Introduction

In recent years we have come a long way in discerning the breakdown of normal cell mechanisms that results in tumour formation. Malignant transformation often involves combinations of mutations which, by necessity, are tissue and context specific, giving rise to a hugely disparate set of diseases.1 However, one property that all cancers share is the ability to proliferate beyond the normal limits and constraints. To do this the cell must acquire new properties, such as being able to deregulate cell proliferation and suppress cell death.2

The survival of long-lived multicellular organisms is dependent on allowing proliferation of their cells when needed, while at the same time suppressing the deregulated growth of mutated cells. Normal somatic cells are totally dependent on continuous receipt of appropriate mitogenic signals for their proliferation. There is also a network of inhibitory factors that serve to check the proliferative response to mitogens and must be overcome for cell cycle entry and proliferation.3 If appropriate signals are not received when cell cycle progression is stimulated, a stress response ensues driving the cell towards programmed cell death (apoptosis).4 A key factor controlling these processes is a nuclear protein called E2F-1.

E2Fs in Cell Cycle Control

E2F-1 belongs to the E2F family of DNA-binding proteins (E2F-1 to E2F-6), which are central regulators of cell cycle progression.5,6 These proteins function as heterodimers with members of the DP family (DP1 and DP2), with the DNA-binding specificity being determined by the E2F subunit. The E2F family regulates overlapping sets of target genes and all contain related DNA binding and dimerisation domains. All the members of the family, except E2F-6, also contain a transactivation domain. Based on structure, transcriptional properties and association with pocket proteins, the E2F family can be divided into three distinct groups. E2F-1, E2F-2 and E2F-3 associate preferentially with pRb (a product of the retinoblastoma susceptibility gene), and are potent transcriptional activators.7 E2F-4, which associates with pRb, and its related ‘pocket’ proteins p107 and p130, and E2F-5 that associates with p130, seem to be primarily involved in the active repression of E2F-responsive genes.7 E2F-6 does not interact with pocket proteins and functions as a negative regulator of E2F-dependent transcription via complexing with chromatin modifiers.8,9,10

In untransformed cells, the ability of pRb to bind to E2F is regulated by its cell-cycle-dependent phosphorylation.11 pRb is unphosphorylated during the G0 and early G1 stages of the cell cycle, and this form binds and inhibits E2F (Figure 1). Mitogenic growth factors induce the sequential activation of the cyclin-dependent kinase (cdk) complexes, cdk4/cdk6–cyclin D and cdk2–cyclin E, which then phosphorylate pRB and causes it to become dissociated from E2F.12,13 The resultant activation of E2F-responsive genes (e.g. those involved in DNA synthesis, cell cycle control, pocket protein expression, etc.) in late G1 seems to be sufficient to commit the cells to initiate DNA replication.14 Conversely, many growth-inhibitory signals such as those from the TGFβ family and from the p53/p21 checkpoint pathway mediate their effects by blocking phosphorylation of pRb. In this way, pRb monitors both positive and negative growth signals and determines if the cell should divide.15,16

Figure 1
figure 1

pRb binds to an E2F–DP complex in cells in the G0/G1 stage of the cell cycle. This leads to repression of E2F-responsive genes, firstly by inhibiting E2F from activating transcription by binding its transactivation domain, and secondly by active repression where it recruits factors such as HDAC, which modifies histone tails and therefore facilitates nucleosome packaging. Cdk4/6-cyclin D and cdk 2–cyclin E phosphorylate pRb and cause it to release E2F, activating E2F-responsive genes

Full size image

In its unphosphorylated state, pRb can regulate E2F-responsive genes through two distinct mechanisms. First, pRb binds to an 18 amino-acid motif within the transactivation domain of E2F, blocking the ability of E2F to recruit the basic transcriptional machinery.17,18 E2F in this context is considered a ‘passive’ repressor as it can occupy E2F DNA-binding sites, but cannot activate gene expression. Second, the pRb–E2F complex, while still bound to DNA, recruits various factors, for example, histone deacetylases (HDACs), SWI/SNF, Polycomb group proteins and histone methyl transferase (SUV39H10), which are able to switch off transcription and as a result effect ‘active’ repression (Figure 1).10,19,20,21 The relative importance of each of these transcriptional mechanisms to the functions of E2Fs is as yet not completely defined and is an area that is duly receiving intense investigation.

A Role in Cancer

E2F was originally discovered as a cellular component that is required for the early region transforming protein (E1A) of adenovirus to mediate transcriptional activation of the viral E2 promoter.22 Subsequent studies have shown that E2F controls the transcription of cellular genes that are essential for cell division, such as enzymes involved in the biosynthesis of nucleotides.23 In the 1990s, Nevins et al24 deduced how E2F is regulated in normal cells by determining the mechanism of E2F activation by E1A. E1A caused a cellular protein to dissociate from E2F, which led them to show that E2F is inhibited by its association with Rb. This early observation that E2F was deregulated by a transforming virus was the first indication that E2F may be associated with cancer. It was subsequently found that Rb is also targeted by other viral oncoproteins, including SV40 large T antigen and E7 proteins from ‘high-risk’ human papilloma viruses.25 It is now considered that deregulation of E2F is an event in most, if not all, cancers. As well as viral infection, this can occur by loss or mutation of Rb, or more often through the upregulation of the cdk/cyclin complexes that phosphorylate pRb or through loss of the cdk inhibitor, p16. Surprisingly, however, E2F itself is rarely found to be mutated.26,27

Oncogene or Tumour Suppressor?

In light of their frequent deregulation, what is the role of E2Fs in tumour development? Studies on the best characterised E2F, E2F-1, have indicated that it may have a unique role compared to other E2Fs, showing characteristics of both an oncogene and a tumour suppressor.28 Several lines of evidence suggest that E2F-1 has the potential to function at an oncogene, promoting the proliferation of cells beyond their normal constraints.29 It is an important part of the circuitry that commits cells to progression through the G1 phase, after which the cell is committed to complete the rest of the cell cycle. As eluded to earlier, many genes that are regulated in a cell-specific manner have E2F-binding sites as their promoters, and in some cases E2F-1 has been demonstrated to induce their expression directly. Some of these gene products play a direct regulatory role in the cell cycle, for example, Cdc2, cdc25a and cyclin E.30,31,32 As well as this, forced expression of E2F-1, as with E2F-2 and E2F-3, in quiescent cells is sufficient to induce entry into DNA synthesis, and each of these E2Fs can function as oncogenes in transforming assays.33,34 It was amazing, therefore, to find that targeted deletion of the E2F-1 gene in mice resulted in animals that spontaneously developed tumours in a number of tissues.35 Although a surprise, this observation gave credence to previous studies that had indicated a role for E2F-1 in pro-grammed cell death. Enforced expression of E2F-1 in vitro had been shown not only to cause cell cycle progression but also cause apoptosis in a number of cell types (Figure 2).36,37 In addition, apoptosis as a result of Rb deletion in mice was shown, by the generation of ‘double’ knockout mice (Rb−/−, E2F-1−/−), to be dependent on E2F-1.38

Figure 2
figure 2

As a result of its upstream activation in tumours and its ability to induce both cell cycle progression and apoptosis, E2F-1 can have both oncogenic- and tumour-suppressive effects. Green factors: upregulated in cancer, blue factors: lost or mutated in cancer

Full size image

For some time, this apoptotic activity of E2F-1 was thought to be similar to that described for another cancer-related protein, c-Myc.39 The elevation of c-Myc occurs in many tumours resulting in potent growth promotion.40 This effect of c-Myc can, however, only occur if the cell is also receiving appropriate survival signals, for example, IGF-1.41 If not, deregulation of c-Myc will cause programmed cell death.42 These findings resulted in the ‘conflict of signals’ model for oncogene activation, indicating that a safeguard mechanism exists within cells to protect against inappropriate growth promotion. This model, however, does not completely hold true for E2F-1 as mutants of E2F-1 have been described, which while unable to promote cell cycle progression, retain the ability to induce programmed cell death.43,44 As a result, interest in the apoptotic function of E2F-1 has resurged and numerous recent studies have provided insights into its death-promoting activities.

Multiple Roads to Cell Death from E2F-1

E2F-1 is capable of inducing apoptosis via several mechanisms (Figure 3). Both overexpression experiments and mutant mouse models of E2F-1 have shown that apoptosis can occur by mechanisms either dependent or independent of the tumour suppressor p53. The p53-dependent mechanism involves transactivation of the p14ARF protein by E2F-1. This can occur either directly through the transactivation of an E2F-binding site in the ARF promoter – an effect that can be augmented by oncogenic Ras, or indirectly through the activation of death-associated protein (DAP) kinase.45,46,47,48 p53 regulation occurs largely at the level of protein stability and in normal unstressed cells, p53 is kept at a low level via a negative feedback loop in which p53 induces the transcription of HDM-2 (MDM2 in mice), which in turn binds to p53 and mediates its degradation.49,50 E2F-1 can stabilise p53 via the induction of the p14ARF protein, which functions by binding directly to HDM-2 and preventing its degradation of p53.51,52 As a result, depending on other signals being received by the cell at that time, this can result in E2F-1 directed, p53-dependent cell death that, to a point, represents death by a ‘conflict of signals’ mechanism.

Figure 3
figure 3

Routes through which E2F-1 can induce apoptosis. E2F-1 can induce cell death by activating the p53 family member p73, or by stabilising the levels of p53 by transactivating the ARF tumour suppressor, with both pathways leading to the activation of caspase 9. In addition, E2F-1 can induce cell death by talking to components of the cell machinery (i.e. Apaf-1 and caspase 7) directly, or by suppressing antiapoptotic signals such as the activation of NF-κB

Full size image

In cell types lacking p53, for example, Saos-2 oesteosarcoma cells, E2F-1 has still been shown to be an effective death inducer. In this situation, the p53 family member, p73, has been shown to play a role in E2F-1-induced cell death.53,54 E2F-1 directly activates the transcription of p73 leading to the activation of its target genes (some of which are shared with p53) and apoptosis.55 p73 is also able to bind to MDM-2, but is not degraded and so cannot be stabilised by ARF expression.56 This provides a direct mechanism for E2F-1-induced cell death, and although p73 is not frequently mutated in human tumours, methylation-dependent silencing has been reported in haematological malignancies.57 Moreover, it has recently been shown that several tumour-derived mutants of p53 that are often retained in cancer are able to bind and inactivate p73, thereby circumventing the necessity to mutate p73 per se.58 This process reduces chemosensitivity and adds more weight to the fact that E2F-1-induced apoptosis is an important tumour-suppressive mechanism as well a prognostic indicator of therapeutic success.59,60

Another mechanism of E2F-1-induced apoptosis is via the inhibition of antiapoptotic signalling. The two main apoptotic pathways in the cell have been defined as the mitochondrial pathway and the death receptor pathway.61 The death receptor pathway activates a caspase cascade in response to external ligands activating the members of the tumour necrosis factor receptor super-family. These receptors have conserved protein–protein binding domains termed death domains, and recruit procaspases such as caspase 8. Cleavage of these procaspases to active forms leads to the subsequent activation of effector caspases and apoptosis. Ligand binding does not always end in cell death, since some of the receptors also activate caspase independent signalling pathways that block apoptosis. The NF-κB family of transcription factors regulate apoptosis in response to many stimuli.62 The activation of NF-κB can lead to tumour cell proliferation, invasion, angiogenesis and metastasis, therefore suppression of NF-κB in tumours may provide an additional target for prevention of cancer.63 In most cases, NF-κB functions as a survival signal, for example, the activation of TNFR results in the activation of NF-κB via Traf2, which contributes to the inhibition of cell death.64 E2F-1 can downregulate Traf2 protein levels and therefore inhibit the activation of antiapoptotic signals, such as NF-κB in response to TNFα.65 It is interesting to note that this effect, similar to E2F-1's ability to inhibit transformation, does not require E2F-1's transactivation domain.65 The extent, however, to which this mechanism is involved in tumour suppression by E2F-1 is very interesting, but has yet to be determined.

In contrast to the role of NF-κB following TNFα treatment, during cell death induced by p53, NF-κB has a surprising proapoptotic role.66 It has been suggested that NF-κB may be a good chemotherapeutic target; however, the inhibition of NF-κB following treatment with chemotherapeutic drugs would affect apoptosis differently depending on the p53 status of the cell – causing more cell death in p53 null or defective cells and less death in wild-type p53 cells. Dominant-negative forms of NF-κB-inducing kinase (NIK) have been shown to enhance TNFα-induced apoptosis in p53-inducible Saos cells where there is no expression of p53, but have no effect on p53-induced death in the same cells.66 This shows that p53 and TNFα utilize distinct pathways to activate NF-κB and presents possible targets for therapeutic intervention depending on the p53 status of the cells to be treated. Interestingly, NIK has recently been shown to be a transcriptional target of E2F-1, although the precise effect this may have on apoptosis and in particular the regulation of Traf-2 by E2F-1 is yet to be ascertained.67 E2F-1 has also been shown to activate the expression of another kinase, PKR, leading to phosphorylation of its downstream target, the translation initiation factor, eIF-2α and subsequent apoptosis.68 Although this factor has previously been reported to interact with NF-κB, its involvement in cell death induced by E2F-1 appears to be independent of NF-κB as well as p53 and p73. It is interesting to speculate that, since translation is known to be regulated by the insulin signalling pathway, whether factors in this pathway, for instance, mTOR are involved in regulating E2F-1 and E2F-1-induced programmed cell death.69,70 It has to be kept in mind though that even once the involvement with E2F-1 is elucidated, these signalling pathways affect many factors and how they all integrate to determine the fate of the cell is undoubtedly going to be complex and situation dependent.

Perhaps more simple to understand are a number of reports that have recently implicated E2F-1 as being a regulator of factors intrinsic to the apoptotic process, for example, apoptosis protein-activating factor (Apaf-1).71,72 When induced, Apaf-1 assembles with cytochrome c, a mitochondrial signal released on receipt of apoptotic signals, and activates caspase 9 leading to the activation of downstream effector caspases eventually leading to apoptosis (Figure 3).73 Although the death receptor and the Apaf pathways can be thought of as distinct, there is accumulating evidence that crosstalk occurs between all the different pathways, so the complex pattern of signal interaction must first be determined before the effects of individual signals can be seen.74

In addition to Apaf-1, DNA microarray studies have demonstrated that ectopic expression of E2F-1 can also upregulate the expression of several members of the caspase family.75 Interestingly, an antiapoptotic member of the bcl-2 family, Mcl-1, has also been found to be affected by E2F-1, but by transcriptional repression rather than activation.76 This repression does not require the transactivation domain of E2F-1 and therefore, together with the involvement of Traf2 regulation, provides some explanation of how E2F-1-induced cell death can occur without target gene activation.

Therapeutic Possibilities

The processes involved in tumour formation present many targets for cancer therapies. As deregulated cell proliferation and inhibition of apoptosis are two of the ‘hallmarks’ of cancer development,2 they present two obvious targets for therapeutic intervention. Many existing cancer drugs interfere with the basic machinery of DNA synthesis or metabolism and induce tumour cell killing by utilising apoptotic pathways.77 The fact that many of these pathways are inactivated during tumour development can lead to existing therapies being compromised through de novo drug resistance.78 Most of the proteins, however, which are inactivated in this way, for example, p53 loss or PTEN activation, are relatively ‘upstream’ components of these pathways and their inactivation leaves the ‘downstream’ apoptotic mechanism functionally intact. This is exemplified, for example, by the observation that spontaneously regressing tumours exhibit an increased frequency of programmed cell death.79 Therefore, since apoptotic programmes can be manipulated to produce changes in cell death, the genes and proteins controlling apoptosis are drug targets with great potential. Due to the ‘double-edged sword’ nature of E2F-1 activation, in many tumours the pathways that link E2F-1 to apoptosis have been interrupted, for example, through the loss of p53 in addition to presumably other as yet unknown mechanisms. This disrupts the balance of apoptosis and proliferation that usually occurs, allowing proliferation without ‘safeguard’ apoptotic mechanisms to prevent tumorigenesis. The factors therefore that suppress E2F-1-induced apoptosis, or the downstream targets in E2F-1-induced apoptosis, could be key targets for therapeutic intervention. As a result, it is an exciting prospect to consider that recent insights into E2F-1-mediated cell death could possibly prove to be of clinical benefit. Its important to note though that an apoptotic role for E2F-1 in cancer is not absolute, and that loss of E2F-1, for example, in Myc-mediated lymphomagenesis, does not necessarily lead to apoptosis resistance.80 In this situation, the inactivation of E2F-1 leads instead to the inhibition of the enhanced cell cycle progression caused by c-Myc, indicating that actually targeting E2F-1 itself may well be useful. The further investigation therefore of the role of E2F-1 in cell death and cell cycle progression, and the way in which it integrates with other factors during tumour development can only serve to enhance and refine the potential of targeting the E2F pathway for therapy design.

Future Prospects

The pathways involving E2F-1 and apoptosis are both multiple and interactive (Figure 3), and although many recent advances discussed here have shed much light on these processes, many questions still remain. An understating, for example, of how E2F-1 can induce cell death either without transcriptional activation or even through repression are two areas that as yet are only minimally studied and undoubtedly worthy of further investigation. In addition, what determines the choice of E2F-1 response following deregulation is also yet to be determined. A number of stress-responsive kinases, including ATM/ATR and Chk2, have recently been implicated in signalling to E2F.81,82 In line with a previous observation showing that E2F-1 is stress responsive, these kinases have been implicated in the upregulation of E2F-1 following cellular stress and in the regulation of E2F-1-mediated death.81,82,83 The tantalising prospect of being able to modulate an E2F response by targeting these signalling pathways ensures that it will be not be long before more information into this level of regulation will be presented.

Ultimately though, perhaps the biggest question that remains is what is the role of E2F-1 in human cancer? Although a genuine tumour-derived somatic cell mutation in E2F-1 has recently been described and the investigation of its activities could indeed prove rewarding, the lack of a panel of E2F-1 mutants makes discerning the role of E2F in tumour development more difficult to determine.84 While the continued study of E2F-1-induced cell death will provide an insight on this issue, it must ultimately be kept in mind that although E2F lies at the core of very significant cell fate decisions, it forms only part of a complex matrix of cell signalling events and interactions. For example, a recent study looking at the role of E2F-1 in the response of keratinocytes to UVB irradiation in mouse skin unexpectedly showed that E2F1−/− mice exhibit enhanced apoptosis following exposure to UVB when compared to wild-type counterparts.85 Moreover, when these mice were crossed with p53−/− mice, it was found that this effect prevailed, thereby reverting the apoptosis resistance caused by loss of p53 alone and indicating that E2F-1 must lie upstream of p53 in this response.85,86 Ultimately therefore, since this finding goes against so much of what we know about the role of E2F-1 in programmed cell death, it seems certain that the ‘life and death decisions’ by E2F-1 and whether it acts as an oncogene or tumour suppressor may well be context specific.