Differential induction of apoptosis and autophagy by pyrrolizidine alkaloid clivorine in human hepatoma Huh-7.5 cells and its toxic implication

Growing evidence suggests that the pyrrolizidine alkaloids (PAs)-induced hepatotoxicity is mediated by multiple cell death/defence modalities. However, the detailed mechanisms are still lacking. In this study, the hepatotoxic effects of four PAs including three retronecine-type ones (senecionine, seneciphylline and monocrotaline) and one otonecine-type (clivorine) on the proliferation of Huh-7.5 cells and the possible mechanisms were investigated. The results showed that all the PAs could inhibit cell proliferation and induce apoptosis in a concentration-dependent manner. Among them clivorine was the most significant one. In addition to its effect on apoptosis, clivorine treatment could promote autophagy in Huh-7.5 cells, as evidenced by the accumulation of autophagosomes, the enhancement of LC3B expression at the concentrations close to its IC0 value, and the increased conversion of LC3B-I to LC3B-II in the presence of lysosomal inhibitor (chloroquine) and decreased formation of green fluorescent protein (GFP)-LC3 positive puncta in the presence of autophagic sequestration inhibitor (3-methyladenine). Among the other tested PAs, senecionine and seneciphylline also activated autophagy at the same concentrations used for clivorine but monocrotaline did not. Furthermore, our study demonstrated that suppression or enhancement of autophagy resulted in the remarkable enhancement or suppression of senecionine, seneciphylline and clivorine-induced apoptosis at the concentration close to the IC10 for clivorine, respectively, indicating a protective role of autophagy against the PA-induced apoptosis at the low level of exposure. Collectively, our data suggest that PAs in different structures may exert different toxic disturbances on the liver cells. Apoptosis may be one of the most common models of the PA-induced cytotoxicity, while autophagy may be a structure-dependent defence model in the early stage of PA intoxication. Differential induction of apoptosis and autophagy probably depending on the concentration is essential for the cytotoxic potency of clivorine.

Senecionine, seneciphylline and monocrotaline are three representatives for retronecinetype PAs. The former twos are commonly found in genera such as Senecio and Gynura of the family Asteraceae, while monocrotaline is mainly present in Crotalaria of the family Fabaceae [3,11]. For example, senecionine and seneciphylline are two major PA components in the roots of Gynura japonica (Thunb.) Juel., which is used for the treatment of physical injuries and traumatic haemorrhages in Chinese herbal medicines (CHM) [11,12]; monocrotaline is a predominant PA in the seeds of Crotalaria assamica Benth., which has effects of dispelling wind-damp syndrome, stopping bleeding and diminishing swelling [9]. As a representative otonecine-type PA, clivorine is abundant in quite a few Ligularia species from Asteraceae, whose roots and rhizomes are usually used to relieve cough and remove excessive phlegm in CHM [13][14][15]. Currently, intake or misuse of PA-containing herbs for medical purposes is still common in China, creating a growing safety concern with the potential exposure to PAs [3,[11][12][13][14][15][16].
In view of the mechanism of PA-induced hepatotoxicity, PAs may suffer metabolic activation by cytochromes P450 (CYP450s) to produce the highly reactive pyrrolic metabolites and then cause hepatotoxicity [17][18][19]. However, how the PAs, either retronecine or otonecinetype, induce hepatotoxicity at the cellular and molecular levels is not well known. In the past twenty years, a few modes of cell death, e.g., necrosis (oncosis) and apoptosis have been found to mediate the PA-induced cytotoxicity [13,[20][21][22]. Several retronecine-type PAs including senecionine, monocrotaline, isoline, retrorsine and riddelliine and the otonecine-type PA clivorine had been reported to cause apoptosis in vitro and/or in vivo [20][21][22][23]. For instance, Ji et al. [20,24] found that clivorine could remarkably inhibit cell proliferation and induce mitochondria-mediated apoptosis in human liver L-02 cells. This effort involved the pathway by degradation of Bcl-xL protein, mitochondrial release of cytochrome c, and activation of caspases-3/-9. The further study indicated that senecionine also involved the same apoptotic signalling pathway as clivorine [20]. Monocrotaline could induce both oncosis and apoptosis of hepatic parenchymal cells of rats in vivo which but occurred differently in distribution [22]. The oncotic lesions primarily occurred in the centrilobular regions with abundant CYP450s, while the caspase inhibition could prevent the development of both oncosis and apoptosis with little effects on the bioactivation of monocrotaline. A hypoxia-regulated cell-death factor,  [7] to LY). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
BNIP3, was found to be up-regulated and implicated in switching the mode of cell death from apoptosis to oncosis after monocrotaline exposure. The study on retrorsine showed that its cytotoxic mode on Huh-7 cells may be dose-dependent with apoptosis at low doses and necrosis at high doses [21]. A recent study also found that the PC12 cells after clivorine exposure involved the apoptotic death at the concentrations higher than 50 μM while suppressed neuronal differentiation via TrkA/Akt signalling pathway at lower doses than it [7]. All the evidence suggests that the modes of PA-induced cell toxicities were complex and diverse with involvements of many cellular factors and/or events, which may depend on chemical structure, concentration, treatment time and cell types or even cellular compartment. Autophagy (hereafter referring to macroautophagy) is the naturally destructive mechanism that disassembles, through a regulated process, unnecessary or dysfunctional cellular components [25]. During this process, targeted cytoplasmic constituents are isolated from the rest of the cell within a double-membraned vesicle known as an autophagosome. The autophagosome then fuses with a lysosome and the contents are degraded and recycled [26]. In the context of disease, autophagy has been seen as an adaptive response to stress, which promotes survival, whereas in some other cases it appears to stimulate cell morbidity and death, or sometimes called autophagic cell death [27]. Many studies have showed that progress of many diseases, e. g., tumour, diabetes and drug-induced liver injuries (DILI) are closely related to the autophagic process in cells [28][29][30]. In the case of DILI, autophagy was found to protect the liver cells by removal of the damaged mitochondria, which regulation may be a new approach for the treatment of liver diseases [31,32], but if too much damaged mitochondria overload autophagy or elicit "autophagic stress", the DILI may be enhanced [33,34]. In China, the PAs-containing herbal medicines have been an important cause of DILI owing to numerous HSOS cases [6,16]. Even though some modes of cell death and signal pathways, e.g. oncotic necrosis and mitochondrial-mediated apoptosis have been reported [20][21][22][23][24], there is still limited knowledge of the liver cell-intrinsic response modes and machineries for the PA-induced toxicity.
In our preliminary study, clivorine could modulate the mRNA expression of the autophagy-associated gene LC3 (corresponding to microtubule-associated protein 1 light chain 3) in Huh-7 cells. The evidence implies that autophagy may have an impact on the toxicity of PAs. In the present study, we continued to study the toxic effects of PAs on the human hepatoma Huh-7.5 cells with three retronecine-type PAs (senecionine, seneciphylline, monocrotaline) and one otonecine-type PA clivorine at different concentrations. Their effects on cell proliferation and underlying mechanism especially involving autophagy were investigated. Our findings demonstrate that all PAs have cytotoxic potency, among them, the most distinct one is clivorine. The same apoptotic pathway may be responsible for their toxicities, while autophagy may play a protective role in the early stage of toxic insults by PAs especially clivorine.

Cell line
Human cancer liver Huh-7.5 cells were kindly provided by Professor Wenzhe Ho at Temple University School of Medicine, USA. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin at 37˚C with 5% CO 2 .

Cell viability assay
Cells in log-phase were seeded in 96 well plates (10, 000 cells/well) overnight and then treated with senecionine, seneciphylline, monocrotaline and clivorine for 48 h. The cells without PA treatments (with only solvent) were used as the negative controls. Eleven concentrations were set for each PA, including 0 μM, 3.125 μM, 6.25 μM, 12.5 μM, 25 μM, 50 μM, 100 μM, 200 μM, 400 μM, 800 μM and 1000 μM; each repeated for 6 wells. Cell viability was assayed using the MTT assay kit according to the standard procedure provided by the supplier. Briefly, 22 μL of MTT staining fluid was added in each well, and then incubated for 4 h. Afterwards, 100 μL of the formazan crystals were dissolved by adding 2-propanol in each well. The absorption of the formazan solution was measured by a Multiskan GO spectrophotometer (Thermo Scientific China, Shanghai, China) at a wavelength of 570 nm [35]. The cell viability was calculated as the following formula.

The cell viability ð%Þ ¼
The OD value of PA À treated group The OD value of control group Â 100 Based on the values of cell viability, the values of IC 50 (the concentration that inhibited cell viability to 50% of the control), IC 20 (the concentration that inhibited cell viability to 20% of the control), IC 10 (the concentration that inhibited cell viability to 10% of the control) and IC 0 (the maximal concentration without inhibition on the cell viability) were also determined using nonlinear regression analysis in GraphPad Prism 4 (GraphPad Software, Inc., LaJolla, USA).

Flow cytometry
For flow cytometry analysis, four concentrations were selected for each PA, including 6.25 μM, 25 μM, 100 μM, and 400 μM. Most of the concentrations were lower than the calculated IC 50 value for each PA (except 400 μM for clivorine) to avoid the non-apoptotic cell death. After treatment, the cells were collected, washed with PBS and digested by 0.25% trypsin solution (without EDTA), and then further incubated at room temperature for 2 minutes. After that, the cells were re-collected and suspended with binding buffer. The suspension was incubated with Annexin V-kFluor488 and stained with PI (Kaiji Biology Corp., China) at room temperature for 20 minutes and then analyzed on flow cytometer (Cyto Flexz, Beckman, USA). The annexin V-positive/PI-negative cells indicated apoptotic cells at the early period, and the annexin V-positive/PI-positive cells showed those at the later period [36]. The ratio of apoptosis or apoptosis index (AI) was calculated using the following equation: AI ð%Þ ¼ Apoptosis cell numbers Apoptosis cell numbers þ the normal cells Â 100 Real-time PCR analysis SYBR Green quantitative real-time PCR (RT-PCR) was used to investigate the mRNA expression levels of the autophagy-associated genes including LC3, ATG3, ATG5, and ATG7. For this analysis, 24 well plates were used by seeding 50, 000 cells (500 μL of cell suspension) per well, which were then treated with or without 56 μL of each PA for 24 h. Three concentrations for each alkaloid, respectively, including 3.125 μM, 6.25 μM and 12.5 μM were used for the experiments. All concentrations were in or lower than the range of the calculated IC 0 and IC 10 values for each PA to avoid the undesirable cytotoxicity. Total RNA was extracted with TRIzol, and the first strand cDNA was synthesized with TAKARA RR036A reverse transcriptase kit following the manufacturer's instructions. Real-time PCR was performed with the Bio-Rad CFX96 thermal cycler (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer's protocol. Primers for all autophagic genes and the internal reference GAPDH were designed by Primer express 2.0, which sequences were as follows: • Each cycle threshold (Ct) was measured and normalized to the average Ct value of GAPDH (n = 6). The results were calculated by 2 -44ct method and given as the relative transcriptional levels of target genes.

Western-blot analysis
In this experiment, three concentrations (3.125 μM, 6.25 μM, and 12.5 μM) were also chosen for each PA, respectively, according to RT-PCR results. Total protein concentrations were determined by NanoDrop2000 spectrophotometer (Thermo Scientific, USA), for which each sample was normalized to the equal total protein concentration. Proteins were separated by SDS-PAGE and probed with appropriate primary and secondary antibodies as previously described by Li et al. [37]. The following antibodies including anti-human antibodies against LC3B (1:3000, Cell Signalling), and internal reference β-actin (1:5000, Millipore) were used. The protein levels including LC3B-I, LC3B-II and β-actin were quantified by Quantity One (Chemical light emitting gel imaging system).

Transmission electron microscope analysis
Electron microscopy was utilized to detect the autophagy response in Huh-7.5 cells treated with or without senecionine, seneciphylline, monocrotaline and clivorine (each 6.25 μM). The cells were digested with trypsin and centrifuged, and then immediately fixed with 2.5% glutaraldehyde/0.1 M sodium cacodylate. The cells were then post-fixed with 1% osmium tetroxide, followed by dehydration with an increasing concentration gradient of ethanol and propylene oxide. The cells were then embedded with the epoxy resin Epon812 and cut to ultrathin sections (50-60 nm) using an ultramicrotome (LKB-I, Bromma, Sweden). Images were examined using a H7650 electron microscope (Hitachi Ltd., Tokyo, Japan) at 80 kV after the samples were stained with 3% uranyl acetate and lead citrate.

Fluorescence microscopic analysis
Transient GFP-LC3-expressing Huh-7.5 cells were generated using lentivirus-mediated GFP-LC3 over-expression. A lentiviral vector containing GFP-LC3 fusion gene and lentivirus were purchased from Genechem Company (Shanghai, China). Infection of Huh-7.5 cells with lentivirus was carried out at a multiplicity of infection (MOI) of 10.0. Autophagy was assessed using GFP-LC3 redistribution in cells that was detected by an inverted fluorescence microscope (Nikon Ti-s, Japan). The number of GFP-LC3 puncta per cell was determined in three independent experiments (30 cells were counted per experiment).

Autophagy inhibitor and activator treatments
To address the PA-induced autophagy flux in Huh-7.5 cells, the autophagy inhibitors were used. Briefly, CQ (5 μM) and 3-MA (5 mM) were added to the cells naive and expressing GFP-LC3 from lentiviruses, respectively, 2 h before senecionine, seneciphylline and clivorine treatments (each 6.25 μM). The PA-treated cells were then subjected to the measurement of autophagy-related protein levels by Western blotting or of green fluorescence signal by fluorescence microscope as described above. As a further effort to address the relationship between apoptosis and autophagy, both the autophagy inhibitor 3-MA and autophagy activator Rapa were used. Huh-7.5 cells were pre-treated with the 3-MA (5 mM), or Rapa (100 nM) for 2 h, and then incubated in the presence of senecionine, seneciphylline, clivorine and monocrotaline (each 12.5 μM) for 24 h. Finally, the cells were harvested and treated for the flow cytometry analysis.

Statistical analysis
Values are expressed as mean ± SD of at least three independent experiments. Statistical analysis was performed using SPSS 17.0 software (SPSS Inc., Chicago, IL) or GraphPad Prism 4 (GraphPad Software, Inc., LaJolla, USA), using one-way analysis of variance followed by twotailed t-test for evaluation, P < 0.05 was considered to be significantly different.

Results
Cytotoxicity of senecionine, seneciphylline, monocrotaline and clivorine in Huh-7.5 cells Senecionine, seneciphylline, monocrotaline and clivorine (Fig 1) are four representatives of hepatotoxic PAs present in many Chinese herbal medicines [3,9,[11][12][13][14][15]. As shown in Table 1 and Fig 2, all four alkaloids could inhibit the Huh-7.5 cell viability. For each PA, the cell survival rate decreased with the increase of concentration, the efficiency, however, was different. The inhibitory effects of senecionine and seneciphylline in the range of the concentrations from 100 μM to 1000 μM, clivorine from 12.5 μM to 400 μM, monocrotaline from 50 μM to 1000 μM, were substantially higher than the negative control. Comparatively, the otonecinetype PA clivorine showed the highest anti-proliferative potency. At 100 μM for example, its average inhibition rate (51.4%) was significantly higher than 16.7%, 19.3% and 28.4% of other three retronecine-type PAs senecionine, seneciphylline and monocrotaline, respectively. Moreover, the IC 50 value calculated in average for clivorine was 141.7 μM, which was remarkably lower than those observed with the other three PAs (509.7 μM, 491.9 μM and 413.2 μM). This result was comparable to that reported [4,20], in which clivorine exhibited the higher cytotoxicity than other types of PAs. Because MTT assay is a common method for assessing the inhibitory effect on cell viability, our result further demonstrated that direct cell death may play an important role in the toxicity of clivorine. It is worth of note that the IC 50 values obtained in different studies may be largely different from each other, e.g., the IC 50 value for clivorine in this study was much higher than 40.8 μM in another study by Ji et al. [20]. The discrepancies may be due to differences in cell lines being used or in vitro model for cytotoxicity screening [4,20]. The Huh-7.5 cell line used in the present study was derived from Huh-7 cell line, which has been reported to have no or mutant expression of bcl-2 and p53 genes, indicating that this cell line may be susceptible to toxins but insensitive to some cytotoxic agentsinduced apoptosis [20,21]. Because different types of PAs may have different modes of cytotoxic effects [4,20,23], the toxic responses and underlying mechanisms of Huh-7.5 cell line to various PAs with different types or necine acid moieties may be different. On the other hand, considering that the PAs with high concentrations may have more direct cytotoxicity [4,21], it is also interesting to know how the cells respond to PAs at low or nontoxic concentrations. Accordingly, the maximal non-toxic concentrations (IC 0 s) and those with 10% and 20% of  20 or above showed remarkable discrepancy between clivorine and all other retronecine-type PAs. Of note, monocrotaline showed comparable cytotoxic potential to that of clivorine at the lower concentrations than IC 20 . The potency order of hepatotoxicity was thus clivorine ! monocrotaline ! seneciphylline ! senecionine.
Senecionine, seneciphylline, monocrotaline and clivorine induce apoptosis of Huh-7.5 cells As for the mechanism of cytotoxicity for PAs, many studies suggested that the apoptotic cell death may play a vital role in it [13,[20][21][22][23]. Senecionine, monocrotaline and clivorine had been found to induce apoptosis in several liver cell lines, to the exclusion of Huh-7.5 cells. In this experiment, the effects of four PAs including senecionine, seneciphylline, monocrotaline and clivorine on the apoptosis of Huh-7.5 cells were further investigated and compared by using flow cytometry. Based on the above analyses, a series of concentrations around or lower than IC 50 s were investigated. As a result, the overall apoptosis ratios or AIs of Huh-7.5 cells increased with increasing concentrations of each PA (Fig 3). All three retronecine-type PAs, senecionine, seneciphylline and monocrotaline showed almost same potency to induce the apoptosis in Huh-7.5 cells. Generally, their AIs against Huh-7.5 cells at all concentrations  were markedly higher than that of the negative control (4.2% in average), but all these values were not very high (around 20% in average at 400 μM) and showed no differences from each other at the same concentrations. This result suggests that the concentration may contribute to their potencies to induce apoptosis. In comparison, clivorine produced relatively higher apoptosis ratios of Huh-7.5 cells than other three retronecine-type PAs at the corresponding concentrations. The discrepancies become much more obvious at the concentrations higher than 25 μM, in which the AI value at 100 μM (22.8 ± 1.7%) was almost equal to or higher than those of senecionine, seneciphylline and monocrotaline at 400 μM and considerably higher than those of them at the same concentration. This effect may partially be derived from the direct cytotoxicity of clivorine, but for the retronecine-type PAs may mainly from the inhibition of cell proliferation [4]. Except for the concentration, the chemical structure especially necine bases may also be a key determinant for the PA-induced apoptosis. Based on above experiment, clivorine may have higher potency to induce apoptosis in Huh-7.5 cells than other PAs, but which effects tend to be consistent at the low concentrations of 25 μM (16.2 ± 0.9%) and 6.25 μM (11.2 ± 1.7%). Also, monocrotaline did not show apparent high effects comparable to those in MTT assay at the concentrations less than IC 20 . These results suggest that apoptosis may not be mere determinant for the PAinduced cytotoxicity, in which some other mechanism of action may be involved. Combined the results from MTT, the PAs intoxication may depend on their chemical structures and concentrations, which was in good agreement with those reported by previous investigators [4,21].
Senecionine, seneciphylline and clivorine enhance the mRNA expression of LC3 in Huh-7.5 cells The PA-induced cytotoxicity in Huh-7.5 cells markedly decreased at low or even nontoxic concentrations, which may not be causally involved by apoptosis or necrosis. In this scenario, some reports suggested that the cell defence but not direct toxic pathway may play a pivotal role [32][33][34]. Under stressful conditions, autophagy is usually activated to protect the cell from damage [38]. A variety of key autophagy-relevant components have been found to participate in the activation by involving in ubiquitin-like conjugation systems, for instance, Atg3, Atg5, Atg7 and LC3 involved in the LC3 lipidation and ATG12-ATG5 conjugation systems that are essential for the autophagosome formation and maturation [39]. Their transcriptional and translational regulation especially that of LC3, the most widely used marker for monitoring autophagy is preferred to provide correlative data related to autophagy [27,[37][38][39]. Accordingly, four autophagy-associated genes including ATG3, ATG5, ATG7 and LC3 were examined by RT-PCR to observe if senecionine, seneciphylline, monocrotaline and clivorine could modulate autophagy at the transcriptional level. To avoid the undesirable direct toxic effects, the choice of the concentration was mainly based on the values in the range of IC 0 and IC 10 of clivorine for each PA. As shown in Fig 4, the relative mRNA levels of LC3 as the fold of control were 1.27 ± 0.06, 1.92 ± 0.05 and 1.75 ± 0.10, in the cells treated by clivorine at 3.125 μM, 6.25 μM and 12.5 μM, 1.73 ± 0.08 and 1.69 ± 0.06 by senecionine at 6.25 μM and 12.5 μM, and 1.81 ± 0.11 and 1.79 ± 0.09 by seneciphylline at the same two concentrations as senecionine, respectively, all of which were significantly higher than that of negative control. For the three PAs, the effects at 6.25 μM and 12.5 μM were also remarkably higher than those at the lower concentrations of 6.25 μM, 25 μM, 100 μM and 400 μM; and C6.25, C25, C100, and C400 represent clivorinetreated cells at the concentrations of 6.25 μM, 25 μM, 100 μM and 400 μM, respectively. * P<0.05; ** P<0.01 vs. negative control. https://doi.org/10.1371/journal.pone.0179379.g003 Clivorine induces autophagy in Huh-7.5 cells concentration of 3.125 μM. The concentration-dependent effects were obvious but slightly dropped at the high concentration of 12.5 μM. For monocrotaline, no unambiguous statistical significance was observed for all concentrations compared with the control. Moreover, there was no substantial change in the mRNA levels of ATG3, ATG5, and ATG7 genes between the PAs-treated and un-treated cells. These results indicated that senecionine, seneciphylline, and clivorine could enhance the mRNA expression of LC3 in Huh-7.5 cells, in which clivorine was more effective than other twos at low concentration close to its IC 0 (6.25 μM) or nontoxic concentration (3.125 μM) (see Fig 4). The transcriptional regulation mediated by PAs may be selective for either ATG genes or PA structures and concentrations in this setting. The up-regulation of the mRNA level of LC3 has been documented to be involved in the elevated autophagy flux probably by replenishing the LC3 protein destroyed during autophagosome fusion with the lysosome [39,40].
Senecionine, seneciphylline and clivorine induce autophagy in Huh-7.5 cells Based on the above results, autophagy may mediate the PA-induced toxicities in Huh-7.5 cells.
To further clarify this effect, we first examined the formation of autophagosomes under transmission electron microscopy (TEM). As can be seen in Fig 5, the closed double-membrane structures or autophagic vacuoles (with a diameter of 300 to 900 nm) were distinctly observed around the nucleus in cells treated by 6.25 μM of senecionine, seneciphylline and clivorine, in comparison with mainly rough endoplasmic reticulum (ER) and mitochondria in the similar region in the untreated cells; also in this concentration clivorine treatment generated some mature autophagosomes containing sequestered materials. In the monocrotalinetreated cells, however, the similar morphological feature of autophagic vacuole accumulation was not observed (Fig 5E). Then we measured the expression of the autophagy marker, autophagosomal membrane protein LC3B, to assess the autophagic activation. The total expression of this protein and the ratio of LC3B-II and LC3B-I, may be elevated owing to increased autophagy induction [37,39]. Comparatively, LC3B-II is more specific than LC3B-I for the assessment of autophagosomes or autophagic flux, which correlation has been demonstrated in many studies of natural compounds involving cytoprotective autophagy [27]. In this study, the protein levels of LC3B-I and LC3B-II were determined by western-blot analysis after exposure to each PA, of which three concentrations (3.125 μM, 6.25 μM and 12.5 μM) were selected between the IC 0 and IC 10 values of clivorine to exclude high toxic disturbances on Huh-7.5 cells. Except for clivorine, senecionine and seneciphylline both largely increased the transcriptional expression of LC3 at high concentrations of 6.25 μM and 12.5 μM but monocrotaline had no effect at any concentration (Fig 4). As a comparison, it was also interesting to know how they affect the protein expression of LC3B especially LC3B-II in Huh-7.5 cells. As can be seen from Fig 6A-6F, treatment with senecionine, seneciphylline and clivorine all resulted in remarkable increases of the relative protein levels of either total LC3B or LC3B-II at the tested concentrations compared with the negative control. Correspondingly, the relative LC3B-II/ LC3B-I ratios were also up-regulated remarkably at all treatments. Unlike these results, however, monocrotaline had no any effect (Fig 6G and 6H). This evidence suggests that clivorine, senecionine and seneciphylline but not monocrotaline could induce autophagy in Huh-7.5 cells. It was noteworthy that the induction by senecionine and seneciphylline were in a concentration-dependent manner but clivorine showed a different trend. From Fig 6E and 6F, the accumulation of either total LC3B or LC3B-II reached the highest level at 6.25 μM but declined at 12.5 μM. Similarly, the relative LC3B-II/LC3B-I ratios significantly increased and also peaked at 6.25 μM in the clivorine-treated cells. This distinct effect by clivorine may be related to its higher toxic potency at the concentrations used in the assay than other PAs (Table 1).
Since the change of LC3B-II level could be caused by either autophagosome formation or degradation in lysosomes, we further investigated senecionine, seneciphylline, and clivorineinduced autophagic flux in the presence and absence of CQ (5 μM), a lysosomal inhibitor that may block the fusion of autophagosomes with lysosomes [39]. The Western blot analysis showed that both the PAs and CQ remarkably induced conversion of LC3B-I to LC3B-II in Huh-7.5 cells, respectively, while their co-treatments with CQ further enhanced the conversion evoked by CQ alone (Fig 7). This suggests that the senecionine, seneciphylline, and clivorineinduced increase in LC3B-II levels may be ascribed to the increased autophagosome formation. Furthermore, we assessed the formation of punctate structures (puncta) with GFP-LC3 reporter, a marker of autophagosomes formation that represents the accumulation of LC3B-II on autophagic vacuoles [39,41]. As a result, a significant increase of cells containing GFP-LC3 puncta was observed in Huh-7.5 cells treated with senecionine, seneciphylline, and clivorine for 24 h, respectively, and the non-specific sequestration inhibitor 3-MA (5 mM) partially blocked the increase of GFP-LC3 puncta per transfected cell induced by them (P < 0.05) (Fig  8). Collectively, these data indicated that senecionine, seneciphylline and clivorine induced autophagy in Huh-7.5 cells especially at the concentrations close to the IC 0 value of clivorine. Moreover, these results suggest that the structural differences in either necine base or necic acid groups of these PAs (Fig 1) may contribute to their differential effects on autophagy in Huh-7.5 cells.
seneciphylline, clivorine and monocrotaline for 24 h, respectively. As indicated by black arrows, the autophagic vacuoles with double-membrane structures were observed in cytoplasm. Asterisks indicate mature autophagosomes containing highly degraded contents. N, nucleus. Scale bar, 500 nm.
https://doi.org/10.1371/journal.pone.0179379.g006 Apoptosis and autophagy are two important and interconnected processes and play important roles in the promotion or inhibition of cell survival in response to many natural products or chemotherapeutic drugs [27,33,[40][41][42]. From the above studies, both apoptosis and autophagy were involved in the PA-induced cellular responses. In order to clarify the relationship between apoptosis and autophagy, the role of autophagy in the PA-treated cell survival was further investigated. After pre-treatment with the autophagy inhibitor 3-MA (5 mM) or autophagy activator Rapa (100 nM), Huh-7.5 cells were incubated with senecionine, seneciphylline, clivorine and monocrotaline for 24 h, respectively. Afterwards, the cell apoptosis was examined. As shown in Fig 9, the Huh-7.5 cell apoptosis induced by senecionine, seneciphylline, clivorine and monocrotaline was all obvious at the concentration of 12.5 μM. The 3-MA treatment (for autophagy suppression) significantly enhanced senecionine, seneciphylline and clivorine-induced cell apoptosis compared with each PA alone, whereas the Rapa treatment (for enhancement of autophagy) markedly weakened these PAs-induced apoptosis. In addition, we did not observe that either 3-MA treatment or Rapa treatment altered monocrotaline-induced apoptosis. These results indicated that autophagy may have a protective role to prevent the hepatic cells from PAs-induced apoptosis.

Discussion
Autophagy as a highly conserved cellular homeostatic process was found to be regulated by many natural products to make either protective or cytotoxic effects across different types of mammalian cells [27]. It may be difficult to exactly predict which effects will be dominant because the autophagic process may not exist alone, but interconnected with many other cellular processes. For example, autophagy and apoptosis can interact on each other in an inhibitory way through some common proteins and signalling pathways; autophagy usually exhibits a dual role in cell fate determination [39][40][41][42]. Previous studies have shown that some PAs can induce apoptosis in isolated mouse hepatocytes [20], human liver L-02 cells [20,24] and human hepatoma cells HepG2 [4] and Huh-7 [21]. In this study, we tested and proved the differential toxic effects of PAs with different types or structures and concentrations on Huh-7.5 cells. In combination with the previous studies, we propose the following hypotheses: (1) PAs induce cytotoxicity by involving in apoptosis and/or necrosis pathways, which may be linked to autophagy; (2) The induction of autophagy may be concentration and structure-dependent; (3) The initial autophagy may have a positive role in the cell survival. Using four PAs with two types, we demonstrated firstly that two retronecine-type PAs, senecionine and seneciphylline, and a representative otonecine-type PA clivorine, did induce autophagy in Huh-7.5 cells at the concentrations close to the IC 0 and IC 10 values of clivorine via RT-PCR, western-blot, fluorescence microscopy and transmission electron microscopy analyses. This effect was accompanied by an up-regulation of the LC3 pathway and accumulation of autophagosomes (Figs 4-8).
On the one hand, it may be concentration-dependent because in many other studies, the high concentrations were usually used to explore the toxic mechanism of PAs, such as 50 μM and 100 μM for clivorine [7,20] and 500 μM for monocrotaline [8], in which no autophagy but apoptosis or necrosis (depending on treatment time) was definitely observed. According to groups. The densitometric intensities of LC3B-I, LC3B-II and β-actin bands were quantified by Image J software. The relative LC3B-II/LC3B-I ratios were calculated and shown as the fold of control (without PA treatment under different conditions, which was defined as 1, respectively). The data shown in Fig 7b, 7d, and 7f are the mean ± SD of the results of three independent experiments. The P value was calculated by Student's t-test ( * P < 0.05).
https://doi.org/10.1371/journal.pone.0179379.g007 our results, the clivorine-induced autophagy occurred at its nontoxic concentration of 3.125 μM, developed or peaked at 6.25 μM (a concentration around its IC 0 value), and decreased at 12.5 μM (around IC 10 ); on the contrary, apoptosis become prominent at 6.25 μM and increased with the increase of concentration (Figs 3 and 6). The concentration around IC 0 may be a turning point to switch autophagy to apoptotic cell death. This effect was distinct for clivorine amongst the tested PAs because senecionine and seneciphylline increased the LC3B protein expression in a concentration-dependent manner whereas monocrotaline had no effect at all concentrations used in the assay (Fig 6). Meanwhile, at these concentrations clivorine displayed the higher toxic potency than other PAs (Table 1, Fig 2). All the evidence suggests that concentration may have a crucial role in the PA-induced autophagy especially by clivorine. The long-recognised mechanism of the PA-induced cytotoxicity involving metabolic  ) and (d) monocrotaline (12.5 μM) treated or untreated Huh-7.5 cells were pre-cultured in the presence of autophagy inhibitor 3-MA (5 mM)) or activator Rapa (100 nM) for 2 h, and then cell apoptosis were subjected to flow cytometry analyses. The data shown in 9a, 9b, 9c and 9d are the mean ± SD of the results of three independent experiments. The P value was calculated by Student's t-test ( * P < 0.05).
https://doi.org/10.1371/journal.pone.0179379.g009 activation by CYP450s may play a key role in this effect [17][18][19]. According to the existing findings about CYP450 activities in Huh-7 cells [43] and further study on the in vitro metabolism of clivorine in pooled human liver microsomes (S1 File), however, we found that the metabolic activation by CYP450s in our Huh-7.5 cell culture system may be too weak to generate a certain level of reactive "pyrrolic" metabolites to make the cellular responses we have observed. Nevertheless, how much exactly the pyrrolic metabolites can be formed at the levels of exposure used in our Huh-7.5 cell culture system and to what extent the metabolic activation and associated CYP450s involving formation of these metabolites is contributing to the autophagy or other cytotoxic effects of PAs warrant further investigation.
On the other hand, the structural factor may also be critical for the PA-induced autophagy. In this study, we used four PAs including three retronecine-type PAs senecionine, seneciphylline and monocrotaline and one otonecine-type PA clivorine, all with different necic acids (Fig 1). Even in the same type, senecionine and seneciphylline are 12-membered macrocyclic diesters while monocrotaline is an 11-membered one. The structural difference may lead to different physicochemical property (e.g., lipophilicity) and/or steric effects, which eventually affect the hepatotoxic potential of PAs [4]. Our study demonstrated that monocrotaline induced apoptosis alone, while clivorine, senecionine and seneciphylline could induce both apoptosis and autophagy. Clivorine can more readily inhibit cell viability, and induce apoptosis than other PAs. Clivorine treatment at its IC 0 level of exposure (6.25 μM) showed the highest autophagic response but flattened ratio of apoptosis, which was different from other retronecine-type PAs (Figs 3 and  6). This result suggests a distinct mechanism for the cell responses to clivorine. Monocrotaline showed the highest inhibitory effects against Huh-7.5 cells among the three retronecine-type PAs but had less apoptotic potency at the concentrations tested, which may rest with activation of other cell death processes such as necrosis [8,22]. The deficiency of autophagy induction by monocrotaline may also partially contribute to its high cytotoxicity at the low levels of exposure. Overall, autophagy is involved in the molecular machinery of PA-induced cytotoxicity in Huh-7.5 cells, which activation/regulation may depend on concentration and structure of PAs.
In addition, the clivorine-induced autophagy as mentioned above may be protective because clivorine at the concentration level close to IC 0 had much less anti-proliferative effects (Fig 2) and apoptosis-inducing potencies on Huh-7.5 cells than those at the high concentrations (Fig 3). More importantly, the treatment by autophagic inhibitor or activator could significantly enhance or weaken clivorine-induced apoptosis at 12.5 μM, respectively (Fig 9). Since this concentration was close to the IC 10 for clivorine (Table 1), showing some degree of direct cytotoxicity and apoptosis (Figs 2 and 3), the regulation of autophagy may have important implication for the cell death/defence responses to this PA (Fig 9). As for the other tested PAs, senecionine and seneciphylline also showed the similar evidence at the same concentration, which is far below the IC 0 s of two PAs (Table 1). In the context of stress conditions, autophagy may be an adaptive response earlier than apoptotic cell death [27,33]. Therefore, the autophagic activation rather than apoptosis or necrosis is likely to serve as a pro-survival mechanism in the initial stage of PA-induced cytotoxicity. In combination with the toxic effects from the PAs, the multiple cell survival and death pathways linking autophagy, apoptosis and necrosis may be intermingled [33,40].
In summary, the present study investigated the effects of three retronecine-type PAs (senecionine, seneciphylline and monocrotaline) and one otonecine-type PA (clivorine) on the proliferation of Huh-7.5 cells and further explored potential cellular and molecular bases on it. Our results indicated that PAs with different types or structures and concentrations may exert different toxic disturbances on Huh-7.5 cells. Apoptosis may be the common cause of cell death shared by them. In comparison, clivorine is the most toxic PA due to its highest direct cytotoxicity and apoptosis-inducing potency. Most notably, clivorine can induce autophagy in Huh-7.5 cells at the concentrations close to its IC 0 and IC 10 values by up-regulating the expression of autophagy markers LC3B in both transcriptional and protein levels as well as autophagosome accumulation. Another two PAs including senecionine and seneciphylline also induced the protective autophagy at the low level of exposure. The autophagy pathway may serve as a protective mechanism in the early stage of PA intoxication. The differential induction of apoptosis and autophagy by clivorine probably depending on the concentration may play an essential role in its cytotoxic potency. This work provides a new mechanistic insight into the cell death/survival modality mediated by PAs or PA-containing products. Also, our findings suggest that the intrinsic cellular signalling pathways of the PA-induced autophagy and interactions with other important cellular processes deserve further exploration for a comprehensive understanding of their toxic implications.
Supporting information S1 File. In vitro microsomal metabolism of clivorine by human liver microsomes. In vitro microsomal metabolism of clivorine was investigated using pooled human liver microsomes (InVitroCYP TM H-class 25-donor mixed gender) in the presence or absence of NADPH. Microsomal incubations were conducted in a total volume of 200 μl containing human liver microsomes (1 mg protein/mL), NADPH (1 mM), GSH (2 mM) and clivorine (0.25 mM), all in phosphate buffered saline at pH 7.4. The incubation mixture was treated with an equal volume of acetonitrile and centrifuged, and then the supernatant was subjected to HPLC analysis. Representative analytical results are shown in the file, and the experimental details about materials and methods, results and discussion are also presented.