Article pubs.acs.org/JAFC
Determination of Aristolochic Acids by High-Performance Liquid Chromatography with Fluorescence Detection Yinan Wang and Wan Chan* Department of Chemistry, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong ABSTRACT: Nephrotoxic and carcinogenic aristolochic acids (AAs) are naturally occurring nitrophenanthrene carboxylic acids in the herbal genus Aristolochia. The misuse of AA-containing herbs in preparing slimming drugs has caused hundred of cases of kidney disease in Belgium women in a slimming regime in the early 1990s. Accumulating evidence also suggested that prolong dietary intake of AA-contaminated food is one of the major causes to the Balkan endemic nephropathy that was first observed in the late 1950s. Therefore, analytical methods of high sensitivity are extremely important for safeguarding human exposure to AAcontaining herbal medicines, herbal remedies, and food composites. In this paper, we describe the development of a new highperformance liquid chromatography coupled fluorescence detector (HPLC−FLD) method for the sensitive determination of AAs. The method makes use of a novel cysteine-induced denitration reaction that “turns on” the fluorescence of AAs for fluorometric detections. Our results showed that the combination of cysteine-induced denitration and HPLC−FLD analysis allows for sensitive quantification of AA-I and AA-II at detection limits of 27.1 and 25.4 ng/g, respectively. The method was validated and has been successfully applied in quantifying AAs in Chinese herbal medicines. KEYWORDS: aristolochic acids, high-performance liquid chromatography, fluorescence, cysteine, hydrogenolysis
■
INTRODUCTION Nephrotoxic aristolochic acids (AAs) are a mixture of nitrophenanthrene carboxylic acids derived from the herbal genus Aristolochia and are classified as members of the most potent carcinogens in the Carcinogenic Potency Database.1,2 Major components of AAs include aristolochic acid I (AA-I) and aristolochic acid II (AA-II) that differ by a methoxy group, as shown in Figure 1. AA-containing herbs had been widely
caused about 100 cases of renal diseases in Belgium in the early 1990s.1,4 The term Chinese herb nephropathy (CHN) has been used since then to describe this unique type of rapid progress renal fibrosis that is associated with prolonged exposure to AAcontaining herbs or herbal products.1,5 Cases of CHN were also reported in Australia, China, Hong Kong, Japan, Taiwan, U.K., and many other areas.1 Emerging evidence also suggested that chronic food poisoning through dietary intake of AA-contaminated flour, maize, and cucumber is responsible for the etiology of Balkan endomic nephropathy (BEN),5−7 a peculiar renal disease that has been observed on thousands of farmers living alongside the Danube River for over the past 50 years. Because of the strong nephrotoxicity and potent human carcinogenicity of AAs and the reason that there is significant evidence showing that AAs are food/environmental contaminants,7−10 analytical methods capable of detecting low concentrations of AAs in a complex matrix of herbal remedies, botanical products, and flour are highly necessary. Current methods for AA determination include enzymelinked immunosorbent assay (ELISA),11 high-performance liquid chromatography with ultraviolet detection (HPLC− UV),8−10 high-performance liquid chromatography with mass spectrometry (HPLC−MS),11,12 capillary electrophoresis with ultraviolet detection (CE−UV),13,14 capillary electrophoresis with electrochemical detection (CE−ECD),15−19 and cyclic voltammetry.20 As one of the most sensitive and selective analytical tools, the application of a fluorescence detector (FLD) on AA determination is only made possible recently by
Figure 1. Cysteine-induced denitration of AA (AA-I, R = OCH3; AAII, R = H) forms of fluorescent aristolic acids for HPLC−FLD analysis.
used as Chinese herbal medicines to treat tumors, rheumatism, snake bites, small pox, and pneumonia, until AAs were observed to be strong human carcinogens and nephrotoxins.1−4 Currently, the use of AA-containing herbs is prohibited in regions including China, Hong Kong, Taiwan, Japan, U.S.A., U.K., and many European countries.1,5 Despite AA-containing herbs being banned from the markets worldwide, they are still available for sale on the Internet and misuse of AA-containing herbs exists.6 For example, inadvertent substitution of Stephania tetrandra with AA-containing Aristolochia fanghi in the preparation of slimming drugs has © 2014 American Chemical Society
Received: Revised: Accepted: Published: 5859
April 3, 2014 June 10, 2014 June 11, 2014 June 11, 2014 dx.doi.org/10.1021/jf501609j | J. Agric. Food Chem. 2014, 62, 5859−5864
Journal of Agricultural and Food Chemistry
Article
programmed to 60% in 10 min, which was then ramped rapidly to 100% B in 0.1 min, and maintained 100% B for 5 min before reconditioning to 10% B for 8 min. Calibration. A stock solution mixture of AA-I and AA-II each at 240 μg/mL was prepared in methanol and stored at −20 °C before use. Working standard solution mixtures of AA-I and AA-II of concentrations of 0.096, 0.24, 1.20, 2.40, and 4.80 μg/mL were prepared by diluting the stock solution with methanol. The calibration standard solutions were prepared by mixing 100 μL of the working standard solutions with 900 μL of the denitrating reagent, reacting, and HPLC analyzing as described above. Calibration curves were established by plotting the peak areas of the denitrated AAs against the AA concentrations in the calibration standard solutions. Method Validation. The optimized method was also validated for analytical sensitivity, precision, and accuracy. The limit of detection (LOD) and limit of quantification (LOQ) were established as the amount of analyte in the blank sample extract that generated a signal 3 and 10 times the signal-to-noise ratio, respectively.22,23 The method precision was evaluated by analyzing herbal extracts spiked with AAs at three different concentrations (0.096, 1.92, and 4.80 μg/mL), on the same day (n = 7) and over separate days in a month (n = 7). The method accuracy was determined by spiking AAs at the three stated concentrations to the blank herbal sample, processed, and analyzed as described above. Method Comparison. The method accuracy was further gauged by direct liquid chromatography−mass spectrometry (LC−MS) analysis of AAs in the underivatized sample extracts on a Waters Acquity ultra-performance liquid chromatograph (UPLC) coupled with a Xevo G2 Q-TOF mass spectrometer with a standard electrospray ionization (ESI) interface (Milford, MA). AAs were resolved with a Waters BEH C18 column (50 × 2.1 mm, 1.8 μm) eluted with the following gradient of methanol in 20 mM ammonium acetate at a flow rate of 0.35 mL/min and 40 °C: 0−7 min, 1−40%; 7.1−9 min, 40−100%; 9−12 min, 100%; and 12.1−15 min, 1%. The LC eluate was analyzed by positive ESI−MS with the following parameters: capillary voltages, 3 kV; cone voltages, 25 V; desolvation temperature, 400 °C; and source temperature 120 °C. The peak areas of the pseudo-molecular ion ([M + NH4]+) of AAs (m/z 359 for AA-I and 329 for AA-II) were adopted quantifying AAs in the sample extracts. The results obtained from LC−MS analyses were compared to that obtained by HPLC−FLD analyses to evaluate the accuracy of the proposed method.
converting the non-fluorescing AAs into their corresponding aristolactams that fluoresce.2,10 However, these methods suffer from the drawbacks that individual samples have to be analyzed twice before the AA content can be determined. The reason underlying this is that the products from the derivatization reaction, i.e., the aristolactams, exist naturally in high abundance in the AA-containing herbs. The concentrations of endogenous aristolactams in the underivatized sample extracts must be determined and deducted from the total aristolactam levels in the derivatized sample extracts before the AA concentrations in the herbal samples can be elucidated. Consequently, although the developed high-performance liquid chromatography with fluorescence detection (HPLC−FLD) methods are sensitive, the necessities of duplicated analysis for a single sample rendered these methods inefficient for AA determination. Herein, we report a novel HPLC−FLD method for the sensitive and efficient determination of AAs. The method makes use of cysteine-induced hydrogenolysis of AAs, in which the fluorescent-inhibiting nitro group is replaced by hydrogen and allows for sensitive and selective fluorometric determination of the otherwise non-fluorescing AAs by HPLC−FLD.21 To the best of our knowledge, the application of the thiolinduced hydrogenolysis converting non-fluorescing nitroaromatics into fluorphores for their quantitative analysis by HPLC−FLD has not been reported before. Our proposed method offering superior sensitivity was validated and has been successfully applied in quantifying AAs in Chinese herbal medicines.
■
EXPERIMENTAL SECTION
Chemicals. All chemicals and reagents of the highest purity available were used without further purification, unless noted otherwise. AA (96% purity), a 1:1 mixture of AA-I and AA-II, was purchased from Acros (Morris Plains, NJ). Alanine, cysteamine, Lcysteine, cysteinylglycine, glutathione (GSH), and 2-mercaptoethanol were obtained from Sigma (St. Louis, MO). Fructus Aristolochiae, Herba Aristolochiae, Aristolochiae mollissima, Radix Aristolochiae, and Aristolochiae cinnabarina were purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Herba Asari and Akebia Stem were obtained from local Chinese drug stores in Hong Kong. HPLC-grade acetonitrile and methanol were purchased from Tedia (Fairfield, OH). Water was produced by a Milli-Q Ultrapure water purification system (18.2 MΩ, Millipore, Billerica, MA). Sample Preparation. Herbal medicines and herbal preparations were powdered and sieved (size smaller than 100 μm) prior to analysis. Approximately 50 mg of the powdered samples were accurately weighted and extracted with 5.0 mL of methanol by sonication. After 30 min of sonication at 50 °C, 100 μL of the sample extracts were mixed with 900 μL of the denitrating reagent comprising 50 mM cysteine in potassium phosphate buffer (50 mM, pH 8.0). The reaction was conducted by heating the reaction mixture at 80 °C for 2 h. The cysteine-treated samples after being cooled to room temperature were centrifuged at 13 800 rpm for 5 min before the supernatant was transferred to a HPLC vial for HPLC analysis. HPLC Analysis. HPLC analysis was conducted on an Agilent 1100 HPLC system coupled with a diode array detector (DAD) and a FLD connected in series (Palo Alto, CA). The DAD was set to monitor the UV absorption of the eluate at 254 nm, whereas the excitation and emission wavelengths for the FLD were set at 254 and 390 nm, respectively. A total of 10 μL of the cysteine-treated sample extract was injected into a Zorbax Eclipse XDB-C8 column (150 × 4.6 mm, 5 μm, Agilent). The HPLC mobile phases consisted of 20 mM, pH 7.0 ammonium acetate (A) and acetonitrile (B). A gradient elution program at a flow rate of 0.70 mL/min was applied to analyze the cysteine-treated samples. The solvent gradient was started from 10% B,
■
RESULTS AND DISCUSSION Fluorometric Properties of AAs. Aromatic compounds generally fluoresce because of the presence of delocalized π electrons. However, the fluorescence of nitroaromatic hydrocarbons is not observed.24 It is believed that the existence of a low-lying n → π* electronic transition in nitroaromatics that favors molecular relaxation by intersystem crossing the competing relaxation by fluorescence is therefore inhibited.24 AAs, although being rigid polycyclic aromatics, do not fluoresce.2 The presence of an electron-withdrawing −NO2 substituent inhibits their molecular fluorescence and, thus, hinders their fluorometric detections. Once the fluorescent inhibiting nitro group of AAs has been removed, the fluorescence of denitro AA products could be “turned on” to allow for sensitive and selective detection of AAs. To test this, we produced aristolic acids by reacting the mixed standard of AA-I and AA-II with cysteine and analyzed the reaction mixture by HPLC−FLD. Characterization of the Cysteine-Induced Hydrogenolysis Product of AAs. Figure 2 depicts the chromatograms obtained from analyzing the underivatized (A and B) and cysteine-treated (C and D) AA-I and AA-II mixture. The HPLC analysis revealed the cysteine treatment of AAs indeed produced two fluorescing products at chromatographic 5860
dx.doi.org/10.1021/jf501609j | J. Agric. Food Chem. 2014, 62, 5859−5864
Journal of Agricultural and Food Chemistry
Article
Selection of a Suitable Hydrogenolysis Reagent. As the basis for our method, Priestap and Barbieri recently reported that cysteine and GSH can induce hydrogenolysis of AAs at physiological relevant conditions, producing aristolic acids as the metabolites (Figure 1).21 Aristolic acids, AAs with the fluorescent inhibiting nitro group replaced by hydrogen, exhibit fluorescence properties that allow for sensitive determination of AAs by spectrofluorometry (Figure 1). With the hypothesis that other thiol-containing compounds may also be candidate denitrating reagents, cysteamine, cysteinylglycine, GSH, and 2-mercaptoethanol were investigated for their capabilites on producing the aristoloic acids from AAs. Our results showed that all of the tested thiolcontaining reagents are capable of inducing AA hydrogenolysis, with cysteine being the most efficient reagent in removing the nitro group of AAs (Figure 4). Cysteine was therefore chosen as the hydrogenolysis reagent in this study.
Figure 2. Typical chromatograms from HPLC analysis of underivatized (A, DAD trace; B, FLD trace) and cysteine-treated (C, DAD trace; D, FLD trace) AAs. Peaks 1, 2, 3, and 4 correspond to AA-II, AA-I, aristolic acid II, and aristolic acid I, respectively.
retention times of 8.71 and 8.91 min (Figure 2D), with their corresponding emission maxima at 385 and 390 nm, respectively (Figure 3A). UV absorbance measurements of
Figure 4. (A) Relative capability of some common thiol-containing compounds in inducing AA denitration and (B) effect of pH on the extent of cysteine-induced AA denitration. GSH, glutathione; BME, 2mercaptoethanol; Cys-Gly, cysteinylglycine; β-MEA, cysteamine; and Cys, cysteine.
As an indirect approach to investigate the mechanism underlying the AA denitration reaction, we have also reacted AAs with alanine, an amino acid similar to cysteine but without the thiol functional group. Our results showed that no aristolic acids were produced in the alanine-catalyzed reaction, indicating that the thiol functional group is essential for the denitration process. Thus, we believed that the thiol functional group is participating in the electron transfer processes, which lead to the hydrogenolysis of AAs, producing the fluorophoric aristolic acids. Optimization of the AA Hydrogenolysis Conditions. After the denitrating reagent was selected, the next step was to optimize the reaction conditions for AA denitration. Specifically, the formation of the aristoloic acids at different cysteine concentrations, pH values, temperatures, and reaction times were investigated. When the cysteine concentrations (from 10 to 150 mM) and pH (from 3.0 to 11.0) were changed in different reactions and the peak areas of aristolic acids in HPLC−FLD analyses were monitored, it was found that cysteine at 50 mM (Figure 5) and in pH 8.0 (Figure 4) is the most efficient for AA denitration. It is worth noting that the denitrating reaction is favored at alkaline conditions. This observation supports our proposed hydrogenolysis mechanism that the conjugated base of cysteine RS− (pKa = 8.36) initiated the AA denitration reaction and that the mechanism underlying the cysteine-induced denitration reaction appears to be
Figure 3. (A) Fluorescence emission and (B) UV absorption spectra of aristolic acids. (C and D) High-accuracy MS spectra of aristolic acid II and arstiolic acid I, respectively.
the chromatographic peak at 8.71 min revealed absorption maxima at 218, 258, 284, 320, 350, and 370 nm (Figure 3B). A similar UV absorption spectrum was also observed for the chromatographic peak at 8.91 min, with absorption maxima at 226, 254, 296, 326, 356, and 374 nm (Figure 3B). High-accuracy MS analyses of the HPLC eluate collected at retention times of 8.71 and 8.91 min showed [M + H]+ ions at m/z 267.0657 (Figure 3C) and m/z 297.0760 (Figure 3D), respectively. The close correlation between the measured (267.0657 and 297.0760) and theoretical (267.0657 and 297.0763) m/z values indicated that the fluorescing products were aristolic acid II (C16H11O4) and aristolic acid I (C17H13O5), respectively.
Ar−NO2 + RS− → Ar−NO2• − + RS• 5861
dx.doi.org/10.1021/jf501609j | J. Agric. Food Chem. 2014, 62, 5859−5864
Journal of Agricultural and Food Chemistry
Article
Figure 5. Influence of the (A) cysteine concentration and (B) reaction time on the denitration reaction of AAs. The data represent the mean ± standard deviation (SD) for three independent experiments with AAs at 4.80 μg/mL.
Ar−NO2• − → Ar • + NO2− Ar • + H 2O → − Ar−H + •OH
The influences of the reaction time and temperature on the extent of hydrogenolysis were studied over 1−5 h and from 25 to 80 °C, respectively. Our results showed that a 2 h reaction (Figure 5) at 80 °C (Figure 6) gave the best detector response.
Figure 7. Typical chromatograms obtained from HPLC−FLD analyses of cysteine-treated AA-I (0.192 μg/mL) and AA-II (0.192 μg/mL), (A) standard mixture and (B) underivatized, and (C) cysteine-treated extracts of Herba Aristolochiae (Tianxianteng). The retention times of aristoic acid II and aristolic acid I were 8.7 and 8.9 min, respectively.
Table 1. Linear Regression Parameters of the Calibration Curves, Precision, Accuracy, and LOD of the Developed HPLC−FLD Method linear range (μg/mL) slope intercept r2 precision intraday (RSD, %) interday (RSD, %) accuracy (% mean deviation) LOD (ng/mL) LOD (ng/g)a
Figure 6. Influence of the reaction temperature on the cysteineinduced denitration reaction of AAs. The data represent the mean ± SD for three independent experiments with AAs at 4.80 μg/mL.
a
The optimized parameters above were selected and used in all studies. Under the developed hydrogenolysis conditions, the reaction yield of aristolic acid I and aristolic acid II from denitration of AA-I and AA-II were determined to be 95.2 and 96.4%, respectively. Calibration and Method Validation. Calibration curves were constructed using standard solution mixture of AAs in triplicate reactions, with the concentrations of AA-I and AA-II ranging from 0.096 to 4.80 μg/mL. Depicted in Figure 7A is a typical chromatogram obtained from HPLC−FLD analysis of the reaction mixture containing AA-I and AA-II at 0.192 μg/ mL. The peak areas of aristolic acids increased linearly over this concentration range, with r2 values better than 0.9992. Table 1 summarizes the calibration slopes, intercepts, and coefficients of determination for the AAs determined in a calibration experiment. The LODs defined as the concentrations of AA-I and AA-II that generated a signal 3 times the signal-to-noise ratio were 0.271 and 0.254 ng/mL,22,25 respectively, corresponding to a method detection limit of 27.1 ng of AA-I and 25.4 ng of AA-II per gram of herbal medicines (Table 1). The LODs obtained
AA-I
AA-II
0.096−4.80 467.69 0.8259 0.9992
0.096−4.80 507.81 0.3741 0.9992
0.45−1.48 0.64−2.95 1.0−5.1 0.271 27.1
0.57−1.31 0.71−2.82 −3.3−8.5 0.254 25.4
The LOD is the sample extract based on 50 mg of sample.
by the HPLC−FLD method are not differentiated from the existing LC−MS/MS (0.14−250 ng/mL)15,16 and HPLC− FLD methods (0.39−26.0 ng/mL).2,10 The intra- and interday reproducibility of the developed method were determined by evaluating three individual reactions containing a mixture of AA-I and AA-II. The relative standard deviations for the intra- and interday reproducibility were less than 1.48 and 2.95%, respectively, at the tested concentrations (0.096, 1.92, and 4.80 μg/mL). The method accuracy as determined by spiking AAs at the three stated concentrations to blank herbal samples, cysteine treating, and HPLC analyzing was less than 8.5% deviated from the true values. The results of the reproducibility and accuracy of the HPLC−FLD method are summarized in Table 1. The robustness and high analytical precision of the data demonstrated a good performance of the HPLC−FLD method for AA determination. HPLC−FLD Quantification of AAs in Herbal Medicine. The developed method combining cysteine-induced denitration 5862
dx.doi.org/10.1021/jf501609j | J. Agric. Food Chem. 2014, 62, 5859−5864
Journal of Agricultural and Food Chemistry
Article
Table 2. Determination of AAs (μg/g) in Selected Herbal Medical Samples by the Developed HPLC−FLD and LC−MS Methodsa AA-I Fructus Aristolochiae (Madouling)b Herba Aristolochiae (Tianxianteng) Aristolochiae mollissima (Xungufeng)b Radix Aristolochiae (Qingmuxiang)b Aristolochiae cinnabarina (Zhushalian)b Herba Asari (Xixin)c Akebia Stem (Mutong)c
AA-II
HPLC−FLD
LC−MS
HPLC−FLD
LC−MS
617.2 ± 22.4 53.7 ± 1.5 1086.7 ± 125.9 965.3 ± 38.4 911.3 ± 8.3
