This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.
Article pubs.acs.org/jnp
LC-MS- and 1H NMR-Based Metabolomic Analysis and in Vitro Toxicological Assessment of 43 Aristolochia Species Johanna Michl,† Geoffrey C. Kite,‡ Stefan Wanke,§ Oliver Zierau,∥ Guenter Vollmer,∥ Christoph Neinhuis,§ Monique S. J. Simmonds,‡ and Michael Heinrich*,† †
Research Cluster Biodiversity and Medicines/Centre for Pharmacognosy and Phytotherapy, UCL School of Pharmacy, 29-39 Brunswick Square, London, WC1N 1AX, United Kingdom ‡ Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3AB, United Kingdom § Institute for Botany and ∥Institute for Zoology, Molecular Cell Physiology and Endocrinology, Technische Universität Dresden, Zellescher Weg 20b, Dresden, 01062, Germany S Supporting Information *
ABSTRACT: Species of Aristolochia are used as herbal medicines worldwide. They cause aristolochic acid nephropathy (AAN), a devastating disease associated with kidney failure and renal cancer. Aristolochic acids I and II (1 and 2) are considered to be responsible for these nephrotoxic and carcinogenic effects. A wide range of other aristolochic acid analogues (AAAs) exist, and their implication in AAN may have been overlooked. An LC-MS- and 1H NMR-based metabolomic analysis was carried out on 43 medicinally used Aristolochia species. The cytotoxicity and genotoxicity of 28 Aristolochia extracts were measured in human kidney (HK-2) cells. Compounds 1 and 2 were found to be the most common AAAs. However, AA IV (3), aristolactam I (4), and aristolactam BI (5) were also widespread. No correlation was found between the amounts of 1 or 2 and extract cytotoxicity against HK-2 cells. The genotoxicity and cytotoxicity of the extracts could be linked to their contents of 5, AA D (8), and AA IIIa (10). These results undermine the assumption that 1 and 2 are exclusively responsible for the toxicity of Aristolochia species. Other analogues are likely to contribute to their toxicity and need to be considered as nephrotoxic agents. These findings facilitate understanding of the nephrotoxic mechanisms of Aristolochia and have significance for the regulation of herbal medicines.
F
clematitis L. seeds has been identified as the causal factor for Balkan endemic nephropathy (BEN).11,12 This disease has affected thousands of patients in the Danube basin, and its clinical expression and pathological lesions are strikingly similar to AAN.13 In Taiwan, recent research has shown that exposure to Aristolochia spp. contributes significantly to the incidence of UUC.2 A characteristic TP53 tumor suppressor gene mutational signature, dominated by otherwise rare A:T to T:A transversions, was found in patients with UUC. The mutational signature was identical to that observed in UUC associated with BEN.2 The carcinogenic effects of species of Aristolochia may not become manifest until 30 or more years after exposure.14 Therefore, health practitioners usually fail to identify the link between the tumors and exposure to Aristolochia species.9 Recent studies showed that Aristolochia spp. intake is a likely cause of tumors previously attributed to other carcinogens, such as smoking and chronic hepatitis infection.1 It has been
or centuries, species of the plant genus Aristolochia have been used to treat a variety of illnesses as components of traditional Chinese medicine (TCM).1 Species of Aristolochia have become a key concern in healthcare since they are known to cause aristolochic acid nephropathy (AAN), a renal fibrosis often associated with upper urothelial cancer (UUC).2 This condition was initially reported in a cohort of Belgian women after the intake of pills used for dieting containing a Chinese herb, Aristolochia fangchi Wu ex L.D. Chow & S.M. Hwang.3,4 The use of this plant resulted in more than 100 cases of chronic tubulointerstitial disease progressing to end-stage renal failure.5 Following this disastrous incident, sporadic cases of AAN have been reported in countries throughout the world.2,6,7 Although the nephrotoxic and carcinogenic effects of species of Aristolochia are well known, in many countries drugs and medical preparations containing Aristolochia species are still widely (and often legally) used and can be purchased via the Internet.8−10 Evidence for AAN being a major public health problem has been demonstrated only in two small regions worldwide, specifically the Balkan peninsula and Taiwan. In the Balkan region, the dietary exposure of flour contaminated with A. © XXXX American Chemical Society and American Society of Pharmacognosy
Received: June 23, 2015
A
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
spectroscopy-based metabolomic approach was chosen in order to detect all AAAs present in a sample set representing important medicinally used Aristolochia species. We included 43 species of Aristolochia and compared the phytochemical profiles of different extracts. This included intraspecific comparisons of different plant parts and source materials from different origins. A principal component analysis (PCA) scores plot for the LC-MS data showed that the samples are clustered into main groups (Figure S1a, Supporting Information). As expected, there is great variation between the phytochemical profiles of different species of Aristolochia, and samples from the same species do not always form clusters in the PCA scores plot. For example, out of five stem samples (samples 48, 49, 50, 54, and 55) of A. manshuriensis Kom. from different origins, only three samples (samples 48, 49, and 50) formed a cluster in the PCA scores plot. This suggests that the intraspecific variation within individual species of Aristolochia can be large and that the phytochemistry of the plants is influenced by factors such as origin or plant part. In the PCA scores plot of the 1H NMR data set (Figure S1b, Supporting Information), the outliers were mainly fruit, flower, and seed samples, e.g., sample 65 and 66 [A. paucinervis Pomel (fruit and seed)], sample 71 [A. serpentaria L. (fruit)], sample 46 [A. acuminata Lam. (flower)], and sample 78 [A. trilobata L. (flower)]. In the LC-MS data set (Figure S1a, Supporting Information), with the exception of sample 40 [A. indica (leaf)], all leaf samples formed a cluster with similar metabolite profiles. Furthermore, all root samples [with the exception of sample 38 (A. indica; root)] clustered together on the right-hand side of the scores plot. However, this cluster of root samples overlapped with a set of stem samples [samples 27 (A. fangchi), 36 (A. guentheri O.C. Schmidt), 42 (A. kaempferi Willd.), 47 (A. macrophylla Lam.), 48−50 (three A. manshuriensis samples), 73 (A. taliscana Hook. & Arn.), 77 (A. triangularis Cham.), and 81 (A. westlandii Hemsl.)]. It is noteworthy that the majority of these stem samples originated from older, woody plants, rather than young, green stems. Therefore, older stem samples have a similar metabolite profile to root samples. These results suggest that apart from genetic factors the phytochemistry of species of Aristolochia is influenced by factors such as the part of the plant used, the age of the plant, and the season of collection and environmental conditions. According to the PCA loadings plots for the LC-MS data set, samples with negative PC1 scores (Figure S2a, Supporting Information) were characterized by higher amounts of 1 (m/z 359.0870, [M + NH4]+, tR 35.93 min), whereas samples with positive PC1 scores were characterized by an alkaloid giving [M]+ at m/z 342.1706 (tR 7.38 min). Among the possible candidate compounds suggested by the molecular formula of this alkaloid (C20H24NO4+; 1.8 ppm) is the aporphine derivative magnoflorine, which was previously reported in Aristolochia.30 Samples with negative PC2 scores were additionally characterized by aristolactam I N-β-D-glucoside (7, [M + H]+, tR 26.18 min) and a component giving a peak at m/z at 372.1441 ([M + NH4]+, tR 35.16 min). The molecular formula and MS/MS fragmentation of the latter peak were in accordance with the lignans asarinin or sesamin known from Asarum species.31 In contrast to the LC-MS data set, the differentiation between samples in the 1H NMR data set was caused mainly by primary metabolites, such as sugars and fatty acids. Samples with positive PC1 loadings (Figure S3a, Supporting Information) in the 1H NMR data set were characterized by a variety of overlapping peaks in the
estimated that in mainland China alone one hundred million people may be at risk of developing AAN.14,15 Given the widespread medicinal use of different species of Aristolochia worldwide, it is likely that AAN and UUC are also prevalent in other countries,9 creating a global public health problem of considerable but largely unknown magnitude.6,14,16 Aristolochic acids I (1) and II (2) are considered to be the cause of the nephrotoxic and genotoxic effects.12,17 Aristolochic acids are metabolized to a reactive cyclic N-acylnitrenium ion to form DNA adducts.7,18,19 However, it has been hypothesized that aristolochic acid-induced nephrotoxicity is not associated with DNA adduct formation.20,21 Mechanisms such as p53 activation,22 oxidative stress,23 and overexpression of transforming growth factor (TGF)-β24 are also implicated in AAN. Interestingly, 1 and 2 have similar genotoxic and carcinogenic potentials, but 1 is solely responsible for the nephrotoxicity associated with AAN.25,26 Although it is clear that compounds 1 and 2 are toxic,27 they are not necessarily the only (or most potent) toxins present in Aristolochia and related genera. At least 178 aristolochic acid analogues (AAAs) have been reported, and so far their possible implications in AAN have not been investigated in detail. These compounds may participate in processes that lead to renal damage28,29 and carcinogenesis. The aim of this work was to assess the toxicological risk of medicinally used species of Aristolochia and to identify constituents responsible for the nephrotoxic and genotoxic effects. Therefore, a comprehensive LC-MS- and 1H NMRbased metabolomic analysis of 43 medicinally used species of Aristolochia was carried out. Using a novel approach, the cytotoxic and genotoxic effects of Aristolochia extracts were analyzed in vitro, and the relationship between phytochemical profiles and in vitro toxicity was studied.
■
RESULTS AND DISCUSSION Phytochemical Variation in Aristolochia Species Using LC-DAD-ESIMS and 1H NMR Metabolomics. Previously, research has mainly focused on the phytochemistry of individual species of Aristolochia used in TCM and a few selected compounds isolated from these species.27 However, many species of Aristolochia are used as herbal medicines worldwide, but so far their phytochemistry is largely unknown. Since other AAAs apart from 1 and 2 may contribute to the toxicity of Aristolochia, a LC-DAD-ESIMS- and 1H NMR B
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
Figure 1. Heat map comparing relative LC-MS peak areas of identified AAAs. Information on the origin of the samples is given in Table S1, Supporting Information. Retention times, UV maxima, and fragmentation ions for all compounds are listed in Table S2, Supporting Information. The samples (columns) are sorted in descending order according to their average peak areas across all aristolochic acid analogues. The compounds (rows) are sorted by descending average peak areas across all samples [aidentified by comparison with reference standard; btentative assignment based on accurate mass, UV spectra, and mass fragmentation, c3-hydroxy-4-methoxy-10-nitrophenanthrene-1-carboxylic acid methyl ester].
carbohydrate region. However, peaks for sucrose [δ 5.415 (d, J = 3.8 Hz), δ 4.17 (d, J = 8.5 Hz)], α-glucose [δ 5.175 (d, J = 3.8 Hz)], and β-glucose [δ 4.55 (d, J = 7.8 Hz)] were apparent. Samples with negative PC1 scores were characterized by peaks in the aliphatic regions of the spectra. These peaks represent protons from fatty acids such as δ 2.31 (CH−CH2-CH), 2.07 (−H2C−CHCH−), 1.59 (−CH2−CH2−COOH), 1.275 (CH2 protons), and 0.885 (CH3 protons). Variation in Aristolochic Acid Analogues. The majority of previously reported analytical techniques have focused on the detection of 1 and 2, as these two compounds are the two main secondary metabolites in many species of Aristolochia. However, in this study, targeted metabolite profiling was carried out to detect all AAAs present in the sample set. Several AAAs were identified by comparison to reference standards [1, 2, aristolochic acid IV (3), aristolactam I (4), aristolochic acid D (8), aristolochic acid IIIa (10), aristolochic acid IIIa 6-O-6-Dglucoside (12), aristolochin (14), and aristolochic acid III (15)]. Tentative assignments based on accurate mass, retention times, UV maxima, and fragmentation ions obtained through LC-MS analysis were made for several other AAAs (Table S2, Supporting Information). Relative LC-MS peak areas of identified AAAs are presented as a heat map in Figure 1. The samples are sorted according to their average peak areas across all AAAs. AAAs were present in most samples. While 1 and 2 are the most common AAAs, other compounds such as 3, 4, and AL BI (5) are widespread. The occurrence of 3 at high levels in species of Aristolochia has not been reported before. This compound may also be of great toxicological concern, since it is known to be mutagenic.32 The presence of these compounds indicates that the nephrotoxic effects of species of Aristolochia are most likely not caused by 1 and 2 alone. Interestingly, some of the samples [e.g., A. guentheri (leaf and stem) or A. elegans Mast. (root)] do not contain 1 and 2, but a large variety of other AAAs such as 4, 5, AL BIII (9), 10, 12, and 14. Therefore, in these cases, quantification of 1 and 2 alone (as it has been carried out in many previous
phytochemical studies and in pharmacopoeias) would have been misleading for a toxicological risk assessment. There are patterns in the variation of AAAs in different plant parts. Root samples generally contain a large variety of AAAs. This is especially evident with sample 59 [A. maxima Jacq. (root)], where more than 10 AAAs could be detected. Furthermore, samples 23 [A. elegans (root)] and 1 [A. acuminata (root)] both have a large chemical diversity. Within leaf samples, there is usually less chemical variety. In most cases, only 1 and 2 were present, and some samples did not even contain AAAs at all. Another interesting observation is that all analyzed seed, fruit, and flower samples contained either high levels of 1 or both 1 and 2. This suggests that such plant parts accumulate these toxins. The amounts of 1, 2, 4, 8, and 10 were quantified using LCMS. The quantities in mg per g crude drug are shown in Table S3 and Figure S4, Supporting Information. The highest amounts of 1 were detected in A. maxima roots (2.467 mg/g crude drug). Interestingly, quantities of compound 2 for some of the samples included were even higher, such as 2.557 mg/g crude drug in A. clematitis roots. In contrast, other compounds such as aristolactam I were present in lower quantities. The maximum quantity of 4 within the data set occurred in A. guentheri leaves (0.212 mg/g crude drug). It is noteworthy that many of the species of Aristolochia reported in this study contain considerably higher amounts of aristolochic acids than A. fangchi, as present in the diet regimen consumed by the Belgian female cohort, which led to the development of urothelial carcinoma when ingested.33 A cumulative dose of 200 g of A. fangchi was reported previously to contain approximately 130 mg of aristolochic acid, and comparably low quantities of 1 were detected (0.609 mg/g crude drug).33 In comparison, levels of 1 as high as 2.467 mg/g crude drug were detected for A. maxima roots. This indicates that consumption of doses as low as 50 g of this plant material could pose a significant risk for developing UUC and AAN. C
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
Table 1. Summary of Toxicity Assessment of Aristolochia Species Extracts Using the Flow Cytometry-Based Micronucleus Test and the Sulforhodamine B Assaya no.
species
1 2 3 4 5 6 7 8 9 10 11 12 13
A. acuminata Lam. A. acuminata Lam. A. argentina Griseb. A. baetica L. A. californica Torr. A. chamissonis (Klotzsch) Duch. A. clematitis L. A. clematitis L. A. cymbifera Mart. A. debilis Siebold & Zucc. A. elegans Mast. A. elegans Mast. A. fangchi Y.C. Wu ex L.D. Chow & S.M. Hwang A. grandif lora Sw. A. guentheri O.C. Schmidt A. guentheri O.C. Schmidt A. labiata Willd. A. manshuriensis Kom. A. maurorum L. A. maxima Jacq. A. odoratissima L. A. paucinervis Pomel A. ringens Vahl A. rotunda L. A. tomentosa Sims A. trilobata L. A. westlandii Hemsl A. zollingeriana Miq.
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
plant part
IC50 [μg/mL] (SRB)b
IC50 [μg/mL] (MNT)b
MN induction (MNT)c
G2/M arrest (MNT)d
apoptosis (MNT)e
root flower stem leaf stem leaf seed root stem stem leaf root stem
182.6 >200 157.4 >200 >200 >200 47.8 163.3 >200 >200 >200 91.1 188.1
14.2 49.0 >200 >200 >200 >200 6.9 >200 78.1 >200 >200 12.5 >200
+++ + − − + − +++ − − − − +++ −
+ + − − + − +++ − − − − ++ +
+ + + − − − +++ − + + − ++ −
leaf leaf stem leaf stem leaf root leaf seed root root stem leaf stem leaf
127.2 >200 85.4 119.7 >200 >200 >200 >200 >200 106.0 >200 >200 >200 142.4 >200
7.2 18.2 3.5 >200 >200 >200 >200 >200 >200 64.2 >200 >200 30.2 163.1 36.8
++ + +++ − − − − − − − − − − + +
− − +++ − − − − − − − − − − − +
++ + + + − + + + − + − − + − +
a Information on the origin of the samples is given in Table S7, Supporting Information. b−: IC50 > 200 μg/mL, +: 100 μg/mL < IC50 < 200 μg/mL, ++: 50 μg/mL < IC50 < 100 μg/mL, +++: IC50 < 50 μg/mL. c−: no significant MN induction at 20 μg/mL; +: significant MN induction from 20 μg/ mL; ++: significant MN induction from 2 μg/mL; +++: significant MN induction from 0.2 μg/mL; d−: no significant G2 arrest at 20 μg/mL; +: significant G2 arrest from 20 μg/mL; ++: significant G2 arrest from 2 μg/mL; +++: significant G2 arrest from 0.2 μg/mL. e−: no significant increase in EMA positive events a 200 μg/mL; +: significant increase in EMA positive events a 200 μg/mL; ++: significant increase in EMA positive events from 20 μg/mL; +++: significant increase in EMA positive events from 2 μg/mL.
In Vitro Toxicity Assessment of Aristolochia Extracts. Table 1 summarizes the outcome of the toxicity assessment of 28 selected extracts of Aristolochia (Table S7, Supporting Information) on human kidney (HK-2) cells. Cytotoxicity was assessed using the sulforhodamine B (SRB) assay. To measure the genotoxic potential of the extracts, the flow cytometrybased micronucleus test (MNT) was carried out. In addition to micronuclei induction, G2 arrest and apoptosis were assessed. Several of the tested Aristolochia extracts were cytotoxic to HK2 cells [e.g., A. clematitis (seed), A. guentheri (stem), and A. grandif lora Sw. (leaf)]. Furthermore, an increase in micronuclei was observed for some of the extracts, including A. clematitis (seed), A. elegans (root), and A. acuminata (root). A large number of extracts caused G2/M phase arrest as well as apoptosis. Based on the in vitro toxicity assessment, the extracts were classified into different categories as described in Table 1 [very toxic (+++), moderately toxic (++), toxic (+), and nontoxic (−)]. These levels were used as dependent (Y) variables for the statistical analysis carried out. Correlation between Aristolochic Acid Analogue Content and Toxicity Using Univariate Regression Analysis. Univariate regression analysis was carried out in
order to analyze the correlation between AAA content and toxicity of the extracts. The correlation coefficients are shown in Figure 2. Values close to 1 indicated that there is a positive linear relationship, values close to −1 suggested that there is a negative linear relationship (anticorrelation), and values close to or equal to 0 suggested there is no linear relationship. Surprisingly, 1 and 2 peak areas did not correlate with the levels of cytotoxicity, micronuclei induction, G2/M arrest, or apoptosis, which were assessed in vitro. On the other hand, higher correlation coefficients between extract toxicity and other constituents such as 5, 8, 10, and 12 were observed. Correlation between Aristolochic Acid Analogue Content and Toxicity Using Multivariate Regression Analysis. Orthogonal projection to latent structures (OPLS) analysis was used to interpret the relationship between AAA peak areas of the extracts and their levels of toxicity. In contrast to univariate regression, where the correlation between only one X and one Y variable is assessed, OPLS is a method for relating two data matrices, X and Y within a linear multivariate model. In this study, the X matrix consists of 28 observations (Aristolochia extracts) and 27 variables (compounds), whereas the Y matrix consisted of only one variable (toxicity level). Five separate OPLS models were calculated, with different Y D
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
between the phytochemical profiles of species of Aristolochia and their toxicity. Using a high-resolution LC-MS- and 1H NMR-based metabolomic analysis we have shown that the majority of medicinally used species of Aristolochia contain a variety of AAAs and, therefore, should be considered nephrotoxic and carcinogenic. Factors (such as the part of the plant used and the age of the plant) that influence the phytochemistry of each plant and can consequently reduce or increase their toxicity were identified. By directly linking results from a metabolomic analysis with an in vitro assessment of cytotoxicity and genotoxicity, we were able to show that the assumption of the toxicity of species of Aristolochia being caused principally by 1 and 2 is incorrect. Other compounds (such as 5) contribute to their toxicity and need to be considered as potentially nephrotoxic agents. Previously, the toxicological risk of AAAs was mainly assessed using in vitro cytotoxicity studies. However, 1 and 2 have only moderate cytotoxicities in the majority of cell lines.27,34,35 In vivo, 2 has been shown to cause no nephrotoxic effects, but its carcinogenic potency is even stronger than that of 1.27 HK-2 cells, an immortalized cell line originating from the kidney proximal tubulus, were used as a cellular model in this study.36 This cell line has been used in previous studies29 since the proximal tubulus is the main tissue affected by Aristolochia toxicity. However, the use of in vitro models for assessing nephrotoxicity has a variety of limitations. Metabolic activation of aristolochic acids is well known to play a role in their biological effects and is an important prerequisite for the formation of aristolochic acid-derived DNA adducts. Since aristolactams may require metabolic activation only to a lesser extent, this could explain why in this study in vitro toxicity was linked primarily to aristolactams, not aristolochic acids. It is noteworthy that in vivo, highly cytotoxic compounds may even decrease the tumorigenic potential of 1 and 2 and prevent cancer development by inducing apoptosis or other forms of cell death. Therefore, a measure for genotoxicity in specific micronuclei induction was also included. Many of the extracts included in this study led to an increase in micronuclei formation and may be carcinogenic to kidney cells. The results of this study should be used as a guide for the selection of potential nephrotoxic compounds occurring in Aristolochia species apart from 1 and 2. Future research on other mechanisms involved in the nephrotoxicity and carcinogenicity (e.g., DNA adduct formation or oxidative stress) of AAAs apart from 1 and 2 is needed. The nephrotoxic effects of these compounds need to be further characterized in vitro and in vivo, to confirm their involvement in the etiology of AAN. In the future, herbal preparations need to be monitored for the presence of a wide range of AAAs in addition to 1 and 2. Taken together, the findings of this study facilitate understanding of the potential nephrotoxic mechanisms of species of Aristolochia and have significance for the regulation of herbal medicines. On a more general level, an innovative approach for determining the bioactive principles of complex mixtures, such as plant extracts, has been described.
Figure 2. Correlation coefficients between LC-MS peak areas and levels of cytotoxicity, genotoxicity, G2/M arrest, and apoptosis assessed in HK-2 cells (see Table 1).
variables depending on which in vitro toxicity outcome was analyzed. The number of components, R2X, R2Y, and Q2 values for the OPLS models are listed in Table S8, Supporting Information, and corresponding scores plots are shown in Figure S6, Supporting Information. The OPLS model loadings plots (Figure 3) were constructed to study whether the AAA peak areas correlated directly or indirectly, significantly or nonsignificantly, with the Y variables (toxicity levels). Interestingly, the amounts of aristolactam BI (5) correlated directly with the toxicity outcomes in all the OPLS models, with the exception of the model where the levels of apoptosis were defined as the Y variable. On the other hand, the levels of 4 and 12 correlated with the cytotoxicity levels as assessed using the SRB assay, while a strong correlation with the levels of 16 was observed for the OPLS model, where cytotoxicity levels were assessed using the micronucleus test. In terms of the correlation with micronuclei induction levels, 5, 12, and 16 were the main compounds predictive of the Y variable. A correlation of piperolactam E (17) and G2/M arrest was found in addition to a correlation with 5 and 1. It is noteworthy that the amount of 1 only correlated with G2/M arrest, but with none of the other outcomes. A correlation of 10 and apoptosis was found, as well as in the case of 12 and 16. The use of species of Aristolochia results in kidney damage and cancers of the urinary tract. Compounds 1 and 2 have been linked to these nephrotoxic and carcinogenic effects. Since a variety of AAAs are present in species of Aristolochia, their nephrotoxic effects are most likely not caused by 1 and 2 alone. Until this study, little was known about the relationship
■
EXPERIMENTAL SECTION
General Experimental Procedures. 1H NMR spectra were recorded on a Bruker Avance 500 spectrometer as previously described.10 The samples were dissolved in 490 μL of deuterated methanol (MeOD) containing 0.01% trimethylsilyl propanoic acid (TSP) and 210 μL of KH2PO4 buffer (pH = 6.0). High-resolution LCE
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
Figure 3. Principal component 1 (p1) loadings plots for OPLS models. The OPLS models were calculated by relating AAA peak areas of 28 Aristolochia extracts to their levels of (a) cytoxicity (SRB), (b) cytotoxicity (MNT), (c) MN induction, (d) G2/M arrest, and (e) apoptosis, as summarized in Table 1. Error bars represent 95% confidence intervals calculated by jackknifing [data mean-centered and Pareto scaled]. Plant Material. Aristolochia samples were collected mainly at the Royal Botanic Gardens (RBG), Kew, UK, and at the botanical garden at Dresden University of Technology (DUoT), Dresden, Germany (Table S1, Supporting Information). Samples from RBG Kew and DUoT were freeze-dried and stored at room temperature until further analysis. Additional samples (A. guentheri O.C. Schmidt and A.
MS and UV spectra were obtained using an LC-DAD-ESIMS system (Thermo Scientific, Waltham, MA, USA) consisting of an “Accela” liquid chromatograph with diode array detector and an “LTQ-Orbitrap XL” hybrid linear ion trap-orbitrap mass spectrometer, as previously described.10 F
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
Article
Cytotoxicity Assessment. Cytotoxicity of the plant extracts in HK-2 cells was carried out using the SRB assay as previously described.38 The cells were treated with test compounds for 72 h. The final concentration of dimethyl sulfoxide (DMSO, VWR, Poole, UK) in the medium did not exceed 1% v/v. All experiments were performed in triplicate wells and were repeated at least three times. The dose− response curve of relative cell viability was plotted to delineate the concentrations that inhibited cell growth to 50% (IC50 value). Colchicine (Sigma, St. Louis, MO, USA) was used as a positive control to ensure the performance of the assay. Flow Cytometry-Based Micronucleus Test. The flow cytometry-based micronucleus test was carried out as described in the In Vitro MicroFlow micronucleus analysis kit (Litron Laboratories, Rochester, NY, USA) instruction manual, with modifications. HK-2 cells (105 cells per well) were seeded in 24-well plates and were allowed to attach overnight. Afterward, the cells were treated with test solutions as indicated in the figure legends for 72 h. The final concentration of DMSO (VWR, Poole, UK) in the medium did not exceed 1% v/v. After treatment with the test compounds, HK-2 cells were stained using a dual-dye staining method with ethidium bromide monoazide (EMA, Molecular Probes, Eugene, OR, USA) and 4′,6diamidino-2-phenylindole (DAPI, Sigma). Micronuclei frequency, cytotoxicity, apoptosis, and cell cycle effects were assessed simultaneously. All experiments were repeated three times. Colchicine (Sigma) was used as a positive control to ensure the performance of the assay. Samples were analyzed with an MACSquant analyzer (Miltenyi Biotec, Bisley, UK) flow cytometer. Data acquisition and analysis were accomplished with MACSquantify software (Miltenyi Biotec) for acquiring flow cytometric data, including configuration of regions and gating logic. Briefly, DAPI-associated fluorescence emission was collected in the VioBlue channel (450/50), and EMA-associated fluorescence was collected in the PI/PE-Cy5.5 channel (655−730). Events were gated as shown in Figure S5 and Table S5, Supporting Information. The gates for each plot in Figure S5, Supporting Information, were specified as listed in Table S6, Supporting Information. Statistical Analysis. Data obtained from this study are expressed as means ± SD. Statistical analyses were performed using one-way analysis of variance (ANOVA) using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA, USA). Multivariate Data Analysis. PCA was carried out on the normalized and Pareto-scaled data set (for both, LC-MS and 1H NMR data sets) using the software SIMCA P+ (v. 12, Umetrics, Umea, Sweden). Univariate regression analysis was used to determine the correlation between AAA peak areas of the extracts (Figure 1) and their levels of toxicity as summarized in Table 1. The Y variable was defined as the level of cytotoxicity/micronuclei induction/G2 arrest/apoptosis ranging from 0 to 3, in accordance with the categorization used in Table 1. Correlation coefficients were determined using MATLAB (v. R2010a, The MathWorks, Inc., Natick, MA, USA). OPLS regression was carried out on the Pareto-scaled data set using SIMCA P+ (v. 12, Umetrics). OPLS was used to interpret the relationship between AAA peak areas of the extracts and their levels of toxicity summarized in Table 1. The goodness-of-fit of the OPLS model was determined by the squared correlation coefficient (R2Y) and the goodness-ofprediction by the 7-fold cross-validated R2Y (Q2).
lagesiana Ule) were collected in Ecuador and identified by Dr. Rocio Alcaron (Market Harborough, UK). A. indica L. samples were collected in Bangladesh (by J.M.) and were identified by Dr. Martin Ingrouille. Further samples were obtained from the Chinese Medicinal Plants Authentication Centre at the RBG, Kew. These samples were identified by Dr. Christine Leon. Extraction of the Plant Material. All samples were frozen and lyophilized, then ground in a mortar. Then, 50 mg of dried plant material was extracted with 1 mL of 70% aqueous methanol. After a 14 h extraction at room temperature the extracts were centrifuged at 10 000 rpm for 10 min. The supernatant liquid was filtered using Whatman 0.45 μm polytetrafluoroethylene filters. Next, 700 μL of the sample solution was used for HPLC-DAD-ESIMS measurements. The samples 40 [A. indica (leaf, BI No. 21583)], 59 [A. maxima (root, BI No. 19209)], 80 [A. trilobata (leaf, BI No. 20832)], and 82 [A. zollingeriana Miq. (leaf, BI No. 20192)] were extracted and analyzed in triplicate to demonstrate reproducibility of the extraction method. Chemicals. Reference standards for 1 and 2 were isolated from Aristolochia repens. 10 was purchased from Phytolab (Vestenbergsgreuth, Germany). Compound 4 was isolated from Asarum sieboldii Miq. in our group.37 The purities of all reference standards were determined using HPLC. The compounds were characterized using high-resolution mass spectrometry, 1D NMR (1H NMR, 13C NMR, 135-DEPT), and 2D NMR (1H 1H COSY, 1H 1H NOESY, 1H 13C HMQC, and 1H 13C HMBC) spectroscopy and their UV spectra. Preparation of the Standard Solutions. For this purpose, 2 mg of 1 (100%), 1.83 mg of 2 (98.82%), 1.12 mg of a 10 + 8 mixture (81.46% 10, 11.63% 8), and 1.24 mg of 4 (96.43%) were dissolved using 5 mL of methanol in volumetric flasks. The mixed standard stock solution was prepared by accurately adding 2.5 mL of 1 stock solution, 2.732 mL of 2 stock solution, 1.785 mL of 10 + 8 stock solution, and 0.806 mL of 4 stock solution into a 10 mL volumetric flask and diluting it to 10 mL with methanol. A series of working standard solutions was prepared by transferring an appropriate volume of the mixed standard stock solution into 5 mL volumetric flasks and diluting them with methanol. Concentrations of the mixed standard solutions are shown in Table S2, Supporting Information. Metabolomics Data Processing. The 1H NMR spectra in the range 0.1−10 ppm were divided into 990 regions of 0.01 ppm using AMIX software (v. 3.5.5, BrukerBiospin). The spectral areas were normalized to the total sum of the spectral integral. Filtering, peak extraction, chromatogram deconvolution, and peak alignment of the high-resolution LC-MS data were performed automatically using Mzmine 2 (http://mzmine.sourceforge.net), as previously described.10 The resulting data set was normalized to the total raw signal. Metabolite Identification Assignment. For the peaks extracted by Mzmine 2, each ion species (i.e., adduct) was determined so as to obtain an experimental accurate molecular mass. Compounds 1, 2, 3, 4, 8, 10, 12, 14, and 15 were identified by comparison with reference standards. Further compounds were identified by comparing their accurate mass data to an in-house database consisting of 178 aristolochic acid analogues. Putative assignments for such AAAs were made based on their accurate mass, retention time, mass fragmentation, and UV spectrum. The accurate mass of all detected base ions was compared to the mass of the [M + NH4]+ ion for aristolochic acids and [M + H]+ ions for aristolactams using the custom database search function in Mzmine2. The identity of potential candidate peaks was further confirmed by the presence of a UV maximum at 390 nm, a UV absorption characteristic for AAAs. Furthermore, the MS2 and MS3 mass fragmentation spectra for each peak were examined for the presence of characteristic losses, such as the [NH3] fragment for aristolochic acids. Cell Culture Conditions. HK-2 cells (American Type Culture Collection, Manassas, VA, USA) were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin (all obtained from Gibco Invitrogen, Paisley, UK). All cells were cultured in a humidified atmosphere with 5% CO2 at 37 °C. For routine cultivation, cells were maintained up to 80% of confluence and the medium was replaced every 3 days.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jnatprod.5b00556. Additional information on the plant material and experimental procedures and results of the quantitative analysis of aristolochic acid contents as well as details on the statistical models used (PDF) G
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products
■
Article
(20) Baudoux, T. E.; Pozdzik, A. A.; Arlt, V. M.; De Prez, E. G.; Antoine, M. H.; Quellard, N.; Goujon, J. M.; Nortier, J. L. Kidney Int. 2012, 82, 1105−1113. (21) Nortier, J. L.; Schmeiser, H. H.; Muniz Martinez, M. C.; Arlt, V. M.; Vervaet, C.; Garbar, C. H.; Daelemans, P.; Vanherweghem, J. L. Nephrol. Dial. Transplant. 2003, 18, 426−428. (22) Zhou, L.; Fu, P.; Huang, X. R.; Liu, F.; Lai, K. N.; Lan, H. Y. J. Am. Soc. Nephrol. 2010, 21, 31−41. (23) Zhu, S.; Wang, Y.; Jin, J.; Guan, C.; Li, M.; Xi, C.; Ouyang, Z.; Chen, M.; Qiu, Y.; Huang, M.; Huang, Z. Toxicol. In Vitro 2012, 26, 663−671. (24) Rui, H. L.; Wang, Y. Y.; Cheng, H.; Chen, Y. P. Am. J. Physiol. 2012, 302, F1569−1575. (25) Shibutani, S.; Dong, H.; Suzuki, N.; Ueda, S.; Miller, F.; Grollman, A. P. Drug Metab. Dispos. 2007, 35, 1217−1222. (26) Sato, N.; Takahashi, D.; Chen, S. M.; Tsuchiya, R.; Mukoyama, T.; Yamagata, S.; Ogawa, M.; Yoshida, M.; Kondo, S.; Satoh, N.; Ueda, S. J. Pharm. Pharmacol. 2004, 56, 221−229. (27) Michl, J.; Ingrouille, M. J.; Simmonds, M. S.; Heinrich, M. Nat. Prod. Rep. 2014, 31, 676−693. (28) Li, B.; Li, X. M.; Zhang, C. Y.; Wang, X.; Cai, S. Q. Beijing Da Xue Xue Bao Yi Xue Ban 2004, 36, 36−40. (29) Wen, Y. J.; Su, T.; Tang, J. W.; Zhang, C. Y.; Wang, X.; Cai, S. Q.; Li, X. M. Nephron 2006, 103, 95−102. (30) El Tahir, K. E. H. Pharm. Biol. 1991, 29, 101−110. (31) Wang, C.-C.; Chen, S.; Wu, T.-S. J. Chin. Chem. Soc. 2003, 50, 261−266. (32) Pistelli, L.; Nieri, E.; Bilia, A. R.; Marsili, A.; Scarpato, R. J. Nat. Prod. 1993, 56, 1605−1608. (33) Nortier, J. l. L.; Martinez, M.-C. M.; Schmeiser, H. H.; Arlt, V. M.; Bieler, C. A.; Petein, M.; Depierreux, M. F.; De Pauw, L.; Abramowicz, D.; Vereerstraeten, P.; Vanherweghem, J.-L. N. Engl. J. Med. 2000, 342, 1686−1692. (34) Li, J.; Zhang, L.; Jiang, Z.; Shu, B.; Li, F.; Bao, Q.; Zhang, L. Toxicol. In Vitro 2010, 24, 1092−1097. (35) Zhang, C. Y.; Wang, X.; Su, T.; Ma, C. M.; Wen, Y. J.; Shang, M. Y.; Li, X. M.; Liu, G. X.; Cai, S. Q. Die Pharmazie 2005, 60, 785−788. (36) Jenkinson, S.; Chung, G.; van Loon, E.; Bakar, N.; Dalzell, A.; Brown, C. A. Pfluegers Arch. 2012, 464, 601−611. (37) Li, Y.; Tian, M.; Yu, J.; Shang, M.; Cai, S. J. Nat. Med. 2010, 64, 442−451. (38) Houghton, P.; Fang, R.; Techatanawat, I.; Steventon, G.; Hylands, P. J.; Lee, C. C. Methods 2007, 42, 377−387.
AUTHOR INFORMATION
Corresponding Author
*Tel: +44 (0) 20 7753 5844. Fax: +44 20 77535909. E-mail: m.
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank C. Leon (Chinese Medicinal Plant Authentication Centre, Royal Botanic Gardens, Kew, Richmond, UK) for providing some of the Aristolochia samples used in this study. We are also thankful to the Bloomsbury Colleges and the London International Development Centre (LIDC) for funding this project.
■
REFERENCES
(1) Hoang, M. L.; Chen, C. H.; Sidorenko, V. S.; He, J.; Dickman, K. G.; Yun, B. H.; Moriya, M.; Niknafs, N.; Douville, C.; Karchin, R.; Turesky, R. J.; Pu, Y. S.; Vogelstein, B.; Papadopoulos, N.; Grollman, A. P.; Kinzler, K. W.; Rosenquist, T. A. Sci. Transl. Med. 2013, 5, 197− 102. (2) Chen, C. H.; Dickman, K. G.; Moriya, M.; Zavadil, J.; Sidorenko, V. S.; Edwards, K. L.; Gnatenko, D. V.; Wu, L.; Turesky, R. J.; Wu, X. R.; Pu, Y. S.; Grollman, A. P. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 8241−8246. (3) Dietz, B.; Bolton, J. L. Chem. Res. Toxicol. 2007, 20, 586−590. (4) Vanherweghem, L. J. J. Altern. Complement. Med. 1998, 4, 9−13. (5) Debelle, F.; Muniz-Martinez, M.-C.; Vanherweghem, J.-L.; Nortier, J. In Clinical Nephrotoxins; Broe, M. E. D., Ed.; Springer: New York, NY, 2008; pp 757−769. (6) Debelle, F. D.; Vanherweghem, J. L.; Nortier, J. L. Kidney Int. 2008, 74, 158−169. (7) Aristolochic Acids. Report on Carcinogens: Carcinogen Profiles; Deptartment of Health and Human Services, Public Health Service, National Toxicology Program: Research Triangle Park, NC, USA, 2011; Vol. 12, pp 45−49. (8) Schaneberg, B. T.; Applequist, W. L.; Khan, I. A. Die Pharmazie 2002, 57, 686−689. (9) Heinrich, M.; Chan, J.; Wanke, S.; Neinhuis, C.; Simmonds, M. S. J. Ethnopharmacol. 2009, 125, 108−144. (10) Michl, J.; Jennings, H. M.; Kite, G. C.; Ingrouille, M. J.; Simmonds, M. S.; Heinrich, M. J. Ethnopharmacol. 2013, 149, 235− 244. (11) Grollman, A. P.; Shibutani, S.; Moriya, M.; Miller, F.; Wu, L.; Moll, U.; Suzuki, N.; Fernandes, A.; Rosenquist, T.; Medverec, Z.; Jakovina, K.; Brdar, B.; Slade, N.; Turesky, R. J.; Goodenough, A. K.; Rieger, R.; Vukelic, M.; Jelakovic, B. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 12129−12134. (12) Jelakovic, B.; Karanovic, S.; Vukovic-Lela, I.; Miller, F.; Edwards, K. L.; Nikolic, J.; Tomic, K.; Slade, N.; Brdar, B.; Turesky, R. J.; Stipancic, Z.; Dittrich, D.; Grollman, A. P.; Dickman, K. G. Kidney Int. 2012, 81, 559−567. (13) De Broe, M. E. Kidney Int. 2012, 81, 513−515. (14) Grollman, A. P. Environ. Mutagen. 2013, 54, 1−7. (15) Hu, S. L.; Zhang, H. Q.; Chan, K.; Mei, Q. X. Toxicology 2004, 198, 195−201. (16) Grollman, A. P.; Scarborough, J.; Jelaković, B. In Advances in Molecular Toxicology; James, C. F., Ed.; Elsevier: Waltham, MA, 2009; Vol. 3, Chapter 7, pp 211−227. (17) Balachandran, P.; Wei, F.; Lin, R. C.; Khan, I. A.; Pasco, D. S. Kidney Int. 2005, 67, 1797−1805. (18) Priestap, H. A.; Torres, M. C.; Rieger, R. A.; Dickman, K. G.; Freshwater, T.; Taft, D. R.; Barbieri, M. A.; Iden, C. R. Chem. Res. Toxicol. 2011, 25, 130−139. (19) Yun, B. H.; Rosenquist, T. A.; Sidorenko, V.; Iden, C. R.; Chen, C.-H.; Pu, Y.-S.; Bonala, R.; Johnson, F.; Dickman, K. G.; Grollman, A. P.; Turesky, R. J. Chem. Res. Toxicol. 2012, 25, 1119−1131. H
DOI: 10.1021/acs.jnatprod.5b00556 J. Nat. Prod. XXXX, XXX, XXX−XXX
