Understanding the interaction of 14‐3‐3 proteins with hDMX and hDM2: a structural and biophysical study

p53 plays a critical role in regulating diverse biological processes: DNA repair, cell cycle arrest, apoptosis and senescence. The p53 pathway has therefore served as the focus of multiple drug‐discovery efforts. p53 is negatively regulated by hDMX and hDM2; prior studies have identified 14‐3‐3 proteins as hDMX and hDM2 client proteins. 14‐3‐3 proteins are adaptor proteins that modulate localization, degradation and interactions of their targets in response to phosphorylation. Thus, 14‐3‐3 proteins may indirectly modulate the interaction between hDMX or hDM2 and p53 and represent potential targets for modulation of the p53 pathway. In this manuscript, we report on the biophysical and structural characterization of peptide/protein interactions that are representative of the interaction between 14‐3‐3 and hDMX or hDM2. The data establish that proximal phosphosites spaced ~20–25 residues apart in both hDMX and hDM2 co‐operate to facilitate high‐affinity 14‐3‐3 binding and provide structural insight that can be utilized in future stabilizer/inhibitor discovery efforts.


Introduction
The transcription factor p53 is often described as 'the guardian of the genome' because of its central role in regulating diverse biological processes: DNA repair, cell cycle arrest, apoptosis and senescence [1]. Thus, the p53 pathway has seen significant attention in developing new anticancer treatments [2]. One such target has been the interaction between p53 and hDM2 or hDMX (Fig. 1A) [3][4][5]. p53 is negatively regulated by hDMX and hDM2 [6][7][8][9][10]; they bind to the transactivation domain of p53 sterically blocking its interaction with DNA [9][10][11][12], p53 is additionally targeted for nuclear export and proteasomal degradation through hDM2 which bears an E3 ligase domain [13][14][15]. Although hDMX lacks such a domain, the E3 ligase activity of hDM2 is enhanced through formation of hDMX/hDM2 heterodimers [16][17][18]. This continuous targeting of p53 for degradation keeps p53 levels low under normal conditions. It is only under stressed conditions (e.g. DNA damage, hypoxia, oncogene activation and ribosomal stress) that p53 is rapidly activated through multiple post-translational modifications; subsequent stabilization and accumulation of p53 upregulates gene expression leading to cell cycle arrest and/or apoptosis. This tightly controlled autoregulatory loop is thus of vital importance; knockout studies of the oncogenes MDMX and MDM2 (the murine homologues of hDMX and hDM2) have shown that these proteins are essential in mice studies [19]. However, hDM2 and hDMX are themselves subject to regulatory control and have additional independent roles in cell cycle regulation [20][21][22]. For instance, in addition to modulating p53 binding, post-translational modification of hDMX and hDM2 such as phosphorylation influences their localization and degradation [21][22][23][24][25].
Several kinases have been reported to phosphorylate hDMX and/or hDM2 whilst cellular studies have shown that phosphorylated hDMX and hDM2 interact with 14-3-3 [26]. First identified in 1967 [27], 14-3-3 proteins are eukaryotic adapter proteins that control diverse physiological functions through interaction with a large network of proteins, for example, they are involved in signal transduction, protein trafficking, apoptosis and cell cycle regulation [28]. Given these functions, they have been shown to play an important role in inflammatory conditions, neurodegenerative diseases, cancer and cystic fibrosis [29][30][31][32][33]. And have consequently received attention as potential drugdiscovery targets [28,34]. The 14-3-3 family of proteins consist of seven isoforms in humans: β, γ, ε, η, ζ, σ and τ/θ, all of which share a high degree of sequence conservation [35] (see Fig. S1). Even though sequence conservation between the isoforms can indicate redundancy in protein function, this is not the case for 14 - 1. Negative regulation of the p53 pathway by hDMX and hDM2: (A) hDMX (turquoise) and hDM2 (green) bind p53 (blue) and subsequently inhibit its transcriptional activity. hDM2 also facilitates nuclear export of p53, and as a ubiquitin ligase additionally tags p53 for degradation, whilst hDMX accelerates hDM2 mediated ubiquitin ligase activity. 14-3-3 (orange) interacts with: (i) phosphorylated p53 to stabilize tetramer formation, (ii) phosphorylated hDMX leading to its nuclear accumulation and degradation, and, possibly modulation of its interaction with hDM2 and (iii) hDM2 to facilitate its nuclear export. (B) domain structure of hDM2 and hDMX highlighting phosphosites on hDM2 and hDMX investigated in this work: Ser342 and Ser367 in intracellular localization, protein degradation and ubiquitination, and additionally Ser367 is associated with cell growth and protein stabilization. Ser166 and Ser186 on hDM2 are involved in apoptosis, signalling pathway regulation and ubiquitination.
protein are clear between the isoforms [36,37]. The 14-3-3 proteins exist as homo or heterodimers where each monomer consists of nine α-helices forming an amphipathic binding groove. Four α-helices are directly responsible for dimer formation, while the rest of the α-helices form the 'W' shape of the protein dimer [38,39]. The N terminal domain of 14-3-3 proteins modulates homo/heterodimerization while the C terminal domain binds client proteins [38]. Activity and modulation of many cellular pathways by 14-3-3 is controlled by phosphorylation events. Target proteins are typically recognized by 14-3-3 through well-defined sequences that can be generalized into three categories: (I) RXX pS/pT XP, (II) RX(Y/F)X pS/pT XP and (III) XX pS/pT X COOH (pS and pT being phosphorylated Ser and Thr, respectively, X being any amino acid). [40] In modes I and II, proline is always located at position +2 in relation to the central pSer/pThr and arginine is at position −3 in mode I. Mode III represents C terminal sequences where pSer/pThr are the penultimate amino acid of the partner protein.
Phosphorylated residues on hDMX or hDM2 are considered relevant to recognition by 14-3-3 proteins, as well as their role in cell signalling is outlined in Fig. 1B [26,[41][42][43][44][45]. Relevant sites on hDMX: Ser342 and Ser367 are targeted by several kinases (e.g. AMP [45], Akt [41], Chk2 [26,44], or Chk1 [42]). hDMX phosphorylation in response to stress or DNA damage with consequent enhancement of hDM2 mediated hDMX degradation has been reported [44,46], together with a role for Ser367 [43]. Phosphorylation of hDMX on Ser367 by Chk2 was shown to promote 14-3-3 binding, hDMX nuclear import and degradation by hDM2 [26]. Similarly, Chk1 was shown to phosphorylate hDMX at Ser367, to promote 14-3-3γ/hDMX interaction and cytoplasmic retention of hDMX to counter hDMX-enhanced p53 ubiquitination, leading to its stabilization and activation [42]. However, the potential relevance of both Ser342 and Ser367 to interaction with 14-3-3 has been noted [44,47]. In contrast, Akt has been shown to phosphorylate hDMX on Ser367, promoting 14-3-3 binding, stabilizing hDMX and consequently enhancing hDM2 stability leading to suppression of p53 transcriptional activity [41]. Adenosine monophosphate (AMP) mediated phosphorylation of hDMX on Ser342 has been shown to induce hDMX/14-3-3 interaction resulting in inhibition of p53 ubiquitylation and consequent stabilization [45]. hDM2 phosphorylation in the NLS/ELS region, by PKB/Akt, can lead to its stabilization [48,49], although the effects are complex; phosphorylation on Ser166 and Ser188 are promoted by Akt kinase [50], and result in positive regulation of hDM2 by inhibiting its self-ubiquitination and nuclear translocation [48,51,52]. Phosphorylation at Ser166 and Ser188 does not appear to result in binding of hDM2 to 14-3-3 [53] whilst such an interaction was observed when phosphorylated at Ser166 and Ser186 by Pim kinase, which simultaneously suppresses phosphorylation of Ser188 [53]. Finally, the interaction between hDM2 phosphorylated in the ring domain and 14-3-3σ was observed to accelerate the self ubiquitination of hDM2, with a resulting downstream stabilization of p53 [54]. Thus cellular studies have linked hDMX and hDM2 to 14-3-3 proteins; 14-3-3 proteins may indirectly modulate the interaction between hDMX or hDM2 and p53 and represent potential targets for pharmacological modulation of p53 [21,23] but the molecular bases of these interactions remain unexplored. In this manuscript, we report on the biophysical and structural characterization of peptide/protein interactions that are representative of the interaction between 14-3-3 and hDMX or hDM2. The data reveal the interactions to be mode I [55,56]; whilst such 14-3-3 interactions are not stabilized by typical 14-3-3 interaction stabilizers, for example, Fusicoccin and Cotelynin A [57]. These data provide structural insights that could be utilized in future stabilizer/inhibitor discovery efforts. More significantly, our results reveal that in both cases proximal phosphosites spaced~20-25 residues apart in both hDMX and hDM2 co-operate to facilitate highaffinity 14-3-3 binding and do so through 1 : 1 stoichiometry.

Results
There are multiple serine and threonine residues in the hDMX and hDM2 proteins that could be phosphorylated (protein sequences with highlighted Ser and Thr residues on hDMX and hDM2, as well as their alignment [35] can be found in Figs S2 and S3). Therefore, 14-3-3Pred [58] was used to identify sequences from hDM2 and hDMX that might serve as 14-3-3 binding sequences. 14-3-3Pred is a webserver that predicts 14-3-3 binding sites in client proteins based on published examples of 14-3-3 interactions with specific amino acids surrounding Ser/Thr. Scores are assigned to each Ser/Thr residue on the protein target whose sequences in FASTA format are queried, using an ensemble approach comprising three different base classification systems: position-specific scoring matrix, support vector machines and artificial neural network. End score values are calculated as the average of three independent methods with a score > 0.5 indicating a high probability of the sequence binding to 14-3-3. The webserver also provides information on the phosphorylation state for each Ser/Thr residue. The phosphorylated sites on hDMX and hDM2 shown to be relevant for the interaction of 14-3-3 in previous cell-based studies [26,[42][43][44][45]50,53] were predicted to have a high probability of 14-3-3 binding (Table 1), except pSer342 on hDMX, which showed a low probability of binding (score 0.19) whereas Ser151 on hDMXa previously unreported site of phosphorylationwas predicted to be a potential 14-3-3 binding site (score 0.75).
Using the 14-3-3Pred results server, several key regions of hDMX and hDM2 were thus identified. Peptides containing these binding regions were designed bearing five/six residues flanking the central phosphorylated residue on either side and prepared by solid-phase peptide synthesis (SPPS). Given the proximity of pSer342 and pSer367 from hDMX, as well as pSer166 and pSer186 from hDM2, we also prepared doubly phosphorylated sequences for both to probe for avidity effects. Peptides were synthesized using a combination of manual or automated Fmoc SPPS on rink amide resin (Rink amide ProTide for doubly phosphorylated sequences) and modified at the N terminus by acetylation for isothermal titration calorimetry (ITC), surface plasmon resonance (SPR) and crystallography studies or with 5,6-carboxyfluorescein following an Ahx linker to function as a tracer in fluorescence anisotropy (FA) assays (Table 1 for peptide sequences, see Table S1 for peptide characterization).
Biophysical characterization of the interaction of hDMX and hDM2 peptides with 14-3-3 To determine the binding affinity between all 7 isoforms of 14-3-3 proteins and hDMX and hDM2 peptides, FA experiments were carried out. 14-3-3 proteins were titrated against a constant concentration of the tracer resulting in an increase of anisotropy upon engaging 14-3-3 proteins ( Fig. 2A-C for 14-3-3 binding hDMX sequences and Figs S4-S6, for 14-3-3 binding hDM2 sequences, non-binding sequence, and unphosphorylated controls). A noteworthy feature of the data is that different maximum anisotropy levels were observed between hDMX or hDM2 peptides and 14-3-3 isoforms, indicating different diffusion properties of the fluorophore when bound to the different 14-3-3 isoforms. This may: (a) reflect differences in the interaction between fluorophore and protein [59] restricting the former's mobility to a greater or lesser degree between isoforms or (b) arise as a consequence of a more complex equilibrium, for example, isoform dependent variation in 14-3-3 dimerization [60] across the concentration range of the assay. This will be investigated in future studies. The consequence of this behaviour is that for weaker binding peptides where saturation was not observed, it was not possible to determine a K d value. Thus, in some cases, for example, hDM2 180-192 pSer186 , K d values (see below) with τ, σ and ε could only be estimated to be greater than limiting values. Similarly, accurate fitting for doubly phosphorylated peptides could be obtained only for η, γ and β isoforms. Peptides hDMX 335-349 pSer342 and hDMX 361-374 pSer367 showed affinity for 14-3-3 proteins, whereas peptide hDMX 144-158 pThr151 exhibited no response under the conditions of the assay (see Fig.  S5). None of the corresponding unphosphorylated control hDMX sequences exhibited evidence of binding under the conditions of the assay confirming the role of phosphorylation for binding of pSer342 and pSer367 to 14-3-3 (see Fig. S6). hDM2 peptides representing sites pSer166 and pSer186 showed low micromolar affinities to all 14-3-3 isoforms (Fig. S4).
Isothermal titration calorimetry was used as an orthogonal assay to confirm the binding affinities of hDMX and hDM2 peptides to 14-3-3η and to gain additional information about the stoichiometry of the interactions. Moreover, such experiments provide thermodynamic data and as a label-free method require no immobilization/functionalization of any of the binding partners. Here data were obtained by titration of the peptide into 14-3-3η protein (Fig. 2D,E and Fig. S7). Thermodynamic parameters for interaction of 14-3-3 with hDMX and hDM2 peptides obtained by ITC are summarized in Table 3.
Isothermal titration calorimetry data showed similar trends for binding of hDMX peptides to 14-3-3η to those observed in the FA assays; binding affinities of hDMX 361-374 pSer367 , hDMX 335-349 pSer342 and hDMX 335- nM. An increase in binding affinity when both binding sites of hDM2 are present was also confirmed, with K d = 151 AE 57 nM. The stoichiometry between doubly phosphorylated hDMX and hDM2 peptides and 14-3-3η was found to be approximately 0.94 indicating a 1 : 1 binding model. If two phosphorylated sites from one peptide were binding simultaneously, the expected stoichiometry would be around 0.5. The contrasting absence of two binding events in the ITC data, when compared to the FA experiments, can be accounted for by the different experimental configurations. In the former, titration of the peptide into protein means that protein begins in excess and becomes saturated with the highest affinity site, whereas for FA, peptide begins in excess and protein (and therefore monomer/dimer ratio) varies in concentration across the experiment. Finally, association and dissociation rates for the binding of hDMX and hDM2 peptides to 14-3-3η were determined by SPR (Fig. 2E,F Table 3. Thermodynamic parameters for hDMX(2) peptides/14-3-3 proteins. and hDM2 180-192 pSer186 ) it was only possible to determine steady-state affinities, but not their kinetic profile (see Supporting Information). The doubly phosphorylated peptides showed the expected increase in affinity, which could be attributed to a much slower k off rate in comparison to singly phosphorylated hDMX 361-374 pSer367 and hDMX 335-349 pSer342 peptides. The 1 : 1 stoichiometry between the doubly phosphorylated peptides and 14-3-3 contrasts with previously reported work [61,62], although it may be that the length and/or accessible conformation of the hDMX and hDM2 peptides are incompatible with the simultaneous engagement of both 14-3-3 protomers in the dimeric structure. To further explore this behaviour and understand the basis of cooperativity and affinity enhancement we performed additional experiments. To exclude the possible formation of higherorder oligomers (e.g. 14-3-3 tetramers held together by two peptides bridging two dimers), an analytical ultracentrifugation experiment was performed. The 14-3-3η (at the concentration of 14 μM, approximately corresponding to the second binding events observed in FA, Fig. 2C) and hDMX 335-373 pSer342/pSer367 peptide in two different concentration ratios (1 : 1 and 1 : 0.5, protein vs. peptide) were analysed. The sedimentation coefficient value remained unchanged for 14-3-3 protein alone and when varying peptide/protein concentrations, indicating that only one dimer of 14-3-3η is involved in binding (see Fig. S10). Secondly, peptide length may also have a significant effect on the affinity and binding mode. This has been demonstrated in a recent study with p53 peptides with 14-3-3σ, where variation in the peptide length (from 9 to 32 residues) correlated with increased binding affinities [64]. Two long variants of hDMX 335-373 pSer342/pSer367 peptide were synthesized, with only one phosphorylation site: pSer367 or pSer342, extending the peptide length by 26 amino acids. The two peptides were tested in FA assays to decouple the influence of peptide length from the co-operative influence of having two phosphosites (Fig. 3). The hDMX 335-373 pSer342 peptide showed an increase in binding affinity with 14-3-3η (K d = 5.4 AE 1.1 µM) as expected, in comparison to the short version of the hDMX 335-349 pSer342 peptide (K d = 20.0 AE 1.2 μM). On the other hand, increasing the peptide length for hDMX 335-373 pSer367 decreased the binding affinity (K d = 0.8 AE 0.1 µM) and significantly lowered the maximum anisotropy signal. The lower maximum anisotropy signal observed might be due to the fluorophore being moved approximately 20 amino acids further from the binding pSer-367 (see earlier discussion on fluorophore mobility). Whilst there is a moderate increase in binding affinity of hDMX 335-373 pSer342 on increasing the peptide length this is insufficient for high-affinity interaction. Similarly, the slight loss in potency for the longer hDMX 335-373 pSer367 peptide relative to hDMX 361-374 pSer367 is noteworthy in considering the interaction with fulllength protein as these data hint at interference from the additional residues, and is therefore likely to be important in relating these results to the behaviour for the full-length proteins. Taken together, the data indicate that proximal phosphorylation of both sites in hDMX 335-373 pSer342/pSer367 is required for high affinity 1 : 1 stoichiometric binding between hDMX and   14-3-3 protomers. Given the absence of 1 : 2 binding, the presence of multiple sites may increase affinity through statistical rebinding or interaction with a second non-canonical phosphate-binding site on the surface of 14-3-3. The second binding event observed only in fluorescence anisotropy experiments, may represent a further non-specific binding event or an interaction linked to 14-3-3 dimerization [60] and will be investigated in future work.
Structural characterization of the interaction between 14-3-3 and hDMX or hDM2 peptides The W-shaped structure of a 14-3-3 dimer provides a groove for phosphorylated client proteins to bind. One dimer usually accommodates two copies of phosphorylated sites from one target, but there are examples of two different targets simultaneously binding to one dimer of 14-3-3 proteins [61,62]. Phosphorylated Ser/ Thr residues from partner proteins generally bind directly between Lys49 and Arg56 in helix α3, and Arg127 and Tyr128 in helix α5 as they form a basic binding pocket in otherwise acidic surroundings. Structural data were obtained by solving three crystal structures of hDMX 361-374 pSer367 (PDB: 6YR5), hDM2 180-192 pSer186 (PDB: 6YR6) and hDMX 335-373 pSer342/pSer367 (PDB: 6YR7) in complex with 14-3-3σΔC (the C-terminus was truncated for the crystallography purposes). Refinement statistics for all three structures are given in the ESI as are additional images of each structure with electron density for peptides in each monomer of 14-3-3σ.
The hDMX 361-374 pSer367 /14-3-3σ structure was solved to a resolution of 2.25Å with a dimer and two additional monomers of 14-3-3σ (in green and blue) in the asymmetric unit, each binding a hDMX 361-374 pSer367 peptide (cyan) in a conserved amphipathic groove ( Fig. 4A and Fig S11 for additional images). Ten amino acids from hDMX 361-374 pSer367 (RTI(pS) APVVRP) could be built into the electron density map. The phosphoserine of hDMX 361-374 pSer367 is positioned in a basic pocket of 14-3-3σ making polar contacts with Lys49, Arg56, Tyr119 and Arg129 (presented as dashed lines), whilst Asn226 makes polar contacts with the backbone of the peptide (Ile-1 with reference to pSer-367), in agreement with previously published 14-3-3 structures [38]. Additional polar contacts were found between Arg-3 of hDMX 361-374 pSer367 and Glu182 of 14-3-3σ, alongside Ala+1 and Thr+2 on the peptide backbone with Asn175 and Trp230 on the protein respectively. Hydrophobic contacts were found between Ile-1, Ala+1 and Pro+2 of the peptide with multiple residues on the protein (presented as spheres).
Structural evidence for the interaction of hDM2 with 14-3-3 proteins was obtained by solving the novel crystal structure of the hDM2 180-192 pSer186 peptide in complex with 14-3-3σ, to a resolution of 1.75Å. The complex is crystallized in space group P1 with the asymmetric unit containing one dimer along with an additional two monomers of 14-3-3σ (monomers A-D) with pSer186 peptide in each groove (in cyan). Six amino acids of the peptide were resolved in the electron density: RHK(pS)DS (Fig. 4B see Fig.  S12 for additional images). pSer of the peptide is always located in the same position, between Lys49, Arg56, Arg129 and Tyr130 and the peptide is stabilized by polar contacts between Lys-1 and Asp225, His-2 and Glu182, Trp230, and the peptide backbone with Asn175 and Asn226. Arg-3 and Ser+2 peptide residues are also stabilized through hydrogen bonds with water molecules.
The structure of a doubly phosphorylated peptide hDMX 335-373 pSer342/pSer367 in complex with 14-3-3σ was solved to 2.1Å resolution (Fig. 4C). Here, 14-3-3σ crystallized as a dimer in space group C121 with the asymmetric unit containing one dimer (green) with additional electron density for the peptide (cyan) observed in both monomeric units of 14-3-3σ ( Fig. 4C and Fig. S13 for additional images). Based on previously published structures of doubly phosphorylated peptides, it was expected to observe pSer342 binding to monomer B and pSer367 binding to monomer A [31,61,65]. The electron density within the binding groove of monomer B could unambiguously be assigned to pSer342, whereby eight amino acids of the peptide could be resolved (HSL(pS)TSDI). The peptide was observed to adopt an unusual conformation by taking a sharp turn. A closer examination of the pSer342 binding site in monomer B of 14-3-3σ shows pSer342 binding to a basic pocket as expected, making polar contacts with Arg56, Arg129 and Tyr130. In this orientation, the Ile+4 from the peptide makes hydrophobic contact with Leu218 and Ile219 on the protein (represented as spheres). Peptide residues Thr+1 and Ser+2 make polar contacts with Lys122, Asn175, Glu182 and Trp230, while Asn226 makes polar contacts with the backbone of the peptide. Overall, the Cterminus of the peptide takes a sharp turn after the pSer-Thr residues and this peptide conformation is stabilized by hydrogen bonds between water molecules and Thr+1, Ser+2 and Asp+3 peptide residues. In contrast, the electron density map within the binding groove of monomer A indicated partial ligand occupancy and could be interpreted in different ways, indicating an overlay of multiple conformations (both pSer342 and pSer367 sites, with the dominant electron density map ascribable to pSer342 binding see Fig. S13 for additional images and discussion).

5349
The  1.9 μM). Interestingly, different maximum anisotropies were observed for peptide binding across the 14-3-3 isoforms, which may reflect a difference in the 14-3-3 dimerization affinity [60] or mobility of the fluorophore. Binding experiments with doubly phosphorylated peptides were observed to result in two binding events. As expected for a multivalent interaction with 14-3-3 proteins [63], the first binding event in both cases was higher affinity than either monovalent interaction, whereas the second binding event was in the micromolar regime (for 14-3-3η, hDMX 335-373 pSer342/pSer367 K d1 = 30.3 AE 6.4 nM and K d2 = 10.67 AE 1.2 μM, hDM2 160-192 pSer166/pSer186 ; K d1 = 49.6 AE 12.1 nM and K d2 = 17.5 AE 2.5 μM). This enhancement in affinity for the first binding event was also observed in ITC experiments which lacked the second binding event (but were performed by titration of peptide into protein in contrast to the FA experiments) and SPR experiments which indicated that a slower k off was associated with co-operative binding of doubly phosphorylated peptides. Importantly, for the doubly phosphorylated peptides, binding was negatively cooperative (i.e. less than additive; Table 4).
Taken together the data can be rationalized by assuming that the dimerization affinity of 14-3-3 differs across isoforms and occurs within the concentration regime of the FA experiments (Fig. 5). As the titration progresses, once 1 : 1 stoichiometry is attained, further addition of 14-3-3 results in unbound 14-3-3 monomer and eventually all peptides will be bound in complexes containing one peptide to two 14-3-3 monomers. For the doubly phosphorylated peptides early in the titration peptides bind with high affinity in 1 : 1 stoichiometryhigher affinity likely results from statistical rebinding or a second phosphosite on the surface of the 14-3-3 protein. The second binding event either derives from a non-specific interaction, or arises from 14-3-3 dimerization [60] but is visible in the titration as a distinct binding event due to the strong affinity of the first event. Such behaviour is not observed in ITC experiments because the peptide is titrated into 14-3-3 protein so its concentration does not vary, and, the peptide becomes the saturating component. Considering the structural data for hDMX 361-374 pSer367 , hDM2 [180][181][182][183][184][185][186][187][188][189][190][191][192] pSer186 and hDMX 335-373 pSer342/pSer367 bind to 14-3-3 directly between K49 and R56 in helix α3, and R127 and Y128 in helix α5 as is typical for 14-3-3 client proteins. The hDMX 361-374 pSer367 peptide is a Mode I sequence (RXXpS/pTXP) which can account for its enhanced affinity in comparison to the other sequences [38,56]. Curiously, for the structure containing the longer hDMX 335-373 pSer342/ pSer367 peptide, the sequence surrounding pSer342 dominates in the electron density for bound ligand despite being the weaker 14-3-3 binding site of the two individual pSer sequences, this may favour the hypothesis of a second phosphorecognition site over the statistical rebinding model. Moreover, in the context of earlier cellular studies [26,[41][42][43][44][45][48][49][50][51][52][53] the data highlight the complex nature by which phosphorylation regulates hDMX in particular [26,[41][42][43][44][45]. pSer367 may be sufficient for 14-3-3 binding and the consequent effects on localization, stability and ultimately p53 function; the presence of both pSer342 and pSer367 confers some increase in 14-3-3 binding affinity (particularly for other isoforms), and it is plausible that accessibility of the binding sites in the full-length protein may play a crucial role that requires phosphorylation at both sites in vivo. In contrast, a much clearer difference in binding affinity for hDM2  pSer166/ pSer186 is observed in contrast to both monophosphorylated peptides and points to a functional requirement for double phosphorylation.
In conclusion, we have performed a biophysical and structural study on the interaction of 14-3-3 with peptides derived from hDMX and hDM2. We have shown that peptides derived from hDMX containing pSer342 or pSer367 and from hDM2 containing pSer166 or pSer186 exhibit affinity for 14-3-3 as demonstrated by FA, ITC and SPR. Structural analyses reveal that the peptides bind to 14-3-3 directly between K49 and R56 in helix α3, and R127 and Y128 in helix α5 as is typical for 14-3-3 client proteins. For high-affinity binding, proximal phosphosites are beneficial and in the case of hDM2 essential. In contrast to other doubly phosphorylated peptides of this length the interaction involves one Table 4. Thermodynamic values for binding of hDM2 and hDMX peptides to 14-3-3η. peptide for each 14-3-3 protomer [61][62][63]. The additional affinity likely derives from a reduced k off rate for doubly phosphorylated peptides. Importantly, the results suggest that sequential phosphorylation of proximal sites in both hDMX and hDM2 plays a role in promoting changes in their localization and degradation in a cellular context. Thus, these data provide structural insight on the indirect effects of 14-3-3 on the p53 pathway, which may inform the development of chemical intervention strategies (e.g. use of kinase inhibitors) and provide starting points for structure-based ligand design of 14-3-3/hDMX and 14-3-3/hDM2 modulators, both stabilizers and inhibitors.

General remarks
Resins and amino acids were purchased from either Sigma-Aldrich (part of Merck Life Science UK Limited, Gillingham, Dorset, UK) or Novabiochem (part of Merck Life Science UK Limited, Gillingham, Dorset, UK). All amino acids were N-Fmoc protected and side chains protected with Boc (His, Lys), t-Bu (Asp, Glu, Ser, Thr), Pbf (Arg), Trt (Asn, Gln). Peptides were synthesized either manually or using a microwave-assisted automated peptide synthesizer (CEM Liberty Blue) on a 0.05 or 0.1 mmol scale. DMF used in peptide synthesis was of ACS grade and from Sigma-Aldrich.

Manual peptide synthesis
Manual peptide synthesis followed this cycle: swelling of a resin (20 min) in cartridge used for solid-phase synthesis, washing (DMF, 3 × 2 mL × 2 min), deprotection (Method A), and coupling of the desired amino acid (Method B), where successful coupling and deprotection were determined by a colour test (Method C). Acetylation (Method D) or coupling of a fluorescent dye (Method E) were performed prior to cleavage (Method F). Method A: Deprotection N-terminal Fmoc-protecting groups were removed by adding 20% piperidine in DMF (5 × 2 mL × 2 min) and washed with DMF (5 × 2 mL × 2 min) after.
Method B: Manual coupling of amino acid and Ahx. The desired amino acid or Ahx (5 equiv.), DIPEA (10 equiv.) and HCTU (5 equiv.) were dissolved in DMF (2 mL) and added to the resin, followed by agitation for 1 h. Reagents were removed by filtration and the resin was washed with DMF (3 × 2 mL × 2 min).
Method C: Kaiser test. Successful coupling or deprotection for any residue coupled manually was determined by the Kaiser test. A few resin beads were transferred into a vial and mixed with 2 drops of each of the solutions: 1 Ninhydrin (5% w/v) in ethanol, 2 Phenol (80% w/v) in ethanol, 3 1 mM KCN (aq.) in pyridine (2% v/v).
The solution was heated at 100°C for 5 min before observing the change in colour. Successful deprotection was observed by colour of the beads changing into blue, where successful coupling gave no change in colour.
Method D: N-terminal acetylation. Acetic anhydride (10 equiv.) and DIPEA (10 equiv.) were dissolved in DMF (2 mL) and the solution was transferred to the resin. After 2 h, the resin was drained and washed with DMF (3 × 2 mL × 2 min). Successful capping was determined by colour test (Method C).
Method F: Cleavage and deprotection of peptides of the resin.
After elongation and acetylation or fluorescent dye coupling were complete, the resin was washed with CH 2 Cl 2 (5 × 2 × 2 min), Et 2 O (5 × 2 mL × 2 min) and dried under vacuum. Peptides were cleaved and side-chain deprotected using 'Reagent K' TFA : EDT : Thioanisole : Phenol : H 2 O 82 : 3 : 5 : 5 : 5 (2 mL × 3 h). The peptide was precipitated in ice-cold Et 2 O (10 mL) and placed in a centrifuge (1107 g × 5 min). The supernatant was removed, the precipitate resuspended in ice-cold Et 2 O and placed in a centrifuge again (3×). The precipitate was dried under a stream of nitrogen overnight, before being dissolved in H 2 O and lyophilized.

Automated peptide synthesis method
Peptides prepared using automated peptide synthesizer followed cycles described below. Resin loading cycle cleans the reaction vessel, washes with DMF : DCM (1 : 1), transfers resin to the reaction vessel, washes with DMF : DCM (1 : 1), and drains the vessel at the end of a cycle. Deprotection and coupling cycle consists of: washing with DMF (4 mL), adding 20% piperidine in DMF (6 mL), microwave deprotection cycle (30 s), washing with DMF (4 + 4 + 4 + 4 mL), the addition of amino acid (2.5 mL, 5 eq or 3 eq for phosphorylated amino acids), coupling reagent (1 mL, 5 eq) and activator base (0.5 mL, 5 eq), coupling microwave cycle (5 min), washing with DMF (2 mL) and draining. HCTU and DIPEA were used during automated peptide synthesis as well. As a rule, all amino acids were coupled using 75°C coupling and deprotection cycles up to Ser(PO(OBzl)OH)-OH or Thr (PO(OBzl)OH)-OH, where conventional coupling and deprotection method was used (coupling at the rt for 2 h, deprotection at rt for 15 min) as well for every amino acid following pSer/pThr. After the final residue was coupled, the resin was ejected from the reaction vessel. Ahx coupling, deprotection, acetylation or florescent dye coupling, and cleavage were performed manually using the methods described above.

Peptide purification
Crude peptides were dissolved in H 2 O or DMSO and purified by UV-or MS-directed HPLC. Jupiter Proteo (250 × 21.2 mm) or a Kinetex EVO C18 (250 × 21.2 mm) preparative column (reversed-phase) was used with an increasing gradient of acetonitrile in water with 0.1% formic acid, over 30 min at the flow rate of 10 mLÁmin −1 . Fractions containing peptides were combined, concentrated and lyophilized. The purity of peptides was assessed by analytical HPLC and HRMS.

Protein expression and purification
The pProEx HTb-His-14-3-3 constructs were expressed in BL21(DE3) cells. A single colony from a freshly made agar plate (8 g LB broth mixed with 8 g agar in 400 mL, autoclaved and poured into Petri dishes to use for the transformation of plasmids) was picked and mixed with 5 mL of LB media with ampicillin to inoculate a starter culture overnight at 37°C. The cells were grown in 2 L of TB media (48 g peptone, 24 g yeast, 4.6 g KH 2 PO 4 , 24 g KHPO 4 , 5 mL glycerol in 2 L of dH 2 O, autoclaved for 20 min at 121°C) at 37°C until the OD reached 0.4-0.6. Expression was induced by adding 0.4 mM IPTG and agitating overnight at 18°C. The expression culture was spun down (7871 g, 20 min, 4°C), resuspended in 200 mL of lysis buffer (50 mM Tris, 300 mM NaCl, 12.5 mM imidazole, 2 mM β-mercaptoethanol) with 5 mM MgCl 2 and DNAse (1 : 1000). The cells were lysed by French press or sonication and the solid fragments were removed by centrifugation (49 192 g, 30 min, 4°C). The cleared lysate was loaded on a Ni 2+ -NTA column, washed with 50 mM Tris, 300 mM NaCl, 12.5 mM imidazole, 2 mM βmercaptoethanol, 0.1% triton X-100, and the protein was eluted with 50 mM Tris, 300 mM NaCl, 250 mM imidazole and 2 mM β-mercaptoethanol. Imidazole was removed by overnight dialysis using the Tris buffer (50 mM Tris, 300 mM NaCl, 2 mM β-mercaptoethanol), full-length proteins were concentrated by centrifugation, rebuffered in HEPES buffer (25 mM HEPES pH 7.5, 100 mM NaCl, 10 mM MgCl 2 and 0.5 mM TCEF) and stored in −80°C freezer. ΔC proteins were dialyzed with TEV protease to remove the expression tag and purified again on Ni 2+ -NTA column, followed by size-exclusion chromatography (HiLoad 16/600 Superdex 75 pg column) in HEPES buffer (20 mM HEPES, 150 mM NaCl, 2 mM DTT). ΔC proteins were concentrated and rebuffered for −80°C storage. All proteins were analyzed by ESI QTOF-MS.

Direct titration assay
All assays were performed in 384 well plates (each experiment was run in triplicates) and data were collected by Per- Collected data were processed in Microsoft Excel using the equations below. Total intensity I and anisotropy r were calculated using Equations 1 and 2 for each well. Average anisotropy was plotted against protein concentration using OriginPro and a logistic curve was fitted to give r min and r max . Using Equation 3 anisotropy was converted into fraction bound and multiplied by peptide concentration to be fitted in Origin using Equation 5 to obtain K d values.

Surface plasmon resonance
Experiments were performed at 25°C using a Biacore T200 instrument (GE Healthcare, Buckinghamshire, UK). For immobilization, the running buffer was HBS buffer (HEPES, NaCl, Tween-20, pH 7.4). 14-3-3η was immobilized on an NTA sensor chip (Series S Sensor Chip GE Healthcare) using NTA reagent and Biacore Amine Coupling Kit. Briefly, 14-3-3η was diluted to 0.2 mgÁmL −1 and 200 µL was injected over a chip surface that had been activated with an injection of Ni 2+ , followed by 140 µL of 1 : 1 NHS/EDC. A 30 µL of 0.5 M ethanolamine was then injected to cap the excess free amine groups. Immobilization levels of the 14-3-3η were found to be around 2000 response units (RU) for kinetic measurements. hDMX and hDM2 peptides were serially diluted 11 times from concentrations 10 times the K d values measured in FA, on a 96-well plate. Then, 20 µL of these solutions were injected over a 14-3-3η immobilized surface for 1 min at the 20 mLÁmin −1 flow rate, followed by a 4 min regeneration period with HBS buffer. Kinetics and the binding affinity were calculated using the BIACORE evaluation software (Buckinghamshire, UK). Standard dose-response curves were fitted by plotting RUs against the peptide concentration. structure determination by molecular replacement with 4DAT (PDB accession code)as a template for the 14-3-3σ structure.

Isothermal titration calorimetry
Further rounds of manual model building and refinement were performed using COOT [67] and REFMAC [68] respectively. Data collection and refinement statistics for each structure are provided in Table 5.

Analytical ultracentrifugation
AUC-SV experiments were performed using a Beckman Coulter Optima XL-I ultracentrifuge. Samples were prepared in HBS buffer (

Circular dichroism spectroscopy
Spectra were recorded on a Chirascan circular dichroism spectropolarimeter (Applied Photophysics, Leatherhead, UK), using 1 mm cells, scan speed of 5 nmÁmin −1 , 2 nm bandwidth and 180-260 nM range. The experiments were performed in 50 mM sodium phosphate buffer, pH 7.5 at 20°C. The spectra were averaged over three repeats with