Cyclin-dependent Kinases Phosphorylate the Cytomegalovirus RNA Export Protein pUL69 and Modulate Its Nuclear Localization and Activity*

Replication of human cytomegalovirus (HCMV) is subject to regulation by cellular protein kinases. Recently, we and others reported that inhibition of cyclin-dependent protein kinases (CDKs) or the viral CDK ortholog pUL97 can induce intranuclear speckled aggregation of the viral mRNA export factor, pUL69. Here we provide the first evidence for a direct regulatory role of CDKs on pUL69 functionality. Although replication of all HCMV strains was dependent on CDK activity, we found strain-specific differences in the amount of CDK inhibitor-induced pUL69 aggregate formation. In all cases analyzed, the inhibitor-induced pUL69 aggregates were clearly localized within viral replication centers but not subnuclear splicing, pore complex, or aggresome structures. The CDK9 and cyclin T1 proteins colocalized with these pUL69 aggregates, whereas other CDKs behaved differently. Phosphorylation analyses in vivo and in vitro demonstrated pUL69 was strongly phosphorylated in HCMV-infected fibroblasts and that CDKs represent a novel class of pUL69-phosphorylating kinases. Moreover, the analysis of CDK inhibitors in a pUL69-dependent nuclear mRNA export assay provided evidence for functional impairment of pUL69 under suppression of CDK activity. Thus, our data underline the crucial importance of CDKs for HCMV replication, and indicate a direct impact of CDK9-cyclin T1 on the nuclear localization and activity of the viral regulator pUL69.

Human cytomegalovirus (HCMV) 2 is a member of the Herpesviridae family and a human pathogen with worldwide distribution. Primary HCMV infection of the immunocompetent host is usually asymptomatic, whereas severe disease can occur upon infection of the immunocompromised and immunonaive. HCMV is a leading cause of complications in transplant recipients and AIDS patients, and congenital infection may result in mental impairment and hearing loss (1).
HCMV replication is differentially regulated in host cell types, and viral replication is dependent on regulation of the cell cycle (2). HCMV infection induces cell cycle arrest, while simultaneously the virus sustains an active cellular metabolic state supporting productive infection (3). Infected cells arrest in a pseudo-G 1 state with high levels of cyclin E and cyclin E-associated kinase activity (4 -6). A number of additional alterations of cyclin-dependent protein kinase (CDK) activity have also been described, such as increased synthesis and reduced degradation of cyclin B1, as well as cytoplasmic translocation of CDK1 in HCMV-infected cells (7). The up-regulation of CDK activity during HCMV replication implies that viral replication requires CDK activity to create an environment favorable for efficient viral transcription, genome replication, and assembly of viral particles. Several regulatory steps in HCMV replication are dependent on CDK activity, particularly those involving CDK1, -2, -7, and -9 (8 -12). Additionally, inhibition of CDK activity affects replication of HCMV and other herpesviruses (13). Roscovitine, a purine analog that preferentially inhibits CDK1, -2, -5, -7 and -9, has been shown to decrease viral DNA synthesis and production of late viral protein and infectious virus (8,9,12,14). Roscovitine is therefore a useful tool to investigate the impact of CDK activity on viral replication and to understand inter-regulation between CDKs and viral proteins. Cross-talk between CDKs and other protein kinases during HCMV replication is one issue of current interest (15).
CDKs, particular serine/threonine kinases that become activated upon binding to cyclins, are involved in the regulation of multiple cellular processes. They can be subdivided into two major functional groups, cell cycle-associated CDKs and transcriptionally regulating CDKs. A prototype of the transcriptionally regulating CDKs is the positive transcription elongation factor b (P-TEFb), which is composed of CDK9 and cyclin T1 (cycT1). This complex is an important regulator of transcription through phosphorylation of the C-terminal domain of the large subunit of RNA polymerase II, thus allowing for transcription elongation (16). The expression of many genes is regulated at the level of transcription elongation, and the activity of the P-TEFb complex is tightly controlled. For example, the association of 7SK small nuclear RNA and HEXIM1 acts as an inhibitor to P-TEFb (17)(18)(19)(20), whereas autophosphorylation of phospho-acceptor sites at the CDK9 C terminus acts to stimulate and promote nuclear translocation of the P-TEFb complex (21).
Among the viral proteins identified as substrates of pUL97, the pluripotent regulator pUL69 appears functionally relevant. pUL69 acts as a transcriptional activator (34,35), a nuclear mRNA export factor (36), and a mediator of cell cycle arrest (37,38). Recent studies show pUL69 binds RNA, has nucleocytoplasmic shuttling activity, and recruits the cellular mRNA export machinery via interaction with the cellular mRNA export factor UAP56/URH49. This latter activity promotes cytoplasmic accumulation of unspliced mRNA (36,39,40). pUL69 is a phosphoprotein subject to phosphorylation by the pUL97 viral kinase (41), although it is still unclear whether CDKs also play an important role in its phosphorylation. Against this background, it is significant that the CDK inhibitor roscovitine influences the intranuclear localization of pUL69 in HCMV-infected fibroblasts by changing pUL69 homogeneous nuclear distribution toward speckled aggregation (9).
In this study, we provide evidence for direct targeting of pUL69 by CDKs, which modulates pUL69 nuclear localization and activity. Findings in support of this concept are the speckled nuclear aggregation of pUL69 induced by CDK inhibitors, colocalization of CDKs and pUL69 in HCMV-infected cells, and the direct in vitro phosphorylation of pUL69 by CDK-cyclin complexes.
qPCR-Quantitative real time PCR (qPCR) was performed (49) using the ABI Prism 7700 sequence detector (Applied Biosystems, Foster City, CA) and corresponding software SDS (sequence detection system version 1.9). qPCR was used to quantify viral genome copies normalized against cellular genome equivalents in initial viral stocks, and supernatant was taken at various time points post-infection. Total DNA was extracted from virus-infected cells using the Wizard DNA purification kit (Promega, Mannheim, Germany). qPCR was performed in a 25-l reaction mixture containing 5 l of either the sample or the standard DNA solution. Additional components of the reaction mixture were 12.5 l of 2ϫ TaqMan PCR Mastermix (Applied Biosystems), 7.5 pmol of each primer complementary to a region within exon 4 of the IE1 gene locus (5Ј-AAGCGGCCTCTGATAACCAAG-3Ј and 5Ј-GAGCA-GACTCTCAGAGGATCG-3Ј), and 5 pmol of probe directed against the HCMV MIE region exon 4 (5Ј-CATGCAGATCTC-CTCAATGCGCGC-3Ј). To calculate the ratio of viral DNA per cellular DNA equivalent, the cellular DNA was quantified in parallel using primers complementary to a region within the albumin gene locus (5Ј-GTGAACAGGCGACCATGCT-3Ј and 5Ј-GCATGGAAGGTGAATGTTTCAG-3Ј) together with an albumin gene-specific probe (5Ј-TCAGTGGAAGAT-GAAACATACGTTC-3Ј). Both probes used were labeled with 6-carboxyfluorescein reporter dye and 6-carboxytetramethylrhodamine quencher dye. The DNA standard for quantification of viral genome copies was prepared by serial dilutions of plasmid pHM123 containing the IE1 cDNA (50). The cellular standard was composed of albumin PCR products obtained from lymphocyte cell extracts (51). The thermal cycling conditions consisted of two initial steps of 2 min at 50°C and 10 min at 95°C followed by 40 amplification cycles (15 s 95°C, 1 min 60°C). DNA extracts were analyzed in triplicate for each sample.
Yeast Two-hybrid Analysis-Protein interactions were analyzed using GAL4 fusion proteins (GAL4-BD, DNA binding domain; GAL4-AD, activation domain) in the yeast two-hybrid system as described previously (24). An expression plasmid for cyclin T1 was provided by L. Lania (52). Expression plasmids for CDK9 and cycT1::CDK9 were subcloned from pACTII-T1 and pRc/CMV-PITALRE-HA (52,53). Saccharomyces cerevisiae strain Y153 was used for interactor analysis, and the selection of clones was achieved by cultivation on media restricting growth to combined tryptophan/leucine prototrophy. Selected colonies were analyzed for ␤-galactosidase activity by filter lift tests.
In Vivo Labeling Assay-Proteins were labeled in vivo in HCMV-infected HFFs by incubation with [␥-33 P]orthophosphate (150 Ci/4.5 ϫ 10 5 cells) in the culture media. Cells were lysed in RIPA buffer and supernatants subject to immunoprecipitation (pAb-UL69). Samples were analyzed by SDS-PAGE/ Western blot as described for the in vitro kinase assay.
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 (39). CAT reporter assays were performed as described by Farjot et al. (54). The plasmid construct pDM128/CMV/RRE, encoding an intron-containing mRNA with the coding sequence of CAT, was used as a reporter of nuclear export activity. CDK inhibitors were added 16 h post-transfection. CAT protein expression was quantified following cell lysis at 48 h post-transfection and analyzed in triplicate using a CAT enzyme-linked immunosorbent assay (Roche Applied Science).

HCMV Replication Is Regulated by CDK Activity, CDK Inhibitors Induce a Pronounced Aggregate Formation of pUL69 That
Is Quantitatively Different for Variants of HCMV-As described previously, the intranuclear localization of the viral regulator pUL69 is significantly altered in the presence of the CDK inhibitor roscovitine, with the formation of speckled nuclear aggregates induced in the late phase of replication (9). This phenomenon was detectable for several strains of HCMV such as Towne (9), AD169 (41) (Fig. 1A), and clinical isolates. 3 For strain AD169, roscovitine treatment produced pUL69 aggregation in 67.3% of virus-positive cells (Fig. 1C). This phenotype varied in quantitative terms between AD169-derived virus variants GDGrXbaF4, GDGrP53, and 759rD100. As depicted in Fig. 1B, GDGrXbaF4 and 759rD100 contain a GCV resistance-conferring mutation in open reading frame UL97 (deletion 590 -593) (44). GDGrP53 and 759rD100 contain a point mutation in open reading frame UL54 (A987G) resulting in cidofovir resistance (43,45) and high level GCV resistance (Fig. 1D, GCV). When analyzing the nuclear localization of pUL69, variants GDGrXbaF4 and GDGrP53 were similar to AD169 with pUL69 aggregate formation in the presence of roscovitine (Fig. 1A, panels a-d and f-i). However in quantitative terms, the pUL69 aggregation of GDGrP53 was significantly reduced (38% of virus-positive cells, p Ͻ 0.01; Fig. 1C) compared with the parental strain AD169. As shown for AD169 and the variants GDGrXbaF4 and GDGrP53, an additional CDK inhibitor, R58, also induced a pUL69 phenotype similar to that induced by roscovitine (Fig. 1A, panels k-n). R58 is a strong inhibitor of CDK2, CDK5 (IC 50 Ͻ 1 M in vitro), and possibly other CDKs. The double mutant 759rD100 behaved differently, showing very little speckled aggregation of pUL69 in infected cells treated with roscovitine or R58 (Fig. 1A, panels e, j, and o). Quantitative immunofluorescence analysis demonstrated only 15% of 759rD100-infected fibroblasts showed pUL69 aggregates under roscovitine treatment, which represented a highly significant reduction (p Ͻ 0.0001; Fig. 1C). Similar results were obtained with inhibitor R58. The UL97/UL54 mutant 759rD100 showed reduced replicative sensitivity toward roscovitine, demonstrated on plaque reduction assay (Fig. 1D, Rosco), but showed a strong sensitivity toward the second CDK inhibitor R58. This indicated that R58, although inefficient in inducing pUL69 aggregates in 759rD100-infected cells, mediated an inhibitory effect on viral replication, albeit through a mode of action that seemed independent from pUL69.
The question whether observed differences between HCMV variants were due to major alterations in viral replication characteristics was addressed by qPCR. HFFs were infected with DNA-normalized viral stocks (i.e. viral DNA copy numbers per cell eq), and the kinetics of viral genomic DNA synthesis were determined as depicted in supplemental Fig. S1. The genomic replication curves of AD169 and GDGrXbaF4 were very similar, whereas 759rD100 and GDGrP53 showed a clear replication deficit throughout the period analyzed. This most likely can be attributed to the pUL54 DNA polymerase mutation of 759rD100 and GDGrP53. However, the qPCR data cannot fully explain the roscovitineinduced pUL69 aggregation phenotypes of the virus variants. Whereas GDGrP53 showed the lowest replication efficiency, 759rD100, but not GDGrP53, showed a clear lack of pUL69 aggregate formation. Thus, the phenotype peculiarity of variant 759rD100, with its double mutation in pUL54 DNA polymerase and pUL97 protein kinase, needs to be further analyzed on a mechanistic basis.
In this context, it was interesting to observe that a known pUL97 kinase inhibitor, Gö6976, but not an unrelated tyrosine kinase inhibitor, AG490, produced a pattern of speckled pUL69 aggregate formation very similar to roscovitine or R58. Parental AD169 and variants GDGrXbaF4 and GDGrP53 showed a pronounced Gö6976-induced pUL69 aggregation (supplemental Fig. S2, t, v, and x), although very little aggregate formation was observed for 759rD100 in response to Gö6976 treatment (supplemental Fig. S2, z; statistically significant, p Ͻ 0.01). Of note, pUL97 was never observed in colocalization with pUL69 aggregates but remained in a nonspeckled, homogeneous nuclear distribution. Additionally, the presence of Gö6976 led to some exclusion of pUL97 from viral replication centers (supplemental Fig. S2, c and  u), an effect that had been described before (55). Thus, inhibition of pUL97 as well as CDKs can induce the pUL69 aggregation phenotype in several variants of HCMV.
To narrow down the number of CDKs associated with the formation of pUL69 aggregates, a series of novel inhibitors with strong inhibitory potential against CDKs in vitro were utilized. These inhibitors fell into two groups with respect to their ability to produce pUL69 speckled aggregates: five compounds (A14, A43, A79, R25, and R58) induced strong effects comparable with roscovitine, whereas two other compounds (A50 and A98) failed to alter pUL69 distribution (supplemental Table S1). These CDK inhibitors all share a strong inhibitory potential against CDK1 and CKD2 in vitro (IC 50 Ͻ 1 M) and possibly further inhibitory effects against other CDKs. Thus, the pattern of CDK inhibition required for pUL69 aggregation could not be deduced from this experiment. However, it is highly suggestive that CDK1/2 inhibition is not sufficient to confer the phenotype of pUL69 aggregation, and additional inhibitory activity is required.
CDK Inhibitor-induced Intranuclear Aggregates of pUL69 Are Localized within Viral Replication Centers-To characterize the speckled aggregation of pUL69 more closely, HCMVinfected fibroblasts were analyzed under roscovitine treatment by costaining of pUL69 with a selection of viral and cellular nuclear proteins (Fig. 2). Viral DNA polymerase pUL54 and its processivity factor pUL44 are prominent markers of viral replication centers, and pUL69 is typically also detectable within these compartments. Following infection of HFFs with HCMV AD169 in the presence of roscovitine, pUL44 (Fig. 2, c and g) and pUL54 (data not shown) did not alter their localization, but markedly, the speckled aggregates of pUL69 fully localized within the area of replication centers (Fig. 2, e-h). This suggests an accumulation of the replication center-associated fraction of pUL69 into subnuclear speckles under conditions of inhibited CDK activity. Additionally, further types of prominent intranuclear structures were examined, such as splicing compartments (marked by splicing factor SC-35), the nuclear pore complex (marked by NUP62/152/90), or aggresome structures (marked by heat shock cognate protein 70, HSC70 (56)). As illustrated in Fig. 2, neither SC-35 (k and o), NUP62/152/90 (s and w), nor HSC70 (data not shown) displayed alterations in their localization or colocalized with pUL69.
CDK Inhibitor-induced Aggregates of pUL69 Colocalize with CDK9 and Cyclin T1-The investigation of CDK distribution patterns in HCMV-infected HFFs under inhibitor treatment showed a specific association of CDKs with pUL69 speckled aggregates ( Fig. 3A and supplemental Fig. S3). CDK1, -2, -7, and -9 were analyzed under roscovitine, R58, or Gö6976 treatment. AG490 and Gö7874, which affect neither pUL97 nor CDK activity, served as specificity controls. Strikingly, CDK9 was found to undergo changes in localization similar to pUL69. CDK9 accumulated in replication centers of HCMV-infected cells as shown by colocalization with pUL69 in the absence of inhibitor (Fig. 3A, panels e-h). This colocalization was further developed in the presence of roscovitine or R58, i.e. CDK9 also aggregated in the form of nuclear speckles (Fig. 3A, panels i-p). CDK9 aggregation was only marginally detectable in the presence of the pUL97 inhibitor Gö6976 (Fig. 3A, panels q-t). Both controls, Gö7874 (Fig. 3A, panels u-x) and AG490 (data not shown), had no impact on the localization of pUL69 and CDK9. In a next step, the regulatory subunit of CDK9, cyclin T1 was analyzed. As depicted in Fig. 3B, cyclin T1 was not only recruited to HCMV replication centers (panels e-h) but also formed speckled aggregates in colocalization with pUL69 under treatment with roscovitine (panels i-l) or R58 but not Gö6976 (data not shown). These findings strongly suggest that although CDK inhibitors as well as pUL97 inhibitors induce a very similar speckled aggregation of pUL69, the composition of the structures and the underlying mechanisms are different. Thus, roscovitine-mediated aggregation is likely to be associated with CDK9-cycT1 activity, although Gö6976-mediated aggregation appears to be CDK-independent.
Interestingly, closer investigation of protein-protein interactions by yeast two-hybrid analysis revealed direct interactions between pUL69 and cyclin T1 as well as a cycT::CDK9 fusion construct, but not with CDK9 alone (Fig. 3C). Signal intensity of the positive scores of the filter lift staining remained at a mod- erate level compared with the positive control (CDK9 and cyclin T1) indicating a dynamic mode of low affinity interaction between pUL69 and cyclin T1.
A putative colocalization between pUL69 and other CDKs was further analyzed. CDK7, functionally related to CDK9, accumulated in replication centers similar to CDK9 and cyclin T1 in HCMV-infected cells (supplemental Fig. S3, e-h). However, neither CDK nor pUL97 inhibitors induced speckled aggregation of CDK7 (supplemental Fig. S3, i-t). For CDK2, HCMV infection did not lead to an incorporation into viral replication centers or any other detectable changes in intranuclear localization. Also, the addition of inhibitors had no impact on the pattern of nuclear localization of CDK2 (data not shown). On the other hand, a previously described translocation of CDK1 from the nucleus to the cytoplasm was detected in HCMV-infected cells (7). Hence, colocalization between CDK1 and pUL69 was not detected in the presence or absence of inhibitors.
Phosphorylation of pUL69 in Vitro and in Vivo-The strong inter-regulation of CDKs with the nuclear localization of pUL69 initiated investigation of CDK-mediated phosphorylation of pUL69. To this end, in vitro kinase assays were performed to analyze the ability of recombinant CDK-cyclin complexes (CDK1-cycB1, CDK2-cycE, CDK7-cycH-MAT1, and CDK9-cycT) to phosphorylate pUL69 immunoprecipitated from transfected 293T cells. The activity of the CDK-cyclin complexes were confirmed via phosphorylation of a reference substrate, RB-CTF (data not shown). Importantly, a clear signal for direct phosphorylation of pUL69 by CDK1-cycB1, CDK7-cycH-MAT1, and CDK9-cycT was detected (Fig. 4A, upper  panel, lanes 1, 3, and 4). Nonspecific phosphorylation activity was excluded by the lack of measurable phosphorylation of HCMV pUL26, used as a specificity control (Fig. 4A, lower  panel). Low level base-line phosphorylation of pUL69 could be detected without addition of CDK-cyclin complexes, which probably indicates traces of pUL69-phosphorylating kinase activity in the immunoprecipitates (Fig. 4A, upper panel, lane  5). The fold increase in phosphorylation of pUL69 by CDKcyclin complexes was then determined via densitometry. As shown in Fig. 4B, the strongest phosphorylation was mediated by CDK1-cycB1 (12.4-fold increase). Pronounced levels of phosphorylation were also measured for CDK9-cycT (7.2-fold) and CDK7-cycH-MAT1 (6.4-fold), whereas CDK2-cycE-mediated phosphorylation was lower (4.1-fold). Thus, these in vitro data indicate that direct phosphorylation of pUL69 can be mediated by several CDK-cyclin complexes. To assess phosphorylation of pUL69 in vivo, we infected HFFs with HCMV AD169 for 2 days and incubated cells with [␥-33 P]orthophosphate to allow for in vivo labeling of proteins. An evaluation of pUL69 immunoprecipitated from these cells revealed a strong signal of phosphorylation (Fig. 4C). Phosphorylation could be partly inhibited by the treatment of infected cells with 15 M roscovitine (signal reduction of ϳ66 Ϯ 13%; data not shown). However, this inhibition of phosphorylation did not occur in a CDK-specific manner and was also observed for other protein kinase inhibitors. This points to a complex regulation of the phosphorylation of pUL69. Thus, the in vivo phosphorylation of pUL69 appears to be dependent on CDK and other protein kinase activities.
The particularly strong CDK1-cycB1-mediated phosphorylation of pUL69, as demonstrated by in vitro data (Fig. 4, A and  B), raised questions about the nucleo-cytoplasmic translocation of CDK1. We determined whether CDK1 was detectable in pUL69-positive nuclei of HCMV-infected fibroblasts by performing immunofluorescence analysis, including confocal laser-scanning microscopy. For this purpose, a kinetic study was performed to investigate localization patterns during the immediate early and early phases of HCMV replication (supplemental Fig. S4). Nuclear pUL69 was observed from 4 hpi, and the percentage of pUL69-positive cells increased continu-ously over time. Under roscovitine treatment, a transient delay of pUL69 expression was detected (supplemental Fig. S4A), with reduced immunofluorescence signal intensities confirming a slightly lower level of pUL69 production. Interestingly, the beginning of nucleo-cytoplasmic translocation of CDK1 was observed at 8 hpi in both roscovitine-treated and untreated cells (supplemental Fig. S4B). Over the period analyzed, the fraction of pUL69-positive cells showing CDK1 in a nucleo-cytoplasmic or cytoplasmic localization steadily increased, with a completion of the translocation at about 24 hpi. Of note, this translocation was slightly retarded through the inhibition of CDK activity by roscovitine. Thus, although a direct colocalization between CDK1 and pUL69 was not detectable, the presence of both CDK1 and pUL69 in the nuclei of HCMV-infected fibroblasts may allow an inter-regulation of the two proteins at early time points of infection.
Inhibition of CDK Activity Reduces mRNA Export Activity of pUL69-A nuclear mRNA export assay was performed to investigate functional aspects of pUL69. This assay determined the ability of pUL69 to export intron-containing CAT mRNA. As shown in Fig. 5, a decline in nuclear export activity was observed when CDKs were inhibited by either roscovitine or R58. The pUL69-mediated export signal was reduced to 43% under roscovitine treatment (statistically significant, p Ͻ 0.01) and to 63% under R58 treatment. An inhibitory effect on the pUL69 nuclear export function was also detected for the pUL97-directed inhibitor, Gö6976 (41). Thus, CDK as well as pUL97 activity is required for the full functionality of pUL69 with regard to mRNA export.

DISCUSSION
The HCMV replication strategy has evolved to an elaborate inter-regulation with factors controlling the cell cycle. On the one hand, HCMV ensures that the regulatory state of the cellular environment efficiently supports viral reproduction, and on the other hand, HCMV reprograms the cellular factors such as regulatory protein kinases from their original function toward virus-specific regulatory pathways (2,3). A number of studies have shown that HCMV replication is functionally linked with CDK activity at various regulatory junctures. In this study, we provide novel insights into the link between cellular CDK activity and the intranuclear localization and functionality of the viral mRNA export factor pUL69. Our findings indicate the following: (i) HCMV-infected fibroblasts treated with CDK inhibitors show an intranuclear speckled aggregation of pUL69.  (ii) Variants of HCMV are differentially sensitive to inhibitors inducing pUL69 aggregation. (iii) Speckled pUL69 aggregates are mainly localized within viral replication centers. (iv) CDK9 and cyclin T1 strictly colocalize with the inhibitor-induced speckled aggregates, whereas other CDKs behave differently.
(v) The HCMV-triggered nucleo-cytoplasmic translocation of CDK1 does not exclude a putative early nuclear interaction with pUL69. (vi) pUL69 is phosphorylated in vivo and in vitro, identifying CDKs (mainly CDK1 and -9) as novel pUL69-phosphorylating kinases. (vii) CDK activity is required to stimulate a high level of nuclear mRNA export activity of pUL69 in a reporter assay.
The importance of CDK activity in the replication cycle of HCMV has been well documented (3, 4, 6 -9, 11, 12). However, for most of these investigations, the description of molecular mechanisms linking CDK activity with viral regulation of replication was still unresolved. We have identified the viral regulatory protein pUL69 as one target of CDK-mediated regulation. This protein contributes to an HCMV-induced cell cycle arrest that may result from interaction with CDKs and/or cyclins but is poorly understood so far (37,38). Additionally, pUL69 acts as a transcriptional transactivator via interaction with the cellular transcription elongation factor hSPT6 (57,58) and as a nuclear RNA export factor via interaction with UAP56, a component of the cellular mRNA export machinery (36,39,40). As pUL69 is a phosphoprotein (35,57), it has been speculated that its activity might be partly regulated through phosphorylation. In this study, we provide evidence for the phosphorylation of pUL69 by CDKs. pUL69 acted as a specific substrate in in vitro kinase reactions with all four analyzed CDKs (CDK1, -2, -7, and -9), whereby CDK1 and -9 exerted the highest pUL69-phosphorylating activity. Our data point to a combined impact of more than one CDK on the phosphorylation and activity of pUL69, based on findings that CDK1-cycB1 and CDK9-cycT exerted main activities in a pUL69-specific in vitro kinase assay, whereas CDK9-cycT exclusively showed colocalization with pUL69 during late phase of infection. In addition, both CDK inhibitors roscovitine and R58 induced speckled aggregates of pUL69, although they possess partly different inhibitory profiles toward individual CDKs. Thus, it remains speculative which of the analyzed CDKs are the key determinants for pUL69 regulation. The impact of CDKs on HCMV replication may be an ordered sequence of events with pUL69-directed activity of CDK1 an early event during viral replication. Consistent with this, both pUL69 and CDK1 were localized in the nucleus prior to CDK1 nucleo-cytoplasmic translocation. This illustrates that CDK1, although not colocalizing with pUL69, may contribute to the regulatory phosphorylation of pUL69 at early time points of infection. In contrast, CDK9 might be required at later time points for regulation of pUL69 activity, as demonstrated by direct colocalization of CDK9-cycT and pUL69 in late phase replication centers. Integrating these findings, this indicates a regulatory impact of CDK9 and CDK1 and, possibly, further CDKs on the functionality of pUL69.
Interestingly, the pUL97 viral protein kinase was characterized as a CDK-related kinase possessing similar functional properties (23). Our studies of other protein kinases involved in the phosphorylation of pUL69 was compatible with these findings. Recently, we provided evidence that the CDK ortholog pUL97 phosphorylates pUL69 (41). Combined with the data of this present study, we hypothesize that cellular CDK and viral pUL97 activities are required to modulate the nuclear localization and function of pUL69 during cytomegalovirus replication.
There is only a limited number of examples describing functional cross-talk between CDKs and herpesviral protein kinases. One prominent example is the sequentially ordered inter-regulation between CDK1-cdc25C and herpes simplex type 1 (HSV-1)-encoded protein kinases UL13 and US3 (59,60). In this case, the two HSV kinases, UL13 and US3, are capable of phosphorylating, and thereby activating the regulatory cellular phosphatase cdc25C, which normally activates CDK1 (cdc2), by removing two inhibitory phosphates. However, in HSV-1infected cells the function of cdc25C is reduced to modulating CDK1 function. In infected cells, CDK1 acquires a new binding partner, the HSV DNA polymerase processivity factor UL42. The CDK1-UL42 complex then functions as an activator of late viral gene expression. In addition, CDK1 is able to phosphorylate UL42 and therefore possibly stimulate its activity in the UL42-DNA polymerase complex (61).
In summary, our data imply that CDK-specific regulatory pathways modulate the multiple functions of pUL69 in HCMV-infected fibroblasts. Further studies will be required to gain a deeper insight into the respective molecular mechanisms and to learn more about contact points of the CDK-HCMV interaction.