Evidence of an interaction between Mos and Hsp70: a role of the Mosresidue serine 3 in mediating Hsp70 association

Hui Liu1,2, Vijayalakshmi B Vuyyuru1,2, Chau D Pham1,2, Yandan Yang1 and Balraj Singh*,1

1Department of Molecular Pathology, Box 172, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd.,Houston, Texas, TX 77030, USA

c-Mos is a germ cell-speci®c MAP kinase kinase kinase(MAPKKK) that plays an essential role during meioticdivisions of oocytes. c-Mos is a key component of anactivity, cytostatic factor, required for metaphase IIarrest of unfertilized eggs in vertebrates. To understandthe regulation of c-Mos, we are investigating c-Mos-interacting proteins. We provide evidence that mouse c-Mos binds to Hsp70, a molecular chaperone. Hsp70 wasfound to associate with Mos ectopically expressed inCOS-1 cells. Mos-Hsp70 complexes could be immuno-precipitated with both Mos and Hsp70 antibodies.Despite a low-abundance of Mos, the Hsp70 antibodyimmunoprecipitated Mos as the major protein. Ofimportance, the Mos protein present in anti-Hsp70immunoprecipitates functioned as an active MAPKKKindicating that it is not grossly misfolded. It is knownthat c-Mos protein kinase activity in cell extracts oftransfected COS-1 or NIH3T3 cells is labile. We foundthat the inclusion of adenosine triphosphate (ATP) in cellextracts protected against the loss of Mos kinaseactivity. In the absence of ATP from cell extracts,protein kinase activity of Mos was lost within 6 h on iceeven though the Mos protein was not degraded andremained bound to Hsp70. Based on our identi®cation ofc-Mos-Hsp70 interaction, one of the roles of ATP maybe to assist the regulation of c-Mos via ATP involvementin the protein-folding function of Hsp70 and possiblyother molecular chaperones. We also detected by co-immunoprecipitation a physical association betweenendogenous c-Mos and Hsp70 in Xenopus eggs. Toprovide further evidence for the functional signi®cance ofHsp70 interaction to Mos function, we show that theresidue serine 3 in Mos, which is important for theregulation of protein kinase activity of Mos is alsoimportant for Hsp70 association.

Keywords: mos oncogene; cytostatic factor; molecularchaperone; protein phosphorylation; enzyme regulation

Introduction

The v-mos gene of the Moloney murine sarcoma virus(Mo-MuSV) encodes a serine/threonine protein kinase(van Beveren et al., 1981; Maxwell and Arlinghaus,1985b). The v-Mos protein contains the entire 343-amino acid c-Mos sequence plus a 31-amino acidsequence at its amino terminus which results from the

fusion of viral env sequences (including the initiationcodon) upstream of the c-mos open reading frame (vanBeveren et al., 1981). The c-Mos protein is expressed ingerm cells. It plays an important role during meioticdivisions of oocytes (reviewed in Singh and Arlinghaus,1997). As an essential component of an activity,cytostatic factor (CSF), c-Mos is responsible for themetaphase II arrest of unfertilized eggs in vertebrates(Sagata et al., 1989). Loss of Mos function causesreduced fertility in female mice and the development ofovarian teratomas as a result of parthenogeneticactivation of oocytes/eggs (Colledge et al., 1994;Hashimoto et al., 1994; Pham et al., 1997). Ectopicexpression of the normal c-mos gene from a foreignpromoter causes neoplastic transformation of NIH3T3cells. The transforming activity of v-mos can beattributed, therefore, to its inappropriate expressionin somatic cells. Both v-Mos and c-Mos are able toactivate the MAP kinase pathway by directlyphosphorylating MAP kinase kinase (Posada et al.,1993; Nebreda et al., 1993; Pham et al., 1995; Chenand Cooper, 1995). Furthermore, because of theirsimilarity in structure, v-Mos is able to substitute c-Mos for its function in oocytes (Freeman et al., 1990).

To understand various aspects of regulation of c-Mos function, we are investigating the proteins thatinteract with c-Mos and v-Mos. Although several c-Mos-interacting proteins have been described (Zhou etal., 1991a, 1992; Bai et al., 1992a,b, 1993; Chen andCooper, 1995, 1997; Chen et al., 1997), which help toexplain the basis of c-Mos function in oocytematuration and cellular transformation, regulation ofc-Mos activity is still poorly understood. In this studywe provide evidence that Hsp70 molecular chaperone isa major Mos-associated protein. Our further experi-ments based on the known biochemical properties ofHsp70 indicated that Hsp70 plays an importantfunctional role in c-Mos regulation. To our knowl-edge, this is the ®rst report of an association betweenHsp70 and a labile serine/threonine protein kinase. Wealso provide evidence that c-Mos residue Ser-3 which isimportant for the regulation of c-Mos protein kinase isalso important for the Mos-Hsp70 interaction.

Results

A number of cellular proteins coprecipitate with Mosduring immunoprecipitation with several antipeptideMos antibodies. One approach to identify these Mos-associated proteins is through their detection withantibodies against the known proteins. In this study,we succeeded in identifying Hsp70 as one of the Mos-associated proteins by this approach. We began by

*Correspondence: B Singh2The ®rst three authors contributed equally to this workReceived 8 June 1998; revised 14 January 1999; accepted 14 January1999

Oncogene (1999) 18, 3461 ± 3470ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00

http://www.stockton-press.co.uk/onc

analysing the cellular proteins that associate with full-length v-Mos and v-Mos residues 32 ± 374 (which isequivalent to c-Mos). Although v-Mos(32 ± 374) en-coded by the Mo-MuSV124 v-mos gene contains 10amino acid substitutions in the protein kinase domainof mouse c-Mos (van Beveren et al., 1981), bothbehave similarly by several criteria that include proteinkinase activity, patterns of phosphorylation andassociated proteins. Therefore, to take optimaladvantage of our antipeptide antibodies, which arebased on the v-Mos sequence, and to examine thee�ect of v-Mos residues 1 ± 31 on the kinase function,we carried out initial analysis using constructsencoding v-Mos(32 ± 374). For the purpose of thisstudy, v-Mos(32 ± 374) will be referred to as Mos todistinguish it from c-Mos and v-Mos.

Coimmunoprecipitation of Mos and Hsp70

v-Mos and Mos were expressed in COS-1 cells bytransfection with plasmids in which the cloned full-length or truncated v-mos gene was expressed fromSV40 late promoter. Two days after transfection, cellswere metabolically labeled with 35S-methionine for30 min and lysed under relatively mild detergentconditions to preserve protein complexes. Immunopre-cipitation was carried out with three di�erentantipeptide antibodies raised against v-Mos residues37 ± 55, 260 ± 271, and 363 ± 374 and with a commer-cially available monoclonal antibody raised againstrecombinant human Hsp70 (Figure 1). The Hsp70antibody recognizes e�ciently both Hsp70 and Hsc70from human and monkey. All three anti-Mosantibodies precipitated v-Mos and Mos e�ciently; anumber of cellular proteins coprecipitated with Mosand v-Mos. One of these proteins, which comigrateswith Hsp70, coprecipitated with Mos more e�cientlythan it did with v-Mos (compare Figure 1 lanes 5, 9,

and 14 with lanes 3, 8, and 12, respectively). Inaddition to the Hsp70-sized protein, a number ofproteins coprecipitated with Mos, including p60, p35,p30, and p25 (not marked in Figure 1). The identity ofthese proteins remains to be determined.

To ®rst determine whether Mos associates withHsp70, we analysed anti-Hsp70 immunoprecipitates forthe presence of Mos. Both Mos- and v-Mos-sizedproteins were present in anti-Hsp70 immunoprecipitatesprepared from the cells expressing Mos and v-Mos,respectively (Figure 1, lanes 17 and 18). In agreementwith the results obtained with the anti-Mos immunopre-cipitates, Hsp70 association with v-Mos was weaker thanwith Mos. We conclude that the 31 amino acids presentat the amino-terminus of v-Mos, but absent in Mos,destabilize the Mos-Hsp70 association. We also exam-ined the ability of mouse c-Mos to associate with Hsp70and found that c-Mos and Mos behave similarly (datanot shown). Association between Mos and Hsp70 wasquite strong as it could survive RIPA bu�er containing1% NP± 40, 1% sodium deoxycholate and 0.1% sodiumdodecyl sulfate during radioimmunoprecipitation (datanot shown). It is important to note that Mos is a low-abundance protein and therefore cannot be detected bySDS ±PAGE among 35S-methionine labeled proteinspresent in cell extract (compare lanes 19 and 20 inFigure 1). Despite the low-abundance of Mos, anti-Hsp70 immunoprecipitates showed Mos as the mostprominent band (Figure 1). This result indicates thatMos has an unusually high preference for Hsp70association.

To establish that the Hsp70-sized protein co-precipitated with Mos is Hsp70, we compared itspartial V8 protease digestion fragments with those ofauthentic Hsp70 immunoprecipitated with anti-Hsp70.All the fragments generated from the Mos-associatedHsp70-sized protein were also generated from theauthentic Hsp70 (Figure 2B). Similarly, V8 protease

Figure 1 Coimmunoprecipitation of Mos and Hsp70. Plasmid constructs expressing full length v-Mos (lanes 3, 4, 8, 12, 13 and 17)or v-Mos residues 32 ± 374 (indicated as Mos since it is similar to c-Mos; lanes 5, 6, 9, 14, 15, 18 and 20) were transfected into COS-1 cells. Two days later, the proteins were metabolically labeled with 35S-methionine for 30 min and analysed byimmunoprecipitation with three di�erent polyclonal anti-Mos peptide antibodies, as indicated, and a monoclonal Hsp70antibody. The controls included mock-transfected cells (lanes 1, 2, 7, 10, 11, 16 and 19) and immunoprecipitation with the peptide-blocked antibodies (peptide+lanes). Lanes: 1 ± 18, immunoprecipitates resolved on 8% gel; 19 and 20, total cell extracts resolved on12% gel. Shown at right are prestained molecular-weight markers in kilodaltons (Diversi®ed Biotech, Boston, MA, USA). Ab,antibodies

Mos association with Hsp70H Liu et al

3462

digestion fragments generated from the Mos-sizedprotein coprecipitated with Hsp70 were identical tothe fragments generated from authentic Mos immuno-precipitated with anti-Mos (Figure 2A). These resultsstrongly suggested that Mos associates with Hsp70.More than one homolog of Hsp70 may be expressed ina given cell type (reviewed in Rassow et al., 1997) but,thus far, we have not identi®ed the particular homologor homologs of Hsp70 that associate with Mos. Someminor V8 protease fragments resulting from thedigestion of authentic Hsp70 were either missing orwere of lower abundance in Mos-associated Hsp70(compare Figure 2B lanes 1 and 2 with lanes 5 and 6).One interpretation of these data is that some but notall homologs of Hsp70 associate with Mos.

It was important to rule out that the Hsp70antibody does not cross-react with Mos. To do so,we immunoblotted the lysate from c-mos-transfectedcells with the Hsp70 antibody (Figure 3A). Lack of a c-Mos-sized band in the Western blot indicates that theHsp70 antibody does not cross-react with Mos.

The data in Figure 1 indicate that Mos is not amajor protein in the lysate of transfected COS-1 cells.To quantitate the Mos protein further, we immuno-blotted the total cell lysates from Mos- and c-Mos-transfected COS-1 cells along with the lysate fromMuSV124-infected NIH3T3 cells (Figure 3B). Theamount of c-Mos, Mos and v-Mos in these extractswas comparable to each other. Each amounted toabout 0.004% of total cellular protein as quantitatedusing the known amounts of GST-v-Mos in Westernblots (data not shown). Our estimate is in agreement

with the estimate of v-Mos in MuSV infected NIH3T3cells by Papko� et al. (1982). Intracellular concentra-tion of v-Mos in MuSV-infected NIH3T3 cells iscomparable to c-Mos concentration in unfertilizedXenopus eggs (Sagata et al., 1989).

To provide further evidence for Mos-Hsp70 associa-tion, we carried out immunoprecipitation-Westernimmunoblotting (IP-Western) analysis. Hsp70 co-precipitated with c-Mos in anti-Mos precipitates wasspeci®cally detected by anti-Hsp70 in Western blot(Figure 4A). Conversely, c-Mos coprecipitated withHsp70 in anti-Hsp70 immunoprecipitate was detectedby anti-Mos in the Western blot (Figure 4B).Appropriate controls included in these experiments(mock-transfected cells and immunoprecipitates withpeptide-blocked anti-Mos antibodies) ruled out thepossibility of these results being merely due to antibodycross-reactivities (Figure 4). In this experiment, theplasmid construct expressing mouse c-Mos was usedfor transfection. c-Mos is recognized poorly by theanti-v-Mos(363 ± 374) antibodies used here because ofan amino acid di�erence in the peptide antigen. Still,the data provided clear evidence of a speci®cassociation between c-Mos and Hsp70.

To determine that c-Mos interacts with Hsp70 inphysiological conditions, we performed IP-Westernanalysis on lysate prepared from Xenopus eggs.According to these data, Hsp70 coimmunoprecipitatedwith c-Mosxe (Figure 5C). Based on Western blottinganalysis, both c-Mosxe and Hsp70 antibodies specifi-cally recognized their respective antigens (Figure5A,B). It is important to note that the Mosxe antibody

Figure 2 Identi®cation of Mos and Hsp70 by partial V8 proteasedigestion. The 35S-methionine-labeled Mos-sized protein that co-immunoprecipitated with Hsp70 (A, lanes 3 and 4), and theHsp70-sized protein that coimmunoprecipitated with Mos (B,lanes 1 and 2) were subjected to partial V8 digestion along withthe authentic Mos (A, lanes 1 and 2) and Hsp70 (B, lanes 3 and4). The digested protein fragments were analysed by SDS±PAGEon a 15% gel. Lanes 5 and 6 in (B) show a shorter exposure oflanes 3 and 4. The source of each protein band subjected to V8digestion is indicated at the bottom of the lanes; Mos and Hsp70indicate anti-Mos and anti-Hsp70 immunoprecipitates, respec-tively

Figure 3 Western immunoblotting analysis of Hsp70 and Mos.(A) Hsp70 present in mock-transfected (lane 1) and c-Mos-transfected COS-1 cells (lane 2) was blotted with the W27 cloneHsp70 antibody. (B) Total cell lysate from MuSV-infectedNIH3T3 cells prepared 2 days post-infection (lane 1), mock-transfected (lane 2), Mos-transfected (lane 3) and c-Mos-transfected COS-1 cells was analysed by Western blotting withthe v-Mos(37 ± 55) antibody

Mos association with Hsp70H Liu et al

3463

did not cross-react with Hsp70 and Hsp70 antibodydid not cross-react with Mos.

Hsp70-Mos complexes can phosphorylate MEK1

Hsp70 homologs play an important role as molecularchaperones in the folding of numerous proteins (Hartl,1996; Rassow et al., 1997). One of these roles is toassist the folding of nascent polypeptide chains (Bukau

et al., 1996). Therefore, in evaluating the signi®cance ofour results, the key question to address was whetherHsp70-associated Mos represented only improperlyfolded newly synthesized Mos molecules. To addressthis question, we determined the protein kinase activityof Mos associated with Hsp70. The v-Mos, Mos, and akinase-inactive K90R mutant of Mos were eitherimmunoprecipitated with anti-v-Mos(37 ± 55) or co-precipitated with Hsp70 in anti-Hsp70 immunoprecipi-tates. Mos protein kinase activity was determined usingGST-MEK1 (K97R kinase-inactive mutant) as asubstrate (Figure 6). Results from this experimentclearly showed that Mos coprecipitated with Hsp70 hasprotein kinase activity. Based upon a parallelexperiment involving metabolic labeling with 35S-methionine, the amount of Mos protein present inanti-Hsp70 immunoprecipitate was one-third of thatpresent in anti-Mos immunoprecipitate. The speci®cactivity of Mos protein kinase in anti-Hsp70 immuno-precipitate was comparable, therefore, to the speci®cactivity of Mos protein kinase in anti-Mos immuno-precipitates (compare Figure 6 lanes 5 and 9). Since theprotein kinase catalytic domain of Mos (residues 67 ±343) encompasses 80% of Mos sequence, this resultindicates that the Hsp70-associated Mos protein is notgrossly denatured. We noted that Mos present in anti-Hsp70 immunoprecipitates autophosphorylated extre-mely poorly, which may be the result of masking ofautophosphorylation sites by Hsp70 association.Signi®cance of Mos autophosphorylation is notknown at present but based on our phosphopeptidemapping data most of the in vitro phosphorylation thatoccurs on Mos in anti-v-Mos(37 ± 55) immunoprecipi-tates does not occur on Mos in transfected COS-1 cells

Figure 4 Detection of c-Mos-Hsp70 complexes by immunopre-cipitation-Western immunoblotting. Cell extracts from mock-transfected (A, lanes 1, 3, and 4; B, lane 1) and from mouse c-mos-transfected COS-1 cells (A, lanes 2, 5, and 6; B, lane 2) wereanalysed by IP-Western immunoblotting. The Mos antibody wasagainst the v-Mos C-terminal 12 amino acid sequence (residues363 ± 374). (A), lanes 1 and 2 contained total cell extracts frommock-transfected and c-mos-transfected cells, respectively, ana-lysed by Western immunoblotting with anti-Hsp70

Figure 5 Coimmunoprecipitation of Hsp70 with endogenous c-Mos from Xenopus egg extract. Total cell lysates from COS-1 cells(lane 1) and unfertilized Xenopus laevis eggs (lane 2) weresubjected to Western blotting with c-Mosxe antibody (C237,Santa Cruz Biotechnology) (A) and with Hsp70 antibody W27clone (B). (C) Cell lysates from COS-1 cells (lane 1) and Xenopuseggs (lane 2) were subjected to immunoprecipitation with theC237 c-Mosxe antibody. The immunoprecipitates were analysedby Western blotting with the W27 Hsp70 antibody

Figure 6 Mos coimmunoprecipitated with Hsp70 can phosphor-ylate its substrate MEK1 in vitro. Mos protein produced intransfected COS-1 cells was precipitated with anti-v-Mos(37 ± 55)(lanes 1 ± 6) or anti-Hsp70 antibody (lanes 7 ± 9). Immunecomplexes prepared from transfected COS-1 cells expressing akinase-inactive K90R Mos mutant (lanes 1, 2 and 7), v-Mos(lanes 3, 4, and 8) and Mos (lanes 5, 6, and 9) were incubatedwith GST-MEK1(K97R mutant) bound to Sepharose in a kinasereaction mixture containing [g-32P]ATP. For lanes indicated aspeptide+, immunoprecipitates were prepared with the peptide-blocked anti-Mos to serve as control

Mos association with Hsp70H Liu et al

3464

(CD Pham, VB Vuyyuru, Y Yang, W Bai and B Singh,manuscript submitted for publication).

Role of Mos-Hsp70 interaction

Among several domains present in Hsp70, the ATPasedomain and the peptide-binding domain are present inall its homologs. Structural change in either of thesetwo domains by ATP or peptide binding causes achange in the other one. Thus, ATP binding causes therelease of the bound peptide, which in turn stimulatesthe ATPase activity of Hsp70. The ADP-bound formof Hsp70 has increased a�nity for peptide binding.The proper folding of an Hsp70-associated proteinrequires cycles of association and dissociation. There-fore, ATP is required for Hsp70 to function as amolecular chaperone. Laboratories working on c-Moshave known that c-Mos protein kinase activity is verylabile in cell extracts. In light of our results, wepostulated that ATP would be required to maintainassociation/dissociation cycles of c-Mos-Hsp70 com-plexes and, therefore, the c-Mos protein kinase activity.The following experiments were carried out toinvestigate this possibility.

The ®rst important experiment was to determinewhether addition of ATP to cell extracts from mos-transfected cells would prevent the loss of Mos proteinkinase activity. Mos protein kinase activity was almosttotally lost upon incubation of the cell extract on icefor 4 h (Figure 7A). In this experiment, Mos kinaseactivity was followed by observing transphosphoryla-tion of GST-MEK1(K97R) as well as the cellularprotein p60. p60 appears to be a true substrate of Mosas it was not phosphorylated by the immunoprecipi-tates containing kinase-inactive Mos (Figure 7A, lane1), although the degree of p60 phosphorylation variedsigni®cantly among experiments. Of interest, additionof ATP (3 mM) and MgCl2 (5 mM) to the cell extracthad a signi®cant protective e�ect as only 50% activitywas lost under these conditions (compare Figure 7a,lanes 3 and 9). ATP alone without MgCl2 had noprotective e�ect (not shown). MgCl2 alone had a slightprotective e�ect. Testing the e�ect of ATP in theconcentration range of 1 mM±3 mM, we found that 2 ±3 mM ATP provided optimal protection against theloss of Mos activity (data not shown). To determinewhether protein phosphorylation-dephosphorylationcontributed to the protection of Mos kinase activityin our system, we tested the e�ect of sodium ¯uoride(NaF), a phosphatase inhibitor. In our system, NaFhad a modest additional protective e�ect when addedalong with ATP and MgCl2 (compare Figure 7A lanes9 and 11). NaF had a less signi®cant e�ect possiblybecause our extraction bu�er already containedanother phosphatase inhibitor, sodium pyrophos-phate, which causes a dramatic improvement in theactivity of Mos protein kinase (Maxwell and Arling-haus, 1985a). The e�ect of pyrophosphate was ®rstdiscovered while attempting to improve the assay for v-Mos protein kinase activity (Maxwell and Arlinghaus,1985a).

Next, we investigated that the hydrolysis of ATPwas required for its protective e�ect on Mos kinase incell extracts. In this experiment the cell lysate from c-Mos-transfected COS-1 cells was incubated at roomtemperature for 4 h which was followed by c-Mos

immune complex autokinase assay. We found that theloss of c-Mos kinase activity was prevented by ATP orthio-ATP (Figure 7B). There was no signi®cant

Figure 7 Protection by ATP against the loss of Mos proteinkinase activity in cell lysates. (A) Cell lysates from mos-transfectedCOS-1 cells were incubated at 48C for 4 h without any addition(lanes 5 and 6); or with 5 mM MgCl2 (lanes 7 and 8); 5 mM

MgCl2 and 3 mM ATP (lanes 9 and 10); or with 5 mM MgCl23 mM ATP and 10 mM NaF (lanes 11 and 12). Then Mos wasimmunoprecipitated with anti-v-Mos(37 ± 55) and subjected toprotein kinase assay in the presence of GST-MEK1 (K97R) as inFigure 6. As negative and positive controls, cell extractscontaining kinase-inactive K90R Mos (lanes 1 and 2) or Mos(lanes 3 and 4) were used immediately after preparation in theusual manner. In addition to autophosphorylation and transpho-sphorylation of MEK1, phosphorylation of a 60-kDa protein(p60) by Mos can also be seen. (B) Cell lysate from c-mos-transfected COS-1 cells was incubated at room temperature for4 h in the presence of 5 mM MgCl2 (lanes 3 and 4), 10 mM

sodium molybdate (lanes 5 and 6), 5 mM MgCl2 plus 3 mM ATP(lanes 9 and 10), 5 mM MgCl2 plus 1 mM thio-ATP (lanes 11 and12) or without any addition (lanes 13 and 4). These lysates alongwith the control lysate without any prior incubation (lanes 1 and2) were subjected to immune complex autokinase assay. The 32P-labeled c-Mos band is indicated as Mos. The lanes marked aspeptide+represent kinase assays with control immunoprecipitatesthat were obtained with the peptide-blocked antibody

Mos association with Hsp70H Liu et al

3465

protection by a non-hydrolysable analog of ATP,AMPPNP. Similarly, Mg2+ alone and a transitionmetal molybdate were without any e�ect. We concludethat the hydrolysis of ATP is important for itsprotective e�ect.

To determine whether the loss of Mos kinase activityin the absence of ATP is caused by enhanceddegradation of Mos, we examined the fate of Mosand Hsp70 metabolically labeled with 35S-methionineafter incubation of the cell extracts with and withoutATP. When Mos-associated Hsp70 were immunopre-cipitated with anti-v-Mos(363 ± 374) and analysed bySDS ±PAGE, we observed that the amount of Mosand associated Hsp70 remained essentially unchangedupon incubation on ice for 4 h with or without ATP(Figure 8). Furthermore, we observed that the loss ofMos kinase activity in COS-1 cell extracts wasreversible in the early stage but not after 24 ± 48 hincubation (data not shown). Based upon what isknown about the Hsp70 chaperoning mechanism(Hartl, 1996; Bukau et al., 1996; Rassow et al.,1997), we suggest that ATP may be needed to keepMos in an active conformation which depends oncycles of association and dissociation with Hsp70. Inthe absence of ATP, the Mos-Hsp70 complex may beunable to dissociate and eventually Mos becomestotally inactive upon prolonged incubation. Althougha lack of exogenously added ATP did not enhance Mos

degradation in cell extracts, experiments reportedbelow utilizing COS-1 cells transfected with mutantversions of Mos are consistent with a role of Hsp70interaction in the regulation of Mos turnover and Mosprotein kinase activity in vivo.

E�ect of the S3A mutation on Mos-Hsp70 interaction

Considering the well known function of Hsp70 as amolecular chaperone, Mos-Hsp70 interaction may beimportant for mediating Mos interaction with othercellular proteins including its regulators and/or targets.To understand the functional signi®cance of Mos-Hsp70 interaction, one can try to create mutations inMos which speci®cally abolish its interaction withHsp70. Alternatively, one can utilize the mutations thatare known to cause speci®c defects in c-Mos functionand regulation. Here we describe the results obtainedwith the later strategy. First, we decided to focus onthe c-Mos mutations involving phosphorylation sitesbecause there are indications that a number ofprotein ± protein interactions with c-Mos may bemediated by phosphorylation. Of particular impor-tance is the c-Mos residue Ser-3 which is phosphory-lated in oocytes after GVBD (Freeman et al., 1992;Nishizawa et al., 1992). Phosphorylation at Ser-3 isbelieved to be important for the stability and proteinkinase activity of c-Mos. The c-Mos residue Pro-2 isimportant for the proteolysis of c-Mos by theubiquitin-proteosome system (Nishizawa et al., 1993).Phosphorylation at the adjacent residue, Ser-3, mayinterfere with the recognition of Pro-2 by the ubiquitinsystem (Nishizawa et al., 1992). Consistent with thismodel, the substitution of Ser-3 with Ala and Gluresults in unstable and stable c-Mos, respectively. Inaddition, the S3A but not the S3E mutation inhibitsthe interaction of c-Mos catalytic domain with itssubstrate MEK1 (Chen and Cooper, 1995). The role ofSer-3 phosphorylation in regulating c-Mos is notcompletely understood. A direct correlation betweenSer-3 phosphorylation and c-Mos stability on itsprotein kinase activity has not been shown. The useof the S3A mutant has also given contradictory resultsas, in contrast to the results obtained by Nishizawa etal. (1992) and Chen and Cooper (1995), Freeman et al.(1992) did not observe any e�ect of the S3A mutationon c-Mos function. Currently, it is not known whetherSer-3 and phosphorylated Ser-3 mediate Mos interac-tion with the proteins involved in Mos regulation.

To determine whether interaction with Hsp70molecular chaperone may be involved in the regula-tion of turnover and protein kinase activity of c-Mos,we examined the e�ects of the S3A and S3E mutationson Mos interaction with Hsp70. As expected from theresults with the analogous Xenopus c-Mos mutantsexpressed in Xenopus oocytes or in somatic cells(Nishizawa et al., 1992; Chen and Cooper, 1995), theS3A but not the S3E mutation inhibits v-Mos proteinkinase and accelerates the degradation of Mosproduced in transfected COS-1 cells (Yang et al.,1998). We immunoprecipitated 35S-methionine-labeledmutant Mos with anti-v-Mos(363 ± 374) to ®nd that theS3A but not the S3E mutation almost completelyabrogated the co-precipitation of Hsp70 with Mos(Figure 9, compare lanes 1 of all three panels). The lossof Hsp70 coprecipitation with Mos occurred only in

Figure 8 Detection of Mos and associated Hsp70 afterincubation of cell extract with or without ATP. The cell lysatesprepared from 35S-methionine-labeled mos-transfected COS-1 cellswere incubated at 48C for 4 h with and without ATP as in Figure7a. Then, the Mos protein was immunoprecipitated with anti-v-Mos(363 ± 374) and analysed by SDS±PAGE followed by¯uorography. Incubation conditions: 0 h (lanes 1, 2); 4 h (lanes3, 4); 4 h with 5 mM MgCl2 (lanes 5, 6); 4 h with 5 mM MgCl2and 3 mM ATP (lanes 7, 8)

Mos association with Hsp70H Liu et al

3466

case of the S3A mutant; all other phosphorylation sitemutants (which included S16A, S16E and S25A)behaved similar to the wild-type (WT) Mos in thisregard (data not shown). The S16A mutationspeci®cally abrogates the coprecipitation of a cellularprotein p35 with Mos (CD Pham, VB Vuyyuru, YYang, W Bai and B Singh manuscript submitted forpublication).

Although Hsp70 did not coprecipitate with the S3Amutant Mos immunoprecipitated with anti-Mos anti-bodies, the S3A mutant Mos coprecipitated e�cientlywith Hsp70 immunoprecipitated with the anti-Hsp70antibody. One possible explanation of this surprisingresult may be that although the S3A mutationsigni®cantly weakens Mos-Hsp70 interaction, thebinding of Hsp70 antibody to Hsp70 somehowstabilizes these weak Mos-Hsp70 complexes. TheHsp70 antibody was raised using whole recombinanthuman Hsp70 but the exact epitope that is recognizedby this monoclonal antibody is not known. We tried touse some other commercially available Hsp70 andHsc70 antibodies (K-20 and K-19, Santa CruzBiotechnology) to resolve this issue (H Liu, VBVuyyuru, CD Pham, and B Singh unpublishedobservations). These antibodies immunoprecipitatedHsp70 and Hsc70, respectively, but failed to co-precipitate Mos possibly because they interfere withthe chaperoning function of Hsp70.

At this point, we chose to utilize a well characterizedHsp70 monoclonal antibody, clone BB70, which has

been used previously to investigate Hsp70-associatedproteins (Uzawa et al., 1995). First we found byWestern blotting that the antibody recognizes Hsp70 inCOS-1 cells with high speci®city and that it does notcross-react with c-Mos (Figure 10A). Upon immuno-precipitation of 35S-methionine-labeled proteins, wefound that c-Mos coimmunoprecipitated with Hsp70e�ciently (Figure 10B). The S3A mutation caused adramatic inhibition in the association of c-Mos withHsp70. In this experiment the relative amount of c-Mos proteins was determined by immunoprecipitationwith anti-v-Mos(37 ± 55) (Figure 10C). The amount ofthe S3A c-Mos was about 50% of WT c-Mospresumably due to increased proteolysis of the mutantc-Mos. These results indicate that the c-Mos residueSer-3 is important for Hsp70 interaction. Phosphoryla-

Figure 9 E�ects of mutating Ser-3 on Mos binding with Hsp70.The wild-type Mos and its mutants S3A and S3E were producedby transfection in COS-1 cells. The proteins were labeled with 35S-methionine and immunoprecipitated as in Figure 1. Antibodiesused for immunoprecipitation, indicated at the bottom, were anti-v-Mos(363 ± 374) (peptide7lanes), peptide-blocked anti-v-Mos(363 ± 374) (peptide+lanes) and anti-Hsp70 W27 clone. Theamount of the S3A mutant Mos was lower than WT Mos,therefore, the ®lm in the middle panel was exposed for a longertime. Exposure times: left and right panels, 1d; middle panel, 2d

Figure 10 Coprecipitation of c-Mos with Hsp70 in BB70immunoprecipitates. (A) Cell lysates from mock-transfected(lane 1) and c-mos-transfected COS-1 cells (lane 2) were subjectedto Western immunoblotting with the BB70 Hsp70 antibody. (B)Transfected COS-1 cells were metabolically labeled with 35S-methionine and cell lysates immunoprecipitated with the BB70antibody. WT, wild-type c-Mos; S3A, the S3A mutant of c-Mos.(C) In order to obtain a relative estimate of c-Mos, a part of celllysate used in panel (B) was immunoprecipitated with the v-Mos(37 ± 55) antibodies. Peptide+lanes show immunoprecipitateswith the peptide-blocked antibody

Mos association with Hsp70H Liu et al

3467

tion of Ser-3 may lead to a modi®ed c-Mos interactionwith Hsp70 and thus alter its fate with regards toproteolysis and kinase regulation.

Discussion

In this paper, we demonstrate that Mos associateswith the Hsp70 molecular chaperone. This associationis analogous to the association between anothermolecular chaperone, Hsp90 and some labile proteinkinases including Src, Raf, Cdk4 and Wee1 (Stancatoet al., 1993; Xu and Lindquist, 1993; Aligue et al.,1994; Schulte et al., 1995; Stepanova et al., 1996). Agood amount of Mos protein observed in anti-Hsp70immunoprecipitate (Figure 1) argues for a speci®c andhigh-a�nity interaction between Mos and Hsp70.Absolute amount of Mos in cells is very low, andSDS ±PAGE analysis of total cell lysate from 35S-methionine-labeled mos- transfected cells did notreveal a Mos band when compared to a lysate frommock-transfected cells (Figure 1). In contrast, Mosprotein was observed as the most prominent band inanti-Hsp70 immunoprecipitate. While the stoichiome-try of Mos-Hsp70 interaction remains to be deter-mined, data such as in Figure 1 indicates that most ofMos is involved. To our knowledge, Mos is the ®rstlabile protein kinase found to associate with Hsp70 tosuch a great extent. In addition to the high degree ofthis protein-protein interaction, what makes itinteresting is the correlation between the Hsp70association and Mos regulation, both processes beingregulated by the known potentially importantphosphorylation site Ser-3 of c-Mos. Hsp70 isgenerally viewed as a molecular chaperone whichinteracts with an unfolded or improperly foldedprotein initially only to transfer the protein later toother molecular chaperones, e.g., Hsp90. We did notobserve any signi®cant amount of Hsp90 in proteincomplexes containing Mos. We believe that like Hsp90in the regulation of Src, Raf, Cdk4 and Wee1 proteinkinases (Stancato et al., 1993; Xu and Lindquist, 1993;Aligue et al., 1994; Schulte et al., 1995; Stepanova etal., 1996), Hsp70 is important in the regulation of c-Mos function. One possibility is that Mos isintrinsically unstable and it needs to stay associatedwith Hsp70 chaperone in order to be protected fromproteolysis. Such a model involving continuedinteraction between a protein and a molecularchaperone would not be unique for Mos; a similarsituation may exist in case of several proteins andmolecular chaperones (reviewed in Pennisi, 1996).

Previously, gel ®ltration experiments from ourlaboratory showed that c-Mos ectopically expressedin somatic cells fractionates predominantly as a500 kDa complex (Bai et al., 1992b). In general,Hsp70 is able to interact with other molecularchaperones, and thus it fractionates as a largemultiprotein complex (reviewed in Pennisi, 1996 andFrydman and HoÈ hfeld, 1997). This property of Hsp70and ATP requirement for its function would beconsistent with the large size of c-Mos-containingprotein complexes and the high amount of ATPrequired to preserve c-Mos protein kinase activity.However, it is also possible that Hsp70 associates withc-Mos in the early stages of multiprotein complex

assembly and is absent from the ®nal c-Mos-contain-ing protein complex. In addition to Hsp70, a numberof proteins (p60, p35, p30 and p25) were foundpresent in protein complexes containing Mos. Whetherany of these proteins are additional molecularchaperones remains to be seen. Interestingly, the p60protein also associates with Hsp70 (Figure 1 and datanot shown). In all our experiments, thus far, p60coprecipitation with Mos parallels Hsp70 coprecipita-tion; we have been unable to dissociate one withoutdissociating the other from Mos. As a separate point,since some other molecular chaperones in addition toHsp70 also require ATP for their function, we cannotconclude that the requirement of ATP (in mM range)for the protection of Mos activity is due to Hsp70only. Furthermore, ATP may play multiple additionalroles in cell extracts; it is possible that ATP coulddirectly or indirectly (e.g., by a�ecting molecularchaperones) in¯uence the phosphorylation state ofMos and/or other proteins. Although 3 mM ATPprotects the loss of CSF activity in egg extracts(Shibuya and Masui, 1988, 1989), the precise role(s) ofATP in this system is not known.

Our results showing that the S3A mutation a�ectsMos-Hsp70 interaction is signi®cant considering thatthe S3A mutation in Xenopus c-Mos inhibits theinteraction between the c-Mos catalytic domain andits substrate MEK1 (Chen and Cooper, 1995). TheS3A mutation also inhibits the kinase activity of v-Mos(Yang et al., 1998). At this time, our results indicate acorrelation between in vitro Hsp70 interaction andkinase activity of Mos. Further studies are needed toappreciate the full scope of Mos-Hsp70 interaction.Our approach is to identify various Mos-interactingproteins so that the biochemical role of a givenprotein ± protein interaction, e.g. Mos-Hsp70, can bebetter understood in relation to other Mos-interactingproteins. At present we also do not know how the S3Amutation a�ects Mos interaction with Hsp70. Is Ser-3 apart of Hsp70 interaction motif or more likely Ser-3mediates other protein ± protein interactions with c-Mos which in turn in¯uence Hsp70 association with c-Mos? Future studies will provide answers to this andother questions regarding the role of Hsp70 in theregulation of c-Mos function. These future studies willalso explain whether pleotropic e�ects of the Ser-3phosphorylation on Mos protein kinase and Mosturnover are due to Hsp70 association.

We observed that the Hsp70 interaction was weakerwith v-Mos than with Mos. This observation would beconsistent with the possibility suggested by our dataobtained with the Ser-3 mutants (Figures 9 and 10)that the amino-terminus of Mos (or c-Mos) isimportant for mediating Hsp70 interaction. Anotherevidence for the involvement of amino-terminalsequence of Mos in Hsp70 interaction comes fromthe less and weaker association of Hsp70 observed inanti-Mos(6-24) immunoprecipitates than in anti-Mos(332 ± 343) immunoprecipitates; strength of pro-tein ± protein interaction was determined by high-saltand detergent washes (H Liu, VB Vuyyuru, CD Phamand B Singh, unpublished observations). Our inter-pretation of these data is that the antibody bindingnear the amino-terminus, analogous to the e�ect of theamino acid sequence extension in v-Mos, weakensHsp70 interaction. The catalytic domain of Mos may

Mos association with Hsp70H Liu et al

3468

not be involved signi®cantly in Hsp70 interaction sinceHsp70-bound Mos is active as a protein kinase (Figure6). c-Mos (or Mos) behaves di�erently from v-Mos inseveral ways: c-Mos is incorporated into largermacromolecular complexes more e�ciently than v-Mos (Bai et al., 1992b); c-Mos associates withmicrotubules better than v-Mos (Zhou et al., 1991b;Bai et al., 1992); and c-Mos is localized into thenucleus more e�ciently than v-Mos (Zhou et al.,1991b). Whether any or all of these properties of c-Mos are due to stronger interaction with and properchaperoning by Hsp70 remains to be seen.

There are several Hsp70 homologs, and some areexpressed in a tissue-speci®c manner. In future studies,determining which Hsp70 homolog or homologsassociate with c-Mos in germ cells will be important.We view the c-Mos-Hsp70 interaction as resulting froma specialized function of Hsp70 that may be similar tothe unique function of Hsp70 ± 2 in male germ cells(Dix et al., 1996). In addition to ®ve Hsp70s commonto germ cells and somatic cells of mammals,spermatogenic cells synthesize Hsp70 ± 2 during meio-sis. In Hsp70 ± 27/7 mice that are otherwise normal, theonly defect is in the meiotic divisions of male germ cells(Dix et al., 1996). Mutagenesis studies currently inprogress in our laboratory are directed at under-standing further the molecular basis and biochemicalrole of Mos-Hsp70 interaction. We believe that theidenti®cation of Hsp70 as a Mos-interacting proteinwill prove to be important in unraveling the regulationof c-Mos function in eggs. Mos localizes to centromere(Wang et al., 1994) but the molecular basis of thislocalization is not yet known. It is conceivable thatMos-Hsp70 association plays a role in the assemblyand subcellular localization of Mos to centromere. Inthis regard, it is interesting that a centromere protein,centrin, also associates with heat shock proteins,including Hsp70 in CSF-arrested Xenopus oocytes(Uzawa et al., 1995).

Materials and methods

Plasmids

Full-length v-mos gene of Mo-MuSV124 and the deletionmutant that would encode Mos, both cloned in theexpression vector pJC119 (Hannink and Donoghue, 1985;Singh et al., 1986), were kindly provided by Daniel JDonoghue (University of California, San Diego, CA, USA)as was the kinase-inactive mutant of v-Mos (K121R)

(Hannink and Donoghue, 1985). The amino-terminaldeletion mutant of the K121R v-Mos was generated bypolymerase chain reaction (PCR). Similarly, mouse c-mos wasampli®ed by PCR using pMS1 (Blair et al., 1981) plasmidDNA as a template and cloned into the XhoI site of pJC119.The sense and antisense primers were designed also tointroduce the XhoI site for cloning. The mutations of the Ser-3 residue of Mos were generated by introducing the S3A andS3E mutations in the PCR primer. The DNA sequence wascon®rmed by sequencing the cloned gene.

Protein analysis

The plasmids were transfected into COS-1 cells by theDEAE-dextran method as described (Sambrook et al., 1989).Two days later, the cells were either labeled with 35S-methionine for 30 min or lysed for carrying out the Mosprotein kinase assay or Western immunoblotting as describedpreviously (Pham et al., 1995; Yang et al., 1996). In allexperiments, cell lysis was carried out by douncing or freezing(at 7808C) and thawing once in a bu�er containing 1% NP-40 (Singh et al., 1988). Immunoprecipitation was carried outwith various v-Mos antipeptide antibodies (Singh et al., 1990)or Hsp70 antibodies (W27 clone from Santa CruzBiotechnology, Santa Cruz, CA, USA and BB70 clone fromDr David Toft, Mayo Clinic Foundation, Rochester, MN,USA). Unfertilized Xenopus laevis eggs frozen on dry ice wereobtained from NASCO (Fort Atkinson, WI, USA). For co-immunoprecipitation analysis, the eggs were lysed bydouncing in the same bu�er as COS-1 cells. c-Mosxe wasimmunoprecipitated with the C237 antipeptide antibodyproduced in rabbit (Santa Cruz Biotechnology). Immunopre-cipitates were washed extensively in a bu�er containing 0.1%NP-40 (Singh et al., 1988).

The V8 protease partial digestion analysis of the 35S-methionine-labeled Mos and Hsp70 was carried out by thestandard procedure as described previously (Maxwell andArlinghaus, 1985b). The kinase assay using anti-v-Mos(37 ±55) or anti-Hsp70 immunoprecipitates and GST-MEK1(K97R) as a Mos kinase substrate was done asdescribed previously (Yang et al., 1996). Quantitation ofrelative signal in protein bands in various experiments wasperformed by densitometry of the bands on X-ray ®lm and/orliquid scintillation counting of the excised gel slices.

AcknowledgmentsWe thank D Toft (Mayo Clinic Foundation, Rochester,MN, USA) for providing the BB70 Hsp70 antibody, RBArlinghaus for suggestions and critical reading of themanuscript, L Feldman for editorial corrections and LKimbrough for her help in manuscript preparation. Thiswork was supported by the grants R01 CA45125 andCA16672 (core) from the National Institutes of Health.

References

Aligue R, Akhavan-Niak H and Russell P. (1994). EMBO J.,13, 6099 ± 6106.

Bai W, Singh B, Yang Y and Arlinghaus RB. (1992a).Oncogene, 7, 1757 ± 1763.

Bai W, Singh B, Yang Y, Ramagli LS, Nash M, Herzog NKand Arlinghaus RB. (1992b). Oncogene, 7, 493 ± 500.

Bai W, Arlinghaus RB and Singh B. (1993). Oncogene, 8,2207 ± 2212.

Blair DG, Oskarsson M, Wood TB, McClements WL,Fischinger PJ and Vande Woude GF. (1981). Science,21, 941 ± 943.

Bukau B, Hesterkamp T and Luirink J. (1996). Trends CellBiol., 6, 480 ± 486.

Chen M and Cooper JC. (1995). Mol. Cell. Biol., 15, 4727 ±4734.

Chen M, Li D, Krebs EG and Cooper JA. (1997). Mol. Cell.Biol., 17, 1904 ± 1912.

Colledge WW, Carlton MBL, Udy GB and Evans MJ.(1994). Nature, 370, 65 ± 68.

Dix DJ, Allen JW, Collins BW, Mori C, Nakamura N,Poorman-Allen P, Goulding EH and Eddy EM. (1996).Proc. Natl. Acad. Sci. USA, 93, 3264 ± 3268.

Freeman RS, Kanki JP, Ballantyne SM, Pikham KM andDonoghue DJ. (1990). J. Cell Biol., 111, 533 ± 541.

Freeman RS, Meyer AN, Li J and Donoghue DJ. (1992). J.Cell Biol., 116, 725 ± 735.

Mos association with Hsp70H Liu et al

3469

Frydman J and HoÈ hfeld J. (1997). Trends Biochem. Sci., 22,87 ± 92.

Hannink M and Donoghue DJ. (1985). Proc. Natl. Acad. Sci.USA, 82, 7894 ± 7898.

Hartl FU. (1996). Nature, 381, 571 ± 580.Hashimoto N, Watanabe N, Furuta Y, Tamemoto H, Sagata

N, Yokoyama M, Okazaki K, Nagayoshi M, Takeda N,Ikawa Y and Aizawa S. (1994). Nature, 370, 68 ± 71.

Maxwell SA and Arlinghaus RB. (1985a). J. Gen. Virol., 66,2135 ± 2146.

Maxwell SA and Arlinghaus RB. (1985b). Virology, 143,321 ± 333.

Nebreda AR and Hunt T. (1993). EMBO J., 12, 1979 ± 1986.Nishizawa M, Furuno N, Okazaki K, Tanaka H, Ogawa Y

and Sagata N. (1993). EMBO J., 12, 4021 ± 4027.Nishizawa M, Okazaki K, Furuno N, Watanabe N and

Sagata N. (1992). EMBO J., 11, 2433 ± 2446.Papko� J, Verma IM and Hunter T. (1982). Cell, 29, 417 ±

426.Pennisi E. (1996). Science, 274, 1613 ± 1614.Pham CD, Arlinghaus RB, Zheng C-F, Guan K-L and Singh

B. (1995). Oncogene, 10, 1683 ± 1688.Pham CD, Rungta M, Chandler DS, Yang Y, Arlinghaus

RB, Rajagopalan S, Schwartz MR and Singh B. (1997).Int. J. Gynecol. Cancer, 7, 314 ± 317.

Posada JN, Yew N, Ahn NG, Vande Woude GF and CooperJA. (1993). Mol. Cell. Biol., 13, 2546 ± 2553.

Rassow J, von Ahsen O, BoÈ mer U and Pfanner N. (1997).Trends Cell Biol., 7, 129 ± 133.

Sagata N, Watanabe N, Vande Woude GF and Ikawa Y.(1989). Nature, 342, 512 ± 518.

Sambrook J, Fritsch EF and Maniatis T. (1989). Molecularcloning: a laboratory manual., Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY.

Schulte TW, Blagosklonny MV, Ingui C and Neckers L.(1995). J. Biol. Chem., 270, 585 ± 588.

Shibuya EK and Masui Y. (1988). Dev. Biol., 192, 253 ± 264.

Shibuya EK and Masui Y. (1989). Development, 106, 799 ±808.

Singh B, Al-Bagdadi F, Liu J and Arlinghaus RB. (1990).Virology, 178, 535 ± 542.

Singh B and Arlinghaus RB. (1997). In: Meijer L, Guidet Sand Philippe M, (eds). Progress in Cell Cycle Research Vol3. Plenum Press: New York, USA. Chapter 20, pp. 251 ±259.

Singh B, Hannink M, Donoghue DJ and Arlinghaus RB.(1986). J. Virol., 60, 1148 ± 1152.

Singh B, Herzog NK, Liu J and Arlinghaus RB. (1988).Oncogene, 3, 79 ± 85.

Stancato LF, Chow Y-H, Hutchison KA, Perdew GH, JoveR and Pratt WB. (1993). J. Biol. Chem., 268, 21711 ±21716.

Stepanova L, Leng X, Parker SB and Harper JW. (1996).Genes Dev., 10, 1491 ± 1502.

Uzawa M, Grams J, Madden B, Toft D and Salisbury JL.(1995). Dev. Biol., 171, 512 ± 559.

van Beveren C, van Straaten F, Galleshaw JA and VermaIM. (1981). Cell, 27, 97 ± 108.

Wang XM, Yew N, Peloquin G, Vande Woude GF andBorisy GG. (1994). Proc. Natl. Acad. Sci. USA, 91, 8329 ±8333.

Xu Y and Lindquist SL. (1993). Proc. Natl. Acad. Sci. USA,90, 7074 ± 7078.

Yang Y, Herrmann CH, Arlinghaus RB and Singh B. (1996).Mol. Cell. Biol., 16, 800 ± 809.

Yang Y, Pham CD, Vuyyuru VB, Liu H, Arlinghaus RB andSingh B. (1998). J. Biol. Chem., 273, 15946 ± 15953.

Zhou R, Daar I, Ferris DK, White G, Paules RS and VandeWoude G. (1992). Mol. Cell. Biol., 12, 3583 ± 3589.

Zhou R, Oskarsson M, Paules RS, Schultz N, Cleveland Dand Vande Woude GF. (1991a). Science, 251, 671 ± 675.

Zhou R, Rulong S, da Silva PP and Vande Woude GF.(1991b). Cell Growth Di�er., 2, 257 ± 265.

Mos association with Hsp70H Liu et al

3470