response pathways and tumorigenesis
Bruins, W.
Citation
Bruins, W. (2007, October 24). The role of p53.S389 phosphorylation in DNA damage response pathways and tumorigenesis. Department Toxicogenetics, Medicine / Leiden University Medical Center (LUMC), Leiden University.
Retrieved from https://hdl.handle.net/1887/12389
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/12389
Note: To cite this publication please use the final published version (if applicable).
Chapter 1
General introduction
Introduction
P53 and cancer
The human p53 gene is localized at chromosome 17 (17p13.1) [1]. Within the gene 11 exons are identified, of which exon 1 is non-coding. Exons 2 to 11 encode the p53 protein comprising 393 amino acids [2]. The p53 protein contains several structural and functional domains [Recently reviewed in [3]] (Figure 1). The N-terminal transactivation domains (TAD1; amino acids 1-42 and TAD2; amino acids 43-63 [4;5]) are required for transcriptional activation. The Proline-rich domain (PRD; amino acids 61-94) regulates the stability and activity of p53, and PXXP motifs in this domain enable protein-protein interactions. Both the TADs and the PRD are furthermore able to interact with co-activators and co-repressors. The DNA binding domain (DBD; amino acids 100-300) is the largest domain and binds specifically to DNA recognition sequences in promoter elements of target genes. The tetramerization domain (4D; amino acids 324-355) regulates the formation of a dimer of two p53 protein dimers into a tetramer; p53 binds the promoter of its target genes as a tetrameric complex. The carboxy-terminal regulatory domain (CTD; amino acids 360-393) binds non-specifically to DNA and is presumably able to influence the binding of DNA by the DBD [6;7]. Finally, a nuclear localization signal (L) and a nuclear export signal (E) are present. Actually there are two nuclear export signals (NES) identified in p53, one located within the C-terminal oligomerization domain and the other in the N-terminus respectively [8;9]. Here we showed the C-terminal NES only, since this one is most studied. Recently, however, there is found that the nuclear export via the C-terminal NES is promoted when DBD and C-terminal lysines are ubiquitinated [10].
P53 was first identified as an oncogene when there was reported that it might be involved in the promotion of cell proliferation, immortalization of cells and transformation of primary rat embryo fibroblasts in cooperation with ras, but later it was recognized as a tumor suppressor gene,
Figure 1 - Model for regulation of p53 The p53 protein is kept inactive and at low levels in unstressed cells by Mdm2, which binds the TAD and in that way blocks transcriptional activation of target genes. Mdm2 also promotes p53 degradation by the proteasome through ubiquitination (upper part).
The p53 protein is stabilized and activated after inhibition of Mdm2 -mediated ubiquitination by various mechanisms, including phosphorylation events on both p53 and Mdm2 by various cellular stress-induced kinases. Tetramers are formed which stay in the nucleus as the nuclear export signals are masked. Accumulated p53 can also indirectly activate apoptosis through Bax and Bak by binding to the anti-apoptotic proteins Bcl2, BclxL (lower part) TAD; transactivation domain, PRD; proline-rich domain, DBD; DNA-binding domain, L; nuclear localization signal, 4D; tetramerization domain, E; nuclear export signal, CTD; C-terminal regulatory domain, A; acetylated lysines, P; phosphorylated serines and threonines, Ub; Ubiquitinated lysines
Adapted from: F. Toledo and G.M. Wahl, Nat Rev Cancer, 2006
because it was frequently found inactivated in human cancer [reviewed in [11]]. Inactivation can be accomplished directly through mutations in the p53 gene, or indirectly through alterations in genes whose products regulate p53 (e.g., Mdm2). Looking at the high incidence of mutations in several human cancers up to now (see also description of available databases in [12;13]), most mutations are found to be located in the DNA binding domain. Furthermore, the cancer- prone inheritable Li-Fraumeni Syndrome (LFS) is characterized by heterozygous mutation in the p53 gene [14]. These LFS patients develop a variety of tumors (for example in the breast, brain, leukemia and others) with an incidence of ~50% at an early age (frequently before 30 years) [15-17]. Although the vast majority of Li-Fraumeni patients carry a heterozygous germ- line mutation in their p53 gene [15], also Li-Fraumeni patients with mutations in Chk2 (an upstream regulator of p53) have been found [18;19].
P53 is a well-known DNA damage sensor preventing the accumulation of genetic lesions and, consequently, tumor development. The protein participates in a number of defense pathways, like cell cycle control, DNA repair, apoptosis, angiogenesis or senescence. Several processes induce p53 protein stabilization. The signals initiated by DNA damage, oncogenic stimuli, cell adhesion and redox stress have been described in detail in recent reviews [20;21]. Hypoxia and heat shock were also shown to cause an increase in p53 protein levels. These stimuli induce cellular stress by inhibiting oxygen-dependent metabolism or denaturing proteins [22]. Furthermore, altered ribonucleotide pools like depletion of CTP, GTP, or UTP caused a p53-dependent G0 or early G1 arrest [23], and disruption of microtubules induces phosphorylation of p53 while distinct p53 phospho-forms were found in a cell specific manner [24]. In Figure 2 a compilation of these cellular stressors identified upto now are depicted, together with the p53-mediated cellular responses that can be induced. After DNA damage, levels of p53 protein in a cell increase owing to protein stabilization [25-27] (see also next section ‘p53 (de-) regulation’).
P53 (de-)regulation
The best known mechanism to destabilize p53 is by the ubiquitin ligase Mdm2. Mdm2 first binds to the N-terminus of p53, where it reduces the ability of p53 to activate gene expression of its target genes. This is caused by the fact that Mdm2 binds within the transactivation domain of
p53 p53 p53 p53 p53 p53 p53p53 p53 p53 p53p53
Extracellular matrix interaction (cell adhesion)
Oxygen reduction (hypoxia) Oncogene expression
DNA damage induction (bulky lesions, strand breaks etc.)
Microtubule disruption
Redox condition Nucleotide pool depletion
p53 accumulation and/or activation
Post-translational modifications Release from inhibitors Binding to co-factors Cell cycle arrest
Apoptosis Senescence
Genetic stability
Inhibition of angiogenesis and metastasis Media depletion (in vitro studies)
Proteasome inhibition
Topoisomerase inhibition Heat/Cold shock activation
Viral infection
Ribosome creation (ribosome biogenesis) Spindle damage induction
Adapted and modified from: M. Ljungman, Neoplasia, 2000 A.C. Blackburn and D.J. Jerry, Breast Cancer Res, 2002 S.L. Harris and A.J. Levine, Oncogene, 2005
Figure 2 - Activation of p53 and the cellular responses
The p53 responses are triggered by many different stressors involving both DNA-damaging and non-DNA-damaging agents and processes. Activation of cellular stress pathways causes several covalent modifications and tetramerization of p53. The ultimate cellular outcome depends on several factors as described in this Introduction chapter.
p53 (specifically at residues 18-28), which is an important region for the interaction of p53 with components of the transcription machinery (see also Figure 1). Examples of these components are TBP with its associated factors (TAFs) and its transcriptional co-activator p300 [28]. In addition to inhibiting p53’s transactivation function, Mdm2 ubiquitinates p53 through its E3 ubiquitin ligase function; thereby targeting the protein for proteasomal degradation. Mdm2 contains nuclear import and export signals and degradation of p53 depends on the ability of Mdm2 to shuttle from the nucleus into the cytoplasm [29;30]. However this is still debatable as there is also shown p53 can be degraded in the nucleus [31]. Some factors can influence this ubiquitination process by either enhancing it (transcription factor Yin Yang 1 or the complex HDAC1 which promotes deacetylation of lysines which can be ubiquitinated [32] or through repression (Abl, p14ARF,L5, L11, L23, PML, Rb [33;34] or deubiquitination by HAUSP [35].
Loss of Mdm2 causes cell death in vitro and in vivo in a p53-dependent manner [36;37]. A protein similar to Mdm2 was reported a few years ago, named Mdm4 (or Mdmx) [38]. Both Mdm2-null and Mdm4-null mice are embryonically lethal. However, mice deficient for both Mdm2 and p53 developed normally and were viable [37]. Interestingly, also the embryonic lethality of Mdm4-null mice was rescued by loss of p53 [39]. Mdm2-null embryos die due to induction of apoptosis at 3.5 dpc, whereas Mdm4-null embryos die from a block in cellular proliferation at 7.5 dpc [39]. Taken together, these results suggest that Mdm2 and Mdm4 are involved in the regulation of p53 function during development. As loss of Mdm2 cannot be compensated by Mdm4 and vice versa; Mdm2 and Mdm4 are therefore suggested to be involved in different processes of regulating p53. However, it has also been described that Mdm4 and Mdm2 are both needed in the same process; i.e., Mdm4 stabilizes the Mdm2 protein so it can degrade p53 and Mdm4 on his turn requires Mdm2 to go into the nucleus in order to inactivate p53 [40]. This might be of importance since Mdm4, in contrast to Mdm2, possesses no detectable E3 ubiquitin ligase activity and by itself cannot stimulate p53 ubiquitination and proteasome-dependent degradation. However, many conflicting data have been published and the role of Mdm4 in the regulation of Mdm2 and p53 stabilization needs further research [reviewed in [41]].
Recently, Mdm2-independent pathways of regulating ubiquitin-mediated p53 destabilization were described [42-44]. Key players like Pirh2, Cop1 and ARF-BP1 were identified as new p53-specific ubiquitin ligases [42-44]. It has been suggested that they are able to control p53 levels (Figure 3). Probably, the cell may need different sets of ubiquitin ligases to degrade p53 upon specific cellular stresses [42;43;45]. However, the need of these newly found ubiquitin ligases, in controlling p53 levels is still debatable. Furthermore, in unstressed and stressed cells JNK-p53 complexes were found which are like Mdm2 inversely correlated with p53 levels.
Where p53-Mdm2 complexes are mostly present in the S/G2/M phases of the cell cycle, p53- JNK complexes accumulate in the G0/G1 phases [46;47]. It seems therefore likely that Mdm2 and JNK regulate different cellular functions of p53 in unstressed cells in a cell cycle-dependent manner.
In summary, besides the well-known targeting molecule Mdm2, other molecules are responsible for the short half life of p53 in both stressed and unstressed cells.
Activation of p53 through post-translational modifications
As described before, several stressors (depicted in Figure 2) induce p53 stabilization, and p53 stabilization and activation is mostly induced by a variety of post-translational modifications. The
modifications are very diverse, as p53 can be phosphorylated, acetylated, ubiquitinated, sumoylated, glycosylated, methylated or neddylated. The N-terminus is mostly phosphorylated and is involved in the stabilization of the protein. Furthermore, the phosphorylated N-terminus is thought to be important for the modifications of the C-terminus and finally for fully p53- mediated responses [33;48].
Phosphorylation of p53
As mentioned previously, the most common post-translational modification of the p53 protein is its phosphorylation. This predominantly occurs at N- or C-terminal ends of the protein, at serine and threonine residues. Modifications of the N-terminal domain are thought to disrupt the p53-Mdm2 interaction, whereas modifications in the C-terminal domain are thought to interfere with non-specific DNA binding or induce conformational changes that prevent interactions between the C-terminus and the core DNA-binding domain, needed for activating and stabilizing the p53 protein.
Several protein kinases have been identified that phosphorylate p53 at its N-terminal part: CK1 and 2, DNA-PK, ATM, ATR, CHK2, PLK3, JNK, AMPK, CDK9, ERK, MAPKAP2, RSK2, VKR1 and p38, whereas at the C-terminal part: CHK1, CHK2, CK2, p38, PKC, CDK9, and PKR are identified as the p53 modifiers [reviewed in [3;48-53]].
The exact function of these individual phosphorylation events as a reaction to the above mentioned cellular stressors is largely unknown. However, knowledge has become available as to what the specific phosphorylation target sites are for the different kinases. Figure 4 is an adapted figure from different reviews (modified to the corresponding murine phosphorylation sites) describing well known N- and C-terminal p53 phosphorylation sites with their specific kinases. One of the major aims of this PhD study was to gain knowledge on how a particular phosphorylation event contributes to the cellular defense against DNA damage. Therefore, post-translational modification of p53 will be discussed here mostly within this context.
In vitro effects of phosphorylation on p53 functions
The role and significance of phosphorylation has initially been investigated using in vitro model systems. These studies showed several interesting findings. Phosphorylation of Ser15 was found to stimulate p53 functioning since a partial DNA-damage induced cell cycle block was found [54]. Phosphorylation of Ser20 seems furthermore to be important for the DNA-damage induced cell cycle control [55]. Phosphorylation of Ser37 was, however, found to play a role in
p53 p53
Atm Chk2
Positive effectors
CKII
p19Arf
Negative regulators Cell cycle arrest Mdm2
Pirh2 Cop1
p21 Mdm4
Apoptotic effectors Puma Noxa Apaf1 Perp1 Bax
Adapted and modified from: G. Lozano and G.P. Zambetti, J Pathol, 2005
ARF-BP1
Figure 3 - Components of the p53 pathway
Several positive and negative regulators, able to control p53 levels, are depicted. An overview of the correlation between these proteins and p53 is shown.
(For details see text.)
the p53-mediated transcriptional regulation of the DNA-damage response [56].
The impact on the regulation of p53 function, specifically transcriptional transactivation and sequence-specific DNA binding, was furthermore examined for a number of phosphorylation events by generating a series of p53 mutants in human and murine cells [57;58]. These sites showed not to be essential for suppression of cell growth or DNA binding for transactivation of different p53 target promoters individually. However, mutations in several residues simultaneously did show a reduced apoptotic response [58], whereas a more recent study showed a combination of several phosphorylation sites are not required for activation of p53 target genes [59].
Although these in vitro experiments revealed important insights, results were highly contradictory.
The overall conclusion is that phosphorylation of p53 is required for some, but not all, biological responses to DNA damages in specific situations. Furthermore, the requirement of a specific phosphorylation site for a certain cellular or molecular p53 response might be highly dependent on the cell type and the nature of the DNA damage introduced. Later, mouse models were used to identify the significance of the phosphorylation in a more biological relevant system. Since this thesis is about one of these mouse models we will review the in vivo model systems more extentensively compared to the in vitro models.
P53 phosphorylation at Ser389
As mentioned before, various kinases are responsible for the post-translational modifications of p53. When we focus on phosphorylation of Ser389 (equivalent of human Ser392), several kinases have been described that can perform this phosphorylation (see Figure 4). Since amino acid positions of the p53 protein differ between mouse and human (i.e., Ser392 in human versus Ser389 in mouse), this site will for convenience of the reader be annotated to Ser389 irrespective of the species.
The in vitro phosphorylation of Ser389 has been attributed to Casein Kinase II (CKII) [60;61], the double stranded RNA activated protein kinase (PKR) [62], p38 MAP kinase [63;64] and the recently discovered Cdk9 [65]. However, the kinases which can phosphorylate p53.S389 in vivo have not been definitively identified yet. Numerous experiments, mainly based on in vitro non-physiological studies with mutant p53 cDNAs, have been performed to analyze the function of Ser389 phosphorylation. Despite the fact that interesting observations were made,
Adapted and modified to murine version from: E. Appella and C.W. Anderson,Eur J Biochem,2001, M. F.Lavin and N. Gueven, Cell Death Differ, 2006 F.T oledo and G.M. Wahl, Nat Rev Cancer, 2006
P P P P P
N-and C-terminal phosphorylation modifications S9 S12 S18 S23
T21
P PP P PP
CK1 CK1 DNAPK
AMPK ATM ATR DNAPK ERK p38 CDK9 RSK2
CK2 CHK2 VRK1 DNAPK
MAPKAPK2 JNK CHK2
PLK3 CHK2
PKC
S363 S373 S389
S374 S375 S384
CHK1 CHK2
CHK1 CHK2 PKCCHK1 PKR
p38 CK2 CDK9
N TAD *****
PRD DBD L 4D
E CTD C
Figure 4 - Cellular stress induced phosphorylation of p53
The most common post-translational modification of the p53 protein is phosphorylation.The bar represents the murine p53 protein with its specific domains (see also figure 1). The kinases responsible for phosphorylation of the different serines (S) and threonines (T) are indicated above the arrows pointing towards their corresponding ‘target’ sites.
results obtained are highly contradictory, still though some of these will be discussed here.
It has been shown that phosphorylation of Ser389 is important for tetramer formation of p53 [66]. Under normal, unstressed conditions the p53 protein is found at low levels in a monomeric fashion. However, upon DNA damage protein levels increase rapidly and p53 starts to assemble into tetrameric complexes. Artificial peptides, either unphosphorylated or phosphorylated at Ser389, were tested for their ability to form tetramers. It was found that tetramer stabilization was positively affected upon Ser389 phosphorylation as probably new polar interactions could be formed [66]. So, after the induction of DNA damage, p53 phosphorylation at Ser389 possibly enables the formation of tetrameric complexes, which are transcriptionally activating downstream target genes.
In a plasmid-based assay the transcriptional activity of p53 was tested using a Ser to Ala mutation at codon 389 to mimick the unphosphorylated form and a Ser to Glutamic Acid (further referred to as ‘Glu’) mutation, mimicking a constitutively phosphorylated form. Introduction of these mutants into NIH3T3 cells showed that only the phosphorylated form but not the non-phosphorylated form was transcriptionally active at high cell density, whereas at low cell density no differences could be detected between the Ala and Glu variants [57]. Apparently, phosphorylation of Ser389 has an important role under certain conditions like cell density, which might be attributed to cell adhesion mechanisms.
Several stress-activated phosphorylation serine sites, including Ser389, were tested for transcriptional activation in HCT116 and RKO cells [59]. In these studies it was found that there were no differences in DNA binding or target gene activation between unphosphorylated and phosphorylated p53. Apparently, activation of the downstream signaling pathways is not related to p53 phosphorylation for these sites studied. A second study [59] described the use of nutlin-3, which is an Mdm2-antagonist leading to accumulation of unphosphorylated p53 and subsequently apoptosis. This response is comparable to the apoptotic induction of cells following treatment with etoposide which led to phosphorylation of p53. Again, no difference in target gene activation was observed in the downstream biological responses between the phosphorylated (etoposide-induced) or unphosphorylated (nutlin-3a treated) forms of p53.
Furthermore, Jackson et al. [67], described experiments in which accumulation of p53 occurred by sequestering Mdm2 without affecting the phosphorylation status except for Ser389 phosphorylation, which did occur. Again, most p53 target genes are regulated in the same way when comparing the fully phosphorylated (doxorubicine-induced) with the unphosphorylated forms of p53. Although, it can be concluded that many phosphorylation events seem dispensable for transcriptional activation, this conclusion cannot, however, be drawn for the Ser389 phosphorylation event since it was still present in this study. Overall, these studies imply that increased levels of p53 are sufficient for transactivation of target genes, rather than the post- translational modification events.
The major criticism on these in vitro studies is that they do not fully recapitulate the in vivo situation as transfection studies were performed, which inevitably lead to over-expression of p53, which normally does not occur in vivo. Furthermore, mostly cancer cell lines were used which might already have a defective and/or disturbed p53 response in for instance the apoptotic pathway and clearly do not represent normal homeostasis. Finally, transcriptional activity of only a limited set of p53 target genes was tested in only a limited number of cell lines, which is evidently not representative for p53-directed overall transcription regulation.
In conclusion, it might be that the N-terminal phosphorylation sites are mostly needed for
release of Mdm2 so the p53 protein can be stabilized. When p53 protein levels reach a certain threshold and co-activators (like p300) are present, transcriptional activation of target genes is initiated. Phosphorylation of the C-terminal region might be essential in a more subtle way to target gene activation through tetramer complex formation.
Another controversy attributed to the mutation of Ser389 into Ala is increased stabilization of p53 by inhibiting its nuclear export signal [68]. Interestingly, under the conditions tested it was found that Mdm2 was still bound to p53, which normally leads to p53 degradation.
The mutated p53.S389A (no phosphorylation) form was mainly localized in the nucleus, whereas the p53.S389E (mimicking phosphorylation) and wild-type p53 form were present in both the cytoplasm as well as in the nucleus. Furthermore, it was shown that both forms (unphosphorylated as well as phosphorylated Ser389) had similar effects on adriamycin and UV- induced cell cycle arrest or apoptosis. One might have expected the opposite as phosphorylated p53 is found to be stabilized and able to induce transcription of genes.
Interestingly, the oncogenic function of mutated p53 protein was also reported to be related to Ser389 phosphorylation. Yap et al. studied the role of Ser389 phosphorylation in combination with tumor-associated p53 mutations [69]. Two p53 hotspot mutants (H175 and W248) were used in combination with a phosphorylation competent (p53H175 and p53W248) and a phosphorylation-deficient (p53H175;S389A, p53W248;S389A) form of Ser389. The S389A variants showed a strong inhibition of the apoptotic response and an increase of the dominant- negative activity by enhancing the transformation potential. The authors claim that their results are in agreement with earlier findings where it was found that human breast tumors express p53 protein in a non-phosphorylated form at Ser389. They concluded that the absence of phosphorylation at Ser389 enhances the oncogenic potential of mutant p53 [69].
On the other hand dominant-negative effects of phosphorylated Ser389 were described by Furihata et al. [70]. As phosphorylated Ser389 seems to have a role in tetramer formation, it seems likely that after Ser389 phosphorylation hetero oligomerization of a mutant form of p53 with wild-type p53 is initiated, promoting the dominant-negative effects of the mutant protein.
Finally, other clinical observations [71] showed corresponding results since (increased) phosphorylation of Ser389 was correlated to tumor growth. In samples of young patients (<35 years) with Vestibular Schwannomas higher levels of phosphorylated Ser389 were identified compared to tumor samples of older patients [71]. The authors linked increased levels of Ser389 phosphorylation to accelerated tumor growth. Although interestingly, this study was performed in low numbers of patients and therefore needs further examination.
In conclusion, an extensive number of in vitro studies have been performed to analyze the significance of Ser389 phosphorylation, and to analyze the adverse effects on p53 functioning when Ser389 phosphorylation is abolished/inhibited (see for an overview Table I). Although, many interesting results were obtained, there still is no clear picture on what the precise function is of Ser389 phosphorylation in various p53-mediated processes. Therefore, a more successful approach might be to study the function of Ser389 phosphorylation in vivo, in a mutant mouse model, which was the major challenge in the studies described in this thesis.
P53 mouse models and cancer
To analyze p53 functioning in vivo, mice with heterozygous or homozygous loss of p53 were generated using homologous recombination in murine embryonic stem cells [72;73]. These
p53 knockout mice were extensively studied, and several reviews described the phenotypes of p53-/- and p53+/- mice in detail [73-77]. Briefly, the homozygous p53-deficient mice show spontaneous development of mainly lymphomas at the age of 6 months, but also sarcomas were found [72;73]. All p53 knock-out models showed that p53 is not absolutely required for normal mouse development overall, since only a small fraction of female embryos displayed defects in neural tube development, also known as exencephaly [78]. The p53 heterozygous mice (p53+/-) are also cancer prone, although to a lesser extent than their p53-/- counterparts. The tumor onset in p53+/- is at a later age (i.e., 12 – 24 months of age) and the main tumor types found were sarcomas [73]. Studies with these models taught us what the role of p53 in tumor suppression in mice is. These studies were important in order to understand what the role of p53 in tumor development in humans is as well. However, in human tumors large deletions of the p53 gene, as introduced into the mouse genome of the p53 knock-out animals, are rare. Therefore, although useful, these first generation p53 knockout mice do not fully recapitulate the human situation.
To accomplish that, other genetic variants introduced into mice are needed, like point mutated p53 as found in human tumors. The majority of p53 missense mutations encountered in human tumors are located in the DNA binding domain [79-81]. As a consequence, the protein cannot bind to DNA and as such no transcriptional activation of target genes will occur.
The functional consequences of missense mutations compared to deletions of p53 have already been studied extensively in vitro (reviewed in [82]). One characteristic of p53 mutations in a
Table 1: Analysis of the role of Ser389 phosphorylation in regulation of p53 functioning in in vitro studies
Assay / Cells
Modification of Ser389 (mouse) or Ser392 (human)*
Results and Conclusions Reference
Synthesis of peptides with phosphorylated and unphosphorylated forms of p53.S389
The phosphorylated status of p53 at Ser389 increased the "association constant"
for reversible tetramer formation nearly 10-fold.
66
Transfection with plasmids containing a Ser389 mutation into NIH3T3 cells
p53.S389A p53.S389E
The Glutamic acid variant is transcriptionally active and has an increased binding activity. The Alanine variant lost these functions.
57
HCT116 cells and RKO cells;
containing a phosphorylated status of p53 after exposure to doxorubicine or etoposide;
containing an
unphosphorylated status of p53 (still stabilized and activated) after treatment with Nutlin3a
Phosphorylation of Ser389 is not required for: 1) activation of p53 target genes 2) induction of apoptosis.
59
Transfection with vectors containing different variants of p53.S389 into the human prostate cancer cell line PC3
p53.S389 p53.S389A p53.S389E
Cell cycle inhibition and apoptosis is normal in both variants. The alanine variant;
1) enhances p53 stability without disrupting Mdm2 binding 2) is mainly localized in the nucleus. The phosphorylated variant is distributed through the cytoplasm and the nucleus just like wild type.
68
Transfection with plasmids containing mutant forms of p53 combined with a different phosphorylation status of p53.S389 into: Primary rat embryo fibroblasts (REF), H1299 cells or Soas-2 cells
p53.H175 p53.H175/S389A p53.H175/S389E p53.W248 p53.W248/S389A p53.W248/S389E
The alanine variants are more: 1) dominant negative 2) active in cooperation with ras oncogene transforming REFs 3) potent inhibitors of apoptosis. In conclusion phosphorylation of Ser389 is involved in the regulation of oncogenic functioning of mutant p53
69
Saos-2 cells p53.S389A
p53.S389D
Both variants have no influence on: 1) p21 or mdm2 transcription 2) DNA binding 3) cell cycle arrest
218
Introduction of different p53.S389 variants into Tg1 bacteria
p53.S389A
p53.S389D Both variants showed: 1) normal immunoreactivity with antibodies 2) specific
binding for the DNA consensus 219
Table 1: Analysis of the role of Ser389 phosphorylation in regulation of p53 functioning in in vitro studies
Assay / Cells
Modification of Ser389 (mouse) or Ser392 (human)*
Results and Conclusions Reference
Synthesis of peptides with phosphorylated and unphosphorylated forms of p53.S389
The phosphorylated status of p53 at Ser389 increased the "association constant"
for reversible tetramer formation nearly 10-fold.
66
Transfection with plasmids containing a Ser389 mutation into NIH3T3 cells
p53.S389A p53.S389E
The Glutamic acid variant is transcriptionally active and has an increased binding activity. The Alanine variant lost these functions.
57
HCT116 cells and RKO cells;
containing a phosphorylated status of p53 after exposure to doxorubicine or etoposide;
containing an
unphosphorylated status of p53 (still stabilized and activated) after treatment with Nutlin3a
Phosphorylation of Ser389 is not required for: 1) activation of p53 target genes 2) induction of apoptosis.
59
Transfection with vectors containing different variants of p53.S389 into the human prostate cancer cell line PC3
p53.S389 p53.S389A p53.S389E
Cell cycle inhibition and apoptosis is normal in both variants. The alanine variant;
1) enhances p53 stability without disrupting Mdm2 binding 2) is mainly localized in the nucleus. The phosphorylated variant is distributed through the cytoplasm and the nucleus just like wild type.
68
Transfection with plasmids containing mutant forms of p53 combined with a different phosphorylation status of p53.S389 into: Primary rat embryo fibroblasts (REF), H1299 cells or Soas-2 cells
p53.H175 p53.H175/S389A p53.H175/S389E p53.W248 p53.W248/S389A p53.W248/S389E
The alanine variants are more: 1) dominant negative 2) active in cooperation with ras oncogene transforming REFs 3) potent inhibitors of apoptosis. In conclusion phosphorylation of Ser389 is involved in the regulation of oncogenic functioning of mutant p53
69
Saos-2 cells p53.S389A
p53.S389D
Both variants have no influence on: 1) p21 or mdm2 transcription 2) DNA binding 3) cell cycle arrest
218
Introduction of different p53.S389 variants into Tg1 bacteria
p53.S389A p53.S389D
Both variants showed: 1) normal immunoreactivity with antibodies 2) specific binding for the DNA consensus
219
Table I - Analysis of the role of Ser389 phosphorylation in regulation of p53 functioning in in vitro studies
heterozygous state can be ‘dominant-negative’ inhibition of wild-type p53, where the function of wild-type p53 is inhibited by the mutant protein through mutual protein-protein interactions [83]. Another described altered function of mutant p53 is the so-called ‘gain-of-function’
variant, where additional (oncogenic) features of mutant p53 are apparent, which are absent in the wild-type form [84].
As a first step to investigate the role on tumorigenesis of these human-derived mutations, several transgenic mouse models were generated with (over-)expression of the specific mutant in a tissue of interest (for example epidermis and mammary gland) [85-87]. Taken all studies together, these transgenic mouse models showed that mutant p53 over-expression results in increased spontaneous and carcinogen-induced tumorigenesis, with a dominant effect of the mutant protein [85].
In an attempt to more closely mimic the human p53 condition as found in LFS patients and several human cancers, germline p53 missense mutations were introduced into mouse models using the endogenous p53 promoter, resulting in expression of mutant p53 proteins in a physiological manner. Recently, (conditional) mutant p53 knock-in mice like p53.R175H and p53.R273H were generated to investigate LFS in mice in a more physiological manner [80;88].
With the mutant p53 protein expressed in all tissues, these mutant mouse models showed clear dominant-negative and gain-of-function properties of the mutant protein. An overall increase in the development of carcinomas and B-cell-lymphomas in p53R270H/+ mice and osteosarcomas in p53R172H/+ mice was found compared to p53+/- mice [80].
Using the same mouse models, the effect of the mutations on tumor development and cellular processes was analyzed in tissues of interest through crosses of conditional p53.R270H variants with Cre-transgenics [89;90]. Heterozygous p53R270H/+ WAP-Cre mice have a mammary gland- specific expression of the mutation at physiological levels [89]. These mice spontaneously developed mammary tumors at high frequencies, whereas mice lacking one p53 allele did not, indicating that the p53.R270H mutation acts in a dominant-negative manner for this specific tumor type. In addition, conditional knock-in mice with a mutation in K-ras combined with the p53.R270H mutant were generated [90]. Here, the p53.R270H mutation acted in a partial dominant-negative way in promoting K-ras-initiated lung adenocarcinomas. In conclusion, expression of mutant p53.R270H protein acts in a dominant-negative fashion with varying degree, depending on the tissue where it is expressed and the tumor type it is involved in [89;90].
Functional p53 mutant mouse models Mouse models with defective phosphorylation sites
Besides analyzing loss of p53 function in knockout or tumor mutant related mouse models, recently studies were described focussing on mutations in specific phosphorylation sites of p53 in mouse models. An overview of these functional mutant mouse models and cells derived from them is depicted in Table II.
Wu et al. published studies with murine embryonic stem cells (ES cells), and MEFs and thymocytes derived through a RAG2 complementation approach. Cells, with a missense mutation in p53 at Ser23 (Ser to Ala), abolishing Ser23 phosphorylation were tested for several cellular functions [91]. It was found that p53.S23A MEFs accumulate p53 as well as p21 and Mdm2 proteins to normal levels after UV radiation. Besides these MEFs, p53.S23A mutant ES cells and thymocytes also accumulate p53 protein and undergo p53-dependent apoptosis
at comparable levels as wild-type cells upon UV or IR radiation. Based on these results it was concluded that the p53.S23A mutation is not required on itself for p53-dependent responses after DNA damages induced by either UV or IR [91].
Next, MacPherson et al. also published the generation of a mouse model in which p53 was modified at Ser23 (from Ser to Ala) [92]. In line with the results described by Wu et al., they also could not find a significant reduction in p53 protein stabilization after IR in MEFs, and neither did they see a difference in p53-dependent cell cycle arrest. However, they found a decrease in the apoptotic response and p53 stabilization in p53.S23A thymocytes after DNA damage (IR or dexamethasone treatment). Next, a decrease in white-pulp apoptosis, a partial resistance of splenic B-cells to irradiation-induced apoptosis, and a significant decrease of apoptosis in p53.S23A CNS after gamma irradiation compared to wild-type mice was found.
Finally, spontaneous tumor development in p53.S23A mice was accelerated, with a tumor spectrum (i.e., B-cell lineage tumors) clearly different from p53-/- mice ([72;73;92-94]). In contrast to p53-/- mice which have an average life span of approximately 6 months, the p53.
S23A have a median survival time of about 15 months. B-cell lymphomas develop with low frequencies in p53-/- mice, possibly because the fact that these mice develop life threatening T- cell lymphomas and sarcomas at earlier stages. From these findings it could be concluded that Ser23 phosphorylation has an important role in regulating some functions of p53 in vivo.
Also the role of Ser18 (equivalent of human Ser15) phosphorylation in p53 was examined in vivo [95]. Wild-type p53 is rapidly phosphorylated at Ser18 after IR as well as after UV. A missense mutation was introduced in ES cells by changing Ser18 into an Ala residue, resulting in a reduced accumulation of p53 protein in mouse embryonic stem cells after DNA damage (both IR and UV). Consequently, p21 protein levels were also reduced under these conditions and the p53-dependent cell cycle arrest response was impaired after IR. Finally, several reports suggested that phosphorylation of p53 at Ser18 might activate the acetylation of C-terminal lysines through increased recruitment of transcription co-activators like p300 and CBP [96;97].
However, no differences in the C-terminal acetylation status of Lys317 and Lys379 were found after UV exposure between wild-type and p53.S18A cells. In conclusion, phosphorylation of p53
Table 2: Functional p53 mutant mouse models
Mouse Model Apoptotic response Cell cycle arrest response Tumor development Reference
p53.S23A Decreased after IR 2,6,7 Normal after IR 1 Spontaneous B-Cell lineage tumors 92
Median survival 63 weeks
p53.S18A Decreased after IR 2,5 Decreased after UV 1* No spontaneous tumor development (1-2 years) 98,99 Normal after IR 1
p53.S18A/S23A Decreased after IR 2 Decreased after IR 1 Wide spectrum of spontaneous tumors 100 Median survival 80 weeks
p53.S389A Decreased after UV 1 Normal after IR 1,2 No spontaneous or IR-induced tumors This thesis;
Increased UV-induced skin and 2-AAF-induced urinary bladder tumors Chapters 2 and 4
p53.QS Decreased after IR 2 Decreased after IR or PALA 1 N.D. 5,103,105
Decreased depending on damage 1 Decreased after doxorubicin or IR 1
p53.deltaP Compromised after etoposide or IR 1,2 Decreased after IR or Adr 1 No spontaneous tumor development (1 year) 107 Increase oncogene-induced tumors (xenograft)
p53.AXXA Normal after IR 1 Normal after IR 1 No spontaneous or oncogene-induced (xenograft) tumor development 109 p53.TTAA
p53.7KR Normal after IR 1 Normal after IR 1 N.D. 110
p53.K317R Increased after IR 1,2,3,4 Normal after UV 1 N.D. 111
p53.super Increased after IR 2,5 Increased after IR 5,8,9 Less susceptible to spontaneous or carcinogen-induced tumor formation 220
N.D. = Not detected, IR = ionizing radiation, Adr = Adriamycin
1MEFs 2Thymocytes 3Small intestinal epithelial cells 4Retina 5Splenocytes 6Splenic white pulp (B- and T-lymphocytes) 7Cerebellum 8Lung 9Kidney
*E1A MEFs; reduced apoptotic response after 2.5 µM etoposide but normal response after 10 µM etoposide.
Table II - Functional p53 mutant mouse models
N.D. = Not detected, IR = ionizing radiation, Adr = Adriamycin
1MEFs2Thymocytes3Small intestinal epithelial cells4Retina5Splenocytes6Splenic white pulp (B- and T-lymphocytes)7Cerebellum8Lung9Kidney
*E1A MEFs; reduced apoptotic response after 2.5 µM etoposide but normal response after 10 µM etoposide
at Ser18 is not essential for acetylation of p53 at the C-terminus in response to UV. [95]. Next, a p53.S18A knock-in mouse model was generated [98]. Thymocytes isolated from these mice showed a normal induction of p53 protein levels upon IR. Interestingly, however, this normal protein response was accompanied by an impaired apoptotic response. To further investigate this impaired response a microarray assay was performed. In general a promoter-specific impact of the p53.S18A mutation was found as some p53 target genes showed a lower induction of gene expression whereas others showed comparable expression levels as seen in wild-type mice.
The cell cycle arrest response after UV was impaired in MEFs isolated from p53.S18A mice, although p53 protein levels were highly induced. Finally, in contrast to what was observed in p53.S18A ES cells, acetylation of p53 at the C-terminus was greatly reduced in p53.S18A MEFs after UV radiation. This could account for the impaired functions of p53.S18A cells in response to DNA damage. However, the role of C-terminal acetylation sites is debated as can be seen from mouse models described in the section ‘additional modified p53 mouse models’. Almost simultaneously, an identical p53.S18A mutant mouse models was generated by Sluss et al.
[99]. Again, an impaired apoptotic response after IR was found in splenocytes and thymocytes.
However, the proliferation rate and the IR-induced cell cycle arrest responses were not affected in p53.S18A MEFs. Finally, a reduction of transcriptional activation of the p53 pro-apoptotic gene Puma after IR was found in p53.S18A thymocytes. Again the p53.S18A mutant mice were not found to be prone to spontaneous tumor development. In conclusion, phosphorylation of Ser18 appears not to be required for normal tumor suppression, but phosphorylation of Ser18 seems necessary for some responses after UV- or IR-induced DNA damage [95;99].
Evidently, single mutations at Ser18 or Ser23 abolishing phosphorylation did not show major effects on p53 functioning. Both Ser18 and Ser23 are phosphorylated more or less simultaneously after different external stress signals like DNA damage [48], suggesting that they have synergistic roles in stress and/or tumor suppression. To investigate this hypothesis a double mutant mouse model was developed in which both Ser18 and Ser23 were mutated into alanine [100]. After IR the p53-dependent apoptotic response was completely abolished in thymocytes of these mice.
Again, the single mutants p53.S18A and p53.S23A, showed only a partial apoptotic defect.
The double mutant mice developed spontaneous tumors in different tissues, however, with longer latency times as compared to p53-/- mice. Apparently, phosphorylation of Ser18 and Ser23 play synergistic roles in the activation of p53-dependent apoptosis, and are essential to suppress spontaneous tumorigenesis [100]. Recently, a serine to alanine missense mutation was introduced into the human knock-in (HUPKI) allele in mice [101] showing phosphorylation of p53.S46 is important in p53-dependent apoptotic responses. However, in contrast to the previous mentioned mouse models this Ser46 site is absent in mice and seems to have a murine homologue at Ser58 [102].
Taken all these in vivo results together, one could conclude that defects in p53 phosphorylation lead to altered, mostly intermediate, cellular and molecular p53 responses, underlining the idea that fine-tuning of p53 functions occurs through post-translational modifications.
Additional modified p53 mouse models
Besides models with defects in phosphorylation sites, series of mouse models and cells were generated that carry other mutations in the p53 protein. A mouse model was generated with mutations at both Leu25 and Trp26 (referred to as p53QS) in order to investigate the mechanism of tumor suppression through p53. Both residues appear in vitro to be essential for transcriptional
transactivation [103]. Well known p53 responses like cell cycle arrest and apoptosis as well as binding to Mdm2 were adversely affected in MEFs and thymocytes isolated from this mutant mouse model [103]. However, later on it appeared that this model also accidentally carried a mutation in the DNA binding domain at codon 135 (Trp53QS-Val135) which likely had a dominant effect on the obtained results. Clearly, the Val135 mutant by itself appears to have compromised p53 functions, since it is a well known mutant showing dominant-negative phenotypes [104].
Recently, a new mutant knock-in transactivation model p53QS was generated in which the same residues crucial for transactivation were mutated, namely L25Q and W26S (referred to as p53QS), but this time without additional p53 modifications [105]. Mdm2 was unable to bind p53QS mutant protein in MEFs, which is in line with the previous study [103]. Upon doxorubicine treatment transcription kinetics of several p53 target genes were compromised in p53QS MEFs. To find out whether this mutant still binds to DNA, the DNA binding activity was tested by performing EMSA analysis on the promoter of Cdkn1a and on the first intron of Perp (both p53 target genes). Even without DNA damage, p53QS was found to be able to bind effectively to these sites. So the p53QS mutant is able to bind to p53 response elements, but appears deficient in transactivation. Apparently, transactivation of target genes needs more than just binding of p53 to its responsive elements. Probably binding of (specific) p53 co-activators is adversely affected in this p53QS mutant. This hypothesis might be true for some target genes, but certainly not all, since transcription of some apoptotic genes like Bax and Apaf1 was still active/measurable, pointing towards different mechanisms of transactivation, which might be explained by the idea that the p53 N-terminal domain has two, largely independent functioning, transactivation domains.
Consequently, well-known functions of p53 were altered, as G1 cell cycle arrest was partially affected and the induction of apoptosis was completely or partially affected depending on the specific apoptotic trigger introduced; i.e., DNA damage-induced apoptosis was completely inhibited whereas serum deprivation was only partially affected. Finally, hypoxia-stress still showed substantial apoptotic activity. In summary, these experiments show that the L25Q and W26S transactivation residues are needed in some but not all apoptotic responses, and co- activators and co-repressors may influence the final outcome in a DNA-damage-dependent way as was discussed earlier by Johnson et al. [5;105;106].
Recently, another mouse model was generated, with a deletion at the N-terminal part of p53. A deletion of the proline-rich domain, also known as PRD, (referred to as p53∆P) was introduced in these mice, removing all PXXP motifs and PIN1 sites of the murine PRD [107]. This PIN1 site is important for enabling the recruitment of CHK2 to phosphorylate Ser20, and in that way reduces the binding of Mdm2 [108]. The PXXP motifs are furthermore needed for optimal p53- p300 interactions. P53∆P is a less efficient transactivator compared to wild-type with also target gene specific differences (1.5 times lower Mdm2 and Puma and 3 times lower p21 and Noxa mRNA levels after exposure to adriamycin). Furthermore, it was shown that PRD is essential for the cell cycle arrest response, as p53∆P MEFs fail to undergo arrest after gamma irradiation and adriamycin treatment that induced cell cycle arrest in wild-type MEFs. However, analysis of the apoptotic response in E1A over-expressing MEFs showed a lower apoptotic response in p53∆P cells after exposure to 2.5 µM etoposide, whereas 10 µM etoposide induced the same apoptotic response in both genotypes. Gamma irradiation-induced apoptosis was also compromised in E1A MEFs as well as in thymocytes.
Two other mouse models were generated to find out whether PIN1 sites in the PRD are
responsible for the phenotype of the p53∆P mutated mouse model; these models were named p53TTAA (eliminating PIN1 sites but leaving PXXP motivs intact) and p53AXXA (mutating proline into alanines but leaving one PIN1 site intact) [109]. P53TTAA accumulation is compromised after gamma irradiation whereas p53AXXA showed a normal accumulation. In that way the PIN1 site seems to affect the regulation of p53 stability after stress. However, the transactivation of Mdm2 and p21 was again normal for both models in contrast to p53∆P mice, which showed decreased induction of Mdm2 en p21 mRNA levels after gamma irradiation. Finally, all 3 mutant models were tested for suppression of oncogene-induced tumors. P53-/- and p53∆P were unable to suppress oncogene-induced tumors whereas p53TTAA and p53AXXA both were able to suppress them.
The PRD seems, therefore, important for the regulation of p53 stability, transactivation ability and induction of transcription-independent apoptosis. The PIN1 sites alter the stabilization of p53, but are not essential for p53 activation or oncogene-induced tumor suppression, whereas PXXP motifs have no effects on any of these processes. An explanation of the authors is that the PRD acts like a spacer between the transactivation domain and the DNA binding domain.
As this spacer is shortened in the p53∆P mutant mouse model this may be changing the p53 structure thereby changing the ability to activate transcription or interaction with other proteins [107;109].
Mutations of functional residues at the N-terminal part of p53 seem to influence transactivation, whereas those found at the C-terminus are thought to be important for both the stability of the protein and/or its transactivation function [110]. The importance of seven C-terminal lysines was tested in mice by changing all the lysines into arginines (Lys367, 369, 370, 378, 379, 383 and 384), resulting in a p537KR mutant mouse model. The mice are viable and phenotypically normal. MEFs and thymocytes of the p537KR mice showed no aberrant changes in stabilization or cell cycle arrest and apoptotic functions upon gamma irradiation compared to wild-type cells. So, the tested Lysine residues are not required for p53 stability or its transactivation function, but are probably needed for fine-tuning the different p53 stress responses. It was shown that in the absence of these C-terminal modifications the protein is equally or more active (in some cases; i.e., thymocytes after gamma irradiation) than wild-type in target gene induction. In the in vivo situation where p537KR is more stable and transcriptionally active, it was suggested that the mutant form is less efficiently cleared from promoter regions of the target genes and that this is the cause of the observed higher transcriptional activity [110]. So, this study indicated that the C-terminal lysine residues are not important for the regulation of p53 stability. However, Chao et al. recently generated a physiologically relevant mouse model with a missense mutation at the C-terminal part of p53 (Lys317 to Arg317), as there were other reports indicating this acetylation site positively regulates the p53 response [111]. MEFs of the p53.K317R mice showed normal G1/S cell cycle arrest after UV radiation and furthermore no difference in expression levels of p53 target genes (p21, Pidd, Noxa and Mdm2) after UV, IR or doxorubicin treatment was found compared to wild-type. The p53-dependent IR-induced apoptotic response and the expression levels of Noxa and Pidd in p53.K317R thymocytes were increased compared to wild-type. This increased apoptotic response compared to wild-type was also found in small intestinal epithelial cells and retina of the mutant mice and also in E1A/Ras expressing MEFs. Finally, to test the effects of the p53.K317R mutation on gene expression a microarray analysis was performed using untreated and ionizing radiated thymocytes. Most of the differentially expressed genes were found repressed after IR. In general it was shown that the
acetylation of lysine 317 negatively modulated the activity of p53 as a transcriptional activator or repressor for a subset of target genes.
In summary, acetylation of Lys317 decreases the apoptotic response in several cell types, probably to prevent too much apoptosis. Furthermore, it affects gene expression of some p53 target genes.
So, a specific modification of p53 can differentially affect the expression of p53-regulated genes and affect the outcome of the cellular stress response.
P53 and transcriptional activation of target genes
The presence of a transactivation domain in p53 was first described around 1990 by several groups [112;113]. With these findings the molecular function of p53 as a transcription factor was postulated. El-Deiry et al. showed that p53 binds in vitro specifically to target DNA fragments having a symmetrical structure of two copies of the DNA sequence elements (5'- PuPuPuC(A/T)(T/A)GPyPyPy-3') [114]. The first identified p53 target gene after the discovery of the p53 binding site was WAF1 (later referred to as p21), which showed to be an important mediator of p53-dependent tumor growth suppression [115]. The second identified p53 target gene was Mdm2 which operates as an antagonist of p53 [116].
P53 was found to cooperate as a transcription factor with an extensive set of co-activators or co-repressors. Examples of these are acetyl transferase enzymes like p300 and CBP that were shown to associate with p53 as co-activators, thereby up-regulating p53 functioning in cell cycle control and apoptosis [117;118]. Furthermore, members of the p300/CBP family were identified as complex integrators of signals that regulate p53. When p53 interacts with p300/
CBP its transactivation function can either be positively or negatively regulated, depending on the cellular context, environment and damage introduced [reviewed in [119]]. The DNA binding affinity of p53 seems to be influenced by complexing with members of the ASPP family, which results in an enhanced (ASPP1 and ASPP2) or decreased (iASPP) transactivation of pro- apoptotic p53 target genes [120;121].
P53-induced apoptotic responses can be divided into an intrinsic and an extrinsic pathway as depicted in Figure 5 [122]. Death receptors belonging to the tumor necrosis factor receptor (TNF-R) family lead to apoptosis through activation of caspases 8 and 3, the intrinsic pathway.
A known p53 target triggered by this receptor-family is Killer/Dr5 (Tnfrsf10b) [123]. Through the extrinsic pathway depolarization of mitochondria followed by release of cytochrome c into the cytoplasm is taking place. In this pathway p53 targets like Bax, Puma and Noxa (Pmaip1) play important roles [reviewed in [21]]. Both pathways are not strictly distinct from each other, in some cases they cooperate in inducing apoptotic responses.
Numerous apoptosis-related p53 target genes have been identified up to now. For a complete overview of the p53-related apoptotic response the reader is referred to a review of Haupt et al.
[122]. Examples of the pro-apoptotic Bcl2-family members are Bax and Bcl-2 [124;125]. Other Bcl2-family members are Noxa and Puma, which are BH3-domain-only proteins [126-128].
These pro-apoptotic proteins act on the mitochondrial membrane resulting in a leakage of cytochrome c into the cytosol. Furthermore, the activation of the p53 target gene Apaf-1 can be triggered [129], followed by binding of Apaf-1 protein to downstream caspases [130].
Besides apoptotic-related p53 target genes also cell cycle control genes were described like p21, which is able to inhibit CDKs, resulting in a G1/S phase inhibition [131;132]. Furthermore, Sfn (14-3-3 sigma) was found to be a p53 target gene involved in cell cycle control, but in contrast to p21 the Sfn protein displays G2/M arrest abilities [133;134]. Besides Sfn, other
p53 target genes are known to play a role in G2 arrest. These are Reprimo [135], B99 [136]
Gadd45 [137] and Mcg10 [138]. Clearly, less p53 target genes are identified playing a role in cell cycle control compared to those involved in apoptotic responses. This might indicate that the repertoire for cell cycle arrest is less complicated and as such requires less protein factors, but one can also envision that many p53 targets are still undetected. Finally, many genes involved in cell cycle arrest might be beyond the control of p53 and act through different, p53-independent, mechanisms.
As previously described, a p53QS model, mutated in the transactivation domains Leu25 and Trp26, showed that p53 was largely compromised for transactivation, but still retains some biological activity depending on stress stimuli and cell types considered [105]. Still though some transcription activity of some apoptotic genes was measured, which is in line with the finding that p53QS still binds to response elements in these genes. As already discussed above, it seems that activation of these target genes depends on both binding to p53 and assisting co-activators.
Probably, these latter factors do not bind efficiently anymore in the p53QS mutant.
Finally, it was shown that p53 employs distinct mechanisms to transactivate different target genes. This feature was discovered in the p53QS mutant, since it is compromised in activating some p53 target genes, but had wild-type activity in activation of the transcription of others.
Another requirement for appropriate transcription is chromatin remodeling [139]. To that end, p53 needs to interact with proteins like TBP and p300 (components of the basal transcriptional apparatus and histone acetyltransferases). In the p53QS mutant these interactions are still intact [5]. As there was already an indication for two activation domains in p53 [140], these results also indicate probably another activation site is needed for the transactivation. AD1 is, for instance, important for most genes whereas AD2 is important for some other genes like Bax and Apaf1 which are still transcriptionally activated by the p53QS mutant possessing mutations in AD1 [5;105].
The p53-dependent decision between apoptosis and cell cycle arrest
There is still extensive debate ongoing on how p53 directs the cell into the apoptotic or the
Adapted from: S. Haupt et al., J Cell Sci, 2003
Figure 5 - Model for p53-mediated apoptosis
Apoptosis can be initiated through two main pathways, either at the membrane (extrinsic) or the mitochondria (intrinsic). Both pathways ultimately result in a common pathway involving the activation of cleavage of several caspases leading to apoptosis. Here, the involvement of p53 in the extrinsic and intrinsic apoptotic pathways is depicted. Known p53 target genes in this figure are: Perp, Dr5, Fas, Bid, t-Bid, Noxa, Puma, Bax, Apaf-1 and Caspase-6.
cell cycle control processes. Different cells respond in an entirely diverse manner to a certain cellular stressor, and specific cells respond differently to a variety of cellular stressors [141].
An explanation for these differences could be the presence of differently post-translationally modified (i.e., phosphorylation, acetylation etc.) p53 proteins, or differences in total p53 protein levels that are thresholds for the apoptotic process. A generally accepted model for the choices of different p53-directed responses was reviewed by Meek [142], showing p53 is able to transactivate promoters for cell cycle control genes or apoptotic genes distinctively based on the expression levels of specific factors, mostly present in the upstream route of p53 regulation and activation. Chk2, JMY, p63/p73 and ASPP1/2 are indicated as favoring the apoptosis route, whereas presence of ATM is indicated as favoring transcription of cell cycle arrest genes.
Furthermore, the mitochondrial apoptosis pathway is favored by activation of WISP-1, which on its turn activates the AKT pathway. (Co-) activators like pro-apoptotic proteins can influence this process by favoring a certain class of promoters [143]. An explanation for another choice between cellular responses, namely apoptosis and DNA repair, was depicted in a model published by Offer et al. [144]. In this study lowly stressed cells were directed towards DNA repair whereas highly stressed cells were directed to the apoptotic pathway. These findings confirmed previously found data [145;146]. Apparently, levels of DNA damage influence the choice of target gene activation, and consequently the appropriate p53-dependent response occurs. Interestingly, recent findings showed phosphorylation of Ser46 is important in switching from cell cycle arrest towards apoptotic conditions [147]
P53 and repression of target genes
Many studies focussed on transactivation of genes by p53 [148;149], but there is also evidence that p53 can repress the transcription of target genes both in apoptosis and cell cycle arrest pathways [150]. It appeared that p53 can selectively repress genes to affect a cellular response through direct binding to a specific consensus DNA binding element [151]. Recently, however, it was shown that p53-dependent repression in response to DNA-damage was also possible without binding to this consensus site. P53 may act directly or indirectly via a DNA binding protein complex at promoters of repressed genes. So, p53-induced cell cycle arrest is a function of not only the transactivation of cell cycle inhibitors as p21, but also the repression of targets that regulate proliferation at several distinct phases of the cell cycle.
Another possible mechanism of repression involves the NF-Y transcription factor [152]. NF-Y is known for its regulation of promoter sites containing multiple CCAAT boxes. Many genes controlling the G2/M transition contain these boxes and are therefore regulated in a cell cycle- dependent manner by NF-Y [153]. Negative regulation of ‘G2/M’ promoters after DNA damage or p53 expression seems to be dependent on NF-Y activity [154;155]. Furthermore, p53 binds to promoters of Cyclin B2, Cdc2, and Cdc25c genes, dependent on NF-Y at regions without consensus p53 DNA-binding sites [156].
Gene expression profiling to study p53-dependent transcriptional responses
As p53 is a well-known transcription factor, and the human and mouse genomic sequences became available, many new innovative technologies like microarray analysis were used to analyze gene expression levels on a genome wide scale in the context of p53 functionality.
For instance, gene-expression profiles of cells carrying the temperature-sensitive p53.Val135 mutant, driving cells into apoptosis upon shifting the temperature to 32 degrees, were compared
