Epstein–Barr virus latent antigen 3C can mediate the degradation of the retinoblastoma protein through an SCF cellular ubiquitin ligase

Abstract

Epstein–Barr virus (EBV) stimulates the proliferation of latently infected B cells and promotes lymphoid malignancies in humans. To address the role of EBV latency protein Epstein–Barr nuclear antigen 3C (EBNA3C) in regulation of the retinoblastoma protein (Rb), we transfected EBNA3C into 293, BJAB, and SAOS-2 cells. In this context, a dominant effect of EBNA3C is to decrease Rb protein levels. EBNA3C also rescues an Rb-induced flat cell phenotype and targets Rb for proteasome- and ubiquitin-dependent degradation. Further, EBNA3C forms a stable complex with Rb in cells when the proteasome machinery is inhibited and interacts with Rb in vitro, mapping to a conserved domain at the terminus of EBNA3C. Deletion analysis of EBNA3C identified a motif within amino acids 140–149 important for both the binding and regulation of Rb. This motif is of particular interest, because it has also been linked to regulation of the Skp1/Cul1/F-box complex, SCF^Skp2. Indeed, inhibition of Skp2 function with a dominant-negative molecule reduces the ability of EBNA3C to degrade Rb. Skp2 has no detectable effect on Rb levels in the absence of EBNA3C, suggesting that SCF^Skp2 is specifically usurped by EBNA3C for the enhancement of Rb degradation. That EBNA3C has exploited this association suggests that other human malignancies might use a similar strategy to regulate the Rb protein.

Epstein–Barr virus (EBV) was the first infectious agent to be strongly associated with cancer in humans (1). EBV is now known to play a definitive and causative role in B cell malignancies of individuals who are immunosuppressed by genetic, disease, or iatrogenic factors, because latent infection with the virus efficiently establishes continuously proliferating lymphoblasts of B cell origin (2, 3). Tumor virus-encoded antigens classically promote cell cycle entry by inactivating the retinoblastoma protein (Rb) (4–6). To date, there has been no direct evidence linking EBV to the regulation of Rb, although some reports have indicated changes in Rb levels or colocalization in the nuclei of transfected cells (7, 8).

Epstein–Barr nuclear antigen 3C (EBNA3C) is essential for B cell transformation and is expressed in EBV-mediated lymphoproliferative disease (9). EBNA3C induces foci formation similar to HPV E7 in a colony formation assay (10). More recently, it has been shown that EBNA3C expression results in efficient accumulation of cells in S/G₂ and stimulates kinase activity associated with cyclin A complexes (11–13). Further, EBNA3C recruits ubiquitination activity dependent upon Skp1/Cul1/F-box complex, the SCF^Skp2 ubiquitin ligase complex (14). Skp1 and Skp2 are integral components of the SCF^Skp2 ubiquitin ligase complex and also form complexes with cyclin A/cdk2 in tumor cells (15). The functional SCF^Skp2 ubiquitin ligase complex consists of additional core components Cul1 and Roc1 and regulates the stability of cell cycle modulatory proteins including p27 and E2F (16, 17). Here, we demonstrate that EBNA3C can destabilize Rb by recruiting the SCF^Skp2 ubiquitin ligase complex, which mediates the ubiquitination and degradation of Rb.

Methods

Plasmids, Antibodies, and Cell Lines. Cytomegalovirus (CMV)-hemagglutinin (HA)-Rb, GST-Rb, pCDNA3-HA-p107, and pCDNA3-HA-p130 were kindly provided by James DeCaprio (Dana–Farber Cancer Institute, Boston) (18). pCDNA3-p27 was provided by Michele Pagano (New York University Medical Center, New York) (19). pCDNA3-myc-Skp2 and pCDNA3-HA-Roc1 were provided by Yue Xiong (University of Rochester Lineberger Cancer Center, Rochester, NY) (20). pCDNA3-HA-Ub was provided by George Mosialos (Alexander Fleming Biomedical Sciences Research Center, Vari, Greece) (21). pA3M-EBNA3C, pSG5-EBNA3C, and GST-EBNA3C constructs have been described (13, 22, 23). pA3M-EBNA3C 1–200 with amino acids 141–145 (ILCFV) mutated to alanines was prepared by standard overlap extension PCR mutagenesis. The dominant-negative pA3M-Skp2ΔF construct was prepared by cloning PCR-amplified cDNA encoding Skp2 amino acids 154–435 into the previously described pA3M vector (22). pGex-Skp2 was prepared by cloning PCR-amplified cDNA encoding either full-length Skp2 or Skp2 amino acids 154–435 into the pGex2TK vector. pA3M-Rb was prepared by cloning PCR-amplified cDNA into the previously described pA3M vector (22). Antibody reactive to Rb (SC-49) was purchased from Santa Cruz Biotechnology. Mouse monoclonal antibody reactive to the HA tag was purchased from Covance (Richmond, CA).

Transfection. In general, 10 million human embryonic kidney (HEK) 293T, BJAB, or U2OS cells were transfected by electroporation with a Bio-Rad Gene Pulser in 0.4-cm-gap cuvettes at 210–220 V and 975 microfarads. Unless otherwise indicated, transfected samples were harvested at 24 h, and SDS/PAGE was performed with 5% of the total normalized protein lysate.

Pulse–Chase Experiments. Ten million HEK 293T cells were transfected by electroporation with 10 μg of pA3M-Rb and 10 μg of pA3M-EBNA3C. Samples were pulsed 2 h in pulse-labeling medium (DMEM deficient for Met and Cys supplemented with 10% FBS) + 150 μCi (1 Ci = 37 GBq) ³⁵S Translabel (PerkinElmer) and then chased with DMEM complete for all amino acids and supplemented with 10% FBS. Samples were harvested 0, 2.5, 7.5, and 20 h after beginning the chase. Rb complexes were precipitated with myc-specific antibody from total protein and then resolved by 10% SDS/PAGE.

Flat Cell Assay. SAOS-2 cells were transfected with 4 μg of pBABE-puro and with pA3M-Rb and pA3M-EBNA3C, as indicated, by Lipofectamine 2000 (Invitrogen). Forty-eight hours after transfection, cells were selected with puromycin and monitored for flat cell formation.

GST Pull-Down Assays, Immunoprecipitation, and Western Blotting. GST pull-downs and immunoprecipitation were as described (14). For Western blotting, bands were visualized by incubation with infrared-conjugated secondary antibodies (Alexa Fluor 680, Molecular Probes). Direct detection of the fluorescence and quantification was with the Li-Cor (Lincoln, NE) Odyssey scanner.

Results

EBNA3C Expression Reduces the Half-Life of the Rb Protein. To test whether EBNA3C might affect the phosphorylation status of Rb, HEK 293T cells were transfected with expression plasmids for Rb, p27, and EBNA3C, as indicated in Fig. 1a. Exogenous expression of p27 was used to reduce the background phosphorylation of Rb in this cell line (data not shown). Although we did not detect changes in Rb phosphorylation with EBNA3C expression, we did note a significant decrease in total Rb levels (Fig. 1a). To determine whether this decrease could be attributed to destabilization of the Rb protein by EBNA3C, we performed a pulse–chase analysis. ³⁵S-labeled Rb was stable for at least 20 h in the absence of EBNA3C with <20% of the Rb degraded in this time (Fig. 1b). In contrast, EBNA3C induced the degradation of >60% of Rb by 20 h (Fig. 1b). We also tested whether the ubiquitination of Rb might be enhanced by EBNA3C. Transfection with expression plasmids for Rb-myc, HA-ubiquitin, and EBNA3C demonstrated a significant and reproducible increase in Rb ubiquitination in EBNA3C-expressing cells (Fig. 1c). This suggests that EBNA3C stimulates Rb degradation by enhancing the polyubiquitination of Rb.

Fig. 1.

EBNA3C destabilizes the Rb protein and enhances Rb ubiquitination. (a) HEK 293T cells were transfected with 10 μg of pA3M-Rb; 1 μg of pCDNA3-p27; and 0, 5, or 15 μg of pA3M-EBNA3C, as indicated. Total protein was normalized by Bradford assay and resolved by 10% SDS/PAGE. (b) HEK 293T cells were transfected with Rb expression plasmid and either EBNA3C expression plasmid (bottom gel) or vector control (top gel). Pulse–chase analysis was as described in Methods. (c) HEK 293T cells were transfected with 10 μg of pA3M-Rb, pCDNA3-HA-Ub, or pSG5-EBNA3C, as indicated. Total protein was immunoprecipitated with myc-specific antibody and resolved by 10% SDS/PAGE. (d) HEK 293T cells were transfected with 10 μg of pCMV-HA-Rb, pCDNA3-HA-p107, or pCDNA3-HA-p130, as indicated. Samples were additionally transfected with 10 μg of either pA3M-EBNA3C or vector control. Total protein was normalized by Bradford assay and resolved by 10% SDS/PAGE.

EBNA3C-Mediated Degradation of Rb Does Not Extend to the p107 and p130 Pocket Proteins. The pocket family proteins p107 and p130 share some of the functions of Rb. Indeed, viral antigens that target Rb, such as adenovirus E1A, SV40 large-T, and HPV-16 E7, may also target p107 and p130 (24, 25). However, in contrast to Rb, which was reduced by ≈50% in the presence of EBNA3C, p107 and p130 were marginally increased with EBNA3C expression (Fig. 1d). These data suggest that the regulation of Rb by EBNA3C is distinct from the other pocket family proteins.

EBNA3C Regulates Rb in Transfected Burkitt Lymphoma Cells. To determine whether the Rb phenotype is also present in an EBV-relevant cell background, we performed the assay in a human B cell line BJAB. The hypophosphorylated form of Rb was eliminated with EBNA3C expression, whereas the hyperphosphorylated form was significantly reduced (Fig. 2a). This confirms that EBNA3C is capable of regulating Rb levels in a B cell background and hints that the hypophosphorylated form may be most potently targeted.

Fig. 2.

EBNA3C additionally regulates Rb in BJAB and SAOS-2 cells. (a) BJAB cells were transfected with 20 μg of pCMV-HA-Rb and 20 μg of either pA3M-EBNA3C or vector control, as indicated. Total protein was normalized by Bradford assay and resolved by 10% SDS/PAGE. (b) SAOS-2 flat cell analysis was as described in Methods. After 2 weeks of puromycin selection, 40 200× fields were monitored for flat cell formation.

SAOS-2 cells are null for Rb; consequently, the introduction of Rb results in dramatic arrest of cells and induction of the flat cell phenotype as cells exit the cell division cycle (26–28). Indeed, Rb expression induced flat cell formation in a dose-responsive fashion (Fig. 2b). EBNA3C abrogated this phenotype, maximally reducing flat cell formation by ≈60% in this assay (Fig. 2b Lower). Importantly, cells that strongly expressed EBNA3C by immunofluorescence never formed flat cells (data not shown). This result corroborates the data above and provides a functional context for the destabilization of Rb by EBNA3C.

Proteasome Inhibition Stabilizes Rb in the Context of EBNA3C Expression. To test whether EBNA3C was using the 26S proteasome to degrade Rb, we treated HEK 293T cells with the proteasome inhibitor MG-132. Importantly, MG-132 treatment had no effect on Rb levels in the absence of EBNA3C (Fig. 3a Upper Left). However, with EBNA3C expression, Rb levels clearly increased over an 8-hr time course, suggesting protection from proteasome-dependent degradation (Fig. 3a Upper Right). It should be noted that we sometimes observed a slight decrease in EBNA3C levels with MG-132 treatment; however, this effect was also seen with the DMSO control, where it never resulted in Rb accumulation. As a further control, Western blotting for p27 levels showed that the p27 protein was stabilized in both the presence and absence of EBNA3C (Fig. 3a Upper).

Fig. 3.

EBNA3C forms complexes with Rb under conditions of proteasome inhibition. (a) HEK 293T cells were transfected with 10 μg of pCMV-HA-Rb and 10 μg of either pA3M vector (Left) or pA3M-EBNA3C (Right). At 36 h, samples were treated with either 20 μg/ml MG-132 (Upper) or DMSO vehicle control (Lower). Total protein was harvested at the times indicated, normalized by Bradford assay, and resolved by 10% SDS/PAGE. (b) HEK 293T cells were transfected with 10 μg of pA3M-EBNA3C and 10 μg of pCMV-HA-Rb. At 24 h, cells were treated with increasing concentrations of the proteasome inhibitor MG-132 for an additional 1 h and then immunoprecipitated for the myc tag on EBNA3C. Samples were resolved by 10% SDS/PAGE. (c) HEK 293T cells were transfected with 10 μg of the indicated pA3M-EBNA3C truncation mutants and 10 μg of pCMV-HA-Rb. Samples were then treated as for b. EBNA3C truncation proteins are marked by asterisks on the myc Western blot.

To test whether Rb/EBNA3C complexes might be stabilized by proteasome inhibition, cells were transfected with EBNA3C-myc and HA-Rb (Fig. 3b). In the presence of 20 μg/ml MG-132, hypophosphorylated forms of Rb preferentially coimmunoprecipitated with EBNA3C (Fig. 3b Lower). EBNA3C truncation mutants were also assayed for their ability to coimmunoprecipitate Rb when cells were treated with 20 μg/ml MG-132 (Fig. 3c). Indeed, the N terminus of EBNA3C, amino acids 1–365, coimmunoprecipitated Rb, an association not convincingly seen with either the 366–620 or 621–992 domains (Fig. 3c Lower).

The N Terminus of EBNA3C Binds Rb and Is Involved in Regulation of Rb Levels. We sought to map the specific region of EBNA3C that mediates its in vitro association with Rb and to determine whether this association contributes to Rb degradation. Both full-length EBNA3C and EBNA3C amino acids 1–365 strongly bound GST-Rb comparable to the 5% input control in intensity (Fig. 4a). To further refine the amino acids mediating this interaction, additional truncation mutants corresponding to EBNA3C amino acids 1–100 and 1–200 were tested. Although amino acids 1–200 gave binding similar to full-length EBNA3C, amino acids 1–100 did not bind GST-Rb at detectable levels (Fig. 4a).

Fig. 4.

EBNA3C amino acids 140–149 are critical regulators of Rb stability. (a) EBNA3C constructs were in vitro-translated and incubated with either GST or GST-Rb. Five percent input is represented in lane 1. (b) HEK 293T cells were transfected with 10 μg of pCMV-HA-Rb and 10 μg of the indicated pA3M-EBNA3C expression plasmids. Total protein was normalized by Bradford assay and resolved by 12% SDS/PAGE. EBNA3C proteins are marked by asterisks on the myc Western blot. (c) HEK 293T cells were transfected and treated as for b.(d)Rbwas in vitro-translated and incubated with 10–20 μg of either GST, GST-EBNA3C 130–159, or GST-EBNA3C 130–159 F144A. Five percent input is represented in lane 1. (e) HEK 293T cells were transfected and treated as for b.

We next determined whether the region of EBNA3C that mediates in vitro binding is critical for the destabilization of Rb by EBNA3C (Fig. 4b). Interestingly, full-length EBNA3C, as well as EBNA3C amino acids 1–365 and 1–200, reproducibly destabilized Rb compared with vector control (Fig. 4b Upper). This effect was not seen for either EBNA3C amino acids 1–100, which was expressed at lower levels, or 621–992 (Fig. 4b Upper). Because amino acids 1–200, but not 1–100, mediated this effect, the data suggest that amino acids 101–200 are critical for this phenotype (Fig. 4b).

EBNA3C Amino Acids 140–149 Are Necessary for Reduction in Rb Levels. EBNA3C amino acids 1–159, but not 1–129, destabilized Rb similar to amino acids 1–365 and 1–200, suggesting that amino acids 130–159 are critical for degradation of Rb (Fig. 4c). Importantly, a GST fusion protein corresponding to EBNA3C amino acids 130–159 bound Rb (Fig. 4d). As a control, the mutation of phenylalanine-144 to alanine, which likely disrupts the higher-order structure of this region of EBNA3C (13), significantly reduced the binding of EBNA3C amino acids 130–159 to Rb by >50% (Fig. 4d). We further defined the specific amino acid residues, within EBNA3C amino acids 130–159, responsible for regulation of Rb. Interestingly, EBNA3C amino acids 1–159 and 1–149 destabilized Rb, an effect not seen with either 1–139 or 1–129 domains (Fig. 4e). This strongly implicates the 10 aa 140–149 in regulation of Rb stability.

EBNA3C 127–149 Shares Some Homology with Residues 8–30 of HPV-16 E7, Which Includes the LxCxE Motif. The E7 protein of HPV type 16 has been clearly implicated in the degradation of Rb (28–31). We therefore aligned HPV type 16 E7 with the conserved N-terminal domain of EBNA3C. Interestingly, this region of EBNA3C aligns with the amino acids that include the LxCxE motif of E7, although it lacks the critical glutamic acid residue of the core LxCxE motif (Fig. 5a). To determine whether this region of homology was important for EBNA3C function in the destabilization of Rb, we mutated the five amino acids that aligned with the LxCxE motif of E7, specifically residues ILCFV (EBNA3C 141–145), to alanines. Although EBNA3C 1–200 potently destabilized Rb, the 141–145 A₅ mutant did not significantly reduce Rb levels, similar to vector control and EBNA3C 1–129, a domain not involved in regulating Rb activity (Fig. 5b).

Fig. 5.

EBNA3C amino acids 141–145 are important for both the regulation of Rb levels and SCF^Skp2 recruitment. (a) Alignment of human papillomavirus type 16 E7 protein with the region of EBNA3C that regulates Rb stability. Boxed amino acids (EBNA3C 141–145) were mutated to alanines in the subsequent experiments. (b) HEK 293T cells were transfected with 10 μg of pCMV-HA-Rb and 10 μg of either pA3M-EBNA3C 1–200, pA3M-EBNA3C 1–200 ILCV₁₄₅ to AAAAA₁₄₅ (A₅), or pA3M-EBNA3C 1–129. Total protein was normalized by Bradford assay and resolved by 10% SDS/PAGE. (c) EBNA3C 1–200 and EBNA3C 1–200 A₅ were in vitro-translated and incubated with either GST or GST-Rb. Five percent input is represented in lane 1. (d) EBNA3C 1–200 and EBNA3C 1–200 A₅ were in vitro-translated and incubated with either GST or GST-Skp2. Five percent input is represented in lane 1. (e) HEK 293T cells were transfected with 10 μg of pCDNA3-HA-Roc1 and 10 μg of either pA3M, pA3M-EBNA3C 1–200, or pA3M EBNA3C 1–200 A₅. After 36 h, samples were immunoprecipitated for the myc tag on EBNA3C and resolved by 12% SDS/PAGE. Immunoprecipitation bands were quantified and presented as the ratio of Roc1 to EBNA3C.

The 141–145 A₅ Mutant of EBNA3C Binds Rb But Lacks the Ability to Recruit Components of the SCF^Skp2 Complex. We next tested whether the aforementioned 141–145 A₅ mutant might disrupt the association between Rb and EBNA3C in vitro. Surprisingly, both EBNA3C 1–200 and the 141–145 A₅ mutant bound strongly to GST-Rb in an in vitro binding assay (Fig. 5c). Because we had previously shown that this specific region of EBNA3C was also important for the regulation of the SCF^Skp2 complex (14), we tested GST-Skp2 for binding to EBNA3C 1–200 and the 141–145 A₅ mutant (Fig. 5d). In contrast to GST-Rb, GST-Skp2 bound with significantly higher affinity to wild-type EBNA3C as compared to the 141–145 A₅ mutant, suggesting that this mutation, which blocks the degradation of Rb, may function by inhibiting the recruitment of SCF^Skp2 (Fig. 5d). We also found that this mutation significantly reduced Roc1 coimmunoprecipitation, further implicating these EBNA3C residues in recruitment of SCF^Skp2 (Fig. 5e).

Expression of a Dominant-Negative Skp2 Abrogates the Destabilization of Rb by EBNA3C. Skp2 lacking the so-called F box domain was shown to function as a dominant negative for full-length Skp2 (32). Because the region of EBNA3C that regulates SCF^Skp2 is linked to Rb degradation, we tested whether dominant-negative Skp2 blocks the degradation of Rb by EBNA3C. Indeed, the expression of dominant-negative Skp2 blocked EBNA3C regulation of Rb (Fig. 6a). A similar effect was seen in BJAB cells (Fig. 6b). In both cell lines, dominant-negative Skp2 had no discernable effect on Rb levels in the absence of EBNA3C, suggesting that regulation of Rb levels involves other cellular mechanisms in this context. Interestingly, in U2OS cells, expression of EBNA3C had no effect on Rb protein levels (Fig. 6c). However, exogenous expression of Skp2 enabled EBNA3C to destabilize Rb (Fig. 6c), indicating that EBNA3C can mediate Rb destabilization in the presence of increased levels of Skp2.

Fig. 6.

Disruption of Skp2 abrogates the destabilization of Rb by EBNA3C. (a) HEK 293T cells were transfected with 10 μg of pA3M-EBNA3C, pA3M-Skp2ΔF, or pCMV-HA-Rb, as indicated. Total protein was normalized by Bradford assay and resolved by 10% SDS/PAGE. (b) BJAB cells were transfected and treated as for a. (c) U2OS cells were transfected and treated as for a.

Discussion

Previously, EBNA3C has been indirectly linked to Rb regulatory pathways (10–14). Specifically, EBNA3C stimulates an E2F-responsive promoter, contributes to transformation of primary rat fibroblasts in vitro, and stimulates Rb phosphorylation under some conditions (10, 12). In the studies presented here, one of the dominant effects of EBNA3C was to decrease Rb protein levels. One motif important for the regulation of Rb protein levels was within the conserved domain of EBNA3C, amino acids 140–149. This region of EBNA3C was of particular interest, because it had also been linked to regulation of the SCF^Skp2 complex (14). Indeed, inhibition of Skp2 function with a dominant-negative molecule abrogated the ability of EBNA3C to degrade Rb. Skp2 had no effect on Rb levels in the absence of EBNA3C, suggesting that SCF^Skp2 is usurped specifically by EBNA3C for targeting and regulating Rb levels in EBV-transformed cells. That EBNA3C has exploited this association suggests that other viral-associated and nonviral human cancers might use a similar strategy to deregulate Rb, leading to cell cycle progression and cell proliferation.

Although other pocket family proteins have been linked to SCF^Skp2 (33–35), this report specifically implicates SCF^Skp2 in Rb degradation, with potential implications for viral oncogenesis. To this end, the HPV E7 protein has been clearly implicated in the proteasome-dependent degradation of Rb (28–31); however, our data are particularly interesting in the context of a recent study linking E7 to Cullin 1- and Skp2-containing E3 ligases (36). Although SCF^Skp2 is implicated in the ubiquitination of E7, this study does not consider the possibility that E7 might be recruiting SCF^Skp2 to regulate Rb stability (36), important in lieu of our results with EBNA3C. Further, other similarities between E7 and EBNA3C do exist, because both proteins predominately bind the hypophosphorylated active form of Rb (37), and both are themselves ubiquitinated (14, 31, 38).

These studies suggest a role for EBNA3C in mediating degradation of Rb in the context of EBV-transformed cells. Indeed, in transfected cell lines, EBNA3C can influence the stability of Rb through recruitment of the SCF^Skp2 complex. It should be noted that the initial steps between the infection of a primary B lymphocyte by EBV and the establishment of EBV-immortalized B cells are complex and not well understood. This is primarily a result of our inability to detect the EBV-infected cells that will ultimately emerge as lymphoblastoid cell lines. It is therefore possible that EBNA3C has a critical role in these early stages, because the resting B cell transitions to a cycling cell destined for immortalization. Further, our only established in vivo system for studying EBV immortalization is the latently infected lymphoblastoid cell line; as such, we are currently developing a tractable EBNA3C-null EBV recombinant system. Without this system, we do not yet have a definitive reference for assessing EBNA3C's affect on Rb stability in an in vivo context.

Finally, the role that EBV plays in the initiation of carcinomas is even more poorly understood than the lymphoblastoid cell line systems. Although EBNA3C is not expressed in many end-stage EBV-infected cancer cells, we cannot rule out a critical role for EBNA3C, perhaps expressed at sufficiently high levels to regulate Rb stability, during the early stages. In conclusion, our studies have clearly shown that EBNA3C can mediate degradation of Rb through the SCF^Skp2 complex. Further studies are needed before we can conclusively state that EBNA3C degrades Rb in LCLs, and that this event is critical for EBV-driven oncogenesis. Nonetheless, the initial studies presented here provide important clues as to a potential mechanism by which EBV can deregulate cell cycle events and drive cell proliferation.