Adenovirus Regulates Sumoylation of Mre11-Rad50-Nbs1 Components through a Paralog-Specific Mechanism

Cells and viruses.

HeLa cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum. Three HeLa cell lines stably expressing hexahistidine (6His)-tagged SUMOs (49), generously provided by R. T. Hay (University of Dundee), were maintained in the same medium containing 1 μg/ml of puromycin. The following viruses were used in these studies: dl309 (phenotypically wild-type Ad5) (27); inORF3 (ΔE4-ORF3) (25); dl355 (ΔE4-ORF6) (25); dl355/inORF3 (ΔE4-ORF3/ΔE4-ORF6) (25); dl355/pmD₁₀₅A/L₁₀₆A (18), dl355/pmL₁₀₃A (19), and dl355/pmN₈₂A (19) (ΔE4-ORF6/point mutations in E4-ORF3); dl338 (ΔE1B-55K) (41); dl1520 (ΔE1B-55K) (4); dl367 (ΔE1B-55K/ΔE4-ORF6) (14); and dl366 (ΔE4) (25). E1 replacement viruses that express hemagglutinin (HA)-tagged E4-ORF3 of Ad3, -4, -5, -9, and -12 and mutant Ad5 E4-ORF3 proteins under the control of the CMV promoter were described previously (18, 19). E1B-55K mutant H354 (55) was a gift from A. J. Berk (University of California—Los Angeles). Viral preparation and infection were performed as described previously (18).

Plasmids.

The E4-ORF3 expression plasmid pcDNA3-E4-ORF3-WT was generated by insertion of the PCR fragment corresponding to the Ad5 E4-ORF3 coding region into the pcDNA3 vector. The SENP-1 coding region was amplified by PCR and inserted into the pEGFP-C1 plasmid (Clontech) at the EcoRI site.

Antibodies.

The anti-E4-ORF3 rat monoclonal antibody (6A-11) was provided by T. Dobner (Heinrich-Pette Institute), the anti-DBP (B6-8) and anti-E1B-55K (2A6) mouse monoclonal antibodies were from A. J. Levine (Princeton University), and the rabbit polyclonal anti-DBP antibody was from P. van der Vliet (University of Utrecht). The anti-E1A rabbit polyclonal antibody and anti-RanGAP1 and anti-p53 mouse monoclonal antibodies were purchased from Santa Cruz Biotechnology, the anti-Mrell and anti-Nbs1 mouse monoclonal antibodies were from Genetex, the anti-Mre11 and anti-Nbs1 rabbit polyclonal antibodies were from Novus, the anti-SUMO-1 and anti-SUMO-2/3 rabbit polyclonal antibodies were from Zymed, and the anti-HA rabbit polyclonal antibody was from Rockland.

In vivo sumoylation assay.

HeLa cells (1 × 10⁷) expressing 6His-tagged SUMO-1 or SUMO-2 were infected with the viruses indicated in the text, and SUMO conjugates were prepared as described previously (49). Briefly, cell pellets were resuspended in lysis buffer (6 M guanidinium-HCl, 10 mM Tris, 100 mM sodium phosphate, pH 8.0, 5 mM imidazole, and 5 mM β-mercaptoethanol) and sonicated (10 strokes at 40% duty cycle). The lysates were incubated with Ni²⁺-nitrilotriacetic acid (NTA) beads (Qiagen) at room temperature for 3 h. The beads were washed once with lysis buffer, once with pH 8.0 wash buffer (8 M urea, 10 mM Tris, 100 mM sodium phosphate buffer, pH 8.0, 0.1% Triton X-100, 5 mM imidazole, and 5 mM β-mercaptoethanol), and three times with pH 6.3 wash buffer (same as pH 8.0 wash buffer except for sodium phosphate buffer, pH 6.3). 6His-SUMOs were eluted from Ni²⁺-NTA beads in buffer containing 200 mM imidazole, and the eluted proteins were analyzed by Western blotting as described previously (45).

Immunofluorescence.

HeLa cells grown on glass coverslips in a 24-well plate were transfected with 0.2 μg pEGFP or pEGFP-SENP1 per well using polyethylenimine (PEI) (Polysciences). Twenty-four hours later, the cells were infected with 200 particles/cell of dl309. At 8 h postinfection (p.i.), the cells were fixed with 4% formaldehyde and permeabilized with phosphate-buffered saline (PBS) buffer containing 0.5% Triton X-100. After blocking in 10% goat serum, Ad DNA-binding protein (DBP) and Mre11 and Nbs1 were detected with rabbit anti-DBP and mouse anti-Mre11 or anti-Nbs1 antibodies. Ad DBP and SUMO-2/3 were detected using mouse anti-DBP and rabbit anti-SUMO-2/3 antibodies. Alexa 350-conjugated (Molecular Probes) and tetramethyl rhodamine isocyanate (TRITC)-conjugated (Zymed) antibodies were used as secondary antibodies. Cell images were acquired on an Axiovert 200 M digital deconvolution microscope (Zeiss) and analyzed using Axiovision software. For the analysis of E4-ORF3 proteins of different Ad serotypes, HeLa cells were infected with 500 particles/cell of E1 replacement viruses expressing HA-tagged E4-ORF3 proteins of Ad serotypes 3, 4, 5, 9, and 12 for 24 h and immunostained using mouse anti-Nbs1 and rabbit anti-HA antibodies. Immunofluorescence assays of SUMO-1 and -2/3 were performed using HeLa cells infected with dl309 or inORF3 for 8 h, and virus-infected cells were monitored using DBP- and E4-ORF3-specific antibodies, respectively.

Immunoprecipitation.

HeLa cells were infected with 200 particles/cell of dl309 and incubated for 8 h. The cell monolayers were lysed in 0.5 ml SDS lysis buffer (50 mM Tris, pH 8.0, 10 mM EDTA, 1% SDS, 1 mM phenylmethylsulfonyl fluoride [PMSF], 10 mM N-ethylmaleimide, and 0.2 mM iodoacetamide) per 10⁷ cells, incubated on ice for 5 min, and boiled for 20 min. After centrifugation at 12,000 × g for 10 min, 2.5 ml immunoprecipitation (IP) dilution buffer (16.7 mM Tris, pH 8.0, 167 mM NaCl, 1.2 mM EDTA, 0.01% SDS, 1.1% Triton X-100, and 1 mM PMSF) was added to the supernatant. The lysates were precleared with protein A-agarose beads (Roche) for 1 h and incubated with anti-SUMO-2/3 antibody overnight, followed by addition of protein A-agarose for 3 h. The beads were washed five times with IP dilution buffer and analyzed by SDS PAGE and Western blotting.

Ad infection induces SUMO-1 and SUMO-2 conjugation of Nbs1 and SUMO-2 conjugation of Mre11.

Previous studies showed that the MRN complex is degraded through a Ub-proteasome pathway during Ad infection (47). To investigate whether the MRN complex is also a substrate of sumoylation, we examined sumoylation using HeLa cells stably expressing 6His-tagged SUMO-1, SUMO-2, or SUMO-3 (49). These cell lines express the mature forms of the SUMO proteins. The SUMO-2 and SUMO-3 proteins share 97% amino acid sequence homology and are often termed SUMO-2/3. The parental HeLa and 6His–SUMO-1- and 6His–SUMO-2-expressing HeLa cells were mock infected or infected with wild-type Ad5 dl309 for 8 h. SUMO conjugates were isolated with Ni²⁺-NTA beads under denaturing conditions, and Mre11 and Nbs1 were analyzed by Western blotting. We detected at least three high-molecular-weight bands corresponding to sumoylated Mre11 and Nbs1 from lysates of Ad5-infected SUMO-2-expressing HeLa cells (Fig. 1A). We also observed SUMO-1 conjugation to Nbs1, but it was weaker than that of SUMO-2; SUMO-1 conjugation to Mre11 was undetectable in these assays. This is the first observation of SUMO modification of Mre11 and Nbs1 in mammalian cells. Sumoylation of Rad50, another component of the MRN complex, was not detected using this assay system. When we examined SUMO-2 and SUMO-3 separately, identical sumoylation patterns were observed (data not shown). No change in the sumoylation pattern of RanGAP1 was observed during Ad infection (Fig. 1A). Unmodified forms of Nbs1 and Mre11 migrate at ∼95 kDa and ∼80 kDa in SDS-PAGE, respectively. Since each SUMO conjugate reduces the mobility of a modified protein by ∼20 kDa, the reductions in Nbs1 and Mre11 mobilities due to Ad-induced sumoylation indicate that two SUMO conjugations occurred with the fastest-migrating SUMO-modified forms and that slower-migrating proteins reflect additional SUMO conjugates on these proteins. Only a small percentage of the total levels of Nbs1 and Mre11 were sumoylated during Ad infection. This is a common observation with sumoylated proteins (49). Sumoylation of Nbs1 was confirmed by immunoprecipitating endogenous SUMO-2/3 from Ad-infected HeLa cell extracts, followed by Western blot analysis (Fig. 1B, IP). We note that the total levels of SUMO-1 and SUMO-2/3 conjugates were not changed in Ad-infected cells (Fig. 1B, input). These results demonstrate that components of the MRN complex, Mre11 and Nbs1, are sumoylated during Ad5 infection.

Ad5 E4-ORF3 is necessary and sufficient to induce sumoylation of Mre11 and Nbs1.

During the early stage of Ad infection, the E4-ORF3 protein reorganizes the MRN complex into filamentous nuclear structures referred to as tracks (11, 16). A previous report showed that ectopically expressed SUMO-1, but not endogenous SUMO-1, is relocalized into nuclear dot- or track-like structures in rat kidney cells stably expressing Ad5 E1B-55K (17). Since we observed Ad-mediated sumoylation of the MRN complex at 8 h p.i. (Fig. 1), we decided to examine the subcellular localization of endogenous SUMO proteins in Ad-infected cells before E1B-55K expression is detectable. The SUMO-1 and SUMO-2/3 proteins colocalized with E4-ORF3 in nuclear tracks in Ad5-infected HeLa cells at 8 h p.i., while infection with an E4-ORF3-deficient mutant virus, inORF3, did not change SUMO localization (Fig. 2A) and displayed a punctate pattern observed in uninfected cells (data not shown).

We examined the effect of the E4-ORF3 protein on Ad-induced sumoylation of the MRN complex. HeLa cells expressing 6His-SUMO proteins were infected with wild-type Ad5 (dl309) or the E4 ORF3-deficient mutant virus (inORF3) for 8 h, and Mre11/Nbs1 sumoylation was analyzed as described above. Interestingly, SUMO-1 and SUMO-2 conjugation of Mre11 and Nbs1 was completely ablated in inORF3-infected cells (Fig. 2B), indicating that the E4-ORF3 protein is essential to induce sumoylation of Mre11 and Nbs1. Since the Ad E1B-55K and E4-ORF6 proteins function to recruit a Ub ligase complex to the MRN complex (24, 42) and E1B-55K itself is a known substrate (17) and ligase (37, 39) of sumoylation, we tested whether E1B-55K and E4-ORF6 are required for E4-ORF3-induced sumoylation of Mre11 and Nbs1. 6His–SUMO-2–HeLa cells infected with mutant viruses dl338 (ΔE1B-55K), dl355 (ΔE4-ORF6), and dl367 (ΔE1B-55K/ΔE4-ORF6) showed levels of Mre11 and Nbs1 sumoylation similar to those in wild-type Ad5 infection (Fig. 2C), demonstrating that E1B-55K and E4-ORF6 do not affect SUMO-2 conjugation of Mre11 and Nbs1.

To further investigate whether E4-ORF3 is sufficient to induce Mre11 and Nbs1 sumoylation, 6His–SUMO-2–HeLa cells were transfected with an empty or an E4-ORF3-containing expression vector, followed by mock infection or infection with dl366 (ΔE4) or inORF3 (ΔE4-ORF3) virus. SUMO-conjugated proteins were analyzed as described above. Interestingly, E4-ORF3 expression alone was sufficient to induce Nbs1 sumoylation; there were no significant differences between uninfected and ΔE4- or ΔE4-ORF3-infected cells (Fig. 2D). These data demonstrate that E4-ORF3 expression induces sumoylation of the Nbs1 and Mre11 proteins and that this process does not require any other Ad gene products.

E4-ORF3-mediated sumoylation of Mre11 and Nbs1 requires relocalization into nuclear track structures.

As the major strategy employed by E4-ORF3 to inactivate components of the cellular DNA damage response is reorganization into nuclear tracks (19, 47), we examined whether relocalization of the MRN complex is critical for E4-ORF3-induced sumoylation of Mre11 and Nbs1. We used several well-characterized E4-ORF3 point mutants, including N82A, which is completely nonfunctional (19, 47), and D105A/L106A, which forms nuclear track structures and relocalizes PML, but not the MRN complex (19). The D105A/L106A mutant is able to relocalize SUMO proteins into nuclear track structures (data not shown). Mre11 and Nbs1 sumoylation was not observed in cells infected with viruses harboring either of these E4-ORF3 point mutations (Fig. 3A), indicating that relocalization of the MRN complex by E4-ORF3 may be required for Mre11 and Nbs1 sumoylation.

It has been shown that only Ad5 E4-ORF3, but not E4-ORF3 from Ad4 and Ad12, relocalizes the MRN complex into nuclear tracks, although the E4-ORF3 proteins of all three Ad serotypes disrupt PML-NB (48). To test the E4-ORF3 serotype specificity for sumoylation of Mre11 and Nbs1 and to examine the correlation between relocalization and sumoylation, 6His–SUMO-2–HeLa cells were infected with E1-deleted recombinant Ad vectors expressing HA-tagged E4-ORF3 proteins of Ad3, -4, -5, -9, and -12 (Ad subgroups B, E, C, D, and A, respectively. Consistent with the results using the E4-ORF3 point mutant D105A/L106A (Fig. 3A), only Ad5 E4-ORF3, but not the E4-ORF3 proteins from any other Ad subgroups, induced sumoylation of Mre11 and Nbs1 (Fig. 3B). As previously reported for Ad4, -5, and -12 (48), only subgroup C Ad5 relocalized Nbs1 into E4-ORF3-containing tracks in infected HeLa cells, while Ad3, -4, -9, and -12 did not (Fig. 3C, top three rows). In contrast, the E4-ORF3 proteins of all of these serotypes were able to redirect SUMO-2/3 from punctate structures observed in uninfected and inORF3-infected cells (Fig. 2A) into nuclear tracks (Fig. 3C, bottom row). Taken together, our findings indicate that relocalization of MRN into nuclear tracks is required for E4-ORF3-induced sumoylation of Mre11 and Nbs1.

These results do not reveal if Mre11 and Nbs1 sumoylation is the result or the cause of MRN relocalization into E4-ORF3-containing nuclear tracks. If inhibition of sumoylation results in a defect in MRN complex relocalization, it would indicate that sumoylation is required for E4-ORF3-mediated MRN relocalization. To test this, the cellular sumoylation system was suppressed by ectopically expressing SUMO-specific protease 1 (SENP1) to direct SUMO deconjugation (2) (Fig. 4). In uninfected cells, SUMO-2/3 expression levels and localization did not change upon transfection of a control enhanced green fluorescent protein (EGFP) expression vector in comparison to nontransfected cells (Fig. 4, uninf. SUMO-2/3, EGFP). In contrast, a significant reduction in nuclear SUMO-2/3 staining was observed in cells transfected with an EGFP-SENP1 expression vector (Fig. 4, uninf. SUMO-2/3, EGFP-SENP1); such a decrease was previously reported with SUMO-1 and ectopic expression of SENP1 (2). In dl309-infected cells, SUMO-2/3 relocalization into nuclear tracks was not altered by EGFP expression (Fig. 4, dl309 SUMO-2/3, EGFP), but SUMO-2/3 expression was significantly ablated by EGFP-SENP1 expression (Fig. 4, dl309 SUMO-2/3, EGFP-SENP1). The same results were obtained with SUMO-1 and EGFP-SENP1 expression (data not shown). Mre11 and Nbs1 localization was examined at 8 h postinfection, as described above. Both Mre11 and Nbs1 were effectively relocalized by Ad5 infection in EGFP-SENP1-expressing cells (Fig. 4, dl309 Mre11/Nbs1, EGFP-SENP1). Since SUMO levels were effectively suppressed by EGFP-SENP1 expression but Mre11 and Nbs1 were still relocalized by E4-ORF3, these results suggest that Mre11 and Nbs1 sumoylation occurs after relocalization into nuclear tracks rather than as a prerequisite for MRN relocalization.

Mre11- and Nbs1-SUMO conjugates are transient during wild-type Ad5 infection.

During the late phase of Ad infection, E4-ORF3 nuclear tracks are maintained (18), but MRN components are translocated into viral replication centers (19) and/or cytoplasmic aggresomes (1, 33). This led us to question if Mre11- and Nbs1-SUMO conjugates were stable into the late phase of Ad infection. 6His–SUMO-1- and -2-expressing HeLa cells were infected with wild-type Ad5 for 4, 8, 12, 16, and 24 h, and Mre11 and Nbs1 SUMO conjugation levels were determined as described above. Mre11 and Nbs1 sumoylation reached a peak at 8 h p.i., when E4-ORF3 expression was evident, and then all Mre11- and Nbs1-SUMO conjugates significantly decreased by 12 h p.i. and became undetectable later than 16 h p.i. (Fig. 5). This was not surprising, since the E1B-55K/E4-ORF6 Ub-ligase complex mediates proteasomal degradation of the MRN complex at late times after infection. Consistent with the previous reports that Mre11 is degraded faster than Nbs1 and Rad50 during Ad infection (29, 47), Mre11 protein levels were greatly diminished by 12 to 16 h p.i., whereas Nbs1 protein levels were constant until 16 h p.i. (Fig. 5). These results are consistent with the observation that E4-ORF3 cannot relocalize Nbs1 in the absence of Mre11 (1) and suggest that sustained Nbs1 sumoylation requires continued colocalization with E4-ORF3 in nuclear tracks. These results also demonstrate that the conjugation of SUMO-1 and SUMO-2 to Nbs1 and SUMO-2 to Mre11 is transient during wild-type Ad5 infection. No changes in the pattern of RanGAP1 sumoylation by SUMO-1 or SUMO-2 were evident throughout the time course of Ad infection (Fig. 5).

Deconjugation of SUMO-1 and SUMO-2 from Nbs1 displays different kinetics during infection with E1B-55K and E4-ORF6 mutant viruses.

To further study desumoylation mechanisms at late times after Ad infection, E1B-55K/E4-ORF6-mediated MRN degradation was prevented by infecting cells with the E4-ORF6 mutant virus dl355. Mre11 and Nbs1 sumoylation levels were determined at 4, 8, 12, 16, 20, and 24 h postinfection. A more complex pattern of Nbs1 SUMO-1 and SUMO-2 conjugation was observed in dl355-infected cells (Fig. 6A and B); we do not know the basis of this effect. Deconjugation of SUMO-2 from Mre11 and Nbs1 showed the same pattern observed during wild-type Ad5 infection (Fig. 6B). In contrast to results observed with wild-type Ad5, E4-ORF3-induced SUMO-1 conjugation to Nbs1 was constant even at late times after infection with the E4-ORF6 mutant virus (Fig. 6A), implying that conjugation/deconjugation of SUMO-1 and SUMO-2 to Nbs1 is regulated by distinct mechanisms during the late phase of Ad infection. We next examined SUMO-1 and SUMO-2 conjugation to Nbs1 at 8 and 18 h p.i. with wild-type Ad5 or E1B-55K and E4-ORF6 mutant viruses. Conjugation of both SUMO-1 and SUMO-2 to Nbs1 was significantly decreased during the late phase of wild-type Ad5 (dl309) infection, whereas only SUMO-2 conjugation was reduced during infection with E4-ORF6 (dl355) or E1B-55K (dl1520) mutant virus infection (Fig. 6C). SUMO-1 conjugation to Nbs1 was not reduced at late times after infection with either mutant virus (Fig. 6C), suggesting that E4-ORF6 and E1B-55K play an important role in deconjugation of SUMO-1 from Nbs1.

To ascertain if E1B-55K/E4-ORF6-mediated proteasomal degradation of Mre11 results in deconjugation of SUMO-1 from Nbs1 during Ad infection, we analyzed Nbs1 sumoylation in cells infected with the E1B-55K insertion mutant H354, that interferes with MRN degradation but still allows the degradation of p53. At 20 h p.i., SUMO-1 conjugation to Nbs1 was retained in cells infected with dl1520 (E1B-55K-null) and H354, while wild-type Ad5 infection abrogated Nbs1 sumoylation (Fig. 7A). Taken together, these results show that E4-ORF3-induced SUMO-1 conjugation to Nbs1 is stable at late times of Ad infection, when the E1B-55K/E4-ORF6 E3 Ub-ligase complex is inactive for Mre11 degradation.

Inhibition of viral DNA replication and late-gene expression restores SUMO-2 conjugation of Nbs1.

To determine whether any early Ad gene products other than E1B-55K and E4-ORF6 play a role in deconjugation of SUMO-2 from Nbs1, we analyzed Nbs1 sumoylation in the presence of AraC, a nucleoside analog that inhibits DNA synthesis and consequently late gene expression. 6His–SUMO-2 HeLa cells were infected with the E4-ORF6 mutant virus dl355, and culture medium containing AraC was added for 8 and 20 h p.i. AraC treatment did not affect Nbs1 sumoylation in uninfected cells (Fig. 7B), although the drug is known to induce stalled replication forks and a DNA damage response. AraC-treated and untreated dl355-infected cells showed similar levels of Nbs1 sumoylation at 8 h p.i., but SUMO-2 conjugation to Nbs1 was retained until 20 h p.i. only in AraC-treated cells (Fig. 7B). This result suggests that viral DNA replication, an Ad late-gene product(s), or a process induced during the late phase of infection is required for deconjugation of SUMO-2 from Nbs1.