The Adenoviral Oncogene E1A-13S Interacts with a Specific Isoform of the Tumor Suppressor PML To Enhance Viral Transcription

Cell lines.

HepaRG (58), HALP/HALP PML-II (59), and H1299 (60) cells and mouse embryonic fibroblasts (MEF) (61) were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) and 100 U of penicillin and 100 μg of streptomycin per ml in a 5% CO₂ atmosphere at 37°C. For HepaRG cells, the medium was additionally supplemented with 5 μg/ml of bovine insulin and 0.5 μM hydrocortisone.

Plasmids and transient transfections.

Ad5 E1A-12S, Ad5 E1A-13S, and the Ad5 E1A mutant proteins were expressed from pcDNA3 vector. N-terminal flag-tagged human PML-isoforms I to VI were expressed from the pLKO.1-puro vector and were kindly provided by Roger Everett (Glasgow, United Kingdom). All PML isoforms are named according to the nomenclature of Jensen et al. (21): PML-I (AAG50180), PML-II (AF230410), PML-III (S50913), PML-IV (AAG50185), PML-V (AAG50181), and PML-VI (AAG50184). The PML-II SIM mutation was introduced by site-directed mutagenesis using forward primer 5′-GGAACGCGGTGGGGGGATCAGCAGC-3′ and reverse primer 5′-GCTGCTGATCCCCCCACCGCGTTCC-3′ (see also Fig. 8A). For transient transfection, subconfluent cells were treated with a mixture of DNA and 25-kDa linear polyethylenimine (Polysciences) as described previously (62).

Viruses.

H5pg4100 served as wild-type virus (63). H5dl347 and H5dl348 carry cloned segments corresponding to E1A-12S and E1A-13S mRNAs, respectively, in place of the E1A gene. H5dl312 lacks a large segment of the E1A gene and therefore does not produce E1A products (64). H5pm4150 carries a frameshift mutation in the E4orf3 open reading frame and H5pm4149 carries stop codons in the E1B-55K open reading frame, so they do not express E4orf3 and E1B-55K, respectively (65, 66). All viruses were propagated, and titers were determined as fluorescence-forming units (FFU), in HEK293 monolayer cultures (63). To measure virus growth, infected cells were harvested at 48 h postinfection (p.i.) and lysed by three cycles of freeze-thawing. The cell lysates were serially diluted in DMEM for infection of HEK293 cells, and virus yield was determined by quantitative E2A immunofluorescence staining at 24 h after infection.

Luciferase reporter assay.

For dual-luciferase assays, subconfluent H1299 cells were transfected as described above, using 1 μg of reporter (pGL-C3G5-luc or pGL-E2-early promoter), 1 μg of pRL-TK (Promega), which expresses Renilla luciferase under the control of the herpes simplex virus (HSV) thymidine kinase promoter, and 1 μg of effector plasmids (E1A-12S, E1A-13S, PML-II, and pG4-p300). Cell extracts were prepared, measured, and normalized as described recently (67).

Antibodies and protein analysis.

Primary antibodies specific for Ad proteins included E1A mouse monoclonal antibody (MAb) M58 and E1A mouse MAb M73 (68) and E2A mouse MAb B6-8 (69). E1A rabbit MAb 610 was a kind gift of Roger Grand. Primary antibodies specific for cellular and epitopically expressed proteins included PML rabbit polyclonal antibody (PAb) NB100-59787 (Novus Biologicals, Inc.), PML mouse MAb clone 36.1-104 (Millipore), flag mouse MAb flag-M2 (Sigma-Aldrich, Inc.), and β-actin mouse MAb AC-15 (Sigma-Aldrich, Inc.).

All protein extracts were prepared in radioimmunoprecipitation assay (RIPA) lysis buffer as published recently (70). For immunoprecipitation, flag-M2-coupled protein A-Sepharose beads (Sigma-Aldrich, Inc.) were used or protein A-Sepharose (3 mg/immunoprecipitation) was coupled with 1 μg of MAb for 1 h at 4°C. The Ab-coupled protein A-Sepharose was added to pansorbin-Sepharose (50 μl/lysate; Calbiochem)-precleared extracts and rotated for 2 h at 4°C. Proteins bound to the Ab-coupled protein A-Sepharose were precipitated by centrifugation, washed three times, boiled for 3 min at 95°C in 2× Laemmli buffer, and analyzed by immunoblotting exactly as described recently (62).

Indirect immunofluorescence.

For indirect immunofluorescence, cells were grown on glass coverslips as already published (71). To visualize intranuclear E1A staining, cells were extracted with 0.5% Triton X-100 in CSK buffer as described by Carvalho and coworkers (72). Cells were fixed in 4% paraformaldehyde (PFA) at 4°C for 20 min and permeabilized in phosphate-buffered saline (PBS)–0.5% Triton X-100 for 30 min at room temperature. After 1 h of blocking in Tris-buffered saline–bovine serum albumin-glycine (TBS-BG) buffer, coverslips were treated for 1 h with the primary antibody diluted in PBS, washed three times in PBS–0.1% Tween 20, and then incubated with the corresponding Alexa 488 (Invitrogen)- or Cy3 (Dianova)-conjugated secondary antibodies. Coverslips were washed three times in PBS–0.1% Tween 20 and mounted in Glow medium (Energene), and digital images were acquired with a confocal laser scanning microscope (Zeiss-CLSM-510). Images were cropped using Adobe Photoshop CS4 and assembled with Adobe Illustrator CS4.

Quantitative RT-PCR analysis.

Subconfluent HepaRG cells were infected with wild-type virus at a multiplicity of 50 FFU/cell and harvested 24 h p.i. Total RNA was isolated with TRIzol reagent (Invitrogen) as described by the manufacturer. The amount of total RNA was measured, and 1 μg of RNA was reverse transcribed using the reverse transcription system from Promega. To amplify specific viral genes, primers were designed as follows: 18S rRNA forward primer, 5′-CGGCTACCACATCCAAGGAA-3′; 18S rRNA reverse primer, 5′-GCTGGAATTACCGCGGCT-3′; E2A forward primer 5′-GAAATTACGGTGATGAACCCG-3′; and E2A reverse primer, 5′-CAGCCTCCATGCCCTTCTCC-3′. Quantitative reverse transcription-PCR (RT-PCR) was performed with a first-strand method in a Rotor-Gene 6000 (Corbett Life Sciences, Sydney, Australia) in 0.5-ml reaction tubes containing a 1/50 dilution of the cDNA template, 10 pmol/μl of each synthetic oligonucleotide primer, and 5 μl/sample SensiMix SYBR (Bioline). The PCR conditions were as follows: 10 min at 95°C, followed by 40 cycles of 30 s at 95°C, 30 s at 62°C, and 30 s at 72°C. The average threshold cycle (C_T) value was determined from triplicate reactions, and levels of viral mRNA relative to cellular 18S rRNA were calculated as described recently. The identities of the products obtained were confirmed by melting curve analysis.

Ad5 E1A-13S localizes to endogenous PML in transiently transfected cells.

Although reports have proposed different functions of E1A-12S and E1A-13S during adenoviral infection (73), E1A has been described to localize to PML nuclear bodies (72). In this context, we decided to express both E1A isoforms separately and visualize their localization with respect to endogenous PML bodies. Both isoforms localized to the nucleus in large amounts (Fig. 1A, panels g and k), whereas PML exhibited the characteristic punctuate nuclear staining (Fig. 1A, panels f and j). Consistent with published results, E1A-12S as well as E1A-13S relocalized PML bodies, causing them to appear larger and fewer in number (74).

Since the large amounts of intranuclear E1A make it very difficult to observe any distinct staining patterns, the cells were treated as described in Materials and Methods (“Indirect immunofluorescence”) prior to fixation (Fig. 1B) (72). After preextraction, E1A-13S accumulates in several brightly stained nuclear aggregates juxtaposed to endogenous PML in ∼30% of the transfected cells (Fig. 1B, panels i to l). For the smaller E1A-12S isoform, no staining juxtaposed to PML could be observed (Fig. 1B, panels e to h). Our observations are consistent with previous reports showing that transcriptionally active proteins often tend to associate with the nuclear matrix (75).

Ad5 E1A-13S localizes to PML nuclear bodies during adenoviral infection.

Detailed studies have revealed an astonishing amount of cross talk between adenoviral proteins in the modulation of host cell factors, e.g., PML and PML-associated proteins. Previously published data illustrate that the viral protein E4orf3 is necessary and sufficient to disrupt the structure of cellular PML NBs during Ad5 infection (72, 76–78). In addition, viral early E1B-55K protein has been shown to interact and functionally cooperate with PML bodies (79–83).

To evaluate whether other adenoviral proteins interfere with or alter the cooperation between E1A and PML, we further analyzed localization during productive infection (Fig. 2). Cells were infected with wild-type virus (H5pg4100) and virus mutants expressing E1A-12S (H5dl347) or E1A-13S (H5dl348). To exclude expression of other adenoviral proteins mediating PML interaction, viruses lacking E1B-55K (H5pm4149) or E4orf3 (H5pm4150) were included as controls.

As expected, PML was relocalized (Fig. 2b, f, j, n, and r) into so-called track-like structures in infected cells. The virus mutant lacking E4orf3 protein (H5pm4150) was no longer able to relocalize PML bodies, although their shape was different from the punctate staining in mock-infected cells (Fig. 2b and v). We detected E1A colocalization with PML track-like structures in wild-type virus (H5pg4100)- as well as H5dl348-infected cells expressing only E1A-13S (Fig. 2g and o), whereas E1A-12S (H5dl347) is distributed much more diffusely in the host cell nucleus (Fig. 2k). E1A also localized to PML after infection with virus lacking E4orf3 or E1B-55K (Fig. 2t and x). Based on these observations, we conclude that E1A-13S localizes to PML nuclear bodies independently of E4orf3 or E1B-55K expression.

Ad5 E1A-13S specifically interacts with the human PML-II isoform in transiently transfected cells.

Human cells express at least seven distinguishable PML isoforms (21), of which PML-I to -VI possess a nuclear localization sequence. To ascertain whether E1A specifically interacts with different PML isoforms, human cells were transfected with E1A-13S or E1A-12S and plasmids encoding human PML isoforms I to VI. By precipitating flag-tagged PML and subsequently staining for E1A, we detected a highly specific interaction between E1A-13S and PML isoform II (Fig. 3D, lane 4). E1A-12S could not be coprecipitated with any PML isoform (Fig. 3B), although the steady-state levels showed equal amounts of transfected PML and E1A isoforms (Fig. 3A and C).

Mutational screening of E1A-13S reveals PML-II isoform interaction within CR3.

Since E1A-13S differs from E1A-12S only in CR3, composed of 46 aa, we assumed that PML-II binding depends on this or adjacent regions in the E1A protein. To more precisely identify the binding domain required for E1A-13S/PML-II interaction, we constructed E1A-13S mutants with overlapping 14- to 16-aa deletions spanning the proposed region (Fig. 4A). Although the E1A-13S mutants had slightly altered gel migration properties, input levels of E1A-13S and PML-II were equal in our coimmunoprecipitation assays (Fig. 4B). These experiments revealed that deletion mutant 7 (E1A-13SΔ176-191) (Fig. 4A) exhibits strongly impaired PML-II binding properties (Fig. 4C, lane 11).

To further narrow down the interaction region, a shorter deletion mutant, E1A-13SΔ182-191, was constructed and tested for PML-II binding (Fig. 5A). We confirmed binding deficiency of deletion mutant E1A-13SΔ182-191 with the PML-II isoform by immunofluorescence analysis and coimmunoprecipitation experiments (data not shown).

Ad5 E1A-13S interacts with endogenous PML in plasmid-transfected and infected cells.

Next, we verified our observations by coimmunoprecipitation experiments with endogenous PML. We transfected human cells with E1A-12S, E1A-13S, or E1A-13SΔ182-191 or infected them with wild-type virus (H5pg4100) and virus mutants expressing only E1A-12S (H5dl347), only E1A-13S (H5dl348), or no E1A protein (H5dl312) as a negative control (Fig. 5B). After precipitation of endogenous PML and subsequent staining for E1A, we could detect specific interaction of endogenous PML with transfected E1A-13S as well as viral E1A in H5pg4100- and H5dl348-infected cells (Fig. 5C, lanes 3, 5, and 6). In contrast, E1A-12S, deletion mutant E1A-13SΔ182-191, and viral E1A encoded by the virus mutant lacking E1A-13S (H5dl347) could not be precipitated with endogenous PML (Fig. 5C).

In sum, interaction of E1A-13S with PML could be mapped to aa 182 to 191 at the end of the CR3 of E1A-13S. This binding region (aa 182 to 191) is highly conserved in different Ad types and has previously been identified as the variable transcription factor binding domain or promoter-targeting domain of E1A-13S (84–86).

Ad5 E1A-13S and PML cooperation impacts viral and cellular transcription.

PML bodies have been implicated in regulating transcription (30, 86). Furthermore, it is known that several viral genomes, including that of adenovirus, localize close to PML bodies in order to start transcription and replication (53, 87). Therefore, E1A likely affects transcriptional regulation from adenoviral promoters in cooperation with PML-II.

To address this idea, we used luciferase reporter assays and transfected luciferase-dependent plasmids carrying adenoviral promoters (67) together with E1A-13S and PML-II (Fig. 6). PML-II alone was not able to stimulate transcription from the Ad5 E2 early promoter, whereas E1A-13S activated luciferase activity from the viral promoter 10-fold. Furthermore, PML-II efficiently stimulated transcription from the Ad5 promoter when coexpressed with E1A-13S. This was further verified by the positive correlation between increasing levels of expressed PML-II and activated transcription at constant E1A-13S levels (Fig. 6B). As a control we also included E1A-12S and E1A-13SΔ182-191 in this reporter gene assay. We observed that neither E1A-12S nor E1A-13SΔ182-191 stimulated transcription from the adenoviral promoter in combination with equal amounts of PML-II (Fig. 6B).

Recently, Pelka et al. reported that cellular transcription factor p300 binds E1A-13S within the CR3 region and is recruited to Ad promoters during infection (88). p300 possesses intrinsic acetyltransferase activity and is therefore an important transcriptional coactivator; it is recruited to gene promoters via its association with numerous DNA binding transcription factors (89). PML was also found to interact with p300/CBP and regulate transcription (90, 91). In this context, we tested whether E1A-13S and PML-II affect p300 coactivator function in transcriptional regulation. We performed luciferase reporter assays with a GAL4-responsive promoter and p300 fused to a GAL4 DNA binding domain, in combination with E1A-13S and PML-II (Fig. 7A). PML-II alone only slightly stimulated transcription from the GAL4 promoter, whereas E1A-13S enhanced p300 transcriptional activity 2-fold. Consistent with our observations described above, coexpression of E1A-13S with PML-II strongly stimulated the luciferase activity from the promoter. Cotransfection of increasing PML-II amounts further stimulated p300-dependent transcriptional activation (Fig. 7B). On the other hand, coexpression of PML-II together with E1A-13SΔ182-191 did not activate transcription from the GAL4 promoter (Fig. 7B).

Additionally, we performed the reporter gene assay shown in Fig. 6 and 7 with the other nuclear PML isoforms. Our results showed that other PML isoforms do positively affect E1A-dependent transactivation, although to a lesser extent than PML-II (Fig. 6C and 7C).

Taken together, our results on transcriptional regulation suggest that PML-II is a dose-dependent coactivator of E1A-13S-dependent transactivation. PML-II not only is able to modulate transcription from Ad5 promoters in combination with E1A-13S but also stimulates cellular gene expression by affecting key cellular coactivators such as p300.

PML-II positively regulates adenovirus progeny production in human cells.

To reveal the effect of PML-II on the production of adenoviral progeny, we determined virus growth in HepaRG cell lines in which endogenous PML was knocked down (HALP cells) and the short hairpin RNA (shRNA)-resistant PML-II isoform was reconstituted (HALP PML-II) (59). In accordance with recent reports, PML knockdown showed a modest effect on adenovirus replication compared to that in the parental cell line (80, 92). However, reconstitution of the single isoform PML-II enhanced virus yield severalfold compared to that in the parental cell line (Fig. 8). Consistent with the data obtained from the reporter assays, this strongly indicates that PML-II is a positive regulator of Ad5 replication in human cells. To further test the positive effect of PML isoforms, we measured early gene expression in an abortive infection in PML-null mouse embryonic fibroblasts (MEF) to confirm the data obtained from the reporter gene experiments. In this context, we performed time course experiments with PML^−/− MEF compared to PML^+/+ MEF cells. We observed a delay in the protein expression of E2A (Fig. 9A) as well as reduced E2A mRNA expression in MEF cells without PML (Fig. 9B).

Transcriptional coactivation by PML does not depend on its functional SIM.

PML formation of PML NBs, its dynamic behavior, and its mode of function are based mainly on SUMOylation of the PML protein itself (24, 26, 93). The PML SIM is required to form PML NBs when exogenously expressed in PML^−/− cells and enables PML to interact noncovalently with SUMOylated proteins, including PML itself (24). In line with these observations, we were interested in whether the PML SIM, which is necessary for interacting with other SUMOylated factors in the PML NBs, plays a role in transcriptional activation of Ad promoters by PML-II together with E1A-13S (Fig. 10A). Therefore, we generated a PML-II mutant with a nonfunctional SIM. Interestingly, this mutant stimulated E1A-13S-dependent transcription from the Ad5 E2 early promoter more efficiently than wild type PML-II (Fig. 10B). We also tested whether the PML-II SIM mutant affected E1A-13S binding properties. The PML-II SIM mutant and wild-type PML-II were expressed at equal amounts (Fig. 10C), and coimmunoprecipitation experiments showed no differences in binding capabilities between wild-type PML-II and the PML-II SIM mutant (Fig. 10D, lanes 3 and 5).