Cell Cycle-dependent Switch of Up- and Down-regulation of Human hsp70 Gene Expression by Interaction between c-Myc and CBF/NF-Y*

A CCAAT box-binding protein subunit, CBF-C/NF-YC, was cloned as a protein involved in the c-Myc complex formed on the G1-specific enhancer in the human hsp70gene. CBF-C/NF-YC directly bound to c-Myc in vitro andin vivo in cultured cells. The CBF/NF-Y·c-Myc complex required the HSP-MYC-B element as well as CCAAT in the hsp70G1-enhancer, while the purified CBF subunits recognized only CCAAT even in the presence of c-Myc. Both the HSP-MYC-B and CCAAT elements were also required for the enhancer activity. In transient transfection experiments, the CBF/NF-Y·c-Myc complex, as well as transcription due to the G1-enhancer, was increased by the introduction of c-Myc at low doses but decreased at high doses. The repression of both complex formation and transcription by c-Myc at high doses was abrogated by the introduction of CBF/NF-Y in a dose-dependent manner. Furthermore, the CBF/NF-Y·c-Myc complex bound to the G1-enhancer appeared in the early G1 phase of the cell cycle when c-Myc was not higly expressed and gradually disappeared after the c-Myc expression reached its maximum. The results indicate that the cell cycle-dependent expression of the hsp70 gene is regulated by the intracellular amount of c-Myc through the complex formation states between CBF/NF-Y and c-Myc.

The human heat shock 70 (hsp70) gene is induced by various stresses, including heat shock (for reviews, see Refs. [1][2][3]. Without stress, the hsp70 gene is expressed in a cell cycle-dependent manner, namely in the G 1 and the S phases (4 -7). In addition to the heat shock element, several transcriptional regulatory elements, including the sites for Sp1, AP2, ATF, and CTF, have been found to contribute to a basal level of hsp70 gene expression in the hsp70 gene promoter region. CTF is a CCAAT box-binding protein, and two CCAAT boxes are present in the hsp70 gene promoter, at about Ϫ150 and Ϫ90 from the transcription start site (8). Not only CTF (9) but also CBF of 114 kDa (10) have been cloned as binding proteins to these CCAAT sequences. The hsp70 gene expression is also induced by several oncogenes, including c-myc (11,12), c-myb, p53, T antigens of SV40 and polyomavirus (13), and adenovirus E1A (14 -17). Although the precise mechanisms of the regulation by these oncogenes have not been clarified, p53, Myb, and E1A have been reported to regulate the hsp70 gene expression by interacting with other proteins that directly bind to the respective transcriptional elements in the hsp70 gene promoter (18). Both E1A and p53 competitively bind to the CBF of 114 kDa to stimulate and repress the hsp70 promoter activity (19). Upregulation of the gene by c-myc has been reported in Drosophila and humans, and the region involved in the regulation was identified to be from Ϫ120 to Ϫ1,250 in the human hsp70 promoter (11,12). We have identified two sequences bound by c-Myc complexes in this region and termed them HSP-MYC-A (from Ϫ232 to Ϫ226) and HSP-MYC-B (from Ϫ157 to Ϫ151) (20). HSP-MYC-B is adjacent to an inverted CCAAT box and one nucleotide is overlapped. Both HSP-MYC-A and HSP-MYC-B functioned as an origin of DNA replication, and the plasmids carrying both or either of the sequences worked as an autonomous replicating plasmid (21). Of the two sequences, only HSP-MYC-B showed a transcriptional enhancer activity which contributed to the G 1 -specific expression of hsp70 in the cell cycle (7). This enhancer function of HSP-MYC-B was confirmed by another group (22). The c-Myc complexe was therefore thought to be involved in the cell cycle-dependent expression of the hsp70 gene. In this study, we have verified the association of c-Myc with CBF/NF-Y, CCAAT-binding proteins, and the involvement of the c-Myc⅐CBF/NF-Y complex formed on the HSP-MYC-B element in the regulation of hsp70 expression.

EXPERIMENTAL PROCEDURES
Plasmids-pACT-CBF-C was obtained from one-hybrid plasmids followed by two-hybrid screening targeting the HSP-MYC-B and C-Myc binding activities. For pCMV-CBF-C, pGEX-CBF-C, and pMBP-CBF-C, the EcoRI-XhoI fragment of pACT-CBF-C was inserted into the EcoRI-XhoI site of pcDNA3 (Invitrogen), pGEX-5X-1 (Amersham Pharmacia Biotech), and pMAL-6P-1, 1 respectively. For pACT-CBF-A and -B the NcoI-XhoI fragment of pCite-2a-CBF-A and -B, respectively (23), was inserted into the respective sites of pACT2. The pCMV-CBF-A, -B, pGEX-CBF-A, -B, and pMBP-CBF-A, -B of the BglII-XhoI fragment of pACT-CBF-A, -B was inserted into the BamHI-XhoI site of pcDNA3, pGEX-5X-1, and pMAL-6P-1, respectively. pCMV-GAL4-CBF-A, B, and C of the HindIII-BamHI fragment of pGBT9 (CLONTECH) and BglII-XhoI fragment of pACT-CBF-A, B, and C, respectively, were inserted into the HindIII-XhoI sites of pcDNA3. The c-Myc plasmids for the mammalian two-hybrid system, wild type, NЈ, ⌬177, ⌬M, and ⌬Zip, have been described previously (24), and ⌬H1 and ⌬LZ were constructed according to the published procedure (25). pGEX-CBF-C(EP) (pGEX-CBF-C) was digested with PstI and the resultant large fragment was self-ligated. pGEX-CBF-C(EN) (pGEX-CBF-C) was digested with NcoI and XhoI, and the resultant large fragment was treated with the Klenow fragment followed by self-ligation. pGEX-CBF-C(NX), the pGEX-CBF-C, was digested with NcoI and the resultant large fragment was self-ligated. pGEX-CBF-C(NP), pGEX-CBF-C(NX), was digested with PstI and the resultant large fragment was self-ligated. pGEX-CBF-C(PX), the CBF-C(PX) cDNA, was prepared by polymerase chain reaction using pACT-CBF-C as a template and the following oligonucleotides as primers: upper, 5Ј-GGAATTCCTGCAGTATATCCGCTTA-3Ј; lower, 5Ј-GTTGAAGTGAACTTGCGGGG-3Ј. The polymerase chain reaction contained 10 ng of pACT-CBF-C, 10 pmol each of the primers, and 2.5 units of Taq polymerase (Nippongene). The reactions were carried out for the first 5 min at 94°C, and then for 30 cycles of 1 min at 94°C followed by 2 min at 55°C, and finally for 3 min at 72°C. The amplified fragment was digested with EcoRI and XhoI, and cloned to the EcoRI-XhoI sites of pGEX-5X-1. pHISi-4xwtB, the complementary oligonucleotides (5Ј-AATTCTGGCCTCTGATTGGATG-3 and 5Ј-TC-GACCAATCAGAGGCCAGAGCT-3Ј) corresponding to HSP-MYC-B, were synthesized, annealed, and inserted into the SacI-XhoI sites of OVEC (26), and multimerization of the elements (2x-and 4xwtB) was carried out according Ref. 26. The tetramerized elements, 4xwtB, were inserted into the SacI-SalI sites of pUC19 (pUC-4xwtB). The HindIII site of pUC-4xwtB was changed to XbaI by linker insertion, and the EcoRI-XbaI fragment of this plasmid containing 4xwtB was inserted into the respective site of pHISi. pLacZi-4xwtB, the EcoRI-SalI fragment of pUC-4xwtB, was inserted into the respective site of pLacZi.
Cloning of cDNAs Encoding HSP-MYC B-binding Protein with Association of c-Myc-pHISi-4xwtB was digested with KpnI to be linearized, and integrated to the chromosome in Saccharomyces cerevisiae YM4271 by Ura selection. pLacZi-4xwtB was digested with ApaI and integrated to the chromosome in S. cerevisiae YM4271 by Ura selection. These two yeast strains were used for one-hybrid screening for His selection and ␤-galactosidase assay.
Cell Culture and Synchronization of the Cells-Mouse Balb 3T3 and hamster CHO 2 cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% calf serum. Human Raji cells were cultured in RPMI 1640 medium with 10% fetal calf serum. Balb 3T3 cells were also cultured under low serum conditions (0.2% calf serum) for 48 h to enter the G 0 phase of the cell cycle.
Antibody-A fusion protein of GST and CBF-C/NF-YC was expressed in Escherichia coli DL21(DE3)pLysS and purified as described previously (27). Rabbits were immunized by the purified GST-CBF-C/NF-YC. The anti-CBF-C/NF-YC antibody from the rabbit serum was prepared by affinity chromatography containing GST-CBF-C/NF-YC, absorbed by GST, and used as a polyclonal anti-CBF-C/NF-YC antibody. Polyclonal antibodies against NF-YA (pR␣YA/C) and NF-YB (pR␣YB) were kindly provided by R. Mantovani (28) or purchased (Rockland Inc.).
In Vitro Binding Assay-GST-CBF-C/NF-YC and GST were purified from 1 liter of E. coli BL21(DE3) culture transformed with pGEX-CBF-C/NF-YC and pGEX-5X-1, respectively, as described previously (27). One g of GST-CBF-C/NF-YC or GST was first applied to glutathione-Sepharose 4B (Amersham Pharmacia Biotech) in a buffer containing 50 mM Tris-HCl (pH 7.5), 40 mM NaCl, 1% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 0.5 g/ml bovine serum albumin, and about 50 g of proteins in the Raji cell extract was then applied to the column and washed with the buffer (20 mM Tris (pH 7.5), 25 mM NaCl, 1% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 50 g/ml bovine serum albumin). The proteins recovered from the resin were separated in a 7.5% polyacrylamide gel containing SDS, blotted onto a nitrocellulose filter, and reacted with a mouse anti-c-Myc antibody (C33, Santa Cruz).
Co-immunoprecipitation Assay-About 10 mg of proteins from Raji cells was first immunoprecipitated with a rabbit anti-CBF-A/NF-YB, -B/-A (Rockland), or C antibody, or nonspecific mouse IgG. The binding reactions for immunoprecipitation of c-Myc and CBF/NF-Y were carried out in a buffer containing 150 mM NaCl, 5 mM EDTA, 50 mM Tris (pH 7.5), 1 mg/ml bovine serum albumin, 150 g/ml phenylmethylsulfonyl fluoride, and 0.25% Nonidet P-40. After washing with the same buffer except for 0.05% Nonidet P-40 instead of 0.25%, the precipitates were separated on a 15 or 7.5% polyacrylamide gel containing SDS, blotted onto a nitrocellulose filter, and reacted with the rabbit anti-CBF-C/ NF-YC antibody or a mouse anti-c-Myc antibody (C33, Santa Cruz), respectively.
Luciferase Assay-Two g of a reporter plasmid and 1 g of CMV-␤-gal, a ␤-galactosidase expression vector, were transfected to cells approximately 60% confluent in a 6-cm dish by the standard calcium phosphate method (29). Two days after transfection, whole cell extracts were prepared by addition of the Triton X-100-containing solution from the Pica gene kit (Wako Pure Chemicals Co. Ltd., Kyoto, Japan) to the cells. About a one-fifth volume of the extract was used for ␤-galactosidase assay to normalize the transfection efficiency as described previously (7), and the luciferase activity due to the reporter plasmid was determined using the Pica gene kit and a luminometer, Luminocounter ATP300 (Advantec Toyo Co. Ltd., Tokyo). The same experiments were repeated 5-10 times.
Mammalian Two-hybrid Assay-CHO cells were transfected with 2 g of effector plasmids, expressing the wild type or mutants of CBF/ NF-Y and c-Myc fused to GALBD or VP16, in addition to 1 g of pCMV-␤-gal and 1 g of pH17-MX-Luc-Luc, and followed by luciferase assay as described previously (24). Gel-mobility Shift (Bandshift) Assay-Preparation of nuclear extracts and bandshift assays were carried out as described previously (20).

Cloning of cDNA Encoding a Protein That Recognizes Together with c-Myc the Sequence Containing HSP-MYC-B and
CCAAT-The c-Myc-containing complex formed on the HSP-MYC-B element was suggested to be responsible for the G 1specific enhancer activity of the element, but c-Myc did not directly bind to the element (20). We thus planned to first clone proteins that directly bind to HSP-MYC-B, and then to select one that associates with c-Myc. Since the wtB oligonucleotide, 5Ј-CTGGCCTCTGATTGG-3Ј, containing the HSP-MYC-B showed a strong enhancer activity (Ref. 12; Fig. 9), tetramerized wtB was used as a targeting sequence for the one-hybrid screening. After introducing pHISi-4xwtB to S. cerevisiae to integrate in yeast genome, the human brain cDNA library in pACT2 was applied to the transformant, and the yeast were cultured in His(Ϫ) media. Among the approximately 4 ϫ 10 6 clones tested, more than 100 His-positive clones were obtained and subjected to ␤-galactosidase assays. Plasmid DNAs were extracted from the 12 ␤-galactosidase-positive clones and used for re-transforming the S. cerevisiae SFY526 previously transformed with pGBT9-c-myc (30). The double transformants surviving in His(Ϫ) media were then tested for ␤-galactosidase activity to see the interaction with c-Myc fused to the GALBD. Finally, 12 clones were obtained and these were classified into two groups, tentatively named as HSM-1 and HSM-2. Through the BLAST search of their nucleic acid sequences, HSM-1 was identified as a human homologue of rat NF-YC/CBF-C (31), while HSM-2 was a novel gene that encodes a protein with slight homology to NF-YC/CBF-C. We further characterized HSM-1, i.e. the human CBF-C/NF-YC, in this study.
CBF/NF-Y is a CCAAT-binding protein with three subunits of CBF, CBF-A, CBF-B, and CBF-C, which correspond to NF-YB, NF-YA, and NF-YC, respectively. The rat, mouse, and human cDNAs for CBF-A, CBF-B, and CBF-C cDNAs have already been cloned (31)(32)(33)(34)(35)(36)(37). Among the three subunits of CBF/ NF-Y, CBF-B/NF-YA directly recognizes the CCAAT sequence, not by itself but in the complex of the three subunits (23,38,39). The cDNA for human CBF-C/NF-YC was cloned in this study as HSM-1 in the screening of HSP-MYC-B-binding proteins, probably because the wtB probe used in the one-hybrid screening contained an inverted CCAAT in addition to the HSP-MYC-B element, and yeast homologues of CBF-B/NF-YA and CBF-A/NF-YB (HAP-2 and HAP-4, respectively) may complement the respective human homologues to sustain the binding activity of the cloned CBF-C/NF-YC to the wtB probe as described previously (40). The HSM-1 cDNA we cloned contained 1,381 nucleotides, with several variations compared with those in the human CBF-C/NF-YC cloned by Bellorini et al. (41), and encoded exactly the same 335 amino acids as the human CBF-C/NF-YC. As in the case of CBF-A/NF-YB and CBF-B/NF-YA, only two amino acids were different between human and rat CBF-C. The expression of human CBF-C/ NF-YC in human tissues was examined by Northern blot analyses. The results revealed that CBF-C/NF-YC was expressed almost ubiquitously in all the human tissues tested (data not shown), as reported previously (41).
Interaction of CBF/NF-Y with c-Myc in Vitro-To verify the interaction between CBF-C/NF-YC and c-Myc, in vitro binding assays were carried out using CBF-C purified as fused to glutathione S-transferase (GST) and the human Raji cell extract containing c-Myc. Various deletion mutants of CBF-C as well as the wild type protein were expressed in E. coli as GST fusion proteins and purified. The fusion proteins and GST alone were, respectively, applied to a glutathione-Sepharose column, and the Raji cell extract was subsequently added to the same column. After extensive washing of the column, the proteins bound to GST-CBF-C, or GST alone, were blotted with a mouse anti-c-Myc monoclonal antibody (C-33, Santa Cruz) (Fig. 1A). The anti-c-Myc antibody used did not react with the purified GST-CBF-C (Fig. 1A, lane 2). c-Myc was detected only in the eluate with the fusion protein of wild type CBF-C (Fig. 1A, lane  3). Neither deletion mutants of CBF-C nor GST alone bound to c-Myc (Fig. 1A, lanes 4 -9). The results suggest that CBF-C/ NF-YC forms a complex with c-Myc, either directly or via other proteins in the Raji cell extract.
Similar binding assays were carried out using antibodies against each subunit of CBF/NF-Y. Since the amino acid sequences of both CBF-A/NF-YB and CBF-B/NF-YA are identical in rat and human except for 2 amino acids, we used rat cDNAs for CBF-A and CBF-B (36,37), instead of the corresponding human cDNAs. Three subunits of CBF, as well as Max as a positive control interacting with c-Myc, were expressed in E. coli as fusion proteins with the maltose-binding protein (MBP) and purified by the affinity column containing maltose. The full-length c-myc cDNA was ligated to the MBP cDNA in a modified MBP expression vector, in which the recognition site for PreScission protease (Amersham Pharmacia Biotech) was introduced in the junction between MBP and an insert cDNA to express. The fusion protein MBP-c-Myc was first expressed in E. coli and purified by the maltose resin. The recombinant full-length c-Myc was then obtained by releasing MBP with PreScission protease and subjected to binding reactions. The full-length c-Myc bound to the E-box sequence together with Max (characterization of the recombinant c-Myc in full-length will be described elsewhere). The MBP fusion protein with either CBF subunits or Max, or MBP alone, was mixed with c-Myc and applied to the agarose resin coupled with a mouse anti-c-Myc monoclonal antibody (C-33, Santa Cruz) or nonspe- cific mouse IgG. The proteins bound or not to either of the antibody-coupled resins were analyzed by Western blotting with an anti-MBP antibody (Fig. 1B). MBP-Max was recovered from the anti-c-Myc antibody-coupled resin (Fig. 1B, lane 1) but not from the resin with nonspecific IgG (Fig. 1B, lane 2). Neither of the resins bound free MBP (Fig. 1B, lanes 13 and 14). A little recovery of MBP-CBF-A from both resins indicated that CBF-A nonspecifically interacts with the resin (Fig. 1B, lanes 4  and 5). MBP-CBF-C was detected in the eluate from the resin with the anti-c-Myc antibody but not with IgG (Fig. 1B, lanes  10 and 11), while MBP-CBF-B was hardly recovered from either resin (Fig. 1B, lanes 7 and 8). The results indicate that CBF-C/NF-YC directly binds to c-Myc in a specific manner.
Interaction of CBF/NF-Y with c-Myc in Vivo-To examine the association of CBF/NF-Y with c-Myc in vivo, a mammalian two-hybrid assay was carried out. The wild type or various deletion mutants of c-Myc were fused to the VP16 transcriptional activation domain (VP16AD), and each subunit of CBF/ NF-Y was fused to the GALBD. The constructs were transfected to hamster CHO cells in various combinations together with pHE17-MX-Luc, which contains 6xGAL4, and HTLV-1 promoter linked to the luciferase gene and luciferase assays were carried out (Fig. 2). With the wild type c-Myc, CBF-A and CBF-C stimulated luciferase activity, while CBF-B did not. c-Myc thus interacted in vivo with CBF-C, but not with CBF-B, in transfected CHO cells as well as in vitro. CBF-A was also suggested to interact with c-Myc in vivo, less strongly than CBF-C. The affinity difference between CBF-A and CBF-C was more significant in the interaction with c-Myc deletion mutants, as described below. Among the various c-Myc mutants examined, ⌬177 and ⌬M deleting both myc boxes-1 and -2 or the myc box-2 followed by the internal region, respectively, interacted strongly with CBF-C and weakly with CBF-A, but did not interact with CBF-B. Deletions in the C-terminal region of c-Myc, not only a big deletion in mutant N but also small deletions in mutants ⌬HZ, ⌬H1, and ⌬Zip, resulted in the loss of CBF binding activity. These results implied that the Cterminal region of c-Myc spanning amino acids 368 -435 is required for the interaction with CBF-C/NF-YC, and additionally with CBF-A/NF-YB.
The in vivo association between c-Myc and CBF/NF-Y was also examined by co-immunoprecipitation assays. A rabbit polyclonal anti-CBF-C antibody was prepared by GST-CBF-C as an immunogen. The antibody specifically detected CBF-C but not CBF-A or CBF-B in human, mouse, and rat cells (data not shown). The human Raji cell extract was first precipitated with the antibody against CBF-A, CBF-B (Rockland Inc.), or   (Fig. 3). The 64-kDa c-Myc in the Raji extract was precipitated with the antibody against CBF-A, CBF-B, or CBF-C (Fig. 3, lanes 3-5), while the precipitate with nonspecific IgG did not contain the protein reacting with the anti-c-Myc antibody (Fig. 3, lane 2). The results indicated that c-Myc associates with all CBF subunits in vivo in human Raji cells. Since the three CBF subunits form a stable complex in vivo (23,38,39), c-Myc was thought to associate with the CBF⅐NF-Y complex in vivo.
DNA-Protein Complex Targeting HSP-MYC-B and CCAAT Sequences-We have previously shown that a protein complex containing c-Myc bound to the HSP-MYC-B sequence (20). To assess the involvement of CBF/NF-Y in the c-Myc complex on HSP-MYC-B, bandshift assays were carried out using specific antibodies. The Balb 3T3 nuclear extract was treated with an antibody against c-Myc (OM11-905, Genosys), CBF-A, -B (28), or -C, prior to binding reactions with a labeled wtB oligonucleotide, and separated in a polyacrylamide gel (Fig. 4). In the absence of antibodies, two specific DNA-protein complexes, complexes I and II, were detected (Fig. 4, A and B, lane 2). The anti-c-Myc antibody abrogated the formation of both complexes, while neither complex was affected by IgG (Fig. 4A,  lanes 3-6). All three antibodies against CBF subunits impaired the formation of complex I but not of complex II (Fig. 4B, lanes  5-10). CBF/NF-Y was thus suggested to be involved in c-Myc containing complex I on the wtB oligonucleotide, but not in complex II. A rabbit polyclonal anti-Max antibody (C-124, Santa Cruz) did not interfere with DNA-protein complex formation (data not shown). Since c-Myc by itself does not bind to the wtB probe, complex I may interact with the sequence via CBF/NF-Y. Complex II may contain an unidentified protein(s) other than CBF/NF-Y between c-Myc and the wtB sequence.
To determine the nucleotides recognized by complexes I and II, binding reactions were carried out in the presence of various FIG. 4. Nucleoprotein complex formation on the wtB sequence. Bandshift assays were carried out using the Balb 3T3 nuclear extract (NE) and the labeled wtB sequence as a probe. Prior to the addition of the probe, the reaction mixture was incubated for 60 min at 0°C in the presence or absence of the indicated amounts of antibodies against c-Myc (OM11-905) (A) or three CBF/NF-Y subunits (B) (pR␣YB, pR␣YA/C, or ␣-CBF-C), or of nonspecific IgG. The labeled probe was then added to the mixture, and the mixture was further incubated at room temperature for 30 min and analyzed in a 4% polyacrylamide gel. Similar bandshift assays on the wtB sequence were carried out using either the Balb 3T3 nuclear extract (C) or purified CBF subunits as MBP fusion proteins (D) in the presence of non-labeled oligonucleotides. Increasing amounts (ϫ 20, 50, and 100 of the probe) of non-labeled oligonucleotides corresponding to the wild type (WW) or various mutants (Wm, mW, mm, and ⌬C) of the wtB sequence (see middle panel) were added to the reaction mixtures as competitors. The positions of two specific DNA-protein complexes are indicated by arrows (I and II). F shows the position of free probes.
hsp70 Gene Expression Regulated by c-Myc and CBF/NF-Y mutants of the wtB probe (Fig. 4C). Mutations were introduced to wtB (WW) within CCAAT (Wm) or HSP-MYC-B (mW), or both (mm). Another variant, ⌬C, in which CCAAT was deleted from wtB, was also constructed. Excess amounts of nonlabeled wtB oligonucleotide (WW) abolished the specific nucleoprotein complexes (Fig. 4C, lanes 2-4). None of the wtB variants (mW, Wm, or mm), on the other hand, affected the formation of the complexes (Fig. 4C, lanes 5-16). The results suggested that not only the inverted CCAAT but also the adjacent nucleotides in the HSP-MYC-B element were involved in DNA-protein association of complexes I and II.
When the CBF/NF-Y proteins purified as MBP fusion proteins were subjected to binding reactions with the labeled wtB probe, a single nucleoprotein complex appeared, with a faint background due to degraded proteins (Fig. 4D, lane 3). The DNA-protein complex of the purified CBF/NF-Y protein on wtB was canceled by excess amounts of nonlabeled oligonucleotides carrying the intact CCAAT (WW and mW; Fig. 4D, lanes 4 -6 and 10 -12) but not by those with mutations or deletion in CCAAT (Wm, ⌬C and mm; Fig. 4D, lanes 7-9 and 13-18). Thus, the purified CBF proteins recognized the CCAAT motif in wtB, while complex I containing the CBF proteins and c-Myc were formed on both HSP-MYC-B and CCAAT. The results of methylation interference experiments also suggested that complex I was formed on almost the whole length of wtB sequence, while complex II and the complex of recombinant CBF proteins interacted with the probe within either the HSP-MYC-B or the CCAAT element, respectively (data not shown).
Although c-Myc was involved in complex I, the purified c-Myc (the MBP-released recombinant protein as described above for Fig. 1B) did not recognize the wtB probe by itself (Fig.  5A, lanes 2-4). The addition of the purified c-Myc did not affect the nucleoprotein complex formed between CCAAT in the wtB sequence and the purified CBF proteins: complex I observed in the nuclear extract was not reconstituted by the purified CBF proteins and the purified c-Myc (Fig. 5A, lanes 8 -10). Complex I, representing a complex in vivo, was thus thought to contain other proteins in addition to the CBF proteins and c-Myc.
We have shown above that c-Myc directly associates with CBF-C (Fig. 1B). A previous report revealed that CBF-A and CBF-C were first associated and that CBF-B possessing the DNA binding activity was recruited to the precomplexed CBF-A and CBF-C (23,42). CBF-A and c-Myc may independently interact with different sites of CBF-C, or alternatively CBF-A and c-Myc may competitively associate with CBF-C. Since the CBF complex on CCAAT was neither decreased nor FIG. 6. Effect of mutations in the wtB sequence on the transcriptional activity. Mutations were introduced into the wtB sequence (see Fig. 4, middle panel) in a reporter luciferase construct, pwtB-TATA-Luc, which carries the wtB sequence linked to the minimal hsp70 promoter (HS-TATA) (7). The mutated plasmids as well as pHS-TATA-Luc were transfected to mouse Balb 3T3 cells, and the luciferase activities were assayed 2 days after transfection. Relative luciferase activities to that due to pHS-TATA-Luc are shown.

FIG. 5. Nucleoprotein complex formation of purified CBF proteins and c-Myc on wtB.
Bandshift assays were carried out using the three CBF subunit proteins purified as fusion proteins to MBP, c-Myc purified as a fusion protein to MBP and released by PreScission protease, and the labeled wtB sequence as a probe. The positions of the CBF complex on wtB and the free probe are indicated as CBF and F, respectively. A, various amounts of the CBF proteins, or c-Myc together with or without CBF, were subjected to the binding reactions. The amounts of c-Myc added to the reactions were 10, 25, and 50 ng in lanes 1 and 8, 3 and 9, and 4 and 10, respectively. The amounts of CBF subunits were 10, 25, 50, and 25 ng each in lanes 5, 6, 7, and 8 -10, respectively. B, various amounts of c-Myc were preincubated with one of the CBF subunits and then incubated with the other two proteins, as indicated. Twenty-five ng of each of the CBF subunits were added to the reactions. The amounts of c-Myc added to the reactions were 5, 10, 25, and 50 ng in lanes 2-5, 7-10, and 12-15, respectively. shifted by the addition of c-Myc (Fig. 5A), it was thought that c-Myc was unable to interact with CBF-C in the preassociated CBF complex on CCAAT. c-Myc was therefore incubated with CBF-A, -B or -C prior to the addition of the other two subunits and the wtB probe (Fig. 5B). When CBF-B and -C were added to the preincubation of c-Myc and CBF-A, the CBF complex on CCAAT was formed as efficiently as in the absence of c-Myc (Fig. 5B, lanes 1-5). The affinity of CBF-C to CBF-A was therefore thought to be stronger than that to c-Myc. The addition of CBF-A and CBF-C to the preincubated CBF-B with c-Myc also yielded the CBF⅐CCAAT complex as in the absence of c-Myc (Fig. 5B, lanes 6 -10), although CBF-C preferred CBF-A to c-Myc. The preincubation of CBF-C with c-Myc, on the other hand, greatly inhibited the formation of the CBF complexes on CCAAT in the wtB probe (Fig. 5B, lanes 11-15). No bands due to nucleoprotein complexes other than the CBF⅐CCAAT complex alternatively appeared. The results suggested that the preceding interaction between CBF-C and c-Myc showed priority to CBF-A, which has a higher affinity to CBF-C. The disruption of the CBF complex on CCAAT did not yield another nucleoprotein complex on wtB as it was thought that an unidentified protein (or proteins) in the nuclear extract, other than c-Myc and the CBF proteins, is involved in complex I on wtB and that the protein (or proteins) mediates between the complex and wtB sequence. It was therefore thought that c-Myc interferes with the association of CBF-C with CBF-A when c-Myc precedes CBF-A in the interaction with CBF-C, but that it cannot intrude into the precomplexed CBF proteins.
Regulation of the Transcriptional Activity of wtB Containing HSP-MYC-B and CCAAT by c-Myc-We have previously reported that the HSP-MYC-B element was required for G 1specific enhancer activity of the wtB sequence (7). Since the wtB sequence contains CCAAT besides HSP-MYC-B, the effects of CCAAT on the transcriptional activity of wtB were examined. The wild type (WW) or wtB variant with mutations within either CCAAT (Wm) or HSP-MYC-B (mW), or both (mm), or deletion of CCAAT (⌬C) (see Fig. 4), was ligated to the TATA box of the human hsp70 gene linked to the luciferase gene, and transfected to mouse Balb 3T3 cells. The results of the luciferase activity assays of the cells are shown in Fig. 6. The wild type wtB (WW) in pwtB-TATA-Luc stimulated luciferase activity 22-fold of that due to pHS-TATA-Luc containing FIG. 7. Effect of additional c-myc expression on the transcriptional activity of the wtB sequence and the specific complex formation on the wtB sequence. A, various amounts of a c-myc expression vector, either in sense (pEF-c-myc) or antisense (pEF-anti-c-myc) orientation (51), were transfected to Balb 3T3 cells together with the reporter plasmid pwtB-TATA-Luc. Two days after transfection, the luciferase activities were assayed. Relative luciferase activities to that due to the reporter plasmid alone are shown. B, Balb 3T3 cells were transfected with various combinations of expression vectors for the CBF subunits, as indicated, in addition to 1 g of pEF-c-myc and the reporter pwtB-TATA-Luc. Two days after transfection, the luciferase activities were assayed. Relative luciferase activities of the reporter plasmid alone are shown. C and D, Balb 3T3 cells were transfected with various amounts of the c-myc expression vector pEF-c-myc as in A. Two days after transfection, the nuclear extracts were prepared from the cells and subjected to bandshift assays using a labeled wtB (C) or Sp1 element (D) as a probe. The positions of the specific CBF/NF-Y complexes I and II, the Sp1 complex, and the free probe are indicated as I, II, Sp1, and F, respectively. E, aliquots of the same nuclear extracts prepared in C and D were also analyzed by Western blotting using the anti-c-Myc antibody (C33) or an anti-CBF-C antibody.
the TATA box alone. All the mutants tested lost the enhancer activity. The luciferase activity of pWm-TATA-Luc, pmW-TATA-Luc, pmm-TATA-Luc, or p⌬C-TATA-Luc was comparable to that of pHS-TATA-Luc. The results suggested that not only the HSP-MYC-B element but also the CCAAT motif are required for the transcriptional enhancer activity. Complex I recognizing both HSP-MYC-B and CCAAT, but not the CBF complexes targeting only CCAAT, was hence implied to be responsible for the transcriptional activity. Complex II which did not contain CBF proteins was also suggested to be involved in the regulation of transcriptional activity, since the complex was formed on both HSP-MYC-B and CCAAT.
Since c-myc protein was shown to be involved in both complexes I and II, the effects of c-Myc on the transcriptional activity were examined. Various amounts of the expression vector for c-Myc either in sense (pEF-c-myc) or antisense orientation (pEF-anti-c-myc) were transfected to Balb 3T3 cells together with pwtB-TATA-Luc, and the luciferase activities were assayed (Fig. 7A). While the c-myc construct in antisense did not affect the luciferase activity at all, the sense construct yielded a biphasic fluctuation of enzyme activity (pEF-c-myc stimulated the luciferase activity in a dose-dependent manner to a peak of 1.8-fold at 10 ng, but higher doses repressed the activity). When 1 g of pEF-c-myc was co-transfected, luciferase activity was about 40% of that in the absence of pEF-c-myc. The 40% activity thus repressed by 1 g of pEF-c-myc was restored to the original level by the co-transfected expression vectors for both CBF-A and CBF-C or the three CBF subunits (Fig. 7B). CBF-C, which directly associates with c-Myc to form the core of complex I, did not release by itself the repression by pEF-c-myc. Neither CBF-A nor CBF-B affected the repressed activity. The results suggest that the ratio among the CBF/ NF-Y subunits and c-Myc is important for the transcriptional activity of the wtB sequence, probably via the formation of either the CBF complexes targeting CCAAT or the c-Myc containing complex I on both HSP-MYC-B and CCAAT.
To further examine the involvement of complex I in the transcriptional activity of wtB and the regulation of the activity by c-Myc, bandshift assays were carried out using the nuclear extract prepared from cells transfected with the expression vector for c-Myc as above. Various amounts of c-myc expression vector, or the vector alone, were transfected to Balb 3T3 cells, and the nuclear extracts were prepared from transfected cells. The extracts were incubated with a labeled wtB probe and analyzed (Fig. 7C). The transfection of the vector alone affected neither the amounts of c-Myc and CBF-C nor the electrophoretic patterns of complexes I and II (data not shown). In the cells transfected with the c-myc expression vector, c-Myc increased with doses of c-myc expression vector in transfected cells, while the amount of CBF-C was almost constant (Fig.  7E). The DNA-protein complex I on the wtB probe was increased in cells transfected with the c-myc expression vector at low doses (5 or 10 ng) (Fig. 7C), where the transcriptional activity of wtB was enhanced (Fig. 7A). Complex I was then decreased in the cells transfected with the plasmid at high doses (50 or 100 ng), where the transcriptional activity was repressed (Fig. 7, A and C). Complex II, on the other hand, was increased in a manner dependent on the dose of the c-myc expression vector (Fig. 7C). The formation of another DNAprotein complex, the Sp1 complex, was not affected by the transfection of the c-myc expression vector (Fig. 7D). These results indicate that the transcriptional enhancer activity of wtB is due to the formation of the specific complex I containing CBF proteins and c-Myc on the sequence and that c-Myc thus regulates the wtB enhancer activity via affecting different nucleoprotein complex formations on the wtB sequence: the balance of the amounts, as well as the binding affinity, of c-Myc, CBF proteins, and unidentified proteins involved in complexes I and II may control the transcriptional activity of wtB contributing to the hsp70 gene expression. . R indicates the samples from the cells at random culture. A, the extracts were Western blotted using the anti-c-Myc antibody (C-33) or the antibody against the subunits of CBF/NF-Y (pR␣YB, pR␣YA/C, or ␣-CBF-C). B, the extracts were subjected to bandshift assays using a labeled wtB sequence as a probe. The positions of the specific complex I, II, and free probes are indicated. C, 2 g of the reporter plasmid pwtB-TATA-Luc was transfected to Balb 3T3 cells, and the cells were synchronized as described previously (12). The luciferase activity was assayed at various times (0 -24 h) after serum addition. Relative luciferase activities to that of the cells at random culture (R) are shown.
Cell Cycle-dependent Expression of DNA Binding Activity of CBF⅐NF-Y Complex-Since the endogenous c-Myc expression changes considerably during the cell cycle, the nucleoprotein complex on the wtB sequence was thought to vary during the cell cycle. To examine this possibility, mouse Balb 3T3 cells were synchronized to the G 0 phase by serum starvation, and the nuclear extracts were prepared at various stages of the cell cycle. The extracts were subjected to bandshift assays with a labeled wtB probe and Western blot analyses (Fig. 8, A and B). c-Myc was accumulated during the S phase to a peak at 8 h after serum readdition, while the amounts of all the CBF/NF-Y subunits were constant during the cell cycle (Fig. 8A). The DNA-protein complex I containing both CBF and c-Myc was considerably altered during the cell cycle. The complex was weakly observed with the extract prepared from the G 0 -arrested cells, but was strongly induced 4 h after the serum addition and then gradually decreased but increased again after the 20 h (Fig. 8B). The c-Myc expression was very low at the peak of complex I formation at 4 h after serum addition, high from 8 to 16 h, and decreased after 20 h, when complex I re-increased (Fig. 8A). The results suggested that complex I formation on the wtB probe is increased when the c-Myc expression is low and that the complex formation is inhibited when the c-Myc expression is high during the cell cycle. The formation of complex II was rather constant up to 12 h and then slightly decreased (Fig. 8B). The cells transfected with the reporter plasmid, pwtB-TATA-Luc, were similarly synchronized and examined for luciferase activity. The luciferase expression showed a sharp peak at 4 h and then decreased with a minor peak at 20 h (Fig. 8C). The complex I formation thus reached a peak at 4 h after serum addition, i.e. in the early G 1 phase, when the introduced luciferase expression due to the wtB sequence showed a strong peak and the hsp70 expression also showed the first maximum (7). These results suggest that cell cycle-dependent hsp70 gene expression is regulated by, at least to some extent, the formation of nucleoprotein complex I containing CBF proteins and c-Myc on both HSP-MYC-B and CCAAT in the wtB sequence and that the complex formation is controlled by the intracellular amounts of c-Myc.
We described the transcriptional regulation of hsp70 by cmyc through the formation of specific DNA-protein complexes. c-Myc associated with CBF/NF-Y to form a specific DNA-protein complex (complex I) on the wtB sequence in the hsp70 gene promoter containing the HSP-MYC-B element and the CCAAT box. c-Myc was also involved in another nucleoprotein complex (complex II) with unidentified proteins other than CBF/NF-Y on the same sequence. The in vivo ratio between complex I and complex II on the wtB sequence varied according to the cell cycle-dependent expression of c-myc expression. Complex I increased when the c-myc expression increased in the early G 1 phase and then decreased when c-Myc was accumulated to reach a peak at the transition of G 1 to S phase. Complex II, on the other hand, increased form the G 1 -S transition. The transcription due to the wtB sequence, as well as the transcription of the hsp70 gene, increased in the G 1 phase to maximal expression at the same time of the maximal formation of complex I and then decreased with the decrease in the formation of complex I and the increase in the formation of complex II. c-Myc was thus thought to switch the up-and down-regulation of the hsp70 gene expression to show a peak in the G 1 phase via transition between complex I and complex II on the wtB sequence.
c-Myc interacted with CBF-C directly and specifically in vitro. Unidentified factor(s) other than c-Myc and CBF were implied to be necessary for the formation of both complexes I and II, commonly or differently. Factor X, supposed to be involved in at least complex I may directly bind to both c-Myc and the HSP-MYC-B element in the wtB sequence to yield complex I (Fig. 9). Factor X may also interact with CBF/NF-Y in complex I. Provided that factor X is a common factor to intermediate c-Myc and the HSP-MYC-B element, complex I containing CBF/NF-Y positively contributes to the transcriptional activity of the wtB sequence, while complex II without CBF/NF-Y is a transcriptionally inactive form. Alternatively, another factor, FIG. 9. A molecular model for transcriptional regulation via the wtB sequence during the cell cycle. The state of specific DNA-protein complexes formed on the wtB sequence is suggested to determine the G 1 -specific transcription due to the sequence. A molecular model is shown. In the G 1 phase of the cell cycle, when the c-myc expression is moderate, complex I containing an unidentified factor, factor X, in addition to c-Myc and subunits A, B, and C of CBF/NF-Y is formed, and thereby the transcription is active. In complex I, factor X and CBF-B/NF-YA directly interact with HSP-MYC-B and CCAAT, respectively. In the S phase, when the c-myc expression is increased, the oversupplied c-Myc absorbs CBF-C/NF-YC and thus disrupts the CBF complex on CCAAT. On the HSP-MYC-B element, complex II containing c-Myc is bound via either factor X or another unidentified factor, factor Y. Complex II which does not contain the CBF/NF-Y proteins is transcriptionally inactive. factor Y, may be a bridge between c-Myc and the HSP-MYC-B element in complex II. The release of CBF/NF-Y from complex I to yield complex II, or the replacement of complex I containing factor X by complex II containing factor Y, was observed in cells transfected with a c-myc expression vector as well as in cells at the transition of the G 1 to S phase of the cell cycle, where the expression level of c-Myc, but not of CBF, changed. During the S phase, the excess amount of c-Myc may upset the proper ratio between c-Myc and CBF/NF-Y to form complex I on the wtB sequence.
CBF/NF-Y is a well characterized transcription factor and expressed ubiquitously in a variety of mammalian cells. More than 170 genes have been found to contain the CBF/NF-Ybinding CCAAT sequence, and some of them were identified as target genes for CBF/NF-Y (43,44). The human hsp70 gene contains two CCAAT sequences in the promoter region, at about Ϫ150 and Ϫ90 from the transcription start site (4). Both the CCAAAT sequences were shown to be bound by NF-1/CTF in vitro (14), and furthermore, the CCAAT sequence around Ϫ90 was essential for the basal expression of the hsp70 gene (21). Another CBF of 114 kDa, different from CBF/NF-Y, was also reported to recognize the CCAAT sequence around Ϫ90 (13), and E1A and p53 competitively bind to the CBF of 114 kDa to stimulate or repress the hsp70 promoter activity, respectively (22,23). In this report, we have shown that another CCAAT at Ϫ150 was recognized by CBF/NF-Y and that c-Myc makes complexes with CBF/NF-Y on the sequence to yield two opposite functions, i.e. activation and repression, on the hsp70 promoter. The up-and down-regulation by c-Myc via the sequence is thought to determine the cell cycle-dependent expression, namely, the maximal expression at the G 1 phase, of the hsp70 gene. The two CCAAT sequences at Ϫ90 and Ϫ150 in the hsp70 promoter are thus thought to be involved in the basal and cell cycle-dependent expression of the gene, respectively.
The involvement of the CCAAT sequence in the cell cycle-dependent expression was also reported in the R2 gene of mouse ribonucleotide reductase (45) and in the cdc25C gene in which CBF/NF-Y interacted with Sp1 (46). Furthermore, CBF/NF-Y has recently been reported to interact with various proteins, including a TATA box-binding protein, TBP (37), histone acetyltransferases GCN5 and P/CAF (47,48), the high mobility group protein HMG-I(Y) (49), and the human T-cell lymphotropic virus type I Tax1 protein (50). The ubiquitous transcription factor CBF/NF-Y may differently regulate a variety of genes in partnership with various proteins.