Adenovirus Replaces Mitotic Checkpoint Controls

ABSTRACT

Infection with adenovirus triggers the cellular DNA damage response, elements of which include cell death and cell cycle arrest. Early adenoviral proteins, including the E1B-55K and E4orf3 proteins, inhibit signaling in response to DNA damage. A fraction of cells infected with an adenovirus mutant unable to express the E1B-55K and E4orf3 genes appeared to arrest in a mitotic-like state. Cells infected early in G₁ of the cell cycle were predisposed to arrest in this state at late times of infection. This arrested state, which displays hallmarks of mitotic catastrophe, was prevented by expression of either the E1B-55K or the E4orf3 genes. However, E1B-55K mutant virus-infected cells became trapped in a mitotic-like state in the presence of the microtubule poison colcemid, suggesting that the two viral proteins restrict entry into mitosis or facilitate exit from mitosis in order to prevent infected cells from arresting in mitosis. The E1B-55K protein appeared to prevent inappropriate entry into mitosis through its interaction with the cellular tumor suppressor protein p53. The E4orf3 protein facilitated exit from mitosis by possibly mislocalizing and functionally inactivating cyclin B1. When expressed in noninfected cells, E4orf3 overcame the mitotic arrest caused by the degradation-resistant R42A cyclin B1 variant.

IMPORTANCE Cells that are infected with adenovirus type 5 early in G₁ of the cell cycle are predisposed to arrest in a mitotic-like state in a p53-dependent manner. The adenoviral E1B-55K protein prevents entry into mitosis. This newly described activity for the E1B-55K protein appears to depend on the interaction between the E1B-55K protein and the tumor suppressor p53. The adenoviral E4orf3 protein facilitates exit from mitosis, possibly by altering the intracellular distribution of cyclin B1. By preventing entry into mitosis and by promoting exit from mitosis, these adenoviral proteins act to prevent the infected cell from arresting in a mitotic-like state.

INTRODUCTION

Adenoviral infection and the ensuing replication of the viral double-stranded DNA genome activate the host DNA damage response (1, 2). Early adenoviral proteins collaborate to dampen this host response (reviewed in reference 3). The initial phase of the DNA damage response proceeds through a phosphorylation cascade, while subsequent recruitment of effector proteins also depends on the conjugation of ubiquitin and the related small ubiquitin-like modifier SUMO (4). Signals initiated by the three apical kinases or DNA-dependent protein kinase (DNA-PK) (5), ataxia telangiectasia mutated protein (ATM) (6), and ATM- and Rad3-related protein (ATR) (7) trigger downstream consequences of DNA damage, such as DNA repair, cell cycle arrest, and cell death. The tumor suppressor protein p53 is centrally positioned in the cellular response to DNA damage. Numerous branches of the DNA damage response are controlled by p53, including cell cycle arrest, cell death, senescence, autophagy, and cell proliferation (8). Not surprisingly, viruses that elicit a robust DNA damage response inevitably target p53. For adenovirus, the transcriptional activity of p53 is suppressed by the E1B-55K protein (9–11), the stability of p53 is decreased by a ubiquitin protein ligase formed by the E1B-55K and E4orf6 protein (12–14), and the expression of p53-responsive genes is epigenetically dampened by the E4orf3 protein (15).

Cell cycle arrest mediated by p53 following DNA damage typically occurs at the G₁/S border (16). However, p53 also inhibits cell cycle progression immediately before mitosis. p53 can prevent entry into mitosis by inhibiting a kinesin involved in the arrangement of condensed chromosomes (17). Polo-like kinase 1 (Plk1) promotes the transition from G₂ into mitosis. The inhibition of Plk1 uncovers p53-dependent outcomes in response to mitotic stress. In p53-deficient cells, Plk1 inhibitors and microtubule poisons elicit mitotic catastrophe and greater DNA damage than in p53-proficient cells (18). This may reflect the absence of p53-dependent apoptosis that would normally eliminate cells arrested in mitosis. It has been suggested that p53-dependent cell cycle arrest at the G₂/M border is the key factor in determining whether a cell undergoes mitotic catastrophe or apoptosis (19).

Although progression through the cell cycle can be stopped at many stages, the intricately orchestrated process of mitosis proceeds once the antephase checkpoint has been cleared or bypassed (20), despite the persistence of damaged DNA (21). Mitosis is regulated by the appropriate localization of cellular proteins and their timely degradation by the anaphase-promoting complex/cyclosome (APC/C). During the G₂ phase of the cell cycle, there is a rise in the levels of cyclin B1, which associates with Cdk1 to form the major mitotic kinase (22). Entry into mitosis begins with the activating phosphorylation of the Cdc25C phosphatase and components of the APC/C as well as the inactivating phosphorylation of the Wee1 and Myt1 kinase by polo-like kinases (23). The cyclin B1-Cdk1 complex is believed to shuttle in and out of the nucleus, with hyperphosphorylation of cyclin B1 inhibiting nuclear export of the complex, leading to an intranuclear increase in cyclin B1-Cdk1 (24, 25). Within the nucleus, this kinase directs mitotic progression by phosphorylating numerous targets (26), such as the nuclear lamins, in order promote nuclear envelope breakdown (27) and condensin II to initiate condensation of the chromosomes (28). Exit from mitosis requires the degradation of proteins ubiquitinated by the APC/C (29). Key mitotic targets of the APC/C are cyclin B1, securin, and Plk1. With the degradation of cyclin B1 and securin, separase is free to cleave cohesin from sister chromatids, leading to the precipitous separation of chromatids and progression out of mitosis (30).

Cell death by mitotic catastrophe was initially described as a caspase-dependent, apoptotic death triggered by aberrant mitosis and the persistence of active mitotic kinases (31). It has been proposed that mitotic catastrophe be considered an oncosuppressive mechanism leading to cell death resulting from the disorder of mitotic machinery that is typified by a period of aberrant mitotic arrest (32). Additional hallmarks of mitotic catastrophe include the formation of multipolar spindles and the appearance of a cleaved form of cyclin B1 that is unable to be degraded and is therefore presumed to maintain activity of the mitotic kinase (33, 34). Aberrant mitotic arrest requires that cells enter mitosis but halt prior to anaphase. Examples of viruses that arrest cells in mitosis and can elicit mitotic catastrophe include adenovirus, adenovirus-associated virus (AAV), and chicken anemia virus. Expression of the adenovirus E4orf4 protein to high levels in H1299 cells led to the accumulation of cells with 4N and greater DNA content followed by cell death characterized as “mitotic catastrophe” (35). Exposure of p53-deficient osteosarcoma cells to UV-inactivated AAV leads to centriole overduplication and mitotic arrest (19). Another example of mitotic catastrophe occurs in response to apoptin, a viral protein from chicken anemia virus. Apoptin directly inhibits the APC/C, thereby preventing exit from mitosis, resulting in mitotic arrest and subsequent apoptosis (36). This arrest was found to be independent of p53 and resulted from inhibition of the metaphase-to-anaphase transition (33).

We show here that the E1B-55K and E4orf3 genes are sufficient to prevent mitotic arrest in the adenovirus-infected cell. The E1B-55K protein circumvents entry into mitosis in a p53-dependent manner. The E4orf3 protein circumvents mitotic arrest by facilitating exit from mitosis, perhaps by mislocalizing cyclin B1. These early adenoviral proteins may prevent an untimely death that would occur in response to inappropriate mitotic arrest.

MATERIALS AND METHODS

Chemicals.

Hydroxyurea (HU) from Sigma-Aldrich (St. Louis, MO) was used at a concentration of 2 mM from a stock solution of 1 M in water. KaryoMAX colcemid from Gibco/Invitrogen (Gaithersburg, MD) was used at a concentration of 0.2 μg/ml from a stock solution of 10 μg/ml in Hanks' balanced salt solution (HBSS).

Cells.

All cell culture media, supplements, and sera were obtained from Invitrogen (Gaithersburg, MD) or Lonza (Hopkinton, MA) through the Tissue Culture and Virus Vector Core Laboratory of the Comprehensive Cancer Center of Wake Forest University. HeLa cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% newborn calf serum. H1299 cells were maintained in DMEM supplemented with 10% fetal bovine serum. Both cell lines were cultured in a 5% CO₂ atmosphere at 37°C by passaging twice weekly at a 1:10 dilution. For all experiments, cells were plated at a density of 5 × 10⁴ cells/ml. For high-resolution immunofluorescence microscopy, cells were grown on nitric acid-cleaned, sterilized glass coverslips in 6-well plates or 35-mm dishes. For cell cycle profile analysis or protein lysate collection, cells were grown in 60-mm dishes.

Viruses.

The phenotypically wild-type virus in this study contains several deletions and a substitution within the E3B region. This virus, dl309, displays wild-type characteristics in cultured cells (37). The virus referred to as the double mutant virus is 3112, a previously described (38) viral recombinant of the dl1520 virus that has an 827-bp deletion in the E1B-55K open reading frame (39) and the dl341 virus that contains an E4orf3 deletion (40). Other E1B-55K deletion mutants include dl338, which bears a 524-bp deletion (41), and dl110, which contains a 472-bp deletion (42). The parental virus for the E1B-55K point mutation mutants listed in Table 2 is H5pg4100, which contains a deletion in the E3 region at nucleotides 28593 to 30471 and has an added endonuclease restriction site (BstBI) at nucleotide 30955. The adenovirus mutants described in Table 2 were constructed by the method described in reference 43. Briefly, point mutations were introduced into the E1B-55K gene in a shuttle plasmid by site-directed mutagenesis. The appropriate fragment was inserted into the parental plasmid pH5pg4100, and the viral genomes were released from the recombinant plasmids by digestion with PacI. Mutant viruses were recovered after transfection of the linear DNA into 293 cells. All viruses were grown in 293 cells, and concentrated virus stocks were prepared by sequential centrifugation through CsCl gradients as described previously (1).

TABLE 2.

Key characteristics of the E1B-55K mutant viruses used in this study

Virus	E1B-55K mutation(s)	Likely defect	Reference(s) or source
dl110	Deletion, frameshift	Null	42
dl1520	Stop codon, deletion	Null	39
dl338	Deletion	Null	41
4149	4 stop codons	Null for all 55K related	104
4100	None	None	43
4108	T255A	E4orf6 binding	Identical mutation to virus described in reference 60
4109	H260A	p53 binding, E4orf6 binding	Identical mutation to virus described in reference 60
4127	C454S C456S	Mre11 binding	105
4174	S490A S491A T495A	Non-phospho C terminus	10, 106, 107
4185	P70T S73A	Unknown (P70T S73A)	108
4197	E472A	Daax interaction 1	55
4198	K185A K187A	Daax interaction 2	55
4216	GVVI233–236AAAS	SUMO interaction	Not published
4217	V339A V341A I342S	SUMO interaction	Not published
4227	S490D S491D T494D	Phosphomimetic C terminus	57, 107

Antibodies.

Primary antibodies included a monoclonal mouse antibody against β-tubulin (clone TUB 2.1; T4026) from Sigma-Aldrich (St. Louis, MO) used at a 1:500 dilution, a monoclonal pantropic mouse antibody against p53 (DO-1) obtained from CalBiochem (Darmstadt, Germany) used at a 1:100 dilution, and a polyclonal rabbit antibody against phospho-histone H3 (Thr11; 9849) from Cell Signaling (Danvers, MA) used at a 1:100 dilution for immunofluorescence. Adenovirus-specific antibodies included the mouse monoclonal antibody Rsa3 (44) for the E4orf6/7 protein and the rat monoclonal antibody 6A11 (45) used as hybridoma culture supernatant fluid diluted 1:5. Five antibody preparations for cyclin B1 were used for this study. The mouse monoclonal antibody against cyclin B1 (CC03) was obtained from Oncogene Research Products (CalBiochem, Cambridge, MA) and used at a 1:250 dilution for microscopy and a 1:1,000 dilution for Western blotting. The remaining cyclin B1 antibodies were obtained from EMD Millipore (Billerica, MA) and included the mouse monoclonal antibody specific for the C terminus of human cyclin B1 (clone Y106), used at a 1:100 dilution for immunofluorescence and a 1:10,000 dilution in 5% milk for Western blotting, a mixture of mouse monoclonal IgGs specific for human cyclin B1 (catalog no. 05-373), used at 10 μg per ml for immunofluorescence and 0.33 μg per ml for Western blotting, a monoclonal antibody raised against hamster cyclin B1 (catalog no. MAB3684), used at a 1:100 dilution for immunofluorescence and a 1:5,000 dilution in 5% milk for Western blotting, and a rabbit monoclonal antibody specific for phosphorylated serine 126 in cyclin B1 (catalog no. MABE490), used at a 1:100 dilution for immunofluorescence and a 1:3,000 dilution for Western blotting. Secondary antibodies used for immunofluorescence microscopy were anti-mouse or anti-rabbit whole IgG conjugated to Alexa Fluor 488 (AF488) or Alexa Fluor 568 (AF568) from Invitrogen used at 2 μg/ml. The secondary antibody used for Western blot analysis was an anti-mouse or anti-rabbit whole IgG raised in goats and conjugated to horseradish peroxidase from Jackson ImmunoResearch Laboratories (West Grove, PA), used at a concentration of 0.1 μg/ml.

Plasmids and PEI transfection.

A plasmid expressing the wild-type human TP53 gene under the control of the cytomegalovirus (CMV) immediate early promoter and enhancer was generously provided by Guangchao Sui (Wake Forest School of Medicine). Plasmids expressing human cyclin B1 fused to the yellow fluorescent protein Venus (pVenus-N1 cyclin B1; Addgene plasmid 26062) and the degradation-resistant form of cyclin B1 (pVenus cyclin B1 R42A; Addgene plasmid 39873) were the gift of Jonathon Pines (Gurdon Institute, Cambridge, United Kingdom). Cells were seeded on glass coverslips in a 6-well culture dish. For each well, 1 μg of DNA was used in a volume of 0.2 ml serum-free medium with poly(ethylenimine) (PEI) at a 1:5 dilution from a 7.5 mM stock dissolved in deionized water with gentle heating. Transfections were carried out in a 5% CO₂ atmosphere at 37°C with gentle rocking and regular rotation for 8 h before being replacing the DNA and PEI mixture with growth medium.

Indirect immunofluorescence.

Cells were washed twice with phosphate-buffered saline (PBS), fixed for 30 min with 2% paraformaldehyde, and permeabilized for 5 min with 0.2% Triton X-100 in PBS at room temperature. All subsequent washes were performed with Tris-buffered saline with bovine serum albumin (BSA), glycine, and Tween 20 (TBS-BGT: 0.137 M NaCl, 0.003 M KCl, 0.025 M Tris-Cl [pH 8.0], 0.0015 M MgCl₂, 0.5% BSA, 0.1% glycine, 0.05% Tween 20, 0.02% sodium azide). Antibodies used for immunofluorescence were diluted in TBS-BGT supplemented with 10% normal goat serum (Invitrogen). Samples were stained for 90 min with primary antibody and for 30 min with secondary antibody with multiple washes with TBS-BGT between. Samples were mounted with ProLong Gold mounting medium (Invitrogen) containing 4′,6-diamidino-2-phenylindole (DAPI).

Microscopy.

Micrographs were obtained by standard epifluorescence microscopy with a Nikon TE300 inverted microscope or by confocal laser scanning microscopy with a Nikon TiE inverted microscope fitted with a Nikon C1si system. A 100×/1.4 NA magnification oil immersion objective was used for all micrographs. Monochromatic images were acquired on the Nikon TE300 microscope with a Retiga EX 1350 digital camera (QImaging Corp., Burnaby, British Columbia, Canada). Confocal images were acquired as a series of 5 sections at 0.2-μm intervals in the center of the nucleus, using sequential excitation for each of the three fluorochromes. The confocal images are presented as a maximum-intensity projection to reduce the three-dimensional information to two dimensions. Merged color images were prepared by assigning red, green, or blue to the appropriate fluorochrome. For single-channel fluorescent images, a pseudocolor scheme was assigned in order to mimic the visual perception with increased saturation at increased fluorescent intensity. Monochromatic images were assigned colors with the open-source program ImageJ (version 1.46). All figures were assembled with the open-source vector graphics editor Inkscape (version 0.48) and raster graphics editor GIMP (version 2.8.4.).

Flow cytometry.

Cells were harvested by trypsinization. 5-Ethynyl-2′-deoxyuridine (EdU)-labeled cells were labeled as indicated below and resuspended in fluorescence-activated cell sorter (FACS) buffer (1% BSA in PBS) with propidium iodide (PI) provided by the Click-iT EdU kit. All other samples were washed twice with PBS and resuspended in 100 μl PBS. Cells were transferred dropwise into 2.5 ml of 70% ethanol with constant mixing. Samples were stored at −20°C for at least 12 h. Ethanol was then removed, and the samples were resuspended at a density of 10⁶ cells per ml in PI solution diluted in water from a 10× stock (1 M NaCl, 0.36 M sodium citrate, 500 μg/ml PI, 6% NP-40) with 0.1 mg/ml RNase A diluted from a stock solution of 10 mg/ml in 0.01 M sodium acetate with 0.1 M Tris-Cl (pH 7.4). Samples were incubated with PI solution for 30 min in the dark at 37°C and then transferred to ice before flow cytometric analysis. A Becton Dickinson FACSCalibur instrument was used to acquire the PI signal in the linear mode. Gating on the peak width and peak area was used to select single cells for DNA profile and cell cycle analysis.

Western blotting.

Cells were washed in PBS supplemented with protease and phosphatase inhibitors (2 mM EDTA, 1 mM NaF, 1 mM sodium pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na₃VO₄, 2 μM leupeptin), harvested by scraping, and suspended in a volume of one-tenth concentrated PBS sufficient to dilute the cells to a density of 2 × 10⁷ cells/ml An equal volume of 2× sodium dodecyl sulfate (SDS) protein sample buffer (4% SDS, 0.135 M Tris [pH 6.8], 20% glycerol, 0.02% bromophenol blue, 6% β-mercaptoethanol) was added to make the final concentration 10⁷ cells/ml. The cell lysate was heated for 5 min at 95°C and subjected to 3 pulses of sonication lasting 20 s each. The lysates were separated by SDS-polyacrylamide gel electrophoresis through 10% acrylamide gels. The proteins were then electrophoretically transferred to nitrocellulose (Whatman [GE Healthcare]) overnight at 4°C. The nitrocellulose was blocked in TBS-BGT containing 5% nonfat dry milk and sodium azide, stained with primary antibody diluted in TBS-BGT with sodium azide overnight at 4°C, and stained with secondary antibody diluted in TBS-BGT without sodium azide for 1 h at room temperature. The stained protein was visualized by a mixture of SuperSignal West Pico and SuperSignal West Femto chemiluminescence substrate from Pierce/ThermoScientific (Rockford, IL) and X-ray film. The density of the specific signal recorded by the X-ray film was quantified by scanning nonsaturated exposures at a 16-bit resolution and measuring the optical density with the tools available in ImageJ.

Click-iT EdU.

Cells were pulse-labeled with an appropriate amount of EdU from a 10 mM stock in (DMSO) diluted in a small volume of warm growth medium to make a final concentration of 100 μM EdU. At the end of the labeling period, the wells were washed and replaced with prewarmed growth medium. At the time of harvest, samples for DNA profile analysis were harvested with trypsin, while samples on glass coverslips were fixed, permeabilized, and processed for immunofluorescence in a humidifying chamber. The samples were fixed with 4% paraformaldehyde in PBS for 15 min, washed with FACS buffer (1% BSA in PBS), and permeabilized with 0.5% Triton X-100 for 20 min at room temperature. The Click-iT reaction cocktail was prepared by mixing Click-iT reaction buffer, CuSO₄, fluorescently labeled azide, and reaction buffer additive as indicated by the manufacturer. Samples were washed with FACS buffer and allowed to incubate with the Click-iT reaction cocktail in a humidifying chamber at room temperature for 30 min in the dark. The samples were washed, and those on coverslips were mounted on slides with Prolong Gold supplemented with DAPI. Samples for DNA profile analysis were pelleted and resuspended in 500 μl of FACS buffer with 0.2 mg/ml RNase A (from 20 mg/ml) and 4 μg/ml PI (from a 1-mg/ml stock solution in water).

Cell synchronization.

Mitotic cells were mechanically harvested from subconfluent cells grown in 75-cm² flasks and pelleted gently. The mitosis-enriched pellet was resuspended at a concentration of 2 × 10⁴ cells per ml in 2 mM HU in growth medium and plated in 60-mm dishes for DNA profile analysis or on acid-treated glass coverslips in 35-mm dishes for immunofluorescence. After 1 h, plates were swirled to dislodge any dead S-phase cells, and the medium was replaced with a fresh solution of 2 mM HU in growth medium. Cells were held in HU for 16 h to allow cells to cycle to the G₁/S border. All samples were released from HU by being washed and replaced with the growth medium. At various times postinfection, cells were collected for DNA profile analysis in order to determine cell cycle distribution in the sample.

RESULTS

Cells infected during early G₁ give rise to cells trapped in a mitotic-like state.

A subset of cells infected with an adenovirus deleted of the E1B-55K and E4orf3 genes develop highly condensed chromatin resembling that of a mitotic cell (46). Because typically fewer than 10% of the infected cells develop the highly condensed chromatin, it seems likely that an underlying process or condition limits the number of infected cells that condensed their chromatin and resemble a mitotic cell. Potential processes include the death and loss of cells following mitotic catastrophe, a small probability of entering mitosis, or the existence of a limited subset of cells that are competent to enter mitosis. Since the outcome of an infection with the E1B-55K-mutant virus is strongly determined by the stage of the cell cycle at the time of infection (38, 47, 48), we performed two experiments to determine if entry into a mitotic-like state was affected by the stage of the cell cycle at infection. Asynchronously dividing HeLa cells were exposed to the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU) for a 4-h interval in order to identify cells that passed through S phase during this period of time. After returning the cells to EdU-free medium, the cells were maintained for various intervals of time before being infected. At the time of infection, a portion of the cells was collected and EdU-positive cells were identified by the Cu(I)-catalyzed reaction between a fluorescent azide and the alkyne in EdU. This sample was analyzed by flow cytometry for DNA and EdU in order to determine the stage of the cell cycle that EdU-positive cells had reached at infection. The labeling scheme and associated flow cytometric analyses are shown in Fig. 1. After 72 h, another fraction of the infected cells was stained for EdU and DNA and the number of mitotic-like nuclei was enumerated by fluorescence microscopy. The results of scoring approximately 600 nuclei at each time point show that cells with condensed DNA were seen most frequently among cells infected during G₂/M and early G₁ (Table 1).

FIG 1.

Click-iT EdU labeling protocol and DNA profile analysis of EdU-labeled cells. (A) HeLa cells were labeled with Click-iT EdU for six 4-h periods (indicated by arrows) over the course of 24 h prior to infection. At time zero h, EdU-labeled cultures were infected with the E1B-55K/E4orf3 double mutant virus at a multiplicity of infection (MOI) of 10 and processed for immunofluorescence at 72 hpi or were collected for flow cytometric analysis. Shading indicates the phase of the cell cycle the EdU-labeled cells are expected to have reached at infection (S phase, gray; G₂/M, black; G₁, white). (B) Samples of cells labeled with Click-iT EdU were collected at the time of infection (0 h) and stained with propidium iodide for DNA content and for EdU as described in Materials and Methods. The shaded density profile represents the DNA profile for the entire population of cells. The unshaded density profile shows the DNA profile of EdU-positive cells.

TABLE 1.

EdU-positive and mitotic-like double mutant virus-infected cells

Labeling period (h)a	EdU-positive cells (%)b			Phasec	Mitotic-like EdU-positive cells (%)
	G1	S	G2/M
−2 to +2	27	52	21	Mid-S	25.6
−6 to −2	6	15	79	G2/M	69.6
−10 to −6	35	7	58	M, early G1	50.0
−14 to −10	68	5	27	Mid-G1	20.0
−18 to −14	61	25	14	Late G1	0.0
−22 to −18	56	25	20	G1, early S	5.8

^aHours exposed to EdU before being infected at an MOI of 10 at 0 h.

^bPercentage of EdU-positive cells with DNA content characteristic of the indicated stage of the cell cycle.

^cPredominant stage of the cell cycle for EdU-positive cells at 0 h.

Although the EdU labeling method permits analysis of unperturbed dividing cells, the temporal resolution afforded by a 4-h labeling period was limited. For better resolution, synchronously dividing HeLa cells were prepared as previously described (47, 49) and infected with the E1B-55K/E4orf3 double mutant virus at hourly intervals from 6 to 14 h following entry into S phase. A portion of the synchronized cells was collected and analyzed for DNA content by flow cytometry in order to determine the stage of the cell cycle at the time of infection (Fig. 2A). After 72 h of infection, the cells were fixed and the frequency of cells with mitotic-like nuclei was determined (Fig. 2B). These results agree with the findings obtained with asynchronously dividing cells (Fig. 1) and indicate that cells infected with the E1B-55K/E4orf3 double mutant virus during a 4-h window in early G₁ gave rise to the subset of cells with highly condensed chromatin. These results suggest that the adenoviral E1B-55K and E4orf3 proteins prevent cells infected early in G₁ from becoming trapped in a mitotic-like state at late times of infection.

FIG 2.

Early G₁ cells give rise to mitotic-like nuclei after infection with the E1B-55K/E4orf3 double mutant virus. Synchronously dividing HeLa cells were obtained by mitotic shake and hydroxyurea selection as described in Materials and Methods. Synchronously dividing cells were either harvested for DNA profile analysis or infected at the indicated time after entering S phase with the E1B-55K/E4orf3 double mutant virus at an MOI of 10. (A) Cell cycle distribution by DNA profile analysis. (B) At 72 hpi, cells were stained for DNA with DAPI and evaluated by fluorescence microscopy to determine the frequency of cells with condensed DNA. The stage of the cell cycle at the time of infection is indicated below. A representative experiment is shown of three that were performed with overlapping times of infection after entering S phase.

Cells infected with the E1B-55K/E4orf3 double mutant virus show evidence of mitotic distress.

Approximately 4% of asynchronously dividing HeLa cells contain condensed chromatin typical of the later stages of mitosis (Fig. 3A, mock panels, and B). The mitotic spindle in these cells, visualized by staining for β-tubulin, occurs in a symmetrical bipolar arrangement about the condensed chromatin in metaphase and anaphase cells (Fig. 3A, panel a). A fraction of the mitotic cells also contain highly phosphorylated histone H3 or phospho-H3 (Fig. 3A, panel b, and B). Phosphorylation of histone H3 initiates late in G₂ phase, is completed in late prophase, and is maintained through metaphase. Histone H3 phosphorylation precipitously decreases during anaphase as the cell exits mitosis (50). In contrast to mock-infected cells, none of the cells infected with the wild-type or E4orf3 mutant adenovirus and less than 0.5% of cells infected with the E1B-55K mutant virus contained condensed chromatin or phospho-H3 when evaluated at 72 h postinfection (hpi) (Fig. 3B). This is consistent with previous observations indicating that adenovirus-infected human epithelial cells cease progression through the cell cycle (51). However, in these asynchronously infected cultures, a significant number (>10%) of cells infected with the E1B-55K and E4orf3 double mutant virus contained chromatin characteristic of a mitotic cell (Fig. 3A, ΔE1B-55K/ΔE4orf3, and B). Some of the double mutant virus-infected cells also stained for phospho-H3, suggesting that these cells progressed into mitosis. Although these cells superficially resembled mitotic cells, many contained an asymmetric distribution of β-tubulin or even a multipolar spindle. The abundance of aberrant mitotic spindles in these cells makes it seems likely that these cells will fail to produce viable daughter cells. Expression of the small avian virus-derived protein apoptin in human tumor cells has been reported to lead to a similar mixture of cells with apparently normal, asymmetric, and multipolar spindles due to a block in the metaphase-to-anaphase transition (33).

FIG 3.

HeLa cells infected with the E1B-55K/E4orf3 double mutant virus show evidence of mitotic distress. HeLa cells were mock infected or infected at an MOI of 10 with the indicated viruses and stained at 72 h postinfection with DAPI to visualize DNA (blue), antibodies to phospho-H3 as a marker of early mitosis (red), and antibodies to β-tubulin to visualize mitotic spindles (green). (A) Representative images of mock (a and b), wild-type (c), and E1B-55K/E4orf3 mutant (d and e) viral infections are shown. (B) The frequency of cells containing mitotic-like condensed DNA, staining for phospho-H3 and asymmetrically stained mitotic spindles, was quantified for at least 500 cells for each viral infection. A representative experiment of three is shown. Error bars indicate the 95% exact binomial confidence interval for the representative experiment. (C) HeLa cells were infected with the E1B 55K/E4orf3 double mutant virus at an MOI of 10. Cells were stained for DNA at the indicated times postinfection, and the frequency of mitotic-like cells was determined for approximately 500 cells at each time point. The symbols with the dashed line for times beyond 72 hpi indicate imprecision due to the probable loss of cells because of death or detachment. The experiment shown is representative of three independent experiments with similar outcomes. Error bars indicate the 95% exact binomial confidence interval for the representative experiment.

Cells infected with the double mutant virus that appear to arrest in mitosis do so only at late times of infection (Fig. 3C). In some experiments, a transient increase in mitotic-like cells occurred at 24 hpi. However, cells with highly condensed DNA do not begin to accumulate until 60 hpi. The number of cells apparently arrested in mitosis reached a maximum at 72 hpi. The decrease in cells with mitotic-like nuclei after 72 hpi appears to be due to the death or selective detachment of these cells from the substrate. Consequently, subsequent experiments evaluated cells at 72 hpi. Taken together, these results suggest that the E1B-55K and E4orf3 proteins prevent cells infected during early G₁ from becoming trapped in a mitotic-like state at late times of infection.

Adenoviral E1B-55K protein prevents entry into mitosis.

Because significant numbers of cells were observed to contain condensed DNA only after infection with the double mutant virus, it seems likely that the E1B-55K and E4orf3 proteins act independently to prevent the infected cell from arresting in a mitotic-like state. Conceptually, these viral oncoproteins could prevent the accumulation of mitotic-like cells by two nonexclusive mechanisms. First, the viral proteins could prevent entry into mitosis. Second, the viral proteins could facilitate exit from mitosis. Cherubini and associates observed that a fraction of E1B-55K mutant virus-infected cells accumulated highly condensed chromosomes after 12 h of exposure to colcemid (52). By depolymerizing microtubules, colcemid prevents the completion of mitosis, thus trapping any cells that entered mitosis during the exposure to colcemid. Since mitotic-like cells did not accumulate in colcemid-treated cells that were infected with the wild-type virus, we reasoned that the E1B-55K protein might prevent the infected cell from entering mitosis. To test this hypothesis, infected cells were treated with colcemid for 12 h prior to fixation and analysis at various times after infection. The number of mitotic-like cells infected with the wild-type or E4orf3 mutant virus did not increase, suggesting that cells infected with the wild-type or E4orf3 mutant virus did not enter mitosis (Fig. 4). As expected, cells with condensed DNA were observed following infection with the double mutant virus; the rate at which these mitotic-like cells accumulated was the same in the presence (Fig. 4) or absence (Fig. 3C) of colcemid. In contrast to the wild-type or E4orf3 mutant virus, colcemid trapped increasing numbers of mitotic-like cells during the course of infection with the E1B-55K mutant virus (Fig. 4). These results, which recapitulate the findings reported by Cherubini (52), suggest that some E1B-55K mutant virus-infected cells enter and exit mitosis at late times of infection. This indicates that another role for the E1B-55K protein is to prevent adenovirus-infected cells from entering into mitosis.

FIG 4.

Colcemid traps cells infected with the E1B-55K mutant virus in a mitotic-like state. HeLa cells were infected at an MOI of 10 with the indicated viruses and treated with colcemid for 12 h at the indicated times postinfection before being fixed and stained with DAPI to visualize DNA. The frequency of mitotic-like cells was determined for approximately 500 cells for each virus at each time point.

E1B-55K prevents entry into mitosis through p53.

The E1B-55K protein serves many roles during an infection. The E1B-55K and E4orf6 proteins collaborate to form a novel SCF (Skp1–Cul1–F-box complex)-type E3 ubiquitin ligase with the scaffold protein Cul5, elongins B and C, and the RING box protein Rbx1 (12–14). This adenovirus-specific complex targets several cellular proteins for degradation, many of which signal or respond to DNA damage (see, for example, references 53 to 58). Acting independently of E4orf6, the E1B-55K protein directs degradation of the cellular transcription factor Daax, mislocalizes the DNA damage-responsive chromatin structure regulator SPOC1, and can directly block p53-mediated transcription (11, 55, 59). In order to identify the activity of the E1B-55K protein that prevents entry into mitosis, HeLa cells were infected with the E1B-55K mutant adenoviruses described in Table 2. At 60 hpi, the infected cells were treated with colcemid to trap any cells entering mitosis and the cells with condensed DNA were counted by fluorescence microscopy. As expected, cells infected with the three E1B-55K-null adenoviruses as well as the virus bearing four nonsense mutations in the E1B-55K open reading frames (H5pm4149) were trapped in a mitotic-like state by colcemid (Fig. 5A and B). Among the viruses bearing missense mutations in the E1B-55K gene, only H5pm4109 was associated with a significant increase in mitotic-like cells (Fig. 5B). This virus encodes an E1B-55K protein that is unable to bind p53. All other E1B-55K variants analyzed in Fig. 5B appeared to prevent entry into mitosis as well as the wild-type virus H5pg4100. Because over 20% of noninfected cells were trapped in a mitotic-like state by this treatment (data not shown), the failure to detect mitotic-like cells among virus-infected cells confirms that the cells were uniformly infected. Because cells infected with the mutant virus H5pm4108 were not trapped in a mitotic-like state by colcemid, we conclude that the ability to block entry into mitosis is independent of the E1B-55K/E4orf6 protein complex since H5pm4108 expresses an E1B-55K protein that fails to bind E4orf6 (see reference 60). These findings suggest that the key property of the E1B-55K protein needed to prevent entry into mitosis is its ability to interact with p53.

FIG 5.

The ability to inhibit entry into a mitotic-like state maps to the p53-binding ability of the E1B-55K protein. (A) HeLa cells were infected at an MOI of 10 with the indicated E1B-55K-null viruses and treated with colcemid or vehicle control 12 h prior to staining with DAPI at 72 hpi to visualize DNA. The frequency of mitotic-like cells was determined by fluorescence microscopy. A representative experiment of three independent experiments is shown. For each of the E1B-55K-null viruses, colcemid significantly increased the fraction of mitotic-like cells (P < 10⁻⁶ by Fisher's exact test). Error bars indicate the 95% exact binomial confidence interval for the representative experiment. (B) HeLa cells were infected at an MOI of 10 with the indicated viruses bearing point mutations in the E1B-55K gene (described in Table 2) and treated with colcemid 12 h prior to staining at 72 hpi with DAPI to visualize DNA. The frequency of mitotic-like cells is shown for each mutant infection. The proportion of mitotic-like nuclei was nonrandomly distributed among the 11 virus-infected samples exclusive of the E1B-55K-null virus H5pm4149 (P < 0.0001, chi-square test). The chi-square test was repeated after systematically excluding each sample. The P value was nonsignificant (P = 0.56) only when the virus H5pm4109 was excluded from the analysis. This analysis was again repeated by excluding individual samples in the collection that also excluded H5pm4109. The chi-square test reported nonsignificant P values for each subset lacking H5pm4109 and one other sample. This indicates that among the non-null viruses, only H5pm4109-infected cells exhibited a significant change in the number of condensed nuclei in the presence of colcemid. Results from a representative experiment of three independent experiments with similar outcomes are shown. (C) p53-null H1299 cells were transfected with a plasmid to express p53 24 h before being mock infected or infected at an MOI of 10 with either the E1B-55K single mutant or E1B-55K/E4orf3 double mutant virus. The cells were either left untreated or were treated with 0.2 μg/ml colcemid 12 h prior to immunostaining for p53 and visualization of DNA with DAPI staining at 72 hpi. Cells were classified as either p53 positive or negative, and the frequency of mitotic-like cells was determined. As noted in the text, condensed nuclei in mock-infected cells resembled pyknotic nuclei of apoptotic cells. The number of p53-positive virus-infected cells with condensed DNA was increased significantly over the number of p53-negative cells with condensed DNA (P = 2 × 10⁻⁶ by Fisher's exact test).

In many HPV-transformed cells, expression of the integrated E6 gene of human papilomavirus (HPV) directs the continual degradation of p53 protein (61). Without the selective pressure to eliminate p53 gene function, the TP53 gene in HeLa cells has remained intact (62). Wild-type p53 protein accumulates to measurable levels in HeLa cells infected with adenovirus and to a high level in cells infected with adenovirus mutants that fail to direct the degradation of p53. To examine the possibility that p53 contributes to mitotic entry in the infected cell, p53-null H1299 cells were infected with the E1B-55K/E4orf3 double mutant virus and evaluated. In sharp contrast to HeLa cells, no mitotic-like cells were observed among H1299 cells infected with this virus at any time after infection (data not shown) (Fig. 5C).

To test directly for a role for p53, H1299 cells were transfected with a p53 expression plasmid and infected after 24 h. At 60 hpi, cells were exposed to colcemid or left untreated. At 72 hpi, cells were stained for p53 and DNA and then evaluated by fluorescence microscopy. Enforced expression of p53 is acutely toxic to H1299 cells (63). Consequently, transfection of the p53 expression plasmid reduced the number of evaluable cells and induced apoptosis in the mock-infected cells after 3 days (data not shown). It seems likely that the increased number of p53-positive, mock-infected H1299 cells with condensed DNA was due to pyknosis or chromatin condensation from apoptosis rather than mitosis (Fig. 5C, mock). Apoptosis is inhibited in the adenovirus-infected cells because all of the viruses studied here express the antiapoptotic E1B-19K protein (64). Accordingly, in the absence of colcemid, neither pyknotic nor mitotic-like nuclei were observed among E1B-55K mutant virus-infected H1299 cells (Fig. 5C). However, colcemid trapped a significant number of p53-positive H1299 cells in a mitotic-like state. Similarly, mitotic-like nuclei were observed only in p53-positive H1299 cells infected with the E1B-55K/E4orf3 double mutant virus. As noted for HeLa cells, colcemid was not required to trap the mitotic-like double mutant virus-infected H1299 cells. At least for the HeLa and H1299 cell lines, these results are consistent with an unexpected role for p53; if the infected cells with condensed nuclei are indeed mitotic, p53 appears to be necessary for the adenovirus-infected cell to enter mitosis. Furthermore, the E1B-55K protein blocks this activity of p53.

E4orf3 targets cyclin B1.

Colcemid is required to trap HeLa cells and p53-positive H1299 cells infected with the E1B-55K-mutant virus in a mitotic-like state. In contrast, p53-positive cells infected with the E1B-55K/E4orf3 double mutant virus are trapped in a mitotic state without colcemid. If the absence of the E1B-55K protein allows infected p53-positive cells to enter mitosis, this result suggests that the presence of E4orf3 protein may facilitate exit from mitosis. Degradation of the major mitotic cyclin cyclin B1 is a critical event that precipitates exit from mitosis (65). The complete degradation of cyclin B1 is required for cells to proceed with cytokinesis (66). Previous studies showed an increase in cyclin B1 in wild-type adenovirus-infected WI-38 and A549 cells as well as an S-phase-dependent increase in cyclin B1 in E1B-55K mutant virus-infected cells (67). We therefore compared by immunoblotting the nature and abundance of cyclin B1 among HeLa cells infected with wild-type and mutant viruses. The level of cyclin B1 was indeed higher in virus-infected cells than in mock-infected cells (Fig. 6), although it seems unlikely that the modest increase in cyclin B1 level in double mutant virus-infected cells could force an apparent mitotic arrest. Adenovirus-infected cells also contained an additional cyclin B1-related protein of slightly greater electrophoretic mobility than the form in mock-infected cells (Fig. 6A). This product was recognized by four different cyclin B1-specific antibody preparations. Both ∼50-kDa forms of cyclin B1 appeared equally abundant when queried by a phosphoserine 126-specific antibody (data not shown). The origin of the two forms of protein of approximately 50 kDa remains unclear. A 35-kDa protein recognized by cyclin B1 antibodies was detected in lysates from cells infected with the E1B-55K/E4orf3 double mutant virus and, to a lesser extent, with the E4orf3-mutant virus. This smaller form of cyclin B1 appears to correspond to a cleavage product found during mitotic catastrophe, termed cyclin B1Δ (34). It was suggested that cyclin B1Δ acts as a dominant-negative inhibitor of cyclin B1 function and sustains the mitotic block in cells that would otherwise exit mitosis (34). Both the elevated levels of cyclin B1 and the presence of a dominant-negative inhibitor could retard E1B-55K/E4orf3 double mutant virus-infected cells in mitosis. In addition, changes in the localization of cyclin B1 in the infected cells point to another mechanism by which E4orf3 could facilitate exit from mitosis.

FIG 6.

Cyclin B1 levels are elevated during adenoviral infections. HeLa cells were mock infected or infected with the indicated viruses at an MOI of 10. Cellular lysates were collected in the presence of protease and phosphatase inhibitors at 72 hpi. Material from identical numbers of infected cells was separated by SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted for cyclin B1 (A) and β-actin (B). An overexposed β-actin blot is presented here to permit visualization of the weaker signals. Nonsaturated exposures were used for quantitative analyses. The position of a cyclin B1-related product of 35 kDa is indicated by the arrowhead. (C) The optical density of the signal for the intact cyclin B1 products was quantified, normalized to β-actin, and then normalized to the value measured from mock-infected cells in three independent experiments. The mean and standard deviation are plotted on a log scale. Application of the t test to log-transformed values shows that levels of cyclin B1 in were significantly greater than in mock-infected cells in dl309- and dl1520-infected cells (P < 0.005). Differences were not significant in dl341- or 3112-infected cells (P = 0.97 and 0.18, respectively).

Because E4orf3 disrupts cell signaling pathways by mislocalizing host proteins (68, 69), we explored the possibility that E4orf3 mislocalizes cyclin B1 in the infected cell. The different localizations of cyclin B1 among mock-infected cells were consistent with fluctuations in the level and movement of cyclin B1 during cell cycle progression. Cells in G₂ contained high levels of cyclin B1 that was found in a diffuse or speckled pattern in the cytoplasm (Fig. 7A, panel a). Early in mitosis, cells contained high levels of cyclin B1 that was largely coincident with chromatin (Fig. 7A, panel b). Finally, a subset of cells with condensed chromatin was judge to be in late mitosis because of the absence of cyclin B1 staining (Fig. 7A, panel c). A subset of cells infected with the wild-type and E1B-55K mutant virus contained nuclear cyclin B1; in many of these cells, cyclin B1 was found in large aggregates throughout the nucleus (Fig. 7B, panels a to d). A fraction of cells infected with E4orf3 mutant viruses also contained nuclear cyclin B1. In contrast to infected cells containing the E4orf3 protein, cyclin B1 was diffusely distributed in the nucleus of cells infected with the E4orf3 mutant viruses (Fig. 7B, panels e to h). The distribution of cyclin B1 in these infected cells more closely resembled the patterns observed in mitotic or G₂ mock-infected cells. These results, which are quantified in Fig. 7C, show that the E4orf3 protein alters the distribution of cyclin B1 in the cell nucleus during an adenoviral infection. E4orf3 may functionally inactivate cyclin B1 in order to facilitate exit from mitosis.

FIG 7.

E4orf3 alters the distribution of cyclin B1 during adenoviral infections. HeLa cells were mock infected or infected with the indicated viruses at an MOI of 10 and stained for DNA (blue) and cyclin B1 (green) at 72 hpi. (A) Representative fields show asynchronously dividing mock-infected cells, with white arrowheads indicating a G₂-phase cell (a), open arrowheads indicate an early mitotic cell (b), and double arrowheads indicating a late mitotic cell. The white bar in panel c indicates 20 nm. (B) Representative cells show patterns of cyclin B1 staining for HeLa cells infected with wild-type (a and b), E1B-55K mutant (c and d), E4orf3 mutant (e and f), and E1B-55K/E4orf3 double mutant (g and h) viruses. (C) The relative frequencies of the three patterns (aggregates, diffuse or speckles, and absent) of cyclin B1 localization are tabulated for a representative experiment. Similar results were obtained in three independent experiments in which approximately 500 cells were evaluated for each virus.

E4orf3 overcomes metaphase arrest imposed by a nondegradable cyclin B1.

Degradation of cyclin B1 during mitosis requires key residues (R₄₂xxL₄₅xxI/V₄₈xN₅₀) in the destruction box (70). Expression of cyclin B1 variants with mutations in the destruction box force cells to accumulate in a mitotic-like state (66, 71). We expressed E4orf3 and the wild-type or degradation-resistant cyclin B1 (CycB1 R42A) by transfection to determine if E4orf3 can overcome the metaphase arrest imposed by CycB1 R42A (71). As expected, enforced expression of wild-type cyclin B1 increased the fraction of cells with condensed DNA (Fig. 8A). Although expression of the E4orf6/7 cDNA reduced the frequency of these mitotic-like cells, the difference was not statistically significant (P = 0.09). E4orf3 decreased the frequency of these mitotic-like cells to statistically significant although modestly reduced levels (7%; P = 0.02). However, the effect of E4orf3 was pronounced in cells expressing the degradation-resistant R42A cyclin B1. More than half of the cells expressing the R42A variant contained condensed DNA after 2 days. This number was not affected by E4orf6/7 (P = 0.32). In contrast, E4orf3 significantly (P = 0.01) and substantially reduced the fraction of mitotic-like cells expressing the R42A variant (Fig. 8A). These results indicate that expression of E4orf3 overcomes the mitotic-like state caused by elevated levels of cyclin B1 without forcing the degradation of cyclin B1.

FIG 8.

E4orf3 overcomes the sustained chromatin condensation elicited by expression of the degradation-resistant R42A cyclin B1 variant. (A) The wild-type cyclin B1 fused to the yellow fluorescent protein Venus (CycB1 wt) or the degradation-resistant form of cyclin B1 fused to Venus (CycB1 R42A) was expressed by transfection in HeLa cells together with E4orf3, E4orf6/7, or an empty vector control. Cells were processed 48 h after transfection and stained for the E4 protein. The cyclin B1 protein was visualized by Venus fluorescence. DNA was visualized by DAPI staining. Chromatin was scored as condensed (mitotic like) or diffuse (interphase like) in approximately 50 cells expressing each combination of the cyclin B1 and E4 constructs in each of three independent experiments. Data are reported as the mean and standard deviation. The effect of the E4 construct (or absence of the construct) on the fraction of cells containing condensed DNA was evaluated with a pairwise t test allowing for different variances among the three groups. The P value was adjusted for multiple comparisons by the Holm-Bonferroni method. Adjusted P values below 0.05 were considered significant. (B) Three representative cells showing that the E4orf6/7 protein does not affect chromatin condensation promoted by CycB1 R42A. The scale bar in panel a indicates 10 nm. (C) Chromatin remains diffuse in three representative cells expressing the R42A cyclin B1 variant and E4orf3. The cyclin B1 fusion protein was excluded from the nucleus in most (>80%) cells expressing both fusion protein and E4orf3, as indicated in panels a and b.

Expression of E4orf3 by transfection did not promote the aggregation of cyclin B1 seen in virus-infected cells (Fig. 7C). We observed no differences in the localization of the cyclin B1 fusion protein among cells expressing E4orf6/7 (Fig. 8B), E4orf3 (Fig. 8C), or no E4 construct (data not shown). Occasional aggregates of cyclin B1 were noted irrespective of the cotransfected viral gene, such as in Fig. 8B, panel a. However, cyclin B1 appeared to be largely excluded from the nucleus in most of the cells expressing both E4orf3 and the cyclin B1 fusion protein. Since so few productively transfected cells contained detectable levels of endogenous cyclin B1, it was not possible to determine if E4orf3 affected the endogenous protein in a similar manner. Because cells expressing E4orf3 and cyclin B1 R42A did not have significant levels of the cyclin B1 in the nucleus after 24, 36, and 72 h of transfection (data not shown), it seems likely that E4orf3 precludes entry of cyclin B1 into the nucleus or promotes nuclear export of cyclin B1.

DISCUSSION

Cells infected by the E1B-55K/E4orf3 double mutant virus during early G₁ are predisposed to arrest in a mitotic-like state. The dependence of this phenomenon on the stage of the cell cycle at the time of infection is unusual but not unexpected. We previously demonstrated that cells infected during S phase by the E1B-55K-deleted virus support a more productive infection (47, 48) and are more rapidly killed than G₁-infected cells (51, 72). The findings reported here reinforce the notion that an infection initiated during G₁ is restrictive for E1B-55K-mutant adenoviruses.

The importance of the E1B-55K and E4orf3 proteins in preventing entry and arrest in mitosis is counterintuitive given that these proteins disable DNA damage checkpoints. In particular, a dysfunctional G₂ checkpoint can allow a cell to enter mitosis inappropriately (73). The G₂ checkpoint prevents mitotic entry by blocking activation of the mitotic kinase Cdk1, which depends on activation of the Cdc25C phosphatase and inactivation of the Wee1 and Myt1 kinases. Cdc25C and Wee1 are regulated by the ATM (6) and ATR (74) kinases. Agents that disable the ATM and ATR pathways thus can force G₂ cells to enter mitosis prematurely. Because both E1B-55K and E4orf3 proteins inactivate ATM (75) and ATR (76), the ability of these two adenoviral proteins to prevent cells from entering mitosis or arresting in mitosis is surprising.

Insight into adenoviral control of mitotic entry was provided by Cherubini and associates, who showed that primary human cells infected with the E1B-55K mutant virus dl1520 accumulated highly condensed chromosomes after 12 h of exposure to colcemid (52). Here we show that additional adenoviruses bearing large deletions in the E1B-55K gene behaved similarly (Fig. 5A). However, among 10 E1B-55K mutant viruses bearing missense mutations (Table 2), all but H5pm4109 prevented cells from entering mitosis (Fig. 5B). We interpret these observations to mean that many E1B-55K properties, including the ability to bind the E4orf6, Daxx, and Mre11 proteins, the ability to interact with SUMO-modified proteins, and C-terminal phosphorylation, are not critical for the E1B-55K protein to block to entry into mitosis. The H5pm4109 E1B-55K protein contains a histidine in place of alanine at position 260. The identical protein expressed by the related virus ONYX-053 is unable to bind both p53 and E4orf6 (60). Because the T255A protein expressed by H5pm4108 fails to bind E4orf6 (60) but prevents entry into mitosis, we conclude that the E1B-55K protein must be able to bind p53 in order to prevent entry into mitosis. Surprisingly, neither the E1B-55K/Eorf3 double mutant virus nor the combination of colcemid and infection with the E1B-55K single mutant virus forced the p53-null H1299 cells into mitosis unless p53 was expressed by transfection (Fig. 5C). In addition to the HeLa cells studied in this report, H460 cells, MCF10A cells, and hTERT-immortalized retinal pigmented epithelial cells also contain a wild-type p53 gene. A significant number of these cells arrested in mitosis after infection with the double mutant virus, whereas the p53-null H358 and PC3 cell lines failed to show a statistically significant increase in arrested cells after infection (data not shown).

The requirement for p53 to promote mitotic entry in the adenovirus-infected cell is at odds with the well-described ability of p53 to enforce cell cycle arrest. For example, p53 precludes entry into S phase by imposing transcriptional and translational blocks to cell cycle progression in response to serum starvation (77) and prevents inappropriate entry into mitosis by suppressing expression of the chromosome alignment protein Kif23 (17). Perhaps p53 is altered in the adenovirus-infected cell, and this altered form of p53 permits cells to slip past the G₂ checkpoint despite DNA damage or chromosomal abnormalities. There is ample precedence for the corruption of normal cellular functions by adenovirus proteins. For example, the E1B-55K proteins of different adenoviruses partner with the E4orf6 protein to reprogram ubiquitin-protein ligases of the Skp1–Cul1–F-box (SCF) family (53, 54). SCF complexes orchestrate progression through the cell cycle and activate checkpoint signaling (recently reviewed in reference 78). It may be significant that the SCF complex associated with the nuclear interaction partner of ALK keeps levels of cyclin B1 low and prevents early mitotic entry (79). Furthermore, the physical interaction between the E1B-55K protein and p53 converts p53 from a transcriptional activator to a potent repressor of transcription (11, 80).

The apparent need for an interaction between the E1B-55K protein and p53 in order to prevent mitotic arrest may underlie observations suggesting that p53 enables the wild-type virus to elicit greater cytopathic effects than the E1B-55K-mutant virus (81). It was later noted that replication of the E1B-55K-null virus dl1520 was enhanced by expression of the gain-of-function R248W p53 variant that could no longer bind DNA in a site-specific manner (82). It would be of interest to determine if the interaction between the E1B-55K protein and the putative form of p53 responsible for preventing entry into mitosis subverts the transcriptional activity of p53 in a manner that phenocopies gain-of-function p53 variants. Some of these variants promote progression into mitosis in response to genotoxic stress (83). Many gain-of-function mutations in p53 map to the DNA-binding domain (84). Coincidentally, Tip60-dependent acetylation of p53 within the DNA-binding domain can determine how p53 governs cell fate in response to DNA damage (85). Recent evidence shows that Tip60 is targeted for degradation by the E1B-55K protein (54). Although the altered function of p53 is seen irrespective of E1B-55K status, other viral proteins may mimic the action of Tip60, either directly or through the action of redirected cellular proteins.

It must be emphasized that the interaction between the E1B-55K protein and p53 cannot simply ablate p53 function: otherwise p53-null cells infected with the E1B-55K/E4orf3 double mutant virus should have arrested in the mitotic-like state. If a modified form of p53 exists during infection, perhaps the E1B-55K protein allows this form of p53 to reinforce the G₂/M checkpoint during the replicative stages of the adenoviral life cycle.

The notion of a protein inhibiting the function of another while simultaneously preserving some of the target's function is not novel. The licensing factor Cdt1 is present during the G₁ and S phases. During DNA synthesis, Cdt1 is expelled from the nucleus, degraded, and inactivated to prevent rereplication. Geminin was initially identified as a mammalian factor that contributed to the inactivation of Cdt1 (86). However, it was later shown that geminin binds Cdt1 to gain nuclear localization (87) and preserve a portion of Cdt1 during late mitosis (88). Perhaps like geminin, the E1B-55K protein is able to inactivate most p53 function while retaining a low level of p53 in order to strengthen the G₂/M checkpoint.

The E1B-55K protein, through its interaction with p53 during an adenovirus infection, acts to prevent entry into mitosis. Because sustained expression of E1B-55K alone does not preclude cell division, this novel viral “checkpoint” must form only in the adenovirus-infected cell. Adenovirus-infected cells that escape this viral G₂ checkpoint and enter mitosis may be prone to arrest because of other adenoviral proteins, such as the E4orf4 protein. The E4orf4 protein inactivates APC/C by reprogramming the activity of protein phosphatase 2A, thereby inducing G₂/M arrest (89, 90). Interestingly, when expressed alone, E4orf4 promotes p53-independent, caspase-independent cell death in tumor cells (91–93) while inhibiting apoptosis in normal cells (94). Since prolonged arrest in a mitotic-like state or mitotic catastrophe is often followed by apoptosis or senescence, it would be advantageous for adenovirus to express an activity that facilitates exit from mitosis. Because colcemid fails to trap E4orf3 single mutant virus-infected cells in a mitotic-like state (Fig. 4), we conclude that E4orf3 does not prevent entry into mitosis. However, because the E4orf3 gene is required to prevent the E1B-55K-mutant virus from arresting in a mitotic-like state (Fig. 3), we suggest that E4orf3 provides an activity that facilitates exit from mitosis.

Progression through mitosis beyond anaphase requires satisfaction of the spindle assembly checkpoint and activation of the APC/C. Two coactivators of APC/C, Cdc20 and Cdh1, dictate substrate specificity and activity. Cdc20 is sequestered by mitotic checkpoint proteins, while unattached kinetochores or perturbations in the mitotic spindle persist. APC/C^Cdc20 initiates anaphase by targeting the cohesins for degradation, while APC/C^Cdh1 promotes the irreversible exit from mitosis into G₁ (reviewed in reference 78). A key target of both APC/C^Cdc20 and APC/C^Cdh1 is the mitotic cyclin B1, whose loss leads to the rapid decline in Cdk1 activity (22). Because cyclin B1 accumulates to high levels in adenovirus-infected cells (Fig. 6), E4orf3 does not promote mitotic exit by simply removing cyclin B1. However, the E4orf3 protein may functionally inactivate cyclin B1 by mislocalizing this protein within the infected cell (Fig. 7). Cells infected with E4orf3-mutant viruses also contain a cyclin B-related protein (Fig. 6) similar in size to a cleaved form of cyclin B1 that sustains the mitotic block (34). In the context of an E1B-55K deletion, E4orf3 may prevent accumulation of this inhibitory product.

E4orf3 inactivates many cellular proteins by altering their localization or directing them to the aggresome (95, 96). For example, mislocalization of Nbs1 by the E4orf3 protein is sufficient to prevent ATR activation, thereby crippling the DNA damage response (76, 97). The activity of cyclin B1 in association with the major mitotic kinase Cdk1 is exquisitely controlled by its intracellular localization, with roles in both the cytoplasm and the nucleus (22). In the infected cell, E4orf3 may promote aggregation of cyclin B1 (Fig. 7). When expressed by transfection, E4orf3 may diminish nuclear levels of cyclin B1 (Fig. 8). E4orf3 may functionally inactivate cyclin B1 to facilitate exit from mitosis. In this manner, E4orf3 may replace a cellular process that was disabled by other adenovirus proteins in order to prevent the untoward consequences of mitotic arrest.

A remaining question is, why are cells infected in early G₁ predisposed to arrest in mitosis after infection with the double mutant virus? Cellular chromatin in the early G₁ cell is distinguished from that in other phases of the cell cycle by the need to reacquire specific proteins. At late stages of mitosis, many proteins, including those that signal DNA damage, are displaced from the condensing chromatin. Consequently, G₁ cells are able to respond to DNA damage only after these proteins reassociate with DNA (98, 99). The reacquisition of chromatin proteins can also be regulated by proteins such as the histone variant H2A.Z, which suppresses transcription during mitosis (100). The histone acetyltransferase Tip60 promotes H2A.Z-mediated association of proteins at sites of DNA damage. Perhaps the ability of the E1B-55K protein to target Tip60 for destruction (54) renders cellular chromatin less able to signal DNA damage and thus unsuited for further progression through the cell cycle. During late mitosis and early G₁, chromatin is also reloaded with replication-critical licensing factors that were displaced or degraded in S phase (101, 102). Early events during the adenovirus infectious cycle may perturb these processes that occur during G₁.

Studies using somatic cell nuclear transfer to generate cloned animal offspring demonstrate how the unique nature of cells in early G₁ can exert consequences over a very long time. Donor cell nuclei derived from serum-starved G₀ cells or early G₁ cells were transferred into enucleated bovine ova to create embryos. Although embryos derived from G₀ nuclei showed greater survival through the blastocyst stage, animals derived from G₁ nuclei showed improved birth weights and greater postnatal survival (103). Both the transfer of an early G₁ nucleus and the infection of a cell in early G₁ have consequences evident much later in time.

While the properties of early G₁ cells that predisposed them to mitotic arrest following infection with adenovirus are unknown, these cells have the potential to enter mitosis despite suspension of normal cell cycle progression. Surprisingly, entry into mitosis by the virus-infected cell requires p53, a protein better known for its ability to halt cell cycle progression. The E1B-55K and E4orf3 proteins function independently of one another to prevent arrest in mitosis. Although adenovirus directs the degradation of p53, it appears that a portion of the p53 protein is repurposed by the E1B-55K protein to reinforce the G₂/M checkpoint. Through the reorganization of cyclin B1, E4orf3 protein may serve as a fail-safe to facilitate exit from mitosis, presumably circumventing cell death associated with prolonged mitotic arrest.