Metabolomics DOI 10.1007/s11306-013-0569-y

ORIGINAL ARTICLE

Annotating unknown components from GC/EI-MS-based metabolite profiling experiments using GC/APCI(+)-QTOFMS Nadine Strehmel • Joachim Kopka Dierk Scheel • Christoph Bo¨ttcher

•

Received: 12 April 2013 / Accepted: 11 July 2013  Springer Science+Business Media New York 2013

Abstract GC/EI-MS-based metabolite profiling of derivatized polar fractions of crude plant extracts typically reveals several hundred components. Thereof, only up to one half can be identified using mass spectral and retention index libraries, the rest remains unknown. In the present work, the utility of GC/APCI(?)-QTOFMS for the annotation of unknown components was explored. Hence, EI and APCI(?) mass spectra of *100 known components were extracted from GC/EI-QMS and GC/APCI(?)-QTOFMS profiles obtained from a methoximated and trimethylsilylated root extract of Arabidopsis thaliana. Based on this reference set, adduct and fragment ion formation under APCI(?) conditions was examined and the calculation of elemental compositions evaluated. During these studies, most of the components formed dominating protonated molecular ions. Despite the high mass accuracy (|Dm| B 3 mDa) and isotopic pattern accuracy (mSigma B 30) the determination of a component’s unique native elemental composition requires additional information, namely the number of trimethylsilyl and methoxime moieties as well as the analysis of corresponding collision-induced dissociation (CID) mass spectra. After all, the reference set was used to develop a strategy for the pairwise assignment of EI and APCI(?) mass spectra. Proceeding Electronic supplementary material The online version of this article (doi:10.1007/s11306-013-0569-y) contains supplementary material, which is available to authorized users. N. Strehmel  D. Scheel  C. Bo¨ttcher (&) Department of Stress and Developmental Biology, Leibniz Institute of Plant Biochemistry, Weinberg 3, 06120 Halle/Saale, Germany e-mail: [email protected] J. Kopka Max Planck Institute of Molecular Plant Physiology, Am Mu¨hlenberg 1, 14476 Golm, Germany

from these findings, the annotation of unidentified components detected by GC/EI-QMS using GC/APCI(?)-QTOFMS and corresponding deuterated derivatization reagents was attempted. For a total of 25 unknown components, pairs of EI and APCI(?) mass spectra were compiled and elemental compositions determined. Integrative interpretation of EI and CID mass spectra resulted in 14 structural hypotheses, of which seven were confirmed using authenticated standards. Keywords Atmospheric pressure chemical ionization  Electron ionization  Gas chromatography/mass spectrometry  Metabolite profiling  Quadrupole timeof-flight mass spectrometry  Structural elucidation

1 Introduction Metabolomics, the comprehensive detection, quantification and identification of low-molecular weight compounds in complex mixtures of biological origin is gaining increasing importance in many branches of life science. Several hyphenated mass spectrometry (MS) platforms have become key technologies for metabolomic analyses due to their high sensitivity, high selectivity and broad applicability for many compound classes (Dunn 2008). Among them, gas chromatography coupled to mass spectrometry (GC/MS) has become a widely spread routine technology, in particular for the profiling of primary metabolites (Fiehn et al. 2000a; Kopka 2006). To facilitate the gas chromatographic separation of these prevailing polar and non-volatile target compounds a derivatization reaction is required prior to analysis, predominantly by methoximation and trimethylsilylation (Dunn et al. 2011; Lisec et al. 2006). Currently, the most popular ionization technique for GC/MS-based metabolite profiling studies is electron ionization (EI). As a hard

123

N. Strehmel et al.

ionization technique, 70 eV EI induces characteristic and often complex fragmentation reactions. However, its reproducibility using a standardized ionization energy permits the establishment of mass spectral libraries and their exchangeability between analytical laboratories (Strehmel et al. 2008). In combination with deposited retention indices (RI), such reference libraries represent powerful tools for component identification of GC/EI-MS-based metabolite profiling experiments, e.g., (Kopka et al. 2005; Sanchez et al. 2010; Schauer et al. 2005; Steinfath et al. 2010). Currently, mass spectral libraries with focus on primary metabolism such as the Golm Metabolom Database (Hummel et al. 2007, 2010) or the FiehnLib (Kind et al. 2009) archive mass spectra and RI of methoxime/trimethylsilyl derivatives of [1000 commercially available reference substances. Although the structural diversity of these reference substances covers main parts of metabolic pathway databases, typically more than half of the deconvoluted mass spectra cannot be identified in metabolite profiling studies by library matching (Dunn et al. 2013). Given the high number of unknown components detectable in GC/EI-MS-based metabolite profiling experiments the need for their comprehensive annotation and subsequent identification is obvious. A starting point for the identification of an unknown component consists in the determination of its elemental composition, which implies the detection and identification of the molecular/quasi-molecular ion or of defined adduct/ fragment ions. Using the so far established GC/MS-based metabolomics workflow this initial step remains challenging, because EI of many trimethylsilyl derivatives leads to extensive fragmentation and results in low abundant or even absent molecular or [M-CH3]? ions (Halket and Zaikin 2003). To overcome this problem, the use of tert-butyldimethylsilyl derivatives has been proposed (Fiehn et al. 2000b). In comparison to trimethylsilyl derivatives, tertbutyldimethylsilyl derivatives are less prone to extensive fragmentation under EI conditions and form [M-C4H9]? ions with a high relative abundance. These ions can be used to determine the elemental composition of a component. However, due to the lack of comprehensive mass spectral and RI libraries and incomplete derivatization, in particular for polyols the use of tert-butyldimethylsilyl derivatives for metabolite profiling is problematic. Other strategies to preserve molecular/quasi-molecular ions for the identification of unknown components rely on the use of trimethylsilyl derivatives in conjunction with softer ionization methods, such as positive ion chemical ionization [CI(?)]. The analytical potential of GC/CI(?)-TOFMS in metabolomic studies was demonstrated by determining the elemental composition of known methoximated/trimethylsilylated metabolites using accurate mass and relative isotopic abundance of protonated molecular ions (Kumari et al. 2011). The utility of CI(?) as a complementary ionization technique to

123

EI for GC/MS-based metabolite profiling was explored recently (Warren 2013). Using reagent gases with varying proton affinity, namely methane and ammonia, CI(?) mass spectra with distinct fragmentation degree were obtained. In conjunction with EI data, CI(?) mass spectra were used to confirm tentative identifications, to detect coeluting analytes and to identify unknown analytes from complex samples. Among other soft ionization techniques applicable for the coupling of GC to MS, e.g., cold EI (Amirav et al. 2008), field ionization, photo ionization (Zimmermann et al. 2008) and electrospray ionization (Brenner et al. 2008), atmospheric pressure chemical ionization (APCI) has already emerged decades ago as a promising alternative (Horning et al. 1973). However, multipurpose APCI interfaces for the coupling of LC and GC to MS only became commercially available over the past years (McEwen and McKay 2005; Schiewek et al. 2008). In particular in combination with high-resolution TOFMS, the application of GC/APCI(?)-MS for the screening and identification of pesticides (Portoles et al. 2010), impurity identification (Bristow et al. 2010), food analysis (Garcia-Villalba et al. 2011), as well as metabolite profiling (Carrasco-Pancorbo et al. 2009; Pacchiarotta et al. 2010; Wachsmuth et al. 2011) has been demonstrated. Positive ion APCI of methoximated/trimethylsilylated primary metabolites generates protonated molecular ions, which can be used to determine the elemental composition (Carrasco-Pancorbo et al. 2009; Wachsmuth et al. 2011). In addition, GC/APCI(?)-TOFMS offers an excellent sensitivity for the analysis of such derivatives. According to Wachsmuth et al. (2011) it is significantly higher than that of GC/EI-TOFMS, GC/EI-QMS and GC/CI(?)-QMS. Given this analytical potential, we explored the use of GC/ APCI(?)-QTOFMS as a complementing tool for the annotation and identification of unknown components within the frame of a standardized GC/EI-MS-based metabolite profiling experiment. Therefore, the applicability of APCI(?)-TOF mass spectra for the determination of elemental compositions was evaluated using a chemically diverse set of components. A strategy for the pairwise assignment of deconvoluted EI and APCI(?) mass spectra was established and used to exploit complementary information of both analytical platforms, namely the EI mass spectrum as well as the elemental composition and the high-resolution collision-induced dissociation (CID) mass spectrum with the objective to derive structural hypotheses of so far unknown components.

2 Materials and methods 2.1 Materials Water, chloroform, toluene, dichloromethane (all LiChrosolv) and pyridine (p.a.) were purchased from Merck,

Annotating unknown components

methanol (LC–MS CHROMASOLV) from Fluka. N,O-Bis (trimethylsilyl)trifluoroacetamide (BSTFA) was obtained from Macherey–Nagel and N-methyl-N-(trimethyl-D9silyl)trifluoroacetamide (MSTFA-D9) from Fluka. Methoxyamine-D3 (MeOx-D3) hydrochloride was from CDN Isotopes. All other reagents and reference compounds were purchased from Sigma-Aldrich/Fluka. Two mixtures of retention index markers were prepared. For the n-alkane mixture, dodecane, pentadecane, nonadecane, docosane, octacosane and dotriacontane were dissolved in pyridine each to a final concentration of 80 lg mL-1. The fatty acid methyl ester (FAME) mixture was obtained from a commercially available C4–C24 even-numbered, saturated straight-chain FAME mixture (Supelco) which was diluted with dichloromethane to a final concentration of 100 lg mL-1 for each component. For mass calibration of the APCI(?)-QTOF instrument a C5–C31 uneven-numbered, saturated straight-chain fatty acid mixture [C5 (50 mg L-1); C7, C9 (40 mg L-1); C11, C13 (25 mg L-1); C15 (20 mg L-1); C17 (15 mg L-1); C19, C21, C23, C25, C27, C29, C31 (10 mg L-1)] in toluene was prepared. 9-(Methylthio)nonanenitrile and 4-methoxyindole-3acetonitrile were prepared by Fe2?-mediated degradation of 8-(methylthio)octyl glucosinolate and 4-methoxy-indol3-ylmethyl glucosinolate (10 mM FeSO4, 10 mM sodium acetate buffer, pH 5) (Bellostas et al. 2008). Both glucosinolates were isolated from Arabidopsis thaliana leaves (accession Col-0) following a procedure described by (Pfalz et al. 2009). The identity of compounds was confirmed by UPLC/ESI-QTOFMS. N-Palmitoyl piperidine was prepared by reaction of palmitoyl chloride (1 mmol) and piperidine (2 mmol) in chloroform (2 mL). O-Palmitoyl diethylene glycol was obtained by reaction of palmitoyl chloride (1 mmol) and diethylene glycol (1 mmol) in the presence of pyridine (2 mmol) in chloroform (2 mL). Arabidopsis thaliana (accession Col-0) was grown in a hydroponic culture system under sterile conditions. Surface-sterilized seeds were individually sown on bottom-cut 0.2-mL PCR tubes, which had been filled with 0.5 % (w/v) agarose before. PCR tubes were fitted into pipette tip boxes filled with 180 mL one-half Murashige and Skoog medium (Duchefa, pH 5.8) containing 0.5 % (w/v) sucrose. Tip boxes were incubated in a growth cabinet at a photoperiod/ temperature of 8 h light/23 C (180–220 lE m-2 s-1) and 16 h dark/21 C. After 3 weeks, individual PCR tubes were transferred to brown 50-mL flasks with perforated cap which had been filled with nutrient solution (von Wiren et al. 1995). Flasks were arrayed in lockable plastic boxes and incubated for additional 3 weeks in a growth cabinet. The nutrient solution was exchanged on a weekly basis. After a total of 6 weeks after sowing, roots were harvested, gently blotted dry using paper towels, shock frozen in liquid nitrogen and stored at –80 C until extraction.

2.2 Extraction and derivatization A pooled sample of root material from 36 plants was ground in liquid nitrogen using a pestle and mortar and aliquots of (50 ± 3) mg were weighed into pre-cooled 2-mL polypropylene tubes. After addition of 360 lL cold methanol (4 C), the samples were thoroughly vortexed and subsequently shaken for 15 min at 70 C. After cooling to room temperature, chloroform (200 lL) was added and samples were shaken at 37 C for additional 5 min. After addition of water (400 lL) and vortexing, phase separation was performed by centrifugation for 5 min at 16,0009g. Multiple preparations of the upper polar phase were pooled. Aliquots (150 lL) were evaporated until dryness under reduced pressure at 40 C using a vacuum centrifuge and stored at –80 C until analysis. A freshly prepared solution (40 lL) of MeOx hydrochloride or MeOx-D3 hydrochloride in pyridine (20 mg mL-1) was added to the dried extract and the mixture shaken at 37 C for 90 min. Afterwards, 10 lL of the correspondig RI mixture and 70 lL BSTFA or MSTFA-D9 were added. The reaction mixture was shaken at 37 C for additional 30 min and finally transferred to a GC vial. 2.3 GC/EI-QMS Gas chromatographic separations were performed on an Agilent 6890N GC equipped with a split/splitless inlet and a ZB-5 column (Phenomenex, 30 m length, 0.25 mm internal diameter, 0.25 lm 95 % dimethyl/5 % diphenyl polysiloxane film, 10 m integrated guard column). Derivatized samples (1 lL) were injected in splitless mode using an injector temperature of 230 C. Helium was used as carrier gas at a constant flow rate of 1 mL min-1 through the column. The purge time was set to 1 min at a purge flow of 20 mL min-1. After injection, the oven temperature was kept at 70 C for 1 min and subsequently raised to 300 C at 9 C min-1 where it was held for 5 min. Eluting components were detected after a solvent delay of 6 min using an Agilent 5975 Series Mass Selective Detector. The transfer line temperature was set at 300 C during the entire analysis. EI at 70 eV was employed at an ion source temperature of 230 C. Mass spectra were recorded from m/z 70–600 at a scan rate of 3 s-1. The mass spectrometer was autotuned and mass calibrated according to the manufacturer0 s recommendations using perfluorotributylamine (FC-43). Data deconvolution was performed using AMDIS 2.7 (National Institute of Standards and Technology, USA) applying the following parameter settings: component width, 6; adjacent peak subtraction, 1; resolution, low; sensitivity, low; shape requirements, low. Identification of deconvoluted component mass spectra was carried out by

123

N. Strehmel et al.

mass spectral matching (reverse match [ 500) using NIST MS Search 2.0 (National Institute of Standards and Technology, USA) and RI matching (DRI B 20) on the basis of the publicly available Golm Metabolome Database (http:// gmd.mpimp-golm.mpg.de/; version April 2011).

2.4 GC/APCI(?)-QTOFMS Gas chromatographic separations were performed on an Agilent 7890A GC equipped with a split/splitless inlet and a ZB-5 column (Phenomenex, 30 m length, 0.25 mm internal diameter, 0.25 lm 95 % dimethyl/5 % diphenyl polysiloxane film, 10 m integrated guard column). The analytical column was extended by a deactivated fused silica capillary (0.5 m length, 0.25 mm internal diameter) which was fitted into the APCI source. Derivatized samples (1 lL) were injected in split mode using split ratios of 1:10 or 1:20. All other GC settings were maintained as described above. Eluting components were detected in positive ion mode using a micrOTOF-Q II hybrid quadrupole time-of-flight mass spectrometer (Bruker Daltonics) equipped with a multipurpose APCI source (Bruker Daltonics). The transfer line temperature was set at 300 C. The following instrument settings were used: sheath gas, nitrogen, 0.6 L min-1; nebulizer gas, nitrogen, 3 bar; vaporizer temperature, 300 C; dry gas, nitrogen, 2 L min-1, 200 C; capillary, –2,000 V; end plate offset, 0 V; corona, ?3,000 nA; funnel 1 RF, 200 Vpp; funnel 2 RF, 200 Vpp; in-source CID energy, 0 V; hexapole RF, 150 Vpp; quadrupole ion energy, 5 eV; collision gas, nitrogen; collision energy, 5 eV; collision RFs, 300/800 Vpp (timing 50/50); transfer time, 70 ls; pre pulse storage, 5 ls; pulser frequency, 10 kHz. Mass spectra were recorded in centroid mode from m/z 90–1,200 at a scan rate of 3 s-1. The resolving power of the TOF analyzer was RFWHM & 6,500 at m/z 300. For initial external mass calibration 1 lL of the C5–C31 fatty acid mixture was introduced manually via the vertically mounted APCI sprayer. For internal recalibration of each chromatographic analysis the calibration mixture was introduced in the same way after finishing the temperature program. For the acquisition of CID mass spectra appropriate precursor ions were selected by Q1 using a m/z-width of ±3 and fragmented inside the collision cell while applying collision energies in the range of 10–50 eV. Nitrogen was used as collision gas. Product ions were detected from m/z 50–1,200 using collision RFs of 50/150 Vpp (timing 50/50). For CID of in-source fragment ions (pseudo-MS3) the insource CID energy was individually increased. Internal recalibration and data evaluation including extraction of ion chromatograms and component mass spectra as well as calculation of elemental compositions was performed using DataAnalysis 4.0 (Bruker Daltonics).

123

3 Results and discussion 3.1 Generation of a reference set of EI and APCI(?) mass spectra using a derivatized plant extract A representative aliquot of the polar fraction of a crude root extract of A. thaliana was derivatized with MeOx/BSTFA and analyzed by GC/EI-QMS with n-alkanes as RI markers. From the resulting profile, mass spectra of 276 putative components with S/N C 50 were deconvoluted using the AMDIS software. Thereof, 116 (42 %) components were identified by mass spectral and RI matching to the publicly accessible Golm Metabolome Database (http://gmd. mpimp-golm.mpg.de). In addition to metabolite derivatives (93), RI markers (6), siloxanes (10) and other artefacts (7) were found. A list of all annotated components is presented in Online Resource 1. In parallel, the derivatized root extract was analyzed by GC/APCI-QTOFMS in positive ion mode using similar gas chromatographic conditions. Due to the noticeable increased sensitivity of this platform in comparison to GC/EI-QMS a split injection (split ratio 1:20) was required. FAMEs were used as RI markers, because n-alkanes evade ionization under APCI(?) conditions. In order to retrieve the already identified components from the acquired GC/APCI(?)QTOFMS profile, narrow window extracted ion chromatograms of the corresponding monoisotopic protonated molecular ion were generated (m/z-tolerance ±0.02). Chromatographic peaks were detected in appropriate retention time windows for a total of 102 out of 110 components. A subsequent analysis of the measured monoisotopic mass confirmed the respective elemental composition of all components. Among the eight components, which could not be retrieved from GC/APCI(?)-QTOFMS data via their protonated molecular ion four metabolite derivatives were found, namely sucrose (8TMS), trehalose (8TMS), galactinol (9TMS) and scopolin (4TMS). The remaining four components most probably represent specific contaminants of the GC/EI-QMS system and thus could not be detected by GC/APCI(?)-QTOFMS. Using authenticated standards, sucrose (8TMS) and trehalose (8TMS) were shown to be prone to extensive in-source fragmentation. Both components exclusively formed fragment ions under the applied APCI conditions. For galactinol (9TMS) and scopolin (4TMS), protonated molecular ions were detected when decreasing the split ratio for sample injection towards 1:10. As observed for sucrose (8TMS) and trehalose (8TMS), APCI(?) mass spectra of galactinol (9TMS) and scopolin (4TMS) were dominated by in-source fragment ions resulting from glycosidic bond cleavage and multiple neutral losses of trimethylsilanol (90.050 Da). For detailed information refer to Online Resource 1.

Annotating unknown components

3.2 Positive ion APCI mass spectra of methoxime/ trimethylsilyl derivatives To study APCI(?) in more detail, extracted mass spectra of all identified components (106) were analyzed. Prominent signals are summarized in Online Resource 1. For a total of 93 mass spectra, the protonated molecular ion was the base peak, for another eight components, its intensity was [40 % relative to the base peak. Thus, it can be concluded that APCI(?) of the majority of the examined components reliably results in the formation of protonated molecular ions with sufficient relative intensity. Exceptions were shikimic acid (4TMS), galactinol (9TMS) and scopolin (4TMS), whose mass spectra exhibited protonated molecular ions with relative intensities \10 %. For sucrose (8TMS) and trehalose (8TMS), protonated molecular ions could not be detected (vide supra). The formation of molecular ions in combination with protonated molecular ions was only observed for a single component, namely indole-3-acetonitrile (1TMS). Besides protonated molecular ions, [M?C3H9Si]?-type adduct ions were detected for a total of 11 components including nine trimethylsilylated metabolites with at least one carboxylic acid moiety as structural feature as well as sulphuric acid (2TMS) and phosphoric acid (3TMS). In accordance with previous findings (Wachsmuth et al. 2011), the formation of [M?C3H9Si]?-type adduct ions does not represent a general phenomenon of APCI(?) of trimethylsilyl derivatives but appears to be restricted to trimethylsilyl esters of carboxylic acids and inorganic acids such as phosphoric acid and sulphuric acid. Although APCI(?) can be considered as a soft ionization technique in-source fragment ions were observed in numerous mass spectra. Frequently occurring fragment ions are related to neutral losses well-known from corresponding CI(?) mass spectra (Warren 2013) including C3H10SiO (90.050 Da), CH4 (16.031 Da), C3H8Si (72.040 Da), C4H10SiO2 (118.045 Da) and combinations thereof. However, the degree of in-source fragmentation strongly depends on the structure of the analyte. In particular methoxime/trimethylsilyl derivatives of multiple hydroxylated metabolites such as monosaccharides, derived alcohols and acids, cyclitols, disaccharides and other O-glycosides exhibited moderate to extensive insource fragmentation impeding in some cases the detection of the protonated molecular ion. 3.3 Mass accuracy and isotopic pattern accuracy To facilitate the calculation of elemental compositions of unknown components, confidence intervals for the monoisotopic mass and the mSigma value are required, the latter characterizing the goodness of fit between the measured

and the calculated isotopic pattern. Therefore, the set of already identified protonated molecular ions covering a broad mass (m/z 167–765) and intensity range (three orders of magnitude) was evaluated (Online Resource 1). After internal recalibration of each chromatographic run the average absolute accuracy of monoisotopic mass measurements was 1.3 mDa with a standard deviation of 0.8 mDa. For more than 95 % of the measured ions, the absolute mass accuracy was found to be below 3.0 mDa irrespective of the signal-to-noise ratio (Fig. 1a). For the evaluation of the corresponding mSigma values, a total of 17 isotopic patterns were omitted from the analysis because of the presence of interfering ion species. In contrast to mass accuracy, mSigma values strongly depended on the signal-to-noise ratio (Fig. 1b). For low abundant ions (S/N \ 250 at the chromatographic peak apex), a strong spreading of mSigma values between 10 and 90 was observed. For ion intensities [ 5,000 cps (S/N [ 250), mSigma values were found to be below 30, in most cases even below 20. In general, detector saturation resulted in the distortion of the isotopic pattern and thus contributed to high mSigma values. 3.4 Positive ion APCI of trimethylsilyl esters of several carboxylic acids results in formation of an uncommon adduct-fragment ion species Among the 17 components whose protonated molecular ion was excluded from the isotopic pattern analysis the complete set of eight identified trimethylsilyl n-alkanoates was found (Online Resource 1). Interestingly, the isotopic pattern of protonated molecular ions of all trimethylsilyl n-alkanoates qualitatively exhibited the same interference which is characterized by a centroid shift to lower m/z values (Dm/z * 0.015) and increased relative intensities of the second and third isotopic peak (Fig. 2a). The systematic occurrence of this interference suggests an additional adduct-fragment ion species, which is formed besides the protonated molecular ion by APCI(?) for all trimethylsilyl n-alkanoates. Using an increased in-source CID voltage, the unknown ion species became the dominating one and accurate mass and isotopic pattern analyses suggested elemental compositions corresponding to a [M?H3O– CH4]?-type ion for all fatty acid derivatives under investigation (Fig. 2b). In addition, CID mass spectra obtained from the putative [M?H3O–CH4]?-type ions displayed a neutral loss of water as well as uniform fragment ions at m/z 131.052 (C5H11O2Si?), m/z 95.037 (C2H9O2Si?) and m/z 75.027 (C2H7OSi?) supporting a molecular structure which results from the substitution of a methyl group by a hydroxyl group at the silicon atom (Fig. 2c). Currently, it can only be speculated whether such adduct-fragment ions might arise from ion/molecule reactions between H3O? or

123

N. Strehmel et al. 100

a

b

3.0 80

2.5

mSigma

⏐m/z calc -m/zmeasured⏐/ mDa

3.5

2.0 1.5

60

40

1.0 20

0.5 0.0

0

4

6

8

10

12

14

16

log2(S/N)

4

6

8

10

12

14

16

log2(S/N)

Fig. 1 Mass accuracy and isotopic pattern accuracy of GC/APCI(?)QTOFMS. a Absolute mass accuracy of monoisotopic protonated molecular ions of 102 identified components detected by GC/ APCI(?)-QTOFMS in a methoximated/trimethylsilylated root extract of Arabidopsis thaliana in dependence on the signal-to-noise ratio. b Goodness of fit between measured and calculated isotopic pattern (mSigma obtained from DataAnalysis, Bruker Daltonics) of

protonated molecular ions of 85 identified components in dependence on the signal-to-noise ratio of the monoisotopic ion. Protonated molecular ions of 17 components were omitted from the analysis because of the presence of interfering [M?H3O–CH4]?-type ions. For protonated molecular ions labeled with solid triangles detector saturation was observed

H?(H2O)n ion clusters and neutral analyte molecules which are accountable for the formation of protonated molecular ions during APCI(?) as well. In addition to trimethylsilyl n-alkanoates, only five other components including fumaric acid (2TMS), pyroglutamic acid (1TMS), methylbenzoic acid (1TMS), 1,4-benzenedicarboxylic acid (2TMS) and sulphuric acid (2TMS) exhibited markedly formation of [M?H3O–CH4]?-type ions upon the applied APCI conditions. This suggests that the occurrence of this adduct-fragment ion species appears to be restricted to certain trimethylsilyl esters. However, a comprehensive study of structural requirements and experimental parameters (in particular the amount of residual water in the sheath gas/ion source) influencing the formation of [M?H3O–CH4]?-type ions is beyond the scope of this work.

elemental compositions were retrieved on average. If additional constraints were applied, which arise from the fact that in metabolite derivatives Si is exclusively occurring as N, O or S-bound trimethylsilyl moiety (nC C 3 nSi, 2 nN ? nO ? nS C nSi) and that the presence of a P atom requires at least the presence of three O atoms, the average number of putative elemental compositions was restricted towards seven. To further narrow down the correct elemental composition, the exact number of Si atoms needs to be determined experimentally for each component of interest. This can be achieved by a comparative analysis of the trimethylsilyl and the corresponding deuterotrimethylsilyl derivative (Herebian et al. 2005). Since BSTFA-D18 was not commercially available, the crude root extract was derivatized with MeOx/MSTFA-D9 and analyzed by GC/APCI(?)QTOFMS. In conjunction with GC/APCI(?)-QTOFMS data obtained from a sample derivatized with MeOx/ BSTFA pairs of unlabeled and deuterium labeled derivatives were identified by inspection of narrow-window extracted ion chromatograms of the corresponding monoisotopic protonated molecular ion which are consecutively shifted by integral multiples of m/z 9.0565. In addition, deuterotrimethylsilyl derivatives exhibit a characteristic shift towards shorter retention times of *5 RI units per trimethylsilyl moiety when compared to their unlabeled analog (Herebian et al. 2005). Due to the different reactivity of BSTFA and MSTFA-D9, isotopologue pairs could only be identified for 19 out of 25 test components. However, knowledge of the number of Si atoms drastically reduced the number of putative elemental compositions to a maximum of three.

3.5 Calculation of elemental compositions In order to test the reliability of GC/APCI(?)-QTOFMS data to retrieve the correct elemental composition, protonated molecular ions of 25 identified metabolite derivatives ranging from m/z 310 to m/z 722 and exhibiting sufficient intensity ([ 5,000 cps) were selected (Online Resource 2). Elemental compositions were determined using the elements C, H, N, O, S, P and Si with reasonable restrictions concerning the number of N, S and P atoms (nN B 5, nS B 2, nP B 1). In addition, an even electron number and a limited number of ring and double bond equivalents (RDBE C -0.5) were applied. Assuming the experimentally derived confidence intervals (Dm B 3 mDa, mSigma B 30) nine putative

123

Annotating unknown components Intens. x10 5

a 259.2111

2.5

calculated

259.2088

measured

O

2.0

9

O

C14H31O2Si+

261.1913

Si

260.2127

260.2120

262.1925 263.1891

1.5

[M+H]+ 259.2111

261.2058

261.1913

262.2090

[M+H-C3H10SiO]+ 169.1589

1.0 0.5

[M+C3H9Si]+ 331.2492

[M+H-CH4]+ 243.1794

[M+H-C3H8Si]+ 187.1695

0.0 100 Intens. x105

150

200

250

b 261.1896

1.25

measured

261.1880

O

1.00 0.75

9

calculated

300

m/z

[M+H3O-CH4]+ 261.1896 [M+C3H9Si]+ 331.2493

C13H29O3Si+

O

Si

259.2103

262.1916

262.1913 263.1850

263.1889

260.2126

[M+H]+ 259.2103

0.50

[M+H-C3H10SiO]+ 169.1587

0.25

131.0523

[M+H-CH4]+ 243.1785

0.00 100 Intens. x104

150

200

250

c

+H

75.0272

1.2

OH+ Si

1.0 0.8

Si OH

m/z

m/z 131

O

OH2+ 93.0368

300

O

O

Si

O

OH2+ Si

9 m/z 261

m/z 243

0.6

131.0523 0.4

261.1885 0.2

111.0477

57.0731

243.1772

0.0 50

75

100

125

150

175

200

225

250

m/z

Fig. 2 Formation of [M?H3O–CH4]?-type ions from trimethylsilyl alkanoates under positive ion APCI exemplified for trimethylsilyl undecanoate. Mass spectra were obtained using GC/APCI-QTOFMS. a Positive ion APCI mass spectrum of trimethysilyl undecanoate acquired with an in-source CID voltage of 0 V. b Positive ion APCI

mass spectrum of trimethysilyl undecanoate acquired with an in-source CID voltage of 55 V. c CID mass spectrum (collision energy 25 eV) obtained from the putative [M?H3O–CH4]?-type ion of trimethylsilyl undecanoate (m/z 261) formed after applying an in-source CID voltage of 55 V

3.6 Pairwise assignment of EI and APCI(?) mass spectra

was established using RI pairs of *100 identified components detected by both platforms. Applying a 3rd order polynomial regression (Strehmel et al. 2008) the search for a proper APCI(?) mass spectrum can be restricted to a RI range of ±15 units (Online Resource 3). Based on the unidentified EI mass spectrum in question, a molecular mass needs to be hypothesized, which is used to screen GC/APCI(?)-QTOFMS data for a corresponding protonated molecular ion within the predicted RI range.

In order to exploit GC/APCI(?)-QTOFMS data for the annotation of unidentified components detected during GC/ EI-QMS experiments an unambiguous pairwise assignment between both types of mass spectra is prerequisite. For this purpose, a prediction model for FAME-based RI (GC/ APCI(?)-QTOFMS) from n-alkane-based RI (GC/EI-QMS)

123

N. Strehmel et al.

GC/EI-QMS [%]

GC/APCI(+)-QTOFMS [%]

147

a

RI(ALKANE) = 1532

80 73

80

TMS

60

1 [M-CH3]+ 317

40 189 202

20 0

100 131 86

100 [%]

150

d

RI(FAME) = 1205 (predicted RI = 1203)

60

205.1002

20

217

200

250

300

350

400

450

500

550

0 100

m/z

[%]

162

RI(ALKANE) = 1516

c

80

60

3

[M-CD3]+ 341

207

82

333.1502 [M+H]+

243.0969 [M+H-TMSOH]+

40

80

TMS-D9

b

40

150

200

250

300 350 2

261.2111 [M+H-TMS-D9OH]+

400

360.3193 [M+H]+

450

500

550 m/z

RI(FAME) = 1190

179.1913

60 40

217

20

20 106

142

189

232 261

0

0

100

150

200

250

300

350

400

450

500

550 m/z

100

150

200

250

300

350

400

450

500

550 m/z

Fig. 3 Pairwise assignment of the EI and APCI(?) mass spectrum of unknown component 12 which was identified afterwards as 5-hydroxyhydantoine (3TMS). a An unidentified EI mass spectrum is deconvoluted from GC/EI-QMS data obtained from a root extract derivatized with MeOx/BSTFA at an alkane-based RI of 1532. Based on a hypothetical [M-CH3]? ion at m/z 317 a molecular mass of 322 Da is postulated. b Using a polynomial regression model a FAME-based RI of 1203 ± 15 is predicted for the unknown component. GC/APCI(?)-QTOFMS data obtained from a root extract derivatized with MeOx/BSTFA is screened for a corresponding protonated molecular ion, which is detected at m/z 333.1502 and a

FAME-based RI of 1205. c GC/APCI(?)-QTOFMS data obtained from a root extract derivatized with MeOx/MSTFA-D9 is screened for a corresponding isotopologue protonated molecular ion, which is detected at m/z 360.3193 and a FAME-based RI of 1190 (Dm/z = 27.169; DRI = 15). Hence, the unknown component contains three trimethylsilyl moieties. d The pairwise assignment of EI and APCI(?) mass spectrum shown in a and b is confirmed by detection of a corresponding shifted [M-CD3]? ion at m/z 341 and an alkane-based RI of 1516 (Dm/z = 24; DRI = 16) in GC/EI-QMS data obtained from a root extract derivatized with MeOx/MSTFA-D9

Trimethylsilyl derivatives usually exhibit low abundant or absent molecular ions under 70 eV EI conditions. Nevertheless, a molecular mass can be hypothesized in numerous cases due to the presence of [M-CH3]? ions. To assure a pairwise assignment of EI and APCI(?) mass spectra relying on the matching of [M-CH3]? and [M?H]? ions GC/MS data of deuterotrimethylsilyl derivatives can be employed (Fig. 3). For this end, the isotopologue protonated molecular ion is firstly identified by comparison of GC/APCI(?)QTOFMS data of trimethylsilylated and deuterotrimethylsilyated components. With the help of the deduced number of trimethylsilyl groups, the identity of the putative [M-CH3]? ion can be confirmed by identification of the corresponding m/z and RI-shifted isotopologue [M-CD3]? ion from GC/EIQMS data of a deuterotrimethylsilylated extract. To test whether EI and APCI(?) mass spectra of the identified components are pairwise assignable following the above described strategy, deconvoluted EI mass spectra of 93 metabolite derivatives and 7 artifacts were investigated (Online Resource 1). A molecular ion along with a

[M-CH3]? ion could be detected for 34 components whereas mass spectra of 42 components displayed a [M-CH3]? ion only. Of the total of these 76 components, 74 formed a protonated molecular ion under APCI(?) conditions, which should in principle allow a pairwise assignment of the EI and APCI(?) mass spectrum. For the remaining 24 components including derivatized sugars, sugar alcohols, acids and phosphates, as well as disaccharides, a molecular mass could not be hypothesized from the EI mass spectrum, which complicates the retrieval of the corresponding APCI(?) mass spectra. In such cases, a pair of consistent EI and in-source CID fragment ions can— assuming its presence—be used for the mapping. In particular, pertrimethylsilylated organic phosphates frequently form fragment ions at m/z 315 (C9H28O4PSi3?) and/or m/z 299 (C8H24O4PSi3?) under both EI and APCI(?) conditions (Online Resource 1). In addition, EI and APCI(?) mass spectra of the investigated disaccharide derivatives can be assigned on the basis of a consistently occurring fragment ion at m/z 361 (C15H33O4Si3?).

123

Annotating unknown components

3.7 Annotation and identification of unknown components Among the 276 mass spectra, which were deconvoluted from the GC/EI-QMS profile of the derivatized root extract, a total of 160 remained without any annotation after mass spectral and RI matching. As proof of concept, EI mass spectra of prominent unknown components (Online Resource 4) were attempted to annotate with the help of GC/APCI(?)-QTOFMS and deuterated derivatization reagents. Following the above described strategy, pairs of EI and APCI(?) mass spectra were unambiguously identified for 25 unknown components (Table 1). In this process, the pairwise assignment of EI and APCI(?) mass spectra was based on the matching of the molecular and/or [M-CH3]? ion with the corresponding protonated molecular ion for 22 out of 25 components and for additional two components (22, 24) on the matching of consistent EI and in-source CID fragment ions. In case of the putative thiohexose derivative 18 a matching of [M-90]?•/[M-105]? and [M?H]? ions allowed the identification of the corresponding APCI(?) mass spectrum. It should be noted, that EI mass spectra and RI of 13 out of 25 unknown components are catalogued within the Golm Metabolome Database. To determine the number of trimethylsilyl and methoxime moieties, GC/APCI(?)-QTOFMS data of differentially derivatized root extracts (MeOx/BSTFA, MeOx-D3/ BSTFA and MeOx/MSTFA-D9) were compared. Whereas the number of methoxime moieties could be deduced for all unknown components, the number of trimethylsilyl moieties could only be retrieved for 17 components. These were detected after reaction with MeOx/MSTFA-D9. Using the identified protonated molecular ions, elemental compositions were determined and filtered according to the above-mentioned criteria (Online Resource 5). Here, unique elemental compositions were obtained for 10 out of 25 unknown components. In order to validate them and to resolve the elemental composition of the remaining unknown components, CID mass spectra of protonated molecular ions of all unknown components were acquired and analyzed (Online Resource 6). A relatively large precursor ion isolation width (m/z ± 3) was applied in order to use the isotopic pattern of fragment ions for the calculation of the elemental composition. Following an approach already described by Konishi et al. (2007), unique elemental compositions of the protonated molecular ion of all unknown components were compiled after analyzing the elemental composition of fragment ions and neutral losses within the CID mass spectra obtained from the protonated molecular ion and their isotopologues (Table 1). After removal of the trimethylsilyl and methoxime moieties native elemental compositions were queried for in

general (e.g., SciFinder, PubChem, ChemSpider) or biochemical (e.g., KEGG, AraCyc, KNApSAcK) compound databases. These typically retrieve a multitude of hits. Hence, the assembly of reasonable structural hypotheses requires the mutual interpretation of EI and CID mass spectra as well as the consideration of other analytical information, such as the derivatization and chromatographic behaviour. Following this approach, for 14 out of 25 unknown components structure or substructure hypotheses were postulated, of which 7 were confirmed by authenticated standards (Table 1). In the following, the identification of two components (20, 22) will be described. 3.7.1 Case study 1 The EI (Fig. 4a, GMD identifier A210003) and APCI(?) mass spectrum (Fig. 4b) of 20 could be pairwise assigned due to the presence of M?•/[M-CH3]? and [M?H]? ions. The application of MeOx-D3/BSTFA and MeOx/MSTFA-D9 revealed that 20 carries only one trimethylsilyl moiety and confirmed the assignment of EI and APCI(?) mass spectrum. Subsequently, the unique elemental composition C14H19N2OSi? was calculated using the protonated molecular ion. The ChemSpider database (www.chemspider.com) returned 685 hits upon querying the native elemental composition of 20 (C11H10N2O, 8 RDBE). In contrast, the KNApSAcK database of phytochemicals (http://kanaya. naist.jp/knapsack_jsp/top.html) returned two hits for A. thaliana, namely 4-methoxy-indole-3-acetonitrile and 1-methoxy-indole-3-acetonitrile. Due to the lack of a derivatizable heteroatom, the latter can be excluded. The CID mass spectrum of 20 (Fig. 4c) showed neutral losses of CH4 (16.032 Da), HCN (27.010 Da), C2H2N• (40.019 Da) and C4H9NSi (99.050 Da) from the protonated molecular ion. These are consistent with a substituted indole-3-acetonitrile derivative. In particular, the neutral loss of C4H9NSi results in formation of a resonance-stabilized skatyl cation (m/z 160.076), which is also observed as a corresponding radical cation (m/z 159) within the EI mass spectrum. 4-Methoxy-indole-3-acetonitrile was prepared by Fe2?mediated degradation of 4-methoxy-indol-3-ylmethyl glucosinolate (Bellostas et al. 2008) and used for authentication. 3.7.2 Case study 2 The EI (Fig. 4d, GMD identifier A260006) and APCI(?) mass spectrum (Fig. 4e) of 22 could be pairwise assigned due to the presence of a consistently occurring fragment ion at m/z 283. In that case, solely the APCI(?) mass spectrum allowed the deduction of a molecular mass. The application of deuterated derivatization reagents only revealed the number of methoxime moieties (0) since 22 was not formed

123

123 B A A

A114002

A116012

–

–

–

–

A140018

A144008

A145015

–c

A167005

A178002

–d

A186003

–

A203003

–

A210003

–

A260006

–

A278013

–

1143

1175

1182

1214

1254

1275

1419

1443

1470

1532

1647

1758

1806

1886

1888

2068

2072

2096

2111

2613

2655

2797

2985

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

2691

2458

2309

2267

1784

1770

1741

1737

1562

1561

1477

1427

1320

1205

1148

1122

1097

954

932

894

860

854

820

811

-115

-44

-20

-17

-7

-7

-2

-2

-4

-4

0

3

0

-2

-6

-6

-5

-7

-6

-8

-6

-6

-7

-5

0

DRIg

531.2613

445.3728

324.3281

417.3411

218.1032

259.1270

382.1528

557.2461

409.1919

423.1706

379.1814

187.0865

186.1326

333.1504

275.1982

263.1140

277.1297

202.1255

283.1458

265.1287

253.1344

230.1760

236.1145

204.1416

284.1125

[M?H]? m/z

0 0 0

–i –i –i

Reverse match of 797 to EI mass spectra of 2-C-methylribonic acid c-lactone (3TMS) from NIST library

0

0

–i

2

0

1

0

0

0

0

0

0

1

2

5

3

3

3

0

0

0

1

–i 3

0

0

2

2

0

0

–i 1

0

0

–i 2

0

1

2

2

0

0

–i 1

MeOxj

TMSh

Reverse match of 711 to EI mass spectra of 5-hydroxyhydantoine (3TMS) from NIST library

Golm Metabolome Database identifier

Measured n-alkane-based RI used for GC/EI-QMS

B

E

D

E

A

A

A

C

A

A

A

A

A

A

A

A

A

A

A

A

735

RI FAMEf

-1.1

C11H25O4Si2? C10H23O4Si2?

-2.7 -0.6 -2.8 -0.8

C16H37O6Si3? C21H53O5SSi5? C17H28NO5Si2? C14H19N2OSi?

-2.1 -2.0

C28H43O6Si2?

-2.1

-1.6

C25H53O4Si?

C21H42NO?

C23H49O4Si?

-0.4

-2.1

C9H20NOSSi?

-2.7

0.1

C11H11N2O? C16H35O7Si3?

-1.5

C10H20NS? C15H35O5Si3?

-2.4

-1.3

C12H29N2O3Si3?

C12H31N2OSi2?

0.2

C9H20NO2Si? -1

1.4

C13H23N2O3Si?

-0.1

-0.5

C11H28NSi2? 2.3

-1.3

C8H22NO3Si2?

C10H25O4Si2?

-0.2

C12H21N2O2Si?

0.8

C9H22NO2Si?

Dm (mDa)

C12H22NO3Si2?

Elem. comp. [M?H]?

18

16

10

10

2

5

12

4

2

4

8

5

7

5

6

12

3

27

14

12

12

11

8

6

24

mSig

C22H26O6

C22H44O4

C21H41NO

C20H40O4

C6H11NOS

C11H10N2O

C10H8O5

C6H12O5S

C7H12O6

C7H10O7

C6H10O5

C11H10N2O

C10H19NS

C3H4N2O3

C5H11NO

C4H6O4

C5H8O4

C6H11NO2

C10H14N2O3

C4H8O4

C9H12N2O2

C5H11N

CH2O3

C6H13NO2

C6H5NO3

Native elem. comp.

4

1

1

1

4

1

4

3

4

4

4

2

1

1

3

1

3

4

4

4

4

3

2

1

4

alk

–

O-Stearoyl diethylene glycol

N-Palmitoyl piperidine

O-Palmitoyl diethylene glycol

–

4-Methoxyindole-3-acetonitrile

–

Thiohexose

–

–

–

1-Methoxyindole-3-acetonitrile

9-(Methylthio)nonanenitrile

5-Hydroxyhydantoine

Primary amino aldehyde/ketone

Erythronic acid c-lactone

Deoxypentonic acid c-lactone

–l

–

–

–

Primary amine, pentenyl amine

Carbonic acid

Ethyl N–n-propylcarbamate

–

Annotation

l

k

j

i

h

g

f

Piperidine-x-carboxylates (x = 2, 3, 4) do not match EI mass spectral and RI data

Annotation level

Number of methoxime moieties determined by comparison of GC/APCI-QTOFMS data obtained from extracts derivatized with MeOx/BSTFA and MeOx-D3/BSTFA

Corresponding derivative not found after derivatization with MeOx/MSTFA-D9

Number of trimethylsilyl moieties determined by comparison of GC/APCI-QTOFMS data obtained from extracts derivatized with MeOx/BSTFA and MeOx/MSTFA-D9

Difference between predicted and measured FAME-based RI

Measured FAME-based RI used for GC/APCI-QTOFMS

Mapping of the EI and APCI mass spectrum based on matching of M?•/[M-CH3]? and [M?H]? ions (A), [M-CH3]? and [M?H]? ions (B), [M-90]?•/[M-105]? and [M?H]? ions (C), M?• and [M?H]? ions (D), consistent fragment ions (E)

e

d

c

b

a

B

A112003

1134

2

A

–

1062

1

Mape

GMD IDb

RIa ALKANE

No.

Table 1 Elemental composition and annotation of unknown components 1–25 obtained by mutual analysis of EI, APCI(?) and related CID mass spectra

N. Strehmel et al.

Annotating unknown components

OMe

CN O

[%]

M+* 258

a

[%]

O

60

14

O m/z 99

N

40

129 84

170

142

102 115

80

100

120

140

185

202

160

180

200

220

99

260 m/z

240

CO+

87

80

80

60

60

40

40

20

20

239

161 179

129

100

259.1270 [M+H]+

14

145

0 [%]

b

103

20

[M-OCH3]+ 227

0 [%]

Si O+

[M-CH3]+ 243

N

20

O

O+

60 73 159

O+

Si

116

OMe

Si

O

283

d

80

80 73

40

O

14 O

N Si

150

200

250

300

e

350

400 m/z

417.3411 [M+H]+

283.2637

160.0750 219.1075

100

200

250

300

350

400

450

500

m/z

550

OMe

80

0 100 [%]

219.107

c

150

200

250

300

f

350

130.063

0 80

100

120

140

160

180

200

220

240

260 m/z

20 0 100

500

550

m/z

O H+

Si

O+ Si

150

H2O

O

O

Si

CO+ 14

239.237

204.083

+

40

179.110

N H+

O

161.099

160.076

C3H9Si+ 20 73.047

O

60

117.072

40

259.126 [M+H]+

N H+

243.094 [M+H-CH4]+

OMe

60

450

O+ 14

N Si

400

283.264

80

232.116 [M+H-HCN]+

[%]

150

417.341 [M+H]+

0

200

250

300

350

400 m/z

Fig. 4 Mutual interpretation of the EI, APCI(?) and derived CID mass spectrum of unidentified components exemplified for 4-methoxy-indole-3-acetonitrile (1TMS) (20) and O-palmitoyl diethylene glycol (1TMS) (22). a Extracted EI mass spectrum of 20, b extracted APCI(?) mass spectrum of 20, c extracted CID mass spectrum

(collision energy 20 eV) of the protonated molecular ion of 20, d extracted EI mass spectrum of 22, e extracted APCI(?) mass spectrum of 22, f extracted CID mass spectra (collision energy 10 eV) of the protonated molecular ion of 22. Mass spectra were obtained either by GC/EI-QMS or GC/APCI(?)-QTOFMS

upon derivatization with MeOx/MSTFA-D9. On basis of the protonated molecular ion, three putative elemental compositions were calculated (C24H49O3S?, C23H49O4Si?, C24H45N4Si?) of which only one (C23H49O4Si?) was supported by the CID mass spectrum obtained from the protonated molecular ion (Fig. 4f). Hence, 22 carries one trimethylsilyl moiety and its native elemental composition is C20H40O4 (1 RDBE) retrieving 50 hits from the ChemSpider database. The CID mass spectrum showed neutral losses of C5H14O2Si (134.077 Da), C7H18O3Si (178.104 Da) and

C16H30O (238.231 Da) from the protonated molecular ion. For the latter two corresponding fragment ions were detected at m/z 179.110 (C7H19O3Si?) and m/z 239.237 (C16H31O?). Assuming complete trimethylsilylation of 22, the observed fragments point to an ester of a C16 alkanoic acid and trimethylsilylated diethylene glycol. Interestingly, the EI and CID mass spectrum show a number of consistent fragment ions at m/z 283, 239, 179, 161, 117. The occurrence of the dominating fragment at m/z 283 (C18H32O2?) can be rationalized by a cyclic 1,3-dioxolanium ion, which is either

123

N. Strehmel et al.

formed by onium cleavage of the protonated molecular ion or by inductive cleavage of the molecular ion. Finally, 22 could be authenticated as palmitoyl diethylene glycol (1TMS) using a synthetic standard.

4 Concluding remarks GC/APCI(?)-QTOFMS represents a valuable technology for the annotation of unidentified components detected during standardized GC/EI-MS-based metabolite profiling experiments. In general, APCI(?)-QTOFMS enables a highly sensitive detection of methoxime/trimethylsilyl derivatives and predominantly provides easily interpretable mass spectra with a dominating protonated molecular ion. These are often accompanied by in-source fragment ions, and in case of trimethylsilyl esters by [M?C3H9Si]?-type adduct and [M?H3O–CH4]?-type adduct-fragment ions. Using protonated molecular ions, the mass and isotopic pattern accuracy of QTOFMS allows the deduction of a small number of candidate elemental compositions. To obtain a unique native elemental composition, the application of deuterated derivatization reagents (MeOx-D3, MSTFA-D9) and the analysis of high-resolution CID mass spectra of the protonated molecular ion are beneficial. Alternatively, in vivo stable isotope labeling can be applied to reduce the number of putative elemental compositions (Birkemeyer et al. 2005). Before the full potential of GC/APCI(?)-QTOFMS can be exploited, a pairwise assignment of the EI and APCI(?) mass spectrum is required. The currently proposed strategy relies on the presence and recognition of the molecular ion and/or well-defined fragment ions (mostly [M-CH3]? ions) within the EI mass spectrum. For a broader applicability, the mapping strategy needs to be refined in the future. Nevertheless, once a pairwise assignment of the EI and APCI(?) mass spectrum has been succeeded a multitude of analytical information can be obtained for the annotation of unknown components including the elemental composition and often complementary fragmentation pattern within EI and CID mass spectra. In particular metabolic databases providing EI mass spectral data will clearly benefit from cataloguing corresponding high-resolution APCI(?) and related CID mass spectra.

References Amirav, A., Gordin, A., Poliak, M., & Fialkov, A. B. (2008). Gas chromatography–mass spectrometry with supersonic molecular beams. Journal of Mass Spectrometry, 43, 141–163. Bellostas, N., Sorensen, A. D., Sorensen, J. C., & Sorensen, H. (2008). Fe2?-catalyzed formation of nitriles and thionamides from intact glucosinolates. Journal of Natural Products, 71, 76–80.

123

Birkemeyer, C., Luedemann, A., Wagner, C., Erban, A., & Kopka, J. (2005). Metabolome analysis: The potential of in vivo labeling with stable isotopes for metabolite profiling. Trends in Biotechnology, 23, 28–33. Brenner, N., Haapala, M., Vuorensola, K., & Kostiainen, R. (2008). Simple coupling of gas chromatography to electrospray ionization mass spectrometry. Analytical Chemistry, 80, 8334–8339. Bristow, T., Harrison, M., & Sims, M. (2010). The application of gas chromatography/atmospheric pressure chemical ionisation timeof-flight mass spectrometry to impurity identification in Pharmaceutical Development. Rapid Communications in Mass Spectrometry, 24, 1673–1681. Carrasco-Pancorbo, A., Nevedomskaya, E., Arthen-Engeland, T., Zey, T., Zurek, G., Baessmann, C., et al. (2009). Gas chromatography/atmospheric pressure chemical ionization-time of flight mass spectrometry: Analytical validation and applicability to metabolic profiling. Analytical Chemistry, 81, 10071–10079. Dunn, W., Erban, A., Weber, R. M., Creek, D., Brown, M., Breitling, R., et al. (2013). Mass appeal: metabolite identification in mass spectrometry-focused untargeted metabolomics. Metabolomics, 9, S44–S66. Dunn, W. B. (2008). Current trends and future requirements for the mass spectrometric investigation of microbial, mammalian and plant metabolomes. Physical Biology, 5, 011001. Dunn, W. B., Broadhurst, D., Begley, P., Zelena, E., FrancisMcIntyre, S., Anderson, N., et al. (2011). Procedures for largescale metabolic profiling of serum and plasma using gas chromatography and liquid chromatography coupled to mass spectrometry. Nature Protocols, 6, 1060–1083. Fiehn, O., Kopka, J., Dormann, P., Altmann, T., Trethewey, R. N., & Willmitzer, L. (2000a). Metabolite profiling for plant functional genomics. Nature Biotechnology, 18, 1157–1161. Fiehn, O., Kopka, J., Trethewey, R. N., & Willmitzer, L. (2000b). Identification of uncommon plant metabolites based on calculation of elemental compositions using gas chromatography and quadrupole mass spectrometry. Analytical Chemistry, 72, 3573–3580. Garcia-Villalba, R., Pacchiarotta, T., Carrasco-Pancorbo, A., SeguraCarretero, A., Fernandez-Gutierrez, A., Deelder, A. M., et al. (2011). Gas chromatography-atmospheric pressure chemical ionization-time of flight mass spectrometry for profiling of phenolic compounds in extra virgin olive oil. Journal of Chromatography A, 1218, 959–971. Halket, J. M., & Zaikin, V. G. (2003). Derivatization in mass spectrometry-1. Silylation. European Journal of Mass Spectrometry, 9, 1–21. Herebian, D., Hanisch, B., & Marner, F. J. (2005). Strategies for gathering structural information on unknown peaks in the GC/ MS analysis of Corynebacterium glutamicum cell extracts. Metabolomics, 1, 317–324. Horning, E. C., Horning, M. G., Carroll, D. I., Dzidic, I., & Stillwell, R. N. (1973). New picogram detection system based on a mass spectrometer with an external ionization source at atmospheric pressure. Analytical Chemistry, 45, 936–943. Hummel, J., Selbig, J., Walther, D., & Kopka, J. (2007). The Golm Metabolome Database: A database for GC-MS based metabolite profiling. Topics in Current Genetics, 18, 75–95. Hummel, J., Strehmel, N., Selbig, J., Walther, D., & Kopka, J. (2010). Decision tree supported substructure prediction of metabolites from GC-MS profiles. Metabolomics, 6, 322–333. Kind, T., Wohlgemuth, G., Lee, D. Y., Lu, Y., Palazoglu, M., Shahbaz, S., et al. (2009). FiehnLib: Mass spectral and retention index libraries for metabolomics based on quadrupole and timeof-flight gas chromatography/mass spectrometry. Analytical Chemistry, 81, 10038–10048.

Annotating unknown components Konishi, Y., Kiyota, T., Draghici, C., Gao, J. M., Yeboah, F., Acoca, S., et al. (2007). Molecular formula analysis by an MS/MS/MS technique to expedite dereplication of natural products. Analytical Chemistry, 79, 1187–1197. Kopka, J. (2006). Current challenges and developments in GC-MS based metabolite profiling technology. Journal of Biotechnology, 124, 312–322. Kopka, J., Schauer, N., Krueger, S., Birkemeyer, C., Usadel, B., Bergmuller, E., et al. (2005). [email protected]: The Golm Metabolome Database. Bioinformatics, 21, 1635–1638. Kumari, S., Stevens, D., Kind, T., Denkert, C., & Fiehn, O. (2011). Applying in-silico retention index and mass spectra matching for identification of unknown metabolites in accurate mass gc-tof mass spectrometry. Analytical Chemistry, 83, 5895–5902. Lisec, J., Schauer, N., Kopka, J., Willmitzer, L., & Fernie, A. R. (2006). Gas chromatography mass spectrometry-based metabolite profiling in plants. Nature Protocols, 1, 387–396. McEwen, C. N., & McKay, R. G. (2005). A combination atmospheric pressure LC/MS:GC/MS ion source: Advantages of dual AP-LC/ MS:GC/MS instrumentation. Journal of the American Society of Mass Spectrometry, 16, 1730–1738. Pacchiarotta, T., Nevedomskaya, E., Carrasco-Pancorbo, A., Deelder, A. M., & Mayboroda, O. A. (2010). Evaluation of GC-APCI/MS and GC-FID as a complementary platform. Journal of Biomolecular Techniques, 21, 205–213. Pfalz, M., Vogel, H., & Kroymann, J. (2009). The gene controlling the indole glucosinolate modifier1 quantitative trait locus alters indole glucosinolate structures and aphid resistance in Arabidopsis. Plant Cell, 21, 985–999. Portoles, T., Sancho, J. V., Hernandez, F., Newton, A., & Hancock, P. (2010). Potential of atmospheric pressure chemical ionization source in GC-QTOF MS for pesticide residue analysis. Journal of Mass Spectrometry, 45, 926–936. Sanchez, D. H., Szymanski, J., Erban, A., Udvardi, M. K., & Kopka, J. (2010). Mining for robust transcriptional and metabolic responses to long-term salt stress: a case study on the model

legume Lotus japonicus. Plant, Cell and Environment, 33, 468–480. Schauer, N., Steinhauser, D., Strelkov, S., Schomburg, D., Allison, G., Moritz, T., et al. (2005). GC-MS libraries for the rapid identification of metabolites in complex biological samples. FEBS Letters, 579, 1332–1337. Schiewek, R., Lorenz, M., Giese, R., Brockmann, K., Benter, T., Gab, S., et al. (2008). Development of a multipurpose ion source for LC-MS and GC-API MS. Analytical and Bioanalytical Chemistry, 392, 87–96. Steinfath, M., Strehmel, N., Peters, R., Schauer, N., Groth, D., Hummel, J., et al. (2010). Discovering plant metabolic biomarkers for phenotype prediction using an untargeted approach. Plant Biotechnology Journal, 8, 900–911. Strehmel, N., Hummel, J., Erban, A., Strassburg, K., & Kopka, J. (2008). Retention index thresholds for compound matching in GC-MS metabolite profiling. Journal of Chromatography B, 871, 182–190. von Wiren, N., Romheld, V., Shioiri, T., & Marschner, H. (1995). Competition between microorganisms and roots of barley and sorghum for iron accumulated in the root apoplasm. New Phytologist, 130, 511–521. Wachsmuth, C. J., Almstetter, M. F., Waldhier, M. C., Gruber, M. A., Nurnberger, N., Oefner, P. J., et al. (2011). Performance evaluation of gas chromatography-atmospheric pressure chemical ionization-time-of-flight mass spectrometry for metabolic fingerprinting and profiling. Analytical Chemistry, 83, 7514–7522. Warren, C. (2013). Use of chemical ionization for GC–MS metabolite profiling. Metabolomics, 9, S110–S120. Zimmermann, R., Welthagen, W., & Groger, T. (2008). Photoionisation mass spectrometry as detection method for gas chromatography. Optical selectivity and multidimensional comprehensive separations. Journal of Chromatography A, 1184, 296–308.

123