Screening for Plant Toxins in Honey and Herbal Beverage by Ultrahigh-Performance Liquid Chromatography-Ion Mobility-Quadrupole Time of Flight Mass Spectrometry


The standards of plant toxins were separated by a C18 column with gradient elution with 0.1% formic acid/water (V/V) and 0.1% formic acid/acetonitrile (V/V) as mobile phase and acquired by ion mobility-quadrupole time of flight mass spectrometry (IM-QTOF MS) in positive ion mode. A database of 308 plant toxins including retention time, collision cross-section (CCS) and its fragment ions was established. Honey dissolved in water or herbal beverage was extracted by acetonitrile and purified with PSA sorbent, and then acquired by ultrahigh-performance liquid chromatography IM-QTOFMS. The acquired data were processed by comparing with the database we established to confirm the target compounds. The average recoveries for samples at two levels ranged from 60.6% - 120.1%, with relative standard deviation (n = 6) less than 25%. The limit of quantitation for plant toxins ranged from 1 - 20 μg/kg. The developed screening method was used in determination of honey, herbal beverage and honey flavored tea beverage samples. The results showed that berberine was detected in one honey with 1 μg/kg and caffeine was present in some beverages with the concentration from 200 and 5500 μg/kg. This method could meet the requirement for rapid screening of plant toxins in honey and herbal beverage. It can be used for the quality control of honey and herbal beverage in enterprises or quality inspection departments. It also can be used in the rapid screening of food poisoning.

Share and Cite:

Guo, Q. , Yang, Y. , Li, J. , Shao, B. and Zhang, J. (2022) Screening for Plant Toxins in Honey and Herbal Beverage by Ultrahigh-Performance Liquid Chromatography-Ion Mobility-Quadrupole Time of Flight Mass Spectrometry. American Journal of Analytical Chemistry, 13, 108-134. doi: 10.4236/ajac.2022.133009.

1. Introduction

Honey and herbal beverage are the favorite functional foods due to natural to meet consumer wish for green, containing “organic and bioactive” components [1] [2]. The consumption of honey was 0.24 kg/person in 2018, and herbal beverage accounts for 13% of beverage consumption in China. For honey and beverage safety risk, one of the key topics related to public health issues, can originate from various factors, such as chemicals used in the production, or contaminants in raw materials [3]. The European Food Safety Authority (EFSA) reported that a tremendously high number of herbal products such as herbal teas, contain significant amounts of toxic alkaloids, such as nicotine, pyrrolizidine, or tropane alkaloids [4] [5].

Plant toxins exist in natural plant and are widely found in roots and fruits of plants. Some poisonous plant flowers are collected by bees and lead to phytotoxins in honey. Poisoning occurred due to eating honey containing Tripterygium wilfordii phytotoxins [6]. There was also report of poisoning due to eating honey contaminated with tutin, a plant-derived neurotoxin [7]. Some poisonous plants are easy to be confused with wild vegetables or Chinese herbal medicine, which makes human beings intake this kind of wild plant accidentally and causes undesirable poisoning. For example, Datura stramonium Linn, also known as Datura metel L, is easily misused as agricultural weed for animals, transferring it into milk and causing tropane alkaloids poisoning [8]. Literatures [9] [10] had shown that tea, honey and traditional medicine contain pyrrolizidine alkaloids (PAs) which are highly toxic to animal, human and infants [11].

Most plant toxins displayed neurotoxicity [12] and developmental toxicity [13] which can cause minor liver damage [9] to the consumers, including chills, sweating, sleepy and other symptoms. Their high toxicity poses a serious health risk for the consumers. Now most of the countries in the world have limit standard for contaminants or pesticide residues in honey and beverages, but there is still no limit standard for detecting plant toxins. Fortunately, the safety of honey and beverages has attracted public attention in recent years.

At present, many methods [9] [14] - [22] were used for the detection of plant toxins. Dawidowicz [14] described a method detecting thujone in alcoholic beverages using gas chromatography combined with solid phase microextraction. Bodi [15] developed liquid chromatography tandem mass spectrometry (LC-MS/MS) with C18 or SCX SPE clean-up detecting PAs in 274 tea and 87 honey samples, suggesting that PA in tea samples are most likely a contamination caused by co-harvesting of PA-producing plants. Hövelmann [21] identified N-caprylhistamine-β-glucoside from tomato and screened for imidazole alkaloids using LC-MS/MS. Lee [20] analyzed of six toxic alkaloids including methyllycaconitine, deltaline, lupanine, anagyrine 5, 6-dehydrolupanine and zygaxine from goats and cows by LC high-resolution MS (LC-HR-MS). In recent years, time-of-flight MS (TOF MS), ion-orbit MS and other HR MS have been applied in the field of hazard chemicals analysis with their high quality accuracy, high throughput and high scanning speed [19] [23]. In this study, UPLC-ion mobility-QTOF MS was used to establish for plant toxins database and an extraction and purification method for plant toxins in honey and beverage was optimization. The established method and database were used for screening the real samples.

2. Materials and Methods

2.1. Chemicals and Reagents

The plant toxin standards were from Enzo life science, Dr. Ehrenstorfer GmbH (Augsburg, Germany) and Sigma-Aldrich (St. Louis, MO, USA). The stock solutions were prepared by acetonitrile (ACN) with the concentration of 1000 μg/mL and stored at −20˚C. The 1 μg/mL working standard solutions were diluted by ACN/H2O (1/9, v/v). Honey and herbal beverage are commercially available. HPLC-MS-grade methanol, ACN and H2O were purchased from J. T. Baker (Deventer, The Netherlands) and Honeywell (Augsburg, Germany). Leucine enkephalin (LE) and formic acid (purity > 99%) were from Sigma-Aldrich (St. Louis, MO, USA). Sodium chloride (NaCl) and magnesium sulfate (MgSO4) of analytical purity were got from the Beijing Chemical Reagent Company (Beijing, China). The Discovery ®DSC-18, primary-secondary amine (PSA), graphitized carbon black (GCB), EMR, Zsep, Zsep+, and Zsep/C18 sorbent were from Supelco (Bellefonte, PA, USA). PSA 50 mg/MgSO4 150 mg was from Dikma. Solid phase extraction (SPE) prime HLB and lipid EMR were got from Waters Corp. and Agilent Corp., respectively.

2.2. UPLC-IM-QTOF MS Conditions

Waters ACQUITYTM UPLC-IM-QTOF MS (Vion, Milford, USA) was used in this experiment. Chromatographic separation was conducted on a BEH C18 column (2.1 mm × 100 mm; 1.7 μm; Waters) at 50˚C. The gradient conditions are shown in Table 1 with the phase A was 0.1% formic acid/ACN (V/V), and phase B was 0.1% formic acid/H2O (V/V). The flow rate was 0.45 mL/min, and

Table 1. The condition of liquid chromatography for plant toxins.

the injection volume was 5 μL.

A QTOF mass spectrometer (Waters Corp., Milford, MA, U.S.A.) equipped with an ESI source was used. The MS parameters were set as follows: capillary voltage, 1.0 kV; source temperature, 120˚C; desolvation temperature, 500˚C; desolvation gas rate, 800 L/h. The m/z range was 50 - 2000 Da. The acquisition mode was HDMSE with a low energy of 6 eV and elevated energy ramping from 15 to 45 eV to obtain the protonated or deprotonated molecule and the fragments ions for the compound in one injection. To ensure the accuracy of mass, the real-time calibration of LE (50 ng/mL, positive ion mode: 556.2771) was carried out. The plant toxins standard was acquired by UPLC-IM-TOF MS with ESI positive mode. The data was processed by UNIFI 1.7 software (Waters Corp.).

2.3. Sample Preparation

Approximately 1.0 g (±0.01 g) of the homogeneous herbal beverage or honey was weighed into a 10 mL polypropylene centrifuge tube. After the honey dissolved in 1 mL water, 5 mL of ACN was added to the herbal beverage or honey and the mixtures were vortexed for 1 min. Added 1.0 g NaCl, then the sample was vortexed for 1 min and centrifuged at 9000 rpm for 5 min. Later, the supernatant was transferred into another tube which containing 50 mg PSA, vortexed for 1 min and centrifuged at 9000 rpm for 5 min. The supernatant was evaporated to near dryness under a gentle stream of nitrogen at 50˚C, and the residues were dissolved with 0.5 mL ACN/H2O (5/95, v/v) before analyzed by UPLC-IM-QTOFMS.

3. Results and Discussion

3.1. Self-Build Database

There are many commercial libraries for searching the information of the chemical compounds, such as the toxin and the target (T3DB) [24], spectral library for organic compounds (SDBS), MassBank [25], NIST mass spectral library and chemistry library. As for searching plant toxins, little information was obtained from those above libraries. To date, there is no commercial library used on LC-MS system. The main drawback is considered to be the variability of the spectra obtained using LC-MS equipment from different manufacturers. The spectrums of LC-MS are greatly affected by ion source or transfer optics. Thus, UPLC-IM-QTOF MS database for plant toxins was established in our laboratory.

The plant toxin standard was acquired by UPLC-IM-TOF MS with ESI positive mode using 2.2 conditions. By entering the.mol file of the compound (loading from, UNIFI software processed the data. The retention time (RT), collision cross-section (CCS) and fragment ions generated from the compound were collected in our database. CCS is the characteristic of the compound. The RT changes with the change of different chromatographic conditions and matrix, but CCS and mass charge ratio parameters are not affected by the different instruments, which can ensure the accuracy of analyte identification and obtain the structural state information of samples. The literature [26] displayed compared with retention time (RT)+MS/MS, RT + MS/MS + CCS can reduce candidate substances 80% effectively when screening. Therefore, CCS can provide additional information to confirm the substances analyzed except RT and fragment ions in this study. In the database of 308 compounds for screening, the CCS results of these 308 compounds were confirmed. Through model test, the relationship between CCS and [M + H]+ was obtained, and the fitting degree R2 reaches 0.9287 (Figure 1) demonstrated that there is a linear relationship between CCS and molecular weight.

The detail information of the database was shown in TableS1. Take aconitine as an example, the chromatography spectrum of aconitine was shown in Figure 2(a). The RT (7.77 min) of aconitine was obtained. Figure 2(b) was the mass spectrum with a low and ramped high energy of collision energy, respectively. We obtained the m/z of the [M + H]+ of aconitine is 646.3211 and the fragment ions are 586.2999, 526.2799, 554.2748, 368.1829. The CCS for [M + H]+ of aconitine is 238.58 Å. The self-build database established in our laboratory contained over 300 compounds (TableS1) including aconitines, pyrrolidines, solanines, colchicine, tripterygium wilfordii toxins. Quality control (QC) and LE also were used to investigate the sensitivity and accuracy of the instrument in the whole experiment.

3.2. Sample Preparation Optimization

In order to extract more compounds from the matrix, the experiment selected 80 representative compounds (TableS2) to optimize the preparation method. The

Figure 1. The relationship of CCS and mass of the compound.

Figure 2. The information of aconitine after processed by software.

log P of these compounds ranged from −1.88 to 7.1.

As an easy-to-use sample preparation, QuEChERS is used to purify the compounds in this study. The original QuEChERS method was designed for samples with water content between 25% and 80%. The honey sample contained little water which leading to disperse difficultly. The experimental parameters need to be optimized.

3.2.1. Amount of Water Added to the Honey

The different amounts of water (0.5, 1, 2 and 3 mL) were estimated from honey spiked with 100 μg/kg of analytes. The results demonstrated that the responses of solasodine were relatively low when the honey was dissolved in 0.5 mL water, may be the honey is not completely dispersed. On the contrary, the responses of mesaconine and seneciphylline decreased when the water volume was larger than 1 mL, may be because these compounds were water-soluble. So this study chose to dissolve 1 g honey in 1 mL water.

3.2.2. Amount of NaCl

Anhydrous magnesium sulfate (MgSO4) and NaCl were used for salting out for determination of pyrrolizidine alkaloids in honey [27]. However, anhydrous MgSO4 can promote the distribution of analytes in the organic phase mainly by absorbing water, thus the responses for some water-soluble compound would lower their responses in this process. Saturated NaCl aqueous solution would increases the solubility of compounds in organic phase. To separate the ACN and water layer, the amount of NaCl salt (0.5, 1, 2, 3 g) was optimized. The responses of the analytes increased when NaCl was from 0.5 g to 1 g, but decreased when 3 g of the NaCl was added. To ensure sufficient ion intensity of NaCl in different matrices, while not reduce the extraction efficiency, 1 g of NaCl was chosen to extraction.

3.2.3. Cleanup

The experiment compared SPE with QuEChERS sorbent (C18, PSA, EMR, Zsep, and Zsep+) to purify the herbal beverage (shown in Figure 3(a)). It was demonstrated that PSA sorbent could give the better result than other sorbents, which may due to PSA remove organic acids, pigments and metal ions in beverages. Meanwhile, the amounts of PSA were optimized (Figure 3(b)). The result found that 50 mg PSA can remove interference for 1.0 g of herbal beverage and honey effectively.

3.3. Method Validation

3.3.1. Matrix Effect

The matrix effect (ME, %) was calculated as the ratio between the area of postextraction spike analytes and the area of standard solution under identical conditions and then multiplied by 100%. It is generally considered that the ME is

Figure 3. The statistics data of the plant toxins for different sorbents (a) and different amounts of PSA (b) in beverage.

ignored if the data is 80% - 120%. It is ME enhancement if the value is above 120% and inhibition if the value is lower than 80%. Compared the response of the standard in ACN/H2O and in blank sample processed by PSA, we found that 10% of the compounds showed ME enhancement, and 25% of the compound showed inhibition (Figure 4). To reduce the interference of the ME, matrix-matched calibration curves were used for quantitative analysis.

3.3.2. Recovery and LOQs

Processed the sample that cannot contain the plant toxins according to the optimized sample preparation to form the matrix solution, and diluted the standard working solution with the matrix solution step by step to obtain a series of matrix-calibration curve with concentrations of 20, 50, 200, 500, 1000 μg/kg. The correlation coefficients were greater than 0.99. The sample spiked with 25 μg/kg and 100 μg/kg concentration was purified to evaluate recovery and accuracy. Six spiked samples were processed for each concentration level, and the average recovery and relative standard deviation (RSD) were calculated. The recovery results displayed that over 80% and 90% of the compounds in beverages for 25 μg/kg and 100 μg/kg were ranged from 80% - 120%, and over 80% and 90% of the compounds in honey for 25 μg/kg and 100 μg/kg were ranged from 70% - 120%. The average recoveries for two matrices at two levels ranged from 60.6% - 120.1%, with relative standard deviation (n = 6) less than 25% which can meet screening requirement. The minimum concentration of the standard curve was diluted with the matrix extract until the signal-to-noise ratio of each drug was equal to 10 (S/N = 10), which was determined as the limit of quantification (LOQ) of the compound. The data was shown in TableS2.

3.4. Detection of the Real Sample

The plant toxins database and the method were applied in detecting 20 honey and 20 herbal beverages and honey flavored tea beverage samples. This study

Figure 4. The matrix effect of the compounds in beverage.

adopts the methodological evaluation criteria for confirmation of identity of plant toxins referred to the office of foods and veterinary medicine (OFVM) of FDA in 2015 including RT, mass extraction window and so on. For confirmation, at least two accurate mass (generally one precursor and one production ion) and mass deviation for precursor ion should be less than or equal to 5 ppm and production ions should be less than or equal to 10 ppm. The set for RT is ±0.5 min and CCS is 5%. The deconvolution technique was used to automatically identify the candidate substances of each peak in the real sample by comparison with the standard spectra in the self-built library, which is easily and quickly identified. The result showed that berberine was detected in one honey sample with content was 1 μg/kg. Caffeine was present in tea beverages. The concentration of caffeine was between 200 and 5500 μg/kg. Studies have shown that women planning to be pregnant and women during pregnancy should not consume more than 300 mg of caffeine per day. The amount for children intaking caffeine does not exceed 2.5 mg/kg body weight/day [1], so there is a certain risk for pregnant women and children who often drink caffeine-containing beverages.

4. Conclusion

The study established a plant toxins database and a method for rapid screening plant toxins in honey and herbal beverages. The sample was extracted by ACN, purified by PSA sorbent and analyzed by UPLC-IM-QTOFMS. The recovery, precision and detection limit of this method met the screening requirements. This method is easy to operate and high practical in screening of plant toxins in honey and herbal beverages.


This study was supported financially by grants from the financial support by the National Key Research and Development Program of China (2018YFC1602700) and National Natural Science Foundation of China (U1736201).

Supplemental Information

Table S1. The information of toxic compounds in database.

Table S2. The recoveries, LODs and LOQs for some plant toxins in honey and beverage.

Conflicts of Interest

The authors declare no conflicts of interest regarding the publication of this paper.


[1] Valduga, A.T., Gonçalves, I.L., Magri, E. and Delalibera Finzer, J.R. (2019) Chemistry, Pharmacology and New Trends in Traditional Functional and Medicinal Beverages. Food Research International, 120, 478-503.
[2] Chandrasekara, A. and Shahidi, F. (2018) Herbal Beverages: Bioactive Compounds and Their Role in Disease Risk Reduction—A Review. Journal of Traditional and Complementary Medicine, 8, 451-458.
[3] Suparmi, S., Ginting, A.J., Mariyam, S., Wesseling, S. and Rietjens, I.M.C.M. (2019) Levels of Methyleugenol and Eugenol in Instant Herbal Beverages Available on the Indonesian Market and Related Risk Assessment. Food and Chemical Toxicology, 125, 467-478.
[4] European Food Safety Authority (2013) Scientific Opinion on Tropane Alkaloids in Food and Feed. EFSA Journal, 11, Article 3386.
[5] Hijazin, T., Radwan, A., Lewerenz, L., Abouzeid, S. and Selmar, D. (2020) The Uptake of Alkaloids by Plants from the Soil Is Determined by Rhizosphere pH. Rhizosphere, 15, Article ID: 100234.
[6] Zhang, Q., Chen, X., Chen, S., Liu, Z., Wan, R. and Li, J. (2016) Fatal Honey Poisoning Caused by Tripterygium wilfordii Hook F in Southwest China: A Case Series. Wilderness & Environmental Medicine, 27, 271-273.
[7] Fields, B.A., Reeve, J., Bartholomaeus, A. and Mueller, U. (2014) Human Pharmacokinetic Study of Tutin in Honey; a Plant-Derived Neurotoxin. Food and Chemical Toxicology, 72, 234-241.
[8] Lamp, J., Knappstein, K., Walte, H.G., Krause, T., Steinberg, P. and Schwake-Anduschus, C. (2021). Transfer of Tropane Alkaloids (Atropine and Scopolamine) into the Milk of Subclinically Exposed Dairy Cows. Food Control, 126, Article ID: 108056.
[9] Chmit, M.S., Wahrig, B. and Beuerle, T. (2019) Quantitative and Qualitative Analysis of Pyrrolizidine Alkaloids in Liqueurs, Elixirs and Herbal Juices. Fitoterapia, 136, Article ID: 104172.
[10] Knutsen, H.K., Alexander, J., Barregård, L., Bignami, M., et al. (2017) Risks for Human Health Related to the Presence of Pyrrolizidine Alkaloids in Honey, Tea, Herbal Infusions and Food Supplements. EFSA Journal, 15, e04908.
[11] Van Schalkwyk, F.J., Stander, M.A., Nsizwane, M., Mathee, A. and Van Wyk, B.E. (2021) Fatal Pyrrolizidine Alkaloid Poisoning of Infants Caused by Adulterated Senecio coronatus. Forensic Science International, 320, Article ID: 110680.
[12] Cox, P. (2014) Botanical Neurotoxins.
[13] Da Motta, L.G., de Morais, J.A., Tavares, A.C.A.M., Vianna, L.M.S., et al. (2018) Maternal and Developmental Toxicity of the Hallucinogenic Plant-Based Beverage Ayahuasca in Rats. Reproductive Toxicology, 77, 143-153.
[14] Dawidowicz, A.L. and Dybowski, M.P. (2011) Fast Determination of α- and β- Thujone in Alcoholic Beverages Using Solid-Phase Extraction and Gas Chromatography. Food Control, 25, 197-201.
[15] Bodi, D., Roncaka, S., Gottschalk, C., et al. (2014) Determination of Pyrrolizidine Alkaloids in Tea, Herbal Drugs and Honey. Food Additives & Contaminants: Part A, 31, 1886-1895.
[16] Klein-Júnior, L.C., Heyden, Y.V. and Henriques, A.T. (2016) Enlarging the Bottleneck in the Analysis of Alkaloids: A Review on Sample Preparation in Herbal Matrices. TrAC Trends in Analytical Chemistry, 80, 66-82.
[17] Cirlini, M., Demuth, T.M., Biancardi, A., Rychlik, M., Dall’Asta, C. and Bruni, R. (2018) Are Tropane Alkaloids Present in Organic Foods? Detection of Scopolamine and Atropine in Organic Buckwheat (Fagopyron esculentum L.) Products by UHPLC—MS/MS. Food Chemistry, 239, 141-147.
[18] Volochanskyi, O., Švecová, M. and Prokopec, V. (2019) Detection and Identification of Medically Important Alkaloids Using the Surface-Enhanced Raman Scattering Spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 207, 143-149.
[19] Fu, Y., Zhao, C., Lu, X. and Xu, G. (2017) Nontargeted Screening of Chemical Contaminants and Illegal Additives in Food Based on Liquid Chromatography—High Resolution Mass Spectrometry. TrAC Trends in Analytical Chemistry, 96, 89-98.
[20] Lee, S.T., Welch, K.D., Stonecipher, C.A., Cook, D., Gardner, D.R. and Pfister, J.A. (2020) Analysis of Rumen Contents and Ocular Fluid for Toxic Alkaloids from Goats and Cows Dosed Larkspur (Delphinium barbeyi), Lupine (Lupinus leucophyllus), and Death Camas (Zigadenus paniculatus). Toxicon, 176, 21-29.
[21] Hövelmann, Y., Steinert, K., Hübner, F. and Humpf, H. (2020) Identification of a Novel N-Caprylhistamine-β-Glucoside from Tomato Fruits and LC-MS/MS-Based Food Screening for Imidazole Alkaloids. Food Chemistry, 312, Article ID: 126068.
[22] Wang, Z., Zheng, P., Wang, J., He, S., Ren, Z., Zhang, Y., Xiong, J. and Jiang H. (2021) Indirect Competitive Enzyme-Linked Immunosorbent Assay Based on a Broad-Spectrum Monoclonal Antibody for Tropane Alkaloids Detection in Pig Urine, Pork and Cereal Flours. Food Chemistry, 337, Article ID: 127617.
[23] Yusop, N.M., Hong, V.C., Phak, C.Z. and Naim, Z. (2020) CO2 Absorption Solvent Degradation Compound Identification Using Liquid Chromatography-Mass Spectrometry Quadrapole-Time of Flight (LCMSQTOF). Journal of Analytical Sciences, Methods and Instrumentation, 10, 78-95.
[24] Lim, E., Pon, A., Djoumbou, Y., Knox, C., Shrivastava, S., Guo, A.C., Neveu, V. and Wishart, D.S. (2010) T3DB: A Comprehensively Annotated Database of Common Toxins and their Targets. Nucleic Acids Research, 38, D781-D786.
[25] Horai, H., Arita, M., Kanaya, S., Nihei, Y., Ikeda, T., et al. (2010) MassBank: A Public Repository for Sharing Mass Spectral Data for Life Sciences. Journal of Mass Spectrometry, 45, 703-714.
[26] Zhou, Z., Luo, M., Chen, X., Yin, Y., Xiong, X., Wang, R. and Zhu, Z. (2020) Ion mobility Collision Cross-Section Atlas for Known and Unknown Metabolite Annotation in Untargeted Metabolomics. Nature Communications, 11, Article No. 4334.
[27] Martinello, M., Borin, A., Stella, R., Bovo, D., Biancotto, G., Gallina, A. and Mutinelli, F. (2017) Development and Validation of a QuEChERS Method Coupled to Liquid Chromatography and High Resolution Mass Spectrometry to Determine Pyrrolizidine and Tropane Alkaloids in Honey. Food Chemistry, 234, 295-302.

Copyright © 2024 by authors and Scientific Research Publishing Inc.

Creative Commons License

This work and the related PDF file are licensed under a Creative Commons Attribution 4.0 International License.