Abstract
Pyrrolizidine alkaloids （PAs） are natural toxins widely distributed in plants， which naturally occur in about 3% of the world's flowering plants. To date， more than 660 PAs and their nitrogen oxides have been identified in over 6 000 plants. Some PAs are hepatotoxic， genotoxic and tumorigenic， posing significant health risks to humans. These alkaloids are commonly detected as contaminants in various food products， including tea， grains， milk， honey， as well as plant-derived pharmaceuticals and dietary supplements. Currently， most studies on the quantitative methods for PAs focus on a limited number of PAs and employ an additive quantification strategy， largely due to the challenges associated with chromatographic separation of isomers. These approaches limit the ability to assess exposure and health risk accurately. Herein， a method was established to quantify 35 PAs individually in dried tea samples using ultra-high performance liquid chromatography-triple quadrupole mass spectrometry （UHPLC-MS/MS）. The 35 target compounds were divided into two groups. The first group included 30 PAs， while the second group consisted of 5 PAs. These compounds were separated on Waters Acquity BEH C18 （150 mm×2.1 mm， 1.7 μm） and Thermo Acclaim(TM) C30 （150 mm×2.1 mm， 3.0 μm） chromatographic columns， respectively. Mobile phases were H(2)O containning 5 mmol/L ammonium formate and 0.13% formic acid （pH=3） and methanol-acetonitrile （4∶6， volume raio） containing 0.1% formic acid. The 33 target compounds were separated and 2 isomers co-eluted. Under positive-ion electrospray ionization （ESI） and multiple reaction monitoring （MRM） mode， target compounds were quantified using the external standard method with the matrix-matched standard curve. The results demonstrated that all target compounds showed good linearity （r(2)>0.99） in their respective mass concentration ranges. The limits of detection （LODs） and quantification （LOQs） of the method were in the range of 0.2-8.0 µg/kg and 0.5-25.0 µg/kg， respectively. The average recoveries of more than 89% compounds were in the range of 70%-130% at spiked levels of 1， 2， and 5 fold-LOQs and the relative standard deviations （RSDs） were less than 20% （RSDs of lasiocarpine and echimidine were less than 30%）. Moreover， the quantitative method of the 35 PAs was applied to 21 dark tea and 30 black tea samples from Yunnan and Fujian Province， where PAs were detected in 4 dark samples with total contents ranging from 5.07 to 15.48 µg/kg and were not detected in black tea samples. The detection rate of PAs in dark tea samples was higher than that of black tea samples. That might be because most dark tea was processed from fresh leaves picked by machines， and this attribution could be associated with the mixing of weeds containing PAs during tea harvesting. And the detected concentrations of all samples were lower than the maximum levels of tea in European Union regulations， indicating that the tea samples involved in this research are basically safe. In brief， the quantitative method of the 35 PAs facilitates the analysis of the occurrence， composition and potential risks in tea samples.