Fifteen batches of Asarum heterotropoides F. Schmidt (Xixin) samples were purchased from different districts of the Chinese mainland with batch numbers XX01–XX15. All were authenticated as the roots and rhizomes of Asarum heterotropoides F. Schmidt (Xixin) by Prof. Jinda Hao, Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences. Traditional Chinese patent medicines containing herbs from the Aristolochiaceae family were collected from online and offline pharmacies.

The AA I reference standard (purity >98%) was purchased from the National Institutes of Food and Drug Control (Beijing, China). AA II, AA IIIa, AA IVa, and AL I (purities >98%) reference standards were obtained from Chengdu Push Bio-technology Co., Ltd. (Chengdu, China). Methanol, acetonitrile, formic acid, and ammonium acetate (HPLC and MS/MS grade) were purchased from Fisher Scientific (MA, United States). Aspartate amino transferase (AST), alanine amino transferase (ALT), alkaline phosphatase activities (ALP), creatinine (CREA), blood urea nitrogen (BUN), total bile acid (TBA), and total bilirubin (TBIL) were purchased from Hua Sin Science Co., Ltd (Guangzhou, China).

The Agilent Eclipse XDB C18 column (4.6 × 150 mm, 5 μm) was used for chromatographic separations of the analogues of AA in Asarum heterotropoides F. Schmidt (Xixin). The mobile phases were acetonitrile containing 0.1% formic acid (A) and water containing 0.1% formic acid (B). Separations were performed as follows: 0–5 min, 30–30% A; 5–5.01 min, 30–36% A; 5.01–26 min, 36–43% A; 26–28 min, 43–43% A; 28–28.01 min, 43–30% A; 28.01–30 min, 30–30% A. The flow rate was 1.0 ml/min. AA analogues were detected at a wavelength of 254 nm. The column temperature was maintained at 20–23°C. Samples (10 μl) were injected into the HPLC system.

Quantitative determination of analogues of AA in traditional Chinese patent medicines was performed with a UPLC-MS/MS system consisting of an ACQUITY UPLC I-Class coupled to a Xevo-TQS detector (Waters, United States). Chromatographic separation was achieved using an ACQUITY UPLC BEH C18 column (2.1 × 50 mm, 1.7 μm; Waters, United States) with a flow rate of 0.3 ml/min at 35°C. Separation was obtained with the following gradient: 0–1.0 min, 10% B; 1.0–7.0 min, 10–75% B; 7.0–7.2 min, 75–95% B; 7.2–10.2 min, 95% B; 10.2–10.3 min, 95–10% B, 10.3–12.0 min, 10% B (A: water with 0.01% formic acid, B: methanol with 0.01% formic acid). Both mobile phases were added with 5 mM ammonium acetate and treated under ultrasonication for 5 min before usage. The injection volume was 1.0 μl.

Multiple reaction monitoring mode (MRM) and positive electrospray ionization were applied for the quantitative determination. The instrument parameters were as follows: source temperature, 150°C; capillary voltage, 3.5 kV; desolvation gas flow, 800 L/h; cone gas flow, 150 L/h; nebulizer gas, 7.0 bar; desolvation temperature, 550°C. Details of the parent/daughter ions, cone voltages, and collision energies for AA I, AA II, AA IIIa, AA IVa, and AL I are listed in Supplementary Table S2.

Asarum heterotropoides F. Schmidt (Xixin) samples were crushed into a powder and 1.0 g was placed in a conical flask, followed by cold immersion with 20 ml of 70% ethanol for 1 h and ultrasonic extraction (250 W, 40 kHz) for 30 min. The supernatant was collected and the residue was extracted again according to the same procedures. Then the supernatants were combined and filtered through a 0.22-μm membrane prior to HPLC determination.

Traditional Chinese patent medicines (including capsules, granules, tablets, powders and water pills) were crushed and filtered through a 60-mesh sieve. Honeyed pill was cut into small pieces. Each 50.0 mg sample was accurately weighed and placed in a 10 ml centrifuge tube, to which 4 ml of methanol was added. The mixture was ultrasonicated for 90 min at room temperature and then centrifuged at 4,000 rpm for 15 min. The supernatant was collected and dried with nitrogen at 45°C. The residue was redissolved with 1.0 ml of 30% methanol and the pH was adjusted to 8.0–9.0 using aqueous ammonia. Biotage Extrahera (Biotage, Sweden) was applied for the automatic solid phase extraction process. Waters Oasis MAX Cartridge 3 cc/60 mg columns (Waters, United States) were first activated by 3.0 ml of methanol and 3.0 ml of water. Then the columns were balanced with 3.0 ml of 30% methanol. The samples were loaded and washed sequentially with 5% aqueous ammonia and 60% methanol, followed by elution with 3.0 ml of methanol and 4.0 ml of methanol containing 8% FA. The eluent was collected and dried with nitrogen. Then the residue was redissolved with 1.0 ml methanol and centrifuged at 13,000×g for 15 min before UPLC-MS/MS detection.

Cytotoxicity analysis was determined using the Cell Counting kit 8 (CCK8) (Bioss, China). HK-2 was cultured in DMEM/F12 (containing 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin; Gibco, Invitrogen) at 37°C in 5% carbon dioxide incubator. The cells were maintained at 80% confluency and the medium was replaced every 3 days for routine cultivation. HK-2 cells (8,000 cells/well) were seeded in 96-well plates overnight. The cells were then treated with analogues of AA for 24 or 48 h at different concentrations (6.8, 27.0, 54.0, 108.0, 216.0, and 432.0 μM for AA I; 12.5, 25.0, 100.0, 200.0, 400.0, and 800.0 μM for AA II; 15.6, 62.5, 125.0, 250.0, 500.0, and 1000.0 μM for AA IIIa; 15.6, 31.3, 62.5, 125.0, 500.0, and 1,000.0 μM for AA IVa; 8.0, 16.0, 32.0, 64.0, 128.0, and 256.0 μM for AL I). Subsequently, the cells were incubated with 10% CCK8 solution for another 2 h. The final concentration of DMSO in the medium did not exceed 1.0% v/v. The absorbance (OD) of each well was measured with a spectrophotometer (Spark; Tecan, Switzerland) at 450 nm and the IC₅₀ values were obtained by fitting the curve through nonlinear regression with GraphPad Prism 7.0 (GraphPad Software, CA, United States).

The animal experiments were approved by the Research Ethics Committee of the Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences (ICMM, CACMS). The experiment was carried out according to ethical guidelines and regulations for the use of laboratory animals. A total of 160 ICR mice were randomly divided into 16 groups, each comprising 10 mice (5 males and 5 females). The drugs were administered in a single dose by oral gavage and the dose volume was 0.2 ml/10 g. The control group was treated with 0.5% CMC-Na. Other groups received AA I, AA II, AA IIIa, AA IVa, or AL I, at concentrations of 10, 20, and 40 mg/kg, respectively. Animal weights were recorded every two days. The animals were fasted overnight before experiments and sacrificed on day 14. Blood samples were collected and centrifuged for 15 min at 3,500×g. AST, ALT, ALP, CREA, BUN, TBA, and TBIL levels in plasma were examined. SPSS 17.0 (IBM Corp., United States) was used for the statistical and significance analyses. The renal and liver tissues were embedded and stained with H&E. Histological images were obtained through Olympus BX63 (Olympus Corporation, Japan) at a magnification of ×200.

ICR mice were separated into seven groups (n = 5 for each group) and administrated with saline, EMS (577.6 mg/kg), AA I (20 mg/kg), AA II (20 mg/kg), AA IIIa (20 mg/kg), AA IVa (20 mg/kg), or AL I (20 mg/kg) for three consecutive days, respectively. After anesthetization and perfusion, renal and liver tissues were obtained. Comet analysis was performed 2–6 h after the last lavage and the processes were according to Günter Speit (Speit and Rothfuss, 2012). The Comet Assay software project (CASP) (V1.2.3) was used to analyze the fraction of tail DNA and the Olive Tail Moment. A total of one hundred cells from kidney or liver tissues of each animal were analyzed. The mean, standard deviation (SD) of tail DNA%, and the median of olive tail moment were calculated. Results were analyzed using SPSS 17.0 by one-way analysis of variance (ANOVA) followed by the least significance difference (LSD) test. The significant differences between the groups were tested and defined by p-value < 0.05.

High-performance liquid chromatography (HPLC) was applied to determine the content of AA analogues (Figure 1) in roots and rhizomes of 15 batches of Asarum heterotropoides F. Schmidt (Xixin). The method showed good separation and reproducibility. AA I, AA II, AA IIIa, IVa, and AL I were eluted at 24.36, 20.81, 9.36, 11.23 and 20.31 min, respectively. The quantitative results are shown in Table 1. AA I, AA II, and AA IIIa were not detected in any of the samples, which may be partially attributed to the detection limit of the HPLC instrument. Relative levels of AA IVa and AL I were found in all samples. The contents of AA IVa and AL I were in the ranges of 66.50–121.03 μg/g and 19.73–43.75 μg/g, respectively. XX04 from Liaoning contained the least AA analogues. There were some regional differences in the contents of AA IVa and AL I, but the changes were within the range of 2 fold.

An effective and automatic SPE method was established and successfully applied in the pretreatment of traditional Chinese patent medicines. Recovery of each analyte was within the range of 86.50–111.88% at three concentration levels (Supplementary Table S4). The contents of AA analogues (AA I, AA II, AA IIIa, AA IVa, and AL I) in 44 commercially available traditional Chinese patent medicines were obtained by UPLC-MS/MS. AA I, AA IVa, and AL I were detected in most traditional Chinese patent medicines evaluated, while AA II and AA IIIa were only detected in few traditional Chinese patent medicines (Table 2, Supplementary Table S5). 70% of the samples tested contained AA I, with concentrations ranging from 0.010–52.450 μg/g. High levels of AA I (3.41–52.45 μg/g) were found in the Ershiwuweisongshi pill, the Duzhongzhuanggu capsule, and the Jingzhikesoutanchuan pill, which contained processed Aristolochia debilis Siebold & Zucc. (Madouling) or Aristolochia mollissima Hance (Xungufeng), which have been reported to have a high content of AA I (Yu et al., 2016; Mao et al., 2017). As for medicines containing only Asarum heterotropoides F. Schmidt (Xixin), a species of the genus Aristolochia, the content of AA I was below 100 ng/g. AA IVa was identified in 86.6% of the samples and the concentrations were within the range of 0.005–1.287 μg/g. Variable content of AL I was detected in all traditional Chinese patent medicines, except for the Shenmei Yangwei and the Shenqi Jianwei Granules. The highest concentration of AL I was found in the Shaqi pill at 5.565 μg/g.

Different dosage forms of traditional Chinese patent medicines with the same composition contained different amounts of AA analogues. For example, the contents of AA I, AA II, AA IVa and AL I in the Shensanqishangyao tablet were 0.020, 0.016, 0.023, and 0.034 µg/g, respectively. However, AA I or AA II were not detected in the Shensanqishangyao capsule. And the contents of AA IVa and AL I were 0.010 and 0.055 µg/g, respectively. The origins of the herb and the processing method of Asarum heterotropoides F. Schmidt (Xixin) may account for the content differences of the AA analogues.

According to the quantitative results, the contents of AA I, AA II, and AA IIIa were low, while the levels of AA IVa and AL I were relatively high in most Asarum heterotropoides F. Schmidt (Xixin)–containing medicines. We questioned whether AA IVa and AL I could play a role in the nephrotoxicity of Aristolochia. Thus, we conducted both in vitro and in vivo toxicity assays. The cytotoxicity of the AA analogues was evaluated using the CCK-8 assay. HK-2 cells were treated with AA analogues for 24 or 48 h. AA analogues showed variable degrees of cytotoxic effects in HK-2 cells, with AA I and AL I exhibiting the strongest cytotoxicity (Figure 2). The IC₅₀ values of AA I were 197.3 μM for 24 h and 76.7 μM for 48 h, respectively. AA II showed relatively weak cytotoxicity with IC₅₀ > 800 μM for 24 h and 306.5 μM for 48 h. The cell viability was not lower than 50% of control at a concentration of 1,000 μM for AA IIIa and AA IVa at 24 or 48 h, suggesting low cytotoxicity in HK-2 cells. AL I exhibited relatively low cytotoxicity at 24 h with IC₅₀ >256 μM. However, the toxicity rapidly enhanced with increasing incubation time. After treatment for 48 h, the IC₅₀ value of AL I was reduced to 37.1 μM, exhibiting even stronger cytotoxicity than AA I. Despite the varying cytotoxicity, all AA analogues showed a time- and concentration-dependent inhibition to HK-2 cells.

To further evaluate the in vivo toxicity of AA analogues, mice were administered single doses of AA I, AA II, AA IIIa, AA IVa, and AL I at 10, 20, and 40 mg/kg, respectively. The control group received 0.05% CMC-Na. Mice administered with AA I achieved a remarkable reduction of body weight compared to the control group. In addition, the body weight decreased with the increasing doses of AA I for both male and female mice (Supplementary Figure S3). Slight weight loss was also observed in male and female mice treated with AA II (40 mg/kg) during the first 10 days. However, the body weight of female mice increased after 10 days and there was no significant difference with the control group. Regarding AA IIIa, AA IVa, and AL I, no distinct changes in body weight were observed between the different dosages and the control group. Female mice treated with AA IVa (40 mg/kg) experienced a drastic weight loss on day 5, but the weight returned to the same level as in the other groups on day 7.

Survival of the control and treated groups were monitored for 14 days. As shown in Figure 3, mice (n = 7 of 10) treated with 40 mg/kg AA I were dead on days 4, 5, and 6 due to the acute kidney failure and the survival rate was 30%. 10% of mice died (n =1 of 10) on day 8 in the group treated with AA II (40 mg/kg). There was no death occurring in control, AA I (10, 20 mg/kg), AA II (10, 20 mg/kg), AA IIIa (10, 20, 40 mg/kg), AA IVa (10, 20, 40 mg/kg), AL I (10, 20, 40 mg/kg) treated groups.

After mice were sacrificed, kidney and liver tissue were obtained for histopathological analysis. Significant lesions were only found in the renal tissues of mice treated with 40 mg/kg AA I. As shown in Figure 4, renal pathological changes were characterized with large area necrosis of renal tubules, epithelial degeneration of renal tubules, and granular casts in renal tubule lumen. Although there were different levels of changes in serum levels of TBIL, ALP, BUN, and CREA in AA I (20 mg/kg, 40 mg/kg) treated mice (Figure 5), no corresponding pathological changes were observed in the liver tissues of the same group (Supplementary Figure S5). The activities of AST, ALT, ALP and the concentrations of TBIL, TBA, CREA and BUN in AA II, AA IIIa, AA IVa, and AL I treated groups showed no significant difference compared to the control group, except for decreased AST, ALT, and ALP activity in the group treated with AL I (10 mg/kg). No pathological changes were observed in the liver tissues of mice treated with AL I.

The comet assay is a simple and rapid method for detecting DNA damage in single cells. After cell lysis and DNA unwinding, undamaged DNA will remain in the nucleus and appear round during electrophoresis; however, cells with DNA damage resemble the shape of a comet as a result of the migration of negatively charged DNA from the nucleus toward the anode. The extent of DNA migration is positively correlated with the amount of DNA damage in cells (Speit and Rothfuss, 2012). We conducted comet assays to study the DNA damage of AA analogues. Ethyl methanesulfonate (EMS) was a strong genotoxic agent (Gocke et al., 2009) and was used as a positive control in this study. Hepatocyte and renal cells in the control group appeared round during electrophoresis, while EMS-treated hepatocyte and renal cells showed significant comet tailing (Figure 6, Supplementary Figure S6). Renal cells from AA I (20 mg/kg) treated mice showed an evident comet shape. The tail DNA% was 15.39 ± 1.75 (p < 0.001) and the olive tail moment was 7.45 ± 1.94 (p < 0.05), respectively (Supplementary Table S6). However, no DNA damage was observed in hepatocyte cells of mice treated with AA I (20 mg/kg). No DNA damage was observed in renal cells nor in hepatocyte cells in mice treated with AA II, AA IIIa, AA IVa, and AL I. These results indicated that AA I could cause direct DNA damage to the renal tissues, while the ability to induce DNA damage for AA II, AA IIIa, AA IVa, and AL I was relatively low at 20 mg/kg.