Importance of Covalent and Noncovalent SUMO Interactions with the Major Human Cytomegalovirus Transactivator IE2p86 for Viral Infection^▿

Cell culture, viruses, and transfection.

Primary human foreskin fibroblasts (HFFs) and HEK 293T cells were cultured and transfected as described previously (17, 27). The cytomegalovirus strains employed in this study as controls for growth kinetics analyses were obtained by reconstitution of infectious viruses using the BACs pHB5, containing the genomic sequence of the HCMV laboratory strain AD169, and FIX-BAC, comprising the genomic sequence of the HCMV clinical isolate VR1814 (4, 12). Stocks of wild-type and recombinant viruses were prepared as described previously (27). Titration of the viral stocks was performed by either IE1p72 fluorescence (for AD169-based viruses), plaque assays (for FIX-BAC-based viruses), or determination of genomic equivalents imported into cells (2, 27). Briefly, for IE1p72 titration, HFFs were infected with serial dilutions of viral supernatants. Twenty-four hours postinfection (p.i.), the cells were fixed and stained with monoclonal antibody (MAb) p63-27 (2). IE1p72-positive cells were counted, and the viral titers (IE units) were calculated. For titration of viral stocks by plaque assays, HFFs were infected with various dilutions of viral supernatants and subsequently overlaid with agarose-containing medium. The infected cells were then incubated for up to 14 days p.i. until the appearance of plaques. For determination of viral genomic equivalents imported into cells upon infection, HFFs were infected with various PFU or IE units of viral stocks. At 10 to 14 h p.i., total cellular DNA was purified using the DNeasy tissue kit (Qiagen, Hilden, Germany). The DNA was then subjected to HCMV-specific real-time PCR as described previously (27, 47). Multistep growth curve analysis of recombinant viruses was also performed as described previously (27).

Oligonucleotides and plasmids.

All oligonucleotides used for PCR, mutagenesis, or sequencing were obtained from Eurogentec (Seraing, Belgium), ARK (Darmstadt, Germany), or Biomers (Ulm, Germany) (Table 1 lists the sequences). Eukaryotic expression vectors encoding the viral tegument protein pp71 (UL82) (pCB6-pp71) and the IE2p86 wild-type protein (pHM134), as well as the SUMOylation-negative mutant IE2 K175/180R (pHM1008), have been described previously (17, 27). For construction of vectors for yeast two-hybrid experiments, the SmaI/SalI fragment of IE2p86, either with mutations of SUMO acceptor sites or not mutated, was inserted into the yeast two-hybrid vector pGBT9 (BD Biosciences Clontech, Mountain View, CA), resulting in plasmids pHM1334 and pHM720, respectively (17). Additional deletion mutants of IE2p86 were generated by PCR amplification of IE2p86 fragments and ligation into the yeast two-hybrid vector pGBT9 via EcoRI and BamHI; primers 5′IE2-200GBT and 3′IE2-579GBT were used to generate plasmid pHM1641 (aa 200 to 579), primers 5′IE2-260GBT and 3′IE2-579GBT resulted in plasmid pHM1607 (aa 260 to 579), primers 5′IE2-260GBT and 3′IE2-402GBT resulted in plasmid pHM1642 (aa 260 to 402), primers 5′IE2-317GBT and 3′IE2-402GBT resulted in plasmid pHM1643 (aa 317 to 402), and the use of primers 5′IE2-468GBT and 3′IE2-579GBT resulted in plasmid pHM1647 (aa 468 to 579). The mammalian expression plasmids pHM2716 and pHM2717, as well as the yeast two-hybrid plasmids pHM2711 and pHM2712, expressing IE2p86 SIM mutants, were generated by site-directed mutagenesis using the primers IE2_SBMmut_motive2_5′ and IE2_SBMmut_motive2_3′, together with plasmid pHM134, pHM1008, pHM720, or pHM1334, respectively, as a template. Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit as instructed by the manufacturer (Stratagene). The yeast plasmid pHM2714, expressing the shortened IE2p86 SIM mutant from aa 200 to 579, was generated by site-directed mutagenesis using plasmid pHM1641 as a template, together with the primers MutapHM1641_5′ and MutapHM1641_3′. The prokaryotic expression plasmid pgalK, as well as cosmid pCM1058 and plasmids pRR47, pHM1438, pHM1469, and pHM1471, used for BAC mutagenesis, were described previously (8, 39, 40, 52).

Yeast two-hybrid analyses.

For yeast two-hybrid analyses, the S. cerevisiae strain AH109 (BD Biosciences Clontech, Mountain View, CA) was employed (19). Yeast cells were transformed according to a protocol described previously (17). The transformed plasmids were maintained within the yeast cells by selection for leucine and tryptophan prototrophy. In order to screen for reporter gene expression, the yeast cells were additionally selected for adenine prototrophy. Beta-galactosidase expression was determined by filter lift assays as described previously (17).

BAC mutagenesis.

BAC mutagenesis of pHB5 was performed as described previously (4, 27, 39). In brief, BAC kΔex5, lacking the exon 5 sequence of the major IE gene ORF UL122, was constructed by inserting a kanamycin resistance cassette, thus replacing exon 5 (15). Thereafter, the kanamycin resistance gene was deleted using FLP recombinase as described previously (6, 15). In order to reinsert ORF UL122-specific sequences into BACΔkΔex5, a recombination strategy involving cointegrate formation was applied (4). The recombination strategy for wild-type ORF UL122, resulting in BAC AD169rev, was performed as described previously (39). Construction of BAC AD169ex5-ΔS was performed in a similar way, with the exception of using the plasmid pHM1008 as the recombination sequence (17). For the construction of FIX-BAC recombinants, a two-step recombination strategy, using the galactokinase gene as a selection marker, was employed (52). For the first recombination, the galK gene with 5′ and 3′ flanking regions homologous to ORF UL122 exon 5 was amplified from plasmid pgalK (52), using the primers 5′IE2ex5galK and 3′IE2ex5galK. In order to accomplish the homologous recombination, the PCR fragment was then transformed into Escherichia coli (strain SW102), already harboring the BAC FIX-BAC. Cells were then plated on minimal medium (M63) agar plates containing 0.2% galactose and chloramphenicol and incubated at 32°C for 5 days. The colonies that appeared were streaked twice on MacConkey agar plates containing 0.2% galactose and chloramphenicol. Bacterial colonies appearing in bright red on MacConkey plates, and hence positive for the use of galactose, were used for the second recombination. For this, the ie2 wild-type sequence, as well as the ie2 K175/180R sequence, was amplified by PCR using primers IE2-ex5-2.Rek_5′ and IE2-ex5-2.Rek_3′, together with plasmids pHM134 and pHM1008, respectively, as templates (17, 21). In order to select for bacteria with loss of the galK gene, the transformed bacteria were plated on M63 agar plates containing 0.2% 2 deoxygalactose and chloramphenicol. Bacterial colonies appearing on these plates were subjected to further characterization (see below). Reconstitution of recombinant cytomegaloviruses using purified BAC DNA was performed as described previously (27).

Viral nucleic acid isolation and analysis.

BAC DNA was isolated either by a standard alkaline lysis minipreparation from 5-ml liquid cultures or by using the Nucleobond BAC100 kit (Machery Nagel, Düren, Germany) according to the manufacturer's instructions. Subsequently, BAC DNA was digested with restriction enzymes and separated by 0.7% agarose gel electrophoresis. For a further characterization of recombinant BACs by PCR, reactions were performed either directly from bacterial colonies or from purified BAC DNA using either Taq or Vent polymerase in the presence of dimethyl sulfoxide and formamide (27). The resulting PCR fragments were purified using the PureLink PCR Purification Kit (Invitrogen, Karlsruhe, Germany) and thereafter subjected to automated sequencing.

Antibodies and immunofluorescence analysis.

The polyclonal anti-IE86 and anti-pHM178 antisera directed against IE2p86, as well as the polyclonal antiserum recognizing ppUL69 of HCMV and the polyclonal rabbit antiserum SA2932 directed against the tegument protein pp71, were described previously (9, 17, 46, 54). MAb810, detecting the major IE proteins IE1p72 and IE2p86, was obtained from Chemicon (Hofheim, Germany) (31). MAb p63-27, recognizing IE1p72, as well as MAb28-4 against the major capsid protein (MCP), MAb-UL44 against the replication protein UL44, and MAb41-18, recognizing pp28, were described previously (2, 35, 51). Human PML was detected with MAb PG-M3 from Santa Cruz Biotechnology (Heidelberg, Germany). MAb AC-15, directed against the cellular protein beta-actin, was purchased from Sigma-Aldrich (Deisenhofen, Germany). For Western blot analyses, anti-mouse and anti-rabbit horseradish peroxidase-conjugated secondary antibodies were obtained from Dianova (Hamburg, Germany), and for immunofluorescence analysis, fluorescence-coupled Alexa Fluor 555 and Alexa Fluor 488 anti-rabbit and anti-mouse secondary antibodies were obtained from Invitrogen (Karlsruhe, Germany). For indirect immunofluorescence analysis, 1 × 10⁵ HFFs were grown on coverslips. The cells were infected with recombinant viruses at the MOIs indicated. The conditions for fixation, staining, and detection of viral and cellular genes were described previously (17, 39).

Western blot and immunoprecipitation analyses.

For Western blot analyses, infected cells were harvested and lysed under strictly denaturing conditions using sodium dodecyl sulfate (SDS) loading buffer. The proteins were separated on 10% to 12.5% polyacrylamide gels containing SDS and transferred onto nitrocellulose membranes (Protran BR 83; Schleicher & Schuell, Dassel, Germany). Western blotting and chemiluminescence detection were described previously (26).

Coimmunoprecipitations were performed according to a protocol modified from those of Lin et al. and Shen et al. (24, 37). Transiently transfected cells were washed once with phosphate-buffered saline containing 10 mM N-ethylmaleimide and then lysed in 100 μl per well of modified RIPA buffer containing 10 mM N-ethylmaleimide, 0.1 mM phenylmethylsulfonyl fluoride, and 1 mg/ml each of aprotinin, pepstatin A, and leupeptin. After a brief sonification, the lysates were cleared by high-speed centrifugation. A 50% protein A-Sepharose slurry, which had been incubated with the appropriate antibody at 4°C overnight, was added to the cleared lysates. The mixture was kept in an overhead tumbler at 4°C for 2 h. Afterward, the Sepharose beads were collected and washed five times in modified RIPA buffer. The precipitated protein complexes were recovered by boiling them in SDS sample buffer and were analyzed by Western blotting.

Identification of a noncovalent SIM within the IE2p86 exon 5 sequence.

We showed previously that IE2p86 can be posttranslationally modified by covalent attachment of SUMO proteins, which occurs at lysine residues located at amino acid positions 175 and 180 of IE2p86 (17). However, since mutation of these covalent SUMO attachment sites did not abrogate binding to SUMO in yeast two-hybrid assays, we and others assumed that additional motifs or factors might contribute to the IE2p86-SUMO interaction (1, 17). A reason for this sustained interaction could be an additional, noncovalent binding of IE2p86 to SUMO. In 2004, Song and colleagues identified an amino acid sequence motif (I/V-X-I/V-I/V) that was hypothesized to be responsible for noncovalent SUMO interactions (38). This SIM has subsequently been detected in several cellular proteins known to be modified by SUMO (e.g., human PML and hDaxx) and was shown to be important for the formation of SUMO-dependent protein networks (25, 37). When we compared the IE2p86 sequence to the amino acid sequences of human PML and hDaxx, a putative SIM starting at amino acid position 200 of IE2p86 was revealed (Fig. 1A). To investigate the functionality of this putative SIM, we performed site-directed mutagenesis, by which we replaced isoleucine 200 and valine 201 with alanine. Based upon previous results for hDaxx, it was anticipated that these mutations would disrupt SIM function (25). Mutagenesis was performed either in the context of wild-type IE2p86, resulting in mutant IE2M2, or using an IE2p86 protein that already harbored mutated acceptor sites for covalent attachment of SUMO, resulting in the double mutant IE2ΔSM2 (Fig. 1B). In order to investigate whether mutation of the SIM resulted in loss of SUMO binding, these point mutants, as well as several N- and C-terminal deletion mutants of IE2p86, were employed in yeast two-hybrid assays. For this, plasmid pHM709, harboring SUMO1 fused to the GAL4 activation domain, was transformed into the yeast strain AH109, together with either wild-type or mutant IE2p86 fused to the GAL4 DNA binding domain (19). After selection for transformants harboring both plasmids, these were restreaked on AWL dropout plates and tested for the activation of reporter genes by growth selection and filter lift assays. As observed previously, mutation of the covalent SUMO attachment sites was not sufficient to abrogate the IE2p86-SUMO interaction (Fig. 1C, II). Similarly, mutation of the putative SIM or deletion of the N-terminal 199 aa of IE2p86 did not affect the binding of IE2p86 to SUMO (Fig. 1C, VIII and III). However, deletion of more than 199 aa from the N terminus, thus removing the putative SIM, together with the SUMO acceptor sites, resulted in loss of interaction (Fig. 1C, IV, V, VI, and VII). The region within IE2p86 responsible for this could be further defined by the use of the IE2ΔSM2 mutant, containing mutations of both the covalent SUMOylation sites and the SIM, as well as the deletion mutant starting at aa 200 and harboring the putative SIM point mutations. Using these plasmids, no interaction could be observed, strongly suggesting that the putative SIM within IE2p86 mediates IE2p86-SUMO interaction in yeast (Fig. 1C, IX and X).

Efficient SUMOylation of IE2p86 depends on noncovalent IE2p86-SUMO interaction.

To examine the effect of the SIM mutation on the covalent modification of IE2p86 by SUMO in mammalian cells, a coimmunoprecipitation analysis was performed. To accomplish this, HEK-293T cells were cotransfected with either wild-type or mutant IE2p86 and FLAG-tagged SUMO1, respectively. The cells were lysed under denaturing conditions as described by Lin et al. (24), followed by coimmunoprecipitation using a polyclonal anti-IE86 antibody (25). The precipitates were analyzed by Western blotting with an anti-Flag MAb to detect FLAG-tagged SUMO, which was coprecipitated with IE2p86 (Fig. 2A). Furthermore, the expression levels of SUMOylated proteins and of IE2p86 were controlled by Western blotting of the cell lysates prior to immunoprecipitation (Fig. 2B). As expected, we could detect a band reacting with the anti-Flag antibody at approximately 105 kDa, corresponding to the mono-SUMOylated IE2p86, as well as several higher-migrating bands, which appeared to be the result of the attachment of multiple SUMO molecules to IE2p86 (Fig. 2A, lanes 6 and 8) (1, 3, 17). IE2p86 variants with mutated SUMO acceptor sites (IE2ΔS and IE2ΔSM2) could not be SUMOylated any longer, since neither the 105-kDa nor the higher-molecular-mass signals were observable after coimmunoprecipitation (Fig. 2A, lanes 7 and 9). Most importantly, when the noncovalent SIM was mutated alone, IE2p86 could still be SUMOylated, but SUMOylation was severely decreased in comparison to wild-type IE2p86 (Fig. 2A and B, compare lanes 6 and 8, and B, compare lanes 1 and 3). This indicated that the noncovalent interaction might be able to modulate the efficacy of IE2p86 SUMOylation. Furthermore, an additional interesting observation in this experiment was the consistent detection of increased protein levels, as well as a faster-migrating isoform of IE2p86, after expression of the double mutant IE2ΔSM2, suggesting that SUMO interactions might modulate IE2p86 phosphorylation and protein stability (Fig. 2B, middle, lanes 4 and 9). In summary, this experiment showed that efficient modification of IE2p86 by covalent attachment of SUMO depends on an intact noncovalent SIM.

Construction and reconstitution of recombinant cytomegaloviruses expressing SUMOylation-negative IE2p86.

Due to our initial observation that SUMOylation increases IE2p86-mediated transactivation, we hypothesized that introduction of SUMOylation-negative IE2p86 into the HCMV genome might lead to a recombinant virus with reduced replication capacity (1, 17). Surprisingly, however, it was reported in a previous publication that mutation of both IE2p86 SUMOylation sites in the context of the HCMV strain Towne exerted no discernible effect on viral replication (23). Already in 2003, Barrasa and coworkers noticed differences in the SUMOylation patterns and the transactivational capacities of IE2p86 proteins derived from the HCMV laboratory strains Towne and AD169 (3). Thus, in order to investigate the possibility of strain-specific differences, we decided to introduce the SUMOylation-negative IE2p86 into both the HCMV laboratory strain AD169 and the clinical isolate VR1814 (available as BAC clones pHB5 and FIX-BAC, respectively) (4, 16). Bioinformatics sequence analysis revealed that, in contrast to the Towne strain, the IE2p86 amino acid sequence encoded by clinical isolate VR1814 was identical to the corresponding sequence of strain AD169. In order to introduce mutations into ORF UL122 of AD169 and VR1814, we utilized a two-step recombination strategy in E. coli for both BAC pHB5 and FIX-BAC. For pHB5, the exon 5 sequence of ie2 was first replaced by a kanamycin resistance marker, which was subsequently exchanged for the mutated ie2 sequence using a shuttle strategy, as described previously (Fig. 3A) (27, 39). In addition to the mutated sequence, carrying the information for lysine-arginine substitutions at amino acid positions 175 and 180 (K175/180R), wild-type exon 5 was reinserted as a positive control. For the generation of the FIX-BAC-derived recombinant viruses, the exon 5 sequence of ie2 was, in the first step, partly replaced by the galactokinase (galK) gene via homologous recombination. In the second recombination step, the galK gene was removed and either the wild-type ie2 sequence or an ie2 sequence harboring the K175/180R mutation was reinserted, resulting in seamless removal of the selection marker (Fig. 3A) (for details, see Materials and Methods) (52). All recombinant BACs, pHB5 as well as FIX-BAC derived, were verified by restriction enzyme cleavage (Fig. 3B, C, and D), PCR, and nucleotide sequence analyses (data not shown). To reconstitute infectious viruses, HFFs were transfected with the purified DNAs of all recombinant BACs containing the desired sequences, as well as BACs pHB5 and FIX-BAC as controls. Approximately 14 days posttransfection, plaques appeared in all cell cultures, indicating that SUMOylation of IE2p86 is not essential for viral replication.

Expression of a SUMOylation-negative IE2p86 during viral replication leads to severely decreased viral growth and an impaired initiation of IE gene expression.

In order to investigate whether the loss of IE2p86 SUMOylation affects viral growth, we first performed multistep growth curve analyses of AD169ex5-ΔS, AD169rev, and AD169 viruses in parallel. For this, HFFs were infected in triplicate at an MOI of 0.01 with viral inocula normalized for equal IE units. After harvesting of the supernatants at 3, 5, 7, and 9 days p.i., they were subjected to IE1p72 immunofluorescence titration as described in Materials and Methods. Figure 4 shows that we could detect a clear decrease in the release of progeny virions of AD169ex5-ΔS compared to AD169 and AD169rev that occurred after day 5 p.i. The growth properties of the revertant virus AD169rev were not altered compared to wild-type AD169 virus (Fig. 4). This strongly suggests a replication defect of the AD169ex5-ΔS virus. As the FIX-BAC-derived viruses exhibited a strongly cell-associated phenotype, the generation of growth curves based on virus release was not possible with this set of viruses.

To further characterize the effects of the loss of IE2p86 SUMOylation, quantitative real-time PCR was performed to analyze viral-genome equivalents in infected cells. This was done in order to detect replication defects prior to the initiation of viral IE gene expression. For this, we infected HFFs with either AD169 and AD169ex5-ΔS or FIX-BAC and FIX-BACex5-ΔS. At 14 h p.i., the infected cells were harvested and total intracellular DNA was isolated. The DNA was analyzed by HCMV-specific real-time PCR using primers homologous to the HCMV major immediate-early gene region. This revealed that for AD169ex5-ΔS at least 5-fold and for FIX-BACex5-ΔS at least 10-fold more viral genomes were necessary to yield a number of IE1p72-positive cells or plaques equivalent to those of AD169 and FIX-BAC-infected cells, respectively (Fig. 5A and B).

Based on this result, HFFs were infected with equal genomic equivalents of the recombinant viruses FIX-BAC, FIX-BACrev, and FIX-BACex5-ΔS, followed by an analysis of plaque formation. All cells were infected with equal viral inocula corresponding to the number of viral genomes contained in 200 PFU of the FIX-BAC virus. After infection, the cells were overlaid with agarose-containing medium and incubated for a total of 14 days p.i. At 5, 7, 9, and 14 days p.i., the plaques were counted (Fig. 5C). As shown in Fig. 5C, cells infected with FIX-BACex5-ΔS showed a clearly reduced number of plaques compared to wild-type and revertant viruses. This was also observed with AD169-derived viruses (data not shown). In summary, these data clearly show that abrogation of IE2p86 SUMOylation in the context of the HCMV strains AD169 and FIX-BAC results in a replication defect that affects the initiation of IE gene expression.

Introduction of a SIM mutation does not further reduce the replication of a recombinant cytomegalovirus expressing SUMOylation-negative IE2p86.

Next, we wanted to analyze the relevance of the noncovalent interaction between IE2p86 and SUMO for viral infection. To achieve this, we generated recombinant cytomegaloviruses harboring an additional mutation of the noncovalent SIM of IE2p86. This was done via galK recombination of BAC pHB5, resulting in the recombinant BACs AD169ex5-M2 (containing a SIM mutation of IE2p86), AD169ex5-ΔSM2 (containing mutations of both covalent attachment sites and the SIM), and AD169rev2 (revertant AD169) (for details on the construction, see Materials and Methods). HFFs transfected with BAC pHB5, AD169rev2, AD169ex5-M2, and AD169ex5-ΔSM2 resulted in the formation of plaques approximately 12 days posttransfection.

In a first multistep growth curve analysis, using the successfully reconstituted viruses AD169, AD169rev2, and AD169ex5-ΔSM2, we observed impaired replication capacity of the double mutant AD169ex5-ΔSM2 (data not shown). In order to investigate whether mutation of the SIM aggravates the replication defect of an IE2p86 SUMOylation-negative virus, we directly compared the replication capacities of AD169ex5-ΔS and AD169ex5-ΔSM2 (Fig. 6A). Again, we were able to observe a distinct replication defect of the double-mutant virus AD169ex5-ΔSM2 compared to the wild-type virus AD169; however, no difference in viral-particle release was detectable between AD169ex5-ΔS and AD169ex5-ΔSM2 (Fig. 6A). Then, the SIM mutant AD169ex5-M2 was compared to the wild-type AD169 and the SUMOylation-negative virus AD169ex5-ΔS (Fig. 6B). AD169ex5-M2 exhibited a slight but significant decrease in viral-particle release at days 5 and 7; however, wild-type titers were reached at later time points, indicating that mutation of the IE2 SIM alone did not result in a major replication defect. Furthermore, the quantification of viral-genome equivalents in cells infected at equal MOIs confirmed our previous observation that significantly more viral genomes of the IE2p86 SUMOylation-negative viruses were necessary to yield an equivalent number of IE1p72-positive cells, but similar to the multistep growth curve experiment, no difference between AD169ex5-ΔS and AD169ex5-ΔSM2 was discernible (Fig. 6C). This shows that the additional loss of the noncovalent interaction of IE2p86 with SUMO does not further compromise viral replication.

Impaired accumulation of viral DNA after infection with IE2p86 SUMOylation-negative viruses.

After having demonstrated the requirement for IE2p86-SUMO interactions for efficient viral replication, we were interested in further characterizing which steps of the replication cycle are affected by the loss of IE2p86 SUMOylation. For this, we decided to analyze the accumulation of viral-progeny DNA during one round of replication. HFFs were infected with equal genomic equivalents of AD169, AD169rev, and AD169ex5-ΔS, as well as FIX-BAC, FIX-BACrev, and FIX-BACex5-ΔS. At various time points p.i. (12, 24, 48, and 72 h p.i. for AD169-derived viruses and 12, 24, 48, 72, 96, and 120 h p.i. for FIX-BAC-derived viruses), intracellular DNA was isolated and subjected to HCMV-specific real-time PCR for the quantification of viral genomes. Figure 7A and B shows a clearly reduced accumulation of viral genomes in cells infected with AD169ex5-ΔS and FIX-BACex5-ΔS in comparison to cells infected with wild-type and revertant viruses, respectively. Furthermore, no difference in the DNA accumulation of AD169ex5-ΔS- and AD169ex5-ΔSM2-infected cells was observable (Fig. 7C). Taken together, these results suggest that the loss of IE2p86 SUMOylation negatively affects the accumulation of viral-progeny DNA, which could be demonstrated for both the laboratory strain AD169 and the clinical strain FIX-BAC.

Localization of IE2p86 during lytic infection is not dependent on covalent or noncovalent SUMO interactions.

We have previously shown that, immediately after viral infection, IE2p86 accumulates in dot-like structures that are dynamically associated with PML bodies and that develop into viral replication compartments as infection progresses (39). Since covalent and noncovalent SUMO interactions have recently been demonstrated to affect the recruitment of proteins into subnuclear structures, we asked whether this was also the case for IE2p86 (30). To investigate this, HFFs were infected with either the wild-type HCMV strain AD169 or the AD169-derived recombinant viruses expressing mutated IE2p86, followed by immunofluorescence detection of proteins. As shown in Fig. 8A and C, the dot-like localization of neither the SUMOylation-negative IE2ΔS, the double mutant IE2ΔSM2, nor the SIM mutant AD169ex5-M2 was abrogated compared to that of wild-type IE2p86, although the sizes of the dots appeared to be slightly reduced (Fig. 8A, compare b and f to k, o, and s, and B, compare b and h to o and u). When we determined the number of IE2 dots per cell, no significant difference could be detected between viruses expressing mutant or genuine IE2p86 (Fig. 8B). This indicates that SUMO interactions do not affect the recruitment of IE2p86 onto parental viral genomes (39). Furthermore, mutant IE2p86 proteins were still efficiently incorporated into viral replication compartments at late times of infection (data not shown for AD169ex5-M2), but the appearance of these structures was delayed compared to cells infected with wild-type virus (Fig. 8C and D). This is most probably due to the impaired accumulation of viral DNA that we observed after infection with AD169ex5-ΔS and AD169ex5-ΔSM2. In summary, these experiments did not reveal a significant effect of SUMO interactions on the subcellular localization of IE2p86.

Viral-protein expression is drastically affected by the loss of covalent and noncovalent interaction of SUMO and IE2p86.

In order to further investigate if there is a delay or decrease in viral-protein expression due to mutations of IE2p86, we examined the expression levels of various viral proteins during lytic viral replication. For this, HFFs were infected with equal IE units of wild-type and recombinant viruses. Total cell lysates were prepared 24, 48, 72, and 96 h p.i. and analyzed by probing the immunoblot membranes with antibodies against different viral and cellular proteins (Fig. 9). First, we compared AD169 with AD169ex5-M2. As shown in Fig. 9A, no gross difference in the accumulation of IE1 or IE2 protein was detectable after infection with these viruses. However, upon closer inspection, we noted a decreased level of IE2p86 SUMOylation (Fig. 9A, compare lanes 3 and 8), as well as a slightly decreased accumulation of the late IE2 isoform IE2p40 (Fig. 9A, compare lanes 4 and 9). This is consistent with our observation made in transient-expression experiments that mutation of the IE2 SIM resulted in a lower level of covalent coupling of SUMO to IE2p86. In contrast, when AD169 was compared to the SUMOylation-negative viruses AD169ex5-ΔS and AD169ex5-ΔSM2, we noted a considerably decreased expression level of viral early (UL44)-, early-late (UL69 and pp71)-, and late (MCP and pp28)-phase proteins (Fig. 9B). For example, the expression of the pp28 protein started with a delay of approximately 24 h (Fig. 9B, second blot from bottom, compare lanes 3 and 4 to 8 and 9 and 13 and 14) in cells infected with the mutant viruses compared to wild-type-virus-infected cells. The level of protein expression was even more decreased in cells infected with the double-mutant virus AD169ex5-ΔSM2 than in cells infected with AD169ex5-ΔS, suggesting that the loss of the noncovalent SUMO interaction leads to a further reduction of viral-protein expression. At 24 h p.i., we could detect comparable expression levels of IE1p72 in all infected cells (Fig. 9B, lanes 1, 6, and 11), as the titration of the viral inocula was based on IE1p72 expression. However, at later time points during infection, the level of IE1p72 seemed to be reduced in cells infected with AD169ex5-ΔSM2 compared to AD169-infected cells (Fig. 9B, compare lanes 2 to 4 to 12 to 14). Probing the membrane against the IE2p86 protein revealed, as expected, no SUMOylated isoforms of IE2p86 in AD169ex5-ΔS- and AD169ex5-ΔSM2-infected cells (Fig. 9B, top blot, lanes 5 to 14). Furthermore, the expression level of IE2p86 was severely reduced and delayed in these cells. Thus, we could confirm the results of earlier immunofluorescence analyses, which showed delayed expression of IE2p86 during viral replication (Fig. 8). As already observed with the proteins of the early and late phases, the expression of IE2p86 was even further decreased in cells infected with AD169ex5-ΔSM2 in comparison to AD169ex5-ΔS-infected cells (Fig. 9B, top blot, compare lanes 6 to 9 to lanes 11 to 14). Interestingly, the decrease in protein expression was also detectable for the IE2p86 isoforms IE2p40 and IE2p60. However, the expression level of IE2p60 seemed to be more affected in AD169ex5-ΔS-infected cells (Fig. 9B, top blot, compare lanes 1 to 4 to lanes 6 to 9 and lanes 11 to 13). Thus, we conclude that the loss of IE2p86 SUMOylation drastically affects the accumulation of several viral proteins, which is consistent with the strong replication defect observed with the respective mutant viruses.