ABSTRACT
The RING domain ubiquitin E3 ligase MDM2 is a key regulator of p53 degradation and a mediator of signals that stabilize p53. The current understanding of the mechanisms by which MDM2 posttranslational modifications and protein binding cause p53 stabilization remains incomplete. Here we present evidence that the MDM2 central acidic region is critical for activating RING domain E3 ligase activity. A 30-amino-acid minimal region of the acidic domain binds to the RING domain through intramolecular interactions and stimulates the catalytic function of the RING domain in promoting ubiquitin release from charged E2. The minimal activation sequence is also the binding site for the ARF tumor suppressor, which inhibits ubiquitination of p53. The acidic domain-RING domain intramolecular interaction is modulated by ATM-mediated phosphorylation near the RING domain or by binding of ARF. These results suggest that MDM2 phosphorylation and association with protein regulators share a mechanism in inhibiting the E3 ligase function and stabilizing p53 and suggest that targeting the MDM2 autoactivation mechanism may be useful for therapeutic modulation of p53 levels.
INTRODUCTION
A unique feature of the p53 tumor suppressor is its stabilization after exposure to many stress signals. This leads to the induction of numerous transcriptional targets that inhibit cell cycle progression, induce apoptosis, and regulate energy metabolism (1). The MDM2 and MDMX proteins are responsible for establishing the dynamic features of the p53 pathway. MDM2 is a RING domain ubiquitin (Ub) E3 ligase for p53 that promotes p53 degradation (2, 3). Mouse models provided strong evidence that MDM2 is indispensable for controlling p53 activity at all stages of life (4–6). The stabilization of p53 by small-molecule inhibitors that disrupt p53-MDM2 binding also confirmed that MDM2 is a major regulator of p53 turnover (7, 8). MDM2-p53 disruptors have antitumor activity in animal models, and their potential as cancer drugs is currently being tested in the clinic (9).
Numerous stress signals have been shown to cause p53 accumulation, mainly by inhibiting its degradation. MDM2 promotes p53 degradation by forming a stable complex through N-terminal domains. The MDM2 C-terminal RING domain recruits ubiquitin-conjugating enzyme E2, which performs a covalent modification of p53 lysine residues (10, 11). The major E2 isoforms involved in MDM2-mediated p53 ubiquitination in cells belong to the UbcH5 family (12). The MDM2-UbcH5 combination promotes the synthesis of mainly K48-linked polyubiquitin chains on p53 that target p53 for degradation by the 26S proteasome. MDM2-mediated ubiquitination of p53 is inhibited by multiple mechanisms. Phosphorylation of the p53 N terminus after DNA damage reduces MDM2 binding and contributes to p53 activation (13, 14). DNA damage also induces ATM-dependent phosphorylation of MDM2, which inhibits RING domain dimerization and p53 polyubiquitination (15–17). Oncogene activation induces the expression of ARF, which binds to MDM2 and inhibits p53 ubiquitination (18). Inhibition of nucleolar ribosomal DNA (rDNA) transcription promotes the release of ribosomal protein L11, which also binds to MDM2 and stabilizes p53 (19, 20).
Ubiquitin E3 ligases bind specifically to substrates, recruit ubiquitin-charged E2 to the substrate, and stimulate the transfer of activated ubiquitin from E2 to lysine residues on the substrate (21). The E2 active-site conformation and spatial proximity to the substrate are important for efficient ubiquitin transfer and chain elongation (22–24). Each step in ubiquitination can be regulated by posttranslational modifications or protein-protein interactions. As expected, the p53-binding domain and RING domain of MDM2 are both essential for p53 degradation. However, the central acidic domain (AD) of MDM2 (residues 220 to 300) is also critical for ubiquitination of p53 (25, 26). The acidic domain has features of a partially unstructured region and contains the binding sites for many MDM2-binding proteins, including chromatin-modifying proteins (p300, YY1, KAP1, SUV39H1, and EHMT1, etc.) (27–29), the deubiquitinating enzyme HAUSP (30), ribosomal proteins (19), and the tumor suppressor ARF (31). Furthermore, the MDM2 acidic domain can bind weakly to the p53 core domain and induces p53 conformational change (32–36). The flexibility of the acidic domain is probably critical for interactions with multiple protein partners (37, 38).
The central region of MDM2 also undergoes constitutive phosphorylation on multiple serine residues that are downregulated by DNA damage (39). Glycogen synthase kinase 3 (GSK3) and casein kinase Iδ (CK1δ) have been shown to modify these sites (40–42). Downregulation of GSK3 by DNA damage may explain the reduction in acidic domain phosphorylation levels (41). Alanine substitutions of some MDM2 acidic domain phosphorylation sites significantly inhibit degradation of p53. A recent study suggests that the acidic domain phosphorylation sites regulate MDM2 interactions with the 19S proteasome regulatory subunit, which mediates delivery of ubiquitinated p53 to the proteasome (43).
In this report, we investigated the mechanism by which the MDM2 acidic domain promotes p53 ubiquitination. Our results showed that the acidic domain functions as an activator of the RING domain through intramolecular interactions. The acidic domain stimulates the binding of the RING domain to a ubiquitin ∼E2 conjugate and promotes the release of ubiquitin from E2. The results suggest that the MDM2 RING domain alone has low catalytic activity and requires binding to the acidic domain to become fully active. Importantly, our findings suggest a common mechanism by which multiple signaling pathways induce p53 stabilization and reveal potential new targets for therapeutic modulation of MDM2 E3 ligase activity.
MATERIALS AND METHODS
Plasmids and cell lines.All MDM2 constructs used in this study were human cDNA clones. Point mutants of MDM2 and UbcH5C were generated by site-directed mutagenesis using the QuikChange kit (Stratagene). The pET-mE1 construct encoding His6-tagged mouse ubiquitin-activating enzyme E1 was purchased from Addgene. MDM2 deletion constructs for expression in mammalian cell culture were generated by multiple steps of PCR cloning and inserted into the pCMV-neo-Bam vector. MDM2 constructs for expression in Escherichia coli were generated by subcloning MDM2 cDNA sequences into the pGEX-6P-1 vector (GE Healthcare Life Sciences), which allows removal of the glutathione S-transferase (GST) tag by cleavage with PreScission protease. The NARF6 cell line (U2OS cell line expressing isopropyl 1-thio-β-d-galactopyranoside [IPTG]-inducible p14ARF) was provided by Dawn Quelle.
Protein analysis.To detect proteins by Western blotting, cells were lysed in lysis buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.5% NP-40, 1 mM phenylmethylsulfonyl fluoride [PMSF], 50 mM NaF) and centrifuged for 5 min at 10,000 × g. The cell lysate (10 to 50 μg protein) was fractionated by SDS-PAGE and transferred onto Immobilon P filters (Millipore). The filter was blocked for 1 h with phosphate-buffered saline (PBS) containing 5% nonfat dry milk and 0.1% Tween 20 and incubated with primary antibodies. The filter was developed by using SuperSignal West Pico chemiluminescent substrate (Thermo Scientific). Monoclonal antibodies 3G9, 4B2, and 4B11 were used for the detection of different MDM2 fragments (44). Antibodies DO-1 and Pab1801 were used for the detection of p53.
Protein expression and purification.MDM2 expression plasmids were transformed into BL21(DE3) cells (Aligent Technologies), cultured to an optical density at 600 nm (OD600) of 0.6 at 37°C, supplemented with 0.1 mM IPTG and 150 μM ZnCl2, and then grown for 20 h at 16°C. Proteins were extracted with lysis buffer (50 mM HEPES [pH 7.5], 150 mM NaCl, 10 μM ZnCl2, and 1 mM dithiothreitol [DTT]) and purified by using a GSTrap FF 1-ml column (GE Healthcare Life Sciences), followed by cleavage with PreScission protease at 4°C for 18 h. The proteins were further purified by chromatography with a GSTrap FF 1-ml column to remove GST and uncleaved fusion proteins. For purification of His6-E1 and His6-UbcH5c, the cells containing E1 or UbcH5C expression plasmids were cultured as described above and induced with 1 mM IPTG. The proteins were extracted in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole [pH 8.0]) and purified by using a HiTrap chelating HP 1-ml column charged with nickel (Ni2+).
In vivo ubiquitination assay.H1299 cells in 10-cm plates were transfected with 5 μg His6-ubiquitin, 1 to 5 μg of MDM2, and 1 μg of p53 expression plasmids by using a calcium phosphate precipitation method. Thirty-two hours after transfection, cells were treated with 50 nM Velcade for 4 h. The cells from each plate were collected into two aliquots. One aliquot (10%) was used for direct Western blotting. The remaining cells (90%) were used for purification of His6-tagged proteins by using Ni2+-nitrilotriacetic acid (Ni2+-NTA) beads. The cell pellet was lysed in buffer A (6 M guanidinium-HCl, 0.1 M Na2HPO4-NaH2PO4, 0.01 M Tris-HCl [pH 8.0], 5 mM imidazole, 10 mM β-mercaptoethanol) and incubated with Ni2+-NTA (Qiagen) for 4 h at room temperature. The beads were washed with buffer A, buffer B (8 M urea, 0.1 M Na2PO4-NaH2PO4, 0.01 M Tris-HCl [pH 8.0], 10 mM β-mercaptoethanol), and buffer C (8 M urea, 0.1 M Na2PO4-NaH2PO4, 0.01 M Tris-HCl [pH 6.3], 10 mM β-mercaptoethanol), and bound proteins were eluted with buffer D (200 mM imidazole, 0.15 M Tris-HCl [pH 6.7], 30% glycerol, 0.72 M β-mercaptoethanol, 5% SDS). The eluted proteins were analyzed by Western blotting for the presence of ubiquitin-conjugated p53.
Ubiquitin release assay.UbcH5c charged with ubiquitin was produced in a reaction mixture (20 μl) containing 0.2 μg His6-E1, 0.2 μg His6-UbcH5c, and 2 μg ubiquitin (Biomol) in a buffer containing 10 mM HEPES (pH 7.5), 100 mM NaCl, 40 μM ATP, and 2 mM MaCl2. The reaction mixture was incubated for 20 min at room temperature, and 0.04 unit/μl apyrase (Sigma) was added and incubated for 10 min at room temperature to deplete the ATP. MDM2 RING proteins (0.25 μM) were added to the reaction mixtures. Samples were incubated for 0 to 45 min at room temperature, and the ubiquitin release reaction was terminated by adding nonreducing Laemmli sample buffer. Boiled samples were fractionated by SDS-PAGE and blotted with anti-UbcH5 antibody (Boston Biochem).
E2-ubiquitin oxyester formation and pulldown assay.The method used for E2-ubiquitin oxyester formation and pulldown was modified from a recently reported protocol (45). The UbcH5c∼ubiquitin oxyester formation reaction mixture (100 μl) contained 1 μg His6-E1, 1 μg His6-UbcH5c (with N77A and C85S substitutions), and 5 μg Myc-ubiquitin (Boston Biochem) in reaction buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 3 mM ATP, 5 mM MgCl2, 1 mM DTT). The mixture was incubated at room temperature for 18 h, and the charging reaction was terminated by adding 0.04 unit/μl apyrase to the mixture. Glutathione-agarose beads (15-μl packed volume) loaded with ∼0.5 μg GST-MDM2-RING or GST-minimal acidic domain-RING (GST-mAD-RING) were incubated with 100 μl of the UbcH5c∼ubiquitin oxyester reaction mixture for 1 h at room temperature. The beads were collected and suspended in 0.5 ml of binding buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM DTT, 5% glycerol) and overlaid on top of 1.4 ml of a solution containing 50 mM Tris-Cl (pH 7.5), 150 mM NaCl, 1 mM DTT, 5% (vol/vol) glycerol, and 10% (wt/vol) sucrose. The beads were quickly collected by centrifugation, and the washing step was repeated once for the detection of binding to UbcH5c∼ubiquitin oxyester by anti-Myc Western blotting. To detect binding to uncharged UbcH5c in the reaction mixture, the beads were further washed with radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% [vol/vol] Triton X-100, 0.1% SDS, and 1% sodium deoxycholate) and boiled in Laemmli sample buffer. The samples were fractionated by SDS-PAGE and blotted with anti-UbcH5 antibody.
RING fragment release assay.SJSA cells were treated with 10 Gy of gamma irradiation (IR) and 50 nM the proteasome inhibitor Velcade, collected at 2 h after IR treatment, and lysed in lysis buffer (50 mM Tris-HCl [pH 8.0], 5 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1 mM PMSF, 50 mM NaF). The cell lysate was immunoprecipitated with a 30-μl slurry of protein A-agarose beads and 4B2 antibody. The beads were incubated with 0.2 μg purified caspase-3 in 30 μl of cytosol buffer {10 mM HEPES (pH 7.5), 150 mM NaCl, 0.1% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 0.5 mM DTT, 2 mM EDTA} for 2 to 10 min at 37°C. The supernatant containing the released MDM2 RING domain was collected. The beads were washed 2 times with lysis buffer and boiled in Laemmli sample buffer. The samples were fractionated by SDS-PAGE and blotted with antibody 4B11 to determine the ratio of the RING domain released into the supernatant to that remained bound to the beads.
RESULTS
Mapping of a minimal MDM2 acidic region required for p53 ubiquitination.The C-terminal RING domain of MDM2 is critical for ubiquitination of p53 and interaction with E2. Previous studies showed that deletion of ∼100 residues in the partially disordered MDM2 central acidic region or its replacement with the corresponding region from MDMX abrogated the ability of MDM2 to promote p53 degradation (25, 26). These results suggested that the acidic region functions as an activating or targeting sequence. To further identify the minimal region important for p53 ubiquitination, a panel of MDM2 acidic domain internal deletion mutants (Fig. 1b) was analyzed for p53 ubiquitination after coexpression with p53 in H1299 cells. The results showed that deletion of residues 210 to 290 significantly reduced p53 ubiquitination efficiency (MDM2-ΔE) (Fig. 1a). The deficiency of smaller internal deletion mutants (MDM2-ΔB and -ΔC) suggested that the sequence from residues 230 to 270 was important for p53 ubiquitination.
Ubiquitination of p53 by MDM2 acidic domain deletion mutants. (a) p53 was cotransfected with MDM2 acidic domain deletion mutants into H1299 cells. Ubiquitination of p53 was detected by p53 IP followed by p53 Western blotting. (b) Diagrams of MDM2 internal deletion mutants used in panel a. Solid lines represent regions of the protein encoded by the mutants, and dotted lines represent deleted sequences. The bottom diagram shows MDM2 polypeptide folding potential using PONDR analysis, suggesting that the central acidic region is partially disordered.
To identify the minimal sequence sufficient for activating p53 ubiquitination by RING, additional MDM2 internal deletion mutants were constructed. The nuclear localization sequence (NLS) was transferred to the N terminus to ensure nuclear import. Internal deletion of other MDM2 regions not involved in p53 or E2 binding (Δ110-230/Δ325-410) did not reduce p53 ubiquitination efficiency (MDM2-ΔN) (Fig. 2a and e), suggesting that the acidic domain alone was sufficient to activate RING. It is noteworthy that all 3 mutants shown in Fig. 2a produced higher-molecular-weight (MW) p53 polyubiquitination products than did wild-type MDM2, at the expense of low-MW monoubiquitination products (for example, MDM2-ΔP). This is likely due to the removal of the ATM phosphorylation region near the RING domain that serves as inhibitory sequences. When the MDM2 sequence between the p53-binding domain and the RING domain was completely removed (MDM2-ΔM) (Fig. 2b and e), the MDM2-ΔM mutant lost most of the p53 polyubiquitination activity (Fig. 2b). When the MDM2 acidic region was reintroduced into the MDM2-ΔM background (MDM2-ΔK), the sequence spanning residues 230 to 260 was sufficient to restore p53 ubiquitination (Fig. 2b). MDM2-ΔK was also able to promote p53 degradation (Fig. 2d). Further removal of parts of the sequence spanning residues 230 to 260 led to a significant loss of activity (MDM2-ΔF/ΔG/ΔH/ΔI) (Fig. 2c), suggesting that the region spanning residues 230 to 260 is the minimal region needed to stimulate p53 ubiquitination. Therefore, the sequence spanning residues 230 to 260 was referred to as the minimal acidic domain (mAD).
Mapping of a minimal MDM2 acidic sequence required for p53 ubiquitination. (a to c) p53 was cotransfected with MDM2 deletion mutants and His6-ubiquitin into H1299 cells. p53 conjugated to His6-Ub was purified by using Ni2+-NTA beads under denaturing conditions and detected by Western blotting with DO-1. The bottom panels are control Western blots of whole-cell extracts to verify expression of p53 and MDM2. (d) p53 was cotransfected with MDM2 at ratios optimized for degradation. p53 degradation was analyzed by Western blotting using DO-1. (e) Diagrams of MDM2 mutants for p53 ubiquitination used in panels a to d and summary of the results. Solid lines represent regions of MDM2 encoded by the mutants, and dotted lines represent internal deletions. The minimal MDM2 acidic sequence required for p53 ubiquitination was mapped to residues 230 to 260 by using the MDM2-ΔK mutant.
The mAD stimulates RING activity.The MDM2 acidic domain may promote p53 degradation through several mechanisms, such as targeting p53 to the proteasome, providing a second binding site for the p53 core domain, and acting as a spacer for optimal orientation of the RING domain. The ubiquitination assay suggested that the mAD sequence promotes mainly p53 polyubiquitination. A linker function for mAD appeared unlikely because extensive unstructured regions of MDM2 were still present in several mutants defective for p53 ubiquitination. In an in vivo ubiquitination assay, MDM2-ΔK was able to ubiquitinate the Gal4–p53-1-82 fusion (Fig. 3a), suggesting that binding to the p53 core domain was not needed for mAD to stimulate ubiquitination. Furthermore, expression of the green fluorescent protein (GFP)-mAD fusion moderately stimulated p53 ubiquitination by MDM2-ΔM (Fig. 3b), suggesting that it was able to stimulate RING activity in trans.
Functional analysis of the minimal acidic sequence. (a) The Gal4-p53 or Gal4–p53-1-82 fusion was coexpressed with MDM2 deletion mutants and His6-ubiquitin in H1299 cells. The ubiquitination of the p53 fusion protein was detected by purification with Ni2+-NTA beads and Western blotting with DO-1. (b) p53 was cotransfected with the MDM2 acidic domain deletion mutant, GFP-AD fusion constructs, and His6-ubiquitin. p53 ubiquitination was detected by Ni2+-NTA pulldown and DO-1 blotting. GFP-AD constructs stimulated the MDM2 acidic domain deletion mutant in trans. (c) MDM2 acidic region point mutants were tested for p53 ubiquitination by cotransfection with p53 and His6-ubiquitin in H1299 cells. (d) MDM2 acidic region point mutants were tested for ARF binding by IP-Western blotting (WB) after coexpression with Myc-ARF in H1299 cells. The MDM2 IP was washed with RIPA buffer. WCE, whole-cell extract. (e) Alignment of the minimal activation region showing sequence conservation and locations of the point mutants.
Residues 230 to 260 of MDM2 are highly conserved compared to other regions of the acidic sequence despite a similar density of acidic residues (Fig. 3e). To test the roles of acidic and hydrophobic residues in the mAD region, MDM2 mutants containing 3 acidic-to-basic substitutions (MDM2-3R) or 8 hydrophobic-to-hydrophilic substitutions (MDM2-8GS) were analyzed (Fig. 3e). Both MDM2-3R and MDM2-8GS were severely deficient in p53 ubiquitination and degradation and partially deficient in self-ubiquitination (Fig. 3c) and MDMX ubiquitination (not shown), suggesting that both the negative charge and hydrophobic characteristics of the mAD are important for its E3-activating function. MDM2-3R was also partially deficient for binding to ARF (Fig. 3d), consistent with previously reported mapping results and the important role of charge in this interaction (31, 46). Further comparison of MDM2-3R (E248R/E250R/E252R) with the intermediate constructs MDM2-2R (E248R/E250R) and MDM2-1R (E248R) revealed the same level of deficiency in p53 ubiquitination (not shown). Therefore, the E248R substitution alone was sufficient to strongly reduce MDM2 E3 ligase activity.
MDM2 with extra copies of the acidic domain is hyperactive.To further confirm the activating function of the acidic domain, the MDM2-3AD construct containing 2 extra tandem copies of the acidic domain (AD) sequence (residues 221 to 280) was created (Fig. 4a). Compared to wild-type MDM2, MDM2-3AD binds to p53 with a similar efficiency (not shown) but had significantly higher activity in promoting p53 and MDMX degradation (Fig. 4b and c). MDM2-3AD was also more active in promoting p53 and MDMX ubiquitination (Fig. 4d and e). Expression of ARF strongly inhibited MDM2 ubiquitination of p53, whereas MDM2-3AD was less sensitive to inhibition by ARF (Fig. 4f). These results provide functional evidence that the mAD region is an activation domain of MDM2 RING E3 activity.
Hyperactive MDM2 containing extra copies of the acidic domain. (a) Diagram of the MDM2-3AD construct containing 2 extra copies of the acidic domain sequence spanning residues 221 to 280. ZF, zinc finger. (b) p53 degradation by MDM2-3AD was analyzed by coexpression in H1299 cells and Western blotting. (c) MDMX degradation by MDM2-3AD was analyzed by coexpression in H1299 cells and Western blotting. (d) p53 ubiquitination by MDM2-3AD was analyzed by coexpression with His6-ubiquitin in H1299 cells followed by Ni2+-NTA pulldown and DO-1 Western blotting. (e) MDMX ubiquitination by MDM2-3AD was analyzed by coexpression with His6-ubiquitin in H1299 cells followed by Ni2+-NTA pulldown and MDMX Western blotting. (f) MDM2-3AD ubiquitination of p53 in the presence of ARF was analyzed by coexpression with His6-ubiquitin in H1299 cells followed by Ni2+-NTA pulldown and DO-1 Western blotting.
The acidic domain stimulates RING domain catalytic activity.Previous studies showed that dimerization of the MDM2 RING domain was important for E3 activity (16). However, chemical cross-linking analysis did not detect a change in MDM2 oligomerization after deletion of the acidic domain (not shown). To test whether the mAD directly activates RING domain catalytic function, it was fused to RING (residues 410 to 491), similar to the MDM2-ΔK configuration (the region spanning residues 410 to 430 serves as a flexible linker), and expressed in E. coli. The purified RING domain and mAD-RING fusion (Fig. 5a) were analyzed for the ability to promote ubiquitin release from Ub∼S-UbcH5c (E2 charged with ubiquitin), which served as a measure of the catalytic efficiency of the RING domain. The results showed that MDM2 RING alone had weak ubiquitin release activity, whereas the mAD-RING fusion was >5-fold more efficient than RING (Fig. 5b).
The MDM2 acidic domain stimulates RING domain catalytic activity and binding to charged E2. (a) MDM2 RING (residues 410 to 491) and the mAD-RING fusion protein (residues 230 to 260 and 410 to 491, respectively) were expressed as GST fusions in E. coli, purified after removal of GST, and quantified by Coomassie staining. MW, molecular weight (in thousands). (b) UbcH5c∼ubiquitin thioester was incubated with MDM2 RING and mAD-RING for the indicated times. The samples were fractionated by nonreducing SDS-PAGE and blotted with UbcH5 antibody. (c) A mixture containing Myc-ubiquitin, UbcH5c, and UbcH5c∼Myc-ubiquitin oxyester was incubated with glutathione beads loaded with GST-RING or GST-mAD-RING. The beads were analyzed by Myc Western blotting to detect binding to UbcH5c∼Myc-ubiquitin, anti-UbcH5 Western blotting to detect binding to UbcH5c, and GST Western blotting to confirm RING protein levels.
To further determine the mechanism of the mAD, the RING and mAD-RING constructs were tested for the ability to bind UbcH5c. Increased binding to charged E2 by RING is a potential mechanism for high E3 activity. Because mAD-RING binding to Ub∼S-UbcH5c caused a rapid release of ubiquitin, a UbcH5c active-site mutant (N77A/C85S) was used to generate Ub∼O-UbcH5c for the analysis of binding efficiency (45). The oxyester linkage serves as a structural mimic of thioester, but the ubiquitin cannot be released after binding to E3. As expected from transient E3-E2 interactions, the binding between GST-RING and Ub∼O-UbcH5c did not survive repeated washing in a conventional pulldown assay (not shown). However, the interaction was detectable after rapid sedimentation of the beads and complex through a sucrose density cushion. This analysis revealed that GST-mAD-RING bound to Ub∼O-UbcH5c with a significantly higher efficiency than GST-RING (Fig. 5c, top). However, both constructs showed similar binding affinities for uncharged UbcH5c (Fig. 5c, middle). These results suggest that the mAD promotes RING binding to charged E2 but does not affect binding to uncharged E2.
The mAD and RING engage in an intramolecular interaction.The MDM2 RING domain has a calculated charge of +9 at pH 7.0, whereas the mAD has a charge of −10, suggesting that these two domains may interact intramolecularly, which results in higher E3 ligase activity (Fig. 6a). Consistent with this notion, the GFP-mAD fusion protein coprecipitated with the MDM2 RING domain fragment in a cotransfection assay (Fig. 6b). GFP–mAD–MDM2-8GS failed to bind the RING fragment in this assay (Fig. 6c), suggesting that the hydrophobic residues of mAD are needed for strong binding to the RING. Analysis of GFP–mAD–MDM2-3R binding to RING was not informative due to undetectable expression (not shown).
The MDM2 acidic region binds to the RING domain. (a) Model of internal binding between the MDM2 AD and RING domains. SQ represents the region with multiple ATM phosphorylation sites (residues 386 to 429). (b) Intramolecular binding between the MDM2 acidic domain and RING domain was analyzed by cotransfection of MDM2 RING and Myc-tagged acidic domain expression plasmids into H1299 cells followed by RING IP and Myc Western blotting. (c) The effect of point mutations in the AD-RING interaction was analyzed by coexpression of the GFP–mAD–MDM2-8GS fusion with the RING fragment followed by IP-Western blotting.
Interactions between isolated protein fragments can be nonspecific due to misfolding or overexpression. To address this caveat, we established a proteolytic fragment release assay to detect the preexisting internal binding between AD and RING domains in full-length MDM2 (Fig. 7a). MDM2 expressed in H1299 cells was immobilized on Sepharose beads using N-terminus-specific 4B2 antibody and cleaved on-bead by using caspase-3 that cuts after residue 361 (47). The dissociation of the C-terminal RING fragment (residues 362 to 491) from the bead was monitored by Western blotting using antibody 4B11. The results showed that after complete cleavage with caspase-3 in a 10-min incubation, ∼50% of the C-terminal fragment remained bound to the 4B2 beads, indicating preexisting internal binding between the N- and C-terminal domains that persisted after cleavage (Fig. 7b, top). As expected, the N-terminal fragment remained immobilized on the bead by the 4B2 antibody after cleavage (Fig. 7b, bottom). Deletion of the AD (Δ210-290) reduced the association between N- and C-terminal fragments, suggesting that the AD was involved in internal binding (Fig. 7b). The MDM2-8GS mutant also showed moderately reduced internal binding (Fig. 7b). MDM2-3R internal binding was not affected, probably due to other acidic sequences in the full-length protein stabilizing the interaction. A time course experiment showed that after cleavage by caspase-3, the RING domain dissociated from the N-terminal fragment slowly, and deletion of the AD significantly accelerated the dissociation (Fig. 7c), confirming that the AD was needed for internal binding with RING.
Detection of intramolecular binding in MDM2 by using a fragment release assay. (a) Illustration of the fragment release assay. MDM2 was mobilized on beads by N-terminus-specific antibody 4B2 and cleaved at residue 361 by using caspase-3. The amount of the C-terminal RING domain that remained bound to the beads was determined by Western blotting using antibody 4B11. (b) Wild-type MDM2 and AD mutants were mobilized by using 4B2 and cleaved with caspase-3. The RING domain that remained associated with the N-terminal fragment or that dissociated into the supernatant was analyzed by 4B11 blotting. The immobilization of the N-terminal fragment was confirmed by Western blotting using a polyclonal MDM2 antibody. (c) The kinetics of C-terminal fragment release was examined by determining the amount of RING fragment remaining bound to the bead at different time points after caspase-3 cleavage.
The AD-RING interaction is inhibited by DNA damage.DNA damage inhibits p53 ubiquitination by ATM-dependent phosphorylation of MDM2 near the C terminus (15). If AD-RING intramolecular binding is important for E3 ligase activity, this interaction may be regulated by DNA damage. When endogenous MDM2 from SJSA cells was immobilized on beads using antibody 4B2 and cleaved with caspase-3, a significant fraction (∼50%) of the C-terminal RING fragment remained associated with the N-terminal fragment, similar to transfected MDM2 (Fig. 8a). Interestingly, cleavage of MDM2 from irradiated SJSA cells caused most of the RING fragment to be released into the supernatant (Fig. 8a). Treatment with the ATM kinase inhibitor KU-55933 after irradiation prevented this release. The results showed that phosphorylation by ATM regulates internal binding in MDM2, possibly contributing to the inhibition of p53 ubiquitination. The MDM2 RING domain has been shown to interact with RNA (rRNA and XIAP and p53 mRNAs). Treatment of MDM2 with RNase A did not affect RING fragment release after caspase-3 cleavage (Fig. 8a), suggesting that RNA was not required for disrupting AD-RING binding after irradiation.
Stress signaling regulates MDM2 intramolecular binding. (a) Endogenous MDM2 from 10-Gy-irradiated SJSA cells was pulled down with N-terminal antibody 4B2 and cleaved on the beads with caspase-3. The C-terminal fragment released into the supernatant or that remained bound to the beads was detected by Western blotting with 4B11. ATM activity was inhibited by the addition of the ATM kinase inhibitor (ATMi) KU-55933 before irradiation. A set of cell lysates was treated with 5 μg/ml RNase A at 23°C for 30 min, followed by MDM2 IP. A fraction of the RING fragment in the released samples appeared as a slower-migrating band, probably due to oxidation in vitro, which occurred only occasionally. (b) MDM2 was cotransfected with ARF into H1299 cells and analyzed by a fragment release assay as described above for panel a. (c) NARF6 cells were treated with 100 μM IPTG for 18 h to induce ARF expression. Endogenous MDM2 AD-RING binding was analyzed by a fragment release assay. (d) U2OS cells were treated with 10 nM actinomycin D (ActD) for 18 h to induce ribosomal stress. Endogenous MDM2 AD-RING binding was analyzed by a fragment release assay. (e) Diagram showing the disruption of MDM2 AD-RING binding by DNA damage and stabilization of binding by ARF, resulting in the inhibition of E3 activity.
Oncogenic stress activates p53 by inducing the expression of ARF. Previous studies suggested that ARF binds to the mAD region (residues 230 to 260) of MDM2 (31, 48), which was corroborated by the MDM2-3R mutant (Fig. 3d). Using the fragment release assay, we found that coexpression of ARF with MDM2 unexpectedly strengthened AD-RING internal binding after caspase-3 cleavage (Fig. 8b). Inducible expression of ARF in the NARF6 cell line also moderately increased AD-RING binding of endogenous MDM2 (Fig. 8c). Induction of ribosomal stress by low-dose actinomycin D treatment also stabilized AD-RING binding in endogenous MDM2 from U2OS cells (Fig. 8d). These results suggest that ARF and ribosomal proteins may be incorporated into the AD-RING complex, forming a stabilized ternary structure that has no E3 ligase activity. Overall, these results suggest that both phosphorylation and protein binding regulate MDM2 E3 function by disrupting or altering the intramolecular interaction between the AD and RING (Fig. 8e).
DISCUSSION
Intramolecular interactions in multidomain proteins often mediate autoinhibition effects (49). Classic examples include the autoinhibition of Src family kinases by SH2-mediated internal binding to a phosphorylated C-terminal tyrosine (50) and the regulation of internal binding in pRb by cyclin/cdk-mediated phosphorylation, which controls binding to E2F1/DP1 (51). Recent studies showed that an intramolecular interaction in MDMX blocks the N-terminal hydrophobic pocket and inhibits binding to p53 (52, 53). DNA damage-mediated phosphorylation stimulates the MDMX internal interaction, thus inhibiting binding to p53 (52).
The results described in this report showed that the MDM2 AD and RING domain engage in an intramolecular interaction that activates the ubiquitination of p53. Previous studies showed that deletion of the MDM2 AD blocked the degradation of p53. Substitution of the MDM2 AD with the corresponding sequence from MDMX did not restore p53 ubiquitination (25, 26). The MDM2 AD has been shown to interact with the p53 core domain and promote p53 ubiquitination (32, 33). A recent study using small internal deletion and point mutations showed that residues 247 to 274 of the AD are important for p53 ubiquitination (54), which is in agreement with our mapping of the mAD region (residues 230 to 260). However, the molecular mechanism of the AD function remains poorly understood. Our results suggest that the AD can function as an activation domain to stimulate RING domain E3 activity. Two RING activities critical for E3 ligase function, i.e., recruitment of Ub∼E2 and activation of ubiquitin release from E2, are activated by the AD. Currently, it is not clear whether increased Ub∼E2 binding is solely responsible for accelerated ubiquitin release or whether they are two distinct functions of the AD.
The structural basis of AD function remains to be further determined. Its interaction with the RING may change the RING conformation to increase the affinity for Ub∼E2, or it may provide additional contacts with Ub∼E2 to facilitate binding. It is also unknown how ATM phosphorylation sites near the RING regulate AD-RING binding. Previous studies showed that phosphorylation of these sites inhibit RING dimerization (16), suggesting that they affect the conformation of the RING domain. Since the ATM sites are in an unstructured and acidic region, they may also interact transiently with the RING domain to alter its conformation or block the RING from binding to the AD through charge-mediated internal competition.
Several regions of MDM2 are predicted to be moderately or significantly disordered, including the AD. Therefore, structural analysis of MDM2 has been limited to the N-terminal p53-binding domain (55), central zinc finger (56), and C-terminal RING domain without the disordered sequences (57). The MDM2 RING domain crystal structure has revealed dimerization and potential E2-binding interfaces. However, MDM2 without the AD promotes only p53 monoubiquitination, which is inefficient for degradation by the proteasome. Our results suggest that the AD is an essential part of active MDM2 that was not captured by crystallographic studies. The weak and dynamic nature of the AD-RING interaction may be critical for MDM2 regulation by multiple modifications and binding partners.
The mAD region has been identified as the ARF-binding site in previous studies (48). Our results showed that ARF stabilizes AD-RING internal binding, which is the opposite of the disruptive effect of ATM-mediated phosphorylation. Other inhibitors of p53 ubiquitination, such as ribosomal proteins, may also function by stabilizing the AD-RING complex during nucleolar stress. These results suggest that disrupting or abnormally stabilizing AD-RING internal binding can achieve similar inhibitory functions. Presumably, the ARF-AD complex is able to interact with the RING in a nonproductive fashion, forming an inactive ARF-AD-RING ternary complex. Proteins that bind to the MDM2 AD are rarely subjected to ubiquitination by MDM2. Our finding suggests that AD-binding proteins may escape ubiquitination by interfering with its RING-activating function.
Intrinsically disordered regions are often the sites of regulatory protein binding and posttranslational modifications (37, 38). Our results from MDM2 and MDMX analyses suggest that these regions may interact with the functional domains in the same protein, forming dynamic structures that block or enhance their activity. Because of the intramolecular nature, even low-affinity interactions can alter a significant fraction of the population. It is possible that although the MDM2 RING domain can fold independently, it does not have the optimal conformation for Ub∼E2 binding and catalysis. The AD forms a dynamic structure with RING that has higher E3 ligase activity needed for polyubiquitin chain synthesis. The complementary net charges in the AD and RING may be important for initiating internal binding, whereas the hydrophobic residues in the AD may contribute to specific interactions that activate the RING. Thus, both types of residues in the mAD are highly conserved. Since MDM2 forms dimers and oligomers, we cannot rule out intermolecular (but intracomplex) AD-RING interactions contributing to activation. However, assuming that there is no unexpected structural constraint that makes trans-activation obligatory, intramolecular contact should be dominant because the two domains are covalently linked.
Currently, therapeutic targeting of MDM2 focuses mainly on disrupting MDM2-p53 interactions (58). It appears that cellular signaling pathways activate p53 mainly by regulating the AD-RING complex. The MDM2 RING domain has become an interesting target of recent drug discovery efforts (59, 60). Strategies that activate p53 in tumors without inducing genotoxic damage or that prevent p53 activation in normal tissues may have therapeutic benefits in the treatment of tumors with wild-type or mutant p53 (7, 61). Our current findings suggest that drug discovery against MDM2 E3 ligase function should consider the AD-RING complex as the target rather than the isolated RING domain. Furthermore, structural analysis of the AD-RING complex may reveal features amenable for therapeutic targeting, which may not be present in the structure of the isolated RING domain.
ACKNOWLEDGMENTS
We thank the Moffitt Molecular Genomics Core for DNA sequence analyses.
This work was supported by grants from the National Institutes of Health (CA141244 and CA109636) and the Florida Department of Health (4BB14) to J.C.
We declare no conflicts of interest.
FOOTNOTES
- Received 18 February 2014.
- Returned for modification 14 March 2014.
- Accepted 1 May 2014.
- Accepted manuscript posted online 19 May 2014.
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
