FOXO4 interacts with p53 TAD and CRD and inhibits its binding to DNA

Abstract

Transcription factor p53 protects cells against tumorigenesis when subjected to various cellular stresses. Under these conditions, p53 interacts with transcription factor Forkhead box O (FOXO) 4, thereby inducing cellular senescence by upregulating the transcription of senescence‐associated protein p21. However, the structural details of this interaction remain unclear. Here, we characterize the interaction between p53 and FOXO4 by NMR, chemical cross‐linking, and analytical ultracentrifugation. Our results reveal that the interaction between p53 TAD and the FOXO4 Forkhead domain is essential for the overall stability of the p53:FOXO4 complex. Furthermore, contacts involving the N‐terminal segment of FOXO4, the C‐terminal negative regulatory domain of p53 and the DNA‐binding domains of both proteins stabilize the complex, whose formation blocks p53 binding to DNA but without affecting the DNA‐binding properties of FOXO4. Therefore, our structural findings may help to understand the intertwined functions of p53 and FOXO4 in cellular homeostasis, longevity, and stress response.

1. INTRODUCTION

Cellular senescence induces permanent cell cycle arrest and the secretion of interleukins, inflammatory cytokines, and growth factors. The resulting adverse effects on the cellular microenvironment contribute to aging and to the onset of age‐related diseases such as tumorigenesis. ¹ , ² Senescence can be induced by multiple factors, such as telomere shortening, oxidative stress and oncogenic signaling, which trigger the expression of p53‐mediated cyclin‐dependent kinase inhibitor (CDKI) p21 or CDKI p16 or both. ³ , ⁴ These CDKIs block retinoblastoma (RB) protein inactivation, thereby activating E2F transcription factors and imposing a permanent cell cycle arrest. ⁵

The transcription factor p53 is a key regulator of apoptosis, senescence and DNA repair, which protects cells against tumorigenesis under various cellular stresses. ⁶ The p53 molecule consists of N‐terminal transactivation domains (TAD) followed by a proline‐rich domain (PRD), a DNA‐binding domain (DBD), a tetramerization domain (TD) and a C‐terminal negative regulatory domain (CRD) ⁴ (Figure 1a). TAD is an intrinsically disordered domain with two motifs prone to form an α‐helix (TAD1 and TAD2), which mediate interactions with several other proteins, including the Mouse double minute 2 homolog (MDM2). ⁷ The p53 DBD recognizes the consensus DNA sequence composed of a tandem of two decameric palindromic sequences (half‐sites) 5’‐RRRCWWGYYY‐3′, where R = purine, Y = pyrimidine and W is either A or T. ⁸ , ⁹ In addition to binding to the target DNA, this domain also interacts with the N‐terminal domains TAD and PRD ¹⁰ , ¹¹ , ¹² and with other proteins, such as the apoptosis‐stimulating protein of p53‐2 (ASPP2), ¹³ also known as Bcl2‐binding protein (Bbp), and the B‐cell lymphoma‐extra large (Bcl‐xL). ¹⁴ CRD is also an intrinsically disordered domain and, similarly to other p53 domains, interacts with numerous other proteins. ¹⁵ , ¹⁶ , ¹⁷ In short, the wide range of p53 interactions facilitated by its multidomain structure accounts for its versatile functions.

FIGURE 1.

Domain structure of p53 (a) and FOXO4 (b) proteins. CR1, conserved region 1; CRD, C‐terminal negative regulatory domain; DBD, DNA binding domain; FH‐DBD, Forkhead DNA‐binding domain; NES, nuclear export sequence; NLS, nuclear localization signal; PRD, proline‐rich domain; TAD, transactivation domain; TD, tetramerization domain. FOXO4 and p53 constructs are indicated by grey boxes

The functions of p53 are closely intertwined with the activity of Forkhead box O (FOXO) transcription factors. FOXO proteins (FOXO1, FOXO3, FOXO4, and FOXO6) regulate cellular homeostasis, longevity and stress response by controlling the transcription of target genes when binding to the consensus recognition motif 5′‐GTAAA(T/C)AA‐3′, also known as the Daf‐16 family member‐binding element (DBE), ¹⁸ , ¹⁹ and to the 5′‐(C/A)(A/C)AAA(C/T)AA‐3′ motif, located in the IGFBP‐1 promoter region and known as the insulin‐responsive element (IRE). ²⁰ FOXO1 and FOXO3 proteins have similar lengths of 655 and 673 amino acid residues, whereas FOXO4 and FOXO6 sequences are shorter and contain 505 and 492 residues, respectively (Figure S1). All FOXO proteins have several highly conserved regions: an N‐terminal CR1, containing the first Akt/protein kinase B (PKB) phosphorylation site and the 14–3‐3 protein motif, a Forkhead DNA‐binding domain (FH‐DBD), which partly overlaps with a nuclear localization signal (NLS), a nuclear export sequence (NES) and a C‐terminal transactivation domain (TAD) (Figure 1b). The N‐ and C‐terminal segments bordering the FH‐DBD are intrinsically disordered and crucial for diverse functions of FOXO proteins because they harbor post‐translational modification sites and participate in protein–protein interactions. ²¹ , ²² , ²³ In turn, the transcriptional activity of FOXO proteins is controlled by the phosphatidylinositol 3‐kinase‐protein kinase B/Akt (PI3K‐PKB/Akt) signaling pathway. Upon its activation by PI3K, PKB/Akt phosphorylates FOXOs at three (two in FOXO6) conserved serine and threonine residues, thereby inducing their binding to the scaffolding 14–3‐3 proteins, followed by nuclear export. ²⁴ In addition to PKB/Akt‐mediated phosphorylation, the function of FOXO proteins is also controlled by other kinases, such as cyclic GMP‐dependent kinase‐1, mammalian Ste‐20 like kinase‐1, dual‐specificity tyrosine‐regulated kinase‐1A, and c‐JUN N‐terminal kinase, acetylation and ubiquitination (reviewed in ²⁵ ). FOXO proteins were initially described as tumor suppressors that inhibit cancer cell growth and survival. ²⁶ However, recent studies have shown that they can also promote tumor development and progression by maintaining cellular homeostasis and cancer stem cells ²⁷ , ²⁸ and by inducing drug‐resistance. ²⁹ , ³⁰ These findings highlight the importance of examining, in detail, structural‐functional relationships between FOXO and p53, as transcription factors.

The interaction between FOXO and p53 was initially reported by Nemoto et al., ³¹ who showed that the FOXO3:p53 complex is responsible for starvation‐induced SIRT1 deacetylase activation. This interaction has also been observed in 293 T cells under oxidative stress conditions in a study focused on SIRT1‐dependent regulation of FOXO3 function. ³² Further insights into the interaction between FOXO3 and p53 have been subsequently provided by Wang et al., ³³ who suggested that p53 DBD weakly interacts with two distinct regions of FOXO3 (FH‐DBD and the C‐terminal TAD) and disrupts their intramolecular interaction. Yet, the roles of other FOXO and p53 domains and regions in this interaction remain unclear. FOXO3 is known to stabilize p53, thus activating p53‐dependent apoptosis. ³⁴ Conversely, the FOXO3:p53 interaction may inhibit the transcriptional activity of FOXO3 under oxidative stress conditions, albeit without affecting p53 functions. ³⁵ Moreover, Baar et al. ³⁶ have demonstrated that the direct interaction between FOXO4 and p53 represses apoptosis in senescent cells and that the inhibition of this interaction by the D‐retro‐inverso peptide, which corresponds to the N‐terminus of FOXO4 FH‐DBD, disrupts the transcription of senescence‐associated CDKI p21 and induces nuclear exclusion of active p53, thereby inducing death in senescent cells.

As described above, numerous studies have suggested links between FOXOs and p53, but the structural details of their interaction remain mostly unclear. In this context, to improve our structural understanding of interactions between these transcription factors, we used the full‐length and several truncated constructs of p53 and FOXO4 towards mapping their interactions by NMR, chemical cross‐linking coupled to MS and analytical ultracentrifugation. Our findings shed light on a complex pattern of contacts between various regions of p53 and FOXO4 by identifying the domains specifically involved in this interaction and its effect on the DNA‐binding affinity of p53.

2. RESULTS

2.1. Both the N‐terminal part of p53 and the N‐terminal part of FOXO4 stabilize the p53:FOXO4 Complex

To map interactions between p53 and FOXO4 and to evaluate the role of individual domains and disordered regions, we prepared the full‐length and several truncated constructs of p53 (p53_1‐393, p53_1‐312, p53_61‐312, p53_94‐312, p53_1‐93) and FOXO4 (FOXO4_1‐505, FOXO4_15‐217, FOXO4_86‐211, FOXO4_198‐505) (Figure 1) and characterized their interactions by sedimentation velocity analytical ultracentrifugation (SV AUC), NMR and chemical cross‐linking coupled to MS.

The apparent equilibrium dissociation constants (K _D) of p53:FOXO4 complexes were determined in SV AUC titrations (Figure 2, S2). Unfortunately, the low stability of full‐length p53 and FOXO4 constructs (p53_1‐393 and FOXO4_1‐505) in solution prevented us from using them in SV AUC measurements. For this reason, these experiments were performed only with more stable truncated constructs.

FIGURE 2.

Sedimentation velocity analytical ultracentrifugation analysis of mixtures of FOXO4 and p53 constructs. (a) FOXO4_15‐217:p53_1‐312 (b) FOXO4_86‐211:p53_1‐312 (c) FOXO4_15‐217:p53_61‐312 (d) FOXO4_15‐217:p53_94‐312. Isotherms of weight‐averaged sedimentation coefficients s _w were constructed from SV AUC experiments with mixtures of p53 (20 μM) and FOXO4 (2–200 μM) variants. The sedimentation coefficient distributions c(s) underlying the s _w data points are shown in Figure S2. All s _w isotherms were fitted using the reversible Langmuir‐type kinetic model A + B ⇌ AB and assuming a 1:1 stoichiometry for the FOXO4:p53 interaction

The analysis of the isotherm of weight‐averaged sedimentation coefficients s _w (s _w isotherm) as a function of the FOXO4_15‐217 concentration revealed an interaction with a K _D of ~100 μM between p53_1‐312 and FOXO4_15‐217 (Figure 2a, S2a). SV AUC titration of p53_1‐312 by the N‐terminally truncated FOXO4_86‐211 (FH‐DBD) showed a considerably reduced binding affinity, with a K _D value of ~800 μM (Figure 2b, S2b). Similarly, both truncated p53 constructs (p53_61‐312 and p53_94‐312), missing TAD and TAD plus PRD, respectively, also exhibited reduced binding affinities for FOXO4_15‐217 (Figures 2c, d and S2c,d). Furthermore, no interaction was observed between p53_61‐312 and p53_94‐312 constructs and FOXO4 FH‐DBD (Figure S2e,f). We also tried to characterize the interaction between the C‐terminal half of the FOXO4 molecule containing TAD (FOXO4_198‐505) and p53 variants. However, given the high sensitivity of this construct for proteolytic degradation, we were unable to perform SV AUC titrations. Taken together, these results suggest that both the N‐terminal part of p53 and the disordered N‐terminal part of FOXO4 stabilize the p53:FOXO4 complex.

2.2. p53 interacts with several regions of the FOXO4 Forkhead domain

FOXO4 and p53 regions involved in their interaction were identified by NMR spectroscopy using uniformly ¹³C,¹⁵N‐labeled FOXO4_15‐217 in standard triple resonance experiments. The analysis of the resulting NMR data enabled the sequence‐specific backbone assignment, including H^N, N^H, C′, C_α and C_β for 143 of the 203 residues (70% of the FOXO4_15‐217 sequence, Figure S3). The ¹⁵N‐labeled FOXO4_15‐217 was titrated with unlabeled p53_1‐312 and p53_1‐93 (p53 TAD), and ¹H and ¹⁵N chemical shift perturbations (CSPs) of the amide groups of FOXO4 were followed in ¹H‐¹⁵N heteronuclear single quantum coherence (HSQC) spectra (Figures 3a–d and S4–S6a).

FIGURE 3.

Mapping of the p53‐binding surface of FOXO4. (a) Distribution of CSPs of 200 μM ¹⁵N‐labeled FOXO4_15‐217 with 200 μM p53_1‐312. Secondary structure of FOXO4 FH‐DBD is indicated at the top. Grey bars represent unassigned residues in the ¹H‐¹⁵N HSQC spectrum. Solid, dashed, and dotted lines correspond to the mean, mean + 1σ _cor, and mean + 2σ _cor, respectively. (b) Detailed view of selected areas of ¹H‐¹⁵N HSQC spectra of ¹⁵N‐labeled FOXO4_15‐217 in the presence of p53_1‐312. (c) Distribution of CSPs of 100 μM ¹⁵N‐labeled FOXO4_15‐217 with 100 μM p53_1‐93. Solid, dashed, and dotted lines correspond to the mean, mean + 1σ _cor, and mean + 2σ _cor, respectively. (d) Detailed view of selected areas of ¹H‐¹⁵N HSQC spectra of ¹⁵N‐labeled FOXO4_15‐217 in the presence of p53_1‐93. (e) CSPs mapped onto the crystal structure of the FOXO4 FH‐DBD:DNA complex. ⁴⁰ Residues with CSPs larger than 2σ _cor above the mean value are highlighted in dark purple, whereas residues with CSPs larger than 1σ _cor above the mean value are highlighted in light purple. Residues that could not be unambiguously assigned are highlighted in ocher

Surprisingly, significant CSPs and changes in intensities of resonances in 2D spectra of ¹⁵N‐labeled FOXO4_15‐217 were observed only in several residues of the Forkhead domain, and the CSP and intensity patterns were similar in the presence of both p53 constructs. The most affected residues were located at the N‐terminus of the Forkhead domain (N99 and G102), the α‐helix H1 (S105) and the loop between helices H2 and H4 (V135 and Y137), thus suggesting that p53 specifically interacts with residues from these FH‐DBD regions. Furthermore, lower but still significant CSPs were also observed in several residues of the α‐helix H3 (N157, L160) and NLS (R194) at the C‐terminus of FH‐DBD. Interestingly, both the N‐terminus of FH‐DBD and the α‐helix H3 are found in the DNA‐binding surface of FH‐DBD (Figure 3e). The gradual shift in resonances of several FH‐DBD residues during the titration (Figure 3b,d) suggests a fast exchange of bound protein on the NMR time scale, in line with the moderate affinity assessed by SV AUC (Figure 2a). The decrease in signal intensity of residues from the Forkhead domain of FOXO4_15‐217 (Figure S6a) is probably due to a slower tumbling time of this domain with bound p53 constructs. Furthermore, the fact that both p53 constructs affected the ¹⁵N‐labeled FOXO4_15‐217 in the same way suggests that TAD of p53 is the primary binding partner for FOXO4 (Figure 3a,c).

Neither significant CSPs nor changes in intensities of resonances in 2D spectra of ¹⁵N‐labeled FOXO4_15‐217 were observed in residues of the N‐terminal unstructured region preceding the Forkhead domain of FOXO4_15‐217 (residues 15–98) (Figures 3a,c and S6a). This region of the FOXO4_15‐217 molecule is thus unlikely to participate in p53 binding. However, SV AUC measurements suggested that the N‐terminal part of FOXO4_15‐217 does stabilize the formation of the complex (Figure 2a,b). Therefore, these results may indicate that FOXO4 interactions with p53 are either highly heterogenous and/or the removal of the N‐terminal part of FOXO4 affects the behavior of FH‐DBD and thus its interaction with p53.

A similar experiment was also performed with the ¹⁵N‐labeled isolated FOXO4 FH‐DBD (FOXO4_86‐211) and unlabeled p53_1‐312 (Figures S6b–d and S7). The previously published NMR sequential assignment of human FOXO4 FH‐DBD ³⁷ was used to evaluate CSPs of residues of the FOXO4_86‐211 construct. This titration revealed the same changes within the Forkhead domain as those identified in the measurements with FOXO4_15‐217. Overall, our ¹H‐¹⁵N HSQC measurements with labeled FOXO4 proteins suggest that several regions of FOXO4 FH‐DBD form a binding surface for p53 TAD, which appears to be the key FOXO4‐interacting domain of p53.

2.3. p53 interacts with FOXO4 via the N‐terminal TAD with additional contacts from PRD and DBD

To map p53 residues involved in FOXO4 binding, we prepared isotopically‐labeled p53_1‐312 and p53_1‐93 (the N‐terminal p53 TAD) (Figure 1a). Because the NMR sequential assignment is available only for isolated p53 domains, namely TAD, DBD and TD, we performed standard triple resonance experiments for sequence‐specific backbone assignment of the p53_1‐312 construct. The analysis of the resulting NMR data enabled the backbone resonance assignment, including H^N, N^H, C′, C_α and C_β of 209 of the 312 residues, thereby identifying 262 systems of the expected 272 (the p53_1‐312 sequence contains 40 proline residues). Supporting Information Figure S8 shows the ¹H‐¹⁵N HSQC spectrum of ²H,¹³C,¹⁵N‐labeled p53_1‐312 with this resonance assignment.

As expected, the analysis of ¹H‐¹⁵N HSQC spectra of ²H,¹³C,¹⁵N‐labeled p53_1‐312 revealed that adding FOXO4_15‐217 induced substantial CSPs in backbone amide groups of residues of the N‐terminal TAD2 of p53 (Figures 4a,b and S9). Smaller but still significant CSPs were also observed in several PRD (L93 and S94) and DBD (T170, C176, T211, G226, C277, G279 and K292) residues (Figure 4c,d). Furthermore, the presence of FOXO4_15‐217 slightly reduced the signal intensity of residues from the TAD2 domain of p53 (Figure S10a). Based on these results, the FOXO4‐binding surface of p53 should primarily consist of TAD, with PRD and DBD playing only minor roles.

FIGURE 4.

The domain TAD forms the crucial region of the FOXO4‐binding surface of p53. (a) Distribution of CSPs observed in residues of 130 μM ²H,¹³C,¹⁵N‐labeled p53_1‐312 in the presence of 590 μM unlabeled FOXO4_15‐217. Grey bars represent unassigned residues in ¹H‐¹⁵N HSQC spectra. Solid, dashed, and dotted lines correspond to the mean, mean + 1σ _cor, and mean + 2σ _cor values of CSPs, respectively. Secondary structure of p53 is indicated at the top. (b) Detailed view of selected peaks from ¹H‐¹⁵N HSQC spectra of ²H,¹³C,¹⁵N‐labeled p53_1‐312 in the presence of FOXO4_15‐217. (c) Detail of CSPs observed in residues 93–293 (DBD). Grey bars represent unassigned residues in ¹H‐¹⁵N HSQC spectra. Solid, dashed, and dotted lines correspond to the mean, mean + 1σ_cor, and mean + 2σ_cor values of CSPs observed in residues 93–293 (CSPs of N‐terminal residues were excluded), respectively. (d) CSPs of ¹⁵N‐p53_1‐312 in the presence of FOXO4_15‐217 mapped onto the crystal structure of the p53 DBD:DNA complex (PDB ID: 2AHI ⁵⁴ ). Residues with CSPs larger than the mean + 2σ _cor (from panel c) are highlighted in blue. Residues that could not be unambiguously assigned are highlighted in ocher. (e) Distribution of CSPs of 100 μM ¹⁵N‐labeled p53_1‐93 in the presence of 100 μM unlabeled FOXO4_15‐217. Grey bars represent unassigned residues in ¹H‐¹⁵N HSQC spectra. Solid, dashed, and dotted lines correspond to the mean, mean + 1σ_cor, and mean + 2σ_cor values of CSPs, respectively. The secondary structure of the N‐terminal region of p53 is indicated at the top. (f) Changes in signal intensities in 2D ¹H‐¹⁵N HSQC spectra of ¹⁵N‐labeled p53_1‐93. Purple bars correspond to signal intensity ratios of 100 μM p53_1‐93 in the presence and absence of 100 μM FOXO4_15‐217. Grey bars correspond to signals with missing assignment. (g) Detailed view of selected peaks from the ¹H‐¹⁵N HSQC spectra of ¹⁵N‐p53_1‐93 in the presence of FOXO4_15‐217

The interaction between p53 TAD2 and FOXO4 was confirmed by titrating ¹⁵N‐labeled p53_1‐93 with unlabeled FOXO4_15‐217 (Figures 4e–g and S11). The pattern of CSPs and changes in signal intensities were similar to those observed in the longer p53 construct, p53_1‐312. The analysis of the backbone chemical shift values of p53_1‐312 revealed a higher propensity for a β‐structure within the TAD2 segment in the presence of FOXO4_15‐217 (Figure S12), that is, in the region with the highest CSPs. Combined, these results further supported the key role of p53 TAD2 in FOXO4 binding.

We also characterized the interaction between p53_1‐312 and the isolated C‐terminal half of the FOXO4 molecule (residues 198–505) containing FOXO4 TAD (Figure 1b). The FOXO4_198‐505 construct is unstable in solution, which prevented us from performing the SV AUC analysis of its complexes. Nevertheless, we successfully recorded ¹H‐¹⁵N HSQC spectra of ²H,¹³C,¹⁵N‐labeled p53_1‐312 with and without unlabeled FOXO4_198‐505 (Figure S13). Although no significant CSPs were observed, the presence of FOXO4_198‐505 slightly reduced the signal intensity of p53 DBD residues (Figure S10b). This suggests the presence of a weak interaction, as in some IDPs, the decrease in signal intensity rather than CSPs reflects the interaction. ³⁸

2.4. p53 DBD and FOXO4 Forkhead domain are located close to each other in the p53:FOXO4 Complex

Because we were unable to perform NMR measurements with full‐length constructs of FOXO4 and p53, we used chemical cross‐linking to map interactions between full‐length p53_1‐393 and FOXO4_1‐505 (Figure S14a). Table S1 lists intermolecularly cross‐linked p53 and FOXO4 residues in the p53_1‐393:FOXO4_1‐505 complex. As noted, most intermolecular cross‐links between p53_1‐393 and FOXO4_1‐505 connect residues of the central region of p53 DBD (region 120–164) to the Forkhead domain and the NLS region of FOXO4 (cross‐links #1, 3, 5, 8, 11–13 in Table S1). Consequently, when the complex is formed, the DNA‐binding domains of both proteins become closer.

Cross‐linked FOXO4 residues are located in the loop between α‐helices H1 and H2 (K120), H2 and H3 (K139 and K141) and the C‐terminal part of the Forkhead domain (K163, K166 and K174) (Figure S15a). In p53 DBD, most cross‐linked lysines (K120, K132 and K139) are close to the α‐helix H3 at the C‐terminus of p53 DBD (Figure S15b). These results match the data from our NMR measurements in which similar regions were identified as the interaction interface (Figure S15c,d).

Several cross‐links are also formed between the C‐terminal p53 CRD and the N‐terminus of FOXO4 (cross‐links #9 and 10), the Forkhead domain (cross‐links #2, 4 and 6), the FOXO4 NLS region (cross‐link #7) and the C‐terminal region of FOXO4 (cross‐link #14). These cross‐links suggest that, when the complex is formed, p53 CRD interacts and/ or is located near these FOXO4 regions. Moreover, no cross‐links were observed between the N‐terminal p53 TAD and the Forkhead domain, likely due to the very limited number of lysine residues in p53 TAD (Figure S16).

Using the same cross‐linking agents, we performed chemical cross‐linking between the C‐terminally truncated p53_1‐312, which remains monomeric in solution, and the full‐length FOXO4_1‐505 (Table S2 and Figure S14b). Our data analysis revealed similar patterns of intermolecular cross‐links between the two complexes, except for cross‐links involving p53 CRD, indicating that p53 tetramerization is not required for the interaction between and/ or closeness of DNA‐binding domains of both proteins in their complex. Furthermore, we detected intermolecular cross‐links involving the N‐ and C‐terminal segments of FOXO4 (cross‐links #9, 10 and 14 in Table S1; cross‐link #5 in Table S2). Our findings not only corroborate the involvement of these disordered segments of FOXO4 in p53 binding but are also in line with our SV AUC results.

Furthermore, chemical cross‐linking between the C‐terminally truncated p53_1‐312 and the C‐terminally truncated FOXO4_15‐217 was also performed (Table S3 and Figure S14c). As noted, most intermolecular cross‐links between these two constructs connect the central region of p53 DBD (region 127–181) to the C‐terminal region of FOXO4 FH‐DBD (region 164–217), thus similarly to the p53_1‐393:FOXO4_1‐505 and p53_1‐312:FOXO4_1‐505 complexes as mentioned above. This indicates that the truncated constructs used in NMR analysis have conformation and the interacting surface similar to full‐length proteins.

2.5. Complex formation reduces the DNA‐binding affinity of p53 but not of FOXO4

Both NMR and chemical cross‐linking coupled to MS suggested that DNA‐binding surfaces of p53 and FOXO4 are involved in the interaction between these two proteins. If so, the formation of the complex could affect their DNA‐binding properties. To test this hypothesis, we performed fluorescence anisotropy measurements using fluorescein‐tagged double‐stranded DNA carrying either the insulin‐response element (DNA^IRE) ²⁰ recognized by FOXO4 or the full site of the p21 promoter (DNA^p21) recognized by p53.

Our results showed that full‐length p53_1‐393 binds to DNA^p21 with a K _D ^HIGH of 10 ± 1 nM and a K _D ^LOW of ~7 ± 4 μM per p53 dimer, as determined by fitting the anisotropy data using a two‐site binding model ³⁹ (Figure 5a). Subsequently, we characterized the DNA‐binding affinity of p53_1‐393 in the presence of FOXO4_1‐505. Surprisingly, our anisotropy measurements revealed that FOXO4_1‐505 alone binds to DNA^p21 with an affinity similar to that of p53_1‐393 with a K _D ^HIGH of 51 ± 4 nM and a K _D ^LOW of 9 ± 2 μM, using the two‐site binding model (Figure S17a), which prevented us from evaluating the p53_1‐393 binding affinity in the presence of FOXO4_1‐505.

FIGURE 5.

Fluorescence anisotropy (FA) measurements. (a) FA of fluorescein‐labeled DNA^p21 titrated by p53_1‐393 with and without FOXO4_1‐505 H3‐Mut. (b) FA of fluorescein‐labeled DNA^IRE titrated by FOXO4_1‐505 with and without p53_1‐393. The binding affinities were determined by fitting changes in fluorescence anisotropy as a function of protein concentration using one‐ or two‐site binding models. All data points are the means ± SD of three replicates

To overcome this problem, we designed a DNA‐binding deficient triple FOXO4_1‐505 R155A, H156A, S159A mutant (denoted as FOXO4_1‐505 H3‐mut) in which all three mutated residues are located within the helix H3 of the Forkhead domain and participate in DNA binding. ⁴⁰ Indeed, our fluorescence anisotropy measurements confirmed that the ability of FOXO4_1‐505 H3‐mut to bind to DNA^p21 is significantly reduced (K _D ~ 2.0 ± 0.6 μM, using the two‐site binding model, Figure S17a). In the presence of 10 μM FOXO4_1‐505 H3‐mut, the DNA^p21‐binding affinity of p53 considerably suppressed (Figure 5a), highlighting the absence of high‐affinity binding.

When using a one‐site binding model, FOXO4_1‐505 alone binds to DNA^IRE with a K _D of 31 ± 2 nM, which slightly decreases in the presence of 40 μM full‐length p53_1‐393 (K _D of 97 ± 3 nM) (Figure 5b). The control titration revealed that p53_1‐393 alone can also bind to DNA^IRE, albeit with a significantly lower binding affinity than FOXO4_1‐505, that is, with a K _D of 530 ± 20 nM per p53 dimer (Figure S17b) and, as such, without a strong interference with the FOXO4:DNA^IRE interaction. Taken together, these data suggest that the formation of the complex between p53 and FOXO4 considerably reduces the DNA‐binding affinity of p53 but not of FOXO4.

3. DISCUSSION

The transcription factors p53 and FOXO4 have multiple domains and inherently disordered segments, and their interaction involves a complex pattern of contacts between various regions of both proteins. The key region of the binding interface is located between p53 TAD (especially TAD2 segment) and several segments of the FOXO4 FH‐DBD DNA‐binding surface. Additional contacts involving the N‐ and C‐terminal disordered segments of FOXO4, the CRD of p53 and the DBDs of both proteins further stabilize the complex. Although a previous study had proposed that the N‐terminal region of FOXO4 FH‐DBD was involved in the interaction with p53, ³⁶ the role of other domains and disordered segments of p53 and FOXO4 in the formation of the complex and the impact of these interactions on DNA‐binding affinities of both proteins have not been elucidated until now.

Our SV AUC analysis of various p53 and FOXO4 constructs revealed that the formation of a stable complex requires N‐terminal disordered segments of both transcription factors (Figure 2). The N‐terminal region of p53 contains TAD and PRD, whereas the N‐terminal segment of FOXO4 has no structured domain other than the conserved region 1 (CR1) with the Akt phosphorylation site/14–3‐3 binding motif (Figure 1). Simultaneously removing both N‐terminal segments (in p53, the N‐terminal 60 residues) completely abrogated the formation of the complex, with no interaction between p53 DBD (or the p53 PRD‐DBD) and FOXO4 FH‐DBD (Figure S2e,f). Accordingly, the first 60 amino acid residues of p53 containing TAD are crucial for p53 binding to FOXO4, and the N‐terminal segment of FOXO4 stabilizes the complex. It should also be noted that full‐length p53 forms tetramers and the presence of four N‐termini in the p53 tetramer may increase FOXO4 binding.

Our mapping of the interaction interface by NMR and chemical cross‐linking confirmed the key role of p53 TAD in FOXO4 binding, with the most significant CSPs in residues of the TAD2 region (Figure 4a,e,f). Our data also showed that p53 TAD interacts with the DNA‐binding surface of FOXO4 FH‐DBD (Figure 3, S6), in line with a previous report, which had identified the N‐terminus of FOXO4 FH‐DBD as an important region of the p53‐binding surface. ³⁶ p53_1‐312 titration by FOXO4_15‐217 also unveiled weak, but significant, CSPs in several residues of p53 PRD and DBD (Figure 4c,d), likely reflecting contacts between p53 DBD and FOXO4 FH‐DBD, as corroborated by chemical cross‐linking (Table S1, Figure S14a). Concurrently, previous studies have reported that p53 TAD interacts with the DNA‐binding surface of the central p53 DBD, including the C‐terminal helix H3. ¹⁰ , ¹¹ , ¹² This region of p53 DBD is also affected by FOXO4_15‐217. These CSPs observed in residues of helix H3 may therefore also reflect the dissociation of TAD from the DNA‐binding surface of p53. Ultimately, chemical cross‐linking experiments revealed connections between the TD‐ and CRD‐containing C‐terminal region of p53 and various regions of FOXO4, thus indicating its involvement in FOXO4 binding. Furthermore, NMR experiments indicated the presence of a weak interaction between the isolated C‐terminal disordered segment of FOXO4_198‐505, which contains TAD, and the C‐terminally truncated p53_1‐312 (Figures S10b and S13). This is consistent with previous reports of interaction between p53 DBD and FOXO TAD domains. ³³ , ⁴¹ However, a yet another interaction of FOXO4 TAD with p53 cannot be ruled out, as TAD of closely related FOXO1 (Figure S1) has been shown to interact with the CRD domain of p53. ⁴²

Our NMR measurements with FOXO4_15‐217 did not reveal any changes in residues of the N‐terminal segment of FOXO4 because we observed significant CSPs only in residues of the FH‐DBD. However, this FOXO4 region appears to be important for the overall stability of the p53:FOXO4 complex (Figure 2a,b) and for its contacts with p53, as corroborated by chemical cross‐linking (Table S1 and Figure S14a). Accordingly, this flexible region of FOXO4 may interact with p53 through highly heterogeneous transient contacts commonly observed in complexes of intrinsically disordered proteins, which can remain unfolded in bound states and adopt multiple conformations. ⁴³ , ⁴⁴ Another possibility could be that the removal of the N‐terminal part of FOXO4 affects the behavior of FH‐DBD and thus its interaction with p53 TAD.

Based on our NMR and chemical cross‐linking experiments, we inferred that the DBDs of both proteins are located close to each other in the p53:FOXO4 complex, which could affect their DNA‐binding properties. In line with these results, our fluorescence anisotropy measurements showed a significant reduction in the high‐affinity binding of p53_1‐393 to DNA containing the full site of the p21 promoter in the presence of FOXO4_1‐505 H3‐Mut (Figure 5a). We also observed that the binding affinity of WT FOXO4_1‐505 to DNA containing the full site of the p21 promoter was similar to that of full‐length p53 (Figure S17a). This motif contains two 5′‐AACA‐3′ repeats, and this sequence is also present in motifs recognized by FOXO proteins. ¹⁸ , ¹⁹ , ²⁰ As a result, FOXO4_1‐505 and p53_1‐393 bind to the motif with similar affinity.

The inhibition of p53_1‐393 binding to the DNA^p21 may derive from steric blocking of the C‐terminal region of p53 DBD by FOXO4 (Figure 4c,d). Furthermore, p53 CRD, which is also involved in FOXO4 binding (Figure S14a), may be responsible for the initial contacts with DNA and for stabilizing the p53:DNA complex through nonspecific interactions. ⁴⁵ In a similar experiment, FOXO4_1‐505 binding to DNA containing the IRE motif decreased by only three‐fold in the presence of p53_1‐393 (Figure 5b). In other words, neither the primary interaction between p53 TAD and the N‐terminal region of FH‐DBD nor other contacts involving FH‐DBD block FOXO4 binding to its target DNA. For now, the biological role of this FOXO4‐mediated inhibition of p53 binding to DNA remains unclear, but this inhibition combined with the ability of FOXO4 to bind to DNA^p21 could explain the recently reported FOXO4‐driven expression of the p53‐target p21 in senescent Leydig cells. ⁴⁶

In conclusion, p53 and FOXO4 interacts through a complex pattern of contacts involving structured and disordered regions of both proteins. The interactions between p53 TAD2 and the N‐terminal region of FOXO4 FH‐DBD are particularly important for the overall stability of the complex. Furthermore, contacts involving the flexible N‐terminal segment of FOXO4, the C‐terminal negative regulatory domain of p53 and the DNA‐binding domains of both proteins stabilize the complex, whose formation significantly reduces the DNA‐binding affinity of p53 but not that of FOXO4. In addition, another study characterizing the interaction between FOXO4 and p53 was published during the revision of this article, which confirmed the interaction between FOXO4 DBD and p53 TAD. ⁴¹

The questions as to whether individual FOXO proteins differ in their interactions with p53 and whether complex formation masks the nuclear export sequence of p53 will undoubtedly be answered in subsequent studies. Further research is also required to assess the effects of post‐translational modifications of both proteins on their interaction and to understand the role of FOXO4‐mediated inhibition of p53 ability to bind DNA. Yet, targeting the interaction between p53 and FOXO4 is an attractive opportunity for drug development aimed at the selective elimination of senescent cells. ⁴ , ³⁶ Thus, our findings should prompt future research on this key protein–protein interaction and on its role in cellular senescence, with a significant translational output.

4. MATERIALS AND METHODS

4.1. Protein preparation

Expression and purification of all p53 and FOXO4 constructs are described in Supporting Information.

4.2. Preparation of isotopically labeled proteins for NMR measurements

Proteins for ¹H‐¹⁵N HSQC NMR experiments were expressed in minimal media supplemented with 1 g/L ¹⁵N‐ammonium chloride as the sole nitrogen source and purified as described in Supporting Information. To prepare ²H,¹⁵N,¹³C‐labeled p53_1‐312 the cells were first pre‐cultured in 200 ml of H₂O minimal media containing 1 g/L ¹⁵N‐ammonium chloride and 4 g/L ¹³C‐glucose as the sole nitrogen and carbon sources. The culture was grown at 37°C until OD⁶⁰⁰ reached 0.3. At this point, the cells were transferred into 1 L of D₂O minimal media supplemented with ¹⁵N‐ammonium chloride and ¹³C‐glucose and grown at 37°C until OD⁶⁰⁰ reached 0.8. Then, ZnSO₄ was added to a final concentration of 150 μM and protein expression was induced with 0.5 mM IPTG, subsequently adding 2.5 mL of 20% ¹³C‐glucose. The protein was expressed overnight (16–18 hours) at 15°C and purified as described in Supporting Information.

4.3. Analytical ultracentrifugation measurements

Sedimentation velocity analytical ultracentrifugation (SV AUC) experiments were performed using a ProteomLab™ XL‐I analytical ultracentrifuge (Beckman Coulter). The proteins FOXO4 and p53 were dialyzed against a buffer containing 20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM TCEP before analysis. Buffer density and the viscosity and partial specific volume of the samples were estimated using the SEDNTERP program. SV experiments were performed at a loading concentration of 20 μM p53 with various loading concentrations of FOXO4 ranging from 2 to 200 μM in charcoal‐filled Epon centrepieces with a12‐mm optical path length at 20°C and 48,000 r.p.m. rotor speed (An‐50 Ti rotor; Beckman Coulter). All sedimentation profiles were recorded with interference optics, and the diffusion‐deconvoluted sedimentation coefficient distributions c(s) were calculated from raw data using the software package SEDFIT. ⁴⁷ The calculated distributions were integrated to establish the weight‐average sedimentation coefficients corrected to 20°C and the density of water, s _w(20,w). The s _w values were plotted as a function of FOXO4 concentration to construct s _w isotherms. These s _w isotherms were fitted to a A + B ⇌ AB model, as implemented in the software package SEDPHAT, ⁴⁸ with known s _w values of individual components.

4.4. NMR data collection and analysis

For ²H,¹³C and ¹⁵N p53_1‐312 resonance assignment, a 350‐μl sample of 290 μM protein in a buffer containing 25 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM TCEP and 5% D₂O was prepared. A series of standard triple resonance experiments (HNCO, HN(CA)CO, HNCACB, CBCA(CO)NH) were acquired at 298 K on a BrukerAvance III™ HD 850 MHz spectrometer equipped with a ¹H/¹³C/¹⁵N cryoprobe. Subsequently, 130 μM ²H,¹³C,¹⁵N‐labeled p53_1‐312 was titrated with increasing stoichiometric equivalents of FOXO4_15‐217 (1:0.5; 1:1; 1:2 and 1:4.5) and FOXO4_198‐505 (1:0.5; 1:1 and 1:2) in a buffer containing 25 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM TCEP and 5% D₂O. Both titrations were measured on a BrukerAvance III™ HD 850 MHz spectrometer equipped with a ¹H/¹³C/¹⁵N cryoprobe. The other NMR spectra were acquired on a Bruker Avance III™ HD 600 MHz spectrometer equipped with a ¹H/¹³C/¹⁵N cryoprobe. All spectra were recorded at 298 K, except for p53_1‐93 spectra, which were recorded at 293 K. The ¹H‐¹⁵N HSQC spectra of 100 μM uniformly labeled ¹⁵N‐labeled p53_1‐93 were measured in a buffer containing 25 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM DTT and 5% D₂O. The sequence‐specific backbone resonance assignment was transferred from published data. ^{49 15}N‐labeled p53_1‐93 was titrated with increasing stoichiometric equivalents of FOXO4_15‐217 (1:0.5; 1:1; 1:1.5 and 1:2, respectively).

For ¹³C and ¹⁵N FOXO4_15‐217 resonance assignment, an approach similar to that described above was used for a 350‐μl sample of 500 μM protein in a buffer containing 20 mM sodium phosphate (pH 6.5), 150 mM NaCl, 1 mM EDTA, 1 mM TCEP and 5% D₂O. ¹H‐¹⁵N HSQC spectra of ¹⁵N‐labeled FOXO4_15‐217 and ¹⁵N‐labeled FOXO4‐DBD_86‐211 were recorded in a buffer containing 50 mM sodium phosphate (pH 6.5), 150 mM NaCl, 1 mM DTT and 5% D₂O. The backbone resonance assignment of FOXO4‐DBD has been published previously. ^{50 15}N‐labeled 300 μM FOXO4‐DBD_86‐211 was titrated with increasing stoichiometric equivalents of p53_1‐312 (1:0.2; 1:0.4; 1:0.8 and 1:1, respectively), and ¹⁵N‐labeled 200 μM FOXO4_15‐217 was titrated with increasing stoichiometric equivalents of p53_1‐312 (1:0.75; 1:1 and 1:1.25, respectively). In addition, 100 μM FOXO4_15‐217 was also titrated with increasing stoichiometric equivalents of p53_1‐93 (1:0.5; 1:1 and 1:1.5 respectively).

All NMR experiments were performed in 5 mm Shigemi NMR tubes (Shigemi Co., LTD, Japan). All spectra were processed using TopSpin software (v3.6) and evaluated using Sparky software (v3.1). ⁵¹ CSP values derived from 2D ¹H‐¹⁵N HSQC experiments were calculated using the following formula: $\mathrm{CSP}=\sqrt{\left[{∆\delta }_{\mathrm{H}}^2+{\left(\frac{1}{5}∆{\delta }_{\mathrm{N}}\right)}^2\right]}$. ⁵² CSPs larger than the mean + 1σ_cor were then taken as significant. The secondary structure of p53_1‐312 was derived from the assigned backbone chemical shifts (HN, HA, CA, CO and N) using TALOS+. ⁵³

NMR resonance assignments of p53_1‐312 and FOXO4_15‐217 were deposited in the Biological Magnetic Resonance Bank (BMRB codes: 50960 and 51081, respectively).

4.5. Chemical cross‐linking coupled to mass spectrometry

Chemical cross‐linking coupled to mass spectrometry is described in Supporting Information.

4.6. Fluorescence anisotropy measurements

Fluorescence anisotropy measurements are described in Supporting Information.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Raju Mandal: Investigation (lead); writing – review and editing (equal). Klara Kohoutova: Investigation (equal); writing – review and editing (equal). Olivia Petrvalska: Investigation (equal); methodology (equal); writing – review and editing (equal). Matej Horvath: Investigation (equal); methodology (equal); writing – review and editing (equal). Pavel Srb: Methodology (equal); writing – review and editing (equal). Vaclav Veverka: Methodology (equal); supervision (equal); validation (equal); writing – review and editing (equal). Veronika Obsilova: Conceptualization (lead); investigation (supporting); supervision (lead); validation (equal); writing – review and editing (equal). Tomas Obsil: Conceptualization (lead); investigation (supporting); supervision (lead); validation (equal); writing – review and editing (equal).

Supporting information

Appendix S1: Supporting Information

Mandal R, Kohoutova K, Petrvalska O, Horvath M, Srb P, Veverka V, et al. FOXO4 interacts with p53 TAD and CRD and inhibits its binding to DNA . Protein Science. 2022;31(5):e4287. 10.1002/pro.4287