Phosphorylation and Inactivation of Myeloid Cell Leukemia 1 by JNK in Response to Oxidative Stress *

From the ‡Laboratory of Cell Signaling, Department of Hard Tissue Engineering, Division of Bio-Matrix, §Homeostasis Medicine and Nephrology, Department of Regulation of Internal Environment and Reproduction, Division of Systemic Organ Regulation, Graduate School, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8549, Japan, and the ¶Department of Pathology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan

Oxidative stress has been implicated in the pathogenesis of several abnormal conditions and diseases including ischemia, cancer, and diabetes mellitus (1)(2)(3). A recent study suggests that stress-activated protein kinases such as JNK 1 and p38 play important roles in triggering apoptosis in response to various cellular stressors including oxidative stress. We have shown that oxidative stress-induced sustained activation of JNK and p38 is required for apoptosis (4). Apoptosis signalregulating kinase 1 (ASK1), a member of the mitogen-activated protein kinase (MAPK) kinase kinase (MAPKKK), specifically mediates the sustained activation of JNK/p38 and apoptosis in response to oxidative stress (5,6). ASK1-dependent apoptosis is mediated by the release of cytochrome c from the mitochondria followed by caspase 9 activation (7). It has also been reported that JNK is required for UV-induced release of cytochrome c and that new gene expression is not required for this process (8). These reports indicate that JNK induces apoptosis in part through the mitochondria-dependent caspase activation. However, the molecular mechanism by which activated JNK induces mitochondrial dysfunction is unclear.
The members of the Bcl-2 family play pivotal roles in cellular decision to undergo apoptosis. Bcl-2 has been reported to be phosphorylated by JNK in response to different stimuli (9 -11). Although the significance of phosphorylation of Bcl-2 is controversial, it was suggested that phosphorylation by JNK within the unstructured loop region of Bcl-2 decreases its anti-apoptotic activity (9,10,12). Anti-apoptotic Bcl-2 family proteins thus may be potential mediators of JNK-induced apoptosis. However, little is known about the relation between JNK and the other anti-apoptotic members of the Bcl-2 family in the context of oxidative stress-induced apoptosis signaling.
The myeloid cell leukemia 1 (Mcl-1) (13), also known as EAT (14), is an anti-apoptotic Bcl-2 family member. Mcl-1 plays an important role in the development of various carcinomas (15)(16)(17). Similar to other Bcl-2 family members, Mcl-1 localizes in the mitochondrion as well as in other intracellular membranes (18) and can associate with other pro-apoptotic family members (19). Mcl-1 differs from Bcl-2 and Bcl-XL in structure (13), in its short half-life (13), in the regulation of its promoter (20 -22), and in its ability to protect cells from a variety of cytotoxic stimuli (23,24). Little is known regarding posttranslational modification and regulation of Mcl-1. In this study, we investigated the potential involvement of phosphorylation regulation of Mcl-1.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HEK293 cells were grown under 5% CO 2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 4.5 g/liter glucose, and 100 units/ml penicillin. Porcine aortic endothelial (PAE) cells were grown under 5% CO 2 in F12 medium (Invitrogen) supplemented with 10% fetal bovine serum, 10 mM HEPES, and 100 units/ml penicillin. Transfection with various constructs in pGEX-Neo was performed using 2 g of plasmid and 8 l of Tfx 50 (Promega). Transfected cells were selected in the presence of 1 mg/ml Geneticin for 2 weeks, and drug-resistant single-cell colonies were chosen and maintained in growth medium containing 0.4 mg/ml Geneticin.
Recombinant adenovirus was constructed as described elsewhere (29,30). MKK4 cDNA was subcloned in pcDNA3 by PCR. Lys-116 was replaced by Arg using a PCR-based site-directed mutagenesis method. Green fluorescent protein-tagged MKK4 mutant cDNA was subcloned into the SwaI site of pAdex1pCAw cassette cosmid. Each cosmid bearing the expression unit and adenovirus DNA-terminal protein complex was cotransfected into the E1 transcomplementing 293 cell clone. The recombinant adenoviruses generated by homologous recombination were isolated, and high titer stocks of recombinant adenoviruses were grown in 293 cells and purified. Nearly 100% infection of PAE cells by recombinant adenoviruses can be achieved at a m.o.i. of 100 as determined by green fluorescent protein fluorescence (data not shown).
Phosphatase Treatment-Cells were lysed in a lysis buffer for phosphatase treatment containing 150 mM NaCl, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1% Nonidet P-40, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, and 1.5% aprotinin. Cellular debris was removed by centrifugation. Lysates were incubated with or without 2 units/l of -protein phosphatase (New England Biolabs) according to the instructions provided by the manufacturer. The reaction was terminated by adding SDS sample buffer and boiling for 3 min.
Metabolic Labeling-Cells were incubated in phosphate-free medium containing 0.1% fetal bovine serum and 10 mM HEPES, pH 7.0, at 37°C for 3 h. [ 32 P]Orthophosphate (Amersham Biosciences) was then added at a final concentration of 1 mCi/ml, and labeling was continued at 37°C for 3 h. The cells were transferred onto ice and washed twice with ice-cold phosphate-buffered saline and then lysed and immunoprecipitated with anti-Myc antibody and analyzed by SDS-PAGE.
Analysis of Cell Viability-Cells were stimulated with 0.5 mM H 2 O 2 containing F12 medium for 3 h. Cell viability was measured by the trypan blue (Sigma) dye exclusion method. Cells were trypsinized, centrifuged, resuspended in phosphate-buffered saline, and counted using a hemocytometer after dilution in trypan blue. Blue cells were considered as dead cells.

RESULTS
To investigate whether Mcl-1 is regulated by phosphorylation in response to oxidative stress, PAE cells were exposed to   1A, top, lane 1). The mobility of both bands was delayed by H 2 O 2 treatment in a time-dependent manner (Fig. 1A, top,  lanes 2-6). The treatment of cell lysates prepared from H 2 O 2stimulated (Fig. 1A, top, lane 7) and unstimulated (data not shown) PAE cells with -protein phosphatase resulted in the acceleration of the mobility of those bands. These findings suggest that endogenous Mcl-1 is partially phosphorylated under non-stressed conditions and that additional phosphorylation occurred after H 2 O 2 treatment.
To identify the kinase responsible for Mcl-1 phosphorylation in response to H 2 O 2 , we examined the activation state of three classes of MAPK, namely, ERK, JNK, and p38, which are all known to be activated by H 2 O 2 (4, 31). The kinetics of activation of JNK correlated with the extent of mobility of Mcl-1 (Fig.  1A), suggesting that JNK might be involved in the H 2 O 2 -in-  (Fig. 1B). The cotransfection of ASK1, a MAPKKK that activates JNK and that p38 MAPK cascades in vivo, strongly enhanced the phosphorylation of Mcl-1 by JNK and p38 (Fig. 1C, top, lanes 1-5). ASK1 itself phosphorylated Mcl-1 very weakly (Fig. 1C, top,  lane 6). These findings suggested that Mcl-1 may serve as a specific substrate for JNK and p38 at least in vitro. We next examined whether Mcl-1 could be phosphorylated by JNK and p38 in mammalian cells. When Mcl-1 was co-transfected with JNK or p38 alone, the phosphorylation of Mcl-1 was undetectable as determined by the band shift analysis (Fig. 1D, lanes 2  and 4). In contrast, Mcl-1 became a shifted doublet by the co-expression of the activated allele of ASK1 (ASK1⌬N) together with JNK or p38 (Fig. 1D, lanes 3 and 5). Phosphatase protein was calculated, and the intensity was shown as fold increase relative to control. C, the absence of gel mobility shift of S121A/T163A Mcl-1 on H 2 O 2 treatment. PAE clones stably expressing WT or S121A/T163A MCL-1 were treated with 0.5 mM H 2 O 2 for the indicated periods of time. Cells were lysed in a lysis buffer for phosphatase treatment. Aliquots of the samples were incubated with or without -protein phosphatase at 30°C for 30 min. Western blots were performed using the indicated specific antibodies. Asterisk indicates nonspecific band. treatment shifted the doublet down to the basal status (Fig. 1D,  lanes 6 -10), indicating that activated JNK and p38 could phosphorylate Mcl-1 in vivo.
A sequence comparison of human and mouse Mcl-1 revealed that Mcl-1 possesses two conserved sites, Ser-121 and Thr-163, in humans that conforms to the consensus motif for the substrate of JNK and p38 (Fig. 2A). These sites are located in the PEST (proline, glutamic acid, serine, and threonine) domain of Mcl-1 ( Fig. 2A) and correspond to the so-called unstructural loop region in Bcl-2, which regulates the anti-apoptotic function of Bcl-2 (32). To examine which sites are phosphorylated by JNK or p38, we constructed three alanine substitution mutants of Mcl-1 (S121A, T163A, and S121A/T163A). The Myctagged wild-type (WT) or alanine-substituted mutant of Mcl-1 was co-transfected with JNK or p38 plus ASK1⌬N. Cells were metabolically labeled with [ 32 P]orthophosphate and analyzed by autoradiography after immunoprecipitation using anti-Myc antibody. WT and single alanine substitution mutants (S121A and T163A) of Mcl-1 were clearly phosphorylated by the coexpression of activated JNK and p38 (Fig. 2B, top, lanes 5-7  and 9 -11). In contrast, little phosphorylation was detected in the double-alanine mutant (S121A/T163A) of Mcl-1 (Fig. 2B,  top, lanes 8 and 12). These findings suggested that when overexpressed, activated JNK and p38 can phosphorylate both Ser-121 and Thr-163 of Mcl-1 and that these two amino acids are the major phosphorylation sites of Mcl-1 in vivo.
To investigate the involvement of Ser-121 and Thr-163 in oxidative stress-induced phosphorylation of Mcl-1 as observed in Fig. 1A, we generated PAE cell clones stably expressing Myc-tagged WT and S121A/T163A mutant of Mcl-1. When these cells were treated with H 2 O 2 , the activations of endogenous JNK and p38 were clearly observed in both cells in a time-dependent manner (Fig. 2C, middle and bottom panels). In parallel with JNK activation, gel mobility of Mcl-1 was retarded in WT but not in S121A/T163A mutant-expressing cells (Fig. 2C, top), and the retardation was canceled by treatment with -protein phosphatase (Fig. 2C, lane 7). We have examined three independently selected clones of WT and mutant Mcl-1 and obtained essentially the same results in independent clones (data not shown). These results suggest that both Ser-121 and Thr-163 of Mcl-1 are phosphorylated in response to oxidative stress.
The overexpression of either activated JNK or p38 phosphorylated Mcl-1 in vivo (Figs. 1D and 2B), and both kinases were activated by H 2 O 2 treatment (Figs. 1A and 2C). However, time course analysis indicated that the activation of JNK coincided with Mcl-1 phosphorylation following H 2 O 2 stimulation much better than that of p38 (Figs. 1A and 2C). To examine which signaling pathway is physiologically required for Mcl-1 phosphorylation in response to H 2 O 2 , we used the p38 inhibitor SB203580 and a recombinant adenovirus encoding dominant negative MKK4. Although p38 was specifically inactivated by SB203580 (data not shown), the treatment of PAE cells with SB203580 before H 2 O 2 stimulation did not alter H 2 O 2 -induced Mcl-1 mobility shift (Fig. 3A). In contrast, the expression of the dominant negative MKK4 significantly reduced the gel mobility shift of Mcl-1 upon H 2 O 2 treatment (Fig. 3B). JNK but not p38 activation was specifically reduced by adenovirus encoding dominant negative MKK4. Taken together, Mcl-1 appears to be phosphorylated mainly via the JNK pathway in response to oxidative stress.
Finally, we assessed the functional importance of Mcl-1 phosphorylation in oxidative stress-induced apoptosis. To this end, the susceptibility to H 2 O 2 -induced apoptosis was examined in PAE clones stably expressing WT and those expressing S121A/T163A Mcl-1. PAE clones were treated with 0.5 mM H 2 O 2 for 3 h, and cell death was determined by the trypan blue exclusion assay (Fig. 4). WT Mcl-1 conferred only minimal resistance compared with the vector control. However, S121A/ T163A Mcl-1 showed substantially stronger anti-apoptotic activity than WT Mcl-1 following H 2 O 2 treatment. We have examined three independently selected clones of WT and mutant Mcl-1 and obtained essentially the same results in independent clones (data not shown). These data indicated that eliminating phosphorylation sites increased anti-apoptotic activity of Mcl-1. In other words, Mcl-1 appears to be negatively regulated through phosphorylation of Ser-121 and Thr-163 by JNK following H 2 O 2 stimulation. DISCUSSION In this study, we demonstrated that Mcl-1 was phosphorylated at Ser-121 and Thr-163 through the JNK pathway and inactivated following H 2 O 2 treatment. We also demonstrated that both JNK and p38 could phosphorylate Mcl-1 in vitro, whereas ERK induced little phosphorylation. A recent study also suggested using an ERK inhibitor that ERK was involved in 12-O-tetradecanoylphorbol-13-acetate-induced Mcl-1 phosphorylation (33). Further investigations will be needed to determine whether the ERK pathway is also involved in Mcl-1 phosphorylation depending on stimuli.
We identified two phosphorylation sites, which regulate the anti-apoptotic function of Mcl-1 in response to H 2 O 2 . Although human Mcl-1 possesses five potential phosphorylation sites that can be phosphorylated by JNK and p38, we could not detect any phosphorylation in the S121A/T163A mutant, sug- gesting that Ser-121 and Thr-163 are the main sites to undergo H 2 O 2 -induced phosphorylation. Bcl-2 has been shown to be phosphorylated by JNK in at least four serine/threonine sites. Especially, Ser-70 and Ser-87 play an important role in negatively regulating the anti-apoptotic function of Bcl-2 (9,11). Recent work suggested that the number of phosphorylated sites of Bcl-2 appeared to depend on the intensity of kinase activation. We detected another shift of gel mobility of the S121A/T163A mutant of Mcl-1 when it was co-expressed with JNK and ASK1⌬N in 293 cells (data not shown). It remains to be determined whether other phosphorylation sites contribute to the regulation of the anti-apoptotic activity of Mcl-1.
The mechanisms by which phosphorylation of Bcl-2 regulates anti-apoptotic function are poorly understood. Several studies have shown that phosphorylated Bcl-2 does not heterodimerize with Bax, and thus, apoptosis is promoted by an increase in the amount of free Bax (34,35). We could not detect any changes in the interaction of Bax and Mcl-1 before or after the phosphorylation of Mcl-1 (data not shown). However, in the phosphorylation sites of Mcl-1 located in the PEST motif, there was no difference in the half-life of WT and S121A/T163A mutant of Mcl-1 after H 2 O 2 stimulation (data not shown). Further studies will be needed to elucidate the mechanism of phosphorylation-mediated inactivation of Mcl-1.
The JNK signaling pathway is essential for exocytotoxic stress-induced apoptosis in neurons and UV-induced apoptosis in mouse embryonic fibroblast (8,36). It seems that activated JNK acts on mitochondria and induces apoptosis through the release of cytochrome c (7,8). The mechanism of cytochrome c release by JNK is not known at all. Although the Bcl-2 family is a potential target of JNK that regulates cytochrome c, several discrepancies have been pointed out. For example, Bcl-2 phosphorylation has been suggested to increase rather than decrease anti-apoptotic function (37). Stimuli that cause JNKinduced apoptosis such as UV do not necessarily cause Bcl-2 phosphorylation (8).
In our study, Mcl-1 was phosphorylated by JNK and its anti-apoptotic function decreased. Phosphorylation and inactivation of Mcl-1 thus may be one of the mechanisms by which JNK induces apoptosis in response to oxidative stress.