Ser/Thr protein phosphatase 5 inactivates hypoxia-induced activation of an apoptosis signal-regulating kinase 1/MKK-4/JNK signaling cascade.

Mitogen-activated protein kinase (MAPK) signaling cascades are multifunctional signaling networks that influence cell growth, differentiation, apoptosis, and cellular responses to stress. Since the activation/propagation of MAPK signaling requires the sequential phosphorylation of many downstream proteins, the phosphatases that dephosphorylate MAPKs represent critical elements in the control of MAPK-signaling networks. Here we show that hypoxia induces a transient increase in the activity of apoptosis signal-regulating kinase 1 (ASK-1), a MAPKKK that responds to oxidative stress by triggering cascades leading to the phosphorylation/activation of c-Jun N-terminal kinases (JNK) and p38-MAPK. Hypoxia-induced ASK-1/MKK-4/JNK signaling is suppressed by serine/threonine protein phosphatase type 5 (PP5), which acts to turn off ASK-1/MKK-4/JNK signaling via two mechanisms. First, in a rapid response hypoxia facilitates the association of endogenous PP5 with ASK-1. PP5 binds to the C-terminal domain of ASK-1, and studies with siRNA targeting PP5 indicate that PP5 acts to suppress the phosphorylation of MKK4 (Thr-261), JNK (Thr-183/Tyr-185), and c-Jun (Ser-63) without affecting the activating phosphorylation of p38 MAPK (Thr-180/Tyr-182), p44/p42-MAPK/ERK1/2 (Thr-202/Tyr-204), or c-Jun protein levels. If hypoxia is prolonged, the expression of PP5 is increased due to the activation of a transcriptional activator, which was identified as hypoxia-inducible factor-1. Together, these studies indicate that PP5 plays an important role in the survival of cells in a low oxygen environment by suppressing a hypoxia-induced ASK-1/MKK4/JNK signaling cascade that promotes an apoptotic response.

In a MCF-7 mouse xenograph tumor model, the overexpression of PP5 is associated with a marked increase in the rate of tumor growth, with a statistical difference in tumor size noted in Ͻ14 days (19). This suggested that the overexpression of PP5 aids xenograph tumor growth. Since the vascular system in mouse xenograph tumors is usually aberrant and provides poor blood flow to the developing tumor, regions within the tumors become hypoxic. Here, we show that hypoxia induces a transient increase in ASK-1/MKK4/JNK activity and that PP5 acts to turn off ASK-1/MKK-4/JNK signaling. Together, the studies presented suggest that the growth advantage of MCF-7 xenograph tumors in which PP5 expression is constitutively elevated (19) may be derived from PP5 acting as a negative regulator of an ASK-1/MKK4/JNK signaling cascade that facilitates an apoptotic response in hypoxic cells.
Western Blotting-Cells grown in 60-mm plates were washed twice with ice-cold phosphate-buffered saline and then lysed by scraping in 250 l of lysis buffer (120 mM Tris-HCl, pH 7.4, 20% glycerol, 4% SDS, 10 mM ␤-glycerol phosphate, and 1 l/ml protease inhibitor mixture (Sigma)). Lysates were clarified by centrifugation at 13,000 ϫ g for 5 min, and an aliquot of the supernatant was removed for protein determination. Protein concentrations were determined using a Bio-Rad D c protein assay, using bovine serum albumin for standards. The remaining supernatant was added to an equal volume of 2ϫ sample buffer (120 mM Tris-HCl, pH 7.4, 200 mM dithiothreitol, 20% glycerol, 4% SDS, and 0.02% bromphenol blue). Typically, 25-50 g of protein was then separated by electrophoresis on 10% SDS-polyacrylamide gels. Proteins were then electrophoretically transferred to Immobilon-P (Millipore Corp.) and placed in blocking buffer (Tris-HCl, pH 7.4, 150 mM NaCl, and 5% nonfat milk or 5% bovine serum albumin) for 1 h. Immunoblotting was performed with the indicated primary antibody using either goat anti-mouse (Pharmingen, San Diego, CA) or goat anti-rabbit secondary (Promega, Madison, WI) antibodies. Antibody association was detected employing ECL (Amersham Biosciences) or Super Signal West Dura (Pierce) Western blotting detection reagents following the protocols of the manufacturer. Quantification of the signal was achieved using a Fuji-LAS-1000 imaging system.
Electrophoretic Mobility Shift Assay-Nuclear extracts for electrophoretic mobility shift assay assays were prepared essentially as described previously (21). Briefly, Hep3B or A549 cells were cultured in 60-mm dishes (three dishes per group) and treated with the indicated agents. At the times indicated, the cells were washed with ice-cold phosphate-buffered saline, scraped in 0.5 ml of ice-cold phosphatebuffered saline, and collected by centrifugation at 4°C for 10 min at 2000 ϫ g. The pellet was washed with five packed cell volumes of buffer A (10 mM Tris-HCl, pH 7.5, 1.5 mM MgCl 2 , 10 mM KCl, 2 mM DTT, 400 M phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 2 g/ml pepstatin, and 1 mM activated Na 3 VO 4 ) and resuspended in 4 packed cell volumes of buffer A. After sitting on ice for 10 min, the nuclei were collected by centrifugation at 10,000 ϫ g for 10 min at 4°C. The nuclear fraction was washed twice, and the nuclei were resuspended in 3 packed nuclear volumes of buffer C (0.42 M KCl, 20 mM Tris-HCl, 20% (v/v) glycerol, 1.5 mM MgCl 2 , pH 7.5). The suspension was rotated for 30 min at 4°C and then subjected to centrifugation at 15,000 ϫ g for 30 min at 4°C. The supernatant was aliquoted and stored at Ϫ80°C. Doublestranded DNA probes for electrophoretic mobility shift assay were generated by radiolabeling synthetic oligonucleotides encoding the HRE contained in the promoter of PP5 or erythropoietin (EPO) (sequences provided in Fig. 2) using a DECA-Prime® labeling kit (Ambion). Binding reactions were performed under stringent conditions by incubating 5.0 g of nuclear extract with ϳ1.0 ng of 32 P-labeled probe (10 4 cpm) in buffer D (25 mM Tris-HCl, 100 mM KCl, 0.2 mM EDTA, 20% glycerol, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1.2 mM sodium vanadate, pH 7.6; final volume 19 l) for 5 min at room temperature and then for 15 min on ice. For competition experiments, unlabeled probes (in 20-fold excess) were added to the mixture. For shift analysis, after the addition of probe, 0.25 g of HIF-1␣ antibody was added to the mixture, and binding was conducted on ice for 20 min. Samples were separated by electrophoresis on a 4% native polyacrylamide gel and visualized by autoradiography. The sequence of the sense strand for each probe is listed: RH309, CCTGCGCAGGCGCGT-GAAGGGC, PP5 wild type; RH325, GCCCTACGTGCTGTCTCA, EPO HIF-1␣ response element; RH335, CCAGCGCAGGCGAAAGAAGGGC, RH309 with the consensus HIF-1 binding site altered in three positions; RH337, CGGCGATGGCCTGCGAGCGGAC, PP5 scramble control (random order of nucleotides contained in RH309).
Transient Transfections and Luciferase Assay-Cell transfections and Luciferase assays were conducted as described previously (20). Briefly, A549 cells were plated at a density of 2 ϫ 10 5 cells/60-mm tissue culture dish and incubated overnight in DMEM supplemented with 10% fetal calf serum. After ϳ24 h, the cells were transfected with 1.0 g of plasmid DNA using LipofectAMINE (Invitrogen) as described previously (20). After 4 h, the transfection medium was replaced with fresh media, and the cells were incubated for another 20 h prior to treatment with the agents indicated. Hypoxic conditions were achieved by incubating cell cultures in an anaerobic pouch (GasPack Pouch; Becton Dickinson). At the times indicated, luciferase activity was measured using the enhanced luciferase assay kit (PharMingen) with a Monolight 2010 luminometer. The ratio of luciferase activity to total protein (as measured with a Bio-Rad Dc assay) was calculated. Measurement of ASK-1 Activity-ASK-1 activity was measured using myelin basic protein (MBP; Sigma) as a substrate as described previously (22). Briefly, cell lysate, 500 g of protein, was mixed with 5 g of anti-ASK1 antibody (Santa Cruz Biotechnology) and 20 l of protein A/G plus agarose (Santa Cruz Biotechnology) at 4°C with gentile rocking for ϳ16 h. ASK-1 was collected by centrifugation (3000 ϫ g for 1 min). The pellet was washed three times with IP buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EGTA, 1% Nonidet P-40, 1 mM DTT, 1 mM activated sodium orthovanadate, 2 l/ml protease inhibitor mixture, 5 M cantharidin, 5 mM NaF, and 10 mM ␤-glycerophosphate) and then one time with kinase buffer (25 mM Tris-HCl, pH 7.5, 5 mM ␤-glycerol phosphate 1 mM sodium orthovanadate, 2 mM DTT, 10 mM MgCl 2 ). Myelin basic protein (3 g) was then added, and the kinase reaction was initiated by the addition of ATP (50 M) containing 1-5 Ci of [␥-32 P]ATP. The reaction was allowed to proceed for 20 min at 30°C. The reaction was then terminated by the addition of 30 l of 2ϫ SDS-PAGE sample buffer and heating. Unincorporated 32 P was separated by electrophoresis (SDS-PAGE) using 12% gels. Radioactivity incorporated into MBP was quantified using a Fuji FLA-5000 phosphorimaging system. ASK-1 activity is reported as the ratio of the amount of phosphate incorporated into MBP divided by the amount of ASK-1 protein in the assay.
Microcystin-Sepharose Pull-down Assays-Microcystin pull-down assays were conducted essentially as described previously (23). Briefly, cells were seeded at a density of 300,000 cells/dish in 60-mm dishes and allowed to grow for ϳ48 h. The dishes were washed twice with phosphate-buffered saline and scraped from the dish in 400 l of extraction buffer (50 mM Tris-HCl, pH 7.0, 2 mM EDTA, 2 mM EGTA, 2 mM MgCl 2 , 1 mM DTT, 6.3 g/ml aprotinin, 2 g/ml leupeptin, 5 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml pepstatin A). The cells were then lysed on ice with a Kontes tissue grinder (30 strokes in a 2-ml tube). The lysates were subjected to centrifugation (15,000 ϫ g for 10 min), the supernatant was collected, and protein concentration in the crude lysate was determined using a Bio-Rad Dc protein assay. Aliquots of lysate containing 1 mg of protein were then mixed with 30 l of 50% Sepharose 4B slurry and mixed at 4°C for 1 h to remove nonspecific binding. The Sepharose 4B was collected by centrifugation, and the supernatant was mixed with 30 l of microcystin-Sepharose (equilibrated with extraction buffer) by circular end-over-end rotation at 4°C for ϳ16 h. The microcystin-Sepharose was collected by centrifugation (10,000 ϫ g for 10 min), and the pellet was washed six times with 600 l of washing buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.0, 0.1 mM EGTA, 1 mM DTT, 6.3 g/ml aprotinin, 4 g/ml leupeptin, 10 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 4 g/ml pepstatin A). Proteins were eluted from the pellets with 2ϫ SDS sample buffer and separated by SDS-PAGE using 10% gels.
Suppression of PP5 Expression with Double-stranded RNA (siRNA)-siRNA-mediated suppression of PP5 was achieved as described previously (24). Briefly, polyribonucleotides (21 bases in length) were synthesized in both the sense and antisense orientation, with 19 bases homologous to the target mRNA and two 2Ј-deoxythymidine nucleotides at 3Ј-ends that produce 3Ј-overhangs upon annealing. The oligonucleotides were annealed to produce siRNA by placing the complementary ribonucleotides (20 M) in RNase-free annealing buffer (100 mM potassium acetate, 30 mM HEPES-KOH at pH 7.4, 2 mM magnesium acetate) for 1 min at 90°C and then at 37°C for 1 h. The sense strand of the siRNA that effectively suppressed PP5 (PP5TG2) is 5Ј-GAG GGU GAG GUG AAG GCC ATT-3Ј. For a control, the order of the bases was scrambled. The sense strand for the mismatched control (PP5 TG2MM) is 5Ј-GGU AGG ACG AGA GGU AGG CTT-3Ј. For treatment, A549 cells were plated at 200,000 -300,000 cells/60-mm dish. Twentyfour hours later, the annealed double-stranded RNA oligonucleotides were transfected using Lipofectin® (Invitrogen), employing the same methods used for the addition of antisense oligonucleotides (6,20,21). Briefly, the day after plating, the cells were washed with DMEM and incubated with 1 ml of DMEM containing 15 l/ml Lipofectin and 2.5 g of siRNA that was mixed in a Falcon tube via shaking by hand for 1 min just prior to the addition. After incubation at 37°C for 4 h, the DMEM was removed and replaced with fresh media supplemented as described above (see "Cell Culture").

RESULTS
Hypoxia Induces the Expression of PP5-To characterize the relationship between PP5 and hypoxia, time course studies were conducted in A549 and Hep3B cells measuring PP5 protein levels by Western analysis. As seen in Fig. 1, when placed in hypoxic conditions, there is a temporal increase in PP5 protein levels in both cell lines, with a statistical difference from controls evident by 5 h. Northern analysis revealed a 1.89 Ϯ 0.2-fold (mean Ϯ S.D., n ϭ 4) increase in PP5 mRNA levels 8 h after the onset of hypoxia, and reverse transcription-PCR revealed a similar 2.04 Ϯ 0.12-fold (p Ͻ 0.01) increase in PP5 mRNA levels in hypoxic cells, suggesting that the increased levels of protein result from an increase in PP5 transcription.
Characterization of the Human PP5 Promoter-Analysis of a 1.9-megabase region of human chromosome 19 q13.3 identified the PP5 gene (PPP5) and 43.5 kilobases 5Ј of PP5 exon 1. 5Ј-rapid amplification of cDNA ends analysis of human cDNA identified ϳ758 base pairs 5Ј of the coding region, and a deletion series of PP5-pLucLink V.2 luciferase reporter plasmids was constructed and tested in transient transfection assays (20). The region between Ϫ184 and Ϫ110 was identified as an important regulatory domain in the PP5 promoter (i.e. PP5-Luc-184 demonstrated comparable luciferase activity to the "full-length" promoter, whereas PP5-Luc-110 had levels similar to the control Luc link vector (20)). A search for sequences with possible biological significance within this region revealed a single estrogen response element (ERE), which has been shown to produce a similar 1.5-2.0-fold induction of PP5 expression in estrogen-responsive tumor cells (20). The region also contains a consensus HIF-1 response element (Fig. 2) (25)(26)(27), and both the HRE and the ERE are contained in a similar orientation in the 5Ј-flanking region of exon 1 in the mouse PP5 gene.
To determine if the HRE contained in the PP5 promoter is functional, we first performed electrophoretic mobility shift analysis using nuclear extracts from hypoxic cells (in which HIF-1 is transcriptionally active) and from control cells (cultured with 21% oxygen and having minimal HIF-1 activity). In these studies, the binding of a doubled-stranded DNA probe encoding the putative HRE from the PP5 promoter (Ϫ143 to Ϫ124) was compared with the binding of a well characterized double-stranded oligonucleotide containing the HRE present in the EPO gene (W18 as described in Ref. 28). As seen in Fig. 3A, when added to nuclear extracts produced from hypoxic cells, the mobility shift obtained with the PP5 probe (lanes 2 and 7) was similar to the shift produced with the EPO probe (lane 11). In contrast, when probes were added to nuclear extracts produced from control cell cultures, the shift was greatly reduced or not detected (lanes 1, 6, and 10). To test the specificity of binding, competition studies were performed assessing 1) the ability of unlabelled EPO probe to compete with the binding of the [ 32 P]PP5 probe, 2) the ability of unlabeled PP5 probe to compete with the binding of the [ 32 P]EPO probe, 3) the ability of an unlabeled scrambled probe (same oligonucleotide composition as the PP5 probe constructed in a random order) to compete with the binding of the [ 32 P]PP5 probe, 4) the ability of a mutated PP5 probe (having three altered bases in the core of the consensus HRE) to compete with the binding of the [ 32 P]PP5 probe, and 5) the effect produced by the addition of an HIF-1␣ specific antibody to the binding assay. These competition studies revealed that the EPO probe was an effective competitor with the PP5 probe and vice versa (compare lane 2 Binding was also sequence-specific, for unlike the EPO probe, the scrambled PP5 probe was not an efficient competitor with the PP5 probe (compare lanes 2, 3, and 5), and mutation of three bases in the core of the PP5 HRE (lane 4) diminished competition. The addition of an antibody that specifically recognizes HIF-1␣ also inhibited the mobility shift produced by both the EPO and the PP5 probe (lanes 9 and 13). Together, these findings suggest that HIF-1␣ obtained from the nuclear extract of hypoxic cells can recognize and bind to the HRE (Ϫ133 to Ϫ130) contained in the 5Ј-flanking region of PP5.
To determine if binding affects transcription, a deletion series of PP5-luciferase reporter plasmids was constructed and tested in A549 cells and Hep3B cells using transient transfection studies. These studies revealed that the deletion of the region between Ϫ643 and Ϫ184 had little effect on transcription (Fig. 3B). However, the deletion of 55 additional bases (PP5-Luc-129), which deleted the HRE, resulted in a marked decrease in transcriptional activity. In contrast, deletion of the ERE revealed no apparent effect on the basal expression of PP5 (data not shown). Next, we tested the effect produced by mutating the PP5-Luc-184 reporter plasmid, altering three bases in the core of the HRE (Ϫ133 to Ϫ130). Studies with the mutated PP5 reporter plasmid (PP5-Luc-184m) revealed that mutation of the HRE markedly suppressed the transcriptional activity of the luciferase-reporter plasmids. This finding suggests that the transcription of PP5 is responsive to HIF-1.
Enhanced PP5 Luciferase Activity Correlates with the Activation of HIF-1-In addition to hypoxia, HIF-1 transcriptional activity can be activated by treatment with chemicals suspected of mimicking some aspect of hypoxia (i.e. cobalt chloride (CoCl 2 ) and DFO). HIF-1-dependent gene transcription can also be inhibited by treatment with agents (i.e. 2,3,7,8-tetrachlorodibenzo-p-dioxin) that activate aryl hydrocarbon receptors, which recruit HIF-1␤/aryl hydrocarbon receptor nuclear translocator protein into a complex with the aryl hydrocarbon receptor and prevents the HIF-1␤/aryl hydrocarbon receptor nuclear translocator protein subunit from forming an active dimer with HIF-1␣ (29). To further characterize the relationship between HIF-1 and PP5, we conducted transient transfection studies in A549 cells using both the PP5-Luc-184 and PP5-Luc-184m reporter plasmids in combination with agents that alter HIF-1 transcriptional activity. Transfection studies conducted 24 h after treatment with CoCl 2 or DFO or 8 h after the onset of hypoxia revealed PP5-Luc-184 luciferase activity that was 170, 230, and 284% of control level, respectively. In contrast, studies with PP5-Luc-184m revealed little change in luciferase activity following hypoxia or treatment with CoCl 2 or DFO (Fig. 3C). Mutation of the HRE also produced a decrease in the basal expression of PP5-luciferase activity, as did treatment with dioxin. Cumulatively, the studies discussed above indicate that transcription of PP5 is responsive to hypoxia and can be mediated by HIF-1.
Hypoxia Induces the Activation of ASK-1-In mammals, hypoxia is associated with the activation of several stress-induced protein kinases, including JNK, p38, and ERK (30). In addition, in a yeast two-hybrid screen PP5 was identified as a binding partner for ASK-1, a MAPKKK that is activated by oxidative stress and has been reported to lead to the activation of both p38 and JNK (7). Therefore, it was suggested that PP5 may act as a negative regulator of ASK1 signaling (7). Still, to date, hypoxia had not been reported to activate ASK-1. Furthermore, the roles of ASK-1 activation in cellular responses to oxidative stress are not clear, for the activation of JNK is generally associated with the promotion of stress-induced apoptosis, whereas the activation of p38 is often associated with an increased chance of cell survival.
To determine if hypoxia influences ASK-1 signaling, we tested the effect of hypoxia on A549 cells, measuring 1) changes in ASK-1 kinase activity directly using [ 32 P]ATP with MBP as a substrate and 2) changes in the phosphorylation of MKK-4 (an immediate downstream target of ASK-1) in cell lysates using Western analysis with antibodies that recognize MKK4 phosphorylated at Thr-261. As seen in Fig. 4A, after 1 h in a hypoxic environment, ASK-1 activity is induced, suggesting that ASK-1 is indeed activated by hypoxia in A549 cells. Fur- thermore, time course studies measuring MKK phosphorylation at Thr-261 indicate that hypoxia is associated with a rapid increase in MKK4 phosphorylation that peaks in ϳ1 h, returns to basal levels by 3-5 h, and remains at basal levels for at least 24 h (Fig. 4B).
Suppression of PP5 Expression with Double-stranded RNA Prevents the Inactivation of Hypoxia-activated MKK4 -To date, PP5 has been shown to associate with Ͼ15 different proteins via immunoprecipitation or yeast two-hybrid-based assays. However, determining the cellular roles of PP5 has proven difficult, because 1) there are no selective small molecule inhibitors of PP5, 2) in cell homogenates PP5 exists predominately in an inactive state, 3) the physiological substrates for PP5 are not known, and 4) when activated by proteasemediated cleavage of the N-terminal autoinhibitory domain or by the addition of polyunsaturated lipids (2,31,32), PP5 demonstrates little substrate specificity in vitro. This makes it impossible to distinguish the activity of PP5 in vivo or in crude cell extracts from that of PP2A and PP1, which are both also expressed at relatively high levels in all human cells examined to date. Therefore, to address the functional consequences of PP5 expression, we developed antisense (20) and, more recently, small double-stranded oligonucleotides of RNA (siRNA) capable of markedly suppressing PP5 mRNA and protein levels in cultured human cells (24). As reported previously (24), siRNAs targeting PP5 (PP5TG2) potently suppress the expression of PP5 protein in A549 cells (Fig. 5A). In contrast, the mismatched controls (which are composed of the same nucleotides with the order of the sense strand scrambled and the antisense strand made to match the scrambled sense strand) had no effect on PP5 protein or mRNA levels. The suppression of PP5 with PP5TG2 is specific, for neither has an effect on the expression of glyceraldehyde-3-phosphate dehydrogenase, actin, or PP2A, which shares 37% identity with PP5 (24).
To determine if PP5 affects hypoxia-induced MKK-4 activation, we suppressed the expression of PP5 with siRNA and then measured changes in the activity of MKK4 by measuring JNK phosphorylation at the activating Thr-183/Tyr-185. As seen in Fig. 5B, the suppression of PP5 did not augment hypoxiainduced MKK4 activity, for after 1 h in hypoxic conditions, JNK phosphorylation was elevated to a similar extent in cells treated with siRNA targeting PP5 (PP5TG2), cells treated with mismatched control siRNA, and untreated control cells (com- pare lanes 2-4). In contrast, the suppression of PP5 had a marked effect on the inactivation of MKK4, for in controls, JNK phosphorylation returned to basal levels in ϳ3 h (lanes 5 and 7), whereas JNK remained highly phosphorylated in cells where PP5 expression was suppressed by treatment with PP5TG2 (lane 6). These studies suggest that in hypoxic cells, PP5 acts to "turn off" hypoxia-induced activation of an ASK1/ To further characterize the relationship of PP5 and hypoxia, we next tested the effect of PP5 suppression on the phosphorylation/activation of several additional MAPKKs, MAPKs, and transcription factors (22,33) using Western analysis with phosphorylation site-specific antibodies. As seen in Fig. 5C, the suppression of PP5 is associated with an increase in the phosphorylation of MKK4 (Thr-261), JNK-(Thr-183/Tyr-185), and c-Jun (Ser-63), yet it had no apparent affect on the phosphorylation of p38 MAPK (Thr-180/Tyr-182), p44/p42MAPK-ERK1/2 (Thr-202/Tyr-204), or total c-Jun protein levels. These findings are consistent with the concept that PP5 acts as a negative regulator of an ASK-1 signaling cascade that leads to the activation of JNK and the phosphorylation of c-Jun at Ser-63, with PP5 acting upstream of JNK and MKK4. However, the relationship between HIF-1-induced PP5 expression and the inactivation of MKK-4/JNK signaling was not clear, because the inactivation of MKK-4 (observed in Ͻ3 h) occurs prior to HIF-1-induced PP5 expression, which is not statistically significant until ϳ5 h after the onset of hypoxia.
Hypoxia Induces the Association of PP5 with ASK-1-To further explore the relationship between PP5 and ASK-1, coimmunoprecipitation studies were conducted using polyclonal anti-ASK-1 antibodies to immunoprecipitate ASK-1 from cell lysates followed by Western analysis to measure PP5-protein levels in the immunoprecipitates and vice versa. When antibod-ies to ASK-1 were used to immunoprecipitate proteins from A549 cell extracts, PP5 was identified by Western blotting of the precipitates (Fig. 6A). Similarly, when antibodies to PP5 were used to immunoprecipitate proteins from crude cell extracts, ASK-1 was identified by Western analysis (Fig. 6B). Immunoprecipitation studies indicate that the association between PP5 and ASK-1 is increased by hypoxia (Fig. 6D), and the suppression of PP5 expression with PP5TG2 results in a slight increase in ASK-1 activity (Fig. 6E). Together, these findings are consistent with PP5 acting as a negative regulator of hypoxia-induced activation of ASK-1. Nonetheless, since the activation of ASK-1 in response to hypoxia is greater than the response induced by the suppression of PP5 expression alone, PP5 does not appear to act as an inhibitor of ASK-activation. Rather, PP5 may function to inactivate ASK-1, preventing the either "accidental" or sustained activation of ASK-1.
The Carboxyl Terminus of ASK-1 Contains Residues Needed for Interaction with PP5-In response to oxidative stress, the activation of ASK-1 is associated with the dissociation of thioredoxin (Trx), a redox-sensitive protein that binds to and sequesters ASK-1 in an inactive state. During oxidative stress, a redox-induced conformational change in Trx induces dissociation of the Trx⅐ASK-1 complex, leading to the autoactivation of ASK-1, possibly by allowing ASK-1 dimerization and the autophosphorylation at Thr-845 (7). To further characterize the interaction of PP5 with ASK-1, expression vectors encoding HA epitope-tagged full-length ASK-1 (ASK-1-HA), a HA-tagged N- IgG was used to detect nonspecific binding (IP; IgG). B, Western blot demonstrating the presence of both ASK1 and PP5 in immunoprecipitations conducted with antibodies that recognize PP5 (IP; PP5). C, Western blot demonstrating the presence of ASK1 and PP5 in microcystin-Sepharose pull-down experiments. A549 cells were transfected with siRNA targeting PP5 (PP5TG2), mismatch control siRNA (PP5TG2MM), or Lipo-fectAMINE alone (control). After 48 h, the cells were subjected to microcystin-Sepharose pull-downs and processed for Western analysis as described under "Experimental Procedures." D, hypoxia increases the association of PP5 and ASK-1. HEK293 cell cultures were then placed in a hypoxic chamber. At the times indicated, the cells were lysed, and ASK-1 was immunoprecipitated from the lysates with an antibody recognizing ASK-1. The amount of PP5 contained in the immunoprecipitate was then determined by Western analysis. To ensure equal loading, the membranes were stripped and reprobed with an anti-ASK1 antibody. The data are plotted as the percentage of the PP5 found associated with ASK-1 in cells cultured with normal levels of oxygen (mean Ϯ S.D.; n ϭ 9). A statistical difference (p Ͻ 0.05) is indicated (*). E, the suppression of PP5 expression increases the basal activity of ASK-1. A549 cells were treated with siRNA targeting PP5 (PP5TG2) or a miss matched control RNA (PP5TG2MM). After 48 h, ASK-1 activity was measured using MBP as a substrate. Representative Western blots showing PP5 protein levels in the crude cell lysates and the relative amount of ASK-1 are shown above the graph. The data are plotted as the percentage of the ASK-1 activity in controls (mean Ϯ S.D. n ϭ 6), and a statistical difference (p Ͻ 0.05) is indicated (*).
terminal deletion mutant of ASK-1 (ASK-⌬ N-HA), and an HA-tagged kinase domain deletion mutant of ASK-1 (ASK-1⌬ K-HA) were used to transfect A549 cells. The transfected cells were then lysed, and anti-HA antibodies were used to immunoprecipitate protein from the crude cell lysates. Western analysis of the immunoprecipitates found PP5 in association with both the N-terminal and kinase domain-deficient forms of ASK-1, indicating that PP5 associates with the C-terminal (residues 936 -1375) region of ASK-1 (Fig. 7). Therefore, PP5 does not appear to bind ASK-1 directly at the N-terminal Trxbinding site.
The activation of ASK-1 is associated with phosphorylation at Thr-845, which is located in the kinase domain, and PP5 can dephosphorylate Thr-845 in vitro (7). These observations lead to the hypothesis that following the activation of ASK-1 by oxidative stress, PP5 binds to and dephosphorylates ASK-1 (7). If the association of PP5 and ASK-1 is based on the ability of PP5 to recognize phosphorylation-activated ASK-1 as a substrate, then one would expect PP5 to have high affinity for the phosphorylated form of ASK-1 and a much lower affinity for the dephosphorylated form. If not, then following catalysis, ASK-1 would not dissociate from PP5, and ASK-1 would act as an inhibitor of PP5. In A549 cells, immunoprecipitation studies indicate that hypoxia is associated with both increased ASK-1 activity and an increase in the association of PP5 with ASK-1. Further, the suppression of PP5 expression is associated with an increase in ASK-1 activity. However, IP studies indicate that the association between PP5 and ASK-1 also occurs in the absence of stress (Fig. 6), and PP5 binds to a mutated form of ASK-1 in which the entire kinase domain (including Thr-845) has been deleted (Fig. 7). In addition, microcystin pull-down studies (in which PP5 is bound in an inactive state to microcystin-Sepharose) indicate that ASK-1 binds to microcystininactivated PP5 (Fig. 6C). Therefore, whereas the binding of PP5 to the C-terminal domain of ASK-1 may position PP5 to inactivate ASK-1 by dephosphorylating phospho-Thr-845, the association between PP5 and ASK-1 represents more than the simple recognition of ASK-1 as a substrate. DISCUSSION During hypoxia, the activation of several stress-activated protein kinases have been linked with postischemic cell death (34,35). One is ASK-1, which is also referred to as c-Jun NH 2 -terminal protein kinase kinase or JNK kinase kinase (22). ASK-1 functions as a MAPKKK that is activated during oxidative stress via a redox-induced dissociation of Trx-1, a small inhibitory redox protein that sequesters ASK-1 in an inactive state in the cytoplasm (36). TNF-␣ and the Fas death receptor adapter protein, Daxx, can also induce the activation of ASK-1, and the activation of ASK-1 by either mechanism is linked to the onset/propagation of an apoptotic response (36 -38). Still, the molecular mechanism(s) regulating ASK-1-mediated apoptosis is not entirely clear. Once activated, ASK-1 has been implicated in the phosphorylation/activation of both MKK4 and MKK3/6 (22). The activation of MKK4 leads to the activation of JNK, the subsequent phosphorylation of c-Jun at Ser 63 , and a cascade culminating in apoptosis (39). In contrast, the activation of MKK3/6 results in the activation of p38 MAPK, which is often associated with increased cell survival during stress (1,40).
The data presented here indicate that hypoxia induces a transient activation of ASK-1/MKK4/JNK that peaks after ϳ1 h and returns to basal levels by ϳ3 h. More importantly, our studies suggest that PP5 acts to turn off hypoxia-induced activation of ASK-1/MKK4/JNK via two related, yet distinct, processes. First, in a fairly rapid response, hypoxia augments the interaction of PP5 and ASK-1, with PP5 binding to the Cterminal region of ASK-1. The suppression of PP5 expression results in an increase in ASK-1 activity, and in vitro PP5 can dephosphorylate Thr-845, a site that becomes phosphorylated in association with the activation of ASK-1 (7,22,40). Therefore, our data support the hypothesis that PP5 acts as a negative regulator of ASK-1. Nonetheless, the suppression of PP5 expression does not prevent or augment hypoxia-induced MKK4/JNK activity, and the association of ASK-1 and PP5 is still observed (to a lesser extent) under normoxic conditions. This suggests that PP5 does not prevent ASK-1 activation.
Rather, PP5 appears to act at the inactivation phase of the cycle, "turning off" hypoxia-induced activation of ASK-1/ MKK4/JNK and thereby preventing hypoxia from inducing the sustained activation of ASK-1.
In a second phase, the expression of PP5 is induced by hypoxia. During prolonged periods of hypoxia (Ͼ30 min to 1 h), HIF-1, a heterodimeric transcriptional activator, becomes activated (41)(42)(43)(44)(45). Our studies indicate that the activation of HIF-1␣ induces the expression of PP5 by binding to a HIF-1 response element contained in the PP5 promoter, which results in an ϳ2-fold increase in both PP5 mRNA and protein levels. This suggests that HIF-1-induced PP5 expression may have a protective role during periods of chronic hypoxia by suppressing the sustained activation of ASK-1 triggered by the production of reactive oxygen species (Fig. 8). The suppression of PP5 expression with siRNA results in an increase in the activity of both MKK4 and JNK. However, the association of PP5 with ASK-1 may argue that PP5 act upstream of MKK4/JNK at the level of ASK-1. This concept is consistent with a recent study suggesting that rapamycin, by suppressing the activity of PP5, produces a response that also leads to the sustained activation of ASK-1 (16). Therefore, PP5 may represent a key regulator of ASK-1. Still, although the activation of ASK-1 commonly results in the activation of both MKK4/JNK-and MKK3/MKK6/ p38 signaling cascades, our studies indicate that the actions of PP5 are limited to the MKK4/JNK arm of the pathway, for the suppression of PP5 had no apparent effect on phosphorylation/ activation on p38 MAPK, p44/p42-MAPK/ERK1/2, or total c-Jun protein levels. Still, at present it is not yet clear how PP5 directs the actions of ASK-1 toward MKK4/JNK without affecting p38.
The ability of HIF-1 to regulate the expression of PP5 may also provide insight into the controversial relationship between HIF-1␣ and p53, where the interaction of HIF-1␣ and p53 has been reported to facilitate either a proapoptotic response or an antiapoptotic response (46 -49). Like HIF-1, p53 acts as a transcription factor that becomes stabilized following cellular stress (including hypoxia) and regulates the expression of many gene products that coordinate a wide variety of cellular responses (50). In proliferating cells, short term p53 activation is often associated with the induction of the cyclin kinase inhibitor protein, p21 WAF1/Cip1 and G 1 /S phase growth arrest. In contrast, prolonged p53 activation, triggered by severe or irreversible genomic damage, can induce an apoptotic response. The activation of p53 is achieved by an increase in p53 levels and by FIG. 8. Schematic representation of a role for PP5 in hypoxia-induced ASK-1 signaling cascades. ASK-1 functions as a MAP kinase kinase kinase that is activated by stress, including treatment with agents that generate reactive oxygen species (ROS) or hypoxia (22). In the absence of oxidative stress, ASK-1 is sequestered in the cytoplasm in an inactive state via association with a small inhibitory redox protein, Trx-1 (36). Redox-induced dissociation of Trx-1 results in the activation of ASK-1, which can phosphorylate/activate both MKK3/6 and MKK4. The activation of MKK3/6 leads to the activation of p38 MAPK, which is often associated with cell survival. The activation of MKK4 leads to the phosphorylation of JNK, the activation of JNK, the hyperphosphorylation of c-Jun at Ser-63 and a JNK-propagated cascade culminating in apoptosis. Under normoxic conditions, PP5 associates with ASK-1, which may help suppress the basal or "accidental activation" of ASK-1/MKK4 signaling. During hypoxia, the association between PP5 and ASK-1 increases, suppressing the prolonged activation of MKK4 by hypoxia. Under periods of prolonged hypoxia, the binding of the von Hippel-Lindau tumor suppressor protein (pVHL) to HIF-1␣ is prevented, which results in an increase in HIF-1␣ protein levels (42,44). This allows the association of HIF-1␣ with the aryl hydrocarbon receptor nuclear translocator protein (ARNT), and the formation of a transcriptional complex that drives the expression of PP5. Other phosphatases (MKP1/4/5, MKP2/6, PP2c, and PTP) suppress ASK-1 signaling by catalyzing the dephosphorylation of JNK and p38 (56). modifications of the p53 protein (e.g. phosphorylation and acetylation). p53 is inactivated via a feedback mechanism in which p53 induces the expression of Mdm2 (an E-3 ligase; murine double minute-2) that facilitates p53 nuclear export and degradation via the ubiquitin-proteasome pathway (51). However, whereas the stability of HIF-1␣ is regulated principally by hydroxylation, the association of p53 with Mdm2 is regulated principally by phosphorylation. During hypoxia, p53 is phosphorylated at Ser-15 via a mechanism that is induced following replication arrest and facilitated by the activation of ATR-kinase (52,53). Ser-15 phosphorylation affects p53 in several ways: 1) it decreases the binding affinity between Mdm2 and p53, which disrupts the negative feedback loop leading to proteolytic degradation; 2) it increases the transcriptional efficiency of certain p53-responsive genes (50); and 3) it allows the nuclear accumulation of p53 by blocking a nuclear export signal contained near the amino terminus of p53 (54).
Other studies have shown that in the nucleus, p53 can form a complex with HIF-1␣ (46,55), which suppresses the nuclear export and degradation of HIF-1␣ and may facilitate the actions of both p53 and HIF-1. Thus, the phosphorylation of p53 at Ser-15 may also affect the actions of HIF-1.
Previous studies with ISIS 15534, a potent and specific suppressor of PP5 expression (20), have shown that the inhibition of PP5 expression aids both p53-induced (6) and GR-induced (21) growth suppression. This resulted in the hypothesis that PP5 acts as a suppressor of both p53-and GR-mediated signaling cascades that affect cell growth. Recently, the p53 and GR pathways were linked when it was shown that 1) ISIS 15534 augments dexamethasone-induced p53 phosphorylation at Ser-15 and 2) dexamethasone-induced growth suppression is dependent on p53 (5). Thus, PP5 is probably acting upstream of p53, possibly acting to suppress the expression/activation of a GR-responsive kinase that acts on p53 and phosphorylates . The observation that HIF-1 regulates the expression of PP5 further links GR/p53-and HIF-1-induced signaling cascades, suggesting that HIF-1-induced PP5 expression contributes to a negative feedback mechanism that suppresses both ASK-1-and GR/p53-mediated responses. Under physiological conditions, HIF-1-induced PP5 expression may represent a protective response that prevents or delays cell death following oxidative stress during hypoxia. This may aid cell survival in hypoxic tissues before and during the growth of new blood vessels into the ischemic area. However, during tumor development, HIF-1-induced PP5 expression may have pathological consequences by aiding tumor cell survival in an ischemic environment while also suppressing actions of the p53 tumor suppressor protein (6) and glucocorticoids (21). Clearly, future studies to determine the role of PP5 tumor development seem warranted, and studies to determine if PP5 is indeed a positive factor in human tumor progression are in progress.