The Peptidyl-isomerase Pin1 Regulates P27 Kip1 Expression through Inhibition of Forkhead Box O Tumor Suppressors

The Forkhead box O (FOXO) protein family is an evolutionarily conserved subclass of transcription factors recently identified as bona fide tumor suppressors. Preventing the accumulation of cellular damage due to oxidative stress is thought to underlie its tumor-suppressive role. Oxidative stress, in turn, also feedback controls FOXO4 function. Regulation of this process, however, is poorly understood but may be relevant to the ability of FOXO to control tumor suppression. Here, we characterize novel FOXO4 phosphorylation sites after increased cellular oxidative stress and identify the isomerase Pin1, a protein frequently found to be overexpressed in cancer, as a critical regulator of p27 kip1 through FOXO4 inhibition. We show that Pin1 requires these phosphorylation events to act negatively on FOXO4 transcriptional activity. Consistent with this, oxidative stress induces binding of Pin1 to FOXO, thereby attenuating its monoubiquitination, a yet uncharacterized mode of substrate modulation by Pin1. We have previously shown that monoubiquitination is involved in controlling nuclear translocation in response to cellular stress, and indeed, Pin1 prevents nuclear FOXO4 accumulation. Interestingly, Pin1 acts on FOXO through stimulation of the activity of the deubiquitinating enzyme HAUSP/USP7. Ultimately, this results in decreased transcriptional activity towards target genes, including the cell cycle arrest gene p27 kip1. Notably, in a primary human breast cancer panel, low p27 kip1 levels inversely correlated with Pin1 expression. Thus, Pin1 is identified as a novel negative FOXO regulator, interconnecting FOXO phos-phorylation and monoubiquitination in response to cellular stress to regulate p27 kip1 .


Introduction
Forkhead box O (FOXO) transcription factors, consisting of mammalian FOXO1, FOXO3a, FOXO4, and FOXO6 are important downstream targets of the evolutionarily conserved phosphoinosi-tide-3-kinase/Akt (PKB) signaling pathway (1,2).Akt/PKB negatively regulates FOXO activity through direct phosphorylation by inducing their nuclear exclusion (3).FOXOs play a critical role in longevity, first shown in the nematode Caenorhabditis elegans, which had an extended life span upon deletion of the daf-2/insulin receptor (reviewed in ref. 4).This requires the FOXO orthologue daf-16 and is characterized by an increase in stress resistance, consistent with the notion that resistance to cellular stress correlates with longevity (5).
Stress resistance is also closely related to the onset of age-related diseases such as cancer (5,6).Indeed, FOXOs were recently shown to be tumor suppressors in a number of cancers (1,7).Mice depleted for FOXO1, FOXO3, and FOXO4 are characterized by the appearance of thymic lymphomas and hemangiomas (8).Intriguingly, cellular stress in turn, also changes FOXO activity towards its target genes, thus allowing for an adaptive response to cellular stress (reviewed in ref. 9).Regulation of this process, however, is poorly understood but may be relevant to the ability of FOXO to control tumor suppression.
FOXOs regulate a number of transcriptional targets involved in stress resistance, survival, and cell proliferation (reviewed in ref. 7).
A key transcriptional FOXO target is the cell cycle arrest gene p27 kip1 (8,10).The cyclin-dependent kinase inhibitor p27 kip1 is a haploinsufficient tumor suppressor that regulates the entry of cells from quiescence to cell cycle through inhibition of CDK2 (11).Interestingly, activation of FOXO in cells induces cell cycle arrest and quiescence, involving p130 and p27 kip1 expression (12).Reentry into the cell cycle involves down-regulation of p27 kip1 , a process that is poorly understood but is thought to involve the phosphorylation and degradation of p27 kip1 (11).In human cancer, expression of p27 kip1 is often found deregulated and numerous therapies are being developed to restore its function.Localization defects and degradation are thought to be the main cause of p27 kip1 deregulation.However, it was recently shown that p27 kip1 levels in human cancer are transcriptionally regulated as well, albeit through unknown mechanisms (13).
Regulation of protein activity often involves signaling through posttranslational modifications.These modifications either induce a structural change in the protein, thereby altering its activity, or induces the exposure of sites recognized by regulatory proteins.Pin1 is a peptidyl-prolyl isomerase that specifically recognizes phosphorylated serines and threonines flanked by a COOHterminal proline residue (14).Pin1-mediated isomerization induces conformational changes in the peptide backbone.This leads to the altered function of its protein substrates and has been shown to be involved in numerous processes, including the regulation of cell proliferation and death (14).Pin1 is found to be overexpressed in many human cancers and is linked to tumorigenesis (15).
Here, we identify a novel regulatory pathway for FOXO signaling.In response to cellular stress, FOXOs are phosphorylated and recognized by Pin1.Evidence is provided that Pin1 negatively regulates FOXO monoubiquitination at the level of deubiquitination through HAUSP/USP7.This inhibits nuclear FOXO translocation in response to hydrogen peroxide-induced stress and ultimately leads to decreased transcription and expression of FOXO4 transcriptional targets, including p27 kip1 .Notably, in a panel of primary human breast cancers, we found an inverse correlation between low p27 kip1 levels and Pin1 expression.
GST pulldown assays, coimmunoprecipitations, and Western blot analysis.GST proteins were coupled to glutathione agarose beads, washed twice with radioimmunoprecipitation assay buffer [RIPA; 20 mmol/L Tris-HCl (pH 8.0), 1% TX-100, 0.5% NaDoC, 5 mmol/L EDTA, 150 mmol/L NaCl, protease and phosphatase inhibitors].Cells were lysed in RIPA, cleared and incubated with the glutathione agarose beads for 2 h at 4jC.Subsequently, beads were washed, boiled in Laemmli sample buffer, and analyzed by Western blotting.Lambda phosphatase was purchased from New England Biolabs.For coimmunoprecipitation studies, 50 AL of Protein-A Sepharose beads were precoupled to 1 Ag of the indicated antibody.Cells were lysed in RIPA or, in the case of endogenous coimmunoprecipitation, in 20 mmol/L of Tris-HCl (pH 8.0), 1% NP40, 10% glycerol, 1 mmol/L of MgCl 2 , 1 mmol/L of EDTA, 150 mmol/L of NaCl, protease and phosphatase inhibitors, and incubated as described previously (17).
Mass spectrometry.Purified Flag-FOXO4 from hydrogen peroxidetreated HEK293T cells (200 Amol/L for 1 h) was digested with trypsin, subtilisin, and/or elastase (Roche).If required, samples were enriched for phosphorylated peptides using TiO 2 microcolumns, as described (18).Samples were subjected to nanoflow LC (Agilent 1100 series) coupled to a quadrupole time-of-flight tandem mass spectrometer (Micromass Waters).Data were processed and subjected to database searches using MASCOT software (Matrixscience).The identified peptides were confirmed by manual interpretation of the spectra.
Protein stability assay.HEK293T cells were transfected with a CMVdriven p27 kip1 construct and/or Pin1.Next, cells were treated with 10 Ag/mL of cycloheximide for the indicated times.Protein levels were detected and corrected for GAPDH levels by Odyssee.
Immunohistochemistry. Paraffin blocks containing formaldehyde-fixed breast cancer tissues were sectioned, deparaffinized, and rehydrated and stained essentially as described (19) with the primary antibodies for Pin1, FOXO3a, FOXO4, and p27 kip1 .These were detected using a poly-Hrp anti-Ms/Rb/Rt (ImmunoLogic) and developed with diaminobenzidine, followed by counterstaining with hematoxylin.
Ubiquitination assay.The monoubiquitination assay was performed as described (16).HEK293T cells were transfected with RNAi oligonucleotides and/or the indicated constructs.Forty-eight hours posttransfection, cells were treated as indicated and lysed in 8 mol/L of urea, 10 mmol/L of Tris-HCl (pH 8.0), 100 mmol/L of Na 2 HPO 4 /NaH 2 PO 4 , 0.2% TX-100, 5 mmol/L of NEM, and protease inhibitors.Ubiquitinated proteins were precipitated using Ni-NTA agarose beads and the experiment was analyzed by Western blotting.

Results
Identification of seven phosphorylated Ser/Thr-Pro sites on FOXO4.To better understand the FOXO4 adaptive response under conditions of cellular stress, a search for posttranslational modifications was initiated on FOXO4 by employing tandem mass spectrometry (MS).Flag-FOXO4 was purified from hydrogen peroxide-treated human embryonic kidney 293T (HEK293T) cells and subjected to proteolytic digestion.Five novel phosphorylated sites for FOXO4 were characterized, in addition to the previously characterized oxidative stress-sensitive phosphorylation sites Thr 447 /Thr 451 (20), suggesting that at least some of these seven sites are regulated through oxidative stress signaling (Fig. 1; Supplementary Table S1).TiO 2 columns, which specifically enrich for phosphorylated peptides (18), were used in MS analysis and confirmed these sites.In addition, we identified two double-phosphorylated peptides, Thr 447 /Thr 451 and Thr 223 /Ser 226 (Supplementary Table S1), indicating that FOXO4 can be multiphosphorylated.
Interestingly, all these sites consisted of phosphorylated Ser/Thr residues, followed by a Pro residue.These phosphorylated Ser/Thr-Pro motifs in proteins are potential recognition sequences for the peptidyl-prolyl isomerase Pin1, which has been proposed to regulate protein function through catalyzed cis-trans isomerization, thereby changing the local structure of its substrates (21).This prompted us to test the hypothesis that FOXO4 activity may be regulated by Pin1.
The FOXO4 transcriptional activity on p27 kip1 is inhibited by Pin1.Previously, we have shown that FOXO4 regulates transcriptional activity of the tumor suppressor gene p27 kip1 (10).Real-time reverse transcriptase qPCR was performed to test if Pin1 could affect FOXO4 transcriptional activity.Expression of FOXO4 increased the abundance of p27 kip1 mRNA by approximately 3-fold (Fig. 2A), similar to previous findings (10).Interestingly, coexpression of Pin1 inhibited the FOXO4-induced increase in p27 kip1 mRNA levels (Fig. 2A).Pin1 harbors two functional domains, an NH 2 -terminal WW domain, involved in substrate interaction (22), and a COOH-terminal PPIase domain critical for its isomerase activity (23).The Pin1 mutant Pin1 W34A , which fails to bind phosphorylated substrates (23), no longer inhibited p27 kip1 transcription, indicating that the WW domain is required for the Pin1-mediated p27 kip1 down-regulation.Similar results were obtained by using primers specific for the Gadd45a gene (Fig. 2A, right), another FOXO transcription target (24).
To determine if p27 kip1 protein levels change accordingly, FOXO4 was transiently expressed in the absence or presence of Pin1 and the mutant Pin1 W34A .Consistent with the reverse transcriptase qPCR results and previous data (10), FOXO4 increased p27 kip1 protein expression.However, Pin1 coexpression, but not mutant Pin1 W34A , abolished this increase in a dose-dependent manner (Fig. 2B).Similar results were found for p21 expression (Supplementary Fig. S1), another FOXO4 transcriptional target (25).Aside from FOXO-regulated transcription of p27 kip1 , Pin1 could also affect p27 kip1 expression at other posttranscriptional levels.To test whether Pin1 could also have a direct effect on p27 kip1 protein stability, we determined the p27 kip1 half-life time by performing cycloheximide experiments.Coexpression of Pin1 did not change the p27 kip1 half-life (Fig. 2C), indicating that Pin1 does not affect p27 kip1 protein stability.
Finally, we wanted to address if the regulation of p27 kip1 is mediated by the isomerase activity of Pin1.For this, we coexpressed FOXO4 and a previously described isomerase-defective Pin1 mutant, Pin1 K63A (21).Whereas Pin1 inhibited FOXO4induced p27 kip1 expression, this mutant did not (Fig. 2D), indicating that the isomerase activity of Pin1 is required for Pin1-mediated inhibition of FOXO4 transcriptional activity.Taken together, these results show that Pin1 expression inhibits FOXO4 transcriptional activity, which is dependent on both the Pin1 substrate interaction domain and the isomerization domain.This prompted us to test if Pin1 and FOXO4 physically interact.
Cellular stress induces Pin1 binding to phosphorylated FOXO4.Glutathione-S-transferase (GST) pulldown experiments were performed to examine if Pin1 could physically interact with FOXO4.As shown in Fig. 3A, GST-Pin1, but not GST alone, specifically precipitated FOXO4 from HEK293T cell lysates.Interestingly, treatment of cells with increasing amounts of hydrogen peroxide prior to lysis strongly enhanced this interaction in a dose-dependent manner.In addition to cellular stress generated by hydrogen peroxide, we tested other stressors as well.GST pulldown experiments were performed on FOXO4-expressing cells that were treated with anisomycin, which is known to activate stress-activated protein kinases (26), doxorubicin and UV, both used to induce DNA damage.Binding of FOXO4 increased only when cells were treated with hydrogen peroxide or anisomycin, indicating specificity for these stressors (Supplementary Fig. S2).Next, GST pulldown experiments were performed to test if Pin1 could also interact with FOXO3a, a closely related FOXO family member.Like FOXO4, FOXO3a was found to bind to Pin1 (Supplementary Fig. S3), suggesting that the interaction is conserved among the FOXO family members.Moreover, consistent with the inability to reduce FOXO4 transcriptional activity, the Pin1 W34A mutant, incapable of binding Pin1 substrates, could no longer interact with FOXO4 and FOXO3a (Fig. 3B; Supplementary Fig. S3).The above experiments indicate that Pin1 interacts with FOXO4 via its WW domain, an interaction that is enhanced in response to hydrogen peroxide.As mentioned, Pin1 binding is specific for phosphorylated Ser/Thr-Pro sites, of which MS on FOXO4 identified seven.Therefore, we analyzed whether the Pin1-FOXO4 interaction is actually dependent on the phosphorylation of FOXO4 Pin1 Regulates p27 kip1 through FOXO4 www.aacrjournals.org on these sites.Pretreatment of hydrogen peroxide-treated FOXO4 with lambda phosphatase, leading to FOXO4 dephosphorylation as indicated by the loss of reduced motility in SDS-PAGE, abolished the ability of FOXO4 to interact with Pin1 (Fig. 3B), indicating that FOXO4 needs to be phosphorylated for its interaction with Pin1.
Next, we set out to determine which phosphorylated FOXO4 Ser/ Thr-Pro sites are involved in the Pin1-FOXO4 interaction.Mutational analysis of single phosphorylation sites to Ala residues did not result in significantly decreased Pin1 binding upon peroxide stress, indicating that multiple phosphorylation sites are likely to be involved (data not shown).Further analysis using combinations of multiple Ala mutations consistently showed that the Pin1-FOXO4 interaction decreased progressively (data not shown) and is impaired for the FOXO4 mutant, in which all putative Pin1 binding sites as identified by MS, are mutated to Ala and is therefore referred to as FOXO4-7S/TA (Fig. 3C).
Finally, coimmunoprecipitation assays on endogenous FOXO4 and Pin1 in HEK293T cells showed that Pin1 specifically interacts with FOXO4 in vivo, an interaction that is increased after hydrogen peroxide-induced stress (Fig. 3D).Taken together, these results show that Pin1 and FOXO4 interact in vitro and in vivo.This interaction is sensitive to cellular stress and is dependent on a functional WW domain of Pin1 as well as phosphorylation of FOXO4.
Pin1 regulates peroxide-induced FOXO4 monoubiquitination and nuclear localization through inhibition of HAUSP/ USP7.Regulation of FOXO activity is often mediated through a change in cellular distribution.For instance, signaling through PKB/Akt inactivates FOXO4 through phosphorylation, resulting in nuclear exclusion (3).Alternatively, increased oxidative stress as generated by hydrogen peroxide results in increased nuclear localization (17,27).The observed Pin1-dependent decrease in FOXO4 transcriptional activity could therefore be the result of a changed nuclear-cytoplasmic FOXO4 localization.Immunofluorescence of FOXO4 in A14 cells shows that FOXO4 is distributed in both the cytoplasm and nucleus under normal growth conditions but is redistributed predominantly to the nucleus when stimulated with hydrogen peroxide, consistent with previous observations (Fig. 4A; refs.17, 27).However, ectopic expression of Pin1 prevented the redistribution of FOXO4 to the nucleus in response to oxidative stress.This phenotype depends on the interaction of FOXO4 with Pin1, as cells coexpressing Pin1 W34A do not inhibit FOXO4 relocalization.Furthermore, the FOXO4-7S/TA mutant that is impaired in binding to Pin1 had a nuclear localization in normal serumcontaining medium (Fig. 4A).Treatment of cells with hydrogen peroxide and Pin1 coexpression did not change FOXO4-7S/TA localization (data not shown), indicating that phosphorylation of these sites is required to retain FOXO4 in the cytoplasm.Taken together, these results indicate that Pin1 inhibits FOXO4-mediated transcriptional effects by inhibiting its nuclear localization.
The nuclear-cytoplasmic shuttling of FOXO factors by growth factor signaling is regulated through binding of 14-3-3 proteins, which leads to the export of FOXO in a Ran/Crm1-dependent manner to the cytoplasm (3).To uncover the mechanism by which Pin1 regulates FOXO4 localization, the interaction between 14-3-3 and FOXO4 was explored.FOXO4, but not a mutant in which all three PKB/Akt sites are mutated (FOXO4-A3), could bind 14-3-3 (Fig. 4B; ref. 3).In the presence of Pin1, however, the interaction of FOXO4 with 14-3-3 is unchanged.Altered 14-3-3 binding therefore, Figure 2. Pin1 inhibits FOXO4-induced p27 kip1 expression.A, Pin1 inhibits FOXO4 transcriptional activity.A14 cells were transfected as indicated together with pBabepuro.Total RNA extracted from puromycin-resistant cells was subjected to real-time reverse transcriptase-qPCR, using specific oligonucleotides for p27 kip1 or Gadd45a mRNA.Samples were normalized against PBDG mRNA.Columns, mean of triplicates; bars, SD (*, P < 0.05, t test).B, transient expression of Pin1 reduces p27 kip1 protein levels.HEK293T cells were transfected as indicated together with pBabePuro.Puromycin-selected cells were analyzed by Western blotting.C, Pin1 has no direct effect on p27 kip1 protein levels.HEK293T cells were transfected with a CMV-driven p27 kip1 construct and/or Flag-Pin1.Prior to lysis, cells were treated for the indicated times with cycloheximide.Protein levels relative to GAPDH were detected and quantified (representative experiment).D, the Pin1 effects are dependent on its isomerase activity.HEK293T cells were transfected as indicated together with pBabePuro.Puromycin-selected cells were analyzed by Western blotting.
could not explain the Pin1-mediated effects observed on FOXO4 localization.Importantly, phosphorylation of FOXO4 on Thr28 which is mediated by PKB/Akt did not change upon Pin1 expression, indicating that Pin1 does not affect PKB-mediated FOXO4 signaling.
Recently, we have shown that FOXO4 can be monoubiquitinated in response to oxidative stress, a process that is reversed by the deubiquitinating enzyme HAUSP/USP7 (16).Monoubiquitination of proteins is a distinct modification leading to a change in cellular localization/signaling, clearly different from polyubiquitination that leads to proteasome-mediated degradation of protein substrates (28,29).Indeed, monoubiquitinated FOXO4 leads to nuclear localization of FOXO4 independent of protein turnover (16).As shown in (Fig. 4C), low amounts of hydrogen peroxide induce FOXO4 monoubiquitination.However, expression of Pin1 completely inhibits hydrogen peroxide-induced FOXO4 monoubiquitination.Because HAUSP/USP7 is the enzyme responsible for deubiquitinating FOXO4, we tested if the Pin1 binding impaired mutant FOXO4-7S/TA mutant, could still be deubiquitinated by USP7.Surprisingly, whereas USP7 can efficiently and completely deubiquitinate FOXO4, deubiquitination is impaired for FOXO4-7S/ TA (Fig. 4D).This observation suggests that USP7-mediated deubiquitination on FOXO4 is in part dependent of Pin1.USP7mediated deubiquitination of FOXO4 involves binding of USP7 to FOXO4.Therefore, we tested whether FOXO4-7S/TA still binds to USP7.In coimmunoprecipitation assays, USP7 was found to interact equally well with both FOXO4 and FOXO4-7S/TA (Supplementary Fig. S4).Because FOXO4-7S/TA shows strongly impaired binding to Pin1, this result suggests that Pin1 enhances substrate, i.e., monoubiquitinated FOXO4 recognition of USP7.This also explains the small amount of FOXO4-7S/TA deubiquitination because USP7 still interacts with FOXO4.Together, these results show that Pin1 prevents hydrogen peroxide-induced FOXO4 nuclear localization and this likely results from the ability of Pin1 to inhibit hydrogen peroxide-induced FOXO4 monoubiquitination through stimulation of HAUSP/USP7-mediated FOXO4 deubiquitination.
Pin1 regulates p27 kip1 through FOXO4.FOXOs are known regulators of p27 kip1 expression, which as shown here, can be inhibited by Pin1.However, to our knowledge, Pin1 has not been described to affect p27 kip1 expression.Therefore, we wished to address the role of endogenous Pin1 in regulating p27 kip1 expression.To this end, we used small interfering RNA (siRNA) oligonucleotides against Pin1.Knockdown of Pin1 by two independent siRNAs increased the expression of p27 kip1 protein, in agreement with the role of Pin1 in p27 kip1 regulation and also in agreement with Pin1 acting as a negative regulator of FOXO (Fig. 5A).To exclude the possibility that the observed increase in p27 kip1 expression is the result of off-target knockdown, we generated a siRNAi-insensitive Pin1 construct by introducing a silent mutation in the RNAi recognition sequence of oligonucleotide no. 1 (Pin1-SM).Expression of this mutant in a Pin1 RNAi #1 background significantly rescued the RNAi-induced p27 kip1 protein levels, indicating that the observed effects are specific for Pin1 (Fig. 5A, right).These results indicate that depletion of Pin1 increases p27 kip1 expression in vivo.FOXO4-Thr28 phosphorylation did not change upon Pin1 depletion, consistent with the notion that Pin1 does not affect PKB-mediated FOXO4 signaling.As shown above, Pin1 overexpression reduces FOXO monoubiquitination and consequent nuclear translocation.To further establish whether endogenous Pin1 can regulate FOXO monoubiquitination, we analyzed ubiquitination of FOXO4 after siRNA against Pin1.Indeed, knockdown of Pin1 increased FOXO4 monoubiquitination (Fig. 5B).Thus, p27 kip1 regulation after Pin1 knockdown correlates with regulation of FOXO4 activity through monoubiquitination.Next, immunofluorescence experiments were performed to test if endogenous Pin1 also affects FOXO4 localization.Knockdown of Pin1 resulted in increased FOXO4 nuclear localization (Fig. 5C), consistent with the observations that FOXO4 monoubiquitination and transcriptional activity are increased upon Pin1 depletion.Because of technical reasons, we were unable to perform double Pin1/FOXO knockdown experiments.Therefore, we used the FOXO4-7S/TA mutant defective in Pin1 binding to test if the effects of Pin1 on p27 kip1 are mediated through FOXO.FOXO4-7S/TA was able to induce p27 kip1 protein expression, indicating that these phosphorylation sites are not required for FOXO4 transcriptional activity per se (Fig. 5D).Importantly, whereas FOXO4-induced p27 kip1 expression is inhibited, Pin1 no longer inhibited FOXO4-7S/TA-induced p27 kip1 expression.Taken together, these results provide evidence that Pin1 regulates p27 kip1 expression through regulation of FOXO.
p27 kip1 expression inversely correlates with Pin1 expression in human breast cancers.Many human cancers, particularly breast and prostate cancers, are characterized by high expression of Pin1 (30).Moreover, Pin1 contributes to oncogenic transformation as, for instance, Pin1 is essential for the transformation of mammary epithelial cells induced by Neu/Ras (31).Loss of p27 kip1 expression is also often found in many human cancers and correlates with poor survival (32,33).This prompted us to determine if p27 kip1 expression correlates with Pin1 levels in human cancer.We had access to a panel of 100 human invasive ductal breast cancer tumors and stained them for p27 kip1 and Pin1.In agreement with previously published work, Pin1 was found to be overexpressed to various degrees in the majority of tumors (n = 77/ 100), as compared with normal ductal breast tissue, in which only weak staining of Pin1 was found (ref.34 and Supplementary Fig. S5).We also stained for FOXO3 and FOXO4 and found no changes in expression staining, indicating that FOXO expression is not affected in our set of tumors (data not shown).
Expression of p27 kip1 was highly variable, varying from a complete loss to a staining comparable with normal breast tissue (Supplementary Fig. S5).Because loss of p27 kip1 is highly correlated with tumorigenesis and poor prognostic outcome, we divided the tumors into two distinct groups; one with normal p27 kip1 levels (defined as tumors with stronger staining than the statistical median, n = 61/100) and one with low p27 kip1 levels (n = 39/100).In the population with low p27 kip1 expression, a significant inverse correlation with Pin1 expression was found (P < 0.01, r = À0.42Spearman correlation test, two-tailed).Importantly, this observation unlikely represents an artifact of Pin1 overexpression in the majority of tumors because we observed no correlation with the high p27 kip1expressing group (Table 1).These results support the notion that Pin1 is important in p27 kip1 regulation and extends our observations to human breast cancer.

Discussion
Here, we provide evidence that the peptidyl-prolyl isomerase Pin1 acts as a negative regulator of FOXO transcriptional activity.Pin1, through its WW domain, binds FOXO directly in a phosphorylation-dependent manner.Furthermore, Pin1 regulation of FOXO requires, next to binding, the isomerase activity of Pin1.This indicates that Pin1-induced conformational changes within FOXO underlie the regulatory action of Pin1.Previously, it had been suggested that Pin1 catalyzes the dephosphorylation of the phosphorylated threonine/serine residues involved in Pin1 binding (14).Thus, Pin1-induced isomerization is thought to facilitate access of phosphatases like PP2A to the phosphorylated threonine/serine residues.Using phosphospecific antibodies against the PKB phosphorylation sites and phosphospecific antibodies against the sites identified in this study, we were unable to obtain any conclusive evidence that Pin1 binding to FOXO4 resulted in the dephosphorylation of these sites.In contrast, we observed that Pin1 binding to FOXO4 regulates its monoubiquitination.Previous studies have indicated that Pin1 can modulate polyubiquitination and subsequent protein degradation of substrates with a short-lifetime, such as Myc, p53, p73, and hcatenin (reviewed in ref. 14).However, FOXOs are relatively stable proteins and monoubiquitination, in contrast, does not lead to proteasome-mediated degradation but instead often results in changes in signaling and subcellular localization (28,29).Indeed, monoubiquitination of FOXO4 leads to nuclear localization and increased transcriptional activity but does not affect FOXO stability (16).Deubiquitination of FOXO4 is mediated by HAUSP/USP7 and similar to phosphatases, deubiquitinating enzymes such as USP7 require a cysteine within their catalytic activity to remove the ubiquitin moiety.We provide evidence that Pin1 increases USP7-mediated FOXO4 deubiquitination, which at least in part leads to decreased FOXO4 monoubiquitination.Thus, analogous to its role in regulating phosphatase activity towards its substrates, Pin1 may also regulate the activity of USP7 towards FOXO4.Nevertheless, Pin1-mediated attenuation of monoubiquitination represents, to the best of our knowledge, a  previously uncharacterized mechanism of Pin1 substrate activity regulation.Previously, we have shown that p27 kip1 is a key transcriptional FOXO target gene in the regulation of cell cycle arrest and quiescence (10,12), a notion that was recently supported by in vivo experiments (8).To maintain cell cycle arrest, FOXOs increase the cellular antioxidant capacity by up-regulating genes like MnSOD and catalase (9).In turn, increased cellular oxidative stress activates FOXOs, thereby creating a feedback loop that can prevent excessive cellular oxidative stress.Recent studies on conditional FoxO1/3/4 knockout mice underscore the critical importance of oxidative stress management by FOXOs in the hematopoietic stem cell compartment in keeping stem cells quiescent (35).Deletion of all three FOXO genes in hematopoietic stem cells results in stem cell depletion due to increased proliferation and concomitant increased intracellular levels of oxidative stress.Thus, FOXOs are important players in cellular oxidative stress management, and as such, it is of importance to understand how cellular oxidative stress impinges on FOXO.Under conditions of cellular stress, FOXO is phosphorylated, and here, we identified several novel sites of phosphorylation, which we show are involved in Pin1 binding after hydrogen peroxide treatment.Therefore, regulation of FOXO activity through Pin1 could present a novel mechanism of how cells re-enter the cell cycle from a quiescent state especially in response to cellular stress.In this respect, it is interesting to note that embryonic fibroblasts derived from Pin1 knockout mice show defective G 0 -G 1 entry (36).In line with this, we observe that knockdown of Pin1 increases p27 kip1 expression.
In contrast, overexpression of Pin1 inhibits FOXO4-induced p27 kip1 expression.p27 kip1 is a haploinsufficient tumor suppressor gene and decreased p27 kip1 expression can result in loss of cell cycle control, a hallmark of carcinogenesis (37).Loss of p27 kip1 is prevalent in human cancer and correlates with poor survival (33).Unlike loss of, for instance, the tumor suppressor p53, p27 kip1 function is lost because of a transcriptional or posttranslational down-regulation rather than a genetic defect (11,13).Moreover, it has been shown that Pin1 is overexpressed in numerous cancer tissues, notably in breast cancer and prostate cancer, and contributes to the malignant transformation of cancer cells (15).The significance of our data, in which Pin1 inhibits p27 kip1 expression is underscored by our findings in human breast cancers, where we find that loss of p27 kip1 expression strongly correlates with high Pin1 expression.Thus, in some percentage of tumors, loss of p27 kip1 may result from increased Pin1 expression inhibiting FOXO function.Reconstitution of nuclear FOXO activity has been shown to arrest both normal and transformed cells in G 1 , to inhibit soft agar growth, and to inhibit xenograft growth in nude mice (38,39).Disruption of the Pin1-FOXO interaction aimed at increasing the nuclear FOXO pool and restoring p27 kip1 transcription could therefore represent a potential therapeutic point for intervention.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.

Figure 1 .
Figure 1.FOXO4 is phosphorylated on multiple Ser/Thr-Pro sites in conditions of oxidative stress.A, schematic representation of FOXO4, with the MS-identified phosphorylated residues.FKH, Forkhead domain; NLS, nuclear localization signal; NES, nuclear export signal; TA, transactivation domain.B, representative tandem MS sequence of a FOXO4 tryptic peptide spanning amino acids 444 to 461.b and y ions are shown as well as the À98 Da H 3 PO 4 loss and the phosphorylated Thr 451 residue.

Figure 3 .
Figure 3. Pin1 interacts with FOXO4 in vitro and in vivo.A, FOXO4 binds Pin1 in a GST pulldown assay.HEK293T cells were transfected with empty vector or Flag-FOXO4.Cells were treated for 1 h as indicated and lysates were subjected to a GST pulldown assay.B, FOXO4 dephosphorylation prevents Pin1 binding.HEK293T cells were transfected with Flag-FOXO4.One hour prior to lysis, cells were treated with 200 Amol/L of H 2 O 2 .FOXO4 was purified with an anti-Flag antibody, eluted off with Flag peptide, and treated with lambda phosphatase (k ) for 30 min, or left untreated and subjected to GST pulldown.C, multiple phosphorylation sites are involved in FOXO4-Pin1 binding.HEK293T cells were transfected as indicated, stressed with 200 Amol/L of H 2 O 2 for 1 h, and subjected to coimmunoprecipitation.D, endogenous FOXO4 and Pin1 interact.HEK293T cells were treated as indicated, lysed, and either immunoprecipitated with a specific antibody for FOXO4 (N19) or an isotype control.

Figure 4 .
Figure 4. FOXO monoubiquitination and nuclear FOXO4 localization is inhibited by Pin1.A, Pin1 inhibits FOXO4 nuclear localization in response to hydrogen peroxide stress.A14 cells were transfected and left untreated or treated as indicated, after which cells were fixed and stained.Transfected cells were used to score the localization of FOXO4.Columns, mean of two independent experiments performed in triplicate; bars, SE.The significance of changes as compared with FOXO4 only was confirmed by t test (**, P < 0.005).B, Pin1 does not change 14-3-3 binding to FOXO4.HEK293T cells were transfected with HA-FOXO4 or HA-FOXO4-A3, a FOXO4 mutant that can no longer bind 14-3-3.Forty-eight hours after transfection, cells were treated as indicated, lysed, and subjected to coimmunoprecipitation.C, Pin1 negatively regulates FOXO4 monoubiquitination.HEK293T cells were transfected and treated as indicated with 50 Amol/L of H 2 O 2 for 15 min, lysed, and analyzed for ubiquitinated FOXO4 (Ub-FOXO4).D, USP7-mediated deubiquitination of FOXO-7S/TA is reduced.HEK293T cells were transfected with His 6 -Ubi, HA-FOXO4, or HA-FOXO4-7S/TA or Myc-USP7 and analyzed as in C .