Regulation of cellular senescence <i>via</i> the <scp>FOXO</scp>4‐p53 axis

Forkhead box O ( FOXO ) and p53 proteins are transcription factors that regulate diverse signalling pathways to control cell cycle, apoptosis and metabolism. In the last decade both FOXO and p53 have been identified as key players in aging, and their misregulation is linked to numerous diseases including cancers. However, many of the underlying molecular mechanisms remain mysterious, including regulation of ageing by FOXO s and p53. Several activities appear to be shared between FOXO s and p53, including their central role in the regulation of cellular senescence. In this review, we will focus on the recent advances on the link between FOXO s and p53, with a particular focus on the FOXO 4‐p53 axis and the role of FOXO 4/p53 in cellular senescence. Moreover, we discuss potential strategies for targeting the FOXO 4‐p53 interaction to modulate cellular senescence as a drug target in treatment of aging‐related diseases and morbidity.

p53 is a homo‐tetrameric transcription factor composed of an N‐terminal trans‐activation domain (TAD, residues 1–61), a prolin‐rich domain (PR, residues 64–92), a central DNA‐binding domain (DBD, residues 93–293) followed by a tetramerization domain (TD, residues 325–355) and the C‐terminal negative regulatory domain (NRD, residues 367–393) (Fig. 1 ). p53 is involved in the regulation of more than 500 target genes and thereby controls a broad range of cellular processes, including metabolic adaptation, DNA repair, cell cycle arrest, apoptosis, and senescence (for review 18 ). The multifaceted function of p53 is closely related to its multi‐functional domain architecture making p53 a scaffold molecule acting as a central protein‐protein interaction hub in a plethora of biological networks. These p53 interaction partners allow for tight regulation of p53 localization, stability and transcriptional activity. Biochemical and structural biology work revealed that these interaction partners share common binding motifs on p53 and through this modulate p53 function and activity (for review 19 ). The N‐terminal p53 transactivation domain, for example, is largely disordered but possesses two binding motifs with α‐helical propensity, named TAD1 (residues 17–29) and TAD2 (residues 40–57). These two motifs act independently or in combination in order to allow p53 binding to several proteins regulating either p53 stability, such as trough the mouse double mutant 2 homolog (MDM2) 20 , or transcriptional activity, such as trough binding of the transcriptional coactivator p300 21 or high mobility group B1 (HMGB1) 22 . As an additional layer of regulation of p53 activity the TAD domain is highly phosphorylated in cells by several kinases including ATM 23 , casein 1‐like kinase (CK1) 24 , downstream checkpoint kinases 1 and 2 (Chk1/2) 25 , 26 and homeodomain‐interacting protein kinase 2 (HIPK2) 27 . Phosphorylation of the TAD has been shown to modulate its binding to MDM2 and p300, therefore acting on the regulation of p53 stability and transcriptional activity (for review 28 ). The DNA‐binding domain of p53 is also involved in mediating protein‐protein interaction. For example, three dimensional structures of the p53‐DBD in complex with two apoptotic factors, the apoptosis‐stimulating of p53 protein 2 (ASPP2) and the B‐cell lymphoma‐extra large (Bcl‐xL) have been solved and revealed that both factors interact with the DNA‐binding interface of p53 and therefore potentially compete with DNA‐binding in the nucleus (Fig. 1 ) 29 , 30 . Interestingly, ASPP2 and Bcl‐xL harbour opposite function in the regulation of p53‐mediated apoptosis, with pro versus anti‐apoptotic function, respectively. This suggests different mode of p53 regulation, such as transcription dependent/independent, despite involving similar binding properties. Communication between p53 domains has been shown to modulate p53 function. For example, binding of the N‐terminal proline‐rich region of p53 to its DNA‐binding domain increases p53 stability by inhibiting aggregation of the tetrameric form 31 . Moreover, intermolecular tetramerization of p53 increases the strength of individual binding interactions trough avidity and is key for the regulation of p53 binding to the promoter region of its target genes and protein partners (for review 32 ). Indeed, p53 tetramerization markedly increases its ability to bind DNA compared to the monomeric form 33 . Understanding the detailed molecular mechanisms of p53 interactions to its different binding partners is primordial in order to fully understand the key steps allowing p53 activation and subsequent regulation of cell fate determination. Figure 1 Structural organization of FOXO s and p53 and post‐translation modifications involved in senescence. Forkhead box class “O” ( FOXO ) proteins are composed of an N‐terminal Forkhead domain ( FH ) followed by a nuclear localization signal ( NLS ) and a C‐terminal transactivation domain ( TAD ). p53 is composed of an N‐terminal TAD domain, a proline‐rich domain ( PR ) followed by a DNA ‐binding domain ( DBD ), a NLS , a C‐terminal tetramerization domain ( TD ) and a negative regulatory domain ( NRD ). Most relevant post‐translationally modified residues in p53 and FOXO s and the corresponding enzymes involved are annotated in blue, red or black if they are either involved in inactivation, activation or both, respectively. FOXO and p53 binding partners discussed in this review are indicated in cyan or green, depending on whether they are common or specific to FOXO s and/or p53. The available three dimensional structure of the corresponding complexes are shown in cartoon ( FOXO 4/ DNA : PDB ID 3L2C 57 ; FOXO 3/ CBP : 2 LQH 64 ; p53/ DNA : 3 IGK 125 ; p53/ MDM 2: 1 YCR 35 ; p53/ HMGB 1: 2 LY 4 22 ; p53/p300: 2 MZD 107 ; p53/ ASPP 2: 1 YCS 29 ; p53/Bcl‐ xL : 2 MEJ 30 ). Because of its pro‐apoptotic function, p53 is recognized as tumour suppressor, and is found mutated in more than hal

Intense oncogenic signalling has been described as a potent tumour suppressive mechanism able to bypass tumour progression by triggering cellular senescence 14 . Given FOXO's functions as tumour suppressor, several studies contributed to assess the role of FOXOs in oncogene‐induced senescence (OIS). In these studies, the B‐raf proto‐oncogene (BRAF) V600E variant has been used as model of OIS. Indeed, BRAF V600E variant is expressed in several tumours, especially in melanoma 76 , 77 . This mutation increase BRAF kinase activity resulting in increased mitogen‐activated protein kinase MEK‐associated signalling. Aberrant activation of BRAF oncogene leads to an increased cell proliferation and cell cycle progression but also ultimately to oncogene‐induced senescence 78 . De Keizer et al ., demonstrated an important role of FOXO4 in BRAF‐associated oncogenic signalling. Indeed, ectopic expression of FOXO4 in cancer cell lines expressing the BRAF mutant induces senescence 74 . The oncogenic form of BRAF phosphorylates the JNK kinase via the MEK pathway. In turn, activated JNK can phosphorylate and activate FOXO4 leading to cell cycle arrest and induction of cellular senescence. This was associated with increased cellular ROS levels which also contribute to MEK‐JNK‐pathway associated active phosphorylation of FOXO4 73 , thus acting as a positive feedback loop. Activation of FOXO4 is correlated with an increased transcriptional activation of p21 and subsequent activation of cellular senescence 74 . Conversely, another study showed that in mice inactivation of FOXO4 results in senescence bypass and consequently restores oncogenic‐BRAF induced melanoma formation 79 . FOXO4‐mediated activation of cellular senescence has also been observed in endothelial progenitor cells 80 , 81 . Ectopic expression of the cleaved high molecular weight kinogen (HKa) accelerates the onset of endothelial progenitor cells senescence and correlates with an increased level of intra‐cellular ROS. As discussed before, increased ROS levels result in JNK‐mediated activation of FOXO4 leading to cellular senescence. All these studies provide strong evidence that oncogenic signals can promote FOXO4 activation of cellular senescence which serves as a defence mechanism in order to prevent cell hyper‐proliferation. Therefore, apart from FOXO function in apoptosis, cellular senescence also contributes to make FOXO transcription factors potent tumour suppressors (Fig. 2 ).

Overlapping functions of FOXOs and p53 in cell‐cycle regulation and tumour suppression are suggested by several studies in mice. Indeed, p53 knockout transgenic mice are prone to increased tumour progression including lymphoma and hemangiomas 85 , 86 . Concomitantly, FOXO1, 3 and 4 triple knockout mice are also characterized by thymic lymphoma and hemangiomas 87 . These observations suggest that FOXOs and p53 play similar functions at least in suppressing tumorigenesis. Consistently, FOXOs and p53 share several downstream target genes including WIF1, FASLG, GADD45, PA26 and p21 88 , 89 . Interestingly, p53 and FOXOs recognize different sequence‐specific DNA‐binding element on the promoter site of their target genes suggesting that they can act in a cooperative manner in order to regulate gene transcription. Indeed, dual binding of p53 and FOXOs at common promoter sites could (a) increase affinity/specificity for DNA binding by modulating the structural conformation of DNA (b) modulate the binding of common transcriptional regulators such as p300/CBP and/or the transcriptional machinery. Given the importance of both FOXO4 and p53‐mediated transcriptional activation of p21 in induction of cellular senescence, one can suggest functional interconnection between the two proteins. Indeed, growing evidence illustrate several links between FOXOs and p53 signalling highlighting by the fact that FOXOs and p53 either share common post‐translational modifications or are involved in inter‐regulation of their specific PTMs and more importantly that they directly interact both in vivo and in vitro . Furthermore, p53 and FOXOs share similar biological functions notably in controlling the balance between cell death and survival therefore limiting cell proliferation in stressed cells. For these reasons, they are both considered as bona fide tumour suppressor. Co‐regulation of p53 and FOXOs by PTMs Acting similarly as Akt, serum/glucocorticoid regulated kinase 1 (SGK1) is involved in inactivation of FOXOs. Indeed, SGK1 is also activated by the PI3K signalling pathway. PI3K‐mediated phosphorylation and activation of SGK1 promote cell survival. In turn, SGK1‐mediated phosphorylation of FOXOs induces FOXO translocation from the nucleus to the cytoplasm thereby suppressing FOXO‐dependent transcription and their function in cell cycle arrest and apoptosis 90 . p53 is involved in the regulation of SGK1 at different levels. First, SGK1 is a downstream target gene of p53 and is upregulated upon p53 activation 91 . Second, genotoxic stresses induce p53‐dependent activation of the extracellular signal‐regulated kinases 1 and 2 (ERK1/2) leading to SGK1 activation and subsequent inactivation of FOXO3 92 . Third, SGK1 activates MDM2‐dependent degradation of p53 suggesting a fine‐tune feedback‐loop where p53 can control its own expression level as well as FOXO transcriptional activity 93 . The E3 ubiquitin ligase MDM2 also harbours distinct functions involved in the regulation of both p53 and FOXOs. Indeed, despite its function in p53 poly‐ubiquitination and proteasomal‐mediated degradation, MDM2 also controls the expression level and activity of FOXOs. MDM2 can directly interact with FOXO4 and catalyses multiple events of FOXO4 mono‐ubiquitination. Expression of increasing amounts of MDM2 in human breast adenocarcinoma MCF‐7 cells results at a low amount of MDM2 in an increase in FOXO4 transcriptional activity whereas expression of high amounts of MDM2 show opposite effects 94 . Furthermore, mono‐ubiquitination of FOXO4 is inducible by oxidative stress. Hydrogen‐peroxide treatment of hypothalamic A14 catecholamine cells induces mono‐ubiquitination of FOXO4 and results in FOXO4 re‐localization towards the nucleus and subsequent increase in transcriptional activity 95 . In contrast to FOXO4, MDM2‐dependent ubiquitination of FOXO1 and 3 results in their proteasomal degradation and transcriptional inactivation in a p53‐dependent manner 38 . The exact mechanisms involving p53 in this process remain to be discovered. The antagonistic function of MDM2 in FOXO4 versus FOXO1 and 3 activities is another clue that could help understanding the opposite functions of FOXOs in cellular senescence. It is also worth noting that the enzyme responsible for de‐ubiquitination of FOXOs and p53 is also identical and named USP7 95 , 96 as well as the histone acetyltransferase p300 and the deacetylase sirtuin‐1 (SIRT1) which are involved in both FOXOs and p53 acetylation and deacetylation, respectively 97 , 98 , 99 , 100 , 101 , 102 .

Given the importance of both FOXO4 and p53‐mediated transcriptional activation of p21 in induction of cellular senescence, it has been hypothesized that FOXOs and p53 could share similar molecular mechanism for the control of cellular fate and senescence. Indeed, several hypotheses could explain how p53 and FOXOs co‐regulate the transcription of common target genes: (a) direct interaction at the promoter site of p21 leading to enhanced transcriptional activity (b) co‐recruitment of common transcriptional co‐activators (c) inter‐regulation of their PTMs resulting in increased stability and nuclear localization. Molecular and structural studies on such networks are lacking in order to provide answer, nevertheless, several studies clearly demonstrate a join function of p53 and FOXOs in the activation of cellular senescence. In human endothelial cells, overexpression of constitutively active form of Akt results in cell growth arrest and senescence‐like phenotype reducing the lifespan of these cells and correlates with and elevation of intra‐cellular levels of p21. This mechanism is p53‐dependent as inactivation of p53 in this system impairs the transcriptional up‐regulation of p21 and subsequent senescence‐like cell growth arrest. Given the fact that FOXO proteins are direct targets of Akt signaling, the authors assessed the role of FOXO3 and demonstrated that constitutive activation of Akt inhibits the transcriptional activity of FOXO3 leading to a decrease in the expression of ROS detoxifying enzymes and therefore increase in intra‐cellular ROS level. The authors conclude that the increase in ROS levels promotes senescence‐like growth arrest via a p53‐mediated activation of p21 expression 104 . Here again, the specific function of FOXO4 in this system was not assessed. Indeed, accumulation of ROS can lead to an activation of the DDR response and subsequent activation of p53 105 , but could also lead to a specific activation of FOXO4 via the ROS/JNK and/or ROS/MDM2 pathways as discussed previously 73 , 74 , 75 , 95 and possibly contributes to the observed senescence‐like phenotype. Recently Baar et al ., characterized the specific function of FOXO4 in p53‐mediated senescence response. They first observed that ionizing‐radiation (IR) induced senescence of human IMR90 fibroblasts results in increase FOXO4 expression and no detectable change in FOXO1 and 3 compared to non‐senescent cells suggesting a specific function of FOXO4 in senescence. They showed that upon IR‐damage FOXO4 promotes senescence over apoptosis and maintains the viability of senescent cells by repressing their apoptotic response. Mechanistically, FOXO4 directly binds to activated p53 in promyelocytic leukemia (PML) bodies at the site of DNA‐damage 106 and activates the p53‐dependent transcription of the senescence‐associated p21 gene 103 . It is worth noting that the promoter of the p21 gene is composed of a canonical FOXO target sequence flanked by two p53 binding sites. The exact molecular mechanisms allowing FOXO4‐dependent p53 activation of p21 transcription are unknown but one can speculate that dual binding of p53 and FOXO4 at the promoter site of p21 can stimulate its transcription compared to a situation where a single transcription factor is available. More studies have to be done to reflect if such mechanism is plausible and in turn how the formation of this potential ternary complex FOXO4/p53/DNA could stimulate gene transcription. FOXO4 and p53 transcriptional activity are tightly regulated via the histones acetyl‐transferase (HATs). Indeed, FOXOs can recruits CREB‐binding protein (CBP)/p300 to the promoter of its target genes leading to the transactivation of the targets and similarly for p53 101 , 102 , 107 . FOXOs binding to p53 could stabilize the HATs complex formation and thereby FOXOs‐p53 dependent gene transcription. Alternatively, FOXO4 binding to p53 can interfere with p53 binding to different co‐repressors and therefore increase p53 activity. For example, FOXOs and MDM2 share a common binding site on p53. This raises the question whether FOXO4 can compete with MDM2 for p53 binding. Allosteric clashes between FOXO4 and MDM2 for p53 binding might lead to MDM2 displacement from p53 resulting in decrease p53 proteasomal degradation and increased stability.

As previously discussed, Baar et al . 103 , showed that FOXO4 can promote senescence by interacting with p53 at the sites of DNA‐damage thereby up‐regulating the transcription of the senescence master regulator p21. Therefore, they decided to design a peptide specifically targeting this interaction in order to prevent senescence. The specificity of this peptide arises from its design that is based on previously published data on the FOXO3/p53 interaction 62 . In order to increase specificity, this peptide contains a region of the p53 binding site on FOXO4 which differs in primary amino‐acids sequence of FOXO1 and 3. The peptide has been designed in a D‐retro‐inverso (DRI) manner that has been shown previously to increase drug potency both in vitro and in vivo 123 . We demonstrated using NMR that FOXO4‐DRI peptide competes with FOXO4 for p53 binding and therefore can efficiently interfere with the FOXO4 ‐ p53 interaction. In turn, when introduced to senescent human IMR90 fibroblasts the FOXO4‐DRI peptide reduces cell viability more than 10 times compared to non‐senescent IMR90 or other cell‐type. At low dose of administration FOXO4‐DRI appears to be more selective toward senescent cells than the pan‐BCL inhibitor ABT‐737 that highlight the efficiency of such targeted approach for drug design. Mechanistically, FOXO4‐DRI promotes nuclear exclusion of active p53. Once released in the cytosol active p53 induces cell‐intrinsic apoptosis by activating a caspase‐dependent pathway likely involved in previously reported transcription‐independent p53‐mediated apoptosis at the mitochondria 124 . In vivo injection of FOXO4‐DRI in fast aged Xpd TTD/TTD and naturally aged p16: MR mice decreases senescence and counteracts subsequent loss of renal function 103 . To summarize, rational design of drug candidates targeting one specific protein‐protein interaction will allow to efficiently and electively target and selectively eliminate senescent cells. Moreover, it will provide insights into the mechanisms involved in FOXO4‐p53 co‐regulation of cellular senescence.

Research data pertaining to this article are located at figshare.com: https://dx.doi.org/10.6084/m9.figshare.6176126 .