Epstein-Barr virus-encoded protein kinase BGLF4 mediates hyperphosphorylation of cellular elongation factor 1{{delta}} (EF-1{{delta}}): EF-1{{delta}} is universally modified by conserved protein kinases of herpesviruses in mammalian cells

Previous studies (described below) suggested that herpesviruses interact with the host cellular translation elongation factor 1δ (EF-1δ), which plays a key role in the regulation of host cell protein synthesis (Merrick, 1992 ; Morales et al., 1992 ; Riis et al., 1990 ; Van Damme et al., 1990 ). Thus, (i) the involvement of EF-1δ was first highlighted by our observations that a regulatory protein of herpes simplex virus type 1 (HSV-1), the prototype alphaherpesvirus, interacts with EF-1δ (Kawaguchi et al., 1997a ). The regulatory protein ICP0, which is known primarily for its function as a promiscuous transactivator (Everett et al., 1991 ; Roizman & Sears, 1996 ), was shown to interact with EF-1δ, and the domain of ICP0 that interacts with EF-1δ was shown to affect translational efficiency in vitro (Kawaguchi et al., 1997a ). This observation was subsequently confirmed by a study of an unrelated virus, human immunodeficiency virus type 1 (HIV-1). A regulatory protein of HIV-1, Tat, also interacts with EF-1δ, and the domain of Tat that interacts with EF-1δ mediates the shut-off of host cell mRNA translation both in vitro and in vivo (Xiao et al., 1998 ). Interestingly, this domain of Tat shows a weak but consistent similarity to the region of ICP0 that interacts with EF-1δ, suggesting that HSV-1 ICP0 and HIV-1 Tat modulate cellular protein synthesis in a similar way. (ii) In the course of studying the interaction between ICP0 and EF-1δ, we also found that HSV-1 infection causes extensive hyperphosphorylation of EF-1δ (Kawaguchi et al., 1997a ). Subsequent experiments revealed that a viral protein kinase encoded by the UL13 gene mediates EF-1δ modification (Kawaguchi et al., 1998 ). The observation that HSV-1 has evolved two viral regulatory proteins for the optimization of EF-1δ function suggests that EF-1δ plays an important role in its life cycle. (iii) The amino acid sequence of the UL13 protein kinase is conserved in all members of the Herpesviridae subfamilies (Chee et al., 1989 ; Smith & Smith, 1989 ) and this led us to investigate whether herpesviruses other than HSV-1 possess the ability to induce EF-1δ hyperphosphorylation. Recently, we demonstrated that various alpha-, beta- and gammaherpesviruses commonly induce EF-1δ modification in cells from various mammalian species (Kawaguchi et al., 1999 ). Furthermore, human cytomegalovirus (HCMV) UL97, a betaherpesvirus homologue of UL13, mediates EF-1δ hyperphosphorylation. These observations suggested that EF-1δ is modified in mammalian cells infected with any herpesviruses subfamily member and that EF-1δ is universally important in herpesvirus infection.

Although HSV-1 UL13 and HCMV UL97 have been shown to mediate modification of EF-1δ (Kawaguchi et al., 1998 , 1999 ), it is not yet known whether conserved protein kinases of herpesviruses (HSV-1 UL13 homologues) are commonly involved in modification of EF-1δ because (i) the modification has not been shown by gammaherpesvirus protein kinases and (ii) conserved protein kinases sometimes show other biological activities (Heineman & Cohen, 1995 ; Moffat et al., 1998 ; Ng et al., 1994 , 1998 ; Ogle et al., 1997 ; Purves & Roizman, 1992 ). We wished, therefore, to examine whether a gammaherpesvirus protein kinase hyperphosphorylates EF-1δ. This would further support the hypothesis that EF-1δ modification is a conserved function that is expressed by all herpesviruses subfamily members in mammalian cells and that conserved protein kinases encoded by herpesviruses universally mediate this modification. In this report, we present evidence that this is in fact the case. We report here that BGLF4, a viral protein kinase of EpsteinBarr virus (EBV), mediates EF-1δ modification at a cellular level.

Cells.
Spodoptera frugiperda Sf9 cells were maintained in TC100 (GibcoBRL) with 10% foetal calf serum (FCS), 0·26% tryptose phosphate broth and 50 µg/ml kanamycin. B95-8 cells, an EBV-producing simian cell line, were maintained in RPMI 1640 medium supplemented with 10% FCS. The monkey kidney epithelial cell line COS-7 was maintained in Dulbeccos modified Eagles medium supplemented with 5% FCS.

Plasmids.
EBV DNA was isolated from B95-8 cells as described previously (Horimoto et al., 1992 ). The entire EBV BGLF4 open reading frame (ORF) was amplified by PCR using viral DNA as a template and the primers GCGAATTCGGAACATGGATGTGAATATG and GCGGATCCTCATCCACGTCGGCCATCTG. The amplified fragments were digested with EcoRI/BamHI and cloned into the EcoRI and BamHI sites of pBluescript II KS+ (Stratagene). The resultant plasmid was designated pBS-BGLF4-stop. pAcGHLT-BGLF4 (Fig. 1C) was generated by inserting an EcoRINotI fragment of pBS-BGLF4-stop into pAcGHLT-B (Pharmingen). The entire EF-1δ ORF without the stop codon was amplified from pBH1003 (Kawaguchi et al., 1997a ) by PCR using the primers GCGAATTCAGAAAAATGGCTACAAACTT and GCGGATCCGATCTTGTTGAAAGCTGCGA. The amplified fragments were digested with EcoRI/BamHI and cloned into the EcoRI and BamHI sites of pBS-Flag-Stop (Kawaguchi et al., 2000 ). The resultant plasmid was designated pBS-EF-1δ(F)-Stop. The EcoRINotI fragment of EF-1δ(F) was inserted into pFASTBAC DUAL (GibcoBRL) to generate pFASTBAC-EF-1δ(F). To construct pBS-BGLF4(F), a fragment of EBV viral DNA encoding the entire coding sequence of BGLF4 without the stop codon was amplified by PCR using the primers GCGAATTCGGAACATGGATGTGAATATG and GCGGATCCTCCACGTCGGCCATCTGGAC. This fragment was then cloned into pBS-Flag-Stop in-frame with the Flag epitope. The EcoRINotI fragment of pBS-BGLF4(F) was inserted into the EcoRI and NotI sites of pME18S (kindly provided by K. Maruyama, Tokyo Medical and Dental University, Bunkyo-ku, Tokyo, Japan) to yield pME-BGLF4(F) (Fig. 1D). In pME-BGLF4(F), the expression of the 3' Flag epitope-tagged BGLF4 was driven by the SRα promoter (Takebe et al., 1988 ).

(14K): Fig. 1. (A) Schematic diagram of the EBV genome. Five unique sequences are indicated as U1U5. The terminal (TR) and internal (IR) repeats flanking the unique sequences are shown as open rectangles. (B) Expanded view of the domain encoding the BGLF4 gene. The direction of ORFs is shown by arrows. (C) The transfer plasmid pAcGHLT-BGLF4, which was used for the construction of recombinant baculovirus expressing GSTBGLF4. (D) The expression plasmid pME-BGLF4, which was used for the expression of BGLF4 in COS-7 cells.

Generation of recombinant baculoviruses.
Either pAcGHLT-BGLF4 or pAcGHLT-B was cotransfected with linearized baculovirus DNA BaculoGold (Pharmingen) into Sf9 cells using Lipofectin (GibcoBRL), as described previously (Kawaguchi et al., 1997b ), to generate recombinant baculoviruses designated either Bac-GST-BGLF4 or Bac-GST. A recombinant baculovirus that expresses human EF-1δ (Bac-hEF-1δ) was also constructed using the BAC-TO-BAC Baculovirus Expression System, according to the manufacturers instructions (GibcoBRL). pFASTBAC-EF-1δ(F) was transfected into E. coli strain DH10BAC (GibcoBRL) and recombinant bacmid DNA was isolated and transfected into Sf9 cells using Lipofectin. The recombinant viruses were subsequently amplified in Sf9 cells.

Purification of EBV BGLF4 protein.
Sf9 cells (1·0x10⁶) infected with each baculovirus (Bac-GST-BGLF4 or Bac-GST) in 0·5 ml of ice-cold buffer C (50 mM TrisHCl, pH 7·5, 100 mM NaCl, 5 mM MgCl₂, 0·1% Nonidet P-40, 10% glycerol and 1 mM PMSF) were lysed by sonication. After insoluble material was removed by centrifugation, the supernatants were mixed with 50 µl of a 50% slurry of glutathioneSepharose beads (Amersham Pharmacia) for 2 h. The beads were extensively washed with buffer C and eluted with elution buffer (10 mM glutathione and 500 mM TrisHCl, pH 8·0). Next, the eluted supernatants were reacted with Ni²⁺NTA agarose beads (Qiagen) for 1 h. The beads were then washed three times with buffer C. Purified protein captured on the beads was separated by 10% SDSPAGE and either silver-stained (Fig. 2A) or immunoblotted (Fig. 2B) with rabbit antiserum containing anti-GST antibody, as described previously (Kawaguchi et al., 1997b ).

(68K): Fig. 2. Silver-stained gels (A) and immunoblots (B) of purified GST or GSTBGLF4 from Sf9 cells infected with the recombinant baculovirus Bac-GST (lanes 13) or Bac-GST-BGLF4 (lanes 46). Total cell extracts (lanes 1 and 4) were subjected to affinity chromatography on glutathioneSepharose (lanes 2 and 5) or Ni2+NTA agarose (lanes 3 and 6), as described in Methods. The proteins were separated on a denaturing gel and either silver-stained or transferred onto a nitrocellulose sheet and reacted with the antibody to EF-1δGST. Molecular masses (kDa) are indicated on the left.

In vitro kinase assays.
Purified GSTBGLF4 or GST captured on Ni²⁺NTA agarose beads was rinsed twice with washing buffer (50 mM TrisHCl, pH 9·0 and 2 mM DTT). Kinase assay reactions were performed with the purified GST proteins at 37 °C for 30 min in a total volume of 50 µl of kinase buffer (50 mM TrisHCl, pH 8·0, 200 mM NaCl, 50 mM MgCl₂, 0·1% Nonidet P-40 and 1 mM DTT) containing 5 µCi [γ-³²P]ATP. After incubation, samples were washed with TNE buffer (20 mM TrisHCl, pH 8·0, 100 mM NaCl and 1 mM EDTA) three times and the phosphorylated proteins were separated by 12% SDSPAGE. The proteins were then transferred onto nitrocellulose sheets, stained with Ponceau S and exposed to X-ray film. Purification and characterization of EBV BGLF4 protein
Chen et al. (2000) used an immune complex kinase assay to demonstrate a protein kinase activity associated with the BGLF4 protein. However, these results could not completely exclude the possibility that a contaminating protein(s) in the immune complex was responsible for this protein kinase activity. As the goal of this study was to determine whether or not BGLF4 mediates EF-1δ hyperphosphorylation, it would be necessary to purify BGLF4 and to show that the purified protein by itself exhibits protein kinase activity.

The objective of the first series of experiments was to purify the EBV BGLF4 gene product. We expressed and purified BGLF4 as a histidine-tagged GST fusion protein using the baculovirus system, as described in Methods. Purified protein was then electrophoretically separated in a denaturing gel and either silver-stained (Fig. 2A) or immunoblotted with rabbit antiserum containing anti-GST antibody (Fig. 2B).

The purified supernatants from Sf9 cells infected with either Bac-GST or Bac-GST-BGLF4 each contained one major purified protein with an M_r of 32000 or 78000, as detected by silver staining (Fig. 2A), and these proteins reacted with antiserum containing anti-GST antibody (Fig. 2B). These results indicated that we had purified the desired GST fusion proteins.

Protein kinase activity of purified BGLF4
Many protein kinases have autophosphorylating activity (Edelman et al., 1987 ). To determine whether purified BGLF4 does in fact possess kinase activity, we examined the ability of purified BGLF4 to autophosphorylate itself in the kinase assay, as described in Methods. The results (Fig. 3) were as follows: (i) electrophoretically separated purified GST protein, which was incubated in kinase buffer, did not contain any labelled bands (Fig. 3A, B). However, in the autoradiographic image of purified BGLF4, a protein band with an apparent M_r of 78000 was labelled (Fig. 3B). The electrophoretic mobility of labelled BGLF4 was the same as that of purified BGLF4 stained with Ponceau S (Fig. 3A, B). (ii) To determine if the labelling of BGLF4 with [γ-³²P]ATP was due to phosphorylation, labelled BGLF4 was boiled to inactivate kinases and incubated with 50 units of alkaline phosphatase at 37 °C for 30 min. As shown in Fig. 3(D), the band of labelled BGLF4 was eliminated by phosphatase treatment, indicating that BGLF4 was labelled with [γ-³²P]ATP by phosphorylation. Fig. 3(C) shows the results of Ponceau S staining of the same blot used in the experiment of Fig. 3(D). These data showed that BGLF4 levels remained relatively unchanged after boiling and indicated that boiling did not result in the loss of BGLF4. Thus, BGLF4 autophosphorylates itself and possesses protein kinase activity.

(27K): Fig. 3. In vitro kinase assays of purified GST and GSTBGLF4 proteins. (A) Purified GST (lane 1) or GSTBGLF4 (lane 2) were incubated in kinase buffer containing [γ-32P]ATP, separated on a denaturing gel, transferred onto a nitrocellulose sheet and stained with Ponceau S. (B) Autoradiograph of the nitrocellulose sheet described in (A). (C) Purified GSTBGLF4 incubated in kinase buffer containing [γ-32P]ATP (lane 2) and the labelled protein treated with alkaline phosphatase (CIP) (lane 1) were separated on a denaturing gel, transferred onto a nitrocellulose sheet and stained with Ponceau S. (D) Autoradiograph of the nitrocellulose sheet described in (C). Molecular masses are indicated on the left.

BGLF4 protein mediates post-translational modification of EF-1δ
To address the question of whether or not BGLF4 is involved in EF-1δ modification, we performed two series of experiments. In the first we constructed a recombinant baculovirus expressing human EF-1δ (Bac-hEF-1δ), as described in Methods. The expression of EF-1δ was confirmed by immunoblotting with Bac-hEF-1δ (Fig. 4A). Using the baculoviruses expressing GSTBGLF4 and human EF-1δ, we examined whether human EF-1δ was modified by BGLF4 in insect cells. As reported previously (Kawaguchi et al., 1997a ), EF-1δ consists of two predominant forms: a hypophosphorylated form (apparent M_r of 38000) and a hyperphosphorylated form (apparent M_r of 40000). The polyclonal antibody (Kawaguchi et al., 1997a ) used in this study can readily detect both forms of EF-1δ and the pattern of bands of EF-1δ radiolabelled by ³²P was exactly the same as that of EF-1δ detected by immunoblotting (Kawaguchi et al., 1998 ). Immunoblotting using the anti-EF-1δ antibody can, therefore, be used to monitor modification of EF-1δ. As shown in Fig. 4(A), when human EF-1δ was expressed in Sf9 cells two predominant forms of protein with an M_r of 38000 and 40000 were produced. The pattern of protein expression was similar to that in mammalian COS-7 cells (Fig. 4A). In Sf9 cells either infected with Bac-hEF-1δ or coinfected with Bac-GST and Bac-hEF-1δ, the hypophosphorylated form of EF-1δ was dominant (Fig. 4B, lane 2). In contrast, the ratio of proteins (upper:lower bands) was significantly increased by coinfection with Bac-GST-BGLF4 and Bac-hEF-1δ (Fig. 4B, lane 3), although the total amount of EF-1δ expressed in COS-7 cells decreased compared with the amount expressed in Sf9 infected with Bac-hEF-1δ and Bac-GST. This change of EF-1δ is very similar to that observed in HFF cells infected with HSV-1(F) (Kawaguchi et al., 1997a ) and, presumably, overexpression of BGLF4 is toxic to the cells as reported for other UL13 homologues (Ng et al., 1996 ). Supporting this hypothesis, the levels of the other cellular proteins slightly decreased when BGLF4 was overexpressed in baculovirus-infected cells (Fig. 4C, lanes 2 and 3). Fig. 4(C) also provides evidence that the levels of the other cellular proteins in baculovirus-infected cells were apparently less than those in mock-infected cells. This would be due to the shutting-off of host cell protein synthesis induced by baculovirus infection (Miller, 1996 ), but not due to the overexpression of EF-1δ: overexpression of EF-1δ did not affect the expression of the other cellular proteins in COS-7 cells, as described below (Fig. 5C).

(26K): Fig. 4. (A) Immunoblots of electrophoretically separated lysates of Sf9 cells infected with recombinant baculoviruses and COS-7 cells. Lysates from Sf9 cells either mock-infected (lane 1) or infected with Bac-hEF-1δ (lane 2) and COS-7 cells (lane 3) were separated in denaturing gels, transferred onto nitrocellulose sheets and reacted with a rabbit polyclonal antibody to EF-1δ. (B) Lysates from Sf9 cells mock-infected (lane 1) or coinfected with either Bac-hEF-1δ and Bac-GST (lane 2) or Bac-hEF-1δ and Bac-GST-BGLF4 (lane 3) were immunoblotted with the antibody to EF-1δ. (C) Ponceau S staining of the nitrocellulose membrane described in (B). Molecular masses are indicated on the left.

(30K): Fig. 5. (A) Immunoblot of electrophoretically separated lysates of transfected COS-7 cells. Lysates of COS-7 cells mock-transfected (lane 1) or transfected with pME-BGLF4(F) (lane 2) were separated on a denaturing gel, transferred onto a nitrocellulose sheet and reacted with mouse monoclonal antibody to Flag epitope (M2). (B) Lysates of COS-7 cells mock-transfected (lane 1) or transfected with either pME18S (lane 2) or pME-BGLF4(F) (lane 3) were immunoblotted with the antibody to EF-1δ. (C) Ponceau S staining of the nitrocellulose membrane described in (B). Molecular masses are indicated on the left.

In the second series of experiments, we constructed a BGLF4 mammalian cell expression vector, pME-BGLF4(F) (Fig. 1D), to examine whether or not BGLF4 protein mediates modification of endogenous EF-1δ. COS-7 cells were transfected with the indicated plasmids using the DEAE-dextran method (Kawaguchi et al., 2000 ). Cells were harvested 3 days post-transfection and the same amount of protein (as measured by Bio-Rad protein assay kit according to manufacturers instructions) was separated in a denaturing gel and immunoblotted using either the anti-Flag epitope antibody M2 (Sigma) or the anti-EF-1δ antibody. In COS-7 cells transfected with pME-BGLF4(F), Flag epitope-tagged BGLF4 was efficiently and specifically expressed (Fig. 5A). As shown in Fig. 5(B), in COS-7 cells either mock-transfected or transfected with pME18S, the hypophosphorylated form of EF-1δ was dominant (Fig. 5B, lanes 1 and 2), whereas the amount of the hyperphosphorylated form of the protein was markedly increased after transfection with pME-BGLF4(F) (Fig. 5B, lane 3). In the same immunoblot shown in Fig. 5(C), the cellular proteins that were stained with Ponceau S served as a loading control for this experiment, eliminating the possibility that overexpression of EF-1δ had an effect on the expression of the other cellular proteins.

We concluded from these results that BGLF4 protein mediates EF-1δ hyperphosphorylation at a cellular level in mammalian cells.

In the present study, we purified EBV-encoded protein kinase BGLF4 and showed that the purified kinase by itself possesses protein kinase activity in vitro. Furthermore, we identified EF-1δ as an in vivo target of the BGLF4 protein. This is the first identification of a cellular target for BGLF4. These results support our hypothesis that EF-1δ modification in infected mammalian cells is a conserved function that is expressed by all subfamilies of herpesviruses and that the conserved protein kinases encoded by herpesviruses universally mediate modification.

Our knowledge regarding the requirements of viruses with respect to host cell biosynthetic processes depends in part on the identification of cellular proteins that interact with viral proteins. Therefore, if many viral proteins commonly interact with the same cellular protein, this protein would be suggested to be important in the life cycle of viruses, as viruses sometimes use similar strategies regardless of their large diversity in size, structure and genome arrangement. For instance, most DNA tumour viruses encode proteins (e.g. adenovirus E1A and E1B, simian virus 40 T antigen and papillomavirus E6 and E7) that interact with and modify key cellular proteins such as retinoblastoma protein and p53, and transform host cells by a common strategy (i.e. by interaction with the cellular regulatory proteins) (Knipe, 1996 ). Our current studies (Kawaguchi et al., 1997a , 1998 , 1999 ) together with reports from other laboratories, which show that HIV Tat also interacts functionally with a cellular protein and affects translation in vivo (Xiao et al., 1998 ) and that the RNA polymerase of vesicular stomatis virus associates with the EF-1 complex for its activity (Das et al., 1998 ), suggest that EF-1δ is one of the cellular proteins that is universally important in various virus replication systems.

This study also provides important information for the field of EBV research. (i) We expressed enzymatically active BGLF4 using the baculovirus expression system and purified it to near homogeneity. The purified protein was labelled with [γ-³²P]ATP in vitro and the labelling was eliminated by phosphatase treatment, indicating that BGLF4 has the ability to autophosphorylate itself. Our experiments using purified BGLF4 also suggested that it functions, without any cofactors, as a protein kinase. These results supplement those of the study by Chen et al. (2000) in which purification and phosphatase treatment were not performed, and confirm the conclusion that BGLF4 protein is a serine/threonine protein kinase. (ii) It is known that HSV-1 UL13 protein kinase and its HCMV counterpart, UL97, regulate viral gene expression by phosphorylation of various viral and cellular proteins (He et al., 1997 ; Kawaguchi et al., 1998 ; Ng et al., 1998 ; Ogle et al., 1997 ; Prichard et al., 1999 ; Purves et al., 1992 , 1993 ). Furthermore, ORF36 of human herpesvirus-8 (HHV-8), another gammaherpesvirus UL13 homologue, and HCMV UL97 have been reported to phosphorylate ganciclovir and thus induce ganciclovir-mediated cell death (Cannon et al., 1999 ; Litter et al., 1992 ; Sullivan et al., 1992 ). The similarity of BGLF4 to HSV-1 UL13, HCMV UL97 and HHV-8 ORF36 (Cannon et al., 1999 ; Chee et al., 1989 ; Smith & Smith, 1989 ) suggests that BGLF4 plays a role in EBV replication in a manner similar to HSV-1 UL13 and HCMV UL97 and, like HHV-8 ORF36 and HCMV UL97, is a target of anti-viral drugs. Supporting this hypothesis, Chen et al. (2000) reported that an EBV regulatory protein, EA-D, is a potential substrate of BGLF4. As it is easy to express large amounts of enzymatically active BGLF4 and to obtain highly purified BGLF4 using our system, this method will be useful for the further characterization of BGLF4. For example, to identify additional cellular and viral targets of the protein and to analyse its sensitivity to anti-viral drugs.

We thank Dr K. Maruyama for pME18S. This study was supported in part by grants for Scientific Research (Y.K., E.T. and K.H.) and a grant for Scientific Research in Priority Areas (K.H.) from the Ministry of Education, Science, Sports and Culture of Japan. Y.K. was supported by a grant from the Inamori Foundation.

Footnotes

^b This article is dedicated to the memory of Kanji Hirai.