Dephosphorylation of the small heat shock protein Hsp27 in vivo by protein phosphatase 2A.

The phosphorylation of the Hsp27 complex is rapidly altered in MRC-5 cells when they are exposed to mitogens, cytokines, stress, or serine/threonine protein phosphatase inhibitors. Here we performed experiments to identify which cellular protein phosphatase (PP1, PP2A, or PP2B) is responsible for the in vivo phosphorylation/dephosphorylation of Hsp27. In their purified forms, PP2A dephosphorylates Hsp27 more effectively than PP2B, whereas PP1 is weakly active. Measurements of enzyme activity of lysates derived from inhibitor-treated cells indicated that Hsp27 phosphatase activity is equally sensitive to okadaic acid (PPI/PP2A inhibitor) and cyclosporin (PP2B inhibitor) and that both okadaic acid and cyclosporin treatment inhibited Hsp27 phosphatase activity additively. Together the in vitro data suggest that both PP2A and PP2B can dephosphorylate Hsp27. However, the phosphorylation of Hsp27 in vivo is only affected when cells are treated with PP1 and PP2A inhibitors (okadaic acid, calyculin A) or cantharidin (PP2A inhibitor), but not the PP2B inhibitor, cyclosporin A, suggesting PP2A to be the main enzyme dephosphorylating Hsp27 in the cells. Purification and immunoblotting of Hsp27 phosphatase from MRC-5 cells also suggest it to be PP2A and not PP1 or PP2B. The ability of PP2A to dephosphorylate Hsp27 is shown to be regulated by the phosphorylation state of PP2A itself.

The phosphorylation of the Hsp27 complex is rapidly altered in MRC-5 cells when they are exposed to mitogens, cytokines, stress, or serinelthreonine protein phosphatase inhibitors. Here we performed experiments to identify which cellular protein phosphatase (PPl, PP2A, or PP2B) is responsible for the in vivo phosphorylation/ dephosphorylation of Hsp27. In their purified forms, PP2A dephosphorylates Hsp27 more effectively than PPBB, whereas PP1 is weakly active. Measurements of enzyme activity of lysates derived from inhibitortreated cells indicated that Hsp27 phosphatase activity is equally sensitive to okadaic acid (PPVPP2A inhibitor) and cyclosporin (PP2B inhibitor) and that both okadaic acid and cyclosporin treatment inhibited Hsp27 phosphatase activity additively. Together the in vitro data suggest that both PP2A and PP2B can dephosphorylate Hsp27. However, the phosphorylation of Hsp27 in vivo is only affected when cells are treated with PP1 and PP2A inhibitors (okadaic acid, calyculin A) or cantharidin (PP2A inhibitor), but not the PP2B inhibitor, cyclosporin A, suggesting PP2A to be the main enzyme dephosphorylating Hsp27 in the cells. Purification and immunoblotting of Hsp27 phosphatase from MRC-5 cells also suggest it to be PP2Aand not PP1 or PP2B. The ability of PP2A to dephosphorylate Hsp27 is shown to be regulated by the phosphorylation state of PP2A itself.
Treatment of eukaryotidprokaryotic cells by higher than normal growth temperature, anoxia, glucose deprivation, or free radicals is associated with an increased synthesis of a family of proteins known as the heat shock proteins (Hsps)' (1-5). They are divided into groups on the basis of their molecular weights (5). Hsp27.belongs to the small molecular weight heat shock proteins which include a-crystallin and the murine analog of Hsp27, Hsp25 (4). Hsp27 is constitutively expressed a t low levels in the cytosol of most human cells (5). In Drosophila, the small Hsps are differentially regulated during development. In Drosophila (6) and certain mammalian cell lines (7) their synthesis is induced by steroid hormones. Hsp27 expression is also reported to correlate with the growth and differentiation of human endometrial and cervical tumors ( 8 ) and with several * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The identification of kinases and phosphatases responsible for the phosphorylation of Hsp27 should provide insights into the cell's response to stress as well as cytokine signaling events. The amino acid sequence of Hsp27 shows that 21% of the protein has potentially phosphorylatable sites (1). TNF, arsenite, and heat shock increase the phosphorylation of serine within a similar amino acid sequence motif where an arginine residue is 3 amino acid residues upstream from the serine residue (RXXS) (27,31). This "signature" is found in three sites in human Hsp27 (Ser15, S e P , and Ser 8 2 ) and in two sites in the murine Hsp25 equivalent (Ser15 and Ser 82). The RXXS sequence motif is also the putative recognition site for several kinases (pp7OS6 kinase (pp70 rsk) and calmodulin kinase I1 (35)). However, recent reports suggest that the Hsp27 kinase to be a novel kinase that does not bear identity with either pp70 or pp90 rsk. (36). Another kinase, MAPKAP kinase 2, is phosphorylated and activated by MAP kinase. MAPKAP kinase 2 catalyzes the phosphorylation of the murine Hsp25 in vitro (38) and recognizes the amino acid sequence LXRXXS, with leucine being essential (38). Hsp27 has two such recognition sites. MAPKAP kinase 2 is also reported to be a specific kinase for the small molecular weight group of Hsps (39).
Relatively little is known of the cellular phosphatase(s) responsible for the dephosphorylation of Hsp27 which acts in concert with a kinase(s) to specify the net phosphorylation of Hsp27 in the cell (40). Recent in vitro studies showed that a calciudcalmodulin-dependent protein phosphatase PP2B (calcineurin) is involved in the dephosphorylation of the murine small heat shock protein Hsp25 in Ehrlich ascites cells (41). This raises the possibility of calcineurin involvement in Hsp27 phosphorylation (42). In the event, phosphorylation of small molecular weight Hsps are potentially regulated by two pathways: the MAP kinase and the calcium/calmodulin/calcineurin pathway.
In this paper, we compare the in vitro and the in vivo phosphorylatioddephosphorylation of Hsp27 by PP1, PP2A, and PP2B and their respective inhibitors. Clearly the in vitro data show that PP2A as well as PP2B dephosphorylate Hsp27. However, the in vivo data suggest PP2A to be the only enzyme that is primarily responsible for the dephosphorylation/ phosphorylation of Hsp27 in MRC-5 cells. Furthermore, when the PP2A catalytic subunit is phosphorylated on tyrosine, it no longer dephosphorylates Hsp27. Together, the data suggest that PP2A is the cellular phosphatase for Hsp27 and that the phosphorylation of Hsp27 is controlled by the opposing action of specific kinases and PP2A. The present data are consistent with the hypothesis that a protein phosphatase recognizing cellular Hsp27 and other phosphoproteins is inactivated during IL-lfI'NF signal transduction (29,40,43,44,46).

EXPERIMENTAL PROCEDURES
Cell Culture-MRC-5 fibroblasts (ATCC, Bethesda, MD) were maintained in modified Eagle's medium supplemented with 2 m~ glutamine, 100 unitsfml each of penicillin and streptomycin, and 10% fetal calf serum (HyClone Laboratories, Logan, UT) in a humidified atmosphere of 95% air and 5% CO, a t 37 "C. Confluent cells were used for all experiments.
Reagents-A 21-amino acid peptide containing two of the known sites of phosphorylation of the full-length Hsp27 (Ser78 and Sefl') was synthesized at the Institute of Molecular and Cell Biology, National University of Singapore. Recombinant Hsp27 and the monoclonal antibody to Hsp27 were purchased from StressGen Biotechnologies Corp, (Victoria, B. C., Canada). The following reagents were from Sigma: streptavidin-agarose, protein A-Sepharose, calcineurin, calmodulin, and 3,3'diaminobenzidine tetrahydrochloride and cantharidin. Okadaic acid was from Biomol (Plymouth Meeting, PA). Heparin-Sepharose was purchased from Pharmacia LKB Biotechnology, Inc. The hydroxysuccinimide ester of biotin was purchased from Pierce Chemical Co. CAMPdependent protein kinase A was from Promega (Madison, WI), and [y-32P]ATP and [y-35S]thio-ATP were from DuPont NEN. Docosohexaenoyl okadaic acid (DOA) was purchased from LC Services Corp, Woburn, MA. Purified PP1, PP2A, and PP2B, and the polyclonal antibodies recognizing the catalytic subunits of PP1, PPBA, and PP2B, respectively, were obtained from UBI, Lake Placid, N Y . Additional samples of purified phosphatases PP1 and PP2A were generous gifts from Dr. J. Goris, Afdeling Biochemie, Campus Gasthuisberg, Leuven, Belgium. Cyclosporin was a generous gift from Drs. T. Payne and E.
Metabolic Labeling a n d Lysis ofMRC-5 Cells-Cells were plated onto 90-mm plastic tissue culture dishes and grown to confluence prior to labeling with [32Plorthophosphate as described previously (17). Briefly, the cells were washed with serum-free medium (10 m~ Tris, pH 7.4, containing 150 mM NaCI, 5 m~ MgCl,, 5 m~ KCI, 1.6 m~ CaCl,, 0.5% glucose) before incubation with 0.5 mCi/ml [32Plorthophosphate for 3 h at 37 "C. In some experiments, agonists were added for the last 15 min of the incubation. The cells were then washed with ice-cold phosphatebuffered saline and lysed in 200 pl of solubilization buffer (10 m~ Tris-HC1, pH 7.4, containing 50 m~ EDTA, 50 mM NaC1, 30 m~ sodium pyrophosphate, 50 m~ NaF, 100 p~ sodium orthovanadate, 0.65% Nonidet P-40, 2 m~ leupeptin, and 2 m~ phenylmethylsulfonyl fluoride). The lysates were centrifuged for 5 min at 14,000 rpm, and the soluble fraction was lyophilized prior to the addition of sample loading buffer (59.7 g of urea in 44.9 ml of water containing 5.5 g of ampholytes (pH range 3-10), 1.45 g of dithiothreitol and 300 pl of 1% bromphenol blue) for first-dimension separation. The protein content of the extract was assayed using a BCA assay kit (Pierce).
no-dimensional Gel Electrophoresis-12 pg of the cell extract was subjected to isoelectric focusing for 18,000 V/h with pH 3-10 ampholytes using the Millipore Investigator two-dimensional electrophoresis system (Millipore, Bedford, MA). After isoelectric focusing, the gels were separated on the second dimension of 12.5% SDS-polyacrylamide gels using constant voltage. The gels were then dried, and 32P polypep-tides were located by autoradiography at -80 "C. The phosphoproteins were analyzed on a Visage 2000 Image system (BioImage Products, Ann Arbor, MI). Preparation of Substrate Phosphoproteins-The Hsp27 peptide containing the sequence AAPAYSRALSRQLSSGVSEIR was biotinylated as follows. One mg of the peptide was dissolved i n 1 ml of 0.1 M NaHCO, buffer, pH 8.4, before the addition of 150 mg of the biotin ester in dimethyl sulfoxide. The mixture was then incubated for 2 h at room temperature before separating the free biotin from the peptide on a Sephadex G-15 column. The biotinylated peptide was eluted into 0.5-ml fractions using 50 m~ potassium phosphate, pH 6.7, containing 6 m~ MgCI,, and the fractions with the highest optical density readings at 280 nm were pooled for phosphorylation. The biotinylated peptide was phosphorylated for 24 h with 25 pCi of [-f2PlATP in the presence of 400 units of cAMP-dependent protein kinase A catalytic subunit. A preliminary experiment indicated that maximum phosphorylation of the peptide is achieved at 24 h. Streptavidin-agarose beads were then added to attach to the biotinylated peptide, and the complex was washed with cold phosphate-buffered saline until the counts in the wash were less than 100 cpm. The immobilized substrate was then stored at 4 "C for the phosphatase assay.
The full-length recombinant Hsp27 was also phosphorylated as above except that the protein was immunoprecipitated with Hsp27specific antibodies and protein A-Sepharose beads. The beads were washed and stored as described above for the Hsp27 peptide.
Determination of PPI a n d PP2A Activity-Phosphatase activity against the Hsp27 phosphopeptide or Hsp27 phosphoprotein was assessed by the liberation of 32P. Assays were conducted by adding 65 pl of total cell lysates or the column fractions to the 32P-labeled substrates immobilized on beads. In some experiments, purified phosphatase catalytic subunits of PP1 and PP2A were used to dephosphorylate the labeled substrates. Dephosphorylation reactions were conducted for 10 min at 30 "C in phosphatase assay buffer (5 m~ Tris-HCI, pH 7.0, 0.1 m~ EDTA, 0.1% 2-mercaptoethanol, and 1 mg/ml bovine serum albumin). The reactions were terminated with 1 p1 of 6 M HCl and immediately spinning the beads in a microcentrifuge. An aliquot of 50 pl of the supernatant was then removed and counted for 32P released. Inhibition of phosphatase activity by okadaic acid and its derivatives was determined by incubating the indicated compound with the enzyme for 15 min prior to the addition of labeled substrate. Where recombinant Hsp27 was used as the phosphoprotein substrate, the dephosphorylated Hsp27 was then eluted off the beads by sonicating briefly in sample loading buffer (detailed above) and analyzed by two-dimensional gel electrophoresis a s described above. The elution protocol was rigorously tested to avoid any artifacts by comparing the isoforms of 32P-labeled Hsp27 that was eluted from immunobeads with that from cytosolic extracts. No differences were observed in any of the multiple Hsp27 isoforms using this protocol.
The activity of the purified PP1 and PP2A was also determined using a kit commercially available from Life Technologies, Inc. Following the manufacturer's instructions, phosphorylase A was labeled with [y-32PlATP and used as the substrate. Purified PP1 or PP2A was then added to the radiolabeled substrate in the absence or presence of either okadaic acid or its derivatives and incubated at 30 "C for 10 min. The reaction was terminated by the addition of 20% trichloroacetic acid. After centrifugation, the supernatants containing released szP were counted in a Beta Counter (LKB, Wallac, Finland). Determination of Calcineurin Activity-A 19-amino acid peptide with the sequence DLDVPIPGRFDRRVSVAAE-NH2 was synthesized (Multiple Peptide Systems) and used as the substrate for the assay. The peptide was biotinylated and phosphorylated using protein kinase A as described above for the Hsp27 peptide. After phosphorylation, the biotinylated peptide was precipitated with streptavidin-agarose beads, and the immobilized peptide was then used as the substrate to assay for calcineurin activity.
In each experiment, identical volumes of phosphorylated substrate immobilized on beads were aliquoted and counted prior to the assay to ensure that the same amount of phosphorylated substrate was used for each reaction. In all dephosphorylation reactions, the measurement was done during the linear phase of labeled phosphate release. Confluent MRC-5 cells were either left untreated or treated with 100 nM cyclosporin A for 1 h a t 37 "C. The cells were then lysed in 400 pl of a 20 m~ Tris-HC1 buffer, pH 8.0, containing 100 m~ NaCI, 6 m~ MgCl,, 0.5 m~ dithiothreitol, 0.1 m~ CaCl,, and 0.1% Nonidet P-40, and the lysates were cleared of insoluble material by centrifugation at 14,000 rpm. To initiate the assay, 65 pl of the lysate was added to the phosphopeptide substrate in the presence of 25 units of calmodulin and incubated for 10 min at 30 "C, after which the reaction was terminated as before, and an aliquot of 50 pl of the supernatant was removed to count for 32P released. I n some assays, 500 m okadaic acid was added to the cyclosporin A-treated cell lysates for 10 min prior to adding the mixture to the peptide substrate to initiate the reaction.
I n Vivo Effect of Cyclosporin A on Dephosphorylation of Hsp27-MRC-5 cells, prelabeled with 32P, were treated with 100 n~ cyclosporin A for 30 min or 60 min after which the cells were washed and prepared for two-dimensional gel electrophoresis, as described above. Separation of Phosphatase Catalytic Subunits from MRC-5 Cytosolic Extracts a n d Location by Western Blotting-Six large flasks of MRC-5 cells were solubilized in sample loading buffer (see above), applied to a heparin-Sepharose column in 20 m~ Tris-HC1, pH 7.4, 0.1 m~ EDTA, and 0.1% Nonidet P-40, and then later eluted with 0.5 M NaCl. The resulting fractions were divided equally, and each fraction was concentrated by lyophilization. One fraction was boiled with Laemmli buffer and run on three identical 10% SDS-PAGE minigels. The gels were then electrophoretically transferred to Immobilon-P, and each was immunoblotted with polyclonal antibodies specific for PP1, PPSA, or PP2B. The immunoreactivity of the primary antibody was visualized after incubation with a secondary anti-rabbit antibody conjugated with horseradish peroxidase and subsequent development with 3,3'-diaminobenzidine tetrahydrochloride. The other fraction was assayed for its activity on immunoprecipitated or recombinant Hsp27.
Assay for Hsp27 Phosphatase Using Immunoprecipitated Hsp27-MRC-5 cells were labeled for 3 h with 32P a s described above. To increase the phosphorylation of the more acidic isoforms of Hsp27 the cells were treated with 100 unitdml IL-16 for 15 min prior to cell lysis, The cells were lysed, Hsp27 was immunoprecipitated, and the immunoprecipitates were washed extensively prior to be used as the substrate for the putative Hsp27 phosphatase. The phosphatase assay took place for 10-20 min a t 30 "C with the Hsp27 still attached to the immunobeads. The beads were divided into aliquots containing approximately 50,000 cpm, and the assay was performed as described for the recombinant protein. After the assay was terminated the Hsp27 protein was removed from the immunobeads by sonication in two-dimensional sample loading buffer, and each sample was subjected to two-dimensional electrophoresis, autoradiography, and image analysis. Comparative experiments were performed previously comparing the immunoprecipitated pattern of Hsp27 phosphorylation with that from cytosolic extracts to show that the phosphorylation pattern was not altered in any way during the immunoprecipitation and subsequent elution processes.
Qrosine Phosphorylation of PP2A a n d Subsequent Assay of Activity-Purified PPZA catalytic subunit was phosphorylated by srckinase for 1 h essentially as described by Chen et al. (47) using either [y-32PlATP or [y-35Slthio-ATP. Hsp27 immunoprecipitated from 32P-labeled, IL-1-treated MRC-5 cells was then added to the reaction mixture, which contained thiophosphate-labeled PP2&=,, for a further 30 min at 30 "C. At the end of this time the whole mixture was denatured in urea containing sample buffer and subjected to two-dimensional electrophoresis, autoradiography, and image analysis to assess the changes in Hsp27 isoform phosphorylation.
Sequence Analysis of Substrate Proteins-Hsp27 that was phosphorylated in vitro by various kinases was separated from the kinase on SDS-PAGE gels. Identical Coomassie Blue-stained bands from dried gels were excised with a scalpel and subjected to in gel proteolytic digestion using trypsin. Excised spots were reswollen in 200 m~ ammonium bicarbonate buffer, pH 8.9, containing 50% acetonitrile. A 2-pl aliquot of trypsin (modified trypsin, Promega) a t a concentration of 0.01 pg/pl was added to each gel piece and allowed to soak into the matrix. This was repeated a second time, after which similar volumes of 200 m~ ammonium bicarbonate buffer, pH 8.9, were added until each gel piece had reswollen to its original size. Gel pieces were then placed into an Eppendorftube, covered with the ammonium bicarbonate buffer (-200 pl), and incubated at 37 "C for 44 h. After incubation the buffer was transferred to a clean tube and pooled with two succesive 30-min washes of 0.1% trifluoroacetic acid containing 60% acetonitrile. These washes were then lyophilyzed. The residue was resuspended in 0.1% trifluoroacetic acid and injected onto a C18 narrow bore (2.1 x 250 mm) reverse phase HPLC column (Vydac, Hesperia, CA). Elution and subsequent separation of the peptides were performed by subjecting the column to a gradient of increasing acetonitrile concentration in 0.1% trifluoroacetic acid. Eluting peptides were monitored at 214 nm and collected into tubes manually.
Subsequent sequence analysis of the purified peptides was carried out using a Milligen 6600 ProSequencer (Millipore). Reverse phase HPLC was used to separate the tryptic peptides using a Waters 600 pump, M990 diode array detector, and incorporating the following pa-rameters. Column size was 2.1 x 250 mm C18 (Vydac). Buffer A was 0.1% trifluoroacetic acid in Milli Q water. Buffer B was 0.08% trifluoroacetic acid in acetonitrile (far UV grade). The elution gradient was linear from 2% B to 60% B over 100 min. The flow rate was 200 pYmin, and the detection was a t 214 nm. The fractions were collected manually and counted for Cerenkov radiation using a liquid scintillation counter.
Sequence AnalysisSequence analysis of labeled peptides was done using a model 6600 ProSequencer (Millipore). Trifluoroacetic acid was used as the ATZ-amino acid extraction solvent on this instrument, which is advantageous during the extraction of the polar phosphoamino acids. The program used  splits the reconstituted phenylthiohydantoin derivative so that 50% can be collected for counting while the remaining 50% is injected into the on-line HPLC system for amino acid determination.

Effect of Cell-permeant Phosphatase Znhibitors on the Phosphorylation of Hsp27 in MRC-5 Fibroblasts-High resolution
two-dimensional gel electrophoretic analysis of early protein phosphorylation in MRC-5 cells treated with TNF/IL-1 or OA indicated that the Hsp27 complex was significantly hyperphosphorylated (29). In this study, MRC-5 cells were treated with two other phosphatase inhibitors unrelated to OA and derived from different sources. Calyculin A inhibits PP2A and PP1, and cantharidin binds the A and C subunits of PP2A (45). The results show that these inhibitors induce a pattern of phosphorylation of cellular proteins in MRC-5 fibroblasts similar to that which is typically induced by OA (Fig. 1). Calyculin A and cantharidin, like OA, induce the hyperphosphorylation of the Hsp27 complex. Treatment of MRC-5 cells with OA, cantharidin, or calyculin A (Fig. 1, b , d , and f ) produces the phosphorylation of at least four acidic Hsp27 isoforms (Fig. 1, which exist as a single species in control cells (Fig. 1, a , c , and e), Experiments were performed t o show which protein phosphatase is the most likely target of these inhibitors or TNF/IL-1 as reported in the next sections.

Dephosphorylation of Recombinant Hsp27 Phosphorylated by Protein Kinase A in Vitro by the Catalytic Units of Purified PPI, PP2A
, and PP2B-A number of protein kinases (MAP-2 kinase, ~3 4 "~"~, casein kinase 2, and the catalytic unit of CAMP kinase) were tested to see which of the kinases was most effective in phosphorylating Hsp27 in vitro. The catalytic subunit of CAMP kinase was the most effective, hence it was used to produce [32PlHsp27 for use in determining which protein phosphatase is the most effective in dephosphorylating Hsp27 in vitro. None of the other kinases catalyzed a significant phosphorylation of Hsp27 in vitro. When CAMP kinase was used in the in vitro phosphorylation of Hsp27 three major isoforms were separated on two-dimensional gels. These were compared with "native" Hsp27 either by comparing the Hsp27 from immunoprecipitations or by corunning the CAMP kinase-catalyzed product with cytosolic extracts of 32P-labeled fibroblasts. The major isoforms ran at the same PI on two-dimensional gels, and microsequencing analysis showed that serines 15,78, and 82 were the major phosphorylated sites (data not shown) as has been demonstrated for other agonists (27,31).
In comparison with PP1 and PP2B, PP2A was most effective in the dephosphorylation of Hsp27 ( Fig. 2A). OAand DOA(both inhibitors of PP2A) blocked or partially blocked the dephosphorylation of Hsp27 by PPSA, whereas the PP1 inhibitor 2 did not block either PP2A or PP2B (data not shown). Cyclosporin could not be used as a n in vitro inhibitor as it forms a complex with cyclophilins in vivo to achieve inhibiton of PP2B (42).
The dephosphorylation of [32PlHsp27 in vitro was also analyzed by two-dimensional electrophoresis (Fig. 2B ). In the controls, [32P]Hsp27 is shown t o consist of at least four isoforms, similar to those observed in vivo (Fig. 1). PP2A preferentially dephosphorylates these phosphorylated isoforms of Hsp27 and is sensitive to OA. PP1 did not appear to cause any significant

FIG. 1. A u t o r a d i o g r a p h s of "2P-la-
beled cellular proteins from h u m a n fibroblasts stimulated with various protein phosphatase inhibitors. MRC-5 cells wrrr prelahrlrd with I:'"Plorthophosphatr (0.5 mCi/ml) for 3 h. Okadaic acid, calyculin A, or cantharidin was added to crlls for the final 30 min of labeling. The cytosolic extracts were suhjrcted to two-dimrnsional gel electrophoresis as descrihed under "Experimental Procedures." Each pair was run in different rxperimcnts, therefore each sample treatrd with a phosphatase inhihitor has its o w n control.  dephosphorylation (less than 5% of control), suggesting that PP1 is not involved in Hsp27 dephosphorylation. Calcineurin (PPSB), in the presence of Ca2+ and calmodulin, dephosphorylated Hsp27 by 23%. albeit to a lesser extent than PP2A but suggesting that PP2B can also dephosphorylate Hsp27. Dephosphorylation of P'2PIHsp27 Immunoprecipitated from MRC--5 Cells by the Catalytic Subunits of Purified PPI, PP2A, and PP2B in Vitro-Attempts using more native phosphorylated Hsp27 substrates were made. MRC-5 cells were phosphorylated in vivo, and ["2PIHsp27 was immunoprecipitated from extracts of MRC-5 cells treated with I G l p to stimulate the phosphorylation of Hsp27 (29). The immunoprecipitated Hsp27 was subjected to dephosphorylation by purified phosphatases while immobilized on Sepharose beads. The radioactive phosphate released was collected and counted, and the results were consistent with those shown in Fig. 2.4 (data not  shown). The substrate was then eluted from the beads and analyzed by two-dimensional gel electrophoresis. The results of this analysis show that PP2A (Fig. 3h) was most active (apparent removal of 90% of :I2P from Hsp27), PP2B (Fig. 3d) had minimal activity (15%), and PPI (Fig. 3c) was the least active (less than 6%) of the three phosphatases in the dephosphorylation of the immunoprecipitated [:'2PIHsp27 complex.
Immunoprecipitation of Hsp27 Phosphatase in Cell Lysates a p e r n e a t m e n t of Cells with Phosphatase Inhibitors-Having shown PP2A and PP2B to dephosphorylate Hsp27 in vitro (Figs. 2 and 3) the activity of the three protein phosphatases was compared in the extracts of MRC-5 cells, using [.12PIHsp27 as substrate. Living MRC-5 cells were treated with phosphatase inhibitors. 500 nw OA inhibited Hsp27 phosphatase activity by 38V (Fig. 4). Since this concentration of OA is known to inhibit 90-9.W of PP2A activity, other OA-insensitive phosphatases such as calcineurin (PP2R) may be involved in the dephosphorylation of Hsp27 as well. The experiment was repeated with cyclosporin A, an inhibitor of PP2R (42). When cells were treated with cyclosporin, the cell extracts show a 34'7 inhibition of Hsp27 phosphatase activity. When cyclosporin Atreated cell lysates were additionally treated with 500 n y OA. the dephosphorylation of Hsp27 was inhihited by 7sr:. Various doses of both OA and cyclosporin were tested, and those used above were found to be optimal. Calyculin A a t 0.1 p~ gave effects similar tn those of OA. Cyclosporin cannot be used in vitro as an inhibitor of PP2R unless it is complexed to a low molecular weight cyclophilin protein.
This additive effect suggests that calcineurin and PP2A may both dephosphorylate Hsp27 in oiuo.
Effvct of Cylosporin A on the Phospho~ylation of Hsp27 Cnmplex in Vivo-MRC-5 fibrohlasts were treated with 100 nw cyclosporin for 30 or 60 min, the cells were lysed, and the lysates were analyzed by two-dimensional gel electrophoresis (Fig. 5 j. No change in the phosphorylation of the Hsp27 complex was observed as a consequence of cyclosporin A treatment. although changes in the phosphorylation of at least six other phosphoproteins were ohserved. This is in contrast tn the results ohtained in MRC-5 cells treated with OA. which causes a sipifi- following which the remaining substrate was released from the heads and run on two-dimensional gels. The resultant autoradiographs were analyzed for changes in the distribution of radioactivity in the Hsp27 isofoms. Panel A shows the radiolabeled phosphate released when labeled Hsp27 was treated with PPI (protrin phosphatase catalytic subunit), PP1 plus OA (okadaic acid a t 400 nw). PPPA (the catalytic suhunit of PPZA). PPPA plus OA 1400 nw), PP2A plus DOA (10 pw). PP2R (phosphatase 2R (calcineurin)) and PP2R plus OA. CON is the control, which consisted of the substrate with reaction buffer only. Panel R shows autoradiographs that resulted from running the recombinant.
:I2P-labeled Hsp27, on two-dimensional gels at the end of the experiment shown in panel A. cant increase in the phosphorylation of the Hsp27 complex (Fig. Id).
To investigate the possibility of PP2B involvement in the phosphorylation status of Hsp27 in vivo we adopted a pulsechase method to determine whether the inhibition of PP2B would prevent the dephosphorylation of Hsp27 (43). Cells were prelabeled with :j2P and activated with TNF-CY for 15 min to induce the phosphorylation of the Hsp27 complex in vivo. The treated cells were washed rapidly and chased with excess unlabeled phosphate. Various treatments were then applied to the TNF-treated cells subsequently divided into different groups; 100 nM cyclosporin was added to one group, 1 p~ OA to another, and nothing to the control group. After a 1-h chase the cells were extracted, [:j2P]Hsp27 was immunoprecipitated and analyzed by two-dimensional PAGE, and the gels were subjected to autoradiography. OA but not cyclosporin inhibited the dephosphorylation of Hsp27 during the cold phosphate chase. There was no difference between the resultant phosphorylation of the Hsp27 complex in control and cyclosporin-treated cells, suggesting that the inhibitor of PP2B has no effect in preventing the dephosphorylation of Hsp27 (not shown) and thus suggesting that PP2B is not primarily responsible for dephosphorylating Hsp27 in vivo at least in IL-1or TNF-treated cells.
It is unlikely that the results in Fig. 5A are due to cyclosporin not inhibiting PP2B in the MRC-5 cells because we tested the ability of cyclosporin to inhibit the activity of PP2R in MRC-5 cells using the synthetic 19-amino acid peptide substrate. Cyclosporin but not OA inhibited PP2R activity of MRC-5 cells (Fig. 5B), and PP2B can be completely inhihited hy treatment of MRC-5 cells with 100 nM cyclosporin A for 30 min.
The above results that demonstrate that the inhibition of PP2B by cyclosporin in vivo seem to disagree with the results from the previous section, where PP2R extracted in lysates from MRC-5 cells could dephosphorylate "2P-labeled Hsp27 in in vitro phosphatase assays, and this could be inhihitrd by Cell lysates from fibroblasts treated with cyclosporin 1100 n.rc) were assayed for their ability to catalyze the release of labeled phosphate from a PP2R suhstrate peptide. BUFFER consisted o f the peptide with assay huffer only. PP2R is purified PP2B plus calmodulin and calcium added to the labrled prptide. LYS. is the lysate from cells that were not treated with cyclosporin. CSA is the lysate from cells treated with cyclosporin (100 nM) for 30 min prior to lysis. OA is the lysates from cells treated with okadaic acid (500 n.rc) for 30 min prior to lysis. The experiment was performed in duplicate and was repeated twice.
pretreating the cells with cyclosporin. The most likely explanation for these observations is that Hsp27 and PP2R are not in the same cellular compartment, but the phosphatase has access to the substrate as a result of the extraction process.
Immunohlotting of Column Chromatographv-puriled Hsp27 Phosphatase Fractions-Cytosolic extract from multiple cultures of MRC-5 fibroblasts was fractionated on a heparin-Sepharose column in a fast protein liquid chromatography apparatus. The various fractions were assayed for phosphatase activity using recombinant [:'"P]Hsp27 as substrate. One major and one minor fraction of Hsp27 phosphatase activity were separated (Fig. 6A ). The fractions corresponding to the major and minor activities were concentrated and analyzed by Western blot. Fig. 6B shows that PP2A was detected in the major fraction, whereas PP1 and PP2B were in the minor fraction. Effect of q r o s i n e Phosphorylation on PP2A Activity in the Dephosphorylation of r12PIHsp27-Experiments were performed to address the question of the effect of phosphorylation of PP2A and its activity. The experiments that were performed were essentially adaptions to those carried out be Chen et al. (47). Purified PP2A was subjected to phosphorylation in vitro by pp60""". The phosphorylated enzyme was tested for protein PP2R. The Hsp27 phosphatase fractions rlutcd from t h r h r p a n n -Sepharose column wrrr concrntratrd and furthrr suhjrctrd to SDS-PAGE separation. Thr protrins on SIX-I'AGF: grls wrrr transfrrrrd to Immohilon-P membranes and blotted with pnlyclonal antihndirs against the catalytic suhunits f r w PJ'1. J'MA. or 1'1'2B. The Incatlrms of the phosphatasr catal.ytic suhunits wrrr indicatrd hv incuhatlng thr blots with a srcondary antibody conjugated to horscradlsh prroxldasr and further incuhation with 3,:~"dinminr)hcnzidinr trtrnhvdrtrhlrlrldr.. phosphatase activity using immunoprecipitated I""PIHsp27 as substrate. The treated enzyme as well as the dephosphorylated substrate were examined on two-dimensional electrophoresis gels. Autoradiographs of the PP2A incubated with pp60""" show that the catalytic subunit of PP2A is phosphorylated Fig.  7A ). Autoradiographs of the digested I""PIHsp27 substrates show the tyrosine-laheled PP2A (labeled with Iy-:'"PIthio-ATP to minimize autodephosphorylation) dephosphorylated Hsp27 by only 14% in comparison with the untreated control enzyme (Fig. 7R).

DISCUSSION
A redox-sensitive Hsp27 phosphatase was shown to be inactivated in TNF/ILl signal transduction (43). This phosphatase was inactivated in vitro by okadaic acid fan inhihitor of PPI and PP2A) which mimicked the effects of TNF/ILl on early protein phosphorylation, suggesting that either PPI and/or PP2A inactivation is an early event of TNF/IL-1 signaling 129. 43 i. expcrimrnt performrd with pp6W'" kinase only, showing a certain amount of autophosphorylation. ii, experiment performed with P2&,,, that was phosphorylated with pp60""' kinase. Thr location of the phosphorylated PP24,,, is indicated by an nrrorcrhmd. I'oncd 13, activity of PP2q.,,, following tyrosine thiophosphorylation. To avoid autodrphosphorylation, PP24.,,, was phosphorylntrd as descrihrd ahovc except I y-.'Tlthio-ATP was uscd as the source of a modified phosphate that is transferred to tyrosinr residues. The thiophosphorylatrd PP2&,,, was then incuhatcd with prrlahelcd Hsp27 from IL-1 stimulated crlls. I, the control :'2PP-lahrlcd Hsp27 immunoprecipitate that was incuhatrd with humer only and suhsrquently run on a two-diii. autoradiograph of the two-dimensional mcnsional gel and exposed to x-ray film.
gcl separation of the immunoprecipitated H s p Z aftrr treatment with unphosphorylatrd PP2A for 80 min. iii, Hsp27 isoforms after PP2&.,,, was thiophosphorylatrd for 60 min prior to incuhation with V"laheled Hsp27. Thc rxprrimrnt was repeated thrrr times.
ii. analog of Hsp27) (41). Notably however, this conclusion was deduced only from in uitro data. In this paper, two lines of investigation utilizing both in vitro and in viuo methods of analysis were undertaken to evaluate which of the three major classes of protein phosphatases is the main target enzyme for the cellular Hsp27 complex during TNF/IL-1 signaling. A comparison of the dephosphorylation of [:"PIHsp27 (recombinant Escherichia coli-derived and immunoprecipitated from MRC-5 cells) by the various enzymes shows that PP2A dephosphorylated Hsp27 much more effectively than PPBB, whereas PP1 was only weakly active (Figs. 2  and 3). Examination of Hsp27 phosphatase activity from cell lysates indicated that the activity was equally sensitive to OA or cyclosporin (an inhibitor of PP2B). and OA and cyclosporin inhibited Hsp27 phosphatase in an additive manner (Fig. 4). The data suggest that both PP2A and PP2B could be responsible for the dephosphorylation of cellular Hsp27 in MRC-5 cells. However, a comparison of the early protein phosphorylation changes in the I''"PIHsp27 complex of MRC-5 cells treated with OA, calyculin A (both inhibitors of PP1 and PP2A), cantharidin (inhibitor of PP2A), and cyclosporin shows that only inhibitors of PP2A (OA, calyculin A, and cantharidin) but not the PP2R inhihitor (cyclosporin) mimicked TNFAL-1 in the hyperphosphorylation of the Hsp27 complex in MRC-5 cells (Figs. 1 and 5 ) . Thus, the in vivo but not in vitro data suggest that PP2A is the enzyme that dephosphorylates Hsp27 during TNF/IL-1 signaling. Partial purification and immunoblotting of Hsp27 phosphatase derived from MRC-5 cells also support this explanation (Fig. 6). The apparent anomaly between the ability of PP2R to drphosphorylate Hsp27 in cell lysates but not in whole crlls could be explained by the enzyme and its substratr hring in diffrrrnt cellular compartments. Although we havr no direct proofof this the data presented are consistrnt with this hypothrsis.
It should also be noted that we havr usrd thr catalytic suhunits of various phosphatases in this work. Thrre is e m r r~n g evidence that the A and R suhunits of phosphatasr 2A can rfTrct the substrate specificity in diffrrrnt ways ( 3 4 1 . Furthermore, the specificity of phosphatases to crrtain suhstratrs is conferred via adaptor protrins (50). and thew would not hr prrsrnt in vitro assays using recombinant protein suhstratrs. With such limitations in phosphatase assays a rolr for PPI in t h r in vivo dephosphorylation of Hsp27 in such circumstnncrs cannot be totally excluded.
A t least two protein kinases arr implicated in t h r phosphorylation of cellular Hsp27 (37-39). One of them. MAPKAI' kinase 2, which recognizes the sequence f3H), has a restricted number of substrate proteins including Hsp27. C/ERPfl, glycogen synthase, and laminin (according to analysis of current, protein sequence data hases). Anothrr kinasr sprcific for Hsp27 and stimulated hy TNF and I L I was rrcrntly described in MRC-5 cells (37). and it is possihle that it is the same as MAPKAP kinase 2. Considering that thr phosphorylation of cellular proteins is maintained in homrostnsis by the protein phosphatases activrly opposing kinases (40). wr suggest that both MAPKAP 2 and MAP kinases and PP2A arr closely linked in the phosphorylation of Hsp27 according to thr scheme in Fig. 8. Activation of the kinasrs or inactivation of t h r