INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Similarly to human alphaherpesviruses, Epstein-Barr virus (EBV), a human gammaherpesvirus, encodes a thymidine kinase (TK). In recent years, alphaherpesvirus TKs have served as important targets for antiviral therapy. Various nucleoside analogs that can be efficiently and preferentially monophosphorylated by these enzymes have been introduced. Following conversion to the nucleoside triphosphate by cellular enzymes, these analogs inhibit the viral DNA polymerase and/or are incorporated into viral DNA causing chain termination (8, 12, 13, 35), selectively inhibiting replication of the virus.

Unlike alphaherpesvirus TKs, characterization of the EBV TK has been slow due to the lack of a tissue culture system in which lytic viral replication occurs efficiently. The existence of this enzyme was initially inferred by demonstration of an EBV-associated TK activity in EBV-infected TK B-cell lines upon chemical treatment or superinfection to induce the EBV lytic cycle (10, 16, 28, 30, 33). Subsequently, the EBV BXLF1 open reading frame (ORF) was demonstrated to encode a protein with TK activity by its ability to convert TK prokaryotic and eukaryotic cells to a TK⁺ phenotype (23-25). The EBV TK is a 607-amino-acid protein, larger than TKs of alphaherpesviruses. In addition to the conserved protein sequences common to herpesviral TKs, the EBV TK has a 243- amino-acid N terminus whose function is unknown (18, 23).

The EBV TK is assumed to accept a broad range of nucleoside analogs as substrates. This assumption is rooted in studies of metabolic activation of antiviral nucleosides in EBV⁺ cell lines which have been treated to induce the lytic cycle in vitro (7, 10, 22, 30) and in clinical reports that ganciclovir {9-([2-hydroxy-1-(hydroxymethyl)ethoxy]methyl) guanine (GCV)} and acyclovir [9-(2-hydroxyethoxymethyl)guanine (ACV)] are useful for in vivo treatment of the EBV lytic disease oral hairy leukoplakia (OHL) (17, 29). Although this evidence is circumstantial and based on analogy with alphaherpesviruses, these observations have been attributed to the EBV TK (1, 23).

Studies with partially purified viral TK from EBV+ B cells have shown the enzyme to be a deoxypyrimidine kinase, capable of phosphorylating 2'-deoxycytidine (dC) as well as thymidine (38). However, such studies are unable to exclude activities contributed by other copurified viral or cellular enzymes. Littler and Arrand (24) report that total cell lysates of bacteria expressing the EBV TK can phosphorylate a broad range of nucleosides and their analogs. In particular, their data show that GCV, a guanosine analog, is a better substrate for EBV TK than for HSV TK. In direct contrast, work by Tung and Summers (39), who also used partially purified lysates of bacteria expressing EBV TK, suggests that EBV TK has a much more restricted substrate specificity. We sought to clarify the activities associated with EBV TK in anticipation that this information might be useful in selecting antiviral therapy for EBV-related diseases.

In this study, we examined the activity of cloned EBV TK expressed in human cells and of an affinity-purified glutathione S-transferase (GST) TK fusion protein. All results are coordinate and demonstrate that EBV TK is able to phosphorylate thymidine (but not dC) and is therefore not a deoxypyrimidine kinase. We present additional biochemical evidence that GCV and ACV are not substrates for the EBV TK. Thymidine analogs, including those used in the treatment of human immunodeficiency virus infection, are effective in inhibiting thymidine phosphorylation by the EBV TK, while analogs of guanosine and cytidine are not. Moreover, removal of the 243-amino-acid N terminus from the EBV TK does not alter its activity, indicating that this structural divergence from alphaherpesvirus TKs is not responsible for its more limited substrate specificity. The implications for treatment of EBV and potentially other gammaherpesvirus-related diseases with nucleoside analogs are discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Nucleosides and cells. Tritium-labeled GCV (14.6 Ci/mmol), ACV (19 Ci/mmol), and dC (25.8 Ci/mmol) were obtained from Moravek Biochemicals, Brea, Calif. Tritium-labeled thymidine (6.7 Ci/mmol) was obtained from New England Nuclear, Boston, Mass. The unlabeled nucleosides 2',3'-dideoxy-2',3'-didehydrothymidine (D4T), 5-bromodeoxyuridine (BrDU), and 3'-azido-2',3'-dideoxythymidine (AZT) were from Sigma, St. Louis, Mo.; 2'-deoxy-2',2'-difluorocythidine (Gemcitabine) was from Eli Lilly, Indianapolis, Ind.; 1-beta-D-arabinofuranosyl-E-5-(2-bromovinyl)uracil (BVaraU) was from Squibb, Princeton, N.J.; GCV was from Syntex Laboratories, Palo Alto, Calif.; ACV was from Glaxo Wellcome, Research Triangle Park, N.C.; and the cytidine nucleotide analog HPMPC (Cidofivir) was obtained from Gilead, Foster City, Calif. All unlabeled nucleosides and nucleotides were stored as concentrated stock solutions at 20°C. 143B TK cells were obtained from the American Type Culture Collection (Rockville, Md.), routinely cultured in Dulbecco's modified Eagle's medium containing 10% heat-inactivated calf serum, 100 Units of penicillin G sodium per ml, and 100 µg of streptomycin sulfate per ml (DMEM-10), and supplemented either with 100 µg of BrDU per ml to maintain a TK phenotype or with 1× hypoxanthine-aminopterin-thymidine (HAT) supplement (Life Technologies, Gaithersburg, Md.) to maintain a TK⁺ phenotype.

Construction of TK vectors. pCMV-HSV-TK, a recombinant adenovirus vector containing the herpes simplex virus type 1 (HSV-1) TK gene under the control of the cytomegalovirus (CMV) immediate-early promoter and enhancer, was a gift of Don Kufe, Dana-Farber Cancer Institute (DFCI), Boston, Mass. (5). An EBV BamHI genomic library was a gift of James Skare, University of Massachusetts, Worcester, Mass., and Jack Strominger, DFCI. The EBV BamHI X fragment, containing the BXLFI ORF which encodes the EBV TK, was subcloned into pBluescript II SK (Stratagene, La Jolla, Calif.) and was designated pBS-BX.

Construction of pGEX-KT expression vectors. The vector pGEX-KT was derived from pGEX-1 (15). To construct pGEX-EBV-TK, the EBV BamHI X fragment was excised from pBS-BX and cloned into the BamHI site of pGEX-KT. The correct orientation was verified by restriction digestion analysis. pGEX-EBV-TK-N was generated by PCR amplification of the internal ORF of the EBV TK gene (18) with primers 5'-cgggatccATGAATGTTCTGAATCTGGATG-3' and 5'-ggaattcCTAGTCCCGATTTCCCCTCTCAA-3', which contain BamHI and EcoRI sites, respectively. These primers amplify a 1,095-bp region of the B95-8 strain of EBV, genome positions 144132 to 143038. The region was amplified by PCR with pGEX-EBV-TK as a template, digested, and directionally cloned into the BamHI/EcoRI site in pGEX-KT. pGEX-HSV-TK was generated by PCR amplification of the HSV TK gene from plasmid pCMV-HSV-TK. The primers used were 5'-cgggatccATGGCTTCGTACCCCTGCCAT-3' and 5'-ggaattcTCAGTTAGCCTCCCCCATCTCCC-3', which contain BamHI and EcoRI sites, respectively. These primers were used to amplify the 1,130-bp HSV TK gene with pCMV-HSV-TK as a template. The product was digested and cloned into pGEX-KT to yield pGEX-HSV-TK.

Expression and purification of GST fusion proteins. The pGEX-KT expression vectors described above were used to express the protein GST and the fusion proteins GST-EBV-TK, GST-EBV-TK-N, and GST-HSV-TK as described by the manufacturer (Pharmacia, Uppsala, Sweden). Briefly, the respective vectors were used to transform competent BL21 Escherichia coli cells. Two-hundred-milliliter cultures were grown to an optical density at 600 nm of 0.6 at 20°C and then induced with 1 mM isopropyl--D-thiogalactoside for 6 h at 20°C. Cells were pelleted at 5,000 × g in a JA-10 rotor (Beckman, Palo Alto, Calif.) and resuspended in 10 ml of lysis buffer (50 mM glucose, 25 mM Tris [pH 8], 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1% Triton X-100). The lysate was clarified by sonication on ice with a 550 Sonic Dismembrator (Fisher Scientific, Pittsburgh, Pa.). Clarified lysates were pelleted for 30 min at 20,000 × g in a JA-20 rotor (Beckman) at 4°C. A GST affinity column was prepared by loading 300 µl of glutathione Sepharose 4B (Pharmacia) into an Econo-Column Chromatography column (Bio-Rad, Richmond, Calif.) and washing with 10 ml of ice-cold phosphate-buffered saline (PBS). All subsequent steps were performed at 4°C. The clarified supernatant was passed over the column twice, after which the column was washed six times with 10 ml of wash buffer (50 mM glucose, 25 mM Tris [pH 8], 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM PMSF). The GST fusion protein was eluted with 200 µl of elution buffer (wash buffer plus 10 mM reduced glutathione [Sigma]) three times, and the eluate was pooled for a 600-µl final volume. The eluted protein was divided into 50-µl aliquots and stored at 70°C. Proteins were quantitated by the method described by Bradford (3), and protein purity was assessed by the intensity of bands in Coomassie blue-stained gels.

Selection of TK-expressing cell lines. 143B TK cells were transfected with equimolar amounts of pCMV-EBV-TK and pCMV-HSV-TK with Lipofectamine (Gibco) according to the manufacturer's directions and cultured in DMEM-10 supplemented with 1× HAT to select for TK⁺ cells. Colonies were isolated and expanded and, respectively, designated 143B EBV TK or 143B HSV TK cells. RNA blot analysis was used to confirm expression of TK RNA, and immunoblot analysis was used to confirm expression of TK protein in the individual clones as described below.

Immunoblot analysis of TK protein. 143B, 143B EBV TK, and 143B HSV TK cells (10⁶) were washed twice with 5 ml of cold PBS and lysed in 1 ml of lysis buffer (1% Nonidet P-40 [NP-40], 0.02% sodium azide, 1 mM PMSF, 20 µg of aprotinin per ml, 1.5 mg of iodoacetamide per ml in PBS) on ice for 20 min. Thirty micrograms of the respective lysates, quantitated by the method of Bradford (3), were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 10% reducing gel and electrophoretically transferred onto nitrocellulose. The nitrocellulose membrane was incubated in blocking solution (PBS, 0.05% Tween 20, 5% nonfat dry milk) for 2 h and then in primary-antibody from serum from a patient with nasopharyngeal carcinoma (NPC) diluted 1:100 in blocking solution for 2 h. After washing with PBS-0.05% Tween 20 (PBS-Tw) twice for 5 min, the membrane was placed in secondary-antibody, horseradish peroxidase-conjugated goat anti-human immunoglobulin G (Sigma) diluted 1:3,000 in blocking solution for 30 min and washed three times for 5 min in PBS-Tw and three times for 5 min in PBS. The blot was developed with the enhanced chemiluminescence system (Amersham, Arlington Heights, Ill.) according to the manufacturer's instructions.

Growth of cells in GCV. 143B TK, 143B EBV TK, and 143B HSV TK cells were plated into 24-well dishes at a density of 2 × 10⁴ cells per well in DMEM-10. Twenty-four hours later, medium was replaced by DMEM-10 containing GCV ranging from 1 to 500 µM. Forty-eight hours after addition of drug, cells were trypsinized, resuspended in DMEM-10, and enumerated in the presence of trypan blue. The concentration of GCV required to reduce the number of viable cells by 50% (ED₅₀) was determined graphically from plots of cell number versus drug concentration.

Preparation of whole-cell lysates. Confluent monolayers of 143B, 143B EBV TK, and 143B HSV TK (10⁸ cells) cells were pelleted at 500 × g in a Sorvall RT 6000D for 5 min, washed once with ice-cold PBS, resuspended in 2 ml of mammalian cell lysis buffer (50 mM Tris [pH 8], 50 mM KCl, 1 mM MgCl₂, 5 mM 2-mercaptoethanol, 1 mM ATP, 1 mM PMSF), and incubated on ice for 10 min. The lysates were clarified by sonication on ice with a 550 Sonic Dismembrator. Sonicated lysates were pelleted at 12,000 × g for 15 min at 4°C. The amount of protein in the supernatants was determined by the Bradford protein assay (3). Aliquots were stored at 80°C.

Phosphorylation assay. (i) Whole-cell lysates. Tritiated nucleosides were used as substrate in kinase reactions utilizing lysates from 143B transfectants as a source of enzyme. Briefly, 40 µg of cell lysate was added to an assay mixture containing 50 mM Tris (pH 7.4), 5 mM MgCl₂, 10 mM NaF, 5 mM 2-mercaptoethanol, 5 mM PMSF, 12.6 mM creatine phosphate, 11.2 U of creatine phosphokinase per ml, 10 mM ATP, and 10 µl of nucleoside (15 µM [³H]thymidine or 7.7 µM [³H]GCV) in a 100-µl final volume. After incubation at 37°C for 30 min, the reactions were stopped and the mixtures were analyzed by disc assaying as follows: 20 µl of each reaction mixture was spotted onto DE-81 discs (Whatman, Maidstone, England) and washed four times for 15 min in 95% ethanol. The discs were dried, and the amount of radioactivity was determined by counting in a LS 6500 Multi-Purpose Scintillation Counter (Beckman). To demonstrate the presence of EBV-specific TK activity, 143B TK, 143B HSV TK, and 143B EBV TK cell lysates were used in disc phosphorylation assays with and without the addition of 10 µl of undiluted serum from a patient with NPC. To determine the relative binding of nucleoside analogs to the EBV TK, cold nucleoside analogs in 10- and 100-fold molar excesses over thymidine were added to phosphorylation reactions, and the disc assay was used for analysis. Results were reported as percent control of the reaction with thymidine only. The 30-min time point was within the linear range of the reaction, which extended beyond 60 min under these conditions.

(ii) GST fusion proteins. Reactions were carried out as described above, except that 100 pmol of GST fusion protein was used as the enzyme source. The assay mixture contained either 15 µM [³H]thymidine, 7.7 µM [³H]GCV, 5.2 µM [³H]ACV, or 3.8 µM [³H]dC with an additional 100 µM respective cold nucleoside or nucleoside analog. After 30 min at 37°C, the reaction mixtures were either analyzed by disc assay as described or prepared for high-performance liquid chromatography (HPLC) as follows: reactions were extracted with 0.5 M perchloric acid (PCA) and neutralized with KHPO₄ and KOH as described in reference 32. Following centrifugation at 12,000 × g for 15 min, supernatants were analyzed by HPLC as described below.

HPLC analysis. HPLC analysis was performed on a 650E Chromatograph using a 600E System Controller and a 484 Tunable Absorbance Detector (Waters). Nucleotides were separated on a POROS 10 SAX anion-exchange column (PerSeptive Biosystems, Framingham, Mass.) with a linear gradient of 5 mM NH₄H₂PO₄buffer (pH 5; buffer A) to 500 mM NH₄H₂PO₄ buffer (pH 5; buffer B) as follows: 5 min of a linear gradient from 100% buffer A to 100% buffer B, 4 min of 100% buffer B, 1 min of linear gradient from 100% buffer B to 100% buffer A, and 2 min of buffer A with a flow rate of 5 ml/min. Elution of nucleosides was monitored by collecting 15-s fractions (1.2 ml) directly into counting vials. To each vial was added 4 ml of ScintiVerse Bio-HP scintillation cocktail (Fisher Scientific). The vials were assayed for radioactivity by scintillation counting. Picomoles of phosphorylated product were calculated by determining the counts per minute per picomole of each [³H]nucleoside before the assay. This procedure could detect 0.03 pmol of [³H]thymidine.

Metabolism of GCV in EBV TK-and HSV TK-transfected cells. 143B, 143B EBV TK, and 143B HSV TK cells were grown to 90% confluence in T-25 flasks. The cells were washed, and the medium was replaced with 3 ml of DMEM-10 containing 50 µCi of [³H]GCV (14.6 Ci/mmol) (final concentration, 1.1 µM) and incubated at 37°C. The cells were harvested at 24 h by trypsin dissociation, pelleting, and washing twice in serum-free DMEM. All traces of supernatant were removed, and the pellets were extracted with ice-cold 0.5 M PCA. The acid-insoluble material was removed by centrifugation, and the supernatant was neutralized by the addition of 1 M KHPO₄ and 4 M KOH. The precipitate was removed by centrifugation at 12,000 × g for 15 min, and the supernatant was used for HPLC analysis.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

TK vector construction. The HSV-1 TK gene and the EBV TK gene were cloned into the adenovirus shuttle vector pCMV for expression in mammalian cells as described in Materials and Methods. Figure 1 shows a map of the EBV BamHI X fragment, indicating the BXLF1 ORF, the internal ORF of BXLF1 which begins at methionine 244 of the full-length EBV TK, and the restriction sites used in cloning. The HSV TK, the EBV TK, and the EBV TK internal ORF were amplified by PCR and cloned into pGEX-KT to facilitate expression and affinity purification of the respective fusion proteins.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Map of the EBV BamHI X fragment showing its position in the EBV genome, the BXLF1 ORF which encodes the EBV TK, and the BXLF1 internal ORF. Sites used in cloning of the EBV TK are indicated. TR, terminal repeat; IR, internal repeat; US, short unique region; UL, long unique region; B, BamHI; B1, BsmfI; N, NheI. Arrows denote position of AUG initiation codons. Map is not drawn to scale.

Expression of EBV TK in mammalian cells. To assess the activity of EBV TK in mammalian cells and to compare its activity with that of HSV TK, both enzymes were expressed from pCMV in 143B TK cells. After transfection of the respective vectors, TK⁺ cells were selected by growth in HAT-containing medium. All of the 143B TK cells grown in HAT medium died. Colonies arose in the EBV TK-transfected cells after selection in HAT medium but consistently grew more slowly and were fewer than colonies arising from the HSV TK-transfected cells (~1:70). Other experiments in our laboratory in which the respective TKs were expressed from the simian virus 40 (SV40) promoter in vector pSG5 (Stratagene) produced identical results (data not shown); in all cases, cells transfected with HSV TK grew more efficiently in HAT medium than cells transfected with EBV TK.

Detection of EBV TK protein. The BXLF1 ORF expresses a protein with an M_r of 67 to 70 when translated in rabbit reticulocyte lysates or in bacteria (24, 27) in accordance with its predicted molecular weight, but the size of the TK protein derived from BXLF1 in mammalian cells has not been described. BXLF1 contains an internal ORF, which when independently expressed in bacteria may also have TK activity (18). This internal ORF has an adjacent favorable Kozak sequence (21) and contains the six consensus regions of herpesvirus TKs identified by Balasubramaniam et al. (2), including the nucleotide and nucleoside binding sites. The protein encoded by the BXLF1 internal ORF is predicted to be equivalent in size to the alphaherpesvirus TKs, i.e., with an M_r of ~37 (18, 23). We therefore investigated whether both species would be synthesized in mammalian cells from BXLF1. Since there are presently no antibodies monospecific to the EBV TK, serum from NPC patients, which is known to contain high-titer antibodies to EBV TK (11), is the single reagent available for detection of EBV TK. Figure 2a shows that NPC serum detects a band with an M_r of ~70 in four EBV TK-transfected 143B cell clones by immunoblotting (lanes 2 to 5). This band is not present in parental 143B cells (lane 1) or in 143B cells transfected with HSV TK (lane 6). No additional protein species migrating at an M_r of ~37 or any other molecular weight was detected by immunoblotting exclusively in BXLF1-transfected 143B cells (data not shown). EBV⁺ B cells stimulated to undergo lytic replication also did not express an inducible protein with an M_r of ~37, whereas induction of a protein with an M_r of ~70 was readily apparent (data not shown).

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Expression and activity of EBV TK in 143B TK cells. (a) Immunoblot analysis of 143B TK cells and viral TK transfectants. Cell lysates were separated on a sodium dodecyl sulfate-10% polyacrylamide gel, transferred to nitrocellulose, and analyzed with serum from a patient with NPC. Lanes: 1, 143B TK; 2 to 5, 143B EBV TK; and 6, 143B HSV TK. The positions of Mr markers are indicated on the left. The arrow denotes the position of the EBV TK. (b) Comparison of thymidine phosphorylation by a representative 143B clone expressing EBV TK with that of 143B cells expressing HSV TK and 143B TK cells. Phosphorylation of [3H]thymidine in counts per minute versus time was measured by the disc assay as described in the text and with lysates of the cell lines 143B (), 143B EBV TK (), and 143B HSV TK (). (c) Phosphorylation of thymidine by 143B cells expressing EBV TK, but not HSV TK or 143B TK cells, is inhibited in the presence of serum containing anti-EBV TK antibodies. Phosphorylation was measured after 1 h by the disc phosphorylation assay in the presence () or absence () of 10 µl of NPC serum.

Activity of EBV TK in mammalian cells. Growth in HAT-containing medium is an indirect indication of TK activity in mammalian cells. As noted by Littler et al. (23), enzymatic activities other than TK may also account for the ability to grow in HAT medium. As a direct test of TK activity, lysates from the 143B and 143B viral TK transfectants were used in an in vitro phosphorylation assay with equal amounts of protein and [³H]thymidine as the substrate. Figure 2b demonstrates the TK activity of lysates from representative clones. No phosphorylation of thymidine occurred when 143B TK cell lysates were used. However, lysates from both of the viral TK-transfected cells showed a linear increase in thymidine phosphorylation throughout the course of the experiment. In every pair of EBV TK and HSV TK clones randomly selected, the slope of the thymidine phosphorylation curve was steeper when HSV TK cell lysates were used, and in no clone did the slope of EBV TK exceed that of HSV TK. Since the expression level of the respective proteins could not be compared adequately with the available antibody reagents, these observations provide statistical evidence that the HSV TK is more efficient at phosphorylating thymidine in these cells than EBV TK.

GCV is not phosphorylated in EBV TK-expressing cells. GCV, a nucleoside analog of guanosine, is known to be efficiently phosphorylated by the HSV TK. However, the literature is inconsistent as to whether GCV is phosphorylated by the EBV TK. Although EBV⁺ B-cell lines have been shown to accumulate GCV di- and triphosphates, the enzymatic activity responsible for their conversion has not been linked directly to EBV TK. Growth of 143B TK, 143B HSV TK, and 143B EBV TK cells in increasing concentrations of GCV was assessed to determine if GCV was toxic in our system, an indication that it had been phosphorylated. No growth inhibition was seen for any of the EBV TK-expressing clones in the presence of GCV compared to 143B TK cells. The ED₅₀s for GCV on representative 143B EBV TK and 143B TK cells were equivalent at ~100 µM. In contrast, all of the HSV TK-expressing 143B clones grew poorly in the presence of GCV. The ED₅₀ of the representative clones was <1 µM (determination of ED₅₀s not shown). Thus, in this system, GCV was toxic only to cells expressing HSV TK.

View larger version (10K): [in this window] [in a new window]   FIG. 3.   Phosphorylation of GCV by 143B TK cells and 143B cells expressing EBV TK and HSV TK. (a) Phosphorylation of GCV detected in 143B HSV TK cells but not in 143B EBV TK or 143B TK cells. Phosphorylation of [3H]GCV in counts per minute versus time was measured by the disc phosphorylation assay as described with lysates of the indicated cell lines. (b) HPLC analysis of acid-soluble metabolites of GCV found in 143B EBV TK, HSV TK, and 143B TK cells after incubation of 1.1 µM [3H]GCV (14.6 Ci/mmol) with the indicated cells for 24 h. , 143B TK; , 143B EBV TK; , 143B HSV TK. MP, DP, and TP, GCV mono-, di-, and triphosphates, respectively.

EBV TK is not a deoxypyrimidine kinase. Prior data concerning the activity of EBV TK have been derived from unpurified or partially purified lysates from EBV-infected B cells or EBV TK-expressing bacteria. Because activity from contaminating proteins could not be excluded, confirmation of enzymatic activity with purified protein would be optimal. For that reason, the EBV TK and HSV TK were expressed as GST fusion proteins in E. coli. Each protein was purified by affinity chromatography and used in phosphorylation assays. In addition, the internal ORF of BXLF1, starting at methionine 244, was similarly cloned and expressed as a GST fusion protein to compare the activity of this N-terminally truncated EBV TK (GST EBV TK N) with that of the full-length protein. Purified GST was used as a control in each experiment. Figure 4a shows the results of a 45-min assay utilizing GST and all three fusion proteins. GST alone had no TK activity, demonstrating that the purification scheme successfully removed bacterial TK. GST HSV TK completed phosphorylation of all substrate by 15 min, while GST EBV TK activity remained linear over the 45-min assay period. GST EBV TK N contained TK activity, extending the results of Holton and Gentry, who first demonstrated this activity in unpurified bacterial lysates (18). The TK activity from GST EBV TK N was essentially indistinguishable from that of the wild-type full-length TK in this assay.

View larger version (10K): [in this window] [in a new window]   FIG. 4.   Kinase activity of GST fusion proteins of EBV TK, HSV TK, and EBV TK N (EBV TK truncated to remove its 243-amino-acid N terminus). Proteins were expressed in bacteria and purified by affinity chromatography on glutathione Sepharose columns as described in Materials and Methods. Phosphorylation of nucleoside versus time was measured by the disc phosphorylation assay as described in the text, with equivalent amounts of GST and GST fusion proteins. Substrates were [3H]thymidine (a) or [3H]deoxycytidine (b). , GST EBV TK; , GST EBV TK N; , GST HSV TK; , GST.

Purified GST EBV TK cannot phosphorylate GCV, ACV or dC. As further verification of whether a particular nucleoside could act as a substrate for EBV TK, phosphorylation assays were performed with GST fusion proteins as the enzyme source and tritiated GCV, ACV, dC, and thymidine as the respective substrates. Products were analyzed by HPLC. The results are summarized in Table 1. The reaction of GST HSV TK with thymidine produces thymidine monophosphate (0.2 pmol) and thymidine diphosphate (13.9 pmol). Thymidylate phosphorylation by HSV TK has been previously reported (40). Although the major product of the reaction of thymidine with GST EBV TK is thymidine monophosphate (14.3 pmol), there is also a minor thymidine diphosphate product (0.25 pmol). Interestingly, reaction of thymidine with GST EBV TK N results in a greater ratio of thymidine diphosphate to thymidine monophosphate than does GST EBV TK. Significantly, neither GST EBV TK nor GST EBV TK N phosphorylated dC, GCV, or ACV, although HSV TK efficiently phosphorylated all three nucleosides.

Competition assay. The ability of a molecule to interfere with the activity of an enzyme is an indication that it may occupy the active site and possibly serve as an alternative substrate for that enzyme. We tested several different nucleoside analogs for their abilities to inhibit thymidine phosphorylation by EBV TK. Lysates of 143B EBV TK cells were used as a source of enzyme in a disc phosphorylation assay with the addition of a 10- or 100-fold molar excess of cold nucleoside analog and compared to an assay with no nucleoside added (Fig. 5). The nucleoside analogs of guanine (GCV and ACV) and cytidine (Gemcitibine and Cidofovir) tested could not interfere with thymidine phosphorylation, even when added in a 100-fold molar excess, indicating they have little or no capacity to block the active site of the EBV TK and cannot act as substrates. On the other hand, all nucleoside analogs of thymidine tested (BVaraU, D4T, AZT, and BrDU) were able to significantly reduce thymidine phosphorylation by the EBV TK, indicating that they do bind the enzyme's active site and may serve as alternative substrates for phosphorylation.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Competition assay measuring the relative inhibition of thymidine phosphorylation by EBV TK by using various nucleoside analogs. Disc phosphorylation assays were performed with 40 µg of protein from 143B EBV TK cells as the enzyme source and 15 µM [3H]thymidine (6.7 Ci/mmol) as the substrate in the presence of nucleoside analog in 10- and 100-fold molar excesses over thymidine as described in Materials and Methods.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

As a step toward developing virus-based treatments for EBV-associated cancers, we undertook a systematic examination of the substrate specificity of the EBV TK. Treatment of alphaherpesvirus infections has relied on selective inhibition of viral replication by nucleoside analogs activated by the virus TK. Based upon limited studies and reasonable extrapolation, it was assumed that the EBV TK would similarly utilize a broad range of antiviral nucleosides. However, functional analyses of the partially purified TK protein from virus-infected cells were inconclusive, and characterization of EBV TK product expressed from bacteria has yielded contradictory results. To better define the biology of EBV TK, we analyzed and compared enzyme activities in two distinct assay systems: (i) following functional expression of the isolated cDNA in human cells and (ii) from purified GST fusion proteins expressed in bacteria. Coordinate assessment of HSV-1 TK activity provided a well-characterized control for all of these experiments.

Expression of EBV TK in 143B TK human cells conferred a TK⁺ phenotype, which was demonstrated by the transfected cell's ability to form colonies in HAT medium. The reduced number and growth rate of colonies that developed following initial transfer of EBV TK into 143B TK cells compared with HSV TK (~1:70) were consistent with biochemical evidence that the HSV TK is the more efficient enzyme. Prior determination of the K_m of HSV TK (~0.5 µM) (6) and EBV TK (22 µM) (39) in independent studies revealed a >40-fold difference in the affinity of each TK for thymidine. Consistent with these observations, TK activity measured in 143B HSV TK-expressing cells was always greater than that in 143B EBV TK cells.

The BXLF1 ORF encoding EBV TK predicts a protein with an M_r of ~70 that differs from alphaherpesvirus TKs by the presence of a 243-amino-acid N terminus of unknown function. An internal ORF present in BXLF1 and beginning at amino acid 244 of the full-length TK has been shown to encode a protein with TK activity upon expression in bacteria. This internal ORF produces protein with an M_r of ~37, which is more similar in size to alphaherpesvirus TKs. 143B EBV TK cells, but neither 143B TK nor 143B HSV TK cells, produced a predicted ~70-M_r protein that could be detected by immunoblotting with a human antiserum containing antibodies to EBV lytic proteins (11). Evidence of an additional 37-M_r TK protein originating from the internal BXLF1 ORF was lacking. Thus, the EBV TK appears to be synthesized only as a full-length protein with a unique N terminus that distinguishes it from the TK of human alphaherpesviruses. Final verification of this observation awaits development of monospecific antibodies to EBV TK.

EBV TK was expressed as a GST fusion protein in bacteria and was purified by affinity chromatography to investigate the activity of the enzyme further. Affinity-purified GST HSV TK was used in the same assays for direct comparison, as was an EBV TK missing its 243-amino-acid N terminus (GST EBV TK N). This permitted independent investigation of the role of the C-terminal protein domain in enzymatic activation. The purified fusion proteins had a high degree of TK activity, and supplementation of the phosphorylation assays with a large excess of cold thymidine was required to prevent immediate depletion of the [³H]thymidine substrate. By use of equimolar amounts of the respective proteins and simultaneous evaluation, the EBV TK was clearly demonstrated to be less efficient than HSV TK (Fig. 4a).

The GST fusion proteins also proved useful for clarification of dC kinase activity associated with EBV TK. Previous investigators had reported that EBV TK, like HSV TK, was a deoxypyrimidine kinase (24, 38). However, an investigation by Tung and Summers (39) indicated that dC was not a substrate for EBV TK. In these determinations, purified enzyme was not utilized; thus, interference or augmentation of activity by contaminating enzymes could not be ruled out. Using affinity-purified enzymes, we observed that GST EBV TK could not phosphorylate dC (Fig. 4b), whereas GST HSV TK had potent dC kinase activity. These results were extended by HPLC analysis of the phosphorylation products (Table 1). No dC monophosphate was detected by this sensitive method, verifying that EBV TK is not a deoxypyrimidine kinase.

Interestingly, this same HPLC analysis suggested that EBV TK has a minor thymidylate kinase activity. Both GST EBV TK and GST EBV TK N converted some thymidine to thymidine diphosphate. This previously unreported activity of EBV TK is of unknown significance, although alterations in both HSV thymidine kinase and thymidylate kinase activities have been shown to determine the sensitivity of HSV to nucleoside antiviral drugs (40).

These results demonstrate significant differences between the activity of EBV TK and HSV TK as assessed in cells or as purified fusion proteins. Such differences may reflect the relative importance of phosphorylated nucleoside pools to the respective life cycles of alphaherpesviruses, which often undergo robust lytic replication, and gammaherpesviruses, which most often are latent. Alternatively, the EBV TK may play a distinct role in the virus life cycle, perhaps involving the N terminus, that is not readily apparent by investigation of the enzyme's ability to phosphorylate thymidine.

The more-limited substrate specificity of EBV TK suggested that the capacity of this enzyme to phosphorylate nucleoside antivirals commonly used in treatment of herpesvirus infections warranted investigation. The antiviral drugs GCV and ACV have been reported to be effective for therapy of the EBV lytic disease OHL (17, 29). GCV in particular has been suggested to variably contribute to the treatment of EBV-associated lymphoproliferative disease and has been shown to be preferentially phosphorylated in EBV⁺ lymphoblastoid cell lines (22). However, these observations have never been linked directly to the EBV TK. The 143B EBV TK-expressing clones permitted examination of viral TK activity independently of other viral proteins in a mammalian cell background. Addition of GCV to these cells would be cytotoxic were they able to phosphorylate this drug. However, growth of 143B EBV TK cells in the presence of increasing concentrations of GCV was indistinguishable from that of 143B TK cells, whereas HSV TK-expressing cells were killed efficiently in <1 µM GCV. In addition, lysates of 143B EBV TK cells were inactive in phosphorylation assays with GCV as the substrate. In EBV TK-expressing cells cultured in the presence of [³H]GCV, HPLC analysis detected only low levels of GCV metabolites, which were similar to those found in 143B TK cells.

Further analysis with affinity-purified GST fusion proteins and HPLC demonstrated that (i) neither GCV nor ACV was a substrate for GST EBV TK or the N-terminal truncated GST EBV TK N and (ii) interference caused by the N-terminal peptide of EBV TK could not account for the distinct substrate specificity of the enzyme compared with that of HSV TK. These studies again show that the activity and substrate specificity of EBV TK expressed in mammalian cells or as a highly purified GST fusion protein were identical and that EBV TK cannot phosphorylate GCV or ACV. If ACV is indeed active in acute infectious mononucleosis, ACV and GCV are active in OHL, or GCV is active in certain EBV-associated tumors, then it is likely that they are being phosphorylated by another mechanism.

At least two alternative mechanisms could account for the capacity of antiviral drugs such as GCV to inhibit virus replication or restrict outgrowth of latently infected cells in the absence of phosphorylation by viral TK. The first is that another viral gene product is responsible for drug activation. A protein kinase gene that is conserved in human herpesviruses has been found (4, 37). The UL97 protein kinase of cytomegalovirus (CMV) and its homologue in varicella-zoster virus (VZV), ORF 47, can phosphorylate GCV (20, 26). Despite low-level GCV phosphorylating activity, UL97 is solely responsible for the susceptibility of CMV to GCV. The BGLF4 ORF encodes the EBV homologue of this protein kinase. Preliminary experiments in our laboratory indicate that BGLF4 is an early gene product, with kinetics of expression very similar to those of EBV TK. If BGLF4 is also a modest GCV kinase, it may account for the minimal phosphorylation described in EBV⁺ B cells (22).

Secondly, these drugs may be converted to their active forms entirely by cellular enzymes. GCV and ACV triphosphates are generated in uninfected cells at low levels, and cellular enzymes have been implicated in their conversion (9, 14, 19, 36). Because of the sensitivity of the EBV DNA polymerase to inhibition by these drugs, minimal phosphorylation may be sufficient to inhibit lytic viral replication, and this idea has been considered (reviewed in reference 31). In latent EBV disease, preferential phosphorylation of GCV in EBV-infected B cells may be a consequence of increased uptake of nutrients into the immortalized cells. Of note, susceptibility of both EBV and the recently described human gammaherpesvirus human herpesvirus 8 (HHV-8) to GCV is evaluated by chemically inducing lytic replication in latently infected cells, since there is no primary lytic replication system for either virus. Such treatment, as well as immortalized cells in vivo, may upregulate cellular enzymes responsible for GCV phosphorylation and consequently phosphorylate GCV to an extent greater than that for uninfected cells. Because the ratio of toxic to therapeutic GCV is narrow as a result of drug recognition by cellular polymerases, GCV could actually be acting in a manner similar to those of chemotherapeutic agents.

Nucleoside analogs of thymidine do effectively prevent thymidine phosphorylation by EBV TK. Based on assessment of competitive inhibition, their rank order for inhibition was BrDU > AZT > D4T > BVaraU. Utility as an inhibitor of EBV lytic replication depends on demonstrated ability to become efficiently phosphorylated (competition is an indirect proof) and selectively recognized by the EBV DNA polymerase. BVaraU, which is useful in the treatment of VZV infections because it selectively inhibits viral replication but is not recognized by the cellular DNA polymerase, would be predicted to be of value for treatment of OHL.

Successful therapy of tumors which harbor latent EBV, on the other hand, should diverge from the normal paradigm of herpes antiviral therapy and result in the selective death of the cell. Such a strategy has two main requirements, i.e., (i) induction of the EBV lytic cycle, or specific lytic proteins, in EBV⁺ tumors and (ii) identification of a drug activated by an EBV protein to produce a cytotoxic compound. While the first requirement is the subject of ongoing investigation in our laboratory, the second is partly investigated here. The results of the competition assay reveal that nucleoside analogs of thymidine appear to best fit the EBV TK active site. Current work is aimed at identifying a thymidine analog capable of being converted to a cytotoxic drug specifically by EBV TK.

In summary, EBV TK has a narrower substrate specificity than the prototype alphaherpesvirus, HSV TK. Using EBV TK expressed in mammalian cells and as a purified GST fusion protein, we have demonstrated that EBV TK does not phosphorylate dC, GCV, or ACV. EBV TK has a minor thymidylate kinase activity. We hypothesize that initiation of lytic replication or selective activation of the TK gene in vivo may enable treatment of EBV-related neoplasms by cytotoxic nucleoside analogs. Our results indicate that nucleoside analogs of thymidine should be investigated for this purpose. The similarity of the newly described HHV-8 TK to EBV TK (38) indicates that a strategy devised to inhibit latent EBV disease may be applicable to HHV-8-infected neoplasms such as Kaposi's sarcoma and body cavity-based lymphomas.