Adenovirus E1B 55-Kilodalton Protein Is a p53-SUMO1 E3 Ligase That Represses p53 and Stimulates Its Nuclear Export through Interactions with Promyelocytic Leukemia Nuclear Bodies

Cell and virus culture, transient transfection luciferase reporter assays, and Western blotting.A549 (p53⁺), H1299 (p53⁻), and 293 (transformed with Ad5 E1A and E1B) (68) cells were cultured in Dulbecco's modified Eagle medium supplemented with penicillin-streptomycin-glutamine and 10% fetal bovine serum (Invitrogen Life Technologies). Ad5 mutants E4inORF3/dl355* (37) and 3112 (E1B-55K⁻ E4orf3⁻) (70) were propagated in 293 cells and assayed by plaque formation. For transfection luciferase assays, H1299 p53⁻ cells were plated onto 60-mm tissue culture plates 12 h before transfection with Lipofectamine 2000 (Invitrogen Life Technologies) according to the manufacturer's protocol using 100 ng of the p21-luc reporter (a gift from Xuan Liu, University of California Riverside), 100 ng of Rluc (Promega), 200 ng of pcDNA3 expression vector for p53, p53K386R, or p53K386R with K-to-R mutations in additional lysines as indicated, and 800 ng of pSRα vector for E1B-55K or empty vector (82). Firefly luciferase activity was assayed and normalized to Renilla luciferase using the Promega dual-luciferase assay system. Western blotting for E1B-55K and p53 was done with monoclonal antibodies (MAbs) 2A6 (67) and D0-1 (sc-126), respectively.

Detection of SUMO-p53 in infected and transfected cell extracts.Two 150-mm plates of A549 cells at 80% confluence were infected with the indicated Ad5 mutants, and cell extracts were prepared at 20 h postinfection (hpi). Cells were washed twice in phosphate-buffered saline (PBS) with 1 mM phenylmethylsulfonyl fluoride (PMSF) and 2 mM N-ethylmaleimide (NEM), extracted with 1 ml 5% sodium dodecyl sulfate (SDS) lysis buffer (5% SDS, 30% glycerol, 150 mM Tris [pH 6.7], 20 mM NEM, 1 mM PMSF, 2× Roche Complete EDTA-free protease inhibitor), and sonicated. Extract from 2.4 × 10⁷ cells was subjected to immunoprecipitation by dilution of the SDS lysis buffer to final concentrations of 0.5% SDS, 1% Triton X-100, 150 mM NaCl, 50 mM Tris (pH 7.4), 1 mM EDTA, 1 mM PMSF, 3 mM NEM, and 1× Roche Complete EDTA-free protease inhibitor. The diluted samples were subjected to preclearing with rabbit IgG (sc-2345) conjugated to agarose for 2 to 4 h at 4°C, followed by centrifugation and removal of the supernatant, which was then subjected to immunoprecipitation with ABCAM rabbit antiserum ab11672 to human SUMO1 bound to eBioscience Rabbit True Blot beads or rabbit IgG conjugated to agarose (sc-2345) at 4°C overnight (∼15 h) in Bio-Rad chromatography columns with nutation. Alternatively, the diluted SDS lysis buffer was subjected to preclearing with Santa Cruz mouse IgG conjugated to agarose (sc-2343), followed by immunoprecipitation as described above with anti-p53 MAb D0-1 coupled to agarose (sc-126 AC). The immunoprecipitates were washed twice in 150 mM NaCl-50 mM Tris (pH 7.4)-1 mM EDTA-1% Triton X-100, twice in 150 mM NaCl-50 mM Tris (pH 7.4), and once with 300 mM NaCl-100 mM Tris (pH 7.4). Immunoprecipitates were dissolved in 3× SDS gel loading buffer and heated to 100°C for 10 min. Fifty percent of the immunoprecipitates were resolved by 8% Tris-glycine polyacrylamide gel electrophoresis (PAGE). Western blotting detection of p53 was with MAb D0-1 (sc 126), detection of hSUMO1 was with ABCAM ab11672, and detection of SUMO2 and/or SUMO3 was with anti-SUMO2/3 MAb (84), a gift from Michael Matunis, Johns Hopkins University.

For assays of transfected H1299 and A549 cells, cells were transfected using Lipofectamine 2000 (Invitrogen Life Technologies) according to the manufacturer's protocol. H1299 cells in 10-cm plates were transfected with 8 μg pSRαE1B-55K plus 8 μg pcDNA3-p53-YFP (a gift from Xuan Liu, University of California Riverside), pcDNA3-p53K386R-YFP (see Fig. 1d), pcDNA3-p53, pcDNA3-p53K386R, or pcDNA3-p53K386R plus additional K-to-R mutations (see Fig. 5). At 24 h posttransfection, cells were lysed and subjected to immunoprecipitation by the protocol for detection of SUMO-p53 in infected-cell extracts described above or cell lysates were subjected to Western blotting without prior immunoprecipitation.

• Open in new tab
• Download powerpoint

FIG. 1.

E1B-55K induces SUMO1 modification of p53 in Ad5-infected cells. (a) A549 cells (p53⁺) were infected with Ad5 mutants (indicated at the top) with inactivating mutations in the indicated viral genes (−). (Top) Anti-p53 Western blot (WB) assay of total cell extract 20 hpi; the molecular mass marker is 55 kDa. (Bottom) Extracts were immunoprecipitated with anti-SUMO1 or control rabbit IgG, as indicated, and the immunoprecipitate (IP) was Western blotted with anti-p53. The red dot in panels a and b indicates the position of SUMO1-p53. HC is IgG heavy chain. The images come from the same exposure of Western blot assays of D0-1 and control IgG immunoprecipitates prepared at the same time. (b and c) The same extracts of infected p53⁺ A549 cells were immunoprecipitated with anti-p53 MAb D0-1, and the immunoprecipitates were Western blotted with antibody to SUMO1 (b) or SUMO2 and SUMO3 (c). The vertical red line in panel c denotes specific bands detected with the anti-SUMO2/3 MAb. (d) H1299 (p53⁻) cells were transfected with the indicated expression vectors, and whole-cell extract was prepared after 24 h and Western blotted with anti-p53 (lanes 1 to 6). In lanes 7 to 10, extracts were immunoprecipitated with anti-p53 MAb and the IP was Western blotted with anti-SUMO1 (7, 8) or anti-Flag (9, 10). (e) H1299 cells were transfected with expression vectors for E1B-55K and p53, whole-cell extract was prepared at 24 h posttransfection, 2-fold dilutions were run in neighboring lanes of an SDS gel, and the gel was subjected to Western blotting with anti-p53 MAb. Arrowheads indicate bands of p53-SUMO1 and p53 of comparable intensities.

In vitro sumoylation assay.His-myc-SUMO1, his-hUBC9, hSAE1 (with the N-terminal 10 amino acids deleted to improve expression), and His-hSAE2 were all expressed in Escherichia coli. (The His-SAE2 vector was a gift from Frauke Melchior, Max Planck Institute of Biochemistry.) His-tagged recombinant proteins were purified by Ni²⁺-nitrilotriacetic acid (NTA) column chromatography. A 10× SUMO cocktail of 0.46 mM (6.2 mg/ml) his-myc-SUMO1, 6.9 μM (0.85 mg/ml) E1, and 0.11 mM (2.2 mg/ml) his-hUBC9 in 5 mM Tris (pH 7.5), 25 mM NaCl, 5 mM β-mercaptoethanol, and 50% glycerol was used. E1B-55K-His₆ was purified from baculovirus vector-infected sf9 cells as described previously (26). Recombinant p53 was purchased from Active Motif. Sumoylation reaction mixtures of 20 μl containing 50 mM Tris (pH 8), 5 mM MgCl₂, 5 mM ATP, 2 mM dithiothreitol (DTT), 12.5 ng p53, 2 μl 10× SUMO cocktail, and the indicated amount of E1B-55K were incubated for 40 min at 30°C.

Immunofluorescence microscopy.Standard immunofluorescence microscopy without confocal imaging was performed as described previously (52). For confocal immunofluorescence microscopy, H1299 cells were plated on sterile 22-by-22-mm coverslips in six-well dishes preincubated with 25 μg/ml fibronectin (Sigma catalog no. F1141-1MG) for 15 min and washed with PBS. Cells were transfected with Lipofectamine 2000 using 200 ng pcDNA3-p53-YFP or p53K386R-YFP and 800 ng pSRαE1B-55K where indicated 24 h before fixation with 2% formaldehyde for 15 min at room temperature or overnight at 4°C, washed with PBS, permeabilized in PBS plus 1% Triton X-100 for 15 min, rinsed three times in PBS plus 0.1% Tween 20, and incubated for 30 to 40 min in PBS-5% bovine serum albumin (BSA)-1% Tween 20. Coverslips were then incubated for 1 h at room temperature in primary antibody diluted in PBS-5% BSA-1% Tween 20 and then incubated for 5 min with each of four successive washes of PBS plus 0.1% Tween 20. Slides were then incubated for 1 h in the dark in secondary antibody diluted in PBS plus 0.1% Tween 20, washed four times for 5 min each in PBS plus 0.1% Tween 20, stained with 0.5 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) for 10 to 15 min, rinsed with PBS, and mounted onto slides using Vectashield mounting medium (Vector Laboratories, catalog no. H-1400). MAb 2A6 was used as the primary antibody against E1B-55K with a secondary Texas Red donkey anti-mouse IgG (715-075-150) from Jackson ImmunoResearch Labs (false colored green). The primary antibody for PML was goat anti-PML (sc-9862) with a secondary Cy3 donkey anti-goat IgG (705-165-147) from Jackson ImmunoResearch Labs (false colored red). Confocal fluorescence microscopy was with a Leica TCS SP2 AOBS single-photon confocal microscope using a 63× 1.4 numerical aperture oil immersion objective.

FRAP.H1299 cells were transfected as described above for immunofluorescence microscopy. The microscope stage and objective were placed in an enclosed chamber preequilibrated to 37°C with a continuous flow of 5% CO₂ over the microscope stage. The laser settings were fly mode at 1,000 Hz, 8× zoom, and an image format of 512 by 256. Fluorescence emission from enhanced YFP was detected with the range of photomultiplier tube 2 set to 525 to 600 nm (with a maximal gain of 680). Bleaching of a 3-μm-diameter circle was performed with laser lines 405, 458, 488, and 514 nm set to maximal power. The average frame rate for prebleaching (10 frames), bleaching (4 frames), and initial postbleaching (16 frames) was 1 frame/0.208 s. The frame rates for additional postbleaching imaging were 30 frames (1 frame/4 s), followed by 50 frames (1 frame/10 s) for monitoring of longer recovery times. Additional experiments were performed with different settings to show that recovery rates were similar independent of the laser lines used for bleaching or the times used for data collection postbleaching. The settings for these experiments were fly mode at 1,000 Hz, 8× zoom, an image format of 512 by 256, and an averaging of 2. Fluorescence emission from YFP was the same as that described above. Bleaching of a 3-μm-diameter circle was performed with laser line 514 nm only set to maximal power. The frame rate for prebleaching (5 frames), bleaching (3 or 4 frames), and postbleaching (2 frames) was 1 frame/0.416 s). The frame rates for additional postbleaching imaging were 30 frames (1 frame/3 s) and 60 frames (1 frame/6 s) or 30 frames (1 frame/3 s) and 100 frames (1 frame/6 s) for monitoring of long recovery times.

Separate FRAP experiments were performed for each construct as indicated, and then the normalized fluorescence from each time point was averaged to generate a single FRAP curve: p53-YFP only (n = 3 experiments), p53-YFP and E1B-55K (n = 4 experiments), p53-YFP and PML-IV (n = 3), p53-YFP with PML-IV and E1B-55K (n = 7), p53K386R-YFP with PML-IV and E1B-55K (n = 12), and p53L344P-YFP and E1B-55K (n = 4). Analyses were performed on two or three separate transfections per construct. Average background values were determined outside cells (region of interest 2 [ROI₂]) and subtracted from all of the raw data points of the bleach region (ROI₁), as well as inside the nucleus outside the bleach area (ROI₃) before further analysis as follows (62): B_n = (ROI₁ − ROI₂)/(ROI₃ − ROI₂), P_b = average of prebleach signal calculated in B_n, and normalized fluorescence intensity (FI_N) = B_n/P_b. These data were then normalized based on the bleach depth, generating curves set to zero (based on the first point after bleaching) to the prebleach intensity, calculated as follows: FI_N − FI_{initial point postbleaching}.

The time starting at the first point postbleaching was set to time zero. Recovery curves report both on the exchange of bound molecules and on the faster exchange of unbound molecules in the bleached volume and are therefore biphasic and must be fitted by a double-exponential equation. Equations were derived from reference 71 and references therein as follows: single exponential − y(t) = A(1 − e^−k1t) and double exponential − y(t) = A(1 − e^−k1t) + B(1 − e^−k2t), where A is the amplitude of the fast phase and B is the amplitude or plateau of recovery of the slow phase, t is time, and k₁ and k₂ are the rate constants for the fast and slow phases of recovery, respectively, derived from curve fitting with Kaleidagraph software. Half-times of recovery were calculated from the fitted rate constants as follows: t_1/2 = ln 0.5/−k. The t_1/2 reported always represents the slow phase from a double-exponential fit. The data averaging for constructs expressing p53-YFP with PML-IV and E1B-55K were noisy after 180 s, and that for p53K386R-YFP with PML-IV and E1B-55K were noisy after 350 s due to cell movement. Therefore, reported half-times for these constructs were estimated based upon extrapolation of data to full recovery.

Gel filtration analysis of E1B-p53 complex.Samples of 1.5 μg of purified rE1B-55K and 1.5 μg of purified p53 (Active Motif) were incubated separately or together at room temperature for 30 min in 300 mM KCl-20 mM HEPES (pH 7.4)-1 mM DTT-0.2 mM EDTA-20% glycerol. The three samples then were separately subjected to gel filtration on a 10-ml Superose 6 column in the same buffer. Fractions of 0.5 ml were collected, and fractions 10 to 37 were subjected to trichloroacetic acid precipitation and analyzed by Western blotting. Dextran blue 2000 eluted in fraction 10 (the void volume), and the peak of thyroglobulin (669 kDa) eluted in fraction 21.8, that of ferritin (440 kDa) eluted in fraction 25.4, that of alcohol dehydrogenase (150 kDa) eluted in fraction 28, and that of BSA (68 kDa) eluted in fraction 32.

E1B-55K stimulates sumoylation of p53 during viral infection.Muller and Dobner (57) reported that transfection of p53⁻ H1299 cells with expression vectors for p53 and Ad5 E1B-55K resulted in p53 conjugation to SUMO1, in contrast to H1299 cells transfected with a p53 expression vector alone. To determine if E1B-55K also stimulates p53 sumoylation during Ad5 infection at the normal cellular concentrations of the proteins involved, we assayed this modification of p53 in extracts from virus-infected p53⁺ A549 cells by immunoprecipitation, followed by Western blotting. When both E1B-55K and E4orf6 are expressed, they assemble with components of the cellular cullin 5 class of E3 ubiquitin ligases to form a multisubunit ubiquitin ligase that polyubiquitinates p53 (33, 63), leading to its proteasomal degradation (64, 73). Consequently, to observe modified versions of p53 induced by Ad5 infection, we infected cells with viruses mutant in the viral E1B-55K or E4orf6 protein required for formation of the viral ubiquitin ligase. Cell extracts were prepared under conditions (5% SDS, 20 mM NEM) that potently inhibit the high activities of desumoylating enzymes present in cell extracts. We observed p53 sumoylation during viral infection most consistently in cells infected with Ad5 mutant in E4orf6 and E4orf3 (E4inORF3/dl355* [37], Fig. 1a, bottom, and b). During infection with wild-type (WT) Ad5, E4orf3 is localized to PML-nb along with E1B-55K (48). Mutation of E4orf3 may have increased the ability to detect sumoylated p53 during infection because sumoylated proteins are enriched in PML-nb and the period of association with PML-nb during viral infection of another E1B-55K bound host cell protein, MRE11, is greatly increased by mutation of E4orf3 (52).

A549 cells infected with an E4orf3 E1B-55K double mutant, 3112 (70), accumulated p53 to levels similar to those observed following infection with a E4orf3 E4orf6 double mutant (Fig. 1a, top) because of the inability of each of these mutants to generate the viral ubiquitin ligase that requires both E1B-55K and E4orf6 and polyubiquitinates p53, targeting it for proteasomal degradation (33, 63). But in the E4orf3⁻ background, the level of SUMO1-modified p53 was clearly higher in the E1B-55K⁺ mutant than in the E4orf6⁺ mutant (Fig. 1a and b). Immunoprecipitation with anti-p53, followed by Western blotting with either anti-SUMO1 antibody or a MAb that recognizes both of the other two closely related major human SUMOs, SUMO2 and SUMO3 (84), showed an E1B-55K-induced increase in p53 conjugation to SUMO1, but not to SUMO2 or SUMO3 (Fig. 1b and c). Infection with both mutant viruses increased the level of SUMO2/3-modified p53 to an extent similar to that seen in mock-infected cells. These results indicate that, in addition to E1B-55K's ability to stimulate SUMO1 modification of p53 in transfected cells (57), E1B-55K can also stimulate p53 sumoylation during viral infection at the normal cellular concentrations of all of the components involved in the process. As for other sumoylated transcriptional regulators, the low level of SUMO1-p53 observed in virus-infected cells was probably a consequence of transient sumoylation resulting from the rapid removal of SUMO1 shortly after its addition (9, 34).

Experiments (described below) using fluorescence microscopy techniques were performed with a p53-YFP fusion. To determine if E1B-55K induces sumoylation of p53-YFP comparably to p53 not fused to YFP, expression vectors for p53-YFP and E1B-55K were cotransfected into p53⁻ H1299 cells. Twenty-four hours posttransfection, whole-cell extracts were prepared under conditions that prevent the loss of SUMO1 from SUMO1-modified proteins and analyzed by Western blotting with an anti-p53 MAb. A band with the mobility expected for p53-YFP conjugated to endogenous cellular SUMO was observed (Fig. 1d, lane 2). That this band corresponds to SUMO1-modified p53 was demonstrated by the observation of a band with the same mobility detected following immunoprecipitation with anti-p53 MAb and Western blotting using SUMO1-specific antibody (lane 8). Further, a doublet near this position was observed when expression vectors for FLAG-tagged SUMO1 and E1B-55K were cotransfected with the YFP-p53 expression vector (Fig. 1d, lane 4). When extract from these cells was immunoprecipitated with anti-p53 MAb and Western blotted with an anti-Flag MAb, the upper band of this doublet was shown to correspond to p53-YFP conjugated to Flag-tagged SUMO1 (lane 10). We conclude that E1B-55K stimulates conjugation of p53-YFP to SUMO1 as it does for untagged p53 in Ad5-infected cells (Fig. 1a to c) and in transiently transfected cells (57). Moreover, the fraction of p53-YFP sumoylated (Fig. 1d, lane 2) was similar to that reported for p53 in a comparable cotransfection experiment (57) and as measured quantitatively (Fig. 1e).

The p53 protein contains a single previously determined site of sumoylation in a SUMO site consensus sequence at K386 (ψKXE, where ψ is a hydrophobic amino acid and X is any amino acid) (30, 65). Nonetheless, we observed that p53K386R-YFP mutated at this site continued to be sumoylated in transfected cells, although the extent of sumoylation was reduced compared to that of WT p53-YFP (Fig. 1d, lane 6). To quantitate the difference in the extent of E1B-55K-induced sumoylation of p53 and p53K386R, extracts from p53⁻ H1299 cells transfected with expression vectors for these proteins were subjected to serial 2-fold dilutions and Western blotting with anti-p53 MAb (Fig. 1e). The intensity of the signal observed for SUMO1-p53 after four 2-fold serial dilutions was similar to the intensity of unmodified p53 detected on the same Western blot after 10 2-fold dilutions, indicating that the fraction of SUMO1-modified p53 at steady state was 2⁻¹⁰/2⁻⁴, about 1.5%. The intensity of the signal observed for SUMO1-p53K386 after 3 2-fold serial dilutions was similar to the intensity of unmodified p53K386R after 10 2-fold dilutions, indicating that the fraction of SUMO1-modified p53K386R was about 0.8%. A similar fraction of SUMO1 conjugation was observed for p53-YFP and p53K386R-YFP (data not shown).

E1B-55K is a SUMO1 E3 ligase for p53.Since E1B-55K stimulates p53 sumoylation in vivo in transiently transfected and Ad5-infected cells, we hypothesized that E1B-55K may function as a SUMO1 E3 ligase and tested for this activity in vitro. We used E. coli-produced, purified human His₆-myc-SUMO1 (recombinant SUMO1 [rSUMO1]), SAE1/His₆-SAE2 (rSAE1/SAE2, human SUMO E1), His₆-UBC9 (rUBC9, human SUMO E2), recombinant p53, and baculovirus vector-produced, purified E1B-55K-His₆ (rE1B-55K). Addition of rE1B-55K stimulated ligation of rSUMO1 to recombinant p53 in proportion to the amount of rE1B-55K added (Fig. 2a). Thus, this E1B-55K preparation had E3 SUMO1 ligase activity for p53. The stimulation of p53 sumoylation produced by E1B-55K was not due to a contaminating activity from insect cells, since eliminating rE1B-55K, E1 (rSAE1/SAE2), or E2 (rUBC9) eliminated the stimulation of p53 sumoylation observed in the complete system (Fig. 2b). Also, a control protein fraction prepared from sf9 cells infected with an empty baculovirus vector prepared in parallel with a second preparation of rE1B-55K had no SUMO1-p53 ligase activity (Fig. 2c). The very low level of SUMO1-p53 observed in the control lane and lane 3 of Fig. 2b was due to the ability of UBC9 to direct a very low level of sumoylation in vitro at a consensus sumoylation site in the absence of an E3 SUMO ligase (34).

• Open in new tab
• Download powerpoint

FIG. 2.

Adenovirus E1B-55K is a SUMO1-p53 ligase. (a) The indicated amount of baculovirus-expressed Ad2 E1B-55K-His₆ purified by Ni²⁺-NTA column chromatography as described previously (55) (rE1B-55K) was incubated at 30°C for 40 min in sumoylation buffer (including ATP) plus purified, E. coli-expressed, Myc-tagged human SUMO1, human SAE1/SAE2 (SUMO E1), human UBC9 (SUMO E2), and human p53 (r p53) for 40 min at 30°C. Reaction products were analyzed by Western blotting with anti-p53 MAb. (b) Two-hundred-nanogram samples of recombinant E1B-55K (rE1B-55K), rSUMO1-Myc, recombinant p53, SUMO E1 (rSAE1/SAE2), and SUMO E2 (rUBC9), as indicated (+, −), were incubated and analyzed as described for panel a. (c) A second preparation of rE1B-55K (∼100 μg/ml) was prepared in parallel with a control protein fraction from sf9 cells infected with the empty baculovirus vector and taken through the same purification steps. BSA (100 μg/ml) was added to the final control fraction, and 2 μl of each was added to reaction mixtures containing the other components shown in panel b. (d) Silver-stained SDS-PAGE of aliquots of fractions 23 to 33 from a Superose 6 column of another preparation of rE1B-55K. (e) Left, dilution series of rE1B-55K. Right, aliquots of fractions 23 to 33 from the Superose 6 column assayed on the same Western blot with anti-E1B-55K MAb. (f) The same volume of aliquots from each Superose 6 column fraction was incubated, and reaction products were assayed by Western blotting as described for panel a. The lane at the left (**) contained the same amount of recombinant p53 before incubation to show background bands (*) in this preparation detected with the anti-p53 MAb. (g) Same as panel e but blotted with anti-Myc antibody to show bands of p53-SUMO1-Myc.

To further test if the rE1B-55K purified from insect cells was associated with a contaminating insect cell activity that stimulated p53 sumoylation, we purified the rE1B-55K protein fraction further by gel filtration chromatography on a Superose 6 column. Column fractions analyzed by SDS-PAGE and silver staining (Fig. 2d) and Western blotting (Fig. 2e) revealed elution of E1B-55K as expected for the dimer of native E1B-55K (54), with a peak in fractions 28 to 31. In vitro p53-SUMO1 ligase activity of each column fraction was assayed with E. coli-expressed recombinant E1, E2, Myc-tagged SUMO1, and recombinant p53. A peak of p53-SUMO1 E3 ligase activity coeluted with the peak of rE1B-55K in fractions 28 to 31 as assayed by Western blotting with anti-p53 (Fig. 2f) and anti-Myc (Fig. 2g) MAbs. This activity did not coelute with two polypeptides of ∼65 and ∼75 kDa that peaked in fraction 28 as detected by silver staining at much weaker intensity than the E1B-55K band or the weakly staining band of ∼30 kDa eluting in fractions 29 to 33 but rather followed the elution profile of the peak of rE1B-55K. If the p53-SUMO1 activity detected in Fig. 2a had been due to a complex formed by rE1B-55K and insect cell SUMO1 E3 or E2, the activity would have eluted at a higher molecular weight than the peak of rE1B-55K in fractions 28 to 31. This coelution on a gel filtration column of p53-SUMO1 E3 ligase activity with rE1B-55K detected by protein staining demonstrates that the activity is intrinsic to E1B-55K.

E1B-55K tethers p53 in PML-nb.To address the question of whether E1B-55K inactivates p53 in part by tethering it in PML-nb, we analyzed the influence of E1B-55K on the intranuclear mobility of p53-YFP in living cells by FRAP, a well-established method for determining the stabilities of macromolecular interactions in vivo (20). H1299 cells (p53⁻) were transfected with an expression vector for p53-YFP plus or minus an E1B-55K expression vector. First, we analyzed the subcellular localization of these proteins, as well as PML, the principal cellular component of PML-nb, by confocal microscopy of fixed cells 24 h after transfection. Direct detection of YFP fluorescence and immunostaining with anti-PML antibody (red) and anti-E1B-55K MAb (green) allowed the visualization of all three proteins in the same cells (Fig. 3a to c). Nuclei were visualized by staining with DAPI. In the absence of E1B-55K, p53-YFP was distributed throughout the nucleoplasm and did not colocalize with PML in PML-nb (Fig. 3a). When E1B-55K was coexpressed with p53-YFP, in ∼50% of the cells, p53-YFP was colocalized with E1B-55K and endogenous PML in structures with a size and morphology typical of PML-nb, as well as in cytoplasmic inclusions with E1B-55K (Fig. 3b). In ∼5% of the cells, p53-YFP and E1B-55K colocalized with PML in PML-nb but no cytoplasmic p53-YFP was observed. In ∼45% of the cells, p53-YFP was observed only in large cytoplasmic inclusions bodies with E1B-55K (see Fig. 6e). By 48 h posttransfection, p53-YFP was observed only in cytoplasmic inclusions with E1B-55K in ∼65% of the cells. In the absence of a cotransfected expression vector for p53-YFP, at 24 h posttransfection, E1B-55K colocalized with PML-nb in ∼50% of the cells, demonstrating that p53 is not required for E1B-55K association with these subnuclear domains in p53⁻ H1299 cells (data not shown). The relative amounts of PML, E1B-55K, and p53-YFP in these nuclear foci varied, so that the overlay of all three colors (Fig. 3b) produced foci with different shades of color from red (mostly PML) to green (mostly E1B-55K) to yellow (mixtures of the three proteins). But the vast majority of p53-YFP foci in nuclei colocalized with both PML and E1B-55K. The E1B-55K-induced p53-YFP association with PML-nb requires direct binding of E1B-55K to p53 since it was not observed with a YFP fusion to p53 with a single amino acid substitution that inhibits E1B-55K binding (P27Y [50]) (Fig. 3c). Note that E1B-55K continued to associate with PML-nb in the absence of an interaction with P27Yp53. As in the case of cells coexpressing WT p53-YFP and E1B-55K (Fig. 3b), the relative amounts of PML and E1B-55K in these nuclear foci varied, so that the overlay produced foci with different shades of color from red (mostly PML) to green (mostly E1B-55K) to yellow (equal staining of PML and E1B-55K). But the vast majority of E1B-55K foci in nuclei colocalized with PML (Fig. 3c).

• Open in new tab
• Download powerpoint

FIG. 3.

E1B-55K-induced tethering of p53 in PML-nb. H1299 cells (p53⁻) were transfected with expression vectors for the indicated proteins and fixed at 24 h posttransfection. Confocal fluorescence microscopy revealed PML and E1B-55K immunostaining (red and green, respectively), YFP fluorescence (yellow), and DAPI staining (blue). (a) p53-YFP alone. (b) p53-YFP and E1B-55K. DAPI staining is shown for the overlay panel only. Arrowheads indicate E1B-55K inclusion bodies in the cytoplasm. (c) P27Yp53-YFP and E1B-55K. The overlay includes DAPI staining. The scale is the same as in panel a. (d) Examples of representative FRAP assays. Only the region of the cell subjected to photobleaching (red circle) is shown prebleaching and at the indicated times postbleaching. (e) Plots of normalized recovery of YFP fluorescence versus time postbleaching for p53-YFP expressed alone (black circles) and with E1B-55K (red circles). (f) Plots of normalized recovery of YFP fluorescence versus time postbleaching in nuclear foci in cells cotransfected with expression vectors for PML-IV and p53-YFP (green diamonds); PML-IV, p53-YFP, and E1B-55K (blue diamonds); and PML-IV, p53K386R-YFP, and E1B-55K (red diamonds).

During adenovirus infection, E1B-55K first appears in PML-nb (19, 48). In cells infected with an E4orf6 mutant where p53 is stabilized, p53 first accumulates with E1B-55K in PML-nb at 8 to 12 hpi, before it is observed in cytoplasmic inclusions in an increasing fraction of cells following 16 hpi (52). Consequently, we interpret E1B-55K and p53-YFP in PML-nb (Fig. 3b) to be intermediates in the nuclear export of E1B-55K-p53-YFP complexes. As discussed below, consistent with this interpretation, we found that in transfected p53⁺ A549 cells, E1B-55K mutated in its sumoylation site was greatly inhibited both in associating with PML-nb and in stimulating p53 nuclear export.

To analyze the dynamics of p53-YFP association with PML-nb directly, we analyzed the mobility of p53-YFP in living cells by FRAP. The t_1/2 for fluorescence recovery of nucleoplasmic p53-YFP expressed alone was ∼1 s (Fig. 3d and e, black circles). This is similar to the t_1/2 of other nucleoplasmic proteins and indicates very rapid mobility throughout the nucleoplasm for the free transcription factor (29). Coexpression of E1B-55K led to the appearance of most of the p53-YFP in nuclear foci in living cells (Fig. 3d, prebleach), presumably associated with PML-nb, as observed in fixed and immunostained transfected cells (Fig. 3b). FRAP of p53-YFP in these foci revealed a reproducible t_1/2 of ∼85 s. Consequently, the E1B-55K-induced association of p53-YFP with PML-nb decreased the mobility of p53-YFP in the nucleus by nearly 2 orders of magnitude. This tethering of p53 in PML-nb likely makes a significant contribution to the inhibition of p53 transcriptional activation function by E1B-55K (82) by restricting its ability to interact with p53-binding sites in target gene control regions.

E1B-55K-induced SUMO1 modification of p53 contributes to p53 tethering in PML-nb and to inhibition of p53 transcriptional activation function.To determine if the low level of p53-SUMO1 modification induced by E1B-55K contributes to E1B-55K-induced tethering of p53 in PML-nb, we used FRAP assays to compare the tethering of WT p53-YFP and p53K386R-YFP. This mutation of the previously determined site of p53 sumoylation at a SUMO modification consensus sequence (30, 65) did not eliminate the low steady-state level of p53 conjugation to SUMO1 induced by E1B-55K (57) but decreased it by a factor of ∼2 (Fig. 1e). The influence of this decrease in the steady-state level of p53-SUMO1 modification on tethering in PML-nb was most apparent in cells in which tethering was increased further by overexpression of the isoform of PML that binds p53. There are several PML isoforms expressed from alternatively spliced PML mRNAs (38). p53 interacts with isoform PML-IV, and transfection of a PML-IV expression vector causes p53 to become partially colocalized with PML-nb (6, 24). Consistent with these observations, we also observed partial colocalization of p53-YFP with PML-nb in H1299 cells cotransfected with expression vectors for p53-YFP and PML-IV and an increase in the t_1/2 of p53-YFP from ∼1 s (Fig. 3e) to ∼15 s in the presence of overexpressed PML-IV (Fig. 3f). When E1B-55K was coexpressed with p53-YFP and PML-IV, the t_1/2 of p53-YFP was decreased further to ∼250 s (Fig. 3f). In contrast, the t_1/2 of p53K386R-YFP in H1299 cells cotransfected with vectors for PML-IV and E1B-55K was ∼80 s. Consequently, under these conditions, a decrease in the steady-state level of SUMO1 conjugation to p53 from 1.5% for WT p53-YFP to 0.8% for p53K386R-YFP led to an ∼3-fold decrease in the stability of the E1B-55K-induced association of p53 with PML-nb. These results indicate that E1B-55K-induced sumoylation of p53 contributes to the E1B-55K-induced sequestration of p53 in PML-nb.

To analyze the influence of the reduced level of p53K386R sumoylation on the ability of E1B-55K to repress the activation function of p53, we performed luciferase reporter assays with p53⁻ H1299 cells. The K386R mutation generated ∼120% of the luciferase activity induced by the transfection of an equal amount of expression vector for WT p53 from a cotransfected reporter with the p21^CIP control region. To compare the extent of repression by a cotransfected E1B-55K expression vector, we normalized the level of luciferase activity induced by WT p53 and p53K386R to 1 and plotted the normalized activity of luciferase in cells with coexpressed E1B-55K (Fig. 4a). In six independent transfection assays performed on 3 separate days, we observed a reproducible decrease in the extent of E1B-55K repression of p53K386R compared to that of WT p53. This is consistent with the reduced tethering in PML-nb of p53K386R-YFP compared to that of WT p53-YFP (Fig. 3f).

• Open in new tab
• Download powerpoint

FIG. 4.

p53 sumoylation is required for maximal p53 repression by E1B-55K. (a) H1299 cells were transfected with a p21^CIP promoter luciferase reporter and expression vectors for the indicated proteins, and luciferase activity in extracts was determined at 24 h posttransfection. Luciferase expression induced by p53K386R was reproducibly ∼20% higher than for WT p53 in the absence of E1B-55K. In these plots, luciferase activities for WT p53 and p53K386R in the absence of E1B-55K were normalized to 1.0 to facilitate comparison of E1B-55K inhibition of transcriptional activation by p53 and p53K386R. Error bars indicate the standard deviation of the mean for six independent assays. (b) Luciferase activity induced by expression vectors for the indicated p53 mutants, normalized to expression of Renilla luciferase from a cotransfected expression vector. Except for the K386R mutant p53 protein, each of the mutant p53 proteins included the K386R mutation and mutations in the additional lysines indicated. +E1B-55K indicates cotransfection with the E1B-55K expression vector. The 6KR mutant protein contains mutations K370R, K372R, K373R, K381R, K382R, and K386R.

E1B-55K induces sumoylation at multiple lysines in mutant p53K386R.These data support a model for maximal repression of p53 that requires E1B-55K-induced sumoylation at p53 lysine 386, which helps tether p53 at PML-nb in the presence of E1B-55K. We postulated that if one could abolish sumoylation of p53 completely, this might further decrease E1B-55K-mediated inhibition of p53 transcriptional activity. Therefore, we made mutations of every additional lysine in p53 in combination with the K386R mutant to determine if we could map a second sumoylation site in p53 that continued to be sumoylated in the p53K386R mutant. There are a total of 20 lysines in p53. Using the K386R-p53 mutant construct as a template, we made single or multiple mutations of all of the other lysines in p53, as shown in Fig. 5a. For simplicity, each construct was labeled for the new mutation but also contained the K386R mutation; K24R was K24R plus K386R-p53, etc. We transfected H1299 p53⁻ cells with expression vectors for E1B-55K and each of these p53 mutants with arginine substitutions for two or more lysines, prepared lysates under conditions that prevent desumoylation, and performed Western blotting using anti-p53 MAb on 2-fold serial dilutions of the lysates (Fig. 5b). For each construct, a band at the position of SUMO1-p53 was evident, although a smaller fraction of mutant K132R K139R K386R was sumoylated compared to the other mutants and therefore required a longer exposure of the Western blot to reveal a band of this mobility clearly. A vector for p53 mutated in the six C-terminal lysines was also prepared (6KR, with the K370R, K372R, K373R, K381R, K382R, and K386R mutations). This mutant was also modified to an extent similar to that of the K386R single mutant (Fig. 5b). Therefore, there is no major single secondary sumoylation site in p53 when K386 is mutated; E1B-55K can induce sumoylation of p53 at a variety of lysines. In this regard, E1B-55K is similar to the MDM2 p53 ubiquitin ligases that can polyubiquitinate p53 at a number of lysines (59). Similarly, no single p53 lysine is required for human papillomavirus E6-induced p53 degradation (14a). Consequently, it was not possible to determine by mutagenesis of additional p53 lysines whether elimination of p53 sumoylation would further affect E1B-55K inhibition of p53 function over that observed for mutation of K386, the major site of p53 sumoylation identified previously (30, 65).

• Open in new tab
• Download powerpoint

FIG. 5.

No single secondary site of E1B-55K-induced sumoylation of p53K386R. (a) Domain map of p53 and positions of lysine residues. TAD, transcriptional activation domain; PRR, proline-rich region; TetD, tetramerization domain; CT, C-terminal domain. (b) H1299 (p53⁻) cells were transfected with expression vectors for E1B-55K and WT p53, p53 K386R, or p53 K386R plus mutations in the other lysines indicated. p53 6KR contains mutations K370R, K372R, K373R, K381R, K382R, and K386R. Whole-cell extracts were prepared after 24 h posttransfection, and 2-fold serial dilutions were subjected to Western blotting with anti-p53 MAb. Blank lanes were included to help distinguish the blots for the individual mutants. When the blank lane was flanked by samples with a high concentration of unmodified p53, a band appeared due to sample spillover from the neighboring lanes. The arrows indicate the sumo-modified form of the p53 mutant K132R, K139R.

Each of these p53 double- or multiple-lysine mutants was assayed for the ability to activate luciferase expression from the p21-luciferase reporter and inhibition of this activity by cotransfection of the E1B-55K expression vector (Fig. 4b). Each mutant stimulated luciferase expression at least as well as WT p53 or up to 2-fold more, with the exception of the K132R K139R K386R triple mutant. This is consistent with an earlier report that K132 is required for activation by p53 (14a). Since this mutant is sumoylated to a lesser extent than other p53 mutants (Fig. 5b), it is possible that sumoylation or another modification of K132 is required for p53 activity. Activation by each of the active lysine double and multiple lysine-to-arginine mutants was inhibited by E1B-55K to a reduced extent compared to WT p53, comparable to the situation observed for p53K386R (Fig. 4b).

Sumoylation of E1B-55K is required for its maximal association with PML-nb and efficient p53 nuclear export and inhibition.Although we were not able to eliminate sumoylation of p53 by mutation of multiple p53 lysines, mutation of the single known site of E1B-55K sumoylation at K104 within a typical sumoylation consensus sequence (VKRE) eliminates E1B-55K sumoylation and greatly inhibits its ability to repress transcriptional activation by p53 and therefore its ability to cooperate with adenovirus E1A in the transformation of primary baby rat kidney cells (22). The K104R mutation also inhibits the ability of E1B-55K to stimulate SUMO1 conjugation to p53 (57). The E1B-55K K104R mutation has also been reported to inhibit E1B-55K nuclear localization (22). But since nuclear E1B-55K is associated with PML-nb (19, 48) (Fig. 3b) and mutation of the sumoylation site interferes with PML-nb association for many PML-associated proteins (34), we wondered if the failure to observe nuclear E1B-55K K104R might be due to its inability to associate with PML-nb rather than to a defect in nuclear import. WT E1B-55K is known to shuttle between the nucleus and cytoplasm (18, 45), and mutation of its nuclear export signal (NES) leads to its association primarily with nuclear PML-nb (21). Consequently, to determine if the E1B-55K K104R mutation influences the association of nuclear E1B-55K with PML-nb, we analyzed the subcellular localization of a double mutant in both the NES and the nearby sumoylation site at K104 (43). This was done by confocal microscopy of transiently transfected p53⁻ H1299 cells with expression vectors for p53-YFP and an E1B-55K NES mutant (21) or the K104R mutant in the background of the NES mutation (43) (Fig. 6a and b). As reported earlier (21), mutation of the E1B-55K NES resulted in localization of the mutant E1B-55K to nuclear PML-nb, with no E1B-55K NES mutant in cytoplasmic inclusions in transiently transfected cells (Fig. 6a). Results for the E1B-55K NES K104R double mutant were somewhat different than previously reported for A549 cells infected with an Ad5 mutant virus with this E1B-55K double mutation (43). This is probably due to the absence of nuclear viral replication centers since, in virus-infected cells, E1B-55K moves from PML-nb to these replication centers once viral DNA replication commences (19, 60). Also, E4orf3 expressed in virus-infected cells influences the subcellular localization of E1B-55K (1, 52). In the absence of other viral proteins in the transiently transfected cells, the double mutant was localized throughout the cytoplasm and in large cytoplasmic inclusions similar to those formed by WT E1B-55K (Fig. 6b). The double mutant also formed nuclear foci with bound p53. However, the NES K104R double mutant associated with PML-nb to a greatly reduced extent compared to the E1B-55K NES mutant (compare Fig. 6a and b). This result is consistent with a requirement for E1B-55K sumoylation for efficient association with PML-nb, as observed for many PML-nb-associated sumoylated proteins (34).

• Open in new tab
• Download powerpoint

FIG. 6.

p53 tethering in PML-nb stimulates E1B-55K-induced p53 nuclear export. (a) H1299 p53⁻ cells were transfected with expression vectors for p53-YFP and E1B-55K with a triple mutation in its previously mapped NES (21). Cells were fixed, immunostained for the indicated proteins, and visualized by confocal microscopy as described in the legend to Fig. 3a. (b) H1299 cells were transfected with expression vectors for p53-YFP and E1B-55K mutant in both its NES and the K104R sumoylation site and visualized as described for panel a. (c) A549 cells were transfected with expression vectors for WT E1B-55K (top row) or E1B-55K K104R (bottom row). At 14 h posttransfection, they were treated with leptomycin B (LMB) and then fixed at 24 h posttransfection, immunostained for the proteins indicated at the top, and visualized by standard fluorescence microscopy. (d) A549 cells were transfected with expression vectors for WT E1B-55K and E1B-55K K104R, fixed at 24 h posttransfection, immunostained for the proteins indicated at the top, and visualized by standard fluorescence microscopy. (e) One hundred cells from the experiment shown in panel d were scored for complete export of p53 to the cytoplasm as in the middle parts of panel d, partial export of p53 to the cytoplasm as in the top parts of panel d for WT E1B-55K and the bottom parts of panel d for E1B-55K K104R, or for no export of p53 to the cytoplasm.

Consistent results also were observed for E1B-55K and E1B-55K K104R in p53⁺ A549 cells treated with the covalent inhibitor of exportin 1 (CRM1) required for nuclear export of WT E1B-55K (18, 45) (Fig. 6c). A549 cells transfected with expression vectors for WT E1B-55K or E1B-55K K104R were treated with leptomycin B at 14 h posttransfection. When the cells were fixed and immunostained for E1B-55K and PML (Fig. 6c, left) and for E1B-55K and p53 (Fig. 6c, right) at 24 h posttransfection and analyzed by standard fluorescence microscopy, WT E1B-55K was entirely localized to intranuclear foci together with PML and p53. In contrast, E1B-55K K104R was also primarily localized in nuclear foci with p53 (Fig. 6c, right) but most of these foci were not associated with PML (Fig. 6c, left).

When p53⁺ A549 cells were transfected with expression vectors for WT E1B-55K or the E1B-55K K104R mutant in the absence of leptomycin B and stained for E1B-55K and p53, it was evident that WT E1B-55K induced the nuclear export of p53 far more efficiently than the E1B-55K K104R mutant that is inhibited from associating with PML-nb (Fig. 6d and e). These results imply that the E1B-55K K104R mutant fails to inhibit transcriptional activation by p53 (22) because it fails to associate with PML-nb where WT E1B-55K causes p53 to become sequestered (Fig. 3b and d). The results of Fig. 6d also imply that E1B-55K induction of p53 nuclear export is greatly enhanced by the interaction of E1B-55K with PML-nb.

An E1B-55K-p53 network contributes to p53 tethering in PML-nb.The E1B-55K NES K104R double mutant and p53-YFP in nuclei continued to colocalize in foci even when they were not associated with PML-nb (Fig. 6b). Similarly, E1B-55K K104R localized to nuclei by treatment with leptomycin B colocalized in foci with endogenous p53 in A549 cells (Fig. 6c, right) even though most foci were not associated with PML-nb (Fig. 6c, left). In considering what molecular interactions might cause such colocalization of p53 and E1B-55K in foci containing large numbers of proteins, we realized that p53, which is a tetramer (39), might form a network cross-linked by dimers of E1B-55K (54) which bind to the N-terminal activation domain region of each p53 monomer (40, 50) (Fig. 7a). Since the p53 N-terminal regions extend from the p53 tetramerization domain with approximately tetrahedral symmetry (39), cross-linking of p53 tetramers by E1B-55K dimers might result in a three-dimensional lattice composed of multiple E1B-55K dimers and p53 tetramers (diagrammed in two dimensions in Fig. 7b). This could explain the association of E1B-55K and p53 into large intranuclear foci in the absence of association with any other nuclear domain such as PML-nb, as observed for the E1B-55K NES K104R double mutant (Fig. 6b) and for E1B-55K K104R in leptomycin B-treated cells (Fig. 6c).

• Open in new tab
• Download powerpoint

FIG. 7.

Multiprotein E1B-55K-p53 molecular network contributes to p53 tethering in PML-nb. (a) Diagram a p53 tetramer (from reference 39) bound by one monomer of the elongated E1B-55K dimer (54). (b) Diagram of a three-dimensional network of E1B dimers bound to p53 tetramers with a small percentage of molecules covalently linked to SUMO1 (red *). (c) Superose 6 gel filtration of purified recombinant E1B-55K and p53 incubated alone or together, as indicated. Column fractions were assayed by immunoblotting for E1B-55K or p53, as indicated. The white asterisk indicates a truncated proteolytic product of p53 present in the purified p53 preparation. (d) H1299 cells were transfected with expression vectors for p53L344P-YFP and E1B-55K, and 24 h later, cells were fixed and immunostained for E1B-55K and PML and YFP fluorescence was detected. The merge panel includes DAPI staining. The arrowhead indicates the position of an E1B-55K aggresome. (e) Plot of normalized recovery of YFP fluorescence versus time postbleaching for p53L344P-YFP in PML-nb (green circles) and plot of recovery of YFP fluorescence from p53-YFP in PML-nb with WT E1B-55K from Fig. 3e for comparison (red circles).

To test the ability of E1B-55K and p53 to form high-molecular-weight complexes such as that diagrammed in Fig. 7b, we incubated purified E1B-55K and p53 in buffer with 0.3 M KCl either separately or together and analyzed the molecular sizes of any resulting complexes by gel filtration on a Superose 6 column (Fig. 7c). E1B-55K incubated alone continued to elute with a peak near a globular protein marker of 150 kDa, as observed for the purified protein (54). E1B-55K elutes ahead of the expected position of a globular protein the size of an E1B-55K dimer (110 kDa) because it has an extended conformation in solution, as revealed by its unexpectedly low sedimentation velocity (54). p53 incubated alone eluted in a broad peak well ahead of globular proteins with the molecular mass of the p53 tetramer (∼200 kDa). This is the expected behavior during gel filtration of the highly extended p53 tetramer with random coil N-terminal and C-terminal domains (39). However, when the two purified proteins were incubated together and subjected to gel filtration, molecular complexes eluting much earlier from the gel filtration column were observed, including in the void volume, corresponding to globular protein structures of >4 MDa (Fig. 7c). These results are consistent with the formation of large molecular networks of the two proteins formed by the cross-linking of p53 tetramers by E1B-55K dimers, as diagrammed in Fig. 7b. Molecules of E1B-55K and p53 that did not shift in their elution on the gel filtration column may have been inactive for cobinding. But the observation that a significant fraction of E1B-55K and p53 did shift to high-molecular-weight fractions after coincubation indicates that the two highly purified proteins can form multiprotein complexes in vitro.

To determine the potential influence of such an E1B-55K-p53 network on the stability of p53 association with PML-nb, we used confocal microscopy and FRAP to characterize the interaction between E1B-55K and p53 with a point mutation, L344P, in its tetramerization domain that inhibits tetramer formation, resulting in p53 monomers (16). p53L344P-YFP did not associate with PML-nb in transfected p53⁻ H1299 cells in the absence of coexpressed E1B-55K but was observed diffusely throughout the nucleoplasm (data not shown). Cotransfection of expression vectors for E1B-55K and p53L344P-YFP into H1299 cells resulted in association of the two proteins in PML-nb with larger diameters than observed for WT p53-YFP (Fig. 7d, compare to Fig. 3b). In many cells, E1B-55K was concentrated on the surface of these enlarged PML-nb (Fig. 7d), and this was occasionally observed for the p53L344P-YFP as well (data not shown). FRAP analysis revealed that the association of the mutant p53-YFP with PML-nb was reproducibly decreased from a t_1/2 of ∼85 s for WT p53-YFP (Fig. 3e) to a t_1/2 of ∼10 s (Fig. 7e). These results, together with the observation of high-molecular-weight complexes formed between the two purified proteins in vitro (Fig. 7c), suggest that formation of a multiprotein network by cross-linking of p53 tetramers by E1B-55K dimers bound to their N termini contributes to the tethering of p53 in PML-nb by E1B-55K.