Cell Biology International 2002, Vol. 26, No. 7, 653–657doi:10.1006/cbir.2002.0882, available online at http://www.idealibrary.com on

SHORT COMMUNICATION

EVIDENCE THAT THE NOVOBIOCIN-SENSITIVE ATP-BINDING SITE OF THEHEAT SHOCK PROTEIN 90 (HSP90) IS NECESSARY FOR ITS

AUTOPHOSPHORYLATION

T. LANGER*, H. SCHLATTER† and H. FASOLD

Institut fur Biochemie der Johann Wolfgang Goethe-Universitat, Frankfurt am Main, Germany

Received 15 November 2001, accepted 8 February 2002

The 90 kDa heat shock protein (Hsp90) is one of the most abundant protein and essential for alleukaryotic cells. Many proteins require the interaction with Hsp90 for proper function. Uponheat stress the expression level of Hsp90 is even enhanced. It is assumed, that under theseconditions Hsp90 is required to protect other proteins from aggregation. One property of Hsp90is its ability to undergo autophosphorylation. The N-terminal domain of Hsp90 has been shownto contain an unusual ATP-binding site. A well-known inhibitor of Hsp90 function isgeldanamycin binding to the N-terminal ATP-binding site with high affinity. Recently it wasshown that Hsp90 possesses a second ATP-binding site in the C-terminal region, which can becompeted with novobiocin. Autophosphorylation of Hsp90 was analysed by incubation with�32P-ATP. Addition of geldanamycin did not interfere with the capability for autophosphor-ylation, while novobiocin indeed did. These results suggest that the C-terminal ATP-binding siteis required for autophosphorylation of Hsp90. � 2002 Elsevier Science Ltd. All rights reserved.

K: Hsp90; novobiocin; geldanamycin; autophosphorylation; ATP.

To whom correspondence should be addressed: Dr Thomas Langer,Institut fur Biochemie der Johann Wolfgang Goethe-UniversitatFrankfurt am Main, Marie Curie-Str. 9, 60439 Frankfurt am Main,Germany. Tel.: +049(0)69-798-29431; E-mail: t.langer@nmr.uni-frankfurt.de†Present address: Protcer & Gamble Service GmbH, SulzbacherStr. 50, 65823 Schwalbach am Taunus, Germany.

INTRODUCTION

One of the most abundant proteins in eukaryoticcells is the heat shock protein 90 (Hsp90). Inunstressed cells, it comprises about 1% of the totalcell protein (Welch and Feramisco, 1982). Afterexposure to stress conditions its synthesis is evenfurther enhanced (Borkovich et al., 1989). Hsp90 isa highly conserved protein. It belongs to the mol-ecular chaperones involved in refolding of unfoldedor denatured proteins. Besides its chaperone func-tion many proteins require interaction with Hsp90for proper function. One of the best characterizedHsp90-interacting proteins are steroid-hormonereceptors. Components of the cytoskeleton, the

1065–6995/02/$-see front matter

NO-synthase, immunophilins FKBP52 and Cyp40as well as several protein kinases, such as caseinkinase II, Raf, vScr, and Wee1 are other cellularproteins known to interact with Hsp90 (for reviewssee Pratt, 1997; Mayer and Bukau, 1999; Pearl andProdromou, 2000; Richter and Buchner, 2001;Young et al., 2001). Hsp90 occurs in vivo in aphosphorylated state. Lees-Miller and Anderson(1989a) determined the ration of 1.7 mol (�0.3)phosphate per mol Hsp90 protein. In a more recentstudy using MALDI-MS by Garnier et al. (2001) itwas demonstrated that the two Hsp90 isoforms,Hsp90� and Hsp90� are phosphorylated at differ-ent amounts. Hsp90� bears up to four phosphategroups, whereas only two phosphate groups havebeen detected in Hsp90� (Garnier et al., 2001).

The ability of CKII and the double strandedDNA activated kinase to phosphorylate Hsp90 iswell established (Lees-Miller and Anderson,1989a,b). Beside phosphorylation of Hsp90 due tothe action of protein kinases, Hsp90 can undergo

� 2002 Elsevier Science Ltd. All rights reserved.

654 Cell Biology International, Vol. 26, No. 7, 2002

Fig. 1. Phosphorylation of Hsp90 purified from rat liver in the presence of varying concentrations of geldanamycin (GA),AMPPNP and ATP. All chemicals were obtained from Sigma–Aldrich, unless otherwise indicated. Hsp90 was purified asdescribed (Langer and Fasold, 2001) and dissolved in incubation buffer (50 mM Tris/HCl, 2.5 mM MgCl2, 5 mM NaCl, 2.5 mMNa2HPO4, 25 mM KCl, 0.5 mM CaCl2, 2 mM DTT, 0.5 mM PMSF, pH 7.4). Protein concentration was adjusted to 1 mg/mlusing ultrafiltration (Centripreps, 10 kDa nominal molecular weight limit, Amicon, Bedford, Mass., U.S.A.) as determined withthe aid of the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, Calif., U.S.A.). 0.5 �l �32P-ATP (100 mCi/ml, specificactivity 30 Ci/mmol, Amersham Pharmacia, Uppsala, Sweden) were added to 30 �l of the protein solution. AMPPNP, ATP andgeldanamycin (obtained from the National Cancer Institute (NCI), U.S.A.) were added to final concentrations as indicated.Dimethylsulfoxide (DMSO) was used as a solvent for geldanamycin. Samples were incubated at 37�C for 1 h and after additionof 20 �l SDS–PAGE sample buffer heated at 60�C for 15 min. 20 �l of each assay was loaded on an SDS–PAGE. SDS–PAGEwas performed as described (Laemmli et al., 1971). After staining with Coomassie Brilliant Blue R250 gels were dried andinserted into a PhosphoImager exposure cassette (Molecular Dynamics). After 2 h exposition plates were analysed with aPhosphoImager 445 SI (Molecular Dynamics) and image processing was done on a personal computer using the programImageQuaNT V.4.2 (Molecular Dynamics). (A) Coomassie stained SDS–PAGE. (B) Autoradiogram of the SDS–PAGE shownin A. ctr.: control with no further additions. M: protein molecular weight marker (Bio-Rad), corresponding molecular weightsare indicated.

Cell Biology International, Vol. 26, No. 7, 2002 655

autophosphorylation (Csermely and Kahn, 1991;Nadeau et al., 1993). Autophosphorylation hasalso been demonstrated for endoplasmin (grp94), amember of the Hsp90 protein family located in thelumen of the endoplasmatic reticulum (Csermelyet al., 1995). Hsp90 contains an unusual ATP-binding site at its N-terminal domain, which is alsothe binding site for the benzoquinon ansamycingeldanamycin (Prodromou et al., 1997; Stebbinset al., 1997; Roe et al., 1999). It is assumedthat Hsp90 requires ATP-hydrolysis to exert itschaperone function efficiently (for review of thesubstrate-binding ATPase cycle of Hsp90 seeYoung et al., 2001). Nevertheless, also ATP-independent chaperone activity of Hsp90 has beenreported (Wiech et al., 1992; Miyata and Yahara,1995; Freeman and Moromoto, 1996; Yoneharaet al., 1996; Minami et al., 2001).

Fig. 2. Phosphorylation of Hsp90 purified from rat liver in the presence of unlabelled ATP as well as geldanamycin. UnlabelledATP was added to all assays at a final concentration of 5 mM. Geldanamycin was added at the indicated concentrations.Experiments were performed as described in the legend of Fig. 1. (A) Coomassie-stained SDS–PAGE. (B) Autoradiogram of theSDS–PAGE shown in A. ctr.: DMSO was added in corresponding amounts only. M: protein molecular weight marker(Bio-Rad), corresponding molecular weights are indicated.

RESULTS AND DISCUSSION

We were interested in further defining the auto-phosphorylation properties of Hsp90. First,autophosphorylation of Hsp90 was analysed withthe aid of radioactive �32P-ATP and non-

radioactive competitors for the ATP-binding-site, namely geldanamycin, the non-hydrolyseableATP-analogue AMPPNP and an excess of ATP.Interestingly, only AMPPNP and ATP were able tocompete for phosphorylation of Hsp90, whereaseven 5 mM geldanamycin did not affect the phos-phorylation state of Hsp90 (Fig. 1). In addition tothe band in the autoradiogram caused by the32P-phosphorylated Hsp90, other bands werebecoming visible as well. Hence, it cannot be ruledout that a contaminant kinase might also bepresent in the Hsp90 preparation. In contrast toprotein kinases, which usually have a high affinityto ATP, Hsp90 has a remarkable low affinity toATP. Nadeau et al. (1993) calculated a kcat of2.5�103 min�1 for the autophosphorylation ofHsp90. The ATP-concentration for half-maximalbinding of ATP to Hsp90 was determined to be400 �M (Scheibel et al., 1997). In order to effi-ciently compete the elusive contaminant kinases inthe Hsp90 preparation, phosphorylation of Hsp90was analysed in the presence of an excess of non-radioactive ATP as well as raising concentration ofgeldanamycin. Under these conditions, only bandscorresponding to Hsp90 can be detected in theautoradiogram. Since Hsp90 binds geldanamycin

656 Cell Biology International, Vol. 26, No. 7, 2002

Fig. 3. Phosphorylation of Hsp90 purified from rat liver in the presence of novobiocin. Novobiocin (Sigma–Aldrich) was addedfrom a 10 mM stock-solution in incubation buffer (see above) and added to final concentrations as indicated in the figure. Forexperimental procedures see legend of Fig. 1. (A) Coomassie-stained SDS–PAGE. (B) Autoradiogram of the SDS–PAGE shownin A. ctr.: control with no further additions. M: protein molecular weight marker (Bio-Rad), corresponding molecular weightsare indicated.

with nanomolar affinity (Young et al., 2001) thebinding of �32P-ATP to the N-terminal ATP-binding site under the experimental conditionsshown in Figure 2 should be efficiently competed.The question raised on the basis of these results is:What is the cause for Hsp90 phosphorylation?Marcu et al. (2000a) recently reported the specificbinding of the antibiotic novobiocin to Hsp90. The

binding site for novobiocin lies within theC-terminal region of Hsp90 and the binding tonovobiocin can be competed with ATP (Marcuet al., 2000b). Considering these findings, thephosphorylation of Hsp90 was analysed in thepresence of raising concentrations of novobiocin.Addition of novobiocin is sufficient to completelysuppress phosphorylation of Hsp90 (Fig. 3). The

Cell Biology International, Vol. 26, No. 7, 2002 657

presented results suggest that the secondC-terminal–ATP-binding site is responsible for theautophosphorylation of Hsp90. As demonstratedin Figure 2, geldanamycin does not interfere withautophosphorylation of Hsp90 and hence does notbind to the C-terminal ATP-binding site of Hsp90.The origin of the additional bands appearing in theautoradiogram besides the bands corresponding tothe phosphorylated Hsp90 as shown in Figs 1 and3 is unknown. Recently, also a kinase function forHsp90 has been reported (Park et al., 1998). Aspeculative interpretation is that the additional 32Plabelled bands stem from contaminant proteinsbeing phosphorylated by Hsp90.

ACKNOWLEDGEMENTS

This work was supported by a grant from theDeutsche Forschungsgemeinschaft SFB474 Intra-zellulare Transport- und Regulationsvorgange.

REFERENCES

B KA, F FW, F DB, T J,L S, 1989. Hsp82 is an essential protein that isrequired in higher concentrations for growth of cells athigher temperatures. Mol Cell Biol 9: 3919–3930.

C P, K CR, 1991. The 90 kDa heat shock protein(Hsp90) possesses an ATP-binding site and autophosphor-ylating activity. J Biol Chem 266: 4943–4950.

C P, M Y, S T, Y I, 1995.Autophosphorylation of grp94 (endoplasmin). J Biol Chem270: 6381–6388.

F BC, M RI, 1996. The human cytosolicmolecular chaperones Hsp90, Hsp70 (Hsc70) and Hdj-1have distinct roles in recognition of a non-native protein andprotein folding. EMBO J 15: 2969–2979.

G C, L D, J TJ, J ON, B C,P V, 2001. Phosphorylation and oligomerization ofnative pig brain Hsp90 studied by mass spectrometry. Eur JBiochem 268: 2402–2407.

L UK, 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 227:680–686.

L T, F H, 2001. Isolation and quantification of theheat shock protein 90 alpha and beta isoforms from rat liver.Protoplasma 218: 54–56.

L-M SP, A CW, 1989a. Two human 90 kDaheat shock proteins are phosphorylated in vivo at conservedserins that are phosphorylated in vitro by casein kinase II.J Biol Chem 264: 2431–2437.

L-M SP, A CW, 1989b. The human double-stranded DNA-activated protein kinase phosphorylates

the 90 kDa heat shock protein, Hsp90alpha, at twoNH2-terminal residues. J Biol Chem 264: 17252–17280.

M MG, C A, B I, C M, NLM, 2000b. The heat shock protein 90 antagonist novo-biocin interacts with a previously unrecognized ATP-binding domain in the C-terminus of the chaperone. J BiolChem 275: 37181–37186.

M MG, S TW, N L, 2000a. Novobiocinand related coumarins and depletion of heat shock protein90-dependent signaling proteins. J Natl Cancer Inst 92:242–248.

M MP, B B, 1999. Molecular chaperons: The busylife of Hsp90. Curr Biol 9: R322–R325.

M M, N M, E Y, M Y, 2001. Boththe N- and C-terminal chaperone sites of Hsp90 participatein protein refolding. Eur J Biochem 268: 2520–2514.

M Y, Y I, 1995. Interaction between casein kinaseII and the 90 kDa stress protein Hsp90. Biochemistry 34:8123–8129.

N K, D A, W CT, 1993. Hsp90 chaperoninspossess ATPase activity and bind heat shock transcriptionfactors and peptidyl prolyl isomerases. J Biol Chem 268:1479–1487.

P M, Y-K C, K P, 1998. Brassica napusHsp90 can autophosphorylate and phosphorylates otherprotein substrates. Mol Cell Biochem 185: 33–38.

P LH, P C, 2000. Structure and in vivo functionof Hsp90. Curr Opin Struct Biol 10: 46–51.

P WB, 1997. The role of the Hsp90-based chaperonesystem in signal transduction by nuclear receptors andreceptors signaling via MASP kinase. Annu Rev PharmacolToxicol 37: 297–326.

P C, R SM, O’B R, L JE, P PW,P LH, 1997. Identification and structural characteriz-ation of the ADP/ATP binding site in the Hsp90 molecularchaperone. Cell 90: 65–75.

R K, B J, 2001. Hsp90: Chaperoning signaltransduction. J Cell Physiol 188: 281–290.

R SM, P C, O’B R, L JE, P PW,P LH, 1999. The structural basis for inhibition of theHsp90 molecular chaperone by the anti-tumor antibioticsradicicol and geldanamycin. J Med Chem 42: 260–266.

S T, N S, W T, M C, R J,V PD, B J, 1997. ATP-binding properties ofhuman Hsp90. J Biol Chem 272: 18008–18013.

S CE, R AA, S C, R N, H FU,P NP, 1997. Crystal structure of an Hsp90-geldanamycin complex: targeting of a protein chaperone byan antitumour agent. Cell 89: 239–250.

W WJ, F JR, 1982. Purification of themajor mammalian heat shock proteins. J Biol Chem 257:14959–14959.

W H, B J, Z M, Z R, JU, 1992. Hsp90 chaperones protein folding in vitro. Nature358: 169–170.

Y M, M Y, K Y, Y I, 1996. Heatinduced chaperone activity of Hsp90. J Biol Chem 271:2641–2645.

Y JC, M I, H FU, 2001. Hsp90: a specializedbut essential protein folding tool. J Cell Biol 154: 267–273.