Cytomegaloviral protein kinase pUL97 interacts with the nuclear mRNA export factor pUL69 to modulate its intranuclear localization and activity

Introduction

Human cytomegalovirus (HCMV), a betaherpesvirus, is a widely distributed human pathogen that causes the majority of infectious complications in immunocompromised individuals, including transplant and AIDS patients. HCMV is also a common cause of congenital infections, leading to neurological damage and hearing loss (reviewed by Mocarski et al., 2007).

Crucial steps in HCMV replication are regulated by various protein kinases, including the product of the HCMV UL97 gene, a nuclear serine/threonine protein kinase, conserved in all mammalian herpesviruses (Chee et al., 1989; Michel et al., 1996; Romaker et al., 2006). Previous studies demonstrated that deletion of the ORF UL97 from the viral genome or pharmacological inhibition of pUL97 kinase activity drastically reduced virus replication in cell culture (Biron et al., 2002; Herget et al., 2004; Marschall et al., 2002; Prichard et al., 1999). pUL97 phosphorylates the viral DNA polymerase processivity factor pUL44; therefore it may be important for viral DNA replication (Marschall et al., 2003). Moreover, pUL97 is involved in the nuclear egress of C capsids. The lamin-phosphorylating activity of pUL97 and its ability to alter the morphology of the nuclear envelope have suggested a role for this viral kinase in the disruption of the nuclear lamina during viral egress (Wolf et al., 2001; Marschall et al., 2005; Milbradt et al., 2009). Recently, two reports described that pUL97 plays an important role in the hyperphosphorylation of retinoblastoma protein (Hume et al., 2008; Prichard et al., 2008) and that this activity inhibits the formation of nuclear aggresomes (Prichard et al., 2008). As far as these specific functions of pUL97 are concerned, it is tempting to speculate that pUL97 is also involved in the activation or inactivation of other phosphorylation-dependent viral regulatory proteins during the viral replication cycle.

HCMV UL69 has counterparts in each mammalian or avian herpesvirus sequenced so far and encodes a multifunctional phosphoprotein (molecular mass approx. 105–116 kDa) capable of transactivating a wide range of promoters (Winkler et al., 1994; Winkler & Stamminger, 1996). We showed previously that pUL69 interacts with itself and the cellular transcription elongation factor hSPT6 via a highly structured central domain of the protein, but the interaction motifs have not been discriminated from each other (Lischka et al., 2007; Winkler et al., 2000). pUL69 has the properties of a viral mRNA export factor: it binds RNA, has nucleocytoplasmic shuttling activity and recruits the cellular mRNA export machinery via interaction with the cellular mRNA export factor UAP56/URH49 in order to promote the cytoplasmic accumulation of unspliced mRNA (Lischka et al., 2001, 2006; Toth et al., 2006). Currently, it is not known which cellular and/or viral kinases are required for phosphorylation of pUL69 and whether phosphorylation is required for any of its multiple functions. Interestingly, after treatment of HCMV-infected fibroblasts with the cyclin-dependent protein kinase (CDK) inhibitor roscovitine, the intranuclear localization of pUL69 changes from a homogeneous distribution to speckled aggregates (Sanchez & Spector, 2006).

The present study provides evidence that the speckled aggregation of pUL69 can also be observed in HCMV-infected human fibroblasts treated with pUL97-specific kinase inhibitors. This observation prompted us to investigate the possibility of a direct pUL97–pUL69 interaction and potential regulatory consequences for pUL69-mediated mRNA export. We provide evidence that pUL97 physically interacts with pUL69, is able to phosphorylate pUL69 in vitro and enhances its mRNA export activity.

Methods

Plasmid constructs.. Deletions in UL97 were generated by PCR amplification (Supplementary Table S1, available in JGV Online) using pcDNA-UL97 as template, followed by subsequent insertion into the vector pcDNA3.1 (Invitrogen). The UL97 and UL69 parental constructs used for this study were described previously (Lischka et al., 2006, 2007; Marschall et al., 2001, 2005; Schregel et al., 2007). Deletions of N-terminal pUL69 coding sequences (i.e. encoding aa 177–744, 269–744, 315–744, 380–744 and 478–744) were constructed by digesting pHM511, pHM515, pHM516, pHM517 or pHM521 (Winkler et al., 1994), respectively, with BamHI/EcoRV, followed by subsequent ligation of the UL69 sequence with pHM972 (Hofmann et al., 2002), which encodes an N-terminal in-frame fusion of a FLAG tag and the simian virus 40 (SV40) T antigen nuclear localization signal (NLS) (Lischka et al., 2006). Further N-terminal deletions of UL69 coding sequences were constructed by inserting BamHI/XhoI-digested PCR-generated fragments (Supplementary Table S1) into pcDNA3-FLAG-NLS (pHM972). C-terminal deletion mutants of the UL69 open reading frame (ORF) were fused in-frame to the FLAG-tag by ligating BamHI/EcoRI-digested PCR products with pcDNA3-FLAG (pHM971; Hofmann et al., 2000). Internal constructs, lacking the N and C terminus of pUL69, were constructed by inserting the relevant BamHI/EcoRI-digested PCR products into pHM972.

Cell culture, HCMV infections and plasmid transfections.. 293T and HeLa cells were cultivated in Dulbecco's modified Eagle's medium containing 10 % fetal calf serum (FCS). Human primary foreskin fibroblasts (HFFs) were cultivated in minimal essential medium containing 7.5 % FCS. HCMV strains AD169, AD169-GFP (Marschall et al., 2000) and BAC213 (AD169delUL97-GFP; Marschall et al., 2005) were propagated in HFFs and used for infection assays as described previously (Marschall et al., 2000). Transfections were performed by using Lipofectamine 2000 (Invitrogen), polyethylenimine (Sigma; Schregel et al., 2007) or calcium phosphate–DNA complexes (Hofmann et al., 2000; Winkler et al., 1994).

Indirect immmunofluorescence analysis.. HFF cells were grown on coverslips and used for infection with HCMV strain AD169 or the recombinant virus BAC213. Protein kinase inhibitors were added to the culture media 24 h (for roscovitine) or immediately post-infection (p.i.) (for others; see Table 1). Culture media containing inhibitors were refreshed every 24 h. Three days p.i., cells were fixed with 4 % paraformaldehyde and permeabilized with PBS/0.2 % Triton X-100, before blocking and incubation with primary antibodies under standard conditions (Marschall et al., 2003). Double staining was achieved by the use of FITC- and Cy3-conjugated secondary antibodies (Dianova); nuclear counterstaining was achieved with DAPI Vectashield mounting medium (Vector Laboratories). Immunofluorescence data were collected by using an Axiovert-135 microscope at magnifications of x400 and x630 (Zeiss).

Table 1. Induction of pUL69 nuclear speckled aggregation by pUL97- and CDK-specific kinase inhibitors as determined by immunofluorescence analyses HFFs were infected with HCMV strain AD169 (m.o.i. of 1) and analysed 72 h p.i. During HCMV replication, the following protein kinase inhibitors were used: indolocarbazoles (2 µM), Gö6976 (active against HCMV pUL97 and PKC) and Gö7874 (active only against PKC; Marschall et al., 2001, 2002); quinazoline (10 µM), Ax7396 [active against HCMV pUL97 and epidermal growth factor receptor (EGFR); Herget et al., 2004]; tyrosine kinase inhibitors (2 µM), AG490 (tyrphostin; active against Janus kinases), AG1478 and PD153035 (active against EGFR); CDK inhibitors (10 µM), Rosco (roscovitine, active against CDKs 1, 2, 7 and 9) and Flavo (flavopiridol, active against CDKs 1, 2, 4 and 9).

Coimmunoprecipitation (CoIP) assay.. 293T cells were transfected in six-well plates or 10 cm dishes. Two days post-transfection (p.t.), cells were lysed in 500–1000 µl CoIP buffer (50 mM Tris/HCl pH 8.0, 150–300 mM NaCl, 5 mM EDTA, 0.5 % NP-40, 1 mM PMSF, 2 µg ml^–1 of each of aprotinin, leupeptin and pepstatin) and used for CoIP as described previously (Schregel et al., 2007). The precipitates were subjected to standard Western blot analysis using tag-specific antibodies [mAb-FLAG M2, Sigma; pAb-haemagglutinin (HA), Covance; mAb-Myc 1-9E10.2, ATCC] for the detection of coimmunoprecipitates by enhanced chemiluminescence (ECL) staining (New England BioLabs).

In vitro kinase assay (IVKA).. After immunoprecipitation of pUL97 from transfected 293T cells, the kinase activity of pUL97 was determined in vitro as described previously (Marschall et al., 2001) using kinase buffer that contained 2.5 µCi (92.5 kBq) [γ-³³P]ATP. Recombinant pUL69, assayed as a putative substrate protein, was coexpressed and either immunoprecipitated by the use of a tag-specific antibody directed to pUL69 (RIPA buffer) or coimmunoprecipitated along with a pUL97 tag-specific antibody (CoIP buffer). CoIP was performed as described above (mAb-FLAG M2; mAb-HA, Roche; mAb-Myc 1-9E10.2; mAb-GFP clones 7.1/13.1, Roche). The immunoprecipitates were recovered by centrifugation and washed before the samples were analysed by IVKA using IVKA buffer (20 mM Tris/HCl pH 7.5, 0.5 mM MnCl₂, 1 µM ATP, 1 mM DTT). As a positive control for substrate phosphorylation, 15 µM purified histone 2B (H2B; Roche) was added to reactions.

Nuclear mRNA export assay for pUL69.. A nuclear mRNA export assay, based on the export activity of recombinantly expressed pUL69, was performed with lysates from transfected HeLa cells as described previously (Lischka et al., 2006). Chloramphenicol acetyl transferase (CAT) reporter assays were performed as described by Farjot et al. (2000) by using plasmid pDM128/CMV/RRE, which encodes CAT by an artifcial intron-containing mRNA. To quantify CAT protein expression, a CAT ELISA was performed in triplicate (Roche).

Results

Roscovitine and pUL97-specific kinase inhibitors induce a nuclear speckled aggregation of pUL69 in HCMV-infected fibroblasts
Previously, it was reported that roscovitine, an inhibitor of cellular CDKs, restricts the replication efficiency of HCMV in infected human fibroblasts. In the presence of roscovitine, the cytomegaloviral nuclear mRNA export factor pUL69 accumulates in intranuclear aggregates (Sanchez & Spector, 2006). This finding prompted us to analyse additional protein kinase inhibitors, particularly those inhibiting the viral protein kinase pUL97. Two inhibitors of pUL97 kinase activity, Ax7396 and Gö6976, belonging to two different chemical classes, quinazoline and indolocarbazole, both of which inhibit pUL97 kinase activity as well as HCMV replication in vitro (Herget et al., 2004; Marschall et al., 2001, 2002; Schleiss et al., 2008), also induced intranuclear microspeckled accumulation of pUL69 similarly to roscovitine (Fig. 1a, v, vi and viii). In the absence of inhibitor, pUL69 could be detected in a diffuse pattern throughout the nucleus and within viral replication centres (Fig. 1a, iv). Gö7874, an inhibitor related to Gö6976, with no activity against pUL97 and HCMV replication (Marschall et al., 2001, 2002), had no effect on pUL69 localization (Table 1). Tyrosine kinase inhibitors, such as AG490, AG1478 and PD153035, were also completely negative (Fig. 1a, vii and Table 1). Two well-characterized CDK inhibitors, roscovitine and flavopiridol, had markedly different effects. While roscovitine (active against CDKs 1, 2, 7 and 9) strongly induced speckled aggregation of pUL69, flavopiridol (active against CDKs 1, 2, 4 and 9) was negative (Table 1). Thus, based on the finding that pUL97 inhibitors, similar to the CDK inhibitor roscovitine, can induce a speckled-localization phenotype of pUL69, it would appear that pUL97 has an influence on pUL69 intranuclear localization. This finding was substantiated by the analysis of the intranuclear distribution of pUL69 in cells infected with BAC213, an AD169-GFP-based virus that has a deletion of the entire ORF UL97 (Marschall et al., 2005). After long-term infection experiments with BAC213 (HFFs harvested 14 days p.i.), pUL69 was detected in nuclear speckles even in the absence of roscovitine or other inhibitors (Fig. 1b, iv). The presence of inhibitors Gö7874, Gö6976, AG490 or Rosco (Fig. 1b, v–viii) did not have an obvious effect on pUL69 localization.

(53K): Fig. 1. Nuclear speckled aggregation of pUL69 in the presence of roscovitine (Rosco) and pUL97 inhibitors in strains with an intact or disrupted ORF UL97. HFFs were infected with HCMV strain AD169 (a) or mutant BAC213 (AD169delUL97-GFP; b). Inhibitors (indicated below the images) were added 24 h p.i. (12 days for BAC213). Seventy-two h p.i. (14 days for BAC213), cells were fixed and analysed for intranuclear distribution of pUL69 by indirect immunofluorescence staining (cell nuclei were counterstained with DAPI; see image i of pair i–ii and iii of pair iii–iv). The absence of pUL97 expression was confirmed in the deletion mutant BAC213 by a staining control (b, i–ii).

Interaction of the pUL97 kinase with pUL69 in transfected and HCMV-infected cells
In order to determine whether there was a direct protein interaction between pUL97 and pUL69, strain AD169 and its recombinant derivative AD169-GFP were used to infect HFFs (m.o.i. of 0.25). Western blots showed similar levels of pUL97 and pUL69 production with both viruses at 3 days p.i. (Fig. 2c). When pUL69 was immunoprecipitated from the cell lysates (Fig. 2b), a distinct pUL97 band could be detected in the coimmunoprecipitates of infected cells but not in the mock-infected control (Fig. 2a, lanes 4–6). Although the signal obtained for coimmunoprecipitated pUL97 was not strong compared with the precipitation control for pUL69 (Fig. 2b), the lack of background reactivity with pre-immune serum demonstrated that it was specific (Fig. 2a, lanes 1–3). The inverse experiment was also performed, in which polyclonal antibody to UL97 was used to immunoprecipitate (Fig. 2d–f). In this case, a higher m.o.i. was used (m.o.i. of 1) and the CoIP detection blot showed a strong band for pUL69; two different batches of the pUL97-specific polyclonal antibody produced comparable results (Fig. 2d, lanes 2 and 3). In addition, a parallel experiment using UL97-deleted HCMV strain (BAC213) for infection and subsequent CoIP did not produce any signal, thus serving as a specificity control (data not shown). These findings indicated the existence of a pUL97–pUL69 interaction complex during the late phase of infection.

(41K): Fig. 2. Interaction of pUL97 and pUL69 in HCMV-infected human fibroblasts (a–f) and plasmid-transfected 293T cells (g–i). HFFs were cultivated in culture flasks (approx. 7x106 cells) and infected with HCMV strain AD169 or AD169-GFP at an m.o.i. of 0.25 (a–c) or 1 (d–f). Three days p.i., cells were lysed and subjected to CoIP analysis using precipitation and detection antibodies as indicated. Reliable precipitation of the analysed protein (pUL69 in b, pUL97 in e) was achieved by subsequent immunostaining of the CoIP blots. (g–i) 293T cells were cultivated in 10 cm dishes (approx. 5x106 cells) and transfected with expression plasmids as indicated. Two days p.t., cells were lysed and subjected to CoIP (g) and Western blot (h) analyses. (c, f and i) Control samples (12 µl) from the total cell lysates (500 µl) were taken prior to immunoprecipitation and used for additional Western blot (Wb) analysis to monitor the levels of expressed proteins. Mock, mock-infected. Size markers are indicated (kDa).

The CoIP experiment was repeated, this time by transient coexpression of epitope-tagged versions of pUL69 and pUL97 in 293T cells, and yielded a clearly detectable signal for coimmunoprecipitated pUL69 (Fig. 2g, lane 3). This transfection experiment indicated that no other viral proteins were required for the formation of a stable pUL69–pUL97 interaction complex. As far as the specificity of the CoIP technique is concerned, we have previously demonstrated that pUL69 is not non-specifically coimmunoprecipitated with FLAG-tagged pUL84 or control antibodies (Lischka et al., 2007). In summary, a protein complex of pUL69 and pUL97 was detected in both HCMV-infected and transient DNA-transfected cells by precipitating either pUL69 or pUL97.

Identification of interaction domains in pUL69 and pUL97
A series of plasmids encoding a range of deletions in the UL69 ORF were produced, each having an N-terminal FLAG tag. The expression of all recombinant proteins was determined by Western blot analysis using lysate control samples taken prior to the addition of the CoIP antibody (Fig. 3b and d). CoIP was used to assess interaction with full-length pUL97-HA (contains a C-terminal HA tag) which determined that N-terminal truncation of pUL69 up to aa 478 did not lead to loss of interaction with pUL97-HA (Fig. 3a, lanes 3–9). Further N-terminal deletion (i.e. mutants 533–744, 547–744, 595–744 and 666–774) led to a loss of interaction (Fig. 3a, lanes 10–13), thus narrowing down the region required for pUL97 binding to aa 478–532. Interestingly, the analysis of C-terminally truncated mutants of pUL69 indicated that at least one additional interaction region for pUL97 exists and is located between aa 1 and 140, since even this smallest pUL69 fragment could be coimmunprecipitated with pUL97 (Fig. 3c). It should be noted that the variability in signal intensity observed for coimmunoprecipitated pUL69 seemed to be due mainly to the varying protein expression levels of the different C-terminally deleted pUL69 proteins (Fig. 3c and d). The existence of further pUL97 binding domains located within the central domain of pUL69 cannot be excluded.

(39K): Fig. 3. Deletion mapping of UL69. (a, c) 293T cells were transfected with expression plasmids as indicated. Cells were lysed 2 days p.t. and subjected to CoIP analysis using the indicated precipitation and detection antibodies. (b, d) Control samples of the cell lysates were used for an additional Western blot (Wb) analysis to monitor the levels of expressed proteins. (e) A schematic summary of CoIP data obtained with N-terminal and C-terminal deletion mutants of pUL69. Size markers are indicated (kDa).

In a second series of CoIP experiments, a series of plasmids encoding deletions of the UL97 ORF expressed with a C-terminal FLAG tag were analysed for interaction with full-length pMyc-UL69 (N-terminal Myc tag). C-terminal truncations of pUL97 were tolerated down to aa 336 (Fig. 4a and c, lanes 5–8 and 4–5, respectively). N-terminal truncation up to aa 231 yielded positive results (Fig. 4a and c, lanes 10-11 and 7, respectively), whereas further loss of sequence produced negative results, e.g. 337–707 (Fig. 4a, lane 9). Fragments with truncations at both termini, e.g. 111–365 and 181–365, were still positive (Fig. 4a and c, lanes 12 and 6, respectively), confirming the presence of an interaction region in the N-terminal half of pUL97. Importantly, an internal deletion (230–280) lost the interaction activity (Fig. 4a, lane 13). This suggests that the pUL69-interaction region was located between aa 231 and 336 of pUL97. In fact, a construct expressing aa 231–336 was sufficient for coprecipitating pUL69, consistent with the mapped interaction region (Fig. 4e, lane 4). The use of another FLAG-tagged HCMV protein (pUL26) did not coimmunoprecipitate pUL69; this supported the specificity of the analysis (Fig. 4e, lane 2). It was apparent that the intensity of the interaction of some mutant forms varied, although the expression levels remained within a comparable range (Fig. 4b and d), which suggests that the overall conformation of the polypeptide has an impact on the functionality of the pUL69 interaction domain.

(26K): Fig. 4. Deletion mapping of UL97. (a, c and e) 293T cells were transfected with the expression plasmids as indicated. Cells were lysed 2 days p.t. and subjected to CoIP analysis using the indicated precipitation and detection antibodies. (b, d and f) Control samples of the cell lysates were used for additional Western blot (Wb) analysis to monitor the levels of expressed proteins. (g) A schematic summary of CoIP data obtained with N-terminal and C-terminal deletion mutants of pUL97.

pUL69 is phosphorylated by pUL97 in vitro
We next addressed whether pUL69 serves as a phosphorylation substrate of pUL97. To achieve this, two functionally active versions of pUL97 were overexpressed in transfected 293T cells: pUL97-FLAG (full-length) and N-terminally truncated-pUL97(181-707)-F with a C-terminal Flag tag (Marschall et al., 2005; Schregel et al., 2007). As controls, two catalytically inactive versions of pUL97 were also expressed: pUL97(K355M)-F, which has a mutation in an essential lysine in the ATP binding site (Marschall et al., 2001; Romaker et al., 2006), and pUL97(1-595)-F, with a C-terminal truncation resulting in a loss of activity (Marschall et al., 2005). When these constructs were expressed individually and immunoprecipitated for analysis in IVKA, clear autophosphorylation bands were detectable for the active versions only (Fig. 5a, lanes 4 and 6). These autophosphorylation bands were the same as the bands in the expression control (Fig. 5b, upper panel, lanes 4–7). In a control reaction, the addition of a purified standard substrate protein (H2B) demonstrated full activity of pUL97 for substrate phosphorylation (Fig. 5a, lane 13). Upon coexpression of pUL97-FLAG (carrying a C-terminal FLAG tag) and pMyc-UL69 (carrying an N-terminal Myc tag), followed by immunoprecipitation with a mixture of two tag-specific antibodies, additional phosphorylation bands were detectable in the IVKA. Strong phosphorylation of pUL69 was seen in the presence of active pUL97 (Fig. 5a, lanes 8 and 10; expression control in Fig. 5b, lower panel). However, the catalytically inactive versions did not produce any background phosphorylation signal, indicating that the assay was specific. Another specificity control was provided by coexpression of pUL26, used as a known HCMV protein that is not phosphorylated by pUL97 (Marschall et al., 2005; Fig. 5a and b, lanes 3 and 12). These data demonstrate for the first time that pUL69 is directly phosphorylated in vitro.

(27K): Fig. 5. Phosphorylation of pUL69 by pUL97 in vitro. (a) 293T cells were transfected with expression plasmids as indicated. Cells were lysed 2 days p.t. and coimmunoprecipitated with mAb-Myc plus mAb-FLAG. The precipitates were used for an IVKA, and labelled phosphorylation products were separated by SDS-PAGE and visualized by Western blotting followed by exposure to autoradiography films. phos., Phosphorylated; autophos., autophosphorylated. (b) Control samples of the cell lysates were used for an additional Western blot (Wb) analysis to monitor the levels of expressed proteins.

The interaction between pUL69 and pUL97 was analysed further in a combined CoIP-IVKA, in which only the fraction of specifically coimmunoprecipitated pUL69 was present in the subsequent IVKA (Fig. 6). Phosphorylation of pUL69 by pUL97 was again detectable (Fig. 6a, lane 5) although, as expected, the signal intensity was somewhat lower compared with Fig. 5a (lane 8). The catalytically inactive point mutant of pUL97, K355M, did not produce any signals (Fig. 6a, lane 6). For intact pUL97, the phosphorylation of control substrate H2B was also positive (Fig. 6a, lower panel, lane 7). Thus, pUL69 serves as a substrate for phosphorylation through its interaction with pUL97.

(42K): Fig. 6. Combined CoIP and IVKA demonstrating pUL97-mediated phosphorylation of the interaction competent pUL69, but not of non-interacting pUL69 fragments or another viral protein, pUL53. (a, c) 293T cells were transfected with the expression plasmids indicated. Cells were lysed 2 days p.t. and subjected to CoIP using the indicated precipitation antibodies directed to pUL97. The coimmunoprecipitates were used for IVKA, and labelled phosphorylation products were separated by SDS-PAGE and visualized by Western blotting followed by exposure to autoradiography films. (d) A control Western blot (Wb) of the coimmunoprecipitated proteins was subsequently performed using the CoIP–IVKA blot. (b, e) Control samples of the cell lysates were used for additional Western blot (Wb) analysis to monitor the levels of expressed proteins. (f) To visualize all phosphorylated bands of pUL69 and pUL97, the N-terminally truncated, catalytically active version of pUL97 (lanes 3 and 4) was analysed in parallel to full-length pUL97 (lanes 1 and 2) in the presence of pUL69. The CoIP–IVKA conditions were identical to those described for (a) and (c). phos., Phosphorylated; autophos., autophosphorylated.

The interpretation of the results was tested by using mutants of pUL69 that were capable or incapable of interacting with pUL97 (see Fig. 3). To this end, full-length pUL69 as well as truncated forms 1–140 and 269–574 were chosen as putative interaction-positive versions, and mutant 595–744 was chosen as an interaction-negative version. Phosphorylation signals were detected from full-length pUL69 and mutant 1–140 (Fig. 6c, lanes 2–3). However, the pUL69-specific band in the expression control (Fig. 6e, lower panel, lane 2) was almost overlapping with the strong upper band of the typical pUL97 double-band (Fig. 6e, upper panel, lane 2; see also Fig. 4e, f and Schregel et al., 2007). To achieve a clearer distinction beween phosphorylated forms of pUL69 and pUL97, the N-terminally truncated, catalytically active version of pUL97 (181–707, containing the pUL69 interaction region) was analysed and compared to full-length pUL97 for its ability to interact with and phosphorylate pUL69 (Fig. 6f). Note the appearance of three phosphorylated bands of pUL69 in lanes 2 and 4. Interestingly, the interaction-positive mutant 1–140 was positive for phosphorylation by pUL97, while mutant 269–574 was not (Fig. 6c, lanes 3–4). The latter finding was not expected, due to the fact that mutant 269–574 contains at least part of one of the two interaction regions. Actually, strong interaction of this mutant with pUL97 was demonstrated by a control staining of the CoIP-IVKA blot using mAb-FLAG (Fig. 6d). Despite the strong interaction, the lack of phosphorylation of mutant 269–574 by pUL97 might be explained by the absence of true phosphorylation sites, i.e. serine/threonine residues recognized as substrate sites by pUL97, in this part of pUL69. Furthermore, the interaction-negative mutant 595–744 did not show a phosphorylation signal (Fig. 6c, lane 5) and another interaction-negative HCMV protein, pUL53, was not phosphorylated (Fig. 6c, lane 6). Thus, phosphorylation of pUL69 by pUL97 was found to be restricted to a subset of those versions of pUL69 detectable by CoIP. Hence, the N terminus of pUL69 (1–140) was identified as a target for phosphorylation.

Nuclear mRNA export activity of pUL69 is modulated by pUL97
Next we investigated whether the interaction with pUL97 influences the mRNA export activity of pUL69 by using our previously described mRNA export assay (Lischka et al., 2006). In cotransfection experiments, the nuclear export of an intron-containing mRNA encoding CAT by pUL69 was determined in the absence or presence of pUL97. In the absence of pUL97, pUL69 produced an intermediate level of CAT expression (Fig. 7a). In the presence of pUL97, pUL69 activity was apparently stimulated, leading to a statistically significant increase of CAT expression (Fig. 7a). However, when a catalytically inactive mutant of pUL97 was used, no increase was observed (Fig. 7a). Instead, a slight decrease was noted; this might be caused by the interaction of pUL69 with inactive pUL97(K355M), which possibly prevents the interaction with other cellular pUL69-phosphorylating kinases (e.g. CDKs). This finding was further illustrated by treatment with protein kinase inhibitors, one inhibitory for pUL97 (Gö6976) and one inhibitory for CDKs (roscovitine) (Fig. 7b). While roscovitine was most effective and almost totally inhibited the export activity of pUL69 in the absence and presence of pUL97 (Fig. 7b), Gö6976 inhibited part of the pUL69 export activity to 76 % of the level without inhibitor and, importantly, prevented the pUL97-dependent increase of activity levels (Fig. 7b). Thus, pUL97 is able to modulate the nuclear mRNA export function of pUL69. It is tempting to speculate that pUL97-mediated phosphorylation of pUL69 is the regulatory trigger of this event.

(31K): Fig. 7. Nuclear RNA export assay for pUL69 in the absence or presence of pUL97 activity. (a) HeLa cells were cultivated in six-well plates (approx. 4x105 cells) and used for transfection with the indicated plasmid constructs (2 µg pUL69 expression plasmid, 0.5 µg pUL97 expression plasmid and 0.15 µg CAT reporter construct). (b) Inhibitors were added to the culture media 1 day p.t. at the following concentrations: 10 µM Rosco (black bars), 2 µM Gö6976 (grey bars) and no inhibitor (hatched bars). Cells were harvested by lysis 2 days p.t. and subjected to the nuclear RNA export assay for pUL69. Transfections were performed in triplicate, and relative values (mean±SD) of CAT activity compared with pUL69 alone (without inhibitor) are given. Statistical significance was calculated by Student's t-test (P<0.0001). Vector indicates cotransfection of the CAT reporter construct with empty vector (i.e. pcDNA3.1) alone.

Discussion

The main findings of this study are that (i) a speckled nuclear aggregation of pUL69 is induced not only by CDK inhibitors but also by pUL97 inhibitors, (ii) pUL97 physically interacts with pUL69, as detectable by CoIP from HCMV-infected primary fibroblasts as well as transiently transfected cells, (iii) pUL69 is a phosphorylation substrate of pUL97 and (iv) the pUL97-mediated phosphorylation of pUL69 has a direct impact on the nuclear mRNA export activity of pUL69. These findings suggest a novel functional role of pUL97 and a phosphorylation-dependent fine-regulation of pUL69 activities.

With respect to the formation of pUL69 aggregates in the presence of pUL97 inhibitors, a recent paper reported that pUL97 affected the formation of large nuclear protein aggregates, termed aggresomes (Prichard et al., 2008). However, the nuclear aggregates of pUL69 observed in this study were morphologically clearly distinct from large aggresomes, since we observed a fine microspeckled distribution pattern. Furthermore, no colocalization of these speckles with promyelocytic leukaemia protein bodies, which may be involved in nuclear aggresome formation, was detectable (S. Rechter and M. Marschall, unpublished results).

Our mapping of interaction-relevant regions revealed interaction motifs at aa 1–140 and 478–532 of pUL69 and aa 231–336 of pUL97. Interestingly, the interaction domain of pUL97, which lies outside of the kinase domain (337–651; Romaker et al., 2006), overlaps with other known protein binding regions, such as those for cellular interactor p32 (181–365; Marschall et al., 2005) and pUL97 self-interaction (231–280; Schregel et al., 2007). However, it can be clearly discriminated from the interaction domain for viral polymerase processivity factor pUL44 (366–459; Marschall et al., 2003). All of these interactors are phosphorylated via interaction with pUL97. This suggests that pUL97 contains at least two distinct substrate binding platforms, one in the region N-terminal of the kinase domain (for phosphorylation of pUL69, p32 and autophosphorylation of pUL97) and a second one located within the kinase domain (for phosphorylation of pUL44). Such an organization of domains is a known feature of protein kinases which seems to confer the ability to bind and phosphorylate a considerable number of different substrates (Goldsmith et al., 2007).

As far as the interaction mapping of pUL69 is concerned, two regions required for binding of pUL97 were identified, one at the N-terminal end (aa 1–140 of pUL69) and one in the C-terminal part (478-532); however, an additional interaction within the central region of pUL69 cannot currently be excluded. In previous studies, we demonstrated that the N terminus of pUL69 consists of multiple overlapping functionally important regions, such as the nuclear localization signal, the UAP56/URH49 interaction motif and the RNA binding site (Lischka et al., 2006; Toth et al., 2006). The data derived from IVKAs in this study provide evidence that the N terminus is phosphorylated by pUL97. It is therefore feasible to speculate that phosphorylation within this domain may trigger a fine-regulation of distinct functional properties of pUL69. Interestingly, in silico analysis of putative phosphorylation sites within the N-terminal 140 aa using the NetPhos 2.0 Server () predicted the existence of 13 serines and three threonines with high scores for potential phosphorylation sites. Three of these residues (S132, S133 and S134) are located within aa 76–140, which was also phosphorylated in our IVKAs (data not shown). Interestingly, these residues are all located within the RS-domain of pUL69 (aa 123–139) which is required for RNA binding (Toth et al., 2006).

In addition to the N-terminal region, we found that a second, C-terminal domain of pUL69 comprising aa 478–532 was also able to interact with pUL97. Importantly, the C terminus of pUL69 has previously been shown to contain the nuclear export signal that is required for nucleo-cytoplasmic shuttling and pUL69-mediated cytoplasmic accumulation of unspliced mRNA (Lischka et al., 2001, 2006). With respect to these findings, it is important to note that our mapping studies would correlate with the presence of a possible bipartite or tripartite interaction surface for pUL97 on pUL69. Such a complex interaction surface might be formed through a defined folding that gives rise to a tertiary structure-based interaction domain. Unfortunately, no crystal structure data for pUL69 are available so far.

A yeast two-hybrid study investigating potential interaction between pUL69 and pUL97 did not detect an interaction. However, negative results in the yeast two-hybrid assay may arise from the lack of post-translational modifications, as well as unfavourable, heterologous protein folding. Furthermore, under physiological conditions, pUL69–pUL97 interaction may occur in HCMV-infected fibroblasts in a manner limited by time and variable affinities that are not reproducible in a yeast cell overexpression system. This notion is also reflected by similar previously reported discrepancies, i.e. the identification of several pUL97 interaction partners such as pUL44, pUL97, pp65 and cellular p32 (Marschall et al., 2003, 2005; Krosky et al., 2003; Schregel et al., 2007; Kamil & Coen, 2007; Milbradt et al., 2007). In these studies that used different experimental approaches with variable sensitivities, some, but not all, interactions could be consistently reproduced. In our hands, yeast two-hybrid assays and other overexpression systems were applied as screening methods which essentially required the confirmation by a CoIP experiment with proteins from HCMV-infected fibroblasts (see the earlier studies involving pUL97–pUL44, pUL97–p32, pUL97–pUL69 and other interactions).

Based on the finding that pUL69 is phosphorylated by pUL97 in vitro, we investigated putative pUL97-mediated effects on the nuclear RNA export activity of pUL69. Data from an established reporter-based assay which determines the nuclear export of an unspliced, intron-containing mRNA (Lischka et al., 2006) strongly suggested that phosphorylation of pUL69 enhances the mRNA export activity of pUL69. Protein kinase inhibitors, such as Gö6976 (pUL97) and roscovitine (CDKs), reduced this activity significantly. In addition, the importance of pUL69 phosphorylation might also be illustrated by other functions of this pleiotropic protein: pUL69 is able to induce a cell cycle arrest in the G1 phase (Lu & Shenk, 1999). Thus, pUL69 might gain access to the cell cycle regulation machinery via interaction with the viral kinase pUL97. This hypothesis is strengthened by a recent report characterizing pUL97 as a functional orthologue of CDKs (Hume et al., 2008). The inter-regulation between pUL97 and CDKs with respect to the function of pUL69 is currently under investigation (S. Rechter and others, manuscript in preparation). Moreover, pUL69 might recruit pUL97 to the transcription elongation complex via its interaction with transcription elongation factor hSPT6. Through this recruitment, pUL97 might contribute to the phosphorylation of the C-terminal domain of RNA polymerase II, as reported by Winkler et al. (2000) and Baek et al. (2004). In conclusion, this study demonstrates that pUL69 interacts with pUL97 as a specific phosphorylation substrate and that the phosphorylation of pUL69 is a determinant of its functionality.

Acknowledgements

The authors wish to thank Peter Lischka (AiCuris GmbH & Co. KG, Wuppertal), Barbara Zielke and Katrin Zielke (Virological Institute, University Erlangen-Nuremberg) for stimulating discussions and contributions to this study. We appreciate the supply of protein kinase inhibitors by GPC Biotech AG (Martinsried, Germany). The study was supported by the Deutsche Forschungsgemeinschaft (MA 1289/4-1 and SFB473), Studienstiftung des Deutschen Volkes (scholarship S. R.), the IZKF Erlangen and the Wilhelm Sander Stiftung.