Oncogene (2000) 19, 1477 ± 1484 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc p53-independent association between SV40 large T antigen and the major cytosolic heat shock protein, HSP90

Yoshihiko Miyata*,1,3 and Ichiro Yahara1,2

1Department of Cell Biology, The Tokyo Metropolitan Institute of Medical Science, 3-18-22, Hon-Komagome, Bunkyo-ku, Tokyo 113-8613, Japan; 2CREST, Japan Science and Technology Corporation, Kawaguchi, Saitama, 332-0012, Japan

The simian double strand DNA tumor virus SV40 and p33cdk2 (Adamczewski et al., 1993). Most of these encodes the 90-kDa multi-functional protein, large T molecules are known to regulate the cell cycle and cell antigen (LT). LT functions by binding to DNA, as well growth, and the interaction of LT with these cellular as to many cellular target proteins such as p53 and proteins is crucial for tumorigenesis. retinoblastoma protein (pRB). We report here the Many viral proteins are known to be associated with identi®cation of a cellular heat shock protein, HSP90, heat shock proteins (HSPs) or molecular chaperones of as a previously undescribed LT-associated protein. host cells. For example, HSP90 associates with viral Immunoprecipitates by anti-HSP90 antibodies from oncoprotein pp60v-src (Brugge et al., 1981; Oppermann LT-expressing cell lysates contained LT protein, as et al., 1981) and hepatitis B virus reverse transcriptase revealed by Western blotting. Conversely, anti-LT (Hu and Seeger, 1996). LT is reported to bind the antibody co-immunoprecipitated HSP90. Co-immunopre- major heat shock protein, HSP70 (Sawai and Butel, cipitation of HSP90 and LT was observed even after 1989). More surprisingly, a previous study has complete immuno-depletion of p53, indicating that the demonstrated that the amino acid sequence of the N- association of LT with HSP90 is p53-independent. LT- terminal region of LT is homologous to the J domain HSP90 complexes can be reconstituted from puri®ed of the DnaJ family of molecular chaperones (Kelley HSP90 and unfolded-LT in vitro in an ATP-independent and Landry, 1994). In fact, this region is capable of manner but not from HSP90 and native LT, suggesting functioning as a J domain (Srinivasan et al., 1997) and that non-mature conformation of LT is required for the is important for ecient viral DNA replication ecient association with HSP90. Moreover, geldanamy- (Campbell et al., 1997), all of which suggests a cin, an anti-tumor drug that speci®cally binds and relationship between the LT function and molecular inhibits HSP90, reduced the intracellular concentration chaperones. of LT by destabilizing newly synthesized LT. The above HSP90 is ubiquitously and abundantly distributed in results suggest that HSP90 associates with immature all organisms from bacteria to humans. The amino acid forms of LT both in vivo and in vitro, and thus might sequence of HSP90 is highly conserved among species, assist LT in the formation of a functional, mature suggesting that HSP90 plays a fundamental role in cells structure. Oncogene (2000) 19, 1477 ± 1484. (Lindquist and Craig, 1988; Welch, 1992). Signi®cant progress has been made recently on the structure and Keywords: stress protein; tumor antigen; molecular function of HSP90 (Csermely et al., 1998; Toft, 1998; chaperone; SV40; p53; geldanamycin Yahara et al., 1998; Buchner, 1999; Mayer and Bukau, 1999), although the precise mechanism underlying the way in which HSP90 functions as a molecular Introduction chaperone remains largely unknown. HSP90 is known to interact with a variety of cellular and viral proteins The simian virus 40 (SV40) Large T antigen (LT) is a (Rutherford and Zuker, 1994). The most well-char- 90-kDa phosphoprotein that is essential for the viral acterized cellular targets of HSP90 are transcription replication and transformation of host cells (Fanning, factors, such as steroid hormone receptors (Joab et al., 1992; Pipas, 1992; Ludlow, 1993). LT possesses various 1984; Catelli et al., 1985; Sanchez et al., 1985), MyoD1 biochemical functions such as ATP-binding, ATPase, (Shaknovich et al., 1992), and heat shock factor (Zou DNA-binding, and DNA-helicase activities (Tjian et et al., 1998). Likely the largest family of HSP90-targets al., 1980; Pipas, 1992). LT has been shown to bind to a consists of protein kinases, such as casein kinase II series of host cellular proteins including p53 (Lane and (Miyata and Yahara, 1992), Raf1 (Stancato et al., Crawford, 1979; Linzer and Levine, 1979), retinoblas- 1993), sevenless tyrosine kinase (Cutforth and Rubin, toma family proteins, such as pRB and p107 1994), and Wee1 (Aligue et al., 1994). The growing list (DeCaprio et al., 1988; Dyson et al., 1989), DNA- of cellular targets of HSP90 now includes endothelial polymerase a (Smale and Tjian, 1986), transcriptional nitric oxide synthase (Garcõ a-CardenÄ a et al., 1998) and coactivators (Avantaggiati et al., 1996), and cyclin A telomerase (Holt et al., 1999). Among the proteins associated with HSP90, of special interest are proteins that translocate from the cytosol to the nucleus, where they bind DNA. HSP90 *Correspondence: Y Miyata has been suggested to modulate the function and 3Current address: Department of Cell and Developmental Biology, subcellular distribution of target proteins. Functional Graduate School of Biostudies, Kyoto University, Kitashirakawa folding, cellular localization, and DNA-binding of Oiwake-cho, Sakyo-ku, Kyoto 606-8502, Japan Received 6 September 1999; revised 20 January 2000; accepted 20 nuclear proteins have been shown to be regulated by January 2000 HSP90 (Shaknovich et al., 1992; Hendrick and Hartl, Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1478 1993; Jakob and Buchner, 1994; Miyata and Yahara, Co-immunoprecipitation of HSP90 with LT 1995; Zou et al., 1998; Holt et al., 1999). We report here evidence that SV40 LT is associated with HSP90 Cell extracts were prepared from three LT-expressing in vivo, and that this association can be reconstituted cell lines (COS1, COS7, and VA4) or from control cell from puri®ed HSP90 and puri®ed unfolded LT in vitro. lines (CV1 and HeLa). HSP90 was immunoprecipitated Consistent with this ®nding, the anti-cancer drug along with its associated proteins from cell extracts and geldanamycin, which is known to function as an resolved by SDS ± PAGE. Nearly equivalent amounts HSP90-speci®c inhibitor, decreased the intracellular of HSP90 were immunoprecipitated from the extracts amount of LT. of these cell lines, as indicated by silver staining (Figure 2a). Western blotting of the immunoprecipitates with anti-LT (PAb419) revealed the presence of LT only in the HSP90-immunocomplexes from SV40-transformed Results cells (Figure 2b, lanes 3 ± 5). No signal was detected in the immunocomplexes from cell lines that do not Specificity of an antibody against LT and antibodies express LT (Figure 2b, lanes 1 and 2). LT and HSP90 against HSP90 were also co-immunoprecipitated from lysates of 3Y1 Crude cell extracts from parental CV1 and SV40- cells transformed with SV40 (A640-3Y1) and from transformed COS7 cells were separately prepared. those transformed with small t antigen-depleted SV40 HSP90 and LT were puri®ed to homogeneity as (dl884-3Y1) (data not shown), indicating that the small described in Materials and methods. The cell lysates t antigen is not responsible for this co-immunopreci- and the puri®ed proteins were subjected to sodium pitation. dodecyl sulfate polyacrylamide gel electrophoresis In a subsequent experiment, PAb419 was used to (SDS ± PAGE) and Western blotting with anti-HSP90 immunoprecipitate LT along with its associated proteins or anti-LT antibodies. As shown in Figure 1a, anti- from the cell extracts described above, and the co- HSP90 antibodies bound to a 90-kDa protein in cell immunoprecipitation of HSP90 was examined by extracts of CV1 (lane 1) and COS7 (lane 2) and to Western blotting. LT was immunoprecipitated only puri®ed HSP90 (lane 3), but not to puri®ed LT (lane from extracts of LT-expressing cell lines as shown by 4). Anti-LT PAb419 monoclonal antibody recognized silver staining (Figure 3a, lanes 3 ± 5). Western blotting the 90-kDa LT protein as well as the 23 kDa small t with anti-HSP90 antibodies revealed that HSP90 was co- antigen expressed only in COS7 cells (Figure 1b, lane immunoprecipitated with LT from extracts of COS1, 2) and also puri®ed LT (lane 4). On the other hand, COS7, and VA4 cell lines (Figure 3b, lanes 3 ± 5) but not PAb419 did not bind to any protein in CV1 cells from those of CV1 and HeLa (Figure 3b, lanes 1 ± 2). (Figure 1b, lane 1) nor to puri®ed HSP90 (lane 3). These ®ndings taken together indicated that LT is These ®ndings clearly indicated that antibodies to associated with HSP90 in SV40-transformed cells. HSP90 and PAb419 speci®cally recognize only the corresponding target proteins. It should be noted that Association between HSP90 and LT is p53-independent the two proteins showed nearly identical mobility on SDS polyacrylamide gels. We also con®rmed that these In addition to the well-characterized association antibodies speci®cally immunoprecipitated the corre- between LT and p53 (Lane and Crawford, 1979; sponding protein (data not shown).

Figure 2 Co-immunoprecipitation of LT with HSP90. (a) HSP90-immunocomplexes were prepared from ®ve cell lines (1, CV1; 2, HeLa; 3, COS1; 4, COS7; 5, VA4). Proteins in the immunocomplexes were analysed by SDS ± PAGE followed by Figure 1 Speci®city of the antibodies against HSP90 or LT. Cell silver staining. The position of HSP90 is indicated. (b) The same extracts (50 mg protein) from parental CV1 cells (lane 1) and LT- set of HSP90-immunocomplexes from ®ve cell lines as in (a) were expressing COS7 cells (lane 2), 100 ng of puri®ed HSP90 (lane 3) separated by SDS ± PAGE, and LT in the immunocomplexes was and puri®ed LT (lane 4) were subjected to SDS ± PAGE. Western revealed by anti-LT Western blotting. The position of LT is blotting with anti-HSP90 antibodies (a) or with anti-LT antibody indicated. The heavy (*50 kDa) and light (*30 kDa) chains of PAb419 (b) was performed. The positions of HSP90 (a) and T- immunoglobulin were also revealed because of the cross-reactivity antigens (b) along with molecular weight markers are indicated of the second antibody

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1479 Linzer and Levine, 1979), HSP90 is also known to be associated with certain mutant forms of p53 (Blagosk- lonny et al., 1996; Whitesell et al., 1998). Thus, the possibility exists that the observed co-immunoprecipi- tation of HSP90 with LT could be mediated by p53. We performed immunodepletion experiments to exam- ine the role of p53 in the HSP90-LT association. When cell lysates were subjected to immunodepletion by anti- LT antibody, essentially all of the LT and p53 were removed from the lysates and no additional LT or p53 was immunoprecipitated from the immunodepleted lysates by corresponding antibodies (Figure 4a, lanes 4 ± 6). p53 was completely immunodepleted with anti- p53 antibody (Figure 4a, lane 8 and 9), whereas the majority of LT was still present in the lysates (Figure 4a, lane 8). This may be explained by the fact that the intracellular amount of LT largely exceeds that of p53 in COS7 cells. Next, we examined the co-immunopre- cipitation of HSP90 with LT using the immunode- pleted cell lysates described above. An association between HSP90 and LT was clearly observed in the non-depleted lysates (Figure 4b, lane 2). LT-associated HSP90 was completely immunodepleted by anti-LT antibody as expected (Figure 4b, lane 5). Importantly, we observed co-immunoprecipitation of HSP90 with LT even after the complete immunodepletion of p53 (Figure 4a,b, lane 8). This result indicates that HSP90 is associated with LT in the absence of p53, suggesting a direct binding of LT with HSP90. HSP90 was also observed in p53-immunocomplexes from the non- depleted lysate (Figure 4b, lane 3) and this co- immunoprecipitation may be attributable to the binding of HSP90 with p53-associated LT. In fact, the amounts of HSP90 in the immunocomplexes (Figure 4b, lanes 2 and 3) were proportional to the amounts of LT immunoprecipitated by anti-LT and anti-p53 antibody, respectively (Figure 4a, lanes 2 and 3). These results suggest the existence of p53-LT- HSP90 ternary complexes. In agreement with this supposition, the amount of HSP90 was found to be appreciably reduced from p53-depleted lysates when compared to the control (Figure 4b, lane 2 vs lane 8). We then examined if HSP70 is included in the immunocomplexes containing HSP90 and LT. Immuno- precipitates were prepared from COS7 cell lysates with antibodies against HSP90 or LT, and then the immunocomplexes were examined by Western blotting with anti-HSP70 antibody. As shown in Figure 4c, HSP70 was detected in both anti-HSP90 and anti-LT Figure 4 (a) Immunodepletion of p53. 35S-Met/Cys-labeled COS7 cell lysates were immunodepleted by anti-LT or anti-p53 antibody along with non-immune antibody as indicated on the top, and the depleted lysates were then used for immunoprecipita- tion with the antibodies indicated at the bottom. The immunoprecipitates were analysed by SDS ± PAGE and ¯uoro- graphy. The positions of LT, p53, small t antigen, and a non- speci®cally bound protein (indicated by an asterisk) are shown on the right, while the positions of molecular weight markers are shown on the left. (b) Co-immunoprecipitation of HSP90 with LT Figure 3 Co-immunoprecipitation of HSP90 with LT. (a) LT- in the depleted lysates. COS7 cells were immunodepleted and then immunocomplexes were isolated from ®ve cell lines (1, CV1; 2, immunoprecipitated as described in (a), and the immunoprecipi- HeLa; 3, COS1; 4, COS7; 5, VA4). Proteins in the immunocom- tates were examined for co-immunoprecipitation of HSP90 by plexes were analysed by SDS ± PAGE followed by silver staining. Western blotting. A part of the membrane corresponding to A part of the gel corresponding to *90 kDa was shown and the *90 kDa is shown and the position of HSP90 is indicated. (c) position of LT is indicated. (b) The same set of LT- Co-immunoprecipitation of HSP70 with LT and HSP90. Im- immunocomplexes from ®ve cell lines as in (a) were separated munocomplexes were prepared from COS7 cell lysates with anti- by SDS ± PAGE and the amount of HSP90 in the immunocom- LT or anti-HSP90 antibodies, and the immunoprecipitates were plexes was revealed by anti-HSP90 Western blotting. A part of examined for co-immunoprecipitation of HSP70 by Western the membrane corresponding to *90 kDa was shown and the blotting. A part of the membrane is shown and the positions of position of HSP90 is indicated HSP70 and immunoglobulin heavy chain (HC) are indicated

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1480 immunocomplexes. This result is in good agreement from mixtures containing either unfolded LT or folded with previous reports that LT is associated with HSP70 LT irrespective of the presence or absence of HSP90 in other systems (Sawai and Butel, 1989; Yang and (Figure 5b, lanes 1, 2, 4, and 5). These results taken DeFranco, 1994). It was dicult to completely together clearly indicated that HSP90 has a higher immunodeplete HSP70 from the lysates because of its anity for unfolded, denatured LT than for folded LT. high abundance and we could not examine if HSP70 We attempted to determine whether ATP is required indirectly mediates the association between HSP90 and for LT-HSP90 reconstitution in the system described LT or not. Thus, we next tested the direct association above. The addition of 3.3 mM ATP to the reconstitu- of HSP90 with LT in an in vitro reconstitution system. tion system did not a€ect the co-precipitation of HSP90 with LT (data not shown). The addition of apyrase (up to 33 units/ml) to the mixture also did not Reconstitution of LT-HSP90 complexes in vitro from a€ect the HSP90-LT reconstitution (data not shown). HSP90 and unfolded LT, but not from native LT These results indicated that ATP is neither required To test whether the LT-HSP90 complex can be recon- for, nor inhibits the complex formation of LT with stituted from HSP90 and LT in vitro, puri®ed LT and HSP90 in vitro. HSP90 were mixed and the association of the two proteins was then examined by co-immunoprecipitation experi- Geldanamycin, an HSP90-inhibiting drug, decreased the ments. However, the results obtained under various intracellular amount of LT conditions were all negative (see Figure 5a, lane 5). We then examined complex formation using native Geldanamycin, a benzoquinoid ansamycin-derivative HSP90 and unfolded LT. Puri®ed LT was denatured drug (Uehara et al., 1986), binds speci®cally to HSP90 with 8 M guanidine-HCl and divided into two parts. and inhibits the chaperone-function of HSP90 (White- One part was then directly diluted 20-fold into a bu€er sell et al., 1994). To determine the requirement of the containing 0.25 mg/ml of HSP90 in the presence of a HSP90 function for maturation and stability of LT, large excess of bovine serum albumin. The other part COS7 cells were treated with 6 mM geldanamycin for 22 was ®rst incubated in the folding bu€er without HSP90 or 46 h and the amount of intracellular LT was then for 150 min at 308C, and then mixed with HSP90 determined by Western blotting. In response to under the same conditions as described above. geldanamycin treatment for 46 h, the amount of LT Subsequently, reconstitution of LT-HSP90 complexes was signi®cantly diminished (Figure 6b, lane 5), was examined by co-immunoprecipitation experiments. whereas the level of LT was not changed as a result As shown in Figure 5a lane 2, LT-immunoprecipitates of treatment for 22 h (Figure 6b, lane 3). Raf1, a target from the mixture of HSP90 and unfolded LT contained of HSP90 which is known to be short-lived, was not HSP90. The reconstituted complexes were observed as detected following treatment with geldanamycin for a very closely adjacent doublet (HSP90 as the upper 22 h (Figure 6d, lanes 3 and 5). This di€erential band and LT as the lower band) by silver staining sensitivity to geldanamycin between LT and Raf1 may (Figure 5c, lane 2). No reconstitution of HSP90-LT be attributable to a di€erence in intracellular life complexes was observed when either HSP90 or LT was duration and/or to a di€erence in intracellular pool omitted (Figure 5a, lanes 1 and 3). Interestingly, the sizes. To more directly illustrate the e€ect of amount of co-immunoprecipitated HSP90 with LT was geldanamycin on LT, we determined the amount of dramatically reduced when LT had been folded prior newly synthesized LT by immunoprecipitating LT from to mixing (compare lane 5 to lane 2 in Figure 5a). COS7 cells that had been labeled with 35S-methionine Equivalent amounts of LT were immunoprecipitated (Met)/cysteine(Cys). Following treatment with gelda-

Figure 5 Reconstitution of HSP90-LT complexes in vitro. Unfolded LT (marked as `U' in lanes 1 and 2) was diluted (1 : 20) in a bu€er containing HSP90, and the reconstitution of LT-HSP90 complexes was examined by co-immunoprecipitation analysis. When indicated, unfolded LT was renatured ®rst (marked as `F' in lanes 4 and 5), and then mixed with HSP90. As a control, HSP90 was omitted in lanes 1 and 4, and LT was omitted in lanes 3 and 6. (a) Association of HSP90 with immunoprecipitated LT was examined by anti-HSP90 Western blotting. (b) Amount of LT in the LT-immunocomplexes was revealed by anti-LT Western blotting. (c) Proteins in the immunocomplexes were analysed by SDS ± PAGE followed by silver staining. The positions of LT, HSP90, as well as heavy chain (HC) and light chain (LC) of antibodies are shown

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1481 proteins (Fanning, 1992; Pipas, 1992; Ludlow, 1993). In the present study, we newly identi®ed HSP90, a major cytosolic molecular chaperone, as a LT-associated cellular protein. No association between the 90-kDa protein and LT has previously been identi®ed, likely due to the fact that its molecular mass is almost the same as that of LT on SDS ± PAGE. We used immunological methods to detect the LT-HSP90 associations both in vitro and in vivo. The signi®cance of our ®ndings can be summarized by the following observations. (i) An antibody against LT co-precipitated HSP90. (ii) Anti- bodies against HSP90 co-precipitated LT. (iii) These co- immunoprecipitations were not observed in normal cells. (iv) LT-HSP90 complexes were reconstituted in vitro by mixing puri®ed HSP90 with puri®ed and unfolded LT without any additional protein components. (v) The anti-tumor drug geldanamycin, a known HSP90-speci®c inhibitor, decreased the amount of cellular LT. One of the major LT-associated cellular proteins is p53 (Lane and Crawford, 1979; Linzer and Levine, 1979). p53 is reported to be associated with HSP90 only when it is mutated (Blagosklonny et al., 1996; Whitesell et al., 1998). In addition, HSP70 is known to be associated with both p53 (Pinhasi-Kimhi et al., 1986) and LT (Sawai and Butel, 1989), and HSP90-containing complexes of Figure 6 E€ect of geldanamycin treatment on the intracellular functional proteins such as steroid hormone receptors concentration of LT. COS7 cells were treated with 6 mM are also associated with HSP70 (Csermely et al., 1998; geldanamycin for 22 (lane 3) or 46 (lane 5) h. As controls, cells Toft, 1998; Buchner, 1999; Mayer & Bukau, 1999). Thus, were treated with DMSO vehicle alone (lanes 1, 2, and 4). (a) The total protein pro®les of the extracts were revealed by SDS ± PAGE the possibility exists that either p53 or HSP70 links and CBB staining. Positions of HSP90, HSP70, and HSP27 are HSP90 with LT. However, we excluded this possibility indicated on the right. (b) The amount of LT in the extracts was based on the following; LT is abundant compared to p53 examined by anti-LT Western blotting. (c) The amount of newly in COS7 cells, and thus we could completely immuno- synthesized LT was shown by ¯uorography of immunoprecipitated deplete p53 without a signi®cant loss of LT. Even after LT from 35S-labeled cell lysates. (d) The amount of Raf1 in the extracts was shown by anti-Raf1 Western blotting p53 had previously been completely immunodepleted from cell lysates, we still detected the co-immunopreci- pitation of HSP90 with LT (Figure 4). In addition, our namycin for 22 h, whereas the total pool of LT was reconstitution experiments clearly demonstrated that no almost completely unchanged (Figure 6b, lane 3), the other protein is required for the HSP90-LT association, newly synthesized LT was signi®cantly reduced (Figure at least in vitro, when LT is in a non-native conformation 6c, lane 3) when compared to that of control cells (Figure 5). Thus, it is reasonable that the association (Figure 6c, lane 2). Hardly any newly synthesized LT between HSP90 and LT is direct, although the potential was detected after treatment for 46 h (Figure 6c, lane existence of ternary complexes such as LT-p53-HSP90 or 5). The total incorporation of radioactivity into cellular LT-HSP70-HSP90 in vivo can not be excluded. proteins was not reduced in response to geldanamycin Our initial attempts to reconstitute the LT-HSP90 treatment (data not shown). In light of these results, we complexes from puri®ed LT and HSP90 were not concluded that inhibition of cellular HSP90 function successful under any of various conditions. We were by geldanamycin lowered the stability and lifetime of able to reconstitute the complexes from native HSP90 newly synthesized LT, and ultimately led to a decrease and unfolded LT, suggesting that HSP90 recognizes a in total LT following treatment for 46 h. non-native LT. Our results also suggested that HSP90 While the overall protein pro®les of cells did not binds and stabilizes newly synthesized LT. Previously, change signi®cantly in response to geldanamycin we determined an HSP90-binding site within casein treatment, the amount of 90-, 70-, and 30-kDa proteins kinase II (Miyata and Yahara, 1995) and the site were found to have increased signi®cantly after contains a positively charged nuclear localization signal, treatment (Figure 6a, lanes 3 and 5). These proteins which is highly homologous to that of LT (Kalderon et were identi®ed as HSP90, HSP70, and HSP27, al., 1984). Thus, the nuclear localization signal sequence respectively, by Western blotting experiments (data of LT may be a candidate binding site for HSP90. not shown). The above result is entirely consistent with We estimated that 5 ± 10% of the total LT within the those of previous reports whereby inhibition of HSP90 cells were co-immunoprecipitated with HSP90 (data not activates the heat shock transcription factor, HSF1 shown). Although the majority of LT is present in nuclei, (Hedge et al., 1995; Zou et al., 1998). a small portion of LT is found in the cytoplasm. Conversely, the majority of HSP90 is cytosolic. Thus, we suppose that LT might be associated with HSP90 in Discussion the cytosol. This type of the correlation between the intracellular distribution of target proteins and associa- Large T antigens are known to transform cells via an tion with HSP90, is similar to the case for glucocorticoid association with many kinds of cell cycle-regulating receptors. Thus, as for glucocorticoid receptors, it might

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1482 be possible that HSP90 plays a role in cytoplasmic then treated with 5% skimmed milk in TBS containing anchoring of LT or in the nuclear import process of LT. 0.05% Tween 20 (TBS+Tw) at room temperature for The cooperative function of several molecular 60 min. The membranes were incubated for 3 h at 378Cor chaperones including HSP70 and HSP90 may be overnight at 48C with appropriate antibodies, washed three required for the correct folding of target proteins times with TBS+Tw, and incubated with alkaline-phospha- tase-conjugated secondary antibodies (Bio-Rad) for 60 min at (Welch, 1992; Schumacher et al., 1994; Frydman and 378C. The membranes were stained with an alkaline- HoÈ hfeld, 1997; Buchner, 1999). HSP70 is reported to phosphatase substrate kit (Vector). In the experiment shown a€ect the structure of LT, and to complement the in Figure 4b, horseradish peroxidase-conjugated secondary localization and functional defects of a LT mutant antibody (Amersham) was used and the membrane was (Jeoung et al., 1991). HSP90 is also reported to a€ect developed using chemiluminescence detection reagents (NEN the conformation of several substrate proteins (Miyata Life Science Products). and Yahara, 1992; Shaknovich et al., 1992; Wiech et al., 1992). Thus, HSP90 and HSP70 may both play Immunoprecipitation important roles in the folding and nuclear transloca- tion of LT in vivo. Cell extracts were precleared by mixing with nonimmune serum and protein G-sepharose beads. The precleared lysates were incubated with appropriate antibodies at 48C for 12 h. Protein G-sepharose was then added to the mixtures and gently rocked for 60 min at 48C. The beads were washed ®ve Materials and methods times with an IPW bu€er containing 0.1% NP40, and the resulting immunoprecipitates were analysed by SDS ± PAGE Cell lines and culture conditions and Western blotting. The cell lines used in this study were HeLa (human), CV1 (monkey), 3Y1 (rat), and SV40-transformed cell lines Proteins including WI26VA4 (human), COS1/COS7 (monkey), A640- 3Y1 (rat), and dl884-3Y1 (rat). All cell lines were cultured in HSP90 was puri®ed from mouse L5178Y and monkey COS7 Dulbecco's modi®ed MEM (Gibco-BRL) supplemented with cells as described previously (Koyasu et al., 1986; Miyata and 10% fetal calf serum. PAb419, a hybridoma cell line Yahara, 1992). LT was puri®ed according to a method producing an antibody against SV40 LT (Harlow et al., previously described (Dixon and Nathans, 1985; Simanis and 1981), was maintained in ASF104 medium (Ajinomoto) Lane, 1985) with some modi®cations. Brie¯y, COS7 cells supplemented with 10% fetal calf serum. WI26VA4, CV1, were extracted in an IPW bu€er containing 1% NP-40, and and COS1 were obtained from JCRB cell bank, A640-3Y1 the extract was applied onto a DEAE-sepharose (Pharmacia) and dl884-3Y1 (Kimura et al., 1975) from RIKEN cell bank column. Bound proteins were eluted by a linear gradient of with the permission of Dr G Kimura, and PAb419 from NaCl (100 ± 1000 mM) in Bu€er A (50 mM Tris-HCl, 2 mM RIKEN cell bank with the permission of Dr Y Murakami. EDTA, 10 mM sodium pyrophosphate, 1 mM dithiothreitol (DTT), and 10% glycerol, pH 7.4). Anti-LT monoclonal antibody (PAb419) was covalently conjugated to protein A- Buffers sepharose using ImmunoPure IgG Orientation Kit (Pierce). IPW bu€er: 20 mM Tris-HCl, 50 mM NaCl, 2 mM ethylene- DEAE-fractions containing LT were collected and passed diaminetetraacetic acid (EDTA), 10 mM sodium pyropho- through a protein A-sepharose column to remove proteins sphate, pH 7.4. Tris-bu€ered saline (TBS); 20 mM Tris-HCl, that non-speci®cally bind to protein A-sepharose, and then 500 mM NaCl, pH 7.4. Phosphate-bu€ered saline (PBS); applied onto the PAb419-protein A-sepharose column. Un- 10 mM sodium phosphate, 150 mM NaCl, pH 7.4. bound proteins were thoroughly washed-out by bu€er A containing 500 mM LiCl and the immuno-adsorbed LT was eluted from the column with a bu€er containing 20 m Antibodies M triethylamine, 1 mM DTT, 2 mM EDTA, 100 mM NaCl, and Rabbit anti-HSP90 antiserum has been described previously 10% glycerol, pH 11. Immuno-puri®ed LT was further (Koyasu et al., 1986). Anti-LT monoclonal antibody PAb419 puri®ed and concentrated by a MonoQ column using a was puri®ed from culture supernatants of the hybridoma NaCl gradient (100 ± 1000 mM) in bu€er A. The puri®ed LT PAb419 with a protein G-sepharose column (Pharmacia). was homogeneous as a 90-kDa band as judged by SDS ± Anti-p53 antibody (clone 421) and anti-Raf1 antibody PAGE and silver staining. The yield was estimated to be (R19120) were obtained from Oncogene Science and approximately 50 mg from 64 dishes (10 cm) of COS7 cells. Transduction Laboratories, respectively. Anti-HSP70 anti- body (clone N27F3-4) and anti-HSP27 antibody (clone G3.1) Immunodepletion of p53 from COS7 lysates were obtained from StressGen. The culture medium was changed to Met/Cys-free DMEM containing 10% fetal calf serum for 8 h, and the same Cell extracts medium containing 6 MBq/dish of Tran35S-label (NEN) was Cells cultured in 10 cm dishes were extracted in a 400 ml/dish then added. Cell lysates were prepared from labeled COS7 of an IPW bu€er containing 0.1% NP40 supplemented with cells and immunodepletion was performed by immunopreci- 10 mg/ml of aprotinin, 1 mM of phenylmethylsulfonyl¯uoride pitating twice with the indicated antibodies. The resulting (PMSF), and 2 mg/ml of leupeptin. The lysates were immuno-depleted supernatants were further used for im- centrifuged at 5000 g for 10 min and further at 100 000 g munoprecipitation with still other antibodies and the co- for 60 min at 28C. The resulting supernatants were used as immunoprecipitation of associated proteins was examined by soluble cell extracts for Western blotting and immunopreci- ¯uorography or by Western blotting. pitation. Reconstitution of LT-HSP90 complexes in vitro Western blotting LT was immunologically isolated from COS7 lysates with Proteins were resolved by SDS ± PAGE and electrotransferred PAb419-coupled Protein A-Sepharose beads as described to polyvinylidene di¯uoride (PVDF) membranes, which were above and the beads were washed extensively with an IPW

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1483 bu€er containing 1 mM DTT, 1% NP40, and 1000 mM NaCl geldanamycin (control cells were mock-treated with the same to remove any LT-associated proteins. LT was eluted from concentration of DMSO vehicle) for the indicated time at the beads with a solution containing 8 M guanidine-HCl and 378C in the culture medium. Cell extraction, immunopreci- 50 mM Tris-HCl, pH 7.4, and stored on ice as `unfolded-LT'. pitation, and Western blotting of treated samples were The unfolded-LT was diluted (1 : 20) in a reconstitution performed as described above. For the 35S-Met/Cys labeling, solution and incubated at 308C for 150 min in the presence or geldanamycin was added in the Met/Cys-free DMEM absence of HSP90 and ATP. The ®nal composition of the containing 10% fetal calf serum and then cells were pulse- reconstitution solution was 5 mM Tris-HCl, 400 mM guani- labeled by adding Tran35S-label during the last 4 h of dine-HCl, 7.5 mM creatine phosphate, 40 mg/ml creatine incubation. phosphokinase, 1.5 mM DTT, 60 mM potassium phosphate, 25 mM KCl, 3.5 mM MgCl2, 0.42 mg/ml bovine serum albumin, 15 mM hemin, 0.08 mM EDTA, 1% glycerol, and 4mM NaCl, pH 7.4, in the presence or absence of 0.25 mg/ ml of HSP90 and/or 3.3 mM ATP. When indicated, LT was ®rst folded without HSP90 in the above solution for 150 min at 308C prior to the addition of HSP90. The reconstitution Acknowledgments mixtures were then diluted eightfold with an IPW bu€er We thank Dr E Nishida for encouragement and support, containing 0.1% NP40 and the association of HSP90 with LT and Drs Y Minami and Y Kimura for their helpful was examined by co-immunoprecipitation experiments as comments. We thank Dr Y Murakami (RIKEN) and Dr described above. G Kimura (Kyushu University) for kindly providing us with cell lines. We thank Ms K Sakakibara and Ms K Kimura for their technical assistance. This work was Geldanamycin treatment of cells supportedinpartbyaCRESTresearchprojectandby Stock geldanamycin (10 mg/ml) was prepared in dimethyl grants-in-aid from the Ministry of Education, Science and sulfoxide (DMSO). COS7 cells were treated with 6 mM Culture of Japan.

References

Adamczewski JP, Gannon JV and Hunt T. (1993). J. Virol., Hu J and Seeger C. (1996). Proc. Natl. Acad. Sci. USA, 93, 67, 6551 ± 6557. 1060 ± 1064. Aligue R, Akhavan-Niak H and Russell P. (1994). EMBO J., Jakob U and Buchner J. (1994). Trends Biochem. Sci., 19, 13, 6099 ± 6106. 205 ± 211. Avantaggiati ML, Carbone M, Graessmann A, Nakatani Y, Jeoung DI, Chen S, Windsor J and Pollack RE. (1991). Genes Howard B and Levine AS. (1996). EMBO J., 15, 2236 ± Dev., 5, 2235 ± 2244. 2248. Joab I, Radanyi C, Renoir M, Buchou T, Catelli M-G, Blagosklonny MV, Toretsky J, Bohen S and Neckers L. Binart N, Mester J and Baulieu E-E. (1984). Nature, 308, (1996). Proc. Natl. Acad. Sci. USA, 93, 8379 ± 8383. 850 ± 853. Brugge JS, Erikson E and Erikson RL. (1981). Cell, 25, 363 ± Kalderon D, Richardson WD, Markham AF and Smith AE. 372. (1984). Nature, 311, 33 ± 38. Buchner J. (1999). Trends Biochem. Sci., 24, 136 ± 141. Kelley WL and Landry SJ. (1994). Trends Biochem. Sci., 19, Campbell KS, Mullane KP, Aksoy IA, Stubdal H, Zalvide J, 277 ± 278. Pipas JM, Silver PA, Roberts TM, Scha€hausen BS and Kimura G, Itagaki A and Summers J. (1975). Int. J. Cancer, DeCaprio JA. (1997). Genes Dev., 11, 1098 ± 1110. 15, 694 ± 706. Catelli MG, Binart N, Jung-Testas I, Renoir JM, Baulieu Koyasu S, Nishida E, Kadowaki T, Matsuzaki F, Iida K, EE, Feramisco JR and Welch WJ. (1985). EMBO J., 4, Harada F, Kasuga M, Sakai H and Yahara I. (1986). Proc. 3131 ± 3135. Natl. Acad. Sci. USA, 83, 8054 ± 8058. Csermely P, Schnaider T, SoÈ ti C, Proha szka Z and Nardai G. Lane DP and Crawford LV. (1979). Nature, 278, 261 ± 263. (1998). Pharmacol. Ther., 79, 129 ± 168. Lindquist S and Craig EA. (1988). Annu.Rev.Genet.,22, Cutforth T and Rubin GM. (1994). Cell, 77, 1027 ± 1036. 631 ± 677. DeCaprio JA, Ludlow JW, Figge J, Shew JY, Huang CM, Linzer DI and Levine AJ. (1979). Cell, 17, 43 ± 52. Lee WH, Marsilio E, Paucha E and Livingston DM. Ludlow JW. (1993). FASEB J., 7, 866 ± 871. (1988). Cell, 54, 275 ± 283. Mayer MP and Bukau B. (1999). Curr. Biol., 9, R322 ± R325. Dixon RA and Nathans D. (1985). J. Virol., 53, 1001 ± 1004. Miyata Y and Yahara I. (1992). J. Biol. Chem., 267, 7042 ± Dyson N, Buchkovich K, Whyte P and Harlow E. (1989). 7047. Cell, 58, 249 ± 255. Miyata Y and Yahara I. (1995). Biochemistry, 34, 8123 ± Fanning E. (1992). J. Virol., 66, 1289 ± 1293. 8129. Frydman J and HoÈ hfeld J. (1997). Trends Biochem. Sci., 22, Oppermann H, Levinson W and Bishop JM. (1981). Proc. 87 ± 92. Natl. Acad. Sci. USA, 78, 1067 ± 1071. Garcõ a-CardenÄ a G, Fan R, Shah V, Sorrentino R, Cirino G, Pinhasi-Kimhi O, Michalovitz D, Ben Zeev A and Oren M. Papapetropoulos A and Sessa WC. (1998). Nature, 392, (1986). Nature, 320, 182 ± 184. 821 ± 824. Pipas JM. (1992). J. Virol., 66, 3979 ± 3985. Harlow E, Crawford LV, Pim DC and Williamson NM. Rutherford SL and Zuker CS. (1994). Cell, 79, 1129 ± 1132. (1981). J. Virol., 39, 861 ± 869. Sanchez ER, Toft DO, Schlesinger MJ and Pratt WB. (1985). Hedge RS, Zuo J, Voellmy R and Welch WJ. (1995). J. Cell. J. Biol. Chem., 260, 12398 ± 12401. Physiol., 165, 186 ± 200. Sawai ET and Butel JS. (1989). J. Virol., 63, 3961 ± 3973. Hendrick JP and Hartl FU. (1993). Annu. Rev. Biochem., 62, Schumacher RJ, Hurst R, Sullivan WP, McMahon NJ, Toft 349 ± 384. DO and Matts RL. (1994). J. Biol. Chem., 269, 9493 ± Holt SE, Aisner DL, Baur J, Tesmer VM, Dy M, Ouellette 9499. M, Trager JB, Morin GB, Toft DO, Shay JW, Wright WE Shaknovich R, Shue G and Kohtz DS. (1992). Mol. Cell. and White MA. (1999). Genes Dev., 13, 817 ± 826. Biol., 12, 5059 ± 5068.

Oncogene Association between SV40 large T antigen and HSP90 Y Miyata and I Yahara 1484 Simanis V and Lane DP. (1985). Virology, 144, 88 ± 100. Welch WJ. (1992). Physiolog. Rev., 72, 1063 ± 1081. Smale ST and Tjian R. (1986). Mol. Cell. Biol., 6, 4077 ± WhitesellL,MimnaughEG,DeCostaB,MyersCEand 4087. Neckers LM. (1994). Proc. Natl. Acad. Sci. USA, 91, Srinivasan A, McClellan AJ, Vartikar J, Marks I, Cantalupo 8324 ± 8328. P, Li Y, Whyte P, Rundell K, Brodsky JL and Pipas JM. Whitesell L, Sutphin PD, Pulcini EJ, Martinez JD and Cook (1997). Mol. Cell. Biol., 17, 4761 ± 4773. PH. (1998). Mol. Cell. Biol., 18, 1517 ± 1524. Stancato LF, Chow Y-H, Hutchison KA, Perdew GH, Jove Wiech H, Buchner J, Zimmermann R and Jakob U. (1992). R and Pratt WB. (1993). J. Biol. Chem., 268, 21711 ± Nature, 358, 169 ± 170. 21716. Yahara I, Minami Y and Miyata Y. (1998). Ann. N.Y. Acad. Tjian R, Robbins A and Clark R. (1980). Cold Spring Sci., 851, 54 ± 60. Harbour Symp. Quant. Biol., 44, 103 ± 111. Yang J and DeFranco DB. (1994). Mol. Cell. Biol., 14, Toft DO. (1998). Trends Endocrin. Metabol., 9, 238 ± 243. 5088 ± 5098. Uehara Y, Hori M, Takeuchi T and Umezawa H. (1986). Zou J, Guo Y, Guettouche T, Smith DF and Voellmy R. Mol. Cell. Biol., 6, 2198 ± 2206. (1998). Cell, 94, 471 ± 480.

Oncogene