Dephosphorylation of the Small Heat Shock Protein hsp25 by Calcium/ Calmodulin-dependent (Type 2B) Protein Phosphatase*

The dephosphorylation of the mouse small heat shock protein hsp25 within an extract obtained from Ehrlich ascites tumor cells is inhibited by the calcium chelator EGTA and at concentrations of microcystin-LR which are characteristic for inhibition of calcium/calmodulin- dependent (2B type) protein phosphatases. Further-more, the dephosphorylation of hsp25 in the cell-free system derived from Ehrlich ascites tumor could be increased specifically by addition of the calcium/cal-modulin-dependent (2B type) protein phosphatase cal- cineurin. Dephosphorylation of the heat shock protein hsp25 is also obtained in an in vitro system containing phosphorylated recombinant hsp25, 1 mM Ca2', cal- modulin, and calcineurin specifying hsp25 as the direct substrate for this enzyme. The expression of two iso- forms of the catalytic subunit of the mouse calcium/ calmodulin-dependent (2B type) protein phosphatases in Ehrlich ascites tumor cells is demonstrated by polymerase chain reaction using specific oligonucleotide primers to the catalytic and calmodulin-binding do- main, respectively. Northern blot analysis using the amplified fragments as probes shows that the mRNA of one isoform of the mouse calcium/calmodulin-de-pendent protein phosphatase is of medium abundance in EAT cells. These data suggest a calcium/calmodulin-dependent

The dephosphorylation of the mouse small heat shock protein hsp25 within an extract obtained from Ehrlich ascites tumor cells is inhibited by the calcium chelator EGTA and at concentrations of microcystin-LR which are characteristic for inhibition of calcium/calmodulindependent (2B type) protein phosphatases. Furthermore, the dephosphorylation of hsp25 in the cell-free system derived from Ehrlich ascites tumor could be increased specifically by addition of the calcium/calmodulin-dependent (2B type) protein phosphatase calcineurin. Dephosphorylation of the heat shock protein hsp25 is also obtained in an in vitro system containing phosphorylated recombinant hsp25, 1 mM Ca2', calmodulin, and calcineurin specifying hsp25 as the direct substrate for this enzyme. The expression of two isoforms of the catalytic subunit of the mouse calcium/ calmodulin-dependent (2B type) protein phosphatases in Ehrlich ascites tumor cells is demonstrated by polymerase chain reaction using specific oligonucleotide primers to the catalytic and calmodulin-binding domain, respectively. Northern blot analysis using the amplified fragments as probes shows that the mRNA of one isoform of the mouse calcium/calmodulin-dependent protein phosphatase is of medium abundance in EAT cells. These data suggest a calcium/calmodulindependent dephosphorylation of the small stress protein in EAT cells also in vivo.
Since it is known that heat shock increases the intracellular calcium level and that thermotolerance is influenced by calcium chelators, ionophores, and anticalmodulin drugs, the changes in the degree of hsp25 phosphorylation induced by thermal stress resulting in a n altered thermoresistance could be explained at least partially by the calcium/calmodulin-dependent dephosphorylation through protein phosphatases 2B.
The small (or low molecular weight) heat shock proteins (hsps)' are a class of heat shock proteins which is, compared to the hsp7O and hsp90 classes, relatively heterogeneous between different organisms (1). However, in mammalians, there is evidently a single small heat shock protein (termed * This work was supported by Grants Ga 453/2-1 and SFB 344 YE1 from the Deutsche Forschungsgemeindschaft and Grant 0310172A from the Bundesministerium fur Forschung und Technologie. 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. 8 To whom correspondence and reprint requests should be addressed. The abbreviations used are: hsp, heat shock protein; EAT, Ehrlich acsites tumor; hsp25, murine small heat shock protein; hsp27, human small heat shock protein; PAGE, polyacrylamide gel electrophoresis; kh, kilobase(s); bp, base pair(s). hsp27 or hsp28 in human and hamster, hsp25 in mouse) which is remarkably conserved in amino acid sequence, especially within the C-terminal half of the molecule. Interestingly, this part of the small hsps is also structurally related to vertebrate a-crystallin (2) and probably responsible for the aggregate forming properties of both the small hsps (3) and the acrystallins (4). The mammalian small hsps are modified by post-translational phosphorylation a t different serine residues (5,6). The phosphorylation state of the mammalian small hsps is changed in dependence on the action of serum factors ( 7 ) , tumor necrosis factor-a (8,9 ) , and tumor promoters (10) as well as in the dependence on the stress conditions (11,12).
It has been shown that the murine small heat shock protein hsp25 is phosphorylated in a cell-free system derived from Ehrlich ascites tumor (EAT) cells at the same sites (serines 15 and 86) as in the EAT cells in uiuo (5). The degree of hsp25 phosphorylation within this cell-free system shows a maximum of hsp25 phosphorylation at about 20 min of incubation. After this time, dephosphorylation of hsp25 surpasses phosphorylation and results in a decreasing degree of phosphorylation of hsp25 (13).
In this report, we characterize the phosphatase activity from EAT cells which is responsible for specific dephosphorylation of hsp25, show that dephosphorylation of hsp25 can be obtained by direct action of calcineurin, and provide evidence for physiological relevance of dephosphorylation of hsp25 by calmodulin-dependent (2B type) protein phosphatases in EAT cells.

Preparation of the Cell-free Phosphorylation/Dephosphorylation
System from EAT Cells and Dephosphorylation Assay-The cell-free system was prepared as described in Ref. 5. The dialyzed cell-free EAT extract containing hsp25, the appropriate hsp25 kinases, and phosphatases in 20 mM HEPES/KOH, pH 7.4, was incubated in the presence of 5 mM MgCl,, 1 mM CaC12, 0.5 mM ATP, and 10 FCi of [y-"'P]ATP (3000 Ci/mmol) a t 37 "C for 30 min. In this time interval, a phosphorylation and appropriate radioactive labeling of hsp25 is obtained by the action of hsp25 kinases. Since the phosphatase activity within the cell-free system surpasses the kinase activity after about 20-30 min, the relative degree of phosphorylation of hsp25 after 30 min was set to he 100% and the incubation in the presence of EGTA (5 mM), microcystin-LR (4 nM-10 PM), and calmodulin/ calcineurin (0.25 unit/pl/0.5 unitlpl) was started. At the times indicated, aliquots were taken and boiled in SDS-PAGE loading buffer for inactivation of phosphorylating and dephosphorylating enzymes. Samples were subjected to SDS-PAGE, and the relative phosphorylation of hsp25 as a quotient between labeling of the hsp25 band and labeling of all proteins in the lane was determined as described below.
Quantification of the p'P/04-labeled Proteins-After separation of the proteins by SDS-gel electrophoresis, quantification of the radioactivity of any desired part of the gel was carried out using a Bio-Imaging Analyzer BAS 2000 (Fuji). The radiation dose of the desired area was proportionally converted into photostimulated luminescence and integrated using the BAS 2000 image analyses software.
Phosphorylation of Recombinant hsp25 and Purification of the Dephosphorylation of Recombinant Phosphorylated hsp25 by Calcineurin-About 20 pg of hsp25 were incubated in 20 I~M HEPES/ KOH, pH 7.4, 5 mM MgCl,, 1 mM CaCl,, 0.25 unit/pl calmodulin in the presence and absence of 0.5 unit/pl bovine brain calcineurin (Sigma). After different times of incubation at 37 "C, samples were processed and subjected to SDS-PAGE as described above.
cDNA Synthesis and Amplification of Sequences Coding for Protein Phosphatases 2s-After first strand synthesis from 5 pg of EATpoly(A) mRNA using oligo(dT)-primer and reverse transcriptase (Stratagene), the polymerase chain reaction was performed in a 100-p1 reaction with 10 ng of single-stranded template cDNA, 100 pmol  Northern Blot Analysis-Hybridization probes were generated from the cloned PCR fragments by random priming and labeling with [CX-~'P]ATP (3000 Ci/mmol) (Megaprime, Amersham). Poly(A)-rich mRNA was purified from EAT cells by the LiCl method and oligo(dT)-cellulose chromatography, separatedunder denaturing conditions in 1% agarose gels, and transferred to nitrocellulose as described (16). Hybridization was carriedout in 6 X SSC, 5 X Denhardt's reagent, 0.5% SDS, 50 pg/ml fragmented salmon sperm DNA at 60 "C overnight. Filters were washed three times for 10 min in 0.1 X SSC, 0.5% SDS at 60 "C and analyzed using the Bio-Imaging Analyser BAS 2000 (Fuji). The abundance of the mRNAs detected was estimated from the amount of hybridizing probe determined using BAS 2000 software taking into account the specific activity of the DNA probe (3.4 X lo9 dpm/pg) and the number of cells used to prepare poly(A)-rich mRNA applied in one slot (5 X lo5).

RESULTS
Dephosphorylation of hsp25 in the Cell-free EAT System-The phosphorylation and dephosphorylation of hsp25 in the cell-free EAT system was first carried out in the presence or absence of different concentrations of the phosphatase inhibitor microcystin-LR (14) to characterize the class(es) of phosphatases (15,24) responsible for the dephosphorylation of hsp25. The degree of hsp25 phosphorylation was determined by SDS-PAGE and subsequent quantification of 32P incorporated into the hsp25 band compared to the 32P incorporated into all proteins of the lane. Fig. 1 demonstrates the influence of increasing concentrations of microcystin-LR on the degree of hsp25 phosphorylation. The degree of phosphorylation is not increased using microcystin-LR concantrations of 4 nM and 40 nM which totally inhibit protein phosphatases of the type 1 and 2A (14). Significant increase of hsp25 phosphorylation is obtained with a microcystin concentration of 1 KM which is known to inhibit calmodulin-dependent (2B type) protein phosphatases but not protein phosphatases of the 2C type. The concentration dependence of inhibition of hsp25 dephosphorylation by microcystin indicates that calmodulin-hsp25 phosphorylation / protein phosphorylation (%) Further characterization of the protein phosphatase activity within the cell-free EAT system was done by determining the influence of the calcium/calmodulin-dependent (2B type) protein phosphatase calcineurin and of different effectors of 2B type protein phosphatases on dephosphorylation of hsp25.
After a preincubation of the EAT cell extract for 30 min at 37 "C in the presence of [ Y -~~P I A T P , in which time period the internal hsp25 kinase(s) phosphorylate(s) hsp25 at the same sites as in vivo, namely at S15 and S86 (5), the time course of dephosphorylation of hsp25 in the cell-free EAT system has been determined with and without adding bovine brain calcineurin, EGTA, or the phosphatase inhibitor microcystin-LR (Fig. 2). Fig. 2B demonstrates that the addition of bovine brain calcineurin to the cell-free EAT system results in a specific dephosphorylation of hsp25, whereas the phosphorylation of the other proteins within the system is not subjected to major changes. In Fig. 2C, the specific phosphorylation of hsp25 which is determined by the ratio between radioactive labeling of the hsp25 band and the radioactive labeling of all other proteins in the lane is represented as a function of time after addition of calcineurin, EGTA, or microcystin. Obviously, there is phosphatase activity already within the control incubation. After the addition of bovine brain calmodulindependent phosphatase calcineurin, the dephosphorylation of hsp25 in the incubation mixture is increased significantly. Addition of microcystin-LR to a final concentration of 1 KM as well as EGTA to a concentration of 5 mM decreases dephosphorylation of hsp25. However, from the data obtained it cannot be excluded that the dephosphorylation of hsp25 is only an indirect result of calmodulin-dependent (type 2B) protein phosphatase, e.g. a result of activation of other protein phosphatases or inhibition of protein kinases in the cell-free extract by calcineurin-mediated dephosphorylation.
Dephosphorylation of Purified Phosphorylated Recombinant hsp25 by the CalciumlCalrnodulin (2B Type) Protein Phosphatase Calcineurin-To demonstrate the direct dephosphorylation of hsp25 by calmodulin-dependent (2B type) protein phosphatase, we have used recombinant hsp25 (16) was phosphorylated at the same phosphorylation sites as in vivo using the EAT cell extract and [y3'P]ATP. After subsequent purification of the phosphorylated recombinant hsp25 by hydroxylapatite chromatography, the phosphorylated hsp25 was incubated in the presence of calcium and calmodulin with and without addition of bovine brain calcineurin (Fig. 3). A quantitative dephosphorylation of hsp25 by this enzyme is obtained (Fig. 3, A and B). After 2 h, less than 20% of the phosphorylated starting material compared to about 90% in the control could be detected (Fig. 30). Twodimensional PAGE of hsp25 reveals that both phosphorylated isoforms ("1") and ("2") are converted to the nonphosphorylated hsp25 isoform ("0") ( Fig. 3C), indicating that both phosphorylation sites S15 and S86 are dephosphorylated by calcineurin.
Expression of Calmodulin-dependent (Type 2B) Protein Phosphatase in EAT-For demonstrating the biological relevance of hsp25 dephosphorylation by 2B type protein phosphatases, the expression of calmodulin-dependent (type 2B) protein phosphatases in EAT cells was detected using polymerase chain reaction and Northern blot analysis (Fig. 4). The oligonucleotide primers used for amplification were complementary to a region of strong homology between all catalytic subunits of protein phosphatases of the 1, 2A, and 2B type and to the calmodulin-binding region of 2B type protein phosphatases, respectively (Fig. 4A). Using cDNA from EAT cells as template, two different types of sequences (EATl and EAT2, Fig. 4A) could be amplified. The sequences are identical with the mouse protein phosphatase 2B catalytic subunit 1 and 2 (17,18).
The amplified cDNA fragments of EAT calcium/calmodulin protein phosphatases EATl and EAT2 were used as probes for Northern blot analysis of the appropriate mRNAs in EAT cells (Fig. 4B). Using EATl (identical with a cDNA fragment of mouse protein phosphatase 2B catalytic subunit 1) as a hybridization probe, two major transcripts with an estimated length of 3.6 and 4.0 kilobases (kb) and one minor transcript with a length of about 1.6 kb can be detected. The two major transcripts correspond in size with the calcineurin mRNAs described in different mouse tissues (17). The amount of these two major transcripts was estimated to be about 50-100 copies/cell in EAT, indicating a medium-abundant mRNA. The hybridization probe EAT2 (identical with a cDNA fragment of mouse protein phosphatase 2B catalytic subunit 2) shows only weak hybridization. The 3.4-kb mRNA known for mouse catalytic subunit 2 (18) cannot clearly be detected, and it seems that the EAT2 probe cross-hybridizes with the same transcripts like the related EATl probe. These data suggest only a minor expression of the catalytic subunit 2 in EAT cells.

DISCUSSION
In this paper we provide evidence for the specific dephosphorylation of the small heat shock protein hsp25 in EAT cells by calcium/calmodulin-dependent (type 2B) protein phosphatase(s) by the following data.
(i) In a cell-free system derived from the EAT, the degree of phosphorylation of hsp25 is influenced by reagents affecting the activity of 2B type protein phosphatases. The increase of the degree of hsp25 phosphorylation at concentrations of microcystin-LR higher that 1 p~ (Fig. 1) (14) and the inhibition of dephosphorylation by the calcium chelating agent EGTA (Fig. 2C) (19) are characteristic for the action of protein phosphatases of the 2B type. The lower efficiency of the inhibition of dephosphorylation by EGTA could be explained by complexing only the free calcium in the assay while leaving the calcium bound to proteins like the regulatory subunit (B subunit) of calcium/calmodulin-dependent protein phosphatases nearly unaffected. (ii) Addition of the 2B type protein phosphatase calcineurin to the cell-free system results in an increased dephosphorylation of hsp25. This dephosphorylation seems to be very specific for hsp25 because no other phosphoprotein of the cell-free system is subject to major changes (cf. Fig. 2B). The hsp25 dephosphorylation could be the result of direct action of calcineurin on hsp25 as well as of activation of another protein phosphatase by dephosphorylation, i.e. of the appropriate inhibitor, or of inhibition of a hsp25 kinase by dephosphorylation.
(iii) Recombinant hsp25, which was phosphorylated in vitro at the same sites as in vivo (5), can be dephosphorylated at both phosphorylation sites directly by calcineurin in a calcium/calmodulin-dependent manner within a reaction mixture containing only recombinant hsp25 and the enzyme (Fig.  3).
(iv) In EAT cells, the expression of two different mRNAs of protein phosphatases of the 2B type, which are identical with the mouse protein phosphatase subtype 2B1 and 2B2 (17,18), could be demonstrated by the polymerase chain reaction (Fig. 4A). Northern  subtype 2B1 are medium-abundant in EAT cells (Fig. 4B). The co-expression of the enzyme protein phosphatase 2B1 and the substrate hsp25 in EAT cells indicates the biological relevance of this enzymatic reaction. Recently, it has been shown that in human fibroblasts the early protein phosphorylation induced by interleukin-1 and tumor necrosis factor-a which includes significant phosphorylation of hsp27 is mimicked by okadaic acid (20), a potent protein phosphatase inhibitor which preferentially inhibits 2A and 1 type phosphatases but at higher concentrations also 2B type phosphatases (21). The okadaic acid concentration necessary to provide significantly increased hsp27 phosphorylation is about 200 nM which is about 10 times higher than the concentration known to inhibit type 1 and 2A phosphatases (21). Although the increased hsp27 phosphorylation in these experiments could be explained by indirect effects of okadaic acid on hsp25 kinases, this finding can be also interpreted as a hint for dephosphorylation of hsp27 by 2B type protein phosphatases.
Apart from this, the loading of mouse C127 cells with different calcium chelators has been demonstrated to result in an increased accumulation of phosphorylated hsp25 isoforms and a decrease of the nonphosphorylated isoform of hsp25 (22). This result could be explained by inhibition of a calcium/calmodulin-dependent phosphatase which recognizes hsp25 as a substrate in these cells.
The preferential dephosphorylation of a 28-kDa protein by the calmodulin-dependent phosphatase calcineurin has already been described for human platelets (23). Although the 28-kDa protein which is rapidly phosphorylated after thrombin stimulation has not been identified, one may assume that this protein is identical with a small heat shock protein.
Calcium/calmodulin-dependent (type 2B) protein phosphatases have been described to be serine/threonine phosphatases with a relatively narrow substrate specificity (24,25). Substrate proteins known to be dephosphorylated by these enzymes are the a-subunit of the phosphorylase kinase, protein phosphatase inhibitor-1, myosin light chains (24), the regulatory subunit of the CAMP-dependent protein kinase (26), and the dopamine-and cyclic AMP-regulated neuronal phosphoprotein DARPP-32 (27). Interestingly, also the A-chain of a-crystallin which has significant sequence homology to the small stress proteins (28) was shown to be dephosphorylated by the calcium/calmodulin-dependent (2B type) protein phosphatase calcineurin (29). The site of dephosphorylation in the A-chain of a-crystallin is located within the sequence motif RLPS(P) which is similar to the phosphorylation site at serine 15 of hsp25 with the motif RSPS(P). Taking further into account the motifs dephosphorylated by 2B type phosphatases in protein phosphatase inhibitor-1 RRPT(P), in the regulatory subunit of CAMP-dependent protein kinase RRVS(P), in phosphorylase kinase RRLS(P), and in DARPP-32 RRPT(P), one may formulate the minimal motif requirement of RXXS/T(P) to be necessary (but not sufficient) for dephosphorylation. Both sites (RQLS(P) and RSPS(P)) dephosphorylated in hsp25 will fit into this minimal motif. Interestingly, the motif RXXS is also known as the consensus motif necessary for phosphorylation by the multifunctional calmodulin-dependent protein kinase I1 (30) and by S6 kinase I1 (31).
It has been demonstrated that the degree of phosphorylation of the small heat shock proteins from mouse hsp25 and human hsp27 is modulated by heat shock (12,32). After a first heat shock, the degree of phosphorylation of hsp25 is increased, while cells surviving a second heat shock by having acquired thermoresistance show an increased amount of de-phosphorylated hsp25 (32). Furthermore, the ability of the cell to maintain a high degree of nonphosphorylated hsp27 under heat shock conditions has been shown to be essential for acquisition of thermoresistance of the cell (12). Since it has been known for several years that heat shock alters the intracellular calcium concentration (33,34) and that cellular thermoresistance is calcium-dependent and can be influenced by calcium chelators, by calcium ionophores, and anticalmodulin drugs (35), as well as by overexpression of calciumbinding proteins (36), one may speculate that the calcium/ calmodulin-dependent activation of dephosphorylation of hsp25/hsp27 plays an important role in the mechanism of acquisition of cellular resistance to thermal stress.