A Critical Role for IκB Kinase β in Metallothionein-1 Expression and Protection against Arsenic Toxicity*

Arsenic is a widespread environmental toxic agent that has been shown to cause diverse tissue and cell damage and at the same time to be an effective anti-cancer therapeutic agent. The objective of this study is to explore the signaling mechanisms involved in arsenic toxicity. We show that the IκB kinase β (IKKβ) plays a crucial role in protecting cells from arsenic toxicity. Ikkβ-/- mouse 3T3 fibroblasts have decreased expression of antioxidant genes, such as metallothionein 1 (Mt1). In contrast to wild type and IKKβ-reconstituted Ikkβ-/- cells, IKKβ-null cells display a marked increase in arsenic-induced reactive oxygen species (ROS) accumulation, which leads to activation of the MKK4-c-Jun NH2-terminal kinase (JNK) pathway, c-Jun phosphorylation, and apoptosis. Pretreatment with the antioxidant N-acetylcysteine (NAC) and expression of MT1 in the Ikkβ-/- cells prevented JNK activation; moreover, NAC pretreatment, MT1 expression, MKK4 ablation, and JNK inhibition all protected cells from death induced by arsenic. Our data show that two signaling pathways appear to be important for modulating arsenic toxicity. First, the IKK-NF-κB pathway is crucial for maintaining cellular metallothionein-1 levels to counteract ROS accumulation, and second, when this pathway fails, excessive ROS leads to activation of the MKK4-JNK pathway, resulting in apoptosis.

Arsenic is a highly toxic ubiquitous environmental contaminant. Chronic low level exposure may cause skin, lung, bladder, and kidney cancer, whereas higher doses of arsenic lead to tissue damage through the induction of cell death. Such dose and cell type-dependent effects are the basis for arsenic trioxide to be used in the treatment of acute promyelocytic leukemia. Initially it was thought that arsenic, like all-trans-retinoic acid, could target the promyelocytic leukemia gene product retinoic acid receptor (PML-RAR) fusion protein, causing its degradation; however, many PML-RAR-negative cell lines and cells from PML knock-out mice are also sensitive to arsenic, suggesting alternative mechanisms for toxicity (1)(2)(3)(4)(5)(6)(7)(8)(9).
As a redox-active metalloid, arsenic in a dose-dependent manner elicits an immediate burst of intracellular ROS (10 -14). Cells normally respond to these changes by activating the defense mechanisms to re-establish redox homeostasis and protect from oxidative damage (15,16). We and others have previously reported the transcriptional activation by arsenite of a battery of antioxidant genes, including hemeoxygenase-1, metallothionein 1 and 2, and thioredoxin reductase 1, which may result in parallel changes in protein levels and enzyme or binding activity (17,18).
Antioxidant up-regulation, however, is insufficient to counteract the ROS 2 caused by high dose and prolonged arsenic exposure. Numerous studies have shown that chronic exposure to arsenic from drinking water in humans and experimental animals are often associated with increased oxidative stress, a state when cellular ROS production exceeds antioxidant capacity (19 -22). Depending on their level and persistence, excessive ROS can cause diverse pathophysiological conditions, such as oxidative damage to DNA, proteins, and lipids, that may contribute to cell transformation and tumorigenesis (23,24). ROS can perturb the mitochondrial permeability transition pore and disrupt electron transfer, triggering apoptosis and necrosis (18,25,26). ROS can also activate death-related signaling effectors, such as the JNK, whose sustained activation results in the phosphorylation of the Bcl family of proteins, leading to cytochrome c release from mitochondria, thereby functioning as an important proapoptotic factor (27)(28)(29). The ultimate toxicity of arsenic, therefore, is dependent largely on the ability of the cell to produce antioxidants and to modulate ROS levels. Tipping the cellular redox balance toward ROS accumulation can enhance arsenic-induced cell damage, as seen in glutamatecysteine ligase modifier subunit-deficient Gclm Ϫ/Ϫ fibroblasts and in ␥-glutamylcysteine synthetase mutant Caenorhabditis elegans (17,30). Conversely, inclusion of ROS scavengers in the culture medium protects cell against arsenic toxicity (31,32).
One of the most important redox modulation cellular constituents is the redox-sensitive transcription factor NF-B. NF-B members are present in the cytoplasm associated with IB, which hinders recognition of the nuclear localization signal on NF-B and prevents its nuclear translocation. Upon stimulation by various growth factors, cytokines and cellular stressors, the IB kinase (IKK) complexes consisting of ␣, ␤, and ␥ subunits are activated, which phosphorylate IB on serine/threonine residues. This phosphorylation leads to IB ubiquitination and subsequent proteolytic degradation, thereby unmasking the nuclear localization signal on NF-B and allowing its nuclear translocation. The nuclear NF-B complex binds to sequence-specific consensus DNA and activates gene transcription (33)(34)(35). NF-B is a critical regulator of the expression of genes involved in adaptive anti-oxidative responses, including glutathione peroxidase, mitochondrial SOD2, and cytosolic SOD1. Cells and mice defective in NF-B signaling are more susceptible to oxidative damage induced by TNF␣, H 2 O 2 , and arsenic (36).
The IKK-NF-B pathway has recently emerged as a master regulator for buffering cellular redox levels and suppressing ROS accumulation during inflammatory responses (37,38). In the current studies we showed that IKK␤ activity was essential for transcriptional activation of antioxidant genes, such as metallothionein 1, whose expression was markedly reduced in Ikk␤ Ϫ/Ϫ cells. As a consequence, Ikk␤ Ϫ/Ϫ cells had higher levels of ROS accumulation than wild type cells. Arsenic caused a ROS-dependent activation of the signaling cascade MKK4-JNK-c-Jun that led to apoptosis. The thiol-containing redox scavenger N-acetyl-L-cysteine (NAC), MKK4 ablation, SP600125, a JNK inhibitor, and reconstitution of IKK␤ and MT1 expression in Ikk␤ Ϫ/Ϫ cells all resulted in a reduction of cell death caused by arsenic exposure. Our data suggest that IKK␤ protects cells from arsenic toxicity by supporting antioxidant gene expression, thereby preventing arsenic-induced ROS and activation of the redox-sensitive signaling pathways that lead to apoptosis.
The IKK␤ adenovirus was a gift from Dr. Yinling Hu (University of Texas, MD Anderson Cancer Center, Houston, TX). Adenoviruses at 10 7 plaque-forming units were used for infection of 90% confluent cells. Viral infection was carried out in serum-free Dulbecco's modified Eagle's medium for 2 h with gentle shaking and then washed with phosphate-buffered saline. After infection, the cells were incubated with growth medium at 37°C and 5% CO 2 for 24 h. When the cells reached 60 -80% confluence, they were exposed to various concentrations of sodium arsenite in fetal bovine serum-free Dulbecco's modified Eagle's medium for different lengths of time.
Cell Toxicity Assays-Cells cultured on cover-glass slides were washed twice with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde in PBS for 10 min at room temperature, washed, and treated with 0.25% Nonidet P-40 for 1 min. Cells were stained with 10 g/ml 4,6-diamidino-2-phenylindole (DAPI) for 30 min, the staining was observed, and pictures were taken under a fluorescence microscope. Approximately 100 -200 cells in 3 different fields/cover-glass were photographed. The percentage of apoptotic cells with condensed or fragmented nuclei was calculated in comparison to the numbers of total cells number.
As a measure of total cell death, lactate dehydrogenase (LDH) activity was determined in a colorimetric assay following the manufacture's instructions (Pierce). Briefly, 50 l of cell culture supernatant were mixed with 200 l of reaction mixture and incubated for 30 min at 37°C in the dark. The remaining cells were lysed with water for 30 min and processed as described above. Absorbance was measured in a microplate reader at 490 nm. Total LDH ϭ LDH in supernatant ϩ LDH in water-treated cells. LDH release (%) ϭ (supernatant LDH/total LDH) ϫ 100.
Fluorescence based terminal deoxynucleotidyltransferase (TdT)-mediated dUTP nick end labeling (TUNEL) assay was performed using the labeling protocol provided by the manufacture (Roche Applied Science).
ROS Measurement-Direct detection of intracellular steadystate levels of ROS was carried out on living cells using 2Ј,7Јdichlorofluorescein-diacetate (CM-H 2 DCFDA). Cells cultured on glass coverslips were incubated with 10 M CM-H 2 DCFDA (Molecular Probes, Inc., OR) in fetal bovine serum-free Dulbecco's modified Eagle's medium in the dark for 30 min. The cells were washed with phosphate-buffered saline and stained with DAPI. ROS generation, resulting in the oxidative production of dichlorofluorescein (excitation, 488 nm; emission, 515-540 nm), was detected under fluorescence microscopy.
Alternatively, cells cultured on 10-cm 2 plates were trypsinized and centrifuged, the cell pellets were incubated in fetal bovine serum-free culture medium with 5 M H 2 -DCFDA for 30 min at 37°C. Cells were washed and suspended at 1 ϫ 10 6 cells/ml in phosphate-buffered saline and analyzed by FACS-Calibur (BD Biosciences Immunocytometry Systems), which was equipped with a 488 argon laser for measurements of intracellular fluorescence. Logarithmic detectors were used for the FL-1 fluorescence channel necessary for DCF detection. Mean log fluorescence intensity values were obtained by the CELLQUEST software program (BD Biosciences).
Cell Lysates and Western Blotting-Whole cell extracts were prepared using radioimmune precipitation assay buffer (25 mM Tris⅐HCl, pH 7.4, 50 mM NaCl, 0.5% sodium deoxycholate, 2% Nonidet P-40, 0.2% SDS, 1 M phenylmethylsulfonyl fluoride, 50 g/ml aprotinin, 50 M leupeptin) at 4°C for 60 min. Samples were boiled and analyzed by SDS-PAGE. The resolved proteins were transferred to a nitrocellulose membrane, and Western blotting was performed using antibodies as indicated.
Reverse Transcription and Quantitative PCR-Total RNA was extracted by using RNeasy mini kit (Qiagen Sciences), and RNA (10 g) was reverse-transcribed to cDNA using random hexamer primers and the Stratascript enzyme following the instructions from the company (Invitrogen). Quantitative PCR was performed using an MX3000p thermal cycler system and SYBR Green QPCR Master Mix (Stratagene, La Jolla, CA). The conditions for the amplification were optimized for specific PCR reactions. At the end of the PCR the samples were subjected to melting curve analysis. All reactions were performed at least in triplicate. Oligonucleotides used as the specific primers to amplify mouse antioxidant genes cDNA are as follows: Mt1, 5Ј-CACGACTTCAACGTCC and 5Ј-CAGCAGGAGCAGCA-GCTCTTCTTGCAG; Mt2, 5Ј-GCTCCTAGAACTCTTC-AAAC and 5Ј-CAGCAGGAGCAGCAGCTCTTCTTGCAG; Gapdh, 5Ј-CCATGGAGAAGGCTGGGG and 5Ј-CAAAGTT-FIGURE 1. IKK␤ protects against arsenite toxicity. A, genomic DNA isolated from wild type (WT), Ikk␤ Ϫ/Ϫ , and Ikk␤ Ϫ/Ϫ -R cells were used for PCR amplification of the Ikk␤-null allele. B, Western blotting (IB) analyses for IKK␤ proteins in WT, Ikk␤ Ϫ/Ϫ , Ikk␤ Ϫ/Ϫ -R, and Ikk␤ Ϫ/Ϫ -Ad. C-F, WT, Ikk␤ Ϫ/Ϫ , and in some experiments IKK␤ reconstituted Ikk␤ Ϫ/Ϫ -R cells were exposed to various concentrations (0 -200 M) of sodium arsenite for 24 h (C) and 6 h (E and F) or exposed to 5 M arsenite for various times as indicated (D). Apoptotic cells with condensed nuclei were identified after DAPI staining by fluorescence microscopy observation (C-E). The percentage of apoptotic cells was calculated from data of three different fields of triplicate samples. F, total apoptotic and necrotic cell death was detected by measuring LDH activity. The relative LDH activity is the percentage of the released versus total cellular LDH activity. Results are presented as mean Ϯ S.D. from three independent experiments. *, indicated significant differences (*, p Ͻ 0.01) compared with the value in wild type cells under the same experimental conditions. G, WT, Ikk␤ Ϫ/Ϫ , and Ikk␤ Ϫ/Ϫ -R cells were treated with 50 M sodium arsenite for various times as indicated. Total cellular proteins were subjected to Western blot assay using anti-PARP. cPARP, cleaved PARP. Arsenite-induced PARP cleavage is evident in Ikk␤ Ϫ/Ϫ but not in WT and Ikk␤ Ϫ/Ϫ -R cells. H, WT, Ikk␤ Ϫ/Ϫ , Ikk␤ Ϫ/Ϫ -R, and Ikk␤ Ϫ/Ϫ -Ad cells were treated with 50 M sodium arsenite for 6 h. TUNELpositive cells were apparently detected in Ikk␤ Ϫ/Ϫ but not in WT, Ikk␤ Ϫ/Ϫ -R and Ikk␤ Ϫ/Ϫ -Ad cells.
Transfection and Luciferase Assay-Cells were plated in 12-well tissue culture plates at 2 ϫ 10 5 cells/well the evening before transfection. The cells were transfected with 1 g of NF-B-dependent luciferase reporter plasmid along with 1 g of ␤-galactosidase plasmid or with expression vectors for MT1 and green fluorescence protein (GFP) following the manufacturer's instructions (Invitrogen). Twenty-four hours after transfection, cells were treated by arsenic and followed by staining by CM-H 2 DCFDA, anti-p-c-Jun, and DAPI.
For the luciferase assay, cells were starved overnight and treated with 20 ng/ml TNF␣ for 16 h. Cells were lysed, and luciferase activity and ␤-galactosidase activity of cell lysates were determined using the luciferase reporter kit and ␤-galactosidase reporter kit (Promega) according to the manufacturer's protocol.
Immunofluorescent Staining for Phospho-c-Jun-Cells grown on cover glasses were fixed with 4% formaldehyde and permeabilized with 0.25% Nonidet P-40. Cells were blocked with 5% bovine serum albumin (BSA) for 1 h and incubated overnight with rabbit polyclonal anti-phospho-c-Jun antibody diluted 1:500 in 1% BSA followed by rhodamine isothiocyanate-conjugated goat anti-rabbit immunoglobulin-G antibody (Molecule Probes) diluted 1:500 in 1% BSA for 30 min. Slides were analyzed using a fluorescent light microscope.
Statistical Analysis-Statistical comparisons were performed with Student's two-tailed paired t test and analysis of variance. Values of p Ͻ 0.01 were considered statistically significant.

IKK␤ Protects Cells from Arsenic
Toxicity-The identity of the cells used for these studies was confirmed by genotypic analyses, which detected the presences of the Ikk␤null alleles (40) in Ikk␤ Ϫ/Ϫ and Ikk␤ Ϫ/Ϫ -R but not in wild type cells (Fig. 1A). Western blot analyses showed that expression of IKK␤ proteins, apparent in wild type, Ikk␤ Ϫ/Ϫ -R, and Ikk␤ Ϫ/Ϫ -Ad, was completely absent in Ikk␤ Ϫ/Ϫ (Fig. 1B).
To evaluate the role of IKK␤ in arsenic toxicity, we exposed wild type and IKK␤-deficient cells to different concentrations (0 -200 M) of sodium arsenite for various length of time, as indicated ( Fig. 1, C-F). Under normal growth condition, IKK status had not much influence on cell survival; when exposed to sodium arsenite, the Ikk␤ Ϫ/Ϫ cells were more sensitive to toxicity than wild type cells, as indicated by the higher levels of LDH release and apoptosis. There was a clear association of arsenic dose and exposure time with increased death of all cells, but in each treatment condition toxicity was more severe in Ikk␤ Ϫ/Ϫ than in wild type cells. Specifically, although 50 M arsenite exposure for 6 h did not cause much toxicity of wild type cells, it induced 20% condensed nuclei, indicative of apo-FIGURE 2. IKK␤ maintains cellular redox levels in arsenite-exposed cells. WT and Ikk␤ Ϫ/Ϫ and in some experiments Ikk␤ Ϫ/Ϫ -Ad and Ikk␤ Ϫ/Ϫ -R cells were exposed to 50 M arsenite for 2 h. A, cells without (green dotted area) or with (blue area) arsenic treatment were labeled for 1 h with 10 M fluorescent probe CM-H 2 DCFDA followed by flow cytometric analysis (A). A total of 10,000 cells were analyzed and divided into low and high DCF fluorescence intensity groups. The percentage of cells in each fluorescence group was quantified, and the figures show a representative result of three independent experiments. B, CM-H 2 DCFDA-labeled cells were examined for ROS accumulation by DCF fluorescence intensity (green), and nuclei were identified after staining by DAPI (blue). C, cell lysates were subjected to Western blotting using anti-hemeoxygenase-1 (HO-1) and anti-␤-actin. D, Western blotting signals were collected by a chemiluminescence imager (UVP), and the relative density of hemooxygenase-1 and ␤-actin was calculated by Atlas Image software (Version 1.5). E, cells were pretreated with various antioxidants, including 5 mM NAC for 4 h or 100 M butylated hydroxytoluene (BHT) for 1 h and 1 mM vitamin E (Vit E) for 4 h before exposed to arsenic or 100 M hydrogen peroxide (H 2 O 2 ) for 6 h. After DAPI staining, apoptotic cells with condensed nuclei were determined under fluorescence microscopy. Apoptotic cell percentage from three different fields was counted for each group, and results are presented as the mean Ϯ S.D. from three independent experiments. The asterisks indicate significant difference (*, p Ͻ 0.01; **, p Ͻ 0.001) of cell apoptosis compared with that in the wild type cells under the same experimental conditions. ptosis, and 80% of the total LDH release, a marker of apoptotic and necrotic death, in the Ikk␤ Ϫ/Ϫ cells. The PARP cleavage and TUNEL staining, measuring caspase activation and genomic DNA damage, respectively, were both markers of apoptosis and were both more highly induced in the arsenictreated Ikk␤ Ϫ/Ϫ than in wild type cells (Fig. 1, G and H). Importantly, reconstituting IKK␤ expression in Ikk␤ Ϫ/Ϫ -R and Ikk␤ Ϫ/Ϫ -Ad completely protected the Ikk␤ Ϫ/Ϫ cells from arsenic-induced apoptosis ( Fig. 1, C, G, and H).
The arsenic treatment conditions applied in these studies ranged from low dose (1 M) and long time (96 h), which might mimic chronic environmental exposure, to high dose (50 M) and short time (6 h), which might resemble acute accidental/ occupational exposure. Nonetheless, in all conditions tested IKK offered equal effective protection against arsenic toxicity. We, therefore, chose the latter treatment condition in most of the subsequent studies.
Ikk␤ Ϫ/Ϫ Cells Are Defective in Maintaining Redox Levels after Arsenic Exposure-Arsenic is known to elicit intracellular ROS, which may be the cause of massive death of Ikk␤ Ϫ/Ϫ cells. To test this possibility, wild type, Ikk␤ Ϫ/Ϫ , Ikk␤ Ϫ/Ϫ -R, and Ikk␤ Ϫ/Ϫ -Ad cells were labeled with CM-H 2 DCFDA after arsenic exposure, and ROS-activated fluorescence positive cells were identified by flow cytometry and fluorescence microscopy. There was a slight increase in fluorescence levels in all arsenic-exposed cells; however, the intensity was significantly higher in Ikk␤ Ϫ/Ϫ than in wild type, Ikk␤ Ϫ/Ϫ -R, and Ikk␤ Ϫ/Ϫ -Ad cells (Fig. 2, A and B). Moreover, Western blot analyses showed that heme oxygenase-1, whose expression is commonly used as a marker for elevated intracellular ROS, was highly induced by arsenic in Ikk␤ Ϫ/Ϫ but not much so in wild type and Ikk␤ Ϫ/Ϫ -R cells (Fig. 2, C and D). Adenoviral-mediated transient IKK␤ expression also significantly reduced apoptosis of the Ikk␤ Ϫ/Ϫ cells in response to arsenic (Fig. 2E).
To investigate whether ROS were responsible for arsenic toxicity, we used antioxidant agents to modulate cellular redox levels; NAC was used to increase the intracellular concentration of cysteine and up-regulate the level of glutathione, and butylated hydroxytoluene and vitamin E were to prevent or scavenge the formation of reactive oxygen radicals. Wild type and Ikk␤ Ϫ/Ϫ cells were treated with antioxidants before arsenic exposure. Although all antioxidants, specifically vitamin E and NAC, significantly prevented apoptosis of Ikk␤ Ϫ/Ϫ cells, none of the antioxidants seemed to affect wild type cells, which did not have high levels of ROS and apoptosis induced by arsenic. The Ikk␤ Ϫ/Ϫ cells were also extremely sensitive to exogenous H 2 O 2 , suggesting that these cells might lack ROS scavengers (Fig. 2E).
The Role of IKK␤ in ROS Modulation Is Likely Mediated through NF-B-Lack of ROS scavenging capacity in Ikk␤ Ϫ/Ϫ cells may be due to defective activation of NF-B, a redoxsensitive transcription factor known to play an important role in antioxidant gene expression. Indeed, TNF␣, a known IKK-NF-B pathway activator, caused an apparent induction of IB␣ degradation and NF-B promoter-driven luciferase activity in wild type but not in Ikk␤ Ϫ/Ϫ cells (Figs. 3, A  and B). On the other hand, neither arsenic nor H 2 O 2 induced IB␣ degradation in wild type or Ikk␤ Ϫ/Ϫ cells, suggesting that these agents did not cause an apparent activation of NF-B through immediate induction of the IKK␤-IB␣ pathway (Fig. 3B).
To determine whether NF-B has a role in arsenic toxicity, we pretreated wild type cells with a NF-B inhibitor for 2 h before they were exposed to sodium arsenite. NF-B inhibitortreated cells behaved like Ikk␤ Ϫ/Ϫ cells, exhibiting increased ROS accumulation and apoptosis after arsenic exposure (Fig. 3,  C and D). Moreover, cells deficient in the p65 (42) NF-B component exhibited an obvious induction of PARP cleavage upon arsenic exposure (Fig. 3E). Thus, there is the existence of an IKK␤-NF-B pathway that helps to maintain cellular redox levels and prevent apoptosis in response to arsenic. . E, WT and p65 Ϫ/Ϫ cells were exposed to 50 M arsenite for the indicated times. Cell lysates were subjected to Western blotting using anti-PARP. Cleaved PARP (cPARP) was apparently induced in the p65 Ϫ/Ϫ cells exposed to arsenite for 6 h. JULY 20, 2007 • VOLUME 282 • NUMBER 29

JOURNAL OF BIOLOGICAL CHEMISTRY 21491
A Role for the IKK␤-NF-B Pathway in Antioxidant Gene Expression-Based on the above observations, we hypothesized that IKK␤ ablation might prevent NF-B activation by cytokines under normal growth conditions, leading to defective antioxidant gene expression in Ikk␤ Ϫ/Ϫ cells (36,38). This idea was tested by examination of the basal transcription levels of key antioxidant genes in wild type and Ikk␤ Ϫ/Ϫ cells using realtime PCR (Fig. 4A). There was less mRNA accumulation for Mt1, Mt2, mitochondrial Sod2, cytosolic Sod1, and glutathione peroxidase in Ikk␤ Ϫ/Ϫ than in wild type cells; conversely, the expression of catalase was slightly increased in IKK␤-null cells. Reconstitution of IKK␤ expression in Ikk␤ Ϫ/Ϫ cells could partially restore the expression of Mt1, Mt2, Sod1, and Sod2, whereas it did not rescue the expression of glutathione peroxidase and catalase.
Among the IKK␤-dependent genes, the most strikingly changed was Mt1, as its expression levels were high in wild type cells but were suppressed by as much as 4000-fold in the Ikk␤ Ϫ/Ϫ cells (Figs. 4, A and B). Although arsenic, like TNF␣, slightly enhanced Mt1 transcription in wild type and IKK␤reconstituted Ikk␤ Ϫ/Ϫ cells, it failed to activate Mt1 expression in Ikk␤ Ϫ/Ϫ cells (Fig. 4B). To determine whether the basal MT1 expression was under the control of NF-B, we pretreated wild type cells with NF-B-I for 2 h, which resulted in a 4-fold suppression of MT1 expression (Fig. 4C). Taken together, our results suggest a crucial role for IKK␤ in MT1 expression, which may be mediated through NF-B.
If lacking MT1 expression were the reason for the arsenic sensitivity of Ikk␤ Ϫ/Ϫ cells, re-introducing MT1 expression should rescue the Ikk␤ Ϫ/Ϫ cells from apoptosis after arsenic exposure. To test this hypothesis, expression vectors for MT1 and a GFP were used to transfect Ikk␤ Ϫ/Ϫ cells. Although the GFP-positive cells in MT1 and GFP co-expression displayed marked resistance to arsenite toxicity, those in GFP expression alone were similar to the parental Ikk␤ Ϫ/Ϫ cells, displaying high sensitivity to arsenite-induced apoptosis (Fig. 4D).
Arsenic-induced ROS Triggers the MKK4-JNK Apoptotic Pathway-To search for the molecular events downstream of ROS that might be responsible for arsenic toxicity, we looked at MAP kinase activation, known to be induced by cellular stress. A strong induction of JNK and ERK phosphorylation but a weak induction of p38 phosphorylation was observed in arsenic-treated Ikk␤ Ϫ/Ϫ but not in wild type cells (Fig. 5A). To evaluate the contributions of MAP kinases to arsenic toxicity, wild type and Ikk␤ Ϫ/Ϫ cells were pretreated with specific MAP kinase inhibitors: SP600125 for JNK, SB202190 for p38, and PD 98059 for ERK. Although the JNK inhibitor significantly reduced arsenic induced apoptosis of the Ikk␤ Ϫ/Ϫ , the p38 inhibitor had less and the ERK inhibitor had no effect on arsenic toxicity (Fig. 5B). Based on these results, we propose that JNK activation is mainly responsible for Ikk␤ Ϫ/Ϫ cell toxicity in response to arsenite.
The onset and level of MAP kinase phosphorylation correlated very well with the ROS status, as induction was strong in Ikk␤ Ϫ/Ϫ but weak if at all in wild type cells (Fig. 5A). The onset of ROS formation was first detectable at 2 h of exposure, and its level increased with time (data not shown); correspondingly, weak JNK phosphorylation was detected at 2 h, became stron-  A and B) or with exposure for 16 h to 20 ng/ml TNF␣ (Tn) (B) and 20 ng/ml TNF␣ and 10 M sodium arsenite (As). cDNA prepared from these samples was used for real-time PCR to detect transcripts of antioxidant genes and Gapdh. A, the relative levels of antioxidant genes/Gapdh in Ikk␤ Ϫ/Ϫ and Ikk␤ Ϫ/Ϫ -R were compared with those in WT cells, designated as 1. Cat, catalase; Gpx, glutathione peroxidase. B, Mt1 mRNA levels were determined by real-time quantitative PCR. The Mt1/Gapdh level in each treatment condition was compared with the untreated samples of the same cells. C, WT cells with or without treatment by NF-B-I for 2 h. RNA was isolated 24 h after removal of the NF-B-I, and Mt1 levels were measured as described above. D, the Ikk␤ Ϫ/Ϫ MEFs with or without MT1 and GFP co-transfection were exposed to 50 M sodium arsenite for 6 h followed by DAPI staining. The condensed nuclei were identified under fluorescence microscopy, and the relative apoptotic cell numbers were quantified in three different fields. Transfection efficiency, evaluated by GFP-positive cells, reached up to 80%. Results are presented as the mean Ϯ S.D. from at least three independent experiments. Statistical analyses were performed. The asterisks indicate significant difference (*, p Ͻ 0.01; **, p Ͻ 0.001; ***, p Ͻ 0.0001) compared with the wild type cells (A) or untreated samples (B-D), whereas # indicates significant difference (#, p Ͻ 0.001) between Ikk␤ Ϫ/Ϫ and Ikk␤ Ϫ/Ϫ -R. ger at 6 h, and persisted for at least 24 h post-exposure. Moreover, pretreatment of cells with NAC completely blocked JNK phosphorylation and reduced hemeoxygenase-1 expression induced by arsenic (Fig. 5C). Hence, arsenic-induced excessive ROS accumulation in Ikk␤ Ϫ/Ϫ cells is likely responsible for JNK activation during the delayed phase of exposure.
To investigate whether an upstream signaling pathway was involved in JNK activation, we examined the phosphorylation of MKK4 and MKK7, the direct kinases for JNK. Arsenite exposure did not cause strong induction of MKK4 and MKK7 phosphorylation in wild type cells, but it induced phospho-MKK4 in Ikk␤ Ϫ/Ϫ cells in a exposure-and time-dependent manner, similar to the induction pattern of phospho-JNK (Fig. 5D). On the contrary, phospho-MKK7 levels were higher in Ikk␤ Ϫ/Ϫ than in wild type cells, but its level was not further increased after arsenite exposure.
To further evaluate the roles of MKK4 and MKK7 in JNK signaling, we examined phospho-JNK levels in wild type, Mkk4 Ϫ/Ϫ , and Mkk7 Ϫ/Ϫ mouse ES cells exposed to various concentrations of sodium arsenite for the indicated times (Fig.  5E). Again, there was an arsenite concentration-and exposure time-dependent increase of phospho-JNK, which was detected only in wild type and Mkk7 Ϫ/Ϫ but not in Mkk4 Ϫ/Ϫ ES cells. In agreement with the notion that JNK activity was essential for arsenic to induce apoptosis, the Mkk4 Ϫ/Ϫ cells were largely resistant to arsenic toxicity, in contrast to wild type and Mkk7 Ϫ/Ϫ ES cells that underwent significant apoptosis after arsenic exposure (Fig. 5F).

NF-B-mediated MT1 Expression Connects the IKK␤ Pathway and the JNK-c-Jun Pathway-To
evaluate the role of NF-B in arsenite-induced JNK activation, wild type cells were pretreated with NF-B-I, which itself caused a slight induction of JNK phosphorylation (Fig. 6A). Exposure to arsenic led to a weak JNK phosphorylation in wild type cells but a much stronger JNK phosphorylation in NF-B-I-treated cells. Hence, similar to IKK␤ ablation, inhibition of NF-B potentiates JNK activation in response to arsenite.
Next, we asked whether JNK activation signals were further transmitted to the nucleus. We examined the phosphorylation of c-Jun, a well defined nuclear phosphorylation target of JNK. In arsenic-exposed Ikk␤ Ϫ/Ϫ cells, there was a strong induction of c-Jun phosphorylation correlated to elevated ROS accumulation; in contrast, in wild type cells there was no detectable ROS and c-Jun phosphorylation (Fig. 6B). Pretreatment of the AP-1-luc cells with NF-B-I also resulted in nearly 2-fold more induction of luciferase activity by arsenic, supporting the notion that inhibition of IKK-NF-B pathway led to enhanced c-Jun and AP-1 activation in the nucleus.
Interestingly, transient overexpression of MT1 in Ikk␤ Ϫ/Ϫ cells completely prevented arsenite from inducing ROS and activating c-Jun, underlying the predominant role of MT1 in modulating the IKK-JNK axis in arsenic toxicity (Fig. 6B). Taken together, our results establish a critical role for the IKK␤ pathway in MT1 expression and redox homeostasis, uncovering the mechanisms by which IKK␤ prevents activation of the apoptotic JNK pathway during arsenic toxicity.

DISCUSSION
Our results clearly support a role for IKK␤ in offering protection against arsenic toxicity (Fig. 6D). Under normal growth conditions, IKK␤ is required for NF-B activation and antioxidant gene expression, which constitutes the redox modulating machinery that counteracts the oxidative stress caused by arsenite. Inactivation of this pathway in Ikk␤ Ϫ/Ϫ cells results in high sensitivity to arsenic-induced ROS accumulation and cell death, whereas reconstituting IKK␤ expression either transiently by adenovirus-mediated gene expression or stably by introducing IKK␤ cDNA reverses the arsenic sensitivity of Ikk␤ Ϫ/Ϫ cells.
The results of our work are consistent with several previous studies. A dominant-inactive IB has been shown to prevent NF-B activation, thereby rendering cells sensitive to arsenic (43). Moreover, IKK␤-null MEFs are highly sensitive to arsenic FIGURE 5. Arsenic-induced JNK activation in Ikk␤ ؊/؊ cells leads to apoptotic cell death. WT and Ikk␤ Ϫ/Ϫ cells were exposed to 50 M sodium arsenite. At various times post-exposure, cell lysates were examined by Western blotting for the phosphorylation (p) of JNK, p38, ERK, and total ␤-actin (A) and the phosphorylation of MKK4 and MKK7, and total ␤-actin (D). B, cells were pretreated with 5 M MAP kinase inhibitors for 1 h before they were exposed to 50 M sodium arsenite for 6 h. The inhibitors used include JNK inhibitor SP600125 (SP), p38 inhibitor SB202190 (SB), and ERK inhibitor PD 98059 (PD). Apoptotic cells were identified under fluorescence microscopy after DAPI staining. C, cells were pretreated with 5 mM NAC for 4 h followed by 50 M sodium arsenite treatment for various times (0 -10 h). Cell lysates were subjected to Western blotting using antibodies to phospho-JNK, hemeoxygenase-1 (HO-1), and ␤-actin. E, WT, Mkk4 Ϫ/Ϫ , and Mkk7 Ϫ/Ϫ ES cells were either left untreated or treated by various concentrations of sodium arsenite for the indicated times. Cell lysates were used for Western blotting for phospho-JNK. F, WT, Mkk4 Ϫ/Ϫ , and Mkk7 Ϫ/Ϫ ES cells were exposed to 50 M arsenite for 6 h and stained by DAPI. Apoptotic and total cells from three different fields were counted for each group of triplicate samples. Results are presented as the mean Ϯ S.D. from three independent experiments. The asterisks indicated significant difference (*, p Ͻ 0.01; **, p Ͻ 0.001) of the same cell compared with control untreated samples; # indicates significant difference (#, p Ͻ 0.001) compared with WT cells under the same treatment conditions. toxicity, displaying excessive H 2 O 2 accumulation (44) and higher levels of GADD45␣ expression after exposure (45). However, a recent study using IKK␤-null MEFs originated from precisely the same source has reached a completely opposite conclusion. In their studies, Song et al. (46) report that IKK␤ ablation protects cells against apoptosis in response to arsenic and that the IKK␤-null MEFs have reduced levels of GADD45␣.
One possible explanation for the apparent contradicting results is that our knock-out cells are mistaken for wild type cells and vice versa. We performed genotyping analyses and detected the mutant Ikk␤ allele only in knock-out but not in wild type cells. The genotypes together with IKK␤ protein expression patterns, confirmed the correct identity of cells used in our studies. When IKK␤ is re-expressed, as shown in Ikk␤ Ϫ/Ϫ -R and Ikk␤ Ϫ/Ϫ -Ad, it is clearly sufficient for protecting the Ikk␤ Ϫ/Ϫ cells from arsenic-induced ROS accumulation, JNK activation, and apoptosis. Moreover, ablation of the NF-B p65 subunit also renders the p65 Ϫ/Ϫ cells highly sensitive to arsenic-induced cell apoptosis. All these results together support a role of the IKK-NF-B pathway in protecting cells from arsenic toxicity.
Several lines of evidence provide novel mechanistic insight of the role of IKK␤ in arsenic toxicity. First, we have identified MT1 as one of the major transcription targets of the IKK␤ pathway. NF-B has been previously shown to regulate the expression of antioxidant defense enzymes, including mitochondrial SOD2 and cytosolic SOD1, glutathione peroxidase, and catalase (36). These enzymes metabolize and detoxify superoxide anions and H 2 O 2 to H 2 O and O 2 . In addition, IKK␤ has been shown to negatively regulate Cyp1b1, whose expression at least partly is responsible for sensitivity to arsenic toxicity (44). We find only a slightly reduction in the expression of detoxification enzymes SOD1 and SOD2 in the Ikk␤ Ϫ/Ϫ cells, whereas the expression of glutathione peroxidase and catalase is not strictly dependent on IKK␤. The most strikingly affected gene by IKK␤ ablation is Mt1, whose expression is markedly decreased in the Ikk␤ Ϫ/Ϫ cells and restored by reintroducing IKK␤ expression in these cells.
Second, we have established a crucial role for MT1 in counteracting arsenic-induced ROS and apoptosis. MTs are low molecular weight cysteine-rich metal-binding proteins that have multiple functions, such as detoxification of nonessential metals and protection of cells against oxidative damage (47,48). Introducing MT1 expression into the Ikk␤ Ϫ/Ϫ cells protects against ROS accumulation, JNK pathway activation, and apoptosis induced by arsenic. Hence, MT1 appears to act as a mediator between IKK and JNK pathways, and high levels of MT1 expression are sufficient for protection against arsenic-induced apoptosis (49 -51).
Third, we have shown that ROS activate the MKK4-JNK pathway, leading to apoptosis; however, JNK activation is not the consequence of a JNK phosphatase inactivation as shown previously (38). ROS act as second messengers to modulate many signal transduction pathways through oxidation of critical target molecules such as protein kinase C and protein-tyrosine phosphatases. Previous studies show that TNF␣-mediated ROS in NF-B-deficient cells cause sustained JNK activation through inactivation of a JNK-specific protein-tyrosine phosphatase (38). In the Ikk␤ Ϫ/Ϫ cells, arsenite-induced ROS accumulation also leads to prolonged JNK activation, which however, is mediated through activation of a specific JNK upstream kinase MKK4 but not MKK7. The activation of MKK4 may in turn result from the activation of members of M NF-B-I pretreatment were exposed to 50 M arsenite for 6 h. Cell lysates were subjected to Western blot (IB) analyses for phospho (p)-JNK. B, WT, Ikk␤ Ϫ/Ϫ cells, and Ikk␤ Ϫ/Ϫ transiently transfected with a Mt1 expression vector were exposed to 50 M arsenite for 6 h. The cells were labeled with 10 M CM-H2DCFDA for 30 min before harvesting and fixation. The cells were used for immunofluorescence staining using anti-phospho-c-Jun, and nuclei were identified by DAPI staining. The DCFDA (green)-and phospho-c-Jun (red)-positive cells and total cells (blue) were identified under fluorescence microscopy. The picture shows a representative of three experiments. C, some AP-1-luc MEFs were exposed to NF-B-I for 2 h. Twenty-four h later the cells were either left untreated or exposed to 50 M sodium arsenite for 12 h, and luciferase activities were measured. Statistical analyses were performed. The asterisks indicate significant difference (*, p Ͻ 0.001) compared with untreated cells. D, a diagram describing the role of the IKK␤-NF-B pathway in the control of MT1 expression, which in turn is crucial for modulation the ROS levels in response to arsenic. Excessive ROS lead to the activation of the MKK4-JNK pathway, resulting in apoptosis.
MAP kinase kinase kinase (MAP3K), as recent findings by Yan et al. (52) show that arsenite activates Ask1, a redox-sensitive MAP3K. Hence, arsenic-elicited ROS likely activate a signaling cascade of MAP3K-MKK4 that leads to JNK activation.
The ROS effect may not be very specific, because we find strong induction of JNK and ERK and to a lesser extent p38 phosphorylation in Ikk␤ Ϫ/Ϫ cells. Interestingly, only JNK activity is required for arsenite toxicity, as a JNK inhibitor or MKK4 ablation markedly decreased apoptosis, whereas inhibition of ERK does not affect toxicity induced by arsenite in the Ikk␤ Ϫ/Ϫ cells. The ERK pathway is commonly associated with cell proliferation and cell cycle progression; hence, its activation may contribute to other toxic outcomes of arsenite exposure, such as tumor promotion.
One caveat is the role of NF-B in the regulation of MT1 expression. Several transcription factors, such as SP1, AP-1, and MTF1, have been shown implicated in the regulation of MT, but a role for the IKK or NF-B in MT regulation has never been previously established. Because the MT1 gene promoter contains NF-B binding sites, it is possible that the IKK␤-mediated NF-B activation transcriptionally up-regulates MT1 expression, which would take place only in wild type and Ikk␤ Ϫ/Ϫ -R cells but not in Ikk␤ Ϫ/Ϫ cells. However, NF-B inhibitor treatment of wild type cells results in a 4-fold reduction of MT1 expression, way below the reduction caused by IKK␤ ablation. One possibility is that the Ikk␤ Ϫ/Ϫ cells represent a state of chronic IKK-NF-B pathway inactivation, having molecular alterations that cannot be mimicked by a transient NF-B inhibition by the chemical compounds. The NF-B inhibitor II exerts cytotoxicity when used for long-term treatment (data not shown). An alternative possibility is that the IKK␤ may control MT1 expression through other downstream effectors in addition to or independently of NF-B. These interesting prospects require further investigation.
Arsenic exerts diverse toxic and therapeutic effects that may be attributed to ROS, which in turn act as critical signaling molecules to modulate biological systems, such as carcinogenesis and apoptosis. In this regard, modulating NF-B activity may be used to directly affect arsenic toxicity. Activation of the IKK-NF-B pathway increases ROS scavenging capacity that may depress tumor growth, whereas inhibition of this pathway may result in excess ROS that would trigger cancer cell apoptosis. Detailed molecular studies suggest additional approaches may exist, such as changing the levels of MT1 expression and JNK activity, which may lead to more specific and convenient means to modulate arsenic toxicity.