Transactivation of Murine Cyclin A by Polyomavirus Large and Small T Antigens

Abstract

Polyomavirus large and small T antigens cooperate in the induction of S phase in serum-deprived Swiss 3T3 cells. While the large T antigen is able to induce S phase-specific enzymes, we have recently shown that both T antigens contribute to the production of the cyclins E and A and that the small T antigen is essential for the induction of cyclin A-dependent cdk2 activity (S. Schüchner and E. Wintersberger, J. Virol. 73:9266–9273, 1999). Here we present our attempts to elucidate the mechanisms by which the large and the small T antigens transactivate the murine cyclin A gene. Using Swiss 3T3 cells carrying the T antigens and various mutants thereof under the hormone-inducible mouse mammary tumor virus promoter, as well as transient-cotransfection experiments with the T antigens and cyclin A promoter-luciferase reporter constructs, we found the following. The large T antigen activates the cyclin A promoter via two transcription factor binding sites, a cyclic AMP responsive element (CRE), and the major negative regulatory site called CDE-CHR. While an intact binding site for pocket proteins is required for the function of this T antigen at the CDE-CHR, its activity at the CRE is largely independent thereof. In contrast, an intact J domain and an intact zinc finger are required at both sites. The small T antigen also appears to have an influence on the cyclin A promoter through the CRE as well as the CDE-CHR. For this an interaction with protein phosphatase 2A is essential; mutation of the J domain does not totally eliminate but greatly reduces the transactivating ability.

The induction of S phase of the cell cycle is a complex reaction usually initiated at the cell surface through binding of ligands to receptors. Signal transduction pathways lead to the synthesis of S phase-specific enzymes and regulators, including the cyclins E and A. DNA tumor viruses require cells in S phase because they heavily depend on the cellular DNA synthesis machinery for the replication of their own DNA. Since they frequently infect differentiated cells, they have to interfere with the cellular mechanisms of growth regulation in order to drive cells out of the G₀ phase and into the S phase (reviewed in reference 19). This is accomplished by viral proteins which interact with various intermediates of the signal transduction pathway downstream of events taking place at the cell surface. One class of target for such viral proteins are the pocket proteins: pRB and its relatives p107 and p130. They negatively regulate members of the transcription factor family E2F (reviewed in reference 8). Adenovirus E1A, human papillomavirus (HPV) E7 protein, and the large T (LT) antigens of simian virus 40 (SV40), and polyomavirus (Py) bind to pocket proteins and cause a dissociation of the repressive complexes, which results in the transactivation of E2F responsive genes (19). Another region of the pleiotropic T antigens which is essential for this reaction is the J domain (reviewed in references 4 and 16), a sequence present in many chaperones capable of binding Escherichia coli DNA K-type proteins such as the mammalian HSP70 and HSC70 proteins. Binding of LT stimulates HSC70's ATPase activity and is supposed to be important for rearrangements of protein complexes.

We have previously shown that PyLT protein can transactivate genes coding for DNA synthesis- and precursor-producing enzymes in serum-starved 3T3 mouse fibroblasts (20, 23). Still, such cells do not enter S phase to a considerable extent (22). This requires the cooperation with the small T (ST) antigen of the virus. ST antigen from SV40 or Py associates with the protein phosphatase 2A (PP2A), which results in the inactivation of the enzyme or in a change of its substrate specificity (5). ST proteins share with LT antigens the N-terminal 82 (SV40) or 79 (Py) amino acids. This region includes the J domain which is therefore a second interaction domain of ST protein.

By investigating the levels of the G₁/S and S phase-specific cyclins we found that cyclin A was undetectable in serum-starved fibroblasts and that both LT and ST antigens were able to efficiently cause the induction of cyclin A protein (32). The amounts of cyclin E, on the other hand, only marginally increased upon the addition of serum or expression of T antigen. We found that LT by itself was very inefficient in inducing cyclin E- and cyclin A-dependent cdk2 activity, a requirement for S phase induction. ST antigen strongly contributed to the development of these cyclin-dependent kinase (cdk) activities by causing the removal of the cdk inhibitor p27 (32). This reaction is therefore a major contribution of ST to the combined S phase-inducing capacities of PyLT and -ST.

The transcriptional regulation of cyclin A has been studied in recent years (13, 14). The minimal promoter of the cyclin A gene carries a cyclic AMP responsive element (CRE), which is a well-known binding site for transcription factors of the CREB-ATF group, a reversed CCAAT box, and a composite element called CDE-CHR. Versions of the latter are also found in other cell cycle-regulated promoters such as those of the cdc25C, the B-myb, and the cdc2 genes (47), as well as that of the p130 gene (9). Proteins binding to this element appear to influence promoter activity negatively in G₀ and most of G₁. The CRE binding proteins, on the other hand, are regulated by phosphorylation; only the phosphorylated form is capable of binding the stimulatory coactivator protein CBP (CREB-binding protein) (reviewed in reference 6). Hence, the cyclin A promoter appears to be regulated by signal transduction pathways in two steps. One is the phosphorylation step of the CRE binding protein (7); this is, however, not sufficient to activate the promoter. In a further step the repressive proteins binding to the CDE-CHR have to be removed by as-yet-unknown mechanisms late in G₁ (14). Although candicate proteins were described (9, 18, 26), it is not quite clear yet which proteins bind to the CDE-CHR. The CDE resembles a binding motif for E2F, and recently published chromatin immunoprecipitation experiments show that E2F 4 in combination with p130 binds to the cyclin A promoter in arrested cells (39). Furthermore, pRB, either directly or indirectly, seems to play a role in the regulation of cyclin A (17, 25, 38, 45).

Cyclin A was shown to be transactivated by E1A, HPV E7, and SV40 and PyLT and PyST antigens (24, 27, 28, 32, 43, 44). So far, however, little is known about the mechanisms involved. We describe here which functions of PyLT and PyST antigens are required to transactivate the cyclin A gene and which transcription factor binding sites of the cyclin A promoter are involved.

We used our previously developed Swiss 3T3 cell lines conditionally expressing LT or ST antigen (22) and established new ones producing mutated versions of the proteins. In addition, cotransfection experiments were carried out using luciferase constructs harboring the wild-type cyclin A promoter or mutated versions thereof. We found that both T antigens affect the murine cyclin A promoter in complex ways. LT affects promoter activity via the CDE-CHR in a way dependent on an intact binding site for pocket proteins. In addition, we obtained evidence that LT also exerts a minor transactivating potential via the CRE site. For this latter activity, the site of interaction with pocket proteins can be mutated with little effect on the activity. On the other hand, the J domain is required for LT activity at both sites. The same holds true for the zinc finger of LT, which is located in the C-terminal half of the protein. ST antigen likewise transactivates the cyclin A gene via the CRE as well as the CDE-CHR and requires the binding site for PP2A and the J domain for this activity.

MATERIALS AND METHODS

Cell culture and stable transfections.

Swiss 3T3 fibroblasts and derived cells conditionally expressing the Py T antigens (22) were maintained in Dulbecco 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 PyLT, PyLTRB⁻, PyST, and PyST(ins107AL) were described previously (22, 32). Cell lines conditionally expressing the J domain mutants of LT (LT_H42Q, LT_P43A, LT_D44N) and of ST (ST_H42Q and ST_P43A) or LT mutated in the zinc finger motif (LT_C452S) were produced by standard Polybrene-assisted transfections with 10 μg of DNA overnight. After selection by either Geneticin (400 μg/ml) or hygromycin (150 μg/ml) for 4 to 6 weeks, resistant colonies were expanded and tested for T antigen expression before and after the addition of dexamethasone. Cell lines exhibiting very low expression in the absence of hormone but high expression after its addition were chosen for further work.

Plasmids.

For stable transfections mouse mammary tumor virus (MMTV)-LT and MMTV-ST expression plasmids were used as in previous work (22). Mutations in the J domain and the zinc finger were introduced into LT with the help of the mutagenesis kit of Stratagene according to the supplier's recommendations. To create the MMTV-ST_P43A and MMTV-ST_H42Q plasmids, an N-terminal fragment of the mutated LT (cut by BstZ17I and coding for amino acids 1 to 66) was ligated to the C-terminal part of ST also digested with BstZ17I. 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 cytomegalovirus promoter. The various murine cyclin A promoter-luciferase constructs were a kind gift of J. M. Blanchard, Montpellier, France.

Antibodies, protein extraction, and immunoblotting.

Cyclin A was detected by a specific goat polyclonal antibody (C-19G) purchased from Santa Cruz. LT and ST antigens were detected by a rabbit polyclonal antibody directed against the LT or ST common N-terminal region of these proteins. Adequate and comparable expression of these proteins after the addition of dexamethasone was assured by immunoblotting in each experiment.

Cells were harvested and lysed in buffer A (20 mM Tris-Cl [pH 8.0], 100 mM NaCl, 1 mM EDTA, 0.5% NP-40, 2 mM phenylmethylsulfonyl fluoride, complete protease inhibitor cocktail [Boehringer Mannheim]) as described elsewhere (1). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by immunoblotting, was performed as described previously (32)

RNA extraction and Northern blotting.

Total RNA was isolated using TRIzol reagent (Life Technologies/GIBCO-BRL) as recommended by the supplier, separated by morpholinepropanesulfonic acid-agarose gel electrophoresis, and blotted to a nylon membrane (GeneScreen; NEN). Hybridization was performed with a ³²P-labeled cDNA fragment of murine cyclin A (EST clone #A414533; Genome Systems, Inc.).

Transient transfections and luciferase reporter assays.

Asynchronously growing REF52 cells were seeded at 10⁵ cells per six-well plate. The next day, the cells were transfected by the polyethylenimine (PEI) method as described previously (3). Briefly, 87 μl of a 50% (wt) solution of PEI 25000 (catalog no. 40827-7; Aldrich) in water were added to 100 ml of water, and the pH was adjusted to 7.0 with HCl. At 60 min before transfection, the cells were put in DMEM without serum and antibiotics (800 μl per six-well plate). Then, 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). Next, 3 μl of PEI was added dropwise to the DNA; the mixture was then 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 after transfection the cells were lysed in luciferase lysis buffer (100 mM potassium 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 on Western blots for the expression levels of cotransfected proteins.

EMSA.

Whole-cell lysates were prepared as described above for immunoblotting. A total of 10 μg of whole-cell protein was used for electrophoretic mobility shift assay (EMSA) with oligonucleotides representing the E2F site of the murine thymidine kinase promoter; EMSA were performed as described elsewhere (15, 23). For EMSA with oligonucleotides containing either the CRE, the CCAAT box, or the CDE-CHR region of the murine cyclin A promoter, nuclear extracts were prepared from serum-deprived cells using a previously described method (36). Then, 3 μg of nuclear proteins was incubated with oligonucleotides. The following oligonucleotides were used: CRE, 5′-GGGTCGCCTTGAATGACGTCAAGGCCGCGA; NF-Y, 5′-GAGCGCTTTCATTGGTCCATTT; CDE-CHR, 5′-TCAATAGTCGCGGGCTACTTGAACTACAAG; and mTK-E2F, 5′-TTTTGAGTTCGCGGGCAAATGCGAGCA. Supershift analyses were performed with the following antibodies: C-20 for p130, C-20 for E2F-4, KH95 for E2F-1, C-21 for CREB-1, X-12 for CREM-1, and C41-5.1 for ATF-1 (all purchased from Santa Cruz); a rabbit polyclonal antibody was raised against a peptide of the B subunit of NF-Y (36).

RESULTS

Swiss 3T3 cells stably transfected with hormone-inducible versions of Py T antigens.

We had previously constructed Swiss 3T3 cell lines expressing PyLT or PyST protein under the MMTV promoter (22). These cells do not produce the viral protein unless dexamethasone is added, resulting in the rapid production of considerable amounts of the viral protein within 4 h regardless of whether the cells are growing or quiescent. Cell lines harboring a mutant of LT incapable of binding pocket proteins (LTRB⁻) were also produced earlier. These cell lines were previously used to study the induction of S phase-specific enzymes in serum-starved fibroblasts (20, 23). For the study of the mechanism of cyclin A transactivation by LT and ST antigens, we have extended our repertoire of cell lines expressing mutant versions of the T antigens. Using previously described methods we created cell lines expressing LT mutated at various positions within the central HPD region of the J domain (H42Q, P43A, and D44N) and a point mutant of the zinc finger region (C452S). Cells carrying an ST mutant with a destroyed binding site for PP2A, ST(ins107AL), were described previously (32). In addition, we also created cell lines expressing the different J domain mutants of ST analogous to those of LT. In each experiment the expression of the T antigens was examined by immunoblotting to assure that the amounts of viral proteins produced are comparable (see Fig. 1C for LT and Fig. 3C for ST).

FIG. 1.

Induction of cyclin A by PyLT. (A) Northern blot analysis of RNA from Swiss 3T3 cells conditionally expressing wild-type LT, LTRB⁻, a J domain mutant (P43A), or a zinc finger mutant (C452S). The cells were serum deprived and treated subsequently with dexamethasone (Dex) for the indicated times. For comparison, arrested cells were reinduced with serum. Equal loading of RNA was assured by methylene blue staining of the 18S rRNA (shown below the cyclin A mRNA blots). (B) Western blot analysis for cyclin A of protein extracts prepared in parallel to the RNA samples. Cyclin A expression was detected by using a rabbit polyclonal antibody. (C) Western blot analysis of LT (wild type and mutants) induced after incubation of cells with dexamethasone for 32 h.

FIG. 3.

Induction of cyclin A by PyST. (A) Northern blot analysis of RNA prepared from Swiss 3T3 cells conditionally expressing ST, ST(ins107AL), or a J domain mutant (P43A). The cells were treated as in Fig. 1. (B) Western blot analysis for cyclin A of protein extracts prepared in parallel to the RNA samples. (C) Western blot analysis of ST expression (wild type and mutants) after treatment of cells with dexamethasone for 32 h.

Transactivation of cyclin A by PyLT antigen.

Cells containing the plasmids for the inducible expression the various versions of the T antigens were arrested by serum withdrawal. Dexamethasone was then added to the medium, and incubation continued for 32 h. Cell extracts were prepared to test for cyclin A expression at the mRNA and the protein level. Results of these experiments are summarized in Fig. 1. They show that cyclin A induction took place at the level of transcription, as can be seen from the amounts of cyclin A mRNA before and after addition of dexamethasone (Fig. 1A). Cells were growth stimulated by the addition of serum, and the levels of cyclin A mRNA were determined. Through an analysis of various mutants of LT it became apparent that mutations at all three sites which were investigated—the LXCXE sequence of the pocket protein binding site, the J domain (shown are the data obtained with one of the mutants, LT_P43A, the other two mutants, LT_H42Q and LT_D44N, yielded identical results), and the zinc finger—greatly reduced or abolished the transactivating ability of the protein. This is also apparent from the analyses of the protein levels with the exception of the LTRB⁻-expressing cells. This mutant exhibited cyclin A protein signals, albeit significantly more weakly than those obtained with wild-type LT (Fig. 1B). This result corresponds well with our previous observation (32). Since this earlier study was done exclusively at the protein level, we concluded that an interaction with a pocket protein is not involved in the transactivation of cyclin A. Our mRNA analyses described here, which showed a weak signal in the LTRB⁻ cells (Fig. 1A, upper right panel), are not in full keeping with this conclusion. Because of this discrepancy we have analyzed two more LTRB⁻ cell lines and found that the signals for cyclin A mRNA and protein vary. They were most significant in the cell line shown in Fig. 1 and used earlier (32) and negligible in another cell line, while the third cell line exhibited levels between the two extremes. Still, the cyclin A mRNA produced in the cell line shown here, although significant compared with that produced in cells carrying the J domain or the zinc finger mutant, was quite small. This may, however, suffice to allow the production of some cyclin A and the protein may accumulate to levels high enough to exhibit the signal seen in immunoblots. Data obtained in the transient-transfection experiments agree with this interpretation (see below).

Next we used transient transfections to analyze the transcriptional activation of the cyclin A promoter by LT in more detail. The cyclin A promoter-luciferase construct contained the upstream region of the mouse cyclin A gene from −177 to +100 (relative to the 3′-most transcription start site) linked to the firefly luciferase gene. Promoter constructs with mutations at the CRE, the CCAAT, and the CDE-CHR (Fig. 2A) were also used. Cyclin A-luciferase constructs were cotransfected with LT antigen and mutants thereof into REF52 cells. We have chosen the rat embryo fibroblast cell line REF52 for transient-transfection experiments because these cells yield a high transfection efficiency (considerably higher than Swiss 3T3 mouse fibroblasts) but still can be very effectively growth arrested by reducing the serum concentration to 0.2% (better than NIH 3T3 cells). Expression of LT (and ST; see below) was always determined by Western blotting and showed similar levels for all the constructs used (see Fig. 2B for LT and Fig. 4A for ST below the graphs). Wild-type LT activated the promoter in quiescent cells ca. 20-fold (average of five independent experiments). Mutations of the pocket protein binding motif, the J domain (P43A), or the zinc finger (C452S) reduced the transactivation to fourfold or less. Importantly, mutation of the J domain had the strongest effect, while mutation of the RB binding site did still allow a small but significant stimulation of the cyclin A promoter (Fig. 2B). This is in accordance with the Northern and Western blot analyses of Fig. 1 and indicates that the promoter is slightly activated by LT even when the RB binding motif is mutated. Additional information was gained when mutated versions of the promoter were used (Fig. 2C). A mutation of the CRE slightly reduced the transactivation by LT (ca. 20% reduction), but a mutation of the CDE had a considerably stronger effect (80% reduction). All these data suggest that LT influences the promoter activity in arrested cells primarily via the CDE site and that an interaction with protein complexes binding to the CRE may play an additional role. Neither mutation of the reversed CCAAT box nor removal of 98 bp at the 5′ end of the promoter, which include a GC box binding Sp1, had a significant effect on the transactivation by LT antigen.

FIG. 2.

Transactivation of the cyclin A promoter by LT. (A) Schematic representation of transcription factor binding sites on the murine cyclin A promoter. (B) A plasmid containing the wild-type cyclin A promoter linked to the luciferase gene as a reporter was cotransfected into REF52 cells together with expression plasmids for various forms of LT. Transfected cells were serum arrested for 48 h, and luciferase activity was measured and corrected for transfection efficiency by measuring β-galactosidase activity in parallel. Fold induction (average of five independent experiments) was calculated by comparing cells expressing LT with cells not expressing LT. Background levels of luciferase activity in mock-transfected cells was 1 to 5% of that in cells transfected with promoter constructs alone. A representative Western blot analysis of LT protein expression in these experiments is shown below the graph. (C) Plasmids containing either the wild-type cyclin A promoter or mutated forms were cotransfected with a plasmid coding for wild-type LT. The fold induction of luciferase activity was determined as in panel B. (D) Basal cyclin A promoter activity (in the absence of T antigen) in transiently transfected, serum-starved REF52 cells. Since these values were determined for each one of the experiments summarized in Fig. 2C and in Fig. 4B, this experiment is a representative example of more than 10 assays. (E) Effect of mutations in LT antigen on the transactivation of wild-type and mutated versions of cyclin A promoter-luciferase constructs. This experiment was done three times, and a representative example is shown.

FIG. 4.

Transactivation of the cyclin A promoter by ST. (A) The wild-type cyclin A promoter was cotransfected into REF52 cells together with expression plasmids for different forms of ST. A representative Western blot analysis of ST protein expression in these experiments is shown below the graph. (B) Various cyclin A promoter constructs were cotransfected with an expression plasmid for wild-type ST. The fold induction (average of five independent experiments) was determined as described in Fig. 2.

There have been reports that mutations of transcription factor binding sites within the cyclin A promoter affect basal transcription (7, 14). For this reason we determined the basal activity of the different promoter constructs in serum-starved, quiescent cells without T antigen expression (Fig. 2D). From this it becomes clear that the mutation of the CRE or the CCAAT box affected the basal promoter activity in arrested REF52 cells only slightly but that the mutation of the CDE had a clear effect resulting in strong promoter activity in quiescent cells. This supports earlier reports showing that the cyclin A promoter is repressed through this site in growth-arrested cells (14). When the CDE was mutated, almost-maximal levels were already attained in the arrested state and there was only little further stimulation by either LT antigen (Fig. 2C) or serum (not shown) which may be due to effects on the CRE only.

Finally, the effect of LT mutations on the transactivation of wild-type and mutated versions of promoter-luciferase constructs is summarized in Fig. 2E. This shows that an inactivation of the J domain had a strong effect on the transactivation of all promoter constructs. The second site whose mutation severely reduced the transactivation potential of LT in every case was the zinc finger, while the pocket protein binding motif was essential for transactivation at the CDE (the CRE being mutated) but largely dispensable for the activity of LT at the CRE (the CDE being mutated).

Transactivation of cyclin A by PyST antigen.

ST antigen also very efficiently transactivated the cylin A gene at the transcriptional level. Cyclin A mRNA and cyclin A protein were induced when ST was generated in quiescent cells (Fig. 3). A mutation of the PP2A binding motif (ins107AL) completely eliminated the transactivation, suggesting that the inactivation of PP2A or a modification of its specificity by ST is involved. Also, the J domain of ST appears to play an important role, since a mutation of the HPD domain greatly reduced the transactivation, although the effect was not as pronounced as that seen by mutation of the PP2A binding site. The influence of the J domain on the transactivating capacity is therefore less pronounced in the case of ST compared to LT protein.

Transient-transfection experiments with cyclin A promoter-luciferase constructs and ST antigen are summarized in Fig. 4. ST protein transactivated the cyclin A promoter on average ca. 10-fold (five independent experiments). This is significantly lower than the transactivation of cyclin A by LT (ca. 20-fold). In cells that express T antigen in a stable manner the situation is opposite. ST induces more strongly than LT. This could be due to the fact that ST is always expressed more weakly than LT in transient transfections. Mutants defective in the J domain (H42Q and P43A) reduced the activity drastically, indicating that the interaction with chaperones is required for transactivation. We observed an even stronger effect when the binding site for PP2A was mutated. These results are in perfect agreement with the data obtained with the cell lines conditionally expressing ST and mutants thereof (Fig. 3).

Using mutated versions of the cyclin A promoter in transient transfections, together with ST, we observed that a mutation of the CRE reduced the fold induction by ST to ca. 50% (Fig. 4B). An even stronger effect was seen when a promoter mutated in the CDE was used. This indicates that, like LT, ST protein acts through the CRE as well as through the CDE-CHR.

In vitro protein binding to transcription factor binding sites.

Since many previous experiments have conclusively shown that viral proteins such as PyLT antigen cause a destruction of protein complexes forming at binding sites for transcription factors such as, for instance, E2F. Thus, we finally carried out EMSA experiments to analyze the effects of viral proteins on complexes building up at the three transcription factor binding sites on the cyclin A promoter (Fig. 5). Using oligonucleotides comprising these binding regions and nuclear protein extracts of arrested Swiss 3T3 cells not expressing a viral protein, as well as specific antibodies directed against candidate transcription factors, we identified the CCAAT binding protein as NF-Y. Furthermore, we show that the CRE was occupied by CREM and CREB, whereas ATF-1 did not appear to be involved to a similar extent. These results agree well with those of Huet et al. (14). Although the CDE sequence resembles a binding motif for E2F, complexes forming at the CDE-CHR were distinctly different from those forming at a genuine E2F site, such as that of the mouse thymidine kinase promoter which was employed for comparison. Use of antibodies indicated that the latter complexes contained E2F4 and p130, in agreement with earlier results in arrested mouse 3T3 cells (37); in contrast, neither antibody provided evidence for an involvement of these proteins in complexes forming on the CDE-CHR. Although this again agrees with published results (14), the possibility cannot be excluded that a much more intricate structure of the protein complexes forming at the CDE-CHR compared with those at an E2F site does not allow sufficient access for antibodies to cause changes in the complex formation or its size (see Discussion). Quite in contrast to the murine thymidine kinase promoter, where LT causes a dissociation of the pocket protein complexes leading to “free” E2F (23, 37; see also Fig. 5), complexes forming on any of the transcription factor binding sites of the cyclin A promoter hardly changed if T antigen was expressed in the cells, as shown for LT in Fig. 5. Identical results for all three transcription factor binding sites were obtained with ST protein, and even the addition of serum did not substantially change the patterns of complexes (data not shown; see also reference 14).

FIG. 5.

Influence of large T on protein complex formation at transcription factor binding sites of the cyclin A promoter in vitro. EMSA was carried out as described in Materials and Methods using oligonucleotides comprising transcription factor binding sites indicated at the bottom. The antibodies used to test for components of the observed complexes are indicated at the top.

DISCUSSION

Cyclin A with its associated protein kinases plays an important role in the cell cycle of mammalian cells from the beginning of S phase until the start of mitosis (10, 11, 12, 29, 31, 34, 35, 41). It is therefore not surprising that DNA tumor viruses code for proteins which are capable of transactivating cyclin A and of inducing cyclin A/cdk2 activity (24, 27, 32, 43, 44). In case of Py this requires the concerted activities of two proteins, LT and ST antigen (32). Although LT and ST can both stimulate the expression of cyclin A, ST is required to activate the affiliated protein kinase. We describe here attempts to analyze the mechanisms by which the two viral proteins transactivate the cyclin A gene. The promoter regions of the human and the murine cyclin A genes were described previously (13, 14), and the sequences of the transcription factor binding sites were found to be identical. Both promoters contain an Sp1 binding site which is upstream of the minimal regulated promoter, a CRE, a reversed CCAAT box, and a composite motif called CDE-CHR. Data in the literature indicate that growth regulation of the promoter is achieved on the one hand through the CRE (7), where phosphorylation of the CRE binding proteins, regulated through signal transduction pathways, causes them to become active as transcription factors (6), and on the other hand, at the CDE-CHR, which functions as a binding site for negative regulators (14, 47). We found that PyLT antigen affects the cyclin A promoter on both of these sites, albeit to different extents. Most of LT's activity is attributable to a derepression at the CDE-CHR region. This requires the pocket protein binding site of LT, the J domain, and the zinc finger region. LT might associate with proteins bound to the CDE-CHR and recruit chaperones, which in an ATP-dependent reaction may cause a change in the chromatin structure at this region. A recent report (33) concluded that the transactivation of cyclin A by PyLT antigen is dependent on the RB binding motif but is independent of an intact J domain. This was based on transient transfections of NIH 3T3 fibroblasts. However, also in this case a J domain mutant exhibited a reduced activity, but this reduction was much less pronounced than in both of our systems. It is therefore possible that NIH 3T3 cells are influenced in slightly different ways.

EMSA shows, at best, minor changes of protein complexes forming at the transcription factor binding sites of the cyclin A promoter when LT antigen is expressed in quiescent cells. This contrasts to the situation at the murine thymidine kinase promoter (Fig. 5). Our data show that despite the great sequence similarity between the CDE and an E2F binding site, complexes forming at regular E2F motifs are at least in vitro distinctly different from those forming at the CDE-CHR. In particular, while E2F4 and p130 can be found by EMSA to bind to the E2F motif of the murine thymidine kinase promoter, these two proteins cannot be detected at the CDE-CHR of the cyclin A promoter with nuclear extracts from arrested cells, although a recent study using chromatin immunoprecipitation provided strong evidence for their presence on the promoter in vivo (39). In the in vitro assays, proteins binding to the CHR part of the composite binding site possibly interfere with the accessibility of antibodies in this assay. Even after serum stimulation there is no significant change to be seen in complexes forming at the CDE-CHR (data not shown). However, in vivo footprints indicate that the site is occupied in quiescent cells, whereas it becomes free when cells are growth stimulated (14). This indicates that EMSAs do not necessarily reveal the actual changes occurring at this site. The same may hold true for the effects of the T antigens because it was shown by in vivo footprinting of another promoter carrying a CDE-CHR, that of the cdc25C gene, that proteins binding to the CDE-CHR are removed by SV40 T antigen (46). Since different promoters which all carry CDE-CHR motifs are activated at different times during the cell cycle and are therefore probably not regulated in exactly the same way, it would be interesting to see, using in vivo footprinting, whether Py T antigens can change protein binding to the CDE-CHR of the cyclin A promoter.

LT antigen also affects the cyclin A promoter through the CRE site. This activity, which is significantly weaker than the effect via the CDE-CHR, is largely independent of an interaction of the viral protein with pocket proteins. This is particulary evident in the transient-transfection experiments in which a cyclin A promoter-luciferase construct is used which carries a mutation in the CDE. The residual activity of LT is then only marginally affected by mutation of the RB binding motif. In contrast, when a promoter mutated at the CRE is used, the viral protein affects the promoter only via the CDE-CHR, and the interaction of LT with pocket proteins is required. Interestingly, the activity of LT on the CRE as well as on the CDE-CHR is strongly dependent on an intact J domain and the zinc finger. The role of the zinc finger is so far unknown. Taking into account that LT binds to the origin of replication on polyomavirus DNA as a double hexamer and that this requires the zinc finger (17a), it is possible that oligomerization is also necessary for transcriptional regulation and may require an intact zinc finger motif. Alternatively the zinc finger might play a role in the interaction with cellular proteins involved in transcription (see, e.g., reference 30).

At the present time it is unknown with which cellular protein(s) bound to the CRE the viral protein interacts. One possibility is CBP which is bound to phosphorylated CREB. PyLT antigen was shown to interact with p300/CBP (21). On the other hand, CREB should be largely unphosphorylated in quiescent cells and as such should not bind CBP. It is interesting to note, however, that PyLT was found to preferentially bind CBP and p300, which is active as a histone acetyltransferase (21) and may thus be able to recruit this particularly active subspecies of the coactivator to the CRE.

Our data indicate that proteins bound to the CRE are one target of ST antigen which potentially cause their enhanced phosphorylation through inhibition of the dephosphorylation reaction, the capacity of ST to bind to PP2A being essential for this reaction. In support of this assumption PP2A was reported to remove phosphate from serine-133 of CREB in HepG2 cells (40), and transforming growth factor β was found to downregulate cyclin A by causing dephosphorylation of CRE binding proteins, a reaction which is counterbalanced by SV40 ST protein (42). On the other hand, peptide inhibitors of protein phosphatase 1 were found to increase phosphorylation of CREB in NIH 3T3 fibroblasts; in this study, SV40 ST antigen was found to have no effect on the phosphorylation status of CREB (2). Furthermore, while SV40 ST antigen was found to transactivate the cyclin A promoter in CV1 cells, in this case an interaction with PP2A was nonessential, whereas the J domain was important (28). Thus, it appears that different enzymes may be responsible for dephosphorylation of CRE binding proteins, depending on the cell type and physiologic conditions and ST proteins may influence cyclin A promoter activity in more than one way. The fact that a promoter mutated at the CDE is only marginally induced by ST indicates that proteins bound to this site are additional targets of ST antigen. So far there is no information as to the nature of these interacting proteins or the mechanism by which ST influences promoter activity via the CDE-CHR except for the fact that in our experimental system an association with PP2A appears to be essential because, as shown in Fig. 3, mutation of the PP2A binding site in ST almost completely abolishes the transactivating capacity of the protein. Again, in vivo footprinting might be helpful in further elucidating the functions of ST on the cyclin A promoter.