The pRB (retinoblastoma protein) has a central role in the control of the G1–S phase transition of the cell cycle that is mediated in part through the regulation of E2F transcription factors. Upon S-phase entry pRB is phosphorylated extensively, which in turn releases bound E2Fs to drive the expression of the genes required for S-phase progression. In the present study, we demonstrate that E2F1-maintains the ability to interact with ppRB (hyperphosphorylated pRB). This interaction is dependent upon the ‘specific’ E2F1-binding site located in the C-terminal domain of pRB. A unique region of the marked box domain of E2F1 contacts the ‘specific’ site to mediate the interaction with ppRB. The mechanistic basis of the interaction between E2F1 and ppRB is subtle. A single substitution between valine and proline residues in the marked box distinguishes E2F1's ability to interact with ppRB from the inability of E2F3 to bind to the ‘specific’ site in ppRB. The E2F1–pRB interaction at the ‘specific’ site also maintains the ability to regulate the transcriptional activation of E2F1 target genes. These data reveal a mechanism by which E2F1 regulation by pRB can persist, when pRB is hyperphosphorylated and presumed to be inactive.
Disruption of the G1 checkpoint of the cell cycle is a ubiquitous event in cancer that allows for inappropriate entry into the cell cycle . The tumour suppressor pRB (retinoblastoma protein) has a central role in the regulation of S-phase entry through its ability to repress the activity of E2Fs . E2Fs are potent transcription factors that function to activate the genes required to progress into the S-phase. Mitogenic signalling results in the activation of CDK (cyclin-dependent kinase) complexes, which phosphorylate pRB and free E2F transcription factors to drive cell cycle progression. The RB1 gene is mutated only in a small subset of cancers, which includes retinoblastoma and small cell lung cancer . Instead, the majority of human cancers express WT (wild-type) pRB that exists predominantly in an inactive phosphorylated state due to the deregulation of CDKs . Thereby, most human cancers disrupt the G1 checkpoint control upstream of pRB through the deregulation of CDK activity.
Inactivation of pRB by phosphorylation requires the activity of both the Cyclin D–CDK4/6 and the Cyclin E–CDK2 complexes . This inactivated form of pRB has often been defined by its slower migration in SDS/PAGE gel and is commonly referred to as ppRB (hyperphosphorylated pRB). ppRB has been shown to contain at least ten distinct phosphopeptides, indicating that it is phosphorylated extensively [6,7], while the faster migrating hypophosphorylated form appears to have limited phosphorylation of some of the same sites [7,8]. These observations, combined with mutational analysis of the 16 predicted CDK phosphorylation sites, has led to a model in which many phosphorylation sites contribute in a redundant manner to the displacement of E2F binding to ppRB [9,10]. The CDK phosphorylation sites are localized to disordered regions of pRB that flank the well-structured pocket domain (Figure 1A) . The small pocket domain consists of two halves termed A and B (Figure 1A) that each adopt a cyclin-like fold to form a large globular domain that is capable of interacting with E2F transcription factors . Co-crystallization studies have revealed an interaction between the transactivation domain of E2Fs and the cleft that forms between the two cyclin-like folds of the pocket domain. This is defined as the ‘general’ interaction (Figure 1B) [12,13]. CDK phosphorylation of pRB results in conformational changes of the unstructured regions containing the CDK phosphorylation sites that block the interaction with the transactivation domain of E2Fs . In this regard, a relatively detailed picture of how phosphorylation regulates pRB–E2F interactions has emerged. However, pRB contains an additional E2F-binding site that is utilized exclusively by E2F1 called the ‘specific’ site (Figure 1B) . This interaction is mediated by the marked box region of E2F1 and the C-terminus of pRB (Figure 1B) . The regulatory effects of CDK phosphorylation on this unique pRB–E2F1 interaction are unknown.
Phosphorylation sites and interactions mapped to the domain structure of pRB
There are eight E2F proteins in mammals that share the ability to regulate E2F target genes through a conserved DNA-binding domain . E2F1, E2F2 and E2F3 are defined as activator E2Fs as they contain strong nuclear localization signals and transactivation domains that allow them to induce the expression of S-phase targets . pRB regulates exclusively the activity of the activator class of E2Fs, indicating that they have an intimate relationship in cell cycle control . Gene-targeting experiments have demonstrated that a single activator E2F can support development in mice, indicating considerable redundancy . E2F1, however, appears to have a unique role in the induction of apoptosis that is distinct from its role in proliferation. The functional significance of this is emphasized by the apoptotic defects that occur in the thymus of E2F1−/− mice, and the susceptibility of these mice to multiple tumours including lymphoma [20,21]. E2F1 has the unique ability to activate the transcription of pro-apoptotic molecules including ARF (ADP-ribosylation factor), Apaf-1 (apoptotic peptidase activating factor 1), caspase 7, caspase 8, Bid (BH3-interacting domain death agonist) and p73 [22–26]. The ability of E2F1 to induce both cell proliferation and cell death necessitates a mechanism by which these contrasting activities can be controlled. Surprisingly, there is little data available to suggest a mechanism by which E2F1-induced apoptosis is controlled independently of proliferation. Using experiments that interchange domains from E2F1 and E2F3, the marked box region of E2F1 has been demonstrated to function in the activation of p73- and p53-induced apoptosis . The marked box region is a protein–protein interaction domain that has been shown to interact with cellular factors such as Jab-1 (Jun activating binding protein), to induce transcription of ARF and apoptosis , as well as the ‘specific’ site in the C-terminus of pRB (Figure 1B), which is capable of attenuating E2F1-induced apoptosis [15,16]. Based on our current understanding of the regulation of pRB–E2F interactions by CDK phosphorylation at the G1- to S-phase transition, it is difficult to reconcile how E2F1's pro-apoptotic activity is restrained in normal cells as they enter S-phase and E2Fs are released from pRB's regulation.
Surprisingly, a number of studies suggest the existence of pRB–E2F1 complexes under circumstances where CDK phosphorylation is expected to disrupt their interaction. First, ectopic expression of G1 Cyclin–CDKs has been shown to have differential effects on E2F1 release from pRB, suggesting that this complex may have an altered sensitivity to the kinases relative to other pRB–E2F complexes . Secondly, pRB and E2F1 have been found at the same E2F-responsive promoters and CpG islands in the S-phase by chromatin immunoprecipitation . Thirdly, following DNA damage, pRB–E2F1 complexes have been reported to assemble while, at least some, phosphorylation sites on pRB remain phosphorylated . Unfortunately, none of these reports offer a mechanistic explanation that accommodates both the release of E2Fs from ppRB and the maintenance of pRB–E2F1 interactions under the same circumstances.
The present study describes a mechanism whereby E2F1 can be bound and regulated by ppRB. This interaction is mediated by the C-terminal ‘specific’ binding site in pRB and the marked box domain of E2F1 (Figure 1B). Despite the high conservation of the marked box region of E2Fs, subtle but important sequence differences render only E2F1 capable of interacting with the C-terminus of pRB. Substitution of a single proline to valine residue in E2F3, to resemble E2F1, is sufficient to create an interaction with phosphorylated pRB. E2F1 interaction with the ‘specific’ site of pRB is also capable of regulating the activation of a pro-apoptotic gene promoter, whereas this interaction has little ability to regulate other E2F1-dependent transcription. Taken together, these data provide a biochemical basis for the ability of proliferative and apoptotic functions to be regulated differentially by pRB during cell cycle advancement.
Site-directed mutagenesis of a pRB cDNA was carried out by PCR as described previously [32,33]. Mutants were introduced into the bacterial GST (glutathione transferase)-RBLP (large pocket domain) or the GST-RB-C (C-terminal domain only) cloned into the pSCodon vector (Delphi Genetics). Mutagenesis of E2Fs was carried out in a similar manner using an E2F cDNA cloned into pBluescript and later subcloned into the CMV (cytomegalovirus)–HA (haemagglutinin) expression vector. All subclones of PCR products were sequenced to ensure that they contained only the desired mutations. CMV-HA, pBluescript, CMV-HA-E2F1, -E2F2, -E2F3, -DP1 and their sources have been described previously . CMV-CDK2, -CDK4 and –DN-CDK2, were reported initially by van den Heuvel and Harlow . CMV-Cyclin D and -Cyclin E were reported initially by Hinds et al. . The p73–Luc plasmid was reported initially by Urist et al. . The Myc–PP1 construct was first reported by Traweger et al. .
C33A and T98G cell lines were obtained from A.T.C.C. and cultured according to standard methods. Cell culture was carried out in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (v/v), 2 mM L-glutamine, penicillin (50 units/ml) and streptomycin (50 μg/ml). The C33A cells were used to generate extracts for GST pull-down and co-transfection immunoprecipitation experiments. The C33A cells were transfected using calcium phosphate with the precipitates left on the cells for 16 h before fresh growth medium was added. The T98G cells were used to characterize endogenous complexes between pRB and E2Fs synchronized by serum starvation for 72 h in medium with 0.1% fetal bovine serum. The cells were stimulated to re-enter the cell cycle with media containing 20% (v/v) fetal bovine serum.
Immunoprecipitation and Western blotting
GST pull-down and co-immunoprecipitation assays were performed as described previously . To generate extracts for these experiments, C33A cells were plated at a density of 6×106 cells per 15 cm dish and transfected with a total of 60 μg of DNA using calcium phosphate. 48 h after transfection the cells were harvested. To generate extracts for the GST pull-down assays, the cells were washed twice with PBS and collected into 1 ml of GSE (gel shift extract) buffer [20 mM Tris, pH 7.5, 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25% glycerol, 5 μg/ml leupeptin, 5 μg/ml aprotinin, 0.1 mM Na3VO4, 0.5 mM NaF and 1 mM DTT (dithiothreitol)]. Extracts were frozen at −80°C. To generate extracts for co-immunoprecipitation of the pRB–E2F complexes nuclear extracts were prepared. Briefly, the cells were washed twice and collected in 1 ml of PBS. The cells were then resuspended in three times the cell volume of hypotonic lysis buffer (20 mM Tris/HCl, pH 7.5, 10 mM KCl, 3 mM MgCl2, 1 mM EDTA, 1 mM PMSF, 5 μg/ml leupeptin, 5 μg/ml aprotinin, 5 mM NaF, 0.1 mM Na3VO4 and 1 mM DTT). The extracts were incubated on ice for 5 min before 0.05% Nonidet P40 was added to the hypotonic lysis buffer and then the extracts were incubated on ice for a further 5 min. The nuclei were pelleted by centrifugation at 4°C at 1700 g for 6 min and washed two times with hypotonic lysis buffer containing 0.05% Nonidet P40. The nuclei were resuspended in GSE buffer and frozen at −80°C.
The extracts were thawed and the cellular debris was removed by centrifugation at 4 °C at 20800 g for 30 min. For the co-transfection immunoprecipitations, C33A extract was diluted in IP wash buffer (20 mM Tris/HCl, pH 7.5, 200 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25 mM DTT and 0.1% Nonidet P40). pRB complexes were immunoprecipitated with 12CA5 for HA-tagged E2Fs, C-20 (Santa Cruz Biotechnology) for E2F1 and C-18 (Santa Cruz Biotechnology) for E2F3, bound to Protein G–Sepharose (GE healthcare). Immunoprecipitations were incubated with rocking for 1 h at 4 °C. The Protein G–Sepharose beads were washed twice with IP wash buffer then eluted in 1×SDS/PAGE sample buffer and resolved by SDS/PAGE (8% gel). Proteins were transferred to a nitrocellulose membrane by standard techniques. HA-tagged E2Fs were detected using 3F10 (Roche), E2F1 by KH20 (Santa Cruz Biotechnology), E2F3 by PG37 (Upstate Biotechnology), Myc-tagged PP1 by 9E10 and pRB by G3–245 (BD Pharmingen).
GST pull-down binding experiments
GST-fusion proteins were expressed in BL21-DE3-Gold Escherichia coli (Stratagene) in 500 ml cultures. Briefly, cells were grown for 2 h at room temperature (22 °C) after which 100 μM IPTG (isopropyl β-D-1-thiogalactopyranoside) was added to the cultures. The cultures were then grown overnight at 16 °C. The following morning, the cells were harvested and GST-fusion proteins were purified using glutathione–Sepharose according to standard protocols. Purified GST-fusion protein (2 μg) was diluted in low salt GSE buffer (20 mM Tris/HCl, pH 7.5, 200 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 1 mM DTT and 0.1% Nonidet P40) and incubated with 100 ml of whole cell C33A extract expressing HA–E2Fs or Myc–PP1. The GST–pRB complexes were precipitated with glutathione–Sepharose and washed twice with low salt GSE buffer and eluted with 1×SDS/PAGE sample buffer. Samples were electrophoresed on SDS/PAGE (8% gel) and blotted using the same antibodies outlined for the immunoprecipitation experiments above.
Luciferase reporter assays
Transcriptional reporter assays were carried out as reported previously . Saos-2 cells were plated in 6-well plates at a density of 5×105 cells per well. The cells were transfected with either 100 ng of the E2F4B-luciferase reporter or 200 ng of the p73–Luc reporter plasmid along with 15 ng of CMV-HA-E2F, 15 ng CMV-HA-DP1 and 200 ng of CMV-β-Gal. Increasing concentrations of CMV–pRB expression plasmid were transfected to block the activity of the transfected E2Fs. The total plasmid DNA was normalized with the addition of CMV–CD20. The cells were harvested 36 h after transfection with 1X Reporter Lysis Buffer (Promega). The luciferase activity was determined with the Luciferase assay system (Promega) and normalized to β-Gal activity. The β-Gal activity was determined using standard techniques to measure the hydrolysis of ONPG (2 nitrophenyl-β-D-galactopyranoside) at 405 nm. The average of three independent transfections is shown and the error bars indicate 1 S.D. from the mean.
A unique interaction between E2F1 and ppRB
The disruption of pRB–E2F complexes by phosphorylation is thought to be a critical event of the G1–S-phase transition of the cell cycle. Gel shift experiments have described the release of E2Fs from pRB upon S-phase entry [38,39]. These experiments have often utilized a double-stranded DNA probe from the adenovirus E2 promoter that contains a canonical E2F site and is bound by E2Fs and pRB–E2F complexes [38,39]. Complexes formed between pRB and E2F1 using the C-terminal ‘specific’ site have a low affinity for this type of probe , and as such pRB–E2F1 ‘specific’ complexes are not observed in gel shift experiments. Thereby, previous work that has described the release of E2Fs following cell cycle entry only pertains to the ‘general’ E2F interaction that is common among E2Fs that interact with pRB. To explore other binding sites that may be regulated independently of phosphorylation at the G1–S transition, we utilized co-immunoprecipitation to assess directly the ability of E2Fs to interact with all binding sites in ppRB.
T98G cells, which have an intact G1 checkpoint , were synchronized by serum starvation for 72 h, then induced to re-enter the cell cycle with medium containing 20% serum. Initially, pRB exists primarily in a hypophosphorylated state and further culture of cells in high serum results in a significant enrichment for ppRB (Figure 2A). Extracts from cells synchronized to enrich for ppRB were immunoprecipitated with E2F1 and E2F3 antibodies. Both E2F1 and E2F3 are capable of interacting with and co-immunoprecipitating pRB (Figure 2B). E2F3 only interacts with the hypophosphorylated form of pRB (Figure 2B) suggesting that phosphorylation disrupts the interaction between ppRB and E2F3. In contrast, E2F1 can immunoprecipitate both pRB and ppRB as determined by the electrophoretic mobility of the precipitated proteins (Figure 2B). The use of the shift in apparent molecular mass to detect ppRB ensures that it is phosphorylated extensively. This provides experimental evidence for complexes between endogenous ppRB and E2F1 and suggests there is functional relevance to E2F1 regulation after S-phase entry.
E2F1 forms a unique complex with ppRB in T98G cells
To further characterize the interaction between E2Fs and ppRB, the RB1-null cell line C33A was utilized. These cells do not respond to the ectopic re-expression of pRB, which allows for its phosphorylation state to be modulated independently of the cell cycle phase. To produce hypophosphorylated pRB, CDK2-DN (dominant negative CDK2) was expressed to block the activity of endogenous CDK complexes. Alternatively, hyperphosphorylated ppRB was produced by expression of Cyclin D–CDK4 and Cyclin E–CDK2 complexes (denoted as E2/D4 in Figure 2C). As shown in Figure 2(C), modulation of the kinase activity is sufficient to shift the ectopically expressed pRB from a hypophosphorylated to hyperphosphorylated state as determined by the electrophoretic mobility shift, as well as with phospho-specific antibodies raised against phosphorylated Ser807/Ser811. HA-tagged E2F and DP1 constructs were co-transfected and immunoprecipitated with an HA-specific antibody. The use of an HA antibody excludes the potential differences in the E2F1 and E2F3 antibodies to recognize E2F–ppRB complexes. This provides a system in which the phosphorylation state of pRB can be modulated to investigate the interaction of ppRB with different E2F transcription factors.
In cells expressing CDK2-DN, both HA–E2F1 and HA–E2F3 are capable of interacting with pRB, confirming that either can immunoprecipitate pRB in its hypophosphorylated state (Figures 2C and 2D). In cells expressing CDKs to produce predominately ppRB, HA–E2F3 is only capable of interacting with and immunoprecipitating the small amount of residual hypophosphorylated pRB that remains (Figure 2D). In contrast, HA–E2F1 is capable of immunoprecipitating ppRB as determined by both the electrophoretic mobility shift and with phospho-specific antibodies shown in Figure 2(C). This suggests that E2F1 has a significant affinity for ppRB that allows for the formation of stable complexes between these proteins. In contrast, hyperphosphorylation of pRB is sufficient to abrogate the binding of E2F3, revealing differential regulation of E2F1 and E2F3 by pRB.
Our experiments demonstrate that complexes between ppRB and E2F1 can be detected at endogenous levels, suggesting that this interaction occurs as part of normal cell cycle progression. While both E2F1 and E2F3 are capable of interacting efficiently with hypophosphorylated pRB, only E2F1 is capable of forming an interaction with ppRB. This demonstrates that the interaction with ppRB is a unique feature of E2F1 that allows for its independent regulation.
The ‘specific’ interaction is a unique feature of pRB and E2F1 that mediates the E2F1–ppRB complex
Detailed reports have described the structural mechanisms by which phosphorylation of pRB results in the release of bound E2F transcription factors. These studies report that phosphorylation induces multiple conformational changes in pRB that function to abrogate interaction with E2Fs [10,14,41,42]. The interaction between ppRB and E2F1 observed in the present study is an apparent contradiction to the structural models of pRB phosphorylation. However, many of these previous studies predate the identification of the ‘specific’ E2F1-binding site found in the C-terminus of pRB (Figure 1B). Therefore we next sought to determine whether the ‘specific’ site mediates the interaction between ppRB and E2F1.
The two distinct E2F-binding sites on pRB can be studied through the use of recombinant proteins that contain the entire RBLP or the RB-C. Recombinant proteins were incubated with extract from C33A cells expressing HA–E2F and HA–DP1 proteins. As shown in Figure 3(A), RBLP is capable of precipitating complexes with HA–E2F1, HA–E2F2, HA–E2F3 and HA–E2F4. In contrast, RB-C only precipitated HA–E2F1 in appreciable amounts (Figure 3A). Taken together this confirms that RBLP contains both the ‘general’ and the ‘specific’ E2F-binding sites while RB-C essentially contains only the ‘specific’ site. Furthermore, the availability of two distinct E2F1-binding sites allows E2F1 to adopt both interaction types interchangeably, as shown in Figure 3(B). However, the mechanism by which E2F1 contacts the ‘specific’ site has not been studied extensively and this raises the possibility that it may mediate the observed interaction between ppRB and E2F1.
Differentiating between ‘specific’ and ‘general’ E2F interactions with pRB
To assess the interaction of E2F1 at the ‘general’ and ‘specific’ binding sites of ppRB, mutants of pRB were utilized that disrupt the individual E2F-binding sites in isolation. Previously, a mutant termed ΔG was reported to disrupt the interaction between pRB and E2Fs at the ‘general’ site . This mutant contains substitutions in the A, B and C-terminal regions including Lys873 and Lys874, which are important contact sites of CDK2-associated cyclins and are necessary for efficient phosphorylation . As a result this mutant, defined as old-ΔG for the present study, is not phosphorylated to the same extent as the WT protein (Figure 3C). For this reason, a new ΔG mutant was created that disrupts selectively the ‘general’ E2F-binding site while other binding sites remain intact. This mutant contains R467E and K548E substitutions and is phosphorylated to a similar extent as WT-pRB when transfected into C33A cells (Figure 3C). In a similar manner to the old-ΔG mutant, the new ΔG mutant disrupts the ability of E2Fs to bind to the ‘general’ site, but maintains the ability to interact with E2F1 through the ‘specific’ site. This is highlighted by the inability of recombinant ΔG-pRB to interact with HA–E2F2, HA–E2F3 or HA–E2F4, but maintain an interaction with HA–E2F1 (Figure 3D).
As depicted in Figure 4(A) the novel ΔG-pRB mutant and the previously reported ΔS mutant , provide a means to study the two distinct E2F-binding sites. The ΔG mutant disrupts selectively the ‘general’ site in order to study the ‘specific’ site, while the ΔS mutant disrupts the ‘specific’ site allowing the ‘general’ site to be studied in isolation. This allows us to determine the binding site that mediates the observed complex between ppRB and E2F1. In a manner similar to our previous experiments, C33A cells were transfected with combinations of CDK complexes to modulate the phosphorylation state of the ΔG and ΔS pRB mutants. As shown in Figures 4(B) and 4(C), both the ΔG and ΔS pRB proteins are phosphorylated extensively by the expression of CDK complexes. As shown in Figure 4(B), E2F1 is capable of immunoprecipitating both the ΔG-pRB and ΔG-ppRB species, suggesting that the ‘specific’ site is sufficient to mediate the observed ppRB–E2F1 complex. To investigate this further, the ΔS mutation was employed to disrupt selectively the ‘specific’ site, thus directing E2F1 to the ‘general’ site. As shown in Figure 4(C), the ΔS mutant is also phosphorylated extensively when CDKs are expressed; however, HA–E2F1 is only capable of interacting with and immunoprecipitating the hypophosphorylated ΔS-pRB. The small amount of residual pRB that is precipitated migrates at the hypophosphorylated size and has almost no detectable phosphorylation at Ser807 and/or Ser811 (Figure 4C). This reveals a critical role for the ‘specific’ site in the ppRB–E2F1 complex formation.
The ‘specific’ site is required for the interaction between E2F1 and hyperphosphorylated pRB
This provides a biochemical basis for the observed ability of E2F1 to maintain an interaction with ppRB while other E2Fs are released. The ‘specific’ site is necessary and sufficient to mediate the interaction with ppRB. Furthermore, it reveals that the ‘general’ and ‘specific’ sites are regulated independently from one another, as the ‘general’ binding site is regulated by CDK phosphorylation while the ‘specific’ binding site is resistant. Therefore the ‘specific’ site provides a means for regulating E2F1 independently of cell cycle position.
Unique structural elements of pRB and E2F1 mediate the ‘specific’ interaction and the ppRB–E2F1 complex
The ‘specific’ site has been previously localized to the C-terminal domain of pRB and the marked box domain of E2F1 ; however, there is still little understanding of the structural basis for E2F1's unique interaction with pRB at this site. The critical role of the ‘specific’ site in mediating the complex between ppRB and E2F1 underscores the importance of understanding the structural basis for the interaction between the C-terminus of pRB and E2F1 and motivated us to investigate it in more detail.
pRB is a member of a family of proteins termed the pocket proteins, which share a well-conserved pocket domain. Interestingly, while the C-terminal region contributes to E2F binding by all pocket proteins, there is little conservation between the C-terminal domains of pRB and its other family members (Figure 5A). Furthermore, the p107 and p130 proteins share common sequence elements that are distinct from pRB suggesting that this region differentiates pRB from its two closest relatives (Figure 5A). To investigate the role of the C-terminus of pocket proteins in E2F1 binding, we generated GST–C-terminal constructs of p107 and p130 termed p107-C and p130-C. These constructs consist of the polypeptides that are aligned in Figure 5(A). They are sequences from p107 and p130 that begin immediately C-terminal to the small pocket domain and extend to the C-terminus. These recombinant proteins were incubated with extracts containing HA–E2F1 and HA–DP1, and complexes were precipitated with glutathione–Sepharose. Only GST-RB-C is capable of precipitating HA–E2F1 complexes, as shown by Figure 5(B). This suggests that the ‘specific’ site is a unique feature of pRB that differentiates its interaction with E2F1 from other pocket protein–E2F1 interactions.
Characterization of critical structural contacts that define the ‘specific’ site
Crystallographic data has described the interaction between a fragment of the C-terminus of pRB and the marked box domain of E2F1 and DP1 . Since these regions have been mapped previously as the site of interaction for the ‘specific’ site , we designed experiments to investigate if its structural features contribute to the ‘specific’ E2F1–pRB interaction. This crystal structure shows how a small pRB fragment interacts with a hydrophobic cleft that is formed by the marked box domain of E2F1 and DP1 (Figure 5C). We used a computational alanine residue scanning mutagenesis approach to identify the critical interaction sites between pRB and E2F1. The residues identified were largely hydrophobic and include Ile831, Leu832, Val833, Ile835, Phe839, Phe845, Ile848, Asn849, Met851 and Val852. Most of these amino acids form contacts found on the α-helix, the loop and β-strand portions of pRB in this structure (Figure 5C). They contact a patch of hydrophobic residues at the E2F1–DP1 interface (Figure 5C). To test the importance of these residues in maintaining the interaction between pRB and the marked box domain of E2F1, RB-C constructs were generated with substitution of the predicted contacts to an alanine residue. As is shown by Figure 5(D), these amino acids appear to have a critical role in maintaining the interaction with E2F1 as mutation of any of them is sufficient to disrupt the interaction with E2F1–DP1 in GST pull-down experiments. To ensure the integrity of the RB-C proteins, their interaction with Myc–PP1 was characterized. The recombinant RB-C mutants were incubated with PP1, which is known to interact with this region of pRB , and the ability of the mutants to precipitate PP1 suggests that these are specifically defective for binding to E2F1 (Figure 5D). This indicates that these residues are essential components of the ‘specific’ E2F1-binding site in the C-terminus of pRB.
The ‘specific’ site of pRB interacts with a region of E2F1 known as the marked box. This region is the site of multiple protein–protein interactions and is well-conserved between distinct E2F family members (Figure 6A). The conservation of this region does not correlate with the selective interaction between the marked box domain of E2F1 and the ‘specific’ site. This region, shown in Figure 6(A), is conserved largely between E2F1, E2F2 and E2F3 with a few exceptions. Of particular interest is Val276 in E2F1. The valine residue at position 276 of E2F1 is conserved in closely related mammals such as human, mice and rats, but in E2F2 and E2F3 this residue is conserved strictly as a proline (Figure 6A). Val276 localizes to the distal end of the β-sheet of E2F1 that is in closest proximity to the co-crystallized pRB fragment shown in Figure 6(B). The marked box domain in E2F2 or E2F3, which contains the conformationally restricted proline residue at this site, probably adopts a distinct structural conformation compared with E2F1 in this region. For this reason, we expect it to be incompatible for interacting with the ‘specific’ site of pRB as depicted by this crystal structure.
Identification of a structural element in E2F1 that defines selectivity for the ‘specific’ site
To investigate the effects of a V276P substitution on the interaction with the ‘specific’ site of pRB, an HA-tagged E2F1-V276P was expressed in C33A cells. Figure 6(C) shows that the V276P mutant maintains the interaction with RBLP, presumably through the ‘general’ site. However, this mutant is unable to interact with RB-C, which measures interactions at the ‘specific’ site. This suggests that the substitution does not disrupt the overall fold of E2F1 or the interaction with DP1, but selectively disrupts the interaction with the ‘specific’ site of pRB. Furthermore, substitution of V276A does not disrupt the interaction with RB-C, suggesting an important role for proline in determining compatibility for binding to the ‘specific’ site. The V276P-E2F1 mutant was transfected into C33A cells along with full-length WT-pRB or ΔG-pRB to further characterize the interaction of V276P-E2F1 with the ‘general’ and ‘specific’ binding sites. As shown by Figure 6(D), the V276P substitution does not disrupt the interaction with WT-pRB suggesting that the overall integrity of E2F1 is maintained. The V276P substitution however, leads to a partial disruption in the interaction with ΔG-pRB, which only contains the ‘specific site’ (Figure 6D). The remaining binding to full-length ΔG-pRB is probably mediated by contact sites or structural features that exist outside of the C-terminal domain of pRB. Taken together this suggests that the introduction of the proline residue in the marked box domain of E2F1 is sufficient to disrupt the interaction with the ‘specific site’.
As shown previously, E2F3 is unable to interact with the ‘specific’ site found in the C-terminal domain of pRB (Figure 3A). Strikingly, substitution of the analogous proline residue (Pro329) to a valine residue in E2F3 results in a gain of interaction with RB-C (Figure 6C). Substitution of Pro329 to an alanine residue in E2F3 is also sufficient to allow E2F3 to interact with RB-C. This suggests that the inability of E2F3 to interact with the ‘specific’ site is due in part to Pro329 and its effect on this region of the marked box domain. To validate further the ability of the proline residue in the marked box domain of E2Fs to prevent interactions with the ‘specific’ site, the ΔG mutant was employed to selectively disrupt the ‘general’ site in order to study the interaction with the ‘specific’ site in isolation. As is shown by Figure 6(D), transfected HA–E2F3 is unable to interact with ΔG-pRB as it is unable to bind to the ‘specific’ site. Once again, the substitution P329V is sufficient to mediate the interaction with the ‘specific’ site of pRB, as the mutant protein is able to interact with both WT- and ΔG-pRB. This confirms further that the presence of a proline residue in the marked box domain creates a distinct conformation in E2F3 that prevents the interaction with the ‘specific’ site in pRB.
Given the requirement of the ‘specific’ site for the complex between ppRB and E2F1, the possibility that the Pro329 in E2F3 functioned to block the interaction with ppRB was investigated. In a similar manner to previous experiments, WT-E2F3 is unable to interact with ppRB and only the residual hypophosphorylated species is precipitated by HA–E2F3 (Figure 6E). However, the substitution P329V in E2F3 results in an enhanced interaction with ppRB as it is capable of immunoprecipitating pRB phosphorylated at Ser807 and/or Ser811 that is shifted partially in migration (Figure 6E). This confirms further that the interaction with ppRB requires the ability to bind to the ‘specific’ site of pRB and this is blocked by the presence of a proline residue in the marked box domain of E2F3.
Taken together this provides novel insight into the mechanism by which E2F1 is capable of forming a unique interaction with the C-terminus of pRB. The ‘specific’ site contains hydrophobic contact sites that interact with a cleft that is formed by both E2F1 and DP1. The specificity of this site for E2F1 is imparted by Val276, since this amino acid is a proline residue in other E2Fs. The proline residue may create a confirmation in the marked box in these E2Fs that does not interact with the ‘specific’ site. This proline residue in E2F3 also prevents the interaction with ppRB, thus supporting the importance of the ‘specific’ site in forming the E2F1–ppRB complex.
The ‘specific’ site maintains the ability to regulate the transcriptional activity of E2F1
The ‘specific’ site is resistant to disruption by phosphorylation and provides a means for pRB to interact selectively with E2F1 beyond the G1-phase of the cell cycle. We sought to investigate the function of these complexes by testing the ability of the ‘specific’ site to regulate E2F-dependent transcription. Saos-2 cells were transfected with HA–E2Fs (and DP1) and a plasmid encoding luciferase under the control of a canonical E2F response element (pE2F4B-Luc). As shown in Figures 7(A) and 7(B), HA–E2F2 and HA–E2F3 can both function as potent activators to stimulate transcription of luciferase from a reporter containing an E2F response element. Co-transfection of increasing amounts of WT-pRB results in a dose-dependent decrease in this activity (Figures 7A and 7B). This indicates that the WT-pRB is capable of repressing the transcriptional activity of E2F2 and E2F3. In contrast, the ΔG mutant is unable to repress the transcription of E2F2 or E2F3 even at the highest level of expression (Figures 7A and 7B), which is consistent with the inability of ΔG to interact with E2F2 or E2F3 (Figure 3D).
Transcriptional regulation of E2Fs by pRB through the ‘general’ and ‘specific’ E2F-binding sites
In a similar experiment, WT-pRB was shown to regulate the transcriptional activity of E2F1 (Figure 7C). However, ΔG-pRB is also capable of regulating the activity of E2F1, albeit to a lesser extent than the WT-pRB as only the highest expression levels of ΔG-pRB affect transcription (Figure 7C). This agrees with the ability of the ‘specific’ site to maintain the interaction with E2F1 in ΔG-pRB (Figure 3D). To test the possibility that the ‘specific’ complex between pRB and E2F1 may regulate selective target genes, a luciferase construct containing the p73 promoter was utilized (p73–Luc). p73 is a well-studied target of E2F1 and activation of E2F1 by DNA damage has been shown to enhance the interaction of E2F1 with this promoter [25,31]. Strikingly, ΔG-pRB was found to regulate the activation of the p73 promoter to a similar extent as WT-pRB (Figure 7D). This suggests that pRB–E2F1 complexes formed through the ‘specific’ site have the ability to regulate the expression of particular E2F1 target genes. This implies that ppRB–E2F1 complexes present in S-phase, or later in the cell cycle, are capable of negatively regulating E2F1-specific transcriptional targets.
Contrary to current understanding of pRB–E2F regulation, the present study suggests that ppRB can maintain an interaction with E2F1. The present study data indicates that E2F1, but not E2F3, is capable of forming an interaction with ppRB, and that this interaction is dependent on the ‘specific’ site of pRB. This suggests a potential mechanism by which phosphorylation can regulate independently the interaction between pRB and distinct E2F proteins (shown in Figure 8). In its hypophosphorylated state, depicted in Figure 8(A), pRB is capable of interacting with E2Fs using the ‘general’ site or the ‘specific’ site. Phosphorylation of pRB by CDK complexes results in well-described structural changes that disrupt E2F binding to the ‘general’ site [9,10,14,42,45]. The present study data, however, indicate that CDK phosphorylation does not disrupt the ‘specific’ site found in the C-terminus of pRB, and as a result, ppRB is capable of maintaining an interaction with E2F1. While previous studies have suggested the existence of E2F1 complexes with phosphorylated pRB, the mechanism by which phosphorylation of pRB could both disrupt interactions with some E2Fs and maintain interactions with E2F1 has been unknown. The present study data provide a mechanism that explains the ability of pRB to be phosphorylated on most CDK-directed sites while disrupting only a portion of pRB–E2F complexes.
Regulation of pRB–E2F interactions by phosphorylation
The present study describes the mechanistic basis for the unique ability of E2F1 to interact with ppRB. The ‘specific’ site of pRB, which is required for the interaction between ppRB and E2F1, provides the observed selectivity for E2F1 complexes. The selectivity of the ‘specific’ site is mediated in part by a valine residue at position 276 in E2F1. All other mammalian E2Fs, and also E2Fs from multiple lower organisms, contain a proline residue at the analogous position to Val276 in E2F1. This suggests that the ancestral E2F protein contained a proline residue at this position. During the divergence of E2F1 from the ancestral E2F proteins, it is probable that the P276V substitution occurred and this was key to introducing a new interaction site between E2F1 and pRB. This sequence difference provides the means by which E2F1 can be regulated distinctly from other E2Fs, and may have further promoted the divergence of E2F1 function from other E2Fs.
E2F1 has a unique role in the induction of apoptosis that is not observed in other E2F transcription factors. This ability has raised the question as to how E2F1-induced apoptosis is attenuated in normal proliferating cells. Traditional models of pRB–E2F regulation suggest that all E2Fs are released from pRB following its phosphorylation at the G1–S transition. The ability of cells to maintain viability as they proceed through the cell cycle suggests that additional mechanisms exist to inhibit the pro-apoptotic potential of E2F1. The present study has refined this model such that the roles of the two distinct E2F-binding sites are included. As shown in Figure 8, the ‘general’ site is disrupted by CDK phosphorylation, but ppRB maintains the ability to interact with E2F1 through the ‘specific’ site. Thereby, phosphorylation of pRB results in the release of E2Fs from the ‘general’ site, which drive the expression of cell cycle genes, but maintains the interaction of E2F1 with the ‘specific’ site, which is capable of regulating E2F1-specific target genes.
While the ‘specific’ site is capable of regulating the transcriptional activity of E2F1, this complex has a relatively low affinity for the canonical E2F DNA-response element . The ability of the ‘specific’ site to effectively control the activity of the p73–Luc promoter suggests that the ‘specific’ site may target pRB–E2F1 complexes to distinct regions of the genome. This is supported by previous work that identified targets for pRB and E2F1 in the S-phase of the cell cycle . Cells synchronized in the S-phase contained largely ppRB that could be immunoprecipitated on DNA with phospho-specific antibodies. Taken with the results from the present study, this suggests that the observed complexes consisted of E2F1 bound to the ‘specific’ site of ppRB. Interestingly the targets of pRB in the S-phase were not observed in other stages of the cell cycle, suggesting that ppRB was capable of localizing to a distinct set of cellular genes . Thereby, phosphorylation of pRB results in complex formation between ppRB and E2F1 through the ‘specific’ site that may impart an altered DNA binding specificity and the regulation of distinct cellular targets.
Given that phosphorylation of pRB largely abrogates E2F binding and blocks the formation of pRB complexes at E2F target genes, few studies have investigated the ability of ppRB to maintain interaction with chromatin-remodelling factors. The ability of ppRB to maintain the interaction with E2F1 described in the present study raises important questions regarding the ability of ppRB–E2F1 complexes to recruit chromatin remodelling factors to E2F1 target genes. Recent work has shown phosphorylated pRB at apoptotic promoters including p73, in response to DNA damage along with E2F1 and the histone acetyl transferase P/CAF [p300/CREB (cAMP-response-element-binding protein)-binding protein-associated factor] . Interestingly the histone deacetylase HDAC1, which is commonly found associated with pRB on cell cycle promoters, is absent from the p73 promoter . This suggests that the phosphorylation state of pRB may allow for the recruitment of distinct chromatin remodelling complexes to E2F target genes. Furthermore, given the distinct interaction surfaces used in the ‘general’ and ‘specific’ complexes, it is possible that these two complexes are capable of associating with different chromatin remodelling factors to give rise to the observed selectivity. Lastly, the ‘specific’ complex containing ppRB and E2F1 may serve as a platform on which to assemble activating or repressive complexes depending on growth status or other cell signals.
The majority of human cancers express functional, but inactivated, pRB that is maintained in a hyperphosphorylated state. The ability of ppRB to interact with E2F1 suggests that tumorigenesis may select for cells that maintain pRB in a hyperphosphorylated state as a means to attenuate E2F1-induced apoptosis while simultaneously deregulating proliferation. In some cases, the depletion of pRB in cells that express predominately ppRB results in cell death , suggesting that the therapeutic disruption of the ‘specific’ site may provide a means to induce apoptosis in cancer cells expressing predominately ppRB. In contrast with the retention of WT-pRB, the majority of human tumours inactivate directly p53 to block the induction of apoptosis . The ability of p53-deficient cancer cells to undergo apoptosis is mediated largely by the p53 homologue p73, which is activated strongly by E2F1 . This further highlights the therapeutic potential of the ‘specific’ site, as it could be utilized as a robust means to sensitize cancer cells to p73-dependent apoptosis. Taken together the present work advances our understanding of the regulation of pRB–E2F interactions by CDK phosphorylation and suggests a selective advantage for retention of WT-pRB during tumorigenesis.
dominant negative CDK2
gel shift extract
large pocket domain
C-terminal domain only
The experiments were designed by Matthew Cecchini and Frederick Dick and performed in their entirety by Matthew Cecchini. Matthew Cecchini and Frederick Dick wrote the manuscript.
We thank numerous colleagues for advice during the course of this research and S. Francis for her critical review of the manuscript prior to submission. M.J.C is a member of the CIHR-Strategic Training Program in Cancer Research and Technology Transfer. F.A.D. was a Research Scientist of the NCIC/CCS during this study.
This work was supported by an operating grant from the Canadian Institutes of Health Research [grant number MOP-89765] that was awarded to F.A.D. M.J.C. acknowledges support from a CIHR MD/PhD studentship.