Assessment of the Cytotoxicity, Mutagenicity, and Genotoxicity of Two Traditional Chinese Herbs: Aristolochia baetica and Magnolia officinalis

Herbal remedies used in traditional medicine often contain several compounds combined in order to potentiate their own intrinsic properties. However, herbs can sometimes cause serious health troubles. In Belgium, patients who developed severe aristolochic acid nephropathy ingested slimming pills containing root extracts of an Aristolochia species, as well as the bark of Magnolia officinalis. The goal of the study was to evaluate, on a human renal cell line, Aristolochia and Magnolia extracts for their cytotoxicity by a resazurin cell viability assay, and their genotoxicity by immunodetection and quantification of the phosphorylated histone γ-H2AX. The present study also sought to assess the mutagenicity of these extracts, employing an OECD recognized test, the Ames test, using four Salmonella typhimurium strains with and without a microsomial fraction. Based on our results, it has been demonstrated that the Aristolochia–Magnolia combination (aqueous extracts) was more genotoxic to human kidney cells, and that this combination (aqueous and methanolic extracts) was more cytotoxic to human kidney cells after 24 and 48 h. Interestingly, it has also been shown that the Aristolochia–Magnolia combination (aqueous extracts) was mutagenic with a TA98 Salmonella typhimurium strain in the presence of a microsomial liver S9 fraction. This mutagenic effect appears to be dose-dependent.


Introduction
Following disappointment with allopathic treatments, patients may turn to other therapies such as phytotherapy. Phytotherapy is increasingly considered by the general public as a safe and reliable therapy. Moreover, plant derivatives (such as Magnolia officinalis or herbal medicinal products that contain aristolochic acids) are often readily available either in supermarkets or on the internet, and sometimes the products' degree of purity is not suitable for pharmaceutical use [1]. For example, mycotoxins, mainly Fusarium toxins, are naturally present in several plants that are widely used as feedstuffs worldwide [2,3]. Although often perceived as innocuous by the general public, many herbs harbor phytochemicals that are either directly reactive towards DNA, or likely to disturb cellular homoeostasis, cell cycle, and/or genome maintenance mechanisms; this may translate into genotoxicity, carcinogenicity, or co-carcinogenicity [4]. When taken orally (in the form of pills, tea, syrup, etc.), plants are excreted by the kidneys. Indeed, the excretory organs, such as the kidney and bladder, are often the sites of toxicity. Moreover, kidney diseases often lead to an increased risk of morbidity from cardiovascular diseases [5,6].
Genotoxicity refers to the deleterious effect of a chemical compound or a physical event on the genetic material; such genotoxic events are considered hallmarks of cancer risk [4].
In the early 1990s, an epidemic of rapidly progressive tubulointerstitial nephritis was reported in young Belgian female patients who inadvertently took slimming pills containing the bark of Magnolia officinalis Rehder & E. H. Wilson, and inadvertently the roots of an Aristolochia species [7][8][9].
At that time, the causative nephrotoxic agent was identified as aristolochic acid (AA), and the renal disease following their ingestion is now worldwide recognized as Aristolochic Acid Nephropathy (AAN) [10,11]. Remarkably, despite the worldwide human consumption of AA, the Materia Medica and Pharmacopoeia before 1995 failed to mention any of the toxic effects of Aristochia spp. [12].
AA was classified as a human carcinogen Class I by the World Health Organization International Agency for Research on Cancer in 2002 [13]. The nephrotoxic effect of aristolochic acid is irreversible. AAN and associated upper urinary tract urothelial carcinoma and bladder cancer may become a major public health threat in the next few years [4,14].
However, despite Aristolochia species prohibition in many countries, more cases of AA intoxication have been reported all over the world, especially in Asia and in Balkan countries [9,15]. For example, there is accumulating evidence that Balkan endemic nephropathy (BEN) is an environmental disease caused by AAs released from the decomposition of Aristolochia clematitis L., an AA-containing weed that grows abundantly in the Balkan Peninsula [16,17].
Nowadays, herbal medicinal products that contain AA continue to be manufactured and marketed worldwide with inadequate regulation, and possible environmental exposure routes receive little attention [1].
In 2015, Nachtergael et al. showed the effect of M. officinalis on Aristolochia genotoxicity. The high potentiation of AA genotoxicity by M. officinalis can be tentatively explained by an increased metabolic activation into aristolactams [18].
In the present study, Magnolia effects in the AAN case were further investigated. Following the nephrotoxicity of AA compounds, studies were performed in HK-2 kidney proximal human tubule cells.
The goal of the present study was to evaluate, on a renal cell line, Aristolochia and Magnolia aqueous and methanolic extracts, alone and in combination, for their cytotoxicity by a resazurine cell viability assay, and their genotoxicity by immunodetection and quantification of the phosphorylated histone γ-H2AX, a genotoxicity biomarker [24,25]. The present study also sought to assess the mutagenicity of these compounds, using an OECD recognized test, the Ames test, using four Salmonella typhimurium strains-TA98, TA100, TA1535, and TA1537-with and without the S9 microsomal fraction, often used to mimic mammal metabolism and to estimate the need for bioactivation [26]. The thin-layer chromatography of raw plant material and the aqueous extract of A. baetica was used according to the European Pharmacopoeia 8.4 monograph "Test for aristolochic acids in herbal drugs," previously detailed by Nachtergael et al. In brief, the following was used: an HPTLC Silica gel 60 plate F254, a mobile phase of anhydrous formic acid, water, ethyl acetate, and toluene (3:3:30:60, v/v/v/v), an upper layer detection method using 100 g/L Tin(II) chloride in diluted hydrochloric acid at 100 • C for 1 min at UV365 nm, as well as aristolochic acid I and II (Rf 0.46 and Rf 0.54) [18].

Magnolia officinalis Rehder & E.H. Wilson
Alkaloid and lignan detection was made using HPTLC-MS. As described by Estevez et al., mass spectrometry can indeed be used to characterize and identify components from toxins [27].

Alkaloids
Following several protocols, the most appropriate mobile phase was selected for the M. officinalis extracts.
Each plate was revealed using a Draggendorff reagent. We found that, following Japanese Pharmacopoeia XVI, the most suitable mobile phase composition was butanol/water/acetic acid (at a proportion of 4/2/1).
Both lignans, magnolol and honokiol, and almost 15 different alkaloids were detected in methanolic and aqueous extracts (see Table 1).

Cytotoxicity of the Plant Extract Alone or in Combination
The cytotoxicity of Aristolochia and Magnolia aqueous and methanol extracts was investigated on HK-2 cells by following resazurin viability curves after 24 and 48 h of exposure to each material alone, or in combination. Results are presented in Figure 1 and Table 2.
Both lignans, magnolol and honokiol, and almost 15 different alkaloids were detected in methanolic and aqueous extracts (see Table 1).

Cytotoxicity of the Plant Extract Alone or in Combination
The cytotoxicity of Aristolochia and Magnolia aqueous and methanol extracts was investigated on HK-2 cells by following resazurin viability curves after 24 and 48 h of exposure to each material alone, or in combination. Results are presented in Figure 1 and Table  2.

After 24h
After 48h   One-way ANOVA (Dunnett's multiple comparison test; p < 0.01; n = 6) confirmed that the viability more significantly decreased with the Magnolia than the Aristolochia extracts, after 24 and 48 h of exposure. The combination of both plant extracts resulted in a more significant deleterious effect on cell viability.

Genotoxicity throughout γ-H2AX Detection
The measurement of the phosphorylated histone γ-H2AX is increasingly considered as an attractive biomarker for either DNA damage or DNA repair. H2AX is a eukaryotic histone protein that, depending on the organism and cell type, constitutes 2-25% of the mammalian histone H2A. Interestingly, this histone is phosphorylated on its 139th serine residue in the presence of DNA damage, mainly a double-strand break, to yield γ-H2AX. The latter form nuclear domains, named "DNA damage foci", can be visualized cytologically using fluorescence microscopy [18,24,25]. γ-H2AX detection by immunofluorescence in HK-2 cells demonstrated that treating cells with bleomycin for 24 and 48 h induced an increase in γ-H2AX foci at all tested concentrations: 50, 100, and 200 µg/mL (data not shown). By comparison with these positive controls, cells treated with the aqueous extract at a 1 mg/mL concentration of a Aristolochia-Magnolia combination (1:1 ratio) demonstrate an increase in γ-H2AX foci, while the Aristolochia or the Magnolia aqueous extracts used alone did not induce any significant increase in γ-H2AX foci ( Figure 2). Cells treated with a combination of Aristolochia and Magnolia aqueous extracts at 0.25 and 0.5 mg/mL did not significantly increase the number of γ-H2AX foci (data not shown).
γ-H2AX fluorescence detection has been quantified by measuring total fluorescence and total fluorescence relative to the number of nuclei. Statistical significance could not be demonstrated for extracts used alone. However, statistical significance could be highlighted by using a combination of the two plants (see Figure 3).
We further investigated which component could explain this increasing genotoxicity. We performed the same experiment with aristolochic acids (I and II), magnolol, honokiol, magnoflorin, and a combination of all of them over 24 h and 48 h exposure. Unfortunately, we did not observe any increase in the presence of γ-H2AX after these different experimental conditions (Figure 4-data not shown for 48 h).

Genotoxicity and Mutagenicity throughout the Ames Test
An Ames test was run using four strains of histidine-deficient Salmonella typhimurium: TA-100, TA-98, TA-1535, and TA-1537, with and without the S9 fraction. The number of revertants was recorded for each sample and compared to a positive control.
With the TA-100 strain, A. baetica seems to be mutagenic at every concentration from 0.25 mg/mL, without the S9 microsomal fraction. In the presence of the S9 fraction, only the concentration of 10 mg/mL was mutagenic. The TA-1535 strain, on its side, does not seem to be sensitive to Aristolochia or Magnolia extracts (data not shown). A statistically significant increase in the mutagenicity induced by Aristolochia extracts alone, as well as the Aristolochia-Magnolia combination, was found with the strain TA-98, but only with the S9 microsomial liver fraction. The TA-98 strains reveal frameshift mutations, meaning that the genotoxicity is based on a genetic mutation caused by the indels (insertions or deletions) of a number of nucleotides in a DNA sequence. Moreover, this mutagenic effect appears to be dose-dependent (Table 3).  Representative immunofluorescence images obtained from control HK-2 cells and cells exposed for over 24 h to aqueous plant extracts alone or in combination, and to 200 µg/mL bleomycin (20x magnification, 1 cm = 500 µm, n = 4).
γ-H2AX fluorescence detection has been quantified by measuring total fluorescence and total fluorescence relative to the number of nuclei. Statistical significance could not be demonstrated for extracts used alone. However, statistical significance could be highlighted by using a combination of the two plants (see Figure 3). γ-H2AX fluorescence detection has been quantified by measuring total fluorescence and total fluorescence relative to the number of nuclei. Statistical significance could not be demonstrated for extracts used alone. However, statistical significance could be highlighted by using a combination of the two plants (see Figure 3).

Genotoxicity and Mutagenicity throughout the Ames Test
An Ames test was run using four strains of histidine-deficient Salmonella typhimurium: TA-100, TA-98, TA-1535, and TA-1537, with and without the S9 fraction. The number of revertants was recorded for each sample and compared to a positive control.
With the TA-100 strain, A. baetica seems to be mutagenic at every concentration from 0.25 mg/mL, without the S9 microsomal fraction. In the presence of the S9 fraction, only the concentration of 10 mg/mL was mutagenic. The TA-1535 strain, on its side, does not seem to be sensitive to Aristolochia or Magnolia extracts (data not shown). A statistically significant increase in the mutagenicity induced by Aristolochia extracts alone, as well as the Aristolochia-Magnolia combination, was found with the strain TA-98, but only with the S9 microsomial liver fraction. The TA-98 strains reveal frameshift mutations, meaning that the genotoxicity is based on a genetic mutation caused by the indels (insertions or deletions) of a number of nucleotides in a DNA sequence. Moreover, this mutagenic effect appears to be dose-dependent (Table 3).

Discussion
The bark of M. officinalis Rehder & Wilson, known under the pinyin name "Hou Po," has been traditionally used in Chinese and Japanese medicines for the treatment of anxiety, asthma, depression, gastrointestinal disorders, and headaches. Many pharmacological activities, including antioxidant, anti-inflammatory, antibiotic, and antispasmodic effects [20], as well as its alleviating effect on depression in postnatal women [28], have been reported for this herb and its major compounds. Many of these activities have been attributed to the lignans magnolol and honokiol, two major constituents of the plant.
In the present in vitro study, we first demonstrated that the M. officinalis aqueous and methanolic extracts show a higher cytotoxicity after 24 and 48 h than the aqueous and methanolic extracts of Aristolochia baetica. Combining both plant extracts resulted in the most deleterious effects. However, we were not able to identify which component association actually increased the generation of γH2AX foci. Therefore, it seems that lignans (i.e., Magnolol and Honokiol) and the tested alkaloid magnoflorin are not the agents responsible for the increased toxicity of Aristolochia. Considering the characterization of our Magnolia officinalis extracts, we can hypothesize that many other components could be candidates for these effects, mainly alkaloid compounds.
Concerning the mutagenic assessment of our extracts, we performed the Ames test. The Ames test is one of the most frequently applied tests in toxicology [29]. Following the OECD guidelines, almost all new pharmaceutical substances and chemicals used in the industry are tested by this assay [30]. The Ames test, or the so-called Salmonella/microsome test, is widely used in investigating the mutagenic effects of chemicals. Not only is it one of the most reliable short-term bacterial test systems, but it is also cheap and provides results rapidly [31,32]. Each herbal extract was tested on four different strains. TA100 and TA1535 were used to detect base substitution mutations, and TA98 and TA1537 were used to detect frameshift mutations. The liver S9 fraction is useful for imitating the mammalian metabolism that activates pro mutagens [26,[31][32][33].
The Salmonella strain TA-98, in the presence of the S9 fraction, showed that the Magnolia and Aristolochia aqueous extract combination is more mutagenic than both aqueous extracts tested separately. This result suggests that the combination generates frameshift mutations and that this effect is dose-dependent. The need for the S9 fraction proves that the combination of plant extracts has to be metabolized by hepatic enzymes to be mutagenic. This pathway is related to the fact that AAN patients ingested root extracts as slimming pills. After oral intake, these pills, following the pharmacokinetic features, were metabolized by the liver before targeting the kidneys where they generated their toxicity.
Although AA might directly cause nephropathy, the enzymatic activation of AAI is required to exert its genotoxic effect. The reduction of the nitro group is now considered to be the major activation pathway of AA [16]. NAD(P)H:quinone oxidoreductase (NQO1) and microsomal enzyme NADPH:cytochrome P450 oxydoreductase are the main enzymes responsible for the metabolic activation of AA in aristolactams by reduction of the nitro group, and they may be involved in the bioactivation of AA, but its exact roles are still a matter of debate [16,[34][35][36][37]. Interestingly, in 2013, Cui et al. showed in a mouse model that kidney NQO1 was significantly increased following treatment with an ethanolic extract of Magnolia officinalis bark [38].
It has been recently demonstrated, for the first time, that cell lines deficient in nucleotide excision repair (NER) machinery accumulated higher adduct levels, indicating that NER is the major mechanism responsible for the repair of these lesions [39].
Following our results, the high potentiation of AA genotoxicity by Magnolia officinalis could then be tentatively explained by an increased metabolic activation into aristolactams. More investigations are needed to determine which components could be responsible for the increase in DNA damage.

Conclusions
Plants are interesting from a health point of view but can be toxic. Moreover, these toxicities are not always direct but can occur months or years later. This further complicates the identification of the origin of these adverse reactions. This study demonstrated that Magnolia officinalis, often described as safe, seems to be relatively cytotoxic, mainly to kidney cells, but also seems to have antibacterial activity (significantly cytotoxic and mutagenic to Salmonella typhimurium TA-100 strains at 10 mg/mL). Moreover, the genotoxicity to the kidney of the association of Aristolochia baetica and Magnolia officinalis extracts is significantly higher compared to the individual plant extracts. This was also confirmed by an Ames test, an OECD-recommended mutagenicity assessment assay. The high potentiation of AA genotoxicity by M. officinalis could be an explanatory factor for Chinese herb nephropathy cases observed in Belgium in the 1990s. Furthermore, plants are, in most cases, used in combination rather than alone. This study proves that plant combinations can be more toxic under specific conditions.

Herbal Extraction
As previously performed by Nachtergael et al. [18], dried Magnolia officinalis Rehder & E. H. Wilson cortices were obtained from Phytax (Schlieren, Switzerland). A certificate of analysis was obtained from the company, indicating a sample free from aflatoxins (detection limit: 0.4 µg/kg) and heavy metals (detection limit: 10 µg/kg). Instead of the Aristolochia fangchi Y. C. Wu ex L. D. Chow & S. M. Hwang from the initial prescription, now prohibited in Belgium, Aristolochia baetica L. radix was used in the present study. It is deposited at the National Herbarium of Morocco (Scientific Institute of Rabat) under the voucher name RAB 78463. Further investigations proved that A. baetica L. also contains quite similar aristolochic acid composition [18]. Each herbal extract has been prepared using polar solvents (water and methanol) ( Table 4). Following the Japanese Pharmacopoeia, 13th Edition (1996), each plant was macerated in a known volume of methanol for 24 h in the dark. Plants were mixed with a Polymix PX-MFC 90 D mixer (1.0 mm diameter). The solvent from the resulting mixture was harvested and removed from the extract by the rotary evaporator (Rotavapor R-210, Vacuum pump V-700, Vacuum controller V-850, Heating Bath B-491).

Extract Characterizations
The raw and lyophilized herbal materials and the methanolic extracts were analyzed by HPTLC thin-layer chromatography according to the European Pharmacopoeia 8.4 monographs "Magnolia officinalis bark" and "Test for aristolochic acids in herbal drugs." Comparable amounts of raw herbs and extracts (calculated from the lyophilization yield) were applied on HPTLC plates (HPTLC silica gel 60 F254 glass plates).
Preliminary cytotoxicity assays were performed by Bunel et al. [12] to determine the working doses of each compound. Based on their recognized cytotoxicity in vivo, AA was 15 1.5% 12.1%

Aristolochia baetica L. Radix
Following the Japanese Pharmacopoeia, 13th Edition (1996), each plant was macerated in a known volume of methanol for 24 h in the dark. Plants were mixed with a Polymix PX-MFC 90 D mixer (1.0 mm diameter). The solvent from the resulting mixture was harvested and removed from the extract by the rotary evaporator (Rotavapor R-210, Vacuum pump V-700, Vacuum controller V-850, Heating Bath B-491).

Extract Characterizations
The raw and lyophilized herbal materials and the methanolic extracts were analyzed by HPTLC thin-layer chromatography according to the European Pharmacopoeia 8.4 monographs "Magnolia officinalis bark" and "Test for aristolochic acids in herbal drugs." Comparable amounts of raw herbs and extracts (calculated from the lyophilization yield) were applied on HPTLC plates (HPTLC silica gel 60 F254 glass plates).
Preliminary cytotoxicity assays were performed by Bunel et al. [12] to determine the working doses of each compound. Based on their recognized cytotoxicity in vivo, AA was 45 23.4% 23.5% For both aqueous and methanolic extracts, the maximal concentration tested was 1 mg/mL. Plant combinations were assessed in a 1:1 ratio. Indeed, the initial prescription The common dosages for Magnolia officinalis and Aristolochia species are 3-9 g and 4.5-9 g, respectively [40]. A study showed that, following a 240 mL dose of water, the intestinal liquid volume ranges from 67 ± 17 to 93 ± 24 mL, respectively, 2 and 45 min after ingestion [41]. The liquid volume is divided into water pockets, where 60% of the total volume is contained in a small number of large pockets (>20 mL), and 40% is contained in a large number of the smallest pockets (0.5-20 mL). Regarding the 100-200 mg of plants ingested in the Belgian cohort patients, the concentration reached in the small intestine can be coarsely estimated at around 100-200 mg/100 mL, or 1-2 mg/mL, for each plant. The concentration used in the present study is in line with the estimated concentration of plants in the small intestine after the ingestion of slimming pills.

Aqueous Extraction
Each material was decocted in water, following the traditional instructions in the Chinese Materia Medica [40]. The decoction was filtered on cellulose 3 times, and the filtrate was lyophilized (Heto PowerDry LL1500, Thermo Fisher Scientific Inc., Waltham, MA, USA) and stored at −20 • C until further use. The extracts were dissolved and diluted with a complete culture medium to the required concentrations.

Methanolic Extraction
Following the Japanese Pharmacopoeia, 13th Edition (1996), each plant was macerated in a known volume of methanol for 24 h in the dark. Plants were mixed with a Polymix PX-MFC 90 D mixer (1.0 mm diameter). The solvent from the resulting mixture was harvested and removed from the extract by the rotary evaporator (Rotavapor R-210, Vacuum pump V-700, Vacuum controller V-850, Heating Bath B-491).

Extract Characterizations
The raw and lyophilized herbal materials and the methanolic extracts were analyzed by HPTLC thin-layer chromatography according to the European Pharmacopoeia 8.4 monographs "Magnolia officinalis bark" and "Test for aristolochic acids in herbal drugs." Comparable amounts of raw herbs and extracts (calculated from the lyophilization yield) were applied on HPTLC plates (HPTLC silica gel 60 F254 glass plates).
Preliminary cytotoxicity assays were performed by Bunel et al. [12] to determine the working doses of each compound. Based on their recognized cytotoxicity in vivo, AA was used at IC25 (≈50 µM); maximum concentrations of magnolol and honokiol were the highest non-lethal doses (10 µM).

Cell Viability Assay by a Resazurin Assay
As described by Bunel et al. [12], wells were treated with test compounds in 96-well plates, washed twice with PBS, and assessed for their viability (metabolic activity) by incubation with 0.44 mM resazurin solution (Sigma-Aldrich, St. Louis, MI, USA) for 1.5 h at 37 • C. Absorbances were measured at wavelengths 540 and 620 nm and percentages of reduced dye were calculated with the following formula: where εOX is the molar extinction coefficient of resazurin (47.6 at 540 nm and 34.8 at 620 nm); εRED is the molar extinction coefficient of resorufin (104.4 at 540 nm and 5.5 at 620 nm); A is the absorbance of the test wells; A' is the mean absorbance of the blank wells; λ1 = 540 nm; λ2 = 620 nm. Metabolic activity was normalized against the control condition.

IF 96-Well on the HK-2 Kidney Cell Line
Cells were seeded in 96-well plates (10,000 cells per well) and grown at 37 • C for 24 h before plant extracts, AA, lignans (both or separately), magnoflorine, or a positive (bleomycin, 200 µg/mL) or negative control (supplemented medium) were added.
After 24 h or 48 h, the cells were washed with PBS for 5 min, fixed with 100 µL/well of 4% PFA for 20 min, washed 4 times consecutively with tap water, distilled water, PBS with triton (0.01%), and PBS alone (100 µL/well) for 5 min. The cells were incubated overnight at 4 • C with 100 µL/well of a 1/1000 dilution of the anti-γ-H2AX primary antibody (goat anti-serum, 20%) for 1 h at room temperature. The cells were washed with 100 µL/well PBS twice. Cells were then incubated overnight on a mild rotative shaker at 4 • C with 30 µL/well of a 1/1000 dilution of the DyLight 488 conjugated secondary antibody (Mouse Phospho-Histone H2A.X pSer140 Antibody (3F2)-species reactivity: mouse, human, and bovine, Thermo Scientific ®® ref MA1-2022) in goat serum, washed with PBS twice for 5 min (100 µL/well).
Work henceforth progressed in the dark. Cells were mounted using a few microliters of ProLong mounting medium with DAPI. A coverslip was added (Cover Glasses, 6 mm diameter, thickness n.0 VWR-Cat. 631-0168), and plates were recovered by adhesive film for microplates (VWR- Ref. 391-1254). Plates were conserved in aluminum at 4 • C.
The fluorescence was quantified using a microscope Axio Observer Z.1 (Zeiss) and Image J.
A t-test was performed (p-values, 1-sided, based on unpaired data). The level of significance was p ≤ 0.05.

Statistical Analysis
Statistical analysis was performed with GraphPad Prism 5 software (GraphPad Software, La Jolla, CA, USA). A probability level of p < 0.05 was considered significant.