Synthesis and structure of a new thiazoline-based palladium(II) complex that promotes cytotoxicity and apoptosis of human promyelocytic leukemia HL-60 cells

Cisplatin is one of the most widely used chemotherapeutic agents in the treatment of different tumors but has high toxicity and side effects. Therefore, the synthesis of new chemotherapeutic agents is necessary, so that they are effective in the treatment of cancer while avoiding such toxicity. In this study, we have synthesized and characterized a palladium(II) complex, [PdCl2(µ-PyTT)2]Cl2·4H2O (PdPyTT), with 2-(2-pyridyl)imine-N-(2-thiazolin-2-yl)thiazolidine (PyTT) as a ligand; besides, its cytotoxicity and pro-apoptotic capacity was tested in human promyelocytic leukemia HL-60 cell line. Similar to cisplatin, PdPyTT produced a time- and dose-dependent decrease in cell viability. Additionally, the palladium complex increased both the proportion of cells with apoptotic morphology and the activation of caspase-3 and -9. PdPyTT, like cisplatin, also increased intracellular ROS production and DNA oxidative damage. Therefore, our findings demonstrated the promising application of palladium(II) complexes as novel anti-leukemic agents.

www.nature.com/scientificreports/ Cationic dinuclear complexes are self-assembled by means of a non-conventional hydrogen bonding network of the type C-H···Cl-Pd and C-H···S (see Supplementary Table S1 online) to generate a 3D supramolecular metal organic matrix having helical channels along a-axis (see Supplementary Figure S1 online). The helical channels are defined with an approximate size of 8.33 × 4.92 Å acting as template for the helical water-chloride chain. These pores are filled by the four non-coordinated chlorine ion and the eight crystallization water molecules of the asymmetric unit resulting either left-handed as right-handed water-chloride helical chains in a 1:1 relation.
Water molecules and chlorine ions are self-assembled in the chain by means of a network of sixteen hydrogen bonding of which twelve are the type O-H···Cl − and four of the type O-H···O in a helical arrangement. Inside the chain, four water molecules of eight of crystallization (O1W, O7W, O8W, O2W) act as donors of hydrogen participating in the formation of two hydrogen bonds, each of which donates two hydrogen (O1W, O2W) to two chloride anions and (O7W, O8W) to two water molecules, while the other four (O3W, O4W, O5W, O6W) are involved in three hydrogen bonds each donating two hydrogen to two chloride ions (O3W, O4W to Cl6, Cl7 and O5W, O6W to Cl5, Cl8) and accepting one hydrogen. The water-chloride chain can be described as an infinite sequence of four-membered ring (2Cl − + 2H 2 O) and six-membered ring (2Cl − + 4H 2 O) arranged alternately joined together sharing a chloride ion and a water molecule with a T4(2)6(2) topology derived of water clusters assembly 27  The crystal packing between the water-chloride chain and metal-organic matrix is maintained by the formation of weak hydrogen bonds among some C-H atoms of pyridine, thiazoline and thiazolidine heterocycles of organic ligands and eight crystallization water of the chain, thereby generating fourteen C-H···O hydrogen bonds [average d(H···O) = 2.48 Å, average d(C···O) = 3.21 Å, average ∠(C-H···O) = 131°]t 28,29 and four non coordinated chlorine ions producing eight C-H···Cl − hydrogen bonds [average d(H··· Cl − ) = 2,81 Å, average d(C··· Cl − ) = 3,21 Å, average ∠(C-H··· Cl − ) = 139.7°] 30,31 (Table S1). Thus, these weak intermolecular interactions within the crystal of the type C-H···O and C-H···Clbetween the water-chloride chain and bimetallic cationic complexes build up a stable supramolecular assembly in which the water-chloride chains (guest) are supported by a metal-organic framework acting as host. . This feature is indicative for coordination through the imine and thiazoline nitrogen atoms 32 . Other relevant band is the one assignable to ν(H 2 O) stretching vibration at 3417 cm −1 due to crystallization water molecules in PdPyTT 26 . In the low-frequency region, the C 1 symmetry of PdPyTT predicts the appearance of four bands assignable to Pd-ligand stretching vibrations. The ν(Pd-Cl) vibration is registered at 325 cm −1 , which is in good agreement with literature data 33,34 . Finally, the ν(Pd-N) vibrations can be attributed to the bands at 465, 285 and 237 cm −1 for ν(M-N imine ), ν(M-N pyridine ) and ν(M-N thiazoline ) vibrations, respectively 33-38 . Biological studies. We have studied the potential cytotoxic effect and the proapoptotic capacity of the complex PdPyTT in human promyelocytic leukemia HL-60 cells, comparing it with cisplatin, a well-known chemotherapeutic agent. First, we analysed the effect of the complex PdPyTT on cellular viability of HL-60 cell line and compared it with that induced by cisplatin. Figure 2 shows the viability of HL-60 cells in the presence of cisplatin and PdPyTT. Stimulation of cells with increasing concentrations (5, 10, 20, 50 and 100 µM) of cisplatin or PdPyTT for 24 h induced a remarkable dose-dependent decrease ( Fig. 2A) in cell viability, reaching the minimum value of viability (12.1 ± 9.2% and 13.1 ± 3.6%, respectively, Fig. 2A) with the dose of 100 µM. The half maximal inhibitory concentrations (IC 50 ) were 20.7 µM and 11.3 µM for PdPyTT and cisplatin, respectively. Figure 2B shows that the effects of both the complex PdPyTT and cisplatin were also time-dependent, although the levels of cell viability were almost completely suppressed after 48 h of incubation with cisplatin (5.1 ± 3.8%, Fig. 2B). In addition, the free ligand (PyTT) produced negligible effect on HL-60 cells viability (see Supplementary Figure S7 online), except for the highest dose (100 µM) and after 72 h of incubation at 20.7 µM (dose comparable to IC 50 of PdPyTT). Similar results were obtained in human histiocytic lymphoma U-937 cells. Both PdPyTT and cisplatin produced a dose-dependent decline in cell viability (see Supplementary Figure S8 online). The IC 50 values in U-937 cells were 22.5 µM and 7.9 µM for PdPyTT and cisplatin, respectively. It is also important to point out that the complex PdPyTT (as well as the free ligand PyTT) was proved innocuous for freshly isolated human leukocytes, whilst cisplatin showed a high toxicity towards these healthy cells in which viability dropped by roughly 70% (Fig. 2C). Even though IC 50 of PdPyTT (20.7 µM) was two-fold higher compared to that of cisplatin (11.3 µM), the palladium(II) complex proved to be harmless for healthy human leukocytes. These findings are in good agreement with the cytotoxic activities found for palladium(II) complexes containing bis(2-pyridylmethyl)amine or saccharinate (IC 50 values ranging between 17.4 and 32.3 mM) in A549 lung carcinoma cells 39,40 , though extremely low IC 50 values (< 1 µM) have been reported for other palladium(II) complexes in brain glioma, breast cancer and fibrosarcoma cell lines [41][42][43] .
To test whether the reduction in cell viability was associated with apoptosis induction, we analysed the percentage of apoptotic nuclei in Hoechst 33342-stained cells. As shown in Fig. 3, nuclear staining with Hoechst 33342 revealed that the treatment of HL-60 cells with 20.7 µM PdPyTT or 11.3 µM cisplatin for 24 h produced nuclear fragmentation and/or condensation (31.2 ± 6.3 and 40.7 ± 7.1% apoptotic cells for PdPyTT and cisplatin, respectively; Fig. 3B), thus indicating that the main form of cell death was apoptosis. These findings were  Fig. 5C). These findings suggest that the complex PdPyTT displayed potent antitumor actions that relied on its ability to trigger caspase-9-dependent mitochondrial apoptosis, on one hand, and to cause excessive formation of cytosolic and mitochondrial ROS, on the other hand. Like results on cell death, the free ligand (PyTT) did not produce any substantial effect on the production of intracellular ROS or mitochondrial superoxide anion (Fig. 5).
Given the pro-apoptotic actions of the complex PdPyTT, its effect on cell cycle progression in HL-60 cells was anticipated and, therefore, a cell cycle distribution analysis was carried out. As shown in Fig. 6, HL-60 cells challenged with 20.7 µM PdPyTT exhibited a moderate, non-significant S phase arrest that was accompanied by noticeable (P < 0.05) diminution in the proportion of cells in both G0/G1 and G2/M subpopulations, while cells treated with 11.3 µM cisplatin depicted a significant (P < 0.05) accumulation of cells at G0/G1 phase and a remarkable (P < 0.05) reduction in cells for G2/M subpopulation. Besides, PdPyTT, but not cisplatin, caused substantial DNA fragmentation in HL-60 cells, as denoted by a significant increase in the subpopulation of cells with hypodiploid DNA content (sub-G1/G0; P < 0.05; Fig. 6). Again, the free ligand (PyTT) did not produce any significant change on cell cycle distribution (Fig. 6). www.nature.com/scientificreports/ Finally, we analysed DNA oxidative damage to study its potential contribution to apoptosis of HL-60 treated with the chemotherapeutic agents. Thus, treatment of HL-60 cells with 20.7 µM PdPyTT or 11.3 µM cisplatin for 24 h produced DNA oxidative damage as revealed by the increase in the proportion of cells positively labelled with 8-OHdG, a marker of DNA oxidative damage (Fig. 7). In this case, the effect of both cisplatin and PdPyTT was comparable since there was no significant statistical difference (45.3 ± 18.1 and 34.5 ± 11.9% of positive cells, respectively; Fig. 7A,B), while the free ligand (PyTT) again produced no effect on DNA oxidative damage. Such induction of DNA oxidative damage by both compounds (20-30% increase vs. control) agrees with the induction www.nature.com/scientificreports/ of mitochondrial ROS production (20% increase vs. control), which may suggest that mitochondrial oxidative stress is specially affecting the mitochondrial pool of DNA as it is extremely susceptible to oxidative damage 44 . Nonetheless, the mechanism of action to trigger the apoptotic process differs amongst both compounds (cisplatin and PdPyTT). In fact, cisplatin is a well-established DNA crosslinking agent that raised p53 (a tumor suppressor that can initiate apoptosis if DNA damage proves to be irreparable) protein expression levels (Fig. 8B,C), while the complex PdPyTT was shown to lack DNA intercalating capacity (Fig. 8A) and did not affect p53 expression (Fig. 8B,C). These findings were also in line with data on cellular accumulation ( Fig. 9) as, once internalized, the palladium(II) complex was preferentially found into the cytosol, wherein it triggered oxidative stress (Fig. 5B,C) that subsequently spread to mitochondria (Fig. 5A,C). Unlike cisplatin and various previously reported palladium(II)-based complexes [45][46][47][48] , PdPyTT did not interact with DNA as demonstrated by the inability of the complex to displace DNA-bound ethidium bromide (Fig. 8A), suggesting that cisplatin and PdPyTT-induced apoptosis in HL-60 cells were driven by different mechanisms. In fact, cisplatin-evoked apoptosis was dependent on genotoxic stress and subsequent p53 activation (the observed ROS overproduction was secondary to apoptosis), while PdPyTT caused p53-independent apoptosis through ROS-mediated mitochondrial dysfunction, ultimately leading to DNA oxidative damage. In this sense, previously reported palladium(II)-based complexes were shown to affect the expression of several key proteins involved in both the extrinsic pathway, such as caspase 8 and Fas death receptor 18,42,49 , and the intrinsic pathway, such as Bax and cytochrome c 50 . Likewise, some palladium(II)-based complexes have exhibited cytotoxicity against www.nature.com/scientificreports/ malignant cells that was associated to their prooxidative actions that led to increases in levels of either cytosolic ROS or mitochondrial superoxide anion 18,51,52 . conclusion Apoptosis induction via the intrinsic pathway is a chief mechanism triggered by different chemotherapeutic agents, including cisplatin 53 . Nevertheless, main limitations of canonical chemotherapy treatment are derived from undesirable side effects and development of resistance during treatment. In this context, new organometallic drugs acting through nonconventional mechanisms and with lower off target toxicity are necessary to improve therapeutic efficiency. Palladium(II)-based complexes represent an appealing alternative to classical antitumor drugs as they possess promising chemotherapeutic properties including covalent interaction with nitrogen bases of DNA, activation of either the extrinsic or the intrinsic apoptosis pathway, blockade of cell cycle progression, and induction of ROS overproduction that ultimately leads to oxidation of biomolecules 54 . Herein, we have synthesized a thiazolinebased palladium(II) complex that exhibited cytotoxicity towards human promyelocytic leukemia HL-60 cells, while it proved to be innocuous in normal non-cancer cells, such as freshly isolated human leukocytes. Additionally, the palladium(II) complex described in the present study showed pro-apoptotic properties dependent on mitochondrial pathway and intracellular ROS production. Therefore, the complex PdPyTT may be considered www.nature.com/scientificreports/ as promising candidate for developing new palladium-based antitumor/antileukemic agents, although its safety profile should be further tested in animal models to identify potential side effects.    and anti-glyceraldhehyde-3-phospate dehydrogenase (GAPDH) mouse monoclonal antibody (clone 6C5) were bought from Sigma Aldrich (Madrid, Spain). Caspase-9 staining kit (Red) was purchased from Abnova (Taipei, Taiwan). CellTiter 96 AQueous One Solution Cell Proliferation Assay was acquired from Promega (Madrid, Spain). anti-8-hydroxy-2′-deoxyguanosine (8-OHdG) mouse monoclonal antibody (clone N45.1) was obtained from JaICa (Fukuroi, Japan). Anti-p53 rabbit polyclonal antibody (clone FL-393) was bought from Santa Cruz Biotechnology (Heidelberg, Germany). Leuko Spin medium was obtained from pluriSelect (Leipzig, Germany). All other reagents were of analytical/commercial grade and were used without further purification.
Elemental analyses (C, H, N and S) were carried out on a Leco CHNS-932 microanalyzer. Infrared spectra were recorded on a Perkin-Elmer FT-IR 1720 spectrophotometer, from KBr pellets (4000-370 cm -1 ) and on a Perkin-Elmer FT-IR 1700X spectrophotometer, from Nujol mulls (500-150 cm -1 ). 1 H NMR spectra were obtained with a Bruker Avance 500 instrument at 500 MHz in DMSO-d 6 . ESI-MS was carried out with an Agilent Q-TOF 6520 instrument with the sample solved in DMSO and using water (0.1% formic acid) as mobile phase.

Synthesis of [PdCl 2 (µ-PyTT) 2 ]Cl 2 ·4H 2 O (PdPyTT). A solution of Na 2 [PdCl 4 ]·H 2 O (157 mg, 0.5 mmol)
in ethanol (5 ml) was mixed with an ethanol solution (30 ml) of PyTT (166 mg, 0.5 mmol), refluxed for 24 h and then cooled at room temperature. The resulting orange solid was filtered off, resulting in a mixture of substance where we did not purify any compound. The filtered solution was allowed to evaporate at room temperature. After 24 h, orange crystals suitable for X-ray diffraction analysis were isolated from the solution by filtration (230 mg, 45%). Orange solid turned out to be a mixture of compounds that could not be separated by recrystallization or chromatography. Anal. Calc. (%) for C 22  crystal structure determination. X-ray diffraction data were collected on a Bruker Kappa APEXII CCD diffractometer with Mo Kα radiation (λ = 0.71073 Å). Absorption corrections were applied using the program SADABS 57 . Using WINGX 58 , the structures were solved by direct methods (SHELXS-14 59 ) and refined by fullmatrix least-squares on F 2 (SHELXL-18 60 ). All non-hydrogen atoms were refined using anisotropic displacement parameters. Hydrogen atoms attached to carbon atoms were located in calculated positions and were refined using a riding model. However, hydrogen atoms of the water molecules were detected by Fourier differences and were refined with fixed O-H and H-H distances (0.957(3) and 1.513(3) Å, respectively). Graphical representations of the molecular structures were generated using ORTEP3 58 . CCDC 1988539 contains the supplementary crystallographic data for PdPyTT (Cambridge Crystallographic Data Center; copies of the data can be obtained, e-mail: deposit@ccdc.cam.ac.uk or www: https ://www.ccdc.cam.ac.uk). Experimental details of the crystal structure determination are listed in Table 2. cell culture and treatments. HL-60 and U-937 cells were grown in RPMI 1640 medium supplemented with 2 mM L-glutamine, 10% heat-inactivated foetal bovine serum, 100 U/ml penicillin and 100 µg/ml streptomycin at 37ºC under a humidified condition of 95% air and 5% CO 2 . Cells were routinely plated at a density of 3 × 10 5 cells/ml into fresh flasks and the viability was routinely kept at > 95% as assayed by the Trypan-blue exclusion method. Cells were treated with the free ligand PyTT, the thiazoline-containing palladium(II) complex PdPyTT, cisplatin (positive control) or vehicle for 24 h at the indicated concentrations. Dimethyl sulfoxide (DMSO) was used as vehicle and its final concentration did not exceed 0.1% (v/v).
Blood collection and isolation of human leukocytes. Venous blood was drawn from healthy volunteers aged 20-45 years. All volunteers gave informed consent. The procedure was carried out according to local ethical committee guidelines and the Declaration of Helsinki. In brief, leucocytes were separated from whole blood by using Leuko Spin medium. Samples were centrifuged at 1000 × g for 30 min at room temperature and then leukocyte cell fraction was isolated from the medium/plasma interface. Leukocytes were washed twice in Scientific RepoRtS | (2020) 10:16745 | https://doi.org/10.1038/s41598-020-73488-0 www.nature.com/scientificreports/ phosphate-buffered saline (PBS) at 300 × g for 10 min at 4 °C, the supernatant was removed, and the cell pellet was finally re-suspended in RPMI-1640 medium.
in vitro cytotoxicity assay. The cytotoxic effects of the different compounds were assayed on human promyelocytic leukemia HL-60 cell line and human leukocytes by means of the CellTiter 96 AQueous One Solution Cell Proliferation Assay, which is based on the reduction of an MTS tetrazolium compound. Cells were seeded in 96-well plates at a density of 2 × 10 4 cells/well. After treating cultures for 24 h, assays were performed by adding 10 µl of the CellTiter 96 AQueous One Solution Reagent directly to culture wells, incubating for 2 h (for HL-60 cells) or 4 h (for leukocytes) at 37 °C, and then recording absorbance on a microplate reader (Infinite M200; Tecan) at both a test wavelength of 490 nm and a reference wavelength of 650 nm to subtract background. All analyses were run in triplicate. The cell viability was calculated as percentage of control values (untreated samples).

Determination of apoptosis.
To evaluate apoptosis, the fraction of apoptotic nuclei was quantified by fluorescence microscopy after DNA staining with Hoechst 33342. The dye was used at a final concentration of 1 µg/ml. Then, at least 100 cells per field were counted at the epifluorescence microscope (Nikon Eclipse TS100 equipped with a LED illumination system CoolLED pE-300 white and a digital camera DS-Qi1Mc) in at least three randomly selected microscopic fields. Hoechst 33342 was excited using a single-band filter set, optimized for DAPI and other like fluorophore (Semrock). The images were analysed with the macro "Apoptosis and Cell Density Macro" available in the open source image processing package Fiji/ImageJ (https ://image jdocu .tudor .lu/ plugi n/morph ology /apopt osis_and_cell_count _macro /start ) and percentage of apoptotic cells was computed. Alternatively, induction of apoptosis was determined by analysing phosphatidylserine externalization via annexin V-FITC/PI assay (ThermoFisher Scientific). Briefly, stimulated cells were harvested (4 × 10 5 cells/ml), washed once with PBS, and centrifuged at 300 × g for 5 min; then, the supernatant was discarded, and the pellet was resuspended in 200 µl binding buffer containing 5 µl of annexin V-FITC. Cells were incubated for 10 min at room temperature, washed once with binding buffer, and finally resuspended in 200 µl binding buffer containing 10 µL of PI. Stained cells were immediately analysed by a MACSQuant X flow cytometer (Miltenyi Biotech). Annexin V-FITC was excited with 488 nm laser and its fluorescence was acquired through 525/50 nm BP filter, while PI was excited with 488 nm laser and its fluorescence was acquired through 585/40 nm BP filter. Ten www.nature.com/scientificreports/ thousand events/cells were analysed per condition. Each sample was tested six times in independent experiments. The sum of early apoptosis and late apoptosis was calculated to obtain the total percentage of apoptotic cells 61,62 . Caspase-3 and -9 analysis. Cells were seeded in 6-well plates at a density of 2 × 10 6 cells/well. After treating cultures for 24 h, caspase-3 and -9 activation was measured with CaspGLOW fluorescein active caspase-3 staining kit and Caspase-9 staining kit (Red), respectively. Flow cytometry analysis was performed using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotech) equipped with 3 lasers (405, 488 and 561 nm). CaspGLOW fluorescein was excited with 488 nm laser and its fluorescence was acquired through 525/50 nm BP filter, while Red-LEHD-FMK (fluorescent marker for caspase-9) was excited with 561 nm laser and their fluorescence was acquired through 586/15 nm BP filter. Analysis was restricted to live cells that were detected by using the vital dye Hoechst 33,258 (10 µg/ml), which was excited with 405 nm laser and its fluorescence acquired through 450/50 nm BP filter. Ten thousand events/cells were analysed per condition. Each sample was tested six times in independent experiments. Data were presented as percentage of stained cells. Western blot. HL-60 cells were grown on 10 cm Petri dishes and challenged with the indicated compounds.

Measurement of reactive oxygen species (RoS
After treatments, cells were washed twice with PBS, re-suspended in RIPA buffer (150 mM NaCl, 50 mM Tris-Cl pH 7.4, 1% Triton-X100, 0.5% sodium deoxycholate, 0.1% SDS and Complete EDTA-free protease inhibitor mixture) and briefly sonicated. Crude extracts were centrifuged at 15,000 × g for 10 min to remove debris, and proteins in the supernatant were then quantified using the BCA Protein Assay Kit (ThermoFisher Scientific). Forty μg of proteins were dissolved in LDS sample buffer, heated for 10 min at 70 °C and loaded on 4-12% Bis-Tris NuPage gels. After electrophoretic separation, proteins were transferred onto nitrocellulose membranes and probed with the indicated primary antibodies. Isotype matched horseradish peroxidase-conjugated secondary antibodies were used following by detection by chemiluminescence (Super-Signal West Dura). Densitometry was performed using the built-in function present in the image processing package Fiji/ImageJ 65 . For the shake of clarity, full-length blots (Supplementary Figure S11) were cropped and merged in a single image (Fig. 8) with the software package Adobe Illustrator CS6 as they were obtained from different parts of the same membrane.
DnA intercalation assay. DNA intercalation was assessed by examining the ability of the complex PdPyTT to displace ethidium bromide from supercoiled and/or relaxed pBR322 plasmid. To do so, DNA unwinding assay kit (Inspiralis) was used following manufacturer's instructions. Briefly, 0.5 µg of DNA (supercoiled or relaxed pBR322) were added to 15 µl of 2 × assay buffer (50 mM Tris-HCl (pH 7.9), 1 mM EDTA, 1 mM DTT, 20% (v/v) glycerol, 50 mM NaCl) and then incubated with 20.7 µM PdPyTT or the vehicle (negative control) for 5 min at RT. Afterwards, 2 µl of 1 × dilution buffer (50 mM Tris-HCl (pH 7.9), 1 mM EDTA, 1 mM DTT, 50% Scientific RepoRtS | (2020) 10:16745 | https://doi.org/10.1038/s41598-020-73488-0 www.nature.com/scientificreports/ (v/v) glycerol, 500 mM NaCl) and 2 µl of wheat germ Topoisomerase I were added to initiate unwinding and samples were incubated at 37 °C for further 30 min. Reactions were terminated by addition of 20 µl of water and 50 µl of butanol. Reactions were separated on 1% (w/v) agarose gel in 1 × TBE (Tris-borate-EDTA) buffer and gel was run at 80 V for approximately 4 h, stained with ethidium bromide, and visualised with a gel documentation system (ChemiDoc XRS + System, Bio-Rad). For the shake of clarity, full-length gels (Supplementary Figure S10) were cropped and merged in a single image (Fig. 8) with the software package Adobe Illustrator CS6 as they were obtained from different parts of the same gel. cellular uptake by icp-MS. To quantify the metal accumulation in different fractions of the cells, HL-60 cells were seeded in culture flasks (2 × 10 6 cells in 5 ml RPMI-1640 culture medium). After 24 h in culture, cells were treated for 5 h with 20.7 µM of PdPyTT, were then harvested and subcellular fractionation was performed to obtain different cell fractions: F1: cytosolic fraction; F2: membrane/organelle fraction; F3: nuclear fraction; F4: cytoskeletal fraction. In this sense, cells were fractionated by means of a ProteoExtract subcellular proteome extraction kit. Afterwards, the different samples were digested with 65% HNO 3 at room temperature and analysed by inductively coupled plasma mass spectrometry (ICP-MS 7900, Agilent Technologies).
Statistical analysis. Data are presented as mean ± standard deviation (S.D.). To compare the different treatments, statistical significance was calculated by one-way analysis of variance (ANOVA) followed by Dunnett's test. The IC 50 of the dose-response curve of each compound was calculated by fitting the curve to the data using nonlinear regression to generate a four-parameter sigmoid dose-response equation. P < 0.05 was considered to indicate a statistically significant difference. The statistics software used was GraphPad Prism 7.04 for Windows.