Allosteric changes in HDM2 by the ATM phosphomimetic S395D mutation: implications on HDM2 function

Intrinsically disordered regions (IDRs) in multi-domain proteins provide large flexibility into the molecule, contributing in this way to include in the same primary sequence, a variety of functions. This is one of the manners that evolution has optimised the genome functionality and diversity. Different conformations are affected allosterically via post-translational modifications, mutations or by the interaction with proteins or other macromolecules like DNA/RNA [1–6]. One example of this concept is the human proto-oncoprotein HDM2 (MDM2 in mouse). HDM2 is a hub in a complex network of interactions, with more than 100 interacting partners [7]. The focus of HDM2 studies has been on its central role in controlling the levels and activity of the p53 tumour suppressor. Mouse models have shown that the MDM2-null mouse develops early embryonic lethality that can be rescued by additional p53 deficiency [8]. HDM2 down-regulates the activity of p53 by acting as an E3 ubiquitin ligase of p53, promoting p53 proteasomal degradation [9,10] or by masking its transactivation domain [11] to control the nuclear export of p53 [12].

The ATM kinase (ataxia telangiectasia mutated kinase) is activated during DNA damage and is a key factor in p53 activation by direct or indirect phosphorylation of p53 N-terminal residues Ser15 and 20 (S15 and S20) [13,14]. However, mouse models show that these events are not sufficient for full activation of the p53-dependent DNA damage response [15–17]. Interestingly, HDM2 is also phosphorylated by ATM after DNA damage [18] on Ser395 (394 in mouse MDM2), which enables p53 stabilisation [19]. The importance of this phosphorylation site is underlined by the MDM2 S394A mice which displayed a defect in p53 activation [20]. The phosphorylation of HDM2 Ser395 is mimicked by the phosphomimetic mutant S395D and results in the binding of the C-terminal RING domain to the p53 mRNA and an enhanced p53 translation [21–23]. The HDM2–p53 mRNA interaction controls the E3 ubiquitin ligase substrate specificity and selectively prevents ubiquitination of p53 while not affecting ubiquitination of itself or of its homologue HDMX [23].

Examples of how post-translational modifications or the binding of macromolecules in multi-domain proteins induce allosteric changes have been reported. For example, the CK2 phosphorylation in Ser392 stabilises the tetramerisation domain of p53, promoting its oligomerisation [24]. The phosphomimetic mutant S392E yields a more thermostable protein with a change in conformation [25,26]. More recently, it has been shown by Raman spectroscopy how mutations in the p53 protein change its conformation [27]. An example of how small molecules induce allosteric changes in HDM2 was shown by Wallace et al. in 2006, who showed that a short peptides of the p53 box I domain, or the small molecule nutlin, provoked a conformational change that promoted a new interphase between MDM2's acidic domain and the p53 core domain [28]. It has also been reported that the phosphomimetic mutation S17D in the lid of MDM2 yields a more thermostable domain [25] and provokes an allosteric change that stimulates the interaction between the acidic domain and the DNA-binding domain of p53 [26].

In the present work, we used hydrogen–deuterium exchange mass spectrometry (HDX MS), limited proteolysis in vitro ubiquitination assay and protein modelling to demonstrate the impact of the phosphorylation site Ser395 on HDM2 structure and the functional repercussion on its E3 ubiquitin ligase activity. Here we show that HDM2 can adopt different conformations induced by post-translational modifications regulating in this way, the function of the protein. By dissecting these conformations, we can propose new interfaces for drug development that specifically target HDM2 activities during different cellular conditions.

Sample preparation: HDM2 proteins either wild type or mutant S395D (final concentration 1 µM in a sample loaded into LC–MS/MS) were diluted either with 50 mM of Tris–base and 150 mM of NaCl buffer in H₂O (pH 7.5) used for the preparation of the un-deuterated control and for the peptide mapping or with 50 mM of Tris–base and 150 mM of NaCl buffer in D₂O (pD 7.1) used for the preparation of the deuterated samples. The HDX was carried out at room temperature (RT) and was quenched at 10 and 45 min by the addition of 0.5 M of TCEP, 0.875 M of HCl in 1 M of glycine and 1 mg/ml of pepsin at a ratio of 10:3:5 (v/v/v), followed with 3 min of incubation at RT and rapid freezing in liquid nitrogen.

Digestion and HPLC separation: Each sample was thawed and injected into an LC-system (UltiMate 3000 RSLCnano, Thermo Scientific Dionex, Massachusetts, USA) to an immobilised nepenthesin enzymatic column (15 µl bed volume, flow rate of 20 µl/min, 2% acetonitrile/0.05% trifluoroacetic acid). Peptides were trapped and desalted on-line on a peptide microtrap (Michrom Bioresources, Auburn, CA, U.S.A.) for 3 min at a flow rate of 20 µl/min. Next, the peptides were eluted onto an analytical column (Jupiter C18, 1.0 × 50 mm, 5 µm, 300 Å, Phenomenex, CA, U.S.A.) and separated using a 2 min linear gradient elution of 10–40% B buffer in A buffer, followed by 31 min isocratic elution at 40% B. The solvents used were: buffer A — 0.1% formic acid in water and buffer B — 80% acetonitrile/0.08% formic acid. The immobilised nepenthesin column, trap cartridge and the analytical column were kept at 1°C.

Mass spectrometry and data analysis: Mass spectrometric analysis was carried out using an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Massachusetts, USA) with ESI ionisation on-line connected with a robotic system based on the HTS-XT platform (CTC Analytics, Zwingen, Switzerland). The instrument was operated in a data-dependent mode for peptide mapping (HPLC–MS/MS). Each MS scan was followed by MS/MS scans of the top three most intensive ions from both the CID and HCD fragmentation spectra. Tandem mass spectra were searched using SequestHT against the cRap protein database (ftp://ftp.thegpm.org/fasta/cRAP), containing the sequence of both HDM2 proteins (wild type and mutant HDM2-S395D) with the following search settings: mass tolerance for precursor ions of 10 ppm, mass tolerance for fragment ions of 0.6 Da, no enzyme specificity, two maximum missed cleavage sites and no-fixed or variable modifications were applied. The false discovery rate at the peptide identification level was set at 1%. The sequence coverage was analysed with the Proteome Discoverer software version 1.4 (Thermo Fisher Scientific, Massachusetts, USA) and was graphically visualised with MS Tools application (http://ms.biomed.cas.cz/MSTools/) [29]. Analysis of the deuterated samples was done in the LC–MS mode with ion detection in the orbital ion trap. The MS raw files together with the list of peptides (peptide pool) identified with high confidence characterised by the requested parameters (amino acid sequence of each peptide, its retention time, XCorr and ion charge) were processed using HDExaminer version 2.2 (Sierra Analytics, Modesto, CA, U.S.A.). The software analysed the peptides behaviour and created the uptake plots that showed peptide deuteration over time with a calculated confidence level (high and medium confidence is accepted, low confidence is rejected). The results from the peptide pool were displayed as a graph showing the evolution of deuteration at individual parts of the protein at the same time, and different protein states plotted using the GraphPad Prism version 5.03 for Windows (GraphPad Software, San Diego, CA, U.S.A.).

Histidine-tagged recombinant proteins, HDM2, HDM2-S395D, HDM2(322–491) and HDM2(322–491)-S395D, were produced in E. coli BL21 (DE3) cells cultured at 37°C and purified using HiTrap nickel columns (GE Healthcare, Chicago, IL, USA). Lysis and buffer A (50 mM of Tris–base 200 mM of NaCl, 40 mM of imidazole, 10 μM of ZnSO₄, 10% of glycerol and were complemented with 10 mg of phenylmethanesulfonyl fluoride (PMSF, Sigma, Merck, Darmstadt, Germany). The pellet was resuspended in 10 ml of urea buffer (8 M urea, 50 mM of Tris–base, 150 mM of NaCl, 10 mM of imidazole, pH 8.5) and incubated for 2 h at RT. After centrifugation, the supernatant was loaded into the nickel column and washed with the following buffers; 25 ml of urea buffer, 25 ml of urea buffer plus 0.2% Triton, 25 ml of buffer A plus 10 mM of β-mercaptoethanol. Recombinant proteins were eluted from nickel columns with buffer B (50 mM of Tris–base, 200 mM of NaCl, 10 µM of ZnSO₄, 300 mM of imidazole, pH 8.5). Proteins were dialysed against phosphate buffer with 10 μM of ZnSO₄ and concentrated in centrifugal filter units Amicon (Merck Millipore, Darmstadt, Germany). Histidine-tagged recombinant p53 was induced at 15°C ON and purified on a HiTrap nickel column from the soluble fraction. Proteins were quantified and analysed in SDS–PAGE.

Thirty micrograms of HDM2(322–491) and HDM2(322–491)-S395D were incubated with 0–100 ng of proteinase K (Thermo Fisher Scientific, Carlsbad, CA, USA) at 30°C for 2 or 5 min. The reaction was arrested by the addition of PMSF (final concentration 3 mM). The samples were analysed by SDS–PAGE.

Ubiquitination reactions contained 25 mM of HEPES pH 8.0, 10 mM of MgCl₂, 4 mM of ATP, 0.5 mM of DTT, 0.05% (v/v) of Triton X-100, 0.25 mM of benzamidine, 10 mM of creatine phosphate, 3.5 units/ml of creatine kinase, ubiquitin or His-tagged ubiquitin (2 μg) and E1 (100 nM), E2 (1 μM) and p53 at 5 μg; the reactions were started with the addition of HDM2 or HDM2-S395D, incubated for 15 min at 30°C and analysed with 4–12% SDS–PAGE followed by immunoblotting with CM1 antibody (rabbit polyclonal sera to human p53 protein).

It has been shown that during genotoxic stress, HDM2 is phosphorylated on Ser395, which switches the activity of HDM2 from an E3 ubiquitin ligase that acts as a negative regulator of p53 to an mRNA translation factor stimulating p53 synthesis [22,23]. To gain understanding into the structural mechanism behind the functional versatility of HDM2, we first studied the impact of the S395D phosphomimetic mutation on the C-terminal domain of (322–491) of HDM2 because this region comprises the site for p53 mRNA binding but also contains the E3 ubiquitin ligase activity (Supplementary Figure S1).

We produced the recombinant C-terminal domain HDM2(322–491) wild type and HDM2(322–491)-S395D phosphomimetic mutant in E. coli. Using HDX MS, we studied the structural changes in the proteins imposed by the phosphomimetic mutation S395D. This is a powerful technique based on the solvent accessibility of the protein surface, measured as the replacement of the amide hydrogen by the deuterium atom on the amino acid [30,31]. The differences in the exchange reaction between the wild type and the phosphomimetic mutant enable the study of the conformational flexibility. Firstly, the peptide mapping approach for both constructs was applied to create the library of peptides which covered the whole amino acid sequence. Either the wild type or the phosphomimetic mutant (final concentration of 1 μM) was diluted with a buffer in H₂O and used as a non-deuterated control. The samples were further processed by acidification and proteolytic digestion with pepsin and nepenthesin which was then measured using liquid chromatography connected with tandem MS (LC–MS/MS) to determine the amino acid sequences of created peptides under conditions used for HDX MS. We obtained 161 unique peptides of HDM2(322–491) wild type that covered 94% of its amino acid sequence in comparison with HDM2(322–491)-S395D when 171 unique peptides were identified with 100% of coverage (Supplementary Figure S2). The deuterium exchange of each construct (final concentration of 1 μM) was initiated by dilution with a buffer in D₂O carried out at RT. Deuteration was quenched at different point-times by the addition of quench buffer and then, rapid freezing in liquid nitrogen. After MS, the data were processed using HDExaminer software to evaluate changes in the deuteration of the peptides from HDM2(322–491) or HDM2(322–491)-S395D as a function of time. The data show a change in the profile of deuteration all along the protein construct and not just in the site of phosphorylation. HDM2(322–491) exhibited elevated HDX at 30 s and 10 min in comparison with HDM2(322–491)-S395D (Figure 1 and Supplementary Figure S3).

The conformational change in the structure of HDM2(322–491)-S395D was also observed using limited proteolysis by proteinases K [32]. The wild type and the phosphomimetic constructs were treated with concentrations ranging from 1 to 100 ng of proteinase K for 2 or 5 min at 30°C (Figure 2 and Supplementary Figure S4). The phosphomimetic mutant shows a higher susceptibility to proteolysis than the wild type construct, indicating a global change in the protein that exposes different cleavage sites. These results are in concordance with the HDX MS data that show a higher amount of unique peptides obtained with the phosphomimetic mutant after proteolytic digestion. Using the free modelling method QUARK [33,34], we built a model of the HDM2(322–491) and the phosphomimetic mutant HDM2(322–491)-S395D. The results revealed that this region of HDM2 has remarkable conformational plasticity and how this feature can be regulated by small changes such as post-translational modification. HDM2(322–491) shows a more globular structure where the RING domain (432–491 aa, in purple) is hidden in the core of the protein by the IDRs (322–431 aa). The Ser395 is in close proximity to the amino acids 464–471 in the α-helix of the RING domain and the model suggests that the phosphomimetic mutation opens the structure of the domain (Figure 3A,B). The surface exposed to the solvent increases almost three times from 419.18 Ų for the HDM2(322–491) until 1128.9 Ų for HDM2(322–491)-S395D, indicating that different surfaces are exposed in the phosphomimetic mutant (Figure 3A,B, right panels). Together, these data help to explain how the phosphorylation on Ser395 alters HDM2 conformation results in the RING domain interacting with the p53 mRNA during the DNA damage response.

Multi-domain proteins often have IDRs that serve as a bridge between properly folded domains [35]. These IDRs are frequently the site of post-translational modifications that can regulate the conformation of the whole protein. To test this hypothesis, we produced full-length HDM2 and HDM2-S395D in a recombinant system and purified them (Supplemental Figure S5). The produced proteins were used in HDX MS to analyse the structural changes in the overall proteins due to the phosphomimetic mutation. Each protein at 1 μM was incubated in D₂O buffer under the same conditions as the C-terminal constructs. The results show that the small change in position 395 near to the RING domain in the C-terminus selectively disturbs two regions of the protein, without affecting the overall profile of the deuteration of the protein. The site next to the phosphorylation event occurs but also the hydrophobic pocket in the N-terminal part of the protein (1–100 aa) (Figure 4A,B and Supplementary Figure S6). The hydrophobic pocket of the HDM2 wild type was suppressed over time whilst part of the RING domain was more deuterated in comparison with HDM2-S395D. Additionally, the disorder tendency variations were performed with the HDM2 wild type and phosphomimetic mutant HDM2-S395D using PrDOS [36] which calculates the disorder probability of the protein. PrDOS characterises the disordered tendency of each amino acid along the protein sequence. The disorder probability can take a value between 0 and 1. The larger disorder probability, the higher disorder tendency of the amino acid is. Residues with a disorder probability above 0.5 are considered disordered, while residues with lower disorder probability considered to be ordered (Figure 4A,C). The data also show changes in the same two regions; the hydrophobic pocket and the Ser395 phosphorylation site near to the RING domain, confirming that there is no change in the overall structure, matching with the HDX MS results.

The functional implications of the phosphomimetic modification were observed by analysing the ubiquitination profile of p53. HDM2's capacity to interact and ubiquitinates p53 protein under normal cellular conditions has been well characterised in vitro [23,37,38]. It has been shown that the phosphorylation of Ser395 or the phosphomimetic S395D mutation has the same affinity towards p53 protein [23] and interestingly, that despite a similar affinity, it promotes the E3 ligase activity towards p53. It is only in the presence of the p53 mRNA that HDM2 no longer is capable of ubiquitinating p53 while it retains the capacity to ubiquitinate its homologous protein HDMX and itself [23]. We compared the p53 ubiquitination profile by HDM2 wild type and HDM2-S395D. The data confirm that the HDM2-S395D phosphomimetic mutant is a better ubiquitin ligase toward p53 and increases the amount of p53 ubiquitinated and the molecular mass of the p53–ubiquitin forms. This suggests that the plasticity of the protein and the domain organisation, rather than the affinity towards p53, are controlled by post-translational modifications and governs HDM2 E3 ligase activity towards p53, presumably by placing the E2 ligase in proximity to p53 lysine residues (Figure 5).

The presented data support other studies showing that HDM2's interactions with its partner proteins are governed by a plasticity that is regulated by post-translation modification. Such events act together with protein–protein interactions to induce conformations that promote specific complex formations and HDM2-regulated functions. Besides, it is likely that allosteric change in one partner of such complex acts together with allosteric changes in the other and, thus, that inter allosteric changes take place between partners in the protein complexes in a reciprocal induced fit. For example, the HDM2 paralogous protein HDMX interacts with the N-terminal p53 binding domain and this first interaction promotes a secondary allosteric interaction between the p53 DNA-binding domain and the acidic domain of HDMX [39]. Similarly, ligands binding the hydrophobic N-terminal pocket like p53, p53-derived short peptides or small molecules promote allosteric changes in HDM2 that allow a secondary interaction between the acidic domain of HDM2 and the DNA-binding domain of p53 [28,40–42]. Additionally, another allosteric rearrangement has been reported involving the interaction of p53 in the N-terminal domain of HDM2 that enhances the HDM2 and HDMX interaction [38].

The deuteration profile of the C-terminal domain of HDM2 shows that the phosphomimetic mutant S395D promotes changes throughout the protein construct and exposes new surfaces. In comparison, the same mimetic change in the full-length HDM2 shows specific changes in two different domains. This likely reflects communication between the N- and C-terminal domains that together with the notion of interchangeable allosteric changes, emphasises that multi-domain proteins with intrinsically disordered domains are more than the sum of their parts (Supplementary Figure S7). The C-terminal constructs carrying the phosphomimetic mutant is more susceptible to limited proteolysis as compared with the wild type. However, Medina-Medina et al., in 2016, showed that the phosphomimetic full-length proteins instead protect HDM2 from proteolytic attacks [23]. This is in line with the idea that more N-terminal domains play a role in the structural rearrangement after phosphorylation. The in silico model of HDM2(322–491) in Figure 3 suggests that the phosphomimetic S395D results in the exposure of a larger surface to solvent and it is conceivable that this would also result in increased proteolysis. The peptide mapping of the C-terminal construct further supports the notion that the phosphomimetic mutation opens the C-terminus. However, the HDX experiments show a decrease in the deuteration profile in the mutant (Figure 1). This could instead suggest that the protein is involved in another type of regulation. HDM2 has the ability to make oligomers and it is possible that the mutant opens up new surfaces of interaction that produce oligomers that can affect the degree of deuteration. One has to bear in mind that these are different assays and it illustrates the importance of a multi-assay approach and it will be interesting to see further structural studies to learn how to best interpret results using different assays.

HDM2 is overexpressed in ∼10% of human cancers, with large variations depending on cancer type, and acts as an oncoprotein when inactivating p53 [43–47]. However, under certain conditions such as ATM activation following DNA damage, HDM2 becomes a positive regulator of p53 protein [22]. The results presented here indicate that HDM2 following ATM-mediated phosphorylation on Ser395 is a better E3 ubiquitin ligase for p53 in the absence of p53 mRNA. In a scenario when HDM2 is amplified and overexpressed in tumour cells, the constitutive activation of ATM due to genomic instability and the imbalance in the HDM2:p53 mRNA ratio would result in no p53 mRNA-bound HDM2 and an increase in p53 degradation and enhanced proliferation and tumour development.

HDM2 is considering as a good target for anti-cancer drug development programmes and several companies have developed drugs that bind the N-terminal hydrophobic pocket to prevent p53 interaction. However, this approach has been shown toxic in the clinic, which has resulted in a need to re-think how HDM2 can best be targeted to minimise the risk of undesirable effects. The results presented here highlight that cell condition-dependent interphases that are generated will help to identify compounds that more specifically target certain conformation of HDM2 and deliver more specific cell biological effects for personalised medicine.

 
• ATM kinase

  ataxia telangiectasia mutated kinase

 
• HDX

  hydrogen–deuterium exchange

 
• IDRs

  intrinsically disordered regions

 
• MS

  mass spectrometry

 
• PMSF

  phenylmethanesulfonyl fluoride

 
• RT

  room temperature

L.U. and Lu.H. performed HDX MS experiments and analysed the data; Le.H. designed HDX MS experiments and analysed the data; L.W. performed the models and the IDR analysis; I.M.-M., S.C. and Y.R.-G. performed the experiments of proteolysis and ubiquitination and analysed the data; R.F. and B.V. supervised the project and analysed the data; V.O.-I. designed experiments, analysed the data and wrote the MS. All authors edited the MS.

This work was supported by Conacyt CB-256637 by projects MEYS — NPS I — LO1413 and MH CZ — DRO (MMCI, 00209805) and ECOS-Nord Program Cooperation Mexico-France No 249553. Cátedra Marcos Moshinsky 2017. B.V. was partially supported by the Grant Agency of the Czech Republic [Grant No. 19-03796S]. Cancerforskningsfonden Norr; Cancerfonden [180296] and Vetenskapsradet.

The Authors declare that there are no competing interests associated with the manuscript.