. Osborn, F. H. Jardine, G. Wilkinson. Nature

1965, 208, 1203.[5] J. F. Young, J. A. Osborn, F. H. Jardine, G. Wilkinso. Chem.

Commun. 1965, 131.[6] R. Adlof. Chem. Phys. Lipids 1997, 88, 107.[7] E. Emken. JAOCS 1988, 65, 373.[8] D. D. Andjelkovic, B. Min, D. Ahn, R. C. Larock. J. Agric.

Food Chem. 2006, 54, 9535.[9] A. August, C. J. Dao, D. Jensen, Q. Zhang, P. Dea.

Microchem. J. 1993, 47, 224.[10] V. Fournier, P. Juaneda, F. Destaillats, F. Dionisi, P. Lambelet,

J. L. Sebedio, O. Berdeaux. J. Chromatogr. A 2006, 1129,21.

[11] V. Fournier, F. Destaillats, B. Hug, P. A. Golay, F. Joffre,P. Juaneda, E. Semon, F. Dionisi, P. Lambelet, J. L. Sebedio,O. Berdeaux. J. Chromatogr. A 2007, 1154, 353.

[12] S. A. Mjos, M. Solvang. Eur. J. Lipid Sci. Technol. 2006, 108,589.

[13] S. A. Mjos. Eur. J. Lipid Sci. Technol. 2008, 110, 547.[14] S. A. Mjos. J. Chromatogr. A 2005, 1100, 185.[15] G. Dobson, W. W. Christie. Trends Anal. Chem. 1996, 15, 130.[16] W. W. Christie. Chem. Phys. Lipids 1998, 94, 35.[17] O. Berdeaux, R. L. Wolff. JAOCS 1996, 73, 1323.[18] W. W. Christie, G. W. Robertson, W. C. McRoberts, J. T. G.

Hamilton. Eur. J. Lipid Sci. Technol. 2000, 102, 23.[19] J. T. G. Hamilton, W. W. Christie. Chem. Phys. Lipids 2000,

105, 93.[20] F. Destaillats, O. Berdeaux, J. L. Sebedio, P. Juaneda, S.

Gregoire, J. M. Chardigny, L. Bretillon, P. Angers. J. Agric.Food Chem. 2005, 53, 1422.

[21] W. W. Christie, E. Y. Brechany, M. S. F. L. Jie. Chem. Phys.Lipids 1988, 46, 225.

[22] W. W. Christie, E. Y. Brechany, F. D. Gunstone, M. S. F. L. Jie,R. T. Holman. Lipids 1987, 22, 664.

[23] W. W. Christie, E. Y. Brechany, R. T. Holman. Lipids 1987, 22,224.

[24] T. W. Mitchell, H. Pham, M. C. Thomas, S. J. Blanksby. J.Chromatogr. B Analyt. Technol. Biomed. Life Sci. 2009, 877,2722.

[25] N. Zehethofer, D. M. Pinto. Anal. Chim. Acta 2008, 627, 62.[26] W. J. Griffiths. Mass Spectrom. Rev. 2003, 22, 81.[27] E. Pittenauer, G. Allmaier. J. Am. Soc. Mass Spectrom. 2009,

20, 1037.



[28] K. B. Tomer, F. W. Crow, M. L. Gross. J. Am. Chem. Soc. 1983,105, 5487.

[29] F. F. Hsu, J. Turk. J. Am. Soc. Mass Spectrom. 2008, 19, 1673.[30] J. S. Crockett, M. L. Gross, W. W. Christie, R. T. Holman.

J. Am. Soc. Mass Spectrom. 1990, 1, 183.[31] F. F. Hsu, J. Turk. J. Am. Soc. Mass Spectrom. 2008, 19, 1681.[32] M. C. Thomas, T. W. Mitchell, S. J. Blanksby. J. Am. Chem.

Soc. 2006, 128, 58.[33] M. C. Thomas, T. W. Mitchell, D. G. Harman, J. M. Deeley,

R. C. Murphy, S. J. Blanksby. Anal. Chem. 2007, 79, 5013.[34] M. C. Thomas, T. W. Mitchell, D. G. Harman, J. M. Deeley,

J. R. Nealon, S. J. Blanksby. Anal. Chem. 2008, 80, 303.[35] B. L. J. Poad, H. T. Pham, M. C. Thomas, J. R. Nealon,

J. L. Campbell, T. W. Mitchell, S. J. Blanksby. J. Am. Soc.Mass Spectrom. 2010, 21, 1989.

[36] A. L.Michaud, G. Y. Diau, R. Abril, J. T. Brenna.Anal. Biochem.2002, 307, 348.


[37] C. K. Van Pelt, J. T. Brenna. Anal. Chem. 1999, 71, 1981.[38] C. K. Van Pelt, B. K. Carpenter, J. T. Brenna. J. Am. Soc. Mass

Spectrom. 1999, 10, 1253.[39] A. L. Michaud, P. Lawrence, R. Adlof, J. T. Brenna. Rapid

Commun. Mass Spectrom. 2005, 19, 363.[40] A. L. Michaud, M. P. Yurawecz, P. Delmonte, B. A. Corl,

D. E. Bauman, J. T. Brenna. Anal. Chem. 2003, 75, 4925.[41] P. Lawrence, J. T. Brenna. Anal. Chem. 2006, 78, 1312.[42] W. W. Christie. Lipid Analysis – Isolation, Separation, Identifica-

tion and Structural Analysis of Lipids, (3rd edn.), The Oily Press,Bridgwater, 2003.

[43] S. P. Alves, R. J. B. Bessa. J. Chromatogr. A 2009, 1216,5130.

[44] V. Fournier, F. Destaillats, P. Juaneda, F. Dionisi, P. Lambelet,J. L. Sebedio, O. Berdeaux. Eur. J. Lipid Sci. Technol. 2006,108, 33.


1941

Documents

Acetonitrile covalent adduct chemical ionization tandem ... · Acetonitrile covalent adduct chemical ionization tandem mass spectrometry of non‐methylene‐interrupted pentaene