Coordination of JNK1 and JNK2 Is Critical for GADD45α Induction and Its Mediated Cell Apoptosis in Arsenite Responses*

Arsenite is a well documented environmental pathogen, whereas it has also been applied as medication to treat various neoplasmas. The pathogenic and therapeutic effects of arsenite are associated with cellular apoptotic responses. However, the molecular mechanisms of arsenite-induced apoptosis are not very well understood. Our previous study has shown that arsenite exposure is able to activate JNKs, which subsequently mediate the apoptotic outcome. The present study further revealed that the coordination of JNK1 and JNK2 was critical for the arsenite-induced expression of GADD45α (growth arrest and DNA damage 45α), which in turn mediated the cellular apoptosis. The arsenite-induced apoptosis and GADD45α expression were significantly impaired in mouse embryonic fibroblasts deficient in either jnk1 (JNK1–/–) or jnk2 (JNK2–/–). Knockdown of GADD45α by its specific small interfering RNA also dramatically reduced the apoptotic responses, and overexpression of GADD45α in either JNK1–/– or JNK2–/– mouse embryonic fibroblasts partially resensitized the cell death. Furthermore, it was found that the regulation of GADD45α by JNK1 and JNK2 was achieved through mediating the activation of c-Jun, since in the JNK1–/– and JNK2–/– cells the c-Jun activation was impaired, and overexpression of the dominant negative mutant of c-Jun (TAM67) in wild type cells could also block GADD45α induction as well as cellular apoptosis. Our results demonstrate that the coordination of JNK1 and JNK2 is critical for c-Jun/GADD45α-mediated cellular apoptosis induced by arsenite.

Arsenite occurs naturally in the earth's crust and is widely distributed in the environment (1). Human exposure to arsenite occurs mainly by ingestion of drinking water contaminated with arsenite from naturally occurring sources or through the inhalation of contaminated dusts in occupational settings (2). As an environmental pathogen, arsenite causes a series of pathophysiological alterations, including immunosuppression and carcinogenesis (3). Paradoxically, arsenite has also been applied as chemotherapeutic reagents to treat various neoplas-mas (4). Further insights into the pathogenic and therapeutic effects of arsenite show that it seems to be associated with the apoptotic induction in both normal and tumor cells (5,6). The inhibition of antiapoptotic Bcl-2 and activation of proapoptotic mitogen-activated protein kinases has recently been shown to participate in the arsenite-induced apoptotic process in various cell models (7)(8)(9)(10)(11)(12). These observations suggest that the alternation of the cascades of cellular survival/proapoptotic signaling pathways may be critical for cell apoptotic responses triggered by arsenite. However, the detailed molecular mechanisms still remain to be elucidated.
In this study, we applied wild type (WT), 2 JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ , three immortalized mouse embryonic fibroblasts (MEFs) to evaluate the roles of JNK1 and JNK2 in arseniteinduced apoptosis. It was found that the arsenite-induced apoptosis required both JNK1 and JNK2, since the deficiency of either impaired the apoptotic responses. Interestingly, GADD45␣ (growth arrest and DNA damage 45␣) was detected to be a downstream target gene transcriptionally regulated by either JNK1 or JNK2, and it functioned as the critical mediator for the JNK1-and JNK2-associated apoptosis by arsenite. In addition, the regulation of GADD45␣ by JNK1 and JNK2 was, at least in part, dependent on c-Jun activation.

EXPERIMENTAL PROCEDURES
Cell Culture-Immortalized WT, JNK1 Ϫ/Ϫ (13), and JNK2 Ϫ/Ϫ (14) MEFs as well as their stable transfectants were maintained at 37°C in 5% CO 2 incubator with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 25 g/ml gentamicin. The cultures were dissociated with trypsin and transferred to new 75-cm 2 culture flasks (Fisher) twice a week. Fetal bovine serum was purchased from Nova-Tech (Grand Island, NE), and the rest of the cell culture reagents were obtained from Sigma.
Plasmids and Transfection-HA-tagged murine jnk1 and jnk2 full-length cDNAs were subcloned into pcDNA3 expression vector, confirmed by DNA sequencing, and then named as HA-JNK1/pcDNA3 and HA-JNK2/pcDNA3, respectively (15,16). A pcDNA3.1 plasmid containing c-Jun dominant negative mutant (pcDNA3.1/His-TAM67) was kindly provided by Dr. Tim G. Bowden (College of Pharmacy, University of Arizona, Tucson, AZ) and Dr. Matthew Young (Center for Cancer Research, NCI-Frederick) (17). An expression construct containing HA-tagged full-length cDNA of murine gadd45␣ (HA-GADD45␣) was described previously (18). HA-tagged JNKK2-JNK1 fusion protein expression vector was a generous gift from Dr. Han-Ming Shen (Department of Community, Occupational, and Family Medicine, Faculty of Medicine, National University of Singapore). Transfection experiments were performed with Lipofectamine2000 (Invitrogen) according to the instructions from the manufacturer. For the transfection of pcDNA3.1/His-TAM67 into WT MEFs, 5 g of plasmids were used, and the stable transfectants were generated by G418 selection (500 g/ml). For the transfection of HA-JNK1/pcDNA3, HA-JNK2/pcDNA3, or HA-GADD45␣ in JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ MEFs, 5 g of the individual plasmids were co-transfected with 0.8 g of the hygromycin-resistant plasmid, respectively. The stable transfectants were established by hygromycin selections (50 g/ml for JNK1 Ϫ/Ϫ cells and 400 g/ml for JNK2 Ϫ/Ϫ cells).
The GADD45␣ siRNA expression plasmids were made by using the GeneSuppressor TM system (Imgenex Co., San Diego, CA). The two siRNA target sequences for gadd45␣ were as follows: 5Ј-GTG CTC AGC AAG GCT CGG A-3Ј (siRNA1) and 5Ј-GCT GCT CAA CGT AGA CCC C-3Ј (siRNA2). Constructs containing the reversed target sequences were used as a negative control. The GADD45␣ siRNA1 and siRNA2 were cotransfected into WT MEFs, and the stable transfectants were established by G418 selection.
Flow Cytometry-To analyze the apoptotic cells with propidium iodide staining, WT, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ cells were plated in 6-well plates with a density of 2 ϫ 10 5 cells/well and cultured in normal 10% serum medium until 70 -80% confluence. After exposure to 20 M sodium arsenite (Fisher) for 24 h, the cells were collected by centrifugation and fixed in ice-cold 80% ethanol at Ϫ20°C overnight. The fixed cells were stained in the buffer containing 100 mM sodium citrate, 0.1% Triton X-100, 0.2 mg/ml RNase A, and 50 g/ml propidium iodide at 4°C for 1 h and then analyzed by an Epics XL fluorescenceactivated cell sorter (Beckman Coulter, Miami, FL) as described in our previous publication (19).
To determine the cell mitochondrial membrane potential by fluorochrome DiOC 6 staining, the WT cells were seeded in 6-well plates. After exposure to 20 M arsenite for 24 h, the cells were applied with 10 g/ml DiOC 6 dye (dissolved in Me 2 SO and diluted with phosphate-buffered saline) for 30 min. The stained cells were then washed twice with phosphate-buffered saline, harvested, and analyzed by flow cytometry (20).
To detect the activated caspase-3, the APO ACTIVE 3 TM kit (Cell Technology Inc., Minneapolis, MN) was applied as previously described (21). Briefly, the cells were harvested by centrifugation 24 h after arsenite exposure, fixed at room temperature for 30 min, labeled with anti-activated caspase-3 antibody for 1 h, and then incubated with fluorescein isothiocyanate-conjugated secondary antibody in the dark for 30 min. The cells labeled with anti-activated caspase-3 antibody were detected by flow cytometry.
Polymerase Chain Reaction (PCR) for jnk1 and jnk2 Gene Knock-out Identification-To identify jnk1 and jnk2 gene deficiencies in JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ cells, genomic DNA was isolated from the cells, and the deleted targets of jnk1 and jnk2 sequences were amplified by PCR using two pairs of primers designed from the jnk1 and jnk2 genomic sequences according to previous publications with some modifications (13,14). The primer sequences for jnk1 were 5Ј-CGT CTG GTG GAA GGA GAG AG-3Ј (sense primer) and 5Ј-TAA TAA CGG GGG TGG AGG AT-3Ј (antisense primer) (13). The primer sequences for jnk2 were 5Ј-TCT GAC GTC CTG GGC TGG AC-3Ј (sense primer) and 5Ј-GCA GCA GCC CTC AGG ATC CT-3Ј (antisense primer) (14). The PCR products were separated on 2% agarose gels and stained with ethidium bromide, and the images were scanned using a UV light.
Reverse Transcription Polymerase Chain Reaction (RT-PCR)-Twelve hours after exposure to arsenite, cells were collected, and total RNA was extracted from the cells using TRIzol reagent (Invitrogen). Total cDNA was synthesized by ThermoScript TM RT-PCR system (Invitrogen). The amount of GADD45␣ mRNA was measured by semiquantitative RT-PCR using a pair of primers (5Ј-ATG ACT TTG GAG GAA TTC TCG-3Ј and 5Ј-CAC TGA TCC ATG TAG CGA CTT-3Ј). The control mouse ␤-actin mRNA was also detected by RT-PCR using the primers (5Ј-GAC GAT GAT ATT GCC GCA CT-3Ј and 5Ј-GAT ACC ACG CT T GCT CTG AG-3Ј). The PCR products were separated on 2% agarose gels and stained with ethidium bromide, and the images were scanned with a UV light.
Western Blotting-MEFs and their transfectants were plated in the 6-well plates and cultured in normal 10% serum medium until 70 -80% confluence. After exposure to arsenite for various doses and time periods as indicated in the figure legends, the cells were washed once with ice-cold phosphate-buffered saline and collected with SDS-sample buffer (22). The cell extracts were sonicated, denatured by heating at 100°C for 5 min, and quantified with a Dc protein assay kit (Bio-Rad). Equal aliquots of cell extracts were separated on SDS-polyacrylamide gels. The proteins were then transferred to polyvinylidene difluoride membranes (Bio-Rad), blocked, and probed with one of the polyclonal antibodies against phospho-specific c-Jun and JNK1/2, nonphosphorylated c-Jun and JNK1/2, HA (Cell Signaling Technology, Beverly, MA), poly(ADP) polymerase (PARP), caspase-3, GADD45␣ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), or ␤-actin (Sigma). Primary antibody-bound proteins were detected by using an alkaline phosphatase-linked secondary antibody and an ECF Western blotting system (Amersham Biosciences) (23).

Arsenite Induces Apoptosis in MEFs-
To study the effect of arsenite in cell apoptosis, MEFs were incubated with 20 M arsenite for 24 h. The key step of the apoptotic process is chromosome DNA degradation. The degraded DNA fragments are then encapsulated in the apoptotic bodies that can be detected by flow cytometry as sub-G 0 /G 1 phase cells (24). As shown in Fig. 1A, exposure of MEFs to arsenite resulted in significant accumulation of sub-G 0 /G 1 phase cells by 55.50% compared with 1.12% in the cells without arsenite exposure, indicating that arsenite caused the cellular DNA fragmentation in the MEFs. Next we examined whether arsenite-induced DNA fragmentation was associated with caspase-3 activation, the key executor of apoptosis, by using an APO ACTIVE 3 TM kit and Western blotting assay. Arsenite exposure led to over 8-fold increases of the activated caspase-3 compared with that of medium control cells detected by flow cytometry (36.87% versus 4.53%) (Fig. 1B). Western blotting data further confirmed the presence of the activated caspase-3 (17-kDa band) (Fig. 1C). Moreover, PARP, an intracellular substrate of the activated caspase-3, was also detected cleaved from 116-kDa to 85-kDa fragments after arsenite treatment (Fig. 1C). The MEFs were also exposed to 5 M arsenite for 24 h; however, no obvious cell death was observed, as indicated by the absence of caspase-3 activation and PARP cleavage (Fig. 1C), so 5 M of arsenite treatment was used as a negative control in the following studies. It may be noticed that the cytotoxicity of arsenite to cells is dependent on the forms of arsenite used (25,26) and cell types as well as species (27,28). For example, the cytotoxicity as well as cellular biological effects of arsenic trioxide and sodium arsenite are quite different when they are compared at the same concentration, which has been observed from the findings of various laboratories (5, 10). Mitochondria potential reduction has been reported to be critical for the apoptotic induction through the mitochondria pathway (29,30); therefore, the mitochondria potential was measured by fluorochrome DiOC 6 staining. As shown in Fig. 1D, DiOC 6 incorporation was 35% less in the arsenite-treated cells compared with that of the medium control cells (53.73% versus 89.25%), indicating that arsenite exposure reduced the mitochondrial membrane potential. These data strongly indicate that arsenite is able to induce apoptosis in MEFs.

Coordination of JNK1 and JNK2 Is Required for the Arseniteinduced Apoptosis-
The proapoptotic role of JNKs has been proposed recently in different cellular conditions (8,10,12,31), but their precise molecular mechanisms still remain unclear. To gain further insight into the JNK1-and/or JNK2-dependent proapoptotic signal pathway, we compared the apoptosis induction by arsenite in WT, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ MEFs. Before performing the comparison experiments, the jnk1 and jnk2 gene deficiencies in their knock-out MEFs were initially confirmed by both Western blotting and PCR ( Fig. 2A). Interestingly, both JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ cells were resistant to arsenite-induced apoptosis, as indicated by the absence of the increased sub-G 0 /G 1 phase cells after the arsenite treatment for 24 h in the two knock-out cell lines (7.88% in JNK1 Ϫ/Ϫ cells and 0.70% in JNK2 Ϫ/Ϫ cells) compared with that in WT cells (53.49%) (Fig. 2B). Similarly, neither caspase-3 activation nor PARP cleavage was observed in the arsenite-treated JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ MEFs (Fig. 2C). Even when the arsenite treatment time was prolonged to 48 h, there was still no obvious cell death observed in the two knock-out cell lines (Fig. 2D). All of these data suggested that JNK1 and JNK2 were both required for arsenite-induced apoptosis, and a deficiency of either one did impair the apoptotic response. However, the cell death induced by other environmental carcinogens, such as nickel chloride, was not different among JNK1 Ϫ/Ϫ , JNK2 Ϫ/Ϫ , and WT MEFs, as indicated by morphological changes (Fig. 2E) or caspase-3 activation and PARP cleavage (Fig. 2F). This demonstrates that the escape from apoptosis due to the deficiency of either jnk1 or jnk2 is a relatively specific feature in the cells, at least in MEFs, when exposed to arsenite. WT MEFs (2 ϫ 10 5 ) were seeded into each well of 6-well plates and cultured until the cell density reached 70 -80% confluence. The cells were then exposed to 5 and/or 20 M of sodium arsenite for 24 h. The chromosome DNA fragmentation indicated by the proportion of sub-G 0 /G 1 phase cells was determined using propidium iodide staining and detected by flow cytometry (A). Caspase-3 activation was detected using an APO ACTIVE 3 TM kit (B) as well as Western blotting (C ). PARP cleavage was detected by Western blotting (C ). The mitochondrial membrane potential was determined using fluorochrome dye DiOC6 and detected by flow cytometry (D).
To further confirm the essential requirement of JNK1 and JNK2 in arsenite-induced apoptosis, we stably reconstituted the HA-tagged JNK1 or JNK2 into their deficient MEFs (Fig. 3A). As expected, the restored expression of either JNK1 or JNK2 in each knock-out cell line mostly returned their abilities as WT MEFs to undergo apoptosis as indicated by caspase-3 activation and PARP cleavage (Fig. 3B). It need be noted here that the stable transfectants of JNK1 or JNK2 reconstitutional cells were established as a mass pool, which might include a proportion of unsuccessfully transfected cells, so the exogenous JNK1 or JNK2 was expressed much lower as compared with those in WT cells (Fig. 3A).
Expression of GADD45␣ Is Regulated by the c-Jun-dependent Pathway-GADD45␣ was originally identified as a gene transcribed in response to DNA damage by UV irradiation (32) and has been reported to be involved in many biological functions, such as suppressing cell growth (33), participating in DNA damage repair (34), regulating cell cycle G 2 /M checkpoint (35), and mediating apoptotic signal pathways (36,37). To determine whether GADD45␣ is the downstream target gene of JNK1 and JNK2 in responses to arsenite exposure, the GADD45␣ induction was detected in WT and the two JNK knock-out MEFs. As shown in Fig. 4A, treatment with 20 M arsenite obviously induced GADD45␣ protein expression in WT MEFs at all of the time points tested.
It was more important to observe that this induction was blocked in either JNK1 Ϫ/Ϫ or JNK2 Ϫ/Ϫ MEFs, indicating that JNK1 and JNK2 were both critical for the GADD45␣ up-regulation in response to arsenite. This notion was further confirmed by the reconstitution experiments, which showed that restoring JNK1 or JNK2 led to an obvious GADD45␣ induction in their respective knock-out cells, although not as strong as those in the WT MEFs (Fig. 4B), which was consistent with the exogenous JNK1 or JNK2 expression levels of the reconstituted cells (Fig. 3A).
To determine the mechanisms of GADD45␣ induction by JNK1 and JNK2, we also tested the changes of the gadd45␣ mRNA level. The RT-PCR data showed that arsenite-induced gadd45␣ mRNA expression was readily observed in the WT cells; however, it was reduced due to jnk1 deficiency and totally blocked in jnk2-deficient cells (Fig. 5A), coinciding with the basal levels of JNKs protein expression in these two knockout cell lines ( Fig. 2A) as well as their sensitivities to arseniteassociated cell death (Fig. 2B), indicating that JNK1 and JNK2 were required for the arsenite-associated GADD45␣ mRNA induction.
Transcription factor c-Jun is a well known downstream target of JNKs (38). To evaluate the relevance of c-Jun for JNK1and JNK2-mediated GADD45␣ induction, the activation of JNK and c-Jun in WT, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ MEFs were ana-FIGURE 2. Coordination of JNK1 and JNK2 mediates the apoptosis induction by arsenite in MEFs. WT, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ MEFs were analyzed for the phenotype identification by Western blotting using antibody against JNK1/2 and by PCR using the primers specific for the targeted deletion sequences (A). After WT and knock-out cells were exposed to 20 M sodium arsenite for 24 h, and the chromosome DNA fragmentation was determined using propidium iodide staining and detected by flow cytometry (B). Caspase-3 activation and PARP cleavage were detected by Western blotting after treatment with sodium arsenite for 24 h (C ) and 48 h (D). WT, JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ cells were exposed to nickel chloride for 24 h, and the morphological changes were observed under an inverted microscope and photographed (E ). Caspase-3 activation and PARP cleavage were detected by Western blotting (F ). lyzed. Arsenite exposure led to phosphorylation of both JNK1 and JNK2 in WT cells. In JNK1 Ϫ/Ϫ cells, the obvious phosphorylation of JNK2 was observed; however, in JNK2 Ϫ/Ϫ cells, the phosphorylated JNK1 was comparably weaker, which may be due to the lower expression level of basal JNK1 protein in JNK2 Ϫ/Ϫ MEFs (Fig. 5B). c-Jun activation was indicated by phophorylations of c-Jun at Ser 63 and Ser 73 , the critical residues for c-Jun activation (39). Compared with the full phosphorylation of c-Jun up to 8 h of arsenite exposure in WT cells, jnk1 deficiency resulted in relative low activation of c-Jun, and jnk2 deficiency led to an even more obvious attenuation of c-Jun activation (Fig. 5B), which was consistent with the phosphorylated JNK levels in JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ cells as well as the GADD45␣ mRNA induction in these cell lines (Fig. 5A), suggesting that the cascade activation of JNKs/c-Jun regulated the induction of GADD45␣. To test whether blocking c-Jun activation was responsible for impairing GADD45␣ induction, a plasmid containing a dominant negative mutant of c-Jun (TAM67) was stably transfected into WT MEFs. As expected, TAM67 totally blocked c-Jun phosphorylations (Fig. 5C) and dramatically reduced GADD45␣ induction (Fig. 5D), indicating that the activation of c-Jun was essential for GADD45␣ induction by arsenite. These results provide strong evidence that blocking GADD45␣ induction upon deficiency of jnk1 or jnk2 is, at least in part, due to repressing the activation of c-Jun. This may suggest that the JNK1/2/c-Jun signaling pathway is one of the signaling pathways involved in GADD45␣ induction in cell response to arsenite exposure. However, the JNK1/2/c-Jun pathway is not the only mediator for GADD45␣ induction by arsenite. This notion was supported by our findings that only activating the JNK1/c-Jun pathway by transfection with a plasmid expressing HA-tagged JNKK2-JNK1 fusion protein in WT cells or by transfection of c-Jun expression vector in JNK1 Ϫ/Ϫ cells (Figs. 5, E and G) was not able to elevate GADD45␣ protein expression (Fig. 5, F and H). Moreover, overexpression of exogenous c-Jun in the JNK1 Ϫ/Ϫ cell (Fig. 5G) could not restore the arsenite-induced apoptotic response (Fig. 5, H and I), indicating that some other pathway(s) may also be critical for GADD45␣ induction as well as apoptotic promotion in cell response to arsenite exposure.
JNK1-and JNK2-mediated GADD45␣ Induction Is Essential for Arsenite-associated Apoptosis-To elucidate the importance of GADD45␣ in the arsenite-induced apoptosis, we applied siRNA technology to inhibit endogenous GADD45␣ expression to determine whether blocking GADD45␣ expression could affect cellular apoptotic responses. The two different GADD45␣ siRNA constructs were co-transfected into WT MEFs, and the siRNA efficiency was shown in Fig. 6A. Arseniteinduced apoptotic related cellular morphological alterations, such as cell shrinkage, bubble, and detachment, were partially impaired by knockdown of endogenous GADD45␣ (Fig. 6B). The interference with apoptosis by the GADD45␣ siRNA was further confirmed by the absence of caspase-3 activation and PARP cleavage (Fig. 6C), suggesting that GADD45␣ is critical for arsenite-induced apoptosis. The relevance of GADD45␣ in JNK1-and JNK2-mediated apoptosis was directly tested by stably expressing GADD45␣ in JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ MEFs (Fig.  7A). Overexpression of HA-tagged GADD45␣ partially sensitized the arsenite-induced apoptosis in both JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ MEFs, indicated by the apoptotic morphological alterations and caspase-3 activation as well as PARP cleavage (Fig. 7, B and C). Therefore, our data provide strong evidence that blocking GADD45␣ induction in either JNK1 Ϫ/Ϫ or JNK2 Ϫ/Ϫ MEFs can protect cells from the arsenite-associated apoptosis. These results further imply that JNK1-and JNK2-dependent cell death stimulated by arsenite, for a large part, is mediated by up-regulated GADD45␣ expression.

DISCUSSION
In this study, we have demonstrated that the coordination of JNK1 and JNK2 is required for the arsenite-induced apoptosis in MEFs. Moreover, GADD45␣ is identified as the downstream target of JNK1/2 signaling, and up-regulation of GADD45␣ is critical for the JNK1/2-mediated apoptosis, at least in part, dependent on the activation of c-Jun. To our knowledge, this is the first time to describe the cascade activation of the JNK1/2/ c-Jun/GADD45␣ signaling pathway in the arsenite-induced apoptosis.
Apoptosis is one of the arsenite-caused multiple stress responses in mammalian cells (40 -42). Arsenite exposure causes programmed cell death in both normal healthy cells (7,43) and malignant transformed cells (44 -46), which accounts for its pathogenic (47) as well as therapeutic properties (4). However, the molecular events in the apoptotic responses of cells exposed to arsenite remain unclear. Initially, we apply WT  WT, JNK1 Ϫ/Ϫ , and JNK2 Ϫ/Ϫ MEFs (2ϫ10 5 ) were seeded into each well of 6-well plates and cultured until the cell density reached 70 -80% confluence. The cells were then exposed to 20 M sodium arsenite. Total RNA was extracted 12 h post-arsenite exposure by TRIzol reagent, and cDNAs were synthesized by ThermoScript TM RT-PCR system. The cDNAs were used as a template for the detection of GADD45␣ expression by PCR (A). c-Jun phosphorylation and basal expression, as well as JNKs phosphorylation were detected by Western blotting after the cells were exposed to 20 M sodium arsenite for the indicted time periods (B) or after the WT cells and stable transfectants with TAM67 (WT/TAM7) were treated with 5 or 20 M arsenite for 6 h (C ). For the detection of GADD45␣ expression, WT and WT/TAM67 cells were treated with 10 or 20 M arsenite for 24 h, and the cells were extracted with SDS-sample buffer and detected by Western blotting (D). WT MEFs and stable transfectants expressing HA-JNKK2-JNK1 fusion protein were subjected to Western blotting for detection of HA and phospho-specific and total JNKs, as well as GADD45␣ protein expression, and arsenite treatment was used as positive control for GADD45␣ induction (E and F). Stable transfectants of JNK1 Ϫ/Ϫ cells with c-Jun expression vector was identified with specific antibody against c-Jun (G). The indicated types of cells were treated with sodium arsenite of 20 M for 12 and 24 h. The apoptotic responses were compared by Western blotting and morphological changes (H and I).
MEFs as a model to replicate arsenite-induced apoptosis. Activation of caspase-3, which functions as a final common pathway in the apoptotic machinery (48), and the cleavage of critical cellular substrate, PARP, are observed in MEFs after arsenite exposure (Fig. 1, B and C), which is consistent with previous reports obtained in other laboratories (49). Therefore, we use caspase-3 activation as well as subsequent PARP cleavage as the indices of apoptotic response in the following studies.
JNK signaling pathway activation has been implicated in apoptosis induction in different cell types (7, 8, 50 -52). The most convincing evidence comes from the observation that the JNK1 and JNK2 double knock-out cells are resistant to the apoptosis induced by UV irradiation, indicating that JNK activation is associated with UV-induced apoptosis (52). Our previous report also demonstrates that sodium arsenite also causes the cell apoptosis in the mouse epidermal cell line JB6 P ϩ Cl 41, and more importantly, the expression of dominant negative JNK1 in Cl41 cells abrogates the apoptotic response (8). Using primary cultured rat cerebellar neurons, Namgung and Xia (7) report that JNK3, the brain-specific expressed JNK isoform, is associated with sodium arsenite-induced apoptosis, whereas JNK1 and JNK2 are not stimulated by arsenite in this cell type, although they exhibit high basal activity, indicating that the activation of JNKs is essential for arsenite-induced apoptosis in different types of cells. The present study further reveals that the coordination of JNK1 and JNK2 is required for the apoptotic responses caused by arsenite in MEFs, because knock-out of either jnk1 or jnk2 totally blocks arsenite-induced DNA fragmentation, caspase-3 activation, and PARP cleavage (Fig. 2,  B-D). JNK2 Ϫ/Ϫ cells (cell death 0.70%) seem even more resist-ant to arsenite-induced apoptosis compared with JNK1 Ϫ/Ϫ cells (cell death 7.88%), which is coincident with the basal JNK1/2 expression levels of these two cell lines (Fig. 2, A and B). It should be mentioned that the JNK1 level in JNK2 Ϫ/Ϫ was somewhat lower than that in WT cells, which might be due to the interaction of JNK1 and JNK2 isoforms. This notion has been supported by the findings from other laboratories that knocking down JNK2 expression by its specific siRNA also reduces the JNK1 protein expression (53)(54)(55). The proapoptotic role of the coordination of JNK1 and JNK2 is further confirmed by restoring either JNK1 or JNK2 in their deficient MEFs (Fig. 3B). However, the deficiency of either jnk does not affect the other environmental carcinogen stimuli, such as nickel chloride-induced apoptosis (Fig. 2, E and F). It seems that the JNK-independent apoptotic outcome caused by nickel might be due to the activation and/or alteration of the different proapoptotic and/or antiapoptotic signaling cascades.
GADD45 plays a prominent role in the control of cell growth, cell cycle checkpoint, and nucleotide excision repair (56 -58). Several lines of evidence directly implicate that GADD45␣, one of the three GADD45 isoforms, is important in the apoptotic process (59). The knock-out of the gadd45␣ gene significantly impairs apoptosis and is susceptible to ionizing radiation, UVB, and 3,12-dimethylbenzaanthracene-induced tumors in mouse models (59,60). The present study has also indicated that arsenite-induced apoptosis is mediated by the up-regulation of GADD45␣. Elimination of endogenous GADD45␣ by siRNA partially impairs the apoptosis response to arsenite (Fig. 6, B  and C). Interestingly, the deficiency of either jnk1 or jnk2 obviously blocks GADD45␣ induction at both the protein and   NOVEMBER (Figs. 4A and 5A). The mRNA induction of gadd45␣ in JNK2 Ϫ/Ϫ cells is even more clearly inhibited compared with that in JNK1 Ϫ/Ϫ cells, which is consistent with the basal JNK protein expression levels as well as the sensitivities of these two cells to arsenite-associated apoptosis (Fig. 2, A and B). Reconstitution of JNK1 or JNK2, respectively, restores GADD45␣ induction partially (Fig. 4B) due to the limited expression levels of the exogenous JNK1 or JNK2 in the transfectant masses (Fig. 3A). These results suggest that the impairment of apoptotic response in JNK1 Ϫ/Ϫ and JNK2 Ϫ/Ϫ cells is most likely associated with the levels of GADD45␣ protein expression. To confirm the hypothesis, exogenous GADD45␣ is transfected into the two knock-out cell lines. Expression of the exogenous GADD45␣ partially sensitizes the apoptotic responses in either jnk1-or jnk2-deficient MEFs (Fig. 7, B and C). Our data demonstrate that the proapoptotic function of either JNK1 or JNK2 is mediated by the GADD45␣ induction, which is consistent with the report that the expression of a JNK1 dominant negative mutant substantially abrogates the UV irradiation-associated GADD45 promoter induction (61). Yin et al. (62) also find that in MCF7 breast carcinoma cells, troglitazone-induced GADD45 up-regulation is achieved through JNK signaling, because inhibiting JNK activity by a JNK inhibitor (SP600125) significantly suppresses gadd45 mRNA expression. In the human bronchial epithelial cell line (BEAS-2B), Chen et al. (63) determine that the inhibition of NFB activation by the stable expression of a kinase-mutated form of IB kinase (IKK␤-KM) causes increased and prolonged GADD45 induction by arsenite; however, transfection of IKK␤-KM cells with a dominant negative mutant of SEK1 (SEK1-KM), which partially reduces JNK activation, suppresses the arsenite-induced GADD45 expression. Our present study further reveals that the coordination of JNK1 and JNK2 is required for the GADD45␣ induction, through which JNK1 and JNK2 promote arsenite-associated cell death. To our knowledge, this is the first time it is shown that the coordination of JNK1 and JNK2 affects the arsenite-associated cell fate through the regulation of GADD45␣ expression.

Role of GADD45␣ in JNK-mediated Cell Apoptosis by Arsenite
Interestingly, Harkin et al. (37) report that BRCA1 triggers apoptosis through up-regulation of GADD45 and subsequent JNK activation, in which case GADD45 functions as the upstream activator of JNKs. Conversely, it is also reported that the activation of JNKs is negatively regulated by another GADD45 isoform, GADD45␤, which binds to MKK7 directly and blocks its catalytic activity, thereby suppressing activation of JNK pathways (64,65). It is to be noted here that our laboratory's most recent findings show that in the same MEF cell model with the same dose of arsenite treatment, GADD45␣ in turn up-regulates JNK pathway activation through activating MKK4, 3 another mitogen-activated protein kinase kinase that is reported to be required for JNK activation. Therefore, it is likely that GADD45␣ and the JNK pathway form a feedback loop, and the activation of one component can positively affect the other. Combined with the other observations of the negative effect of GADD45␤ on JNK pathway activation, it seems that different GADD45 isoforms possess different or sometimes contradictory functions on JNK signaling pathway activation.
Since c-Jun is a well documented downstream target of JNK signaling, we chose to examine whether the coordination of JNK1 and JNK2 regulates GADD45␣ expression through c-Jun activation. The promoter region of c-Jun contains the TRE binding domain, so c-Jun regulates its gene transcription by itself (66). Therefore, overexpression of the c-Jun dominant negative mutant (TAM67) impairs not only its phosphorylation but also the basal protein level of c-Jun (Fig. 5C). More importantly, expression of TAM67 obviously impairs GADD45␣ induction in WT MEFs (Fig. 5D), indicating that the activation of c-Jun is required for GADD45␣ induction in response to arsenite treatment. Similarly, the activation of c-Jun induced by arsenite is attenuated in JNK1 Ϫ/Ϫ cells and even totally blocked in JNK2 Ϫ/Ϫ cells, which coincides with the basal and activated JNK protein levels of these two cell lines as well as the GADD45␣ mRNA induction levels (Fig. 5, A and B). So JNK1 and JNK2 regulate GADD45␣ expression in a c-Jun-dependent manner. Similar to GADD45␣ siRNA and JNK deficiency, TAM67 also impairs the apoptotic response by arsenite (data not shown). Arsenite-induced apoptosis requires cascade activation of c-Jun/GADD45␣ to a certain level. Before reaching the threshold level of c-Jun/GADD45␣ induction, the apoptotic response still cannot be initiated, like the situations in JNK1 Ϫ/Ϫ cells. Although certain proportions of c-Jun and GADD45␣ were induced by activated JNK2 protein in JNK1 Ϫ/Ϫ MEFs, it is still not enough for full scale apoptosis induction.
In summary, this study suggests that JNK1 and JNK2, in response to arsenite exposure, are coordinated to activate the c-Jun/GADD45␣ pathway, which subsequently mediates the proapoptotic pathway.