Determination of free selenomethionine in nutritional supplements by high-performance liquid chromatography coupled with thermochemical hydride generation atomic absorption spectrometry

Abstract

Selective isolation and HPLC–AAS protocols were developed to determine free selenomethionine (SeMet) in complex matrices such as nutritional supplements and mixtures of free amino acids. The selenoamino acid in alkaline solution was first derivatized with 1-fluoro-2,4-dinitrobenzene (FDNB, Sanger's reagent). After removal of excess of reagent by partitioning with diethyl ether, the N-dinitrophenyl (DNP)-derivatized SeMet was protonated by acidification and extracted with diethyl ether. DNP-SeMet was determined selectively by HPLC–thermochemical hydride generation (THG) AAS. A selective chromatographic mechanism based on π-electron interactions was optimized using a silica stationary phase derivatized with p-nitrophenyl moieties. Primary analytical validation parameters including stability, linearity and limit of detection were obtained using a purified DNP-SeMet standard which was characterized by ¹H/¹³C/⁷⁷Se NMR and fast atom bombardment MS techniques. The calibration graphs for sequential dilutions of this standard were linear (average r= 0.997) and the limit of detection from the resultant calibration graphs was estimated to be 17 ng as Se or 81 ng as DNP-SeMet. The standard was found to decompose slowly (0.08% h^–1) via photolytic cleavage of a Se—C bond. MS and HPLC–THG-AAS studies demonstrated that the cleavage intermediates (CH₃Se·) combined to form dimethyl diselenide. Ascorbic acid present in large amounts in nutritional supplements was revealed to be the major interferent during the DNP derivatization step. This interference was removed using a cation-exchange treatment of the crude extract prior to derivatization. The recovery of the SeMet from stock solutions and nutritional supplements was virtually quantitative (99 ± 4.5%). In the presence of a 500-fold excess of other amino acids, this recovery decreased (95 ± 2.6%), but remained in an acceptable range to allow the accurate determination of SeMet in a protein hydrolysate.