Specific Residues of a Conserved Domain in the N Terminus of the Human Cytomegalovirus pUL50 Protein Determine Its Intranuclear Interaction with pUL53*

Background: Interaction between the cytomegalovirus proteins pUL50 and pUL53 is essential for formation of a nuclear egress complex. Results: Mutations within a globular domain interfere with the function of pUL50. Conclusion: Residues Glu-56 and Tyr-57 of pUL50 are essential for binding to pUL53. Significance: Identification of the mode of important viral protein interactions promotes the development of novel antiviral strategies. Herpesviral capsids are assembled in the host cell nucleus and are subsequently translocated to the cytoplasm. During this process it has been demonstrated that the human cytomegalovirus proteins pUL50 and pUL53 interact and form, together with other viral and cellular proteins, the nuclear egress complex at the nuclear envelope. In this study we provide evidence that specific residues of a conserved N-terminal region of pUL50 determine its intranuclear interaction with pUL53. In silico evaluation and biophysical analyses suggested that the conserved region forms a regular secondary structure adopting a globular fold. Importantly, site-directed replacement of individual amino acids by alanine indicated a strong functional influence of specific residues inside this globular domain. In particular, mutation of the widely conserved residues Glu-56 or Tyr-57 led to a loss of interaction with pUL53. Consistent with the loss of binding properties, mutants E56A and Y57A showed a defective function in the recruitment of pUL53 to the nuclear envelope in expression plasmid-transfected and human cytomegalovirus-infected cells. In addition, in silico analysis suggested that residues 3–20 form an amphipathic α-helix that appears to be conserved among Herpesviridae. Point mutants revealed a structural role of this N-terminal α-helix for pUL50 stability rather than a direct role in the binding of pUL53. In contrast, the central part of the globular domain including Glu-56 and Tyr-57 is directly responsible for the functional interaction with pUL53 and thus determines formation of the basic nuclear egress complex.

Herpesviral capsids are assembled in the host cell nucleus and are subsequently translocated to the cytoplasm. During this process it has been demonstrated that the human cytomegalovirus proteins pUL50 and pUL53 interact and form, together with other viral and cellular proteins, the nuclear egress complex at the nuclear envelope. In this study we provide evidence that specific residues of a conserved N-terminal region of pUL50 determine its intranuclear interaction with pUL53. In silico evaluation and biophysical analyses suggested that the conserved region forms a regular secondary structure adopting a globular fold. Importantly, site-directed replacement of individual amino acids by alanine indicated a strong functional influence of specific residues inside this globular domain. In particular, mutation of the widely conserved residues Glu-56 or Tyr-57 led to a loss of interaction with pUL53. Consistent with the loss of binding properties, mutants E56A and Y57A showed a defective function in the recruitment of pUL53 to the nuclear envelope in expression plasmid-transfected and human cytomegalovirus-infected cells. In addition, in silico analysis suggested that residues 3-20 form an amphipathic ␣-helix that appears to be conserved among Herpesviridae. Point mutants revealed a structural role of this N-terminal ␣-helix for pUL50 stability rather than a direct role in the binding of pUL53. In contrast, the central part of the globular domain including Glu-56 and Tyr-57 is directly responsible for the functional interaction with pUL53 and thus determines formation of the basic nuclear egress complex.
Human cytomegalovirus (HCMV) 2 is the type species of ␤-Herpesvirinae (family Herpesviridae) and represents a ubiq-uitous, clinically highly important human pathogen. HCMV infection is mostly associated with mild pathogenesis in immunocompetent hosts but may cause severe systemic or even lifethreatening disease in immunosuppressed hosts and prenatally infected children (1). As characteristic for most DNA viruses, HCMV replicates genomes in the host cell nucleus. Thereafter, preformed viral capsids that are packaged with genomic DNA have to be transported to the cytoplasm for final maturation and release of infectious virions. Due to the large size of HCMV capsids (ϳ130 nm) (2), these cannot be transported through the nuclear pore complex (ϳ40 nm) (3). The currently accepted model for nuclear egress of HCMV and other herpesviruses is based on a complex envelopment/de-envelopment/re-envelopment process (4 -6). Hereby, a transient primary envelopment is achieved by budding through the inner nuclear membrane (INM) (4,7,8,9). Before herpesviral capsids gain access to the INM, however, the proteinaceous network of the nuclear lamina provides a major obstacle. Thus, the locally restricted destabilization of the nuclear lamina is a rate-limiting step during the viral replication process (10). As reported previously, the HCMV-specific nuclear egress complex (NEC) is composed of viral and cellular proteins, in particular protein kinases with the capacity to induce destabilization of the nuclear lamina (11)(12)(13)(14). Viral protein kinase pUL97 and cellular protein kinase C (PKC) play important roles by phosphorylating several types of nuclear lamins. In lamin A/C, a phosphorylation-dependent binding motif for the peptidyl-prolyl cis/trans-isomerase Pin1 leads to the local recruitment of Pin1, which is suggestive to contribute to the reorganization of the nuclear lamina (12). As two basic viral proteins essentially involved in the complex formation of a functional NEC, pUL50 and pUL53 have been characterized in their interaction properties (11,(15)(16)(17). The question if both proteins form a stable or transient complex, the latter possibly characterized by a dynamic change in composition over the time of infection, remains unclear so far. However, pUL50 and pUL53 directly interact with each other to form heterodimers (15,17). Furthermore, pUL50 possesses interaction domains for other NEC proteins, such as PKC and p32 (11,15). The region of pUL50 responsible for pUL53 interaction was previously attributed to the N-terminal amino acids 1-250 (11). In the case of pUL53, a short region comprising amino acids 50 to 84 was identified as the site required for binding to pUL50 (17) (Fig. 1A). The interaction of pUL50 with pUL53 appears as a prerequisite for a powerful relocalization activity of pUL50; full-length pUL50 relocalizes pUL53 toward a colocalization at the nuclear rim while evenly distributed throughout the nucleus in the absence of pUL50 (or the presence of mutant pUL50 that is lacking its C-terminal trans-membrane domain) (15,16). Proteins with homology to HCMV pUL50 and pUL53 are present in members of all three herpesvirus subfamilies, and a pairwise interaction between the respective proteins could be demonstrated for various examples (20 -23). In this report we identified a globular domain as well as an amphipathic helix within the N terminus of pUL50, and we investigated the impact of these regions on the intranuclear interaction with pUL53. Experimental evidence is provided demonstrating the functional importance of individual residues inside the globular domain. In addition, novel computational data for pUL50 is presented supporting our current model on combined structure-function relationship. A comparison with the respective elements in homologous proteins of other herpesviruses is discussed.
Indirect Immunofluorescence Assay and Confocal Laser Scanning Microscopy-HeLa cells were cultivated and grown on coverslips for transient transfection by the use of Lipofectamine 2000 (Invitrogen) under previously described conditions (11). Primary human foreskin fibroblasts (HFFs) were used for combined transfection-infection experiments. First, HFFs were transfected with expression plasmids by the use of FuGENE HD (Roche Applied Science) according to the manufacturer's protocol. One day later transfected HFFs were infected with HCMV laboratory strain AD169 at a multiplicity of infection of 1.0. At 2 days post-transfection (transfected HeLa cells) or at 3 days post-infection (transfected-infected HFFs) cells were fixed and permeabilized after indirect immunofluorescence staining as described previously (11). The mouse monoclonal antibody (mAb) anti-FLAG (M2; Sigma) and the rabbit polyclonal antiserum (pAb) anti-hemagglutinin (HA.11; Covance Inc.) were used to detect the transiently expressed tagged proteins. The mouse pAb anti-UL53 was used for the detection of pUL53 of HCMV (kindly provided by Dr. P. Dal Monte, Dept. of Haematology, St. Orsola Malpighi General Hospital, University of Bologna, Bologna, Italy). Secondary antibodies used for double staining were Alexa Fluor 488-conjugated goat anti-rabbit IgG (HϩL) and Alexa Fluor 555-conjugated goat anti-mouse IgG (HϩL; New England Biolabs GmbH). Images were acquired using a Leica TCS SP5 confocal laser scanning microscope equipped with a 63ϫ HCX PL APO CS oil immersion objective lens (Leica). Images were analyzed, and signal intensities were quantified using LAS AF software (Leica).

Prokaryotic Expression and Purification of pUL50(1-181)-
The N-terminal fragment of pUL50 containing residues 1-181 was expressed in Escherichia coli BL21(DE3) as a SUMO fusion protein using the pE-SUMOpro system (LifeSensors Inc.) containing a His tag at the N terminus. Bacterial cells were grown in LB medium in the presence of 100 g/ml ampicillin at 28°C to an optical density A 600 of 0.5-0.6 before the temperature was lowered to 20°C, and protein expression was induced with 0.25 mM isopropyl ␤-D-thiogalactopyranoside (IPTG). The protein was expressed overnight, and bacterial cells were harvested by centrifugation and resuspended in His-trap binding buffer (50 mM phosphate buffer (pH 7.4), 300 mM NaCl, 30 mM imidazole, 3 mM DTT) containing protease inhibitors and disrupted by sonication. After centrifugation at 100,000 ϫ g for 1 h at 4°C, the supernatant containing His-SUMO-pUL50(1-181) was purified by affinity chromatography using a His-Trap column (GE Healthcare). After the first His-trap column, the SUMO tag was cleaved using 5 units of SUMO protease/100 g of fusion protein for 5-7 days at 4°C in 20 mM Tris (pH 8.0), 150 mM NaCl, and 3 mM DTT. Both the SUMO tag and SUMO protease were removed via a second His-trap chromatography. pUL50(1-181) was further purified by size exclusion chromatography and concentrated to 10 mg/ml in 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 3 mM DTT. Typically, 1 liter of bacterial culture yielded about 2.5 mg of pure protein.
Circular Dichroism Spectroscopy and Thermal Denaturation Analysis-Circular dichroism (CD) measurements were performed at 20°C using a Jasco J-810 spectropolarimeter (Jasco International Co.) and a cuvette with a 0.1-cm path length. All experiments were performed in 20 mM potassium phosphate buffer (pH 7.6) at a protein concentration of 0.25 mg/ml. Spectra were registered between 185 and 260 nm and corrected for the contributions from the phosphate buffer. Spectra were accumulated eight times with a bandwidth of 1.0 nm and scan speed of 20 nm/min, time response of 1 s, and data pitch 0.1 nm. CD spectra were analyzed using the deconvolution program CDSSTR within the Jasco CDPro program. Heat-induced unfolding transitions were also measured under similar buffer conditions and protein concentration. Ellipticity values were recorded at 222 nm by raising the temperature from 20 to 96°C using a PTC 423S/15 Peltier element in 1°C steps, and denaturation analysis was performed using the Jasco Spectra analysis program.
Secondary Structure Prediction-A secondary structure consensus prediction of HCMV-encoded pUL50 and homologous proteins was performed at the NPS@ server (network protein sequence analysis) (26) and at the PSIPRED server (protein structure prediction) (27) using several independent methods: DSC (discrimination of secondary structure class) (28), PHD (profile network from Heidelberg) (29), MLRC (multivariate linear regression combination) (30), and PSIPRED (two feedforward neuronal networks) (31). Potential globular domains by Russell/Linding definition were assessed by GlobPlot 2.3 (predictor of intrinsic protein disorder and globularity) (32). The HeliQuest server (33) was used to determine the amphiphilicity of predicted ␣-helices by analyzing the distribution and properties of their side chains.

RESULTS
The N-terminal Region 10 -169 of pUL50 Is Responsible for pUL53 Binding-Our previous studies demonstrated a high affinity interaction between HCMV pUL50 and pUL53, both harboring several conserved functional elements and domains ( Fig. 1, A and B). To narrow down the domain responsible for pUL53 interaction, C-and N-terminal truncation mutants of pUL50 were generated and fused to a C-terminal coding region of GFP or HA tag, respectively. First, pUL50 truncation mutants were tested for their ability to interact with pUL53 in a CoIP assay (Fig. 2). With C-terminal truncations, specifically the construct comprising amino acids 1-181 of pUL50 was coimmunoprecipitated by pUL53 ( Fig. 2A, CoIP, lane 4), whereas further C-terminal deletions of pUL50 abrogated the interaction (lanes 5-8). As an additional important feature, truncation mutants 1-181 and 1-150 of pUL50 were expressed at high levels ( Fig. 2A, expression controls, lanes 4 and 5 and lanes 9 and 10), whereas truncations of more than N-terminal 150 amino acids showed decreased expression levels (lanes 6 -8 and 11-13). It should be stressed that the interaction-negative mutant 1-150 ( Fig. 2A, lanes 5 and 10) was expressed to similar levels as the interaction-positive mutant 1-181 (lanes 4 and 9), thus illustrating the reliability of this experiment even under variable expression situations (also note the reliable amounts of proteins present in the immunoprecipitates; Fig. 2A, precipitation controls). In addition, any unspecific binding of the GFP tag was excluded by control settings, i.e. using GFP-tagged PKC␣ ( Fig. 2A, lane 2) and empty-vector controls (lanes 9 -13). With N-terminal truncations, deletions up to amino acid 10 were tolerable for interaction (Fig. 2B, CoIP, lanes 4 and 5), although the CoIP signal was weaker compared with full-length pUL50 (lane 3). Notably, the expression was lower for mutants 15-397 and 20 -397 (Fig. 2B, expression controls, lanes 6 and 7) than for mutants 5-397 and 10 -397 (lanes 4 and 5) possibly indicating a stabilizing role of amino acids 11-14. Here again, the reliability of the experiment was provided by an additional staining of the CoIP blot with a pUL50-specific antibody demonstrating equal amounts of precipitated proteins (Fig. 2B, precipitation controls, lanes 3-7). Thus, CoIP analyses with these N-and C-terminal truncation mutants suggest that amino acids 10 -181 of pUL50 are sufficient for binding to pUL53.
To further examine the pUL53 interaction domain, constructs were generated that should transfer the interaction region of pUL50 into a GFP-␤-gal fusion protein. These constructs were also analyzed concerning their ability to interact with pUL53 (Fig. 2C). In contrast to GFP-␤-gal alone (vector; lane 1) or a fusion protein containing a putative non-interacting pUL50 sequence (i.e. amino acids 236 -358; lane 4), fusion proteins that comprise amino acids 1-358, 1-204, or 10 -169 were all able to coimmunoprecipitate pUL53 (lanes 2, 3, and 5). Notably, the CoIP signal in the case of 10 -169 was weaker compared with 1-358 and 1-204. This demonstrates that a minimal interaction domain (amino acids 10 -169) can be transferred to a large non-interacting GFP-␤-gal fusion protein to confer binding to pUL53, thus confirming the domain mapping reported by others (17). Taken together, we therefore concluded that amino acids 10 -169 are sufficient to mediate interaction with pUL53 (summarized in Fig. 3B), but the complete N terminus of pUL50 (mutant 1-181) is necessary for an optimal, high affinity interaction equivalent to full-length pUL50.
The Interaction of pUL53-pUL50 Is Essential for the Relocalization of pUL53 to the Nuclear Rim-In a next step, pUL50 truncation mutants were analyzed with respect to their ability to recruit pUL53 to the nuclear envelope using immunofluorescence analysis and confocal laser-scanning microscopy (Fig. 3).
Notably, all analyzed C-terminal truncation mutants showed a diffuse nuclear localization due to the lack of the transmembrane domain and, therefore, were not able to recruit pUL53 to the nuclear rim (summarized schematically in Fig. 3B). In the cases of N-terminal truncation, all mutants expressed alone were localized at the nuclear rim and cytoplasmic membranes similar to full-length pUL50 (data not shown). When coexpressed with pUL53, the interaction-competent mutant pUL50(5-397)-HA showed a perfect colocalization with pUL53 at the nuclear rim (Fig. 3A, panels i-m; compare fulllength in panels e-h), whereas no recruitment was noted for non-interacting mutant pUL50(20 -397)-HA (panels n-q). This correlation between detectable pUL50-pUL53 interaction and pUL53 relocalization was a general feature of all mutants analyzed, i.e. pUL53 recruitment to the nuclear rim was restricted to those mutants of pUL50 that were positive for interaction in the CoIP analysis ( Fig. 3B) (11). In addition, coexpression experiments demonstrated that the complete nuclear rim localization of pUL50 itself was also restricted to those versions of pUL50 that are interaction-competent for pUL53 (Fig.  3A, panels e-m compared with panels n-q). These findings underline that the interaction pUL50-pUL53 is a prerequisite for the correct nuclear localization of both proteins. Recently, a similar finding for the homologous proteins of HSV-1 was pub-  (19) divided the N termini of pUL50 homologs into three conserved regions probably due to a different selection of analyzed sequences and the use of another alignment algorithm. The N terminus of pUL50 was initially identified as the binding region for pUL53 (dashed box) (11,17); essential residues of this interaction region were mapped in this study (Glu-56, Tyr-57). C, shown are far UV-CD spectra of pUL50(1-181) measured at 20°C. D, temperature-induced denaturation was monitored by change of ellipticity at 222 nm for pUL50(1-181), which exists as a monomer in solution. The blue line denotes experimental data, and the dashed black line shows a two-state fitting model. Note, that the melting temperature determined from the fitting is 41.47°C, and the heat-induced denaturation was not reversible.
lished by Roller et al. (34), thus suggesting that this functional aspect may be conserved among herpesviruses.
Comparison between Conserved Regions Using Sequence Alignments of pUL50 Homologs-Multiple sequence alignments of pUL50 and homologous proteins of human herpesviruses (i.e. HSV-1, HSV-2, VZV, HHV-6A, HHV-6B, HHV-7, EBV, and HHV-8) and animal herpesviruses (i.e. MCMV and PrV) were generated to address the question of a conserved mode of interaction (Fig. 4A). Interestingly, the N terminus is the most conserved part of pUL50 (Fig. 1B). In particular, the consensus sequence within this region revealed a high degree of conservation of the corresponding residues. Within this conserved part, more than 30% (␤-Herpesvirinae) or 3% (Herpesviridae) of the residues were identical and were found in all analyzed sequences. Thus, the N terminus of pUL50 and homologous proteins is relatively well conserved and, in case of pUL50, contains the pUL53 interaction region. We next addressed the issue, in which structural features determine the interaction properties of the N-terminal part of pUL50.
Structural Information on pUL50 Determined by Bioinformatic and Biophysical Evaluation-A modeling approach derived from the secondary structure prediction of the HCMVencoded pUL50 amino acid sequence (11) suggested that two distinct structural elements could be present in the N-terminal region of amino acids 1-181 (Fig. 1B). These elements, i.e. an N-terminal ␣-helix (3-20) and a globular domain (45-181), appeared to have an impact on the interaction properties of pUL50 based on the finding that truncation of either of these two elements was incompatible with pUL53 binding. However, the use of novel globularity scales for globular domain prediction suggested that the globular domain comprises amino acids 1-209, indicating that the N-terminal ␣-helix is, rather, a part of the globular domain than a distinct structural element (data not shown). To confirm this new concept, pUL50(1-181) was expressed in E. coli, affinity-purified via an N-terminal His tag, and analyzed by CD spectroscopy. Far UV-CD spectra of purified pUL50(1-181) suggested a folded protein with a mixed ␣-helix and ␤-sheet fold (Fig. 1C). Importantly, the temperature-induced change of ellipticity at 222 nm reveals a highly cooperative unfolding transition pointing to the existence of a single globular fold within the N-terminal residues 1-181 of pUL50 (Fig. 1D). Thus, the globular domain appears to comprise at least residues 1-181 including the ␣-helix in the N terminus of pUL50.
The relatively high degree of sequence conservation in the N terminus of pUL50 and homologs, i.e. ␣-herpesviral (HSV-1 and PrV UL34), ␤-herpesviral (MCMV pM50), and ␥-herpesviral (EBV BFRF1) homologs, allowed a comparative setting to predict their tendency to adopt a common regular secondary structure in the corresponding regions. Interestingly, N-terminal ␣-helices were predicted in all cases (supplemental Fig. S1). Closer inspection of side chain properties disclosed segregation of hydrophobic and polar residues opposed to each of the two faces of these ␣-helices (Fig. 4B), which is characteristic for amphipathic helices (37). The mean helical hydrophobic moment H (representing a degree of amphiphilicity of helices) (38) was in a similar range for all analyzed proteins (i.e. H of ϳ0.4 for MCMV pM50 to ϳ0.65 for EBV BFRF1).
Specific Residues within the Globular Domain of pUL50 Are Responsible for the Functional Interaction with pUL53-To investigate which residues within the interaction domain are directly involved in binding to pUL53, we performed alanine replacement mutations of single amino acids. First, a possible involvement of the predicted N-terminal amphipathic helix was analyzed. Region 10 -20 was subjected to mutation, as deletion of these residues resulted in the complete loss of interaction with pUL53 (Fig. 2B). In particular, we produced three mutants with replacements on the predicted polar side (i.e. D10A, Q13A, and K20A) and four mutants with replacements on the predicted hydrophobic side (i.e. L11A, V12A, T15A, and I18A) (Fig. 4C). Additionally, two double-mutants were gener-ated with double-replacements on either side (i.e. the polar, D10A/Q13A, or hydrophobic side, L11A/V12A, respectively). These point mutants were used for CoIP and immunofluorescence analyses to determine their interaction and recruitment properties ( Fig. 5A and Fig. 6A). CoIP analysis demonstrated that coimmunoprecipitation of pUL53 was positive for the four mutants with replacements on the polar side (Fig. 5A, CoIP, lanes 3-5 and 9) and the two hydrophobic side mutants V12A and T15A (lanes 6 and 7). A significantly reduced CoIP signal was detected for mutant I18A (Fig. 5A, CoIP, lane 8). Importantly, the Western blot expression controls revealed a decreased expression level for mutant I18A (Fig. 5A, expression controls, lane 8; compare wild-type in lane 2). Staining of ␤-actin served as a loading control (Fig. 5A, expression controls, bottom panel). Interestingly, this imbalance in expression levels was mostly compensated during the CoIP procedure by the limited binding capacity of the precipitation antibody, finally leading to an immunoprecipitation of all mutants quantitatively similar to wild-type pUL50 (Fig. 5A, precipitation controls, lanes 2-9). Notably, no CoIP signal was observed for mutants with hydrophobic side mutation L11 (supplemental Fig. S2A). However, the loss of interaction properties of singlemutant L11A and double-mutant L11A/V12A was rather a result of inaccurate protein folding than caused by a direct involvement of L11 in binding to pUL53. This was concluded from three observations: mutants L11A and L11A/V12A were expressed at significant lower levels as wild-type pUL50 (supplemental Fig. S2A and data not shown); L11A showed defective intracellular localization in transfected HeLa cells (i.e. L11A is not only localized at the INM but accumulates at aggregates in the cytoplasm; supplemental Fig. S2B); L11A was not able to interact with PKC␣ as another pUL50 binding partner FIGURE 3. Interaction with pUL50 is required for pUL53 recruitment to the nuclear envelope. A, HA-tagged full-length pUL50 or N-terminal deletion mutants were transiently coexpressed with FLAG-tagged pUL53 in HeLa cells. At 2 days post-transfection, cells were subjected to indirect immunofluorescence analysis using the antibodies indicated. Samples were subsequently analyzed by confocal laser-scanning microscopy. A selection of representative images of transfected cell nuclei is shown (the depicted phenotype was observed in more than 99% of pUL50-pUL53-positve cells in each case). DAPI, 4Ј,6-diamidino-2-phenylindole. B, shown is a schematic summary of combined CoIP and immunofluorescence (IF) data obtained with N-and C-terminal deletion mutants of pUL50 (this study and Milbradt et al. (11)). Based on these results, the region required for binding to and recruitment of pUL53 to the nuclear envelope was determined. Structural elements within pUL50, which are present in regions important for protein function, are indicated. ϩ, CoIP positive/recruitment of pUL53 to the nuclear envelope; Ϫ, CoIP negative/no recruitment; n.d., not determined.
(supplemental Fig. S2C). In the case of the stable mutants of the amphipathic helix, analysis of recruitment properties produced positive results consistent with the CoIP findings with excep-tion of mutant I18A that showed an intermediate phenotype: as expected, all pUL53-interacting mutants were able to recruit pUL53 to the nuclear envelope (see examples in Fig. 6A; sum-  (35) and PrV (21) are marked with colored bars. Conserved residues Glu-56 and Tyr-57 of MCMV (36) and HCMV pUL50 have been shown to be essential for binding to pUL53 (red framing). Alignment coloring scheme: black, non-similar residues; blue on cyan, consensus derived from a block of similar residues; black on green, consensus derived from the occurrence of greater than 50% of a single residue; red on yellow, consensus derived from a completely conserved residue. B and C, putative amphipathic helices in the N terminus of pUL50 homologs are shown. B, the HeliQuest server (33) was used to analyze side chain properties and their distribution in the predicted ␣-helices. Various sequence stretches for each pUL50 homolog were analyzed that correspond to the results of the secondary structure prediction (supplemental Fig. S1). Helical wheels are depicted giving the highest mean helical hydrophobic moment (H). Amino acids (aa) are colored according to side chain properties: yellow, hydrophobic; red/blue, acidic/basic (amount of red/blue decreases proportionally to the acidity/basicity level); gray, neutral. C, shown is a schematic presentation of side chain properties of the N-terminal ␣-helix formed by amino acids 3-20 of HCMV-encoded pUL50. Amino acids are illustrated in stick presentation and colored according to side chain properties as explained for B. Only amino acids were labeled that were chosen for site-directed replacement by alanine. marized in Table 1). Surprisingly, mutant I18A was not impaired in its recruitment properties (Fig. 6A, panels l-p) even though its ability to interact with pUL53 was reduced in CoIP assay (Fig. 5A). The latter observation suggested that mutation of specific residues on the hydrophobic side might affect the folding of pUL50 but that the amphipathic helix is not directly involved in the functional interaction with pUL53.
In a second step we concentrated on the most conserved part of the globular domain to identify a region of pUL50 that is directly involved in the binding of pUL53. Therefore, six residues within the central part of the globular domain, fully conserved among all analyzed sequences (i.e. Glu-56, Asn-76, Gly-78, Pro-90, Lys-123, and Arg-136), as well as four residues conserved only among ␣and/or ␤-Herpesvirinae (i.e. Tyr-57, Leu-116, Gly-152, and Pro-153) were replaced by alanine. Analyzing the interaction and recruitment properties of these mutants, CoIP and relocalization analyses produced consistent results. In the CoIP analysis, mutants E56A and Y57A were negative for a wild-type-like, efficient interaction with pUL53 but retained a marginal level of residual binding activity (Fig.  5B, CoIP, lanes 1 and 2). Moreover, all these mutants of the central part of the globular domain showed mostly normal, wild-type-like expression levels (Fig. 5B, expression controls). As an exception in this regard, the slightly reduced protein levels detected for the pUL53 non-interacting point mutants E56A and Y57A might result from the lack of a stabilizing binding to pUL53, which might indicate the importance of interaction for protein stability of both partners (Fig. 5B, expression controls,   lanes 1 and 2, compared with lanes 3-10). However, in clear contrast to I18A, mutants E56A and Y57A had lost the ability to relocalize pUL53 to the nuclear rim (Fig. 6B, panels a-k), whereas other mutants of the central part of the globular domain were not impaired in this activity (see examples in Fig.  6B). A summary of the combined interaction and recruitment analyses is depicted in Table 1.
To exclude that the reduced ability of mutants E56A and Y57A to interact with pUL53 was caused by general defects in protein folding, mutants were analyzed in regard of their interaction with PKC␣ as another known pUL50 binding partner (supplemental Fig. S2C). Interestingly, CoIP analysis demonstrated that mutations E56A and Y57A were tolerable for interaction with PKC␣ (supplemental Fig. S2C, CoIP, lanes 5 and 6), although the CoIP signal was weaker compared with wild-type pUL50 (supplemental Fig. S2C, lane 2). Notably, pUL53 and PKC␣ interaction regions within pUL50 partly overlap (11). This indicates that pUL50 mutants E56A and Y57A had lost the pUL53 interaction phenotype most probably as a direct consequence of the amino acid replacements but not an overall loss of protein folding. This assumption was confirmed by the fact that unstable mutant L11A not only showed a loss of pUL53 binding but also PKC␣ binding (as represented by a very low CoIP signal at background levels; supplemental Fig. S2C, CoIP, lanes 1 and 3). The reliability of this experiment was provided by additional immunostaining of the CoIP blot with a pUL50-specific antibody that demonstrated equal amounts of pUL50 mutants in the precipitates (supplemental Fig.  S2C, precipitation controls, lanes 2-6). In conclusion, among all analyzed stable mutants, only E56A and Y57A were impaired in both their pUL53 interaction and recruitment properties, suggesting that the central part of the globular domain (including Glu-56 and Tyr-57) is directly involved in pUL53 binding.
Residues Glu-56 and Tyr-57 of pUL50 Are Essential for Colocalization with pUL53 in HCMV-infected Cells-To confirm the importance of residues Glu-56 and Tyr-57 of pUL50 for the functional interaction with pUL53 in the native environment, combined transfection-infection experiments were performed. Therefore, primary HFFs were individually transfected with expression constructs coding for HA-tagged wild-type pUL50 or the non-interacting mutants E56A and Y57A. The transfected HFFs were additionally infected with HCMV strain AD169. Three days post-infection, the transfected-infected HFFs were used for immunofluorescence analysis and analyzed by confocal laser-scanning microscopy to detect viral proteins in HCMV-infected cells. Thus, the experimental setting allowed a mixed expression of native viral proteins and transiently coexpressed pUL50-HA (Fig. 7). Notably, in transfected HeLa cells overexpressing pUL53 without any other viral proteins, pUL53 shows a diffuse nuclear localization (15). However, in HFFs infected with HCMV, pUL53 is incorporated into the NEC (6,11,14), which is represented here by a speckled aggregation of pUL53 at specific sites of the nuclear envelope (Fig. 7A, panels e, p, and u). In the combined setting of transfected-infected cells, transiently expressed pUL50-HA colocal- ized with accumulations of the virus-produced pUL53 (Fig. 7A, panels a-e, filled arrowheads). In contrast, the non-interacting mutants E56A and Y57A were distributed homogenously at the nuclear envelope of infected cells (Fig. 7A, panels l-u). As expected, pUL53 still accumulated at distinct sites of the nuclear envelope in these cells (Fig. 7A, panels p and u, open  arrowheads). This was explained by the fact that virus-produced wild-type pUL50 (unstained) was also present, but the pUL50 mutants were missing in these speckled sites (panels p and u). Notably, the homogenous distribution of E56A and Y57A in HCMV-infected cells was similar to the localization of these mutants (data not shown) or wild-type pUL50-HA in uninfected cells (Fig. 7A, panels f-k). In addition, sequential optical planes in the z axis (z stacks) were acquired from these samples, and signal intensities were quantified (Fig. 7, B-D). Thereby, single focal planes at the bottom of the infected cell nuclei were used (Fig. 7, B-D, indicated by dashed lines in xz and yz). Consistent with images taken from the center of the nuclei (Fig. 7A), transiently expressed pUL50-HA and virusproduced pUL53 accumulated at distinct sites of the nuclear envelope (Fig. 7B, inset in xy). Quantification of signal intensities demonstrated that pUL50-HA and pUL53 colocalize at these sites (Fig. 7B, graph; indicated by covering peaks of pixel intensities). As expected, signal intensities of mutants E56A (Fig. 7C) and Y57A (Fig. 7D) remained at a bottom line and did not show any peak of colocalization, as also represented by a diffuse green staining at the bottom of the nuclei (Fig. 7, C and D, inset in xy). In summary, we concluded that amino acids Glu-56 and Tyr-57 of pUL50 are essential for the functional interaction with pUL53 in the native environment of HCMVinfected cells.

Multiple Activities of pUL50 and Homologs during Virus
Replication-The conserved herpesviral proteins pUL50 and pUL53 play important roles during the nuclear egress of viral capsids. For ␤-herpesviruses, a focus of investigations was directed to the initial local disruption of the nuclear lamina, which allows capsids access to the INM. In particular, pUL50 and pUL53 represent the core of a viral-cellular NEC, which mediates the recruitment of protein kinases essential for the phosphorylation-dependent disassembly of the nuclear lamina (11)(12)(13)(14)16). For the ␣-herpesviral homologs of pUL50 and pUL53, additional evidence suggested that both proteins function at later stages of nuclear egress including capsid docking at the INM, curvature of the membrane around capsids, and deenvelopment at the outer nuclear membrane. Furthermore, a putative role for cell-to-cell spread was discussed for the pUL50 homolog of HSV-1 (UL34) (14, 19, 34, 39 -42). All these functions rely on the pUL50-pUL53 interaction as an essential step of complex formation at the nuclear envelope. Thus, the identification of essential structural elements required for interaction has been aspired.
Conserved Residues Glu-56 and Tyr-57 of pUL50 Are Essential for NEC Formation-Data in this study suggest that the complete N terminus of pUL50 (amino acids 1-181) is necessary for efficient binding to pUL53 (Figs. 2 and 3). Nevertheless, a domain-swapping experiment demonstrated a transfer of amino acids 10 -169 to a large GFP-␤-gal fusion protein sufficient to confer pUL53 interaction. Recently, an N-terminal ␣-helix (amino acids 3-20) and a putative globular domain, including at least residues 45-181, were identified by bioinformatic analyses in this region (11). Notably, novel data derived from thermal denaturation experiments for pUL50 indicate that the complete N terminus adopts a globular fold (i.e. amino acids 1-209), and there is no evidence that the N-terminal ␣-helix represents an independent structural element. A short region within the globular domain including highly conserved Glu-56 and Tyr-57 appears to be required for high affinity binding to pUL53, as amino acid exchanges of both residues to alanine abrogated pUL50-pUL53 coimmunoprecipitation (Fig. 5). This suggests that Glu-56 and Tyr-57 may be exposed at the surface of the globular domain. General defects in protein folding produced by E56A and Y57A appear not to be likely as the mutants (Figs. 5 and 6 and supplemental Fig. S2C) (i) are stably expressed to similar levels compared with wild-type pUL50, (ii) show a regular localization at the nuclear envelope, and (iii) are still able to bind PKC␣. In addition, residues Glu-56 and Tyr-57 of pUL50 were not only proven to be important for the physical interaction with pUL53 but also for the recruitment of pUL53 to the HCMV-specific NEC. In cotransfected cells, the noninteracting mutants E56A and Y57A failed to recruit pUL53 to the nuclear envelope (Fig. 6); in the transfection-infection setting, mutants E56A and Y57A were not incorporated into the HCMV-specific NEC (Fig. 7). In conclusion, the central part of the globular domain including Glu-56 and Tyr-57 mediates binding to pUL53 and is, therefore, essential for pUL50 functionality.
The Mode of Interaction between pUL50 and pUL53 Is Conserved among Cytomegaloviruses-The general importance of Glu-56 and Tyr-57 was illustrated by Bubeck et al. (36), demonstrating that the corresponding residues in the MCMV pro-tein pM50 were similarly essential for coimmunoprecipitation of pM53. Surprisingly, despite high conservation of these residues among Herpesviridae, the pUL50-homologous region of the HSV-1 UL34 protein proved to be dispensable for UL31 interaction (35). However, the residues homologous to Glu-56 and Tyr-57 (HSV-1 UL34 residue Glu-67 and Tyr-68) are critical for virus replication in HSV-1 (19,43). In particular, a charged cluster mutant of HSV-1 UL34, CL06, also containing E67A, was not able to complement infection with a UL34-null virus. Surprisingly, CL06 failed to localize to the INM but still colocalized with UL31 suggesting some residual activity in UL34-UL31 interaction (43). Moreover, residue Tyr-68 of UL34 showed similar functional importance, as mutation Y68A produced a major defect in virus replication, impairment in nuclear capsid egress, and defects in cell-to-cell spread. Interestingly, indirect evidence suggested that mutation Y68A did not interfere with UL31 interaction (19). Taken together, we conclude that the mode of interaction of viral egress proteins is conserved among cytomegaloviruses (dependent on the highly conserved residues Glu-56 and Tyr-57) but differs in several aspects from HSV-1 and possibly other herpesviruses.
Importance of the N-terminal Amphipathic Helix of pUL50-The presence of an amphipathic helix in the extreme N terminus of pUL50 was predicted by bioinformatics analyses (also including conservation among homologous proteins of Herpesviridae). This prediction proved to be compatible with a functional analysis of replacement mutants. Residue Ile-18 appeared to be functionally important as it is situated on the hydrophobic face of the putative amphipathic helix of pUL50. In our experimental evaluation, mutant I18A showed an impairment of binding to pUL53 but still recruited pUL53 to the nuclear envelope. In accordance with a reduced expression level in transfection experiments, this intermediate phenotype might result from inaccurate protein folding than from direct involvement in pUL53 binding. Interestingly, mutation of amino acid Leu-11, positioned adjacent to Ile-18 (Fig. 4, B and C), resulted in a more drastic phenotype. Mutant L11A completely lost its functionality in interaction (neither binding to pUL53 nor to PKC␣) and recruitment of pUL53 (supplemental Fig. S2), most probably due to a loss of correct folding and stability. Taken together, the hydrophobic side of the amphipathic helix appears to possess a stabilizing function for pUL50 but is not directly involved in the interaction with pUL53. Nevertheless, it is tempting to speculate that the conserved N-terminal amphipathic helix is functionally critical for pUL50 and its herpesviral homologs. The charged cluster mutant CL02 of HSV-1 UL34 carries mutations in the predicted helix (supplemental Fig. S1; QRIRLϾQAIAL) and exhibits a defect in virus replication (43). Interestingly, the function of UL34 to support INM curvature around capsids was mapped to its N terminus adjacent to the putative ␣-helix (42). In this scenario the helices might induce local curvature of the INM by insertion of their amphipathic moieties between the polar headgroups of lipid molecules (37,44).
The Functional Importance of the Conserved N-terminal Domain of pUL50-Summarized, we conclude on the functional importance of the pUL50 N terminus (amino acids 1-181) as follows. (i) The globular domain (including the amphipathic helix) determines interaction with pUL53, (ii) the amphipathic helix is not directly involved in binding but attributes to protein stability and folding, (iii) residues Glu-56 and Tyr-57 are essential for pUL53 interaction and recruitment to the nuclear envelope. Thus, the residues essential for pUL50-pUL53 interaction are determinants of functional association of the entire NEC. This crucial contact point for multiple NEC interactions may represent a highly interesting target for future antiviral strategies.