Transactivation of E2F-Regulated Genes by Polyomavirus Large T Antigen: Evidence for a Two-Step Mechanism

Abstract

Polyomavirus large T antigen transactivates a variety of genes whose products are involved in S phase induction. These genes are regulated by the E2F family of transcription factors, which are under the control of the pocket protein retinoblastoma protein and its relatives p130 and p107. The viral protein causes a dissociation of E2F-pocket protein complexes that results in transactivation of the genes. This reaction requires the N-terminal binding site for pocket proteins and the J domain that binds chaperones. We found earlier that a mutation of the zinc finger located within the C-terminal domain, a region assumed to function mainly in the replication of viral DNA, also interferes with transactivation. Here we show that binding of the histone acetyltransferase coactivator complex CBP/p300-PCAF to the C terminus correlates with the ability of large T antigen to transactivate genes. This interaction results in promoter-specific acetylation of histones. Inactive mutant proteins with changes within the C-terminal domain were nevertheless able to dissociate the E2F pocket protein complexes, indicating that this dissociation is a necessary but insufficient step in the T antigen-induced transactivation of genes. It has to be accompanied by a second step involving the T antigen-mediated recruitment of a histone acetyltransferase complex.

DNA tumor viruses depend on the replication machinery of the host cell for the synthesis of their DNA. As these enzymes are present in effective amounts only during S phase, the viruses have evolved mechanisms to meet this requirement. They encode proteins that interfere with the growth regulation of infected cells, allowing them to drive cells from the quiescent, growth-arrested state into S phase. Main players in this reaction are the E1A protein of adenoviruses, the large T (LT) antigens of simian virus 40 (SV40) and polyomavirus (Py), and the E7 protein of human papillomaviruses (21). All of these proteins have binding sites for the pocket protein retinoblastoma protein (pRb) and its relatives p130 and p107. In their underphosphorylated form, pocket proteins are negative regulators of the E2F transcription factor family, which plays a decisive role in the regulation of the expression of G₁ and S phase-specific genes (17). Under physiologic conditions of growth induction, the pocket proteins are phosphorylated by cyclin D- and E-specific kinases. They then dissociate from E2F, thus allowing gene expression. The viral proteins circumvent this signal transduction-dependent activation pathway by binding to the underphosphorylated form of the pocket proteins, which results in dissociation of the E2F-pocket protein complexes and consequently in the transactivation of S phase-specific genes even in cells that are growth arrested. In addition to the pocket protein binding site, the N-terminal region of the T antigens carries a motif called the J domain that interacts with dnaK-type chaperones such as HSC70 (5, 18). Mutations in either one of these sites abolish the transactivation of S phase-specific genes by LT antigen. In our previous work on Py LT antigen-mediated transactivation of the genes coding for thymidine kinase (25) and cyclin A (31), we observed that, in addition to the N terminus, regions within the C-terminal part of the LT protein play an important role (31). A mutation of the zinc finger present within this domain (C452S) interferes with transactivation. The C terminus of the LT antigens of Py and SV40 was originally thought to be mainly responsible for the replication function of the proteins. It contains a region involved in specific binding to the origin of replication of viral DNA, and it harbors the ATP binding and helicase activities that are essential for viral DNA replication (12). In the case of SV40, this region can also bind p53, a capacity that is absent in Py LT antigen (26). Moreover, both proteins were found to interact through the C-terminal domain with the coactivator proteins CREB-binding proteins (CPB) and p300 (11, 23). A similar capacity was previously found within the N-terminal part of the E1A protein, where mutations of this region were reported to affect functions of the viral protein (13). The fact that LT proteins interact with CPB/p300 suggests that these coactivators, which exhibit histone acetyltransferase (HAT) activity, may play an important role in the function of the viral proteins. This assumption is supported by the observation that a mutation within the C-terminal domain of the Py LT antigen, which eliminates p300 binding, interferes with the ability of the protein to transform cells (8).

Here we describe our attempts to elucidate the role of the C-terminal region of the Py LT protein in the transactivation of G₁ and S phase-specific genes in quiescent mouse fibroblasts. We show that this region operates in conjunction with the N terminus in the activation of E2F-regulated promoters and that the dissociation of E2F-pocket protein complexes is necessary but not sufficient. Rather, a second step is required that involves the binding to LT antigen of a HAT complex consisting of CPB/p300 and PCAF and that results in promoter-specific acetylation of histones.

MATERIALS AND METHODS

Plasmids.

Mutations in the C-terminal domain, 413AA, Y442S, Q491E, L545S, E607A, and P670S, were introduced into the LT antigen with the help of the Stratagene mutagenesis kit in accordance with the supplier's recommendations. All mutations were verified by sequencing. The expression plasmids for transient transfections carried the cDNA for the respective T antigen under the control of the CMV promoter. The murine cyclin A promoter-luciferase construct was a gift from J. M. Blanchard, Montpellier, France; expression vectors CMV-HA-p300_wt and CMV-HA p300_d33 were a gift from R. Eckner, Zürich, Switzerland. The expression vector pCI-Flag-PCAF was a gift from R. L. Schiltz, Bethesda, Md.

N-terminal Myc tagging of mutant LT proteins was achieved by replacing the hemagglutinin (HA) tag in the HA-pCI_neo vector with a c-myc tag sequence.

Cell culture and stable transfections.

REF52 rat embryo fibroblasts and Swiss 3T3 cells conditionally expressing the Py LT antigen under the control of the mouse mammary tumor virus (MMTV) promoter (24) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), penicillin (60 μg/ml), and streptomycin (100 μg/ml) in a 7.5% CO₂ atmosphere. For growth arrest, asynchronously growing cells were seeded at 5 × 10⁵ cells per 100-mm-diameter petri dish; the next day, the serum concentration was reduced to 0.2% for 72 h. The cells were then either reinduced by addition of fresh medium containing 20% FCS or treated with dexamethasone at a concentration of 10⁻⁶ mol/liter. Cell lines expressing the wild-type Py LT antigen (LTwt) and the E145D, P43A, and C452S mutant forms were described previously (31). Cell lines conditionally expressing LT antigens with other point mutations in the C-terminal region (Y442S and P670S) were produced by Polybrene-assisted transfections as described previously (24, 31).

Transient transfection and luciferase assay.

Asynchronously growing REF52 cells were seeded at 10⁵ per six-well plate. The next day, the cells were transfected by the polyethylenimine (PEI) method (4). Briefly, 2 μg of total DNA was dissolved in 125 μl of HBS (140 mM NaCl, 25 mM HEPES, 0.75 mM Na₂HPO₄ [pH 7.1]). Three microliters of PEI diluted in HBS was then added dropwise to the DNA, and the mixture was incubated at room temperature for 20 min and added to the cells. After 4 h, the medium was changed to DMEM containing 0.2% FCS. At 48 h posttransfection, the cells were lysed in luciferase lysis buffer (100 mM K-phosphate [pH 7.8], 0.2% Triton X-100). Luciferase activity and β-galactosidase activity (as a control for transfection efficiency) were assayed in parallel by using the Dual Light Chemoluminescent Reporter Gene Assay System (Tropix, Bedford, Mass.). An aliquot of each extract was analyzed by immunoblotting for the expression levels of cotransfected proteins.

Immunoprecipitation and immunoblotting.

REF52 cells (6.5 × 10⁵/10-cm-diameter dish) were transiently transfected by the PEI method with expression vectors for wild-type or mutant LT proteins either alone (8.0 μg) or as HA-tagged LT antigen constructs (4.0 μg) in combination with Flag-PCAF (4.0 μg). After 28 h of incubation in DMEM containing 10% FCS, the cells were extracted with 500 μl of IP lysis buffer (10% glycerol, 20 mM Tris-Cl [pH 8.0], 135 mM NaCl, 1% NP-40, 1 mM phenylmethylsulfonyl fluoride, complete protease inhibitor cocktail [Roche Diagnostics]) for 20 min at 4°C, and lysates were cleared by centrifugation at 14,000 × g for 15 min. The supernatants were incubated with 10 μg of rabbit anti-CBP polyclonal antibody (C1; Santa Cruz), rabbit anti-E1A polyclonal antibody (13S5; Santa Cruz), or mouse anti-HA tag monoclonal antibody (12CA5; Santa Cruz) for 2 h at 4°C. Immunocomplexes were collected by adding 20 μl of protein A-Sepharose beads and further incubation for 3 h at 4°C, followed by centrifugation. The beads were then washed two times with IP lysis buffer and three times with TBS (25 mM Tris [pH 7.4], 137 mM NaCl, 3 mM KCl), resuspended in 15 μl of sodium dodecyl sulfate (SDS) sample buffer, and boiled for 5 min. Samples were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and detected by immunoblotting with the corresponding antibodies, followed by ECL (Perkin-Elmer). LT antigen was detected by a rabbit polyclonal antibody directed against the LT-ST common N-terminal region of Py T antigens (a gift from E. Ogris); for detection of HA-tagged proteins, mouse monoclonal antibody 16B12 (Santa Cruz) was used; and for Flag-PCAF detection, rabbit polyclonal anti-FLAG antibody Ab-1 (Sigma) was used.

For LT-LT antigen interactions, coimmunoprecipitation assays of mutated LT antigens were performed with HA- and Myc-tagged versions, respectively. Transient transfection of REF52 cells and cell lysis were done as described above. For immunoprecipitation, REF52 cell extract was incubated with 1 μl of mouse anti-HA tag monoclonal antibody 12CA5 (Santa Cruz) or 5 μl of mouse anti-c-myc tag monoclonal antibody (kind gift from E. Ogris) for 2 h at 4°C. Immunocomplexes were collected by adding 30 μl of protein A-Sepharose beads for the anti-HA antibody or 30 μl of protein G-Sepharose beads for the anti-Myc antibody. After further incubation for 3 h at 4°C, beads were spun down and washed two times with IP lysis buffer and two times with TBS buffer. Beads were then resuspended in 30 μl of SDS sample buffer and boiled for 5 min at 95°C. SDS-PAGE and Western blot analysis were performed as described above. For detection of Myc- and HA-tagged LT proteins, the same antibodies as for immunoprecipitation assays were used.

RNA extraction and Northern blotting.

Total RNA was isolated with TRIzol reagent (Life Technologies, GIBCO BRL) as recommended by the supplier, separated by morpholinepropanesulfonic acid (MOPS)-agarose gel electrophoresis, and blotted onto a nylon membrane (GeneScreen; NEN). Hybridization was performed with a ³²P-labeled cDNA fragment of murine thymidine kinase and visualized by autoradiography.

ChIP experiments.

The cells used in chromatin immunoprecipitation (ChIP) experiments were 3T3 cells as a control and 3T3 cells carrying the information for the LT antigen or mutant forms thereof under the control of the hormone-inducible MMTV promoter (24). ChIP assays were performed as described previously (32). Immunoprecipitation of chromatin fragments was done with anti-acetyl-histone H3 antibody (catalog no. 06-599; Upstate Biotechnology), and complexes were purified by incubation with 30 μl of protein A-Sepharose beads and subsequent washing as described previously (32). Chromatin-antibody complexes were eluted from the beads by addition of freshly prepared elution buffer (2% SDS, 0.1 M NaHCO₃, 10 mM dithiothreitol), and DNA was released from the complex by reversing the cross-linkage with 4 M NaCl and incubation for at least 6 h at 65°C. After proteinase K digestion, the DNA was extracted with phenol-chloroform, precipitated with ethanol, and dissolved in water.

PCR analysis of immunoprecipitated DNA.

PCR was performed with the Biometra D3 thermocycler with PCR Master Mix (Promega) and primers designed for the mouse thymidine kinase promoter region (tk1, 5′-AGACCCCGCACCTGAATCTG-3′; tk2, 5′-TTCACGTAGCTGAGAGGTGG-3′). The linear range of the PCR was determined empirically with different amounts of genomic DNA or different numbers of cycles. PCR products were resolved on 2% agarose-40mM Tris-acetate-1mM EDTA (pH 8.0) gels.

EMSA.

Cells were harvested and lysed in buffer containing 20 mM Tris-Cl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 0.5% NP-40, 2 mM phenylmethylsulfonyl fluoride, and complete protease inhibitor cocktail. Ten microliters of whole-cell protein extract was used for electrophoretic mobility shift assays (EMSA) with oligonucleotides representing the E2F site of the murine thymidine kinase promoter, which was performed as described earlier (31). The antibodies used were anti-p130 (C-20; Santa Cruz) and anti-E2F4 (RK-13; Santa Cruz).

RESULTS

Construction and activity of mutant LT proteins with changes in the C-terminal domain.

We found previously that a mutation within the zinc finger (C452S) of the Py LT antigen resulted in a defect in the transactivation of cyclin A (31). The zinc finger region of the LT antigen is known to play a role in viral DNA replication. Mutation of this part of the LT antigen results in an inability of the protein to form double hexamers at the origin of replication on viral DNA and to function as a replicative helicase (28). In this work, we have created a number of further mutations in the C-terminal region of the protein, including one (P670S) that was previously shown to interfere with the binding of CBP/p300 (8) and another one (E607A) whose homologue in the SV40 LT antigen interferes with the helicase activity of the protein (38). In the 413AA mutant protein, two glutamic acid residues at the border of the DNA binding to the helicase domain of the Py LT antigen are replaced with two alanine residues. The other amino acid changes (Y442S, Q491E, and L545S) were selected because of their high conservation among Pys (Fig. 1 contains a schematic representation of the functional domains of the LT antigen and the mutant forms used in this work). The effects of all of the C-terminal mutations on the transactivation of the cyclin A and thymidine kinase promoter were tested by transient cotransfection of promoter-luciferase constructs and the LT antigen under the control of the cytomegalovirus (CMV) promoter. The results are summarized in Fig. 2. Besides the C452S mutation within the zinc finger domain, several others were found to produce defective transactivation, whereby the extent of the insufficiency was comparable for both of the E2F-regulated promoters examined. It is noteworthy that the mutant form with altered helicase activity (E607A) and the 413AA mutant form were completely functional in transactivation. While the C452S, Y442S, and P670S mutant forms had greatly reduced transactivating ability, the Q491E mutant form and in particular the L545S mutant form exhibited residual activity.

FIG. 1.

Schematic representation of domains of the LT antigen and the locations of the mutations used in this work.

FIG. 2.

Transactivation of the cyclin A (CycA) and thymidine kinase (TK) promoters by the wild-type LT protein (LTwt) and mutant forms thereof. Transient cotransfection was carried out as described in Materials and Methods. All experiments were done at least five times. For the complementation assay, two mutant forms of LT antigen were cotransfected together with the cyclin A promoter-luciferase construct. The activity of mutant forms transfected alone with the reporter construct is shown for comparison. The locations of the mutations within the LT antigen are shown in the schematic representation in Fig. 1.

Mutant proteins with changes in the pocket protein binding site and in the C-terminal domain can complement each other.

In order to find out whether the inactive mutant proteins with changes in the C terminus can be complemented by LT proteins mutated in the binding site for pocket proteins (E146D) or within the J domain (P43A), a complementation analysis was carried out in which the two mutant forms were cotransfected with the cyclin A promoter-luciferase construct (Fig. 2). The results of this experiment show that cotransfection of mutant forms with changes in the pocket protein binding site with either one of the inactive mutant forms with changes in the C-terminal region resulted in transactivation. In contrast, the mutant form with the change within the J domain was unable to complement any of the other mutant forms. This indicates that the J domain is required in cis not only for its activity involving the interaction with pocket proteins (33) but also for the function involving the C terminus. To the contrary, the activities linked to the binding of pocket proteins and that of the C terminus can be localized to two different molecules of the LT antigen. The two promoters used in this study differ in their transcription factor requirements except for the common presence of a binding site for E2F, which is occupied by E2F4/p130 in serum-starved cells (34). Given the similar effects of the mutant LT proteins on both promoters, it is likely that pocket protein p130 is the target for the LT antigen in both cases. The viral protein may bind to p130 as a dimer consisting of one mutant version with an intact pocket protein binding site but carrying a mutation within the C-terminal domain and a second mutant form with a defect in the pocket protein binding site but an intact C terminus. One of these, the C452S mutant protein, is known to be defective in producing oligomers (28); it was therefore of interest to determine whether this mutant protein was still capable of forming a dimer with a partner that is wild type in its C-terminal domain. This was tested in coimmunoprecipitation experiments with mutant forms of LT antigen to which different tags were attached. The results of this experiment are shown in Fig. 3. Clearly, while the mutant protein with an altered zinc finger domain by itself cannot form homodimers, it can dimerize with mutant proteins with changes within the N-terminal domain but having a wild-type C terminus. It is notable that both the mutant with a changed LXCXE region and the one with a changed J domain can form heterodimers with the zinc finger mutant form, although only the former combination is able to transactivate. Therefore the failure of mutant proteins with changes in the J domain to complement C-terminal mutant proteins is not due to a lack of dimerization.

FIG. 3.

Coimmunoprecipitation (Co-IP) of the E146D, P43A, and C452S mutant forms of LT antigen. Immunoprecipitations (IP) of HA- or Myc-tagged mutant forms of LT antigen and detection of LT antigen were carried out as described in Materials and Methods. Proteins from cell extracts (200 μg) were precipitated with anti-HA antibody, and the precipitate was divided 1:1 for SDS-PAGE to detect HA-tagged LT antigen and Myc-tagged LT antigen separately. Twenty micrograms of cell extract (Input) was loaded as an expression control for each transfected LT antigen. Precipitations with anti-Myc (α-Myc) antibody gave identical results. α-HA, anti-HA.

Mutant proteins with changes in the C-terminal domain are defective in the interaction with CBP/p300 and PCAF.

Considering that the P670S mutant form was reported to be defective in binding the coactivator CBP/p300, we examined all of our mutant proteins for the ability to coimmunoprecipitate with CBP/p300 (Fig. 4A). Clearly, the defect in transactivation correlated with a failure to bind CBP/p300. No interaction with CBP/p300 was observed with the C452S, Y442S, and P670S mutant forms, which were fully defective in transactivation. The Q491E and L545S mutant forms, on the other hand, appear to contradict this connection. Both of these mutant forms exhibit diminished activity, but they differ in the capacity to bind p300. While in the experiment shown in Fig. 4A the Q491E mutant protein shows hardly visible binding of CBP/p300, in several similar experiments this mutant protein disclosed variable but always weak binding to the coactivator. The reason for this variability and the low activity of this mutant protein might be explained by its location. Q491 is localized in an area of the LT antigen corresponding to helix α5 of the helicase of SV40. This helix was found to play an important role in the oligomerization of the protein (19) and, in agreement with the data on the SV40 LT antigen, Q491E was found to be highly defective in oligomerization, including dimerization (M. Smutny, unpublished data). L545S, in contrast, was always found to bind CBP/p300. Why this mutant protein is lower in its transactivation potential than the wild-type protein is not clear. One possibility is that it is less stable. Since coimmunoprecipitation experiments were routinely carried out 28 h after transfection while extracts of cells 48 h after transfection were used for luciferase assays, a reduced stability of the protein could explain our results.

FIG. 4.

Coimmunoprecipitation experiment. (A) REF52 fibroblasts were transiently transfected with expression plasmids for wild-type (LTwt) or mutant LT proteins under the control of the CMV promoter, and extracts were prepared 28 h thereafter. The antibody used for immunoprecipitation (IP) was anti-CBP (αCBP), and coprecipitated LT antigen was visualized by immunoblotting. The anti-E1A (αE1A) antibody served as a control, and the input represents the signal obtained by 1/10 of the amount of cell extract used for immunoprecipitation. (B) Cells were cotransfected with HA-tagged T antigen and Flag-PCAF. Immunoprecipitation was carried out with anti-HA antibody, and immunoblotting was done consecutively with anti-Flag, anti-CBP, and anti-HA antibodies. Wt, wild type.

If binding of these coactivators was indeed functionally relevant, overexpression of p300 should cause an increase in the transactivation by the LT antigen of E2F-regulated promoters. To test this, cotransfection experiments were carried out with cyclin A and thymidine kinase promoter-luciferase constructs, LT antigen, and p300 (Fig. 5). Cotransfection of p300 increased the transactivation of the cyclin A promoter construct by LT antigen about twofold. p300 did not stimulate promoter activity in the absence of LT antigen. A mutant form of p300 (p300d33) defective in the binding of a variety of proteins, notably PCAF (30, 37), did not support transactivation, which is in agreement with earlier observations (23). This may indicate an involvement of a complex of coactivators including PCAF. To examine this possibility, we used wild-type and mutant versions of LT antigen and looked for coimmunoprecipitation of both CBP/p300 and PCAF (Fig. 4B). Clearly, mutant LT proteins had lost interaction not only with p300 but also with PCAF, supporting the assumption that CBP/p300 binding to LT antigen recruits PCAF and that this HAT complex stimulates the transactivation of genes by LT antigen. Confirming this conjecture, we found in transient transfection assays that not only cotransfection of p300 but also that of PCAF stimulated transactivation of the cyclin A promoter-luciferase construct (Fig. 5). Cotransfection of both PCAF and p300 resulted in strong stimulation of the transcriptional activation by Py LT antigen, while cotransfection of PCAF and p300d33 led to considerably diminished activity, apparently because of a dominant negative effect of the mutated p300 under these conditions.

FIG. 5.

Stimulation of the transactivation of the murine cyclin A (cycA) and thymidine kinase (mTK) promoters by cotransfected p300, mutant p300d33, and PCAF. For experimental details, see Materials and Methods. The amounts of plasmid DNA used in cotransfections are shown at the bottom. All transient transfection experiments were done at least five times.

Promoter-specific acetylation of histones by LT antigen-mediated recruitment of HAT complex.

If the HAT activity of the coactivator complex were essential for transactivation by LT antigen, this should result in promoter-specific acetylation of histones by the wild-type protein and an absence of this modification when mutant proteins are used. We tested this assumption by carrying out a ChIP analysis of the endogenous thymidine kinase promoter with an antibody directed against acetylated histone H3 prior to and after induction of LT antigen (wild type or mutated). For this experiment, we used quiescent mouse fibroblasts carrying the information for viral proteins under the MMTV promoter (24). The results (Fig. 6) show that the wild-type LT antigen, like serum induction, leads to the formation of acetylated H3 but that all of the transfection-deficient mutant proteins examined, including two with changes within the C-terminal domain, failed to do so. These results strongly support our conclusion that the C-terminal domain functions by recruiting a coactivator complex with HAT activity to transactivate E2F-regulated promoters.

FIG. 6.

ChIP. Chromatin was cross-linked and prepared from exponentially growing cells (exp), from serum-starved quiescent cells (arr), and from quiescent cells after addition of the hormone dexamethasone (Dex) for induction of LT antigen. Immunoprecipitation was carried out with an antibody (Ab) directed against acetylated histone H3. Promoter-specific amplification of precipitated DNA by PCR was done with oligonucleotides specific for the promoter of the murine thymidine kinase gene. Quantification means the difference in the measured intensities of the bands obtained with arrested cells versus those obtained with growing cells or with arrested cells after LT protein induction. The data were corrected for the slight difference in the input amounts. preim, preimmune serum; αAcH3, anti-acetyl-histone H3.

Mutant proteins with changes in the C-terminal domain are competent in the disruption of E2F-pocket protein complexes.

We next asked whether the interaction with CBP/p300 is essential for the dissociation of E2F-pocket protein complexes present in quiescent cells. To answer this question, band shift experiments were carried out with an oligonucleotide corresponding to the E2F site of the murine thymidine kinase promoter. This oligonucleotide was earlier shown to form a complex with E2F4/p130 in arrested cells (31), which is confirmed in Fig. 7. As in the former work, antibodies against p130 or E2F4 eliminated the band completely or partially. This is compatible with the results of ChIP analyses carried out with a variety of E2F-regulated promoters that identify E2F4 and p130 on the promoters in growth-arrested cells (34). In cells growth stimulated by addition of serum, this complex is destroyed as expected. The same occurs if the wild-type LT protein is expressed in the cells. As a source of LT protein we again used our cell lines carrying the inducible LT antigen, which allows induction of the viral protein in growth-arrested, quiescent cells. Contrary to the wild-type LT antigen, neither the E146D mutant protein, which is defective in the interaction with pocket proteins, nor the P43A mutant protein, which has an altered J domain, is able to do so (Fig. 7). Significantly, however, all transactivation-defective mutant proteins with changes within the C terminus were able to almost completely dissociate the E2F-p130 complexes. The P670S mutant protein was produced in smaller amounts than the other mutant proteins, and this led to incomplete dissociation of the E2F complexes. This indicates that the amounts of LT antigen produced in our cell lines conditionally expressing the viral protein are not unphysiologically high. A Northern blot analysis with RNA from the same cells was performed to verify that no thymidine kinase mRNA was produced from the endogenous gene, confirming the lack of transactivation by the mutated viral proteins. These data strongly suggest that the transactivation of E2F-regulated genes by the Py LT antigen is a two-step process. One step, the dissociation of E2F-pocket protein complexes, is essential but, in contrast to the current model, not sufficient. It has to be followed by another, possibly simultaneously occurring, step that involves a function localized to the C-terminal domain of the viral oncoprotein and that appears to require an interaction of this domain with the coactivator proteins CBP/p300 and PCAF. Recruitment of this HAT complex to the promoters results in the acetylation of histone H3, a reaction correlating with promoter activation. Furthermore, we provide evidence (Fig. 2B) that also in this second step an interaction of the LT antigen with chaperones such as HSC70 is required in cis.

FIG. 7.

EMSA. Cells expressing the wild-type LT protein (LTwt) or the mutant LT protein indicated were made quiescent by serum withdrawal for 72 h. Viral protein was then induced by addition of dexamethasone (final concentration, 1 μmol/liter). After 30 h, cell extracts were prepared for EMSA with oligonucleotides comprising the E2F site of the murine thymidine kinase (mTK) promoter. Shown are the complexes formed in quiescent cells without LT antigen (arr) and those formed after induction of LT antigen (Dex). In control experiments, the complex was not present in exponentially growing cells (exp) and was totally or partially removed by addition of antibodies against p130 (αp130) and E2F4 (αE2F4). The Western blot assay was performed on aliquots to verify the synthesis of viral protein after induction by hormone. A further control experiment (bottom) shows the absence of thymidine kinase mRNA synthesis in cells carrying mutated versions of LT antigen while this RNA was induced by either serum addition or the activity of wild-type LT antigen. The 18S rRNA band served as a control for equal loading of RNA.

DISCUSSION

The transcription factor family E2F (35) is a major target of the oncoproteins of DNA tumor viruses (21). E2F is bound by underphosphorylated pocket proteins in concert with histone deacetylases and methylases (14, 27), thereby blocking gene expression. When phosphorylated by cyclin D- and E-associated kinases, pocket proteins are released from E2F, which results in promoter activation.

Viral oncoproteins can stimulate gene expression in quiescent cells by binding to underphosphorylated pocket proteins. This results in dissociation of the E2F-pocket protein complexes, thereby evading the requirement of phosphorylation steps. The capacity of the viral oncoproteins to interact with pocket proteins is essential and has been considered sufficient for transactivation. There is, however, some evidence from previous studies that growth induction and transformation by SV40 requires an unknown function(s) of LT protein in addition to the interactions with chaperones, pocket proteins, and p53 (6, 7, 10, 29). Here we extend our earlier observation that the transactivation of cyclin A by Py LT antigen is abolished by a mutation in the zinc finger of the protein (31). We present evidence that the model described above, specifically that the transactivation by T antigens is merely due to the dissociation of E2F-pocket protein complexes, is too simple because all of those mutant Py LT proteins with changes within the C-terminal region that were found to be defective in transactivation still caused a dissociation of E2F complexes. This region was shown to interact with the coactivator proteins CBP and p300, and our analysis indicates that this contact may be as essential for the transactivating potential of the LT antigen as the interaction with pocket proteins. Our data agree with the report (8) that mutation of the proline at position 670 (671 in reference 8) eliminates binding of p300. This mutant protein was shown to be defective in transforming activity, suggesting that the transactivation of E2F-regulated genes plays an important part in this LT antigen-mediated process. We can therefore assume that the other mutant proteins, which are inactive in binding of the histone acetylase complex and in transactivation, are likewise defective in transformation.

The amino acids whose mutation resulted in a lack of binding to CBP/p300 are spread out over a range of more than 200 amino acids, indicating that the binding domain for p300 on the LT antigen is complex and may depend more on the three-dimensional structure of the protein than on a short, linear amino acid sequence. A similar situation exists in the area of p53 used for binding to the helicase domain of the SV40 LT antigen (19) and the interaction surface between the papillomavirus helicase E1 and its matchmaker E2 (1). This contrasts with the well-defined binding motifs for pocket proteins and chaperones in the SV40 and Py LT antigens (5), as well as the recently reported binding sites for Bub1 (9) and Cul7 (3) within the N terminus of the SV40 LT protein. This difference may be explained by the dominant structural requirements for the formation of the hexameric replicative helicase. This part of the LT antigen molecule is under strong evolutionary pressure because of the vital role of the origin binding capacity and the helicase activity for the replication of viral DNA and contrasts with the apparently much more flexible N terminus of the protein. The amino acids we have chosen for the generation of mutant proteins are conserved in SV40, which makes it likely that the results obtained with the Py LT antigen can be extended to the SV40 LT antigen. In contrast, all of the amino acids known to be involved in the binding of p53 to the SV40 LT antigen are not conserved in the Py LT antigen, explaining the difference between the two proteins with regard to this property.

Considering that histone deacetylases are recruited by pocket proteins to E2F-regulated promoters, which leads to deacetylation of histones (27), HATs have to reverse this reaction for the onset of transcription. Together with the dissociation of the E2F-pocket protein complexes and the accompanying removal of histone deacetylases, reacetylation of histones has to take place by the LT antigen-mediated recruitment of CBP/p300 together with PCAF. CBP/p300 was shown earlier to promote the expression from E2F-regulated genes in serum-stimulated cells (2, 15, 36). Under these conditions, binding of CBP/p300 might occur after phosphorylation events mediated by signal transduction. In accordance with this assumption, it was reported that phosphorylation of E2F5 allows its interaction with CPB/p300 (22). In serum-starved quiescent cells, such reactions are unlikely to take place, making a second function of the LT antigen necessary in order to bring HATs to the promoters. In this context it is interesting that a chromatin remodeling function was recently also shown for the adenovirus E1A protein (16). Considering the fact that the two different E2F-regulated promoters used here show identical requirements, we propose that following binding of the LT antigen to the pocket protein (probably in the form of a dimer), the dissociation of the E2F complex and the recruitment of HATs, followed by histone acetylation, are concerted reactions (Fig. 8 shows a model). The latter assumption is supported by the results of the complementation assays (Fig. 2), by the capacity of the complementing mutant proteins to form dimers, and by our observation that histone acetylation does not occur in cells expressing a mutant form of LT antigen that is unable to bind pocket proteins (Fig. 6). Final proof of this model would require ChIP analyses with antibodies against the LT antigen, p300, and PCAF. The presumably very transient binding of this complex to promoters may, however, render such a task difficult. Although our data show a correlation between the absence of histone H3 acetylation and a defect in the binding of CBP/p300-PCAF to the LT antigen, they do not exclude the possibility that proteins other than histones are acetylated in addition. For instance, PCAF was shown to acetylate E2F1 (20) and to bind to and acetylate the LT antigen in a p300-independent manner (39), a reaction that appears to play a role in the replication of viral DNA but not in transactivation.

FIG. 8.

Model for the transactivation of E2F-regulated genes. Binding of LT antigen to the pocket protein, followed by dissociation of E2F-pocket protein complexes, as shown in panel A, is insufficient; rather, LT antigen probably binds as a dimer and two reactions, the dissociation of E2F complexes and histone acetylation, occur as concerted reactions (B). Pocket protein p130 (itself a complex with histone deacetylases and methylases) is shown to bind to E2F (E2F4) in serum-starved cells.