Identification of an essential cis-element near the transcription start site for transcriptional activation of the proliferating cell nuclear antigen gene.

Interleukin 2 (IL-2) stimulates T lymphocyte proliferation and induces the expression of proliferating cell nuclear antigen (PCNA), a processivity factor for DNA polymerase . Previously, deletion analysis suggested cis-element(s) in the proximal region of the PCNA promoter (−40 to +143) are required for IL-2 induction in cloned T lymphocytes. The sequence 5′-TTGCGGGC-3′ located at +10 to +17 is similar to the E2F consensus binding site and is required for optimal PCNA promoter activity. In IL-2-stimulated T cells, nuclear proteins are induced to bind to this sequence as demonstrated using electrophoretic mobility shift assay (EMSA), competition EMSA, and methylation interference analysis. A 180-kDa polypeptide was detected by UV cross-linking to bind specifically to the PCNA E2F-like sequence. Our data indicate that the protein bound to the PCNA E2F-like site is not one of the transcription factor E2F proteins. Our results demonstrate that the E2F-like sequence and the protein(s) binding to it are required for optimal PCNA promoter activity and IL-2 induction of PCNA expression.

Interleukin 2 (IL-2) stimulates T lymphocyte proliferation and induces the expression of proliferating cell nuclear antigen (PCNA), a processivity factor for DNA polymerase ␦. Previously, deletion analysis suggested cis-element(s) in the proximal region of the PCNA promoter (؊40 to ؉143) are required for IL-2 induction in cloned T lymphocytes. The sequence 5-TTGCGGGC-3 located at ؉10 to ؉17 is similar to the E2F consensus binding site and is required for optimal PCNA promoter activity. In IL-2-stimulated T cells, nuclear proteins are induced to bind to this sequence as demonstrated using electrophoretic mobility shift assay (EMSA), competition EMSA, and methylation interference analysis. A 180-kDa polypeptide was detected by UV cross-linking to bind specifically to the PCNA E2F-like sequence. Our data indicate that the protein bound to the PCNA E2Flike site is not one of the transcription factor E2F proteins. Our results demonstrate that the E2F-like sequence and the protein(s) binding to it are required for optimal PCNA promoter activity and IL-2 induction of PCNA expression.
PCNA was first described by Miyachi et al. (1978) as a nuclear antigen restricted to proliferating cells that reacts with sera from some patients with the autoimmune disorder systemic lupus erythematosus, hence the name proliferating cell nuclear antigen. PCNA 1 is an auxiliary factor for DNA polymerase ␦ (Bravo et al., 1987;Prelich and Stillman, 1988). It is required for both DNA replication and DNA repair (Shivji et al., 1992). The gene for PCNA has been highly conserved throughout the course of evolution (Mathews, 1989). The ability of PCNA-specific antibodies to inhibit inducible DNA synthesis in isolated nuclei (Wong et al., 1987), and of PCNA antisense oligonucleotides to inhibit proliferation of Balb/c 3T3 cells (Jaskulski et al., 1988), supports the importance of this protein in cell cycle progression. However, to date all of the cellular functions of PCNA during the cell cycle have not been elucidated. In addition to its role as a processivity factor in DNA replication, PCNA may function in the regulation of cell cycle progression (McAlear et al., 1994;Xiong et al., 1993). Recently, PCNA has been found in complexes with a variety of cyclins, cyclin-dependent kinases, and p21, one of the inhibitors of cyclin-dependent kinases, also known as Waf1 or Cip1 (Xiong et al., 1992;Waga et al., 1994). Also p21 controls DNA replication by direct interaction with PCNA (Waga et al., 1994;Flores-Rozas et al., 1994). Therefore it is important to understand the regulation of PCNA expression and the function(s) of PCNA within the cell cycle.
Previous studies have shown transcriptional regulation of PCNA (Morris and Mathews, 1990;Shipman-Appasamy et al., 1990). The promoter region and the introns of the PCNA gene are involved in the transcriptional regulation of PCNA (Alder et al., 1992;Huang et al., 1994). Post-transcriptional regulation also seems to play a role in its regulation. During growth factor stimulation of lymphocytes or 3T3 cells, there is an increase in PCNA mRNA stability (Chang et al., 1990;Shipman-Appasamy et al., 1990;Baserga, 1991). Increased PCNA transcription has been demonstrated in IL-2-stimulated T cells (Shipman-Appasamy et al., 1990), but it could not be found in serum-stimulated 3T3 cells (Baserga, 1991). This suggests that lymphocytes and fibroblasts use different mechanisms for regulating levels of PCNA mRNA. We have shown transcriptional regulation of PCNA during IL-2 induced proliferation of T lymphocytes. Tandem CRE binding sites in the murine PCNA promoter are required for optimal promoter activity and the IL-2-induced PCNA transcriptional activation Orten et al., 1994;Feuerstein et al., 1995). Morris and Matthews (1991) have shown that the corresponding CREbinding site in the human PCNA gene is required for adenovirus E1A-driven transactivation.
Proteins and enzymes involved in DNA synthesis are coordinately regulated with growth in mammalian cells: they are induced in late G 1 to early S phase of the cell cycle. These include PCNA, dihydrofolate reductase (DHFR), thymidine kinase, and DNA polymerase ␣ among others (Hofbauer and Denhardt, 1991). The molecular basis of this regulation has yet to be established. Promoter sequences known for some of these genes differ markedly. Recent studies of these promoters have shown that the binding site for transcription factor E2F is important for DHFR and thymidine kinase expression (Blake and Azizkhan, 1989;Dou et al., 1994). Similar motifs are known to be present in the promoter of other DNA synthesis genes, namely, those of DNA polymerase ␣ and possibly thymidine synthase (Pearson et al., 1991). The presence of an E2F binding site in promoters of several genes encoding DNA synthesis proteins suggests that growth regulation of these proteins may take place by a common mechanism (Nevins, 1992). This is similar to the regulation of the DNA synthesis genes in the yeast Saccharomyces cerevisiae, in which a regulatory element, identical in sequence with the recognition site for restriction enzyme MluI (ACGCGT), plays a central role (Lowndes et * This work was supported in part by American Cancer Society Grant DB-52B. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
al., 1991). One hypothesis for the mechanism of coordinate regulation of DNA synthesis genes in mammalian cells is that the transcription factor E2F plays a central role in activating DNA synthesis genes during growth stimulation (Nevins, 1992). E2F represents a group of related transcription factors with similar DNA-binding specificity that interacts with pRBrelated pocket proteins (La Thangue, 1994). Sequence comparison indicates that an E2F-like binding site is present between ϩ10 and ϩ17 of the murine PCNA promoter. In addition to the tandem CRE binding site, our previous results suggested that the proximal PCNA promoter containing the E2F-like site is critical for IL-2-induced transcriptional activation . Thus, we analyzed the requirement of this E2F-like site for optimal promoter activity and for protein-DNA interactions in the PCNA promoter proximal region and demonstrate a functional role for the E2F-like site in PCNA transcriptional regulation.

MATERIALS AND METHODS
Plasmids-A 1.8-kilobase PCNA genomic sequence was isolated previously from a Balb/c liver genomic library and subcloned into the pBluescript vector (Shipman-Appasamy et al., 1990). The pPCNA5-LUC plasmids were constructed by inserting polymerase chain reaction amplified DNA fragments containing the PCNA promoter (Ϫ182 to ϩ143) into a HindIII site of the promoterless pSVOA vector (De Wet et al., 1987;Huang et al., 1994).
Cells-The cloned murine T helper cell line, L2, was maintained as described previously (Prystowsky, 1989;Huang et al., 1994). Eight days following exposure to irradiated allogenic spleen cells and 10 -20 units/ml IL-2, L2 cells were purified by Ficoll-Hypaque density gradient centrifugation. At this time, the cells are in a G 1 -resting state (Sabath et al., 1986). To insure that the cells were in a resting state, L2 cells were cultured for 24 h in Dulbecco's modified Eagle's medium with additives without IL-2. After 24 h, 1.0 ϫ 10 7 L2 cells in 10 ml of medium were stimulated with 100 -150 units/ml of IL-2. Human recombinant IL-2 was a gift of Chiron (Emeryville, CA) (Rosenberg et al., 1984). This substance was 98% pure and contained 0.01 ng of endotoxin/3.6 ϫ 10 6 units of IL-2; specific activity was evaluated by the manufacturer.
Freshly prepared splenocytes were from CBA/J mice, separated from erythrocytes using density gradient centrifugation through Ficoll-Hypaque (1.090 g/ml). Splenocytes at a concentration of 6 ϫ 10 6 /ml in Dulbecco's modified Eagle's medium with additives containing 500 ng/ml ConA were incubated for 72 h prior to extraction of proteins.
DNA Transfection and Luciferase Assay-PCNA-LUC constructs were transfected transiently into L2 cells by the DEAE-dextran method as described previously (Ausubel et al., 1992). To increase the transfection efficiency, L2 cells were stimulated with 100 units/ml IL-2 for 20 h prior to transfection. Then L2 cells were transfected using the DEAEdextran method . Briefly, resting cells were washed with Tris-buffered saline twice, pelleted, and resuspended in 600 l of a solution containing equal parts of DNA and DEAE-dextran (1 mg/ml) and incubated at 37°C for 30 min. The cells were washed twice with Tris-buffered saline and incubated at 37°C with 150 units/ml IL-2 until harvesting at 53 h post-transfection, which was optimal for IL-2-induced luciferase expression in L2 cells. Cells were harvested, washed with Ca 2ϩ ,Mg 2ϩ -free phosphate-buffered saline and lysed by freezethawing. Luciferase activity was assayed as described previously using a Berthold Lumat LB 9501 luminometer and expressed in relative light units (Ausubel et al., 1992). Quantitation of transfected luciferase DNA by slot blotting was used to correct for variation in transfection efficiency (Maniatis et al., 1982).
Preparation of Nuclear Extracts-Cellular extracts were prepared from induced and non-induced cells as described previously (Dignam et al., 1983;Shreiber et al., 1989). 0.5-1 ϫ 10 6 cells were collected, washed with 10 ml of phosphate-buffered saline, and pelleted by centrifugation at 1500 ϫ g for 5 min. The pellet was resuspended and washed in 1 ml of Tris-buffered saline. Then the cell pellet was resuspended in 400 l of cold buffer A (10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) by gentle pipetting. The cells were allowed to swell on ice for 15 min, after which 25 l of a 10% solution of Nonidet P-40 was added and the tube was vigorously vortexed for 10 s. The homogenate was centrifuged for 30 s in a microcentrifuge. The nuclear pellet was resuspended in 50 l of ice-cold buffer C (20 mM HEPES pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) and the tube was vigorously rocked at 4°C for 15 min on a shaking platform. The nuclear extract was centrifuged for 5 min in a microcentrifuge at maximum speed at 4°C and the supernatant was frozen in aliquots at Ϫ70°C. Protein concentration was quantitated using the Bradford reagent (Bio-Rad).
Electrophoretic Mobility Shift Assay (EMSA)-Electrophoretic mobility shift assay was performed as described previously (Garner et al., 1981). Briefly, 5,000 -15,000 cpm of 32 P-labeled duplex oligonucleotide probe ( Fig. 1) and 3-5 g of cell extract were mixed at room temperature for 15 min in 20 l of binding buffer containing 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, and 5% glycerol. Either poly(dI-dC) or salmon sperm DNA was used as nonspecific competitor DNA. The binding complexes were separated by 4.5% polyacrylamide gel electrophoresis (PAGE) at 4°C in buffer containing 25 mM Tris, 190 mM glycine, 1 mM EDTA at 25 mA. Gels were then dried and autoradiographed. Competition experiments were performed as described previously .
Sequence of oligonucleotides (sense strand) used for EMSA are listed below with numbers indicating the positions of start and end bases of the oligonucleotides within the PCNA promoter. The binding sites for PEA3, Inr, and E2F are underlined. Mutated nucleotides are indicated as lower case letters. ST, 5Ј-TAGGAAGCCGCGGCATTAGACGGTT- Polymerase Chain Reaction Primer-mediated Mutagenesis-Site-specific mutagenesis at PCNA E2F-like site was accomplished as described (Prelich, 1993) using a two-step polymerase chain reaction method. This method produced a fragment (nt Ϫ182 to ϩ143) of the PCNA promoter (pPCNA5) with the desired mutations. HindIII sites flanked this fragment, allowing cloning into the pSVOA vector. The DNA sequence of all mutated PCNA promoters was determined to insure that the desired product was prepared and to confirm that the subcloned DNA fragment is in the correct orientation in the vector.
Methylation Interference-The methylation interference assay was performed essentially as described previously . Partially methylated PCNA ST probes were 32 P-labeled at 5Ј-ends and incubated with L2 cell nuclear extract and poly(dI-dC). After gel electrophoresis, free DNA and protein bound DNA were cut and electrophoretically transferred to a DEAE membrane (Schleicher and Schuell) and eluted according to the supplier's protocol. After cleavage with piperidine, the various DNA fractions were analyzed by urea-PAGE (15%) electrophoresis and autoradiography (Ausubel et al., 1992). Maxam and Gilbert DNA sequencing reactions were performed as described previously (Ausubel et al., 1992) and G, G ϩ A, T ϩ C, and C reactions were carried using a commercial Maxam and Gilbert DNA sequencing kit (Du Pont, Westwood, MA) according to the manufacturer's instructions.
UV Cross-linking-The procedures for UV cross-linking were essentially as described by Chodosh et al. (1986). PCNA ST was labeled with both [␣-32 P]dCTP and bromodeoxyuridine triphosphate (Sigma). The binding reactions were scaled up 5-fold and 90 g of nuclear extracts were used for each reaction. Initial studies showed maximal binding by 30 min. The reactions were irradiated, the products were mixed with SDS-PAGE gel loading buffer, and separated on a 10% SDS-polyacrylamide gel. The specificity of the photoaffinity labeling reaction was established by including 500 g of either a nonspecific oligonucleotide or PCNA ST. Only the latter oligonucleotide should eliminate specific labeling of PCNA-specific binding. We used an oligonucleotide containing a CCAAT box motif derived from the PCNA promoter (Ϫ99 to Ϫ70) as a nonspecific competitor (5Ј-CCACGGGTACGATTGGTCCTTGA-GGAGAGG-3Ј).

Fine Mapping of Cis-Acting Elements in the PCNA Promoter
Proximal Region-Sequence analysis of the region around the PCNA transcription start site (ϩ1) suggests the presence of putative transcription factor binding sites, including a PEA3 site (5Ј-AGGAAGC-3Ј) between Ϫ14 and Ϫ8, a sequence sharing homology with the TdT (Smale and Baltimore, 1989) initiator element (Inr) around ϩ1 (CATT), and an E2F-like site (5Ј-TTGCGGGC-3Ј) between ϩ10 and ϩ17 (Fig. 1).
To determine the complexity of the protein(s) interacting with these sequences, a 36-base pair double-stranded oligonucleotide (ST) spanning all three sites (PCNA Ϫ14 to ϩ22) was radiolabeled and used in an EMSA (Fig. 1). A series of DNAprotein binding complexes were formed when this probe was incubated with nuclear extracts from the cloned T lymphocyte L2 and mouse splenocytes (Fig. 2). Notably, IL-2 and ConA induce the formation of several complexes with ST in stimulated L2 cells (Fig. 2B, lanes 1 and 2) and in splenocytes ( Fig.  2A), respectively. In contrast to PCNA ST, when the same extracts from L2 cells were used with a probe containing a USF binding motif (Gregor et al., 1990;Huang et al., 1994), USF binding activity did not change significantly during IL-2 stimulation (Fig. 2B, lanes 3 and 4). The binding pattern of these induced complexes is very similar in IL-2 stimulated L2 cells and ConA-stimulated splenocytes ( Fig. 2A). Initial competition experiments using unlabeled ST as well as a non-related oligonucleotide as competitors of complex formation indicated that complexes I, II, and III are specific for PCNA ST (data not shown). Among them, complex I and II were consistently present in all nuclear extracts we prepared but complex III binding activity varied with different nuclear extract preparations. In addition, three fast migrating bands (Fig. 2B) are not detected reproducibly in other experiments.
To determine where the various IL-2-inducible nuclear proteins bind to the PCNA promoter, a series of short oligonucleotides ( Fig. 1) each containing only one of the three putative protein binding sites was made: ST-1 (PEA3), ST-2 (Inr), and ST-3 (E2F). Complex formation with radiolabeled ST-1, ST-2, ST-3, and ST were compared by EMSA using a 24-h IL-2stimulated L2 cell nuclear extract (Fig. 3). Results in Fig. 3A demonstrated that only ST-3 retains the ability to form all three complexes (complexs I, II, and III). The fast migrating band present in this analysis was not found reproducibly in other experiments.
To determine whether the putative binding sites are required for complex formation, a series of binding site mutations (see "Materials and Methods") was made in these oligonucleotides based on previously defined consensus sequences of PEA, Inr, and E2F sites and on mutation studies (Yoo et al., 1991;Smale and Baltimore, 1989;Mudryj et al., 1990). Fig. 3B shows a competition binding experiment in which ST was 32 P-labeled and incubated with IL-2-stimulated L2 cell extract. Excess amounts of unlabeled oligonucleotides containing the non-mutated or mutated binding site were used as competitors of complex formation. Among them, ST-3 was the most efficient competitor. In contrast, ST-3m1, an oligonucleotide with a mutation in the E2F-like site (see Fig. 3B) failed to inhibit any of the binding activities. This result indicates that the E2F-like site is essential for complex formation.
It is interesting that both ST-1 and ST-2 show some dose-de- pendent inhibition of the binding activities (Fig. 3B). Since there is no obvious common sequence among ST-1, ST-2, and ST-3, this result suggests that the binding protein(s) may have multiple contact sites along the PCNA promoter sequence. It is more striking that ST-2m which has a mutated Inr element failed to inhibit complex formation, suggesting that the protein complexes may also contact the Inr element. To reconcile these data with Fig. 3A where there was no detectable binding to radiolabeled ST-1 and ST-2, we propose that the interaction of the three complexes with sequences from ST-1 and ST-2 (including the Inr element) is of low enough affinity that the interactions can only be inferred in the presence of high concentrations of oligonucleotides used during competition. On the basis of the data presented in Fig. 3, we conclude that the E2F-like site is required for initiating complex formation and the protein complexes may contact other sequences within the PCNA promoter.
Nuclear Proteins Interact Specifically with the PCNA E2Flike Site-We noted that a sequence (5Ј-GGGCGC-3Ј) between ϩ14 and ϩ19, which overlaps with the E2F-like site, has only one mismatch out of 6 bases when compared to the SP1 consensus sequence (5Ј-GGGCGG-3Ј). To determine the contribution of the E2F-like site and the SP1 site to complex formation, one additional E2F-like site mutation ST-3m2 (see "Materials and Methods"), was made. Unlike ST-3m1, ST-3m2 should only interfere with E2F binding but not the binding to the putative SP1 site. Results in Fig. 4A showed that both ST-3m1 and ST-3m2 abolished the formation of all three complexes, suggesting that the overlapping SP1 site is not required for complex formation. In addition to the three bands, some faster migrating bands appear variably in different experiments. A fast migrating band may represent an SP1-binding complex because this complex apparently still forms with the ST-3m2 probe in which the SP-1 site is intact, but not with the ST-3m1 probe.
The results of a competition binding experiment are shown in Fig. 4B. Consistent with the results from Fig. 4A, mutations in the E2F-like site, both ST-3m1 and ST-3m2, reduce the inhibition of complex formation with the wild type ST-3 probe. The relative binding affinity for the wild type and mutated E2F-like sequences is: ST-3 Ͼ Ͼ ST-3m2 Ͼ ST-3m1. In addition to complexes I, II, and III, a series of fast migrating complexes could be detected with the radiolabeled ST-3 probe. The two bands immediately below complex II (a1 and a2) and the fastest migrating band (a3) show a competition pattern similar to complex II suggesting that they may also form on the E2F-like site. However, the nature of these binding complexes is not ascertained because of their low abundance and because they are only present in some of our nuclear extract preparations. The relatively abundant fast migrating bands (b and c) were judged to be nonspecific for the E2F-like site, band c might represent an SP-1 related binding factor since band c is competed by ST-3 m2 but only very poorly by ST-3m1. The functional relevance of these binding activities is unknown.
Functional Role of the Interactions between Nuclear Factors and the PCNA E2F-like Site-To determine the contribution of this E2F-like site to PCNA promoter activity, two PCNA mutants (M-1 and M-2) corresponding to the mutations in ST-3m1 and ST-3m2 were prepared in the context of the pPCNA5-LUC construct (Ϫ182 to ϩ143) as described previously . In these constructs, PCNA genomic sequences were inserted into a HindIII site of a promoterless vector pSVOA. These constructs were transiently transfected into IL-2-stimulated L2 cells by the DEAE-dextran method. Because of the difficulties in transfecting quiescent L2 cells, we stimulated L2 cells with IL-2 for 22 h prior to transfection, as described under "Materials and Methods." Fig. 5 shows that mutations in the E2F-like site (M-1 and M-2) cause partial reduction in PCNA promoter activity. The relative promoter strengths of the wild type promoter and the two promoter mutants are consistent with the results of the relative binding affinity determined by EMSA (Fig. 4C), wt Ͼ M-2 Ͼ M-1. These data demonstrate that the E2F-like binding activity is required for optimal PCNA promoter activity and transcriptional activation of the PCNA gene in IL-2-stimulated T cells.
Methylation Interference Analysis of the Interaction of Nu-clear Proteins with the E2F-like Site-To determine the exact contact bases of these binding complexes within the PCNA promoter, methylation interference analysis was performed. In this experiment, only the two relatively abundant complexes, I and II, were analyzed (Fig. 6). ST was end-labeled and partially methylated by dimethyl sulfate before being subjected to EMSA. Both the protein-bound and free oligonucleotides were recovered from the gel and treated with piperidine, which specifically cleaves all the methylated guanidine (G) residues. Then the cleavage products were resolved on a sequencing gel. Methylation interference analysis for both of the DNA strands is shown in Fig. 6. Maxam and Gilbert DNA sequencing reactions were performed to confirm the positions of the G-residues in the PCNA promoter sequence. Methylation interference analysis of the shifted bands corresponding to complex I and complex II resulted in identical interference patterns centered at the E2F-like site on both strands. The pattern shows specificity for the E2F-like site on both strands with the top (coding) strand results more clearly overlapping with the E2F-like motif. The top strand contains three G-residues within the E2F site that interfere with E2F binding when methylated. On the bottom strand, one of the two G-residues within the E2F-like site shows methylation interference. The second G residue in the center of the E2F-like motif is very poorly cleaved in the free DNA and, thus, it is impossible to determine whether there is methylation interference at this position.
This pattern is consistent with the results obtained from both EMSA and competition experiments (Fig. 4). Notably, the Inr sequence CATT located at Ϫ2 to ϩ2 is not protected in the methylation interference analysis suggesting that the nuclear protein is in close contact with DNA at the E2F-like site. Our data show variable and weak methylation interference at the PEA3 site which we believe is inconclusive and requires further experimentation to determine if there is protein binding at this site. Taken together, we conclude that the major binding com- FIG. 4. Specificity of nuclear factors binding to PCNA E2F-like site. A, EMSA of nuclear factors binding to PCNA E2F-like site. PCNA ST-3 (wt), ST-3m1 (m1), and ST-3m2 (m2) were 32 P-labeled and incubated with L2 cell nuclear extract. B, 32 P-labeled ST-3 was incubated with 24 h IL-2-stimulated L2 cell nuclear extract. Unlabeled oligonucleotides ST-3, ST-3m1, and ST-3m2 were used as competitors. Increasing amounts of each competitor (0, 1, 5, and 20 ng) were used in the binding reactions. Complex I and II are the most prominent bands. Low amounts of complex III can also be found. The faster migrating bands (a1, a2, a3, b, and c) were variable in repeated experiments. Free probe (free) runs at the bottom of the gel.
FIG. 5. Effect of mutations at the E2F-like site on PCNA promoter activity in L2 cells. pPCNA5-LUC construct containing Ϫ182 to ϩ143 genomic sequence of the mouse PCNA gene and two E2F-like site mutation constructs (M-1 and M-2) were transiently transfected into L2 cells. The promoterless vector pSVOA was also transfected as a negative control. To assure efficient transfection, L2 cells were stimulated with IL-2 (10 -20 h) prior to transfection. Luciferase activities are presented as the percentage of luciferase activity derived from transfection of the wild type PCNA construct, pPCNA5-LUC (pPCNA5). All results were normalized to luciferase DNA content as measured by DNA slot blotting. Results represent the average of duplicates. The experiment was repeated three times with similar results. Standard deviations of luciferase activity are indicated. plexes form at the E2F-like sequence between ϩ10 and ϩ17.
The Nuclear Proteins That Interact with the E2F-like Site Are Distinct from E2F-To determine whether a member of the transcription factor E2F family binds to the PCNA E2F-like site, nuclear proteins binding to oligonucleotides containing the E2F consensus sequence derived from the hamster DHFR promoter were compared with those binding to the PCNA E2F-like site. We concluded on the basis of these data (Fig. 7) that the E2F family members do not bind to the PCNA E2F-like site because: 1) there is no cross-competition between the PCNA E2F-like site and DHFR-E2F site (Fig. 7, lanes 3 and 10). 2) E2F binds to the hamster DHFR-E2F oligonucleotide only in the presence of salmon sperm DNA but not with poly(dI-dC) as carrier DNA (Fig. 7, lanes 7-11). In contrast, complexes binding to PCNA E2F-like sequence only form with poly(dI-dC) but not with salmon sperm DNA as carrier DNA (compare lane 1 and 6). In addition, anti-E2F-1 antibody failed to supershift the complex formed with the PCNA E2F-like site. 2 These results demonstrate that the proteins that bind to the PCNA E2F-like site are clearly distinct from the E2F family of proteins.
Biochemical Characterization of the PCNA E2F-like Binding Protein-To provide additional support for the contention that nuclear proteins binding to the PCNA E2F-like site are distinct from previously characterized E2F family members, the DNAbinding component of this complex was identified biochemically by photoaffinity labeling using a bromodeoxyuridine-substituted radiolabeled probe. Complexes formed between ST probe and IL-2-stimulated L2 cell nuclear extract were re-solved in an EMSA and were covalently cross-linked by UV irradiation; subsequently, the cross-linked species were subjected to SDS-polyarylamide gel electrophoresis.
Several labeled bands were obtained with different doses of UV irradiation. The specificity of the photoaffinity labeling was determined by adding either nonspecific or specific DNA to the binding reactions. The labeling of the 180-kDa band was not competed with an excess of nonspecific oligonucleotide in the binding reaction (Fig. 8, lanes 3 and 7) but was efficiently competed with a specific competitor: an oligonucleotide bearing PCNA E2F-like site (Fig. 8, lanes 4 and 8). The binding specificity of the labeled species of 30-and 43-kDa proteins were not clear. These data demonstrated that a 180-kDa protein binds specifically to PCNA E2F-like site. It is distinct from the known E2F-binding proteins because E2F family members have been characterized to be 40 -60 kDa.

DISCUSSION
In mammalian cells, stimulation of quiescent cells by serum or growth factors causes a marked increase in the mRNA levels of PCNA, and other genes coding for proteins of the DNA synthesis machinery, such as DHFR, thymidine kinase, DNA polymerase ␣, and thymidine synthase (Pardee, 1989;Nevins, 1992).
The growth factor-responsive expression of the mouse PCNA gene in L2 cells requires elements within a 180-base pair region immediately upstream of the transcriptional start site . Previously we have shown that tandem cAMP response element binding protein/ATF binding sites located at nucleotides from Ϫ37 to Ϫ52 in the PCNA promoter are critical for IL-2-induced PCNA promoter activity (Huang et 2 D. Huang and M. B. Prystowsky, unpublished data. FIG. 6. Methylation interference analysis of nuclear proteins binding to PCNA E2F-like binding site. A, coding strand of the PCNA ST was end-labeled at the 5Ј-end. The probe DNA was partially methylated with dimethyl sulfate before addition to the binding reaction. Labeled DNA in the free (F) and bound (I and II) bands were recovered, cleaved at sites of methylation, and analyzed by urea-PAGE. The methylated guanine residues that interfere with binding of proteins are depicted at the side (G). Maxam and Gilbert sequencing, G, G ϩ A, T ϩ C, and C reactions were performed and loaded in parallel to confirm the positions of each guanine residue in the PCNA sequence. B, analysis of the non-coding strand of the PCNA probe (5Ј-end labeled), the same procedure was performed as above. C, sequence of PCNA ST and its methylation interference pattern are summarized.

FIG. 7.
Comparison of complex formation with the PCNA E2Flike site and DHFR-E2F motif. PCNA ST-3 (PCNA), hamster DHFR-E2F oligonucleotides (E2F), and DHFR-E2F with mutation at the E2F motif (E2Fm) were 32 P-labeled and incubated with 24-h IL-2-stimulated L2 cell nuclear extract. Poly(dI-dC) or salmon sperm DNA was used as nonspecific competitor DNA. They each gave different binding complexes. In addition to nonspecific DNA, various oligonucleotides (50 ng) containing PCNA ST-3 (PCNA), DHRF-E2F, or DHFR-E2F mutation (DHFR-E2Fm) were used as competitors in the binding reactions. The specific PCNA binding complexes (I and II) and specific E2F binding complex are indicated. al., 1994). In this report, we identify an E2F-like motif with the sequence 5Ј-TTGCGGGC-3Ј located between ϩ10 and ϩ17 of the murine PCNA promoter that is required for PCNA transcriptional regulation. It is worth noting that in several other genes, including the hamster and murine DHFR promoters, E2F elements are found within a sequence near transcription start sites and are critical for transcriptional regulation (Blake and Azizkhan, 1989;Means et al., 1992).
Transcription factor E2F was originally identified as a protein binding to a cis-element needed for the activation of the adenoviral early E2 gene promoter (Kovesdi et al., 1986;Berk, 1986). There are at least six similar but not identical E2F-like sites found in promoters of several cellular genes, with the sequence often being TTTCGCGC or TTTGGCGC (Mudryj et al., 1990;Nevins, 1992). Regulation of E2F activity involves its interactions with multiple cellular proteins including pRB, cyclin A, and cdk2 (Nevins, 1992). The retinoblastoma protein pRB as a tumor suppressor forms a specific complex with E2F providing an important mechanism for pRB regulation of cell growth. E2F-1 cDNA was cloned by virtue of its ability to bind to pRB (Helin et al., 1992). Soon after its cloning, it became apparent that the term E2F represents a family of transcription factors including E2F-1, DP-1, E2F-2, E2F-3, E2F-4, and E2F-5 (La Thangue, 1994;Lees et al., 1993;Ginsberg et al., 1994;Sardet et al., 1995). Recent studies have shown E2F-1 and DP-1 normally form heterodimers in binding to the E2F site. Therefore, E2F is an expanding group of related transcription factors with similar DNA-binding specificity that appear to function in cell cycle regulation.
Although there is sequence similarity between the PCNA promoter ϩ10 to ϩ17 region and the consensus E2F binding site, we concluded that distinct proteins bind to the PCNA E2F-like site based on the following observations: 1) there is no cross-competition between the PCNA E2F-like site and an oligonucleotide containing the E2F consensus sequence from the DHFR promoter (Fig. 7) or the E2F site from the adenoviral E2 promoter (data not shown). 2) The optimal condition for com-plex formation differs, when used as a nonspecific competitor DNA in the binding reactions, poly(dI-dC) favors complex formation with the PCNA E2F-like site while salmon sperm DNA favors complex formation with the DHFR-E2F site. 3) Results from a UV cross-linking experiment indicate that a protein of 180 kDa binds to the PCNA E2F-like site. In contrast E2F-1 is a 54-kDa nuclear protein previously identified in cross-linking experiments by several groups (Mudryj et al., 1990;Chellappan et al., 1991). The other E2F proteins have molecular mass ranging from 40 to 60 kDa. Therefore, this 180-kDa protein binding to the PCNA E2F-like site is distinct from all known E2F family proteins. However, we cannot rule out the possibility that this 180-kDa protein is related to the E2F family of transcription factors.
RNA polymerase is known to bind to DNA at the transcription start site to form an initiation complex with general transcription factors such as TFIID. Three forms of RNA polymerase II large subunit have been described, each differing in apparent molecular mass, 180, 210, and 240 kDa (Sawadogo and Sentenac, 1990;Young, 1991). Is it possible that the 180-kDa protein binding next to the initiation site represents RNA polymerase? Our results argue that this protein is not RNA polymerase because: first, methylation interference data indicate that the 180-kDa protein contacts a small region (nt ϩ12 to ϩ19) within or adjacent to the E2F-like site, while previous analyses have revealed that RNA polymerase protects a much larger region of DNA on both sides of the initiation site (Sawadogo and Sentenac, 1990); second, the interaction of the 180-kDa protein with DNA requires the E2F-like site, while RNA polymerase is thought to bind to DNA in a nonspecific manner.
A similar situation exists with studies of the DHFR promoter and our findings with the mouse PCNA promoter. E2F proteins have been shown to bind to the E2F site in the murine DHFR promoter between nt Ϫ10 and Ϫ3 (Slansky et al., 1993) but Means et al. (1992) have shown that HIP1, a 180-kDa protein also binds near this site at Ϫ9 to Ϫ1. This sequence has been shown to function as the initiator element that controls the start site for murine DHFR transcription (Means and Farnham, 1990). The protein binding to the PCNA E2F-like site that we characterized in this report is similar to HIP1 protein. 1) Both proteins have molecular mass of approximately 180 kDa. 2) Both bind to a sequence at or adjacent to the transcription initiation site. 3) Both proteins contact sequences which share homology to the E2F site. Our data suggest that the 180-kDa protein does not bind to the hamster DHFR E2F site (Fig. 7). We noticed that the murine DHFR E2F motif contains a G as the second base 5Ј of the E2F site, in contrast, the hamster E2F motif contains an A at the Ϫ2 position. Previously it has been shown that the sequence adjacent to the murine DHFR E2F core sequence is required for complex formation of HIP1 (Means et al., 1992). Therefore a difference of the sequences in the Ϫ2 position of the DHFR E2F elements may account for the observation that this 180-kDa protein binds to the PCNA E2F-like site as well as the murine DHFR E2F motif but not to the hamster DHFR E2F motif.
Mutations that abolish complex formation on this E2F-like site only reduced PCNA promoter activity moderately, to about 60 -80% of the wild type promoter activity. One of the explanations for this is that other cis-elements including the tandem CRE/ATF binding sites in the PCNA promoter also contribute substantially to optimal promoter activity. Further experiments making double or multiple mutations of these cis-elements in the PCNA promoter should help to determine the cooperation between these sites for PCNA promoter activity.
The identification of an E2F-like binding site in the PCNA FIG. 8. Biochemical characterization of the E2F-like binding activities. UV cross-linking of proteins to 32 P-labeled PCNA ST probe was performed as described under "Materials and Methods." Samples were UV irradiated for either 30 min (lanes 1-4) or 60 min (lanes 5-8) as indicated. A band of approximately 180 kDa is labeled (Ϫ). Competition experiments were performed in the presence of different unlabeled oligonucleotides. Lanes 1 and 5, with no nuclear extract; lanes 2 and 6, with nuclear extracts in the absence of competitor; lanes 3 and 7, with nuclear extracts in the presence of nonspecific-competitor (NS) oligonucleotides (containing a CCAAT box); lanes 4 and 8, with nuclear extracts in the presence of ST as a specific competitor (S). promoter and of nuclear proteins interacting with this E2F-like site in this report provides a basis for further study on the mechanism of PCNA transcriptional regulation and contributes to our understanding of the mechanisms by which IL-2 activates T lymphocyte proliferation.