Secondary interaction between MDMX and p53 core domain inhibits p53 DNA binding

Significance

MDMX is a critical regulator of p53 and a potential drug target. The mechanisms by which MDMX inhibit p53 are not fully understood. Results in this report suggest that MDMX inhibits p53 DNA-binding function. Using a protein fragment release assay, MDMX and p53 were found to engage in multiple strong secondary interactions following initial binding through the canonical binding domains. These secondary interactions are involved in blocking p53 DNA binding and stabilizing the MDMX–p53 complex. The results suggest that secondary interactions play important roles in regulating the function of multidomain protein complexes.

Abstract

The MDMX oncoprotein is an important regulator of tumor suppressor p53 activity during embryonic development. Despite sequence homology to the ubiquitin E3 ligase MDM2, MDMX depletion activates p53 without significant increase in p53 level, implicating a degradation-independent mechanism. We present evidence that MDMX inhibits the sequence-specific DNA binding activity of p53. This function requires the cooperation between MDMX and CK1α, and phosphorylation of S289 on MDMX. Depletion of MDMX or CK1α increases p53 DNA binding without stabilization of p53. A proteolytic fragment release assay revealed that in the MDMX–p53 complex, the MDMX acidic domain and RING domain interact stably with the p53 DNA binding domain. These interactions are referred to as secondary interactions because they only occur after the canonical-specific binding between the MDMX and p53 N termini, but exhibit significant binding stability in the mature complex. CK1α cooperates with MDMX to inhibit p53 DNA binding by further stabilizing the MDMX acidic domain and p53 core domain interaction. These results suggest that secondary intermolecular interaction is important in p53 regulation by MDMX, which may represent a common phenomenon in complexes containing multidomain proteins.

The p53 tumor suppressor is a transcription factor that plays critical roles in promoting DNA repair, cell cycle arrest, or apoptosis in response to various types of damage and stress (1). p53 binds to specific DNA sequences as a tetramer. The core domain involved in DNA binding has poor thermostability and can undergo rapid spontaneous denaturation at physiological temperatures (2). Single amino acid mutations in the p53 DNA binding domain occur in more than 50% of human tumors, resulting in p53 misfolding and accumulation to high levels (3–5). The p53 mutants found in tumors are generally deficient for DNA binding and fail to regulate expression of its target genes.

p53 is present at low levels in unstressed tissues because of rapid turnover. This regulation is achieved mainly through MDM2 binding to p53 and acting as an ubiquitin E3 ligase to promote its proteasomal degradation (6, 7). MDMX is a p53 binding protein with strong sequence homology to MDM2 (8). Similar to MDM2, MDMX can bind to p53 N-terminal transactivation domain and inhibit p53 transactivation of target genes (9). However, MDMX lacks robust ubiquitin ligase activity and is unable to target p53 for proteasomal degradation. The prevailing view is that MDMX mainly functions by regulating p53 transcriptional activity, whereas MDM2 regulates p53 degradation (10). MDMX forms a heterodimer with MDM2 through the C-terminal RING domains. An important role of MDMX–MDM2 interaction is the regulation of MDMX stability. MDMX level is controlled by MDM2-mediated ubiquitination in a stress-dependent fashion (11, 12). Significant degradation of MDMX occurs after DNA damage through phosphorylation on several C-terminal sites, with S367 phosphorylation by Chk2 being most critical (13). MDMX knockdown in cell culture generally failed to cause significant change in p53 level. Furthermore, MDMX knockout in mice leads to p53 activation without significant stabilization (10).

In addition to promoting p53 ubiquitination through the RING domain, MDM2 inhibits p53 binding to DNA. The MDM2 acidic domain (AD) has a weak interaction with p53 core domain (14). MDM2 binding induces conformational change in the p53 core domain, which is detectable by using the Pab240 antibody that recognizes misfolded p53 (15). In contrast, MDMX alone fails to induce the Pab240 epitope on p53 and does not inhibit p53 DNA binding in vitro (15). However, earlier findings showed that MDMX knockdown caused a significant increase in p53 transcriptional output (16). These results led us to investigate the possibility that MDMX inhibits p53 DNA binding in vivo in cooperation with other factors.

A major MDMX-associated protein is casein kinase 1 alpha (CK1α) (17). CK1α interacts with the central region of MDMX, including the acidic domain and zinc finger, promotes phosphorylation of S289, and stimulates MDMX-p53 binding (17). Recent studies suggest that CK1α disrupts an intramolecular autoinhibitory interaction between the MDMX N-terminal domain and acidic domain, thus enhancing MDMX-p53 binding (18). DNA damage induces Chk2-mediated phosphorylation of MDMX at S367 that inhibits CK1α binding, which leads to inhibition of MDMX–p53 interaction (18).

In this report, we show that in the presence of CK1α, MDMX inhibits the DNA binding activity of p53. Using a proteolytic fragment release assay (PFR), we detected robust interaction between the MDMX AD and RING domain with p53 in the MDMX–p53 complex. CK1α promotes the binding of MDMX AD to the p53 core domain, suggesting a mechanism by which CK1α and MDMX cooperate to inhibit p53. The RING domain of MDMX is also important for stabilizing MDMX–p53 interaction and inactivation of p53. Our results reveal the importance of secondary interactions between MDMX and p53 in regulating p53 activity, which may represent a mechanism commonly used in other protein complexes.

Results

MDMX and CK1α Cooperate To Inhibit p53 DNA Binding in Vitro.

To determine how MDMX inhibits p53 activity, we tested whether MDMX and CK1α cooperate to regulate p53 DNA binding by using an affinity pull-down assay (15). When the MDM2–p53 complex was purified from cotransfected cells and incubated with oligonucleotide-containing p53-binding site, poor binding to DNA was detected (Fig. 1A). This result is consistent with the ability of MDM2 to inhibit p53 DNA binding (15). In contrast, MDMX–p53 complex only showed moderate loss of DNA-binding activity compared with FLAG-p53. When MDMX and CK1α were coexpressed with p53, the MDMX–p53–CK1α complex showed a further loss of DNA binding (Fig. 1A), suggesting that MDMX inhibits p53 DNA binding after forming a complex with CK1α.

Fig. 1.

MDMX and CK1α cooperate to inhibit p53 DNA binding in vitro. (A) H1299 cells were transiently transfected with indicated plasmids. FLAG-tagged protein complexes were immunopurified and eluted with FLAG peptide. Eluates containing equal amounts of p53 were captured with biotinylated oligonucleotide DNA containing p53-binding site. An aliquot of eluate was analyzed to confirm similar levels of p53 expression. (B) H1299 cells were transiently transfected with p53, FLAG-MDMX, and CK1α plasmids. MDMX–p53 complexes were purified, eluted with FLAG peptide, and analyzed for binding to biotinylated oligonucleotide DNA containing p53-binding site. (C) Quantitation of p53 DNA binding in B and two repeat experiments.

To further test the role of MDMX–CK1α interaction, several MDMX mutants were analyzed. The MDMX-C306S mutant (does not bind CK1α because of zinc finger mutation) and MDMX-S289A (CK1α phosphorylation site mutant) failed to inhibit p53 DNA binding in the presence of CK1α. MDMX-S367A cooperation with CK1α was more efficient than wild-type MDMX, presumably because it is resistant to Chk2-mediated phosphorylation that blocks CK1α binding (Fig. 1 B and C). Therefore, MDMX-CK1α binding and phosphorylation of S289 by CK1α is important for inhibiting p53 DNA binding. Furthermore, deleting the AD (∆200–304) abrogated the ability of MDMX to inhibit p53 DNA binding, whereas deleting the RING (1–430) had no effect, suggesting that the AD is critical for inhibiting p53 DNA binding (Fig. 1B). MDMX inhibition of p53 DNA binding was also confirmed by using electrophoretic mobility shift assay (EMSA). The ability of p53 to induce mobility shift of a probe containing p53-binding site was inhibited by the addition of purified MDMX–CK1α complex. The activity of MDMX was abrogated by mutation of CK1α phosphorylation site and mutation of the p53 binding pocket (Fig. S1).

Fig. S1.

MDMX–CK1α complex inhibits p53 DNA binding. (A) EMSA analysis of p53 binding to DNA. Purified MDMX and MDMX/CK1α complexes were added to DNA-binding reaction containing in vitro translated p53, Pab421 antibody, and IRDye 700-labeled DNA probe. p53 DNA binding was detected by probe mobility shift. MDMX-V92W contains a bulky substitution in the N-terminal hydrophobic pocket that prevents binding to p53. (B) Coomassie stain of purified FLAG-MDMX. (C) GST-CK1α copurification with FLAG-MDMX. FLAG-MDMX was coexpressed with GST-CK1α in E. coli and purified by M2 beads. The presence of copurified GST-CK1α was confirmed by CK1α Western blot. (D) Formation of p53–MDMX complex in vitro. p53 and purified FLAG-MDMX were mixed under EMSA reaction condition, followed by p53 IP using Pab1801 antibody and detection of coprecipitated MDMX and CK1α by Western blot to confirm interaction in vitro.

Secondary Binding of MDMX Acidic and RING Domains to p53.

To determine how MDMX inhibits p53 DNA binding, we tested whether the MDMX AD binds to the p53 core domain. When p53-loaded beads were used to capture MDMX mutants, only the mutants containing intact N-terminal 1–100 region (canonical p53BD) were able to bind p53 (Fig. S2A). In the second assay, a cleavable MDMXc3 construct was used for the analysis. MDMXc3 contains PreScission cleavage sites and epitope tags inserted into three flexible regions of MDMX (19). Cleavage by PreScission produces four fragments containing different eiptopes (Fig. S2B). When MDMXc3 expressed in H1299 cells was cleaved with PreScission and then incubated with p53 beads, significant capture of the MDMX N terminus was detected by p53-1-393 and p53-1-80 (Fig. S2C, Bottom), both containing the canonical-binding site for MDMX. Other MDMX fragments were not captured by p53, indicating that only the MDMX N terminus has strong binding to the p53 N terminus when presented individually to p53.

Fig. S2.

Analysis of p53 binding to separated MDMX domains. (A) Beads loaded with GST-p53 were incubated with H1299 lysate-expressing MDMX mutants. The captured MDMX mutants were detected by Western blot. (B) Lysate of H1299 transfected with MDMXc3 was digested with PreScission, incubated with GST-p53 beads for 2 h and analyzed by Western blot to detect the pulldown of fragments. (C) PreScission cleavage site and epitope tags were inserted after residues 140, 350, and 429 of MDMX to create MDMXc3. Cleavage by PreScission produces four fragments containing unique epitopes for detection of binding to GST-p53 loaded beads.

To test whether there are intermolecular interactions in the MDMX–p53 complex that conventional pull-down assays failed to detect, MDMXc3 was first captured with GST-p53 beads and then subjected to on-bead cleavage by PreScission. We hypothesized that after cleavage of the complex, MDMX fragments that engage in binding to p53 will dissociate slowly compared with fragments that have no affinity for p53. We refer to such experiments as PFR assay (Fig. 2A). The PFR analysis revealed that in addition to the canonical N-terminal binding, the MDMX AD and RING fragments also showed significant association with p53 (Fig. 2B). Furthermore, the RING domain binding to p53 was as stable as the p53BD fragment, with >75% remaining bound to the beads after cleavage of MDMXc3. In contrast, the SQ fragment was completely released in the same time span (Fig. 2B), providing an internal control. Therefore, the PFR analysis uncovered stable binding of MDMX AD and RING domains to p53 in the full-length complex. The AD-p53 binding was detergent-sensitive, suggesting it is mainly mediated by hydrophobic interactions (Fig. S3A).

Fig. 2.

PFR assay. (A) Diagram of PFR assay for detecting secondary interactions. MDMXc3 in H1299 lysate was captured onto GST-p53 beads and cleaved on-bead with PreScission for 60 min. The beads and supernatant were analyzed for MDMX fragments by Western blot to detect their release from the beads. (B) Result of PFR assay to detect MDMX fragment retention on beads loaded with GST-p53. (C) MDMXc3 and p53 were cotransfected into H1299 cells and the MDMXc3–p53 complex was immobilized by using p53 antibody Pab421. The beads were incubated with PreScission for 30 min, and the release of MDMX AD fragment was analyzed by FLAG Western blot.

Fig. S3.

Characterization of MDMX AD–p53 interaction. (A and B) Effects of detergent and salt concentrations on MDMX–p53 interactions. MDMXc3 expressed in H1299 lysate was captured onto GST-p53 beads. The beads were washed and suspended in PreScission buffer (10 mM Hepes pH 7.5, 0.5 mM DTT, 10% glycerol, NaCl and Nonidet P-40 concentrations as indicated) and cleaved on-bead with PreScission for 60 min at 23 °C. The beads and supernatant were analyzed for the release of MDMX fragments by Western blot. (C) MDMXc3 expressed in H1299 lysate was captured onto GST-p53 beads. The beads were washed and suspended in PreScission buffer (10 mM Hepes pH 7.5, 0.5 mM DTT, 10% glycerol, 150 mM NaCl, 0.05% Nonidet P-40), containing indicated peptides and cleaved on-bead with PreScission for 60 min at 23 °C. The beads and supernatant were analyzed for the release of MDMX fragments by Western blot. pDI (LTFEHYWAQLTS) is a p53-mimetic peptide that inhibits the binding of MDMX N-terminal domain to p53. pDI-3A (LTAEHYAAQATS) is a mutated control of pDI that does not inhibit MDMX-p53 N-terminal binding.

To confirm that the novel interactions also occur in MDMX–p53 complex formed in vivo, the p53 antibody Pab421 was used to capture the p53–MDMXc3 complex from H1299 cells. The p53–MDMXc3 complex was digested with PreScission, and the release of MDMX fragments was analyzed. The result also showed strong AD and RING fragment binding to immobilized p53 (Fig. 2C). Overall, the results suggest that after specific binding mediated by the MDMX and p53 N termini, the MDMX AD and RING also establish robust binding to p53. We refer to these interactions as secondary binding, because they require initial interaction through the N terminus of MDMX (Fig. 2A). Once the MDMX–p53 complex was formed, the AD and RING binding to p53 no longer required the p53BD, because adding the MDMX N-terminal pocket binding peptide pDI during PreScission cleavage promoted the release of p53BD fragment, but did not promote AD and RING dissociation (Fig. S3C).

MDMX Acidic and RING Domains Interact with p53 Core Domain.

To map the secondary binding sites for the MDMX AD and RING domains on p53, p53 mutants were used to capture MDMXc3, followed by PFR analysis. The results showed that p53-1-300 and p53-1-393 bound to the p53BD, AD, and RING. P53-1-82 bound to the p53BD but not AD and RING (Fig. 3A). Therefore, the core domain of p53 (83–300) contains the secondary binding sites for MDMX AD and RING.

Fig. 3.

MDMX AD and RING bind to the p53 core domain. (A) Beads loaded with GST-p53 mutants were used to capture MDMXc3 from H1299 lysate, cleaved with PreScission for 60 min, and analyzed for the release of MDMX fragments from the beads by Western blot. (B) MDMXc3 containing W200S/W201G mutations were captured by GST-p53 beads, cleaved with PreScission, and analyzed for fragment release by Western blot. (C) H1299 cells were cotransfected with MDMXc3 and CK1α mutants. MDMXc3 was captured onto beads loaded with GST-p53 and cleaved with PreScission in the presence of 0.5% Nonidet P-40. The release of MDMX AD fragment and CK1α from p53 was detected by Western blot. (D) H1299 cells were cotransfected with MDMXc2 and CK1α. MDMXc2 was captured onto beads loaded with GST-p53 and cleaved with PreScission in the presence of 0.5% Nonidet P-40. The release of MDMX AD and SQ-RING fragments from p53 was detected by Western blot. MDMXc2 is similar to MDMXc3 except without the PreScission site between SQ and RING.

Recent analysis showed that the MDMX AD engages in intramolecular interactions with the N terminus and RING (19). The MDMX intramolecular interactions can be disrupted by mutating two conserved residues in the AD (W200S/W201G). To test whether W200/W201 are important for AD binding to the p53 core, PFR analysis was performed by using MDMXc3-W200S/W201G mutant. The AD containing SG mutation showed significantly reduced binding to p53 (Fig. 3B, compare ratio of lanes 1/2 vs. 3/4), whereas the RING domain from the SG mutant retained strong binding to p53. Therefore, the conserved W200/W201 residues are important for AD binding to p53.

CK1α disrupts the intramolecular interaction between MDMX N terminus and AD, thus stimulating the N-terminal binding to p53 (18, 19). We found that under stringent PFR assay condition (0.5% Nonidet P-40; Fig. S3 A and B), CK1α also stabilized the secondary MDMX AD-p53 binding (Fig. 3C, note the change in ratio of lanes 1/2 vs. 3/4). Mutation of the CK1α phosphorylation site on MDMX (MDMXc3-S289A) abrogated the stimulation by CK1α (Fig. 3C, lanes 7/8 vs. 9/10). The CK1α-D136N kinase-dead mutant also failed to stimulate AD-p53 binding (Fig. 3C, lanes 1/2 vs. 5/6). The effect of CK1α on RING-p53 binding was not determined in this experiment because of difficulty in detecting the small RING fragment. To address this issue, we repeated the analysis by using the MDMXc2 construct (without the C-terminal PreScission site) that produced a larger C-terminal SQ-RING fragment. CK1α increased AD-p53 binding without significant effect on RING-p53 binding (Fig. 3D, lanes 1/2 vs. 3/4). The results suggest that CK1α cooperates with MDMX to inhibit p53 DNA binding by phosphorylating S289, enabling the AD to bind p53 core domain with higher affinity.

RING–p53 Interaction Contributes to p53 Inactivation.

In the PFR assay, the binding of RING to p53 was as stable as the canonical N-terminal interaction (Fig. 2B). However, the RING was not needed for inhibiting p53 DNA binding, as suggested by the RING deletion mutant (MDMX-1-430, Fig. 1B). We tested whether the RING-p53 binding has other roles such as preventing dissociation of the MDMX–p53 complex. p53-loaded beads were used to capture MDMX deletion mutants (Fig. 4A). The beads containing the preformed complex were incubated in buffer, and the dissociation of MDMX from the beads was monitored. The MDMX constructs containing the RING domain had no significant dissociation from p53 during the 4-h span. MDMX mutants without the RING domain (1–200, 1–300) showed significant dissociation (Fig. 4B). The results suggest that the canonical p53-binding domain (1–100) interacts with p53 in a reversible manner. Presence of the RING domain stabilizes the MDMX–p53 complex by providing a strong secondary interaction with p53. When GST-p53-1-82 (without the core domain) was used for the binding stability analysis, all MDMX constructs showed similar degrees of dissociation, consistent with the absence of stabilizing effect from the RING (Fig. 4C).

Fig. 4.

The RING domain increases MDMX-p53 binding stability. (A) Diagram of MDMX deletion mutants. (B and C) Beads loaded with GST-p53 or GST-p53-1-82 were used to capture MDMX mutants expressed in H1299 lysate. After removal of unbound MDMX by washing, the beads were incubated with excess buffer for indicated times at 4 °C, and the MDMX remaining bound to the beads were analyzed by Western blot. Quantitation is shown below the blots. (D and E) Gal4-p53 and Gal4-p53-1-52 activation of Gal4-TK-luc reporter in MDMX/p53 double-null MEF cells was used to measure the inhibitory effect of different MDMX mutants. Statistically significant differences from the controls are marked by asterisk (Student's t test; *P < 0.05).

To test whether the RING–p53 interaction contributes to p53 inhibition, luciferase assay was used to compare the ability of MDMX mutants to inhibit p53. Against the full-length Gal4-p53-1-393, the MDMX-∆200-304 mutant was a stronger p53 inhibitor than MDMX-1-200 and MDMX-1-300, consistent with stabilization of the MDMX–p53 complex by the RING (Fig. 4D). When the MDMX mutants were tested for inhibition of p53 transactivation domain alone (Gal4-p53-1-52), MDMX-∆200-304 no longer had an advantage over MDMX-1-200 and MDMX-1-300, consistent with the absence of secondary interactions (Fig. 4E). Full-length MDMX activity was moderate in this assay despite containing the RING domain, probably due to multiple intramolecular interactions (19). The results suggest that the RING contributes to p53 inhibition by stabilizing the MDMX–p53 complex, thus helping to conceal the p53 transactivation domain. This function may cooperate with AD inhibition of p53 DNA binding to regulate transcriptional output.

MDMX Inhibits p53 DNA Binding in Vivo.

Knockdown of MDMX in tumor cell lines did not cause a significant increase in p53, but induced the expression of p53 targets (p21, PUMA, MDM2) at both protein and mRNA levels (Fig. 5 A and B). Knockdown of CK1α also induced p53 targets without affecting p53 level (Fig. 5 A and B and Fig. S4). ChIP analysis showed that p53 binding to p21/PUMA/MDM2 promoters increased after depletion of MDMX or CK1α (Fig. 5C). Therefore, MDMX and CK1α appeared to inhibit p53 in part by blocking its DNA binding.

Fig. 5.

MDMX and CK1α inhibit p53 DNA binding in vivo. (A) Cells were treated with MDMX and CK1α siRNA for 48 h and analyzed for the expression of indicated markers by Western blot. (B and C) A549 cells were treated with MDMX and CK1α siRNA for 48 h and analyzed by RT-PCR for p21, PUMA, and MDM2 transcription (B), and by ChIP to detect p53 binding to the promoters (C).

Fig. S4.

Characterization of CK1α siRNA. (A and B) A549 and U2OS cells were transfected with CK1α siRNAs for 48 h and analyzed for the level of indicated markers by Western blot. (C) U2OS cells stably infected with MDMX-S367A mutants were used to test the effects of different CK1α siRNAs on MDMX level. SiRNA #2 was selected for further experiments.

MDMX has been shown to block p53 activation by stress such as ARF expression, ribosomal stress, and DNA damage. To test whether ARF overcomes MDMX inhibition of p53 DNA binding, MDMX was stably transfected into a U2OS cell line expressing IPTG-inducible ARF (NARF6) (20). Treatment of NARF6 cells with IPTG induced ARF, resulting in p53 accumulation and induction of p21. MDMX strongly inhibited p53 activation by ARF (Fig. S5A). Unlike ARF, IR was able to moderately activate p53 in the presence of MDMX, but failed to overcome the inhibition by MDMX-S367A (Fig. S5B). ChIP analysis showed that in the presence of MDMX and MDMX-S367A, the p53 stabilized by ARF and IR had significantly reduced DNA binding (Fig. S5 C and D). Activation of p53 DNA binding by IR requires phosphorylation of S367. The results suggest that MDMX inhibits p53 DNA binding in vivo. This function is regulated by phosphorylation of S367, which disrupts MDMX–CK1α interaction (18).

Fig. S5.

MDMX inhibits ARF-mediated p53 activation. (A) NARF6 cells were stably infected with lentivirus-expressing MDMX and MDMX-S367A mutant. ARF expression was induced by treatment with IPTG for 24 h. Induction of p21 expression by p53 was analyzed by Western blot. (B) NARF6 cells stably infected with MDMX and MDMX-S367A lentiviruses were treated with 10 Gy IR for 4 h and analyzed for p21 induction by Western blot. (C and D) NARF6 cells stably infected with MDMX and MDMX-S367A lentiviruses were treated with IPTG to induce ARF expression for 24 h or treated with 10 Gy IR for 4 h. p53 binding to the p21 promoter was determined by ChIP. Statistically significant differences from the controls are marked by asterisks (Student's t test; *P < 0.05).

CK1α Is Required for MDMX Inhibition of p53 DNA Binding in Vivo.

To further test the role of CK1α, we took advantage of the MDMX-S367A mutant that binds CK1α constitutively after IR (18). We generated U2OS cells stably expressing MDMX and MDMX-S367A mutant. MDMX and MDMX-S367A binding to p53 were reduced after knockdown of CK1α, confirming previous reports that CK1α is involved in stimulating MDMX-p53 binding (Fig. 6A). Next, the impact of CK1α knockdown on MDMX-S367A and MDMX-S289A/S367A activities after DNA damage was determined. MDMX-S367A strongly inhibited p21 induction by IR (Fig. 6B). Knockdown of CK1α or mutating the CK1α phosphorylation site (S289A/S367A double mutant) partially abrogated the ability of S367A mutant to block p21 induction. The S289A/S367A double mutant did not respond further to CK1α knockdown (Fig. 6B). ChIP and RT-PCR analyses showed that both CK1α knockdown and S289A mutation partially abrogated the ability of S367A mutant to inhibit p53 DNA binding (Fig. 6 C and D) and p21/PUMA mRNA induction (Fig. S6 A and B). Therefore, MDMX inhibition of p53 DNA binding in vivo requires the cooperation of CK1α and phosphorylation of S289. It is noteworthy that although MDMX-S289A/S367A did not inhibit p53 binding to the PUMA promoter (Fig. 6D), it still partially inhibited PUMA mRNA induction by IR (Fig. 6B), suggesting that DNA binding is a contributing factor but not the only mechanism by which MDMX inhibits p53 transcriptional output.

Fig. 6.

Endogenous CK1α regulates MDMX-p53 binding. (A) U2OS cells stably infected with MDMX and MDMX-S367A lentiviruses were treated with CK1α siRNA for 48 h. MDMX-p53 binding was determined by IP-Western blot. (B) U2OS cells infected with MDMX-S367A, and MDMX-S289A/S367A lentivirus were treated with CK1α siRNA for 48 h, 10 Gy IR for 4 h, and analyzed for p21 induction by Western blot. (C and D) U2OS infected with MDMX mutant lentivirus were treated with CK1α siRNA for 48 h, 10 Gy IR for 4 h, and p53 binding to p21 and PUMA promoter was analyzed by ChIP. (E) A model of p53 inhibition by MDMX. The MDMX AD engages in intramolecular interactions with the p53BD and RING domains in the absence of p53. CK1α binding to MDMX releases the p53BD from autoinhibition, allowing it to bind p53. Following formation of the initial MDMX–p53 complex, the AD and RING also establish interactions with the p53 core domain, stabilizing the complex and blocking p53 DNA-binding function. CK1α-mediated phosphorylation of MDMX S289 increases the binding affinity of AD to p53 core domain, which is necessary for inhibiting p53 DNA binding.

Fig. S6.

Endogenous CK1α regulates MDMX inhibition of p53. (A and B) U2OS stably expressing MDMX and MDMX mutants were treated with CK1α siRNA for 48 h, followed by treatment with 10 Gy IR for 4 h. The induction of p21 and PUMA mRNA expression was analyzed by RT-PCR. See Fig. 6 B–D for corresponding protein and ChIP analyses.

Discussion

Several studies showed that MDMX inhibits p53 transcriptional function without increasing its degradation. The results described above suggest a two-step model of p53 inhibition by MDMX (Fig. 6E): The binding between MDMX and p53 is initiated by the canonical interaction between the N-terminal domains. Subsequently, the MDMX AD and RING domains also interact with the p53 core domain, causing the loss of sequence-specific DNA binding. CK1α plays a critical role in this process by stimulating the primary p53 binding by the MDMX N-terminal pocket, and promoting the secondary binding of MDMX AD to p53 core domain. In this model, the MDMX AD has opposite effects on p53, depending on whether CK1α is present. In the absence of CK1α, the AD acts as a p53 mimetic and autoinhibitory domain to reduce MDMX-p53 binding. In the presence of CK1α, the AD becomes an active participant in suppressing p53 DNA binding activity. This mechanism is subjected to tight regulation by DNA damage signaling, because Chk2-mediated phosphorylation of S367 disrupts MDMX–CK1α interaction and inhibits MDMX-p53 binding (18). Our model does not rule out other mechanisms of p53 inhibition, because MDMX mutants without the AD can inhibit p53 by blocking the transactivation domain.

Peptides and small molecule inhibitors of the MDMX N-terminal pocket efficiently inhibit MDMX-p53 binding. Therefore, MDMX–p53 complex formation strictly depends on the MDMX N-terminal p53BD. Although the separated AD and RING fragments interact with p53 poorly in vitro, the PFR assay revealed that in the mature complex, these domains bind to p53 with significant stability. It is possible that only the MDMX N-terminal domain has the conformation and rapid on/off binding rates that enable it to seek out p53 from a crowded intracellular environment. In contrast, the AD and RING domains are optimized for rapid binding to other partners such as CK1α and MDM2. Their binding to p53 may require slow conformational changes, which is only relevant after formation of the initial complex through the N terminus.

The partially unstructured MDM2 AD has significant binding affinity for ribosomal proteins, ARF, transcription repressors, and chromatin modifying enzymes. MDM2 AD also provides a second binding site to p53 and Notch, thus leading to the suggestion of a two-site binding model for these interactions (21, 22). The MDM2 AD binds to p53 core domain with measurable affinity in pull-down and ITC assays, induces a Pab240-reactive conformational change, and inhibits p53 DNA binding in the full-length complex (14, 15). In comparison, MDMX alone was incapable of inducing these changes, and requires CK1α cooperation to inhibit p53 DNA binding. The MDM2 AD–p53 core interaction is mainly charge-mediated (14). The MDMX AD region has lower density of negative charges compared with MDM2, thus it may require phosphorylation on S289 by CK1α, which increases negative charges and binding affinity to p53. The MDM2 AD has been shown to bind to the DNA-binding surface of p53 core domain as a nucleic acid mimic (14). It is possible that the MDMX AD acts in a similar fashion after phosphorylation by CK1α.

In addition to the AD–core interaction, the PFR assay also identified stable binding between the RING and p53 core domain. This interaction is important for stabilizing the full-length MDMX–p53 complex. The significance of the MDMX RING was demonstrated by mutant mouse models with point mutation or internal deletion in the RING (23, 24). The RING internal deletion results in p53 activation and embryonic lethality without increase of p53 level (23). Although the study was interpreted mainly from the perspective of MDM2-MDMX heterodimer formation, our results suggest that the phenotypes may also partly result from the disruption of RING–p53 interaction. In fact, the MDMX RING internal deletion mutant showed reduced p53 binding, which is consistent with our results (23).

It remains to be determined whether the secondary interactions described here are merely byproducts of the specific N-terminal binding, or represent optimized interactions evolved for important functions. It is conceivable that following binding by the N termini, other domains of p53 and MDMX are forced to interact. Some of the lowest energy configurations may involve forming new domain interactions. Because such structural rearrangements will have functional consequences, they will be fine-tuned by mutations and natural selection. The unique characteristics of secondary interactions may make them difficult to detect by using conventional binding assays. The PFR assay allows the discovery of interactions that preexist in full-length protein complexes, providing clues for further investigation. We speculate that secondary interactions may occur frequently in multidomain protein complexes and play important roles in mediating signaling.

Materials and Methods

Plasmids and Cell Lines.

The MDMXc3 sequence contains LEVLFQGPDYKDDDDK, LEVLFQGPEEQKLISEEDL, and LEVLFQGPYPYDVPDYA inserted after MDMX residues 140, 350, and 429 respectively. Cell lines H1299 (p53-null), A549, JEG3, MCF7 and U2OS (p53 wild type), MDMX/p53 double null MEF cells (41.4), NARF6 (U2OS with inducible ARF), and U2OS stably infected with MDMX lentiviruses were maintained in Dulbecco’s modified Eagle medium (DMEM) with 10% (vol/vol) FBS. GST-PreScission protease fusion was purified from Escherichia coli by using glutathione agarose column. This study did not involve human subjects. Therefore no Institutional Review Board approval was necessary.

Proteolytic Fragment Release Assay.

H1299 cells were transiently transfected with MDMXc3 plasmid by using standard calcium phosphate precipitation protocol. Cells were lysed by using IP buffer [150 mM NaCl, 50 mM Tris⋅HCl pH 8.0, 0.5% Nonidet P-40, 0.5 mM DTT, 10% (vol/vol) glycerol]. Cell lysate (1 mL) from ∼2 × 10⁶ cells (a 10-cm plate) was incubated with 20 µL of packed glutathione agarose beads loaded with ∼5 µg of GST-p53 for 2 h at 4 °C. The beads were washed two times with PreScission buffer (150 mM NaCl, 10 mM Hepes pH 7.5, 0.05% Nonidet P-40, 0.5 mM DTT, 10% glycerol) and suspended in 100 µL of PreScission buffer. PreScission protease was added to 0.2 µg/µL final concentration, and the beads were incubated at 23 °C with shaking for 20–60 min. The protease digestion mixture was centrifuged for 10 s, and the beads (bound material) and supernatant (released material) were separated. The beads were washed once with PreScission buffer to remove residual supernatant. The beads and supernatant were boiled in Laemmli sample buffer, and analyzed by SDS/PAGE and Western blot by using 8C6, FLAG, Myc, or HA antibodies to determine the bound/released ratio of each fragment.

Additional experimental procedures are described in SI Materials and Methods.

SI Materials and Methods

Western Blot.

Cells were lysed in lysis buffer [50 mM Tris⋅HCl (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40, 1× protease inhibitor mixture], centrifuged for 10 min at 14,000 × g, and the insoluble debris was discarded. Cell lysate (10–50 µg of protein) was fractionated by SDS-polyacrylamide gel electrophoresis and transferred to Immobilion P filters (Millipore). The filter was blocked for 1 h with PBS containing 5% (wt/vol) nonfat dry milk and 0.1% Tween 20, incubated with primary and secondary antibodies, and the filter was developed by using the Supersignal reagent (Thermo Scientific). MDM2 was detected by using antibody 3G9. MDMX was detected with antibody 8C6. Anti-actin and HA tag antibodies were purchased from Santa Cruz Biotechnology. FLAG tag antibody was from Sigma-Aldrich. DO-1 for p53 and p21 antibody was from BD Pharmingen.

Purification of MDMX-Associated p53 and DNA Affinity Immunoblotting.

H1299 cells were transiently transfected with FLAG-tagged MDMX and p53. Cells from a 10-cm plate were lysed in 1 mL of lysis buffer [50 mM Tris⋅HCl (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 0.7 µg/mL pepstatin A], centrifuged for 10 min at 14,000 × g, and the insoluble debris was discarded. The lysate was incubated with 40 µL of slurry of M2-agarose beads (Sigma) for 18 h at 4 °C. The beads were washed with lysis buffer, and the FLAG-tagged proteins with their binding partners were eluted with 150 µL of lysis buffer containing 50 µg/mL FLAG epitope peptide (Sigma) for 2 h at 4 °C. An aliquot of the eluted proteins was analyzed for expression levels by Western blot. Lysate containing equal levels of p53 was added to a 400 µL of DNA binding reaction mixture and incubated at 4 °C for 30 min. The DNA binding reaction mixture contains 25 nM (0.01 nmol) double-stranded biotinylated oligonucleotide DNA containing a consensus p53 binding site (Biotin-5′-TCGAGAGGCATGTCTAGGCATGTCTC annealed with 5′-GAGACATGCCTAGACATGCCTCTCGA), 2 µg of poly(dI•dC), 5 mM DTT, 150 mM NaCl, 20 mM Tris⋅HCl (pH 7.2), and 4% glycerol. The DNA/protein complexes were captured with 0.1 mg of magnetic Streptavidin beads (Promega) at 4 °C for 30 min. The beads were collected by using a magnet and washed three times with DNA binding buffer. The bound proteins were eluted by boiling in Laemmli sample buffer [4% (wt/vol) SDS, 20% glycerol, 200 mM DTT, 120 mM Tris (pH 6.8), 0.002% bromophenol blue]. The protein complexes were resolved by SDS/PAGE, and p53 was detected by Western blot using DO-1 antibody.

Luciferase Reporter Assay.

Cells (50,000 per well) were cultured in 24-well plates and transfected with a mixture containing 50 ng of luciferase reporter plasmid, 5 ng of CMV-lacZ plasmid, 50 ng of MDMX expression plasmids, and 5 ng of Gal4-p53 fusion expression plasmid. Transfection was achieved by using Lipofectamine PLUS reagents (Invitrogen). Cell lysate was analyzed for luciferase and β-gal expression after 24 h. The ratio of luciferase/β-gal activity was used as an indicator of transcription activity.

GST Pulldown Assay.

H1299 transiently transfected with MDMXc3 was lysed in lysis buffer [50 mM Tris⋅HCl (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40, 1× protease inhibitor mixture] and digested with PreScission. Bacterial lysate expressing GST-p53 fusion proteins were applied to glutathione-agarose beads according to the manufacturer instruction (Pierce). The beads (20 µL of packed volume) loaded with ∼1 µg of GST fusion proteins were incubated for 2 h at 4 °C with PreScission-digested H1299 lysate. The beads were washed with lysis buffer, boiled in Laemmli sample buffer, and detected by Western blot for MDMX fragments.

GST Pulldown and Dissociation Assay.

H1299 transiently transfected with MDMX deletion mutants were lysed in lysis buffer [50 mM Tris⋅HCl (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40, 1× protease inhibitor mixture]. Bacterial lysate expressing GST-p53 fusion proteins were applied to glutathione-agarose beads. The beads (20 µL of packed volume) loaded with ∼1 µg of GST fusion proteins were incubated for 2 h at 4 °C with H1299 lysate. The beads were washed in lysis buffer and incubated in 2.0 mL of lysis buffer for up to 4 h with constant end-to-end rotation. The beads were recovered and boiled in Laemmli sample buffer. The MDMX mutants that remained bound to the beads were detected by Western blot and quantified by densitometry.

RNA Interference.

Cells were transfected with 100 nM control siRNA or 100 nM MDMX and CK1α (Dharmacon) by using RNAiMAX (Invitrogen) according to instructions from the supplier. After 48 h of transfection, cells were treated with IR for 4 h and analyzed for protein expression or chromatin immunoprecipitation. The siRNAs used were as follows: Control siRNA (AATTCTCCGAACGTGTCACGT). CK1α siRNAs [1. AAGCAAGCTCTATAAGATTCT (used in Fig. 6A); 2. AAACGGUGGUAUGGUCAGGAA (used in Fig. 6 B and C); 3. AACAAGGCAACACAUACCAUA]. MDMX siRNA (AGATTCAGCTGGTTATTAA).

Chromatin Immunoprecipitation and Quantitative PCR.

ChIP assay was performed by using standard procedure. Proteins were cross-linked to genomic DNA with 1% formaldehyde for 10 min at 23 °C. The cross-link was stopped by 0.125 M glycine for 5 min at 23 °C. The cells were washed three times with ice-cold PBS and lysed in 0.45 mL of RIPA buffer with proteinase inhibitor mixtures on ice for 10 min. Cell pellet was sonicated five cycles by using Bioruptor XL (8 min/cycle, 30 s on and 30 s off) and then spun down at 14,000 × g for 10 min to remove debris. The lysates were precleared by incubating with a salmon sperm DNA/protein A agarose slurry for 30 min at 4 °C with rotation. The cleared lysates were diluted 1:6 in ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris⋅HCl, pH 8.0, and 167 mM NaCl). The samples were incubated with 1 μg of antibody for 18 h at 4 °C with rotation. Protein A agarose slurry (50 µL) were added and incubated for 2 h at 4 °C. The beads were washed once each with low salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris⋅HCl pH 8.0, 150 mM NaCl), high salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris⋅HCl pH 8.0, 500 mM NaCl), LiCl buffer (250 mM LiCl, 1% Nonidet P-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris⋅HCl pH 8.0), and TE buffer (10 mM Tris⋅HCl pH 8.0, 1 mM EDTA). The beads were eluted twice with a total of 200 μL of elution buffer (1% SDS, 100 mM NaHCO₃) at 25 °C for 15 min with shaking (1,000 rpm in an orbital mixer). NaCl was added to 200 mM, and the samples were decross-linked at 65 °C for 18 h. Proteinase K (0.1 mg/mL), EDTA (10 mM), and Tris⋅HCl pH 6.8 (40 mM) were added and incubated at 42 °C for 1 h. DNA was extracted by using GenElute PCR clean-up kit (Sigma). The samples were subjected to SYBR Green real-time PCR analysis by using the following primers: MDM2 promoter (5′-CGG GAG TTC AGG GTA AAG GT and 5′-CCT TTT ACT GCA GTT TCG). p21 promoter (5′-TGG CTC TGA TTG GCT TTC TG and 5′-TTC AGA GTA AGA GGC TAA GG). PUMA promoter (5′-CTG TGG CCT TGT GTC TGT GAG TAC and 5′-CCT AGC CCA AGG CAA GGA GGA C).

RNA Isolation and Quantitative PCR.

Total RNA was extracted by using the RNeasy Mini kit (Qiagen). cDNAs were prepared by reverse transcription of total RNA by using the SuperScript III First-Strand Synthesis System (Invitrogen). The products were used for real-time PCR by using the following primers: MDM2 (5′-CCC TTA ATG CCA TTG AAC CT and 5′-CAT ACT GGG CAG GGC TTA TT), p21 (5′-CAG ACC AGC ATG ACA GAT TTC and 5′-TTA GGG CTT CCT CTT GGA GA), PUMA (5′-AGA GGG AGG AGT CTG GGA GTG and 5′-GCA GCG CAT ATA CAG TAT CTT ACA GG), Actin (5′-GCT CGT CGT CGA CAA CGG CTC and 5′-CAA ACA TGA TCT GGG TCA TCT TCT C).

EMSA.

Oligonucleotide probe containing p53-binding site (TACAGAACATGTCTAAGCATGCTGGGGACT) labeled with IRDye 700 (LI-COR) was used for p53 EMSA analysis. p53 and GFP were produced by using the pIVEX vector and PURExpress purified E. coli in vitro translation system (New England Biolabs). FLAG-MDMX, FLAG-MDMX-S289A (CK1α phosphorylation site mutant), and FLAG-MDMX-V92W (p53 binding pocket mutant) were expressed in E. coli BL21(DE3) by using pET28 vector, purified using M2 beads, and eluted with FLAG peptide. MDMX–CK1α complex was purified from BL21(DE3) cotransformed with pET28-FLAG-MDMX and pGEX-2T-CK1α by using M2 beads and eluted with FLAG peptide. Copurification of GST-CK1α with FLAG-MDMX was confirmed by Western blot of purified FLAG-MDMX. p53 concentration in the in vitro translation system was determined by DO1 Western blot using purified GST-p53-1-82 as a standard. The EMSA reaction (20 µL of final volume) contains IRDye 700-labeled DNA probe, 100 ng of p53 (5 µL of in vitro translation product), 1–5 µg of FLAG-MDMX, 20 ng of Pab421, 10 mM Tris pH 7.5, 30 mM NaCl, 1 mM DTT, and 1 μg of Poly (dI.dC). The mixture was incubated for 25 min at 4 °C, fractionated on a 5% (wt/vol) native polyacrylamide gel, and imaged by using a Licor Odyssey Fc imager. Specificity of the DNA binding was verified by using GFP on the p53-specific probe, and using p53 on a NFkB-specific probe as negative controls.