Downregulation of Mdm2 and Mdm4 enhances viral gene expression during adenovirus infection

Abstract

Successful viral replication entails elimination or bypass of host antiviral mechanisms. Here, we show that shRNA-mediated knockdown of murine double minute (Mdm2) and its paralog Mdm4 enhanced the expression of early and late viral gene products during adenovirus (HAdV) infection. Remarkably, whereas the expression of HAdV genes was low in p53-deficient mouse embryonic fibroblasts (p53KO MEFs), the HAdV early gene products were efficiently expressed in Mdm2/p53 double-knockout (DKO) and Mdm4/p53 DKO MEFs, and viral capsid proteins were produced in Mdm2/p53 DKO MEFs. Thus, Mdm2 and Mdm4 seem to have potent antiviral property. In cells infected with wt HAdV or a mutant virus lacking the E1B-55K gene (dl1520), both Mdm2 and Mdm4 were rapidly depleted, whereas replication-deficient mutant viruses (Ad-GFP) or ΔpTP with deletions within the coding sequence of preterminal binding protein failed to induce their downregulation. Reduced expression of Mdm2 and Mdm4 was not due to general shutoff of host protein synthesis. Additionally, expression of a dominant-negative mutant of Cul5 did not affect Mdm2/Mdm4 downregulation. Thus, viral replication but not the presence of E1B-55K is required for Mdm2/Mdm4 degradation. Surprisingly, treatment of HAdV-infected cells with proteasome inhibitor MG132 only partially restored the protein levels of Mdm2 and Mdm4, suggesting that they may also be downregulated through an additional mechanism independent of proteasome. Interestingly, cyclin D1 and p21 appear to be downregulated similarly during HAdV infection. Collectively, our work provides the first biochemical evidence for antiviral function of Mdm2 and Mdm4 and that viruses employ efficient countermeasure to ensure viral replication.

Introduction

Human adenoviruses (HAdVs) have a small, linear double-stranded genome of about 36-kb. The viral genome is transported to the nucleus, where it serves as a template for producing viral mRNAs by host transcriptional machinery.¹ The immediate early gene, E1A, encodes a potent transcriptional activator that activates the expression of early viral genes (E1 to E4),² which sets the stage for viral DNA replication, virion assembly and the release of viral progeny. In permissive host cells, HAdV replication is highly efficient, and a lytic cycle is completed within a few days after initial viral entry. The early viral gene products enable viral replication by inhibiting host antiviral responses through diverse mechanisms. One well-studied mechanism is the degradation of cellular proteins involved in antiviral responses through specific E3 ubiquitin ligases consisting of viral and cellular proteins. For example, E1B-55K and E4orf6-34K proteins assemble an E3 ligase complex with cullin 2 or 5 (Cul2 or Cul5), Elongins B and C and RING box protein 1 or 2 (Rbx1 or 2).^3–6 A number of substrates of this ubiquitin ligase complex, such as p53,^3,4 Mre11,⁷ DNA ligase IV (lig4)⁸ and TOPBP1,⁶ have been identified, although precise biochemical mechanisms of substrate ubiquitination and degradation for these cellular proteins may differ.^6,9

Mdm2 is RING domain-containing E3 ubiquitin ligase that binds to p53 and mediates its ubiquitination and degradation.^10–12 This mechanism is essential for mouse embryonic development, as homozygous Mdm2 knockout is lethal, and simultaneous deletion of p53 rescues this phenotype.^13,14 Cellular mechanisms exist to suppress the E3 ubiquitin ligase function of Mdm2, thereby enhancing p53 activation.¹⁵ Mdm4 (also known as Mdmx, Hdmx and Hdm4), a protein paralogous to Mdm2, contains the conserved p53 binding domain (also known as the SWIB domain) in the N-terminal region, a zinc finger in the central region and a RING domain in the C terminus.¹⁶ Although it lacks intrinsic E3 ubiquitin ligase function,¹⁷ Mdm4 is also a potent p53 inhibitor.^18,19 Like Mdm2, Mdm4 is essential for mouse embryonic development, and concomitant deletion of p53 gene rescues the lethal phenotype of homozygous Mdm4-knockout mice.²⁰ Mdm2 gene transcription is activated by p53,²¹ thus forming an autoregulatory feedback loop. Similar mutual regulation between Mdm4 and p53 might also exist, as Mdm4 also appears to be a p53 target gene.²²

Emerging evidence suggests that Mdm2 appears to have antiviral property. In a screen using an shRNA library, Mdm2 was identified as a restriction factor against infection of N-tropic murine leukemia virus.²³ In humans, a naturally occurring single nucleotide polymorphism known as SNP309 was found in the promoter of the Mdm2 gene, resulting in a T-to-G change and the creation of a binding site for specificity factor 1 (Sp1).^24,25 Individuals with the G allele tend to have higher Mdm2 expression and increased susceptibility to tumor development,²⁴ and a mouse model confirmed that G allele confers an increased cancer risk.²⁶ Consistently, a nearby SNP, known as SNP285C that reduced Sp1-binding to the promoter of the Mdm2 gene, appears to lower the risk for breast and ovarian cancer in Caucasian populations.²⁵ Surprisingly, epidemiologic studies revealed that patients with the G allele of SNP309 seem to be associated with a reduced risk of oral squamous cell carcinoma (OSCC) due to human papillomavirus (HPV) infection compared with individuals with homozygous T/T alleles and HPV seropositivity.²⁷ Although this epidemiological finding awaits experimental validation, it seems that increased Mdm2 expression might confer resistance to HPV infection, thereby reducing the risk of OSCC. Downregulation of Mdm2 has been observed in cells infected with human cytomegalovirus (HCMV) ^28,29 and HAdVs.⁹ Downregulation of Mdm2 could be due to inactivation and destruction of p53 in HAdV-infected cells, as Mdm2 gene expression is activated by p53. However, in no cases has the potential functional importance of reducing Mdm2 protein levels for viral infection been determined. Here, we provide evidence that downregulation of both Mdm2 and Mdm4 enhanced HAdV gene expression, and distinct mechanisms might exist to inhibit Mdm2 and Mdm4 for efficient viral replication.

Results

shRNA-mediated knockdown of Mdm2 and Mdm4 augments HA dV gene expression.

We have stably transduced HCT116 cell line with lentiviral vectors carrying control, Mdm2 or Mdm4 shRNA constructs under the control of the H1 promoter. As shown in Figure 1A and D, we were able to stably deplete Mdm2 and Mdm4 using previously published shRNA sequences.^30,31 The transduced cells were uninfected (mock) or infected with varying doses of wt Ad5 or Ad12. At 48 h post Ad5 infection, higher levels of capsid protein hexon were produced in Mdm4- (lanes 5–8) or Mdm2-depleted cells (lanes 9–12) than in control cells (Fig. 1B, lower part). Higher levels of hexon in Mdm-depleted cells can also be seen in Coomassie blue-stained gel compared with control (Fig. 1B, boxed section in the upper part). Notably, the levels of cellular proteins remained largely unchanged in the infected cells compared with uninfected (mock) (Fig. 1B), indicating that virus-induced shutoff of cellular protein synthesis that is generally observed during the late phase of HAdV infection had not yet occurred under our experimental conditions. Higher levels of hexon production were also observed in Ad12-infected cells 48 hpi in Mdm4- (lanes 5–8) or Mdm2-depleted cells (lanes 9–12) compared with control cells (lanes 1–4) as shown in Figure 1D.

Figure 1.

Depletion of Mdm2 or Mdm4 enhances HAdV gene expression. Lentiviral vector carrying a control, Mdm2 or Mdm4 shRNA was stably transduced in HCT116. The cells were uninfected (mock) or infected with the indicated MOIs of wt Ad5 or Ad12. Cells were harvested at either 48 hpi (A, B and D) or 24 hpi (C and E) for western blotting analysis using antibodies against the indicated proteins. In (B), the SDS-PAGE gel was stained with Colloid blue staining kit (Invitrogen). The band corresponding to the Ad5 hexon in the boxed gel region is clearly visible. In (D), 0.1x denotes that 10-fold less cell extracts were loaded for the Ad12 hexon blot than for the other blots. PCNA, Erk1/2, and the mitochondrial heat shock protein 60 (Hsp60) were detected as loading controls. (F) shRNA-mediated Mdm2 knockdown slowed HAdV-mediated downregulation of Mdm4. The intensities of Mdm4 bands (A and D) were quantified using the NIH ImageJ software. Plotted are relative band intensities of Mdm4 in infected cells over mock-treated cells in control cells as well as in cells with Mdm2 knockdown. Error bars are standard deviation of data derived from three independent images.

To determine whether Mdm depletion also impacts the expression of early viral genes, we analyzed the levels of various viral proteins at 24 hpi. As shown in Figure 1C and E, knockdown of Mdm2 or Mdm4 resulted in increased levels of E1A of both Ad5 and Ad12, especially at low MOIs, compared with the control (compare lane 2 vs. 6 and 10, and 3 vs. 7 and 11 in Fig. 1C and E). Similar results were observed for Ad5 DBP (Fig. 1C). Interestingly, elevated levels of Ad5 or Ad12 E1B-55K expression were most evident in Mdm4-depleted cells at 24 hpi (compare lane 8 with lanes 4 and 12 in Fig. 1C and E) as well as at 48 hpi (Fig. 1A and D). For Ad5, viral capsid proteins, such as hexon, penton and fiber, were not abundantly expressed in HCT116 at 24 hpi (Fig. 1C); however, L2 pVII was clearly seen in HCT116 expressing Mdm4 shRNA (lanes 7 and 8 of Fig. 1C; trace amounts of penton and fiber were also visible in lane 8). Low level of L2 pVII was also detected in HCT116 cells expressing Mdm2 shRNA (lane 12), but no capsid proteins were detectable in control cells (lanes 2–4 of Fig. 1C). For Ad12, the highest levels of hexon were clearly seen in Mdm4-knockdown HCT116 cells (lane 8 of Fig. 1E), and Mdm2 knockdown also resulted in higher levels of hexon in comparison to control cells (compare lane 12 with 4), in agreement with the Ad12 hexon expression pattern at 48 hpi (Fig. 1D). Collectively, these data indicate that depletion of either Mdm2 or Mdm4 led to increased expression of early as well as late HAdV genes. Interestingly, the data presented in Figure 1 show that the levels of viral gene expression were higher in HCT116 cells with Mdm4 depletion than in those with Mdm2 depletion. This phenomenon could be due to persistent Mdm4 expression in Mdm2-depleted HCT116 cells (compare lanes 9–12 with other lanes of the Mdm4 part in Fig. 1A and D), as Mdm2 depletion should stabilize Mdm4, because it is a substrate of Mdm2-medited ubiquitination.^32–35

Absence of Mdm2 and Mdm4 enhances HAdV gene expression in mouse cells.

HAdVs can enter but do not replicate in mouse cells.^36,37 Since shRNA-mediated Mdm2 or Mdm4 knockdown improved viral gene expression in permissive human cells (Fig. 1), we wished to assess whether complete knockout of the Mdm2 or Mdm4 gene in mouse cells could also enhance HAdV gene expression. We infected mouse embryonic fibroblasts (MEFs) of three different genotypes [homozygous p53 knockout (p53KO), homozygous knockout of both p53 and Mdm2 (Mdm2/p53 DKO) and homozygous knockout of both Mdm4 and p53 (Mdm4/p53 DKO)] with wt Ad5 and Ad12 as well as a replication-deficient Ad-GFP. As shown in Figure 2, GFP was efficiently expressed as early as 24 hpi in MEFs of all three genotypes, indicative of efficient entry of HAdVs into these MEFs in agreement with previous results regarding infectivity of mouse cells by HAdVs.^36,38 Low levels of E1A of both Ad12 and Ad5 were detected in p53KO MEFs. In contrast, markedly higher levels of E1A were seen in Mdm2/p53 and Mdm4/p53 DKO MEFs. For the Ad5 E2A gene product DBP, a low level was detected in p53 KO MEFs (lane 8), and increased levels were found in Mdm4/p53 KO MEFs (lanes 16, 20 and 24), with highest levels of DBP being seen in Mdm2/p53 MEFs (lanes 28, 32 and 36). Interestingly, in MEFs infected with wt Ad12, DBP was only seen in Mdm2/p53 KO cells, and the expression levels were higher at early time points during infection (lanes 27, 31 and 35 in Fig. 2). Nonetheless, a faint band of Ad12 DBP was detected in Mdm4/p53 DKO MEFs at 24 hpi (lane 15). However, Ad12 DBP was undetectable in p53KO MEFs. We then examined the expression of viral late gene products. While Ad12 capsid protein hexon was undetectable in p53KO and Mdm4/p53 DKO MEFs, low levels of this protein were seen in Mdm2/p53 DKO MEFs at 24 and 48 hpi (lanes 27 and 31 in Fig. 2). Likewise, Ad5 capsid proteins were only detected in infected Mdm2/p53 DKO MEFs with higher levels at the later time points (48 and 72 hpi, lanes 32 and 36 in Fig. 2). The production of the late gene products, such as the viral capsid proteins (hexon and fiber), is indicative of productive adenoviral reproduction. Taken together, these results suggest that the absence of the Mdm2 gene seems to render MEFs more permissive for efficient gene expression of HAdVs than p53KO MEFs (Fig. 2) or 3T3 cells.³⁷ Interestingly, Mdm4/p53 DKO MEFs displayed an intermediate permissivity for HAdV infection, in which early gene products were highly expressed, while little or no late gene products were produced (Fig. 2).

Figure 2.

Absence of Mdm2 permits efficient HAdV gene expression in mouse embryonic fibroblasts (MEFs). MEFs of the indicated genotypes (p53 KO, Mdm2/p53 or Mdm4/p53 double knockout, DKO) were uninfected or infected with the indicated viruses (MOI of 10). The infected cells were harvested at the indicated times and analyzed with western blotting using antibodies against the indicated proteins. In the Ad5 virion blot, specific capsid proteins are denoted. Erk1/2 was detected as a loading control.

Mdm2 and Mdm4 are downregulated in HAdV-infected cells.

Notably, the protein levels of both Mdm2 and Mdm4 were reduced in HAdV-infected HCT116 cells in a dose-dependent manner (Fig. 1A and D). A recent study also reported Mdm2 downregulation in HAdV-infected cells.⁹ To further validate this observation, A549 cells were uninfected (mock) or infected with the indicated viruses. The cells were harvested for western blotting analysis 48 hpi. As shown in Figure 3A, Mdm2 protein levels were markedly reduced in cells infected with wt Ad5 (lane 2), E1B-55K-deleted dl1520 (lane 3) and wt Ad12 (lane 4). In cells infected with a replication-deficient recombinant Ad5 (Ad-GFP, lane 6), Mdm2 levels were not reduced and only slightly decreased in cells infected with another replication-deficient virus Ad-GFP-E1B, in which the GFP fusion of Ad12 E1B-55K under the control of the CMV IE promoter was inserted in the Ad5 vector with deleted E1 the and E3 regions (lane 5). Thus, replication competence but not the presence of E1B-55K appears to be critical for Mdm2 downregulation during HAdV infection. We also examined the levels of Mdm4 in the infected A549 cells. Mdm4 was degraded efficiently in cells infected with dl1520 and wt Ad12 (lanes 3 and 4). Surprisingly, although the Mdm4 levels were clearly reduced in cells infected with wt Ad5, a significant amount of Mdm4 was still present (lane 2), suggesting that wt Ad5 might be less potent to trigger Mdm4 downregulation. The replication efficiency appears similar for both wt Ad5 and dl1520, as a comparable amount of Ad5 capsid proteins was produced (lanes 2 and 3 of the Ad5 virion part of Fig. 3), although Ad5 DBP levels were slightly higher in dl1520-infected cells than in wt Ad5-infected cells (see the Ad5 DBP part of Fig. 3A). Ad5 DBP was also produced in A549 cells infected with replicationdeficient Ad-GFP-E1B or Ad-GFP, although no Ad5 capsid proteins were detected in cells infected with these recombinant viruses. Notably, compared with mock-infected cells, Mdm4 levels were also slightly reduced in cells infected with Ad-GFP-E1B-55K (lane 5) and Ad-GFP (lane 6 of Fig. 3A). As a control, we determined the Mre11 protein levels in the infected cells. It was clear that both wt Ad5 and Ad12, but not dl1520, efficiently degraded Mre11, in agreement with previous findings that E1B-55K is critical for Mre11 degradation.³⁹ Interestingly, Mre11 was also significantly downregulated in cells infected with Ad-GFP-E1B-55K and Ad-GFP (lanes 5 and 6 of Fig. 3A), suggesting that in the absence of both E1 and E3 regions of the Ad5 genome, replication-defective viruses could also trigger Mre11 degradation and that E1B-55K is dispensable.

Figure 3.

HAdV infection downregulates Mdm2 and Mdm4. (A) A549 cells were uninfected (mock) or infected (at the MOI of 10 for each) with wt Ad5, E1B-55K-deleted Ad5 mutant (dl1520, also known as ONYX-015), wt Ad12, replication-deficient Ad5 expressing GFP fusion of Ad12 E1B-55K (Ad-GFP-E1B), or GFP (Ad-GFP). The cells were harvested for western blotting 48 hpi using the antibodies against the indicated proteins. (B) SJSA-1 or HCT116 cells were infected with wt Ad12 with increasing doses (MOI of 10 and 50 respectively). The cells were harvested at 24 hpi or 48 hpi for western blotting with antibodies against the indicated proteins. PCNA was detected as a loading control.

To assess the kinetics of Mdm downregulation during HAdV infection, we determined Mdm2 and Mdm4 levels at 24 and 48 hpi in osteosarcoma cell line SJSA-1, in which Mdm2 is overexpressed, and HCT116 cells, which display relatively high levels of Mdm4. In SJSA-1 cells infected with wt Ad12, moderate downregulation of Mdm2 and marked depletion of Mdm4 were observed at 24 hpi, although p53 was not degraded (lanes 1–3 of Fig. 3B). Marked downregulation of Mdm2 and Mdm4 was detected at 48 hpi, but p53 degradation occurred only at a higher MOI in SJSA-1 cells (lanes 4–6 of Fig. 3B). There was no correlation between the expression of E1B-55K and the downregulation of Mdm2 and Mdm4, since little or no E1B-55K was detected at 24 hpi in SJSA-1 cells (lanes 1–3), while p53 degradation appears to correlate with E1B-55K expression (lanes 4–6 of Fig. 3B), supporting the contention that E1B-55K is dispensable for Mdm2 and Mdm4 downregulation, but is required for p53 degradation. These results also suggest that the downregulation of Mdm2 and Mdm4 has a faster kinetics than p53 degradation and might proceed through different mechanisms (see Discussion).

In HCT116 cells, downregulation of Mdm2, Mdm4 and p53 was observed at 24 hpi in a dose-dependent manner (lanes 7–9 of Fig. 3B). At 48 hpi in cells infected at a higher MOI, these proteins were barely detectable, indicative of their quantitative degradation. E1B-55K along with E2A DBP and hexon was already visible at 24 hpi in HCT116 cells, although fiber appeared at a later time (lanes 7–12 of Fig. 3B; also see Fig. 1D and E). Compared with SJSA-1, the expression of Ad12 gene products exhibited faster kinetics and accumulated to higher levels in HCT116 cells. Thus, there appears a correlation between viral gene expression and the efficiency of degradation of Mdm2, Mdm4 and p53 (Fig. 3B). Because no obvious shutoff of host protein synthesis occurred at 48 hpi (Fig. 1B), the observed reduction of Mdm2 and Mdm4 protein levels was probably due to a virus-mediated active mechanism of protein degradation, but not as a passive consequence of host shutoff.

Interestingly, in HCT116 cells with stable Mdm2 knockdown, depletion of Mdm4 was impaired in both Ad5 and Ad12-infected cells (Fig. 1A and D). We plotted the rate of Mdm4 depletion in control cells and in Mdm2-knockdown cells (Fig. 1F). It appears clear that reduced Mdm2 slowed Mdm4 downregulation. One intriguing possibility is that HAdV infection promotes Mdm2 to degrade Mdm4.

dl1520 and wt Ad12 were more efficient in triggering Mdm2 and Mdm4 downregulation.

Data presented in Figure 3A indicate that dl1520 and wt Ad12 seemed to be more effective in downregulating Mdm4 than wt Ad5 in A549 cells. To assess whether this is also applicable to Mdm2, we chose to examine the effects of HAdV infection on Mdm2 expression in SJSA-1 cell line that overexpresses Mdm2 (see Fig. 3). As shown in Figure 4A, Mdm2 was depleted in all infected cells, but Mdm2 levels were lower in cells infected with dl1520 and wt Ad12 compared with wt Ad5-infected cells, suggesting that dl1520 and Ad12 might be more potent to induce Mdm2 downregulation than wt Ad5 (compare lanes 3 and 4 with 2). We then assessed whether Mdm2 downregulation is mediated through ubiquitin-proteasome pathway. SJSA-1 cells were infected with wt Ad12 and Ad5, and 6 h before harvesting, cell cultures were untreated or treated with proteasomal inhibitor MG132. As shown in Figure 4B, marked stabilization of Mdm2 and p53 and, to a lesser extent, Mdm4 was observed in MG132-treated cells that were mock-infected or infected with HAdVs (lanes 4–6 of Fig. 4B), indicating that the stability of these proteins is largely regulated by the ubiquitin-proteasome pathway. Interestingly, in cells infected with Ad12 and, to a lesser extent in wt Ad5-infected cells, the levels of Mdm2, Mdm4 and p53 were not accumulated to the same extent as in mock after MG132 treatment, suggesting that a proteasome-independent mechanism might also play a significant role in suppressing the expression of these proteins during Ad12 infection. Notably, cyclin D1 and cyclin-dependent kinase inhibitor p21 exhibited a similar pattern of downregulation during HAdV infection (Fig. 4B; also see Fig. 3B). Surprisingly, although Mre11 was efficiently downregulated in cells infected with Ad5 or Ad12, MG132 treatment failed to stabilize Mre11 in SJSA-1 cells.

Figure 4.

HAdV-mediated downregulation of Mdm2 and Mdm4 partially depends on proteasome but not Cul5. (A) SASJ-1 cells were uninfected (mock) or infected with the indicated viruses (MOI of 10). Cells were harvested 48 hpi for western blotting analysis with the indicated antibodies. (B) SJSA-1 cells were uninfected or infected with wt Ad12 or Ad5 (wtD). Cells were first infected with Ad-GFP, Ψ5-Cul5 encoding full-length Cul5 and Ψ5-NTD encoding the N-terminal domain of Cul5, and then superinfected with wt Ad12 or Ad5 as indicated. One set of cultures was treated with proteasomal inhibitor MG132 6 h before harvesting. Cells were harvested 48 hpi for western blotting with antibodies against the indicated proteins. Hsp60 was used as a loading control.

Cul5 is not involved in Mdm2 and Mdm4 downregulation.

In Ad5-infected cells, E1B-55K and E4orf6-34K form a complex with Cul5, Elongins B and C and Rbx1 to promote ubiquitination and degradation of p53 and other substrates.^4,39,40 To assess whether Cul5 is involved in downregulation of Mdm2 and Mdm4, we infected SJSA-1 cells with recombinant adenoviruses carrying the full-length Cul5 cDNA (ψ5-Cul5) or the N-terminal domain of Cul5 (ψ5-Cul5 NTD). Cul5 NTD was shown to act as a dominant-negative mutant of Cul5.⁴⁰ As a control, cells were infected with Ad-GFP. The cells were then superinfected with wt Ad12 or Ad5. As shown in Figure 4B (lanes 7–12), Cul5 NTD expression stabilized p53 in Ad5-infected cells (lane 12 of Fig. 4B), in agreement with previous findings.⁴⁰ However, Cul5 NTD did not stabilize p53 in Ad12-infected cells (lane 11 of Fig. 4B), suggesting that Cul5 is not required for p53 degradation in Ad12-infected cells. Indeed, Cul2-containing E3 ligase has been shown recently to mediate p53 degradation in Ad12-infected cells.^5,6,9 Similarly, Cul5 NTD expression failed to stabilize Mdm2 and Mdm4 in Ad12-infected cells. In contrast, in Ad5-infected cells, Cul5 NTD expression resulted in reduced levels of Mdm2 (Fig. 4B, lane 12). Intriguingly, it appears that Mdm4 was slightly stabilized in cells infected with ψ5-Cul5 NTD and superinfected with wt Ad5 (the Mdm4 part in Fig. 4B, lane 12). Notably, Cul5 NTD did not stabilize Mre11 in SJSA-1 cells infected with either Ad5 or Ad12, indicating that Cul5 is probably not involved in Mre11 downregulation in this cell line. Consistent with this finding, Forrester et al. reported recently that siRNA-mediated depletion of Cul2 or Cul5 did not affect Mre11 degradation during HAdV infection.⁹ In addition, Nichols et al. showed that Mre11 downregulation was impaired in HeLa cells infected with a replication-deficient mutant Ad5 (ΔpTP) with a deletion of the E2B gene encoding the preterminal binding protein (pTP). The early gene products, including E1A, E1B-55K, DBP and E4orf6 were produced in cells infected with this mutant virus.⁴¹ Thus, it seems likely that viral replication might play an important role in Mre11 degradation during viral infection. Collectively, these results suggest that Cul5 is not required for downregulation of Mdm2 and Mdm4 in Ad12-infected cells, although it remains a possibility that Cul5 might have a role in Mdm4, but not Mdm2, downregulation during Ad5 infection.

Mdm4 is not downregulated in cells infected with ΔpTP mutant virus.

As described above, Mdm downregulation appears to require viral replication. To further assess this requirement, we examined Mdm4 protein levels in HCT116 cells infected with the mutant Ad5 ΔpTP that contains deletion of the pTP E2B gene. As shown in Figure 5, Mdm4 was effectively depleted in wt Ad5-infected cells but not in cells infected with ΔpTP (compare lanes 6 and 7 with 4 and 5). In fact, Mdm4 levels were increased in cells infected with a higher dose of ΔpTP (lane 7). E1A and E2A DBP were expressed in ΔpTP-infected cells, although at lower levels than in cells infected with wt Ad5. As expected, viral capsid proteins were produced in wt Ad5-infected cells but not in ΔpTP-infected cells (Fig. 5). Altogether, our data support the notion that active viral replication is critical for Mdm downregulation during HAdV infection.

Figure 5.

Mdm4 is not depleted in cells infected with replication-deficient mutant virus ΔpTP. HCT116 cells were uninfected (mock) or infected with the indicated viruses and MOIs. The cells were then harvested for western blotting analysis 24 hpi using antibodies against the indicated proteins. Ad-GFP is a replication-deficient Ad5 with deletions of the E1 and E3 region of the viral genome, whereas ΔpTP carries two small deletions of the coding region of the preterminal binding protein (pTP) without affecting the coding sequences of any other viral genes.

Discussion

In this study, we found that Mdm2 and Mdm4 are rapidly downregulated during HAdV infection. Our data indicated that shRNA-mediated depletion of both proteins enhanced HAdV gene expression. Strikingly, mouse embryonic fibroblasts lacking Mdm2 and, to a lesser extent, Mdm4, permitted efficient expression of early and late HAdV genes in non-permissive mouse cells. Thus, Mdm2 and Mdm4 may have antiviral property. Consistent with this suggestion, Malbec et al. identified Mdm2 as a restriction factor for retrovirus replication during a screening with an shRNA library.²³ An epidemiologic study suggests that SNP at the Mdm2 promoter that enhances Mdm2 expression appears to correlate with decreased risk of HPV-induced OSCC,²⁷ providing a circumstantial link to restriction of HPV infection by Mdm2. Zhang et al. reported that the HCMV IE2-86 protein binds Mdm2 and seems to facilitate its degradation in a proteasome-independent manner.²⁸ In another study, Chen et al. showed that Mdm2 was downregulated in HCMV-infected cells, and its abundance was only slightly restored in the presence of proteasome inhibitor MG132,²⁹ a phenomenon similar to what we have observed with HAdV-infected cells (Fig. 4). Downregulation of Mdm2 is important for p53 accumulation in HCMV-infected cells,²⁹ although whether this enhances HCMV replication remains to be determined.

These observations raise an important question as to how Mdm2 and Mdm4 would restrict viral replication. Mdm2 is a RING finger domain E3 ubiquitin ligase that targets various cellular substrates such as p53 ^10–12 and Mdm4 ^32,35 for ubiquitin-mediated degradation. Via their C-terminal RING domains, Mdm2 can heterodimerize with Mdm4 and several studies suggest that Mdm2/Mdm4 heterodimer is more stable than Mdm2 homodimer and highly active in promoting polyubiquitination of a substrate protein.^42–45 Thus, one obvious possibility is that the Mdm proteins could target viral proteins for degradation, thereby restricting viral replication. Targeting the early regulatory proteins such as HAdV E1A and E1B-55K proteins by the Mdms is likely, as shRNA-mediated knockdown of either Mdm2 or Mdm4 results in increased levels of E1A and E1B-55K (Fig. 1). Degradation of these early gene products would severely impair viral replication. Our preliminary results also indicate that the Mdm proteins interact with both E1A and E1B-55K (Liao D, et al. unpublished data). Thus, it is reasonable to suggest that Mdm2 or Mdm2/Mdm4 complex is capable of catalyzing polyubiquitination of these early viral gene products, leading to their degradation. Alternatively, the Mdm proteins could suppress gene expression of the early regions of the viral genome, as both Mdm2 and Mdm4 can potently inhibit transcription.^18,46,47 Further studies will be required to define mechanism by which Mdm2 and Mdm4 restrict viral replication.

Our study indicates that HAdVs seem to use distinct mechanisms to downregulate Mdm2 and Mdm4. First, viral replication appears to be important for downregulation of both Mdm2 and Mdm4, as replication-competent viruses, such as wt Ad5 and Ad12 as well as E1B-55K-deleted mutant Ad5 dl1520 but not the replication-deficient mutant viruses Ad-GFP and ΔpTP, trigger efficient degradation of Mdm2 and Mdm4 (Figs. 3 and 5). How is Mdm2 degraded during viral replication? One possibility is that viral replication induces host DNA damage signaling,⁴¹ which can presumably lead to degradation of Mdm2 and Mdm4. Mdm2 can undergo auto-ubiquitination, which becomes more intense in response to DNA damage signaling,⁴⁸ as does Mdm2's ability to ubiquitinate Mdm4.^33,34 Additionally, efficient Mdm2 degradation can be executed in the absence of a functional Mdm2 E3 ligase,⁴⁹ suggesting that Mdm2 abundance can also be regulated by other E3 ubiquitin ligases. In response to DNA damaging signaling, Mdm2 and Mdm4 are phosphorylated by multiple kinases, including Ataxia telangiectasia mutated (ATM).^33,34,50 Interestingly, mutating the ATM phosphorylation site Ser³⁹⁵ of Mdm2 only partially impairs Mdm2 downregulation on DNA damage, but treatment of cells with PI3 kinase inhibitor wortmannin blocked Mdm2 degradation to a greater extent.⁴⁸ Hence, multiple DNA damage-activated kinases of the PI3 kinase family, including ATM and ATR are likely involved in regulating Mdm2 degradation in response to DNA damage. Recently, Inuzuka et al. reported that casein kinase I (CKI) phosphorylates multiple sites in Mdm2 and triggers its destruction through Cul1-β-TrCP1 E3 ubiquitin ligase in response to DNA damage.⁵¹ Thus, independent cellular mechanisms are implicated in inducing Mdm2 degradation. In HAdV-infected cells, E1B-55K and E4-34K form a complex with Cul-5 or Cul-2-containing E3 ubiquitin ligases to degrade cellular substrates such as p53.^4,5,9 However, this virus-commandeered E3 ligase may not play a major role in degrading Mdm2 and Mdm4, as both are efficiently degraded in cells infected with a dl1520 mutant virus that does not produce E1B-55K (Fig. 3). Additionally, the expression of a dominant-negative Cul5 mutant (NTD) did not affect Mdm2 downregulation (Fig. 4). Intriguingly, depleting Mdm2 markedly reduced Mdm4 downregulation during HAdV infection (Fig. 1A, D and F), indicating that Mdm2 might play an active role in virus-induced Mdm4 degradation. This reduced kinetics of Mdm4 downregulation appears to correlate with lower levels of early and late HAdV gene products (Fig. 1), reinforcing the notion that both Mdm paralogs can inhibit viral infection. We suggest that cellular mechanism such as DNA-damage-induced Mdm2 self-ubiquitination and trans-ubiquitination of Mdm4 could efficiently promote their degradation during viral infection.

Our data indicate that cyclin D1 and p21 appear to undergo similar downregulation during HAdV infection (Figs. 3 and 4). During G₁ phase of the cell cycle, cyclin D1 and other D-type cyclins form a complex with cyclin-dependent kinases (CDK) 4 and 6, resulting in phosphorylation of retinoblastoma protein (Rb) and cell cycle progression.⁵² Surprisingly, a recent study revealed that cyclin D1 plays a role in DNA repair through interacting with BRAC2 and RAD51.⁵³ Although entry into S phase is important for viral replication, downregulation of cyclin D1 during HAdV replication suggests that suppression of cyclin D1-mediated DNA repair might be of particular importance for efficient viral genome replication and packaging, and that its function in cell cycle progression might be dispensable. Instead, HAdV induces cyclin E expression to promote S-phase entry and viral replication.⁵⁴ p21 is an inhibitor of CDK2/cyclin E and plays an important role in DNA damage-induced cell cycle arrest. Although p21 has been shown to be directly involved in nucleotide and base excision repair,⁵⁵ relief of p21-imposed G₁ arrest might be important for efficient viral replication. It is worth noting that p21 and Mdm2 are p53 target genes. Whereas the role of p53 in viral infection is complex and likely context-dependent,^56–58 it has been well established that HAdVs and oncogenic viruses such as HPV produce proteins that suppress p53 functions through diverse biochemical mechanisms.^2,59–61 Moreover, HAdV infection induces proteasome-mediated degradation of p53.^2,4 Thus, the downregulation of Mdm2 and p21 could be a consequence of p53 inhibition and degradation during HAdV infection. However, our data showed that both proteins are drastically stabilized in cells treated with MG132, suggesting that transcriptional repression might only play a minor role in suppressing the expression of Mdm2 and p21. Thus, HAdVs evolve to elicit multiple powerful mechanisms to degrade an increasing number of cellular proteins including several new substrates described herein to maximize their replication.

Materials and Methods

Viruses, cell culture and DNA constructs.

HAdVs type 5 (Ad5) and type 12 (Ad12) were obtained from the American Type Culture Collection (ATCC), amplified and purified by ViraQuest, Inc. Ad5 wild-type strain wtD, the E1B-55K deleted mutant virus dl1520 (ONYX-015), E1-substituted Ad5 carrying full-length Cul5 cDNA (Ψ5-FL) and that encoding Cul5 N-terminal domain 384 residues (Ψ5-NTD) were provided by Arnold Berk (University of California, Los Angeles). The replication-deficient Ad5 H5wt300DpTP (ΔpTP) that has two small deletions in the gene encoding the terminal binding protein (pTP) was provided by Jerry Schaack (University of Colorado).⁶² These deletions do not affect any other known coding region or any of the leader sequences for the major late RNAs. Replication-deficient recombinant Ad5 (with deletions of both E1 and E3 regions) carrying a GFP transgene under the control of the cytomegalovirus immediate-early (CMV IE) promoter in the E1 region of the modified viral genome (Ad-GFP) was purchased from ViraQuest, Inc. Human cancer cell lines used in this study were obtained from ATCC. Mouse embryonic fibroblasts (MEFs) lacking p53 (p53KO) were provided by Ronald DePinho (Harvard Medical School), and the p53/Mdm2 and p53/Mdm4 double knockout MEFs as well as p53/Mdm2/Mdm4 triple knockout MEFs⁶³ were provided by Gigi Lozano (University of Texas M. D. Anderson Cancer Center). All cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 units/ml penicillin, 10 µg/ml streptomycin sulfate and 10% bovine calf serum.

Stable expression of Mdm2 and Mdm4 shRNA.

Oligonucleotides carrying shRNA sequences targeting Mdm2 or Mdm4 were annealed and ligated to a lentivial vector. Viral particles were produced by transfecting the corresponding vector along with pCMV-VSV-G and pCMVδR8.2 plasmids into 293T cells as described previously in reference ⁶⁴. The viral supernatants were harvested 48 h after transfection and used directly to transduce various cell lines. The transduced cells were selected in medium containing 2 µg/ml of puromycin or 5 µg/ml of blasticidin. The resulting resistant cells were pooled and passed continuously in the presence of a proper antibiotic. The sequences for Mdm2 shRNA (GGA ATT TAG ACA ACC TGAA) and Mdm4 (GTG ATG ATA CCG ATG TAG A) were as previously published in references 30 and 31, respectively.

Western blotting.

Cultured cells were harvested by trypsinization and the resulting cell pellets were lysed with the RadioImmunoPrecipitation Assay (RIPA) buffer [50 mM TRIS-HCl (pH 8.0), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS] supplemented with a 100-fold diluted protease inhibitor cocktail (Sigma-Aldrich, P8340). Cells in multiple-well plates were also lysed in situ using 1x Passive Lysis Buffer (Promega). Cell lysates were frozen at −80°C and thawed at room temperature. The lysates were cleared by centrifugation with a microcentrifuge at the maximal speed (13,500 RPM) for 15 min. The protein concentration of the lysates was determined using the Bradford method with Bio-Rad protein assay reagent. A total of 30 µg cellular proteins were loaded in each lane of a SDS-PAGE gel. Antibodies used for western blotting detection included anti-Ad12 E1B-55K (custom-made rabbit antiserum⁶⁵), custom-made rabbit antisera against Ad12 proteins DBP, fiber (GenScript), and hexon (YenZym Antibodies, LLC); rabbit anti-Ad12 E1A antiserum (provided by Hancheng Guan, University of Pennsylvania), mouse monoclonal anti-Ad2/5 E1B-55K antibody (2A6), mouse monoclonal anti-Ad5 E1A (M73, Santa Cruz Biotechnology), mouse monoclonal anti-Ad5 DBP (B6-8, provided by David Ornelles, Wake Forest University), rabbit anti-Ad5 virion (provided by Arnold Berk, University of California, Los Angeles),⁴⁰ anti-p53 (DO1, Santa Cruz Biotechnology), anti-Mdm2 (custom-made antisera, YenZym Antibodies LLC), anti-Mdm2 (4B11, CalBiochem), anti-Mdm4 (A300-287A, Bethyl Laboratories; 8C6, Millipore), anti-MRE11 antibody (NB100-142, Novus Biologics), anti-p21 (clone SXM30, BD Biosciences), anti-cyclin D1 (1677-1, Epitomics), anti-GFP (MMS-118P, Covance), anti-Erk1/2 (M5670, Sigma-Aldrich), anti-Hsp60 (H99020, BD Biosciences) and anti-PCNA (EPR3821, Epitomics).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.