Thiabendazole inhibits glioblastoma cell proliferation and invasion targeting MCM2

Thiabendazole (TBZ), approved by the U.S. Food and Drug Administration (FDA) for human oral use, elicits a potential anti-cancer activity on cancer cells in vitro and in animal models. Here, we evaluated the efficacy of TBZ in the treatment of human glioblastoma multiforme (GBM). TBZ reduced the viability of GBM cells (P3, U251, LN229, A172, and U118MG) relative to controls in a dose- and time-dependent manner. However, normal human astrocytes (NHA) exhibited a greater IC50 than tumor cells lines and were thus, more resistant to its cytotoxic effects. EdU positive cells and the number of colonies formed was decreased in TBZ-treated cells (at 150 μM, P < 0.05 and at 150 μM, P < 0.001, respectively). This decrease in proliferation was associated with a G2/M arrest as assessed with flow cytometry, and the downregulation of G2/M check point proteins. In addition, TBZ suppressed GBM cell invasion. Analysis of RNA sequencing data comparing TBZ treated cells with controls yielded a group of differentially expressed genes, the functions of which were associated with the cell cycle and DNA replication. The most significantly downregulated gene in TBZ-treated cells was mini-chromosome maintenance protein 2 (MCM2). SiRNA knockdown of MCM2 inhibited proliferation, causing a G2/M arrest in GBM cell lines and suppressed invasion. Taken together, our results demonstrated that TBZ inhibited proliferation and invasion in GBM cells through targeting of MCM2. Significance Statement TBZ inhibits the proliferation and invasion of glioblastoma cells by downregulating the expression of MCM2. These results support the repurposing of TBZ as a possible therapeutic drug in the treatment of GBM.


Introduction
Glioblastoma multiforme (GBM) is the most common and malignant primary brain tumor in human adults with only 14.6 months of median survival after primary diagnosis (Darlix et al., 2017), despite a standard therapeutic regimen consisting of surgery, radiotherapy and chemotherapy (Ostrom et al., 2020). The current chemotherapy used is often temozolomide, an oral DNA alkylating agent, which in combination with radiotherapy following surgery has increased patient survival from 12.1 to 14.6 months (Stupp et al., 2005). However, at least 50% of GBM patients do not respond (Lee, 2016). Several biological properties of GBM render the disease resistant to treatment. First, GBM cells filtrate the peripheral normal brain tissue, making complete removal of the tumor with surgery impossible (Bell and Karnosh, 1949;Shergalis et al., 2018). Second, most chemotherapeutic molecules insufficiently permeate the brain due to the blood-brain barrier (BBB) (Shergalis et al., 2018). Third, targeting of key molecular pathways is ineffective because of the high cellular and genetic heterogeneity within GBM (Brennan et al., 2013). Thus, novel effective drugs and therapeutic targets are urgently needed for GBM treatment.
Drug repurposing has become a widely accepted strategy in oncology to identify new therapies. Drugs already known to be safe in humans accelerate the initiation of clinical trials needed especially in the case of cancers with few treatment options. For instance, flubendazole and mebendazole are benzimidazole carbamate family compounds approved for use as anthelminthics in humans and have been studied for their anticancer properties against diverse cancers including human glioma.

6
Flubendazole has been shown to inhibit glioma proliferation and tumorigenesis.
Mebendazole was shown to be cytotoxic to glioma and significantly prolonged mean survival in syngeneic and xenograft orthotopic animal glioma models (Bai et al., 2011).
In a phase 1 clinical trial, mebendazole demonstrated long-term safety and acceptable toxicity at doses of up to 200 mg/kg (Gallia et al., 2021). Another benzimidazole, thiabendazole (TBZ; tiabendazole; 2-(thiazol-4-yl) benzimidazole), has been used to treat gut parasites in humans for over 50 years (Campbell and Cuckler, 1969;Whalen et al., 1971). TBZ inhibits blastocysts, candida albicans, penicillium and psoriasis, and prevents the formation of aflatoxin in plant feed, but it does not affect carcinogenesis and fertility in animals (Gosselin et al., 1984). A previous study demonstrated that TBZ reduced the growth of human fibrosarcoma (Cha et al., 2012). Therefore, as a non-toxic member of the family of benzimidazole compounds, TBZ has gained interest for its potential as an anticancer therapy in humans.
In this study, we examined the anticancer effects of TBZ and investigated its potential molecular mechanisms in GBM cells in vitro and in vivo. We demonstrated that TBZ induces G2/M arrest in GBM cells and inhibits invasion. We performed RNA sequencing on the tumor cells treated with TBZ to identify differentially expressed genes and found mini-chromosome maintenance protein 2 (MCM2) to be a key transcriptional factor downregulated by TBZ, showing the MCM2 as a molecular target of TBZ. Finally, we determined that TBZ inhibits GBM cells growth in vivo.
After 24 h, the medium was replaced with 100 μL of fresh culture medium containing different concentrations of TBZ or vehicle control (DMSO). At 24 h, 48 h, 72 h, and 96 h after dosing, GBM cells were incubated with 10 μL of CCK-8 reagent in 100 μL of serum-free DMEM at 37°C for 1 h and an EnSight Multimode Plate Reader (PerkinElmer; Singapore) was used to measure the absorbance at 450 nm. Cell viability of GBM cells transfected with MCM2 siRNA and overexpression constructs was also assessed with the CCK-8 assay.

Cell proliferation assay
1×10 4 of GBM cells were seeded into each well of 24-well plates (Corning) and cultured in a humidified incubator at 37°C, 5% CO2. 5-ethynyl-2'-deoxyuridine (EdU) was diluted 1:1000 in DMEM complete medium, and GBM cells were treated with the This article has not been copyedited and formatted. The final version may differ from this version.

Colony formation assay
U251 and P3 cells were counted, the cell density was diluted to 500 cells/mL, and 2 mL of the cell suspension was added to each well of a 6-well plate (Corning). The drug concentrations used were the following: 0 μM (DMSO), 150 μM and 300 μM, following 7 days culture. The medium was then replaced with fresh medium and cells were continued culture for an additional 7 days. The culture medium was discarded, each well of the 6-well plate was rinsed with 500 μL PBS (3x), and the cells were fixed with 4% paraformaldehyde for 15 min. Each well of the 6-well plate was rinsed with 500 μL PBS (3x) and cells were stained for 30 min with crystal violet. The wells were slowly rinsed with double distilled water. Clones were counted after air drying the wells.
Clones were counted if the number of cells was > 50.

Protein lysates and Western blotting
GBM cells were lysed with RIPA Lysis Buffer (Beyotime; Shanghai, China) supplemented with phenylmethanesulfonyl fluoride (PMSF, Beyotime) for 30 minutes This article has not been copyedited and formatted. The final version may differ from this version. Cell lysates (20 μg protein) were subjected to western blot analysis, according to previously described protocols (Kong et al., 2019). Membranes were incubated with primary antibody at 4 overnight followed by incubation with appropriate secondary antibodies (1:2000) for 1 h at room temperature. Chemiluminescent signals were imaged using the Chemiluminescence Imager (Bio-Rad; Hercules, CA, USA) according to the manufacturer's protocol. Band density was quantified with ImageJ software and normalized to GAPDH or -actin. All experiments were repeated three times.

Cell cycle
U251 and P3 cells were diluted to 4×10 5 cells /mL and 2 mL of the cell suspension was seeded into each well of a 6-well plate, and cultured overnight. TBZ were added to cells at the following concentrations: 0 μM (DMSO), 150 μM and 300 μM. After 2 days, cells were rinsed and harvested at 4000 rpm/min for 5 min, and gently fixed in fresh 300 μL PBS and 700 μL 75% ethanol. Cells were incubated at 4℃ overnight, harvested at 4,000 rpm/min for 5 min, rinsed with PBS at 4,000 rpm/min for 5 min. The cell This article has not been copyedited and formatted. The final version may differ from this version. dye solution for 20 min and then followed by flow cytometry for cell cycle analysis. In gate P1, the linear relationship was set for F2L-A and F2L-H. Gate P2 is looped to select diploid and tetraploid cells, and 10,000 events are collected under P2 conditions. Data was exported, and Modifit 2.0 software was used to determine the cell cycle distribution. GBM cells transfected with MCM2 siRNA and MCM2 overexpression constructs were similarly processed to obtain cell cycle parameters.

Trans-well invasion assay
U251 and P3 cells were diluted to 4×10 5 cells /mL, and 2 mL of cell suspension was seeded into each well of a 6-well plate. The cells were incubated under different conditions for 48 h. Trans-well migration plates with 8 μm pore size (Corning; Sigma-Aldrich) were coated with Matrigel (Becton-Dickinson; Bedford, MA, USA) for 4 h. 20,000 cells in 100 μL DMEM without FBS were seeded into the upper chamber of a trans-well apparatus and 600 μL medium containing 10% FBS was added to the lower chamber. After 24 h incubation at 37°C, cells remaining were removed from the top side of the insert with a cotton swab and the migrated cells were fixed with 4% paraformaldehyde for 15 min, rinsed twice with PBS and stained with crystal violet for 30 min. The dye was removed, and cells were rinsed with double distilled water. Images from 3 random views under a light microscope were acquired and used to count migrated cells.

Cell invasion in 3D culture
3,000 cells were seeded into each well of 3D Culture Qualified 96-well spheroid This article has not been copyedited and formatted. The final version may differ from this version. Plates were placed on ice for 15 min, and 50 μL of invasion matrix (Trevigen, 3500-096-03) was added to each well in the plates. Plates were centrifuged at 300×g at 4℃ for 5 min and incubated at 37℃ for 1 h. Conditioned medium (100 μL) with different concentrations of TBZ was added to each well of the plates. P3 tumor spheroids were incubated for 3 days and U251 tumor spheroids were incubated for 10 days. Images of the spheroids were captured every 24 h under bright field microscopy with a 4× objective. The 192-hour images of U251 and the 48-hour image of P3 were analyzed with the software ImageJ. GBM cells after transfection with MCM2 siRNA or MCM2 overexpression constructs were also assessed in 3D invasion culture as described above.

RNA-seq and bioinformatics analysis
The RNA-Seq libraries were prepared using the Illumina TruSeq™ RNA sample preparation Kit (Illumina, San Diego, CA), and sequenced through paired-end (150 base paired-end reads) sequencing performed on the Illumina NovaSeq 6000 platform.
Raw data was then quality filtered to generate "clean reads" for further analysis. The "clean reads" were then aligned to the human genome reference (hg19) using STAR software and the reference-based assembly of transcripts was conducted using HISAT2.
We used picard to compare the results and to remove redundancy, and used Sentieon software to detect single nucleotide variations (SNVs) and InDels. All previously identified SNVs and InDels were determined by using the dbsnp database. Gene expression values were expressed as reads per kilobase of exon per million fragments This article has not been copyedited and formatted. The final version may differ from this version.
The DEGs (P-value ≤ 0.05, |Log2FC|≥1) were subjected to enrichment analyses of GO and KEGG pathways. Protein-to-protein interaction network analyses of DEGs was performed using the STRING database and the protein-protein interaction networks were visualized with Cytoscape software.

Lentiviral transduction
This article has not been copyedited and formatted. The final version may differ from this version. Lentiviral vectors expressing human mRNA targeting MCM2 (GenePharma, Shanghai) or scrambled-control (negative control) were used to generate stable cell clones overexpressing MCM2 or a nonspecific RNA as the control. Transfected clones were selected in 1 mg/mL of puromycin (Selleckchem; Houston, TX, USA) for 2 weeks.
Western blot analysis was used to evaluate the transduction efficiency.

Orthotopic xnograft model
P3 cells expressing luciferase-GFP (X×10X; OBiO Technology; Shanghai, China) were implanted into the brains of nude mice. After 7 days, tumor was determined using bioluminescence imaging (PerkinElmer IVIS Spectrum; Waltham, MA, USA), and the mice were divided into the following 2 groups: control, n = 5; TBZ, n = 5. Mice were intraperitoneally injected with diluted DMSO alone (control) and TBZ (50 mg/kg/day) every day. Tumor volume was monitored using the bioluminescence imaging every week for 3 weeks, and the weight of each mouse was recorded every week for 4 weeks.
Tumor bearing nude mice were treated until severe symptoms or death appeared imminent: the body weight had decreased > 10% and mice were unable to return upright after being pushed down. Survival (in days) was determined as the number of days starting from implantation (day 1) to death. Mice were euthanized at the end of the experiment. Excised tumor tissue was snap frozen in liquid nitrogen or formalin-fixed for further analysis.

Liquid chromatography-tandem mass spectrometry analysis
The nude mice were separated into TBZ-treated group (intraperitoneal injection, 3 mice) and control group (3 mice). Two hours after the injection, the mice were This article has not been copyedited and formatted. The final version may differ from this version. anesthetized with chloral hydrate solution (Qilu hospital, China). PBS was perfused through the heart. The mice were sacrificed by CO2 inhalation and the brain samples were collected and stored at -80°C. The TBZ and the brain samples were further analyzed by liquid chromatography-tandem mass spectrometry to examine the distribution of TBZ within the brain tissue. In brief, the tissue samples were weighed and appropriate amounts of methanol (chromatographically pure; Thermo Fisher, USA) and zirconia grinding beads were added (Servicebio, Wuhan, China). The samples were ground for 5 min after vortexing for 10 min, and then centrifuged at 15000 RPM for 10 min (centrifuge: D3024R, Dragonlab, Beijing, China). The supernatant was collected and diluted for the analysis on the UltiMate 3000 RS (ThermoFisher Scientific, USA) and TSQ Quantum (ThermoFisher Scientific, USA) instruments.

Plotting and statistical analysis
At least three times for each assay was independently conducted. All analyses were performed using GraphPad Prism 8.02 software (San Diego, CA, USA). Data were reported as the mean ± SD. The statistical significance of data was evaluated using a Student's t-test and the following p-values: *P < 0.05; **P < 0.01; ***P < 0.001 were considered to be indicated as significant differences.

Thiabendazole induces G2/M arrest in GBM cells
To determine whether TBZ is cytotoxic to GBM, we first exposed GBM cell lines and NHA to TBZ in vitro. The viability of all cells tested including P3, U251, LN229, This article has not been copyedited and formatted. The final version may differ from this version. A172, U118MG and NHA, decreased in a dose-dependent manner with increasing concentrations of TBZ ( Figure 1A). The IC50 of NHA was at least 100 μM greater than for all other cell lines indicating TBZ might be selective for tumor cells at certain concentrations ( Figure 1B). In the functional experiments, we then chose P3, representing a primary GBM cell line and U251, representing one of most common GBM laboratory cell lines. In the colony forming assay, colony numbers were decreased by ~ 50% for P3 and U251 cells with 150 μM TBZ, and decreased by 90% for P3 and 75% for U251 with 300 μM TBZ (Supp. Fig. 1A and 1B). jpet.aspetjournals.org Downloaded from more than 60% in 300 μM TBZ ( Figure 1F). These results demonstrated that levels of key checkpoint proteins paralleled cell cycle arrest induced by TBZ in GBM cell lines.

Thiabendazole inhibits invasion of GBM cells
To determine whether TBZ might inhibit infiltration capabilities of GBM cells, we examined GBM cells under TBZ treatment in trans-well and Matrigel assays. In transwell assays, the number of P3 cells penetrating the membrane was reduced from > 250 (0 μM TBZ) to ~ 130 (150μM TBZ) and to no more than 60 (300 μM TBZ). The number of invasive U251 cells was reduced from ~ 250 (0 μM TBZ) to ~ 100 (150 μM TBZ) and ~ 35 (300 μM TBZ). After 48 h of TBZ treatment, the number of cells in both cell lines were reduced in a dose-dependent manner relative to controls (0 μM; Figure 2A).
We also measured the invasive ability of GBM spheroids derived from P3 and U251 cells in suspension culture. The invasive areas of P3 spheres in Matrigel were also decreased to 65% (150 μM TBZ) and 27% (300 μM TBZ), and the invasive areas of U251 were decreased to 35% (150 μM TBZ) and 17% (300 μM TBZ) after exposure to TBZ relative to controls (0 μM TBZ; Figure 2B). Increasing TBZ concentrations furthermore led to reduced invasion ( Figure 2B). In western blots performed on lysates prepared from P3 and U251 cells treated with TBZ, invasion-related proteins associated with EMT, such as N-cadherin, ZEB1 and MMP2, were downregulated > 30% with 150 μM TBZ and > 50% with 300 μM TBZ. Protein levels decreased in response to TBZ in a dose-dependent manner ( Figure 2C). These results indicated that TBZ suppressed invasion of GBM cells and inhibited expression of proteins involved in EMT.

MCM2 is significantly downregulated in TBZ-treated glioma cells
This article has not been copyedited and formatted. The final version may differ from this version. To identify potential gene targets of TBZ, we performed RNA sequencing on RNA isolated from GBM cells treated with the molecule. GO and KEGG analysis of the resultant differentially expressed genes showed that TBZ treatment altered expression of genes associated with the cell cycle, mitosis and DNA replication ( Figure   3A and 3B). Through protein-protein interaction enrichment analysis, we selected out a protein-protein interaction network involving genes regulating the G2/M phase of the cell cycle ( Figure 3C and 3D). Of the top 10 differentially expressed genes in both P3 and U251 cell lines, MCM2, UHRF1, and MCM5 showed the greatest difference in expression levels between treated and untreated cells ( Figure 3E). However, we found only MCM2 to be significantly downregulated also at the protein level in both TBZtreated P3 and U251 cell lines (Supp. Fig. 2). Moreover, MCM2 protein levels decreased in a dose-dependent manner ( Figure 3F).
In Kaplan-Meier analysis performed with expression data from the TCGA and CGGA datasets, we found MCM2 to be increased in GBM and low-grade gliomas.
Furthermore, high expression of MCM2 was related to poor survival of glioma patients ( Figure 3G and 3H). These results indicate that TBZ plays a role in causing G2/M cell cycle arrest in GBM cells which is possibly mediated through the downregulation of MCM2. Thus, MCM might be a novel therapeutic target for the treatment of human glioma.

Overexpression of MCM2 reverses TBZ induced suppression of GBM cell proliferation and invasion
To determine whether increased MCM2 expression interfered with TBZ induced growth arrest in GBM cells, we created stably expressing cells through infection with lentiviral constructs expressing MCM2. P3-and U251-MCM2-OE cells showed enhanced proliferation relative to uninfected or TBZ-treated cells ( Figure 5A).
Overexpression of MCM2 led to a reduced percentage of cells (Negative control; MCM2 overexpression, MCM2 OE; TBZ 300 μM; TBZ 300 μM + MCM2 OE) in This article has not been copyedited and formatted. The final version may differ from this version.  (Figure 5B and Supp. Fig. 4) Figure 5D and Supp. Fig. 5 Figure 5E). In conclusion, these results indicated that overexpression of MCM2 reversed TBZ induced inhibition of proliferation and invasion in GBM cells.

TBZ inhibits growth of GBM cells in vivo
To examine whether TBZ inhibits GBM cell growth in vivo, we assessed its effects in an GBM tumor model derived from P3-luciferase expressing cells intracranially implanted in nude mice. Tumor growth was monitored using luciferase bioluminescence. TBZ treatment significantly inhibited tumor growth compared to vehicle control in the mice ( Figures 6A and 6B), and the weight of TBZ-treated mice also did not decrease as rapidly relative to controls at the 2-and 3-week time points after treatment ( Figure 6C). The day of death of control mice (DMSO) were the following: day 32, 35, 39, 39, and 41. The day of death of TBZ-treated mice were the following: day 39, 43, 44, 48, and 55. overall survival between control and TBZ-treated animals Kaplan-Meier analysis of the survival data also demonstrated a statistically significant difference for (median survival time 39 days vs. 44 days, controls vs. treated animals; Figure 6D). Immunohistochemistry performed on tissue sections from xenografts demonstrated that Ki67, a marker of cell proliferation, was decreased by ~ 50% in TBZ-treated tumors compared to untreated controls ( Figure 6E). In addition, the expression of MCM2 was significantly decreased in xenografts from TBZ-treated mice relative to controls. Thus, TBZ inhibited tumor cell growth in vivo, possibly This article has not been copyedited and formatted. The final version may differ from this version. JPET Fast Forward. Published on November 8, 2021as DOI: 10.1124 at ASPET Journals on January 11, 2022 jpet.aspetjournals.org Downloaded from 23 through the suppression of MCM2 expression. We also examined the TBZ distribution in the brain tissue of nude mice by the liquid chromatography-tandem mass spectrometry assay. We found that the TBZ-treated sample had a transition of m/z 174.53-175.53 for an ion peak at 2.69 min (Supp. Fig. 6). This value was consistent with that of the TBZ standard solution, confirming that TBZ is capable of delivery into the tumor area in the brain. This article has not been copyedited and formatted. The final version may differ from this version.

Discussion
Benzimidazole carbamate derivatives are approved compounds for the treatment of parasitic diseases in humans. Such compounds include mebendazole, albendazole, fenbendazole, thiabendazole and flubendazole. They kill worms by binding and inhibiting beta-tubulin (Cumino et al., 2009). TBZ was FDA approved in 1967 for use in humans and has been used as antifungal treatment for a half century. Previous reports indicated that TBZ has anti-tumor effects in melanoma and fibrosarcoma, including inhibition of proliferation and migration in melanoma B16F10 cells and angiogenesis in fibrosarcoma   Loss of cell cycle checkpoint control underlies the aggressive proliferation and dysregulation of the cell cycle associated with GBM. Thus, therapies have been designed to inhibit the cell cycle (Dominguez-Brauer et al., 2015). The mechanism of action of many microtubule inhibitors involves inhibition of the G2/M phase (Castro-Gamero et al., 2018). These drugs may be synergistic with the current standard of GBM therapy (temozolomide or radiotherapy) either by facilitating DNA damage or sensitizing malignant cells to standard therapy (Vitovcova et al., 2020). In future studies, we plan to explore the effect of the combination treatment of TBZ, TMZ and This article has not been copyedited and formatted. The final version may differ from this version. JPET Fast Forward. Published on November 8, 2021as DOI: 10.1124 at ASPET Journals on January 11, 2022 jpet.aspetjournals.org Downloaded from radiotherapy on GBM.
RNA sequencing revealed potential targets of TBZ in GBM cells. Using GO, KEGG and protein-protein interaction network analysis, we found that TBZ regulates the expression of proteins that function in cell proliferation and the cell cycle. MCM2 was among the most highly differentially expressed genes under TBZ treatment, and overexpression of the gene rescued TBZ-treated cells from inhibition of cell growth. As the cytotoxic effect of TBZ (> IC50) was lower in NHA than in GBM cells, the molecule might be selective for tumor cells at certain concentrations in the clinical management of patients.
The function of MCM is regulated at elongation and termination of DNA replication (Brewster and Chen, 2010;Li et al., 2015;Seo and Kang, 2018). In the process of carcinogenesis, the dysfunction of MCM generates instability in the structure of the DNA fork, thus creating conditions for the acquisition of the gene mutations driving tumor development. As a member of the MCM family, MCM2 has been shown to be overexpressed in various tumors, including hepatocellular carcinoma (Yang et al., 2018;Yang et al., 2019), pancreatic adenocarcinoma (Peng et al., 2016;Xi and Zhang, 2018), lung cancer (Cheung et al., 2017), breast carcinoma (Yousef et al., 2017;Issac et al., 2019), ovarian cancer (Deng et al., 2019) and cervical cancer (Mukherjee et al., 2007;Amaro Filho et al., 2014). MCM2 was predicted to be a valuable prognostic biomarker in breast cancer , cervical cancer (Wu and Xi, 2021) and neuroendocrine prostate cancer (Hsu et al., 2021). In addition, MCM2 was suggested to be a potential treatment target to breast cancer and prostate cancer (Hsu et al., 2021; This article has not been copyedited and formatted. The final version may differ from this version. JPET Fast Forward. Published on November 8, 2021as DOI: 10.1124 at ASPET Journals on January 11, 2022 jpet.aspetjournals.org Downloaded from Liu et al., 2021). In neuroblastoma, MCM2 expression is positively correlated with tumor growth, and thus the gene is a novel potential target for neuroblastoma pharmacological treatment (Garbati et al., 2020). Our bioinformatic analysis showed that MCM2 is upregulated in GBM tissue (TCGA) and related to decreased survival in glioma patients (CGGA). Silencing MCM2 through siRNA knockdown inhibited proliferation and invasion of GBM cells. In contrast, the overexpression of MCM2 partially rescued GBM cells from cell cycle arrest and reduced invasion under TBZ treatment. These results suggest that MCM2 is a critical molecular target of TBZ, and warrants further study as a biomarker for TBZ as a potential treatment for GBM.  (F) Western blot to detect levels of cyclin B1, cyclin B2, CDK1 and PCNA and β-actin in P3 and U251. All data are expressed as the mean ± SD of values from triplicate experiments. * P < 0.05, ** P < 0.01 and *** P < 0.001 compared to controls. concentrations for 48 h. All data are expressed as the mean ± SD of values from triplicate experiments. * P < 0.05, ** P < 0.01 and *** P < 0.001 compared to controls.