Phosphorylation of p27Kip1 by Epstein-Barr Virus Protein Kinase Induces Its Degradation through SCFSkp2 Ubiquitin Ligase Actions during Viral Lytic Replication*

Epstein-Barr virus (EBV) productive replication occurs in an S-phase-like cellular environment with high cyclin-dependent kinase (CDK) activity. The EBV protein kinase (PK), encoded by the viral BGLF4 gene, is a Ser/Thr protein kinase, which phosphorylates both viral and cellular proteins, modifying the cellular environment for efficient viral productive replication. We here provide evidence that the EBV PK phosphorylates the CDK inhibitor p27Kip1, resulting in ubiquitination and degradation in a proteasome-dependent manner during EBV productive replication. Experiments with BGLF4 knockdown by small interfering RNA and BGLF4 knock-out viruses clarified that EBV PK is involved in p27Kip1 degradation upon lytic replication. Transfection of the BGLF4 expression vector revealed that EBV PK alone could phosphorylate the Thr-187 residue of p27Kip1 and that the ubiquitination and degradation of p27Kip1 occurred in an SCFSkp2 ubiquitin ligase-dependent manner. In vitro, EBV PK proved capable of phosphorylating p27Kip1 at Thr-187. Unlike cyclin E-CDK2 activity, the EBV PK activity was not inhibited by p27Kip1. Overall, EBV PK enhances p27Kip1 degradation effectively upon EBV productive replication, contributing to establishment of an S-phase-like cellular environment with high CDK activity.

The Epstein-Barr virus (EBV) is a human lymphotropic herpesvirus with a linear double-stranded DNA, 172 kb in length (17). Primary EBV infection targets resting B lymphocytes, inducing their continuous proliferation. In resultant B lymphoblastoid cell lines, a limited number of viral proteins are usually expressed, and there is no production of virus particles, this being termed latent infection. Reactivation from latency can occur spontaneously or be induced artificially. A virus-productive lytic stage is initiated by the expression of the viral immediate early BZLF1 protein, a b-Zip transcriptional factor that binds to AP-1-like sequences present in the promoters of early lytic genes, followed by an ordered cascade of viral gene expression, viral DNA replication, and virion production (18). During lytic replication, the levels of cyclin E and cyclin A continue to be elevated, and cyclin E-and cyclin A-associated CDK activities increase, thereby causing accumulation of hyperphosphorylated forms of retinoblastoma (Rb) protein and an increase of the level of E2F-1 protein (19). Chemical CDK inhibitors, like purvalanol A and roscovitine, block viral lytic replication through prevention of viral immediate early and early gene expression (20). Thus, a cellular environment with high CDK activity is required for efficient viral replication.
The EBV protein kinase (PK) encoded by the BGLF4 gene is a Ser/Thr protein kinase conserved throughout all subfamilies of Herpesviridae, targeting proline-directed Ser/Thr (21)(22)(23), expressed from an early stage of viral lytic replication and localized in viral replication compartments. EBV PK phosphorylates not only viral proteins, such as EBV BZLF1, BMRF1 (EA-D), EBNA-LP, EBNA2, and protein kinase itself, but also several cellular proteins, including EF-1␦, lamin A/C, MCM4, and MCM6 (22, 24 -29). The phosphorylation sites targeted by the BGLF4 protein include those with CDK1 (Cdc2) and CDK2. Especially, phosphorylation of Thr-19 and Thr-110 residues on MCM4 results in loss of helicase activity of MCM4-6-7 complexes (26), providing an example of EBV PK-mediated protein phosphorylation causing a dramatic functional change. Knockdown of the EBV PK protein by small interfering RNA reduces the yield of infectious virus particles (30).
In this paper, we document evidence that, during EBV productive replication, EBV PK phosphorylates p27 Kip1 so that it is ubiquitinated by SCF Skp2 ubiquitin ligase and degraded in a proteasome-dependent manner. Unlike the cyclin E-CDK2 activity, the EBV PK activity is not inhibited by p27 Kip1 . Overall, EBV possesses its own strategy to degrade p27 Kip1 upon onset of productive replication, contributing to provision of an S-phaselike cellular environment with high CDK activity.
Immunoblot Analysis-Cells were suspended in lysis buffer (20 mM Tris-HCl (pH 7.4), 0.5% Triton X-100, 300 mM NaCl, 1 mM EDTA, 0.1% SDS, 100 mM NaF, 2 mM Na 3 VO 4 , protease inhibitor mixture (Sigma)) and incubated on ice for 40 min followed by centrifugation. Equal amounts of proteins were separated by 15% (acrylamide (A)/bisacrylamide (B) ϭ 72:1) SDS-PAGE and transferred onto Immobilon transfer membranes (Millipore). Immunoreactivity was detected by Western Lightning (PerkinElmer Life Sciences), and images were processed with LumiVision PRO 400EX (Aisin/Taitec Inc.). Signal intensity was quantified with a LumiVision Analyzer 400. The system used in this study mounts a cooled CCD camera that has a 16-bit (65,535) grayscale wide dynamic range. It enhances the accuracy of the quantitative analysis up to 100 times compared with ordinary quantitative analysis scanning of an x-ray film into the personal computer after exposing the signal to the film.
In Vitro Kinase Assay-In vitro kinase assays of BGLF4 were performed as described previously (26). Briefly, GST-tagged human p27 Kip1 protein (400 ng; Abcam) was incubated with 20 and 50 ng of GST-fused wild-type BGLF4 or kinase-dead BGLF4 expressed in insect cells (26,27) in a 50-l reaction mixture containing 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 mM dithiothreitol, 10 mM MgCl 2 , 1 mM ATP, and 0.2 mM Na 3 VO 4 at 37°C for 60 min. The reactions were terminated by the addition of SDS gel loading buffer, and samples were subjected to immunoblot analysis with anti-p27 Kip1 -phospho-Thr-187 and p27 Kip1 antibodies. To assess inhibition by p27 Kip1 , inhibition assays were performed with some modifications from the previous report (37). A 30-ng aliquot of GST-fused BGLF4 and 50 ng of cyclin E-CDK2 (Upstate Biotechnology) were preincubated with various amounts of GST-fused p27 Kip1 (8,40,200, and 1000 ng) in the absence of ATP for 10 min at 30°C. Phosphorylation reactions (40 l) were conducted in 50 mM Tris-HCl (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM EGTA, 0.5 g of histone H1, 100 M ATP, and 5 Ci of [␥-32 P]ATP in each p27 Kip1 /kinase mixture for 5 min at 30°C. The reaction was terminated by the addition of SDS gel loading buffer, and the samples were separated by SDS-15% PAGE, followed by autoradiography.
Titration of Virus Yields from 293/EBV Cells-293/EBV-WT, 293/EBV-dBGLF4/NeoSt, and 293/EBV-dBGLF4/NeoSt/R cells were transfected with the BZLF1 expression vector to induce lytic replication. Alternatively, 293/EBV-WT cells were transfected with BZLF1 expression plasmid together with non-targeting/control siRNA, Skp2 siRNA, or various amounts of pcDNA-FLAG/p27 Kip1 . Cells and the culture supernatant were harvested at 60 hpt, freezethawed, and centrifuged. The supernatant from centrifugation was filtered and used as a virus stock. EBV-negative Akata(Ϫ) cells (38) were infected with the virus, and enhanced green fluorescent protein-positive cells were counted by fluorescenceactivated cell sorting.

RESULTS
Degradation of p27 Kip1 upon EBV Lytic Infection-EBV-latently infected Tet-BZLF1/B95-8 cells, in which exogenous BZLF1 protein is conditionally expressed under the control of a tetracycline-regulated promoter (31), were treated with doxycycline to induce lytic replication. As shown in Fig. 1A, BZLF1 protein was detected by 10 h postinduction (hpi) and reached a plateau at 24 hpi. The early EBV proteins, BMRF1 (polymerase accessory protein) and BGLF4 (protein kinase), became strongly detectable at 24 hpi. Interestingly, the protein level of p27 Kip1 decreased gradually from 24 hpi, and only 30% remained at 72 hpi. A proteasome inhibitor, MG132, restored the p27 Kip1 protein level in doxycycline-treated Tet-BZLF1/B95-8 cells (Fig. 1B), suggesting that down-regulation of p27 Kip1 upon EBV lytic infection is proteasome-dependent. Similarly, when Akata(ϩ) cells, an EBV-positive B cell line derived from Burkitt's lymphoma, were treated with polyclonal rabbit anti-human IgG to induce lytic replication, the p27 Kip1 protein level became reduced remarkably upon lytic replication (Fig. 1C).
Lack of EBV PK Prevents Proteasome-dependent Degradation of p27 Kip1 upon EBV Lytic Replication-There are two different pathways for the proteasome-dependent degradation of p27 Kip1 . During the G 1 phase, p27 Kip1 is phosphorylated at Ser-10, exported to the cytoplasm, and then ubiquitinated by KPC (14 -16). On the other hand, p27 Kip1 is phosphorylated at Thr-187 by cyclin E-CDK2 and ubiquitinated by the ubiquitin ligase SCF Skp2 in the nucleus, followed by degradation (4 -8). Since EBV PK can phosphorylate CDK2 or CDK1 target sites (21,26), there is a possibility that the EBV PK is involved in p27 Kip1 phosphorylation and subsequent degradation. At first, we examined whether silencing of BGLF4 expression by an siRNA strategy affects p27 Kip1 levels after induction of lytic replication. Tet-BZLF1/B95-8 cells were transfected with BGLF4-targeting siRNA (si-BGLF4) or nontargeting/control siRNA. At 17 hpt, cells were treated with doxycycline for induction of lytic replication ( Fig. 2A). In control siRNA-transfected cells, BGLF4 protein became detectable by 24 hpi, whereas the BGLF4-targeting siRNA prevented accumulation of BGLF4 protein throughout the lytic replication. The level of p27 Kip1 decreased significantly by 24 hpi in control siRNA-transfected cells. In contrast, the p27 Kip1 levels in cells transfected with BGLF4-targeting siRNA proved constant throughout the lytic replication. The expression levels of BZLF1 and BMRF1 proteins were comparable between cells transfected with either control siRNA or BGLF4 targeting siRNA ( Fig. 2A), confirming a previous report (30). Since the BMRF1 protein is a target of BGLF4 (24), the level of slower migrating bands of the BMRF1 protein in SDS-PAGE decreased with progression of the lytic replication in cells transfected with BGLF4-targeting siRNA, proving its efficacy. Next, p27 Kip1 protein levels were compared in lytic replication-induced HEK293 cells with a wild-type or BGLF4-deficient EBV-bacmid genome. When HEK293 cells harboring the wildtype EBV-bacmid (293/EBV-WT) were transfected with the BZLF1 expression vector to induce lytic replication, the p27 Kip1 protein level decreased (Fig. 2B), corresponding well with the data for Tet-BZLF1/B95-8 and Akata(ϩ) cells (Fig. 1). In contrast, no reduction of p27 Kip1 was observed in cells harboring the BGLF4 knock-out virus, whereas reduction was again observed in cells harboring revertant virus. Lytic replication appeared to be equally induced among the three viruses, since comparable amounts of BZLF1 protein were expressed. Lack of BGLF4 expression and change in phosphorylation states of BMRF1 protein in the cells with the knock-out virus were clearly observed, as expected. It was also confirmed that levels of viral DNA synthesis among those viruses are almost the same (data not shown), but the virus yield with the knock-out virus was significantly impaired as compared with wild-type and revertant virus cases (Fig. 2C). These results strongly suggest that EBV PK is involved in p27 Kip1 degradation in the context of lytic replication.
EBV PK Enhances Ubiquitination and Proteasome-dependent Degradation of p27 Kip1 -We next investigated direct effects of BGLF4 expression on p27 Kip1 protein levels in HEK293T and HeLa cells. Cells were transfected with empty vector or expres- sion vectors for wild-type BGLF4 (WT BGLF4) or kinase-dead BGLF4 (kd BGLF4) containing a Lys-102 to Ile mutation (Fig.  3A). The level of p27 Kip1 in both HEK293T and HeLa cells transfected with the wild-type BGLF4 expression vector was significantly decreased as compared with that in cells transfected with empty vector, whereas transfection of kinase-dead BGLF4 expression vector did not affect the level of p27 Kip1 (Fig.  3A, lanes 1, 3, and 5). Treatment with MG132 increased p27 Kip1 protein to comparable levels, regardless of the types of transfected plasmid (Fig. 3A, lanes 2, 4, and 6). Thus, p27 Kip1 is subjected to degradation via cellular mechanisms, and the expression of EBV PK can further enhance proteasome-dependent degradation of p27 Kip1 in the absence of other viral factor(s). Next, to examine whether EBV PK promotes ubiquitination of p27 Kip1 , expression vectors for FLAG-tagged p27 Kip1 and HAtagged ubiquitin (Ub) were cotransfected into HEK293T cells with expression vectors for wild-type or kinase-dead BGLF4 in the presence of MG132 (Fig. 3B). Exogenously expressed p27 Kip1 proteins were immunoprecipitated with anti-FLAG antibodies and subjected to immunoblotting with anti-p27 Kip1 and HA antibodies. Successful precipitation of p27 Kip1 protein was confirmed in both WT BGLF4-and kd BGLF4-transfected cells (Fig. 3B, lanes 5 and 6). In contrast, HA-ubiquitinated p27 Kip1 was detectable in WT BGLF4-transfected cells but not in kd BGLF4-transfected cells (Fig. 3B, lanes 7 and 8). These observations indicate that kinase activity of BGLF4 protein enhances ubiquitination and proteasome-dependent degradation of p27 Kip1 in cells.
EBV PK Phosphorylates p27 Kip1 on Thr-187 in Vivo and in Vitro-BGLF4 protein is a Ser/Thr-kinase that phosphorylates several cellular and viral substrates and often targets a prolinedirected Ser/Thr (21)(22)(23). Human p27 Kip1 possesses three serine/threonine sites followed by proline residues (Ser-10, Ser-178, and Thr-187). To determine the phosphorylation site of p27 Kip1 targeted by EBV PK, we investigated the protein stability of p27 Kip1 with mutation in Ser-10, Ser-178, or Thr-187 (Fig. 4A). Wildtype BGLF4 (WT BGLF4) or kinasedead BGLF4 (kd BGLF4) expression vectors were cotransfected into HEK293T cells with expression vectors for FLAG-tagged wild-type p27 Kip1 or FLAG-tagged mutant p27 Kip1 with an alanine mutation in either Ser-10, Ser-178, or Thr-187. The protein level of FLAG-tagged wild-type p27 Kip1 was significantly decreased when cotransfected with the WT BGLF4 expression vector. A specific antibody against phospho-Thr-187 on p27 Kip1 showed that wild-type p27 Kip1 could be phosphorylated at Thr-187 when cotransfected with the wild-type BGLF4 expression vector, whereas a specific antibody against phospho-Ser-10 showed that overexpressed p27 Kip1 was phosphorylated at Ser-10 on p27 Kip1 even in the absence of expression of BGLF4 protein. Phosphorylation of Ser-10 is known to be responsible for a shift in the electrophoretic mobility of p27 Kip1 (34). Consistent with this, the mobility of the p27 Kip1 Ser-10 to Ala mutant on SDS-PAGE was faster than that of wild-type p27 Kip1 . These observations indicate overexpressed p27 Kip1 to be phosphorylated at Ser-10 on p27 Kip1 even without expression of the BGLF4 protein. Regarding the stability of p27 Kip1 protein, p27 Kip1 protein levels containing alanine mutations of Ser-10 or Ser-178 as well as wildtype p27 Kip1 were significantly decreased when WT BGLF4 protein was co-expressed. Thus, it was proved that Ser-10 and Ser-178 residues are not involved in degradation of p27 Kip1 enhanced by BGLF4 protein. In contrast, the protein level of the p27 Kip1 Thr-187 to Ala mutant was not so affected even when the WT BGLF4 protein was co-expressed. Thus, these observations strongly suggest that phosphorylation of the Thr-187 residue is mainly required for enhanced p27 Kip1 degradation. It should be noted that there is a possibility that other Ser/Thr residue(s) are phosphorylated by EBV PK, because the protein level of the p27 Kip1 Thr-187 to Ala mutant with expression of BGLF4 protein did not recover completely.
Next, we examined whether the BGLF4 protein can directly phosphorylate the Thr-187 residue on p27 Kip1 in vitro. As shown in Fig. 4B, immunoblot analysis using specific antibody against phospho-Thr-187 of p27 Kip1 demonstrated this to be the case.
We then tried to determine the phosphorylation state of Thr-187 on p27 Kip1 after induction of lytic replication. Unfortunately, phospho-Thr-187 of endogenous p27 Kip1 was not clearly detected (data not shown), probably due to rapid degradation of the phosphorylated p27 Kip1 . Therefore, the phosphorylation states of overexpressed p27 Kip1 were examined in 293/ EBV-WT cells before and after the induction of lytic

Degradation of p27 Kip1 Enhanced by EBV PK BGLF4
replication. FLAG-tagged wild-type p27 Kip1 or FLAG-tagged p27 Kip1 Thr-187 to Ala mutant were expressed in 293/ EBV-WT cells, which were induced to enter lytic replication by transfection of BZLF1 expression vector (Fig. 4C). The results revealed that the amounts of exogenously expressed p27 Kip1 did not change before and after the induction of lytic replication. Phosphorylation levels of Ser-10 were constant as well. In contrast, phosphorylation of Thr-187 on p27 Kip1 increased significantly after the induction of lytic replication. Similarly, the phosphorylation states of overexpressed p27 Kip1 were examined in Tet-BZLF1/B95-8 cells that were transfected with the expression vector of FLAG-tagged wildtype p27 Kip1 and then treated with doxycycline at 26 hpt (Fig.  4D). The expression levels of exogenous p27 Kip1 protein were constant until 48 hpi. Phosphorylation levels of Ser-10 on p27 Kip1 also proved constant until 48 hpi. In contrast, phosphorylation of Thr-187 on p27 Kip1 increased by 24 hpi, this correlating with the appearance of BGLF4 protein. Thus, it was demonstrated that p27 Kip1 is phosphorylated at Thr-187 upon EBV lytic replication.
EBV PK Enhances p27 Kip1 Degradation Mediated by the Ubiquitin Ligase SCF Skp2 -It was previously reported that Skp2 (S-phase kinase-associated protein 2), an F-box substrate recognition subunit of the SCF ubiquitin ligase complex, recognizes Thr-187-phosphorylated p27 Kip1 and promotes its ubiquitination and subsequent degradation (4,5,7,8). We therefore examined whether SCF Skp2 ubiquitin ligase is involved in BGLF4-mediated p27 Kip1 degradation. Nontargeting/control siRNA or Skp2-targeting siRNA were cotransfected into HEK293T cells with the FLAG-tagged p27 Kip1 expression vector and wild-type or kinase-dead BGLF4 expression vectors (Fig. 5A). Treatment with Skp2 siRNA reduced the Skp2 protein level by 54% compared with that of control siRNA and prevented the degradation of p27 Kip1 even when wild-type BGLF4 protein was expressed, strongly suggesting that Skp2 is involved in the EBV PKmediated p27 Kip1 degradation. , together with expression vectors for FLAG-tagged WT p27 Kip1 , mutant p27 Kip1 -T187A, S10A, or S178A. Cells were harvested at 27 hpt, and whole cell lysates were prepared and subjected to immunoblot analysis with anti-GAPDH, p27 Kip1 -pT187, or p27 Kip1 -pS10 antibody or anti-FLAG antibody to detect exogenously expressed BGLF4 (FLAG-BGLF4) and p27 Kip1 (FLAG-p27). B, GST-tagged WT and kd BGLF4 proteins isolated from insect cells were incubated with p27 Kip1 in the presence of 1 mM ATP at 37°C for 60 min. The samples were applied for immunoblot analysis with anti-p27 Kip1 or p27 Kip1 -pT187 antibodies. C, expression vectors for FLAG-tagged WT p27 Kip1 or p27 Kip1 -T187A were cotransfected with the BZLF1 expression vector (Z) or empty vector (Ev) into 293/EBV-WT cells. Cells were harvested at 48 hpt, and equal amounts of whole cell lysates from each sample were subjected to immunoblot analysis with the indicated antibodies. D, FLAG-tagged WT p27 Kip1 -overexpressed Tet-BZLF1/B95-8 cells were cultured in the presence of doxycycline (DOX) and harvested at the indicated times. Equal amounts of whole cell lysates were subjected to immunoblot analysis with the indicated antibodies. FIGURE 5. The BGLF4 protein enhances p27 Kip1 degradation mediated by the ubiquitin ligase, SCF Skp2 . A, HEK293T cells transfected with the expression vector for FLAG-tagged p27 Kip1 were transfected with non-targeting/ control siRNA (si-control) or Skp2 siRNA (si-Skp2) together with expression vectors for FLAG-tagged WT BGLF4, kd BGLF4, or empty vector (Ϫ). Cells were harvested at 32 hpt, and whole cell lysates were prepared and subjected to immunoblot analysis with anti-Skp2, GAPDH antibodies, and anti-FLAG antibody to detect exogenously expressed BGLF4 (Flag-BGLF4) and p27 Kip1 (Flag-p27). B, 293/EBV-WT cells were subjected to transfection with the BZLF1 expression vector for induction of lytic EBV replication together with control siRNA or Skp2 siRNA and were harvested at 0, 24, and 48 hpt. Equal amounts of whole cell lysates were subjected to immunoblot analysis with the indicated antibodies.
We next examined the effect of Skp2 on the p27 Kip1 degradation after induction of lytic replication in 293/EBV-WT cells by transfection of the BZLF1 expression vector in the presence of non-targeting/control siRNA or Skp2-targeting siRNA (Fig.  5B). When cells were treated with Skp2 siRNA, the protein level of Skp2 was reduced apparently by 24 hpt, whereas the protein level of p27 Kip1 increased markedly (Fig. 5B). The amounts of exogenously expressed BZLF1 and subsequently expressed BMRF1 and BGLF4 proteins were almost the same in non-targeting/control siRNA-and Skp2 siRNA-treated cells. Thus, these observations strongly suggest that the SCF Skp2 ubiquitin ligase is involved in the degradation of p27 Kip1 during EBV lytic replication.
EBV PK Activity Is Not Inhibited by p27 Kip1 and Phosphorylates p27 Kip1 Efficiently-Although the p27 Kip1 -bound CDK2 is catalytically inactive in vitro, several groups have recently discovered that phosphorylation of p27 Kip1 at Tyr-74 and Tyr-88 residues by Abl and Src family kinases is a prerequisite for p27 Kip1 to become a substrate for cyclin E-CDK2 (9 -11). To determine the difference between p27 Kip1 phos-phorylation by EBV PK and by cyclin E-CDK2, purified BGLF4 protein and cyclin E-CDK2 proteins were assayed for their ability to phosphorylate p27 Kip1 in vitro (Fig. 6). BGLF4 protein and cyclin E-CDK2 were preincubated with various amounts of p27 Kip1 in the absence of ATP, and then ATP and histone H1 as a substrate were added to the reaction. Cyclin E-CDK2 could phosphorylate histone H1 in vitro, but the kinase activity was inhibited by the addition of p27 Kip1 in a dose-dependent manner. In contrast, BGLF4 protein phosphorylated histone H1 even at a concentration of p27 Kip1 that completely inhibited cyclin E-CDK2 activity. With high concentrations of purified p27 Kip1 , cyclin E-CDK2 did not phosphorylate p27 Kip1 at all. In contrast, BGLF4 protein still phosphorylated p27 Kip1 , correlating with the data in cells (Fig. 4A). Thus, unlike cyclin E-CDK2, EBV PK escapes from the inhibition of p27 Kip1 and efficiently phosphorylates p27 Kip1 , thereby leading to p27 Kip1 degradation.
Stable Retention of p27 Kip1 Prevents Efficient Viral Lytic Replication-To examine whether inhibition of p27 Kip1 degradation affects the virus yield, lytic replication was induced in 293/EBV-WT cells that were treated with Skp2-targeting siRNA (Fig. 7A). Treatment of Skp2 siRNA reduced the level of Skp2 while the protein level of p27 Kip1 was increased, as shown in Fig. 5B. The virus yield from cells treated with Skp2 siRNA was 64%, compared with that with control siRNA at 60 hpt, taken as 100% (Fig. 7A), correlating with the impairment of virus production in cells with the BGLF4-deficient virus (Fig.  2C). Next, we investigated whether overexpression of p27 Kip1 protein affects the virus yield by transfecting into 293/EBV-WT cells with the BZLF1 expression vector together with increasing amounts of p27 Kip1 expression vector (Fig. 7B). Virus yields in 293/EBV-WT cells transfected with 1, 10, 50, and 100 ng of p27 Kip1 expression vector were 86, 61, 59, and 47%, respectively, compared with that in the absence of p27 Kip1 expression vector. Overall, p27 Kip1 degradation by EBV PK should be required for efficient viral lytic replication.

DISCUSSION
EBV lytic replication occurs in an S-phase-like cellular environment with high CDK activity (20,31). In this report, we document evidence that EBV PK phosphorylates Thr-187 on p27 Kip1 , a CDK inhibitor, so that it is ubiquitinated by SCF Skp2 ubiquitin ligase and degraded to allow viral replication to proceed. The experimental data from siRNA-mediated EBV PKknockdown and with the BGLF4knock-out virus thus clearly demonstrate that EBV PK is involved in p27 Kip1 degradation in the context of viral lytic replication. Although cyclin E-CDK2 regulates the pathways of phosphorylation-dependent ubiquitination and degrada-  tion of p27 Kip1 during the cell cycle, the expression of EBV PK disrupts the pathways, leading to continuous degradation. To maintain a cellular environment with high CDK activity, which is advantageous for viral lytic replication, EBV might possess a variety of strategies to down-regulate CDK inhibitors like p27 Kip1 . We reported previously that the levels of cyclin E and cyclin A continue to be elevated and cyclin E-and cyclin A-associated CDK activities increase as lytic replication progresses (19). This elevated activity might be achieved by not only increased levels of cyclin E and cyclin A but also p27 Kip1 degradation. Moreover, we recently found that during the lytic infection, p53 was actively degraded, resulting in an undetectable level of another CDK inhibitor protein p21 Cip1/Waf1 (39). Thus, EBV lytic replication possesses complex mechanisms to evade inhibition of CDK activity by p21 Cip1/Waf1 and p27 Kip1 to promote an S-phase-like cellular environment.
Knockdown of Skp2 expression by means of siRNA technology and p27 Kip1 -overexpression resulted in reduction of virus yields, corresponding with impaired virus production with the BGLF4-deficient virus (Figs. 2C and 7). This suggests that p27 Kip1 degradation is required for efficient viral lytic replication and that it contributes to establishment of an S-phase-like cellular environment with high CDK activity for viral lytic replication. Actually, we earlier reported that purvalanol A and roscovitine, inhibitors of S-phase CDKs, blocked viral lytic replication when cells were treated at early stages of lytic infection (20). This observation supports the idea that cyclin A-CDK2 and cyclin E-CDK2 activities are somehow critical.
Virus production with the BGLF4-deficient virus (Fig. 2C) was here found to be severely impaired, whereas that of knockdown of Skp2 expression with siRNA and p27 Kip1 overexpression was moderately reduced (Fig. 7). So far, it has been reported that BGLF4 phosphorylates a variety of viral and cellular proteins (22, 24 -29). Lee et al. (22) recently showed that EBV PK phosphorylates lamin A/C to promote reorganization of nuclear inner membrane, which in turn may facilitate BFLF2 and BFRF1 interaction, thereby enhancing virion maturation. Thus, virus production in the BGLF4-deficient virus was presumably affected by not only p27 overexpression but also loss of phosphorylation of target proteins by EBV PK.
We and others have previously reported that expression of EBV PK, BGL4, results in cell growth arrest alone (26,40). Furthermore, Lee et al. (41) reported cell cycle profiles of asynchronized HeLa cells with or without expression of BGLF4 and showed that the cell population of G 1 /S border increased ϳ10%, whereas the cell population of G 2 /M border was slightly reduced with BGLF4 expression. Targets of BGLF4 are not only p27 Kip1 found in this study but also several cellular proteins affecting cell growth, such as EF-1␦, lamin A/C, MCM complex, etc., as described in the Introduction. For example, phosphorylation of MCM4 by BGLF4 results in loss of helicase activity of MCM complex (26). Therefore, the cell cycle distribution seen in BGLF4-expressing cells could be caused not only by the enhanced degradation of p27 Kip1 but through complex mechanisms.
Other ␥-herpesviruses possess their own strategies to degrade p27 Kip1 . Kaposi's sarcoma-associated herpesvirus virus-encoded cyclin (v-cyclin), a latent viral protein, forms a complex with CDK6 and phosphorylates Thr-187 on p27 Kip1 and leads to down-regulation at the protein level (42,43). Also, the viral cyclin encoded by murine herpesvirus 68 preferentially associates with CDK2, and the complex phosphorylates Thr-187 on p27 Kip1 and leads to down-regulation (44). Although EBV does not encode any v-cyclin homologue in its genome, it is reported that latent protein EBNA3C expression prevents the accumulation of p27 Kip1 (45). EBNA3C associates with the cyclin A-CDK2 complex and with the Skp2 subunit of SCF Skp2 , and the recruitment of SCF Skp2 ubiquitin ligase to cyclin A-CDK2-p27 Kip1 complex by EBNA3C results in ubiquitination and degradation of p27 Kip1 . Thus, with EBV latent infection, EBNA3C regulates p27 Kip1 stability by manipulating Skp2 (46). On the other hand, EBV PK mainly appears to regulate the level of p27 Kip1 in lytic replication.