Bente Gammelgaard, Kim Grimstrup Madsen, Jesper Bjerrum, Lars Bendahl, Ole Jøns, Jørgen Olsen and Ulrik Sidenius
Department of Analytical Chemistry, The Royal Danish School of Pharmacy, Universitetsparken 2, DK-2100 Copenhagen, Denmark. E-mail: bg@dfh.dk
First published on 27th November 2002
When humans are supplied with selenium-containing nutritional preparations, one of the selenium-containing metabolites in urine increases relatively more than the other selenium metabolites. The purpose of this study was to identify this major selenium metabolite. Urine samples from six male volunteers were collected and analysed by ion-pair chromatography with ICP-MS detection for this major selenium metabolite. Samples containing the metabolite were pooled and solid phase extracted to remove ionic substances. The extracted pool was purified and preconcentrated twice by preparative reversed-phase chromatography. The fractions containing the selenium metabolite were collected and further purified by size exclusion chromatography. It was not possible to ionize the selenium metabolite by electrospray ionization mass spectrometry, ESI-MS. Instead, atmospheric pressure chemical ionization, APCI, was applied. The m/z of the selenium metabolite was 300 for the 80Se isotope. MS/MS experiments indicated that the metabolite was a selenosugar, and it is proposed that the selenium metabolite is a Se-methyl-N-acetylselenohexosamine.
The metabolism of selenium from the administration of a selenium compound until the incorporation of selenium in the essential selenoproteins is fairly well elucidated and understood. The metabolism is often described by a model developed by Ganther.5,6 According to this model, the ingested selenium is converted to the hydrogen selenide ion (HSe−) prior to incorporation as selenocysteine into the essential selenoproteins. Excess selenium is then excreted after consecutive methylation processes to monomethylselenol (CH3SeH), dimethyl selenide ((CH3)2Se) and dimethyl diselenide ((CH3)2Se2), which are exhaled during respiration. The only suggested urinary excretion product in this model is the further methylated trimethylselenonium ion, TMSe ((CH3)3Se+). This model, however, is primarily based on experiments with rats after supplementation of their diet with large doses of selenium.
After the introduction of hyphenated techniques with element selective detection, especially HPLC-ICP-MS, several reports on selenium speciation in biological material using different chromatographic systems appeared. Extensive reviews on selenium speciation in biological material have been given by Lobinski et al.7and Uden.8 From these studies it appears that TMSe is not the only selenium excretion product in human urine. Several selenium metabolites have been separated in basal human urine9–12 as well as human urine from individuals supplemented with different selenium preparations.13–16
The identities of some of these metabolites were suggested on the basis of co-elution with available standards in various chromatographic systems. Thus, the selenium species suggested to be present in human urine are TMSe11–13,14,17 in small amounts and not in all samples, SeMet14,15,16 after supplementation with SeMet, and selenite in low concentrations in some samples of basal human urine.18
The complex urine matrix and the low concentrations of selenium in basal human urine makes purification and preconcentration of the urine samples mandatory prior to identification by mass spectrometry. However, Cao et al. identified selenomethionine and selenocystamine in a human urine sample from an individual who was supplemented with 400 µg of SeMet. They used tandem MS in the multi-reaction monitoring mode, in which one or more transitions of precursor to product ions were monitored in tandem.16 In a recent study, Ogra et al.19 identified a novel metabolite in rat urine by ESI tandem MS. The authors suggested that the metabolite was a Se-methyl-N-acetylselenohexosamine. They also synthesized 2-acetamide-1,2-dideoxy-β-D-glucopyranosyl methylselenide and observed that the mass spectrum of this compound was identical to the spectrum of the metabolite.
In previous studies we observed that although several selenium metabolites are present in human urine, the amount of one of these metabolites increased distinctively more than the rest after supplementation with selenomethionine.15 This unidentified selenium metabolite did not co-elute with any available standards in chromatographic systems. Thus, the purpose of this work was to separate, purify and preconcentrate urine in order to identify this metabolite by mass spectrometry.
Sampler and skimmer cones were made of platinum. The plasma and auxilliary argon flow rates were 14 and 1.2 l min−1, respectively. The nebulization argon flow was optimized in each eluent. The RF power was 1400 W.
The data acquisition parameters were: dwell time, 500 ms; sweeps/reading, 1; readings/replicate, between 400 and 700. The 78Se and 82Se isotopes were monitored.
The Agilent 1100 series HPLC system consisted of a G1376A capillary pump, a G1313A autosampler, G1314A wavelength detector, G1316A column compartment and a G1379A de-gasser controlled by the ChemStation software, all from the Agilent 1100 series (Agilent Technologies, Waldbron, Germany). The UV detector was operated at 254 nm. The outlet of the analytical columns was coupled to the ICP-MS system by a laboratory made direct injection nebulizer, which has previously been described,20via a 65 µm id, 508 µm od length of capillary PEEK tubing (Upchurch Scientific, Oak Harbor, WA, USA).
The pump for preparative chromatography was a Jasco 880-PU (Jasco, Gross-Umstadt, Germany) equipped with a Rheodyne 7125 injection valve (Rheodyne, Cotati, CA, USA).
Electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) mass spectrometry was performed by direct infusion into a Finnigan MAT LCQ ion-trap mass spectrometer (Finnigan, San Jose, CA, USA) using the following settings.
ESI: sheath/auxiliary gas flow 60/0, spray voltage: +4.5 kV, capillary temperature: 200 °C, flow rate: 5 µl min−1.
APCI: sheath/auxiliary gas flow 60/0, capillary temperature: 150 °C, octapole RF amplitude: 400, flow rate: 50 µl min−1, discharge current: 5 µA, vaporizer temperature: 490 °C.
Evaporations were performed on a rotary evaporator (Büchi Rotavapor, Büchi Laboratoriums Technik AG, Flawil, Schwitzerland) by heating to 45 °C.
Stock standard solutions of 10 mg l−1 Se were prepared in water from selenomethionine, SeMet, selenocystine, SeCys, Se-(methyl)selenocysteine hydrochloride, MeSeCys, selenourea, SeU, selenocystamine, SeCA, selenoethionine, SeEt (all Sigma, St. Louis, MO, USA), Se-(methyl)selenomethionine iodide, MeSeMet, synthesized according to ref. 21 and trimethylselenonium iodide, TMSe, synthesized according to ref. 22. The stock standard solutions were standardized against a 1.001 g l−1 PE pure atomic spectroscopy standard (PerkinElmer). Working standard solutions were prepared daily in water.
Preparative reversed-phase chromatography. The column was a Luna C18(2) pore size 100Å, 250 mm × 15 mm id (Phenomenex) protected by a SecurityGuard Cartridge C18, 10 mm × 10 mm, in a SecurityGuard Cartridge Holder (Phenomenex). The eluent consisted of 10 mM ammonium formate (ammonium hydroxide 25% (Merck), formic acid (Ferak, Berlin, Germany)), pH 7, and 5% methanol. The flow rate was 10 ml min−1 and the injection volume was 1000 µl.
Size exclusion chromatography. The column was a Shodex Asahipak GS-320 HQ, 300 mm × 7.6 mm id. The eluent was 0.1% (v/v) formic acid in 3% methanol. The column temperature was kept constant at 40 °C. The flow rate was 600 µl min−1 and the injection volume was 100 µl.
The cartridges were conditioned with 3 ml of methanol, followed by 3 ml of 10 mM ammonium formate, pH 7. 10 ml of urine pool diluted 1 + 1 with 10 mM ammonium formate, pH 7, was applied stepwise followed by elution with 6 ml of 25% methanol. The combined methanol eluates were evaporated to dryness and the residue was reconstituted in 20 ml of 10 mM ammonium formate in 5% methanol.
The total selenium concentrations in the urine samples varied between <2.2 and 144 µg l−1, while the concentrations of the metabolite estimated on the basis of the peak height varied between 0 and 100 µg l−1. The time for maximum excretion varied between 4 and 10 h and after 15 h the selenium concentration had reached the basal level for most individuals. There was no difference in the profiles of the excretion curves of total selenium and the selenium metabolite, i.e. high total selenium concentrations resulted in high concentrations of the metabolite.
An example of analysis of one of the urine samples by ion-pair chromatography with UV and ICP-MS detection is shown in Fig. 1. The selenium compound eluting at 2 min is the metabolite to be identified. It appears from the UV-absorbance that the selenium compound in this system eluted together with several other compounds in large concentrations as the UV absorbance exceeded 2 absorbance units.
Fig. 1 Chromatogram of a urine sample analysed by ion-pairing chromatography. Column: Luna C8, eluent: 0.1% HFBA in 20% methanol. Thin line: UV absorbance, bold line: 82Se. |
Elution with 100% methanol resulted in very coloured effluents. Different concentrations of methanol were examined in the elution step in order to improve the selectivity of the purification step and thus reduce the elution of the coloured compounds. 25% methanol was chosen as the eluent as a change in the eluent from 100% methanol to 25% methanol did not decrease the yield of the SPE step, while the cartridge still retained a large part of the coloured compounds. Hence, the cartridges were eluted with 6 ml of 25% methanol in the following procedures. However, the total yield of the SPE step is only about 30% as part of the metabolite runs through the cartridge during application of the urine sample. A washing step between sample application and elution was not included in the procedure as this resulted in substantial loss of the selenium metabolite owing to its hydrophilic properties. The sodium and potassium concentrations of the urine pool were compared to the concentrations of the SPE eluate by flame emission spectrometry analysis. The concentrations of sodium and potassium in the eluate were 41 mM and 28 mM, respectively, which corresponded to 50% of the salt concentrations in the urine pool.
The eluates were combined and evaporated to dryness on a rotary evaporator. The evaporation process did not decompose the selenium metabolite as chromatography of the sample before and after evaporation did not change the retention time of the compound.
A chromatogram of the urine pool of the collected samples analysed in the reversed phase system is shown in Fig. 2. It appears that separation of the main selenium metabolite from several other selenium species could be achieved. Thus, urine samples contain several selenium species which is in accordance with our earlier findings and results by others.9,11,12,13,16 The species that have previously been suggested to be present in human urine-selenite, TMSe and SeMet, are charged compounds which would elute before this unknown, probably neutral, compound. This was confirmed by analysing standards. In this system, selenite, TMSe and SeCys eluted in the front just after 1.7 min, selenourea eluted after 2.5 min and SeMet after 3.7 min (results not shown). Selenocystamine, which was identified by Cao et al. in a human urine sample by electrospray ionization mass spectrometry,16 eluted at 9.3 min. However, in the ion-pair chromatographic system, selenocystamine eluted after 16 min. We have never detected selenium species in urine with that long retention time in this chromatographic system.
Fig. 2 Chromatogram of the urine pool analysed by reversed phase chromatography. Column: Luna C18, Eluent: 10 mM ammonium formate, pH 7 in 5% methanol. Thin line: UV-absorbance, bold line: 82Se. |
The retention time of the main selenium metabolite was 6 min. This was confirmed by analysing the urine pool after solid-phase extraction in the HFBA system. Only the peak eluting after 2 min was now present. Analysing the same extract in the reversed phase system resulted in a peak eluting after 6 min. A chromatogram of the eluate from the solid phase extraction together with the trace from the UV absorbance is shown in Fig. 3. When comparing Figs. 2 and 3, it appears that the selenium concentration had increased about 20 fold while the UV absorbance only increased by 50%. However, the selenium metabolite eluted together with some UV-absorbing species. Hence, attempts to identify the selenium metabolite by ESI-MS at this stage were not successful because of extensive background.
Fig. 3 Chromatogram of eluate from the solid phase extraction analysed by reversed phase chromatography. Column: Luna C18, Eluent: 10 mM ammonium formate, pH 7 in 5% methanol. Thin line: UV absorbance, bold line: 82Se. |
When the eluate from the solid phase extraction was applied to the preparative system, the collected fractions were pale yellow, but after repeating this preparative chromatographic step once the collected fractions were colourless, indicating a further purification of the sample.
In preparative chromatography, long retention times are preferable as this often improves the probability of obtaining a selective separation. When analysing simultaneously for sodium during the preparative chromatography, it appeared that the sodium concentration was efficiently reduced during the first 6 min of elution, resulting in a concentration less than 0.3 µM of Na in the eluate when fractions were collected.
Fig. 4 Chromatograms of selenium fraction collected from reversed phase chromatography before and after size exclusion chromatography. Column: Asahipak GS-320 HQ, Eluent: 0.1% formic acid in 3% methanol. Thin line: UV-absorbance, bold line: 82Se. |
However, any attempt to detect the unidentified selenium metabolite by ESI-MS was in vain, probably because the metabolite is a neutral compound and difficult to ionize.
An attempt to identify the metabolite with electron impact MS was not succesful as no selenium isotope pattern was detectable. This fierce ionization mode probably totally decomposed the molecule. However, the compounds suggested by the instruments database showed ring structures containing oxygen indicating the compound could contain a sugar moiety.
The mass spectrum obtained with APCI ionization is shown in Fig. 5. The m/z of the 80Se metabolite was 300 and the intensities of the m/z 298, 297, 296 and 302 peaks corresponds to natural abundances of the selenium isotopes (76Se: 9%, 77Se: 8%, 78Se: 24%, 80Se: 50% and 82Se: 9%). MS-MS fragmentation on m/z 298, 300 and 302 all resulted in the same product ion, m/z 204, indicating that the MS-MS fragment did not contain selenium. This means that the selenium containing moiety was lost by the first fragmentation, probably as CH3SeH. Further fragmentation on m/z 204 resulted in the fragment ions m/z 186, 168, 144 and 126. This is shown in Fig. 6. The m/z of the parent ions together with the product ions and the possible leaving groups are summarized in Table 1.
Fig. 5 APCI mass spectrum of selenium containing fraction from size exclusion chromatography. |
Fig. 6 Fragmentation of m/z 298, 300 and 302 by (MS)2, (MS)3 and (MS)4. |
Parent ion m/z | Product ion m/z | Product |
---|---|---|
M | 298, 300, 302 | [M + H]+ |
298, 300, 302 | 204 | [M − CH3Se]+ |
204 | 186 | [204 − H2O]+ |
186 | 168 | [186 − H2O]+ |
186 | 144 | [186 − CH3CO + H]+ |
186 | 126 | [144 − H2O]+ |
186 | 138 | ? |
A selenium metabolite with the same mass has very recently been identified in rat urine by Ogra et al.19 by ESI tandem MS. The authors suggested that the rat metabolite was a Se-methyl-N-acetyl-selenohexosamine. They also synthesized 2-acetamide-1,2-dideoxy-β-D-glucopyranosyl methylselenide and observed that the mass spectrum of this compound was identical to the spectrum of the metabolite. However, there was a slight difference in the chromatographic retention times of the metabolite and the synthetic product.19
The fragmentation of the compound identified by Ogra et al.19 is in accordance with our results as we also observed m/z 204 as the first fragmentation product, which is known as the marker ion of hexose-N-acetyl+ in the ESI-MS detection of glycopeptides.24 Hence, the metabolite in human urine is probably the same as the metabolite from rat urine.
The (MS)2 and (MS)3 of m/z 204 to 186 and 168, respectively, corresponding to loss of water from hydroxy groups, and the (MS)3 fragmentation of m/z 186 to 144, indicating the presence of an acetyl group, are in accordance with the results obtained by Ogra et al.,19 although their results were obtained by a general increase in collision energy. The m/z of the parent ion and all product ions are odd numbers, indicating an uneven number of nitrogen atoms in the parent ion as well as in the product ions. The presence of an amide group is supported by the experiments with retainment on solid phase extraction cartridges, which showed that the molecule had no acidic or alkaline functional groups.
The metabolite is not a conjugation product of glucuronic acid, which is a common metabolic product, as incubation of urine samples with glucuronidases did not change the chromatographic behaviour of the selenium metabolite. Hence, our experiments support that the structure of the human selenium metabolite is a Se-methyl-N-acetylselenohexosamine.
The pathways for formation of aminosugars and subsequent acetylation of the amino group via the UDP-N-acetylhexosamines is well known as being a part of the biosynthesis of glycoproteins and glycolipids. Also, the conversion of galactose to glucose via the activated UDP-glucose is described.25 Hence, a possible way of formation of the selenosugar is addition of selenide to the UDP-carbon resulting in the leaving of the UDP-group followed by a methylation of the selenium or direct addition of methylselenide to the UDP-carbon. Identification of the sugar, however, is not possible by MS techniques as the N-acetylglucosamine as well as the N-acetylgalactosamine would generate the m/z 204.
Hence, the metabolite now found in human urine is probably the same as the metabolite from rat urine. This suggests that rat selenium metabolism is an appropriate model for human selenium metabolism. However, it seems as if the rat metabolism is more simple as it has been reported that in same chromatographic system rat urine showed only one peak, while human urine showed three peaks.10
The rat in the work of Ogra et al.19 was supplied with selenium as 2 mg l−1 sodium selenite in the drinking water for 5 weeks. The volunteers in this work were supplied with 200 µg of selenium daily for a week and 500 µg on the day of delivery as Selenoprecise, which are selenium enriched yeast tablets. In spite of this, the rats and the humans produced the same metabolite. This supports the view that selenium metabolism is independent of the ingested chemical form of selenium as ingested selenium is reduced to selenide and traverses this selenide pool before incorporation in selenoproteins or excretion.5
In conclusion, a selenosugar, probably a Se-methyl-N-acetylselenohexosamine has, for the first time, been isolated and identified in human urine.
This journal is © The Royal Society of Chemistry 2003 |