The N-terminal region of E2F-1 is required for transcriptional activation of a new class of target promoter.

Because of its expression in numerous cells, the herpes simplex virus thymidine kinase promoter (HSV-TK) is one of the best characterized promoters. Using the HSV-TK promoter as a model system, we have defined a new mode of E2F-1 transcriptional activation which utilizes the N-terminal region of E2F-1. We demonstrate that E2F-1 strongly activated HSV-TK, but in the absence of consensus E2F DNA elements. Nonetheless, E2F-1 could bind to GC-rich elements, which were conclusively identified in classic studies of HSV-TK as SP-1 sites. Second, the transcriptional activation of HSV-TK required the entire E2F-1 protein, including the N-terminal 89 amino acids. In contrast, the N-terminal 89 amino acids of E2F-1 were dispensable for transcriptional activation through consensus E2F sites. Third, we demonstrated that S phase entry is not sufficient for activation of HSV-TK by E2F-1, while the activation through consensus E2F sites is strictly linked to the cell cycle. Taken together, the activation of HSV-TK by E2F-1 proceeds by a different mechanism directed in part through the N-terminal region of E2F-1 and may be uncoupled from the known cell cycle regulatory role.

The E2F transcription factors have a major role in cellular pathways, including the regulation of the G 1 /S transition and the activation of the S phase gene expression program (for reviews, see Refs. [1][2][3][4]. Numerous growth control genes are regulated by E2F (e.g. N-myc, cdc-2, c-myc, dihydrofolate reductase). E2F function is regulated by protein-protein interaction with the retinoblastoma (RB) 1 family of tumor suppressor proteins (RB, p107, p130) and with the cell cycle machinery (cyclin E, cyclin A, CDK2). The regulation of E2F by RB has served as a paradigm for growth suppressor function. Expression of RB leads to arrest in G 1 , and one mechanism works through the inhibition of E2F and the majority of the S phase gene expression program. Conversely, disruption of the RB-E2F interaction leads to activation of S phase genes and progression through the cell cycle. Either displacement of RB by viral oncogenes (e.g. E1A) or the phosphorylation of RB by cell cycle-dependent kinases (CDKs) can dissociate the RB⅐E2F complex. Both events trigger S phase entry. Because E2F is a key regulatory target, we and others have demonstrated combinatorial complexity in E2F interactions with RB family members. Different E2F⅐RB family complexes dictate the G 1 /S and G 0 /G 1 cell cycle transitions in concert with the cell cycle machinery (5)(6)(7)(8)(9). E2F is a critical cellular regulatory protein since perturbation in expression levels lead to apoptosis or oncogenicity (10 -13).
We and others have demonstrated that the N-terminal region of E2F-1 is an important regulatory region. We have demonstrated that cyclin A and CDK2 can interact directly with the extreme N-terminal region of E2F-1. The association results in direct phosphorylation of E2F-1 and the heterodimeric partner DP-1. The phosphorylation leads to inhibition of DNA binding and subsequent inactivation of E2F-1-dependent function. We and others have hypothesized that the phosphorylation and subsequent inactivation of E2F-1/DP-1 is critical to the inactivation of the S phase gene expression program and to S phase exit (14 -16).
In this report, we have defined a new mode of E2F-1 transcriptional activation which utilizes the N-terminal region of E2F-1. Using the herpesvirus simplex thymidine kinase (HSV-TK) promoter as a model system, we demonstrate that E2F-1 strongly activates HSV-TK, but in the absence of consensus E2F DNA elements (TTTSSCGC, where S is C or G). Nonetheless, we demonstrate that E2F-1 can bind to GC-rich elements suggesting a nonconsensus E2F-1 interaction. Intriguingly, these sites were previously and conclusively identified in classic studies as SP-1 sites (17,18). Second, the transcriptional activation of HSV-TK required the entire E2F-1 protein, including the N-terminal 89 amino acids. In contrast, numerous studies have demonstrated that the N-terminal 89 amino acids of E2F-1 was dispensable for transcriptional activation through consensus E2F sites (19,20). Third, we demonstrate that S phase entry is not sufficient for activation of HSV-TK by E2F-1. The activation through consensus E2F sites is strictly linked to the cell cycle. We demonstrate that the N-terminal 89 amino acids of E2F-1 are also dispensable for cell cycle progression and is consistent with the known requirement of the consensus E2F sites in dictating the S phase gene expression program. Taken together, the activation of HSV-TK by E2F-1 proceeds by a different mechanism directed in part through the Nterminal of E2F-1 and is uncoupled from the known cell cycle regulatory role. These observations provide further evidence for the importance of the N-terminal region of E2F-1 and significantly broaden the repertoire of promoters that can be regulated by E2F-1.

Materials
Plasmids and Constructs-The 4R-TK-CAT plasmid and TK-CAT (or p(18)-TK-CAT) plasmids were gifts of the late Dr. Harold Weintraub. Both the 4R-TK-CAT and TK-CAT contain the herpes simplex virus promoter sequence Ϫ105 to ϩ51 (18,49). The 4R mutant-TK-CAT was made by synthesis of the 4R sequence in which there were no E-box or consensus E2F sequences. The internal control for transfections was pGLC (SV-luciferase) and was purchased from Promega.
Cell Lines-C33A cells, a human cervical carcinoma cell line, and all other cell lines were obtained from American Type Culture Collection (ATCC). They were grown in the presence of 5% CO 2 and Dulbecco's modified Eagle's medium supplemented with 10% calf serum. The C2 myoblasts were a generous gift of Dr. Zach Hall. The C2 cells were maintained in the presence of 5% CO 2 and 20% fetal calf serum-Dulbecco's modified Eagle's medium.

Methods
Transcriptional Activation Assays-Transfections were performed by the BES method (21). Transfections were performed on C33A cells in 60-mm plates for 48 h. The cell extracts were assayed by both CAT ELISA (Boehringer Mannheim) and Luciferase (Promega) assay kits. A standard curve of 0 -800 pg of purified CAT protein and of 10 Ϫ8 to 10 Ϫ15 fmol of purified luciferase (U. S. Biochemical Corp.) was simultaneously assayed and was used to calculate all data. The final transfection data was expressed as a ratio of picograms of CAT/femtomoles of luciferase protein and was corrected for transfection efficiency.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA conditions were exactly as described elsewhere (16,22). The E2F probe was a single E2F site distal to the TATA element of the E2 promoter (gift of Dr. SriLata Bagchi (23). The TK probe encompassed Ϫ105 to ϩ21 of the TK promoter and was a Pst1 fragment. Competitor wild-type DNA consists of E2F sequences from the E2 promoter (TTTCGCGC) in an inverted repeat orientation. The E2F mutant competitor site consists of a C-to-A mutation in the fourth position of each E2F site (TTTAGCGC). This E2F site mutant fails to bind purified E2F (22,24).
DNase Footprinting-The TK promoter DNA was end-labeled by Klenow fill-in reaction at either the 5Ј or 3Ј end. The DNA-protein binding reaction was done in standard E2F EMSA conditions with 0 -150 ng of purified protein. Five units of DNase I were were placed in 150 l of a solution containing 25 mM CaCl 2 and 50 mM MgCl 2 . For the protein-DNA binding reactions, 1 l of diluted DNase solution (0.03 unit) was added to a 50-l binding reaction for 30 s at 20°C. For the no protein control, 0.5 l of the DNase I solution was incubated for 30 min at 20°C. The DNA was purified and resolved on a 6% sequencing gel. The position of each footprint was based on a Maxam-Gilbert sequencing ladder that was run in an adjacent lane.
S Phase Immunofluorescence Assay-The cell cycle assay was performed essentially as described by Nevins and colleagues (25). C2 cells were plated on 60-mm tissue culture plates containing several 12-mm circular glass coverslips (Fisherbrand). C2 myoblasts were transiently transfected at 40 -50% confluence in 60-mm tissue culture dishes by the Hepes-buffered saline-CaPO 4 method with glycerol shock. Cells were transfected with 3 g of ␤-galactosidase and either 1 g of Nflu-E2F-1, E2F mutant d1-88, or E2F mutant E132. The DNA concentration was brought up to 15 g with puc12 plasmid DNA. After 24 h, the transfected cells were pulsed for 1 h with medium containing 10 mM BrdUrd. The plates were then rinsed with 1 ϫ PBS three times. The coverslips were removed and fixed in a 3:7 methanol:acetone solution for 15 min at 0°C. The coverslips were allowed to air dry and then were stored in 1 ϫ PBS at 4°C.
For staining transfected cells, the coverslips were blocked in Western blocking solution (1 ϫ PBS, 0.01% Triton X-100, 1% bovine serum albumin, 0.1 M Tris, pH 7.5, 0.2% milk powder (Carnation), 0.02% azide) for 15-30 min at 37°C. Next, the coverslips were incubated in a 1:80 dilution of rabbit anti-␤-galactosidase for 1 h at 37°C. All dilutions of antibodies were done with 2% fetal calf serum-Dulbecco's modified Eagle's medium. The coverslips were rinsed in 1 ϫ PBS and then incubated in 1:500 dilution of goat anti-rabbit conjugated TRIC for 30 min at 37°C. The coverslips were rinsed in 1 ϫ PBS again and then fixed with 3:7 methanol:acetone for 15 min at 0°C. Afterward, the methanol:acetone was removed, and the coverslips were allowed to air dry.
For staining the BrdUrd-positive cells, the coverslips were treated with 2 N HCl for 1 h at 37°C, removed, and neutralized with sodium borate (pH 8.55) twice for 10 min each. The coverslip was washed with 1 ϫ PBS again and then incubated with a 1:50 dilution of mouse anti-BrdUrd for 1 h at 37°C. The coverslips were washed in 1 ϫ PBS and incubated with a 1:25 dilution of donkey anti-mouse conjugated fluorescein (FITC) for 30 min at 37°C. To visualize all nuclei in the field, the coverslips were stained with 2.5 g/ml Hoechst dye (Sigma B2883) for 2 min at room temperature.

E2F Transactivates the Herpes Simplex Virus Thymidine
Kinase Promoter in the Absence of Consensus E2F sites-During the course of experiments examining the interplay of muscle differentiation and E2F-1, we found that a muscle-specific reporter plasmid (4R-TK-CAT) was consistently and significantly transactivated by E2F-1 protein in transient transfections. 4R-TK-CAT contains four copies of a muscle-specific E box sequences (CAGGTG) fused to HSV-TK promotor. Upon closer inspection, the activation by E2F-1 actually proceeded through the HSV-TK portion of the reporter plasmid. However, inspection of the HSV-TK promoter revealed no consensus E2F sites (TTTSSCGC), suggesting a different mode of activation.
In these experiments, three HSV-TK derivative reporter constructs were used: 1) 4R-TK-CAT, 2) 4R mutant-TK-CAT in which the E-boxes were mutated, and 3) TK-CAT. The common denominator of all constructs are the Ϫ105 to ϩ51 sequences of the HSV-TK promoter. In all transfections of the non-muscle C33A cell line, SV-luciferase was used as a control to normalize for transfection efficiency in a given experiment. All transfections were done in duplicate with less than 10% variation.
A representative experiment is shown in Fig. 1. In the upper table in Fig. 1A, each HSV-TK derivative construct was activated 6 -9-fold in the presence of E2F-1, regardless of the E-box sequence. Extensive mutation of the E-box did not abolish the induction (4Rmut-TK-CAT), suggesting that the E2F-1 induction was independent of the upstream element. We have observed 5-7-fold activation by E2F-1 in three other TK derivative reporters with different, unrelated promoter elements (data not shown). To conclusively demonstrate that the E-box was not involved in the induction by E2F-1, the 4R and 4R mutant sequences were placed upstream of a minimal E1B TATA promoter. As shown in the bottom table of Fig. 1A, neither E1B or its derivative E-box constructs was significantly activated by E2F-1. As expected, E2F-1 expression gave a 4-fold activation of an E2F-E1B reporter construct.
The only difference between various TK derivative promoters is shown in the upper table in Fig. 1A. The 4R-TK-CAT containing the muscle specific E-box has a lower basal level than the parent TK reporter. Because E1B-luciferase has a lower basal level than TK-CAT, the E-box mediated repression was not detectable. This apparent repression of E-box sequences in non-muscle cell lines has been observed previously (e.g. see Ref. 26), but does not appreciably affect the observed activation through HSV-TK.
While Fig. 1A is a representative experiment, the observation that E2F-1 activates HSV-TK derivative plasmids has been repeated numerous times and each point is the average of duplicate transfections that varied less than 10%. With 4R-TK-CAT, the fold activation by E2F-1 was 9.5 Ϯ 1, and these data were derived from 18 sets of duplicate transfections. With TK-CAT, the fold activation was 8.4 Ϯ 2, and these data were derived from 11 sets of duplicate transfections. As shown in Fig. 1B, the activation of the 4R-TK-CAT and TK-CAT report-ers was also dependent on the concentration of input E2F-1 expression plasmids.
As shown in Fig. 1C, the activation by E2F-1 by HSV-TK was observed in numerous cell lines and was not limited to C33A cells, which is our test line. Furthermore, in control transfections, expression of irrelevant proteins (adenovirus E1A or polyoma large T) did not lead to activation of the TK-CAT or 4R-TK-CAT reporter constructs (data not shown). Thus, the activation of HSV-TK was specific for E2F-1 and is a general effect in numerous cell lines. Fig. 1 establish that E2F-1 can activate through the HSV-TK promoter from Ϫ105 to ϩ54. Inspection of the TK promoter sequence revealed no consensus E2F sites of the consensus sequence TTTSSCGC where S is C or G (1,22,24). We also independently sequenced the reporter constructs to confirm the absence of consensus E2F sites, and the entire sequence of this HSV-TK region is shown in Fig. 3. The HSV-TK promoter is one of the best characterized eukaryotic promoters. It was one of the first promoters subjected to extensive mutagenesis and its functional elements have been established both in vitro and in vivo. As summarized in Fig. 3, early work in Tjian and McKnight's laboratories established the presence of SP-1 and CTF sites as critical elements for the expression of HSV-TK by several different criteria both in vitro and in vivo (17,18).

E2F-1 Binds to GC-rich Regions in the HSV-TK Promoter-The experiments in
To address the mechanism of the activation in Fig. 1, we asked whether E2F-1 could nonetheless bind the HSV-TK promoter in the absence of consensus E2F sites and direct subsequent transcriptional activation. If so, the region of binding would define a new type of E2F-DNA interaction. For these binding experiment, purified, recombinant GST-E2F-1/DP-1 was used. E2F-1⅐DP1 is a heterodimeric complex which exhibits high affinity binding to DNA. The complex of GST⅐E2F-1 and DP-1 was co-expressed by baculovirus infection of Sf9 cells (generous gift of Min Xu and Helen Piwnica-Worms) and subsequently purified by glutathione-affinity chromatography.
As shown in Fig. 2, purified GST-E2F-1/DP1 bound efficiently to the HSV-TK promoter even at low protein concentrations in EMSA assays with labeled HSV-TK promoter (Fig. 2,  lane 4). Accounting for differences in the specifc activity of the two probes, E2F-1⅐DP1 shifted approximately the same percentage of the TK and E2F probes. Lanes 2 and 5 show competition with wild-type E2F sites. Lanes 3 and 6 show competition with a mutant E2F site that does not compete in E2F gel shifts. These competitions indicate that the E2F-1⅐DP1 complex is not binding nonspecifically to the TK promoter, i.e. any competitor DNA did not compete (Fig. 2, lane 6). E2F-1⅐DP1 bound to an E2F probe derived from the E2 promoter with the same competition profile, which is in agreement with our published data (9,22,24,27). E2F-1⅐DP1 appeared to bind as two complexes (two gel shifted bands) on the TK promoter, indicating multiple potential binding sites. Purified, recombinant GST-E2F-1 alone also bound to E2F and HSV-TK gel shift probes under these conditions, suggesting that DP1 did not affect the specificity for binding to both consensus and nonconsensus E2F sites (data not shown).
Because inspection of HSV-TK revealed no consensus E2F sites, a next step focused on the precise location of E2F-1/DP-1 binding in the HSV-TK promoter using DNase I footprinting.
Increasing amounts of GST-E2F-1/DP1 protein was incubated with end-labeled TK promoter and analyzed in Fig. 3. Both stands of the HSV-TK promoter were analyzed. Higher affinity sites would be protected at low protein concentration and lower affinity site would be protected at higher protein concentrations. For reference, the unprotected DNase ladder is shown in lane 1 and 8 for both strands. The regions of protection were precisely assigned based on Maxam-Gilbert sequencing ladders (Fig. 3, lanes 6, 7, 13, and 14).
As shown in Fig. 3, the binding of E2F-1/DP1 gave three areas of protection on HSV-TK and were labeled regions 1, 2, and 3 based upon proximity from the 5Ј end of the promoter.  1. E2F transactivates the HSV-TK promoter. A, upper table, C33A cells were transiently transfected in 100 mM plates with 5 g of the indicated reporter and 3 g of pCMV-E2F-1 as described under "Experimental Procedures." The pGLC plasmid containing the SV-luciferase gene served as the transfection efficiency control. The CAT protein was quantitated by CAT ELISA assays, and luciferase activity was quantitated by a standard assay. The results were calculated using CAT and luciferase standard curves, and a normalized ratio is shown. The values represent averaged duplicates of individual transfections, which varied less than 10%. Bottom table, same as the top table, except the reporters were all derivatives of a minimal E1B luciferase. The transfection efficiency was normalized with RSV-CAT. E2F-E1B-luciferase was constructed with the E2100 fragment of the E2 promoter and contains 2 E2F sites (22). Quantitation was exactly as described as under "Experimental Procedures." B, same as top table in A, except that the input E2F-1 expression plasmid was varied. Each point is the average of a duplicate transfection. C, same conditions as top table in A, except that the indicated cell lines were used.
Based on protection at increasing GST E2F-1/DP1 protein levels, region 2 represents the highest affinity site and regions 1 and 3 are the weaker sites. The region 2 footprint was evident at only 30 ng of purified protein and corresponded to Ϫ54 to Ϫ39 on the HSV-TK promoter. The footprint defined as region 1 mapped to Ϫ105 to Ϫ95. The region 3 footprint mapped to ϩ21 to ϩ38, but was evident only at the highest concentration of E2F-1/DP1 protein and represented the weakest site.
Thus, GST-E2F-1/DP1 selectively bound only to GC-rich elements of HSV-TK (see Fig. 3B). Intriguingly, the region 2 and 1 E2F-1 footprints were clearly established in classic transcription studies as Sp-1 sites (17). Thus, these protected regions could be defined in two ways. Either they represent a new class of E2F site or they represent E2F interaction with established Sp-1 sites. For simplicity of discussion, we will refer to these protected regions as GC-rich sites.
Activation of HSV-TK Requires the Entire E2F-1 Protein, Including the N-terminal Region-There are three possible mechanisms for the activation of E2F-1 through the GC rich regions of HSV-TK. The first is that E2F-1 binds directly to HSV-TK and directs the transcriptional activation. This mechanism is supported by the direct binding experiments in Figs. 2 and 3. However, there are two other possibilities in which HSV-TK activation is an indirect consequence of standard E2F induction. A second possibility is that E2F-1 is inducing the synthesis of the actual factor that induces transcription

FIG. 2. GST-E2F-1/DP1 can bind to the HSV-TK promoter.
EMSA were performed with 30 ng of purified baculoviral expressed GST-E2F-1/DP1. The probes used were the 32 P-labeled single distal E2F site from the E2 promoter and a Pst1-Pst1 fragment from the HSV-TK promoter encoding part of the polylinker and Ϫ105 to Ϫ10 of the TK promoter. The TK probe has a lower specific activity than the E2F probe, accounting for the difference in signal intensity. The competitors were in 100-fold excess over the probe. The competitors were wild-type E2F sites (TTTCGCGC; lanes 2 and 5) and mutant E2F (TTTAGCGC; lanes 3 and 6).  1 and 8, lanes 2 and 9, 3 and 10, 4 and 11, 5 and 12, respectively) to bind to end-labeled HSV-TK promoter. The promoter fragment contained the entire HSV-TK promoter region (Ϫ105 to ϩ51) and was excised from the TK-CAT plasmid with XhoI and HindIII. DNase I footprinting was carried out as described under "Experimental Procedures." Lanes 1 and 8 are negative controls with no protein addition for comparison. Guanine and purine specific markers are labeled as indicated (lanes 6 and 13, 7 and 14, respectively). through the GC-rich regions. It should be noted that numerous studies have demonstrated that E2F activation through consensus sites does not require the N-terminal 89 amino acids of E2F-1. A third possibility is that the HSV-TK activation is a response to an enhanced cell cycle, which is an established property of E2F-1 expression and activation of the S phase gene expression program (25). The requirement for the E2F-1 Nterminal region in activation of the cell cycle has not been previously demonstrated.
To address these possibilities, we next compared the genetic requirements in E2F-1 for activation through the GC-rich regions and through the consensus E2F sites. As will be shown in the next two figures, the mechanism of activation through the GC-rich regions of HSV-TK differ from the standard E2F-1 transcriptional activation and has a strict requirement for the N-terminal 89 amino acids of E2F-1. An E2F-1 mutant that lacks the first 89 amino acids (d1-88) is functional in the transactivation of standard E2F sites, but is defective in activation of HSV-TK.
Several studies have defined the regions of E2F-1 that are necessary for activation through its usual sites (19, 20, 28 -30). A full series of E2F-1 mutants were the generous gift of Joe Nevins (20) and Bill Kaelin (15).The functional regions are diagrammed schematically in Fig. 6 and have been exhaustively characterized for activation on standard E2F sites by Nevins and co-workers. The DNA binding region is located at amino acids 120 -191 and contains a putative basic and helixloop-helix region (amino acids 117-128). A leucine-rich region at amino acids 191-284 is the region of heterodimerization to DP-1. The transcriptional activation is encoded in the C terminus from amino acids 368 -437 with the RB interaction region at amino acids 409 -437. Lastly, the cyclin A/CDK2 binding site in the N-terminal region is located in the first 89 amino acids, specifically from amino acids 67-108 (15,16)  To compare the activation on HSV-TK, we used two TK reporter derivatives (TK-CAT and 4R-TK-CAT). To allow a direct comparison of activation on standard E2F sites and on the GC-rich sites, we also repeated the published analysis on E2F sites and this was in excellent agreement with the results published by Nevins and co-workers (20). The activation of each reporter construct was approximately 6 -9-fold with wildtype E2F-1 and was set to 100% for comparison.
As shown in Fig. 4, the activation of each TK reporter construct by E2F-1 required the entire E2F-1 protein. Destruction of several known functional motif impaired activation of HSV-TK. The E132 point mutant is defective in DNA binding and activation of HSV-TK. Together with the results in Figs. 2 and 3, this data indicates that the DNA binding ability of E2F-1 is necessary for HSV-TK activation. FS409 is defective in activation on both E2F and GC-rich sites, suggesting a requirement for the transcriptional activation region.
While there was no consistent requirement for the marked box region (e.g. d284 -358), the DP-1 binding site (d206 -220) did not appear to be required for the activation of HSV-TK in both reporters. The data for activation on standard E2F sites on the leucine-rich, DP-1 region (d206 -220) and for the marked box (d284 -358) was consistent with published work (e.g. see Ref. 20). The function of the marked box region is currently unknown. However, for HSV-TK activation, heterodimeric interaction with DP-1 may not be essential and represents another difference from activation of consensus E2F sites.
However, the activation of HSV-TK required the N-terminal 88 amino acids of E2F-1 (d1-88). In contrast, the N-terminal 89 amino acids were dispensable for activation through standard E2F sites. As shown in Fig. 4B, the expression of wild-type and d1-88 E2F-1 are equivalent, suggesting that the functional difference is not due to differing protein levels. Furthermore, this difference was observed on both TK derivative constructs. Furthermore, deletion of the cyclin A/CDK2 site (⌬24; deletion of amino acids 67-108) did not affect the activation of HSV-TK, suggesting that there is no overlap. Deletion of amino acids 113-120 in E2F-1 also impaired HSV-TK activation, and this mutant is functional for standard E2F-dependent activation. By comparison, E2F-4, which has a divergent N terminus, does not activate HSV-TK (data not shown). Thus, the strict dependence on the N-terminal region of E2F-1 represented the major difference between activation through standard E2F and the GC-rich sites. Cell Cycle Progression Is Not Sufficient for HSV-TK Activation by E2F-1-Studies by Nevins and co-workers have demonstrated that overexpression of E2F-1 can drive quiescent cells into S phase and activates numerous genes important for FIG. 4. E2F deletion 1-88 is defective in transactivation of the TK promoter. A, C33A cells were transfected with 2.5 g of the indicated reporter and 1.5 g of E2F-1 or deletion mutant. SV-luciferase (pGLC) was cotransfected as a transfection efficiency control. The transfection were done in duplicate. Data were calculated and normalized as previously stated (Fig. 1). B, Western blots of the transfected proteins from A transfections were normalized by protein concentration and loaded onto a 7.5% gel (for the 4R blot) or a 10% SDS-PAGE gel (for the E2F and TK blots). The blots were probed at 1:5000 dilution with rabbit polyclonal antiserum made against GST-E2F-1 and visualized by ECL (9). cell cycle progression (25,31). This observation is consistent with the known role of E2F-1 in the activation of numerous genes that are required for progression to S phase. A mutant E2F-1 that was defective in DNA binding (E132) failed to elicit the enhanced S phase entry and is consistent with the notion that transcriptional activation of E2F-dependent genes was required for the S phase phenotype (25).
The results in Fig. 4 indicated that the N-terminal region and other functional elements of the E2F-1 proteins are all required for activation through GC-rich regions. In contrast, Nevins and others (e.g. see Ref. 20) had demonstrated that the N-terminal region was dispensable for activation through consensus E2F sites. The additional requirement for the N-terminal region in the activation through GC-rich regions suggested that the mechanism differs from activation through the consensus E2F sites. We asked whether the N-terminal region was also required for enhanced S phase entry. If expression of the N-terminal mutant (d1-88) did elicit the S phase phenotype, this would indicate that this region was dispensable for S phase entry, but was required for activation through GC-rich elements. This result would provide further evidence that activation through GC-rich and consensus E2F elements differ and that activation through GC-rich elements may not be strictly dependent on the cell cycle.
To assess the cell cycle phenotype of the N-terminal mutant (E2F-1 (d1-88)), we modified the cell cycle assay originally developed by Nevins and co-workers. Asynchronously growing C2 cells were transiently co-transfected with the indicated mutants of E2F-1 and ␤-galactosidase, which served as a marker for the transfected cells. After 24 h, S phase was visualized by a 1-h pulse of BrdUrd, which should "capture" the cells in S phase. Double immunofluorescence was then used to visualize transfected cells that were in S phase (␤-galactosidase-positive and BrdUrd-positive). Induction of S phase was quantitated as a ratio of BrdUrd-positive/␤-galactosidase-positive cells. Wild-type E2F-1 was used as a positive control and DNA-binding defective mutant (E132) was used as a negative control. C2 cells were chosen because these cells grow and bind well to coverslips used in immunofluorescence assays.
As shown in Fig. 5, the N-terminal mutant (E2F-1 (d1-88)) resembled wild-type E2F-1 in the ability to induce S phase. Expression of full-length E2F-1 gave approximately 50% S phase cells; expression of d1-88 gave 47% S phase cells. In contrast, untransfected cells or expression of a DNA-binding defective E2F-1 gave 28 and 29% S phase cells, respectively. Because the N-terminal mutant d1-88 could induce S phase and activate E2F-dependent genes, these independent assays indicate that a fully functional protein for activation of E2F elements was produced. However, while functional on consensus elements, this N-terminal deletion was defective for activation of GC-rich regions, suggesting that cell cycle progression was not sufficient for the activation. DISCUSSION Because of its expression in numerous cell types, HSV-TK has been used as a model promoter for cellular transcription studies and remains one of the best characterized eukaryotic promoters. The classic studies of Tjian and McKnight established that multiple promoter elements were necessary for its basal expression (17,18). Their work established the GC-rich regions of HSV-TK as SP-1 sites by numerous functional criteria, in vivo transfection assays, in vitro transcription assays, and DNase footprinting with purified SP-1. McKnight and colleagues further demonstrated that mutation of the SP-1 sites partially impaired the function of HSV-TK promoter in transient assays. These studies indicated that multiple elements are required for transcription of HSV-TK. It is interesting to note that the Ϫ105 to 54 region of HSV-TK has served as a "minimal" promoter for numerous reporter constructs, including those for E2F.
In this study, we report a new mode of E2F-1 activation utilizing the N-terminal regulatory region and nonconsensus DNA sites. Expression of E2F-1 leads to efficient activation of the HSV-TK promoter and a requirement for DNA binding is supported by two observations. We demonstrate that E2F-1 can directly interact with the GC-rich, apparent SP-1 elements in the context of the well established HSV-TK model promoter. Each site differs from the consensus E2F sites extensively outlined in several cell cycle-regulated promoters (for recent review, see Ref. 1). Furthermore, an E2F-1 mutant that was defective in DNA binding fails to activate HSV-TK. The activation of HSV-TK requires the entire E2F-1 protein, including the N-terminal 89 amino acids. In contrast, transcriptional activation through the consensus E2F sites and the subsequent S phase entry do not require the N-terminal 89 amino acids of E2F-1. Taken together, these observations suggest that E2F-1 activation of HSV-TK proceeds through a different mechanism than generic E2F transcriptional activation.
This and previous studies highlight the importance of the N-terminal region as a regulatory region for E2F-1 and suggest at least two distinct functions. We have shown previously that the N-terminal region can dictate a regulatory interaction with cyclin A/CDK2 (16). The data in the current study indicated that the N-terminal 89 amino acids were essential for activation of HSV-TK, presumably through binding of the GC-rich elements. These two regulatory modes appear to be distinct, since an E2F-1 mutant that cannot bind cyclin A/CDK2 (⌬24) (15) can still activate HSV-TK. E2F-2 and E2F-3 also support activation of HSV-TK, but E2F-4 fails to activate HSV-TK (data not shown). E2F-4 contains a divergent N-terminal region and lacks the cyclin A/CDK2 binding site that is present in E2F-1, -2, and-3 (for review, see Ref. 2). Finer mutagenesis of the N-terminal region E2F-1 will be required to conclusively dissect the various regulatory regions.
Because the N-terminal 89 amino acids are required and the cyclin A/CDK2 binding site is dispensable, the transcriptional activation of HSV-TK may only be loosely linked to the cell cycle. The entry into S phase is certainly not sufficient for activation of HSV-TK by E2F-1. This is supported by the observation that an E2F-1 mutant lacking the N-terminal 89 amino acids can still induce S phase, but cannot support acti- 5. E2F mutant d1-88 can trigger entry into S phase. C2 cells attached to coverslips were co-transfected with a PRC CMV-E2F-1 expression vectors and RSV-␤-galactosidase as described under "Experimental Procedures." Twenty-four hours after transfection, the cells were labeled for 1 h with 10 M BrdUrd. Transfected cells were visualized by immunofluorescence for ␤-galactosidase (anti-␤-galactosidase antisera (5Ј 3 3Ј Corp.); TRIC-anti-rabbit IgG (Jackson)) and for Br-dUrd incorporation (anti-BrdUrd (Boehringer Mannheim); FITC-anti mouse IgG (Jackson)). The percentage of cells in S phase is calculated as the number of cells in the total population of ␤-galactosidase-positive cells that are also BrdUrd-positive. vation of HSV-TK transcription. The precise relationship of GC-rich activation and the cell cycle is difficult to address, since impairment of E2F-1 functional domains will affect both HSV-TK activation and cell cycle. While the role of E2Fs in G 1 /S regulation is well established, several observations hint at an even broader role in cellular regulation. We and others have demonstrated the presence of both E2F-1 protein in a postmitotic, nondividing cell (9). Furthermore, the developmental profile of E2F-1 mRNA revealed a peak of expression relatively late in murine embryogenesis and in tissues that are not significantly dividing. Thus, it is possible that E2F-1 may have a function apart from its well-established role in G 1 /S control.
The probable mechanism for E2F-1 activation of HSV-TK is through direct binding of the GC-rich regions in HSV-TK. This is supported by the footprinting experiments and the requirement for a functional DNA binding region in E2F-1. Conclusive proof of a direct E2F-1 mechanism requires the systematic mutation of each GC-rich site in the HSV-TK promoter and a demonstration that E2F-1 mediated activation is abolished. A possible difficulty is the previous demonstration that the SP-1 sites were required for basal activity of HSV-TK (17,18,32), so impairment of basal activity may obscure effects on E2F-1dependent activated transcription. Regardless, this represents an important future study. In addition, these observations argue against an indirect mechanism in which E2F-1 is triggering the synthesis of a factor such as SP-1, which is known to activate through the GC-rich elements (e.g. see Refs. [33][34][35]. A formal possibility is that synthesis of an SP-1-like factor also requires the N-terminal region of E2F-1, but there is no evidence for this mechanism. Numerous studies with DNA tumor viruses have demonstrated interaction of viral and cellular transcription factors. The E2F protein itself is targeted by numerous DNA tumor viral regulatory proteins (36). Early work by Nevins and colleagues indicated that herpesvirus infection can stimulate the adenoviral E2 promoter, likely through activation of cellular E2F activity (37,38). Studies with herpesvirus demonstrated conclusively that these GC-rich, SP-1 sites were necessary for basal expression, but the mechanisms of viral activation of HSV-TK are complex (32,39). The existence of nonconsensus E2F sites may be a recurrent theme in viral promoters and has been demonstrated in the context of the Epstein-Barr virus (EBV) function (40). While both studies utilized E2F-1 protein, the sites in the current study do differ from the EBV promoter sites. In the EBV study, no genetic analysis of E2F-1 function on the target EBV promoter was reported, so it is difficult to difficult to compare the present study and their work. Clearly, more work is required to understand the role of E2F in the function of large DNA virus, such as EBV and HSV.
Nonetheless, this study expands the repertoire of promoters that can be targeted by E2F-1. The GC-rich SP-1 sites are in numerous promoters that are not restricted to cell cycle regulation. Most previous studies on E2F have only focused on its role in the cell cycle; SP-1 and E2F sites do co-exist in many promoters, such as DHFR (41,42). It is interesting to note that RB family members can regulate both E2F and SP family members (43)(44)(45)(46). The current study raises an interesting interplay of E2F and SP family members in controlling transcription. An attractive model would be that E2F-1 recruits SP-1 to the GC-rich elements or vice versa. Any putative functional interactions of E2F-1 and SP-1 should occur through the Nterminal region of E2F-1. This event would then trigger assembly of a complicated promoter complex on the HSV-TK promoter and lead to subsequent transcriptional activation. Previous work has demonstrated that multiple elements and factors are required to activate this short segment of HSV-TK promoters in cells, and elements other than SP-1 sites appear to be required (17). More work is required to establish the role of E2F-1 and SP-1 in triggering the overall transcriptional activation of HSV-TK. The current study suggests a potentially interesting and complex interaction of the SP-1 and E2F transcription factor families.