Apoptosis Signal-regulating Kinase 1 promotes Ochratoxin A-induced renal cytotoxicity

Oxidative stress and apoptosis are involved in Ochratoxin A (OTA)-induced renal cytotoxicity. Apoptosis signal-regulating kinase 1 (ASK1) is a Mitogen-Activated Protein Kinase Kinase Kinase (MAPKKK, MAP3K) family member that plays an important role in oxidative stress-induced cell apoptosis. In this study, we performed RNA interference of ASK1 in HEK293 cells and employed an iTRAQ-based quantitative proteomics approach to globally investigate the regulatory mechanism of ASK1 in OTA-induced renal cytotoxicity. Our results showed that ASK1 knockdown alleviated OTA-induced ROS generation and Δψm loss and thus desensitized the cells to OTA-induced apoptosis. We identified 33 and 24 differentially expressed proteins upon OTA treatment in scrambled and ASK1 knockdown cells, respectively. Pathway classification and analysis revealed that ASK1 participated in OTA-induced inhibition of mRNA splicing, nucleotide metabolism, the cell cycle, DNA repair, and the activation of lipid metabolism. We concluded that ASK1 plays an essential role in promoting OTA-induced renal cytotoxicity.

In the present study, we combined RNAi of ASK1 with an iTRAQbased quantitative proteomics approach to globally profile the function of ASK1 in OTA-induced renal cytotoxicity. We performed a stable knockdown of ASK1 in the human embryonic kidney (HEK293) cell line and compared the proteome between ASK1 knockdown cells and scrambled cells following OTA treatment. In summary, this study, for the first time, showed the function of ASK1 in OTA-induced renal cytotoxicity using a combination of RNAi technology and iTRAQ-based quantitative proteomics.

Results
OTA induced ASK1 activation. Since ASK1 was discovered by   25 , it has drawn much attention for its role in cell apoptosis. ASK1 plays a key role in oxidative stress-induced apoptosis through Thr838 phosphorylation 26,27 . Because OTA is capable of inducing oxidative stress and apoptosis, we speculated that ASK1 might be involved in OTA-induced apoptosis. Western Blot analysis of ASK1 phosphorylation confirmed this hypothesis. As shown in Figure 1, ASK1 activity reached its peak at 1 h following OTA treatment and then decreased with the duration of OTA treatment.
Confirmation of RNA interference efficiency. To further investigate the role of ASK1 in OTA-induced renal cytotoxicity, we knocked down ASK1 expression using RNA interference. The interference efficiency of ASK1 knockdown cells versus scrambled cells was confirmed by Western blot. As shown in Figure 2, ASK1 shRNA transfection markedly reduced the expression of ASK1 to approximately 54% compared with that of cells transfected with scrambled shRNA.
ASK1 knockdown desensitized cells to OTA. The cell viability of ASK1 knockdown and scrambled cells after 24 h exposure to increasing concentrations of OTA were determined by WST-8 staining. As shown in Figure 3, OTA treatment caused a decrease of cell viability in a dose-dependent manner; 20 mM OTA treatment caused the cell viability to decrease to 46.4% and 54.7% in scrambled and ASK1 knockdown cells, respectively (Figure 3). Because a dose of 20 mM was close to the IC 50 for both cell lines, it was chosen for the following study. Interestingly, the cell viability of ASK1 knockdown cells after OTA treatment was slightly higher than that of scrambled cells. Sturchler et al. showed that ASK1 overexpression promoted cell death in HEK293 cells 28 . Our results showed, from another point of view, that ASK1 has a promoting role in OTA-induced cell death.
ASK1 knockdown reduced OTA-induced ROS generation. It is well documented that ASK1 can be activated by oxidative stress 29 , but how ASK1 affects ROS generation is unknown. Therefore, we examined ROS generation in scrambled and ASK1 knockdown cells in response to 1 h or 24 h of OTA treatment using DCFH-DA fluorescent staining. As shown in Figure 4, OTA treatment markedly increased ROS generation in scrambled cells. However, ASK1 knockdown alleviated this effect at 1 h after OTA treatment. No significant difference was observed between scrambled cells and knockdown cells at 24 h. It has been reported that the targeting of JNK to the mitochondria enhances ROS production through oxidant generation by mitochondrial complex 30,31 . Our results suggested that ASK1 enhances ROS production at the transient time point, possibly due to the downstream activation of JNK.
ASK1 knockdown alleviated OTA-induced Dym loss. Mitochondrial membrane potential (MMP, Dym) loss is a symbolic feature of early apoptosis 32 . We measured the Dym of scrambled and ASK1 knockdown cells in response to 1 h or 24 h of OTA treatment. As shown in Figure 5, OTA exposure significantly reduced the Dym in HEK293 cells. No significant difference was observed between transient (1 h) and sustained (24 h) OTAinduced Dym loss. The Dym loss was significantly less in ASK1 knockdown cells compared with that in scrambled cells. These results suggested that ASK1 might promote OTA-induced mitochondrial membrane potential loss.
iTRAQ profiling of differentially expressed proteomes. To investigate the global effect of ASK1 on OTA-induced renal cytotoxicity, we performed iTRAQ-based quantitative proteomics on scrambled cells and ASK1 knockdown cells, with or without 20 mM OTA treatment for 24 h ( Figure 6). There were 3146, 2846 and 2837 nonredundant proteins (unused protein score $ 1.3) identified for three biological replicates, respectively, among which 2224 proteins overlapped ( Figure 7A). To assess the variance in the iTRAQ quantification experiments, we introduced a statistical approach described by Ferret-Bernard et al. 33 Log ratios obtained by comparing differentially expressed proteins (p , 0.05) in response to OTA versus the control were plotted against the number of identified peptides ( Figure 7B). The extent to which the values deviate from unity is an indication of the variance. We chose a jlog (ratio)j value of 0.08 as the threshold, which corresponds to ratios of 1.2 and 0.83, because the vast majority of proteins were within this range (nearly 99%). With this filter, we identified 33 and 24 significantly altered proteins in scrambled cells (1185117, OTA treatment versus control) and ASK1 knockdown cells (1215119, OTA treatment versus control), respectively (Tables 1 and 2).
Effect of ASK1 on OTA-induced proteomic changes. As shown in Tables 1 and 2, the expression of 33 proteins was significantly altered in scrambled cells in response to OTA treatment, among which 13 were up-regulated and 20 were down-regulated. Twenty-four proteins were differentially expressed in ASK1 knockdown cells, among which 8 were up-regulated and 16 were down-regulated. All of the proteins were classified according to annotations from  the UniProt knowledge base and the GO database. Pathways associated with these proteins were elucidated according to REACTOME_PATHWAY using DAVID functional annotation 34 ( Figure 8). In general, sustained OTA exposure (24 h) exerted influences on mRNA splicing, nucleotide metabolism, the cell cycle, DNA repair, and lipid and lipoprotein metabolism, and ASK1 was implicated in all of these metabolic pathways.
mRNA splicing. As shown in Figure 8, OTA treatment down-regulated most of the proteins related to the spliceosome pathway, including EIF4A3, DDX39B, CDC5L, XAB2 and PRPF19. CDC5L and XAB2 are also involved in transcription. HNRNPL is a component of the heterogeneous nuclear ribonucleoprotein (hnRNP) complexes, which provide substrates for pre-mRNA processing events, and is a critical inducible regulator of CD45 alternative splicing 35 . As for the up-regulated proteins, DDX5 and SRRM2 also participate in transcription or the regulation of pre-mRNA splicing. Interestingly, although DDX5 has been reported to be a transcriptional activator, it also functions as a repressor in a promoter-specific manner 36 . Additionally, SRRM2 (also known as SRm300) must form a complex with SRm160 to function as a splicing coactivator. However, the specific depletion of SRm300 does not prevent the splicing of pre-mRNAs, indicating that SRm160 might be the more critical component of the SRm160/300 coactivator 37 . Our results indicated that OTA affects mRNA splicing and transcription, possibly via regulating associated proteins.
Among the 8 differentially expressed proteins, only DDX5, DDX39B and PRPF19 were not affected by ASK1 knockdown. This suggested that ASK1 knockdown partially protected HEK293 cells from OTA-induced mRNA splicing impairment.
Nucleotide metabolism. Six proteins involved in nucleotide metabolism were down-regulated by OTA ( Figure 8). Among them, RRM1 and RRM2 catalyze the biosynthesis of deoxyribonucleotides from the corresponding ribonucleotides 38 . POLA1, POLR2A and POLR2B play essential roles in pyrimidine or purine metabolism, and TYMS contributes to the de novo mitochondrial thymidylate biosynthesis pathway 39 . Interestingly, PKM, which catalyzes the inter-conversion of ATP and ADP in purine metabolism, was up-regulated upon OTA treatment. However, PKM plays a much more critical role in glycolysis, and overexpression of PKM also contributes to tumorigenesis 40 . Our results generally suggested that OTA inhibits nucleotide metabolism, which might further affect DNA biogenesis and cell proliferation.
Similarly to the role of ASK1 in mRNA splicing, ASK1 knockdown prevented the alteration of 6 proteins (PKM, RRM1, RRM2, POLA1, POLR2A and POLR2B), and partially alleviated the down-regulation fold change of TYMS (from 0.290 to 0.376, p , 0.05). These results indicated the participation of ASK1 in OTA-induced nucleotide metabolism impairment.
Cell cycle. All 9 proteins involved in the cell cycle were down-regulated by OTA ( Figure 8). Among these proteins, CDC5L has been   shown to act as a positive regulator of cell cycle G2/M progression 41 ; CDK1 plays a key role in promoting the G2-M transition, as well as G1 progression and the G1-S transition 42 . KIF11 is a motor protein required for establishing a bipolar spindle during mitosis 43 . POLA1, PCNA, RRM1, RRM2, TYMS and MCMBP are key players in DNA replication during S phase in the cell cycle. The down-regulation of these proteins suggested that OTA suppresses the cell cycle, especially DNA replication.
ASK1 knockdown prevented the OTA-induced down-regulation of 5 proteins (CDC5L, POLA1, RRM1, RRM2 and KIF11) and alleviated the down-regulation of 4 proteins, of which 3 were significantly different (CDK1 from 0.646 to 0.732, PCNA from 0.652 to 0.681, and TYMS from 0.290 to 0.376, p , 0.05). In addition, ASK1 knock-down resulted in the up-regulation of KRT18 and down-regulation of STAG2, SKP1, MKI67 and LIG1. Among them, KRT18 is associated with 14-3-3 protein in vitro in a phosphorylation-and cell cycle-dependent manner; this association occurs during the S/G2/M phases of the cell cycle, when KRT18 become hyperphosphorylated 44 . SKP1 (also known as p19) is a cyclin-dependent kinase inhibitor that inhibits CDK4/6 and controls cell cycle exit 45 . The alteration of these two proteins indicates that ASK1 has an inhibitory role in the cell cycle. However, the down-regulation of STAG2, MKI67 and LIG1, all of which are required for DNA replication or maintaining cell proliferation, indicates that ASK1 also has a promoting role in the cell cycle to some extent.
DNA repair. Six proteins related to DNA repair were down-regulated by OTA ( Figure 8). Among them, POLR2A, POLR2B, and PCNA play critical roles in nucleotide excision repair (NER) 46 . PCNA also plays a key role in the oxidative DNA damage response and the promotion of post-replication repair 47 . XAB2 is involved in transcription-coupled repair (TCR) 48 . PRPF19 plays a role in DNA double-strand break (DSB) repair and is a component of the PSO4 complex, which is required in the DNA interstrand cross-link (ICL) repair process 49 . CDK1 was inactivated upon DNA damage to stop the cell cycle and genome replication at the G2 checkpoint, thus facilitating DNA repair 50 . It has been shown that OTA induces oxidative DNA damage in various cell lines 8,51 . Our results further demonstrated that OTA-induced DNA damage might partly arise from the impaired DNA repair system through the down-regulation of key proteins involved in this system.
Among the 6 altered proteins, 2 (POLR2B and XAB2) were reversed by ASK1 knockdown, and 2 (CDK1 and PCNA) were alleviated in their fold changes. ASK1 knockdown also resulted in the downregulation of LIG1 and DDB2, both of which are also involved in NER. DDB2 variants from HeLa cells, mostly D1 and D2, were shown to be dominant negative inhibitors of DNA repair 52 . These results suggested that ASK1 has an inhibitory role in DNA repair.
Lipid and lipoprotein metabolism. Four proteins in lipid and lipoprotein metabolism were up-regulated upon OTA treatment ( Figure 8). Among them, ACACA catalyzes the rate-limiting reaction in the biogenesis of long-chain fatty acids 53 , P4HB acts as a structural subunit of microsomal triacylglycerol transfer protein (MTTP) and SCP2 mediates the transfer of phospholipids between membranes 54,55 . ALB is the main protein of plasma and has a high binding capacity for fatty acids, hormones and drugs. No research has been conducted to show whether OTA has an effect on lipid biogenesis and transfer, whereas studies have shown that OTA could induce lipid peroxidation 56 . The up-regulation of lipid metabolism-relevant proteins might be a response mechanism to protect cells against lipid peroxidation. Preventing the up-regulation of these proteins by ASK1 knockdown indicated the involvement of ASK1 in the OTAinduced enhancement of lipid metabolism.
Proteins involved in oxidative stress and apoptosis. Because ASK1 is stimulated by oxidative stress and contributes largely to the regulation of apoptosis 15 , we further studied the proteomics data for proteins involved in oxidative stress and apoptosis according to the UniProt database. Three proteins (P4HB, DHRS2 and Prx4) were involved in oxidative stress, and 3 (CDK1, PKM and DHRS2) were involved in the regulation of apoptosis (Tables 1 and 2).
Among the proteins involved in oxidative stress, P4HB (protein disulfide isomerase A1) is a multifunctional protein that catalyzes the formation and rearrangement of disulfide bonds. It was also shown to function in the maintenance of cell redox homeostasis and oxidative protein folding 57 . Intriguingly, P4HB is a thioredoxin (Trx) superfamily protein containing two active Trx domains and is a primary inhibitor of ASK1 in its reduced form 15,58 . The oxidizing redox condition in the cell favors the formation of disulfide bonds, leading to the oxidation of reduced Trx into oxidative Trx, whereas P4HB is reduced to an inactive form. This might explain why P4HB is up-regulated upon OTA treatment and how ASK1 is activated in response to OTA treatment.
DHRS2 belongs to the SDR family and was found to possess a protective role against apoptosis induced by oxidative stress 59 . Thus, we assumed that OTA-induced DHRS2 down-regulation could    partially result in cell apoptosis; however, ASK1 had little effect on DHRS2 expression. ASK1 knockdown increased the expression of Prx4, indicating that ASK1 has an inhibitory role on Prx4. Prx4 is an antioxidant enzyme that belongs to the peroxiredoxin family. Prx4 was reported to play a regulatory role in the activation of the transcription factor NF-kappaB 60 . In addition, it was reported that the inhibition of NFkB allows not only prolonged JNK activation but also amplification of ROS stress due to decreased antioxidant transcription 61,62 . Taken together, ASK1 might amplify ROS generation through the inhibition of Prx4 and, subsequently, NFkB; this is in accordance with the ROS assay results mentioned above ( Figure 4).
As for the proteins related to cell death, CDK1 and DHRS2, both of which are involved in the negative regulation of apoptosis, were down-regulated by OTA, and PKM, which plays a general role in caspase-independent cell death, was up-regulated by OTA. These results suggested that OTA might induce cell apoptosis through the regulation of these proteins. Additionally, ASK1 knockdown reversed the up-regulation of PKM and alleviated the down-regulation fold change of CDK1, confirming the role of ASK1 in OTAinduced cell death.
Verification of iTRAQ data by Western blot. Four differentially expressed proteins identified in the proteomics data were validated by Western blotting (Figure 9). CDK1 (cdc2) and KRT18 were involved in the cell cycle process, and PKM1/2 and DHRS2 were involved in the regulation of cell death. KRT18 was up-regulated in ASK1 knockdown cells, and PKM was up-regulated in scrambled cells, which was in accordance with the iTRAQ results. In addition, CDK1 and DHRS2 were both down-regulated independently of ASK1, which was also in agreement with the proteomics data. These results confirmed the reliability and accuracy of the iTRAQ-based proteomics technique.

Discussion
ASK1 has been shown to be an important regulator of oxidative stress-induced cell death and has drawn much attention in apoptosis research 17 . OTA induces a variety of cytotoxic effects, including oxidative stress and apoptosis 7,9 . However, to the best of our knowledge, no research has been performed to study the role of ASK1 in OTA-induced cytotoxicity. This study proposed that ASK1 contributes to OTA-induced cell death and impacts other cell signaling or metabolic pathways, as observed in the iTRAQ-based proteomics results.
As oxidative stress is induced by OTA and is one of the major triggers of ASK1 activation, we asked whether ASK1 has a reverse effect on oxidative stress. Intriguingly, although ASK1 is clearly activated by oxidative stress 17 , our result showed that ASK1 is also capable of amplifying ROS generation (Figure 4), probably due to the generation of oxidants and the decreased expression of transcription factors of certain antioxidants. JNK silencing mediated the amplification of ROS production during stress, but p38 silencing had no effect on ROS production 31 . Thus, we speculate that the role of ASK1 in ROS production is mainly through JNK activation.
The loss of Dym is an apoptotic signal that precedes DNA fragmentation in various cell types and is thus considered a sign of early apoptosis 63 . Our results suggested that ASK1 is required in the breakdown of Dym and initiation of the mitochondrial-dependent intrinsic apoptotic pathway ( Figure 5). This was in accordance with a previous report showing that the early generation of ROS and subsequent signaling through the ASK1 and JNK pathway initiated irreversible intracellular changes, including caspase-3 activation and mitochondrial membrane potential loss, thus contributing to the induction of the death process 64 . Existing researches of the influence of OTA on metabolic pathways have focused on oxidative stress, apoptosis, cell cycle arrest, DNA damage and repair, and lipid peroxidation 8,56,65 . Our proteomics data revealed new metabolic pathways affected by OTA such as mRNA splicing, nucleotide metabolism, and lipid metabolism (especially lipid biosynthesis and transfer). The prominent cause of OTAinduced DNA damage was the formation of DNA adducts initiated by the covalent binding of carcinogens to DNA 3 . However, our proteomics data suggested that OTA-induced DNA damage might also be caused by the inhibition of nucleotide biosynthesis. For instance, RRM1 and RRM2 catalyze the biosynthesis of deoxyribonucleotides, and TYMS contributes to the de novo mitochondrial thymidylate biosynthesis. Blocking the formation of the four basic DNA substrates might result in unscheduled DNA synthesis and replication. Angelika et al. 66 reported that low concentrations of OTA (750 nM-1 mM) caused the dose-dependent induction of DNA repair in rat hepatocyte and porcine urinary bladder epithelial cell cultures. However, based on our proteomics data, a higher concentration of OTA (20 mM) resulted in the inhibition rather than the induction of DNA repair, which is evidenced by the down-regulation of POLR2A, POLR2B, and PCNA in nucleotide excision repair (NER), XAB2 in transcription-coupled repair (TCR), and PRPF19 in double-strand break (DSB) repair. Stefan et al. 67 reported that OTA-induced DNA damage is increased by blocking repair mechanisms in MDCK cells. Taken together, OTA not only induces DNA damage per se but also impairs DNA repair mechanisms, which further aggravates the aftermath of DNA damage.
The down-regulation of proteins involved in mRNA splicing implied that OTA also induces damage at the RNA level, in addition to DNA damage. Pre-mRNA splicing is one of the three major pre-mRNA processing events and can occur post-transcriptionally or co-transcriptionally 68 . Additionally, the down-regulated protein HNRNPL is a critical inducible regulator of alternative splicing, which generates different mRNAs encoding distinct protein products, including those involved in apoptotic cell death 69 . Altogether, our results suggested that OTA blocks the formation of mature mRNA and subsequently affects the translation and expression of critical proteins.
Hibi et al. 65 fed male rats with 5 ppm OTA for 4 weeks and observed an up-regulation of cell cycle-related genes. However, our proteomics data showed that the cell cycle was inhibited by OTA. For instance, among all of the down-regulated proteins, CDC5L acts as a positive regulator of G2/M progression, CDK1 plays essential roles in promoting the G2-M transition, G1 progression and the G1-S transition, and KIF11 is required to establish a bipolar spindle during M phase. The inconsistency between the results might arise from dose differences and variations between in vivo and in vitro experiments. OTA might be partially metabolized in vivo, thereby decreasing its toxicity. Last but not least, OTA increased the expression of proteins involved in lipid biosynthesis (ACACA) and lipid transfer (P4HB and SCP2). OTA is capable of inducing lipid peroxidation 56 , and the elevation in lipid metabolism might be a response mechanism for cells to adapt to the severe environment.
ASK1 silencing changed almost all of the proteins in the OTAinduced inhibition or activation of the above pathways ( Figure 8). Among them, the alteration of proteins involved in nucleotide and lipid metabolism was completely offset by ASK1 silencing. This suggested that the OTA-induced inhibition of nucleotide metabolism and enhancement of lipid metabolism were likely dependent on ASK1 activation.
On the other hand, ASK1 partially participated in the regulation of the OTA-induced inhibition of mRNA splicing, the cell cycle and DNA repair. Kuo et al. 70 reported that Isoobtusilactone A (IOA)induced cell cycle arrest, associated with the up-regulation of p21 and down-regulation of cyclin B1, cyclin A, cdc2 and cdc25, was regulated by the ROS and ASK1 signaling pathways in human breast cancer cells. Our results further confirmed that ASK1 regulated the OTA-induced cell cycle blockade, coupled with the down-regulation of CDK1 (cdc2). However, the proteomics data showed that ASK1 also has an activating effect on cell cycle progression, along with the down-regulation of STAG2, MKI67 and LIG1 after ASK1 silencing. p38 can function as a dual mediator of ROS signaling and either activate or suppress cell cycle progression depending on the activation stimulus 71 . As ASK1 is the upstream activator of p38, it is reasonable to conclude that ASK1 has a dual role in the cell cycle. Cell death (typically apoptosis) provoked by genotoxins is largely due to DNA damage, including O 6 -methylguanine, N-methylation lesions, bulky DNA adducts, DNA cross-links and DNA doublestrand breaks (DSBs) 72 . Cells respond to DNA lesions by activating cell cycle checkpoint and repair mechanisms; however, failure of the repair system causes tumorigenesis or cell death 73 . At cell cycle checkpoints, including DNA damage, DNA replication and spindle assembly checkpoints, damaged cells are examined, and the cell cycle is arrested until the damage is repaired 74 . If the damage is too serious to repair or the repair system is inhibited, cells undergo apoptosis or other types of cell death 75,76 . The inhibition of RNA synthesis is another trigger of cell death upon DNA damage, which leads to a decline in MKP1 (mitogen-activated protein kinase phosphatase 1) and sustained JNK activation 72 . In human fibroblasts, Nucleotide Excision Repair (NER)-defective mutants display sustained and higher levels of JNK activation compared with repair-competent cells following cisplatin treatment 77 . Our results showed that ASK1 regulated the inhibition of DNA repair mechanisms by down-regulating proteins involved in NER (POLR2B, PCNA) and TCR (XAB2),  which might further cause prolonged JNK activation and apoptosis initiation. Moreover, protein-protein interactions, including p53 binding to the SH3 domains of some proteins such as Bcl2, play an important role in p53-dependent apoptosis after RNA synthesis is blocked 78 . Altogether, we speculate that the ASK1-mediated blockade of RNA synthesis (predominantly mRNA splicing) can induce apoptosis through the repression of certain genes or through proteinprotein interactions. A hypothetical model of the detailed pathway network is shown in Figure 10.

Conclusion
In summary, our results indicated that OTA activates ASK1 through oxidative stress, whereas ASK1 also has a role in the amplification of ROS generation. Activated ASK1 mediates apoptosis initiation through multiple pathways, including DNA damage (impaired nucleotide metabolism), cell cycle checkpoints, DNA repair mechanisms, and the blockade of RNA synthesis. The cell cycle checkpoints initiate the repair cascade when DNA is damaged. If repair fails, the apoptotic pathway is activated. In the meantime, DNA damage induces the inhibition of RNA synthesis, leading to the repression of certain anti-apoptotic genes. This study shed light on the underlying mechanism by which ASK1 regulates OTAinduced cytotoxicity, providing new ideas for understanding the toxification mechanisms of OTA and the function of ASK1 in mediating cell death.
RNA interference. The ASK1 shRNA and scrambled shRNA were purchased from Santa Cruz Biotechnologies (Santa Cruz Biotechnologies, Dallas, Texas, USA). The ASK1 shRNA Plasmid (h) is a pool of 3 different shRNA plasmids with the following hairpin sequences: GATCCGTACCTCAAGTCTATTGTATTCAAGAGATACAA-TAGACTTGAGGTACTTTTT, GATCCCAAGGCATTCATACTGAAATTCAAG-AGATTTCAGTATGAATGCCTTGTTTTT, and GATCCGGAAGGCTATCATT-GACTTTTCAAGAGAAAGTCAATGATAGCCTTCCTTTTT. All sequences are provided in 59R39 orientation. shRNA was transfected into cells using VigoFect (Vigorous Biotechnology, PRC) following protocols provided by the manufacturer. Briefly, 2 3 10 5 cells were seeded into six-well plates and grown to 40-60% confluence. The medium was renewed 1 h before transfection. The mixture of 10 ml of shRNA (1 mg) diluted in 90 ml of normal saline and 1 ml of VigoFect diluted in 99 ml of normal saline was added dropwise to the medium. At 48 h post-transfection, the medium was replaced with fresh selective medium containing 1 mg/mL puromycin (Santa Cruz Biotechnologies) to screen for stably transfected cells. The cultures were replaced with fresh selective medium every two days until no cells were killed.
Cell viability assay. The cell viability was assessed by WST-8 staining with the Cell Counting Kit-8 (CCK-8) (Beyotime, PRC) according to the manufacturer's instructions. In brief, 1 3 10 4 cells were seeded in 96-well plates and treated with 0, 5, 10, 20 or 40 mM of OTA. After incubation for 24 h, 10 ml of WST-8 dye diluted in 100 ml of PBS was added to each well, and the cells were incubated at 37uC for 1 h. The absorbance was read at 450 nm using a Varioskan Flash microplate reader (Thermo Scientific, USA). Measurement of mitochondrial membrane potential (Dym). The mitochondrial membrane potential (Dym) was measured by 5,50,6,60-tetrachloro-1,10,3,30tetraethylbenzimidazolycarbocyanine iodide (JC-1, Beyotime, PRC) staining, following the manufacturer's instructions. Briefly, 2 3 10 5 cells were seeded into a 6well plate and incubated for 24 h. After OTA treatment, the cells were incubated with JC-1 (Beyotime, PRC) staining solution (5 mg/ml) for 20 min at 37uC. The cells were then washed twice with the JC-1 staining buffer, and the fluorescence intensity of JC-1 monomers (lex 5 488 nm, lem 5 529 nm) and JC-1 aggregates (lex 5 524 nm, lem 5 594 nm) was measured using a microplate reader. The Dym was expressed as the fluorescence ratio of JC-1 aggregates versus JC-1 monomers.
iTRAQ labeling, mass spectrometry identification and data analysis. Protein extraction and iTRAQ labeling. The flow chart of iTRAQ-based proteomics is shown in Figure 6. ASK1 knockdown or scrambled HEK293 cells were treated with or without 20 mM OTA for 24 h. Samples were ground in liquid nitrogen. One milliliter of lysis buffer (7 M urea, 2 M thiourea, 13 Protease Inhibitor Cocktail (Roche Ltd. Basel, Switzerland)) was added to each sample, followed by sonication on ice and centrifugation at 13,000 rpm for 10 min at 4uC. The supernatant was transferred to a fresh tube and stored at 280uC until use. For each sample, proteins were precipitated with ice-cold acetone and then redissolved in dissolution buffer (0.5 M triethylammonium bicarbonate, 0.1% SDS). The proteins were quantified by the BCA protein assay. Then, 100 mg of protein was tryptically digested, and the resultant peptide mixture was labeled using the iTRAQ reagent (Applied Biosystems, California, USA). Briefly, disulfide bonds were reduced in 5 mM Tris-(2carboxyethy) phosphine (TCEP) at 60uC for 1 h. Cysteine residues were blocked in 10 mM methyl methanethiosulfonate (MMTS) for 30 min at room temperature. The protein solution was further digested with sequence-grade modified trypsin (Promega, Madison, WI). For labeling, each iTRAQ reagent was dissolved in 50 mL of isopropanol and added to the respective peptide mixture. Proteins samples were labeled with iTRAQ isobaric tags 117, 118, 119, and 121, respectively ( Figure 6). The labeled samples were combined and dried in vacuo. A SepPac TM C18 cartridge (1 cm 3 / 50 mg, Waters Corporation, Milford, MA) was used to remove the salt buffer, and the samples were then dried in a vacuum concentrator for the next step.
High pH reverse phase separation. We employed a two-dimensional RP-RP-HPLC separation method with different pHs in two dimensions; this method represents a promising tool for proteomics research and compares favorably with the traditional SCX-RP-HPLC approach in overall performance 79 . In brief, the peptide mixture was redissolved in buffer A (buffer A: 20 mM ammonium formate in water, adjusted with ammonium hydroxide to pH 10.0) and was then fractionated by high pH separation using a Aquity UPLC system (Waters Corporation, Milford, MA) connected to a reverse phase column (XBridge C18 column, 2.1 mm 3 150 mm, 3.5 mm, 300 Å , Waters Corporation, Milford, MA). High pH separation was performed using a linear gradient, starting from 5% B and going to 35% B, for 40 min (B: 20 mM ammonium formate in 90% ACN, adjusted with ammonium hydroxide to pH 10.0). The column was re-equilibrated to the initial conditions for 15 min. The column flow rate was maintained at 200 mL/min, and the column temperature was maintained at room temperature. Twelve fractions were collected, and each fraction was dried in a vacuum concentrator before further analysis.
Low pH nano-HPLC-MS/MS analysis. The peptides were resuspended with 80 ml solvent C (C: water with 0.1% formic acid; D: ACN with 0.1% formic acid), separated by nanoLC and analyzed by on-line electrospray tandem mass spectrometry. The experiments were performed on a Nano Aquity UPLC system (Waters Corporation, Milford, MA) connected to an LTQ Orbitrap XL mass spectrometer (Thermo Electron Corp., Bremen, Germany) equipped with an online nanoelectrospray ion source (Michrom Bioresources, Auburn, USA). First, 18 ml of each peptide sample was loaded onto the Thermo Scientific Acclaim PepMap C18 column (100 mm 3 2 cm, 3-mm particle size), with a flow of 10 ml/min for 5 min, and subsequently separated on the analytical column (Acclaim PepMap C18, 75 mm 3 15 cm) with a linear gradient, from 5% D to 45% D, for 165 min. The column was re-equilibrated to the initial conditions for 15 min. The column flow rate was maintained at 300 nL/ min, and the column temperature was maintained at 35uC. An electrospray voltage of 1.9 kV versus the inlet of the mass spectrometer was used.
The LTQ Orbitrap XL mass spectrometer was operated in data-dependent mode to automatically switch between MS and MS/MS acquisition 80 . Survey full-scan MS spectra (m/z 400-1600) were acquired in the Orbitrap, with a mass resolution of 30,000 at m/z 400, followed by five sequential HCD-MS/MS scans. The automatic gain control (AGC) was set to 500,000 ions to prevent over-filling of the ion trap. The minimum MS signal for triggering MS/MS was set to 1000. In all cases, one microscan was recorded. MS/MS scans were acquired in the Orbitrap with a mass resolution of 7500. The dissociation mode was higher energy C-trap dissociation (HCD). Dynamic exclusion was used with two repeat counts with a 10-s repeat duration, and the m/z values triggering MS/MS were put on an exclusion list for 120 s. For MS/MS, precursor ions were activated using 40% normalized collision energy and an activation time of 30 ms.
Database searching and iTRAQ quantification. Protein identification and quantification for the iTRAQ experiment were performed with the ProteinPilot software version 4.0 (Applied Biosystems, California, USA). The database used was the Human UniProtKB/Swiss-Prot database (Release 2012_12_27, with 20233 sequences). The Paragon Algorithm in the ProteinPilot software was used for peptide identification and isoform specific quantification. The data search parameters were set up as follows: trypsin (KR) cleavage with two missed cleavages and fixed modification of cysteines by MMTS were considered, and iTRAQ modification of peptide N termini, methionine oxidation and iTRAQ modification of lysine residues were set as variable modifications. To minimize false positive results, a strict cutoff for protein identification was applied with the unused ProtScore $ 1.3, which corresponds to a confidence limit of 95%, and at least two peptides with 95% confidence were considered for protein quantification. The resulting data set was automatically bias-corrected to remove any variations imparted due to unequal mixing during the combination of different labeled samples. For iTRAQ quantitation, the peptide for quantification was automatically selected by the Pro Group algorithm (at least two peptides with 99% confidence) to calculate the reporter peak area, error factor (EF), and p-value. For the selection of differentially expressed proteins, the proteins had to contain at least two unique high-scoring peptides.
Analysis of proteomics data. The differentially expressed proteins were analyzed according to the functional annotations from the UniProt knowledge base (Swiss-Prot/TrEMBL, http://www.uniprot.org/) and GO database (http://www. geneontology.org/). The metabolic pathways associated with the differentially expressed proteins were classified using KEGG_PATHWAY, PANTHER_PATHWAY and REACTOME_PATHWAY in the DAVID database (http://david.abcc.ncifcrf.gov/). Statistical analysis. Microsoft Excel 2010 and SPSS 20 (SPSS, Inc) were used for the statistical analysis. Data were subjected to an analysis of variance (ANOVA), and a comparison of means was carried out by Duncan's multiple-range test. p , 0.05 is considered to be statistically significant.