|Home | About | Journals | Submit | Contact Us | Français|
The ability of adenovirus early region proteins, E1B-55K and E4orf6, to usurp control of cellular ubiquitin ligases and target proteins for proteasome-dependent degradation during infection is well established. Here we show that the E4 gene product, E4orf3 can, independently of E1B-55K and E4orf6, target the transcriptional corepressor transcriptional intermediary factor 1γ (TIF1γ) for proteasome-mediated degradation during infection. Initial mass spectrometric studies identified TIF1 family members—TIF1α, TIF1β, and TIF1γ—as E1B-55K-binding proteins in both transformed and infected cells, but analyses revealed that, akin to TIF1α, TIF1γ is reorganized in an E4orf3-dependent manner to promyelocytic leukemia protein-containing nuclear tracks during infection. The use of a number of different adenovirus early region mutants identified the specific and sole requirement for E4orf3 in mediating TIF1γ degradation. Further analyses revealed that TIF1γ is targeted for degradation by a number of divergent human adenoviruses, suggesting that the ability of E4orf3 to regulate TIF1γ expression is evolutionarily conserved. We also determined that E4orf3 does not utilize the Cullin-based ubiquitin ligases, CRL2 and CRL5, or the TIF1α ubiquitin ligase in order to promote TIF1γ degradation. Further studies suggested that TIF1γ possesses antiviral activity and limits adenovirus early and late gene product expression during infection. Indeed, TIF1γ knockdown accelerates the adenovirus-mediated degradation of MRE11, while TIF1γ overexpression delays the adenovirus-mediated degradation of MRE11. Taken together, these studies have identified novel adenovirus targets and have established a new role for the E4orf3 protein during infection.
Human adenoviruses (Ad) are small, nonenveloped viruses with a linear double-stranded DNA genome and are classified into species A to F according to various criteria (7). Since the observation that Ad12 could induce tumors in newborn rodents, Ad has served as a reliable model for dissecting the molecular basis of the key cellular signaling pathways that underlie the transformation process (28, 33, 69, 70). Studies investigating the roles of the Ad early region proteins in both Ad-transformed and Ad-infected cells have led to key advances in the understanding of basic cellular processes and how Ad usurps control of these pathways in order to promote viral replication (8, 33, 67).
The Ad early region proteins E1B-55K, E4orf3, and E4orf6 have a complex inter-relationship and serve together to regulate RNA processing, late viral mRNA nuclear export, the shutoff of host-cell protein synthesis, and neutralization of the host cell DNA damage response during infection (4, 29, 57, 61, 67, 73). They can also function synergistically and cooperate with E1A to promote Ad-induced cellular transformation (47–49). It is perhaps not surprising, therefore, that they share many common functions. For instance, E1B-55K interacts directly with p53 to repress transcriptional activity and also promotes p53 sumoylation and targeting to cytoplasmic aggresomes for degradation (41, 46, 53, 76, 77). E4orf6 also interacts directly with p53 to repress p53 transcriptional activity, while E4orf3 inhibits p53 function, by inducing the selective trimethylation of histone H3 K9 at p53 promoters and preventing p53 association with p53-responsive promoter elements (24, 62).
Ad E1B-55K and E4orf6 interact directly and cooperate functionally during infection. It has been established that Ad5 E1B-55K/E4orf6 recruit subcomplexes of the Cullin 5-containing ubiquitin ligase (CRL5), minimally comprising CUL5, elongins B and C, and Rbx1, to promote the ubiquitin- and proteasome-dependent degradation of p53 (35, 54). A model has been proposed whereby E1B-55K serves as substrate receptor for p53, while E4orf6 recruits functional CRL5 to the E1B-55K/E4orf6 complex through elongin-interacting BC boxes within its primary sequence (11, 54). Targeting of CRL5 by Ad E1B-55K/E4orf6 has also been shown to be important for viral mRNA export (12, 75). Further studies with Ad5 have indicated that E1B-55K and E4orf6 cooperate to promote the degradation of the MRE11-RAD50-NBS1 (MRN) component, MRE11, in order to inhibit ATM and ATR activation and also promote the degradation of DNA ligase IV in order to prevent nonhomologous end joining (5, 18, 64). The E1B-55K binding protein, E1B-AP5 (hnRNPUL1), also plays an important role in regulating ATR during Ad5 and Ad12 infection (9). By targeting ATM, ATR, and DNA ligase IV, E1B-55K and E4orf6 prevent the concatenation of linear double-stranded viral DNA during infection (73). More recent work suggests that E1B-55K/E4orf6 also targets the BLM helicase for degradation in order to inhibit DNA damage repair pathways and integrin α3 to possibly inhibit viral reinfection (23, 51).
It appears, however, that the relationship between E1B-55K and E4orf6 and the ubiquitin-proteasome pathway is more complicated than previously thought. It has been determined, for instance, that Ad12 and Ad40 E1B-55K and E4orf6 utilize exclusively the CUL2-containing CRL2 to promote the degradation of p53 during infection, while Ad16 can utilize either CRL2 or CRL5 (10, 22). Further investigation has revealed that different Ad serotypes have evolved different strategies in order to neutralize the DNA damage response during infection such that the cohort of substrates targeted for degradation varies between viral serotypes (22, 32). In this regard, all Ads studied to date promote the degradation of DNA ligase IV, while they do not necessarily target p53 or MRE11 for degradation; the reasons for these differences await further investigation. Further complexity has been established, such that Ad12 E4orf6 can, independently of Ad12 E1B-55K, utilize CRL2 to promote the degradation of the ATR activator, TOPBP1, and inhibit CHK1 activation during infection (10). In this regard, Ad12 E4orf6 not only recruits CRL2 but also acts as a substrate receptor for TOPBP1 (10). Similarly, Ad5 E1B-55K can function independently of E4orf6 to recruit CRL5 and target the promyelocytic leukemia (PML) nuclear body component, Daxx, for proteasome-dependent degradation (59). Indeed, the SUMO-1-conjugated E1B-55K-mediated degradation of Daxx is important for the ability of Ad to promote cellular transformation (60).
It has been suggested that E4orf6 shares a number of redundant functions with the E4orf3 protein during infection. Indeed, these proteins have been shown to enhance viral DNA replication, initiate shutoff host-cell protein synthesis, stabilize the nuclear accumulation of late viral mRNAs, and promote late viral protein synthesis, enhancing the production of new virus particles (15, 16, 34, 39). Early work with E4orf3 indicated that it associated with the PML protein and was solely responsible for reorganizing PML-containing nuclear bodies (also called PML oncogenic domains or ND10) into distinctive nuclear track-like structures (19, 25). The importance of PML reorganization during infection was highlighted by observations indicating that overexpression of PML prevented the E4orf3-mediated reorganization of PML bodies and severely delayed adenovirus replication (25). Other work has established that E4orf3 targets the PML II isoform specifically in order to reorganize PML bodies into nuclear tracks and also recruits E1B-55K to these tracks during infection (38, 43). More recent work has determined that the transcriptional intermediary factor 1 (TIF1) family member, TIF1α, is also recruited in an E4orf3-dependent manner to PML nuclear tracks during infection (78). Although E4orf3s from the divergent Ad serotypes Ad4, Ad5, Ad9, and Ad12 all recruit TIF1α to PML tracks, the functional importance of TIF1α reorganization during infection remains to be established (78). E4orf3, akin to E1B-55K/E4orf6, also targets DNA damage response and repair pathways during infection in order to prevent viral DNA concatenation and double-strand break repair (DSBR). Although not conserved among other Ad serotypes, Ad5 E4orf3 sequesters MRN components in PML nuclear tracks prior to their degradation in cytoplasmic aggresomes, and E4orf3, like E4orf6, also interacts with the DNA-PK catalytic subunit, presumably to inhibit DSBR during infection (14, 18, 30, 42, 65).
The TRIM/RBCC family of proteins are characterized by an N-terminal tripartite motif that contains a RING finger moiety, one or two zinc-binding motifs named B-boxes, and a coiled-coil domain which is necessary for oligomerization and association with subcellular structures (45, 55). The most prominent TRIM family member is TRIM19, the PML tumor suppressor protein, which is an integral component of PML nuclear bodies and involved in a number of diverse cellular processes, including cellular senescence, apoptosis, and DNA repair (13). TRIM proteins can, based on homologies, be divided into two main groups, and subdivided into 11 subfamilies (52). The TIF1 subfamily contains four members in mammals α (TRIM24), β (TRIM28), γ (TRIM33), and δ (TRIM66). TIF1 family members possess an N-terminal tripartite motif, C-terminal plant homeobox (PHD) and bromodomains, and a unique TIF1 signature sequence which is likely to participate in TIF1-dependent transcriptional regulation (72).
Here we report that TIF1 family members are evolutionarily conserved targets for Ad early region proteins E1B-55K and E4orf3 during infection. We have determined that E1B-55K associates with TIF1α, TIF1β, and TIF1γ in Ad5- and Ad12-transformed and infected cells and that E4orf3 can, independent of E1B-55K, selectively reorganize TIF1α and TIF1γ to PML nuclear bodies during infection. We have also identified a new function for E4orf3, namely, the ability to promote, independent of E1B-55K and E4orf6, the proteasome-mediated degradation of TIF1γ. We show that TIF1γ possesses antiviral activities and suggest that Ad ablates these activities by targeting TIF1γ for degradation during infection. Taken together, these results identify TIF1 proteins as major targets for Ad during infection and establish that E4orf3 utilizes the ubiquitin-proteasome pathway in order to promote the degradation of TIF1γ.
Ad12 human embryo retinoblast (HER2) cells, Ad5 human embryo kidney 293 (HEK293) cells, A549 human small cell lung carcinoma cells, and HeLa human cervical carcinoma cells were all grown in HEPES-modified Dulbecco modified Eagle medium (DMEM; Sigma-Aldrich) supplemented with 8% (vol/vol) fetal calf serum (FCS) and 2 mM l-glutamine (Invitrogen). All cells were kept at 37°C in a humidified 5% CO2 atmosphere. A549 and HeLa cells were used interchangeably throughout the present study as reliable cell models for studying Ad infection and viral gene function.
Human wild-type (wt) Ad5 and wt Ad12 Huie viruses were from the American Type Culture Collection, while wt Ad3, wt Ad7, and wt Ad11 were generously provided by Joe Mymryk. The Ad5 and Ad12 E1B-55K deletion viruses (Ad5 dl1520, Ad12 dl620, and Ad12 hr703) and the E4orf3 (H5pm4150), E4orf6 (H5pm4154), and E4orf3/E4orf6 (H5pm4155) deletion viruses have all been described previously (6, 12, 17, 32, 63). Ad5 and Ad12 viruses were propagated on permissive HEK293 cells and HER3 cells, respectively, and titered by plaque assay on HER911 and HER3 cells, respectively.
Ad5 expression constructs for E1B-55K, HA-E4orf3, and HA-E4orf6 have all been described previously (47–49, 56). FLAG-TIF1γ, Ad12 E1B-55K, HA-E4orf3, and HA-E4orf6 were cloned into pcDNA3 for mammalian expression. Plasmids were transfected into cells using Lipofectamine 2000 according to the manufacturer's instructions.
The monoclonal antibodies against Ad5 E1A (M58), Ad12 E1A (#13), Ad5 E1B-55K (2A6), Ad12 E1B-55K (XPH9), Ad5 E4orf3 (6A11), Ad5 E4orf6 (RSA3), p53 (DO-1), and hemagglutinin (HA) tag were all obtained as supernatant fluid from cultures of the appropriate hybridoma cell lines. The anti-TIF1γ, anti-Ad5 E1B-55K, and anti-Ad12 E1B-55K polyclonal antibodies were raised against respective glutathione S-transferase fusion proteins in conjunction with Eurogentec. Anti-TIF1α, TIF1β, p-TIF1β S824, and CUL5 antibodies were obtained from Bethyl Laboratories. Anti-RPA32 antibodies were from Calbiochem, anti-MRE11 and anti-NBS1 antibodies were from GeneTex, the CUL2 antibody was from Abcam, and the anti-β-actin and anti-FLAG antibodies were from Sigma. The TOPBP1 and DNA ligase IV antibodies were gifts from Iain Morgan and Stephen Jackson, respectively. The anti-Ad5 penton and fiber antibody was from Vivien Mautner. Horseradish peroxidase-conjugated secondary anti-mouse and anti-rabbit antibodies used for Western blotting were from Dako. Secondary anti-mouse and anti-rabbit Alexa 488/546 antibodies used for immunofluorescence were from Molecular Probes.
Infection was carried out using subconfluent monolayers of cultured cells in serum-free HEPES-buffered DMEM containing 2 mM l-glutamine. Prior to infection, the cells were washed twice in warmed phosphate-buffered saline (PBS), before addition of the virus at the appropriate infectivity ratio. After 2 h at 37°C, medium containing virus was removed and replaced with HEPES-buffered DMEM supplemented with 8% (vol/vol) FCS and 2 mM l-glutamine.
The cells were washed twice in ice-cold saline, and whole-cell extracts were prepared in immunoprecipitation lysis buffer (50 mM Tris-HCl [pH 7.4], 1% [vol/vol] Nonidet-P40, 0.825 M NaCl). Lysates were clarified by sonication and centrifugation. Immunoprecipitation was carried out overnight at 4°C, typically with 10 μg of antibody per 5 mg of cell lysate. Antigen-antibody complexes were isolated using protein G-Sepharose (Sigma-Aldrich). Immunoprecipitates were washed five times in lysis buffer, eluted in sample buffer (9 M urea, 50 mM Tris-HCl [pH 7.4], and 0.15 M β-mercaptoethanol–10% sodium dodecyl sulfate [SDS; 2:1, vol/vol] containing 0.1% [wt/vol] bromophenol blue), and then separated by SDS-PAGE.
Cells were washed twice in ice-cold saline, and whole-cell extracts were prepared in lysis buffer containing 9 M urea, 50 mM Tris-HCl (pH 7.4), and 0.15 M β-mercaptoethanol. Samples were sonicated and then cleared by centrifugation. Protein concentrations were subsequently determined by Bradford assay (Bio-Rad). Typically, 50-μg protein samples and immunoprecipitated samples were separated on 12% (wt/vol) polyacrylamide gels, run in the presence of 0.1 M Tris, 0.1 M Bicine, and 0.1% (wt/vol) SDS. Separated proteins were transferred onto nitrocellulose filters (Pall), followed by incubation for 1 h in TBST (0.1% [vol/vol] Tween 20 in Tris-buffered saline containing 150 mM NaCl and 20 mM Tris-HCl [pH 7.3]) buffer containing 5% (wt/vol) dried milk powder. Nitrocellulose filters were subsequently incubated in TBST-milk with the appropriate primary and secondary antibodies. Antigens were visualized by chemiluminescence.
The cells were grown on glass slides and infected with either wt Ad5 or wt Ad12 at 10 PFU/cell. At the appropriate time postinfection, cells were washed with PBS, treated with pre-extraction buffer (10 mM PIPES [pH 6.8], 20 mM NaCl, 300 mM sucrose, 3 mM MgCl2, 0.5% Triton X-100) for 5 min, and then fixed in 4% (wt/vol) paraformaldehyde in PBS for 10 min. The cells were then rehydrated in PBS and blocked in blocking buffer (10% [vol/vol] FCS in PBS) for 1 h. After further washing in PBS, the cells were incubated with primary antibodies in 0.1% (vol/vol) FCS in PBS for 2 h. The slides were washed three times in PBS before incubation with secondary antibodies for 1 h. The cells were washed further in PBS and then mounted in Vectashield mounting medium containing DAPI (4′,6′-diamidino-2-phenylindole; Vector Laboratories). Images were obtained using a Zeiss LSM510-Meta laser-scanning confocal microscope and processed with associated software. In instances where images of single cells are presented, these images are representative of the population of cells studied.
TIF1α small interfering RNA (siRNA) oligonucleotides (CGACUGAUUACAUACCGGUtt) and TIF1γ siRNA oligonucleotides (CCUGCAUCUAGAAAGUGAAdTdT) were purchased from Ambion. CUL2 and CUL5 smart pools were purchased from Dharmacon: the CUL2 sequences were GGAAGUGCAUGGUAAAUUU, CAUCCAAGUUCAUAUACUA, GCAGAAAGACACACCACAA, and UGGUUUACCUCAUAUGAUU, and the CUL5 sequences were GACACGACGUCUUAUAUUA, GCAAAUAGAGUGGCUAAUA, UAAACAAGCUUGCUAGAAU, and CGUCUAAUCUGUUAAAGAA. AllStars Negative Control siRNA was purchased from Qiagen. The cells were plated 24 h prior to transfection, so as to be 30% confluent the next day. Oligofectamine reagent (Invitrogen) was used to deliver siRNA duplexes into cells according to the manufacturer's instructions. The cells were then incubated for 48 h prior to viral infection.
Protein bands detected by Coomassie Brilliant Blue G-250 staining were excised and then washed twice with a solution containing 50 mM ammonium bicarbonate and 50% (vol/vol) acetonitrile by agitation for 45 min at 37°C. The protein bands were then incubated under reducing conditions (50 mM dithiothreitol and 50 mM ammonium bicarbonate in 10% [vol/vol] acetonitrile) for 1 h at 56°C prior to 30 min of incubation at room temperature in the dark in an alkylating solution containing 200 mM iodoacetamide, 50 mM ammonium bicarbonate, and 10% (vol/vol) acetonitrile. Protein samples were then washed three times in a solution containing 40 mM ammonium bicarbonate and 10% (vol/vol) acetonitrile and dried in a DNA–mini-vacuum centrifuge for 1 h. The proteins were then digested by rehydration in 12.5 μg of modified trypsin (Sigma)/ml. After 1 h of incubation at room temperature, an equal volume of 40 mM ammonium bicarbonate in a 10% (vol/vol) acetonitrile solution was added to each sample and left to incubate with agitation overnight at 37°C. The resulting peptides were then separated using a Bruker amaZon ion trap mass spectrometer and processed and analyzed by using the ProteinScape central bioinformatic platform (Bruker).
Since known E1B-55K binding proteins function predominantly in fundamental cellular signaling pathways, we undertook an investigation to define more completely the cellular proteins that interact with the less-well-characterized Ad12 E1B-55K protein in Ad12 E1-transformed HER2 cells. In order to identify novel E1B-55K-interacting proteins, we performed immunoprecipitation with an anti-Ad12 E1B-55K antibody, isolated protein bands following SDS-PAGE, and subsequently analyzed tryptic peptides on a Bruker amaZon ion trap mass spectrometer. The data were processed using the Bruker DatAnalysis software to select peaks for a Mascot database search to identify peptides using the Swiss-Prot protein database. In addition to identifying the known E1B-55K-binding partners, E1B-AP5 and MRE11, we also isolated members of the TIF1 family of transcriptional regulators (unpublished data). Of these, TIF1γ was the most abundant E1B-55K interacting protein isolated (unpublished data).
To validate these findings, we performed reciprocal coimmunoprecipitation, Western blot analyses. In accordance with our mass spectrometry data, these analyses revealed that the TIF1 family member, TIF1γ was found associated with E1B-55K in Ad12 HER2 cells (Fig. 1A, panel i). In order to establish whether TIF1γ is a general target for E1B-55K from different Ad serotypes, we also investigated whether TIF1γ associated with E1B-55K in Ad5 HEK293 cells. Consistent with the notion that E1B-55K from different Ad serotypes target TIF1γ, these analyses revealed that Ad5 E1B-55K also associated with TIF1γ in Ad-transformed cells (Fig. 1A, panel ii). Interestingly, the anti-Ad5 E1B-55K polyclonal antibody used for the Western blot identified two E1B-55K-specific bands in Ad5 HEK293 cells; TIF1γ associated exclusively with the lower-molecular-weight form (Fig. 1A, panel ii). In agreement with the mass spectrometry data, we also determined that TIF1β interacted with E1B-55K in both Ad12 HER2 and Ad5 HEK293 cells (Fig. 1B, panels i and ii).
In order to establish whether TIF1γ was also targeted by Ad E1B-55K during infection, we performed coimmunoprecipitation experiments from Ad12- and Ad5-infected HeLa cells and assessed the binding of TIF1γ to E1B-55K by Western blot analysis. In agreement with the binding studies from Ad-transformed cells, TIF1γ was found to be associated with Ad12 and Ad5 E1B-55K in Ad-infected cells despite an overall reduction in TIF1γ levels after infection (Fig. 1C, panel i [Ad12] and panel ii [Ad5]). The ability of anti-TIF1γ antibodies to immunoprecipitate TIF1γ following infection reflects the limiting amount of TIF1γ antibody used, and the apparently large pool of TIF1γ in the cell extracts; overexposure of Western blots indicates that TIF1γ is expressed in infected cells, albeit at considerably lower levels (data not shown). Consistent with these data, TIF1γ coimmunoprecipitated E1B-55K from both Ad12- and Ad5-infected HeLa cells (Fig. 1C, panel iii [Ad12] and panel iv [Ad5]). Akin to the binding studies performed in Ad5 HEK293 cells, TIF1γ also associated with the lower-molecular-weight form of E1B-55K in Ad5-infected HeLa cells (Fig. 1C, panel iv). Taken together, these data establish TIF1γ is a common target for evolutionarily divergent Ad E1B-55K species in both Ad-infected and AdE1-transformed cells.
Since cellular proteins targeted by Ad during infection are often reorganized into discrete intranuclear locations by the particular viral proteins, they associate with, we next used immunofluorescence confocal microscopy to determine the subcellular localization of TIF1γ early during Ad infection. Since E1B-55K can be found within viral replication centers (VRCs) during infection, we initially assessed whether TIF1γ also colocalizes to VRCs during infection. To do this, we infected HeLa cells with either Ad5 or Ad12 and fixed the cells at 18 h postinfection. We then costained infected cells with TIF1γ and the surrogate VRC marker, RPA32. In mock-infected cells TIF1γ had a pan-nuclear staining distribution (Fig. 2A). Interestingly, TIF1γ did not appear to associate with VRCs in infected cells, but instead tended to reorganize into extensive nuclear track-like structures in both Ad5- and Ad12-infected cells (Fig. 2A). Since it has previously been established that the TIF1 family member, TIF1α is reorganized by E4orf3 to PML-containing nuclear tracks in Ad-infected cells (78) we next assessed whether TIF1γ colocalizes with E4orf3 and PML in Ad-infected cells. Consistent with the notion that TIF1γ might be recruited along with TIF1α to PML-containing nuclear tracks, these analyses revealed that TIF1γ colocalized with PML in both Ad5- and Ad12-infected cells (Fig. 2B). Indeed, further analyses revealed that TIF1γ colocalized with Ad5 E4orf3 in Ad5-infected cells (Fig. 2B), although the lack of an Ad12 E4orf3 antibody precluded such an analysis in Ad12-infected cells. In agreement with published work, we also showed that TIF1α was recruited to PML-nuclear tracks following both Ad5 and Ad12 infection (data not shown). To investigate whether TIF1γ recruitment to PML nuclear tracks during infection is solely dependent upon E4orf3 expression, we transfected A549 cells with either Ad5 HA-E4orf3 or Ad12 HA-E4orf3 and assessed the distribution of TIF1γ at 6 h posttransfection (Fig. 2C). Consistent with our earlier observations, TIF1γ had a pan-nuclear distribution in the absence of viral gene expression (Fig. 2C). However, in response to either Ad5 or Ad12 HA-E4orf3 expression, TIF1γ reorganized to nuclear track-like structures, where it colocalized with HA-E4orf3 (Fig. 2C). These data are consistent with the hypothesis that E4orf3 is solely responsible for TIF1γ redistribution into PML-containing nuclear tracks.
Since TIF1α and TIF1γ are both reorganized to PML-containing nuclear tracks during infection, we next investigated whether TIF1β is also targeted to these structures. Akin to TIF1α and TIF1γ, TIF1β displayed a general pan-nuclear staining pattern with a few discrete foci in mock-infected cells (Fig. 2D). In contrast to TIF1α and TIF1γ, however, TIF1β was not reorganized to PML-containing nuclear structures during infection (Fig. 2D). Since we have shown previously that a proportion of TIF1β is phosphorylated after both Ad5 and Ad12 infection (32), we next investigated whether this phospho-TIF1β species was recruited to PML-tracks during infection. Thus, we costained mock-infected and Ad-infected cells for PML and phospho-TIF1β using a phospho-specific TIF1β antibody. As expected, we were unable to detect any phosphorylated TIF1β in mock-infected cells (Fig. 2E). Consistent with our previous work, we were able to detect phospho-TIF1β in both Ad5- and Ad12-infected cells, but significantly, phospho-TIF1β did not colocalize with PML-containing nuclear tracks in these cells (Fig. 2E). Taken together, these results establish that TIF1γ is reorganized, along with TIF1α, to E4orf3 and PML-containing nuclear tracks in Ad-infected cells and suggest that Ad effects upon TIF1β and TIF1γ are functionally distinct.
During the course of our studies characterizing the ability of E1B-55K to interact with TIF1γ in Ad-infected cells, we noted that there was appreciably less TIF1γ protein in Ad-infected cells than mock-infected cells (Fig. 1C, panels i and ii, cf. lanes 7 and 8). We therefore decided to investigate these findings in more detail. To do this, we initially infected A549 cells with either Ad5 or Ad12 and determined TIF1γ protein levels at appropriate times postinfection. In accordance with our earlier observations, there was considerably less TIF1γ protein in A549 cells after both Ad5 and Ad12 infection (Fig. 3A). Significantly, and in agreement with earlier studies (78) TIF1α protein levels were not affected by Ad infection (Fig. 3A). In line with observations from a number of laboratories, p53 and the MRN component, MRE11, were both targeted for degradation during infection (Fig. 3A).
Since we have recently identified considerable variation in the abilities of different Ad serotypes to target cellular proteins for proteasome-dependent degradation, we next investigated the capabilities of other Ad serotypes to regulate the expression of the TIF1γ protein (32). In this regard, we chose to investigate the ability of the group B viruses Ad3, Ad7, and Ad11, which we have shown previously to have a very restrictive capacity to target protein substrates for degradation. Indeed, to date, these viruses have only been shown to target DNA ligase IV for degradation (22, 32). Thus, we infected HeLa cells with the group B viruses Ad3, Ad7, and Ad11, along with Ad5 and Ad12, and determined the effects of infection upon TIF1γ protein levels. Significantly, Western blot analysis revealed that the protein levels of TIF1γ were reduced considerably following infection with Ad3, Ad7, and Ad11 (Fig. 3B). In agreement with our previous findings, Ad3, Ad7, and Ad11 were unable, however, to target the MRN component, NBS1, for degradation (Fig. 3B). Given that we also identified TIF1β as an E1B-55K binding protein (Fig. 1B), we next investigated whether infection with any of these Ad serotypes would affect TIF1β protein levels. Interestingly, infection with Ad3, Ad5, Ad7, Ad11, and Ad12 had no effect upon the levels of the TIF1β protein (Fig. 3B). Taken together, these data suggest that the TIF1 family member, TIF1γ, is a preferential and common target for divergent Ad serotypes and that Ad serotypes differentially regulate TIF1α, TIF1β, and TIF1γ.
Since a number of cellular proteins are targeted for proteasome-dependent degradation during Ad infection, we next decided to examine the role of the proteasome in regulating TIF1γ protein expression during infection. To do this, we first infected HeLa cells with Ad5 and Ad12 and subsequently incubated infected cells in the absence or presence of the proteasome inhibitor, MG132. In the absence of proteasome inhibitor, Ad5 and Ad12 infection promoted the loss of the TIF1γ protein (Fig. 4). Significantly, however, treatment of cells with MG132 limited the loss of the TIF1γ protein following Ad5 and Ad12 infection (Fig. 4). In support of these findings, MG132 had a similar propensity to limit the Ad5- and Ad12-mediated degradation of the known Ad substrate, DNA ligase IV (Fig. 4). Taken together, this series of experiments have revealed that Ad5 and Ad12 target TIF1γ for proteasome-dependent degradation during infection.
We and others have previously shown that different Ad serotypes can selectively utilize cellular Cullin ring ligases, CRL2 or CRL5 to promote the degradation of protein substrates, such as p53 and TOPBP1 (10, 22, 54). However, our previous studies have also suggested that the Ad-induced degradation of substrates, such as MRE11 and DNA ligase IV, are not dependent exclusively upon CRL2 or CRL5 (32). In light of these findings, we wanted to establish whether Ad utilized CRL2 or CRL5 to promote the degradation of TIF1γ. To do this, we initially ablated the expression of the CUL2 or CUL5 proteins in HeLa cells by RNAi, using siRNAs targeted against the CUL2 and CUL5 mRNAs. We then infected these cells with Ad5 or Ad12 and determined the cellular levels of the TIF1γ protein at appropriate times postinfection (Fig. 5). These experiments revealed that knockdown of CUL2 or CUL5 had no effect on the ability of Ad5 or Ad12 to promote the proteasome-mediated degradation of TIF1γ (Fig. 5). Consistent with our previous findings, however, CUL5 knockdown did inhibit the ability of Ad5 to promote the degradation of p53 (Fig. 5A), and CUL2 knockdown did inhibit the ability of Ad12 to promote the degradation of TOPBP1 (Fig. 5B). In accordance with our previous findings, the knockdown of CUL2 or CUL5 did not, however, prevent the Ad5- or Ad12-dependent degradation of MRE11 (Fig. 5). These data suggest that Ad5 and Ad12 do not utilize CRL2 or CRL5 in order to promote the degradation of TIF1γ during infection.
Since TIF1α is reorganized along with TIF1γ to PML-containing nuclear tracks during infection but is not targeted for proteasomal degradation, we reasoned that Ad might utilize TIF1α ubiquitin ligase activity in order to promote the degradation of TIF1γ. To investigate this possibility, we first ablated the expression of the TIF1α protein in HeLa cells by RNA interference (RNAi), subsequently infected knockdown cells with Ad5 and Ad12, and then monitored TIF1γ protein levels at appropriate times postinfection (Fig. 6). Although Western blot analyses revealed that TIF1α expression in HeLa cells was successfully ablated following treatment with siRNAs specific for TIF1α (Fig. 6), both Ad5 and Ad12 retained the ability to target TIF1γ for degradation (Fig. 6). It appears, therefore, that Ad does not utilize TIF1α in order to promote the degradation of TIF1γ during infection.
It is well established that Ad E1B-55K and E4orf6 proteins cooperate to promote the ubiquitin- and proteasome-dependent degradation of numerous cellular proteins during infection. More recent data suggest that E1B-55K and E4orf6 can also function independently to target an, as yet, more limited pool of cellular proteins for degradation. Indeed, the findings presented here suggest that E1B-55K and E4orf3 both have the capacity to interact with TIF1γ during infection. Given this added complexity, we deemed it important to establish which adenoviral proteins were required to promote the degradation of TIF1γ during infection. To resolve this, we initially infected HeLa cells with wt Ad5 and a selection of Ad5 mutants that express different complements of E1B-55K, E4orf3, and E4orf6 proteins and then monitored the expression of TIF1γ at appropriate times postinfection. Interestingly, akin to wt Ad5, E1B-55K (Ad5 dl1520), and E4orf6 (H5pm4154) deletion viruses retained their ability to promote the proteasome-mediated degradation of TIF1γ, suggesting that E1B-55K and E4orf6 do not cooperate in this process (Fig. 7A). Additional Western blots for early region proteins confirmed the authenticity of the viruses used and that these proteins were expressed to similar levels (Fig. 7A).
To delineate further the requirement for early region proteins in the degradation of TIF1γ, we next compared the abilities of wt Ad5, the E4orf3 deletion mutant, H5pm4150, and the E4orf3/E4orf6 deletion mutant, H5pm4155, to promote the degradation of TIF1γ. As before, we infected A549 cells with these viruses and monitored TIF1γ protein levels at suitable times after infection. Intriguingly, these experiments indicated that the E4orf3 and E4orf3/E4orf6 deletion viruses lost their capacity to target TIF1γ for degradation (Fig. 7B). This experiment also reaffirmed the requirement for E4orf6 in the degradation of p53 (Fig. 7B). Again, Western blot analyses for early region proteins confirmed the authenticity and the infectivity ratios of the viruses used (Fig. 7B). Since there are no Ad12 E4 mutants available, we assessed the ability of Ad12 E1B-55K deletion viruses (dl620 and hr703) to promote TIF1γ degradation. Consistent with the Ad5 studies, Ad12 E1B-55K deletion viruses retained their capacity to target TIF1γ for degradation (Fig. 7C). As expected, however, these viruses lost their capacity to target MRE11, p53, and DNA ligase IV for degradation (Fig. 7C). Taken together, these data suggest that the viral proteins normally required for Ad-dependent degradation of target substrates, E1B-55K and E4orf6, are not required to promote the degradation of TIF1γ. Rather, these data suggest that degradation of TIF1γ during infection is dependent upon the expression of the E4orf3 protein.
To investigate further the requirement for early region proteins E1B-55K, E4orf3, and E4orf6 in the Ad-induced degradation of TIF1γ, we assessed individually the ability of these proteins to target TIF1γ for degradation. To do this, we transfected HeLa cells with expression plasmids for either Ad5 or Ad12, E1B-55K, HA-E4orf3, and HA-E4orf6. Since E1B-55K is known to associate with both E4orf3 and E4orf6, we also transfected HeLa cells with Ad5 or Ad12, E1B-55K/HA-E4orf3, and E1B-55K/HA-E4orf6. To evaluate the effects of early gene products upon TIF1γ, we harvested cell lysates at 24 h posttransfection and quantified TIF1γ protein expression by Western blotting. Significantly, and in accordance with the data obtained using mutant viruses, only the expression of Ad5 E4orf3 or Ad12 E4orf3 was able to promote the degradation of TIF1γ (Fig. 8A). Expression of E1B-55K or E4orf6 alone did not affect TIF1γ protein levels (Fig. 8A). The coexpression of Ad5 or Ad12 E1B-55K with E4orf3 did not affect the ability of E4orf3 to promote TIF1γ degradation, while coexpression of E1B-55K/E4orf6 did not affect TIF1γ protein levels (Fig. 8A). These expression studies also revealed that expression of Ad12 E4orf3, but not Ad5 E4orf3, upregulated the expression of the p53 protein, which was more evident when E1B-55K was coexpressed with E4orf3 (Fig. 8A). As expected, the expression of Ad5 E1B-55K also upregulated p53 protein levels, while the expression of Ad12 E1B-55K only upregulated p53 modestly (Fig. 8A). As anticipated, the expression of Ad5 or Ad12 E1B-55K and E4orf6 expression plasmids promoted p53 degradation (Fig. 8A).
Since CUL2 and CUL5 knockdown did not affect the Ad-mediated degradation of TIF1γ (Fig. 5), we next investigated whether CUL2 or CUL5 knockdown affected specifically the E4orf3-dependent degradation of TIFIγ. To do this, we first ablated CUL2 and CUL5 expression in HeLa cells by RNAi and then transfected knockdown cells with either pcDNA3, Ad5 HA-E4orf3, or Ad12 HA-E4orf3 (Fig. 8B). Consistent with our earlier observations, CUL2 or CUL5 knockdown had no affect on the ability of either Ad5 or Ad12 HA-E4orf3 to promote the degradation of TIF1γ (Fig. 8B). In order to extend these observations and the TIF1γ-E4orf3 colocalization work presented in Fig. 2C, we next investigated by coimmunoprecipitation whether Ad E4orf3 associates with TIF1γ in the absence of other Ad proteins in vivo. Therefore, we transfected HeLa cells with FLAG-tagged TIF1γ and either Ad5 or Ad12 HA-tagged E4orf3 and then immunoprecipitated FLAG-tagged TIF1γ and assessed E4orf3 binding by Western blotting. Consistent with the TIF1γ-E4orf3 colocalization data, immunoprecipitation of FLAG-tagged TIF1γ coimmunoprecipitated HA-tagged Ad5 and Ad12 E4orf3 (Fig. 8C).
Taken together, these data establish that E4orf3 binds specifically to TIF1γ and is solely responsible for targeting TIF1γ for degradation. In this regard, we have determined that E4orf3 does not employ the CRLs utilized by E1B-55K and E4orf6 to promote TIF1γ degradation.
In order to determine the significance of the Ad-mediated degradation of TIF1γ during infection, we first investigated the consequences of TIF1γ knockdown upon Ad protein expression and the Ad-mediated degradation of MRE11. To do this, we treated HeLa cells with either nonsilencing siRNA or TIF1γ siRNA, infected knockdown cells with either Ad5 or Ad12, and then monitored Ad and MRE11 protein levels by Western blotting (Fig. 9A and B). Initial observations revealed that TIF1γ expression had been successfully ablated by siRNA and that both Ad5 and Ad12 promoted TIF1γ degradation in cells treated with nonsilencing siRNA (Fig. 9A and B). Significantly, although Ad5 and Ad12 both promoted the rapid degradation of MRE11 in cells treated with nonsilencing siRNA, the Ad-mediated degradation of MRE11 proceeded more rapidly in cells, where TIF1γ expression had been ablated by siRNA (Fig. 9A and B). Consistent with these observations, Ad early region proteins were expressed at increased levels and, in some instances, at earlier times during infection in cells where TIF1γ expression had been knocked down by siRNA (Fig. 9A and B). Late viral proteins were similarly expressed earlier during Ad5 infection of TIF1γ-knockdown cells relative to control cells, although the lack of appropriate antibodies precluded such analyses for Ad12 (Fig. 9A and B).
These data suggested that TIF1γ expression might restrict Ad infection. To test this directly, we overexpressed TIF1γ in HeLa cells and then, following either Ad5 or Ad12 infection, monitored Ad and MRE11 protein levels by Western blotting (Fig. 9C and D). In agreement with the notion that TIF1γ might restrict Ad infection, the Ad5- and Ad12-mediated degradation of MRE11 was delayed, relative to control cells, in cells overexpressing TIF1γ (Fig. 9C and D). Consistent with these observations, the expression of Ad5 and Ad12 early region proteins was similarly delayed in cells where TIF1γ was overexpressed (Fig. 9C and D). Late viral proteins were also expressed later, relative to control cells, in Ad5-infected cells where TIF1γ was overexpressed (Fig. 9C and D). Taken in their entirety, these data are supportive of the proposition that TIF1γ restricts Ad infection and suggests that Ad has evolved to ablate the antiviral function of TIF1γ by selectively targeting it for degradation during infection.
It has long been established that early during infection E4orf3 reorganizes PML nuclear bodies to form distinct nuclear track-like structures that contain the cellular proteins PML, SP-100, the MRN complex, and TIF1α (19, 25, 65, 78). Of these, only the MRN component, MRE11, is subsequently targeted to cytoplasmic aggresomes for E1B-55K/E4orf6-dependent degradation (2, 30, 44). More recently, it has been determined that E4orf3 can also repress specifically p53-dependent transcription epigenetically by promoting histone H3K9 trimethylation at p53-responsive genes (62).
Work presented here identifies a new molecular function for E4orf3, namely, the evolutionarily conserved and selective ability, to promote the proteasome-dependent degradation of TIF1 family member, TIF1γ (Fig. 3 to to8).8). Crucial in this respect, we have shown that the E4orf3-dependent degradation of TIF1γ is wholly independent of E1B-55K and E4orf6 expression (Fig. 7 and and8).8). Indeed, expression of AdE4orf3 alone was shown to be sufficient to promote the degradation of TIF1γ (Fig. 8). Immunofluorescence studies revealed that Ad E4orf3 reorganized TIF1γ to PML tracks during infection (Fig. 2). Consistent with the ability of E4orf3 to reorganize TIF1γ to nuclear tracks during infection, immunoprecipitation analyses revealed that E4orf3 associated with TIF1γ in vivo (Fig. 8C). Given that TIF1γ is not targeted to cytoplasmic aggresomes, we propose that E4orf3 promotes the degradation of TIF1γ in the PML-containing nuclear tracks formed during infection.
Given the established role of CUL2 and CUL5 containing CRLs in promoting E1B-55K-, E4orf6-, and E1B-55K/E4orf6-dependent degradation of cellular substrates targeted by Ad, we investigated whether E4orf3 also utilized CRL2 or CRL5 to promote the degradation of TIF1γ. Interestingly, although we could prevent the Ad-dependent and CRL-mediated degradation of p53 and TOPBP1 during infection, we could not inhibit the Ad-induced degradation of TIF1γ (Fig. 5). We also determined that knockdown of CUL2 or CUL5 could not prevent the E4orf3-dependent degradation of TIF1γ (Fig. 8B). These data are consistent with observations suggesting that CUL2 or CUL5 are not required for the Ad-induced degradation of MRE11 or DNA ligase IV (32). Since TIF1α is also recruited to PML nuclear tracks during infection, we also investigated whether the TIF1α ubiquitin ligase promoted the AdE4orf3-mediated degradation of TIF1γ (Fig. 6). These experiments revealed, however, that AdE4orf3 does not utilize TIF1α to promote TIF1γ degradation during infection. Taken together, these data suggest that E4orf3 might utilize another cellular ubiquitin ligase activity in order to promote the degradation of TIF1γ. In support of this idea, other viruses have been shown to utilize distinct ubiquitin ligases in order to promote the degradation of specific cellular substrates. For instance, HPV E7 protein utilizes CRL2 to promote the degradation of the retinoblastoma protein, while the HPV E6 protein targets the HECT domain ubiquitin ligase, E6-AP to promote the degradation of p53 (40, 58). It is also possible that E4orf3 association with TIF1γ activates the intrinsic ubiquitin ligase activity of TIF1γ to promote TIF1γ auto-ubiquitylation and degradation.
Initial mass spectrometric analyses identified TIF1 family members TIF1α, TIF1β, and TIF1γ as Ad12 E1B-55K binding proteins (unpublished data). The ability of Ad5 and Ad12 E1B-55K to bind TIF1β and TIF1γ was corroborated following reciprocal immunoprecipitation-Western blot analyses (Fig. 1). Since E1B-55K can bind to TIF1 family members in Ad5 and Ad12 E1-transformed cells, as well as Ad-infected cells, these data suggest that E1B-55K might be able to regulate the function of these proteins independently of E4orf3 (Fig. 1). Indeed, as E1B-55K associates directly with TIF1β, whereas E4orf3 is unable to reorganize TIF1β to PML-containing nuclear tracks during infection, it is possible that E1B-55K and E4orf3 differentially and selectively regulate the functions of TIF1 family members (Fig. 1B and Fig. 2). Undoubtedly, the differential ability of E1B-55K and E4orf3 to regulate TIF1 family members is highlighted by the ability of Ad E4orf3 to promote TIF1γ degradation independently of E1B-55K expression (Fig. 7). However, since E1B-55K can cooperate functionally with E4orf3 during infection and during the transformation process, it is possible that these proteins also function together to regulate TIF1 family member function. It is interesting that the polyclonal antiserum raised against the Ad5 E1B-55K protein recognizes two specific E1B-55K protein species, of which TIF1γ associates preferentially with the lower-molecular-weight form (Fig. 1). It is tempting to speculate that the higher-molecular-weight species might represent a posttranslationally modified form of Ad5 E1B-55K. Since TIF1γ does not interact with the higher-molecular-weight form of E1B-55K these data suggest that these distinct E1B-55K species perform different roles in transformed and infected cells.
Given that TIF1α, TIF1β, and TIF1γ were all initially characterized as transcriptional regulators and that both E1B-55K and E4orf3 can function as transcriptional repressors, it is possible that E1B-55K and E4orf3 target TIF1 family members to coordinately regulate host cell transcription programs during infection or transformation. It is interesting, however, that distinct functions for TIF1 family members have also been established. Recent work has determined that TIF1α ubiquitylates p53, targets it for proteasome-dependent degradation, and consequently inhibits p53-dependent apoptosis (1). It is possible, therefore, that at least part of the ability of E1B-55K and E4orf3 to repress p53 transcriptional activity is mediated through their interaction with TIF1α. Ιt has also recently been determined that TIF1β function is essential for heterochromatic DNA DSBR. It has been established that ionizing radiation induces the ATM-dependent phosphorylation of TIF1β and the 53BP1-dependent recruitment of phosphorylated TIF1β to ionization-radiation induced foci (50). Since E1B-55K functions in concert with E4orf6 during infection to target the MRN component, MRE11, for proteasomal degradation (22, 64), it is possible that E1B-55K also targets TIF1β to further disrupt ATM signaling pathways during infection. The work presented here also suggests that E1B-55K and E4orf3 differentially regulate TIF1γ (Fig. 1, ,7,7, and and8).8). In addition to its known ability to regulate transcription, it has been determined that TIF1γ also functions to modulate transforming growth factor β (TGF-β) signaling (26, 27, 36). It has recently been proposed that TIF1γ also functions as a tumor suppressor for mouse and human chronic myelomonocytic leukemia and that TIF1α, TIF1β, and TIF1γ suppress murine hepatocellular carcinoma (3, 37). Given these observations, it is possible that E1B-55K and E4orf3 interact with TIF1γ to modulate TGF-β signaling during infection or cellular transformation.
It is becoming increasingly apparent that a large number of TRIM proteins exhibit antiviral activities. For instance, the prototypic TRIM protein, PML (TRIM19), can mediate the interferon response to viral infection and impairs virus replication by sequestering viral proteins, inhibiting the synthesis of viral mRNA, and inducing p53-dependent apoptosis (31). Viruses have therefore evolved to counteract such antiviral responses. Indeed, AdE4orf3 is believed to suppress PML function directly by disrupting PML bodies and reorganizing PML into nuclear track-like structures, whereas the herpes simplex virus type 1 ICP0 ubiquitin ligase antagonizes PML function by promoting its ubiquitin-mediated degradation (21).
Given these findings, we investigated whether TIF1γ possesses inherent antiviral activity during Ad infection (Fig. 9). In this regard, we determined that knockdown of TIF1γ accelerates the Ad-mediated degradation of MRE11 during infection, while overexpression of TIF1γ delays the Ad-mediated degradation of MRE11 (Fig. 9). In support of these findings, the expression of Ad early region, and late proteins, was evident earlier during infection in TIF1γ-knockdown cells relative to control cells (Fig. 9A and B) and, later, relative to control cells, in cells where TIF1γ was overexpressed (Fig. 9C and D). Taken together, these data suggest that TIF1γ limits Ad replication. Since TIF1γ has not previously been reported to limit viral gene expression and modulate viral responses during infection, the work presented here is the first indication that TIF1γ actively participates in antiviral responses.
In support of the antiviral properties of TRIM proteins, one systematic screen of 55 TRIM proteins identified 20 TRIM proteins which had antiviral activities toward HIV-1, murine leukemia virus (MLV) and avian leukosis virus by affecting viral entry or release (71). In the case of HIV-1, TRIM5α, TRIM11, and TRIM31 were found to restrict virus entry, TRIM22 and TRIM32 attenuated transcription of the HIV-1 long terminal repeat transcriptional promoter, and TRIM25, TRIM31, and TRIM62 inhibited virus release from cells (66, 68, 71). The transcriptional repressor properties of TIF1β have also been implicated in the antiviral response toward MLV and KSHV. TIF1β restricts MLV replication in embryonic stem and embryonic carcinoma cells, while KSHV has been found to silence gene expression and maintain a state of latency by exploiting the chromatin remodeling functions of TIF1β (20, 74). Furthermore, it has been suggested that the switch from viral latency to lytic replication during KSHV infection is mediated via the viral protein kinase (vPK)-dependent phosphorylation of TIF1β, which affects the ability of TIF1β to condense chromatin on viral promoters (20). Given our observations and the known functions of TIF1γ, it is possible that TIF1γ limits viral replication by modulating gene expression programs and remodeling chromatin through ubiquitylation.
In summary, we have identified TIF1 family members as key targets for Ad early region proteins during infection. We have also determined a novel role for E4orf3 in promoting the proteasome-mediated degradation of TIF1γ. We propose that Ad has evolved specifically to inhibit the antiviral functions of TIF1 proteins in order to promote viral replication.
We thank Stephen Jackson, Vivien Mautner, Iain Morgan, Joe Mymryk, and Stefano Piccolo for providing some of the reagents used in this study. We also thank Sally Roberts and Elena Odintsova for help with confocal microscopy.
This study was supported by a University of Birmingham Medical School Rowbotham Bequest studentship to N.A.F., and a University of Birmingham Medical School studentship to R.N.P. The Heinrich-Pette Institute is supported by the Freie und Hansestadt Hamburg and the Bundesministerium für Gesundheit.
Published ahead of print 28 December 2011