Main

Facts

  • p53 causes cell cycle arrest

  • p21/CDKN1A is required for indirect transcriptional repression by p53

  • The DREAM protein complex is a transcriptional repressor

  • CHR and E2F promoter elements bind the DREAM complex

  • p21/CDKN1A initiates a switch from activating B-MYB- and FOXM1-containing complexes to the repressing DREAM complex

  • p53 indirectly downregulates many cell cycle genes

Open Questions

  • How do p63, p73 and p53 variants influence the p21–DREAM–E2F/CHR (p53–DREAM) pathway?

  • Are cellular kinase inhibitors other than p21/CDKN1A regulating this pathway?

  • Which clinical benefits can be achieved in cancer treatment with small-molecule CDK inhibitors by compensating for defects in the p53–DREAM pathway?

  • What are the overlaps or differences in pRB and DREAM function?

Prologue

One central role of the tumor suppressor p53 is to arrest the cell cycle. p53 indirectly downregulates the expression of many genes which are essential for progression through the cell division cycle. The detailed mechanism of indirect transcriptional repression by p53 has only recently become clear. p53 employs a protein complex named DREAM to downregulate gene expression. DREAM functions as a transcriptional repressor complex. With the advent of genome-wide experimental and bioinformatic analyses, we are now in the position to assess the wide spectrum of genes controlled through the newly defined p53–DREAM pathway.

p53 Downregulates Expression of Cell Cycle Genes

p53 is at the heart of several fundamental cellular signaling pathways.1, 2, 3, 4 The most important of these pathways for p53’s tumor-suppressive role are induction of apoptosis and cell cycle arrest.5, 6

Cell cycle arrest can be achieved by depleting regulatory proteins required for cell cycle progression. Upon p53 activation, genes for many central cell cycle proteins are transcriptionally downregulated. Key examples for genes repressed after induction of p53 are cyclin A,7 polo-like kinase 1 (PLK1),7 cyclin B1,8, 9, 10 cyclin B2,10 cyclin-dependent kinase 1 (CDK1),11 CDC20,12 cell cycle phosphatases CDC25A13 and CDC25C,14 DNA replication licensing factor MCM5,7, 15 CKS116 and antiapoptotic Survivin (BIRC5).7 Even such a small selection of genes exemplifies that p53-dependent downregulation of expression affects many aspects of cell cycle regulation.

Transcriptional Repression by p53 is Indirect

Transcriptional regulation is essential to the function of p53 as a tumor suppressor.2 Interestingly, the number of genes downregulated after p53 activation (approximately 2700) is larger than the number of genes activated by p53 (approximately 2200).17 Before this enigma was finally solved, several mechanisms had been proposed to explain how p53 can serve as a transcriptional activator as well as a repressor.2, 4, 18 However, experimental data obtained for particular genes were often not consistent with the suggested mechanism or results published for certain genes by different groups were contradictory.17, 18, 19 Furthermore, different models for p53-dependent repression require direct binding of p53 to the downregulated gene. However, genome-wide mRNA expression and chromatin immunoprecipitation (ChIP) results demonstrated that this requirement is not fulfilled for most repressed genes. Only about 3% of the genes downregulated by p53 are also bound by p53.17 Thus, essentially all genes are downregulated by p53 indirectly.

Prior to the availability of genome-wide ChIP data on binding of p53 and other factors potentially involved in transcriptional repression, it was not evident by which mechanism p53 downregulates a plethora of cell cycle genes. This changed when the mammalian DREAM complex together with its target genes was discovered20, 21 and the observation was made that DREAM is formed following p53 induction.22

DREAM is a Transcriptional Repressor

The DREAM transcriptional complex displays two remarkable features. It changes its composition to exert opposing functions in gene regulation and it contains two subunits that bind to distinct DNA elements.

DREAM is composed of the MuvB core complex, E2F4-5/DP, and p130 or p107 proteins, which are related to the retinoblastoma tumor suppressor pRB20, 21 (Figure 1). E2F4, E2F5 and p130/p107 had long been implicated in transcriptional repression via E2F sites.23 Consistently, DREAM was initially identified as a complex which binds promoters through E2F sites.20, 21, 24 However, DREAM loses its E2F/pRB-related components to associate with the transcriptional activators B-MYB and FOXM1 during the cell cycle.20, 22, 25, 26, 27 Thus, these MuvB-based complexes cannot bind E2F sites. DREAM as all other MuvB-derived complexes binds DNA through cell cycle genes homology regions (CHRs).28, 29, 30 CHR transcriptional elements are distinct from E2F sites and are bound by the LIN54 component of MuvB31, 32 (Figure 1).

Figure 1
figure 1

Cell cycle and transcription factor complexes. The protein complexes binding to DNA change during the cell cycle. Gene expression is repressed in the early phases of the cell cycle and becomes activated during the later phases. For this change, E2F and CHR (cell cycle genes homology region) promoter elements switch from repressor to activator sites. In G0 and early G1 phase the DREAM complex binds E2F, CHR, CDE (cell cycle-dependent element), and CLE (CHR-like element) sites to repress transcription. In G2 phase and mitosis transcriptional repression is released and activation occurs via CHR sites. Only promoters with CHR sites can bind the MuvB-based complexes MMB (B-MYB-MuvB), FOXM1-MMB and FOXM1-MuvB. The MuvB core complex is composed of LIN9, LIN37, LIN52, LIN54 and RBBP4 proteins. LIN54 is the component which binds to CHR elements. For the switch from repressing to activating complexes, B-MYB and FOXM1 are recruited to the MuvB core when E2F4-5/DP and p107/p130 dissociate from the complex. B-MYB-MuvB (MMB), FOXM1-MMB and FOXM1-MuvB complexes serve as activators of late cell cycle genes which carry functional CHR elements. Early cell cycle genes with maximum expression in the S phase are activated by E2F1-3/DP heterodimers through E2F sites

Full size image

MuvB-based complexes can switch their function. Association of MuvB with B-MYB or FOXM1 switches DREAM to B-MYB-MuvB (MMB), FOXM1-MMB or FOXM1-MuvB complexes and turns the MuvB core from repressor to activator. This change in protein composition of MuvB-based complexes is connected to progression through the cell cycle and explains the switch from repression to activation via the same DNA site in the target promoters, that is, the CHR element (Figure 1).

It has been discussed whether B-MYB and FOXM1 require additional direct DNA binding when they are in a complex with MuvB.19, 26, 29, 31, 33, 34, 35, 36, 37 Generally, MYB consensus sites or forkhead binding sites are not observed close to the MuvB-binding CHR elements. For FOXM1 it was reported that it mostly binds to non-forkhead binding sites in the genome and that this nonspecific DNA binding may support association of MuvB with DNA.35, 36 Possibly, also B-MYB binds to sites far from CHR elements to augment MMB-LIN54 binding to DNA.

Recently, the importance of CHR sites in cancer signaling pathways yet again has been demonstrated when the computer software SWItchMiner (SWIM) was employed to search for crucial nodes in signaling networks – called switch genes – out of a large panel of cancer data sets from The Cancer Genome Atlas.38 The analysis yielded 100 significant switch genes which are mostly upregulated in a panel of different tumor types. With this selection of genes a de novo motif search for promoter elements was carried out. Interestingly, the CHR element emerged as a crucial site central to the regulation of the switch genes from the cancer signaling nodes.38

In addition to binding to single E2F or CHR sites, DREAM binding can be supported by two other elements, CDE (cell cycle-dependent element) and CLE (CHR-like element) sites (Figure 2). CLE sites are weak CHR-like elements and augment binding of DREAM to E2F sites. In general, affinity of CLE sites toward MuvB-based complexes, also the activating complexes, is not sufficient for binding. CLE sites alone cannot bind DREAM and an E2F element is required in tandem. Also, promoters require a spacer of four nucleotides between E2F and CLE sites.33 Similarly, CDE sites support binding of DREAM only when a CHR element is present in the promoter. Again, a spacer of four bases is found between CDE and CHR sites.33 CHR and CLE sites are contacted by LIN54 of the MuvB core complex.28, 32 Thus, DREAM binds to promoter DNA by four different modes33 (Figure 2).

Figure 2
figure 2

Modes of DREAM binding. DREAM can form two distinct contacts with DNA. It can bind to DNA via single E2F (a) or CHR (b) sites. E2F sites are contacted through E2F4-5/DP heterodimers. Distinct from this binding, contacts to CHR elements are made via the LIN54 protein. In the figure, the LIN54 component of the MuvB core complex is the only constituent that is labeled. Binding to E2F or CHR elements can be supported by CLE (c) or CDE (d) sites, respectively. CDE and CLE sites differ from E2F and CHR elements as CDE and CLE sites are unable to bind DREAM as single elements

Full size image

The p53–p21–DREAM–E2F/CHR Pathway

After the discovery of DREAM binding to E2F and CHR elements, the pathway by which p53 downregulates many genes became evident.39 In short, this pathway requires transcriptional upregulation of p21/CDKN1A. p21/CDKN1A inhibits cyclin-dependent kinases (CDKs) which phosphorylate the pRB-related proteins p107 and p130. Thus, p21/CDKN1A expression results in hypophosphorylation of p107 and p130. In this hypophosphorylated state, p107 and p130 can join other proteins to form the DREAM complex and thereby repress transcription through DREAM binding to E2F or CHR promoter sites (Figure 3).

Figure 3
figure 3

The p53–p21–DREAM–E2F/CHR pathway. Indirect p53-dependent repression through DREAM. Induction of p53 leads to downregulation of genes. This regulation is indirect as p53 does not bind to the regulated genes. Instead, induction of p21/CDKN1A expression by p53 causes hypophosphorylation of p107 and p130. Hypophosphorylation of these pRB-related pocket proteins facilitates DREAM formation. DREAM complexes then displace the activating complexes FOXM1–B-MYB-MuvB (FOXM1-MMB) and E2F1-3/DP on the target promoters. (In the figure, LIN54 is the only labeled MuvB component.) Overall, this switch causes previously activated genes to be indirectly downregulation by p53

Full size image

The CDK inhibitor p21/CDKN1A (WAF1, CIP1) was the first transcriptional target identified for p53.40, 41 And with this target the more detailed description of the pathway starts. Upon p53 activation, p21/CDKN1A is transcriptionally upregulated through direct binding of p53 to sites in the p21/CDKN1A promoter (Figure 3).39

One question that still needs to be addressed systemically is how the p53-related p63 and p73 protein families influence transcription of p21/CDKN1A. Especially the TAp63/TAp73 variants have similar functions in regulating gene transcription as p53.42, 43 The DNA binding motifs for p63, p73 and p53 are apparently essentially identical,44, 45, 46, 47 suggesting that the transcriptionally active members of the p63 and p73 families may contribute to cell cycle arrest through activating p21/CDKN1A.43, 48, 49 However, early experiments with overexpression of p63 and p73 variants indicated a reduced ability to induce p21/CDKN1A expression compared with p53 and showed only minor effects on genes which are downregulated by p53.13, 16, 48, 50, 51

Another challenge in delineating activation of p21/CDKN1A is the formation of hetero-tetramers between p53 isoforms and the various proteins of the p53/p63/p73 family.52, 53 In particular, tetramer formation including isoforms such as Δ40p53 and Δ133p53 may compromise activation of p21/CDKN1A by other p53 family members.43 Which combination of p53 isoforms and other p53/p63/p73 family members compete for binding sites in the p21/CDKN1A promoter depends on cell type and developmental stage-specific expression of these factors.

As the next step in the p53–DREAM pathway, p21/CDKN1A inhibits cyclin-dependent kinase complexes such as cyclin E/A-CDK2 and cyclin D-CDK4/6.54, 55 In turn, these cyclin/CDK complexes are no longer able to phosphorylate p107 and p130.56 The resulting hypophosphorylated p107 or p130 proteins attach to the MuvB core complex and shift the equilibrium from FOXM1-MMB to DREAM.22, 35, 39 Concomitant to this shift in MuvB-derived complex composition, transcriptional activation through FOXM1-MMB switches to repression by DREAM. Thus, genes active before p53 activation become repressed following p53 induction (Figure 3). At this step, the DREAM pathway shows a parallel regulation to the control by pRB because hypophosphorylation of pRB leads to pRB/E2F complex formation.57

p21/CDKN1A is most likely not the only protein which can inhibit cyclin/CDK complexes that can phosphorylate p107 and p130, thereby promoting DREAM formation.56 Other CDK inhibitor proteins can substitute for p21/CDKN1A to inhibit cyclin E/A-CDK2 and cyclin D-CDK4/6 combinations.

These inhibitors include p27/Kip1/CDKN1B and p57/Kip2/CDKN1C, both members of the Cip/Kip family with broad complex formation capacity, as well as p16/INK4A/CDKN2A, p15/INK4B/CDKN2B, p18/INK4C/CDKN2C and p19/INK4D/CDKN2D of the INK4 family with narrow binding specificity towards cyclin D-CDK4/6 complexes.58

Although the function of p21/CDKN1A in cell cycle checkpoint control and thus a possible role in tumor suppression has been confirmed many times, one observation that may be related to this possible cdk inhibitor redundancy is the absence of p21/CDKN1A mutations in tumors and the lack of spontaneous tumorigenesis in p21/Cdkn1a (−/−) mice.59, 60 Consistently, recent results from several knockout models show that loss of p21/CDKN1A function alone is not sufficient for tumor development.61, 62

Cyclin-dependent kinase regulation may even be more complex. Contrasting the canonical CDK inhibitor function, potential activation of cyclin/CDK complexes by p21/CDKN1A and p27/Kip1/CDKN1B has been discussed, with CDK inhibitors functioning as cyclin/CDK assembly factors, mediating nuclear localization of D-type cyclins, and contributing to stability of cyclin D-CDK4 complexes.58, 63, 64 Thus, it remains open whether additional signaling steps aside from the p53–p21/CDKN1A axis signal into the DREAM pathway.

Target gene selection by DREAM dictates the cellular response of indirect p53-mediated gene repression. Four types of binding represented by the two main classes of target genes with either E2F or CHR sites can be distinguished (Figure 2). Depending on the specific promoter of the gene, either E2F or CHR elements bind the complexes independently or with the support of CDE or CLE sites, respectively (Figures 2 and 3).

Before the p53-dependent switch to transcriptional repression, target genes are activated by two different mechanisms. E2F elements bind E2F1-3/DP proteins for activation, whereas promoters carrying CHR sites are activated by FOXM1-MMB. Both groups of promoter elements then switch to DREAM binding for repression (Figure 3).

Taken together, this sequence of reactions constitutes the p53–p21–DREAM–E2F/CHR or short the p53–DREAM pathway.39

Target Genes for Indirect p53-Dependent Repression

With the p53–DREAM pathway as a basis, criteria for identification of targets for indirect transcriptional downregulation by p53 are straightforward to define. Downregulation of target mRNA following p53 activation, DREAM binding to the target gene, and the presence of E2F or CHR sites in the proximal promoter are pivotal criteria for identification of target genes. There are many studies describing changes in mRNA levels employing a few different cell systems to compare expression with or without active p53.17, 65 Furthermore, the binding of DREAM components to these target genes can be assayed by ChIP. Subsequently, this information can be combined with the presence of E2F or CHR sites in the promoters. Of course, the quality of p53–DREAM target identification improves considerably the more results from independent studies are combined. We have employed bioinformatic tools to search for overlaps in a large number of reports on differential mRNA expression after p53 induction, on the binding of DREAM components by ChIP, and whether the potential target genes display E2F or CHR elements.17, 66 With a more recent analysis, the www.targetgenereg.org website was established. This site is updated with links to new data reports and allows retrieving results from genome-wide analyses easily.65, 67

Here, a compilation of p53–DREAM target genes is provided (Table 1). In order to obtain a catalog of high-confidence targets, criteria for inclusion as targets were binding of p130, E2F4, LIN9, LIN54, and the lack of binding by p53 as assayed by ChIP in combination with downregulation of target gene mRNA after activation of p53. The data for individual genes were retrieved from www.targetgenereg.org and several meta-analyses.17, 29, 31, 65, 66, 67 Although most of the p53–DREAM target genes were identified merely by such meta-analyses, several genes such as CCNB1, CDK1, CCNB2, KIF23, PLK4, BIRC5, CDC25C and PLK1 have already been found or confirmed in detailed experiments as targets of the p53–DREAM pathway.19, 22, 39, 68, 69 Nevertheless, meta-analyses of genome-wide studies bypass such experimental efforts for individual genes and yield more than 250 high-confidence p53–DREAM targets (Table 1).

Table 1 Genes regulated by the p53–DREAM pathway
Full size table

The compilation of p53–DREAM targets represents numerous cellular functions (Table 1 and Figure 4). The many protein classes found among the p53–DREAM targets are illustrated by examples such as kinases, protein chaperones, DNA helicases, ubiquitin ligases, phosphatases, methyltransferases, nucleases, ATPases and transcription factors (Table 1). Most gene products participate in cell cycle control. Examples for particular functions are DNA replication, nucleosome packaging, mitotic spindle assembly and chromosome segregation. Thus, it is becoming evident that the p53–DREAM pathway coordinately downregulates a plethora of genes which are categorized into functional groups (Figure 4).

Figure 4
figure 4

Cellular functions of the p53–p21–DREAM–E2F/CHR pathway. In order to summarize cellular functions regulated by the pathway, gene ontology terms for p53–p21–DREAM–E2F/CHR targets from Table 1 were compiled

Full size image

Checkpoint Control from DNA Synthesis to Cytokinesis

p53 can induce cell cycle arrest at several stages, including G1/S and G2/M checkpoints.1, 2, 3, 4 For example, it has been shown that p53 can induce G1 arrest via p21/CDKN1A-dependent inhibition of cyclin A/E-CDK2.55 Also progression through G2 phase and mitosis can be affected by p53, as several early studies showed that p53 is responsible for the downregulation of many genes important for checkpoint control from G1 through cytokinesis.70, 71, 72, 73 However, at the time it was not evident that such checkpoint control by p53 is based on a common mechanism17, 39, 66 (Figure 3). Now it is apparent that many proteins controlling cell cycle checkpoints are regulated by the p53–DREAM pathway and are clustered in functional groups (Table 1 and Figure 4).

Coordinated Transcriptional Repression by the p53–DREAM Pathway

A major feature of p53-dependent repression is that whole groups of functionally related genes are indirectly downregulated. Many such groups are defined by their function and timing of expression during the cell cycle. DREAM-dependent transcriptional repression employs binding to E2F or CHR sites as a determinant for early or late expression in the cell cycle, respectively.31, 33 Genes with maximum expression in the G1 and S phases are controlled through E2F or E2F/CLE sites and can be activated by E2F1-3/DP complexes, whereas genes expressed in the G2 phase and mitosis are upregulated by MMB and FOXM1-MuvB activator complexes through CHR or CDE/CHR elements (Figure 1).

G1/S Checkpoint Genes are Repressed by DREAM Binding to E2F Sites

One group of DREAM target genes important for the G1/S checkpoint is represented by POLA1, MCM2 and ORC174, 75 (Table 1). Furthermore, several DREAM targets, that is, cyclin A, CDK2, CDC6 and CDT1, are active in a checkpoint preventing rereplication (Table 1).73, 76 Interestingly, many genes previously described as classical E2F targets and hallmark genes for S phase progression such as TK1 and DHFR are now considered p53–DREAM targets77, 78 (Table 1). The E2F or E2F/CLE sites in their promoters are bound by DREAM for repression in resting cells and the E2F elements may bind activating E2F complexes at later stages of the cell cycle33 (Figure 1).

p53-Repressed Genes Required for the G2 Phase and Mitosis are Controlled by CHR Elements

In addition to controlling the G1/S checkpoint, p53 also has a role in regulating genes required for progression through G2 phase and mitosis.30, 66, 79 Cell cycle-dependent expression of these genes is controlled by CHR or CDE/CHR sites in their promoters31, 33 (Figure 1). Prominent examples for p53–DREAM-regulated genes involved in G2/M checkpoint control and progression through mitosis are CHEK2, CDK1, CCNB1, CCNB2 and CDC25C10, 14, 31, 39, 66, 80, 81 (Table 1).

In addition to such central regulators, also genes coding for proteins required in the mechanical execution of mitosis are controlled by the p53–DREAM pathway, such as kinesins.82 Of the many kinesins discovered in bioinformatic screens as p53–DREAM targets, KIF2C, KIF23 and KIF24 have been studied in detail and were validated to be controlled by DREAM29, 31, 68, 83 (Table 1).

The p53–DREAM Pathway and its Role in the Spindle Assembly Checkpoint, Chromosomal Instability, Aneuploidy in Cancer Cells and Mitotic Catastrophe

Several gene products mentioned above together with many additional cell cycle proteins are required for accurate segregation of chromosomes. Deregulation of their genes can perturb the spindle assembly checkpoint and lead to chromosomal instability (CIN).84, 85, 86, 87 CIN and resulting aneuploidy are considered hallmarks of cancer cells. Deregulated expression of mitosis genes has been shown in numerous studies to cause aneuploidy and tumor development.88 Importantly, many genes involved in chromosome segregation are p53–DREAM targets (Table 1). Similarly, several genes important for mitosis which are downregulated by the p53–DREAM pathway are part of the DNA damage response. Repression of these genes leads to perturbations in the mitotic machinery. As a consequence of depriving cells of these regulators, cells can arrest in mitosis and undergo the death program of mitotic catastrophe.89

As chromosomal missegregation causes elevated levels of p53 and p21/CDKN1A,90 the p53–DREAM pathway becomes activated and many genes required for segregation of chromosomes are downregulated (Table 1). The lack of expression of segregation regulators results in cell cycle arrest. In cells that have lost p53 or p21/CDKN1A function, the ability to arrest the cell cycle is compromised causing CIN and aneuploidy.90

Numerous gene products involved in mitotic spindle formation, kinetochore function, microtubule binding, centromere organization and centrosome formation such as CENP-A/C/E/F/L/N/O/W, CAF1A, MCM2-8, INCENP and CEP152/295 are implicated as p53–DREAM targets (Table 1). Furthermore, several genes involved in these processes – BIRC5 (Survivin), CEP55, PLK1, GAS2L3 and PRC1 – have been established as DREAM targets in detailed experimental studies.19, 91, 92, 93

As two examples, the histone H3-like CENP-A protein (CenH3) and its chaperone HJURP (Holliday junction recognition protein) have important functions in centromere formation. Their expression peaks in the G2 phase. CENP-A is incorporated into centromeric chromatin between telophase and early G1 to form centromere-specific nucleosomes and to facilitate kinetochore binding to the centromere.94 CENPA and HJURP genes had been predicted as targets repressed by the p53–DREAM pathway.66 Recently, it has been confirmed that these two factors are indeed downregulated indirectly by p53 requiring CDE/CHR sites in their promoters and a functional p21/CDKN1A CDK inhibitor.95 Consistently, expression of CENPA and HJURP mRNA was found increased in several tumor types which lack functional p53 compared with samples with wild-type p53. Notably, the report suggests that overexpression of CENPA and HJURP is not simply a consequence but may be one of the causes of cell cycle deregulation after p53 inactivation and cellular transformation. This assumption stems from the observation that mRNA levels of the G2/M genes CENPA and HJURP remain high even when a decreasing proportion of cells enters G2/M and an increasing proportion of cells undergoes apoptosis.95

Also in the context of preventing supernumerary centrosomes, the formation of the PIDDosome from its components together with its regulatory effect on p53 displays a balancing network of feedback loops. The PIDDosome via Caspase-2 mediates MDM2 cleavage leading to p53 stabilization and p21/CDKN1A activation.96 While expression of the PIDDosome constituent PIDD1 is strongly induced, expression of another component, CRADD (RAIDD), is not significantly affected by p53.17 In contrast, the CASP2 (Caspase-2) component is clearly downregulated, possibly via p53–DREAM.17

Furthermore, it has been shown that loss of p53 causes centrosome amplification.97 Particularly overexpression of cell cycle regulators such as PLK4, which is also downregulated by the p53–DREAM pathway, was reported to be central to the amplification of centrosomes.69, 98 More importantly, overexpression of these genes was implicated not as a consequence but rather as a cause contributing to the formation of tumors92, 98

As a result, deregulation of p53 cell cycle targets leads to centrosome amplification which promotes aneuploidy and ultimately tumorigenesis.98 In general, these observations suggest a tumor-suppressive function of the p53–DREAM pathway.

Entire functional Groups of Genes are Downregulated by the p53–DREAM Pathway: DNA Repair, Telomere Maintenance and Fanconi Anemia

Bioinformatic analysis of mRNA expression, ChIP and promoter element conservation data pointed at several genes involved in DNA repair and telomere maintenance to be downregulated by the DREAM pathway.17 Among the genes suggested to be regulated by DREAM were examples such as FANCB, DCLRE1B (Apollo), RAD54L, RAD18 and CHEK217, 31, 66 (Table 1). Interestingly, some of the DREAM targets are genes of the Fanconi anemia complementation groups (Table 2).17, 99

Table 2 DREAM targets among Fanconi anemia, dyskeratosis congenita, and related DNA repair and telomere maintenance genes
Full size table

Fanconi anemia is the most common inherited bone marrow failure syndrome. It causes constitutive genomic instability and predisposes for myelodysplasia, myeloid leukemia and solid tumors such as squamous cell carcinomas.100 Recently, expression of Fanconi anemia genes in the context of truncated vs full-length p53 was investigated in a mouse model. A truncated variant of p53 missing the C-terminal 31 amino acids was employed and its transcriptional program in comparison with full-length p53 was tested.101 The p53Δ31 mutant lacks the C-terminal domain which interferes with DNA binding reducing p53 activity.102 Thus, p53Δ31 displays elevated transcriptional activity compared with full-length p53 resulting in enhanced p21/CDKN1A activation and concomitant induction of the p53–DREAM pathway.39, 101 It was shown that several Fanconi anemia genes are repressed by p53, bind E2F4 after induction of p53 and contain candidate CDE/CHR sites in their promoters (Table 2). Detailed analyses were performed with FANCD2, FANCI and RAD51 (FANCR) by testing mutants of CDE/CHR sites in their promoters.101 Consistently, all Fanconi anemia genes which were experimentally confirmed to be controlled through DREAM had also been predicted by bioinformatic analyses as DREAM targets17, 101 (Table 2). These data suggest that an entire group of functionally related genes is coordinately downregulated by the p53–DREAM pathway. The coordinate regulation of whole functional groups implies that the p53–DREAM pathway controls not just a partial aspect but an entire function of a cell (Figure 4).

Another group of genes associated with telomere maintenance partially overlaps with the Fanconi anemia gene family as some genes from both groups are involved in DNA repair.101 The telomere-related DKC1 (Dyskerin), RTEL1 and TINF2 genes are found mutated in dyskeratosis congenita. From the meta-analysis data it is unclear whether they are also DREAM targets17, 101 (Table 2). However, many genes with functions in telomere maintenance, length or replication as well as DNA repair – e.g. DEK, FEN1, RECQL4, TIMELESS, BLM, RIF1, ACD, RPA2, WRAP53 (TCAB1), TRAIP and PIF1 (RRM3) – are indirectly downregulated by p53. Correspondingly, binding of DREAM components to these genes was observed by genome-wide ChIP experiments, again indicating that a functionally related gene set is controlled by the p53–DREAM pathway17, 101, 103, 104 (Table 2).

Also breast and ovarian cancer susceptibility genes BRCA1 and BRCA2 are among the genes implied as DREAM targets by observations from several genome-wide screens.17 BRCA1 and BRCA2 were originally described as Fanconi complementation groups FANCS and FANCD1, respectively.100 This identity has been challenged recently.105 Yet, downregulation of these genes by p53 and binding of DREAM components has been shown in a compilation of genome-wide expression and ChIP protein binding data.17 Furthermore, before the discovery of mammalian DREAM, observations suggested that E2F4, p130 and p107 can bind at the BRCA1 promoter after induction of hypoxia.106 p53-dependent repression of BRCA1 and binding of E2F4 to the gene was later confirmed.107

In general, these data suggest that gene groups representing pathways controlling important cell functions such as cell cycle checkpoint regulation, DNA repair, telomere maintenance and other functions are coordinately regulated by the p53–DREAM pathway (Figure 4). Again, this implies that p53 employs DREAM to exert its master regulator function by controlling entire sets of genes responsible for complete cell functions.

Cancer treatment: CDK Inhibitor Drugs and Rescue of the p53–DREAM Pathway

Cell cycle checkpoint control is in the focus of cancer treatment. Prominent examples for drugs targeting the cell cycle are Palbociclib (PD-0332991, tradename: Ibrance), Abemaciclib (LY2835219) and Ribociclib (LEE 011, Kisqali).108 Palbociclib was the first of these small-molecule inhibitors to obtain FDA approval for the treatment of breast cancer. The drugs inhibit CDK4/6 cell cycle kinases and compensate for the loss of checkpoint control in cancerous cells. The CDK inhibitors were originally aimed at primarily decreasing pRB phosphorylation in order to promote formation pRB/E2F transcriptional repressor complexes. The classical view is that hypophosphorylation of pRB is an important step in G1/S checkpoint control.58, 108

However, it has been established early – analogous to pRB itself – that the pRB-related proteins p107 and p130 are substrates for cyclin D/CDK4/6-dependent phosphorylation.109, 110 Thus, inhibition by drugs such as Palbociclib will lead to DREAM formation and cause downregulation of its target genes. As DREAM controls genes which are required for G1/S transition, the G2/M checkpoint and for progression through mitosis (Table 1), CDK inhibitors such as Palbociclib will address several cell cycle checkpoints by causing DREAM formation. This suggests that the therapeutic effect of the CDK inhibitors may depend on DREAM.

Human Papilloma Virus HPV E7 – Destruction of DREAM Function

Human papilloma virus (HPV)-16 E7 has been shown to bind the retinoblastoma protein pRB and impair its tumor-suppressive function.111 Consistently, also DREAM was reported to be disrupted by E7 binding to the pRB-related protein p130.112 Moreover, it is established that HPV E6 targets p53 for ubiquitin-mediated destruction.113 Also, the HPV E7 protein will compromise the function of p53 as a tumor suppressor through binding to the DREAM components p107 and p130. A genome-wide study listed the genes with their change in expression following HPV E7 protein expression.114 In a report on PLK4 transcription, we showed in regard to the mechanism that transcriptional deregulation by HPV E7 is mediated through the DREAM complex and CDE/CHR elements in the promoter of the gene.69 In general, these results implied that all genes controlled by DREAM through E2F or CHR sites in their promoters are deregulated by HPV E7.69 This notion emerged also from earlier data sets and a recent report on gene deregulation upon E7 expression in keratinocytes.69, 114, 115 In a recent genome-wide meta-analysis we identified more than 90 genes, mostly coding for cell cycle regulators, which are upregulated following E7 expression.116 Thus, these data suggest that deregulation of DREAM substantially contributes to HPV E7-mediated tumorigenesis.

DREAM and Epigenetics

DREAM also regulates genes involved in DNA methylation, nucleosome formation and histone modification, including CHAF1A, EZH2, H2AFX, KMT5A, SMCHD1 and SUZ12 (Table 1). Recently, it was shown that p53-dependent regulation of enzymes is required for DNA methylation homeostasis.117 In p53-deficient cells, an imbalance in DNA methylation causes clonal heterogeneity in naïve embryonal stem cells and upon differentiation. The DNA methyltransferase gene DNMT3B contributes to DNA methylation homeostasis and appears to be – according to meta-analysis data – downregulated by the p53–DREAM pathway (Table 1). Thus, with their role in epigenetics, DREAM and the p53–DREAM pathway contribute to gene regulation on a global level.

Implications of the p53–p21–DREAM–E2F/CHR pathway

p53 is a key mediator of cell cycle arrest in response to cellular stress. With the plethora of genes downregulated by the p53–DREAM pathway, it has become likely that this signaling pathway is central to cell cycle arrest (Table 1). Considering that regulator functions of these genes span from the G1 phase to the end of mitosis, it is evident that p53-dependent cell cycle arrest is not restricted to G1/S transition but is also important for all checkpoints up to the completion of cell division (Figure 4).

An unresolved issue in cell cycle checkpoint control is how functions of pRB and DREAM differ or overlap. Both pRB/E2F complexes and DREAM bind DNA through E2F sites. However, DREAM also employs CHR elements without E2F sites. Thus, gene sets controlled by pRB/E2F or DREAM will overlap but a separate set will be controlled by DREAM and CHR sites (Figure 2). It has been shown that pRB and p21/CDKN1A have additive effects on G1 phase regulation, which may suggest that pRB and DREAM are both required to control G1/S transition.118 Consistently, triple knockout cells for the pRB-related genes cannot undergo cell cycle arrest, in contrast to Rb−/− single or p130−/−; p107−/− double knockout cells which still arrest.119 Genome-wide expression and protein/DNA binding studies will be instrumental in defining the distinct functions of pRB and p130/p107 – and thus DREAM.65

Another feature of the DREAM pathway may be quality of the induced cell cycle arrest. While transcriptional regulation of cell cycle proteins is slower than regulation via, for example, phosphorylation as employed by other pathways, the response to the p53–DREAM pathway may be more sustained, possibly leading to senescence as an irreversible cell cycle arrest or to induction of apoptosis.2, 6, 74

The function of many oncogenic factors is to stimulate cell division or, as seen from another perspective, to counteract cell cycle arrest. Consistently, p53 functions as a tumor suppressor through the p53–DREAM pathway by downregulating many oncogenic proteins such as B-Myb, FOXM1, Cyclin B1/2 and CDK1/2 (Table 1). Thus, many of the genes repressed by p53 are found overexpressed in tumors once the p53–DREAM pathway is impaired. Expression signatures for many cancer types comprises genes downregulation by the p53–DREAM pathway.120 In numerous studies on many cancer types, p53–DREAM targets head the list of signature genes whose aberrant expression is predictive for poor clinical outcome of cancer patients.121, 122, 123 Considering that CDK inhibitors promote repression of these genes by DREAM, functional defects of p21/CDKN1A or upstream pathway elements can be attenuated by these drugs.

In summary, the p53–p21–DREAM–E2F/CHR pathway downregulates a plethora of cell cycle genes, contributes to cell cycle arrest and is a target for cancer therapy. Researchers working on p53 function, cell cycle regulation or cancer treatment may soon join in saluting: We have a DREAM!