BACKGROUND Selenium is an essential trace element and a typical bifunctional element that can affect human health if consumed in insufficient or excessive amounts. The biological activity of selenium depends not only on its intake level but also on its chemical speciation. Selenium comes in various speciation and is divided mainly into inorganic and organic selenium. Inorganic selenium includes selenate [Se(Ⅵ)] and selenite [Se(Ⅳ)], and organic selenium mainly includes selenocysteine (SeCys2), selenomethionine (SeMet), and methylselenocysteine (MeSeCys). It has been found that organic selenium has high bioactivity and bioavailability. At present, while the nutritional effects of selenium are drawing more and more attention, it is very important to analyze and study the different speciation of selenium in food. Since the analysis of selenium speciation is closely related to the sample matrix, the extraction efficiency and stability of different selenium speciation are also related to many factors. At present, the analysis method of selenium speciation in food is still in the research stage. Broccoli is rich in nutrients, such as protein, flavonoids, polyphenols, and vitamins, and is widely loved by people because it contains many kinds of thioglucosides and has a strong ability to gather selenium, which has antioxidant and anti-cancer medical values. Therefore, the analysis and study of selenium speciation in broccoli is of some significance. OBJECTIVES To establish a method for the determination of Se(Ⅵ), Se(Ⅳ), SeCys2, MeSeCys, and SeMet in commercial broccoli by high performance liquid chromatography-inductively coupled plasma-mass spectrometry (HPLC-ICP-MS). METHODS Firstly, the chromatographic conditions were selected by examining the separation and sensitivity of Se(Ⅵ), Se(Ⅳ), SeCys2, MeSeCys, and SeMet on a Hamilton PRP-X100 anion column with 40mmol/L diammonium hydrogen phosphate (pH=5 with 1% methanol) as the mobile phase and on a ZORBAX SB-Aq C18 reversed-phase column with 10mmol/L citric acid plus 5mmol/L sodium hexane-sulphonate (pH=4 with 1% methanol) as the mobile phase. Secondly, the sample pretreatment conditions were optimized, including the selection of extraction reagents, the amount of extraction reagents, and extraction time. Four extraction reagents (ultrapure water, Tris-HCl buffer solution, proteinase XIV and complex proteinase) were selected for optimization. The effect of proteinase XIV concentration on the extraction was investigated by adding 2, 4, 6, and 8mg/mL proteinase XIV to broccoli samples with selenium content of 0.81mg/kg (calculated as Se). The effect of adding 6mL, 10mL, 12mL, and 15mL of Tris-HCl buffer solution on the extraction was compared. The extraction time of the samples also had a great influence on the extraction efficiency of the selenium speciation. The effects of four extraction times of 1h, 3h, 5h and 7h on the extraction were compared. RESULTS The ZORBAX SB-Aq C18 separation system was used in this study because of the short analysis time and high sensitivity of each selenium speciation. Protease XIV was the most effective extraction reagent for selenium; therefore proteinase XIV was chosen as the extraction reagent. The concentration of selenium speciation increased with the concentration of proteinase XIV. The maximum concentration of selenium speciation was reached when the concentration of proteinase XIV was 6mg/mL. It was reported that the use of Tris-HCl buffer solution with proteinase XIV at appropriate pH conditions could further improve the extraction efficiency and maintain the stability of selenium speciation. The volume of Tris-HCl buffer increased, the extraction efficiency of each selenium speciation gradually increased and then decreased, and the final selection of Tris-HCl buffer solution addition was 12mL. A longer extraction time would help to increase the extraction effect, but too long an enzymatic digestion time would also cause a decrease in the stability of SeMet and SeCys2. To ensure high extraction efficiency and reduce the conversion of selenium speciation, an extraction time of 3h was preferred. After optimization and selection, the final analysis method was determined as follows: weighing a certain amount of broccoli sample into 12mL of Tris-HCl (pH=7.4, containing 6mg/mL proteinase XIV) at a concentration of 100mmol/L, vortexing and mixing, and then sonicating at 37℃ for 3h. After centrifugation, the extraction were eluted with 10mmol/L citric acid and 5mmol/L sodium hexane sulfonate (pH=4 with 1% methanol) on ZORBAX SB-Aq C18 reversed-phase column. ICP/MS was used for analysis and determination.This method can achieve effective separation and determination of five selenium speciation within 8 minutes. The linearity range of the method was 0.3-100.0μg/L, with linear correlation coefficients (r) greater than 0.999. The detection limits of Se(Ⅳ), Se(Ⅵ), MeSeCys, and SeMet were within the range of 1.2-6.0μg/kg (calculated as Se). The standard recovery tests were carried out on broccoli samples at low, medium, and high concentration. The recoveries of these four selenium speciation, Se(Ⅵ), Se(Ⅳ), MeSeCys and SeMet, were 81.9%-105.3% with relative standard deviations (RSD) less than 5%. The method established in this study was used to determine SeMet in the EU-certified reference material (ERM BC210a, wheat flour), and the measured value of SeMet was within the range of its standard values.More than 20 commercially available broccoli samples collected from different regions of China were analyzed and determined. The results showed that the selenium speciation in commercially available broccoli was mainly MeSeCys, with small amounts of Se(Ⅵ), Se(Ⅳ), and SeMet, and also a small amount of unknown selenium-containing compounds was also present.Two problems identified in the methodological study were explored. (1) The effect of proteinase XIV dosage on the stability of SeCys2 was investigated by adding 1, 2, 4, and 6mg/mL of proteinase XIV to SeCys2 standard solution, respectively. The results showed that as the concentration of proteinase XIV increased, the signal value of SeCys2 gradually decreased and the signal value of three unknown peaks gradually increased. At the same time, the recovery of SeCys2 in broccoli samples decreased to 10%. Based on the above conditions, it is assumed that the content of proteinase XIV and the matrix of broccoli samples affect the stability of SeCys2.(2) Three different broccoli samples were selected for Se(Ⅳ) standard recovery tests: fresh commercially available broccoli samples, freeze-dried powder of commercially available broccoli, and freeze-dried powder of broccoli fortified with Se(Ⅳ) selenium fertilizer. A certain amount of the above three samples was added with Se(Ⅳ) standard solution and 100mmol/L Tris-HCl (pH=7.4, containing 6mg/mL of proteinase XIV). The determination was then carried out according to the proposed analytical method and the mean recoveries of the three samples were found to be 81.1%, 69.5% and 1.53%, respectively. The Kruskal-Wallis rank sum test showed that the recoveries of Se(Ⅳ) were significantly different among the three samples (p < 0.05). Previous investigations have found that phenolic substances can affect the stability of Se(Ⅳ) and that the addition of selenium fertilizer during the growth of broccoli can change the phenolics. Based on the above, it is assumed that the presence of phenolics in broccoli samples may affect the determination of Se(IV). CONCLUSIONS A method for the determination of Se(Ⅳ), Se(Ⅵ), MeSeCys, and SeMet in commercially available broccoli by HPLC-ICP-MS is established by selecting and optimizing the sample pretreatment and analytical conditions. The Tris-HCl buffer solution containing proteinase XIV is chosen for the extraction of samples.It is found that the stability of SeCys2 is affected by the concentration of proteinase XIV and broccoli samples matrix. It is hypothesized that the presence of large amounts of phenolics in the samples can affect the determination of Se(Ⅳ) for reasons to be further explored.