E1A 13S Research Articles

Structural Limitations of the Ad5 E1A 12S Nuclear Localization Signal

The Ad5 E1A 12S gene encodes an oncoprotein with the ability to immortalize and cooperate with other viral or cellular oncoproteins to transform primary epithelial cells. The immortalizing function is dependent on the protein's efficient localization to the nucleus. A five amino acid nuclear localization signal (NLS), Lys-Arg-Pro-Arg-Pro, has been identified at the extreme COOH-terminus. This signal is necessary but not sufficient for efficient nuclear localization. A mutational analysis has been undertaken to further characterize the 12S NLS. The individual amino acids of the signal appear to have varying functional relevance. The lysine residue (a.a. 239) and the first arginine residue (a.a. 240) are the most critical. Changing the second arginine (a.a. 242) to threonine or either proline (a.a. 241 or 243) to alanine marginally diminishes signal function. Replacing the 12S NLS with the SV40 large T antigen (LT) NLS does not measurably affect the protein's nuclear localization. Sequences directly upstream of the NLS have a significant role in the proper localization of the 12S protein as illustrated by inefficiently localized mutants that have deletions of these sequences. Analyses of these mutants using a monoclonal antibody that recognizes the COOH-terminal four amino acids of the NLS have revealed that their signals are probably masked. To further investigate the importance of protein context in signal function, several NLS insertion mutants were constructed. Two regions in the first exon with predicted high surface probabilities and no known functions were chosen as sites for NLS insertions. Neither a wild-type 12S- nor a SV40 LT-NLS was functional in any of the new locations, indicating that for 12S, positioning of the NLS in the protein is critical.

Adenovirus E1A 13S gene product upregulates tumor necrosis factor gene.

Tumor necrosis factor (TNF) is a potential mediator of adenovirus-mediated lung inflammation. I postulated that early genes of adenovirus transactivate the TNF gene as a possible mechanism. To examine this hypothesis, I transfected T-lymphocyte-like Jurkat cells and monocyte/macrophage-like THP-1 cells with plasmids coding for adenovirus E1A 12S or 13S proteins along with a plasmid containing the TNF promoter linked to chloramphenicol acetyltransferase (CAT). In unstimulated Jurkat cells, E1A 13S increased TNF CAT activity 21-fold over cells transfected with control E1A plasmid, whereas 12S had a minimal effect. In unstimulated THP-1 cells, 13S increased TNF CAT activity by almost twofold over cells transfected with the control E1A plasmid; 12S had no effect. The effect of 13S was present in both cell lines when stimulated [Jurkat cells by phorbol 12-myristate 13-acetate; THP-1 cells, by lipopolysaccharide (LPS)]. E1A 13S also increased endogenous TNF mRNA production in LPS-stimulated THP-1 cells. These studies show adenovirus E1A 13S stimulates the TNF gene in inflammatory cell lines.

Adenovirus E1A activates cyclin A gene transcription in the absence of growth factors through interaction with p107.

Using the infection of quiescent human fibroblasts with adenovirus type 5 and various deletion mutants, we show that E1A can stimulate transcription of the cyclin A gene in the absence of exogenous growth factors. Required for this activity is conserved region 2 (CR2), while both the N-terminal part of E1A and CR1 are dispensable. This indicates that activation of cyclin A gene expression requires the binding of E1A to p107, while binding to either pRB or p300 is not involved in transcriptional activation. We demonstrate that p107 represses the cyclin A promoter through its cell cycle-regulatory E2F binding site and that 12S E1A can activate the cyclin A promoter, essentially by counteracting its repression by p107. Since Cr2 is required for cell transformation, transcriptional activation of the cyclin A gene by E1A appears to be important for its capacity to override control of cellular growth.

The adenoviral transcription factor, E1A 13S, trans-activates the human tumor necrosis factor-α promoter

The 1311 bp TNF-α promoter region fused to a luciferase reporter vector was used in a transient transfection system to study the regulation of TNF-α promoter activity by E1A 13S in the U937 macrophage cell line and the MLA 144 T cell line. Co-transfections of the TNF-α promoter with an E1A expression vector resulted in a strong trans-activation of the promoter in both cell lines. Sequential truncation of the promoter mapped the E1A responsive region to sequences contained between −120 bp and the transcription start site. Truncation to −95 bp caused a dramatic 87% reduction of E1A activation in MLA 144 cells and further truncation to −36 bp caused a complete loss of E1A activation. In U937 cells, each truncation lowered E1A responsiveness but activity was never completely abolished. Site-directed mutagenesis of putative cis-acting sequences in the TNF-α promoter identified the AP-1 site as important for E1A trans-activation in the U937 cell line; the AP-2 and CRE sites also appeared to contribute to a lesser degree. In contrast, only the CRE mutation caused a reduction in E1A induced activity in the MLA 144 cell line. Co-transfection of the E1A expression vector with expression vectors for the cellular transcription factors AP-1, AP-2 and CREB indicated that none of these transcription factors showed any co-operativity with E1A. Thus, cis-acting sequences which contribute to E1A trans-activation of the TNF-α promoter have been delineated.

E1A-induced immortalization of rat type II alveolar epithelial cells.

Using a retroviral vector expressing the adenoviral 12S E1A gene product the authors have immortalized rat type II alveolar epithelial cells. For a period of time, the immortalized cells retain many of the ultrastructural characteristics of type II cells in situ, including the presence of lamellar bodies. By 250 days in culture, however, neither lamellar bodies, SP-A, nor a phospholipid profile characteristic of surfactant were present. The cell bind the lectin Maclura pomifera and stably express cytokeratins and the E1A gene product. The cell line also has a diploid karyotype, exhibits contact inhibition of growth, and does not grow in soft agar. E1A-immortalized cell lines should prove useful as models for study of certain aspects of type II alveolar epithelial cell function.

Phenotypic Determinants of Adenovirus E1A Gene Autoregulation: Variable Region between Conserved Coding Domains 2 and 3

Different serotypes and evolutionary variants of human adenoviruses exhibit distinctive patterns of positive and negative autoregulation of the viral E1A gene. An autoregulatory E1A promoter mutation of the adenovirus type 3 (Ad3) E1A gene renders Ad3hr15 incapable of growth in normally permissive cells. The promoter mutation is complementedin transby E1A products of the heterologous helper adenovirus type 5 (Ad5). Second-site revertants of Ad3hr15 restore viability with high levels of E1A gene expression. The revertant E1A genotypes retain the mutant E1A promoter and have small in-frame deletions in the nonconserved region between the repression- and activation-associated conserved domains CR2 and CR3. Plasmid expression vectors were constructed as 12S and 13S cDNA forms of revertant E1A genes. These were used in cotransfection experiments with a reporter gene plasmid under transcriptional control of the mutant Ad3hr15 E1A promoter. The repression of the Ad3hr15 E1A promoter by helper Ad5 or revertant 12S E1A cDNA was consistently greater than that effected by wild-type Ad3 12S E1A cDNA expression. Significantly greater levels of positive transactivation were observed in cotransfections with 13S cDNAs of Ad5 or with the 13S E1A cDNA of Ad3hr15 revertants, compared to the transactivation observed with the mutant-encoded wild-type Ad3 13S E1A cDNA. The Ad5 helper anddl-revertant phenotype of Ad3hr15 appear to be related to transactivation activities of coexpressed E1A genes. The nonconserved region which separates the conserved coding regions CR2 and CR3 of the type 3 E1A gene acts to attenuate E1A-mediated repression and transactivation of transcription.

CBP-induced stimulation of c-Fos activity is abrogated by E1A.

The CBP protein stimulates transcription of cAMP-responsive genes by binding to the phosphorylated activation domain of the CREB transcription factor. Here we show that CBP stimulates transcription of Fos/Jun activity in F9 cells and that this response of mediated, at least partly, via c-Fos. We show that CBP binds c-Fos in a phosphorylation-independent manner in vitro, using a domain distinct from that required to bind CREB. When this CBP domain is linked to the activation domain of VP16 it can stimulate GAL4-Fos activity in vivo. The domain of CBP that binds c-Fos is also used to contact the E1A protein. We therefore asked whether the documented repression of AP1 activity by E1A is due to sequestration of CBP from c-Fos. We show that E1A 12S can repress c-Fos activation functions. The use of E1A mutants indicates that binding of CBP, but not RB, to E1A is essential for E1A-mediated repression. These data support a model whereby E1A can modulate AP1 activity by directly competing for the CBP co-activator protein.

Suppression of Mutations in Two Saccharomyces cerevisiae Genes by the Adenovirus E1A Protein

The protein products of the adenoviral E1A gene are implicated in a variety of transcriptional and cell cycle events, involving interactions with several proteins present in human cells, including parts of the transcriptional machinery and negative regulators of cell division such as the Rb gene product and p107. To determine if there are functional homologs of E1A in Saccharomyces cerevisiae, we have developed a genetic screen for mutants that depend on E1A for growth. The screen is based on a colony color sectoring assay which allows the identification of mutants dependent on the maintenance and expression of an E1A-containing plasmid. Using this screen, we have isolated five mutants that depend on expression of the 12S or 13S cDNA of E1A for growth. A plasmid shuffle assay confirms that the plasmid-dependent phenotype is due to the presence of either the 12S or the 13S E1A cDNA and that both forms of E1A rescue growth of all mutants equally well. The five mutants fall into two classes that were named web1 and web2 (for "wants E1A badly"). Plasmid shuffle assays with mutant forms of E1A show that conserved region 1 (CR1) is required for rescue of the growth of the web1 and web2 E1A-dependent yeast mutants, while the N-terminal 22 amino acids are only partially required; conserved region 2 (CR2) and the C terminus are dispensable. The phenotypes of mutants in both the web1 and the web2 groups are due to a single gene defect, and the yeast genes that fully complement the mutant phenotypes of both groups were cloned. The WEB1 gene sequence encodes a 1,273-amino-acid protein that is identical to SEC31, a protein involved in the budding of transport vesicles from the endoplasmic reticulum. The WEB2 gene encodes a 1,522-amino-acid protein with homology to nucleic acid-dependent ATPases. Deletion of either WEB1 or WEB2 is lethal. Expression of E1A is not able to rescue the lethality of either the web1 or the web2 null allele, implying allele-specific mutations that lead to E1A dependence.

The N-terminal region of the adenovirus type 5 E1A proteins can repress expression of cellular genes via two distinct but overlapping domains

The transforming E1A 12S and E1A 13S proteins of human adenovirus type 5 (Ad5) contain two and three conserved regions, respectively. In the present study, the contribution of sequences in the nonconserved N-terminal region of the E1A proteins to morphological transformation and to down-regulation of a number of mitogen-inducible genes was investigated. As described previously, transformation of NRK cells (an established normal rat kidney cell line) results in denser cell growth and a cuboidal cellular morphology. None of the cells expressing N-terminally mutated E1A proteins, however, show these transformed properties, which suggests an important role for sequences in that domain. The decrease in cyclin D1 levels requires exactly the same sequences. The ability to transform NRK cells and to reduce cyclin D1 levels does not correlate with the presence in the E1A proteins of binding domains for p300, CBP, p107, pRb, cyclin A, or cdk2. In contrast, down-regulation of expression of the JE gene in NRK cells and repression of transcription of the collagenase gene in human HeLa cells does correlate with the presence in the E1A proteins of an intact binding domain for p300 and CBP. The results suggest that the N-terminal domain of the E1A proteins can repress expression of cellular genes by at least two different mechanisms.

Repression in vitro, by human adenovirus E1A protein domains, of basal or Tat-activated transcription of the human immunodeficiency virus type 1 long terminal repeat.

Human adenovirus E1A proteins can repress the expression of several viral and cellular genes. By using a cell-free transcription system, we demonstrated that the gene product of the E1A 12S mRNA, the 243-residue protein E1A243R, inhibits basal transcription from the human immunodeficiency virus type 1 (HIV-1) long terminal repeat (LTR). The HIV-1 transactivator protein Tat greatly stimulates transcription from the viral promoter in vitro. However, E1A243R can repress Tat-activated transcription in vitro. Strong repression of both basal and Tat-activated transcriptions requires only E1A N-terminal amino acid residues 1 to 80. Deletion analysis showed that E1A N-terminal amino acids 4 to 25 are essential for repression, whereas amino acid residues 30 to 49 and 70 to 80 are dispensable. Transcriptional repression by E1A in the cell-free transcription system is promoter specific, since under identical conditions, transcription of the adenovirus major late promoter and the Rous sarcoma virus LTR promoter was unaffected. The repression of transcription by small E1A peptides in vitro provides an assay for investigation of molecular mechanisms governing E1A-mediated repression of both basal and Tat-activated transcriptions of the HIV-1 LTR promoter.

Adenovirus E1A proteins interact with the cellular YY1 transcription factor.

The adenovirus 12S and 13S E1A proteins have been shown to relieve repression mediated by the cellular transcription factor YY1. The 13S E1A protein not only relieves repression but also activates transcription through YY1 binding sites. In this study, using a variety of in vivo and in vitro assays, we demonstrate that both E1A proteins can bind to YY1, although the 13S E1A protein binds more efficiently than the 12S E1A protein. Two domains on the E1A proteins interact with YY1: an amino-terminal sequence (residues 15 to 35) that is present in both E1A proteins and a domain that includes at least a portion of conserved region 3 (residues 140 to 188) that is present in the 13S but not the 12S E1A protein. Two domains on YY1 interact with E1A proteins: one is contained within residues 54 to 260, and the other is contained within the carboxy-terminal domain of YY1 (residues 332 to 414). Cotransfection of a plasmid expressing carboxy-terminal amino acids 332 to 414 of YY1 fused to the GAL4 DNA-binding domain can inhibit expression from a reporter construct with GAL4 DNA binding sites in its promoter, and inclusion of a third plasmid expressing E1A proteins can relieve the repression. Thus, we find a correlation between the ability of E1A to interact with the carboxy-terminal domain of YY1 and its ability to relieve repression caused by the carboxy-terminal domain of YY1. We propose that E1A proteins normally relieve YY1-mediated transcriptional repression by binding directly to the cellular transcription factor.

Two domains of p53 interact with the TATA-binding protein, and the adenovirus 13S E1A protein disrupts the association, relieving p53-mediated transcriptional repression.

The tumor suppressor gene product p53 can activate and repress transcription. Both transcriptional activation and repression are thought to involve the direct interaction of p53 with the basal transcriptional machinery. Previous work has demonstrated an in vitro interaction between p53 and the TATA-binding protein that requires amino acids 20 to 57 of p53 and amino acids 220 to 271 of the TATA-binding protein. The present results show that a 75-amino-acid segment from the carboxy terminus of p53 also can bind to the TATA-binding protein in vitro, and this interaction requires amino acids 217 to 268 of the TATA-binding protein, essentially the same domain that is required for interaction with the amino-terminal domain of p53. A carboxy-terminal segment of p53 can mediate repression when bound to DNA as a GAL4-p53 fusion protein. The amino- and carboxy-terminal p53 interactions occur within the domain on the TATA-binding protein to which the adenovirus 13S E1A oncoprotein has previously been shown to bind. The 13S E1A oncoprotein can dissociate the complex formed between the carboxy-terminal domain of p53 and the TATA-binding protein and relieve p53-mediated transcriptional repression. These results demonstrate that two independent domains of p53 can potentially interact with the TATA-binding protein, and they define a mechanism--relief of repression--by which the 13S E1A oncoprotein can activate transcription through the TATA motif.

Mitotic cycle reactivation in terminally differentiated cells by adenovirus infection

Different cell types (e.g., neurons, skeletal and heart myocytes, adipocytes, keratinocytes) undergo terminal differentiation, in which acquisition of specialized functions entails definitive withdrawal from the cell cycle. Such cells are distinct from quiescent (reversibly growth-arrested) cells, such as contact-inhibited fibroblasts. Terminally differentiated cells can not be induced to proliferate by means of growth factor stimulation or transduction of cellular oncogenes. An important first step toward defining the molecular basis for such unresponsiveness is to find a practical means to overcome the proliferative block. Furthermore, determining whether terminally differentiated, postmitotic cells still retain a potential competence for proliferation that can be reactivated would have important theoretical and practical implications. To address these questions, we exploited the properties of adenoviruses. These viruses can infect postmitotic cells and express E1A, a powerful activator of proliferation in reversibly growth-arrested cells. We infected terminally differentiated skeletal muscle cells and adipocytes with human adenovirus type 5 or 12, obtaining full reentry into the cell cycle, including DNA synthesis, mitosis, cytokinesis, and extended proliferation. Similar results were obtained with established cell lines and primary cells belonging to several species, from quail to humans. Genetic analysis indicated that the smaller splice product of E1A, E1A 12S, is sufficient to induce cell cycle reactivation in otherwise permanently nonmitotic cells. These results demonstrate that terminally differentiated cells retain proliferative potential and establish adenovirus as a convenient and powerful means to force such cells to reenter the cell cycle.

Two pathways for repression of human papillomavirus transcription by the adenovirus 5 E1A 12S product

Two pathways for repression of human papillomavirus transcription by the adenovirus 5 E1A 12S product

Protein domains connect cell cycle stimulation directly to initiation of DNA replication.

Polyoma large T antigen (LT) is the only viral gene product required for viral DNA replication. LT can be divided into two domains, one N-terminal (NT) spanning residues 1-260 and one C-terminal (CT) comprising approximately residues 264-785. NT is known to immortalize primary cells in a manner dependent on binding of pRB/p107. Here a CT construct comprising residues 264-785 was shown to have independent function in DNA replication. CT is entirely sufficient for driving viral DNA replication in vivo in growing mouse cells at a level approaching that of full-length LT. In contrast, CT is strikingly deficient for replication in serum-starved cells. However, this deficiency can be complemented by coexpression of NT. BrdUrd incorporation in transfected, starved cells showed that NT was sufficient for inducing S phase, suggesting a mechanism for complementation. By contrast, CT was unable to induce S phase when tested in the same assay. NT also promotes phosphorylation of sites in CT that are likely to be important for replication. Other DNA tumor virus gene products such as adenovirus E1A 12S and human papillomavirus 16 E7 could also complement CT for replication. Although NT, E1A 12S, and E7 all bind the retinoblastoma gene product (pRB) and p107, genetic analysis demonstrates an additional function, independent of that binding, is responsible for complementation.

E1A repression of IL-6-induced gene activation by blocking the assembly of IL-6 response element binding complexes.

Some viral products interfere with host antiviral defense mechanisms. Adenovirus E1A represses IFN signal transduction pathways which induces gene activation and an antiviral state. Both IFN and IL-6 activate Jak/Tyk protein tyrosine kinases and the STAT (signal transducer and activator of transcription) family proteins. We showed that 12S E1A repressed IL-6 signals activating the junB promoter and the two IL-6 response elements (REs), JRE-IL6 and type II IL-6 RE (also called acute phase response element), required for IL-6-induced activation of the junB promoter and the type II acute phase reactant genes, respectively, in hepatocytes. Conserved region 1 of the 12S E1A was responsible for the repression. Target molecules of the repression by E1A appeared to be IL-6-inducible DNA-binding proteins acting on the IL-6 REs. In a rat 3Y1 cell line stably expressing E1A, the levels of IL-6-induced IL-6 RE binding proteins were severely reduced compared with those in a parental 3Y1 cell line. Moreover, we found that the levels of the STAT family proteins including Stat1-alpha (p91), Stat1-beta (p84), Stat2 (p113), and Stat3 were decreased by the stable expression of adenovirus E1A. The E1A-induced reduction in the amount of DNA-binding proteins seemed to be partly responsible for the decreased transcriptional activity of the IL-6 RE-driven gene expression in response to IL-6. This repression mechanism may be applicable to the E1A repression of IFN-gamma-induced gene activation.

Adenovirus E1A 13S gene product up-regulates the cytomegalovirus major immediate early promoter.

Latent cytomegalovirus (CMV) infection is often reactivated in the lung. We postulated that this reactivation could occur by stimulation of the CMV major immediate early (IE) promoter by other viruses that infect the lung. The specific aim of this study was to investigate whether adenovirus early proteins could stimulate the CMV IE promoter in inflammatory cells. We transfected the monocyte/macrophage THP-1 cell line and the T-lymphocyte Jurkat cell line with plasmids coding for adenovirus E1A 12S or 13S proteins, along with a plasmid containing the CMV IE promoter region linked to the chloramphenicol acetyltransferase (CAT) reporter gene. In unstimulated THP-1 cells, the E1A 13S gene product increased CMV IE CAT activity by 18-fold compared with cells containing the control E1A plasmid. This effect was not seen in cells transfected with the E1A 12S plasmid. There was a similar effect of the E1A 13S gene product in LPS-stimulated THP-1 cells. In unstimulated Jurkat cells, the E1A 13S gene product stimulated CMV IE CAT activity by 19-fold compared with cells containing the E1A control plasmid; the E1A 12S gene product had no effect. There was a similar effect of the 13S E1A gene product in phorbol myristate acetate-stimulated Jurkat cells. These findings demonstrate that the CMV IE promoter can be stimulated by early viral proteins of adenovirus in inflammatory cells. These observations could be important for understanding the reactivation of latent CMV infection.

Modulation of transcriptional activation of the proliferating cell nuclear antigen promoter by the adenovirus E1A 243-residue oncoprotein depends on proximal activators.

Previous analyses defined a proliferating cell nuclear antigen (PCNA) E1A-responsive element (PERE) in the PCNA promoter that is essential for transactivation by the 243-residue product of the adenovirus type 2 E1A 12S mRNA (E1A 243R). In this report, we show that the PERE activates a heterologous basal promoter and confers susceptibility to transactivation by E1A 243R, indicating that the PERE is both necessary and sufficient for the response of the PCNA promoter to this oncoprotein. Insertion of linker sequences between the PERE and the site of transcription initiation in the PCNA promoter severely impairs the promoter's response to E1A 243R transactivation. GAL4 sites can replace the function of the PERE in the E1A 243R response of the PCNA basal promoter if transcriptional activators of suitable strength are supplied as GAL4 fusion proteins. Weak transcriptional activators render the PCNA basal promoter subject to transactivation by E1A 243R but do not endow the adenovirus E1B basal promoter with a similar response. Strong transcriptional activators do not support transactivation by E1A 243R, however; instead, E1A reduces the ability of the strong activators to activate both the PCNA and E1B basal promoters. Although other mechanistic differences might determine the response, the data imply a relationship between the activation strength of promoter-proximal effectors and the response of the PCNA basal promoter to E1A 243R. These experiments indicate that the PERE can function autonomously in mediating transactivation by E1A 243R and that the PCNA basal promoter is configured in a manner that permits modulation by E1A 243R of transcriptional activation by promoter-proximal effectors.

Modulation of transcriptional activation of the proliferating cell nuclear antigen promoter by the adenovirus E1A 243-residue oncoprotein depends on proximal activators.

Previous analyses defined a proliferating cell nuclear antigen (PCNA) E1A-responsive element (PERE) in the PCNA promoter that is essential for transactivation by the 243-residue product of the adenovirus type 2 E1A 12S mRNA (E1A 243R). In this report, we show that the PERE activates a heterologous basal promoter and confers susceptibility to transactivation by E1A 243R, indicating that the PERE is both necessary and sufficient for the response of the PCNA promoter to this oncoprotein. Insertion of linker sequences between the PERE and the site of transcription initiation in the PCNA promoter severely impairs the promoter's response to E1A 243R transactivation. GAL4 sites can replace the function of the PERE in the E1A 243R response of the PCNA basal promoter if transcriptional activators of suitable strength are supplied as GAL4 fusion proteins. Weak transcriptional activators render the PCNA basal promoter subject to transactivation by E1A 243R but do not endow the adenovirus E1B basal promoter with a similar response. Strong transcriptional activators do not support transactivation by E1A 243R, however; instead, E1A reduces the ability of the strong activators to activate both the PCNA and E1B basal promoters. Although other mechanistic differences might determine the response, the data imply a relationship between the activation strength of promoter-proximal effectors and the response of the PCNA basal promoter to E1A 243R. These experiments indicate that the PERE can function autonomously in mediating transactivation by E1A 243R and that the PCNA basal promoter is configured in a manner that permits modulation by E1A 243R of transcriptional activation by promoter-proximal effectors.

Differential Activation of Cyclin and Cyclin-Dependent Kinase Genes by Adenovirus E1A 12S cDNA Product

The differential activation of cyclin and cyclin-dependent kinase genes by the adenovirus E1A gene product (E1A) or serum factors was studied with a rat 3Y1 derivative cell line, g12-21, in which the E1A 12S cDNA can be expressed in response to dexamethasone (dex). The induction of DNA synthesis in quiescent g12-21 cells occurred within 12 h after serum stimulation, while it occurred within 8 h after treatment with dex. The expression of cyclin D1 and E genes in the serum-stimulated cells was induced in mid G1 and mid to late G1, respectively, while that of the cyclin D1 gene was not induced and the induction of the cyclin E gene was shifted to the G1/S boundary in the dex-treated cells. The cdk2 gene was induced in late G1 and cdc2 and cyclin A genes at the G1/S boundary in both serum-stimulated and dex-treated cells. These results suggest that E1A skips cell cycle events which normally occur in early to mid G1 and may directly activate late-response genes. Analysis of the transcription factor E2F complexes formed in the promoter regions of cdc2 and dihydrofolate reductase genes showed that the amount of complexes formed is maximal at the G1/S boundary, but decreases in S phase when these genes are transcribed extensively.