| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Published online before print May 22, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Physiological Chemistry and Center for Biomedical Genetics, University Medical Center Utrecht, The Netherlands; and
Division of Molecular Biology, The Netherlands Cancer Institute, Amsterdam
Correspondence: René H. Medema, Division of Molecular Biology, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands. E-mail: r.medema{at}nki.nl
 |
ABSTRACT |
|---|
Key Words: apoptosis • cell cycle • oxidative stress • cytokines
|
INTRODUCTION |
|---|
In fact, constitutive activation of PI-3K and its effector PKB/Akt is a very frequent event in human cancers [5 ]. Such activation can occur through mutational loss of the tumor suppressor PTEN, Ras activation, or oncogenic receptor tyrosine kinase activation. Much research over the last two decades has addressed the consequences of activating PKB/Akt. This has unequivocally demonstrated that active PKB/Akt can drive cellular proliferation and protect cells from apoptosis and has a stimulating effect on glucose uptake and glucose metabolism [6 ].
What happens, however, when PKB/Akt is silent, a situation with which the majority of the cells in our bodies are confronted? Recent work has demonstrated that a subfamily of Forkhead transcription factors, the FOXO factors, then regulate a genetic program that is not only essential for the induction of quiescence but is also required for long-term survival of quiescent cells [7 ,8 ]. In addition, FOXO factors play an important, active role in the apoptosis of hematopoietic cell types, which cease to proliferate in the absence of appropriate mitogenic signals [9 10 11 12 ].
This review summarizes our current understanding how FOXO factors can regulate these very diverse processes. We will first discuss the function and regulation of FOXO activity itself, followed by an overview of the genes that appear to be crucial targets of FOXO in the control of quiescence, oxidative stress, and apoptosis.
|
FORKHEAD TRANSCRIPTION FACTORS |
|---|
-helices, three ß-sheets, and two loop regions located at the C-terminal end that shape the wings of the structure [16
17
18
19
20
]. As a result of these structural characteristics, the Forkhead box is sometimes referred to as the winged-helix motif. Recently, a unifying nomenclature has been adopted, and Forkhead factors are now denoted in general as Fox (Forkhead box) factors [21
]. Outside the Forkhead box, the different members are highly divergent, and based on phylogenetic analysis, Forkhead proteins have been assigned to 17 subfamilies ranging from FOXA to FOXQ [21
] (see <www.biology.pomona.edu/fox.html> for the most recent update). Of these, the FOXO factors are the only ones known to date to be regulated by the PKB/Akt pathway, and others clearly function in distinct, cellular processes. DNA binding of Forkhead proteins relies on interactions between the third helix in the Forkhead box, the so-called "recognition helix" H3, and DNA bases within the major groove of double-stranded DNA [16 ]. Residues present in the two loops make additional contacts with the DNA-binding element, and these interactions may contribute to binding-site selectivity of the different Forkhead proteins [16 ,22 ]. Also, from the three-dimensional structures, it is clear that the orientation of helix 3 is not identical in all Forkhead members, indicating that this could allow discrimination among different binding elements [16 17 18 19 20 ]. Binding-site selection performed with a variety of Forkhead proteins has led to the identification of a core recognition motif of 7 bp: T-(G/A)-T-T-(G/T)-(G/A)-(C/T) [22 23 24 25 ]. This core region is necessary for Forkhead binding, and bases immediately flanking the core contribute to binding specificity of the different family members [22 ,24 25 26 ]. The optimal DNA-binding site for the FOXO members for example has been determined as TTGTTTAC [27 ].
Members of the Forkhead family of transcription factors have been shown to play important roles in cell proliferation, differentiation of a variety of cell lineages, tissue-specific gene expression, and embryogenesis [28 ]. As a matter of fact, many homozygous knockout mice of Forkhead factors have been described that show some type of developmental defect. For example, the knockout of Foxa2 (HNF-3ß) is embryonic-lethal and shows defects in the formation of notochord, neurotube, gut endoderm, and somites [29 ,30 ]. This mutant phenotype is reproduced in mice deficient in Foxh1 (FAST1), and it was demonstrated that Foxh1 regulates the expression of Foxa2 to control patterning of the anterior primitive streak and node [31 ]. Knocking out Foxd1 (BF-2) results in retarded kidney development [32 ], and Foxg1-deficiency (BF-1) leads to aberrant development of the cerebral hemispheres [33 ]. Mice deficient in Foxm1 (HFH-11/Trident) displayed a defect in the developing myocardium [34 ], and mice deficient in Foxc2 (MFH-1) show defects in aortic arch patterning and skeletogenesis [35 ]. Knocking out Foxf1 (FREAC) causes embryos to die at midgestation as a result of defects in mesoderm differentiation and cell adhesion [36 ], whereas Foxb1a (TWH) plays a role in the early differentiation of neuronal progenitors [37 ]. Finally, the nude mouse and the scurfy mouse represent two very well known examples of Forkhead-deficient mice. The immunological deficiencies and lack of hair that is so typical for the nude mouse are a result of a mutation in the Foxn1 gene (whn) [38 ], and the lymphoproliferative disorder that causes the embryonic lethality in the scurfy mouse is a result of disruption of the FoxP3 gene [39 ]. The human counterparts of these genes are involved in immune dysregulation, polyendocrinopathy, and enteropathy with X-linked inheritance syndrome and nude/severe combined immunodeficiency disease, respectively [40 41 42 ]. Clearly, Forkhead factors regulate different aspects of development and are required for the establishment of distinct embryonic tissues.
In addition to their diverse roles in development, Forkhead proteins have also been suggested to play a role in neoplasia. First, the Qin oncogene (v-qin) was derived from avian sarcoma virus 31 and is essential for the transforming capacity of this retrovirus [43 ]. The v-qin gene encodes a Forkhead transcription factor that is highly homologous to Foxg1 (BF-1). It is interesting that Qin is a potent transcriptional repressor, suggesting that its transforming capacity depends on the suppression of growth-inhibitory genes [44 ]. Second, a number of chromosomal translocations have been shown to involve Forkhead genes. The t(2;13)(q35;q14) translocation that is frequently found in alveolar rhabdomyosarcoma fuses the PAX3 gene to the C terminus of FOXO1 (FKHR) [45 46 47 ]. Similarly, in a small subset of alveolar rhabdomyosarcomas, the same carboxy-terminal portion of FOXO1 is fused to the coding region of PAX7 in a variant t(1;13)(p36q14) translocation [48 ]. PAX3 and PAX7 are members of the paired box transcription factor family, and the resulting PAX–FOXO1 fusions retain an intact PAX DNA-binding domain [49 ]. In addition, the t(X;11)(q13;q23) chromosomal translocation found in acute lymphoblastic leukemia gives rise to a fusion between the coding region of the mixed-lineage leukemia (MLL) gene, a thritorax-related transcription factor with FOXO4 [50 ,51 ]. Finally, the t(6;11)(q21;q23) chromosomal translocation causes the fusion of MLL to FOXO2 (AF6q21) [52 ].
In case of the fusions involving the PAX paired box transcription factors, it has been proposed that the DNA-binding properties of the fusion protein closely follow those of the PAX proteins, as the complete PAX DNA-binding domain has been retained in the fusions [49 ]. The Forkhead fragment would supply the fusion with a strong, transcriptional activation domain, resulting in aberrant regulation of PAX3/PAX7 target genes. For the fusions involving the MLL gene, this is less clear, as it is commonly believed that rearrangement of MLL disrupts the function of the MLL protein. It is noteworthy that in each of these translocations, a FOXO member is involved rather than another member of the Forkhead family, suggesting that some function of the FOXO factors is relevant to transformation by these particular fusions. In all of the fusions, the FOXO factors are fused to their respective partners at exactly the same amino acid, just N-terminal of helix 3, where the coding sequence is interrupted by an intron. As a result, all fusions retain the DNA-binding helix of the Forkhead box as well as the transactivation domain of FOXO, present in the carboxy-terminal end of the protein. Recent data indeed demonstrate that the FOXO transactivation domain is essential for the transforming activity of these fusions [53 ,54 ]. Moreover, it was shown that these fusions could competitively interfere with transcription and apoptosis induced by wild-type FOXO proteins and as such, contribute to malignant transformation [53 ,54 ].
|
REGULATION OF FOXO FACTORS BY PKB/Akt |
|---|
Several consensus sequences for phosphorylation by PKB/Akt could be identified in DAF-16, and three of these are conserved in the FOXO factors, the mammalian orthologs of DAF-16. Indeed, it was demonstrated by several laboratories that PKB/Akt phosphorylated FOXO1, FOXO3, and FOXO4 in vitro at three residues equivalent to T24, Ser256, and Ser319 of FOXO1 [9 ,59 60 61 ] (see Fig. 1 ). Moreover, treatment of cells with growth factors that activate PKB/Akt causes phosphorylation of FOXO members on multiple sites in vivo, including those phosphorylated by PKB/Akt in vitro [9 ,59 60 61 ]. In vivo phosphorylation of FOXO was blocked in the presence of pharmacological inhibitors of PI-3K, as well as by expression of dominant-negative mutants of PI-3K and PBK/Akt [9 ,59 60 61 ]. In contrast, expression of a constitutive, active mutant of PKB/Akt was sufficient to induce in vivo phosphorylation of FOXO on the same sites that were phosphorylated in vitro [59 ].
|
Shuttling proteins between nucleus and cytoplasm is a highly regulated process and requires accessory proteins such as importins and exportins [63 ]. At present, there is little detailed knowledge as to how FOXO shuttling is regulated. Domains within these FOXO members that could function as a nuclear export signal (NES) or nuclear localization signal (NLS) have been assigned but are supported by little experimental evidence. Only FOXO4, a nonclassical NLS that surrounds the PKB/Akt phosphorylation site Ser193 (corresponding to S256 in FOXO1), has been clearly defined [62 ]. Shuttling FOXO4 was shown to be a Ran- and Crm1-dependent mechanism, and PKB/Akt-mediated phosphorylation of FOXO4 on Ser193 functionally inactivates the NLS, causing the transcription factor to be detained in the cytoplasm [62 ]. The primary sequence of the FOXO4 NLS shows that it is not a classical NLS and indeed, none of the classical importins were found to bind FOXO4, and at present, the importin responsible for nuclear import is unknown.
Although all three FOXO members harbor a classical NES sequence, its function remains unclear. Nuclear export of FOXO members is sensitive to leptomycin B treatment, indicating a Crm1-dependent, export mechanism [60 ,62 ]. Deletion of the putative NES clearly affects nuclear export of FOXO4, yet surprisingly, Crm-1 binding appears unaffected by phosphorylation, and this deletion does not prevent binding between FOXO4 and Crm1 [62 ]. This leaves several alternative explanations. Proteins such as 14-3-3 that bind to these FOXO members could provide a NES, as was suggested for the phosphorylation-dependent, nuclear export of the cdc25 protein by the 14-3-3 family member Rad24 [64 ]. However, other reports suggest that the function of 14-3-3 binding in cdc25 shuttling might be the masking of an NLS, and the little experimental evidence available on the function of 14-3-3 binding to FOXO members suggests this a more likely possibility for FOXO regulation as well [61 ,62 ,65 ].
At present, it is not fully established at what location PKB/Akt phosphorylates FOXO. For FOXO4, there is evidence that phosphorylation occurs in the nucleus. For example, Leptomycin B treatment, which detains FOXO4 in the nucleus, does not prevent insulin-induced phosphorylation of Ser193 (corresponding to S256 in FOXO1) and also has little effect on transcriptional repression by insulin [62 ]. Indeed, PKB/Akt appears to translocate to the nucleus following growth-factor stimulation in various cell systems, although this process appears to be rather slow [66 ]. Alternatively, as will be discussed in more detail below, different kinases could phosphorylate FOXO members in different locations within the cells, and a distinct kinase could be responsible for FOXO phosphorylation in the nucleus.
Besides directly regulating nuclear import or export, FOXO phosphorylation can regulate its transcriptional activity in other ways. First, phosphorylation of Ser256 in FOXO1 was shown to directly affect DNA-binding activity in addition to its effect on nuclear/cytoplasmic shuttling [67 ]. Second, phosphorylation might disturb FOXO interaction with cofactor(s). Indeed, a recent study described the histone acetylase p300/cyclic AMP response element-binding protein (CBP) as coactivator for the C. elegans Forkhead transcription factor DAF-16 as well as for FOXO1 [68 ]. Whether phosphorylation modulates the interaction between FOXO members and p300/CBP remains to be determined. Third, phosphorylation might affect the stability of the protein, but our own experiments would suggest that at least FOXO4 is a highly stable protein (t1/2>10 h) Finally, phosphorylation in the C terminus (Ser319 in FOXO1) might directly affect the affinity of the transactivating domain with the basal transcription machinery.
|
OTHER REGULATORS OF FOXO |
|---|
|
Ras signaling can also affect FOXO function, as was demonstrated for FOXO4 [75 ]. This pathway depends on the activation of Ral, a Ras-related GTPase involved in Ras-dependent cellular proliferation and transformation. Ras/Ral-dependent signaling result in phosphorylation of Thr447 and Thr451 of FOXO4, and mutational analysis has shown that this site is essential to FOXO4 activity [76 ]. It is interesting that Ral activation by itself enhances the transactivating capacity of FOXO4 but appears to act synergistically with PKB/Akt in the inhibition of FOXO4 [76 ]. As such, Ral seems to prime FOXO4 for inhibition by PKB/Akt and to play an important role in the integration of multiple signal inputs on FOXO4. At present, it is unknown whether this regulation is specific for FOXO4 or also applies to FOXO1 and FOXO3A. Also, the Ral-dependent kinase responsible for phosphorylation of FOXO4 is unknown.
More recently, it was demonstrated that Ser322 and Ser325 in FOXO1 are phosphorylated in vivo, most likely by CK1 [77 ]. These residues are conserved in FOXO3 and FOXO4, suggesting that this is a more general event that affects all FOXO members. Phosphorylation at these sites does not occur in the absence of Ser319 phosphorylation, a target site of PKB/Akt and SGK [77 ]. This means that PKB/Akt or SGK has to prime FOXO factors for further phosphorylation by the Ser322/Ser325 kinase. As the Ser322 and Ser325 residues are flanked by two other phosphorylation sites (Ser319 and Ser329; the latter a target of DYRK1A), their phosphorylation creates a highly acidic patch. This acidic patch stimulates the interaction of FOXO1 with the Ran-containing protein complex that controls nuclear export [77 ]. In doing so, sequential phosphorylation of this cluster promotes nuclear exclusion of FOXO1.
|
FOXO-MEDIATED REGULATION OF CELL PROLIFERATION |
|---|
Several laboratories have investigated the FOXO-induced cell-cycle arrest and demonstrated that FOXO members increase the expression of the cyclin-dependent kinase (cdk) inhibitor p27kip1 [80 ,81 ]. Up-regulation of p27kip1 is linked to a cell-cycle arrest in G0/G1 through its inhibitory effect on cyclin/cdk complexes that are essential for progression to S-phase (Fig. 3 ). There is some discrepancy, however, as to the mechanism by which the FOXO-induced increase in p27kip1 protein levels comes about. Clearly, in a number of cell types, regulation of p27kip1 occurs through direct, transcriptional activation of the p27kip1 gene by FOXO members [10 ,80 ]. A moderate-to-strong transactivation of the p27kip1 promoter has been observed in response to FOXO activation [10 ,80 ]. This induction of p27kip promoter activity has been further corroborated by Northern blot analysis of p27kip1 mRNA, confirming the transcriptional up-regulation of p27kip1 [10 ,80 ]. Nevertheless, in other cell types, regulation of p27kip1 by FOXO appears to occur predominantly post-translationally, through stabilization of the p27kip1 protein [81 ]. This apparent contradiction is most likely a result of the way p27kip1 can regulate its own stability. As long as p27kip1 is in excess over cyclin E/cdk2 complexes, it can inhibit cdk activity through direct binding to the complex [82 ] (Fig. 3) . However, once the balance turns in favor of the cyclin E/cdk2 complexes, the active cyclin E/cdk2 complexes can phosphorylate p27kip1 at Thr187, resulting in its dissociation from the complex [82 ] (Fig. 3) . Thr187-phosphorylated p27kip1 is subsequently targeted for ubiquitin-mediated proteosomal degradation [82 ]. Accordingly, a relatively small increase in p27kip1 mRNA may result in an increase in p27kip1 protein, sufficient to reach a threshold value above which the p27kip1 protein becomes resistant to regulation by degradation. As a consequence of this intimate link between protein levels and stability, it may be difficult to separate transcriptional regulation of p27kip1 from protein stability. Clearly, any mechanism that causes up-regulation of p27kip1 to levels sufficient to completely inhibit cyclin E/cdk2 will also result in stabilization of the protein. To add to this complexity, it was recently shown by several groups that PKB/Akt directly phosphorylates p27kip1, resulting in nuclear exclusion of p27kip1 and relieving cyclin/cdk complexes from inhibition by p27kip1 [83 84 85 ]. Thus, PKB/Akt appears to regulate p27kip1 at multiple levels to ensure a tight control of progression through the G1 phase of the cell cycle.
|
The first evidence for transcriptional regulation of D-type cyclins by FOXO factors came from transcriptional profiling of PTEN-negative cells transduced with an adenoviral construct encoding a phosphosite mutant of FOXO1, in which all three PKB/Akt phosphorylation sites had been changed to alanine [86 ]. As described above, such mutants are constitutively nuclear and strong activators of FOXO-responsive promoters. Expression of this mutant led to a strong inhibition of expression of cyclin D1 and cyclin D2, as demonstrated by quantitative real-time polymerase chain reaction and immunoblotting [86 ]. A similar repression of cyclin D1 and cyclin D2 expression was observed in cells expressing a similar phosphosite mutant of FOXO3, as well as in cells overexpressing wild-type FOXO4 [87 ]. Thus, repression of cyclin D1/D2 appears to be a general mechanism of growth suppression by FOXO factors. It is presently not clear if the third member of the D-type cyclins is also prone to repression by FOXO. It is a distinct possibility that cyclin D3 expression is very low in the cell types studied thus far. Expression of cyclin D3 is mostly restricted to cells of the hematopoietic system, and it has not yet been investigated if FOXO factors are involved in transcriptional regulation of cyclin D3 in the appropriate cell lines. Also, the mechanism through which FOXO-mediated repression of cyclin D comes about is presently unclear. From the transcriptional profiling and analysis of mRNA levels, it is well established that regulation of cyclin D by FOXO is mediated through transcriptional repression [86 ,87 ]. This is further supported by promoter analysis of the cyclin D1 and cyclin D2 promoters, both of which are repressed by FOXO factors [87 ]. What is very striking, however, is the fact that no bona fide FOXO DNA-binding elements are present in those promoters. To add to this, a helix3 mutant of FOXO1, which fails to bind to DNA, is equally capable of repressing cyclin D expression [86 ]. This would suggest that repression does not involve direct binding of FOXO to the cyclin D promoter. Nevertheless, FOXO-mediated repression of cyclin D does not require the transcriptional regulation of other transcriptional regulators, as it is not affected by the addition of cyclohexamide [87 ]. Moreover, chromatin immunoprecipitation assays indicate that FOXO1 does bind to the cyclin D1 promoter [86 ]. This would suggest that FOXO members could be recruited to promoters by other transcriptional regulators, independent of FOXO DNA-binding activity. In this respect, it is of interest to note that FOXO1 has been described to interact with a number of nuclear hormone receptors [89 ]. Depending on the receptor type with which it interacts, FOXO1 can act as a transcriptional coactivator or corepressor [89 ]. In addition, it was recently shown that complex formation of FOXO1 with the androgen receptor results in repression of FOXO function [90 ]. At present, it is unclear if such interactions would be involved in regulation of cyclin D expression.
Expression of cyclin D is induced as quiescent cells are stimulated to enter the cell cycle [88 ]. At the same time, the expression of p27kip drops, allowing full activation of G1 cyclin/cdk complexes and eventually entry into S-phase [91 ]. Evidently, FOXO members can regulate the expression of two important regulators of cell-cycle entry, which raised the question if the FOXO-induced growth arrest is in effect a reflection of the quiescent state.
This seems to be the case, as it was demonstrated that the FOXO-induced arrest is reversible, an important criterion for quiescent cells [79 ]. In addition, a number of characteristic hallmarks of quiescence were observed in cells arrested by FOXO. The expression of the pRb-related pocket protein p130 was induced, apparently through direct, transcriptional regulation by FOXO, protein synthesis was inhibited, and G0-specific E2F complexes appeared [79 ]. Obviously, this situation bears great resemblance to the process of DAF-16-regulated Dauer formation in C. elegans, which is a reversible state of developmental arrest. What is more, these findings indicate that negative regulation of FOXO members by PKB/Akt functions as a switch in cell-cycle regulation controlling cell-cycle entry and exit.
|
FOXO-MEDIATED PROTECTION FROM OXIDATIVE STRESS |
|---|
As indicated, the paradigm for Forkhead regulation by the PI-3K/PKB/Akt signaling pathway has been set by studies in C. elegans. Here, DAF-16 controls Dauer formation, a phenotype characterized by a complex series of changes in thermoregulation, metabolism regulation, reproduction regulation, and oxidative stress control. Especially the latter aspect has been associated with the longevity phenotype that accompanies Dauer formation [93 ]. Also, in other animal model systems, oxidative stress control has been shown to correlate with aging, and these and other observations provide the basis for the free-radical theory of aging [94 ].
During the last year, several laboratories have now obtained evidence that at least in mammalian cell systems FOXO factors can be involved in protecting cells against the damaging effects of oxidative stress. First, FOXO factors were shown to control the expression of the antioxidant enzymes manganese superoxide dismutase (MnSOD) [8 ] and catalase [95 ]. Second, FOXO was shown to control the expression of growth arrest and DNA damage (GADD)45, a protein involved in DNA repair mechanisms [96 ,97 ]. Taken together, these data demonstrate that FOXO factors regulate the expression of a number of genes that are important in the protection against oxidative insults (Fig. 4 ).
|
30%, display increased resistance against oxidative stress [95
]. This correlated with a reduced ability of H2O2 to activate PKB/Akt and hence, inactivate FOXO3a, as measured by phosphorylation of T32 [95
]. This seems rather contra-intuitive, as it would suggest that oxidative stress causes inhibition of FOXO factors, which would counteract their protective effects. Indeed, the observed, stress-induced cytoplasmic retention of FOXO3a is at odds with observations in C. elegans, which show that a multitude of stresses, including caloric restriction, induce nuclear accumulation of DAF-16–green fluorescent protein [98
]. One caution would be that PKB/Akt activation by oxidative stress in mammalian cell systems is usually accomplished by adding high concentrations of H2O2 (100 µM or more) and that the oxidative stress generated by, for example, caloric restriction may compare with a much lower dose of H2O2. Several important distinctions should be made when dealing with oxidative stress. First of all, the type of oxidative stress generated by various treatments differs substantially. For example, H2O2 generates OH·- through the Fenton reaction, whereas UVA radiation generates singlet oxygen (1O2; named so because of its null spin value). It is possible that these different types of oxidative stress differ in their ability to regulate intracellular signaling pathways. Moreover, the mechanism by which cells detoxify these oxygen metabolites will be different. Second, increased expression of MnSOD and catalase induced by FOXO factors suggests protection against O2·- by a simple linear chain of events. MnSOD can convert O2·- into H2O2, and catalase can convert H2O2 into H2O and O2. However, it is important to note that MnSOD is localized in mitochondria, whereas catalase is predominantly localized in peroxisomes. Although H2O2 has a relatively high diffusion rate (compared with O2·- and OH·-), this clearly imposes a localization problem, and it is likely that up-regulation of catalase and MnSOD serves distinct purposes. As a matter of fact, an alternative perspective toward understanding the combined up-regulation of these two antioxidant enzymes by FOXO factors is provided by the Dauer phenotype. Long-term survival under the conditions that trigger Dauer formation requires an organismal switch from a glucose-based metabolism to a fat-based metabolism. As a large part of the fatty acids will be metabolized within the peroxisomes, this will result in increased production of H2O2 in the peroxisomes (as a result of ß-oxidation). Thus, the up-regulation of catalase is more likely to be required to accommodate increased peroxisome function. Another interesting aspect of this metabolic switch model is that the ability of PKB/Akt to protect cells from apoptosis relies on the ability of PKB/Akt to regulate glucose metabolism. Recent work has demonstrated that activated PKB/Akt inhibits closure of the VDAC to prevent mitochondrial hyperpolarization and cytochrome c release [99 ]. This is accomplished via PKB/Akt-stimulated binding of a hexokinase to VDACs at the outer mitochondrial membrane, and this interaction crucially depends on the first step of glycolysis [99 ]. In accordance with this, it was shown that in the absence of glucose, PKB/Akt is unable to protect cells from apoptosis [99 ]. The induction of MnSOD by FOXO factors could therefore function as a compensatory mechanism to deal with the oxygen radicals that are produced within the mitochondria when PKB/Akt is silenced. In agreement with this model, it has been shown that mammalian cells show increased resistance to glucose starvation as a result of the up-regulation of MnSOD [8 ]. As PKB/Akt inhibits FOXO function, this suggests that cells rely on two different programs of coupling metabolism, cell-cycle regulation and apoptosis protection (Fig. 4) . The first, PKB/Akt-active glucose metabolism, increased proliferation and PKB/Akt-dependent apoptosis protection (glucose-dependent); the second is PKB/Akt-inactive and therefore, FOXO-active fat/fatty acid metabolism, quiescence, and FOXO-dependent protection (MnSOD, catalase, GADD45).
Thus, the up-regulation of MnSOD and catalase represent part of an integrated change in survival, metabolism and cell cycle, which are set into motion by FOXO activation. In C. elegans, this results ultimately in Dauer formation and lifespan extension. The up-regulation of GADD45 may be viewed in a similar perspective. Whereas MnSOD is up-regulated to reduce the level of oxidative stress generated by the change in cell cycle and metabolism as a result of FOXO activation, GADD45 may function as a safeguard in the event that DNA damage may still occur.
|
FOXO-INDUCED APOPTOSIS |
|---|
Mere activation of FOXO factors appears to be sufficient to trigger programmed cell death, which stirred many to look for FOXO target genes that could be responsible for the induction of cell death. The first candidate to be identified was the Fas ligand [9 ]. Three putative FOXO-binding sites were identified in the Fas-ligand promoter, and fusion of the portion of the Fas-ligand promoter to a luciferase reporter plasmid demonstrated that this fragment is responsive to FOXO3 activation, at least in CCL39 fibroblasts [9 ]. Further support for a critical role for the Fas ligand came from experiments demonstrating that FOXO-induced apoptosis in cerebellar granule cells and Jurkat T cells is dependent on Fas-receptor signaling [9 ]. However, it was subsequently demonstrated that FOXO3 is unable to transactivate the full-length Fas-ligand promoter in BaF3 pre-B cells, yet FOXO3 is a very potent inducer of cell death in these cells [100 ]. In fact, FOXO-induced cell death occurs independent of death-receptor signaling in BaF3 cells, much like apoptosis induced by cytokine deprivation [100 ]. Instead, FOXO3 activation stimulates the expression of Bim, a proapoptotic Bcl-2 family member containing a single BH3 domain. Interleukin (IL)-3-deprivation of BaF3 cells also leads to transcriptional regulation of Bim [11 ], and similar results are seen upon cytokine deprivation of IL-2-dependent T cells as well as embryonic fetal liver cells [11 ,12 ]. Induction of Bim expression, at the level of mRNA and protein, could be reproduced by introduction of an active mutant of FOXO3 in BaF3 and cytotoxic T2 cells [11 ,12 ]. Moreover, the increase in Bim mRNA induced by FOXO3 was not dependent on de novo protein synthesis, indicating this occurs via direct transcriptional activation. In support of a role for Bim in mediating cytokine-dependent cell survival is the finding that cytokine dependency is reduced in Bim-/- fetal liver cultures [100 ]. This is in apparent discrepancy to the proposed role of the Fas ligand as the mediator of FOXO-induced apoptosis, but it is very well possible that FOXO factors activate different genes depending on the cell type. As a matter of fact, Bim is not up-regulated in cells that arrest upon introduction of active FOXO mutants (our own unpublished observations). Also, Bim was not amongst the regulated genes in the PTEN-deficient cells transduced with a FOXO1-encoding adenovirus [86 ]. The mechanism underlying this differential regulation of FOXO targets is presently unclear, just as we do not understand at present why certain cells die in response to FOXO activation, and others enter a state of quiescence. Of course, this could depend on the presence of the appropriate cofactors, but this option has not been addressed experimentally.
Since these earlier reports, other targets of FOXO members that could be involved in the induction of apoptosis have been reported (Fig. 5 ). First, the expression of Bcl-6, a strong transcriptional repressor, was shown to be induced by FOXO4 [101 ] and later also by FOXO3a [87 ]. FOXO4 was shown to bind specific binding sites in the Bcl-6 promoter that could activate transcription [101 ]. Bcl-6 in its turn negatively regulates the expression of the antiapoptotic protein BCL-XL, suggesting that FOXO factors can induce apoptosis through Bcl-6-mediated repression of BCL-XL. Indeed, macrophages isolated from Bcl-6-deficient mice undergo apoptosis with slightly delayed kinetics, indicating that FOXO-induced apoptosis may partly depend on Bcl-6 [101 ]. Second, transcriptional profiling in prostate cancer cell lines transduced with adenoviruses encoding FOXO1 or FOXO3a led to the identification of TRAIL as a target of FOXO factors [102 ]. Analysis of the TRAIL promoter showed that it contains a FOXO-responsive element and demonstrated it is a direct, transcriptional target of FOXO [102 ]. However, the exact contribution of TRAIL to FOXO-induced apoptosis remains to be determined. It is interesting that it was recently reported that expression of the TRADD gene is also under direct, transcriptional control of FOXO1 [103 ]. Transcriptional induction of TRADD by FOXO1 was shown to play an important role in chemotherapeutic drug-induced apoptosis, as dominant-negative TRADD mutants in HT1080 human fibrosarcoma cells could attenuate drug-induced apoptosis [103 ]. However, although mere activation of endogenous FOXO1 is sufficient to induce TRADD expression, it is unable to trigger apoptosis in the cell lines used in these studies. This would suggest that the induction of TRADD by FOXO1 is required to collaborate with other chemotherapeutic drug-induced genes to induce apoptosis.
|
Finally, it is of interest to note that a helix3 mutant of FOXO1 that fails to bind to the FOXO DNA-binding element directly is unable to induce apoptosis in PTEN-deficient cells [86 ]. This observation would suggest that FOXO-dependent transactivation of proapoptotic targets requires direct binding to FOXO-binding elements in the promoter of these genes. In contrast, a subset of cell-cycle regulatory genes, most notably the D-type cyclins, does not dependent on this and can apparently be regulated independent of FOXO DNA binding. However, it still has to be determined if helix3 mutants are similarly impaired in other "apoptotic" cell types, such as neuronal cells and cytokine-dependent lymphocytes.
|
CONCLUDING REMARKS |
|---|
Several questions remain, however. Why do multiple homologues exist? At present, it seems that the various functions of FOXO factors are shared by all of the members of the family. However, most of the studies done so far have relied on overexpression or expression of constitutive active mutants, which may have caused them to behave more promiscuously. Future experiments in which single FOXO factors are deleted through homologous recombination or RNA interference-mediated knockdown should help us separate out specific functions of the distinct members. Second, how do these transcription factors distinguish a quiescent state from an apoptotic state? This is a very intriguing question, the answer to which will help us devise intervening strategies that will induce cell death rather than quiescence upon inhibition of the PKB/Akt/Forkhead pathway. Such a result will of course greatly benefit any potential antitumor therapy directed at this pathway.
Future experiments should be directed at understanding the (different) roles of these FOXO factors in the whole organism. Studies in conditional knockout models may help us understand if each FOXO factor performs a distinct function. Moreover, such animal models will allow us to study the role of FOXO factors in tissue homeostasis, immune modulation by apoptosis, and tumor suppression.
|
ACKNOWLEDGEMENTS |
|---|
Received December 31, 2002; accepted February 12, 2003.
|
REFERENCES |
|---|
TOP
ABSTRACT
INTRODUCTION
