Development of A LC-MS/MS Method for Simultaneously Determining Nucleosides and Methy-Nucleosides in Mice Liver mRNA of Epimedin C-Induced Liver Injury Model

Epimedium, the crude drug used in clinical, has been proved to have the potential to cause liver damage in patients. As one of the main active ingredient of Epimedium, Epimedin C is reported to have the potential hepatotoxicity. However, the mechanism of Epimedin C-induced mice liver injury has not been studied. mRNA methylation is implicated in the regulation of many biological processes and diseases. The study of mRNA methylation in Epimedin C-induced liver injury may contribute to clarify the liver toxicity mechanism of Epimedin C. Therefore, an analysis method needs to be established to determine liver mRNA nucleoside and methy-nucleoside levels in epigenetic studies. and vacuolar degeneration in liver, indicated the liver injury in mice. This method was successfully applied to the detection of six liver mRNA nucleosides levels in Epimedin C-induced liver injury with good precision and accuracy. The results indicated that mRNA methylation might be associated with Epimedin C-induced liver injury. This study offers a method for the research on the mechanism of Epimedin C-induced liver injury, and will facilitate the research on the hepatotoxicity of herbal medicine in epigenetic studies. and sensitive method was developed using high performance liquid chromatography-tandem mass spectrometry to simultaneous determination of six nucleosides (adenosine, uridine, cytidine, guanosine, m6A and m5C) in mice liver mRNA in Epimedin C-induced liver injury model. It might provide new idea for further research on the mechanism of herbal medicine-induced liver injury.


Conclusion
This method was successfully applied to the detection of six liver mRNA nucleosides levels in Epimedin C-induced liver injury with good precision and accuracy. The results indicated that mRNA methylation might be associated with Epimedin Cinduced liver injury. This study offers a method for the research on the mechanism of Epimedin C-induced liver injury, and will facilitate the research on the hepatotoxicity of herbal medicine in epigenetic studies.

Background
In recent years, the incidence of drug-induced liver injury has been increasing year by year due to the surging demand for herbal medicine and health care products [1]. Epimedium, the crude drug used in clinical, is the dry leaf of Epimedium brevicornum Maxim., Epimedium sagittatum (Sieb.et Zucc.) Maxim., Epimedium pubescens Maxim., and Epimedium koreanum Nakai. Epimedium preparation is widely used for rheumatism, arthritis, osteoporosis and other diseases in China [2][3]. However, herbal preparations containing Epimedium have been reported to have the potential to cause liver damage in patients [4][5]. As one of the main active ingredient and quality indicator of Epimedium, Epimedin C is metabolized to 2''-O-rhamnosylicariside in the body, which has strong cytotoxicity to HL-7702 and HepG2 cells [6][7][8]. Despite some reports to the potential hepatotoxicity, the mechanism of Epimedin C-induced liver injury has not been studied.
Methylation modi cation of messenger RNAs (mRNA), as an important RNA modi cation in mRNA, has been widely proved to effect on a variety of biological processes and related diseases, including metabolism, circadian rhythm, immune response, gametogenesis, neurologic development, and cancer [9][10][11]. At present, the methylation modi cations of mRNA have been found mainly include N6-methyladenosine (m6A), N5-methylcytidine (m5C), N1-methyladenosine (m1A), N7methylguanosine (m7G), and so on. A large number of studies have shown that methylation modi cation disorders of mRNA are closely related to the occurrence and development of a variety of diseases. Studies indicated that m6A is associated with the progression and metastasis of hepatocellular carcinoma [12][13]. Alteration of m5C in Neurons may lead to cell stress response and apoptosis [14]. The level of m5C modi cation is closely related to PI3K/Akt signaling pathway in gastrointestinal malignancies [15]. However, the relationship between mRNA methylation and Epimedin C-induced liver injury is still unknown. m6A and m5C are the dominant modi cations in mRNA methylation in vivo. The study of m6A and m5C in Epimedin C-induced liver injury may contribute to clarify the liver toxicity mechanism of Epimedin C. In consequence, the content of mRNA methylation in Epimedin C-induced liver injury should be quanti ed.
Although some studies described the detection method of m6A and m5C in total RNA and DNA, the method for the simultaneous quanti cation of six nucleosides (adenosine, uridine, cytidine, guanosine, m6A and m5C) in mRNA was less reported [16][17][18][19][20][21][22]. Because of the fewer levels of m6A and m5C in mRNA than total RNA, new method should be established to enhance the detection sensitivity. A reliable method to simultaneous determination of m6A and m5C could contribute to the epigenetic research of hepatotoxicity or other complex diseases.
In this study, a selective, rapid and sensitive method was developed using high performance liquid chromatography-tandem mass spectrometry to simultaneous determination of six nucleosides (adenosine, uridine, cytidine, guanosine, m6A and m5C) in mice liver mRNA in Epimedin C-induced liver injury model. It might provide new idea for further research on the mechanism of herbal medicine-induced liver injury.

Animals
Male Balb/c mice (6-8 weeks old and weighing 18.0-22.0 g) were obtained from Beijing HFK bioscience Co., Ltd. (Beijing, China). Mice were kept in cages under controlled conditions of 22 ± 0.5 • C, 50 ± 2.0% RH and maintained with free access to standard laboratory food and water for one week before experiments.

Establishment of the Epimedin C-induced hepatitis model and experimental groups
Animals were randomly divided into three groups (n = 7 each): the normal control group, the Epimedin C (10 mg/kg) group, and the Epimedin C (40 mg/kg) group. Epimedin C were completely dissolved in 0.9% saline before passing through a 0.22 µm cell strainer. The Epimedin C groups were intragastrically administered single doses of 10 and 40 mg/kg body weight of Epimedin C per day, and the normal control group were given the same volume of saline. After four weeks of intragastric administration, all mice were sacri ced, and blood samples and liver tissues were collected. The procedures for the present study were approved by the Guide for the Care set by the National Institutes of Health.

Assessment of liver injury
Serum alanine transaminase (ALT) and aspartate transaminase (AST) contents in serum were analyzed using colorimetric tests (Nanjing Jiancheng Bioengineering Research Institute, Nanjing, China). Liver tissues xed in 4% paraformaldehyde were embedded in para n using a tissue procedure, and 4-µm-thick slices were cut and stained with hematoxylin and eosin (H&E) reagent. Photomicrographs were observed with a light microscope to evaluate liver injury.

LC-MS/MS instruments and conditions
The LC was performed using an ExionLCTM analytical (UPLC) system (AB Sciex, USA). Chromatographic separation was carried out on a Kinetex® 2.6 µm Polar C18 100A LC column (100 mm × 2.1 mm i.d.). The ow rate was 0.3 mL/min. The mobile phase included ultra-puri ed water containing 0.1% formic acid (solvent A) and methanol (solvent B) in a linear gradient. The gradient program was as follows: 0 to 0.5 min, 95% A; 0.5 to 3 min, 95 to 30% A; 3 to 4 min, 30% A; 4 to 4.1 min, 30 to 95% A; 4.1 to 6 min, 95% A. The injection volume was 10 µL and the total run time was 8 min. The temperature of the autosampler was set at 4 • C, and the column temperature was maintained at 40 • C. MS/MS analysis was carried out on a Qtrap 6500 mass spectrometer (AB Sciex, Redwood City, CA, USA) equipped with Turbo Ionspray interface operating in positive ESI mode. The instrument was operated with an ion spray voltage of 4.5 kV and a heater gas temperature of 500 • C.
Mass-dependent parameters (declustering potential, entrance potential, collision energy, and collision cell exit potential) were set to the optimal values obtained by automated optimization. Data acquisition was achieved by multiple reaction monitoring (MRM). The precursor-product ion pair and the optimal values of mass parameters are listed in Table 1. Positive ion mode was used and the dwell time was set at 100 ms. Data acquisition was generated and processed using the Analyst 1.6.2 software (AB Sciex). Table 1 Multiple reaction monitoring transitions and optimized mass parameters for the analytes

Analytes
Precursor ion Stock solutions for calibration and quality control (QC) were accurately weighed and dissolved in dimethylsulfoxide (2% of the total volume) before adding an appropriate volume of methanol to nal concentration of 1 mg/mL. Working solutions were prepared by serially diluting the stock solutions with water, and then the corresponding working solutions were mixed to prepare mixed working solutions with concentration in the ranges of 160-80000 pg/mL for A, C, m5C and m6A, 0.8-400 ng/mL for G, and 8-4000 ng/mL for U. The stock solution (1 mg/mL) of the IS was dissolved in water to 4 ng/mL containing 0.4% formic acid. All solutions were kept at -20 • C and brought to room temperature before use. The calibration standards were prepared by spiking 10 μL of the corresponding working solutions mentioned above into 30 μL of mixtures of nuclease P1 (0.1 U) and alkaline phosphatase (2 U) to yield concentrations of 40, 120, 500, 1000, 2000, 4000, 8000 and 20000 pg/mL for A, C, m5C and m6A, 0.2, 0.6, 2.5, 5, 10, 20, 40 and 100 ng/mL for G , and 2, 6, 25, 50, 100, 200 400 and 1000 ng/mL for U.
The QC samples were prepared in the same way as the calibration samples at three concentrations 120, 1600 and 16000 pg/mL for A, C, m5C and m6A, 0.6, 8, 80 ng/mL for G and 6, 80 and 800 ng/mL for U.
RNA isolation from liver tissues and Enzymatic digestion of the mRNA 100 mg of liver tissue was completely disrupted and homogenized into 1 mL TRIzol reagent. Then, the total RNA of liver tissue was isolated according to the manufacturer's instructions. After analyzed by a NanoDrop One (Thermo Scienti c), the Dynabeads ® mRNA Puri cation Kit (Ambion) was used to enrich mRNA. The contaminant DNA was removed from the mRNA samples by using DNase, and the concentration was analyzed by a Qubit 3.0 Fluorometer (Invitrogen).
Referring to the methods reported, enzymatic digestion of the mRNA was performed

Method validation
The analytical methodology was under the guidelines set by the United States Food and Drug Administration [23] and the Chinese Pharmacopoeia Commission [24].
The selectivity was evaluated by comparing chromatograms of mRNA-free blank enzymolysis matrix, blank enzymolysis matrix with all analytes, and a liver mRNA enzymolysis sample containing IS from a mouse after Epimedin C treatment. The absence of peaks at retention times of seven analytes indicated no interference in the test samples.
After detection of the upper limit of the quanti cation (ULOQ) samples, blank samples were injected to evaluate the carryover effects. The response peak of any analytes must be < 10% of the lower limit of the quanti cation (LLOQ) samples.
The linearity was investigated by plotting the peak-area ratios of the analytes (A, U, C, G, m6A and m5C) to the IS versus the concentrations of the calibration standards. The calibration equations were tted using a weighed least-squares linear regression analysis (weighing factor of 1/x 2 ). The accuracy, expressed as the mean relative error (RE, %), should be ≤ 20% for LLOQ and ≤ 15% for the other seven concentrations of the calibration standards.
To assess the precision and accuracy of the method, ve replicates of QC (at three concentration levels) and LLOQ were prepared and analyzed within three validation days. Both the accuracy (RE, %) and intra-and inter-precision (RSD, %) for LLOQ should be less than or equal to 20%. The accuracy and precision for the QC levels should be within ± 15%.
To determine whether matrix components affected the ion suppression or enhancement in the method, the matrix effect (ME) was assessed by comparing the corresponding peak area responses of enzymolysis matrix with all analytes and the blank samples in which the enzymolysis matrix was replaced with water. In this method, the variability values of the MEs (RSD, %) should be less than 15%.
To evaluate the stability of the analytes in the enzymolysis matrix during sample preparation and storage, the low and high QC concentration levels in different storage conditions were detected. The storage conditions included room temperature for Page 6/16 6 h, three freeze-thaw cycles, autosampler at ambient temperature for 20 h, and freezing at -20 • C for 30 days. The analytes were stable when 85-115% of the initial concentration was retained.
The dilution integrity was assessed by testing the solution which was diluted 100-fold with blank enzymolysis matrix before ultra-ltration from highly concentrated samples above the upper limit of standard curves. The accuracy and precision should be within ± 15%.

Data analysis
The data were presented as the arithmetic mean ± SD. Statistical analyses were performed using SPSS software for Windows. Statistical signi cance was assessed by unpaired two-tailed Student's t-test between two samples. p value < 0.05 was considered statistically signi cant.

Results
Chromatography and mass spectrometry

Accuracy and precision
Three batches of LLOQ and QC samples were evaluated to obtain the intra-and interday precision and accuracy with the current method. The validation results of accuracy and precision for A, U, C, G, m6A and m5C are listed in Table 2, which demonstrated that the values of RSD and RE were all inside the acceptable variability limits. It indicates that the method is accurate and precise. and 98.5 ± 3.7% for m6A, and 100.5 ± 5.0 and 95.0 ± 3.9% for m5C, respectively. These data indicated that the MEs for all the analytes were negligible following the current method.
The stability results of the analytes are summarized in Table 3. A, U, C, G, m6A and m5C remained at room temperature for 6 h and after three freeze-thaw cycles. All the analytes were also stable in the autosampler at ambient temperature for 20 h and after freezing at -20 • C for 30 days.  Table 4 shows the results of the dilution integrity. The precision and accuracy of the dilution test at the low and high concentration levels were within the acceptable criteria, indicated that A, U, C, G, m6A and m5C were assayed reliably by diluting 100-fold with blank enzymolysis matrix. Samples could be tested by dilution when the analyte concentration exceeded the linear range of the standard curve. 16540.0 ± 320.9 1.9 3.4

Epimedin C induced liver injury
Hematoxylin and eosin staining of the liver was performed in order to evaluate the pathological changes in liver. Epimedin C challenge increased the in ammatory cell in ltration (the red arrow) and vacuolar degeneration (the black arrow) observed in liver, and these effects were induced with the increased dose of Epimedin C (Fig. 3). Serum aminotransferase levels were in good agreement with histopathological changes in our study. The serum levels of transaminase were measured to evaluate hepatocellular damage. The results showed that serum ALT and AST levels increased after intravenous Epimedin C administration compared with the normal control group (Fig. 3).
Application for quantifying the nucleosides The mean concentration levels of m6A and m5C in prepared test samples of mice liver mRNA in the control and Epimedin Cinduced liver injury model groups are presented in Fig. 4. The ratio of m6A to A (m6A/A% = C m6A /C A × %) and m5C to C (m5C/C% = C m5C /C C × %) expressed the content of modi ed nucleoside in mouse liver mRNA. The results showed that compared with the normal control group, the concentration levels of m6A and m5C signi cantly increased, respectively.
Furthermore, the ratio of m6A/A and m5C/C in the mice liver mRNA in the Epimedin C groups were higher than that in the control group. According to these results, it could be indicated that epigenetic modi cation may change may in mice liver after Epimedin C treatment.

Discussion
As the widely use in China, Epimedium has been frequently reported to cause liver injury [4][5]. Epimedin C is one of the main active ingredient of Epimedium. Some studies have indicated the potential hepatotoxicity[6-8], but the mechanism of Epimedin C-induced liver damage has not been studied. To the best of our knowledge, methylation modi cation disorders of mRNA are closely related to the occurrence and development of many liver diseases. The study of mRNA methylation in Epimedin C-induced liver injury may contribute to clarify the liver toxicity mechanism of Epimedin C, and provide guidance for Chinese herbal medicine in clinical practice. For this reason, it is important to precisely quantify the content of mRNA methylation in Epimedin C-induced liver injury.
Because of the high selectivity and sensitivity, high performance liquid chromatography-tandem mass spectrometry is one of the best means for the quanti cation of bio-samples at present [18]. Some methods have been utilized to detected m6A or m5C in DNA and RNA samples, but it is lack of method for the simultaneous determination of six nucleosides to evaluate the content of modi ed nucleosides using the ratio of m6A to A and m5C to C. Combined with the change of m6A and m5C levels in samples, the ratio of m6A to A and m5C to C can better indicate the level of mRNA methylation in liver injury model.
It is known that mRNA is unstable and easily decomposed. Consequently, an puri cation and enzymatic digestion method was performed before detection by LC-MS/MS. This present method was successfully applied to the determination of six nucleoside levels in mice liver mRNA in Epimedin Cinduced liver injury model. Due to consisting of multiple compounds in mice plasma, the simultaneous quantitation of six nucleoside encounters the great challenge in short running time using UPLC. Multiple reaction monitoring, a highly speci c technique, can be used for quantifying the targeted analyte without the considering of baseline chromatographic separation.
The chromatographic conditions were optimized for the high sensitivity and fast separation to avoid the low MS response caused by the potential mutual ionization suppression in ESI [25]. A Kinetex® 2.6 µm Polar C18 100A LC column was proved to be more suitable for the separation of the targeted compounds in the sample. And as results, ultra-puri ed water containing 0.1% formic acid / methanol in a linear gradient was the optimum mobile phase to achieve the chromatogram.
In this study, a selective, rapid and sensitive method was established using high performance liquid chromatography-tandem mass spectrometry to simultaneous determination of six nucleosides in mRNA. The precision and accuracy of this method were adequately validated under the guidelines for bioanalytical method validation. The differences of mass spectrometry response values between these six nucleosides led to the different ranges of standard curve. Furthermore, because of the much higher concentrations of A, U, C and G than that of m6A and m5C, the highly sensitive detector is prone to saturation effect when the concentrations of m6A and m5C are within the standard curve range but the remaining part is at high concentration. Therefore, an alternative solution is diluting the concentration when detecting A, U, C and G.
Our study showed the changes of m6A and m5C in mice liver mRNA after Epimedin C treatment, but the relationship between modi ed nucleosides and the mechanism of Epimedin C-induced liver injury is not well understood, which requires further investigation. This study may offer a new idea and approach for studying the mechanism of Epimedin C-induced liver injury.

Conclusions
This analysis method was successfully applied to the detection of six liver mRNA nucleosides levels in Epimedin C-induced liver injury. The method has quali ed precision and accuracy with the guidelines for bio-analytical method validation. The results showed that Epimedin C led to liver injury, and epigenetic modi cation changed in mice liver after Epimedin C treatment. Therefore, it is suggested that the modi ed nucleoside m6A and m5C might be associated with liver injury. It might contribute to prevent the hepatotoxicity of Epimedium in clinical treatment and provide new idea for further research on the mechanism of herbal medicine-induced liver injury.    Representative photomicrographs (H&E stain 100× magni cation) of liver tissues (A) and serum aminotransferase levels (B) after Epimedin C challenge. Epimedin C was absent in the normal control group. The mice were treated with Epimedin C (10 mg/kg) in the EL group. The mice were treated with Epimedin C (40 mg/kg) in the EH group. *P <0.05 vs. the normal control group. The black arrow represents the vacuolar degeneration and the red arrow stands for the in ammatory cell in ltration.

Figure 4
The concentration levels of six nucleosides and the relative contents of m6A and m5C in test samples of the mice liver mRNA in the control and Epimedin C groups. (A) The concentration levels of A, U, C, G, m6A and m5C in test samples; (B) the content of modi ed nucleoside m6A (ratio of m6A/A) and m5C (ratio of m5C/C). Epimedin C was absent in the normal control group. The mice were treated with Epimedin C (10 mg/kg) in the EL group. The mice were treated with Epimedin C (40 mg/kg) in the EH group. *P <0.05 and **P <0.01 vs. the normal control group.