Elucidation of the mass fragmentation pathways of

Research Article Received: 22 April 2010

Accepted: 18 June 2010

Published online in Wiley Online Library: 16 July 2010

(wileyonlinelibrary.com) DOI 10.1002/jms.1785

Elucidation of the mass fragmentation pathways of potato glycoalkaloids and aglycons using Orbitrap mass spectrometry† Michael G. Cahill,a Giovanni Caprioli,b Sauro Vittorib and Kevin J. Jamesa,c∗ The mass fragmentation of potato glycoalkaloids, α-solanine and α-chaconine, and the aglycons, demissidine and solasodine were studied using the Orbitrap Fourier transform (FT) mass spectrometer. Using the linear ion trap (LIT) mass spectrometry, multistage collisional-induced dissociation (CID) experiments (MSn ) on the [M + H]+ precursor ions were performed to aid the elucidation of the mass fragmentation pathways. In addition, higher energy collisional-induced dissociation (HCD) mass spectra were generated for these toxins at a high resolution setting [100 000 FWHM (full width at half maximum)] using the Orbitrap. This hybrid mass spectrometry instrumentation was exploited to produce MS3 spectra by selecting MS2 product ions, generated using LIT MS, and fragmentation using HCD. The accurate mass data in the MS3 spectra aided the confirmation of proposed product ion formulae. The precursor and product ions from glycoalkaloids lost up to four sugars from different regions during MSn experiments. Mass fragmentation of the six-ring aglycons were similar, generating major product ions that resulted from c 2010 John Wiley & Sons, Ltd. cleavages at the B-rings and E-rings. Copyright Keywords: high resolution MS; solanines; hybrid mass analyzers; HCD fragmentation; potato toxins

Introduction

J. Mass. Spectrom. 2010, 45, 1019–1025

∗

Correspondence to: Kevin J. James, PROTEOBIO (Mass Spectrometry Centre), Cork Institute of Technology, Bishopstown, Cork, Ireland. E-mail: [email protected]

† This work was presented at the 1st Mass Spectrometry Food Day 2009, Dec 2nd and 3 rd 2009, Parma, Italy a PROTEOBIO (Mass Spectrometry Centre), Cork Institute of Technology, Bishopstown, Cork, Ireland b Dipartimento di Scienze Chimiche, Facoltà di Farmacia, Università di Camerino, via S.Agostino 1, 62032 Camerino, Italy c Environmental Research Institute, University College Cork, Lee Road, Cork, Ireland

c 2010 John Wiley & Sons, Ltd. Copyright

1019

Potatoes (Solanum tuberosum) are considered as one of the most important crops for human consumption and are regarded as the stable diet of choice in many cultures due to its nutritional characteristics.[1] However, toxic steroidal glycoalkaloids are ubiquitous in potato products because these biotoxins play key roles in natural plant protection from mould and insect infestation.[2] The two major toxins, α-solanine and α-chaconine, usually comprise more than 95% of the total steroidal glycoalkaloid content in potatoes. Although these toxins are present in all parts of the potato plant, the skins and the sprouts contain the highest concentration of glycoalkaloids.[1] These compounds play an important role in natural plant protection,[3] and it has also been established that stresses such as exposure to light, frost damage and tissue wounding during harvesting can lead to an accumulation of these toxic glycoalkaloids in potatoes.[2,4,5] Glycoalkaloid content of ca 15 mg/kg can result in a bitter taste in potatoes while levels at or greater than 200 mg/kg are considered hazardous to human health.[5 – 7] Lachman et al. reported that higher levels of α-solanine can cause a burning sensation in the throat and mouth.[8] Other symptoms of glycoalkaloid poisoning include intestinal pain, gastroenteritis, diarrhea, vomiting, fever, rapid pulse, low blood pressure and neurological disorders.[1] Structurally, α-solanine and α-chaconine are esters of the aglycon base (solanidine) and a trisaccharide, differing only in one of the three attached monosaccharides (Fig. 1A). One of the early studies of the mass fragmentation processes in glycoalkaloids employed sector instruments with both lowenergy and high-energy collision induced dissociation (CID) of the [M + H]+ ions. The product ion data included the carbohydrate sequence and linkage sites together with ions that were characteristic of the aglycon moieties.[9] MS methods that have been developed for the determination of glycoalkaloids

include GC-MS,[10] capillary electrophoresis (CE) MS[11] and LCMS/MS.[6] Previous studies of mass fragmentation pathways using multiple analogs of various natural toxin groups, including azapiracids,[12] fumonisins[13] and pyrrolizidine alkaloids,[14] have employed iontrap MS and/or high resolution MS. The main aims of the studies reported herein were to elucidate the mass fragmentation pathways of potato glycoalkaloids by combining the power of MSn using a linear ion-trap (LIT) MS to determine the sequence of fragmentations, together with the high resolution (100 000 FWHM) and the high mass accuracy capabilities of the Orbitrap MS to confirm the formulae of proposed product ions. The simultaneous generation of data using this hybrid MS instrumentation[15] will be useful in future studies to investigate new analogs and bioconversion products of glycoalkaloids, especially using fullscan MS.

M. G. Cahill et al. interface (HESI) source, operated in positive ionization mode. The following optimized source settings were used: capillary temperature 240 ◦ C, vaporization temperature 50 ◦ C, sheath gas flow 35, aux gas flow 30, source voltage 4 kV, source current 100 µA, capillary voltage 52 V and tube lens 120 V. MSn (n = 2–4) studies were performed using CID in the LIT MS. The product ion accurate mass data were generated using the LTQ Orbitrap XL mass spectrometer, in Fourier transform (FT) mode, at a setting of 100 000 (FWHM) resolution. For these purposes, the fragmentation was carried out on ions, isolated using LIT MS, in a higher energy collisional dissociation (HCD) cell.

Results and Discussion

Figure 1. (A) α-Solanine structure: R1 = D-galactose, R2 = D-glucose, and R3 = L-rhamnose. α-Chaconine structure: R1 = D-glucose; R2 = Lrhamnose; and R3 = L-rhamnose. (B) Demissidine structure. (C) Solasodine structure.

Chemicals and reagents α-Solanine and demissidine were purchased from Sigma Aldrich (Dublin, Ireland); solasodine and α-chaconine were purchased from ABCR GmbH (Karlsruhe, Germany). Stock standard solutions (1000 µg/ml) were prepared using methanol containing formic acid (0.01%). Standard working solutions, at various concentrations, were prepared daily by appropriate dilution with methanol from aliquots of the stock solutions. All solvents, including HPLC grade methanol and HPLC grade water were purchased from Fisher Scientific (Dublin, Ireland). Mass spectrometric conditions The LTQ Orbitrap XL hybrid mass spectrometer (ThermoFisher Scientific, Hemel Hemstead, UK), was calibrated using a solution containing caffeine, methionine-arginine-phenylalanine-alanine acetate (MRFA) and Ultramark 1621, according to the manufacturer’s instructions. Toxin stock solutions were diluted using methanol containing formic acid (0.01%) and each standard (1 µg/ml) was infused for 2 min through a heated electrospray

The instrumentation employed in these studies was a LIT MS coupled with an Orbitrap FT mass analyzer. This hybrid configuration greatly facilitates the identification of analytes and the elucidation of mass fragmentation pathways because the parallel data acquisition on two mass analyzers deliver MSn data with high mass accuracy determinations of precursor and product ions.[16] Historically, these studies were carried out using separate ion- trap and high resolution mass spectrometers.[17] The HCD in the LTQ Orbitrap is an alternative dissociation method produced by an additional collision cell situated at the far end of the C-trap region.[18] A drawback typical of most ion-trap mass analyzers is the lack of retention of ions with m/z values less than one-third of the precursor m/z value but the HCD effectively eliminates this problem.[19] Mass fragmentation pathways of α-solanine and α-chaconine Because α-solanine and α-chaconine differ only by one of the three attached monosaccharides, their mass spectra and mass fragmentation pathways are very similar. The HCD MS/MS spectra of these compounds are shown in Fig. 2 and were generated by fragmentation of their [M + H]+ ions using relative collision energy (RCE) values of 60%. The use of an optimized RCE allowed a portion of precursor ions to be observed in the HCD spectra, which have characteristics similar to the spectra obtained using a quadrupole collision cell. However, since ions generated using the HCD were analyzed in the Orbitrap Fourier transform mass spectrometry (FTMS), high mass accuracy data were acquired for all precursor and product ions. As expected, product ions due to the fragmentation of each of the monosaccharides that are attached to the A-ring of these toxins dominated their MS/MS spectra. In the α-solanine spectrum

1020

Figure 2. Orbitrap (100 000 resolution) MS/MS spectra of (A) α-solanine and (B) α-chaconine, generated by trapping the [M + H]+ ions at m/z 868 and 852, respectively and fragmenting in the HCD at 60% RCE.

wileyonlinelibrary.com/journal/jms


J. Mass. Spectrom. 2010, 45, 1019–1025

Mass fragmentation of potato glycoalkaloids and aglycons using Orbitrap MS spectrum that was obtained for α-chaconine by targeting the ion at m/z 560 and a similar spectrum was obtained with α-solanine using LIT MS. In general, fragmentation of the six-ring steroidal backbone produced low abundant ions apart from the ions at m/z 98 and 126 (Fig. 2), which are due to fragmentation of the E-ring. The mass fragmentation pathways proposed for these glycoalkaloids are shown in Scheme 1, together with the accurate mass values that were obtained using HCD MS/MS in the Orbitrap. The data shown for the precursor and product ion masses of α-solanine are designated with the superscript, A, and those for α-chaconine are designated with a superscript, B. The quality of the accurate mass determinations using the Orbitrap MS is demonstrated by the fact that all of the precursor and product ions greater than m/z 300 were determined with error values less than 1 ppm. Demissidine

Figure 3. MS4 spectrum of α-chaconine generated in the linear ion-trap MS by sequential trapping and fragmentation of the ions, m/z 852, 706 and 560.

(Fig. 2A), the product ion at m/z 722 is attributed to a facile loss of the monosaccharide, L-rhamnose and similarly, the ion at m/z 706 is attributed to a loss of the D-glucose monosaccharide. The MS3 spectra were obtained using CID by trapping and fragmenting of the ions, m/z 722 (α-solanine) and m/z 706 (α-chaconine) in the LIT, and both produced similar spectral data. The product ion, m/z 560, from fragmentation of the ion, m/z 722, is attributed to a loss of the D-glucose moiety and a similar ion was observed in the MS3 spectra of α-chaconine (not shown), produced by targeting the ion, m/z 706. In the latter case, the m/z 560 is produced by a loss of the L-rhamnose moiety from the m/z 706 ion. The next stage in the fragmentation pathways for these toxins is the loss of the last monosacharride from the ion, m/z 560, to produce the aglycon ion, m/z 398, which was the base peak in the MS4 spectra, together with the water-loss ion, m/z 380, and an ion at m/z 366, formed by the loss of methanol from the A-ring. Figure 3 shows the MS4

Demissidine (Fig. 1B) has a more complex fragmentation pathway than the glycoalkaloids, and many product ions were observed using both CID MSn in the LIT MS and HCD in the Orbitrap. Unusually, the CID MS/MS spectrum (not shown) contained a greater range and abundance of product ions across the entire observable m/z range than were observed using the HCD FTMS (Fig. 4A). The [M + H]+ precursor ion (m/z 400) of demissidine is the dihydro equivalent of the ion at m/z 398 which is present in the MS2 spectra of α-solanine (Fig. 2A) and α-chaconine (Fig. 2B). The product ion at m/z 382, due to a water loss, was observed in the demissidine spectrum (Fig. 4A) and the fragmentation of the E-ring produced the most abundant ions. The ions, m/z 273 and 255 are attributed to a loss of C8 H17 N from the precursor ion, m/z 400, and the water-loss ion (m/z 382), respectively. Fragmentation at the E-ring also leads to the lower mass ions, m/z 126 and 98, which both contain the nitrogenous F-ring (Scheme 2). These ions were also observed in the glycoalkaloid spectra presented previously (Figs 2A and B). The most abundant hydrocarbon ion in the CID MS2 spectrum (Fig. 4A) using LIT MS was observed at m/z 161 and, using CID MS3 , this was shown to be derived from fragmentation of the ion, m/z 273. The MS3 spectrum (Fig. 4B) was obtained by trapping the precursor ion, m/z 400, and fragmentation in the LIT MS using CID at 40% RCE and then selecting the abundant product ion, m/z 255, for further fragmentation using HCD FTMS at 60% RCE. By these means, all of the ions in this spectrum were determined with accurate mass measurements and the product ion formulae and mass errors (ppm) are shown (Fig. 4B). A wide range of hydrocarbon ions were observed but on this occasion the major product ion was at m/z 159 rather than m/z 161 (Fig. 4B).

J. Mass. Spectrom. 2010, 45, 1019–1025



1021

Figure 4. Orbitrap (100 000 resolution) spectra of demissidine: (A) MS/MS spectrum generated by trapping the ion at m/z 400 and fragmenting in the HCD at 75% RCE. (B) MS3 spectrum generated by trapping the ion at m/z 400 at 40% RCE in the LIT and selecting the ion at m/z 255 for fragmentation using HCD at 60% RCE.

M. G. Cahill et al.

Scheme 1. Mass fragmentation pathways for (A) α-solanine and (B) α-choconine. The accurate mass values were determined using HCD MS/MS with the LTQ Orbitrap and a resolution setting of 100 000 FWHM.

This observation led us to propose that these ions are as a result of fragmentation of the B-ring which occurs after the E-ring fragmentation although other possible fragmentation processes cannot be excluded. Solasodine

1022

The aglycon, solasodine (Fig. 1C), is structurally different from the other compounds examined in this study by having a


spiro E-/F-ring system rather than fused rings. The HCD MS2 spectrum (Fig. 5A) of solasodine is dominated by the water-loss ion, m/z 396, and ions due to fragmentation of the E-ring that produce the hydrocarbon ions, m/z 271 and 253. Interestingly, this E-ring fragmentation also leads to the same nitrogenous ions, m/z 126 and 98, that were observed in the HCD MS2 spectra of α-solanine (Fig. 2A), α-chaconine (Fig. 2B) and demissidine (Fig. 4A). Also evident in the solasodine spectrum (Fig. 5A) are the hydrocarbon ions, m/z 175 and 157. These ions can be formed


J. Mass. Spectrom. 2010, 45, 1019–1025

Mass fragmentation of potato glycoalkaloids and aglycons using Orbitrap MS

Scheme 2. Mass fragmentation pathway for demissidine. The accurate mass values were determined using HCD MS/MS with the LTQ Orbitrap and a resolution setting of 100 000 FWHM.

Figure 5. Orbitrap (100 000 resolution) spectra of solasodine: (A) MS/MS spectrum generated by trapping the ion at m/z 414 and fragmenting in the HCD at 60% RCE. (B) MS3 spectrum generated by trapping the ion at m/z 414 at 40% RCE in the LIT and selecting the ion at m/z 253 for fragmentation using HCD at 60% RCE.

J. Mass. Spectrom. 2010, 45, 1019–1025

demissidine (Scheme 2) demonstrates that B-ring fragmentation is a common process in these aglycons.

Conclusions MSn (n = 2–4) studies on glycoalkaloids using LIT MS generated data showing the sequence of cleavages of the different sugar moieties. The limitations inherent in the CID spectra using the



1023

by fragmentation at the B-rings of ions, m/z 271 and 253 (Scheme 3). Taking advantage of the hybrid MS instrumentation, the MS3 spectrum (Fig. 5B) was produced by trapping and fragmenting the [M + H]+ ion of solasodine, m/z 414, in the LIT MS and selecting the hydrocarbon ion, m/z 253, to produce the HCD high resolution spectrum. A series of unsaturated hydrocarbon ions were observed but the major product ion was at m/z 157, which confirms the proposed B-ring fragmentation (Scheme 3). Thus, a comparison with the fragmentation processes proposed for

M. G. Cahill et al.

Scheme 3. Mass fragmentation pathway for solasodine. The accurate mass values were determined using HCD MS/MS with the LTQ Orbitrap and a resolution setting of 100 000 FWHM.

LIT MS were overcome by using HCD fragmentation to generate accurate mass product ion data. The HCD spectra obtained using the Orbitrap FTMS, at 100 000 resolution, produced high mass accuracy data across a wide spectral range, including low mass diagnostic ions. All of the precursor and product ions greater than m/z 300 were determined with error values less than 1 ppm. The hybrid instrumentation was also exploited using the HCD cell to fragment selected CID generated ions from the LIT MS and thus produce high resolution MS3 spectra of the aglycons. The main fragmentation processes that led to the most abundant product ions were identified as emanating from cleavages at the B-ring and E-ring regions of these six-ring systems. The high quality of the accurate mass data provided supporting evidence for the structures and formulae in the proposed mass fragmentation schemes. Acknowledgements This research was funded by the Higher Education Authority of Ireland, as part ofIreland’s EU Structural Funds Programmes (20072013) and the European Regional Development Fund; Programme for Research in Third Level Institutions (PRTLI-4), Environment and Climate Change: Impacts and Responses.

References

1024

[1] R. M. D. Machado, M. C. F. Toledo, L. C. Garcia. Effect of light and temperature on the formation of glycoalkaloids in potato tubers. Food Control 2007, 18, 503. [2] Y. Chuda, S. Tsuda, A. Ohara-Takada, A. Kobayashi, K. Suzuki, H. Ono, M. Yoshida, T. Nagata, S. Kobayashi, M. Mori. Quantification of lightinduced glycoalkaloids, α-Solanine and α-Chaconine, in fourpotato cultivars (Solanum tuberosum L.) distributed in Japan by LC/MS. Food Sci. Technol. 2004, 10, 341. [3] M. Friedman. Potato glycoalkaloids and metabolites: Roles in the plant and in the diet. J. Agric. Food Chem. 2006, 54, 8655.


[4] D. W. Griffiths, M. F. B. Dale. Effect of light exposure on the glycoalkaloid content of Solanum phureja tubers. J. Agric. Food Chem. 2001, 49, 5223. ´ [5] A. Tajner-Czopek, M. Jarych-Szyszka, G. Lisinska. Changes in glycoalkaloids content of potatoes destined for consumption. Food Chem. 2008, 106, 706. [6] B. Zywicki, G. Catchpole, J. Draper, O. Fiehn. Comparison of rapid liquid chromatography-electrospray ionization-tandem mass spectrometry methods for determination of glycoalkaloids in transgenic field-grown potatoes. Anal. Biochem. 2005, 36, 178. [7] C. L. Pariera Dinkins, R. K. D. Peterson. A human dietary risk assessment associated with glycoalkaloid responses of potato to Colorado potato beetle defoliation. Food Chem. Toxicol. 2008, 46, 2837. [8] J. Lachman, K. Hamouz, M. Orsák, V. Pivec. Potato glycoalkaloids and their significance in plant protection and human nutrition Review. Plant, Soil and Environment (Rostlinná výroba) 2001, 47, 181. [9] M. Claeys, H. Van den Heuvel, S. Chen, P. J. Derrick, F. A. Mellon, K. R. Price. Comparison of high- and low-energy collision-induced dissociation tandem mass spectrometry in the analysis of glycoalkaloids and their aglycons. J. Am. Soc. Mass Spectrom. 1996, 7, 173. [10] N. Kozukue, K.-S. Yoon, G.-I. Byun, S. Misoo, C. E. Levin, M. Friedman. Distribution of glycoalkaloids in potato tubers of 59 accessions of two wild and five cultivated Solanum species. J. Agric. Food Chem. 2008, 56, 11920. [11] G. Bianco, P. Schmitt-Kopplin, A. Crescenzi, S. Comes, A. Kettrup, T. R. I. Cataldi. Evaluation of glycoalkaloids in tubers of genetically modified virus Y-resistant potato plants (var. Désirée) by nonaqueous capillary electrophoresis coupled with electrospray ionization mass spectrometry (NACE-ESI-MS). Anal. Bioanal. Chem. 2003, 375, 799. [12] M. Diaz-Sierra, A. Furey, B. Hamilton, M. Lehane, K. J. James. Elucidation of the fragmentation pathways of azaspiracids, using electrospray ionisation, hydrogen/deuterium exchange, and multiple-stage mass spectrometry. J. Mass Spectrom. 2003, 38, 1178. ´ Szécsi, A. Szekeres, A. ´ Mesterházy, M. Bartok. ´ ´ [13] T. Bartok, A. Detection of new fumonisin mycotoxins and fumonisinlike compounds by reversed-phase high-performance liquid chromatography/electrospray ionization ion trap mass spectrometry. Rapid Commun. Mass Spectrom. 2006, 20, 2447.


J. Mass. Spectrom. 2010, 45, 1019–1025

Mass fragmentation of potato glycoalkaloids and aglycons using Orbitrap MS [14] A. Xiong, L. Yang, Y. He, F. Zhang, J. Wang, H. Han, C. Wang, S. W. A. Bligh, Z. Wang. Identification of metabolites of adonifoline, a hepatotoxic pyrrolizidine alkaloid, by liquid chromatography/tandem and high-resolution mass spectrometry. Rapid Commun. Mass Spectrom. 2007, 21, 869. [15] M. Kellmann, H. Muenster, P. Zomer, H. Mol. Full scan MS in comprehensive qualitative and quantitative residue analysis in food and feed matrices: How much resolving power is required? J. Am. Soc. Mass Spectrom. 2009, 20, 1464. [16] A. R. Godfrey, C. M. Williams, E. Dudley, R. P. Newton, P. Willshaw, A. Mikhail, L. Bastin, A. G. Brenton. Investigation of uremic analytes in hemodialysate and their structural elucidation from accurate mass maps generated by a multi-dimensional liquid chromatography/mass spectrometry approach. Rapid Commun. Mass Spectrom. 2009, 23, 3194.

[17] B. Hamilton, M. Díaz Sierra, M. Lehane, F. Furey, K. J. James. The fragmentation pathways of azaspiracids elucidated using positive nanospray hybrid quadrupole time-of-flight (QqTOF) mass spectrometry. Spectrosc. Int. J. 2004, 18, 355. [18] J. S. Olsen, B. Macek, O. Lange, A. Makarov, S. Horning, M. Mann. Higher-energy C-trap dissociation for peptide modification analysis. Nat. Methods 2007, 4, 709. [19] E. van der Heeft, Y. J. C. Bolck, B. Beumer, A.W.J.M. Nijrolder, A. A. M. Stolker, M. W. F. Nielena. Full-scan accurate mass selectivity of ultra-performance liquid chromatography combined with timeof-flight and Orbitrap mass spectrometry in hormone and veterinary drug residue analysis. J. Am. Soc. Mass Spectrom. 2009, 20, 451.

1025

J. Mass. Spectrom. 2010, 45, 1019–1025

