Identification of anionic selenium species in Se-rich yeast by electrospray QTOF MS/MS and hybrid linear ion trap/orbitrapMSⁿ

Abstract

An analytical approach allowing the identification of unknown selenium metabolites in selenium-rich yeast was described. Anion-exchange HPLC of the Se-metabolome fraction co-eluting with salts in size-exclusion chromatography allowed the separation of nine selenium species (excluding isomers and selenate) as monitored by inductively coupled plasma mass spectrometry (ICP MS). The individual fractions were analyzed by electrospray QTOF MS/MS and hybrid linear ion trap/OrbitrapMSⁿ after sample introduction by reversed-phase nanoHPLC and by hydrophilic interaction LC (HILIC), respectively. Out of the nine detected species, eight were identified on the basis of accurate mass measurements and collision induced dissociation/fragmentation information. Seven Se-species (selenohomolanthionine, γ-Glu-selenocystathionine, 2,3-DHP-selenocystathionine, N-acetyl-selenocystathionine, 2,3-DHP-selenohomolanthionine, Se-methyl-selenoglutathione, and 2,3-DHP-Se-methylselenocysteine) were reported for the first time in Se-rich yeast, five of them have never been reported in any biological sample before.

---

Introduction

The beneficial role of selenium in the supplementation of the human and animal diet is increasingly recognized.^1,2 One of the most popular selenium supplements is yeast grown in the presence of selenite.³ The scientific and industrial interest in this product has already matured into numerous analytical and physiological studies.^4–7 Advances in analytical chemistry^8–12 have allowed the preparation of a Se-rich yeast certified reference material¹³ and a proficiency testing campaign regarding selenomethionine determination in Se-rich yeast tablets.¹⁴

Selenium is principally incorporated into yeast proteins in the form of selenomethionine (SeMet) in a rather non-specific way.^15,16 This incorporation leaves behind a characteristic selenium metabolomic blueprint accounting for 10–30% of selenium distributed amongst a variety of low-molecular weight species.¹⁷ The identification of these low molecular weight (and mostly water-soluble) Se-species in Se-rich yeast (selenometabolome) is becoming of paramount importance. The main reason for this is the recently reported failure of the SELECT clinical trial¹⁸ that attempted to attribute to selenomethionine (supplemented as 200 μg L-SeMet day⁻¹) the decrease in the cancer incidence observed by Clark et al.⁶ in a clinical trial using Se-rich yeast. One of the official reasons disseminated for this unsuccessful trial was that the formulation of selenium, high-selenium yeast, used in the NPC trial of Clark et al. may have been more active than the L-SeMet used in SELECT.¹⁸ Of note are some of the reasons cited by the designers of the SELECT trial on why SeMet had been chosen instead of high-selenium yeast. For example, evidence indicated substantial batch-to-batch variations in specific organoselenium compounds in samples of NPC yeast, making it unlikely to duplicate for the goals of the SELECT study, and practical and safety concerns over newer selenium compounds, such as monomethylated forms (i.e., lacking availability, investigational new drug certification, and clinical data).¹⁹ These reasons clearly point to the need for the characterization of Se-rich yeast in terms of the quality, safety and origin of marketed preparations, which can be achieved via the description of the Se-metabolome.

Electrospray mass spectrometry, first proposed for the standardless identification of Se-metabolites by Casiot et al.,²⁰ has been the principal technique for the characterization of Se-containing metabolites in Se-rich yeast,^9,10,21–23 especially when assisted by multidimensional chromatographic purification.^24–26 Numerous reports of “shoot-and-see” approaches in which a or some Se-species were, sometimes accidentally, identified in extracts of different purity by HPLC–ESI MS/MS give a false impression of the ease of use of these hyphenated techniques for Se-metabolomics studies. Indeed, systematic approaches in which the mass balance of selenium was carefully monitored and all the existing species above a certain threshold concentration were identified have been rare. Recently, we proposed the anion-exchange HPLC separation of Se-species in the most abundant size-exclusion metabolite fraction under strict Se mass balance control by ICP MS, which allowed the identification of all seven species present in this fraction by electrospray QTOF MS.²⁷ The identity of all the detected species could be confirmed by HILIC–hybrid linear ion trap/Orbitrap MSⁿ.²⁸ The objective of this study was to investigate the feasibility of these protocols for the characterization of another, less abundant and richer in co-eluting salts, lower molecular weight selenium fraction separated by size-exclusion HPLC in order to advance the characterization of the Se-metabolome in Se-rich yeast.

Experimental

Reagents and standards

All the chromatographic eluents and reagents were purchased from the Sigma group (Sigma-Aldrich-Fluka-Riedel-de Haën; St. Quentin Fallavier, France). De-ionised water (18.2 MΩ cm; MilliQ Millipore; Guyancourt, France) was used throughout.

Sample

A selenised yeast sample, Sel-Plex (Alltech, Nicholasville, KY, USA) containing 2.0 ± 0.1 g Se kg⁻¹ corresponding to a yeast strain Saccharomyces cerevisiae CNCM I-3060, batch ES-453, was used.

Instrumentation

The normal bore anion exchange HPLC–ICP MS coupling was achieved by using an Agilent 1100 HPLC system (Agilent Technologies, Palo Alto, CA) connected to an Elan 6000 ICP MS (PE-SCIEX, Ontario, Canada) for element-specific detection of ³⁴S, ⁷⁷Se, ⁸²Se, and ¹⁰³Rh.

For the HILIC–ESI MS experiments, an Accela High Speed liquid chromatograph (Thermo Fisher Scientific, Waltham, MA) system was connected to a hybrid linear ion trap/Orbitrap mass analyzer (Thermo Fisher Scientific) used in either full-scan mode or in product ion (MSⁿ) mode. The ion source was operated in the positive ion mode at 4 kV. Capillary temperature was set to 300 °C. The resolving power of the Orbitrap (full width half-height, FWHM) was set to nominal 60 000 (at m/z = 400; 1 s scan cycle time) in full scan mode. For the MSⁿ experiments, the product ions from the [M + H]⁺ charged target ions were generated in the LTQ trap at a collision energy setting of 35% and using an isolation width of 10 Da (i.e., the target monoisotopic mass ± 5 Da), in order to include and recognize the selenium isotopes in the fragments. The other operating parameters of the instrument have been presented in detail elsewhere.²⁸

For nanoflow HPLC an Ultimate nanoHPLC system (LC Packings, Amsterdam, The Netherlands) consisting of a binary nanoflow pump and a He-degasser was used. The sample was introduced by means of a 0.3 μl loop of a model CN2 injection valve (Valco Instruments, Houston, TX, USA) serving as a manual injection port. All connections were made of fused silica capillaries (i.d. 20 μm, Polymicro Technologies, Phoenix, AZ).

For the nanoHPLC–ESI MS experiments, the nanoHPLC system was connected to a hybrid quadrupole/time-of-flight mass spectrometer Applied Biosystems QSTAR XL (Applied Biosystems, Foster City, CA, USA) used in either full-scan TOF mode with a 1 s integration time or in product ion (MS/MS) mode. The nanoHPLC–ESI MS coupling was carried out via a nanospray interface (Applied Biosystems), operating in the positive ion mode. TOF MSmass spectra were recorded across the range m/z 100–2000. QTOF MS/MS experiments were carried out in the range m/z 50–1000. MS/MS experiments were accomplished with collision energies automatically set by the software. The other operating parameters of the instrument have been presented in detail elsewhere.²⁶

For total digestion of samples a Multiwave 3000 microwave digestion system (Anton Paar; Courtaboeuf, France) was used.

Procedures

Determination of total selenium and column recovery. A sample (or aliquots to analyse) was mixed with 4.0 ml of conc. HNO₃ and 4.0 ml of H₂O₂ in digestion tubes. The temperature was raised to 210 °C within 15 min and held for 20 min. The total Se concentration was determined (Elan 6000) using the ⁷⁷Se and ⁸²Se isotopes by the method of standard additions using Rh as an internal standard.

Column recovery data were determined according to the procedure described by Polatajko et al.²⁹ Briefly, separate injections were done without coupling the HPLC system to the ICP MS to collect the column effluent during the gradient elution program. After collection, the effluent was frozen, lyophilised and digested as presented above. Column recovery was calculated as the mass ratio of Se in the collected effluent/injected sample.

Water extraction and size exclusion chromatography (SEC). The protocols of waterextraction of the selenised yeast sample and the preparative scale SEC clean-up procedure, based on the use of a HiLoad 26/60 Superdex 30 pg column (GE Healthcare-Amersham Biosciences, Uppsala, Sweden) have been described elsewhere.^27,30 The SEC fraction selected for further analyses was frozen and lyophilised.

Anion exchange chromatography (SAX). The procedure of SAX separation has been published by Dernovics and Lobinski,²⁷ and only major modifications are listed here. A PRPX-100 SAX column (250 mm × 4.1 mm × 10 μm; Hamilton, Reno, NV) was used. Gradient elution was made with ammonium acetate (buffer A: 10 mM, buffer B: 100 mM; pH 5.5) delivered at 1.5 ml min⁻¹. The program was: 0–4 min 100% A, 4–40 min up to 100% B, 40–60 min 100% B. The sample was dissolved in 0.5 ml of buffer A. For fraction collection, the injection volume was 200 μl.

Hydrophilic interaction liquid chromatography (HILIC) and nanoRPLC sample introduction systems

A TSK-Gel Amide-80 (250 mm × 4.6 mm × 5 μm; Tosoh Biosciences, Stuttgart, Germany) HILIC column equipped with a matching guard column was used. Gradient elution (0.8 ml min⁻¹) was carried out using eluent A: 0.1 v/v% trifluoroacetic acid (TFA) in acetonitrile and eluent B: 0.1 v/v% TFA in water. The program was: 0–4 min 97% A, 4–35 min up to 100% B, 35–45 min 100% B. Before injection, the sample was dissolved in a solution containing 97% of eluent A and 3% of eluent B.

Nano-scale RPLC–ESI MS analysis of the purified Se-containing fractions was carried out according to the procedure presented elsewhere.²⁷

Accurate mass analysis and the use of elementary calculator tool (ECT)

In order to assign structures to complete the HILIC–ESI MSⁿ (Orbitrap) analyses, a bottom-up approach was used. It started from the analysis of the MS⁴ data and continued with MS³, MS², and MS. For all the calculations, the elementary calculator tool (ECT) was set as follows: mass error [<5 ppm], double bond and ring equivalent (DBE) [−15–+15], electron state [even], charge [+1], C [1–50], H [1–80], N [0–10], O [0–20], P [0–2], S [0–2], and Se [0–1]. The last value was set individually according to the observed pattern of the given fragment. This was possible due to the fragmentation isolation window of 10 set: the fragments containing either 0 or 1 Se atom could be routinely spotted owing to the isotopic pattern, which made the elementary calculation process produce less possible compositions. In the descriptions provided, the data in square brackets refer to the theoretical mass and to its difference from the measured value (in ppm).

Results and discussion

Fractionation of the Se-yeast water extract by SEC and SAX

The water extract was found to contain 13 ± 1% of selenium originally present in the sample. The extract was first submitted to preparative scale SEC. As shown in Fig. 1a, the elution resulted in several Se peaks of which those at 100, 130, 200, and 310 ml have been extensively characterized elsewhere.^27,31–33 The object of this study was the third most abundant fraction eluting from 220 to 240 ml. It was found to contain 16 ± 4% of the total selenium injected on the column and was collected for further analyses. The S elution profile shows clearly that the targeted fraction contains a significant amount of sulfur as well.

Fig. 1 (a) Preparative scale SEC–ICP MS chromatogram of the Se-rich yeast water extract. ⁸²Se and ³⁴S were monitored. Vertical lines indicate the fraction collected for further analyses. The ³⁴S signal is set off for clarity of presentation. (b) SAX–ICP MS chromatogram of the collected SEC fraction. The ³⁴S signal is set off for clarity of presentation. The arrows indicate the retention time of three Se-standards: I-selenomethionine, II-Se(IV), III-Se(VI). Fractions numbered from 1 to 9 were collected separately for further analyses.

The collected peak was further characterized by analytical scale anion exchange chromatography that provides high recovery for Se-species²⁹ and is robust enough for successive fraction collection even with an increased sample load. As shown in Fig. 1b, 10 selenium-containing peaks were detected, along with a sulfur-containing peak. Note that the sulfur chromatogram is noisier because of the ca. 100-fold higher detection limit for ³⁴S than for ⁸²Se.

Retention time matching with the available Se standards allowed the identification of only one peak (No. 7). It corresponded to selenate that is further corroborated by its relative elution order with regard to sulfate, the only peak detected in the ³⁴S chromatogram. Therefore, fraction collection of each individual peak and subsequent ESI MS analyses were carried out to identify the unknown Se-species. The column recovery was 94%. It could be increased up to 100% (at the expense of resolution, however) offering the guarantee that none of the Se-species more abundant than any of the species seen in the chromatogram escaped the analysis.

Analyses of the SEC–SAX fractions by nanoHPLC–ESI TOF MS, QTOF MS/MS and HILIC–OrbitrapMSⁿ instrumentation

In order to increase the purity of the selenium compounds at the moment of their arrival at the ionization source, nanoRP HPLC was chosen for sample introduction into the ESI MS. It was reported in our previous works to improve the signal-to-noise (S/N) ratio of Se-containing analytes from complex mixtures due to its high sensitivity, the feasibility of an efficient background correction and the elimination of a number of matrix interferences.²⁷ NanoRP HPLC sample introduction into ESI QTOF MS allowed the detection of minor Se-species at a concentration of 1 μg ml⁻¹ (0.3 ng per injection) recently.²⁶ In order not to miss any Se-species in this work the injections were carried out at a concentration of at least 10 μg ml⁻¹ (ca. 3 ng per injection) of Se in each individual fraction.

Each of the fractions was analyzed individually with a preliminary off-line full scan ESI QTOF MS acquisition. The full scan spectra were then checked for the presence of Se-containing compounds through the search for the isotopic pattern and mass defect of selenium. All the observed Se-species were then subjected to collision induced dissociation (CID) experiments. In the case of fractions 3, 4, 4′ and 8, for which QTOF MS data were judged inconclusive, Orbitrap-MSⁿ analyses were carried out. The latter proved to be of vital importance in the case of unknown Se-species when low and sub-ppm mass accuracy was required for the unambiguous identification of the species found.²⁸

No selenium-containing peak was detected in fraction No. 7; however, this fraction contained selenate, an inorganic Se-species that presents no intensity when analyzed in the positive ion mode.

Identification of Se-species in SAX fraction No. 1. The full scan spectrum of this fraction eluting in the void volume of the column (see ESI† -Fig. 1a) reveals the presence of a low intensity monoselenised compound with the characteristic isotopic pattern centered at m/z = 285.0472. Although the compound suffered from isobaric overlaps its CID fragmentation on the ⁸⁰Se isotope could be carried out. The analysis of the resulting fragments allowed the identification of selenohomolanthionine, a Se-species recently reported in Japanese pungent radish by Ogra et al.³⁴ ESI† -Fig. 1b presents the proposed fragmentation pathway, along with the accurate mass information in Table 1.

Table 1 Detected and identified ions and fragments using ESI QTOF MS/MS and LTQ OrbitrapMS (in bold) originated from the Se-species presented in ESI¹ -Fig. 1,2 and 4 and Fig. 2, 3 and 5–7

Se-species description	No. of ions/fragments	Level of MS^n	Notes on detected ions and fragments	Elemental composition	Theoretical mass, m/z	Measured mass, m/z	Difference, ppm
Selenohomolanthionine (see ESI^1 -Fig. 1)	1	1	Entire molecule	C8H17N2O4Se^+	285.03481	285.0472	43
	2	2	Selenohomocysteine residue	C4H10NO2Se^+	183.98713	183.9965	51
	3	2	Loss of formic acid from the selenohomocysteine residue	C3H8NSe^+	137.98165	137.9828	8
	4	2	Heteroelement-free homocysteine residue	C4H8NO2^+	102.05495	102.0548	−1
Isomers of γ-Glu-selenocystathionine (see Fig. 2 and ESI^1 -Fig. 2)	1	1	Entire molecule	C12H22N3O7Se^+	400.06175	400.0409	−52
		1				400.06174	−0.02
	2	2	Losses of NH3 and formic acid	C11H17N2O5Se^+	337.02972	337.0279	−5
	3	2	Loss of γ-Glu	C7H15N2O4Se^+	271.01916	271.0300	40
		2				271.01865	−1.88
	4	2	Selenohomocysteine residue	C4H8NO2Se^+	181.97148	181.9660	−30
		2				181.97113	−1.92
	5	2	Loss of formic acid from the selenohomocysteine residue	C3H6NSe^+	135.96600	135.9557	−76
		4				135.96568	−2.35
	6	2	γ-Glu residue	C5H8NO3^+	130.04987	130.0407	−71
	7	2	Heteroelement-free homocysteine residue	C4H8NO2^+	102.05495	102.0478	−70
	8	2	Loss of formic acid from the γ-Glu residue	C4H6NO^+	84.04439	84.0400	−52
	9	1	Entire molecule	C12H22N3O7Se^+	400.06175	400.0418	−50
	10	2	Loss of heteroelement-free homocysteine residue	C8H15N2O5Se^+	299.01407	299.0201	20
	11	2	Selenohomocysteine residue	C4H8NO2Se^+	181.97148	181.9630	−47
	12	2	Loss of formic acid from the selenohomocysteine residue	C3H6NSe^+	135.96600	135.9548	−82
	13	2	γ-Glu residue	C5H8NO3^+	130.04987	130.0403	−74
	14	2	Heteroelement-free homocysteine residue	C4H8NO2^+	102.05495	102.0449	−98
	15	2	Loss of formic acid from the γ-Glu residue	C4H6NO^+	84.04439	84.0364	−95
Isomers of 2,3-DHP-selenocystathionine (see Fig. 3)	1	1	Entire molecule	C10H19N2O7Se^+	359.03520	359.0161	−53
		1				359.03519	−0.03
	2	2	Loss of H2O	C10H17N2O6Se^+	341.02463	341.02454	−0.26
	3	2	Loss of formic acid	C9H17N2O5Se^+	313.02972	313.0215	−26
		2				313.02927	−1.44
	4	2	Selenocystathionine residue (by the loss of 2,3-DHP)	C7H15N2O4Se^+	271.01916	271.0127	−24
		2				271.01883	−1.22
	5	2	Loss of heteroelement-free cysteine residue	C7H12NO5Se^+	269.98752	269.9839	−13
		2				269.98703	−1.81
	6	2	Loss of heteroelement-free homocysteine residue	C6H12NO5Se^+	257.98752	257.9825	−19
		2				257.98737	−0.58
	7	2	Losses of heteroelement-free homocysteine and formic acid	C6H10NO2Se^+	207.98713	207.98691	−1.06
	8	2	Selenohomocysteine residue	C4H10NO2Se^+	183.98713	183.9772	−54
		2				183.98701	−0.65
	9	2	Selenohomocysteine residue	C4H8NO2Se^+	181.97148	181.9619	−53
		3				181.97159	0.60
	10	2	Loss of selenohomocysteine residue	C6H10NO5^+	176.05535	176.0471	−47
		2				176.05516	−1.08
	11	2	Losses of selenohomocysteine residue and H2O	C6H8NO4^+	158.04478	158.0382	−42
		3				158.04456	−1.39
	12	2	Loss of formic acid from the selenohomocysteine residue	C3H8NSe^+	137.98165	137.9733	−61
		2				137.98155	−0.72
	13	2	Loss of formic acid from the selenohomocysteine residue	C3H6NSe^+	135.96600	135.9572	−65
		4				135.96590	−0.74
	14	3	Loss of formic acid from fragment No. 10	C5H8NO3^+	130.04987	130.04958	−2.23
	15	2	Losses of 2,3-DHP, formic acid, NH3 and heteroelement-free Cys	C3H5Se^+	120.95509	120.95502	−0.58
	16	2	Loss of selenohomocysteine residue from selenocystathionine	C3H6NO2^+	88.03930	88.0329	−73
		3				88.03914	−1.82
	17	2	Losses of formic acid and selenium from selenohomocysteine residue	C3H6N^+	56.04948	56.0451	−78
Isomers of N-acetyl-selenocystathionine (see Fig. 5)	1	1	Entire molecule	C9H17N2O5Se^+	313.02972	313.0169	−41
	2	2	Loss of heteroelement-free cysteine residue	C6H10NO3Se^+	223.98204	223.9761	−27
	3	2	Selenohomocysteine residue	C4H8NO2Se^+	181.97148	181.9619	−53
	4	2	Loss of ammonia from the selenohomocysteine residue	C4H7O2Se^+	166.96058	166.9532	−44
	5	2	Loss of formic acid from the selenohomocysteine residue	C3H6NSe^+	135.96600	135.9564	−71
	6	2	Loss of selenohomocysteine residue	C5H8NO3^+	130.04987	130.0390	−84
2,3-DHP-selenohomolanthionine (see Fig. 5)	1	1	Entire molecule	C11H21N2O7Se^+	373.05085	373.0305	−55
	2	2	Losses of 2,3-DHP, NH3 and formic acid	C7H12NO2Se^+	222.00278	221.9912	−52
	3	2	Loss of selenohomocysteine residue	C7H12NO5^+	190.07100	190.0587	−65
	4	2	Selenohomocysteine residue	C4H10NO2Se^+	183.98713	183.9726	−79
	5	2	Losses of formic acid and the selenohomocysteine residue	C6H10NO3^+	144.06552	144.0592	−44
	6	2	Losses of formic acid, a CH2 group and the selenohomocysteine residue	C5H8NO3^+	130.04987	130.0389	−84
γ-Glu-Se-methylselenocysteine (see ESI^1 -Fig. 4)	1	1	Entire molecule	C9H17N2O5Se^+	313.02972	313.0147	−48
	2	2	Losses of NH3 and formic acid	C8H12NO3Se^+	249.99769	249.9863	−46
	3	2	Losses of C2H2, NH3 and formic acid	C6H10NO3Se^+	223.98204	223.9664	−70
	4	2	Loss of γ-Glu residue	C4H10NO2Se^+	183.98713	183.9808	−34
	5	2	Losses of C2H2, NH3 and two formic acid groups	C5H8NOSe^+	177.97656	177.9694	−40
	6	2	losses of γ-Glu residue and NH3	C4H7O2Se^+	166.96058	166.9535	−42
	7	2	Losses of γ-Glu residue, H2O and NH3	C4H5OSe^+	148.95001	148.9423	−52
	8	2	Losses of γ-Glu residue and formic acid	C3H8NSe^+	137.98165	137.9718	−71
	9	2	γ-Glu residue	C5H8NO3^+	130.04987	130.0395	−80
	10	2	Loss of formic acid from γ-Glu residue	C4H6NO^+	84.04439	84.0388	−67
Se-methylselenoglutathione (see Fig. 6)	1	1	Entire molecule	C11H20N3O6Se^+	370.05118	370.0406	−29
		1				370.05124	0.16
	2	2	Losses of NH3 and formic acid	C9H15N2O4Se^+	295.01916	295.01935	0.64
	3	2	Loss of γ-Glu	C6H13N2O3Se^+	241.00859	241.00856	−0.12
	4	2	Losses of γ-Glu residue and NH3	C6H10NO3Se^+	223.98204	223.9743	−35
		2				223.98195	−0.40
	5	2	Losses of Gly, γ-Glu and NH3	C4H5OSe^+	148.95001	148.9438	−42
	6	2	Losses of Gly, γ-Glu and CO	C3H8NSe^+	137.98165	137.9732	−61
	7	2	γ-Glu residue	C5H8NO3^+	130.04987	130.0402	−74
		2				130.04987	0.00
	8	3	Loss of Se-methyl-group from fragment No. 4	C5H8NO3^+	130.04987	130.04976	−0.85
	9	2	Losses of Gly, γ-Glu, CO and NH3	C3H7Se^+	122.97075	122.9598	−89
	10	2	Loss of H2O from the γ-Glu residue	C5H6NO2^+	112.03930	112.0289	−93
	11	3, 4	Loss of H2O from the fragment No. 8			112.03922	−0.71
	12	2	Loss of formic acid from the γ-Glu residue	C4H6NO^+	84.04439	84.0403	−49
	13	3, 4	Loss of formic acid from the fragment No. 8			84.04430	−1.07
2,3-DHP-Se-methylselenocysteine (see Fig. 7)	1	1	Entire molecule	C7H14NO5Se^+	272.00317	271.9963	25
	2	1 (in source)	Losses of 2,3-DHP and NH3	C4H7O2Se^+	166.96058	166.9557	29
	3	2	Se-Methyl residue	CH3Se^+	94.93945	94.9344	53

---

Identification of Se-species in SAX fractions No. 2 and 3. The full scan spectra of these fractions contained a low-intensity monoselenised compound with the characteristic isotopic pattern centered at m/z = 400.04 in both fractions (see ESI† -Fig. 2a and b). Although the relevant CID spectra contained only a few fragments, they were all compatible with the isomers of γ-Glu-selenocystathionine. The two isomers of this species were reported in nut and seed samples^26,35 elsewhere but this is, to our knowledge, the first time they have been detected in Se-rich yeast.

For unambiguous identification, fraction No. 3 was also analysed with the HILIC–Orbitrap hyphenated system, by a targeted search for the accurate mass of γ-Glu-selenocystathionine (C₁₂H₂₂N₃O₇Se⁺), m/z = 400.06175 ± 0.0025 Da. This process resulted in the detection of a monoselenised Se-compound with the close accurate mass (m/z = 400.06174; −0.02 ppm]) of γ-Glu-selenocystathionine. The cascade of MSⁿ acquisitions presented in Fig. 2 and the information published on the fragmentation characteristics²⁶ proved the compound to be one of the isomers of γ-Glu-selenocystathionine. Briefly, the ECT gave only one possible elemental composition for the single Se-atom containing fragment detected at MS⁴ within the 5 ppm threshold, C₃H₆NSe⁺ [m/z = 135.96600; −2.35 ppm]. This information is equivalent to only one possible elemental composition of the relevant parent ion observed in the MS³spectrum (m/z = 181.97205, Fig. 2); this is C₄H₈NO₂Se⁺ [m/z = 181.97148; 3.13 ppm], assuming the loss of formic acid at the trapping step of MS³ → MS⁴. This composition helps to limit the number of possible hits on the MS² level to a unique possibility, C₇H₁₅N₂O₄Se⁺ [m/z = 271.01916; −1.88 ppm], the composition of selenocystathionine. Finally, the parent ion on the MS¹ (full scan) level could be identified as C₁₂H₂₂N₃O₇Se⁺, which refers to a characteristic loss of a γ-Glu group at the first fragmentation process of MS¹ → MS².

Fig. 2 LTQ OrbitrapMS analysis of SAX fraction No. 3 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic pattern of selenium at m/z = 400.06174 ([M + H]⁺), and the spectra of the MSⁿ (where n = 2–4) trapping and fragmentation events. For accurate mass information see Table 1.

ESI† -Fig. 2c and d present the proposed fragmentation pathways of the isomers, along with the accurate mass information in Table 1.

Identification of Se-species in SAX fraction No. 4

The full scan spectrum of the SAX fraction with the highest selenium content contained an abundant monoselenised compound at m/z = 359.0161 (Fig. 3a). Although this compound has never been identified before, its MS/MSspectrum showed a series of characteristic and known fragments that could be referred to previously identified species and residues, such as m/z = 181.96 and 135.95 (selenohomocysteine) and m/z = 271.01 (selenocystathionine). For the assignment of a possible structure, the fragment detected at m/z = 257.98 was of crucial importance: this fragment was observed during the analysis of an unusual selenoglutathione conjugate,^27,28 referring to a 2,3-dihydroxypropionyl (DHP) residue. Taking into account the asymmetry of selenocystathionine, two isomers of 2,3-dihydroxy-propionyl-selenocystathionine could be proposed, similarly to the isomers of γ-Glu-selenocystathionine.

Fig. 3 (a) ESI QTOF MS/MS analysis of the SAX fraction No. 4 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic pattern of selenium at m/z = 359.01 ([M + H]⁺) and the CID spectrum of the pseudo-molecular ion. (b) LTQ OrbitrapMS analysis of SAX fraction No. 4 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The inset shows the isotopic pattern of selenium at m/z = 359.03519 ([M + H]⁺). (c) HILIC–LTQ Orbitrap TIC (Total Ion Chromatogram) of the Se-containing SAX fraction No. 4 (cf. Fig. 1b). The inset shows the XIC (Extracted Ion Chromatogram) of the accurate mass m/z = 359.03495–359.03545. (d) Spectra of the HILIC–LTQ Orbitrap MSⁿ (where n = 2–3) trapping & fragmentation events of the pseudomolecular ion m/z = 359.03519 eluting in the first part of the XIC peak, t_R = 16.23 min. (e) Spectra of the HILIC–LTQ Orbitrap MSⁿ (where n = 2–4) trapping & fragmentation events of the pseudomolecular ion m/z = 359.03519 eluting in the second part of the XIC peak, t_R = 16.55 min. (f) Proposed fragmentation pathways of the detected Se-compounds of SAX fraction No. 4, the isomers of 2,3-DHP-selenocystathionine. For accurate mass information see Table 1.

Although this approach could explain all fragments detected, the mass accuracies of the pseudomolecular ion and the daughter ions were often low (13–78 ppm, see Table 1). Therefore, a confirmation by OrbitrapMS was sought. The mass filtering process for the accurate mass of the proposed compound, 2,3-dihydroxy-propionyl-selenocystathionine ((C₁₀H₁₉N₂O₇Se⁺), m/z = 359.03520 ± 0.0025 Da), resulted in the detection of a monoselenised compound (Fig. 3b) eluting in a double peak in the extracted ion chromatogram (XIC) at t_R = 16.23 and 16.55 min (see Fig. 3c). In the whole peak, the detected Se-species possessed the same and close accurate mass to the proposed compound (m/z = 359.03519; −0.03 ppm; Fig. 3b). In order to explore the phenomenon of the double peak, two cascades of MSⁿ were recorded covering each of the peaks. The first set (Fig. 3d) showed that the two isomers of the target compound partly co-eluted, while the second set of data corresponding to the later eluting less abundant peak showed the presence of only one isomer (see Fig. 3e).

The detection of the two isomers of 2,3-dihydroxy-propionyl-selenocystathionine was supported by the ECT information as well. Regarding the first set (Fig. 3d), the MS³ level originating from the parent ion of m/z = 176.05516 ± 5 Da contained four fragments (the MS⁴ data were of too low abundance for any assignments). The non-selenised fragments at m/z = 88.03914, m/z = 130.04958 and m/z = 158.04456 could each be determined as C₃H₆NO₂⁺ [m/z = 88.03930; −1.82 ppm], C₅H₈NO₃⁺ [m/z = 130.04987; −2.23 ppm], and C₆H₈NO₄⁺ [m/z = 158.04478; −1.39 ppm], respectively. Due to the wide fragmentation window, a ⁷⁸Se isotope containing fragment could also be assigned as C₃H₆N⁷⁸Se⁺ [m/z = 133.96680; −2.39 ppm]. The non-selenium containing fragments provide a unique hit by the ECT for their parent ion, C₆H₁₀NO₅⁺ [m/z = 176.05535; −1.08 ppm] on the MS² level. At the same level, eight selenised fragments can be detected from [m/z = 120.95502 up to m/z = 341.02454]. In the increasing order of mass from these, the elemental composition of seven (up to m/z = 313.02927) fragments could be assigned by the ECT with unique hits (see Table 1). In the case of the m/z = 341.02454 fragment, several possible hits were calculated and its composition could not be determined. However, the elemental composition of the successfully assigned fragments of the MS² level unambiguously determined the composition of the unknown Se-species as C₁₀H₁₉N₂O₇Se⁺ [m/z = 359.03520; −0.03 ppm].

In the second set of MSⁿ data (Fig. 3e), the most abundant fragments of the MS³ and MS⁴ levels can be described with only one couple of elemental composition, that of selenohomocysteine and its −46 Da (formic acid) loss residue: C₄H₈NO₂Se⁺ [m/z = 181.97148; 0.60 ppm], and C₃H₆NSe⁺ [m/z = 135.96600; −0.74 ppm], respectively. The selenohomocysteine residue on the MS³ level limits the possible elemental compositions of its parent ion on the MS² level for C₇H₁₂NO₅Se⁺ [m/z = 269.98752; −1.81 ppm], that unambiguously determines the composition of the main pseudomolecular ion on the MS¹ level as C₁₀H₁₉N₂O₇Se⁺ [m/z = 359.03520; −0.03 ppm]. The proposed fragmentation pathways of both isomers are presented in Fig. 3f, while the related accurate mass information is listed in Table 1.

It is to be emphasized that the high resolution and the high mass accuracy of the Orbitrap instrument even at low intensity fragments provided important information for the successful assignment of the co-eluting isomers. As presented in ESI† -Fig. 3, fragments differing by only 17–34 mDa (requiring mass resolution (R) over 10500 and 7600, respectively) could be resolved and assigned with high (<3 ppm) mass accuracy.

Identification of Se-species in SAX fraction No. 4′

A less abundant selenised compound eluting in peak No. 4′ in the SAX run was co-fractionated with peak No. 4. However, in this peak no Se-species other than the m/z = 359.03519 compound were found with QTOF MS detection. Therefore, the search for this other Se-species was conducted further in the HILIC–Orbitrap set-up. Because this species was less retained on the SAX column, its possible elution was estimated after the previously detected m/z = 359.03519 species. Indeed, a monoselenised compound was spotted at t_R = 17.11 min with the accurate mass of m/z = 345.01965 (see Fig. 4). However its fragmentation was not successful because of its low abundance, its possible elemental composition can be set to C₉H₁₇N₂O₇Se⁺ [m/z = 345.019548; −0.30 ppm], supported by the ECT as the most probable combination. Interestingly, this proposed composition differs from the 2,3-dihydroxy-propionyl-selenocystathionine species by a –CH₂– group, which may refer to a hypothetical structure presented in the inset of Fig. 4. Indeed, this Se-species may be formed through the oxidative cleavage of a sulfur or a selenium atom from a sulfur-selenium or a diselenide bridge, followed by a structural rearrangement. This reaction may take place during the down-stream fermentation processes.

Fig. 4 Full-scan MS spectrum taken at the retention time of the HILIC–LTQ OrbitrapMS chromatogram corresponding to the SAX fraction No. 4′ (collected together with No. 4; cf. Fig. 1b). The insets show the isotopic pattern of selenium at m/z = 345.01965 ([M + H]⁺) and a possible hypothetical structure of the Se-compound.

Identification of Se-species in SAX fraction No. 5

The full scan spectrum of this low intensity SAX fraction contained two compounds with the characteristic Se-pattern, at m/z = 313.0169 and at m/z = 373.0305, respectively (see Fig. 5a). The mass of the first species was close to a well known Se-species, γ-Glu-Se-methylselenocysteine [C₉H₁₇N₂O₅Se⁺; m/z = 313.02972; −41 ppm] and its MS/MSspectrum contained several known fragments of this compound.^36–38 However, the relative intensities of the fragments (especially the unexpectedly low intensity of the m/z = 166.95 fragment and the interestingly high intensity of the m/z = 181.96 fragment) and the low abundance of this species in comparison with a previous report³⁶ raised doubts on this assignment. Finally, recalling the structure of an abundant Se-containing residue, selenocystathionine, in the analysed SEC fraction, this compound was assigned as an N-acetyl derivative of selenocystathionine (see Fig. 5b and Table 1). Possible explanations for the presence of this compound might be either its formation during the down-stream fermentation processes through the action of free acetyl-CoA or decomposition of the previously presented selenocystathionine derivatives. Note that the elemental composition and the mass are the same as those of γ-Glu-Se-methylselenocysteine.

Fig. 5 (a) ESI QTOF MS/MS analysis of SAX fraction No. 5 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic patterns of selenium at m/z = 313.01 ([M + H]⁺) and at m/z = 373.03 ([M + H]⁺), together with the CID spectra of these two pseudomolecular ions. (b) Proposed fragmentation pathways of the Se-compound at m/z = 313.01, N-acetyl-selenocystathionine. (c) Proposed fragmentation pathways of the Se-compound at m/z = 373.03, 2,3-DHP-selenohomolanthionine. For accurate mass information see Table 1.

The other Se-species at m/z = 373.0305 was present in a very low concentration and the background of its MS/MSspectrum was hence relatively high (see Fig. 5a). On the other hand, some of the fragments provided useful hints regarding the possible structure: for example, the common appearance of the fragments at m/z = 102.05 and m/z = 183.97 is characteristic of selenohomolanthionine.³⁴ Therefore, the assignment of 2,3-dihydroxy-propionyl-selenohomolanthionine was privileged; its proposed fragmentation pathway is presented in Fig. 5c, while the accurate mass information is listed in Table 1. Interestingly, this proposed composition differs from the 2,3-dihydroxy-propionyl-selenocystathionine species by a –CH₂– group, possibly creating a homologue series of the m/z = 345, 359, and 373 species.

It is to be noted that a Se-species with the mass of m/z = 373 has been reported by McSheehy et al.³⁹ from a Se-yeast sample, without a successful structure elucidation. As the intensity of the actual species was low and a different instrumental set-up was used, the fragmentation data cannot be compared and no information about their possible identity can be provided.

Identification of Se-species in SAX fraction No. 6

The full scan spectrum of this SAX fraction contained one abundant compound with the characteristic Se-pattern, at m/z = 313.0147 (see ESI† -Fig. 4a). The obtained MS/MS data proved this species to be γ-Glu-Se-methylselenocysteine, a compound formerly identified in several Se-enriched samples such as yeast,³⁶ garlic,³⁸ and shallot.⁴⁰ Interestingly, its relative concentration compared to the other Se-species of this water-soluble SEC fraction is low (see Fig. 1b), contrary to the observation of Goenaga-Infante et al.,³⁶ where γ-Glu-Se-methylselenocysteine was found to be one of the major water-soluble Se-metabolites in Se-yeast. Clearly, the concentration of Se-metabolites depends on the actual fermentation parameters and the yeast strain used. The proposed structures of the fragments of this Se-species are presented in ESI† -Fig. 4b, while the related accurate mass information is listed in Table 1.

Identification of Se-species in SAX fraction No. 8

The full scan spectrum of this SAX fraction contained one abundant compound with the characteristic Se-pattern, at m/z = 370.0406 (see Fig. 6a). The MS/MS data showed several fragments with characteristic masses, such as the triplet of m/z = 84.04, 112.03 and 130.04 and the doublet of m/z = 137.97 and 223.97. These five fragments can be common with γ-Glu-Se-methylselenocysteine. The mass difference between the unknown compound and γ-Glu-Se-methylselenocysteine is 57 Da—this value may refer to a Gly residue, similarly to the couple of m/z = 661 and 604 species of selenoglutathione conjugates.²⁷ Indeed, all the fragments detected can be assigned by supposing the structure of a novel Se-species, Se-methylselenoglutathione. On the other hand, the mass accuracy data of both the pseudomolecular ion and the daughter ions were often low (29–93 ppm). Therefore, the HILIC–Orbitrap set-up was finally addressed due to the relative purity and high selenium concentration.

Fig. 6 (a) ESI QTOF MS/MS analysis of the SAX fraction No. 8 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic pattern of selenium at m/z = 370.04 ([M + H]⁺) and the CID spectrum of the pseudo-molecular ion. (b) LTQ OrbitrapMS analysis of the SAX fraction No. 8 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic pattern of selenium at m/z = 370.05124 ([M + H]⁺), and the spectra of the MSⁿ (where n = 2–4) trapping & fragmentation events. (c) Proposed fragmentation pathways of the detected Se-compound, Se-methylselenoglutathione. For accurate mass information see Table 1.

The mass filtering of the HPLC data for the accurate mass of the proposed compound, Se-methylselenoglutathione ((C₁₁H₂₀N₃O₆Se⁺), m/z = 370.05118 ± 0.0025 Da), resulted in the detection of a monoselenised Se-compound with a close accurate mass (m/z = 370.05124; 0.16 ppm; Fig. 6b). Due to the relatively low mass of the pseudomolecular ion and the parent ion chosen for the series of MSⁿ experiments, and the bottom-up subsequent use²⁸ of identified fragments, all the fragments detected on the MS⁴, MS³ and MS² levels could be identified with unique hits by the ECT. On the MS⁴ and MS³ levels, all three non-selenised fragments could be assigned as C₄H₆NO⁺ [m/z = 84.04439; −1.07 ppm], C₅H₆NO₂⁺ [m/z = 112.03930; −0.71 ppm], and C₅H₈NO₃⁺ [m/z = 130.04987; −0.85 ppm], respectively. In the MS² experiment, the three selenised fragments were unambiguously assigned due to the significant mass defect of Se and the 5 ppm mass accuracy provided by the Orbitrap instrument as C₆H₁₀NO₃Se⁺ [m/z = 223.98204; −0.40 ppm], C₆H₁₃N₂O₃Se⁺ [m/z = 241.00859; −0.12 ppm], and C₉H₁₅N₂O₄Se⁺ [m/z = 295.01916; 0.64 ppm], respectively. This information limits the list of possible elemental formulae of the pseudomolecular ion to the unique possibility: C₁₁H₂₀N₃O₆Se⁺ [m/z = 370.05118; 0.16 ppm]. Fig. 6c presents the structures of the proposed fragments of Se-methylselenoglutathione, while Table 1 carries the relevant accurate mass information.

Identification of Se-species in SAX fraction No. 9

The full scan spectrum of this SAX fraction contained two compounds with the characteristic isotopic pattern of compounds with one Se-atom, one at m/z = 166.9557 and a second one at m/z = 271.9963 (see Fig. 7a). The first compound was found in the MS/MSspectrum of the m/z = 271.9963 ion (Fig. 7b), therefore it was assigned as its abundant in-source fragment. Another important hint for the possible structure of this new Se-species was its relatively high mass defect compared to, e.g., selenocystathionine [C₇H₁₅N₂O₄Se⁺; m/z = 271.01916]. This possibly refers to a larger number of oxygen atoms in the elemental formula. The MS/MSspectrum is of low abundance, which may be attributed to the decomposition of the compound in the source, reflected by the similar intensity of the in-source fragment at m/z = 166.9557. A possible structure of 2,3-dihydroxy-propionyl-Se-methylselenocysteine can be proposed (see Fig. 7c and Table 1) with the elemental composition of C₇H₁₄NO₅Se⁺ [m/z = 272.00317; −25 ppm].

Fig. 7 (a) ESI QTOF MS/MS analysis of SAX fraction No. 9 (cf. Fig. 1b). Full-scan mass spectrum taken at the TIC apex. The insets show the isotopic patterns of selenium at m/z = 271.99 ([M + H]⁺) and at m/z = 166.96 ([M + H]⁺), together with the CID spectrum of the ion at m/z = 271.99. (b) Proposed fragmentation pathways of the Se-compound, 2,3-DHP-methylselenocysteine. For accurate mass information see Table 1.

Conclusions

Electrospray MS analysis of the nine Se-containing fractions separated by strong anion-exchange chromatography allowed the assignment of structures to nine species (≤400 Da), in addition to selenate and a Se-species of which the identity was deduced on the basis of the empiric formula. Out of the nine compounds, only one (γ-Glu-Se-methylselenocysteine) has been reported previously in Se-rich yeast;³⁶ two others (selenohomolanthionine, γ-Glu-selenocystathionine) have been reported in other biological samples. All nine Se-species belong to the group of non-proteinaceous amino acid derivatives. Future studies should focus on these, likely to be present in marketed supplements, water-soluble low molecular weight Se-species (including three derivatives of Se-methylselenocysteine, evoked as a putative anti-cancer agent)⁷ rather than on selenomethionine.

The success of screening procedures based on an orthogonal chromatographic clean-up depends critically on the stability of the analyte. The stability, especially against oxidation, can either depend directly on the molecular environment of the Se atom or may be conferred by the matrix (e.g., by providing authentic or artificially supplemented reducing agents in excess). All the Se-species identified in this study contain the selenium atom either in Se-methyl form or in the –CH₂–Se–CH₂– sequence that confers them high stability owing to the absence of a directly accessible prone-to-oxidation selenol group. Failing to stabilize selenol groups can be responsible for the unsuccessful ESI- MS analyses of water-soluble Se-species, as e.g. in a recent study of garlic.⁴¹ On the other hand, even Se-methyl groups may undergo a usually reversible oxidation process as studied in details by Pedrero et al.⁴² and the possibility of artifact formation cannot be totally excluded.⁴³ It is clear that the Se-species identified in final formulations and products of Se-yeast origin can differ from the genuine Se-species synthesized by yeast cells.

The study confirmed the potential of electrospray QTOF MS/MS for the identification of selenium metabolites as well as its major limitation which is the low intrascan dynamic range demanding an extreme purity of the analyte at the moment of its arrival at the detector. The 10–50 ppm mass accuracy achieved in practice with the QTOF instrument combined with the MS² fragmentation pattern is in many cases sufficient to confirm the identity of expected species and even to identify the new ones—however, the lack of the corresponding authentic standards definitely hampers unambiguous identification. Therefore, hybrid linear ion trap/Orbitrap MSⁿ was indispensable for the detection and identification of less abundant species in less pure fractions. Clearly, the sub-ppm mass accuracy and the MSⁿ structural analysis make this technique (or FT ICR MSⁿ) a privileged tool for selenium metabolomics in Se-rich yeast.

Acknowledgements

M. D. acknowledges a Marie-Curie Fellowship (Grant MERG-CT-2006-044951). The authors thank Eric Génin (Thermo France) and Dr Hugues Preud’homme (UMR 5254) for their help with the analyses, Prof. Dr Judit Kosáry (CUB) for her help in structure elucidation and Dr Gérard Bertin (Alltech) for providing the Se-rich yeast sample.

References

1. M. P. Rayman, Lancet, 2000, 356, 233–241.
2. M. P. Rayman, Br. J. Nutr., 2004, 92, 557–573.
3. G. N. Schrauzer, J. Am. Coll. Nutr., 2001, 20, 1–4.
4. M. E. Reid, A. J. Duffield-Lillico, E. Slate, N. Natarajan, B. Turnbull, E. Jacobs, G. F. Combs, D. S. Alberts, L. C. Clark and J. R. Marshall, Nutr. Cancer, 2008, 60, 155–163.
5. R. F. Burk, B. K. Norsworthy, K. E. Hill, A. K. Motley and D. W. Byrne, Cancer Epidemiol. Biomarkers Prev., 2006, 15, 804–810.
6. L. C. Clark, G. F. Combs, B. W. Turnbull, E. H. Slate, D. K. Chalker, J. Chow, L. S. Davis, R. A. Glover, G. F. Graham, E. G. Gross, A. Krongrad, J. L. Lesher, H. K. Park, B. B. Sanders, C. L. Smith and J. R. Taylor, J. Am. Med. Assoc., 1996, 276, 1957–1963.
7. M. P. Rayman, H. G. Infante and M. Sargent, Br. J. Nutr., 2008, 100, 238–253.
8. R. Lobinski, J. S. Edmonds, K. T. Suzuki and P. C. Uden, Pure Appl. Chem., 2000, 72, 447–461.
9. A. Polatajko, N. Jakubowski and J. Szpunar, J. Anal. At. Spectrom., 2006, 21, 639–654.
10. H. G. Infante, R. Hearn and T. Catterick, Anal. Bioanal. Chem., 2005, 382, 957–967.
11. S. McSheehy and Z. Mester, TrAC, Trends Anal. Chem., 2003, 22, 210–224.
12. Y. Ogra, Anal. Bioanal. Chem., 2008, 390, 1685–1689.
13. Z. Mester, S. Willie, L. Yang, R. Sturgeon, J. A. Caruso, M. L. Fernandez, P. Fodor, R. J. Goldschmidt, H. Goenaga-Infante, R. Lobinski, P. Maxwell, S. McSheehy, A. Polatajko, B. B. M. Sadi, A. Sanz-Medel, C. Scriver, J. Szpunar, R. Wahlen and W. Wolf, Anal. Bioanal. Chem., 2006, 385, 168–180.
14. H. Goenaga-Infante, R. Sturgeon, J. Turner, R. Hearn, M. Sargent, P. Maxwell, L. Yang, A. Barzev, Z. Pedrero, C. Cámara, V. Díaz Huerta, M. Fernández Sánchez, A. Sanz-Medel, K. Emese, P. Fodor, W. Wolf, R. Goldschmidt, V. Vacchina, J. Szpunar, L. Valiente, R. Huertas, G. Labarraque, C. Davis, R. Zeisler, G. Turk, E. Rizzio, L. Mackay, R. Myors, D. Saxby, S. Askew, W. Chao and W. Jun, Anal. Bioanal. Chem., 2008, 390, 629–642.
15. S. McSheehy, J. Kelly, L. Tessier and Z. Mester, Analyst, 2005, 130, 35–37.
16. L. Ouerdane and Z. Mester, J. Agric. Food Chem., 2008, 56, 11792–11799.
17. E. H. Larsen, M. Hansen, H. Paulin, S. Moesgaard, M. Reid and M. Rayman, J. AOAC Int., 2004, 87, 225–232.
18. S. M. Lippman, E. A. Klein, P. J. Goodman, M. S. Lucia, I. M. Thompson, L. G. Ford, H. L. Parnes, L. M. Minasian, J. M. Gaziano, J. A. Hartline, J. K. Parsons, J. D. Bearden III, E. D. Crawford, G. E. Goodman, J. Claudio, E. Winquist, E. D. Cook, D. D. Karp, P. Walther, M. M. Lieber, A. R. Kristal, A. K. Darke, K. B. Arnold, P. A. Ganz, R. M. Santella, D. Albanes, P. R. Taylor, J. L. Probstfield, T. J. Jagpal, J. J. Crowley, F. L. Meyskens Jr, L. H. Baker and C. A. Coltman Jr, J. Am. Med. Assoc., 2009, 301, 39–51.
19. S. M. Lippman, P. J. Goodman, E. A. Klein, H. L. Parnes, I. M. Thompson, A. R. Kristal, R. M. Santella, J. L. Probstfield, C. M. Moinpour, D. Albanes, P. R. Taylor, L. M. Minasian, A. Hoque, S. M. Thomas, J. J. Crowley, J. M. Gaziano, J. L. Stanford, E. D. Cook, N. E. Fleshner, M. M. Lieber, P. J. Walther, F. R. Khuri, D. D. Karp, G. G. Schwartz, L. G. Ford and C. A. Coltman, J. Natl. Cancer Inst., 2005, 97, 94–102.
20. C. Casiot, V. Vacchina, H. Chassaigne, J. Szpunar, M. Potin-Gautier and R. Lobinski, Anal. Commun., 1999, 36, 77–80.
21. B. Gammelgaard, C. Gabel-Jensen, S. Sturup and H. R. Hansen, Anal. Bioanal. Chem., 2008, 390, 1691–1706.
22. E. Dumont, F. Vanhaecke and R. Cornelis, Anal. Bioanal. Chem., 2006, 385, 1304–1323.
23. J. Szpunar, Analyst, 2005, 130, 442–465.
24. J. Szpunar and R. Lobinski, Anal. Bioanal. Chem., 2002, 373, 404–411.
25. S. McSheehy, F. Pannier, J. Szpunar, M. Potin-Gautier and R. Lobinski, Analyst, 2002, 127, 223–229.
26. M. Dernovics, T. Garcia-Barrera, K. Bierla, H. Preud'homme and R. Lobinski, Analyst, 2007, 132, 439–449.
27. M. Dernovics and R. Lobinski, J. Anal. At. Spectrom., 2008, 23, 72–83.
28. M. Dernovics and R. Lobinski, Anal. Chem., 2008, 80, 3975–3984.
29. A. Polatajko, B. Banas, J. R. Encinar and J. Szpunar, Anal. Bioanal. Chem., 2005, 381, 844–849.
30. J. R. Encinar, M. Sliwka-Kaszynska, A. Polatajko, V. Vacchina and J. Szpunar, Anal. Chim. Acta, 2003, 500, 171–183.
31. J. F. García-Reyes, M. Dernovics, P. Giusti and R. Lobinski, J. Anal. At. Spectrom., 2006, 21, 655–665.
32. J. F. García-Reyes, M. Dernovics, P. Ortega-Barrales, A. R. Fernandez-Alba and A. Molina-Diaz, J. Anal. At. Spectrom., 2007, 22, 947–959.
33. J. R. Encinar, L. Ouerdane, W. Buchmann, J. Tortajada, R. Lobinski and J. Szpunar, Anal. Chem., 2003, 75, 3765–3774.
34. Y. Ogra, T. Kitaguchi, K. Ishiwata, N. Suzuki, Y. Iwashita and K. T. Suzuki, J. Anal. At. Spectrom., 2007, 22, 1390–1396.
35. S. N. Nigam and W. B. McConnell, Biochim. Biophys. Acta, 1976, 437, 116–121.
36. H. Goenaga-Infante, G. O'Connor, M. Rayman, R. Wahlen, J. E. Spallholz, R. Hearn and T. Catterick, J. Anal. At. Spectrom., 2005, 20, 864–870.
37. S. McSheehy, W. Yang, F. Pannier, J. Szpunar, R. Lobinski, J. Auger and M. Potin-Gautier, Anal. Chim. Acta, 2000, 421, 147–153.
38. E. H. Larsen, R. Lobinski, K. Burger-Meyer, M. Hansen, R. Ruzik, L. Mazurowska, P. H. Rasmussen, J. J. Sloth, O. Scholten and C. Kik, Anal. Bioanal. Chem., 2006, 385, 1098–1108.
39. S. McSheehy, P. Pohl, J. Szpunar, M. Potin-Gautier and R. Lobinski, J. Anal. At. Spectrom., 2001, 16, 68–73.
40. Y. Ogra, K. Ishiwata, Y. Iwashita and K. T. Suzuki, J. Chromatogr., A, 2005, 1093, 118–125.
41. S. Mounicou, M. Dernovics, K. Bierla and J. Szpunar, Talanta, 2009, 77, 1877–1882.
42. Z. Pedrero, J. Ruiz Encinar, Y. Madrid and C. Cámara, J. Anal. At. Spectrom., 2007, 22, 1061–1066.
43. M. Dernovics, L. Ouerdane, L. Tastet, P. Giusti, H. Preud'homme and R. Lobinski, J. Anal. At. Spectrom., 2006, 21, 703–707.

---

Footnote

† Electronic supplementary information (ESI) available: Additional spectra (see text for details). See DOI: 10.1039/b901184f

---