Ca2 /Calmodulin-dependent Protein Kinase II Is Required for ...

20 downloads 165611 Views 367KB Size Report
Sep 19, 2001 - bers of this class of compounds may have evolved separately because they are ... Greengard (Rockefeller University, New York) and Dr. I. Walaas (Uni- versity of Oslo .... viewed in Adobe Photoshop software. Apoptosis ...
THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2002 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 277, No. 4, Issue of January 25, pp. 2804 –2811, 2002 Printed in U.S.A.

Ca2ⴙ/Calmodulin-dependent Protein Kinase II Is Required for Microcystin-induced Apoptosis* Received for publication, September 19, 2001, and in revised form, November 5, 2001 Published, JBC Papers in Press, November 16, 2001, DOI 10.1074/jbc.M109049200

Kari E. Fladmark‡, Odd T. Brustugun‡, Gunnar Mellgren‡, Camilla Krakstad‡, Roald Bøe‡, Olav K. Vintermyr‡, Howard Schulman§, and Stein O. Døskeland‡¶ From the ‡Department of Anatomy and Cell Biology, University of Bergen, Årstadveien 19, N-5009 Bergen, Norway and §Department of Neurobiology, Stanford University School of Medicine, Fairchild D223, Stanford, California 94305-5125

The potent natural toxins microcystin, nodularin, and okadaic acid act rapidly to induce apoptotic cell death. Here we show that the apoptosis correlates with protein phosphorylation events and can be blocked by protein kinase inhibitors directed against the multifunctional Ca2ⴙ/calmodulin-dependent protein kinase II (CaMKII). The inhibitors used comprised a battery of cell-permeable protein kinase antagonists and CaMKII-directed peptide inhibitors introduced by microinjection or enforced expression. Furthermore, apoptosis could be induced by enforced expression of active forms of CaMKII but not with inactive CaMKII. It is concluded that the apoptogenic toxins, presumably through their known ability to inhibit serine/threonine protein phosphatases, can cause CaMKII-dependent phosphorylation events leading to cell death.

The endogenous suicide machinery of multicellular eukaryotes is an obvious target for natural toxic compounds. Toxins such as okadaic acid, calyculin A, tautomycin, motuporin, cantharidin, microcystins, and nodularin all target the catalytic subunits of the well conserved serine/threonine protein phosphatases (PP)1 1 and 2A (1, 2). The toxins are structurally diverse, sharing only the common motif required for high affinity binding to and inhibition of PP1 and PP2A. Members of this class of compounds may have evolved separately because they are produced by marine, freshwater, and soil micro-organisms as well as by insects. The PP inhibitors induce apoptosis in a wide range of cell types (3– 6), and completely resistant cell types have not been reported (7–9). We have previously found that the PP inhibitors can induce apoptosis with unprecedented rapidity, independently of new gene transcription or protein synthesis. Although apoptosis could be induced in the absence of caspase activity, it was amplified by caspases, notably caspase-3 (10). It is controversial whether these toxins can induce complete apoptosis through PP inhibition, since tautomycin analogs without phosphatase inhibitory * The work was supported by the Norwegian Research Council, the Norwegian Cancer Society, the Novo Nordic Foundation, and the Marine Science and Technology (MAST III) Program of the European Commission. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed. Tel.: 47 55 586375; Fax: 47 55 586360; E-mail: [email protected]. 1 The abbreviations used are: PP, protein phosphatase; CaMK, Ca2⫹/ calmodulin-dependent protein kinase; MLC, myosin light chain; AIP, autocamtide-2-related inhibitory peptide; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate; MLC, myosin light chain; MLCK, MLC kinase; HA, hemagglutinin; HEK cells, human embryonic kidney cells; NRK cells, normal rat kidney cells.

action have been shown to induce apoptotic cell death (11, 12). Furthermore, nodularin action has been ascribed to the creation of membrane pores (13). To establish whether protein phosphorylation is necessary for the apoptogenic action of PP inhibitors, we attempted to determine if protein kinase antagonists could counteract both apoptosis and protein phosphorylation events in PP inhibitortreated cells. Next, we searched to identify the protein kinase(s) required for the PP inhibitor-induced death. In view of the general nature of PP inhibitor-induced death, the responsible kinase(s) would be expected to be well conserved during evolution and expressed ubiquitously. Evidence will be presented that the PP inhibitors induced calmodulin-dependent protein kinase activity and that such activity was necessary for apoptosis. The calmodulin-dependent protein kinases can be divided into mono- or multifunctional depending on the number of known substrates (14, 15). The multifunctional CaMK subfamily comprises CaMKI, -II, and -IV. The generally expressed CaMKI and the less generally expressed CaMKIV require activation by CaMK kinase. CaMKIV has a survival promoting (anti-apoptotic) action in cerebellar neurons (16), whereas CaMKI has not been implicated in apoptosis. CaMKII is generally expressed. In mammalian cells, four CaMKII genes (␣, ␤, ␥, ␦) give rise to a number of different isoforms (17). Ca2⫹/ calmodulin-dependent autophosphorylation (intermolecular) of a specific Thr residue (Thr286 of the ␣ isoform) activates the enzyme and to some extent makes it autonomous, i.e. active in the absence of Ca2⫹/calmodulin (15, 18). Several protein phosphatases, including the PP inhibitor-sensitive PP1 and PP2A, can dephosphorylate CaMKII at Thr286 and, thus, reverse the autonomous state (15). CaMKII is a major mediator of cellular Ca2⫹ effects, and its biological actions have been studied in most detail in intermediate metabolism and neural signaling (14, 19). CaMKII has received little attention in relation to apoptosis but has been reported to be activated downstream of protease activation in apoptosis induced by UV light and TNF␣ (20). The present study will show that CaMKII is necessary for rapid PP inhibitor-induced death and that enforced expression of active CaMKII can lead to apoptosis. EXPERIMENTAL PROCEDURES

Reagents—Nodularin was from LC Services. Microcystin-LR, calphostin C, HA100, W7, and bisbenzimide (H33342) were obtained from Calbiochem. KT5720 and KT5926 were from Kamiya Biomedical Co. H89, KN62, KN93, and ML7 were from Seikagaku America, Inc. Myosin light chain and myosin light chain kinase, purified as previously described (21), were kindly provided by Dr. J. T. Stull (University of Texas Southwestern, Dallas). The peptide sequence CaMKII (273–302; T286A) and CaMKII isolated from rat brain were provided from Dr. P. Greengard (Rockefeller University, New York) and Dr. I. Walaas (University of Oslo, Norway). The CaMKII inhibitor peptide autocamtide-

2804

This paper is available on line at http://www.jbc.org

Microcystin-induced Apoptosis Depends on CaMKII ⫹

2-related inhibitory peptide AIP (H -KKALRRQEAVDAL-OH) was synthesized by Macromolecular Resources. Myosin light chain kinase (MLCK)-(480 –501) and MLCK-(493–512) were kindly provided by Dr. J. Tooze (EMBL, Germany). Stock solutions (5–25 mM) of the peptides were made in 20 –100% Me2SO, and small aliquots (20 –50 ␮l) were stored in liquid nitrogen. All solutions and compounds were handled in the dark, except for calphostin C, which was light-activated according to the instructions from the manufacturer. [␥-32P]ATP (3000 Ci/mmol) and [32P]orthophosphate (10 mCi/ml) were from Amersham Bioscience, Inc. Pharmalyte and linear immobilized pH 4.0 –7.0 gradients were from Pharmacia Biotechnology. Ponasterone A was from Invitrogen. Plasmids—The plasmid used for expression of both wild type and mutant forms of ␣CaM kinase II was the pCD derivative SR␣ (22). The plasmids encoding the ␣CaMKII mutant forms T286D, T286D/T305D/ T306D, and K42M and the truncated kinase (1–326) and its T286D form have been described earlier by Hanson et al. (23). Some of the ␣CaMKII forms were tagged by insertion of a 18-amino acid sequence from the influenza hemagglutinin HA1 protein between Thr3 and Ile4. For inducible-expression, ␣CaMKII T286D was subcloned into the pIND vector (ecdysone-inducible system, Invitrogen) at EcoRI and XhoI sites. The CaMKII auto-inhibitory peptides AIP, AC3-I, and AC3-C were described earlier by Braun and Schulman (24). They were expressed in pUC19 with a Rous sarcoma virus promotor. Kinase Assays—CaMKII and MLCK (at final concentrations of 11 nM) were assayed in a 60-␮l volume of 15 mM HEPES (pH 7.2), 10 mM MgCl2, 2 mM [␥-32P] ATP (1.5 ␮Ci/ml), 1 mM CaCl2, 10 ␮g/ml calmodulin, 100 mM KCl, and bovine serum albumin (0.5 mg/ml). The high ATP concentration was chosen to mimic intracellular conditions. The substrate was either 20 ␮M syntide-2 (CaMKII) or 5 ␮M MLC (MLCK). The reaction was arrested after 5 or 10 min by removing 25-␮l aliquots, which were spotted on phosphocellulose paper (CaMKII assay) or Whatman No. 3MM paper (MLCK assay) discs. The phosphocellulose paper was washed repeatedly (8 times) in dilute (50 mM) phosphoric acid, and the filter paper discs were washed in 5% trichloroacetic acid with 2 mM pyrophosphate. Both types of paper were rinsed in ethanol, dried, and subjected to scintillation counting. Cell Culture and Microinjection—Hepatocytes were isolated from male Wistar rats (120 –200 g) by in vitro collagenase perfusion and either kept in suspension or cultured in monolayers on collagen gel as previously described (25). For short term suspension culture the hepatocytes were resuspended (1.2 ⫻ 106 cells/ml) in pre-gassed (5% CO2, 95% O2) low phosphate (0.1 mM) Krebs-Ringer bicarbonate buffer (10 mM Hepes (pH 7.4), 120 mM NaCl, 5.3 mM KCl, 0.01 mM KH2PO4, 1.2 mM MgSO4, 1.0 mM CaCl2) with 5 mM lactate, 5 mM pyruvate, and 0.5% bovine serum albumin. Incubations were in capped vials with gyratory shaking (175 cycles/min, 37 °C). Human embryonic kidney (HEK 293) cells, mouse embryonic fibroblasts (C3H/10T1/2 Cl8), monkey kidney cells (COS-1), and normal rat kidney (NRK) cells were grown in Dulbecco’s modification of Eagle’s minimal essential medium with 10% fetal calf serum, 50 units/ml of penicillin, and 50 ␮g/ml streptomycin. Mouse embryonic fibroblasts (Swiss 3T3) were grown in Eagle’s minimal essential medium and RPMI 1640 (50/50 v/v) and supplemented with serum and antibiotics as described above. IPC-81 rat promyelocytic leukemia cells were cultured in Eagle’s minimal essential medium with 7% horse serum. Microinjection was performed using an Eppendorf 5171 micromanipulator and 5246 microinjector mounted on a Nikon Diaphot 300 inverted microscope equipped with an incubator providing 5% CO2 and 37 °C. Microcapillaries (type BF100 –10, 1.00/0.78) and puller (Model P-87) were from Sutter Instrument Co. Phosphatase inhibitors were delivered in a buffer with 5% Me2SO but otherwise approaching the intracellular composition of electrolytes (26). HEK 293, NRK, and Swiss 3T3 cells (10,000 cells/cm2) were injected 24 – 48 h after seeding. Hepatocytes (30,000 cells/cm2) were injected after 44 –56 h of culture, as described earlier (27). Labeling of Cellular Phosphoproteins and Two-dimensional Gel Electrophoresis—Suspension cultures of hepatocytes (2 ⫻ 106/ml) were preincubated for 35 min in a low phosphate (0.1 mM) Krebs-Ringer bicarbonate buffer as described above, supplemented with 0.5 mCi/ml 32P. Cells were incubated with agents (phosphatase and kinase inhibitors) to be tested for their ability to modulate protein phosphorylation. The incubations were terminated by adding a 10-fold excess of ice-cold 8% aqueous trichloroacetic acid. Samples were spun (15,000 ⫻ g for 15 min) and washed in 5% trichloroacetate. Cell pellets were resuspended in 100 ␮l of a solution containing 9.8 M urea, 100 mM 1,4-dithioerythreitol, 1.5% v/v Pharmalyte (pH 3.5–10), 0.5% v/v Pharmalyte (pH 5– 6), 4% w/v CHAPS, and 0.2% w/v SDS. Sample separation was by isoelectric

2805

focusing (80,000 Volt hours) in linear-immobilized pH gradients (IPG strips, pH range 4.0 –7.0; Amersham Bioscience, Inc.). Strips were equilibrated for 12 min in solution (0.05 M Tris HCl (pH 6.8), 2% w/v SDS, 6.5 M urea, 26% glycerol) with 100 mM 1,4-dithioerythreitol and thereafter for 5 min in the same solution with 0.24 M iodoacetamide. The strips were then subjected to SDS-electrophoresis (13.8% polyacrylamide separation gel), and the gels were dried and exposed to DuPont NEF-496 autoradiography films at ⫺80 °C with intensifying screens. For further details see Gjertsen et al. (28). The exposed autoradiographic films were scanned on a Agfa Arcus II flatbed scanner and viewed in Adobe Photoshop software. Apoptosis Evaluation—For routine assessment of apoptosis, the cell morphology was evaluated by inverted phase microscopy using phaseand Hoffman-modulated optics. Apoptotic cells were easily discriminated from non-apoptotic (both normal and necrotic) cells by the appearance of multiple surface buds. For scoring of the chromatin condensation, cells fixed in 0.1 M sodium cacodylate buffer (pH 7.4) with 1.5% glutaraldehyde were stained with 1 ␮g/ml DNA-specific dye Hoechst 33342 (bisbenzimide). For transmission, electron microscopy cells were fixed in 1.5% glutaraldehyde buffered with 0.1 M sodium cacodylate (pH 7.4) and processed as described previously (3). Transfections—COS-1 cells were transfected in 6-well dishes the day after seeding. Cells were transfected using either SuperFect (Qiagen) or FuGene6TM (Roche Molecular Biochemicals) according to the manufacturer’s instructions using between 1 and 4 ␮g of DNA. NRK cells were stably transfected with pVgRxR vector (ecdysoneinducible system, Invitrogen), and the resulting clones were tested by transiently transfecting them with a pIND-lacZ construct and determining the percentage of cells with ␤-galactosidase expression after induction with ponasterone A. The clone with the highest level of ␤-galactosidase expression was chosen for stable transfection with pIND-␣CaMKII T286D. Colonies were picked after 3 weeks in selective media (0.6 mg/ml ZeocineTM for pVgRxR selection and 0.6 mg/ml G-418 for pIND-␣CaMKII T286D selection) and re-seeded by dilution to obtain new colonies from single cells. Immunofluorescence—Cells were fixed and processed for immunofluorescence as described in Srinivasan et al. (29) using a monoclonal antibody to the hemagglutinin tag (clone 12CA5, Roche Molecular Biochemicals). Cells were extensively washed before exposure to the secondary antibody, anti-mouse CY3 (Zymed Laboratories Inc.). The cells were washed and counterstained with 5 ␮g/ml Hoechst 33258 before mounting onto glass slides using Vectashield (Vector Laboratories). Fluorescence was observed using a Zeiss Axioplan microscope, and photomicrographs were adjusted using Photoshop. Western Blotting—For myosin light chain detection, proteins were transferred from two-dimensional gels by electroblotting onto nitrocellulose membranes. Blots were incubated with anti-myosin light chain antibody (M4401, Sigma-Aldrich) followed by horseradish peroxidaseconjugated sheep anti-mouse antibody (NA931) from Amersham Bioscience, Inc. For CaMKII detection, transfected COS-cells were washed in cold phosphate-buffered saline and homogenized and in buffer (10 mM potassium phosphate buffer (pH 6.8) with 1 mM EDTA) containing 10 mM CHAPS and a protease inhibitor mixture (0.1 mg/ml soybean trypsin inhibitor, 50 ␮g/ml aprotinin, 50 ␮g/ml leupeptin, and 10 ␮g/ml of chymostatin, antipain, and pepstatin). Cells were disrupted by sonication, and extracts were prepared by centrifugation at 10,000 ⫻ g for 5 min. Proteins (60 ␮g/lane) were separated on 9% SDS-polyacrylamide gels and electroblotted onto nitrocellulose membranes. Tagged CaMKII was detected using the anti-HA antibody and secondary antibody as described above. The blots were developed using chemiluminescence (ECL, Amersham Bioscience, Inc.). RESULTS

Protein Kinase Inhibitors Counteract Microcystin-induced Hepatocyte Apoptosis According to Their Potency as CaMKII Antagonists—Inhibition of protein phosphatases can only increase protein phosphorylation in the presence of kinase activity. It is therefore possible to prevent the biological action of phosphatase inhibitors by antagonizing the relevant protein kinases. Based on this consideration, protein kinase inhibitors were tested for their ability to counteract the apoptogenic effect of microcystin. Cell-permeable inhibitors directed against tyrosine kinases, casein kinases, and cyclin-dependent protein kinases had no effect. By contrast, anti-apoptotic effect was observed with the general calmodulin inhibitor W7 and a number

2806

Microcystin-induced Apoptosis Depends on CaMKII

FIG. 1. Protein kinase inhibitors protected against microcystin-induced hepatocyte apoptosis according to their potency as CaMKII antagonists. The kinase inhibitors KT5926, calphostin C, KN93, KN62, KT5720, H89, ML7, HA100, and the calmodulin antagonist W7 were tested at various concentrations for the ability to protect rat hepatocytes against the apoptogenic effect of microcystin-LR and as antagonists of CaMKII and MLCK. A–C, representative experiments for the three inhibitors, KT5926 (‚), calphostin C (E), and KN93 (䡺). The insets of A show prevention by KN93 of the ultrastructural features accompanying microcystin-induced apoptosis. MC, microcystin. Bar, 10 ␮m. Apoptosis was determined by differential interference contrast microscopy of cells after 45 min in suspension culture with 0.4 ␮M microcystin. The kinase assays were at pH 7.2, 37 °C, 2 mM ATP, and near physiological ionic strength. D, the IC50 value for apoptosis protection plotted against the IC50 values for inhibition of CaMKII (●) and MLCK (ƒ) for the nine kinase inhibitors tested. The diagonal is the regression line between the IC50 values for CaMKII antagonism and apoptosis protection. KN62 and KN93 were without effect on MLCK in the concentration range (indicated by a dashed line with an arrowhead) in which they were soluble. Panel D represents the average of 3–7 determinations for each compound and parameter. Further experimental details are given under “Experimental Procedures.” TABLE I Modulation of microcystin-induced hepatocyte apoptosis by microinjection of protein kinase antagonist peptides Hepatocytes in monolayer were injected 48 h after plating. The percentage of injected cells with apoptotic budding was quantified 20 min after injection. Each experiment (n) involves 70 injected cells. MCYSTLR, microcystin. Compound injected

Apoptotic cells

n

% ⫾ S.E.

FIG. 2. Apoptotic induction rate of primary rat hepatocytes in monolayers after injection of 200 ␮M microcystin alone (E) or together with 4.4 mM general calmodulin inhibitory peptide MLCK-(493–512) (‚) or 25 mM CaMKII-preferring inhibitor peptide AIP (Œ). Further details are given under “Experimental Procedures.” Values are the mean ⫾ S.E. from at least 3 separate experiments, each including ⬎50 injections.

of CaMK inhibitors as well as with high concentrations of the protein kinase C inhibitor calphostin C (Figs. 1 and 2; data not shown). This suggested that a CaMK sensitive to high concentrations of calphostin C was required for the apoptogenic effect of microcystin. The most abundant members of the CaMK subfamilies are MLCK and CaMKII (14). Antagonists were therefore assayed for the ability to inhibit MLCK, CaMKII, and microcystininduced apoptosis in hepatocytes in suspension culture. Because some of the kinase inhibitors are competitive with respect to ATP and others target calmodulin binding as well, the enzyme assay was run at near physiological ATP (about 50-fold

Vehicle 0.1 mM MCYST-LR 0.2 mM MCYST-LR 0.2 mM MCYST-LR ⫹ 4.4 mM MLCK-(493–512) 0.2 mM MCYST-LR ⫹ 4.0 mM CaMKII (273–302; T286A) 0.2 mM MCYST-LR ⫹ 5.0 mM MLCK-(480–501) 0.2 mM MCYST-LR ⫹ 4.8 mM protein kinase C-(283–312)

0 21.2 ⫾ 5 73.5 ⫾ 6 21.0 ⫾ 6

3 3 6 4

51.2 ⫾ 4

6

74.8 ⫾ 6

3

69.6 ⫾ 9

3

higher than commonly used in kinase assays) and calmodulin concentrations. The concentration dependence of the kinase inhibitors calphostin C, KN93, and KT5926 as apoptosis antagonists correlated with their inhibitory potency against CaMKII. This selective correlation was further strengthened when a battery of other kinase inhibitors spanning several orders of magnitude of IC50 values for CaMKII were included in the study (Fig. 1D). The CaMKII inhibitors also protected hepatocytes against apoptosis induced by the protein phosphatase inhibitors nodularin and okadaic acid (not shown). Cells treated with microcystin with or without the kinase inhibitors KN93 (Fig. 1A, insets), KN62, or KT5926 (not shown) were studied by electron microscopy. The cells treated with microcystin plus inhibitor showed normal ultrastructure.

Microcystin-induced Apoptosis Depends on CaMKII

2807

FIG. 3. CaMKII inhibitors prevented microcystin- and nodularin-induced protein phosphorylation events in intact cells. Freshly isolated rat hepatocytes in suspension culture were prelabeled for 35 min with 32Pi and thereafter incubated with various agents in the continued presence of 32Pi before being analyzed for protein phosphorylation pattern. A, autoradiograms of two-dimensional electrophoresis separations of proteins from hepatocytes incubated 12 min without either phosphatase (a) or kinase inhibitor (Control), with 1 ␮M microcystin-LR (b), or with 1 ␮M microcystin-LR plus 30 ␮M KT5926 (added 5 min before the microcystin) (c). The circles numbered 1 and 2 surround phosphoproteins observed in control cells whose labeling intensity increased after microcystin treatment (arrows in b). After exposure to microcystin, several new phosphorylated proteins also appeared (circles numbered 3–7). Note that KT5926 prevented the microcystin-induced increases of protein phosphorylation. The dotted lines in panel A surround myosin light chain. B, Western blot of myosin light chain from control hepatocytes (upper panel), hepatocytes treated for 3 min with 5 ␮M nodularin (middle panel), and cells pretreated for 10 min with 30 ␮M CaMKII inhibitor KN93 and then for 3 min with nodularin (lower panel). The percentage of apoptosis was determined in aliquots taken from parallel cell incubations. Values are the mean ⫾ S.E., n ⫽ 4.

These cells could attach to culture dishes and undergo several rounds of DNA replication, in contrast to cells treated with microcystin alone, which failed to attach (data not shown). The data of Fig. 1 indicated that CaMKII or a kinase with an extremely similar inhibitor specificity was required for microcystin-induced hepatocyte apoptosis. To further pinpoint the apoptosis-promoting kinase, microcystin and peptide inhibitors of protein kinases were co-injected into primary hepatocytes in a monolayer culture (Fig. 2). Cells injected with inhibitors directed preferentially against MLCK or protein kinase C were not protected. Cells injected with a general inhibitor of calmodulin-dependent protein kinases (MLCK-(493–512) (30) or with inhibitors directed preferentially against CaMKII (31, 32)) were protected, as evidenced by the delayed appearance of apoptotic cells (Fig. 2; Table I). CaMK Antagonists Block Phosphatase Inhibitor-induced Protein Phosphorylation in Hepatocytes—Hepatocytes preincubated with radioactive phosphate were challenged with microcystin with or without the kinase inhibitor KT5926, and the labeled phosphoproteins were separated by two-dimensional gel electrophoresis. Several proteins showed increased phosphorylation in cells incubated with microcystin alone, but KT5926 effectively prevented these phosphorylation events (Fig. 3A). Under the conditions used, 12 min of incubation with 1 ␮M microcystin induced 65% apoptosis in the absence of KT5926 against less than 2% in the presence of the inhibitor.

This suggested a close link between microcystin-induced protein phosphorylation and apoptosis. A puzzling finding was microcystin-induced, KT5926-inhibitable phosphorylation of proteins at the position of myosin light chain on two-dimensional gels (surrounded by dashed white lines in Fig. 3A), since the inhibitor studies presented in Fig. 1 indicated that KT5926 action correlated with inhibition of CaMKII rather than of MLCK. This phenomenon was studied further using a high concentration (5 ␮M) of nodularin to induce apoptosis and replacing KT5926 by KN93, which had no inhibitory action toward MLCK (Fig. 1). The position of myosin light chain variants was determined by immunoblotting. KN93 was found to inhibit MLC phosphorylation, indicating a CaMKII-dependent pathway of MLC phosphorylation. It was noted that apoptosis development was inhibited by KN93 to a similar extent (Fig. 3B). The Ability of CaMK Antagonists to Block the Acute Apoptogenic Action of Phosphatase Inhibitors Is Not Restricted to Hepatocytes and May Be General—The experiments reported so far suggested that hepatocyte apoptosis acutely induced by phosphatase inhibitors was dependent on CaMKII. To determine whether this was a hepatocyte-specific or a general phenomenon, a number of cell types were tested for CaMK involvement in apoptosis induced by microinjected microcystin or nodularin. Microinjection was required because only hepatocytes among mammalian cells have an efficient uptake system for microcystin and nodularin. C3H/10T1/2 Cl8 cells (Fig. 4),

2808

Microcystin-induced Apoptosis Depends on CaMKII

FIG. 4. CaMKII inhibitors protected C3H10T1/2 Cl8, NRK, and HEK 293 cells against induction of apoptosis by microinjected phosphatase inhibitor. C3H10T1/2 Cl8 fibroblasts (upper panel), NRK cells (middle panel), and HEK 293 cells (lower panel) were injected with 250, 200, or 100 ␮M nodularin, respectively, and scored for apoptotic morphology after 5, 3, or 1 min, respectively. The cells had either been pretreated for 40 min with 25 ␮M KT 5926 (right columns), 25 ␮M KN 93 (middle columns), or had not been pretreated (left columns). Values are the mean ⫾ S.E. from 3 separate experiments, each including ⬎50 injections.

NRK cells (Fig. 4), HEK293 cells (Fig. 4), Swiss 3T3 (Fig. 5), COS-1 cells, and rat promyelocytic IPC-81 cells (not shown) were all protected against nodularin- or microcystin-induced apoptosis by KN93 as well as by KT5926. When injected with a low concentration of nodularin (25 ␮M), fibroblasts preincubated with KN93 showed not only strongly increased latency before apoptosis onset but in many cases were protected completely against apoptosis (Fig. 5B). The CaMK antagonists inhibited all apoptotic hallmarks observed, including cell budding and fragmentation, cell shrinkage, disappearance of microvilli, organelle segregation, externalization of phosphatidylserine, and hypercondensation of chromatin (data not shown). As for the hepatocytes (Figs. 1 and 2), KT5926 protected more efficiently than KN93 at equimolar concentrations (Fig. 5). This is compatible with CaMKII involvement, since KT5926 is a more potent CaMKII inhibitor than KN93 (Fig. 1B). The CaMKII-specific auto-inhibitory peptide AIP, when comicroinjected with nodularin, strongly delayed apoptosis in all cell types (HEK293, COS-1, rat promyelocytic IPC-81) tested (not shown). The AIP peptide presumably was more stable in fibroblasts than in hepatocytes, since Swiss 3T3 cells co-injected with 12.5 ␮M nodularin and AIP peptide did not undergo apoptosis during the 60 min after injection that they were studied, whereas more than 90% of cells injected with 12.5 ␮M nodularin alone had undergone apoptosis after 30 min. An alternative way of introducing the inhibitory peptide was to transiently transfect COS-1 cells with an AIP expression vector. Such cells showed significantly decreased apoptosis response to the phosphatase inhibitor okadaic acid, which, unlike microcystin, penetrates COS cells and, therefore, does not have to be injected (Fig. 6). Also, cells with enforced expression of autoinhibitory CaMKII peptide, AC3-I (Fig. 6), were partially protected against okadaic acid-induced apoptosis. Transfection with a control, inactive peptide (AC3-C), in which the amino acid sequence HRQ in AC3-I was substituted with DGE, failed to protect (Fig. 6).

FIG. 5. CaMKII inhibitors protected Swiss 3T3 fibroblasts against apoptosis induced by microinjected nodularin. A, Swiss 3T3 fibroblasts preincubated with 25 ␮M KN93 (right column) were compared with non-preincubated cells (left column) with respect to apoptotic morphology 60, 75, and 120 s after microinjection of 200 ␮M nodularin. B, cells that had been preincubated with 25 ␮M CaMKII-inhibitors KT5926 (䡺) or KN93 (E) or left untreated (●) were microinjected with 25 ␮M nodularin. The percentage of apoptotic cells was determined as a function of time after injection. The CaMKII inhibitors were added to the medium 30 – 60 min before injections. The lower inset of panel B shows apoptotic cells 6 min after injection of nodularin. The upper inset shows non-apoptotic cells pretreated with KT5926 15 min after nodularin injection. Values are the mean ⫾ S.E. from at least three separate experiments, each including ⬎50 injections. Cells were photographed using Hoffmann-modulated interference optics. Bars, 10 ␮m.

Microcystin-induced Apoptosis Depends on CaMKII

2809

FIG. 6. The effect of CaMKII-specific peptide inhibitors on okadaic acid-induced apoptosis in COS-cells. COS-1 cells were transfected with AIP-based inhibitor, autoinhibitory peptide (AC3-I), autoinhibitory inactive control peptide (AC3-C), or empty vector. Six hours after transfection, the cells received either okadaic acid or vehicle. Apoptosis was determined 8 h after transfection by the presence of cell surface budding and chromatin condensation. Values are the mean ⫾ S.E. of 3– 6 separate experiments.

Enforced Cell Expression of Active CaMKII Can Induce Apoptosis—Having found CaMKII to be pivotal for phosphatase inhibitor-induced death, we next wanted to know if activation of endogenous CaMKII was sufficient on its own to induce apoptosis. Activation of endogenous CaMKII in hepatocytes by the Ca2⫹-elevating agents vasopressin or the Ca2⫹ ionophore A23187 (33–35) did not lead to apoptosis. This might be due to Ca2⫹-induced survival signaling through other mediators than CaMKII. Next, wild type and several mutagenized variants of ␣-CaMKII were expressed in COS-1 and NRK cells to determine if increased CaMKII activity without a Ca2⫹ signal could promote apoptosis. Apoptosis was induced in COS-1 cells within a few hours after transfection with monomeric ␣CaMKII-(1–326) (Fig. 7A), truncated to delete the enzyme self-association domain (36). The CaMKII inhibitor KN93 could prevent the start of apoptosis when given just after transfection (Fig. 7B) and could halt the apoptosis development when given after apoptosis had begun (Fig. 7C). This indicated that it was the kinase activity of the transfected gene product that killed the cells. The antiapoptotic effect of KN93 was reversed after transfer to KN93-free medium (Fig. 7, B and C), as expected for a competitive inhibitor of CaMKII. The phase contrast morphology of mock-transfected cells (Fig. 7D) and CaMKII-transfected cells (Fig. 7E) is shown. The transfection method itself gave a certain degree of apoptosis (Fig. 7), and to exclude that the transfection reagent used contributed to the apoptogenic effect of ␣CaMKII-(1–326), the latter was also introduced by direct microinjection into the COS-1 cells. About 100 COS-1 cells were injected with 1 ␮g/␮l CaMKII construct on three different occasions, and each time between 10 and 30% of the injected cells developed apoptosis within 12 h after injection. Cells injected with empty vector failed to develop apoptosis (data not shown). To know what part of the CaMKII molecule was responsible for apoptosis induction, COS-1 cells were transfected with wild type CaMKII and a number of mutant forms of CaMKII subunit (Fig. 8A). These experiments supported the supposition that catalytic activity was essential for apoptosis, since the inactive ␣ CaMKII K42M failed to induce apoptosis. The autonomously active ␣ CaMKII T286D was at least as active as wild type as an apoptosis inducer. Apoptosis was inhibited by co-transfection with the CaMKII-specific auto-inhibitory peptide AIP. Inactivation of the calmodulin binding domain in ␣ CaMKII T286/305/306D did not interfere negatively with apoptosis induction. It was noted that apoptosis development was highest just after

FIG. 7. COS-cells transfected with CaMKII developed KN93inhibitable apoptosis. A, COS-1 cells were transfected with ␣CaMKII-(1–326) (E) or with the expression vector alone (ƒ), and scored for apoptosis. Values are mean ⫾ S.E. from four separate experiments. B, cells transfected with ␣CaMKII-(1–326) as described above and then exposed to medium without (E) or with (●) 40 ␮M KN93. The medium was replaced by new medium (without KN93) 4 h later. C, cells transfected with CaMKII and left in medium without KN93 (E) or supplemented with KN93 3.7 h after start of transfection (●). Seven hours later the cells were transferred to KN93-free medium. D, COS-1 cells 10 h after transfection with expression vector alone. E, COS-1 cells 8 h after transfection with ␣CaMKII-(1–326). Bars, 10 ␮m.

the cells had begun to express detectable autonomously active CaMKII and that the recruitment of new cells to apoptosis declined (not shown) even if CaMKII expression increased further (Fig. 8, B–D). It is possible that CaMKII activation could induce counter-regulation, diminishing its apoptogenic action. Similar results to those shown for COS-1 cells in Figs. 7 and 8 were found in HEK 293 and NRK cells, but only 3– 6% of the results became apoptotic after lipid-mediated CaMKII transfection. Parallel experiments using green fluorescent protein revealed that the transfection rate was only 20 –30% in these cells and that the rate of fluorescence accumulation was much slower than in the COS cells. To achieve a higher percentage of expressing cells and a more rapid protein expression, autoactive ␣CaMKII T286D was expressed in an ecdysone-inducible construct in NRK-cells. In such cells, moderate, but significant, apoptosis was observed after induction with the ecdysone analog ponasterone A (Fig. 9). In conclusion, autoactive CaMKII, when abruptly expressed, can induce apoptosis in more than one cell type.

2810

Microcystin-induced Apoptosis Depends on CaMKII

FIG. 8. Apoptosis in COS-cells after transfection with mutant forms of ␣CaMKII. A, COS-1 cells were transfected with a catalytically inactive ␣CaMKII (column 2), plasmid expressing the inhibitory peptide AIP (column 3), ␣CaMKII (column 4), autoactive kinase alone (column 5) and together with AIP (column 6), truncated autoactive kinase (column 7), and autoactive kinase with deficient calmodulin binding (column 8). Column 1 shows COS-1 cells treated with the transfection agent FuGene6 only. Cells were fixed 7 h after transfection and scored for apoptosis by the appearance of surface budding and chromatin condensation. Values are the mean ⫾ S.E. of 3–5 different experiments. w.t., wild type. B, CaMKII expression in COS-1 cells, analyzed by Western blot. Cell lysates were prepared at different time intervals from COS-1 cells after transfection with a hemagglutinin-tagged autoactive form of the ␣CaMKII (␣CaMKII T286D tag). The immunoblot was developed with anti-HA monoclonal antibody. C, COS-1 cells transfected with tagged-␣CaMKII T286D. The expression of CaMKII was determined 7 h after transfection by immunostaining with an anti-HA monoclonal antibody as described under “Experimental Procedures.” D, Hoechst staining of the same field as in C. Apoptotic COS-1 cells showing chromatin condensation in D can be recognized in C as rounded cells with a high expression of CaMKII.

FIG. 9. Apoptosis induction by ponasterone A in NRK cells with ecdysone-regulated expression of CaMKII. NRK cells stably transfected with an ecdysone-regulatable CaMKII expression system (see under “Experimental Procedures”) were incubated with various concentrations of the ecdysone analog ponasterone A. Apoptosis was determined 8 h after the addition of ponasterone A or ethanol (vehicle). The slight spontaneous apoptosis (similar in transfected NRK cells treated with vehicle or wild type NRK cells treated with ponasterone A) was subtracted. Values are the mean ⫾ S.E. of 3– 4 different experiments.

DISCUSSION

The major questions addressed by the present study were whether toxins like microcystin and nodularin induce apoptosis through inhibition of protein phosphatases, and if so, which kinase(s) was responsible for the protein phosphorylation events critical for apoptosis. The protein kinase antagonists KT5926 and KN93 inhibited toxin-induced apoptosis and toxininduced protein phosphorylation to a similar degree. This provided evidence that protein phosphorylation was essential for the induced apoptosis but could not tell whether protein phosphorylation was instrumental only in apoptosis execution or, also, in its triggering. A number of recent studies demonstrate

caspase-dependent activation of protein kinases during apoptosis (37, 38). We have previously shown that the nodularininduced protein phosphorylation occurred in the presence of enough caspase inhibitor to completely block apoptosis (10). We conclude therefore that toxin-induced protein phosphorylation occurs upstream of caspase activation. It appears therefore that nodularin and microcystin induced apoptosis through increased protein phosphorylation, which is a logical consequence of protein phosphatase inhibition. In view of the broad substrate specificity of the toxin targets PP1 and PP2A (39) and the multitude of protein kinases known to exist, we expected that more than one protein kinase was responsible for catalyzing the toxin-induced protein phosphorylation events critical for apoptosis. Cell-permeable protein kinase antagonists showed a strong correlation between the ability to protect against microcystin-induced apoptosis and to inhibit the multifunctional CaMKII. The apoptosis was also counteracted by microinjected or overexpressed auto-inhibitory peptides of CaMKII. Finally, apoptosis could be induced by overexpression of active, but not inactive, CaMKII. Surprisingly, it appeared that a single protein kinase, CaMKII, was pivotal for phosphatase inhibitor-induced apoptosis. It should be considered, however, that the enzyme is unusually well suited to mediating the rapid effects of PP inhibitors. CaMKII is subjected to rapid cycles of phosphorylation/dephosphorylation at Thr286, which may enable it to respond to Ca2⫹ oscillations (40). The activating phosphorylation of Thr286 is rapidly reversed by either PP1 or PP2A, explaining the activation of CaMKII in hepatocytes incubated with microcystin (34). The protein phosphorylation induced by microcystin was stronger than that induced by CaMKII activators like vasopressin (34). Microcystin may therefore not only activate CaMKII but also may stabilize the phosphorylated state of CaMKII substrates. This may explain why we observed less apoptosis in CaMKII overexpressing cells than in cells exposed to PP inhibitors.

Microcystin-induced Apoptosis Depends on CaMKII Given the numerous substrates phosphorylated in microcystin-exposed cells (Fig. 3), one would expect some of them to play a more important role in apoptosis development than others. Phosphorylation of myosin light chain (MLC) was observed early after exposure to microcystin or nodularin, and the extent of MLC phosphorylation was correlated with apoptosis. CaMKII is not capable of phosphorylating MLC (17), but CaMKII is an efficient activator of MLCK (41). Disruption of CaMKII activation of MLCK, therefore, is the most plausible explanation of the observed inhibition of nodularin-caused MLC phosphorylation by the CaMKII inhibitor KN-93 (Fig. 3B). MLC phosphorylation catalyzed by MLCK has been reported to be necessary for the formation of apoptotic blebs in cells after serum withdrawal (42). The death-associated protein kinase members have also been shown to target MLC (43, 44). Recently, the Rho effector ROCK I was shown to up-regulate MLC phosphorylation and, thereby, induce membrane blebbing. The activation of ROCKI was dependent on cleavage by caspase-3 (45, 46), and its contribution was more probably at a late stage of apoptosis. In contrast, CaMKII-dependent phosphorylation in microcystin-induced apoptosis was independent of caspase activation (10).2 CaMKII can also be activated through proteolysis, as in cells undergoing apoptosis induced by tumor necrosis factor or UV light (20). It is possible that proteolytically activated CaMKII (47, 48) has a physiological role in apoptosis execution and that microcystin short-cuts this pathway by direct activation of CaMKII. In conclusion, the rapid apoptosis induced by phosphatase inhibitors requires the activity of CaMKII. Work is in progress to resolve the identity of proteins other than MLC phosphorylated early in the process of phosphatase inhibitor-induced apoptosis. Acknowledgments—We are grateful to Drs. James T. Stull and Ivar Walaas for providing enzymes at an early stage of these studies. We are also indebted to Nina Lied Larsen, Erna Finsås, and Berit Hausvik for superior technical assistance. REFERENCES 1. Holmes, C. F. B., and Boland, M. P. (1993) Curr. Opin. Struct. Biol. 3, 934 –943 2. Goldberg, J., Huang, H.-B., and Kwon, Y.-G. (1995) Nature 376, 745–753 3. Bøe, R., Gjertsen, B. T., Vintermyr, O. K., Houge, G., Lanotte, M., and Døskeland, S. O. (1991) Exp. Cell Res. 195, 237–246 4. Gjertsen, B. T., Cressey, L. I., Ruchaud, S., Houge, G., Lanotte, M., and Døskeland, S. O. (1994) J. Cell Sci. 107, 3363–3377 5. Li, D. W., Xiang, H., Mao, Y. W., Wang, J., Fass, U., Zhang, X. Y., and Xu, C. (2001) Exp. Cell Res. 266, 279 –291 6. Morimoto, Y., Morimoto, H., Kobayashi, S., Ohba, T., and Haneji, T. (1999) Oral Dis. 5, 104 –110 7. Chambers, T. C., Raynor, R. L., and Kuo, J. F. (1993) Int. J. Cancer 53, 323–327 8. Zheng, B., Chambers, T. C., Raynor, R. L., Markham, P. N., Gebel, H. M., Vogler, W. R., and Kuo, J. F. (1994) J. Biol. Chem. 269, 12332–12338 9. Sandal, T., Ahlgren, R., Lillehaug, J., and Døskeland, S. O. (2001) Cell Death Differ. 8, 754 –766 2

K. E. Fladmark and S. O. Døskeland, unpublished observations.

2811

10. Fladmark, K. E., Brustugun, O. T., Hovland, R., Boe, R., Gjertsen, B. T., Zhivotovsky, B., and Doskeland, S. O. (1999) Cell Death Differ. 6, 1099 –1108 11. Kikuchi, K., Shima, H., Mitsuhashi, S., Suzuki, M., and Oikawa, H. (1999) Int. J. Mol. Med. 4, 395– 401 12. Kawamura, T., Matsuzuwa, S., Mizuno, Y., Kikuchi, K., Oikawa, H., Oikawa, M., Ubukata, M., and Ichihara, A. (1998) Biochem. Pharmacol. 55, 995–1003 13. Spassova, M., Mellor, I. R., Petrov, A. G., Beattie, K. A., Codd, G. A., Vais, H., and Usherwood, P. N. R. (1995) Eur. Biophys. J. 24, 69 –76 14. Hook, S. S., and Means, A. R. (2001) Annu. Rev. Pharmacol. Toxicol. 41, 471–505 15. Ishida, A., Shigeri, Y., Tatsu, Y., Endo, Y., Kameshita, I., Okuno, S., Kitani, T., Takeuchi, M., Yumoto, N., and Fujisawa, H. (2001) J. Biochem. (Tokyo) 129, 745–753 16. See, V., Boutillier, A. L., Bito, H., and Loeffler, J. P. (2001) FASEB J. 15, 134 –144 17. Braun, A. P., and Schulman, H. (1995) Annu. Rev. Physiol. 57, 419 – 445 18. Hanson, P. I., and Schulman, H. (1992) Annu. Rev. Biochem. 61, 559 – 601 19. Soderling, T. R., Chang, B., and Brickey, D. (2001) J. Biol. Chem. 276, 3719 –3722 20. Wright, S. C., Schellenberger, U., Ji, L., Wang, H., and Larrick, J. W. (1997) FASEB J. 11, 843– 849 21. Blumenthal, D. K., and Stull, J. T. (1980) Biochemistry 19, 5608 –5614 22. Takebe, Y., Seiki, M., Fujisawa, J., Hoy, P., Yokota, K., Arai, K., Yoshida, M., and Arai, N. (1988) Mol. Cell. Biol. 8, 466 – 472 23. Hanson, P. I., Meyer, T., Stryer, L., and Schulman, H. (1994) Neuron 12, 943–956 24. Braun, A. P., and Schulman, H. (1995) J. Physiol. 488, 37–55 25. Mellgren, G., Vintermyr, O. K., and Døskeland, S. O. (1995) J. Cell. Physiol. 163, 232–240 26. Graessmann, M., and Graessmann, A. (1983) Methods Enzymol. 101, 482– 492 27. Vintermyr, O. K., Bøe, R., Bruland, T., Houge, G., and Døskeland, S. O. (1993) J. Cell. Physiol. 156, 160 –170 28. Gjertsen, B. T., Mellgren, G., Otten, A., Maronde, E., Genieser, H.-G., Jastorff, B., Vintermyr, O. K., McKnight, G. S., and Døskeland, S. O. (1995) J. Biol. Chem. 270, 20599 –20607 29. Srinivasan, M., Edman, C., and Schulman, H. (1994) J. Cell Biol. 126, 839 – 852 30. Foster, C. J., Johnston, S. A., Sunday, B., and Gaeta, F. C. (1990) Arch. Biochem. Biophys. 280, 397– 404 31. Ishida, A., Kameshita, I., Okuno, S., Kitani, T., and Fujisawa, H. (1995) Biochem. Biophys. Res. Commun. 212, 806 – 812 32. Ishida, A., Shigeri, Y., Tatsu, Y., Uegaki, K., Kameshita, I., Okuno, S., Kitani, T., Yumoto, N., and Fujisawa, H. (1998) FEBS Lett. 427, 115–118 33. Døskeland, A. P., Vintermyr, O. K., Flatmark, T., Cotton, R. G. H., and Døskeland, S. O. (1992) Eur. J. Biochem. 206, 161–170 34. Mellgren, G., Bruland, T., Døskeland, A. P., Flatmark, T., Vintermyr, O. K., and Døskeland, S. O. (1997) Endocrinology 138, 4373– 4383 35. Connelly, P., Sisk, R., Schulman, H., and Garrison, J. (1987) J. Cell Biol. 262, 10154 –10163 36. Rich, R. C., and Schulman, H. (1998) J. Biol. Chem. 273, 28424 –28429 37. Saelens, X., Kalai, M., and Vandenabeele, P. (2001) J. Biol. Chem. 276, 41620 – 41628 38. Sabourin, L. A., Tamai, K., Seale, P., Wagner, J., and Rudnicki, M. A. (2000) Mol. Cell. Biol. 20, 684 – 696 39. Hubbard, M. J., and Cohen, P. (1993) Trends Biochem. Sci. 18, 172–177 40. Damer, C. K., Partridge, J., Pearson, W. R., and Haystead, T. A. J. (1998) J. Biol. Chem. 273, 24396 –24405 41. Ikebe, M., and Reardon, S. (1990) J. Biol. Chem. 265, 8975– 8978 42. Mills, J. C., Stone, N. L., Erhardt, J., and Pittman, R. N. (1998) J. Cell Biol. 140, 627– 636 43. Cohen, O., Feinstein, E., and Kimchi, A. (1997) EMBO J. 16, 998 –1008 44. Kawai, T., Nomura, F., Hoshino, K., Copeland, N. G., Gilbert, D. J., Jenkins, N. A., and Akira, S. (1999) Oncogene 18, 3471–3480 45. Coleman, M. L., Sahai, E. A., Yeo, M., Bosch, M., Dewar, A., and Olson, M. F. (2001) Nat. Cell Biol. 3, 339 –345 46. Sebbagh, M., Renvoize, C., Hamelin, J., Riche, N., Bertoglio, J., and Breard, J. (2001) Nat. Cell Biol. 3, 346 –352 47. Kwiatkowski, A. P., and King, M. M. (1989) Biochemistry 28, 5380 –5385 48. Krebs, E. G., and Graves, J. D. (2000) Adv. Enzyme Regul. 40, 441– 470