Tracking toxic hypoglycin A over two maturity stages of different positions in four Chinese litchi cultivars by UPLC-MS/MS

Background Litchi (Litchi chinensis Sonn.), a member of Sapindaceae family, is a common fruit with deliciously fragrant, sweet avors and high commercial value, besides edible arils, the pericarps and seeds of fruits and the leaves of litchi all exhibited meaningful bioactivities, while as the bibliography reported, hypoglycin A (HGA) as a toxic amino acid was naturally occurring in some members of Sapindaceae family. Methods In this work, an UPLC-MS/MS method was developed to quantied HGA in different parts of four litchi cultivars which at two maturity stages. Meanwhile, the methodological indicators of the established method were evaluated by selectivity, linearity, precious, accuracy, recovery and stability. Results The consequences expressed that the levels of HGA were highly associated with litchi cultivars and maturity stages. The positions of seeds and branches were the major source of HGA in the four litchi cultivars been detected, as for the edible arils, especially of ‘Fenghua’ and ‘Linglan’ were relative safety to be taken in. This developed method can provide scientic technical support for relevant research of content determination and this founding can offer scientic data for the further research of human health that was related to litchi cultivars.

All reagents used in the UPLC-MS/MS analysis were of HPLC grade. Methanol and formic acid were purchased from Macklin (Shanghai, China), ammonium acetate was bought from Kermel (Tianjing, China), HGA was obtained from TRC (Toronto, ON, Canada).

Standard preparation
Standard HGA was dissolved in methanol and prepared at a series concentrations of 10, 20, 50, 100, 200, 500 ng/mL.

Sample preparation
As shown in Fig.1, fresh fruits were divided into the parts of pericarps, edible arils and seeds and then all of these parts and leaves, branches were ground into powder by an YF-114B grinder (Yongli Pharmaceutical Machinery Co., Ltd, Zhejiang, China), these sample powder were stored at -18 °C. 1 g leaves, branches, pericarps and seeds powders were distributed in 15 mL centrifuge tubes and mixed with 5 mL methanol, respectively. Similarly, 1 g arils powders were suspended in 5 mL of methanol/water (10:90, v/v) mixture. Mulit reax oscillator (Heidolph Instruments GmbH & Co. KG, Schwabach, Germany) was used to extract HGA from samples at 1200 rpm for 3 h. After that, the extract solution was centrifuged at 10,000 rpm for 5 min by the using of H1850 centrifuge (Cence Laboratory Instrument Exploitation Co., Ltd, Hunan, China). In the end, the supernatant was collected and storage at -18 °C before the analysis of HPLC-MS/MS.

UPLC-MS/MS analysis
The qualitative and quantitative analyses of HGA were determined on a UPLC-MS/MS 8045 system (Shimadzu, Kyoto, Japan), which equipped with a triple quadrupole mass spectrometer using positive ionization mode and coupled with a UPLC equipped with a Shimadzu Shim-pack GIST-HP C18 column (2.1× 100 mm, 3 μm). Mobile phase A consisted of 0.1% formic acid and 10 mmoL/L ammonium acetate in HPLC-grade water, mobile phase B composed of HPLCgrade methanol. The ow rate was 0.3 mL/min and the elution gradient was programmed as follows: 0 min, 22% B; 1 min, 22% B; 3 min, 25% B; 6 min, 30% B; 8 min, 60% B; 13 min, 80% B; and nally, the initial conditions were held for 3 min to re-equilibrate the column. In addition, the instrument parameters were set as follows: sheath gas ow was 3 L/min, aux gas ow was 10 L/min and heated gas ow was 10 L/min. The temperature of ion transfer tube, DL and heating block were 300, 250 and 400 °C, respectively. Multiple-reaction monitoring (MRM) was used to obtain the quantitative and qualitative data. The quantitation ion of HGA was based on the transition m/z 142.10 → 74.05, with regard to the conformation ion, was on the grounds of the transition m/z 142.10 → 96.05. The collision energies (CE) of HGA were11.0 V.

Method validation
Linearity range, selectivity, sensitivity, precision, accuracy, stability and recovery were used to assess the validation of the method we used [20].
A series of standard solutions were dissolved in methanol at 10, 20, 50, 100, 200, 500 ng/mL for HGA. The standard curve was built with the mass concentration as the abscissa and the peak areas of quanti cation ion as the ordinate. Meanwhile, the correlation coe cient (R 2 ) was utilized to response the tting degree of curvilinear regression equation [21]. As for selectivity, it was investigated by comparing the retention time of HGA in the samples. In addition, the sensitivity of this method was characterized by two parameters, limits of detections (LOD) and limits of quanti cations (LOQ). LOD was assessed as the concentration of analyte that with signal-to-noise (S/N) ratio of 3:1, simultaneously, LOQ with S/N ratio of 10:1 [22].
The values of precision and accuracy were determined through the calculation of percent relative error (%RE) and percent relative standard deviation (%RSD) which with 6 separate measurements for three calibrators [16]. Furthermore, the stability of HGA was evaluated by three QC samples which were stood for 2 h, 4 h, 8 h, 12 h, 24 h at 4 °C on the same day (intra-day) and stored for 1, 3, 7, 15 and 30 days (inter-day) at -20 °C [23].
Extraction recovery was con rmed by comparing the average HGA concentrations of samples which were chosen from four litchi species data randomly with the average HGA concentrations of samples that were prepared by removing half volume of chosen litchi samples and then spiked with corresponding volumes of validation solutions representing 100% recovery [24]. The samples chosen from four litchi species were the branches of 'Fengchuiliao', the maturity pericarps of 'Fenghua', the immaturity seeds of 'Gualv' and the immaturity seeds of 'Linglan', respectively, and six replicates were carried at different samples.

Statistical analysis
The statistical analysis was calculated by SPSS version 20.0 in this work. ANOVA analysis and Independent-Samples T Test were used. Data was expressed as means ± standard deviation of triplicate detections. The determined data were owned signi cant differences. The p-value < 0.05 represented that the results were signi cant, and p-value < 0.01 on behalf of the results were very signi cant.

Detection and separation
Compared with the analytic system gas chromatography-mass spectrometry (GC-MS) which samples needed to be derivatized [25,26], in this work, samples were prepared by brie y steps of ultrasonic extraction and centrifugation without any chemical derivatization. As for the detection, it was performed by UPLC-MS/MS (Fig. 2), at a CE of 11.0 V, the precursor ion of HGA at m/z 142.10 in the positive mode was switched into quantitation ion at m/z 74.05 and conformation ion at m/z 96.05.
Selectivity was a parameter been set to con rm the method speci c to the detection of the unknown compounds. It was able to be assessed through comparing the retention time between standards and samples [27]. As demonstrated in Fig. 3, the retention time of HGA in the standards and determinands were both around 1.3 min In addition, throughout all the analysis process, the retention time without being observed a signi cant offset, which hinted that this method was tted to the analysis of HGA with great selectivity.
The linearity of HGA was evaluated by standard substance being dissolved into methanol in the range of 10-500 μg/L. The standard curve was established as with R 2 = 0.9992. The correlation coe cient (R 2 >0.99) represented that in the concentration range between 10 μg/L and 500 μg/L, the linearity was ideal [28].
Sensitivity was estimated by LOD and LOQ, the amount of LOD was three times as much as the peak height of baseline background, while for LOQ, it was ten times. LOD and LOQ can also be calculated through the slope of linear equation and the standard deviation of noise of the baseline background, the equation as follow [29].
See equation 1 in the supplementary les.
where σ represents the standard deviation of the response, and a on behalf of the slope of the standard curve.
According to the concentration observed at an S/N greater than three, the LOD value of the method was 6 μg/L or 0.03 μg/g when converted unit, meanwhile, the LOQ value of the method was 18 μg/L (0.10 μg/g). As Fig.3 shown, the peak signal intensity of the matrix blank was almost 15-fold less than that of the lowest standard point, which implied that the value of LOD of this method was rather low and this method was appropriate for the analysis of HGA in the litchi.

Precision, accuracy and recovery
As Table 1 demonstrated, three validation samples with the concentrations of 10 μg/L, 150 μg/L and 3500 μg/L respectively, were detected six times to determine the precision (%RSD) and accuracy (%RE) values of this method for HGA. The results shown that the %REs of validation samples were -0.17%, 0.02% and 0.14%, respectively, with corresponding %RSDs were 1.37%, 1.32% and 1.44%. Meanwhile, the mean of these concentrations were not exceeding 15% of the nominal values and the precision were within 15% [30].
According to the results of HGA contents in the litchi parts which were chosen to evaluate the extraction recovery, the concentrations of analyte solutions added to the sample solutions were as followed: 500 μg/L, 40 μg/L, 2000 μg/L and 4000 μg/L standard solutions were mixed with the sample solutions of 'Fengchuiliao', 'Fenghua', 'Gualv' and 'Linglan' chosen above in the equivalent volume, respectively. On the basis of the request, the allowable bias range of the extraction recovery is from 85% to 115%, and the RSD of extraction recovery should within 15% [31]. As Table 2 exhibited, 104.16%, 106.64%, 98.45% and 101.60% corresponding to the recovery of 'Fengchuiliao', 'Fenghua', 'Gualv' and 'Linglan' and all of these data were in the range between 85% to 115%, simultaneously, the corresponding %RSD were 2.18%, 3.49%, 3.03% and 2.98% which were all less than 15%.

Stability
HGA was proven to be stable in QC samples under different storage conditions at low, medium and high concentration levels. As shown in Table 3, the relative error was less than 3.07% for low, medium and high QC levels whether stored intra-day or inter-day, namely the bias of HGA concentrations in QC samples were within 15% of the theoretical concentrations [32].

Detection of HGA in four litchi cultivars
The HPLC-MS/MS method was applied to quantify the HGA contents of four litchi species 'Linglan', 'Fengchuiliao', 'Fenghua' and 'Gualv' at two degree of maturity in different parts including pericarps, arils, seeds, leaves and branches. The consequences were exhibited in Table 4. At the cultivar level, 'Fengchuiliao' showed the highest HGA concentrations in immature pericarps, immature arils, immature seeds and leaves. In addition, the mature arils and mature seeds also expressed rather high contents of HGA. As for the HGA level of 'Fenghua', the mature seeds and branches parts hinted maximum, it is notable that the branches of 'Fenghua' displayed the maximal HGA concentration among these data which was up to 114.165 ± 1.456 μg/g, while it was reported that the toxic dose of HGA for male and female rats was 231.19 ± 62.55 mg/kg BW [33]. Converting to a 50 kg adult, the toxic dose was as high as 101.252 kg which suggested that all the positions of these four litchi cultivars were relatively safe. The remaining positions of litchi, mature pericarps and mature arils, were revealed highest HGA concentrations in 'Gualv'. With regard to 'Linglan', the HGA contents in any part were lower than that of other species.
Apart from the immature seeds of 'Linglan' was 15.092 ± 0.217 μg/g, the rest parts were less than 1.000 μg/g, that implied that in comparison with other cultivars, 'Linglan' expressed the less toxicity. Moreover, ANOVA analysis was employed to found the signi cant differences of various litchi cultivars. As Table   4 displayed, either in variety of parts or in different maturity stages, the concentrations of HGA were exhibited high correlation with litchi cultivars. From the perspective of position level, Independent-Samples T Test was applied to ensure whether the HGA concentrations of two maturity stages in various positions were signi cant or not, the consequences shown in Table 4 with p-value < 0.01, represented that the HGA concentrations in different maturity stages exist quite great differences. The consequences exhibited that the maximum HGA contents of mature and immature pericarps, mature and immature arils, mature and immature seeds, leaves and branches were 1.653 ± 0.036 μg/g, 0.869 ± 0.012 μg/g, 16.076 ± 0.161 μg/g, 25.231 ± 0.247 μg/g, 54.325 ± 0.747 μg/g, 35.140 ± 0.421 μg/g, 1.532 ± 0.024 μg/g and 114.165 ± 1.456 μg/g, severally.
Litchi (Litchi chinensis Sonn.) as a member of the Sapindaceae family is widely cultivated in Asia, East Africa, Oceania, some Paci c Islands and so forth [34]. As the literature reported [4,5,9], not only the litchi owns edible arils, but also possesses the bioactive compounds in its pericarp and seed of fruits, leaves and branches. However, just like the other species in this family, it caused numerous public health issues in various counties, such as India, Vietnam, Bangladesh, and so on [14]. Actually to date [10], HGA has been deemed to relate to the deaths caused by Sapindaceae family. Results of this study con rm that HGA exists in the pericarp, seed and aril of fruits, leaves and branches.
HGA as a speci c inhibitor of isovaleryl CoA dehydrogenase and α-methylbutyrate dehydrogenase has been identi ed as the cause of atypical myopathy in horse and Père David's deer [37,38], and Jamaican vomiting sickness, viral encephalitis and neurologic illness in human [39,40]. In the cases of these diseases, extreme hypoglycemia was observed [10]. Our work implied that seeds and branches were the major source of HGA in the four litchi cultivars been detected, while the edible arils, especially of 'Fenghua' and 'Linglan' were relative safety to be taken in.

Conclusions
In this work, a simple and straightforward sample preparation method, using the UPLC-MS/MS, was established to detect the concentrations of HGA in different positions (Pericarps, arils, seeds, leaves and branches) of four litchi cultivars, Linglan', 'Fengchuiliao', 'Fenghua' and 'gualv', and to monitor HGA concentrations in different maturity stages of litchi fruits. According to the consequences of methodological indicators, the established method expressed great selectivity, linearity, precious, accuracy, recovery and stability, represented that this method is suitable for the analysis of HGA in litchi. As a result, the levels of HGA were highly associated with litchi cultivars and maturity stages. In addition, the branches of 'Fenghua' displayed the maximal HGA concentration which was up to 114.165 ± 1.456 µg/g, by the way, the positions of seeds and branches were the major source of HGA in the four litchi cultivars been detected, as for the edible arils, especially of 'Fenghua' and 'Linglan' were relative safety to be taken in. Actually, this developed method can provide scienti c technical support for relevant research of content determination and can offer scienti c data for the further research of human health that was related to litchi cultivars.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
All authors have read and agreed to the published version of this manscript.

Availability of data and materials
The datasets used during this work are available from the corresponding author upon reasonable request.

Competing interests
The authors declare no con ict of interest.  Tables   Table 1 Accuracy (%RE) and precision (%RSD) data (n=6) for HGA in validation samples.  Table 4 The UPLC-MS/MS method was applied to detect HGA contents in different parts (pericarp, pulp, seed, leaf and branch) of four litchi cultivars at two maturity stages, ANOVA analysis and Independent-Samples T Test were used to found the significant differences of HGA concentrations within various cultivars and between different maturity stages, respectively.  Figure 1 Pericarps, arils, seeds, leaves and branches which were divided from four different Litchi cultivars in two maturity stages.