The Mre11 Cellular Protein Is Modified by Conjugation of Both SUMO-1 and SUMO-2/3 during Adenovirus Infection

The adenovirus type 5 (Ad5) E1B 55 kDa and E4 Orf6 proteins assemble a Cullin 5-E3 ubiquitin (Ub) ligase that targets, among other cellular proteins, p53 and the Mre11-Rad50-Nbs1 (MRN) complex for degradation. The latter is also inhibited by the E4 Orf3 protein, which promotes the recruitment of Mre11 into specific nuclear sites to promote viral DNA replication. The activities associated with the E1B 55 kDa and E4 Orf6 viral proteins depend mostly on the assembly of this E3-Ub ligase. However, E1B 55 kDa can also function as an E3-SUMO ligase, suggesting not only that regulation of cellular proteins by these viral early proteins may depend on polyubiquitination and proteasomal degradation but also that SUMOylation of target proteins may play a key role in their activities. Since Mre11 is a target of both the E1B/E4 Orf6 complex and E4 Orf3, we decided to determine whether Mre11 displayed similar properties to those of other cellular targets, in Ad5-infected cells. We have found that during Ad5-infection, Mre11 is modified by SUMO-1 and SUMO-2/3 conjugation. Unexpectedly, SUMOylation of Mre11 is not exclusively dependent on E1B 55 kDa, E4 Orf6, or E4 Orf3, rather it seems to be influenced by a molecular interplay that involves each of these viral early proteins.


ISRN Virology
The PML protein is modified by the small ubiquitin-like modifier (SUMO) system, and this modification is critical for the formation and regulation of PML NBs [30][31][32][33][34][35][36]. Like PML many proteins that localize in PML NBs are SUMOylated [31,36]. The SUMO system is responsible for the modification of a large pool of cellular proteins [30,32,37,38]. Modification of proteins by SUMO-1 or by SUMO-2 or -3, which display a high degree of similarity to each other and form a family distinct from SUMO-1, achieves a diverse range of effects from cellular stress responses to regulating subcellular localization, transcription factor activity, and protein stability [37][38][39][40][41]. SUMO modification is operated by an enzymatic pathway consisting of an E1 activating enzyme (SAE-2/1), a unique E2 conjugating enzyme (Ubc9), and a number of E3 ligases (reviewed in [30]). While Ubc9 is capable of directly modifying substrates through interaction with a SUMO conjugation motif (ΨKXD/E, where Ψ is a large hydrophobic amino acid and X is any amino acid), E3 ligases add specificity and increase the efficiency of the conjugation reaction. In contrast to SUMO-1, both SUMO-2 and -3 possess an N-terminal SUMO conjugation motif (VKTE) that allows their polymerization to form SUMO chains [42,43], while conjugation of SUMO-1 to a SUMO-2/3 moiety leads to chain termination [44]. An important role of SUMO modification is to protect modified proteins from proteasomal degradation or conversely to promote their polyubiquitination and degradation [40,42,[45][46][47][48].
These observations confirmed that regulation of p53 in Ad5-infected cells is achieved by processes that do not depend solely on its polyubiquitination and degradation [16][17][18], suggesting that other cellular targets of the early adenoviral proteins may be regulated by SUMOylation. Therefore, we decided to investigate whether Mre11 is modified by SUMO-1 and/or SUMO-2/3 in Ad5-infected cells. We found that during Ad5 infection of nontumor cells (HFF) Mre11 degradation was not complete and initiated very late during infection, similar to previous observations on p53 [16]. As expected, Mre11 was relocalized to E4 Orf3-and PML-containing structures, where these proteins could be seen throughout viral replication. Significantly, we have found not only that Mre11 was subject to SUMO-2/3-modification [56] but also that the protein was modified by SUMO-1. The redistribution of Mre11 in adenovirus-infected cells correlated with changes in the nuclear localization of the SUMO-1 and SUMO-2/3 cellular pools, and the protein colocalized extensively but displayed distinct SUMO-paralogue colocalization in PML NBs. In contrast to p53 [55], Mre11 SUMOylation was not dependent solely on E1B 55 kDa; rather, our data indicate that during adenovirus infection of nontumor cells, Mre11 is modified by the addition of both SUMO-1 and SUMO-2/3 paralogues and that these modifications depend on concerted activities of the E1B 55 kDa, E4 Orf6, and E4 Orf3 proteins.

Materials and Methods
2.1. Cells and Viruses. H1299, HeLa, HEK-293, and human foreskin fibroblast (HFF) cells were grown as monolayer cultures in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% or 10% fetal bovine serum. The E1B 55 kDa mutant virus A143, carrying a linker insertion in the hybrid Ad2/Ad5 E1B gene, has been described previously [57]. The Ad5 mutant Hr6 carries a frameshift mutation in the E1B coding sequence leading to an E1B 55 kDa null mutant virus [58]. The RF6 mutant carries a point mutation that inhibits E1B 55 kDa binding to Mre11 [59]. The Ad5 wildtype dl309 virus [60] and E1B mutants were propagated in monolayers of HeLa and HEK-293 cells, respectively. Viruses were titered by plaque assay on HEK-293 cells as described [61]. The H5pm4149 virus carries four stop codons in the E1B 55 kDa open reading frame and does not express the E1B 55 kDa protein [62]; H5pm4154 does not express the E4 Orf6 protein and H5pm4150 does not express the E4 Orf3 protein [13]. The three latter virus mutants were propagated in 2E2 cells and titered as fluorescent forming units (FFU) in HEK-293 cells [63].

Immunoblotting.
To analyze the steady-state concentrations of the cellular Mre11 protein, H1299 and HFF cells at approximately 80-90% confluence were infected with Ad5 or the E1B 55 kDa mutants at 30 plaque forming units (PFU)/cell. For immunoblotting, cells were harvested at indicated times after infection, washed with phosphatebuffered saline (PBS), and extracted with 25 mM Tris-HCl, pH 8.0, 50 mM NaCl, 0.5% (w/v) sodium deoxycholate, 0.5% (v/v) Nonidet P-40 (NP-40), and 1 mM phenylmethylsulfonyl fluoride for 30 min at 4 ∘ C. Cell debris was removed by centrifugation at 10,000 g at 4 ∘ C for 5 min. The extracts were analyzed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and immunoblotting. For immunoblotting equal amounts of total protein were separated by SDS-PAGE, transferred to nitrocellulose membranes, and incubated as described previously [68]. The bands were visualized by enhanced chemiluminescence as recommended by the manufacturer (Pierce, Thermo Fisher Scientific, Bonn, Germany) on X-ray films (CEA RP, medical X-ray film). Autoradiograms were scanned and cropped using Adobe Photoshop Elements 8.0.

Percentage of Mre11 SUMOylation. To measure in vivo
SUMOylation of Mre11 in Ad infected cells, the EpiSeeker Global Protein SUMOylation Assay Kit (Abcam) was used as described by the manufacturer. Briefly, HFF cells were infected with Ad5 wt or the A143, E4 Orf3-, E4 Orf6-, or E1B 55 K-mutants. Cells were harvested at 16 and 28 h p.i. and nuclei were isolated using nonidet NP-40 as described previously [15]. Nuclear extracts were incubated with the anti-Mre11 rabbit pAB (pNB 100-142) (Novus Biologicals, Ltd., Cambridge, UK) in 8-well assay microwell strips, and SUMO was measured with the anti-SUMO antibody included in the kit by colorimetry. Positive and negative controls included in the kit were used. To calculate SUMOylation of Mre11, the OD 450 nm values from the negative control were subtracted from the values obtained for extracts from infected cells and were compared to the level of mock-infected cells. Values are expressed as the percentage of SUMO-conjugated Mre11 normalized to the values obtained for Ad5 wt-infected cells. The error bars indicate the standard deviation of the duplicates of two independent measurements.

Plasmid and Transient
Transfection. The expression vectors pcDNA3-empty, pcDNA3-E1B 55 kDa wt [70], pcDNA3-HA-Ubc9, pcDNA3-E4 Orf6 wt, and pcDNA3-E4 Orf3 wt were used for the transfection assays. One hour before transfection, the medium was replaced by DMEM without antibiotics. For transient transfection, the transfection solution containing 1 L DNA (1 g/ L), 10 L of 25 kDa linear polyethylenimine (PEI; Polysciences Inc., Eppelheim, Germany), and 100 L DMEM was incubated for 30 min at room temperature. After application of the solution, cells were incubated for 6 h before replacing the medium with standard culture medium.

2.7.
Immunofluorescence. HFF cells grown on coverslips to approximately 90% confluence were mock-infected or infected with Ad5 wt or E1B 55 kDa, E4 Orf6, or E4 Orf3 null mutant viruses. Cells were processed for immunofluorescence as described previously [68]. After application of specific primary antibodies, cells were incubated with secondary antibodies (Alexa-Fluor 488, 568, and 680, Invitrogen). The coverslips were mounted on glass slides in 10% glycerol and samples were examined by optical section microscopy (Apotome. Zeiss Axiovert 200 M). Images were taken with an Axiocam MRM using Axiovision 3.1 software (Carl Zeiss, Inc.). Images were assembled using Adobe Photoshop Elements 8.0.

The Mre11 Protein Is not Efficiently Degraded and Is Posttranslationally Modified in Ad5-Infected Normal Human
Cells. During Ad5-infection of cultured human cells the DNA damage response (DDR) is initially inhibited by the E4 Orf3 protein, which disrupts PML nuclear bodies (PML NBs) and induces the intracellular redistribution of Mre11 to these sites, prior to its targeted polyubiquitination and proteasomal degradation directed by E1B 55 kDa and E4 Orf6 [8,14,[22][23][24]71]. A similar order of events occurs to regulate p53. In addition, this cellular protein is SUMOylated by E1B 55 kDa and relocalized to nuclear structures [54], where p53 and E1B 55 kDa colocalize [16,55].
In order to investigate if Mre11 is subject to similar processes, we decided to analyze whether degradation of Mre11 is induced in Ad5-infected human foreskin fibroblasts (HFF). Therefore, we analyzed soluble extracts of Ad5infected HFF cells by immunoblotting. Adenovirus infection reduced the Mre11 protein levels; however, such reductions could only be observed at time points corresponding to the late phase of infection in HFF cells (28 h postinfection (p.i.)) ( Figure 1). Since Mre11 degradation requires association of E1B 55 kDa with E4 Orf6, we examined Mre11 protein levels in cells infected with the E1B 55 kDa insertion mutant (A143) and the E1B 55 kDa null mutant Hr6 [58]. The A143 mutant virus carries a small insertion in the E1B 55 kDa coding sequence [57], which impairs binding to the E4 Orf6 protein [72]. A band of 55 kDa corresponding to E1B 55 kDa was detectable in Ad5 wild-type (wt) and A143-infected cells after immunostaining with the anti-E1B 55 kDa monoclonal antibody (mAb) 2A6. Additionally, we observed bands of higher molecular mass that correspond to the previously observed pattern for the SUMOylated forms of E1B 55 kDa [49,[51][52][53]73]. In agreement with previous findings [52], absence of the E1B 55 kDa-E4 Orf6 interaction during infection with the A143 mutant led to more abundant E1B 55 kDa-SUMOylation than during Ad5 infection ( Figure 1). Higher levels of Mre11 could be observed in cells infected with the E1B 55 kDa mutants compared to the Ad5 wt virus, indicating that, as expected, degradation of the Mre11 protein was impaired by each of the mutations ( Figure 1). This effect on Mre11 protein levels was significantly more pronounced late in infection.
In agreement with previous reports, we thus observed that degradation of Mre11 in HFF cells depends on formation of the E1B/E4 Orf6 complex. However, a clear difference in the Mre11 protein pattern, which was particularly noticeable for Mre11 in response to Ad5 and A143 infection compared to the mock-infected control, could be observed in these cells ( Figure 1). After immunostaining with the anti-Mre11 mAb (12D7), both Ad5 and A143-infected cell lysates showed additional bands with higher molecular mass (approximately 100 kDa and 200 kDa) than the corresponding Mre11 protein band (81 kDa). These high molecular weight bands were present at low levels in Ad5-infected cells and at significantly increased levels in A143-infected cells. The accumulation of slower-migrating bands detected by the anti-Mre11 mAb suggests this protein may be posttranslationally modified in the course of Ad5 infection.

The Mre11 Protein Colocalizes with E4 Orf3 and E1B 55 kDa in Nuclear and Cytoplasmic Structures throughout
the Ad5 Replication Cycle. The inactivation of MRN during the early phase of infection correlates with the E4 Orf3induced relocalization of Mre11 to nuclear sites adjacent or juxtaposed to PML NBs [23]. During the late phase Mre11 is redistributed to the cytoplasm where it has been described to colocalize with the viral E1B 55 kDa, E4 Orf6, and E4 Orf3 proteins, as well as cellular aggresome components, in a cage-like structure formed by the reorganization of vimentin filaments [8,23,27,74]. Since the degradation of Mre11 was not efficiently induced in Ad5-infected HFFs (Figure 1), we decided to analyze the intracellular distribution of Mre11 and the intermediate filament protein, vimentin, as well as the viral proteins E1B 55 kDa and E4 Orf3 (Figures 2(a) and 2(b)) in these cells, at early and late times postinfection. Therefore, immunfluorescence analyses were performed as described in Section 2. In mock-infected cells, the Mre11 protein was diffusely distributed throughout the nucleoplasm (Figure 2(a)), in agreement with previous reports [23,71]. After Ad5-infection the Mre11 protein was mostly relocalized to a number of nuclear speckles or tracks during the early phase, displaying extensive colocalization with E4 Orf3 (Figure 2(a)). Interestingly, in HFF cells the Mre11 and E4 Orf3 proteins also colocalized in numerous punctuate cytoplasmic structures (Figure 2(a)). Significantly, in contrast to the staining pattern reported in Ad5-infected cell lines, such as HeLa, a different staining pattern was observed during the late phase of infection by 28 and 48 h p.i., as the Mre11 protein was still prominently detected in nuclear dots colocalizing with the adenoviral E4 Orf3 protein (Figure 2(a)). We also examined the intracellular distribution of the E1B 55 kDa and E4 Orf3 proteins (Figure 2(b)) during the viral replication cycle and their localization resembled the patterns previously described for Ad5-infected HeLa and HFF cells [68]. Interestingly, since the Mre11 protein was clearly detected at late times postinfection, these results suggest that inhibition of nuclear Mre11, by E4 Orf3 or additional mechanisms, should be sustained well into the late phase. For the pull down assays anti-Mre11 was used, and subsequent Western blot analyses were performed using monoclonal antibodies directed specifically to SUMO-1 or SUMO-2/3 ( Figure 3(a)) thus allowing us to distinguish SUMO-1 from SUMO-2/3 paralogues. Immunoprecipitated Mre11 and subsequent SUMO-2/3 immunostaining showed bands with molecular masses of approximately 130 kDa, 180 kDa, and 250 kDa, which correspond to different forms of poly-SUMO-2/3 chains covalently attached to Mre11. As shown in Figure 3(a), the pattern of SUMO-2/3-modified Mre11 bands was very dynamic throughout the Ad5 replication cycle. The SUMO-2/3 forms of Mre11 were present from the early phase (10 and 16 h p.i.) and remained well into the late phase (28 and 36 h p.i.), in agreement with the sustained nuclear Mre11 signal observed by immunofluorescence analysis (Figure 2). Interestingly, different patterns were observed between the Ad5 and mutant viruses. The 130 kDa band was the most abundant form of SUMO-2/3-Mre11 detected during infection with either Ad5 or A143 (Figure 3(a)  the early phase of the viral replication cycle (16 h p.i.). Throughout infection, this 250 kDa band was significantly more abundant in cells infected with the E1B 55 kDa-or E4 Orf3 null mutants. In addition a 180 kDa band which was observed in cells infected with A143, as well as with the E1B 55 kDa-or E4 Orf3-minus mutant viruses, was not detected during infection with Ad5 wt (Figure 3(a)). A clear difference in the pattern and a higher overall abundance of SUMOylated bands was observed in cells infected with the E4 Orf6 null mutant, where multiple SUMO-2/3 forms of Mre11 were detected and a band with a molecular weight higher than 250 kDa was prominent. As control for total Mre11 input levels the same samples were also analyzed by Western blot using the anti-Mre11 (Mre11 long exposure). These results clearly show that Mre11 is modified by addition of SUMO-2/3, in agreement with recently published results [56]. However, these data also demonstrate that during infection the SUMO-2/3 modification of Mre11 can occur in the absence of either E1B 55 kDa, E4 Orf6, or E4 Orf3. Nevertheless, since we observed a clear difference in the pattern of Mre11-SUMO-2/3 bands between Ad5 and the different mutant viruses (Figure 3(a)), these data suggest that all three proteins have the ability to modulate SUMOylation of Mre11 in the context of infection. We then decided to test the addition of SUMO-1 to Mre11. In mock-infected cells, we observed a 110 kDa band that corresponded to SUMO-1-modified Mre11 (Figure 3(a)).  we observed approximately 20% decrease in the total level of SUMO-modified Mre11 in the late phase of infection compared with the early phase (Figure 3(b)). This is consistent with the lower levels of SUMO-1-Mre11, which are more prominent in Ad5 infection than the SUMO-2 and -3 modified protein, detected at 36 h p.i. (Figure 3(a)) and with the degradation of Mre11 induced during the late phase of the replication cycle. To compare the relative effect of E1B 55 kDa, E4 Orf3, E4 Orf6, and the E1B/E4 Orf6 complex, the SUMO-Mre11 values obtained for each mutant virus were normalized to Ad5 levels, which were set to 100% at 16 h p.i. (Figure 3(b)).  ( Figure 3(b)). The higher levels of SUMO-Mre11 observed in the absence of E1B 55 kDa or E4 Orf6 cannot be attributed to decreased deconjugation of SUMO, since similar higher percentages of SUMO-Mre11 were obtained in A143-infected cells, when both proteins are produced. However, consistently a functional E1B/E4 Orf6 E3-Ub ligase could be involved in the regulation of SUMO-modified Mre11. Taken together, these data indicate that Mre11 is SUMOylated by covalent addition of both SUMO-1 and SUMO-2/3 paralogues during Ad5-infection (Figure 3(a)). In addition, our data show that the modification of Mre11 by SUMO is highly dynamic and displays different requirements for E1B, E4 Orf3, and E4 Orf6 during the early and late phases of infection (Figures 3(a) and  3(b)).

The E1B 55 kDa, E4 Orf6, and E4 Orf3 Proteins Can Each Modulate Mre11 SUMOylation during Adenovirus Infection.
To further investigate the role of E1B 55 kDa, E4 Orf6, or E4 Orf3 in Mre11-SUMOylation, we decided to determine whether any one of these viral early proteins was sufficient to induce SUMO addition to Mre11 independent of an adenoviral infection environment. We overexpressed each of these viral proteins independently (E1B 55 kDa, E4 Orf6, and E4 Orf3) and in combination with Ubc9. For these experiments we transfected H1299 cells with either an empty vector or a plasmid encoding HA-Ubc9 as controls or plasmids encoding E1B 55 kDa, E4 Orf6, or E4 Orf3 plus the HA-Ubc9 plasmid. The soluble extracts of the transfected cells were analyzed by Western blot (Figure 4(a)). Significantly, when we transfected an empty pcDNA plasmid or pcDNA in combination with the HA-Ubc9 plasmid, as controls, we observed a higher molecular weight band of approximately 180 kDa, matching the molecular mass of one of the SUMO-2/3-Mre11 bands observed in previous experiments (Figure 3(a)). This observation suggested that the stress caused by the transfection assays was enough to induce formation of the 180 kDa form of SUMO-2/3-Mre11. When each of the viral early proteins was overexpressed in combination with the HA-Ubc9 protein, we detected an additional band of higher molecular mass of approximately 250 kDa (Figure 4(a)). No differences in the pattern of bands corresponding to SUMO-2/3-Mre11 were observed between cells transfected with any of the three viral early proteins, suggesting that outside the context of virus infection and when Ubc9 is overexpressed E1B 55 kDa, E4 Orf6, and E4 Orf3 can each induce addition of SUMO-2/3 to Mre11 to form the 250 kDa SUMO-2/3-modified Mre11 independently of one another. As described in the introduction, the E1B 55 kDa protein can function as an E3 SUMO-ligase for p53, and it provides target specificity to the Cullin 5-E3 ubiquitin ligase inducing degradation of p53 and other targets, including Mre11. To explore whether the interaction between E1B 55 kDa and Mre11 may be directly implicated in its SUMOylation, we analyzed the SUMO status of Mre11 in cells infected with the RF6 mutant, which carries a point mutation that abrogates binding of E1B 55 kDa to Mre11 [59,75] Figure 5: Ad5-infection induces the nuclear reorganization of SUMO-1 and SUMO-2/3 paralogues. (a) The intracellular distribution of E4 Orf3, Mre11, and SUMO-1 was analyzed in HFF cells infected with Ad5, the A143, or the E1B 55 kDa, E4 Orf6, or E4 Orf3 null mutants at 28 h p.i. After infection cells were processed for immunofluorescence as described in Section 2. E4 Orf3 is shown in magenta, Mre11 in red, and SUMO-1 in green. (b) The intracellular distribution of E4 Orf3 and SUMO-2/3 was analyzed in HFF cells infected with Ad5, A143, or the null mutants for E1B 55 kDa, E4 Orf6, or E4 Orf3. E4 Orf3 is shown in red and SUMO-2/3 in green; DAPI-stained nuclei are shown in blue. Insets were enlarged 5X to show the localization of SUMO-2/3 in ring-like structures, adjacent to E4 Orf3 tracks. (c) The intracellular distribution of DBP and SUMO-2/3 was analyzed in HFF cells infected with the Ad5, A143, or the null mutants for E1B 55 kDa or E4 Orf3 viruses, at 28 h p.i. DBP is shown in red and SUMO-2/3 in green; DAPI-stained nuclei are shown in blue.
were infected with Ad5, the E1B 55 kDa null, or the RF6 mutant virus, and infected-cell extracts were analyzed by Western blot (Figure 4(b)). As expected, both E1B 55 kDa virus mutants failed to induce decreased levels of Mre11 compared to Ad5 wt, since they either do not form the E3-Ub ligase (E1B-) or cannot recruit Mre11 for ubiquitination (RF6) [53,76]. Interestingly, the patterns of bands observed during the early and late phase of infection were different between the Ad5 wt and E1B 55 kDa mutants. At 16 h p.i. two high molecular mass bands of approximately 180 kDa and 250 kDa were observed in extracts of cells infected with either of the E1B 55 kDa mutants. In contrast, during late infection with each of the E1B 55 kDa mutants (28 h p.i.), the 110 kDa form of Mre11 accumulated to higher levels than in Ad5 wt infection, indicating that the levels of SUMO-1 modified Mre11 increased during the late phase of infection in the absence of the E1B 55 kDa protein or the E1B 55 kDa-Mre11 interaction. In addition, during late infection with RF6 only the 110 kDa band was visible; infection with the E1B 55 kDa null mutant resulted in the appearance of both the 110 kDa and 130 kDa bands that correspond to SUMO-1 and SUMO-2/3 forms of Mre11, respectively (Figure 4(b)). These results suggest that a different SUMOylation pattern of Mre11 was induced in the absence of E1B 55 kDa and that the E1B 55 kDa-Mre11 interaction may be necessary for SUMO-2/3 addition to Mre11 only in the late phase of infection. Taken together these data indicate that although E1B 55 kDa is not absolutely required to induce SUMOylation of Mre11 in infected cells, it can influence the level and pattern of SUMO addition to this cellular protein at different times during the viral replication cycle.

Ad5-Infection Induces Reorganization of SUMO-1 and SUMO-2/3
Paralogues. SUMOylation of PML plays an important role for the scaffold function of this protein in functional PML NB organization. Furthermore, SUMO-1 or SUMO-2/3 attachment to PML is responsible for the correct localization of PML in the PML NBs [36]. SUMO paralogues display distinct localization in the PML NBs when analyzed by immunofluorescence microscopy. While SUMO-1 modified proteins are located in the periphery of PML NBs forming a shell-like appearance, SUMO-2/3 proteins seem to occupy the interior of these structures [33]. We decided to investigate whether Ad5-infection induces the nuclear redistribution of SUMO-1 and SUMO-2/3. We therefore performed immunofluorescence analysis of the intracellular distribution of SUMO-1, SUMO-2/3, Mre11, and E4 Orf3 in HFF cells infected with Ad5 wt, A143, or the E1B 55 kDa, E4 Orf6, or E4 Orf3 null mutants ( Figure 5). Besides a slight increase in the SUMO-1 and SUMO-2/3 signals upon adenovirus infection, we observed that SUMO-1 was redistributed from small nuclear dots in mock-infected cells to speck-or track-like structures in the cell nucleus of Ad5 wt-infected HFF (90% of all cells examined, > 60). The redistribution of SUMO-1 seemed to be independent of the association between E1B 55 kDa and E4 Orf6 or the presence of either of these viral proteins, since it was also induced in cells infected with the A143, E1B 55 kDa-, or E4 Orf6-mutant virus ( Figure 5(a)). As expected Mre11 colocalized with E4 Orf3 [27], and the staining pattern of both of these proteins overlapped with SUMO-1 in the majority of nuclear tracklike structures in cells infected with either Ad5 wt, A143, E1B 55 kDa-null, or E4 Orf6-null mutants (90% of all cells examined, > 60) (Figure 5(a)). During infection with the E4 Orf3 null mutant, however, we found that neither Mre11 nor SUMO-1 was reorganized into nuclear tracks but relocalized to large dot-like structures showing a stronger Mre11 and SUMO-1 staining pattern. These data demonstrate that while E1B 55 kDa and E4 Orf6 are dispensable, E4 Orf3 is necessary for the nuclear reorganization of both Mre11 and SUMO-1 to tracks, consistent with previous observations [23,27,56,71].
We next analyzed the intracellular distribution of SUMO-2/3 during infection with Ad5 and the viral mutants using the same setup as before (Figures 5(b) and 5(c)). In mockinfected cells, we observed a diffuse SUMO-2/3 distribution throughout the nucleoplasm, with small dots displaying a slightly stronger signal. In Ad5 wt-infected cells, the SUMO-2/3 signal was reorganized into a reticular distribution where the protein accumulated in ring-like spherical structures ( Figures 5(b)-insets and 5(c)). These structures were closely associated and displayed partial colocalization with adjacent E4 Orf3 tracks (80% of all cells examined, > 30) (Figure 5(b)). Furthermore, this distribution of SUMO-2/3 was similar to viral replication centers marked by the E2 DNA-binding protein (DBP) (Figure 5(c)), which was additionally juxtaposed with E4 Orf3 (Figure 5(b)) or PML tracks (not shown). During infection with the E1B 55 kDa or E4 Orf6 null mutants, SUMO-2/3 was reorganized into nuclear tracks and colocalized with E4 Orf3 (Figure 5(b)). In contrast, during infection with the A143 or the E4 Orf3 null mutant viruses, SUMO-2/3 colocalized more extensively with DBP ( Figure 5(c)). Interestingly, relocalization of SUMO-2/3 to sites of viral DNA replication (marked by DBP) was enhanced in the absence of a functional E1B/E4 Orf6 complex (A143) or the absence of E4 Orf3 (Figure 5(c)). These data indicate that in contrast to SUMO-1, whose reorganization and colocalization depends on E4 Orf3, the extensive relocalization of SUMO-2/3 to viral replication centers is dependent on the presence of both E1B 55 kDa and E4 Orf6 (Figures 5(b) and 5(c)). Altogether, these observations suggest a compensatory effect between the three viral proteins in the modulation of the SUMO paralogues during infection. Moreover, the localization of SUMO-1 was influenced by E4 Orf3 and unaffected by E1B 55 kDa or E4 Orf6, while SUMO-2/3 relocalization to viral replication centers was dependent on E1B 55 kDa and E4 Orf6. These results suggest that a complex molecular interplay between the E1B 55 kDa, E4 Orf6, and E4 Orf3 proteins is responsible for the concerted regulation and modification of protein targets with the different SUMO paralogues during adenovirus infection.

Discussion
Various components of the DDR are regulated by posttranslational modifications, including phosphorylation, ubiquitylation, acetylation, methylation, NEDDylation, and SUMOylation [77]. Following DNA damage, several DNA repair factors transit to and from PML NBs, and the PML NBs themselves colocalize with sites of unscheduled DNA synthesis [78][79][80]. In the case of Mre11, its phosphorylation and methylation have been studied in detail [81][82][83][84][85][86][87], and the protein is known to shuttle to PML NBs in response to DNA damage, in an arginine methylation-dependent manner [86,88]. However, although Mre11 is predicted to possess up to two potential SUMOylation sites, the role of this modification in the regulation of the protein's activities is not known [80,89].
Here we have shown that Mre11 is modified by both SUMO-1 and SUMO-2/3 during infection with Ad5. The SUMOylated forms of Mre11 reproducibly displayed a complex pattern of well-defined bands in Western blot analyses that varied depending on the presence of the E1B 55 kDa, E4 Orf6, or E4 Orf3 proteins and on the ability of E1B 55 kDa to interact with E4 Orf6 or Mre11. The pulldown of Mre11 and the use of antibodies directed against either the SUMO-1 or SUMO-2/3 paralogues yielded bands with approximate molecular masses of 110, 130, 180, and 250 kDa. The 110 kDa band was observed only with antibodies directed against SUMO-1, suggesting that Mre11 was monoSUMOylated with this paralogue during infection. Prediction of potential SUMOylation sites in the Mre11 sequence has shown that there exist two putative lysines within consensus SUMO motifs (lysines 255 and 661) [90]. We do not yet know whether only one putative lysine or each lysine may be modified, with the latter resulting in a combined population of Mre11 proteins monoSUMOylated in different lysines. In contrast, higher molecular weight bands were observed only with the anti-SUMO-2/3 antibody, indicating that modification by both mono-and poly-SUMO-2/3 can occur either on one or more lysines.
As described in the introduction, E1B 55 kDa, E4 Orf6, and E4 Orf3 are required to inhibit the antiviral cellular defenses and to promote the expression of viral late genes, resulting in efficient viral replication [91]. Two key processes they modulate are the relocalization and the proteasomal degradation of cellular targets that would otherwise interfere with viral replication [2, 7-9, 14, 22, 23]. Interestingly, these viral early proteins inhibit an overlapping, but nonidentical, set of cellular target proteins. All three participate in degradation of cellular targets. Protein ubiquitination by E1B 55 kDa and E4 Orf6 depends on Cul5 or Cul2 containing E3ubiquitin ligases whose assembly is directed by E4 Orf6, while E1B 55 kDa provides specificity for substrate recognition [2,92]. In addition, E1B 55 kDa can induce degradation of the cellular transcription repressor, Daxx, independently of E4 Orf6 [93], and the E4 Orf3 protein can mediate turnover of the transcriptional intermediary factor 1 (TIF 1 ), in the absence of either E4 Orf6 or E1B 55 kDa [94], suggesting that each can modulate ubiquitination of cellular targets by both common and independent mechanisms. Interestingly, E1B 55 kDa and E4 Orf3 can also modulate the SUMOylation machinery, suggesting that regulation of cellular proteins by these viral gene products is likely to be influenced by a crosstalk between ubiquitination and SUMOylation. The data presented here now show that addition of SUMO to Mre11 during adenoviral infection is highly dynamic. The absence of E1B 55 kDa, E4 Orf6, or E4 Orf3, when neither degradation (by the E1B/E4 Orf6 complex) nor relocalization (by E4 Orf3) of Mre11 can be induced, led to the accumulation of slower migrating SUMO-modified Mre11 forms already early during infection and increased as infection progressed. However, infection with the Ad5 and each of the virus mutants produced different patterns of SUMOylated Mre11 bands, indicating that each of these proteins can differentially modulate the cellular SUMOylation system.
During infection, different adenoviral proteins modulate the cellular SUMO machinery. In the early phase, E1A interacts with Ubc9 interfering with polySUMOylation of cellular substrates, while monoSUMOylation is unaffected [95]. In the case of E1B 55 kDa, the protein is targeted by both, SUMO-1 and SUMO-2/3 addition [73], and in addition the E1B 55 kDa protein can act as an E3-SUMO-1-ligase for p53 [55]. E1B 55 kDa also interacts with Ubc9, and the degradation of p53 and Mre11 are both independent of the SUMOylation or phosphorylation status of E1B 55 kDa [49]. The SUMOylated form of E1B 55 kDa seems to be associated with an insoluble nuclear matrix fraction early during infection, when the protein associates with PML isoform IV in the PML NBs [52]. E1B 55 kDa can also interact with E4 Orf3 favoring its association with PML [26]. Upon formation of the E1B/E4 Orf6 complex E4 Orf6 has been suggested to displace E1B 55 kDa from E4 Orf3 and from the insoluble fraction, resulting in a soluble complex that associates with viral replication centers or the Cul5-E3-Ubiquitin ligase [49,52]. A crucial player in the reorganization of PML NBs during infection is the E4 Orf3 protein, which can also impact the SUMOylation machinery [56]. Intriguingly, our results have shown that SUMOylation of Mre11 is influenced by the E1B 55 kDa, E4 Orf6, and E4 Orf3 proteins. Transient expression of each of these viral early proteins was sufficient to induce SUMO addition to Mre11, while in the context of a viral infection none of them was absolutely required for Mre11 SUMOylation. Rather, the pattern of SUMOylated bands observed during the early and late phases of infection indicated that a temporal regulation of SUMO addition was influenced by a dynamic molecular interplay that involves each of these viral early proteins.
Our data showed that, as expected, the relocalization of Mre11 to E4 Orf3 and PML containing tracks (Figure 2; data not shown) was independent of E1B 55 kDa or E4 Orf6 or formation of the E1B/E4 Orf6 complex (Figures 2 and 5(a)) and that E4 Orf3 was necessary to induce the relocalization of both Mre11 and SUMO-1 to PML tracks in infected cells ( Figure 5(a)). Intriguingly, the colocalization of E4 Orf3 with SUMO-2/3 was not as abundant, as the latter accumulated mostly in ring-like spherical structures that colocalized with DBP in viral replication centers ( Figure 5(c)). The distribution of SUMO-1 and SUMO-2/3 associated to PML NBs is highly dynamic but has been shown to display distinct patterns, where SUMO-1 appears as a spherical shell that surrounds PML NBs and colocalizes with PML and SP100, while SUMO-2/3 can be located in the interior of the spherical PML NBs [33]. Our findings therefore suggest that SUMO-1 may be retained in the E4 Orf3-reorganized PML tracks that form adjacent to the viral replication centers, while SUMO-2/3 colocalizes only partially with these tracks and is instead mostly relocated to sites of viral DNA replication. The colocalization of SUMO-2/3 with DBP in these spherical structures suggests that SUMO-2/3 may be localized to the interior of the viral replication centers where it could function as a scaffold as it does in the PML NBs [33]. The more extensive recruitment of SUMO-2/3 to the interior of viral replication centers occurred only in the absence of E4 Orf3 or in the presence of both E1B 55 kDa and E4 Orf6, regardless of whether the E1B/E4 Orf6 complex can form, suggesting that each of these proteins is required for the concerted recruitment of SUMO-2/3 to the viral replication centers. These findings suggest that these three viral early proteins modulate the relocalization of SUMO paralogues to counteract antiviral host defenses that are coordinated by the PML NBs. Since other components of DDR are known to be modified by SUMO [77], it will be interesting to determine whether the different SUMO paralogues are used in Ad infected cells to modify and regulate specific DDR proteins in the PML NBs or viral replication centers.
The SUMO-modified forms of E1B 55 kDa were also more readily observed with the A143 mutant than with the Ad5 wt ( Figure 1). This was in agreement with previous findings showing that E1B 55 kDa SUMOylation increases in the absence of E4 Orf6, when association of E1B 55 kDa with E4 Orf3 and PML NBs is enhanced [52]. Interestingly, in the absence of the E3-Ub ligase, the SUMO-modified forms of both E1B 55 kDa and Mre11 accumulated to a higher extent during the late phase of infection. Taken together these findings indicate that the SUMOylated forms of Mre11 increase when degradation of the protein is precluded, suggesting that SUMOylation may be induced as a cellular mechanism that protects Mre11 from degradation. An alternative possibility is that addition of SUMO-2/3 modulates Mre11 polyubiquitination. Interestingly, while SUMO-1 addition does not correlate extensively with polyubiquitination and protein degradation, SUMO-2-conjugated proteins are more commonly ubiquitylated, linking SUMO-2 modification to the ubiquitinproteasome system [42]. In our experiments, the pattern of SUMO-modified Mre11 bands changed as viral replication progressed. While during infection with Ad5, when Mre11 degradation can be induced, slower migrating bands corresponding to the highest SUMO-2/3 modified Mre11 forms were noticeable at early times postinfection; these bands decreased by the late phase correlating with the decrease in Mre11 levels. In contrast, these bands increased in intensity in the absence of E1B 55 kDa or when E1B 55 kDa cannot interact with Mre11 (RF6), when degradation of Mre11 cannot be induced (Figure 4(b)). Interestingly at late times, the pattern of multi-SUMO-modified Mre11 switched from the slower migrating 180 and 250 kDa bands to the faster migrating, 110 and 130 kDa bands, which correspond to monoSUMO-1 and monoSUMO-2/3, respectively (Figure 4(b)). Therefore, it will be interesting to investigate whether E1B 55 kDa, E4 Orf6, and E4 Orf3 may be responsible for the temporal regulation of the SUMO-ubiquitin switch.
Sohn and Hearing recently reported [56] that Ad infection induces modification of Mre11 by SUMO-2 conjugation and of Nbs1 by both SUMO-1 and -2 in HeLa cells. In their experiments E4 Orf3 was shown to be necessary and sufficient for this modification, while deconjugation of SUMO from Nbs1 and not Mre11 was influenced by E1B 55 kDa or E4 Orf6 [56]. Our results now confirm that Mre11 SUMOylation is induced in Ad infected normal HFF and in transformed H1299 cells. However, we have demonstrated that Mre11 is also modified by the addition of SUMO-1. The reason for the difference in the SUMO-1 modification of Mre11 is not clear and should be explored to determine whether it may depend on experimental conditions. However, while Sohn and Hearing [56] used transfected HeLa cells and overexpression of both SUMO paralogues and E4 Orf3, in our experiments SUMO-1 conjugation to Mre11 was observed in normal human foreskin fibroblast infected cells. More importantly our results now show that E1B 55 kDa and the E1B/E4 Orf6 complex influence the redistribution of SUMO-2/3 in the cell nucleus as well as SUMOylation of Mre11. Therefore, these data indicate for the first time that each of these proteins can modulate SUMO modification or degradation of cellular substrates underscoring the complexity of the molecular interplay that underlies the activities of the E1B 55 kDa, E4 Orf6, and E4 Orf3 proteins in regulation of SUMOylation and ubiquitination during Ad infection.